{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 25000,
  "global_step": 183345,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016362595107584062,
      "grad_norm": 9.97589111328125,
      "learning_rate": 5.4540496318516504e-09,
      "loss": 0.8674,
      "step": 10
    },
    {
      "epoch": 0.00032725190215168124,
      "grad_norm": 9.567214012145996,
      "learning_rate": 1.0908099263703301e-08,
      "loss": 0.8385,
      "step": 20
    },
    {
      "epoch": 0.0004908778532275219,
      "grad_norm": 11.185924530029297,
      "learning_rate": 1.636214889555495e-08,
      "loss": 0.8346,
      "step": 30
    },
    {
      "epoch": 0.0006545038043033625,
      "grad_norm": 7.626499652862549,
      "learning_rate": 2.1816198527406602e-08,
      "loss": 0.8504,
      "step": 40
    },
    {
      "epoch": 0.0008181297553792031,
      "grad_norm": 13.738503456115723,
      "learning_rate": 2.7270248159258253e-08,
      "loss": 0.902,
      "step": 50
    },
    {
      "epoch": 0.0009817557064550437,
      "grad_norm": 11.20395565032959,
      "learning_rate": 3.27242977911099e-08,
      "loss": 0.7915,
      "step": 60
    },
    {
      "epoch": 0.0011453816575308843,
      "grad_norm": 9.65567398071289,
      "learning_rate": 3.8178347422961555e-08,
      "loss": 0.8506,
      "step": 70
    },
    {
      "epoch": 0.001309007608606725,
      "grad_norm": 9.179412841796875,
      "learning_rate": 4.3632397054813203e-08,
      "loss": 0.8165,
      "step": 80
    },
    {
      "epoch": 0.0014726335596825656,
      "grad_norm": 9.80648136138916,
      "learning_rate": 4.908644668666486e-08,
      "loss": 0.8035,
      "step": 90
    },
    {
      "epoch": 0.0016362595107584062,
      "grad_norm": 11.501489639282227,
      "learning_rate": 5.4540496318516506e-08,
      "loss": 0.842,
      "step": 100
    },
    {
      "epoch": 0.0017998854618342468,
      "grad_norm": 10.568979263305664,
      "learning_rate": 5.999454595036815e-08,
      "loss": 0.8199,
      "step": 110
    },
    {
      "epoch": 0.0019635114129100874,
      "grad_norm": 10.183390617370605,
      "learning_rate": 6.54485955822198e-08,
      "loss": 0.8713,
      "step": 120
    },
    {
      "epoch": 0.002127137363985928,
      "grad_norm": 10.735555648803711,
      "learning_rate": 7.090264521407145e-08,
      "loss": 0.7941,
      "step": 130
    },
    {
      "epoch": 0.0022907633150617687,
      "grad_norm": 9.965042114257812,
      "learning_rate": 7.635669484592311e-08,
      "loss": 0.8091,
      "step": 140
    },
    {
      "epoch": 0.0024543892661376093,
      "grad_norm": 11.738472938537598,
      "learning_rate": 8.181074447777476e-08,
      "loss": 0.7949,
      "step": 150
    },
    {
      "epoch": 0.00261801521721345,
      "grad_norm": 10.930262565612793,
      "learning_rate": 8.726479410962641e-08,
      "loss": 0.8119,
      "step": 160
    },
    {
      "epoch": 0.0027816411682892906,
      "grad_norm": 7.951657772064209,
      "learning_rate": 9.271884374147805e-08,
      "loss": 0.7635,
      "step": 170
    },
    {
      "epoch": 0.002945267119365131,
      "grad_norm": 10.602276802062988,
      "learning_rate": 9.817289337332972e-08,
      "loss": 0.758,
      "step": 180
    },
    {
      "epoch": 0.003108893070440972,
      "grad_norm": 8.189801216125488,
      "learning_rate": 1.0362694300518136e-07,
      "loss": 0.7383,
      "step": 190
    },
    {
      "epoch": 0.0032725190215168124,
      "grad_norm": 9.993020057678223,
      "learning_rate": 1.0908099263703301e-07,
      "loss": 0.7053,
      "step": 200
    },
    {
      "epoch": 0.003436144972592653,
      "grad_norm": 8.978446960449219,
      "learning_rate": 1.1453504226888466e-07,
      "loss": 0.713,
      "step": 210
    },
    {
      "epoch": 0.0035997709236684937,
      "grad_norm": 7.887515068054199,
      "learning_rate": 1.199890919007363e-07,
      "loss": 0.6923,
      "step": 220
    },
    {
      "epoch": 0.0037633968747443343,
      "grad_norm": 7.2815632820129395,
      "learning_rate": 1.2544314153258796e-07,
      "loss": 0.6419,
      "step": 230
    },
    {
      "epoch": 0.003927022825820175,
      "grad_norm": 7.335437774658203,
      "learning_rate": 1.308971911644396e-07,
      "loss": 0.5948,
      "step": 240
    },
    {
      "epoch": 0.004090648776896016,
      "grad_norm": 4.979640960693359,
      "learning_rate": 1.3635124079629125e-07,
      "loss": 0.5236,
      "step": 250
    },
    {
      "epoch": 0.004254274727971856,
      "grad_norm": 5.992700099945068,
      "learning_rate": 1.418052904281429e-07,
      "loss": 0.5299,
      "step": 260
    },
    {
      "epoch": 0.004417900679047697,
      "grad_norm": 4.274743556976318,
      "learning_rate": 1.4725934005999455e-07,
      "loss": 0.479,
      "step": 270
    },
    {
      "epoch": 0.004581526630123537,
      "grad_norm": 4.86069917678833,
      "learning_rate": 1.5271338969184622e-07,
      "loss": 0.4442,
      "step": 280
    },
    {
      "epoch": 0.004745152581199378,
      "grad_norm": 4.358114719390869,
      "learning_rate": 1.5816743932369784e-07,
      "loss": 0.399,
      "step": 290
    },
    {
      "epoch": 0.004908778532275219,
      "grad_norm": 3.6695806980133057,
      "learning_rate": 1.6362148895554952e-07,
      "loss": 0.3686,
      "step": 300
    },
    {
      "epoch": 0.00507240448335106,
      "grad_norm": 3.6369130611419678,
      "learning_rate": 1.6907553858740117e-07,
      "loss": 0.3371,
      "step": 310
    },
    {
      "epoch": 0.0052360304344269,
      "grad_norm": 3.5566537380218506,
      "learning_rate": 1.7452958821925281e-07,
      "loss": 0.2841,
      "step": 320
    },
    {
      "epoch": 0.005399656385502741,
      "grad_norm": 2.6107022762298584,
      "learning_rate": 1.7998363785110446e-07,
      "loss": 0.2614,
      "step": 330
    },
    {
      "epoch": 0.005563282336578581,
      "grad_norm": 2.713139295578003,
      "learning_rate": 1.854376874829561e-07,
      "loss": 0.2089,
      "step": 340
    },
    {
      "epoch": 0.005726908287654422,
      "grad_norm": 2.8379340171813965,
      "learning_rate": 1.9089173711480776e-07,
      "loss": 0.231,
      "step": 350
    },
    {
      "epoch": 0.005890534238730262,
      "grad_norm": 2.4595868587493896,
      "learning_rate": 1.9634578674665943e-07,
      "loss": 0.2114,
      "step": 360
    },
    {
      "epoch": 0.006054160189806103,
      "grad_norm": 2.988358736038208,
      "learning_rate": 2.0179983637851105e-07,
      "loss": 0.2023,
      "step": 370
    },
    {
      "epoch": 0.006217786140881944,
      "grad_norm": 3.147031545639038,
      "learning_rate": 2.0725388601036273e-07,
      "loss": 0.1946,
      "step": 380
    },
    {
      "epoch": 0.006381412091957785,
      "grad_norm": 2.546103000640869,
      "learning_rate": 2.1270793564221435e-07,
      "loss": 0.1573,
      "step": 390
    },
    {
      "epoch": 0.006545038043033625,
      "grad_norm": 2.4364805221557617,
      "learning_rate": 2.1816198527406602e-07,
      "loss": 0.1953,
      "step": 400
    },
    {
      "epoch": 0.006708663994109466,
      "grad_norm": 2.171234130859375,
      "learning_rate": 2.2361603490591767e-07,
      "loss": 0.1712,
      "step": 410
    },
    {
      "epoch": 0.006872289945185306,
      "grad_norm": 2.1505839824676514,
      "learning_rate": 2.2907008453776932e-07,
      "loss": 0.1713,
      "step": 420
    },
    {
      "epoch": 0.007035915896261147,
      "grad_norm": 2.621340036392212,
      "learning_rate": 2.3452413416962097e-07,
      "loss": 0.1461,
      "step": 430
    },
    {
      "epoch": 0.007199541847336987,
      "grad_norm": 2.4786622524261475,
      "learning_rate": 2.399781838014726e-07,
      "loss": 0.1646,
      "step": 440
    },
    {
      "epoch": 0.007363167798412828,
      "grad_norm": 2.546891927719116,
      "learning_rate": 2.4543223343332426e-07,
      "loss": 0.1606,
      "step": 450
    },
    {
      "epoch": 0.0075267937494886685,
      "grad_norm": 2.1033828258514404,
      "learning_rate": 2.508862830651759e-07,
      "loss": 0.1464,
      "step": 460
    },
    {
      "epoch": 0.00769041970056451,
      "grad_norm": 2.094399929046631,
      "learning_rate": 2.5634033269702756e-07,
      "loss": 0.133,
      "step": 470
    },
    {
      "epoch": 0.00785404565164035,
      "grad_norm": 2.470501661300659,
      "learning_rate": 2.617943823288792e-07,
      "loss": 0.1537,
      "step": 480
    },
    {
      "epoch": 0.00801767160271619,
      "grad_norm": 2.9982750415802,
      "learning_rate": 2.6724843196073086e-07,
      "loss": 0.1262,
      "step": 490
    },
    {
      "epoch": 0.008181297553792032,
      "grad_norm": 2.8229923248291016,
      "learning_rate": 2.727024815925825e-07,
      "loss": 0.127,
      "step": 500
    },
    {
      "epoch": 0.008344923504867872,
      "grad_norm": 2.3655219078063965,
      "learning_rate": 2.781565312244342e-07,
      "loss": 0.1187,
      "step": 510
    },
    {
      "epoch": 0.008508549455943712,
      "grad_norm": 3.2591352462768555,
      "learning_rate": 2.836105808562858e-07,
      "loss": 0.1421,
      "step": 520
    },
    {
      "epoch": 0.008672175407019552,
      "grad_norm": 2.98222017288208,
      "learning_rate": 2.890646304881375e-07,
      "loss": 0.1271,
      "step": 530
    },
    {
      "epoch": 0.008835801358095394,
      "grad_norm": 1.7123140096664429,
      "learning_rate": 2.945186801199891e-07,
      "loss": 0.1121,
      "step": 540
    },
    {
      "epoch": 0.008999427309171235,
      "grad_norm": 2.169299602508545,
      "learning_rate": 2.9997272975184074e-07,
      "loss": 0.1065,
      "step": 550
    },
    {
      "epoch": 0.009163053260247075,
      "grad_norm": 2.1806955337524414,
      "learning_rate": 3.0542677938369244e-07,
      "loss": 0.1074,
      "step": 560
    },
    {
      "epoch": 0.009326679211322917,
      "grad_norm": 2.582362174987793,
      "learning_rate": 3.1088082901554404e-07,
      "loss": 0.0972,
      "step": 570
    },
    {
      "epoch": 0.009490305162398757,
      "grad_norm": 2.4557204246520996,
      "learning_rate": 3.163348786473957e-07,
      "loss": 0.1209,
      "step": 580
    },
    {
      "epoch": 0.009653931113474597,
      "grad_norm": 2.2875640392303467,
      "learning_rate": 3.217889282792474e-07,
      "loss": 0.1062,
      "step": 590
    },
    {
      "epoch": 0.009817557064550437,
      "grad_norm": 2.3615801334381104,
      "learning_rate": 3.2724297791109904e-07,
      "loss": 0.0952,
      "step": 600
    },
    {
      "epoch": 0.00998118301562628,
      "grad_norm": 2.200690746307373,
      "learning_rate": 3.3269702754295063e-07,
      "loss": 0.0958,
      "step": 610
    },
    {
      "epoch": 0.01014480896670212,
      "grad_norm": 2.3095052242279053,
      "learning_rate": 3.3815107717480233e-07,
      "loss": 0.0965,
      "step": 620
    },
    {
      "epoch": 0.01030843491777796,
      "grad_norm": 2.548759937286377,
      "learning_rate": 3.43605126806654e-07,
      "loss": 0.09,
      "step": 630
    },
    {
      "epoch": 0.0104720608688538,
      "grad_norm": 2.3012022972106934,
      "learning_rate": 3.4905917643850563e-07,
      "loss": 0.1115,
      "step": 640
    },
    {
      "epoch": 0.010635686819929642,
      "grad_norm": 2.221707582473755,
      "learning_rate": 3.545132260703572e-07,
      "loss": 0.0783,
      "step": 650
    },
    {
      "epoch": 0.010799312771005482,
      "grad_norm": 2.049013137817383,
      "learning_rate": 3.599672757022089e-07,
      "loss": 0.0842,
      "step": 660
    },
    {
      "epoch": 0.010962938722081322,
      "grad_norm": 2.9410202503204346,
      "learning_rate": 3.6542132533406057e-07,
      "loss": 0.0937,
      "step": 670
    },
    {
      "epoch": 0.011126564673157162,
      "grad_norm": 2.20444393157959,
      "learning_rate": 3.708753749659122e-07,
      "loss": 0.0988,
      "step": 680
    },
    {
      "epoch": 0.011290190624233004,
      "grad_norm": 1.746721863746643,
      "learning_rate": 3.7632942459776387e-07,
      "loss": 0.0854,
      "step": 690
    },
    {
      "epoch": 0.011453816575308844,
      "grad_norm": 2.119723320007324,
      "learning_rate": 3.817834742296155e-07,
      "loss": 0.0854,
      "step": 700
    },
    {
      "epoch": 0.011617442526384684,
      "grad_norm": 2.0076067447662354,
      "learning_rate": 3.8723752386146716e-07,
      "loss": 0.0821,
      "step": 710
    },
    {
      "epoch": 0.011781068477460525,
      "grad_norm": 2.034360647201538,
      "learning_rate": 3.9269157349331886e-07,
      "loss": 0.0891,
      "step": 720
    },
    {
      "epoch": 0.011944694428536367,
      "grad_norm": 2.2256293296813965,
      "learning_rate": 3.9814562312517046e-07,
      "loss": 0.0774,
      "step": 730
    },
    {
      "epoch": 0.012108320379612207,
      "grad_norm": 2.375505208969116,
      "learning_rate": 4.035996727570221e-07,
      "loss": 0.0891,
      "step": 740
    },
    {
      "epoch": 0.012271946330688047,
      "grad_norm": 2.013505458831787,
      "learning_rate": 4.0905372238887375e-07,
      "loss": 0.0904,
      "step": 750
    },
    {
      "epoch": 0.012435572281763887,
      "grad_norm": 2.3231635093688965,
      "learning_rate": 4.1450777202072546e-07,
      "loss": 0.0816,
      "step": 760
    },
    {
      "epoch": 0.012599198232839729,
      "grad_norm": 1.8532401323318481,
      "learning_rate": 4.1996182165257705e-07,
      "loss": 0.0987,
      "step": 770
    },
    {
      "epoch": 0.01276282418391557,
      "grad_norm": 1.8238180875778198,
      "learning_rate": 4.254158712844287e-07,
      "loss": 0.0807,
      "step": 780
    },
    {
      "epoch": 0.01292645013499141,
      "grad_norm": 2.107550859451294,
      "learning_rate": 4.308699209162804e-07,
      "loss": 0.0948,
      "step": 790
    },
    {
      "epoch": 0.01309007608606725,
      "grad_norm": 3.030282974243164,
      "learning_rate": 4.3632397054813205e-07,
      "loss": 0.0847,
      "step": 800
    },
    {
      "epoch": 0.013253702037143092,
      "grad_norm": 2.112107515335083,
      "learning_rate": 4.4177802017998364e-07,
      "loss": 0.0639,
      "step": 810
    },
    {
      "epoch": 0.013417327988218932,
      "grad_norm": 1.8931057453155518,
      "learning_rate": 4.4723206981183534e-07,
      "loss": 0.0842,
      "step": 820
    },
    {
      "epoch": 0.013580953939294772,
      "grad_norm": 2.5364418029785156,
      "learning_rate": 4.52686119443687e-07,
      "loss": 0.0791,
      "step": 830
    },
    {
      "epoch": 0.013744579890370612,
      "grad_norm": 1.8763220310211182,
      "learning_rate": 4.5814016907553864e-07,
      "loss": 0.0784,
      "step": 840
    },
    {
      "epoch": 0.013908205841446454,
      "grad_norm": 1.9687457084655762,
      "learning_rate": 4.6359421870739023e-07,
      "loss": 0.0688,
      "step": 850
    },
    {
      "epoch": 0.014071831792522294,
      "grad_norm": 2.191291570663452,
      "learning_rate": 4.6904826833924193e-07,
      "loss": 0.0728,
      "step": 860
    },
    {
      "epoch": 0.014235457743598134,
      "grad_norm": 2.8170077800750732,
      "learning_rate": 4.745023179710936e-07,
      "loss": 0.0672,
      "step": 870
    },
    {
      "epoch": 0.014399083694673975,
      "grad_norm": 2.217576265335083,
      "learning_rate": 4.799563676029452e-07,
      "loss": 0.0806,
      "step": 880
    },
    {
      "epoch": 0.014562709645749817,
      "grad_norm": 1.9850986003875732,
      "learning_rate": 4.854104172347969e-07,
      "loss": 0.0594,
      "step": 890
    },
    {
      "epoch": 0.014726335596825657,
      "grad_norm": 2.1633453369140625,
      "learning_rate": 4.908644668666485e-07,
      "loss": 0.0686,
      "step": 900
    },
    {
      "epoch": 0.014889961547901497,
      "grad_norm": 1.5597481727600098,
      "learning_rate": 4.963185164985001e-07,
      "loss": 0.0588,
      "step": 910
    },
    {
      "epoch": 0.015053587498977337,
      "grad_norm": 2.029129981994629,
      "learning_rate": 5.017725661303518e-07,
      "loss": 0.0626,
      "step": 920
    },
    {
      "epoch": 0.015217213450053179,
      "grad_norm": 1.7281333208084106,
      "learning_rate": 5.072266157622035e-07,
      "loss": 0.0641,
      "step": 930
    },
    {
      "epoch": 0.01538083940112902,
      "grad_norm": 2.016092538833618,
      "learning_rate": 5.126806653940551e-07,
      "loss": 0.0639,
      "step": 940
    },
    {
      "epoch": 0.01554446535220486,
      "grad_norm": 1.8831826448440552,
      "learning_rate": 5.181347150259067e-07,
      "loss": 0.0686,
      "step": 950
    },
    {
      "epoch": 0.0157080913032807,
      "grad_norm": 1.9618314504623413,
      "learning_rate": 5.235887646577584e-07,
      "loss": 0.0683,
      "step": 960
    },
    {
      "epoch": 0.01587171725435654,
      "grad_norm": 2.9861929416656494,
      "learning_rate": 5.290428142896101e-07,
      "loss": 0.0488,
      "step": 970
    },
    {
      "epoch": 0.01603534320543238,
      "grad_norm": 1.9134918451309204,
      "learning_rate": 5.344968639214617e-07,
      "loss": 0.0535,
      "step": 980
    },
    {
      "epoch": 0.016198969156508224,
      "grad_norm": 1.8413556814193726,
      "learning_rate": 5.399509135533134e-07,
      "loss": 0.0769,
      "step": 990
    },
    {
      "epoch": 0.016362595107584064,
      "grad_norm": 1.905432939529419,
      "learning_rate": 5.45404963185165e-07,
      "loss": 0.048,
      "step": 1000
    },
    {
      "epoch": 0.016526221058659904,
      "grad_norm": 2.4195327758789062,
      "learning_rate": 5.508590128170167e-07,
      "loss": 0.0615,
      "step": 1010
    },
    {
      "epoch": 0.016689847009735744,
      "grad_norm": 1.89035964012146,
      "learning_rate": 5.563130624488684e-07,
      "loss": 0.0616,
      "step": 1020
    },
    {
      "epoch": 0.016853472960811584,
      "grad_norm": 2.0280380249023438,
      "learning_rate": 5.6176711208072e-07,
      "loss": 0.0625,
      "step": 1030
    },
    {
      "epoch": 0.017017098911887425,
      "grad_norm": 2.0723822116851807,
      "learning_rate": 5.672211617125716e-07,
      "loss": 0.0631,
      "step": 1040
    },
    {
      "epoch": 0.017180724862963265,
      "grad_norm": 2.0635242462158203,
      "learning_rate": 5.726752113444233e-07,
      "loss": 0.0686,
      "step": 1050
    },
    {
      "epoch": 0.017344350814039105,
      "grad_norm": 2.308194398880005,
      "learning_rate": 5.78129260976275e-07,
      "loss": 0.0552,
      "step": 1060
    },
    {
      "epoch": 0.01750797676511495,
      "grad_norm": 2.0093746185302734,
      "learning_rate": 5.835833106081266e-07,
      "loss": 0.0667,
      "step": 1070
    },
    {
      "epoch": 0.01767160271619079,
      "grad_norm": 2.412339210510254,
      "learning_rate": 5.890373602399782e-07,
      "loss": 0.0595,
      "step": 1080
    },
    {
      "epoch": 0.01783522866726663,
      "grad_norm": 2.2292864322662354,
      "learning_rate": 5.944914098718299e-07,
      "loss": 0.0558,
      "step": 1090
    },
    {
      "epoch": 0.01799885461834247,
      "grad_norm": 1.8276525735855103,
      "learning_rate": 5.999454595036815e-07,
      "loss": 0.0693,
      "step": 1100
    },
    {
      "epoch": 0.01816248056941831,
      "grad_norm": 2.1351993083953857,
      "learning_rate": 6.053995091355332e-07,
      "loss": 0.0501,
      "step": 1110
    },
    {
      "epoch": 0.01832610652049415,
      "grad_norm": 1.4929412603378296,
      "learning_rate": 6.108535587673849e-07,
      "loss": 0.047,
      "step": 1120
    },
    {
      "epoch": 0.01848973247156999,
      "grad_norm": 2.781839609146118,
      "learning_rate": 6.163076083992365e-07,
      "loss": 0.0629,
      "step": 1130
    },
    {
      "epoch": 0.018653358422645833,
      "grad_norm": 1.4084773063659668,
      "learning_rate": 6.217616580310881e-07,
      "loss": 0.0459,
      "step": 1140
    },
    {
      "epoch": 0.018816984373721674,
      "grad_norm": 2.2878854274749756,
      "learning_rate": 6.272157076629398e-07,
      "loss": 0.0582,
      "step": 1150
    },
    {
      "epoch": 0.018980610324797514,
      "grad_norm": 1.6368273496627808,
      "learning_rate": 6.326697572947914e-07,
      "loss": 0.0461,
      "step": 1160
    },
    {
      "epoch": 0.019144236275873354,
      "grad_norm": 1.8522087335586548,
      "learning_rate": 6.381238069266431e-07,
      "loss": 0.0426,
      "step": 1170
    },
    {
      "epoch": 0.019307862226949194,
      "grad_norm": 1.780583143234253,
      "learning_rate": 6.435778565584948e-07,
      "loss": 0.0485,
      "step": 1180
    },
    {
      "epoch": 0.019471488178025034,
      "grad_norm": 1.912996530532837,
      "learning_rate": 6.490319061903464e-07,
      "loss": 0.0395,
      "step": 1190
    },
    {
      "epoch": 0.019635114129100874,
      "grad_norm": 1.9334734678268433,
      "learning_rate": 6.544859558221981e-07,
      "loss": 0.0631,
      "step": 1200
    },
    {
      "epoch": 0.019798740080176715,
      "grad_norm": 1.8987680673599243,
      "learning_rate": 6.599400054540498e-07,
      "loss": 0.0527,
      "step": 1210
    },
    {
      "epoch": 0.01996236603125256,
      "grad_norm": 2.10577392578125,
      "learning_rate": 6.653940550859013e-07,
      "loss": 0.0457,
      "step": 1220
    },
    {
      "epoch": 0.0201259919823284,
      "grad_norm": 1.821528434753418,
      "learning_rate": 6.70848104717753e-07,
      "loss": 0.0524,
      "step": 1230
    },
    {
      "epoch": 0.02028961793340424,
      "grad_norm": 1.9784972667694092,
      "learning_rate": 6.763021543496047e-07,
      "loss": 0.0503,
      "step": 1240
    },
    {
      "epoch": 0.02045324388448008,
      "grad_norm": 1.4815479516983032,
      "learning_rate": 6.817562039814563e-07,
      "loss": 0.0535,
      "step": 1250
    },
    {
      "epoch": 0.02061686983555592,
      "grad_norm": 1.542924404144287,
      "learning_rate": 6.87210253613308e-07,
      "loss": 0.0532,
      "step": 1260
    },
    {
      "epoch": 0.02078049578663176,
      "grad_norm": 2.2649707794189453,
      "learning_rate": 6.926643032451597e-07,
      "loss": 0.0492,
      "step": 1270
    },
    {
      "epoch": 0.0209441217377076,
      "grad_norm": 1.3413527011871338,
      "learning_rate": 6.981183528770113e-07,
      "loss": 0.0468,
      "step": 1280
    },
    {
      "epoch": 0.02110774768878344,
      "grad_norm": 1.7332874536514282,
      "learning_rate": 7.03572402508863e-07,
      "loss": 0.0549,
      "step": 1290
    },
    {
      "epoch": 0.021271373639859283,
      "grad_norm": 2.0860042572021484,
      "learning_rate": 7.090264521407144e-07,
      "loss": 0.0499,
      "step": 1300
    },
    {
      "epoch": 0.021434999590935123,
      "grad_norm": 1.7805142402648926,
      "learning_rate": 7.144805017725661e-07,
      "loss": 0.0485,
      "step": 1310
    },
    {
      "epoch": 0.021598625542010964,
      "grad_norm": 2.4757297039031982,
      "learning_rate": 7.199345514044178e-07,
      "loss": 0.0543,
      "step": 1320
    },
    {
      "epoch": 0.021762251493086804,
      "grad_norm": 1.8308428525924683,
      "learning_rate": 7.253886010362694e-07,
      "loss": 0.0392,
      "step": 1330
    },
    {
      "epoch": 0.021925877444162644,
      "grad_norm": 1.7218717336654663,
      "learning_rate": 7.308426506681211e-07,
      "loss": 0.0411,
      "step": 1340
    },
    {
      "epoch": 0.022089503395238484,
      "grad_norm": 1.776108980178833,
      "learning_rate": 7.362967002999728e-07,
      "loss": 0.0522,
      "step": 1350
    },
    {
      "epoch": 0.022253129346314324,
      "grad_norm": 2.626009225845337,
      "learning_rate": 7.417507499318244e-07,
      "loss": 0.05,
      "step": 1360
    },
    {
      "epoch": 0.022416755297390165,
      "grad_norm": 2.433746337890625,
      "learning_rate": 7.472047995636761e-07,
      "loss": 0.0584,
      "step": 1370
    },
    {
      "epoch": 0.022580381248466008,
      "grad_norm": 2.144934892654419,
      "learning_rate": 7.526588491955277e-07,
      "loss": 0.0535,
      "step": 1380
    },
    {
      "epoch": 0.02274400719954185,
      "grad_norm": 1.6727941036224365,
      "learning_rate": 7.581128988273793e-07,
      "loss": 0.0455,
      "step": 1390
    },
    {
      "epoch": 0.02290763315061769,
      "grad_norm": 2.0695009231567383,
      "learning_rate": 7.63566948459231e-07,
      "loss": 0.0511,
      "step": 1400
    },
    {
      "epoch": 0.02307125910169353,
      "grad_norm": 1.9832948446273804,
      "learning_rate": 7.690209980910827e-07,
      "loss": 0.0535,
      "step": 1410
    },
    {
      "epoch": 0.02323488505276937,
      "grad_norm": 1.469416618347168,
      "learning_rate": 7.744750477229343e-07,
      "loss": 0.053,
      "step": 1420
    },
    {
      "epoch": 0.02339851100384521,
      "grad_norm": 1.8490004539489746,
      "learning_rate": 7.79929097354786e-07,
      "loss": 0.0485,
      "step": 1430
    },
    {
      "epoch": 0.02356213695492105,
      "grad_norm": 1.8629415035247803,
      "learning_rate": 7.853831469866377e-07,
      "loss": 0.0539,
      "step": 1440
    },
    {
      "epoch": 0.02372576290599689,
      "grad_norm": 2.072378158569336,
      "learning_rate": 7.908371966184892e-07,
      "loss": 0.0429,
      "step": 1450
    },
    {
      "epoch": 0.023889388857072733,
      "grad_norm": 1.342645287513733,
      "learning_rate": 7.962912462503409e-07,
      "loss": 0.0463,
      "step": 1460
    },
    {
      "epoch": 0.024053014808148573,
      "grad_norm": 1.9463469982147217,
      "learning_rate": 8.017452958821926e-07,
      "loss": 0.0368,
      "step": 1470
    },
    {
      "epoch": 0.024216640759224414,
      "grad_norm": 1.7826083898544312,
      "learning_rate": 8.071993455140442e-07,
      "loss": 0.0462,
      "step": 1480
    },
    {
      "epoch": 0.024380266710300254,
      "grad_norm": 1.5456538200378418,
      "learning_rate": 8.126533951458959e-07,
      "loss": 0.0388,
      "step": 1490
    },
    {
      "epoch": 0.024543892661376094,
      "grad_norm": 1.6366924047470093,
      "learning_rate": 8.181074447777475e-07,
      "loss": 0.0421,
      "step": 1500
    },
    {
      "epoch": 0.024707518612451934,
      "grad_norm": 2.5628910064697266,
      "learning_rate": 8.235614944095992e-07,
      "loss": 0.0434,
      "step": 1510
    },
    {
      "epoch": 0.024871144563527774,
      "grad_norm": 2.053734302520752,
      "learning_rate": 8.290155440414509e-07,
      "loss": 0.0476,
      "step": 1520
    },
    {
      "epoch": 0.025034770514603615,
      "grad_norm": 1.4935264587402344,
      "learning_rate": 8.344695936733024e-07,
      "loss": 0.0545,
      "step": 1530
    },
    {
      "epoch": 0.025198396465679458,
      "grad_norm": 1.3891876935958862,
      "learning_rate": 8.399236433051541e-07,
      "loss": 0.0539,
      "step": 1540
    },
    {
      "epoch": 0.0253620224167553,
      "grad_norm": 1.82400381565094,
      "learning_rate": 8.453776929370058e-07,
      "loss": 0.0452,
      "step": 1550
    },
    {
      "epoch": 0.02552564836783114,
      "grad_norm": 1.3610445261001587,
      "learning_rate": 8.508317425688574e-07,
      "loss": 0.0426,
      "step": 1560
    },
    {
      "epoch": 0.02568927431890698,
      "grad_norm": 1.5624148845672607,
      "learning_rate": 8.562857922007091e-07,
      "loss": 0.0408,
      "step": 1570
    },
    {
      "epoch": 0.02585290026998282,
      "grad_norm": 1.5950453281402588,
      "learning_rate": 8.617398418325608e-07,
      "loss": 0.0389,
      "step": 1580
    },
    {
      "epoch": 0.02601652622105866,
      "grad_norm": 1.7113897800445557,
      "learning_rate": 8.671938914644124e-07,
      "loss": 0.0466,
      "step": 1590
    },
    {
      "epoch": 0.0261801521721345,
      "grad_norm": 1.0999279022216797,
      "learning_rate": 8.726479410962641e-07,
      "loss": 0.045,
      "step": 1600
    },
    {
      "epoch": 0.02634377812321034,
      "grad_norm": 1.6611063480377197,
      "learning_rate": 8.781019907281158e-07,
      "loss": 0.0429,
      "step": 1610
    },
    {
      "epoch": 0.026507404074286183,
      "grad_norm": 1.202569842338562,
      "learning_rate": 8.835560403599673e-07,
      "loss": 0.0399,
      "step": 1620
    },
    {
      "epoch": 0.026671030025362023,
      "grad_norm": 1.280436396598816,
      "learning_rate": 8.89010089991819e-07,
      "loss": 0.0308,
      "step": 1630
    },
    {
      "epoch": 0.026834655976437864,
      "grad_norm": 1.6949858665466309,
      "learning_rate": 8.944641396236707e-07,
      "loss": 0.0402,
      "step": 1640
    },
    {
      "epoch": 0.026998281927513704,
      "grad_norm": 2.375936269760132,
      "learning_rate": 8.999181892555223e-07,
      "loss": 0.0474,
      "step": 1650
    },
    {
      "epoch": 0.027161907878589544,
      "grad_norm": 1.9269670248031616,
      "learning_rate": 9.05372238887374e-07,
      "loss": 0.0431,
      "step": 1660
    },
    {
      "epoch": 0.027325533829665384,
      "grad_norm": 1.5133509635925293,
      "learning_rate": 9.108262885192257e-07,
      "loss": 0.0392,
      "step": 1670
    },
    {
      "epoch": 0.027489159780741224,
      "grad_norm": 1.6996502876281738,
      "learning_rate": 9.162803381510773e-07,
      "loss": 0.0453,
      "step": 1680
    },
    {
      "epoch": 0.027652785731817068,
      "grad_norm": 1.4514844417572021,
      "learning_rate": 9.217343877829289e-07,
      "loss": 0.0418,
      "step": 1690
    },
    {
      "epoch": 0.027816411682892908,
      "grad_norm": 1.8763824701309204,
      "learning_rate": 9.271884374147805e-07,
      "loss": 0.0431,
      "step": 1700
    },
    {
      "epoch": 0.02798003763396875,
      "grad_norm": 1.4516658782958984,
      "learning_rate": 9.326424870466322e-07,
      "loss": 0.0424,
      "step": 1710
    },
    {
      "epoch": 0.02814366358504459,
      "grad_norm": 1.6959811449050903,
      "learning_rate": 9.380965366784839e-07,
      "loss": 0.05,
      "step": 1720
    },
    {
      "epoch": 0.02830728953612043,
      "grad_norm": 1.4374184608459473,
      "learning_rate": 9.435505863103355e-07,
      "loss": 0.0345,
      "step": 1730
    },
    {
      "epoch": 0.02847091548719627,
      "grad_norm": 1.1023731231689453,
      "learning_rate": 9.490046359421872e-07,
      "loss": 0.033,
      "step": 1740
    },
    {
      "epoch": 0.02863454143827211,
      "grad_norm": 1.9677906036376953,
      "learning_rate": 9.544586855740388e-07,
      "loss": 0.0373,
      "step": 1750
    },
    {
      "epoch": 0.02879816738934795,
      "grad_norm": 1.621667742729187,
      "learning_rate": 9.599127352058905e-07,
      "loss": 0.0407,
      "step": 1760
    },
    {
      "epoch": 0.028961793340423793,
      "grad_norm": 1.875444769859314,
      "learning_rate": 9.653667848377422e-07,
      "loss": 0.0507,
      "step": 1770
    },
    {
      "epoch": 0.029125419291499633,
      "grad_norm": 1.7537697553634644,
      "learning_rate": 9.708208344695939e-07,
      "loss": 0.0415,
      "step": 1780
    },
    {
      "epoch": 0.029289045242575473,
      "grad_norm": 2.206434965133667,
      "learning_rate": 9.762748841014454e-07,
      "loss": 0.0352,
      "step": 1790
    },
    {
      "epoch": 0.029452671193651313,
      "grad_norm": 1.4959696531295776,
      "learning_rate": 9.81728933733297e-07,
      "loss": 0.0398,
      "step": 1800
    },
    {
      "epoch": 0.029616297144727154,
      "grad_norm": 1.8054625988006592,
      "learning_rate": 9.871829833651488e-07,
      "loss": 0.0243,
      "step": 1810
    },
    {
      "epoch": 0.029779923095802994,
      "grad_norm": 1.892923355102539,
      "learning_rate": 9.926370329970002e-07,
      "loss": 0.0268,
      "step": 1820
    },
    {
      "epoch": 0.029943549046878834,
      "grad_norm": 1.5711841583251953,
      "learning_rate": 9.98091082628852e-07,
      "loss": 0.0472,
      "step": 1830
    },
    {
      "epoch": 0.030107174997954674,
      "grad_norm": 1.6437585353851318,
      "learning_rate": 1.0035451322607036e-06,
      "loss": 0.0335,
      "step": 1840
    },
    {
      "epoch": 0.030270800949030518,
      "grad_norm": 1.483176589012146,
      "learning_rate": 1.0089991818925553e-06,
      "loss": 0.0417,
      "step": 1850
    },
    {
      "epoch": 0.030434426900106358,
      "grad_norm": 1.8856254816055298,
      "learning_rate": 1.014453231524407e-06,
      "loss": 0.0391,
      "step": 1860
    },
    {
      "epoch": 0.030598052851182198,
      "grad_norm": 1.3388725519180298,
      "learning_rate": 1.0199072811562587e-06,
      "loss": 0.0483,
      "step": 1870
    },
    {
      "epoch": 0.03076167880225804,
      "grad_norm": 2.1064863204956055,
      "learning_rate": 1.0253613307881102e-06,
      "loss": 0.0434,
      "step": 1880
    },
    {
      "epoch": 0.03092530475333388,
      "grad_norm": 1.2316155433654785,
      "learning_rate": 1.030815380419962e-06,
      "loss": 0.0371,
      "step": 1890
    },
    {
      "epoch": 0.03108893070440972,
      "grad_norm": 1.4669277667999268,
      "learning_rate": 1.0362694300518134e-06,
      "loss": 0.0337,
      "step": 1900
    },
    {
      "epoch": 0.03125255665548556,
      "grad_norm": 1.5199886560440063,
      "learning_rate": 1.0417234796836651e-06,
      "loss": 0.0402,
      "step": 1910
    },
    {
      "epoch": 0.0314161826065614,
      "grad_norm": 1.8782730102539062,
      "learning_rate": 1.0471775293155168e-06,
      "loss": 0.0347,
      "step": 1920
    },
    {
      "epoch": 0.03157980855763724,
      "grad_norm": 1.899007797241211,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 0.0346,
      "step": 1930
    },
    {
      "epoch": 0.03174343450871308,
      "grad_norm": 1.5304501056671143,
      "learning_rate": 1.0580856285792202e-06,
      "loss": 0.0347,
      "step": 1940
    },
    {
      "epoch": 0.03190706045978892,
      "grad_norm": 1.7002626657485962,
      "learning_rate": 1.063539678211072e-06,
      "loss": 0.0275,
      "step": 1950
    },
    {
      "epoch": 0.03207068641086476,
      "grad_norm": 1.3978679180145264,
      "learning_rate": 1.0689937278429234e-06,
      "loss": 0.0398,
      "step": 1960
    },
    {
      "epoch": 0.032234312361940604,
      "grad_norm": 1.7616475820541382,
      "learning_rate": 1.0744477774747751e-06,
      "loss": 0.0295,
      "step": 1970
    },
    {
      "epoch": 0.03239793831301645,
      "grad_norm": 1.1628918647766113,
      "learning_rate": 1.0799018271066268e-06,
      "loss": 0.0403,
      "step": 1980
    },
    {
      "epoch": 0.032561564264092284,
      "grad_norm": 1.6942625045776367,
      "learning_rate": 1.0853558767384783e-06,
      "loss": 0.0373,
      "step": 1990
    },
    {
      "epoch": 0.03272519021516813,
      "grad_norm": 1.9778265953063965,
      "learning_rate": 1.09080992637033e-06,
      "loss": 0.0366,
      "step": 2000
    },
    {
      "epoch": 0.032888816166243964,
      "grad_norm": 1.5971571207046509,
      "learning_rate": 1.0962639760021817e-06,
      "loss": 0.0407,
      "step": 2010
    },
    {
      "epoch": 0.03305244211731981,
      "grad_norm": 1.4534926414489746,
      "learning_rate": 1.1017180256340334e-06,
      "loss": 0.0404,
      "step": 2020
    },
    {
      "epoch": 0.033216068068395645,
      "grad_norm": 1.2644355297088623,
      "learning_rate": 1.1071720752658851e-06,
      "loss": 0.029,
      "step": 2030
    },
    {
      "epoch": 0.03337969401947149,
      "grad_norm": 1.8251043558120728,
      "learning_rate": 1.1126261248977368e-06,
      "loss": 0.0329,
      "step": 2040
    },
    {
      "epoch": 0.03354331997054733,
      "grad_norm": 1.685950756072998,
      "learning_rate": 1.1180801745295883e-06,
      "loss": 0.0399,
      "step": 2050
    },
    {
      "epoch": 0.03370694592162317,
      "grad_norm": 1.51814866065979,
      "learning_rate": 1.12353422416144e-06,
      "loss": 0.0299,
      "step": 2060
    },
    {
      "epoch": 0.03387057187269901,
      "grad_norm": 1.473401427268982,
      "learning_rate": 1.1289882737932917e-06,
      "loss": 0.0308,
      "step": 2070
    },
    {
      "epoch": 0.03403419782377485,
      "grad_norm": 1.423195242881775,
      "learning_rate": 1.1344423234251432e-06,
      "loss": 0.0305,
      "step": 2080
    },
    {
      "epoch": 0.03419782377485069,
      "grad_norm": 1.4249560832977295,
      "learning_rate": 1.139896373056995e-06,
      "loss": 0.0289,
      "step": 2090
    },
    {
      "epoch": 0.03436144972592653,
      "grad_norm": 1.7325375080108643,
      "learning_rate": 1.1453504226888466e-06,
      "loss": 0.0306,
      "step": 2100
    },
    {
      "epoch": 0.03452507567700237,
      "grad_norm": 1.6845966577529907,
      "learning_rate": 1.1508044723206983e-06,
      "loss": 0.052,
      "step": 2110
    },
    {
      "epoch": 0.03468870162807821,
      "grad_norm": 1.3788546323776245,
      "learning_rate": 1.15625852195255e-06,
      "loss": 0.0316,
      "step": 2120
    },
    {
      "epoch": 0.034852327579154053,
      "grad_norm": 1.4934451580047607,
      "learning_rate": 1.1617125715844015e-06,
      "loss": 0.0375,
      "step": 2130
    },
    {
      "epoch": 0.0350159535302299,
      "grad_norm": 1.194960594177246,
      "learning_rate": 1.1671666212162532e-06,
      "loss": 0.0256,
      "step": 2140
    },
    {
      "epoch": 0.035179579481305734,
      "grad_norm": 1.423689842224121,
      "learning_rate": 1.1726206708481049e-06,
      "loss": 0.0419,
      "step": 2150
    },
    {
      "epoch": 0.03534320543238158,
      "grad_norm": 0.9441061615943909,
      "learning_rate": 1.1780747204799564e-06,
      "loss": 0.0321,
      "step": 2160
    },
    {
      "epoch": 0.035506831383457414,
      "grad_norm": 1.4725145101547241,
      "learning_rate": 1.183528770111808e-06,
      "loss": 0.0309,
      "step": 2170
    },
    {
      "epoch": 0.03567045733453326,
      "grad_norm": 1.6861231327056885,
      "learning_rate": 1.1889828197436598e-06,
      "loss": 0.0348,
      "step": 2180
    },
    {
      "epoch": 0.035834083285609095,
      "grad_norm": 2.181194543838501,
      "learning_rate": 1.1944368693755113e-06,
      "loss": 0.0321,
      "step": 2190
    },
    {
      "epoch": 0.03599770923668494,
      "grad_norm": 2.2255334854125977,
      "learning_rate": 1.199890919007363e-06,
      "loss": 0.0312,
      "step": 2200
    },
    {
      "epoch": 0.03616133518776078,
      "grad_norm": 1.7597378492355347,
      "learning_rate": 1.2053449686392147e-06,
      "loss": 0.0369,
      "step": 2210
    },
    {
      "epoch": 0.03632496113883662,
      "grad_norm": 2.365129232406616,
      "learning_rate": 1.2107990182710664e-06,
      "loss": 0.0312,
      "step": 2220
    },
    {
      "epoch": 0.03648858708991246,
      "grad_norm": 1.986585021018982,
      "learning_rate": 1.216253067902918e-06,
      "loss": 0.0395,
      "step": 2230
    },
    {
      "epoch": 0.0366522130409883,
      "grad_norm": 1.7252691984176636,
      "learning_rate": 1.2217071175347698e-06,
      "loss": 0.0272,
      "step": 2240
    },
    {
      "epoch": 0.03681583899206414,
      "grad_norm": 1.6250042915344238,
      "learning_rate": 1.2271611671666213e-06,
      "loss": 0.0317,
      "step": 2250
    },
    {
      "epoch": 0.03697946494313998,
      "grad_norm": 1.8350541591644287,
      "learning_rate": 1.232615216798473e-06,
      "loss": 0.0283,
      "step": 2260
    },
    {
      "epoch": 0.03714309089421582,
      "grad_norm": 1.3847283124923706,
      "learning_rate": 1.2380692664303247e-06,
      "loss": 0.0288,
      "step": 2270
    },
    {
      "epoch": 0.03730671684529167,
      "grad_norm": 0.9745829105377197,
      "learning_rate": 1.2435233160621762e-06,
      "loss": 0.0284,
      "step": 2280
    },
    {
      "epoch": 0.0374703427963675,
      "grad_norm": 2.5010721683502197,
      "learning_rate": 1.2489773656940279e-06,
      "loss": 0.0276,
      "step": 2290
    },
    {
      "epoch": 0.03763396874744335,
      "grad_norm": 2.3063294887542725,
      "learning_rate": 1.2544314153258796e-06,
      "loss": 0.0329,
      "step": 2300
    },
    {
      "epoch": 0.037797594698519184,
      "grad_norm": 0.9539628624916077,
      "learning_rate": 1.2598854649577313e-06,
      "loss": 0.0316,
      "step": 2310
    },
    {
      "epoch": 0.03796122064959503,
      "grad_norm": 1.117640495300293,
      "learning_rate": 1.2653395145895827e-06,
      "loss": 0.0294,
      "step": 2320
    },
    {
      "epoch": 0.038124846600670864,
      "grad_norm": 1.8013001680374146,
      "learning_rate": 1.2707935642214344e-06,
      "loss": 0.0338,
      "step": 2330
    },
    {
      "epoch": 0.03828847255174671,
      "grad_norm": 1.5713682174682617,
      "learning_rate": 1.2762476138532861e-06,
      "loss": 0.03,
      "step": 2340
    },
    {
      "epoch": 0.038452098502822545,
      "grad_norm": 1.7282088994979858,
      "learning_rate": 1.2817016634851378e-06,
      "loss": 0.0311,
      "step": 2350
    },
    {
      "epoch": 0.03861572445389839,
      "grad_norm": 0.9448469877243042,
      "learning_rate": 1.2871557131169895e-06,
      "loss": 0.0281,
      "step": 2360
    },
    {
      "epoch": 0.03877935040497423,
      "grad_norm": 0.909325122833252,
      "learning_rate": 1.2926097627488413e-06,
      "loss": 0.0263,
      "step": 2370
    },
    {
      "epoch": 0.03894297635605007,
      "grad_norm": 1.1525700092315674,
      "learning_rate": 1.2980638123806927e-06,
      "loss": 0.0279,
      "step": 2380
    },
    {
      "epoch": 0.03910660230712591,
      "grad_norm": 1.975926160812378,
      "learning_rate": 1.3035178620125444e-06,
      "loss": 0.0257,
      "step": 2390
    },
    {
      "epoch": 0.03927022825820175,
      "grad_norm": 1.2981781959533691,
      "learning_rate": 1.3089719116443961e-06,
      "loss": 0.0249,
      "step": 2400
    },
    {
      "epoch": 0.03943385420927759,
      "grad_norm": 1.5993543863296509,
      "learning_rate": 1.3144259612762478e-06,
      "loss": 0.0248,
      "step": 2410
    },
    {
      "epoch": 0.03959748016035343,
      "grad_norm": 1.5007202625274658,
      "learning_rate": 1.3198800109080995e-06,
      "loss": 0.0269,
      "step": 2420
    },
    {
      "epoch": 0.03976110611142927,
      "grad_norm": 1.7699395418167114,
      "learning_rate": 1.3253340605399508e-06,
      "loss": 0.0301,
      "step": 2430
    },
    {
      "epoch": 0.03992473206250512,
      "grad_norm": 1.59510338306427,
      "learning_rate": 1.3307881101718025e-06,
      "loss": 0.0332,
      "step": 2440
    },
    {
      "epoch": 0.04008835801358095,
      "grad_norm": 1.471999168395996,
      "learning_rate": 1.3362421598036542e-06,
      "loss": 0.0361,
      "step": 2450
    },
    {
      "epoch": 0.0402519839646568,
      "grad_norm": 1.2164160013198853,
      "learning_rate": 1.341696209435506e-06,
      "loss": 0.032,
      "step": 2460
    },
    {
      "epoch": 0.040415609915732634,
      "grad_norm": 1.95022451877594,
      "learning_rate": 1.3471502590673576e-06,
      "loss": 0.0239,
      "step": 2470
    },
    {
      "epoch": 0.04057923586680848,
      "grad_norm": 1.029660940170288,
      "learning_rate": 1.3526043086992093e-06,
      "loss": 0.0245,
      "step": 2480
    },
    {
      "epoch": 0.040742861817884314,
      "grad_norm": 1.6159366369247437,
      "learning_rate": 1.3580583583310608e-06,
      "loss": 0.0324,
      "step": 2490
    },
    {
      "epoch": 0.04090648776896016,
      "grad_norm": 1.67780601978302,
      "learning_rate": 1.3635124079629125e-06,
      "loss": 0.0249,
      "step": 2500
    },
    {
      "epoch": 0.041070113720035994,
      "grad_norm": 0.9726197719573975,
      "learning_rate": 1.3689664575947642e-06,
      "loss": 0.0316,
      "step": 2510
    },
    {
      "epoch": 0.04123373967111184,
      "grad_norm": 1.2978094816207886,
      "learning_rate": 1.374420507226616e-06,
      "loss": 0.0266,
      "step": 2520
    },
    {
      "epoch": 0.04139736562218768,
      "grad_norm": 1.2126712799072266,
      "learning_rate": 1.3798745568584676e-06,
      "loss": 0.0277,
      "step": 2530
    },
    {
      "epoch": 0.04156099157326352,
      "grad_norm": 1.6619802713394165,
      "learning_rate": 1.3853286064903193e-06,
      "loss": 0.0293,
      "step": 2540
    },
    {
      "epoch": 0.04172461752433936,
      "grad_norm": 1.3217777013778687,
      "learning_rate": 1.3907826561221708e-06,
      "loss": 0.0281,
      "step": 2550
    },
    {
      "epoch": 0.0418882434754152,
      "grad_norm": 1.415819764137268,
      "learning_rate": 1.3962367057540225e-06,
      "loss": 0.0266,
      "step": 2560
    },
    {
      "epoch": 0.04205186942649104,
      "grad_norm": 1.6335690021514893,
      "learning_rate": 1.4016907553858742e-06,
      "loss": 0.0199,
      "step": 2570
    },
    {
      "epoch": 0.04221549537756688,
      "grad_norm": 1.4788322448730469,
      "learning_rate": 1.407144805017726e-06,
      "loss": 0.0174,
      "step": 2580
    },
    {
      "epoch": 0.04237912132864272,
      "grad_norm": 1.1941033601760864,
      "learning_rate": 1.4125988546495776e-06,
      "loss": 0.0286,
      "step": 2590
    },
    {
      "epoch": 0.04254274727971857,
      "grad_norm": 0.6530599594116211,
      "learning_rate": 1.4180529042814289e-06,
      "loss": 0.0206,
      "step": 2600
    },
    {
      "epoch": 0.0427063732307944,
      "grad_norm": 0.6440643668174744,
      "learning_rate": 1.4235069539132806e-06,
      "loss": 0.0249,
      "step": 2610
    },
    {
      "epoch": 0.04286999918187025,
      "grad_norm": 2.7297959327697754,
      "learning_rate": 1.4289610035451323e-06,
      "loss": 0.0205,
      "step": 2620
    },
    {
      "epoch": 0.043033625132946084,
      "grad_norm": 1.141425609588623,
      "learning_rate": 1.434415053176984e-06,
      "loss": 0.0246,
      "step": 2630
    },
    {
      "epoch": 0.04319725108402193,
      "grad_norm": 1.098841905593872,
      "learning_rate": 1.4398691028088357e-06,
      "loss": 0.0293,
      "step": 2640
    },
    {
      "epoch": 0.043360877035097764,
      "grad_norm": 1.3283329010009766,
      "learning_rate": 1.4453231524406874e-06,
      "loss": 0.0249,
      "step": 2650
    },
    {
      "epoch": 0.04352450298617361,
      "grad_norm": 1.0602494478225708,
      "learning_rate": 1.4507772020725389e-06,
      "loss": 0.0306,
      "step": 2660
    },
    {
      "epoch": 0.043688128937249444,
      "grad_norm": 1.495590329170227,
      "learning_rate": 1.4562312517043906e-06,
      "loss": 0.0217,
      "step": 2670
    },
    {
      "epoch": 0.04385175488832529,
      "grad_norm": 1.8322491645812988,
      "learning_rate": 1.4616853013362423e-06,
      "loss": 0.0221,
      "step": 2680
    },
    {
      "epoch": 0.04401538083940113,
      "grad_norm": 1.0096715688705444,
      "learning_rate": 1.467139350968094e-06,
      "loss": 0.0268,
      "step": 2690
    },
    {
      "epoch": 0.04417900679047697,
      "grad_norm": 1.028315544128418,
      "learning_rate": 1.4725934005999457e-06,
      "loss": 0.0266,
      "step": 2700
    },
    {
      "epoch": 0.04434263274155281,
      "grad_norm": 1.7434896230697632,
      "learning_rate": 1.4780474502317974e-06,
      "loss": 0.0268,
      "step": 2710
    },
    {
      "epoch": 0.04450625869262865,
      "grad_norm": 1.291272521018982,
      "learning_rate": 1.4835014998636489e-06,
      "loss": 0.0239,
      "step": 2720
    },
    {
      "epoch": 0.04466988464370449,
      "grad_norm": 1.2723766565322876,
      "learning_rate": 1.4889555494955006e-06,
      "loss": 0.0246,
      "step": 2730
    },
    {
      "epoch": 0.04483351059478033,
      "grad_norm": 1.0887057781219482,
      "learning_rate": 1.4944095991273523e-06,
      "loss": 0.0229,
      "step": 2740
    },
    {
      "epoch": 0.04499713654585617,
      "grad_norm": 1.857032060623169,
      "learning_rate": 1.4998636487592038e-06,
      "loss": 0.0301,
      "step": 2750
    },
    {
      "epoch": 0.045160762496932016,
      "grad_norm": 1.183394193649292,
      "learning_rate": 1.5053176983910555e-06,
      "loss": 0.0211,
      "step": 2760
    },
    {
      "epoch": 0.04532438844800785,
      "grad_norm": 1.164291262626648,
      "learning_rate": 1.5107717480229072e-06,
      "loss": 0.0236,
      "step": 2770
    },
    {
      "epoch": 0.0454880143990837,
      "grad_norm": 2.5396125316619873,
      "learning_rate": 1.5162257976547587e-06,
      "loss": 0.0354,
      "step": 2780
    },
    {
      "epoch": 0.045651640350159534,
      "grad_norm": 1.3356671333312988,
      "learning_rate": 1.5216798472866104e-06,
      "loss": 0.0248,
      "step": 2790
    },
    {
      "epoch": 0.04581526630123538,
      "grad_norm": 1.0268887281417847,
      "learning_rate": 1.527133896918462e-06,
      "loss": 0.0263,
      "step": 2800
    },
    {
      "epoch": 0.045978892252311214,
      "grad_norm": 0.8928141593933105,
      "learning_rate": 1.5325879465503138e-06,
      "loss": 0.0276,
      "step": 2810
    },
    {
      "epoch": 0.04614251820338706,
      "grad_norm": 1.9548661708831787,
      "learning_rate": 1.5380419961821655e-06,
      "loss": 0.02,
      "step": 2820
    },
    {
      "epoch": 0.0463061441544629,
      "grad_norm": 1.2414358854293823,
      "learning_rate": 1.543496045814017e-06,
      "loss": 0.0249,
      "step": 2830
    },
    {
      "epoch": 0.04646977010553874,
      "grad_norm": 1.6838544607162476,
      "learning_rate": 1.5489500954458687e-06,
      "loss": 0.0255,
      "step": 2840
    },
    {
      "epoch": 0.04663339605661458,
      "grad_norm": 1.1277804374694824,
      "learning_rate": 1.5544041450777204e-06,
      "loss": 0.028,
      "step": 2850
    },
    {
      "epoch": 0.04679702200769042,
      "grad_norm": 0.892866313457489,
      "learning_rate": 1.559858194709572e-06,
      "loss": 0.0247,
      "step": 2860
    },
    {
      "epoch": 0.04696064795876626,
      "grad_norm": 1.0380139350891113,
      "learning_rate": 1.5653122443414238e-06,
      "loss": 0.0271,
      "step": 2870
    },
    {
      "epoch": 0.0471242739098421,
      "grad_norm": 1.732856273651123,
      "learning_rate": 1.5707662939732755e-06,
      "loss": 0.0326,
      "step": 2880
    },
    {
      "epoch": 0.04728789986091794,
      "grad_norm": 1.3144307136535645,
      "learning_rate": 1.5762203436051267e-06,
      "loss": 0.0247,
      "step": 2890
    },
    {
      "epoch": 0.04745152581199378,
      "grad_norm": 0.7988929152488708,
      "learning_rate": 1.5816743932369784e-06,
      "loss": 0.0276,
      "step": 2900
    },
    {
      "epoch": 0.04761515176306962,
      "grad_norm": 1.2821091413497925,
      "learning_rate": 1.5871284428688301e-06,
      "loss": 0.0275,
      "step": 2910
    },
    {
      "epoch": 0.047778777714145466,
      "grad_norm": 1.8612990379333496,
      "learning_rate": 1.5925824925006818e-06,
      "loss": 0.0295,
      "step": 2920
    },
    {
      "epoch": 0.0479424036652213,
      "grad_norm": 1.0249279737472534,
      "learning_rate": 1.5980365421325335e-06,
      "loss": 0.0216,
      "step": 2930
    },
    {
      "epoch": 0.04810602961629715,
      "grad_norm": 1.6574149131774902,
      "learning_rate": 1.6034905917643852e-06,
      "loss": 0.0283,
      "step": 2940
    },
    {
      "epoch": 0.048269655567372984,
      "grad_norm": 0.7182159423828125,
      "learning_rate": 1.6089446413962367e-06,
      "loss": 0.0252,
      "step": 2950
    },
    {
      "epoch": 0.04843328151844883,
      "grad_norm": 1.0928566455841064,
      "learning_rate": 1.6143986910280884e-06,
      "loss": 0.023,
      "step": 2960
    },
    {
      "epoch": 0.048596907469524664,
      "grad_norm": 1.2761528491973877,
      "learning_rate": 1.6198527406599401e-06,
      "loss": 0.0226,
      "step": 2970
    },
    {
      "epoch": 0.04876053342060051,
      "grad_norm": 1.541603446006775,
      "learning_rate": 1.6253067902917918e-06,
      "loss": 0.0237,
      "step": 2980
    },
    {
      "epoch": 0.04892415937167635,
      "grad_norm": 1.1588711738586426,
      "learning_rate": 1.6307608399236435e-06,
      "loss": 0.0235,
      "step": 2990
    },
    {
      "epoch": 0.04908778532275219,
      "grad_norm": 1.2908154726028442,
      "learning_rate": 1.636214889555495e-06,
      "loss": 0.0221,
      "step": 3000
    },
    {
      "epoch": 0.04925141127382803,
      "grad_norm": 1.1763355731964111,
      "learning_rate": 1.6416689391873467e-06,
      "loss": 0.0259,
      "step": 3010
    },
    {
      "epoch": 0.04941503722490387,
      "grad_norm": 0.9689708352088928,
      "learning_rate": 1.6471229888191984e-06,
      "loss": 0.0227,
      "step": 3020
    },
    {
      "epoch": 0.04957866317597971,
      "grad_norm": 1.828675389289856,
      "learning_rate": 1.6525770384510501e-06,
      "loss": 0.0184,
      "step": 3030
    },
    {
      "epoch": 0.04974228912705555,
      "grad_norm": 1.0045007467269897,
      "learning_rate": 1.6580310880829018e-06,
      "loss": 0.0283,
      "step": 3040
    },
    {
      "epoch": 0.04990591507813139,
      "grad_norm": 1.3750196695327759,
      "learning_rate": 1.6634851377147535e-06,
      "loss": 0.0251,
      "step": 3050
    },
    {
      "epoch": 0.05006954102920723,
      "grad_norm": 1.3335890769958496,
      "learning_rate": 1.6689391873466048e-06,
      "loss": 0.0276,
      "step": 3060
    },
    {
      "epoch": 0.05023316698028307,
      "grad_norm": 1.7532286643981934,
      "learning_rate": 1.6743932369784565e-06,
      "loss": 0.0206,
      "step": 3070
    },
    {
      "epoch": 0.050396792931358916,
      "grad_norm": 1.1638535261154175,
      "learning_rate": 1.6798472866103082e-06,
      "loss": 0.0281,
      "step": 3080
    },
    {
      "epoch": 0.05056041888243475,
      "grad_norm": 1.4859319925308228,
      "learning_rate": 1.68530133624216e-06,
      "loss": 0.0245,
      "step": 3090
    },
    {
      "epoch": 0.0507240448335106,
      "grad_norm": 0.9853307008743286,
      "learning_rate": 1.6907553858740116e-06,
      "loss": 0.0219,
      "step": 3100
    },
    {
      "epoch": 0.05088767078458643,
      "grad_norm": 2.2381179332733154,
      "learning_rate": 1.6962094355058633e-06,
      "loss": 0.0213,
      "step": 3110
    },
    {
      "epoch": 0.05105129673566228,
      "grad_norm": 1.471472144126892,
      "learning_rate": 1.7016634851377148e-06,
      "loss": 0.0279,
      "step": 3120
    },
    {
      "epoch": 0.051214922686738114,
      "grad_norm": 0.9398375749588013,
      "learning_rate": 1.7071175347695665e-06,
      "loss": 0.0221,
      "step": 3130
    },
    {
      "epoch": 0.05137854863781396,
      "grad_norm": 1.147047758102417,
      "learning_rate": 1.7125715844014182e-06,
      "loss": 0.0202,
      "step": 3140
    },
    {
      "epoch": 0.0515421745888898,
      "grad_norm": 1.226751446723938,
      "learning_rate": 1.71802563403327e-06,
      "loss": 0.0224,
      "step": 3150
    },
    {
      "epoch": 0.05170580053996564,
      "grad_norm": 1.1850850582122803,
      "learning_rate": 1.7234796836651216e-06,
      "loss": 0.0243,
      "step": 3160
    },
    {
      "epoch": 0.05186942649104148,
      "grad_norm": 1.0775951147079468,
      "learning_rate": 1.7289337332969733e-06,
      "loss": 0.0188,
      "step": 3170
    },
    {
      "epoch": 0.05203305244211732,
      "grad_norm": 1.3040891885757446,
      "learning_rate": 1.7343877829288248e-06,
      "loss": 0.0269,
      "step": 3180
    },
    {
      "epoch": 0.05219667839319316,
      "grad_norm": 1.0553785562515259,
      "learning_rate": 1.7398418325606765e-06,
      "loss": 0.0243,
      "step": 3190
    },
    {
      "epoch": 0.052360304344269,
      "grad_norm": 1.4058150053024292,
      "learning_rate": 1.7452958821925282e-06,
      "loss": 0.0284,
      "step": 3200
    },
    {
      "epoch": 0.05252393029534484,
      "grad_norm": 1.0478510856628418,
      "learning_rate": 1.7507499318243799e-06,
      "loss": 0.0258,
      "step": 3210
    },
    {
      "epoch": 0.05268755624642068,
      "grad_norm": 1.5595393180847168,
      "learning_rate": 1.7562039814562316e-06,
      "loss": 0.0262,
      "step": 3220
    },
    {
      "epoch": 0.05285118219749652,
      "grad_norm": 1.3992499113082886,
      "learning_rate": 1.7616580310880829e-06,
      "loss": 0.0224,
      "step": 3230
    },
    {
      "epoch": 0.053014808148572366,
      "grad_norm": 1.3505454063415527,
      "learning_rate": 1.7671120807199346e-06,
      "loss": 0.0256,
      "step": 3240
    },
    {
      "epoch": 0.0531784340996482,
      "grad_norm": 1.5510951280593872,
      "learning_rate": 1.7725661303517863e-06,
      "loss": 0.0228,
      "step": 3250
    },
    {
      "epoch": 0.05334206005072405,
      "grad_norm": 1.2564942836761475,
      "learning_rate": 1.778020179983638e-06,
      "loss": 0.0243,
      "step": 3260
    },
    {
      "epoch": 0.05350568600179988,
      "grad_norm": 1.0546241998672485,
      "learning_rate": 1.7834742296154897e-06,
      "loss": 0.0234,
      "step": 3270
    },
    {
      "epoch": 0.05366931195287573,
      "grad_norm": 0.5394402146339417,
      "learning_rate": 1.7889282792473414e-06,
      "loss": 0.0229,
      "step": 3280
    },
    {
      "epoch": 0.053832937903951564,
      "grad_norm": 0.988962709903717,
      "learning_rate": 1.7943823288791929e-06,
      "loss": 0.0273,
      "step": 3290
    },
    {
      "epoch": 0.05399656385502741,
      "grad_norm": 0.838750422000885,
      "learning_rate": 1.7998363785110446e-06,
      "loss": 0.0241,
      "step": 3300
    },
    {
      "epoch": 0.05416018980610325,
      "grad_norm": 0.8716315031051636,
      "learning_rate": 1.8052904281428963e-06,
      "loss": 0.0189,
      "step": 3310
    },
    {
      "epoch": 0.05432381575717909,
      "grad_norm": 1.0897934436798096,
      "learning_rate": 1.810744477774748e-06,
      "loss": 0.0188,
      "step": 3320
    },
    {
      "epoch": 0.05448744170825493,
      "grad_norm": 0.9270868301391602,
      "learning_rate": 1.8161985274065997e-06,
      "loss": 0.0187,
      "step": 3330
    },
    {
      "epoch": 0.05465106765933077,
      "grad_norm": 0.9894445538520813,
      "learning_rate": 1.8216525770384514e-06,
      "loss": 0.0181,
      "step": 3340
    },
    {
      "epoch": 0.05481469361040661,
      "grad_norm": 0.8768455386161804,
      "learning_rate": 1.8271066266703029e-06,
      "loss": 0.021,
      "step": 3350
    },
    {
      "epoch": 0.05497831956148245,
      "grad_norm": 0.981567919254303,
      "learning_rate": 1.8325606763021546e-06,
      "loss": 0.0179,
      "step": 3360
    },
    {
      "epoch": 0.05514194551255829,
      "grad_norm": 0.9962620735168457,
      "learning_rate": 1.838014725934006e-06,
      "loss": 0.0201,
      "step": 3370
    },
    {
      "epoch": 0.055305571463634136,
      "grad_norm": 1.4500595331192017,
      "learning_rate": 1.8434687755658577e-06,
      "loss": 0.0326,
      "step": 3380
    },
    {
      "epoch": 0.05546919741470997,
      "grad_norm": 0.881920337677002,
      "learning_rate": 1.8489228251977094e-06,
      "loss": 0.0291,
      "step": 3390
    },
    {
      "epoch": 0.055632823365785816,
      "grad_norm": 0.9697723984718323,
      "learning_rate": 1.854376874829561e-06,
      "loss": 0.0172,
      "step": 3400
    },
    {
      "epoch": 0.05579644931686165,
      "grad_norm": 1.1145638227462769,
      "learning_rate": 1.8598309244614126e-06,
      "loss": 0.0247,
      "step": 3410
    },
    {
      "epoch": 0.0559600752679375,
      "grad_norm": 0.8215819001197815,
      "learning_rate": 1.8652849740932643e-06,
      "loss": 0.0241,
      "step": 3420
    },
    {
      "epoch": 0.05612370121901333,
      "grad_norm": 1.275476336479187,
      "learning_rate": 1.870739023725116e-06,
      "loss": 0.0194,
      "step": 3430
    },
    {
      "epoch": 0.05628732717008918,
      "grad_norm": 1.3093003034591675,
      "learning_rate": 1.8761930733569677e-06,
      "loss": 0.0227,
      "step": 3440
    },
    {
      "epoch": 0.056450953121165014,
      "grad_norm": 1.2893853187561035,
      "learning_rate": 1.8816471229888194e-06,
      "loss": 0.0252,
      "step": 3450
    },
    {
      "epoch": 0.05661457907224086,
      "grad_norm": 0.7612252235412598,
      "learning_rate": 1.887101172620671e-06,
      "loss": 0.0199,
      "step": 3460
    },
    {
      "epoch": 0.0567782050233167,
      "grad_norm": 0.9209619760513306,
      "learning_rate": 1.8925552222525226e-06,
      "loss": 0.0252,
      "step": 3470
    },
    {
      "epoch": 0.05694183097439254,
      "grad_norm": 1.7326873540878296,
      "learning_rate": 1.8980092718843743e-06,
      "loss": 0.0217,
      "step": 3480
    },
    {
      "epoch": 0.05710545692546838,
      "grad_norm": 0.8247413635253906,
      "learning_rate": 1.903463321516226e-06,
      "loss": 0.0204,
      "step": 3490
    },
    {
      "epoch": 0.05726908287654422,
      "grad_norm": 1.678443431854248,
      "learning_rate": 1.9089173711480775e-06,
      "loss": 0.0184,
      "step": 3500
    },
    {
      "epoch": 0.05743270882762006,
      "grad_norm": 1.1757315397262573,
      "learning_rate": 1.9143714207799292e-06,
      "loss": 0.0225,
      "step": 3510
    },
    {
      "epoch": 0.0575963347786959,
      "grad_norm": 0.8371372222900391,
      "learning_rate": 1.919825470411781e-06,
      "loss": 0.0223,
      "step": 3520
    },
    {
      "epoch": 0.05775996072977174,
      "grad_norm": 1.302064061164856,
      "learning_rate": 1.9252795200436326e-06,
      "loss": 0.0233,
      "step": 3530
    },
    {
      "epoch": 0.057923586680847586,
      "grad_norm": 0.9722303748130798,
      "learning_rate": 1.9307335696754843e-06,
      "loss": 0.0256,
      "step": 3540
    },
    {
      "epoch": 0.05808721263192342,
      "grad_norm": 0.8694469332695007,
      "learning_rate": 1.936187619307336e-06,
      "loss": 0.0235,
      "step": 3550
    },
    {
      "epoch": 0.058250838582999266,
      "grad_norm": 1.2569630146026611,
      "learning_rate": 1.9416416689391877e-06,
      "loss": 0.0193,
      "step": 3560
    },
    {
      "epoch": 0.0584144645340751,
      "grad_norm": 1.2085784673690796,
      "learning_rate": 1.947095718571039e-06,
      "loss": 0.0219,
      "step": 3570
    },
    {
      "epoch": 0.058578090485150947,
      "grad_norm": 1.3117491006851196,
      "learning_rate": 1.9525497682028907e-06,
      "loss": 0.0229,
      "step": 3580
    },
    {
      "epoch": 0.05874171643622678,
      "grad_norm": 1.3743319511413574,
      "learning_rate": 1.9580038178347424e-06,
      "loss": 0.0198,
      "step": 3590
    },
    {
      "epoch": 0.05890534238730263,
      "grad_norm": 1.4419586658477783,
      "learning_rate": 1.963457867466594e-06,
      "loss": 0.0227,
      "step": 3600
    },
    {
      "epoch": 0.059068968338378464,
      "grad_norm": 1.910510778427124,
      "learning_rate": 1.968911917098446e-06,
      "loss": 0.023,
      "step": 3610
    },
    {
      "epoch": 0.05923259428945431,
      "grad_norm": 0.9994873404502869,
      "learning_rate": 1.9743659667302975e-06,
      "loss": 0.0194,
      "step": 3620
    },
    {
      "epoch": 0.05939622024053015,
      "grad_norm": 1.761000633239746,
      "learning_rate": 1.9798200163621488e-06,
      "loss": 0.0262,
      "step": 3630
    },
    {
      "epoch": 0.05955984619160599,
      "grad_norm": 0.8472606539726257,
      "learning_rate": 1.9852740659940005e-06,
      "loss": 0.0231,
      "step": 3640
    },
    {
      "epoch": 0.05972347214268183,
      "grad_norm": 1.8871433734893799,
      "learning_rate": 1.990728115625852e-06,
      "loss": 0.0205,
      "step": 3650
    },
    {
      "epoch": 0.05988709809375767,
      "grad_norm": 1.7018417119979858,
      "learning_rate": 1.996182165257704e-06,
      "loss": 0.0262,
      "step": 3660
    },
    {
      "epoch": 0.06005072404483351,
      "grad_norm": 1.1097400188446045,
      "learning_rate": 2.0016362148895556e-06,
      "loss": 0.0232,
      "step": 3670
    },
    {
      "epoch": 0.06021434999590935,
      "grad_norm": 0.8129346370697021,
      "learning_rate": 2.0070902645214073e-06,
      "loss": 0.0188,
      "step": 3680
    },
    {
      "epoch": 0.06037797594698519,
      "grad_norm": 1.0512713193893433,
      "learning_rate": 2.012544314153259e-06,
      "loss": 0.0178,
      "step": 3690
    },
    {
      "epoch": 0.060541601898061036,
      "grad_norm": 0.9962353706359863,
      "learning_rate": 2.0179983637851107e-06,
      "loss": 0.0267,
      "step": 3700
    },
    {
      "epoch": 0.06070522784913687,
      "grad_norm": 1.0234490633010864,
      "learning_rate": 2.0234524134169624e-06,
      "loss": 0.0146,
      "step": 3710
    },
    {
      "epoch": 0.060868853800212716,
      "grad_norm": 0.7434457540512085,
      "learning_rate": 2.028906463048814e-06,
      "loss": 0.0251,
      "step": 3720
    },
    {
      "epoch": 0.06103247975128855,
      "grad_norm": 1.0221445560455322,
      "learning_rate": 2.034360512680666e-06,
      "loss": 0.0227,
      "step": 3730
    },
    {
      "epoch": 0.061196105702364396,
      "grad_norm": 1.1963093280792236,
      "learning_rate": 2.0398145623125175e-06,
      "loss": 0.0216,
      "step": 3740
    },
    {
      "epoch": 0.06135973165344023,
      "grad_norm": 1.6195560693740845,
      "learning_rate": 2.0452686119443688e-06,
      "loss": 0.0229,
      "step": 3750
    },
    {
      "epoch": 0.06152335760451608,
      "grad_norm": 0.9843617677688599,
      "learning_rate": 2.0507226615762205e-06,
      "loss": 0.0237,
      "step": 3760
    },
    {
      "epoch": 0.061686983555591914,
      "grad_norm": 1.0382856130599976,
      "learning_rate": 2.056176711208072e-06,
      "loss": 0.0232,
      "step": 3770
    },
    {
      "epoch": 0.06185060950666776,
      "grad_norm": 1.4391368627548218,
      "learning_rate": 2.061630760839924e-06,
      "loss": 0.0204,
      "step": 3780
    },
    {
      "epoch": 0.0620142354577436,
      "grad_norm": 0.9887288212776184,
      "learning_rate": 2.0670848104717756e-06,
      "loss": 0.0198,
      "step": 3790
    },
    {
      "epoch": 0.06217786140881944,
      "grad_norm": 1.3060282468795776,
      "learning_rate": 2.072538860103627e-06,
      "loss": 0.0221,
      "step": 3800
    },
    {
      "epoch": 0.06234148735989528,
      "grad_norm": 0.882520318031311,
      "learning_rate": 2.0779929097354786e-06,
      "loss": 0.0198,
      "step": 3810
    },
    {
      "epoch": 0.06250511331097112,
      "grad_norm": 1.0814740657806396,
      "learning_rate": 2.0834469593673303e-06,
      "loss": 0.0172,
      "step": 3820
    },
    {
      "epoch": 0.06266873926204695,
      "grad_norm": 1.0201282501220703,
      "learning_rate": 2.088901008999182e-06,
      "loss": 0.0303,
      "step": 3830
    },
    {
      "epoch": 0.0628323652131228,
      "grad_norm": 1.3715721368789673,
      "learning_rate": 2.0943550586310337e-06,
      "loss": 0.0214,
      "step": 3840
    },
    {
      "epoch": 0.06299599116419864,
      "grad_norm": 1.0639108419418335,
      "learning_rate": 2.0998091082628854e-06,
      "loss": 0.0174,
      "step": 3850
    },
    {
      "epoch": 0.06315961711527449,
      "grad_norm": 1.855281949043274,
      "learning_rate": 2.105263157894737e-06,
      "loss": 0.0277,
      "step": 3860
    },
    {
      "epoch": 0.06332324306635033,
      "grad_norm": 0.84293133020401,
      "learning_rate": 2.1107172075265888e-06,
      "loss": 0.0203,
      "step": 3870
    },
    {
      "epoch": 0.06348686901742616,
      "grad_norm": 1.3143112659454346,
      "learning_rate": 2.1161712571584405e-06,
      "loss": 0.0155,
      "step": 3880
    },
    {
      "epoch": 0.063650494968502,
      "grad_norm": 1.0552860498428345,
      "learning_rate": 2.121625306790292e-06,
      "loss": 0.0246,
      "step": 3890
    },
    {
      "epoch": 0.06381412091957785,
      "grad_norm": 0.819998025894165,
      "learning_rate": 2.127079356422144e-06,
      "loss": 0.0154,
      "step": 3900
    },
    {
      "epoch": 0.06397774687065369,
      "grad_norm": 0.787436842918396,
      "learning_rate": 2.1325334060539956e-06,
      "loss": 0.0182,
      "step": 3910
    },
    {
      "epoch": 0.06414137282172952,
      "grad_norm": 1.3263076543807983,
      "learning_rate": 2.137987455685847e-06,
      "loss": 0.0205,
      "step": 3920
    },
    {
      "epoch": 0.06430499877280536,
      "grad_norm": 1.0753942728042603,
      "learning_rate": 2.1434415053176985e-06,
      "loss": 0.0226,
      "step": 3930
    },
    {
      "epoch": 0.06446862472388121,
      "grad_norm": 1.1934694051742554,
      "learning_rate": 2.1488955549495502e-06,
      "loss": 0.0152,
      "step": 3940
    },
    {
      "epoch": 0.06463225067495705,
      "grad_norm": 0.8780665993690491,
      "learning_rate": 2.154349604581402e-06,
      "loss": 0.0182,
      "step": 3950
    },
    {
      "epoch": 0.0647958766260329,
      "grad_norm": 0.690985918045044,
      "learning_rate": 2.1598036542132536e-06,
      "loss": 0.0183,
      "step": 3960
    },
    {
      "epoch": 0.06495950257710872,
      "grad_norm": 1.1936982870101929,
      "learning_rate": 2.165257703845105e-06,
      "loss": 0.0208,
      "step": 3970
    },
    {
      "epoch": 0.06512312852818457,
      "grad_norm": 1.1310224533081055,
      "learning_rate": 2.1707117534769566e-06,
      "loss": 0.0223,
      "step": 3980
    },
    {
      "epoch": 0.06528675447926041,
      "grad_norm": 1.5926493406295776,
      "learning_rate": 2.1761658031088083e-06,
      "loss": 0.0218,
      "step": 3990
    },
    {
      "epoch": 0.06545038043033626,
      "grad_norm": 0.8800981044769287,
      "learning_rate": 2.18161985274066e-06,
      "loss": 0.0243,
      "step": 4000
    },
    {
      "epoch": 0.0656140063814121,
      "grad_norm": 1.2548387050628662,
      "learning_rate": 2.1870739023725117e-06,
      "loss": 0.0187,
      "step": 4010
    },
    {
      "epoch": 0.06577763233248793,
      "grad_norm": 1.114944338798523,
      "learning_rate": 2.1925279520043634e-06,
      "loss": 0.0186,
      "step": 4020
    },
    {
      "epoch": 0.06594125828356377,
      "grad_norm": 1.1938918828964233,
      "learning_rate": 2.197982001636215e-06,
      "loss": 0.0251,
      "step": 4030
    },
    {
      "epoch": 0.06610488423463962,
      "grad_norm": 1.7698301076889038,
      "learning_rate": 2.203436051268067e-06,
      "loss": 0.0169,
      "step": 4040
    },
    {
      "epoch": 0.06626851018571546,
      "grad_norm": 0.8285405039787292,
      "learning_rate": 2.2088901008999185e-06,
      "loss": 0.0234,
      "step": 4050
    },
    {
      "epoch": 0.06643213613679129,
      "grad_norm": 1.1454505920410156,
      "learning_rate": 2.2143441505317702e-06,
      "loss": 0.0176,
      "step": 4060
    },
    {
      "epoch": 0.06659576208786713,
      "grad_norm": 1.0285584926605225,
      "learning_rate": 2.219798200163622e-06,
      "loss": 0.0157,
      "step": 4070
    },
    {
      "epoch": 0.06675938803894298,
      "grad_norm": 0.9753805994987488,
      "learning_rate": 2.2252522497954736e-06,
      "loss": 0.0177,
      "step": 4080
    },
    {
      "epoch": 0.06692301399001882,
      "grad_norm": 1.5349905490875244,
      "learning_rate": 2.230706299427325e-06,
      "loss": 0.022,
      "step": 4090
    },
    {
      "epoch": 0.06708663994109466,
      "grad_norm": 0.9370464086532593,
      "learning_rate": 2.2361603490591766e-06,
      "loss": 0.0147,
      "step": 4100
    },
    {
      "epoch": 0.0672502658921705,
      "grad_norm": 1.2721103429794312,
      "learning_rate": 2.2416143986910283e-06,
      "loss": 0.02,
      "step": 4110
    },
    {
      "epoch": 0.06741389184324634,
      "grad_norm": 0.795366108417511,
      "learning_rate": 2.24706844832288e-06,
      "loss": 0.0256,
      "step": 4120
    },
    {
      "epoch": 0.06757751779432218,
      "grad_norm": 1.9020237922668457,
      "learning_rate": 2.2525224979547317e-06,
      "loss": 0.0192,
      "step": 4130
    },
    {
      "epoch": 0.06774114374539802,
      "grad_norm": 1.2364789247512817,
      "learning_rate": 2.2579765475865834e-06,
      "loss": 0.0175,
      "step": 4140
    },
    {
      "epoch": 0.06790476969647385,
      "grad_norm": 1.258829116821289,
      "learning_rate": 2.2634305972184347e-06,
      "loss": 0.02,
      "step": 4150
    },
    {
      "epoch": 0.0680683956475497,
      "grad_norm": 1.2653374671936035,
      "learning_rate": 2.2688846468502864e-06,
      "loss": 0.0183,
      "step": 4160
    },
    {
      "epoch": 0.06823202159862554,
      "grad_norm": 0.4251483976840973,
      "learning_rate": 2.274338696482138e-06,
      "loss": 0.0118,
      "step": 4170
    },
    {
      "epoch": 0.06839564754970139,
      "grad_norm": 0.4919903874397278,
      "learning_rate": 2.27979274611399e-06,
      "loss": 0.0213,
      "step": 4180
    },
    {
      "epoch": 0.06855927350077723,
      "grad_norm": 1.0949044227600098,
      "learning_rate": 2.2852467957458415e-06,
      "loss": 0.013,
      "step": 4190
    },
    {
      "epoch": 0.06872289945185306,
      "grad_norm": 1.0280132293701172,
      "learning_rate": 2.290700845377693e-06,
      "loss": 0.023,
      "step": 4200
    },
    {
      "epoch": 0.0688865254029289,
      "grad_norm": 0.8359214663505554,
      "learning_rate": 2.296154895009545e-06,
      "loss": 0.0198,
      "step": 4210
    },
    {
      "epoch": 0.06905015135400475,
      "grad_norm": 1.1138992309570312,
      "learning_rate": 2.3016089446413966e-06,
      "loss": 0.0163,
      "step": 4220
    },
    {
      "epoch": 0.06921377730508059,
      "grad_norm": 1.1474568843841553,
      "learning_rate": 2.3070629942732483e-06,
      "loss": 0.0156,
      "step": 4230
    },
    {
      "epoch": 0.06937740325615642,
      "grad_norm": 1.2642734050750732,
      "learning_rate": 2.3125170439051e-06,
      "loss": 0.0202,
      "step": 4240
    },
    {
      "epoch": 0.06954102920723226,
      "grad_norm": 0.8628132343292236,
      "learning_rate": 2.3179710935369517e-06,
      "loss": 0.0369,
      "step": 4250
    },
    {
      "epoch": 0.06970465515830811,
      "grad_norm": 0.5450739860534668,
      "learning_rate": 2.323425143168803e-06,
      "loss": 0.022,
      "step": 4260
    },
    {
      "epoch": 0.06986828110938395,
      "grad_norm": 0.7991898655891418,
      "learning_rate": 2.3288791928006547e-06,
      "loss": 0.019,
      "step": 4270
    },
    {
      "epoch": 0.0700319070604598,
      "grad_norm": 1.1290448904037476,
      "learning_rate": 2.3343332424325064e-06,
      "loss": 0.0165,
      "step": 4280
    },
    {
      "epoch": 0.07019553301153562,
      "grad_norm": 1.0869791507720947,
      "learning_rate": 2.339787292064358e-06,
      "loss": 0.0218,
      "step": 4290
    },
    {
      "epoch": 0.07035915896261147,
      "grad_norm": 0.9370965957641602,
      "learning_rate": 2.3452413416962098e-06,
      "loss": 0.0155,
      "step": 4300
    },
    {
      "epoch": 0.07052278491368731,
      "grad_norm": 0.9847501516342163,
      "learning_rate": 2.3506953913280615e-06,
      "loss": 0.0148,
      "step": 4310
    },
    {
      "epoch": 0.07068641086476316,
      "grad_norm": 1.3345344066619873,
      "learning_rate": 2.3561494409599128e-06,
      "loss": 0.0139,
      "step": 4320
    },
    {
      "epoch": 0.070850036815839,
      "grad_norm": 1.071442723274231,
      "learning_rate": 2.3616034905917645e-06,
      "loss": 0.0274,
      "step": 4330
    },
    {
      "epoch": 0.07101366276691483,
      "grad_norm": 0.9193683862686157,
      "learning_rate": 2.367057540223616e-06,
      "loss": 0.0211,
      "step": 4340
    },
    {
      "epoch": 0.07117728871799067,
      "grad_norm": 0.9321464896202087,
      "learning_rate": 2.372511589855468e-06,
      "loss": 0.0155,
      "step": 4350
    },
    {
      "epoch": 0.07134091466906652,
      "grad_norm": 1.0057884454727173,
      "learning_rate": 2.3779656394873196e-06,
      "loss": 0.0182,
      "step": 4360
    },
    {
      "epoch": 0.07150454062014236,
      "grad_norm": 1.0530040264129639,
      "learning_rate": 2.383419689119171e-06,
      "loss": 0.0252,
      "step": 4370
    },
    {
      "epoch": 0.07166816657121819,
      "grad_norm": 0.2803080081939697,
      "learning_rate": 2.3888737387510225e-06,
      "loss": 0.0255,
      "step": 4380
    },
    {
      "epoch": 0.07183179252229403,
      "grad_norm": 0.7525278329849243,
      "learning_rate": 2.3943277883828742e-06,
      "loss": 0.0197,
      "step": 4390
    },
    {
      "epoch": 0.07199541847336988,
      "grad_norm": 1.0929770469665527,
      "learning_rate": 2.399781838014726e-06,
      "loss": 0.0202,
      "step": 4400
    },
    {
      "epoch": 0.07215904442444572,
      "grad_norm": 1.0513255596160889,
      "learning_rate": 2.4052358876465776e-06,
      "loss": 0.0126,
      "step": 4410
    },
    {
      "epoch": 0.07232267037552156,
      "grad_norm": 0.7591415047645569,
      "learning_rate": 2.4106899372784293e-06,
      "loss": 0.0182,
      "step": 4420
    },
    {
      "epoch": 0.0724862963265974,
      "grad_norm": 1.4326552152633667,
      "learning_rate": 2.416143986910281e-06,
      "loss": 0.0212,
      "step": 4430
    },
    {
      "epoch": 0.07264992227767324,
      "grad_norm": 1.0587764978408813,
      "learning_rate": 2.4215980365421327e-06,
      "loss": 0.0187,
      "step": 4440
    },
    {
      "epoch": 0.07281354822874908,
      "grad_norm": 1.3421663045883179,
      "learning_rate": 2.4270520861739844e-06,
      "loss": 0.0242,
      "step": 4450
    },
    {
      "epoch": 0.07297717417982492,
      "grad_norm": 0.5612053275108337,
      "learning_rate": 2.432506135805836e-06,
      "loss": 0.017,
      "step": 4460
    },
    {
      "epoch": 0.07314080013090075,
      "grad_norm": 1.1192561388015747,
      "learning_rate": 2.437960185437688e-06,
      "loss": 0.0169,
      "step": 4470
    },
    {
      "epoch": 0.0733044260819766,
      "grad_norm": 0.8539515733718872,
      "learning_rate": 2.4434142350695395e-06,
      "loss": 0.016,
      "step": 4480
    },
    {
      "epoch": 0.07346805203305244,
      "grad_norm": 1.0323679447174072,
      "learning_rate": 2.448868284701391e-06,
      "loss": 0.0154,
      "step": 4490
    },
    {
      "epoch": 0.07363167798412829,
      "grad_norm": 1.013827919960022,
      "learning_rate": 2.4543223343332425e-06,
      "loss": 0.0177,
      "step": 4500
    },
    {
      "epoch": 0.07379530393520413,
      "grad_norm": 0.909745454788208,
      "learning_rate": 2.4597763839650942e-06,
      "loss": 0.0234,
      "step": 4510
    },
    {
      "epoch": 0.07395892988627996,
      "grad_norm": 0.6795579791069031,
      "learning_rate": 2.465230433596946e-06,
      "loss": 0.0214,
      "step": 4520
    },
    {
      "epoch": 0.0741225558373558,
      "grad_norm": 1.1066290140151978,
      "learning_rate": 2.4706844832287976e-06,
      "loss": 0.025,
      "step": 4530
    },
    {
      "epoch": 0.07428618178843165,
      "grad_norm": 0.8269834518432617,
      "learning_rate": 2.4761385328606493e-06,
      "loss": 0.0174,
      "step": 4540
    },
    {
      "epoch": 0.07444980773950749,
      "grad_norm": 0.977855920791626,
      "learning_rate": 2.4815925824925006e-06,
      "loss": 0.0175,
      "step": 4550
    },
    {
      "epoch": 0.07461343369058333,
      "grad_norm": 0.9118476510047913,
      "learning_rate": 2.4870466321243523e-06,
      "loss": 0.0206,
      "step": 4560
    },
    {
      "epoch": 0.07477705964165916,
      "grad_norm": 0.6684682965278625,
      "learning_rate": 2.492500681756204e-06,
      "loss": 0.0156,
      "step": 4570
    },
    {
      "epoch": 0.074940685592735,
      "grad_norm": 0.7412125468254089,
      "learning_rate": 2.4979547313880557e-06,
      "loss": 0.0149,
      "step": 4580
    },
    {
      "epoch": 0.07510431154381085,
      "grad_norm": 1.01152503490448,
      "learning_rate": 2.5034087810199074e-06,
      "loss": 0.0156,
      "step": 4590
    },
    {
      "epoch": 0.0752679374948867,
      "grad_norm": 1.2604584693908691,
      "learning_rate": 2.508862830651759e-06,
      "loss": 0.0171,
      "step": 4600
    },
    {
      "epoch": 0.07543156344596252,
      "grad_norm": 0.8666785955429077,
      "learning_rate": 2.514316880283611e-06,
      "loss": 0.0166,
      "step": 4610
    },
    {
      "epoch": 0.07559518939703837,
      "grad_norm": 1.079569697380066,
      "learning_rate": 2.5197709299154625e-06,
      "loss": 0.0166,
      "step": 4620
    },
    {
      "epoch": 0.07575881534811421,
      "grad_norm": 1.1537538766860962,
      "learning_rate": 2.525224979547314e-06,
      "loss": 0.0208,
      "step": 4630
    },
    {
      "epoch": 0.07592244129919005,
      "grad_norm": 1.1425799131393433,
      "learning_rate": 2.5306790291791655e-06,
      "loss": 0.0228,
      "step": 4640
    },
    {
      "epoch": 0.0760860672502659,
      "grad_norm": 1.1060936450958252,
      "learning_rate": 2.536133078811017e-06,
      "loss": 0.0206,
      "step": 4650
    },
    {
      "epoch": 0.07624969320134173,
      "grad_norm": 0.9448750615119934,
      "learning_rate": 2.541587128442869e-06,
      "loss": 0.0198,
      "step": 4660
    },
    {
      "epoch": 0.07641331915241757,
      "grad_norm": 0.7279178500175476,
      "learning_rate": 2.5470411780747206e-06,
      "loss": 0.0197,
      "step": 4670
    },
    {
      "epoch": 0.07657694510349342,
      "grad_norm": 0.9074788689613342,
      "learning_rate": 2.5524952277065723e-06,
      "loss": 0.0182,
      "step": 4680
    },
    {
      "epoch": 0.07674057105456926,
      "grad_norm": 1.2299587726593018,
      "learning_rate": 2.557949277338424e-06,
      "loss": 0.0202,
      "step": 4690
    },
    {
      "epoch": 0.07690419700564509,
      "grad_norm": 0.9038825035095215,
      "learning_rate": 2.5634033269702757e-06,
      "loss": 0.0189,
      "step": 4700
    },
    {
      "epoch": 0.07706782295672093,
      "grad_norm": 1.1351773738861084,
      "learning_rate": 2.5688573766021274e-06,
      "loss": 0.0186,
      "step": 4710
    },
    {
      "epoch": 0.07723144890779678,
      "grad_norm": 0.8688490390777588,
      "learning_rate": 2.574311426233979e-06,
      "loss": 0.0145,
      "step": 4720
    },
    {
      "epoch": 0.07739507485887262,
      "grad_norm": 0.9342639446258545,
      "learning_rate": 2.579765475865831e-06,
      "loss": 0.0179,
      "step": 4730
    },
    {
      "epoch": 0.07755870080994846,
      "grad_norm": 0.8001863956451416,
      "learning_rate": 2.5852195254976825e-06,
      "loss": 0.0196,
      "step": 4740
    },
    {
      "epoch": 0.0777223267610243,
      "grad_norm": 1.0900238752365112,
      "learning_rate": 2.5906735751295338e-06,
      "loss": 0.0259,
      "step": 4750
    },
    {
      "epoch": 0.07788595271210014,
      "grad_norm": 0.6990840435028076,
      "learning_rate": 2.5961276247613855e-06,
      "loss": 0.0122,
      "step": 4760
    },
    {
      "epoch": 0.07804957866317598,
      "grad_norm": 0.6042187809944153,
      "learning_rate": 2.601581674393237e-06,
      "loss": 0.016,
      "step": 4770
    },
    {
      "epoch": 0.07821320461425182,
      "grad_norm": 0.8377915620803833,
      "learning_rate": 2.607035724025089e-06,
      "loss": 0.0134,
      "step": 4780
    },
    {
      "epoch": 0.07837683056532765,
      "grad_norm": 1.0652810335159302,
      "learning_rate": 2.6124897736569406e-06,
      "loss": 0.0173,
      "step": 4790
    },
    {
      "epoch": 0.0785404565164035,
      "grad_norm": 1.0956909656524658,
      "learning_rate": 2.6179438232887923e-06,
      "loss": 0.017,
      "step": 4800
    },
    {
      "epoch": 0.07870408246747934,
      "grad_norm": 1.2536448240280151,
      "learning_rate": 2.623397872920644e-06,
      "loss": 0.0155,
      "step": 4810
    },
    {
      "epoch": 0.07886770841855519,
      "grad_norm": 0.9264737963676453,
      "learning_rate": 2.6288519225524957e-06,
      "loss": 0.0186,
      "step": 4820
    },
    {
      "epoch": 0.07903133436963103,
      "grad_norm": 1.7796748876571655,
      "learning_rate": 2.6343059721843474e-06,
      "loss": 0.0171,
      "step": 4830
    },
    {
      "epoch": 0.07919496032070686,
      "grad_norm": 1.0903502702713013,
      "learning_rate": 2.639760021816199e-06,
      "loss": 0.0169,
      "step": 4840
    },
    {
      "epoch": 0.0793585862717827,
      "grad_norm": 0.7941282987594604,
      "learning_rate": 2.6452140714480508e-06,
      "loss": 0.0157,
      "step": 4850
    },
    {
      "epoch": 0.07952221222285855,
      "grad_norm": 0.9998249411582947,
      "learning_rate": 2.6506681210799016e-06,
      "loss": 0.0199,
      "step": 4860
    },
    {
      "epoch": 0.07968583817393439,
      "grad_norm": 0.5043797492980957,
      "learning_rate": 2.6561221707117533e-06,
      "loss": 0.0158,
      "step": 4870
    },
    {
      "epoch": 0.07984946412501023,
      "grad_norm": 0.6337336897850037,
      "learning_rate": 2.661576220343605e-06,
      "loss": 0.015,
      "step": 4880
    },
    {
      "epoch": 0.08001309007608606,
      "grad_norm": 0.759972095489502,
      "learning_rate": 2.6670302699754567e-06,
      "loss": 0.0114,
      "step": 4890
    },
    {
      "epoch": 0.0801767160271619,
      "grad_norm": 0.8151446580886841,
      "learning_rate": 2.6724843196073084e-06,
      "loss": 0.0177,
      "step": 4900
    },
    {
      "epoch": 0.08034034197823775,
      "grad_norm": 0.7381401658058167,
      "learning_rate": 2.67793836923916e-06,
      "loss": 0.0143,
      "step": 4910
    },
    {
      "epoch": 0.0805039679293136,
      "grad_norm": 0.5805257558822632,
      "learning_rate": 2.683392418871012e-06,
      "loss": 0.0121,
      "step": 4920
    },
    {
      "epoch": 0.08066759388038942,
      "grad_norm": 1.1568772792816162,
      "learning_rate": 2.6888464685028635e-06,
      "loss": 0.0182,
      "step": 4930
    },
    {
      "epoch": 0.08083121983146527,
      "grad_norm": 1.5141230821609497,
      "learning_rate": 2.6943005181347152e-06,
      "loss": 0.0172,
      "step": 4940
    },
    {
      "epoch": 0.08099484578254111,
      "grad_norm": 0.7351788282394409,
      "learning_rate": 2.699754567766567e-06,
      "loss": 0.0152,
      "step": 4950
    },
    {
      "epoch": 0.08115847173361695,
      "grad_norm": 0.769402801990509,
      "learning_rate": 2.7052086173984187e-06,
      "loss": 0.0197,
      "step": 4960
    },
    {
      "epoch": 0.0813220976846928,
      "grad_norm": 0.3709261119365692,
      "learning_rate": 2.71066266703027e-06,
      "loss": 0.0155,
      "step": 4970
    },
    {
      "epoch": 0.08148572363576863,
      "grad_norm": 0.8174160718917847,
      "learning_rate": 2.7161167166621216e-06,
      "loss": 0.0159,
      "step": 4980
    },
    {
      "epoch": 0.08164934958684447,
      "grad_norm": 1.1775890588760376,
      "learning_rate": 2.7215707662939733e-06,
      "loss": 0.0161,
      "step": 4990
    },
    {
      "epoch": 0.08181297553792032,
      "grad_norm": 0.9500192999839783,
      "learning_rate": 2.727024815925825e-06,
      "loss": 0.0171,
      "step": 5000
    },
    {
      "epoch": 0.08197660148899616,
      "grad_norm": 0.9321677684783936,
      "learning_rate": 2.7324788655576767e-06,
      "loss": 0.0167,
      "step": 5010
    },
    {
      "epoch": 0.08214022744007199,
      "grad_norm": 1.0055530071258545,
      "learning_rate": 2.7379329151895284e-06,
      "loss": 0.0121,
      "step": 5020
    },
    {
      "epoch": 0.08230385339114783,
      "grad_norm": 1.2070386409759521,
      "learning_rate": 2.74338696482138e-06,
      "loss": 0.0165,
      "step": 5030
    },
    {
      "epoch": 0.08246747934222368,
      "grad_norm": 0.5914405584335327,
      "learning_rate": 2.748841014453232e-06,
      "loss": 0.018,
      "step": 5040
    },
    {
      "epoch": 0.08263110529329952,
      "grad_norm": 0.6340383291244507,
      "learning_rate": 2.7542950640850835e-06,
      "loss": 0.0189,
      "step": 5050
    },
    {
      "epoch": 0.08279473124437536,
      "grad_norm": 0.9640193581581116,
      "learning_rate": 2.7597491137169352e-06,
      "loss": 0.0157,
      "step": 5060
    },
    {
      "epoch": 0.0829583571954512,
      "grad_norm": 0.8477957248687744,
      "learning_rate": 2.765203163348787e-06,
      "loss": 0.0172,
      "step": 5070
    },
    {
      "epoch": 0.08312198314652704,
      "grad_norm": 1.026456594467163,
      "learning_rate": 2.7706572129806386e-06,
      "loss": 0.0146,
      "step": 5080
    },
    {
      "epoch": 0.08328560909760288,
      "grad_norm": 0.7711597681045532,
      "learning_rate": 2.77611126261249e-06,
      "loss": 0.0173,
      "step": 5090
    },
    {
      "epoch": 0.08344923504867872,
      "grad_norm": 0.9673550128936768,
      "learning_rate": 2.7815653122443416e-06,
      "loss": 0.0142,
      "step": 5100
    },
    {
      "epoch": 0.08361286099975457,
      "grad_norm": 1.2038995027542114,
      "learning_rate": 2.7870193618761933e-06,
      "loss": 0.0156,
      "step": 5110
    },
    {
      "epoch": 0.0837764869508304,
      "grad_norm": 0.878197193145752,
      "learning_rate": 2.792473411508045e-06,
      "loss": 0.022,
      "step": 5120
    },
    {
      "epoch": 0.08394011290190624,
      "grad_norm": 1.0659819841384888,
      "learning_rate": 2.7979274611398967e-06,
      "loss": 0.0136,
      "step": 5130
    },
    {
      "epoch": 0.08410373885298209,
      "grad_norm": 1.1639721393585205,
      "learning_rate": 2.8033815107717484e-06,
      "loss": 0.0187,
      "step": 5140
    },
    {
      "epoch": 0.08426736480405793,
      "grad_norm": 0.8449233174324036,
      "learning_rate": 2.8088355604036e-06,
      "loss": 0.0138,
      "step": 5150
    },
    {
      "epoch": 0.08443099075513376,
      "grad_norm": 0.8939120173454285,
      "learning_rate": 2.814289610035452e-06,
      "loss": 0.0213,
      "step": 5160
    },
    {
      "epoch": 0.0845946167062096,
      "grad_norm": 1.4400274753570557,
      "learning_rate": 2.8197436596673035e-06,
      "loss": 0.0216,
      "step": 5170
    },
    {
      "epoch": 0.08475824265728545,
      "grad_norm": 0.8084188103675842,
      "learning_rate": 2.8251977092991552e-06,
      "loss": 0.0167,
      "step": 5180
    },
    {
      "epoch": 0.08492186860836129,
      "grad_norm": 0.7128744721412659,
      "learning_rate": 2.830651758931007e-06,
      "loss": 0.014,
      "step": 5190
    },
    {
      "epoch": 0.08508549455943713,
      "grad_norm": 0.9861514568328857,
      "learning_rate": 2.8361058085628578e-06,
      "loss": 0.0223,
      "step": 5200
    },
    {
      "epoch": 0.08524912051051296,
      "grad_norm": 0.5894780158996582,
      "learning_rate": 2.8415598581947095e-06,
      "loss": 0.0178,
      "step": 5210
    },
    {
      "epoch": 0.0854127464615888,
      "grad_norm": 0.7652517557144165,
      "learning_rate": 2.847013907826561e-06,
      "loss": 0.0223,
      "step": 5220
    },
    {
      "epoch": 0.08557637241266465,
      "grad_norm": 0.7430261373519897,
      "learning_rate": 2.852467957458413e-06,
      "loss": 0.0153,
      "step": 5230
    },
    {
      "epoch": 0.0857399983637405,
      "grad_norm": 0.831869900226593,
      "learning_rate": 2.8579220070902646e-06,
      "loss": 0.0159,
      "step": 5240
    },
    {
      "epoch": 0.08590362431481632,
      "grad_norm": 0.7985550761222839,
      "learning_rate": 2.8633760567221163e-06,
      "loss": 0.019,
      "step": 5250
    },
    {
      "epoch": 0.08606725026589217,
      "grad_norm": 0.784176766872406,
      "learning_rate": 2.868830106353968e-06,
      "loss": 0.0163,
      "step": 5260
    },
    {
      "epoch": 0.08623087621696801,
      "grad_norm": 0.9921072125434875,
      "learning_rate": 2.8742841559858197e-06,
      "loss": 0.0109,
      "step": 5270
    },
    {
      "epoch": 0.08639450216804385,
      "grad_norm": 1.004798173904419,
      "learning_rate": 2.8797382056176714e-06,
      "loss": 0.0152,
      "step": 5280
    },
    {
      "epoch": 0.0865581281191197,
      "grad_norm": 1.0165233612060547,
      "learning_rate": 2.885192255249523e-06,
      "loss": 0.016,
      "step": 5290
    },
    {
      "epoch": 0.08672175407019553,
      "grad_norm": 0.8232343792915344,
      "learning_rate": 2.8906463048813748e-06,
      "loss": 0.0169,
      "step": 5300
    },
    {
      "epoch": 0.08688538002127137,
      "grad_norm": 0.7885742783546448,
      "learning_rate": 2.8961003545132265e-06,
      "loss": 0.0132,
      "step": 5310
    },
    {
      "epoch": 0.08704900597234722,
      "grad_norm": 0.602729856967926,
      "learning_rate": 2.9015544041450778e-06,
      "loss": 0.0174,
      "step": 5320
    },
    {
      "epoch": 0.08721263192342306,
      "grad_norm": 0.6252309679985046,
      "learning_rate": 2.9070084537769295e-06,
      "loss": 0.0187,
      "step": 5330
    },
    {
      "epoch": 0.08737625787449889,
      "grad_norm": 1.046807050704956,
      "learning_rate": 2.912462503408781e-06,
      "loss": 0.0161,
      "step": 5340
    },
    {
      "epoch": 0.08753988382557473,
      "grad_norm": 0.8579831719398499,
      "learning_rate": 2.917916553040633e-06,
      "loss": 0.0141,
      "step": 5350
    },
    {
      "epoch": 0.08770350977665058,
      "grad_norm": 0.7528521418571472,
      "learning_rate": 2.9233706026724846e-06,
      "loss": 0.0127,
      "step": 5360
    },
    {
      "epoch": 0.08786713572772642,
      "grad_norm": 0.6912550330162048,
      "learning_rate": 2.9288246523043363e-06,
      "loss": 0.0153,
      "step": 5370
    },
    {
      "epoch": 0.08803076167880226,
      "grad_norm": 0.9151894450187683,
      "learning_rate": 2.934278701936188e-06,
      "loss": 0.0218,
      "step": 5380
    },
    {
      "epoch": 0.0881943876298781,
      "grad_norm": 0.7298992872238159,
      "learning_rate": 2.9397327515680397e-06,
      "loss": 0.0164,
      "step": 5390
    },
    {
      "epoch": 0.08835801358095394,
      "grad_norm": 0.8278002738952637,
      "learning_rate": 2.9451868011998914e-06,
      "loss": 0.0154,
      "step": 5400
    },
    {
      "epoch": 0.08852163953202978,
      "grad_norm": 0.8794282078742981,
      "learning_rate": 2.950640850831743e-06,
      "loss": 0.0128,
      "step": 5410
    },
    {
      "epoch": 0.08868526548310562,
      "grad_norm": 0.856492280960083,
      "learning_rate": 2.9560949004635948e-06,
      "loss": 0.0109,
      "step": 5420
    },
    {
      "epoch": 0.08884889143418147,
      "grad_norm": 0.6660822629928589,
      "learning_rate": 2.961548950095446e-06,
      "loss": 0.0127,
      "step": 5430
    },
    {
      "epoch": 0.0890125173852573,
      "grad_norm": 0.8414649367332458,
      "learning_rate": 2.9670029997272978e-06,
      "loss": 0.0137,
      "step": 5440
    },
    {
      "epoch": 0.08917614333633314,
      "grad_norm": 0.4392111301422119,
      "learning_rate": 2.9724570493591495e-06,
      "loss": 0.0108,
      "step": 5450
    },
    {
      "epoch": 0.08933976928740898,
      "grad_norm": 1.1808362007141113,
      "learning_rate": 2.977911098991001e-06,
      "loss": 0.016,
      "step": 5460
    },
    {
      "epoch": 0.08950339523848483,
      "grad_norm": 0.6742119789123535,
      "learning_rate": 2.983365148622853e-06,
      "loss": 0.0121,
      "step": 5470
    },
    {
      "epoch": 0.08966702118956066,
      "grad_norm": 0.9016894102096558,
      "learning_rate": 2.9888191982547046e-06,
      "loss": 0.019,
      "step": 5480
    },
    {
      "epoch": 0.0898306471406365,
      "grad_norm": 0.802516520023346,
      "learning_rate": 2.9942732478865563e-06,
      "loss": 0.0134,
      "step": 5490
    },
    {
      "epoch": 0.08999427309171235,
      "grad_norm": 0.5612902641296387,
      "learning_rate": 2.9997272975184075e-06,
      "loss": 0.0134,
      "step": 5500
    },
    {
      "epoch": 0.09015789904278819,
      "grad_norm": 0.5004104375839233,
      "learning_rate": 3.0051813471502592e-06,
      "loss": 0.0145,
      "step": 5510
    },
    {
      "epoch": 0.09032152499386403,
      "grad_norm": 0.9202992916107178,
      "learning_rate": 3.010635396782111e-06,
      "loss": 0.0127,
      "step": 5520
    },
    {
      "epoch": 0.09048515094493986,
      "grad_norm": 0.6290443539619446,
      "learning_rate": 3.0160894464139626e-06,
      "loss": 0.0159,
      "step": 5530
    },
    {
      "epoch": 0.0906487768960157,
      "grad_norm": 0.672428548336029,
      "learning_rate": 3.0215434960458143e-06,
      "loss": 0.014,
      "step": 5540
    },
    {
      "epoch": 0.09081240284709155,
      "grad_norm": 0.7635077834129333,
      "learning_rate": 3.0269975456776656e-06,
      "loss": 0.0182,
      "step": 5550
    },
    {
      "epoch": 0.0909760287981674,
      "grad_norm": 0.7806611657142639,
      "learning_rate": 3.0324515953095173e-06,
      "loss": 0.0147,
      "step": 5560
    },
    {
      "epoch": 0.09113965474924322,
      "grad_norm": 0.6791344285011292,
      "learning_rate": 3.037905644941369e-06,
      "loss": 0.0179,
      "step": 5570
    },
    {
      "epoch": 0.09130328070031907,
      "grad_norm": 0.9148838520050049,
      "learning_rate": 3.0433596945732207e-06,
      "loss": 0.0154,
      "step": 5580
    },
    {
      "epoch": 0.09146690665139491,
      "grad_norm": 0.36368292570114136,
      "learning_rate": 3.0488137442050724e-06,
      "loss": 0.013,
      "step": 5590
    },
    {
      "epoch": 0.09163053260247075,
      "grad_norm": 0.7524011135101318,
      "learning_rate": 3.054267793836924e-06,
      "loss": 0.0156,
      "step": 5600
    },
    {
      "epoch": 0.0917941585535466,
      "grad_norm": 0.834479033946991,
      "learning_rate": 3.059721843468776e-06,
      "loss": 0.0119,
      "step": 5610
    },
    {
      "epoch": 0.09195778450462243,
      "grad_norm": 0.921504557132721,
      "learning_rate": 3.0651758931006275e-06,
      "loss": 0.0203,
      "step": 5620
    },
    {
      "epoch": 0.09212141045569827,
      "grad_norm": 1.429835319519043,
      "learning_rate": 3.0706299427324792e-06,
      "loss": 0.0169,
      "step": 5630
    },
    {
      "epoch": 0.09228503640677412,
      "grad_norm": 0.5858962535858154,
      "learning_rate": 3.076083992364331e-06,
      "loss": 0.0206,
      "step": 5640
    },
    {
      "epoch": 0.09244866235784996,
      "grad_norm": 0.8736156225204468,
      "learning_rate": 3.0815380419961826e-06,
      "loss": 0.0129,
      "step": 5650
    },
    {
      "epoch": 0.0926122883089258,
      "grad_norm": 0.7214558124542236,
      "learning_rate": 3.086992091628034e-06,
      "loss": 0.0122,
      "step": 5660
    },
    {
      "epoch": 0.09277591426000163,
      "grad_norm": 1.244158148765564,
      "learning_rate": 3.0924461412598856e-06,
      "loss": 0.0161,
      "step": 5670
    },
    {
      "epoch": 0.09293954021107748,
      "grad_norm": 0.7762056589126587,
      "learning_rate": 3.0979001908917373e-06,
      "loss": 0.0154,
      "step": 5680
    },
    {
      "epoch": 0.09310316616215332,
      "grad_norm": 0.8118436336517334,
      "learning_rate": 3.103354240523589e-06,
      "loss": 0.0134,
      "step": 5690
    },
    {
      "epoch": 0.09326679211322916,
      "grad_norm": 0.6386100053787231,
      "learning_rate": 3.1088082901554407e-06,
      "loss": 0.0154,
      "step": 5700
    },
    {
      "epoch": 0.093430418064305,
      "grad_norm": 0.6355448961257935,
      "learning_rate": 3.1142623397872924e-06,
      "loss": 0.0234,
      "step": 5710
    },
    {
      "epoch": 0.09359404401538084,
      "grad_norm": 0.3626098334789276,
      "learning_rate": 3.119716389419144e-06,
      "loss": 0.0129,
      "step": 5720
    },
    {
      "epoch": 0.09375766996645668,
      "grad_norm": 0.6233932375907898,
      "learning_rate": 3.125170439050996e-06,
      "loss": 0.0134,
      "step": 5730
    },
    {
      "epoch": 0.09392129591753252,
      "grad_norm": 1.3197227716445923,
      "learning_rate": 3.1306244886828475e-06,
      "loss": 0.0157,
      "step": 5740
    },
    {
      "epoch": 0.09408492186860837,
      "grad_norm": 0.7796733975410461,
      "learning_rate": 3.136078538314699e-06,
      "loss": 0.0141,
      "step": 5750
    },
    {
      "epoch": 0.0942485478196842,
      "grad_norm": 0.7965986132621765,
      "learning_rate": 3.141532587946551e-06,
      "loss": 0.0184,
      "step": 5760
    },
    {
      "epoch": 0.09441217377076004,
      "grad_norm": 0.8424684405326843,
      "learning_rate": 3.1469866375784018e-06,
      "loss": 0.0147,
      "step": 5770
    },
    {
      "epoch": 0.09457579972183588,
      "grad_norm": 0.8431611657142639,
      "learning_rate": 3.1524406872102535e-06,
      "loss": 0.0157,
      "step": 5780
    },
    {
      "epoch": 0.09473942567291173,
      "grad_norm": 0.6252585053443909,
      "learning_rate": 3.157894736842105e-06,
      "loss": 0.0146,
      "step": 5790
    },
    {
      "epoch": 0.09490305162398756,
      "grad_norm": 0.7879089117050171,
      "learning_rate": 3.163348786473957e-06,
      "loss": 0.0155,
      "step": 5800
    },
    {
      "epoch": 0.0950666775750634,
      "grad_norm": 0.9320424795150757,
      "learning_rate": 3.1688028361058086e-06,
      "loss": 0.0137,
      "step": 5810
    },
    {
      "epoch": 0.09523030352613925,
      "grad_norm": 0.7367252111434937,
      "learning_rate": 3.1742568857376603e-06,
      "loss": 0.0136,
      "step": 5820
    },
    {
      "epoch": 0.09539392947721509,
      "grad_norm": 0.7247695326805115,
      "learning_rate": 3.179710935369512e-06,
      "loss": 0.0188,
      "step": 5830
    },
    {
      "epoch": 0.09555755542829093,
      "grad_norm": 0.4898044168949127,
      "learning_rate": 3.1851649850013637e-06,
      "loss": 0.0126,
      "step": 5840
    },
    {
      "epoch": 0.09572118137936676,
      "grad_norm": 0.7382405400276184,
      "learning_rate": 3.1906190346332154e-06,
      "loss": 0.0154,
      "step": 5850
    },
    {
      "epoch": 0.0958848073304426,
      "grad_norm": 0.9127718806266785,
      "learning_rate": 3.196073084265067e-06,
      "loss": 0.0159,
      "step": 5860
    },
    {
      "epoch": 0.09604843328151845,
      "grad_norm": 0.5369975566864014,
      "learning_rate": 3.2015271338969188e-06,
      "loss": 0.0175,
      "step": 5870
    },
    {
      "epoch": 0.0962120592325943,
      "grad_norm": 0.7133941054344177,
      "learning_rate": 3.2069811835287705e-06,
      "loss": 0.0093,
      "step": 5880
    },
    {
      "epoch": 0.09637568518367012,
      "grad_norm": 1.0156655311584473,
      "learning_rate": 3.2124352331606218e-06,
      "loss": 0.0127,
      "step": 5890
    },
    {
      "epoch": 0.09653931113474597,
      "grad_norm": 1.0250358581542969,
      "learning_rate": 3.2178892827924735e-06,
      "loss": 0.0164,
      "step": 5900
    },
    {
      "epoch": 0.09670293708582181,
      "grad_norm": 0.5811936855316162,
      "learning_rate": 3.223343332424325e-06,
      "loss": 0.0137,
      "step": 5910
    },
    {
      "epoch": 0.09686656303689765,
      "grad_norm": 0.9672229290008545,
      "learning_rate": 3.228797382056177e-06,
      "loss": 0.0145,
      "step": 5920
    },
    {
      "epoch": 0.0970301889879735,
      "grad_norm": 0.7883620858192444,
      "learning_rate": 3.2342514316880286e-06,
      "loss": 0.0115,
      "step": 5930
    },
    {
      "epoch": 0.09719381493904933,
      "grad_norm": 0.5272173285484314,
      "learning_rate": 3.2397054813198803e-06,
      "loss": 0.013,
      "step": 5940
    },
    {
      "epoch": 0.09735744089012517,
      "grad_norm": 0.78303462266922,
      "learning_rate": 3.245159530951732e-06,
      "loss": 0.0176,
      "step": 5950
    },
    {
      "epoch": 0.09752106684120102,
      "grad_norm": 1.213394284248352,
      "learning_rate": 3.2506135805835837e-06,
      "loss": 0.0171,
      "step": 5960
    },
    {
      "epoch": 0.09768469279227686,
      "grad_norm": 1.1699204444885254,
      "learning_rate": 3.2560676302154354e-06,
      "loss": 0.0164,
      "step": 5970
    },
    {
      "epoch": 0.0978483187433527,
      "grad_norm": 0.8411902785301208,
      "learning_rate": 3.261521679847287e-06,
      "loss": 0.0137,
      "step": 5980
    },
    {
      "epoch": 0.09801194469442853,
      "grad_norm": 0.6784468293190002,
      "learning_rate": 3.2669757294791388e-06,
      "loss": 0.0113,
      "step": 5990
    },
    {
      "epoch": 0.09817557064550438,
      "grad_norm": 1.2000222206115723,
      "learning_rate": 3.27242977911099e-06,
      "loss": 0.0151,
      "step": 6000
    },
    {
      "epoch": 0.09833919659658022,
      "grad_norm": 0.4644208252429962,
      "learning_rate": 3.2778838287428417e-06,
      "loss": 0.0173,
      "step": 6010
    },
    {
      "epoch": 0.09850282254765606,
      "grad_norm": 0.6437108516693115,
      "learning_rate": 3.2833378783746934e-06,
      "loss": 0.0141,
      "step": 6020
    },
    {
      "epoch": 0.09866644849873189,
      "grad_norm": 0.42634034156799316,
      "learning_rate": 3.288791928006545e-06,
      "loss": 0.011,
      "step": 6030
    },
    {
      "epoch": 0.09883007444980774,
      "grad_norm": 0.706085205078125,
      "learning_rate": 3.294245977638397e-06,
      "loss": 0.0144,
      "step": 6040
    },
    {
      "epoch": 0.09899370040088358,
      "grad_norm": 0.6038877367973328,
      "learning_rate": 3.2997000272702485e-06,
      "loss": 0.0117,
      "step": 6050
    },
    {
      "epoch": 0.09915732635195942,
      "grad_norm": 0.20743539929389954,
      "learning_rate": 3.3051540769021002e-06,
      "loss": 0.0113,
      "step": 6060
    },
    {
      "epoch": 0.09932095230303527,
      "grad_norm": 0.8119137287139893,
      "learning_rate": 3.310608126533952e-06,
      "loss": 0.0121,
      "step": 6070
    },
    {
      "epoch": 0.0994845782541111,
      "grad_norm": 0.8317829370498657,
      "learning_rate": 3.3160621761658036e-06,
      "loss": 0.0102,
      "step": 6080
    },
    {
      "epoch": 0.09964820420518694,
      "grad_norm": 0.8082549571990967,
      "learning_rate": 3.3215162257976553e-06,
      "loss": 0.0169,
      "step": 6090
    },
    {
      "epoch": 0.09981183015626278,
      "grad_norm": 0.8954679369926453,
      "learning_rate": 3.326970275429507e-06,
      "loss": 0.0102,
      "step": 6100
    },
    {
      "epoch": 0.09997545610733863,
      "grad_norm": 0.580042839050293,
      "learning_rate": 3.3324243250613587e-06,
      "loss": 0.0126,
      "step": 6110
    },
    {
      "epoch": 0.10013908205841446,
      "grad_norm": 1.0125596523284912,
      "learning_rate": 3.3378783746932096e-06,
      "loss": 0.0082,
      "step": 6120
    },
    {
      "epoch": 0.1003027080094903,
      "grad_norm": 0.5418968200683594,
      "learning_rate": 3.3433324243250613e-06,
      "loss": 0.0165,
      "step": 6130
    },
    {
      "epoch": 0.10046633396056615,
      "grad_norm": 0.917887270450592,
      "learning_rate": 3.348786473956913e-06,
      "loss": 0.0131,
      "step": 6140
    },
    {
      "epoch": 0.10062995991164199,
      "grad_norm": 1.1862261295318604,
      "learning_rate": 3.3542405235887647e-06,
      "loss": 0.0158,
      "step": 6150
    },
    {
      "epoch": 0.10079358586271783,
      "grad_norm": 0.7814752459526062,
      "learning_rate": 3.3596945732206164e-06,
      "loss": 0.0156,
      "step": 6160
    },
    {
      "epoch": 0.10095721181379366,
      "grad_norm": 1.0431448221206665,
      "learning_rate": 3.365148622852468e-06,
      "loss": 0.016,
      "step": 6170
    },
    {
      "epoch": 0.1011208377648695,
      "grad_norm": 1.0791957378387451,
      "learning_rate": 3.37060267248432e-06,
      "loss": 0.0157,
      "step": 6180
    },
    {
      "epoch": 0.10128446371594535,
      "grad_norm": 0.9360935091972351,
      "learning_rate": 3.3760567221161715e-06,
      "loss": 0.0162,
      "step": 6190
    },
    {
      "epoch": 0.1014480896670212,
      "grad_norm": 1.493471622467041,
      "learning_rate": 3.381510771748023e-06,
      "loss": 0.0144,
      "step": 6200
    },
    {
      "epoch": 0.10161171561809704,
      "grad_norm": 0.6363555192947388,
      "learning_rate": 3.386964821379875e-06,
      "loss": 0.0105,
      "step": 6210
    },
    {
      "epoch": 0.10177534156917287,
      "grad_norm": 0.7255340218544006,
      "learning_rate": 3.3924188710117266e-06,
      "loss": 0.0152,
      "step": 6220
    },
    {
      "epoch": 0.10193896752024871,
      "grad_norm": 1.0318663120269775,
      "learning_rate": 3.397872920643578e-06,
      "loss": 0.0143,
      "step": 6230
    },
    {
      "epoch": 0.10210259347132455,
      "grad_norm": 0.6117664575576782,
      "learning_rate": 3.4033269702754296e-06,
      "loss": 0.013,
      "step": 6240
    },
    {
      "epoch": 0.1022662194224004,
      "grad_norm": 1.5089282989501953,
      "learning_rate": 3.4087810199072813e-06,
      "loss": 0.0116,
      "step": 6250
    },
    {
      "epoch": 0.10242984537347623,
      "grad_norm": 0.6036588549613953,
      "learning_rate": 3.414235069539133e-06,
      "loss": 0.0096,
      "step": 6260
    },
    {
      "epoch": 0.10259347132455207,
      "grad_norm": 0.561668872833252,
      "learning_rate": 3.4196891191709847e-06,
      "loss": 0.0129,
      "step": 6270
    },
    {
      "epoch": 0.10275709727562791,
      "grad_norm": 0.7205589413642883,
      "learning_rate": 3.4251431688028364e-06,
      "loss": 0.0136,
      "step": 6280
    },
    {
      "epoch": 0.10292072322670376,
      "grad_norm": 0.7881771922111511,
      "learning_rate": 3.430597218434688e-06,
      "loss": 0.014,
      "step": 6290
    },
    {
      "epoch": 0.1030843491777796,
      "grad_norm": 0.5417429208755493,
      "learning_rate": 3.43605126806654e-06,
      "loss": 0.0121,
      "step": 6300
    },
    {
      "epoch": 0.10324797512885543,
      "grad_norm": 0.7294362783432007,
      "learning_rate": 3.4415053176983915e-06,
      "loss": 0.0133,
      "step": 6310
    },
    {
      "epoch": 0.10341160107993128,
      "grad_norm": 0.9415963888168335,
      "learning_rate": 3.446959367330243e-06,
      "loss": 0.0205,
      "step": 6320
    },
    {
      "epoch": 0.10357522703100712,
      "grad_norm": 0.7842119336128235,
      "learning_rate": 3.452413416962095e-06,
      "loss": 0.0168,
      "step": 6330
    },
    {
      "epoch": 0.10373885298208296,
      "grad_norm": 0.6940158605575562,
      "learning_rate": 3.4578674665939466e-06,
      "loss": 0.0156,
      "step": 6340
    },
    {
      "epoch": 0.10390247893315879,
      "grad_norm": 0.6025082468986511,
      "learning_rate": 3.463321516225798e-06,
      "loss": 0.0121,
      "step": 6350
    },
    {
      "epoch": 0.10406610488423464,
      "grad_norm": 0.3900011479854584,
      "learning_rate": 3.4687755658576496e-06,
      "loss": 0.0159,
      "step": 6360
    },
    {
      "epoch": 0.10422973083531048,
      "grad_norm": 1.0236190557479858,
      "learning_rate": 3.4742296154895013e-06,
      "loss": 0.0126,
      "step": 6370
    },
    {
      "epoch": 0.10439335678638632,
      "grad_norm": 0.6410085558891296,
      "learning_rate": 3.479683665121353e-06,
      "loss": 0.0099,
      "step": 6380
    },
    {
      "epoch": 0.10455698273746217,
      "grad_norm": 0.6382144093513489,
      "learning_rate": 3.4851377147532047e-06,
      "loss": 0.014,
      "step": 6390
    },
    {
      "epoch": 0.104720608688538,
      "grad_norm": 1.0055736303329468,
      "learning_rate": 3.4905917643850564e-06,
      "loss": 0.0145,
      "step": 6400
    },
    {
      "epoch": 0.10488423463961384,
      "grad_norm": 1.1841650009155273,
      "learning_rate": 3.496045814016908e-06,
      "loss": 0.0153,
      "step": 6410
    },
    {
      "epoch": 0.10504786059068968,
      "grad_norm": 0.5654468536376953,
      "learning_rate": 3.5014998636487598e-06,
      "loss": 0.0132,
      "step": 6420
    },
    {
      "epoch": 0.10521148654176553,
      "grad_norm": 0.9150072336196899,
      "learning_rate": 3.5069539132806115e-06,
      "loss": 0.0149,
      "step": 6430
    },
    {
      "epoch": 0.10537511249284136,
      "grad_norm": 0.4784562289714813,
      "learning_rate": 3.512407962912463e-06,
      "loss": 0.0114,
      "step": 6440
    },
    {
      "epoch": 0.1055387384439172,
      "grad_norm": 0.6295190453529358,
      "learning_rate": 3.517862012544315e-06,
      "loss": 0.0112,
      "step": 6450
    },
    {
      "epoch": 0.10570236439499305,
      "grad_norm": 0.5648689866065979,
      "learning_rate": 3.5233160621761657e-06,
      "loss": 0.0156,
      "step": 6460
    },
    {
      "epoch": 0.10586599034606889,
      "grad_norm": 1.2486478090286255,
      "learning_rate": 3.5287701118080174e-06,
      "loss": 0.0151,
      "step": 6470
    },
    {
      "epoch": 0.10602961629714473,
      "grad_norm": 0.2829054594039917,
      "learning_rate": 3.534224161439869e-06,
      "loss": 0.014,
      "step": 6480
    },
    {
      "epoch": 0.10619324224822056,
      "grad_norm": 1.5381412506103516,
      "learning_rate": 3.539678211071721e-06,
      "loss": 0.0175,
      "step": 6490
    },
    {
      "epoch": 0.1063568681992964,
      "grad_norm": 0.5836203098297119,
      "learning_rate": 3.5451322607035725e-06,
      "loss": 0.0099,
      "step": 6500
    },
    {
      "epoch": 0.10652049415037225,
      "grad_norm": 0.5512748956680298,
      "learning_rate": 3.5505863103354242e-06,
      "loss": 0.0149,
      "step": 6510
    },
    {
      "epoch": 0.1066841201014481,
      "grad_norm": 0.8949612379074097,
      "learning_rate": 3.556040359967276e-06,
      "loss": 0.0142,
      "step": 6520
    },
    {
      "epoch": 0.10684774605252394,
      "grad_norm": 0.6022013425827026,
      "learning_rate": 3.5614944095991276e-06,
      "loss": 0.013,
      "step": 6530
    },
    {
      "epoch": 0.10701137200359977,
      "grad_norm": 0.757368266582489,
      "learning_rate": 3.5669484592309793e-06,
      "loss": 0.0156,
      "step": 6540
    },
    {
      "epoch": 0.10717499795467561,
      "grad_norm": 0.646453857421875,
      "learning_rate": 3.572402508862831e-06,
      "loss": 0.0095,
      "step": 6550
    },
    {
      "epoch": 0.10733862390575145,
      "grad_norm": 1.159908413887024,
      "learning_rate": 3.5778565584946827e-06,
      "loss": 0.0106,
      "step": 6560
    },
    {
      "epoch": 0.1075022498568273,
      "grad_norm": 0.9116016030311584,
      "learning_rate": 3.583310608126534e-06,
      "loss": 0.0156,
      "step": 6570
    },
    {
      "epoch": 0.10766587580790313,
      "grad_norm": 0.705751359462738,
      "learning_rate": 3.5887646577583857e-06,
      "loss": 0.0101,
      "step": 6580
    },
    {
      "epoch": 0.10782950175897897,
      "grad_norm": 0.6236238479614258,
      "learning_rate": 3.5942187073902374e-06,
      "loss": 0.0158,
      "step": 6590
    },
    {
      "epoch": 0.10799312771005481,
      "grad_norm": 0.8098599910736084,
      "learning_rate": 3.599672757022089e-06,
      "loss": 0.0135,
      "step": 6600
    },
    {
      "epoch": 0.10815675366113066,
      "grad_norm": 1.268117070198059,
      "learning_rate": 3.605126806653941e-06,
      "loss": 0.0235,
      "step": 6610
    },
    {
      "epoch": 0.1083203796122065,
      "grad_norm": 0.9548227190971375,
      "learning_rate": 3.6105808562857925e-06,
      "loss": 0.0128,
      "step": 6620
    },
    {
      "epoch": 0.10848400556328233,
      "grad_norm": 0.6587949991226196,
      "learning_rate": 3.6160349059176442e-06,
      "loss": 0.0091,
      "step": 6630
    },
    {
      "epoch": 0.10864763151435818,
      "grad_norm": 0.785673975944519,
      "learning_rate": 3.621488955549496e-06,
      "loss": 0.0179,
      "step": 6640
    },
    {
      "epoch": 0.10881125746543402,
      "grad_norm": 0.6535679697990417,
      "learning_rate": 3.6269430051813476e-06,
      "loss": 0.009,
      "step": 6650
    },
    {
      "epoch": 0.10897488341650986,
      "grad_norm": 0.5656495094299316,
      "learning_rate": 3.6323970548131993e-06,
      "loss": 0.0128,
      "step": 6660
    },
    {
      "epoch": 0.10913850936758569,
      "grad_norm": 0.5163063406944275,
      "learning_rate": 3.637851104445051e-06,
      "loss": 0.0134,
      "step": 6670
    },
    {
      "epoch": 0.10930213531866154,
      "grad_norm": 1.133750557899475,
      "learning_rate": 3.6433051540769027e-06,
      "loss": 0.013,
      "step": 6680
    },
    {
      "epoch": 0.10946576126973738,
      "grad_norm": 1.0261653661727905,
      "learning_rate": 3.648759203708754e-06,
      "loss": 0.0123,
      "step": 6690
    },
    {
      "epoch": 0.10962938722081322,
      "grad_norm": 0.4472079575061798,
      "learning_rate": 3.6542132533406057e-06,
      "loss": 0.0173,
      "step": 6700
    },
    {
      "epoch": 0.10979301317188907,
      "grad_norm": 0.643719494342804,
      "learning_rate": 3.6596673029724574e-06,
      "loss": 0.012,
      "step": 6710
    },
    {
      "epoch": 0.1099566391229649,
      "grad_norm": 1.0469650030136108,
      "learning_rate": 3.665121352604309e-06,
      "loss": 0.0211,
      "step": 6720
    },
    {
      "epoch": 0.11012026507404074,
      "grad_norm": 0.4100765287876129,
      "learning_rate": 3.670575402236161e-06,
      "loss": 0.011,
      "step": 6730
    },
    {
      "epoch": 0.11028389102511658,
      "grad_norm": 0.8248699307441711,
      "learning_rate": 3.676029451868012e-06,
      "loss": 0.0111,
      "step": 6740
    },
    {
      "epoch": 0.11044751697619243,
      "grad_norm": 0.5361427068710327,
      "learning_rate": 3.681483501499864e-06,
      "loss": 0.0101,
      "step": 6750
    },
    {
      "epoch": 0.11061114292726827,
      "grad_norm": 0.84926837682724,
      "learning_rate": 3.6869375511317155e-06,
      "loss": 0.0142,
      "step": 6760
    },
    {
      "epoch": 0.1107747688783441,
      "grad_norm": 0.618086040019989,
      "learning_rate": 3.692391600763567e-06,
      "loss": 0.011,
      "step": 6770
    },
    {
      "epoch": 0.11093839482941995,
      "grad_norm": 0.7282034158706665,
      "learning_rate": 3.697845650395419e-06,
      "loss": 0.0126,
      "step": 6780
    },
    {
      "epoch": 0.11110202078049579,
      "grad_norm": 0.5833308696746826,
      "learning_rate": 3.7032997000272706e-06,
      "loss": 0.0086,
      "step": 6790
    },
    {
      "epoch": 0.11126564673157163,
      "grad_norm": 0.5305668115615845,
      "learning_rate": 3.708753749659122e-06,
      "loss": 0.0094,
      "step": 6800
    },
    {
      "epoch": 0.11142927268264746,
      "grad_norm": 0.6998348832130432,
      "learning_rate": 3.7142077992909736e-06,
      "loss": 0.019,
      "step": 6810
    },
    {
      "epoch": 0.1115928986337233,
      "grad_norm": 0.7521585822105408,
      "learning_rate": 3.7196618489228253e-06,
      "loss": 0.0169,
      "step": 6820
    },
    {
      "epoch": 0.11175652458479915,
      "grad_norm": 1.0378811359405518,
      "learning_rate": 3.725115898554677e-06,
      "loss": 0.0155,
      "step": 6830
    },
    {
      "epoch": 0.111920150535875,
      "grad_norm": 0.5311976671218872,
      "learning_rate": 3.7305699481865287e-06,
      "loss": 0.0151,
      "step": 6840
    },
    {
      "epoch": 0.11208377648695084,
      "grad_norm": 0.7835608720779419,
      "learning_rate": 3.7360239978183804e-06,
      "loss": 0.0134,
      "step": 6850
    },
    {
      "epoch": 0.11224740243802667,
      "grad_norm": 0.7429501414299011,
      "learning_rate": 3.741478047450232e-06,
      "loss": 0.0155,
      "step": 6860
    },
    {
      "epoch": 0.11241102838910251,
      "grad_norm": 0.788200855255127,
      "learning_rate": 3.7469320970820838e-06,
      "loss": 0.0205,
      "step": 6870
    },
    {
      "epoch": 0.11257465434017835,
      "grad_norm": 1.2643417119979858,
      "learning_rate": 3.7523861467139355e-06,
      "loss": 0.016,
      "step": 6880
    },
    {
      "epoch": 0.1127382802912542,
      "grad_norm": 0.457520455121994,
      "learning_rate": 3.757840196345787e-06,
      "loss": 0.0071,
      "step": 6890
    },
    {
      "epoch": 0.11290190624233003,
      "grad_norm": 0.2600027620792389,
      "learning_rate": 3.763294245977639e-06,
      "loss": 0.0186,
      "step": 6900
    },
    {
      "epoch": 0.11306553219340587,
      "grad_norm": 0.6887505650520325,
      "learning_rate": 3.7687482956094906e-06,
      "loss": 0.0127,
      "step": 6910
    },
    {
      "epoch": 0.11322915814448171,
      "grad_norm": 0.4825083911418915,
      "learning_rate": 3.774202345241342e-06,
      "loss": 0.01,
      "step": 6920
    },
    {
      "epoch": 0.11339278409555756,
      "grad_norm": 0.723751962184906,
      "learning_rate": 3.7796563948731936e-06,
      "loss": 0.0128,
      "step": 6930
    },
    {
      "epoch": 0.1135564100466334,
      "grad_norm": 0.7400292158126831,
      "learning_rate": 3.7851104445050453e-06,
      "loss": 0.0122,
      "step": 6940
    },
    {
      "epoch": 0.11372003599770923,
      "grad_norm": 0.8067269921302795,
      "learning_rate": 3.790564494136897e-06,
      "loss": 0.0103,
      "step": 6950
    },
    {
      "epoch": 0.11388366194878508,
      "grad_norm": 0.7380013465881348,
      "learning_rate": 3.7960185437687487e-06,
      "loss": 0.0128,
      "step": 6960
    },
    {
      "epoch": 0.11404728789986092,
      "grad_norm": 0.9683087468147278,
      "learning_rate": 3.8014725934006004e-06,
      "loss": 0.0173,
      "step": 6970
    },
    {
      "epoch": 0.11421091385093676,
      "grad_norm": 0.5647590756416321,
      "learning_rate": 3.806926643032452e-06,
      "loss": 0.0153,
      "step": 6980
    },
    {
      "epoch": 0.11437453980201259,
      "grad_norm": 0.5487265586853027,
      "learning_rate": 3.8123806926643038e-06,
      "loss": 0.013,
      "step": 6990
    },
    {
      "epoch": 0.11453816575308844,
      "grad_norm": 0.7692117094993591,
      "learning_rate": 3.817834742296155e-06,
      "loss": 0.0113,
      "step": 7000
    },
    {
      "epoch": 0.11470179170416428,
      "grad_norm": 0.7856493592262268,
      "learning_rate": 3.823288791928007e-06,
      "loss": 0.0104,
      "step": 7010
    },
    {
      "epoch": 0.11486541765524012,
      "grad_norm": 1.0421538352966309,
      "learning_rate": 3.8287428415598584e-06,
      "loss": 0.0129,
      "step": 7020
    },
    {
      "epoch": 0.11502904360631597,
      "grad_norm": 0.684248685836792,
      "learning_rate": 3.83419689119171e-06,
      "loss": 0.0143,
      "step": 7030
    },
    {
      "epoch": 0.1151926695573918,
      "grad_norm": 0.9292379021644592,
      "learning_rate": 3.839650940823562e-06,
      "loss": 0.0174,
      "step": 7040
    },
    {
      "epoch": 0.11535629550846764,
      "grad_norm": 1.2012156248092651,
      "learning_rate": 3.8451049904554135e-06,
      "loss": 0.0114,
      "step": 7050
    },
    {
      "epoch": 0.11551992145954348,
      "grad_norm": 0.4060277044773102,
      "learning_rate": 3.850559040087265e-06,
      "loss": 0.0112,
      "step": 7060
    },
    {
      "epoch": 0.11568354741061933,
      "grad_norm": 0.5306548476219177,
      "learning_rate": 3.856013089719117e-06,
      "loss": 0.0151,
      "step": 7070
    },
    {
      "epoch": 0.11584717336169517,
      "grad_norm": 0.8730077743530273,
      "learning_rate": 3.861467139350969e-06,
      "loss": 0.017,
      "step": 7080
    },
    {
      "epoch": 0.116010799312771,
      "grad_norm": 0.2362775057554245,
      "learning_rate": 3.86692118898282e-06,
      "loss": 0.0142,
      "step": 7090
    },
    {
      "epoch": 0.11617442526384684,
      "grad_norm": 1.0708410739898682,
      "learning_rate": 3.872375238614672e-06,
      "loss": 0.0127,
      "step": 7100
    },
    {
      "epoch": 0.11633805121492269,
      "grad_norm": 0.8410795331001282,
      "learning_rate": 3.877829288246524e-06,
      "loss": 0.0138,
      "step": 7110
    },
    {
      "epoch": 0.11650167716599853,
      "grad_norm": 0.7324991822242737,
      "learning_rate": 3.8832833378783755e-06,
      "loss": 0.0136,
      "step": 7120
    },
    {
      "epoch": 0.11666530311707436,
      "grad_norm": 0.8725026249885559,
      "learning_rate": 3.888737387510227e-06,
      "loss": 0.014,
      "step": 7130
    },
    {
      "epoch": 0.1168289290681502,
      "grad_norm": 0.6370802521705627,
      "learning_rate": 3.894191437142078e-06,
      "loss": 0.0095,
      "step": 7140
    },
    {
      "epoch": 0.11699255501922605,
      "grad_norm": 0.6782422065734863,
      "learning_rate": 3.89964548677393e-06,
      "loss": 0.0156,
      "step": 7150
    },
    {
      "epoch": 0.11715618097030189,
      "grad_norm": 0.8580349683761597,
      "learning_rate": 3.905099536405781e-06,
      "loss": 0.015,
      "step": 7160
    },
    {
      "epoch": 0.11731980692137774,
      "grad_norm": 1.2276225090026855,
      "learning_rate": 3.910553586037633e-06,
      "loss": 0.0122,
      "step": 7170
    },
    {
      "epoch": 0.11748343287245357,
      "grad_norm": 0.5423526167869568,
      "learning_rate": 3.916007635669485e-06,
      "loss": 0.0139,
      "step": 7180
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.8889505863189697,
      "learning_rate": 3.9214616853013365e-06,
      "loss": 0.0144,
      "step": 7190
    },
    {
      "epoch": 0.11781068477460525,
      "grad_norm": 0.7310557961463928,
      "learning_rate": 3.926915734933188e-06,
      "loss": 0.0094,
      "step": 7200
    },
    {
      "epoch": 0.1179743107256811,
      "grad_norm": 0.5209395885467529,
      "learning_rate": 3.93236978456504e-06,
      "loss": 0.0128,
      "step": 7210
    },
    {
      "epoch": 0.11813793667675693,
      "grad_norm": 1.2866625785827637,
      "learning_rate": 3.937823834196892e-06,
      "loss": 0.0123,
      "step": 7220
    },
    {
      "epoch": 0.11830156262783277,
      "grad_norm": 1.1112022399902344,
      "learning_rate": 3.943277883828743e-06,
      "loss": 0.01,
      "step": 7230
    },
    {
      "epoch": 0.11846518857890861,
      "grad_norm": 0.7369223237037659,
      "learning_rate": 3.948731933460595e-06,
      "loss": 0.0154,
      "step": 7240
    },
    {
      "epoch": 0.11862881452998446,
      "grad_norm": 0.6381825804710388,
      "learning_rate": 3.954185983092447e-06,
      "loss": 0.0065,
      "step": 7250
    },
    {
      "epoch": 0.1187924404810603,
      "grad_norm": 0.2766132652759552,
      "learning_rate": 3.9596400327242976e-06,
      "loss": 0.0161,
      "step": 7260
    },
    {
      "epoch": 0.11895606643213613,
      "grad_norm": 0.6037293076515198,
      "learning_rate": 3.965094082356149e-06,
      "loss": 0.0105,
      "step": 7270
    },
    {
      "epoch": 0.11911969238321198,
      "grad_norm": 0.34659820795059204,
      "learning_rate": 3.970548131988001e-06,
      "loss": 0.0116,
      "step": 7280
    },
    {
      "epoch": 0.11928331833428782,
      "grad_norm": 0.9719557166099548,
      "learning_rate": 3.976002181619853e-06,
      "loss": 0.0129,
      "step": 7290
    },
    {
      "epoch": 0.11944694428536366,
      "grad_norm": 0.49065473675727844,
      "learning_rate": 3.981456231251704e-06,
      "loss": 0.011,
      "step": 7300
    },
    {
      "epoch": 0.11961057023643949,
      "grad_norm": 1.0903617143630981,
      "learning_rate": 3.986910280883556e-06,
      "loss": 0.0136,
      "step": 7310
    },
    {
      "epoch": 0.11977419618751534,
      "grad_norm": 0.6037145853042603,
      "learning_rate": 3.992364330515408e-06,
      "loss": 0.0124,
      "step": 7320
    },
    {
      "epoch": 0.11993782213859118,
      "grad_norm": 0.8191437721252441,
      "learning_rate": 3.9978183801472595e-06,
      "loss": 0.0117,
      "step": 7330
    },
    {
      "epoch": 0.12010144808966702,
      "grad_norm": 0.6067626476287842,
      "learning_rate": 4.003272429779111e-06,
      "loss": 0.0113,
      "step": 7340
    },
    {
      "epoch": 0.12026507404074287,
      "grad_norm": 0.831027626991272,
      "learning_rate": 4.008726479410963e-06,
      "loss": 0.0187,
      "step": 7350
    },
    {
      "epoch": 0.1204286999918187,
      "grad_norm": 0.6959092617034912,
      "learning_rate": 4.014180529042815e-06,
      "loss": 0.0127,
      "step": 7360
    },
    {
      "epoch": 0.12059232594289454,
      "grad_norm": 1.0949978828430176,
      "learning_rate": 4.019634578674666e-06,
      "loss": 0.0153,
      "step": 7370
    },
    {
      "epoch": 0.12075595189397038,
      "grad_norm": 0.704740047454834,
      "learning_rate": 4.025088628306518e-06,
      "loss": 0.0117,
      "step": 7380
    },
    {
      "epoch": 0.12091957784504623,
      "grad_norm": 0.9571103453636169,
      "learning_rate": 4.03054267793837e-06,
      "loss": 0.0202,
      "step": 7390
    },
    {
      "epoch": 0.12108320379612207,
      "grad_norm": 0.8239849209785461,
      "learning_rate": 4.035996727570221e-06,
      "loss": 0.0088,
      "step": 7400
    },
    {
      "epoch": 0.1212468297471979,
      "grad_norm": 0.625403106212616,
      "learning_rate": 4.041450777202073e-06,
      "loss": 0.0097,
      "step": 7410
    },
    {
      "epoch": 0.12141045569827374,
      "grad_norm": 0.4190848171710968,
      "learning_rate": 4.046904826833925e-06,
      "loss": 0.016,
      "step": 7420
    },
    {
      "epoch": 0.12157408164934959,
      "grad_norm": 0.7225220203399658,
      "learning_rate": 4.0523588764657765e-06,
      "loss": 0.0133,
      "step": 7430
    },
    {
      "epoch": 0.12173770760042543,
      "grad_norm": 0.2791774868965149,
      "learning_rate": 4.057812926097628e-06,
      "loss": 0.0098,
      "step": 7440
    },
    {
      "epoch": 0.12190133355150126,
      "grad_norm": 0.30875951051712036,
      "learning_rate": 4.06326697572948e-06,
      "loss": 0.0127,
      "step": 7450
    },
    {
      "epoch": 0.1220649595025771,
      "grad_norm": 0.7031594514846802,
      "learning_rate": 4.068721025361332e-06,
      "loss": 0.0146,
      "step": 7460
    },
    {
      "epoch": 0.12222858545365295,
      "grad_norm": 0.4741123914718628,
      "learning_rate": 4.074175074993183e-06,
      "loss": 0.0083,
      "step": 7470
    },
    {
      "epoch": 0.12239221140472879,
      "grad_norm": 0.5694860219955444,
      "learning_rate": 4.079629124625035e-06,
      "loss": 0.0085,
      "step": 7480
    },
    {
      "epoch": 0.12255583735580464,
      "grad_norm": 0.5967519283294678,
      "learning_rate": 4.085083174256886e-06,
      "loss": 0.0108,
      "step": 7490
    },
    {
      "epoch": 0.12271946330688047,
      "grad_norm": 0.7985425591468811,
      "learning_rate": 4.0905372238887375e-06,
      "loss": 0.0141,
      "step": 7500
    },
    {
      "epoch": 0.12288308925795631,
      "grad_norm": 0.4691658616065979,
      "learning_rate": 4.095991273520589e-06,
      "loss": 0.0189,
      "step": 7510
    },
    {
      "epoch": 0.12304671520903215,
      "grad_norm": 1.1882740259170532,
      "learning_rate": 4.101445323152441e-06,
      "loss": 0.0125,
      "step": 7520
    },
    {
      "epoch": 0.123210341160108,
      "grad_norm": 0.85139399766922,
      "learning_rate": 4.106899372784293e-06,
      "loss": 0.0096,
      "step": 7530
    },
    {
      "epoch": 0.12337396711118383,
      "grad_norm": 0.7582197189331055,
      "learning_rate": 4.112353422416144e-06,
      "loss": 0.014,
      "step": 7540
    },
    {
      "epoch": 0.12353759306225967,
      "grad_norm": 0.5662003755569458,
      "learning_rate": 4.117807472047996e-06,
      "loss": 0.0137,
      "step": 7550
    },
    {
      "epoch": 0.12370121901333551,
      "grad_norm": 0.40338239073753357,
      "learning_rate": 4.123261521679848e-06,
      "loss": 0.0111,
      "step": 7560
    },
    {
      "epoch": 0.12386484496441136,
      "grad_norm": 0.857072651386261,
      "learning_rate": 4.1287155713116995e-06,
      "loss": 0.0119,
      "step": 7570
    },
    {
      "epoch": 0.1240284709154872,
      "grad_norm": 0.4618501663208008,
      "learning_rate": 4.134169620943551e-06,
      "loss": 0.0105,
      "step": 7580
    },
    {
      "epoch": 0.12419209686656303,
      "grad_norm": 0.8550952672958374,
      "learning_rate": 4.139623670575403e-06,
      "loss": 0.0168,
      "step": 7590
    },
    {
      "epoch": 0.12435572281763888,
      "grad_norm": 0.7462050318717957,
      "learning_rate": 4.145077720207254e-06,
      "loss": 0.0125,
      "step": 7600
    },
    {
      "epoch": 0.12451934876871472,
      "grad_norm": 0.9552479982376099,
      "learning_rate": 4.150531769839105e-06,
      "loss": 0.0119,
      "step": 7610
    },
    {
      "epoch": 0.12468297471979056,
      "grad_norm": 0.744327187538147,
      "learning_rate": 4.155985819470957e-06,
      "loss": 0.0134,
      "step": 7620
    },
    {
      "epoch": 0.1248466006708664,
      "grad_norm": 0.8605981469154358,
      "learning_rate": 4.161439869102809e-06,
      "loss": 0.012,
      "step": 7630
    },
    {
      "epoch": 0.12501022662194225,
      "grad_norm": 0.968020498752594,
      "learning_rate": 4.1668939187346605e-06,
      "loss": 0.0099,
      "step": 7640
    },
    {
      "epoch": 0.12517385257301808,
      "grad_norm": 1.0020442008972168,
      "learning_rate": 4.172347968366512e-06,
      "loss": 0.0102,
      "step": 7650
    },
    {
      "epoch": 0.1253374785240939,
      "grad_norm": 0.9526031017303467,
      "learning_rate": 4.177802017998364e-06,
      "loss": 0.0114,
      "step": 7660
    },
    {
      "epoch": 0.12550110447516977,
      "grad_norm": 0.3526127338409424,
      "learning_rate": 4.183256067630216e-06,
      "loss": 0.0106,
      "step": 7670
    },
    {
      "epoch": 0.1256647304262456,
      "grad_norm": 1.0006099939346313,
      "learning_rate": 4.188710117262067e-06,
      "loss": 0.0151,
      "step": 7680
    },
    {
      "epoch": 0.12582835637732145,
      "grad_norm": 0.6407269835472107,
      "learning_rate": 4.194164166893919e-06,
      "loss": 0.0172,
      "step": 7690
    },
    {
      "epoch": 0.12599198232839728,
      "grad_norm": 0.9559716582298279,
      "learning_rate": 4.199618216525771e-06,
      "loss": 0.0117,
      "step": 7700
    },
    {
      "epoch": 0.1261556082794731,
      "grad_norm": 0.39195576310157776,
      "learning_rate": 4.205072266157622e-06,
      "loss": 0.0085,
      "step": 7710
    },
    {
      "epoch": 0.12631923423054897,
      "grad_norm": 0.4272684156894684,
      "learning_rate": 4.210526315789474e-06,
      "loss": 0.0104,
      "step": 7720
    },
    {
      "epoch": 0.1264828601816248,
      "grad_norm": 0.30527105927467346,
      "learning_rate": 4.215980365421326e-06,
      "loss": 0.0105,
      "step": 7730
    },
    {
      "epoch": 0.12664648613270066,
      "grad_norm": 0.7783389687538147,
      "learning_rate": 4.2214344150531775e-06,
      "loss": 0.013,
      "step": 7740
    },
    {
      "epoch": 0.1268101120837765,
      "grad_norm": 1.2681180238723755,
      "learning_rate": 4.226888464685029e-06,
      "loss": 0.0116,
      "step": 7750
    },
    {
      "epoch": 0.12697373803485232,
      "grad_norm": 0.7202063798904419,
      "learning_rate": 4.232342514316881e-06,
      "loss": 0.0147,
      "step": 7760
    },
    {
      "epoch": 0.12713736398592818,
      "grad_norm": 0.59439617395401,
      "learning_rate": 4.237796563948733e-06,
      "loss": 0.0155,
      "step": 7770
    },
    {
      "epoch": 0.127300989937004,
      "grad_norm": 0.8765865564346313,
      "learning_rate": 4.243250613580584e-06,
      "loss": 0.0101,
      "step": 7780
    },
    {
      "epoch": 0.12746461588807986,
      "grad_norm": 0.6685693264007568,
      "learning_rate": 4.248704663212436e-06,
      "loss": 0.0137,
      "step": 7790
    },
    {
      "epoch": 0.1276282418391557,
      "grad_norm": 8.568276405334473,
      "learning_rate": 4.254158712844288e-06,
      "loss": 0.0159,
      "step": 7800
    },
    {
      "epoch": 0.12779186779023152,
      "grad_norm": 0.7810708284378052,
      "learning_rate": 4.2596127624761394e-06,
      "loss": 0.0156,
      "step": 7810
    },
    {
      "epoch": 0.12795549374130738,
      "grad_norm": 0.5474937558174133,
      "learning_rate": 4.265066812107991e-06,
      "loss": 0.0068,
      "step": 7820
    },
    {
      "epoch": 0.1281191196923832,
      "grad_norm": 0.6665557026863098,
      "learning_rate": 4.270520861739842e-06,
      "loss": 0.012,
      "step": 7830
    },
    {
      "epoch": 0.12828274564345904,
      "grad_norm": 1.1190690994262695,
      "learning_rate": 4.275974911371694e-06,
      "loss": 0.0127,
      "step": 7840
    },
    {
      "epoch": 0.1284463715945349,
      "grad_norm": 0.34522131085395813,
      "learning_rate": 4.281428961003545e-06,
      "loss": 0.0075,
      "step": 7850
    },
    {
      "epoch": 0.12860999754561073,
      "grad_norm": 0.5141227841377258,
      "learning_rate": 4.286883010635397e-06,
      "loss": 0.0136,
      "step": 7860
    },
    {
      "epoch": 0.12877362349668658,
      "grad_norm": 0.9033347368240356,
      "learning_rate": 4.292337060267249e-06,
      "loss": 0.0099,
      "step": 7870
    },
    {
      "epoch": 0.12893724944776241,
      "grad_norm": 0.48658838868141174,
      "learning_rate": 4.2977911098991005e-06,
      "loss": 0.0129,
      "step": 7880
    },
    {
      "epoch": 0.12910087539883824,
      "grad_norm": 0.8204610347747803,
      "learning_rate": 4.303245159530952e-06,
      "loss": 0.012,
      "step": 7890
    },
    {
      "epoch": 0.1292645013499141,
      "grad_norm": 0.642123281955719,
      "learning_rate": 4.308699209162804e-06,
      "loss": 0.013,
      "step": 7900
    },
    {
      "epoch": 0.12942812730098993,
      "grad_norm": 0.7733058333396912,
      "learning_rate": 4.314153258794656e-06,
      "loss": 0.0152,
      "step": 7910
    },
    {
      "epoch": 0.1295917532520658,
      "grad_norm": 0.7520296573638916,
      "learning_rate": 4.319607308426507e-06,
      "loss": 0.0094,
      "step": 7920
    },
    {
      "epoch": 0.12975537920314162,
      "grad_norm": 0.7927498817443848,
      "learning_rate": 4.325061358058359e-06,
      "loss": 0.0111,
      "step": 7930
    },
    {
      "epoch": 0.12991900515421745,
      "grad_norm": 0.5819029211997986,
      "learning_rate": 4.33051540769021e-06,
      "loss": 0.0094,
      "step": 7940
    },
    {
      "epoch": 0.1300826311052933,
      "grad_norm": 0.527286946773529,
      "learning_rate": 4.3359694573220615e-06,
      "loss": 0.0131,
      "step": 7950
    },
    {
      "epoch": 0.13024625705636914,
      "grad_norm": 1.093508005142212,
      "learning_rate": 4.341423506953913e-06,
      "loss": 0.0104,
      "step": 7960
    },
    {
      "epoch": 0.130409883007445,
      "grad_norm": 0.536765456199646,
      "learning_rate": 4.346877556585765e-06,
      "loss": 0.0102,
      "step": 7970
    },
    {
      "epoch": 0.13057350895852082,
      "grad_norm": 0.7686751484870911,
      "learning_rate": 4.352331606217617e-06,
      "loss": 0.0107,
      "step": 7980
    },
    {
      "epoch": 0.13073713490959665,
      "grad_norm": 0.7606210112571716,
      "learning_rate": 4.357785655849468e-06,
      "loss": 0.013,
      "step": 7990
    },
    {
      "epoch": 0.1309007608606725,
      "grad_norm": 0.720030665397644,
      "learning_rate": 4.36323970548132e-06,
      "loss": 0.0147,
      "step": 8000
    },
    {
      "epoch": 0.13106438681174834,
      "grad_norm": 1.2556663751602173,
      "learning_rate": 4.368693755113172e-06,
      "loss": 0.0121,
      "step": 8010
    },
    {
      "epoch": 0.1312280127628242,
      "grad_norm": 0.6101266145706177,
      "learning_rate": 4.3741478047450235e-06,
      "loss": 0.0091,
      "step": 8020
    },
    {
      "epoch": 0.13139163871390003,
      "grad_norm": 1.2447086572647095,
      "learning_rate": 4.379601854376875e-06,
      "loss": 0.0112,
      "step": 8030
    },
    {
      "epoch": 0.13155526466497586,
      "grad_norm": 0.2675427198410034,
      "learning_rate": 4.385055904008727e-06,
      "loss": 0.0102,
      "step": 8040
    },
    {
      "epoch": 0.13171889061605171,
      "grad_norm": 0.6429067254066467,
      "learning_rate": 4.3905099536405786e-06,
      "loss": 0.0101,
      "step": 8050
    },
    {
      "epoch": 0.13188251656712754,
      "grad_norm": 0.5738160014152527,
      "learning_rate": 4.39596400327243e-06,
      "loss": 0.0075,
      "step": 8060
    },
    {
      "epoch": 0.13204614251820337,
      "grad_norm": 0.9842446446418762,
      "learning_rate": 4.401418052904282e-06,
      "loss": 0.0105,
      "step": 8070
    },
    {
      "epoch": 0.13220976846927923,
      "grad_norm": 0.5476564168930054,
      "learning_rate": 4.406872102536134e-06,
      "loss": 0.0112,
      "step": 8080
    },
    {
      "epoch": 0.13237339442035506,
      "grad_norm": 0.4678577482700348,
      "learning_rate": 4.412326152167985e-06,
      "loss": 0.0101,
      "step": 8090
    },
    {
      "epoch": 0.13253702037143092,
      "grad_norm": 0.7975099682807922,
      "learning_rate": 4.417780201799837e-06,
      "loss": 0.0095,
      "step": 8100
    },
    {
      "epoch": 0.13270064632250675,
      "grad_norm": 0.5060747861862183,
      "learning_rate": 4.423234251431689e-06,
      "loss": 0.0094,
      "step": 8110
    },
    {
      "epoch": 0.13286427227358258,
      "grad_norm": 0.8256040811538696,
      "learning_rate": 4.4286883010635405e-06,
      "loss": 0.0093,
      "step": 8120
    },
    {
      "epoch": 0.13302789822465844,
      "grad_norm": 0.6360666751861572,
      "learning_rate": 4.434142350695392e-06,
      "loss": 0.0126,
      "step": 8130
    },
    {
      "epoch": 0.13319152417573427,
      "grad_norm": 0.9268854260444641,
      "learning_rate": 4.439596400327244e-06,
      "loss": 0.009,
      "step": 8140
    },
    {
      "epoch": 0.13335515012681012,
      "grad_norm": 0.4691714346408844,
      "learning_rate": 4.4450504499590956e-06,
      "loss": 0.0115,
      "step": 8150
    },
    {
      "epoch": 0.13351877607788595,
      "grad_norm": 0.8491403460502625,
      "learning_rate": 4.450504499590947e-06,
      "loss": 0.0156,
      "step": 8160
    },
    {
      "epoch": 0.13368240202896178,
      "grad_norm": 0.5958046317100525,
      "learning_rate": 4.455958549222798e-06,
      "loss": 0.0096,
      "step": 8170
    },
    {
      "epoch": 0.13384602798003764,
      "grad_norm": 0.8047823905944824,
      "learning_rate": 4.46141259885465e-06,
      "loss": 0.013,
      "step": 8180
    },
    {
      "epoch": 0.13400965393111347,
      "grad_norm": 0.5287350416183472,
      "learning_rate": 4.4668666484865015e-06,
      "loss": 0.0094,
      "step": 8190
    },
    {
      "epoch": 0.13417327988218933,
      "grad_norm": 0.8187251687049866,
      "learning_rate": 4.472320698118353e-06,
      "loss": 0.0171,
      "step": 8200
    },
    {
      "epoch": 0.13433690583326516,
      "grad_norm": 0.8458852171897888,
      "learning_rate": 4.477774747750205e-06,
      "loss": 0.0119,
      "step": 8210
    },
    {
      "epoch": 0.134500531784341,
      "grad_norm": 0.6979679465293884,
      "learning_rate": 4.483228797382057e-06,
      "loss": 0.0121,
      "step": 8220
    },
    {
      "epoch": 0.13466415773541685,
      "grad_norm": 0.573388397693634,
      "learning_rate": 4.488682847013908e-06,
      "loss": 0.0115,
      "step": 8230
    },
    {
      "epoch": 0.13482778368649267,
      "grad_norm": 0.6398588418960571,
      "learning_rate": 4.49413689664576e-06,
      "loss": 0.0118,
      "step": 8240
    },
    {
      "epoch": 0.1349914096375685,
      "grad_norm": 0.4842328429222107,
      "learning_rate": 4.499590946277612e-06,
      "loss": 0.0092,
      "step": 8250
    },
    {
      "epoch": 0.13515503558864436,
      "grad_norm": 0.6921097636222839,
      "learning_rate": 4.5050449959094634e-06,
      "loss": 0.01,
      "step": 8260
    },
    {
      "epoch": 0.1353186615397202,
      "grad_norm": 0.7509792447090149,
      "learning_rate": 4.510499045541315e-06,
      "loss": 0.0095,
      "step": 8270
    },
    {
      "epoch": 0.13548228749079605,
      "grad_norm": 0.7071582674980164,
      "learning_rate": 4.515953095173167e-06,
      "loss": 0.0103,
      "step": 8280
    },
    {
      "epoch": 0.13564591344187188,
      "grad_norm": 0.987937331199646,
      "learning_rate": 4.521407144805018e-06,
      "loss": 0.0126,
      "step": 8290
    },
    {
      "epoch": 0.1358095393929477,
      "grad_norm": 1.0022671222686768,
      "learning_rate": 4.526861194436869e-06,
      "loss": 0.015,
      "step": 8300
    },
    {
      "epoch": 0.13597316534402357,
      "grad_norm": 0.4881025552749634,
      "learning_rate": 4.532315244068721e-06,
      "loss": 0.0153,
      "step": 8310
    },
    {
      "epoch": 0.1361367912950994,
      "grad_norm": 0.6487968564033508,
      "learning_rate": 4.537769293700573e-06,
      "loss": 0.0116,
      "step": 8320
    },
    {
      "epoch": 0.13630041724617525,
      "grad_norm": 0.5809783935546875,
      "learning_rate": 4.5432233433324245e-06,
      "loss": 0.0104,
      "step": 8330
    },
    {
      "epoch": 0.13646404319725108,
      "grad_norm": 0.45486441254615784,
      "learning_rate": 4.548677392964276e-06,
      "loss": 0.0118,
      "step": 8340
    },
    {
      "epoch": 0.1366276691483269,
      "grad_norm": 0.5191390514373779,
      "learning_rate": 4.554131442596128e-06,
      "loss": 0.0091,
      "step": 8350
    },
    {
      "epoch": 0.13679129509940277,
      "grad_norm": 0.649986982345581,
      "learning_rate": 4.55958549222798e-06,
      "loss": 0.0079,
      "step": 8360
    },
    {
      "epoch": 0.1369549210504786,
      "grad_norm": 0.7410485744476318,
      "learning_rate": 4.565039541859831e-06,
      "loss": 0.009,
      "step": 8370
    },
    {
      "epoch": 0.13711854700155446,
      "grad_norm": 0.961081326007843,
      "learning_rate": 4.570493591491683e-06,
      "loss": 0.0122,
      "step": 8380
    },
    {
      "epoch": 0.1372821729526303,
      "grad_norm": 0.668884813785553,
      "learning_rate": 4.575947641123535e-06,
      "loss": 0.0112,
      "step": 8390
    },
    {
      "epoch": 0.13744579890370612,
      "grad_norm": 0.5774788856506348,
      "learning_rate": 4.581401690755386e-06,
      "loss": 0.0119,
      "step": 8400
    },
    {
      "epoch": 0.13760942485478198,
      "grad_norm": 0.8657992482185364,
      "learning_rate": 4.586855740387238e-06,
      "loss": 0.0094,
      "step": 8410
    },
    {
      "epoch": 0.1377730508058578,
      "grad_norm": 0.6547164916992188,
      "learning_rate": 4.59230979001909e-06,
      "loss": 0.0103,
      "step": 8420
    },
    {
      "epoch": 0.13793667675693366,
      "grad_norm": 0.783018946647644,
      "learning_rate": 4.5977638396509415e-06,
      "loss": 0.0143,
      "step": 8430
    },
    {
      "epoch": 0.1381003027080095,
      "grad_norm": 0.7820207476615906,
      "learning_rate": 4.603217889282793e-06,
      "loss": 0.0114,
      "step": 8440
    },
    {
      "epoch": 0.13826392865908532,
      "grad_norm": 0.6100033521652222,
      "learning_rate": 4.608671938914645e-06,
      "loss": 0.011,
      "step": 8450
    },
    {
      "epoch": 0.13842755461016118,
      "grad_norm": 0.4886954128742218,
      "learning_rate": 4.614125988546497e-06,
      "loss": 0.0104,
      "step": 8460
    },
    {
      "epoch": 0.138591180561237,
      "grad_norm": 0.31829285621643066,
      "learning_rate": 4.619580038178348e-06,
      "loss": 0.0137,
      "step": 8470
    },
    {
      "epoch": 0.13875480651231284,
      "grad_norm": 0.8497973084449768,
      "learning_rate": 4.6250340878102e-06,
      "loss": 0.0107,
      "step": 8480
    },
    {
      "epoch": 0.1389184324633887,
      "grad_norm": 0.42186859250068665,
      "learning_rate": 4.630488137442052e-06,
      "loss": 0.0145,
      "step": 8490
    },
    {
      "epoch": 0.13908205841446453,
      "grad_norm": 0.341957688331604,
      "learning_rate": 4.635942187073903e-06,
      "loss": 0.0091,
      "step": 8500
    },
    {
      "epoch": 0.13924568436554038,
      "grad_norm": 0.8097950220108032,
      "learning_rate": 4.641396236705755e-06,
      "loss": 0.0114,
      "step": 8510
    },
    {
      "epoch": 0.13940931031661621,
      "grad_norm": 0.5531971454620361,
      "learning_rate": 4.646850286337606e-06,
      "loss": 0.0086,
      "step": 8520
    },
    {
      "epoch": 0.13957293626769204,
      "grad_norm": 0.42774277925491333,
      "learning_rate": 4.652304335969458e-06,
      "loss": 0.0135,
      "step": 8530
    },
    {
      "epoch": 0.1397365622187679,
      "grad_norm": 0.630917489528656,
      "learning_rate": 4.657758385601309e-06,
      "loss": 0.0105,
      "step": 8540
    },
    {
      "epoch": 0.13990018816984373,
      "grad_norm": 0.35983219742774963,
      "learning_rate": 4.663212435233161e-06,
      "loss": 0.0097,
      "step": 8550
    },
    {
      "epoch": 0.1400638141209196,
      "grad_norm": 0.909866452217102,
      "learning_rate": 4.668666484865013e-06,
      "loss": 0.0092,
      "step": 8560
    },
    {
      "epoch": 0.14022744007199542,
      "grad_norm": 0.5690034031867981,
      "learning_rate": 4.6741205344968645e-06,
      "loss": 0.0121,
      "step": 8570
    },
    {
      "epoch": 0.14039106602307125,
      "grad_norm": 0.9370225071907043,
      "learning_rate": 4.679574584128716e-06,
      "loss": 0.0124,
      "step": 8580
    },
    {
      "epoch": 0.1405546919741471,
      "grad_norm": 1.0086448192596436,
      "learning_rate": 4.685028633760568e-06,
      "loss": 0.014,
      "step": 8590
    },
    {
      "epoch": 0.14071831792522294,
      "grad_norm": 0.49329251050949097,
      "learning_rate": 4.6904826833924196e-06,
      "loss": 0.013,
      "step": 8600
    },
    {
      "epoch": 0.1408819438762988,
      "grad_norm": 0.5797995924949646,
      "learning_rate": 4.695936733024271e-06,
      "loss": 0.0088,
      "step": 8610
    },
    {
      "epoch": 0.14104556982737462,
      "grad_norm": 0.498033732175827,
      "learning_rate": 4.701390782656123e-06,
      "loss": 0.0098,
      "step": 8620
    },
    {
      "epoch": 0.14120919577845045,
      "grad_norm": 0.8719580769538879,
      "learning_rate": 4.706844832287974e-06,
      "loss": 0.0127,
      "step": 8630
    },
    {
      "epoch": 0.1413728217295263,
      "grad_norm": 0.6618049740791321,
      "learning_rate": 4.7122988819198255e-06,
      "loss": 0.015,
      "step": 8640
    },
    {
      "epoch": 0.14153644768060214,
      "grad_norm": 0.41451624035835266,
      "learning_rate": 4.717752931551677e-06,
      "loss": 0.012,
      "step": 8650
    },
    {
      "epoch": 0.141700073631678,
      "grad_norm": 0.479082852602005,
      "learning_rate": 4.723206981183529e-06,
      "loss": 0.0095,
      "step": 8660
    },
    {
      "epoch": 0.14186369958275383,
      "grad_norm": 0.6567007303237915,
      "learning_rate": 4.728661030815381e-06,
      "loss": 0.0146,
      "step": 8670
    },
    {
      "epoch": 0.14202732553382966,
      "grad_norm": 1.1912174224853516,
      "learning_rate": 4.734115080447232e-06,
      "loss": 0.0105,
      "step": 8680
    },
    {
      "epoch": 0.14219095148490551,
      "grad_norm": 0.5104548931121826,
      "learning_rate": 4.739569130079084e-06,
      "loss": 0.0093,
      "step": 8690
    },
    {
      "epoch": 0.14235457743598134,
      "grad_norm": 0.5026189088821411,
      "learning_rate": 4.745023179710936e-06,
      "loss": 0.0119,
      "step": 8700
    },
    {
      "epoch": 0.14251820338705717,
      "grad_norm": 0.8047844767570496,
      "learning_rate": 4.750477229342787e-06,
      "loss": 0.0112,
      "step": 8710
    },
    {
      "epoch": 0.14268182933813303,
      "grad_norm": 0.933950662612915,
      "learning_rate": 4.755931278974639e-06,
      "loss": 0.0142,
      "step": 8720
    },
    {
      "epoch": 0.14284545528920886,
      "grad_norm": 0.31917062401771545,
      "learning_rate": 4.761385328606491e-06,
      "loss": 0.0174,
      "step": 8730
    },
    {
      "epoch": 0.14300908124028472,
      "grad_norm": 1.0437222719192505,
      "learning_rate": 4.766839378238342e-06,
      "loss": 0.0111,
      "step": 8740
    },
    {
      "epoch": 0.14317270719136055,
      "grad_norm": 0.7204347252845764,
      "learning_rate": 4.772293427870193e-06,
      "loss": 0.0136,
      "step": 8750
    },
    {
      "epoch": 0.14333633314243638,
      "grad_norm": 0.5480690598487854,
      "learning_rate": 4.777747477502045e-06,
      "loss": 0.0108,
      "step": 8760
    },
    {
      "epoch": 0.14349995909351224,
      "grad_norm": 1.1413639783859253,
      "learning_rate": 4.783201527133897e-06,
      "loss": 0.0132,
      "step": 8770
    },
    {
      "epoch": 0.14366358504458807,
      "grad_norm": 0.5628373622894287,
      "learning_rate": 4.7886555767657485e-06,
      "loss": 0.0095,
      "step": 8780
    },
    {
      "epoch": 0.14382721099566392,
      "grad_norm": 0.8018162250518799,
      "learning_rate": 4.7941096263976e-06,
      "loss": 0.0095,
      "step": 8790
    },
    {
      "epoch": 0.14399083694673975,
      "grad_norm": 0.8129953145980835,
      "learning_rate": 4.799563676029452e-06,
      "loss": 0.0104,
      "step": 8800
    },
    {
      "epoch": 0.14415446289781558,
      "grad_norm": 0.572077751159668,
      "learning_rate": 4.805017725661304e-06,
      "loss": 0.0072,
      "step": 8810
    },
    {
      "epoch": 0.14431808884889144,
      "grad_norm": 0.47636815905570984,
      "learning_rate": 4.810471775293155e-06,
      "loss": 0.0102,
      "step": 8820
    },
    {
      "epoch": 0.14448171479996727,
      "grad_norm": 0.7629075646400452,
      "learning_rate": 4.815925824925007e-06,
      "loss": 0.0113,
      "step": 8830
    },
    {
      "epoch": 0.14464534075104313,
      "grad_norm": 0.18702660501003265,
      "learning_rate": 4.821379874556859e-06,
      "loss": 0.007,
      "step": 8840
    },
    {
      "epoch": 0.14480896670211896,
      "grad_norm": 0.4570651948451996,
      "learning_rate": 4.82683392418871e-06,
      "loss": 0.0113,
      "step": 8850
    },
    {
      "epoch": 0.1449725926531948,
      "grad_norm": 0.3744865655899048,
      "learning_rate": 4.832287973820562e-06,
      "loss": 0.0122,
      "step": 8860
    },
    {
      "epoch": 0.14513621860427064,
      "grad_norm": 0.49121788144111633,
      "learning_rate": 4.837742023452414e-06,
      "loss": 0.009,
      "step": 8870
    },
    {
      "epoch": 0.14529984455534647,
      "grad_norm": 0.41438502073287964,
      "learning_rate": 4.8431960730842655e-06,
      "loss": 0.0119,
      "step": 8880
    },
    {
      "epoch": 0.14546347050642233,
      "grad_norm": 0.4890996813774109,
      "learning_rate": 4.848650122716117e-06,
      "loss": 0.0101,
      "step": 8890
    },
    {
      "epoch": 0.14562709645749816,
      "grad_norm": 0.7709494233131409,
      "learning_rate": 4.854104172347969e-06,
      "loss": 0.0091,
      "step": 8900
    },
    {
      "epoch": 0.145790722408574,
      "grad_norm": 0.5240187048912048,
      "learning_rate": 4.859558221979821e-06,
      "loss": 0.0093,
      "step": 8910
    },
    {
      "epoch": 0.14595434835964985,
      "grad_norm": 0.8679760694503784,
      "learning_rate": 4.865012271611672e-06,
      "loss": 0.0117,
      "step": 8920
    },
    {
      "epoch": 0.14611797431072568,
      "grad_norm": 0.44508063793182373,
      "learning_rate": 4.870466321243524e-06,
      "loss": 0.0114,
      "step": 8930
    },
    {
      "epoch": 0.1462816002618015,
      "grad_norm": 0.6625714898109436,
      "learning_rate": 4.875920370875376e-06,
      "loss": 0.009,
      "step": 8940
    },
    {
      "epoch": 0.14644522621287737,
      "grad_norm": 0.6329509615898132,
      "learning_rate": 4.881374420507227e-06,
      "loss": 0.0093,
      "step": 8950
    },
    {
      "epoch": 0.1466088521639532,
      "grad_norm": 0.7100282907485962,
      "learning_rate": 4.886828470139079e-06,
      "loss": 0.008,
      "step": 8960
    },
    {
      "epoch": 0.14677247811502905,
      "grad_norm": 0.37761127948760986,
      "learning_rate": 4.89228251977093e-06,
      "loss": 0.0102,
      "step": 8970
    },
    {
      "epoch": 0.14693610406610488,
      "grad_norm": 0.8917548656463623,
      "learning_rate": 4.897736569402782e-06,
      "loss": 0.01,
      "step": 8980
    },
    {
      "epoch": 0.1470997300171807,
      "grad_norm": 0.7606556415557861,
      "learning_rate": 4.903190619034633e-06,
      "loss": 0.0094,
      "step": 8990
    },
    {
      "epoch": 0.14726335596825657,
      "grad_norm": 0.4368397891521454,
      "learning_rate": 4.908644668666485e-06,
      "loss": 0.0087,
      "step": 9000
    },
    {
      "epoch": 0.1474269819193324,
      "grad_norm": 0.624626100063324,
      "learning_rate": 4.914098718298337e-06,
      "loss": 0.012,
      "step": 9010
    },
    {
      "epoch": 0.14759060787040826,
      "grad_norm": 0.3818283975124359,
      "learning_rate": 4.9195527679301885e-06,
      "loss": 0.0101,
      "step": 9020
    },
    {
      "epoch": 0.1477542338214841,
      "grad_norm": 0.7135481834411621,
      "learning_rate": 4.92500681756204e-06,
      "loss": 0.0112,
      "step": 9030
    },
    {
      "epoch": 0.14791785977255992,
      "grad_norm": 0.8082696199417114,
      "learning_rate": 4.930460867193892e-06,
      "loss": 0.0108,
      "step": 9040
    },
    {
      "epoch": 0.14808148572363578,
      "grad_norm": 0.5904209017753601,
      "learning_rate": 4.9359149168257436e-06,
      "loss": 0.0105,
      "step": 9050
    },
    {
      "epoch": 0.1482451116747116,
      "grad_norm": 0.6689426302909851,
      "learning_rate": 4.941368966457595e-06,
      "loss": 0.0102,
      "step": 9060
    },
    {
      "epoch": 0.14840873762578746,
      "grad_norm": 0.6649727821350098,
      "learning_rate": 4.946823016089447e-06,
      "loss": 0.0116,
      "step": 9070
    },
    {
      "epoch": 0.1485723635768633,
      "grad_norm": 1.2194387912750244,
      "learning_rate": 4.952277065721299e-06,
      "loss": 0.012,
      "step": 9080
    },
    {
      "epoch": 0.14873598952793912,
      "grad_norm": 0.656234622001648,
      "learning_rate": 4.9577311153531495e-06,
      "loss": 0.0072,
      "step": 9090
    },
    {
      "epoch": 0.14889961547901498,
      "grad_norm": 0.7400660514831543,
      "learning_rate": 4.963185164985001e-06,
      "loss": 0.0095,
      "step": 9100
    },
    {
      "epoch": 0.1490632414300908,
      "grad_norm": 0.7004467844963074,
      "learning_rate": 4.968639214616853e-06,
      "loss": 0.0092,
      "step": 9110
    },
    {
      "epoch": 0.14922686738116667,
      "grad_norm": 0.40032580494880676,
      "learning_rate": 4.974093264248705e-06,
      "loss": 0.0149,
      "step": 9120
    },
    {
      "epoch": 0.1493904933322425,
      "grad_norm": 0.5339474678039551,
      "learning_rate": 4.979547313880556e-06,
      "loss": 0.0125,
      "step": 9130
    },
    {
      "epoch": 0.14955411928331833,
      "grad_norm": 0.43809598684310913,
      "learning_rate": 4.985001363512408e-06,
      "loss": 0.0089,
      "step": 9140
    },
    {
      "epoch": 0.14971774523439418,
      "grad_norm": 0.753343939781189,
      "learning_rate": 4.99045541314426e-06,
      "loss": 0.0098,
      "step": 9150
    },
    {
      "epoch": 0.14988137118547,
      "grad_norm": 0.6842641830444336,
      "learning_rate": 4.995909462776111e-06,
      "loss": 0.0141,
      "step": 9160
    },
    {
      "epoch": 0.15004499713654584,
      "grad_norm": 0.339585542678833,
      "learning_rate": 5.001363512407963e-06,
      "loss": 0.0094,
      "step": 9170
    },
    {
      "epoch": 0.1502086230876217,
      "grad_norm": 0.8428782820701599,
      "learning_rate": 5.006817562039815e-06,
      "loss": 0.0106,
      "step": 9180
    },
    {
      "epoch": 0.15037224903869753,
      "grad_norm": 0.9628096222877502,
      "learning_rate": 5.0122716116716665e-06,
      "loss": 0.0098,
      "step": 9190
    },
    {
      "epoch": 0.1505358749897734,
      "grad_norm": 0.866793692111969,
      "learning_rate": 5.017725661303518e-06,
      "loss": 0.0116,
      "step": 9200
    },
    {
      "epoch": 0.15069950094084922,
      "grad_norm": 0.8987458944320679,
      "learning_rate": 5.02317971093537e-06,
      "loss": 0.0096,
      "step": 9210
    },
    {
      "epoch": 0.15086312689192505,
      "grad_norm": 0.22411911189556122,
      "learning_rate": 5.028633760567222e-06,
      "loss": 0.0065,
      "step": 9220
    },
    {
      "epoch": 0.1510267528430009,
      "grad_norm": 1.012649655342102,
      "learning_rate": 5.034087810199073e-06,
      "loss": 0.0116,
      "step": 9230
    },
    {
      "epoch": 0.15119037879407674,
      "grad_norm": 0.48535045981407166,
      "learning_rate": 5.039541859830925e-06,
      "loss": 0.008,
      "step": 9240
    },
    {
      "epoch": 0.1513540047451526,
      "grad_norm": 0.35569506883621216,
      "learning_rate": 5.044995909462777e-06,
      "loss": 0.0077,
      "step": 9250
    },
    {
      "epoch": 0.15151763069622842,
      "grad_norm": 0.43013277649879456,
      "learning_rate": 5.050449959094628e-06,
      "loss": 0.0088,
      "step": 9260
    },
    {
      "epoch": 0.15168125664730425,
      "grad_norm": 0.4324105679988861,
      "learning_rate": 5.05590400872648e-06,
      "loss": 0.0083,
      "step": 9270
    },
    {
      "epoch": 0.1518448825983801,
      "grad_norm": 0.6981995105743408,
      "learning_rate": 5.061358058358331e-06,
      "loss": 0.0071,
      "step": 9280
    },
    {
      "epoch": 0.15200850854945594,
      "grad_norm": 0.7040393948554993,
      "learning_rate": 5.0668121079901835e-06,
      "loss": 0.0086,
      "step": 9290
    },
    {
      "epoch": 0.1521721345005318,
      "grad_norm": 0.6929276585578918,
      "learning_rate": 5.072266157622034e-06,
      "loss": 0.0085,
      "step": 9300
    },
    {
      "epoch": 0.15233576045160763,
      "grad_norm": 0.8001575469970703,
      "learning_rate": 5.077720207253887e-06,
      "loss": 0.0141,
      "step": 9310
    },
    {
      "epoch": 0.15249938640268346,
      "grad_norm": 0.6251505017280579,
      "learning_rate": 5.083174256885738e-06,
      "loss": 0.0114,
      "step": 9320
    },
    {
      "epoch": 0.15266301235375931,
      "grad_norm": 0.44184187054634094,
      "learning_rate": 5.08862830651759e-06,
      "loss": 0.0083,
      "step": 9330
    },
    {
      "epoch": 0.15282663830483514,
      "grad_norm": 0.7383847236633301,
      "learning_rate": 5.094082356149441e-06,
      "loss": 0.0128,
      "step": 9340
    },
    {
      "epoch": 0.15299026425591097,
      "grad_norm": 0.8107274770736694,
      "learning_rate": 5.099536405781294e-06,
      "loss": 0.0166,
      "step": 9350
    },
    {
      "epoch": 0.15315389020698683,
      "grad_norm": 0.6202419400215149,
      "learning_rate": 5.104990455413145e-06,
      "loss": 0.0109,
      "step": 9360
    },
    {
      "epoch": 0.15331751615806266,
      "grad_norm": 0.9023040533065796,
      "learning_rate": 5.1104445050449954e-06,
      "loss": 0.0121,
      "step": 9370
    },
    {
      "epoch": 0.15348114210913852,
      "grad_norm": 0.6068389415740967,
      "learning_rate": 5.115898554676848e-06,
      "loss": 0.0104,
      "step": 9380
    },
    {
      "epoch": 0.15364476806021435,
      "grad_norm": 0.8096945285797119,
      "learning_rate": 5.121352604308699e-06,
      "loss": 0.0083,
      "step": 9390
    },
    {
      "epoch": 0.15380839401129018,
      "grad_norm": 0.597368597984314,
      "learning_rate": 5.126806653940551e-06,
      "loss": 0.0112,
      "step": 9400
    },
    {
      "epoch": 0.15397201996236604,
      "grad_norm": 0.5406811237335205,
      "learning_rate": 5.132260703572402e-06,
      "loss": 0.011,
      "step": 9410
    },
    {
      "epoch": 0.15413564591344187,
      "grad_norm": 0.6746290326118469,
      "learning_rate": 5.137714753204255e-06,
      "loss": 0.0076,
      "step": 9420
    },
    {
      "epoch": 0.15429927186451772,
      "grad_norm": 0.9298248887062073,
      "learning_rate": 5.143168802836106e-06,
      "loss": 0.0108,
      "step": 9430
    },
    {
      "epoch": 0.15446289781559355,
      "grad_norm": 0.2614902853965759,
      "learning_rate": 5.148622852467958e-06,
      "loss": 0.0121,
      "step": 9440
    },
    {
      "epoch": 0.15462652376666938,
      "grad_norm": 0.8659275770187378,
      "learning_rate": 5.154076902099809e-06,
      "loss": 0.0131,
      "step": 9450
    },
    {
      "epoch": 0.15479014971774524,
      "grad_norm": 1.1671043634414673,
      "learning_rate": 5.159530951731662e-06,
      "loss": 0.0126,
      "step": 9460
    },
    {
      "epoch": 0.15495377566882107,
      "grad_norm": 0.8133409023284912,
      "learning_rate": 5.1649850013635125e-06,
      "loss": 0.0111,
      "step": 9470
    },
    {
      "epoch": 0.15511740161989693,
      "grad_norm": 0.8114234209060669,
      "learning_rate": 5.170439050995365e-06,
      "loss": 0.012,
      "step": 9480
    },
    {
      "epoch": 0.15528102757097276,
      "grad_norm": 0.47942283749580383,
      "learning_rate": 5.175893100627216e-06,
      "loss": 0.0121,
      "step": 9490
    },
    {
      "epoch": 0.1554446535220486,
      "grad_norm": 0.749294102191925,
      "learning_rate": 5.1813471502590676e-06,
      "loss": 0.0086,
      "step": 9500
    },
    {
      "epoch": 0.15560827947312444,
      "grad_norm": 0.621303141117096,
      "learning_rate": 5.186801199890919e-06,
      "loss": 0.0094,
      "step": 9510
    },
    {
      "epoch": 0.15577190542420027,
      "grad_norm": 0.380519837141037,
      "learning_rate": 5.192255249522771e-06,
      "loss": 0.0069,
      "step": 9520
    },
    {
      "epoch": 0.15593553137527613,
      "grad_norm": 0.37362000346183777,
      "learning_rate": 5.197709299154623e-06,
      "loss": 0.0143,
      "step": 9530
    },
    {
      "epoch": 0.15609915732635196,
      "grad_norm": 0.5878266096115112,
      "learning_rate": 5.203163348786474e-06,
      "loss": 0.0094,
      "step": 9540
    },
    {
      "epoch": 0.1562627832774278,
      "grad_norm": 0.45064565539360046,
      "learning_rate": 5.208617398418326e-06,
      "loss": 0.0097,
      "step": 9550
    },
    {
      "epoch": 0.15642640922850365,
      "grad_norm": 0.3453048765659332,
      "learning_rate": 5.214071448050178e-06,
      "loss": 0.0085,
      "step": 9560
    },
    {
      "epoch": 0.15659003517957948,
      "grad_norm": 0.3360031247138977,
      "learning_rate": 5.2195254976820295e-06,
      "loss": 0.0129,
      "step": 9570
    },
    {
      "epoch": 0.1567536611306553,
      "grad_norm": 0.3435528874397278,
      "learning_rate": 5.224979547313881e-06,
      "loss": 0.0094,
      "step": 9580
    },
    {
      "epoch": 0.15691728708173117,
      "grad_norm": 0.5522723197937012,
      "learning_rate": 5.230433596945733e-06,
      "loss": 0.0115,
      "step": 9590
    },
    {
      "epoch": 0.157080913032807,
      "grad_norm": 0.6665348410606384,
      "learning_rate": 5.2358876465775846e-06,
      "loss": 0.0128,
      "step": 9600
    },
    {
      "epoch": 0.15724453898388285,
      "grad_norm": 0.6444919109344482,
      "learning_rate": 5.241341696209435e-06,
      "loss": 0.009,
      "step": 9610
    },
    {
      "epoch": 0.15740816493495868,
      "grad_norm": 0.39760738611221313,
      "learning_rate": 5.246795745841288e-06,
      "loss": 0.0081,
      "step": 9620
    },
    {
      "epoch": 0.1575717908860345,
      "grad_norm": 0.27387863397598267,
      "learning_rate": 5.252249795473139e-06,
      "loss": 0.0065,
      "step": 9630
    },
    {
      "epoch": 0.15773541683711037,
      "grad_norm": 0.7389279007911682,
      "learning_rate": 5.257703845104991e-06,
      "loss": 0.0146,
      "step": 9640
    },
    {
      "epoch": 0.1578990427881862,
      "grad_norm": 0.6964746117591858,
      "learning_rate": 5.263157894736842e-06,
      "loss": 0.01,
      "step": 9650
    },
    {
      "epoch": 0.15806266873926206,
      "grad_norm": 0.6602382659912109,
      "learning_rate": 5.268611944368695e-06,
      "loss": 0.0097,
      "step": 9660
    },
    {
      "epoch": 0.1582262946903379,
      "grad_norm": 0.6786006689071655,
      "learning_rate": 5.274065994000546e-06,
      "loss": 0.0127,
      "step": 9670
    },
    {
      "epoch": 0.15838992064141372,
      "grad_norm": 0.5183022022247314,
      "learning_rate": 5.279520043632398e-06,
      "loss": 0.0112,
      "step": 9680
    },
    {
      "epoch": 0.15855354659248957,
      "grad_norm": 0.7277314066886902,
      "learning_rate": 5.284974093264249e-06,
      "loss": 0.0131,
      "step": 9690
    },
    {
      "epoch": 0.1587171725435654,
      "grad_norm": 0.6161181330680847,
      "learning_rate": 5.2904281428961016e-06,
      "loss": 0.0096,
      "step": 9700
    },
    {
      "epoch": 0.15888079849464126,
      "grad_norm": 0.6199920773506165,
      "learning_rate": 5.2958821925279524e-06,
      "loss": 0.0091,
      "step": 9710
    },
    {
      "epoch": 0.1590444244457171,
      "grad_norm": 0.43742048740386963,
      "learning_rate": 5.301336242159803e-06,
      "loss": 0.0077,
      "step": 9720
    },
    {
      "epoch": 0.15920805039679292,
      "grad_norm": 0.8142781853675842,
      "learning_rate": 5.306790291791656e-06,
      "loss": 0.0105,
      "step": 9730
    },
    {
      "epoch": 0.15937167634786878,
      "grad_norm": 0.626279354095459,
      "learning_rate": 5.312244341423507e-06,
      "loss": 0.0071,
      "step": 9740
    },
    {
      "epoch": 0.1595353022989446,
      "grad_norm": 0.7502890229225159,
      "learning_rate": 5.317698391055359e-06,
      "loss": 0.0095,
      "step": 9750
    },
    {
      "epoch": 0.15969892825002047,
      "grad_norm": 0.2243766188621521,
      "learning_rate": 5.32315244068721e-06,
      "loss": 0.0075,
      "step": 9760
    },
    {
      "epoch": 0.1598625542010963,
      "grad_norm": 0.15477269887924194,
      "learning_rate": 5.328606490319063e-06,
      "loss": 0.0097,
      "step": 9770
    },
    {
      "epoch": 0.16002618015217213,
      "grad_norm": 0.32237935066223145,
      "learning_rate": 5.3340605399509135e-06,
      "loss": 0.01,
      "step": 9780
    },
    {
      "epoch": 0.16018980610324798,
      "grad_norm": 0.8506962060928345,
      "learning_rate": 5.339514589582766e-06,
      "loss": 0.0109,
      "step": 9790
    },
    {
      "epoch": 0.1603534320543238,
      "grad_norm": 0.5185884237289429,
      "learning_rate": 5.344968639214617e-06,
      "loss": 0.0119,
      "step": 9800
    },
    {
      "epoch": 0.16051705800539964,
      "grad_norm": 0.529741108417511,
      "learning_rate": 5.3504226888464694e-06,
      "loss": 0.0117,
      "step": 9810
    },
    {
      "epoch": 0.1606806839564755,
      "grad_norm": 0.37562671303749084,
      "learning_rate": 5.35587673847832e-06,
      "loss": 0.009,
      "step": 9820
    },
    {
      "epoch": 0.16084430990755133,
      "grad_norm": 0.9291324019432068,
      "learning_rate": 5.361330788110172e-06,
      "loss": 0.0118,
      "step": 9830
    },
    {
      "epoch": 0.1610079358586272,
      "grad_norm": 0.5216712951660156,
      "learning_rate": 5.366784837742024e-06,
      "loss": 0.01,
      "step": 9840
    },
    {
      "epoch": 0.16117156180970302,
      "grad_norm": 0.486609548330307,
      "learning_rate": 5.372238887373875e-06,
      "loss": 0.0139,
      "step": 9850
    },
    {
      "epoch": 0.16133518776077885,
      "grad_norm": 0.4292732775211334,
      "learning_rate": 5.377692937005727e-06,
      "loss": 0.012,
      "step": 9860
    },
    {
      "epoch": 0.1614988137118547,
      "grad_norm": 0.8930361270904541,
      "learning_rate": 5.383146986637579e-06,
      "loss": 0.0123,
      "step": 9870
    },
    {
      "epoch": 0.16166243966293053,
      "grad_norm": 0.5616849064826965,
      "learning_rate": 5.3886010362694305e-06,
      "loss": 0.0103,
      "step": 9880
    },
    {
      "epoch": 0.1618260656140064,
      "grad_norm": 0.6051303148269653,
      "learning_rate": 5.394055085901282e-06,
      "loss": 0.0058,
      "step": 9890
    },
    {
      "epoch": 0.16198969156508222,
      "grad_norm": 0.7344470024108887,
      "learning_rate": 5.399509135533134e-06,
      "loss": 0.0121,
      "step": 9900
    },
    {
      "epoch": 0.16215331751615805,
      "grad_norm": 0.45683553814888,
      "learning_rate": 5.404963185164986e-06,
      "loss": 0.0111,
      "step": 9910
    },
    {
      "epoch": 0.1623169434672339,
      "grad_norm": 0.23688705265522003,
      "learning_rate": 5.410417234796837e-06,
      "loss": 0.0093,
      "step": 9920
    },
    {
      "epoch": 0.16248056941830974,
      "grad_norm": 0.5689866542816162,
      "learning_rate": 5.415871284428689e-06,
      "loss": 0.0101,
      "step": 9930
    },
    {
      "epoch": 0.1626441953693856,
      "grad_norm": 0.4035464823246002,
      "learning_rate": 5.42132533406054e-06,
      "loss": 0.0113,
      "step": 9940
    },
    {
      "epoch": 0.16280782132046143,
      "grad_norm": 0.646021842956543,
      "learning_rate": 5.426779383692392e-06,
      "loss": 0.0117,
      "step": 9950
    },
    {
      "epoch": 0.16297144727153726,
      "grad_norm": 0.1906503587961197,
      "learning_rate": 5.432233433324243e-06,
      "loss": 0.0057,
      "step": 9960
    },
    {
      "epoch": 0.16313507322261311,
      "grad_norm": 0.5345288515090942,
      "learning_rate": 5.437687482956096e-06,
      "loss": 0.0096,
      "step": 9970
    },
    {
      "epoch": 0.16329869917368894,
      "grad_norm": 0.5528846979141235,
      "learning_rate": 5.443141532587947e-06,
      "loss": 0.0085,
      "step": 9980
    },
    {
      "epoch": 0.1634623251247648,
      "grad_norm": 0.4603172242641449,
      "learning_rate": 5.448595582219799e-06,
      "loss": 0.0115,
      "step": 9990
    },
    {
      "epoch": 0.16362595107584063,
      "grad_norm": 0.5774797797203064,
      "learning_rate": 5.45404963185165e-06,
      "loss": 0.0081,
      "step": 10000
    },
    {
      "epoch": 0.16378957702691646,
      "grad_norm": 0.5093913078308105,
      "learning_rate": 5.459503681483503e-06,
      "loss": 0.0123,
      "step": 10010
    },
    {
      "epoch": 0.16395320297799232,
      "grad_norm": 0.32274413108825684,
      "learning_rate": 5.4649577311153535e-06,
      "loss": 0.0115,
      "step": 10020
    },
    {
      "epoch": 0.16411682892906815,
      "grad_norm": 0.6581262350082397,
      "learning_rate": 5.470411780747206e-06,
      "loss": 0.0125,
      "step": 10030
    },
    {
      "epoch": 0.16428045488014398,
      "grad_norm": 0.7366822957992554,
      "learning_rate": 5.475865830379057e-06,
      "loss": 0.0106,
      "step": 10040
    },
    {
      "epoch": 0.16444408083121984,
      "grad_norm": 0.5786474943161011,
      "learning_rate": 5.481319880010909e-06,
      "loss": 0.0091,
      "step": 10050
    },
    {
      "epoch": 0.16460770678229567,
      "grad_norm": 0.5384429097175598,
      "learning_rate": 5.48677392964276e-06,
      "loss": 0.0088,
      "step": 10060
    },
    {
      "epoch": 0.16477133273337152,
      "grad_norm": 0.8376790285110474,
      "learning_rate": 5.492227979274611e-06,
      "loss": 0.0113,
      "step": 10070
    },
    {
      "epoch": 0.16493495868444735,
      "grad_norm": 0.5172038078308105,
      "learning_rate": 5.497682028906464e-06,
      "loss": 0.0105,
      "step": 10080
    },
    {
      "epoch": 0.16509858463552318,
      "grad_norm": 0.5650915503501892,
      "learning_rate": 5.5031360785383145e-06,
      "loss": 0.0094,
      "step": 10090
    },
    {
      "epoch": 0.16526221058659904,
      "grad_norm": 0.7378827333450317,
      "learning_rate": 5.508590128170167e-06,
      "loss": 0.0109,
      "step": 10100
    },
    {
      "epoch": 0.16542583653767487,
      "grad_norm": 0.8093585968017578,
      "learning_rate": 5.514044177802018e-06,
      "loss": 0.0126,
      "step": 10110
    },
    {
      "epoch": 0.16558946248875073,
      "grad_norm": 0.5197385549545288,
      "learning_rate": 5.5194982274338705e-06,
      "loss": 0.0092,
      "step": 10120
    },
    {
      "epoch": 0.16575308843982656,
      "grad_norm": 0.632228672504425,
      "learning_rate": 5.524952277065721e-06,
      "loss": 0.011,
      "step": 10130
    },
    {
      "epoch": 0.1659167143909024,
      "grad_norm": 0.5733054876327515,
      "learning_rate": 5.530406326697574e-06,
      "loss": 0.0092,
      "step": 10140
    },
    {
      "epoch": 0.16608034034197824,
      "grad_norm": 0.5369895100593567,
      "learning_rate": 5.535860376329425e-06,
      "loss": 0.012,
      "step": 10150
    },
    {
      "epoch": 0.16624396629305407,
      "grad_norm": 0.6807331442832947,
      "learning_rate": 5.541314425961277e-06,
      "loss": 0.0105,
      "step": 10160
    },
    {
      "epoch": 0.16640759224412993,
      "grad_norm": 0.3529227375984192,
      "learning_rate": 5.546768475593128e-06,
      "loss": 0.0093,
      "step": 10170
    },
    {
      "epoch": 0.16657121819520576,
      "grad_norm": 0.6981804966926575,
      "learning_rate": 5.55222252522498e-06,
      "loss": 0.0135,
      "step": 10180
    },
    {
      "epoch": 0.1667348441462816,
      "grad_norm": 0.9030161499977112,
      "learning_rate": 5.5576765748568315e-06,
      "loss": 0.0063,
      "step": 10190
    },
    {
      "epoch": 0.16689847009735745,
      "grad_norm": 0.3628447651863098,
      "learning_rate": 5.563130624488683e-06,
      "loss": 0.0079,
      "step": 10200
    },
    {
      "epoch": 0.16706209604843328,
      "grad_norm": 0.8995673060417175,
      "learning_rate": 5.568584674120535e-06,
      "loss": 0.0096,
      "step": 10210
    },
    {
      "epoch": 0.16722572199950914,
      "grad_norm": 0.6258118748664856,
      "learning_rate": 5.574038723752387e-06,
      "loss": 0.0083,
      "step": 10220
    },
    {
      "epoch": 0.16738934795058497,
      "grad_norm": 0.5971655249595642,
      "learning_rate": 5.579492773384238e-06,
      "loss": 0.0095,
      "step": 10230
    },
    {
      "epoch": 0.1675529739016608,
      "grad_norm": 0.9196944832801819,
      "learning_rate": 5.58494682301609e-06,
      "loss": 0.0127,
      "step": 10240
    },
    {
      "epoch": 0.16771659985273665,
      "grad_norm": 0.30177071690559387,
      "learning_rate": 5.590400872647942e-06,
      "loss": 0.0091,
      "step": 10250
    },
    {
      "epoch": 0.16788022580381248,
      "grad_norm": 0.6043103933334351,
      "learning_rate": 5.5958549222797934e-06,
      "loss": 0.0099,
      "step": 10260
    },
    {
      "epoch": 0.1680438517548883,
      "grad_norm": 0.6831212639808655,
      "learning_rate": 5.601308971911645e-06,
      "loss": 0.0133,
      "step": 10270
    },
    {
      "epoch": 0.16820747770596417,
      "grad_norm": 0.5136745572090149,
      "learning_rate": 5.606763021543497e-06,
      "loss": 0.0124,
      "step": 10280
    },
    {
      "epoch": 0.16837110365704,
      "grad_norm": 0.4598332941532135,
      "learning_rate": 5.612217071175348e-06,
      "loss": 0.0097,
      "step": 10290
    },
    {
      "epoch": 0.16853472960811586,
      "grad_norm": 0.5498688817024231,
      "learning_rate": 5.6176711208072e-06,
      "loss": 0.0123,
      "step": 10300
    },
    {
      "epoch": 0.1686983555591917,
      "grad_norm": 0.48969921469688416,
      "learning_rate": 5.623125170439051e-06,
      "loss": 0.0102,
      "step": 10310
    },
    {
      "epoch": 0.16886198151026752,
      "grad_norm": 0.18896116316318512,
      "learning_rate": 5.628579220070904e-06,
      "loss": 0.0102,
      "step": 10320
    },
    {
      "epoch": 0.16902560746134337,
      "grad_norm": 0.3152559995651245,
      "learning_rate": 5.6340332697027545e-06,
      "loss": 0.0115,
      "step": 10330
    },
    {
      "epoch": 0.1691892334124192,
      "grad_norm": 0.7210392951965332,
      "learning_rate": 5.639487319334607e-06,
      "loss": 0.009,
      "step": 10340
    },
    {
      "epoch": 0.16935285936349506,
      "grad_norm": 0.9130593538284302,
      "learning_rate": 5.644941368966458e-06,
      "loss": 0.0126,
      "step": 10350
    },
    {
      "epoch": 0.1695164853145709,
      "grad_norm": 0.5895740985870361,
      "learning_rate": 5.6503954185983104e-06,
      "loss": 0.0104,
      "step": 10360
    },
    {
      "epoch": 0.16968011126564672,
      "grad_norm": 0.3843470513820648,
      "learning_rate": 5.655849468230161e-06,
      "loss": 0.0079,
      "step": 10370
    },
    {
      "epoch": 0.16984373721672258,
      "grad_norm": 0.2323817014694214,
      "learning_rate": 5.661303517862014e-06,
      "loss": 0.0088,
      "step": 10380
    },
    {
      "epoch": 0.1700073631677984,
      "grad_norm": 0.3774637281894684,
      "learning_rate": 5.666757567493865e-06,
      "loss": 0.016,
      "step": 10390
    },
    {
      "epoch": 0.17017098911887427,
      "grad_norm": 0.40302830934524536,
      "learning_rate": 5.6722116171257156e-06,
      "loss": 0.009,
      "step": 10400
    },
    {
      "epoch": 0.1703346150699501,
      "grad_norm": 0.36535724997520447,
      "learning_rate": 5.677665666757568e-06,
      "loss": 0.009,
      "step": 10410
    },
    {
      "epoch": 0.17049824102102593,
      "grad_norm": 0.3651065230369568,
      "learning_rate": 5.683119716389419e-06,
      "loss": 0.0076,
      "step": 10420
    },
    {
      "epoch": 0.17066186697210178,
      "grad_norm": 0.7712186574935913,
      "learning_rate": 5.6885737660212715e-06,
      "loss": 0.0113,
      "step": 10430
    },
    {
      "epoch": 0.1708254929231776,
      "grad_norm": 0.6908636093139648,
      "learning_rate": 5.694027815653122e-06,
      "loss": 0.007,
      "step": 10440
    },
    {
      "epoch": 0.17098911887425344,
      "grad_norm": 0.4428759515285492,
      "learning_rate": 5.699481865284975e-06,
      "loss": 0.008,
      "step": 10450
    },
    {
      "epoch": 0.1711527448253293,
      "grad_norm": 0.4941957890987396,
      "learning_rate": 5.704935914916826e-06,
      "loss": 0.011,
      "step": 10460
    },
    {
      "epoch": 0.17131637077640513,
      "grad_norm": 0.5548277497291565,
      "learning_rate": 5.710389964548678e-06,
      "loss": 0.0105,
      "step": 10470
    },
    {
      "epoch": 0.171479996727481,
      "grad_norm": 0.4565807282924652,
      "learning_rate": 5.715844014180529e-06,
      "loss": 0.0128,
      "step": 10480
    },
    {
      "epoch": 0.17164362267855682,
      "grad_norm": 0.5146138668060303,
      "learning_rate": 5.721298063812382e-06,
      "loss": 0.0148,
      "step": 10490
    },
    {
      "epoch": 0.17180724862963265,
      "grad_norm": 0.6249204874038696,
      "learning_rate": 5.7267521134442326e-06,
      "loss": 0.0073,
      "step": 10500
    },
    {
      "epoch": 0.1719708745807085,
      "grad_norm": 0.49384769797325134,
      "learning_rate": 5.732206163076084e-06,
      "loss": 0.0116,
      "step": 10510
    },
    {
      "epoch": 0.17213450053178433,
      "grad_norm": 1.056188702583313,
      "learning_rate": 5.737660212707936e-06,
      "loss": 0.0113,
      "step": 10520
    },
    {
      "epoch": 0.1722981264828602,
      "grad_norm": 0.5830293893814087,
      "learning_rate": 5.743114262339788e-06,
      "loss": 0.0095,
      "step": 10530
    },
    {
      "epoch": 0.17246175243393602,
      "grad_norm": 0.8748631477355957,
      "learning_rate": 5.748568311971639e-06,
      "loss": 0.0118,
      "step": 10540
    },
    {
      "epoch": 0.17262537838501185,
      "grad_norm": 0.6459181904792786,
      "learning_rate": 5.754022361603491e-06,
      "loss": 0.0104,
      "step": 10550
    },
    {
      "epoch": 0.1727890043360877,
      "grad_norm": 0.28606754541397095,
      "learning_rate": 5.759476411235343e-06,
      "loss": 0.0111,
      "step": 10560
    },
    {
      "epoch": 0.17295263028716354,
      "grad_norm": 0.1575181782245636,
      "learning_rate": 5.7649304608671945e-06,
      "loss": 0.0118,
      "step": 10570
    },
    {
      "epoch": 0.1731162562382394,
      "grad_norm": 0.42970117926597595,
      "learning_rate": 5.770384510499046e-06,
      "loss": 0.0094,
      "step": 10580
    },
    {
      "epoch": 0.17327988218931523,
      "grad_norm": 0.26632261276245117,
      "learning_rate": 5.775838560130898e-06,
      "loss": 0.0097,
      "step": 10590
    },
    {
      "epoch": 0.17344350814039106,
      "grad_norm": 0.6736506819725037,
      "learning_rate": 5.7812926097627496e-06,
      "loss": 0.0122,
      "step": 10600
    },
    {
      "epoch": 0.1736071340914669,
      "grad_norm": 0.5773181915283203,
      "learning_rate": 5.786746659394601e-06,
      "loss": 0.0059,
      "step": 10610
    },
    {
      "epoch": 0.17377076004254274,
      "grad_norm": 0.6499444246292114,
      "learning_rate": 5.792200709026453e-06,
      "loss": 0.0096,
      "step": 10620
    },
    {
      "epoch": 0.1739343859936186,
      "grad_norm": 0.7350925207138062,
      "learning_rate": 5.797654758658305e-06,
      "loss": 0.0073,
      "step": 10630
    },
    {
      "epoch": 0.17409801194469443,
      "grad_norm": 0.5747483372688293,
      "learning_rate": 5.8031088082901555e-06,
      "loss": 0.009,
      "step": 10640
    },
    {
      "epoch": 0.17426163789577026,
      "grad_norm": 0.6138623356819153,
      "learning_rate": 5.808562857922008e-06,
      "loss": 0.0073,
      "step": 10650
    },
    {
      "epoch": 0.17442526384684612,
      "grad_norm": 0.5799426436424255,
      "learning_rate": 5.814016907553859e-06,
      "loss": 0.0072,
      "step": 10660
    },
    {
      "epoch": 0.17458888979792195,
      "grad_norm": 0.5614217519760132,
      "learning_rate": 5.8194709571857115e-06,
      "loss": 0.0074,
      "step": 10670
    },
    {
      "epoch": 0.17475251574899778,
      "grad_norm": 0.5787138938903809,
      "learning_rate": 5.824925006817562e-06,
      "loss": 0.0138,
      "step": 10680
    },
    {
      "epoch": 0.17491614170007364,
      "grad_norm": 0.7112692594528198,
      "learning_rate": 5.830379056449415e-06,
      "loss": 0.007,
      "step": 10690
    },
    {
      "epoch": 0.17507976765114946,
      "grad_norm": 0.3200352191925049,
      "learning_rate": 5.835833106081266e-06,
      "loss": 0.0095,
      "step": 10700
    },
    {
      "epoch": 0.17524339360222532,
      "grad_norm": 0.5317075848579407,
      "learning_rate": 5.841287155713118e-06,
      "loss": 0.0084,
      "step": 10710
    },
    {
      "epoch": 0.17540701955330115,
      "grad_norm": 0.7819271087646484,
      "learning_rate": 5.846741205344969e-06,
      "loss": 0.0108,
      "step": 10720
    },
    {
      "epoch": 0.17557064550437698,
      "grad_norm": 0.36027494072914124,
      "learning_rate": 5.852195254976822e-06,
      "loss": 0.0091,
      "step": 10730
    },
    {
      "epoch": 0.17573427145545284,
      "grad_norm": 0.8856569528579712,
      "learning_rate": 5.8576493046086725e-06,
      "loss": 0.0087,
      "step": 10740
    },
    {
      "epoch": 0.17589789740652867,
      "grad_norm": 0.3918082118034363,
      "learning_rate": 5.863103354240523e-06,
      "loss": 0.0055,
      "step": 10750
    },
    {
      "epoch": 0.17606152335760453,
      "grad_norm": 0.6352339386940002,
      "learning_rate": 5.868557403872376e-06,
      "loss": 0.0108,
      "step": 10760
    },
    {
      "epoch": 0.17622514930868036,
      "grad_norm": 0.5247442722320557,
      "learning_rate": 5.874011453504227e-06,
      "loss": 0.009,
      "step": 10770
    },
    {
      "epoch": 0.1763887752597562,
      "grad_norm": 0.7295508980751038,
      "learning_rate": 5.879465503136079e-06,
      "loss": 0.0087,
      "step": 10780
    },
    {
      "epoch": 0.17655240121083204,
      "grad_norm": 0.5269325971603394,
      "learning_rate": 5.88491955276793e-06,
      "loss": 0.0069,
      "step": 10790
    },
    {
      "epoch": 0.17671602716190787,
      "grad_norm": 0.11139679700136185,
      "learning_rate": 5.890373602399783e-06,
      "loss": 0.0075,
      "step": 10800
    },
    {
      "epoch": 0.17687965311298373,
      "grad_norm": 0.4286610186100006,
      "learning_rate": 5.895827652031634e-06,
      "loss": 0.0076,
      "step": 10810
    },
    {
      "epoch": 0.17704327906405956,
      "grad_norm": 0.15217582881450653,
      "learning_rate": 5.901281701663486e-06,
      "loss": 0.0101,
      "step": 10820
    },
    {
      "epoch": 0.1772069050151354,
      "grad_norm": 0.13957741856575012,
      "learning_rate": 5.906735751295337e-06,
      "loss": 0.0072,
      "step": 10830
    },
    {
      "epoch": 0.17737053096621125,
      "grad_norm": 0.24470998346805573,
      "learning_rate": 5.9121898009271895e-06,
      "loss": 0.0058,
      "step": 10840
    },
    {
      "epoch": 0.17753415691728708,
      "grad_norm": 0.4903556704521179,
      "learning_rate": 5.91764385055904e-06,
      "loss": 0.0101,
      "step": 10850
    },
    {
      "epoch": 0.17769778286836294,
      "grad_norm": 0.5044123530387878,
      "learning_rate": 5.923097900190892e-06,
      "loss": 0.0077,
      "step": 10860
    },
    {
      "epoch": 0.17786140881943877,
      "grad_norm": 0.6955841779708862,
      "learning_rate": 5.928551949822744e-06,
      "loss": 0.0092,
      "step": 10870
    },
    {
      "epoch": 0.1780250347705146,
      "grad_norm": 0.6097860932350159,
      "learning_rate": 5.9340059994545955e-06,
      "loss": 0.0083,
      "step": 10880
    },
    {
      "epoch": 0.17818866072159045,
      "grad_norm": 0.6501186490058899,
      "learning_rate": 5.939460049086447e-06,
      "loss": 0.0073,
      "step": 10890
    },
    {
      "epoch": 0.17835228667266628,
      "grad_norm": 0.7196681499481201,
      "learning_rate": 5.944914098718299e-06,
      "loss": 0.0091,
      "step": 10900
    },
    {
      "epoch": 0.1785159126237421,
      "grad_norm": 0.8038938641548157,
      "learning_rate": 5.950368148350151e-06,
      "loss": 0.0089,
      "step": 10910
    },
    {
      "epoch": 0.17867953857481797,
      "grad_norm": 0.4989466369152069,
      "learning_rate": 5.955822197982002e-06,
      "loss": 0.0074,
      "step": 10920
    },
    {
      "epoch": 0.1788431645258938,
      "grad_norm": 0.4776281416416168,
      "learning_rate": 5.961276247613854e-06,
      "loss": 0.0103,
      "step": 10930
    },
    {
      "epoch": 0.17900679047696966,
      "grad_norm": 0.4999660849571228,
      "learning_rate": 5.966730297245706e-06,
      "loss": 0.0077,
      "step": 10940
    },
    {
      "epoch": 0.1791704164280455,
      "grad_norm": 0.6260465383529663,
      "learning_rate": 5.972184346877557e-06,
      "loss": 0.0111,
      "step": 10950
    },
    {
      "epoch": 0.17933404237912132,
      "grad_norm": 0.5149052143096924,
      "learning_rate": 5.977638396509409e-06,
      "loss": 0.0106,
      "step": 10960
    },
    {
      "epoch": 0.17949766833019717,
      "grad_norm": 0.648848831653595,
      "learning_rate": 5.98309244614126e-06,
      "loss": 0.0106,
      "step": 10970
    },
    {
      "epoch": 0.179661294281273,
      "grad_norm": 0.45320019125938416,
      "learning_rate": 5.9885464957731125e-06,
      "loss": 0.0078,
      "step": 10980
    },
    {
      "epoch": 0.17982492023234886,
      "grad_norm": 0.44847699999809265,
      "learning_rate": 5.994000545404963e-06,
      "loss": 0.0097,
      "step": 10990
    },
    {
      "epoch": 0.1799885461834247,
      "grad_norm": 0.5372036099433899,
      "learning_rate": 5.999454595036815e-06,
      "loss": 0.01,
      "step": 11000
    },
    {
      "epoch": 0.18015217213450052,
      "grad_norm": 0.516180694103241,
      "learning_rate": 6.004908644668667e-06,
      "loss": 0.0099,
      "step": 11010
    },
    {
      "epoch": 0.18031579808557638,
      "grad_norm": 0.30993184447288513,
      "learning_rate": 6.0103626943005185e-06,
      "loss": 0.0112,
      "step": 11020
    },
    {
      "epoch": 0.1804794240366522,
      "grad_norm": 0.33052492141723633,
      "learning_rate": 6.01581674393237e-06,
      "loss": 0.0115,
      "step": 11030
    },
    {
      "epoch": 0.18064304998772807,
      "grad_norm": 0.7691507935523987,
      "learning_rate": 6.021270793564222e-06,
      "loss": 0.0103,
      "step": 11040
    },
    {
      "epoch": 0.1808066759388039,
      "grad_norm": 0.48284903168678284,
      "learning_rate": 6.0267248431960736e-06,
      "loss": 0.0075,
      "step": 11050
    },
    {
      "epoch": 0.18097030188987973,
      "grad_norm": 0.6617900729179382,
      "learning_rate": 6.032178892827925e-06,
      "loss": 0.0116,
      "step": 11060
    },
    {
      "epoch": 0.18113392784095558,
      "grad_norm": 0.3081563115119934,
      "learning_rate": 6.037632942459777e-06,
      "loss": 0.0107,
      "step": 11070
    },
    {
      "epoch": 0.1812975537920314,
      "grad_norm": 0.4150715470314026,
      "learning_rate": 6.043086992091629e-06,
      "loss": 0.0132,
      "step": 11080
    },
    {
      "epoch": 0.18146117974310727,
      "grad_norm": 0.34169572591781616,
      "learning_rate": 6.04854104172348e-06,
      "loss": 0.0101,
      "step": 11090
    },
    {
      "epoch": 0.1816248056941831,
      "grad_norm": 0.34667065739631653,
      "learning_rate": 6.053995091355331e-06,
      "loss": 0.0056,
      "step": 11100
    },
    {
      "epoch": 0.18178843164525893,
      "grad_norm": 0.749790608882904,
      "learning_rate": 6.059449140987184e-06,
      "loss": 0.009,
      "step": 11110
    },
    {
      "epoch": 0.1819520575963348,
      "grad_norm": 0.8240759372711182,
      "learning_rate": 6.064903190619035e-06,
      "loss": 0.0083,
      "step": 11120
    },
    {
      "epoch": 0.18211568354741062,
      "grad_norm": 0.3187662363052368,
      "learning_rate": 6.070357240250887e-06,
      "loss": 0.0066,
      "step": 11130
    },
    {
      "epoch": 0.18227930949848645,
      "grad_norm": 1.5054137706756592,
      "learning_rate": 6.075811289882738e-06,
      "loss": 0.0093,
      "step": 11140
    },
    {
      "epoch": 0.1824429354495623,
      "grad_norm": 0.8990772366523743,
      "learning_rate": 6.081265339514591e-06,
      "loss": 0.0074,
      "step": 11150
    },
    {
      "epoch": 0.18260656140063813,
      "grad_norm": 0.4735490679740906,
      "learning_rate": 6.0867193891464414e-06,
      "loss": 0.0089,
      "step": 11160
    },
    {
      "epoch": 0.182770187351714,
      "grad_norm": 0.39772114157676697,
      "learning_rate": 6.092173438778294e-06,
      "loss": 0.0082,
      "step": 11170
    },
    {
      "epoch": 0.18293381330278982,
      "grad_norm": 0.540266215801239,
      "learning_rate": 6.097627488410145e-06,
      "loss": 0.0085,
      "step": 11180
    },
    {
      "epoch": 0.18309743925386565,
      "grad_norm": 0.2835488021373749,
      "learning_rate": 6.103081538041997e-06,
      "loss": 0.0117,
      "step": 11190
    },
    {
      "epoch": 0.1832610652049415,
      "grad_norm": 0.4021994471549988,
      "learning_rate": 6.108535587673848e-06,
      "loss": 0.0057,
      "step": 11200
    },
    {
      "epoch": 0.18342469115601734,
      "grad_norm": 0.6735053658485413,
      "learning_rate": 6.113989637305699e-06,
      "loss": 0.0099,
      "step": 11210
    },
    {
      "epoch": 0.1835883171070932,
      "grad_norm": 0.6619128584861755,
      "learning_rate": 6.119443686937552e-06,
      "loss": 0.009,
      "step": 11220
    },
    {
      "epoch": 0.18375194305816903,
      "grad_norm": 0.7361101508140564,
      "learning_rate": 6.1248977365694025e-06,
      "loss": 0.0075,
      "step": 11230
    },
    {
      "epoch": 0.18391556900924486,
      "grad_norm": 0.5200756192207336,
      "learning_rate": 6.130351786201255e-06,
      "loss": 0.0085,
      "step": 11240
    },
    {
      "epoch": 0.1840791949603207,
      "grad_norm": 0.47258487343788147,
      "learning_rate": 6.135805835833106e-06,
      "loss": 0.0085,
      "step": 11250
    },
    {
      "epoch": 0.18424282091139654,
      "grad_norm": 0.22633542120456696,
      "learning_rate": 6.1412598854649584e-06,
      "loss": 0.0087,
      "step": 11260
    },
    {
      "epoch": 0.1844064468624724,
      "grad_norm": 0.7213738560676575,
      "learning_rate": 6.146713935096809e-06,
      "loss": 0.0094,
      "step": 11270
    },
    {
      "epoch": 0.18457007281354823,
      "grad_norm": 0.5219589471817017,
      "learning_rate": 6.152167984728662e-06,
      "loss": 0.0079,
      "step": 11280
    },
    {
      "epoch": 0.18473369876462406,
      "grad_norm": 0.5819460153579712,
      "learning_rate": 6.157622034360513e-06,
      "loss": 0.0091,
      "step": 11290
    },
    {
      "epoch": 0.18489732471569992,
      "grad_norm": 0.3599722385406494,
      "learning_rate": 6.163076083992365e-06,
      "loss": 0.0087,
      "step": 11300
    },
    {
      "epoch": 0.18506095066677575,
      "grad_norm": 0.49820250272750854,
      "learning_rate": 6.168530133624216e-06,
      "loss": 0.0133,
      "step": 11310
    },
    {
      "epoch": 0.1852245766178516,
      "grad_norm": 0.5457605123519897,
      "learning_rate": 6.173984183256068e-06,
      "loss": 0.0065,
      "step": 11320
    },
    {
      "epoch": 0.18538820256892743,
      "grad_norm": 0.5832045078277588,
      "learning_rate": 6.1794382328879195e-06,
      "loss": 0.0092,
      "step": 11330
    },
    {
      "epoch": 0.18555182852000326,
      "grad_norm": 0.42447108030319214,
      "learning_rate": 6.184892282519771e-06,
      "loss": 0.0104,
      "step": 11340
    },
    {
      "epoch": 0.18571545447107912,
      "grad_norm": 0.5838319063186646,
      "learning_rate": 6.190346332151623e-06,
      "loss": 0.0069,
      "step": 11350
    },
    {
      "epoch": 0.18587908042215495,
      "grad_norm": 0.37487807869911194,
      "learning_rate": 6.195800381783475e-06,
      "loss": 0.0094,
      "step": 11360
    },
    {
      "epoch": 0.18604270637323078,
      "grad_norm": 0.45324909687042236,
      "learning_rate": 6.201254431415326e-06,
      "loss": 0.0108,
      "step": 11370
    },
    {
      "epoch": 0.18620633232430664,
      "grad_norm": 0.6902481317520142,
      "learning_rate": 6.206708481047178e-06,
      "loss": 0.0119,
      "step": 11380
    },
    {
      "epoch": 0.18636995827538247,
      "grad_norm": 0.4646401107311249,
      "learning_rate": 6.21216253067903e-06,
      "loss": 0.0129,
      "step": 11390
    },
    {
      "epoch": 0.18653358422645833,
      "grad_norm": 0.49047648906707764,
      "learning_rate": 6.217616580310881e-06,
      "loss": 0.01,
      "step": 11400
    },
    {
      "epoch": 0.18669721017753416,
      "grad_norm": 0.6002476811408997,
      "learning_rate": 6.223070629942733e-06,
      "loss": 0.009,
      "step": 11410
    },
    {
      "epoch": 0.18686083612861,
      "grad_norm": 1.2768020629882812,
      "learning_rate": 6.228524679574585e-06,
      "loss": 0.0109,
      "step": 11420
    },
    {
      "epoch": 0.18702446207968584,
      "grad_norm": 0.4653293490409851,
      "learning_rate": 6.233978729206436e-06,
      "loss": 0.0081,
      "step": 11430
    },
    {
      "epoch": 0.18718808803076167,
      "grad_norm": 0.37930235266685486,
      "learning_rate": 6.239432778838288e-06,
      "loss": 0.0087,
      "step": 11440
    },
    {
      "epoch": 0.18735171398183753,
      "grad_norm": 0.6293927431106567,
      "learning_rate": 6.244886828470139e-06,
      "loss": 0.012,
      "step": 11450
    },
    {
      "epoch": 0.18751533993291336,
      "grad_norm": 0.38692766427993774,
      "learning_rate": 6.250340878101992e-06,
      "loss": 0.0076,
      "step": 11460
    },
    {
      "epoch": 0.1876789658839892,
      "grad_norm": 0.28060996532440186,
      "learning_rate": 6.2557949277338425e-06,
      "loss": 0.0097,
      "step": 11470
    },
    {
      "epoch": 0.18784259183506505,
      "grad_norm": 0.27393385767936707,
      "learning_rate": 6.261248977365695e-06,
      "loss": 0.0092,
      "step": 11480
    },
    {
      "epoch": 0.18800621778614088,
      "grad_norm": 0.36885377764701843,
      "learning_rate": 6.266703026997546e-06,
      "loss": 0.0072,
      "step": 11490
    },
    {
      "epoch": 0.18816984373721674,
      "grad_norm": 0.48685377836227417,
      "learning_rate": 6.272157076629398e-06,
      "loss": 0.0075,
      "step": 11500
    },
    {
      "epoch": 0.18833346968829257,
      "grad_norm": 0.3773362636566162,
      "learning_rate": 6.277611126261249e-06,
      "loss": 0.0074,
      "step": 11510
    },
    {
      "epoch": 0.1884970956393684,
      "grad_norm": 0.6746549606323242,
      "learning_rate": 6.283065175893102e-06,
      "loss": 0.0128,
      "step": 11520
    },
    {
      "epoch": 0.18866072159044425,
      "grad_norm": 0.5694946050643921,
      "learning_rate": 6.288519225524953e-06,
      "loss": 0.0115,
      "step": 11530
    },
    {
      "epoch": 0.18882434754152008,
      "grad_norm": 0.36031264066696167,
      "learning_rate": 6.2939732751568035e-06,
      "loss": 0.0087,
      "step": 11540
    },
    {
      "epoch": 0.1889879734925959,
      "grad_norm": 0.6497841477394104,
      "learning_rate": 6.299427324788656e-06,
      "loss": 0.0104,
      "step": 11550
    },
    {
      "epoch": 0.18915159944367177,
      "grad_norm": 0.5403355956077576,
      "learning_rate": 6.304881374420507e-06,
      "loss": 0.0095,
      "step": 11560
    },
    {
      "epoch": 0.1893152253947476,
      "grad_norm": 0.7536537647247314,
      "learning_rate": 6.3103354240523595e-06,
      "loss": 0.0107,
      "step": 11570
    },
    {
      "epoch": 0.18947885134582346,
      "grad_norm": 0.34756720066070557,
      "learning_rate": 6.31578947368421e-06,
      "loss": 0.0101,
      "step": 11580
    },
    {
      "epoch": 0.1896424772968993,
      "grad_norm": 0.1733919233083725,
      "learning_rate": 6.321243523316063e-06,
      "loss": 0.0062,
      "step": 11590
    },
    {
      "epoch": 0.18980610324797512,
      "grad_norm": 0.303161084651947,
      "learning_rate": 6.326697572947914e-06,
      "loss": 0.0125,
      "step": 11600
    },
    {
      "epoch": 0.18996972919905097,
      "grad_norm": 0.3984090983867645,
      "learning_rate": 6.332151622579766e-06,
      "loss": 0.007,
      "step": 11610
    },
    {
      "epoch": 0.1901333551501268,
      "grad_norm": 0.2361973524093628,
      "learning_rate": 6.337605672211617e-06,
      "loss": 0.0105,
      "step": 11620
    },
    {
      "epoch": 0.19029698110120266,
      "grad_norm": 0.2629629373550415,
      "learning_rate": 6.34305972184347e-06,
      "loss": 0.007,
      "step": 11630
    },
    {
      "epoch": 0.1904606070522785,
      "grad_norm": 0.6978625655174255,
      "learning_rate": 6.3485137714753205e-06,
      "loss": 0.0097,
      "step": 11640
    },
    {
      "epoch": 0.19062423300335432,
      "grad_norm": 0.2649421989917755,
      "learning_rate": 6.353967821107173e-06,
      "loss": 0.0078,
      "step": 11650
    },
    {
      "epoch": 0.19078785895443018,
      "grad_norm": 0.33282360434532166,
      "learning_rate": 6.359421870739024e-06,
      "loss": 0.0073,
      "step": 11660
    },
    {
      "epoch": 0.190951484905506,
      "grad_norm": 0.5424314737319946,
      "learning_rate": 6.364875920370876e-06,
      "loss": 0.0081,
      "step": 11670
    },
    {
      "epoch": 0.19111511085658187,
      "grad_norm": 0.1660630702972412,
      "learning_rate": 6.370329970002727e-06,
      "loss": 0.0078,
      "step": 11680
    },
    {
      "epoch": 0.1912787368076577,
      "grad_norm": 0.6327295303344727,
      "learning_rate": 6.375784019634579e-06,
      "loss": 0.0109,
      "step": 11690
    },
    {
      "epoch": 0.19144236275873353,
      "grad_norm": 0.32394713163375854,
      "learning_rate": 6.381238069266431e-06,
      "loss": 0.0061,
      "step": 11700
    },
    {
      "epoch": 0.19160598870980938,
      "grad_norm": 0.5135193467140198,
      "learning_rate": 6.3866921188982824e-06,
      "loss": 0.01,
      "step": 11710
    },
    {
      "epoch": 0.1917696146608852,
      "grad_norm": 0.6632450819015503,
      "learning_rate": 6.392146168530134e-06,
      "loss": 0.0097,
      "step": 11720
    },
    {
      "epoch": 0.19193324061196107,
      "grad_norm": 0.6025798320770264,
      "learning_rate": 6.397600218161986e-06,
      "loss": 0.0077,
      "step": 11730
    },
    {
      "epoch": 0.1920968665630369,
      "grad_norm": 0.5043540000915527,
      "learning_rate": 6.4030542677938375e-06,
      "loss": 0.0056,
      "step": 11740
    },
    {
      "epoch": 0.19226049251411273,
      "grad_norm": 0.14362969994544983,
      "learning_rate": 6.408508317425689e-06,
      "loss": 0.0086,
      "step": 11750
    },
    {
      "epoch": 0.1924241184651886,
      "grad_norm": 0.22613579034805298,
      "learning_rate": 6.413962367057541e-06,
      "loss": 0.0058,
      "step": 11760
    },
    {
      "epoch": 0.19258774441626442,
      "grad_norm": 0.7183791399002075,
      "learning_rate": 6.419416416689393e-06,
      "loss": 0.0111,
      "step": 11770
    },
    {
      "epoch": 0.19275137036734025,
      "grad_norm": 0.6961106061935425,
      "learning_rate": 6.4248704663212435e-06,
      "loss": 0.0076,
      "step": 11780
    },
    {
      "epoch": 0.1929149963184161,
      "grad_norm": 0.328006386756897,
      "learning_rate": 6.430324515953096e-06,
      "loss": 0.0068,
      "step": 11790
    },
    {
      "epoch": 0.19307862226949193,
      "grad_norm": 0.6311428546905518,
      "learning_rate": 6.435778565584947e-06,
      "loss": 0.0083,
      "step": 11800
    },
    {
      "epoch": 0.1932422482205678,
      "grad_norm": 0.3945896029472351,
      "learning_rate": 6.4412326152167994e-06,
      "loss": 0.0066,
      "step": 11810
    },
    {
      "epoch": 0.19340587417164362,
      "grad_norm": 0.6109340190887451,
      "learning_rate": 6.44668666484865e-06,
      "loss": 0.0104,
      "step": 11820
    },
    {
      "epoch": 0.19356950012271945,
      "grad_norm": 0.5317077040672302,
      "learning_rate": 6.452140714480503e-06,
      "loss": 0.0082,
      "step": 11830
    },
    {
      "epoch": 0.1937331260737953,
      "grad_norm": 0.787543773651123,
      "learning_rate": 6.457594764112354e-06,
      "loss": 0.0066,
      "step": 11840
    },
    {
      "epoch": 0.19389675202487114,
      "grad_norm": 0.7824800610542297,
      "learning_rate": 6.463048813744206e-06,
      "loss": 0.0074,
      "step": 11850
    },
    {
      "epoch": 0.194060377975947,
      "grad_norm": 0.4073673486709595,
      "learning_rate": 6.468502863376057e-06,
      "loss": 0.0094,
      "step": 11860
    },
    {
      "epoch": 0.19422400392702283,
      "grad_norm": 0.5223649740219116,
      "learning_rate": 6.47395691300791e-06,
      "loss": 0.0116,
      "step": 11870
    },
    {
      "epoch": 0.19438762987809866,
      "grad_norm": 0.696689784526825,
      "learning_rate": 6.4794109626397605e-06,
      "loss": 0.0118,
      "step": 11880
    },
    {
      "epoch": 0.1945512558291745,
      "grad_norm": 0.7363672852516174,
      "learning_rate": 6.484865012271611e-06,
      "loss": 0.008,
      "step": 11890
    },
    {
      "epoch": 0.19471488178025034,
      "grad_norm": 0.7384046912193298,
      "learning_rate": 6.490319061903464e-06,
      "loss": 0.0117,
      "step": 11900
    },
    {
      "epoch": 0.1948785077313262,
      "grad_norm": 0.4886281192302704,
      "learning_rate": 6.495773111535315e-06,
      "loss": 0.007,
      "step": 11910
    },
    {
      "epoch": 0.19504213368240203,
      "grad_norm": 0.45536166429519653,
      "learning_rate": 6.501227161167167e-06,
      "loss": 0.0084,
      "step": 11920
    },
    {
      "epoch": 0.19520575963347786,
      "grad_norm": 0.1903892308473587,
      "learning_rate": 6.506681210799018e-06,
      "loss": 0.0065,
      "step": 11930
    },
    {
      "epoch": 0.19536938558455372,
      "grad_norm": 0.395683616399765,
      "learning_rate": 6.512135260430871e-06,
      "loss": 0.0074,
      "step": 11940
    },
    {
      "epoch": 0.19553301153562955,
      "grad_norm": 0.3251670300960541,
      "learning_rate": 6.5175893100627216e-06,
      "loss": 0.0102,
      "step": 11950
    },
    {
      "epoch": 0.1956966374867054,
      "grad_norm": 0.48893827199935913,
      "learning_rate": 6.523043359694574e-06,
      "loss": 0.0069,
      "step": 11960
    },
    {
      "epoch": 0.19586026343778123,
      "grad_norm": 0.5587809681892395,
      "learning_rate": 6.528497409326425e-06,
      "loss": 0.0081,
      "step": 11970
    },
    {
      "epoch": 0.19602388938885706,
      "grad_norm": 0.6132639050483704,
      "learning_rate": 6.5339514589582775e-06,
      "loss": 0.0088,
      "step": 11980
    },
    {
      "epoch": 0.19618751533993292,
      "grad_norm": 0.5497077107429504,
      "learning_rate": 6.539405508590128e-06,
      "loss": 0.0149,
      "step": 11990
    },
    {
      "epoch": 0.19635114129100875,
      "grad_norm": 0.42586514353752136,
      "learning_rate": 6.54485955822198e-06,
      "loss": 0.0087,
      "step": 12000
    },
    {
      "epoch": 0.19651476724208458,
      "grad_norm": 0.2691415548324585,
      "learning_rate": 6.550313607853832e-06,
      "loss": 0.0102,
      "step": 12010
    },
    {
      "epoch": 0.19667839319316044,
      "grad_norm": 0.37258654832839966,
      "learning_rate": 6.5557676574856835e-06,
      "loss": 0.0105,
      "step": 12020
    },
    {
      "epoch": 0.19684201914423627,
      "grad_norm": 0.43392255902290344,
      "learning_rate": 6.561221707117535e-06,
      "loss": 0.0095,
      "step": 12030
    },
    {
      "epoch": 0.19700564509531213,
      "grad_norm": 0.1617450714111328,
      "learning_rate": 6.566675756749387e-06,
      "loss": 0.0088,
      "step": 12040
    },
    {
      "epoch": 0.19716927104638796,
      "grad_norm": 0.48435476422309875,
      "learning_rate": 6.572129806381239e-06,
      "loss": 0.0101,
      "step": 12050
    },
    {
      "epoch": 0.19733289699746379,
      "grad_norm": 0.6208023428916931,
      "learning_rate": 6.57758385601309e-06,
      "loss": 0.0132,
      "step": 12060
    },
    {
      "epoch": 0.19749652294853964,
      "grad_norm": 0.4030143618583679,
      "learning_rate": 6.583037905644942e-06,
      "loss": 0.0091,
      "step": 12070
    },
    {
      "epoch": 0.19766014889961547,
      "grad_norm": 0.38547268509864807,
      "learning_rate": 6.588491955276794e-06,
      "loss": 0.0085,
      "step": 12080
    },
    {
      "epoch": 0.19782377485069133,
      "grad_norm": 0.6603232622146606,
      "learning_rate": 6.593946004908645e-06,
      "loss": 0.0092,
      "step": 12090
    },
    {
      "epoch": 0.19798740080176716,
      "grad_norm": 0.5414279103279114,
      "learning_rate": 6.599400054540497e-06,
      "loss": 0.0113,
      "step": 12100
    },
    {
      "epoch": 0.198151026752843,
      "grad_norm": 0.45296400785446167,
      "learning_rate": 6.604854104172348e-06,
      "loss": 0.0077,
      "step": 12110
    },
    {
      "epoch": 0.19831465270391885,
      "grad_norm": 0.6430184841156006,
      "learning_rate": 6.6103081538042005e-06,
      "loss": 0.0082,
      "step": 12120
    },
    {
      "epoch": 0.19847827865499468,
      "grad_norm": 0.1737295538187027,
      "learning_rate": 6.615762203436051e-06,
      "loss": 0.0059,
      "step": 12130
    },
    {
      "epoch": 0.19864190460607054,
      "grad_norm": 0.3895517587661743,
      "learning_rate": 6.621216253067904e-06,
      "loss": 0.0093,
      "step": 12140
    },
    {
      "epoch": 0.19880553055714636,
      "grad_norm": 1.083504319190979,
      "learning_rate": 6.626670302699755e-06,
      "loss": 0.0112,
      "step": 12150
    },
    {
      "epoch": 0.1989691565082222,
      "grad_norm": 0.7300825119018555,
      "learning_rate": 6.632124352331607e-06,
      "loss": 0.0073,
      "step": 12160
    },
    {
      "epoch": 0.19913278245929805,
      "grad_norm": 0.44121623039245605,
      "learning_rate": 6.637578401963458e-06,
      "loss": 0.0126,
      "step": 12170
    },
    {
      "epoch": 0.19929640841037388,
      "grad_norm": 0.421705961227417,
      "learning_rate": 6.643032451595311e-06,
      "loss": 0.0104,
      "step": 12180
    },
    {
      "epoch": 0.19946003436144974,
      "grad_norm": 0.33037325739860535,
      "learning_rate": 6.6484865012271615e-06,
      "loss": 0.0079,
      "step": 12190
    },
    {
      "epoch": 0.19962366031252557,
      "grad_norm": 1.2830380201339722,
      "learning_rate": 6.653940550859014e-06,
      "loss": 0.0089,
      "step": 12200
    },
    {
      "epoch": 0.1997872862636014,
      "grad_norm": 0.4163344204425812,
      "learning_rate": 6.659394600490865e-06,
      "loss": 0.0055,
      "step": 12210
    },
    {
      "epoch": 0.19995091221467726,
      "grad_norm": 0.9814919829368591,
      "learning_rate": 6.6648486501227175e-06,
      "loss": 0.0103,
      "step": 12220
    },
    {
      "epoch": 0.2001145381657531,
      "grad_norm": 0.6290691494941711,
      "learning_rate": 6.670302699754568e-06,
      "loss": 0.01,
      "step": 12230
    },
    {
      "epoch": 0.20027816411682892,
      "grad_norm": 0.5497564673423767,
      "learning_rate": 6.675756749386419e-06,
      "loss": 0.0101,
      "step": 12240
    },
    {
      "epoch": 0.20044179006790477,
      "grad_norm": 0.5460956692695618,
      "learning_rate": 6.681210799018272e-06,
      "loss": 0.0104,
      "step": 12250
    },
    {
      "epoch": 0.2006054160189806,
      "grad_norm": 0.45882344245910645,
      "learning_rate": 6.686664848650123e-06,
      "loss": 0.0066,
      "step": 12260
    },
    {
      "epoch": 0.20076904197005646,
      "grad_norm": 0.3451559245586395,
      "learning_rate": 6.692118898281975e-06,
      "loss": 0.0063,
      "step": 12270
    },
    {
      "epoch": 0.2009326679211323,
      "grad_norm": 0.6180064082145691,
      "learning_rate": 6.697572947913826e-06,
      "loss": 0.0084,
      "step": 12280
    },
    {
      "epoch": 0.20109629387220812,
      "grad_norm": 0.5369217991828918,
      "learning_rate": 6.7030269975456786e-06,
      "loss": 0.0116,
      "step": 12290
    },
    {
      "epoch": 0.20125991982328398,
      "grad_norm": 0.4721609354019165,
      "learning_rate": 6.708481047177529e-06,
      "loss": 0.0077,
      "step": 12300
    },
    {
      "epoch": 0.2014235457743598,
      "grad_norm": 0.5312516689300537,
      "learning_rate": 6.713935096809382e-06,
      "loss": 0.0097,
      "step": 12310
    },
    {
      "epoch": 0.20158717172543567,
      "grad_norm": 0.32041916251182556,
      "learning_rate": 6.719389146441233e-06,
      "loss": 0.008,
      "step": 12320
    },
    {
      "epoch": 0.2017507976765115,
      "grad_norm": 0.4275837242603302,
      "learning_rate": 6.724843196073085e-06,
      "loss": 0.0095,
      "step": 12330
    },
    {
      "epoch": 0.20191442362758732,
      "grad_norm": 0.5569990277290344,
      "learning_rate": 6.730297245704936e-06,
      "loss": 0.007,
      "step": 12340
    },
    {
      "epoch": 0.20207804957866318,
      "grad_norm": 0.37230172753334045,
      "learning_rate": 6.735751295336788e-06,
      "loss": 0.0093,
      "step": 12350
    },
    {
      "epoch": 0.202241675529739,
      "grad_norm": 0.4427005648612976,
      "learning_rate": 6.74120534496864e-06,
      "loss": 0.0076,
      "step": 12360
    },
    {
      "epoch": 0.20240530148081487,
      "grad_norm": 0.29859134554862976,
      "learning_rate": 6.746659394600491e-06,
      "loss": 0.0082,
      "step": 12370
    },
    {
      "epoch": 0.2025689274318907,
      "grad_norm": 0.33112832903862,
      "learning_rate": 6.752113444232343e-06,
      "loss": 0.0075,
      "step": 12380
    },
    {
      "epoch": 0.20273255338296653,
      "grad_norm": 0.13960298895835876,
      "learning_rate": 6.757567493864195e-06,
      "loss": 0.0155,
      "step": 12390
    },
    {
      "epoch": 0.2028961793340424,
      "grad_norm": 0.19399921596050262,
      "learning_rate": 6.763021543496046e-06,
      "loss": 0.0055,
      "step": 12400
    },
    {
      "epoch": 0.20305980528511822,
      "grad_norm": 0.6567419767379761,
      "learning_rate": 6.768475593127898e-06,
      "loss": 0.01,
      "step": 12410
    },
    {
      "epoch": 0.20322343123619407,
      "grad_norm": 1.0741888284683228,
      "learning_rate": 6.77392964275975e-06,
      "loss": 0.0084,
      "step": 12420
    },
    {
      "epoch": 0.2033870571872699,
      "grad_norm": 0.6558672189712524,
      "learning_rate": 6.7793836923916015e-06,
      "loss": 0.0084,
      "step": 12430
    },
    {
      "epoch": 0.20355068313834573,
      "grad_norm": 0.6887688040733337,
      "learning_rate": 6.784837742023453e-06,
      "loss": 0.0089,
      "step": 12440
    },
    {
      "epoch": 0.2037143090894216,
      "grad_norm": 0.5970726013183594,
      "learning_rate": 6.790291791655305e-06,
      "loss": 0.0084,
      "step": 12450
    },
    {
      "epoch": 0.20387793504049742,
      "grad_norm": 0.534174382686615,
      "learning_rate": 6.795745841287156e-06,
      "loss": 0.0099,
      "step": 12460
    },
    {
      "epoch": 0.20404156099157325,
      "grad_norm": 0.45492666959762573,
      "learning_rate": 6.801199890919008e-06,
      "loss": 0.0094,
      "step": 12470
    },
    {
      "epoch": 0.2042051869426491,
      "grad_norm": 0.6736888885498047,
      "learning_rate": 6.806653940550859e-06,
      "loss": 0.0083,
      "step": 12480
    },
    {
      "epoch": 0.20436881289372494,
      "grad_norm": 0.42749863862991333,
      "learning_rate": 6.812107990182712e-06,
      "loss": 0.0079,
      "step": 12490
    },
    {
      "epoch": 0.2045324388448008,
      "grad_norm": 0.21442770957946777,
      "learning_rate": 6.8175620398145626e-06,
      "loss": 0.0114,
      "step": 12500
    },
    {
      "epoch": 0.20469606479587663,
      "grad_norm": 0.36209985613822937,
      "learning_rate": 6.823016089446415e-06,
      "loss": 0.0097,
      "step": 12510
    },
    {
      "epoch": 0.20485969074695246,
      "grad_norm": 0.41109931468963623,
      "learning_rate": 6.828470139078266e-06,
      "loss": 0.0081,
      "step": 12520
    },
    {
      "epoch": 0.2050233166980283,
      "grad_norm": 0.30309098958969116,
      "learning_rate": 6.8339241887101185e-06,
      "loss": 0.0115,
      "step": 12530
    },
    {
      "epoch": 0.20518694264910414,
      "grad_norm": 0.34034353494644165,
      "learning_rate": 6.839378238341969e-06,
      "loss": 0.0083,
      "step": 12540
    },
    {
      "epoch": 0.20535056860018,
      "grad_norm": 0.3176431953907013,
      "learning_rate": 6.844832287973822e-06,
      "loss": 0.0058,
      "step": 12550
    },
    {
      "epoch": 0.20551419455125583,
      "grad_norm": 0.5689629912376404,
      "learning_rate": 6.850286337605673e-06,
      "loss": 0.0082,
      "step": 12560
    },
    {
      "epoch": 0.20567782050233166,
      "grad_norm": 0.702111005783081,
      "learning_rate": 6.855740387237524e-06,
      "loss": 0.0069,
      "step": 12570
    },
    {
      "epoch": 0.20584144645340752,
      "grad_norm": 0.5140200257301331,
      "learning_rate": 6.861194436869376e-06,
      "loss": 0.0113,
      "step": 12580
    },
    {
      "epoch": 0.20600507240448335,
      "grad_norm": 0.5493470430374146,
      "learning_rate": 6.866648486501227e-06,
      "loss": 0.0089,
      "step": 12590
    },
    {
      "epoch": 0.2061686983555592,
      "grad_norm": 0.6805787086486816,
      "learning_rate": 6.87210253613308e-06,
      "loss": 0.0091,
      "step": 12600
    },
    {
      "epoch": 0.20633232430663503,
      "grad_norm": 0.6136297583580017,
      "learning_rate": 6.8775565857649304e-06,
      "loss": 0.0078,
      "step": 12610
    },
    {
      "epoch": 0.20649595025771086,
      "grad_norm": 0.3267809748649597,
      "learning_rate": 6.883010635396783e-06,
      "loss": 0.008,
      "step": 12620
    },
    {
      "epoch": 0.20665957620878672,
      "grad_norm": 0.29001057147979736,
      "learning_rate": 6.888464685028634e-06,
      "loss": 0.0097,
      "step": 12630
    },
    {
      "epoch": 0.20682320215986255,
      "grad_norm": 0.689616322517395,
      "learning_rate": 6.893918734660486e-06,
      "loss": 0.0091,
      "step": 12640
    },
    {
      "epoch": 0.20698682811093838,
      "grad_norm": 0.37942513823509216,
      "learning_rate": 6.899372784292337e-06,
      "loss": 0.0115,
      "step": 12650
    },
    {
      "epoch": 0.20715045406201424,
      "grad_norm": 0.2932007908821106,
      "learning_rate": 6.90482683392419e-06,
      "loss": 0.0079,
      "step": 12660
    },
    {
      "epoch": 0.20731408001309007,
      "grad_norm": 0.36570748686790466,
      "learning_rate": 6.910280883556041e-06,
      "loss": 0.0105,
      "step": 12670
    },
    {
      "epoch": 0.20747770596416593,
      "grad_norm": 0.5460178852081299,
      "learning_rate": 6.915734933187893e-06,
      "loss": 0.0116,
      "step": 12680
    },
    {
      "epoch": 0.20764133191524176,
      "grad_norm": 0.2802366018295288,
      "learning_rate": 6.921188982819744e-06,
      "loss": 0.0066,
      "step": 12690
    },
    {
      "epoch": 0.20780495786631759,
      "grad_norm": 0.4743119478225708,
      "learning_rate": 6.926643032451596e-06,
      "loss": 0.0084,
      "step": 12700
    },
    {
      "epoch": 0.20796858381739344,
      "grad_norm": 0.5068359375,
      "learning_rate": 6.9320970820834474e-06,
      "loss": 0.0091,
      "step": 12710
    },
    {
      "epoch": 0.20813220976846927,
      "grad_norm": 0.21520860493183136,
      "learning_rate": 6.937551131715299e-06,
      "loss": 0.0056,
      "step": 12720
    },
    {
      "epoch": 0.20829583571954513,
      "grad_norm": 0.28961360454559326,
      "learning_rate": 6.943005181347151e-06,
      "loss": 0.0077,
      "step": 12730
    },
    {
      "epoch": 0.20845946167062096,
      "grad_norm": 0.5005819201469421,
      "learning_rate": 6.9484592309790025e-06,
      "loss": 0.0107,
      "step": 12740
    },
    {
      "epoch": 0.2086230876216968,
      "grad_norm": 0.31138646602630615,
      "learning_rate": 6.953913280610854e-06,
      "loss": 0.0086,
      "step": 12750
    },
    {
      "epoch": 0.20878671357277265,
      "grad_norm": 0.6689060926437378,
      "learning_rate": 6.959367330242706e-06,
      "loss": 0.0103,
      "step": 12760
    },
    {
      "epoch": 0.20895033952384848,
      "grad_norm": 0.5686567425727844,
      "learning_rate": 6.964821379874558e-06,
      "loss": 0.011,
      "step": 12770
    },
    {
      "epoch": 0.20911396547492433,
      "grad_norm": 0.18714195489883423,
      "learning_rate": 6.970275429506409e-06,
      "loss": 0.0064,
      "step": 12780
    },
    {
      "epoch": 0.20927759142600016,
      "grad_norm": 0.5293023586273193,
      "learning_rate": 6.975729479138261e-06,
      "loss": 0.011,
      "step": 12790
    },
    {
      "epoch": 0.209441217377076,
      "grad_norm": 0.29385289549827576,
      "learning_rate": 6.981183528770113e-06,
      "loss": 0.0114,
      "step": 12800
    },
    {
      "epoch": 0.20960484332815185,
      "grad_norm": 0.31326308846473694,
      "learning_rate": 6.986637578401964e-06,
      "loss": 0.0074,
      "step": 12810
    },
    {
      "epoch": 0.20976846927922768,
      "grad_norm": 0.7566695213317871,
      "learning_rate": 6.992091628033816e-06,
      "loss": 0.0079,
      "step": 12820
    },
    {
      "epoch": 0.20993209523030354,
      "grad_norm": 0.267514169216156,
      "learning_rate": 6.997545677665667e-06,
      "loss": 0.0082,
      "step": 12830
    },
    {
      "epoch": 0.21009572118137937,
      "grad_norm": 0.0504985935986042,
      "learning_rate": 7.0029997272975196e-06,
      "loss": 0.0091,
      "step": 12840
    },
    {
      "epoch": 0.2102593471324552,
      "grad_norm": 0.4737120270729065,
      "learning_rate": 7.00845377692937e-06,
      "loss": 0.0082,
      "step": 12850
    },
    {
      "epoch": 0.21042297308353106,
      "grad_norm": 0.23650555312633514,
      "learning_rate": 7.013907826561223e-06,
      "loss": 0.0076,
      "step": 12860
    },
    {
      "epoch": 0.21058659903460689,
      "grad_norm": 0.41000521183013916,
      "learning_rate": 7.019361876193074e-06,
      "loss": 0.0051,
      "step": 12870
    },
    {
      "epoch": 0.21075022498568272,
      "grad_norm": 0.18648408353328705,
      "learning_rate": 7.024815925824926e-06,
      "loss": 0.0059,
      "step": 12880
    },
    {
      "epoch": 0.21091385093675857,
      "grad_norm": 0.34689322113990784,
      "learning_rate": 7.030269975456777e-06,
      "loss": 0.0071,
      "step": 12890
    },
    {
      "epoch": 0.2110774768878344,
      "grad_norm": 0.7265233397483826,
      "learning_rate": 7.03572402508863e-06,
      "loss": 0.0092,
      "step": 12900
    },
    {
      "epoch": 0.21124110283891026,
      "grad_norm": 0.4350171685218811,
      "learning_rate": 7.041178074720481e-06,
      "loss": 0.0096,
      "step": 12910
    },
    {
      "epoch": 0.2114047287899861,
      "grad_norm": 0.5271282196044922,
      "learning_rate": 7.0466321243523315e-06,
      "loss": 0.0084,
      "step": 12920
    },
    {
      "epoch": 0.21156835474106192,
      "grad_norm": 0.6984720230102539,
      "learning_rate": 7.052086173984184e-06,
      "loss": 0.0063,
      "step": 12930
    },
    {
      "epoch": 0.21173198069213778,
      "grad_norm": 0.6105838418006897,
      "learning_rate": 7.057540223616035e-06,
      "loss": 0.0074,
      "step": 12940
    },
    {
      "epoch": 0.2118956066432136,
      "grad_norm": 0.3719397485256195,
      "learning_rate": 7.062994273247887e-06,
      "loss": 0.0065,
      "step": 12950
    },
    {
      "epoch": 0.21205923259428947,
      "grad_norm": 0.35808417201042175,
      "learning_rate": 7.068448322879738e-06,
      "loss": 0.0087,
      "step": 12960
    },
    {
      "epoch": 0.2122228585453653,
      "grad_norm": 0.4282531142234802,
      "learning_rate": 7.073902372511591e-06,
      "loss": 0.0077,
      "step": 12970
    },
    {
      "epoch": 0.21238648449644112,
      "grad_norm": 0.5499398708343506,
      "learning_rate": 7.079356422143442e-06,
      "loss": 0.0098,
      "step": 12980
    },
    {
      "epoch": 0.21255011044751698,
      "grad_norm": 0.5832967758178711,
      "learning_rate": 7.084810471775294e-06,
      "loss": 0.0113,
      "step": 12990
    },
    {
      "epoch": 0.2127137363985928,
      "grad_norm": 0.6456331014633179,
      "learning_rate": 7.090264521407145e-06,
      "loss": 0.0067,
      "step": 13000
    },
    {
      "epoch": 0.21287736234966867,
      "grad_norm": 0.42318278551101685,
      "learning_rate": 7.095718571038998e-06,
      "loss": 0.0045,
      "step": 13010
    },
    {
      "epoch": 0.2130409883007445,
      "grad_norm": 0.2748035788536072,
      "learning_rate": 7.1011726206708485e-06,
      "loss": 0.0059,
      "step": 13020
    },
    {
      "epoch": 0.21320461425182033,
      "grad_norm": 0.4589381814002991,
      "learning_rate": 7.1066266703027e-06,
      "loss": 0.0091,
      "step": 13030
    },
    {
      "epoch": 0.2133682402028962,
      "grad_norm": 0.5025801658630371,
      "learning_rate": 7.112080719934552e-06,
      "loss": 0.0093,
      "step": 13040
    },
    {
      "epoch": 0.21353186615397202,
      "grad_norm": 1.3389508724212646,
      "learning_rate": 7.117534769566404e-06,
      "loss": 0.009,
      "step": 13050
    },
    {
      "epoch": 0.21369549210504787,
      "grad_norm": 0.28795158863067627,
      "learning_rate": 7.122988819198255e-06,
      "loss": 0.0075,
      "step": 13060
    },
    {
      "epoch": 0.2138591180561237,
      "grad_norm": 0.628476619720459,
      "learning_rate": 7.128442868830107e-06,
      "loss": 0.0079,
      "step": 13070
    },
    {
      "epoch": 0.21402274400719953,
      "grad_norm": 1.038084626197815,
      "learning_rate": 7.133896918461959e-06,
      "loss": 0.0093,
      "step": 13080
    },
    {
      "epoch": 0.2141863699582754,
      "grad_norm": 0.44432532787323,
      "learning_rate": 7.13935096809381e-06,
      "loss": 0.0071,
      "step": 13090
    },
    {
      "epoch": 0.21434999590935122,
      "grad_norm": 0.37490004301071167,
      "learning_rate": 7.144805017725662e-06,
      "loss": 0.0101,
      "step": 13100
    },
    {
      "epoch": 0.21451362186042705,
      "grad_norm": 0.3405896723270416,
      "learning_rate": 7.150259067357514e-06,
      "loss": 0.0065,
      "step": 13110
    },
    {
      "epoch": 0.2146772478115029,
      "grad_norm": 0.3738662302494049,
      "learning_rate": 7.1557131169893655e-06,
      "loss": 0.0065,
      "step": 13120
    },
    {
      "epoch": 0.21484087376257874,
      "grad_norm": 0.5956369638442993,
      "learning_rate": 7.161167166621217e-06,
      "loss": 0.0088,
      "step": 13130
    },
    {
      "epoch": 0.2150044997136546,
      "grad_norm": 0.3642753064632416,
      "learning_rate": 7.166621216253068e-06,
      "loss": 0.0084,
      "step": 13140
    },
    {
      "epoch": 0.21516812566473043,
      "grad_norm": 0.3063509166240692,
      "learning_rate": 7.172075265884921e-06,
      "loss": 0.0093,
      "step": 13150
    },
    {
      "epoch": 0.21533175161580626,
      "grad_norm": 0.26861312985420227,
      "learning_rate": 7.1775293155167714e-06,
      "loss": 0.0088,
      "step": 13160
    },
    {
      "epoch": 0.2154953775668821,
      "grad_norm": 0.5493288636207581,
      "learning_rate": 7.182983365148624e-06,
      "loss": 0.009,
      "step": 13170
    },
    {
      "epoch": 0.21565900351795794,
      "grad_norm": 0.47742828726768494,
      "learning_rate": 7.188437414780475e-06,
      "loss": 0.0104,
      "step": 13180
    },
    {
      "epoch": 0.2158226294690338,
      "grad_norm": 0.36240002512931824,
      "learning_rate": 7.193891464412327e-06,
      "loss": 0.0112,
      "step": 13190
    },
    {
      "epoch": 0.21598625542010963,
      "grad_norm": 0.48867344856262207,
      "learning_rate": 7.199345514044178e-06,
      "loss": 0.0082,
      "step": 13200
    },
    {
      "epoch": 0.21614988137118546,
      "grad_norm": 0.41958922147750854,
      "learning_rate": 7.204799563676031e-06,
      "loss": 0.0087,
      "step": 13210
    },
    {
      "epoch": 0.21631350732226132,
      "grad_norm": 0.6330770254135132,
      "learning_rate": 7.210253613307882e-06,
      "loss": 0.0077,
      "step": 13220
    },
    {
      "epoch": 0.21647713327333715,
      "grad_norm": 0.342072993516922,
      "learning_rate": 7.215707662939734e-06,
      "loss": 0.0092,
      "step": 13230
    },
    {
      "epoch": 0.216640759224413,
      "grad_norm": 0.16671772301197052,
      "learning_rate": 7.221161712571585e-06,
      "loss": 0.0069,
      "step": 13240
    },
    {
      "epoch": 0.21680438517548883,
      "grad_norm": 0.6191067099571228,
      "learning_rate": 7.226615762203438e-06,
      "loss": 0.0091,
      "step": 13250
    },
    {
      "epoch": 0.21696801112656466,
      "grad_norm": 0.5939600467681885,
      "learning_rate": 7.2320698118352885e-06,
      "loss": 0.01,
      "step": 13260
    },
    {
      "epoch": 0.21713163707764052,
      "grad_norm": 0.3789665400981903,
      "learning_rate": 7.237523861467139e-06,
      "loss": 0.0077,
      "step": 13270
    },
    {
      "epoch": 0.21729526302871635,
      "grad_norm": 0.42645567655563354,
      "learning_rate": 7.242977911098992e-06,
      "loss": 0.007,
      "step": 13280
    },
    {
      "epoch": 0.2174588889797922,
      "grad_norm": 0.3949757516384125,
      "learning_rate": 7.248431960730843e-06,
      "loss": 0.0077,
      "step": 13290
    },
    {
      "epoch": 0.21762251493086804,
      "grad_norm": 0.311788409948349,
      "learning_rate": 7.253886010362695e-06,
      "loss": 0.0076,
      "step": 13300
    },
    {
      "epoch": 0.21778614088194387,
      "grad_norm": 0.7898632287979126,
      "learning_rate": 7.259340059994546e-06,
      "loss": 0.0088,
      "step": 13310
    },
    {
      "epoch": 0.21794976683301973,
      "grad_norm": 0.5830650329589844,
      "learning_rate": 7.264794109626399e-06,
      "loss": 0.0058,
      "step": 13320
    },
    {
      "epoch": 0.21811339278409556,
      "grad_norm": 0.5990749597549438,
      "learning_rate": 7.2702481592582495e-06,
      "loss": 0.0097,
      "step": 13330
    },
    {
      "epoch": 0.21827701873517139,
      "grad_norm": 0.6458852291107178,
      "learning_rate": 7.275702208890102e-06,
      "loss": 0.0073,
      "step": 13340
    },
    {
      "epoch": 0.21844064468624724,
      "grad_norm": 0.35592833161354065,
      "learning_rate": 7.281156258521953e-06,
      "loss": 0.0103,
      "step": 13350
    },
    {
      "epoch": 0.21860427063732307,
      "grad_norm": 0.5433804988861084,
      "learning_rate": 7.2866103081538055e-06,
      "loss": 0.0145,
      "step": 13360
    },
    {
      "epoch": 0.21876789658839893,
      "grad_norm": 0.6583028435707092,
      "learning_rate": 7.292064357785656e-06,
      "loss": 0.0107,
      "step": 13370
    },
    {
      "epoch": 0.21893152253947476,
      "grad_norm": 0.34199726581573486,
      "learning_rate": 7.297518407417508e-06,
      "loss": 0.0067,
      "step": 13380
    },
    {
      "epoch": 0.2190951484905506,
      "grad_norm": 0.611171543598175,
      "learning_rate": 7.30297245704936e-06,
      "loss": 0.0085,
      "step": 13390
    },
    {
      "epoch": 0.21925877444162645,
      "grad_norm": 0.3668595850467682,
      "learning_rate": 7.308426506681211e-06,
      "loss": 0.0089,
      "step": 13400
    },
    {
      "epoch": 0.21942240039270228,
      "grad_norm": 0.6533665657043457,
      "learning_rate": 7.313880556313063e-06,
      "loss": 0.0101,
      "step": 13410
    },
    {
      "epoch": 0.21958602634377813,
      "grad_norm": 0.24867713451385498,
      "learning_rate": 7.319334605944915e-06,
      "loss": 0.0051,
      "step": 13420
    },
    {
      "epoch": 0.21974965229485396,
      "grad_norm": 0.5480397939682007,
      "learning_rate": 7.3247886555767665e-06,
      "loss": 0.0134,
      "step": 13430
    },
    {
      "epoch": 0.2199132782459298,
      "grad_norm": 0.2016737312078476,
      "learning_rate": 7.330242705208618e-06,
      "loss": 0.0109,
      "step": 13440
    },
    {
      "epoch": 0.22007690419700565,
      "grad_norm": 0.28459542989730835,
      "learning_rate": 7.33569675484047e-06,
      "loss": 0.0079,
      "step": 13450
    },
    {
      "epoch": 0.22024053014808148,
      "grad_norm": 0.3175831735134125,
      "learning_rate": 7.341150804472322e-06,
      "loss": 0.0059,
      "step": 13460
    },
    {
      "epoch": 0.22040415609915734,
      "grad_norm": 0.5060707926750183,
      "learning_rate": 7.346604854104173e-06,
      "loss": 0.0086,
      "step": 13470
    },
    {
      "epoch": 0.22056778205023317,
      "grad_norm": 0.4972872734069824,
      "learning_rate": 7.352058903736024e-06,
      "loss": 0.0069,
      "step": 13480
    },
    {
      "epoch": 0.220731408001309,
      "grad_norm": 0.3736521005630493,
      "learning_rate": 7.357512953367876e-06,
      "loss": 0.0081,
      "step": 13490
    },
    {
      "epoch": 0.22089503395238486,
      "grad_norm": 0.3874336779117584,
      "learning_rate": 7.362967002999728e-06,
      "loss": 0.0107,
      "step": 13500
    },
    {
      "epoch": 0.22105865990346069,
      "grad_norm": 0.21691761910915375,
      "learning_rate": 7.368421052631579e-06,
      "loss": 0.0049,
      "step": 13510
    },
    {
      "epoch": 0.22122228585453654,
      "grad_norm": 0.42537644505500793,
      "learning_rate": 7.373875102263431e-06,
      "loss": 0.0093,
      "step": 13520
    },
    {
      "epoch": 0.22138591180561237,
      "grad_norm": 1.0397299528121948,
      "learning_rate": 7.379329151895283e-06,
      "loss": 0.0101,
      "step": 13530
    },
    {
      "epoch": 0.2215495377566882,
      "grad_norm": 0.55242919921875,
      "learning_rate": 7.384783201527134e-06,
      "loss": 0.0086,
      "step": 13540
    },
    {
      "epoch": 0.22171316370776406,
      "grad_norm": 0.4072169065475464,
      "learning_rate": 7.390237251158986e-06,
      "loss": 0.0055,
      "step": 13550
    },
    {
      "epoch": 0.2218767896588399,
      "grad_norm": 0.198331817984581,
      "learning_rate": 7.395691300790838e-06,
      "loss": 0.0076,
      "step": 13560
    },
    {
      "epoch": 0.22204041560991572,
      "grad_norm": 0.6559755802154541,
      "learning_rate": 7.4011453504226895e-06,
      "loss": 0.0123,
      "step": 13570
    },
    {
      "epoch": 0.22220404156099158,
      "grad_norm": 0.26538601517677307,
      "learning_rate": 7.406599400054541e-06,
      "loss": 0.0069,
      "step": 13580
    },
    {
      "epoch": 0.2223676675120674,
      "grad_norm": 0.5764954686164856,
      "learning_rate": 7.412053449686393e-06,
      "loss": 0.0071,
      "step": 13590
    },
    {
      "epoch": 0.22253129346314326,
      "grad_norm": 0.7161535024642944,
      "learning_rate": 7.417507499318244e-06,
      "loss": 0.0089,
      "step": 13600
    },
    {
      "epoch": 0.2226949194142191,
      "grad_norm": 0.49198058247566223,
      "learning_rate": 7.422961548950096e-06,
      "loss": 0.0089,
      "step": 13610
    },
    {
      "epoch": 0.22285854536529492,
      "grad_norm": 0.2610313594341278,
      "learning_rate": 7.428415598581947e-06,
      "loss": 0.0092,
      "step": 13620
    },
    {
      "epoch": 0.22302217131637078,
      "grad_norm": 0.35409772396087646,
      "learning_rate": 7.4338696482138e-06,
      "loss": 0.0081,
      "step": 13630
    },
    {
      "epoch": 0.2231857972674466,
      "grad_norm": 0.5974505543708801,
      "learning_rate": 7.4393236978456505e-06,
      "loss": 0.0067,
      "step": 13640
    },
    {
      "epoch": 0.22334942321852247,
      "grad_norm": 0.29375317692756653,
      "learning_rate": 7.444777747477503e-06,
      "loss": 0.005,
      "step": 13650
    },
    {
      "epoch": 0.2235130491695983,
      "grad_norm": 0.35343772172927856,
      "learning_rate": 7.450231797109354e-06,
      "loss": 0.0075,
      "step": 13660
    },
    {
      "epoch": 0.22367667512067413,
      "grad_norm": 0.4877508878707886,
      "learning_rate": 7.4556858467412065e-06,
      "loss": 0.007,
      "step": 13670
    },
    {
      "epoch": 0.22384030107175,
      "grad_norm": 0.38897445797920227,
      "learning_rate": 7.461139896373057e-06,
      "loss": 0.0088,
      "step": 13680
    },
    {
      "epoch": 0.22400392702282582,
      "grad_norm": 0.4827202260494232,
      "learning_rate": 7.46659394600491e-06,
      "loss": 0.0064,
      "step": 13690
    },
    {
      "epoch": 0.22416755297390167,
      "grad_norm": 0.4307715594768524,
      "learning_rate": 7.472047995636761e-06,
      "loss": 0.0086,
      "step": 13700
    },
    {
      "epoch": 0.2243311789249775,
      "grad_norm": 0.4649055302143097,
      "learning_rate": 7.477502045268612e-06,
      "loss": 0.0056,
      "step": 13710
    },
    {
      "epoch": 0.22449480487605333,
      "grad_norm": 0.565535306930542,
      "learning_rate": 7.482956094900464e-06,
      "loss": 0.0063,
      "step": 13720
    },
    {
      "epoch": 0.2246584308271292,
      "grad_norm": 0.3925972878932953,
      "learning_rate": 7.488410144532315e-06,
      "loss": 0.0097,
      "step": 13730
    },
    {
      "epoch": 0.22482205677820502,
      "grad_norm": 0.38579362630844116,
      "learning_rate": 7.4938641941641676e-06,
      "loss": 0.0086,
      "step": 13740
    },
    {
      "epoch": 0.22498568272928085,
      "grad_norm": 0.34197136759757996,
      "learning_rate": 7.499318243796018e-06,
      "loss": 0.007,
      "step": 13750
    },
    {
      "epoch": 0.2251493086803567,
      "grad_norm": 0.42359283566474915,
      "learning_rate": 7.504772293427871e-06,
      "loss": 0.0071,
      "step": 13760
    },
    {
      "epoch": 0.22531293463143254,
      "grad_norm": 0.5756303071975708,
      "learning_rate": 7.510226343059722e-06,
      "loss": 0.0147,
      "step": 13770
    },
    {
      "epoch": 0.2254765605825084,
      "grad_norm": 0.5594773888587952,
      "learning_rate": 7.515680392691574e-06,
      "loss": 0.0072,
      "step": 13780
    },
    {
      "epoch": 0.22564018653358422,
      "grad_norm": 0.5141112804412842,
      "learning_rate": 7.521134442323425e-06,
      "loss": 0.0093,
      "step": 13790
    },
    {
      "epoch": 0.22580381248466005,
      "grad_norm": 0.524696409702301,
      "learning_rate": 7.526588491955278e-06,
      "loss": 0.0083,
      "step": 13800
    },
    {
      "epoch": 0.2259674384357359,
      "grad_norm": 0.2597118020057678,
      "learning_rate": 7.532042541587129e-06,
      "loss": 0.0073,
      "step": 13810
    },
    {
      "epoch": 0.22613106438681174,
      "grad_norm": 0.43524059653282166,
      "learning_rate": 7.537496591218981e-06,
      "loss": 0.0092,
      "step": 13820
    },
    {
      "epoch": 0.2262946903378876,
      "grad_norm": 0.38005179166793823,
      "learning_rate": 7.542950640850832e-06,
      "loss": 0.0079,
      "step": 13830
    },
    {
      "epoch": 0.22645831628896343,
      "grad_norm": 0.33128640055656433,
      "learning_rate": 7.548404690482684e-06,
      "loss": 0.0106,
      "step": 13840
    },
    {
      "epoch": 0.22662194224003926,
      "grad_norm": 0.7541969418525696,
      "learning_rate": 7.553858740114535e-06,
      "loss": 0.0093,
      "step": 13850
    },
    {
      "epoch": 0.22678556819111512,
      "grad_norm": 0.5331867933273315,
      "learning_rate": 7.559312789746387e-06,
      "loss": 0.0077,
      "step": 13860
    },
    {
      "epoch": 0.22694919414219095,
      "grad_norm": 0.47539660334587097,
      "learning_rate": 7.564766839378239e-06,
      "loss": 0.0108,
      "step": 13870
    },
    {
      "epoch": 0.2271128200932668,
      "grad_norm": 0.40205758810043335,
      "learning_rate": 7.5702208890100905e-06,
      "loss": 0.0098,
      "step": 13880
    },
    {
      "epoch": 0.22727644604434263,
      "grad_norm": 0.25452202558517456,
      "learning_rate": 7.575674938641942e-06,
      "loss": 0.0083,
      "step": 13890
    },
    {
      "epoch": 0.22744007199541846,
      "grad_norm": 0.24174167215824127,
      "learning_rate": 7.581128988273794e-06,
      "loss": 0.0095,
      "step": 13900
    },
    {
      "epoch": 0.22760369794649432,
      "grad_norm": 0.35427331924438477,
      "learning_rate": 7.586583037905646e-06,
      "loss": 0.009,
      "step": 13910
    },
    {
      "epoch": 0.22776732389757015,
      "grad_norm": 0.3100494146347046,
      "learning_rate": 7.592037087537497e-06,
      "loss": 0.0079,
      "step": 13920
    },
    {
      "epoch": 0.227930949848646,
      "grad_norm": 0.3048965632915497,
      "learning_rate": 7.597491137169349e-06,
      "loss": 0.0096,
      "step": 13930
    },
    {
      "epoch": 0.22809457579972184,
      "grad_norm": 0.21039795875549316,
      "learning_rate": 7.602945186801201e-06,
      "loss": 0.0065,
      "step": 13940
    },
    {
      "epoch": 0.22825820175079767,
      "grad_norm": 0.355277955532074,
      "learning_rate": 7.608399236433052e-06,
      "loss": 0.0057,
      "step": 13950
    },
    {
      "epoch": 0.22842182770187353,
      "grad_norm": 0.47577816247940063,
      "learning_rate": 7.613853286064904e-06,
      "loss": 0.0068,
      "step": 13960
    },
    {
      "epoch": 0.22858545365294936,
      "grad_norm": 0.5222656726837158,
      "learning_rate": 7.619307335696755e-06,
      "loss": 0.0076,
      "step": 13970
    },
    {
      "epoch": 0.22874907960402519,
      "grad_norm": 0.8510044813156128,
      "learning_rate": 7.6247613853286075e-06,
      "loss": 0.0079,
      "step": 13980
    },
    {
      "epoch": 0.22891270555510104,
      "grad_norm": 0.3962283432483673,
      "learning_rate": 7.63021543496046e-06,
      "loss": 0.0096,
      "step": 13990
    },
    {
      "epoch": 0.22907633150617687,
      "grad_norm": 0.23457308113574982,
      "learning_rate": 7.63566948459231e-06,
      "loss": 0.0086,
      "step": 14000
    },
    {
      "epoch": 0.22923995745725273,
      "grad_norm": 0.7131142020225525,
      "learning_rate": 7.641123534224163e-06,
      "loss": 0.0075,
      "step": 14010
    },
    {
      "epoch": 0.22940358340832856,
      "grad_norm": 0.6043888330459595,
      "learning_rate": 7.646577583856013e-06,
      "loss": 0.0113,
      "step": 14020
    },
    {
      "epoch": 0.2295672093594044,
      "grad_norm": 0.40674737095832825,
      "learning_rate": 7.652031633487866e-06,
      "loss": 0.0045,
      "step": 14030
    },
    {
      "epoch": 0.22973083531048025,
      "grad_norm": 0.49257761240005493,
      "learning_rate": 7.657485683119717e-06,
      "loss": 0.009,
      "step": 14040
    },
    {
      "epoch": 0.22989446126155608,
      "grad_norm": 0.5192036628723145,
      "learning_rate": 7.66293973275157e-06,
      "loss": 0.0083,
      "step": 14050
    },
    {
      "epoch": 0.23005808721263193,
      "grad_norm": 0.5333401560783386,
      "learning_rate": 7.66839378238342e-06,
      "loss": 0.0081,
      "step": 14060
    },
    {
      "epoch": 0.23022171316370776,
      "grad_norm": 0.4272986650466919,
      "learning_rate": 7.673847832015273e-06,
      "loss": 0.007,
      "step": 14070
    },
    {
      "epoch": 0.2303853391147836,
      "grad_norm": 0.807560920715332,
      "learning_rate": 7.679301881647124e-06,
      "loss": 0.0139,
      "step": 14080
    },
    {
      "epoch": 0.23054896506585945,
      "grad_norm": 0.4795917570590973,
      "learning_rate": 7.684755931278976e-06,
      "loss": 0.0072,
      "step": 14090
    },
    {
      "epoch": 0.23071259101693528,
      "grad_norm": 0.6201799511909485,
      "learning_rate": 7.690209980910827e-06,
      "loss": 0.0078,
      "step": 14100
    },
    {
      "epoch": 0.23087621696801114,
      "grad_norm": 0.5711755156517029,
      "learning_rate": 7.69566403054268e-06,
      "loss": 0.0053,
      "step": 14110
    },
    {
      "epoch": 0.23103984291908697,
      "grad_norm": 0.50462406873703,
      "learning_rate": 7.70111808017453e-06,
      "loss": 0.0078,
      "step": 14120
    },
    {
      "epoch": 0.2312034688701628,
      "grad_norm": 0.6455571055412292,
      "learning_rate": 7.706572129806383e-06,
      "loss": 0.0101,
      "step": 14130
    },
    {
      "epoch": 0.23136709482123866,
      "grad_norm": 0.2799656093120575,
      "learning_rate": 7.712026179438234e-06,
      "loss": 0.0088,
      "step": 14140
    },
    {
      "epoch": 0.23153072077231449,
      "grad_norm": 0.13593356311321259,
      "learning_rate": 7.717480229070086e-06,
      "loss": 0.0103,
      "step": 14150
    },
    {
      "epoch": 0.23169434672339034,
      "grad_norm": 0.3700540065765381,
      "learning_rate": 7.722934278701937e-06,
      "loss": 0.0065,
      "step": 14160
    },
    {
      "epoch": 0.23185797267446617,
      "grad_norm": 0.3780127465724945,
      "learning_rate": 7.728388328333788e-06,
      "loss": 0.0071,
      "step": 14170
    },
    {
      "epoch": 0.232021598625542,
      "grad_norm": 0.779640257358551,
      "learning_rate": 7.73384237796564e-06,
      "loss": 0.0114,
      "step": 14180
    },
    {
      "epoch": 0.23218522457661786,
      "grad_norm": 0.793234646320343,
      "learning_rate": 7.739296427597492e-06,
      "loss": 0.0098,
      "step": 14190
    },
    {
      "epoch": 0.2323488505276937,
      "grad_norm": 0.49452006816864014,
      "learning_rate": 7.744750477229344e-06,
      "loss": 0.0088,
      "step": 14200
    },
    {
      "epoch": 0.23251247647876952,
      "grad_norm": 0.4760637581348419,
      "learning_rate": 7.750204526861195e-06,
      "loss": 0.0091,
      "step": 14210
    },
    {
      "epoch": 0.23267610242984538,
      "grad_norm": 0.5735825300216675,
      "learning_rate": 7.755658576493048e-06,
      "loss": 0.0085,
      "step": 14220
    },
    {
      "epoch": 0.2328397283809212,
      "grad_norm": 0.4971274733543396,
      "learning_rate": 7.761112626124898e-06,
      "loss": 0.0112,
      "step": 14230
    },
    {
      "epoch": 0.23300335433199706,
      "grad_norm": 0.7157065868377686,
      "learning_rate": 7.766566675756751e-06,
      "loss": 0.0076,
      "step": 14240
    },
    {
      "epoch": 0.2331669802830729,
      "grad_norm": 0.34946438670158386,
      "learning_rate": 7.772020725388602e-06,
      "loss": 0.0095,
      "step": 14250
    },
    {
      "epoch": 0.23333060623414872,
      "grad_norm": 0.6798130869865417,
      "learning_rate": 7.777474775020454e-06,
      "loss": 0.0071,
      "step": 14260
    },
    {
      "epoch": 0.23349423218522458,
      "grad_norm": 0.275322288274765,
      "learning_rate": 7.782928824652305e-06,
      "loss": 0.0096,
      "step": 14270
    },
    {
      "epoch": 0.2336578581363004,
      "grad_norm": 0.4094807207584381,
      "learning_rate": 7.788382874284156e-06,
      "loss": 0.0062,
      "step": 14280
    },
    {
      "epoch": 0.23382148408737627,
      "grad_norm": 0.33121684193611145,
      "learning_rate": 7.793836923916009e-06,
      "loss": 0.0065,
      "step": 14290
    },
    {
      "epoch": 0.2339851100384521,
      "grad_norm": 0.6018854975700378,
      "learning_rate": 7.79929097354786e-06,
      "loss": 0.0095,
      "step": 14300
    },
    {
      "epoch": 0.23414873598952793,
      "grad_norm": 0.51739901304245,
      "learning_rate": 7.804745023179712e-06,
      "loss": 0.0074,
      "step": 14310
    },
    {
      "epoch": 0.23431236194060379,
      "grad_norm": 0.8751764297485352,
      "learning_rate": 7.810199072811563e-06,
      "loss": 0.013,
      "step": 14320
    },
    {
      "epoch": 0.23447598789167962,
      "grad_norm": 0.36460477113723755,
      "learning_rate": 7.815653122443415e-06,
      "loss": 0.0085,
      "step": 14330
    },
    {
      "epoch": 0.23463961384275547,
      "grad_norm": 0.410590261220932,
      "learning_rate": 7.821107172075266e-06,
      "loss": 0.0077,
      "step": 14340
    },
    {
      "epoch": 0.2348032397938313,
      "grad_norm": 0.2005511224269867,
      "learning_rate": 7.826561221707119e-06,
      "loss": 0.0105,
      "step": 14350
    },
    {
      "epoch": 0.23496686574490713,
      "grad_norm": 0.486274391412735,
      "learning_rate": 7.83201527133897e-06,
      "loss": 0.0104,
      "step": 14360
    },
    {
      "epoch": 0.235130491695983,
      "grad_norm": 0.3867475986480713,
      "learning_rate": 7.837469320970822e-06,
      "loss": 0.0082,
      "step": 14370
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.38502442836761475,
      "learning_rate": 7.842923370602673e-06,
      "loss": 0.0099,
      "step": 14380
    },
    {
      "epoch": 0.23545774359813468,
      "grad_norm": 0.29765263199806213,
      "learning_rate": 7.848377420234526e-06,
      "loss": 0.0112,
      "step": 14390
    },
    {
      "epoch": 0.2356213695492105,
      "grad_norm": 0.5395899415016174,
      "learning_rate": 7.853831469866376e-06,
      "loss": 0.0108,
      "step": 14400
    },
    {
      "epoch": 0.23578499550028634,
      "grad_norm": 0.5860231518745422,
      "learning_rate": 7.859285519498227e-06,
      "loss": 0.0086,
      "step": 14410
    },
    {
      "epoch": 0.2359486214513622,
      "grad_norm": 0.38418200612068176,
      "learning_rate": 7.86473956913008e-06,
      "loss": 0.0062,
      "step": 14420
    },
    {
      "epoch": 0.23611224740243802,
      "grad_norm": 0.5440731048583984,
      "learning_rate": 7.87019361876193e-06,
      "loss": 0.0088,
      "step": 14430
    },
    {
      "epoch": 0.23627587335351385,
      "grad_norm": 0.43382129073143005,
      "learning_rate": 7.875647668393783e-06,
      "loss": 0.0082,
      "step": 14440
    },
    {
      "epoch": 0.2364394993045897,
      "grad_norm": 0.2749595046043396,
      "learning_rate": 7.881101718025634e-06,
      "loss": 0.0111,
      "step": 14450
    },
    {
      "epoch": 0.23660312525566554,
      "grad_norm": 0.1973612755537033,
      "learning_rate": 7.886555767657487e-06,
      "loss": 0.006,
      "step": 14460
    },
    {
      "epoch": 0.2367667512067414,
      "grad_norm": 0.913813054561615,
      "learning_rate": 7.892009817289337e-06,
      "loss": 0.0076,
      "step": 14470
    },
    {
      "epoch": 0.23693037715781723,
      "grad_norm": 0.4277299642562866,
      "learning_rate": 7.89746386692119e-06,
      "loss": 0.0061,
      "step": 14480
    },
    {
      "epoch": 0.23709400310889306,
      "grad_norm": 0.62294602394104,
      "learning_rate": 7.902917916553041e-06,
      "loss": 0.0062,
      "step": 14490
    },
    {
      "epoch": 0.23725762905996892,
      "grad_norm": 0.2680823504924774,
      "learning_rate": 7.908371966184893e-06,
      "loss": 0.0051,
      "step": 14500
    },
    {
      "epoch": 0.23742125501104475,
      "grad_norm": 0.11393973231315613,
      "learning_rate": 7.913826015816744e-06,
      "loss": 0.0068,
      "step": 14510
    },
    {
      "epoch": 0.2375848809621206,
      "grad_norm": 0.36629122495651245,
      "learning_rate": 7.919280065448595e-06,
      "loss": 0.0067,
      "step": 14520
    },
    {
      "epoch": 0.23774850691319643,
      "grad_norm": 0.4813237190246582,
      "learning_rate": 7.924734115080448e-06,
      "loss": 0.0106,
      "step": 14530
    },
    {
      "epoch": 0.23791213286427226,
      "grad_norm": 0.46408146619796753,
      "learning_rate": 7.930188164712299e-06,
      "loss": 0.0056,
      "step": 14540
    },
    {
      "epoch": 0.23807575881534812,
      "grad_norm": 0.615135133266449,
      "learning_rate": 7.935642214344151e-06,
      "loss": 0.0073,
      "step": 14550
    },
    {
      "epoch": 0.23823938476642395,
      "grad_norm": 0.5712845325469971,
      "learning_rate": 7.941096263976002e-06,
      "loss": 0.0061,
      "step": 14560
    },
    {
      "epoch": 0.2384030107174998,
      "grad_norm": 0.36727845668792725,
      "learning_rate": 7.946550313607854e-06,
      "loss": 0.0066,
      "step": 14570
    },
    {
      "epoch": 0.23856663666857564,
      "grad_norm": 0.2183261513710022,
      "learning_rate": 7.952004363239705e-06,
      "loss": 0.0072,
      "step": 14580
    },
    {
      "epoch": 0.23873026261965147,
      "grad_norm": 0.4431883990764618,
      "learning_rate": 7.957458412871558e-06,
      "loss": 0.0109,
      "step": 14590
    },
    {
      "epoch": 0.23889388857072733,
      "grad_norm": 0.1619562953710556,
      "learning_rate": 7.962912462503409e-06,
      "loss": 0.0051,
      "step": 14600
    },
    {
      "epoch": 0.23905751452180315,
      "grad_norm": 0.6055853962898254,
      "learning_rate": 7.968366512135261e-06,
      "loss": 0.0103,
      "step": 14610
    },
    {
      "epoch": 0.23922114047287898,
      "grad_norm": 0.6191697120666504,
      "learning_rate": 7.973820561767112e-06,
      "loss": 0.0072,
      "step": 14620
    },
    {
      "epoch": 0.23938476642395484,
      "grad_norm": 0.3025716245174408,
      "learning_rate": 7.979274611398965e-06,
      "loss": 0.0066,
      "step": 14630
    },
    {
      "epoch": 0.23954839237503067,
      "grad_norm": 0.5540299415588379,
      "learning_rate": 7.984728661030816e-06,
      "loss": 0.0076,
      "step": 14640
    },
    {
      "epoch": 0.23971201832610653,
      "grad_norm": 0.7998112440109253,
      "learning_rate": 7.990182710662668e-06,
      "loss": 0.0078,
      "step": 14650
    },
    {
      "epoch": 0.23987564427718236,
      "grad_norm": 0.31662535667419434,
      "learning_rate": 7.995636760294519e-06,
      "loss": 0.0105,
      "step": 14660
    },
    {
      "epoch": 0.2400392702282582,
      "grad_norm": 0.3014910817146301,
      "learning_rate": 8.001090809926372e-06,
      "loss": 0.0086,
      "step": 14670
    },
    {
      "epoch": 0.24020289617933405,
      "grad_norm": 0.4394930303096771,
      "learning_rate": 8.006544859558222e-06,
      "loss": 0.0061,
      "step": 14680
    },
    {
      "epoch": 0.24036652213040988,
      "grad_norm": 0.19743195176124573,
      "learning_rate": 8.011998909190075e-06,
      "loss": 0.0043,
      "step": 14690
    },
    {
      "epoch": 0.24053014808148573,
      "grad_norm": 0.09330590814352036,
      "learning_rate": 8.017452958821926e-06,
      "loss": 0.006,
      "step": 14700
    },
    {
      "epoch": 0.24069377403256156,
      "grad_norm": 0.7159872651100159,
      "learning_rate": 8.022907008453777e-06,
      "loss": 0.0095,
      "step": 14710
    },
    {
      "epoch": 0.2408573999836374,
      "grad_norm": 0.6109306812286377,
      "learning_rate": 8.02836105808563e-06,
      "loss": 0.0066,
      "step": 14720
    },
    {
      "epoch": 0.24102102593471325,
      "grad_norm": 0.1929505616426468,
      "learning_rate": 8.03381510771748e-06,
      "loss": 0.0078,
      "step": 14730
    },
    {
      "epoch": 0.24118465188578908,
      "grad_norm": 0.23274843394756317,
      "learning_rate": 8.039269157349333e-06,
      "loss": 0.0065,
      "step": 14740
    },
    {
      "epoch": 0.24134827783686494,
      "grad_norm": 0.5227785706520081,
      "learning_rate": 8.044723206981183e-06,
      "loss": 0.0085,
      "step": 14750
    },
    {
      "epoch": 0.24151190378794077,
      "grad_norm": 1.1571177244186401,
      "learning_rate": 8.050177256613036e-06,
      "loss": 0.0095,
      "step": 14760
    },
    {
      "epoch": 0.2416755297390166,
      "grad_norm": 0.34047138690948486,
      "learning_rate": 8.055631306244887e-06,
      "loss": 0.008,
      "step": 14770
    },
    {
      "epoch": 0.24183915569009246,
      "grad_norm": 0.25412407517433167,
      "learning_rate": 8.06108535587674e-06,
      "loss": 0.0061,
      "step": 14780
    },
    {
      "epoch": 0.24200278164116829,
      "grad_norm": 0.3473801612854004,
      "learning_rate": 8.06653940550859e-06,
      "loss": 0.0052,
      "step": 14790
    },
    {
      "epoch": 0.24216640759224414,
      "grad_norm": 0.22174876928329468,
      "learning_rate": 8.071993455140443e-06,
      "loss": 0.0097,
      "step": 14800
    },
    {
      "epoch": 0.24233003354331997,
      "grad_norm": 0.6153866052627563,
      "learning_rate": 8.077447504772294e-06,
      "loss": 0.0079,
      "step": 14810
    },
    {
      "epoch": 0.2424936594943958,
      "grad_norm": 0.481741726398468,
      "learning_rate": 8.082901554404146e-06,
      "loss": 0.0086,
      "step": 14820
    },
    {
      "epoch": 0.24265728544547166,
      "grad_norm": 0.42188334465026855,
      "learning_rate": 8.088355604035997e-06,
      "loss": 0.0113,
      "step": 14830
    },
    {
      "epoch": 0.2428209113965475,
      "grad_norm": 0.3054935932159424,
      "learning_rate": 8.09380965366785e-06,
      "loss": 0.0082,
      "step": 14840
    },
    {
      "epoch": 0.24298453734762332,
      "grad_norm": 0.23599357903003693,
      "learning_rate": 8.0992637032997e-06,
      "loss": 0.0078,
      "step": 14850
    },
    {
      "epoch": 0.24314816329869918,
      "grad_norm": 0.38109397888183594,
      "learning_rate": 8.104717752931553e-06,
      "loss": 0.0092,
      "step": 14860
    },
    {
      "epoch": 0.243311789249775,
      "grad_norm": 0.4762151837348938,
      "learning_rate": 8.110171802563404e-06,
      "loss": 0.006,
      "step": 14870
    },
    {
      "epoch": 0.24347541520085086,
      "grad_norm": 0.5742477178573608,
      "learning_rate": 8.115625852195256e-06,
      "loss": 0.007,
      "step": 14880
    },
    {
      "epoch": 0.2436390411519267,
      "grad_norm": 0.34737327694892883,
      "learning_rate": 8.121079901827107e-06,
      "loss": 0.0059,
      "step": 14890
    },
    {
      "epoch": 0.24380266710300252,
      "grad_norm": 0.27185627818107605,
      "learning_rate": 8.12653395145896e-06,
      "loss": 0.0051,
      "step": 14900
    },
    {
      "epoch": 0.24396629305407838,
      "grad_norm": 0.2783239781856537,
      "learning_rate": 8.13198800109081e-06,
      "loss": 0.0112,
      "step": 14910
    },
    {
      "epoch": 0.2441299190051542,
      "grad_norm": 0.2922504246234894,
      "learning_rate": 8.137442050722663e-06,
      "loss": 0.005,
      "step": 14920
    },
    {
      "epoch": 0.24429354495623007,
      "grad_norm": 0.386493980884552,
      "learning_rate": 8.142896100354514e-06,
      "loss": 0.0104,
      "step": 14930
    },
    {
      "epoch": 0.2444571709073059,
      "grad_norm": 0.16024218499660492,
      "learning_rate": 8.148350149986367e-06,
      "loss": 0.0054,
      "step": 14940
    },
    {
      "epoch": 0.24462079685838173,
      "grad_norm": 0.43841415643692017,
      "learning_rate": 8.153804199618217e-06,
      "loss": 0.0065,
      "step": 14950
    },
    {
      "epoch": 0.24478442280945759,
      "grad_norm": 0.3885124623775482,
      "learning_rate": 8.15925824925007e-06,
      "loss": 0.0073,
      "step": 14960
    },
    {
      "epoch": 0.24494804876053342,
      "grad_norm": 0.42438313364982605,
      "learning_rate": 8.16471229888192e-06,
      "loss": 0.0094,
      "step": 14970
    },
    {
      "epoch": 0.24511167471160927,
      "grad_norm": 0.5339972972869873,
      "learning_rate": 8.170166348513772e-06,
      "loss": 0.0104,
      "step": 14980
    },
    {
      "epoch": 0.2452753006626851,
      "grad_norm": 0.24884118139743805,
      "learning_rate": 8.175620398145624e-06,
      "loss": 0.0076,
      "step": 14990
    },
    {
      "epoch": 0.24543892661376093,
      "grad_norm": 0.4015233814716339,
      "learning_rate": 8.181074447777475e-06,
      "loss": 0.009,
      "step": 15000
    },
    {
      "epoch": 0.2456025525648368,
      "grad_norm": 0.4313192069530487,
      "learning_rate": 8.186528497409328e-06,
      "loss": 0.0055,
      "step": 15010
    },
    {
      "epoch": 0.24576617851591262,
      "grad_norm": 0.4145377576351166,
      "learning_rate": 8.191982547041178e-06,
      "loss": 0.0083,
      "step": 15020
    },
    {
      "epoch": 0.24592980446698848,
      "grad_norm": 0.3716921806335449,
      "learning_rate": 8.197436596673031e-06,
      "loss": 0.0084,
      "step": 15030
    },
    {
      "epoch": 0.2460934304180643,
      "grad_norm": 0.41910916566848755,
      "learning_rate": 8.202890646304882e-06,
      "loss": 0.0109,
      "step": 15040
    },
    {
      "epoch": 0.24625705636914014,
      "grad_norm": 0.24850162863731384,
      "learning_rate": 8.208344695936734e-06,
      "loss": 0.0096,
      "step": 15050
    },
    {
      "epoch": 0.246420682320216,
      "grad_norm": 0.4973500072956085,
      "learning_rate": 8.213798745568585e-06,
      "loss": 0.0076,
      "step": 15060
    },
    {
      "epoch": 0.24658430827129182,
      "grad_norm": 0.16750168800354004,
      "learning_rate": 8.219252795200438e-06,
      "loss": 0.0071,
      "step": 15070
    },
    {
      "epoch": 0.24674793422236765,
      "grad_norm": 0.514723002910614,
      "learning_rate": 8.224706844832289e-06,
      "loss": 0.0101,
      "step": 15080
    },
    {
      "epoch": 0.2469115601734435,
      "grad_norm": 0.16123224794864655,
      "learning_rate": 8.23016089446414e-06,
      "loss": 0.0057,
      "step": 15090
    },
    {
      "epoch": 0.24707518612451934,
      "grad_norm": 0.5143868327140808,
      "learning_rate": 8.235614944095992e-06,
      "loss": 0.0064,
      "step": 15100
    },
    {
      "epoch": 0.2472388120755952,
      "grad_norm": 0.3330678343772888,
      "learning_rate": 8.241068993727843e-06,
      "loss": 0.0097,
      "step": 15110
    },
    {
      "epoch": 0.24740243802667103,
      "grad_norm": 0.3562721610069275,
      "learning_rate": 8.246523043359696e-06,
      "loss": 0.0065,
      "step": 15120
    },
    {
      "epoch": 0.24756606397774686,
      "grad_norm": 0.360736221075058,
      "learning_rate": 8.251977092991546e-06,
      "loss": 0.0067,
      "step": 15130
    },
    {
      "epoch": 0.24772968992882272,
      "grad_norm": 0.304109126329422,
      "learning_rate": 8.257431142623399e-06,
      "loss": 0.0072,
      "step": 15140
    },
    {
      "epoch": 0.24789331587989855,
      "grad_norm": 0.09711117297410965,
      "learning_rate": 8.26288519225525e-06,
      "loss": 0.0056,
      "step": 15150
    },
    {
      "epoch": 0.2480569418309744,
      "grad_norm": 0.31720617413520813,
      "learning_rate": 8.268339241887102e-06,
      "loss": 0.0066,
      "step": 15160
    },
    {
      "epoch": 0.24822056778205023,
      "grad_norm": 0.3169492781162262,
      "learning_rate": 8.273793291518953e-06,
      "loss": 0.0065,
      "step": 15170
    },
    {
      "epoch": 0.24838419373312606,
      "grad_norm": 0.5174160003662109,
      "learning_rate": 8.279247341150806e-06,
      "loss": 0.0083,
      "step": 15180
    },
    {
      "epoch": 0.24854781968420192,
      "grad_norm": 0.603222131729126,
      "learning_rate": 8.284701390782657e-06,
      "loss": 0.0088,
      "step": 15190
    },
    {
      "epoch": 0.24871144563527775,
      "grad_norm": 0.48851442337036133,
      "learning_rate": 8.290155440414507e-06,
      "loss": 0.0071,
      "step": 15200
    },
    {
      "epoch": 0.2488750715863536,
      "grad_norm": 0.14064426720142365,
      "learning_rate": 8.29560949004636e-06,
      "loss": 0.0038,
      "step": 15210
    },
    {
      "epoch": 0.24903869753742944,
      "grad_norm": 0.37737977504730225,
      "learning_rate": 8.30106353967821e-06,
      "loss": 0.0072,
      "step": 15220
    },
    {
      "epoch": 0.24920232348850527,
      "grad_norm": 0.25355586409568787,
      "learning_rate": 8.306517589310063e-06,
      "loss": 0.0096,
      "step": 15230
    },
    {
      "epoch": 0.24936594943958112,
      "grad_norm": 0.47630444169044495,
      "learning_rate": 8.311971638941914e-06,
      "loss": 0.0076,
      "step": 15240
    },
    {
      "epoch": 0.24952957539065695,
      "grad_norm": 0.6583170890808105,
      "learning_rate": 8.317425688573767e-06,
      "loss": 0.0064,
      "step": 15250
    },
    {
      "epoch": 0.2496932013417328,
      "grad_norm": 0.6292118430137634,
      "learning_rate": 8.322879738205618e-06,
      "loss": 0.0088,
      "step": 15260
    },
    {
      "epoch": 0.24985682729280864,
      "grad_norm": 0.13277125358581543,
      "learning_rate": 8.32833378783747e-06,
      "loss": 0.0071,
      "step": 15270
    },
    {
      "epoch": 0.2500204532438845,
      "grad_norm": 0.4996962547302246,
      "learning_rate": 8.333787837469321e-06,
      "loss": 0.0078,
      "step": 15280
    },
    {
      "epoch": 0.25018407919496033,
      "grad_norm": 0.555380642414093,
      "learning_rate": 8.339241887101174e-06,
      "loss": 0.0067,
      "step": 15290
    },
    {
      "epoch": 0.25034770514603616,
      "grad_norm": 0.7256217002868652,
      "learning_rate": 8.344695936733024e-06,
      "loss": 0.0081,
      "step": 15300
    },
    {
      "epoch": 0.250511331097112,
      "grad_norm": 0.3761509358882904,
      "learning_rate": 8.350149986364875e-06,
      "loss": 0.0067,
      "step": 15310
    },
    {
      "epoch": 0.2506749570481878,
      "grad_norm": 0.36821871995925903,
      "learning_rate": 8.355604035996728e-06,
      "loss": 0.0075,
      "step": 15320
    },
    {
      "epoch": 0.2508385829992637,
      "grad_norm": 0.20835179090499878,
      "learning_rate": 8.361058085628579e-06,
      "loss": 0.0071,
      "step": 15330
    },
    {
      "epoch": 0.25100220895033953,
      "grad_norm": 0.4317692816257477,
      "learning_rate": 8.366512135260431e-06,
      "loss": 0.0069,
      "step": 15340
    },
    {
      "epoch": 0.25116583490141536,
      "grad_norm": 0.7908754944801331,
      "learning_rate": 8.371966184892282e-06,
      "loss": 0.0084,
      "step": 15350
    },
    {
      "epoch": 0.2513294608524912,
      "grad_norm": 0.5661348700523376,
      "learning_rate": 8.377420234524135e-06,
      "loss": 0.0113,
      "step": 15360
    },
    {
      "epoch": 0.251493086803567,
      "grad_norm": 0.5994781255722046,
      "learning_rate": 8.382874284155985e-06,
      "loss": 0.0066,
      "step": 15370
    },
    {
      "epoch": 0.2516567127546429,
      "grad_norm": 0.13631018996238708,
      "learning_rate": 8.388328333787838e-06,
      "loss": 0.0068,
      "step": 15380
    },
    {
      "epoch": 0.25182033870571874,
      "grad_norm": 0.6861046552658081,
      "learning_rate": 8.393782383419689e-06,
      "loss": 0.0082,
      "step": 15390
    },
    {
      "epoch": 0.25198396465679457,
      "grad_norm": 0.22847877442836761,
      "learning_rate": 8.399236433051541e-06,
      "loss": 0.0085,
      "step": 15400
    },
    {
      "epoch": 0.2521475906078704,
      "grad_norm": 0.36022627353668213,
      "learning_rate": 8.404690482683392e-06,
      "loss": 0.0064,
      "step": 15410
    },
    {
      "epoch": 0.2523112165589462,
      "grad_norm": 0.3559449315071106,
      "learning_rate": 8.410144532315245e-06,
      "loss": 0.009,
      "step": 15420
    },
    {
      "epoch": 0.2524748425100221,
      "grad_norm": 0.9223658442497253,
      "learning_rate": 8.415598581947096e-06,
      "loss": 0.0062,
      "step": 15430
    },
    {
      "epoch": 0.25263846846109794,
      "grad_norm": 0.4064810276031494,
      "learning_rate": 8.421052631578948e-06,
      "loss": 0.0083,
      "step": 15440
    },
    {
      "epoch": 0.2528020944121738,
      "grad_norm": 0.5033938884735107,
      "learning_rate": 8.426506681210799e-06,
      "loss": 0.0061,
      "step": 15450
    },
    {
      "epoch": 0.2529657203632496,
      "grad_norm": 0.1993841975927353,
      "learning_rate": 8.431960730842652e-06,
      "loss": 0.0112,
      "step": 15460
    },
    {
      "epoch": 0.25312934631432543,
      "grad_norm": 0.8136267066001892,
      "learning_rate": 8.437414780474502e-06,
      "loss": 0.0042,
      "step": 15470
    },
    {
      "epoch": 0.2532929722654013,
      "grad_norm": 0.45578423142433167,
      "learning_rate": 8.442868830106355e-06,
      "loss": 0.0094,
      "step": 15480
    },
    {
      "epoch": 0.25345659821647715,
      "grad_norm": 0.7431492805480957,
      "learning_rate": 8.448322879738206e-06,
      "loss": 0.0117,
      "step": 15490
    },
    {
      "epoch": 0.253620224167553,
      "grad_norm": 0.4185831844806671,
      "learning_rate": 8.453776929370058e-06,
      "loss": 0.0123,
      "step": 15500
    },
    {
      "epoch": 0.2537838501186288,
      "grad_norm": 0.10859210789203644,
      "learning_rate": 8.45923097900191e-06,
      "loss": 0.0085,
      "step": 15510
    },
    {
      "epoch": 0.25394747606970464,
      "grad_norm": 0.31318360567092896,
      "learning_rate": 8.464685028633762e-06,
      "loss": 0.0058,
      "step": 15520
    },
    {
      "epoch": 0.2541111020207805,
      "grad_norm": 0.4922339618206024,
      "learning_rate": 8.470139078265613e-06,
      "loss": 0.0064,
      "step": 15530
    },
    {
      "epoch": 0.25427472797185635,
      "grad_norm": 0.29603078961372375,
      "learning_rate": 8.475593127897465e-06,
      "loss": 0.006,
      "step": 15540
    },
    {
      "epoch": 0.2544383539229322,
      "grad_norm": 0.39668726921081543,
      "learning_rate": 8.481047177529316e-06,
      "loss": 0.0123,
      "step": 15550
    },
    {
      "epoch": 0.254601979874008,
      "grad_norm": 0.7039226293563843,
      "learning_rate": 8.486501227161169e-06,
      "loss": 0.008,
      "step": 15560
    },
    {
      "epoch": 0.25476560582508384,
      "grad_norm": 0.5413767695426941,
      "learning_rate": 8.49195527679302e-06,
      "loss": 0.005,
      "step": 15570
    },
    {
      "epoch": 0.2549292317761597,
      "grad_norm": 0.2888321578502655,
      "learning_rate": 8.497409326424872e-06,
      "loss": 0.0065,
      "step": 15580
    },
    {
      "epoch": 0.25509285772723556,
      "grad_norm": 1.0297186374664307,
      "learning_rate": 8.502863376056723e-06,
      "loss": 0.0082,
      "step": 15590
    },
    {
      "epoch": 0.2552564836783114,
      "grad_norm": 0.5334463119506836,
      "learning_rate": 8.508317425688575e-06,
      "loss": 0.007,
      "step": 15600
    },
    {
      "epoch": 0.2554201096293872,
      "grad_norm": 0.4290338158607483,
      "learning_rate": 8.513771475320426e-06,
      "loss": 0.0089,
      "step": 15610
    },
    {
      "epoch": 0.25558373558046305,
      "grad_norm": 0.5034772753715515,
      "learning_rate": 8.519225524952279e-06,
      "loss": 0.0075,
      "step": 15620
    },
    {
      "epoch": 0.2557473615315389,
      "grad_norm": 0.6412157416343689,
      "learning_rate": 8.52467957458413e-06,
      "loss": 0.006,
      "step": 15630
    },
    {
      "epoch": 0.25591098748261476,
      "grad_norm": 0.4565969705581665,
      "learning_rate": 8.530133624215982e-06,
      "loss": 0.0051,
      "step": 15640
    },
    {
      "epoch": 0.2560746134336906,
      "grad_norm": 0.6854017972946167,
      "learning_rate": 8.535587673847833e-06,
      "loss": 0.0087,
      "step": 15650
    },
    {
      "epoch": 0.2562382393847664,
      "grad_norm": 0.5298134088516235,
      "learning_rate": 8.541041723479684e-06,
      "loss": 0.0058,
      "step": 15660
    },
    {
      "epoch": 0.25640186533584225,
      "grad_norm": 0.277515709400177,
      "learning_rate": 8.546495773111537e-06,
      "loss": 0.0094,
      "step": 15670
    },
    {
      "epoch": 0.2565654912869181,
      "grad_norm": 0.08279644697904587,
      "learning_rate": 8.551949822743387e-06,
      "loss": 0.0061,
      "step": 15680
    },
    {
      "epoch": 0.25672911723799396,
      "grad_norm": 0.1460450440645218,
      "learning_rate": 8.55740387237524e-06,
      "loss": 0.0063,
      "step": 15690
    },
    {
      "epoch": 0.2568927431890698,
      "grad_norm": 0.36915767192840576,
      "learning_rate": 8.56285792200709e-06,
      "loss": 0.0066,
      "step": 15700
    },
    {
      "epoch": 0.2570563691401456,
      "grad_norm": 0.28821367025375366,
      "learning_rate": 8.568311971638943e-06,
      "loss": 0.0082,
      "step": 15710
    },
    {
      "epoch": 0.25721999509122145,
      "grad_norm": 0.47287726402282715,
      "learning_rate": 8.573766021270794e-06,
      "loss": 0.0085,
      "step": 15720
    },
    {
      "epoch": 0.2573836210422973,
      "grad_norm": 0.3398825228214264,
      "learning_rate": 8.579220070902647e-06,
      "loss": 0.0096,
      "step": 15730
    },
    {
      "epoch": 0.25754724699337317,
      "grad_norm": 0.3178982436656952,
      "learning_rate": 8.584674120534498e-06,
      "loss": 0.0134,
      "step": 15740
    },
    {
      "epoch": 0.257710872944449,
      "grad_norm": 0.5679348111152649,
      "learning_rate": 8.59012817016635e-06,
      "loss": 0.0089,
      "step": 15750
    },
    {
      "epoch": 0.25787449889552483,
      "grad_norm": 0.12127234041690826,
      "learning_rate": 8.595582219798201e-06,
      "loss": 0.0054,
      "step": 15760
    },
    {
      "epoch": 0.25803812484660066,
      "grad_norm": 0.5184342265129089,
      "learning_rate": 8.601036269430052e-06,
      "loss": 0.0088,
      "step": 15770
    },
    {
      "epoch": 0.2582017507976765,
      "grad_norm": 0.2941882908344269,
      "learning_rate": 8.606490319061904e-06,
      "loss": 0.007,
      "step": 15780
    },
    {
      "epoch": 0.2583653767487524,
      "grad_norm": 0.21098819375038147,
      "learning_rate": 8.611944368693755e-06,
      "loss": 0.0047,
      "step": 15790
    },
    {
      "epoch": 0.2585290026998282,
      "grad_norm": 0.5106900930404663,
      "learning_rate": 8.617398418325608e-06,
      "loss": 0.0065,
      "step": 15800
    },
    {
      "epoch": 0.25869262865090403,
      "grad_norm": 0.1790807992219925,
      "learning_rate": 8.622852467957459e-06,
      "loss": 0.006,
      "step": 15810
    },
    {
      "epoch": 0.25885625460197986,
      "grad_norm": 0.25317224860191345,
      "learning_rate": 8.628306517589311e-06,
      "loss": 0.0059,
      "step": 15820
    },
    {
      "epoch": 0.2590198805530557,
      "grad_norm": 0.37723788619041443,
      "learning_rate": 8.633760567221162e-06,
      "loss": 0.006,
      "step": 15830
    },
    {
      "epoch": 0.2591835065041316,
      "grad_norm": 0.4963933825492859,
      "learning_rate": 8.639214616853015e-06,
      "loss": 0.0062,
      "step": 15840
    },
    {
      "epoch": 0.2593471324552074,
      "grad_norm": 0.42260056734085083,
      "learning_rate": 8.644668666484865e-06,
      "loss": 0.0071,
      "step": 15850
    },
    {
      "epoch": 0.25951075840628324,
      "grad_norm": 0.24291133880615234,
      "learning_rate": 8.650122716116718e-06,
      "loss": 0.0062,
      "step": 15860
    },
    {
      "epoch": 0.25967438435735907,
      "grad_norm": 0.17453806102275848,
      "learning_rate": 8.655576765748569e-06,
      "loss": 0.0063,
      "step": 15870
    },
    {
      "epoch": 0.2598380103084349,
      "grad_norm": 0.30748218297958374,
      "learning_rate": 8.66103081538042e-06,
      "loss": 0.0075,
      "step": 15880
    },
    {
      "epoch": 0.2600016362595108,
      "grad_norm": 0.5070960521697998,
      "learning_rate": 8.666484865012272e-06,
      "loss": 0.0052,
      "step": 15890
    },
    {
      "epoch": 0.2601652622105866,
      "grad_norm": 0.4176754951477051,
      "learning_rate": 8.671938914644123e-06,
      "loss": 0.0052,
      "step": 15900
    },
    {
      "epoch": 0.26032888816166244,
      "grad_norm": 0.43345314264297485,
      "learning_rate": 8.677392964275976e-06,
      "loss": 0.0079,
      "step": 15910
    },
    {
      "epoch": 0.26049251411273827,
      "grad_norm": 0.40920785069465637,
      "learning_rate": 8.682847013907826e-06,
      "loss": 0.0117,
      "step": 15920
    },
    {
      "epoch": 0.2606561400638141,
      "grad_norm": 0.37591618299484253,
      "learning_rate": 8.688301063539679e-06,
      "loss": 0.0097,
      "step": 15930
    },
    {
      "epoch": 0.26081976601489,
      "grad_norm": 0.42752519249916077,
      "learning_rate": 8.69375511317153e-06,
      "loss": 0.0046,
      "step": 15940
    },
    {
      "epoch": 0.2609833919659658,
      "grad_norm": 0.2730713188648224,
      "learning_rate": 8.699209162803382e-06,
      "loss": 0.0052,
      "step": 15950
    },
    {
      "epoch": 0.26114701791704165,
      "grad_norm": 0.6756895780563354,
      "learning_rate": 8.704663212435233e-06,
      "loss": 0.0145,
      "step": 15960
    },
    {
      "epoch": 0.2613106438681175,
      "grad_norm": 0.3605157434940338,
      "learning_rate": 8.710117262067086e-06,
      "loss": 0.0091,
      "step": 15970
    },
    {
      "epoch": 0.2614742698191933,
      "grad_norm": 0.3654633164405823,
      "learning_rate": 8.715571311698937e-06,
      "loss": 0.0064,
      "step": 15980
    },
    {
      "epoch": 0.2616378957702692,
      "grad_norm": 0.9474996328353882,
      "learning_rate": 8.72102536133079e-06,
      "loss": 0.009,
      "step": 15990
    },
    {
      "epoch": 0.261801521721345,
      "grad_norm": 0.09037787467241287,
      "learning_rate": 8.72647941096264e-06,
      "loss": 0.0077,
      "step": 16000
    },
    {
      "epoch": 0.26196514767242085,
      "grad_norm": 0.43435654044151306,
      "learning_rate": 8.731933460594491e-06,
      "loss": 0.0081,
      "step": 16010
    },
    {
      "epoch": 0.2621287736234967,
      "grad_norm": 0.3199719488620758,
      "learning_rate": 8.737387510226343e-06,
      "loss": 0.0071,
      "step": 16020
    },
    {
      "epoch": 0.2622923995745725,
      "grad_norm": 0.25331613421440125,
      "learning_rate": 8.742841559858194e-06,
      "loss": 0.0092,
      "step": 16030
    },
    {
      "epoch": 0.2624560255256484,
      "grad_norm": 0.3290269076824188,
      "learning_rate": 8.748295609490047e-06,
      "loss": 0.007,
      "step": 16040
    },
    {
      "epoch": 0.2626196514767242,
      "grad_norm": 0.4656314253807068,
      "learning_rate": 8.753749659121898e-06,
      "loss": 0.0078,
      "step": 16050
    },
    {
      "epoch": 0.26278327742780005,
      "grad_norm": 0.4236758351325989,
      "learning_rate": 8.75920370875375e-06,
      "loss": 0.0072,
      "step": 16060
    },
    {
      "epoch": 0.2629469033788759,
      "grad_norm": 0.23814909160137177,
      "learning_rate": 8.764657758385601e-06,
      "loss": 0.0054,
      "step": 16070
    },
    {
      "epoch": 0.2631105293299517,
      "grad_norm": 0.4015258252620697,
      "learning_rate": 8.770111808017454e-06,
      "loss": 0.0052,
      "step": 16080
    },
    {
      "epoch": 0.26327415528102754,
      "grad_norm": 0.677941083908081,
      "learning_rate": 8.775565857649305e-06,
      "loss": 0.0078,
      "step": 16090
    },
    {
      "epoch": 0.26343778123210343,
      "grad_norm": 0.4373498260974884,
      "learning_rate": 8.781019907281157e-06,
      "loss": 0.0086,
      "step": 16100
    },
    {
      "epoch": 0.26360140718317926,
      "grad_norm": 0.41308775544166565,
      "learning_rate": 8.786473956913008e-06,
      "loss": 0.0082,
      "step": 16110
    },
    {
      "epoch": 0.2637650331342551,
      "grad_norm": 0.2800823450088501,
      "learning_rate": 8.79192800654486e-06,
      "loss": 0.0072,
      "step": 16120
    },
    {
      "epoch": 0.2639286590853309,
      "grad_norm": 0.3212105929851532,
      "learning_rate": 8.797382056176711e-06,
      "loss": 0.0074,
      "step": 16130
    },
    {
      "epoch": 0.26409228503640675,
      "grad_norm": 0.43777480721473694,
      "learning_rate": 8.802836105808564e-06,
      "loss": 0.0064,
      "step": 16140
    },
    {
      "epoch": 0.26425591098748263,
      "grad_norm": 0.10684596002101898,
      "learning_rate": 8.808290155440415e-06,
      "loss": 0.0078,
      "step": 16150
    },
    {
      "epoch": 0.26441953693855846,
      "grad_norm": 0.3250572681427002,
      "learning_rate": 8.813744205072267e-06,
      "loss": 0.0065,
      "step": 16160
    },
    {
      "epoch": 0.2645831628896343,
      "grad_norm": 0.2860829830169678,
      "learning_rate": 8.819198254704118e-06,
      "loss": 0.0053,
      "step": 16170
    },
    {
      "epoch": 0.2647467888407101,
      "grad_norm": 0.7128570675849915,
      "learning_rate": 8.82465230433597e-06,
      "loss": 0.0103,
      "step": 16180
    },
    {
      "epoch": 0.26491041479178595,
      "grad_norm": 0.2804383933544159,
      "learning_rate": 8.830106353967822e-06,
      "loss": 0.0146,
      "step": 16190
    },
    {
      "epoch": 0.26507404074286184,
      "grad_norm": 0.12914694845676422,
      "learning_rate": 8.835560403599674e-06,
      "loss": 0.0034,
      "step": 16200
    },
    {
      "epoch": 0.26523766669393767,
      "grad_norm": 0.36098477244377136,
      "learning_rate": 8.841014453231525e-06,
      "loss": 0.007,
      "step": 16210
    },
    {
      "epoch": 0.2654012926450135,
      "grad_norm": 0.6338185667991638,
      "learning_rate": 8.846468502863378e-06,
      "loss": 0.0063,
      "step": 16220
    },
    {
      "epoch": 0.2655649185960893,
      "grad_norm": 0.4452401399612427,
      "learning_rate": 8.851922552495228e-06,
      "loss": 0.0085,
      "step": 16230
    },
    {
      "epoch": 0.26572854454716516,
      "grad_norm": 0.6460739970207214,
      "learning_rate": 8.857376602127081e-06,
      "loss": 0.0069,
      "step": 16240
    },
    {
      "epoch": 0.26589217049824104,
      "grad_norm": 0.4192502498626709,
      "learning_rate": 8.862830651758932e-06,
      "loss": 0.0056,
      "step": 16250
    },
    {
      "epoch": 0.2660557964493169,
      "grad_norm": 0.3042267858982086,
      "learning_rate": 8.868284701390784e-06,
      "loss": 0.0047,
      "step": 16260
    },
    {
      "epoch": 0.2662194224003927,
      "grad_norm": 0.5498833060264587,
      "learning_rate": 8.873738751022635e-06,
      "loss": 0.0091,
      "step": 16270
    },
    {
      "epoch": 0.26638304835146853,
      "grad_norm": 0.3096252381801605,
      "learning_rate": 8.879192800654488e-06,
      "loss": 0.0055,
      "step": 16280
    },
    {
      "epoch": 0.26654667430254436,
      "grad_norm": 0.45774009823799133,
      "learning_rate": 8.884646850286339e-06,
      "loss": 0.0066,
      "step": 16290
    },
    {
      "epoch": 0.26671030025362025,
      "grad_norm": 0.15017341077327728,
      "learning_rate": 8.890100899918191e-06,
      "loss": 0.0066,
      "step": 16300
    },
    {
      "epoch": 0.2668739262046961,
      "grad_norm": 0.4404623508453369,
      "learning_rate": 8.895554949550042e-06,
      "loss": 0.0068,
      "step": 16310
    },
    {
      "epoch": 0.2670375521557719,
      "grad_norm": 0.3334588408470154,
      "learning_rate": 8.901008999181895e-06,
      "loss": 0.0083,
      "step": 16320
    },
    {
      "epoch": 0.26720117810684774,
      "grad_norm": 0.6172933578491211,
      "learning_rate": 8.906463048813745e-06,
      "loss": 0.0082,
      "step": 16330
    },
    {
      "epoch": 0.26736480405792357,
      "grad_norm": 0.22365346550941467,
      "learning_rate": 8.911917098445596e-06,
      "loss": 0.0093,
      "step": 16340
    },
    {
      "epoch": 0.26752843000899945,
      "grad_norm": 0.28134772181510925,
      "learning_rate": 8.917371148077449e-06,
      "loss": 0.0056,
      "step": 16350
    },
    {
      "epoch": 0.2676920559600753,
      "grad_norm": 0.3498561382293701,
      "learning_rate": 8.9228251977093e-06,
      "loss": 0.0083,
      "step": 16360
    },
    {
      "epoch": 0.2678556819111511,
      "grad_norm": 0.2254149168729782,
      "learning_rate": 8.928279247341152e-06,
      "loss": 0.0075,
      "step": 16370
    },
    {
      "epoch": 0.26801930786222694,
      "grad_norm": 0.27590957283973694,
      "learning_rate": 8.933733296973003e-06,
      "loss": 0.0084,
      "step": 16380
    },
    {
      "epoch": 0.26818293381330277,
      "grad_norm": 0.40174850821495056,
      "learning_rate": 8.939187346604856e-06,
      "loss": 0.0133,
      "step": 16390
    },
    {
      "epoch": 0.26834655976437866,
      "grad_norm": 0.2599905729293823,
      "learning_rate": 8.944641396236706e-06,
      "loss": 0.0071,
      "step": 16400
    },
    {
      "epoch": 0.2685101857154545,
      "grad_norm": 0.4147341251373291,
      "learning_rate": 8.950095445868559e-06,
      "loss": 0.0054,
      "step": 16410
    },
    {
      "epoch": 0.2686738116665303,
      "grad_norm": 0.4697679281234741,
      "learning_rate": 8.95554949550041e-06,
      "loss": 0.0106,
      "step": 16420
    },
    {
      "epoch": 0.26883743761760615,
      "grad_norm": 0.34753677248954773,
      "learning_rate": 8.961003545132262e-06,
      "loss": 0.0062,
      "step": 16430
    },
    {
      "epoch": 0.269001063568682,
      "grad_norm": 0.8293981552124023,
      "learning_rate": 8.966457594764113e-06,
      "loss": 0.0082,
      "step": 16440
    },
    {
      "epoch": 0.26916468951975786,
      "grad_norm": 0.45402100682258606,
      "learning_rate": 8.971911644395966e-06,
      "loss": 0.008,
      "step": 16450
    },
    {
      "epoch": 0.2693283154708337,
      "grad_norm": 0.6655663847923279,
      "learning_rate": 8.977365694027817e-06,
      "loss": 0.0079,
      "step": 16460
    },
    {
      "epoch": 0.2694919414219095,
      "grad_norm": 0.6291821002960205,
      "learning_rate": 8.982819743659667e-06,
      "loss": 0.0109,
      "step": 16470
    },
    {
      "epoch": 0.26965556737298535,
      "grad_norm": 0.541570782661438,
      "learning_rate": 8.98827379329152e-06,
      "loss": 0.0109,
      "step": 16480
    },
    {
      "epoch": 0.2698191933240612,
      "grad_norm": 0.32790982723236084,
      "learning_rate": 8.993727842923371e-06,
      "loss": 0.0071,
      "step": 16490
    },
    {
      "epoch": 0.269982819275137,
      "grad_norm": 0.2682746946811676,
      "learning_rate": 8.999181892555223e-06,
      "loss": 0.0072,
      "step": 16500
    },
    {
      "epoch": 0.2701464452262129,
      "grad_norm": 0.6405606865882874,
      "learning_rate": 9.004635942187074e-06,
      "loss": 0.0098,
      "step": 16510
    },
    {
      "epoch": 0.2703100711772887,
      "grad_norm": 0.4136946499347687,
      "learning_rate": 9.010089991818927e-06,
      "loss": 0.0088,
      "step": 16520
    },
    {
      "epoch": 0.27047369712836455,
      "grad_norm": 0.9547814726829529,
      "learning_rate": 9.015544041450778e-06,
      "loss": 0.0072,
      "step": 16530
    },
    {
      "epoch": 0.2706373230794404,
      "grad_norm": 0.4734065532684326,
      "learning_rate": 9.02099809108263e-06,
      "loss": 0.0056,
      "step": 16540
    },
    {
      "epoch": 0.2708009490305162,
      "grad_norm": 0.3656756579875946,
      "learning_rate": 9.026452140714481e-06,
      "loss": 0.0131,
      "step": 16550
    },
    {
      "epoch": 0.2709645749815921,
      "grad_norm": 0.41437825560569763,
      "learning_rate": 9.031906190346334e-06,
      "loss": 0.0092,
      "step": 16560
    },
    {
      "epoch": 0.27112820093266793,
      "grad_norm": 0.4583395719528198,
      "learning_rate": 9.037360239978185e-06,
      "loss": 0.0061,
      "step": 16570
    },
    {
      "epoch": 0.27129182688374376,
      "grad_norm": 0.4452890157699585,
      "learning_rate": 9.042814289610035e-06,
      "loss": 0.008,
      "step": 16580
    },
    {
      "epoch": 0.2714554528348196,
      "grad_norm": 0.2936899960041046,
      "learning_rate": 9.048268339241888e-06,
      "loss": 0.0061,
      "step": 16590
    },
    {
      "epoch": 0.2716190787858954,
      "grad_norm": 0.6065975427627563,
      "learning_rate": 9.053722388873739e-06,
      "loss": 0.0081,
      "step": 16600
    },
    {
      "epoch": 0.2717827047369713,
      "grad_norm": 0.43540599942207336,
      "learning_rate": 9.059176438505591e-06,
      "loss": 0.0091,
      "step": 16610
    },
    {
      "epoch": 0.27194633068804713,
      "grad_norm": 0.371324360370636,
      "learning_rate": 9.064630488137442e-06,
      "loss": 0.0055,
      "step": 16620
    },
    {
      "epoch": 0.27210995663912296,
      "grad_norm": 0.2378314733505249,
      "learning_rate": 9.070084537769295e-06,
      "loss": 0.0052,
      "step": 16630
    },
    {
      "epoch": 0.2722735825901988,
      "grad_norm": 0.2827133536338806,
      "learning_rate": 9.075538587401146e-06,
      "loss": 0.0069,
      "step": 16640
    },
    {
      "epoch": 0.2724372085412746,
      "grad_norm": 0.49517032504081726,
      "learning_rate": 9.080992637032998e-06,
      "loss": 0.0066,
      "step": 16650
    },
    {
      "epoch": 0.2726008344923505,
      "grad_norm": 0.43881359696388245,
      "learning_rate": 9.086446686664849e-06,
      "loss": 0.0079,
      "step": 16660
    },
    {
      "epoch": 0.27276446044342634,
      "grad_norm": 0.4152918756008148,
      "learning_rate": 9.091900736296702e-06,
      "loss": 0.0072,
      "step": 16670
    },
    {
      "epoch": 0.27292808639450217,
      "grad_norm": 0.3023996949195862,
      "learning_rate": 9.097354785928552e-06,
      "loss": 0.0045,
      "step": 16680
    },
    {
      "epoch": 0.273091712345578,
      "grad_norm": 0.7321740984916687,
      "learning_rate": 9.102808835560403e-06,
      "loss": 0.0082,
      "step": 16690
    },
    {
      "epoch": 0.2732553382966538,
      "grad_norm": 0.7708824872970581,
      "learning_rate": 9.108262885192256e-06,
      "loss": 0.0098,
      "step": 16700
    },
    {
      "epoch": 0.2734189642477297,
      "grad_norm": 0.2992950975894928,
      "learning_rate": 9.113716934824107e-06,
      "loss": 0.0073,
      "step": 16710
    },
    {
      "epoch": 0.27358259019880554,
      "grad_norm": 0.11642202734947205,
      "learning_rate": 9.11917098445596e-06,
      "loss": 0.0058,
      "step": 16720
    },
    {
      "epoch": 0.27374621614988137,
      "grad_norm": 0.34320154786109924,
      "learning_rate": 9.12462503408781e-06,
      "loss": 0.0109,
      "step": 16730
    },
    {
      "epoch": 0.2739098421009572,
      "grad_norm": 0.35033077001571655,
      "learning_rate": 9.130079083719663e-06,
      "loss": 0.0072,
      "step": 16740
    },
    {
      "epoch": 0.27407346805203303,
      "grad_norm": 0.48964419960975647,
      "learning_rate": 9.135533133351513e-06,
      "loss": 0.0052,
      "step": 16750
    },
    {
      "epoch": 0.2742370940031089,
      "grad_norm": 0.7566307783126831,
      "learning_rate": 9.140987182983366e-06,
      "loss": 0.0075,
      "step": 16760
    },
    {
      "epoch": 0.27440071995418475,
      "grad_norm": 0.33908167481422424,
      "learning_rate": 9.146441232615217e-06,
      "loss": 0.0079,
      "step": 16770
    },
    {
      "epoch": 0.2745643459052606,
      "grad_norm": 0.2446870505809784,
      "learning_rate": 9.15189528224707e-06,
      "loss": 0.0065,
      "step": 16780
    },
    {
      "epoch": 0.2747279718563364,
      "grad_norm": 0.4471055269241333,
      "learning_rate": 9.15734933187892e-06,
      "loss": 0.0076,
      "step": 16790
    },
    {
      "epoch": 0.27489159780741224,
      "grad_norm": 0.8024077415466309,
      "learning_rate": 9.162803381510773e-06,
      "loss": 0.006,
      "step": 16800
    },
    {
      "epoch": 0.2750552237584881,
      "grad_norm": 0.484811931848526,
      "learning_rate": 9.168257431142624e-06,
      "loss": 0.0044,
      "step": 16810
    },
    {
      "epoch": 0.27521884970956395,
      "grad_norm": 0.19515646994113922,
      "learning_rate": 9.173711480774476e-06,
      "loss": 0.0057,
      "step": 16820
    },
    {
      "epoch": 0.2753824756606398,
      "grad_norm": 0.46697765588760376,
      "learning_rate": 9.179165530406327e-06,
      "loss": 0.0065,
      "step": 16830
    },
    {
      "epoch": 0.2755461016117156,
      "grad_norm": 0.42592164874076843,
      "learning_rate": 9.18461958003818e-06,
      "loss": 0.0088,
      "step": 16840
    },
    {
      "epoch": 0.27570972756279144,
      "grad_norm": 0.48190489411354065,
      "learning_rate": 9.19007362967003e-06,
      "loss": 0.004,
      "step": 16850
    },
    {
      "epoch": 0.2758733535138673,
      "grad_norm": 0.3762807250022888,
      "learning_rate": 9.195527679301883e-06,
      "loss": 0.01,
      "step": 16860
    },
    {
      "epoch": 0.27603697946494316,
      "grad_norm": 0.17557047307491302,
      "learning_rate": 9.200981728933734e-06,
      "loss": 0.0069,
      "step": 16870
    },
    {
      "epoch": 0.276200605416019,
      "grad_norm": 0.7334563136100769,
      "learning_rate": 9.206435778565586e-06,
      "loss": 0.0082,
      "step": 16880
    },
    {
      "epoch": 0.2763642313670948,
      "grad_norm": 0.6913423538208008,
      "learning_rate": 9.211889828197437e-06,
      "loss": 0.0054,
      "step": 16890
    },
    {
      "epoch": 0.27652785731817064,
      "grad_norm": 0.8012818098068237,
      "learning_rate": 9.21734387782929e-06,
      "loss": 0.0113,
      "step": 16900
    },
    {
      "epoch": 0.27669148326924653,
      "grad_norm": 0.5596820116043091,
      "learning_rate": 9.22279792746114e-06,
      "loss": 0.0081,
      "step": 16910
    },
    {
      "epoch": 0.27685510922032236,
      "grad_norm": 0.4714412987232208,
      "learning_rate": 9.228251977092993e-06,
      "loss": 0.0083,
      "step": 16920
    },
    {
      "epoch": 0.2770187351713982,
      "grad_norm": 0.6238222718238831,
      "learning_rate": 9.233706026724844e-06,
      "loss": 0.0062,
      "step": 16930
    },
    {
      "epoch": 0.277182361122474,
      "grad_norm": 0.3426477611064911,
      "learning_rate": 9.239160076356697e-06,
      "loss": 0.008,
      "step": 16940
    },
    {
      "epoch": 0.27734598707354985,
      "grad_norm": 0.6691911220550537,
      "learning_rate": 9.244614125988547e-06,
      "loss": 0.0089,
      "step": 16950
    },
    {
      "epoch": 0.2775096130246257,
      "grad_norm": 0.40745797753334045,
      "learning_rate": 9.2500681756204e-06,
      "loss": 0.0089,
      "step": 16960
    },
    {
      "epoch": 0.27767323897570156,
      "grad_norm": 0.5076944231987,
      "learning_rate": 9.255522225252251e-06,
      "loss": 0.0082,
      "step": 16970
    },
    {
      "epoch": 0.2778368649267774,
      "grad_norm": 0.07181976735591888,
      "learning_rate": 9.260976274884103e-06,
      "loss": 0.0098,
      "step": 16980
    },
    {
      "epoch": 0.2780004908778532,
      "grad_norm": 0.36731722950935364,
      "learning_rate": 9.266430324515954e-06,
      "loss": 0.0068,
      "step": 16990
    },
    {
      "epoch": 0.27816411682892905,
      "grad_norm": 0.5639378428459167,
      "learning_rate": 9.271884374147807e-06,
      "loss": 0.0058,
      "step": 17000
    },
    {
      "epoch": 0.2783277427800049,
      "grad_norm": 0.3828292489051819,
      "learning_rate": 9.277338423779658e-06,
      "loss": 0.0097,
      "step": 17010
    },
    {
      "epoch": 0.27849136873108077,
      "grad_norm": 0.3754061162471771,
      "learning_rate": 9.28279247341151e-06,
      "loss": 0.0126,
      "step": 17020
    },
    {
      "epoch": 0.2786549946821566,
      "grad_norm": 0.4050922095775604,
      "learning_rate": 9.288246523043361e-06,
      "loss": 0.0042,
      "step": 17030
    },
    {
      "epoch": 0.27881862063323243,
      "grad_norm": 0.04091784730553627,
      "learning_rate": 9.293700572675212e-06,
      "loss": 0.0053,
      "step": 17040
    },
    {
      "epoch": 0.27898224658430826,
      "grad_norm": 0.7078526616096497,
      "learning_rate": 9.299154622307064e-06,
      "loss": 0.0061,
      "step": 17050
    },
    {
      "epoch": 0.2791458725353841,
      "grad_norm": 0.5788561105728149,
      "learning_rate": 9.304608671938915e-06,
      "loss": 0.0081,
      "step": 17060
    },
    {
      "epoch": 0.27930949848646,
      "grad_norm": 0.17846155166625977,
      "learning_rate": 9.310062721570768e-06,
      "loss": 0.0058,
      "step": 17070
    },
    {
      "epoch": 0.2794731244375358,
      "grad_norm": 0.28911489248275757,
      "learning_rate": 9.315516771202619e-06,
      "loss": 0.0082,
      "step": 17080
    },
    {
      "epoch": 0.27963675038861163,
      "grad_norm": 0.4038585126399994,
      "learning_rate": 9.320970820834471e-06,
      "loss": 0.0047,
      "step": 17090
    },
    {
      "epoch": 0.27980037633968746,
      "grad_norm": 0.3644547164440155,
      "learning_rate": 9.326424870466322e-06,
      "loss": 0.0094,
      "step": 17100
    },
    {
      "epoch": 0.2799640022907633,
      "grad_norm": 0.07796856015920639,
      "learning_rate": 9.331878920098175e-06,
      "loss": 0.0067,
      "step": 17110
    },
    {
      "epoch": 0.2801276282418392,
      "grad_norm": 0.2838761806488037,
      "learning_rate": 9.337332969730026e-06,
      "loss": 0.0103,
      "step": 17120
    },
    {
      "epoch": 0.280291254192915,
      "grad_norm": 0.33975324034690857,
      "learning_rate": 9.342787019361878e-06,
      "loss": 0.0059,
      "step": 17130
    },
    {
      "epoch": 0.28045488014399084,
      "grad_norm": 0.3931236267089844,
      "learning_rate": 9.348241068993729e-06,
      "loss": 0.0064,
      "step": 17140
    },
    {
      "epoch": 0.28061850609506667,
      "grad_norm": 0.5075228810310364,
      "learning_rate": 9.35369511862558e-06,
      "loss": 0.0118,
      "step": 17150
    },
    {
      "epoch": 0.2807821320461425,
      "grad_norm": 0.31072351336479187,
      "learning_rate": 9.359149168257432e-06,
      "loss": 0.0083,
      "step": 17160
    },
    {
      "epoch": 0.2809457579972184,
      "grad_norm": 0.33038461208343506,
      "learning_rate": 9.364603217889283e-06,
      "loss": 0.007,
      "step": 17170
    },
    {
      "epoch": 0.2811093839482942,
      "grad_norm": 0.32689762115478516,
      "learning_rate": 9.370057267521136e-06,
      "loss": 0.0074,
      "step": 17180
    },
    {
      "epoch": 0.28127300989937004,
      "grad_norm": 0.4697996973991394,
      "learning_rate": 9.375511317152987e-06,
      "loss": 0.0069,
      "step": 17190
    },
    {
      "epoch": 0.28143663585044587,
      "grad_norm": 0.573824942111969,
      "learning_rate": 9.380965366784839e-06,
      "loss": 0.0095,
      "step": 17200
    },
    {
      "epoch": 0.2816002618015217,
      "grad_norm": 0.5459651947021484,
      "learning_rate": 9.38641941641669e-06,
      "loss": 0.0079,
      "step": 17210
    },
    {
      "epoch": 0.2817638877525976,
      "grad_norm": 0.3976970613002777,
      "learning_rate": 9.391873466048543e-06,
      "loss": 0.0084,
      "step": 17220
    },
    {
      "epoch": 0.2819275137036734,
      "grad_norm": 1.0190106630325317,
      "learning_rate": 9.397327515680393e-06,
      "loss": 0.0074,
      "step": 17230
    },
    {
      "epoch": 0.28209113965474925,
      "grad_norm": 0.13550855219364166,
      "learning_rate": 9.402781565312246e-06,
      "loss": 0.0077,
      "step": 17240
    },
    {
      "epoch": 0.2822547656058251,
      "grad_norm": 0.4691948890686035,
      "learning_rate": 9.408235614944097e-06,
      "loss": 0.0067,
      "step": 17250
    },
    {
      "epoch": 0.2824183915569009,
      "grad_norm": 0.5311391949653625,
      "learning_rate": 9.413689664575948e-06,
      "loss": 0.0064,
      "step": 17260
    },
    {
      "epoch": 0.2825820175079768,
      "grad_norm": 0.23525945842266083,
      "learning_rate": 9.4191437142078e-06,
      "loss": 0.0076,
      "step": 17270
    },
    {
      "epoch": 0.2827456434590526,
      "grad_norm": 0.8739100694656372,
      "learning_rate": 9.424597763839651e-06,
      "loss": 0.007,
      "step": 17280
    },
    {
      "epoch": 0.28290926941012845,
      "grad_norm": 0.31303292512893677,
      "learning_rate": 9.430051813471504e-06,
      "loss": 0.0078,
      "step": 17290
    },
    {
      "epoch": 0.2830728953612043,
      "grad_norm": 0.22775372862815857,
      "learning_rate": 9.435505863103354e-06,
      "loss": 0.0059,
      "step": 17300
    },
    {
      "epoch": 0.2832365213122801,
      "grad_norm": 0.21936200559139252,
      "learning_rate": 9.440959912735207e-06,
      "loss": 0.0051,
      "step": 17310
    },
    {
      "epoch": 0.283400147263356,
      "grad_norm": 0.24235956370830536,
      "learning_rate": 9.446413962367058e-06,
      "loss": 0.0059,
      "step": 17320
    },
    {
      "epoch": 0.2835637732144318,
      "grad_norm": 0.5034245848655701,
      "learning_rate": 9.45186801199891e-06,
      "loss": 0.0065,
      "step": 17330
    },
    {
      "epoch": 0.28372739916550765,
      "grad_norm": 0.3012360632419586,
      "learning_rate": 9.457322061630761e-06,
      "loss": 0.0072,
      "step": 17340
    },
    {
      "epoch": 0.2838910251165835,
      "grad_norm": 0.256351113319397,
      "learning_rate": 9.462776111262614e-06,
      "loss": 0.0054,
      "step": 17350
    },
    {
      "epoch": 0.2840546510676593,
      "grad_norm": 0.5906171798706055,
      "learning_rate": 9.468230160894465e-06,
      "loss": 0.0068,
      "step": 17360
    },
    {
      "epoch": 0.28421827701873514,
      "grad_norm": 0.2543254792690277,
      "learning_rate": 9.473684210526315e-06,
      "loss": 0.0054,
      "step": 17370
    },
    {
      "epoch": 0.28438190296981103,
      "grad_norm": 0.4507140517234802,
      "learning_rate": 9.479138260158168e-06,
      "loss": 0.0053,
      "step": 17380
    },
    {
      "epoch": 0.28454552892088686,
      "grad_norm": 0.36774569749832153,
      "learning_rate": 9.484592309790019e-06,
      "loss": 0.0083,
      "step": 17390
    },
    {
      "epoch": 0.2847091548719627,
      "grad_norm": 0.42855939269065857,
      "learning_rate": 9.490046359421871e-06,
      "loss": 0.0073,
      "step": 17400
    },
    {
      "epoch": 0.2848727808230385,
      "grad_norm": 0.18056045472621918,
      "learning_rate": 9.495500409053722e-06,
      "loss": 0.0053,
      "step": 17410
    },
    {
      "epoch": 0.28503640677411435,
      "grad_norm": 0.45547768473625183,
      "learning_rate": 9.500954458685575e-06,
      "loss": 0.008,
      "step": 17420
    },
    {
      "epoch": 0.28520003272519023,
      "grad_norm": 0.40392816066741943,
      "learning_rate": 9.506408508317426e-06,
      "loss": 0.0074,
      "step": 17430
    },
    {
      "epoch": 0.28536365867626606,
      "grad_norm": 0.2421988844871521,
      "learning_rate": 9.511862557949278e-06,
      "loss": 0.0052,
      "step": 17440
    },
    {
      "epoch": 0.2855272846273419,
      "grad_norm": 0.36258062720298767,
      "learning_rate": 9.517316607581129e-06,
      "loss": 0.0081,
      "step": 17450
    },
    {
      "epoch": 0.2856909105784177,
      "grad_norm": 0.2071145623922348,
      "learning_rate": 9.522770657212982e-06,
      "loss": 0.0085,
      "step": 17460
    },
    {
      "epoch": 0.28585453652949355,
      "grad_norm": 0.7389194965362549,
      "learning_rate": 9.528224706844833e-06,
      "loss": 0.0075,
      "step": 17470
    },
    {
      "epoch": 0.28601816248056944,
      "grad_norm": 0.22596819698810577,
      "learning_rate": 9.533678756476683e-06,
      "loss": 0.0052,
      "step": 17480
    },
    {
      "epoch": 0.28618178843164527,
      "grad_norm": 0.6479752063751221,
      "learning_rate": 9.539132806108536e-06,
      "loss": 0.0065,
      "step": 17490
    },
    {
      "epoch": 0.2863454143827211,
      "grad_norm": 0.2373426854610443,
      "learning_rate": 9.544586855740387e-06,
      "loss": 0.0061,
      "step": 17500
    },
    {
      "epoch": 0.2865090403337969,
      "grad_norm": 0.22162528336048126,
      "learning_rate": 9.55004090537224e-06,
      "loss": 0.0036,
      "step": 17510
    },
    {
      "epoch": 0.28667266628487276,
      "grad_norm": 0.6735430955886841,
      "learning_rate": 9.55549495500409e-06,
      "loss": 0.0112,
      "step": 17520
    },
    {
      "epoch": 0.28683629223594864,
      "grad_norm": 0.2916731536388397,
      "learning_rate": 9.560949004635943e-06,
      "loss": 0.0059,
      "step": 17530
    },
    {
      "epoch": 0.28699991818702447,
      "grad_norm": 0.25756481289863586,
      "learning_rate": 9.566403054267794e-06,
      "loss": 0.0055,
      "step": 17540
    },
    {
      "epoch": 0.2871635441381003,
      "grad_norm": 0.25046148896217346,
      "learning_rate": 9.571857103899646e-06,
      "loss": 0.0066,
      "step": 17550
    },
    {
      "epoch": 0.28732717008917613,
      "grad_norm": 0.49978095293045044,
      "learning_rate": 9.577311153531497e-06,
      "loss": 0.0099,
      "step": 17560
    },
    {
      "epoch": 0.28749079604025196,
      "grad_norm": 0.33705317974090576,
      "learning_rate": 9.58276520316335e-06,
      "loss": 0.0072,
      "step": 17570
    },
    {
      "epoch": 0.28765442199132785,
      "grad_norm": 0.49843356013298035,
      "learning_rate": 9.5882192527952e-06,
      "loss": 0.0085,
      "step": 17580
    },
    {
      "epoch": 0.2878180479424037,
      "grad_norm": 0.3631350100040436,
      "learning_rate": 9.593673302427053e-06,
      "loss": 0.006,
      "step": 17590
    },
    {
      "epoch": 0.2879816738934795,
      "grad_norm": 0.7583593130111694,
      "learning_rate": 9.599127352058904e-06,
      "loss": 0.0118,
      "step": 17600
    },
    {
      "epoch": 0.28814529984455534,
      "grad_norm": 0.07032642513513565,
      "learning_rate": 9.604581401690756e-06,
      "loss": 0.0059,
      "step": 17610
    },
    {
      "epoch": 0.28830892579563117,
      "grad_norm": 0.2079721987247467,
      "learning_rate": 9.610035451322607e-06,
      "loss": 0.0052,
      "step": 17620
    },
    {
      "epoch": 0.28847255174670705,
      "grad_norm": 0.5555415749549866,
      "learning_rate": 9.61548950095446e-06,
      "loss": 0.0103,
      "step": 17630
    },
    {
      "epoch": 0.2886361776977829,
      "grad_norm": 0.40738505125045776,
      "learning_rate": 9.62094355058631e-06,
      "loss": 0.0066,
      "step": 17640
    },
    {
      "epoch": 0.2887998036488587,
      "grad_norm": 0.4552417993545532,
      "learning_rate": 9.626397600218163e-06,
      "loss": 0.0058,
      "step": 17650
    },
    {
      "epoch": 0.28896342959993454,
      "grad_norm": 0.20900996029376984,
      "learning_rate": 9.631851649850014e-06,
      "loss": 0.0059,
      "step": 17660
    },
    {
      "epoch": 0.28912705555101037,
      "grad_norm": 1.4825387001037598,
      "learning_rate": 9.637305699481867e-06,
      "loss": 0.0161,
      "step": 17670
    },
    {
      "epoch": 0.28929068150208626,
      "grad_norm": 0.37469592690467834,
      "learning_rate": 9.642759749113717e-06,
      "loss": 0.0069,
      "step": 17680
    },
    {
      "epoch": 0.2894543074531621,
      "grad_norm": 0.17680047452449799,
      "learning_rate": 9.64821379874557e-06,
      "loss": 0.0054,
      "step": 17690
    },
    {
      "epoch": 0.2896179334042379,
      "grad_norm": 0.27439039945602417,
      "learning_rate": 9.65366784837742e-06,
      "loss": 0.0051,
      "step": 17700
    },
    {
      "epoch": 0.28978155935531374,
      "grad_norm": 0.6349892020225525,
      "learning_rate": 9.659121898009273e-06,
      "loss": 0.0051,
      "step": 17710
    },
    {
      "epoch": 0.2899451853063896,
      "grad_norm": 0.6358305215835571,
      "learning_rate": 9.664575947641124e-06,
      "loss": 0.0064,
      "step": 17720
    },
    {
      "epoch": 0.29010881125746546,
      "grad_norm": 0.4132031798362732,
      "learning_rate": 9.670029997272977e-06,
      "loss": 0.0077,
      "step": 17730
    },
    {
      "epoch": 0.2902724372085413,
      "grad_norm": 0.3340214490890503,
      "learning_rate": 9.675484046904828e-06,
      "loss": 0.0076,
      "step": 17740
    },
    {
      "epoch": 0.2904360631596171,
      "grad_norm": 0.26891523599624634,
      "learning_rate": 9.68093809653668e-06,
      "loss": 0.008,
      "step": 17750
    },
    {
      "epoch": 0.29059968911069295,
      "grad_norm": 1.0061402320861816,
      "learning_rate": 9.686392146168531e-06,
      "loss": 0.0055,
      "step": 17760
    },
    {
      "epoch": 0.2907633150617688,
      "grad_norm": 0.1788272261619568,
      "learning_rate": 9.691846195800384e-06,
      "loss": 0.008,
      "step": 17770
    },
    {
      "epoch": 0.29092694101284466,
      "grad_norm": 0.3419720232486725,
      "learning_rate": 9.697300245432234e-06,
      "loss": 0.0069,
      "step": 17780
    },
    {
      "epoch": 0.2910905669639205,
      "grad_norm": 0.16235148906707764,
      "learning_rate": 9.702754295064087e-06,
      "loss": 0.0072,
      "step": 17790
    },
    {
      "epoch": 0.2912541929149963,
      "grad_norm": 0.3408327102661133,
      "learning_rate": 9.708208344695938e-06,
      "loss": 0.0049,
      "step": 17800
    },
    {
      "epoch": 0.29141781886607215,
      "grad_norm": 0.42627087235450745,
      "learning_rate": 9.71366239432779e-06,
      "loss": 0.0059,
      "step": 17810
    },
    {
      "epoch": 0.291581444817148,
      "grad_norm": 0.29981982707977295,
      "learning_rate": 9.719116443959641e-06,
      "loss": 0.0107,
      "step": 17820
    },
    {
      "epoch": 0.2917450707682238,
      "grad_norm": 0.18751360476016998,
      "learning_rate": 9.724570493591492e-06,
      "loss": 0.0049,
      "step": 17830
    },
    {
      "epoch": 0.2919086967192997,
      "grad_norm": 0.18362772464752197,
      "learning_rate": 9.730024543223345e-06,
      "loss": 0.0042,
      "step": 17840
    },
    {
      "epoch": 0.29207232267037553,
      "grad_norm": 0.3897661566734314,
      "learning_rate": 9.735478592855195e-06,
      "loss": 0.0059,
      "step": 17850
    },
    {
      "epoch": 0.29223594862145136,
      "grad_norm": 0.272909939289093,
      "learning_rate": 9.740932642487048e-06,
      "loss": 0.008,
      "step": 17860
    },
    {
      "epoch": 0.2923995745725272,
      "grad_norm": 0.330464631319046,
      "learning_rate": 9.746386692118899e-06,
      "loss": 0.0056,
      "step": 17870
    },
    {
      "epoch": 0.292563200523603,
      "grad_norm": 0.5986437201499939,
      "learning_rate": 9.751840741750751e-06,
      "loss": 0.0059,
      "step": 17880
    },
    {
      "epoch": 0.2927268264746789,
      "grad_norm": 0.1420435756444931,
      "learning_rate": 9.757294791382602e-06,
      "loss": 0.0067,
      "step": 17890
    },
    {
      "epoch": 0.29289045242575473,
      "grad_norm": 0.29462817311286926,
      "learning_rate": 9.762748841014455e-06,
      "loss": 0.0065,
      "step": 17900
    },
    {
      "epoch": 0.29305407837683056,
      "grad_norm": 0.6878054738044739,
      "learning_rate": 9.768202890646306e-06,
      "loss": 0.0094,
      "step": 17910
    },
    {
      "epoch": 0.2932177043279064,
      "grad_norm": 0.3721540570259094,
      "learning_rate": 9.773656940278158e-06,
      "loss": 0.0063,
      "step": 17920
    },
    {
      "epoch": 0.2933813302789822,
      "grad_norm": 0.39864540100097656,
      "learning_rate": 9.779110989910009e-06,
      "loss": 0.0056,
      "step": 17930
    },
    {
      "epoch": 0.2935449562300581,
      "grad_norm": 0.2694548964500427,
      "learning_rate": 9.78456503954186e-06,
      "loss": 0.0064,
      "step": 17940
    },
    {
      "epoch": 0.29370858218113394,
      "grad_norm": 0.2016744613647461,
      "learning_rate": 9.790019089173712e-06,
      "loss": 0.0065,
      "step": 17950
    },
    {
      "epoch": 0.29387220813220977,
      "grad_norm": 0.24006912112236023,
      "learning_rate": 9.795473138805563e-06,
      "loss": 0.0035,
      "step": 17960
    },
    {
      "epoch": 0.2940358340832856,
      "grad_norm": 0.037999190390110016,
      "learning_rate": 9.800927188437416e-06,
      "loss": 0.0096,
      "step": 17970
    },
    {
      "epoch": 0.2941994600343614,
      "grad_norm": 0.3086879551410675,
      "learning_rate": 9.806381238069267e-06,
      "loss": 0.0072,
      "step": 17980
    },
    {
      "epoch": 0.2943630859854373,
      "grad_norm": 0.3267678916454315,
      "learning_rate": 9.81183528770112e-06,
      "loss": 0.0087,
      "step": 17990
    },
    {
      "epoch": 0.29452671193651314,
      "grad_norm": 0.2561754882335663,
      "learning_rate": 9.81728933733297e-06,
      "loss": 0.0058,
      "step": 18000
    },
    {
      "epoch": 0.29469033788758897,
      "grad_norm": 0.13376009464263916,
      "learning_rate": 9.822743386964823e-06,
      "loss": 0.0048,
      "step": 18010
    },
    {
      "epoch": 0.2948539638386648,
      "grad_norm": 0.4218680262565613,
      "learning_rate": 9.828197436596674e-06,
      "loss": 0.007,
      "step": 18020
    },
    {
      "epoch": 0.29501758978974063,
      "grad_norm": 0.28427305817604065,
      "learning_rate": 9.833651486228526e-06,
      "loss": 0.0096,
      "step": 18030
    },
    {
      "epoch": 0.2951812157408165,
      "grad_norm": 0.4316966235637665,
      "learning_rate": 9.839105535860377e-06,
      "loss": 0.0054,
      "step": 18040
    },
    {
      "epoch": 0.29534484169189235,
      "grad_norm": 0.5705351233482361,
      "learning_rate": 9.844559585492228e-06,
      "loss": 0.0058,
      "step": 18050
    },
    {
      "epoch": 0.2955084676429682,
      "grad_norm": 0.4316052794456482,
      "learning_rate": 9.85001363512408e-06,
      "loss": 0.0088,
      "step": 18060
    },
    {
      "epoch": 0.295672093594044,
      "grad_norm": 0.06517240405082703,
      "learning_rate": 9.855467684755931e-06,
      "loss": 0.0124,
      "step": 18070
    },
    {
      "epoch": 0.29583571954511984,
      "grad_norm": 0.3734110891819,
      "learning_rate": 9.860921734387784e-06,
      "loss": 0.007,
      "step": 18080
    },
    {
      "epoch": 0.2959993454961957,
      "grad_norm": 0.4664648175239563,
      "learning_rate": 9.866375784019635e-06,
      "loss": 0.0082,
      "step": 18090
    },
    {
      "epoch": 0.29616297144727155,
      "grad_norm": 0.5983517169952393,
      "learning_rate": 9.871829833651487e-06,
      "loss": 0.0089,
      "step": 18100
    },
    {
      "epoch": 0.2963265973983474,
      "grad_norm": 0.2989373803138733,
      "learning_rate": 9.877283883283338e-06,
      "loss": 0.0079,
      "step": 18110
    },
    {
      "epoch": 0.2964902233494232,
      "grad_norm": 0.5892401337623596,
      "learning_rate": 9.88273793291519e-06,
      "loss": 0.0095,
      "step": 18120
    },
    {
      "epoch": 0.29665384930049904,
      "grad_norm": 0.25288066267967224,
      "learning_rate": 9.888191982547041e-06,
      "loss": 0.0064,
      "step": 18130
    },
    {
      "epoch": 0.2968174752515749,
      "grad_norm": 0.20859119296073914,
      "learning_rate": 9.893646032178894e-06,
      "loss": 0.0087,
      "step": 18140
    },
    {
      "epoch": 0.29698110120265075,
      "grad_norm": 0.4292795956134796,
      "learning_rate": 9.899100081810745e-06,
      "loss": 0.0093,
      "step": 18150
    },
    {
      "epoch": 0.2971447271537266,
      "grad_norm": 0.3262891173362732,
      "learning_rate": 9.904554131442597e-06,
      "loss": 0.0079,
      "step": 18160
    },
    {
      "epoch": 0.2973083531048024,
      "grad_norm": 0.35146385431289673,
      "learning_rate": 9.910008181074448e-06,
      "loss": 0.008,
      "step": 18170
    },
    {
      "epoch": 0.29747197905587824,
      "grad_norm": 0.29222363233566284,
      "learning_rate": 9.915462230706299e-06,
      "loss": 0.0083,
      "step": 18180
    },
    {
      "epoch": 0.29763560500695413,
      "grad_norm": 0.4378841519355774,
      "learning_rate": 9.920916280338152e-06,
      "loss": 0.0082,
      "step": 18190
    },
    {
      "epoch": 0.29779923095802996,
      "grad_norm": 0.3232148587703705,
      "learning_rate": 9.926370329970002e-06,
      "loss": 0.0069,
      "step": 18200
    },
    {
      "epoch": 0.2979628569091058,
      "grad_norm": 0.3009403645992279,
      "learning_rate": 9.931824379601855e-06,
      "loss": 0.0058,
      "step": 18210
    },
    {
      "epoch": 0.2981264828601816,
      "grad_norm": 0.5111549496650696,
      "learning_rate": 9.937278429233706e-06,
      "loss": 0.0051,
      "step": 18220
    },
    {
      "epoch": 0.29829010881125745,
      "grad_norm": 0.3713594973087311,
      "learning_rate": 9.942732478865558e-06,
      "loss": 0.0049,
      "step": 18230
    },
    {
      "epoch": 0.29845373476233333,
      "grad_norm": 0.5472033619880676,
      "learning_rate": 9.94818652849741e-06,
      "loss": 0.0066,
      "step": 18240
    },
    {
      "epoch": 0.29861736071340916,
      "grad_norm": 0.4773631691932678,
      "learning_rate": 9.953640578129262e-06,
      "loss": 0.0055,
      "step": 18250
    },
    {
      "epoch": 0.298780986664485,
      "grad_norm": 0.29816052317619324,
      "learning_rate": 9.959094627761113e-06,
      "loss": 0.0073,
      "step": 18260
    },
    {
      "epoch": 0.2989446126155608,
      "grad_norm": 0.4285408854484558,
      "learning_rate": 9.964548677392965e-06,
      "loss": 0.0041,
      "step": 18270
    },
    {
      "epoch": 0.29910823856663665,
      "grad_norm": 0.5150445699691772,
      "learning_rate": 9.970002727024816e-06,
      "loss": 0.0056,
      "step": 18280
    },
    {
      "epoch": 0.2992718645177125,
      "grad_norm": 0.5367032885551453,
      "learning_rate": 9.975456776656669e-06,
      "loss": 0.0072,
      "step": 18290
    },
    {
      "epoch": 0.29943549046878837,
      "grad_norm": 0.5799115896224976,
      "learning_rate": 9.98091082628852e-06,
      "loss": 0.0062,
      "step": 18300
    },
    {
      "epoch": 0.2995991164198642,
      "grad_norm": 0.22263823449611664,
      "learning_rate": 9.986364875920372e-06,
      "loss": 0.007,
      "step": 18310
    },
    {
      "epoch": 0.29976274237094,
      "grad_norm": 0.2263379991054535,
      "learning_rate": 9.991818925552223e-06,
      "loss": 0.009,
      "step": 18320
    },
    {
      "epoch": 0.29992636832201586,
      "grad_norm": 0.4348183870315552,
      "learning_rate": 9.997272975184075e-06,
      "loss": 0.0078,
      "step": 18330
    },
    {
      "epoch": 0.3000899942730917,
      "grad_norm": 0.2248072326183319,
      "learning_rate": 9.999999977345253e-06,
      "loss": 0.0081,
      "step": 18340
    },
    {
      "epoch": 0.30025362022416757,
      "grad_norm": 0.5009114742279053,
      "learning_rate": 9.999999796107271e-06,
      "loss": 0.0062,
      "step": 18350
    },
    {
      "epoch": 0.3004172461752434,
      "grad_norm": 0.21323329210281372,
      "learning_rate": 9.999999433631314e-06,
      "loss": 0.0079,
      "step": 18360
    },
    {
      "epoch": 0.30058087212631923,
      "grad_norm": 0.28179433941841125,
      "learning_rate": 9.999998889917394e-06,
      "loss": 0.0065,
      "step": 18370
    },
    {
      "epoch": 0.30074449807739506,
      "grad_norm": 0.32057973742485046,
      "learning_rate": 9.999998164965532e-06,
      "loss": 0.0074,
      "step": 18380
    },
    {
      "epoch": 0.3009081240284709,
      "grad_norm": 0.43141409754753113,
      "learning_rate": 9.999997258775753e-06,
      "loss": 0.0073,
      "step": 18390
    },
    {
      "epoch": 0.3010717499795468,
      "grad_norm": 0.6593682765960693,
      "learning_rate": 9.99999617134809e-06,
      "loss": 0.0046,
      "step": 18400
    },
    {
      "epoch": 0.3012353759306226,
      "grad_norm": 0.21974030137062073,
      "learning_rate": 9.999994902682584e-06,
      "loss": 0.0056,
      "step": 18410
    },
    {
      "epoch": 0.30139900188169844,
      "grad_norm": 0.6519566774368286,
      "learning_rate": 9.99999345277928e-06,
      "loss": 0.0062,
      "step": 18420
    },
    {
      "epoch": 0.30156262783277427,
      "grad_norm": 0.14988838136196136,
      "learning_rate": 9.999991821638232e-06,
      "loss": 0.0101,
      "step": 18430
    },
    {
      "epoch": 0.3017262537838501,
      "grad_norm": 0.4918505847454071,
      "learning_rate": 9.999990009259495e-06,
      "loss": 0.007,
      "step": 18440
    },
    {
      "epoch": 0.301889879734926,
      "grad_norm": 0.33565282821655273,
      "learning_rate": 9.999988015643138e-06,
      "loss": 0.0059,
      "step": 18450
    },
    {
      "epoch": 0.3020535056860018,
      "grad_norm": 0.3530016541481018,
      "learning_rate": 9.999985840789232e-06,
      "loss": 0.0076,
      "step": 18460
    },
    {
      "epoch": 0.30221713163707764,
      "grad_norm": 0.2299710512161255,
      "learning_rate": 9.99998348469786e-06,
      "loss": 0.0076,
      "step": 18470
    },
    {
      "epoch": 0.30238075758815347,
      "grad_norm": 0.3344787657260895,
      "learning_rate": 9.9999809473691e-06,
      "loss": 0.0049,
      "step": 18480
    },
    {
      "epoch": 0.3025443835392293,
      "grad_norm": 0.6200084686279297,
      "learning_rate": 9.99997822880305e-06,
      "loss": 0.0077,
      "step": 18490
    },
    {
      "epoch": 0.3027080094903052,
      "grad_norm": 0.36080074310302734,
      "learning_rate": 9.999975328999805e-06,
      "loss": 0.0102,
      "step": 18500
    },
    {
      "epoch": 0.302871635441381,
      "grad_norm": 0.296271413564682,
      "learning_rate": 9.99997224795947e-06,
      "loss": 0.0062,
      "step": 18510
    },
    {
      "epoch": 0.30303526139245685,
      "grad_norm": 0.562339186668396,
      "learning_rate": 9.999968985682161e-06,
      "loss": 0.0091,
      "step": 18520
    },
    {
      "epoch": 0.3031988873435327,
      "grad_norm": 0.2859978973865509,
      "learning_rate": 9.999965542167992e-06,
      "loss": 0.006,
      "step": 18530
    },
    {
      "epoch": 0.3033625132946085,
      "grad_norm": 0.5382653474807739,
      "learning_rate": 9.99996191741709e-06,
      "loss": 0.0071,
      "step": 18540
    },
    {
      "epoch": 0.3035261392456844,
      "grad_norm": 0.16001670062541962,
      "learning_rate": 9.999958111429584e-06,
      "loss": 0.008,
      "step": 18550
    },
    {
      "epoch": 0.3036897651967602,
      "grad_norm": 0.5494171977043152,
      "learning_rate": 9.999954124205615e-06,
      "loss": 0.0079,
      "step": 18560
    },
    {
      "epoch": 0.30385339114783605,
      "grad_norm": 0.5336729288101196,
      "learning_rate": 9.999949955745324e-06,
      "loss": 0.0103,
      "step": 18570
    },
    {
      "epoch": 0.3040170170989119,
      "grad_norm": 0.15034836530685425,
      "learning_rate": 9.999945606048867e-06,
      "loss": 0.0065,
      "step": 18580
    },
    {
      "epoch": 0.3041806430499877,
      "grad_norm": 0.2007811814546585,
      "learning_rate": 9.999941075116399e-06,
      "loss": 0.0049,
      "step": 18590
    },
    {
      "epoch": 0.3043442690010636,
      "grad_norm": 0.2226434350013733,
      "learning_rate": 9.999936362948082e-06,
      "loss": 0.0066,
      "step": 18600
    },
    {
      "epoch": 0.3045078949521394,
      "grad_norm": 0.5107800364494324,
      "learning_rate": 9.99993146954409e-06,
      "loss": 0.0056,
      "step": 18610
    },
    {
      "epoch": 0.30467152090321525,
      "grad_norm": 0.5736075043678284,
      "learning_rate": 9.999926394904597e-06,
      "loss": 0.008,
      "step": 18620
    },
    {
      "epoch": 0.3048351468542911,
      "grad_norm": 0.35707661509513855,
      "learning_rate": 9.999921139029792e-06,
      "loss": 0.0094,
      "step": 18630
    },
    {
      "epoch": 0.3049987728053669,
      "grad_norm": 0.17102129757404327,
      "learning_rate": 9.999915701919862e-06,
      "loss": 0.0091,
      "step": 18640
    },
    {
      "epoch": 0.3051623987564428,
      "grad_norm": 0.3269667327404022,
      "learning_rate": 9.999910083575004e-06,
      "loss": 0.0065,
      "step": 18650
    },
    {
      "epoch": 0.30532602470751863,
      "grad_norm": 0.07784273475408554,
      "learning_rate": 9.999904283995422e-06,
      "loss": 0.0081,
      "step": 18660
    },
    {
      "epoch": 0.30548965065859446,
      "grad_norm": 0.31072136759757996,
      "learning_rate": 9.999898303181329e-06,
      "loss": 0.0063,
      "step": 18670
    },
    {
      "epoch": 0.3056532766096703,
      "grad_norm": 0.4528925120830536,
      "learning_rate": 9.999892141132938e-06,
      "loss": 0.0059,
      "step": 18680
    },
    {
      "epoch": 0.3058169025607461,
      "grad_norm": 0.42534682154655457,
      "learning_rate": 9.999885797850475e-06,
      "loss": 0.0065,
      "step": 18690
    },
    {
      "epoch": 0.30598052851182195,
      "grad_norm": 0.15108926594257355,
      "learning_rate": 9.999879273334166e-06,
      "loss": 0.0095,
      "step": 18700
    },
    {
      "epoch": 0.30614415446289783,
      "grad_norm": 0.4689083695411682,
      "learning_rate": 9.999872567584253e-06,
      "loss": 0.0069,
      "step": 18710
    },
    {
      "epoch": 0.30630778041397366,
      "grad_norm": 0.3014495074748993,
      "learning_rate": 9.999865680600975e-06,
      "loss": 0.0065,
      "step": 18720
    },
    {
      "epoch": 0.3064714063650495,
      "grad_norm": 0.2090315818786621,
      "learning_rate": 9.999858612384583e-06,
      "loss": 0.0098,
      "step": 18730
    },
    {
      "epoch": 0.3066350323161253,
      "grad_norm": 0.20829305052757263,
      "learning_rate": 9.999851362935334e-06,
      "loss": 0.0067,
      "step": 18740
    },
    {
      "epoch": 0.30679865826720115,
      "grad_norm": 0.3801248371601105,
      "learning_rate": 9.999843932253487e-06,
      "loss": 0.0067,
      "step": 18750
    },
    {
      "epoch": 0.30696228421827704,
      "grad_norm": 0.07702095061540604,
      "learning_rate": 9.999836320339318e-06,
      "loss": 0.0066,
      "step": 18760
    },
    {
      "epoch": 0.30712591016935287,
      "grad_norm": 0.9566939473152161,
      "learning_rate": 9.999828527193097e-06,
      "loss": 0.0051,
      "step": 18770
    },
    {
      "epoch": 0.3072895361204287,
      "grad_norm": 0.2995423674583435,
      "learning_rate": 9.99982055281511e-06,
      "loss": 0.0066,
      "step": 18780
    },
    {
      "epoch": 0.3074531620715045,
      "grad_norm": 0.5662979483604431,
      "learning_rate": 9.999812397205643e-06,
      "loss": 0.0077,
      "step": 18790
    },
    {
      "epoch": 0.30761678802258036,
      "grad_norm": 0.3930233120918274,
      "learning_rate": 9.999804060364995e-06,
      "loss": 0.0046,
      "step": 18800
    },
    {
      "epoch": 0.30778041397365624,
      "grad_norm": 0.3896386921405792,
      "learning_rate": 9.999795542293466e-06,
      "loss": 0.0096,
      "step": 18810
    },
    {
      "epoch": 0.30794403992473207,
      "grad_norm": 0.3319282531738281,
      "learning_rate": 9.999786842991366e-06,
      "loss": 0.0051,
      "step": 18820
    },
    {
      "epoch": 0.3081076658758079,
      "grad_norm": 0.5183929204940796,
      "learning_rate": 9.999777962459007e-06,
      "loss": 0.0058,
      "step": 18830
    },
    {
      "epoch": 0.30827129182688373,
      "grad_norm": 0.262056827545166,
      "learning_rate": 9.999768900696716e-06,
      "loss": 0.0059,
      "step": 18840
    },
    {
      "epoch": 0.30843491777795956,
      "grad_norm": 0.43029534816741943,
      "learning_rate": 9.999759657704817e-06,
      "loss": 0.0071,
      "step": 18850
    },
    {
      "epoch": 0.30859854372903545,
      "grad_norm": 0.21654173731803894,
      "learning_rate": 9.99975023348365e-06,
      "loss": 0.0053,
      "step": 18860
    },
    {
      "epoch": 0.3087621696801113,
      "grad_norm": 0.3761322498321533,
      "learning_rate": 9.999740628033552e-06,
      "loss": 0.0083,
      "step": 18870
    },
    {
      "epoch": 0.3089257956311871,
      "grad_norm": 0.04160545766353607,
      "learning_rate": 9.999730841354871e-06,
      "loss": 0.0034,
      "step": 18880
    },
    {
      "epoch": 0.30908942158226294,
      "grad_norm": 0.36300981044769287,
      "learning_rate": 9.999720873447966e-06,
      "loss": 0.0097,
      "step": 18890
    },
    {
      "epoch": 0.30925304753333877,
      "grad_norm": 0.4820455312728882,
      "learning_rate": 9.999710724313195e-06,
      "loss": 0.0086,
      "step": 18900
    },
    {
      "epoch": 0.30941667348441465,
      "grad_norm": 0.3691643476486206,
      "learning_rate": 9.999700393950926e-06,
      "loss": 0.0073,
      "step": 18910
    },
    {
      "epoch": 0.3095802994354905,
      "grad_norm": 0.46267586946487427,
      "learning_rate": 9.999689882361536e-06,
      "loss": 0.005,
      "step": 18920
    },
    {
      "epoch": 0.3097439253865663,
      "grad_norm": 0.5345110297203064,
      "learning_rate": 9.999679189545402e-06,
      "loss": 0.0044,
      "step": 18930
    },
    {
      "epoch": 0.30990755133764214,
      "grad_norm": 0.34804192185401917,
      "learning_rate": 9.999668315502914e-06,
      "loss": 0.0067,
      "step": 18940
    },
    {
      "epoch": 0.31007117728871797,
      "grad_norm": 0.29948872327804565,
      "learning_rate": 9.999657260234469e-06,
      "loss": 0.0065,
      "step": 18950
    },
    {
      "epoch": 0.31023480323979385,
      "grad_norm": 0.5881202220916748,
      "learning_rate": 9.999646023740462e-06,
      "loss": 0.0086,
      "step": 18960
    },
    {
      "epoch": 0.3103984291908697,
      "grad_norm": 0.43824076652526855,
      "learning_rate": 9.999634606021304e-06,
      "loss": 0.006,
      "step": 18970
    },
    {
      "epoch": 0.3105620551419455,
      "grad_norm": 0.3577647805213928,
      "learning_rate": 9.999623007077407e-06,
      "loss": 0.0078,
      "step": 18980
    },
    {
      "epoch": 0.31072568109302134,
      "grad_norm": 0.2878649830818176,
      "learning_rate": 9.999611226909193e-06,
      "loss": 0.0078,
      "step": 18990
    },
    {
      "epoch": 0.3108893070440972,
      "grad_norm": 0.3834591209888458,
      "learning_rate": 9.999599265517087e-06,
      "loss": 0.0058,
      "step": 19000
    },
    {
      "epoch": 0.31105293299517306,
      "grad_norm": 0.17713794112205505,
      "learning_rate": 9.999587122901526e-06,
      "loss": 0.0068,
      "step": 19010
    },
    {
      "epoch": 0.3112165589462489,
      "grad_norm": 0.5166522264480591,
      "learning_rate": 9.999574799062946e-06,
      "loss": 0.0071,
      "step": 19020
    },
    {
      "epoch": 0.3113801848973247,
      "grad_norm": 0.5418213605880737,
      "learning_rate": 9.999562294001796e-06,
      "loss": 0.0099,
      "step": 19030
    },
    {
      "epoch": 0.31154381084840055,
      "grad_norm": 0.32879868149757385,
      "learning_rate": 9.99954960771853e-06,
      "loss": 0.0052,
      "step": 19040
    },
    {
      "epoch": 0.3117074367994764,
      "grad_norm": 0.2878336012363434,
      "learning_rate": 9.999536740213607e-06,
      "loss": 0.0063,
      "step": 19050
    },
    {
      "epoch": 0.31187106275055226,
      "grad_norm": 0.28318139910697937,
      "learning_rate": 9.999523691487492e-06,
      "loss": 0.0059,
      "step": 19060
    },
    {
      "epoch": 0.3120346887016281,
      "grad_norm": 0.3776904344558716,
      "learning_rate": 9.99951046154066e-06,
      "loss": 0.008,
      "step": 19070
    },
    {
      "epoch": 0.3121983146527039,
      "grad_norm": 0.3345799446105957,
      "learning_rate": 9.99949705037359e-06,
      "loss": 0.0101,
      "step": 19080
    },
    {
      "epoch": 0.31236194060377975,
      "grad_norm": 0.4610148072242737,
      "learning_rate": 9.999483457986767e-06,
      "loss": 0.006,
      "step": 19090
    },
    {
      "epoch": 0.3125255665548556,
      "grad_norm": 0.20774437487125397,
      "learning_rate": 9.999469684380685e-06,
      "loss": 0.0042,
      "step": 19100
    },
    {
      "epoch": 0.31268919250593147,
      "grad_norm": 0.5823769569396973,
      "learning_rate": 9.999455729555842e-06,
      "loss": 0.0065,
      "step": 19110
    },
    {
      "epoch": 0.3128528184570073,
      "grad_norm": 0.28292059898376465,
      "learning_rate": 9.999441593512747e-06,
      "loss": 0.0082,
      "step": 19120
    },
    {
      "epoch": 0.3130164444080831,
      "grad_norm": 0.7109828591346741,
      "learning_rate": 9.999427276251907e-06,
      "loss": 0.0061,
      "step": 19130
    },
    {
      "epoch": 0.31318007035915896,
      "grad_norm": 0.09907187521457672,
      "learning_rate": 9.999412777773845e-06,
      "loss": 0.0044,
      "step": 19140
    },
    {
      "epoch": 0.3133436963102348,
      "grad_norm": 0.28128212690353394,
      "learning_rate": 9.999398098079087e-06,
      "loss": 0.0066,
      "step": 19150
    },
    {
      "epoch": 0.3135073222613106,
      "grad_norm": 0.16690900921821594,
      "learning_rate": 9.999383237168163e-06,
      "loss": 0.0068,
      "step": 19160
    },
    {
      "epoch": 0.3136709482123865,
      "grad_norm": 0.44006332755088806,
      "learning_rate": 9.999368195041612e-06,
      "loss": 0.0077,
      "step": 19170
    },
    {
      "epoch": 0.31383457416346233,
      "grad_norm": 0.26346489787101746,
      "learning_rate": 9.99935297169998e-06,
      "loss": 0.0062,
      "step": 19180
    },
    {
      "epoch": 0.31399820011453816,
      "grad_norm": 0.22483132779598236,
      "learning_rate": 9.999337567143819e-06,
      "loss": 0.0071,
      "step": 19190
    },
    {
      "epoch": 0.314161826065614,
      "grad_norm": 0.3907037675380707,
      "learning_rate": 9.999321981373686e-06,
      "loss": 0.0079,
      "step": 19200
    },
    {
      "epoch": 0.3143254520166898,
      "grad_norm": 0.4094851315021515,
      "learning_rate": 9.999306214390147e-06,
      "loss": 0.006,
      "step": 19210
    },
    {
      "epoch": 0.3144890779677657,
      "grad_norm": 0.47151899337768555,
      "learning_rate": 9.999290266193773e-06,
      "loss": 0.0094,
      "step": 19220
    },
    {
      "epoch": 0.31465270391884154,
      "grad_norm": 0.33708736300468445,
      "learning_rate": 9.999274136785141e-06,
      "loss": 0.0065,
      "step": 19230
    },
    {
      "epoch": 0.31481632986991737,
      "grad_norm": 0.7017020583152771,
      "learning_rate": 9.999257826164839e-06,
      "loss": 0.0059,
      "step": 19240
    },
    {
      "epoch": 0.3149799558209932,
      "grad_norm": 0.4297424256801605,
      "learning_rate": 9.999241334333457e-06,
      "loss": 0.0071,
      "step": 19250
    },
    {
      "epoch": 0.315143581772069,
      "grad_norm": 0.26415255665779114,
      "learning_rate": 9.999224661291588e-06,
      "loss": 0.0075,
      "step": 19260
    },
    {
      "epoch": 0.3153072077231449,
      "grad_norm": 0.42910659313201904,
      "learning_rate": 9.999207807039843e-06,
      "loss": 0.0064,
      "step": 19270
    },
    {
      "epoch": 0.31547083367422074,
      "grad_norm": 0.17327217757701874,
      "learning_rate": 9.99919077157883e-06,
      "loss": 0.0054,
      "step": 19280
    },
    {
      "epoch": 0.31563445962529657,
      "grad_norm": 0.4476442039012909,
      "learning_rate": 9.999173554909167e-06,
      "loss": 0.0049,
      "step": 19290
    },
    {
      "epoch": 0.3157980855763724,
      "grad_norm": 0.36961737275123596,
      "learning_rate": 9.999156157031477e-06,
      "loss": 0.0075,
      "step": 19300
    },
    {
      "epoch": 0.31596171152744823,
      "grad_norm": 0.5618547797203064,
      "learning_rate": 9.999138577946393e-06,
      "loss": 0.0092,
      "step": 19310
    },
    {
      "epoch": 0.3161253374785241,
      "grad_norm": 0.3292243480682373,
      "learning_rate": 9.99912081765455e-06,
      "loss": 0.0051,
      "step": 19320
    },
    {
      "epoch": 0.31628896342959995,
      "grad_norm": 0.3421551287174225,
      "learning_rate": 9.999102876156592e-06,
      "loss": 0.0053,
      "step": 19330
    },
    {
      "epoch": 0.3164525893806758,
      "grad_norm": 0.36465516686439514,
      "learning_rate": 9.999084753453171e-06,
      "loss": 0.0082,
      "step": 19340
    },
    {
      "epoch": 0.3166162153317516,
      "grad_norm": 0.5549119710922241,
      "learning_rate": 9.999066449544941e-06,
      "loss": 0.0074,
      "step": 19350
    },
    {
      "epoch": 0.31677984128282743,
      "grad_norm": 0.33259063959121704,
      "learning_rate": 9.999047964432568e-06,
      "loss": 0.006,
      "step": 19360
    },
    {
      "epoch": 0.3169434672339033,
      "grad_norm": 0.31920158863067627,
      "learning_rate": 9.999029298116722e-06,
      "loss": 0.0075,
      "step": 19370
    },
    {
      "epoch": 0.31710709318497915,
      "grad_norm": 0.5421958565711975,
      "learning_rate": 9.999010450598076e-06,
      "loss": 0.0096,
      "step": 19380
    },
    {
      "epoch": 0.317270719136055,
      "grad_norm": 0.4390092194080353,
      "learning_rate": 9.99899142187732e-06,
      "loss": 0.0071,
      "step": 19390
    },
    {
      "epoch": 0.3174343450871308,
      "grad_norm": 0.7274793982505798,
      "learning_rate": 9.998972211955137e-06,
      "loss": 0.0063,
      "step": 19400
    },
    {
      "epoch": 0.31759797103820664,
      "grad_norm": 0.4780811369419098,
      "learning_rate": 9.998952820832227e-06,
      "loss": 0.0054,
      "step": 19410
    },
    {
      "epoch": 0.3177615969892825,
      "grad_norm": 0.5134124159812927,
      "learning_rate": 9.99893324850929e-06,
      "loss": 0.006,
      "step": 19420
    },
    {
      "epoch": 0.31792522294035835,
      "grad_norm": 0.25927048921585083,
      "learning_rate": 9.99891349498704e-06,
      "loss": 0.0101,
      "step": 19430
    },
    {
      "epoch": 0.3180888488914342,
      "grad_norm": 0.3968285918235779,
      "learning_rate": 9.998893560266188e-06,
      "loss": 0.0069,
      "step": 19440
    },
    {
      "epoch": 0.31825247484251,
      "grad_norm": 0.48260748386383057,
      "learning_rate": 9.998873444347461e-06,
      "loss": 0.0072,
      "step": 19450
    },
    {
      "epoch": 0.31841610079358584,
      "grad_norm": 0.2677678167819977,
      "learning_rate": 9.998853147231585e-06,
      "loss": 0.0067,
      "step": 19460
    },
    {
      "epoch": 0.31857972674466173,
      "grad_norm": 0.36443576216697693,
      "learning_rate": 9.998832668919297e-06,
      "loss": 0.0073,
      "step": 19470
    },
    {
      "epoch": 0.31874335269573756,
      "grad_norm": 0.3514650762081146,
      "learning_rate": 9.99881200941134e-06,
      "loss": 0.0045,
      "step": 19480
    },
    {
      "epoch": 0.3189069786468134,
      "grad_norm": 0.3199397623538971,
      "learning_rate": 9.998791168708462e-06,
      "loss": 0.0084,
      "step": 19490
    },
    {
      "epoch": 0.3190706045978892,
      "grad_norm": 0.3068281412124634,
      "learning_rate": 9.998770146811416e-06,
      "loss": 0.0075,
      "step": 19500
    },
    {
      "epoch": 0.31923423054896505,
      "grad_norm": 0.24740198254585266,
      "learning_rate": 9.998748943720968e-06,
      "loss": 0.0067,
      "step": 19510
    },
    {
      "epoch": 0.31939785650004093,
      "grad_norm": 0.5325202345848083,
      "learning_rate": 9.998727559437886e-06,
      "loss": 0.0051,
      "step": 19520
    },
    {
      "epoch": 0.31956148245111676,
      "grad_norm": 0.19455644488334656,
      "learning_rate": 9.998705993962943e-06,
      "loss": 0.0068,
      "step": 19530
    },
    {
      "epoch": 0.3197251084021926,
      "grad_norm": 0.5948394536972046,
      "learning_rate": 9.998684247296921e-06,
      "loss": 0.0063,
      "step": 19540
    },
    {
      "epoch": 0.3198887343532684,
      "grad_norm": 0.7155441045761108,
      "learning_rate": 9.998662319440611e-06,
      "loss": 0.0048,
      "step": 19550
    },
    {
      "epoch": 0.32005236030434425,
      "grad_norm": 0.6412357091903687,
      "learning_rate": 9.998640210394804e-06,
      "loss": 0.0044,
      "step": 19560
    },
    {
      "epoch": 0.3202159862554201,
      "grad_norm": 0.11753785610198975,
      "learning_rate": 9.998617920160304e-06,
      "loss": 0.0051,
      "step": 19570
    },
    {
      "epoch": 0.32037961220649597,
      "grad_norm": 0.3821144700050354,
      "learning_rate": 9.998595448737919e-06,
      "loss": 0.0089,
      "step": 19580
    },
    {
      "epoch": 0.3205432381575718,
      "grad_norm": 0.49600300192832947,
      "learning_rate": 9.998572796128462e-06,
      "loss": 0.0061,
      "step": 19590
    },
    {
      "epoch": 0.3207068641086476,
      "grad_norm": 0.24403300881385803,
      "learning_rate": 9.998549962332756e-06,
      "loss": 0.0072,
      "step": 19600
    },
    {
      "epoch": 0.32087049005972346,
      "grad_norm": 0.3227308392524719,
      "learning_rate": 9.998526947351625e-06,
      "loss": 0.0073,
      "step": 19610
    },
    {
      "epoch": 0.3210341160107993,
      "grad_norm": 0.2587415874004364,
      "learning_rate": 9.998503751185908e-06,
      "loss": 0.0073,
      "step": 19620
    },
    {
      "epoch": 0.32119774196187517,
      "grad_norm": 0.46121665835380554,
      "learning_rate": 9.998480373836444e-06,
      "loss": 0.0094,
      "step": 19630
    },
    {
      "epoch": 0.321361367912951,
      "grad_norm": 0.12452620267868042,
      "learning_rate": 9.998456815304078e-06,
      "loss": 0.0053,
      "step": 19640
    },
    {
      "epoch": 0.32152499386402683,
      "grad_norm": 0.6648384928703308,
      "learning_rate": 9.998433075589667e-06,
      "loss": 0.0083,
      "step": 19650
    },
    {
      "epoch": 0.32168861981510266,
      "grad_norm": 0.5547574758529663,
      "learning_rate": 9.998409154694071e-06,
      "loss": 0.0061,
      "step": 19660
    },
    {
      "epoch": 0.3218522457661785,
      "grad_norm": 0.28415676951408386,
      "learning_rate": 9.998385052618156e-06,
      "loss": 0.0067,
      "step": 19670
    },
    {
      "epoch": 0.3220158717172544,
      "grad_norm": 0.3386670649051666,
      "learning_rate": 9.998360769362795e-06,
      "loss": 0.0075,
      "step": 19680
    },
    {
      "epoch": 0.3221794976683302,
      "grad_norm": 0.7446482181549072,
      "learning_rate": 9.998336304928872e-06,
      "loss": 0.0079,
      "step": 19690
    },
    {
      "epoch": 0.32234312361940604,
      "grad_norm": 0.3214438855648041,
      "learning_rate": 9.998311659317269e-06,
      "loss": 0.0047,
      "step": 19700
    },
    {
      "epoch": 0.32250674957048187,
      "grad_norm": 0.2506428062915802,
      "learning_rate": 9.998286832528882e-06,
      "loss": 0.0038,
      "step": 19710
    },
    {
      "epoch": 0.3226703755215577,
      "grad_norm": 0.12644913792610168,
      "learning_rate": 9.99826182456461e-06,
      "loss": 0.0056,
      "step": 19720
    },
    {
      "epoch": 0.3228340014726336,
      "grad_norm": 0.27583956718444824,
      "learning_rate": 9.99823663542536e-06,
      "loss": 0.0042,
      "step": 19730
    },
    {
      "epoch": 0.3229976274237094,
      "grad_norm": 0.22835558652877808,
      "learning_rate": 9.998211265112046e-06,
      "loss": 0.0065,
      "step": 19740
    },
    {
      "epoch": 0.32316125337478524,
      "grad_norm": 0.2603740692138672,
      "learning_rate": 9.998185713625586e-06,
      "loss": 0.0085,
      "step": 19750
    },
    {
      "epoch": 0.32332487932586107,
      "grad_norm": 0.20266957581043243,
      "learning_rate": 9.998159980966906e-06,
      "loss": 0.0048,
      "step": 19760
    },
    {
      "epoch": 0.3234885052769369,
      "grad_norm": 0.35531818866729736,
      "learning_rate": 9.998134067136939e-06,
      "loss": 0.0063,
      "step": 19770
    },
    {
      "epoch": 0.3236521312280128,
      "grad_norm": 0.21686922013759613,
      "learning_rate": 9.998107972136626e-06,
      "loss": 0.0032,
      "step": 19780
    },
    {
      "epoch": 0.3238157571790886,
      "grad_norm": 0.5397780537605286,
      "learning_rate": 9.998081695966913e-06,
      "loss": 0.0062,
      "step": 19790
    },
    {
      "epoch": 0.32397938313016444,
      "grad_norm": 0.5007716417312622,
      "learning_rate": 9.998055238628748e-06,
      "loss": 0.0084,
      "step": 19800
    },
    {
      "epoch": 0.3241430090812403,
      "grad_norm": 0.40624916553497314,
      "learning_rate": 9.998028600123095e-06,
      "loss": 0.006,
      "step": 19810
    },
    {
      "epoch": 0.3243066350323161,
      "grad_norm": 0.17299938201904297,
      "learning_rate": 9.998001780450917e-06,
      "loss": 0.0058,
      "step": 19820
    },
    {
      "epoch": 0.324470260983392,
      "grad_norm": 0.2824023365974426,
      "learning_rate": 9.997974779613186e-06,
      "loss": 0.0062,
      "step": 19830
    },
    {
      "epoch": 0.3246338869344678,
      "grad_norm": 0.43262213468551636,
      "learning_rate": 9.997947597610884e-06,
      "loss": 0.0061,
      "step": 19840
    },
    {
      "epoch": 0.32479751288554365,
      "grad_norm": 0.25371748208999634,
      "learning_rate": 9.99792023444499e-06,
      "loss": 0.0054,
      "step": 19850
    },
    {
      "epoch": 0.3249611388366195,
      "grad_norm": 0.2944333255290985,
      "learning_rate": 9.997892690116503e-06,
      "loss": 0.0069,
      "step": 19860
    },
    {
      "epoch": 0.3251247647876953,
      "grad_norm": 0.4182451665401459,
      "learning_rate": 9.997864964626417e-06,
      "loss": 0.0067,
      "step": 19870
    },
    {
      "epoch": 0.3252883907387712,
      "grad_norm": 0.3526301681995392,
      "learning_rate": 9.997837057975738e-06,
      "loss": 0.009,
      "step": 19880
    },
    {
      "epoch": 0.325452016689847,
      "grad_norm": 0.29576632380485535,
      "learning_rate": 9.99780897016548e-06,
      "loss": 0.0069,
      "step": 19890
    },
    {
      "epoch": 0.32561564264092285,
      "grad_norm": 0.5171969532966614,
      "learning_rate": 9.997780701196655e-06,
      "loss": 0.0081,
      "step": 19900
    },
    {
      "epoch": 0.3257792685919987,
      "grad_norm": 0.7963478565216064,
      "learning_rate": 9.997752251070294e-06,
      "loss": 0.0087,
      "step": 19910
    },
    {
      "epoch": 0.3259428945430745,
      "grad_norm": 0.32467925548553467,
      "learning_rate": 9.997723619787425e-06,
      "loss": 0.0064,
      "step": 19920
    },
    {
      "epoch": 0.3261065204941504,
      "grad_norm": 0.5840205550193787,
      "learning_rate": 9.997694807349085e-06,
      "loss": 0.0064,
      "step": 19930
    },
    {
      "epoch": 0.32627014644522623,
      "grad_norm": 0.5618772506713867,
      "learning_rate": 9.997665813756323e-06,
      "loss": 0.0054,
      "step": 19940
    },
    {
      "epoch": 0.32643377239630206,
      "grad_norm": 0.28135111927986145,
      "learning_rate": 9.997636639010184e-06,
      "loss": 0.0079,
      "step": 19950
    },
    {
      "epoch": 0.3265973983473779,
      "grad_norm": 0.5040724873542786,
      "learning_rate": 9.99760728311173e-06,
      "loss": 0.0053,
      "step": 19960
    },
    {
      "epoch": 0.3267610242984537,
      "grad_norm": 0.36021214723587036,
      "learning_rate": 9.997577746062022e-06,
      "loss": 0.0077,
      "step": 19970
    },
    {
      "epoch": 0.3269246502495296,
      "grad_norm": 0.25864776968955994,
      "learning_rate": 9.997548027862132e-06,
      "loss": 0.0051,
      "step": 19980
    },
    {
      "epoch": 0.32708827620060543,
      "grad_norm": 0.38222116231918335,
      "learning_rate": 9.99751812851314e-06,
      "loss": 0.006,
      "step": 19990
    },
    {
      "epoch": 0.32725190215168126,
      "grad_norm": 0.4257170557975769,
      "learning_rate": 9.997488048016124e-06,
      "loss": 0.0099,
      "step": 20000
    },
    {
      "epoch": 0.3274155281027571,
      "grad_norm": 0.21867577731609344,
      "learning_rate": 9.997457786372179e-06,
      "loss": 0.0056,
      "step": 20010
    },
    {
      "epoch": 0.3275791540538329,
      "grad_norm": 0.4962421953678131,
      "learning_rate": 9.9974273435824e-06,
      "loss": 0.0066,
      "step": 20020
    },
    {
      "epoch": 0.32774278000490875,
      "grad_norm": 0.2248992621898651,
      "learning_rate": 9.997396719647889e-06,
      "loss": 0.0041,
      "step": 20030
    },
    {
      "epoch": 0.32790640595598464,
      "grad_norm": 0.33530429005622864,
      "learning_rate": 9.997365914569762e-06,
      "loss": 0.0076,
      "step": 20040
    },
    {
      "epoch": 0.32807003190706047,
      "grad_norm": 0.3895592987537384,
      "learning_rate": 9.997334928349128e-06,
      "loss": 0.0071,
      "step": 20050
    },
    {
      "epoch": 0.3282336578581363,
      "grad_norm": 0.5960682034492493,
      "learning_rate": 9.997303760987114e-06,
      "loss": 0.0094,
      "step": 20060
    },
    {
      "epoch": 0.3283972838092121,
      "grad_norm": 0.24708130955696106,
      "learning_rate": 9.99727241248485e-06,
      "loss": 0.0079,
      "step": 20070
    },
    {
      "epoch": 0.32856090976028796,
      "grad_norm": 0.21781188249588013,
      "learning_rate": 9.997240882843472e-06,
      "loss": 0.0036,
      "step": 20080
    },
    {
      "epoch": 0.32872453571136384,
      "grad_norm": 0.6580252647399902,
      "learning_rate": 9.997209172064123e-06,
      "loss": 0.0105,
      "step": 20090
    },
    {
      "epoch": 0.32888816166243967,
      "grad_norm": 0.22322967648506165,
      "learning_rate": 9.99717728014795e-06,
      "loss": 0.0059,
      "step": 20100
    },
    {
      "epoch": 0.3290517876135155,
      "grad_norm": 0.34823718667030334,
      "learning_rate": 9.997145207096113e-06,
      "loss": 0.0062,
      "step": 20110
    },
    {
      "epoch": 0.32921541356459133,
      "grad_norm": 0.4263283312320709,
      "learning_rate": 9.997112952909771e-06,
      "loss": 0.0089,
      "step": 20120
    },
    {
      "epoch": 0.32937903951566716,
      "grad_norm": 0.10307332128286362,
      "learning_rate": 9.997080517590096e-06,
      "loss": 0.0089,
      "step": 20130
    },
    {
      "epoch": 0.32954266546674305,
      "grad_norm": 0.5239238739013672,
      "learning_rate": 9.997047901138262e-06,
      "loss": 0.0079,
      "step": 20140
    },
    {
      "epoch": 0.3297062914178189,
      "grad_norm": 0.4170688986778259,
      "learning_rate": 9.997015103555452e-06,
      "loss": 0.0061,
      "step": 20150
    },
    {
      "epoch": 0.3298699173688947,
      "grad_norm": 0.13897143304347992,
      "learning_rate": 9.996982124842853e-06,
      "loss": 0.0083,
      "step": 20160
    },
    {
      "epoch": 0.33003354331997053,
      "grad_norm": 0.5043318867683411,
      "learning_rate": 9.996948965001662e-06,
      "loss": 0.0119,
      "step": 20170
    },
    {
      "epoch": 0.33019716927104636,
      "grad_norm": 0.4660276174545288,
      "learning_rate": 9.996915624033082e-06,
      "loss": 0.0054,
      "step": 20180
    },
    {
      "epoch": 0.33036079522212225,
      "grad_norm": 0.17434312403202057,
      "learning_rate": 9.996882101938321e-06,
      "loss": 0.0068,
      "step": 20190
    },
    {
      "epoch": 0.3305244211731981,
      "grad_norm": 0.5469229221343994,
      "learning_rate": 9.996848398718593e-06,
      "loss": 0.0064,
      "step": 20200
    },
    {
      "epoch": 0.3306880471242739,
      "grad_norm": 0.2864542007446289,
      "learning_rate": 9.99681451437512e-06,
      "loss": 0.0081,
      "step": 20210
    },
    {
      "epoch": 0.33085167307534974,
      "grad_norm": 0.4709395170211792,
      "learning_rate": 9.99678044890913e-06,
      "loss": 0.0094,
      "step": 20220
    },
    {
      "epoch": 0.33101529902642557,
      "grad_norm": 0.5131140351295471,
      "learning_rate": 9.996746202321857e-06,
      "loss": 0.0057,
      "step": 20230
    },
    {
      "epoch": 0.33117892497750145,
      "grad_norm": 0.3796265125274658,
      "learning_rate": 9.996711774614545e-06,
      "loss": 0.0037,
      "step": 20240
    },
    {
      "epoch": 0.3313425509285773,
      "grad_norm": 0.4387819766998291,
      "learning_rate": 9.996677165788442e-06,
      "loss": 0.0075,
      "step": 20250
    },
    {
      "epoch": 0.3315061768796531,
      "grad_norm": 0.16999395191669464,
      "learning_rate": 9.996642375844799e-06,
      "loss": 0.0061,
      "step": 20260
    },
    {
      "epoch": 0.33166980283072894,
      "grad_norm": 0.3768167495727539,
      "learning_rate": 9.996607404784878e-06,
      "loss": 0.0044,
      "step": 20270
    },
    {
      "epoch": 0.3318334287818048,
      "grad_norm": 0.2779279053211212,
      "learning_rate": 9.996572252609948e-06,
      "loss": 0.0062,
      "step": 20280
    },
    {
      "epoch": 0.33199705473288066,
      "grad_norm": 0.2242118865251541,
      "learning_rate": 9.996536919321285e-06,
      "loss": 0.0053,
      "step": 20290
    },
    {
      "epoch": 0.3321606806839565,
      "grad_norm": 0.17876176536083221,
      "learning_rate": 9.996501404920166e-06,
      "loss": 0.0091,
      "step": 20300
    },
    {
      "epoch": 0.3323243066350323,
      "grad_norm": 0.3561529219150543,
      "learning_rate": 9.99646570940788e-06,
      "loss": 0.0063,
      "step": 20310
    },
    {
      "epoch": 0.33248793258610815,
      "grad_norm": 0.47531288862228394,
      "learning_rate": 9.996429832785722e-06,
      "loss": 0.0042,
      "step": 20320
    },
    {
      "epoch": 0.332651558537184,
      "grad_norm": 0.602146565914154,
      "learning_rate": 9.996393775054988e-06,
      "loss": 0.0073,
      "step": 20330
    },
    {
      "epoch": 0.33281518448825986,
      "grad_norm": 0.31655630469322205,
      "learning_rate": 9.996357536216991e-06,
      "loss": 0.0076,
      "step": 20340
    },
    {
      "epoch": 0.3329788104393357,
      "grad_norm": 0.3042921721935272,
      "learning_rate": 9.99632111627304e-06,
      "loss": 0.0048,
      "step": 20350
    },
    {
      "epoch": 0.3331424363904115,
      "grad_norm": 0.14022202789783478,
      "learning_rate": 9.996284515224459e-06,
      "loss": 0.0061,
      "step": 20360
    },
    {
      "epoch": 0.33330606234148735,
      "grad_norm": 0.5969071388244629,
      "learning_rate": 9.99624773307257e-06,
      "loss": 0.0057,
      "step": 20370
    },
    {
      "epoch": 0.3334696882925632,
      "grad_norm": 0.9815368056297302,
      "learning_rate": 9.996210769818711e-06,
      "loss": 0.006,
      "step": 20380
    },
    {
      "epoch": 0.33363331424363907,
      "grad_norm": 0.11673838645219803,
      "learning_rate": 9.996173625464218e-06,
      "loss": 0.0066,
      "step": 20390
    },
    {
      "epoch": 0.3337969401947149,
      "grad_norm": 0.23781123757362366,
      "learning_rate": 9.996136300010441e-06,
      "loss": 0.0055,
      "step": 20400
    },
    {
      "epoch": 0.3339605661457907,
      "grad_norm": 0.20059452950954437,
      "learning_rate": 9.99609879345873e-06,
      "loss": 0.0058,
      "step": 20410
    },
    {
      "epoch": 0.33412419209686656,
      "grad_norm": 0.28475067019462585,
      "learning_rate": 9.996061105810445e-06,
      "loss": 0.0085,
      "step": 20420
    },
    {
      "epoch": 0.3342878180479424,
      "grad_norm": 0.27808523178100586,
      "learning_rate": 9.996023237066953e-06,
      "loss": 0.0081,
      "step": 20430
    },
    {
      "epoch": 0.33445144399901827,
      "grad_norm": 0.37142521142959595,
      "learning_rate": 9.995985187229626e-06,
      "loss": 0.006,
      "step": 20440
    },
    {
      "epoch": 0.3346150699500941,
      "grad_norm": 0.5273996591567993,
      "learning_rate": 9.995946956299845e-06,
      "loss": 0.0059,
      "step": 20450
    },
    {
      "epoch": 0.33477869590116993,
      "grad_norm": 0.022713884711265564,
      "learning_rate": 9.995908544278993e-06,
      "loss": 0.0045,
      "step": 20460
    },
    {
      "epoch": 0.33494232185224576,
      "grad_norm": 0.248569056391716,
      "learning_rate": 9.995869951168462e-06,
      "loss": 0.0047,
      "step": 20470
    },
    {
      "epoch": 0.3351059478033216,
      "grad_norm": 0.43785926699638367,
      "learning_rate": 9.995831176969655e-06,
      "loss": 0.0132,
      "step": 20480
    },
    {
      "epoch": 0.3352695737543974,
      "grad_norm": 0.6996095776557922,
      "learning_rate": 9.995792221683975e-06,
      "loss": 0.0059,
      "step": 20490
    },
    {
      "epoch": 0.3354331997054733,
      "grad_norm": 0.5188438892364502,
      "learning_rate": 9.995753085312833e-06,
      "loss": 0.0072,
      "step": 20500
    },
    {
      "epoch": 0.33559682565654914,
      "grad_norm": 0.3539886772632599,
      "learning_rate": 9.99571376785765e-06,
      "loss": 0.0067,
      "step": 20510
    },
    {
      "epoch": 0.33576045160762497,
      "grad_norm": 0.29389485716819763,
      "learning_rate": 9.995674269319849e-06,
      "loss": 0.0072,
      "step": 20520
    },
    {
      "epoch": 0.3359240775587008,
      "grad_norm": 0.24048228561878204,
      "learning_rate": 9.995634589700861e-06,
      "loss": 0.0101,
      "step": 20530
    },
    {
      "epoch": 0.3360877035097766,
      "grad_norm": 0.39865291118621826,
      "learning_rate": 9.995594729002126e-06,
      "loss": 0.0079,
      "step": 20540
    },
    {
      "epoch": 0.3362513294608525,
      "grad_norm": 0.44991105794906616,
      "learning_rate": 9.995554687225092e-06,
      "loss": 0.0064,
      "step": 20550
    },
    {
      "epoch": 0.33641495541192834,
      "grad_norm": 0.279826283454895,
      "learning_rate": 9.995514464371203e-06,
      "loss": 0.0053,
      "step": 20560
    },
    {
      "epoch": 0.33657858136300417,
      "grad_norm": 0.1260901540517807,
      "learning_rate": 9.995474060441923e-06,
      "loss": 0.0047,
      "step": 20570
    },
    {
      "epoch": 0.33674220731408,
      "grad_norm": 0.30685099959373474,
      "learning_rate": 9.995433475438715e-06,
      "loss": 0.0077,
      "step": 20580
    },
    {
      "epoch": 0.33690583326515583,
      "grad_norm": 0.41488710045814514,
      "learning_rate": 9.99539270936305e-06,
      "loss": 0.0062,
      "step": 20590
    },
    {
      "epoch": 0.3370694592162317,
      "grad_norm": 3.008312225341797,
      "learning_rate": 9.995351762216404e-06,
      "loss": 0.0066,
      "step": 20600
    },
    {
      "epoch": 0.33723308516730754,
      "grad_norm": 0.38788747787475586,
      "learning_rate": 9.995310634000264e-06,
      "loss": 0.0053,
      "step": 20610
    },
    {
      "epoch": 0.3373967111183834,
      "grad_norm": 0.19717708230018616,
      "learning_rate": 9.995269324716119e-06,
      "loss": 0.0048,
      "step": 20620
    },
    {
      "epoch": 0.3375603370694592,
      "grad_norm": 0.19570958614349365,
      "learning_rate": 9.995227834365465e-06,
      "loss": 0.0048,
      "step": 20630
    },
    {
      "epoch": 0.33772396302053503,
      "grad_norm": 0.23481355607509613,
      "learning_rate": 9.99518616294981e-06,
      "loss": 0.0054,
      "step": 20640
    },
    {
      "epoch": 0.3378875889716109,
      "grad_norm": 0.26487621665000916,
      "learning_rate": 9.995144310470661e-06,
      "loss": 0.0076,
      "step": 20650
    },
    {
      "epoch": 0.33805121492268675,
      "grad_norm": 0.2517988085746765,
      "learning_rate": 9.995102276929537e-06,
      "loss": 0.0042,
      "step": 20660
    },
    {
      "epoch": 0.3382148408737626,
      "grad_norm": 0.21403658390045166,
      "learning_rate": 9.995060062327961e-06,
      "loss": 0.0047,
      "step": 20670
    },
    {
      "epoch": 0.3383784668248384,
      "grad_norm": 0.28809064626693726,
      "learning_rate": 9.99501766666746e-06,
      "loss": 0.0081,
      "step": 20680
    },
    {
      "epoch": 0.33854209277591424,
      "grad_norm": 0.2755778729915619,
      "learning_rate": 9.99497508994958e-06,
      "loss": 0.0049,
      "step": 20690
    },
    {
      "epoch": 0.3387057187269901,
      "grad_norm": 0.34083205461502075,
      "learning_rate": 9.994932332175853e-06,
      "loss": 0.0076,
      "step": 20700
    },
    {
      "epoch": 0.33886934467806595,
      "grad_norm": 0.29603058099746704,
      "learning_rate": 9.994889393347835e-06,
      "loss": 0.0058,
      "step": 20710
    },
    {
      "epoch": 0.3390329706291418,
      "grad_norm": 0.0946914330124855,
      "learning_rate": 9.994846273467083e-06,
      "loss": 0.0047,
      "step": 20720
    },
    {
      "epoch": 0.3391965965802176,
      "grad_norm": 0.6166210770606995,
      "learning_rate": 9.994802972535158e-06,
      "loss": 0.008,
      "step": 20730
    },
    {
      "epoch": 0.33936022253129344,
      "grad_norm": 0.16732168197631836,
      "learning_rate": 9.994759490553626e-06,
      "loss": 0.0064,
      "step": 20740
    },
    {
      "epoch": 0.33952384848236933,
      "grad_norm": 0.3816123306751251,
      "learning_rate": 9.994715827524073e-06,
      "loss": 0.0051,
      "step": 20750
    },
    {
      "epoch": 0.33968747443344516,
      "grad_norm": 0.5703874826431274,
      "learning_rate": 9.994671983448073e-06,
      "loss": 0.0059,
      "step": 20760
    },
    {
      "epoch": 0.339851100384521,
      "grad_norm": 0.16485288739204407,
      "learning_rate": 9.994627958327216e-06,
      "loss": 0.005,
      "step": 20770
    },
    {
      "epoch": 0.3400147263355968,
      "grad_norm": 0.29808759689331055,
      "learning_rate": 9.994583752163103e-06,
      "loss": 0.0051,
      "step": 20780
    },
    {
      "epoch": 0.34017835228667265,
      "grad_norm": 0.2384519875049591,
      "learning_rate": 9.994539364957333e-06,
      "loss": 0.0053,
      "step": 20790
    },
    {
      "epoch": 0.34034197823774853,
      "grad_norm": 0.5800129771232605,
      "learning_rate": 9.994494796711512e-06,
      "loss": 0.0085,
      "step": 20800
    },
    {
      "epoch": 0.34050560418882436,
      "grad_norm": 0.1808997392654419,
      "learning_rate": 9.99445004742726e-06,
      "loss": 0.0044,
      "step": 20810
    },
    {
      "epoch": 0.3406692301399002,
      "grad_norm": 0.4030112326145172,
      "learning_rate": 9.9944051171062e-06,
      "loss": 0.0064,
      "step": 20820
    },
    {
      "epoch": 0.340832856090976,
      "grad_norm": 0.1567230373620987,
      "learning_rate": 9.994360005749955e-06,
      "loss": 0.0066,
      "step": 20830
    },
    {
      "epoch": 0.34099648204205185,
      "grad_norm": 0.2962917387485504,
      "learning_rate": 9.994314713360166e-06,
      "loss": 0.0063,
      "step": 20840
    },
    {
      "epoch": 0.34116010799312774,
      "grad_norm": 0.2686673700809479,
      "learning_rate": 9.99426923993847e-06,
      "loss": 0.0094,
      "step": 20850
    },
    {
      "epoch": 0.34132373394420357,
      "grad_norm": 0.2440519779920578,
      "learning_rate": 9.994223585486518e-06,
      "loss": 0.0084,
      "step": 20860
    },
    {
      "epoch": 0.3414873598952794,
      "grad_norm": 0.1828407347202301,
      "learning_rate": 9.994177750005966e-06,
      "loss": 0.0057,
      "step": 20870
    },
    {
      "epoch": 0.3416509858463552,
      "grad_norm": 0.21303662657737732,
      "learning_rate": 9.994131733498472e-06,
      "loss": 0.0041,
      "step": 20880
    },
    {
      "epoch": 0.34181461179743106,
      "grad_norm": 0.306277334690094,
      "learning_rate": 9.994085535965707e-06,
      "loss": 0.0063,
      "step": 20890
    },
    {
      "epoch": 0.3419782377485069,
      "grad_norm": 0.3458658754825592,
      "learning_rate": 9.994039157409343e-06,
      "loss": 0.013,
      "step": 20900
    },
    {
      "epoch": 0.34214186369958277,
      "grad_norm": 0.7181628346443176,
      "learning_rate": 9.993992597831064e-06,
      "loss": 0.009,
      "step": 20910
    },
    {
      "epoch": 0.3423054896506586,
      "grad_norm": 0.33908718824386597,
      "learning_rate": 9.993945857232555e-06,
      "loss": 0.0073,
      "step": 20920
    },
    {
      "epoch": 0.34246911560173443,
      "grad_norm": 0.23480235040187836,
      "learning_rate": 9.993898935615511e-06,
      "loss": 0.007,
      "step": 20930
    },
    {
      "epoch": 0.34263274155281026,
      "grad_norm": 0.3750567138195038,
      "learning_rate": 9.993851832981633e-06,
      "loss": 0.0061,
      "step": 20940
    },
    {
      "epoch": 0.3427963675038861,
      "grad_norm": 0.4308174252510071,
      "learning_rate": 9.993804549332628e-06,
      "loss": 0.0046,
      "step": 20950
    },
    {
      "epoch": 0.342959993454962,
      "grad_norm": 0.20649023354053497,
      "learning_rate": 9.993757084670213e-06,
      "loss": 0.0065,
      "step": 20960
    },
    {
      "epoch": 0.3431236194060378,
      "grad_norm": 0.5056235194206238,
      "learning_rate": 9.993709438996103e-06,
      "loss": 0.0121,
      "step": 20970
    },
    {
      "epoch": 0.34328724535711364,
      "grad_norm": 0.1302957534790039,
      "learning_rate": 9.993661612312029e-06,
      "loss": 0.0061,
      "step": 20980
    },
    {
      "epoch": 0.34345087130818946,
      "grad_norm": 0.4491141140460968,
      "learning_rate": 9.993613604619723e-06,
      "loss": 0.0055,
      "step": 20990
    },
    {
      "epoch": 0.3436144972592653,
      "grad_norm": 0.2363044023513794,
      "learning_rate": 9.993565415920927e-06,
      "loss": 0.004,
      "step": 21000
    },
    {
      "epoch": 0.3437781232103412,
      "grad_norm": 0.2518073320388794,
      "learning_rate": 9.993517046217385e-06,
      "loss": 0.0072,
      "step": 21010
    },
    {
      "epoch": 0.343941749161417,
      "grad_norm": 0.28679320216178894,
      "learning_rate": 9.993468495510853e-06,
      "loss": 0.0048,
      "step": 21020
    },
    {
      "epoch": 0.34410537511249284,
      "grad_norm": 0.43081966042518616,
      "learning_rate": 9.99341976380309e-06,
      "loss": 0.0049,
      "step": 21030
    },
    {
      "epoch": 0.34426900106356867,
      "grad_norm": 0.19454362988471985,
      "learning_rate": 9.993370851095859e-06,
      "loss": 0.0062,
      "step": 21040
    },
    {
      "epoch": 0.3444326270146445,
      "grad_norm": 0.31099480390548706,
      "learning_rate": 9.993321757390938e-06,
      "loss": 0.0048,
      "step": 21050
    },
    {
      "epoch": 0.3445962529657204,
      "grad_norm": 0.27064645290374756,
      "learning_rate": 9.993272482690104e-06,
      "loss": 0.0058,
      "step": 21060
    },
    {
      "epoch": 0.3447598789167962,
      "grad_norm": 0.41544073820114136,
      "learning_rate": 9.993223026995144e-06,
      "loss": 0.007,
      "step": 21070
    },
    {
      "epoch": 0.34492350486787204,
      "grad_norm": 0.2886449694633484,
      "learning_rate": 9.99317339030785e-06,
      "loss": 0.0042,
      "step": 21080
    },
    {
      "epoch": 0.3450871308189479,
      "grad_norm": 0.19265399873256683,
      "learning_rate": 9.993123572630021e-06,
      "loss": 0.0034,
      "step": 21090
    },
    {
      "epoch": 0.3452507567700237,
      "grad_norm": 0.1833256483078003,
      "learning_rate": 9.993073573963463e-06,
      "loss": 0.0082,
      "step": 21100
    },
    {
      "epoch": 0.3454143827210996,
      "grad_norm": 0.29618486762046814,
      "learning_rate": 9.993023394309988e-06,
      "loss": 0.0068,
      "step": 21110
    },
    {
      "epoch": 0.3455780086721754,
      "grad_norm": 0.4258694648742676,
      "learning_rate": 9.992973033671417e-06,
      "loss": 0.0069,
      "step": 21120
    },
    {
      "epoch": 0.34574163462325125,
      "grad_norm": 0.6530909538269043,
      "learning_rate": 9.992922492049573e-06,
      "loss": 0.0074,
      "step": 21130
    },
    {
      "epoch": 0.3459052605743271,
      "grad_norm": 0.5368636250495911,
      "learning_rate": 9.992871769446289e-06,
      "loss": 0.006,
      "step": 21140
    },
    {
      "epoch": 0.3460688865254029,
      "grad_norm": 0.30355000495910645,
      "learning_rate": 9.992820865863403e-06,
      "loss": 0.0061,
      "step": 21150
    },
    {
      "epoch": 0.3462325124764788,
      "grad_norm": 0.46395808458328247,
      "learning_rate": 9.99276978130276e-06,
      "loss": 0.005,
      "step": 21160
    },
    {
      "epoch": 0.3463961384275546,
      "grad_norm": 0.1760774552822113,
      "learning_rate": 9.992718515766213e-06,
      "loss": 0.0084,
      "step": 21170
    },
    {
      "epoch": 0.34655976437863045,
      "grad_norm": 0.3363729417324066,
      "learning_rate": 9.99266706925562e-06,
      "loss": 0.0094,
      "step": 21180
    },
    {
      "epoch": 0.3467233903297063,
      "grad_norm": 0.2082325965166092,
      "learning_rate": 9.992615441772845e-06,
      "loss": 0.0063,
      "step": 21190
    },
    {
      "epoch": 0.3468870162807821,
      "grad_norm": 0.3925575911998749,
      "learning_rate": 9.992563633319759e-06,
      "loss": 0.0047,
      "step": 21200
    },
    {
      "epoch": 0.347050642231858,
      "grad_norm": 0.4477984607219696,
      "learning_rate": 9.992511643898242e-06,
      "loss": 0.009,
      "step": 21210
    },
    {
      "epoch": 0.3472142681829338,
      "grad_norm": 0.23646071553230286,
      "learning_rate": 9.992459473510175e-06,
      "loss": 0.0055,
      "step": 21220
    },
    {
      "epoch": 0.34737789413400966,
      "grad_norm": 0.28470760583877563,
      "learning_rate": 9.992407122157452e-06,
      "loss": 0.0063,
      "step": 21230
    },
    {
      "epoch": 0.3475415200850855,
      "grad_norm": 0.717832088470459,
      "learning_rate": 9.99235458984197e-06,
      "loss": 0.0043,
      "step": 21240
    },
    {
      "epoch": 0.3477051460361613,
      "grad_norm": 0.2909640073776245,
      "learning_rate": 9.992301876565634e-06,
      "loss": 0.009,
      "step": 21250
    },
    {
      "epoch": 0.3478687719872372,
      "grad_norm": 0.531152606010437,
      "learning_rate": 9.992248982330352e-06,
      "loss": 0.0067,
      "step": 21260
    },
    {
      "epoch": 0.34803239793831303,
      "grad_norm": 0.4054419994354248,
      "learning_rate": 9.992195907138045e-06,
      "loss": 0.0059,
      "step": 21270
    },
    {
      "epoch": 0.34819602388938886,
      "grad_norm": 0.23385083675384521,
      "learning_rate": 9.992142650990633e-06,
      "loss": 0.0043,
      "step": 21280
    },
    {
      "epoch": 0.3483596498404647,
      "grad_norm": 0.4649423658847809,
      "learning_rate": 9.992089213890048e-06,
      "loss": 0.0044,
      "step": 21290
    },
    {
      "epoch": 0.3485232757915405,
      "grad_norm": 0.3326658308506012,
      "learning_rate": 9.992035595838228e-06,
      "loss": 0.0048,
      "step": 21300
    },
    {
      "epoch": 0.3486869017426164,
      "grad_norm": 0.19111086428165436,
      "learning_rate": 9.991981796837116e-06,
      "loss": 0.0068,
      "step": 21310
    },
    {
      "epoch": 0.34885052769369224,
      "grad_norm": 0.2604122459888458,
      "learning_rate": 9.99192781688866e-06,
      "loss": 0.0061,
      "step": 21320
    },
    {
      "epoch": 0.34901415364476807,
      "grad_norm": 0.06770627200603485,
      "learning_rate": 9.99187365599482e-06,
      "loss": 0.005,
      "step": 21330
    },
    {
      "epoch": 0.3491777795958439,
      "grad_norm": 0.396420955657959,
      "learning_rate": 9.991819314157558e-06,
      "loss": 0.009,
      "step": 21340
    },
    {
      "epoch": 0.3493414055469197,
      "grad_norm": 0.31943750381469727,
      "learning_rate": 9.991764791378842e-06,
      "loss": 0.0046,
      "step": 21350
    },
    {
      "epoch": 0.34950503149799556,
      "grad_norm": 0.11120828986167908,
      "learning_rate": 9.99171008766065e-06,
      "loss": 0.0074,
      "step": 21360
    },
    {
      "epoch": 0.34966865744907144,
      "grad_norm": 0.17532435059547424,
      "learning_rate": 9.991655203004966e-06,
      "loss": 0.0062,
      "step": 21370
    },
    {
      "epoch": 0.34983228340014727,
      "grad_norm": 0.14777524769306183,
      "learning_rate": 9.991600137413777e-06,
      "loss": 0.0051,
      "step": 21380
    },
    {
      "epoch": 0.3499959093512231,
      "grad_norm": 0.19834992289543152,
      "learning_rate": 9.991544890889079e-06,
      "loss": 0.0125,
      "step": 21390
    },
    {
      "epoch": 0.35015953530229893,
      "grad_norm": 0.25837865471839905,
      "learning_rate": 9.991489463432877e-06,
      "loss": 0.0057,
      "step": 21400
    },
    {
      "epoch": 0.35032316125337476,
      "grad_norm": 0.4817623198032379,
      "learning_rate": 9.991433855047177e-06,
      "loss": 0.0068,
      "step": 21410
    },
    {
      "epoch": 0.35048678720445064,
      "grad_norm": 0.3779519200325012,
      "learning_rate": 9.991378065733999e-06,
      "loss": 0.006,
      "step": 21420
    },
    {
      "epoch": 0.3506504131555265,
      "grad_norm": 0.20369569957256317,
      "learning_rate": 9.991322095495359e-06,
      "loss": 0.005,
      "step": 21430
    },
    {
      "epoch": 0.3508140391066023,
      "grad_norm": 0.20326143503189087,
      "learning_rate": 9.991265944333292e-06,
      "loss": 0.0034,
      "step": 21440
    },
    {
      "epoch": 0.35097766505767813,
      "grad_norm": 0.19558410346508026,
      "learning_rate": 9.99120961224983e-06,
      "loss": 0.0091,
      "step": 21450
    },
    {
      "epoch": 0.35114129100875396,
      "grad_norm": 0.4241901636123657,
      "learning_rate": 9.991153099247018e-06,
      "loss": 0.0092,
      "step": 21460
    },
    {
      "epoch": 0.35130491695982985,
      "grad_norm": 0.5111272931098938,
      "learning_rate": 9.9910964053269e-06,
      "loss": 0.0064,
      "step": 21470
    },
    {
      "epoch": 0.3514685429109057,
      "grad_norm": 0.446339875459671,
      "learning_rate": 9.99103953049153e-06,
      "loss": 0.0055,
      "step": 21480
    },
    {
      "epoch": 0.3516321688619815,
      "grad_norm": 0.2024042308330536,
      "learning_rate": 9.990982474742977e-06,
      "loss": 0.0081,
      "step": 21490
    },
    {
      "epoch": 0.35179579481305734,
      "grad_norm": 0.30922695994377136,
      "learning_rate": 9.990925238083304e-06,
      "loss": 0.0069,
      "step": 21500
    },
    {
      "epoch": 0.35195942076413317,
      "grad_norm": 0.420624315738678,
      "learning_rate": 9.990867820514584e-06,
      "loss": 0.0058,
      "step": 21510
    },
    {
      "epoch": 0.35212304671520905,
      "grad_norm": 0.307954341173172,
      "learning_rate": 9.990810222038904e-06,
      "loss": 0.0042,
      "step": 21520
    },
    {
      "epoch": 0.3522866726662849,
      "grad_norm": 0.3861580789089203,
      "learning_rate": 9.990752442658347e-06,
      "loss": 0.006,
      "step": 21530
    },
    {
      "epoch": 0.3524502986173607,
      "grad_norm": 0.11174075305461884,
      "learning_rate": 9.99069448237501e-06,
      "loss": 0.0062,
      "step": 21540
    },
    {
      "epoch": 0.35261392456843654,
      "grad_norm": 0.13128244876861572,
      "learning_rate": 9.99063634119099e-06,
      "loss": 0.0042,
      "step": 21550
    },
    {
      "epoch": 0.3527775505195124,
      "grad_norm": 0.2859407067298889,
      "learning_rate": 9.990578019108401e-06,
      "loss": 0.0053,
      "step": 21560
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.303383469581604,
      "learning_rate": 9.99051951612935e-06,
      "loss": 0.0053,
      "step": 21570
    },
    {
      "epoch": 0.3531048024216641,
      "grad_norm": 0.25243061780929565,
      "learning_rate": 9.990460832255964e-06,
      "loss": 0.0066,
      "step": 21580
    },
    {
      "epoch": 0.3532684283727399,
      "grad_norm": 0.44662171602249146,
      "learning_rate": 9.990401967490366e-06,
      "loss": 0.0053,
      "step": 21590
    },
    {
      "epoch": 0.35343205432381575,
      "grad_norm": 0.48392048478126526,
      "learning_rate": 9.990342921834691e-06,
      "loss": 0.0049,
      "step": 21600
    },
    {
      "epoch": 0.3535956802748916,
      "grad_norm": 0.21543946862220764,
      "learning_rate": 9.990283695291077e-06,
      "loss": 0.0071,
      "step": 21610
    },
    {
      "epoch": 0.35375930622596746,
      "grad_norm": 0.37745505571365356,
      "learning_rate": 9.990224287861676e-06,
      "loss": 0.0073,
      "step": 21620
    },
    {
      "epoch": 0.3539229321770433,
      "grad_norm": 0.23981940746307373,
      "learning_rate": 9.990164699548636e-06,
      "loss": 0.0066,
      "step": 21630
    },
    {
      "epoch": 0.3540865581281191,
      "grad_norm": 0.26420408487319946,
      "learning_rate": 9.990104930354118e-06,
      "loss": 0.0081,
      "step": 21640
    },
    {
      "epoch": 0.35425018407919495,
      "grad_norm": 0.33763372898101807,
      "learning_rate": 9.990044980280293e-06,
      "loss": 0.0052,
      "step": 21650
    },
    {
      "epoch": 0.3544138100302708,
      "grad_norm": 0.4525817930698395,
      "learning_rate": 9.989984849329329e-06,
      "loss": 0.0116,
      "step": 21660
    },
    {
      "epoch": 0.35457743598134667,
      "grad_norm": 0.2738310396671295,
      "learning_rate": 9.989924537503408e-06,
      "loss": 0.0082,
      "step": 21670
    },
    {
      "epoch": 0.3547410619324225,
      "grad_norm": 0.15360699594020844,
      "learning_rate": 9.989864044804715e-06,
      "loss": 0.0047,
      "step": 21680
    },
    {
      "epoch": 0.3549046878834983,
      "grad_norm": 0.31382203102111816,
      "learning_rate": 9.989803371235442e-06,
      "loss": 0.0116,
      "step": 21690
    },
    {
      "epoch": 0.35506831383457416,
      "grad_norm": 0.14262357354164124,
      "learning_rate": 9.989742516797791e-06,
      "loss": 0.0052,
      "step": 21700
    },
    {
      "epoch": 0.35523193978565,
      "grad_norm": 0.4709635376930237,
      "learning_rate": 9.989681481493966e-06,
      "loss": 0.0064,
      "step": 21710
    },
    {
      "epoch": 0.35539556573672587,
      "grad_norm": 0.29232409596443176,
      "learning_rate": 9.98962026532618e-06,
      "loss": 0.0058,
      "step": 21720
    },
    {
      "epoch": 0.3555591916878017,
      "grad_norm": 0.20618091523647308,
      "learning_rate": 9.98955886829665e-06,
      "loss": 0.0056,
      "step": 21730
    },
    {
      "epoch": 0.35572281763887753,
      "grad_norm": 0.2509346306324005,
      "learning_rate": 9.989497290407606e-06,
      "loss": 0.0063,
      "step": 21740
    },
    {
      "epoch": 0.35588644358995336,
      "grad_norm": 0.26989296078681946,
      "learning_rate": 9.989435531661275e-06,
      "loss": 0.0061,
      "step": 21750
    },
    {
      "epoch": 0.3560500695410292,
      "grad_norm": 0.3324601352214813,
      "learning_rate": 9.989373592059898e-06,
      "loss": 0.0053,
      "step": 21760
    },
    {
      "epoch": 0.356213695492105,
      "grad_norm": 0.335764616727829,
      "learning_rate": 9.98931147160572e-06,
      "loss": 0.0079,
      "step": 21770
    },
    {
      "epoch": 0.3563773214431809,
      "grad_norm": 0.5803649425506592,
      "learning_rate": 9.989249170300993e-06,
      "loss": 0.0082,
      "step": 21780
    },
    {
      "epoch": 0.35654094739425674,
      "grad_norm": 0.19377391040325165,
      "learning_rate": 9.989186688147977e-06,
      "loss": 0.0038,
      "step": 21790
    },
    {
      "epoch": 0.35670457334533257,
      "grad_norm": 0.5861106514930725,
      "learning_rate": 9.989124025148931e-06,
      "loss": 0.0101,
      "step": 21800
    },
    {
      "epoch": 0.3568681992964084,
      "grad_norm": 0.39570486545562744,
      "learning_rate": 9.989061181306132e-06,
      "loss": 0.005,
      "step": 21810
    },
    {
      "epoch": 0.3570318252474842,
      "grad_norm": 0.13977734744548798,
      "learning_rate": 9.988998156621857e-06,
      "loss": 0.0063,
      "step": 21820
    },
    {
      "epoch": 0.3571954511985601,
      "grad_norm": 0.2916661500930786,
      "learning_rate": 9.988934951098388e-06,
      "loss": 0.0057,
      "step": 21830
    },
    {
      "epoch": 0.35735907714963594,
      "grad_norm": 0.16173802316188812,
      "learning_rate": 9.988871564738017e-06,
      "loss": 0.0042,
      "step": 21840
    },
    {
      "epoch": 0.35752270310071177,
      "grad_norm": 0.21764524281024933,
      "learning_rate": 9.988807997543044e-06,
      "loss": 0.0065,
      "step": 21850
    },
    {
      "epoch": 0.3576863290517876,
      "grad_norm": 0.22217422723770142,
      "learning_rate": 9.988744249515769e-06,
      "loss": 0.0043,
      "step": 21860
    },
    {
      "epoch": 0.35784995500286343,
      "grad_norm": 0.2889345586299896,
      "learning_rate": 9.988680320658508e-06,
      "loss": 0.0057,
      "step": 21870
    },
    {
      "epoch": 0.3580135809539393,
      "grad_norm": 0.3199993073940277,
      "learning_rate": 9.988616210973574e-06,
      "loss": 0.0044,
      "step": 21880
    },
    {
      "epoch": 0.35817720690501514,
      "grad_norm": 0.41283565759658813,
      "learning_rate": 9.988551920463293e-06,
      "loss": 0.0061,
      "step": 21890
    },
    {
      "epoch": 0.358340832856091,
      "grad_norm": 0.6704068779945374,
      "learning_rate": 9.988487449129993e-06,
      "loss": 0.008,
      "step": 21900
    },
    {
      "epoch": 0.3585044588071668,
      "grad_norm": 0.3587813377380371,
      "learning_rate": 9.988422796976013e-06,
      "loss": 0.0071,
      "step": 21910
    },
    {
      "epoch": 0.35866808475824263,
      "grad_norm": 0.3461415767669678,
      "learning_rate": 9.988357964003696e-06,
      "loss": 0.0069,
      "step": 21920
    },
    {
      "epoch": 0.3588317107093185,
      "grad_norm": 0.28841766715049744,
      "learning_rate": 9.988292950215393e-06,
      "loss": 0.0078,
      "step": 21930
    },
    {
      "epoch": 0.35899533666039435,
      "grad_norm": 0.3538677990436554,
      "learning_rate": 9.988227755613458e-06,
      "loss": 0.0047,
      "step": 21940
    },
    {
      "epoch": 0.3591589626114702,
      "grad_norm": 0.3300904929637909,
      "learning_rate": 9.988162380200257e-06,
      "loss": 0.0074,
      "step": 21950
    },
    {
      "epoch": 0.359322588562546,
      "grad_norm": 0.06635846942663193,
      "learning_rate": 9.988096823978158e-06,
      "loss": 0.0039,
      "step": 21960
    },
    {
      "epoch": 0.35948621451362184,
      "grad_norm": 0.29696011543273926,
      "learning_rate": 9.988031086949536e-06,
      "loss": 0.0054,
      "step": 21970
    },
    {
      "epoch": 0.3596498404646977,
      "grad_norm": 0.15317226946353912,
      "learning_rate": 9.987965169116778e-06,
      "loss": 0.007,
      "step": 21980
    },
    {
      "epoch": 0.35981346641577355,
      "grad_norm": 0.2748468518257141,
      "learning_rate": 9.987899070482269e-06,
      "loss": 0.0056,
      "step": 21990
    },
    {
      "epoch": 0.3599770923668494,
      "grad_norm": 0.2048082798719406,
      "learning_rate": 9.987832791048408e-06,
      "loss": 0.0057,
      "step": 22000
    },
    {
      "epoch": 0.3601407183179252,
      "grad_norm": 0.1489567756652832,
      "learning_rate": 9.987766330817598e-06,
      "loss": 0.0041,
      "step": 22010
    },
    {
      "epoch": 0.36030434426900104,
      "grad_norm": 0.2870359420776367,
      "learning_rate": 9.987699689792244e-06,
      "loss": 0.006,
      "step": 22020
    },
    {
      "epoch": 0.3604679702200769,
      "grad_norm": 0.2506335973739624,
      "learning_rate": 9.987632867974764e-06,
      "loss": 0.0065,
      "step": 22030
    },
    {
      "epoch": 0.36063159617115276,
      "grad_norm": 0.19328466057777405,
      "learning_rate": 9.987565865367582e-06,
      "loss": 0.0075,
      "step": 22040
    },
    {
      "epoch": 0.3607952221222286,
      "grad_norm": 0.1783657819032669,
      "learning_rate": 9.987498681973121e-06,
      "loss": 0.004,
      "step": 22050
    },
    {
      "epoch": 0.3609588480733044,
      "grad_norm": 0.16120240092277527,
      "learning_rate": 9.987431317793823e-06,
      "loss": 0.0051,
      "step": 22060
    },
    {
      "epoch": 0.36112247402438025,
      "grad_norm": 0.4215061664581299,
      "learning_rate": 9.987363772832125e-06,
      "loss": 0.0081,
      "step": 22070
    },
    {
      "epoch": 0.36128609997545613,
      "grad_norm": 0.5700271725654602,
      "learning_rate": 9.987296047090477e-06,
      "loss": 0.0066,
      "step": 22080
    },
    {
      "epoch": 0.36144972592653196,
      "grad_norm": 0.21919545531272888,
      "learning_rate": 9.987228140571335e-06,
      "loss": 0.0053,
      "step": 22090
    },
    {
      "epoch": 0.3616133518776078,
      "grad_norm": 0.23127950727939606,
      "learning_rate": 9.987160053277158e-06,
      "loss": 0.0054,
      "step": 22100
    },
    {
      "epoch": 0.3617769778286836,
      "grad_norm": 0.20025378465652466,
      "learning_rate": 9.987091785210416e-06,
      "loss": 0.0057,
      "step": 22110
    },
    {
      "epoch": 0.36194060377975945,
      "grad_norm": 0.5790795087814331,
      "learning_rate": 9.987023336373584e-06,
      "loss": 0.0047,
      "step": 22120
    },
    {
      "epoch": 0.36210422973083534,
      "grad_norm": 0.22350969910621643,
      "learning_rate": 9.986954706769142e-06,
      "loss": 0.0073,
      "step": 22130
    },
    {
      "epoch": 0.36226785568191117,
      "grad_norm": 0.2985982298851013,
      "learning_rate": 9.986885896399577e-06,
      "loss": 0.0067,
      "step": 22140
    },
    {
      "epoch": 0.362431481632987,
      "grad_norm": 0.18830154836177826,
      "learning_rate": 9.986816905267385e-06,
      "loss": 0.0063,
      "step": 22150
    },
    {
      "epoch": 0.3625951075840628,
      "grad_norm": 0.07355014979839325,
      "learning_rate": 9.986747733375064e-06,
      "loss": 0.0069,
      "step": 22160
    },
    {
      "epoch": 0.36275873353513866,
      "grad_norm": 0.6007991433143616,
      "learning_rate": 9.986678380725123e-06,
      "loss": 0.0044,
      "step": 22170
    },
    {
      "epoch": 0.36292235948621454,
      "grad_norm": 0.20156943798065186,
      "learning_rate": 9.986608847320077e-06,
      "loss": 0.0073,
      "step": 22180
    },
    {
      "epoch": 0.36308598543729037,
      "grad_norm": 0.21956665813922882,
      "learning_rate": 9.986539133162444e-06,
      "loss": 0.0072,
      "step": 22190
    },
    {
      "epoch": 0.3632496113883662,
      "grad_norm": 0.3403819501399994,
      "learning_rate": 9.986469238254754e-06,
      "loss": 0.0062,
      "step": 22200
    },
    {
      "epoch": 0.36341323733944203,
      "grad_norm": 0.28590965270996094,
      "learning_rate": 9.986399162599538e-06,
      "loss": 0.0076,
      "step": 22210
    },
    {
      "epoch": 0.36357686329051786,
      "grad_norm": 0.311116099357605,
      "learning_rate": 9.986328906199337e-06,
      "loss": 0.0076,
      "step": 22220
    },
    {
      "epoch": 0.3637404892415937,
      "grad_norm": 0.3970174789428711,
      "learning_rate": 9.986258469056697e-06,
      "loss": 0.0088,
      "step": 22230
    },
    {
      "epoch": 0.3639041151926696,
      "grad_norm": 0.2976403534412384,
      "learning_rate": 9.986187851174171e-06,
      "loss": 0.0056,
      "step": 22240
    },
    {
      "epoch": 0.3640677411437454,
      "grad_norm": 0.23498351871967316,
      "learning_rate": 9.98611705255432e-06,
      "loss": 0.0073,
      "step": 22250
    },
    {
      "epoch": 0.36423136709482123,
      "grad_norm": 0.23971201479434967,
      "learning_rate": 9.986046073199711e-06,
      "loss": 0.006,
      "step": 22260
    },
    {
      "epoch": 0.36439499304589706,
      "grad_norm": 0.2019869089126587,
      "learning_rate": 9.985974913112914e-06,
      "loss": 0.0057,
      "step": 22270
    },
    {
      "epoch": 0.3645586189969729,
      "grad_norm": 0.30041128396987915,
      "learning_rate": 9.98590357229651e-06,
      "loss": 0.0053,
      "step": 22280
    },
    {
      "epoch": 0.3647222449480488,
      "grad_norm": 0.1984991729259491,
      "learning_rate": 9.985832050753085e-06,
      "loss": 0.0052,
      "step": 22290
    },
    {
      "epoch": 0.3648858708991246,
      "grad_norm": 0.467449814081192,
      "learning_rate": 9.985760348485232e-06,
      "loss": 0.01,
      "step": 22300
    },
    {
      "epoch": 0.36504949685020044,
      "grad_norm": 0.15146076679229736,
      "learning_rate": 9.985688465495549e-06,
      "loss": 0.0057,
      "step": 22310
    },
    {
      "epoch": 0.36521312280127627,
      "grad_norm": 0.1645674705505371,
      "learning_rate": 9.985616401786641e-06,
      "loss": 0.0056,
      "step": 22320
    },
    {
      "epoch": 0.3653767487523521,
      "grad_norm": 0.38499942421913147,
      "learning_rate": 9.985544157361122e-06,
      "loss": 0.0049,
      "step": 22330
    },
    {
      "epoch": 0.365540374703428,
      "grad_norm": 0.21839094161987305,
      "learning_rate": 9.985471732221609e-06,
      "loss": 0.0035,
      "step": 22340
    },
    {
      "epoch": 0.3657040006545038,
      "grad_norm": 0.023548418655991554,
      "learning_rate": 9.985399126370729e-06,
      "loss": 0.0083,
      "step": 22350
    },
    {
      "epoch": 0.36586762660557964,
      "grad_norm": 0.19656912982463837,
      "learning_rate": 9.985326339811113e-06,
      "loss": 0.0046,
      "step": 22360
    },
    {
      "epoch": 0.3660312525566555,
      "grad_norm": 0.4063679277896881,
      "learning_rate": 9.985253372545396e-06,
      "loss": 0.0063,
      "step": 22370
    },
    {
      "epoch": 0.3661948785077313,
      "grad_norm": 0.19815057516098022,
      "learning_rate": 9.985180224576228e-06,
      "loss": 0.0055,
      "step": 22380
    },
    {
      "epoch": 0.3663585044588072,
      "grad_norm": 0.4024091362953186,
      "learning_rate": 9.98510689590626e-06,
      "loss": 0.0056,
      "step": 22390
    },
    {
      "epoch": 0.366522130409883,
      "grad_norm": 0.12570717930793762,
      "learning_rate": 9.985033386538146e-06,
      "loss": 0.0037,
      "step": 22400
    },
    {
      "epoch": 0.36668575636095885,
      "grad_norm": 0.36101505160331726,
      "learning_rate": 9.984959696474555e-06,
      "loss": 0.0056,
      "step": 22410
    },
    {
      "epoch": 0.3668493823120347,
      "grad_norm": 0.08656579256057739,
      "learning_rate": 9.984885825718155e-06,
      "loss": 0.0053,
      "step": 22420
    },
    {
      "epoch": 0.3670130082631105,
      "grad_norm": 0.5284775495529175,
      "learning_rate": 9.984811774271624e-06,
      "loss": 0.0044,
      "step": 22430
    },
    {
      "epoch": 0.3671766342141864,
      "grad_norm": 0.09360700845718384,
      "learning_rate": 9.984737542137648e-06,
      "loss": 0.0066,
      "step": 22440
    },
    {
      "epoch": 0.3673402601652622,
      "grad_norm": 0.2975481152534485,
      "learning_rate": 9.984663129318917e-06,
      "loss": 0.0045,
      "step": 22450
    },
    {
      "epoch": 0.36750388611633805,
      "grad_norm": 0.3162071406841278,
      "learning_rate": 9.984588535818126e-06,
      "loss": 0.0063,
      "step": 22460
    },
    {
      "epoch": 0.3676675120674139,
      "grad_norm": 0.48850497603416443,
      "learning_rate": 9.984513761637982e-06,
      "loss": 0.0053,
      "step": 22470
    },
    {
      "epoch": 0.3678311380184897,
      "grad_norm": 0.25399038195610046,
      "learning_rate": 9.984438806781194e-06,
      "loss": 0.0078,
      "step": 22480
    },
    {
      "epoch": 0.3679947639695656,
      "grad_norm": 0.28231891989707947,
      "learning_rate": 9.98436367125048e-06,
      "loss": 0.0078,
      "step": 22490
    },
    {
      "epoch": 0.3681583899206414,
      "grad_norm": 0.21071882545948029,
      "learning_rate": 9.98428835504856e-06,
      "loss": 0.0041,
      "step": 22500
    },
    {
      "epoch": 0.36832201587171726,
      "grad_norm": 0.1944991797208786,
      "learning_rate": 9.984212858178169e-06,
      "loss": 0.009,
      "step": 22510
    },
    {
      "epoch": 0.3684856418227931,
      "grad_norm": 0.06897173821926117,
      "learning_rate": 9.984137180642039e-06,
      "loss": 0.0058,
      "step": 22520
    },
    {
      "epoch": 0.3686492677738689,
      "grad_norm": 0.17861177027225494,
      "learning_rate": 9.984061322442917e-06,
      "loss": 0.0046,
      "step": 22530
    },
    {
      "epoch": 0.3688128937249448,
      "grad_norm": 0.45208612084388733,
      "learning_rate": 9.983985283583551e-06,
      "loss": 0.0053,
      "step": 22540
    },
    {
      "epoch": 0.36897651967602063,
      "grad_norm": 0.21381516754627228,
      "learning_rate": 9.983909064066695e-06,
      "loss": 0.0067,
      "step": 22550
    },
    {
      "epoch": 0.36914014562709646,
      "grad_norm": 0.15785840153694153,
      "learning_rate": 9.983832663895115e-06,
      "loss": 0.0061,
      "step": 22560
    },
    {
      "epoch": 0.3693037715781723,
      "grad_norm": 0.13654188811779022,
      "learning_rate": 9.983756083071581e-06,
      "loss": 0.0057,
      "step": 22570
    },
    {
      "epoch": 0.3694673975292481,
      "grad_norm": 0.25045299530029297,
      "learning_rate": 9.983679321598865e-06,
      "loss": 0.0031,
      "step": 22580
    },
    {
      "epoch": 0.369631023480324,
      "grad_norm": 0.23947203159332275,
      "learning_rate": 9.983602379479752e-06,
      "loss": 0.0059,
      "step": 22590
    },
    {
      "epoch": 0.36979464943139984,
      "grad_norm": 0.36851584911346436,
      "learning_rate": 9.98352525671703e-06,
      "loss": 0.0065,
      "step": 22600
    },
    {
      "epoch": 0.36995827538247567,
      "grad_norm": 0.2738041877746582,
      "learning_rate": 9.983447953313497e-06,
      "loss": 0.0059,
      "step": 22610
    },
    {
      "epoch": 0.3701219013335515,
      "grad_norm": 0.39628249406814575,
      "learning_rate": 9.983370469271951e-06,
      "loss": 0.0069,
      "step": 22620
    },
    {
      "epoch": 0.3702855272846273,
      "grad_norm": 0.7322098612785339,
      "learning_rate": 9.983292804595205e-06,
      "loss": 0.0092,
      "step": 22630
    },
    {
      "epoch": 0.3704491532357032,
      "grad_norm": 0.2309613823890686,
      "learning_rate": 9.983214959286071e-06,
      "loss": 0.0056,
      "step": 22640
    },
    {
      "epoch": 0.37061277918677904,
      "grad_norm": 0.33896979689598083,
      "learning_rate": 9.983136933347371e-06,
      "loss": 0.0077,
      "step": 22650
    },
    {
      "epoch": 0.37077640513785487,
      "grad_norm": 0.38524144887924194,
      "learning_rate": 9.983058726781934e-06,
      "loss": 0.009,
      "step": 22660
    },
    {
      "epoch": 0.3709400310889307,
      "grad_norm": 0.39209043979644775,
      "learning_rate": 9.982980339592594e-06,
      "loss": 0.0109,
      "step": 22670
    },
    {
      "epoch": 0.37110365704000653,
      "grad_norm": 0.34785038232803345,
      "learning_rate": 9.982901771782195e-06,
      "loss": 0.0087,
      "step": 22680
    },
    {
      "epoch": 0.37126728299108236,
      "grad_norm": 0.19190365076065063,
      "learning_rate": 9.982823023353581e-06,
      "loss": 0.0056,
      "step": 22690
    },
    {
      "epoch": 0.37143090894215824,
      "grad_norm": 0.30775558948516846,
      "learning_rate": 9.982744094309608e-06,
      "loss": 0.0073,
      "step": 22700
    },
    {
      "epoch": 0.3715945348932341,
      "grad_norm": 0.2999928891658783,
      "learning_rate": 9.982664984653138e-06,
      "loss": 0.0045,
      "step": 22710
    },
    {
      "epoch": 0.3717581608443099,
      "grad_norm": 0.2771570682525635,
      "learning_rate": 9.982585694387036e-06,
      "loss": 0.0037,
      "step": 22720
    },
    {
      "epoch": 0.37192178679538573,
      "grad_norm": 0.18505041301250458,
      "learning_rate": 9.982506223514181e-06,
      "loss": 0.0053,
      "step": 22730
    },
    {
      "epoch": 0.37208541274646156,
      "grad_norm": 0.23630085587501526,
      "learning_rate": 9.982426572037449e-06,
      "loss": 0.0037,
      "step": 22740
    },
    {
      "epoch": 0.37224903869753745,
      "grad_norm": 0.25697746872901917,
      "learning_rate": 9.98234673995973e-06,
      "loss": 0.0055,
      "step": 22750
    },
    {
      "epoch": 0.3724126646486133,
      "grad_norm": 0.3935088515281677,
      "learning_rate": 9.982266727283915e-06,
      "loss": 0.0059,
      "step": 22760
    },
    {
      "epoch": 0.3725762905996891,
      "grad_norm": 0.2000584602355957,
      "learning_rate": 9.982186534012905e-06,
      "loss": 0.0037,
      "step": 22770
    },
    {
      "epoch": 0.37273991655076494,
      "grad_norm": 0.19533513486385345,
      "learning_rate": 9.982106160149609e-06,
      "loss": 0.0041,
      "step": 22780
    },
    {
      "epoch": 0.37290354250184077,
      "grad_norm": 0.21803849935531616,
      "learning_rate": 9.982025605696939e-06,
      "loss": 0.0059,
      "step": 22790
    },
    {
      "epoch": 0.37306716845291665,
      "grad_norm": 0.1725212186574936,
      "learning_rate": 9.981944870657813e-06,
      "loss": 0.0069,
      "step": 22800
    },
    {
      "epoch": 0.3732307944039925,
      "grad_norm": 0.23436790704727173,
      "learning_rate": 9.981863955035163e-06,
      "loss": 0.0056,
      "step": 22810
    },
    {
      "epoch": 0.3733944203550683,
      "grad_norm": 0.1467805802822113,
      "learning_rate": 9.981782858831914e-06,
      "loss": 0.0053,
      "step": 22820
    },
    {
      "epoch": 0.37355804630614414,
      "grad_norm": 0.2499193698167801,
      "learning_rate": 9.981701582051011e-06,
      "loss": 0.0051,
      "step": 22830
    },
    {
      "epoch": 0.37372167225722,
      "grad_norm": 0.265960156917572,
      "learning_rate": 9.9816201246954e-06,
      "loss": 0.0046,
      "step": 22840
    },
    {
      "epoch": 0.37388529820829586,
      "grad_norm": 0.3409392237663269,
      "learning_rate": 9.981538486768032e-06,
      "loss": 0.0052,
      "step": 22850
    },
    {
      "epoch": 0.3740489241593717,
      "grad_norm": 0.37110787630081177,
      "learning_rate": 9.981456668271867e-06,
      "loss": 0.0077,
      "step": 22860
    },
    {
      "epoch": 0.3742125501104475,
      "grad_norm": 0.2040826827287674,
      "learning_rate": 9.981374669209869e-06,
      "loss": 0.0057,
      "step": 22870
    },
    {
      "epoch": 0.37437617606152335,
      "grad_norm": 0.24345263838768005,
      "learning_rate": 9.981292489585013e-06,
      "loss": 0.0063,
      "step": 22880
    },
    {
      "epoch": 0.3745398020125992,
      "grad_norm": 0.34455573558807373,
      "learning_rate": 9.981210129400276e-06,
      "loss": 0.004,
      "step": 22890
    },
    {
      "epoch": 0.37470342796367506,
      "grad_norm": 0.1967410147190094,
      "learning_rate": 9.981127588658644e-06,
      "loss": 0.0047,
      "step": 22900
    },
    {
      "epoch": 0.3748670539147509,
      "grad_norm": 0.11044573038816452,
      "learning_rate": 9.981044867363107e-06,
      "loss": 0.0052,
      "step": 22910
    },
    {
      "epoch": 0.3750306798658267,
      "grad_norm": 0.146162748336792,
      "learning_rate": 9.98096196551667e-06,
      "loss": 0.0048,
      "step": 22920
    },
    {
      "epoch": 0.37519430581690255,
      "grad_norm": 0.05192472040653229,
      "learning_rate": 9.980878883122329e-06,
      "loss": 0.0036,
      "step": 22930
    },
    {
      "epoch": 0.3753579317679784,
      "grad_norm": 0.3095148205757141,
      "learning_rate": 9.9807956201831e-06,
      "loss": 0.006,
      "step": 22940
    },
    {
      "epoch": 0.37552155771905427,
      "grad_norm": 0.22653578221797943,
      "learning_rate": 9.980712176702002e-06,
      "loss": 0.0046,
      "step": 22950
    },
    {
      "epoch": 0.3756851836701301,
      "grad_norm": 0.3559248745441437,
      "learning_rate": 9.980628552682056e-06,
      "loss": 0.0043,
      "step": 22960
    },
    {
      "epoch": 0.3758488096212059,
      "grad_norm": 0.1362975537776947,
      "learning_rate": 9.9805447481263e-06,
      "loss": 0.0054,
      "step": 22970
    },
    {
      "epoch": 0.37601243557228176,
      "grad_norm": 0.2152615487575531,
      "learning_rate": 9.980460763037763e-06,
      "loss": 0.0036,
      "step": 22980
    },
    {
      "epoch": 0.3761760615233576,
      "grad_norm": 0.5250092148780823,
      "learning_rate": 9.980376597419496e-06,
      "loss": 0.0065,
      "step": 22990
    },
    {
      "epoch": 0.37633968747443347,
      "grad_norm": 0.24772511422634125,
      "learning_rate": 9.980292251274546e-06,
      "loss": 0.0071,
      "step": 23000
    },
    {
      "epoch": 0.3765033134255093,
      "grad_norm": 0.1705523431301117,
      "learning_rate": 9.980207724605972e-06,
      "loss": 0.0054,
      "step": 23010
    },
    {
      "epoch": 0.37666693937658513,
      "grad_norm": 0.07682956010103226,
      "learning_rate": 9.980123017416838e-06,
      "loss": 0.0054,
      "step": 23020
    },
    {
      "epoch": 0.37683056532766096,
      "grad_norm": 0.18210425972938538,
      "learning_rate": 9.980038129710214e-06,
      "loss": 0.0053,
      "step": 23030
    },
    {
      "epoch": 0.3769941912787368,
      "grad_norm": 0.1908276379108429,
      "learning_rate": 9.979953061489179e-06,
      "loss": 0.0076,
      "step": 23040
    },
    {
      "epoch": 0.3771578172298127,
      "grad_norm": 0.44596797227859497,
      "learning_rate": 9.979867812756811e-06,
      "loss": 0.0069,
      "step": 23050
    },
    {
      "epoch": 0.3773214431808885,
      "grad_norm": 0.29015466570854187,
      "learning_rate": 9.979782383516206e-06,
      "loss": 0.0058,
      "step": 23060
    },
    {
      "epoch": 0.37748506913196433,
      "grad_norm": 0.17226456105709076,
      "learning_rate": 9.979696773770458e-06,
      "loss": 0.0043,
      "step": 23070
    },
    {
      "epoch": 0.37764869508304016,
      "grad_norm": 0.06641542166471481,
      "learning_rate": 9.97961098352267e-06,
      "loss": 0.005,
      "step": 23080
    },
    {
      "epoch": 0.377812321034116,
      "grad_norm": 0.3365395963191986,
      "learning_rate": 9.97952501277595e-06,
      "loss": 0.0065,
      "step": 23090
    },
    {
      "epoch": 0.3779759469851918,
      "grad_norm": 0.799506425857544,
      "learning_rate": 9.97943886153342e-06,
      "loss": 0.0046,
      "step": 23100
    },
    {
      "epoch": 0.3781395729362677,
      "grad_norm": 0.18472743034362793,
      "learning_rate": 9.979352529798195e-06,
      "loss": 0.0045,
      "step": 23110
    },
    {
      "epoch": 0.37830319888734354,
      "grad_norm": 0.13883015513420105,
      "learning_rate": 9.979266017573411e-06,
      "loss": 0.0045,
      "step": 23120
    },
    {
      "epoch": 0.37846682483841937,
      "grad_norm": 0.3223952054977417,
      "learning_rate": 9.9791793248622e-06,
      "loss": 0.0072,
      "step": 23130
    },
    {
      "epoch": 0.3786304507894952,
      "grad_norm": 0.499763160943985,
      "learning_rate": 9.979092451667707e-06,
      "loss": 0.0054,
      "step": 23140
    },
    {
      "epoch": 0.37879407674057103,
      "grad_norm": 0.4987930655479431,
      "learning_rate": 9.979005397993078e-06,
      "loss": 0.0069,
      "step": 23150
    },
    {
      "epoch": 0.3789577026916469,
      "grad_norm": 0.1661977916955948,
      "learning_rate": 9.97891816384147e-06,
      "loss": 0.0052,
      "step": 23160
    },
    {
      "epoch": 0.37912132864272274,
      "grad_norm": 0.30234283208847046,
      "learning_rate": 9.978830749216046e-06,
      "loss": 0.0061,
      "step": 23170
    },
    {
      "epoch": 0.3792849545937986,
      "grad_norm": 0.4478006660938263,
      "learning_rate": 9.978743154119975e-06,
      "loss": 0.0052,
      "step": 23180
    },
    {
      "epoch": 0.3794485805448744,
      "grad_norm": 0.15942557156085968,
      "learning_rate": 9.978655378556429e-06,
      "loss": 0.0045,
      "step": 23190
    },
    {
      "epoch": 0.37961220649595023,
      "grad_norm": 0.37059086561203003,
      "learning_rate": 9.97856742252859e-06,
      "loss": 0.0047,
      "step": 23200
    },
    {
      "epoch": 0.3797758324470261,
      "grad_norm": 0.3322787284851074,
      "learning_rate": 9.97847928603965e-06,
      "loss": 0.0069,
      "step": 23210
    },
    {
      "epoch": 0.37993945839810195,
      "grad_norm": 0.4209263026714325,
      "learning_rate": 9.978390969092802e-06,
      "loss": 0.0065,
      "step": 23220
    },
    {
      "epoch": 0.3801030843491778,
      "grad_norm": 0.1979064792394638,
      "learning_rate": 9.978302471691245e-06,
      "loss": 0.0041,
      "step": 23230
    },
    {
      "epoch": 0.3802667103002536,
      "grad_norm": 0.20982396602630615,
      "learning_rate": 9.978213793838188e-06,
      "loss": 0.0053,
      "step": 23240
    },
    {
      "epoch": 0.38043033625132944,
      "grad_norm": 0.13042889535427094,
      "learning_rate": 9.978124935536848e-06,
      "loss": 0.0041,
      "step": 23250
    },
    {
      "epoch": 0.3805939622024053,
      "grad_norm": 0.20270025730133057,
      "learning_rate": 9.978035896790442e-06,
      "loss": 0.0034,
      "step": 23260
    },
    {
      "epoch": 0.38075758815348115,
      "grad_norm": 0.6419389247894287,
      "learning_rate": 9.977946677602198e-06,
      "loss": 0.0062,
      "step": 23270
    },
    {
      "epoch": 0.380921214104557,
      "grad_norm": 0.20245814323425293,
      "learning_rate": 9.977857277975353e-06,
      "loss": 0.0045,
      "step": 23280
    },
    {
      "epoch": 0.3810848400556328,
      "grad_norm": 0.3386939764022827,
      "learning_rate": 9.977767697913145e-06,
      "loss": 0.0067,
      "step": 23290
    },
    {
      "epoch": 0.38124846600670864,
      "grad_norm": 0.19438815116882324,
      "learning_rate": 9.977677937418821e-06,
      "loss": 0.0042,
      "step": 23300
    },
    {
      "epoch": 0.3814120919577845,
      "grad_norm": 0.3810281753540039,
      "learning_rate": 9.977587996495636e-06,
      "loss": 0.0068,
      "step": 23310
    },
    {
      "epoch": 0.38157571790886036,
      "grad_norm": 0.5735651254653931,
      "learning_rate": 9.97749787514685e-06,
      "loss": 0.0045,
      "step": 23320
    },
    {
      "epoch": 0.3817393438599362,
      "grad_norm": 0.16405542194843292,
      "learning_rate": 9.977407573375728e-06,
      "loss": 0.0049,
      "step": 23330
    },
    {
      "epoch": 0.381902969811012,
      "grad_norm": 0.1767434924840927,
      "learning_rate": 9.977317091185545e-06,
      "loss": 0.006,
      "step": 23340
    },
    {
      "epoch": 0.38206659576208785,
      "grad_norm": 0.19959715008735657,
      "learning_rate": 9.97722642857958e-06,
      "loss": 0.0052,
      "step": 23350
    },
    {
      "epoch": 0.38223022171316373,
      "grad_norm": 0.4158332347869873,
      "learning_rate": 9.97713558556112e-06,
      "loss": 0.0057,
      "step": 23360
    },
    {
      "epoch": 0.38239384766423956,
      "grad_norm": 0.1551690548658371,
      "learning_rate": 9.977044562133458e-06,
      "loss": 0.0044,
      "step": 23370
    },
    {
      "epoch": 0.3825574736153154,
      "grad_norm": 0.4347027838230133,
      "learning_rate": 9.97695335829989e-06,
      "loss": 0.0066,
      "step": 23380
    },
    {
      "epoch": 0.3827210995663912,
      "grad_norm": 0.33756881952285767,
      "learning_rate": 9.976861974063728e-06,
      "loss": 0.0052,
      "step": 23390
    },
    {
      "epoch": 0.38288472551746705,
      "grad_norm": 0.07451612502336502,
      "learning_rate": 9.976770409428278e-06,
      "loss": 0.0048,
      "step": 23400
    },
    {
      "epoch": 0.38304835146854294,
      "grad_norm": 0.5526024699211121,
      "learning_rate": 9.976678664396862e-06,
      "loss": 0.0052,
      "step": 23410
    },
    {
      "epoch": 0.38321197741961877,
      "grad_norm": 0.22901882231235504,
      "learning_rate": 9.976586738972806e-06,
      "loss": 0.0059,
      "step": 23420
    },
    {
      "epoch": 0.3833756033706946,
      "grad_norm": 0.39812806248664856,
      "learning_rate": 9.976494633159444e-06,
      "loss": 0.0084,
      "step": 23430
    },
    {
      "epoch": 0.3835392293217704,
      "grad_norm": 0.14845570921897888,
      "learning_rate": 9.97640234696011e-06,
      "loss": 0.0043,
      "step": 23440
    },
    {
      "epoch": 0.38370285527284625,
      "grad_norm": 0.15554490685462952,
      "learning_rate": 9.976309880378152e-06,
      "loss": 0.0077,
      "step": 23450
    },
    {
      "epoch": 0.38386648122392214,
      "grad_norm": 0.471924751996994,
      "learning_rate": 9.976217233416921e-06,
      "loss": 0.0048,
      "step": 23460
    },
    {
      "epoch": 0.38403010717499797,
      "grad_norm": 0.4548797309398651,
      "learning_rate": 9.976124406079775e-06,
      "loss": 0.0057,
      "step": 23470
    },
    {
      "epoch": 0.3841937331260738,
      "grad_norm": 0.4401860535144806,
      "learning_rate": 9.976031398370078e-06,
      "loss": 0.0066,
      "step": 23480
    },
    {
      "epoch": 0.38435735907714963,
      "grad_norm": 0.2766363024711609,
      "learning_rate": 9.975938210291206e-06,
      "loss": 0.0044,
      "step": 23490
    },
    {
      "epoch": 0.38452098502822546,
      "grad_norm": 0.366016685962677,
      "learning_rate": 9.97584484184653e-06,
      "loss": 0.0055,
      "step": 23500
    },
    {
      "epoch": 0.38468461097930134,
      "grad_norm": 0.4660361409187317,
      "learning_rate": 9.975751293039437e-06,
      "loss": 0.0045,
      "step": 23510
    },
    {
      "epoch": 0.3848482369303772,
      "grad_norm": 0.24821795523166656,
      "learning_rate": 9.975657563873322e-06,
      "loss": 0.0103,
      "step": 23520
    },
    {
      "epoch": 0.385011862881453,
      "grad_norm": 0.2739928066730499,
      "learning_rate": 9.975563654351577e-06,
      "loss": 0.0049,
      "step": 23530
    },
    {
      "epoch": 0.38517548883252883,
      "grad_norm": 0.04939445108175278,
      "learning_rate": 9.975469564477606e-06,
      "loss": 0.0038,
      "step": 23540
    },
    {
      "epoch": 0.38533911478360466,
      "grad_norm": 0.3742635250091553,
      "learning_rate": 9.975375294254824e-06,
      "loss": 0.0076,
      "step": 23550
    },
    {
      "epoch": 0.3855027407346805,
      "grad_norm": 0.25204190611839294,
      "learning_rate": 9.975280843686645e-06,
      "loss": 0.0044,
      "step": 23560
    },
    {
      "epoch": 0.3856663666857564,
      "grad_norm": 0.28611651062965393,
      "learning_rate": 9.975186212776493e-06,
      "loss": 0.0063,
      "step": 23570
    },
    {
      "epoch": 0.3858299926368322,
      "grad_norm": 0.2678607702255249,
      "learning_rate": 9.975091401527799e-06,
      "loss": 0.0048,
      "step": 23580
    },
    {
      "epoch": 0.38599361858790804,
      "grad_norm": 0.2967936396598816,
      "learning_rate": 9.974996409943998e-06,
      "loss": 0.0045,
      "step": 23590
    },
    {
      "epoch": 0.38615724453898387,
      "grad_norm": 0.41232213377952576,
      "learning_rate": 9.974901238028536e-06,
      "loss": 0.0087,
      "step": 23600
    },
    {
      "epoch": 0.3863208704900597,
      "grad_norm": 0.26939794421195984,
      "learning_rate": 9.974805885784859e-06,
      "loss": 0.006,
      "step": 23610
    },
    {
      "epoch": 0.3864844964411356,
      "grad_norm": 0.32232874631881714,
      "learning_rate": 9.974710353216427e-06,
      "loss": 0.0043,
      "step": 23620
    },
    {
      "epoch": 0.3866481223922114,
      "grad_norm": 0.22538581490516663,
      "learning_rate": 9.974614640326699e-06,
      "loss": 0.0039,
      "step": 23630
    },
    {
      "epoch": 0.38681174834328724,
      "grad_norm": 0.19105230271816254,
      "learning_rate": 9.974518747119148e-06,
      "loss": 0.0061,
      "step": 23640
    },
    {
      "epoch": 0.3869753742943631,
      "grad_norm": 0.16645918786525726,
      "learning_rate": 9.974422673597248e-06,
      "loss": 0.0046,
      "step": 23650
    },
    {
      "epoch": 0.3871390002454389,
      "grad_norm": 0.18031635880470276,
      "learning_rate": 9.974326419764482e-06,
      "loss": 0.0059,
      "step": 23660
    },
    {
      "epoch": 0.3873026261965148,
      "grad_norm": 0.26605668663978577,
      "learning_rate": 9.97422998562434e-06,
      "loss": 0.0056,
      "step": 23670
    },
    {
      "epoch": 0.3874662521475906,
      "grad_norm": 0.13481472432613373,
      "learning_rate": 9.974133371180314e-06,
      "loss": 0.008,
      "step": 23680
    },
    {
      "epoch": 0.38762987809866645,
      "grad_norm": 0.22045265138149261,
      "learning_rate": 9.974036576435909e-06,
      "loss": 0.006,
      "step": 23690
    },
    {
      "epoch": 0.3877935040497423,
      "grad_norm": 0.23106570541858673,
      "learning_rate": 9.973939601394634e-06,
      "loss": 0.0041,
      "step": 23700
    },
    {
      "epoch": 0.3879571300008181,
      "grad_norm": 0.09335854649543762,
      "learning_rate": 9.97384244606e-06,
      "loss": 0.0046,
      "step": 23710
    },
    {
      "epoch": 0.388120755951894,
      "grad_norm": 0.17957039177417755,
      "learning_rate": 9.973745110435536e-06,
      "loss": 0.0064,
      "step": 23720
    },
    {
      "epoch": 0.3882843819029698,
      "grad_norm": 0.1819697767496109,
      "learning_rate": 9.973647594524762e-06,
      "loss": 0.0051,
      "step": 23730
    },
    {
      "epoch": 0.38844800785404565,
      "grad_norm": 0.46709126234054565,
      "learning_rate": 9.973549898331219e-06,
      "loss": 0.0038,
      "step": 23740
    },
    {
      "epoch": 0.3886116338051215,
      "grad_norm": 0.2450115829706192,
      "learning_rate": 9.973452021858445e-06,
      "loss": 0.0047,
      "step": 23750
    },
    {
      "epoch": 0.3887752597561973,
      "grad_norm": 0.3151560127735138,
      "learning_rate": 9.973353965109987e-06,
      "loss": 0.0047,
      "step": 23760
    },
    {
      "epoch": 0.3889388857072732,
      "grad_norm": 0.26489320397377014,
      "learning_rate": 9.973255728089403e-06,
      "loss": 0.0075,
      "step": 23770
    },
    {
      "epoch": 0.389102511658349,
      "grad_norm": 0.4936962127685547,
      "learning_rate": 9.973157310800251e-06,
      "loss": 0.0054,
      "step": 23780
    },
    {
      "epoch": 0.38926613760942486,
      "grad_norm": 0.1364099383354187,
      "learning_rate": 9.9730587132461e-06,
      "loss": 0.0061,
      "step": 23790
    },
    {
      "epoch": 0.3894297635605007,
      "grad_norm": 0.43895190954208374,
      "learning_rate": 9.972959935430522e-06,
      "loss": 0.0076,
      "step": 23800
    },
    {
      "epoch": 0.3895933895115765,
      "grad_norm": 0.2118120640516281,
      "learning_rate": 9.972860977357099e-06,
      "loss": 0.0065,
      "step": 23810
    },
    {
      "epoch": 0.3897570154626524,
      "grad_norm": 0.28332260251045227,
      "learning_rate": 9.972761839029416e-06,
      "loss": 0.0094,
      "step": 23820
    },
    {
      "epoch": 0.38992064141372823,
      "grad_norm": 0.18957801163196564,
      "learning_rate": 9.972662520451071e-06,
      "loss": 0.0069,
      "step": 23830
    },
    {
      "epoch": 0.39008426736480406,
      "grad_norm": 0.10194365680217743,
      "learning_rate": 9.972563021625658e-06,
      "loss": 0.0042,
      "step": 23840
    },
    {
      "epoch": 0.3902478933158799,
      "grad_norm": 0.07737588882446289,
      "learning_rate": 9.972463342556789e-06,
      "loss": 0.0076,
      "step": 23850
    },
    {
      "epoch": 0.3904115192669557,
      "grad_norm": 0.07712378352880478,
      "learning_rate": 9.972363483248075e-06,
      "loss": 0.0055,
      "step": 23860
    },
    {
      "epoch": 0.3905751452180316,
      "grad_norm": 0.26023125648498535,
      "learning_rate": 9.972263443703134e-06,
      "loss": 0.0057,
      "step": 23870
    },
    {
      "epoch": 0.39073877116910744,
      "grad_norm": 0.6190078258514404,
      "learning_rate": 9.972163223925595e-06,
      "loss": 0.0078,
      "step": 23880
    },
    {
      "epoch": 0.39090239712018326,
      "grad_norm": 0.24326372146606445,
      "learning_rate": 9.972062823919088e-06,
      "loss": 0.0047,
      "step": 23890
    },
    {
      "epoch": 0.3910660230712591,
      "grad_norm": 0.29368191957473755,
      "learning_rate": 9.971962243687254e-06,
      "loss": 0.0052,
      "step": 23900
    },
    {
      "epoch": 0.3912296490223349,
      "grad_norm": 0.43202829360961914,
      "learning_rate": 9.971861483233738e-06,
      "loss": 0.0041,
      "step": 23910
    },
    {
      "epoch": 0.3913932749734108,
      "grad_norm": 0.415345698595047,
      "learning_rate": 9.971760542562195e-06,
      "loss": 0.0034,
      "step": 23920
    },
    {
      "epoch": 0.39155690092448664,
      "grad_norm": 0.40836766362190247,
      "learning_rate": 9.97165942167628e-06,
      "loss": 0.0032,
      "step": 23930
    },
    {
      "epoch": 0.39172052687556247,
      "grad_norm": 0.2623162567615509,
      "learning_rate": 9.971558120579661e-06,
      "loss": 0.0064,
      "step": 23940
    },
    {
      "epoch": 0.3918841528266383,
      "grad_norm": 0.18547464907169342,
      "learning_rate": 9.971456639276007e-06,
      "loss": 0.0069,
      "step": 23950
    },
    {
      "epoch": 0.39204777877771413,
      "grad_norm": 0.3880669176578522,
      "learning_rate": 9.971354977769e-06,
      "loss": 0.0045,
      "step": 23960
    },
    {
      "epoch": 0.39221140472878996,
      "grad_norm": 0.4460575580596924,
      "learning_rate": 9.971253136062323e-06,
      "loss": 0.0063,
      "step": 23970
    },
    {
      "epoch": 0.39237503067986584,
      "grad_norm": 0.12719711661338806,
      "learning_rate": 9.971151114159668e-06,
      "loss": 0.0044,
      "step": 23980
    },
    {
      "epoch": 0.3925386566309417,
      "grad_norm": 0.0935448482632637,
      "learning_rate": 9.971048912064733e-06,
      "loss": 0.0061,
      "step": 23990
    },
    {
      "epoch": 0.3927022825820175,
      "grad_norm": 0.3725396692752838,
      "learning_rate": 9.970946529781223e-06,
      "loss": 0.0055,
      "step": 24000
    },
    {
      "epoch": 0.39286590853309333,
      "grad_norm": 0.24368101358413696,
      "learning_rate": 9.970843967312848e-06,
      "loss": 0.0052,
      "step": 24010
    },
    {
      "epoch": 0.39302953448416916,
      "grad_norm": 0.21469877660274506,
      "learning_rate": 9.970741224663326e-06,
      "loss": 0.0045,
      "step": 24020
    },
    {
      "epoch": 0.39319316043524505,
      "grad_norm": 0.19806301593780518,
      "learning_rate": 9.970638301836382e-06,
      "loss": 0.006,
      "step": 24030
    },
    {
      "epoch": 0.3933567863863209,
      "grad_norm": 0.18765555322170258,
      "learning_rate": 9.970535198835748e-06,
      "loss": 0.0044,
      "step": 24040
    },
    {
      "epoch": 0.3935204123373967,
      "grad_norm": 0.29159626364707947,
      "learning_rate": 9.970431915665156e-06,
      "loss": 0.0058,
      "step": 24050
    },
    {
      "epoch": 0.39368403828847254,
      "grad_norm": 0.2907009422779083,
      "learning_rate": 9.970328452328354e-06,
      "loss": 0.0031,
      "step": 24060
    },
    {
      "epoch": 0.39384766423954837,
      "grad_norm": 0.38016605377197266,
      "learning_rate": 9.970224808829092e-06,
      "loss": 0.004,
      "step": 24070
    },
    {
      "epoch": 0.39401129019062425,
      "grad_norm": 0.2068358212709427,
      "learning_rate": 9.970120985171127e-06,
      "loss": 0.008,
      "step": 24080
    },
    {
      "epoch": 0.3941749161417001,
      "grad_norm": 0.09498483687639236,
      "learning_rate": 9.970016981358222e-06,
      "loss": 0.0063,
      "step": 24090
    },
    {
      "epoch": 0.3943385420927759,
      "grad_norm": 0.22719818353652954,
      "learning_rate": 9.969912797394144e-06,
      "loss": 0.0028,
      "step": 24100
    },
    {
      "epoch": 0.39450216804385174,
      "grad_norm": 0.8270195722579956,
      "learning_rate": 9.969808433282675e-06,
      "loss": 0.0053,
      "step": 24110
    },
    {
      "epoch": 0.39466579399492757,
      "grad_norm": 0.2571798861026764,
      "learning_rate": 9.969703889027593e-06,
      "loss": 0.0106,
      "step": 24120
    },
    {
      "epoch": 0.39482941994600346,
      "grad_norm": 0.36881402134895325,
      "learning_rate": 9.969599164632689e-06,
      "loss": 0.0044,
      "step": 24130
    },
    {
      "epoch": 0.3949930458970793,
      "grad_norm": 0.4917198717594147,
      "learning_rate": 9.969494260101762e-06,
      "loss": 0.0042,
      "step": 24140
    },
    {
      "epoch": 0.3951566718481551,
      "grad_norm": 0.4277592599391937,
      "learning_rate": 9.969389175438609e-06,
      "loss": 0.0054,
      "step": 24150
    },
    {
      "epoch": 0.39532029779923095,
      "grad_norm": 0.2754640281200409,
      "learning_rate": 9.969283910647042e-06,
      "loss": 0.0098,
      "step": 24160
    },
    {
      "epoch": 0.3954839237503068,
      "grad_norm": 0.31599128246307373,
      "learning_rate": 9.969178465730879e-06,
      "loss": 0.0046,
      "step": 24170
    },
    {
      "epoch": 0.39564754970138266,
      "grad_norm": 0.2879290282726288,
      "learning_rate": 9.969072840693938e-06,
      "loss": 0.0045,
      "step": 24180
    },
    {
      "epoch": 0.3958111756524585,
      "grad_norm": 0.09363383054733276,
      "learning_rate": 9.968967035540051e-06,
      "loss": 0.0044,
      "step": 24190
    },
    {
      "epoch": 0.3959748016035343,
      "grad_norm": 0.3829444646835327,
      "learning_rate": 9.968861050273049e-06,
      "loss": 0.0045,
      "step": 24200
    },
    {
      "epoch": 0.39613842755461015,
      "grad_norm": 0.2220722734928131,
      "learning_rate": 9.968754884896778e-06,
      "loss": 0.0063,
      "step": 24210
    },
    {
      "epoch": 0.396302053505686,
      "grad_norm": 0.3211401402950287,
      "learning_rate": 9.968648539415085e-06,
      "loss": 0.0061,
      "step": 24220
    },
    {
      "epoch": 0.39646567945676187,
      "grad_norm": 0.49462971091270447,
      "learning_rate": 9.968542013831824e-06,
      "loss": 0.0058,
      "step": 24230
    },
    {
      "epoch": 0.3966293054078377,
      "grad_norm": 0.1990286409854889,
      "learning_rate": 9.968435308150856e-06,
      "loss": 0.0055,
      "step": 24240
    },
    {
      "epoch": 0.3967929313589135,
      "grad_norm": 0.22975203394889832,
      "learning_rate": 9.96832842237605e-06,
      "loss": 0.0049,
      "step": 24250
    },
    {
      "epoch": 0.39695655730998936,
      "grad_norm": 0.3899569809436798,
      "learning_rate": 9.968221356511278e-06,
      "loss": 0.0072,
      "step": 24260
    },
    {
      "epoch": 0.3971201832610652,
      "grad_norm": 0.21084776520729065,
      "learning_rate": 9.968114110560424e-06,
      "loss": 0.0059,
      "step": 24270
    },
    {
      "epoch": 0.39728380921214107,
      "grad_norm": 0.42953652143478394,
      "learning_rate": 9.968006684527374e-06,
      "loss": 0.0042,
      "step": 24280
    },
    {
      "epoch": 0.3974474351632169,
      "grad_norm": 0.25477996468544006,
      "learning_rate": 9.967899078416022e-06,
      "loss": 0.0033,
      "step": 24290
    },
    {
      "epoch": 0.39761106111429273,
      "grad_norm": 0.36159929633140564,
      "learning_rate": 9.967791292230268e-06,
      "loss": 0.0052,
      "step": 24300
    },
    {
      "epoch": 0.39777468706536856,
      "grad_norm": 0.3326166570186615,
      "learning_rate": 9.96768332597402e-06,
      "loss": 0.0093,
      "step": 24310
    },
    {
      "epoch": 0.3979383130164444,
      "grad_norm": 0.5496561527252197,
      "learning_rate": 9.967575179651191e-06,
      "loss": 0.0054,
      "step": 24320
    },
    {
      "epoch": 0.3981019389675203,
      "grad_norm": 0.1711733490228653,
      "learning_rate": 9.967466853265701e-06,
      "loss": 0.0032,
      "step": 24330
    },
    {
      "epoch": 0.3982655649185961,
      "grad_norm": 0.25165197253227234,
      "learning_rate": 9.967358346821476e-06,
      "loss": 0.0056,
      "step": 24340
    },
    {
      "epoch": 0.39842919086967193,
      "grad_norm": 0.19688613712787628,
      "learning_rate": 9.96724966032245e-06,
      "loss": 0.0037,
      "step": 24350
    },
    {
      "epoch": 0.39859281682074776,
      "grad_norm": 0.25742360949516296,
      "learning_rate": 9.967140793772563e-06,
      "loss": 0.0061,
      "step": 24360
    },
    {
      "epoch": 0.3987564427718236,
      "grad_norm": 0.40701621770858765,
      "learning_rate": 9.96703174717576e-06,
      "loss": 0.0053,
      "step": 24370
    },
    {
      "epoch": 0.3989200687228995,
      "grad_norm": 0.3093240261077881,
      "learning_rate": 9.966922520535993e-06,
      "loss": 0.0049,
      "step": 24380
    },
    {
      "epoch": 0.3990836946739753,
      "grad_norm": 0.2780626714229584,
      "learning_rate": 9.966813113857223e-06,
      "loss": 0.0045,
      "step": 24390
    },
    {
      "epoch": 0.39924732062505114,
      "grad_norm": 0.3644535541534424,
      "learning_rate": 9.966703527143415e-06,
      "loss": 0.0045,
      "step": 24400
    },
    {
      "epoch": 0.39941094657612697,
      "grad_norm": 0.11512038111686707,
      "learning_rate": 9.966593760398542e-06,
      "loss": 0.0032,
      "step": 24410
    },
    {
      "epoch": 0.3995745725272028,
      "grad_norm": 0.19217462837696075,
      "learning_rate": 9.966483813626581e-06,
      "loss": 0.0057,
      "step": 24420
    },
    {
      "epoch": 0.39973819847827863,
      "grad_norm": 0.33124637603759766,
      "learning_rate": 9.966373686831518e-06,
      "loss": 0.0059,
      "step": 24430
    },
    {
      "epoch": 0.3999018244293545,
      "grad_norm": 0.19283248484134674,
      "learning_rate": 9.966263380017348e-06,
      "loss": 0.0028,
      "step": 24440
    },
    {
      "epoch": 0.40006545038043034,
      "grad_norm": 0.320963054895401,
      "learning_rate": 9.966152893188064e-06,
      "loss": 0.0058,
      "step": 24450
    },
    {
      "epoch": 0.4002290763315062,
      "grad_norm": 0.3013334572315216,
      "learning_rate": 9.966042226347674e-06,
      "loss": 0.0052,
      "step": 24460
    },
    {
      "epoch": 0.400392702282582,
      "grad_norm": 0.2365194410085678,
      "learning_rate": 9.965931379500188e-06,
      "loss": 0.0038,
      "step": 24470
    },
    {
      "epoch": 0.40055632823365783,
      "grad_norm": 0.1528308242559433,
      "learning_rate": 9.965820352649626e-06,
      "loss": 0.0048,
      "step": 24480
    },
    {
      "epoch": 0.4007199541847337,
      "grad_norm": 0.0778537392616272,
      "learning_rate": 9.965709145800011e-06,
      "loss": 0.0066,
      "step": 24490
    },
    {
      "epoch": 0.40088358013580955,
      "grad_norm": 0.26332536339759827,
      "learning_rate": 9.965597758955375e-06,
      "loss": 0.0044,
      "step": 24500
    },
    {
      "epoch": 0.4010472060868854,
      "grad_norm": 0.24666878581047058,
      "learning_rate": 9.965486192119754e-06,
      "loss": 0.0048,
      "step": 24510
    },
    {
      "epoch": 0.4012108320379612,
      "grad_norm": 0.3467451333999634,
      "learning_rate": 9.965374445297192e-06,
      "loss": 0.0046,
      "step": 24520
    },
    {
      "epoch": 0.40137445798903704,
      "grad_norm": 0.25522151589393616,
      "learning_rate": 9.965262518491742e-06,
      "loss": 0.0045,
      "step": 24530
    },
    {
      "epoch": 0.4015380839401129,
      "grad_norm": 0.4946388006210327,
      "learning_rate": 9.965150411707458e-06,
      "loss": 0.0092,
      "step": 24540
    },
    {
      "epoch": 0.40170170989118875,
      "grad_norm": 0.14260706305503845,
      "learning_rate": 9.965038124948406e-06,
      "loss": 0.0057,
      "step": 24550
    },
    {
      "epoch": 0.4018653358422646,
      "grad_norm": 0.5055813193321228,
      "learning_rate": 9.964925658218655e-06,
      "loss": 0.0062,
      "step": 24560
    },
    {
      "epoch": 0.4020289617933404,
      "grad_norm": 0.14866884052753448,
      "learning_rate": 9.964813011522281e-06,
      "loss": 0.0053,
      "step": 24570
    },
    {
      "epoch": 0.40219258774441624,
      "grad_norm": 0.313092976808548,
      "learning_rate": 9.964700184863368e-06,
      "loss": 0.0053,
      "step": 24580
    },
    {
      "epoch": 0.4023562136954921,
      "grad_norm": 0.40639904141426086,
      "learning_rate": 9.964587178246006e-06,
      "loss": 0.0047,
      "step": 24590
    },
    {
      "epoch": 0.40251983964656796,
      "grad_norm": 0.35632413625717163,
      "learning_rate": 9.964473991674291e-06,
      "loss": 0.0061,
      "step": 24600
    },
    {
      "epoch": 0.4026834655976438,
      "grad_norm": 0.12481741607189178,
      "learning_rate": 9.964360625152326e-06,
      "loss": 0.0048,
      "step": 24610
    },
    {
      "epoch": 0.4028470915487196,
      "grad_norm": 0.10687965154647827,
      "learning_rate": 9.96424707868422e-06,
      "loss": 0.0022,
      "step": 24620
    },
    {
      "epoch": 0.40301071749979545,
      "grad_norm": 0.19886836409568787,
      "learning_rate": 9.964133352274086e-06,
      "loss": 0.0041,
      "step": 24630
    },
    {
      "epoch": 0.40317434345087133,
      "grad_norm": 0.3724556267261505,
      "learning_rate": 9.964019445926052e-06,
      "loss": 0.0077,
      "step": 24640
    },
    {
      "epoch": 0.40333796940194716,
      "grad_norm": 0.1827770471572876,
      "learning_rate": 9.963905359644243e-06,
      "loss": 0.0054,
      "step": 24650
    },
    {
      "epoch": 0.403501595353023,
      "grad_norm": 0.6065165400505066,
      "learning_rate": 9.963791093432794e-06,
      "loss": 0.0051,
      "step": 24660
    },
    {
      "epoch": 0.4036652213040988,
      "grad_norm": 0.18687261641025543,
      "learning_rate": 9.96367664729585e-06,
      "loss": 0.007,
      "step": 24670
    },
    {
      "epoch": 0.40382884725517465,
      "grad_norm": 0.3484671413898468,
      "learning_rate": 9.963562021237555e-06,
      "loss": 0.0044,
      "step": 24680
    },
    {
      "epoch": 0.40399247320625054,
      "grad_norm": 0.32855385541915894,
      "learning_rate": 9.963447215262067e-06,
      "loss": 0.0062,
      "step": 24690
    },
    {
      "epoch": 0.40415609915732637,
      "grad_norm": 0.22319531440734863,
      "learning_rate": 9.963332229373548e-06,
      "loss": 0.0061,
      "step": 24700
    },
    {
      "epoch": 0.4043197251084022,
      "grad_norm": 0.43020373582839966,
      "learning_rate": 9.963217063576165e-06,
      "loss": 0.0049,
      "step": 24710
    },
    {
      "epoch": 0.404483351059478,
      "grad_norm": 0.2594371438026428,
      "learning_rate": 9.963101717874091e-06,
      "loss": 0.0048,
      "step": 24720
    },
    {
      "epoch": 0.40464697701055385,
      "grad_norm": 0.2644640803337097,
      "learning_rate": 9.962986192271508e-06,
      "loss": 0.0044,
      "step": 24730
    },
    {
      "epoch": 0.40481060296162974,
      "grad_norm": 0.17103584110736847,
      "learning_rate": 9.962870486772605e-06,
      "loss": 0.0061,
      "step": 24740
    },
    {
      "epoch": 0.40497422891270557,
      "grad_norm": 0.2089265137910843,
      "learning_rate": 9.962754601381574e-06,
      "loss": 0.0056,
      "step": 24750
    },
    {
      "epoch": 0.4051378548637814,
      "grad_norm": 0.34814906120300293,
      "learning_rate": 9.962638536102617e-06,
      "loss": 0.0057,
      "step": 24760
    },
    {
      "epoch": 0.40530148081485723,
      "grad_norm": 0.27355480194091797,
      "learning_rate": 9.96252229093994e-06,
      "loss": 0.004,
      "step": 24770
    },
    {
      "epoch": 0.40546510676593306,
      "grad_norm": 0.19638685882091522,
      "learning_rate": 9.962405865897757e-06,
      "loss": 0.0052,
      "step": 24780
    },
    {
      "epoch": 0.40562873271700894,
      "grad_norm": 0.36804428696632385,
      "learning_rate": 9.96228926098029e-06,
      "loss": 0.0063,
      "step": 24790
    },
    {
      "epoch": 0.4057923586680848,
      "grad_norm": 0.19375556707382202,
      "learning_rate": 9.962172476191761e-06,
      "loss": 0.0049,
      "step": 24800
    },
    {
      "epoch": 0.4059559846191606,
      "grad_norm": 0.33867430686950684,
      "learning_rate": 9.962055511536406e-06,
      "loss": 0.0035,
      "step": 24810
    },
    {
      "epoch": 0.40611961057023643,
      "grad_norm": 0.2779190242290497,
      "learning_rate": 9.961938367018467e-06,
      "loss": 0.0046,
      "step": 24820
    },
    {
      "epoch": 0.40628323652131226,
      "grad_norm": 0.4429919719696045,
      "learning_rate": 9.961821042642186e-06,
      "loss": 0.0073,
      "step": 24830
    },
    {
      "epoch": 0.40644686247238815,
      "grad_norm": 0.24504682421684265,
      "learning_rate": 9.961703538411817e-06,
      "loss": 0.0063,
      "step": 24840
    },
    {
      "epoch": 0.406610488423464,
      "grad_norm": 0.5340507626533508,
      "learning_rate": 9.961585854331622e-06,
      "loss": 0.004,
      "step": 24850
    },
    {
      "epoch": 0.4067741143745398,
      "grad_norm": 0.10961088538169861,
      "learning_rate": 9.961467990405865e-06,
      "loss": 0.0043,
      "step": 24860
    },
    {
      "epoch": 0.40693774032561564,
      "grad_norm": 0.18699420988559723,
      "learning_rate": 9.961349946638816e-06,
      "loss": 0.0054,
      "step": 24870
    },
    {
      "epoch": 0.40710136627669147,
      "grad_norm": 0.25895199179649353,
      "learning_rate": 9.961231723034756e-06,
      "loss": 0.0047,
      "step": 24880
    },
    {
      "epoch": 0.4072649922277673,
      "grad_norm": 0.19572466611862183,
      "learning_rate": 9.96111331959797e-06,
      "loss": 0.0068,
      "step": 24890
    },
    {
      "epoch": 0.4074286181788432,
      "grad_norm": 0.20218001306056976,
      "learning_rate": 9.960994736332751e-06,
      "loss": 0.0035,
      "step": 24900
    },
    {
      "epoch": 0.407592244129919,
      "grad_norm": 0.17006951570510864,
      "learning_rate": 9.960875973243396e-06,
      "loss": 0.0035,
      "step": 24910
    },
    {
      "epoch": 0.40775587008099484,
      "grad_norm": 0.20042966306209564,
      "learning_rate": 9.96075703033421e-06,
      "loss": 0.0036,
      "step": 24920
    },
    {
      "epoch": 0.40791949603207067,
      "grad_norm": 0.44041380286216736,
      "learning_rate": 9.960637907609505e-06,
      "loss": 0.0059,
      "step": 24930
    },
    {
      "epoch": 0.4080831219831465,
      "grad_norm": 0.5424758791923523,
      "learning_rate": 9.960518605073596e-06,
      "loss": 0.0053,
      "step": 24940
    },
    {
      "epoch": 0.4082467479342224,
      "grad_norm": 0.24550536274909973,
      "learning_rate": 9.960399122730813e-06,
      "loss": 0.0066,
      "step": 24950
    },
    {
      "epoch": 0.4084103738852982,
      "grad_norm": 0.32169726490974426,
      "learning_rate": 9.960279460585482e-06,
      "loss": 0.006,
      "step": 24960
    },
    {
      "epoch": 0.40857399983637405,
      "grad_norm": 0.12515616416931152,
      "learning_rate": 9.960159618641942e-06,
      "loss": 0.0074,
      "step": 24970
    },
    {
      "epoch": 0.4087376257874499,
      "grad_norm": 0.310051828622818,
      "learning_rate": 9.960039596904538e-06,
      "loss": 0.0068,
      "step": 24980
    },
    {
      "epoch": 0.4089012517385257,
      "grad_norm": 0.5180619359016418,
      "learning_rate": 9.95991939537762e-06,
      "loss": 0.0045,
      "step": 24990
    },
    {
      "epoch": 0.4090648776896016,
      "grad_norm": 0.19713109731674194,
      "learning_rate": 9.959799014065546e-06,
      "loss": 0.0042,
      "step": 25000
    },
    {
      "epoch": 0.4090648776896016,
      "eval_loss": 0.0036062640137970448,
      "eval_runtime": 3.0986,
      "eval_samples_per_second": 64.545,
      "eval_steps_per_second": 16.136,
      "step": 25000
    },
    {
      "epoch": 0.4092285036406774,
      "grad_norm": 0.29881104826927185,
      "learning_rate": 9.959678452972675e-06,
      "loss": 0.0071,
      "step": 25010
    },
    {
      "epoch": 0.40939212959175325,
      "grad_norm": 0.2008904218673706,
      "learning_rate": 9.959557712103383e-06,
      "loss": 0.0051,
      "step": 25020
    },
    {
      "epoch": 0.4095557555428291,
      "grad_norm": 0.3467547297477722,
      "learning_rate": 9.959436791462043e-06,
      "loss": 0.0054,
      "step": 25030
    },
    {
      "epoch": 0.4097193814939049,
      "grad_norm": 0.08272959291934967,
      "learning_rate": 9.959315691053039e-06,
      "loss": 0.0078,
      "step": 25040
    },
    {
      "epoch": 0.4098830074449808,
      "grad_norm": 0.3445785343647003,
      "learning_rate": 9.959194410880761e-06,
      "loss": 0.0058,
      "step": 25050
    },
    {
      "epoch": 0.4100466333960566,
      "grad_norm": 0.0949714183807373,
      "learning_rate": 9.959072950949603e-06,
      "loss": 0.0077,
      "step": 25060
    },
    {
      "epoch": 0.41021025934713246,
      "grad_norm": 0.012899404391646385,
      "learning_rate": 9.958951311263971e-06,
      "loss": 0.0035,
      "step": 25070
    },
    {
      "epoch": 0.4103738852982083,
      "grad_norm": 0.20581640303134918,
      "learning_rate": 9.958829491828273e-06,
      "loss": 0.0061,
      "step": 25080
    },
    {
      "epoch": 0.4105375112492841,
      "grad_norm": 0.19220706820487976,
      "learning_rate": 9.958707492646921e-06,
      "loss": 0.005,
      "step": 25090
    },
    {
      "epoch": 0.41070113720036,
      "grad_norm": 0.32810795307159424,
      "learning_rate": 9.958585313724345e-06,
      "loss": 0.0059,
      "step": 25100
    },
    {
      "epoch": 0.41086476315143583,
      "grad_norm": 0.16524268686771393,
      "learning_rate": 9.958462955064967e-06,
      "loss": 0.0052,
      "step": 25110
    },
    {
      "epoch": 0.41102838910251166,
      "grad_norm": 0.21436114609241486,
      "learning_rate": 9.958340416673222e-06,
      "loss": 0.0054,
      "step": 25120
    },
    {
      "epoch": 0.4111920150535875,
      "grad_norm": 0.29300281405448914,
      "learning_rate": 9.958217698553556e-06,
      "loss": 0.004,
      "step": 25130
    },
    {
      "epoch": 0.4113556410046633,
      "grad_norm": 0.1949840486049652,
      "learning_rate": 9.958094800710417e-06,
      "loss": 0.0082,
      "step": 25140
    },
    {
      "epoch": 0.4115192669557392,
      "grad_norm": 0.42509058117866516,
      "learning_rate": 9.957971723148254e-06,
      "loss": 0.0046,
      "step": 25150
    },
    {
      "epoch": 0.41168289290681503,
      "grad_norm": 0.35835668444633484,
      "learning_rate": 9.957848465871536e-06,
      "loss": 0.0057,
      "step": 25160
    },
    {
      "epoch": 0.41184651885789086,
      "grad_norm": 0.29913750290870667,
      "learning_rate": 9.957725028884724e-06,
      "loss": 0.008,
      "step": 25170
    },
    {
      "epoch": 0.4120101448089667,
      "grad_norm": 0.20176512002944946,
      "learning_rate": 9.957601412192297e-06,
      "loss": 0.0059,
      "step": 25180
    },
    {
      "epoch": 0.4121737707600425,
      "grad_norm": 0.15453588962554932,
      "learning_rate": 9.957477615798732e-06,
      "loss": 0.0043,
      "step": 25190
    },
    {
      "epoch": 0.4123373967111184,
      "grad_norm": 0.13700389862060547,
      "learning_rate": 9.957353639708522e-06,
      "loss": 0.008,
      "step": 25200
    },
    {
      "epoch": 0.41250102266219424,
      "grad_norm": 0.17809541523456573,
      "learning_rate": 9.957229483926155e-06,
      "loss": 0.0071,
      "step": 25210
    },
    {
      "epoch": 0.41266464861327007,
      "grad_norm": 0.41866710782051086,
      "learning_rate": 9.957105148456133e-06,
      "loss": 0.004,
      "step": 25220
    },
    {
      "epoch": 0.4128282745643459,
      "grad_norm": 0.6000007390975952,
      "learning_rate": 9.956980633302963e-06,
      "loss": 0.0075,
      "step": 25230
    },
    {
      "epoch": 0.41299190051542173,
      "grad_norm": 0.22050197422504425,
      "learning_rate": 9.956855938471162e-06,
      "loss": 0.0034,
      "step": 25240
    },
    {
      "epoch": 0.4131555264664976,
      "grad_norm": 0.14122271537780762,
      "learning_rate": 9.956731063965244e-06,
      "loss": 0.0052,
      "step": 25250
    },
    {
      "epoch": 0.41331915241757344,
      "grad_norm": 0.18973210453987122,
      "learning_rate": 9.95660600978974e-06,
      "loss": 0.0052,
      "step": 25260
    },
    {
      "epoch": 0.4134827783686493,
      "grad_norm": 0.34211334586143494,
      "learning_rate": 9.95648077594918e-06,
      "loss": 0.0038,
      "step": 25270
    },
    {
      "epoch": 0.4136464043197251,
      "grad_norm": 0.3392001986503601,
      "learning_rate": 9.956355362448104e-06,
      "loss": 0.0054,
      "step": 25280
    },
    {
      "epoch": 0.41381003027080093,
      "grad_norm": 0.9375589489936829,
      "learning_rate": 9.95622976929106e-06,
      "loss": 0.0052,
      "step": 25290
    },
    {
      "epoch": 0.41397365622187676,
      "grad_norm": 0.3106217384338379,
      "learning_rate": 9.956103996482596e-06,
      "loss": 0.0047,
      "step": 25300
    },
    {
      "epoch": 0.41413728217295265,
      "grad_norm": 0.22567102313041687,
      "learning_rate": 9.955978044027276e-06,
      "loss": 0.0051,
      "step": 25310
    },
    {
      "epoch": 0.4143009081240285,
      "grad_norm": 0.27923932671546936,
      "learning_rate": 9.955851911929662e-06,
      "loss": 0.0073,
      "step": 25320
    },
    {
      "epoch": 0.4144645340751043,
      "grad_norm": 0.18773263692855835,
      "learning_rate": 9.955725600194328e-06,
      "loss": 0.0075,
      "step": 25330
    },
    {
      "epoch": 0.41462816002618014,
      "grad_norm": 0.6497575640678406,
      "learning_rate": 9.955599108825853e-06,
      "loss": 0.007,
      "step": 25340
    },
    {
      "epoch": 0.41479178597725597,
      "grad_norm": 0.062236130237579346,
      "learning_rate": 9.95547243782882e-06,
      "loss": 0.0044,
      "step": 25350
    },
    {
      "epoch": 0.41495541192833185,
      "grad_norm": 0.12795664370059967,
      "learning_rate": 9.95534558720782e-06,
      "loss": 0.0052,
      "step": 25360
    },
    {
      "epoch": 0.4151190378794077,
      "grad_norm": 0.22792263329029083,
      "learning_rate": 9.955218556967453e-06,
      "loss": 0.0066,
      "step": 25370
    },
    {
      "epoch": 0.4152826638304835,
      "grad_norm": 0.2341441959142685,
      "learning_rate": 9.955091347112324e-06,
      "loss": 0.005,
      "step": 25380
    },
    {
      "epoch": 0.41544628978155934,
      "grad_norm": 0.302999883890152,
      "learning_rate": 9.954963957647042e-06,
      "loss": 0.0053,
      "step": 25390
    },
    {
      "epoch": 0.41560991573263517,
      "grad_norm": 0.09005288779735565,
      "learning_rate": 9.954836388576224e-06,
      "loss": 0.0046,
      "step": 25400
    },
    {
      "epoch": 0.41577354168371106,
      "grad_norm": 0.07903607189655304,
      "learning_rate": 9.954708639904498e-06,
      "loss": 0.006,
      "step": 25410
    },
    {
      "epoch": 0.4159371676347869,
      "grad_norm": 0.3181702494621277,
      "learning_rate": 9.95458071163649e-06,
      "loss": 0.0052,
      "step": 25420
    },
    {
      "epoch": 0.4161007935858627,
      "grad_norm": 0.3068331778049469,
      "learning_rate": 9.954452603776841e-06,
      "loss": 0.0045,
      "step": 25430
    },
    {
      "epoch": 0.41626441953693855,
      "grad_norm": 0.40216565132141113,
      "learning_rate": 9.954324316330193e-06,
      "loss": 0.0066,
      "step": 25440
    },
    {
      "epoch": 0.4164280454880144,
      "grad_norm": 0.34599897265434265,
      "learning_rate": 9.954195849301195e-06,
      "loss": 0.0068,
      "step": 25450
    },
    {
      "epoch": 0.41659167143909026,
      "grad_norm": 0.06795250624418259,
      "learning_rate": 9.954067202694505e-06,
      "loss": 0.006,
      "step": 25460
    },
    {
      "epoch": 0.4167552973901661,
      "grad_norm": 0.18456536531448364,
      "learning_rate": 9.953938376514785e-06,
      "loss": 0.0054,
      "step": 25470
    },
    {
      "epoch": 0.4169189233412419,
      "grad_norm": 0.4055711030960083,
      "learning_rate": 9.953809370766705e-06,
      "loss": 0.0037,
      "step": 25480
    },
    {
      "epoch": 0.41708254929231775,
      "grad_norm": 0.43466177582740784,
      "learning_rate": 9.953680185454943e-06,
      "loss": 0.006,
      "step": 25490
    },
    {
      "epoch": 0.4172461752433936,
      "grad_norm": 0.5423679947853088,
      "learning_rate": 9.95355082058418e-06,
      "loss": 0.0058,
      "step": 25500
    },
    {
      "epoch": 0.41740980119446947,
      "grad_norm": 0.16508300602436066,
      "learning_rate": 9.953421276159103e-06,
      "loss": 0.0048,
      "step": 25510
    },
    {
      "epoch": 0.4175734271455453,
      "grad_norm": 0.16442042589187622,
      "learning_rate": 9.953291552184413e-06,
      "loss": 0.0046,
      "step": 25520
    },
    {
      "epoch": 0.4177370530966211,
      "grad_norm": 0.5984448790550232,
      "learning_rate": 9.953161648664807e-06,
      "loss": 0.005,
      "step": 25530
    },
    {
      "epoch": 0.41790067904769695,
      "grad_norm": 0.799819827079773,
      "learning_rate": 9.953031565604996e-06,
      "loss": 0.0051,
      "step": 25540
    },
    {
      "epoch": 0.4180643049987728,
      "grad_norm": 0.16470642387866974,
      "learning_rate": 9.952901303009696e-06,
      "loss": 0.0053,
      "step": 25550
    },
    {
      "epoch": 0.41822793094984867,
      "grad_norm": 0.22018200159072876,
      "learning_rate": 9.952770860883627e-06,
      "loss": 0.0054,
      "step": 25560
    },
    {
      "epoch": 0.4183915569009245,
      "grad_norm": 0.4381274878978729,
      "learning_rate": 9.95264023923152e-06,
      "loss": 0.0047,
      "step": 25570
    },
    {
      "epoch": 0.41855518285200033,
      "grad_norm": 0.16278640925884247,
      "learning_rate": 9.952509438058104e-06,
      "loss": 0.0046,
      "step": 25580
    },
    {
      "epoch": 0.41871880880307616,
      "grad_norm": 0.05348186939954758,
      "learning_rate": 9.952378457368126e-06,
      "loss": 0.0059,
      "step": 25590
    },
    {
      "epoch": 0.418882434754152,
      "grad_norm": 0.3621346950531006,
      "learning_rate": 9.952247297166334e-06,
      "loss": 0.0101,
      "step": 25600
    },
    {
      "epoch": 0.4190460607052279,
      "grad_norm": 0.3846125900745392,
      "learning_rate": 9.952115957457475e-06,
      "loss": 0.0038,
      "step": 25610
    },
    {
      "epoch": 0.4192096866563037,
      "grad_norm": 0.34133437275886536,
      "learning_rate": 9.951984438246317e-06,
      "loss": 0.0076,
      "step": 25620
    },
    {
      "epoch": 0.41937331260737953,
      "grad_norm": 0.29336848855018616,
      "learning_rate": 9.951852739537626e-06,
      "loss": 0.0065,
      "step": 25630
    },
    {
      "epoch": 0.41953693855845536,
      "grad_norm": 0.11607229709625244,
      "learning_rate": 9.951720861336174e-06,
      "loss": 0.0054,
      "step": 25640
    },
    {
      "epoch": 0.4197005645095312,
      "grad_norm": 0.34873926639556885,
      "learning_rate": 9.951588803646741e-06,
      "loss": 0.0035,
      "step": 25650
    },
    {
      "epoch": 0.4198641904606071,
      "grad_norm": 0.3979949951171875,
      "learning_rate": 9.951456566474116e-06,
      "loss": 0.0047,
      "step": 25660
    },
    {
      "epoch": 0.4200278164116829,
      "grad_norm": 0.13472244143486023,
      "learning_rate": 9.95132414982309e-06,
      "loss": 0.0062,
      "step": 25670
    },
    {
      "epoch": 0.42019144236275874,
      "grad_norm": 0.18795715272426605,
      "learning_rate": 9.951191553698464e-06,
      "loss": 0.0042,
      "step": 25680
    },
    {
      "epoch": 0.42035506831383457,
      "grad_norm": 0.23879875242710114,
      "learning_rate": 9.951058778105045e-06,
      "loss": 0.0048,
      "step": 25690
    },
    {
      "epoch": 0.4205186942649104,
      "grad_norm": 0.13192777335643768,
      "learning_rate": 9.950925823047644e-06,
      "loss": 0.0034,
      "step": 25700
    },
    {
      "epoch": 0.4206823202159863,
      "grad_norm": 0.5185250639915466,
      "learning_rate": 9.95079268853108e-06,
      "loss": 0.0055,
      "step": 25710
    },
    {
      "epoch": 0.4208459461670621,
      "grad_norm": 0.15754516422748566,
      "learning_rate": 9.95065937456018e-06,
      "loss": 0.0037,
      "step": 25720
    },
    {
      "epoch": 0.42100957211813794,
      "grad_norm": 0.3642040193080902,
      "learning_rate": 9.950525881139778e-06,
      "loss": 0.0051,
      "step": 25730
    },
    {
      "epoch": 0.42117319806921377,
      "grad_norm": 0.48439934849739075,
      "learning_rate": 9.95039220827471e-06,
      "loss": 0.003,
      "step": 25740
    },
    {
      "epoch": 0.4213368240202896,
      "grad_norm": 0.20476143062114716,
      "learning_rate": 9.950258355969825e-06,
      "loss": 0.0044,
      "step": 25750
    },
    {
      "epoch": 0.42150044997136543,
      "grad_norm": 0.4189421832561493,
      "learning_rate": 9.950124324229969e-06,
      "loss": 0.0062,
      "step": 25760
    },
    {
      "epoch": 0.4216640759224413,
      "grad_norm": 0.5145852565765381,
      "learning_rate": 9.949990113060003e-06,
      "loss": 0.0049,
      "step": 25770
    },
    {
      "epoch": 0.42182770187351715,
      "grad_norm": 0.1941939741373062,
      "learning_rate": 9.949855722464797e-06,
      "loss": 0.0052,
      "step": 25780
    },
    {
      "epoch": 0.421991327824593,
      "grad_norm": 0.2683684527873993,
      "learning_rate": 9.949721152449213e-06,
      "loss": 0.0053,
      "step": 25790
    },
    {
      "epoch": 0.4221549537756688,
      "grad_norm": 0.17171452939510345,
      "learning_rate": 9.949586403018136e-06,
      "loss": 0.0062,
      "step": 25800
    },
    {
      "epoch": 0.42231857972674464,
      "grad_norm": 0.3140377402305603,
      "learning_rate": 9.949451474176446e-06,
      "loss": 0.0046,
      "step": 25810
    },
    {
      "epoch": 0.4224822056778205,
      "grad_norm": 0.0775977075099945,
      "learning_rate": 9.949316365929038e-06,
      "loss": 0.0036,
      "step": 25820
    },
    {
      "epoch": 0.42264583162889635,
      "grad_norm": 0.5512652397155762,
      "learning_rate": 9.949181078280806e-06,
      "loss": 0.0045,
      "step": 25830
    },
    {
      "epoch": 0.4228094575799722,
      "grad_norm": 0.1899128258228302,
      "learning_rate": 9.949045611236655e-06,
      "loss": 0.0053,
      "step": 25840
    },
    {
      "epoch": 0.422973083531048,
      "grad_norm": 0.1803659051656723,
      "learning_rate": 9.948909964801495e-06,
      "loss": 0.0068,
      "step": 25850
    },
    {
      "epoch": 0.42313670948212384,
      "grad_norm": 0.1032019630074501,
      "learning_rate": 9.948774138980244e-06,
      "loss": 0.0043,
      "step": 25860
    },
    {
      "epoch": 0.4233003354331997,
      "grad_norm": 0.48989441990852356,
      "learning_rate": 9.948638133777825e-06,
      "loss": 0.0057,
      "step": 25870
    },
    {
      "epoch": 0.42346396138427556,
      "grad_norm": 0.22366076707839966,
      "learning_rate": 9.948501949199165e-06,
      "loss": 0.0049,
      "step": 25880
    },
    {
      "epoch": 0.4236275873353514,
      "grad_norm": 0.10869742929935455,
      "learning_rate": 9.948365585249204e-06,
      "loss": 0.0047,
      "step": 25890
    },
    {
      "epoch": 0.4237912132864272,
      "grad_norm": 0.23598021268844604,
      "learning_rate": 9.948229041932884e-06,
      "loss": 0.005,
      "step": 25900
    },
    {
      "epoch": 0.42395483923750305,
      "grad_norm": 0.27074310183525085,
      "learning_rate": 9.948092319255155e-06,
      "loss": 0.0057,
      "step": 25910
    },
    {
      "epoch": 0.42411846518857893,
      "grad_norm": 0.3151009976863861,
      "learning_rate": 9.94795541722097e-06,
      "loss": 0.0052,
      "step": 25920
    },
    {
      "epoch": 0.42428209113965476,
      "grad_norm": 0.20643682777881622,
      "learning_rate": 9.947818335835293e-06,
      "loss": 0.0059,
      "step": 25930
    },
    {
      "epoch": 0.4244457170907306,
      "grad_norm": 0.11311911046504974,
      "learning_rate": 9.947681075103095e-06,
      "loss": 0.0063,
      "step": 25940
    },
    {
      "epoch": 0.4246093430418064,
      "grad_norm": 0.058958884328603745,
      "learning_rate": 9.947543635029347e-06,
      "loss": 0.0033,
      "step": 25950
    },
    {
      "epoch": 0.42477296899288225,
      "grad_norm": 0.20372574031352997,
      "learning_rate": 9.947406015619036e-06,
      "loss": 0.005,
      "step": 25960
    },
    {
      "epoch": 0.42493659494395813,
      "grad_norm": 0.1304374486207962,
      "learning_rate": 9.947268216877146e-06,
      "loss": 0.0042,
      "step": 25970
    },
    {
      "epoch": 0.42510022089503396,
      "grad_norm": 0.47513291239738464,
      "learning_rate": 9.947130238808673e-06,
      "loss": 0.0041,
      "step": 25980
    },
    {
      "epoch": 0.4252638468461098,
      "grad_norm": 0.1874646544456482,
      "learning_rate": 9.94699208141862e-06,
      "loss": 0.0032,
      "step": 25990
    },
    {
      "epoch": 0.4254274727971856,
      "grad_norm": 0.4611448347568512,
      "learning_rate": 9.946853744711995e-06,
      "loss": 0.0049,
      "step": 26000
    },
    {
      "epoch": 0.42559109874826145,
      "grad_norm": 0.20589487254619598,
      "learning_rate": 9.94671522869381e-06,
      "loss": 0.0074,
      "step": 26010
    },
    {
      "epoch": 0.42575472469933734,
      "grad_norm": 0.47777605056762695,
      "learning_rate": 9.946576533369086e-06,
      "loss": 0.0063,
      "step": 26020
    },
    {
      "epoch": 0.42591835065041317,
      "grad_norm": 0.3368110954761505,
      "learning_rate": 9.946437658742854e-06,
      "loss": 0.0052,
      "step": 26030
    },
    {
      "epoch": 0.426081976601489,
      "grad_norm": 0.14642922580242157,
      "learning_rate": 9.946298604820143e-06,
      "loss": 0.0037,
      "step": 26040
    },
    {
      "epoch": 0.42624560255256483,
      "grad_norm": 0.45016244053840637,
      "learning_rate": 9.946159371605996e-06,
      "loss": 0.006,
      "step": 26050
    },
    {
      "epoch": 0.42640922850364066,
      "grad_norm": 0.0836092010140419,
      "learning_rate": 9.946019959105462e-06,
      "loss": 0.0031,
      "step": 26060
    },
    {
      "epoch": 0.42657285445471654,
      "grad_norm": 0.1861150711774826,
      "learning_rate": 9.945880367323589e-06,
      "loss": 0.0039,
      "step": 26070
    },
    {
      "epoch": 0.4267364804057924,
      "grad_norm": 0.3985017240047455,
      "learning_rate": 9.945740596265441e-06,
      "loss": 0.0062,
      "step": 26080
    },
    {
      "epoch": 0.4269001063568682,
      "grad_norm": 0.15152108669281006,
      "learning_rate": 9.945600645936084e-06,
      "loss": 0.0058,
      "step": 26090
    },
    {
      "epoch": 0.42706373230794403,
      "grad_norm": 0.506436288356781,
      "learning_rate": 9.945460516340589e-06,
      "loss": 0.0051,
      "step": 26100
    },
    {
      "epoch": 0.42722735825901986,
      "grad_norm": 0.31075620651245117,
      "learning_rate": 9.945320207484035e-06,
      "loss": 0.0038,
      "step": 26110
    },
    {
      "epoch": 0.42739098421009575,
      "grad_norm": 0.2524474859237671,
      "learning_rate": 9.945179719371511e-06,
      "loss": 0.0068,
      "step": 26120
    },
    {
      "epoch": 0.4275546101611716,
      "grad_norm": 0.4262329041957855,
      "learning_rate": 9.945039052008108e-06,
      "loss": 0.0057,
      "step": 26130
    },
    {
      "epoch": 0.4277182361122474,
      "grad_norm": 0.27304673194885254,
      "learning_rate": 9.944898205398923e-06,
      "loss": 0.0057,
      "step": 26140
    },
    {
      "epoch": 0.42788186206332324,
      "grad_norm": 0.47253692150115967,
      "learning_rate": 9.944757179549064e-06,
      "loss": 0.0032,
      "step": 26150
    },
    {
      "epoch": 0.42804548801439907,
      "grad_norm": 0.176385760307312,
      "learning_rate": 9.94461597446364e-06,
      "loss": 0.0045,
      "step": 26160
    },
    {
      "epoch": 0.4282091139654749,
      "grad_norm": 0.24605652689933777,
      "learning_rate": 9.944474590147773e-06,
      "loss": 0.007,
      "step": 26170
    },
    {
      "epoch": 0.4283727399165508,
      "grad_norm": 0.22253073751926422,
      "learning_rate": 9.944333026606585e-06,
      "loss": 0.0047,
      "step": 26180
    },
    {
      "epoch": 0.4285363658676266,
      "grad_norm": 0.24287839233875275,
      "learning_rate": 9.944191283845209e-06,
      "loss": 0.005,
      "step": 26190
    },
    {
      "epoch": 0.42869999181870244,
      "grad_norm": 0.26345324516296387,
      "learning_rate": 9.944049361868781e-06,
      "loss": 0.0049,
      "step": 26200
    },
    {
      "epoch": 0.42886361776977827,
      "grad_norm": 0.475466251373291,
      "learning_rate": 9.943907260682445e-06,
      "loss": 0.0046,
      "step": 26210
    },
    {
      "epoch": 0.4290272437208541,
      "grad_norm": 0.15794482827186584,
      "learning_rate": 9.943764980291355e-06,
      "loss": 0.0056,
      "step": 26220
    },
    {
      "epoch": 0.42919086967193,
      "grad_norm": 0.14782395958900452,
      "learning_rate": 9.943622520700666e-06,
      "loss": 0.0042,
      "step": 26230
    },
    {
      "epoch": 0.4293544956230058,
      "grad_norm": 0.2919323146343231,
      "learning_rate": 9.943479881915543e-06,
      "loss": 0.0039,
      "step": 26240
    },
    {
      "epoch": 0.42951812157408165,
      "grad_norm": 0.5810136795043945,
      "learning_rate": 9.943337063941157e-06,
      "loss": 0.0057,
      "step": 26250
    },
    {
      "epoch": 0.4296817475251575,
      "grad_norm": 0.3564576804637909,
      "learning_rate": 9.94319406678268e-06,
      "loss": 0.0081,
      "step": 26260
    },
    {
      "epoch": 0.4298453734762333,
      "grad_norm": 0.41717562079429626,
      "learning_rate": 9.9430508904453e-06,
      "loss": 0.003,
      "step": 26270
    },
    {
      "epoch": 0.4300089994273092,
      "grad_norm": 0.12562857568264008,
      "learning_rate": 9.942907534934207e-06,
      "loss": 0.004,
      "step": 26280
    },
    {
      "epoch": 0.430172625378385,
      "grad_norm": 0.3914327323436737,
      "learning_rate": 9.942764000254595e-06,
      "loss": 0.0049,
      "step": 26290
    },
    {
      "epoch": 0.43033625132946085,
      "grad_norm": 0.13155989348888397,
      "learning_rate": 9.942620286411667e-06,
      "loss": 0.0047,
      "step": 26300
    },
    {
      "epoch": 0.4304998772805367,
      "grad_norm": 0.39990848302841187,
      "learning_rate": 9.942476393410632e-06,
      "loss": 0.0075,
      "step": 26310
    },
    {
      "epoch": 0.4306635032316125,
      "grad_norm": 0.09514040499925613,
      "learning_rate": 9.942332321256707e-06,
      "loss": 0.0052,
      "step": 26320
    },
    {
      "epoch": 0.4308271291826884,
      "grad_norm": 0.45336511731147766,
      "learning_rate": 9.942188069955116e-06,
      "loss": 0.0058,
      "step": 26330
    },
    {
      "epoch": 0.4309907551337642,
      "grad_norm": 0.35713547468185425,
      "learning_rate": 9.942043639511085e-06,
      "loss": 0.0078,
      "step": 26340
    },
    {
      "epoch": 0.43115438108484005,
      "grad_norm": 0.38602957129478455,
      "learning_rate": 9.94189902992985e-06,
      "loss": 0.0037,
      "step": 26350
    },
    {
      "epoch": 0.4313180070359159,
      "grad_norm": 0.26430776715278625,
      "learning_rate": 9.941754241216651e-06,
      "loss": 0.0056,
      "step": 26360
    },
    {
      "epoch": 0.4314816329869917,
      "grad_norm": 0.2346680462360382,
      "learning_rate": 9.941609273376739e-06,
      "loss": 0.0046,
      "step": 26370
    },
    {
      "epoch": 0.4316452589380676,
      "grad_norm": 0.2572612464427948,
      "learning_rate": 9.941464126415367e-06,
      "loss": 0.0054,
      "step": 26380
    },
    {
      "epoch": 0.43180888488914343,
      "grad_norm": 0.28751105070114136,
      "learning_rate": 9.941318800337798e-06,
      "loss": 0.0031,
      "step": 26390
    },
    {
      "epoch": 0.43197251084021926,
      "grad_norm": 0.36122652888298035,
      "learning_rate": 9.9411732951493e-06,
      "loss": 0.0055,
      "step": 26400
    },
    {
      "epoch": 0.4321361367912951,
      "grad_norm": 0.22892005741596222,
      "learning_rate": 9.941027610855143e-06,
      "loss": 0.0055,
      "step": 26410
    },
    {
      "epoch": 0.4322997627423709,
      "grad_norm": 0.23452228307724,
      "learning_rate": 9.94088174746061e-06,
      "loss": 0.0084,
      "step": 26420
    },
    {
      "epoch": 0.4324633886934468,
      "grad_norm": 0.051503460854291916,
      "learning_rate": 9.940735704970992e-06,
      "loss": 0.0142,
      "step": 26430
    },
    {
      "epoch": 0.43262701464452263,
      "grad_norm": 0.27087894082069397,
      "learning_rate": 9.940589483391578e-06,
      "loss": 0.0045,
      "step": 26440
    },
    {
      "epoch": 0.43279064059559846,
      "grad_norm": 0.34042420983314514,
      "learning_rate": 9.94044308272767e-06,
      "loss": 0.0044,
      "step": 26450
    },
    {
      "epoch": 0.4329542665466743,
      "grad_norm": 0.29429343342781067,
      "learning_rate": 9.940296502984574e-06,
      "loss": 0.0045,
      "step": 26460
    },
    {
      "epoch": 0.4331178924977501,
      "grad_norm": 0.5314293503761292,
      "learning_rate": 9.940149744167602e-06,
      "loss": 0.0043,
      "step": 26470
    },
    {
      "epoch": 0.433281518448826,
      "grad_norm": 0.057193126529455185,
      "learning_rate": 9.940002806282077e-06,
      "loss": 0.0037,
      "step": 26480
    },
    {
      "epoch": 0.43344514439990184,
      "grad_norm": 0.43834778666496277,
      "learning_rate": 9.939855689333322e-06,
      "loss": 0.0064,
      "step": 26490
    },
    {
      "epoch": 0.43360877035097767,
      "grad_norm": 1.0762265920639038,
      "learning_rate": 9.939708393326672e-06,
      "loss": 0.0117,
      "step": 26500
    },
    {
      "epoch": 0.4337723963020535,
      "grad_norm": 0.2556087076663971,
      "learning_rate": 9.939560918267465e-06,
      "loss": 0.0065,
      "step": 26510
    },
    {
      "epoch": 0.4339360222531293,
      "grad_norm": 0.2927626371383667,
      "learning_rate": 9.939413264161047e-06,
      "loss": 0.004,
      "step": 26520
    },
    {
      "epoch": 0.4340996482042052,
      "grad_norm": 0.13159334659576416,
      "learning_rate": 9.939265431012769e-06,
      "loss": 0.0054,
      "step": 26530
    },
    {
      "epoch": 0.43426327415528104,
      "grad_norm": 0.11064198613166809,
      "learning_rate": 9.93911741882799e-06,
      "loss": 0.0031,
      "step": 26540
    },
    {
      "epoch": 0.4344269001063569,
      "grad_norm": 0.21083663403987885,
      "learning_rate": 9.938969227612076e-06,
      "loss": 0.0026,
      "step": 26550
    },
    {
      "epoch": 0.4345905260574327,
      "grad_norm": 0.13292013108730316,
      "learning_rate": 9.938820857370397e-06,
      "loss": 0.0032,
      "step": 26560
    },
    {
      "epoch": 0.43475415200850853,
      "grad_norm": 0.4474940299987793,
      "learning_rate": 9.938672308108333e-06,
      "loss": 0.0055,
      "step": 26570
    },
    {
      "epoch": 0.4349177779595844,
      "grad_norm": 0.27184492349624634,
      "learning_rate": 9.938523579831267e-06,
      "loss": 0.0038,
      "step": 26580
    },
    {
      "epoch": 0.43508140391066025,
      "grad_norm": 0.2035464495420456,
      "learning_rate": 9.93837467254459e-06,
      "loss": 0.0076,
      "step": 26590
    },
    {
      "epoch": 0.4352450298617361,
      "grad_norm": 0.2861744165420532,
      "learning_rate": 9.938225586253703e-06,
      "loss": 0.0055,
      "step": 26600
    },
    {
      "epoch": 0.4354086558128119,
      "grad_norm": 0.13672231137752533,
      "learning_rate": 9.938076320964006e-06,
      "loss": 0.0042,
      "step": 26610
    },
    {
      "epoch": 0.43557228176388774,
      "grad_norm": 0.22904746234416962,
      "learning_rate": 9.93792687668091e-06,
      "loss": 0.0038,
      "step": 26620
    },
    {
      "epoch": 0.43573590771496357,
      "grad_norm": 0.48036491870880127,
      "learning_rate": 9.937777253409832e-06,
      "loss": 0.0123,
      "step": 26630
    },
    {
      "epoch": 0.43589953366603945,
      "grad_norm": 0.08625561743974686,
      "learning_rate": 9.937627451156197e-06,
      "loss": 0.0066,
      "step": 26640
    },
    {
      "epoch": 0.4360631596171153,
      "grad_norm": 0.38035905361175537,
      "learning_rate": 9.937477469925434e-06,
      "loss": 0.0055,
      "step": 26650
    },
    {
      "epoch": 0.4362267855681911,
      "grad_norm": 0.2845790982246399,
      "learning_rate": 9.937327309722979e-06,
      "loss": 0.0054,
      "step": 26660
    },
    {
      "epoch": 0.43639041151926694,
      "grad_norm": 0.4349279999732971,
      "learning_rate": 9.937176970554278e-06,
      "loss": 0.0072,
      "step": 26670
    },
    {
      "epoch": 0.43655403747034277,
      "grad_norm": 0.5188882350921631,
      "learning_rate": 9.937026452424776e-06,
      "loss": 0.0047,
      "step": 26680
    },
    {
      "epoch": 0.43671766342141866,
      "grad_norm": 0.16659517586231232,
      "learning_rate": 9.936875755339931e-06,
      "loss": 0.0081,
      "step": 26690
    },
    {
      "epoch": 0.4368812893724945,
      "grad_norm": 0.21452824771404266,
      "learning_rate": 9.936724879305206e-06,
      "loss": 0.0042,
      "step": 26700
    },
    {
      "epoch": 0.4370449153235703,
      "grad_norm": 0.1589195877313614,
      "learning_rate": 9.93657382432607e-06,
      "loss": 0.0067,
      "step": 26710
    },
    {
      "epoch": 0.43720854127464615,
      "grad_norm": 0.3009520471096039,
      "learning_rate": 9.936422590407994e-06,
      "loss": 0.0038,
      "step": 26720
    },
    {
      "epoch": 0.437372167225722,
      "grad_norm": 0.507652223110199,
      "learning_rate": 9.936271177556467e-06,
      "loss": 0.0073,
      "step": 26730
    },
    {
      "epoch": 0.43753579317679786,
      "grad_norm": 0.12943866848945618,
      "learning_rate": 9.936119585776973e-06,
      "loss": 0.0048,
      "step": 26740
    },
    {
      "epoch": 0.4376994191278737,
      "grad_norm": 0.3426955044269562,
      "learning_rate": 9.935967815075007e-06,
      "loss": 0.0065,
      "step": 26750
    },
    {
      "epoch": 0.4378630450789495,
      "grad_norm": 0.3047889471054077,
      "learning_rate": 9.935815865456073e-06,
      "loss": 0.0075,
      "step": 26760
    },
    {
      "epoch": 0.43802667103002535,
      "grad_norm": 0.36037224531173706,
      "learning_rate": 9.935663736925675e-06,
      "loss": 0.0041,
      "step": 26770
    },
    {
      "epoch": 0.4381902969811012,
      "grad_norm": 0.5686681866645813,
      "learning_rate": 9.935511429489328e-06,
      "loss": 0.0058,
      "step": 26780
    },
    {
      "epoch": 0.43835392293217706,
      "grad_norm": 0.5435603260993958,
      "learning_rate": 9.935358943152557e-06,
      "loss": 0.0059,
      "step": 26790
    },
    {
      "epoch": 0.4385175488832529,
      "grad_norm": 0.28427553176879883,
      "learning_rate": 9.935206277920884e-06,
      "loss": 0.0111,
      "step": 26800
    },
    {
      "epoch": 0.4386811748343287,
      "grad_norm": 0.17619702219963074,
      "learning_rate": 9.935053433799846e-06,
      "loss": 0.0034,
      "step": 26810
    },
    {
      "epoch": 0.43884480078540455,
      "grad_norm": 0.12450452148914337,
      "learning_rate": 9.934900410794981e-06,
      "loss": 0.0049,
      "step": 26820
    },
    {
      "epoch": 0.4390084267364804,
      "grad_norm": 0.2546733319759369,
      "learning_rate": 9.934747208911837e-06,
      "loss": 0.0054,
      "step": 26830
    },
    {
      "epoch": 0.43917205268755627,
      "grad_norm": 0.2232169359922409,
      "learning_rate": 9.934593828155966e-06,
      "loss": 0.0039,
      "step": 26840
    },
    {
      "epoch": 0.4393356786386321,
      "grad_norm": 0.33902662992477417,
      "learning_rate": 9.934440268532931e-06,
      "loss": 0.0071,
      "step": 26850
    },
    {
      "epoch": 0.43949930458970793,
      "grad_norm": 0.24697628617286682,
      "learning_rate": 9.934286530048294e-06,
      "loss": 0.0045,
      "step": 26860
    },
    {
      "epoch": 0.43966293054078376,
      "grad_norm": 0.31107616424560547,
      "learning_rate": 9.934132612707631e-06,
      "loss": 0.0044,
      "step": 26870
    },
    {
      "epoch": 0.4398265564918596,
      "grad_norm": 0.3301653265953064,
      "learning_rate": 9.93397851651652e-06,
      "loss": 0.0066,
      "step": 26880
    },
    {
      "epoch": 0.4399901824429355,
      "grad_norm": 0.2726227045059204,
      "learning_rate": 9.933824241480545e-06,
      "loss": 0.0041,
      "step": 26890
    },
    {
      "epoch": 0.4401538083940113,
      "grad_norm": 0.24102412164211273,
      "learning_rate": 9.9336697876053e-06,
      "loss": 0.0041,
      "step": 26900
    },
    {
      "epoch": 0.44031743434508713,
      "grad_norm": 0.18632589280605316,
      "learning_rate": 9.933515154896383e-06,
      "loss": 0.0042,
      "step": 26910
    },
    {
      "epoch": 0.44048106029616296,
      "grad_norm": 0.1324058473110199,
      "learning_rate": 9.933360343359399e-06,
      "loss": 0.0042,
      "step": 26920
    },
    {
      "epoch": 0.4406446862472388,
      "grad_norm": 0.32733872532844543,
      "learning_rate": 9.933205352999958e-06,
      "loss": 0.0078,
      "step": 26930
    },
    {
      "epoch": 0.4408083121983147,
      "grad_norm": 0.2764756381511688,
      "learning_rate": 9.933050183823681e-06,
      "loss": 0.0054,
      "step": 26940
    },
    {
      "epoch": 0.4409719381493905,
      "grad_norm": 0.1801196038722992,
      "learning_rate": 9.932894835836191e-06,
      "loss": 0.0047,
      "step": 26950
    },
    {
      "epoch": 0.44113556410046634,
      "grad_norm": 0.11624041944742203,
      "learning_rate": 9.93273930904312e-06,
      "loss": 0.0035,
      "step": 26960
    },
    {
      "epoch": 0.44129919005154217,
      "grad_norm": 0.48371636867523193,
      "learning_rate": 9.932583603450103e-06,
      "loss": 0.0077,
      "step": 26970
    },
    {
      "epoch": 0.441462816002618,
      "grad_norm": 0.4422214925289154,
      "learning_rate": 9.932427719062787e-06,
      "loss": 0.0048,
      "step": 26980
    },
    {
      "epoch": 0.4416264419536939,
      "grad_norm": 0.23202046751976013,
      "learning_rate": 9.93227165588682e-06,
      "loss": 0.0051,
      "step": 26990
    },
    {
      "epoch": 0.4417900679047697,
      "grad_norm": 0.4644702970981598,
      "learning_rate": 9.93211541392786e-06,
      "loss": 0.0042,
      "step": 27000
    },
    {
      "epoch": 0.44195369385584554,
      "grad_norm": 0.1372828185558319,
      "learning_rate": 9.93195899319157e-06,
      "loss": 0.004,
      "step": 27010
    },
    {
      "epoch": 0.44211731980692137,
      "grad_norm": 0.24187833070755005,
      "learning_rate": 9.93180239368362e-06,
      "loss": 0.0075,
      "step": 27020
    },
    {
      "epoch": 0.4422809457579972,
      "grad_norm": 0.48835405707359314,
      "learning_rate": 9.931645615409687e-06,
      "loss": 0.0047,
      "step": 27030
    },
    {
      "epoch": 0.4424445717090731,
      "grad_norm": 0.12844069302082062,
      "learning_rate": 9.931488658375454e-06,
      "loss": 0.0052,
      "step": 27040
    },
    {
      "epoch": 0.4426081976601489,
      "grad_norm": 0.13725093007087708,
      "learning_rate": 9.931331522586608e-06,
      "loss": 0.0044,
      "step": 27050
    },
    {
      "epoch": 0.44277182361122475,
      "grad_norm": 0.3767290413379669,
      "learning_rate": 9.931174208048849e-06,
      "loss": 0.0047,
      "step": 27060
    },
    {
      "epoch": 0.4429354495623006,
      "grad_norm": 0.14739324152469635,
      "learning_rate": 9.931016714767874e-06,
      "loss": 0.0041,
      "step": 27070
    },
    {
      "epoch": 0.4430990755133764,
      "grad_norm": 0.18387803435325623,
      "learning_rate": 9.930859042749397e-06,
      "loss": 0.0061,
      "step": 27080
    },
    {
      "epoch": 0.44326270146445224,
      "grad_norm": 0.1818620264530182,
      "learning_rate": 9.930701191999127e-06,
      "loss": 0.0042,
      "step": 27090
    },
    {
      "epoch": 0.4434263274155281,
      "grad_norm": 0.33416905999183655,
      "learning_rate": 9.930543162522794e-06,
      "loss": 0.0072,
      "step": 27100
    },
    {
      "epoch": 0.44358995336660395,
      "grad_norm": 0.18846382200717926,
      "learning_rate": 9.930384954326118e-06,
      "loss": 0.0039,
      "step": 27110
    },
    {
      "epoch": 0.4437535793176798,
      "grad_norm": 0.1889810413122177,
      "learning_rate": 9.930226567414838e-06,
      "loss": 0.0036,
      "step": 27120
    },
    {
      "epoch": 0.4439172052687556,
      "grad_norm": 0.1603645533323288,
      "learning_rate": 9.930068001794696e-06,
      "loss": 0.0049,
      "step": 27130
    },
    {
      "epoch": 0.44408083121983144,
      "grad_norm": 0.11223021149635315,
      "learning_rate": 9.929909257471437e-06,
      "loss": 0.0043,
      "step": 27140
    },
    {
      "epoch": 0.4442444571709073,
      "grad_norm": 0.08833596855401993,
      "learning_rate": 9.929750334450816e-06,
      "loss": 0.0042,
      "step": 27150
    },
    {
      "epoch": 0.44440808312198316,
      "grad_norm": 0.19301727414131165,
      "learning_rate": 9.929591232738595e-06,
      "loss": 0.0042,
      "step": 27160
    },
    {
      "epoch": 0.444571709073059,
      "grad_norm": 1.130907654762268,
      "learning_rate": 9.929431952340539e-06,
      "loss": 0.0048,
      "step": 27170
    },
    {
      "epoch": 0.4447353350241348,
      "grad_norm": 0.2923726737499237,
      "learning_rate": 9.92927249326242e-06,
      "loss": 0.0039,
      "step": 27180
    },
    {
      "epoch": 0.44489896097521064,
      "grad_norm": 0.19610510766506195,
      "learning_rate": 9.929112855510023e-06,
      "loss": 0.0051,
      "step": 27190
    },
    {
      "epoch": 0.44506258692628653,
      "grad_norm": 0.3318614363670349,
      "learning_rate": 9.928953039089131e-06,
      "loss": 0.0081,
      "step": 27200
    },
    {
      "epoch": 0.44522621287736236,
      "grad_norm": 0.3307742774486542,
      "learning_rate": 9.928793044005538e-06,
      "loss": 0.0053,
      "step": 27210
    },
    {
      "epoch": 0.4453898388284382,
      "grad_norm": 0.36365392804145813,
      "learning_rate": 9.928632870265044e-06,
      "loss": 0.0059,
      "step": 27220
    },
    {
      "epoch": 0.445553464779514,
      "grad_norm": 0.16435521841049194,
      "learning_rate": 9.928472517873454e-06,
      "loss": 0.0036,
      "step": 27230
    },
    {
      "epoch": 0.44571709073058985,
      "grad_norm": 0.2523607313632965,
      "learning_rate": 9.928311986836578e-06,
      "loss": 0.0036,
      "step": 27240
    },
    {
      "epoch": 0.44588071668166573,
      "grad_norm": 0.16454745829105377,
      "learning_rate": 9.92815127716024e-06,
      "loss": 0.0033,
      "step": 27250
    },
    {
      "epoch": 0.44604434263274156,
      "grad_norm": 0.677886426448822,
      "learning_rate": 9.927990388850261e-06,
      "loss": 0.0075,
      "step": 27260
    },
    {
      "epoch": 0.4462079685838174,
      "grad_norm": 0.16890017688274384,
      "learning_rate": 9.927829321912477e-06,
      "loss": 0.0057,
      "step": 27270
    },
    {
      "epoch": 0.4463715945348932,
      "grad_norm": 0.2555929124355316,
      "learning_rate": 9.927668076352721e-06,
      "loss": 0.0044,
      "step": 27280
    },
    {
      "epoch": 0.44653522048596905,
      "grad_norm": 0.20007185637950897,
      "learning_rate": 9.927506652176842e-06,
      "loss": 0.0051,
      "step": 27290
    },
    {
      "epoch": 0.44669884643704494,
      "grad_norm": 0.1796633005142212,
      "learning_rate": 9.927345049390688e-06,
      "loss": 0.0038,
      "step": 27300
    },
    {
      "epoch": 0.44686247238812077,
      "grad_norm": 0.3198784589767456,
      "learning_rate": 9.92718326800012e-06,
      "loss": 0.0039,
      "step": 27310
    },
    {
      "epoch": 0.4470260983391966,
      "grad_norm": 0.09753880649805069,
      "learning_rate": 9.927021308011003e-06,
      "loss": 0.0049,
      "step": 27320
    },
    {
      "epoch": 0.44718972429027243,
      "grad_norm": 0.24289223551750183,
      "learning_rate": 9.926859169429201e-06,
      "loss": 0.0032,
      "step": 27330
    },
    {
      "epoch": 0.44735335024134826,
      "grad_norm": 0.16761480271816254,
      "learning_rate": 9.926696852260598e-06,
      "loss": 0.0066,
      "step": 27340
    },
    {
      "epoch": 0.44751697619242414,
      "grad_norm": 0.2215719074010849,
      "learning_rate": 9.926534356511075e-06,
      "loss": 0.0047,
      "step": 27350
    },
    {
      "epoch": 0.4476806021435,
      "grad_norm": 0.3459492325782776,
      "learning_rate": 9.926371682186522e-06,
      "loss": 0.005,
      "step": 27360
    },
    {
      "epoch": 0.4478442280945758,
      "grad_norm": 0.14066335558891296,
      "learning_rate": 9.926208829292837e-06,
      "loss": 0.0048,
      "step": 27370
    },
    {
      "epoch": 0.44800785404565163,
      "grad_norm": 0.4356427490711212,
      "learning_rate": 9.92604579783592e-06,
      "loss": 0.0042,
      "step": 27380
    },
    {
      "epoch": 0.44817147999672746,
      "grad_norm": 0.18119563162326813,
      "learning_rate": 9.925882587821682e-06,
      "loss": 0.0054,
      "step": 27390
    },
    {
      "epoch": 0.44833510594780335,
      "grad_norm": 0.43294450640678406,
      "learning_rate": 9.92571919925604e-06,
      "loss": 0.0048,
      "step": 27400
    },
    {
      "epoch": 0.4484987318988792,
      "grad_norm": 0.20888350903987885,
      "learning_rate": 9.925555632144916e-06,
      "loss": 0.0038,
      "step": 27410
    },
    {
      "epoch": 0.448662357849955,
      "grad_norm": 0.23409788310527802,
      "learning_rate": 9.925391886494237e-06,
      "loss": 0.0047,
      "step": 27420
    },
    {
      "epoch": 0.44882598380103084,
      "grad_norm": 0.19678986072540283,
      "learning_rate": 9.925227962309942e-06,
      "loss": 0.0028,
      "step": 27430
    },
    {
      "epoch": 0.44898960975210667,
      "grad_norm": 0.23898477852344513,
      "learning_rate": 9.92506385959797e-06,
      "loss": 0.003,
      "step": 27440
    },
    {
      "epoch": 0.44915323570318255,
      "grad_norm": 0.2357511669397354,
      "learning_rate": 9.92489957836427e-06,
      "loss": 0.0062,
      "step": 27450
    },
    {
      "epoch": 0.4493168616542584,
      "grad_norm": 0.3089338541030884,
      "learning_rate": 9.924735118614797e-06,
      "loss": 0.0071,
      "step": 27460
    },
    {
      "epoch": 0.4494804876053342,
      "grad_norm": 0.19275996088981628,
      "learning_rate": 9.924570480355513e-06,
      "loss": 0.0065,
      "step": 27470
    },
    {
      "epoch": 0.44964411355641004,
      "grad_norm": 0.02500307373702526,
      "learning_rate": 9.924405663592384e-06,
      "loss": 0.0059,
      "step": 27480
    },
    {
      "epoch": 0.44980773950748587,
      "grad_norm": 0.6308009624481201,
      "learning_rate": 9.924240668331386e-06,
      "loss": 0.0038,
      "step": 27490
    },
    {
      "epoch": 0.4499713654585617,
      "grad_norm": 0.380270779132843,
      "learning_rate": 9.924075494578498e-06,
      "loss": 0.0039,
      "step": 27500
    },
    {
      "epoch": 0.4501349914096376,
      "grad_norm": 0.37860944867134094,
      "learning_rate": 9.923910142339708e-06,
      "loss": 0.0051,
      "step": 27510
    },
    {
      "epoch": 0.4502986173607134,
      "grad_norm": 0.2461448758840561,
      "learning_rate": 9.92374461162101e-06,
      "loss": 0.004,
      "step": 27520
    },
    {
      "epoch": 0.45046224331178925,
      "grad_norm": 0.35521289706230164,
      "learning_rate": 9.923578902428403e-06,
      "loss": 0.0058,
      "step": 27530
    },
    {
      "epoch": 0.4506258692628651,
      "grad_norm": 0.11499812453985214,
      "learning_rate": 9.923413014767897e-06,
      "loss": 0.0052,
      "step": 27540
    },
    {
      "epoch": 0.4507894952139409,
      "grad_norm": 0.30423709750175476,
      "learning_rate": 9.923246948645499e-06,
      "loss": 0.0037,
      "step": 27550
    },
    {
      "epoch": 0.4509531211650168,
      "grad_norm": 0.17959947884082794,
      "learning_rate": 9.923080704067233e-06,
      "loss": 0.0039,
      "step": 27560
    },
    {
      "epoch": 0.4511167471160926,
      "grad_norm": 0.25133025646209717,
      "learning_rate": 9.922914281039124e-06,
      "loss": 0.0034,
      "step": 27570
    },
    {
      "epoch": 0.45128037306716845,
      "grad_norm": 0.38148918747901917,
      "learning_rate": 9.922747679567206e-06,
      "loss": 0.0043,
      "step": 27580
    },
    {
      "epoch": 0.4514439990182443,
      "grad_norm": 0.6596552729606628,
      "learning_rate": 9.922580899657514e-06,
      "loss": 0.0116,
      "step": 27590
    },
    {
      "epoch": 0.4516076249693201,
      "grad_norm": 0.2443445473909378,
      "learning_rate": 9.922413941316095e-06,
      "loss": 0.0065,
      "step": 27600
    },
    {
      "epoch": 0.451771250920396,
      "grad_norm": 0.36659279465675354,
      "learning_rate": 9.922246804549002e-06,
      "loss": 0.005,
      "step": 27610
    },
    {
      "epoch": 0.4519348768714718,
      "grad_norm": 0.32651492953300476,
      "learning_rate": 9.922079489362294e-06,
      "loss": 0.0068,
      "step": 27620
    },
    {
      "epoch": 0.45209850282254765,
      "grad_norm": 0.2040097564458847,
      "learning_rate": 9.921911995762032e-06,
      "loss": 0.0046,
      "step": 27630
    },
    {
      "epoch": 0.4522621287736235,
      "grad_norm": 0.4699828624725342,
      "learning_rate": 9.921744323754292e-06,
      "loss": 0.0051,
      "step": 27640
    },
    {
      "epoch": 0.4524257547246993,
      "grad_norm": 0.2855335772037506,
      "learning_rate": 9.92157647334515e-06,
      "loss": 0.0048,
      "step": 27650
    },
    {
      "epoch": 0.4525893806757752,
      "grad_norm": 0.3780006468296051,
      "learning_rate": 9.921408444540687e-06,
      "loss": 0.0051,
      "step": 27660
    },
    {
      "epoch": 0.45275300662685103,
      "grad_norm": 0.30700623989105225,
      "learning_rate": 9.921240237346996e-06,
      "loss": 0.0047,
      "step": 27670
    },
    {
      "epoch": 0.45291663257792686,
      "grad_norm": 0.3167283833026886,
      "learning_rate": 9.921071851770177e-06,
      "loss": 0.0056,
      "step": 27680
    },
    {
      "epoch": 0.4530802585290027,
      "grad_norm": 0.26490625739097595,
      "learning_rate": 9.920903287816329e-06,
      "loss": 0.0042,
      "step": 27690
    },
    {
      "epoch": 0.4532438844800785,
      "grad_norm": 0.13398846983909607,
      "learning_rate": 9.920734545491564e-06,
      "loss": 0.0051,
      "step": 27700
    },
    {
      "epoch": 0.4534075104311544,
      "grad_norm": 0.23275910317897797,
      "learning_rate": 9.920565624801997e-06,
      "loss": 0.0038,
      "step": 27710
    },
    {
      "epoch": 0.45357113638223023,
      "grad_norm": 0.296029657125473,
      "learning_rate": 9.920396525753755e-06,
      "loss": 0.0034,
      "step": 27720
    },
    {
      "epoch": 0.45373476233330606,
      "grad_norm": 0.24191559851169586,
      "learning_rate": 9.920227248352963e-06,
      "loss": 0.0044,
      "step": 27730
    },
    {
      "epoch": 0.4538983882843819,
      "grad_norm": 0.24207191169261932,
      "learning_rate": 9.92005779260576e-06,
      "loss": 0.0056,
      "step": 27740
    },
    {
      "epoch": 0.4540620142354577,
      "grad_norm": 0.1859743744134903,
      "learning_rate": 9.919888158518286e-06,
      "loss": 0.0046,
      "step": 27750
    },
    {
      "epoch": 0.4542256401865336,
      "grad_norm": 0.4077349007129669,
      "learning_rate": 9.919718346096692e-06,
      "loss": 0.004,
      "step": 27760
    },
    {
      "epoch": 0.45438926613760944,
      "grad_norm": 0.39792948961257935,
      "learning_rate": 9.91954835534713e-06,
      "loss": 0.0059,
      "step": 27770
    },
    {
      "epoch": 0.45455289208868527,
      "grad_norm": 0.46911177039146423,
      "learning_rate": 9.919378186275767e-06,
      "loss": 0.0045,
      "step": 27780
    },
    {
      "epoch": 0.4547165180397611,
      "grad_norm": 0.14708006381988525,
      "learning_rate": 9.919207838888767e-06,
      "loss": 0.004,
      "step": 27790
    },
    {
      "epoch": 0.4548801439908369,
      "grad_norm": 0.5174422264099121,
      "learning_rate": 9.919037313192305e-06,
      "loss": 0.0037,
      "step": 27800
    },
    {
      "epoch": 0.4550437699419128,
      "grad_norm": 0.20993366837501526,
      "learning_rate": 9.918866609192563e-06,
      "loss": 0.0056,
      "step": 27810
    },
    {
      "epoch": 0.45520739589298864,
      "grad_norm": 0.22028259932994843,
      "learning_rate": 9.91869572689573e-06,
      "loss": 0.0055,
      "step": 27820
    },
    {
      "epoch": 0.45537102184406447,
      "grad_norm": 0.076836496591568,
      "learning_rate": 9.918524666307998e-06,
      "loss": 0.0033,
      "step": 27830
    },
    {
      "epoch": 0.4555346477951403,
      "grad_norm": 0.04713664948940277,
      "learning_rate": 9.918353427435568e-06,
      "loss": 0.0041,
      "step": 27840
    },
    {
      "epoch": 0.45569827374621613,
      "grad_norm": 0.2616214454174042,
      "learning_rate": 9.918182010284645e-06,
      "loss": 0.0062,
      "step": 27850
    },
    {
      "epoch": 0.455861899697292,
      "grad_norm": 0.13661831617355347,
      "learning_rate": 9.918010414861448e-06,
      "loss": 0.0056,
      "step": 27860
    },
    {
      "epoch": 0.45602552564836785,
      "grad_norm": 0.19420835375785828,
      "learning_rate": 9.917838641172193e-06,
      "loss": 0.0046,
      "step": 27870
    },
    {
      "epoch": 0.4561891515994437,
      "grad_norm": 0.17217794060707092,
      "learning_rate": 9.917666689223106e-06,
      "loss": 0.0051,
      "step": 27880
    },
    {
      "epoch": 0.4563527775505195,
      "grad_norm": 0.17375342547893524,
      "learning_rate": 9.91749455902042e-06,
      "loss": 0.0044,
      "step": 27890
    },
    {
      "epoch": 0.45651640350159534,
      "grad_norm": 0.23703983426094055,
      "learning_rate": 9.917322250570378e-06,
      "loss": 0.0035,
      "step": 27900
    },
    {
      "epoch": 0.4566800294526712,
      "grad_norm": 0.5515260696411133,
      "learning_rate": 9.91714976387922e-06,
      "loss": 0.0052,
      "step": 27910
    },
    {
      "epoch": 0.45684365540374705,
      "grad_norm": 0.06264668703079224,
      "learning_rate": 9.916977098953202e-06,
      "loss": 0.0063,
      "step": 27920
    },
    {
      "epoch": 0.4570072813548229,
      "grad_norm": 0.12721134722232819,
      "learning_rate": 9.916804255798584e-06,
      "loss": 0.0051,
      "step": 27930
    },
    {
      "epoch": 0.4571709073058987,
      "grad_norm": 0.27503839135169983,
      "learning_rate": 9.916631234421627e-06,
      "loss": 0.0076,
      "step": 27940
    },
    {
      "epoch": 0.45733453325697454,
      "grad_norm": 0.12606807053089142,
      "learning_rate": 9.916458034828605e-06,
      "loss": 0.0053,
      "step": 27950
    },
    {
      "epoch": 0.45749815920805037,
      "grad_norm": 0.28796064853668213,
      "learning_rate": 9.916284657025795e-06,
      "loss": 0.004,
      "step": 27960
    },
    {
      "epoch": 0.45766178515912626,
      "grad_norm": 0.19446000456809998,
      "learning_rate": 9.91611110101948e-06,
      "loss": 0.0037,
      "step": 27970
    },
    {
      "epoch": 0.4578254111102021,
      "grad_norm": 0.35852789878845215,
      "learning_rate": 9.915937366815958e-06,
      "loss": 0.0063,
      "step": 27980
    },
    {
      "epoch": 0.4579890370612779,
      "grad_norm": 0.2826062738895416,
      "learning_rate": 9.915763454421519e-06,
      "loss": 0.003,
      "step": 27990
    },
    {
      "epoch": 0.45815266301235374,
      "grad_norm": 0.2767725884914398,
      "learning_rate": 9.91558936384247e-06,
      "loss": 0.0036,
      "step": 28000
    },
    {
      "epoch": 0.4583162889634296,
      "grad_norm": 0.5194330215454102,
      "learning_rate": 9.91541509508512e-06,
      "loss": 0.0043,
      "step": 28010
    },
    {
      "epoch": 0.45847991491450546,
      "grad_norm": 0.16330242156982422,
      "learning_rate": 9.915240648155787e-06,
      "loss": 0.0073,
      "step": 28020
    },
    {
      "epoch": 0.4586435408655813,
      "grad_norm": 0.09129302203655243,
      "learning_rate": 9.915066023060794e-06,
      "loss": 0.0048,
      "step": 28030
    },
    {
      "epoch": 0.4588071668166571,
      "grad_norm": 0.31927719712257385,
      "learning_rate": 9.91489121980647e-06,
      "loss": 0.0058,
      "step": 28040
    },
    {
      "epoch": 0.45897079276773295,
      "grad_norm": 0.10923999547958374,
      "learning_rate": 9.914716238399153e-06,
      "loss": 0.0081,
      "step": 28050
    },
    {
      "epoch": 0.4591344187188088,
      "grad_norm": 0.1889079511165619,
      "learning_rate": 9.914541078845186e-06,
      "loss": 0.0039,
      "step": 28060
    },
    {
      "epoch": 0.45929804466988466,
      "grad_norm": 0.22986048460006714,
      "learning_rate": 9.914365741150915e-06,
      "loss": 0.0058,
      "step": 28070
    },
    {
      "epoch": 0.4594616706209605,
      "grad_norm": 0.10280491411685944,
      "learning_rate": 9.914190225322697e-06,
      "loss": 0.0038,
      "step": 28080
    },
    {
      "epoch": 0.4596252965720363,
      "grad_norm": 0.7463960647583008,
      "learning_rate": 9.914014531366894e-06,
      "loss": 0.0066,
      "step": 28090
    },
    {
      "epoch": 0.45978892252311215,
      "grad_norm": 0.04366424307227135,
      "learning_rate": 9.913838659289875e-06,
      "loss": 0.006,
      "step": 28100
    },
    {
      "epoch": 0.459952548474188,
      "grad_norm": 0.17364706099033356,
      "learning_rate": 9.913662609098015e-06,
      "loss": 0.0055,
      "step": 28110
    },
    {
      "epoch": 0.46011617442526387,
      "grad_norm": 0.40444594621658325,
      "learning_rate": 9.913486380797696e-06,
      "loss": 0.0052,
      "step": 28120
    },
    {
      "epoch": 0.4602798003763397,
      "grad_norm": 0.2098781317472458,
      "learning_rate": 9.913309974395305e-06,
      "loss": 0.0038,
      "step": 28130
    },
    {
      "epoch": 0.46044342632741553,
      "grad_norm": 0.10730168223381042,
      "learning_rate": 9.913133389897235e-06,
      "loss": 0.0043,
      "step": 28140
    },
    {
      "epoch": 0.46060705227849136,
      "grad_norm": 0.1922830194234848,
      "learning_rate": 9.912956627309888e-06,
      "loss": 0.004,
      "step": 28150
    },
    {
      "epoch": 0.4607706782295672,
      "grad_norm": 0.1791941225528717,
      "learning_rate": 9.912779686639671e-06,
      "loss": 0.0064,
      "step": 28160
    },
    {
      "epoch": 0.4609343041806431,
      "grad_norm": 0.2935792803764343,
      "learning_rate": 9.912602567893e-06,
      "loss": 0.0084,
      "step": 28170
    },
    {
      "epoch": 0.4610979301317189,
      "grad_norm": 0.28350767493247986,
      "learning_rate": 9.912425271076291e-06,
      "loss": 0.0053,
      "step": 28180
    },
    {
      "epoch": 0.46126155608279473,
      "grad_norm": 0.12251194566488266,
      "learning_rate": 9.912247796195975e-06,
      "loss": 0.0052,
      "step": 28190
    },
    {
      "epoch": 0.46142518203387056,
      "grad_norm": 0.21796534955501556,
      "learning_rate": 9.91207014325848e-06,
      "loss": 0.0041,
      "step": 28200
    },
    {
      "epoch": 0.4615888079849464,
      "grad_norm": 0.11030439287424088,
      "learning_rate": 9.911892312270248e-06,
      "loss": 0.0049,
      "step": 28210
    },
    {
      "epoch": 0.4617524339360223,
      "grad_norm": 0.3622218072414398,
      "learning_rate": 9.911714303237728e-06,
      "loss": 0.0033,
      "step": 28220
    },
    {
      "epoch": 0.4619160598870981,
      "grad_norm": 0.3663948178291321,
      "learning_rate": 9.911536116167367e-06,
      "loss": 0.0069,
      "step": 28230
    },
    {
      "epoch": 0.46207968583817394,
      "grad_norm": 0.16379572451114655,
      "learning_rate": 9.911357751065628e-06,
      "loss": 0.007,
      "step": 28240
    },
    {
      "epoch": 0.46224331178924977,
      "grad_norm": 0.23335902392864227,
      "learning_rate": 9.911179207938972e-06,
      "loss": 0.0039,
      "step": 28250
    },
    {
      "epoch": 0.4624069377403256,
      "grad_norm": 0.22836972773075104,
      "learning_rate": 9.911000486793876e-06,
      "loss": 0.0047,
      "step": 28260
    },
    {
      "epoch": 0.4625705636914015,
      "grad_norm": 1.1119749546051025,
      "learning_rate": 9.910821587636814e-06,
      "loss": 0.0039,
      "step": 28270
    },
    {
      "epoch": 0.4627341896424773,
      "grad_norm": 0.3211715519428253,
      "learning_rate": 9.910642510474273e-06,
      "loss": 0.0071,
      "step": 28280
    },
    {
      "epoch": 0.46289781559355314,
      "grad_norm": 0.2997148931026459,
      "learning_rate": 9.910463255312744e-06,
      "loss": 0.0039,
      "step": 28290
    },
    {
      "epoch": 0.46306144154462897,
      "grad_norm": 0.16480353474617004,
      "learning_rate": 9.910283822158724e-06,
      "loss": 0.0058,
      "step": 28300
    },
    {
      "epoch": 0.4632250674957048,
      "grad_norm": 0.5872746109962463,
      "learning_rate": 9.910104211018716e-06,
      "loss": 0.0096,
      "step": 28310
    },
    {
      "epoch": 0.4633886934467807,
      "grad_norm": 0.2717980146408081,
      "learning_rate": 9.909924421899232e-06,
      "loss": 0.0048,
      "step": 28320
    },
    {
      "epoch": 0.4635523193978565,
      "grad_norm": 0.3067272901535034,
      "learning_rate": 9.909744454806789e-06,
      "loss": 0.0057,
      "step": 28330
    },
    {
      "epoch": 0.46371594534893235,
      "grad_norm": 0.22667890787124634,
      "learning_rate": 9.909564309747908e-06,
      "loss": 0.0057,
      "step": 28340
    },
    {
      "epoch": 0.4638795713000082,
      "grad_norm": 0.23339539766311646,
      "learning_rate": 9.909383986729121e-06,
      "loss": 0.0036,
      "step": 28350
    },
    {
      "epoch": 0.464043197251084,
      "grad_norm": 0.28694164752960205,
      "learning_rate": 9.909203485756966e-06,
      "loss": 0.0055,
      "step": 28360
    },
    {
      "epoch": 0.46420682320215984,
      "grad_norm": 0.04660428687930107,
      "learning_rate": 9.909022806837983e-06,
      "loss": 0.0038,
      "step": 28370
    },
    {
      "epoch": 0.4643704491532357,
      "grad_norm": 0.046696942299604416,
      "learning_rate": 9.90884194997872e-06,
      "loss": 0.0051,
      "step": 28380
    },
    {
      "epoch": 0.46453407510431155,
      "grad_norm": 0.259512335062027,
      "learning_rate": 9.908660915185734e-06,
      "loss": 0.0036,
      "step": 28390
    },
    {
      "epoch": 0.4646977010553874,
      "grad_norm": 0.6218687295913696,
      "learning_rate": 9.90847970246559e-06,
      "loss": 0.0059,
      "step": 28400
    },
    {
      "epoch": 0.4648613270064632,
      "grad_norm": 0.3486070930957794,
      "learning_rate": 9.908298311824853e-06,
      "loss": 0.0065,
      "step": 28410
    },
    {
      "epoch": 0.46502495295753904,
      "grad_norm": 0.2569338381290436,
      "learning_rate": 9.908116743270098e-06,
      "loss": 0.0045,
      "step": 28420
    },
    {
      "epoch": 0.4651885789086149,
      "grad_norm": 0.25672534108161926,
      "learning_rate": 9.907934996807908e-06,
      "loss": 0.0035,
      "step": 28430
    },
    {
      "epoch": 0.46535220485969075,
      "grad_norm": 0.09729569405317307,
      "learning_rate": 9.907753072444871e-06,
      "loss": 0.0034,
      "step": 28440
    },
    {
      "epoch": 0.4655158308107666,
      "grad_norm": 0.17873574793338776,
      "learning_rate": 9.907570970187582e-06,
      "loss": 0.0044,
      "step": 28450
    },
    {
      "epoch": 0.4656794567618424,
      "grad_norm": 0.15932105481624603,
      "learning_rate": 9.907388690042636e-06,
      "loss": 0.0041,
      "step": 28460
    },
    {
      "epoch": 0.46584308271291824,
      "grad_norm": 0.0673116147518158,
      "learning_rate": 9.907206232016648e-06,
      "loss": 0.0064,
      "step": 28470
    },
    {
      "epoch": 0.46600670866399413,
      "grad_norm": 0.11695891618728638,
      "learning_rate": 9.907023596116228e-06,
      "loss": 0.0044,
      "step": 28480
    },
    {
      "epoch": 0.46617033461506996,
      "grad_norm": 0.14565777778625488,
      "learning_rate": 9.906840782347995e-06,
      "loss": 0.0048,
      "step": 28490
    },
    {
      "epoch": 0.4663339605661458,
      "grad_norm": 0.3451426923274994,
      "learning_rate": 9.906657790718577e-06,
      "loss": 0.0036,
      "step": 28500
    },
    {
      "epoch": 0.4664975865172216,
      "grad_norm": 0.2421916425228119,
      "learning_rate": 9.906474621234607e-06,
      "loss": 0.006,
      "step": 28510
    },
    {
      "epoch": 0.46666121246829745,
      "grad_norm": 0.27039963006973267,
      "learning_rate": 9.906291273902726e-06,
      "loss": 0.0053,
      "step": 28520
    },
    {
      "epoch": 0.46682483841937333,
      "grad_norm": 0.14194738864898682,
      "learning_rate": 9.906107748729577e-06,
      "loss": 0.0043,
      "step": 28530
    },
    {
      "epoch": 0.46698846437044916,
      "grad_norm": 0.15197667479515076,
      "learning_rate": 9.905924045721814e-06,
      "loss": 0.0037,
      "step": 28540
    },
    {
      "epoch": 0.467152090321525,
      "grad_norm": 0.45280808210372925,
      "learning_rate": 9.905740164886095e-06,
      "loss": 0.0046,
      "step": 28550
    },
    {
      "epoch": 0.4673157162726008,
      "grad_norm": 0.06453012675046921,
      "learning_rate": 9.905556106229085e-06,
      "loss": 0.0063,
      "step": 28560
    },
    {
      "epoch": 0.46747934222367665,
      "grad_norm": 0.1308635026216507,
      "learning_rate": 9.905371869757458e-06,
      "loss": 0.0067,
      "step": 28570
    },
    {
      "epoch": 0.46764296817475254,
      "grad_norm": 0.40309959650039673,
      "learning_rate": 9.90518745547789e-06,
      "loss": 0.0073,
      "step": 28580
    },
    {
      "epoch": 0.46780659412582837,
      "grad_norm": 0.3105323016643524,
      "learning_rate": 9.905002863397066e-06,
      "loss": 0.0037,
      "step": 28590
    },
    {
      "epoch": 0.4679702200769042,
      "grad_norm": 0.3754200339317322,
      "learning_rate": 9.904818093521678e-06,
      "loss": 0.0034,
      "step": 28600
    },
    {
      "epoch": 0.46813384602798,
      "grad_norm": 0.11845695227384567,
      "learning_rate": 9.90463314585842e-06,
      "loss": 0.0042,
      "step": 28610
    },
    {
      "epoch": 0.46829747197905586,
      "grad_norm": 0.05855477228760719,
      "learning_rate": 9.904448020414e-06,
      "loss": 0.0059,
      "step": 28620
    },
    {
      "epoch": 0.46846109793013174,
      "grad_norm": 0.09374213218688965,
      "learning_rate": 9.904262717195126e-06,
      "loss": 0.0036,
      "step": 28630
    },
    {
      "epoch": 0.46862472388120757,
      "grad_norm": 0.05167952552437782,
      "learning_rate": 9.904077236208516e-06,
      "loss": 0.009,
      "step": 28640
    },
    {
      "epoch": 0.4687883498322834,
      "grad_norm": 0.20160222053527832,
      "learning_rate": 9.903891577460894e-06,
      "loss": 0.0059,
      "step": 28650
    },
    {
      "epoch": 0.46895197578335923,
      "grad_norm": 0.3289794921875,
      "learning_rate": 9.903705740958986e-06,
      "loss": 0.0063,
      "step": 28660
    },
    {
      "epoch": 0.46911560173443506,
      "grad_norm": 0.06347501277923584,
      "learning_rate": 9.903519726709533e-06,
      "loss": 0.0063,
      "step": 28670
    },
    {
      "epoch": 0.46927922768551095,
      "grad_norm": 0.3217048943042755,
      "learning_rate": 9.903333534719275e-06,
      "loss": 0.0025,
      "step": 28680
    },
    {
      "epoch": 0.4694428536365868,
      "grad_norm": 0.23959380388259888,
      "learning_rate": 9.90314716499496e-06,
      "loss": 0.0042,
      "step": 28690
    },
    {
      "epoch": 0.4696064795876626,
      "grad_norm": 0.3354378938674927,
      "learning_rate": 9.902960617543345e-06,
      "loss": 0.0081,
      "step": 28700
    },
    {
      "epoch": 0.46977010553873844,
      "grad_norm": 0.1581249237060547,
      "learning_rate": 9.902773892371193e-06,
      "loss": 0.0057,
      "step": 28710
    },
    {
      "epoch": 0.46993373148981427,
      "grad_norm": 0.10706600546836853,
      "learning_rate": 9.902586989485269e-06,
      "loss": 0.0056,
      "step": 28720
    },
    {
      "epoch": 0.47009735744089015,
      "grad_norm": 0.12246666103601456,
      "learning_rate": 9.90239990889235e-06,
      "loss": 0.0048,
      "step": 28730
    },
    {
      "epoch": 0.470260983391966,
      "grad_norm": 0.13429774343967438,
      "learning_rate": 9.902212650599218e-06,
      "loss": 0.0025,
      "step": 28740
    },
    {
      "epoch": 0.4704246093430418,
      "grad_norm": 0.2532491087913513,
      "learning_rate": 9.902025214612658e-06,
      "loss": 0.0073,
      "step": 28750
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.14271461963653564,
      "learning_rate": 9.901837600939468e-06,
      "loss": 0.004,
      "step": 28760
    },
    {
      "epoch": 0.47075186124519347,
      "grad_norm": 0.11834526062011719,
      "learning_rate": 9.901649809586445e-06,
      "loss": 0.0033,
      "step": 28770
    },
    {
      "epoch": 0.47091548719626936,
      "grad_norm": 0.390371710062027,
      "learning_rate": 9.901461840560396e-06,
      "loss": 0.0046,
      "step": 28780
    },
    {
      "epoch": 0.4710791131473452,
      "grad_norm": 0.522347092628479,
      "learning_rate": 9.901273693868136e-06,
      "loss": 0.0058,
      "step": 28790
    },
    {
      "epoch": 0.471242739098421,
      "grad_norm": 0.3997916281223297,
      "learning_rate": 9.901085369516485e-06,
      "loss": 0.0067,
      "step": 28800
    },
    {
      "epoch": 0.47140636504949684,
      "grad_norm": 0.2708657681941986,
      "learning_rate": 9.90089686751227e-06,
      "loss": 0.0042,
      "step": 28810
    },
    {
      "epoch": 0.4715699910005727,
      "grad_norm": 0.6393135786056519,
      "learning_rate": 9.900708187862321e-06,
      "loss": 0.0062,
      "step": 28820
    },
    {
      "epoch": 0.4717336169516485,
      "grad_norm": 0.17552627623081207,
      "learning_rate": 9.90051933057348e-06,
      "loss": 0.0048,
      "step": 28830
    },
    {
      "epoch": 0.4718972429027244,
      "grad_norm": 0.12909357249736786,
      "learning_rate": 9.900330295652589e-06,
      "loss": 0.0079,
      "step": 28840
    },
    {
      "epoch": 0.4720608688538002,
      "grad_norm": 0.2059362530708313,
      "learning_rate": 9.900141083106505e-06,
      "loss": 0.0034,
      "step": 28850
    },
    {
      "epoch": 0.47222449480487605,
      "grad_norm": 0.43750259280204773,
      "learning_rate": 9.899951692942082e-06,
      "loss": 0.005,
      "step": 28860
    },
    {
      "epoch": 0.4723881207559519,
      "grad_norm": 0.17548465728759766,
      "learning_rate": 9.89976212516619e-06,
      "loss": 0.0042,
      "step": 28870
    },
    {
      "epoch": 0.4725517467070277,
      "grad_norm": 0.06503050774335861,
      "learning_rate": 9.899572379785693e-06,
      "loss": 0.0074,
      "step": 28880
    },
    {
      "epoch": 0.4727153726581036,
      "grad_norm": 0.1215314120054245,
      "learning_rate": 9.899382456807475e-06,
      "loss": 0.0046,
      "step": 28890
    },
    {
      "epoch": 0.4728789986091794,
      "grad_norm": 0.15483234822750092,
      "learning_rate": 9.899192356238418e-06,
      "loss": 0.0043,
      "step": 28900
    },
    {
      "epoch": 0.47304262456025525,
      "grad_norm": 0.6191183924674988,
      "learning_rate": 9.899002078085415e-06,
      "loss": 0.0038,
      "step": 28910
    },
    {
      "epoch": 0.4732062505113311,
      "grad_norm": 0.5984652638435364,
      "learning_rate": 9.898811622355358e-06,
      "loss": 0.009,
      "step": 28920
    },
    {
      "epoch": 0.4733698764624069,
      "grad_norm": 0.1429656744003296,
      "learning_rate": 9.898620989055156e-06,
      "loss": 0.0047,
      "step": 28930
    },
    {
      "epoch": 0.4735335024134828,
      "grad_norm": 0.17204104363918304,
      "learning_rate": 9.898430178191716e-06,
      "loss": 0.0072,
      "step": 28940
    },
    {
      "epoch": 0.47369712836455863,
      "grad_norm": 0.2196149230003357,
      "learning_rate": 9.898239189771955e-06,
      "loss": 0.0074,
      "step": 28950
    },
    {
      "epoch": 0.47386075431563446,
      "grad_norm": 0.1460697203874588,
      "learning_rate": 9.898048023802795e-06,
      "loss": 0.0057,
      "step": 28960
    },
    {
      "epoch": 0.4740243802667103,
      "grad_norm": 0.07840320467948914,
      "learning_rate": 9.897856680291168e-06,
      "loss": 0.0027,
      "step": 28970
    },
    {
      "epoch": 0.4741880062177861,
      "grad_norm": 0.39064085483551025,
      "learning_rate": 9.897665159244008e-06,
      "loss": 0.0041,
      "step": 28980
    },
    {
      "epoch": 0.474351632168862,
      "grad_norm": 0.3270370364189148,
      "learning_rate": 9.897473460668256e-06,
      "loss": 0.0065,
      "step": 28990
    },
    {
      "epoch": 0.47451525811993783,
      "grad_norm": 0.1632394641637802,
      "learning_rate": 9.897281584570861e-06,
      "loss": 0.0057,
      "step": 29000
    },
    {
      "epoch": 0.47467888407101366,
      "grad_norm": 0.2010754495859146,
      "learning_rate": 9.897089530958782e-06,
      "loss": 0.0045,
      "step": 29010
    },
    {
      "epoch": 0.4748425100220895,
      "grad_norm": 0.28294432163238525,
      "learning_rate": 9.896897299838974e-06,
      "loss": 0.004,
      "step": 29020
    },
    {
      "epoch": 0.4750061359731653,
      "grad_norm": 0.11046252399682999,
      "learning_rate": 9.89670489121841e-06,
      "loss": 0.0055,
      "step": 29030
    },
    {
      "epoch": 0.4751697619242412,
      "grad_norm": 0.29268643260002136,
      "learning_rate": 9.896512305104063e-06,
      "loss": 0.0039,
      "step": 29040
    },
    {
      "epoch": 0.47533338787531704,
      "grad_norm": 0.22420576214790344,
      "learning_rate": 9.896319541502912e-06,
      "loss": 0.0054,
      "step": 29050
    },
    {
      "epoch": 0.47549701382639287,
      "grad_norm": 0.35233908891677856,
      "learning_rate": 9.896126600421945e-06,
      "loss": 0.0058,
      "step": 29060
    },
    {
      "epoch": 0.4756606397774687,
      "grad_norm": 0.2954990565776825,
      "learning_rate": 9.895933481868158e-06,
      "loss": 0.0043,
      "step": 29070
    },
    {
      "epoch": 0.4758242657285445,
      "grad_norm": 0.09990601241588593,
      "learning_rate": 9.89574018584855e-06,
      "loss": 0.0035,
      "step": 29080
    },
    {
      "epoch": 0.4759878916796204,
      "grad_norm": 0.42652109265327454,
      "learning_rate": 9.895546712370126e-06,
      "loss": 0.0068,
      "step": 29090
    },
    {
      "epoch": 0.47615151763069624,
      "grad_norm": 0.18180683255195618,
      "learning_rate": 9.895353061439899e-06,
      "loss": 0.0033,
      "step": 29100
    },
    {
      "epoch": 0.47631514358177207,
      "grad_norm": 0.12242705374956131,
      "learning_rate": 9.89515923306489e-06,
      "loss": 0.0044,
      "step": 29110
    },
    {
      "epoch": 0.4764787695328479,
      "grad_norm": 0.16486947238445282,
      "learning_rate": 9.894965227252123e-06,
      "loss": 0.005,
      "step": 29120
    },
    {
      "epoch": 0.47664239548392373,
      "grad_norm": 0.1478731632232666,
      "learning_rate": 9.894771044008633e-06,
      "loss": 0.0049,
      "step": 29130
    },
    {
      "epoch": 0.4768060214349996,
      "grad_norm": 0.25755301117897034,
      "learning_rate": 9.894576683341455e-06,
      "loss": 0.0063,
      "step": 29140
    },
    {
      "epoch": 0.47696964738607545,
      "grad_norm": 0.2787677049636841,
      "learning_rate": 9.894382145257637e-06,
      "loss": 0.0047,
      "step": 29150
    },
    {
      "epoch": 0.4771332733371513,
      "grad_norm": 0.197701096534729,
      "learning_rate": 9.89418742976423e-06,
      "loss": 0.0038,
      "step": 29160
    },
    {
      "epoch": 0.4772968992882271,
      "grad_norm": 0.09459872543811798,
      "learning_rate": 9.893992536868291e-06,
      "loss": 0.0036,
      "step": 29170
    },
    {
      "epoch": 0.47746052523930294,
      "grad_norm": 0.1716059446334839,
      "learning_rate": 9.893797466576886e-06,
      "loss": 0.0055,
      "step": 29180
    },
    {
      "epoch": 0.4776241511903788,
      "grad_norm": 0.48710349202156067,
      "learning_rate": 9.893602218897083e-06,
      "loss": 0.0048,
      "step": 29190
    },
    {
      "epoch": 0.47778777714145465,
      "grad_norm": 0.18181033432483673,
      "learning_rate": 9.893406793835962e-06,
      "loss": 0.0083,
      "step": 29200
    },
    {
      "epoch": 0.4779514030925305,
      "grad_norm": 0.1216956079006195,
      "learning_rate": 9.893211191400605e-06,
      "loss": 0.0042,
      "step": 29210
    },
    {
      "epoch": 0.4781150290436063,
      "grad_norm": 0.24959330260753632,
      "learning_rate": 9.893015411598104e-06,
      "loss": 0.0047,
      "step": 29220
    },
    {
      "epoch": 0.47827865499468214,
      "grad_norm": 0.4371202886104584,
      "learning_rate": 9.892819454435555e-06,
      "loss": 0.0039,
      "step": 29230
    },
    {
      "epoch": 0.47844228094575797,
      "grad_norm": 0.2708517909049988,
      "learning_rate": 9.892623319920059e-06,
      "loss": 0.0055,
      "step": 29240
    },
    {
      "epoch": 0.47860590689683385,
      "grad_norm": 0.07128234207630157,
      "learning_rate": 9.892427008058727e-06,
      "loss": 0.004,
      "step": 29250
    },
    {
      "epoch": 0.4787695328479097,
      "grad_norm": 0.11701580137014389,
      "learning_rate": 9.892230518858674e-06,
      "loss": 0.0066,
      "step": 29260
    },
    {
      "epoch": 0.4789331587989855,
      "grad_norm": 0.1885947585105896,
      "learning_rate": 9.892033852327023e-06,
      "loss": 0.0047,
      "step": 29270
    },
    {
      "epoch": 0.47909678475006134,
      "grad_norm": 0.4105403423309326,
      "learning_rate": 9.891837008470902e-06,
      "loss": 0.0064,
      "step": 29280
    },
    {
      "epoch": 0.4792604107011372,
      "grad_norm": 0.1233682855963707,
      "learning_rate": 9.891639987297448e-06,
      "loss": 0.0034,
      "step": 29290
    },
    {
      "epoch": 0.47942403665221306,
      "grad_norm": 0.3517018258571625,
      "learning_rate": 9.891442788813801e-06,
      "loss": 0.0046,
      "step": 29300
    },
    {
      "epoch": 0.4795876626032889,
      "grad_norm": 0.2979389727115631,
      "learning_rate": 9.891245413027108e-06,
      "loss": 0.0033,
      "step": 29310
    },
    {
      "epoch": 0.4797512885543647,
      "grad_norm": 0.2267286777496338,
      "learning_rate": 9.891047859944527e-06,
      "loss": 0.0034,
      "step": 29320
    },
    {
      "epoch": 0.47991491450544055,
      "grad_norm": 0.1396339237689972,
      "learning_rate": 9.890850129573215e-06,
      "loss": 0.006,
      "step": 29330
    },
    {
      "epoch": 0.4800785404565164,
      "grad_norm": 0.26734721660614014,
      "learning_rate": 9.89065222192034e-06,
      "loss": 0.0044,
      "step": 29340
    },
    {
      "epoch": 0.48024216640759226,
      "grad_norm": 0.06409502774477005,
      "learning_rate": 9.890454136993077e-06,
      "loss": 0.0066,
      "step": 29350
    },
    {
      "epoch": 0.4804057923586681,
      "grad_norm": 0.03526555746793747,
      "learning_rate": 9.890255874798605e-06,
      "loss": 0.0045,
      "step": 29360
    },
    {
      "epoch": 0.4805694183097439,
      "grad_norm": 0.1905120462179184,
      "learning_rate": 9.89005743534411e-06,
      "loss": 0.0051,
      "step": 29370
    },
    {
      "epoch": 0.48073304426081975,
      "grad_norm": 0.18688473105430603,
      "learning_rate": 9.88985881863679e-06,
      "loss": 0.0045,
      "step": 29380
    },
    {
      "epoch": 0.4808966702118956,
      "grad_norm": 0.3594113290309906,
      "learning_rate": 9.889660024683836e-06,
      "loss": 0.0041,
      "step": 29390
    },
    {
      "epoch": 0.48106029616297147,
      "grad_norm": 0.250940203666687,
      "learning_rate": 9.889461053492459e-06,
      "loss": 0.004,
      "step": 29400
    },
    {
      "epoch": 0.4812239221140473,
      "grad_norm": 0.060092538595199585,
      "learning_rate": 9.889261905069873e-06,
      "loss": 0.0032,
      "step": 29410
    },
    {
      "epoch": 0.4813875480651231,
      "grad_norm": 0.3339327871799469,
      "learning_rate": 9.889062579423292e-06,
      "loss": 0.0049,
      "step": 29420
    },
    {
      "epoch": 0.48155117401619896,
      "grad_norm": 0.1326649785041809,
      "learning_rate": 9.888863076559942e-06,
      "loss": 0.0041,
      "step": 29430
    },
    {
      "epoch": 0.4817147999672748,
      "grad_norm": 0.14469024538993835,
      "learning_rate": 9.888663396487057e-06,
      "loss": 0.004,
      "step": 29440
    },
    {
      "epoch": 0.4818784259183507,
      "grad_norm": 0.17382074892520905,
      "learning_rate": 9.888463539211873e-06,
      "loss": 0.0032,
      "step": 29450
    },
    {
      "epoch": 0.4820420518694265,
      "grad_norm": 0.049370840191841125,
      "learning_rate": 9.888263504741635e-06,
      "loss": 0.0045,
      "step": 29460
    },
    {
      "epoch": 0.48220567782050233,
      "grad_norm": 0.17935901880264282,
      "learning_rate": 9.888063293083593e-06,
      "loss": 0.0058,
      "step": 29470
    },
    {
      "epoch": 0.48236930377157816,
      "grad_norm": 0.13717053830623627,
      "learning_rate": 9.887862904245005e-06,
      "loss": 0.0062,
      "step": 29480
    },
    {
      "epoch": 0.482532929722654,
      "grad_norm": 0.25411278009414673,
      "learning_rate": 9.887662338233135e-06,
      "loss": 0.003,
      "step": 29490
    },
    {
      "epoch": 0.4826965556737299,
      "grad_norm": 0.41999804973602295,
      "learning_rate": 9.887461595055251e-06,
      "loss": 0.0056,
      "step": 29500
    },
    {
      "epoch": 0.4828601816248057,
      "grad_norm": 0.32119449973106384,
      "learning_rate": 9.887260674718633e-06,
      "loss": 0.0063,
      "step": 29510
    },
    {
      "epoch": 0.48302380757588154,
      "grad_norm": 0.5428754687309265,
      "learning_rate": 9.88705957723056e-06,
      "loss": 0.0059,
      "step": 29520
    },
    {
      "epoch": 0.48318743352695737,
      "grad_norm": 0.11517683416604996,
      "learning_rate": 9.886858302598323e-06,
      "loss": 0.0043,
      "step": 29530
    },
    {
      "epoch": 0.4833510594780332,
      "grad_norm": 0.19034425914287567,
      "learning_rate": 9.88665685082922e-06,
      "loss": 0.0041,
      "step": 29540
    },
    {
      "epoch": 0.4835146854291091,
      "grad_norm": 0.17365509271621704,
      "learning_rate": 9.886455221930548e-06,
      "loss": 0.0035,
      "step": 29550
    },
    {
      "epoch": 0.4836783113801849,
      "grad_norm": 0.23726002871990204,
      "learning_rate": 9.886253415909622e-06,
      "loss": 0.0056,
      "step": 29560
    },
    {
      "epoch": 0.48384193733126074,
      "grad_norm": 0.1593717634677887,
      "learning_rate": 9.88605143277375e-06,
      "loss": 0.0047,
      "step": 29570
    },
    {
      "epoch": 0.48400556328233657,
      "grad_norm": 0.23736333847045898,
      "learning_rate": 9.885849272530257e-06,
      "loss": 0.0057,
      "step": 29580
    },
    {
      "epoch": 0.4841691892334124,
      "grad_norm": 0.19513937830924988,
      "learning_rate": 9.885646935186473e-06,
      "loss": 0.0052,
      "step": 29590
    },
    {
      "epoch": 0.4843328151844883,
      "grad_norm": 0.1746743768453598,
      "learning_rate": 9.88544442074973e-06,
      "loss": 0.008,
      "step": 29600
    },
    {
      "epoch": 0.4844964411355641,
      "grad_norm": 0.25018948316574097,
      "learning_rate": 9.885241729227366e-06,
      "loss": 0.0032,
      "step": 29610
    },
    {
      "epoch": 0.48466006708663995,
      "grad_norm": 0.036033306270837784,
      "learning_rate": 9.885038860626732e-06,
      "loss": 0.0038,
      "step": 29620
    },
    {
      "epoch": 0.4848236930377158,
      "grad_norm": 0.3072309195995331,
      "learning_rate": 9.88483581495518e-06,
      "loss": 0.0049,
      "step": 29630
    },
    {
      "epoch": 0.4849873189887916,
      "grad_norm": 0.09974529594182968,
      "learning_rate": 9.884632592220071e-06,
      "loss": 0.005,
      "step": 29640
    },
    {
      "epoch": 0.4851509449398675,
      "grad_norm": 0.25555264949798584,
      "learning_rate": 9.884429192428771e-06,
      "loss": 0.0039,
      "step": 29650
    },
    {
      "epoch": 0.4853145708909433,
      "grad_norm": 0.4265265166759491,
      "learning_rate": 9.884225615588651e-06,
      "loss": 0.0042,
      "step": 29660
    },
    {
      "epoch": 0.48547819684201915,
      "grad_norm": 0.2798185646533966,
      "learning_rate": 9.884021861707091e-06,
      "loss": 0.0046,
      "step": 29670
    },
    {
      "epoch": 0.485641822793095,
      "grad_norm": 0.2792820930480957,
      "learning_rate": 9.883817930791477e-06,
      "loss": 0.0048,
      "step": 29680
    },
    {
      "epoch": 0.4858054487441708,
      "grad_norm": 0.2896956205368042,
      "learning_rate": 9.883613822849201e-06,
      "loss": 0.0069,
      "step": 29690
    },
    {
      "epoch": 0.48596907469524664,
      "grad_norm": 0.1544959545135498,
      "learning_rate": 9.883409537887662e-06,
      "loss": 0.0036,
      "step": 29700
    },
    {
      "epoch": 0.4861327006463225,
      "grad_norm": 0.2987891137599945,
      "learning_rate": 9.883205075914263e-06,
      "loss": 0.0056,
      "step": 29710
    },
    {
      "epoch": 0.48629632659739835,
      "grad_norm": 0.16056248545646667,
      "learning_rate": 9.883000436936418e-06,
      "loss": 0.006,
      "step": 29720
    },
    {
      "epoch": 0.4864599525484742,
      "grad_norm": 0.24921059608459473,
      "learning_rate": 9.882795620961544e-06,
      "loss": 0.0044,
      "step": 29730
    },
    {
      "epoch": 0.48662357849955,
      "grad_norm": 0.3647303283214569,
      "learning_rate": 9.882590627997061e-06,
      "loss": 0.0058,
      "step": 29740
    },
    {
      "epoch": 0.48678720445062584,
      "grad_norm": 0.2851385176181793,
      "learning_rate": 9.882385458050406e-06,
      "loss": 0.0056,
      "step": 29750
    },
    {
      "epoch": 0.48695083040170173,
      "grad_norm": 0.15056554973125458,
      "learning_rate": 9.882180111129012e-06,
      "loss": 0.0038,
      "step": 29760
    },
    {
      "epoch": 0.48711445635277756,
      "grad_norm": 0.17657239735126495,
      "learning_rate": 9.881974587240321e-06,
      "loss": 0.0045,
      "step": 29770
    },
    {
      "epoch": 0.4872780823038534,
      "grad_norm": 0.3117468059062958,
      "learning_rate": 9.881768886391787e-06,
      "loss": 0.008,
      "step": 29780
    },
    {
      "epoch": 0.4874417082549292,
      "grad_norm": 0.6961111426353455,
      "learning_rate": 9.881563008590861e-06,
      "loss": 0.003,
      "step": 29790
    },
    {
      "epoch": 0.48760533420600505,
      "grad_norm": 0.23479531705379486,
      "learning_rate": 9.881356953845011e-06,
      "loss": 0.0079,
      "step": 29800
    },
    {
      "epoch": 0.48776896015708093,
      "grad_norm": 0.20981964468955994,
      "learning_rate": 9.881150722161703e-06,
      "loss": 0.0039,
      "step": 29810
    },
    {
      "epoch": 0.48793258610815676,
      "grad_norm": 0.04459293186664581,
      "learning_rate": 9.880944313548413e-06,
      "loss": 0.0033,
      "step": 29820
    },
    {
      "epoch": 0.4880962120592326,
      "grad_norm": 0.05605221912264824,
      "learning_rate": 9.880737728012622e-06,
      "loss": 0.0032,
      "step": 29830
    },
    {
      "epoch": 0.4882598380103084,
      "grad_norm": 0.24640503525733948,
      "learning_rate": 9.880530965561818e-06,
      "loss": 0.0058,
      "step": 29840
    },
    {
      "epoch": 0.48842346396138425,
      "grad_norm": 0.3623533248901367,
      "learning_rate": 9.880324026203498e-06,
      "loss": 0.0045,
      "step": 29850
    },
    {
      "epoch": 0.48858708991246014,
      "grad_norm": 0.2251889407634735,
      "learning_rate": 9.880116909945162e-06,
      "loss": 0.0051,
      "step": 29860
    },
    {
      "epoch": 0.48875071586353597,
      "grad_norm": 0.29930582642555237,
      "learning_rate": 9.879909616794316e-06,
      "loss": 0.0053,
      "step": 29870
    },
    {
      "epoch": 0.4889143418146118,
      "grad_norm": 0.14795860648155212,
      "learning_rate": 9.879702146758477e-06,
      "loss": 0.0054,
      "step": 29880
    },
    {
      "epoch": 0.4890779677656876,
      "grad_norm": 0.14577575027942657,
      "learning_rate": 9.87949449984516e-06,
      "loss": 0.0042,
      "step": 29890
    },
    {
      "epoch": 0.48924159371676346,
      "grad_norm": 0.2441176474094391,
      "learning_rate": 9.879286676061897e-06,
      "loss": 0.0032,
      "step": 29900
    },
    {
      "epoch": 0.48940521966783934,
      "grad_norm": 0.26571568846702576,
      "learning_rate": 9.879078675416218e-06,
      "loss": 0.0041,
      "step": 29910
    },
    {
      "epoch": 0.48956884561891517,
      "grad_norm": 0.33371320366859436,
      "learning_rate": 9.878870497915664e-06,
      "loss": 0.0039,
      "step": 29920
    },
    {
      "epoch": 0.489732471569991,
      "grad_norm": 0.47707465291023254,
      "learning_rate": 9.878662143567782e-06,
      "loss": 0.0034,
      "step": 29930
    },
    {
      "epoch": 0.48989609752106683,
      "grad_norm": 0.288194477558136,
      "learning_rate": 9.878453612380122e-06,
      "loss": 0.003,
      "step": 29940
    },
    {
      "epoch": 0.49005972347214266,
      "grad_norm": 0.15214216709136963,
      "learning_rate": 9.87824490436024e-06,
      "loss": 0.0043,
      "step": 29950
    },
    {
      "epoch": 0.49022334942321855,
      "grad_norm": 0.3886477053165436,
      "learning_rate": 9.878036019515708e-06,
      "loss": 0.0031,
      "step": 29960
    },
    {
      "epoch": 0.4903869753742944,
      "grad_norm": 0.12571264803409576,
      "learning_rate": 9.877826957854095e-06,
      "loss": 0.0045,
      "step": 29970
    },
    {
      "epoch": 0.4905506013253702,
      "grad_norm": 0.2895621955394745,
      "learning_rate": 9.877617719382978e-06,
      "loss": 0.0047,
      "step": 29980
    },
    {
      "epoch": 0.49071422727644604,
      "grad_norm": 0.17046798765659332,
      "learning_rate": 9.87740830410994e-06,
      "loss": 0.0053,
      "step": 29990
    },
    {
      "epoch": 0.49087785322752187,
      "grad_norm": 0.37388870120048523,
      "learning_rate": 9.877198712042576e-06,
      "loss": 0.0026,
      "step": 30000
    },
    {
      "epoch": 0.49104147917859775,
      "grad_norm": 0.40467822551727295,
      "learning_rate": 9.876988943188479e-06,
      "loss": 0.0035,
      "step": 30010
    },
    {
      "epoch": 0.4912051051296736,
      "grad_norm": 0.5484909415245056,
      "learning_rate": 9.876778997555255e-06,
      "loss": 0.0055,
      "step": 30020
    },
    {
      "epoch": 0.4913687310807494,
      "grad_norm": 0.12758401036262512,
      "learning_rate": 9.876568875150515e-06,
      "loss": 0.005,
      "step": 30030
    },
    {
      "epoch": 0.49153235703182524,
      "grad_norm": 0.21156053245067596,
      "learning_rate": 9.87635857598187e-06,
      "loss": 0.0052,
      "step": 30040
    },
    {
      "epoch": 0.49169598298290107,
      "grad_norm": 0.16070495545864105,
      "learning_rate": 9.87614810005695e-06,
      "loss": 0.003,
      "step": 30050
    },
    {
      "epoch": 0.49185960893397696,
      "grad_norm": 0.13787518441677094,
      "learning_rate": 9.87593744738338e-06,
      "loss": 0.0076,
      "step": 30060
    },
    {
      "epoch": 0.4920232348850528,
      "grad_norm": 0.2332271784543991,
      "learning_rate": 9.875726617968794e-06,
      "loss": 0.0048,
      "step": 30070
    },
    {
      "epoch": 0.4921868608361286,
      "grad_norm": 0.3964560329914093,
      "learning_rate": 9.875515611820839e-06,
      "loss": 0.0047,
      "step": 30080
    },
    {
      "epoch": 0.49235048678720444,
      "grad_norm": 0.15190444886684418,
      "learning_rate": 9.87530442894716e-06,
      "loss": 0.0046,
      "step": 30090
    },
    {
      "epoch": 0.4925141127382803,
      "grad_norm": 0.04631827399134636,
      "learning_rate": 9.875093069355414e-06,
      "loss": 0.0028,
      "step": 30100
    },
    {
      "epoch": 0.49267773868935616,
      "grad_norm": 0.23227931559085846,
      "learning_rate": 9.874881533053262e-06,
      "loss": 0.0051,
      "step": 30110
    },
    {
      "epoch": 0.492841364640432,
      "grad_norm": 0.20683607459068298,
      "learning_rate": 9.874669820048369e-06,
      "loss": 0.0041,
      "step": 30120
    },
    {
      "epoch": 0.4930049905915078,
      "grad_norm": 0.18111008405685425,
      "learning_rate": 9.874457930348411e-06,
      "loss": 0.0039,
      "step": 30130
    },
    {
      "epoch": 0.49316861654258365,
      "grad_norm": 0.07248686254024506,
      "learning_rate": 9.874245863961071e-06,
      "loss": 0.004,
      "step": 30140
    },
    {
      "epoch": 0.4933322424936595,
      "grad_norm": 0.39866340160369873,
      "learning_rate": 9.874033620894032e-06,
      "loss": 0.0059,
      "step": 30150
    },
    {
      "epoch": 0.4934958684447353,
      "grad_norm": 0.3728131949901581,
      "learning_rate": 9.873821201154988e-06,
      "loss": 0.0052,
      "step": 30160
    },
    {
      "epoch": 0.4936594943958112,
      "grad_norm": 0.22588664293289185,
      "learning_rate": 9.87360860475164e-06,
      "loss": 0.0042,
      "step": 30170
    },
    {
      "epoch": 0.493823120346887,
      "grad_norm": 0.2180001437664032,
      "learning_rate": 9.873395831691692e-06,
      "loss": 0.0046,
      "step": 30180
    },
    {
      "epoch": 0.49398674629796285,
      "grad_norm": 0.15236681699752808,
      "learning_rate": 9.87318288198286e-06,
      "loss": 0.0045,
      "step": 30190
    },
    {
      "epoch": 0.4941503722490387,
      "grad_norm": 0.11469181627035141,
      "learning_rate": 9.87296975563286e-06,
      "loss": 0.0047,
      "step": 30200
    },
    {
      "epoch": 0.4943139982001145,
      "grad_norm": 0.360554963350296,
      "learning_rate": 9.872756452649417e-06,
      "loss": 0.0039,
      "step": 30210
    },
    {
      "epoch": 0.4944776241511904,
      "grad_norm": 0.30228957533836365,
      "learning_rate": 9.872542973040266e-06,
      "loss": 0.0061,
      "step": 30220
    },
    {
      "epoch": 0.49464125010226623,
      "grad_norm": 0.16423900425434113,
      "learning_rate": 9.872329316813143e-06,
      "loss": 0.0049,
      "step": 30230
    },
    {
      "epoch": 0.49480487605334206,
      "grad_norm": 0.11911007761955261,
      "learning_rate": 9.87211548397579e-06,
      "loss": 0.0042,
      "step": 30240
    },
    {
      "epoch": 0.4949685020044179,
      "grad_norm": 0.20877450704574585,
      "learning_rate": 9.871901474535964e-06,
      "loss": 0.0038,
      "step": 30250
    },
    {
      "epoch": 0.4951321279554937,
      "grad_norm": 0.09507668018341064,
      "learning_rate": 9.871687288501418e-06,
      "loss": 0.0048,
      "step": 30260
    },
    {
      "epoch": 0.4952957539065696,
      "grad_norm": 0.1973990499973297,
      "learning_rate": 9.871472925879914e-06,
      "loss": 0.0039,
      "step": 30270
    },
    {
      "epoch": 0.49545937985764543,
      "grad_norm": 0.2648104131221771,
      "learning_rate": 9.871258386679228e-06,
      "loss": 0.0031,
      "step": 30280
    },
    {
      "epoch": 0.49562300580872126,
      "grad_norm": 0.18706491589546204,
      "learning_rate": 9.871043670907132e-06,
      "loss": 0.0042,
      "step": 30290
    },
    {
      "epoch": 0.4957866317597971,
      "grad_norm": 0.49145370721817017,
      "learning_rate": 9.870828778571408e-06,
      "loss": 0.0033,
      "step": 30300
    },
    {
      "epoch": 0.4959502577108729,
      "grad_norm": 0.25051698088645935,
      "learning_rate": 9.87061370967985e-06,
      "loss": 0.0046,
      "step": 30310
    },
    {
      "epoch": 0.4961138836619488,
      "grad_norm": 0.12890347838401794,
      "learning_rate": 9.870398464240251e-06,
      "loss": 0.0042,
      "step": 30320
    },
    {
      "epoch": 0.49627750961302464,
      "grad_norm": 0.2778070867061615,
      "learning_rate": 9.870183042260414e-06,
      "loss": 0.0037,
      "step": 30330
    },
    {
      "epoch": 0.49644113556410047,
      "grad_norm": 0.33920711278915405,
      "learning_rate": 9.869967443748145e-06,
      "loss": 0.0052,
      "step": 30340
    },
    {
      "epoch": 0.4966047615151763,
      "grad_norm": 0.23967944085597992,
      "learning_rate": 9.869751668711262e-06,
      "loss": 0.0055,
      "step": 30350
    },
    {
      "epoch": 0.4967683874662521,
      "grad_norm": 0.0423504002392292,
      "learning_rate": 9.869535717157585e-06,
      "loss": 0.005,
      "step": 30360
    },
    {
      "epoch": 0.496932013417328,
      "grad_norm": 0.1894690841436386,
      "learning_rate": 9.869319589094943e-06,
      "loss": 0.0034,
      "step": 30370
    },
    {
      "epoch": 0.49709563936840384,
      "grad_norm": 0.22895008325576782,
      "learning_rate": 9.869103284531167e-06,
      "loss": 0.0045,
      "step": 30380
    },
    {
      "epoch": 0.49725926531947967,
      "grad_norm": 0.3898073732852936,
      "learning_rate": 9.8688868034741e-06,
      "loss": 0.0076,
      "step": 30390
    },
    {
      "epoch": 0.4974228912705555,
      "grad_norm": 0.7379021644592285,
      "learning_rate": 9.868670145931589e-06,
      "loss": 0.0056,
      "step": 30400
    },
    {
      "epoch": 0.49758651722163133,
      "grad_norm": 0.20226426422595978,
      "learning_rate": 9.868453311911487e-06,
      "loss": 0.0059,
      "step": 30410
    },
    {
      "epoch": 0.4977501431727072,
      "grad_norm": 0.2828306257724762,
      "learning_rate": 9.868236301421653e-06,
      "loss": 0.007,
      "step": 30420
    },
    {
      "epoch": 0.49791376912378305,
      "grad_norm": 0.14391179382801056,
      "learning_rate": 9.868019114469953e-06,
      "loss": 0.0041,
      "step": 30430
    },
    {
      "epoch": 0.4980773950748589,
      "grad_norm": 0.318920373916626,
      "learning_rate": 9.867801751064262e-06,
      "loss": 0.0026,
      "step": 30440
    },
    {
      "epoch": 0.4982410210259347,
      "grad_norm": 0.1021595448255539,
      "learning_rate": 9.867584211212456e-06,
      "loss": 0.006,
      "step": 30450
    },
    {
      "epoch": 0.49840464697701053,
      "grad_norm": 0.2872607707977295,
      "learning_rate": 9.86736649492242e-06,
      "loss": 0.0043,
      "step": 30460
    },
    {
      "epoch": 0.4985682729280864,
      "grad_norm": 0.08095104992389679,
      "learning_rate": 9.867148602202049e-06,
      "loss": 0.0044,
      "step": 30470
    },
    {
      "epoch": 0.49873189887916225,
      "grad_norm": 0.15095247328281403,
      "learning_rate": 9.866930533059237e-06,
      "loss": 0.0044,
      "step": 30480
    },
    {
      "epoch": 0.4988955248302381,
      "grad_norm": 0.08777711540460587,
      "learning_rate": 9.866712287501891e-06,
      "loss": 0.003,
      "step": 30490
    },
    {
      "epoch": 0.4990591507813139,
      "grad_norm": 0.10970611125230789,
      "learning_rate": 9.866493865537922e-06,
      "loss": 0.0061,
      "step": 30500
    },
    {
      "epoch": 0.49922277673238974,
      "grad_norm": 0.18514953553676605,
      "learning_rate": 9.866275267175248e-06,
      "loss": 0.0054,
      "step": 30510
    },
    {
      "epoch": 0.4993864026834656,
      "grad_norm": 0.1681094616651535,
      "learning_rate": 9.866056492421789e-06,
      "loss": 0.0049,
      "step": 30520
    },
    {
      "epoch": 0.49955002863454145,
      "grad_norm": 0.32745957374572754,
      "learning_rate": 9.865837541285479e-06,
      "loss": 0.0032,
      "step": 30530
    },
    {
      "epoch": 0.4997136545856173,
      "grad_norm": 0.3193792700767517,
      "learning_rate": 9.865618413774251e-06,
      "loss": 0.0051,
      "step": 30540
    },
    {
      "epoch": 0.4998772805366931,
      "grad_norm": 0.2464493066072464,
      "learning_rate": 9.865399109896051e-06,
      "loss": 0.0054,
      "step": 30550
    },
    {
      "epoch": 0.500040906487769,
      "grad_norm": 0.2660923898220062,
      "learning_rate": 9.865179629658827e-06,
      "loss": 0.0045,
      "step": 30560
    },
    {
      "epoch": 0.5002045324388448,
      "grad_norm": 0.1223653182387352,
      "learning_rate": 9.864959973070534e-06,
      "loss": 0.0059,
      "step": 30570
    },
    {
      "epoch": 0.5003681583899207,
      "grad_norm": 0.3126308023929596,
      "learning_rate": 9.864740140139135e-06,
      "loss": 0.0044,
      "step": 30580
    },
    {
      "epoch": 0.5005317843409964,
      "grad_norm": 0.11172177642583847,
      "learning_rate": 9.864520130872599e-06,
      "loss": 0.0044,
      "step": 30590
    },
    {
      "epoch": 0.5006954102920723,
      "grad_norm": 0.17950640618801117,
      "learning_rate": 9.8642999452789e-06,
      "loss": 0.0036,
      "step": 30600
    },
    {
      "epoch": 0.5008590362431482,
      "grad_norm": 0.22110264003276825,
      "learning_rate": 9.864079583366016e-06,
      "loss": 0.0059,
      "step": 30610
    },
    {
      "epoch": 0.501022662194224,
      "grad_norm": 0.43560776114463806,
      "learning_rate": 9.86385904514194e-06,
      "loss": 0.0051,
      "step": 30620
    },
    {
      "epoch": 0.5011862881452999,
      "grad_norm": 0.43948209285736084,
      "learning_rate": 9.863638330614663e-06,
      "loss": 0.0041,
      "step": 30630
    },
    {
      "epoch": 0.5013499140963756,
      "grad_norm": 0.09683792293071747,
      "learning_rate": 9.863417439792187e-06,
      "loss": 0.0059,
      "step": 30640
    },
    {
      "epoch": 0.5015135400474515,
      "grad_norm": 0.0957476869225502,
      "learning_rate": 9.863196372682516e-06,
      "loss": 0.0049,
      "step": 30650
    },
    {
      "epoch": 0.5016771659985274,
      "grad_norm": 0.22459042072296143,
      "learning_rate": 9.862975129293666e-06,
      "loss": 0.0064,
      "step": 30660
    },
    {
      "epoch": 0.5018407919496032,
      "grad_norm": 0.049262214452028275,
      "learning_rate": 9.862753709633654e-06,
      "loss": 0.005,
      "step": 30670
    },
    {
      "epoch": 0.5020044179006791,
      "grad_norm": 0.13619364798069,
      "learning_rate": 9.862532113710509e-06,
      "loss": 0.0029,
      "step": 30680
    },
    {
      "epoch": 0.5021680438517548,
      "grad_norm": 0.15392081439495087,
      "learning_rate": 9.862310341532263e-06,
      "loss": 0.0043,
      "step": 30690
    },
    {
      "epoch": 0.5023316698028307,
      "grad_norm": 0.22392538189888,
      "learning_rate": 9.86208839310695e-06,
      "loss": 0.0039,
      "step": 30700
    },
    {
      "epoch": 0.5024952957539066,
      "grad_norm": 0.5785654783248901,
      "learning_rate": 9.861866268442618e-06,
      "loss": 0.0079,
      "step": 30710
    },
    {
      "epoch": 0.5026589217049824,
      "grad_norm": 0.04843432083725929,
      "learning_rate": 9.861643967547323e-06,
      "loss": 0.0059,
      "step": 30720
    },
    {
      "epoch": 0.5028225476560583,
      "grad_norm": 0.27586302161216736,
      "learning_rate": 9.861421490429116e-06,
      "loss": 0.0041,
      "step": 30730
    },
    {
      "epoch": 0.502986173607134,
      "grad_norm": 0.37647032737731934,
      "learning_rate": 9.861198837096064e-06,
      "loss": 0.0039,
      "step": 30740
    },
    {
      "epoch": 0.5031497995582099,
      "grad_norm": 0.17302639782428741,
      "learning_rate": 9.86097600755624e-06,
      "loss": 0.0046,
      "step": 30750
    },
    {
      "epoch": 0.5033134255092858,
      "grad_norm": 0.07171842455863953,
      "learning_rate": 9.860753001817716e-06,
      "loss": 0.0033,
      "step": 30760
    },
    {
      "epoch": 0.5034770514603616,
      "grad_norm": 0.24697132408618927,
      "learning_rate": 9.86052981988858e-06,
      "loss": 0.0049,
      "step": 30770
    },
    {
      "epoch": 0.5036406774114375,
      "grad_norm": 0.0713362991809845,
      "learning_rate": 9.86030646177692e-06,
      "loss": 0.0034,
      "step": 30780
    },
    {
      "epoch": 0.5038043033625133,
      "grad_norm": 0.12136947363615036,
      "learning_rate": 9.860082927490831e-06,
      "loss": 0.0044,
      "step": 30790
    },
    {
      "epoch": 0.5039679293135891,
      "grad_norm": 0.3462497591972351,
      "learning_rate": 9.859859217038418e-06,
      "loss": 0.0042,
      "step": 30800
    },
    {
      "epoch": 0.504131555264665,
      "grad_norm": 0.2680438458919525,
      "learning_rate": 9.859635330427789e-06,
      "loss": 0.0032,
      "step": 30810
    },
    {
      "epoch": 0.5042951812157408,
      "grad_norm": 0.22928646206855774,
      "learning_rate": 9.859411267667059e-06,
      "loss": 0.003,
      "step": 30820
    },
    {
      "epoch": 0.5044588071668167,
      "grad_norm": 0.10160088539123535,
      "learning_rate": 9.859187028764348e-06,
      "loss": 0.0051,
      "step": 30830
    },
    {
      "epoch": 0.5046224331178925,
      "grad_norm": 0.24085910618305206,
      "learning_rate": 9.858962613727787e-06,
      "loss": 0.004,
      "step": 30840
    },
    {
      "epoch": 0.5047860590689683,
      "grad_norm": 0.1074313223361969,
      "learning_rate": 9.85873802256551e-06,
      "loss": 0.0033,
      "step": 30850
    },
    {
      "epoch": 0.5049496850200442,
      "grad_norm": 0.10650540888309479,
      "learning_rate": 9.858513255285657e-06,
      "loss": 0.0054,
      "step": 30860
    },
    {
      "epoch": 0.50511331097112,
      "grad_norm": 0.41054514050483704,
      "learning_rate": 9.858288311896375e-06,
      "loss": 0.0051,
      "step": 30870
    },
    {
      "epoch": 0.5052769369221959,
      "grad_norm": 0.4507800340652466,
      "learning_rate": 9.85806319240582e-06,
      "loss": 0.0051,
      "step": 30880
    },
    {
      "epoch": 0.5054405628732717,
      "grad_norm": 0.20634064078330994,
      "learning_rate": 9.857837896822149e-06,
      "loss": 0.0037,
      "step": 30890
    },
    {
      "epoch": 0.5056041888243475,
      "grad_norm": 0.2859896719455719,
      "learning_rate": 9.857612425153529e-06,
      "loss": 0.0048,
      "step": 30900
    },
    {
      "epoch": 0.5057678147754234,
      "grad_norm": 0.23569360375404358,
      "learning_rate": 9.857386777408135e-06,
      "loss": 0.0041,
      "step": 30910
    },
    {
      "epoch": 0.5059314407264992,
      "grad_norm": 0.3860852122306824,
      "learning_rate": 9.857160953594144e-06,
      "loss": 0.0028,
      "step": 30920
    },
    {
      "epoch": 0.5060950666775751,
      "grad_norm": 0.20479625463485718,
      "learning_rate": 9.856934953719742e-06,
      "loss": 0.004,
      "step": 30930
    },
    {
      "epoch": 0.5062586926286509,
      "grad_norm": 0.11897959560155869,
      "learning_rate": 9.856708777793122e-06,
      "loss": 0.0026,
      "step": 30940
    },
    {
      "epoch": 0.5064223185797267,
      "grad_norm": 0.3292287588119507,
      "learning_rate": 9.856482425822482e-06,
      "loss": 0.0045,
      "step": 30950
    },
    {
      "epoch": 0.5065859445308026,
      "grad_norm": 0.3212360739707947,
      "learning_rate": 9.856255897816024e-06,
      "loss": 0.0036,
      "step": 30960
    },
    {
      "epoch": 0.5067495704818784,
      "grad_norm": 0.20862159132957458,
      "learning_rate": 9.856029193781964e-06,
      "loss": 0.0031,
      "step": 30970
    },
    {
      "epoch": 0.5069131964329543,
      "grad_norm": 0.2558629810810089,
      "learning_rate": 9.855802313728517e-06,
      "loss": 0.0044,
      "step": 30980
    },
    {
      "epoch": 0.5070768223840301,
      "grad_norm": 0.2657250165939331,
      "learning_rate": 9.855575257663905e-06,
      "loss": 0.0061,
      "step": 30990
    },
    {
      "epoch": 0.507240448335106,
      "grad_norm": 0.1743023544549942,
      "learning_rate": 9.855348025596362e-06,
      "loss": 0.0046,
      "step": 31000
    },
    {
      "epoch": 0.5074040742861818,
      "grad_norm": 0.25727203488349915,
      "learning_rate": 9.85512061753412e-06,
      "loss": 0.0032,
      "step": 31010
    },
    {
      "epoch": 0.5075677002372576,
      "grad_norm": 0.4770571291446686,
      "learning_rate": 9.854893033485427e-06,
      "loss": 0.0054,
      "step": 31020
    },
    {
      "epoch": 0.5077313261883335,
      "grad_norm": 0.3970147967338562,
      "learning_rate": 9.854665273458527e-06,
      "loss": 0.0046,
      "step": 31030
    },
    {
      "epoch": 0.5078949521394093,
      "grad_norm": 0.29395201802253723,
      "learning_rate": 9.854437337461682e-06,
      "loss": 0.0073,
      "step": 31040
    },
    {
      "epoch": 0.5080585780904852,
      "grad_norm": 0.33381879329681396,
      "learning_rate": 9.854209225503148e-06,
      "loss": 0.0035,
      "step": 31050
    },
    {
      "epoch": 0.508222204041561,
      "grad_norm": 0.3724513351917267,
      "learning_rate": 9.853980937591199e-06,
      "loss": 0.0063,
      "step": 31060
    },
    {
      "epoch": 0.5083858299926368,
      "grad_norm": 0.3101649880409241,
      "learning_rate": 9.853752473734107e-06,
      "loss": 0.0065,
      "step": 31070
    },
    {
      "epoch": 0.5085494559437127,
      "grad_norm": 0.1964121311903,
      "learning_rate": 9.853523833940152e-06,
      "loss": 0.0033,
      "step": 31080
    },
    {
      "epoch": 0.5087130818947885,
      "grad_norm": 0.1315566897392273,
      "learning_rate": 9.853295018217623e-06,
      "loss": 0.0045,
      "step": 31090
    },
    {
      "epoch": 0.5088767078458644,
      "grad_norm": 0.23415246605873108,
      "learning_rate": 9.853066026574816e-06,
      "loss": 0.0031,
      "step": 31100
    },
    {
      "epoch": 0.5090403337969402,
      "grad_norm": 0.3314189910888672,
      "learning_rate": 9.852836859020027e-06,
      "loss": 0.0058,
      "step": 31110
    },
    {
      "epoch": 0.509203959748016,
      "grad_norm": 0.051704198122024536,
      "learning_rate": 9.852607515561567e-06,
      "loss": 0.0069,
      "step": 31120
    },
    {
      "epoch": 0.5093675856990919,
      "grad_norm": 0.5094984173774719,
      "learning_rate": 9.852377996207748e-06,
      "loss": 0.0052,
      "step": 31130
    },
    {
      "epoch": 0.5095312116501677,
      "grad_norm": 0.19634106755256653,
      "learning_rate": 9.852148300966888e-06,
      "loss": 0.0051,
      "step": 31140
    },
    {
      "epoch": 0.5096948376012436,
      "grad_norm": 0.15406464040279388,
      "learning_rate": 9.851918429847315e-06,
      "loss": 0.0031,
      "step": 31150
    },
    {
      "epoch": 0.5098584635523195,
      "grad_norm": 0.504865288734436,
      "learning_rate": 9.851688382857359e-06,
      "loss": 0.004,
      "step": 31160
    },
    {
      "epoch": 0.5100220895033952,
      "grad_norm": 0.16970601677894592,
      "learning_rate": 9.851458160005361e-06,
      "loss": 0.0043,
      "step": 31170
    },
    {
      "epoch": 0.5101857154544711,
      "grad_norm": 0.26412877440452576,
      "learning_rate": 9.851227761299662e-06,
      "loss": 0.0028,
      "step": 31180
    },
    {
      "epoch": 0.5103493414055469,
      "grad_norm": 0.11685189604759216,
      "learning_rate": 9.85099718674862e-06,
      "loss": 0.0036,
      "step": 31190
    },
    {
      "epoch": 0.5105129673566228,
      "grad_norm": 0.34057921171188354,
      "learning_rate": 9.850766436360587e-06,
      "loss": 0.0055,
      "step": 31200
    },
    {
      "epoch": 0.5106765933076987,
      "grad_norm": 0.14303185045719147,
      "learning_rate": 9.85053551014393e-06,
      "loss": 0.0031,
      "step": 31210
    },
    {
      "epoch": 0.5108402192587744,
      "grad_norm": 0.10585922747850418,
      "learning_rate": 9.850304408107019e-06,
      "loss": 0.0036,
      "step": 31220
    },
    {
      "epoch": 0.5110038452098503,
      "grad_norm": 0.24011142551898956,
      "learning_rate": 9.85007313025823e-06,
      "loss": 0.0031,
      "step": 31230
    },
    {
      "epoch": 0.5111674711609261,
      "grad_norm": 0.047218699008226395,
      "learning_rate": 9.849841676605946e-06,
      "loss": 0.0048,
      "step": 31240
    },
    {
      "epoch": 0.511331097112002,
      "grad_norm": 0.1564851552248001,
      "learning_rate": 9.849610047158558e-06,
      "loss": 0.003,
      "step": 31250
    },
    {
      "epoch": 0.5114947230630777,
      "grad_norm": 0.804195761680603,
      "learning_rate": 9.849378241924464e-06,
      "loss": 0.0036,
      "step": 31260
    },
    {
      "epoch": 0.5116583490141536,
      "grad_norm": 0.21099935472011566,
      "learning_rate": 9.84914626091206e-06,
      "loss": 0.0045,
      "step": 31270
    },
    {
      "epoch": 0.5118219749652295,
      "grad_norm": 0.16904276609420776,
      "learning_rate": 9.848914104129761e-06,
      "loss": 0.0041,
      "step": 31280
    },
    {
      "epoch": 0.5119856009163053,
      "grad_norm": 0.10721002519130707,
      "learning_rate": 9.84868177158598e-06,
      "loss": 0.0035,
      "step": 31290
    },
    {
      "epoch": 0.5121492268673812,
      "grad_norm": 0.2088523656129837,
      "learning_rate": 9.848449263289137e-06,
      "loss": 0.0051,
      "step": 31300
    },
    {
      "epoch": 0.512312852818457,
      "grad_norm": 0.27241212129592896,
      "learning_rate": 9.84821657924766e-06,
      "loss": 0.0041,
      "step": 31310
    },
    {
      "epoch": 0.5124764787695328,
      "grad_norm": 0.25362634658813477,
      "learning_rate": 9.847983719469987e-06,
      "loss": 0.0049,
      "step": 31320
    },
    {
      "epoch": 0.5126401047206087,
      "grad_norm": 0.3286561369895935,
      "learning_rate": 9.847750683964555e-06,
      "loss": 0.0047,
      "step": 31330
    },
    {
      "epoch": 0.5128037306716845,
      "grad_norm": 0.19310718774795532,
      "learning_rate": 9.847517472739813e-06,
      "loss": 0.0041,
      "step": 31340
    },
    {
      "epoch": 0.5129673566227604,
      "grad_norm": 0.26821574568748474,
      "learning_rate": 9.847284085804212e-06,
      "loss": 0.0048,
      "step": 31350
    },
    {
      "epoch": 0.5131309825738362,
      "grad_norm": 0.3510286808013916,
      "learning_rate": 9.847050523166213e-06,
      "loss": 0.0071,
      "step": 31360
    },
    {
      "epoch": 0.513294608524912,
      "grad_norm": 0.4269711673259735,
      "learning_rate": 9.846816784834284e-06,
      "loss": 0.0067,
      "step": 31370
    },
    {
      "epoch": 0.5134582344759879,
      "grad_norm": 0.30935657024383545,
      "learning_rate": 9.846582870816893e-06,
      "loss": 0.0045,
      "step": 31380
    },
    {
      "epoch": 0.5136218604270637,
      "grad_norm": 0.2932969331741333,
      "learning_rate": 9.846348781122522e-06,
      "loss": 0.003,
      "step": 31390
    },
    {
      "epoch": 0.5137854863781396,
      "grad_norm": 0.445114403963089,
      "learning_rate": 9.846114515759656e-06,
      "loss": 0.0057,
      "step": 31400
    },
    {
      "epoch": 0.5139491123292154,
      "grad_norm": 0.1886824667453766,
      "learning_rate": 9.845880074736787e-06,
      "loss": 0.0031,
      "step": 31410
    },
    {
      "epoch": 0.5141127382802912,
      "grad_norm": 0.1140340119600296,
      "learning_rate": 9.84564545806241e-06,
      "loss": 0.0064,
      "step": 31420
    },
    {
      "epoch": 0.5142763642313671,
      "grad_norm": 0.08062157034873962,
      "learning_rate": 9.845410665745033e-06,
      "loss": 0.0049,
      "step": 31430
    },
    {
      "epoch": 0.5144399901824429,
      "grad_norm": 0.26562511920928955,
      "learning_rate": 9.845175697793164e-06,
      "loss": 0.0033,
      "step": 31440
    },
    {
      "epoch": 0.5146036161335188,
      "grad_norm": 0.17411820590496063,
      "learning_rate": 9.844940554215323e-06,
      "loss": 0.0053,
      "step": 31450
    },
    {
      "epoch": 0.5147672420845946,
      "grad_norm": 0.18165600299835205,
      "learning_rate": 9.84470523502003e-06,
      "loss": 0.0054,
      "step": 31460
    },
    {
      "epoch": 0.5149308680356705,
      "grad_norm": 0.2793065011501312,
      "learning_rate": 9.844469740215817e-06,
      "loss": 0.005,
      "step": 31470
    },
    {
      "epoch": 0.5150944939867463,
      "grad_norm": 0.2496550977230072,
      "learning_rate": 9.844234069811217e-06,
      "loss": 0.0065,
      "step": 31480
    },
    {
      "epoch": 0.5152581199378221,
      "grad_norm": 0.10200822353363037,
      "learning_rate": 9.843998223814777e-06,
      "loss": 0.0107,
      "step": 31490
    },
    {
      "epoch": 0.515421745888898,
      "grad_norm": 0.5874437093734741,
      "learning_rate": 9.843762202235042e-06,
      "loss": 0.007,
      "step": 31500
    },
    {
      "epoch": 0.5155853718399738,
      "grad_norm": 0.2364024817943573,
      "learning_rate": 9.84352600508057e-06,
      "loss": 0.0043,
      "step": 31510
    },
    {
      "epoch": 0.5157489977910497,
      "grad_norm": 0.3517710268497467,
      "learning_rate": 9.84328963235992e-06,
      "loss": 0.0035,
      "step": 31520
    },
    {
      "epoch": 0.5159126237421255,
      "grad_norm": 0.38161441683769226,
      "learning_rate": 9.843053084081664e-06,
      "loss": 0.0048,
      "step": 31530
    },
    {
      "epoch": 0.5160762496932013,
      "grad_norm": 0.2573119103908539,
      "learning_rate": 9.842816360254372e-06,
      "loss": 0.0033,
      "step": 31540
    },
    {
      "epoch": 0.5162398756442772,
      "grad_norm": 0.22384265065193176,
      "learning_rate": 9.842579460886626e-06,
      "loss": 0.0053,
      "step": 31550
    },
    {
      "epoch": 0.516403501595353,
      "grad_norm": 0.2548035681247711,
      "learning_rate": 9.842342385987014e-06,
      "loss": 0.0081,
      "step": 31560
    },
    {
      "epoch": 0.5165671275464289,
      "grad_norm": 0.2440217286348343,
      "learning_rate": 9.842105135564129e-06,
      "loss": 0.0035,
      "step": 31570
    },
    {
      "epoch": 0.5167307534975047,
      "grad_norm": 0.18748502433300018,
      "learning_rate": 9.84186770962657e-06,
      "loss": 0.0082,
      "step": 31580
    },
    {
      "epoch": 0.5168943794485805,
      "grad_norm": 0.5127915740013123,
      "learning_rate": 9.841630108182944e-06,
      "loss": 0.0066,
      "step": 31590
    },
    {
      "epoch": 0.5170580053996564,
      "grad_norm": 0.1906578242778778,
      "learning_rate": 9.841392331241863e-06,
      "loss": 0.0054,
      "step": 31600
    },
    {
      "epoch": 0.5172216313507322,
      "grad_norm": 0.2827862799167633,
      "learning_rate": 9.841154378811946e-06,
      "loss": 0.0059,
      "step": 31610
    },
    {
      "epoch": 0.5173852573018081,
      "grad_norm": 0.43559354543685913,
      "learning_rate": 9.840916250901818e-06,
      "loss": 0.005,
      "step": 31620
    },
    {
      "epoch": 0.517548883252884,
      "grad_norm": 0.10356776416301727,
      "learning_rate": 9.84067794752011e-06,
      "loss": 0.005,
      "step": 31630
    },
    {
      "epoch": 0.5177125092039597,
      "grad_norm": 0.19057238101959229,
      "learning_rate": 9.840439468675462e-06,
      "loss": 0.0042,
      "step": 31640
    },
    {
      "epoch": 0.5178761351550356,
      "grad_norm": 0.1440851241350174,
      "learning_rate": 9.840200814376516e-06,
      "loss": 0.004,
      "step": 31650
    },
    {
      "epoch": 0.5180397611061114,
      "grad_norm": 0.26910269260406494,
      "learning_rate": 9.839961984631926e-06,
      "loss": 0.0045,
      "step": 31660
    },
    {
      "epoch": 0.5182033870571873,
      "grad_norm": 0.16595585644245148,
      "learning_rate": 9.839722979450343e-06,
      "loss": 0.0041,
      "step": 31670
    },
    {
      "epoch": 0.5183670130082632,
      "grad_norm": 0.1385856419801712,
      "learning_rate": 9.839483798840437e-06,
      "loss": 0.0052,
      "step": 31680
    },
    {
      "epoch": 0.5185306389593389,
      "grad_norm": 0.30816200375556946,
      "learning_rate": 9.839244442810874e-06,
      "loss": 0.0038,
      "step": 31690
    },
    {
      "epoch": 0.5186942649104148,
      "grad_norm": 0.16760753095149994,
      "learning_rate": 9.839004911370329e-06,
      "loss": 0.0035,
      "step": 31700
    },
    {
      "epoch": 0.5188578908614906,
      "grad_norm": 0.06721753627061844,
      "learning_rate": 9.838765204527487e-06,
      "loss": 0.005,
      "step": 31710
    },
    {
      "epoch": 0.5190215168125665,
      "grad_norm": 0.15035152435302734,
      "learning_rate": 9.838525322291036e-06,
      "loss": 0.0056,
      "step": 31720
    },
    {
      "epoch": 0.5191851427636424,
      "grad_norm": 0.252145379781723,
      "learning_rate": 9.838285264669672e-06,
      "loss": 0.0031,
      "step": 31730
    },
    {
      "epoch": 0.5193487687147181,
      "grad_norm": 0.22252891957759857,
      "learning_rate": 9.838045031672096e-06,
      "loss": 0.0039,
      "step": 31740
    },
    {
      "epoch": 0.519512394665794,
      "grad_norm": 0.19901922345161438,
      "learning_rate": 9.837804623307014e-06,
      "loss": 0.0061,
      "step": 31750
    },
    {
      "epoch": 0.5196760206168698,
      "grad_norm": 0.20793382823467255,
      "learning_rate": 9.837564039583143e-06,
      "loss": 0.0041,
      "step": 31760
    },
    {
      "epoch": 0.5198396465679457,
      "grad_norm": 0.15770792961120605,
      "learning_rate": 9.837323280509202e-06,
      "loss": 0.0036,
      "step": 31770
    },
    {
      "epoch": 0.5200032725190216,
      "grad_norm": 0.2245488315820694,
      "learning_rate": 9.837082346093919e-06,
      "loss": 0.006,
      "step": 31780
    },
    {
      "epoch": 0.5201668984700973,
      "grad_norm": 0.4283466935157776,
      "learning_rate": 9.836841236346024e-06,
      "loss": 0.0079,
      "step": 31790
    },
    {
      "epoch": 0.5203305244211732,
      "grad_norm": 0.1798923760652542,
      "learning_rate": 9.83659995127426e-06,
      "loss": 0.0046,
      "step": 31800
    },
    {
      "epoch": 0.520494150372249,
      "grad_norm": 0.09698275476694107,
      "learning_rate": 9.836358490887375e-06,
      "loss": 0.0033,
      "step": 31810
    },
    {
      "epoch": 0.5206577763233249,
      "grad_norm": 0.12913936376571655,
      "learning_rate": 9.836116855194116e-06,
      "loss": 0.0057,
      "step": 31820
    },
    {
      "epoch": 0.5208214022744008,
      "grad_norm": 0.10202960669994354,
      "learning_rate": 9.835875044203245e-06,
      "loss": 0.0047,
      "step": 31830
    },
    {
      "epoch": 0.5209850282254765,
      "grad_norm": 0.2646859884262085,
      "learning_rate": 9.835633057923526e-06,
      "loss": 0.0045,
      "step": 31840
    },
    {
      "epoch": 0.5211486541765524,
      "grad_norm": 0.49617961049079895,
      "learning_rate": 9.835390896363731e-06,
      "loss": 0.0042,
      "step": 31850
    },
    {
      "epoch": 0.5213122801276282,
      "grad_norm": 0.15102189779281616,
      "learning_rate": 9.835148559532638e-06,
      "loss": 0.0034,
      "step": 31860
    },
    {
      "epoch": 0.5214759060787041,
      "grad_norm": 0.2633160352706909,
      "learning_rate": 9.83490604743903e-06,
      "loss": 0.0058,
      "step": 31870
    },
    {
      "epoch": 0.52163953202978,
      "grad_norm": 0.4304446876049042,
      "learning_rate": 9.834663360091698e-06,
      "loss": 0.0053,
      "step": 31880
    },
    {
      "epoch": 0.5218031579808557,
      "grad_norm": 0.21876242756843567,
      "learning_rate": 9.83442049749944e-06,
      "loss": 0.0044,
      "step": 31890
    },
    {
      "epoch": 0.5219667839319316,
      "grad_norm": 0.12237274646759033,
      "learning_rate": 9.834177459671059e-06,
      "loss": 0.0044,
      "step": 31900
    },
    {
      "epoch": 0.5221304098830074,
      "grad_norm": 0.28060805797576904,
      "learning_rate": 9.833934246615363e-06,
      "loss": 0.0045,
      "step": 31910
    },
    {
      "epoch": 0.5222940358340833,
      "grad_norm": 0.09831776469945908,
      "learning_rate": 9.833690858341167e-06,
      "loss": 0.0035,
      "step": 31920
    },
    {
      "epoch": 0.5224576617851592,
      "grad_norm": 0.17224234342575073,
      "learning_rate": 9.833447294857295e-06,
      "loss": 0.0038,
      "step": 31930
    },
    {
      "epoch": 0.522621287736235,
      "grad_norm": 0.22440104186534882,
      "learning_rate": 9.833203556172576e-06,
      "loss": 0.0035,
      "step": 31940
    },
    {
      "epoch": 0.5227849136873108,
      "grad_norm": 0.20979750156402588,
      "learning_rate": 9.832959642295845e-06,
      "loss": 0.0049,
      "step": 31950
    },
    {
      "epoch": 0.5229485396383866,
      "grad_norm": 0.5015364289283752,
      "learning_rate": 9.832715553235943e-06,
      "loss": 0.0045,
      "step": 31960
    },
    {
      "epoch": 0.5231121655894625,
      "grad_norm": 0.11745341122150421,
      "learning_rate": 9.832471289001718e-06,
      "loss": 0.0031,
      "step": 31970
    },
    {
      "epoch": 0.5232757915405384,
      "grad_norm": 0.34065333008766174,
      "learning_rate": 9.832226849602022e-06,
      "loss": 0.0041,
      "step": 31980
    },
    {
      "epoch": 0.5234394174916142,
      "grad_norm": 0.2888043522834778,
      "learning_rate": 9.831982235045716e-06,
      "loss": 0.0033,
      "step": 31990
    },
    {
      "epoch": 0.52360304344269,
      "grad_norm": 0.03164900839328766,
      "learning_rate": 9.831737445341667e-06,
      "loss": 0.0021,
      "step": 32000
    },
    {
      "epoch": 0.5237666693937658,
      "grad_norm": 0.17371325194835663,
      "learning_rate": 9.83149248049875e-06,
      "loss": 0.004,
      "step": 32010
    },
    {
      "epoch": 0.5239302953448417,
      "grad_norm": 0.13506458699703217,
      "learning_rate": 9.831247340525842e-06,
      "loss": 0.0038,
      "step": 32020
    },
    {
      "epoch": 0.5240939212959176,
      "grad_norm": 0.28841733932495117,
      "learning_rate": 9.83100202543183e-06,
      "loss": 0.0057,
      "step": 32030
    },
    {
      "epoch": 0.5242575472469934,
      "grad_norm": 0.5486514568328857,
      "learning_rate": 9.830756535225605e-06,
      "loss": 0.0056,
      "step": 32040
    },
    {
      "epoch": 0.5244211731980692,
      "grad_norm": 0.1623668670654297,
      "learning_rate": 9.830510869916068e-06,
      "loss": 0.0034,
      "step": 32050
    },
    {
      "epoch": 0.524584799149145,
      "grad_norm": 0.24796000123023987,
      "learning_rate": 9.83026502951212e-06,
      "loss": 0.0046,
      "step": 32060
    },
    {
      "epoch": 0.5247484251002209,
      "grad_norm": 0.247153639793396,
      "learning_rate": 9.830019014022674e-06,
      "loss": 0.0061,
      "step": 32070
    },
    {
      "epoch": 0.5249120510512968,
      "grad_norm": 0.28486204147338867,
      "learning_rate": 9.829772823456646e-06,
      "loss": 0.0032,
      "step": 32080
    },
    {
      "epoch": 0.5250756770023726,
      "grad_norm": 0.35897791385650635,
      "learning_rate": 9.829526457822964e-06,
      "loss": 0.0029,
      "step": 32090
    },
    {
      "epoch": 0.5252393029534485,
      "grad_norm": 0.26973316073417664,
      "learning_rate": 9.829279917130555e-06,
      "loss": 0.0045,
      "step": 32100
    },
    {
      "epoch": 0.5254029289045242,
      "grad_norm": 0.21221882104873657,
      "learning_rate": 9.829033201388353e-06,
      "loss": 0.0032,
      "step": 32110
    },
    {
      "epoch": 0.5255665548556001,
      "grad_norm": 0.16623906791210175,
      "learning_rate": 9.828786310605306e-06,
      "loss": 0.0037,
      "step": 32120
    },
    {
      "epoch": 0.5257301808066759,
      "grad_norm": 0.17663079500198364,
      "learning_rate": 9.828539244790361e-06,
      "loss": 0.0066,
      "step": 32130
    },
    {
      "epoch": 0.5258938067577518,
      "grad_norm": 0.31358128786087036,
      "learning_rate": 9.828292003952473e-06,
      "loss": 0.0042,
      "step": 32140
    },
    {
      "epoch": 0.5260574327088277,
      "grad_norm": 0.03548801690340042,
      "learning_rate": 9.828044588100605e-06,
      "loss": 0.0019,
      "step": 32150
    },
    {
      "epoch": 0.5262210586599034,
      "grad_norm": 0.3755888044834137,
      "learning_rate": 9.827796997243724e-06,
      "loss": 0.0041,
      "step": 32160
    },
    {
      "epoch": 0.5263846846109793,
      "grad_norm": 0.22414498031139374,
      "learning_rate": 9.827549231390803e-06,
      "loss": 0.0044,
      "step": 32170
    },
    {
      "epoch": 0.5265483105620551,
      "grad_norm": 0.16670654714107513,
      "learning_rate": 9.827301290550828e-06,
      "loss": 0.0065,
      "step": 32180
    },
    {
      "epoch": 0.526711936513131,
      "grad_norm": 0.30883532762527466,
      "learning_rate": 9.827053174732783e-06,
      "loss": 0.004,
      "step": 32190
    },
    {
      "epoch": 0.5268755624642069,
      "grad_norm": 0.34060797095298767,
      "learning_rate": 9.82680488394566e-06,
      "loss": 0.0057,
      "step": 32200
    },
    {
      "epoch": 0.5270391884152826,
      "grad_norm": 0.250896692276001,
      "learning_rate": 9.826556418198463e-06,
      "loss": 0.0035,
      "step": 32210
    },
    {
      "epoch": 0.5272028143663585,
      "grad_norm": 0.3988708555698395,
      "learning_rate": 9.826307777500196e-06,
      "loss": 0.0042,
      "step": 32220
    },
    {
      "epoch": 0.5273664403174343,
      "grad_norm": 0.2920372188091278,
      "learning_rate": 9.826058961859872e-06,
      "loss": 0.0047,
      "step": 32230
    },
    {
      "epoch": 0.5275300662685102,
      "grad_norm": 0.17561347782611847,
      "learning_rate": 9.825809971286509e-06,
      "loss": 0.0028,
      "step": 32240
    },
    {
      "epoch": 0.5276936922195861,
      "grad_norm": 0.19744329154491425,
      "learning_rate": 9.825560805789131e-06,
      "loss": 0.0032,
      "step": 32250
    },
    {
      "epoch": 0.5278573181706618,
      "grad_norm": 0.2579793930053711,
      "learning_rate": 9.825311465376775e-06,
      "loss": 0.0039,
      "step": 32260
    },
    {
      "epoch": 0.5280209441217377,
      "grad_norm": 0.3522445559501648,
      "learning_rate": 9.825061950058474e-06,
      "loss": 0.0056,
      "step": 32270
    },
    {
      "epoch": 0.5281845700728135,
      "grad_norm": 0.27567538619041443,
      "learning_rate": 9.824812259843275e-06,
      "loss": 0.0048,
      "step": 32280
    },
    {
      "epoch": 0.5283481960238894,
      "grad_norm": 0.3310495913028717,
      "learning_rate": 9.824562394740227e-06,
      "loss": 0.0044,
      "step": 32290
    },
    {
      "epoch": 0.5285118219749653,
      "grad_norm": 0.37350690364837646,
      "learning_rate": 9.824312354758389e-06,
      "loss": 0.005,
      "step": 32300
    },
    {
      "epoch": 0.528675447926041,
      "grad_norm": 0.2704145312309265,
      "learning_rate": 9.824062139906821e-06,
      "loss": 0.0058,
      "step": 32310
    },
    {
      "epoch": 0.5288390738771169,
      "grad_norm": 0.08880201727151871,
      "learning_rate": 9.823811750194597e-06,
      "loss": 0.0038,
      "step": 32320
    },
    {
      "epoch": 0.5290026998281927,
      "grad_norm": 0.18930236995220184,
      "learning_rate": 9.82356118563079e-06,
      "loss": 0.0053,
      "step": 32330
    },
    {
      "epoch": 0.5291663257792686,
      "grad_norm": 0.28846243023872375,
      "learning_rate": 9.82331044622448e-06,
      "loss": 0.0063,
      "step": 32340
    },
    {
      "epoch": 0.5293299517303445,
      "grad_norm": 0.4184209108352661,
      "learning_rate": 9.823059531984764e-06,
      "loss": 0.0051,
      "step": 32350
    },
    {
      "epoch": 0.5294935776814202,
      "grad_norm": 0.08133115619421005,
      "learning_rate": 9.822808442920728e-06,
      "loss": 0.0027,
      "step": 32360
    },
    {
      "epoch": 0.5296572036324961,
      "grad_norm": 0.28720521926879883,
      "learning_rate": 9.82255717904148e-06,
      "loss": 0.0043,
      "step": 32370
    },
    {
      "epoch": 0.5298208295835719,
      "grad_norm": 0.1819610595703125,
      "learning_rate": 9.822305740356123e-06,
      "loss": 0.0053,
      "step": 32380
    },
    {
      "epoch": 0.5299844555346478,
      "grad_norm": 0.15878736972808838,
      "learning_rate": 9.822054126873775e-06,
      "loss": 0.0049,
      "step": 32390
    },
    {
      "epoch": 0.5301480814857237,
      "grad_norm": 0.20396092534065247,
      "learning_rate": 9.821802338603552e-06,
      "loss": 0.0039,
      "step": 32400
    },
    {
      "epoch": 0.5303117074367995,
      "grad_norm": 0.30080902576446533,
      "learning_rate": 9.821550375554586e-06,
      "loss": 0.0033,
      "step": 32410
    },
    {
      "epoch": 0.5304753333878753,
      "grad_norm": 0.193630650639534,
      "learning_rate": 9.821298237736004e-06,
      "loss": 0.0054,
      "step": 32420
    },
    {
      "epoch": 0.5306389593389511,
      "grad_norm": 0.2316664755344391,
      "learning_rate": 9.82104592515695e-06,
      "loss": 0.0024,
      "step": 32430
    },
    {
      "epoch": 0.530802585290027,
      "grad_norm": 0.08750326931476593,
      "learning_rate": 9.820793437826568e-06,
      "loss": 0.0051,
      "step": 32440
    },
    {
      "epoch": 0.5309662112411029,
      "grad_norm": 0.3620743453502655,
      "learning_rate": 9.82054077575401e-06,
      "loss": 0.0054,
      "step": 32450
    },
    {
      "epoch": 0.5311298371921787,
      "grad_norm": 0.09510017186403275,
      "learning_rate": 9.820287938948437e-06,
      "loss": 0.0054,
      "step": 32460
    },
    {
      "epoch": 0.5312934631432545,
      "grad_norm": 0.3773811161518097,
      "learning_rate": 9.820034927419009e-06,
      "loss": 0.0054,
      "step": 32470
    },
    {
      "epoch": 0.5314570890943303,
      "grad_norm": 0.37230491638183594,
      "learning_rate": 9.819781741174901e-06,
      "loss": 0.0053,
      "step": 32480
    },
    {
      "epoch": 0.5316207150454062,
      "grad_norm": 0.11946693062782288,
      "learning_rate": 9.819528380225287e-06,
      "loss": 0.0034,
      "step": 32490
    },
    {
      "epoch": 0.5317843409964821,
      "grad_norm": 0.5339857339859009,
      "learning_rate": 9.819274844579354e-06,
      "loss": 0.0048,
      "step": 32500
    },
    {
      "epoch": 0.5319479669475579,
      "grad_norm": 0.1704990714788437,
      "learning_rate": 9.819021134246292e-06,
      "loss": 0.0036,
      "step": 32510
    },
    {
      "epoch": 0.5321115928986337,
      "grad_norm": 0.12953652441501617,
      "learning_rate": 9.818767249235296e-06,
      "loss": 0.0052,
      "step": 32520
    },
    {
      "epoch": 0.5322752188497095,
      "grad_norm": 0.3123488426208496,
      "learning_rate": 9.818513189555565e-06,
      "loss": 0.0054,
      "step": 32530
    },
    {
      "epoch": 0.5324388448007854,
      "grad_norm": 0.4499529004096985,
      "learning_rate": 9.818258955216316e-06,
      "loss": 0.0053,
      "step": 32540
    },
    {
      "epoch": 0.5326024707518613,
      "grad_norm": 0.05439787730574608,
      "learning_rate": 9.818004546226758e-06,
      "loss": 0.0044,
      "step": 32550
    },
    {
      "epoch": 0.5327660967029371,
      "grad_norm": 0.19502659142017365,
      "learning_rate": 9.817749962596115e-06,
      "loss": 0.0049,
      "step": 32560
    },
    {
      "epoch": 0.532929722654013,
      "grad_norm": 0.13590234518051147,
      "learning_rate": 9.817495204333615e-06,
      "loss": 0.0027,
      "step": 32570
    },
    {
      "epoch": 0.5330933486050887,
      "grad_norm": 0.3587286174297333,
      "learning_rate": 9.817240271448494e-06,
      "loss": 0.004,
      "step": 32580
    },
    {
      "epoch": 0.5332569745561646,
      "grad_norm": 0.1723286360502243,
      "learning_rate": 9.81698516394999e-06,
      "loss": 0.004,
      "step": 32590
    },
    {
      "epoch": 0.5334206005072405,
      "grad_norm": 0.16744588315486908,
      "learning_rate": 9.816729881847351e-06,
      "loss": 0.0029,
      "step": 32600
    },
    {
      "epoch": 0.5335842264583163,
      "grad_norm": 0.2356855720281601,
      "learning_rate": 9.816474425149831e-06,
      "loss": 0.0045,
      "step": 32610
    },
    {
      "epoch": 0.5337478524093922,
      "grad_norm": 0.1434105634689331,
      "learning_rate": 9.81621879386669e-06,
      "loss": 0.0043,
      "step": 32620
    },
    {
      "epoch": 0.5339114783604679,
      "grad_norm": 0.20921590924263,
      "learning_rate": 9.815962988007192e-06,
      "loss": 0.0039,
      "step": 32630
    },
    {
      "epoch": 0.5340751043115438,
      "grad_norm": 0.04490165039896965,
      "learning_rate": 9.81570700758061e-06,
      "loss": 0.003,
      "step": 32640
    },
    {
      "epoch": 0.5342387302626197,
      "grad_norm": 0.17965340614318848,
      "learning_rate": 9.815450852596225e-06,
      "loss": 0.0032,
      "step": 32650
    },
    {
      "epoch": 0.5344023562136955,
      "grad_norm": 0.1089031994342804,
      "learning_rate": 9.815194523063318e-06,
      "loss": 0.0026,
      "step": 32660
    },
    {
      "epoch": 0.5345659821647714,
      "grad_norm": 0.30800139904022217,
      "learning_rate": 9.814938018991184e-06,
      "loss": 0.0028,
      "step": 32670
    },
    {
      "epoch": 0.5347296081158471,
      "grad_norm": 0.5891119837760925,
      "learning_rate": 9.81468134038912e-06,
      "loss": 0.0035,
      "step": 32680
    },
    {
      "epoch": 0.534893234066923,
      "grad_norm": 0.235852912068367,
      "learning_rate": 9.814424487266427e-06,
      "loss": 0.0032,
      "step": 32690
    },
    {
      "epoch": 0.5350568600179989,
      "grad_norm": 0.4114832282066345,
      "learning_rate": 9.814167459632418e-06,
      "loss": 0.0034,
      "step": 32700
    },
    {
      "epoch": 0.5352204859690747,
      "grad_norm": 0.08472443372011185,
      "learning_rate": 9.81391025749641e-06,
      "loss": 0.0033,
      "step": 32710
    },
    {
      "epoch": 0.5353841119201506,
      "grad_norm": 0.1922909915447235,
      "learning_rate": 9.813652880867725e-06,
      "loss": 0.0037,
      "step": 32720
    },
    {
      "epoch": 0.5355477378712263,
      "grad_norm": 0.26681867241859436,
      "learning_rate": 9.813395329755693e-06,
      "loss": 0.0047,
      "step": 32730
    },
    {
      "epoch": 0.5357113638223022,
      "grad_norm": 0.3745154142379761,
      "learning_rate": 9.813137604169649e-06,
      "loss": 0.0033,
      "step": 32740
    },
    {
      "epoch": 0.5358749897733781,
      "grad_norm": 0.2619208097457886,
      "learning_rate": 9.812879704118934e-06,
      "loss": 0.004,
      "step": 32750
    },
    {
      "epoch": 0.5360386157244539,
      "grad_norm": 0.11802687495946884,
      "learning_rate": 9.812621629612896e-06,
      "loss": 0.0048,
      "step": 32760
    },
    {
      "epoch": 0.5362022416755298,
      "grad_norm": 0.21977204084396362,
      "learning_rate": 9.812363380660892e-06,
      "loss": 0.0043,
      "step": 32770
    },
    {
      "epoch": 0.5363658676266055,
      "grad_norm": 0.13824838399887085,
      "learning_rate": 9.812104957272281e-06,
      "loss": 0.0093,
      "step": 32780
    },
    {
      "epoch": 0.5365294935776814,
      "grad_norm": 0.0591072253882885,
      "learning_rate": 9.811846359456433e-06,
      "loss": 0.003,
      "step": 32790
    },
    {
      "epoch": 0.5366931195287573,
      "grad_norm": 0.10084434598684311,
      "learning_rate": 9.811587587222716e-06,
      "loss": 0.0032,
      "step": 32800
    },
    {
      "epoch": 0.5368567454798331,
      "grad_norm": 0.24370764195919037,
      "learning_rate": 9.811328640580517e-06,
      "loss": 0.007,
      "step": 32810
    },
    {
      "epoch": 0.537020371430909,
      "grad_norm": 0.3512839376926422,
      "learning_rate": 9.811069519539217e-06,
      "loss": 0.004,
      "step": 32820
    },
    {
      "epoch": 0.5371839973819847,
      "grad_norm": 0.3090587854385376,
      "learning_rate": 9.81081022410821e-06,
      "loss": 0.0033,
      "step": 32830
    },
    {
      "epoch": 0.5373476233330606,
      "grad_norm": 0.0773339569568634,
      "learning_rate": 9.810550754296894e-06,
      "loss": 0.0045,
      "step": 32840
    },
    {
      "epoch": 0.5375112492841365,
      "grad_norm": 0.15011484920978546,
      "learning_rate": 9.810291110114676e-06,
      "loss": 0.0025,
      "step": 32850
    },
    {
      "epoch": 0.5376748752352123,
      "grad_norm": 0.2257227897644043,
      "learning_rate": 9.810031291570967e-06,
      "loss": 0.0029,
      "step": 32860
    },
    {
      "epoch": 0.5378385011862882,
      "grad_norm": 0.5098716616630554,
      "learning_rate": 9.809771298675184e-06,
      "loss": 0.0048,
      "step": 32870
    },
    {
      "epoch": 0.538002127137364,
      "grad_norm": 0.20705170929431915,
      "learning_rate": 9.80951113143675e-06,
      "loss": 0.0046,
      "step": 32880
    },
    {
      "epoch": 0.5381657530884398,
      "grad_norm": 0.20500947535037994,
      "learning_rate": 9.809250789865097e-06,
      "loss": 0.0035,
      "step": 32890
    },
    {
      "epoch": 0.5383293790395157,
      "grad_norm": 0.11755549907684326,
      "learning_rate": 9.808990273969662e-06,
      "loss": 0.0065,
      "step": 32900
    },
    {
      "epoch": 0.5384930049905915,
      "grad_norm": 0.18920952081680298,
      "learning_rate": 9.80872958375989e-06,
      "loss": 0.0028,
      "step": 32910
    },
    {
      "epoch": 0.5386566309416674,
      "grad_norm": 0.3245505094528198,
      "learning_rate": 9.808468719245225e-06,
      "loss": 0.0039,
      "step": 32920
    },
    {
      "epoch": 0.5388202568927432,
      "grad_norm": 0.2592693269252777,
      "learning_rate": 9.808207680435127e-06,
      "loss": 0.0053,
      "step": 32930
    },
    {
      "epoch": 0.538983882843819,
      "grad_norm": 0.17564500868320465,
      "learning_rate": 9.807946467339059e-06,
      "loss": 0.0053,
      "step": 32940
    },
    {
      "epoch": 0.5391475087948949,
      "grad_norm": 0.23559381067752838,
      "learning_rate": 9.807685079966483e-06,
      "loss": 0.003,
      "step": 32950
    },
    {
      "epoch": 0.5393111347459707,
      "grad_norm": 0.0649779886007309,
      "learning_rate": 9.807423518326879e-06,
      "loss": 0.0028,
      "step": 32960
    },
    {
      "epoch": 0.5394747606970466,
      "grad_norm": 0.2339847832918167,
      "learning_rate": 9.807161782429729e-06,
      "loss": 0.0048,
      "step": 32970
    },
    {
      "epoch": 0.5396383866481224,
      "grad_norm": 0.3472442924976349,
      "learning_rate": 9.806899872284517e-06,
      "loss": 0.0036,
      "step": 32980
    },
    {
      "epoch": 0.5398020125991982,
      "grad_norm": 0.24282602965831757,
      "learning_rate": 9.806637787900735e-06,
      "loss": 0.0039,
      "step": 32990
    },
    {
      "epoch": 0.539965638550274,
      "grad_norm": 0.3652968406677246,
      "learning_rate": 9.806375529287888e-06,
      "loss": 0.0044,
      "step": 33000
    },
    {
      "epoch": 0.5401292645013499,
      "grad_norm": 0.6360982656478882,
      "learning_rate": 9.80611309645548e-06,
      "loss": 0.0062,
      "step": 33010
    },
    {
      "epoch": 0.5402928904524258,
      "grad_norm": 0.1364808827638626,
      "learning_rate": 9.805850489413025e-06,
      "loss": 0.0038,
      "step": 33020
    },
    {
      "epoch": 0.5404565164035016,
      "grad_norm": 0.12798751890659332,
      "learning_rate": 9.805587708170036e-06,
      "loss": 0.006,
      "step": 33030
    },
    {
      "epoch": 0.5406201423545774,
      "grad_norm": 0.22705502808094025,
      "learning_rate": 9.805324752736047e-06,
      "loss": 0.0034,
      "step": 33040
    },
    {
      "epoch": 0.5407837683056532,
      "grad_norm": 0.1974307894706726,
      "learning_rate": 9.805061623120583e-06,
      "loss": 0.0029,
      "step": 33050
    },
    {
      "epoch": 0.5409473942567291,
      "grad_norm": 0.21818414330482483,
      "learning_rate": 9.804798319333184e-06,
      "loss": 0.0051,
      "step": 33060
    },
    {
      "epoch": 0.541111020207805,
      "grad_norm": 0.16655340790748596,
      "learning_rate": 9.804534841383394e-06,
      "loss": 0.0032,
      "step": 33070
    },
    {
      "epoch": 0.5412746461588808,
      "grad_norm": 0.0988406091928482,
      "learning_rate": 9.804271189280763e-06,
      "loss": 0.0046,
      "step": 33080
    },
    {
      "epoch": 0.5414382721099567,
      "grad_norm": 0.41690224409103394,
      "learning_rate": 9.804007363034849e-06,
      "loss": 0.0035,
      "step": 33090
    },
    {
      "epoch": 0.5416018980610324,
      "grad_norm": 0.2624330520629883,
      "learning_rate": 9.803743362655213e-06,
      "loss": 0.0043,
      "step": 33100
    },
    {
      "epoch": 0.5417655240121083,
      "grad_norm": 0.24833914637565613,
      "learning_rate": 9.803479188151426e-06,
      "loss": 0.0044,
      "step": 33110
    },
    {
      "epoch": 0.5419291499631842,
      "grad_norm": 0.15569384396076202,
      "learning_rate": 9.803214839533065e-06,
      "loss": 0.0052,
      "step": 33120
    },
    {
      "epoch": 0.54209277591426,
      "grad_norm": 0.1732093244791031,
      "learning_rate": 9.802950316809707e-06,
      "loss": 0.0057,
      "step": 33130
    },
    {
      "epoch": 0.5422564018653359,
      "grad_norm": 0.16405533254146576,
      "learning_rate": 9.802685619990945e-06,
      "loss": 0.0044,
      "step": 33140
    },
    {
      "epoch": 0.5424200278164116,
      "grad_norm": 0.3588194251060486,
      "learning_rate": 9.802420749086373e-06,
      "loss": 0.0041,
      "step": 33150
    },
    {
      "epoch": 0.5425836537674875,
      "grad_norm": 0.2474457174539566,
      "learning_rate": 9.802155704105591e-06,
      "loss": 0.0062,
      "step": 33160
    },
    {
      "epoch": 0.5427472797185634,
      "grad_norm": 0.25152698159217834,
      "learning_rate": 9.801890485058206e-06,
      "loss": 0.0043,
      "step": 33170
    },
    {
      "epoch": 0.5429109056696392,
      "grad_norm": 0.1282421201467514,
      "learning_rate": 9.801625091953832e-06,
      "loss": 0.0034,
      "step": 33180
    },
    {
      "epoch": 0.5430745316207151,
      "grad_norm": 0.23702237010002136,
      "learning_rate": 9.80135952480209e-06,
      "loss": 0.0045,
      "step": 33190
    },
    {
      "epoch": 0.5432381575717908,
      "grad_norm": 0.2506360709667206,
      "learning_rate": 9.801093783612603e-06,
      "loss": 0.0038,
      "step": 33200
    },
    {
      "epoch": 0.5434017835228667,
      "grad_norm": 0.12637735903263092,
      "learning_rate": 9.800827868395006e-06,
      "loss": 0.0042,
      "step": 33210
    },
    {
      "epoch": 0.5435654094739426,
      "grad_norm": 0.2550637125968933,
      "learning_rate": 9.800561779158937e-06,
      "loss": 0.0049,
      "step": 33220
    },
    {
      "epoch": 0.5437290354250184,
      "grad_norm": 0.21127969026565552,
      "learning_rate": 9.800295515914044e-06,
      "loss": 0.0021,
      "step": 33230
    },
    {
      "epoch": 0.5438926613760943,
      "grad_norm": 0.27860361337661743,
      "learning_rate": 9.800029078669973e-06,
      "loss": 0.01,
      "step": 33240
    },
    {
      "epoch": 0.54405628732717,
      "grad_norm": 0.24480634927749634,
      "learning_rate": 9.799762467436384e-06,
      "loss": 0.0041,
      "step": 33250
    },
    {
      "epoch": 0.5442199132782459,
      "grad_norm": 0.13082528114318848,
      "learning_rate": 9.799495682222944e-06,
      "loss": 0.0047,
      "step": 33260
    },
    {
      "epoch": 0.5443835392293218,
      "grad_norm": 0.3724941611289978,
      "learning_rate": 9.79922872303932e-06,
      "loss": 0.0041,
      "step": 33270
    },
    {
      "epoch": 0.5445471651803976,
      "grad_norm": 0.37125328183174133,
      "learning_rate": 9.798961589895188e-06,
      "loss": 0.0032,
      "step": 33280
    },
    {
      "epoch": 0.5447107911314735,
      "grad_norm": 0.18387950956821442,
      "learning_rate": 9.798694282800234e-06,
      "loss": 0.0044,
      "step": 33290
    },
    {
      "epoch": 0.5448744170825492,
      "grad_norm": 0.47088825702667236,
      "learning_rate": 9.798426801764144e-06,
      "loss": 0.0041,
      "step": 33300
    },
    {
      "epoch": 0.5450380430336251,
      "grad_norm": 0.0341993011534214,
      "learning_rate": 9.798159146796617e-06,
      "loss": 0.0068,
      "step": 33310
    },
    {
      "epoch": 0.545201668984701,
      "grad_norm": 0.19534042477607727,
      "learning_rate": 9.797891317907352e-06,
      "loss": 0.0027,
      "step": 33320
    },
    {
      "epoch": 0.5453652949357768,
      "grad_norm": 0.3997628688812256,
      "learning_rate": 9.797623315106057e-06,
      "loss": 0.0038,
      "step": 33330
    },
    {
      "epoch": 0.5455289208868527,
      "grad_norm": 0.05273151770234108,
      "learning_rate": 9.79735513840245e-06,
      "loss": 0.0061,
      "step": 33340
    },
    {
      "epoch": 0.5456925468379284,
      "grad_norm": 0.3081619441509247,
      "learning_rate": 9.797086787806249e-06,
      "loss": 0.0032,
      "step": 33350
    },
    {
      "epoch": 0.5458561727890043,
      "grad_norm": 0.10779151320457458,
      "learning_rate": 9.79681826332718e-06,
      "loss": 0.0021,
      "step": 33360
    },
    {
      "epoch": 0.5460197987400802,
      "grad_norm": 0.22849293053150177,
      "learning_rate": 9.796549564974978e-06,
      "loss": 0.0042,
      "step": 33370
    },
    {
      "epoch": 0.546183424691156,
      "grad_norm": 0.0554744191467762,
      "learning_rate": 9.796280692759384e-06,
      "loss": 0.0042,
      "step": 33380
    },
    {
      "epoch": 0.5463470506422319,
      "grad_norm": 0.2788725197315216,
      "learning_rate": 9.79601164669014e-06,
      "loss": 0.0045,
      "step": 33390
    },
    {
      "epoch": 0.5465106765933077,
      "grad_norm": 0.1635172814130783,
      "learning_rate": 9.795742426777002e-06,
      "loss": 0.0048,
      "step": 33400
    },
    {
      "epoch": 0.5466743025443835,
      "grad_norm": 0.1375514566898346,
      "learning_rate": 9.795473033029727e-06,
      "loss": 0.0062,
      "step": 33410
    },
    {
      "epoch": 0.5468379284954594,
      "grad_norm": 0.10028968751430511,
      "learning_rate": 9.795203465458081e-06,
      "loss": 0.0036,
      "step": 33420
    },
    {
      "epoch": 0.5470015544465352,
      "grad_norm": 0.27354317903518677,
      "learning_rate": 9.794933724071834e-06,
      "loss": 0.0036,
      "step": 33430
    },
    {
      "epoch": 0.5471651803976111,
      "grad_norm": 0.32796114683151245,
      "learning_rate": 9.794663808880764e-06,
      "loss": 0.004,
      "step": 33440
    },
    {
      "epoch": 0.5473288063486869,
      "grad_norm": 0.4347842335700989,
      "learning_rate": 9.794393719894655e-06,
      "loss": 0.0035,
      "step": 33450
    },
    {
      "epoch": 0.5474924322997627,
      "grad_norm": 0.25115635991096497,
      "learning_rate": 9.794123457123296e-06,
      "loss": 0.0033,
      "step": 33460
    },
    {
      "epoch": 0.5476560582508386,
      "grad_norm": 0.23931430280208588,
      "learning_rate": 9.793853020576484e-06,
      "loss": 0.0024,
      "step": 33470
    },
    {
      "epoch": 0.5478196842019144,
      "grad_norm": 0.13950666785240173,
      "learning_rate": 9.793582410264023e-06,
      "loss": 0.0036,
      "step": 33480
    },
    {
      "epoch": 0.5479833101529903,
      "grad_norm": 0.3506321609020233,
      "learning_rate": 9.793311626195718e-06,
      "loss": 0.0042,
      "step": 33490
    },
    {
      "epoch": 0.5481469361040661,
      "grad_norm": 0.23200082778930664,
      "learning_rate": 9.793040668381388e-06,
      "loss": 0.0039,
      "step": 33500
    },
    {
      "epoch": 0.548310562055142,
      "grad_norm": 0.6095235347747803,
      "learning_rate": 9.792769536830855e-06,
      "loss": 0.0039,
      "step": 33510
    },
    {
      "epoch": 0.5484741880062178,
      "grad_norm": 0.32369744777679443,
      "learning_rate": 9.792498231553943e-06,
      "loss": 0.0032,
      "step": 33520
    },
    {
      "epoch": 0.5486378139572936,
      "grad_norm": 0.1714697927236557,
      "learning_rate": 9.792226752560492e-06,
      "loss": 0.0074,
      "step": 33530
    },
    {
      "epoch": 0.5488014399083695,
      "grad_norm": 0.15778622031211853,
      "learning_rate": 9.791955099860335e-06,
      "loss": 0.0027,
      "step": 33540
    },
    {
      "epoch": 0.5489650658594453,
      "grad_norm": 0.28746315836906433,
      "learning_rate": 9.791683273463327e-06,
      "loss": 0.0057,
      "step": 33550
    },
    {
      "epoch": 0.5491286918105212,
      "grad_norm": 0.3084624409675598,
      "learning_rate": 9.791411273379314e-06,
      "loss": 0.0034,
      "step": 33560
    },
    {
      "epoch": 0.549292317761597,
      "grad_norm": 0.3382299244403839,
      "learning_rate": 9.79113909961816e-06,
      "loss": 0.0039,
      "step": 33570
    },
    {
      "epoch": 0.5494559437126728,
      "grad_norm": 0.16015921533107758,
      "learning_rate": 9.790866752189725e-06,
      "loss": 0.0051,
      "step": 33580
    },
    {
      "epoch": 0.5496195696637487,
      "grad_norm": 0.2711232900619507,
      "learning_rate": 9.790594231103888e-06,
      "loss": 0.004,
      "step": 33590
    },
    {
      "epoch": 0.5497831956148245,
      "grad_norm": 0.5179532170295715,
      "learning_rate": 9.790321536370525e-06,
      "loss": 0.0039,
      "step": 33600
    },
    {
      "epoch": 0.5499468215659004,
      "grad_norm": 0.12061049789190292,
      "learning_rate": 9.790048667999517e-06,
      "loss": 0.0038,
      "step": 33610
    },
    {
      "epoch": 0.5501104475169762,
      "grad_norm": 0.3108615577220917,
      "learning_rate": 9.789775626000757e-06,
      "loss": 0.0029,
      "step": 33620
    },
    {
      "epoch": 0.550274073468052,
      "grad_norm": 0.26426491141319275,
      "learning_rate": 9.789502410384145e-06,
      "loss": 0.0056,
      "step": 33630
    },
    {
      "epoch": 0.5504376994191279,
      "grad_norm": 0.04900655150413513,
      "learning_rate": 9.789229021159581e-06,
      "loss": 0.0035,
      "step": 33640
    },
    {
      "epoch": 0.5506013253702037,
      "grad_norm": 0.16743774712085724,
      "learning_rate": 9.788955458336976e-06,
      "loss": 0.0048,
      "step": 33650
    },
    {
      "epoch": 0.5507649513212796,
      "grad_norm": 0.5871763229370117,
      "learning_rate": 9.788681721926243e-06,
      "loss": 0.0046,
      "step": 33660
    },
    {
      "epoch": 0.5509285772723554,
      "grad_norm": 0.3706881105899811,
      "learning_rate": 9.788407811937311e-06,
      "loss": 0.0032,
      "step": 33670
    },
    {
      "epoch": 0.5510922032234312,
      "grad_norm": 0.2674044668674469,
      "learning_rate": 9.7881337283801e-06,
      "loss": 0.0036,
      "step": 33680
    },
    {
      "epoch": 0.5512558291745071,
      "grad_norm": 0.19376537203788757,
      "learning_rate": 9.787859471264552e-06,
      "loss": 0.0035,
      "step": 33690
    },
    {
      "epoch": 0.5514194551255829,
      "grad_norm": 0.16932715475559235,
      "learning_rate": 9.787585040600605e-06,
      "loss": 0.0043,
      "step": 33700
    },
    {
      "epoch": 0.5515830810766588,
      "grad_norm": 0.7847943305969238,
      "learning_rate": 9.787310436398208e-06,
      "loss": 0.0036,
      "step": 33710
    },
    {
      "epoch": 0.5517467070277347,
      "grad_norm": 0.3932487368583679,
      "learning_rate": 9.787035658667313e-06,
      "loss": 0.0031,
      "step": 33720
    },
    {
      "epoch": 0.5519103329788104,
      "grad_norm": 0.14990553259849548,
      "learning_rate": 9.78676070741788e-06,
      "loss": 0.0035,
      "step": 33730
    },
    {
      "epoch": 0.5520739589298863,
      "grad_norm": 0.6247564554214478,
      "learning_rate": 9.786485582659876e-06,
      "loss": 0.0076,
      "step": 33740
    },
    {
      "epoch": 0.5522375848809621,
      "grad_norm": 0.09049864858388901,
      "learning_rate": 9.786210284403274e-06,
      "loss": 0.0032,
      "step": 33750
    },
    {
      "epoch": 0.552401210832038,
      "grad_norm": 0.07898964732885361,
      "learning_rate": 9.785934812658053e-06,
      "loss": 0.0033,
      "step": 33760
    },
    {
      "epoch": 0.5525648367831139,
      "grad_norm": 0.059457868337631226,
      "learning_rate": 9.785659167434197e-06,
      "loss": 0.0061,
      "step": 33770
    },
    {
      "epoch": 0.5527284627341896,
      "grad_norm": 0.12354923039674759,
      "learning_rate": 9.785383348741699e-06,
      "loss": 0.0024,
      "step": 33780
    },
    {
      "epoch": 0.5528920886852655,
      "grad_norm": 0.41873377561569214,
      "learning_rate": 9.785107356590555e-06,
      "loss": 0.0029,
      "step": 33790
    },
    {
      "epoch": 0.5530557146363413,
      "grad_norm": 0.33212020993232727,
      "learning_rate": 9.78483119099077e-06,
      "loss": 0.0053,
      "step": 33800
    },
    {
      "epoch": 0.5532193405874172,
      "grad_norm": 0.10025341063737869,
      "learning_rate": 9.784554851952354e-06,
      "loss": 0.0041,
      "step": 33810
    },
    {
      "epoch": 0.5533829665384931,
      "grad_norm": 0.36780327558517456,
      "learning_rate": 9.784278339485327e-06,
      "loss": 0.0052,
      "step": 33820
    },
    {
      "epoch": 0.5535465924895688,
      "grad_norm": 0.19688554108142853,
      "learning_rate": 9.784001653599707e-06,
      "loss": 0.0029,
      "step": 33830
    },
    {
      "epoch": 0.5537102184406447,
      "grad_norm": 0.23417767882347107,
      "learning_rate": 9.783724794305524e-06,
      "loss": 0.0022,
      "step": 33840
    },
    {
      "epoch": 0.5538738443917205,
      "grad_norm": 0.4738108217716217,
      "learning_rate": 9.783447761612816e-06,
      "loss": 0.0062,
      "step": 33850
    },
    {
      "epoch": 0.5540374703427964,
      "grad_norm": 0.07847896963357925,
      "learning_rate": 9.783170555531623e-06,
      "loss": 0.0033,
      "step": 33860
    },
    {
      "epoch": 0.5542010962938722,
      "grad_norm": 0.20875278115272522,
      "learning_rate": 9.782893176071993e-06,
      "loss": 0.0042,
      "step": 33870
    },
    {
      "epoch": 0.554364722244948,
      "grad_norm": 0.388008713722229,
      "learning_rate": 9.78261562324398e-06,
      "loss": 0.0058,
      "step": 33880
    },
    {
      "epoch": 0.5545283481960239,
      "grad_norm": 0.20827463269233704,
      "learning_rate": 9.782337897057648e-06,
      "loss": 0.0038,
      "step": 33890
    },
    {
      "epoch": 0.5546919741470997,
      "grad_norm": 0.22781474888324738,
      "learning_rate": 9.782059997523058e-06,
      "loss": 0.0032,
      "step": 33900
    },
    {
      "epoch": 0.5548556000981756,
      "grad_norm": 0.21710669994354248,
      "learning_rate": 9.78178192465029e-06,
      "loss": 0.0044,
      "step": 33910
    },
    {
      "epoch": 0.5550192260492514,
      "grad_norm": 0.23331332206726074,
      "learning_rate": 9.781503678449417e-06,
      "loss": 0.0037,
      "step": 33920
    },
    {
      "epoch": 0.5551828520003272,
      "grad_norm": 0.3796658217906952,
      "learning_rate": 9.781225258930529e-06,
      "loss": 0.0052,
      "step": 33930
    },
    {
      "epoch": 0.5553464779514031,
      "grad_norm": 0.2554663121700287,
      "learning_rate": 9.780946666103717e-06,
      "loss": 0.0042,
      "step": 33940
    },
    {
      "epoch": 0.5555101039024789,
      "grad_norm": 0.06200270354747772,
      "learning_rate": 9.780667899979077e-06,
      "loss": 0.0057,
      "step": 33950
    },
    {
      "epoch": 0.5556737298535548,
      "grad_norm": 0.37055638432502747,
      "learning_rate": 9.780388960566719e-06,
      "loss": 0.004,
      "step": 33960
    },
    {
      "epoch": 0.5558373558046306,
      "grad_norm": 0.10813041776418686,
      "learning_rate": 9.780109847876749e-06,
      "loss": 0.0049,
      "step": 33970
    },
    {
      "epoch": 0.5560009817557064,
      "grad_norm": 0.20948156714439392,
      "learning_rate": 9.779830561919285e-06,
      "loss": 0.003,
      "step": 33980
    },
    {
      "epoch": 0.5561646077067823,
      "grad_norm": 0.3225252032279968,
      "learning_rate": 9.77955110270445e-06,
      "loss": 0.0052,
      "step": 33990
    },
    {
      "epoch": 0.5563282336578581,
      "grad_norm": 0.13214503228664398,
      "learning_rate": 9.779271470242377e-06,
      "loss": 0.0046,
      "step": 34000
    },
    {
      "epoch": 0.556491859608934,
      "grad_norm": 0.23867765069007874,
      "learning_rate": 9.7789916645432e-06,
      "loss": 0.0039,
      "step": 34010
    },
    {
      "epoch": 0.5566554855600098,
      "grad_norm": 0.10316136479377747,
      "learning_rate": 9.77871168561706e-06,
      "loss": 0.0032,
      "step": 34020
    },
    {
      "epoch": 0.5568191115110857,
      "grad_norm": 0.07306062430143356,
      "learning_rate": 9.778431533474107e-06,
      "loss": 0.0044,
      "step": 34030
    },
    {
      "epoch": 0.5569827374621615,
      "grad_norm": 0.1291823536157608,
      "learning_rate": 9.778151208124496e-06,
      "loss": 0.0038,
      "step": 34040
    },
    {
      "epoch": 0.5571463634132373,
      "grad_norm": 0.19382648169994354,
      "learning_rate": 9.777870709578388e-06,
      "loss": 0.0043,
      "step": 34050
    },
    {
      "epoch": 0.5573099893643132,
      "grad_norm": 0.20677529275417328,
      "learning_rate": 9.777590037845949e-06,
      "loss": 0.003,
      "step": 34060
    },
    {
      "epoch": 0.557473615315389,
      "grad_norm": 0.2174794226884842,
      "learning_rate": 9.777309192937356e-06,
      "loss": 0.0037,
      "step": 34070
    },
    {
      "epoch": 0.5576372412664649,
      "grad_norm": 0.27535009384155273,
      "learning_rate": 9.777028174862784e-06,
      "loss": 0.005,
      "step": 34080
    },
    {
      "epoch": 0.5578008672175407,
      "grad_norm": 0.24558651447296143,
      "learning_rate": 9.776746983632424e-06,
      "loss": 0.0049,
      "step": 34090
    },
    {
      "epoch": 0.5579644931686165,
      "grad_norm": 0.17967216670513153,
      "learning_rate": 9.776465619256465e-06,
      "loss": 0.0046,
      "step": 34100
    },
    {
      "epoch": 0.5581281191196924,
      "grad_norm": 0.12029209733009338,
      "learning_rate": 9.776184081745108e-06,
      "loss": 0.0043,
      "step": 34110
    },
    {
      "epoch": 0.5582917450707682,
      "grad_norm": 0.0979442149400711,
      "learning_rate": 9.775902371108556e-06,
      "loss": 0.0051,
      "step": 34120
    },
    {
      "epoch": 0.5584553710218441,
      "grad_norm": 0.16064612567424774,
      "learning_rate": 9.775620487357023e-06,
      "loss": 0.0029,
      "step": 34130
    },
    {
      "epoch": 0.55861899697292,
      "grad_norm": 0.2996758222579956,
      "learning_rate": 9.775338430500724e-06,
      "loss": 0.0057,
      "step": 34140
    },
    {
      "epoch": 0.5587826229239957,
      "grad_norm": 0.05495762452483177,
      "learning_rate": 9.775056200549885e-06,
      "loss": 0.0029,
      "step": 34150
    },
    {
      "epoch": 0.5589462488750716,
      "grad_norm": 0.3233628571033478,
      "learning_rate": 9.774773797514737e-06,
      "loss": 0.0033,
      "step": 34160
    },
    {
      "epoch": 0.5591098748261474,
      "grad_norm": 0.08589668571949005,
      "learning_rate": 9.774491221405512e-06,
      "loss": 0.0027,
      "step": 34170
    },
    {
      "epoch": 0.5592735007772233,
      "grad_norm": 0.31055769324302673,
      "learning_rate": 9.774208472232456e-06,
      "loss": 0.0045,
      "step": 34180
    },
    {
      "epoch": 0.5594371267282992,
      "grad_norm": 0.042775653302669525,
      "learning_rate": 9.773925550005818e-06,
      "loss": 0.004,
      "step": 34190
    },
    {
      "epoch": 0.5596007526793749,
      "grad_norm": 0.07803049683570862,
      "learning_rate": 9.773642454735852e-06,
      "loss": 0.0047,
      "step": 34200
    },
    {
      "epoch": 0.5597643786304508,
      "grad_norm": 0.24542154371738434,
      "learning_rate": 9.773359186432822e-06,
      "loss": 0.0042,
      "step": 34210
    },
    {
      "epoch": 0.5599280045815266,
      "grad_norm": 0.07078225910663605,
      "learning_rate": 9.773075745106992e-06,
      "loss": 0.0031,
      "step": 34220
    },
    {
      "epoch": 0.5600916305326025,
      "grad_norm": 0.08734782785177231,
      "learning_rate": 9.772792130768638e-06,
      "loss": 0.0025,
      "step": 34230
    },
    {
      "epoch": 0.5602552564836784,
      "grad_norm": 0.19833236932754517,
      "learning_rate": 9.772508343428042e-06,
      "loss": 0.0041,
      "step": 34240
    },
    {
      "epoch": 0.5604188824347541,
      "grad_norm": 0.09172920882701874,
      "learning_rate": 9.772224383095487e-06,
      "loss": 0.0021,
      "step": 34250
    },
    {
      "epoch": 0.56058250838583,
      "grad_norm": 0.2891700565814972,
      "learning_rate": 9.77194024978127e-06,
      "loss": 0.0032,
      "step": 34260
    },
    {
      "epoch": 0.5607461343369058,
      "grad_norm": 0.10319384932518005,
      "learning_rate": 9.771655943495685e-06,
      "loss": 0.004,
      "step": 34270
    },
    {
      "epoch": 0.5609097602879817,
      "grad_norm": 0.34937310218811035,
      "learning_rate": 9.771371464249043e-06,
      "loss": 0.0042,
      "step": 34280
    },
    {
      "epoch": 0.5610733862390576,
      "grad_norm": 0.07920363545417786,
      "learning_rate": 9.771086812051652e-06,
      "loss": 0.0047,
      "step": 34290
    },
    {
      "epoch": 0.5612370121901333,
      "grad_norm": 0.39690378308296204,
      "learning_rate": 9.77080198691383e-06,
      "loss": 0.0049,
      "step": 34300
    },
    {
      "epoch": 0.5614006381412092,
      "grad_norm": 0.4145283102989197,
      "learning_rate": 9.770516988845906e-06,
      "loss": 0.0038,
      "step": 34310
    },
    {
      "epoch": 0.561564264092285,
      "grad_norm": 0.31935012340545654,
      "learning_rate": 9.770231817858205e-06,
      "loss": 0.0054,
      "step": 34320
    },
    {
      "epoch": 0.5617278900433609,
      "grad_norm": 0.15822888910770416,
      "learning_rate": 9.769946473961064e-06,
      "loss": 0.0042,
      "step": 34330
    },
    {
      "epoch": 0.5618915159944368,
      "grad_norm": 0.2548803389072418,
      "learning_rate": 9.769660957164828e-06,
      "loss": 0.0059,
      "step": 34340
    },
    {
      "epoch": 0.5620551419455125,
      "grad_norm": 0.25123193860054016,
      "learning_rate": 9.769375267479847e-06,
      "loss": 0.005,
      "step": 34350
    },
    {
      "epoch": 0.5622187678965884,
      "grad_norm": 0.2755795121192932,
      "learning_rate": 9.769089404916475e-06,
      "loss": 0.0041,
      "step": 34360
    },
    {
      "epoch": 0.5623823938476642,
      "grad_norm": 0.08042644709348679,
      "learning_rate": 9.768803369485074e-06,
      "loss": 0.0025,
      "step": 34370
    },
    {
      "epoch": 0.5625460197987401,
      "grad_norm": 0.4067413806915283,
      "learning_rate": 9.768517161196012e-06,
      "loss": 0.0057,
      "step": 34380
    },
    {
      "epoch": 0.562709645749816,
      "grad_norm": 0.32594916224479675,
      "learning_rate": 9.768230780059664e-06,
      "loss": 0.0073,
      "step": 34390
    },
    {
      "epoch": 0.5628732717008917,
      "grad_norm": 0.11061684042215347,
      "learning_rate": 9.767944226086413e-06,
      "loss": 0.0024,
      "step": 34400
    },
    {
      "epoch": 0.5630368976519676,
      "grad_norm": 0.26365336775779724,
      "learning_rate": 9.76765749928664e-06,
      "loss": 0.0024,
      "step": 34410
    },
    {
      "epoch": 0.5632005236030434,
      "grad_norm": 0.1506739854812622,
      "learning_rate": 9.767370599670743e-06,
      "loss": 0.0042,
      "step": 34420
    },
    {
      "epoch": 0.5633641495541193,
      "grad_norm": 0.19781681895256042,
      "learning_rate": 9.767083527249118e-06,
      "loss": 0.0035,
      "step": 34430
    },
    {
      "epoch": 0.5635277755051952,
      "grad_norm": 0.14687316119670868,
      "learning_rate": 9.766796282032173e-06,
      "loss": 0.0037,
      "step": 34440
    },
    {
      "epoch": 0.563691401456271,
      "grad_norm": 0.1054859310388565,
      "learning_rate": 9.76650886403032e-06,
      "loss": 0.0033,
      "step": 34450
    },
    {
      "epoch": 0.5638550274073468,
      "grad_norm": 0.206525519490242,
      "learning_rate": 9.766221273253978e-06,
      "loss": 0.0063,
      "step": 34460
    },
    {
      "epoch": 0.5640186533584226,
      "grad_norm": 0.16110996901988983,
      "learning_rate": 9.76593350971357e-06,
      "loss": 0.0038,
      "step": 34470
    },
    {
      "epoch": 0.5641822793094985,
      "grad_norm": 0.2761392891407013,
      "learning_rate": 9.765645573419525e-06,
      "loss": 0.004,
      "step": 34480
    },
    {
      "epoch": 0.5643459052605744,
      "grad_norm": 0.2078597992658615,
      "learning_rate": 9.765357464382282e-06,
      "loss": 0.0058,
      "step": 34490
    },
    {
      "epoch": 0.5645095312116502,
      "grad_norm": 0.32887399196624756,
      "learning_rate": 9.765069182612286e-06,
      "loss": 0.0049,
      "step": 34500
    },
    {
      "epoch": 0.564673157162726,
      "grad_norm": 0.2882753014564514,
      "learning_rate": 9.764780728119983e-06,
      "loss": 0.003,
      "step": 34510
    },
    {
      "epoch": 0.5648367831138018,
      "grad_norm": 0.17291636765003204,
      "learning_rate": 9.764492100915832e-06,
      "loss": 0.0041,
      "step": 34520
    },
    {
      "epoch": 0.5650004090648777,
      "grad_norm": 0.2206551879644394,
      "learning_rate": 9.76420330101029e-06,
      "loss": 0.0053,
      "step": 34530
    },
    {
      "epoch": 0.5651640350159536,
      "grad_norm": 0.07411439716815948,
      "learning_rate": 9.763914328413833e-06,
      "loss": 0.0032,
      "step": 34540
    },
    {
      "epoch": 0.5653276609670294,
      "grad_norm": 0.03896886855363846,
      "learning_rate": 9.76362518313693e-06,
      "loss": 0.0034,
      "step": 34550
    },
    {
      "epoch": 0.5654912869181052,
      "grad_norm": 0.17319238185882568,
      "learning_rate": 9.763335865190062e-06,
      "loss": 0.0045,
      "step": 34560
    },
    {
      "epoch": 0.565654912869181,
      "grad_norm": 0.4278779923915863,
      "learning_rate": 9.76304637458372e-06,
      "loss": 0.0052,
      "step": 34570
    },
    {
      "epoch": 0.5658185388202569,
      "grad_norm": 0.20016364753246307,
      "learning_rate": 9.762756711328391e-06,
      "loss": 0.0044,
      "step": 34580
    },
    {
      "epoch": 0.5659821647713328,
      "grad_norm": 0.20062294602394104,
      "learning_rate": 9.762466875434579e-06,
      "loss": 0.0033,
      "step": 34590
    },
    {
      "epoch": 0.5661457907224086,
      "grad_norm": 0.3652080297470093,
      "learning_rate": 9.76217686691279e-06,
      "loss": 0.0042,
      "step": 34600
    },
    {
      "epoch": 0.5663094166734844,
      "grad_norm": 0.22696973383426666,
      "learning_rate": 9.761886685773536e-06,
      "loss": 0.0034,
      "step": 34610
    },
    {
      "epoch": 0.5664730426245602,
      "grad_norm": 0.11855337023735046,
      "learning_rate": 9.761596332027334e-06,
      "loss": 0.0065,
      "step": 34620
    },
    {
      "epoch": 0.5666366685756361,
      "grad_norm": 0.09045524150133133,
      "learning_rate": 9.761305805684708e-06,
      "loss": 0.0018,
      "step": 34630
    },
    {
      "epoch": 0.566800294526712,
      "grad_norm": 0.31754690408706665,
      "learning_rate": 9.761015106756192e-06,
      "loss": 0.0053,
      "step": 34640
    },
    {
      "epoch": 0.5669639204777878,
      "grad_norm": 0.10781533271074295,
      "learning_rate": 9.76072423525232e-06,
      "loss": 0.003,
      "step": 34650
    },
    {
      "epoch": 0.5671275464288636,
      "grad_norm": 0.07485975325107574,
      "learning_rate": 9.760433191183636e-06,
      "loss": 0.0037,
      "step": 34660
    },
    {
      "epoch": 0.5672911723799394,
      "grad_norm": 0.36674466729164124,
      "learning_rate": 9.760141974560692e-06,
      "loss": 0.0055,
      "step": 34670
    },
    {
      "epoch": 0.5674547983310153,
      "grad_norm": 0.2253779172897339,
      "learning_rate": 9.75985058539404e-06,
      "loss": 0.0047,
      "step": 34680
    },
    {
      "epoch": 0.5676184242820912,
      "grad_norm": 0.21675962209701538,
      "learning_rate": 9.759559023694246e-06,
      "loss": 0.0039,
      "step": 34690
    },
    {
      "epoch": 0.567782050233167,
      "grad_norm": 0.23162716627120972,
      "learning_rate": 9.759267289471876e-06,
      "loss": 0.0032,
      "step": 34700
    },
    {
      "epoch": 0.5679456761842429,
      "grad_norm": 0.19693949818611145,
      "learning_rate": 9.758975382737505e-06,
      "loss": 0.0035,
      "step": 34710
    },
    {
      "epoch": 0.5681093021353186,
      "grad_norm": 0.3346758186817169,
      "learning_rate": 9.758683303501714e-06,
      "loss": 0.0032,
      "step": 34720
    },
    {
      "epoch": 0.5682729280863945,
      "grad_norm": 0.18781304359436035,
      "learning_rate": 9.758391051775092e-06,
      "loss": 0.0029,
      "step": 34730
    },
    {
      "epoch": 0.5684365540374703,
      "grad_norm": 0.08256262540817261,
      "learning_rate": 9.75809862756823e-06,
      "loss": 0.0045,
      "step": 34740
    },
    {
      "epoch": 0.5686001799885462,
      "grad_norm": 0.26031774282455444,
      "learning_rate": 9.757806030891728e-06,
      "loss": 0.0026,
      "step": 34750
    },
    {
      "epoch": 0.5687638059396221,
      "grad_norm": 0.18769051134586334,
      "learning_rate": 9.757513261756193e-06,
      "loss": 0.0037,
      "step": 34760
    },
    {
      "epoch": 0.5689274318906978,
      "grad_norm": 0.16542164981365204,
      "learning_rate": 9.757220320172237e-06,
      "loss": 0.0054,
      "step": 34770
    },
    {
      "epoch": 0.5690910578417737,
      "grad_norm": 0.2837505638599396,
      "learning_rate": 9.756927206150477e-06,
      "loss": 0.0061,
      "step": 34780
    },
    {
      "epoch": 0.5692546837928495,
      "grad_norm": 0.22357076406478882,
      "learning_rate": 9.75663391970154e-06,
      "loss": 0.0055,
      "step": 34790
    },
    {
      "epoch": 0.5694183097439254,
      "grad_norm": 0.2976042330265045,
      "learning_rate": 9.756340460836055e-06,
      "loss": 0.0052,
      "step": 34800
    },
    {
      "epoch": 0.5695819356950013,
      "grad_norm": 0.5176551342010498,
      "learning_rate": 9.75604682956466e-06,
      "loss": 0.0049,
      "step": 34810
    },
    {
      "epoch": 0.569745561646077,
      "grad_norm": 0.22788123786449432,
      "learning_rate": 9.755753025897999e-06,
      "loss": 0.0041,
      "step": 34820
    },
    {
      "epoch": 0.5699091875971529,
      "grad_norm": 0.3112598955631256,
      "learning_rate": 9.75545904984672e-06,
      "loss": 0.0047,
      "step": 34830
    },
    {
      "epoch": 0.5700728135482287,
      "grad_norm": 0.30748066306114197,
      "learning_rate": 9.75516490142148e-06,
      "loss": 0.0043,
      "step": 34840
    },
    {
      "epoch": 0.5702364394993046,
      "grad_norm": 0.27794697880744934,
      "learning_rate": 9.75487058063294e-06,
      "loss": 0.0026,
      "step": 34850
    },
    {
      "epoch": 0.5704000654503805,
      "grad_norm": 0.23995459079742432,
      "learning_rate": 9.754576087491772e-06,
      "loss": 0.0026,
      "step": 34860
    },
    {
      "epoch": 0.5705636914014562,
      "grad_norm": 0.37012141942977905,
      "learning_rate": 9.754281422008647e-06,
      "loss": 0.0069,
      "step": 34870
    },
    {
      "epoch": 0.5707273173525321,
      "grad_norm": 0.20414775609970093,
      "learning_rate": 9.753986584194246e-06,
      "loss": 0.0055,
      "step": 34880
    },
    {
      "epoch": 0.5708909433036079,
      "grad_norm": 0.17128324508666992,
      "learning_rate": 9.753691574059258e-06,
      "loss": 0.0041,
      "step": 34890
    },
    {
      "epoch": 0.5710545692546838,
      "grad_norm": 0.2541649639606476,
      "learning_rate": 9.753396391614377e-06,
      "loss": 0.0062,
      "step": 34900
    },
    {
      "epoch": 0.5712181952057597,
      "grad_norm": 0.04486648738384247,
      "learning_rate": 9.753101036870298e-06,
      "loss": 0.0046,
      "step": 34910
    },
    {
      "epoch": 0.5713818211568354,
      "grad_norm": 0.14260824024677277,
      "learning_rate": 9.752805509837733e-06,
      "loss": 0.0022,
      "step": 34920
    },
    {
      "epoch": 0.5715454471079113,
      "grad_norm": 0.2230953574180603,
      "learning_rate": 9.75250981052739e-06,
      "loss": 0.004,
      "step": 34930
    },
    {
      "epoch": 0.5717090730589871,
      "grad_norm": 0.49457302689552307,
      "learning_rate": 9.75221393894999e-06,
      "loss": 0.006,
      "step": 34940
    },
    {
      "epoch": 0.571872699010063,
      "grad_norm": 0.23775379359722137,
      "learning_rate": 9.751917895116255e-06,
      "loss": 0.0025,
      "step": 34950
    },
    {
      "epoch": 0.5720363249611389,
      "grad_norm": 0.14535000920295715,
      "learning_rate": 9.751621679036917e-06,
      "loss": 0.0041,
      "step": 34960
    },
    {
      "epoch": 0.5721999509122146,
      "grad_norm": 0.17630788683891296,
      "learning_rate": 9.751325290722715e-06,
      "loss": 0.0052,
      "step": 34970
    },
    {
      "epoch": 0.5723635768632905,
      "grad_norm": 0.1856747269630432,
      "learning_rate": 9.751028730184392e-06,
      "loss": 0.0035,
      "step": 34980
    },
    {
      "epoch": 0.5725272028143663,
      "grad_norm": 0.1989014744758606,
      "learning_rate": 9.750731997432691e-06,
      "loss": 0.0034,
      "step": 34990
    },
    {
      "epoch": 0.5726908287654422,
      "grad_norm": 0.269896924495697,
      "learning_rate": 9.75043509247838e-06,
      "loss": 0.0061,
      "step": 35000
    },
    {
      "epoch": 0.5728544547165181,
      "grad_norm": 0.09226461499929428,
      "learning_rate": 9.750138015332208e-06,
      "loss": 0.0036,
      "step": 35010
    },
    {
      "epoch": 0.5730180806675939,
      "grad_norm": 0.2072274088859558,
      "learning_rate": 9.749840766004953e-06,
      "loss": 0.0049,
      "step": 35020
    },
    {
      "epoch": 0.5731817066186697,
      "grad_norm": 0.013144253753125668,
      "learning_rate": 9.749543344507385e-06,
      "loss": 0.0029,
      "step": 35030
    },
    {
      "epoch": 0.5733453325697455,
      "grad_norm": 0.1586284339427948,
      "learning_rate": 9.749245750850287e-06,
      "loss": 0.0031,
      "step": 35040
    },
    {
      "epoch": 0.5735089585208214,
      "grad_norm": 0.11746500432491302,
      "learning_rate": 9.748947985044443e-06,
      "loss": 0.0037,
      "step": 35050
    },
    {
      "epoch": 0.5736725844718973,
      "grad_norm": 0.22508607804775238,
      "learning_rate": 9.748650047100651e-06,
      "loss": 0.0034,
      "step": 35060
    },
    {
      "epoch": 0.5738362104229731,
      "grad_norm": 0.2871067523956299,
      "learning_rate": 9.748351937029707e-06,
      "loss": 0.0052,
      "step": 35070
    },
    {
      "epoch": 0.5739998363740489,
      "grad_norm": 0.03430821746587753,
      "learning_rate": 9.748053654842417e-06,
      "loss": 0.003,
      "step": 35080
    },
    {
      "epoch": 0.5741634623251247,
      "grad_norm": 0.35699984431266785,
      "learning_rate": 9.747755200549594e-06,
      "loss": 0.0056,
      "step": 35090
    },
    {
      "epoch": 0.5743270882762006,
      "grad_norm": 0.18932944536209106,
      "learning_rate": 9.747456574162055e-06,
      "loss": 0.0048,
      "step": 35100
    },
    {
      "epoch": 0.5744907142272765,
      "grad_norm": 0.14558500051498413,
      "learning_rate": 9.747157775690628e-06,
      "loss": 0.004,
      "step": 35110
    },
    {
      "epoch": 0.5746543401783523,
      "grad_norm": 0.2817893326282501,
      "learning_rate": 9.74685880514614e-06,
      "loss": 0.004,
      "step": 35120
    },
    {
      "epoch": 0.5748179661294281,
      "grad_norm": 0.4181196987628937,
      "learning_rate": 9.746559662539427e-06,
      "loss": 0.0029,
      "step": 35130
    },
    {
      "epoch": 0.5749815920805039,
      "grad_norm": 0.5082183480262756,
      "learning_rate": 9.746260347881335e-06,
      "loss": 0.0035,
      "step": 35140
    },
    {
      "epoch": 0.5751452180315798,
      "grad_norm": 0.3064762055873871,
      "learning_rate": 9.745960861182716e-06,
      "loss": 0.0034,
      "step": 35150
    },
    {
      "epoch": 0.5753088439826557,
      "grad_norm": 0.514564037322998,
      "learning_rate": 9.745661202454419e-06,
      "loss": 0.0035,
      "step": 35160
    },
    {
      "epoch": 0.5754724699337315,
      "grad_norm": 0.5590274333953857,
      "learning_rate": 9.745361371707309e-06,
      "loss": 0.0041,
      "step": 35170
    },
    {
      "epoch": 0.5756360958848074,
      "grad_norm": 0.17944744229316711,
      "learning_rate": 9.745061368952257e-06,
      "loss": 0.002,
      "step": 35180
    },
    {
      "epoch": 0.5757997218358831,
      "grad_norm": 0.26431846618652344,
      "learning_rate": 9.744761194200133e-06,
      "loss": 0.007,
      "step": 35190
    },
    {
      "epoch": 0.575963347786959,
      "grad_norm": 0.22643274068832397,
      "learning_rate": 9.74446084746182e-06,
      "loss": 0.004,
      "step": 35200
    },
    {
      "epoch": 0.5761269737380349,
      "grad_norm": 0.431440144777298,
      "learning_rate": 9.744160328748204e-06,
      "loss": 0.0069,
      "step": 35210
    },
    {
      "epoch": 0.5762905996891107,
      "grad_norm": 0.16164946556091309,
      "learning_rate": 9.74385963807018e-06,
      "loss": 0.0039,
      "step": 35220
    },
    {
      "epoch": 0.5764542256401866,
      "grad_norm": 0.14265932142734528,
      "learning_rate": 9.743558775438644e-06,
      "loss": 0.0061,
      "step": 35230
    },
    {
      "epoch": 0.5766178515912623,
      "grad_norm": 0.17127510905265808,
      "learning_rate": 9.743257740864502e-06,
      "loss": 0.0039,
      "step": 35240
    },
    {
      "epoch": 0.5767814775423382,
      "grad_norm": 0.10949498414993286,
      "learning_rate": 9.742956534358669e-06,
      "loss": 0.0041,
      "step": 35250
    },
    {
      "epoch": 0.5769451034934141,
      "grad_norm": 0.3070734441280365,
      "learning_rate": 9.742655155932061e-06,
      "loss": 0.0027,
      "step": 35260
    },
    {
      "epoch": 0.5771087294444899,
      "grad_norm": 0.408394455909729,
      "learning_rate": 9.742353605595602e-06,
      "loss": 0.0028,
      "step": 35270
    },
    {
      "epoch": 0.5772723553955658,
      "grad_norm": 0.391413152217865,
      "learning_rate": 9.742051883360222e-06,
      "loss": 0.0031,
      "step": 35280
    },
    {
      "epoch": 0.5774359813466415,
      "grad_norm": 0.015228900127112865,
      "learning_rate": 9.74174998923686e-06,
      "loss": 0.0034,
      "step": 35290
    },
    {
      "epoch": 0.5775996072977174,
      "grad_norm": 0.36725103855133057,
      "learning_rate": 9.741447923236456e-06,
      "loss": 0.0023,
      "step": 35300
    },
    {
      "epoch": 0.5777632332487933,
      "grad_norm": 0.3825167417526245,
      "learning_rate": 9.741145685369962e-06,
      "loss": 0.004,
      "step": 35310
    },
    {
      "epoch": 0.5779268591998691,
      "grad_norm": 0.41490596532821655,
      "learning_rate": 9.74084327564833e-06,
      "loss": 0.0045,
      "step": 35320
    },
    {
      "epoch": 0.578090485150945,
      "grad_norm": 0.3690876066684723,
      "learning_rate": 9.740540694082524e-06,
      "loss": 0.0044,
      "step": 35330
    },
    {
      "epoch": 0.5782541111020207,
      "grad_norm": 0.19393190741539001,
      "learning_rate": 9.740237940683512e-06,
      "loss": 0.0053,
      "step": 35340
    },
    {
      "epoch": 0.5784177370530966,
      "grad_norm": 0.08227064460515976,
      "learning_rate": 9.739935015462267e-06,
      "loss": 0.0051,
      "step": 35350
    },
    {
      "epoch": 0.5785813630041725,
      "grad_norm": 0.33437106013298035,
      "learning_rate": 9.73963191842977e-06,
      "loss": 0.0036,
      "step": 35360
    },
    {
      "epoch": 0.5787449889552483,
      "grad_norm": 0.3459462821483612,
      "learning_rate": 9.73932864959701e-06,
      "loss": 0.005,
      "step": 35370
    },
    {
      "epoch": 0.5789086149063242,
      "grad_norm": 0.5705816149711609,
      "learning_rate": 9.739025208974974e-06,
      "loss": 0.0049,
      "step": 35380
    },
    {
      "epoch": 0.5790722408574,
      "grad_norm": 0.1507769674062729,
      "learning_rate": 9.738721596574666e-06,
      "loss": 0.0055,
      "step": 35390
    },
    {
      "epoch": 0.5792358668084758,
      "grad_norm": 0.33719950914382935,
      "learning_rate": 9.738417812407085e-06,
      "loss": 0.0065,
      "step": 35400
    },
    {
      "epoch": 0.5793994927595517,
      "grad_norm": 0.2390710860490799,
      "learning_rate": 9.738113856483251e-06,
      "loss": 0.0062,
      "step": 35410
    },
    {
      "epoch": 0.5795631187106275,
      "grad_norm": 0.027272971346974373,
      "learning_rate": 9.737809728814178e-06,
      "loss": 0.0042,
      "step": 35420
    },
    {
      "epoch": 0.5797267446617034,
      "grad_norm": 0.3524695038795471,
      "learning_rate": 9.737505429410888e-06,
      "loss": 0.0032,
      "step": 35430
    },
    {
      "epoch": 0.5798903706127791,
      "grad_norm": 0.2170034945011139,
      "learning_rate": 9.737200958284412e-06,
      "loss": 0.0048,
      "step": 35440
    },
    {
      "epoch": 0.580053996563855,
      "grad_norm": 0.38364577293395996,
      "learning_rate": 9.736896315445787e-06,
      "loss": 0.0043,
      "step": 35450
    },
    {
      "epoch": 0.5802176225149309,
      "grad_norm": 0.28469038009643555,
      "learning_rate": 9.736591500906055e-06,
      "loss": 0.0039,
      "step": 35460
    },
    {
      "epoch": 0.5803812484660067,
      "grad_norm": 0.09250683337450027,
      "learning_rate": 9.736286514676265e-06,
      "loss": 0.0042,
      "step": 35470
    },
    {
      "epoch": 0.5805448744170826,
      "grad_norm": 0.1815451979637146,
      "learning_rate": 9.735981356767475e-06,
      "loss": 0.0039,
      "step": 35480
    },
    {
      "epoch": 0.5807085003681584,
      "grad_norm": 0.28491681814193726,
      "learning_rate": 9.73567602719074e-06,
      "loss": 0.0039,
      "step": 35490
    },
    {
      "epoch": 0.5808721263192342,
      "grad_norm": 0.10905561596155167,
      "learning_rate": 9.735370525957133e-06,
      "loss": 0.0028,
      "step": 35500
    },
    {
      "epoch": 0.5810357522703101,
      "grad_norm": 0.06161411479115486,
      "learning_rate": 9.735064853077724e-06,
      "loss": 0.0033,
      "step": 35510
    },
    {
      "epoch": 0.5811993782213859,
      "grad_norm": 0.21105897426605225,
      "learning_rate": 9.734759008563596e-06,
      "loss": 0.0041,
      "step": 35520
    },
    {
      "epoch": 0.5813630041724618,
      "grad_norm": 0.058447372168302536,
      "learning_rate": 9.734452992425834e-06,
      "loss": 0.0042,
      "step": 35530
    },
    {
      "epoch": 0.5815266301235376,
      "grad_norm": 0.21096155047416687,
      "learning_rate": 9.734146804675528e-06,
      "loss": 0.0035,
      "step": 35540
    },
    {
      "epoch": 0.5816902560746134,
      "grad_norm": 0.09213443100452423,
      "learning_rate": 9.733840445323778e-06,
      "loss": 0.0037,
      "step": 35550
    },
    {
      "epoch": 0.5818538820256893,
      "grad_norm": 0.20286403596401215,
      "learning_rate": 9.733533914381692e-06,
      "loss": 0.0032,
      "step": 35560
    },
    {
      "epoch": 0.5820175079767651,
      "grad_norm": 0.0792188048362732,
      "learning_rate": 9.733227211860378e-06,
      "loss": 0.0044,
      "step": 35570
    },
    {
      "epoch": 0.582181133927841,
      "grad_norm": 0.09005182236433029,
      "learning_rate": 9.732920337770953e-06,
      "loss": 0.005,
      "step": 35580
    },
    {
      "epoch": 0.5823447598789168,
      "grad_norm": 0.18099474906921387,
      "learning_rate": 9.73261329212454e-06,
      "loss": 0.0037,
      "step": 35590
    },
    {
      "epoch": 0.5825083858299926,
      "grad_norm": 0.23028463125228882,
      "learning_rate": 9.732306074932271e-06,
      "loss": 0.0058,
      "step": 35600
    },
    {
      "epoch": 0.5826720117810685,
      "grad_norm": 0.1142086535692215,
      "learning_rate": 9.731998686205279e-06,
      "loss": 0.0042,
      "step": 35610
    },
    {
      "epoch": 0.5828356377321443,
      "grad_norm": 0.26584315299987793,
      "learning_rate": 9.73169112595471e-06,
      "loss": 0.0041,
      "step": 35620
    },
    {
      "epoch": 0.5829992636832202,
      "grad_norm": 0.19755353033542633,
      "learning_rate": 9.731383394191708e-06,
      "loss": 0.0062,
      "step": 35630
    },
    {
      "epoch": 0.583162889634296,
      "grad_norm": 0.49182742834091187,
      "learning_rate": 9.73107549092743e-06,
      "loss": 0.0057,
      "step": 35640
    },
    {
      "epoch": 0.5833265155853719,
      "grad_norm": 0.272129088640213,
      "learning_rate": 9.730767416173038e-06,
      "loss": 0.0051,
      "step": 35650
    },
    {
      "epoch": 0.5834901415364476,
      "grad_norm": 0.3390112519264221,
      "learning_rate": 9.730459169939697e-06,
      "loss": 0.0051,
      "step": 35660
    },
    {
      "epoch": 0.5836537674875235,
      "grad_norm": 0.048624761402606964,
      "learning_rate": 9.730150752238578e-06,
      "loss": 0.0032,
      "step": 35670
    },
    {
      "epoch": 0.5838173934385994,
      "grad_norm": 0.1803896576166153,
      "learning_rate": 9.729842163080866e-06,
      "loss": 0.003,
      "step": 35680
    },
    {
      "epoch": 0.5839810193896752,
      "grad_norm": 0.21148480474948883,
      "learning_rate": 9.72953340247774e-06,
      "loss": 0.004,
      "step": 35690
    },
    {
      "epoch": 0.5841446453407511,
      "grad_norm": 0.08211598545312881,
      "learning_rate": 9.729224470440397e-06,
      "loss": 0.0014,
      "step": 35700
    },
    {
      "epoch": 0.5843082712918268,
      "grad_norm": 0.24902889132499695,
      "learning_rate": 9.728915366980033e-06,
      "loss": 0.0048,
      "step": 35710
    },
    {
      "epoch": 0.5844718972429027,
      "grad_norm": 0.4118281304836273,
      "learning_rate": 9.728606092107854e-06,
      "loss": 0.0039,
      "step": 35720
    },
    {
      "epoch": 0.5846355231939786,
      "grad_norm": 0.1850242018699646,
      "learning_rate": 9.728296645835069e-06,
      "loss": 0.0036,
      "step": 35730
    },
    {
      "epoch": 0.5847991491450544,
      "grad_norm": 0.10127825289964676,
      "learning_rate": 9.727987028172892e-06,
      "loss": 0.0038,
      "step": 35740
    },
    {
      "epoch": 0.5849627750961303,
      "grad_norm": 0.24307718873023987,
      "learning_rate": 9.727677239132552e-06,
      "loss": 0.0039,
      "step": 35750
    },
    {
      "epoch": 0.585126401047206,
      "grad_norm": 0.32504841685295105,
      "learning_rate": 9.727367278725272e-06,
      "loss": 0.0091,
      "step": 35760
    },
    {
      "epoch": 0.5852900269982819,
      "grad_norm": 0.07922003418207169,
      "learning_rate": 9.727057146962291e-06,
      "loss": 0.0052,
      "step": 35770
    },
    {
      "epoch": 0.5854536529493578,
      "grad_norm": 0.39402493834495544,
      "learning_rate": 9.726746843854851e-06,
      "loss": 0.0035,
      "step": 35780
    },
    {
      "epoch": 0.5856172789004336,
      "grad_norm": 0.1367437243461609,
      "learning_rate": 9.726436369414196e-06,
      "loss": 0.0037,
      "step": 35790
    },
    {
      "epoch": 0.5857809048515095,
      "grad_norm": 0.15938787162303925,
      "learning_rate": 9.726125723651584e-06,
      "loss": 0.0046,
      "step": 35800
    },
    {
      "epoch": 0.5859445308025852,
      "grad_norm": 0.12762023508548737,
      "learning_rate": 9.725814906578272e-06,
      "loss": 0.0032,
      "step": 35810
    },
    {
      "epoch": 0.5861081567536611,
      "grad_norm": 0.2957753837108612,
      "learning_rate": 9.72550391820553e-06,
      "loss": 0.0041,
      "step": 35820
    },
    {
      "epoch": 0.586271782704737,
      "grad_norm": 0.34607669711112976,
      "learning_rate": 9.725192758544628e-06,
      "loss": 0.0026,
      "step": 35830
    },
    {
      "epoch": 0.5864354086558128,
      "grad_norm": 0.3427541255950928,
      "learning_rate": 9.724881427606842e-06,
      "loss": 0.0033,
      "step": 35840
    },
    {
      "epoch": 0.5865990346068887,
      "grad_norm": 0.1023557260632515,
      "learning_rate": 9.724569925403465e-06,
      "loss": 0.003,
      "step": 35850
    },
    {
      "epoch": 0.5867626605579644,
      "grad_norm": 0.1070786640048027,
      "learning_rate": 9.724258251945779e-06,
      "loss": 0.0067,
      "step": 35860
    },
    {
      "epoch": 0.5869262865090403,
      "grad_norm": 0.1273365169763565,
      "learning_rate": 9.723946407245089e-06,
      "loss": 0.0028,
      "step": 35870
    },
    {
      "epoch": 0.5870899124601162,
      "grad_norm": 0.18796522915363312,
      "learning_rate": 9.723634391312694e-06,
      "loss": 0.0034,
      "step": 35880
    },
    {
      "epoch": 0.587253538411192,
      "grad_norm": 0.1727188676595688,
      "learning_rate": 9.723322204159906e-06,
      "loss": 0.0049,
      "step": 35890
    },
    {
      "epoch": 0.5874171643622679,
      "grad_norm": 0.013268476352095604,
      "learning_rate": 9.723009845798039e-06,
      "loss": 0.0032,
      "step": 35900
    },
    {
      "epoch": 0.5875807903133436,
      "grad_norm": 0.15942905843257904,
      "learning_rate": 9.722697316238419e-06,
      "loss": 0.0027,
      "step": 35910
    },
    {
      "epoch": 0.5877444162644195,
      "grad_norm": 0.258378267288208,
      "learning_rate": 9.72238461549237e-06,
      "loss": 0.0043,
      "step": 35920
    },
    {
      "epoch": 0.5879080422154954,
      "grad_norm": 0.27693480253219604,
      "learning_rate": 9.722071743571229e-06,
      "loss": 0.0028,
      "step": 35930
    },
    {
      "epoch": 0.5880716681665712,
      "grad_norm": 0.3835853338241577,
      "learning_rate": 9.721758700486336e-06,
      "loss": 0.0043,
      "step": 35940
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.16655398905277252,
      "learning_rate": 9.72144548624904e-06,
      "loss": 0.0033,
      "step": 35950
    },
    {
      "epoch": 0.5883989200687229,
      "grad_norm": 0.15240994095802307,
      "learning_rate": 9.721132100870692e-06,
      "loss": 0.003,
      "step": 35960
    },
    {
      "epoch": 0.5885625460197987,
      "grad_norm": 0.08357996493577957,
      "learning_rate": 9.720818544362652e-06,
      "loss": 0.0037,
      "step": 35970
    },
    {
      "epoch": 0.5887261719708746,
      "grad_norm": 0.2652566730976105,
      "learning_rate": 9.720504816736285e-06,
      "loss": 0.0043,
      "step": 35980
    },
    {
      "epoch": 0.5888897979219504,
      "grad_norm": 0.1606847494840622,
      "learning_rate": 9.720190918002965e-06,
      "loss": 0.0046,
      "step": 35990
    },
    {
      "epoch": 0.5890534238730263,
      "grad_norm": 0.0774414911866188,
      "learning_rate": 9.71987684817407e-06,
      "loss": 0.0029,
      "step": 36000
    },
    {
      "epoch": 0.5892170498241021,
      "grad_norm": 0.31056180596351624,
      "learning_rate": 9.719562607260982e-06,
      "loss": 0.0043,
      "step": 36010
    },
    {
      "epoch": 0.5893806757751779,
      "grad_norm": 0.15677975118160248,
      "learning_rate": 9.719248195275092e-06,
      "loss": 0.0029,
      "step": 36020
    },
    {
      "epoch": 0.5895443017262538,
      "grad_norm": 0.23312224447727203,
      "learning_rate": 9.718933612227799e-06,
      "loss": 0.0049,
      "step": 36030
    },
    {
      "epoch": 0.5897079276773296,
      "grad_norm": 0.15039682388305664,
      "learning_rate": 9.718618858130502e-06,
      "loss": 0.0043,
      "step": 36040
    },
    {
      "epoch": 0.5898715536284055,
      "grad_norm": 0.2911875545978546,
      "learning_rate": 9.718303932994614e-06,
      "loss": 0.0042,
      "step": 36050
    },
    {
      "epoch": 0.5900351795794813,
      "grad_norm": 0.3128955364227295,
      "learning_rate": 9.717988836831548e-06,
      "loss": 0.0041,
      "step": 36060
    },
    {
      "epoch": 0.5901988055305571,
      "grad_norm": 0.211540088057518,
      "learning_rate": 9.717673569652726e-06,
      "loss": 0.0034,
      "step": 36070
    },
    {
      "epoch": 0.590362431481633,
      "grad_norm": 0.11134030669927597,
      "learning_rate": 9.717358131469576e-06,
      "loss": 0.0042,
      "step": 36080
    },
    {
      "epoch": 0.5905260574327088,
      "grad_norm": 0.21403639018535614,
      "learning_rate": 9.717042522293531e-06,
      "loss": 0.002,
      "step": 36090
    },
    {
      "epoch": 0.5906896833837847,
      "grad_norm": 0.6550421118736267,
      "learning_rate": 9.716726742136034e-06,
      "loss": 0.003,
      "step": 36100
    },
    {
      "epoch": 0.5908533093348605,
      "grad_norm": 0.2903192341327667,
      "learning_rate": 9.716410791008526e-06,
      "loss": 0.0043,
      "step": 36110
    },
    {
      "epoch": 0.5910169352859364,
      "grad_norm": 0.18604125082492828,
      "learning_rate": 9.716094668922464e-06,
      "loss": 0.0032,
      "step": 36120
    },
    {
      "epoch": 0.5911805612370122,
      "grad_norm": 0.08308400958776474,
      "learning_rate": 9.715778375889304e-06,
      "loss": 0.0027,
      "step": 36130
    },
    {
      "epoch": 0.591344187188088,
      "grad_norm": 0.26171454787254333,
      "learning_rate": 9.715461911920513e-06,
      "loss": 0.0049,
      "step": 36140
    },
    {
      "epoch": 0.5915078131391639,
      "grad_norm": 0.09076100587844849,
      "learning_rate": 9.71514527702756e-06,
      "loss": 0.0024,
      "step": 36150
    },
    {
      "epoch": 0.5916714390902397,
      "grad_norm": 0.14372113347053528,
      "learning_rate": 9.714828471221924e-06,
      "loss": 0.0039,
      "step": 36160
    },
    {
      "epoch": 0.5918350650413156,
      "grad_norm": 0.11785857379436493,
      "learning_rate": 9.714511494515089e-06,
      "loss": 0.004,
      "step": 36170
    },
    {
      "epoch": 0.5919986909923914,
      "grad_norm": 0.2088654637336731,
      "learning_rate": 9.714194346918542e-06,
      "loss": 0.0062,
      "step": 36180
    },
    {
      "epoch": 0.5921623169434672,
      "grad_norm": 0.2550960183143616,
      "learning_rate": 9.713877028443779e-06,
      "loss": 0.0043,
      "step": 36190
    },
    {
      "epoch": 0.5923259428945431,
      "grad_norm": 0.058798037469387054,
      "learning_rate": 9.713559539102304e-06,
      "loss": 0.003,
      "step": 36200
    },
    {
      "epoch": 0.5924895688456189,
      "grad_norm": 0.25456178188323975,
      "learning_rate": 9.713241878905626e-06,
      "loss": 0.0056,
      "step": 36210
    },
    {
      "epoch": 0.5926531947966948,
      "grad_norm": 0.29534563422203064,
      "learning_rate": 9.712924047865255e-06,
      "loss": 0.0052,
      "step": 36220
    },
    {
      "epoch": 0.5928168207477706,
      "grad_norm": 0.15518364310264587,
      "learning_rate": 9.712606045992717e-06,
      "loss": 0.0039,
      "step": 36230
    },
    {
      "epoch": 0.5929804466988464,
      "grad_norm": 0.08899492770433426,
      "learning_rate": 9.712287873299535e-06,
      "loss": 0.0066,
      "step": 36240
    },
    {
      "epoch": 0.5931440726499223,
      "grad_norm": 0.1559165120124817,
      "learning_rate": 9.711969529797245e-06,
      "loss": 0.0045,
      "step": 36250
    },
    {
      "epoch": 0.5933076986009981,
      "grad_norm": 0.22468434274196625,
      "learning_rate": 9.711651015497382e-06,
      "loss": 0.0032,
      "step": 36260
    },
    {
      "epoch": 0.593471324552074,
      "grad_norm": 0.15269573032855988,
      "learning_rate": 9.711332330411495e-06,
      "loss": 0.0046,
      "step": 36270
    },
    {
      "epoch": 0.5936349505031498,
      "grad_norm": 0.336885005235672,
      "learning_rate": 9.711013474551135e-06,
      "loss": 0.0031,
      "step": 36280
    },
    {
      "epoch": 0.5937985764542256,
      "grad_norm": 0.22085560858249664,
      "learning_rate": 9.710694447927858e-06,
      "loss": 0.0031,
      "step": 36290
    },
    {
      "epoch": 0.5939622024053015,
      "grad_norm": 0.16883236169815063,
      "learning_rate": 9.71037525055323e-06,
      "loss": 0.0035,
      "step": 36300
    },
    {
      "epoch": 0.5941258283563773,
      "grad_norm": 0.9199559092521667,
      "learning_rate": 9.71005588243882e-06,
      "loss": 0.006,
      "step": 36310
    },
    {
      "epoch": 0.5942894543074532,
      "grad_norm": 0.09754136204719543,
      "learning_rate": 9.709736343596206e-06,
      "loss": 0.0033,
      "step": 36320
    },
    {
      "epoch": 0.594453080258529,
      "grad_norm": 0.0830918550491333,
      "learning_rate": 9.709416634036966e-06,
      "loss": 0.0053,
      "step": 36330
    },
    {
      "epoch": 0.5946167062096048,
      "grad_norm": 0.428536593914032,
      "learning_rate": 9.709096753772694e-06,
      "loss": 0.0055,
      "step": 36340
    },
    {
      "epoch": 0.5947803321606807,
      "grad_norm": 0.12406786531209946,
      "learning_rate": 9.708776702814982e-06,
      "loss": 0.0032,
      "step": 36350
    },
    {
      "epoch": 0.5949439581117565,
      "grad_norm": 0.34767746925354004,
      "learning_rate": 9.708456481175433e-06,
      "loss": 0.0036,
      "step": 36360
    },
    {
      "epoch": 0.5951075840628324,
      "grad_norm": 0.3666210472583771,
      "learning_rate": 9.708136088865651e-06,
      "loss": 0.0048,
      "step": 36370
    },
    {
      "epoch": 0.5952712100139083,
      "grad_norm": 0.1579841524362564,
      "learning_rate": 9.707815525897253e-06,
      "loss": 0.0031,
      "step": 36380
    },
    {
      "epoch": 0.595434835964984,
      "grad_norm": 0.19056245684623718,
      "learning_rate": 9.707494792281856e-06,
      "loss": 0.0034,
      "step": 36390
    },
    {
      "epoch": 0.5955984619160599,
      "grad_norm": 0.5286058783531189,
      "learning_rate": 9.707173888031087e-06,
      "loss": 0.0054,
      "step": 36400
    },
    {
      "epoch": 0.5957620878671357,
      "grad_norm": 0.4452703297138214,
      "learning_rate": 9.70685281315658e-06,
      "loss": 0.0046,
      "step": 36410
    },
    {
      "epoch": 0.5959257138182116,
      "grad_norm": 0.2830319106578827,
      "learning_rate": 9.70653156766997e-06,
      "loss": 0.0042,
      "step": 36420
    },
    {
      "epoch": 0.5960893397692875,
      "grad_norm": 0.07279734313488007,
      "learning_rate": 9.706210151582901e-06,
      "loss": 0.0055,
      "step": 36430
    },
    {
      "epoch": 0.5962529657203632,
      "grad_norm": 0.5065034031867981,
      "learning_rate": 9.705888564907027e-06,
      "loss": 0.0065,
      "step": 36440
    },
    {
      "epoch": 0.5964165916714391,
      "grad_norm": 0.09429576992988586,
      "learning_rate": 9.705566807654e-06,
      "loss": 0.0052,
      "step": 36450
    },
    {
      "epoch": 0.5965802176225149,
      "grad_norm": 0.14021041989326477,
      "learning_rate": 9.70524487983549e-06,
      "loss": 0.0022,
      "step": 36460
    },
    {
      "epoch": 0.5967438435735908,
      "grad_norm": 0.13279235363006592,
      "learning_rate": 9.704922781463159e-06,
      "loss": 0.0023,
      "step": 36470
    },
    {
      "epoch": 0.5969074695246667,
      "grad_norm": 0.09888876229524612,
      "learning_rate": 9.704600512548685e-06,
      "loss": 0.0026,
      "step": 36480
    },
    {
      "epoch": 0.5970710954757424,
      "grad_norm": 0.21202786266803741,
      "learning_rate": 9.70427807310375e-06,
      "loss": 0.0035,
      "step": 36490
    },
    {
      "epoch": 0.5972347214268183,
      "grad_norm": 0.268649160861969,
      "learning_rate": 9.703955463140042e-06,
      "loss": 0.0025,
      "step": 36500
    },
    {
      "epoch": 0.5973983473778941,
      "grad_norm": 0.13659344613552094,
      "learning_rate": 9.703632682669255e-06,
      "loss": 0.0048,
      "step": 36510
    },
    {
      "epoch": 0.59756197332897,
      "grad_norm": 0.1162571832537651,
      "learning_rate": 9.703309731703088e-06,
      "loss": 0.0039,
      "step": 36520
    },
    {
      "epoch": 0.5977255992800458,
      "grad_norm": 0.322752982378006,
      "learning_rate": 9.702986610253244e-06,
      "loss": 0.0033,
      "step": 36530
    },
    {
      "epoch": 0.5978892252311216,
      "grad_norm": 0.06960514187812805,
      "learning_rate": 9.702663318331443e-06,
      "loss": 0.0047,
      "step": 36540
    },
    {
      "epoch": 0.5980528511821975,
      "grad_norm": 0.30746543407440186,
      "learning_rate": 9.702339855949396e-06,
      "loss": 0.0058,
      "step": 36550
    },
    {
      "epoch": 0.5982164771332733,
      "grad_norm": 0.09129039943218231,
      "learning_rate": 9.702016223118832e-06,
      "loss": 0.0043,
      "step": 36560
    },
    {
      "epoch": 0.5983801030843492,
      "grad_norm": 0.2228856384754181,
      "learning_rate": 9.701692419851483e-06,
      "loss": 0.0032,
      "step": 36570
    },
    {
      "epoch": 0.598543729035425,
      "grad_norm": 0.012627189978957176,
      "learning_rate": 9.701368446159081e-06,
      "loss": 0.0046,
      "step": 36580
    },
    {
      "epoch": 0.5987073549865009,
      "grad_norm": 0.12724575400352478,
      "learning_rate": 9.701044302053373e-06,
      "loss": 0.0025,
      "step": 36590
    },
    {
      "epoch": 0.5988709809375767,
      "grad_norm": 0.11833687126636505,
      "learning_rate": 9.700719987546108e-06,
      "loss": 0.0052,
      "step": 36600
    },
    {
      "epoch": 0.5990346068886525,
      "grad_norm": 0.1939002275466919,
      "learning_rate": 9.700395502649042e-06,
      "loss": 0.004,
      "step": 36610
    },
    {
      "epoch": 0.5991982328397284,
      "grad_norm": 0.09875940531492233,
      "learning_rate": 9.700070847373937e-06,
      "loss": 0.0035,
      "step": 36620
    },
    {
      "epoch": 0.5993618587908042,
      "grad_norm": 0.09660714119672775,
      "learning_rate": 9.699746021732559e-06,
      "loss": 0.0033,
      "step": 36630
    },
    {
      "epoch": 0.59952548474188,
      "grad_norm": 0.38685843348503113,
      "learning_rate": 9.69942102573668e-06,
      "loss": 0.0047,
      "step": 36640
    },
    {
      "epoch": 0.5996891106929559,
      "grad_norm": 0.28250786662101746,
      "learning_rate": 9.699095859398088e-06,
      "loss": 0.0027,
      "step": 36650
    },
    {
      "epoch": 0.5998527366440317,
      "grad_norm": 0.39558863639831543,
      "learning_rate": 9.698770522728564e-06,
      "loss": 0.0039,
      "step": 36660
    },
    {
      "epoch": 0.6000163625951076,
      "grad_norm": 0.07542917132377625,
      "learning_rate": 9.698445015739902e-06,
      "loss": 0.0049,
      "step": 36670
    },
    {
      "epoch": 0.6001799885461834,
      "grad_norm": 0.27011069655418396,
      "learning_rate": 9.6981193384439e-06,
      "loss": 0.0043,
      "step": 36680
    },
    {
      "epoch": 0.6003436144972593,
      "grad_norm": 0.2498656064271927,
      "learning_rate": 9.697793490852364e-06,
      "loss": 0.0036,
      "step": 36690
    },
    {
      "epoch": 0.6005072404483351,
      "grad_norm": 0.15047931671142578,
      "learning_rate": 9.697467472977104e-06,
      "loss": 0.0045,
      "step": 36700
    },
    {
      "epoch": 0.6006708663994109,
      "grad_norm": 0.15089407563209534,
      "learning_rate": 9.697141284829939e-06,
      "loss": 0.0049,
      "step": 36710
    },
    {
      "epoch": 0.6008344923504868,
      "grad_norm": 0.1703311949968338,
      "learning_rate": 9.69681492642269e-06,
      "loss": 0.0027,
      "step": 36720
    },
    {
      "epoch": 0.6009981183015626,
      "grad_norm": 0.04145166277885437,
      "learning_rate": 9.69648839776719e-06,
      "loss": 0.0059,
      "step": 36730
    },
    {
      "epoch": 0.6011617442526385,
      "grad_norm": 0.06942036747932434,
      "learning_rate": 9.696161698875274e-06,
      "loss": 0.0027,
      "step": 36740
    },
    {
      "epoch": 0.6013253702037143,
      "grad_norm": 0.1678304374217987,
      "learning_rate": 9.695834829758782e-06,
      "loss": 0.0043,
      "step": 36750
    },
    {
      "epoch": 0.6014889961547901,
      "grad_norm": 0.562107503414154,
      "learning_rate": 9.695507790429564e-06,
      "loss": 0.0048,
      "step": 36760
    },
    {
      "epoch": 0.601652622105866,
      "grad_norm": 0.28181058168411255,
      "learning_rate": 9.695180580899475e-06,
      "loss": 0.0031,
      "step": 36770
    },
    {
      "epoch": 0.6018162480569418,
      "grad_norm": 0.11822933703660965,
      "learning_rate": 9.694853201180372e-06,
      "loss": 0.005,
      "step": 36780
    },
    {
      "epoch": 0.6019798740080177,
      "grad_norm": 0.15481255948543549,
      "learning_rate": 9.694525651284126e-06,
      "loss": 0.0036,
      "step": 36790
    },
    {
      "epoch": 0.6021434999590936,
      "grad_norm": 0.10693569481372833,
      "learning_rate": 9.69419793122261e-06,
      "loss": 0.0029,
      "step": 36800
    },
    {
      "epoch": 0.6023071259101693,
      "grad_norm": 0.6296468377113342,
      "learning_rate": 9.693870041007698e-06,
      "loss": 0.0034,
      "step": 36810
    },
    {
      "epoch": 0.6024707518612452,
      "grad_norm": 0.10660327225923538,
      "learning_rate": 9.693541980651278e-06,
      "loss": 0.0035,
      "step": 36820
    },
    {
      "epoch": 0.602634377812321,
      "grad_norm": 0.44996020197868347,
      "learning_rate": 9.693213750165244e-06,
      "loss": 0.0042,
      "step": 36830
    },
    {
      "epoch": 0.6027980037633969,
      "grad_norm": 0.08215480297803879,
      "learning_rate": 9.69288534956149e-06,
      "loss": 0.0071,
      "step": 36840
    },
    {
      "epoch": 0.6029616297144728,
      "grad_norm": 0.3710929751396179,
      "learning_rate": 9.692556778851922e-06,
      "loss": 0.0023,
      "step": 36850
    },
    {
      "epoch": 0.6031252556655485,
      "grad_norm": 0.41604313254356384,
      "learning_rate": 9.692228038048449e-06,
      "loss": 0.0057,
      "step": 36860
    },
    {
      "epoch": 0.6032888816166244,
      "grad_norm": 0.05612803250551224,
      "learning_rate": 9.691899127162986e-06,
      "loss": 0.0049,
      "step": 36870
    },
    {
      "epoch": 0.6034525075677002,
      "grad_norm": 0.14243543148040771,
      "learning_rate": 9.691570046207458e-06,
      "loss": 0.0049,
      "step": 36880
    },
    {
      "epoch": 0.6036161335187761,
      "grad_norm": 0.1722288578748703,
      "learning_rate": 9.69124079519379e-06,
      "loss": 0.0043,
      "step": 36890
    },
    {
      "epoch": 0.603779759469852,
      "grad_norm": 0.2720753252506256,
      "learning_rate": 9.69091137413392e-06,
      "loss": 0.0053,
      "step": 36900
    },
    {
      "epoch": 0.6039433854209277,
      "grad_norm": 0.12692444026470184,
      "learning_rate": 9.690581783039786e-06,
      "loss": 0.0051,
      "step": 36910
    },
    {
      "epoch": 0.6041070113720036,
      "grad_norm": 0.22577054798603058,
      "learning_rate": 9.690252021923337e-06,
      "loss": 0.0039,
      "step": 36920
    },
    {
      "epoch": 0.6042706373230794,
      "grad_norm": 0.24379587173461914,
      "learning_rate": 9.689922090796525e-06,
      "loss": 0.0039,
      "step": 36930
    },
    {
      "epoch": 0.6044342632741553,
      "grad_norm": 0.19056163728237152,
      "learning_rate": 9.689591989671308e-06,
      "loss": 0.0047,
      "step": 36940
    },
    {
      "epoch": 0.6045978892252312,
      "grad_norm": 0.0722750797867775,
      "learning_rate": 9.689261718559652e-06,
      "loss": 0.0041,
      "step": 36950
    },
    {
      "epoch": 0.6047615151763069,
      "grad_norm": 0.15594319999217987,
      "learning_rate": 9.68893127747353e-06,
      "loss": 0.0027,
      "step": 36960
    },
    {
      "epoch": 0.6049251411273828,
      "grad_norm": 0.1811019480228424,
      "learning_rate": 9.688600666424917e-06,
      "loss": 0.0058,
      "step": 36970
    },
    {
      "epoch": 0.6050887670784586,
      "grad_norm": 0.12612028419971466,
      "learning_rate": 9.6882698854258e-06,
      "loss": 0.0054,
      "step": 36980
    },
    {
      "epoch": 0.6052523930295345,
      "grad_norm": 0.2595139741897583,
      "learning_rate": 9.687938934488169e-06,
      "loss": 0.0028,
      "step": 36990
    },
    {
      "epoch": 0.6054160189806104,
      "grad_norm": 0.19056661427021027,
      "learning_rate": 9.687607813624017e-06,
      "loss": 0.0035,
      "step": 37000
    },
    {
      "epoch": 0.6055796449316861,
      "grad_norm": 0.20385126769542694,
      "learning_rate": 9.687276522845348e-06,
      "loss": 0.0043,
      "step": 37010
    },
    {
      "epoch": 0.605743270882762,
      "grad_norm": 0.38853633403778076,
      "learning_rate": 9.686945062164173e-06,
      "loss": 0.0042,
      "step": 37020
    },
    {
      "epoch": 0.6059068968338378,
      "grad_norm": 0.1379052847623825,
      "learning_rate": 9.686613431592503e-06,
      "loss": 0.0034,
      "step": 37030
    },
    {
      "epoch": 0.6060705227849137,
      "grad_norm": 0.0941978469491005,
      "learning_rate": 9.686281631142357e-06,
      "loss": 0.003,
      "step": 37040
    },
    {
      "epoch": 0.6062341487359896,
      "grad_norm": 0.2398861199617386,
      "learning_rate": 9.685949660825768e-06,
      "loss": 0.0086,
      "step": 37050
    },
    {
      "epoch": 0.6063977746870653,
      "grad_norm": 0.40306493639945984,
      "learning_rate": 9.685617520654767e-06,
      "loss": 0.0062,
      "step": 37060
    },
    {
      "epoch": 0.6065614006381412,
      "grad_norm": 0.18662330508232117,
      "learning_rate": 9.685285210641391e-06,
      "loss": 0.0038,
      "step": 37070
    },
    {
      "epoch": 0.606725026589217,
      "grad_norm": 0.2957414388656616,
      "learning_rate": 9.684952730797687e-06,
      "loss": 0.0048,
      "step": 37080
    },
    {
      "epoch": 0.6068886525402929,
      "grad_norm": 0.06518682837486267,
      "learning_rate": 9.684620081135707e-06,
      "loss": 0.0023,
      "step": 37090
    },
    {
      "epoch": 0.6070522784913688,
      "grad_norm": 0.1331016719341278,
      "learning_rate": 9.684287261667506e-06,
      "loss": 0.0033,
      "step": 37100
    },
    {
      "epoch": 0.6072159044424446,
      "grad_norm": 0.1152452602982521,
      "learning_rate": 9.683954272405154e-06,
      "loss": 0.005,
      "step": 37110
    },
    {
      "epoch": 0.6073795303935204,
      "grad_norm": 0.2517909109592438,
      "learning_rate": 9.683621113360715e-06,
      "loss": 0.0035,
      "step": 37120
    },
    {
      "epoch": 0.6075431563445962,
      "grad_norm": 0.11674592643976212,
      "learning_rate": 9.683287784546267e-06,
      "loss": 0.0034,
      "step": 37130
    },
    {
      "epoch": 0.6077067822956721,
      "grad_norm": 0.3762049674987793,
      "learning_rate": 9.682954285973893e-06,
      "loss": 0.0027,
      "step": 37140
    },
    {
      "epoch": 0.607870408246748,
      "grad_norm": 0.1305396556854248,
      "learning_rate": 9.682620617655682e-06,
      "loss": 0.0057,
      "step": 37150
    },
    {
      "epoch": 0.6080340341978238,
      "grad_norm": 0.13146241009235382,
      "learning_rate": 9.68228677960373e-06,
      "loss": 0.0034,
      "step": 37160
    },
    {
      "epoch": 0.6081976601488996,
      "grad_norm": 0.12051857262849808,
      "learning_rate": 9.681952771830132e-06,
      "loss": 0.0029,
      "step": 37170
    },
    {
      "epoch": 0.6083612860999754,
      "grad_norm": 0.16091465950012207,
      "learning_rate": 9.681618594347e-06,
      "loss": 0.0063,
      "step": 37180
    },
    {
      "epoch": 0.6085249120510513,
      "grad_norm": 0.25111061334609985,
      "learning_rate": 9.681284247166446e-06,
      "loss": 0.0044,
      "step": 37190
    },
    {
      "epoch": 0.6086885380021272,
      "grad_norm": 0.3670584559440613,
      "learning_rate": 9.68094973030059e-06,
      "loss": 0.0032,
      "step": 37200
    },
    {
      "epoch": 0.608852163953203,
      "grad_norm": 0.04304076358675957,
      "learning_rate": 9.680615043761557e-06,
      "loss": 0.0044,
      "step": 37210
    },
    {
      "epoch": 0.6090157899042788,
      "grad_norm": 0.4578183889389038,
      "learning_rate": 9.680280187561477e-06,
      "loss": 0.0101,
      "step": 37220
    },
    {
      "epoch": 0.6091794158553546,
      "grad_norm": 0.32647621631622314,
      "learning_rate": 9.67994516171249e-06,
      "loss": 0.0033,
      "step": 37230
    },
    {
      "epoch": 0.6093430418064305,
      "grad_norm": 0.3882319927215576,
      "learning_rate": 9.679609966226737e-06,
      "loss": 0.0031,
      "step": 37240
    },
    {
      "epoch": 0.6095066677575064,
      "grad_norm": 0.18886899948120117,
      "learning_rate": 9.679274601116372e-06,
      "loss": 0.0033,
      "step": 37250
    },
    {
      "epoch": 0.6096702937085822,
      "grad_norm": 0.2093072384595871,
      "learning_rate": 9.678939066393547e-06,
      "loss": 0.0032,
      "step": 37260
    },
    {
      "epoch": 0.609833919659658,
      "grad_norm": 0.5415998101234436,
      "learning_rate": 9.67860336207043e-06,
      "loss": 0.0061,
      "step": 37270
    },
    {
      "epoch": 0.6099975456107338,
      "grad_norm": 0.07922516763210297,
      "learning_rate": 9.678267488159182e-06,
      "loss": 0.0024,
      "step": 37280
    },
    {
      "epoch": 0.6101611715618097,
      "grad_norm": 0.12227454036474228,
      "learning_rate": 9.677931444671983e-06,
      "loss": 0.0043,
      "step": 37290
    },
    {
      "epoch": 0.6103247975128856,
      "grad_norm": 0.08456715941429138,
      "learning_rate": 9.677595231621012e-06,
      "loss": 0.0029,
      "step": 37300
    },
    {
      "epoch": 0.6104884234639614,
      "grad_norm": 0.6788985729217529,
      "learning_rate": 9.677258849018457e-06,
      "loss": 0.0036,
      "step": 37310
    },
    {
      "epoch": 0.6106520494150373,
      "grad_norm": 0.17371058464050293,
      "learning_rate": 9.676922296876509e-06,
      "loss": 0.0039,
      "step": 37320
    },
    {
      "epoch": 0.610815675366113,
      "grad_norm": 0.2948424816131592,
      "learning_rate": 9.67658557520737e-06,
      "loss": 0.0027,
      "step": 37330
    },
    {
      "epoch": 0.6109793013171889,
      "grad_norm": 0.3183381259441376,
      "learning_rate": 9.676248684023242e-06,
      "loss": 0.0056,
      "step": 37340
    },
    {
      "epoch": 0.6111429272682648,
      "grad_norm": 0.08708910644054413,
      "learning_rate": 9.675911623336338e-06,
      "loss": 0.0039,
      "step": 37350
    },
    {
      "epoch": 0.6113065532193406,
      "grad_norm": 0.4103429317474365,
      "learning_rate": 9.675574393158877e-06,
      "loss": 0.0048,
      "step": 37360
    },
    {
      "epoch": 0.6114701791704165,
      "grad_norm": 0.06971242278814316,
      "learning_rate": 9.67523699350308e-06,
      "loss": 0.0035,
      "step": 37370
    },
    {
      "epoch": 0.6116338051214922,
      "grad_norm": 0.2830762267112732,
      "learning_rate": 9.67489942438118e-06,
      "loss": 0.0049,
      "step": 37380
    },
    {
      "epoch": 0.6117974310725681,
      "grad_norm": 0.43115437030792236,
      "learning_rate": 9.674561685805413e-06,
      "loss": 0.0032,
      "step": 37390
    },
    {
      "epoch": 0.6119610570236439,
      "grad_norm": 0.10677501559257507,
      "learning_rate": 9.674223777788018e-06,
      "loss": 0.0021,
      "step": 37400
    },
    {
      "epoch": 0.6121246829747198,
      "grad_norm": 0.3374081254005432,
      "learning_rate": 9.673885700341248e-06,
      "loss": 0.0036,
      "step": 37410
    },
    {
      "epoch": 0.6122883089257957,
      "grad_norm": 0.17241886258125305,
      "learning_rate": 9.673547453477352e-06,
      "loss": 0.003,
      "step": 37420
    },
    {
      "epoch": 0.6124519348768714,
      "grad_norm": 0.07272238284349442,
      "learning_rate": 9.673209037208596e-06,
      "loss": 0.0045,
      "step": 37430
    },
    {
      "epoch": 0.6126155608279473,
      "grad_norm": 0.303509920835495,
      "learning_rate": 9.67287045154724e-06,
      "loss": 0.0035,
      "step": 37440
    },
    {
      "epoch": 0.6127791867790231,
      "grad_norm": 0.1668083518743515,
      "learning_rate": 9.672531696505564e-06,
      "loss": 0.0024,
      "step": 37450
    },
    {
      "epoch": 0.612942812730099,
      "grad_norm": 0.23766106367111206,
      "learning_rate": 9.672192772095843e-06,
      "loss": 0.0056,
      "step": 37460
    },
    {
      "epoch": 0.6131064386811749,
      "grad_norm": 0.29345017671585083,
      "learning_rate": 9.671853678330365e-06,
      "loss": 0.0036,
      "step": 37470
    },
    {
      "epoch": 0.6132700646322506,
      "grad_norm": 0.16987884044647217,
      "learning_rate": 9.67151441522142e-06,
      "loss": 0.0029,
      "step": 37480
    },
    {
      "epoch": 0.6134336905833265,
      "grad_norm": 0.2603413462638855,
      "learning_rate": 9.671174982781302e-06,
      "loss": 0.0056,
      "step": 37490
    },
    {
      "epoch": 0.6135973165344023,
      "grad_norm": 0.13226963579654694,
      "learning_rate": 9.67083538102232e-06,
      "loss": 0.0049,
      "step": 37500
    },
    {
      "epoch": 0.6137609424854782,
      "grad_norm": 0.2015450894832611,
      "learning_rate": 9.67049560995678e-06,
      "loss": 0.0054,
      "step": 37510
    },
    {
      "epoch": 0.6139245684365541,
      "grad_norm": 0.36624816060066223,
      "learning_rate": 9.670155669597002e-06,
      "loss": 0.0064,
      "step": 37520
    },
    {
      "epoch": 0.6140881943876298,
      "grad_norm": 0.09674598276615143,
      "learning_rate": 9.669815559955302e-06,
      "loss": 0.0025,
      "step": 37530
    },
    {
      "epoch": 0.6142518203387057,
      "grad_norm": 0.029832294210791588,
      "learning_rate": 9.669475281044013e-06,
      "loss": 0.0062,
      "step": 37540
    },
    {
      "epoch": 0.6144154462897815,
      "grad_norm": 0.1951131522655487,
      "learning_rate": 9.669134832875469e-06,
      "loss": 0.0023,
      "step": 37550
    },
    {
      "epoch": 0.6145790722408574,
      "grad_norm": 0.17030267417430878,
      "learning_rate": 9.668794215462009e-06,
      "loss": 0.0025,
      "step": 37560
    },
    {
      "epoch": 0.6147426981919333,
      "grad_norm": 0.26466241478919983,
      "learning_rate": 9.668453428815978e-06,
      "loss": 0.0039,
      "step": 37570
    },
    {
      "epoch": 0.614906324143009,
      "grad_norm": 0.26683205366134644,
      "learning_rate": 9.668112472949732e-06,
      "loss": 0.0049,
      "step": 37580
    },
    {
      "epoch": 0.6150699500940849,
      "grad_norm": 0.28840070962905884,
      "learning_rate": 9.667771347875628e-06,
      "loss": 0.006,
      "step": 37590
    },
    {
      "epoch": 0.6152335760451607,
      "grad_norm": 0.09830860793590546,
      "learning_rate": 9.667430053606032e-06,
      "loss": 0.0092,
      "step": 37600
    },
    {
      "epoch": 0.6153972019962366,
      "grad_norm": 0.11398457735776901,
      "learning_rate": 9.667088590153314e-06,
      "loss": 0.0029,
      "step": 37610
    },
    {
      "epoch": 0.6155608279473125,
      "grad_norm": 0.21779796481132507,
      "learning_rate": 9.666746957529851e-06,
      "loss": 0.0042,
      "step": 37620
    },
    {
      "epoch": 0.6157244538983883,
      "grad_norm": 0.1915261447429657,
      "learning_rate": 9.666405155748029e-06,
      "loss": 0.0035,
      "step": 37630
    },
    {
      "epoch": 0.6158880798494641,
      "grad_norm": 0.2669447064399719,
      "learning_rate": 9.666063184820234e-06,
      "loss": 0.0046,
      "step": 37640
    },
    {
      "epoch": 0.6160517058005399,
      "grad_norm": 0.13493488729000092,
      "learning_rate": 9.665721044758863e-06,
      "loss": 0.0043,
      "step": 37650
    },
    {
      "epoch": 0.6162153317516158,
      "grad_norm": 0.3728507459163666,
      "learning_rate": 9.665378735576317e-06,
      "loss": 0.0044,
      "step": 37660
    },
    {
      "epoch": 0.6163789577026917,
      "grad_norm": 0.20458364486694336,
      "learning_rate": 9.665036257285006e-06,
      "loss": 0.0044,
      "step": 37670
    },
    {
      "epoch": 0.6165425836537675,
      "grad_norm": 0.3213384747505188,
      "learning_rate": 9.664693609897342e-06,
      "loss": 0.0047,
      "step": 37680
    },
    {
      "epoch": 0.6167062096048433,
      "grad_norm": 0.32243579626083374,
      "learning_rate": 9.664350793425748e-06,
      "loss": 0.0032,
      "step": 37690
    },
    {
      "epoch": 0.6168698355559191,
      "grad_norm": 0.14428888261318207,
      "learning_rate": 9.664007807882647e-06,
      "loss": 0.003,
      "step": 37700
    },
    {
      "epoch": 0.617033461506995,
      "grad_norm": 0.3491305112838745,
      "learning_rate": 9.663664653280473e-06,
      "loss": 0.0049,
      "step": 37710
    },
    {
      "epoch": 0.6171970874580709,
      "grad_norm": 0.08974195271730423,
      "learning_rate": 9.663321329631664e-06,
      "loss": 0.0041,
      "step": 37720
    },
    {
      "epoch": 0.6173607134091467,
      "grad_norm": 0.42107313871383667,
      "learning_rate": 9.662977836948663e-06,
      "loss": 0.0047,
      "step": 37730
    },
    {
      "epoch": 0.6175243393602226,
      "grad_norm": 0.07104238122701645,
      "learning_rate": 9.662634175243923e-06,
      "loss": 0.0056,
      "step": 37740
    },
    {
      "epoch": 0.6176879653112983,
      "grad_norm": 0.33224251866340637,
      "learning_rate": 9.662290344529902e-06,
      "loss": 0.0041,
      "step": 37750
    },
    {
      "epoch": 0.6178515912623742,
      "grad_norm": 0.446054607629776,
      "learning_rate": 9.661946344819062e-06,
      "loss": 0.0037,
      "step": 37760
    },
    {
      "epoch": 0.6180152172134501,
      "grad_norm": 0.4401337802410126,
      "learning_rate": 9.66160217612387e-06,
      "loss": 0.0058,
      "step": 37770
    },
    {
      "epoch": 0.6181788431645259,
      "grad_norm": 0.22429263591766357,
      "learning_rate": 9.661257838456804e-06,
      "loss": 0.0045,
      "step": 37780
    },
    {
      "epoch": 0.6183424691156018,
      "grad_norm": 0.14740681648254395,
      "learning_rate": 9.660913331830344e-06,
      "loss": 0.0027,
      "step": 37790
    },
    {
      "epoch": 0.6185060950666775,
      "grad_norm": 0.01172002125531435,
      "learning_rate": 9.66056865625698e-06,
      "loss": 0.0047,
      "step": 37800
    },
    {
      "epoch": 0.6186697210177534,
      "grad_norm": 0.13470607995986938,
      "learning_rate": 9.6602238117492e-06,
      "loss": 0.0036,
      "step": 37810
    },
    {
      "epoch": 0.6188333469688293,
      "grad_norm": 0.3673902451992035,
      "learning_rate": 9.659878798319508e-06,
      "loss": 0.006,
      "step": 37820
    },
    {
      "epoch": 0.6189969729199051,
      "grad_norm": 0.5027232766151428,
      "learning_rate": 9.659533615980411e-06,
      "loss": 0.0043,
      "step": 37830
    },
    {
      "epoch": 0.619160598870981,
      "grad_norm": 1.0106658935546875,
      "learning_rate": 9.65918826474442e-06,
      "loss": 0.0074,
      "step": 37840
    },
    {
      "epoch": 0.6193242248220567,
      "grad_norm": 0.2663329839706421,
      "learning_rate": 9.65884274462405e-06,
      "loss": 0.0057,
      "step": 37850
    },
    {
      "epoch": 0.6194878507731326,
      "grad_norm": 0.37738049030303955,
      "learning_rate": 9.658497055631827e-06,
      "loss": 0.0053,
      "step": 37860
    },
    {
      "epoch": 0.6196514767242085,
      "grad_norm": 0.1489264816045761,
      "learning_rate": 9.658151197780285e-06,
      "loss": 0.0041,
      "step": 37870
    },
    {
      "epoch": 0.6198151026752843,
      "grad_norm": 0.2652358114719391,
      "learning_rate": 9.657805171081958e-06,
      "loss": 0.0026,
      "step": 37880
    },
    {
      "epoch": 0.6199787286263602,
      "grad_norm": 0.3128514289855957,
      "learning_rate": 9.657458975549386e-06,
      "loss": 0.0056,
      "step": 37890
    },
    {
      "epoch": 0.6201423545774359,
      "grad_norm": 0.14601120352745056,
      "learning_rate": 9.657112611195121e-06,
      "loss": 0.0036,
      "step": 37900
    },
    {
      "epoch": 0.6203059805285118,
      "grad_norm": 0.5487610697746277,
      "learning_rate": 9.656766078031717e-06,
      "loss": 0.0047,
      "step": 37910
    },
    {
      "epoch": 0.6204696064795877,
      "grad_norm": 0.2473091185092926,
      "learning_rate": 9.656419376071736e-06,
      "loss": 0.0034,
      "step": 37920
    },
    {
      "epoch": 0.6206332324306635,
      "grad_norm": 0.298767626285553,
      "learning_rate": 9.656072505327741e-06,
      "loss": 0.0041,
      "step": 37930
    },
    {
      "epoch": 0.6207968583817394,
      "grad_norm": 0.23336118459701538,
      "learning_rate": 9.65572546581231e-06,
      "loss": 0.0076,
      "step": 37940
    },
    {
      "epoch": 0.6209604843328151,
      "grad_norm": 0.1006658673286438,
      "learning_rate": 9.65537825753802e-06,
      "loss": 0.0033,
      "step": 37950
    },
    {
      "epoch": 0.621124110283891,
      "grad_norm": 0.22614005208015442,
      "learning_rate": 9.65503088051746e-06,
      "loss": 0.0052,
      "step": 37960
    },
    {
      "epoch": 0.6212877362349669,
      "grad_norm": 0.2829948365688324,
      "learning_rate": 9.654683334763216e-06,
      "loss": 0.0031,
      "step": 37970
    },
    {
      "epoch": 0.6214513621860427,
      "grad_norm": 0.0900287851691246,
      "learning_rate": 9.654335620287886e-06,
      "loss": 0.0029,
      "step": 37980
    },
    {
      "epoch": 0.6216149881371186,
      "grad_norm": 0.27310997247695923,
      "learning_rate": 9.65398773710408e-06,
      "loss": 0.0034,
      "step": 37990
    },
    {
      "epoch": 0.6217786140881943,
      "grad_norm": 0.15457694232463837,
      "learning_rate": 9.653639685224403e-06,
      "loss": 0.0036,
      "step": 38000
    },
    {
      "epoch": 0.6219422400392702,
      "grad_norm": 0.15960688889026642,
      "learning_rate": 9.653291464661473e-06,
      "loss": 0.0036,
      "step": 38010
    },
    {
      "epoch": 0.6221058659903461,
      "grad_norm": 0.6965922713279724,
      "learning_rate": 9.65294307542791e-06,
      "loss": 0.0027,
      "step": 38020
    },
    {
      "epoch": 0.6222694919414219,
      "grad_norm": 0.14204402267932892,
      "learning_rate": 9.652594517536345e-06,
      "loss": 0.0036,
      "step": 38030
    },
    {
      "epoch": 0.6224331178924978,
      "grad_norm": 0.22957226634025574,
      "learning_rate": 9.652245790999409e-06,
      "loss": 0.0043,
      "step": 38040
    },
    {
      "epoch": 0.6225967438435736,
      "grad_norm": 0.13951456546783447,
      "learning_rate": 9.651896895829745e-06,
      "loss": 0.0044,
      "step": 38050
    },
    {
      "epoch": 0.6227603697946494,
      "grad_norm": 0.05449066311120987,
      "learning_rate": 9.65154783204e-06,
      "loss": 0.0025,
      "step": 38060
    },
    {
      "epoch": 0.6229239957457253,
      "grad_norm": 0.3685210049152374,
      "learning_rate": 9.651198599642826e-06,
      "loss": 0.004,
      "step": 38070
    },
    {
      "epoch": 0.6230876216968011,
      "grad_norm": 0.18782642483711243,
      "learning_rate": 9.65084919865088e-06,
      "loss": 0.0037,
      "step": 38080
    },
    {
      "epoch": 0.623251247647877,
      "grad_norm": 0.039590705186128616,
      "learning_rate": 9.650499629076829e-06,
      "loss": 0.0042,
      "step": 38090
    },
    {
      "epoch": 0.6234148735989528,
      "grad_norm": 0.1745484620332718,
      "learning_rate": 9.650149890933345e-06,
      "loss": 0.0033,
      "step": 38100
    },
    {
      "epoch": 0.6235784995500286,
      "grad_norm": 0.4898908734321594,
      "learning_rate": 9.649799984233102e-06,
      "loss": 0.0077,
      "step": 38110
    },
    {
      "epoch": 0.6237421255011045,
      "grad_norm": 0.0713873952627182,
      "learning_rate": 9.649449908988786e-06,
      "loss": 0.0021,
      "step": 38120
    },
    {
      "epoch": 0.6239057514521803,
      "grad_norm": 0.20937389135360718,
      "learning_rate": 9.649099665213086e-06,
      "loss": 0.0024,
      "step": 38130
    },
    {
      "epoch": 0.6240693774032562,
      "grad_norm": 0.0904269590973854,
      "learning_rate": 9.648749252918697e-06,
      "loss": 0.0047,
      "step": 38140
    },
    {
      "epoch": 0.624233003354332,
      "grad_norm": 0.15568599104881287,
      "learning_rate": 9.648398672118318e-06,
      "loss": 0.0058,
      "step": 38150
    },
    {
      "epoch": 0.6243966293054078,
      "grad_norm": 0.10997980833053589,
      "learning_rate": 9.648047922824661e-06,
      "loss": 0.0057,
      "step": 38160
    },
    {
      "epoch": 0.6245602552564837,
      "grad_norm": 0.11430040746927261,
      "learning_rate": 9.64769700505044e-06,
      "loss": 0.0037,
      "step": 38170
    },
    {
      "epoch": 0.6247238812075595,
      "grad_norm": 0.26384666562080383,
      "learning_rate": 9.64734591880837e-06,
      "loss": 0.004,
      "step": 38180
    },
    {
      "epoch": 0.6248875071586354,
      "grad_norm": 0.012330407276749611,
      "learning_rate": 9.646994664111183e-06,
      "loss": 0.0054,
      "step": 38190
    },
    {
      "epoch": 0.6250511331097112,
      "grad_norm": 0.1942172646522522,
      "learning_rate": 9.646643240971605e-06,
      "loss": 0.0026,
      "step": 38200
    },
    {
      "epoch": 0.625214759060787,
      "grad_norm": 0.18183833360671997,
      "learning_rate": 9.64629164940238e-06,
      "loss": 0.0031,
      "step": 38210
    },
    {
      "epoch": 0.6253783850118629,
      "grad_norm": 0.09610706567764282,
      "learning_rate": 9.645939889416249e-06,
      "loss": 0.0032,
      "step": 38220
    },
    {
      "epoch": 0.6255420109629387,
      "grad_norm": 0.22070765495300293,
      "learning_rate": 9.645587961025961e-06,
      "loss": 0.0029,
      "step": 38230
    },
    {
      "epoch": 0.6257056369140146,
      "grad_norm": 0.1928679198026657,
      "learning_rate": 9.645235864244278e-06,
      "loss": 0.0022,
      "step": 38240
    },
    {
      "epoch": 0.6258692628650904,
      "grad_norm": 0.31935417652130127,
      "learning_rate": 9.644883599083959e-06,
      "loss": 0.0037,
      "step": 38250
    },
    {
      "epoch": 0.6260328888161663,
      "grad_norm": 0.13179603219032288,
      "learning_rate": 9.64453116555777e-06,
      "loss": 0.0039,
      "step": 38260
    },
    {
      "epoch": 0.626196514767242,
      "grad_norm": 0.07257692515850067,
      "learning_rate": 9.644178563678493e-06,
      "loss": 0.0031,
      "step": 38270
    },
    {
      "epoch": 0.6263601407183179,
      "grad_norm": 0.5487810373306274,
      "learning_rate": 9.643825793458904e-06,
      "loss": 0.0049,
      "step": 38280
    },
    {
      "epoch": 0.6265237666693938,
      "grad_norm": 0.30215689539909363,
      "learning_rate": 9.64347285491179e-06,
      "loss": 0.0043,
      "step": 38290
    },
    {
      "epoch": 0.6266873926204696,
      "grad_norm": 0.11180410534143448,
      "learning_rate": 9.643119748049945e-06,
      "loss": 0.0036,
      "step": 38300
    },
    {
      "epoch": 0.6268510185715455,
      "grad_norm": 0.31685671210289,
      "learning_rate": 9.64276647288617e-06,
      "loss": 0.0031,
      "step": 38310
    },
    {
      "epoch": 0.6270146445226212,
      "grad_norm": 0.2429034411907196,
      "learning_rate": 9.642413029433268e-06,
      "loss": 0.0061,
      "step": 38320
    },
    {
      "epoch": 0.6271782704736971,
      "grad_norm": 0.2383284866809845,
      "learning_rate": 9.642059417704052e-06,
      "loss": 0.0034,
      "step": 38330
    },
    {
      "epoch": 0.627341896424773,
      "grad_norm": 0.16095753014087677,
      "learning_rate": 9.641705637711341e-06,
      "loss": 0.0033,
      "step": 38340
    },
    {
      "epoch": 0.6275055223758488,
      "grad_norm": 0.057584162801504135,
      "learning_rate": 9.641351689467954e-06,
      "loss": 0.0029,
      "step": 38350
    },
    {
      "epoch": 0.6276691483269247,
      "grad_norm": 0.10399903357028961,
      "learning_rate": 9.640997572986723e-06,
      "loss": 0.0056,
      "step": 38360
    },
    {
      "epoch": 0.6278327742780004,
      "grad_norm": 0.2439599633216858,
      "learning_rate": 9.640643288280487e-06,
      "loss": 0.0033,
      "step": 38370
    },
    {
      "epoch": 0.6279964002290763,
      "grad_norm": 0.2845558226108551,
      "learning_rate": 9.640288835362082e-06,
      "loss": 0.0033,
      "step": 38380
    },
    {
      "epoch": 0.6281600261801522,
      "grad_norm": 0.0931815654039383,
      "learning_rate": 9.639934214244362e-06,
      "loss": 0.0028,
      "step": 38390
    },
    {
      "epoch": 0.628323652131228,
      "grad_norm": 0.1392059177160263,
      "learning_rate": 9.639579424940178e-06,
      "loss": 0.0026,
      "step": 38400
    },
    {
      "epoch": 0.6284872780823039,
      "grad_norm": 0.11994877457618713,
      "learning_rate": 9.63922446746239e-06,
      "loss": 0.0046,
      "step": 38410
    },
    {
      "epoch": 0.6286509040333796,
      "grad_norm": 0.1402147114276886,
      "learning_rate": 9.638869341823865e-06,
      "loss": 0.006,
      "step": 38420
    },
    {
      "epoch": 0.6288145299844555,
      "grad_norm": 0.14142590761184692,
      "learning_rate": 9.638514048037476e-06,
      "loss": 0.0036,
      "step": 38430
    },
    {
      "epoch": 0.6289781559355314,
      "grad_norm": 0.03410758823156357,
      "learning_rate": 9.638158586116102e-06,
      "loss": 0.0035,
      "step": 38440
    },
    {
      "epoch": 0.6291417818866072,
      "grad_norm": 0.10523486137390137,
      "learning_rate": 9.637802956072627e-06,
      "loss": 0.0045,
      "step": 38450
    },
    {
      "epoch": 0.6293054078376831,
      "grad_norm": 0.08307523280382156,
      "learning_rate": 9.637447157919939e-06,
      "loss": 0.003,
      "step": 38460
    },
    {
      "epoch": 0.6294690337887588,
      "grad_norm": 0.1560235172510147,
      "learning_rate": 9.637091191670939e-06,
      "loss": 0.0043,
      "step": 38470
    },
    {
      "epoch": 0.6296326597398347,
      "grad_norm": 0.0628463625907898,
      "learning_rate": 9.63673505733853e-06,
      "loss": 0.0039,
      "step": 38480
    },
    {
      "epoch": 0.6297962856909106,
      "grad_norm": 0.3192249834537506,
      "learning_rate": 9.636378754935615e-06,
      "loss": 0.0032,
      "step": 38490
    },
    {
      "epoch": 0.6299599116419864,
      "grad_norm": 0.1372830867767334,
      "learning_rate": 9.636022284475117e-06,
      "loss": 0.0033,
      "step": 38500
    },
    {
      "epoch": 0.6301235375930623,
      "grad_norm": 0.07836729288101196,
      "learning_rate": 9.635665645969952e-06,
      "loss": 0.0028,
      "step": 38510
    },
    {
      "epoch": 0.630287163544138,
      "grad_norm": 0.10574813932180405,
      "learning_rate": 9.635308839433048e-06,
      "loss": 0.0046,
      "step": 38520
    },
    {
      "epoch": 0.6304507894952139,
      "grad_norm": 0.12654341757297516,
      "learning_rate": 9.63495186487734e-06,
      "loss": 0.0032,
      "step": 38530
    },
    {
      "epoch": 0.6306144154462898,
      "grad_norm": 0.28303995728492737,
      "learning_rate": 9.634594722315768e-06,
      "loss": 0.0047,
      "step": 38540
    },
    {
      "epoch": 0.6307780413973656,
      "grad_norm": 0.22968409955501556,
      "learning_rate": 9.634237411761275e-06,
      "loss": 0.0029,
      "step": 38550
    },
    {
      "epoch": 0.6309416673484415,
      "grad_norm": 0.0440322607755661,
      "learning_rate": 9.633879933226815e-06,
      "loss": 0.0049,
      "step": 38560
    },
    {
      "epoch": 0.6311052932995173,
      "grad_norm": 0.23966920375823975,
      "learning_rate": 9.633522286725344e-06,
      "loss": 0.0035,
      "step": 38570
    },
    {
      "epoch": 0.6312689192505931,
      "grad_norm": 0.3072430491447449,
      "learning_rate": 9.633164472269828e-06,
      "loss": 0.007,
      "step": 38580
    },
    {
      "epoch": 0.631432545201669,
      "grad_norm": 0.2619354724884033,
      "learning_rate": 9.632806489873231e-06,
      "loss": 0.0054,
      "step": 38590
    },
    {
      "epoch": 0.6315961711527448,
      "grad_norm": 0.226179838180542,
      "learning_rate": 9.632448339548536e-06,
      "loss": 0.0028,
      "step": 38600
    },
    {
      "epoch": 0.6317597971038207,
      "grad_norm": 0.10131746530532837,
      "learning_rate": 9.632090021308725e-06,
      "loss": 0.0028,
      "step": 38610
    },
    {
      "epoch": 0.6319234230548965,
      "grad_norm": 0.14819808304309845,
      "learning_rate": 9.631731535166781e-06,
      "loss": 0.0026,
      "step": 38620
    },
    {
      "epoch": 0.6320870490059723,
      "grad_norm": 0.185590460896492,
      "learning_rate": 9.6313728811357e-06,
      "loss": 0.0038,
      "step": 38630
    },
    {
      "epoch": 0.6322506749570482,
      "grad_norm": 0.255806028842926,
      "learning_rate": 9.631014059228486e-06,
      "loss": 0.0038,
      "step": 38640
    },
    {
      "epoch": 0.632414300908124,
      "grad_norm": 0.21789845824241638,
      "learning_rate": 9.630655069458141e-06,
      "loss": 0.005,
      "step": 38650
    },
    {
      "epoch": 0.6325779268591999,
      "grad_norm": 0.28489506244659424,
      "learning_rate": 9.63029591183768e-06,
      "loss": 0.0033,
      "step": 38660
    },
    {
      "epoch": 0.6327415528102757,
      "grad_norm": 0.19765153527259827,
      "learning_rate": 9.629936586380119e-06,
      "loss": 0.0029,
      "step": 38670
    },
    {
      "epoch": 0.6329051787613516,
      "grad_norm": 0.061400216072797775,
      "learning_rate": 9.629577093098486e-06,
      "loss": 0.0032,
      "step": 38680
    },
    {
      "epoch": 0.6330688047124274,
      "grad_norm": 0.23604781925678253,
      "learning_rate": 9.629217432005811e-06,
      "loss": 0.0037,
      "step": 38690
    },
    {
      "epoch": 0.6332324306635032,
      "grad_norm": 0.28271064162254333,
      "learning_rate": 9.628857603115129e-06,
      "loss": 0.0038,
      "step": 38700
    },
    {
      "epoch": 0.6333960566145791,
      "grad_norm": 0.2158617377281189,
      "learning_rate": 9.628497606439484e-06,
      "loss": 0.0036,
      "step": 38710
    },
    {
      "epoch": 0.6335596825656549,
      "grad_norm": 0.09813366085290909,
      "learning_rate": 9.628137441991926e-06,
      "loss": 0.0041,
      "step": 38720
    },
    {
      "epoch": 0.6337233085167308,
      "grad_norm": 0.05281127244234085,
      "learning_rate": 9.627777109785507e-06,
      "loss": 0.0027,
      "step": 38730
    },
    {
      "epoch": 0.6338869344678066,
      "grad_norm": 0.088776133954525,
      "learning_rate": 9.627416609833294e-06,
      "loss": 0.0035,
      "step": 38740
    },
    {
      "epoch": 0.6340505604188824,
      "grad_norm": 0.2632506787776947,
      "learning_rate": 9.627055942148347e-06,
      "loss": 0.0037,
      "step": 38750
    },
    {
      "epoch": 0.6342141863699583,
      "grad_norm": 0.07854590564966202,
      "learning_rate": 9.626695106743744e-06,
      "loss": 0.0033,
      "step": 38760
    },
    {
      "epoch": 0.6343778123210341,
      "grad_norm": 0.1894259750843048,
      "learning_rate": 9.626334103632564e-06,
      "loss": 0.0045,
      "step": 38770
    },
    {
      "epoch": 0.63454143827211,
      "grad_norm": 0.32894113659858704,
      "learning_rate": 9.625972932827892e-06,
      "loss": 0.0049,
      "step": 38780
    },
    {
      "epoch": 0.6347050642231858,
      "grad_norm": 0.08008615672588348,
      "learning_rate": 9.625611594342819e-06,
      "loss": 0.0024,
      "step": 38790
    },
    {
      "epoch": 0.6348686901742616,
      "grad_norm": 0.08497855812311172,
      "learning_rate": 9.625250088190444e-06,
      "loss": 0.005,
      "step": 38800
    },
    {
      "epoch": 0.6350323161253375,
      "grad_norm": 0.2697611153125763,
      "learning_rate": 9.624888414383868e-06,
      "loss": 0.0047,
      "step": 38810
    },
    {
      "epoch": 0.6351959420764133,
      "grad_norm": 0.15111055970191956,
      "learning_rate": 9.624526572936203e-06,
      "loss": 0.0046,
      "step": 38820
    },
    {
      "epoch": 0.6353595680274892,
      "grad_norm": 0.1395147293806076,
      "learning_rate": 9.624164563860565e-06,
      "loss": 0.0036,
      "step": 38830
    },
    {
      "epoch": 0.635523193978565,
      "grad_norm": 0.26285654306411743,
      "learning_rate": 9.623802387170076e-06,
      "loss": 0.0029,
      "step": 38840
    },
    {
      "epoch": 0.6356868199296408,
      "grad_norm": 0.08407123386859894,
      "learning_rate": 9.623440042877863e-06,
      "loss": 0.0047,
      "step": 38850
    },
    {
      "epoch": 0.6358504458807167,
      "grad_norm": 0.15698997676372528,
      "learning_rate": 9.62307753099706e-06,
      "loss": 0.0039,
      "step": 38860
    },
    {
      "epoch": 0.6360140718317925,
      "grad_norm": 0.1328202188014984,
      "learning_rate": 9.62271485154081e-06,
      "loss": 0.004,
      "step": 38870
    },
    {
      "epoch": 0.6361776977828684,
      "grad_norm": 0.14202706515789032,
      "learning_rate": 9.622352004522255e-06,
      "loss": 0.0031,
      "step": 38880
    },
    {
      "epoch": 0.6363413237339443,
      "grad_norm": 0.3476366400718689,
      "learning_rate": 9.621988989954549e-06,
      "loss": 0.0038,
      "step": 38890
    },
    {
      "epoch": 0.63650494968502,
      "grad_norm": 0.24466153979301453,
      "learning_rate": 9.621625807850851e-06,
      "loss": 0.0034,
      "step": 38900
    },
    {
      "epoch": 0.6366685756360959,
      "grad_norm": 0.10046984255313873,
      "learning_rate": 9.621262458224324e-06,
      "loss": 0.004,
      "step": 38910
    },
    {
      "epoch": 0.6368322015871717,
      "grad_norm": 0.3870885670185089,
      "learning_rate": 9.620898941088144e-06,
      "loss": 0.0037,
      "step": 38920
    },
    {
      "epoch": 0.6369958275382476,
      "grad_norm": 0.12600962817668915,
      "learning_rate": 9.62053525645548e-06,
      "loss": 0.0047,
      "step": 38930
    },
    {
      "epoch": 0.6371594534893235,
      "grad_norm": 0.2957196831703186,
      "learning_rate": 9.62017140433952e-06,
      "loss": 0.0047,
      "step": 38940
    },
    {
      "epoch": 0.6373230794403992,
      "grad_norm": 0.18455517292022705,
      "learning_rate": 9.619807384753449e-06,
      "loss": 0.0038,
      "step": 38950
    },
    {
      "epoch": 0.6374867053914751,
      "grad_norm": 0.12891778349876404,
      "learning_rate": 9.619443197710465e-06,
      "loss": 0.0044,
      "step": 38960
    },
    {
      "epoch": 0.6376503313425509,
      "grad_norm": 0.2780033051967621,
      "learning_rate": 9.619078843223766e-06,
      "loss": 0.0076,
      "step": 38970
    },
    {
      "epoch": 0.6378139572936268,
      "grad_norm": 0.18556208908557892,
      "learning_rate": 9.618714321306563e-06,
      "loss": 0.0038,
      "step": 38980
    },
    {
      "epoch": 0.6379775832447027,
      "grad_norm": 0.10493610054254532,
      "learning_rate": 9.618349631972064e-06,
      "loss": 0.0038,
      "step": 38990
    },
    {
      "epoch": 0.6381412091957784,
      "grad_norm": 0.244174987077713,
      "learning_rate": 9.617984775233493e-06,
      "loss": 0.0037,
      "step": 39000
    },
    {
      "epoch": 0.6383048351468543,
      "grad_norm": 0.11400426179170609,
      "learning_rate": 9.617619751104072e-06,
      "loss": 0.0066,
      "step": 39010
    },
    {
      "epoch": 0.6384684610979301,
      "grad_norm": 0.1196909248828888,
      "learning_rate": 9.617254559597034e-06,
      "loss": 0.0045,
      "step": 39020
    },
    {
      "epoch": 0.638632087049006,
      "grad_norm": 0.08639112114906311,
      "learning_rate": 9.616889200725615e-06,
      "loss": 0.0024,
      "step": 39030
    },
    {
      "epoch": 0.6387957130000819,
      "grad_norm": 0.2317192554473877,
      "learning_rate": 9.616523674503057e-06,
      "loss": 0.0055,
      "step": 39040
    },
    {
      "epoch": 0.6389593389511576,
      "grad_norm": 0.3311014175415039,
      "learning_rate": 9.616157980942616e-06,
      "loss": 0.0026,
      "step": 39050
    },
    {
      "epoch": 0.6391229649022335,
      "grad_norm": 0.25754231214523315,
      "learning_rate": 9.615792120057539e-06,
      "loss": 0.0026,
      "step": 39060
    },
    {
      "epoch": 0.6392865908533093,
      "grad_norm": 0.1464586704969406,
      "learning_rate": 9.615426091861091e-06,
      "loss": 0.0027,
      "step": 39070
    },
    {
      "epoch": 0.6394502168043852,
      "grad_norm": 0.04481315240263939,
      "learning_rate": 9.615059896366542e-06,
      "loss": 0.0028,
      "step": 39080
    },
    {
      "epoch": 0.6396138427554611,
      "grad_norm": 0.2647063136100769,
      "learning_rate": 9.614693533587162e-06,
      "loss": 0.0049,
      "step": 39090
    },
    {
      "epoch": 0.6397774687065368,
      "grad_norm": 0.21080859005451202,
      "learning_rate": 9.614327003536235e-06,
      "loss": 0.0049,
      "step": 39100
    },
    {
      "epoch": 0.6399410946576127,
      "grad_norm": 0.1746206134557724,
      "learning_rate": 9.613960306227043e-06,
      "loss": 0.0039,
      "step": 39110
    },
    {
      "epoch": 0.6401047206086885,
      "grad_norm": 0.12504762411117554,
      "learning_rate": 9.613593441672879e-06,
      "loss": 0.0025,
      "step": 39120
    },
    {
      "epoch": 0.6402683465597644,
      "grad_norm": 0.46945270895957947,
      "learning_rate": 9.613226409887042e-06,
      "loss": 0.0041,
      "step": 39130
    },
    {
      "epoch": 0.6404319725108402,
      "grad_norm": 0.08291307091712952,
      "learning_rate": 9.612859210882835e-06,
      "loss": 0.003,
      "step": 39140
    },
    {
      "epoch": 0.640595598461916,
      "grad_norm": 0.08960380405187607,
      "learning_rate": 9.612491844673568e-06,
      "loss": 0.003,
      "step": 39150
    },
    {
      "epoch": 0.6407592244129919,
      "grad_norm": 0.12569643557071686,
      "learning_rate": 9.612124311272558e-06,
      "loss": 0.0029,
      "step": 39160
    },
    {
      "epoch": 0.6409228503640677,
      "grad_norm": 0.14201229810714722,
      "learning_rate": 9.611756610693124e-06,
      "loss": 0.0042,
      "step": 39170
    },
    {
      "epoch": 0.6410864763151436,
      "grad_norm": 0.10764100402593613,
      "learning_rate": 9.6113887429486e-06,
      "loss": 0.0038,
      "step": 39180
    },
    {
      "epoch": 0.6412501022662194,
      "grad_norm": 0.42665010690689087,
      "learning_rate": 9.611020708052316e-06,
      "loss": 0.004,
      "step": 39190
    },
    {
      "epoch": 0.6414137282172953,
      "grad_norm": 0.23785610496997833,
      "learning_rate": 9.610652506017615e-06,
      "loss": 0.0035,
      "step": 39200
    },
    {
      "epoch": 0.6415773541683711,
      "grad_norm": 0.1503949612379074,
      "learning_rate": 9.61028413685784e-06,
      "loss": 0.0055,
      "step": 39210
    },
    {
      "epoch": 0.6417409801194469,
      "grad_norm": 0.12189337611198425,
      "learning_rate": 9.609915600586347e-06,
      "loss": 0.0063,
      "step": 39220
    },
    {
      "epoch": 0.6419046060705228,
      "grad_norm": 0.06078196316957474,
      "learning_rate": 9.609546897216493e-06,
      "loss": 0.0064,
      "step": 39230
    },
    {
      "epoch": 0.6420682320215986,
      "grad_norm": 0.1902364045381546,
      "learning_rate": 9.609178026761643e-06,
      "loss": 0.0062,
      "step": 39240
    },
    {
      "epoch": 0.6422318579726745,
      "grad_norm": 0.3740362823009491,
      "learning_rate": 9.608808989235165e-06,
      "loss": 0.0058,
      "step": 39250
    },
    {
      "epoch": 0.6423954839237503,
      "grad_norm": 0.1996176540851593,
      "learning_rate": 9.60843978465044e-06,
      "loss": 0.0045,
      "step": 39260
    },
    {
      "epoch": 0.6425591098748261,
      "grad_norm": 0.28526821732521057,
      "learning_rate": 9.60807041302085e-06,
      "loss": 0.0023,
      "step": 39270
    },
    {
      "epoch": 0.642722735825902,
      "grad_norm": 0.12540899217128754,
      "learning_rate": 9.607700874359781e-06,
      "loss": 0.0034,
      "step": 39280
    },
    {
      "epoch": 0.6428863617769778,
      "grad_norm": 0.14467529952526093,
      "learning_rate": 9.607331168680632e-06,
      "loss": 0.0033,
      "step": 39290
    },
    {
      "epoch": 0.6430499877280537,
      "grad_norm": 0.32027381658554077,
      "learning_rate": 9.606961295996798e-06,
      "loss": 0.0038,
      "step": 39300
    },
    {
      "epoch": 0.6432136136791295,
      "grad_norm": 0.04616748169064522,
      "learning_rate": 9.606591256321693e-06,
      "loss": 0.0028,
      "step": 39310
    },
    {
      "epoch": 0.6433772396302053,
      "grad_norm": 0.4567842483520508,
      "learning_rate": 9.606221049668725e-06,
      "loss": 0.0042,
      "step": 39320
    },
    {
      "epoch": 0.6435408655812812,
      "grad_norm": 0.23921778798103333,
      "learning_rate": 9.605850676051314e-06,
      "loss": 0.0057,
      "step": 39330
    },
    {
      "epoch": 0.643704491532357,
      "grad_norm": 0.1458958089351654,
      "learning_rate": 9.605480135482886e-06,
      "loss": 0.0034,
      "step": 39340
    },
    {
      "epoch": 0.6438681174834329,
      "grad_norm": 0.12824426591396332,
      "learning_rate": 9.605109427976873e-06,
      "loss": 0.0043,
      "step": 39350
    },
    {
      "epoch": 0.6440317434345088,
      "grad_norm": 0.2134348303079605,
      "learning_rate": 9.60473855354671e-06,
      "loss": 0.0021,
      "step": 39360
    },
    {
      "epoch": 0.6441953693855845,
      "grad_norm": 0.27929699420928955,
      "learning_rate": 9.604367512205843e-06,
      "loss": 0.0041,
      "step": 39370
    },
    {
      "epoch": 0.6443589953366604,
      "grad_norm": 0.04714587703347206,
      "learning_rate": 9.60399630396772e-06,
      "loss": 0.0038,
      "step": 39380
    },
    {
      "epoch": 0.6445226212877362,
      "grad_norm": 0.12128560990095139,
      "learning_rate": 9.603624928845796e-06,
      "loss": 0.0051,
      "step": 39390
    },
    {
      "epoch": 0.6446862472388121,
      "grad_norm": 0.14154492318630219,
      "learning_rate": 9.603253386853533e-06,
      "loss": 0.0022,
      "step": 39400
    },
    {
      "epoch": 0.644849873189888,
      "grad_norm": 0.34802916646003723,
      "learning_rate": 9.602881678004398e-06,
      "loss": 0.0057,
      "step": 39410
    },
    {
      "epoch": 0.6450134991409637,
      "grad_norm": 0.15548211336135864,
      "learning_rate": 9.602509802311865e-06,
      "loss": 0.0061,
      "step": 39420
    },
    {
      "epoch": 0.6451771250920396,
      "grad_norm": 0.11623035371303558,
      "learning_rate": 9.602137759789415e-06,
      "loss": 0.0031,
      "step": 39430
    },
    {
      "epoch": 0.6453407510431154,
      "grad_norm": 0.13361559808254242,
      "learning_rate": 9.60176555045053e-06,
      "loss": 0.0032,
      "step": 39440
    },
    {
      "epoch": 0.6455043769941913,
      "grad_norm": 0.36541488766670227,
      "learning_rate": 9.601393174308705e-06,
      "loss": 0.0031,
      "step": 39450
    },
    {
      "epoch": 0.6456680029452672,
      "grad_norm": 0.18346063792705536,
      "learning_rate": 9.601020631377435e-06,
      "loss": 0.005,
      "step": 39460
    },
    {
      "epoch": 0.6458316288963429,
      "grad_norm": 0.251495897769928,
      "learning_rate": 9.600647921670227e-06,
      "loss": 0.0071,
      "step": 39470
    },
    {
      "epoch": 0.6459952548474188,
      "grad_norm": 0.3548803925514221,
      "learning_rate": 9.600275045200589e-06,
      "loss": 0.0049,
      "step": 39480
    },
    {
      "epoch": 0.6461588807984946,
      "grad_norm": 0.09060274064540863,
      "learning_rate": 9.599902001982035e-06,
      "loss": 0.0038,
      "step": 39490
    },
    {
      "epoch": 0.6463225067495705,
      "grad_norm": 0.23502935469150543,
      "learning_rate": 9.599528792028091e-06,
      "loss": 0.0043,
      "step": 39500
    },
    {
      "epoch": 0.6464861327006464,
      "grad_norm": 0.18305960297584534,
      "learning_rate": 9.599155415352284e-06,
      "loss": 0.0088,
      "step": 39510
    },
    {
      "epoch": 0.6466497586517221,
      "grad_norm": 0.19083371758460999,
      "learning_rate": 9.598781871968145e-06,
      "loss": 0.0037,
      "step": 39520
    },
    {
      "epoch": 0.646813384602798,
      "grad_norm": 0.189573273062706,
      "learning_rate": 9.598408161889217e-06,
      "loss": 0.0045,
      "step": 39530
    },
    {
      "epoch": 0.6469770105538738,
      "grad_norm": 0.14892877638339996,
      "learning_rate": 9.598034285129045e-06,
      "loss": 0.0042,
      "step": 39540
    },
    {
      "epoch": 0.6471406365049497,
      "grad_norm": 0.32697221636772156,
      "learning_rate": 9.59766024170118e-06,
      "loss": 0.0058,
      "step": 39550
    },
    {
      "epoch": 0.6473042624560256,
      "grad_norm": 0.39728957414627075,
      "learning_rate": 9.597286031619185e-06,
      "loss": 0.0036,
      "step": 39560
    },
    {
      "epoch": 0.6474678884071013,
      "grad_norm": 0.12987461686134338,
      "learning_rate": 9.596911654896616e-06,
      "loss": 0.0053,
      "step": 39570
    },
    {
      "epoch": 0.6476315143581772,
      "grad_norm": 0.0808095708489418,
      "learning_rate": 9.596537111547051e-06,
      "loss": 0.0033,
      "step": 39580
    },
    {
      "epoch": 0.647795140309253,
      "grad_norm": 0.10307874530553818,
      "learning_rate": 9.596162401584062e-06,
      "loss": 0.0047,
      "step": 39590
    },
    {
      "epoch": 0.6479587662603289,
      "grad_norm": 0.11977291107177734,
      "learning_rate": 9.595787525021234e-06,
      "loss": 0.003,
      "step": 39600
    },
    {
      "epoch": 0.6481223922114048,
      "grad_norm": 0.18306361138820648,
      "learning_rate": 9.595412481872152e-06,
      "loss": 0.0036,
      "step": 39610
    },
    {
      "epoch": 0.6482860181624805,
      "grad_norm": 0.25254836678504944,
      "learning_rate": 9.595037272150414e-06,
      "loss": 0.0056,
      "step": 39620
    },
    {
      "epoch": 0.6484496441135564,
      "grad_norm": 0.1175762414932251,
      "learning_rate": 9.594661895869618e-06,
      "loss": 0.0041,
      "step": 39630
    },
    {
      "epoch": 0.6486132700646322,
      "grad_norm": 0.1297464370727539,
      "learning_rate": 9.594286353043369e-06,
      "loss": 0.0029,
      "step": 39640
    },
    {
      "epoch": 0.6487768960157081,
      "grad_norm": 0.1492602527141571,
      "learning_rate": 9.593910643685284e-06,
      "loss": 0.0031,
      "step": 39650
    },
    {
      "epoch": 0.648940521966784,
      "grad_norm": 0.146444633603096,
      "learning_rate": 9.59353476780898e-06,
      "loss": 0.0026,
      "step": 39660
    },
    {
      "epoch": 0.6491041479178598,
      "grad_norm": 0.15150833129882812,
      "learning_rate": 9.59315872542808e-06,
      "loss": 0.0038,
      "step": 39670
    },
    {
      "epoch": 0.6492677738689356,
      "grad_norm": 0.10827313363552094,
      "learning_rate": 9.592782516556216e-06,
      "loss": 0.0039,
      "step": 39680
    },
    {
      "epoch": 0.6494313998200114,
      "grad_norm": 0.17238198220729828,
      "learning_rate": 9.592406141207024e-06,
      "loss": 0.0026,
      "step": 39690
    },
    {
      "epoch": 0.6495950257710873,
      "grad_norm": 0.05940546840429306,
      "learning_rate": 9.592029599394147e-06,
      "loss": 0.0029,
      "step": 39700
    },
    {
      "epoch": 0.6497586517221632,
      "grad_norm": 0.12967193126678467,
      "learning_rate": 9.591652891131233e-06,
      "loss": 0.004,
      "step": 39710
    },
    {
      "epoch": 0.649922277673239,
      "grad_norm": 0.33647042512893677,
      "learning_rate": 9.591276016431938e-06,
      "loss": 0.0027,
      "step": 39720
    },
    {
      "epoch": 0.6500859036243148,
      "grad_norm": 0.09277136623859406,
      "learning_rate": 9.590898975309924e-06,
      "loss": 0.0029,
      "step": 39730
    },
    {
      "epoch": 0.6502495295753906,
      "grad_norm": 0.17488396167755127,
      "learning_rate": 9.590521767778853e-06,
      "loss": 0.0035,
      "step": 39740
    },
    {
      "epoch": 0.6504131555264665,
      "grad_norm": 0.050710029900074005,
      "learning_rate": 9.590144393852403e-06,
      "loss": 0.0066,
      "step": 39750
    },
    {
      "epoch": 0.6505767814775424,
      "grad_norm": 0.013859692960977554,
      "learning_rate": 9.589766853544248e-06,
      "loss": 0.0037,
      "step": 39760
    },
    {
      "epoch": 0.6507404074286182,
      "grad_norm": 0.43363457918167114,
      "learning_rate": 9.58938914686808e-06,
      "loss": 0.0035,
      "step": 39770
    },
    {
      "epoch": 0.650904033379694,
      "grad_norm": 0.1285753697156906,
      "learning_rate": 9.589011273837582e-06,
      "loss": 0.0035,
      "step": 39780
    },
    {
      "epoch": 0.6510676593307698,
      "grad_norm": 0.14213183522224426,
      "learning_rate": 9.588633234466459e-06,
      "loss": 0.0049,
      "step": 39790
    },
    {
      "epoch": 0.6512312852818457,
      "grad_norm": 0.13648316264152527,
      "learning_rate": 9.588255028768405e-06,
      "loss": 0.0048,
      "step": 39800
    },
    {
      "epoch": 0.6513949112329216,
      "grad_norm": 0.22802965342998505,
      "learning_rate": 9.587876656757136e-06,
      "loss": 0.0037,
      "step": 39810
    },
    {
      "epoch": 0.6515585371839974,
      "grad_norm": 0.06534487754106522,
      "learning_rate": 9.587498118446365e-06,
      "loss": 0.0057,
      "step": 39820
    },
    {
      "epoch": 0.6517221631350733,
      "grad_norm": 0.17357707023620605,
      "learning_rate": 9.587119413849812e-06,
      "loss": 0.005,
      "step": 39830
    },
    {
      "epoch": 0.651885789086149,
      "grad_norm": 0.15947392582893372,
      "learning_rate": 9.586740542981206e-06,
      "loss": 0.003,
      "step": 39840
    },
    {
      "epoch": 0.6520494150372249,
      "grad_norm": 0.1313660442829132,
      "learning_rate": 9.58636150585428e-06,
      "loss": 0.0034,
      "step": 39850
    },
    {
      "epoch": 0.6522130409883008,
      "grad_norm": 0.12863187491893768,
      "learning_rate": 9.585982302482772e-06,
      "loss": 0.0044,
      "step": 39860
    },
    {
      "epoch": 0.6523766669393766,
      "grad_norm": 0.2548704147338867,
      "learning_rate": 9.585602932880427e-06,
      "loss": 0.0073,
      "step": 39870
    },
    {
      "epoch": 0.6525402928904525,
      "grad_norm": 0.2391563504934311,
      "learning_rate": 9.585223397060996e-06,
      "loss": 0.0042,
      "step": 39880
    },
    {
      "epoch": 0.6527039188415282,
      "grad_norm": 0.41539669036865234,
      "learning_rate": 9.58484369503824e-06,
      "loss": 0.0049,
      "step": 39890
    },
    {
      "epoch": 0.6528675447926041,
      "grad_norm": 0.2751244604587555,
      "learning_rate": 9.584463826825916e-06,
      "loss": 0.0028,
      "step": 39900
    },
    {
      "epoch": 0.65303117074368,
      "grad_norm": 0.09683585911989212,
      "learning_rate": 9.584083792437798e-06,
      "loss": 0.006,
      "step": 39910
    },
    {
      "epoch": 0.6531947966947558,
      "grad_norm": 0.2703448235988617,
      "learning_rate": 9.583703591887661e-06,
      "loss": 0.0045,
      "step": 39920
    },
    {
      "epoch": 0.6533584226458317,
      "grad_norm": 0.08823627978563309,
      "learning_rate": 9.583323225189286e-06,
      "loss": 0.0027,
      "step": 39930
    },
    {
      "epoch": 0.6535220485969074,
      "grad_norm": 0.5508773922920227,
      "learning_rate": 9.582942692356458e-06,
      "loss": 0.0031,
      "step": 39940
    },
    {
      "epoch": 0.6536856745479833,
      "grad_norm": 0.08115766197443008,
      "learning_rate": 9.582561993402972e-06,
      "loss": 0.0047,
      "step": 39950
    },
    {
      "epoch": 0.6538493004990592,
      "grad_norm": 0.11804918199777603,
      "learning_rate": 9.58218112834263e-06,
      "loss": 0.005,
      "step": 39960
    },
    {
      "epoch": 0.654012926450135,
      "grad_norm": 0.2557043731212616,
      "learning_rate": 9.581800097189233e-06,
      "loss": 0.0039,
      "step": 39970
    },
    {
      "epoch": 0.6541765524012109,
      "grad_norm": 0.1431952267885208,
      "learning_rate": 9.581418899956594e-06,
      "loss": 0.0056,
      "step": 39980
    },
    {
      "epoch": 0.6543401783522866,
      "grad_norm": 0.5375110507011414,
      "learning_rate": 9.581037536658533e-06,
      "loss": 0.0045,
      "step": 39990
    },
    {
      "epoch": 0.6545038043033625,
      "grad_norm": 0.42734554409980774,
      "learning_rate": 9.58065600730887e-06,
      "loss": 0.0024,
      "step": 40000
    },
    {
      "epoch": 0.6546674302544384,
      "grad_norm": 0.19054429233074188,
      "learning_rate": 9.580274311921437e-06,
      "loss": 0.0034,
      "step": 40010
    },
    {
      "epoch": 0.6548310562055142,
      "grad_norm": 0.14221473038196564,
      "learning_rate": 9.57989245051007e-06,
      "loss": 0.006,
      "step": 40020
    },
    {
      "epoch": 0.6549946821565901,
      "grad_norm": 0.14326852560043335,
      "learning_rate": 9.579510423088608e-06,
      "loss": 0.0045,
      "step": 40030
    },
    {
      "epoch": 0.6551583081076658,
      "grad_norm": 0.2570078670978546,
      "learning_rate": 9.579128229670898e-06,
      "loss": 0.0032,
      "step": 40040
    },
    {
      "epoch": 0.6553219340587417,
      "grad_norm": 0.08198045194149017,
      "learning_rate": 9.578745870270796e-06,
      "loss": 0.0029,
      "step": 40050
    },
    {
      "epoch": 0.6554855600098175,
      "grad_norm": 0.3780317008495331,
      "learning_rate": 9.578363344902162e-06,
      "loss": 0.0038,
      "step": 40060
    },
    {
      "epoch": 0.6556491859608934,
      "grad_norm": 0.2337639480829239,
      "learning_rate": 9.57798065357886e-06,
      "loss": 0.0049,
      "step": 40070
    },
    {
      "epoch": 0.6558128119119693,
      "grad_norm": 0.07617715746164322,
      "learning_rate": 9.577597796314762e-06,
      "loss": 0.0032,
      "step": 40080
    },
    {
      "epoch": 0.655976437863045,
      "grad_norm": 0.09904934465885162,
      "learning_rate": 9.577214773123746e-06,
      "loss": 0.0024,
      "step": 40090
    },
    {
      "epoch": 0.6561400638141209,
      "grad_norm": 0.3449536859989166,
      "learning_rate": 9.576831584019697e-06,
      "loss": 0.0024,
      "step": 40100
    },
    {
      "epoch": 0.6563036897651967,
      "grad_norm": 0.34825918078422546,
      "learning_rate": 9.576448229016501e-06,
      "loss": 0.0029,
      "step": 40110
    },
    {
      "epoch": 0.6564673157162726,
      "grad_norm": 0.33838579058647156,
      "learning_rate": 9.576064708128056e-06,
      "loss": 0.0031,
      "step": 40120
    },
    {
      "epoch": 0.6566309416673485,
      "grad_norm": 0.07733172178268433,
      "learning_rate": 9.575681021368265e-06,
      "loss": 0.0058,
      "step": 40130
    },
    {
      "epoch": 0.6567945676184243,
      "grad_norm": 0.2668304443359375,
      "learning_rate": 9.575297168751035e-06,
      "loss": 0.0045,
      "step": 40140
    },
    {
      "epoch": 0.6569581935695001,
      "grad_norm": 0.5039167404174805,
      "learning_rate": 9.574913150290277e-06,
      "loss": 0.0094,
      "step": 40150
    },
    {
      "epoch": 0.6571218195205759,
      "grad_norm": 0.13782106339931488,
      "learning_rate": 9.574528965999913e-06,
      "loss": 0.0031,
      "step": 40160
    },
    {
      "epoch": 0.6572854454716518,
      "grad_norm": 0.2670063078403473,
      "learning_rate": 9.574144615893869e-06,
      "loss": 0.004,
      "step": 40170
    },
    {
      "epoch": 0.6574490714227277,
      "grad_norm": 0.08842339366674423,
      "learning_rate": 9.573760099986077e-06,
      "loss": 0.0041,
      "step": 40180
    },
    {
      "epoch": 0.6576126973738035,
      "grad_norm": 0.06852489709854126,
      "learning_rate": 9.573375418290474e-06,
      "loss": 0.0025,
      "step": 40190
    },
    {
      "epoch": 0.6577763233248793,
      "grad_norm": 0.11243218183517456,
      "learning_rate": 9.572990570821003e-06,
      "loss": 0.0057,
      "step": 40200
    },
    {
      "epoch": 0.6579399492759551,
      "grad_norm": 0.18241071701049805,
      "learning_rate": 9.572605557591614e-06,
      "loss": 0.004,
      "step": 40210
    },
    {
      "epoch": 0.658103575227031,
      "grad_norm": 0.18272794783115387,
      "learning_rate": 9.572220378616266e-06,
      "loss": 0.005,
      "step": 40220
    },
    {
      "epoch": 0.6582672011781069,
      "grad_norm": 0.13834288716316223,
      "learning_rate": 9.571835033908915e-06,
      "loss": 0.0028,
      "step": 40230
    },
    {
      "epoch": 0.6584308271291827,
      "grad_norm": 0.22063976526260376,
      "learning_rate": 9.571449523483536e-06,
      "loss": 0.0039,
      "step": 40240
    },
    {
      "epoch": 0.6585944530802585,
      "grad_norm": 0.0930066704750061,
      "learning_rate": 9.571063847354096e-06,
      "loss": 0.0041,
      "step": 40250
    },
    {
      "epoch": 0.6587580790313343,
      "grad_norm": 0.2133239358663559,
      "learning_rate": 9.570678005534579e-06,
      "loss": 0.0038,
      "step": 40260
    },
    {
      "epoch": 0.6589217049824102,
      "grad_norm": 0.042088355869054794,
      "learning_rate": 9.570291998038972e-06,
      "loss": 0.0021,
      "step": 40270
    },
    {
      "epoch": 0.6590853309334861,
      "grad_norm": 0.20709596574306488,
      "learning_rate": 9.56990582488126e-06,
      "loss": 0.0045,
      "step": 40280
    },
    {
      "epoch": 0.6592489568845619,
      "grad_norm": 0.2154361605644226,
      "learning_rate": 9.569519486075448e-06,
      "loss": 0.0053,
      "step": 40290
    },
    {
      "epoch": 0.6594125828356378,
      "grad_norm": 0.1612267941236496,
      "learning_rate": 9.569132981635536e-06,
      "loss": 0.0053,
      "step": 40300
    },
    {
      "epoch": 0.6595762087867135,
      "grad_norm": 0.25180354714393616,
      "learning_rate": 9.568746311575536e-06,
      "loss": 0.0033,
      "step": 40310
    },
    {
      "epoch": 0.6597398347377894,
      "grad_norm": 0.09668106585741043,
      "learning_rate": 9.568359475909464e-06,
      "loss": 0.0042,
      "step": 40320
    },
    {
      "epoch": 0.6599034606888653,
      "grad_norm": 0.09837563335895538,
      "learning_rate": 9.567972474651338e-06,
      "loss": 0.0062,
      "step": 40330
    },
    {
      "epoch": 0.6600670866399411,
      "grad_norm": 0.24022804200649261,
      "learning_rate": 9.56758530781519e-06,
      "loss": 0.0026,
      "step": 40340
    },
    {
      "epoch": 0.660230712591017,
      "grad_norm": 0.27872827649116516,
      "learning_rate": 9.567197975415052e-06,
      "loss": 0.004,
      "step": 40350
    },
    {
      "epoch": 0.6603943385420927,
      "grad_norm": 0.3612937927246094,
      "learning_rate": 9.566810477464965e-06,
      "loss": 0.0055,
      "step": 40360
    },
    {
      "epoch": 0.6605579644931686,
      "grad_norm": 0.21453024446964264,
      "learning_rate": 9.566422813978974e-06,
      "loss": 0.005,
      "step": 40370
    },
    {
      "epoch": 0.6607215904442445,
      "grad_norm": 0.03908712416887283,
      "learning_rate": 9.56603498497113e-06,
      "loss": 0.0035,
      "step": 40380
    },
    {
      "epoch": 0.6608852163953203,
      "grad_norm": 0.17099544405937195,
      "learning_rate": 9.565646990455494e-06,
      "loss": 0.0038,
      "step": 40390
    },
    {
      "epoch": 0.6610488423463962,
      "grad_norm": 0.0667608231306076,
      "learning_rate": 9.565258830446128e-06,
      "loss": 0.0029,
      "step": 40400
    },
    {
      "epoch": 0.6612124682974719,
      "grad_norm": 0.10545757412910461,
      "learning_rate": 9.5648705049571e-06,
      "loss": 0.0034,
      "step": 40410
    },
    {
      "epoch": 0.6613760942485478,
      "grad_norm": 0.1538141369819641,
      "learning_rate": 9.564482014002487e-06,
      "loss": 0.0033,
      "step": 40420
    },
    {
      "epoch": 0.6615397201996237,
      "grad_norm": 0.3816950023174286,
      "learning_rate": 9.564093357596373e-06,
      "loss": 0.0028,
      "step": 40430
    },
    {
      "epoch": 0.6617033461506995,
      "grad_norm": 0.2177911102771759,
      "learning_rate": 9.563704535752844e-06,
      "loss": 0.0041,
      "step": 40440
    },
    {
      "epoch": 0.6618669721017754,
      "grad_norm": 0.17686927318572998,
      "learning_rate": 9.563315548485993e-06,
      "loss": 0.0021,
      "step": 40450
    },
    {
      "epoch": 0.6620305980528511,
      "grad_norm": 0.37919172644615173,
      "learning_rate": 9.562926395809923e-06,
      "loss": 0.0038,
      "step": 40460
    },
    {
      "epoch": 0.662194224003927,
      "grad_norm": 0.41865721344947815,
      "learning_rate": 9.562537077738737e-06,
      "loss": 0.0032,
      "step": 40470
    },
    {
      "epoch": 0.6623578499550029,
      "grad_norm": 0.12198913097381592,
      "learning_rate": 9.562147594286549e-06,
      "loss": 0.0014,
      "step": 40480
    },
    {
      "epoch": 0.6625214759060787,
      "grad_norm": 0.08148010820150375,
      "learning_rate": 9.561757945467474e-06,
      "loss": 0.0026,
      "step": 40490
    },
    {
      "epoch": 0.6626851018571546,
      "grad_norm": 0.11657990515232086,
      "learning_rate": 9.561368131295637e-06,
      "loss": 0.0033,
      "step": 40500
    },
    {
      "epoch": 0.6628487278082303,
      "grad_norm": 0.3662223517894745,
      "learning_rate": 9.560978151785167e-06,
      "loss": 0.0061,
      "step": 40510
    },
    {
      "epoch": 0.6630123537593062,
      "grad_norm": 0.2581077218055725,
      "learning_rate": 9.560588006950204e-06,
      "loss": 0.0032,
      "step": 40520
    },
    {
      "epoch": 0.6631759797103821,
      "grad_norm": 0.30310264229774475,
      "learning_rate": 9.560197696804885e-06,
      "loss": 0.0043,
      "step": 40530
    },
    {
      "epoch": 0.6633396056614579,
      "grad_norm": 0.203278049826622,
      "learning_rate": 9.55980722136336e-06,
      "loss": 0.0027,
      "step": 40540
    },
    {
      "epoch": 0.6635032316125338,
      "grad_norm": 0.0290276687592268,
      "learning_rate": 9.559416580639783e-06,
      "loss": 0.0035,
      "step": 40550
    },
    {
      "epoch": 0.6636668575636095,
      "grad_norm": 0.2354142963886261,
      "learning_rate": 9.559025774648315e-06,
      "loss": 0.0042,
      "step": 40560
    },
    {
      "epoch": 0.6638304835146854,
      "grad_norm": 0.1471615433692932,
      "learning_rate": 9.558634803403117e-06,
      "loss": 0.0053,
      "step": 40570
    },
    {
      "epoch": 0.6639941094657613,
      "grad_norm": 0.21577584743499756,
      "learning_rate": 9.558243666918362e-06,
      "loss": 0.0042,
      "step": 40580
    },
    {
      "epoch": 0.6641577354168371,
      "grad_norm": 0.30954912304878235,
      "learning_rate": 9.557852365208234e-06,
      "loss": 0.0041,
      "step": 40590
    },
    {
      "epoch": 0.664321361367913,
      "grad_norm": 0.37253338098526,
      "learning_rate": 9.55746089828691e-06,
      "loss": 0.004,
      "step": 40600
    },
    {
      "epoch": 0.6644849873189888,
      "grad_norm": 0.17140115797519684,
      "learning_rate": 9.557069266168585e-06,
      "loss": 0.0045,
      "step": 40610
    },
    {
      "epoch": 0.6646486132700646,
      "grad_norm": 0.04907820373773575,
      "learning_rate": 9.556677468867448e-06,
      "loss": 0.0026,
      "step": 40620
    },
    {
      "epoch": 0.6648122392211405,
      "grad_norm": 0.02016725391149521,
      "learning_rate": 9.556285506397705e-06,
      "loss": 0.0041,
      "step": 40630
    },
    {
      "epoch": 0.6649758651722163,
      "grad_norm": 0.150553897023201,
      "learning_rate": 9.555893378773564e-06,
      "loss": 0.0031,
      "step": 40640
    },
    {
      "epoch": 0.6651394911232922,
      "grad_norm": 0.176218643784523,
      "learning_rate": 9.555501086009236e-06,
      "loss": 0.0047,
      "step": 40650
    },
    {
      "epoch": 0.665303117074368,
      "grad_norm": 0.1619766801595688,
      "learning_rate": 9.555108628118945e-06,
      "loss": 0.004,
      "step": 40660
    },
    {
      "epoch": 0.6654667430254438,
      "grad_norm": 0.3033839762210846,
      "learning_rate": 9.554716005116913e-06,
      "loss": 0.0035,
      "step": 40670
    },
    {
      "epoch": 0.6656303689765197,
      "grad_norm": 0.2725808620452881,
      "learning_rate": 9.554323217017373e-06,
      "loss": 0.005,
      "step": 40680
    },
    {
      "epoch": 0.6657939949275955,
      "grad_norm": 0.1550709456205368,
      "learning_rate": 9.553930263834563e-06,
      "loss": 0.0039,
      "step": 40690
    },
    {
      "epoch": 0.6659576208786714,
      "grad_norm": 0.36677852272987366,
      "learning_rate": 9.553537145582726e-06,
      "loss": 0.0031,
      "step": 40700
    },
    {
      "epoch": 0.6661212468297472,
      "grad_norm": 0.195937842130661,
      "learning_rate": 9.553143862276111e-06,
      "loss": 0.0053,
      "step": 40710
    },
    {
      "epoch": 0.666284872780823,
      "grad_norm": 0.22115139663219452,
      "learning_rate": 9.552750413928973e-06,
      "loss": 0.0043,
      "step": 40720
    },
    {
      "epoch": 0.6664484987318989,
      "grad_norm": 0.16133953630924225,
      "learning_rate": 9.552356800555575e-06,
      "loss": 0.0039,
      "step": 40730
    },
    {
      "epoch": 0.6666121246829747,
      "grad_norm": 0.1360553503036499,
      "learning_rate": 9.551963022170186e-06,
      "loss": 0.0038,
      "step": 40740
    },
    {
      "epoch": 0.6667757506340506,
      "grad_norm": 0.09650354832410812,
      "learning_rate": 9.551569078787077e-06,
      "loss": 0.0021,
      "step": 40750
    },
    {
      "epoch": 0.6669393765851264,
      "grad_norm": 0.08582047373056412,
      "learning_rate": 9.55117497042053e-06,
      "loss": 0.0026,
      "step": 40760
    },
    {
      "epoch": 0.6671030025362022,
      "grad_norm": 0.1923443078994751,
      "learning_rate": 9.550780697084826e-06,
      "loss": 0.0025,
      "step": 40770
    },
    {
      "epoch": 0.6672666284872781,
      "grad_norm": 0.311466246843338,
      "learning_rate": 9.55038625879426e-06,
      "loss": 0.0045,
      "step": 40780
    },
    {
      "epoch": 0.6674302544383539,
      "grad_norm": 0.1760646551847458,
      "learning_rate": 9.549991655563129e-06,
      "loss": 0.0026,
      "step": 40790
    },
    {
      "epoch": 0.6675938803894298,
      "grad_norm": 0.1887482851743698,
      "learning_rate": 9.549596887405736e-06,
      "loss": 0.0025,
      "step": 40800
    },
    {
      "epoch": 0.6677575063405056,
      "grad_norm": 0.2653856873512268,
      "learning_rate": 9.549201954336392e-06,
      "loss": 0.0041,
      "step": 40810
    },
    {
      "epoch": 0.6679211322915815,
      "grad_norm": 0.1052040383219719,
      "learning_rate": 9.548806856369407e-06,
      "loss": 0.0014,
      "step": 40820
    },
    {
      "epoch": 0.6680847582426573,
      "grad_norm": 0.07335983961820602,
      "learning_rate": 9.548411593519112e-06,
      "loss": 0.0025,
      "step": 40830
    },
    {
      "epoch": 0.6682483841937331,
      "grad_norm": 0.06771315634250641,
      "learning_rate": 9.548016165799824e-06,
      "loss": 0.0031,
      "step": 40840
    },
    {
      "epoch": 0.668412010144809,
      "grad_norm": 0.2092895209789276,
      "learning_rate": 9.547620573225882e-06,
      "loss": 0.0091,
      "step": 40850
    },
    {
      "epoch": 0.6685756360958848,
      "grad_norm": 0.14687207341194153,
      "learning_rate": 9.547224815811624e-06,
      "loss": 0.0051,
      "step": 40860
    },
    {
      "epoch": 0.6687392620469607,
      "grad_norm": 0.22625429928302765,
      "learning_rate": 9.546828893571395e-06,
      "loss": 0.0043,
      "step": 40870
    },
    {
      "epoch": 0.6689028879980365,
      "grad_norm": 0.10107409209012985,
      "learning_rate": 9.546432806519548e-06,
      "loss": 0.0035,
      "step": 40880
    },
    {
      "epoch": 0.6690665139491123,
      "grad_norm": 0.08284055441617966,
      "learning_rate": 9.546036554670438e-06,
      "loss": 0.0029,
      "step": 40890
    },
    {
      "epoch": 0.6692301399001882,
      "grad_norm": 0.033760856837034225,
      "learning_rate": 9.54564013803843e-06,
      "loss": 0.0036,
      "step": 40900
    },
    {
      "epoch": 0.669393765851264,
      "grad_norm": 0.08804108202457428,
      "learning_rate": 9.545243556637892e-06,
      "loss": 0.0031,
      "step": 40910
    },
    {
      "epoch": 0.6695573918023399,
      "grad_norm": 0.30136144161224365,
      "learning_rate": 9.544846810483197e-06,
      "loss": 0.003,
      "step": 40920
    },
    {
      "epoch": 0.6697210177534156,
      "grad_norm": 0.17357979714870453,
      "learning_rate": 9.54444989958873e-06,
      "loss": 0.0034,
      "step": 40930
    },
    {
      "epoch": 0.6698846437044915,
      "grad_norm": 0.11663365364074707,
      "learning_rate": 9.544052823968876e-06,
      "loss": 0.0063,
      "step": 40940
    },
    {
      "epoch": 0.6700482696555674,
      "grad_norm": 0.09595862776041031,
      "learning_rate": 9.54365558363803e-06,
      "loss": 0.0035,
      "step": 40950
    },
    {
      "epoch": 0.6702118956066432,
      "grad_norm": 0.0594622977077961,
      "learning_rate": 9.543258178610587e-06,
      "loss": 0.0089,
      "step": 40960
    },
    {
      "epoch": 0.6703755215577191,
      "grad_norm": 0.2683613896369934,
      "learning_rate": 9.542860608900958e-06,
      "loss": 0.0036,
      "step": 40970
    },
    {
      "epoch": 0.6705391475087948,
      "grad_norm": 0.15494564175605774,
      "learning_rate": 9.542462874523549e-06,
      "loss": 0.0022,
      "step": 40980
    },
    {
      "epoch": 0.6707027734598707,
      "grad_norm": 0.40871626138687134,
      "learning_rate": 9.542064975492776e-06,
      "loss": 0.0036,
      "step": 40990
    },
    {
      "epoch": 0.6708663994109466,
      "grad_norm": 0.13518080115318298,
      "learning_rate": 9.541666911823066e-06,
      "loss": 0.0038,
      "step": 41000
    },
    {
      "epoch": 0.6710300253620224,
      "grad_norm": 0.08975598216056824,
      "learning_rate": 9.541268683528846e-06,
      "loss": 0.0026,
      "step": 41010
    },
    {
      "epoch": 0.6711936513130983,
      "grad_norm": 0.19219611585140228,
      "learning_rate": 9.540870290624551e-06,
      "loss": 0.003,
      "step": 41020
    },
    {
      "epoch": 0.671357277264174,
      "grad_norm": 0.07628770172595978,
      "learning_rate": 9.54047173312462e-06,
      "loss": 0.003,
      "step": 41030
    },
    {
      "epoch": 0.6715209032152499,
      "grad_norm": 0.11276376247406006,
      "learning_rate": 9.540073011043504e-06,
      "loss": 0.0037,
      "step": 41040
    },
    {
      "epoch": 0.6716845291663258,
      "grad_norm": 0.2766047418117523,
      "learning_rate": 9.539674124395652e-06,
      "loss": 0.005,
      "step": 41050
    },
    {
      "epoch": 0.6718481551174016,
      "grad_norm": 0.040906086564064026,
      "learning_rate": 9.539275073195522e-06,
      "loss": 0.0022,
      "step": 41060
    },
    {
      "epoch": 0.6720117810684775,
      "grad_norm": 0.12412739545106888,
      "learning_rate": 9.538875857457583e-06,
      "loss": 0.0029,
      "step": 41070
    },
    {
      "epoch": 0.6721754070195533,
      "grad_norm": 0.10310149192810059,
      "learning_rate": 9.538476477196302e-06,
      "loss": 0.0037,
      "step": 41080
    },
    {
      "epoch": 0.6723390329706291,
      "grad_norm": 0.10124338418245316,
      "learning_rate": 9.538076932426157e-06,
      "loss": 0.0027,
      "step": 41090
    },
    {
      "epoch": 0.672502658921705,
      "grad_norm": 0.25430813431739807,
      "learning_rate": 9.53767722316163e-06,
      "loss": 0.0037,
      "step": 41100
    },
    {
      "epoch": 0.6726662848727808,
      "grad_norm": 0.15678854286670685,
      "learning_rate": 9.53727734941721e-06,
      "loss": 0.0043,
      "step": 41110
    },
    {
      "epoch": 0.6728299108238567,
      "grad_norm": 0.3007122576236725,
      "learning_rate": 9.536877311207391e-06,
      "loss": 0.0035,
      "step": 41120
    },
    {
      "epoch": 0.6729935367749325,
      "grad_norm": 0.2486082762479782,
      "learning_rate": 9.536477108546674e-06,
      "loss": 0.0037,
      "step": 41130
    },
    {
      "epoch": 0.6731571627260083,
      "grad_norm": 0.23905012011528015,
      "learning_rate": 9.536076741449563e-06,
      "loss": 0.0031,
      "step": 41140
    },
    {
      "epoch": 0.6733207886770842,
      "grad_norm": 0.08154893666505814,
      "learning_rate": 9.535676209930574e-06,
      "loss": 0.003,
      "step": 41150
    },
    {
      "epoch": 0.67348441462816,
      "grad_norm": 0.19384953379631042,
      "learning_rate": 9.535275514004222e-06,
      "loss": 0.0039,
      "step": 41160
    },
    {
      "epoch": 0.6736480405792359,
      "grad_norm": 0.16193881630897522,
      "learning_rate": 9.534874653685036e-06,
      "loss": 0.0033,
      "step": 41170
    },
    {
      "epoch": 0.6738116665303117,
      "grad_norm": 0.3026077449321747,
      "learning_rate": 9.534473628987542e-06,
      "loss": 0.0035,
      "step": 41180
    },
    {
      "epoch": 0.6739752924813875,
      "grad_norm": 0.12581059336662292,
      "learning_rate": 9.534072439926276e-06,
      "loss": 0.0042,
      "step": 41190
    },
    {
      "epoch": 0.6741389184324634,
      "grad_norm": 0.21835175156593323,
      "learning_rate": 9.533671086515781e-06,
      "loss": 0.0041,
      "step": 41200
    },
    {
      "epoch": 0.6743025443835392,
      "grad_norm": 0.0905032530426979,
      "learning_rate": 9.533269568770607e-06,
      "loss": 0.002,
      "step": 41210
    },
    {
      "epoch": 0.6744661703346151,
      "grad_norm": 0.3723595440387726,
      "learning_rate": 9.532867886705305e-06,
      "loss": 0.0044,
      "step": 41220
    },
    {
      "epoch": 0.6746297962856909,
      "grad_norm": 0.3325466513633728,
      "learning_rate": 9.532466040334438e-06,
      "loss": 0.0051,
      "step": 41230
    },
    {
      "epoch": 0.6747934222367667,
      "grad_norm": 0.2079477608203888,
      "learning_rate": 9.53206402967257e-06,
      "loss": 0.0044,
      "step": 41240
    },
    {
      "epoch": 0.6749570481878426,
      "grad_norm": 0.36569222807884216,
      "learning_rate": 9.531661854734275e-06,
      "loss": 0.0027,
      "step": 41250
    },
    {
      "epoch": 0.6751206741389184,
      "grad_norm": 0.42900118231773376,
      "learning_rate": 9.531259515534127e-06,
      "loss": 0.0081,
      "step": 41260
    },
    {
      "epoch": 0.6752843000899943,
      "grad_norm": 0.17705038189888,
      "learning_rate": 9.530857012086712e-06,
      "loss": 0.0048,
      "step": 41270
    },
    {
      "epoch": 0.6754479260410701,
      "grad_norm": 0.17323188483715057,
      "learning_rate": 9.53045434440662e-06,
      "loss": 0.0028,
      "step": 41280
    },
    {
      "epoch": 0.675611551992146,
      "grad_norm": 0.1745935082435608,
      "learning_rate": 9.530051512508448e-06,
      "loss": 0.0061,
      "step": 41290
    },
    {
      "epoch": 0.6757751779432218,
      "grad_norm": 0.21723127365112305,
      "learning_rate": 9.529648516406798e-06,
      "loss": 0.0041,
      "step": 41300
    },
    {
      "epoch": 0.6759388038942976,
      "grad_norm": 0.2178402692079544,
      "learning_rate": 9.529245356116273e-06,
      "loss": 0.0049,
      "step": 41310
    },
    {
      "epoch": 0.6761024298453735,
      "grad_norm": 0.205748051404953,
      "learning_rate": 9.528842031651491e-06,
      "loss": 0.0031,
      "step": 41320
    },
    {
      "epoch": 0.6762660557964493,
      "grad_norm": 0.04753841832280159,
      "learning_rate": 9.52843854302707e-06,
      "loss": 0.0042,
      "step": 41330
    },
    {
      "epoch": 0.6764296817475252,
      "grad_norm": 0.1326904147863388,
      "learning_rate": 9.528034890257635e-06,
      "loss": 0.0066,
      "step": 41340
    },
    {
      "epoch": 0.676593307698601,
      "grad_norm": 0.11043822765350342,
      "learning_rate": 9.527631073357818e-06,
      "loss": 0.0021,
      "step": 41350
    },
    {
      "epoch": 0.6767569336496768,
      "grad_norm": 0.4312784671783447,
      "learning_rate": 9.527227092342258e-06,
      "loss": 0.0027,
      "step": 41360
    },
    {
      "epoch": 0.6769205596007527,
      "grad_norm": 0.07799684256315231,
      "learning_rate": 9.526822947225597e-06,
      "loss": 0.0018,
      "step": 41370
    },
    {
      "epoch": 0.6770841855518285,
      "grad_norm": 0.39989206194877625,
      "learning_rate": 9.52641863802248e-06,
      "loss": 0.0043,
      "step": 41380
    },
    {
      "epoch": 0.6772478115029044,
      "grad_norm": 0.17345643043518066,
      "learning_rate": 9.52601416474757e-06,
      "loss": 0.0022,
      "step": 41390
    },
    {
      "epoch": 0.6774114374539802,
      "grad_norm": 0.040480829775333405,
      "learning_rate": 9.525609527415524e-06,
      "loss": 0.0029,
      "step": 41400
    },
    {
      "epoch": 0.677575063405056,
      "grad_norm": 0.2246304303407669,
      "learning_rate": 9.525204726041009e-06,
      "loss": 0.0058,
      "step": 41410
    },
    {
      "epoch": 0.6777386893561319,
      "grad_norm": 0.19318869709968567,
      "learning_rate": 9.524799760638698e-06,
      "loss": 0.004,
      "step": 41420
    },
    {
      "epoch": 0.6779023153072077,
      "grad_norm": 0.1640588790178299,
      "learning_rate": 9.524394631223272e-06,
      "loss": 0.0027,
      "step": 41430
    },
    {
      "epoch": 0.6780659412582836,
      "grad_norm": 0.15712791681289673,
      "learning_rate": 9.523989337809417e-06,
      "loss": 0.003,
      "step": 41440
    },
    {
      "epoch": 0.6782295672093595,
      "grad_norm": 0.2708931863307953,
      "learning_rate": 9.523583880411817e-06,
      "loss": 0.0022,
      "step": 41450
    },
    {
      "epoch": 0.6783931931604352,
      "grad_norm": 0.13915149867534637,
      "learning_rate": 9.523178259045176e-06,
      "loss": 0.0047,
      "step": 41460
    },
    {
      "epoch": 0.6785568191115111,
      "grad_norm": 0.13243943452835083,
      "learning_rate": 9.522772473724194e-06,
      "loss": 0.0039,
      "step": 41470
    },
    {
      "epoch": 0.6787204450625869,
      "grad_norm": 0.24703291058540344,
      "learning_rate": 9.52236652446358e-06,
      "loss": 0.0035,
      "step": 41480
    },
    {
      "epoch": 0.6788840710136628,
      "grad_norm": 0.1995680183172226,
      "learning_rate": 9.521960411278049e-06,
      "loss": 0.0041,
      "step": 41490
    },
    {
      "epoch": 0.6790476969647387,
      "grad_norm": 0.19879138469696045,
      "learning_rate": 9.52155413418232e-06,
      "loss": 0.006,
      "step": 41500
    },
    {
      "epoch": 0.6792113229158144,
      "grad_norm": 0.32783085107803345,
      "learning_rate": 9.521147693191121e-06,
      "loss": 0.0027,
      "step": 41510
    },
    {
      "epoch": 0.6793749488668903,
      "grad_norm": 0.16779537498950958,
      "learning_rate": 9.520741088319188e-06,
      "loss": 0.0024,
      "step": 41520
    },
    {
      "epoch": 0.6795385748179661,
      "grad_norm": 0.12358634173870087,
      "learning_rate": 9.520334319581252e-06,
      "loss": 0.0024,
      "step": 41530
    },
    {
      "epoch": 0.679702200769042,
      "grad_norm": 0.22999624907970428,
      "learning_rate": 9.519927386992063e-06,
      "loss": 0.0045,
      "step": 41540
    },
    {
      "epoch": 0.6798658267201179,
      "grad_norm": 0.11387041956186295,
      "learning_rate": 9.51952029056637e-06,
      "loss": 0.0034,
      "step": 41550
    },
    {
      "epoch": 0.6800294526711936,
      "grad_norm": 0.25392088294029236,
      "learning_rate": 9.519113030318927e-06,
      "loss": 0.0057,
      "step": 41560
    },
    {
      "epoch": 0.6801930786222695,
      "grad_norm": 0.4185047745704651,
      "learning_rate": 9.518705606264497e-06,
      "loss": 0.004,
      "step": 41570
    },
    {
      "epoch": 0.6803567045733453,
      "grad_norm": 0.16185013949871063,
      "learning_rate": 9.518298018417852e-06,
      "loss": 0.003,
      "step": 41580
    },
    {
      "epoch": 0.6805203305244212,
      "grad_norm": 0.18516352772712708,
      "learning_rate": 9.517890266793762e-06,
      "loss": 0.0035,
      "step": 41590
    },
    {
      "epoch": 0.6806839564754971,
      "grad_norm": 0.2935173213481903,
      "learning_rate": 9.51748235140701e-06,
      "loss": 0.0034,
      "step": 41600
    },
    {
      "epoch": 0.6808475824265728,
      "grad_norm": 0.09066560119390488,
      "learning_rate": 9.517074272272378e-06,
      "loss": 0.0041,
      "step": 41610
    },
    {
      "epoch": 0.6810112083776487,
      "grad_norm": 0.8889887928962708,
      "learning_rate": 9.516666029404661e-06,
      "loss": 0.0043,
      "step": 41620
    },
    {
      "epoch": 0.6811748343287245,
      "grad_norm": 0.3778586983680725,
      "learning_rate": 9.516257622818657e-06,
      "loss": 0.0034,
      "step": 41630
    },
    {
      "epoch": 0.6813384602798004,
      "grad_norm": 0.13633938133716583,
      "learning_rate": 9.515849052529167e-06,
      "loss": 0.0061,
      "step": 41640
    },
    {
      "epoch": 0.6815020862308763,
      "grad_norm": 0.1451253741979599,
      "learning_rate": 9.515440318551002e-06,
      "loss": 0.0036,
      "step": 41650
    },
    {
      "epoch": 0.681665712181952,
      "grad_norm": 0.06916622817516327,
      "learning_rate": 9.515031420898979e-06,
      "loss": 0.0037,
      "step": 41660
    },
    {
      "epoch": 0.6818293381330279,
      "grad_norm": 0.11693932861089706,
      "learning_rate": 9.51462235958792e-06,
      "loss": 0.0038,
      "step": 41670
    },
    {
      "epoch": 0.6819929640841037,
      "grad_norm": 0.1786661595106125,
      "learning_rate": 9.51421313463265e-06,
      "loss": 0.0041,
      "step": 41680
    },
    {
      "epoch": 0.6821565900351796,
      "grad_norm": 0.17368175089359283,
      "learning_rate": 9.513803746048002e-06,
      "loss": 0.0021,
      "step": 41690
    },
    {
      "epoch": 0.6823202159862555,
      "grad_norm": 0.18363507091999054,
      "learning_rate": 9.51339419384882e-06,
      "loss": 0.0023,
      "step": 41700
    },
    {
      "epoch": 0.6824838419373312,
      "grad_norm": 0.2775869369506836,
      "learning_rate": 9.512984478049942e-06,
      "loss": 0.0036,
      "step": 41710
    },
    {
      "epoch": 0.6826474678884071,
      "grad_norm": 0.3735685348510742,
      "learning_rate": 9.512574598666226e-06,
      "loss": 0.0023,
      "step": 41720
    },
    {
      "epoch": 0.6828110938394829,
      "grad_norm": 0.13980357348918915,
      "learning_rate": 9.512164555712526e-06,
      "loss": 0.0022,
      "step": 41730
    },
    {
      "epoch": 0.6829747197905588,
      "grad_norm": 0.19643956422805786,
      "learning_rate": 9.511754349203705e-06,
      "loss": 0.0029,
      "step": 41740
    },
    {
      "epoch": 0.6831383457416347,
      "grad_norm": 0.07771702855825424,
      "learning_rate": 9.511343979154635e-06,
      "loss": 0.0062,
      "step": 41750
    },
    {
      "epoch": 0.6833019716927105,
      "grad_norm": 0.05773426964879036,
      "learning_rate": 9.510933445580185e-06,
      "loss": 0.0021,
      "step": 41760
    },
    {
      "epoch": 0.6834655976437863,
      "grad_norm": 0.2177582085132599,
      "learning_rate": 9.510522748495243e-06,
      "loss": 0.0037,
      "step": 41770
    },
    {
      "epoch": 0.6836292235948621,
      "grad_norm": 0.19544053077697754,
      "learning_rate": 9.51011188791469e-06,
      "loss": 0.0033,
      "step": 41780
    },
    {
      "epoch": 0.683792849545938,
      "grad_norm": 0.11291540414094925,
      "learning_rate": 9.50970086385342e-06,
      "loss": 0.0076,
      "step": 41790
    },
    {
      "epoch": 0.6839564754970138,
      "grad_norm": 0.18975688517093658,
      "learning_rate": 9.509289676326334e-06,
      "loss": 0.0033,
      "step": 41800
    },
    {
      "epoch": 0.6841201014480897,
      "grad_norm": 0.22611401975154877,
      "learning_rate": 9.508878325348334e-06,
      "loss": 0.003,
      "step": 41810
    },
    {
      "epoch": 0.6842837273991655,
      "grad_norm": 0.3169178068637848,
      "learning_rate": 9.508466810934332e-06,
      "loss": 0.0052,
      "step": 41820
    },
    {
      "epoch": 0.6844473533502413,
      "grad_norm": 0.363726407289505,
      "learning_rate": 9.508055133099244e-06,
      "loss": 0.0045,
      "step": 41830
    },
    {
      "epoch": 0.6846109793013172,
      "grad_norm": 0.14350302517414093,
      "learning_rate": 9.507643291857992e-06,
      "loss": 0.0036,
      "step": 41840
    },
    {
      "epoch": 0.684774605252393,
      "grad_norm": 0.12181004881858826,
      "learning_rate": 9.507231287225504e-06,
      "loss": 0.0031,
      "step": 41850
    },
    {
      "epoch": 0.6849382312034689,
      "grad_norm": 0.04048116132616997,
      "learning_rate": 9.506819119216717e-06,
      "loss": 0.0029,
      "step": 41860
    },
    {
      "epoch": 0.6851018571545447,
      "grad_norm": 0.24783559143543243,
      "learning_rate": 9.506406787846567e-06,
      "loss": 0.0021,
      "step": 41870
    },
    {
      "epoch": 0.6852654831056205,
      "grad_norm": 0.07389844954013824,
      "learning_rate": 9.50599429313e-06,
      "loss": 0.0067,
      "step": 41880
    },
    {
      "epoch": 0.6854291090566964,
      "grad_norm": 0.3724473714828491,
      "learning_rate": 9.505581635081972e-06,
      "loss": 0.0034,
      "step": 41890
    },
    {
      "epoch": 0.6855927350077722,
      "grad_norm": 0.5259312391281128,
      "learning_rate": 9.50516881371744e-06,
      "loss": 0.0052,
      "step": 41900
    },
    {
      "epoch": 0.6857563609588481,
      "grad_norm": 0.19595074653625488,
      "learning_rate": 9.504755829051364e-06,
      "loss": 0.0034,
      "step": 41910
    },
    {
      "epoch": 0.685919986909924,
      "grad_norm": 0.2707808017730713,
      "learning_rate": 9.504342681098717e-06,
      "loss": 0.0037,
      "step": 41920
    },
    {
      "epoch": 0.6860836128609997,
      "grad_norm": 0.1302749514579773,
      "learning_rate": 9.503929369874473e-06,
      "loss": 0.0042,
      "step": 41930
    },
    {
      "epoch": 0.6862472388120756,
      "grad_norm": 0.16718685626983643,
      "learning_rate": 9.503515895393615e-06,
      "loss": 0.0044,
      "step": 41940
    },
    {
      "epoch": 0.6864108647631514,
      "grad_norm": 0.25853249430656433,
      "learning_rate": 9.503102257671128e-06,
      "loss": 0.0032,
      "step": 41950
    },
    {
      "epoch": 0.6865744907142273,
      "grad_norm": 0.13117872178554535,
      "learning_rate": 9.502688456722008e-06,
      "loss": 0.0049,
      "step": 41960
    },
    {
      "epoch": 0.6867381166653032,
      "grad_norm": 0.3355722725391388,
      "learning_rate": 9.502274492561256e-06,
      "loss": 0.0039,
      "step": 41970
    },
    {
      "epoch": 0.6869017426163789,
      "grad_norm": 0.11014098674058914,
      "learning_rate": 9.501860365203873e-06,
      "loss": 0.0035,
      "step": 41980
    },
    {
      "epoch": 0.6870653685674548,
      "grad_norm": 0.3031855523586273,
      "learning_rate": 9.501446074664871e-06,
      "loss": 0.0061,
      "step": 41990
    },
    {
      "epoch": 0.6872289945185306,
      "grad_norm": 0.11191651225090027,
      "learning_rate": 9.501031620959267e-06,
      "loss": 0.0039,
      "step": 42000
    },
    {
      "epoch": 0.6873926204696065,
      "grad_norm": 0.2681221663951874,
      "learning_rate": 9.500617004102086e-06,
      "loss": 0.0042,
      "step": 42010
    },
    {
      "epoch": 0.6875562464206824,
      "grad_norm": 0.11693637073040009,
      "learning_rate": 9.500202224108355e-06,
      "loss": 0.0025,
      "step": 42020
    },
    {
      "epoch": 0.6877198723717581,
      "grad_norm": 0.2281288355588913,
      "learning_rate": 9.499787280993109e-06,
      "loss": 0.0023,
      "step": 42030
    },
    {
      "epoch": 0.687883498322834,
      "grad_norm": 0.13584215939044952,
      "learning_rate": 9.499372174771389e-06,
      "loss": 0.0045,
      "step": 42040
    },
    {
      "epoch": 0.6880471242739098,
      "grad_norm": 0.3787764310836792,
      "learning_rate": 9.498956905458244e-06,
      "loss": 0.0041,
      "step": 42050
    },
    {
      "epoch": 0.6882107502249857,
      "grad_norm": 0.2026219516992569,
      "learning_rate": 9.49854147306872e-06,
      "loss": 0.0036,
      "step": 42060
    },
    {
      "epoch": 0.6883743761760616,
      "grad_norm": 0.18095897138118744,
      "learning_rate": 9.498125877617884e-06,
      "loss": 0.0042,
      "step": 42070
    },
    {
      "epoch": 0.6885380021271373,
      "grad_norm": 0.1485137939453125,
      "learning_rate": 9.497710119120793e-06,
      "loss": 0.0048,
      "step": 42080
    },
    {
      "epoch": 0.6887016280782132,
      "grad_norm": 0.32534754276275635,
      "learning_rate": 9.49729419759252e-06,
      "loss": 0.0055,
      "step": 42090
    },
    {
      "epoch": 0.688865254029289,
      "grad_norm": 0.0838620588183403,
      "learning_rate": 9.496878113048143e-06,
      "loss": 0.0018,
      "step": 42100
    },
    {
      "epoch": 0.6890288799803649,
      "grad_norm": 4.149435520172119,
      "learning_rate": 9.49646186550274e-06,
      "loss": 0.008,
      "step": 42110
    },
    {
      "epoch": 0.6891925059314408,
      "grad_norm": 0.23951877653598785,
      "learning_rate": 9.496045454971403e-06,
      "loss": 0.0032,
      "step": 42120
    },
    {
      "epoch": 0.6893561318825165,
      "grad_norm": 0.2085265964269638,
      "learning_rate": 9.495628881469225e-06,
      "loss": 0.0052,
      "step": 42130
    },
    {
      "epoch": 0.6895197578335924,
      "grad_norm": 0.30439940094947815,
      "learning_rate": 9.495212145011303e-06,
      "loss": 0.0052,
      "step": 42140
    },
    {
      "epoch": 0.6896833837846682,
      "grad_norm": 0.05532262846827507,
      "learning_rate": 9.494795245612747e-06,
      "loss": 0.0028,
      "step": 42150
    },
    {
      "epoch": 0.6898470097357441,
      "grad_norm": 0.20707818865776062,
      "learning_rate": 9.494378183288665e-06,
      "loss": 0.006,
      "step": 42160
    },
    {
      "epoch": 0.69001063568682,
      "grad_norm": 0.09967313706874847,
      "learning_rate": 9.493960958054175e-06,
      "loss": 0.0035,
      "step": 42170
    },
    {
      "epoch": 0.6901742616378957,
      "grad_norm": 0.20637306571006775,
      "learning_rate": 9.493543569924403e-06,
      "loss": 0.0021,
      "step": 42180
    },
    {
      "epoch": 0.6903378875889716,
      "grad_norm": 0.2893702983856201,
      "learning_rate": 9.493126018914476e-06,
      "loss": 0.0061,
      "step": 42190
    },
    {
      "epoch": 0.6905015135400474,
      "grad_norm": 0.2046521008014679,
      "learning_rate": 9.492708305039529e-06,
      "loss": 0.0041,
      "step": 42200
    },
    {
      "epoch": 0.6906651394911233,
      "grad_norm": 0.18706543743610382,
      "learning_rate": 9.492290428314703e-06,
      "loss": 0.0026,
      "step": 42210
    },
    {
      "epoch": 0.6908287654421992,
      "grad_norm": 0.31529700756073,
      "learning_rate": 9.491872388755148e-06,
      "loss": 0.0025,
      "step": 42220
    },
    {
      "epoch": 0.690992391393275,
      "grad_norm": 0.17167064547538757,
      "learning_rate": 9.491454186376015e-06,
      "loss": 0.0049,
      "step": 42230
    },
    {
      "epoch": 0.6911560173443508,
      "grad_norm": 0.25095802545547485,
      "learning_rate": 9.49103582119246e-06,
      "loss": 0.002,
      "step": 42240
    },
    {
      "epoch": 0.6913196432954266,
      "grad_norm": 0.11241482198238373,
      "learning_rate": 9.490617293219652e-06,
      "loss": 0.002,
      "step": 42250
    },
    {
      "epoch": 0.6914832692465025,
      "grad_norm": 0.36931461095809937,
      "learning_rate": 9.49019860247276e-06,
      "loss": 0.004,
      "step": 42260
    },
    {
      "epoch": 0.6916468951975784,
      "grad_norm": 0.06430893391370773,
      "learning_rate": 9.489779748966962e-06,
      "loss": 0.0058,
      "step": 42270
    },
    {
      "epoch": 0.6918105211486542,
      "grad_norm": 0.19548408687114716,
      "learning_rate": 9.489360732717438e-06,
      "loss": 0.0058,
      "step": 42280
    },
    {
      "epoch": 0.69197414709973,
      "grad_norm": 0.4175778329372406,
      "learning_rate": 9.488941553739377e-06,
      "loss": 0.0074,
      "step": 42290
    },
    {
      "epoch": 0.6921377730508058,
      "grad_norm": 0.1676895022392273,
      "learning_rate": 9.488522212047972e-06,
      "loss": 0.0047,
      "step": 42300
    },
    {
      "epoch": 0.6923013990018817,
      "grad_norm": 0.09545018523931503,
      "learning_rate": 9.488102707658428e-06,
      "loss": 0.0037,
      "step": 42310
    },
    {
      "epoch": 0.6924650249529576,
      "grad_norm": 0.15663392841815948,
      "learning_rate": 9.487683040585945e-06,
      "loss": 0.0036,
      "step": 42320
    },
    {
      "epoch": 0.6926286509040334,
      "grad_norm": 0.23286955058574677,
      "learning_rate": 9.487263210845739e-06,
      "loss": 0.0034,
      "step": 42330
    },
    {
      "epoch": 0.6927922768551092,
      "grad_norm": 0.13943170011043549,
      "learning_rate": 9.486843218453026e-06,
      "loss": 0.0033,
      "step": 42340
    },
    {
      "epoch": 0.692955902806185,
      "grad_norm": 0.15861579775810242,
      "learning_rate": 9.486423063423031e-06,
      "loss": 0.0031,
      "step": 42350
    },
    {
      "epoch": 0.6931195287572609,
      "grad_norm": 0.09583029896020889,
      "learning_rate": 9.48600274577098e-06,
      "loss": 0.0042,
      "step": 42360
    },
    {
      "epoch": 0.6932831547083368,
      "grad_norm": 0.11641132831573486,
      "learning_rate": 9.485582265512115e-06,
      "loss": 0.0039,
      "step": 42370
    },
    {
      "epoch": 0.6934467806594126,
      "grad_norm": 0.2458818256855011,
      "learning_rate": 9.485161622661672e-06,
      "loss": 0.0033,
      "step": 42380
    },
    {
      "epoch": 0.6936104066104885,
      "grad_norm": 0.07750624418258667,
      "learning_rate": 9.4847408172349e-06,
      "loss": 0.003,
      "step": 42390
    },
    {
      "epoch": 0.6937740325615642,
      "grad_norm": 0.22216324508190155,
      "learning_rate": 9.484319849247053e-06,
      "loss": 0.0043,
      "step": 42400
    },
    {
      "epoch": 0.6939376585126401,
      "grad_norm": 0.050752464681863785,
      "learning_rate": 9.483898718713389e-06,
      "loss": 0.0024,
      "step": 42410
    },
    {
      "epoch": 0.694101284463716,
      "grad_norm": 0.31082597374916077,
      "learning_rate": 9.48347742564917e-06,
      "loss": 0.0026,
      "step": 42420
    },
    {
      "epoch": 0.6942649104147918,
      "grad_norm": 0.192427858710289,
      "learning_rate": 9.483055970069674e-06,
      "loss": 0.0031,
      "step": 42430
    },
    {
      "epoch": 0.6944285363658677,
      "grad_norm": 0.25390905141830444,
      "learning_rate": 9.482634351990172e-06,
      "loss": 0.0025,
      "step": 42440
    },
    {
      "epoch": 0.6945921623169434,
      "grad_norm": 0.1646750420331955,
      "learning_rate": 9.482212571425951e-06,
      "loss": 0.0033,
      "step": 42450
    },
    {
      "epoch": 0.6947557882680193,
      "grad_norm": 0.07395486533641815,
      "learning_rate": 9.481790628392294e-06,
      "loss": 0.0026,
      "step": 42460
    },
    {
      "epoch": 0.6949194142190952,
      "grad_norm": 0.12713567912578583,
      "learning_rate": 9.481368522904499e-06,
      "loss": 0.0054,
      "step": 42470
    },
    {
      "epoch": 0.695083040170171,
      "grad_norm": 0.18395251035690308,
      "learning_rate": 9.480946254977866e-06,
      "loss": 0.0055,
      "step": 42480
    },
    {
      "epoch": 0.6952466661212469,
      "grad_norm": 0.0654275119304657,
      "learning_rate": 9.4805238246277e-06,
      "loss": 0.0052,
      "step": 42490
    },
    {
      "epoch": 0.6954102920723226,
      "grad_norm": 0.218764066696167,
      "learning_rate": 9.480101231869317e-06,
      "loss": 0.0023,
      "step": 42500
    },
    {
      "epoch": 0.6955739180233985,
      "grad_norm": 0.12461761385202408,
      "learning_rate": 9.47967847671803e-06,
      "loss": 0.0038,
      "step": 42510
    },
    {
      "epoch": 0.6957375439744744,
      "grad_norm": 0.31489256024360657,
      "learning_rate": 9.479255559189164e-06,
      "loss": 0.003,
      "step": 42520
    },
    {
      "epoch": 0.6959011699255502,
      "grad_norm": 0.3809584081172943,
      "learning_rate": 9.478832479298051e-06,
      "loss": 0.0022,
      "step": 42530
    },
    {
      "epoch": 0.6960647958766261,
      "grad_norm": 0.19557444751262665,
      "learning_rate": 9.478409237060024e-06,
      "loss": 0.0034,
      "step": 42540
    },
    {
      "epoch": 0.6962284218277018,
      "grad_norm": 0.285917729139328,
      "learning_rate": 9.477985832490427e-06,
      "loss": 0.0059,
      "step": 42550
    },
    {
      "epoch": 0.6963920477787777,
      "grad_norm": 0.10538316518068314,
      "learning_rate": 9.477562265604606e-06,
      "loss": 0.003,
      "step": 42560
    },
    {
      "epoch": 0.6965556737298536,
      "grad_norm": 0.08849726617336273,
      "learning_rate": 9.477138536417914e-06,
      "loss": 0.0033,
      "step": 42570
    },
    {
      "epoch": 0.6967192996809294,
      "grad_norm": 0.13097114861011505,
      "learning_rate": 9.476714644945711e-06,
      "loss": 0.0053,
      "step": 42580
    },
    {
      "epoch": 0.6968829256320053,
      "grad_norm": 0.17211544513702393,
      "learning_rate": 9.476290591203361e-06,
      "loss": 0.0035,
      "step": 42590
    },
    {
      "epoch": 0.697046551583081,
      "grad_norm": 0.3398285210132599,
      "learning_rate": 9.475866375206235e-06,
      "loss": 0.0022,
      "step": 42600
    },
    {
      "epoch": 0.6972101775341569,
      "grad_norm": 0.07803649455308914,
      "learning_rate": 9.475441996969712e-06,
      "loss": 0.0041,
      "step": 42610
    },
    {
      "epoch": 0.6973738034852328,
      "grad_norm": 0.12283594161272049,
      "learning_rate": 9.475017456509172e-06,
      "loss": 0.0022,
      "step": 42620
    },
    {
      "epoch": 0.6975374294363086,
      "grad_norm": 0.3924367427825928,
      "learning_rate": 9.474592753840004e-06,
      "loss": 0.0034,
      "step": 42630
    },
    {
      "epoch": 0.6977010553873845,
      "grad_norm": 0.09592388570308685,
      "learning_rate": 9.474167888977605e-06,
      "loss": 0.0027,
      "step": 42640
    },
    {
      "epoch": 0.6978646813384602,
      "grad_norm": 0.09245303273200989,
      "learning_rate": 9.473742861937372e-06,
      "loss": 0.0047,
      "step": 42650
    },
    {
      "epoch": 0.6980283072895361,
      "grad_norm": 0.3495856821537018,
      "learning_rate": 9.473317672734711e-06,
      "loss": 0.0055,
      "step": 42660
    },
    {
      "epoch": 0.6981919332406119,
      "grad_norm": 0.29541856050491333,
      "learning_rate": 9.472892321385039e-06,
      "loss": 0.0025,
      "step": 42670
    },
    {
      "epoch": 0.6983555591916878,
      "grad_norm": 0.18021422624588013,
      "learning_rate": 9.472466807903769e-06,
      "loss": 0.0029,
      "step": 42680
    },
    {
      "epoch": 0.6985191851427637,
      "grad_norm": 0.14795605838298798,
      "learning_rate": 9.472041132306327e-06,
      "loss": 0.0032,
      "step": 42690
    },
    {
      "epoch": 0.6986828110938395,
      "grad_norm": 0.26293906569480896,
      "learning_rate": 9.471615294608142e-06,
      "loss": 0.0024,
      "step": 42700
    },
    {
      "epoch": 0.6988464370449153,
      "grad_norm": 0.34060484170913696,
      "learning_rate": 9.47118929482465e-06,
      "loss": 0.0044,
      "step": 42710
    },
    {
      "epoch": 0.6990100629959911,
      "grad_norm": 0.05026577040553093,
      "learning_rate": 9.470763132971292e-06,
      "loss": 0.0022,
      "step": 42720
    },
    {
      "epoch": 0.699173688947067,
      "grad_norm": 0.21517863869667053,
      "learning_rate": 9.470336809063516e-06,
      "loss": 0.0038,
      "step": 42730
    },
    {
      "epoch": 0.6993373148981429,
      "grad_norm": 0.024979673326015472,
      "learning_rate": 9.469910323116774e-06,
      "loss": 0.0037,
      "step": 42740
    },
    {
      "epoch": 0.6995009408492187,
      "grad_norm": 0.24875278770923615,
      "learning_rate": 9.469483675146526e-06,
      "loss": 0.0027,
      "step": 42750
    },
    {
      "epoch": 0.6996645668002945,
      "grad_norm": 0.0886520966887474,
      "learning_rate": 9.469056865168238e-06,
      "loss": 0.0024,
      "step": 42760
    },
    {
      "epoch": 0.6998281927513703,
      "grad_norm": 0.2391182780265808,
      "learning_rate": 9.468629893197378e-06,
      "loss": 0.004,
      "step": 42770
    },
    {
      "epoch": 0.6999918187024462,
      "grad_norm": 0.2099774181842804,
      "learning_rate": 9.468202759249428e-06,
      "loss": 0.0034,
      "step": 42780
    },
    {
      "epoch": 0.7001554446535221,
      "grad_norm": 0.1299414485692978,
      "learning_rate": 9.467775463339862e-06,
      "loss": 0.0017,
      "step": 42790
    },
    {
      "epoch": 0.7003190706045979,
      "grad_norm": 0.1556621491909027,
      "learning_rate": 9.467348005484179e-06,
      "loss": 0.0034,
      "step": 42800
    },
    {
      "epoch": 0.7004826965556737,
      "grad_norm": 0.3767586648464203,
      "learning_rate": 9.466920385697862e-06,
      "loss": 0.0035,
      "step": 42810
    },
    {
      "epoch": 0.7006463225067495,
      "grad_norm": 0.11935310810804367,
      "learning_rate": 9.466492603996421e-06,
      "loss": 0.0037,
      "step": 42820
    },
    {
      "epoch": 0.7008099484578254,
      "grad_norm": 0.13375429809093475,
      "learning_rate": 9.466064660395357e-06,
      "loss": 0.0028,
      "step": 42830
    },
    {
      "epoch": 0.7009735744089013,
      "grad_norm": 0.12701818346977234,
      "learning_rate": 9.465636554910183e-06,
      "loss": 0.003,
      "step": 42840
    },
    {
      "epoch": 0.7011372003599771,
      "grad_norm": 0.08951638638973236,
      "learning_rate": 9.465208287556417e-06,
      "loss": 0.0037,
      "step": 42850
    },
    {
      "epoch": 0.701300826311053,
      "grad_norm": 0.3227202892303467,
      "learning_rate": 9.464779858349582e-06,
      "loss": 0.0037,
      "step": 42860
    },
    {
      "epoch": 0.7014644522621287,
      "grad_norm": 0.04804162681102753,
      "learning_rate": 9.464351267305207e-06,
      "loss": 0.004,
      "step": 42870
    },
    {
      "epoch": 0.7016280782132046,
      "grad_norm": 0.23942385613918304,
      "learning_rate": 9.46392251443883e-06,
      "loss": 0.0045,
      "step": 42880
    },
    {
      "epoch": 0.7017917041642805,
      "grad_norm": 0.22162482142448425,
      "learning_rate": 9.463493599765989e-06,
      "loss": 0.0031,
      "step": 42890
    },
    {
      "epoch": 0.7019553301153563,
      "grad_norm": 0.17343223094940186,
      "learning_rate": 9.463064523302234e-06,
      "loss": 0.0045,
      "step": 42900
    },
    {
      "epoch": 0.7021189560664322,
      "grad_norm": 0.12951546907424927,
      "learning_rate": 9.462635285063116e-06,
      "loss": 0.0026,
      "step": 42910
    },
    {
      "epoch": 0.7022825820175079,
      "grad_norm": 0.3482697904109955,
      "learning_rate": 9.462205885064196e-06,
      "loss": 0.0039,
      "step": 42920
    },
    {
      "epoch": 0.7024462079685838,
      "grad_norm": 0.1096210703253746,
      "learning_rate": 9.461776323321035e-06,
      "loss": 0.002,
      "step": 42930
    },
    {
      "epoch": 0.7026098339196597,
      "grad_norm": 0.34508779644966125,
      "learning_rate": 9.461346599849209e-06,
      "loss": 0.0046,
      "step": 42940
    },
    {
      "epoch": 0.7027734598707355,
      "grad_norm": 0.032945528626441956,
      "learning_rate": 9.460916714664288e-06,
      "loss": 0.0041,
      "step": 42950
    },
    {
      "epoch": 0.7029370858218114,
      "grad_norm": 0.15586833655834198,
      "learning_rate": 9.460486667781858e-06,
      "loss": 0.0034,
      "step": 42960
    },
    {
      "epoch": 0.7031007117728871,
      "grad_norm": 0.25630250573158264,
      "learning_rate": 9.460056459217509e-06,
      "loss": 0.0025,
      "step": 42970
    },
    {
      "epoch": 0.703264337723963,
      "grad_norm": 0.18831761181354523,
      "learning_rate": 9.459626088986831e-06,
      "loss": 0.0045,
      "step": 42980
    },
    {
      "epoch": 0.7034279636750389,
      "grad_norm": 0.5098690986633301,
      "learning_rate": 9.459195557105427e-06,
      "loss": 0.0062,
      "step": 42990
    },
    {
      "epoch": 0.7035915896261147,
      "grad_norm": 0.19844293594360352,
      "learning_rate": 9.4587648635889e-06,
      "loss": 0.0016,
      "step": 43000
    },
    {
      "epoch": 0.7037552155771906,
      "grad_norm": 0.19846397638320923,
      "learning_rate": 9.458334008452866e-06,
      "loss": 0.0042,
      "step": 43010
    },
    {
      "epoch": 0.7039188415282663,
      "grad_norm": 0.18688060343265533,
      "learning_rate": 9.457902991712936e-06,
      "loss": 0.0022,
      "step": 43020
    },
    {
      "epoch": 0.7040824674793422,
      "grad_norm": 0.14431153237819672,
      "learning_rate": 9.457471813384737e-06,
      "loss": 0.0023,
      "step": 43030
    },
    {
      "epoch": 0.7042460934304181,
      "grad_norm": 0.20458489656448364,
      "learning_rate": 9.4570404734839e-06,
      "loss": 0.0064,
      "step": 43040
    },
    {
      "epoch": 0.7044097193814939,
      "grad_norm": 0.205845445394516,
      "learning_rate": 9.456608972026057e-06,
      "loss": 0.0028,
      "step": 43050
    },
    {
      "epoch": 0.7045733453325698,
      "grad_norm": 0.0740434005856514,
      "learning_rate": 9.45617730902685e-06,
      "loss": 0.0063,
      "step": 43060
    },
    {
      "epoch": 0.7047369712836455,
      "grad_norm": 0.07615919411182404,
      "learning_rate": 9.455745484501927e-06,
      "loss": 0.0034,
      "step": 43070
    },
    {
      "epoch": 0.7049005972347214,
      "grad_norm": 0.16960659623146057,
      "learning_rate": 9.455313498466937e-06,
      "loss": 0.0029,
      "step": 43080
    },
    {
      "epoch": 0.7050642231857973,
      "grad_norm": 0.13934388756752014,
      "learning_rate": 9.454881350937544e-06,
      "loss": 0.0027,
      "step": 43090
    },
    {
      "epoch": 0.7052278491368731,
      "grad_norm": 0.12008245289325714,
      "learning_rate": 9.454449041929406e-06,
      "loss": 0.0033,
      "step": 43100
    },
    {
      "epoch": 0.705391475087949,
      "grad_norm": 0.1778629869222641,
      "learning_rate": 9.454016571458198e-06,
      "loss": 0.0037,
      "step": 43110
    },
    {
      "epoch": 0.7055551010390247,
      "grad_norm": 0.1880698800086975,
      "learning_rate": 9.453583939539593e-06,
      "loss": 0.0031,
      "step": 43120
    },
    {
      "epoch": 0.7057187269901006,
      "grad_norm": 0.026169197633862495,
      "learning_rate": 9.453151146189275e-06,
      "loss": 0.0036,
      "step": 43130
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.41859370470046997,
      "learning_rate": 9.45271819142293e-06,
      "loss": 0.002,
      "step": 43140
    },
    {
      "epoch": 0.7060459788922523,
      "grad_norm": 0.12803144752979279,
      "learning_rate": 9.452285075256251e-06,
      "loss": 0.0039,
      "step": 43150
    },
    {
      "epoch": 0.7062096048433282,
      "grad_norm": 0.22554750740528107,
      "learning_rate": 9.451851797704941e-06,
      "loss": 0.0032,
      "step": 43160
    },
    {
      "epoch": 0.706373230794404,
      "grad_norm": 0.20387428998947144,
      "learning_rate": 9.451418358784703e-06,
      "loss": 0.0035,
      "step": 43170
    },
    {
      "epoch": 0.7065368567454798,
      "grad_norm": 0.10441246628761292,
      "learning_rate": 9.450984758511248e-06,
      "loss": 0.0048,
      "step": 43180
    },
    {
      "epoch": 0.7067004826965557,
      "grad_norm": 0.06095161288976669,
      "learning_rate": 9.450550996900295e-06,
      "loss": 0.0036,
      "step": 43190
    },
    {
      "epoch": 0.7068641086476315,
      "grad_norm": 0.4035806357860565,
      "learning_rate": 9.450117073967561e-06,
      "loss": 0.0059,
      "step": 43200
    },
    {
      "epoch": 0.7070277345987074,
      "grad_norm": 0.6646978259086609,
      "learning_rate": 9.44968298972878e-06,
      "loss": 0.0056,
      "step": 43210
    },
    {
      "epoch": 0.7071913605497832,
      "grad_norm": 0.14821387827396393,
      "learning_rate": 9.449248744199687e-06,
      "loss": 0.0056,
      "step": 43220
    },
    {
      "epoch": 0.707354986500859,
      "grad_norm": 0.2699081301689148,
      "learning_rate": 9.44881433739602e-06,
      "loss": 0.009,
      "step": 43230
    },
    {
      "epoch": 0.7075186124519349,
      "grad_norm": 0.11489080637693405,
      "learning_rate": 9.448379769333524e-06,
      "loss": 0.0032,
      "step": 43240
    },
    {
      "epoch": 0.7076822384030107,
      "grad_norm": 0.24692316353321075,
      "learning_rate": 9.447945040027954e-06,
      "loss": 0.0027,
      "step": 43250
    },
    {
      "epoch": 0.7078458643540866,
      "grad_norm": 0.0392230786383152,
      "learning_rate": 9.447510149495065e-06,
      "loss": 0.0039,
      "step": 43260
    },
    {
      "epoch": 0.7080094903051624,
      "grad_norm": 0.09486676007509232,
      "learning_rate": 9.447075097750626e-06,
      "loss": 0.0032,
      "step": 43270
    },
    {
      "epoch": 0.7081731162562382,
      "grad_norm": 0.11020425707101822,
      "learning_rate": 9.4466398848104e-06,
      "loss": 0.0025,
      "step": 43280
    },
    {
      "epoch": 0.7083367422073141,
      "grad_norm": 0.10121571272611618,
      "learning_rate": 9.446204510690167e-06,
      "loss": 0.0027,
      "step": 43290
    },
    {
      "epoch": 0.7085003681583899,
      "grad_norm": 0.185047447681427,
      "learning_rate": 9.445768975405704e-06,
      "loss": 0.0028,
      "step": 43300
    },
    {
      "epoch": 0.7086639941094658,
      "grad_norm": 0.4259854555130005,
      "learning_rate": 9.445333278972804e-06,
      "loss": 0.0049,
      "step": 43310
    },
    {
      "epoch": 0.7088276200605416,
      "grad_norm": 0.34021973609924316,
      "learning_rate": 9.444897421407255e-06,
      "loss": 0.003,
      "step": 43320
    },
    {
      "epoch": 0.7089912460116174,
      "grad_norm": 0.2663522958755493,
      "learning_rate": 9.444461402724858e-06,
      "loss": 0.004,
      "step": 43330
    },
    {
      "epoch": 0.7091548719626933,
      "grad_norm": 0.25476956367492676,
      "learning_rate": 9.444025222941416e-06,
      "loss": 0.007,
      "step": 43340
    },
    {
      "epoch": 0.7093184979137691,
      "grad_norm": 0.34113025665283203,
      "learning_rate": 9.443588882072743e-06,
      "loss": 0.0073,
      "step": 43350
    },
    {
      "epoch": 0.709482123864845,
      "grad_norm": 0.0735788568854332,
      "learning_rate": 9.44315238013465e-06,
      "loss": 0.0028,
      "step": 43360
    },
    {
      "epoch": 0.7096457498159208,
      "grad_norm": 0.027225032448768616,
      "learning_rate": 9.442715717142963e-06,
      "loss": 0.0026,
      "step": 43370
    },
    {
      "epoch": 0.7098093757669967,
      "grad_norm": 0.1284790188074112,
      "learning_rate": 9.44227889311351e-06,
      "loss": 0.0031,
      "step": 43380
    },
    {
      "epoch": 0.7099730017180725,
      "grad_norm": 0.30750277638435364,
      "learning_rate": 9.441841908062124e-06,
      "loss": 0.0035,
      "step": 43390
    },
    {
      "epoch": 0.7101366276691483,
      "grad_norm": 0.4352801442146301,
      "learning_rate": 9.441404762004644e-06,
      "loss": 0.0056,
      "step": 43400
    },
    {
      "epoch": 0.7103002536202242,
      "grad_norm": 0.8874611258506775,
      "learning_rate": 9.440967454956915e-06,
      "loss": 0.0046,
      "step": 43410
    },
    {
      "epoch": 0.7104638795713,
      "grad_norm": 0.1412227600812912,
      "learning_rate": 9.44052998693479e-06,
      "loss": 0.0034,
      "step": 43420
    },
    {
      "epoch": 0.7106275055223759,
      "grad_norm": 0.19500382244586945,
      "learning_rate": 9.440092357954127e-06,
      "loss": 0.0051,
      "step": 43430
    },
    {
      "epoch": 0.7107911314734517,
      "grad_norm": 0.337568074464798,
      "learning_rate": 9.439654568030785e-06,
      "loss": 0.0022,
      "step": 43440
    },
    {
      "epoch": 0.7109547574245275,
      "grad_norm": 0.03067833185195923,
      "learning_rate": 9.439216617180638e-06,
      "loss": 0.0027,
      "step": 43450
    },
    {
      "epoch": 0.7111183833756034,
      "grad_norm": 0.27238795161247253,
      "learning_rate": 9.438778505419557e-06,
      "loss": 0.0031,
      "step": 43460
    },
    {
      "epoch": 0.7112820093266792,
      "grad_norm": 0.2768290936946869,
      "learning_rate": 9.438340232763422e-06,
      "loss": 0.0037,
      "step": 43470
    },
    {
      "epoch": 0.7114456352777551,
      "grad_norm": 0.29424840211868286,
      "learning_rate": 9.437901799228121e-06,
      "loss": 0.0064,
      "step": 43480
    },
    {
      "epoch": 0.711609261228831,
      "grad_norm": 0.10928195714950562,
      "learning_rate": 9.437463204829549e-06,
      "loss": 0.0038,
      "step": 43490
    },
    {
      "epoch": 0.7117728871799067,
      "grad_norm": 0.2692672312259674,
      "learning_rate": 9.437024449583598e-06,
      "loss": 0.0025,
      "step": 43500
    },
    {
      "epoch": 0.7119365131309826,
      "grad_norm": 0.29281020164489746,
      "learning_rate": 9.436585533506176e-06,
      "loss": 0.0028,
      "step": 43510
    },
    {
      "epoch": 0.7121001390820584,
      "grad_norm": 0.4347575902938843,
      "learning_rate": 9.436146456613192e-06,
      "loss": 0.005,
      "step": 43520
    },
    {
      "epoch": 0.7122637650331343,
      "grad_norm": 0.1944822520017624,
      "learning_rate": 9.43570721892056e-06,
      "loss": 0.0038,
      "step": 43530
    },
    {
      "epoch": 0.71242739098421,
      "grad_norm": 0.38200923800468445,
      "learning_rate": 9.435267820444203e-06,
      "loss": 0.0042,
      "step": 43540
    },
    {
      "epoch": 0.7125910169352859,
      "grad_norm": 0.30823472142219543,
      "learning_rate": 9.434828261200046e-06,
      "loss": 0.0029,
      "step": 43550
    },
    {
      "epoch": 0.7127546428863618,
      "grad_norm": 0.16902115941047668,
      "learning_rate": 9.434388541204025e-06,
      "loss": 0.0024,
      "step": 43560
    },
    {
      "epoch": 0.7129182688374376,
      "grad_norm": 0.07349956780672073,
      "learning_rate": 9.433948660472075e-06,
      "loss": 0.0035,
      "step": 43570
    },
    {
      "epoch": 0.7130818947885135,
      "grad_norm": 0.07912345975637436,
      "learning_rate": 9.433508619020144e-06,
      "loss": 0.0029,
      "step": 43580
    },
    {
      "epoch": 0.7132455207395892,
      "grad_norm": 0.19981731474399567,
      "learning_rate": 9.433068416864182e-06,
      "loss": 0.0041,
      "step": 43590
    },
    {
      "epoch": 0.7134091466906651,
      "grad_norm": 0.2753267288208008,
      "learning_rate": 9.432628054020143e-06,
      "loss": 0.0038,
      "step": 43600
    },
    {
      "epoch": 0.713572772641741,
      "grad_norm": 0.11021915078163147,
      "learning_rate": 9.432187530503993e-06,
      "loss": 0.0059,
      "step": 43610
    },
    {
      "epoch": 0.7137363985928168,
      "grad_norm": 0.22448433935642242,
      "learning_rate": 9.431746846331694e-06,
      "loss": 0.0038,
      "step": 43620
    },
    {
      "epoch": 0.7139000245438927,
      "grad_norm": 0.11207084357738495,
      "learning_rate": 9.431306001519227e-06,
      "loss": 0.0049,
      "step": 43630
    },
    {
      "epoch": 0.7140636504949684,
      "grad_norm": 0.034532126039266586,
      "learning_rate": 9.430864996082565e-06,
      "loss": 0.0039,
      "step": 43640
    },
    {
      "epoch": 0.7142272764460443,
      "grad_norm": 0.20960968732833862,
      "learning_rate": 9.430423830037699e-06,
      "loss": 0.0047,
      "step": 43650
    },
    {
      "epoch": 0.7143909023971202,
      "grad_norm": 0.47492852807044983,
      "learning_rate": 9.429982503400614e-06,
      "loss": 0.0033,
      "step": 43660
    },
    {
      "epoch": 0.714554528348196,
      "grad_norm": 0.31388694047927856,
      "learning_rate": 9.429541016187312e-06,
      "loss": 0.0029,
      "step": 43670
    },
    {
      "epoch": 0.7147181542992719,
      "grad_norm": 0.046706922352313995,
      "learning_rate": 9.429099368413794e-06,
      "loss": 0.0034,
      "step": 43680
    },
    {
      "epoch": 0.7148817802503477,
      "grad_norm": 0.10895107686519623,
      "learning_rate": 9.42865756009607e-06,
      "loss": 0.0028,
      "step": 43690
    },
    {
      "epoch": 0.7150454062014235,
      "grad_norm": 0.21794630587100983,
      "learning_rate": 9.428215591250151e-06,
      "loss": 0.0047,
      "step": 43700
    },
    {
      "epoch": 0.7152090321524994,
      "grad_norm": 0.13827158510684967,
      "learning_rate": 9.427773461892063e-06,
      "loss": 0.0024,
      "step": 43710
    },
    {
      "epoch": 0.7153726581035752,
      "grad_norm": 0.0689816102385521,
      "learning_rate": 9.427331172037826e-06,
      "loss": 0.0029,
      "step": 43720
    },
    {
      "epoch": 0.7155362840546511,
      "grad_norm": 0.20435881614685059,
      "learning_rate": 9.426888721703477e-06,
      "loss": 0.0034,
      "step": 43730
    },
    {
      "epoch": 0.7156999100057269,
      "grad_norm": 0.2274610549211502,
      "learning_rate": 9.42644611090505e-06,
      "loss": 0.0047,
      "step": 43740
    },
    {
      "epoch": 0.7158635359568027,
      "grad_norm": 0.3494691848754883,
      "learning_rate": 9.426003339658591e-06,
      "loss": 0.0027,
      "step": 43750
    },
    {
      "epoch": 0.7160271619078786,
      "grad_norm": 0.3882181942462921,
      "learning_rate": 9.42556040798015e-06,
      "loss": 0.0038,
      "step": 43760
    },
    {
      "epoch": 0.7161907878589544,
      "grad_norm": 0.1835632622241974,
      "learning_rate": 9.42511731588578e-06,
      "loss": 0.0019,
      "step": 43770
    },
    {
      "epoch": 0.7163544138100303,
      "grad_norm": 0.25872087478637695,
      "learning_rate": 9.42467406339154e-06,
      "loss": 0.0034,
      "step": 43780
    },
    {
      "epoch": 0.7165180397611061,
      "grad_norm": 0.14610989391803741,
      "learning_rate": 9.424230650513501e-06,
      "loss": 0.0023,
      "step": 43790
    },
    {
      "epoch": 0.716681665712182,
      "grad_norm": 0.18413010239601135,
      "learning_rate": 9.423787077267737e-06,
      "loss": 0.0037,
      "step": 43800
    },
    {
      "epoch": 0.7168452916632578,
      "grad_norm": 0.10271312296390533,
      "learning_rate": 9.423343343670321e-06,
      "loss": 0.0045,
      "step": 43810
    },
    {
      "epoch": 0.7170089176143336,
      "grad_norm": 0.06794589757919312,
      "learning_rate": 9.422899449737339e-06,
      "loss": 0.0031,
      "step": 43820
    },
    {
      "epoch": 0.7171725435654095,
      "grad_norm": 0.29964444041252136,
      "learning_rate": 9.422455395484886e-06,
      "loss": 0.0035,
      "step": 43830
    },
    {
      "epoch": 0.7173361695164853,
      "grad_norm": 0.4265291094779968,
      "learning_rate": 9.42201118092905e-06,
      "loss": 0.0052,
      "step": 43840
    },
    {
      "epoch": 0.7174997954675612,
      "grad_norm": 0.13908329606056213,
      "learning_rate": 9.42156680608594e-06,
      "loss": 0.003,
      "step": 43850
    },
    {
      "epoch": 0.717663421418637,
      "grad_norm": 0.30660727620124817,
      "learning_rate": 9.42112227097166e-06,
      "loss": 0.0027,
      "step": 43860
    },
    {
      "epoch": 0.7178270473697128,
      "grad_norm": 0.12608669698238373,
      "learning_rate": 9.42067757560232e-06,
      "loss": 0.0024,
      "step": 43870
    },
    {
      "epoch": 0.7179906733207887,
      "grad_norm": 0.2620046138763428,
      "learning_rate": 9.420232719994044e-06,
      "loss": 0.0036,
      "step": 43880
    },
    {
      "epoch": 0.7181542992718645,
      "grad_norm": 0.06332611292600632,
      "learning_rate": 9.419787704162957e-06,
      "loss": 0.0046,
      "step": 43890
    },
    {
      "epoch": 0.7183179252229404,
      "grad_norm": 0.08811411261558533,
      "learning_rate": 9.419342528125188e-06,
      "loss": 0.0039,
      "step": 43900
    },
    {
      "epoch": 0.7184815511740162,
      "grad_norm": 0.004064450040459633,
      "learning_rate": 9.418897191896873e-06,
      "loss": 0.0031,
      "step": 43910
    },
    {
      "epoch": 0.718645177125092,
      "grad_norm": 0.208746075630188,
      "learning_rate": 9.418451695494158e-06,
      "loss": 0.0026,
      "step": 43920
    },
    {
      "epoch": 0.7188088030761679,
      "grad_norm": 0.3180221915245056,
      "learning_rate": 9.418006038933187e-06,
      "loss": 0.0052,
      "step": 43930
    },
    {
      "epoch": 0.7189724290272437,
      "grad_norm": 0.08794719725847244,
      "learning_rate": 9.417560222230115e-06,
      "loss": 0.0021,
      "step": 43940
    },
    {
      "epoch": 0.7191360549783196,
      "grad_norm": 0.18991270661354065,
      "learning_rate": 9.417114245401103e-06,
      "loss": 0.0033,
      "step": 43950
    },
    {
      "epoch": 0.7192996809293954,
      "grad_norm": 0.05480005964636803,
      "learning_rate": 9.416668108462316e-06,
      "loss": 0.0026,
      "step": 43960
    },
    {
      "epoch": 0.7194633068804712,
      "grad_norm": 0.1448337286710739,
      "learning_rate": 9.416221811429924e-06,
      "loss": 0.0036,
      "step": 43970
    },
    {
      "epoch": 0.7196269328315471,
      "grad_norm": 0.16126029193401337,
      "learning_rate": 9.415775354320106e-06,
      "loss": 0.0041,
      "step": 43980
    },
    {
      "epoch": 0.7197905587826229,
      "grad_norm": 0.2131180763244629,
      "learning_rate": 9.415328737149045e-06,
      "loss": 0.002,
      "step": 43990
    },
    {
      "epoch": 0.7199541847336988,
      "grad_norm": 0.5852167010307312,
      "learning_rate": 9.414881959932929e-06,
      "loss": 0.0092,
      "step": 44000
    },
    {
      "epoch": 0.7201178106847747,
      "grad_norm": 0.20315052568912506,
      "learning_rate": 9.414435022687954e-06,
      "loss": 0.0027,
      "step": 44010
    },
    {
      "epoch": 0.7202814366358504,
      "grad_norm": 0.06888915598392487,
      "learning_rate": 9.413987925430317e-06,
      "loss": 0.003,
      "step": 44020
    },
    {
      "epoch": 0.7204450625869263,
      "grad_norm": 0.022122442722320557,
      "learning_rate": 9.413540668176226e-06,
      "loss": 0.0112,
      "step": 44030
    },
    {
      "epoch": 0.7206086885380021,
      "grad_norm": 0.1346496194601059,
      "learning_rate": 9.413093250941895e-06,
      "loss": 0.0037,
      "step": 44040
    },
    {
      "epoch": 0.720772314489078,
      "grad_norm": 0.06178409978747368,
      "learning_rate": 9.412645673743542e-06,
      "loss": 0.0042,
      "step": 44050
    },
    {
      "epoch": 0.7209359404401539,
      "grad_norm": 0.07769902050495148,
      "learning_rate": 9.412197936597386e-06,
      "loss": 0.0021,
      "step": 44060
    },
    {
      "epoch": 0.7210995663912296,
      "grad_norm": 0.1280083805322647,
      "learning_rate": 9.41175003951966e-06,
      "loss": 0.0041,
      "step": 44070
    },
    {
      "epoch": 0.7212631923423055,
      "grad_norm": 0.15115448832511902,
      "learning_rate": 9.411301982526598e-06,
      "loss": 0.0036,
      "step": 44080
    },
    {
      "epoch": 0.7214268182933813,
      "grad_norm": 0.15542688965797424,
      "learning_rate": 9.410853765634443e-06,
      "loss": 0.0025,
      "step": 44090
    },
    {
      "epoch": 0.7215904442444572,
      "grad_norm": 0.1315353810787201,
      "learning_rate": 9.410405388859439e-06,
      "loss": 0.0038,
      "step": 44100
    },
    {
      "epoch": 0.7217540701955331,
      "grad_norm": 0.11461956799030304,
      "learning_rate": 9.40995685221784e-06,
      "loss": 0.0036,
      "step": 44110
    },
    {
      "epoch": 0.7219176961466088,
      "grad_norm": 0.23914045095443726,
      "learning_rate": 9.409508155725903e-06,
      "loss": 0.0038,
      "step": 44120
    },
    {
      "epoch": 0.7220813220976847,
      "grad_norm": 0.0565275214612484,
      "learning_rate": 9.409059299399895e-06,
      "loss": 0.0029,
      "step": 44130
    },
    {
      "epoch": 0.7222449480487605,
      "grad_norm": 0.02471841126680374,
      "learning_rate": 9.408610283256084e-06,
      "loss": 0.0024,
      "step": 44140
    },
    {
      "epoch": 0.7224085739998364,
      "grad_norm": 0.09245089441537857,
      "learning_rate": 9.408161107310746e-06,
      "loss": 0.0027,
      "step": 44150
    },
    {
      "epoch": 0.7225721999509123,
      "grad_norm": 0.3753833472728729,
      "learning_rate": 9.407711771580162e-06,
      "loss": 0.0047,
      "step": 44160
    },
    {
      "epoch": 0.722735825901988,
      "grad_norm": 0.09006118029356003,
      "learning_rate": 9.40726227608062e-06,
      "loss": 0.0034,
      "step": 44170
    },
    {
      "epoch": 0.7228994518530639,
      "grad_norm": 0.07483652234077454,
      "learning_rate": 9.406812620828413e-06,
      "loss": 0.0024,
      "step": 44180
    },
    {
      "epoch": 0.7230630778041397,
      "grad_norm": 0.2620481252670288,
      "learning_rate": 9.40636280583984e-06,
      "loss": 0.0036,
      "step": 44190
    },
    {
      "epoch": 0.7232267037552156,
      "grad_norm": 0.39716148376464844,
      "learning_rate": 9.405912831131207e-06,
      "loss": 0.0028,
      "step": 44200
    },
    {
      "epoch": 0.7233903297062915,
      "grad_norm": 0.1393423080444336,
      "learning_rate": 9.405462696718824e-06,
      "loss": 0.0028,
      "step": 44210
    },
    {
      "epoch": 0.7235539556573672,
      "grad_norm": 0.17728587985038757,
      "learning_rate": 9.405012402619004e-06,
      "loss": 0.002,
      "step": 44220
    },
    {
      "epoch": 0.7237175816084431,
      "grad_norm": 0.13728955388069153,
      "learning_rate": 9.404561948848072e-06,
      "loss": 0.0037,
      "step": 44230
    },
    {
      "epoch": 0.7238812075595189,
      "grad_norm": 0.15387779474258423,
      "learning_rate": 9.404111335422356e-06,
      "loss": 0.0033,
      "step": 44240
    },
    {
      "epoch": 0.7240448335105948,
      "grad_norm": 0.14162440598011017,
      "learning_rate": 9.40366056235819e-06,
      "loss": 0.0039,
      "step": 44250
    },
    {
      "epoch": 0.7242084594616707,
      "grad_norm": 0.1799604594707489,
      "learning_rate": 9.40320962967191e-06,
      "loss": 0.0028,
      "step": 44260
    },
    {
      "epoch": 0.7243720854127464,
      "grad_norm": 0.09081440418958664,
      "learning_rate": 9.402758537379866e-06,
      "loss": 0.003,
      "step": 44270
    },
    {
      "epoch": 0.7245357113638223,
      "grad_norm": 0.15042786300182343,
      "learning_rate": 9.402307285498405e-06,
      "loss": 0.0035,
      "step": 44280
    },
    {
      "epoch": 0.7246993373148981,
      "grad_norm": 0.20422321557998657,
      "learning_rate": 9.401855874043887e-06,
      "loss": 0.0032,
      "step": 44290
    },
    {
      "epoch": 0.724862963265974,
      "grad_norm": 0.09770471602678299,
      "learning_rate": 9.401404303032673e-06,
      "loss": 0.0025,
      "step": 44300
    },
    {
      "epoch": 0.7250265892170499,
      "grad_norm": 0.18016590178012848,
      "learning_rate": 9.40095257248113e-06,
      "loss": 0.0028,
      "step": 44310
    },
    {
      "epoch": 0.7251902151681257,
      "grad_norm": 0.19632486999034882,
      "learning_rate": 9.400500682405635e-06,
      "loss": 0.003,
      "step": 44320
    },
    {
      "epoch": 0.7253538411192015,
      "grad_norm": 0.08654646575450897,
      "learning_rate": 9.400048632822566e-06,
      "loss": 0.0033,
      "step": 44330
    },
    {
      "epoch": 0.7255174670702773,
      "grad_norm": 0.10829994082450867,
      "learning_rate": 9.399596423748311e-06,
      "loss": 0.002,
      "step": 44340
    },
    {
      "epoch": 0.7256810930213532,
      "grad_norm": 0.117234006524086,
      "learning_rate": 9.399144055199258e-06,
      "loss": 0.0025,
      "step": 44350
    },
    {
      "epoch": 0.7258447189724291,
      "grad_norm": 0.1839020848274231,
      "learning_rate": 9.398691527191808e-06,
      "loss": 0.0023,
      "step": 44360
    },
    {
      "epoch": 0.7260083449235049,
      "grad_norm": 0.2735215425491333,
      "learning_rate": 9.39823883974236e-06,
      "loss": 0.0034,
      "step": 44370
    },
    {
      "epoch": 0.7261719708745807,
      "grad_norm": 0.045950066298246384,
      "learning_rate": 9.397785992867325e-06,
      "loss": 0.0021,
      "step": 44380
    },
    {
      "epoch": 0.7263355968256565,
      "grad_norm": 0.2915136516094208,
      "learning_rate": 9.39733298658312e-06,
      "loss": 0.0035,
      "step": 44390
    },
    {
      "epoch": 0.7264992227767324,
      "grad_norm": 0.2397589087486267,
      "learning_rate": 9.396879820906159e-06,
      "loss": 0.0025,
      "step": 44400
    },
    {
      "epoch": 0.7266628487278082,
      "grad_norm": 0.120446115732193,
      "learning_rate": 9.396426495852875e-06,
      "loss": 0.0053,
      "step": 44410
    },
    {
      "epoch": 0.7268264746788841,
      "grad_norm": 0.13101591169834137,
      "learning_rate": 9.395973011439696e-06,
      "loss": 0.0026,
      "step": 44420
    },
    {
      "epoch": 0.72699010062996,
      "grad_norm": 0.34294649958610535,
      "learning_rate": 9.39551936768306e-06,
      "loss": 0.0043,
      "step": 44430
    },
    {
      "epoch": 0.7271537265810357,
      "grad_norm": 0.24978701770305634,
      "learning_rate": 9.395065564599413e-06,
      "loss": 0.0035,
      "step": 44440
    },
    {
      "epoch": 0.7273173525321116,
      "grad_norm": 0.2867276668548584,
      "learning_rate": 9.394611602205202e-06,
      "loss": 0.0036,
      "step": 44450
    },
    {
      "epoch": 0.7274809784831874,
      "grad_norm": 0.26214414834976196,
      "learning_rate": 9.394157480516883e-06,
      "loss": 0.0029,
      "step": 44460
    },
    {
      "epoch": 0.7276446044342633,
      "grad_norm": 0.2770514488220215,
      "learning_rate": 9.393703199550916e-06,
      "loss": 0.0072,
      "step": 44470
    },
    {
      "epoch": 0.7278082303853391,
      "grad_norm": 0.17170031368732452,
      "learning_rate": 9.393248759323768e-06,
      "loss": 0.002,
      "step": 44480
    },
    {
      "epoch": 0.7279718563364149,
      "grad_norm": 0.18607540428638458,
      "learning_rate": 9.392794159851912e-06,
      "loss": 0.0024,
      "step": 44490
    },
    {
      "epoch": 0.7281354822874908,
      "grad_norm": 0.1586611568927765,
      "learning_rate": 9.392339401151824e-06,
      "loss": 0.0034,
      "step": 44500
    },
    {
      "epoch": 0.7282991082385666,
      "grad_norm": 0.1039513424038887,
      "learning_rate": 9.391884483239991e-06,
      "loss": 0.0041,
      "step": 44510
    },
    {
      "epoch": 0.7284627341896425,
      "grad_norm": 0.2333914041519165,
      "learning_rate": 9.3914294061329e-06,
      "loss": 0.0043,
      "step": 44520
    },
    {
      "epoch": 0.7286263601407184,
      "grad_norm": 0.07784456759691238,
      "learning_rate": 9.390974169847048e-06,
      "loss": 0.0022,
      "step": 44530
    },
    {
      "epoch": 0.7287899860917941,
      "grad_norm": 0.3544568717479706,
      "learning_rate": 9.390518774398935e-06,
      "loss": 0.0062,
      "step": 44540
    },
    {
      "epoch": 0.72895361204287,
      "grad_norm": 0.24859455227851868,
      "learning_rate": 9.39006321980507e-06,
      "loss": 0.0032,
      "step": 44550
    },
    {
      "epoch": 0.7291172379939458,
      "grad_norm": 0.14105965197086334,
      "learning_rate": 9.389607506081964e-06,
      "loss": 0.0037,
      "step": 44560
    },
    {
      "epoch": 0.7292808639450217,
      "grad_norm": 0.14568167924880981,
      "learning_rate": 9.389151633246137e-06,
      "loss": 0.0045,
      "step": 44570
    },
    {
      "epoch": 0.7294444898960976,
      "grad_norm": 0.11969415843486786,
      "learning_rate": 9.38869560131411e-06,
      "loss": 0.0037,
      "step": 44580
    },
    {
      "epoch": 0.7296081158471733,
      "grad_norm": 0.2141263782978058,
      "learning_rate": 9.388239410302418e-06,
      "loss": 0.0048,
      "step": 44590
    },
    {
      "epoch": 0.7297717417982492,
      "grad_norm": 0.2572079002857208,
      "learning_rate": 9.387783060227592e-06,
      "loss": 0.0038,
      "step": 44600
    },
    {
      "epoch": 0.729935367749325,
      "grad_norm": 0.2962052822113037,
      "learning_rate": 9.387326551106177e-06,
      "loss": 0.0037,
      "step": 44610
    },
    {
      "epoch": 0.7300989937004009,
      "grad_norm": 0.23166991770267487,
      "learning_rate": 9.386869882954718e-06,
      "loss": 0.0041,
      "step": 44620
    },
    {
      "epoch": 0.7302626196514768,
      "grad_norm": 0.035091642290353775,
      "learning_rate": 9.386413055789772e-06,
      "loss": 0.0024,
      "step": 44630
    },
    {
      "epoch": 0.7304262456025525,
      "grad_norm": 0.1557222157716751,
      "learning_rate": 9.385956069627893e-06,
      "loss": 0.0031,
      "step": 44640
    },
    {
      "epoch": 0.7305898715536284,
      "grad_norm": 0.18170341849327087,
      "learning_rate": 9.385498924485648e-06,
      "loss": 0.0055,
      "step": 44650
    },
    {
      "epoch": 0.7307534975047042,
      "grad_norm": 0.28071677684783936,
      "learning_rate": 9.385041620379607e-06,
      "loss": 0.0024,
      "step": 44660
    },
    {
      "epoch": 0.7309171234557801,
      "grad_norm": 0.4183295965194702,
      "learning_rate": 9.384584157326347e-06,
      "loss": 0.0035,
      "step": 44670
    },
    {
      "epoch": 0.731080749406856,
      "grad_norm": 0.152989000082016,
      "learning_rate": 9.384126535342451e-06,
      "loss": 0.0027,
      "step": 44680
    },
    {
      "epoch": 0.7312443753579317,
      "grad_norm": 0.32978740334510803,
      "learning_rate": 9.383668754444503e-06,
      "loss": 0.002,
      "step": 44690
    },
    {
      "epoch": 0.7314080013090076,
      "grad_norm": 0.2996117174625397,
      "learning_rate": 9.383210814649098e-06,
      "loss": 0.0049,
      "step": 44700
    },
    {
      "epoch": 0.7315716272600834,
      "grad_norm": 0.24867761135101318,
      "learning_rate": 9.382752715972837e-06,
      "loss": 0.0028,
      "step": 44710
    },
    {
      "epoch": 0.7317352532111593,
      "grad_norm": 0.051696084439754486,
      "learning_rate": 9.382294458432324e-06,
      "loss": 0.0026,
      "step": 44720
    },
    {
      "epoch": 0.7318988791622352,
      "grad_norm": 0.34324270486831665,
      "learning_rate": 9.381836042044168e-06,
      "loss": 0.0039,
      "step": 44730
    },
    {
      "epoch": 0.732062505113311,
      "grad_norm": 0.16919779777526855,
      "learning_rate": 9.381377466824989e-06,
      "loss": 0.0025,
      "step": 44740
    },
    {
      "epoch": 0.7322261310643868,
      "grad_norm": 0.09304846078157425,
      "learning_rate": 9.380918732791406e-06,
      "loss": 0.0046,
      "step": 44750
    },
    {
      "epoch": 0.7323897570154626,
      "grad_norm": 0.44175365567207336,
      "learning_rate": 9.380459839960048e-06,
      "loss": 0.0032,
      "step": 44760
    },
    {
      "epoch": 0.7325533829665385,
      "grad_norm": 0.25202420353889465,
      "learning_rate": 9.380000788347549e-06,
      "loss": 0.0048,
      "step": 44770
    },
    {
      "epoch": 0.7327170089176144,
      "grad_norm": 0.09570915997028351,
      "learning_rate": 9.379541577970547e-06,
      "loss": 0.003,
      "step": 44780
    },
    {
      "epoch": 0.7328806348686902,
      "grad_norm": 0.15347380936145782,
      "learning_rate": 9.37908220884569e-06,
      "loss": 0.0027,
      "step": 44790
    },
    {
      "epoch": 0.733044260819766,
      "grad_norm": 0.18991000950336456,
      "learning_rate": 9.378622680989631e-06,
      "loss": 0.0025,
      "step": 44800
    },
    {
      "epoch": 0.7332078867708418,
      "grad_norm": 0.12974664568901062,
      "learning_rate": 9.37816299441902e-06,
      "loss": 0.002,
      "step": 44810
    },
    {
      "epoch": 0.7333715127219177,
      "grad_norm": 0.2423994094133377,
      "learning_rate": 9.377703149150523e-06,
      "loss": 0.0019,
      "step": 44820
    },
    {
      "epoch": 0.7335351386729936,
      "grad_norm": 0.2589271664619446,
      "learning_rate": 9.37724314520081e-06,
      "loss": 0.0043,
      "step": 44830
    },
    {
      "epoch": 0.7336987646240694,
      "grad_norm": 0.319818377494812,
      "learning_rate": 9.376782982586553e-06,
      "loss": 0.0034,
      "step": 44840
    },
    {
      "epoch": 0.7338623905751452,
      "grad_norm": 0.17180155217647552,
      "learning_rate": 9.37632266132443e-06,
      "loss": 0.0039,
      "step": 44850
    },
    {
      "epoch": 0.734026016526221,
      "grad_norm": 0.1244574636220932,
      "learning_rate": 9.375862181431132e-06,
      "loss": 0.0028,
      "step": 44860
    },
    {
      "epoch": 0.7341896424772969,
      "grad_norm": 0.09944602847099304,
      "learning_rate": 9.375401542923346e-06,
      "loss": 0.003,
      "step": 44870
    },
    {
      "epoch": 0.7343532684283728,
      "grad_norm": 0.36826252937316895,
      "learning_rate": 9.374940745817769e-06,
      "loss": 0.0058,
      "step": 44880
    },
    {
      "epoch": 0.7345168943794486,
      "grad_norm": 0.07497650384902954,
      "learning_rate": 9.374479790131106e-06,
      "loss": 0.0045,
      "step": 44890
    },
    {
      "epoch": 0.7346805203305244,
      "grad_norm": 0.12750157713890076,
      "learning_rate": 9.374018675880063e-06,
      "loss": 0.0028,
      "step": 44900
    },
    {
      "epoch": 0.7348441462816002,
      "grad_norm": 0.17026767134666443,
      "learning_rate": 9.373557403081357e-06,
      "loss": 0.003,
      "step": 44910
    },
    {
      "epoch": 0.7350077722326761,
      "grad_norm": 0.26842716336250305,
      "learning_rate": 9.373095971751706e-06,
      "loss": 0.0044,
      "step": 44920
    },
    {
      "epoch": 0.735171398183752,
      "grad_norm": 0.08554580062627792,
      "learning_rate": 9.372634381907837e-06,
      "loss": 0.0047,
      "step": 44930
    },
    {
      "epoch": 0.7353350241348278,
      "grad_norm": 0.284843385219574,
      "learning_rate": 9.37217263356648e-06,
      "loss": 0.0031,
      "step": 44940
    },
    {
      "epoch": 0.7354986500859036,
      "grad_norm": 0.04252878949046135,
      "learning_rate": 9.371710726744375e-06,
      "loss": 0.0025,
      "step": 44950
    },
    {
      "epoch": 0.7356622760369794,
      "grad_norm": 0.06299583613872528,
      "learning_rate": 9.371248661458263e-06,
      "loss": 0.0028,
      "step": 44960
    },
    {
      "epoch": 0.7358259019880553,
      "grad_norm": 0.1219446137547493,
      "learning_rate": 9.370786437724892e-06,
      "loss": 0.0041,
      "step": 44970
    },
    {
      "epoch": 0.7359895279391312,
      "grad_norm": 0.19243265688419342,
      "learning_rate": 9.370324055561017e-06,
      "loss": 0.0037,
      "step": 44980
    },
    {
      "epoch": 0.736153153890207,
      "grad_norm": 0.3377157747745514,
      "learning_rate": 9.3698615149834e-06,
      "loss": 0.0021,
      "step": 44990
    },
    {
      "epoch": 0.7363167798412829,
      "grad_norm": 0.10912178456783295,
      "learning_rate": 9.369398816008805e-06,
      "loss": 0.0029,
      "step": 45000
    },
    {
      "epoch": 0.7364804057923586,
      "grad_norm": 0.06433393806219101,
      "learning_rate": 9.368935958654006e-06,
      "loss": 0.0055,
      "step": 45010
    },
    {
      "epoch": 0.7366440317434345,
      "grad_norm": 0.20445063710212708,
      "learning_rate": 9.368472942935778e-06,
      "loss": 0.0039,
      "step": 45020
    },
    {
      "epoch": 0.7368076576945104,
      "grad_norm": 0.14766888320446014,
      "learning_rate": 9.368009768870906e-06,
      "loss": 0.0036,
      "step": 45030
    },
    {
      "epoch": 0.7369712836455862,
      "grad_norm": 0.07786289602518082,
      "learning_rate": 9.367546436476176e-06,
      "loss": 0.003,
      "step": 45040
    },
    {
      "epoch": 0.7371349095966621,
      "grad_norm": 0.058195970952510834,
      "learning_rate": 9.367082945768386e-06,
      "loss": 0.0035,
      "step": 45050
    },
    {
      "epoch": 0.7372985355477378,
      "grad_norm": 0.16606906056404114,
      "learning_rate": 9.366619296764335e-06,
      "loss": 0.0027,
      "step": 45060
    },
    {
      "epoch": 0.7374621614988137,
      "grad_norm": 0.4443644881248474,
      "learning_rate": 9.36615548948083e-06,
      "loss": 0.0033,
      "step": 45070
    },
    {
      "epoch": 0.7376257874498896,
      "grad_norm": 0.1841696947813034,
      "learning_rate": 9.365691523934682e-06,
      "loss": 0.005,
      "step": 45080
    },
    {
      "epoch": 0.7377894134009654,
      "grad_norm": 0.24207483232021332,
      "learning_rate": 9.36522740014271e-06,
      "loss": 0.0044,
      "step": 45090
    },
    {
      "epoch": 0.7379530393520413,
      "grad_norm": 0.13808202743530273,
      "learning_rate": 9.364763118121734e-06,
      "loss": 0.0041,
      "step": 45100
    },
    {
      "epoch": 0.738116665303117,
      "grad_norm": 0.23255422711372375,
      "learning_rate": 9.364298677888586e-06,
      "loss": 0.0028,
      "step": 45110
    },
    {
      "epoch": 0.7382802912541929,
      "grad_norm": 0.1403762549161911,
      "learning_rate": 9.363834079460101e-06,
      "loss": 0.004,
      "step": 45120
    },
    {
      "epoch": 0.7384439172052688,
      "grad_norm": 0.38627293705940247,
      "learning_rate": 9.36336932285312e-06,
      "loss": 0.0049,
      "step": 45130
    },
    {
      "epoch": 0.7386075431563446,
      "grad_norm": 0.31459856033325195,
      "learning_rate": 9.362904408084486e-06,
      "loss": 0.0029,
      "step": 45140
    },
    {
      "epoch": 0.7387711691074205,
      "grad_norm": 0.21671229600906372,
      "learning_rate": 9.362439335171053e-06,
      "loss": 0.0024,
      "step": 45150
    },
    {
      "epoch": 0.7389347950584962,
      "grad_norm": 0.1458991914987564,
      "learning_rate": 9.36197410412968e-06,
      "loss": 0.0017,
      "step": 45160
    },
    {
      "epoch": 0.7390984210095721,
      "grad_norm": 0.11365485191345215,
      "learning_rate": 9.36150871497723e-06,
      "loss": 0.0035,
      "step": 45170
    },
    {
      "epoch": 0.739262046960648,
      "grad_norm": 0.30745479464530945,
      "learning_rate": 9.36104316773057e-06,
      "loss": 0.0037,
      "step": 45180
    },
    {
      "epoch": 0.7394256729117238,
      "grad_norm": 0.30352458357810974,
      "learning_rate": 9.360577462406579e-06,
      "loss": 0.0031,
      "step": 45190
    },
    {
      "epoch": 0.7395892988627997,
      "grad_norm": 0.11075519770383835,
      "learning_rate": 9.360111599022133e-06,
      "loss": 0.0043,
      "step": 45200
    },
    {
      "epoch": 0.7397529248138754,
      "grad_norm": 0.22227925062179565,
      "learning_rate": 9.359645577594122e-06,
      "loss": 0.0067,
      "step": 45210
    },
    {
      "epoch": 0.7399165507649513,
      "grad_norm": 0.4556516110897064,
      "learning_rate": 9.359179398139437e-06,
      "loss": 0.002,
      "step": 45220
    },
    {
      "epoch": 0.7400801767160272,
      "grad_norm": 0.10171130299568176,
      "learning_rate": 9.358713060674975e-06,
      "loss": 0.0029,
      "step": 45230
    },
    {
      "epoch": 0.740243802667103,
      "grad_norm": 0.1968950778245926,
      "learning_rate": 9.35824656521764e-06,
      "loss": 0.0032,
      "step": 45240
    },
    {
      "epoch": 0.7404074286181789,
      "grad_norm": 0.28342896699905396,
      "learning_rate": 9.357779911784343e-06,
      "loss": 0.0032,
      "step": 45250
    },
    {
      "epoch": 0.7405710545692546,
      "grad_norm": 0.15065467357635498,
      "learning_rate": 9.357313100391998e-06,
      "loss": 0.0031,
      "step": 45260
    },
    {
      "epoch": 0.7407346805203305,
      "grad_norm": 0.37738969922065735,
      "learning_rate": 9.356846131057526e-06,
      "loss": 0.0039,
      "step": 45270
    },
    {
      "epoch": 0.7408983064714064,
      "grad_norm": 0.06322255730628967,
      "learning_rate": 9.356379003797853e-06,
      "loss": 0.0018,
      "step": 45280
    },
    {
      "epoch": 0.7410619324224822,
      "grad_norm": 0.14703792333602905,
      "learning_rate": 9.35591171862991e-06,
      "loss": 0.0039,
      "step": 45290
    },
    {
      "epoch": 0.7412255583735581,
      "grad_norm": 0.08885933458805084,
      "learning_rate": 9.355444275570637e-06,
      "loss": 0.0021,
      "step": 45300
    },
    {
      "epoch": 0.7413891843246339,
      "grad_norm": 0.32444217801094055,
      "learning_rate": 9.354976674636977e-06,
      "loss": 0.0063,
      "step": 45310
    },
    {
      "epoch": 0.7415528102757097,
      "grad_norm": 0.35946983098983765,
      "learning_rate": 9.35450891584588e-06,
      "loss": 0.0047,
      "step": 45320
    },
    {
      "epoch": 0.7417164362267855,
      "grad_norm": 0.419741690158844,
      "learning_rate": 9.3540409992143e-06,
      "loss": 0.0053,
      "step": 45330
    },
    {
      "epoch": 0.7418800621778614,
      "grad_norm": 0.20957323908805847,
      "learning_rate": 9.3535729247592e-06,
      "loss": 0.003,
      "step": 45340
    },
    {
      "epoch": 0.7420436881289373,
      "grad_norm": 0.02961203269660473,
      "learning_rate": 9.353104692497544e-06,
      "loss": 0.0031,
      "step": 45350
    },
    {
      "epoch": 0.7422073140800131,
      "grad_norm": 0.03943047672510147,
      "learning_rate": 9.352636302446305e-06,
      "loss": 0.0027,
      "step": 45360
    },
    {
      "epoch": 0.7423709400310889,
      "grad_norm": 0.12195006757974625,
      "learning_rate": 9.35216775462246e-06,
      "loss": 0.0052,
      "step": 45370
    },
    {
      "epoch": 0.7425345659821647,
      "grad_norm": 0.24874858558177948,
      "learning_rate": 9.351699049042995e-06,
      "loss": 0.0036,
      "step": 45380
    },
    {
      "epoch": 0.7426981919332406,
      "grad_norm": 0.1881808340549469,
      "learning_rate": 9.3512301857249e-06,
      "loss": 0.0032,
      "step": 45390
    },
    {
      "epoch": 0.7428618178843165,
      "grad_norm": 0.360744446516037,
      "learning_rate": 9.350761164685167e-06,
      "loss": 0.0044,
      "step": 45400
    },
    {
      "epoch": 0.7430254438353923,
      "grad_norm": 0.1845385730266571,
      "learning_rate": 9.350291985940799e-06,
      "loss": 0.0019,
      "step": 45410
    },
    {
      "epoch": 0.7431890697864681,
      "grad_norm": 0.0915493592619896,
      "learning_rate": 9.349822649508801e-06,
      "loss": 0.0023,
      "step": 45420
    },
    {
      "epoch": 0.7433526957375439,
      "grad_norm": 0.42872482538223267,
      "learning_rate": 9.349353155406189e-06,
      "loss": 0.0049,
      "step": 45430
    },
    {
      "epoch": 0.7435163216886198,
      "grad_norm": 0.3561018705368042,
      "learning_rate": 9.348883503649976e-06,
      "loss": 0.0036,
      "step": 45440
    },
    {
      "epoch": 0.7436799476396957,
      "grad_norm": 0.0710846334695816,
      "learning_rate": 9.34841369425719e-06,
      "loss": 0.0017,
      "step": 45450
    },
    {
      "epoch": 0.7438435735907715,
      "grad_norm": 0.15055815875530243,
      "learning_rate": 9.347943727244858e-06,
      "loss": 0.0023,
      "step": 45460
    },
    {
      "epoch": 0.7440071995418474,
      "grad_norm": 0.27672040462493896,
      "learning_rate": 9.347473602630018e-06,
      "loss": 0.0016,
      "step": 45470
    },
    {
      "epoch": 0.7441708254929231,
      "grad_norm": 0.259103000164032,
      "learning_rate": 9.347003320429706e-06,
      "loss": 0.0035,
      "step": 45480
    },
    {
      "epoch": 0.744334451443999,
      "grad_norm": 0.2535003125667572,
      "learning_rate": 9.346532880660972e-06,
      "loss": 0.0034,
      "step": 45490
    },
    {
      "epoch": 0.7444980773950749,
      "grad_norm": 0.3494164049625397,
      "learning_rate": 9.346062283340869e-06,
      "loss": 0.0043,
      "step": 45500
    },
    {
      "epoch": 0.7446617033461507,
      "grad_norm": 0.03437132015824318,
      "learning_rate": 9.345591528486452e-06,
      "loss": 0.0025,
      "step": 45510
    },
    {
      "epoch": 0.7448253292972266,
      "grad_norm": 0.15139859914779663,
      "learning_rate": 9.345120616114787e-06,
      "loss": 0.0024,
      "step": 45520
    },
    {
      "epoch": 0.7449889552483023,
      "grad_norm": 0.19978711009025574,
      "learning_rate": 9.344649546242943e-06,
      "loss": 0.0021,
      "step": 45530
    },
    {
      "epoch": 0.7451525811993782,
      "grad_norm": 0.0410379134118557,
      "learning_rate": 9.344178318887995e-06,
      "loss": 0.0015,
      "step": 45540
    },
    {
      "epoch": 0.7453162071504541,
      "grad_norm": 0.051625676453113556,
      "learning_rate": 9.343706934067026e-06,
      "loss": 0.0028,
      "step": 45550
    },
    {
      "epoch": 0.7454798331015299,
      "grad_norm": 0.16650187969207764,
      "learning_rate": 9.343235391797117e-06,
      "loss": 0.0027,
      "step": 45560
    },
    {
      "epoch": 0.7456434590526058,
      "grad_norm": 0.08861218392848969,
      "learning_rate": 9.342763692095366e-06,
      "loss": 0.0043,
      "step": 45570
    },
    {
      "epoch": 0.7458070850036815,
      "grad_norm": 0.13995987176895142,
      "learning_rate": 9.342291834978868e-06,
      "loss": 0.0032,
      "step": 45580
    },
    {
      "epoch": 0.7459707109547574,
      "grad_norm": 0.01826346106827259,
      "learning_rate": 9.341819820464727e-06,
      "loss": 0.0056,
      "step": 45590
    },
    {
      "epoch": 0.7461343369058333,
      "grad_norm": 0.2519683241844177,
      "learning_rate": 9.341347648570053e-06,
      "loss": 0.002,
      "step": 45600
    },
    {
      "epoch": 0.7462979628569091,
      "grad_norm": 0.07952021807432175,
      "learning_rate": 9.340875319311962e-06,
      "loss": 0.0036,
      "step": 45610
    },
    {
      "epoch": 0.746461588807985,
      "grad_norm": 0.3329220414161682,
      "learning_rate": 9.340402832707572e-06,
      "loss": 0.0034,
      "step": 45620
    },
    {
      "epoch": 0.7466252147590607,
      "grad_norm": 0.25139468908309937,
      "learning_rate": 9.339930188774012e-06,
      "loss": 0.0025,
      "step": 45630
    },
    {
      "epoch": 0.7467888407101366,
      "grad_norm": 0.05135999247431755,
      "learning_rate": 9.339457387528412e-06,
      "loss": 0.0043,
      "step": 45640
    },
    {
      "epoch": 0.7469524666612125,
      "grad_norm": 0.4018520712852478,
      "learning_rate": 9.338984428987915e-06,
      "loss": 0.0051,
      "step": 45650
    },
    {
      "epoch": 0.7471160926122883,
      "grad_norm": 0.11271344125270844,
      "learning_rate": 9.338511313169657e-06,
      "loss": 0.0024,
      "step": 45660
    },
    {
      "epoch": 0.7472797185633642,
      "grad_norm": 0.2551169991493225,
      "learning_rate": 9.338038040090793e-06,
      "loss": 0.002,
      "step": 45670
    },
    {
      "epoch": 0.74744334451444,
      "grad_norm": 0.2805965542793274,
      "learning_rate": 9.337564609768477e-06,
      "loss": 0.003,
      "step": 45680
    },
    {
      "epoch": 0.7476069704655158,
      "grad_norm": 0.04204365983605385,
      "learning_rate": 9.337091022219866e-06,
      "loss": 0.0018,
      "step": 45690
    },
    {
      "epoch": 0.7477705964165917,
      "grad_norm": 0.09109525382518768,
      "learning_rate": 9.336617277462131e-06,
      "loss": 0.0017,
      "step": 45700
    },
    {
      "epoch": 0.7479342223676675,
      "grad_norm": 0.33344176411628723,
      "learning_rate": 9.336143375512445e-06,
      "loss": 0.0039,
      "step": 45710
    },
    {
      "epoch": 0.7480978483187434,
      "grad_norm": 0.14615987241268158,
      "learning_rate": 9.335669316387981e-06,
      "loss": 0.0036,
      "step": 45720
    },
    {
      "epoch": 0.7482614742698191,
      "grad_norm": 0.2519879639148712,
      "learning_rate": 9.335195100105924e-06,
      "loss": 0.0052,
      "step": 45730
    },
    {
      "epoch": 0.748425100220895,
      "grad_norm": 0.18035291135311127,
      "learning_rate": 9.334720726683466e-06,
      "loss": 0.0039,
      "step": 45740
    },
    {
      "epoch": 0.7485887261719709,
      "grad_norm": 0.3346319794654846,
      "learning_rate": 9.334246196137798e-06,
      "loss": 0.0044,
      "step": 45750
    },
    {
      "epoch": 0.7487523521230467,
      "grad_norm": 0.230926513671875,
      "learning_rate": 9.333771508486122e-06,
      "loss": 0.0029,
      "step": 45760
    },
    {
      "epoch": 0.7489159780741226,
      "grad_norm": 0.09823073446750641,
      "learning_rate": 9.333296663745648e-06,
      "loss": 0.0034,
      "step": 45770
    },
    {
      "epoch": 0.7490796040251984,
      "grad_norm": 0.11959690600633621,
      "learning_rate": 9.332821661933582e-06,
      "loss": 0.0023,
      "step": 45780
    },
    {
      "epoch": 0.7492432299762742,
      "grad_norm": 0.13075628876686096,
      "learning_rate": 9.332346503067146e-06,
      "loss": 0.0021,
      "step": 45790
    },
    {
      "epoch": 0.7494068559273501,
      "grad_norm": 0.12976419925689697,
      "learning_rate": 9.33187118716356e-06,
      "loss": 0.0043,
      "step": 45800
    },
    {
      "epoch": 0.7495704818784259,
      "grad_norm": 0.22187969088554382,
      "learning_rate": 9.331395714240056e-06,
      "loss": 0.0032,
      "step": 45810
    },
    {
      "epoch": 0.7497341078295018,
      "grad_norm": 0.12239314615726471,
      "learning_rate": 9.330920084313868e-06,
      "loss": 0.0016,
      "step": 45820
    },
    {
      "epoch": 0.7498977337805776,
      "grad_norm": 0.06627237051725388,
      "learning_rate": 9.330444297402234e-06,
      "loss": 0.0014,
      "step": 45830
    },
    {
      "epoch": 0.7500613597316534,
      "grad_norm": 0.15680819749832153,
      "learning_rate": 9.329968353522405e-06,
      "loss": 0.0042,
      "step": 45840
    },
    {
      "epoch": 0.7502249856827293,
      "grad_norm": 0.15960851311683655,
      "learning_rate": 9.329492252691628e-06,
      "loss": 0.0027,
      "step": 45850
    },
    {
      "epoch": 0.7503886116338051,
      "grad_norm": 0.1933201253414154,
      "learning_rate": 9.329015994927164e-06,
      "loss": 0.0024,
      "step": 45860
    },
    {
      "epoch": 0.750552237584881,
      "grad_norm": 0.12402519583702087,
      "learning_rate": 9.328539580246274e-06,
      "loss": 0.0033,
      "step": 45870
    },
    {
      "epoch": 0.7507158635359568,
      "grad_norm": 0.2637035846710205,
      "learning_rate": 9.328063008666226e-06,
      "loss": 0.005,
      "step": 45880
    },
    {
      "epoch": 0.7508794894870326,
      "grad_norm": 0.0336226224899292,
      "learning_rate": 9.327586280204298e-06,
      "loss": 0.0028,
      "step": 45890
    },
    {
      "epoch": 0.7510431154381085,
      "grad_norm": 0.18882043659687042,
      "learning_rate": 9.327109394877768e-06,
      "loss": 0.0041,
      "step": 45900
    },
    {
      "epoch": 0.7512067413891843,
      "grad_norm": 0.15450966358184814,
      "learning_rate": 9.326632352703922e-06,
      "loss": 0.0073,
      "step": 45910
    },
    {
      "epoch": 0.7513703673402602,
      "grad_norm": 0.5061607360839844,
      "learning_rate": 9.326155153700053e-06,
      "loss": 0.0077,
      "step": 45920
    },
    {
      "epoch": 0.751533993291336,
      "grad_norm": 0.22285473346710205,
      "learning_rate": 9.325677797883457e-06,
      "loss": 0.0034,
      "step": 45930
    },
    {
      "epoch": 0.7516976192424119,
      "grad_norm": 0.1413329839706421,
      "learning_rate": 9.325200285271437e-06,
      "loss": 0.0042,
      "step": 45940
    },
    {
      "epoch": 0.7518612451934877,
      "grad_norm": 0.2556937336921692,
      "learning_rate": 9.3247226158813e-06,
      "loss": 0.0031,
      "step": 45950
    },
    {
      "epoch": 0.7520248711445635,
      "grad_norm": 0.14354334771633148,
      "learning_rate": 9.324244789730365e-06,
      "loss": 0.0051,
      "step": 45960
    },
    {
      "epoch": 0.7521884970956394,
      "grad_norm": 0.1389549970626831,
      "learning_rate": 9.32376680683595e-06,
      "loss": 0.0026,
      "step": 45970
    },
    {
      "epoch": 0.7523521230467152,
      "grad_norm": 0.12958888709545135,
      "learning_rate": 9.323288667215377e-06,
      "loss": 0.0036,
      "step": 45980
    },
    {
      "epoch": 0.7525157489977911,
      "grad_norm": 0.13619212806224823,
      "learning_rate": 9.322810370885984e-06,
      "loss": 0.0023,
      "step": 45990
    },
    {
      "epoch": 0.7526793749488669,
      "grad_norm": 0.07780390977859497,
      "learning_rate": 9.322331917865104e-06,
      "loss": 0.0023,
      "step": 46000
    },
    {
      "epoch": 0.7528430008999427,
      "grad_norm": 0.2370491474866867,
      "learning_rate": 9.32185330817008e-06,
      "loss": 0.0032,
      "step": 46010
    },
    {
      "epoch": 0.7530066268510186,
      "grad_norm": 0.16289196908473969,
      "learning_rate": 9.32137454181826e-06,
      "loss": 0.0028,
      "step": 46020
    },
    {
      "epoch": 0.7531702528020944,
      "grad_norm": 0.18555554747581482,
      "learning_rate": 9.320895618827e-06,
      "loss": 0.0033,
      "step": 46030
    },
    {
      "epoch": 0.7533338787531703,
      "grad_norm": 0.04601273313164711,
      "learning_rate": 9.320416539213656e-06,
      "loss": 0.0024,
      "step": 46040
    },
    {
      "epoch": 0.7534975047042461,
      "grad_norm": 0.15345878899097443,
      "learning_rate": 9.3199373029956e-06,
      "loss": 0.0046,
      "step": 46050
    },
    {
      "epoch": 0.7536611306553219,
      "grad_norm": 0.15434256196022034,
      "learning_rate": 9.319457910190197e-06,
      "loss": 0.0041,
      "step": 46060
    },
    {
      "epoch": 0.7538247566063978,
      "grad_norm": 0.07933124154806137,
      "learning_rate": 9.318978360814827e-06,
      "loss": 0.0031,
      "step": 46070
    },
    {
      "epoch": 0.7539883825574736,
      "grad_norm": 0.21763266623020172,
      "learning_rate": 9.318498654886874e-06,
      "loss": 0.0035,
      "step": 46080
    },
    {
      "epoch": 0.7541520085085495,
      "grad_norm": 0.07607048004865646,
      "learning_rate": 9.318018792423722e-06,
      "loss": 0.0022,
      "step": 46090
    },
    {
      "epoch": 0.7543156344596254,
      "grad_norm": 0.06933566927909851,
      "learning_rate": 9.317538773442766e-06,
      "loss": 0.0031,
      "step": 46100
    },
    {
      "epoch": 0.7544792604107011,
      "grad_norm": 0.13332277536392212,
      "learning_rate": 9.317058597961408e-06,
      "loss": 0.0022,
      "step": 46110
    },
    {
      "epoch": 0.754642886361777,
      "grad_norm": 0.1752205640077591,
      "learning_rate": 9.316578265997051e-06,
      "loss": 0.0049,
      "step": 46120
    },
    {
      "epoch": 0.7548065123128528,
      "grad_norm": 0.16984455287456512,
      "learning_rate": 9.316097777567105e-06,
      "loss": 0.0031,
      "step": 46130
    },
    {
      "epoch": 0.7549701382639287,
      "grad_norm": 0.19780689477920532,
      "learning_rate": 9.315617132688991e-06,
      "loss": 0.0031,
      "step": 46140
    },
    {
      "epoch": 0.7551337642150046,
      "grad_norm": 0.07790403068065643,
      "learning_rate": 9.315136331380126e-06,
      "loss": 0.003,
      "step": 46150
    },
    {
      "epoch": 0.7552973901660803,
      "grad_norm": 0.25169387459754944,
      "learning_rate": 9.314655373657942e-06,
      "loss": 0.0024,
      "step": 46160
    },
    {
      "epoch": 0.7554610161171562,
      "grad_norm": 0.15293337404727936,
      "learning_rate": 9.31417425953987e-06,
      "loss": 0.0027,
      "step": 46170
    },
    {
      "epoch": 0.755624642068232,
      "grad_norm": 0.09488566219806671,
      "learning_rate": 9.313692989043349e-06,
      "loss": 0.0024,
      "step": 46180
    },
    {
      "epoch": 0.7557882680193079,
      "grad_norm": 0.2919982969760895,
      "learning_rate": 9.313211562185826e-06,
      "loss": 0.0017,
      "step": 46190
    },
    {
      "epoch": 0.7559518939703836,
      "grad_norm": 0.05947905406355858,
      "learning_rate": 9.312729978984749e-06,
      "loss": 0.0035,
      "step": 46200
    },
    {
      "epoch": 0.7561155199214595,
      "grad_norm": 0.14913895726203918,
      "learning_rate": 9.312248239457578e-06,
      "loss": 0.0024,
      "step": 46210
    },
    {
      "epoch": 0.7562791458725354,
      "grad_norm": 0.10114043951034546,
      "learning_rate": 9.31176634362177e-06,
      "loss": 0.0021,
      "step": 46220
    },
    {
      "epoch": 0.7564427718236112,
      "grad_norm": 0.09346242249011993,
      "learning_rate": 9.311284291494796e-06,
      "loss": 0.0019,
      "step": 46230
    },
    {
      "epoch": 0.7566063977746871,
      "grad_norm": 0.07998554408550262,
      "learning_rate": 9.310802083094128e-06,
      "loss": 0.0035,
      "step": 46240
    },
    {
      "epoch": 0.7567700237257629,
      "grad_norm": 0.2284916639328003,
      "learning_rate": 9.310319718437247e-06,
      "loss": 0.004,
      "step": 46250
    },
    {
      "epoch": 0.7569336496768387,
      "grad_norm": 0.17484083771705627,
      "learning_rate": 9.309837197541634e-06,
      "loss": 0.0036,
      "step": 46260
    },
    {
      "epoch": 0.7570972756279146,
      "grad_norm": 0.21179313957691193,
      "learning_rate": 9.30935452042478e-06,
      "loss": 0.0071,
      "step": 46270
    },
    {
      "epoch": 0.7572609015789904,
      "grad_norm": 0.10206786543130875,
      "learning_rate": 9.308871687104184e-06,
      "loss": 0.0029,
      "step": 46280
    },
    {
      "epoch": 0.7574245275300663,
      "grad_norm": 0.6523825526237488,
      "learning_rate": 9.308388697597346e-06,
      "loss": 0.0037,
      "step": 46290
    },
    {
      "epoch": 0.7575881534811421,
      "grad_norm": 0.08849554508924484,
      "learning_rate": 9.307905551921769e-06,
      "loss": 0.0036,
      "step": 46300
    },
    {
      "epoch": 0.7577517794322179,
      "grad_norm": 0.15587395429611206,
      "learning_rate": 9.307422250094974e-06,
      "loss": 0.0035,
      "step": 46310
    },
    {
      "epoch": 0.7579154053832938,
      "grad_norm": 0.19340258836746216,
      "learning_rate": 9.306938792134471e-06,
      "loss": 0.0059,
      "step": 46320
    },
    {
      "epoch": 0.7580790313343696,
      "grad_norm": 0.32111769914627075,
      "learning_rate": 9.306455178057791e-06,
      "loss": 0.0025,
      "step": 46330
    },
    {
      "epoch": 0.7582426572854455,
      "grad_norm": 0.25783252716064453,
      "learning_rate": 9.305971407882462e-06,
      "loss": 0.0035,
      "step": 46340
    },
    {
      "epoch": 0.7584062832365213,
      "grad_norm": 0.1390017569065094,
      "learning_rate": 9.305487481626017e-06,
      "loss": 0.0036,
      "step": 46350
    },
    {
      "epoch": 0.7585699091875971,
      "grad_norm": 0.27731814980506897,
      "learning_rate": 9.305003399305997e-06,
      "loss": 0.0032,
      "step": 46360
    },
    {
      "epoch": 0.758733535138673,
      "grad_norm": 0.0679384097456932,
      "learning_rate": 9.304519160939954e-06,
      "loss": 0.0024,
      "step": 46370
    },
    {
      "epoch": 0.7588971610897488,
      "grad_norm": 0.3053310513496399,
      "learning_rate": 9.304034766545433e-06,
      "loss": 0.0025,
      "step": 46380
    },
    {
      "epoch": 0.7590607870408247,
      "grad_norm": 0.28973039984703064,
      "learning_rate": 9.303550216140001e-06,
      "loss": 0.0021,
      "step": 46390
    },
    {
      "epoch": 0.7592244129919005,
      "grad_norm": 0.10076013952493668,
      "learning_rate": 9.303065509741214e-06,
      "loss": 0.0037,
      "step": 46400
    },
    {
      "epoch": 0.7593880389429764,
      "grad_norm": 0.13185669481754303,
      "learning_rate": 9.302580647366643e-06,
      "loss": 0.0042,
      "step": 46410
    },
    {
      "epoch": 0.7595516648940522,
      "grad_norm": 0.06140631064772606,
      "learning_rate": 9.302095629033866e-06,
      "loss": 0.0035,
      "step": 46420
    },
    {
      "epoch": 0.759715290845128,
      "grad_norm": 0.30516910552978516,
      "learning_rate": 9.301610454760463e-06,
      "loss": 0.0048,
      "step": 46430
    },
    {
      "epoch": 0.7598789167962039,
      "grad_norm": 0.11276406049728394,
      "learning_rate": 9.301125124564021e-06,
      "loss": 0.0028,
      "step": 46440
    },
    {
      "epoch": 0.7600425427472797,
      "grad_norm": 0.09542270749807358,
      "learning_rate": 9.300639638462128e-06,
      "loss": 0.0058,
      "step": 46450
    },
    {
      "epoch": 0.7602061686983556,
      "grad_norm": 0.08172499388456345,
      "learning_rate": 9.300153996472386e-06,
      "loss": 0.0033,
      "step": 46460
    },
    {
      "epoch": 0.7603697946494314,
      "grad_norm": 0.08125490695238113,
      "learning_rate": 9.299668198612396e-06,
      "loss": 0.0028,
      "step": 46470
    },
    {
      "epoch": 0.7605334206005072,
      "grad_norm": 0.4125836193561554,
      "learning_rate": 9.299182244899768e-06,
      "loss": 0.0052,
      "step": 46480
    },
    {
      "epoch": 0.7606970465515831,
      "grad_norm": 0.16170723736286163,
      "learning_rate": 9.298696135352116e-06,
      "loss": 0.0036,
      "step": 46490
    },
    {
      "epoch": 0.7608606725026589,
      "grad_norm": 0.1094004362821579,
      "learning_rate": 9.298209869987062e-06,
      "loss": 0.0026,
      "step": 46500
    },
    {
      "epoch": 0.7610242984537348,
      "grad_norm": 0.22159212827682495,
      "learning_rate": 9.297723448822229e-06,
      "loss": 0.0031,
      "step": 46510
    },
    {
      "epoch": 0.7611879244048106,
      "grad_norm": 0.12661929428577423,
      "learning_rate": 9.297236871875253e-06,
      "loss": 0.0026,
      "step": 46520
    },
    {
      "epoch": 0.7613515503558864,
      "grad_norm": 0.3631120026111603,
      "learning_rate": 9.296750139163766e-06,
      "loss": 0.0048,
      "step": 46530
    },
    {
      "epoch": 0.7615151763069623,
      "grad_norm": 0.13580480217933655,
      "learning_rate": 9.296263250705415e-06,
      "loss": 0.003,
      "step": 46540
    },
    {
      "epoch": 0.7616788022580381,
      "grad_norm": 0.09591087698936462,
      "learning_rate": 9.295776206517847e-06,
      "loss": 0.0021,
      "step": 46550
    },
    {
      "epoch": 0.761842428209114,
      "grad_norm": 0.14948637783527374,
      "learning_rate": 9.295289006618713e-06,
      "loss": 0.0031,
      "step": 46560
    },
    {
      "epoch": 0.7620060541601898,
      "grad_norm": 0.45437610149383545,
      "learning_rate": 9.294801651025678e-06,
      "loss": 0.0038,
      "step": 46570
    },
    {
      "epoch": 0.7621696801112656,
      "grad_norm": 0.07336080819368362,
      "learning_rate": 9.294314139756406e-06,
      "loss": 0.0024,
      "step": 46580
    },
    {
      "epoch": 0.7623333060623415,
      "grad_norm": 0.03860901668667793,
      "learning_rate": 9.293826472828567e-06,
      "loss": 0.004,
      "step": 46590
    },
    {
      "epoch": 0.7624969320134173,
      "grad_norm": 0.0506257563829422,
      "learning_rate": 9.293338650259838e-06,
      "loss": 0.0027,
      "step": 46600
    },
    {
      "epoch": 0.7626605579644932,
      "grad_norm": 0.3378849923610687,
      "learning_rate": 9.292850672067904e-06,
      "loss": 0.0064,
      "step": 46610
    },
    {
      "epoch": 0.762824183915569,
      "grad_norm": 0.1885622888803482,
      "learning_rate": 9.292362538270448e-06,
      "loss": 0.0054,
      "step": 46620
    },
    {
      "epoch": 0.7629878098666448,
      "grad_norm": 0.12624213099479675,
      "learning_rate": 9.291874248885167e-06,
      "loss": 0.0033,
      "step": 46630
    },
    {
      "epoch": 0.7631514358177207,
      "grad_norm": 0.2975305914878845,
      "learning_rate": 9.29138580392976e-06,
      "loss": 0.0035,
      "step": 46640
    },
    {
      "epoch": 0.7633150617687965,
      "grad_norm": 0.29391130805015564,
      "learning_rate": 9.290897203421931e-06,
      "loss": 0.0027,
      "step": 46650
    },
    {
      "epoch": 0.7634786877198724,
      "grad_norm": 0.04558788239955902,
      "learning_rate": 9.290408447379393e-06,
      "loss": 0.0038,
      "step": 46660
    },
    {
      "epoch": 0.7636423136709483,
      "grad_norm": 0.3681044280529022,
      "learning_rate": 9.28991953581986e-06,
      "loss": 0.0021,
      "step": 46670
    },
    {
      "epoch": 0.763805939622024,
      "grad_norm": 0.21886494755744934,
      "learning_rate": 9.289430468761054e-06,
      "loss": 0.0035,
      "step": 46680
    },
    {
      "epoch": 0.7639695655730999,
      "grad_norm": 0.1452564299106598,
      "learning_rate": 9.288941246220701e-06,
      "loss": 0.0025,
      "step": 46690
    },
    {
      "epoch": 0.7641331915241757,
      "grad_norm": 0.28494179248809814,
      "learning_rate": 9.288451868216538e-06,
      "loss": 0.0035,
      "step": 46700
    },
    {
      "epoch": 0.7642968174752516,
      "grad_norm": 0.22606857120990753,
      "learning_rate": 9.287962334766302e-06,
      "loss": 0.0034,
      "step": 46710
    },
    {
      "epoch": 0.7644604434263275,
      "grad_norm": 0.3624116778373718,
      "learning_rate": 9.287472645887737e-06,
      "loss": 0.0031,
      "step": 46720
    },
    {
      "epoch": 0.7646240693774032,
      "grad_norm": 0.13198938965797424,
      "learning_rate": 9.286982801598595e-06,
      "loss": 0.0024,
      "step": 46730
    },
    {
      "epoch": 0.7647876953284791,
      "grad_norm": 0.3184228837490082,
      "learning_rate": 9.286492801916626e-06,
      "loss": 0.0043,
      "step": 46740
    },
    {
      "epoch": 0.7649513212795549,
      "grad_norm": 0.06921160966157913,
      "learning_rate": 9.286002646859598e-06,
      "loss": 0.0031,
      "step": 46750
    },
    {
      "epoch": 0.7651149472306308,
      "grad_norm": 0.18555854260921478,
      "learning_rate": 9.285512336445275e-06,
      "loss": 0.0035,
      "step": 46760
    },
    {
      "epoch": 0.7652785731817067,
      "grad_norm": 0.1534012258052826,
      "learning_rate": 9.285021870691429e-06,
      "loss": 0.0038,
      "step": 46770
    },
    {
      "epoch": 0.7654421991327824,
      "grad_norm": 0.3063267469406128,
      "learning_rate": 9.284531249615838e-06,
      "loss": 0.0046,
      "step": 46780
    },
    {
      "epoch": 0.7656058250838583,
      "grad_norm": 0.6759603023529053,
      "learning_rate": 9.284040473236288e-06,
      "loss": 0.0048,
      "step": 46790
    },
    {
      "epoch": 0.7657694510349341,
      "grad_norm": 0.321036159992218,
      "learning_rate": 9.283549541570568e-06,
      "loss": 0.0028,
      "step": 46800
    },
    {
      "epoch": 0.76593307698601,
      "grad_norm": 0.20663334429264069,
      "learning_rate": 9.28305845463647e-06,
      "loss": 0.0025,
      "step": 46810
    },
    {
      "epoch": 0.7660967029370859,
      "grad_norm": 0.11464294046163559,
      "learning_rate": 9.2825672124518e-06,
      "loss": 0.0046,
      "step": 46820
    },
    {
      "epoch": 0.7662603288881616,
      "grad_norm": 0.18782205879688263,
      "learning_rate": 9.28207581503436e-06,
      "loss": 0.0031,
      "step": 46830
    },
    {
      "epoch": 0.7664239548392375,
      "grad_norm": 0.03613877296447754,
      "learning_rate": 9.281584262401963e-06,
      "loss": 0.0066,
      "step": 46840
    },
    {
      "epoch": 0.7665875807903133,
      "grad_norm": 0.09105803817510605,
      "learning_rate": 9.281092554572428e-06,
      "loss": 0.0022,
      "step": 46850
    },
    {
      "epoch": 0.7667512067413892,
      "grad_norm": 0.2554304003715515,
      "learning_rate": 9.280600691563577e-06,
      "loss": 0.0038,
      "step": 46860
    },
    {
      "epoch": 0.7669148326924651,
      "grad_norm": 0.2795160710811615,
      "learning_rate": 9.28010867339324e-06,
      "loss": 0.0037,
      "step": 46870
    },
    {
      "epoch": 0.7670784586435409,
      "grad_norm": 0.15072228014469147,
      "learning_rate": 9.279616500079249e-06,
      "loss": 0.0032,
      "step": 46880
    },
    {
      "epoch": 0.7672420845946167,
      "grad_norm": 0.30160611867904663,
      "learning_rate": 9.279124171639446e-06,
      "loss": 0.0032,
      "step": 46890
    },
    {
      "epoch": 0.7674057105456925,
      "grad_norm": 0.49359646439552307,
      "learning_rate": 9.278631688091675e-06,
      "loss": 0.0042,
      "step": 46900
    },
    {
      "epoch": 0.7675693364967684,
      "grad_norm": 0.06053291633725166,
      "learning_rate": 9.278139049453792e-06,
      "loss": 0.003,
      "step": 46910
    },
    {
      "epoch": 0.7677329624478443,
      "grad_norm": 0.17476892471313477,
      "learning_rate": 9.277646255743648e-06,
      "loss": 0.0031,
      "step": 46920
    },
    {
      "epoch": 0.76789658839892,
      "grad_norm": 0.1530722677707672,
      "learning_rate": 9.27715330697911e-06,
      "loss": 0.0036,
      "step": 46930
    },
    {
      "epoch": 0.7680602143499959,
      "grad_norm": 0.16645754873752594,
      "learning_rate": 9.276660203178043e-06,
      "loss": 0.0048,
      "step": 46940
    },
    {
      "epoch": 0.7682238403010717,
      "grad_norm": 0.12035121023654938,
      "learning_rate": 9.276166944358322e-06,
      "loss": 0.0043,
      "step": 46950
    },
    {
      "epoch": 0.7683874662521476,
      "grad_norm": 0.27420181035995483,
      "learning_rate": 9.275673530537827e-06,
      "loss": 0.0049,
      "step": 46960
    },
    {
      "epoch": 0.7685510922032235,
      "grad_norm": 0.22867822647094727,
      "learning_rate": 9.275179961734444e-06,
      "loss": 0.0024,
      "step": 46970
    },
    {
      "epoch": 0.7687147181542993,
      "grad_norm": 0.21577972173690796,
      "learning_rate": 9.274686237966063e-06,
      "loss": 0.0029,
      "step": 46980
    },
    {
      "epoch": 0.7688783441053751,
      "grad_norm": 0.1786114126443863,
      "learning_rate": 9.27419235925058e-06,
      "loss": 0.0024,
      "step": 46990
    },
    {
      "epoch": 0.7690419700564509,
      "grad_norm": 0.05456943437457085,
      "learning_rate": 9.273698325605894e-06,
      "loss": 0.0042,
      "step": 47000
    },
    {
      "epoch": 0.7692055960075268,
      "grad_norm": 0.46462979912757874,
      "learning_rate": 9.273204137049916e-06,
      "loss": 0.0026,
      "step": 47010
    },
    {
      "epoch": 0.7693692219586027,
      "grad_norm": 0.08982861787080765,
      "learning_rate": 9.27270979360056e-06,
      "loss": 0.0023,
      "step": 47020
    },
    {
      "epoch": 0.7695328479096785,
      "grad_norm": 0.17524635791778564,
      "learning_rate": 9.272215295275743e-06,
      "loss": 0.0043,
      "step": 47030
    },
    {
      "epoch": 0.7696964738607543,
      "grad_norm": 0.1675378978252411,
      "learning_rate": 9.27172064209339e-06,
      "loss": 0.0035,
      "step": 47040
    },
    {
      "epoch": 0.7698600998118301,
      "grad_norm": 0.11323246359825134,
      "learning_rate": 9.271225834071428e-06,
      "loss": 0.0071,
      "step": 47050
    },
    {
      "epoch": 0.770023725762906,
      "grad_norm": 0.09345797449350357,
      "learning_rate": 9.270730871227796e-06,
      "loss": 0.0022,
      "step": 47060
    },
    {
      "epoch": 0.7701873517139818,
      "grad_norm": 0.07348580658435822,
      "learning_rate": 9.270235753580436e-06,
      "loss": 0.0043,
      "step": 47070
    },
    {
      "epoch": 0.7703509776650577,
      "grad_norm": 0.1204700842499733,
      "learning_rate": 9.26974048114729e-06,
      "loss": 0.0023,
      "step": 47080
    },
    {
      "epoch": 0.7705146036161336,
      "grad_norm": 0.1438237875699997,
      "learning_rate": 9.269245053946318e-06,
      "loss": 0.0035,
      "step": 47090
    },
    {
      "epoch": 0.7706782295672093,
      "grad_norm": 0.1774197667837143,
      "learning_rate": 9.268749471995472e-06,
      "loss": 0.0038,
      "step": 47100
    },
    {
      "epoch": 0.7708418555182852,
      "grad_norm": 0.14052198827266693,
      "learning_rate": 9.268253735312716e-06,
      "loss": 0.003,
      "step": 47110
    },
    {
      "epoch": 0.771005481469361,
      "grad_norm": 0.09925854206085205,
      "learning_rate": 9.26775784391602e-06,
      "loss": 0.0028,
      "step": 47120
    },
    {
      "epoch": 0.7711691074204369,
      "grad_norm": 0.03176082670688629,
      "learning_rate": 9.267261797823362e-06,
      "loss": 0.0046,
      "step": 47130
    },
    {
      "epoch": 0.7713327333715128,
      "grad_norm": 0.2940159738063812,
      "learning_rate": 9.266765597052718e-06,
      "loss": 0.0033,
      "step": 47140
    },
    {
      "epoch": 0.7714963593225885,
      "grad_norm": 0.07327362149953842,
      "learning_rate": 9.266269241622077e-06,
      "loss": 0.0034,
      "step": 47150
    },
    {
      "epoch": 0.7716599852736644,
      "grad_norm": 0.1620359569787979,
      "learning_rate": 9.26577273154943e-06,
      "loss": 0.0025,
      "step": 47160
    },
    {
      "epoch": 0.7718236112247402,
      "grad_norm": 0.1572045236825943,
      "learning_rate": 9.265276066852774e-06,
      "loss": 0.0023,
      "step": 47170
    },
    {
      "epoch": 0.7719872371758161,
      "grad_norm": 1.5911974906921387,
      "learning_rate": 9.264779247550111e-06,
      "loss": 0.0027,
      "step": 47180
    },
    {
      "epoch": 0.772150863126892,
      "grad_norm": 0.22655841708183289,
      "learning_rate": 9.26428227365945e-06,
      "loss": 0.0041,
      "step": 47190
    },
    {
      "epoch": 0.7723144890779677,
      "grad_norm": 0.16720829904079437,
      "learning_rate": 9.263785145198806e-06,
      "loss": 0.0044,
      "step": 47200
    },
    {
      "epoch": 0.7724781150290436,
      "grad_norm": 0.15368755161762238,
      "learning_rate": 9.263287862186198e-06,
      "loss": 0.0022,
      "step": 47210
    },
    {
      "epoch": 0.7726417409801194,
      "grad_norm": 0.08605608344078064,
      "learning_rate": 9.262790424639651e-06,
      "loss": 0.0034,
      "step": 47220
    },
    {
      "epoch": 0.7728053669311953,
      "grad_norm": 0.25703421235084534,
      "learning_rate": 9.2622928325772e-06,
      "loss": 0.0039,
      "step": 47230
    },
    {
      "epoch": 0.7729689928822712,
      "grad_norm": 0.5112340450286865,
      "learning_rate": 9.261795086016873e-06,
      "loss": 0.003,
      "step": 47240
    },
    {
      "epoch": 0.7731326188333469,
      "grad_norm": 0.13069429993629456,
      "learning_rate": 9.261297184976721e-06,
      "loss": 0.0034,
      "step": 47250
    },
    {
      "epoch": 0.7732962447844228,
      "grad_norm": 0.25137898325920105,
      "learning_rate": 9.260799129474785e-06,
      "loss": 0.0019,
      "step": 47260
    },
    {
      "epoch": 0.7734598707354986,
      "grad_norm": 0.2937735617160797,
      "learning_rate": 9.260300919529123e-06,
      "loss": 0.0048,
      "step": 47270
    },
    {
      "epoch": 0.7736234966865745,
      "grad_norm": 0.19685785472393036,
      "learning_rate": 9.259802555157792e-06,
      "loss": 0.0029,
      "step": 47280
    },
    {
      "epoch": 0.7737871226376504,
      "grad_norm": 0.08265804499387741,
      "learning_rate": 9.259304036378856e-06,
      "loss": 0.0037,
      "step": 47290
    },
    {
      "epoch": 0.7739507485887261,
      "grad_norm": 0.018450217321515083,
      "learning_rate": 9.258805363210386e-06,
      "loss": 0.0033,
      "step": 47300
    },
    {
      "epoch": 0.774114374539802,
      "grad_norm": 0.051352452486753464,
      "learning_rate": 9.258306535670457e-06,
      "loss": 0.0019,
      "step": 47310
    },
    {
      "epoch": 0.7742780004908778,
      "grad_norm": 0.3335985243320465,
      "learning_rate": 9.25780755377715e-06,
      "loss": 0.0028,
      "step": 47320
    },
    {
      "epoch": 0.7744416264419537,
      "grad_norm": 0.1750146746635437,
      "learning_rate": 9.257308417548553e-06,
      "loss": 0.0031,
      "step": 47330
    },
    {
      "epoch": 0.7746052523930296,
      "grad_norm": 0.080111064016819,
      "learning_rate": 9.256809127002759e-06,
      "loss": 0.0032,
      "step": 47340
    },
    {
      "epoch": 0.7747688783441053,
      "grad_norm": 0.15874405205249786,
      "learning_rate": 9.256309682157863e-06,
      "loss": 0.0028,
      "step": 47350
    },
    {
      "epoch": 0.7749325042951812,
      "grad_norm": 0.3410603404045105,
      "learning_rate": 9.255810083031973e-06,
      "loss": 0.0026,
      "step": 47360
    },
    {
      "epoch": 0.775096130246257,
      "grad_norm": 0.07764293253421783,
      "learning_rate": 9.255310329643195e-06,
      "loss": 0.0029,
      "step": 47370
    },
    {
      "epoch": 0.7752597561973329,
      "grad_norm": 0.0763319730758667,
      "learning_rate": 9.254810422009645e-06,
      "loss": 0.0025,
      "step": 47380
    },
    {
      "epoch": 0.7754233821484088,
      "grad_norm": 0.0930410772562027,
      "learning_rate": 9.254310360149444e-06,
      "loss": 0.0022,
      "step": 47390
    },
    {
      "epoch": 0.7755870080994846,
      "grad_norm": 0.22531259059906006,
      "learning_rate": 9.253810144080716e-06,
      "loss": 0.0026,
      "step": 47400
    },
    {
      "epoch": 0.7757506340505604,
      "grad_norm": 0.1400601714849472,
      "learning_rate": 9.253309773821595e-06,
      "loss": 0.0043,
      "step": 47410
    },
    {
      "epoch": 0.7759142600016362,
      "grad_norm": 0.2750956118106842,
      "learning_rate": 9.252809249390218e-06,
      "loss": 0.0025,
      "step": 47420
    },
    {
      "epoch": 0.7760778859527121,
      "grad_norm": 0.1686827540397644,
      "learning_rate": 9.252308570804725e-06,
      "loss": 0.0028,
      "step": 47430
    },
    {
      "epoch": 0.776241511903788,
      "grad_norm": 0.4864672124385834,
      "learning_rate": 9.251807738083268e-06,
      "loss": 0.0048,
      "step": 47440
    },
    {
      "epoch": 0.7764051378548638,
      "grad_norm": 0.17237213253974915,
      "learning_rate": 9.251306751244e-06,
      "loss": 0.0028,
      "step": 47450
    },
    {
      "epoch": 0.7765687638059396,
      "grad_norm": 0.47553494572639465,
      "learning_rate": 9.250805610305079e-06,
      "loss": 0.0047,
      "step": 47460
    },
    {
      "epoch": 0.7767323897570154,
      "grad_norm": 0.19099704921245575,
      "learning_rate": 9.25030431528467e-06,
      "loss": 0.0042,
      "step": 47470
    },
    {
      "epoch": 0.7768960157080913,
      "grad_norm": 0.19183245301246643,
      "learning_rate": 9.249802866200948e-06,
      "loss": 0.0024,
      "step": 47480
    },
    {
      "epoch": 0.7770596416591672,
      "grad_norm": 0.15168438851833344,
      "learning_rate": 9.249301263072083e-06,
      "loss": 0.004,
      "step": 47490
    },
    {
      "epoch": 0.777223267610243,
      "grad_norm": 0.26051074266433716,
      "learning_rate": 9.248799505916261e-06,
      "loss": 0.0033,
      "step": 47500
    },
    {
      "epoch": 0.7773868935613188,
      "grad_norm": 0.202837735414505,
      "learning_rate": 9.24829759475167e-06,
      "loss": 0.0029,
      "step": 47510
    },
    {
      "epoch": 0.7775505195123946,
      "grad_norm": 0.17700165510177612,
      "learning_rate": 9.247795529596501e-06,
      "loss": 0.0026,
      "step": 47520
    },
    {
      "epoch": 0.7777141454634705,
      "grad_norm": 0.2127302885055542,
      "learning_rate": 9.247293310468954e-06,
      "loss": 0.0034,
      "step": 47530
    },
    {
      "epoch": 0.7778777714145464,
      "grad_norm": 0.13977067172527313,
      "learning_rate": 9.246790937387231e-06,
      "loss": 0.0057,
      "step": 47540
    },
    {
      "epoch": 0.7780413973656222,
      "grad_norm": 0.10118871182203293,
      "learning_rate": 9.246288410369545e-06,
      "loss": 0.003,
      "step": 47550
    },
    {
      "epoch": 0.778205023316698,
      "grad_norm": 0.17957332730293274,
      "learning_rate": 9.24578572943411e-06,
      "loss": 0.0033,
      "step": 47560
    },
    {
      "epoch": 0.7783686492677738,
      "grad_norm": 0.2622086703777313,
      "learning_rate": 9.245282894599145e-06,
      "loss": 0.0038,
      "step": 47570
    },
    {
      "epoch": 0.7785322752188497,
      "grad_norm": 0.10515914112329483,
      "learning_rate": 9.24477990588288e-06,
      "loss": 0.0031,
      "step": 47580
    },
    {
      "epoch": 0.7786959011699256,
      "grad_norm": 0.283098042011261,
      "learning_rate": 9.244276763303544e-06,
      "loss": 0.0042,
      "step": 47590
    },
    {
      "epoch": 0.7788595271210014,
      "grad_norm": 0.295404314994812,
      "learning_rate": 9.243773466879377e-06,
      "loss": 0.002,
      "step": 47600
    },
    {
      "epoch": 0.7790231530720773,
      "grad_norm": 0.08057957887649536,
      "learning_rate": 9.24327001662862e-06,
      "loss": 0.0024,
      "step": 47610
    },
    {
      "epoch": 0.779186779023153,
      "grad_norm": 0.2663296163082123,
      "learning_rate": 9.242766412569526e-06,
      "loss": 0.0038,
      "step": 47620
    },
    {
      "epoch": 0.7793504049742289,
      "grad_norm": 0.14266766607761383,
      "learning_rate": 9.242262654720345e-06,
      "loss": 0.0022,
      "step": 47630
    },
    {
      "epoch": 0.7795140309253048,
      "grad_norm": 0.22005395591259003,
      "learning_rate": 9.24175874309934e-06,
      "loss": 0.0034,
      "step": 47640
    },
    {
      "epoch": 0.7796776568763806,
      "grad_norm": 0.07073821872472763,
      "learning_rate": 9.241254677724773e-06,
      "loss": 0.0033,
      "step": 47650
    },
    {
      "epoch": 0.7798412828274565,
      "grad_norm": 0.4702405631542206,
      "learning_rate": 9.24075045861492e-06,
      "loss": 0.0029,
      "step": 47660
    },
    {
      "epoch": 0.7800049087785322,
      "grad_norm": 0.12413022667169571,
      "learning_rate": 9.240246085788055e-06,
      "loss": 0.0029,
      "step": 47670
    },
    {
      "epoch": 0.7801685347296081,
      "grad_norm": 0.11062261462211609,
      "learning_rate": 9.23974155926246e-06,
      "loss": 0.004,
      "step": 47680
    },
    {
      "epoch": 0.780332160680684,
      "grad_norm": 0.2187574952840805,
      "learning_rate": 9.239236879056423e-06,
      "loss": 0.0033,
      "step": 47690
    },
    {
      "epoch": 0.7804957866317598,
      "grad_norm": 0.1472589373588562,
      "learning_rate": 9.23873204518824e-06,
      "loss": 0.0028,
      "step": 47700
    },
    {
      "epoch": 0.7806594125828357,
      "grad_norm": 0.21112596988677979,
      "learning_rate": 9.238227057676204e-06,
      "loss": 0.0051,
      "step": 47710
    },
    {
      "epoch": 0.7808230385339114,
      "grad_norm": 0.09035231173038483,
      "learning_rate": 9.237721916538625e-06,
      "loss": 0.004,
      "step": 47720
    },
    {
      "epoch": 0.7809866644849873,
      "grad_norm": 0.2239183932542801,
      "learning_rate": 9.237216621793812e-06,
      "loss": 0.0043,
      "step": 47730
    },
    {
      "epoch": 0.7811502904360632,
      "grad_norm": 0.14135926961898804,
      "learning_rate": 9.236711173460082e-06,
      "loss": 0.0056,
      "step": 47740
    },
    {
      "epoch": 0.781313916387139,
      "grad_norm": 0.10764330625534058,
      "learning_rate": 9.236205571555752e-06,
      "loss": 0.0019,
      "step": 47750
    },
    {
      "epoch": 0.7814775423382149,
      "grad_norm": 0.14232297241687775,
      "learning_rate": 9.235699816099153e-06,
      "loss": 0.0045,
      "step": 47760
    },
    {
      "epoch": 0.7816411682892906,
      "grad_norm": 0.09331347048282623,
      "learning_rate": 9.235193907108617e-06,
      "loss": 0.0038,
      "step": 47770
    },
    {
      "epoch": 0.7818047942403665,
      "grad_norm": 0.28311392664909363,
      "learning_rate": 9.234687844602479e-06,
      "loss": 0.0066,
      "step": 47780
    },
    {
      "epoch": 0.7819684201914424,
      "grad_norm": 0.15356793999671936,
      "learning_rate": 9.234181628599083e-06,
      "loss": 0.0021,
      "step": 47790
    },
    {
      "epoch": 0.7821320461425182,
      "grad_norm": 0.2692772448062897,
      "learning_rate": 9.233675259116783e-06,
      "loss": 0.0056,
      "step": 47800
    },
    {
      "epoch": 0.7822956720935941,
      "grad_norm": 0.23091362416744232,
      "learning_rate": 9.233168736173928e-06,
      "loss": 0.0049,
      "step": 47810
    },
    {
      "epoch": 0.7824592980446698,
      "grad_norm": 0.06629378348588943,
      "learning_rate": 9.232662059788881e-06,
      "loss": 0.0043,
      "step": 47820
    },
    {
      "epoch": 0.7826229239957457,
      "grad_norm": 0.1617034375667572,
      "learning_rate": 9.232155229980007e-06,
      "loss": 0.0021,
      "step": 47830
    },
    {
      "epoch": 0.7827865499468216,
      "grad_norm": 0.19493144750595093,
      "learning_rate": 9.23164824676568e-06,
      "loss": 0.0034,
      "step": 47840
    },
    {
      "epoch": 0.7829501758978974,
      "grad_norm": 0.060315508395433426,
      "learning_rate": 9.231141110164273e-06,
      "loss": 0.0029,
      "step": 47850
    },
    {
      "epoch": 0.7831138018489733,
      "grad_norm": 0.05947552248835564,
      "learning_rate": 9.230633820194168e-06,
      "loss": 0.0026,
      "step": 47860
    },
    {
      "epoch": 0.783277427800049,
      "grad_norm": 0.19408294558525085,
      "learning_rate": 9.230126376873757e-06,
      "loss": 0.0039,
      "step": 47870
    },
    {
      "epoch": 0.7834410537511249,
      "grad_norm": 0.20794975757598877,
      "learning_rate": 9.229618780221433e-06,
      "loss": 0.0039,
      "step": 47880
    },
    {
      "epoch": 0.7836046797022008,
      "grad_norm": 0.23113468289375305,
      "learning_rate": 9.229111030255591e-06,
      "loss": 0.0016,
      "step": 47890
    },
    {
      "epoch": 0.7837683056532766,
      "grad_norm": 0.18818505108356476,
      "learning_rate": 9.22860312699464e-06,
      "loss": 0.0035,
      "step": 47900
    },
    {
      "epoch": 0.7839319316043525,
      "grad_norm": 0.0857291892170906,
      "learning_rate": 9.228095070456989e-06,
      "loss": 0.0037,
      "step": 47910
    },
    {
      "epoch": 0.7840955575554283,
      "grad_norm": 0.04636092111468315,
      "learning_rate": 9.227586860661053e-06,
      "loss": 0.0032,
      "step": 47920
    },
    {
      "epoch": 0.7842591835065041,
      "grad_norm": 0.08256290853023529,
      "learning_rate": 9.227078497625253e-06,
      "loss": 0.0034,
      "step": 47930
    },
    {
      "epoch": 0.7844228094575799,
      "grad_norm": 0.09330001473426819,
      "learning_rate": 9.226569981368018e-06,
      "loss": 0.0073,
      "step": 47940
    },
    {
      "epoch": 0.7845864354086558,
      "grad_norm": 0.21023191511631012,
      "learning_rate": 9.22606131190778e-06,
      "loss": 0.005,
      "step": 47950
    },
    {
      "epoch": 0.7847500613597317,
      "grad_norm": 0.042733803391456604,
      "learning_rate": 9.225552489262977e-06,
      "loss": 0.0031,
      "step": 47960
    },
    {
      "epoch": 0.7849136873108075,
      "grad_norm": 0.09576703608036041,
      "learning_rate": 9.225043513452052e-06,
      "loss": 0.0039,
      "step": 47970
    },
    {
      "epoch": 0.7850773132618833,
      "grad_norm": 0.5880565643310547,
      "learning_rate": 9.224534384493453e-06,
      "loss": 0.0033,
      "step": 47980
    },
    {
      "epoch": 0.7852409392129591,
      "grad_norm": 0.24097760021686554,
      "learning_rate": 9.224025102405636e-06,
      "loss": 0.0036,
      "step": 47990
    },
    {
      "epoch": 0.785404565164035,
      "grad_norm": 0.12412737309932709,
      "learning_rate": 9.223515667207061e-06,
      "loss": 0.0044,
      "step": 48000
    },
    {
      "epoch": 0.7855681911151109,
      "grad_norm": 0.15705639123916626,
      "learning_rate": 9.223006078916196e-06,
      "loss": 0.0034,
      "step": 48010
    },
    {
      "epoch": 0.7857318170661867,
      "grad_norm": 0.198961541056633,
      "learning_rate": 9.222496337551509e-06,
      "loss": 0.0031,
      "step": 48020
    },
    {
      "epoch": 0.7858954430172626,
      "grad_norm": 0.21408711373806,
      "learning_rate": 9.221986443131479e-06,
      "loss": 0.0035,
      "step": 48030
    },
    {
      "epoch": 0.7860590689683383,
      "grad_norm": 0.19704949855804443,
      "learning_rate": 9.221476395674586e-06,
      "loss": 0.0038,
      "step": 48040
    },
    {
      "epoch": 0.7862226949194142,
      "grad_norm": 0.10040422528982162,
      "learning_rate": 9.220966195199321e-06,
      "loss": 0.0021,
      "step": 48050
    },
    {
      "epoch": 0.7863863208704901,
      "grad_norm": 0.04798143729567528,
      "learning_rate": 9.220455841724176e-06,
      "loss": 0.0018,
      "step": 48060
    },
    {
      "epoch": 0.7865499468215659,
      "grad_norm": 0.18889322876930237,
      "learning_rate": 9.21994533526765e-06,
      "loss": 0.0035,
      "step": 48070
    },
    {
      "epoch": 0.7867135727726418,
      "grad_norm": 0.1173163652420044,
      "learning_rate": 9.21943467584825e-06,
      "loss": 0.0028,
      "step": 48080
    },
    {
      "epoch": 0.7868771987237175,
      "grad_norm": 0.14069336652755737,
      "learning_rate": 9.21892386348448e-06,
      "loss": 0.0021,
      "step": 48090
    },
    {
      "epoch": 0.7870408246747934,
      "grad_norm": 0.1269276887178421,
      "learning_rate": 9.218412898194865e-06,
      "loss": 0.0027,
      "step": 48100
    },
    {
      "epoch": 0.7872044506258693,
      "grad_norm": 0.24161434173583984,
      "learning_rate": 9.217901779997918e-06,
      "loss": 0.0022,
      "step": 48110
    },
    {
      "epoch": 0.7873680765769451,
      "grad_norm": 0.07993855327367783,
      "learning_rate": 9.217390508912169e-06,
      "loss": 0.0033,
      "step": 48120
    },
    {
      "epoch": 0.787531702528021,
      "grad_norm": 0.00963317696005106,
      "learning_rate": 9.216879084956152e-06,
      "loss": 0.0021,
      "step": 48130
    },
    {
      "epoch": 0.7876953284790967,
      "grad_norm": 0.09157359600067139,
      "learning_rate": 9.216367508148401e-06,
      "loss": 0.0022,
      "step": 48140
    },
    {
      "epoch": 0.7878589544301726,
      "grad_norm": 0.13302327692508698,
      "learning_rate": 9.215855778507463e-06,
      "loss": 0.0026,
      "step": 48150
    },
    {
      "epoch": 0.7880225803812485,
      "grad_norm": 0.10796001553535461,
      "learning_rate": 9.215343896051885e-06,
      "loss": 0.0028,
      "step": 48160
    },
    {
      "epoch": 0.7881862063323243,
      "grad_norm": 0.15687522292137146,
      "learning_rate": 9.214831860800221e-06,
      "loss": 0.003,
      "step": 48170
    },
    {
      "epoch": 0.7883498322834002,
      "grad_norm": 0.019233981147408485,
      "learning_rate": 9.214319672771036e-06,
      "loss": 0.0033,
      "step": 48180
    },
    {
      "epoch": 0.7885134582344759,
      "grad_norm": 0.09423023462295532,
      "learning_rate": 9.213807331982887e-06,
      "loss": 0.0026,
      "step": 48190
    },
    {
      "epoch": 0.7886770841855518,
      "grad_norm": 0.08601681888103485,
      "learning_rate": 9.213294838454352e-06,
      "loss": 0.0049,
      "step": 48200
    },
    {
      "epoch": 0.7888407101366277,
      "grad_norm": 0.0757230892777443,
      "learning_rate": 9.212782192204007e-06,
      "loss": 0.0032,
      "step": 48210
    },
    {
      "epoch": 0.7890043360877035,
      "grad_norm": 0.25804081559181213,
      "learning_rate": 9.21226939325043e-06,
      "loss": 0.0018,
      "step": 48220
    },
    {
      "epoch": 0.7891679620387794,
      "grad_norm": 0.06971985101699829,
      "learning_rate": 9.211756441612211e-06,
      "loss": 0.0033,
      "step": 48230
    },
    {
      "epoch": 0.7893315879898551,
      "grad_norm": 0.12275729328393936,
      "learning_rate": 9.211243337307947e-06,
      "loss": 0.0026,
      "step": 48240
    },
    {
      "epoch": 0.789495213940931,
      "grad_norm": 0.5567205548286438,
      "learning_rate": 9.210730080356232e-06,
      "loss": 0.0032,
      "step": 48250
    },
    {
      "epoch": 0.7896588398920069,
      "grad_norm": 0.048873912543058395,
      "learning_rate": 9.210216670775671e-06,
      "loss": 0.0041,
      "step": 48260
    },
    {
      "epoch": 0.7898224658430827,
      "grad_norm": 0.06869818270206451,
      "learning_rate": 9.209703108584876e-06,
      "loss": 0.0036,
      "step": 48270
    },
    {
      "epoch": 0.7899860917941586,
      "grad_norm": 0.09845008701086044,
      "learning_rate": 9.20918939380246e-06,
      "loss": 0.0033,
      "step": 48280
    },
    {
      "epoch": 0.7901497177452343,
      "grad_norm": 0.09102193266153336,
      "learning_rate": 9.208675526447044e-06,
      "loss": 0.0024,
      "step": 48290
    },
    {
      "epoch": 0.7903133436963102,
      "grad_norm": 0.29814067482948303,
      "learning_rate": 9.208161506537257e-06,
      "loss": 0.0048,
      "step": 48300
    },
    {
      "epoch": 0.7904769696473861,
      "grad_norm": 0.08440706878900528,
      "learning_rate": 9.207647334091729e-06,
      "loss": 0.0026,
      "step": 48310
    },
    {
      "epoch": 0.7906405955984619,
      "grad_norm": 0.27004656195640564,
      "learning_rate": 9.207133009129098e-06,
      "loss": 0.0041,
      "step": 48320
    },
    {
      "epoch": 0.7908042215495378,
      "grad_norm": 0.10873109847307205,
      "learning_rate": 9.206618531668007e-06,
      "loss": 0.0029,
      "step": 48330
    },
    {
      "epoch": 0.7909678475006136,
      "grad_norm": 0.168745756149292,
      "learning_rate": 9.206103901727104e-06,
      "loss": 0.0045,
      "step": 48340
    },
    {
      "epoch": 0.7911314734516894,
      "grad_norm": 0.1363922506570816,
      "learning_rate": 9.205589119325045e-06,
      "loss": 0.0043,
      "step": 48350
    },
    {
      "epoch": 0.7912950994027653,
      "grad_norm": 0.1527826189994812,
      "learning_rate": 9.205074184480488e-06,
      "loss": 0.0043,
      "step": 48360
    },
    {
      "epoch": 0.7914587253538411,
      "grad_norm": 0.11487593501806259,
      "learning_rate": 9.204559097212097e-06,
      "loss": 0.0047,
      "step": 48370
    },
    {
      "epoch": 0.791622351304917,
      "grad_norm": 0.2888651490211487,
      "learning_rate": 9.204043857538546e-06,
      "loss": 0.0047,
      "step": 48380
    },
    {
      "epoch": 0.7917859772559928,
      "grad_norm": 0.09659779816865921,
      "learning_rate": 9.203528465478509e-06,
      "loss": 0.0037,
      "step": 48390
    },
    {
      "epoch": 0.7919496032070686,
      "grad_norm": 0.07407717406749725,
      "learning_rate": 9.203012921050667e-06,
      "loss": 0.0023,
      "step": 48400
    },
    {
      "epoch": 0.7921132291581445,
      "grad_norm": 0.20125484466552734,
      "learning_rate": 9.202497224273708e-06,
      "loss": 0.0026,
      "step": 48410
    },
    {
      "epoch": 0.7922768551092203,
      "grad_norm": 0.12445491552352905,
      "learning_rate": 9.201981375166326e-06,
      "loss": 0.0024,
      "step": 48420
    },
    {
      "epoch": 0.7924404810602962,
      "grad_norm": 0.17671945691108704,
      "learning_rate": 9.201465373747219e-06,
      "loss": 0.0039,
      "step": 48430
    },
    {
      "epoch": 0.792604107011372,
      "grad_norm": 0.24571382999420166,
      "learning_rate": 9.20094922003509e-06,
      "loss": 0.0031,
      "step": 48440
    },
    {
      "epoch": 0.7927677329624478,
      "grad_norm": 0.11097700893878937,
      "learning_rate": 9.200432914048647e-06,
      "loss": 0.0019,
      "step": 48450
    },
    {
      "epoch": 0.7929313589135237,
      "grad_norm": 0.1446879804134369,
      "learning_rate": 9.199916455806606e-06,
      "loss": 0.0021,
      "step": 48460
    },
    {
      "epoch": 0.7930949848645995,
      "grad_norm": 0.12244662642478943,
      "learning_rate": 9.19939984532769e-06,
      "loss": 0.0034,
      "step": 48470
    },
    {
      "epoch": 0.7932586108156754,
      "grad_norm": 0.3136637806892395,
      "learning_rate": 9.198883082630621e-06,
      "loss": 0.0023,
      "step": 48480
    },
    {
      "epoch": 0.7934222367667512,
      "grad_norm": 0.060511503368616104,
      "learning_rate": 9.198366167734133e-06,
      "loss": 0.0039,
      "step": 48490
    },
    {
      "epoch": 0.793585862717827,
      "grad_norm": 0.2677363455295563,
      "learning_rate": 9.197849100656961e-06,
      "loss": 0.004,
      "step": 48500
    },
    {
      "epoch": 0.7937494886689029,
      "grad_norm": 0.17779919505119324,
      "learning_rate": 9.197331881417847e-06,
      "loss": 0.0037,
      "step": 48510
    },
    {
      "epoch": 0.7939131146199787,
      "grad_norm": 0.32646405696868896,
      "learning_rate": 9.196814510035542e-06,
      "loss": 0.0076,
      "step": 48520
    },
    {
      "epoch": 0.7940767405710546,
      "grad_norm": 0.18580962717533112,
      "learning_rate": 9.1962969865288e-06,
      "loss": 0.0019,
      "step": 48530
    },
    {
      "epoch": 0.7942403665221304,
      "grad_norm": 0.2901622951030731,
      "learning_rate": 9.195779310916374e-06,
      "loss": 0.0037,
      "step": 48540
    },
    {
      "epoch": 0.7944039924732063,
      "grad_norm": 0.313329815864563,
      "learning_rate": 9.195261483217034e-06,
      "loss": 0.0047,
      "step": 48550
    },
    {
      "epoch": 0.7945676184242821,
      "grad_norm": 0.1565229892730713,
      "learning_rate": 9.194743503449551e-06,
      "loss": 0.0045,
      "step": 48560
    },
    {
      "epoch": 0.7947312443753579,
      "grad_norm": 0.12905044853687286,
      "learning_rate": 9.194225371632693e-06,
      "loss": 0.004,
      "step": 48570
    },
    {
      "epoch": 0.7948948703264338,
      "grad_norm": 0.253761351108551,
      "learning_rate": 9.193707087785248e-06,
      "loss": 0.0033,
      "step": 48580
    },
    {
      "epoch": 0.7950584962775096,
      "grad_norm": 0.2196241021156311,
      "learning_rate": 9.193188651926e-06,
      "loss": 0.003,
      "step": 48590
    },
    {
      "epoch": 0.7952221222285855,
      "grad_norm": 0.21832162141799927,
      "learning_rate": 9.192670064073743e-06,
      "loss": 0.0026,
      "step": 48600
    },
    {
      "epoch": 0.7953857481796613,
      "grad_norm": 0.012147759087383747,
      "learning_rate": 9.19215132424727e-06,
      "loss": 0.0021,
      "step": 48610
    },
    {
      "epoch": 0.7955493741307371,
      "grad_norm": 0.08017176389694214,
      "learning_rate": 9.19163243246539e-06,
      "loss": 0.0029,
      "step": 48620
    },
    {
      "epoch": 0.795713000081813,
      "grad_norm": 0.3405068516731262,
      "learning_rate": 9.191113388746908e-06,
      "loss": 0.0031,
      "step": 48630
    },
    {
      "epoch": 0.7958766260328888,
      "grad_norm": 0.3401029407978058,
      "learning_rate": 9.190594193110639e-06,
      "loss": 0.0029,
      "step": 48640
    },
    {
      "epoch": 0.7960402519839647,
      "grad_norm": 0.21912595629692078,
      "learning_rate": 9.1900748455754e-06,
      "loss": 0.0035,
      "step": 48650
    },
    {
      "epoch": 0.7962038779350405,
      "grad_norm": 0.12028355896472931,
      "learning_rate": 9.18955534616002e-06,
      "loss": 0.0043,
      "step": 48660
    },
    {
      "epoch": 0.7963675038861163,
      "grad_norm": 0.1338496059179306,
      "learning_rate": 9.189035694883326e-06,
      "loss": 0.0032,
      "step": 48670
    },
    {
      "epoch": 0.7965311298371922,
      "grad_norm": 0.013301242142915726,
      "learning_rate": 9.188515891764159e-06,
      "loss": 0.0047,
      "step": 48680
    },
    {
      "epoch": 0.796694755788268,
      "grad_norm": 0.09481485188007355,
      "learning_rate": 9.187995936821357e-06,
      "loss": 0.0033,
      "step": 48690
    },
    {
      "epoch": 0.7968583817393439,
      "grad_norm": 0.4685578942298889,
      "learning_rate": 9.187475830073768e-06,
      "loss": 0.0068,
      "step": 48700
    },
    {
      "epoch": 0.7970220076904198,
      "grad_norm": 0.5887050628662109,
      "learning_rate": 9.186955571540244e-06,
      "loss": 0.0087,
      "step": 48710
    },
    {
      "epoch": 0.7971856336414955,
      "grad_norm": 0.17826080322265625,
      "learning_rate": 9.186435161239645e-06,
      "loss": 0.0038,
      "step": 48720
    },
    {
      "epoch": 0.7973492595925714,
      "grad_norm": 0.17676234245300293,
      "learning_rate": 9.18591459919083e-06,
      "loss": 0.0034,
      "step": 48730
    },
    {
      "epoch": 0.7975128855436472,
      "grad_norm": 0.14908593893051147,
      "learning_rate": 9.185393885412675e-06,
      "loss": 0.004,
      "step": 48740
    },
    {
      "epoch": 0.7976765114947231,
      "grad_norm": 0.16212302446365356,
      "learning_rate": 9.184873019924048e-06,
      "loss": 0.0033,
      "step": 48750
    },
    {
      "epoch": 0.797840137445799,
      "grad_norm": 0.19955624639987946,
      "learning_rate": 9.184352002743835e-06,
      "loss": 0.0022,
      "step": 48760
    },
    {
      "epoch": 0.7980037633968747,
      "grad_norm": 0.08536086231470108,
      "learning_rate": 9.183830833890916e-06,
      "loss": 0.0025,
      "step": 48770
    },
    {
      "epoch": 0.7981673893479506,
      "grad_norm": 0.16760310530662537,
      "learning_rate": 9.183309513384186e-06,
      "loss": 0.0046,
      "step": 48780
    },
    {
      "epoch": 0.7983310152990264,
      "grad_norm": 0.27276304364204407,
      "learning_rate": 9.18278804124254e-06,
      "loss": 0.0037,
      "step": 48790
    },
    {
      "epoch": 0.7984946412501023,
      "grad_norm": 0.07501021027565002,
      "learning_rate": 9.182266417484881e-06,
      "loss": 0.003,
      "step": 48800
    },
    {
      "epoch": 0.798658267201178,
      "grad_norm": 0.1924322098493576,
      "learning_rate": 9.181744642130117e-06,
      "loss": 0.0029,
      "step": 48810
    },
    {
      "epoch": 0.7988218931522539,
      "grad_norm": 0.1344781070947647,
      "learning_rate": 9.18122271519716e-06,
      "loss": 0.0018,
      "step": 48820
    },
    {
      "epoch": 0.7989855191033298,
      "grad_norm": 0.1316998302936554,
      "learning_rate": 9.180700636704927e-06,
      "loss": 0.006,
      "step": 48830
    },
    {
      "epoch": 0.7991491450544056,
      "grad_norm": 0.23098084330558777,
      "learning_rate": 9.180178406672346e-06,
      "loss": 0.0048,
      "step": 48840
    },
    {
      "epoch": 0.7993127710054815,
      "grad_norm": 0.22029943764209747,
      "learning_rate": 9.179656025118344e-06,
      "loss": 0.0025,
      "step": 48850
    },
    {
      "epoch": 0.7994763969565573,
      "grad_norm": 0.2143545001745224,
      "learning_rate": 9.179133492061856e-06,
      "loss": 0.0035,
      "step": 48860
    },
    {
      "epoch": 0.7996400229076331,
      "grad_norm": 0.131936177611351,
      "learning_rate": 9.178610807521824e-06,
      "loss": 0.003,
      "step": 48870
    },
    {
      "epoch": 0.799803648858709,
      "grad_norm": 0.10593626648187637,
      "learning_rate": 9.178087971517194e-06,
      "loss": 0.0035,
      "step": 48880
    },
    {
      "epoch": 0.7999672748097848,
      "grad_norm": 0.13795939087867737,
      "learning_rate": 9.177564984066915e-06,
      "loss": 0.0034,
      "step": 48890
    },
    {
      "epoch": 0.8001309007608607,
      "grad_norm": 0.08845298737287521,
      "learning_rate": 9.177041845189947e-06,
      "loss": 0.0037,
      "step": 48900
    },
    {
      "epoch": 0.8002945267119365,
      "grad_norm": 0.2725151479244232,
      "learning_rate": 9.17651855490525e-06,
      "loss": 0.0033,
      "step": 48910
    },
    {
      "epoch": 0.8004581526630123,
      "grad_norm": 0.07664602994918823,
      "learning_rate": 9.175995113231794e-06,
      "loss": 0.0039,
      "step": 48920
    },
    {
      "epoch": 0.8006217786140882,
      "grad_norm": 0.12458325922489166,
      "learning_rate": 9.175471520188552e-06,
      "loss": 0.0026,
      "step": 48930
    },
    {
      "epoch": 0.800785404565164,
      "grad_norm": 0.1353135108947754,
      "learning_rate": 9.174947775794504e-06,
      "loss": 0.0036,
      "step": 48940
    },
    {
      "epoch": 0.8009490305162399,
      "grad_norm": 0.06275133043527603,
      "learning_rate": 9.174423880068633e-06,
      "loss": 0.0017,
      "step": 48950
    },
    {
      "epoch": 0.8011126564673157,
      "grad_norm": 0.12637528777122498,
      "learning_rate": 9.173899833029928e-06,
      "loss": 0.0036,
      "step": 48960
    },
    {
      "epoch": 0.8012762824183915,
      "grad_norm": 0.34527072310447693,
      "learning_rate": 9.173375634697387e-06,
      "loss": 0.0052,
      "step": 48970
    },
    {
      "epoch": 0.8014399083694674,
      "grad_norm": 0.10525242239236832,
      "learning_rate": 9.17285128509001e-06,
      "loss": 0.0021,
      "step": 48980
    },
    {
      "epoch": 0.8016035343205432,
      "grad_norm": 0.031754203140735626,
      "learning_rate": 9.172326784226802e-06,
      "loss": 0.0025,
      "step": 48990
    },
    {
      "epoch": 0.8017671602716191,
      "grad_norm": 0.2534172534942627,
      "learning_rate": 9.171802132126777e-06,
      "loss": 0.0031,
      "step": 49000
    },
    {
      "epoch": 0.8019307862226949,
      "grad_norm": 0.03221935033798218,
      "learning_rate": 9.17127732880895e-06,
      "loss": 0.0048,
      "step": 49010
    },
    {
      "epoch": 0.8020944121737708,
      "grad_norm": 0.159492626786232,
      "learning_rate": 9.170752374292347e-06,
      "loss": 0.0031,
      "step": 49020
    },
    {
      "epoch": 0.8022580381248466,
      "grad_norm": 0.1019534170627594,
      "learning_rate": 9.170227268595992e-06,
      "loss": 0.0048,
      "step": 49030
    },
    {
      "epoch": 0.8024216640759224,
      "grad_norm": 0.5864030718803406,
      "learning_rate": 9.169702011738923e-06,
      "loss": 0.0044,
      "step": 49040
    },
    {
      "epoch": 0.8025852900269983,
      "grad_norm": 0.12144570797681808,
      "learning_rate": 9.169176603740178e-06,
      "loss": 0.0038,
      "step": 49050
    },
    {
      "epoch": 0.8027489159780741,
      "grad_norm": 0.14907336235046387,
      "learning_rate": 9.168651044618802e-06,
      "loss": 0.0035,
      "step": 49060
    },
    {
      "epoch": 0.80291254192915,
      "grad_norm": 0.24622511863708496,
      "learning_rate": 9.168125334393843e-06,
      "loss": 0.0027,
      "step": 49070
    },
    {
      "epoch": 0.8030761678802258,
      "grad_norm": 0.3472890853881836,
      "learning_rate": 9.167599473084358e-06,
      "loss": 0.0032,
      "step": 49080
    },
    {
      "epoch": 0.8032397938313016,
      "grad_norm": 0.49071112275123596,
      "learning_rate": 9.167073460709411e-06,
      "loss": 0.0023,
      "step": 49090
    },
    {
      "epoch": 0.8034034197823775,
      "grad_norm": 0.12722264230251312,
      "learning_rate": 9.166547297288064e-06,
      "loss": 0.004,
      "step": 49100
    },
    {
      "epoch": 0.8035670457334533,
      "grad_norm": 0.5407263040542603,
      "learning_rate": 9.166020982839394e-06,
      "loss": 0.0025,
      "step": 49110
    },
    {
      "epoch": 0.8037306716845292,
      "grad_norm": 0.11173861473798752,
      "learning_rate": 9.165494517382474e-06,
      "loss": 0.0028,
      "step": 49120
    },
    {
      "epoch": 0.803894297635605,
      "grad_norm": 0.14655622839927673,
      "learning_rate": 9.164967900936392e-06,
      "loss": 0.0049,
      "step": 49130
    },
    {
      "epoch": 0.8040579235866808,
      "grad_norm": 0.24831844866275787,
      "learning_rate": 9.16444113352023e-06,
      "loss": 0.0018,
      "step": 49140
    },
    {
      "epoch": 0.8042215495377567,
      "grad_norm": 0.03322440758347511,
      "learning_rate": 9.16391421515309e-06,
      "loss": 0.0036,
      "step": 49150
    },
    {
      "epoch": 0.8043851754888325,
      "grad_norm": 0.1715569645166397,
      "learning_rate": 9.163387145854063e-06,
      "loss": 0.0038,
      "step": 49160
    },
    {
      "epoch": 0.8045488014399084,
      "grad_norm": 0.1208120658993721,
      "learning_rate": 9.162859925642263e-06,
      "loss": 0.0062,
      "step": 49170
    },
    {
      "epoch": 0.8047124273909843,
      "grad_norm": 0.18374855816364288,
      "learning_rate": 9.162332554536793e-06,
      "loss": 0.0026,
      "step": 49180
    },
    {
      "epoch": 0.80487605334206,
      "grad_norm": 0.10204806178808212,
      "learning_rate": 9.161805032556771e-06,
      "loss": 0.0056,
      "step": 49190
    },
    {
      "epoch": 0.8050396792931359,
      "grad_norm": 0.5097343325614929,
      "learning_rate": 9.161277359721322e-06,
      "loss": 0.0029,
      "step": 49200
    },
    {
      "epoch": 0.8052033052442117,
      "grad_norm": 0.21722246706485748,
      "learning_rate": 9.160749536049567e-06,
      "loss": 0.0025,
      "step": 49210
    },
    {
      "epoch": 0.8053669311952876,
      "grad_norm": 0.20632918179035187,
      "learning_rate": 9.160221561560643e-06,
      "loss": 0.0033,
      "step": 49220
    },
    {
      "epoch": 0.8055305571463635,
      "grad_norm": 0.2717243731021881,
      "learning_rate": 9.159693436273687e-06,
      "loss": 0.0034,
      "step": 49230
    },
    {
      "epoch": 0.8056941830974392,
      "grad_norm": 0.19756168127059937,
      "learning_rate": 9.159165160207842e-06,
      "loss": 0.0032,
      "step": 49240
    },
    {
      "epoch": 0.8058578090485151,
      "grad_norm": 0.4554564952850342,
      "learning_rate": 9.158636733382253e-06,
      "loss": 0.0056,
      "step": 49250
    },
    {
      "epoch": 0.8060214349995909,
      "grad_norm": 0.05653040483593941,
      "learning_rate": 9.15810815581608e-06,
      "loss": 0.0037,
      "step": 49260
    },
    {
      "epoch": 0.8061850609506668,
      "grad_norm": 0.08855530619621277,
      "learning_rate": 9.15757942752848e-06,
      "loss": 0.0016,
      "step": 49270
    },
    {
      "epoch": 0.8063486869017427,
      "grad_norm": 0.04314913600683212,
      "learning_rate": 9.157050548538617e-06,
      "loss": 0.0022,
      "step": 49280
    },
    {
      "epoch": 0.8065123128528184,
      "grad_norm": 0.18650372326374054,
      "learning_rate": 9.156521518865665e-06,
      "loss": 0.0027,
      "step": 49290
    },
    {
      "epoch": 0.8066759388038943,
      "grad_norm": 0.17142866551876068,
      "learning_rate": 9.155992338528796e-06,
      "loss": 0.0033,
      "step": 49300
    },
    {
      "epoch": 0.8068395647549701,
      "grad_norm": 0.3487672805786133,
      "learning_rate": 9.155463007547193e-06,
      "loss": 0.0066,
      "step": 49310
    },
    {
      "epoch": 0.807003190706046,
      "grad_norm": 0.045737870037555695,
      "learning_rate": 9.154933525940045e-06,
      "loss": 0.0031,
      "step": 49320
    },
    {
      "epoch": 0.8071668166571219,
      "grad_norm": 0.3658466339111328,
      "learning_rate": 9.154403893726541e-06,
      "loss": 0.0021,
      "step": 49330
    },
    {
      "epoch": 0.8073304426081976,
      "grad_norm": 0.08573342114686966,
      "learning_rate": 9.153874110925882e-06,
      "loss": 0.0022,
      "step": 49340
    },
    {
      "epoch": 0.8074940685592735,
      "grad_norm": 0.1402796506881714,
      "learning_rate": 9.15334417755727e-06,
      "loss": 0.0029,
      "step": 49350
    },
    {
      "epoch": 0.8076576945103493,
      "grad_norm": 0.3999117612838745,
      "learning_rate": 9.152814093639914e-06,
      "loss": 0.0039,
      "step": 49360
    },
    {
      "epoch": 0.8078213204614252,
      "grad_norm": 0.09455102682113647,
      "learning_rate": 9.152283859193027e-06,
      "loss": 0.0022,
      "step": 49370
    },
    {
      "epoch": 0.8079849464125011,
      "grad_norm": 0.008731618523597717,
      "learning_rate": 9.151753474235832e-06,
      "loss": 0.0021,
      "step": 49380
    },
    {
      "epoch": 0.8081485723635768,
      "grad_norm": 0.1310076117515564,
      "learning_rate": 9.151222938787549e-06,
      "loss": 0.003,
      "step": 49390
    },
    {
      "epoch": 0.8083121983146527,
      "grad_norm": 0.062411386519670486,
      "learning_rate": 9.150692252867414e-06,
      "loss": 0.0034,
      "step": 49400
    },
    {
      "epoch": 0.8084758242657285,
      "grad_norm": 0.36796289682388306,
      "learning_rate": 9.150161416494661e-06,
      "loss": 0.004,
      "step": 49410
    },
    {
      "epoch": 0.8086394502168044,
      "grad_norm": 0.24233144521713257,
      "learning_rate": 9.14963042968853e-06,
      "loss": 0.0029,
      "step": 49420
    },
    {
      "epoch": 0.8088030761678803,
      "grad_norm": 0.15177123248577118,
      "learning_rate": 9.149099292468268e-06,
      "loss": 0.0039,
      "step": 49430
    },
    {
      "epoch": 0.808966702118956,
      "grad_norm": 0.08123491704463959,
      "learning_rate": 9.148568004853131e-06,
      "loss": 0.0016,
      "step": 49440
    },
    {
      "epoch": 0.8091303280700319,
      "grad_norm": 0.2115696370601654,
      "learning_rate": 9.148036566862375e-06,
      "loss": 0.0035,
      "step": 49450
    },
    {
      "epoch": 0.8092939540211077,
      "grad_norm": 0.25085505843162537,
      "learning_rate": 9.147504978515262e-06,
      "loss": 0.0031,
      "step": 49460
    },
    {
      "epoch": 0.8094575799721836,
      "grad_norm": 0.12272707372903824,
      "learning_rate": 9.14697323983106e-06,
      "loss": 0.0023,
      "step": 49470
    },
    {
      "epoch": 0.8096212059232595,
      "grad_norm": 0.12907728552818298,
      "learning_rate": 9.146441350829048e-06,
      "loss": 0.0024,
      "step": 49480
    },
    {
      "epoch": 0.8097848318743353,
      "grad_norm": 0.08441904187202454,
      "learning_rate": 9.145909311528502e-06,
      "loss": 0.0023,
      "step": 49490
    },
    {
      "epoch": 0.8099484578254111,
      "grad_norm": 0.10625497996807098,
      "learning_rate": 9.145377121948707e-06,
      "loss": 0.0045,
      "step": 49500
    },
    {
      "epoch": 0.8101120837764869,
      "grad_norm": 0.12744413316249847,
      "learning_rate": 9.144844782108954e-06,
      "loss": 0.0033,
      "step": 49510
    },
    {
      "epoch": 0.8102757097275628,
      "grad_norm": 0.1864137500524521,
      "learning_rate": 9.144312292028542e-06,
      "loss": 0.0026,
      "step": 49520
    },
    {
      "epoch": 0.8104393356786387,
      "grad_norm": 0.07206683605909348,
      "learning_rate": 9.143779651726768e-06,
      "loss": 0.0044,
      "step": 49530
    },
    {
      "epoch": 0.8106029616297145,
      "grad_norm": 0.2719031572341919,
      "learning_rate": 9.143246861222941e-06,
      "loss": 0.0049,
      "step": 49540
    },
    {
      "epoch": 0.8107665875807903,
      "grad_norm": 0.1794460564851761,
      "learning_rate": 9.142713920536375e-06,
      "loss": 0.003,
      "step": 49550
    },
    {
      "epoch": 0.8109302135318661,
      "grad_norm": 0.0942332074046135,
      "learning_rate": 9.142180829686384e-06,
      "loss": 0.0029,
      "step": 49560
    },
    {
      "epoch": 0.811093839482942,
      "grad_norm": 0.09986443817615509,
      "learning_rate": 9.141647588692295e-06,
      "loss": 0.0031,
      "step": 49570
    },
    {
      "epoch": 0.8112574654340179,
      "grad_norm": 0.20012661814689636,
      "learning_rate": 9.141114197573435e-06,
      "loss": 0.0029,
      "step": 49580
    },
    {
      "epoch": 0.8114210913850937,
      "grad_norm": 0.12359360605478287,
      "learning_rate": 9.140580656349138e-06,
      "loss": 0.0044,
      "step": 49590
    },
    {
      "epoch": 0.8115847173361695,
      "grad_norm": 0.059907495975494385,
      "learning_rate": 9.140046965038743e-06,
      "loss": 0.0018,
      "step": 49600
    },
    {
      "epoch": 0.8117483432872453,
      "grad_norm": 0.05025084689259529,
      "learning_rate": 9.139513123661596e-06,
      "loss": 0.0034,
      "step": 49610
    },
    {
      "epoch": 0.8119119692383212,
      "grad_norm": 0.0567028634250164,
      "learning_rate": 9.138979132237047e-06,
      "loss": 0.0042,
      "step": 49620
    },
    {
      "epoch": 0.8120755951893971,
      "grad_norm": 0.04546193405985832,
      "learning_rate": 9.138444990784455e-06,
      "loss": 0.0028,
      "step": 49630
    },
    {
      "epoch": 0.8122392211404729,
      "grad_norm": 0.05458003655076027,
      "learning_rate": 9.137910699323175e-06,
      "loss": 0.0041,
      "step": 49640
    },
    {
      "epoch": 0.8124028470915488,
      "grad_norm": 0.0964931845664978,
      "learning_rate": 9.137376257872578e-06,
      "loss": 0.0026,
      "step": 49650
    },
    {
      "epoch": 0.8125664730426245,
      "grad_norm": 0.058308687061071396,
      "learning_rate": 9.136841666452037e-06,
      "loss": 0.0041,
      "step": 49660
    },
    {
      "epoch": 0.8127300989937004,
      "grad_norm": 0.06652742624282837,
      "learning_rate": 9.136306925080929e-06,
      "loss": 0.0025,
      "step": 49670
    },
    {
      "epoch": 0.8128937249447763,
      "grad_norm": 0.1931263655424118,
      "learning_rate": 9.135772033778634e-06,
      "loss": 0.003,
      "step": 49680
    },
    {
      "epoch": 0.8130573508958521,
      "grad_norm": 0.18454749882221222,
      "learning_rate": 9.135236992564545e-06,
      "loss": 0.0017,
      "step": 49690
    },
    {
      "epoch": 0.813220976846928,
      "grad_norm": 0.16015540063381195,
      "learning_rate": 9.134701801458052e-06,
      "loss": 0.0027,
      "step": 49700
    },
    {
      "epoch": 0.8133846027980037,
      "grad_norm": 0.07232258468866348,
      "learning_rate": 9.134166460478555e-06,
      "loss": 0.0031,
      "step": 49710
    },
    {
      "epoch": 0.8135482287490796,
      "grad_norm": 0.1756167709827423,
      "learning_rate": 9.133630969645463e-06,
      "loss": 0.0028,
      "step": 49720
    },
    {
      "epoch": 0.8137118547001554,
      "grad_norm": 0.19829516112804413,
      "learning_rate": 9.133095328978181e-06,
      "loss": 0.0047,
      "step": 49730
    },
    {
      "epoch": 0.8138754806512313,
      "grad_norm": 0.1780901700258255,
      "learning_rate": 9.132559538496128e-06,
      "loss": 0.0029,
      "step": 49740
    },
    {
      "epoch": 0.8140391066023072,
      "grad_norm": 0.13632865250110626,
      "learning_rate": 9.132023598218724e-06,
      "loss": 0.0024,
      "step": 49750
    },
    {
      "epoch": 0.8142027325533829,
      "grad_norm": 0.2727030813694,
      "learning_rate": 9.131487508165397e-06,
      "loss": 0.0044,
      "step": 49760
    },
    {
      "epoch": 0.8143663585044588,
      "grad_norm": 0.14128953218460083,
      "learning_rate": 9.130951268355576e-06,
      "loss": 0.0017,
      "step": 49770
    },
    {
      "epoch": 0.8145299844555346,
      "grad_norm": 0.1939658522605896,
      "learning_rate": 9.1304148788087e-06,
      "loss": 0.0044,
      "step": 49780
    },
    {
      "epoch": 0.8146936104066105,
      "grad_norm": 0.09017079323530197,
      "learning_rate": 9.129878339544213e-06,
      "loss": 0.0043,
      "step": 49790
    },
    {
      "epoch": 0.8148572363576864,
      "grad_norm": 0.14885945618152618,
      "learning_rate": 9.12934165058156e-06,
      "loss": 0.0023,
      "step": 49800
    },
    {
      "epoch": 0.8150208623087621,
      "grad_norm": 0.008533250540494919,
      "learning_rate": 9.1288048119402e-06,
      "loss": 0.0026,
      "step": 49810
    },
    {
      "epoch": 0.815184488259838,
      "grad_norm": 0.12529003620147705,
      "learning_rate": 9.128267823639585e-06,
      "loss": 0.0048,
      "step": 49820
    },
    {
      "epoch": 0.8153481142109138,
      "grad_norm": 0.09427347779273987,
      "learning_rate": 9.127730685699186e-06,
      "loss": 0.0031,
      "step": 49830
    },
    {
      "epoch": 0.8155117401619897,
      "grad_norm": 0.06731635332107544,
      "learning_rate": 9.12719339813847e-06,
      "loss": 0.002,
      "step": 49840
    },
    {
      "epoch": 0.8156753661130656,
      "grad_norm": 0.12200576066970825,
      "learning_rate": 9.126655960976912e-06,
      "loss": 0.0029,
      "step": 49850
    },
    {
      "epoch": 0.8158389920641413,
      "grad_norm": 0.19231025874614716,
      "learning_rate": 9.126118374233994e-06,
      "loss": 0.0081,
      "step": 49860
    },
    {
      "epoch": 0.8160026180152172,
      "grad_norm": 0.17219844460487366,
      "learning_rate": 9.125580637929203e-06,
      "loss": 0.0028,
      "step": 49870
    },
    {
      "epoch": 0.816166243966293,
      "grad_norm": 0.10270903259515762,
      "learning_rate": 9.125042752082027e-06,
      "loss": 0.0037,
      "step": 49880
    },
    {
      "epoch": 0.8163298699173689,
      "grad_norm": 0.06245635077357292,
      "learning_rate": 9.124504716711968e-06,
      "loss": 0.0036,
      "step": 49890
    },
    {
      "epoch": 0.8164934958684448,
      "grad_norm": 0.3058038055896759,
      "learning_rate": 9.123966531838525e-06,
      "loss": 0.0043,
      "step": 49900
    },
    {
      "epoch": 0.8166571218195205,
      "grad_norm": 0.0785287469625473,
      "learning_rate": 9.123428197481205e-06,
      "loss": 0.0046,
      "step": 49910
    },
    {
      "epoch": 0.8168207477705964,
      "grad_norm": 0.0849079117178917,
      "learning_rate": 9.122889713659526e-06,
      "loss": 0.0032,
      "step": 49920
    },
    {
      "epoch": 0.8169843737216722,
      "grad_norm": 0.0672604963183403,
      "learning_rate": 9.122351080393003e-06,
      "loss": 0.0024,
      "step": 49930
    },
    {
      "epoch": 0.8171479996727481,
      "grad_norm": 0.2374507188796997,
      "learning_rate": 9.121812297701162e-06,
      "loss": 0.0023,
      "step": 49940
    },
    {
      "epoch": 0.817311625623824,
      "grad_norm": 0.06891575455665588,
      "learning_rate": 9.12127336560353e-06,
      "loss": 0.0039,
      "step": 49950
    },
    {
      "epoch": 0.8174752515748998,
      "grad_norm": 0.14454405009746552,
      "learning_rate": 9.120734284119645e-06,
      "loss": 0.003,
      "step": 49960
    },
    {
      "epoch": 0.8176388775259756,
      "grad_norm": 0.052260179072618484,
      "learning_rate": 9.120195053269046e-06,
      "loss": 0.0026,
      "step": 49970
    },
    {
      "epoch": 0.8178025034770514,
      "grad_norm": 0.17448776960372925,
      "learning_rate": 9.119655673071278e-06,
      "loss": 0.0024,
      "step": 49980
    },
    {
      "epoch": 0.8179661294281273,
      "grad_norm": 0.21683809161186218,
      "learning_rate": 9.119116143545894e-06,
      "loss": 0.0037,
      "step": 49990
    },
    {
      "epoch": 0.8181297553792032,
      "grad_norm": 0.25295209884643555,
      "learning_rate": 9.11857646471245e-06,
      "loss": 0.0026,
      "step": 50000
    },
    {
      "epoch": 0.8181297553792032,
      "eval_loss": 0.002267897129058838,
      "eval_runtime": 3.0955,
      "eval_samples_per_second": 64.609,
      "eval_steps_per_second": 16.152,
      "step": 50000
    },
    {
      "epoch": 0.818293381330279,
      "grad_norm": 0.1907290369272232,
      "learning_rate": 9.118036636590507e-06,
      "loss": 0.0039,
      "step": 50010
    },
    {
      "epoch": 0.8184570072813548,
      "grad_norm": 0.08578497916460037,
      "learning_rate": 9.117496659199633e-06,
      "loss": 0.0034,
      "step": 50020
    },
    {
      "epoch": 0.8186206332324306,
      "grad_norm": 0.4064314067363739,
      "learning_rate": 9.116956532559403e-06,
      "loss": 0.004,
      "step": 50030
    },
    {
      "epoch": 0.8187842591835065,
      "grad_norm": 0.17687946557998657,
      "learning_rate": 9.116416256689393e-06,
      "loss": 0.0039,
      "step": 50040
    },
    {
      "epoch": 0.8189478851345824,
      "grad_norm": 0.22434987127780914,
      "learning_rate": 9.115875831609187e-06,
      "loss": 0.0037,
      "step": 50050
    },
    {
      "epoch": 0.8191115110856582,
      "grad_norm": 0.029149552807211876,
      "learning_rate": 9.115335257338374e-06,
      "loss": 0.0034,
      "step": 50060
    },
    {
      "epoch": 0.819275137036734,
      "grad_norm": 0.09738941490650177,
      "learning_rate": 9.114794533896549e-06,
      "loss": 0.003,
      "step": 50070
    },
    {
      "epoch": 0.8194387629878098,
      "grad_norm": 0.08028104901313782,
      "learning_rate": 9.114253661303311e-06,
      "loss": 0.0027,
      "step": 50080
    },
    {
      "epoch": 0.8196023889388857,
      "grad_norm": 0.13614508509635925,
      "learning_rate": 9.113712639578268e-06,
      "loss": 0.0035,
      "step": 50090
    },
    {
      "epoch": 0.8197660148899616,
      "grad_norm": 0.3525117337703705,
      "learning_rate": 9.113171468741028e-06,
      "loss": 0.0042,
      "step": 50100
    },
    {
      "epoch": 0.8199296408410374,
      "grad_norm": 0.1918036788702011,
      "learning_rate": 9.112630148811207e-06,
      "loss": 0.0038,
      "step": 50110
    },
    {
      "epoch": 0.8200932667921133,
      "grad_norm": 0.13757005333900452,
      "learning_rate": 9.112088679808429e-06,
      "loss": 0.0022,
      "step": 50120
    },
    {
      "epoch": 0.820256892743189,
      "grad_norm": 0.21057748794555664,
      "learning_rate": 9.11154706175232e-06,
      "loss": 0.0046,
      "step": 50130
    },
    {
      "epoch": 0.8204205186942649,
      "grad_norm": 0.2428678423166275,
      "learning_rate": 9.11100529466251e-06,
      "loss": 0.0038,
      "step": 50140
    },
    {
      "epoch": 0.8205841446453408,
      "grad_norm": 0.09326040744781494,
      "learning_rate": 9.11046337855864e-06,
      "loss": 0.0029,
      "step": 50150
    },
    {
      "epoch": 0.8207477705964166,
      "grad_norm": 0.12856489419937134,
      "learning_rate": 9.109921313460351e-06,
      "loss": 0.0032,
      "step": 50160
    },
    {
      "epoch": 0.8209113965474925,
      "grad_norm": 0.06457704305648804,
      "learning_rate": 9.109379099387292e-06,
      "loss": 0.0027,
      "step": 50170
    },
    {
      "epoch": 0.8210750224985682,
      "grad_norm": 0.10474004596471786,
      "learning_rate": 9.10883673635912e-06,
      "loss": 0.0033,
      "step": 50180
    },
    {
      "epoch": 0.8212386484496441,
      "grad_norm": 0.20883670449256897,
      "learning_rate": 9.108294224395488e-06,
      "loss": 0.0025,
      "step": 50190
    },
    {
      "epoch": 0.82140227440072,
      "grad_norm": 0.1303151696920395,
      "learning_rate": 9.107751563516065e-06,
      "loss": 0.004,
      "step": 50200
    },
    {
      "epoch": 0.8215659003517958,
      "grad_norm": 0.19828058779239655,
      "learning_rate": 9.107208753740522e-06,
      "loss": 0.0029,
      "step": 50210
    },
    {
      "epoch": 0.8217295263028717,
      "grad_norm": 0.4396517276763916,
      "learning_rate": 9.106665795088533e-06,
      "loss": 0.0033,
      "step": 50220
    },
    {
      "epoch": 0.8218931522539474,
      "grad_norm": 0.1698131412267685,
      "learning_rate": 9.106122687579779e-06,
      "loss": 0.0027,
      "step": 50230
    },
    {
      "epoch": 0.8220567782050233,
      "grad_norm": 0.03647887706756592,
      "learning_rate": 9.105579431233946e-06,
      "loss": 0.0027,
      "step": 50240
    },
    {
      "epoch": 0.8222204041560992,
      "grad_norm": 0.2587306797504425,
      "learning_rate": 9.105036026070726e-06,
      "loss": 0.0035,
      "step": 50250
    },
    {
      "epoch": 0.822384030107175,
      "grad_norm": 0.1795492321252823,
      "learning_rate": 9.104492472109816e-06,
      "loss": 0.0024,
      "step": 50260
    },
    {
      "epoch": 0.8225476560582509,
      "grad_norm": 0.0518924742937088,
      "learning_rate": 9.103948769370918e-06,
      "loss": 0.0027,
      "step": 50270
    },
    {
      "epoch": 0.8227112820093266,
      "grad_norm": 0.11034813523292542,
      "learning_rate": 9.103404917873742e-06,
      "loss": 0.0025,
      "step": 50280
    },
    {
      "epoch": 0.8228749079604025,
      "grad_norm": 0.14446285367012024,
      "learning_rate": 9.102860917638e-06,
      "loss": 0.002,
      "step": 50290
    },
    {
      "epoch": 0.8230385339114784,
      "grad_norm": 0.1639178991317749,
      "learning_rate": 9.10231676868341e-06,
      "loss": 0.0038,
      "step": 50300
    },
    {
      "epoch": 0.8232021598625542,
      "grad_norm": 0.04208364337682724,
      "learning_rate": 9.101772471029697e-06,
      "loss": 0.0025,
      "step": 50310
    },
    {
      "epoch": 0.8233657858136301,
      "grad_norm": 0.10019554197788239,
      "learning_rate": 9.10122802469659e-06,
      "loss": 0.0038,
      "step": 50320
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.16462786495685577,
      "learning_rate": 9.100683429703827e-06,
      "loss": 0.0032,
      "step": 50330
    },
    {
      "epoch": 0.8236930377157817,
      "grad_norm": 0.25822529196739197,
      "learning_rate": 9.100138686071143e-06,
      "loss": 0.0023,
      "step": 50340
    },
    {
      "epoch": 0.8238566636668576,
      "grad_norm": 0.1642756164073944,
      "learning_rate": 9.099593793818285e-06,
      "loss": 0.0064,
      "step": 50350
    },
    {
      "epoch": 0.8240202896179334,
      "grad_norm": 0.19849711656570435,
      "learning_rate": 9.099048752965005e-06,
      "loss": 0.0026,
      "step": 50360
    },
    {
      "epoch": 0.8241839155690093,
      "grad_norm": 0.23884549736976624,
      "learning_rate": 9.098503563531063e-06,
      "loss": 0.0022,
      "step": 50370
    },
    {
      "epoch": 0.824347541520085,
      "grad_norm": 0.12391776591539383,
      "learning_rate": 9.097958225536214e-06,
      "loss": 0.0014,
      "step": 50380
    },
    {
      "epoch": 0.8245111674711609,
      "grad_norm": 0.09041792899370193,
      "learning_rate": 9.09741273900023e-06,
      "loss": 0.0023,
      "step": 50390
    },
    {
      "epoch": 0.8246747934222368,
      "grad_norm": 0.2077135592699051,
      "learning_rate": 9.096867103942881e-06,
      "loss": 0.0034,
      "step": 50400
    },
    {
      "epoch": 0.8248384193733126,
      "grad_norm": 0.3110194504261017,
      "learning_rate": 9.096321320383948e-06,
      "loss": 0.0036,
      "step": 50410
    },
    {
      "epoch": 0.8250020453243885,
      "grad_norm": 0.2612054646015167,
      "learning_rate": 9.09577538834321e-06,
      "loss": 0.0033,
      "step": 50420
    },
    {
      "epoch": 0.8251656712754643,
      "grad_norm": 0.11150077730417252,
      "learning_rate": 9.095229307840459e-06,
      "loss": 0.0029,
      "step": 50430
    },
    {
      "epoch": 0.8253292972265401,
      "grad_norm": 0.17644207179546356,
      "learning_rate": 9.09468307889549e-06,
      "loss": 0.0033,
      "step": 50440
    },
    {
      "epoch": 0.825492923177616,
      "grad_norm": 0.21658693253993988,
      "learning_rate": 9.094136701528098e-06,
      "loss": 0.002,
      "step": 50450
    },
    {
      "epoch": 0.8256565491286918,
      "grad_norm": 0.08151908218860626,
      "learning_rate": 9.093590175758091e-06,
      "loss": 0.002,
      "step": 50460
    },
    {
      "epoch": 0.8258201750797677,
      "grad_norm": 0.10395552963018417,
      "learning_rate": 9.09304350160528e-06,
      "loss": 0.0045,
      "step": 50470
    },
    {
      "epoch": 0.8259838010308435,
      "grad_norm": 0.22743777930736542,
      "learning_rate": 9.092496679089479e-06,
      "loss": 0.0033,
      "step": 50480
    },
    {
      "epoch": 0.8261474269819193,
      "grad_norm": 0.33236101269721985,
      "learning_rate": 9.09194970823051e-06,
      "loss": 0.0033,
      "step": 50490
    },
    {
      "epoch": 0.8263110529329952,
      "grad_norm": 0.11979358643293381,
      "learning_rate": 9.091402589048197e-06,
      "loss": 0.0044,
      "step": 50500
    },
    {
      "epoch": 0.826474678884071,
      "grad_norm": 0.4249751567840576,
      "learning_rate": 9.090855321562375e-06,
      "loss": 0.0043,
      "step": 50510
    },
    {
      "epoch": 0.8266383048351469,
      "grad_norm": 0.04183148592710495,
      "learning_rate": 9.090307905792879e-06,
      "loss": 0.0026,
      "step": 50520
    },
    {
      "epoch": 0.8268019307862227,
      "grad_norm": 0.14201410114765167,
      "learning_rate": 9.089760341759552e-06,
      "loss": 0.0028,
      "step": 50530
    },
    {
      "epoch": 0.8269655567372985,
      "grad_norm": 0.13985739648342133,
      "learning_rate": 9.089212629482244e-06,
      "loss": 0.0031,
      "step": 50540
    },
    {
      "epoch": 0.8271291826883744,
      "grad_norm": 0.054805707186460495,
      "learning_rate": 9.088664768980806e-06,
      "loss": 0.0028,
      "step": 50550
    },
    {
      "epoch": 0.8272928086394502,
      "grad_norm": 0.13985218107700348,
      "learning_rate": 9.088116760275095e-06,
      "loss": 0.0031,
      "step": 50560
    },
    {
      "epoch": 0.8274564345905261,
      "grad_norm": 0.1397520750761032,
      "learning_rate": 9.087568603384977e-06,
      "loss": 0.002,
      "step": 50570
    },
    {
      "epoch": 0.8276200605416019,
      "grad_norm": 0.10762417316436768,
      "learning_rate": 9.087020298330324e-06,
      "loss": 0.002,
      "step": 50580
    },
    {
      "epoch": 0.8277836864926778,
      "grad_norm": 0.10410875082015991,
      "learning_rate": 9.086471845131007e-06,
      "loss": 0.0026,
      "step": 50590
    },
    {
      "epoch": 0.8279473124437535,
      "grad_norm": 0.14096710085868835,
      "learning_rate": 9.085923243806906e-06,
      "loss": 0.002,
      "step": 50600
    },
    {
      "epoch": 0.8281109383948294,
      "grad_norm": 0.12788322567939758,
      "learning_rate": 9.085374494377908e-06,
      "loss": 0.0024,
      "step": 50610
    },
    {
      "epoch": 0.8282745643459053,
      "grad_norm": 0.36692094802856445,
      "learning_rate": 9.084825596863904e-06,
      "loss": 0.0037,
      "step": 50620
    },
    {
      "epoch": 0.8284381902969811,
      "grad_norm": 0.07190009951591492,
      "learning_rate": 9.08427655128479e-06,
      "loss": 0.0026,
      "step": 50630
    },
    {
      "epoch": 0.828601816248057,
      "grad_norm": 0.27206164598464966,
      "learning_rate": 9.083727357660466e-06,
      "loss": 0.0038,
      "step": 50640
    },
    {
      "epoch": 0.8287654421991327,
      "grad_norm": 0.11852709203958511,
      "learning_rate": 9.08317801601084e-06,
      "loss": 0.0032,
      "step": 50650
    },
    {
      "epoch": 0.8289290681502086,
      "grad_norm": 0.3692225515842438,
      "learning_rate": 9.082628526355828e-06,
      "loss": 0.0044,
      "step": 50660
    },
    {
      "epoch": 0.8290926941012845,
      "grad_norm": 0.0759807676076889,
      "learning_rate": 9.08207888871534e-06,
      "loss": 0.0028,
      "step": 50670
    },
    {
      "epoch": 0.8292563200523603,
      "grad_norm": 0.15307533740997314,
      "learning_rate": 9.081529103109306e-06,
      "loss": 0.0028,
      "step": 50680
    },
    {
      "epoch": 0.8294199460034362,
      "grad_norm": 0.3663370907306671,
      "learning_rate": 9.080979169557652e-06,
      "loss": 0.0023,
      "step": 50690
    },
    {
      "epoch": 0.8295835719545119,
      "grad_norm": 0.2570980489253998,
      "learning_rate": 9.08042908808031e-06,
      "loss": 0.0037,
      "step": 50700
    },
    {
      "epoch": 0.8297471979055878,
      "grad_norm": 0.06891855597496033,
      "learning_rate": 9.07987885869722e-06,
      "loss": 0.0028,
      "step": 50710
    },
    {
      "epoch": 0.8299108238566637,
      "grad_norm": 0.1795913577079773,
      "learning_rate": 9.07932848142833e-06,
      "loss": 0.004,
      "step": 50720
    },
    {
      "epoch": 0.8300744498077395,
      "grad_norm": 0.07551542669534683,
      "learning_rate": 9.078777956293584e-06,
      "loss": 0.0018,
      "step": 50730
    },
    {
      "epoch": 0.8302380757588154,
      "grad_norm": 0.16215987503528595,
      "learning_rate": 9.078227283312942e-06,
      "loss": 0.0027,
      "step": 50740
    },
    {
      "epoch": 0.8304017017098911,
      "grad_norm": 0.4029521644115448,
      "learning_rate": 9.077676462506361e-06,
      "loss": 0.0031,
      "step": 50750
    },
    {
      "epoch": 0.830565327660967,
      "grad_norm": 0.043793685734272,
      "learning_rate": 9.07712549389381e-06,
      "loss": 0.0033,
      "step": 50760
    },
    {
      "epoch": 0.8307289536120429,
      "grad_norm": 0.40310558676719666,
      "learning_rate": 9.07657437749526e-06,
      "loss": 0.0031,
      "step": 50770
    },
    {
      "epoch": 0.8308925795631187,
      "grad_norm": 0.18080948293209076,
      "learning_rate": 9.076023113330685e-06,
      "loss": 0.0024,
      "step": 50780
    },
    {
      "epoch": 0.8310562055141946,
      "grad_norm": 0.14649733901023865,
      "learning_rate": 9.07547170142007e-06,
      "loss": 0.0037,
      "step": 50790
    },
    {
      "epoch": 0.8312198314652703,
      "grad_norm": 0.41183507442474365,
      "learning_rate": 9.074920141783398e-06,
      "loss": 0.0048,
      "step": 50800
    },
    {
      "epoch": 0.8313834574163462,
      "grad_norm": 0.025577817112207413,
      "learning_rate": 9.074368434440668e-06,
      "loss": 0.0052,
      "step": 50810
    },
    {
      "epoch": 0.8315470833674221,
      "grad_norm": 0.30468156933784485,
      "learning_rate": 9.073816579411871e-06,
      "loss": 0.0043,
      "step": 50820
    },
    {
      "epoch": 0.8317107093184979,
      "grad_norm": 0.1438124030828476,
      "learning_rate": 9.073264576717019e-06,
      "loss": 0.0018,
      "step": 50830
    },
    {
      "epoch": 0.8318743352695738,
      "grad_norm": 0.18616560101509094,
      "learning_rate": 9.072712426376113e-06,
      "loss": 0.0044,
      "step": 50840
    },
    {
      "epoch": 0.8320379612206495,
      "grad_norm": 0.08166428655385971,
      "learning_rate": 9.072160128409168e-06,
      "loss": 0.0027,
      "step": 50850
    },
    {
      "epoch": 0.8322015871717254,
      "grad_norm": 0.22281168401241302,
      "learning_rate": 9.07160768283621e-06,
      "loss": 0.0026,
      "step": 50860
    },
    {
      "epoch": 0.8323652131228013,
      "grad_norm": 0.16007889807224274,
      "learning_rate": 9.071055089677256e-06,
      "loss": 0.0049,
      "step": 50870
    },
    {
      "epoch": 0.8325288390738771,
      "grad_norm": 0.14258666336536407,
      "learning_rate": 9.07050234895234e-06,
      "loss": 0.0031,
      "step": 50880
    },
    {
      "epoch": 0.832692465024953,
      "grad_norm": 0.23663482069969177,
      "learning_rate": 9.0699494606815e-06,
      "loss": 0.0036,
      "step": 50890
    },
    {
      "epoch": 0.8328560909760288,
      "grad_norm": 0.09731143712997437,
      "learning_rate": 9.069396424884771e-06,
      "loss": 0.0033,
      "step": 50900
    },
    {
      "epoch": 0.8330197169271046,
      "grad_norm": 0.29777204990386963,
      "learning_rate": 9.068843241582204e-06,
      "loss": 0.0047,
      "step": 50910
    },
    {
      "epoch": 0.8331833428781805,
      "grad_norm": 0.12265215069055557,
      "learning_rate": 9.068289910793846e-06,
      "loss": 0.0029,
      "step": 50920
    },
    {
      "epoch": 0.8333469688292563,
      "grad_norm": 0.0338740237057209,
      "learning_rate": 9.06773643253976e-06,
      "loss": 0.0033,
      "step": 50930
    },
    {
      "epoch": 0.8335105947803322,
      "grad_norm": 0.0460546649992466,
      "learning_rate": 9.067182806840003e-06,
      "loss": 0.0036,
      "step": 50940
    },
    {
      "epoch": 0.833674220731408,
      "grad_norm": 0.26102080941200256,
      "learning_rate": 9.066629033714644e-06,
      "loss": 0.0039,
      "step": 50950
    },
    {
      "epoch": 0.8338378466824838,
      "grad_norm": 0.17924408614635468,
      "learning_rate": 9.066075113183757e-06,
      "loss": 0.0034,
      "step": 50960
    },
    {
      "epoch": 0.8340014726335597,
      "grad_norm": 0.14767718315124512,
      "learning_rate": 9.065521045267418e-06,
      "loss": 0.0037,
      "step": 50970
    },
    {
      "epoch": 0.8341650985846355,
      "grad_norm": 0.20270393788814545,
      "learning_rate": 9.064966829985715e-06,
      "loss": 0.0021,
      "step": 50980
    },
    {
      "epoch": 0.8343287245357114,
      "grad_norm": 0.3154531717300415,
      "learning_rate": 9.064412467358733e-06,
      "loss": 0.0047,
      "step": 50990
    },
    {
      "epoch": 0.8344923504867872,
      "grad_norm": 0.20131541788578033,
      "learning_rate": 9.063857957406567e-06,
      "loss": 0.0032,
      "step": 51000
    },
    {
      "epoch": 0.834655976437863,
      "grad_norm": 0.1727403998374939,
      "learning_rate": 9.06330330014932e-06,
      "loss": 0.0037,
      "step": 51010
    },
    {
      "epoch": 0.8348196023889389,
      "grad_norm": 0.4080318212509155,
      "learning_rate": 9.062748495607092e-06,
      "loss": 0.0062,
      "step": 51020
    },
    {
      "epoch": 0.8349832283400147,
      "grad_norm": 0.14040258526802063,
      "learning_rate": 9.062193543799996e-06,
      "loss": 0.0041,
      "step": 51030
    },
    {
      "epoch": 0.8351468542910906,
      "grad_norm": 0.2805325984954834,
      "learning_rate": 9.061638444748148e-06,
      "loss": 0.0029,
      "step": 51040
    },
    {
      "epoch": 0.8353104802421664,
      "grad_norm": 0.02529214508831501,
      "learning_rate": 9.061083198471667e-06,
      "loss": 0.003,
      "step": 51050
    },
    {
      "epoch": 0.8354741061932422,
      "grad_norm": 0.14584942162036896,
      "learning_rate": 9.060527804990682e-06,
      "loss": 0.0026,
      "step": 51060
    },
    {
      "epoch": 0.8356377321443181,
      "grad_norm": 0.27215349674224854,
      "learning_rate": 9.059972264325321e-06,
      "loss": 0.0035,
      "step": 51070
    },
    {
      "epoch": 0.8358013580953939,
      "grad_norm": 0.0744057446718216,
      "learning_rate": 9.059416576495725e-06,
      "loss": 0.0031,
      "step": 51080
    },
    {
      "epoch": 0.8359649840464698,
      "grad_norm": 0.03862448409199715,
      "learning_rate": 9.058860741522035e-06,
      "loss": 0.003,
      "step": 51090
    },
    {
      "epoch": 0.8361286099975456,
      "grad_norm": 0.8369637131690979,
      "learning_rate": 9.058304759424396e-06,
      "loss": 0.0022,
      "step": 51100
    },
    {
      "epoch": 0.8362922359486215,
      "grad_norm": 0.2615121006965637,
      "learning_rate": 9.057748630222966e-06,
      "loss": 0.0046,
      "step": 51110
    },
    {
      "epoch": 0.8364558618996973,
      "grad_norm": 0.21378803253173828,
      "learning_rate": 9.057192353937901e-06,
      "loss": 0.0026,
      "step": 51120
    },
    {
      "epoch": 0.8366194878507731,
      "grad_norm": 0.17934903502464294,
      "learning_rate": 9.056635930589362e-06,
      "loss": 0.0036,
      "step": 51130
    },
    {
      "epoch": 0.836783113801849,
      "grad_norm": 0.22878490388393402,
      "learning_rate": 9.056079360197522e-06,
      "loss": 0.0038,
      "step": 51140
    },
    {
      "epoch": 0.8369467397529248,
      "grad_norm": 0.13659565150737762,
      "learning_rate": 9.055522642782553e-06,
      "loss": 0.0041,
      "step": 51150
    },
    {
      "epoch": 0.8371103657040007,
      "grad_norm": 0.0801631361246109,
      "learning_rate": 9.054965778364637e-06,
      "loss": 0.0016,
      "step": 51160
    },
    {
      "epoch": 0.8372739916550765,
      "grad_norm": 0.1493367850780487,
      "learning_rate": 9.054408766963956e-06,
      "loss": 0.0035,
      "step": 51170
    },
    {
      "epoch": 0.8374376176061523,
      "grad_norm": 0.10300180315971375,
      "learning_rate": 9.053851608600702e-06,
      "loss": 0.0045,
      "step": 51180
    },
    {
      "epoch": 0.8376012435572282,
      "grad_norm": 0.06744793802499771,
      "learning_rate": 9.053294303295069e-06,
      "loss": 0.0046,
      "step": 51190
    },
    {
      "epoch": 0.837764869508304,
      "grad_norm": 0.07411858439445496,
      "learning_rate": 9.052736851067261e-06,
      "loss": 0.0037,
      "step": 51200
    },
    {
      "epoch": 0.8379284954593799,
      "grad_norm": 0.20761512219905853,
      "learning_rate": 9.052179251937482e-06,
      "loss": 0.0042,
      "step": 51210
    },
    {
      "epoch": 0.8380921214104557,
      "grad_norm": 0.2238827794790268,
      "learning_rate": 9.051621505925945e-06,
      "loss": 0.0027,
      "step": 51220
    },
    {
      "epoch": 0.8382557473615315,
      "grad_norm": 0.03649295121431351,
      "learning_rate": 9.051063613052865e-06,
      "loss": 0.003,
      "step": 51230
    },
    {
      "epoch": 0.8384193733126074,
      "grad_norm": 0.20875859260559082,
      "learning_rate": 9.050505573338467e-06,
      "loss": 0.0046,
      "step": 51240
    },
    {
      "epoch": 0.8385829992636832,
      "grad_norm": 0.10120797902345657,
      "learning_rate": 9.049947386802975e-06,
      "loss": 0.0028,
      "step": 51250
    },
    {
      "epoch": 0.8387466252147591,
      "grad_norm": 0.22312410175800323,
      "learning_rate": 9.049389053466625e-06,
      "loss": 0.0033,
      "step": 51260
    },
    {
      "epoch": 0.838910251165835,
      "grad_norm": 0.05088645592331886,
      "learning_rate": 9.048830573349656e-06,
      "loss": 0.0042,
      "step": 51270
    },
    {
      "epoch": 0.8390738771169107,
      "grad_norm": 0.1280689388513565,
      "learning_rate": 9.048271946472306e-06,
      "loss": 0.0034,
      "step": 51280
    },
    {
      "epoch": 0.8392375030679866,
      "grad_norm": 0.09160040318965912,
      "learning_rate": 9.047713172854832e-06,
      "loss": 0.0034,
      "step": 51290
    },
    {
      "epoch": 0.8394011290190624,
      "grad_norm": 0.4399280846118927,
      "learning_rate": 9.047154252517482e-06,
      "loss": 0.003,
      "step": 51300
    },
    {
      "epoch": 0.8395647549701383,
      "grad_norm": 0.2895500957965851,
      "learning_rate": 9.046595185480517e-06,
      "loss": 0.0031,
      "step": 51310
    },
    {
      "epoch": 0.8397283809212142,
      "grad_norm": 0.13036033511161804,
      "learning_rate": 9.046035971764202e-06,
      "loss": 0.0031,
      "step": 51320
    },
    {
      "epoch": 0.8398920068722899,
      "grad_norm": 0.273827463388443,
      "learning_rate": 9.045476611388808e-06,
      "loss": 0.0023,
      "step": 51330
    },
    {
      "epoch": 0.8400556328233658,
      "grad_norm": 0.31248846650123596,
      "learning_rate": 9.04491710437461e-06,
      "loss": 0.0032,
      "step": 51340
    },
    {
      "epoch": 0.8402192587744416,
      "grad_norm": 0.13093781471252441,
      "learning_rate": 9.04435745074189e-06,
      "loss": 0.0052,
      "step": 51350
    },
    {
      "epoch": 0.8403828847255175,
      "grad_norm": 0.35147008299827576,
      "learning_rate": 9.04379765051093e-06,
      "loss": 0.0028,
      "step": 51360
    },
    {
      "epoch": 0.8405465106765934,
      "grad_norm": 0.14003416895866394,
      "learning_rate": 9.043237703702026e-06,
      "loss": 0.0023,
      "step": 51370
    },
    {
      "epoch": 0.8407101366276691,
      "grad_norm": 0.36045512557029724,
      "learning_rate": 9.042677610335473e-06,
      "loss": 0.0034,
      "step": 51380
    },
    {
      "epoch": 0.840873762578745,
      "grad_norm": 0.04314054176211357,
      "learning_rate": 9.042117370431573e-06,
      "loss": 0.0017,
      "step": 51390
    },
    {
      "epoch": 0.8410373885298208,
      "grad_norm": 0.03455325588583946,
      "learning_rate": 9.041556984010633e-06,
      "loss": 0.0028,
      "step": 51400
    },
    {
      "epoch": 0.8412010144808967,
      "grad_norm": 0.11042078584432602,
      "learning_rate": 9.040996451092966e-06,
      "loss": 0.0025,
      "step": 51410
    },
    {
      "epoch": 0.8413646404319726,
      "grad_norm": 0.08467378467321396,
      "learning_rate": 9.040435771698887e-06,
      "loss": 0.0034,
      "step": 51420
    },
    {
      "epoch": 0.8415282663830483,
      "grad_norm": 0.29322943091392517,
      "learning_rate": 9.039874945848726e-06,
      "loss": 0.0042,
      "step": 51430
    },
    {
      "epoch": 0.8416918923341242,
      "grad_norm": 0.1992769092321396,
      "learning_rate": 9.039313973562807e-06,
      "loss": 0.0027,
      "step": 51440
    },
    {
      "epoch": 0.8418555182852,
      "grad_norm": 0.09897596389055252,
      "learning_rate": 9.038752854861464e-06,
      "loss": 0.0025,
      "step": 51450
    },
    {
      "epoch": 0.8420191442362759,
      "grad_norm": 0.08596764504909515,
      "learning_rate": 9.038191589765038e-06,
      "loss": 0.0032,
      "step": 51460
    },
    {
      "epoch": 0.8421827701873517,
      "grad_norm": 0.1174035295844078,
      "learning_rate": 9.037630178293871e-06,
      "loss": 0.0041,
      "step": 51470
    },
    {
      "epoch": 0.8423463961384275,
      "grad_norm": 0.1402428299188614,
      "learning_rate": 9.037068620468315e-06,
      "loss": 0.0027,
      "step": 51480
    },
    {
      "epoch": 0.8425100220895034,
      "grad_norm": 0.18283773958683014,
      "learning_rate": 9.036506916308723e-06,
      "loss": 0.0023,
      "step": 51490
    },
    {
      "epoch": 0.8426736480405792,
      "grad_norm": 0.13473039865493774,
      "learning_rate": 9.035945065835459e-06,
      "loss": 0.0027,
      "step": 51500
    },
    {
      "epoch": 0.8428372739916551,
      "grad_norm": 0.047684382647275925,
      "learning_rate": 9.035383069068884e-06,
      "loss": 0.0037,
      "step": 51510
    },
    {
      "epoch": 0.8430008999427309,
      "grad_norm": 0.15443789958953857,
      "learning_rate": 9.034820926029374e-06,
      "loss": 0.0024,
      "step": 51520
    },
    {
      "epoch": 0.8431645258938067,
      "grad_norm": 0.23368026316165924,
      "learning_rate": 9.034258636737302e-06,
      "loss": 0.0019,
      "step": 51530
    },
    {
      "epoch": 0.8433281518448826,
      "grad_norm": 0.08092086762189865,
      "learning_rate": 9.033696201213051e-06,
      "loss": 0.0053,
      "step": 51540
    },
    {
      "epoch": 0.8434917777959584,
      "grad_norm": 0.10714762657880783,
      "learning_rate": 9.033133619477006e-06,
      "loss": 0.0023,
      "step": 51550
    },
    {
      "epoch": 0.8436554037470343,
      "grad_norm": 0.24213559925556183,
      "learning_rate": 9.032570891549562e-06,
      "loss": 0.0028,
      "step": 51560
    },
    {
      "epoch": 0.8438190296981101,
      "grad_norm": 0.17505262792110443,
      "learning_rate": 9.032008017451114e-06,
      "loss": 0.0029,
      "step": 51570
    },
    {
      "epoch": 0.843982655649186,
      "grad_norm": 0.07047256082296371,
      "learning_rate": 9.031444997202068e-06,
      "loss": 0.0023,
      "step": 51580
    },
    {
      "epoch": 0.8441462816002618,
      "grad_norm": 0.21796371042728424,
      "learning_rate": 9.030881830822829e-06,
      "loss": 0.0031,
      "step": 51590
    },
    {
      "epoch": 0.8443099075513376,
      "grad_norm": 0.02481258288025856,
      "learning_rate": 9.03031851833381e-06,
      "loss": 0.0046,
      "step": 51600
    },
    {
      "epoch": 0.8444735335024135,
      "grad_norm": 0.014050047844648361,
      "learning_rate": 9.029755059755433e-06,
      "loss": 0.0017,
      "step": 51610
    },
    {
      "epoch": 0.8446371594534893,
      "grad_norm": 0.24635645747184753,
      "learning_rate": 9.029191455108121e-06,
      "loss": 0.0047,
      "step": 51620
    },
    {
      "epoch": 0.8448007854045652,
      "grad_norm": 0.14891695976257324,
      "learning_rate": 9.0286277044123e-06,
      "loss": 0.0024,
      "step": 51630
    },
    {
      "epoch": 0.844964411355641,
      "grad_norm": 0.14402726292610168,
      "learning_rate": 9.02806380768841e-06,
      "loss": 0.0035,
      "step": 51640
    },
    {
      "epoch": 0.8451280373067168,
      "grad_norm": 0.0439751110970974,
      "learning_rate": 9.027499764956888e-06,
      "loss": 0.0048,
      "step": 51650
    },
    {
      "epoch": 0.8452916632577927,
      "grad_norm": 0.09451764822006226,
      "learning_rate": 9.026935576238178e-06,
      "loss": 0.0037,
      "step": 51660
    },
    {
      "epoch": 0.8454552892088685,
      "grad_norm": 0.3115558326244354,
      "learning_rate": 9.026371241552732e-06,
      "loss": 0.0068,
      "step": 51670
    },
    {
      "epoch": 0.8456189151599444,
      "grad_norm": 0.1732897311449051,
      "learning_rate": 9.025806760921007e-06,
      "loss": 0.0029,
      "step": 51680
    },
    {
      "epoch": 0.8457825411110202,
      "grad_norm": 0.16963207721710205,
      "learning_rate": 9.025242134363462e-06,
      "loss": 0.0042,
      "step": 51690
    },
    {
      "epoch": 0.845946167062096,
      "grad_norm": 0.11961447447538376,
      "learning_rate": 9.024677361900563e-06,
      "loss": 0.0031,
      "step": 51700
    },
    {
      "epoch": 0.8461097930131719,
      "grad_norm": 0.1747836470603943,
      "learning_rate": 9.024112443552784e-06,
      "loss": 0.0035,
      "step": 51710
    },
    {
      "epoch": 0.8462734189642477,
      "grad_norm": 0.20227637887001038,
      "learning_rate": 9.0235473793406e-06,
      "loss": 0.0033,
      "step": 51720
    },
    {
      "epoch": 0.8464370449153236,
      "grad_norm": 0.26923802495002747,
      "learning_rate": 9.022982169284494e-06,
      "loss": 0.0024,
      "step": 51730
    },
    {
      "epoch": 0.8466006708663995,
      "grad_norm": 0.2732408046722412,
      "learning_rate": 9.022416813404954e-06,
      "loss": 0.004,
      "step": 51740
    },
    {
      "epoch": 0.8467642968174752,
      "grad_norm": 0.056906960904598236,
      "learning_rate": 9.021851311722472e-06,
      "loss": 0.0024,
      "step": 51750
    },
    {
      "epoch": 0.8469279227685511,
      "grad_norm": 0.15122783184051514,
      "learning_rate": 9.021285664257544e-06,
      "loss": 0.0039,
      "step": 51760
    },
    {
      "epoch": 0.8470915487196269,
      "grad_norm": 0.2581248879432678,
      "learning_rate": 9.020719871030678e-06,
      "loss": 0.0029,
      "step": 51770
    },
    {
      "epoch": 0.8472551746707028,
      "grad_norm": 0.12173102796077728,
      "learning_rate": 9.020153932062379e-06,
      "loss": 0.0047,
      "step": 51780
    },
    {
      "epoch": 0.8474188006217787,
      "grad_norm": 0.10298284888267517,
      "learning_rate": 9.019587847373163e-06,
      "loss": 0.002,
      "step": 51790
    },
    {
      "epoch": 0.8475824265728544,
      "grad_norm": 0.4102534353733063,
      "learning_rate": 9.019021616983548e-06,
      "loss": 0.0044,
      "step": 51800
    },
    {
      "epoch": 0.8477460525239303,
      "grad_norm": 0.13126803934574127,
      "learning_rate": 9.01845524091406e-06,
      "loss": 0.0042,
      "step": 51810
    },
    {
      "epoch": 0.8479096784750061,
      "grad_norm": 0.1789359599351883,
      "learning_rate": 9.017888719185226e-06,
      "loss": 0.0044,
      "step": 51820
    },
    {
      "epoch": 0.848073304426082,
      "grad_norm": 0.16379286348819733,
      "learning_rate": 9.017322051817584e-06,
      "loss": 0.006,
      "step": 51830
    },
    {
      "epoch": 0.8482369303771579,
      "grad_norm": 0.13456517457962036,
      "learning_rate": 9.016755238831671e-06,
      "loss": 0.0024,
      "step": 51840
    },
    {
      "epoch": 0.8484005563282336,
      "grad_norm": 0.11003376543521881,
      "learning_rate": 9.016188280248034e-06,
      "loss": 0.0033,
      "step": 51850
    },
    {
      "epoch": 0.8485641822793095,
      "grad_norm": 0.10518805682659149,
      "learning_rate": 9.015621176087228e-06,
      "loss": 0.0032,
      "step": 51860
    },
    {
      "epoch": 0.8487278082303853,
      "grad_norm": 0.2536409795284271,
      "learning_rate": 9.015053926369803e-06,
      "loss": 0.003,
      "step": 51870
    },
    {
      "epoch": 0.8488914341814612,
      "grad_norm": 0.31367555260658264,
      "learning_rate": 9.014486531116324e-06,
      "loss": 0.003,
      "step": 51880
    },
    {
      "epoch": 0.8490550601325371,
      "grad_norm": 0.06684146076440811,
      "learning_rate": 9.013918990347357e-06,
      "loss": 0.0026,
      "step": 51890
    },
    {
      "epoch": 0.8492186860836128,
      "grad_norm": 0.19261233508586884,
      "learning_rate": 9.013351304083475e-06,
      "loss": 0.0035,
      "step": 51900
    },
    {
      "epoch": 0.8493823120346887,
      "grad_norm": 0.11500176787376404,
      "learning_rate": 9.012783472345253e-06,
      "loss": 0.0019,
      "step": 51910
    },
    {
      "epoch": 0.8495459379857645,
      "grad_norm": 0.26511287689208984,
      "learning_rate": 9.012215495153273e-06,
      "loss": 0.0047,
      "step": 51920
    },
    {
      "epoch": 0.8497095639368404,
      "grad_norm": 0.07897260785102844,
      "learning_rate": 9.011647372528125e-06,
      "loss": 0.0033,
      "step": 51930
    },
    {
      "epoch": 0.8498731898879163,
      "grad_norm": 0.041406337171792984,
      "learning_rate": 9.011079104490403e-06,
      "loss": 0.003,
      "step": 51940
    },
    {
      "epoch": 0.850036815838992,
      "grad_norm": 0.1999581903219223,
      "learning_rate": 9.010510691060704e-06,
      "loss": 0.0029,
      "step": 51950
    },
    {
      "epoch": 0.8502004417900679,
      "grad_norm": 0.274642676115036,
      "learning_rate": 9.009942132259629e-06,
      "loss": 0.0097,
      "step": 51960
    },
    {
      "epoch": 0.8503640677411437,
      "grad_norm": 0.13690555095672607,
      "learning_rate": 9.00937342810779e-06,
      "loss": 0.0027,
      "step": 51970
    },
    {
      "epoch": 0.8505276936922196,
      "grad_norm": 0.1821022778749466,
      "learning_rate": 9.008804578625803e-06,
      "loss": 0.0038,
      "step": 51980
    },
    {
      "epoch": 0.8506913196432955,
      "grad_norm": 0.214046448469162,
      "learning_rate": 9.008235583834283e-06,
      "loss": 0.0037,
      "step": 51990
    },
    {
      "epoch": 0.8508549455943712,
      "grad_norm": 0.1042679101228714,
      "learning_rate": 9.007666443753856e-06,
      "loss": 0.0025,
      "step": 52000
    },
    {
      "epoch": 0.8510185715454471,
      "grad_norm": 0.06679016351699829,
      "learning_rate": 9.007097158405153e-06,
      "loss": 0.0025,
      "step": 52010
    },
    {
      "epoch": 0.8511821974965229,
      "grad_norm": 0.10288181155920029,
      "learning_rate": 9.006527727808811e-06,
      "loss": 0.0047,
      "step": 52020
    },
    {
      "epoch": 0.8513458234475988,
      "grad_norm": 0.12589871883392334,
      "learning_rate": 9.005958151985466e-06,
      "loss": 0.0025,
      "step": 52030
    },
    {
      "epoch": 0.8515094493986747,
      "grad_norm": 0.13085179030895233,
      "learning_rate": 9.005388430955767e-06,
      "loss": 0.0034,
      "step": 52040
    },
    {
      "epoch": 0.8516730753497505,
      "grad_norm": 0.10764490067958832,
      "learning_rate": 9.004818564740362e-06,
      "loss": 0.0041,
      "step": 52050
    },
    {
      "epoch": 0.8518367013008263,
      "grad_norm": 0.10860119014978409,
      "learning_rate": 9.004248553359911e-06,
      "loss": 0.002,
      "step": 52060
    },
    {
      "epoch": 0.8520003272519021,
      "grad_norm": 0.1613011360168457,
      "learning_rate": 9.003678396835076e-06,
      "loss": 0.0023,
      "step": 52070
    },
    {
      "epoch": 0.852163953202978,
      "grad_norm": 0.04479849338531494,
      "learning_rate": 9.003108095186519e-06,
      "loss": 0.0018,
      "step": 52080
    },
    {
      "epoch": 0.8523275791540539,
      "grad_norm": 0.06335500627756119,
      "learning_rate": 9.002537648434914e-06,
      "loss": 0.0021,
      "step": 52090
    },
    {
      "epoch": 0.8524912051051297,
      "grad_norm": 0.3290322422981262,
      "learning_rate": 9.001967056600941e-06,
      "loss": 0.0026,
      "step": 52100
    },
    {
      "epoch": 0.8526548310562055,
      "grad_norm": 0.1582181304693222,
      "learning_rate": 9.001396319705282e-06,
      "loss": 0.0022,
      "step": 52110
    },
    {
      "epoch": 0.8528184570072813,
      "grad_norm": 0.03260332718491554,
      "learning_rate": 9.000825437768622e-06,
      "loss": 0.0029,
      "step": 52120
    },
    {
      "epoch": 0.8529820829583572,
      "grad_norm": 0.13020731508731842,
      "learning_rate": 9.000254410811654e-06,
      "loss": 0.0032,
      "step": 52130
    },
    {
      "epoch": 0.8531457089094331,
      "grad_norm": 0.02632703073322773,
      "learning_rate": 8.99968323885508e-06,
      "loss": 0.0024,
      "step": 52140
    },
    {
      "epoch": 0.8533093348605089,
      "grad_norm": 0.12739895284175873,
      "learning_rate": 8.999111921919601e-06,
      "loss": 0.0033,
      "step": 52150
    },
    {
      "epoch": 0.8534729608115847,
      "grad_norm": 0.09220941364765167,
      "learning_rate": 8.998540460025926e-06,
      "loss": 0.0026,
      "step": 52160
    },
    {
      "epoch": 0.8536365867626605,
      "grad_norm": 0.2832767963409424,
      "learning_rate": 8.99796885319477e-06,
      "loss": 0.0024,
      "step": 52170
    },
    {
      "epoch": 0.8538002127137364,
      "grad_norm": 0.1344723403453827,
      "learning_rate": 8.997397101446852e-06,
      "loss": 0.0021,
      "step": 52180
    },
    {
      "epoch": 0.8539638386648123,
      "grad_norm": 0.35045450925827026,
      "learning_rate": 8.996825204802895e-06,
      "loss": 0.0017,
      "step": 52190
    },
    {
      "epoch": 0.8541274646158881,
      "grad_norm": 0.08236874639987946,
      "learning_rate": 8.996253163283633e-06,
      "loss": 0.0044,
      "step": 52200
    },
    {
      "epoch": 0.854291090566964,
      "grad_norm": 0.3529832661151886,
      "learning_rate": 8.995680976909798e-06,
      "loss": 0.0028,
      "step": 52210
    },
    {
      "epoch": 0.8544547165180397,
      "grad_norm": 0.3102409541606903,
      "learning_rate": 8.995108645702128e-06,
      "loss": 0.0047,
      "step": 52220
    },
    {
      "epoch": 0.8546183424691156,
      "grad_norm": 0.11922548711299896,
      "learning_rate": 8.994536169681376e-06,
      "loss": 0.0042,
      "step": 52230
    },
    {
      "epoch": 0.8547819684201915,
      "grad_norm": 0.12232251465320587,
      "learning_rate": 8.993963548868286e-06,
      "loss": 0.0018,
      "step": 52240
    },
    {
      "epoch": 0.8549455943712673,
      "grad_norm": 0.17454329133033752,
      "learning_rate": 8.993390783283617e-06,
      "loss": 0.0048,
      "step": 52250
    },
    {
      "epoch": 0.8551092203223432,
      "grad_norm": 0.1342998445034027,
      "learning_rate": 8.99281787294813e-06,
      "loss": 0.0026,
      "step": 52260
    },
    {
      "epoch": 0.8552728462734189,
      "grad_norm": 0.15353047847747803,
      "learning_rate": 8.992244817882592e-06,
      "loss": 0.0026,
      "step": 52270
    },
    {
      "epoch": 0.8554364722244948,
      "grad_norm": 0.164621502161026,
      "learning_rate": 8.991671618107773e-06,
      "loss": 0.0023,
      "step": 52280
    },
    {
      "epoch": 0.8556000981755707,
      "grad_norm": 0.11834081262350082,
      "learning_rate": 8.991098273644453e-06,
      "loss": 0.0038,
      "step": 52290
    },
    {
      "epoch": 0.8557637241266465,
      "grad_norm": 0.04167987033724785,
      "learning_rate": 8.990524784513412e-06,
      "loss": 0.0048,
      "step": 52300
    },
    {
      "epoch": 0.8559273500777224,
      "grad_norm": 0.13660024106502533,
      "learning_rate": 8.989951150735438e-06,
      "loss": 0.002,
      "step": 52310
    },
    {
      "epoch": 0.8560909760287981,
      "grad_norm": 0.1240919679403305,
      "learning_rate": 8.989377372331327e-06,
      "loss": 0.0026,
      "step": 52320
    },
    {
      "epoch": 0.856254601979874,
      "grad_norm": 0.11201497912406921,
      "learning_rate": 8.988803449321872e-06,
      "loss": 0.0025,
      "step": 52330
    },
    {
      "epoch": 0.8564182279309498,
      "grad_norm": 0.29503849148750305,
      "learning_rate": 8.98822938172788e-06,
      "loss": 0.0031,
      "step": 52340
    },
    {
      "epoch": 0.8565818538820257,
      "grad_norm": 0.1383712738752365,
      "learning_rate": 8.987655169570157e-06,
      "loss": 0.0025,
      "step": 52350
    },
    {
      "epoch": 0.8567454798331016,
      "grad_norm": 0.15909098088741302,
      "learning_rate": 8.98708081286952e-06,
      "loss": 0.0023,
      "step": 52360
    },
    {
      "epoch": 0.8569091057841773,
      "grad_norm": 0.17658723890781403,
      "learning_rate": 8.986506311646785e-06,
      "loss": 0.0028,
      "step": 52370
    },
    {
      "epoch": 0.8570727317352532,
      "grad_norm": 0.07994277775287628,
      "learning_rate": 8.985931665922778e-06,
      "loss": 0.0042,
      "step": 52380
    },
    {
      "epoch": 0.857236357686329,
      "grad_norm": 0.07423728704452515,
      "learning_rate": 8.98535687571833e-06,
      "loss": 0.0026,
      "step": 52390
    },
    {
      "epoch": 0.8573999836374049,
      "grad_norm": 0.09365329146385193,
      "learning_rate": 8.98478194105427e-06,
      "loss": 0.0027,
      "step": 52400
    },
    {
      "epoch": 0.8575636095884808,
      "grad_norm": 0.14266206324100494,
      "learning_rate": 8.984206861951443e-06,
      "loss": 0.0028,
      "step": 52410
    },
    {
      "epoch": 0.8577272355395565,
      "grad_norm": 0.16483327746391296,
      "learning_rate": 8.983631638430694e-06,
      "loss": 0.0041,
      "step": 52420
    },
    {
      "epoch": 0.8578908614906324,
      "grad_norm": 0.1589885801076889,
      "learning_rate": 8.983056270512874e-06,
      "loss": 0.0036,
      "step": 52430
    },
    {
      "epoch": 0.8580544874417082,
      "grad_norm": 0.09253576397895813,
      "learning_rate": 8.982480758218835e-06,
      "loss": 0.0036,
      "step": 52440
    },
    {
      "epoch": 0.8582181133927841,
      "grad_norm": 0.17469380795955658,
      "learning_rate": 8.98190510156944e-06,
      "loss": 0.0029,
      "step": 52450
    },
    {
      "epoch": 0.85838173934386,
      "grad_norm": 0.08178523182868958,
      "learning_rate": 8.981329300585556e-06,
      "loss": 0.0026,
      "step": 52460
    },
    {
      "epoch": 0.8585453652949357,
      "grad_norm": 0.3584573566913605,
      "learning_rate": 8.980753355288055e-06,
      "loss": 0.0056,
      "step": 52470
    },
    {
      "epoch": 0.8587089912460116,
      "grad_norm": 0.08372239768505096,
      "learning_rate": 8.98017726569781e-06,
      "loss": 0.0026,
      "step": 52480
    },
    {
      "epoch": 0.8588726171970874,
      "grad_norm": 0.2778053879737854,
      "learning_rate": 8.979601031835705e-06,
      "loss": 0.0032,
      "step": 52490
    },
    {
      "epoch": 0.8590362431481633,
      "grad_norm": 0.30251872539520264,
      "learning_rate": 8.97902465372263e-06,
      "loss": 0.0033,
      "step": 52500
    },
    {
      "epoch": 0.8591998690992392,
      "grad_norm": 0.09378662705421448,
      "learning_rate": 8.978448131379474e-06,
      "loss": 0.0042,
      "step": 52510
    },
    {
      "epoch": 0.859363495050315,
      "grad_norm": 0.19447028636932373,
      "learning_rate": 8.977871464827133e-06,
      "loss": 0.0038,
      "step": 52520
    },
    {
      "epoch": 0.8595271210013908,
      "grad_norm": 0.26850658655166626,
      "learning_rate": 8.977294654086514e-06,
      "loss": 0.0048,
      "step": 52530
    },
    {
      "epoch": 0.8596907469524666,
      "grad_norm": 0.08822564780712128,
      "learning_rate": 8.976717699178521e-06,
      "loss": 0.0022,
      "step": 52540
    },
    {
      "epoch": 0.8598543729035425,
      "grad_norm": 0.21378576755523682,
      "learning_rate": 8.976140600124071e-06,
      "loss": 0.0037,
      "step": 52550
    },
    {
      "epoch": 0.8600179988546184,
      "grad_norm": 0.09782931208610535,
      "learning_rate": 8.97556335694408e-06,
      "loss": 0.0025,
      "step": 52560
    },
    {
      "epoch": 0.8601816248056942,
      "grad_norm": 0.13276933133602142,
      "learning_rate": 8.974985969659473e-06,
      "loss": 0.0035,
      "step": 52570
    },
    {
      "epoch": 0.86034525075677,
      "grad_norm": 0.32663071155548096,
      "learning_rate": 8.974408438291178e-06,
      "loss": 0.0037,
      "step": 52580
    },
    {
      "epoch": 0.8605088767078458,
      "grad_norm": 0.4171352982521057,
      "learning_rate": 8.973830762860128e-06,
      "loss": 0.0033,
      "step": 52590
    },
    {
      "epoch": 0.8606725026589217,
      "grad_norm": 0.1790311336517334,
      "learning_rate": 8.973252943387265e-06,
      "loss": 0.0023,
      "step": 52600
    },
    {
      "epoch": 0.8608361286099976,
      "grad_norm": 0.11530248075723648,
      "learning_rate": 8.972674979893531e-06,
      "loss": 0.0023,
      "step": 52610
    },
    {
      "epoch": 0.8609997545610734,
      "grad_norm": 0.13855765759944916,
      "learning_rate": 8.972096872399879e-06,
      "loss": 0.0081,
      "step": 52620
    },
    {
      "epoch": 0.8611633805121492,
      "grad_norm": 0.043049514293670654,
      "learning_rate": 8.971518620927263e-06,
      "loss": 0.0032,
      "step": 52630
    },
    {
      "epoch": 0.861327006463225,
      "grad_norm": 0.4812357425689697,
      "learning_rate": 8.970940225496639e-06,
      "loss": 0.0043,
      "step": 52640
    },
    {
      "epoch": 0.8614906324143009,
      "grad_norm": 0.08960550278425217,
      "learning_rate": 8.970361686128976e-06,
      "loss": 0.0048,
      "step": 52650
    },
    {
      "epoch": 0.8616542583653768,
      "grad_norm": 0.10986928641796112,
      "learning_rate": 8.969783002845247e-06,
      "loss": 0.0044,
      "step": 52660
    },
    {
      "epoch": 0.8618178843164526,
      "grad_norm": 0.17350338399410248,
      "learning_rate": 8.969204175666424e-06,
      "loss": 0.0032,
      "step": 52670
    },
    {
      "epoch": 0.8619815102675285,
      "grad_norm": 0.18268626928329468,
      "learning_rate": 8.968625204613488e-06,
      "loss": 0.0024,
      "step": 52680
    },
    {
      "epoch": 0.8621451362186042,
      "grad_norm": 0.06502486020326614,
      "learning_rate": 8.968046089707427e-06,
      "loss": 0.0027,
      "step": 52690
    },
    {
      "epoch": 0.8623087621696801,
      "grad_norm": 0.23045609891414642,
      "learning_rate": 8.967466830969234e-06,
      "loss": 0.0025,
      "step": 52700
    },
    {
      "epoch": 0.862472388120756,
      "grad_norm": 0.1758364588022232,
      "learning_rate": 8.966887428419903e-06,
      "loss": 0.0028,
      "step": 52710
    },
    {
      "epoch": 0.8626360140718318,
      "grad_norm": 0.2817644476890564,
      "learning_rate": 8.966307882080436e-06,
      "loss": 0.0032,
      "step": 52720
    },
    {
      "epoch": 0.8627996400229077,
      "grad_norm": 0.10124043375253677,
      "learning_rate": 8.965728191971843e-06,
      "loss": 0.003,
      "step": 52730
    },
    {
      "epoch": 0.8629632659739834,
      "grad_norm": 0.4323711097240448,
      "learning_rate": 8.965148358115131e-06,
      "loss": 0.0031,
      "step": 52740
    },
    {
      "epoch": 0.8631268919250593,
      "grad_norm": 0.3049001097679138,
      "learning_rate": 8.964568380531325e-06,
      "loss": 0.0036,
      "step": 52750
    },
    {
      "epoch": 0.8632905178761352,
      "grad_norm": 0.13822802901268005,
      "learning_rate": 8.963988259241443e-06,
      "loss": 0.0027,
      "step": 52760
    },
    {
      "epoch": 0.863454143827211,
      "grad_norm": 0.46392422914505005,
      "learning_rate": 8.963407994266512e-06,
      "loss": 0.0041,
      "step": 52770
    },
    {
      "epoch": 0.8636177697782869,
      "grad_norm": 0.23145531117916107,
      "learning_rate": 8.962827585627569e-06,
      "loss": 0.0024,
      "step": 52780
    },
    {
      "epoch": 0.8637813957293626,
      "grad_norm": 0.07914381474256516,
      "learning_rate": 8.962247033345648e-06,
      "loss": 0.0031,
      "step": 52790
    },
    {
      "epoch": 0.8639450216804385,
      "grad_norm": 0.08892399817705154,
      "learning_rate": 8.961666337441797e-06,
      "loss": 0.0024,
      "step": 52800
    },
    {
      "epoch": 0.8641086476315144,
      "grad_norm": 0.19455312192440033,
      "learning_rate": 8.961085497937061e-06,
      "loss": 0.0045,
      "step": 52810
    },
    {
      "epoch": 0.8642722735825902,
      "grad_norm": 0.174282506108284,
      "learning_rate": 8.960504514852498e-06,
      "loss": 0.0037,
      "step": 52820
    },
    {
      "epoch": 0.8644358995336661,
      "grad_norm": 0.06773357093334198,
      "learning_rate": 8.959923388209164e-06,
      "loss": 0.005,
      "step": 52830
    },
    {
      "epoch": 0.8645995254847418,
      "grad_norm": 0.11846756190061569,
      "learning_rate": 8.959342118028124e-06,
      "loss": 0.0024,
      "step": 52840
    },
    {
      "epoch": 0.8647631514358177,
      "grad_norm": 0.016769489273428917,
      "learning_rate": 8.95876070433045e-06,
      "loss": 0.0035,
      "step": 52850
    },
    {
      "epoch": 0.8649267773868936,
      "grad_norm": 0.08660092204809189,
      "learning_rate": 8.958179147137215e-06,
      "loss": 0.0024,
      "step": 52860
    },
    {
      "epoch": 0.8650904033379694,
      "grad_norm": 0.0785297304391861,
      "learning_rate": 8.957597446469498e-06,
      "loss": 0.0037,
      "step": 52870
    },
    {
      "epoch": 0.8652540292890453,
      "grad_norm": 0.10576910525560379,
      "learning_rate": 8.957015602348387e-06,
      "loss": 0.0034,
      "step": 52880
    },
    {
      "epoch": 0.865417655240121,
      "grad_norm": 0.11605571955442429,
      "learning_rate": 8.956433614794968e-06,
      "loss": 0.0018,
      "step": 52890
    },
    {
      "epoch": 0.8655812811911969,
      "grad_norm": 0.2694227397441864,
      "learning_rate": 8.955851483830342e-06,
      "loss": 0.003,
      "step": 52900
    },
    {
      "epoch": 0.8657449071422728,
      "grad_norm": 0.1706376075744629,
      "learning_rate": 8.955269209475607e-06,
      "loss": 0.0027,
      "step": 52910
    },
    {
      "epoch": 0.8659085330933486,
      "grad_norm": 0.47280558943748474,
      "learning_rate": 8.954686791751868e-06,
      "loss": 0.0028,
      "step": 52920
    },
    {
      "epoch": 0.8660721590444245,
      "grad_norm": 0.4548928439617157,
      "learning_rate": 8.954104230680238e-06,
      "loss": 0.0024,
      "step": 52930
    },
    {
      "epoch": 0.8662357849955002,
      "grad_norm": 0.06332504004240036,
      "learning_rate": 8.953521526281834e-06,
      "loss": 0.0017,
      "step": 52940
    },
    {
      "epoch": 0.8663994109465761,
      "grad_norm": 0.3046591579914093,
      "learning_rate": 8.952938678577777e-06,
      "loss": 0.0046,
      "step": 52950
    },
    {
      "epoch": 0.866563036897652,
      "grad_norm": 0.1686720997095108,
      "learning_rate": 8.952355687589194e-06,
      "loss": 0.0029,
      "step": 52960
    },
    {
      "epoch": 0.8667266628487278,
      "grad_norm": 0.2987702488899231,
      "learning_rate": 8.951772553337215e-06,
      "loss": 0.0029,
      "step": 52970
    },
    {
      "epoch": 0.8668902887998037,
      "grad_norm": 0.07991890609264374,
      "learning_rate": 8.95118927584298e-06,
      "loss": 0.0039,
      "step": 52980
    },
    {
      "epoch": 0.8670539147508795,
      "grad_norm": 0.11487843841314316,
      "learning_rate": 8.950605855127627e-06,
      "loss": 0.0035,
      "step": 52990
    },
    {
      "epoch": 0.8672175407019553,
      "grad_norm": 0.20543067157268524,
      "learning_rate": 8.95002229121231e-06,
      "loss": 0.0052,
      "step": 53000
    },
    {
      "epoch": 0.8673811666530312,
      "grad_norm": 0.17694096267223358,
      "learning_rate": 8.949438584118178e-06,
      "loss": 0.0037,
      "step": 53010
    },
    {
      "epoch": 0.867544792604107,
      "grad_norm": 0.08411154896020889,
      "learning_rate": 8.94885473386639e-06,
      "loss": 0.0034,
      "step": 53020
    },
    {
      "epoch": 0.8677084185551829,
      "grad_norm": 0.10121965408325195,
      "learning_rate": 8.948270740478107e-06,
      "loss": 0.0016,
      "step": 53030
    },
    {
      "epoch": 0.8678720445062587,
      "grad_norm": 0.2577684223651886,
      "learning_rate": 8.947686603974501e-06,
      "loss": 0.0036,
      "step": 53040
    },
    {
      "epoch": 0.8680356704573345,
      "grad_norm": 0.153594508767128,
      "learning_rate": 8.947102324376742e-06,
      "loss": 0.0037,
      "step": 53050
    },
    {
      "epoch": 0.8681992964084104,
      "grad_norm": 0.20292901992797852,
      "learning_rate": 8.946517901706012e-06,
      "loss": 0.0019,
      "step": 53060
    },
    {
      "epoch": 0.8683629223594862,
      "grad_norm": 0.15858130156993866,
      "learning_rate": 8.945933335983491e-06,
      "loss": 0.0021,
      "step": 53070
    },
    {
      "epoch": 0.8685265483105621,
      "grad_norm": 0.06600574404001236,
      "learning_rate": 8.945348627230373e-06,
      "loss": 0.003,
      "step": 53080
    },
    {
      "epoch": 0.8686901742616379,
      "grad_norm": 0.03988082706928253,
      "learning_rate": 8.944763775467848e-06,
      "loss": 0.0039,
      "step": 53090
    },
    {
      "epoch": 0.8688538002127137,
      "grad_norm": 0.29407060146331787,
      "learning_rate": 8.944178780717118e-06,
      "loss": 0.003,
      "step": 53100
    },
    {
      "epoch": 0.8690174261637896,
      "grad_norm": 0.1616322100162506,
      "learning_rate": 8.943593642999386e-06,
      "loss": 0.0035,
      "step": 53110
    },
    {
      "epoch": 0.8691810521148654,
      "grad_norm": 0.10926707834005356,
      "learning_rate": 8.943008362335864e-06,
      "loss": 0.0025,
      "step": 53120
    },
    {
      "epoch": 0.8693446780659413,
      "grad_norm": 0.1041572317481041,
      "learning_rate": 8.942422938747763e-06,
      "loss": 0.0027,
      "step": 53130
    },
    {
      "epoch": 0.8695083040170171,
      "grad_norm": 0.2496742159128189,
      "learning_rate": 8.941837372256307e-06,
      "loss": 0.0021,
      "step": 53140
    },
    {
      "epoch": 0.869671929968093,
      "grad_norm": 0.13319042325019836,
      "learning_rate": 8.941251662882722e-06,
      "loss": 0.0027,
      "step": 53150
    },
    {
      "epoch": 0.8698355559191688,
      "grad_norm": 0.11574677377939224,
      "learning_rate": 8.940665810648235e-06,
      "loss": 0.0059,
      "step": 53160
    },
    {
      "epoch": 0.8699991818702446,
      "grad_norm": 0.1485525369644165,
      "learning_rate": 8.940079815574083e-06,
      "loss": 0.0031,
      "step": 53170
    },
    {
      "epoch": 0.8701628078213205,
      "grad_norm": 0.25555598735809326,
      "learning_rate": 8.939493677681509e-06,
      "loss": 0.0022,
      "step": 53180
    },
    {
      "epoch": 0.8703264337723963,
      "grad_norm": 0.18181513249874115,
      "learning_rate": 8.938907396991756e-06,
      "loss": 0.0031,
      "step": 53190
    },
    {
      "epoch": 0.8704900597234722,
      "grad_norm": 0.22958390414714813,
      "learning_rate": 8.938320973526078e-06,
      "loss": 0.0037,
      "step": 53200
    },
    {
      "epoch": 0.8706536856745479,
      "grad_norm": 0.16609273850917816,
      "learning_rate": 8.937734407305729e-06,
      "loss": 0.0043,
      "step": 53210
    },
    {
      "epoch": 0.8708173116256238,
      "grad_norm": 0.15709634125232697,
      "learning_rate": 8.937147698351973e-06,
      "loss": 0.0018,
      "step": 53220
    },
    {
      "epoch": 0.8709809375766997,
      "grad_norm": 0.09992536157369614,
      "learning_rate": 8.936560846686076e-06,
      "loss": 0.0061,
      "step": 53230
    },
    {
      "epoch": 0.8711445635277755,
      "grad_norm": 0.2760920822620392,
      "learning_rate": 8.935973852329309e-06,
      "loss": 0.0038,
      "step": 53240
    },
    {
      "epoch": 0.8713081894788514,
      "grad_norm": 0.03906510770320892,
      "learning_rate": 8.93538671530295e-06,
      "loss": 0.0012,
      "step": 53250
    },
    {
      "epoch": 0.8714718154299271,
      "grad_norm": 0.17667658627033234,
      "learning_rate": 8.93479943562828e-06,
      "loss": 0.0026,
      "step": 53260
    },
    {
      "epoch": 0.871635441381003,
      "grad_norm": 0.08896145224571228,
      "learning_rate": 8.934212013326589e-06,
      "loss": 0.0021,
      "step": 53270
    },
    {
      "epoch": 0.8717990673320789,
      "grad_norm": 0.14169903099536896,
      "learning_rate": 8.933624448419165e-06,
      "loss": 0.0034,
      "step": 53280
    },
    {
      "epoch": 0.8719626932831547,
      "grad_norm": 0.2344195544719696,
      "learning_rate": 8.933036740927313e-06,
      "loss": 0.0034,
      "step": 53290
    },
    {
      "epoch": 0.8721263192342306,
      "grad_norm": 0.10642287135124207,
      "learning_rate": 8.932448890872329e-06,
      "loss": 0.003,
      "step": 53300
    },
    {
      "epoch": 0.8722899451853063,
      "grad_norm": 0.39549535512924194,
      "learning_rate": 8.931860898275526e-06,
      "loss": 0.0045,
      "step": 53310
    },
    {
      "epoch": 0.8724535711363822,
      "grad_norm": 0.3656351864337921,
      "learning_rate": 8.931272763158214e-06,
      "loss": 0.0035,
      "step": 53320
    },
    {
      "epoch": 0.8726171970874581,
      "grad_norm": 0.14767798781394958,
      "learning_rate": 8.930684485541713e-06,
      "loss": 0.0053,
      "step": 53330
    },
    {
      "epoch": 0.8727808230385339,
      "grad_norm": 0.24421629309654236,
      "learning_rate": 8.930096065447348e-06,
      "loss": 0.0033,
      "step": 53340
    },
    {
      "epoch": 0.8729444489896098,
      "grad_norm": 0.18194125592708588,
      "learning_rate": 8.929507502896446e-06,
      "loss": 0.0034,
      "step": 53350
    },
    {
      "epoch": 0.8731080749406855,
      "grad_norm": 0.1932399719953537,
      "learning_rate": 8.92891879791034e-06,
      "loss": 0.0019,
      "step": 53360
    },
    {
      "epoch": 0.8732717008917614,
      "grad_norm": 0.17778369784355164,
      "learning_rate": 8.928329950510371e-06,
      "loss": 0.0037,
      "step": 53370
    },
    {
      "epoch": 0.8734353268428373,
      "grad_norm": 0.2531684935092926,
      "learning_rate": 8.927740960717883e-06,
      "loss": 0.0034,
      "step": 53380
    },
    {
      "epoch": 0.8735989527939131,
      "grad_norm": 0.21164102852344513,
      "learning_rate": 8.927151828554226e-06,
      "loss": 0.0081,
      "step": 53390
    },
    {
      "epoch": 0.873762578744989,
      "grad_norm": 0.21142245829105377,
      "learning_rate": 8.926562554040754e-06,
      "loss": 0.0027,
      "step": 53400
    },
    {
      "epoch": 0.8739262046960647,
      "grad_norm": 0.24757084250450134,
      "learning_rate": 8.925973137198824e-06,
      "loss": 0.0027,
      "step": 53410
    },
    {
      "epoch": 0.8740898306471406,
      "grad_norm": 0.21466612815856934,
      "learning_rate": 8.925383578049806e-06,
      "loss": 0.0024,
      "step": 53420
    },
    {
      "epoch": 0.8742534565982165,
      "grad_norm": 0.25179845094680786,
      "learning_rate": 8.924793876615067e-06,
      "loss": 0.0023,
      "step": 53430
    },
    {
      "epoch": 0.8744170825492923,
      "grad_norm": 0.281654953956604,
      "learning_rate": 8.924204032915984e-06,
      "loss": 0.0043,
      "step": 53440
    },
    {
      "epoch": 0.8745807085003682,
      "grad_norm": 0.08162378519773483,
      "learning_rate": 8.923614046973935e-06,
      "loss": 0.002,
      "step": 53450
    },
    {
      "epoch": 0.874744334451444,
      "grad_norm": 0.38746851682662964,
      "learning_rate": 8.923023918810308e-06,
      "loss": 0.0025,
      "step": 53460
    },
    {
      "epoch": 0.8749079604025198,
      "grad_norm": 0.17148450016975403,
      "learning_rate": 8.922433648446491e-06,
      "loss": 0.0052,
      "step": 53470
    },
    {
      "epoch": 0.8750715863535957,
      "grad_norm": 0.25914186239242554,
      "learning_rate": 8.921843235903884e-06,
      "loss": 0.0043,
      "step": 53480
    },
    {
      "epoch": 0.8752352123046715,
      "grad_norm": 0.09895703196525574,
      "learning_rate": 8.921252681203884e-06,
      "loss": 0.0018,
      "step": 53490
    },
    {
      "epoch": 0.8753988382557474,
      "grad_norm": 0.1403970718383789,
      "learning_rate": 8.9206619843679e-06,
      "loss": 0.0032,
      "step": 53500
    },
    {
      "epoch": 0.8755624642068232,
      "grad_norm": 0.2040606439113617,
      "learning_rate": 8.92007114541734e-06,
      "loss": 0.0037,
      "step": 53510
    },
    {
      "epoch": 0.875726090157899,
      "grad_norm": 0.20477795600891113,
      "learning_rate": 8.919480164373623e-06,
      "loss": 0.0023,
      "step": 53520
    },
    {
      "epoch": 0.8758897161089749,
      "grad_norm": 0.47427594661712646,
      "learning_rate": 8.918889041258172e-06,
      "loss": 0.0027,
      "step": 53530
    },
    {
      "epoch": 0.8760533420600507,
      "grad_norm": 0.39713793992996216,
      "learning_rate": 8.918297776092412e-06,
      "loss": 0.0047,
      "step": 53540
    },
    {
      "epoch": 0.8762169680111266,
      "grad_norm": 0.38035979866981506,
      "learning_rate": 8.917706368897774e-06,
      "loss": 0.0026,
      "step": 53550
    },
    {
      "epoch": 0.8763805939622024,
      "grad_norm": 0.29770806431770325,
      "learning_rate": 8.917114819695695e-06,
      "loss": 0.0037,
      "step": 53560
    },
    {
      "epoch": 0.8765442199132782,
      "grad_norm": 0.08524775505065918,
      "learning_rate": 8.91652312850762e-06,
      "loss": 0.0016,
      "step": 53570
    },
    {
      "epoch": 0.8767078458643541,
      "grad_norm": 0.09087630361318588,
      "learning_rate": 8.915931295354995e-06,
      "loss": 0.002,
      "step": 53580
    },
    {
      "epoch": 0.8768714718154299,
      "grad_norm": 0.280171275138855,
      "learning_rate": 8.91533932025927e-06,
      "loss": 0.002,
      "step": 53590
    },
    {
      "epoch": 0.8770350977665058,
      "grad_norm": 0.2526150643825531,
      "learning_rate": 8.914747203241908e-06,
      "loss": 0.0031,
      "step": 53600
    },
    {
      "epoch": 0.8771987237175816,
      "grad_norm": 0.3521006107330322,
      "learning_rate": 8.914154944324367e-06,
      "loss": 0.0033,
      "step": 53610
    },
    {
      "epoch": 0.8773623496686574,
      "grad_norm": 0.16499029099941254,
      "learning_rate": 8.913562543528116e-06,
      "loss": 0.0023,
      "step": 53620
    },
    {
      "epoch": 0.8775259756197333,
      "grad_norm": 0.03129316493868828,
      "learning_rate": 8.91297000087463e-06,
      "loss": 0.0028,
      "step": 53630
    },
    {
      "epoch": 0.8776896015708091,
      "grad_norm": 0.23531906306743622,
      "learning_rate": 8.912377316385386e-06,
      "loss": 0.0024,
      "step": 53640
    },
    {
      "epoch": 0.877853227521885,
      "grad_norm": 0.24655872583389282,
      "learning_rate": 8.911784490081867e-06,
      "loss": 0.0021,
      "step": 53650
    },
    {
      "epoch": 0.8780168534729608,
      "grad_norm": 0.09599456191062927,
      "learning_rate": 8.911191521985562e-06,
      "loss": 0.0029,
      "step": 53660
    },
    {
      "epoch": 0.8781804794240367,
      "grad_norm": 0.3862043023109436,
      "learning_rate": 8.910598412117965e-06,
      "loss": 0.0045,
      "step": 53670
    },
    {
      "epoch": 0.8783441053751125,
      "grad_norm": 0.04599544405937195,
      "learning_rate": 8.910005160500575e-06,
      "loss": 0.0041,
      "step": 53680
    },
    {
      "epoch": 0.8785077313261883,
      "grad_norm": 0.2448827475309372,
      "learning_rate": 8.909411767154894e-06,
      "loss": 0.0039,
      "step": 53690
    },
    {
      "epoch": 0.8786713572772642,
      "grad_norm": 0.15530982613563538,
      "learning_rate": 8.908818232102433e-06,
      "loss": 0.0026,
      "step": 53700
    },
    {
      "epoch": 0.87883498322834,
      "grad_norm": 0.2259899526834488,
      "learning_rate": 8.908224555364707e-06,
      "loss": 0.0028,
      "step": 53710
    },
    {
      "epoch": 0.8789986091794159,
      "grad_norm": 0.20185403525829315,
      "learning_rate": 8.907630736963231e-06,
      "loss": 0.0018,
      "step": 53720
    },
    {
      "epoch": 0.8791622351304917,
      "grad_norm": 0.04519975930452347,
      "learning_rate": 8.907036776919536e-06,
      "loss": 0.0023,
      "step": 53730
    },
    {
      "epoch": 0.8793258610815675,
      "grad_norm": 0.061125755310058594,
      "learning_rate": 8.906442675255145e-06,
      "loss": 0.004,
      "step": 53740
    },
    {
      "epoch": 0.8794894870326434,
      "grad_norm": 0.3056689500808716,
      "learning_rate": 8.905848431991597e-06,
      "loss": 0.0028,
      "step": 53750
    },
    {
      "epoch": 0.8796531129837192,
      "grad_norm": 0.36371874809265137,
      "learning_rate": 8.905254047150431e-06,
      "loss": 0.0048,
      "step": 53760
    },
    {
      "epoch": 0.8798167389347951,
      "grad_norm": 0.03165926784276962,
      "learning_rate": 8.904659520753192e-06,
      "loss": 0.0032,
      "step": 53770
    },
    {
      "epoch": 0.879980364885871,
      "grad_norm": 0.057995691895484924,
      "learning_rate": 8.90406485282143e-06,
      "loss": 0.003,
      "step": 53780
    },
    {
      "epoch": 0.8801439908369467,
      "grad_norm": 0.22391577064990997,
      "learning_rate": 8.903470043376701e-06,
      "loss": 0.0043,
      "step": 53790
    },
    {
      "epoch": 0.8803076167880226,
      "grad_norm": 0.007737902924418449,
      "learning_rate": 8.902875092440564e-06,
      "loss": 0.0015,
      "step": 53800
    },
    {
      "epoch": 0.8804712427390984,
      "grad_norm": 0.09004518389701843,
      "learning_rate": 8.902280000034584e-06,
      "loss": 0.0031,
      "step": 53810
    },
    {
      "epoch": 0.8806348686901743,
      "grad_norm": 0.29592758417129517,
      "learning_rate": 8.901684766180335e-06,
      "loss": 0.0037,
      "step": 53820
    },
    {
      "epoch": 0.8807984946412502,
      "grad_norm": 0.04724936932325363,
      "learning_rate": 8.901089390899388e-06,
      "loss": 0.0027,
      "step": 53830
    },
    {
      "epoch": 0.8809621205923259,
      "grad_norm": 0.14172354340553284,
      "learning_rate": 8.90049387421333e-06,
      "loss": 0.0048,
      "step": 53840
    },
    {
      "epoch": 0.8811257465434018,
      "grad_norm": 0.10313286632299423,
      "learning_rate": 8.89989821614374e-06,
      "loss": 0.0014,
      "step": 53850
    },
    {
      "epoch": 0.8812893724944776,
      "grad_norm": 0.0624067522585392,
      "learning_rate": 8.899302416712217e-06,
      "loss": 0.0023,
      "step": 53860
    },
    {
      "epoch": 0.8814529984455535,
      "grad_norm": 0.07169128954410553,
      "learning_rate": 8.89870647594035e-06,
      "loss": 0.0023,
      "step": 53870
    },
    {
      "epoch": 0.8816166243966294,
      "grad_norm": 0.15577267110347748,
      "learning_rate": 8.898110393849744e-06,
      "loss": 0.0025,
      "step": 53880
    },
    {
      "epoch": 0.8817802503477051,
      "grad_norm": 0.19391897320747375,
      "learning_rate": 8.897514170462005e-06,
      "loss": 0.0031,
      "step": 53890
    },
    {
      "epoch": 0.881943876298781,
      "grad_norm": 0.12142909318208694,
      "learning_rate": 8.896917805798746e-06,
      "loss": 0.0029,
      "step": 53900
    },
    {
      "epoch": 0.8821075022498568,
      "grad_norm": 0.08712892234325409,
      "learning_rate": 8.896321299881582e-06,
      "loss": 0.0036,
      "step": 53910
    },
    {
      "epoch": 0.8822711282009327,
      "grad_norm": 0.03784019127488136,
      "learning_rate": 8.895724652732133e-06,
      "loss": 0.0018,
      "step": 53920
    },
    {
      "epoch": 0.8824347541520086,
      "grad_norm": 0.08243045210838318,
      "learning_rate": 8.895127864372033e-06,
      "loss": 0.0034,
      "step": 53930
    },
    {
      "epoch": 0.8825983801030843,
      "grad_norm": 0.18200337886810303,
      "learning_rate": 8.894530934822908e-06,
      "loss": 0.0043,
      "step": 53940
    },
    {
      "epoch": 0.8827620060541602,
      "grad_norm": 0.11088070273399353,
      "learning_rate": 8.893933864106394e-06,
      "loss": 0.0018,
      "step": 53950
    },
    {
      "epoch": 0.882925632005236,
      "grad_norm": 0.2738510072231293,
      "learning_rate": 8.89333665224414e-06,
      "loss": 0.0029,
      "step": 53960
    },
    {
      "epoch": 0.8830892579563119,
      "grad_norm": 0.11092813313007355,
      "learning_rate": 8.892739299257786e-06,
      "loss": 0.0028,
      "step": 53970
    },
    {
      "epoch": 0.8832528839073878,
      "grad_norm": 0.2351102977991104,
      "learning_rate": 8.892141805168992e-06,
      "loss": 0.0036,
      "step": 53980
    },
    {
      "epoch": 0.8834165098584635,
      "grad_norm": 0.1342463344335556,
      "learning_rate": 8.89154416999941e-06,
      "loss": 0.0027,
      "step": 53990
    },
    {
      "epoch": 0.8835801358095394,
      "grad_norm": 0.07385512441396713,
      "learning_rate": 8.890946393770707e-06,
      "loss": 0.0026,
      "step": 54000
    },
    {
      "epoch": 0.8837437617606152,
      "grad_norm": 0.1501883864402771,
      "learning_rate": 8.890348476504547e-06,
      "loss": 0.0032,
      "step": 54010
    },
    {
      "epoch": 0.8839073877116911,
      "grad_norm": 0.144461527466774,
      "learning_rate": 8.889750418222605e-06,
      "loss": 0.0021,
      "step": 54020
    },
    {
      "epoch": 0.884071013662767,
      "grad_norm": 0.2875502109527588,
      "learning_rate": 8.88915221894656e-06,
      "loss": 0.002,
      "step": 54030
    },
    {
      "epoch": 0.8842346396138427,
      "grad_norm": 0.037042923271656036,
      "learning_rate": 8.888553878698093e-06,
      "loss": 0.0046,
      "step": 54040
    },
    {
      "epoch": 0.8843982655649186,
      "grad_norm": 0.11383754760026932,
      "learning_rate": 8.887955397498895e-06,
      "loss": 0.0045,
      "step": 54050
    },
    {
      "epoch": 0.8845618915159944,
      "grad_norm": 0.4360297620296478,
      "learning_rate": 8.887356775370657e-06,
      "loss": 0.0026,
      "step": 54060
    },
    {
      "epoch": 0.8847255174670703,
      "grad_norm": 0.27691003680229187,
      "learning_rate": 8.88675801233508e-06,
      "loss": 0.0037,
      "step": 54070
    },
    {
      "epoch": 0.8848891434181462,
      "grad_norm": 0.2729971408843994,
      "learning_rate": 8.886159108413865e-06,
      "loss": 0.0026,
      "step": 54080
    },
    {
      "epoch": 0.885052769369222,
      "grad_norm": 0.19147582352161407,
      "learning_rate": 8.885560063628724e-06,
      "loss": 0.0025,
      "step": 54090
    },
    {
      "epoch": 0.8852163953202978,
      "grad_norm": 0.10575056076049805,
      "learning_rate": 8.88496087800137e-06,
      "loss": 0.0043,
      "step": 54100
    },
    {
      "epoch": 0.8853800212713736,
      "grad_norm": 0.17590594291687012,
      "learning_rate": 8.884361551553522e-06,
      "loss": 0.0018,
      "step": 54110
    },
    {
      "epoch": 0.8855436472224495,
      "grad_norm": 0.16460944712162018,
      "learning_rate": 8.883762084306901e-06,
      "loss": 0.003,
      "step": 54120
    },
    {
      "epoch": 0.8857072731735253,
      "grad_norm": 0.24677202105522156,
      "learning_rate": 8.88316247628324e-06,
      "loss": 0.0052,
      "step": 54130
    },
    {
      "epoch": 0.8858708991246012,
      "grad_norm": 0.16806961596012115,
      "learning_rate": 8.882562727504272e-06,
      "loss": 0.0019,
      "step": 54140
    },
    {
      "epoch": 0.886034525075677,
      "grad_norm": 0.3846401870250702,
      "learning_rate": 8.881962837991738e-06,
      "loss": 0.005,
      "step": 54150
    },
    {
      "epoch": 0.8861981510267528,
      "grad_norm": 0.21025578677654266,
      "learning_rate": 8.88136280776738e-06,
      "loss": 0.0023,
      "step": 54160
    },
    {
      "epoch": 0.8863617769778287,
      "grad_norm": 0.19421720504760742,
      "learning_rate": 8.880762636852949e-06,
      "loss": 0.0035,
      "step": 54170
    },
    {
      "epoch": 0.8865254029289045,
      "grad_norm": 0.07441296428442001,
      "learning_rate": 8.880162325270199e-06,
      "loss": 0.0028,
      "step": 54180
    },
    {
      "epoch": 0.8866890288799804,
      "grad_norm": 0.25098496675491333,
      "learning_rate": 8.879561873040891e-06,
      "loss": 0.0045,
      "step": 54190
    },
    {
      "epoch": 0.8868526548310562,
      "grad_norm": 0.20665836334228516,
      "learning_rate": 8.878961280186789e-06,
      "loss": 0.004,
      "step": 54200
    },
    {
      "epoch": 0.887016280782132,
      "grad_norm": 0.053514499217271805,
      "learning_rate": 8.878360546729664e-06,
      "loss": 0.004,
      "step": 54210
    },
    {
      "epoch": 0.8871799067332079,
      "grad_norm": 0.07332173734903336,
      "learning_rate": 8.877759672691291e-06,
      "loss": 0.0031,
      "step": 54220
    },
    {
      "epoch": 0.8873435326842837,
      "grad_norm": 0.23525884747505188,
      "learning_rate": 8.877158658093448e-06,
      "loss": 0.0026,
      "step": 54230
    },
    {
      "epoch": 0.8875071586353596,
      "grad_norm": 0.3173944652080536,
      "learning_rate": 8.876557502957923e-06,
      "loss": 0.0032,
      "step": 54240
    },
    {
      "epoch": 0.8876707845864354,
      "grad_norm": 0.03907003626227379,
      "learning_rate": 8.875956207306507e-06,
      "loss": 0.0025,
      "step": 54250
    },
    {
      "epoch": 0.8878344105375112,
      "grad_norm": 0.21646857261657715,
      "learning_rate": 8.875354771160991e-06,
      "loss": 0.0042,
      "step": 54260
    },
    {
      "epoch": 0.8879980364885871,
      "grad_norm": 0.040403611958026886,
      "learning_rate": 8.874753194543181e-06,
      "loss": 0.0036,
      "step": 54270
    },
    {
      "epoch": 0.8881616624396629,
      "grad_norm": 0.24802431464195251,
      "learning_rate": 8.874151477474881e-06,
      "loss": 0.0029,
      "step": 54280
    },
    {
      "epoch": 0.8883252883907388,
      "grad_norm": 0.18411269783973694,
      "learning_rate": 8.8735496199779e-06,
      "loss": 0.004,
      "step": 54290
    },
    {
      "epoch": 0.8884889143418147,
      "grad_norm": 0.41890931129455566,
      "learning_rate": 8.872947622074056e-06,
      "loss": 0.0029,
      "step": 54300
    },
    {
      "epoch": 0.8886525402928904,
      "grad_norm": 0.161081463098526,
      "learning_rate": 8.872345483785168e-06,
      "loss": 0.0019,
      "step": 54310
    },
    {
      "epoch": 0.8888161662439663,
      "grad_norm": 0.37078016996383667,
      "learning_rate": 8.871743205133063e-06,
      "loss": 0.0048,
      "step": 54320
    },
    {
      "epoch": 0.8889797921950421,
      "grad_norm": 0.28318163752555847,
      "learning_rate": 8.871140786139574e-06,
      "loss": 0.0028,
      "step": 54330
    },
    {
      "epoch": 0.889143418146118,
      "grad_norm": 0.056515686213970184,
      "learning_rate": 8.870538226826534e-06,
      "loss": 0.0022,
      "step": 54340
    },
    {
      "epoch": 0.8893070440971939,
      "grad_norm": 0.18007394671440125,
      "learning_rate": 8.869935527215787e-06,
      "loss": 0.0031,
      "step": 54350
    },
    {
      "epoch": 0.8894706700482696,
      "grad_norm": 0.08672183007001877,
      "learning_rate": 8.869332687329177e-06,
      "loss": 0.0034,
      "step": 54360
    },
    {
      "epoch": 0.8896342959993455,
      "grad_norm": 0.2902868092060089,
      "learning_rate": 8.868729707188558e-06,
      "loss": 0.0038,
      "step": 54370
    },
    {
      "epoch": 0.8897979219504213,
      "grad_norm": 0.03184123337268829,
      "learning_rate": 8.868126586815785e-06,
      "loss": 0.0031,
      "step": 54380
    },
    {
      "epoch": 0.8899615479014972,
      "grad_norm": 0.42973706126213074,
      "learning_rate": 8.86752332623272e-06,
      "loss": 0.0058,
      "step": 54390
    },
    {
      "epoch": 0.8901251738525731,
      "grad_norm": 0.14185434579849243,
      "learning_rate": 8.866919925461229e-06,
      "loss": 0.0022,
      "step": 54400
    },
    {
      "epoch": 0.8902887998036488,
      "grad_norm": 0.2531667649745941,
      "learning_rate": 8.866316384523185e-06,
      "loss": 0.0027,
      "step": 54410
    },
    {
      "epoch": 0.8904524257547247,
      "grad_norm": 0.16212552785873413,
      "learning_rate": 8.865712703440464e-06,
      "loss": 0.0043,
      "step": 54420
    },
    {
      "epoch": 0.8906160517058005,
      "grad_norm": 0.1044308990240097,
      "learning_rate": 8.86510888223495e-06,
      "loss": 0.0027,
      "step": 54430
    },
    {
      "epoch": 0.8907796776568764,
      "grad_norm": 0.05360344052314758,
      "learning_rate": 8.864504920928527e-06,
      "loss": 0.0041,
      "step": 54440
    },
    {
      "epoch": 0.8909433036079523,
      "grad_norm": 0.05559851601719856,
      "learning_rate": 8.86390081954309e-06,
      "loss": 0.0015,
      "step": 54450
    },
    {
      "epoch": 0.891106929559028,
      "grad_norm": 0.8955914974212646,
      "learning_rate": 8.863296578100535e-06,
      "loss": 0.0043,
      "step": 54460
    },
    {
      "epoch": 0.8912705555101039,
      "grad_norm": 0.017737558111548424,
      "learning_rate": 8.862692196622761e-06,
      "loss": 0.0028,
      "step": 54470
    },
    {
      "epoch": 0.8914341814611797,
      "grad_norm": 0.41233164072036743,
      "learning_rate": 8.862087675131684e-06,
      "loss": 0.0066,
      "step": 54480
    },
    {
      "epoch": 0.8915978074122556,
      "grad_norm": 0.06687049567699432,
      "learning_rate": 8.861483013649207e-06,
      "loss": 0.0037,
      "step": 54490
    },
    {
      "epoch": 0.8917614333633315,
      "grad_norm": 0.19879239797592163,
      "learning_rate": 8.860878212197252e-06,
      "loss": 0.0055,
      "step": 54500
    },
    {
      "epoch": 0.8919250593144072,
      "grad_norm": 0.27851152420043945,
      "learning_rate": 8.860273270797743e-06,
      "loss": 0.0031,
      "step": 54510
    },
    {
      "epoch": 0.8920886852654831,
      "grad_norm": 0.2351270616054535,
      "learning_rate": 8.859668189472605e-06,
      "loss": 0.0026,
      "step": 54520
    },
    {
      "epoch": 0.8922523112165589,
      "grad_norm": 0.14612500369548798,
      "learning_rate": 8.859062968243773e-06,
      "loss": 0.0024,
      "step": 54530
    },
    {
      "epoch": 0.8924159371676348,
      "grad_norm": 0.11215360462665558,
      "learning_rate": 8.85845760713318e-06,
      "loss": 0.0026,
      "step": 54540
    },
    {
      "epoch": 0.8925795631187107,
      "grad_norm": 0.35958877205848694,
      "learning_rate": 8.857852106162774e-06,
      "loss": 0.0049,
      "step": 54550
    },
    {
      "epoch": 0.8927431890697864,
      "grad_norm": 0.08596529811620712,
      "learning_rate": 8.857246465354502e-06,
      "loss": 0.0037,
      "step": 54560
    },
    {
      "epoch": 0.8929068150208623,
      "grad_norm": 0.14215393364429474,
      "learning_rate": 8.856640684730315e-06,
      "loss": 0.0035,
      "step": 54570
    },
    {
      "epoch": 0.8930704409719381,
      "grad_norm": 0.22472992539405823,
      "learning_rate": 8.856034764312173e-06,
      "loss": 0.0023,
      "step": 54580
    },
    {
      "epoch": 0.893234066923014,
      "grad_norm": 0.09880702942609787,
      "learning_rate": 8.85542870412204e-06,
      "loss": 0.0026,
      "step": 54590
    },
    {
      "epoch": 0.8933976928740899,
      "grad_norm": 0.13637830317020416,
      "learning_rate": 8.854822504181881e-06,
      "loss": 0.0028,
      "step": 54600
    },
    {
      "epoch": 0.8935613188251657,
      "grad_norm": 0.14674554765224457,
      "learning_rate": 8.854216164513671e-06,
      "loss": 0.0046,
      "step": 54610
    },
    {
      "epoch": 0.8937249447762415,
      "grad_norm": 0.05734406039118767,
      "learning_rate": 8.85360968513939e-06,
      "loss": 0.0034,
      "step": 54620
    },
    {
      "epoch": 0.8938885707273173,
      "grad_norm": 0.09534914791584015,
      "learning_rate": 8.85300306608102e-06,
      "loss": 0.0023,
      "step": 54630
    },
    {
      "epoch": 0.8940521966783932,
      "grad_norm": 0.02331485226750374,
      "learning_rate": 8.852396307360548e-06,
      "loss": 0.0024,
      "step": 54640
    },
    {
      "epoch": 0.8942158226294691,
      "grad_norm": 0.22868740558624268,
      "learning_rate": 8.851789408999968e-06,
      "loss": 0.0055,
      "step": 54650
    },
    {
      "epoch": 0.8943794485805449,
      "grad_norm": 0.1800401657819748,
      "learning_rate": 8.851182371021278e-06,
      "loss": 0.003,
      "step": 54660
    },
    {
      "epoch": 0.8945430745316207,
      "grad_norm": 1.7709912061691284,
      "learning_rate": 8.850575193446487e-06,
      "loss": 0.0047,
      "step": 54670
    },
    {
      "epoch": 0.8947067004826965,
      "grad_norm": 0.1639859825372696,
      "learning_rate": 8.849967876297598e-06,
      "loss": 0.0017,
      "step": 54680
    },
    {
      "epoch": 0.8948703264337724,
      "grad_norm": 0.06918571144342422,
      "learning_rate": 8.849360419596626e-06,
      "loss": 0.0022,
      "step": 54690
    },
    {
      "epoch": 0.8950339523848483,
      "grad_norm": 0.11224681884050369,
      "learning_rate": 8.848752823365593e-06,
      "loss": 0.0035,
      "step": 54700
    },
    {
      "epoch": 0.8951975783359241,
      "grad_norm": 0.10752326250076294,
      "learning_rate": 8.848145087626518e-06,
      "loss": 0.004,
      "step": 54710
    },
    {
      "epoch": 0.895361204287,
      "grad_norm": 0.29402297735214233,
      "learning_rate": 8.847537212401432e-06,
      "loss": 0.0034,
      "step": 54720
    },
    {
      "epoch": 0.8955248302380757,
      "grad_norm": 0.10052228718996048,
      "learning_rate": 8.846929197712372e-06,
      "loss": 0.0033,
      "step": 54730
    },
    {
      "epoch": 0.8956884561891516,
      "grad_norm": 0.08744067698717117,
      "learning_rate": 8.846321043581373e-06,
      "loss": 0.0024,
      "step": 54740
    },
    {
      "epoch": 0.8958520821402275,
      "grad_norm": 0.08275303244590759,
      "learning_rate": 8.84571275003048e-06,
      "loss": 0.0025,
      "step": 54750
    },
    {
      "epoch": 0.8960157080913033,
      "grad_norm": 0.585598349571228,
      "learning_rate": 8.845104317081743e-06,
      "loss": 0.0034,
      "step": 54760
    },
    {
      "epoch": 0.8961793340423791,
      "grad_norm": 0.13055624067783356,
      "learning_rate": 8.844495744757215e-06,
      "loss": 0.0026,
      "step": 54770
    },
    {
      "epoch": 0.8963429599934549,
      "grad_norm": 0.11452317982912064,
      "learning_rate": 8.843887033078959e-06,
      "loss": 0.0041,
      "step": 54780
    },
    {
      "epoch": 0.8965065859445308,
      "grad_norm": 0.1353876143693924,
      "learning_rate": 8.843278182069034e-06,
      "loss": 0.0028,
      "step": 54790
    },
    {
      "epoch": 0.8966702118956067,
      "grad_norm": 0.15332217514514923,
      "learning_rate": 8.842669191749514e-06,
      "loss": 0.0031,
      "step": 54800
    },
    {
      "epoch": 0.8968338378466825,
      "grad_norm": 0.09307501465082169,
      "learning_rate": 8.84206006214247e-06,
      "loss": 0.0029,
      "step": 54810
    },
    {
      "epoch": 0.8969974637977584,
      "grad_norm": 0.04865473136305809,
      "learning_rate": 8.841450793269984e-06,
      "loss": 0.0024,
      "step": 54820
    },
    {
      "epoch": 0.8971610897488341,
      "grad_norm": 0.027074819430708885,
      "learning_rate": 8.840841385154137e-06,
      "loss": 0.0023,
      "step": 54830
    },
    {
      "epoch": 0.89732471569991,
      "grad_norm": 0.16106519103050232,
      "learning_rate": 8.840231837817024e-06,
      "loss": 0.0026,
      "step": 54840
    },
    {
      "epoch": 0.8974883416509859,
      "grad_norm": 0.29977506399154663,
      "learning_rate": 8.839622151280736e-06,
      "loss": 0.0045,
      "step": 54850
    },
    {
      "epoch": 0.8976519676020617,
      "grad_norm": 0.15003938972949982,
      "learning_rate": 8.839012325567372e-06,
      "loss": 0.0053,
      "step": 54860
    },
    {
      "epoch": 0.8978155935531376,
      "grad_norm": 0.06466463953256607,
      "learning_rate": 8.838402360699039e-06,
      "loss": 0.0025,
      "step": 54870
    },
    {
      "epoch": 0.8979792195042133,
      "grad_norm": 0.15948380529880524,
      "learning_rate": 8.837792256697846e-06,
      "loss": 0.0044,
      "step": 54880
    },
    {
      "epoch": 0.8981428454552892,
      "grad_norm": 0.2415420114994049,
      "learning_rate": 8.837182013585907e-06,
      "loss": 0.0023,
      "step": 54890
    },
    {
      "epoch": 0.8983064714063651,
      "grad_norm": 0.19046470522880554,
      "learning_rate": 8.836571631385342e-06,
      "loss": 0.0021,
      "step": 54900
    },
    {
      "epoch": 0.8984700973574409,
      "grad_norm": 0.20351184904575348,
      "learning_rate": 8.835961110118276e-06,
      "loss": 0.0028,
      "step": 54910
    },
    {
      "epoch": 0.8986337233085168,
      "grad_norm": 0.14510388672351837,
      "learning_rate": 8.835350449806842e-06,
      "loss": 0.0025,
      "step": 54920
    },
    {
      "epoch": 0.8987973492595925,
      "grad_norm": 0.17106325924396515,
      "learning_rate": 8.834739650473169e-06,
      "loss": 0.0044,
      "step": 54930
    },
    {
      "epoch": 0.8989609752106684,
      "grad_norm": 0.27555814385414124,
      "learning_rate": 8.834128712139402e-06,
      "loss": 0.0031,
      "step": 54940
    },
    {
      "epoch": 0.8991246011617443,
      "grad_norm": 0.14989875257015228,
      "learning_rate": 8.833517634827683e-06,
      "loss": 0.0043,
      "step": 54950
    },
    {
      "epoch": 0.8992882271128201,
      "grad_norm": 0.22055450081825256,
      "learning_rate": 8.832906418560165e-06,
      "loss": 0.0016,
      "step": 54960
    },
    {
      "epoch": 0.899451853063896,
      "grad_norm": 0.41166171431541443,
      "learning_rate": 8.832295063359e-06,
      "loss": 0.0037,
      "step": 54970
    },
    {
      "epoch": 0.8996154790149717,
      "grad_norm": 0.19088685512542725,
      "learning_rate": 8.83168356924635e-06,
      "loss": 0.0027,
      "step": 54980
    },
    {
      "epoch": 0.8997791049660476,
      "grad_norm": 0.10413893312215805,
      "learning_rate": 8.83107193624438e-06,
      "loss": 0.0034,
      "step": 54990
    },
    {
      "epoch": 0.8999427309171234,
      "grad_norm": 0.20757974684238434,
      "learning_rate": 8.830460164375261e-06,
      "loss": 0.0028,
      "step": 55000
    },
    {
      "epoch": 0.9001063568681993,
      "grad_norm": 0.09585452079772949,
      "learning_rate": 8.829848253661166e-06,
      "loss": 0.0031,
      "step": 55010
    },
    {
      "epoch": 0.9002699828192752,
      "grad_norm": 0.07869952917098999,
      "learning_rate": 8.829236204124276e-06,
      "loss": 0.005,
      "step": 55020
    },
    {
      "epoch": 0.900433608770351,
      "grad_norm": 0.16095200181007385,
      "learning_rate": 8.82862401578678e-06,
      "loss": 0.0033,
      "step": 55030
    },
    {
      "epoch": 0.9005972347214268,
      "grad_norm": 0.17637254297733307,
      "learning_rate": 8.828011688670862e-06,
      "loss": 0.0026,
      "step": 55040
    },
    {
      "epoch": 0.9007608606725026,
      "grad_norm": 0.1566551923751831,
      "learning_rate": 8.827399222798722e-06,
      "loss": 0.003,
      "step": 55050
    },
    {
      "epoch": 0.9009244866235785,
      "grad_norm": 0.2678000330924988,
      "learning_rate": 8.82678661819256e-06,
      "loss": 0.0036,
      "step": 55060
    },
    {
      "epoch": 0.9010881125746544,
      "grad_norm": 0.12816756963729858,
      "learning_rate": 8.826173874874578e-06,
      "loss": 0.0027,
      "step": 55070
    },
    {
      "epoch": 0.9012517385257302,
      "grad_norm": 0.021010136231780052,
      "learning_rate": 8.82556099286699e-06,
      "loss": 0.0059,
      "step": 55080
    },
    {
      "epoch": 0.901415364476806,
      "grad_norm": 0.07008292526006699,
      "learning_rate": 8.824947972192012e-06,
      "loss": 0.0025,
      "step": 55090
    },
    {
      "epoch": 0.9015789904278818,
      "grad_norm": 0.2838267982006073,
      "learning_rate": 8.824334812871862e-06,
      "loss": 0.0033,
      "step": 55100
    },
    {
      "epoch": 0.9017426163789577,
      "grad_norm": 0.17677299678325653,
      "learning_rate": 8.823721514928766e-06,
      "loss": 0.0033,
      "step": 55110
    },
    {
      "epoch": 0.9019062423300336,
      "grad_norm": 0.11584359407424927,
      "learning_rate": 8.823108078384956e-06,
      "loss": 0.0034,
      "step": 55120
    },
    {
      "epoch": 0.9020698682811094,
      "grad_norm": 0.11494273692369461,
      "learning_rate": 8.822494503262668e-06,
      "loss": 0.0028,
      "step": 55130
    },
    {
      "epoch": 0.9022334942321852,
      "grad_norm": 0.12606169283390045,
      "learning_rate": 8.821880789584138e-06,
      "loss": 0.0033,
      "step": 55140
    },
    {
      "epoch": 0.902397120183261,
      "grad_norm": 0.124476857483387,
      "learning_rate": 8.821266937371618e-06,
      "loss": 0.0025,
      "step": 55150
    },
    {
      "epoch": 0.9025607461343369,
      "grad_norm": 0.06272819638252258,
      "learning_rate": 8.820652946647355e-06,
      "loss": 0.0031,
      "step": 55160
    },
    {
      "epoch": 0.9027243720854128,
      "grad_norm": 0.2096959799528122,
      "learning_rate": 8.820038817433605e-06,
      "loss": 0.0024,
      "step": 55170
    },
    {
      "epoch": 0.9028879980364886,
      "grad_norm": 0.09812069684267044,
      "learning_rate": 8.81942454975263e-06,
      "loss": 0.0031,
      "step": 55180
    },
    {
      "epoch": 0.9030516239875644,
      "grad_norm": 0.04358232766389847,
      "learning_rate": 8.818810143626693e-06,
      "loss": 0.0031,
      "step": 55190
    },
    {
      "epoch": 0.9032152499386402,
      "grad_norm": 0.08749408274888992,
      "learning_rate": 8.818195599078068e-06,
      "loss": 0.0029,
      "step": 55200
    },
    {
      "epoch": 0.9033788758897161,
      "grad_norm": 0.056532811373472214,
      "learning_rate": 8.817580916129028e-06,
      "loss": 0.0022,
      "step": 55210
    },
    {
      "epoch": 0.903542501840792,
      "grad_norm": 0.14252284169197083,
      "learning_rate": 8.816966094801856e-06,
      "loss": 0.0031,
      "step": 55220
    },
    {
      "epoch": 0.9037061277918678,
      "grad_norm": 0.417191743850708,
      "learning_rate": 8.816351135118836e-06,
      "loss": 0.0049,
      "step": 55230
    },
    {
      "epoch": 0.9038697537429436,
      "grad_norm": 0.11611729860305786,
      "learning_rate": 8.815736037102262e-06,
      "loss": 0.0023,
      "step": 55240
    },
    {
      "epoch": 0.9040333796940194,
      "grad_norm": 0.06029880419373512,
      "learning_rate": 8.815120800774426e-06,
      "loss": 0.0045,
      "step": 55250
    },
    {
      "epoch": 0.9041970056450953,
      "grad_norm": 0.11495212465524673,
      "learning_rate": 8.81450542615763e-06,
      "loss": 0.0035,
      "step": 55260
    },
    {
      "epoch": 0.9043606315961712,
      "grad_norm": 0.17255397140979767,
      "learning_rate": 8.81388991327418e-06,
      "loss": 0.0033,
      "step": 55270
    },
    {
      "epoch": 0.904524257547247,
      "grad_norm": 0.16232207417488098,
      "learning_rate": 8.81327426214639e-06,
      "loss": 0.0022,
      "step": 55280
    },
    {
      "epoch": 0.9046878834983229,
      "grad_norm": 0.10415556281805038,
      "learning_rate": 8.812658472796569e-06,
      "loss": 0.0021,
      "step": 55290
    },
    {
      "epoch": 0.9048515094493986,
      "grad_norm": 0.7202304601669312,
      "learning_rate": 8.812042545247044e-06,
      "loss": 0.002,
      "step": 55300
    },
    {
      "epoch": 0.9050151354004745,
      "grad_norm": 0.07775498926639557,
      "learning_rate": 8.811426479520138e-06,
      "loss": 0.0038,
      "step": 55310
    },
    {
      "epoch": 0.9051787613515504,
      "grad_norm": 0.009761802852153778,
      "learning_rate": 8.810810275638183e-06,
      "loss": 0.0041,
      "step": 55320
    },
    {
      "epoch": 0.9053423873026262,
      "grad_norm": 0.1412675529718399,
      "learning_rate": 8.810193933623514e-06,
      "loss": 0.0037,
      "step": 55330
    },
    {
      "epoch": 0.9055060132537021,
      "grad_norm": 0.17761661112308502,
      "learning_rate": 8.809577453498474e-06,
      "loss": 0.0033,
      "step": 55340
    },
    {
      "epoch": 0.9056696392047778,
      "grad_norm": 0.16839146614074707,
      "learning_rate": 8.808960835285405e-06,
      "loss": 0.0021,
      "step": 55350
    },
    {
      "epoch": 0.9058332651558537,
      "grad_norm": 0.02789127826690674,
      "learning_rate": 8.808344079006663e-06,
      "loss": 0.0017,
      "step": 55360
    },
    {
      "epoch": 0.9059968911069296,
      "grad_norm": 0.08321288973093033,
      "learning_rate": 8.8077271846846e-06,
      "loss": 0.0017,
      "step": 55370
    },
    {
      "epoch": 0.9061605170580054,
      "grad_norm": 0.1785695105791092,
      "learning_rate": 8.807110152341577e-06,
      "loss": 0.0042,
      "step": 55380
    },
    {
      "epoch": 0.9063241430090813,
      "grad_norm": 0.2165706306695938,
      "learning_rate": 8.806492981999964e-06,
      "loss": 0.0029,
      "step": 55390
    },
    {
      "epoch": 0.906487768960157,
      "grad_norm": 0.33078229427337646,
      "learning_rate": 8.805875673682127e-06,
      "loss": 0.0038,
      "step": 55400
    },
    {
      "epoch": 0.9066513949112329,
      "grad_norm": 0.052473850548267365,
      "learning_rate": 8.805258227410444e-06,
      "loss": 0.0023,
      "step": 55410
    },
    {
      "epoch": 0.9068150208623088,
      "grad_norm": 0.08056936413049698,
      "learning_rate": 8.804640643207296e-06,
      "loss": 0.0018,
      "step": 55420
    },
    {
      "epoch": 0.9069786468133846,
      "grad_norm": 0.055580999702215195,
      "learning_rate": 8.80402292109507e-06,
      "loss": 0.0057,
      "step": 55430
    },
    {
      "epoch": 0.9071422727644605,
      "grad_norm": 0.217872753739357,
      "learning_rate": 8.803405061096154e-06,
      "loss": 0.0039,
      "step": 55440
    },
    {
      "epoch": 0.9073058987155362,
      "grad_norm": 0.039096899330616,
      "learning_rate": 8.802787063232947e-06,
      "loss": 0.0031,
      "step": 55450
    },
    {
      "epoch": 0.9074695246666121,
      "grad_norm": 0.1173924058675766,
      "learning_rate": 8.80216892752785e-06,
      "loss": 0.0034,
      "step": 55460
    },
    {
      "epoch": 0.907633150617688,
      "grad_norm": 0.0818713903427124,
      "learning_rate": 8.801550654003263e-06,
      "loss": 0.0023,
      "step": 55470
    },
    {
      "epoch": 0.9077967765687638,
      "grad_norm": 0.13284772634506226,
      "learning_rate": 8.800932242681605e-06,
      "loss": 0.0027,
      "step": 55480
    },
    {
      "epoch": 0.9079604025198397,
      "grad_norm": 0.05722169205546379,
      "learning_rate": 8.80031369358529e-06,
      "loss": 0.0019,
      "step": 55490
    },
    {
      "epoch": 0.9081240284709154,
      "grad_norm": 0.07156434655189514,
      "learning_rate": 8.799695006736735e-06,
      "loss": 0.0033,
      "step": 55500
    },
    {
      "epoch": 0.9082876544219913,
      "grad_norm": 0.18691639602184296,
      "learning_rate": 8.799076182158368e-06,
      "loss": 0.0048,
      "step": 55510
    },
    {
      "epoch": 0.9084512803730672,
      "grad_norm": 0.10891542583703995,
      "learning_rate": 8.798457219872622e-06,
      "loss": 0.0021,
      "step": 55520
    },
    {
      "epoch": 0.908614906324143,
      "grad_norm": 0.4028400480747223,
      "learning_rate": 8.797838119901931e-06,
      "loss": 0.004,
      "step": 55530
    },
    {
      "epoch": 0.9087785322752189,
      "grad_norm": 0.38665276765823364,
      "learning_rate": 8.797218882268734e-06,
      "loss": 0.0046,
      "step": 55540
    },
    {
      "epoch": 0.9089421582262946,
      "grad_norm": 0.09540103375911713,
      "learning_rate": 8.796599506995481e-06,
      "loss": 0.0021,
      "step": 55550
    },
    {
      "epoch": 0.9091057841773705,
      "grad_norm": 0.04197270795702934,
      "learning_rate": 8.795979994104619e-06,
      "loss": 0.0035,
      "step": 55560
    },
    {
      "epoch": 0.9092694101284464,
      "grad_norm": 0.0794285386800766,
      "learning_rate": 8.795360343618608e-06,
      "loss": 0.002,
      "step": 55570
    },
    {
      "epoch": 0.9094330360795222,
      "grad_norm": 0.16556252539157867,
      "learning_rate": 8.794740555559905e-06,
      "loss": 0.0026,
      "step": 55580
    },
    {
      "epoch": 0.9095966620305981,
      "grad_norm": 0.05589031055569649,
      "learning_rate": 8.794120629950977e-06,
      "loss": 0.0023,
      "step": 55590
    },
    {
      "epoch": 0.9097602879816739,
      "grad_norm": 0.2827475070953369,
      "learning_rate": 8.793500566814295e-06,
      "loss": 0.0042,
      "step": 55600
    },
    {
      "epoch": 0.9099239139327497,
      "grad_norm": 0.12089551985263824,
      "learning_rate": 8.792880366172337e-06,
      "loss": 0.0029,
      "step": 55610
    },
    {
      "epoch": 0.9100875398838256,
      "grad_norm": 0.05288970470428467,
      "learning_rate": 8.79226002804758e-06,
      "loss": 0.0024,
      "step": 55620
    },
    {
      "epoch": 0.9102511658349014,
      "grad_norm": 0.014621092937886715,
      "learning_rate": 8.791639552462509e-06,
      "loss": 0.0027,
      "step": 55630
    },
    {
      "epoch": 0.9104147917859773,
      "grad_norm": 0.07463811337947845,
      "learning_rate": 8.791018939439622e-06,
      "loss": 0.002,
      "step": 55640
    },
    {
      "epoch": 0.9105784177370531,
      "grad_norm": 0.08513284474611282,
      "learning_rate": 8.790398189001407e-06,
      "loss": 0.0025,
      "step": 55650
    },
    {
      "epoch": 0.9107420436881289,
      "grad_norm": 0.044549815356731415,
      "learning_rate": 8.789777301170368e-06,
      "loss": 0.0032,
      "step": 55660
    },
    {
      "epoch": 0.9109056696392048,
      "grad_norm": 0.11088958382606506,
      "learning_rate": 8.789156275969011e-06,
      "loss": 0.0032,
      "step": 55670
    },
    {
      "epoch": 0.9110692955902806,
      "grad_norm": 0.12103458493947983,
      "learning_rate": 8.788535113419844e-06,
      "loss": 0.0029,
      "step": 55680
    },
    {
      "epoch": 0.9112329215413565,
      "grad_norm": 0.19530555605888367,
      "learning_rate": 8.787913813545388e-06,
      "loss": 0.0047,
      "step": 55690
    },
    {
      "epoch": 0.9113965474924323,
      "grad_norm": 0.07805991917848587,
      "learning_rate": 8.787292376368158e-06,
      "loss": 0.0014,
      "step": 55700
    },
    {
      "epoch": 0.9115601734435081,
      "grad_norm": 0.04136421158909798,
      "learning_rate": 8.786670801910682e-06,
      "loss": 0.0021,
      "step": 55710
    },
    {
      "epoch": 0.911723799394584,
      "grad_norm": 0.07676675170660019,
      "learning_rate": 8.786049090195491e-06,
      "loss": 0.0016,
      "step": 55720
    },
    {
      "epoch": 0.9118874253456598,
      "grad_norm": 0.11405820399522781,
      "learning_rate": 8.785427241245118e-06,
      "loss": 0.0021,
      "step": 55730
    },
    {
      "epoch": 0.9120510512967357,
      "grad_norm": 0.16153748333454132,
      "learning_rate": 8.78480525508211e-06,
      "loss": 0.0037,
      "step": 55740
    },
    {
      "epoch": 0.9122146772478115,
      "grad_norm": 0.07060796022415161,
      "learning_rate": 8.784183131729005e-06,
      "loss": 0.0021,
      "step": 55750
    },
    {
      "epoch": 0.9123783031988874,
      "grad_norm": 0.2343321293592453,
      "learning_rate": 8.783560871208356e-06,
      "loss": 0.0032,
      "step": 55760
    },
    {
      "epoch": 0.9125419291499632,
      "grad_norm": 0.030766038224101067,
      "learning_rate": 8.78293847354272e-06,
      "loss": 0.0018,
      "step": 55770
    },
    {
      "epoch": 0.912705555101039,
      "grad_norm": 0.14081358909606934,
      "learning_rate": 8.782315938754657e-06,
      "loss": 0.0026,
      "step": 55780
    },
    {
      "epoch": 0.9128691810521149,
      "grad_norm": 0.029263710603117943,
      "learning_rate": 8.78169326686673e-06,
      "loss": 0.0018,
      "step": 55790
    },
    {
      "epoch": 0.9130328070031907,
      "grad_norm": 0.13216765224933624,
      "learning_rate": 8.781070457901513e-06,
      "loss": 0.0031,
      "step": 55800
    },
    {
      "epoch": 0.9131964329542666,
      "grad_norm": 0.06182245910167694,
      "learning_rate": 8.780447511881578e-06,
      "loss": 0.0034,
      "step": 55810
    },
    {
      "epoch": 0.9133600589053424,
      "grad_norm": 0.051942311227321625,
      "learning_rate": 8.779824428829507e-06,
      "loss": 0.0038,
      "step": 55820
    },
    {
      "epoch": 0.9135236848564182,
      "grad_norm": 0.13555654883384705,
      "learning_rate": 8.779201208767885e-06,
      "loss": 0.0031,
      "step": 55830
    },
    {
      "epoch": 0.9136873108074941,
      "grad_norm": 0.18750157952308655,
      "learning_rate": 8.778577851719302e-06,
      "loss": 0.0028,
      "step": 55840
    },
    {
      "epoch": 0.9138509367585699,
      "grad_norm": 0.2095109522342682,
      "learning_rate": 8.777954357706356e-06,
      "loss": 0.0042,
      "step": 55850
    },
    {
      "epoch": 0.9140145627096458,
      "grad_norm": 0.18244940042495728,
      "learning_rate": 8.777330726751643e-06,
      "loss": 0.003,
      "step": 55860
    },
    {
      "epoch": 0.9141781886607215,
      "grad_norm": 0.32803255319595337,
      "learning_rate": 8.77670695887777e-06,
      "loss": 0.0028,
      "step": 55870
    },
    {
      "epoch": 0.9143418146117974,
      "grad_norm": 0.09819483757019043,
      "learning_rate": 8.776083054107346e-06,
      "loss": 0.003,
      "step": 55880
    },
    {
      "epoch": 0.9145054405628733,
      "grad_norm": 0.2654581665992737,
      "learning_rate": 8.775459012462986e-06,
      "loss": 0.0035,
      "step": 55890
    },
    {
      "epoch": 0.9146690665139491,
      "grad_norm": 0.08447451889514923,
      "learning_rate": 8.774834833967312e-06,
      "loss": 0.0023,
      "step": 55900
    },
    {
      "epoch": 0.914832692465025,
      "grad_norm": 0.16487756371498108,
      "learning_rate": 8.774210518642948e-06,
      "loss": 0.0037,
      "step": 55910
    },
    {
      "epoch": 0.9149963184161007,
      "grad_norm": 0.1917775273323059,
      "learning_rate": 8.773586066512524e-06,
      "loss": 0.0026,
      "step": 55920
    },
    {
      "epoch": 0.9151599443671766,
      "grad_norm": 0.1459125280380249,
      "learning_rate": 8.772961477598674e-06,
      "loss": 0.002,
      "step": 55930
    },
    {
      "epoch": 0.9153235703182525,
      "grad_norm": 0.44084975123405457,
      "learning_rate": 8.772336751924038e-06,
      "loss": 0.0019,
      "step": 55940
    },
    {
      "epoch": 0.9154871962693283,
      "grad_norm": 0.15338511765003204,
      "learning_rate": 8.771711889511263e-06,
      "loss": 0.0026,
      "step": 55950
    },
    {
      "epoch": 0.9156508222204042,
      "grad_norm": 0.20346993207931519,
      "learning_rate": 8.771086890382996e-06,
      "loss": 0.0046,
      "step": 55960
    },
    {
      "epoch": 0.91581444817148,
      "grad_norm": 0.3131324350833893,
      "learning_rate": 8.770461754561891e-06,
      "loss": 0.0038,
      "step": 55970
    },
    {
      "epoch": 0.9159780741225558,
      "grad_norm": 0.1511867791414261,
      "learning_rate": 8.769836482070613e-06,
      "loss": 0.005,
      "step": 55980
    },
    {
      "epoch": 0.9161417000736317,
      "grad_norm": 0.27550679445266724,
      "learning_rate": 8.769211072931823e-06,
      "loss": 0.0042,
      "step": 55990
    },
    {
      "epoch": 0.9163053260247075,
      "grad_norm": 0.08323934674263,
      "learning_rate": 8.768585527168187e-06,
      "loss": 0.0019,
      "step": 56000
    },
    {
      "epoch": 0.9164689519757834,
      "grad_norm": 0.3180956542491913,
      "learning_rate": 8.767959844802387e-06,
      "loss": 0.0043,
      "step": 56010
    },
    {
      "epoch": 0.9166325779268591,
      "grad_norm": 0.04906865581870079,
      "learning_rate": 8.767334025857097e-06,
      "loss": 0.0037,
      "step": 56020
    },
    {
      "epoch": 0.916796203877935,
      "grad_norm": 0.12980033457279205,
      "learning_rate": 8.766708070355003e-06,
      "loss": 0.002,
      "step": 56030
    },
    {
      "epoch": 0.9169598298290109,
      "grad_norm": 0.3076556324958801,
      "learning_rate": 8.766081978318794e-06,
      "loss": 0.0063,
      "step": 56040
    },
    {
      "epoch": 0.9171234557800867,
      "grad_norm": 0.21607230603694916,
      "learning_rate": 8.765455749771166e-06,
      "loss": 0.0033,
      "step": 56050
    },
    {
      "epoch": 0.9172870817311626,
      "grad_norm": 0.13356612622737885,
      "learning_rate": 8.764829384734817e-06,
      "loss": 0.0038,
      "step": 56060
    },
    {
      "epoch": 0.9174507076822384,
      "grad_norm": 0.08096982538700104,
      "learning_rate": 8.76420288323245e-06,
      "loss": 0.0032,
      "step": 56070
    },
    {
      "epoch": 0.9176143336333142,
      "grad_norm": 0.19389745593070984,
      "learning_rate": 8.763576245286777e-06,
      "loss": 0.0026,
      "step": 56080
    },
    {
      "epoch": 0.9177779595843901,
      "grad_norm": 0.08451075851917267,
      "learning_rate": 8.76294947092051e-06,
      "loss": 0.0031,
      "step": 56090
    },
    {
      "epoch": 0.9179415855354659,
      "grad_norm": 0.07520750164985657,
      "learning_rate": 8.762322560156369e-06,
      "loss": 0.0026,
      "step": 56100
    },
    {
      "epoch": 0.9181052114865418,
      "grad_norm": 0.16138632595539093,
      "learning_rate": 8.761695513017077e-06,
      "loss": 0.0046,
      "step": 56110
    },
    {
      "epoch": 0.9182688374376176,
      "grad_norm": 0.15444326400756836,
      "learning_rate": 8.761068329525363e-06,
      "loss": 0.0022,
      "step": 56120
    },
    {
      "epoch": 0.9184324633886934,
      "grad_norm": 0.06740682572126389,
      "learning_rate": 8.760441009703964e-06,
      "loss": 0.0041,
      "step": 56130
    },
    {
      "epoch": 0.9185960893397693,
      "grad_norm": 0.130819171667099,
      "learning_rate": 8.759813553575614e-06,
      "loss": 0.0025,
      "step": 56140
    },
    {
      "epoch": 0.9187597152908451,
      "grad_norm": 0.11753668636083603,
      "learning_rate": 8.75918596116306e-06,
      "loss": 0.0015,
      "step": 56150
    },
    {
      "epoch": 0.918923341241921,
      "grad_norm": 0.16721569001674652,
      "learning_rate": 8.758558232489051e-06,
      "loss": 0.0017,
      "step": 56160
    },
    {
      "epoch": 0.9190869671929968,
      "grad_norm": 0.1588633954524994,
      "learning_rate": 8.75793036757634e-06,
      "loss": 0.002,
      "step": 56170
    },
    {
      "epoch": 0.9192505931440726,
      "grad_norm": 0.1676766276359558,
      "learning_rate": 8.757302366447682e-06,
      "loss": 0.0037,
      "step": 56180
    },
    {
      "epoch": 0.9194142190951485,
      "grad_norm": 0.03938222676515579,
      "learning_rate": 8.756674229125845e-06,
      "loss": 0.0018,
      "step": 56190
    },
    {
      "epoch": 0.9195778450462243,
      "grad_norm": 0.069462351500988,
      "learning_rate": 8.756045955633598e-06,
      "loss": 0.0021,
      "step": 56200
    },
    {
      "epoch": 0.9197414709973002,
      "grad_norm": 0.0305546373128891,
      "learning_rate": 8.75541754599371e-06,
      "loss": 0.0036,
      "step": 56210
    },
    {
      "epoch": 0.919905096948376,
      "grad_norm": 0.07955552637577057,
      "learning_rate": 8.754789000228963e-06,
      "loss": 0.0039,
      "step": 56220
    },
    {
      "epoch": 0.9200687228994519,
      "grad_norm": 0.41352513432502747,
      "learning_rate": 8.75416031836214e-06,
      "loss": 0.003,
      "step": 56230
    },
    {
      "epoch": 0.9202323488505277,
      "grad_norm": 0.06355006992816925,
      "learning_rate": 8.753531500416027e-06,
      "loss": 0.0025,
      "step": 56240
    },
    {
      "epoch": 0.9203959748016035,
      "grad_norm": 0.15708458423614502,
      "learning_rate": 8.752902546413418e-06,
      "loss": 0.0022,
      "step": 56250
    },
    {
      "epoch": 0.9205596007526794,
      "grad_norm": 0.06660819053649902,
      "learning_rate": 8.752273456377114e-06,
      "loss": 0.0018,
      "step": 56260
    },
    {
      "epoch": 0.9207232267037552,
      "grad_norm": 0.35426750779151917,
      "learning_rate": 8.751644230329913e-06,
      "loss": 0.0035,
      "step": 56270
    },
    {
      "epoch": 0.9208868526548311,
      "grad_norm": 0.11654836684465408,
      "learning_rate": 8.751014868294628e-06,
      "loss": 0.0025,
      "step": 56280
    },
    {
      "epoch": 0.9210504786059069,
      "grad_norm": 0.0773138627409935,
      "learning_rate": 8.750385370294065e-06,
      "loss": 0.0033,
      "step": 56290
    },
    {
      "epoch": 0.9212141045569827,
      "grad_norm": 0.20983535051345825,
      "learning_rate": 8.74975573635105e-06,
      "loss": 0.0027,
      "step": 56300
    },
    {
      "epoch": 0.9213777305080586,
      "grad_norm": 0.13140127062797546,
      "learning_rate": 8.7491259664884e-06,
      "loss": 0.0022,
      "step": 56310
    },
    {
      "epoch": 0.9215413564591344,
      "grad_norm": 0.25688043236732483,
      "learning_rate": 8.748496060728945e-06,
      "loss": 0.0039,
      "step": 56320
    },
    {
      "epoch": 0.9217049824102103,
      "grad_norm": 0.03212318569421768,
      "learning_rate": 8.74786601909552e-06,
      "loss": 0.0016,
      "step": 56330
    },
    {
      "epoch": 0.9218686083612861,
      "grad_norm": 0.20308706164360046,
      "learning_rate": 8.747235841610956e-06,
      "loss": 0.003,
      "step": 56340
    },
    {
      "epoch": 0.9220322343123619,
      "grad_norm": 0.3436363637447357,
      "learning_rate": 8.7466055282981e-06,
      "loss": 0.0024,
      "step": 56350
    },
    {
      "epoch": 0.9221958602634378,
      "grad_norm": 0.4917292296886444,
      "learning_rate": 8.745975079179799e-06,
      "loss": 0.003,
      "step": 56360
    },
    {
      "epoch": 0.9223594862145136,
      "grad_norm": 0.07783607393503189,
      "learning_rate": 8.745344494278903e-06,
      "loss": 0.0041,
      "step": 56370
    },
    {
      "epoch": 0.9225231121655895,
      "grad_norm": 0.10799320042133331,
      "learning_rate": 8.744713773618272e-06,
      "loss": 0.0024,
      "step": 56380
    },
    {
      "epoch": 0.9226867381166654,
      "grad_norm": 0.2581583857536316,
      "learning_rate": 8.744082917220766e-06,
      "loss": 0.0032,
      "step": 56390
    },
    {
      "epoch": 0.9228503640677411,
      "grad_norm": 0.0991620346903801,
      "learning_rate": 8.743451925109254e-06,
      "loss": 0.0027,
      "step": 56400
    },
    {
      "epoch": 0.923013990018817,
      "grad_norm": 0.06590348482131958,
      "learning_rate": 8.742820797306606e-06,
      "loss": 0.0022,
      "step": 56410
    },
    {
      "epoch": 0.9231776159698928,
      "grad_norm": 0.1018282026052475,
      "learning_rate": 8.7421895338357e-06,
      "loss": 0.0021,
      "step": 56420
    },
    {
      "epoch": 0.9233412419209687,
      "grad_norm": 0.19533893465995789,
      "learning_rate": 8.741558134719418e-06,
      "loss": 0.0018,
      "step": 56430
    },
    {
      "epoch": 0.9235048678720446,
      "grad_norm": 0.2927216589450836,
      "learning_rate": 8.740926599980646e-06,
      "loss": 0.0036,
      "step": 56440
    },
    {
      "epoch": 0.9236684938231203,
      "grad_norm": 0.11482112854719162,
      "learning_rate": 8.740294929642275e-06,
      "loss": 0.002,
      "step": 56450
    },
    {
      "epoch": 0.9238321197741962,
      "grad_norm": 0.26916325092315674,
      "learning_rate": 8.739663123727203e-06,
      "loss": 0.0025,
      "step": 56460
    },
    {
      "epoch": 0.923995745725272,
      "grad_norm": 0.039856091141700745,
      "learning_rate": 8.739031182258331e-06,
      "loss": 0.0027,
      "step": 56470
    },
    {
      "epoch": 0.9241593716763479,
      "grad_norm": 0.07528948038816452,
      "learning_rate": 8.738399105258563e-06,
      "loss": 0.004,
      "step": 56480
    },
    {
      "epoch": 0.9243229976274238,
      "grad_norm": 0.13336074352264404,
      "learning_rate": 8.737766892750817e-06,
      "loss": 0.0039,
      "step": 56490
    },
    {
      "epoch": 0.9244866235784995,
      "grad_norm": 0.1255578100681305,
      "learning_rate": 8.737134544758001e-06,
      "loss": 0.0029,
      "step": 56500
    },
    {
      "epoch": 0.9246502495295754,
      "grad_norm": 0.011512682773172855,
      "learning_rate": 8.736502061303041e-06,
      "loss": 0.0023,
      "step": 56510
    },
    {
      "epoch": 0.9248138754806512,
      "grad_norm": 0.1771090030670166,
      "learning_rate": 8.735869442408862e-06,
      "loss": 0.0024,
      "step": 56520
    },
    {
      "epoch": 0.9249775014317271,
      "grad_norm": 0.2767779529094696,
      "learning_rate": 8.735236688098393e-06,
      "loss": 0.003,
      "step": 56530
    },
    {
      "epoch": 0.925141127382803,
      "grad_norm": 0.22273513674736023,
      "learning_rate": 8.734603798394571e-06,
      "loss": 0.0019,
      "step": 56540
    },
    {
      "epoch": 0.9253047533338787,
      "grad_norm": 0.028323331847786903,
      "learning_rate": 8.73397077332034e-06,
      "loss": 0.0032,
      "step": 56550
    },
    {
      "epoch": 0.9254683792849546,
      "grad_norm": 0.07312364876270294,
      "learning_rate": 8.733337612898642e-06,
      "loss": 0.0024,
      "step": 56560
    },
    {
      "epoch": 0.9256320052360304,
      "grad_norm": 0.3575909435749054,
      "learning_rate": 8.732704317152427e-06,
      "loss": 0.0031,
      "step": 56570
    },
    {
      "epoch": 0.9257956311871063,
      "grad_norm": 0.1104685440659523,
      "learning_rate": 8.732070886104653e-06,
      "loss": 0.0034,
      "step": 56580
    },
    {
      "epoch": 0.9259592571381822,
      "grad_norm": 0.14856906235218048,
      "learning_rate": 8.73143731977828e-06,
      "loss": 0.0028,
      "step": 56590
    },
    {
      "epoch": 0.9261228830892579,
      "grad_norm": 0.3008265793323517,
      "learning_rate": 8.73080361819627e-06,
      "loss": 0.0046,
      "step": 56600
    },
    {
      "epoch": 0.9262865090403338,
      "grad_norm": 0.1314973086118698,
      "learning_rate": 8.730169781381597e-06,
      "loss": 0.004,
      "step": 56610
    },
    {
      "epoch": 0.9264501349914096,
      "grad_norm": 0.10367554426193237,
      "learning_rate": 8.729535809357236e-06,
      "loss": 0.0031,
      "step": 56620
    },
    {
      "epoch": 0.9266137609424855,
      "grad_norm": 0.24642257392406464,
      "learning_rate": 8.728901702146164e-06,
      "loss": 0.0024,
      "step": 56630
    },
    {
      "epoch": 0.9267773868935614,
      "grad_norm": 0.022402064874768257,
      "learning_rate": 8.728267459771366e-06,
      "loss": 0.0042,
      "step": 56640
    },
    {
      "epoch": 0.9269410128446371,
      "grad_norm": 0.19319544732570648,
      "learning_rate": 8.727633082255837e-06,
      "loss": 0.0037,
      "step": 56650
    },
    {
      "epoch": 0.927104638795713,
      "grad_norm": 0.14075158536434174,
      "learning_rate": 8.726998569622565e-06,
      "loss": 0.0023,
      "step": 56660
    },
    {
      "epoch": 0.9272682647467888,
      "grad_norm": 0.15454579889774323,
      "learning_rate": 8.726363921894555e-06,
      "loss": 0.0023,
      "step": 56670
    },
    {
      "epoch": 0.9274318906978647,
      "grad_norm": 0.08716664463281631,
      "learning_rate": 8.725729139094805e-06,
      "loss": 0.0022,
      "step": 56680
    },
    {
      "epoch": 0.9275955166489406,
      "grad_norm": 0.026017392054200172,
      "learning_rate": 8.725094221246331e-06,
      "loss": 0.0016,
      "step": 56690
    },
    {
      "epoch": 0.9277591426000164,
      "grad_norm": 0.11958318203687668,
      "learning_rate": 8.724459168372144e-06,
      "loss": 0.0023,
      "step": 56700
    },
    {
      "epoch": 0.9279227685510922,
      "grad_norm": 0.14724621176719666,
      "learning_rate": 8.723823980495264e-06,
      "loss": 0.0027,
      "step": 56710
    },
    {
      "epoch": 0.928086394502168,
      "grad_norm": 0.23805959522724152,
      "learning_rate": 8.723188657638715e-06,
      "loss": 0.0043,
      "step": 56720
    },
    {
      "epoch": 0.9282500204532439,
      "grad_norm": 0.10530497133731842,
      "learning_rate": 8.722553199825525e-06,
      "loss": 0.0044,
      "step": 56730
    },
    {
      "epoch": 0.9284136464043197,
      "grad_norm": 0.36908823251724243,
      "learning_rate": 8.721917607078729e-06,
      "loss": 0.0028,
      "step": 56740
    },
    {
      "epoch": 0.9285772723553956,
      "grad_norm": 0.4372231364250183,
      "learning_rate": 8.721281879421363e-06,
      "loss": 0.0025,
      "step": 56750
    },
    {
      "epoch": 0.9287408983064714,
      "grad_norm": 0.07128158956766129,
      "learning_rate": 8.720646016876474e-06,
      "loss": 0.0034,
      "step": 56760
    },
    {
      "epoch": 0.9289045242575472,
      "grad_norm": 0.361255943775177,
      "learning_rate": 8.72001001946711e-06,
      "loss": 0.0051,
      "step": 56770
    },
    {
      "epoch": 0.9290681502086231,
      "grad_norm": 0.11073429137468338,
      "learning_rate": 8.719373887216322e-06,
      "loss": 0.0021,
      "step": 56780
    },
    {
      "epoch": 0.9292317761596989,
      "grad_norm": 0.049818575382232666,
      "learning_rate": 8.718737620147173e-06,
      "loss": 0.0039,
      "step": 56790
    },
    {
      "epoch": 0.9293954021107748,
      "grad_norm": 0.2616579234600067,
      "learning_rate": 8.71810121828272e-06,
      "loss": 0.003,
      "step": 56800
    },
    {
      "epoch": 0.9295590280618506,
      "grad_norm": 0.20634043216705322,
      "learning_rate": 8.717464681646035e-06,
      "loss": 0.0033,
      "step": 56810
    },
    {
      "epoch": 0.9297226540129264,
      "grad_norm": 0.14242428541183472,
      "learning_rate": 8.71682801026019e-06,
      "loss": 0.0025,
      "step": 56820
    },
    {
      "epoch": 0.9298862799640023,
      "grad_norm": 0.26235896348953247,
      "learning_rate": 8.716191204148262e-06,
      "loss": 0.0024,
      "step": 56830
    },
    {
      "epoch": 0.9300499059150781,
      "grad_norm": 0.19843776524066925,
      "learning_rate": 8.715554263333337e-06,
      "loss": 0.0029,
      "step": 56840
    },
    {
      "epoch": 0.930213531866154,
      "grad_norm": 0.26787033677101135,
      "learning_rate": 8.714917187838497e-06,
      "loss": 0.0022,
      "step": 56850
    },
    {
      "epoch": 0.9303771578172298,
      "grad_norm": 0.1914227306842804,
      "learning_rate": 8.714279977686839e-06,
      "loss": 0.0042,
      "step": 56860
    },
    {
      "epoch": 0.9305407837683056,
      "grad_norm": 0.09505842626094818,
      "learning_rate": 8.713642632901459e-06,
      "loss": 0.0017,
      "step": 56870
    },
    {
      "epoch": 0.9307044097193815,
      "grad_norm": 0.243962362408638,
      "learning_rate": 8.713005153505458e-06,
      "loss": 0.0021,
      "step": 56880
    },
    {
      "epoch": 0.9308680356704573,
      "grad_norm": 0.05524337291717529,
      "learning_rate": 8.712367539521945e-06,
      "loss": 0.0033,
      "step": 56890
    },
    {
      "epoch": 0.9310316616215332,
      "grad_norm": 0.15506170690059662,
      "learning_rate": 8.711729790974031e-06,
      "loss": 0.0012,
      "step": 56900
    },
    {
      "epoch": 0.931195287572609,
      "grad_norm": 0.15791694819927216,
      "learning_rate": 8.711091907884833e-06,
      "loss": 0.0037,
      "step": 56910
    },
    {
      "epoch": 0.9313589135236848,
      "grad_norm": 0.05830807983875275,
      "learning_rate": 8.710453890277473e-06,
      "loss": 0.0035,
      "step": 56920
    },
    {
      "epoch": 0.9315225394747607,
      "grad_norm": 0.22038206458091736,
      "learning_rate": 8.709815738175077e-06,
      "loss": 0.0027,
      "step": 56930
    },
    {
      "epoch": 0.9316861654258365,
      "grad_norm": 0.21177567541599274,
      "learning_rate": 8.709177451600776e-06,
      "loss": 0.0034,
      "step": 56940
    },
    {
      "epoch": 0.9318497913769124,
      "grad_norm": 0.07490844279527664,
      "learning_rate": 8.70853903057771e-06,
      "loss": 0.0026,
      "step": 56950
    },
    {
      "epoch": 0.9320134173279883,
      "grad_norm": 0.15769219398498535,
      "learning_rate": 8.707900475129014e-06,
      "loss": 0.0029,
      "step": 56960
    },
    {
      "epoch": 0.932177043279064,
      "grad_norm": 0.25551092624664307,
      "learning_rate": 8.707261785277839e-06,
      "loss": 0.0038,
      "step": 56970
    },
    {
      "epoch": 0.9323406692301399,
      "grad_norm": 0.06905394047498703,
      "learning_rate": 8.706622961047333e-06,
      "loss": 0.0017,
      "step": 56980
    },
    {
      "epoch": 0.9325042951812157,
      "grad_norm": 0.15709897875785828,
      "learning_rate": 8.705984002460655e-06,
      "loss": 0.0042,
      "step": 56990
    },
    {
      "epoch": 0.9326679211322916,
      "grad_norm": 0.08229276537895203,
      "learning_rate": 8.705344909540965e-06,
      "loss": 0.0026,
      "step": 57000
    },
    {
      "epoch": 0.9328315470833675,
      "grad_norm": 0.02851949632167816,
      "learning_rate": 8.704705682311424e-06,
      "loss": 0.0035,
      "step": 57010
    },
    {
      "epoch": 0.9329951730344432,
      "grad_norm": 0.059260979294776917,
      "learning_rate": 8.70406632079521e-06,
      "loss": 0.0027,
      "step": 57020
    },
    {
      "epoch": 0.9331587989855191,
      "grad_norm": 0.14293326437473297,
      "learning_rate": 8.703426825015493e-06,
      "loss": 0.0026,
      "step": 57030
    },
    {
      "epoch": 0.9333224249365949,
      "grad_norm": 0.09775244444608688,
      "learning_rate": 8.702787194995455e-06,
      "loss": 0.0022,
      "step": 57040
    },
    {
      "epoch": 0.9334860508876708,
      "grad_norm": 0.11891881376504898,
      "learning_rate": 8.70214743075828e-06,
      "loss": 0.0045,
      "step": 57050
    },
    {
      "epoch": 0.9336496768387467,
      "grad_norm": 0.21938017010688782,
      "learning_rate": 8.701507532327157e-06,
      "loss": 0.0044,
      "step": 57060
    },
    {
      "epoch": 0.9338133027898224,
      "grad_norm": 0.2713831067085266,
      "learning_rate": 8.700867499725283e-06,
      "loss": 0.003,
      "step": 57070
    },
    {
      "epoch": 0.9339769287408983,
      "grad_norm": 0.09283127635717392,
      "learning_rate": 8.700227332975859e-06,
      "loss": 0.0034,
      "step": 57080
    },
    {
      "epoch": 0.9341405546919741,
      "grad_norm": 0.01693669892847538,
      "learning_rate": 8.699587032102084e-06,
      "loss": 0.0036,
      "step": 57090
    },
    {
      "epoch": 0.93430418064305,
      "grad_norm": 0.3564091622829437,
      "learning_rate": 8.698946597127174e-06,
      "loss": 0.0034,
      "step": 57100
    },
    {
      "epoch": 0.9344678065941259,
      "grad_norm": 0.1506435126066208,
      "learning_rate": 8.698306028074339e-06,
      "loss": 0.0038,
      "step": 57110
    },
    {
      "epoch": 0.9346314325452016,
      "grad_norm": 0.24460448324680328,
      "learning_rate": 8.697665324966798e-06,
      "loss": 0.0042,
      "step": 57120
    },
    {
      "epoch": 0.9347950584962775,
      "grad_norm": 0.16294312477111816,
      "learning_rate": 8.697024487827776e-06,
      "loss": 0.0032,
      "step": 57130
    },
    {
      "epoch": 0.9349586844473533,
      "grad_norm": 0.05100912228226662,
      "learning_rate": 8.696383516680505e-06,
      "loss": 0.0022,
      "step": 57140
    },
    {
      "epoch": 0.9351223103984292,
      "grad_norm": 0.12327504903078079,
      "learning_rate": 8.695742411548212e-06,
      "loss": 0.0032,
      "step": 57150
    },
    {
      "epoch": 0.9352859363495051,
      "grad_norm": 0.31645238399505615,
      "learning_rate": 8.695101172454141e-06,
      "loss": 0.0039,
      "step": 57160
    },
    {
      "epoch": 0.9354495623005809,
      "grad_norm": 0.15630635619163513,
      "learning_rate": 8.694459799421533e-06,
      "loss": 0.0027,
      "step": 57170
    },
    {
      "epoch": 0.9356131882516567,
      "grad_norm": 0.10846993327140808,
      "learning_rate": 8.693818292473637e-06,
      "loss": 0.0024,
      "step": 57180
    },
    {
      "epoch": 0.9357768142027325,
      "grad_norm": 0.21625679731369019,
      "learning_rate": 8.693176651633704e-06,
      "loss": 0.002,
      "step": 57190
    },
    {
      "epoch": 0.9359404401538084,
      "grad_norm": 0.09098803251981735,
      "learning_rate": 8.692534876924996e-06,
      "loss": 0.0027,
      "step": 57200
    },
    {
      "epoch": 0.9361040661048843,
      "grad_norm": 0.09953711181879044,
      "learning_rate": 8.691892968370771e-06,
      "loss": 0.003,
      "step": 57210
    },
    {
      "epoch": 0.93626769205596,
      "grad_norm": 0.050231534987688065,
      "learning_rate": 8.691250925994301e-06,
      "loss": 0.0023,
      "step": 57220
    },
    {
      "epoch": 0.9364313180070359,
      "grad_norm": 0.21652771532535553,
      "learning_rate": 8.690608749818857e-06,
      "loss": 0.0022,
      "step": 57230
    },
    {
      "epoch": 0.9365949439581117,
      "grad_norm": 0.13481707870960236,
      "learning_rate": 8.689966439867715e-06,
      "loss": 0.0032,
      "step": 57240
    },
    {
      "epoch": 0.9367585699091876,
      "grad_norm": 0.10447494685649872,
      "learning_rate": 8.689323996164158e-06,
      "loss": 0.003,
      "step": 57250
    },
    {
      "epoch": 0.9369221958602635,
      "grad_norm": 0.32264572381973267,
      "learning_rate": 8.688681418731474e-06,
      "loss": 0.004,
      "step": 57260
    },
    {
      "epoch": 0.9370858218113393,
      "grad_norm": 0.3122585713863373,
      "learning_rate": 8.688038707592955e-06,
      "loss": 0.0041,
      "step": 57270
    },
    {
      "epoch": 0.9372494477624151,
      "grad_norm": 0.24094940721988678,
      "learning_rate": 8.687395862771894e-06,
      "loss": 0.003,
      "step": 57280
    },
    {
      "epoch": 0.9374130737134909,
      "grad_norm": 0.11071789264678955,
      "learning_rate": 8.686752884291599e-06,
      "loss": 0.0022,
      "step": 57290
    },
    {
      "epoch": 0.9375766996645668,
      "grad_norm": 0.4375678300857544,
      "learning_rate": 8.686109772175368e-06,
      "loss": 0.0026,
      "step": 57300
    },
    {
      "epoch": 0.9377403256156427,
      "grad_norm": 0.12545320391654968,
      "learning_rate": 8.68546652644652e-06,
      "loss": 0.0037,
      "step": 57310
    },
    {
      "epoch": 0.9379039515667185,
      "grad_norm": 0.26050540804862976,
      "learning_rate": 8.68482314712837e-06,
      "loss": 0.003,
      "step": 57320
    },
    {
      "epoch": 0.9380675775177943,
      "grad_norm": 0.1724195033311844,
      "learning_rate": 8.684179634244234e-06,
      "loss": 0.0018,
      "step": 57330
    },
    {
      "epoch": 0.9382312034688701,
      "grad_norm": 0.11418847739696503,
      "learning_rate": 8.683535987817441e-06,
      "loss": 0.0041,
      "step": 57340
    },
    {
      "epoch": 0.938394829419946,
      "grad_norm": 0.09722284227609634,
      "learning_rate": 8.682892207871322e-06,
      "loss": 0.0028,
      "step": 57350
    },
    {
      "epoch": 0.9385584553710219,
      "grad_norm": 0.03472256287932396,
      "learning_rate": 8.682248294429214e-06,
      "loss": 0.0026,
      "step": 57360
    },
    {
      "epoch": 0.9387220813220977,
      "grad_norm": 0.2636689245700836,
      "learning_rate": 8.681604247514455e-06,
      "loss": 0.0026,
      "step": 57370
    },
    {
      "epoch": 0.9388857072731736,
      "grad_norm": 0.22415636479854584,
      "learning_rate": 8.68096006715039e-06,
      "loss": 0.0037,
      "step": 57380
    },
    {
      "epoch": 0.9390493332242493,
      "grad_norm": 0.027264980599284172,
      "learning_rate": 8.680315753360368e-06,
      "loss": 0.0025,
      "step": 57390
    },
    {
      "epoch": 0.9392129591753252,
      "grad_norm": 0.08876129984855652,
      "learning_rate": 8.679671306167748e-06,
      "loss": 0.0023,
      "step": 57400
    },
    {
      "epoch": 0.9393765851264011,
      "grad_norm": 0.11450862884521484,
      "learning_rate": 8.679026725595886e-06,
      "loss": 0.0032,
      "step": 57410
    },
    {
      "epoch": 0.9395402110774769,
      "grad_norm": 0.3536887466907501,
      "learning_rate": 8.678382011668145e-06,
      "loss": 0.0027,
      "step": 57420
    },
    {
      "epoch": 0.9397038370285528,
      "grad_norm": 0.2696461081504822,
      "learning_rate": 8.6777371644079e-06,
      "loss": 0.003,
      "step": 57430
    },
    {
      "epoch": 0.9398674629796285,
      "grad_norm": 0.21755576133728027,
      "learning_rate": 8.67709218383852e-06,
      "loss": 0.0033,
      "step": 57440
    },
    {
      "epoch": 0.9400310889307044,
      "grad_norm": 0.16868865489959717,
      "learning_rate": 8.676447069983387e-06,
      "loss": 0.0038,
      "step": 57450
    },
    {
      "epoch": 0.9401947148817803,
      "grad_norm": 0.05901087820529938,
      "learning_rate": 8.675801822865883e-06,
      "loss": 0.0025,
      "step": 57460
    },
    {
      "epoch": 0.9403583408328561,
      "grad_norm": 0.2886706292629242,
      "learning_rate": 8.6751564425094e-06,
      "loss": 0.0061,
      "step": 57470
    },
    {
      "epoch": 0.940521966783932,
      "grad_norm": 0.6585190296173096,
      "learning_rate": 8.674510928937327e-06,
      "loss": 0.0045,
      "step": 57480
    },
    {
      "epoch": 0.9406855927350077,
      "grad_norm": 0.2559524178504944,
      "learning_rate": 8.673865282173064e-06,
      "loss": 0.002,
      "step": 57490
    },
    {
      "epoch": 0.9408492186860836,
      "grad_norm": 0.15461988747119904,
      "learning_rate": 8.673219502240014e-06,
      "loss": 0.0034,
      "step": 57500
    },
    {
      "epoch": 0.9410128446371595,
      "grad_norm": 0.14877702295780182,
      "learning_rate": 8.672573589161586e-06,
      "loss": 0.0029,
      "step": 57510
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.13619624078273773,
      "learning_rate": 8.671927542961194e-06,
      "loss": 0.0027,
      "step": 57520
    },
    {
      "epoch": 0.9413400965393112,
      "grad_norm": 0.1986972689628601,
      "learning_rate": 8.671281363662252e-06,
      "loss": 0.0017,
      "step": 57530
    },
    {
      "epoch": 0.9415037224903869,
      "grad_norm": 0.10938013345003128,
      "learning_rate": 8.670635051288182e-06,
      "loss": 0.0058,
      "step": 57540
    },
    {
      "epoch": 0.9416673484414628,
      "grad_norm": 0.08242452889680862,
      "learning_rate": 8.669988605862419e-06,
      "loss": 0.0031,
      "step": 57550
    },
    {
      "epoch": 0.9418309743925387,
      "grad_norm": 0.08062282204627991,
      "learning_rate": 8.669342027408385e-06,
      "loss": 0.003,
      "step": 57560
    },
    {
      "epoch": 0.9419946003436145,
      "grad_norm": 0.16516931354999542,
      "learning_rate": 8.668695315949524e-06,
      "loss": 0.0038,
      "step": 57570
    },
    {
      "epoch": 0.9421582262946904,
      "grad_norm": 0.19662010669708252,
      "learning_rate": 8.668048471509274e-06,
      "loss": 0.0036,
      "step": 57580
    },
    {
      "epoch": 0.9423218522457661,
      "grad_norm": 0.03410712257027626,
      "learning_rate": 8.667401494111083e-06,
      "loss": 0.0025,
      "step": 57590
    },
    {
      "epoch": 0.942485478196842,
      "grad_norm": 0.0899646133184433,
      "learning_rate": 8.666754383778403e-06,
      "loss": 0.0027,
      "step": 57600
    },
    {
      "epoch": 0.9426491041479178,
      "grad_norm": 0.13241948187351227,
      "learning_rate": 8.666107140534688e-06,
      "loss": 0.0012,
      "step": 57610
    },
    {
      "epoch": 0.9428127300989937,
      "grad_norm": 0.17516230046749115,
      "learning_rate": 8.665459764403402e-06,
      "loss": 0.0027,
      "step": 57620
    },
    {
      "epoch": 0.9429763560500696,
      "grad_norm": 0.16082867980003357,
      "learning_rate": 8.664812255408008e-06,
      "loss": 0.0035,
      "step": 57630
    },
    {
      "epoch": 0.9431399820011453,
      "grad_norm": 0.23999162018299103,
      "learning_rate": 8.664164613571979e-06,
      "loss": 0.0026,
      "step": 57640
    },
    {
      "epoch": 0.9433036079522212,
      "grad_norm": 0.2453894168138504,
      "learning_rate": 8.663516838918788e-06,
      "loss": 0.0027,
      "step": 57650
    },
    {
      "epoch": 0.943467233903297,
      "grad_norm": 0.24244779348373413,
      "learning_rate": 8.662868931471919e-06,
      "loss": 0.0032,
      "step": 57660
    },
    {
      "epoch": 0.9436308598543729,
      "grad_norm": 0.431749552488327,
      "learning_rate": 8.662220891254851e-06,
      "loss": 0.0024,
      "step": 57670
    },
    {
      "epoch": 0.9437944858054488,
      "grad_norm": 0.0839313417673111,
      "learning_rate": 8.66157271829108e-06,
      "loss": 0.0114,
      "step": 57680
    },
    {
      "epoch": 0.9439581117565246,
      "grad_norm": 0.12909993529319763,
      "learning_rate": 8.660924412604098e-06,
      "loss": 0.0027,
      "step": 57690
    },
    {
      "epoch": 0.9441217377076004,
      "grad_norm": 0.09456297755241394,
      "learning_rate": 8.660275974217406e-06,
      "loss": 0.0028,
      "step": 57700
    },
    {
      "epoch": 0.9442853636586762,
      "grad_norm": 0.4690358340740204,
      "learning_rate": 8.659627403154506e-06,
      "loss": 0.0035,
      "step": 57710
    },
    {
      "epoch": 0.9444489896097521,
      "grad_norm": 0.22912617027759552,
      "learning_rate": 8.658978699438909e-06,
      "loss": 0.005,
      "step": 57720
    },
    {
      "epoch": 0.944612615560828,
      "grad_norm": 0.1195877194404602,
      "learning_rate": 8.658329863094128e-06,
      "loss": 0.0023,
      "step": 57730
    },
    {
      "epoch": 0.9447762415119038,
      "grad_norm": 0.1556454449892044,
      "learning_rate": 8.65768089414368e-06,
      "loss": 0.0022,
      "step": 57740
    },
    {
      "epoch": 0.9449398674629796,
      "grad_norm": 0.29133787751197815,
      "learning_rate": 8.657031792611092e-06,
      "loss": 0.003,
      "step": 57750
    },
    {
      "epoch": 0.9451034934140554,
      "grad_norm": 0.3909377157688141,
      "learning_rate": 8.656382558519893e-06,
      "loss": 0.0046,
      "step": 57760
    },
    {
      "epoch": 0.9452671193651313,
      "grad_norm": 0.17427799105644226,
      "learning_rate": 8.655733191893611e-06,
      "loss": 0.0039,
      "step": 57770
    },
    {
      "epoch": 0.9454307453162072,
      "grad_norm": 0.22057141363620758,
      "learning_rate": 8.655083692755789e-06,
      "loss": 0.0027,
      "step": 57780
    },
    {
      "epoch": 0.945594371267283,
      "grad_norm": 0.07879922538995743,
      "learning_rate": 8.654434061129969e-06,
      "loss": 0.0022,
      "step": 57790
    },
    {
      "epoch": 0.9457579972183588,
      "grad_norm": 0.04771272838115692,
      "learning_rate": 8.653784297039694e-06,
      "loss": 0.0019,
      "step": 57800
    },
    {
      "epoch": 0.9459216231694346,
      "grad_norm": 0.3877989947795868,
      "learning_rate": 8.653134400508523e-06,
      "loss": 0.0044,
      "step": 57810
    },
    {
      "epoch": 0.9460852491205105,
      "grad_norm": 0.34246203303337097,
      "learning_rate": 8.652484371560008e-06,
      "loss": 0.002,
      "step": 57820
    },
    {
      "epoch": 0.9462488750715864,
      "grad_norm": 0.2034769356250763,
      "learning_rate": 8.651834210217713e-06,
      "loss": 0.003,
      "step": 57830
    },
    {
      "epoch": 0.9464125010226622,
      "grad_norm": 0.2086765170097351,
      "learning_rate": 8.651183916505205e-06,
      "loss": 0.0034,
      "step": 57840
    },
    {
      "epoch": 0.946576126973738,
      "grad_norm": 0.1092161312699318,
      "learning_rate": 8.650533490446057e-06,
      "loss": 0.0032,
      "step": 57850
    },
    {
      "epoch": 0.9467397529248138,
      "grad_norm": 0.07904884964227676,
      "learning_rate": 8.649882932063843e-06,
      "loss": 0.003,
      "step": 57860
    },
    {
      "epoch": 0.9469033788758897,
      "grad_norm": 0.21050243079662323,
      "learning_rate": 8.649232241382146e-06,
      "loss": 0.0035,
      "step": 57870
    },
    {
      "epoch": 0.9470670048269656,
      "grad_norm": 0.09052559733390808,
      "learning_rate": 8.648581418424549e-06,
      "loss": 0.0048,
      "step": 57880
    },
    {
      "epoch": 0.9472306307780414,
      "grad_norm": 0.09729443490505219,
      "learning_rate": 8.647930463214646e-06,
      "loss": 0.0028,
      "step": 57890
    },
    {
      "epoch": 0.9473942567291173,
      "grad_norm": 0.15841154754161835,
      "learning_rate": 8.64727937577603e-06,
      "loss": 0.0021,
      "step": 57900
    },
    {
      "epoch": 0.947557882680193,
      "grad_norm": 0.1975061297416687,
      "learning_rate": 8.646628156132304e-06,
      "loss": 0.0052,
      "step": 57910
    },
    {
      "epoch": 0.9477215086312689,
      "grad_norm": 0.07624470442533493,
      "learning_rate": 8.645976804307071e-06,
      "loss": 0.0032,
      "step": 57920
    },
    {
      "epoch": 0.9478851345823448,
      "grad_norm": 0.2026941180229187,
      "learning_rate": 8.645325320323942e-06,
      "loss": 0.0024,
      "step": 57930
    },
    {
      "epoch": 0.9480487605334206,
      "grad_norm": 0.032917656004428864,
      "learning_rate": 8.644673704206533e-06,
      "loss": 0.0016,
      "step": 57940
    },
    {
      "epoch": 0.9482123864844965,
      "grad_norm": 0.2713351547718048,
      "learning_rate": 8.644021955978458e-06,
      "loss": 0.003,
      "step": 57950
    },
    {
      "epoch": 0.9483760124355722,
      "grad_norm": 0.07845479249954224,
      "learning_rate": 8.643370075663348e-06,
      "loss": 0.0033,
      "step": 57960
    },
    {
      "epoch": 0.9485396383866481,
      "grad_norm": 0.06043951213359833,
      "learning_rate": 8.64271806328483e-06,
      "loss": 0.0018,
      "step": 57970
    },
    {
      "epoch": 0.948703264337724,
      "grad_norm": 0.1575866937637329,
      "learning_rate": 8.642065918866536e-06,
      "loss": 0.0047,
      "step": 57980
    },
    {
      "epoch": 0.9488668902887998,
      "grad_norm": 0.14943064749240875,
      "learning_rate": 8.641413642432106e-06,
      "loss": 0.0027,
      "step": 57990
    },
    {
      "epoch": 0.9490305162398757,
      "grad_norm": 0.1311347484588623,
      "learning_rate": 8.640761234005183e-06,
      "loss": 0.0023,
      "step": 58000
    },
    {
      "epoch": 0.9491941421909514,
      "grad_norm": 0.21922120451927185,
      "learning_rate": 8.640108693609416e-06,
      "loss": 0.0025,
      "step": 58010
    },
    {
      "epoch": 0.9493577681420273,
      "grad_norm": 0.2126881331205368,
      "learning_rate": 8.639456021268459e-06,
      "loss": 0.0036,
      "step": 58020
    },
    {
      "epoch": 0.9495213940931032,
      "grad_norm": 0.29935869574546814,
      "learning_rate": 8.638803217005966e-06,
      "loss": 0.002,
      "step": 58030
    },
    {
      "epoch": 0.949685020044179,
      "grad_norm": 0.2871516942977905,
      "learning_rate": 8.638150280845603e-06,
      "loss": 0.0028,
      "step": 58040
    },
    {
      "epoch": 0.9498486459952549,
      "grad_norm": 0.2636096179485321,
      "learning_rate": 8.637497212811036e-06,
      "loss": 0.0018,
      "step": 58050
    },
    {
      "epoch": 0.9500122719463306,
      "grad_norm": 0.053623076528310776,
      "learning_rate": 8.63684401292594e-06,
      "loss": 0.0016,
      "step": 58060
    },
    {
      "epoch": 0.9501758978974065,
      "grad_norm": 0.34329134225845337,
      "learning_rate": 8.636190681213988e-06,
      "loss": 0.0018,
      "step": 58070
    },
    {
      "epoch": 0.9503395238484824,
      "grad_norm": 0.051948998123407364,
      "learning_rate": 8.635537217698862e-06,
      "loss": 0.0038,
      "step": 58080
    },
    {
      "epoch": 0.9505031497995582,
      "grad_norm": 0.10153673589229584,
      "learning_rate": 8.634883622404251e-06,
      "loss": 0.0018,
      "step": 58090
    },
    {
      "epoch": 0.9506667757506341,
      "grad_norm": 0.21416954696178436,
      "learning_rate": 8.634229895353843e-06,
      "loss": 0.0053,
      "step": 58100
    },
    {
      "epoch": 0.9508304017017098,
      "grad_norm": 0.20511853694915771,
      "learning_rate": 8.633576036571338e-06,
      "loss": 0.0023,
      "step": 58110
    },
    {
      "epoch": 0.9509940276527857,
      "grad_norm": 0.26264485716819763,
      "learning_rate": 8.632922046080434e-06,
      "loss": 0.0027,
      "step": 58120
    },
    {
      "epoch": 0.9511576536038616,
      "grad_norm": 0.09430819749832153,
      "learning_rate": 8.632267923904836e-06,
      "loss": 0.0031,
      "step": 58130
    },
    {
      "epoch": 0.9513212795549374,
      "grad_norm": 0.048129092901945114,
      "learning_rate": 8.631613670068257e-06,
      "loss": 0.0035,
      "step": 58140
    },
    {
      "epoch": 0.9514849055060133,
      "grad_norm": 0.12212701886892319,
      "learning_rate": 8.63095928459441e-06,
      "loss": 0.003,
      "step": 58150
    },
    {
      "epoch": 0.951648531457089,
      "grad_norm": 0.053956203162670135,
      "learning_rate": 8.630304767507018e-06,
      "loss": 0.0019,
      "step": 58160
    },
    {
      "epoch": 0.9518121574081649,
      "grad_norm": 0.29734739661216736,
      "learning_rate": 8.629650118829803e-06,
      "loss": 0.0019,
      "step": 58170
    },
    {
      "epoch": 0.9519757833592408,
      "grad_norm": 0.16599032282829285,
      "learning_rate": 8.628995338586492e-06,
      "loss": 0.003,
      "step": 58180
    },
    {
      "epoch": 0.9521394093103166,
      "grad_norm": 0.25429263710975647,
      "learning_rate": 8.628340426800825e-06,
      "loss": 0.0026,
      "step": 58190
    },
    {
      "epoch": 0.9523030352613925,
      "grad_norm": 0.17607857286930084,
      "learning_rate": 8.627685383496536e-06,
      "loss": 0.0026,
      "step": 58200
    },
    {
      "epoch": 0.9524666612124683,
      "grad_norm": 0.17603527009487152,
      "learning_rate": 8.62703020869737e-06,
      "loss": 0.0025,
      "step": 58210
    },
    {
      "epoch": 0.9526302871635441,
      "grad_norm": 0.0441669337451458,
      "learning_rate": 8.626374902427079e-06,
      "loss": 0.0023,
      "step": 58220
    },
    {
      "epoch": 0.95279391311462,
      "grad_norm": 0.11860113590955734,
      "learning_rate": 8.625719464709414e-06,
      "loss": 0.0019,
      "step": 58230
    },
    {
      "epoch": 0.9529575390656958,
      "grad_norm": 0.40554141998291016,
      "learning_rate": 8.62506389556813e-06,
      "loss": 0.0025,
      "step": 58240
    },
    {
      "epoch": 0.9531211650167717,
      "grad_norm": 0.2965421676635742,
      "learning_rate": 8.624408195026993e-06,
      "loss": 0.0026,
      "step": 58250
    },
    {
      "epoch": 0.9532847909678475,
      "grad_norm": 0.17998652160167694,
      "learning_rate": 8.62375236310977e-06,
      "loss": 0.0028,
      "step": 58260
    },
    {
      "epoch": 0.9534484169189233,
      "grad_norm": 0.0767703726887703,
      "learning_rate": 8.623096399840234e-06,
      "loss": 0.0031,
      "step": 58270
    },
    {
      "epoch": 0.9536120428699992,
      "grad_norm": 0.19731241464614868,
      "learning_rate": 8.622440305242161e-06,
      "loss": 0.0022,
      "step": 58280
    },
    {
      "epoch": 0.953775668821075,
      "grad_norm": 0.2790059447288513,
      "learning_rate": 8.621784079339334e-06,
      "loss": 0.002,
      "step": 58290
    },
    {
      "epoch": 0.9539392947721509,
      "grad_norm": 0.1442946493625641,
      "learning_rate": 8.621127722155537e-06,
      "loss": 0.0032,
      "step": 58300
    },
    {
      "epoch": 0.9541029207232267,
      "grad_norm": 0.31856971979141235,
      "learning_rate": 8.620471233714565e-06,
      "loss": 0.0026,
      "step": 58310
    },
    {
      "epoch": 0.9542665466743026,
      "grad_norm": 0.1451014280319214,
      "learning_rate": 8.619814614040211e-06,
      "loss": 0.0026,
      "step": 58320
    },
    {
      "epoch": 0.9544301726253784,
      "grad_norm": 0.15975750982761383,
      "learning_rate": 8.619157863156278e-06,
      "loss": 0.0036,
      "step": 58330
    },
    {
      "epoch": 0.9545937985764542,
      "grad_norm": 0.1194847971200943,
      "learning_rate": 8.61850098108657e-06,
      "loss": 0.0035,
      "step": 58340
    },
    {
      "epoch": 0.9547574245275301,
      "grad_norm": 0.12278693169355392,
      "learning_rate": 8.6178439678549e-06,
      "loss": 0.0041,
      "step": 58350
    },
    {
      "epoch": 0.9549210504786059,
      "grad_norm": 0.24334728717803955,
      "learning_rate": 8.61718682348508e-06,
      "loss": 0.0041,
      "step": 58360
    },
    {
      "epoch": 0.9550846764296818,
      "grad_norm": 0.37377408146858215,
      "learning_rate": 8.616529548000934e-06,
      "loss": 0.0039,
      "step": 58370
    },
    {
      "epoch": 0.9552483023807576,
      "grad_norm": 0.10425055027008057,
      "learning_rate": 8.61587214142628e-06,
      "loss": 0.0045,
      "step": 58380
    },
    {
      "epoch": 0.9554119283318334,
      "grad_norm": 0.11645399779081345,
      "learning_rate": 8.615214603784955e-06,
      "loss": 0.0034,
      "step": 58390
    },
    {
      "epoch": 0.9555755542829093,
      "grad_norm": 0.09546682238578796,
      "learning_rate": 8.614556935100786e-06,
      "loss": 0.0024,
      "step": 58400
    },
    {
      "epoch": 0.9557391802339851,
      "grad_norm": 0.1266881823539734,
      "learning_rate": 8.613899135397618e-06,
      "loss": 0.0014,
      "step": 58410
    },
    {
      "epoch": 0.955902806185061,
      "grad_norm": 0.10030030459165573,
      "learning_rate": 8.613241204699291e-06,
      "loss": 0.0036,
      "step": 58420
    },
    {
      "epoch": 0.9560664321361368,
      "grad_norm": 0.07037805020809174,
      "learning_rate": 8.612583143029654e-06,
      "loss": 0.0033,
      "step": 58430
    },
    {
      "epoch": 0.9562300580872126,
      "grad_norm": 0.133728489279747,
      "learning_rate": 8.611924950412562e-06,
      "loss": 0.0018,
      "step": 58440
    },
    {
      "epoch": 0.9563936840382885,
      "grad_norm": 0.08356554806232452,
      "learning_rate": 8.611266626871871e-06,
      "loss": 0.003,
      "step": 58450
    },
    {
      "epoch": 0.9565573099893643,
      "grad_norm": 0.11183792352676392,
      "learning_rate": 8.610608172431444e-06,
      "loss": 0.0019,
      "step": 58460
    },
    {
      "epoch": 0.9567209359404402,
      "grad_norm": 0.10106891393661499,
      "learning_rate": 8.609949587115149e-06,
      "loss": 0.0032,
      "step": 58470
    },
    {
      "epoch": 0.9568845618915159,
      "grad_norm": 0.13128845393657684,
      "learning_rate": 8.609290870946857e-06,
      "loss": 0.0035,
      "step": 58480
    },
    {
      "epoch": 0.9570481878425918,
      "grad_norm": 0.10600393265485764,
      "learning_rate": 8.608632023950448e-06,
      "loss": 0.0023,
      "step": 58490
    },
    {
      "epoch": 0.9572118137936677,
      "grad_norm": 0.08835292607545853,
      "learning_rate": 8.6079730461498e-06,
      "loss": 0.0029,
      "step": 58500
    },
    {
      "epoch": 0.9573754397447435,
      "grad_norm": 0.059927403926849365,
      "learning_rate": 8.6073139375688e-06,
      "loss": 0.0027,
      "step": 58510
    },
    {
      "epoch": 0.9575390656958194,
      "grad_norm": 0.11428363621234894,
      "learning_rate": 8.60665469823134e-06,
      "loss": 0.0021,
      "step": 58520
    },
    {
      "epoch": 0.9577026916468951,
      "grad_norm": 0.13040103018283844,
      "learning_rate": 8.605995328161314e-06,
      "loss": 0.0027,
      "step": 58530
    },
    {
      "epoch": 0.957866317597971,
      "grad_norm": 0.27606552839279175,
      "learning_rate": 8.605335827382626e-06,
      "loss": 0.002,
      "step": 58540
    },
    {
      "epoch": 0.9580299435490469,
      "grad_norm": 0.03385399654507637,
      "learning_rate": 8.60467619591918e-06,
      "loss": 0.0033,
      "step": 58550
    },
    {
      "epoch": 0.9581935695001227,
      "grad_norm": 0.07648373395204544,
      "learning_rate": 8.604016433794884e-06,
      "loss": 0.002,
      "step": 58560
    },
    {
      "epoch": 0.9583571954511986,
      "grad_norm": 0.02985292486846447,
      "learning_rate": 8.603356541033655e-06,
      "loss": 0.003,
      "step": 58570
    },
    {
      "epoch": 0.9585208214022743,
      "grad_norm": 0.24194005131721497,
      "learning_rate": 8.602696517659412e-06,
      "loss": 0.0046,
      "step": 58580
    },
    {
      "epoch": 0.9586844473533502,
      "grad_norm": 0.07669809460639954,
      "learning_rate": 8.602036363696078e-06,
      "loss": 0.0017,
      "step": 58590
    },
    {
      "epoch": 0.9588480733044261,
      "grad_norm": 0.2314564287662506,
      "learning_rate": 8.601376079167585e-06,
      "loss": 0.0018,
      "step": 58600
    },
    {
      "epoch": 0.9590116992555019,
      "grad_norm": 0.08914574980735779,
      "learning_rate": 8.600715664097863e-06,
      "loss": 0.0027,
      "step": 58610
    },
    {
      "epoch": 0.9591753252065778,
      "grad_norm": 0.06118670478463173,
      "learning_rate": 8.600055118510852e-06,
      "loss": 0.0025,
      "step": 58620
    },
    {
      "epoch": 0.9593389511576536,
      "grad_norm": 0.18590016663074493,
      "learning_rate": 8.599394442430497e-06,
      "loss": 0.003,
      "step": 58630
    },
    {
      "epoch": 0.9595025771087294,
      "grad_norm": 0.22793278098106384,
      "learning_rate": 8.598733635880744e-06,
      "loss": 0.0022,
      "step": 58640
    },
    {
      "epoch": 0.9596662030598053,
      "grad_norm": 0.28833839297294617,
      "learning_rate": 8.598072698885546e-06,
      "loss": 0.0031,
      "step": 58650
    },
    {
      "epoch": 0.9598298290108811,
      "grad_norm": 0.3035374879837036,
      "learning_rate": 8.597411631468861e-06,
      "loss": 0.0044,
      "step": 58660
    },
    {
      "epoch": 0.959993454961957,
      "grad_norm": 0.0633229985833168,
      "learning_rate": 8.596750433654649e-06,
      "loss": 0.0029,
      "step": 58670
    },
    {
      "epoch": 0.9601570809130328,
      "grad_norm": 0.22874128818511963,
      "learning_rate": 8.596089105466878e-06,
      "loss": 0.0041,
      "step": 58680
    },
    {
      "epoch": 0.9603207068641086,
      "grad_norm": 0.09386830776929855,
      "learning_rate": 8.595427646929521e-06,
      "loss": 0.0027,
      "step": 58690
    },
    {
      "epoch": 0.9604843328151845,
      "grad_norm": 0.11341681331396103,
      "learning_rate": 8.594766058066555e-06,
      "loss": 0.003,
      "step": 58700
    },
    {
      "epoch": 0.9606479587662603,
      "grad_norm": 0.2628450393676758,
      "learning_rate": 8.594104338901957e-06,
      "loss": 0.0024,
      "step": 58710
    },
    {
      "epoch": 0.9608115847173362,
      "grad_norm": 0.0664951279759407,
      "learning_rate": 8.593442489459714e-06,
      "loss": 0.0034,
      "step": 58720
    },
    {
      "epoch": 0.960975210668412,
      "grad_norm": 0.17633040249347687,
      "learning_rate": 8.59278050976382e-06,
      "loss": 0.0025,
      "step": 58730
    },
    {
      "epoch": 0.9611388366194878,
      "grad_norm": 0.13413134217262268,
      "learning_rate": 8.592118399838265e-06,
      "loss": 0.003,
      "step": 58740
    },
    {
      "epoch": 0.9613024625705637,
      "grad_norm": 0.1787150651216507,
      "learning_rate": 8.591456159707053e-06,
      "loss": 0.0021,
      "step": 58750
    },
    {
      "epoch": 0.9614660885216395,
      "grad_norm": 0.12327458709478378,
      "learning_rate": 8.590793789394186e-06,
      "loss": 0.0039,
      "step": 58760
    },
    {
      "epoch": 0.9616297144727154,
      "grad_norm": 0.12161486595869064,
      "learning_rate": 8.590131288923676e-06,
      "loss": 0.0025,
      "step": 58770
    },
    {
      "epoch": 0.9617933404237912,
      "grad_norm": 0.1718994826078415,
      "learning_rate": 8.589468658319535e-06,
      "loss": 0.0027,
      "step": 58780
    },
    {
      "epoch": 0.961956966374867,
      "grad_norm": 0.08942586183547974,
      "learning_rate": 8.588805897605781e-06,
      "loss": 0.0027,
      "step": 58790
    },
    {
      "epoch": 0.9621205923259429,
      "grad_norm": 0.17843519151210785,
      "learning_rate": 8.58814300680644e-06,
      "loss": 0.0033,
      "step": 58800
    },
    {
      "epoch": 0.9622842182770187,
      "grad_norm": 0.06974881887435913,
      "learning_rate": 8.587479985945537e-06,
      "loss": 0.0022,
      "step": 58810
    },
    {
      "epoch": 0.9624478442280946,
      "grad_norm": 0.13175734877586365,
      "learning_rate": 8.586816835047108e-06,
      "loss": 0.0022,
      "step": 58820
    },
    {
      "epoch": 0.9626114701791704,
      "grad_norm": 0.39447152614593506,
      "learning_rate": 8.58615355413519e-06,
      "loss": 0.0033,
      "step": 58830
    },
    {
      "epoch": 0.9627750961302463,
      "grad_norm": 0.07321786880493164,
      "learning_rate": 8.585490143233824e-06,
      "loss": 0.0024,
      "step": 58840
    },
    {
      "epoch": 0.9629387220813221,
      "grad_norm": 0.15211625397205353,
      "learning_rate": 8.584826602367058e-06,
      "loss": 0.0049,
      "step": 58850
    },
    {
      "epoch": 0.9631023480323979,
      "grad_norm": 0.23689234256744385,
      "learning_rate": 8.584162931558944e-06,
      "loss": 0.0036,
      "step": 58860
    },
    {
      "epoch": 0.9632659739834738,
      "grad_norm": 0.10103043913841248,
      "learning_rate": 8.583499130833536e-06,
      "loss": 0.0025,
      "step": 58870
    },
    {
      "epoch": 0.9634295999345496,
      "grad_norm": 0.17015494406223297,
      "learning_rate": 8.5828352002149e-06,
      "loss": 0.0035,
      "step": 58880
    },
    {
      "epoch": 0.9635932258856255,
      "grad_norm": 0.06825874745845795,
      "learning_rate": 8.582171139727098e-06,
      "loss": 0.0032,
      "step": 58890
    },
    {
      "epoch": 0.9637568518367013,
      "grad_norm": 0.4128832519054413,
      "learning_rate": 8.581506949394201e-06,
      "loss": 0.0019,
      "step": 58900
    },
    {
      "epoch": 0.9639204777877771,
      "grad_norm": 0.22421450912952423,
      "learning_rate": 8.580842629240288e-06,
      "loss": 0.0024,
      "step": 58910
    },
    {
      "epoch": 0.964084103738853,
      "grad_norm": 0.13705191016197205,
      "learning_rate": 8.58017817928943e-06,
      "loss": 0.002,
      "step": 58920
    },
    {
      "epoch": 0.9642477296899288,
      "grad_norm": 0.08183940500020981,
      "learning_rate": 8.579513599565723e-06,
      "loss": 0.0046,
      "step": 58930
    },
    {
      "epoch": 0.9644113556410047,
      "grad_norm": 0.05310506373643875,
      "learning_rate": 8.578848890093249e-06,
      "loss": 0.003,
      "step": 58940
    },
    {
      "epoch": 0.9645749815920805,
      "grad_norm": 0.3109532594680786,
      "learning_rate": 8.578184050896106e-06,
      "loss": 0.002,
      "step": 58950
    },
    {
      "epoch": 0.9647386075431563,
      "grad_norm": 0.3090008795261383,
      "learning_rate": 8.577519081998389e-06,
      "loss": 0.0028,
      "step": 58960
    },
    {
      "epoch": 0.9649022334942322,
      "grad_norm": 0.376575231552124,
      "learning_rate": 8.576853983424206e-06,
      "loss": 0.0048,
      "step": 58970
    },
    {
      "epoch": 0.965065859445308,
      "grad_norm": 0.266393780708313,
      "learning_rate": 8.57618875519766e-06,
      "loss": 0.0026,
      "step": 58980
    },
    {
      "epoch": 0.9652294853963839,
      "grad_norm": 0.24883608520030975,
      "learning_rate": 8.575523397342868e-06,
      "loss": 0.0031,
      "step": 58990
    },
    {
      "epoch": 0.9653931113474598,
      "grad_norm": 0.12536068260669708,
      "learning_rate": 8.574857909883946e-06,
      "loss": 0.0026,
      "step": 59000
    },
    {
      "epoch": 0.9655567372985355,
      "grad_norm": 0.1851234883069992,
      "learning_rate": 8.574192292845019e-06,
      "loss": 0.0026,
      "step": 59010
    },
    {
      "epoch": 0.9657203632496114,
      "grad_norm": 0.22310858964920044,
      "learning_rate": 8.573526546250209e-06,
      "loss": 0.0028,
      "step": 59020
    },
    {
      "epoch": 0.9658839892006872,
      "grad_norm": 0.05807534232735634,
      "learning_rate": 8.572860670123651e-06,
      "loss": 0.0022,
      "step": 59030
    },
    {
      "epoch": 0.9660476151517631,
      "grad_norm": 0.06185042858123779,
      "learning_rate": 8.57219466448948e-06,
      "loss": 0.0049,
      "step": 59040
    },
    {
      "epoch": 0.966211241102839,
      "grad_norm": 0.5232572555541992,
      "learning_rate": 8.571528529371842e-06,
      "loss": 0.0035,
      "step": 59050
    },
    {
      "epoch": 0.9663748670539147,
      "grad_norm": 0.07686765491962433,
      "learning_rate": 8.570862264794875e-06,
      "loss": 0.0024,
      "step": 59060
    },
    {
      "epoch": 0.9665384930049906,
      "grad_norm": 0.08040457218885422,
      "learning_rate": 8.570195870782734e-06,
      "loss": 0.0028,
      "step": 59070
    },
    {
      "epoch": 0.9667021189560664,
      "grad_norm": 0.0629778727889061,
      "learning_rate": 8.569529347359574e-06,
      "loss": 0.0028,
      "step": 59080
    },
    {
      "epoch": 0.9668657449071423,
      "grad_norm": 0.07602206617593765,
      "learning_rate": 8.568862694549554e-06,
      "loss": 0.0025,
      "step": 59090
    },
    {
      "epoch": 0.9670293708582182,
      "grad_norm": 0.13948380947113037,
      "learning_rate": 8.568195912376838e-06,
      "loss": 0.0039,
      "step": 59100
    },
    {
      "epoch": 0.9671929968092939,
      "grad_norm": 0.27862828969955444,
      "learning_rate": 8.567529000865598e-06,
      "loss": 0.0014,
      "step": 59110
    },
    {
      "epoch": 0.9673566227603698,
      "grad_norm": 0.09404278546571732,
      "learning_rate": 8.566861960040005e-06,
      "loss": 0.0043,
      "step": 59120
    },
    {
      "epoch": 0.9675202487114456,
      "grad_norm": 0.10638560354709625,
      "learning_rate": 8.566194789924237e-06,
      "loss": 0.002,
      "step": 59130
    },
    {
      "epoch": 0.9676838746625215,
      "grad_norm": 0.1763361692428589,
      "learning_rate": 8.565527490542482e-06,
      "loss": 0.0027,
      "step": 59140
    },
    {
      "epoch": 0.9678475006135974,
      "grad_norm": 0.2545202672481537,
      "learning_rate": 8.564860061918924e-06,
      "loss": 0.0036,
      "step": 59150
    },
    {
      "epoch": 0.9680111265646731,
      "grad_norm": 0.12541721761226654,
      "learning_rate": 8.564192504077755e-06,
      "loss": 0.0023,
      "step": 59160
    },
    {
      "epoch": 0.968174752515749,
      "grad_norm": 0.21643178164958954,
      "learning_rate": 8.563524817043176e-06,
      "loss": 0.0021,
      "step": 59170
    },
    {
      "epoch": 0.9683383784668248,
      "grad_norm": 0.06688152253627777,
      "learning_rate": 8.562857000839386e-06,
      "loss": 0.0022,
      "step": 59180
    },
    {
      "epoch": 0.9685020044179007,
      "grad_norm": 0.08326565474271774,
      "learning_rate": 8.562189055490593e-06,
      "loss": 0.0024,
      "step": 59190
    },
    {
      "epoch": 0.9686656303689766,
      "grad_norm": 0.11828333884477615,
      "learning_rate": 8.561520981021009e-06,
      "loss": 0.0031,
      "step": 59200
    },
    {
      "epoch": 0.9688292563200523,
      "grad_norm": 0.10928668826818466,
      "learning_rate": 8.560852777454849e-06,
      "loss": 0.004,
      "step": 59210
    },
    {
      "epoch": 0.9689928822711282,
      "grad_norm": 0.21463769674301147,
      "learning_rate": 8.560184444816333e-06,
      "loss": 0.0039,
      "step": 59220
    },
    {
      "epoch": 0.969156508222204,
      "grad_norm": 0.28723961114883423,
      "learning_rate": 8.55951598312969e-06,
      "loss": 0.004,
      "step": 59230
    },
    {
      "epoch": 0.9693201341732799,
      "grad_norm": 0.205383762717247,
      "learning_rate": 8.558847392419145e-06,
      "loss": 0.0021,
      "step": 59240
    },
    {
      "epoch": 0.9694837601243558,
      "grad_norm": 0.06544926762580872,
      "learning_rate": 8.558178672708937e-06,
      "loss": 0.0026,
      "step": 59250
    },
    {
      "epoch": 0.9696473860754315,
      "grad_norm": 0.6217775940895081,
      "learning_rate": 8.557509824023303e-06,
      "loss": 0.0044,
      "step": 59260
    },
    {
      "epoch": 0.9698110120265074,
      "grad_norm": 0.2745841145515442,
      "learning_rate": 8.556840846386489e-06,
      "loss": 0.0022,
      "step": 59270
    },
    {
      "epoch": 0.9699746379775832,
      "grad_norm": 0.07420367002487183,
      "learning_rate": 8.556171739822742e-06,
      "loss": 0.0061,
      "step": 59280
    },
    {
      "epoch": 0.9701382639286591,
      "grad_norm": 0.22379092872142792,
      "learning_rate": 8.555502504356317e-06,
      "loss": 0.0025,
      "step": 59290
    },
    {
      "epoch": 0.970301889879735,
      "grad_norm": 0.21570985019207,
      "learning_rate": 8.554833140011473e-06,
      "loss": 0.0015,
      "step": 59300
    },
    {
      "epoch": 0.9704655158308108,
      "grad_norm": 0.15618379414081573,
      "learning_rate": 8.55416364681247e-06,
      "loss": 0.0032,
      "step": 59310
    },
    {
      "epoch": 0.9706291417818866,
      "grad_norm": 0.09433108568191528,
      "learning_rate": 8.553494024783578e-06,
      "loss": 0.0028,
      "step": 59320
    },
    {
      "epoch": 0.9707927677329624,
      "grad_norm": 0.17836423218250275,
      "learning_rate": 8.552824273949067e-06,
      "loss": 0.0053,
      "step": 59330
    },
    {
      "epoch": 0.9709563936840383,
      "grad_norm": 0.2055259644985199,
      "learning_rate": 8.552154394333216e-06,
      "loss": 0.0024,
      "step": 59340
    },
    {
      "epoch": 0.9711200196351142,
      "grad_norm": 0.18296679854393005,
      "learning_rate": 8.551484385960304e-06,
      "loss": 0.0036,
      "step": 59350
    },
    {
      "epoch": 0.97128364558619,
      "grad_norm": 0.486174613237381,
      "learning_rate": 8.55081424885462e-06,
      "loss": 0.0042,
      "step": 59360
    },
    {
      "epoch": 0.9714472715372658,
      "grad_norm": 0.053949691355228424,
      "learning_rate": 8.550143983040454e-06,
      "loss": 0.0016,
      "step": 59370
    },
    {
      "epoch": 0.9716108974883416,
      "grad_norm": 0.044979531317949295,
      "learning_rate": 8.549473588542101e-06,
      "loss": 0.0026,
      "step": 59380
    },
    {
      "epoch": 0.9717745234394175,
      "grad_norm": 0.2997584939002991,
      "learning_rate": 8.54880306538386e-06,
      "loss": 0.0026,
      "step": 59390
    },
    {
      "epoch": 0.9719381493904933,
      "grad_norm": 0.0504860021173954,
      "learning_rate": 8.548132413590038e-06,
      "loss": 0.0021,
      "step": 59400
    },
    {
      "epoch": 0.9721017753415692,
      "grad_norm": 0.27527013421058655,
      "learning_rate": 8.547461633184945e-06,
      "loss": 0.0043,
      "step": 59410
    },
    {
      "epoch": 0.972265401292645,
      "grad_norm": 0.1653563231229782,
      "learning_rate": 8.546790724192892e-06,
      "loss": 0.0034,
      "step": 59420
    },
    {
      "epoch": 0.9724290272437208,
      "grad_norm": 0.1331002116203308,
      "learning_rate": 8.546119686638202e-06,
      "loss": 0.0028,
      "step": 59430
    },
    {
      "epoch": 0.9725926531947967,
      "grad_norm": 0.13089890778064728,
      "learning_rate": 8.545448520545195e-06,
      "loss": 0.0022,
      "step": 59440
    },
    {
      "epoch": 0.9727562791458725,
      "grad_norm": 0.040818240493535995,
      "learning_rate": 8.544777225938199e-06,
      "loss": 0.0031,
      "step": 59450
    },
    {
      "epoch": 0.9729199050969484,
      "grad_norm": 0.03818698972463608,
      "learning_rate": 8.54410580284155e-06,
      "loss": 0.0021,
      "step": 59460
    },
    {
      "epoch": 0.9730835310480243,
      "grad_norm": 0.6097914576530457,
      "learning_rate": 8.543434251279583e-06,
      "loss": 0.0033,
      "step": 59470
    },
    {
      "epoch": 0.9732471569991,
      "grad_norm": 0.24187172949314117,
      "learning_rate": 8.542762571276641e-06,
      "loss": 0.0031,
      "step": 59480
    },
    {
      "epoch": 0.9734107829501759,
      "grad_norm": 0.24722379446029663,
      "learning_rate": 8.542090762857071e-06,
      "loss": 0.004,
      "step": 59490
    },
    {
      "epoch": 0.9735744089012517,
      "grad_norm": 0.185979425907135,
      "learning_rate": 8.541418826045223e-06,
      "loss": 0.0031,
      "step": 59500
    },
    {
      "epoch": 0.9737380348523276,
      "grad_norm": 0.09226076304912567,
      "learning_rate": 8.540746760865455e-06,
      "loss": 0.0035,
      "step": 59510
    },
    {
      "epoch": 0.9739016608034035,
      "grad_norm": 0.17793650925159454,
      "learning_rate": 8.540074567342126e-06,
      "loss": 0.0028,
      "step": 59520
    },
    {
      "epoch": 0.9740652867544792,
      "grad_norm": 0.12207861244678497,
      "learning_rate": 8.539402245499603e-06,
      "loss": 0.0031,
      "step": 59530
    },
    {
      "epoch": 0.9742289127055551,
      "grad_norm": 0.1498616486787796,
      "learning_rate": 8.538729795362255e-06,
      "loss": 0.0033,
      "step": 59540
    },
    {
      "epoch": 0.9743925386566309,
      "grad_norm": 0.12463486939668655,
      "learning_rate": 8.538057216954456e-06,
      "loss": 0.0036,
      "step": 59550
    },
    {
      "epoch": 0.9745561646077068,
      "grad_norm": 0.289331316947937,
      "learning_rate": 8.53738451030059e-06,
      "loss": 0.0031,
      "step": 59560
    },
    {
      "epoch": 0.9747197905587827,
      "grad_norm": 0.23395973443984985,
      "learning_rate": 8.536711675425034e-06,
      "loss": 0.004,
      "step": 59570
    },
    {
      "epoch": 0.9748834165098584,
      "grad_norm": 0.056549884378910065,
      "learning_rate": 8.536038712352181e-06,
      "loss": 0.0028,
      "step": 59580
    },
    {
      "epoch": 0.9750470424609343,
      "grad_norm": 0.17443841695785522,
      "learning_rate": 8.535365621106425e-06,
      "loss": 0.0024,
      "step": 59590
    },
    {
      "epoch": 0.9752106684120101,
      "grad_norm": 0.06441398710012436,
      "learning_rate": 8.53469240171216e-06,
      "loss": 0.0015,
      "step": 59600
    },
    {
      "epoch": 0.975374294363086,
      "grad_norm": 0.1061554029583931,
      "learning_rate": 8.53401905419379e-06,
      "loss": 0.003,
      "step": 59610
    },
    {
      "epoch": 0.9755379203141619,
      "grad_norm": 0.25653916597366333,
      "learning_rate": 8.533345578575724e-06,
      "loss": 0.0047,
      "step": 59620
    },
    {
      "epoch": 0.9757015462652376,
      "grad_norm": 0.37085962295532227,
      "learning_rate": 8.532671974882374e-06,
      "loss": 0.0033,
      "step": 59630
    },
    {
      "epoch": 0.9758651722163135,
      "grad_norm": 0.13060380518436432,
      "learning_rate": 8.531998243138155e-06,
      "loss": 0.0023,
      "step": 59640
    },
    {
      "epoch": 0.9760287981673893,
      "grad_norm": 0.11418633162975311,
      "learning_rate": 8.531324383367488e-06,
      "loss": 0.0028,
      "step": 59650
    },
    {
      "epoch": 0.9761924241184652,
      "grad_norm": 0.2217169851064682,
      "learning_rate": 8.530650395594801e-06,
      "loss": 0.003,
      "step": 59660
    },
    {
      "epoch": 0.9763560500695411,
      "grad_norm": 0.005907972808927298,
      "learning_rate": 8.529976279844521e-06,
      "loss": 0.002,
      "step": 59670
    },
    {
      "epoch": 0.9765196760206168,
      "grad_norm": 0.19005122780799866,
      "learning_rate": 8.529302036141087e-06,
      "loss": 0.0021,
      "step": 59680
    },
    {
      "epoch": 0.9766833019716927,
      "grad_norm": 0.17092862725257874,
      "learning_rate": 8.528627664508935e-06,
      "loss": 0.0056,
      "step": 59690
    },
    {
      "epoch": 0.9768469279227685,
      "grad_norm": 0.23192507028579712,
      "learning_rate": 8.527953164972511e-06,
      "loss": 0.0034,
      "step": 59700
    },
    {
      "epoch": 0.9770105538738444,
      "grad_norm": 0.03187968581914902,
      "learning_rate": 8.527278537556265e-06,
      "loss": 0.0026,
      "step": 59710
    },
    {
      "epoch": 0.9771741798249203,
      "grad_norm": 0.16410218179225922,
      "learning_rate": 8.52660378228465e-06,
      "loss": 0.0041,
      "step": 59720
    },
    {
      "epoch": 0.977337805775996,
      "grad_norm": 0.1285248100757599,
      "learning_rate": 8.52592889918212e-06,
      "loss": 0.0031,
      "step": 59730
    },
    {
      "epoch": 0.9775014317270719,
      "grad_norm": 0.22292162477970123,
      "learning_rate": 8.525253888273146e-06,
      "loss": 0.0022,
      "step": 59740
    },
    {
      "epoch": 0.9776650576781477,
      "grad_norm": 0.016681628301739693,
      "learning_rate": 8.524578749582189e-06,
      "loss": 0.0032,
      "step": 59750
    },
    {
      "epoch": 0.9778286836292236,
      "grad_norm": 0.35194265842437744,
      "learning_rate": 8.523903483133725e-06,
      "loss": 0.0029,
      "step": 59760
    },
    {
      "epoch": 0.9779923095802995,
      "grad_norm": 0.036682259291410446,
      "learning_rate": 8.523228088952229e-06,
      "loss": 0.0021,
      "step": 59770
    },
    {
      "epoch": 0.9781559355313753,
      "grad_norm": 0.25140687823295593,
      "learning_rate": 8.522552567062182e-06,
      "loss": 0.0032,
      "step": 59780
    },
    {
      "epoch": 0.9783195614824511,
      "grad_norm": 0.3055187165737152,
      "learning_rate": 8.52187691748807e-06,
      "loss": 0.0033,
      "step": 59790
    },
    {
      "epoch": 0.9784831874335269,
      "grad_norm": 0.2021966278553009,
      "learning_rate": 8.521201140254385e-06,
      "loss": 0.0018,
      "step": 59800
    },
    {
      "epoch": 0.9786468133846028,
      "grad_norm": 0.10952519625425339,
      "learning_rate": 8.520525235385622e-06,
      "loss": 0.0028,
      "step": 59810
    },
    {
      "epoch": 0.9788104393356787,
      "grad_norm": 0.050762154161930084,
      "learning_rate": 8.51984920290628e-06,
      "loss": 0.0026,
      "step": 59820
    },
    {
      "epoch": 0.9789740652867545,
      "grad_norm": 0.03261469304561615,
      "learning_rate": 8.519173042840866e-06,
      "loss": 0.0035,
      "step": 59830
    },
    {
      "epoch": 0.9791376912378303,
      "grad_norm": 0.09155498445034027,
      "learning_rate": 8.518496755213885e-06,
      "loss": 0.0038,
      "step": 59840
    },
    {
      "epoch": 0.9793013171889061,
      "grad_norm": 0.12844696640968323,
      "learning_rate": 8.517820340049853e-06,
      "loss": 0.0029,
      "step": 59850
    },
    {
      "epoch": 0.979464943139982,
      "grad_norm": 0.4480613172054291,
      "learning_rate": 8.51714379737329e-06,
      "loss": 0.003,
      "step": 59860
    },
    {
      "epoch": 0.9796285690910579,
      "grad_norm": 0.16266338527202606,
      "learning_rate": 8.516467127208718e-06,
      "loss": 0.0026,
      "step": 59870
    },
    {
      "epoch": 0.9797921950421337,
      "grad_norm": 0.06926774978637695,
      "learning_rate": 8.515790329580664e-06,
      "loss": 0.0023,
      "step": 59880
    },
    {
      "epoch": 0.9799558209932095,
      "grad_norm": 0.20574086904525757,
      "learning_rate": 8.51511340451366e-06,
      "loss": 0.0032,
      "step": 59890
    },
    {
      "epoch": 0.9801194469442853,
      "grad_norm": 0.13596323132514954,
      "learning_rate": 8.514436352032242e-06,
      "loss": 0.0026,
      "step": 59900
    },
    {
      "epoch": 0.9802830728953612,
      "grad_norm": 0.10957399755716324,
      "learning_rate": 8.513759172160955e-06,
      "loss": 0.0031,
      "step": 59910
    },
    {
      "epoch": 0.9804466988464371,
      "grad_norm": 0.22817716002464294,
      "learning_rate": 8.513081864924343e-06,
      "loss": 0.0027,
      "step": 59920
    },
    {
      "epoch": 0.9806103247975129,
      "grad_norm": 0.2074064165353775,
      "learning_rate": 8.512404430346955e-06,
      "loss": 0.0024,
      "step": 59930
    },
    {
      "epoch": 0.9807739507485888,
      "grad_norm": 0.007189332041889429,
      "learning_rate": 8.51172686845335e-06,
      "loss": 0.0029,
      "step": 59940
    },
    {
      "epoch": 0.9809375766996645,
      "grad_norm": 0.12334074825048447,
      "learning_rate": 8.511049179268085e-06,
      "loss": 0.0021,
      "step": 59950
    },
    {
      "epoch": 0.9811012026507404,
      "grad_norm": 0.05503307655453682,
      "learning_rate": 8.510371362815726e-06,
      "loss": 0.0029,
      "step": 59960
    },
    {
      "epoch": 0.9812648286018163,
      "grad_norm": 0.18143554031848907,
      "learning_rate": 8.509693419120844e-06,
      "loss": 0.0027,
      "step": 59970
    },
    {
      "epoch": 0.9814284545528921,
      "grad_norm": 0.07888347655534744,
      "learning_rate": 8.50901534820801e-06,
      "loss": 0.0022,
      "step": 59980
    },
    {
      "epoch": 0.981592080503968,
      "grad_norm": 0.3083827793598175,
      "learning_rate": 8.508337150101801e-06,
      "loss": 0.0047,
      "step": 59990
    },
    {
      "epoch": 0.9817557064550437,
      "grad_norm": 0.12481352686882019,
      "learning_rate": 8.507658824826805e-06,
      "loss": 0.0047,
      "step": 60000
    },
    {
      "epoch": 0.9819193324061196,
      "grad_norm": 0.4041292071342468,
      "learning_rate": 8.506980372407605e-06,
      "loss": 0.0041,
      "step": 60010
    },
    {
      "epoch": 0.9820829583571955,
      "grad_norm": 0.13590195775032043,
      "learning_rate": 8.506301792868798e-06,
      "loss": 0.003,
      "step": 60020
    },
    {
      "epoch": 0.9822465843082713,
      "grad_norm": 0.08283043652772903,
      "learning_rate": 8.505623086234976e-06,
      "loss": 0.0031,
      "step": 60030
    },
    {
      "epoch": 0.9824102102593472,
      "grad_norm": 0.08336950838565826,
      "learning_rate": 8.504944252530743e-06,
      "loss": 0.0034,
      "step": 60040
    },
    {
      "epoch": 0.9825738362104229,
      "grad_norm": 0.24332402646541595,
      "learning_rate": 8.504265291780706e-06,
      "loss": 0.0019,
      "step": 60050
    },
    {
      "epoch": 0.9827374621614988,
      "grad_norm": 0.10673798620700836,
      "learning_rate": 8.503586204009474e-06,
      "loss": 0.0035,
      "step": 60060
    },
    {
      "epoch": 0.9829010881125747,
      "grad_norm": 0.06880920380353928,
      "learning_rate": 8.502906989241662e-06,
      "loss": 0.0031,
      "step": 60070
    },
    {
      "epoch": 0.9830647140636505,
      "grad_norm": 0.07995596528053284,
      "learning_rate": 8.50222764750189e-06,
      "loss": 0.0016,
      "step": 60080
    },
    {
      "epoch": 0.9832283400147264,
      "grad_norm": 0.18849436938762665,
      "learning_rate": 8.501548178814785e-06,
      "loss": 0.0052,
      "step": 60090
    },
    {
      "epoch": 0.9833919659658021,
      "grad_norm": 0.1039353460073471,
      "learning_rate": 8.500868583204974e-06,
      "loss": 0.0029,
      "step": 60100
    },
    {
      "epoch": 0.983555591916878,
      "grad_norm": 0.673334538936615,
      "learning_rate": 8.500188860697089e-06,
      "loss": 0.0036,
      "step": 60110
    },
    {
      "epoch": 0.9837192178679539,
      "grad_norm": 0.1642826348543167,
      "learning_rate": 8.499509011315772e-06,
      "loss": 0.0038,
      "step": 60120
    },
    {
      "epoch": 0.9838828438190297,
      "grad_norm": 0.13992322981357574,
      "learning_rate": 8.498829035085663e-06,
      "loss": 0.0049,
      "step": 60130
    },
    {
      "epoch": 0.9840464697701056,
      "grad_norm": 0.18044134974479675,
      "learning_rate": 8.498148932031413e-06,
      "loss": 0.0026,
      "step": 60140
    },
    {
      "epoch": 0.9842100957211813,
      "grad_norm": 0.08346856385469437,
      "learning_rate": 8.497468702177669e-06,
      "loss": 0.0013,
      "step": 60150
    },
    {
      "epoch": 0.9843737216722572,
      "grad_norm": 0.29542285203933716,
      "learning_rate": 8.496788345549092e-06,
      "loss": 0.0044,
      "step": 60160
    },
    {
      "epoch": 0.9845373476233331,
      "grad_norm": 0.29454171657562256,
      "learning_rate": 8.496107862170341e-06,
      "loss": 0.0037,
      "step": 60170
    },
    {
      "epoch": 0.9847009735744089,
      "grad_norm": 0.026632007211446762,
      "learning_rate": 8.495427252066084e-06,
      "loss": 0.0019,
      "step": 60180
    },
    {
      "epoch": 0.9848645995254848,
      "grad_norm": 0.05433214455842972,
      "learning_rate": 8.49474651526099e-06,
      "loss": 0.0024,
      "step": 60190
    },
    {
      "epoch": 0.9850282254765605,
      "grad_norm": 0.027612656354904175,
      "learning_rate": 8.494065651779733e-06,
      "loss": 0.0019,
      "step": 60200
    },
    {
      "epoch": 0.9851918514276364,
      "grad_norm": 0.05650606006383896,
      "learning_rate": 8.493384661646995e-06,
      "loss": 0.0015,
      "step": 60210
    },
    {
      "epoch": 0.9853554773787123,
      "grad_norm": 0.26662370562553406,
      "learning_rate": 8.49270354488746e-06,
      "loss": 0.0017,
      "step": 60220
    },
    {
      "epoch": 0.9855191033297881,
      "grad_norm": 0.06338777393102646,
      "learning_rate": 8.492022301525812e-06,
      "loss": 0.0034,
      "step": 60230
    },
    {
      "epoch": 0.985682729280864,
      "grad_norm": 0.2385823279619217,
      "learning_rate": 8.491340931586753e-06,
      "loss": 0.0026,
      "step": 60240
    },
    {
      "epoch": 0.9858463552319398,
      "grad_norm": 0.36735841631889343,
      "learning_rate": 8.490659435094975e-06,
      "loss": 0.0024,
      "step": 60250
    },
    {
      "epoch": 0.9860099811830156,
      "grad_norm": 0.14980167150497437,
      "learning_rate": 8.489977812075181e-06,
      "loss": 0.0027,
      "step": 60260
    },
    {
      "epoch": 0.9861736071340914,
      "grad_norm": 0.20571166276931763,
      "learning_rate": 8.48929606255208e-06,
      "loss": 0.0022,
      "step": 60270
    },
    {
      "epoch": 0.9863372330851673,
      "grad_norm": 0.1867142617702484,
      "learning_rate": 8.488614186550382e-06,
      "loss": 0.0019,
      "step": 60280
    },
    {
      "epoch": 0.9865008590362432,
      "grad_norm": 0.15288801491260529,
      "learning_rate": 8.487932184094806e-06,
      "loss": 0.0044,
      "step": 60290
    },
    {
      "epoch": 0.986664484987319,
      "grad_norm": 0.13992652297019958,
      "learning_rate": 8.48725005521007e-06,
      "loss": 0.0032,
      "step": 60300
    },
    {
      "epoch": 0.9868281109383948,
      "grad_norm": 0.03404860198497772,
      "learning_rate": 8.486567799920901e-06,
      "loss": 0.0035,
      "step": 60310
    },
    {
      "epoch": 0.9869917368894706,
      "grad_norm": 0.031207703053951263,
      "learning_rate": 8.485885418252032e-06,
      "loss": 0.0043,
      "step": 60320
    },
    {
      "epoch": 0.9871553628405465,
      "grad_norm": 0.15165279805660248,
      "learning_rate": 8.48520291022819e-06,
      "loss": 0.0026,
      "step": 60330
    },
    {
      "epoch": 0.9873189887916224,
      "grad_norm": 0.2213481366634369,
      "learning_rate": 8.484520275874123e-06,
      "loss": 0.0024,
      "step": 60340
    },
    {
      "epoch": 0.9874826147426982,
      "grad_norm": 0.04767608642578125,
      "learning_rate": 8.483837515214568e-06,
      "loss": 0.003,
      "step": 60350
    },
    {
      "epoch": 0.987646240693774,
      "grad_norm": 0.26207292079925537,
      "learning_rate": 8.483154628274277e-06,
      "loss": 0.0059,
      "step": 60360
    },
    {
      "epoch": 0.9878098666448498,
      "grad_norm": 0.12950855493545532,
      "learning_rate": 8.482471615078002e-06,
      "loss": 0.0026,
      "step": 60370
    },
    {
      "epoch": 0.9879734925959257,
      "grad_norm": 0.1369142383337021,
      "learning_rate": 8.481788475650502e-06,
      "loss": 0.0031,
      "step": 60380
    },
    {
      "epoch": 0.9881371185470016,
      "grad_norm": 0.036217134445905685,
      "learning_rate": 8.481105210016538e-06,
      "loss": 0.002,
      "step": 60390
    },
    {
      "epoch": 0.9883007444980774,
      "grad_norm": 0.11564240604639053,
      "learning_rate": 8.480421818200874e-06,
      "loss": 0.002,
      "step": 60400
    },
    {
      "epoch": 0.9884643704491533,
      "grad_norm": 0.26017099618911743,
      "learning_rate": 8.479738300228287e-06,
      "loss": 0.0029,
      "step": 60410
    },
    {
      "epoch": 0.988627996400229,
      "grad_norm": 0.16226722300052643,
      "learning_rate": 8.47905465612355e-06,
      "loss": 0.0023,
      "step": 60420
    },
    {
      "epoch": 0.9887916223513049,
      "grad_norm": 0.25014492869377136,
      "learning_rate": 8.478370885911442e-06,
      "loss": 0.0023,
      "step": 60430
    },
    {
      "epoch": 0.9889552483023808,
      "grad_norm": 0.21477025747299194,
      "learning_rate": 8.477686989616751e-06,
      "loss": 0.0029,
      "step": 60440
    },
    {
      "epoch": 0.9891188742534566,
      "grad_norm": 0.1658000499010086,
      "learning_rate": 8.477002967264262e-06,
      "loss": 0.0024,
      "step": 60450
    },
    {
      "epoch": 0.9892825002045325,
      "grad_norm": 0.08058953285217285,
      "learning_rate": 8.476318818878775e-06,
      "loss": 0.0029,
      "step": 60460
    },
    {
      "epoch": 0.9894461261556082,
      "grad_norm": 0.34774577617645264,
      "learning_rate": 8.475634544485084e-06,
      "loss": 0.0024,
      "step": 60470
    },
    {
      "epoch": 0.9896097521066841,
      "grad_norm": 0.06716769188642502,
      "learning_rate": 8.474950144107998e-06,
      "loss": 0.0014,
      "step": 60480
    },
    {
      "epoch": 0.98977337805776,
      "grad_norm": 0.28167226910591125,
      "learning_rate": 8.474265617772317e-06,
      "loss": 0.003,
      "step": 60490
    },
    {
      "epoch": 0.9899370040088358,
      "grad_norm": 0.15282514691352844,
      "learning_rate": 8.47358096550286e-06,
      "loss": 0.0029,
      "step": 60500
    },
    {
      "epoch": 0.9901006299599117,
      "grad_norm": 0.09227076917886734,
      "learning_rate": 8.47289618732444e-06,
      "loss": 0.0023,
      "step": 60510
    },
    {
      "epoch": 0.9902642559109874,
      "grad_norm": 0.05908651277422905,
      "learning_rate": 8.47221128326188e-06,
      "loss": 0.0027,
      "step": 60520
    },
    {
      "epoch": 0.9904278818620633,
      "grad_norm": 0.02395007200539112,
      "learning_rate": 8.471526253340007e-06,
      "loss": 0.0028,
      "step": 60530
    },
    {
      "epoch": 0.9905915078131392,
      "grad_norm": 0.12977057695388794,
      "learning_rate": 8.470841097583651e-06,
      "loss": 0.0028,
      "step": 60540
    },
    {
      "epoch": 0.990755133764215,
      "grad_norm": 0.21846355497837067,
      "learning_rate": 8.470155816017646e-06,
      "loss": 0.0038,
      "step": 60550
    },
    {
      "epoch": 0.9909187597152909,
      "grad_norm": 0.05971357226371765,
      "learning_rate": 8.469470408666836e-06,
      "loss": 0.0024,
      "step": 60560
    },
    {
      "epoch": 0.9910823856663666,
      "grad_norm": 0.10828137397766113,
      "learning_rate": 8.46878487555606e-06,
      "loss": 0.0033,
      "step": 60570
    },
    {
      "epoch": 0.9912460116174425,
      "grad_norm": 0.0945863202214241,
      "learning_rate": 8.468099216710168e-06,
      "loss": 0.0025,
      "step": 60580
    },
    {
      "epoch": 0.9914096375685184,
      "grad_norm": 0.07669234275817871,
      "learning_rate": 8.467413432154018e-06,
      "loss": 0.0032,
      "step": 60590
    },
    {
      "epoch": 0.9915732635195942,
      "grad_norm": 0.2728372812271118,
      "learning_rate": 8.466727521912463e-06,
      "loss": 0.0039,
      "step": 60600
    },
    {
      "epoch": 0.9917368894706701,
      "grad_norm": 0.029599549248814583,
      "learning_rate": 8.466041486010365e-06,
      "loss": 0.0025,
      "step": 60610
    },
    {
      "epoch": 0.9919005154217458,
      "grad_norm": 0.08870405703783035,
      "learning_rate": 8.465355324472597e-06,
      "loss": 0.0027,
      "step": 60620
    },
    {
      "epoch": 0.9920641413728217,
      "grad_norm": 0.21733129024505615,
      "learning_rate": 8.464669037324026e-06,
      "loss": 0.0024,
      "step": 60630
    },
    {
      "epoch": 0.9922277673238976,
      "grad_norm": 0.09045561403036118,
      "learning_rate": 8.46398262458953e-06,
      "loss": 0.0028,
      "step": 60640
    },
    {
      "epoch": 0.9923913932749734,
      "grad_norm": 0.13382533192634583,
      "learning_rate": 8.46329608629399e-06,
      "loss": 0.0022,
      "step": 60650
    },
    {
      "epoch": 0.9925550192260493,
      "grad_norm": 0.18012291193008423,
      "learning_rate": 8.46260942246229e-06,
      "loss": 0.0042,
      "step": 60660
    },
    {
      "epoch": 0.992718645177125,
      "grad_norm": 0.23046180605888367,
      "learning_rate": 8.46192263311932e-06,
      "loss": 0.0024,
      "step": 60670
    },
    {
      "epoch": 0.9928822711282009,
      "grad_norm": 0.02615022286772728,
      "learning_rate": 8.461235718289976e-06,
      "loss": 0.0023,
      "step": 60680
    },
    {
      "epoch": 0.9930458970792768,
      "grad_norm": 0.110099658370018,
      "learning_rate": 8.460548677999153e-06,
      "loss": 0.0039,
      "step": 60690
    },
    {
      "epoch": 0.9932095230303526,
      "grad_norm": 0.18425911664962769,
      "learning_rate": 8.45986151227176e-06,
      "loss": 0.0026,
      "step": 60700
    },
    {
      "epoch": 0.9933731489814285,
      "grad_norm": 0.05597611516714096,
      "learning_rate": 8.459174221132704e-06,
      "loss": 0.0028,
      "step": 60710
    },
    {
      "epoch": 0.9935367749325043,
      "grad_norm": 0.10822423547506332,
      "learning_rate": 8.458486804606895e-06,
      "loss": 0.0025,
      "step": 60720
    },
    {
      "epoch": 0.9937004008835801,
      "grad_norm": 0.12967842817306519,
      "learning_rate": 8.457799262719254e-06,
      "loss": 0.0029,
      "step": 60730
    },
    {
      "epoch": 0.993864026834656,
      "grad_norm": 0.18580889701843262,
      "learning_rate": 8.457111595494697e-06,
      "loss": 0.0059,
      "step": 60740
    },
    {
      "epoch": 0.9940276527857318,
      "grad_norm": 0.08482278883457184,
      "learning_rate": 8.456423802958154e-06,
      "loss": 0.0029,
      "step": 60750
    },
    {
      "epoch": 0.9941912787368077,
      "grad_norm": 0.13644836843013763,
      "learning_rate": 8.455735885134558e-06,
      "loss": 0.0036,
      "step": 60760
    },
    {
      "epoch": 0.9943549046878835,
      "grad_norm": 0.06407059729099274,
      "learning_rate": 8.455047842048839e-06,
      "loss": 0.0031,
      "step": 60770
    },
    {
      "epoch": 0.9945185306389593,
      "grad_norm": 0.26606786251068115,
      "learning_rate": 8.45435967372594e-06,
      "loss": 0.0016,
      "step": 60780
    },
    {
      "epoch": 0.9946821565900352,
      "grad_norm": 0.19546149671077728,
      "learning_rate": 8.453671380190807e-06,
      "loss": 0.0042,
      "step": 60790
    },
    {
      "epoch": 0.994845782541111,
      "grad_norm": 0.19736286997795105,
      "learning_rate": 8.452982961468385e-06,
      "loss": 0.0037,
      "step": 60800
    },
    {
      "epoch": 0.9950094084921869,
      "grad_norm": 0.2795342206954956,
      "learning_rate": 8.45229441758363e-06,
      "loss": 0.0036,
      "step": 60810
    },
    {
      "epoch": 0.9951730344432627,
      "grad_norm": 0.18036170303821564,
      "learning_rate": 8.451605748561502e-06,
      "loss": 0.0014,
      "step": 60820
    },
    {
      "epoch": 0.9953366603943385,
      "grad_norm": 0.11996934562921524,
      "learning_rate": 8.45091695442696e-06,
      "loss": 0.0036,
      "step": 60830
    },
    {
      "epoch": 0.9955002863454144,
      "grad_norm": 0.08755335211753845,
      "learning_rate": 8.450228035204972e-06,
      "loss": 0.0024,
      "step": 60840
    },
    {
      "epoch": 0.9956639122964902,
      "grad_norm": 0.1050562858581543,
      "learning_rate": 8.44953899092051e-06,
      "loss": 0.0029,
      "step": 60850
    },
    {
      "epoch": 0.9958275382475661,
      "grad_norm": 0.3128628730773926,
      "learning_rate": 8.44884982159855e-06,
      "loss": 0.0053,
      "step": 60860
    },
    {
      "epoch": 0.9959911641986419,
      "grad_norm": 0.16530530154705048,
      "learning_rate": 8.448160527264075e-06,
      "loss": 0.0028,
      "step": 60870
    },
    {
      "epoch": 0.9961547901497178,
      "grad_norm": 0.203591451048851,
      "learning_rate": 8.447471107942066e-06,
      "loss": 0.0026,
      "step": 60880
    },
    {
      "epoch": 0.9963184161007936,
      "grad_norm": 0.11729732155799866,
      "learning_rate": 8.446781563657517e-06,
      "loss": 0.0014,
      "step": 60890
    },
    {
      "epoch": 0.9964820420518694,
      "grad_norm": 0.3393700420856476,
      "learning_rate": 8.44609189443542e-06,
      "loss": 0.0022,
      "step": 60900
    },
    {
      "epoch": 0.9966456680029453,
      "grad_norm": 0.05291201174259186,
      "learning_rate": 8.445402100300772e-06,
      "loss": 0.0033,
      "step": 60910
    },
    {
      "epoch": 0.9968092939540211,
      "grad_norm": 0.06167982146143913,
      "learning_rate": 8.444712181278583e-06,
      "loss": 0.0027,
      "step": 60920
    },
    {
      "epoch": 0.996972919905097,
      "grad_norm": 0.23836767673492432,
      "learning_rate": 8.444022137393854e-06,
      "loss": 0.0041,
      "step": 60930
    },
    {
      "epoch": 0.9971365458561728,
      "grad_norm": 0.21049663424491882,
      "learning_rate": 8.4433319686716e-06,
      "loss": 0.0035,
      "step": 60940
    },
    {
      "epoch": 0.9973001718072486,
      "grad_norm": 0.16214777529239655,
      "learning_rate": 8.442641675136838e-06,
      "loss": 0.0033,
      "step": 60950
    },
    {
      "epoch": 0.9974637977583245,
      "grad_norm": 0.08187725394964218,
      "learning_rate": 8.441951256814588e-06,
      "loss": 0.0041,
      "step": 60960
    },
    {
      "epoch": 0.9976274237094003,
      "grad_norm": 0.23726464807987213,
      "learning_rate": 8.441260713729879e-06,
      "loss": 0.0017,
      "step": 60970
    },
    {
      "epoch": 0.9977910496604762,
      "grad_norm": 0.029425758868455887,
      "learning_rate": 8.44057004590774e-06,
      "loss": 0.003,
      "step": 60980
    },
    {
      "epoch": 0.997954675611552,
      "grad_norm": 0.17834286391735077,
      "learning_rate": 8.439879253373207e-06,
      "loss": 0.0044,
      "step": 60990
    },
    {
      "epoch": 0.9981183015626278,
      "grad_norm": 0.08476901799440384,
      "learning_rate": 8.439188336151318e-06,
      "loss": 0.0016,
      "step": 61000
    },
    {
      "epoch": 0.9982819275137037,
      "grad_norm": 0.1598939448595047,
      "learning_rate": 8.438497294267117e-06,
      "loss": 0.0021,
      "step": 61010
    },
    {
      "epoch": 0.9984455534647795,
      "grad_norm": 0.1463746875524521,
      "learning_rate": 8.437806127745654e-06,
      "loss": 0.0025,
      "step": 61020
    },
    {
      "epoch": 0.9986091794158554,
      "grad_norm": 0.1795947551727295,
      "learning_rate": 8.437114836611981e-06,
      "loss": 0.002,
      "step": 61030
    },
    {
      "epoch": 0.9987728053669312,
      "grad_norm": 0.426880806684494,
      "learning_rate": 8.436423420891155e-06,
      "loss": 0.0027,
      "step": 61040
    },
    {
      "epoch": 0.998936431318007,
      "grad_norm": 0.2773195505142212,
      "learning_rate": 8.435731880608242e-06,
      "loss": 0.0018,
      "step": 61050
    },
    {
      "epoch": 0.9991000572690829,
      "grad_norm": 0.03795728459954262,
      "learning_rate": 8.435040215788303e-06,
      "loss": 0.0017,
      "step": 61060
    },
    {
      "epoch": 0.9992636832201587,
      "grad_norm": 0.3899232745170593,
      "learning_rate": 8.434348426456414e-06,
      "loss": 0.002,
      "step": 61070
    },
    {
      "epoch": 0.9994273091712346,
      "grad_norm": 0.04569382593035698,
      "learning_rate": 8.433656512637648e-06,
      "loss": 0.0029,
      "step": 61080
    },
    {
      "epoch": 0.9995909351223105,
      "grad_norm": 0.08163157105445862,
      "learning_rate": 8.432964474357085e-06,
      "loss": 0.0033,
      "step": 61090
    },
    {
      "epoch": 0.9997545610733862,
      "grad_norm": 0.14875875413417816,
      "learning_rate": 8.432272311639814e-06,
      "loss": 0.0016,
      "step": 61100
    },
    {
      "epoch": 0.9999181870244621,
      "grad_norm": 0.12114173918962479,
      "learning_rate": 8.431580024510917e-06,
      "loss": 0.0024,
      "step": 61110
    },
    {
      "epoch": 1.000081812975538,
      "grad_norm": 0.11625766009092331,
      "learning_rate": 8.430887612995495e-06,
      "loss": 0.0019,
      "step": 61120
    },
    {
      "epoch": 1.0002454389266138,
      "grad_norm": 0.17931455373764038,
      "learning_rate": 8.43019507711864e-06,
      "loss": 0.0029,
      "step": 61130
    },
    {
      "epoch": 1.0004090648776895,
      "grad_norm": 0.12281627207994461,
      "learning_rate": 8.42950241690546e-06,
      "loss": 0.002,
      "step": 61140
    },
    {
      "epoch": 1.0005726908287655,
      "grad_norm": 0.2563318908214569,
      "learning_rate": 8.428809632381059e-06,
      "loss": 0.0056,
      "step": 61150
    },
    {
      "epoch": 1.0007363167798413,
      "grad_norm": 0.3004699647426605,
      "learning_rate": 8.42811672357055e-06,
      "loss": 0.0014,
      "step": 61160
    },
    {
      "epoch": 1.000899942730917,
      "grad_norm": 0.1399531215429306,
      "learning_rate": 8.427423690499047e-06,
      "loss": 0.0026,
      "step": 61170
    },
    {
      "epoch": 1.0010635686819929,
      "grad_norm": 0.122503861784935,
      "learning_rate": 8.426730533191676e-06,
      "loss": 0.0015,
      "step": 61180
    },
    {
      "epoch": 1.0012271946330689,
      "grad_norm": 0.30409717559814453,
      "learning_rate": 8.426037251673558e-06,
      "loss": 0.0023,
      "step": 61190
    },
    {
      "epoch": 1.0013908205841446,
      "grad_norm": 0.07384887337684631,
      "learning_rate": 8.425343845969824e-06,
      "loss": 0.0034,
      "step": 61200
    },
    {
      "epoch": 1.0015544465352204,
      "grad_norm": 0.1534566730260849,
      "learning_rate": 8.424650316105607e-06,
      "loss": 0.0038,
      "step": 61210
    },
    {
      "epoch": 1.0017180724862964,
      "grad_norm": 0.1939021646976471,
      "learning_rate": 8.423956662106048e-06,
      "loss": 0.0025,
      "step": 61220
    },
    {
      "epoch": 1.0018816984373722,
      "grad_norm": 0.03150082379579544,
      "learning_rate": 8.42326288399629e-06,
      "loss": 0.0016,
      "step": 61230
    },
    {
      "epoch": 1.002045324388448,
      "grad_norm": 0.0625978410243988,
      "learning_rate": 8.422568981801478e-06,
      "loss": 0.0026,
      "step": 61240
    },
    {
      "epoch": 1.002208950339524,
      "grad_norm": 0.16247856616973877,
      "learning_rate": 8.421874955546769e-06,
      "loss": 0.0033,
      "step": 61250
    },
    {
      "epoch": 1.0023725762905997,
      "grad_norm": 0.0780845358967781,
      "learning_rate": 8.421180805257315e-06,
      "loss": 0.0015,
      "step": 61260
    },
    {
      "epoch": 1.0025362022416755,
      "grad_norm": 0.015654591843485832,
      "learning_rate": 8.420486530958278e-06,
      "loss": 0.0013,
      "step": 61270
    },
    {
      "epoch": 1.0026998281927513,
      "grad_norm": 0.18486538529396057,
      "learning_rate": 8.41979213267483e-06,
      "loss": 0.0029,
      "step": 61280
    },
    {
      "epoch": 1.0028634541438273,
      "grad_norm": 0.08054941892623901,
      "learning_rate": 8.419097610432133e-06,
      "loss": 0.0034,
      "step": 61290
    },
    {
      "epoch": 1.003027080094903,
      "grad_norm": 0.2680058777332306,
      "learning_rate": 8.418402964255366e-06,
      "loss": 0.004,
      "step": 61300
    },
    {
      "epoch": 1.0031907060459788,
      "grad_norm": 0.031576067209243774,
      "learning_rate": 8.417708194169707e-06,
      "loss": 0.0015,
      "step": 61310
    },
    {
      "epoch": 1.0033543319970548,
      "grad_norm": 0.0929732620716095,
      "learning_rate": 8.41701330020034e-06,
      "loss": 0.0023,
      "step": 61320
    },
    {
      "epoch": 1.0035179579481306,
      "grad_norm": 0.18532471358776093,
      "learning_rate": 8.416318282372455e-06,
      "loss": 0.0022,
      "step": 61330
    },
    {
      "epoch": 1.0036815838992064,
      "grad_norm": 0.2866421341896057,
      "learning_rate": 8.415623140711242e-06,
      "loss": 0.0031,
      "step": 61340
    },
    {
      "epoch": 1.0038452098502824,
      "grad_norm": 0.06344977021217346,
      "learning_rate": 8.4149278752419e-06,
      "loss": 0.0042,
      "step": 61350
    },
    {
      "epoch": 1.0040088358013581,
      "grad_norm": 0.3297552466392517,
      "learning_rate": 8.414232485989628e-06,
      "loss": 0.0026,
      "step": 61360
    },
    {
      "epoch": 1.004172461752434,
      "grad_norm": 0.0796428918838501,
      "learning_rate": 8.413536972979635e-06,
      "loss": 0.0023,
      "step": 61370
    },
    {
      "epoch": 1.0043360877035097,
      "grad_norm": 0.13391965627670288,
      "learning_rate": 8.412841336237134e-06,
      "loss": 0.0025,
      "step": 61380
    },
    {
      "epoch": 1.0044997136545857,
      "grad_norm": 0.15371493995189667,
      "learning_rate": 8.412145575787334e-06,
      "loss": 0.0058,
      "step": 61390
    },
    {
      "epoch": 1.0046633396056615,
      "grad_norm": 0.1355946958065033,
      "learning_rate": 8.411449691655457e-06,
      "loss": 0.0028,
      "step": 61400
    },
    {
      "epoch": 1.0048269655567372,
      "grad_norm": 0.16831043362617493,
      "learning_rate": 8.41075368386673e-06,
      "loss": 0.0023,
      "step": 61410
    },
    {
      "epoch": 1.0049905915078132,
      "grad_norm": 0.10733257979154587,
      "learning_rate": 8.41005755244638e-06,
      "loss": 0.0023,
      "step": 61420
    },
    {
      "epoch": 1.005154217458889,
      "grad_norm": 0.20408815145492554,
      "learning_rate": 8.409361297419639e-06,
      "loss": 0.003,
      "step": 61430
    },
    {
      "epoch": 1.0053178434099648,
      "grad_norm": 0.22214002907276154,
      "learning_rate": 8.408664918811744e-06,
      "loss": 0.0027,
      "step": 61440
    },
    {
      "epoch": 1.0054814693610408,
      "grad_norm": 0.07550564408302307,
      "learning_rate": 8.40796841664794e-06,
      "loss": 0.0041,
      "step": 61450
    },
    {
      "epoch": 1.0056450953121165,
      "grad_norm": 0.2276439368724823,
      "learning_rate": 8.407271790953471e-06,
      "loss": 0.0018,
      "step": 61460
    },
    {
      "epoch": 1.0058087212631923,
      "grad_norm": 0.12818531692028046,
      "learning_rate": 8.406575041753588e-06,
      "loss": 0.0015,
      "step": 61470
    },
    {
      "epoch": 1.005972347214268,
      "grad_norm": 0.30886825919151306,
      "learning_rate": 8.405878169073546e-06,
      "loss": 0.0021,
      "step": 61480
    },
    {
      "epoch": 1.006135973165344,
      "grad_norm": 0.21089695394039154,
      "learning_rate": 8.405181172938609e-06,
      "loss": 0.002,
      "step": 61490
    },
    {
      "epoch": 1.0062995991164199,
      "grad_norm": 0.4853508174419403,
      "learning_rate": 8.404484053374038e-06,
      "loss": 0.0043,
      "step": 61500
    },
    {
      "epoch": 1.0064632250674956,
      "grad_norm": 0.08074169605970383,
      "learning_rate": 8.4037868104051e-06,
      "loss": 0.0029,
      "step": 61510
    },
    {
      "epoch": 1.0066268510185716,
      "grad_norm": 0.21607299149036407,
      "learning_rate": 8.403089444057074e-06,
      "loss": 0.0023,
      "step": 61520
    },
    {
      "epoch": 1.0067904769696474,
      "grad_norm": 0.14379531145095825,
      "learning_rate": 8.402391954355235e-06,
      "loss": 0.0014,
      "step": 61530
    },
    {
      "epoch": 1.0069541029207232,
      "grad_norm": 0.12721051275730133,
      "learning_rate": 8.401694341324863e-06,
      "loss": 0.0025,
      "step": 61540
    },
    {
      "epoch": 1.007117728871799,
      "grad_norm": 0.069223552942276,
      "learning_rate": 8.400996604991247e-06,
      "loss": 0.0014,
      "step": 61550
    },
    {
      "epoch": 1.007281354822875,
      "grad_norm": 0.1583051234483719,
      "learning_rate": 8.400298745379679e-06,
      "loss": 0.0021,
      "step": 61560
    },
    {
      "epoch": 1.0074449807739507,
      "grad_norm": 0.07988758385181427,
      "learning_rate": 8.399600762515454e-06,
      "loss": 0.0036,
      "step": 61570
    },
    {
      "epoch": 1.0076086067250265,
      "grad_norm": 0.17742285132408142,
      "learning_rate": 8.398902656423872e-06,
      "loss": 0.0033,
      "step": 61580
    },
    {
      "epoch": 1.0077722326761025,
      "grad_norm": 0.08552506566047668,
      "learning_rate": 8.398204427130238e-06,
      "loss": 0.0034,
      "step": 61590
    },
    {
      "epoch": 1.0079358586271783,
      "grad_norm": 0.07694020122289658,
      "learning_rate": 8.397506074659861e-06,
      "loss": 0.0015,
      "step": 61600
    },
    {
      "epoch": 1.008099484578254,
      "grad_norm": 0.028427090495824814,
      "learning_rate": 8.396807599038053e-06,
      "loss": 0.0025,
      "step": 61610
    },
    {
      "epoch": 1.00826311052933,
      "grad_norm": 0.11284486204385757,
      "learning_rate": 8.396109000290133e-06,
      "loss": 0.0035,
      "step": 61620
    },
    {
      "epoch": 1.0084267364804058,
      "grad_norm": 0.052602652460336685,
      "learning_rate": 8.395410278441427e-06,
      "loss": 0.0016,
      "step": 61630
    },
    {
      "epoch": 1.0085903624314816,
      "grad_norm": 0.07618909329175949,
      "learning_rate": 8.394711433517257e-06,
      "loss": 0.0012,
      "step": 61640
    },
    {
      "epoch": 1.0087539883825574,
      "grad_norm": 0.10430283844470978,
      "learning_rate": 8.394012465542958e-06,
      "loss": 0.0043,
      "step": 61650
    },
    {
      "epoch": 1.0089176143336334,
      "grad_norm": 0.05425436794757843,
      "learning_rate": 8.393313374543863e-06,
      "loss": 0.0018,
      "step": 61660
    },
    {
      "epoch": 1.0090812402847091,
      "grad_norm": 0.05373930558562279,
      "learning_rate": 8.392614160545313e-06,
      "loss": 0.003,
      "step": 61670
    },
    {
      "epoch": 1.009244866235785,
      "grad_norm": 0.2939392924308777,
      "learning_rate": 8.391914823572655e-06,
      "loss": 0.0015,
      "step": 61680
    },
    {
      "epoch": 1.009408492186861,
      "grad_norm": 0.16351237893104553,
      "learning_rate": 8.391215363651237e-06,
      "loss": 0.002,
      "step": 61690
    },
    {
      "epoch": 1.0095721181379367,
      "grad_norm": 0.02641957253217697,
      "learning_rate": 8.390515780806412e-06,
      "loss": 0.0026,
      "step": 61700
    },
    {
      "epoch": 1.0097357440890125,
      "grad_norm": 0.12252381443977356,
      "learning_rate": 8.38981607506354e-06,
      "loss": 0.0025,
      "step": 61710
    },
    {
      "epoch": 1.0098993700400885,
      "grad_norm": 0.0973534882068634,
      "learning_rate": 8.389116246447981e-06,
      "loss": 0.0019,
      "step": 61720
    },
    {
      "epoch": 1.0100629959911642,
      "grad_norm": 0.21252557635307312,
      "learning_rate": 8.388416294985105e-06,
      "loss": 0.003,
      "step": 61730
    },
    {
      "epoch": 1.01022662194224,
      "grad_norm": 0.0555267371237278,
      "learning_rate": 8.387716220700282e-06,
      "loss": 0.001,
      "step": 61740
    },
    {
      "epoch": 1.0103902478933158,
      "grad_norm": 0.1361568421125412,
      "learning_rate": 8.387016023618889e-06,
      "loss": 0.0033,
      "step": 61750
    },
    {
      "epoch": 1.0105538738443918,
      "grad_norm": 0.32422080636024475,
      "learning_rate": 8.386315703766303e-06,
      "loss": 0.0016,
      "step": 61760
    },
    {
      "epoch": 1.0107174997954675,
      "grad_norm": 0.02642049267888069,
      "learning_rate": 8.385615261167912e-06,
      "loss": 0.0027,
      "step": 61770
    },
    {
      "epoch": 1.0108811257465433,
      "grad_norm": 0.08894288539886475,
      "learning_rate": 8.384914695849106e-06,
      "loss": 0.0016,
      "step": 61780
    },
    {
      "epoch": 1.0110447516976193,
      "grad_norm": 0.10418354719877243,
      "learning_rate": 8.384214007835278e-06,
      "loss": 0.0018,
      "step": 61790
    },
    {
      "epoch": 1.011208377648695,
      "grad_norm": 0.07337412983179092,
      "learning_rate": 8.383513197151825e-06,
      "loss": 0.0013,
      "step": 61800
    },
    {
      "epoch": 1.0113720035997709,
      "grad_norm": 0.06198417395353317,
      "learning_rate": 8.382812263824151e-06,
      "loss": 0.0023,
      "step": 61810
    },
    {
      "epoch": 1.0115356295508469,
      "grad_norm": 0.22408807277679443,
      "learning_rate": 8.382111207877663e-06,
      "loss": 0.001,
      "step": 61820
    },
    {
      "epoch": 1.0116992555019226,
      "grad_norm": 0.01795016974210739,
      "learning_rate": 8.381410029337774e-06,
      "loss": 0.0019,
      "step": 61830
    },
    {
      "epoch": 1.0118628814529984,
      "grad_norm": 0.11608357727527618,
      "learning_rate": 8.380708728229896e-06,
      "loss": 0.0015,
      "step": 61840
    },
    {
      "epoch": 1.0120265074040742,
      "grad_norm": 0.1601213961839676,
      "learning_rate": 8.380007304579451e-06,
      "loss": 0.0016,
      "step": 61850
    },
    {
      "epoch": 1.0121901333551502,
      "grad_norm": 0.29916322231292725,
      "learning_rate": 8.379305758411868e-06,
      "loss": 0.0025,
      "step": 61860
    },
    {
      "epoch": 1.012353759306226,
      "grad_norm": 0.18652862310409546,
      "learning_rate": 8.37860408975257e-06,
      "loss": 0.0024,
      "step": 61870
    },
    {
      "epoch": 1.0125173852573017,
      "grad_norm": 0.14501667022705078,
      "learning_rate": 8.377902298626997e-06,
      "loss": 0.0037,
      "step": 61880
    },
    {
      "epoch": 1.0126810112083777,
      "grad_norm": 0.04337180405855179,
      "learning_rate": 8.377200385060583e-06,
      "loss": 0.0021,
      "step": 61890
    },
    {
      "epoch": 1.0128446371594535,
      "grad_norm": 0.09240813553333282,
      "learning_rate": 8.376498349078772e-06,
      "loss": 0.0026,
      "step": 61900
    },
    {
      "epoch": 1.0130082631105293,
      "grad_norm": 0.0846519023180008,
      "learning_rate": 8.375796190707011e-06,
      "loss": 0.0024,
      "step": 61910
    },
    {
      "epoch": 1.0131718890616053,
      "grad_norm": 0.07384341955184937,
      "learning_rate": 8.375093909970751e-06,
      "loss": 0.0019,
      "step": 61920
    },
    {
      "epoch": 1.013335515012681,
      "grad_norm": 0.06110379099845886,
      "learning_rate": 8.37439150689545e-06,
      "loss": 0.0017,
      "step": 61930
    },
    {
      "epoch": 1.0134991409637568,
      "grad_norm": 0.2784845232963562,
      "learning_rate": 8.373688981506564e-06,
      "loss": 0.0038,
      "step": 61940
    },
    {
      "epoch": 1.0136627669148326,
      "grad_norm": 0.02236044779419899,
      "learning_rate": 8.372986333829565e-06,
      "loss": 0.0017,
      "step": 61950
    },
    {
      "epoch": 1.0138263928659086,
      "grad_norm": 0.09520578384399414,
      "learning_rate": 8.372283563889916e-06,
      "loss": 0.0026,
      "step": 61960
    },
    {
      "epoch": 1.0139900188169844,
      "grad_norm": 0.12372705340385437,
      "learning_rate": 8.371580671713092e-06,
      "loss": 0.0019,
      "step": 61970
    },
    {
      "epoch": 1.0141536447680601,
      "grad_norm": 0.17561574280261993,
      "learning_rate": 8.370877657324574e-06,
      "loss": 0.0017,
      "step": 61980
    },
    {
      "epoch": 1.0143172707191361,
      "grad_norm": 0.07626992464065552,
      "learning_rate": 8.37017452074984e-06,
      "loss": 0.0018,
      "step": 61990
    },
    {
      "epoch": 1.014480896670212,
      "grad_norm": 0.08037300407886505,
      "learning_rate": 8.369471262014383e-06,
      "loss": 0.0032,
      "step": 62000
    },
    {
      "epoch": 1.0146445226212877,
      "grad_norm": 0.19899636507034302,
      "learning_rate": 8.368767881143689e-06,
      "loss": 0.0012,
      "step": 62010
    },
    {
      "epoch": 1.0148081485723637,
      "grad_norm": 0.22281433641910553,
      "learning_rate": 8.368064378163255e-06,
      "loss": 0.0032,
      "step": 62020
    },
    {
      "epoch": 1.0149717745234395,
      "grad_norm": 0.16252335906028748,
      "learning_rate": 8.367360753098585e-06,
      "loss": 0.0027,
      "step": 62030
    },
    {
      "epoch": 1.0151354004745152,
      "grad_norm": 0.273134708404541,
      "learning_rate": 8.366657005975179e-06,
      "loss": 0.0025,
      "step": 62040
    },
    {
      "epoch": 1.015299026425591,
      "grad_norm": 0.09302778542041779,
      "learning_rate": 8.365953136818548e-06,
      "loss": 0.0021,
      "step": 62050
    },
    {
      "epoch": 1.015462652376667,
      "grad_norm": 0.1567562073469162,
      "learning_rate": 8.365249145654207e-06,
      "loss": 0.0025,
      "step": 62060
    },
    {
      "epoch": 1.0156262783277428,
      "grad_norm": 0.10178086906671524,
      "learning_rate": 8.364545032507672e-06,
      "loss": 0.0022,
      "step": 62070
    },
    {
      "epoch": 1.0157899042788185,
      "grad_norm": 0.16511395573616028,
      "learning_rate": 8.363840797404466e-06,
      "loss": 0.0021,
      "step": 62080
    },
    {
      "epoch": 1.0159535302298945,
      "grad_norm": 0.12475978583097458,
      "learning_rate": 8.363136440370115e-06,
      "loss": 0.0016,
      "step": 62090
    },
    {
      "epoch": 1.0161171561809703,
      "grad_norm": 0.2665903568267822,
      "learning_rate": 8.362431961430152e-06,
      "loss": 0.0024,
      "step": 62100
    },
    {
      "epoch": 1.016280782132046,
      "grad_norm": 0.13130570948123932,
      "learning_rate": 8.361727360610112e-06,
      "loss": 0.0032,
      "step": 62110
    },
    {
      "epoch": 1.016444408083122,
      "grad_norm": 0.04836617782711983,
      "learning_rate": 8.361022637935534e-06,
      "loss": 0.0029,
      "step": 62120
    },
    {
      "epoch": 1.0166080340341979,
      "grad_norm": 0.28147563338279724,
      "learning_rate": 8.360317793431964e-06,
      "loss": 0.0021,
      "step": 62130
    },
    {
      "epoch": 1.0167716599852736,
      "grad_norm": 0.05762926861643791,
      "learning_rate": 8.35961282712495e-06,
      "loss": 0.0028,
      "step": 62140
    },
    {
      "epoch": 1.0169352859363494,
      "grad_norm": 0.06845203787088394,
      "learning_rate": 8.358907739040046e-06,
      "loss": 0.003,
      "step": 62150
    },
    {
      "epoch": 1.0170989118874254,
      "grad_norm": 0.09512495249509811,
      "learning_rate": 8.358202529202809e-06,
      "loss": 0.0013,
      "step": 62160
    },
    {
      "epoch": 1.0172625378385012,
      "grad_norm": 0.19341616332530975,
      "learning_rate": 8.357497197638802e-06,
      "loss": 0.0024,
      "step": 62170
    },
    {
      "epoch": 1.017426163789577,
      "grad_norm": 0.31077906489372253,
      "learning_rate": 8.35679174437359e-06,
      "loss": 0.0017,
      "step": 62180
    },
    {
      "epoch": 1.017589789740653,
      "grad_norm": 0.2906704545021057,
      "learning_rate": 8.356086169432747e-06,
      "loss": 0.0024,
      "step": 62190
    },
    {
      "epoch": 1.0177534156917287,
      "grad_norm": 0.008857743814587593,
      "learning_rate": 8.355380472841845e-06,
      "loss": 0.0029,
      "step": 62200
    },
    {
      "epoch": 1.0179170416428045,
      "grad_norm": 0.00857104454189539,
      "learning_rate": 8.354674654626466e-06,
      "loss": 0.0044,
      "step": 62210
    },
    {
      "epoch": 1.0180806675938805,
      "grad_norm": 0.08437494188547134,
      "learning_rate": 8.353968714812193e-06,
      "loss": 0.0023,
      "step": 62220
    },
    {
      "epoch": 1.0182442935449563,
      "grad_norm": 0.23425538837909698,
      "learning_rate": 8.353262653424615e-06,
      "loss": 0.0031,
      "step": 62230
    },
    {
      "epoch": 1.018407919496032,
      "grad_norm": 0.06720089912414551,
      "learning_rate": 8.352556470489326e-06,
      "loss": 0.003,
      "step": 62240
    },
    {
      "epoch": 1.0185715454471078,
      "grad_norm": 0.256036639213562,
      "learning_rate": 8.351850166031922e-06,
      "loss": 0.002,
      "step": 62250
    },
    {
      "epoch": 1.0187351713981838,
      "grad_norm": 0.02549031563103199,
      "learning_rate": 8.351143740078005e-06,
      "loss": 0.0031,
      "step": 62260
    },
    {
      "epoch": 1.0188987973492596,
      "grad_norm": 1.3627597093582153,
      "learning_rate": 8.350437192653182e-06,
      "loss": 0.0026,
      "step": 62270
    },
    {
      "epoch": 1.0190624233003354,
      "grad_norm": 0.10078544169664383,
      "learning_rate": 8.349730523783065e-06,
      "loss": 0.0025,
      "step": 62280
    },
    {
      "epoch": 1.0192260492514114,
      "grad_norm": 0.2564089298248291,
      "learning_rate": 8.349023733493265e-06,
      "loss": 0.0032,
      "step": 62290
    },
    {
      "epoch": 1.0193896752024871,
      "grad_norm": 0.1104068011045456,
      "learning_rate": 8.348316821809405e-06,
      "loss": 0.0031,
      "step": 62300
    },
    {
      "epoch": 1.019553301153563,
      "grad_norm": 0.013330202549695969,
      "learning_rate": 8.347609788757107e-06,
      "loss": 0.0016,
      "step": 62310
    },
    {
      "epoch": 1.019716927104639,
      "grad_norm": 0.2409638613462448,
      "learning_rate": 8.346902634362002e-06,
      "loss": 0.0036,
      "step": 62320
    },
    {
      "epoch": 1.0198805530557147,
      "grad_norm": 0.17588280141353607,
      "learning_rate": 8.346195358649718e-06,
      "loss": 0.0031,
      "step": 62330
    },
    {
      "epoch": 1.0200441790067905,
      "grad_norm": 0.017134644091129303,
      "learning_rate": 8.345487961645896e-06,
      "loss": 0.0041,
      "step": 62340
    },
    {
      "epoch": 1.0202078049578662,
      "grad_norm": 0.224152073264122,
      "learning_rate": 8.344780443376176e-06,
      "loss": 0.0022,
      "step": 62350
    },
    {
      "epoch": 1.0203714309089422,
      "grad_norm": 0.11352168023586273,
      "learning_rate": 8.344072803866204e-06,
      "loss": 0.0015,
      "step": 62360
    },
    {
      "epoch": 1.020535056860018,
      "grad_norm": 0.07339297980070114,
      "learning_rate": 8.343365043141629e-06,
      "loss": 0.0026,
      "step": 62370
    },
    {
      "epoch": 1.0206986828110938,
      "grad_norm": 0.08674383163452148,
      "learning_rate": 8.342657161228108e-06,
      "loss": 0.0024,
      "step": 62380
    },
    {
      "epoch": 1.0208623087621698,
      "grad_norm": 0.2911551892757416,
      "learning_rate": 8.341949158151298e-06,
      "loss": 0.0033,
      "step": 62390
    },
    {
      "epoch": 1.0210259347132455,
      "grad_norm": 0.12789283692836761,
      "learning_rate": 8.341241033936862e-06,
      "loss": 0.0036,
      "step": 62400
    },
    {
      "epoch": 1.0211895606643213,
      "grad_norm": 0.06214087828993797,
      "learning_rate": 8.340532788610471e-06,
      "loss": 0.0022,
      "step": 62410
    },
    {
      "epoch": 1.0213531866153973,
      "grad_norm": 0.08382989466190338,
      "learning_rate": 8.339824422197796e-06,
      "loss": 0.0029,
      "step": 62420
    },
    {
      "epoch": 1.021516812566473,
      "grad_norm": 0.27731460332870483,
      "learning_rate": 8.33911593472451e-06,
      "loss": 0.0042,
      "step": 62430
    },
    {
      "epoch": 1.0216804385175489,
      "grad_norm": 0.11863048374652863,
      "learning_rate": 8.338407326216299e-06,
      "loss": 0.0025,
      "step": 62440
    },
    {
      "epoch": 1.0218440644686246,
      "grad_norm": 0.3180948793888092,
      "learning_rate": 8.337698596698844e-06,
      "loss": 0.004,
      "step": 62450
    },
    {
      "epoch": 1.0220076904197006,
      "grad_norm": 0.07720239460468292,
      "learning_rate": 8.336989746197839e-06,
      "loss": 0.0071,
      "step": 62460
    },
    {
      "epoch": 1.0221713163707764,
      "grad_norm": 0.147596076130867,
      "learning_rate": 8.336280774738974e-06,
      "loss": 0.0025,
      "step": 62470
    },
    {
      "epoch": 1.0223349423218522,
      "grad_norm": 0.44029510021209717,
      "learning_rate": 8.33557168234795e-06,
      "loss": 0.0027,
      "step": 62480
    },
    {
      "epoch": 1.0224985682729282,
      "grad_norm": 0.02467416413128376,
      "learning_rate": 8.334862469050471e-06,
      "loss": 0.0024,
      "step": 62490
    },
    {
      "epoch": 1.022662194224004,
      "grad_norm": 0.1042647659778595,
      "learning_rate": 8.334153134872242e-06,
      "loss": 0.0015,
      "step": 62500
    },
    {
      "epoch": 1.0228258201750797,
      "grad_norm": 0.15093326568603516,
      "learning_rate": 8.333443679838972e-06,
      "loss": 0.0022,
      "step": 62510
    },
    {
      "epoch": 1.0229894461261555,
      "grad_norm": 0.09556972980499268,
      "learning_rate": 8.332734103976385e-06,
      "loss": 0.0019,
      "step": 62520
    },
    {
      "epoch": 1.0231530720772315,
      "grad_norm": 0.48387524485588074,
      "learning_rate": 8.332024407310195e-06,
      "loss": 0.0018,
      "step": 62530
    },
    {
      "epoch": 1.0233166980283073,
      "grad_norm": 0.2792567014694214,
      "learning_rate": 8.331314589866127e-06,
      "loss": 0.0025,
      "step": 62540
    },
    {
      "epoch": 1.023480323979383,
      "grad_norm": 0.06885558366775513,
      "learning_rate": 8.330604651669913e-06,
      "loss": 0.0015,
      "step": 62550
    },
    {
      "epoch": 1.023643949930459,
      "grad_norm": 0.25647929310798645,
      "learning_rate": 8.329894592747285e-06,
      "loss": 0.0027,
      "step": 62560
    },
    {
      "epoch": 1.0238075758815348,
      "grad_norm": 0.31204888224601746,
      "learning_rate": 8.329184413123982e-06,
      "loss": 0.0022,
      "step": 62570
    },
    {
      "epoch": 1.0239712018326106,
      "grad_norm": 0.21856515109539032,
      "learning_rate": 8.328474112825745e-06,
      "loss": 0.0027,
      "step": 62580
    },
    {
      "epoch": 1.0241348277836866,
      "grad_norm": 0.0903751477599144,
      "learning_rate": 8.327763691878321e-06,
      "loss": 0.0015,
      "step": 62590
    },
    {
      "epoch": 1.0242984537347624,
      "grad_norm": 0.0380975604057312,
      "learning_rate": 8.327053150307462e-06,
      "loss": 0.0033,
      "step": 62600
    },
    {
      "epoch": 1.0244620796858381,
      "grad_norm": 0.15065494179725647,
      "learning_rate": 8.326342488138921e-06,
      "loss": 0.003,
      "step": 62610
    },
    {
      "epoch": 1.024625705636914,
      "grad_norm": 0.055887166410684586,
      "learning_rate": 8.325631705398461e-06,
      "loss": 0.0021,
      "step": 62620
    },
    {
      "epoch": 1.02478933158799,
      "grad_norm": 0.09244561940431595,
      "learning_rate": 8.324920802111844e-06,
      "loss": 0.0016,
      "step": 62630
    },
    {
      "epoch": 1.0249529575390657,
      "grad_norm": 0.06201239302754402,
      "learning_rate": 8.32420977830484e-06,
      "loss": 0.002,
      "step": 62640
    },
    {
      "epoch": 1.0251165834901415,
      "grad_norm": 0.13963104784488678,
      "learning_rate": 8.32349863400322e-06,
      "loss": 0.0021,
      "step": 62650
    },
    {
      "epoch": 1.0252802094412174,
      "grad_norm": 0.07053521275520325,
      "learning_rate": 8.322787369232763e-06,
      "loss": 0.0049,
      "step": 62660
    },
    {
      "epoch": 1.0254438353922932,
      "grad_norm": 0.03840421885251999,
      "learning_rate": 8.322075984019248e-06,
      "loss": 0.0029,
      "step": 62670
    },
    {
      "epoch": 1.025607461343369,
      "grad_norm": 0.09541652351617813,
      "learning_rate": 8.321364478388467e-06,
      "loss": 0.002,
      "step": 62680
    },
    {
      "epoch": 1.025771087294445,
      "grad_norm": 0.08917645364999771,
      "learning_rate": 8.320652852366202e-06,
      "loss": 0.0015,
      "step": 62690
    },
    {
      "epoch": 1.0259347132455208,
      "grad_norm": 0.11140826344490051,
      "learning_rate": 8.319941105978255e-06,
      "loss": 0.003,
      "step": 62700
    },
    {
      "epoch": 1.0260983391965965,
      "grad_norm": 0.18213094770908356,
      "learning_rate": 8.319229239250422e-06,
      "loss": 0.0024,
      "step": 62710
    },
    {
      "epoch": 1.0262619651476723,
      "grad_norm": 0.07034864276647568,
      "learning_rate": 8.318517252208505e-06,
      "loss": 0.0018,
      "step": 62720
    },
    {
      "epoch": 1.0264255910987483,
      "grad_norm": 0.05921407416462898,
      "learning_rate": 8.317805144878315e-06,
      "loss": 0.0025,
      "step": 62730
    },
    {
      "epoch": 1.026589217049824,
      "grad_norm": 0.1356448382139206,
      "learning_rate": 8.31709291728566e-06,
      "loss": 0.0022,
      "step": 62740
    },
    {
      "epoch": 1.0267528430008999,
      "grad_norm": 0.0674639642238617,
      "learning_rate": 8.316380569456363e-06,
      "loss": 0.0015,
      "step": 62750
    },
    {
      "epoch": 1.0269164689519759,
      "grad_norm": 0.23950588703155518,
      "learning_rate": 8.315668101416239e-06,
      "loss": 0.0026,
      "step": 62760
    },
    {
      "epoch": 1.0270800949030516,
      "grad_norm": 0.13180133700370789,
      "learning_rate": 8.314955513191114e-06,
      "loss": 0.0014,
      "step": 62770
    },
    {
      "epoch": 1.0272437208541274,
      "grad_norm": 0.00595868518576026,
      "learning_rate": 8.314242804806822e-06,
      "loss": 0.0045,
      "step": 62780
    },
    {
      "epoch": 1.0274073468052034,
      "grad_norm": 0.034348130226135254,
      "learning_rate": 8.313529976289193e-06,
      "loss": 0.0012,
      "step": 62790
    },
    {
      "epoch": 1.0275709727562792,
      "grad_norm": 0.060016658157110214,
      "learning_rate": 8.312817027664065e-06,
      "loss": 0.0023,
      "step": 62800
    },
    {
      "epoch": 1.027734598707355,
      "grad_norm": 0.03407631069421768,
      "learning_rate": 8.312103958957282e-06,
      "loss": 0.0039,
      "step": 62810
    },
    {
      "epoch": 1.0278982246584307,
      "grad_norm": 0.04801015183329582,
      "learning_rate": 8.311390770194691e-06,
      "loss": 0.0023,
      "step": 62820
    },
    {
      "epoch": 1.0280618506095067,
      "grad_norm": 0.544994592666626,
      "learning_rate": 8.310677461402143e-06,
      "loss": 0.0024,
      "step": 62830
    },
    {
      "epoch": 1.0282254765605825,
      "grad_norm": 0.03717778995633125,
      "learning_rate": 8.309964032605495e-06,
      "loss": 0.0026,
      "step": 62840
    },
    {
      "epoch": 1.0283891025116583,
      "grad_norm": 0.023024654015898705,
      "learning_rate": 8.309250483830607e-06,
      "loss": 0.0015,
      "step": 62850
    },
    {
      "epoch": 1.0285527284627343,
      "grad_norm": 0.12822191417217255,
      "learning_rate": 8.308536815103341e-06,
      "loss": 0.0037,
      "step": 62860
    },
    {
      "epoch": 1.02871635441381,
      "grad_norm": 0.2804183065891266,
      "learning_rate": 8.307823026449568e-06,
      "loss": 0.0028,
      "step": 62870
    },
    {
      "epoch": 1.0288799803648858,
      "grad_norm": 0.0640629380941391,
      "learning_rate": 8.30710911789516e-06,
      "loss": 0.0012,
      "step": 62880
    },
    {
      "epoch": 1.0290436063159618,
      "grad_norm": 0.06334434449672699,
      "learning_rate": 8.306395089465995e-06,
      "loss": 0.0024,
      "step": 62890
    },
    {
      "epoch": 1.0292072322670376,
      "grad_norm": 0.12159447371959686,
      "learning_rate": 8.305680941187955e-06,
      "loss": 0.0026,
      "step": 62900
    },
    {
      "epoch": 1.0293708582181134,
      "grad_norm": 0.17732298374176025,
      "learning_rate": 8.304966673086925e-06,
      "loss": 0.0025,
      "step": 62910
    },
    {
      "epoch": 1.0295344841691891,
      "grad_norm": 0.15575121343135834,
      "learning_rate": 8.304252285188797e-06,
      "loss": 0.0029,
      "step": 62920
    },
    {
      "epoch": 1.0296981101202651,
      "grad_norm": 0.12282532453536987,
      "learning_rate": 8.303537777519466e-06,
      "loss": 0.0027,
      "step": 62930
    },
    {
      "epoch": 1.029861736071341,
      "grad_norm": 0.1006195917725563,
      "learning_rate": 8.302823150104829e-06,
      "loss": 0.0021,
      "step": 62940
    },
    {
      "epoch": 1.0300253620224167,
      "grad_norm": 0.09268511086702347,
      "learning_rate": 8.302108402970792e-06,
      "loss": 0.002,
      "step": 62950
    },
    {
      "epoch": 1.0301889879734927,
      "grad_norm": 0.08966106921434402,
      "learning_rate": 8.30139353614326e-06,
      "loss": 0.002,
      "step": 62960
    },
    {
      "epoch": 1.0303526139245685,
      "grad_norm": 0.03228648006916046,
      "learning_rate": 8.30067854964815e-06,
      "loss": 0.0019,
      "step": 62970
    },
    {
      "epoch": 1.0305162398756442,
      "grad_norm": 0.011323303915560246,
      "learning_rate": 8.299963443511373e-06,
      "loss": 0.0016,
      "step": 62980
    },
    {
      "epoch": 1.0306798658267202,
      "grad_norm": 0.06857464462518692,
      "learning_rate": 8.299248217758853e-06,
      "loss": 0.0023,
      "step": 62990
    },
    {
      "epoch": 1.030843491777796,
      "grad_norm": 0.22043435275554657,
      "learning_rate": 8.298532872416515e-06,
      "loss": 0.0019,
      "step": 63000
    },
    {
      "epoch": 1.0310071177288718,
      "grad_norm": 0.10400668531656265,
      "learning_rate": 8.297817407510289e-06,
      "loss": 0.0015,
      "step": 63010
    },
    {
      "epoch": 1.0311707436799475,
      "grad_norm": 0.20460723340511322,
      "learning_rate": 8.297101823066106e-06,
      "loss": 0.0036,
      "step": 63020
    },
    {
      "epoch": 1.0313343696310235,
      "grad_norm": 0.12914440035820007,
      "learning_rate": 8.296386119109908e-06,
      "loss": 0.0017,
      "step": 63030
    },
    {
      "epoch": 1.0314979955820993,
      "grad_norm": 0.12779386341571808,
      "learning_rate": 8.295670295667635e-06,
      "loss": 0.0026,
      "step": 63040
    },
    {
      "epoch": 1.031661621533175,
      "grad_norm": 0.1792525351047516,
      "learning_rate": 8.294954352765235e-06,
      "loss": 0.0027,
      "step": 63050
    },
    {
      "epoch": 1.031825247484251,
      "grad_norm": 0.19937963783740997,
      "learning_rate": 8.29423829042866e-06,
      "loss": 0.0019,
      "step": 63060
    },
    {
      "epoch": 1.0319888734353269,
      "grad_norm": 0.03238700330257416,
      "learning_rate": 8.293522108683863e-06,
      "loss": 0.0018,
      "step": 63070
    },
    {
      "epoch": 1.0321524993864026,
      "grad_norm": 0.15098980069160461,
      "learning_rate": 8.292805807556806e-06,
      "loss": 0.0058,
      "step": 63080
    },
    {
      "epoch": 1.0323161253374786,
      "grad_norm": 0.06254418939352036,
      "learning_rate": 8.292089387073452e-06,
      "loss": 0.0025,
      "step": 63090
    },
    {
      "epoch": 1.0324797512885544,
      "grad_norm": 0.1866140365600586,
      "learning_rate": 8.29137284725977e-06,
      "loss": 0.0024,
      "step": 63100
    },
    {
      "epoch": 1.0326433772396302,
      "grad_norm": 0.11695242673158646,
      "learning_rate": 8.290656188141733e-06,
      "loss": 0.0021,
      "step": 63110
    },
    {
      "epoch": 1.032807003190706,
      "grad_norm": 0.16276447474956512,
      "learning_rate": 8.289939409745317e-06,
      "loss": 0.0019,
      "step": 63120
    },
    {
      "epoch": 1.032970629141782,
      "grad_norm": 0.1279701590538025,
      "learning_rate": 8.289222512096508e-06,
      "loss": 0.0022,
      "step": 63130
    },
    {
      "epoch": 1.0331342550928577,
      "grad_norm": 0.19693182408809662,
      "learning_rate": 8.288505495221286e-06,
      "loss": 0.0023,
      "step": 63140
    },
    {
      "epoch": 1.0332978810439335,
      "grad_norm": 0.13077808916568756,
      "learning_rate": 8.287788359145643e-06,
      "loss": 0.0024,
      "step": 63150
    },
    {
      "epoch": 1.0334615069950095,
      "grad_norm": 0.17228233814239502,
      "learning_rate": 8.287071103895575e-06,
      "loss": 0.0031,
      "step": 63160
    },
    {
      "epoch": 1.0336251329460853,
      "grad_norm": 0.03512703627347946,
      "learning_rate": 8.28635372949708e-06,
      "loss": 0.0015,
      "step": 63170
    },
    {
      "epoch": 1.033788758897161,
      "grad_norm": 0.23893366754055023,
      "learning_rate": 8.28563623597616e-06,
      "loss": 0.003,
      "step": 63180
    },
    {
      "epoch": 1.033952384848237,
      "grad_norm": 0.20334574580192566,
      "learning_rate": 8.284918623358824e-06,
      "loss": 0.002,
      "step": 63190
    },
    {
      "epoch": 1.0341160107993128,
      "grad_norm": 0.16438986361026764,
      "learning_rate": 8.284200891671083e-06,
      "loss": 0.002,
      "step": 63200
    },
    {
      "epoch": 1.0342796367503886,
      "grad_norm": 0.06547048687934875,
      "learning_rate": 8.283483040938954e-06,
      "loss": 0.0021,
      "step": 63210
    },
    {
      "epoch": 1.0344432627014644,
      "grad_norm": 0.19469082355499268,
      "learning_rate": 8.282765071188453e-06,
      "loss": 0.0019,
      "step": 63220
    },
    {
      "epoch": 1.0346068886525404,
      "grad_norm": 0.13101544976234436,
      "learning_rate": 8.282046982445611e-06,
      "loss": 0.0024,
      "step": 63230
    },
    {
      "epoch": 1.0347705146036161,
      "grad_norm": 0.11263741552829742,
      "learning_rate": 8.281328774736452e-06,
      "loss": 0.0021,
      "step": 63240
    },
    {
      "epoch": 1.034934140554692,
      "grad_norm": 0.13953259587287903,
      "learning_rate": 8.280610448087013e-06,
      "loss": 0.0022,
      "step": 63250
    },
    {
      "epoch": 1.035097766505768,
      "grad_norm": 0.14145800471305847,
      "learning_rate": 8.27989200252333e-06,
      "loss": 0.0024,
      "step": 63260
    },
    {
      "epoch": 1.0352613924568437,
      "grad_norm": 0.20280882716178894,
      "learning_rate": 8.279173438071444e-06,
      "loss": 0.0023,
      "step": 63270
    },
    {
      "epoch": 1.0354250184079195,
      "grad_norm": 0.27344730496406555,
      "learning_rate": 8.278454754757403e-06,
      "loss": 0.0028,
      "step": 63280
    },
    {
      "epoch": 1.0355886443589952,
      "grad_norm": 0.2773624062538147,
      "learning_rate": 8.277735952607257e-06,
      "loss": 0.0024,
      "step": 63290
    },
    {
      "epoch": 1.0357522703100712,
      "grad_norm": 0.07628427445888519,
      "learning_rate": 8.277017031647059e-06,
      "loss": 0.0018,
      "step": 63300
    },
    {
      "epoch": 1.035915896261147,
      "grad_norm": 0.13781200349330902,
      "learning_rate": 8.27629799190287e-06,
      "loss": 0.0013,
      "step": 63310
    },
    {
      "epoch": 1.0360795222122228,
      "grad_norm": 0.0606485940515995,
      "learning_rate": 8.275578833400754e-06,
      "loss": 0.0027,
      "step": 63320
    },
    {
      "epoch": 1.0362431481632988,
      "grad_norm": 0.18993821740150452,
      "learning_rate": 8.274859556166779e-06,
      "loss": 0.0025,
      "step": 63330
    },
    {
      "epoch": 1.0364067741143745,
      "grad_norm": 0.26492953300476074,
      "learning_rate": 8.274140160227016e-06,
      "loss": 0.0026,
      "step": 63340
    },
    {
      "epoch": 1.0365704000654503,
      "grad_norm": 0.1297852098941803,
      "learning_rate": 8.273420645607539e-06,
      "loss": 0.0025,
      "step": 63350
    },
    {
      "epoch": 1.0367340260165263,
      "grad_norm": 0.1854548156261444,
      "learning_rate": 8.272701012334432e-06,
      "loss": 0.0027,
      "step": 63360
    },
    {
      "epoch": 1.036897651967602,
      "grad_norm": 0.08247770369052887,
      "learning_rate": 8.271981260433779e-06,
      "loss": 0.0027,
      "step": 63370
    },
    {
      "epoch": 1.0370612779186779,
      "grad_norm": 0.02969399094581604,
      "learning_rate": 8.271261389931668e-06,
      "loss": 0.0011,
      "step": 63380
    },
    {
      "epoch": 1.0372249038697539,
      "grad_norm": 0.39014166593551636,
      "learning_rate": 8.270541400854197e-06,
      "loss": 0.0026,
      "step": 63390
    },
    {
      "epoch": 1.0373885298208296,
      "grad_norm": 0.1571214348077774,
      "learning_rate": 8.269821293227459e-06,
      "loss": 0.0029,
      "step": 63400
    },
    {
      "epoch": 1.0375521557719054,
      "grad_norm": 0.32384729385375977,
      "learning_rate": 8.269101067077558e-06,
      "loss": 0.0019,
      "step": 63410
    },
    {
      "epoch": 1.0377157817229812,
      "grad_norm": 0.0747719407081604,
      "learning_rate": 8.2683807224306e-06,
      "loss": 0.0027,
      "step": 63420
    },
    {
      "epoch": 1.0378794076740572,
      "grad_norm": 0.06614229828119278,
      "learning_rate": 8.267660259312697e-06,
      "loss": 0.0018,
      "step": 63430
    },
    {
      "epoch": 1.038043033625133,
      "grad_norm": 0.019833263009786606,
      "learning_rate": 8.266939677749964e-06,
      "loss": 0.0018,
      "step": 63440
    },
    {
      "epoch": 1.0382066595762087,
      "grad_norm": 0.15740786492824554,
      "learning_rate": 8.266218977768518e-06,
      "loss": 0.0025,
      "step": 63450
    },
    {
      "epoch": 1.0383702855272847,
      "grad_norm": 0.046031709760427475,
      "learning_rate": 8.265498159394486e-06,
      "loss": 0.0019,
      "step": 63460
    },
    {
      "epoch": 1.0385339114783605,
      "grad_norm": 0.20547205209732056,
      "learning_rate": 8.26477722265399e-06,
      "loss": 0.0063,
      "step": 63470
    },
    {
      "epoch": 1.0386975374294363,
      "grad_norm": 0.16334377229213715,
      "learning_rate": 8.26405616757317e-06,
      "loss": 0.0018,
      "step": 63480
    },
    {
      "epoch": 1.038861163380512,
      "grad_norm": 0.047372184693813324,
      "learning_rate": 8.263334994178158e-06,
      "loss": 0.0021,
      "step": 63490
    },
    {
      "epoch": 1.039024789331588,
      "grad_norm": 0.07817479223012924,
      "learning_rate": 8.262613702495096e-06,
      "loss": 0.0054,
      "step": 63500
    },
    {
      "epoch": 1.0391884152826638,
      "grad_norm": 0.024200119078159332,
      "learning_rate": 8.26189229255013e-06,
      "loss": 0.0033,
      "step": 63510
    },
    {
      "epoch": 1.0393520412337396,
      "grad_norm": 0.18235932290554047,
      "learning_rate": 8.261170764369405e-06,
      "loss": 0.0017,
      "step": 63520
    },
    {
      "epoch": 1.0395156671848156,
      "grad_norm": 0.40467819571495056,
      "learning_rate": 8.26044911797908e-06,
      "loss": 0.0033,
      "step": 63530
    },
    {
      "epoch": 1.0396792931358914,
      "grad_norm": 0.16961003839969635,
      "learning_rate": 8.25972735340531e-06,
      "loss": 0.0027,
      "step": 63540
    },
    {
      "epoch": 1.0398429190869671,
      "grad_norm": 0.2214461714029312,
      "learning_rate": 8.259005470674257e-06,
      "loss": 0.0019,
      "step": 63550
    },
    {
      "epoch": 1.0400065450380431,
      "grad_norm": 0.14852003753185272,
      "learning_rate": 8.258283469812091e-06,
      "loss": 0.0059,
      "step": 63560
    },
    {
      "epoch": 1.040170170989119,
      "grad_norm": 0.09031914174556732,
      "learning_rate": 8.257561350844977e-06,
      "loss": 0.001,
      "step": 63570
    },
    {
      "epoch": 1.0403337969401947,
      "grad_norm": 0.08425884693861008,
      "learning_rate": 8.256839113799097e-06,
      "loss": 0.0059,
      "step": 63580
    },
    {
      "epoch": 1.0404974228912705,
      "grad_norm": 0.15355518460273743,
      "learning_rate": 8.256116758700625e-06,
      "loss": 0.0026,
      "step": 63590
    },
    {
      "epoch": 1.0406610488423464,
      "grad_norm": 0.13370011746883392,
      "learning_rate": 8.255394285575746e-06,
      "loss": 0.0019,
      "step": 63600
    },
    {
      "epoch": 1.0408246747934222,
      "grad_norm": 0.07177018374204636,
      "learning_rate": 8.25467169445065e-06,
      "loss": 0.0016,
      "step": 63610
    },
    {
      "epoch": 1.040988300744498,
      "grad_norm": 0.2018873542547226,
      "learning_rate": 8.253948985351525e-06,
      "loss": 0.0017,
      "step": 63620
    },
    {
      "epoch": 1.041151926695574,
      "grad_norm": 0.1159633919596672,
      "learning_rate": 8.253226158304571e-06,
      "loss": 0.0024,
      "step": 63630
    },
    {
      "epoch": 1.0413155526466498,
      "grad_norm": 0.08404206484556198,
      "learning_rate": 8.252503213335989e-06,
      "loss": 0.0031,
      "step": 63640
    },
    {
      "epoch": 1.0414791785977255,
      "grad_norm": 0.030515529215335846,
      "learning_rate": 8.251780150471981e-06,
      "loss": 0.0032,
      "step": 63650
    },
    {
      "epoch": 1.0416428045488015,
      "grad_norm": 0.238947793841362,
      "learning_rate": 8.251056969738757e-06,
      "loss": 0.0017,
      "step": 63660
    },
    {
      "epoch": 1.0418064304998773,
      "grad_norm": 0.18806204199790955,
      "learning_rate": 8.250333671162534e-06,
      "loss": 0.0041,
      "step": 63670
    },
    {
      "epoch": 1.041970056450953,
      "grad_norm": 0.21754586696624756,
      "learning_rate": 8.249610254769527e-06,
      "loss": 0.0013,
      "step": 63680
    },
    {
      "epoch": 1.0421336824020289,
      "grad_norm": 0.09268508851528168,
      "learning_rate": 8.248886720585958e-06,
      "loss": 0.0037,
      "step": 63690
    },
    {
      "epoch": 1.0422973083531049,
      "grad_norm": 0.2234266847372055,
      "learning_rate": 8.248163068638055e-06,
      "loss": 0.0024,
      "step": 63700
    },
    {
      "epoch": 1.0424609343041806,
      "grad_norm": 0.19320163130760193,
      "learning_rate": 8.247439298952044e-06,
      "loss": 0.0035,
      "step": 63710
    },
    {
      "epoch": 1.0426245602552564,
      "grad_norm": 0.04447995871305466,
      "learning_rate": 8.246715411554167e-06,
      "loss": 0.0028,
      "step": 63720
    },
    {
      "epoch": 1.0427881862063324,
      "grad_norm": 0.02140369638800621,
      "learning_rate": 8.245991406470658e-06,
      "loss": 0.0021,
      "step": 63730
    },
    {
      "epoch": 1.0429518121574082,
      "grad_norm": 0.043271198868751526,
      "learning_rate": 8.245267283727762e-06,
      "loss": 0.003,
      "step": 63740
    },
    {
      "epoch": 1.043115438108484,
      "grad_norm": 0.08963941782712936,
      "learning_rate": 8.244543043351729e-06,
      "loss": 0.0027,
      "step": 63750
    },
    {
      "epoch": 1.04327906405956,
      "grad_norm": 0.0702454149723053,
      "learning_rate": 8.243818685368805e-06,
      "loss": 0.0026,
      "step": 63760
    },
    {
      "epoch": 1.0434426900106357,
      "grad_norm": 0.0729159265756607,
      "learning_rate": 8.243094209805254e-06,
      "loss": 0.0038,
      "step": 63770
    },
    {
      "epoch": 1.0436063159617115,
      "grad_norm": 0.0551057867705822,
      "learning_rate": 8.242369616687329e-06,
      "loss": 0.0032,
      "step": 63780
    },
    {
      "epoch": 1.0437699419127873,
      "grad_norm": 0.1939677894115448,
      "learning_rate": 8.241644906041299e-06,
      "loss": 0.0061,
      "step": 63790
    },
    {
      "epoch": 1.0439335678638633,
      "grad_norm": 0.05500328168272972,
      "learning_rate": 8.240920077893433e-06,
      "loss": 0.0016,
      "step": 63800
    },
    {
      "epoch": 1.044097193814939,
      "grad_norm": 0.294474333524704,
      "learning_rate": 8.240195132270003e-06,
      "loss": 0.0024,
      "step": 63810
    },
    {
      "epoch": 1.0442608197660148,
      "grad_norm": 0.07188540697097778,
      "learning_rate": 8.239470069197289e-06,
      "loss": 0.0014,
      "step": 63820
    },
    {
      "epoch": 1.0444244457170908,
      "grad_norm": 0.19571511447429657,
      "learning_rate": 8.23874488870157e-06,
      "loss": 0.003,
      "step": 63830
    },
    {
      "epoch": 1.0445880716681666,
      "grad_norm": 0.06561805307865143,
      "learning_rate": 8.238019590809131e-06,
      "loss": 0.0014,
      "step": 63840
    },
    {
      "epoch": 1.0447516976192424,
      "grad_norm": 0.05515031889081001,
      "learning_rate": 8.237294175546266e-06,
      "loss": 0.0021,
      "step": 63850
    },
    {
      "epoch": 1.0449153235703184,
      "grad_norm": 0.12624773383140564,
      "learning_rate": 8.236568642939267e-06,
      "loss": 0.0023,
      "step": 63860
    },
    {
      "epoch": 1.0450789495213941,
      "grad_norm": 0.13351145386695862,
      "learning_rate": 8.235842993014434e-06,
      "loss": 0.0028,
      "step": 63870
    },
    {
      "epoch": 1.04524257547247,
      "grad_norm": 0.06946561485528946,
      "learning_rate": 8.235117225798069e-06,
      "loss": 0.0034,
      "step": 63880
    },
    {
      "epoch": 1.0454062014235457,
      "grad_norm": 0.20091894268989563,
      "learning_rate": 8.23439134131648e-06,
      "loss": 0.0021,
      "step": 63890
    },
    {
      "epoch": 1.0455698273746217,
      "grad_norm": 0.08368875831365585,
      "learning_rate": 8.23366533959598e-06,
      "loss": 0.002,
      "step": 63900
    },
    {
      "epoch": 1.0457334533256974,
      "grad_norm": 0.2012326419353485,
      "learning_rate": 8.232939220662881e-06,
      "loss": 0.003,
      "step": 63910
    },
    {
      "epoch": 1.0458970792767732,
      "grad_norm": 0.03791200742125511,
      "learning_rate": 8.232212984543507e-06,
      "loss": 0.0018,
      "step": 63920
    },
    {
      "epoch": 1.0460607052278492,
      "grad_norm": 0.08582803606987,
      "learning_rate": 8.231486631264181e-06,
      "loss": 0.003,
      "step": 63930
    },
    {
      "epoch": 1.046224331178925,
      "grad_norm": 0.23705624043941498,
      "learning_rate": 8.23076016085123e-06,
      "loss": 0.0055,
      "step": 63940
    },
    {
      "epoch": 1.0463879571300008,
      "grad_norm": 0.10906633734703064,
      "learning_rate": 8.230033573330989e-06,
      "loss": 0.0027,
      "step": 63950
    },
    {
      "epoch": 1.0465515830810768,
      "grad_norm": 0.12108849734067917,
      "learning_rate": 8.229306868729793e-06,
      "loss": 0.0035,
      "step": 63960
    },
    {
      "epoch": 1.0467152090321525,
      "grad_norm": 0.06064796447753906,
      "learning_rate": 8.228580047073986e-06,
      "loss": 0.0021,
      "step": 63970
    },
    {
      "epoch": 1.0468788349832283,
      "grad_norm": 0.10400663316249847,
      "learning_rate": 8.227853108389911e-06,
      "loss": 0.0025,
      "step": 63980
    },
    {
      "epoch": 1.047042460934304,
      "grad_norm": 0.170183464884758,
      "learning_rate": 8.22712605270392e-06,
      "loss": 0.0031,
      "step": 63990
    },
    {
      "epoch": 1.04720608688538,
      "grad_norm": 0.12702810764312744,
      "learning_rate": 8.226398880042364e-06,
      "loss": 0.0028,
      "step": 64000
    },
    {
      "epoch": 1.0473697128364559,
      "grad_norm": 0.23696675896644592,
      "learning_rate": 8.225671590431604e-06,
      "loss": 0.0018,
      "step": 64010
    },
    {
      "epoch": 1.0475333387875316,
      "grad_norm": 0.054465774446725845,
      "learning_rate": 8.224944183898002e-06,
      "loss": 0.0032,
      "step": 64020
    },
    {
      "epoch": 1.0476969647386076,
      "grad_norm": 0.06987177580595016,
      "learning_rate": 8.224216660467924e-06,
      "loss": 0.0013,
      "step": 64030
    },
    {
      "epoch": 1.0478605906896834,
      "grad_norm": 0.12339454144239426,
      "learning_rate": 8.223489020167741e-06,
      "loss": 0.0018,
      "step": 64040
    },
    {
      "epoch": 1.0480242166407592,
      "grad_norm": 0.17792271077632904,
      "learning_rate": 8.222761263023829e-06,
      "loss": 0.0014,
      "step": 64050
    },
    {
      "epoch": 1.048187842591835,
      "grad_norm": 0.046114180237054825,
      "learning_rate": 8.222033389062565e-06,
      "loss": 0.0026,
      "step": 64060
    },
    {
      "epoch": 1.048351468542911,
      "grad_norm": 0.04991060122847557,
      "learning_rate": 8.221305398310336e-06,
      "loss": 0.0018,
      "step": 64070
    },
    {
      "epoch": 1.0485150944939867,
      "grad_norm": 0.22806178033351898,
      "learning_rate": 8.220577290793529e-06,
      "loss": 0.0027,
      "step": 64080
    },
    {
      "epoch": 1.0486787204450625,
      "grad_norm": 0.0364716611802578,
      "learning_rate": 8.219849066538535e-06,
      "loss": 0.0013,
      "step": 64090
    },
    {
      "epoch": 1.0488423463961385,
      "grad_norm": 0.1451956331729889,
      "learning_rate": 8.21912072557175e-06,
      "loss": 0.0041,
      "step": 64100
    },
    {
      "epoch": 1.0490059723472143,
      "grad_norm": 0.07945800572633743,
      "learning_rate": 8.218392267919578e-06,
      "loss": 0.0038,
      "step": 64110
    },
    {
      "epoch": 1.04916959829829,
      "grad_norm": 0.09665296226739883,
      "learning_rate": 8.217663693608419e-06,
      "loss": 0.0026,
      "step": 64120
    },
    {
      "epoch": 1.049333224249366,
      "grad_norm": 0.0728873461484909,
      "learning_rate": 8.216935002664688e-06,
      "loss": 0.0017,
      "step": 64130
    },
    {
      "epoch": 1.0494968502004418,
      "grad_norm": 0.15877701342105865,
      "learning_rate": 8.216206195114792e-06,
      "loss": 0.002,
      "step": 64140
    },
    {
      "epoch": 1.0496604761515176,
      "grad_norm": 0.10207909345626831,
      "learning_rate": 8.215477270985152e-06,
      "loss": 0.0023,
      "step": 64150
    },
    {
      "epoch": 1.0498241021025936,
      "grad_norm": 0.08115428686141968,
      "learning_rate": 8.21474823030219e-06,
      "loss": 0.0019,
      "step": 64160
    },
    {
      "epoch": 1.0499877280536694,
      "grad_norm": 0.19729632139205933,
      "learning_rate": 8.21401907309233e-06,
      "loss": 0.0029,
      "step": 64170
    },
    {
      "epoch": 1.0501513540047451,
      "grad_norm": 0.13712078332901,
      "learning_rate": 8.213289799382002e-06,
      "loss": 0.0031,
      "step": 64180
    },
    {
      "epoch": 1.050314979955821,
      "grad_norm": 0.1926317662000656,
      "learning_rate": 8.212560409197645e-06,
      "loss": 0.0022,
      "step": 64190
    },
    {
      "epoch": 1.050478605906897,
      "grad_norm": 0.20843687653541565,
      "learning_rate": 8.211830902565693e-06,
      "loss": 0.003,
      "step": 64200
    },
    {
      "epoch": 1.0506422318579727,
      "grad_norm": 0.23143619298934937,
      "learning_rate": 8.21110127951259e-06,
      "loss": 0.0023,
      "step": 64210
    },
    {
      "epoch": 1.0508058578090484,
      "grad_norm": 0.31759482622146606,
      "learning_rate": 8.210371540064783e-06,
      "loss": 0.0033,
      "step": 64220
    },
    {
      "epoch": 1.0509694837601244,
      "grad_norm": 0.053235169500112534,
      "learning_rate": 8.209641684248723e-06,
      "loss": 0.0014,
      "step": 64230
    },
    {
      "epoch": 1.0511331097112002,
      "grad_norm": 0.07178881764411926,
      "learning_rate": 8.208911712090868e-06,
      "loss": 0.0022,
      "step": 64240
    },
    {
      "epoch": 1.051296735662276,
      "grad_norm": 0.12044107913970947,
      "learning_rate": 8.208181623617675e-06,
      "loss": 0.0026,
      "step": 64250
    },
    {
      "epoch": 1.0514603616133518,
      "grad_norm": 0.1554144322872162,
      "learning_rate": 8.207451418855607e-06,
      "loss": 0.0027,
      "step": 64260
    },
    {
      "epoch": 1.0516239875644278,
      "grad_norm": 0.05538433417677879,
      "learning_rate": 8.206721097831136e-06,
      "loss": 0.0021,
      "step": 64270
    },
    {
      "epoch": 1.0517876135155035,
      "grad_norm": 0.10453752428293228,
      "learning_rate": 8.205990660570733e-06,
      "loss": 0.0018,
      "step": 64280
    },
    {
      "epoch": 1.0519512394665793,
      "grad_norm": 0.08280295878648758,
      "learning_rate": 8.205260107100872e-06,
      "loss": 0.0022,
      "step": 64290
    },
    {
      "epoch": 1.0521148654176553,
      "grad_norm": 0.08725742250680923,
      "learning_rate": 8.204529437448036e-06,
      "loss": 0.0026,
      "step": 64300
    },
    {
      "epoch": 1.052278491368731,
      "grad_norm": 0.0965590700507164,
      "learning_rate": 8.203798651638713e-06,
      "loss": 0.0026,
      "step": 64310
    },
    {
      "epoch": 1.0524421173198069,
      "grad_norm": 0.23796500265598297,
      "learning_rate": 8.203067749699384e-06,
      "loss": 0.0014,
      "step": 64320
    },
    {
      "epoch": 1.0526057432708829,
      "grad_norm": 0.020064018666744232,
      "learning_rate": 8.202336731656551e-06,
      "loss": 0.0029,
      "step": 64330
    },
    {
      "epoch": 1.0527693692219586,
      "grad_norm": 0.08776815235614777,
      "learning_rate": 8.201605597536707e-06,
      "loss": 0.0017,
      "step": 64340
    },
    {
      "epoch": 1.0529329951730344,
      "grad_norm": 0.12225066125392914,
      "learning_rate": 8.200874347366354e-06,
      "loss": 0.0023,
      "step": 64350
    },
    {
      "epoch": 1.0530966211241102,
      "grad_norm": 0.29338234663009644,
      "learning_rate": 8.200142981172001e-06,
      "loss": 0.002,
      "step": 64360
    },
    {
      "epoch": 1.0532602470751862,
      "grad_norm": 0.15350478887557983,
      "learning_rate": 8.199411498980155e-06,
      "loss": 0.0023,
      "step": 64370
    },
    {
      "epoch": 1.053423873026262,
      "grad_norm": 0.2017962485551834,
      "learning_rate": 8.198679900817333e-06,
      "loss": 0.0017,
      "step": 64380
    },
    {
      "epoch": 1.0535874989773377,
      "grad_norm": 0.011517062783241272,
      "learning_rate": 8.197948186710049e-06,
      "loss": 0.0018,
      "step": 64390
    },
    {
      "epoch": 1.0537511249284137,
      "grad_norm": 0.25941595435142517,
      "learning_rate": 8.197216356684833e-06,
      "loss": 0.002,
      "step": 64400
    },
    {
      "epoch": 1.0539147508794895,
      "grad_norm": 0.32127970457077026,
      "learning_rate": 8.196484410768206e-06,
      "loss": 0.003,
      "step": 64410
    },
    {
      "epoch": 1.0540783768305653,
      "grad_norm": 0.27373647689819336,
      "learning_rate": 8.195752348986703e-06,
      "loss": 0.0031,
      "step": 64420
    },
    {
      "epoch": 1.0542420027816413,
      "grad_norm": 0.1393657773733139,
      "learning_rate": 8.195020171366857e-06,
      "loss": 0.0024,
      "step": 64430
    },
    {
      "epoch": 1.054405628732717,
      "grad_norm": 0.18474701046943665,
      "learning_rate": 8.194287877935211e-06,
      "loss": 0.0017,
      "step": 64440
    },
    {
      "epoch": 1.0545692546837928,
      "grad_norm": 0.09622348099946976,
      "learning_rate": 8.193555468718306e-06,
      "loss": 0.0021,
      "step": 64450
    },
    {
      "epoch": 1.0547328806348686,
      "grad_norm": 0.14394673705101013,
      "learning_rate": 8.192822943742689e-06,
      "loss": 0.0021,
      "step": 64460
    },
    {
      "epoch": 1.0548965065859446,
      "grad_norm": 0.13619962334632874,
      "learning_rate": 8.192090303034916e-06,
      "loss": 0.002,
      "step": 64470
    },
    {
      "epoch": 1.0550601325370204,
      "grad_norm": 0.04721641167998314,
      "learning_rate": 8.19135754662154e-06,
      "loss": 0.0032,
      "step": 64480
    },
    {
      "epoch": 1.0552237584880961,
      "grad_norm": 0.1592308133840561,
      "learning_rate": 8.190624674529124e-06,
      "loss": 0.0015,
      "step": 64490
    },
    {
      "epoch": 1.0553873844391721,
      "grad_norm": 0.22303423285484314,
      "learning_rate": 8.189891686784232e-06,
      "loss": 0.0015,
      "step": 64500
    },
    {
      "epoch": 1.055551010390248,
      "grad_norm": 0.12353459745645523,
      "learning_rate": 8.189158583413432e-06,
      "loss": 0.0015,
      "step": 64510
    },
    {
      "epoch": 1.0557146363413237,
      "grad_norm": 0.1472569704055786,
      "learning_rate": 8.1884253644433e-06,
      "loss": 0.0037,
      "step": 64520
    },
    {
      "epoch": 1.0558782622923997,
      "grad_norm": 0.09601698815822601,
      "learning_rate": 8.18769202990041e-06,
      "loss": 0.0024,
      "step": 64530
    },
    {
      "epoch": 1.0560418882434754,
      "grad_norm": 0.08635096251964569,
      "learning_rate": 8.186958579811347e-06,
      "loss": 0.0022,
      "step": 64540
    },
    {
      "epoch": 1.0562055141945512,
      "grad_norm": 0.06166418641805649,
      "learning_rate": 8.186225014202694e-06,
      "loss": 0.0017,
      "step": 64550
    },
    {
      "epoch": 1.056369140145627,
      "grad_norm": 0.05661597102880478,
      "learning_rate": 8.185491333101043e-06,
      "loss": 0.0025,
      "step": 64560
    },
    {
      "epoch": 1.056532766096703,
      "grad_norm": 0.15421047806739807,
      "learning_rate": 8.184757536532985e-06,
      "loss": 0.0027,
      "step": 64570
    },
    {
      "epoch": 1.0566963920477788,
      "grad_norm": 0.20517005026340485,
      "learning_rate": 8.184023624525123e-06,
      "loss": 0.0026,
      "step": 64580
    },
    {
      "epoch": 1.0568600179988545,
      "grad_norm": 0.22030770778656006,
      "learning_rate": 8.183289597104058e-06,
      "loss": 0.0024,
      "step": 64590
    },
    {
      "epoch": 1.0570236439499305,
      "grad_norm": 0.05827048048377037,
      "learning_rate": 8.182555454296394e-06,
      "loss": 0.0014,
      "step": 64600
    },
    {
      "epoch": 1.0571872699010063,
      "grad_norm": 0.14805006980895996,
      "learning_rate": 8.181821196128745e-06,
      "loss": 0.0021,
      "step": 64610
    },
    {
      "epoch": 1.057350895852082,
      "grad_norm": 0.13481366634368896,
      "learning_rate": 8.181086822627724e-06,
      "loss": 0.0022,
      "step": 64620
    },
    {
      "epoch": 1.057514521803158,
      "grad_norm": 0.07633001357316971,
      "learning_rate": 8.180352333819953e-06,
      "loss": 0.0031,
      "step": 64630
    },
    {
      "epoch": 1.0576781477542339,
      "grad_norm": 0.19227145612239838,
      "learning_rate": 8.17961772973205e-06,
      "loss": 0.002,
      "step": 64640
    },
    {
      "epoch": 1.0578417737053096,
      "grad_norm": 0.038855452090501785,
      "learning_rate": 8.17888301039065e-06,
      "loss": 0.0021,
      "step": 64650
    },
    {
      "epoch": 1.0580053996563854,
      "grad_norm": 0.15322133898735046,
      "learning_rate": 8.17814817582238e-06,
      "loss": 0.0021,
      "step": 64660
    },
    {
      "epoch": 1.0581690256074614,
      "grad_norm": 0.1207018569111824,
      "learning_rate": 8.177413226053876e-06,
      "loss": 0.0015,
      "step": 64670
    },
    {
      "epoch": 1.0583326515585372,
      "grad_norm": 0.05391998961567879,
      "learning_rate": 8.17667816111178e-06,
      "loss": 0.0022,
      "step": 64680
    },
    {
      "epoch": 1.058496277509613,
      "grad_norm": 0.27931055426597595,
      "learning_rate": 8.175942981022737e-06,
      "loss": 0.0029,
      "step": 64690
    },
    {
      "epoch": 1.058659903460689,
      "grad_norm": 0.045851610600948334,
      "learning_rate": 8.175207685813392e-06,
      "loss": 0.0021,
      "step": 64700
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 0.07769758254289627,
      "learning_rate": 8.1744722755104e-06,
      "loss": 0.0018,
      "step": 64710
    },
    {
      "epoch": 1.0589871553628405,
      "grad_norm": 0.13252119719982147,
      "learning_rate": 8.17373675014042e-06,
      "loss": 0.0026,
      "step": 64720
    },
    {
      "epoch": 1.0591507813139165,
      "grad_norm": 0.3611391484737396,
      "learning_rate": 8.173001109730107e-06,
      "loss": 0.0018,
      "step": 64730
    },
    {
      "epoch": 1.0593144072649923,
      "grad_norm": 0.11499658226966858,
      "learning_rate": 8.172265354306132e-06,
      "loss": 0.0018,
      "step": 64740
    },
    {
      "epoch": 1.059478033216068,
      "grad_norm": 0.3381210267543793,
      "learning_rate": 8.171529483895161e-06,
      "loss": 0.0032,
      "step": 64750
    },
    {
      "epoch": 1.0596416591671438,
      "grad_norm": 0.0493021160364151,
      "learning_rate": 8.170793498523871e-06,
      "loss": 0.0025,
      "step": 64760
    },
    {
      "epoch": 1.0598052851182198,
      "grad_norm": 0.17737528681755066,
      "learning_rate": 8.170057398218936e-06,
      "loss": 0.0023,
      "step": 64770
    },
    {
      "epoch": 1.0599689110692956,
      "grad_norm": 0.01416482962667942,
      "learning_rate": 8.169321183007039e-06,
      "loss": 0.0027,
      "step": 64780
    },
    {
      "epoch": 1.0601325370203714,
      "grad_norm": 0.18234539031982422,
      "learning_rate": 8.168584852914869e-06,
      "loss": 0.0018,
      "step": 64790
    },
    {
      "epoch": 1.0602961629714474,
      "grad_norm": 0.049760852009058,
      "learning_rate": 8.16784840796911e-06,
      "loss": 0.0021,
      "step": 64800
    },
    {
      "epoch": 1.0604597889225231,
      "grad_norm": 0.11862653493881226,
      "learning_rate": 8.167111848196462e-06,
      "loss": 0.0037,
      "step": 64810
    },
    {
      "epoch": 1.060623414873599,
      "grad_norm": 0.021947648376226425,
      "learning_rate": 8.166375173623621e-06,
      "loss": 0.0042,
      "step": 64820
    },
    {
      "epoch": 1.060787040824675,
      "grad_norm": 0.20204272866249084,
      "learning_rate": 8.16563838427729e-06,
      "loss": 0.0023,
      "step": 64830
    },
    {
      "epoch": 1.0609506667757507,
      "grad_norm": 0.2470373958349228,
      "learning_rate": 8.164901480184177e-06,
      "loss": 0.0031,
      "step": 64840
    },
    {
      "epoch": 1.0611142927268264,
      "grad_norm": 0.10044754296541214,
      "learning_rate": 8.164164461370991e-06,
      "loss": 0.0033,
      "step": 64850
    },
    {
      "epoch": 1.0612779186779022,
      "grad_norm": 0.047854505479335785,
      "learning_rate": 8.163427327864449e-06,
      "loss": 0.0029,
      "step": 64860
    },
    {
      "epoch": 1.0614415446289782,
      "grad_norm": 0.1003831997513771,
      "learning_rate": 8.162690079691269e-06,
      "loss": 0.0019,
      "step": 64870
    },
    {
      "epoch": 1.061605170580054,
      "grad_norm": 0.11221947520971298,
      "learning_rate": 8.161952716878174e-06,
      "loss": 0.0017,
      "step": 64880
    },
    {
      "epoch": 1.0617687965311298,
      "grad_norm": 0.2193416953086853,
      "learning_rate": 8.161215239451893e-06,
      "loss": 0.0029,
      "step": 64890
    },
    {
      "epoch": 1.0619324224822058,
      "grad_norm": 0.016209201887249947,
      "learning_rate": 8.160477647439159e-06,
      "loss": 0.002,
      "step": 64900
    },
    {
      "epoch": 1.0620960484332815,
      "grad_norm": 0.2978039085865021,
      "learning_rate": 8.159739940866704e-06,
      "loss": 0.0022,
      "step": 64910
    },
    {
      "epoch": 1.0622596743843573,
      "grad_norm": 0.07247889041900635,
      "learning_rate": 8.15900211976127e-06,
      "loss": 0.0021,
      "step": 64920
    },
    {
      "epoch": 1.0624233003354333,
      "grad_norm": 0.15218892693519592,
      "learning_rate": 8.158264184149602e-06,
      "loss": 0.0019,
      "step": 64930
    },
    {
      "epoch": 1.062586926286509,
      "grad_norm": 0.13460765779018402,
      "learning_rate": 8.157526134058452e-06,
      "loss": 0.0018,
      "step": 64940
    },
    {
      "epoch": 1.0627505522375849,
      "grad_norm": 0.11842235177755356,
      "learning_rate": 8.156787969514564e-06,
      "loss": 0.0041,
      "step": 64950
    },
    {
      "epoch": 1.0629141781886606,
      "grad_norm": 0.1161569207906723,
      "learning_rate": 8.156049690544701e-06,
      "loss": 0.0028,
      "step": 64960
    },
    {
      "epoch": 1.0630778041397366,
      "grad_norm": 0.09794294834136963,
      "learning_rate": 8.15531129717562e-06,
      "loss": 0.0021,
      "step": 64970
    },
    {
      "epoch": 1.0632414300908124,
      "grad_norm": 0.06227337196469307,
      "learning_rate": 8.15457278943409e-06,
      "loss": 0.0025,
      "step": 64980
    },
    {
      "epoch": 1.0634050560418882,
      "grad_norm": 0.11521907150745392,
      "learning_rate": 8.15383416734688e-06,
      "loss": 0.0018,
      "step": 64990
    },
    {
      "epoch": 1.0635686819929642,
      "grad_norm": 0.15379422903060913,
      "learning_rate": 8.153095430940758e-06,
      "loss": 0.0025,
      "step": 65000
    },
    {
      "epoch": 1.06373230794404,
      "grad_norm": 0.14501148462295532,
      "learning_rate": 8.152356580242507e-06,
      "loss": 0.0032,
      "step": 65010
    },
    {
      "epoch": 1.0638959338951157,
      "grad_norm": 0.10077831894159317,
      "learning_rate": 8.151617615278907e-06,
      "loss": 0.003,
      "step": 65020
    },
    {
      "epoch": 1.0640595598461915,
      "grad_norm": 0.1052640825510025,
      "learning_rate": 8.15087853607674e-06,
      "loss": 0.0026,
      "step": 65030
    },
    {
      "epoch": 1.0642231857972675,
      "grad_norm": 0.05653892830014229,
      "learning_rate": 8.150139342662801e-06,
      "loss": 0.0038,
      "step": 65040
    },
    {
      "epoch": 1.0643868117483433,
      "grad_norm": 0.11214754730463028,
      "learning_rate": 8.149400035063883e-06,
      "loss": 0.0021,
      "step": 65050
    },
    {
      "epoch": 1.064550437699419,
      "grad_norm": 0.11909761279821396,
      "learning_rate": 8.148660613306783e-06,
      "loss": 0.0022,
      "step": 65060
    },
    {
      "epoch": 1.064714063650495,
      "grad_norm": 0.08854610472917557,
      "learning_rate": 8.147921077418301e-06,
      "loss": 0.002,
      "step": 65070
    },
    {
      "epoch": 1.0648776896015708,
      "grad_norm": 0.18064527213573456,
      "learning_rate": 8.147181427425248e-06,
      "loss": 0.0027,
      "step": 65080
    },
    {
      "epoch": 1.0650413155526466,
      "grad_norm": 0.029910102486610413,
      "learning_rate": 8.146441663354431e-06,
      "loss": 0.0015,
      "step": 65090
    },
    {
      "epoch": 1.0652049415037226,
      "grad_norm": 0.26644033193588257,
      "learning_rate": 8.145701785232666e-06,
      "loss": 0.0031,
      "step": 65100
    },
    {
      "epoch": 1.0653685674547984,
      "grad_norm": 0.3423719108104706,
      "learning_rate": 8.144961793086775e-06,
      "loss": 0.0016,
      "step": 65110
    },
    {
      "epoch": 1.0655321934058741,
      "grad_norm": 0.13085444271564484,
      "learning_rate": 8.144221686943574e-06,
      "loss": 0.0039,
      "step": 65120
    },
    {
      "epoch": 1.0656958193569501,
      "grad_norm": 0.11731607466936111,
      "learning_rate": 8.143481466829895e-06,
      "loss": 0.0016,
      "step": 65130
    },
    {
      "epoch": 1.065859445308026,
      "grad_norm": 0.11845797300338745,
      "learning_rate": 8.142741132772567e-06,
      "loss": 0.0025,
      "step": 65140
    },
    {
      "epoch": 1.0660230712591017,
      "grad_norm": 0.13626961410045624,
      "learning_rate": 8.142000684798427e-06,
      "loss": 0.0027,
      "step": 65150
    },
    {
      "epoch": 1.0661866972101774,
      "grad_norm": 0.08413999527692795,
      "learning_rate": 8.141260122934313e-06,
      "loss": 0.0019,
      "step": 65160
    },
    {
      "epoch": 1.0663503231612534,
      "grad_norm": 0.12726691365242004,
      "learning_rate": 8.140519447207072e-06,
      "loss": 0.0019,
      "step": 65170
    },
    {
      "epoch": 1.0665139491123292,
      "grad_norm": 0.1418721228837967,
      "learning_rate": 8.139778657643547e-06,
      "loss": 0.0036,
      "step": 65180
    },
    {
      "epoch": 1.066677575063405,
      "grad_norm": 0.16693459451198578,
      "learning_rate": 8.13903775427059e-06,
      "loss": 0.0028,
      "step": 65190
    },
    {
      "epoch": 1.066841201014481,
      "grad_norm": 0.18969066441059113,
      "learning_rate": 8.138296737115063e-06,
      "loss": 0.0031,
      "step": 65200
    },
    {
      "epoch": 1.0670048269655568,
      "grad_norm": 0.1474536806344986,
      "learning_rate": 8.13755560620382e-06,
      "loss": 0.0019,
      "step": 65210
    },
    {
      "epoch": 1.0671684529166325,
      "grad_norm": 0.01634121686220169,
      "learning_rate": 8.136814361563728e-06,
      "loss": 0.0027,
      "step": 65220
    },
    {
      "epoch": 1.0673320788677083,
      "grad_norm": 0.05115247890353203,
      "learning_rate": 8.136073003221654e-06,
      "loss": 0.0023,
      "step": 65230
    },
    {
      "epoch": 1.0674957048187843,
      "grad_norm": 0.4415786564350128,
      "learning_rate": 8.13533153120447e-06,
      "loss": 0.0018,
      "step": 65240
    },
    {
      "epoch": 1.06765933076986,
      "grad_norm": 0.06870345026254654,
      "learning_rate": 8.134589945539055e-06,
      "loss": 0.0037,
      "step": 65250
    },
    {
      "epoch": 1.0678229567209359,
      "grad_norm": 0.0712616816163063,
      "learning_rate": 8.133848246252288e-06,
      "loss": 0.0026,
      "step": 65260
    },
    {
      "epoch": 1.0679865826720119,
      "grad_norm": 0.11226426810026169,
      "learning_rate": 8.133106433371055e-06,
      "loss": 0.0024,
      "step": 65270
    },
    {
      "epoch": 1.0681502086230876,
      "grad_norm": 0.08850929141044617,
      "learning_rate": 8.132364506922241e-06,
      "loss": 0.0029,
      "step": 65280
    },
    {
      "epoch": 1.0683138345741634,
      "grad_norm": 0.226271390914917,
      "learning_rate": 8.131622466932744e-06,
      "loss": 0.0031,
      "step": 65290
    },
    {
      "epoch": 1.0684774605252394,
      "grad_norm": 0.10922946035861969,
      "learning_rate": 8.130880313429459e-06,
      "loss": 0.0033,
      "step": 65300
    },
    {
      "epoch": 1.0686410864763152,
      "grad_norm": 0.14066576957702637,
      "learning_rate": 8.130138046439287e-06,
      "loss": 0.0031,
      "step": 65310
    },
    {
      "epoch": 1.068804712427391,
      "grad_norm": 0.14621934294700623,
      "learning_rate": 8.129395665989135e-06,
      "loss": 0.0022,
      "step": 65320
    },
    {
      "epoch": 1.068968338378467,
      "grad_norm": 0.4749991297721863,
      "learning_rate": 8.128653172105911e-06,
      "loss": 0.0054,
      "step": 65330
    },
    {
      "epoch": 1.0691319643295427,
      "grad_norm": 0.08464115858078003,
      "learning_rate": 8.127910564816527e-06,
      "loss": 0.0037,
      "step": 65340
    },
    {
      "epoch": 1.0692955902806185,
      "grad_norm": 0.20063425600528717,
      "learning_rate": 8.127167844147908e-06,
      "loss": 0.0021,
      "step": 65350
    },
    {
      "epoch": 1.0694592162316943,
      "grad_norm": 0.059448108077049255,
      "learning_rate": 8.126425010126966e-06,
      "loss": 0.0021,
      "step": 65360
    },
    {
      "epoch": 1.0696228421827703,
      "grad_norm": 0.17158937454223633,
      "learning_rate": 8.125682062780634e-06,
      "loss": 0.0022,
      "step": 65370
    },
    {
      "epoch": 1.069786468133846,
      "grad_norm": 0.1435229331254959,
      "learning_rate": 8.124939002135837e-06,
      "loss": 0.0022,
      "step": 65380
    },
    {
      "epoch": 1.0699500940849218,
      "grad_norm": 0.29025959968566895,
      "learning_rate": 8.124195828219513e-06,
      "loss": 0.0027,
      "step": 65390
    },
    {
      "epoch": 1.0701137200359978,
      "grad_norm": 0.09354247897863388,
      "learning_rate": 8.1234525410586e-06,
      "loss": 0.0029,
      "step": 65400
    },
    {
      "epoch": 1.0702773459870736,
      "grad_norm": 0.27978625893592834,
      "learning_rate": 8.122709140680039e-06,
      "loss": 0.0029,
      "step": 65410
    },
    {
      "epoch": 1.0704409719381494,
      "grad_norm": 0.0676610991358757,
      "learning_rate": 8.121965627110775e-06,
      "loss": 0.0056,
      "step": 65420
    },
    {
      "epoch": 1.0706045978892251,
      "grad_norm": 0.23127678036689758,
      "learning_rate": 8.121222000377763e-06,
      "loss": 0.0033,
      "step": 65430
    },
    {
      "epoch": 1.0707682238403011,
      "grad_norm": 0.22223244607448578,
      "learning_rate": 8.120478260507954e-06,
      "loss": 0.003,
      "step": 65440
    },
    {
      "epoch": 1.070931849791377,
      "grad_norm": 0.1291331946849823,
      "learning_rate": 8.119734407528308e-06,
      "loss": 0.0021,
      "step": 65450
    },
    {
      "epoch": 1.0710954757424527,
      "grad_norm": 0.18976299464702606,
      "learning_rate": 8.118990441465789e-06,
      "loss": 0.0024,
      "step": 65460
    },
    {
      "epoch": 1.0712591016935287,
      "grad_norm": 0.09180238097906113,
      "learning_rate": 8.118246362347361e-06,
      "loss": 0.0043,
      "step": 65470
    },
    {
      "epoch": 1.0714227276446044,
      "grad_norm": 0.11240053176879883,
      "learning_rate": 8.117502170199997e-06,
      "loss": 0.002,
      "step": 65480
    },
    {
      "epoch": 1.0715863535956802,
      "grad_norm": 0.061869967728853226,
      "learning_rate": 8.116757865050674e-06,
      "loss": 0.0021,
      "step": 65490
    },
    {
      "epoch": 1.0717499795467562,
      "grad_norm": 0.19311775267124176,
      "learning_rate": 8.116013446926367e-06,
      "loss": 0.0027,
      "step": 65500
    },
    {
      "epoch": 1.071913605497832,
      "grad_norm": 0.06645571440458298,
      "learning_rate": 8.115268915854062e-06,
      "loss": 0.0025,
      "step": 65510
    },
    {
      "epoch": 1.0720772314489078,
      "grad_norm": 0.14953108131885529,
      "learning_rate": 8.114524271860746e-06,
      "loss": 0.0024,
      "step": 65520
    },
    {
      "epoch": 1.0722408573999835,
      "grad_norm": 0.06905048340559006,
      "learning_rate": 8.113779514973412e-06,
      "loss": 0.0016,
      "step": 65530
    },
    {
      "epoch": 1.0724044833510595,
      "grad_norm": 0.12818819284439087,
      "learning_rate": 8.113034645219053e-06,
      "loss": 0.0029,
      "step": 65540
    },
    {
      "epoch": 1.0725681093021353,
      "grad_norm": 0.5096604824066162,
      "learning_rate": 8.112289662624673e-06,
      "loss": 0.004,
      "step": 65550
    },
    {
      "epoch": 1.072731735253211,
      "grad_norm": 0.19868259131908417,
      "learning_rate": 8.11154456721727e-06,
      "loss": 0.0031,
      "step": 65560
    },
    {
      "epoch": 1.072895361204287,
      "grad_norm": 0.007214093115180731,
      "learning_rate": 8.110799359023855e-06,
      "loss": 0.0028,
      "step": 65570
    },
    {
      "epoch": 1.0730589871553629,
      "grad_norm": 0.05731329321861267,
      "learning_rate": 8.11005403807144e-06,
      "loss": 0.0015,
      "step": 65580
    },
    {
      "epoch": 1.0732226131064386,
      "grad_norm": 0.14137881994247437,
      "learning_rate": 8.109308604387043e-06,
      "loss": 0.0025,
      "step": 65590
    },
    {
      "epoch": 1.0733862390575146,
      "grad_norm": 0.29019469022750854,
      "learning_rate": 8.10856305799768e-06,
      "loss": 0.0017,
      "step": 65600
    },
    {
      "epoch": 1.0735498650085904,
      "grad_norm": 0.11052026599645615,
      "learning_rate": 8.107817398930377e-06,
      "loss": 0.0018,
      "step": 65610
    },
    {
      "epoch": 1.0737134909596662,
      "grad_norm": 0.24654839932918549,
      "learning_rate": 8.107071627212166e-06,
      "loss": 0.0036,
      "step": 65620
    },
    {
      "epoch": 1.073877116910742,
      "grad_norm": 0.08177804201841354,
      "learning_rate": 8.106325742870072e-06,
      "loss": 0.0035,
      "step": 65630
    },
    {
      "epoch": 1.074040742861818,
      "grad_norm": 0.1591581553220749,
      "learning_rate": 8.105579745931139e-06,
      "loss": 0.004,
      "step": 65640
    },
    {
      "epoch": 1.0742043688128937,
      "grad_norm": 0.17649294435977936,
      "learning_rate": 8.104833636422402e-06,
      "loss": 0.0028,
      "step": 65650
    },
    {
      "epoch": 1.0743679947639695,
      "grad_norm": 0.05311176925897598,
      "learning_rate": 8.104087414370908e-06,
      "loss": 0.0022,
      "step": 65660
    },
    {
      "epoch": 1.0745316207150455,
      "grad_norm": 0.22333522140979767,
      "learning_rate": 8.103341079803707e-06,
      "loss": 0.0028,
      "step": 65670
    },
    {
      "epoch": 1.0746952466661213,
      "grad_norm": 0.13097503781318665,
      "learning_rate": 8.10259463274785e-06,
      "loss": 0.0027,
      "step": 65680
    },
    {
      "epoch": 1.074858872617197,
      "grad_norm": 0.04455450549721718,
      "learning_rate": 8.101848073230393e-06,
      "loss": 0.0037,
      "step": 65690
    },
    {
      "epoch": 1.075022498568273,
      "grad_norm": 0.2615099251270294,
      "learning_rate": 8.101101401278402e-06,
      "loss": 0.0027,
      "step": 65700
    },
    {
      "epoch": 1.0751861245193488,
      "grad_norm": 0.15217532217502594,
      "learning_rate": 8.100354616918936e-06,
      "loss": 0.0016,
      "step": 65710
    },
    {
      "epoch": 1.0753497504704246,
      "grad_norm": 0.07289324700832367,
      "learning_rate": 8.099607720179067e-06,
      "loss": 0.002,
      "step": 65720
    },
    {
      "epoch": 1.0755133764215004,
      "grad_norm": 0.2076735496520996,
      "learning_rate": 8.098860711085867e-06,
      "loss": 0.0023,
      "step": 65730
    },
    {
      "epoch": 1.0756770023725764,
      "grad_norm": 0.08935001492500305,
      "learning_rate": 8.098113589666416e-06,
      "loss": 0.0018,
      "step": 65740
    },
    {
      "epoch": 1.0758406283236521,
      "grad_norm": 0.2966357469558716,
      "learning_rate": 8.097366355947793e-06,
      "loss": 0.0019,
      "step": 65750
    },
    {
      "epoch": 1.076004254274728,
      "grad_norm": 0.29122012853622437,
      "learning_rate": 8.096619009957082e-06,
      "loss": 0.0018,
      "step": 65760
    },
    {
      "epoch": 1.076167880225804,
      "grad_norm": 0.11934593319892883,
      "learning_rate": 8.095871551721377e-06,
      "loss": 0.0018,
      "step": 65770
    },
    {
      "epoch": 1.0763315061768797,
      "grad_norm": 0.4460866451263428,
      "learning_rate": 8.095123981267767e-06,
      "loss": 0.003,
      "step": 65780
    },
    {
      "epoch": 1.0764951321279554,
      "grad_norm": 0.1869867444038391,
      "learning_rate": 8.094376298623354e-06,
      "loss": 0.0024,
      "step": 65790
    },
    {
      "epoch": 1.0766587580790312,
      "grad_norm": 0.07907191663980484,
      "learning_rate": 8.093628503815236e-06,
      "loss": 0.0024,
      "step": 65800
    },
    {
      "epoch": 1.0768223840301072,
      "grad_norm": 0.08090656250715256,
      "learning_rate": 8.092880596870518e-06,
      "loss": 0.0017,
      "step": 65810
    },
    {
      "epoch": 1.076986009981183,
      "grad_norm": 0.9056613445281982,
      "learning_rate": 8.092132577816314e-06,
      "loss": 0.0044,
      "step": 65820
    },
    {
      "epoch": 1.0771496359322588,
      "grad_norm": 0.3343521058559418,
      "learning_rate": 8.091384446679736e-06,
      "loss": 0.0036,
      "step": 65830
    },
    {
      "epoch": 1.0773132618833348,
      "grad_norm": 0.02946481667459011,
      "learning_rate": 8.090636203487902e-06,
      "loss": 0.0022,
      "step": 65840
    },
    {
      "epoch": 1.0774768878344105,
      "grad_norm": 0.2516515254974365,
      "learning_rate": 8.089887848267933e-06,
      "loss": 0.0026,
      "step": 65850
    },
    {
      "epoch": 1.0776405137854863,
      "grad_norm": 0.007942191325128078,
      "learning_rate": 8.089139381046955e-06,
      "loss": 0.0012,
      "step": 65860
    },
    {
      "epoch": 1.0778041397365623,
      "grad_norm": 0.05471820384263992,
      "learning_rate": 8.088390801852101e-06,
      "loss": 0.0011,
      "step": 65870
    },
    {
      "epoch": 1.077967765687638,
      "grad_norm": 0.34158727526664734,
      "learning_rate": 8.087642110710501e-06,
      "loss": 0.0034,
      "step": 65880
    },
    {
      "epoch": 1.0781313916387139,
      "grad_norm": 0.14389953017234802,
      "learning_rate": 8.086893307649296e-06,
      "loss": 0.002,
      "step": 65890
    },
    {
      "epoch": 1.0782950175897899,
      "grad_norm": 0.1541469395160675,
      "learning_rate": 8.086144392695629e-06,
      "loss": 0.0034,
      "step": 65900
    },
    {
      "epoch": 1.0784586435408656,
      "grad_norm": 0.15493831038475037,
      "learning_rate": 8.085395365876644e-06,
      "loss": 0.0021,
      "step": 65910
    },
    {
      "epoch": 1.0786222694919414,
      "grad_norm": 0.09754495322704315,
      "learning_rate": 8.084646227219493e-06,
      "loss": 0.0029,
      "step": 65920
    },
    {
      "epoch": 1.0787858954430172,
      "grad_norm": 0.10205633193254471,
      "learning_rate": 8.083896976751332e-06,
      "loss": 0.0028,
      "step": 65930
    },
    {
      "epoch": 1.0789495213940932,
      "grad_norm": 0.12074751406908035,
      "learning_rate": 8.083147614499314e-06,
      "loss": 0.0036,
      "step": 65940
    },
    {
      "epoch": 1.079113147345169,
      "grad_norm": 0.07306011766195297,
      "learning_rate": 8.082398140490608e-06,
      "loss": 0.0025,
      "step": 65950
    },
    {
      "epoch": 1.0792767732962447,
      "grad_norm": 0.3032417297363281,
      "learning_rate": 8.081648554752374e-06,
      "loss": 0.005,
      "step": 65960
    },
    {
      "epoch": 1.0794403992473207,
      "grad_norm": 0.10294811427593231,
      "learning_rate": 8.080898857311789e-06,
      "loss": 0.0019,
      "step": 65970
    },
    {
      "epoch": 1.0796040251983965,
      "grad_norm": 0.0685034766793251,
      "learning_rate": 8.080149048196027e-06,
      "loss": 0.0014,
      "step": 65980
    },
    {
      "epoch": 1.0797676511494723,
      "grad_norm": 0.14835114777088165,
      "learning_rate": 8.079399127432263e-06,
      "loss": 0.0031,
      "step": 65990
    },
    {
      "epoch": 1.079931277100548,
      "grad_norm": 0.005194514058530331,
      "learning_rate": 8.078649095047681e-06,
      "loss": 0.0018,
      "step": 66000
    },
    {
      "epoch": 1.080094903051624,
      "grad_norm": 0.11142878979444504,
      "learning_rate": 8.07789895106947e-06,
      "loss": 0.0017,
      "step": 66010
    },
    {
      "epoch": 1.0802585290026998,
      "grad_norm": 0.18695776164531708,
      "learning_rate": 8.077148695524819e-06,
      "loss": 0.0029,
      "step": 66020
    },
    {
      "epoch": 1.0804221549537756,
      "grad_norm": 0.05895831808447838,
      "learning_rate": 8.076398328440922e-06,
      "loss": 0.0019,
      "step": 66030
    },
    {
      "epoch": 1.0805857809048516,
      "grad_norm": 0.2404588907957077,
      "learning_rate": 8.075647849844981e-06,
      "loss": 0.0025,
      "step": 66040
    },
    {
      "epoch": 1.0807494068559274,
      "grad_norm": 0.07785984873771667,
      "learning_rate": 8.074897259764198e-06,
      "loss": 0.002,
      "step": 66050
    },
    {
      "epoch": 1.0809130328070031,
      "grad_norm": 0.10359875857830048,
      "learning_rate": 8.07414655822578e-06,
      "loss": 0.0016,
      "step": 66060
    },
    {
      "epoch": 1.0810766587580791,
      "grad_norm": 0.04366911202669144,
      "learning_rate": 8.073395745256937e-06,
      "loss": 0.0024,
      "step": 66070
    },
    {
      "epoch": 1.081240284709155,
      "grad_norm": 0.12163852900266647,
      "learning_rate": 8.072644820884884e-06,
      "loss": 0.0022,
      "step": 66080
    },
    {
      "epoch": 1.0814039106602307,
      "grad_norm": 0.1838352084159851,
      "learning_rate": 8.071893785136842e-06,
      "loss": 0.0035,
      "step": 66090
    },
    {
      "epoch": 1.0815675366113067,
      "grad_norm": 0.088449627161026,
      "learning_rate": 8.071142638040035e-06,
      "loss": 0.0021,
      "step": 66100
    },
    {
      "epoch": 1.0817311625623824,
      "grad_norm": 0.045412205159664154,
      "learning_rate": 8.070391379621687e-06,
      "loss": 0.0021,
      "step": 66110
    },
    {
      "epoch": 1.0818947885134582,
      "grad_norm": 0.3884061276912689,
      "learning_rate": 8.06964000990903e-06,
      "loss": 0.0055,
      "step": 66120
    },
    {
      "epoch": 1.082058414464534,
      "grad_norm": 0.08041780441999435,
      "learning_rate": 8.068888528929303e-06,
      "loss": 0.0017,
      "step": 66130
    },
    {
      "epoch": 1.08222204041561,
      "grad_norm": 0.2716643214225769,
      "learning_rate": 8.06813693670974e-06,
      "loss": 0.0023,
      "step": 66140
    },
    {
      "epoch": 1.0823856663666858,
      "grad_norm": 0.17752836644649506,
      "learning_rate": 8.067385233277589e-06,
      "loss": 0.0028,
      "step": 66150
    },
    {
      "epoch": 1.0825492923177615,
      "grad_norm": 0.3403341472148895,
      "learning_rate": 8.066633418660095e-06,
      "loss": 0.0038,
      "step": 66160
    },
    {
      "epoch": 1.0827129182688375,
      "grad_norm": 0.06830795854330063,
      "learning_rate": 8.06588149288451e-06,
      "loss": 0.0016,
      "step": 66170
    },
    {
      "epoch": 1.0828765442199133,
      "grad_norm": 0.15338610112667084,
      "learning_rate": 8.065129455978087e-06,
      "loss": 0.0042,
      "step": 66180
    },
    {
      "epoch": 1.083040170170989,
      "grad_norm": 0.06503737717866898,
      "learning_rate": 8.064377307968092e-06,
      "loss": 0.0025,
      "step": 66190
    },
    {
      "epoch": 1.0832037961220649,
      "grad_norm": 0.3710920512676239,
      "learning_rate": 8.063625048881782e-06,
      "loss": 0.0023,
      "step": 66200
    },
    {
      "epoch": 1.0833674220731409,
      "grad_norm": 0.1289042979478836,
      "learning_rate": 8.062872678746429e-06,
      "loss": 0.0025,
      "step": 66210
    },
    {
      "epoch": 1.0835310480242166,
      "grad_norm": 0.06448682397603989,
      "learning_rate": 8.062120197589303e-06,
      "loss": 0.002,
      "step": 66220
    },
    {
      "epoch": 1.0836946739752924,
      "grad_norm": 0.3412668704986572,
      "learning_rate": 8.061367605437677e-06,
      "loss": 0.0024,
      "step": 66230
    },
    {
      "epoch": 1.0838582999263684,
      "grad_norm": 0.19381700456142426,
      "learning_rate": 8.060614902318835e-06,
      "loss": 0.0015,
      "step": 66240
    },
    {
      "epoch": 1.0840219258774442,
      "grad_norm": 0.0045209359377622604,
      "learning_rate": 8.05986208826006e-06,
      "loss": 0.0027,
      "step": 66250
    },
    {
      "epoch": 1.08418555182852,
      "grad_norm": 0.12231379002332687,
      "learning_rate": 8.059109163288637e-06,
      "loss": 0.0024,
      "step": 66260
    },
    {
      "epoch": 1.084349177779596,
      "grad_norm": 0.26199087500572205,
      "learning_rate": 8.058356127431858e-06,
      "loss": 0.0017,
      "step": 66270
    },
    {
      "epoch": 1.0845128037306717,
      "grad_norm": 0.07341767102479935,
      "learning_rate": 8.057602980717021e-06,
      "loss": 0.002,
      "step": 66280
    },
    {
      "epoch": 1.0846764296817475,
      "grad_norm": 0.01413731463253498,
      "learning_rate": 8.056849723171426e-06,
      "loss": 0.0029,
      "step": 66290
    },
    {
      "epoch": 1.0848400556328233,
      "grad_norm": 0.2918248772621155,
      "learning_rate": 8.056096354822374e-06,
      "loss": 0.0024,
      "step": 66300
    },
    {
      "epoch": 1.0850036815838993,
      "grad_norm": 0.10700853168964386,
      "learning_rate": 8.055342875697175e-06,
      "loss": 0.0022,
      "step": 66310
    },
    {
      "epoch": 1.085167307534975,
      "grad_norm": 0.5969021320343018,
      "learning_rate": 8.054589285823137e-06,
      "loss": 0.005,
      "step": 66320
    },
    {
      "epoch": 1.0853309334860508,
      "grad_norm": 0.1458989828824997,
      "learning_rate": 8.053835585227582e-06,
      "loss": 0.0019,
      "step": 66330
    },
    {
      "epoch": 1.0854945594371268,
      "grad_norm": 0.07172498106956482,
      "learning_rate": 8.053081773937828e-06,
      "loss": 0.0026,
      "step": 66340
    },
    {
      "epoch": 1.0856581853882026,
      "grad_norm": 0.14115053415298462,
      "learning_rate": 8.052327851981195e-06,
      "loss": 0.0022,
      "step": 66350
    },
    {
      "epoch": 1.0858218113392784,
      "grad_norm": 0.18594765663146973,
      "learning_rate": 8.051573819385015e-06,
      "loss": 0.002,
      "step": 66360
    },
    {
      "epoch": 1.0859854372903543,
      "grad_norm": 0.07159534841775894,
      "learning_rate": 8.050819676176616e-06,
      "loss": 0.0039,
      "step": 66370
    },
    {
      "epoch": 1.0861490632414301,
      "grad_norm": 0.3044295012950897,
      "learning_rate": 8.050065422383337e-06,
      "loss": 0.0013,
      "step": 66380
    },
    {
      "epoch": 1.086312689192506,
      "grad_norm": 0.12141037732362747,
      "learning_rate": 8.049311058032519e-06,
      "loss": 0.0024,
      "step": 66390
    },
    {
      "epoch": 1.0864763151435817,
      "grad_norm": 0.49803096055984497,
      "learning_rate": 8.048556583151504e-06,
      "loss": 0.0021,
      "step": 66400
    },
    {
      "epoch": 1.0866399410946577,
      "grad_norm": 0.1595144420862198,
      "learning_rate": 8.047801997767638e-06,
      "loss": 0.004,
      "step": 66410
    },
    {
      "epoch": 1.0868035670457334,
      "grad_norm": 0.11515387892723083,
      "learning_rate": 8.047047301908275e-06,
      "loss": 0.0037,
      "step": 66420
    },
    {
      "epoch": 1.0869671929968092,
      "grad_norm": 0.10269689559936523,
      "learning_rate": 8.04629249560077e-06,
      "loss": 0.0022,
      "step": 66430
    },
    {
      "epoch": 1.0871308189478852,
      "grad_norm": 0.17100584506988525,
      "learning_rate": 8.045537578872486e-06,
      "loss": 0.0024,
      "step": 66440
    },
    {
      "epoch": 1.087294444898961,
      "grad_norm": 0.2069419026374817,
      "learning_rate": 8.044782551750784e-06,
      "loss": 0.0023,
      "step": 66450
    },
    {
      "epoch": 1.0874580708500368,
      "grad_norm": 0.0044179982505738735,
      "learning_rate": 8.044027414263032e-06,
      "loss": 0.0034,
      "step": 66460
    },
    {
      "epoch": 1.0876216968011128,
      "grad_norm": 0.14417746663093567,
      "learning_rate": 8.043272166436602e-06,
      "loss": 0.003,
      "step": 66470
    },
    {
      "epoch": 1.0877853227521885,
      "grad_norm": 0.24433189630508423,
      "learning_rate": 8.042516808298871e-06,
      "loss": 0.0028,
      "step": 66480
    },
    {
      "epoch": 1.0879489487032643,
      "grad_norm": 0.20231904089450836,
      "learning_rate": 8.041761339877216e-06,
      "loss": 0.0026,
      "step": 66490
    },
    {
      "epoch": 1.08811257465434,
      "grad_norm": 0.23650319874286652,
      "learning_rate": 8.041005761199026e-06,
      "loss": 0.0024,
      "step": 66500
    },
    {
      "epoch": 1.088276200605416,
      "grad_norm": 0.06208028271794319,
      "learning_rate": 8.040250072291683e-06,
      "loss": 0.0028,
      "step": 66510
    },
    {
      "epoch": 1.0884398265564919,
      "grad_norm": 0.059307318180799484,
      "learning_rate": 8.039494273182584e-06,
      "loss": 0.0018,
      "step": 66520
    },
    {
      "epoch": 1.0886034525075676,
      "grad_norm": 0.060293685644865036,
      "learning_rate": 8.038738363899124e-06,
      "loss": 0.0021,
      "step": 66530
    },
    {
      "epoch": 1.0887670784586436,
      "grad_norm": 0.06537779420614243,
      "learning_rate": 8.037982344468698e-06,
      "loss": 0.0022,
      "step": 66540
    },
    {
      "epoch": 1.0889307044097194,
      "grad_norm": 0.157841756939888,
      "learning_rate": 8.037226214918716e-06,
      "loss": 0.0018,
      "step": 66550
    },
    {
      "epoch": 1.0890943303607952,
      "grad_norm": 0.2087858021259308,
      "learning_rate": 8.036469975276583e-06,
      "loss": 0.0023,
      "step": 66560
    },
    {
      "epoch": 1.089257956311871,
      "grad_norm": 0.04547343775629997,
      "learning_rate": 8.035713625569712e-06,
      "loss": 0.0025,
      "step": 66570
    },
    {
      "epoch": 1.089421582262947,
      "grad_norm": 0.19873729348182678,
      "learning_rate": 8.034957165825518e-06,
      "loss": 0.0026,
      "step": 66580
    },
    {
      "epoch": 1.0895852082140227,
      "grad_norm": 0.20767268538475037,
      "learning_rate": 8.034200596071422e-06,
      "loss": 0.0017,
      "step": 66590
    },
    {
      "epoch": 1.0897488341650985,
      "grad_norm": 0.2506837546825409,
      "learning_rate": 8.033443916334846e-06,
      "loss": 0.0015,
      "step": 66600
    },
    {
      "epoch": 1.0899124601161745,
      "grad_norm": 0.09592298418283463,
      "learning_rate": 8.032687126643218e-06,
      "loss": 0.0031,
      "step": 66610
    },
    {
      "epoch": 1.0900760860672503,
      "grad_norm": 0.2865099310874939,
      "learning_rate": 8.031930227023972e-06,
      "loss": 0.0029,
      "step": 66620
    },
    {
      "epoch": 1.090239712018326,
      "grad_norm": 0.044565241783857346,
      "learning_rate": 8.03117321750454e-06,
      "loss": 0.0027,
      "step": 66630
    },
    {
      "epoch": 1.090403337969402,
      "grad_norm": 0.266215443611145,
      "learning_rate": 8.030416098112366e-06,
      "loss": 0.0019,
      "step": 66640
    },
    {
      "epoch": 1.0905669639204778,
      "grad_norm": 0.04512857273221016,
      "learning_rate": 8.029658868874892e-06,
      "loss": 0.0012,
      "step": 66650
    },
    {
      "epoch": 1.0907305898715536,
      "grad_norm": 0.27870652079582214,
      "learning_rate": 8.028901529819564e-06,
      "loss": 0.0023,
      "step": 66660
    },
    {
      "epoch": 1.0908942158226296,
      "grad_norm": 0.13263162970542908,
      "learning_rate": 8.028144080973836e-06,
      "loss": 0.0025,
      "step": 66670
    },
    {
      "epoch": 1.0910578417737054,
      "grad_norm": 0.01751447282731533,
      "learning_rate": 8.027386522365163e-06,
      "loss": 0.0025,
      "step": 66680
    },
    {
      "epoch": 1.0912214677247811,
      "grad_norm": 0.048395249992609024,
      "learning_rate": 8.026628854021005e-06,
      "loss": 0.0014,
      "step": 66690
    },
    {
      "epoch": 1.091385093675857,
      "grad_norm": 0.0751519426703453,
      "learning_rate": 8.025871075968828e-06,
      "loss": 0.0028,
      "step": 66700
    },
    {
      "epoch": 1.091548719626933,
      "grad_norm": 0.10317564010620117,
      "learning_rate": 8.025113188236094e-06,
      "loss": 0.0018,
      "step": 66710
    },
    {
      "epoch": 1.0917123455780087,
      "grad_norm": 0.17076021432876587,
      "learning_rate": 8.024355190850276e-06,
      "loss": 0.0025,
      "step": 66720
    },
    {
      "epoch": 1.0918759715290844,
      "grad_norm": 0.04208666458725929,
      "learning_rate": 8.023597083838854e-06,
      "loss": 0.002,
      "step": 66730
    },
    {
      "epoch": 1.0920395974801604,
      "grad_norm": 0.3747827410697937,
      "learning_rate": 8.022838867229304e-06,
      "loss": 0.007,
      "step": 66740
    },
    {
      "epoch": 1.0922032234312362,
      "grad_norm": 0.028362950310111046,
      "learning_rate": 8.022080541049111e-06,
      "loss": 0.0022,
      "step": 66750
    },
    {
      "epoch": 1.092366849382312,
      "grad_norm": 0.20069214701652527,
      "learning_rate": 8.02132210532576e-06,
      "loss": 0.0028,
      "step": 66760
    },
    {
      "epoch": 1.0925304753333878,
      "grad_norm": 0.10993585735559464,
      "learning_rate": 8.020563560086745e-06,
      "loss": 0.0048,
      "step": 66770
    },
    {
      "epoch": 1.0926941012844638,
      "grad_norm": 0.23520207405090332,
      "learning_rate": 8.01980490535956e-06,
      "loss": 0.002,
      "step": 66780
    },
    {
      "epoch": 1.0928577272355395,
      "grad_norm": 0.05501289665699005,
      "learning_rate": 8.019046141171706e-06,
      "loss": 0.0043,
      "step": 66790
    },
    {
      "epoch": 1.0930213531866153,
      "grad_norm": 0.30154648423194885,
      "learning_rate": 8.018287267550686e-06,
      "loss": 0.0038,
      "step": 66800
    },
    {
      "epoch": 1.0931849791376913,
      "grad_norm": 0.12688271701335907,
      "learning_rate": 8.017528284524007e-06,
      "loss": 0.0018,
      "step": 66810
    },
    {
      "epoch": 1.093348605088767,
      "grad_norm": 0.28403204679489136,
      "learning_rate": 8.016769192119178e-06,
      "loss": 0.0016,
      "step": 66820
    },
    {
      "epoch": 1.0935122310398429,
      "grad_norm": 0.10046349465847015,
      "learning_rate": 8.016009990363718e-06,
      "loss": 0.0019,
      "step": 66830
    },
    {
      "epoch": 1.0936758569909188,
      "grad_norm": 0.060041528195142746,
      "learning_rate": 8.015250679285144e-06,
      "loss": 0.0026,
      "step": 66840
    },
    {
      "epoch": 1.0938394829419946,
      "grad_norm": 0.11048177629709244,
      "learning_rate": 8.014491258910982e-06,
      "loss": 0.0013,
      "step": 66850
    },
    {
      "epoch": 1.0940031088930704,
      "grad_norm": 0.2221585363149643,
      "learning_rate": 8.013731729268754e-06,
      "loss": 0.0023,
      "step": 66860
    },
    {
      "epoch": 1.0941667348441464,
      "grad_norm": 0.13272634148597717,
      "learning_rate": 8.012972090385996e-06,
      "loss": 0.0029,
      "step": 66870
    },
    {
      "epoch": 1.0943303607952222,
      "grad_norm": 0.12997838854789734,
      "learning_rate": 8.01221234229024e-06,
      "loss": 0.0037,
      "step": 66880
    },
    {
      "epoch": 1.094493986746298,
      "grad_norm": 0.2496015429496765,
      "learning_rate": 8.011452485009026e-06,
      "loss": 0.0027,
      "step": 66890
    },
    {
      "epoch": 1.0946576126973737,
      "grad_norm": 0.07141968607902527,
      "learning_rate": 8.010692518569899e-06,
      "loss": 0.0029,
      "step": 66900
    },
    {
      "epoch": 1.0948212386484497,
      "grad_norm": 0.2587110102176666,
      "learning_rate": 8.009932443000403e-06,
      "loss": 0.0023,
      "step": 66910
    },
    {
      "epoch": 1.0949848645995255,
      "grad_norm": 0.5316643118858337,
      "learning_rate": 8.009172258328091e-06,
      "loss": 0.002,
      "step": 66920
    },
    {
      "epoch": 1.0951484905506013,
      "grad_norm": 0.26069799065589905,
      "learning_rate": 8.008411964580517e-06,
      "loss": 0.0018,
      "step": 66930
    },
    {
      "epoch": 1.0953121165016773,
      "grad_norm": 0.19477787613868713,
      "learning_rate": 8.007651561785236e-06,
      "loss": 0.0024,
      "step": 66940
    },
    {
      "epoch": 1.095475742452753,
      "grad_norm": 0.018018892034888268,
      "learning_rate": 8.006891049969821e-06,
      "loss": 0.0014,
      "step": 66950
    },
    {
      "epoch": 1.0956393684038288,
      "grad_norm": 0.17283983528614044,
      "learning_rate": 8.006130429161828e-06,
      "loss": 0.0037,
      "step": 66960
    },
    {
      "epoch": 1.0958029943549046,
      "grad_norm": 0.06775011122226715,
      "learning_rate": 8.005369699388833e-06,
      "loss": 0.0028,
      "step": 66970
    },
    {
      "epoch": 1.0959666203059806,
      "grad_norm": 0.1613825261592865,
      "learning_rate": 8.004608860678409e-06,
      "loss": 0.0022,
      "step": 66980
    },
    {
      "epoch": 1.0961302462570564,
      "grad_norm": 0.1508966088294983,
      "learning_rate": 8.003847913058135e-06,
      "loss": 0.0026,
      "step": 66990
    },
    {
      "epoch": 1.0962938722081321,
      "grad_norm": 0.13129541277885437,
      "learning_rate": 8.003086856555596e-06,
      "loss": 0.0028,
      "step": 67000
    },
    {
      "epoch": 1.0964574981592081,
      "grad_norm": 0.22265157103538513,
      "learning_rate": 8.002325691198375e-06,
      "loss": 0.0031,
      "step": 67010
    },
    {
      "epoch": 1.096621124110284,
      "grad_norm": 0.1172194853425026,
      "learning_rate": 8.001564417014062e-06,
      "loss": 0.0026,
      "step": 67020
    },
    {
      "epoch": 1.0967847500613597,
      "grad_norm": 0.09978342056274414,
      "learning_rate": 8.000803034030256e-06,
      "loss": 0.0016,
      "step": 67030
    },
    {
      "epoch": 1.0969483760124357,
      "grad_norm": 0.15964211523532867,
      "learning_rate": 8.000041542274549e-06,
      "loss": 0.0028,
      "step": 67040
    },
    {
      "epoch": 1.0971120019635114,
      "grad_norm": 0.24837394058704376,
      "learning_rate": 7.999279941774549e-06,
      "loss": 0.0019,
      "step": 67050
    },
    {
      "epoch": 1.0972756279145872,
      "grad_norm": 0.08646401762962341,
      "learning_rate": 7.99851823255786e-06,
      "loss": 0.002,
      "step": 67060
    },
    {
      "epoch": 1.0974392538656632,
      "grad_norm": 0.06600673496723175,
      "learning_rate": 7.997756414652089e-06,
      "loss": 0.0018,
      "step": 67070
    },
    {
      "epoch": 1.097602879816739,
      "grad_norm": 0.05896184593439102,
      "learning_rate": 7.996994488084855e-06,
      "loss": 0.0044,
      "step": 67080
    },
    {
      "epoch": 1.0977665057678148,
      "grad_norm": 0.058279041200876236,
      "learning_rate": 7.996232452883772e-06,
      "loss": 0.003,
      "step": 67090
    },
    {
      "epoch": 1.0979301317188905,
      "grad_norm": 0.2605247497558594,
      "learning_rate": 7.995470309076466e-06,
      "loss": 0.0016,
      "step": 67100
    },
    {
      "epoch": 1.0980937576699665,
      "grad_norm": 0.18463708460330963,
      "learning_rate": 7.99470805669056e-06,
      "loss": 0.0027,
      "step": 67110
    },
    {
      "epoch": 1.0982573836210423,
      "grad_norm": 0.059139564633369446,
      "learning_rate": 7.993945695753684e-06,
      "loss": 0.0012,
      "step": 67120
    },
    {
      "epoch": 1.098421009572118,
      "grad_norm": 0.156093567609787,
      "learning_rate": 7.993183226293473e-06,
      "loss": 0.0023,
      "step": 67130
    },
    {
      "epoch": 1.098584635523194,
      "grad_norm": 0.16678935289382935,
      "learning_rate": 7.992420648337564e-06,
      "loss": 0.0023,
      "step": 67140
    },
    {
      "epoch": 1.0987482614742698,
      "grad_norm": 0.16258345544338226,
      "learning_rate": 7.991657961913599e-06,
      "loss": 0.0029,
      "step": 67150
    },
    {
      "epoch": 1.0989118874253456,
      "grad_norm": 0.14234188199043274,
      "learning_rate": 7.990895167049222e-06,
      "loss": 0.0013,
      "step": 67160
    },
    {
      "epoch": 1.0990755133764214,
      "grad_norm": 0.7577089667320251,
      "learning_rate": 7.990132263772083e-06,
      "loss": 0.0017,
      "step": 67170
    },
    {
      "epoch": 1.0992391393274974,
      "grad_norm": 0.20397929847240448,
      "learning_rate": 7.989369252109838e-06,
      "loss": 0.0021,
      "step": 67180
    },
    {
      "epoch": 1.0994027652785732,
      "grad_norm": 0.17214415967464447,
      "learning_rate": 7.988606132090141e-06,
      "loss": 0.0022,
      "step": 67190
    },
    {
      "epoch": 1.099566391229649,
      "grad_norm": 0.3056853413581848,
      "learning_rate": 7.987842903740654e-06,
      "loss": 0.0034,
      "step": 67200
    },
    {
      "epoch": 1.099730017180725,
      "grad_norm": 0.32553645968437195,
      "learning_rate": 7.987079567089045e-06,
      "loss": 0.002,
      "step": 67210
    },
    {
      "epoch": 1.0998936431318007,
      "grad_norm": 0.06345076858997345,
      "learning_rate": 7.98631612216298e-06,
      "loss": 0.0063,
      "step": 67220
    },
    {
      "epoch": 1.1000572690828765,
      "grad_norm": 0.10789083689451218,
      "learning_rate": 7.985552568990134e-06,
      "loss": 0.0029,
      "step": 67230
    },
    {
      "epoch": 1.1002208950339525,
      "grad_norm": 0.13412733376026154,
      "learning_rate": 7.98478890759818e-06,
      "loss": 0.0027,
      "step": 67240
    },
    {
      "epoch": 1.1003845209850283,
      "grad_norm": 0.07595600187778473,
      "learning_rate": 7.984025138014803e-06,
      "loss": 0.0031,
      "step": 67250
    },
    {
      "epoch": 1.100548146936104,
      "grad_norm": 0.10003720968961716,
      "learning_rate": 7.98326126026769e-06,
      "loss": 0.003,
      "step": 67260
    },
    {
      "epoch": 1.1007117728871798,
      "grad_norm": 0.13010506331920624,
      "learning_rate": 7.982497274384523e-06,
      "loss": 0.003,
      "step": 67270
    },
    {
      "epoch": 1.1008753988382558,
      "grad_norm": 0.10623190551996231,
      "learning_rate": 7.981733180392999e-06,
      "loss": 0.0029,
      "step": 67280
    },
    {
      "epoch": 1.1010390247893316,
      "grad_norm": 0.2330292910337448,
      "learning_rate": 7.980968978320812e-06,
      "loss": 0.0026,
      "step": 67290
    },
    {
      "epoch": 1.1012026507404074,
      "grad_norm": 0.18196021020412445,
      "learning_rate": 7.980204668195665e-06,
      "loss": 0.0026,
      "step": 67300
    },
    {
      "epoch": 1.1013662766914833,
      "grad_norm": 0.15365463495254517,
      "learning_rate": 7.979440250045262e-06,
      "loss": 0.0015,
      "step": 67310
    },
    {
      "epoch": 1.1015299026425591,
      "grad_norm": 0.11036017537117004,
      "learning_rate": 7.978675723897312e-06,
      "loss": 0.0016,
      "step": 67320
    },
    {
      "epoch": 1.101693528593635,
      "grad_norm": 0.15332701802253723,
      "learning_rate": 7.977911089779525e-06,
      "loss": 0.0021,
      "step": 67330
    },
    {
      "epoch": 1.101857154544711,
      "grad_norm": 0.3528757095336914,
      "learning_rate": 7.977146347719618e-06,
      "loss": 0.0039,
      "step": 67340
    },
    {
      "epoch": 1.1020207804957867,
      "grad_norm": 0.056583061814308167,
      "learning_rate": 7.976381497745312e-06,
      "loss": 0.0015,
      "step": 67350
    },
    {
      "epoch": 1.1021844064468624,
      "grad_norm": 0.1486535668373108,
      "learning_rate": 7.97561653988433e-06,
      "loss": 0.0018,
      "step": 67360
    },
    {
      "epoch": 1.1023480323979382,
      "grad_norm": 0.1272255927324295,
      "learning_rate": 7.974851474164399e-06,
      "loss": 0.0025,
      "step": 67370
    },
    {
      "epoch": 1.1025116583490142,
      "grad_norm": 0.1979738473892212,
      "learning_rate": 7.974086300613252e-06,
      "loss": 0.0016,
      "step": 67380
    },
    {
      "epoch": 1.10267528430009,
      "grad_norm": 0.013305054977536201,
      "learning_rate": 7.973321019258627e-06,
      "loss": 0.0022,
      "step": 67390
    },
    {
      "epoch": 1.1028389102511658,
      "grad_norm": 0.0996328741312027,
      "learning_rate": 7.972555630128258e-06,
      "loss": 0.0026,
      "step": 67400
    },
    {
      "epoch": 1.1030025362022418,
      "grad_norm": 0.11364498734474182,
      "learning_rate": 7.971790133249895e-06,
      "loss": 0.0023,
      "step": 67410
    },
    {
      "epoch": 1.1031661621533175,
      "grad_norm": 0.0554550401866436,
      "learning_rate": 7.971024528651278e-06,
      "loss": 0.0018,
      "step": 67420
    },
    {
      "epoch": 1.1033297881043933,
      "grad_norm": 0.21192406117916107,
      "learning_rate": 7.970258816360165e-06,
      "loss": 0.0021,
      "step": 67430
    },
    {
      "epoch": 1.1034934140554693,
      "grad_norm": 0.23106718063354492,
      "learning_rate": 7.96949299640431e-06,
      "loss": 0.0029,
      "step": 67440
    },
    {
      "epoch": 1.103657040006545,
      "grad_norm": 0.1673358529806137,
      "learning_rate": 7.96872706881147e-06,
      "loss": 0.0029,
      "step": 67450
    },
    {
      "epoch": 1.1038206659576209,
      "grad_norm": 0.21540683507919312,
      "learning_rate": 7.967961033609407e-06,
      "loss": 0.0029,
      "step": 67460
    },
    {
      "epoch": 1.1039842919086966,
      "grad_norm": 0.241883784532547,
      "learning_rate": 7.967194890825893e-06,
      "loss": 0.0023,
      "step": 67470
    },
    {
      "epoch": 1.1041479178597726,
      "grad_norm": 0.0331779383122921,
      "learning_rate": 7.966428640488695e-06,
      "loss": 0.0025,
      "step": 67480
    },
    {
      "epoch": 1.1043115438108484,
      "grad_norm": 0.047347042709589005,
      "learning_rate": 7.965662282625586e-06,
      "loss": 0.0016,
      "step": 67490
    },
    {
      "epoch": 1.1044751697619242,
      "grad_norm": 0.14074267446994781,
      "learning_rate": 7.964895817264349e-06,
      "loss": 0.0023,
      "step": 67500
    },
    {
      "epoch": 1.1046387957130002,
      "grad_norm": 0.3148994743824005,
      "learning_rate": 7.964129244432763e-06,
      "loss": 0.0029,
      "step": 67510
    },
    {
      "epoch": 1.104802421664076,
      "grad_norm": 0.23355428874492645,
      "learning_rate": 7.963362564158619e-06,
      "loss": 0.0022,
      "step": 67520
    },
    {
      "epoch": 1.1049660476151517,
      "grad_norm": 0.11430051177740097,
      "learning_rate": 7.9625957764697e-06,
      "loss": 0.0032,
      "step": 67530
    },
    {
      "epoch": 1.1051296735662275,
      "grad_norm": 0.22647179663181305,
      "learning_rate": 7.961828881393808e-06,
      "loss": 0.003,
      "step": 67540
    },
    {
      "epoch": 1.1052932995173035,
      "grad_norm": 0.20279039442539215,
      "learning_rate": 7.961061878958736e-06,
      "loss": 0.0041,
      "step": 67550
    },
    {
      "epoch": 1.1054569254683793,
      "grad_norm": 0.0628499686717987,
      "learning_rate": 7.960294769192289e-06,
      "loss": 0.0019,
      "step": 67560
    },
    {
      "epoch": 1.105620551419455,
      "grad_norm": 0.0232760701328516,
      "learning_rate": 7.95952755212227e-06,
      "loss": 0.0021,
      "step": 67570
    },
    {
      "epoch": 1.105784177370531,
      "grad_norm": 0.14633068442344666,
      "learning_rate": 7.958760227776492e-06,
      "loss": 0.0015,
      "step": 67580
    },
    {
      "epoch": 1.1059478033216068,
      "grad_norm": 0.15686815977096558,
      "learning_rate": 7.957992796182765e-06,
      "loss": 0.0018,
      "step": 67590
    },
    {
      "epoch": 1.1061114292726826,
      "grad_norm": 0.04825974628329277,
      "learning_rate": 7.957225257368909e-06,
      "loss": 0.0026,
      "step": 67600
    },
    {
      "epoch": 1.1062750552237586,
      "grad_norm": 0.013961097225546837,
      "learning_rate": 7.956457611362747e-06,
      "loss": 0.0013,
      "step": 67610
    },
    {
      "epoch": 1.1064386811748343,
      "grad_norm": 0.15498970448970795,
      "learning_rate": 7.955689858192099e-06,
      "loss": 0.0059,
      "step": 67620
    },
    {
      "epoch": 1.1066023071259101,
      "grad_norm": 0.009553973563015461,
      "learning_rate": 7.954921997884799e-06,
      "loss": 0.0015,
      "step": 67630
    },
    {
      "epoch": 1.1067659330769861,
      "grad_norm": 0.058502793312072754,
      "learning_rate": 7.954154030468678e-06,
      "loss": 0.0017,
      "step": 67640
    },
    {
      "epoch": 1.106929559028062,
      "grad_norm": 0.10463238507509232,
      "learning_rate": 7.953385955971574e-06,
      "loss": 0.0024,
      "step": 67650
    },
    {
      "epoch": 1.1070931849791377,
      "grad_norm": 0.21524441242218018,
      "learning_rate": 7.952617774421328e-06,
      "loss": 0.0026,
      "step": 67660
    },
    {
      "epoch": 1.1072568109302134,
      "grad_norm": 0.09608050435781479,
      "learning_rate": 7.951849485845784e-06,
      "loss": 0.0046,
      "step": 67670
    },
    {
      "epoch": 1.1074204368812894,
      "grad_norm": 0.1355513632297516,
      "learning_rate": 7.95108109027279e-06,
      "loss": 0.0019,
      "step": 67680
    },
    {
      "epoch": 1.1075840628323652,
      "grad_norm": 0.14819177985191345,
      "learning_rate": 7.950312587730199e-06,
      "loss": 0.0028,
      "step": 67690
    },
    {
      "epoch": 1.107747688783441,
      "grad_norm": 0.23417548835277557,
      "learning_rate": 7.949543978245867e-06,
      "loss": 0.002,
      "step": 67700
    },
    {
      "epoch": 1.107911314734517,
      "grad_norm": 0.14641791582107544,
      "learning_rate": 7.948775261847656e-06,
      "loss": 0.0027,
      "step": 67710
    },
    {
      "epoch": 1.1080749406855928,
      "grad_norm": 0.10514626652002335,
      "learning_rate": 7.94800643856343e-06,
      "loss": 0.0035,
      "step": 67720
    },
    {
      "epoch": 1.1082385666366685,
      "grad_norm": 0.004406503867357969,
      "learning_rate": 7.947237508421054e-06,
      "loss": 0.0029,
      "step": 67730
    },
    {
      "epoch": 1.1084021925877443,
      "grad_norm": 0.15475668013095856,
      "learning_rate": 7.946468471448403e-06,
      "loss": 0.0034,
      "step": 67740
    },
    {
      "epoch": 1.1085658185388203,
      "grad_norm": 0.10073676705360413,
      "learning_rate": 7.94569932767335e-06,
      "loss": 0.0027,
      "step": 67750
    },
    {
      "epoch": 1.108729444489896,
      "grad_norm": 0.07992584258317947,
      "learning_rate": 7.944930077123777e-06,
      "loss": 0.002,
      "step": 67760
    },
    {
      "epoch": 1.1088930704409719,
      "grad_norm": 0.23091736435890198,
      "learning_rate": 7.944160719827569e-06,
      "loss": 0.0033,
      "step": 67770
    },
    {
      "epoch": 1.1090566963920478,
      "grad_norm": 0.11401800066232681,
      "learning_rate": 7.943391255812608e-06,
      "loss": 0.0021,
      "step": 67780
    },
    {
      "epoch": 1.1092203223431236,
      "grad_norm": 0.11418555676937103,
      "learning_rate": 7.942621685106789e-06,
      "loss": 0.0038,
      "step": 67790
    },
    {
      "epoch": 1.1093839482941994,
      "grad_norm": 0.043558269739151,
      "learning_rate": 7.941852007738006e-06,
      "loss": 0.004,
      "step": 67800
    },
    {
      "epoch": 1.1095475742452754,
      "grad_norm": 0.0932808667421341,
      "learning_rate": 7.94108222373416e-06,
      "loss": 0.003,
      "step": 67810
    },
    {
      "epoch": 1.1097112001963512,
      "grad_norm": 0.21740348637104034,
      "learning_rate": 7.94031233312315e-06,
      "loss": 0.0039,
      "step": 67820
    },
    {
      "epoch": 1.109874826147427,
      "grad_norm": 0.19150827825069427,
      "learning_rate": 7.939542335932886e-06,
      "loss": 0.0028,
      "step": 67830
    },
    {
      "epoch": 1.110038452098503,
      "grad_norm": 0.29575255513191223,
      "learning_rate": 7.938772232191276e-06,
      "loss": 0.0025,
      "step": 67840
    },
    {
      "epoch": 1.1102020780495787,
      "grad_norm": 0.21679075062274933,
      "learning_rate": 7.938002021926237e-06,
      "loss": 0.003,
      "step": 67850
    },
    {
      "epoch": 1.1103657040006545,
      "grad_norm": 0.12247305363416672,
      "learning_rate": 7.937231705165687e-06,
      "loss": 0.0025,
      "step": 67860
    },
    {
      "epoch": 1.1105293299517303,
      "grad_norm": 0.2624988257884979,
      "learning_rate": 7.936461281937545e-06,
      "loss": 0.0027,
      "step": 67870
    },
    {
      "epoch": 1.1106929559028063,
      "grad_norm": 0.23529274761676788,
      "learning_rate": 7.93569075226974e-06,
      "loss": 0.0021,
      "step": 67880
    },
    {
      "epoch": 1.110856581853882,
      "grad_norm": 0.34660282731056213,
      "learning_rate": 7.934920116190202e-06,
      "loss": 0.0013,
      "step": 67890
    },
    {
      "epoch": 1.1110202078049578,
      "grad_norm": 0.01306147687137127,
      "learning_rate": 7.934149373726862e-06,
      "loss": 0.0024,
      "step": 67900
    },
    {
      "epoch": 1.1111838337560338,
      "grad_norm": 0.0663568302989006,
      "learning_rate": 7.93337852490766e-06,
      "loss": 0.0017,
      "step": 67910
    },
    {
      "epoch": 1.1113474597071096,
      "grad_norm": 0.06737388670444489,
      "learning_rate": 7.932607569760537e-06,
      "loss": 0.0016,
      "step": 67920
    },
    {
      "epoch": 1.1115110856581853,
      "grad_norm": 0.10520081967115402,
      "learning_rate": 7.931836508313438e-06,
      "loss": 0.0021,
      "step": 67930
    },
    {
      "epoch": 1.1116747116092611,
      "grad_norm": 0.1524018794298172,
      "learning_rate": 7.93106534059431e-06,
      "loss": 0.003,
      "step": 67940
    },
    {
      "epoch": 1.1118383375603371,
      "grad_norm": 0.15310916304588318,
      "learning_rate": 7.93029406663111e-06,
      "loss": 0.0011,
      "step": 67950
    },
    {
      "epoch": 1.112001963511413,
      "grad_norm": 0.08758015930652618,
      "learning_rate": 7.929522686451793e-06,
      "loss": 0.0026,
      "step": 67960
    },
    {
      "epoch": 1.1121655894624887,
      "grad_norm": 0.15671318769454956,
      "learning_rate": 7.928751200084319e-06,
      "loss": 0.0031,
      "step": 67970
    },
    {
      "epoch": 1.1123292154135647,
      "grad_norm": 0.16461004316806793,
      "learning_rate": 7.927979607556652e-06,
      "loss": 0.0038,
      "step": 67980
    },
    {
      "epoch": 1.1124928413646404,
      "grad_norm": 0.1444018930196762,
      "learning_rate": 7.927207908896763e-06,
      "loss": 0.0027,
      "step": 67990
    },
    {
      "epoch": 1.1126564673157162,
      "grad_norm": 0.07200793921947479,
      "learning_rate": 7.926436104132622e-06,
      "loss": 0.0023,
      "step": 68000
    },
    {
      "epoch": 1.1128200932667922,
      "grad_norm": 0.03750094026327133,
      "learning_rate": 7.925664193292206e-06,
      "loss": 0.003,
      "step": 68010
    },
    {
      "epoch": 1.112983719217868,
      "grad_norm": 0.05523361265659332,
      "learning_rate": 7.924892176403494e-06,
      "loss": 0.0037,
      "step": 68020
    },
    {
      "epoch": 1.1131473451689438,
      "grad_norm": 0.19418194890022278,
      "learning_rate": 7.924120053494471e-06,
      "loss": 0.0042,
      "step": 68030
    },
    {
      "epoch": 1.1133109711200195,
      "grad_norm": 0.06008940562605858,
      "learning_rate": 7.923347824593125e-06,
      "loss": 0.0032,
      "step": 68040
    },
    {
      "epoch": 1.1134745970710955,
      "grad_norm": 0.1261216104030609,
      "learning_rate": 7.922575489727444e-06,
      "loss": 0.0027,
      "step": 68050
    },
    {
      "epoch": 1.1136382230221713,
      "grad_norm": 0.10495466738939285,
      "learning_rate": 7.921803048925427e-06,
      "loss": 0.0022,
      "step": 68060
    },
    {
      "epoch": 1.113801848973247,
      "grad_norm": 0.050722554326057434,
      "learning_rate": 7.921030502215072e-06,
      "loss": 0.0025,
      "step": 68070
    },
    {
      "epoch": 1.113965474924323,
      "grad_norm": 0.14870263636112213,
      "learning_rate": 7.920257849624382e-06,
      "loss": 0.0021,
      "step": 68080
    },
    {
      "epoch": 1.1141291008753988,
      "grad_norm": 0.050933562219142914,
      "learning_rate": 7.919485091181362e-06,
      "loss": 0.0022,
      "step": 68090
    },
    {
      "epoch": 1.1142927268264746,
      "grad_norm": 0.15519891679286957,
      "learning_rate": 7.918712226914027e-06,
      "loss": 0.0019,
      "step": 68100
    },
    {
      "epoch": 1.1144563527775506,
      "grad_norm": 0.025211861357092857,
      "learning_rate": 7.917939256850387e-06,
      "loss": 0.0023,
      "step": 68110
    },
    {
      "epoch": 1.1146199787286264,
      "grad_norm": 0.10136483609676361,
      "learning_rate": 7.917166181018464e-06,
      "loss": 0.0024,
      "step": 68120
    },
    {
      "epoch": 1.1147836046797022,
      "grad_norm": 0.2363646924495697,
      "learning_rate": 7.916392999446274e-06,
      "loss": 0.0016,
      "step": 68130
    },
    {
      "epoch": 1.114947230630778,
      "grad_norm": 0.1900477111339569,
      "learning_rate": 7.91561971216185e-06,
      "loss": 0.0033,
      "step": 68140
    },
    {
      "epoch": 1.115110856581854,
      "grad_norm": 0.14923186600208282,
      "learning_rate": 7.91484631919322e-06,
      "loss": 0.0021,
      "step": 68150
    },
    {
      "epoch": 1.1152744825329297,
      "grad_norm": 0.13170966506004333,
      "learning_rate": 7.914072820568415e-06,
      "loss": 0.0015,
      "step": 68160
    },
    {
      "epoch": 1.1154381084840055,
      "grad_norm": 0.1346287876367569,
      "learning_rate": 7.913299216315474e-06,
      "loss": 0.0027,
      "step": 68170
    },
    {
      "epoch": 1.1156017344350815,
      "grad_norm": 0.10992290824651718,
      "learning_rate": 7.912525506462439e-06,
      "loss": 0.0018,
      "step": 68180
    },
    {
      "epoch": 1.1157653603861573,
      "grad_norm": 0.0763528048992157,
      "learning_rate": 7.911751691037353e-06,
      "loss": 0.0018,
      "step": 68190
    },
    {
      "epoch": 1.115928986337233,
      "grad_norm": 0.1582881212234497,
      "learning_rate": 7.910977770068267e-06,
      "loss": 0.0033,
      "step": 68200
    },
    {
      "epoch": 1.116092612288309,
      "grad_norm": 0.10619562119245529,
      "learning_rate": 7.910203743583234e-06,
      "loss": 0.0015,
      "step": 68210
    },
    {
      "epoch": 1.1162562382393848,
      "grad_norm": 0.11039046198129654,
      "learning_rate": 7.90942961161031e-06,
      "loss": 0.0019,
      "step": 68220
    },
    {
      "epoch": 1.1164198641904606,
      "grad_norm": 0.17310935258865356,
      "learning_rate": 7.908655374177555e-06,
      "loss": 0.0031,
      "step": 68230
    },
    {
      "epoch": 1.1165834901415364,
      "grad_norm": 0.1316337138414383,
      "learning_rate": 7.907881031313032e-06,
      "loss": 0.0018,
      "step": 68240
    },
    {
      "epoch": 1.1167471160926123,
      "grad_norm": 0.0757899135351181,
      "learning_rate": 7.90710658304481e-06,
      "loss": 0.0016,
      "step": 68250
    },
    {
      "epoch": 1.1169107420436881,
      "grad_norm": 0.07081680744886398,
      "learning_rate": 7.906332029400964e-06,
      "loss": 0.0024,
      "step": 68260
    },
    {
      "epoch": 1.117074367994764,
      "grad_norm": 0.3196355998516083,
      "learning_rate": 7.905557370409566e-06,
      "loss": 0.0041,
      "step": 68270
    },
    {
      "epoch": 1.11723799394584,
      "grad_norm": 0.051009662449359894,
      "learning_rate": 7.904782606098698e-06,
      "loss": 0.0018,
      "step": 68280
    },
    {
      "epoch": 1.1174016198969157,
      "grad_norm": 0.18096141517162323,
      "learning_rate": 7.90400773649644e-06,
      "loss": 0.0019,
      "step": 68290
    },
    {
      "epoch": 1.1175652458479914,
      "grad_norm": 0.1862032115459442,
      "learning_rate": 7.903232761630881e-06,
      "loss": 0.002,
      "step": 68300
    },
    {
      "epoch": 1.1177288717990674,
      "grad_norm": 0.17516174912452698,
      "learning_rate": 7.902457681530114e-06,
      "loss": 0.0012,
      "step": 68310
    },
    {
      "epoch": 1.1178924977501432,
      "grad_norm": 0.1434076577425003,
      "learning_rate": 7.901682496222231e-06,
      "loss": 0.0036,
      "step": 68320
    },
    {
      "epoch": 1.118056123701219,
      "grad_norm": 0.0856248065829277,
      "learning_rate": 7.900907205735331e-06,
      "loss": 0.0035,
      "step": 68330
    },
    {
      "epoch": 1.1182197496522948,
      "grad_norm": 0.10628880560398102,
      "learning_rate": 7.900131810097517e-06,
      "loss": 0.0024,
      "step": 68340
    },
    {
      "epoch": 1.1183833756033708,
      "grad_norm": 0.0780029296875,
      "learning_rate": 7.899356309336895e-06,
      "loss": 0.0021,
      "step": 68350
    },
    {
      "epoch": 1.1185470015544465,
      "grad_norm": 0.045987654477357864,
      "learning_rate": 7.898580703481577e-06,
      "loss": 0.0037,
      "step": 68360
    },
    {
      "epoch": 1.1187106275055223,
      "grad_norm": 0.02341105043888092,
      "learning_rate": 7.897804992559674e-06,
      "loss": 0.0023,
      "step": 68370
    },
    {
      "epoch": 1.1188742534565983,
      "grad_norm": 0.0918494388461113,
      "learning_rate": 7.897029176599304e-06,
      "loss": 0.0021,
      "step": 68380
    },
    {
      "epoch": 1.119037879407674,
      "grad_norm": 0.17768150568008423,
      "learning_rate": 7.89625325562859e-06,
      "loss": 0.0029,
      "step": 68390
    },
    {
      "epoch": 1.1192015053587498,
      "grad_norm": 0.0587909072637558,
      "learning_rate": 7.895477229675655e-06,
      "loss": 0.0024,
      "step": 68400
    },
    {
      "epoch": 1.1193651313098258,
      "grad_norm": 0.1557624340057373,
      "learning_rate": 7.894701098768631e-06,
      "loss": 0.0024,
      "step": 68410
    },
    {
      "epoch": 1.1195287572609016,
      "grad_norm": 0.04897768795490265,
      "learning_rate": 7.893924862935648e-06,
      "loss": 0.0028,
      "step": 68420
    },
    {
      "epoch": 1.1196923832119774,
      "grad_norm": 0.08152521401643753,
      "learning_rate": 7.893148522204847e-06,
      "loss": 0.0028,
      "step": 68430
    },
    {
      "epoch": 1.1198560091630532,
      "grad_norm": 0.13534341752529144,
      "learning_rate": 7.892372076604363e-06,
      "loss": 0.0021,
      "step": 68440
    },
    {
      "epoch": 1.1200196351141292,
      "grad_norm": 0.2423546314239502,
      "learning_rate": 7.891595526162344e-06,
      "loss": 0.0022,
      "step": 68450
    },
    {
      "epoch": 1.120183261065205,
      "grad_norm": 0.26700547337532043,
      "learning_rate": 7.890818870906937e-06,
      "loss": 0.0027,
      "step": 68460
    },
    {
      "epoch": 1.1203468870162807,
      "grad_norm": 0.06260558217763901,
      "learning_rate": 7.890042110866293e-06,
      "loss": 0.002,
      "step": 68470
    },
    {
      "epoch": 1.1205105129673567,
      "grad_norm": 0.02388470619916916,
      "learning_rate": 7.889265246068569e-06,
      "loss": 0.0024,
      "step": 68480
    },
    {
      "epoch": 1.1206741389184325,
      "grad_norm": 0.16415877640247345,
      "learning_rate": 7.888488276541923e-06,
      "loss": 0.0029,
      "step": 68490
    },
    {
      "epoch": 1.1208377648695083,
      "grad_norm": 0.10648050904273987,
      "learning_rate": 7.88771120231452e-06,
      "loss": 0.0019,
      "step": 68500
    },
    {
      "epoch": 1.121001390820584,
      "grad_norm": 0.11352626979351044,
      "learning_rate": 7.886934023414527e-06,
      "loss": 0.0017,
      "step": 68510
    },
    {
      "epoch": 1.12116501677166,
      "grad_norm": 0.22070206701755524,
      "learning_rate": 7.886156739870114e-06,
      "loss": 0.0018,
      "step": 68520
    },
    {
      "epoch": 1.1213286427227358,
      "grad_norm": 0.08376161754131317,
      "learning_rate": 7.885379351709455e-06,
      "loss": 0.0032,
      "step": 68530
    },
    {
      "epoch": 1.1214922686738116,
      "grad_norm": 0.14602339267730713,
      "learning_rate": 7.88460185896073e-06,
      "loss": 0.0024,
      "step": 68540
    },
    {
      "epoch": 1.1216558946248876,
      "grad_norm": 0.17993013560771942,
      "learning_rate": 7.88382426165212e-06,
      "loss": 0.0024,
      "step": 68550
    },
    {
      "epoch": 1.1218195205759633,
      "grad_norm": 0.1910533607006073,
      "learning_rate": 7.883046559811813e-06,
      "loss": 0.0032,
      "step": 68560
    },
    {
      "epoch": 1.1219831465270391,
      "grad_norm": 0.22462201118469238,
      "learning_rate": 7.882268753467997e-06,
      "loss": 0.0031,
      "step": 68570
    },
    {
      "epoch": 1.1221467724781151,
      "grad_norm": 0.1692439168691635,
      "learning_rate": 7.881490842648864e-06,
      "loss": 0.0025,
      "step": 68580
    },
    {
      "epoch": 1.122310398429191,
      "grad_norm": 0.09960126131772995,
      "learning_rate": 7.880712827382615e-06,
      "loss": 0.0039,
      "step": 68590
    },
    {
      "epoch": 1.1224740243802667,
      "grad_norm": 0.03694210201501846,
      "learning_rate": 7.879934707697449e-06,
      "loss": 0.0034,
      "step": 68600
    },
    {
      "epoch": 1.1226376503313427,
      "grad_norm": 0.3607424199581146,
      "learning_rate": 7.879156483621573e-06,
      "loss": 0.0036,
      "step": 68610
    },
    {
      "epoch": 1.1228012762824184,
      "grad_norm": 0.03478715941309929,
      "learning_rate": 7.878378155183192e-06,
      "loss": 0.0016,
      "step": 68620
    },
    {
      "epoch": 1.1229649022334942,
      "grad_norm": 0.03479054197669029,
      "learning_rate": 7.877599722410524e-06,
      "loss": 0.0018,
      "step": 68630
    },
    {
      "epoch": 1.12312852818457,
      "grad_norm": 0.1400851309299469,
      "learning_rate": 7.87682118533178e-06,
      "loss": 0.0026,
      "step": 68640
    },
    {
      "epoch": 1.123292154135646,
      "grad_norm": 0.14542587101459503,
      "learning_rate": 7.876042543975183e-06,
      "loss": 0.0032,
      "step": 68650
    },
    {
      "epoch": 1.1234557800867218,
      "grad_norm": 0.16380688548088074,
      "learning_rate": 7.875263798368957e-06,
      "loss": 0.0027,
      "step": 68660
    },
    {
      "epoch": 1.1236194060377975,
      "grad_norm": 0.03666701912879944,
      "learning_rate": 7.874484948541328e-06,
      "loss": 0.002,
      "step": 68670
    },
    {
      "epoch": 1.1237830319888735,
      "grad_norm": 0.13301661610603333,
      "learning_rate": 7.873705994520527e-06,
      "loss": 0.0011,
      "step": 68680
    },
    {
      "epoch": 1.1239466579399493,
      "grad_norm": 0.19392654299736023,
      "learning_rate": 7.872926936334792e-06,
      "loss": 0.0024,
      "step": 68690
    },
    {
      "epoch": 1.124110283891025,
      "grad_norm": 0.1347409337759018,
      "learning_rate": 7.87214777401236e-06,
      "loss": 0.0018,
      "step": 68700
    },
    {
      "epoch": 1.1242739098421008,
      "grad_norm": 0.07486072182655334,
      "learning_rate": 7.871368507581474e-06,
      "loss": 0.0013,
      "step": 68710
    },
    {
      "epoch": 1.1244375357931768,
      "grad_norm": 0.09588923305273056,
      "learning_rate": 7.870589137070382e-06,
      "loss": 0.0018,
      "step": 68720
    },
    {
      "epoch": 1.1246011617442526,
      "grad_norm": 0.05466753616929054,
      "learning_rate": 7.869809662507331e-06,
      "loss": 0.0022,
      "step": 68730
    },
    {
      "epoch": 1.1247647876953284,
      "grad_norm": 0.057994551956653595,
      "learning_rate": 7.869030083920578e-06,
      "loss": 0.0023,
      "step": 68740
    },
    {
      "epoch": 1.1249284136464044,
      "grad_norm": 0.0489218533039093,
      "learning_rate": 7.86825040133838e-06,
      "loss": 0.0024,
      "step": 68750
    },
    {
      "epoch": 1.1250920395974802,
      "grad_norm": 0.08175227791070938,
      "learning_rate": 7.867470614788998e-06,
      "loss": 0.0034,
      "step": 68760
    },
    {
      "epoch": 1.125255665548556,
      "grad_norm": 0.02568182907998562,
      "learning_rate": 7.8666907243007e-06,
      "loss": 0.0023,
      "step": 68770
    },
    {
      "epoch": 1.125419291499632,
      "grad_norm": 0.04513145238161087,
      "learning_rate": 7.865910729901751e-06,
      "loss": 0.0018,
      "step": 68780
    },
    {
      "epoch": 1.1255829174507077,
      "grad_norm": 0.13794945180416107,
      "learning_rate": 7.865130631620428e-06,
      "loss": 0.0022,
      "step": 68790
    },
    {
      "epoch": 1.1257465434017835,
      "grad_norm": 0.1073811948299408,
      "learning_rate": 7.864350429485004e-06,
      "loss": 0.0044,
      "step": 68800
    },
    {
      "epoch": 1.1259101693528595,
      "grad_norm": 0.009210161864757538,
      "learning_rate": 7.863570123523763e-06,
      "loss": 0.0023,
      "step": 68810
    },
    {
      "epoch": 1.1260737953039353,
      "grad_norm": 0.04893968626856804,
      "learning_rate": 7.862789713764987e-06,
      "loss": 0.002,
      "step": 68820
    },
    {
      "epoch": 1.126237421255011,
      "grad_norm": 0.06745315343141556,
      "learning_rate": 7.862009200236962e-06,
      "loss": 0.0025,
      "step": 68830
    },
    {
      "epoch": 1.1264010472060868,
      "grad_norm": 0.14464573562145233,
      "learning_rate": 7.861228582967985e-06,
      "loss": 0.0039,
      "step": 68840
    },
    {
      "epoch": 1.1265646731571628,
      "grad_norm": 0.035705361515283585,
      "learning_rate": 7.860447861986347e-06,
      "loss": 0.0018,
      "step": 68850
    },
    {
      "epoch": 1.1267282991082386,
      "grad_norm": 0.11243464797735214,
      "learning_rate": 7.859667037320351e-06,
      "loss": 0.0021,
      "step": 68860
    },
    {
      "epoch": 1.1268919250593143,
      "grad_norm": 0.11256809532642365,
      "learning_rate": 7.858886108998296e-06,
      "loss": 0.0021,
      "step": 68870
    },
    {
      "epoch": 1.1270555510103903,
      "grad_norm": 0.328066349029541,
      "learning_rate": 7.85810507704849e-06,
      "loss": 0.0027,
      "step": 68880
    },
    {
      "epoch": 1.1272191769614661,
      "grad_norm": 0.2756465971469879,
      "learning_rate": 7.857323941499247e-06,
      "loss": 0.0022,
      "step": 68890
    },
    {
      "epoch": 1.127382802912542,
      "grad_norm": 0.058144621551036835,
      "learning_rate": 7.856542702378875e-06,
      "loss": 0.0024,
      "step": 68900
    },
    {
      "epoch": 1.1275464288636177,
      "grad_norm": 0.0866839587688446,
      "learning_rate": 7.8557613597157e-06,
      "loss": 0.0012,
      "step": 68910
    },
    {
      "epoch": 1.1277100548146937,
      "grad_norm": 0.12420640885829926,
      "learning_rate": 7.854979913538035e-06,
      "loss": 0.0016,
      "step": 68920
    },
    {
      "epoch": 1.1278736807657694,
      "grad_norm": 0.017458591610193253,
      "learning_rate": 7.854198363874212e-06,
      "loss": 0.0025,
      "step": 68930
    },
    {
      "epoch": 1.1280373067168452,
      "grad_norm": 0.060509685426950455,
      "learning_rate": 7.853416710752556e-06,
      "loss": 0.0036,
      "step": 68940
    },
    {
      "epoch": 1.1282009326679212,
      "grad_norm": 0.08177191019058228,
      "learning_rate": 7.852634954201405e-06,
      "loss": 0.0019,
      "step": 68950
    },
    {
      "epoch": 1.128364558618997,
      "grad_norm": 0.23244555294513702,
      "learning_rate": 7.851853094249093e-06,
      "loss": 0.0033,
      "step": 68960
    },
    {
      "epoch": 1.1285281845700728,
      "grad_norm": 0.23035791516304016,
      "learning_rate": 7.851071130923959e-06,
      "loss": 0.0024,
      "step": 68970
    },
    {
      "epoch": 1.1286918105211488,
      "grad_norm": 0.4902045428752899,
      "learning_rate": 7.85028906425435e-06,
      "loss": 0.0034,
      "step": 68980
    },
    {
      "epoch": 1.1288554364722245,
      "grad_norm": 0.06800995022058487,
      "learning_rate": 7.849506894268613e-06,
      "loss": 0.0019,
      "step": 68990
    },
    {
      "epoch": 1.1290190624233003,
      "grad_norm": 0.14782725274562836,
      "learning_rate": 7.8487246209951e-06,
      "loss": 0.002,
      "step": 69000
    },
    {
      "epoch": 1.1291826883743763,
      "grad_norm": 0.16419872641563416,
      "learning_rate": 7.847942244462165e-06,
      "loss": 0.002,
      "step": 69010
    },
    {
      "epoch": 1.129346314325452,
      "grad_norm": 0.1362604796886444,
      "learning_rate": 7.84715976469817e-06,
      "loss": 0.0046,
      "step": 69020
    },
    {
      "epoch": 1.1295099402765278,
      "grad_norm": 0.3837820887565613,
      "learning_rate": 7.846377181731475e-06,
      "loss": 0.0038,
      "step": 69030
    },
    {
      "epoch": 1.1296735662276036,
      "grad_norm": 0.040347859263420105,
      "learning_rate": 7.845594495590448e-06,
      "loss": 0.0023,
      "step": 69040
    },
    {
      "epoch": 1.1298371921786796,
      "grad_norm": 0.12403661012649536,
      "learning_rate": 7.844811706303461e-06,
      "loss": 0.0022,
      "step": 69050
    },
    {
      "epoch": 1.1300008181297554,
      "grad_norm": 0.02095261961221695,
      "learning_rate": 7.844028813898887e-06,
      "loss": 0.0025,
      "step": 69060
    },
    {
      "epoch": 1.1301644440808312,
      "grad_norm": 0.0585404597222805,
      "learning_rate": 7.843245818405103e-06,
      "loss": 0.0016,
      "step": 69070
    },
    {
      "epoch": 1.130328070031907,
      "grad_norm": 0.10304427146911621,
      "learning_rate": 7.842462719850492e-06,
      "loss": 0.0022,
      "step": 69080
    },
    {
      "epoch": 1.130491695982983,
      "grad_norm": 0.04839986562728882,
      "learning_rate": 7.841679518263439e-06,
      "loss": 0.0027,
      "step": 69090
    },
    {
      "epoch": 1.1306553219340587,
      "grad_norm": 0.06605979055166245,
      "learning_rate": 7.840896213672334e-06,
      "loss": 0.0023,
      "step": 69100
    },
    {
      "epoch": 1.1308189478851345,
      "grad_norm": 0.1461275964975357,
      "learning_rate": 7.840112806105568e-06,
      "loss": 0.0021,
      "step": 69110
    },
    {
      "epoch": 1.1309825738362105,
      "grad_norm": 0.0795038491487503,
      "learning_rate": 7.83932929559154e-06,
      "loss": 0.0031,
      "step": 69120
    },
    {
      "epoch": 1.1311461997872863,
      "grad_norm": 0.14406965672969818,
      "learning_rate": 7.838545682158647e-06,
      "loss": 0.0023,
      "step": 69130
    },
    {
      "epoch": 1.131309825738362,
      "grad_norm": 0.008858182467520237,
      "learning_rate": 7.837761965835298e-06,
      "loss": 0.0021,
      "step": 69140
    },
    {
      "epoch": 1.131473451689438,
      "grad_norm": 0.13654375076293945,
      "learning_rate": 7.836978146649897e-06,
      "loss": 0.0031,
      "step": 69150
    },
    {
      "epoch": 1.1316370776405138,
      "grad_norm": 0.06601133197546005,
      "learning_rate": 7.836194224630857e-06,
      "loss": 0.0013,
      "step": 69160
    },
    {
      "epoch": 1.1318007035915896,
      "grad_norm": 0.509337306022644,
      "learning_rate": 7.835410199806592e-06,
      "loss": 0.0025,
      "step": 69170
    },
    {
      "epoch": 1.1319643295426656,
      "grad_norm": 0.18314862251281738,
      "learning_rate": 7.834626072205522e-06,
      "loss": 0.0026,
      "step": 69180
    },
    {
      "epoch": 1.1321279554937413,
      "grad_norm": 0.1151198074221611,
      "learning_rate": 7.83384184185607e-06,
      "loss": 0.0028,
      "step": 69190
    },
    {
      "epoch": 1.1322915814448171,
      "grad_norm": 0.057028695940971375,
      "learning_rate": 7.833057508786664e-06,
      "loss": 0.0013,
      "step": 69200
    },
    {
      "epoch": 1.132455207395893,
      "grad_norm": 0.13587886095046997,
      "learning_rate": 7.832273073025732e-06,
      "loss": 0.0036,
      "step": 69210
    },
    {
      "epoch": 1.132618833346969,
      "grad_norm": 0.1952548623085022,
      "learning_rate": 7.831488534601706e-06,
      "loss": 0.0021,
      "step": 69220
    },
    {
      "epoch": 1.1327824592980447,
      "grad_norm": 0.15132951736450195,
      "learning_rate": 7.830703893543026e-06,
      "loss": 0.002,
      "step": 69230
    },
    {
      "epoch": 1.1329460852491204,
      "grad_norm": 0.08181136846542358,
      "learning_rate": 7.829919149878134e-06,
      "loss": 0.0018,
      "step": 69240
    },
    {
      "epoch": 1.1331097112001964,
      "grad_norm": 0.17251145839691162,
      "learning_rate": 7.829134303635474e-06,
      "loss": 0.0024,
      "step": 69250
    },
    {
      "epoch": 1.1332733371512722,
      "grad_norm": 0.17258213460445404,
      "learning_rate": 7.828349354843498e-06,
      "loss": 0.0025,
      "step": 69260
    },
    {
      "epoch": 1.133436963102348,
      "grad_norm": 0.250765860080719,
      "learning_rate": 7.827564303530653e-06,
      "loss": 0.0039,
      "step": 69270
    },
    {
      "epoch": 1.1336005890534238,
      "grad_norm": 0.12064200639724731,
      "learning_rate": 7.826779149725398e-06,
      "loss": 0.0024,
      "step": 69280
    },
    {
      "epoch": 1.1337642150044998,
      "grad_norm": 0.16128379106521606,
      "learning_rate": 7.825993893456192e-06,
      "loss": 0.0014,
      "step": 69290
    },
    {
      "epoch": 1.1339278409555755,
      "grad_norm": 0.2249792516231537,
      "learning_rate": 7.825208534751501e-06,
      "loss": 0.0018,
      "step": 69300
    },
    {
      "epoch": 1.1340914669066513,
      "grad_norm": 0.13834290206432343,
      "learning_rate": 7.824423073639791e-06,
      "loss": 0.0018,
      "step": 69310
    },
    {
      "epoch": 1.1342550928577273,
      "grad_norm": 0.32640811800956726,
      "learning_rate": 7.823637510149533e-06,
      "loss": 0.0029,
      "step": 69320
    },
    {
      "epoch": 1.134418718808803,
      "grad_norm": 0.49388423562049866,
      "learning_rate": 7.822851844309201e-06,
      "loss": 0.0018,
      "step": 69330
    },
    {
      "epoch": 1.1345823447598788,
      "grad_norm": 0.02632729709148407,
      "learning_rate": 7.822066076147274e-06,
      "loss": 0.002,
      "step": 69340
    },
    {
      "epoch": 1.1347459707109548,
      "grad_norm": 0.059555403888225555,
      "learning_rate": 7.821280205692234e-06,
      "loss": 0.002,
      "step": 69350
    },
    {
      "epoch": 1.1349095966620306,
      "grad_norm": 0.3162457346916199,
      "learning_rate": 7.82049423297257e-06,
      "loss": 0.0024,
      "step": 69360
    },
    {
      "epoch": 1.1350732226131064,
      "grad_norm": 0.09541064500808716,
      "learning_rate": 7.819708158016765e-06,
      "loss": 0.0021,
      "step": 69370
    },
    {
      "epoch": 1.1352368485641824,
      "grad_norm": 0.15407849848270416,
      "learning_rate": 7.818921980853317e-06,
      "loss": 0.0026,
      "step": 69380
    },
    {
      "epoch": 1.1354004745152582,
      "grad_norm": 0.06315534561872482,
      "learning_rate": 7.818135701510725e-06,
      "loss": 0.0031,
      "step": 69390
    },
    {
      "epoch": 1.135564100466334,
      "grad_norm": 0.08403895795345306,
      "learning_rate": 7.817349320017485e-06,
      "loss": 0.0026,
      "step": 69400
    },
    {
      "epoch": 1.1357277264174097,
      "grad_norm": 0.06529895216226578,
      "learning_rate": 7.816562836402104e-06,
      "loss": 0.0029,
      "step": 69410
    },
    {
      "epoch": 1.1358913523684857,
      "grad_norm": 0.1551051288843155,
      "learning_rate": 7.81577625069309e-06,
      "loss": 0.0021,
      "step": 69420
    },
    {
      "epoch": 1.1360549783195615,
      "grad_norm": 0.482984334230423,
      "learning_rate": 7.814989562918954e-06,
      "loss": 0.0027,
      "step": 69430
    },
    {
      "epoch": 1.1362186042706373,
      "grad_norm": 0.18356969952583313,
      "learning_rate": 7.814202773108212e-06,
      "loss": 0.0018,
      "step": 69440
    },
    {
      "epoch": 1.1363822302217133,
      "grad_norm": 0.17813226580619812,
      "learning_rate": 7.813415881289384e-06,
      "loss": 0.0024,
      "step": 69450
    },
    {
      "epoch": 1.136545856172789,
      "grad_norm": 0.04675832763314247,
      "learning_rate": 7.81262888749099e-06,
      "loss": 0.0028,
      "step": 69460
    },
    {
      "epoch": 1.1367094821238648,
      "grad_norm": 0.16692353785037994,
      "learning_rate": 7.811841791741561e-06,
      "loss": 0.0024,
      "step": 69470
    },
    {
      "epoch": 1.1368731080749406,
      "grad_norm": 0.06796851754188538,
      "learning_rate": 7.811054594069625e-06,
      "loss": 0.0018,
      "step": 69480
    },
    {
      "epoch": 1.1370367340260166,
      "grad_norm": 0.16330274939537048,
      "learning_rate": 7.810267294503715e-06,
      "loss": 0.0032,
      "step": 69490
    },
    {
      "epoch": 1.1372003599770923,
      "grad_norm": 0.1672714352607727,
      "learning_rate": 7.809479893072369e-06,
      "loss": 0.0024,
      "step": 69500
    },
    {
      "epoch": 1.1373639859281681,
      "grad_norm": 0.02578585222363472,
      "learning_rate": 7.80869238980413e-06,
      "loss": 0.0021,
      "step": 69510
    },
    {
      "epoch": 1.1375276118792441,
      "grad_norm": 0.12554201483726501,
      "learning_rate": 7.807904784727544e-06,
      "loss": 0.0034,
      "step": 69520
    },
    {
      "epoch": 1.13769123783032,
      "grad_norm": 0.015130684711039066,
      "learning_rate": 7.807117077871157e-06,
      "loss": 0.0024,
      "step": 69530
    },
    {
      "epoch": 1.1378548637813957,
      "grad_norm": 0.06824557483196259,
      "learning_rate": 7.806329269263523e-06,
      "loss": 0.002,
      "step": 69540
    },
    {
      "epoch": 1.1380184897324717,
      "grad_norm": 0.038978103548288345,
      "learning_rate": 7.805541358933197e-06,
      "loss": 0.002,
      "step": 69550
    },
    {
      "epoch": 1.1381821156835474,
      "grad_norm": 0.04791449010372162,
      "learning_rate": 7.804753346908742e-06,
      "loss": 0.0012,
      "step": 69560
    },
    {
      "epoch": 1.1383457416346232,
      "grad_norm": 0.06030956283211708,
      "learning_rate": 7.803965233218717e-06,
      "loss": 0.002,
      "step": 69570
    },
    {
      "epoch": 1.1385093675856992,
      "grad_norm": 0.11117040365934372,
      "learning_rate": 7.803177017891693e-06,
      "loss": 0.002,
      "step": 69580
    },
    {
      "epoch": 1.138672993536775,
      "grad_norm": 0.13186952471733093,
      "learning_rate": 7.802388700956238e-06,
      "loss": 0.0025,
      "step": 69590
    },
    {
      "epoch": 1.1388366194878508,
      "grad_norm": 0.06334838271141052,
      "learning_rate": 7.801600282440929e-06,
      "loss": 0.0011,
      "step": 69600
    },
    {
      "epoch": 1.1390002454389265,
      "grad_norm": 0.013611716218292713,
      "learning_rate": 7.800811762374342e-06,
      "loss": 0.0016,
      "step": 69610
    },
    {
      "epoch": 1.1391638713900025,
      "grad_norm": 0.060349758714437485,
      "learning_rate": 7.800023140785061e-06,
      "loss": 0.0017,
      "step": 69620
    },
    {
      "epoch": 1.1393274973410783,
      "grad_norm": 0.11975428462028503,
      "learning_rate": 7.799234417701672e-06,
      "loss": 0.0023,
      "step": 69630
    },
    {
      "epoch": 1.139491123292154,
      "grad_norm": 0.10301206260919571,
      "learning_rate": 7.798445593152761e-06,
      "loss": 0.0024,
      "step": 69640
    },
    {
      "epoch": 1.13965474924323,
      "grad_norm": 0.07847728580236435,
      "learning_rate": 7.797656667166923e-06,
      "loss": 0.0036,
      "step": 69650
    },
    {
      "epoch": 1.1398183751943058,
      "grad_norm": 0.043589264154434204,
      "learning_rate": 7.796867639772755e-06,
      "loss": 0.0016,
      "step": 69660
    },
    {
      "epoch": 1.1399820011453816,
      "grad_norm": 0.23483453691005707,
      "learning_rate": 7.79607851099886e-06,
      "loss": 0.0029,
      "step": 69670
    },
    {
      "epoch": 1.1401456270964574,
      "grad_norm": 0.0905677080154419,
      "learning_rate": 7.795289280873836e-06,
      "loss": 0.0034,
      "step": 69680
    },
    {
      "epoch": 1.1403092530475334,
      "grad_norm": 0.1493551880121231,
      "learning_rate": 7.794499949426293e-06,
      "loss": 0.0036,
      "step": 69690
    },
    {
      "epoch": 1.1404728789986092,
      "grad_norm": 0.21221008896827698,
      "learning_rate": 7.793710516684847e-06,
      "loss": 0.0027,
      "step": 69700
    },
    {
      "epoch": 1.140636504949685,
      "grad_norm": 0.16252242028713226,
      "learning_rate": 7.792920982678107e-06,
      "loss": 0.0023,
      "step": 69710
    },
    {
      "epoch": 1.140800130900761,
      "grad_norm": 0.18081463873386383,
      "learning_rate": 7.792131347434695e-06,
      "loss": 0.0027,
      "step": 69720
    },
    {
      "epoch": 1.1409637568518367,
      "grad_norm": 0.1700911670923233,
      "learning_rate": 7.791341610983232e-06,
      "loss": 0.0018,
      "step": 69730
    },
    {
      "epoch": 1.1411273828029125,
      "grad_norm": 0.5025933980941772,
      "learning_rate": 7.790551773352344e-06,
      "loss": 0.0019,
      "step": 69740
    },
    {
      "epoch": 1.1412910087539885,
      "grad_norm": 0.09026019275188446,
      "learning_rate": 7.789761834570661e-06,
      "loss": 0.003,
      "step": 69750
    },
    {
      "epoch": 1.1414546347050643,
      "grad_norm": 0.31321465969085693,
      "learning_rate": 7.788971794666816e-06,
      "loss": 0.0024,
      "step": 69760
    },
    {
      "epoch": 1.14161826065614,
      "grad_norm": 0.475144624710083,
      "learning_rate": 7.788181653669448e-06,
      "loss": 0.0022,
      "step": 69770
    },
    {
      "epoch": 1.141781886607216,
      "grad_norm": 0.1627066433429718,
      "learning_rate": 7.787391411607195e-06,
      "loss": 0.0021,
      "step": 69780
    },
    {
      "epoch": 1.1419455125582918,
      "grad_norm": 0.10785926878452301,
      "learning_rate": 7.786601068508703e-06,
      "loss": 0.0025,
      "step": 69790
    },
    {
      "epoch": 1.1421091385093676,
      "grad_norm": 0.2096908539533615,
      "learning_rate": 7.78581062440262e-06,
      "loss": 0.0018,
      "step": 69800
    },
    {
      "epoch": 1.1422727644604433,
      "grad_norm": 0.1953137367963791,
      "learning_rate": 7.785020079317597e-06,
      "loss": 0.0028,
      "step": 69810
    },
    {
      "epoch": 1.1424363904115193,
      "grad_norm": 0.10830351710319519,
      "learning_rate": 7.784229433282291e-06,
      "loss": 0.0021,
      "step": 69820
    },
    {
      "epoch": 1.1426000163625951,
      "grad_norm": 0.3584763705730438,
      "learning_rate": 7.783438686325359e-06,
      "loss": 0.0024,
      "step": 69830
    },
    {
      "epoch": 1.142763642313671,
      "grad_norm": 0.18040603399276733,
      "learning_rate": 7.782647838475464e-06,
      "loss": 0.0019,
      "step": 69840
    },
    {
      "epoch": 1.1429272682647467,
      "grad_norm": 0.11206971853971481,
      "learning_rate": 7.781856889761274e-06,
      "loss": 0.0018,
      "step": 69850
    },
    {
      "epoch": 1.1430908942158227,
      "grad_norm": 0.21511170268058777,
      "learning_rate": 7.781065840211456e-06,
      "loss": 0.0019,
      "step": 69860
    },
    {
      "epoch": 1.1432545201668984,
      "grad_norm": 0.08971478790044785,
      "learning_rate": 7.780274689854687e-06,
      "loss": 0.0027,
      "step": 69870
    },
    {
      "epoch": 1.1434181461179742,
      "grad_norm": 0.11468818038702011,
      "learning_rate": 7.779483438719642e-06,
      "loss": 0.002,
      "step": 69880
    },
    {
      "epoch": 1.1435817720690502,
      "grad_norm": 0.09300553798675537,
      "learning_rate": 7.778692086835003e-06,
      "loss": 0.0013,
      "step": 69890
    },
    {
      "epoch": 1.143745398020126,
      "grad_norm": 0.005916991271078587,
      "learning_rate": 7.777900634229452e-06,
      "loss": 0.0023,
      "step": 69900
    },
    {
      "epoch": 1.1439090239712018,
      "grad_norm": 0.20844735205173492,
      "learning_rate": 7.777109080931682e-06,
      "loss": 0.0037,
      "step": 69910
    },
    {
      "epoch": 1.1440726499222778,
      "grad_norm": 0.17260238528251648,
      "learning_rate": 7.776317426970383e-06,
      "loss": 0.0023,
      "step": 69920
    },
    {
      "epoch": 1.1442362758733535,
      "grad_norm": 0.11354384571313858,
      "learning_rate": 7.775525672374247e-06,
      "loss": 0.0043,
      "step": 69930
    },
    {
      "epoch": 1.1443999018244293,
      "grad_norm": 0.04650479555130005,
      "learning_rate": 7.774733817171979e-06,
      "loss": 0.0013,
      "step": 69940
    },
    {
      "epoch": 1.1445635277755053,
      "grad_norm": 0.38329213857650757,
      "learning_rate": 7.773941861392277e-06,
      "loss": 0.0021,
      "step": 69950
    },
    {
      "epoch": 1.144727153726581,
      "grad_norm": 0.14534588158130646,
      "learning_rate": 7.773149805063851e-06,
      "loss": 0.0022,
      "step": 69960
    },
    {
      "epoch": 1.1448907796776568,
      "grad_norm": 0.22051934897899628,
      "learning_rate": 7.77235764821541e-06,
      "loss": 0.0035,
      "step": 69970
    },
    {
      "epoch": 1.1450544056287328,
      "grad_norm": 0.14047622680664062,
      "learning_rate": 7.771565390875665e-06,
      "loss": 0.002,
      "step": 69980
    },
    {
      "epoch": 1.1452180315798086,
      "grad_norm": 0.1916721612215042,
      "learning_rate": 7.77077303307334e-06,
      "loss": 0.0024,
      "step": 69990
    },
    {
      "epoch": 1.1453816575308844,
      "grad_norm": 0.08109626919031143,
      "learning_rate": 7.769980574837148e-06,
      "loss": 0.0013,
      "step": 70000
    },
    {
      "epoch": 1.1455452834819602,
      "grad_norm": 0.31952276825904846,
      "learning_rate": 7.76918801619582e-06,
      "loss": 0.0025,
      "step": 70010
    },
    {
      "epoch": 1.1457089094330362,
      "grad_norm": 0.06412523239850998,
      "learning_rate": 7.76839535717808e-06,
      "loss": 0.0016,
      "step": 70020
    },
    {
      "epoch": 1.145872535384112,
      "grad_norm": 0.05007540434598923,
      "learning_rate": 7.767602597812663e-06,
      "loss": 0.0019,
      "step": 70030
    },
    {
      "epoch": 1.1460361613351877,
      "grad_norm": 0.17421574890613556,
      "learning_rate": 7.766809738128303e-06,
      "loss": 0.0019,
      "step": 70040
    },
    {
      "epoch": 1.1461997872862635,
      "grad_norm": 0.012598136439919472,
      "learning_rate": 7.766016778153741e-06,
      "loss": 0.0017,
      "step": 70050
    },
    {
      "epoch": 1.1463634132373395,
      "grad_norm": 0.08941764384508133,
      "learning_rate": 7.765223717917719e-06,
      "loss": 0.0021,
      "step": 70060
    },
    {
      "epoch": 1.1465270391884153,
      "grad_norm": 0.1546737402677536,
      "learning_rate": 7.764430557448981e-06,
      "loss": 0.0026,
      "step": 70070
    },
    {
      "epoch": 1.146690665139491,
      "grad_norm": 0.2542679011821747,
      "learning_rate": 7.763637296776281e-06,
      "loss": 0.0045,
      "step": 70080
    },
    {
      "epoch": 1.146854291090567,
      "grad_norm": 0.03860015422105789,
      "learning_rate": 7.762843935928371e-06,
      "loss": 0.0031,
      "step": 70090
    },
    {
      "epoch": 1.1470179170416428,
      "grad_norm": 0.14806663990020752,
      "learning_rate": 7.762050474934009e-06,
      "loss": 0.0026,
      "step": 70100
    },
    {
      "epoch": 1.1471815429927186,
      "grad_norm": 0.613617479801178,
      "learning_rate": 7.761256913821955e-06,
      "loss": 0.0027,
      "step": 70110
    },
    {
      "epoch": 1.1473451689437946,
      "grad_norm": 0.11001269519329071,
      "learning_rate": 7.760463252620975e-06,
      "loss": 0.0029,
      "step": 70120
    },
    {
      "epoch": 1.1475087948948703,
      "grad_norm": 0.14136607944965363,
      "learning_rate": 7.759669491359837e-06,
      "loss": 0.0017,
      "step": 70130
    },
    {
      "epoch": 1.1476724208459461,
      "grad_norm": 0.2886188328266144,
      "learning_rate": 7.758875630067312e-06,
      "loss": 0.0027,
      "step": 70140
    },
    {
      "epoch": 1.1478360467970221,
      "grad_norm": 0.1937817633152008,
      "learning_rate": 7.758081668772176e-06,
      "loss": 0.0015,
      "step": 70150
    },
    {
      "epoch": 1.1479996727480979,
      "grad_norm": 0.116033636033535,
      "learning_rate": 7.75728760750321e-06,
      "loss": 0.0024,
      "step": 70160
    },
    {
      "epoch": 1.1481632986991737,
      "grad_norm": 0.14908021688461304,
      "learning_rate": 7.756493446289193e-06,
      "loss": 0.002,
      "step": 70170
    },
    {
      "epoch": 1.1483269246502494,
      "grad_norm": 0.13305851817131042,
      "learning_rate": 7.755699185158914e-06,
      "loss": 0.002,
      "step": 70180
    },
    {
      "epoch": 1.1484905506013254,
      "grad_norm": 0.2889268398284912,
      "learning_rate": 7.754904824141162e-06,
      "loss": 0.0026,
      "step": 70190
    },
    {
      "epoch": 1.1486541765524012,
      "grad_norm": 0.010579154826700687,
      "learning_rate": 7.754110363264735e-06,
      "loss": 0.0036,
      "step": 70200
    },
    {
      "epoch": 1.148817802503477,
      "grad_norm": 0.5627008676528931,
      "learning_rate": 7.753315802558424e-06,
      "loss": 0.0034,
      "step": 70210
    },
    {
      "epoch": 1.148981428454553,
      "grad_norm": 0.17042863368988037,
      "learning_rate": 7.752521142051033e-06,
      "loss": 0.0051,
      "step": 70220
    },
    {
      "epoch": 1.1491450544056288,
      "grad_norm": 0.04580502584576607,
      "learning_rate": 7.751726381771365e-06,
      "loss": 0.0028,
      "step": 70230
    },
    {
      "epoch": 1.1493086803567045,
      "grad_norm": 0.033817559480667114,
      "learning_rate": 7.75093152174823e-06,
      "loss": 0.0019,
      "step": 70240
    },
    {
      "epoch": 1.1494723063077803,
      "grad_norm": 0.19072742760181427,
      "learning_rate": 7.750136562010439e-06,
      "loss": 0.0028,
      "step": 70250
    },
    {
      "epoch": 1.1496359322588563,
      "grad_norm": 0.08436397463083267,
      "learning_rate": 7.749341502586807e-06,
      "loss": 0.0024,
      "step": 70260
    },
    {
      "epoch": 1.149799558209932,
      "grad_norm": 0.08733312040567398,
      "learning_rate": 7.748546343506156e-06,
      "loss": 0.0014,
      "step": 70270
    },
    {
      "epoch": 1.1499631841610078,
      "grad_norm": 0.23128721117973328,
      "learning_rate": 7.747751084797303e-06,
      "loss": 0.0021,
      "step": 70280
    },
    {
      "epoch": 1.1501268101120838,
      "grad_norm": 0.12886542081832886,
      "learning_rate": 7.746955726489079e-06,
      "loss": 0.0024,
      "step": 70290
    },
    {
      "epoch": 1.1502904360631596,
      "grad_norm": 0.08301883190870285,
      "learning_rate": 7.74616026861031e-06,
      "loss": 0.003,
      "step": 70300
    },
    {
      "epoch": 1.1504540620142354,
      "grad_norm": 0.07706377655267715,
      "learning_rate": 7.745364711189834e-06,
      "loss": 0.0019,
      "step": 70310
    },
    {
      "epoch": 1.1506176879653114,
      "grad_norm": 0.08896639198064804,
      "learning_rate": 7.744569054256483e-06,
      "loss": 0.002,
      "step": 70320
    },
    {
      "epoch": 1.1507813139163872,
      "grad_norm": 0.259365439414978,
      "learning_rate": 7.743773297839102e-06,
      "loss": 0.0038,
      "step": 70330
    },
    {
      "epoch": 1.150944939867463,
      "grad_norm": 0.050292663276195526,
      "learning_rate": 7.742977441966532e-06,
      "loss": 0.0027,
      "step": 70340
    },
    {
      "epoch": 1.151108565818539,
      "grad_norm": 0.09711932390928268,
      "learning_rate": 7.742181486667623e-06,
      "loss": 0.0033,
      "step": 70350
    },
    {
      "epoch": 1.1512721917696147,
      "grad_norm": 0.06031165271997452,
      "learning_rate": 7.741385431971227e-06,
      "loss": 0.0023,
      "step": 70360
    },
    {
      "epoch": 1.1514358177206905,
      "grad_norm": 0.21656079590320587,
      "learning_rate": 7.740589277906197e-06,
      "loss": 0.0021,
      "step": 70370
    },
    {
      "epoch": 1.1515994436717663,
      "grad_norm": 0.15291477739810944,
      "learning_rate": 7.739793024501393e-06,
      "loss": 0.0022,
      "step": 70380
    },
    {
      "epoch": 1.1517630696228423,
      "grad_norm": 0.12854357063770294,
      "learning_rate": 7.738996671785675e-06,
      "loss": 0.0014,
      "step": 70390
    },
    {
      "epoch": 1.151926695573918,
      "grad_norm": 0.03711121156811714,
      "learning_rate": 7.738200219787913e-06,
      "loss": 0.0031,
      "step": 70400
    },
    {
      "epoch": 1.1520903215249938,
      "grad_norm": 0.16242754459381104,
      "learning_rate": 7.73740366853697e-06,
      "loss": 0.0033,
      "step": 70410
    },
    {
      "epoch": 1.1522539474760698,
      "grad_norm": 0.12611828744411469,
      "learning_rate": 7.736607018061728e-06,
      "loss": 0.0016,
      "step": 70420
    },
    {
      "epoch": 1.1524175734271456,
      "grad_norm": 0.09919460117816925,
      "learning_rate": 7.735810268391058e-06,
      "loss": 0.0019,
      "step": 70430
    },
    {
      "epoch": 1.1525811993782213,
      "grad_norm": 0.03554411977529526,
      "learning_rate": 7.73501341955384e-06,
      "loss": 0.0022,
      "step": 70440
    },
    {
      "epoch": 1.1527448253292971,
      "grad_norm": 0.05773541331291199,
      "learning_rate": 7.734216471578958e-06,
      "loss": 0.0028,
      "step": 70450
    },
    {
      "epoch": 1.1529084512803731,
      "grad_norm": 0.1583910584449768,
      "learning_rate": 7.733419424495301e-06,
      "loss": 0.0023,
      "step": 70460
    },
    {
      "epoch": 1.1530720772314489,
      "grad_norm": 0.057962872087955475,
      "learning_rate": 7.73262227833176e-06,
      "loss": 0.0026,
      "step": 70470
    },
    {
      "epoch": 1.1532357031825247,
      "grad_norm": 0.030212948098778725,
      "learning_rate": 7.731825033117231e-06,
      "loss": 0.0016,
      "step": 70480
    },
    {
      "epoch": 1.1533993291336007,
      "grad_norm": 0.18173915147781372,
      "learning_rate": 7.731027688880608e-06,
      "loss": 0.0013,
      "step": 70490
    },
    {
      "epoch": 1.1535629550846764,
      "grad_norm": 0.039716802537441254,
      "learning_rate": 7.730230245650795e-06,
      "loss": 0.002,
      "step": 70500
    },
    {
      "epoch": 1.1537265810357522,
      "grad_norm": 0.03940742090344429,
      "learning_rate": 7.729432703456697e-06,
      "loss": 0.0023,
      "step": 70510
    },
    {
      "epoch": 1.1538902069868282,
      "grad_norm": 0.2081855982542038,
      "learning_rate": 7.728635062327225e-06,
      "loss": 0.0017,
      "step": 70520
    },
    {
      "epoch": 1.154053832937904,
      "grad_norm": 0.1273934543132782,
      "learning_rate": 7.72783732229129e-06,
      "loss": 0.0019,
      "step": 70530
    },
    {
      "epoch": 1.1542174588889798,
      "grad_norm": 0.13988907635211945,
      "learning_rate": 7.727039483377807e-06,
      "loss": 0.0024,
      "step": 70540
    },
    {
      "epoch": 1.1543810848400557,
      "grad_norm": 0.07988998293876648,
      "learning_rate": 7.726241545615698e-06,
      "loss": 0.002,
      "step": 70550
    },
    {
      "epoch": 1.1545447107911315,
      "grad_norm": 0.2179417610168457,
      "learning_rate": 7.725443509033885e-06,
      "loss": 0.0018,
      "step": 70560
    },
    {
      "epoch": 1.1547083367422073,
      "grad_norm": 0.12431062012910843,
      "learning_rate": 7.724645373661295e-06,
      "loss": 0.0017,
      "step": 70570
    },
    {
      "epoch": 1.154871962693283,
      "grad_norm": 0.035822976380586624,
      "learning_rate": 7.72384713952686e-06,
      "loss": 0.0024,
      "step": 70580
    },
    {
      "epoch": 1.155035588644359,
      "grad_norm": 0.04636373370885849,
      "learning_rate": 7.723048806659512e-06,
      "loss": 0.002,
      "step": 70590
    },
    {
      "epoch": 1.1551992145954348,
      "grad_norm": 0.1291734278202057,
      "learning_rate": 7.722250375088189e-06,
      "loss": 0.0028,
      "step": 70600
    },
    {
      "epoch": 1.1553628405465106,
      "grad_norm": 0.13300809264183044,
      "learning_rate": 7.721451844841834e-06,
      "loss": 0.0032,
      "step": 70610
    },
    {
      "epoch": 1.1555264664975866,
      "grad_norm": 0.33549442887306213,
      "learning_rate": 7.72065321594939e-06,
      "loss": 0.0031,
      "step": 70620
    },
    {
      "epoch": 1.1556900924486624,
      "grad_norm": 0.2980208992958069,
      "learning_rate": 7.719854488439804e-06,
      "loss": 0.0026,
      "step": 70630
    },
    {
      "epoch": 1.1558537183997382,
      "grad_norm": 0.07061202079057693,
      "learning_rate": 7.719055662342032e-06,
      "loss": 0.001,
      "step": 70640
    },
    {
      "epoch": 1.156017344350814,
      "grad_norm": 0.036851078271865845,
      "learning_rate": 7.718256737685026e-06,
      "loss": 0.0022,
      "step": 70650
    },
    {
      "epoch": 1.15618097030189,
      "grad_norm": 0.12601856887340546,
      "learning_rate": 7.717457714497745e-06,
      "loss": 0.0032,
      "step": 70660
    },
    {
      "epoch": 1.1563445962529657,
      "grad_norm": 0.13237760961055756,
      "learning_rate": 7.716658592809156e-06,
      "loss": 0.0014,
      "step": 70670
    },
    {
      "epoch": 1.1565082222040415,
      "grad_norm": 0.18847492337226868,
      "learning_rate": 7.715859372648223e-06,
      "loss": 0.0018,
      "step": 70680
    },
    {
      "epoch": 1.1566718481551175,
      "grad_norm": 0.0651775524020195,
      "learning_rate": 7.715060054043913e-06,
      "loss": 0.0025,
      "step": 70690
    },
    {
      "epoch": 1.1568354741061933,
      "grad_norm": 0.44074270129203796,
      "learning_rate": 7.714260637025201e-06,
      "loss": 0.0033,
      "step": 70700
    },
    {
      "epoch": 1.156999100057269,
      "grad_norm": 0.19722644984722137,
      "learning_rate": 7.713461121621065e-06,
      "loss": 0.0048,
      "step": 70710
    },
    {
      "epoch": 1.157162726008345,
      "grad_norm": 0.019424978643655777,
      "learning_rate": 7.712661507860486e-06,
      "loss": 0.0016,
      "step": 70720
    },
    {
      "epoch": 1.1573263519594208,
      "grad_norm": 0.33553346991539,
      "learning_rate": 7.711861795772446e-06,
      "loss": 0.0026,
      "step": 70730
    },
    {
      "epoch": 1.1574899779104966,
      "grad_norm": 0.027482450008392334,
      "learning_rate": 7.711061985385935e-06,
      "loss": 0.0035,
      "step": 70740
    },
    {
      "epoch": 1.1576536038615726,
      "grad_norm": 0.07315513491630554,
      "learning_rate": 7.71026207672994e-06,
      "loss": 0.0021,
      "step": 70750
    },
    {
      "epoch": 1.1578172298126483,
      "grad_norm": 0.11577493697404861,
      "learning_rate": 7.709462069833461e-06,
      "loss": 0.0048,
      "step": 70760
    },
    {
      "epoch": 1.1579808557637241,
      "grad_norm": 0.025808915495872498,
      "learning_rate": 7.708661964725495e-06,
      "loss": 0.0025,
      "step": 70770
    },
    {
      "epoch": 1.1581444817148,
      "grad_norm": 0.2006557583808899,
      "learning_rate": 7.707861761435041e-06,
      "loss": 0.0031,
      "step": 70780
    },
    {
      "epoch": 1.1583081076658759,
      "grad_norm": 0.17072750627994537,
      "learning_rate": 7.707061459991106e-06,
      "loss": 0.0038,
      "step": 70790
    },
    {
      "epoch": 1.1584717336169517,
      "grad_norm": 0.21064308285713196,
      "learning_rate": 7.7062610604227e-06,
      "loss": 0.0021,
      "step": 70800
    },
    {
      "epoch": 1.1586353595680274,
      "grad_norm": 0.23670190572738647,
      "learning_rate": 7.705460562758836e-06,
      "loss": 0.0026,
      "step": 70810
    },
    {
      "epoch": 1.1587989855191032,
      "grad_norm": 0.12288448214530945,
      "learning_rate": 7.704659967028527e-06,
      "loss": 0.0015,
      "step": 70820
    },
    {
      "epoch": 1.1589626114701792,
      "grad_norm": 0.07719285041093826,
      "learning_rate": 7.703859273260796e-06,
      "loss": 0.0014,
      "step": 70830
    },
    {
      "epoch": 1.159126237421255,
      "grad_norm": 0.14321456849575043,
      "learning_rate": 7.703058481484665e-06,
      "loss": 0.0015,
      "step": 70840
    },
    {
      "epoch": 1.1592898633723308,
      "grad_norm": 0.09547753632068634,
      "learning_rate": 7.702257591729161e-06,
      "loss": 0.0018,
      "step": 70850
    },
    {
      "epoch": 1.1594534893234067,
      "grad_norm": 0.14161425828933716,
      "learning_rate": 7.701456604023312e-06,
      "loss": 0.0014,
      "step": 70860
    },
    {
      "epoch": 1.1596171152744825,
      "grad_norm": 0.15306857228279114,
      "learning_rate": 7.700655518396157e-06,
      "loss": 0.0029,
      "step": 70870
    },
    {
      "epoch": 1.1597807412255583,
      "grad_norm": 0.08018869906663895,
      "learning_rate": 7.699854334876727e-06,
      "loss": 0.0024,
      "step": 70880
    },
    {
      "epoch": 1.1599443671766343,
      "grad_norm": 0.020031031221151352,
      "learning_rate": 7.699053053494068e-06,
      "loss": 0.0024,
      "step": 70890
    },
    {
      "epoch": 1.16010799312771,
      "grad_norm": 0.07199593633413315,
      "learning_rate": 7.698251674277221e-06,
      "loss": 0.0032,
      "step": 70900
    },
    {
      "epoch": 1.1602716190787858,
      "grad_norm": 0.04047798365354538,
      "learning_rate": 7.697450197255239e-06,
      "loss": 0.0013,
      "step": 70910
    },
    {
      "epoch": 1.1604352450298618,
      "grad_norm": 0.357229083776474,
      "learning_rate": 7.69664862245717e-06,
      "loss": 0.0021,
      "step": 70920
    },
    {
      "epoch": 1.1605988709809376,
      "grad_norm": 0.05563312768936157,
      "learning_rate": 7.695846949912067e-06,
      "loss": 0.0038,
      "step": 70930
    },
    {
      "epoch": 1.1607624969320134,
      "grad_norm": 0.12651728093624115,
      "learning_rate": 7.695045179648995e-06,
      "loss": 0.0034,
      "step": 70940
    },
    {
      "epoch": 1.1609261228830892,
      "grad_norm": 0.13400737941265106,
      "learning_rate": 7.694243311697009e-06,
      "loss": 0.0019,
      "step": 70950
    },
    {
      "epoch": 1.1610897488341652,
      "grad_norm": 0.057702917605638504,
      "learning_rate": 7.693441346085179e-06,
      "loss": 0.002,
      "step": 70960
    },
    {
      "epoch": 1.161253374785241,
      "grad_norm": 0.1612233966588974,
      "learning_rate": 7.692639282842575e-06,
      "loss": 0.0027,
      "step": 70970
    },
    {
      "epoch": 1.1614170007363167,
      "grad_norm": 0.1036149188876152,
      "learning_rate": 7.691837121998268e-06,
      "loss": 0.0022,
      "step": 70980
    },
    {
      "epoch": 1.1615806266873927,
      "grad_norm": 0.04147518053650856,
      "learning_rate": 7.691034863581336e-06,
      "loss": 0.0016,
      "step": 70990
    },
    {
      "epoch": 1.1617442526384685,
      "grad_norm": 0.18594759702682495,
      "learning_rate": 7.690232507620859e-06,
      "loss": 0.0016,
      "step": 71000
    },
    {
      "epoch": 1.1619078785895443,
      "grad_norm": 0.04953709617257118,
      "learning_rate": 7.689430054145917e-06,
      "loss": 0.0019,
      "step": 71010
    },
    {
      "epoch": 1.16207150454062,
      "grad_norm": 0.06181958690285683,
      "learning_rate": 7.6886275031856e-06,
      "loss": 0.0022,
      "step": 71020
    },
    {
      "epoch": 1.162235130491696,
      "grad_norm": 0.10076411068439484,
      "learning_rate": 7.687824854768998e-06,
      "loss": 0.002,
      "step": 71030
    },
    {
      "epoch": 1.1623987564427718,
      "grad_norm": 0.13595019280910492,
      "learning_rate": 7.687022108925204e-06,
      "loss": 0.0013,
      "step": 71040
    },
    {
      "epoch": 1.1625623823938476,
      "grad_norm": 0.1439180225133896,
      "learning_rate": 7.686219265683319e-06,
      "loss": 0.0018,
      "step": 71050
    },
    {
      "epoch": 1.1627260083449236,
      "grad_norm": 0.09472621232271194,
      "learning_rate": 7.685416325072439e-06,
      "loss": 0.0013,
      "step": 71060
    },
    {
      "epoch": 1.1628896342959993,
      "grad_norm": 0.2254975587129593,
      "learning_rate": 7.684613287121673e-06,
      "loss": 0.0024,
      "step": 71070
    },
    {
      "epoch": 1.1630532602470751,
      "grad_norm": 0.07248338311910629,
      "learning_rate": 7.683810151860129e-06,
      "loss": 0.0026,
      "step": 71080
    },
    {
      "epoch": 1.1632168861981511,
      "grad_norm": 0.06300617754459381,
      "learning_rate": 7.683006919316915e-06,
      "loss": 0.0024,
      "step": 71090
    },
    {
      "epoch": 1.1633805121492269,
      "grad_norm": 0.19227281212806702,
      "learning_rate": 7.682203589521149e-06,
      "loss": 0.0027,
      "step": 71100
    },
    {
      "epoch": 1.1635441381003027,
      "grad_norm": 0.08374682813882828,
      "learning_rate": 7.68140016250195e-06,
      "loss": 0.0028,
      "step": 71110
    },
    {
      "epoch": 1.1637077640513787,
      "grad_norm": 0.27543559670448303,
      "learning_rate": 7.68059663828844e-06,
      "loss": 0.004,
      "step": 71120
    },
    {
      "epoch": 1.1638713900024544,
      "grad_norm": 0.06532328575849533,
      "learning_rate": 7.679793016909745e-06,
      "loss": 0.0036,
      "step": 71130
    },
    {
      "epoch": 1.1640350159535302,
      "grad_norm": 0.15541760623455048,
      "learning_rate": 7.678989298394991e-06,
      "loss": 0.0018,
      "step": 71140
    },
    {
      "epoch": 1.164198641904606,
      "grad_norm": 0.08065922558307648,
      "learning_rate": 7.678185482773317e-06,
      "loss": 0.0021,
      "step": 71150
    },
    {
      "epoch": 1.164362267855682,
      "grad_norm": 0.11972736567258835,
      "learning_rate": 7.677381570073855e-06,
      "loss": 0.0037,
      "step": 71160
    },
    {
      "epoch": 1.1645258938067578,
      "grad_norm": 0.13993096351623535,
      "learning_rate": 7.676577560325745e-06,
      "loss": 0.0015,
      "step": 71170
    },
    {
      "epoch": 1.1646895197578335,
      "grad_norm": 0.19419540464878082,
      "learning_rate": 7.675773453558133e-06,
      "loss": 0.0017,
      "step": 71180
    },
    {
      "epoch": 1.1648531457089095,
      "grad_norm": 0.09884954988956451,
      "learning_rate": 7.674969249800162e-06,
      "loss": 0.0025,
      "step": 71190
    },
    {
      "epoch": 1.1650167716599853,
      "grad_norm": 0.11203473061323166,
      "learning_rate": 7.674164949080986e-06,
      "loss": 0.0031,
      "step": 71200
    },
    {
      "epoch": 1.165180397611061,
      "grad_norm": 0.1971210390329361,
      "learning_rate": 7.673360551429759e-06,
      "loss": 0.002,
      "step": 71210
    },
    {
      "epoch": 1.1653440235621368,
      "grad_norm": 0.050767164677381516,
      "learning_rate": 7.672556056875638e-06,
      "loss": 0.0015,
      "step": 71220
    },
    {
      "epoch": 1.1655076495132128,
      "grad_norm": 0.6226209402084351,
      "learning_rate": 7.67175146544778e-06,
      "loss": 0.0043,
      "step": 71230
    },
    {
      "epoch": 1.1656712754642886,
      "grad_norm": 0.058105431497097015,
      "learning_rate": 7.670946777175354e-06,
      "loss": 0.0015,
      "step": 71240
    },
    {
      "epoch": 1.1658349014153644,
      "grad_norm": 0.15004487335681915,
      "learning_rate": 7.670141992087528e-06,
      "loss": 0.0024,
      "step": 71250
    },
    {
      "epoch": 1.1659985273664404,
      "grad_norm": 0.0597807839512825,
      "learning_rate": 7.66933711021347e-06,
      "loss": 0.0023,
      "step": 71260
    },
    {
      "epoch": 1.1661621533175162,
      "grad_norm": 0.21553616225719452,
      "learning_rate": 7.66853213158236e-06,
      "loss": 0.0029,
      "step": 71270
    },
    {
      "epoch": 1.166325779268592,
      "grad_norm": 0.061164747923612595,
      "learning_rate": 7.667727056223374e-06,
      "loss": 0.001,
      "step": 71280
    },
    {
      "epoch": 1.166489405219668,
      "grad_norm": 0.08249186724424362,
      "learning_rate": 7.666921884165693e-06,
      "loss": 0.0016,
      "step": 71290
    },
    {
      "epoch": 1.1666530311707437,
      "grad_norm": 0.05865321308374405,
      "learning_rate": 7.666116615438501e-06,
      "loss": 0.0011,
      "step": 71300
    },
    {
      "epoch": 1.1668166571218195,
      "grad_norm": 0.19730877876281738,
      "learning_rate": 7.665311250070992e-06,
      "loss": 0.0024,
      "step": 71310
    },
    {
      "epoch": 1.1669802830728955,
      "grad_norm": 0.11595425754785538,
      "learning_rate": 7.664505788092356e-06,
      "loss": 0.0016,
      "step": 71320
    },
    {
      "epoch": 1.1671439090239712,
      "grad_norm": 0.12209577113389969,
      "learning_rate": 7.66370022953179e-06,
      "loss": 0.0026,
      "step": 71330
    },
    {
      "epoch": 1.167307534975047,
      "grad_norm": 0.2315395176410675,
      "learning_rate": 7.662894574418491e-06,
      "loss": 0.0019,
      "step": 71340
    },
    {
      "epoch": 1.1674711609261228,
      "grad_norm": 0.11272265017032623,
      "learning_rate": 7.662088822781663e-06,
      "loss": 0.0016,
      "step": 71350
    },
    {
      "epoch": 1.1676347868771988,
      "grad_norm": 0.3565988838672638,
      "learning_rate": 7.661282974650515e-06,
      "loss": 0.0037,
      "step": 71360
    },
    {
      "epoch": 1.1677984128282746,
      "grad_norm": 0.22530940175056458,
      "learning_rate": 7.660477030054256e-06,
      "loss": 0.0023,
      "step": 71370
    },
    {
      "epoch": 1.1679620387793503,
      "grad_norm": 0.13648036122322083,
      "learning_rate": 7.659670989022098e-06,
      "loss": 0.0021,
      "step": 71380
    },
    {
      "epoch": 1.1681256647304263,
      "grad_norm": 0.03399841487407684,
      "learning_rate": 7.65886485158326e-06,
      "loss": 0.001,
      "step": 71390
    },
    {
      "epoch": 1.1682892906815021,
      "grad_norm": 0.05450304225087166,
      "learning_rate": 7.658058617766958e-06,
      "loss": 0.0018,
      "step": 71400
    },
    {
      "epoch": 1.1684529166325779,
      "grad_norm": 0.16956022381782532,
      "learning_rate": 7.657252287602423e-06,
      "loss": 0.0023,
      "step": 71410
    },
    {
      "epoch": 1.1686165425836537,
      "grad_norm": 0.06380729377269745,
      "learning_rate": 7.656445861118878e-06,
      "loss": 0.0035,
      "step": 71420
    },
    {
      "epoch": 1.1687801685347297,
      "grad_norm": 0.07962498068809509,
      "learning_rate": 7.655639338345557e-06,
      "loss": 0.0017,
      "step": 71430
    },
    {
      "epoch": 1.1689437944858054,
      "grad_norm": 0.13164030015468597,
      "learning_rate": 7.654832719311691e-06,
      "loss": 0.0034,
      "step": 71440
    },
    {
      "epoch": 1.1691074204368812,
      "grad_norm": 0.05280362069606781,
      "learning_rate": 7.654026004046518e-06,
      "loss": 0.0013,
      "step": 71450
    },
    {
      "epoch": 1.1692710463879572,
      "grad_norm": 0.36253800988197327,
      "learning_rate": 7.653219192579283e-06,
      "loss": 0.0028,
      "step": 71460
    },
    {
      "epoch": 1.169434672339033,
      "grad_norm": 0.07338973134756088,
      "learning_rate": 7.65241228493923e-06,
      "loss": 0.0035,
      "step": 71470
    },
    {
      "epoch": 1.1695982982901088,
      "grad_norm": 0.0691860020160675,
      "learning_rate": 7.651605281155606e-06,
      "loss": 0.0014,
      "step": 71480
    },
    {
      "epoch": 1.1697619242411847,
      "grad_norm": 0.1511940062046051,
      "learning_rate": 7.650798181257662e-06,
      "loss": 0.0023,
      "step": 71490
    },
    {
      "epoch": 1.1699255501922605,
      "grad_norm": 0.39839473366737366,
      "learning_rate": 7.649990985274657e-06,
      "loss": 0.0022,
      "step": 71500
    },
    {
      "epoch": 1.1700891761433363,
      "grad_norm": 0.23905257880687714,
      "learning_rate": 7.649183693235847e-06,
      "loss": 0.0023,
      "step": 71510
    },
    {
      "epoch": 1.1702528020944123,
      "grad_norm": 0.23282641172409058,
      "learning_rate": 7.648376305170494e-06,
      "loss": 0.0058,
      "step": 71520
    },
    {
      "epoch": 1.170416428045488,
      "grad_norm": 0.24495255947113037,
      "learning_rate": 7.647568821107868e-06,
      "loss": 0.0022,
      "step": 71530
    },
    {
      "epoch": 1.1705800539965638,
      "grad_norm": 0.2635161280632019,
      "learning_rate": 7.646761241077233e-06,
      "loss": 0.0031,
      "step": 71540
    },
    {
      "epoch": 1.1707436799476396,
      "grad_norm": 0.10612886399030685,
      "learning_rate": 7.645953565107864e-06,
      "loss": 0.0021,
      "step": 71550
    },
    {
      "epoch": 1.1709073058987156,
      "grad_norm": 0.13540451228618622,
      "learning_rate": 7.645145793229039e-06,
      "loss": 0.0019,
      "step": 71560
    },
    {
      "epoch": 1.1710709318497914,
      "grad_norm": 0.07193749397993088,
      "learning_rate": 7.644337925470036e-06,
      "loss": 0.0024,
      "step": 71570
    },
    {
      "epoch": 1.1712345578008672,
      "grad_norm": 0.03849054127931595,
      "learning_rate": 7.643529961860139e-06,
      "loss": 0.0027,
      "step": 71580
    },
    {
      "epoch": 1.171398183751943,
      "grad_norm": 0.041433919221162796,
      "learning_rate": 7.642721902428634e-06,
      "loss": 0.0037,
      "step": 71590
    },
    {
      "epoch": 1.171561809703019,
      "grad_norm": 0.1281454712152481,
      "learning_rate": 7.641913747204811e-06,
      "loss": 0.0018,
      "step": 71600
    },
    {
      "epoch": 1.1717254356540947,
      "grad_norm": 0.08238737285137177,
      "learning_rate": 7.641105496217966e-06,
      "loss": 0.0028,
      "step": 71610
    },
    {
      "epoch": 1.1718890616051705,
      "grad_norm": 0.06016228348016739,
      "learning_rate": 7.640297149497396e-06,
      "loss": 0.0017,
      "step": 71620
    },
    {
      "epoch": 1.1720526875562465,
      "grad_norm": 0.5093915462493896,
      "learning_rate": 7.639488707072399e-06,
      "loss": 0.0026,
      "step": 71630
    },
    {
      "epoch": 1.1722163135073222,
      "grad_norm": 0.21576476097106934,
      "learning_rate": 7.638680168972277e-06,
      "loss": 0.0052,
      "step": 71640
    },
    {
      "epoch": 1.172379939458398,
      "grad_norm": 0.12510138750076294,
      "learning_rate": 7.637871535226344e-06,
      "loss": 0.0024,
      "step": 71650
    },
    {
      "epoch": 1.172543565409474,
      "grad_norm": 0.03411099314689636,
      "learning_rate": 7.637062805863906e-06,
      "loss": 0.0016,
      "step": 71660
    },
    {
      "epoch": 1.1727071913605498,
      "grad_norm": 0.25399377942085266,
      "learning_rate": 7.636253980914278e-06,
      "loss": 0.0026,
      "step": 71670
    },
    {
      "epoch": 1.1728708173116256,
      "grad_norm": 0.14685702323913574,
      "learning_rate": 7.635445060406783e-06,
      "loss": 0.0016,
      "step": 71680
    },
    {
      "epoch": 1.1730344432627016,
      "grad_norm": 0.05005675554275513,
      "learning_rate": 7.634636044370738e-06,
      "loss": 0.0017,
      "step": 71690
    },
    {
      "epoch": 1.1731980692137773,
      "grad_norm": 0.2060636579990387,
      "learning_rate": 7.633826932835466e-06,
      "loss": 0.0027,
      "step": 71700
    },
    {
      "epoch": 1.1733616951648531,
      "grad_norm": 0.22579315304756165,
      "learning_rate": 7.633017725830301e-06,
      "loss": 0.0032,
      "step": 71710
    },
    {
      "epoch": 1.173525321115929,
      "grad_norm": 0.014119814150035381,
      "learning_rate": 7.63220842338457e-06,
      "loss": 0.0025,
      "step": 71720
    },
    {
      "epoch": 1.1736889470670049,
      "grad_norm": 0.0524456650018692,
      "learning_rate": 7.63139902552761e-06,
      "loss": 0.0014,
      "step": 71730
    },
    {
      "epoch": 1.1738525730180807,
      "grad_norm": 0.1027493104338646,
      "learning_rate": 7.63058953228876e-06,
      "loss": 0.0027,
      "step": 71740
    },
    {
      "epoch": 1.1740161989691564,
      "grad_norm": 0.18032707273960114,
      "learning_rate": 7.629779943697362e-06,
      "loss": 0.0018,
      "step": 71750
    },
    {
      "epoch": 1.1741798249202324,
      "grad_norm": 0.07175762951374054,
      "learning_rate": 7.628970259782761e-06,
      "loss": 0.0028,
      "step": 71760
    },
    {
      "epoch": 1.1743434508713082,
      "grad_norm": 0.21787424385547638,
      "learning_rate": 7.628160480574307e-06,
      "loss": 0.0025,
      "step": 71770
    },
    {
      "epoch": 1.174507076822384,
      "grad_norm": 0.3073088526725769,
      "learning_rate": 7.627350606101353e-06,
      "loss": 0.0022,
      "step": 71780
    },
    {
      "epoch": 1.1746707027734598,
      "grad_norm": 0.06641089916229248,
      "learning_rate": 7.626540636393254e-06,
      "loss": 0.0025,
      "step": 71790
    },
    {
      "epoch": 1.1748343287245357,
      "grad_norm": 0.1391206681728363,
      "learning_rate": 7.625730571479369e-06,
      "loss": 0.003,
      "step": 71800
    },
    {
      "epoch": 1.1749979546756115,
      "grad_norm": 0.10353788733482361,
      "learning_rate": 7.624920411389061e-06,
      "loss": 0.0037,
      "step": 71810
    },
    {
      "epoch": 1.1751615806266873,
      "grad_norm": 0.07975750416517258,
      "learning_rate": 7.624110156151696e-06,
      "loss": 0.0024,
      "step": 71820
    },
    {
      "epoch": 1.1753252065777633,
      "grad_norm": 0.0043273488990962505,
      "learning_rate": 7.6232998057966476e-06,
      "loss": 0.0049,
      "step": 71830
    },
    {
      "epoch": 1.175488832528839,
      "grad_norm": 0.14998240768909454,
      "learning_rate": 7.6224893603532845e-06,
      "loss": 0.0026,
      "step": 71840
    },
    {
      "epoch": 1.1756524584799148,
      "grad_norm": 0.10849238187074661,
      "learning_rate": 7.621678819850986e-06,
      "loss": 0.0021,
      "step": 71850
    },
    {
      "epoch": 1.1758160844309908,
      "grad_norm": 0.05144664645195007,
      "learning_rate": 7.62086818431913e-06,
      "loss": 0.0022,
      "step": 71860
    },
    {
      "epoch": 1.1759797103820666,
      "grad_norm": 0.09146712720394135,
      "learning_rate": 7.620057453787103e-06,
      "loss": 0.0027,
      "step": 71870
    },
    {
      "epoch": 1.1761433363331424,
      "grad_norm": 0.16517168283462524,
      "learning_rate": 7.619246628284289e-06,
      "loss": 0.003,
      "step": 71880
    },
    {
      "epoch": 1.1763069622842184,
      "grad_norm": 0.2248610556125641,
      "learning_rate": 7.618435707840082e-06,
      "loss": 0.004,
      "step": 71890
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.0688471719622612,
      "learning_rate": 7.617624692483872e-06,
      "loss": 0.0037,
      "step": 71900
    },
    {
      "epoch": 1.17663421418637,
      "grad_norm": 0.0887303575873375,
      "learning_rate": 7.616813582245058e-06,
      "loss": 0.0023,
      "step": 71910
    },
    {
      "epoch": 1.1767978401374457,
      "grad_norm": 0.24531736969947815,
      "learning_rate": 7.616002377153042e-06,
      "loss": 0.0031,
      "step": 71920
    },
    {
      "epoch": 1.1769614660885217,
      "grad_norm": 0.026818662881851196,
      "learning_rate": 7.615191077237227e-06,
      "loss": 0.0022,
      "step": 71930
    },
    {
      "epoch": 1.1771250920395975,
      "grad_norm": 0.008812198415398598,
      "learning_rate": 7.61437968252702e-06,
      "loss": 0.003,
      "step": 71940
    },
    {
      "epoch": 1.1772887179906733,
      "grad_norm": 0.024305429309606552,
      "learning_rate": 7.613568193051834e-06,
      "loss": 0.0015,
      "step": 71950
    },
    {
      "epoch": 1.1774523439417492,
      "grad_norm": 0.08036768436431885,
      "learning_rate": 7.612756608841081e-06,
      "loss": 0.0022,
      "step": 71960
    },
    {
      "epoch": 1.177615969892825,
      "grad_norm": 0.1106012836098671,
      "learning_rate": 7.611944929924182e-06,
      "loss": 0.0028,
      "step": 71970
    },
    {
      "epoch": 1.1777795958439008,
      "grad_norm": 0.0743735283613205,
      "learning_rate": 7.611133156330558e-06,
      "loss": 0.0016,
      "step": 71980
    },
    {
      "epoch": 1.1779432217949766,
      "grad_norm": 0.07157034426927567,
      "learning_rate": 7.610321288089631e-06,
      "loss": 0.0019,
      "step": 71990
    },
    {
      "epoch": 1.1781068477460526,
      "grad_norm": 0.10042784363031387,
      "learning_rate": 7.609509325230831e-06,
      "loss": 0.0023,
      "step": 72000
    },
    {
      "epoch": 1.1782704736971283,
      "grad_norm": 0.12921467423439026,
      "learning_rate": 7.608697267783591e-06,
      "loss": 0.0029,
      "step": 72010
    },
    {
      "epoch": 1.1784340996482041,
      "grad_norm": 0.16908426582813263,
      "learning_rate": 7.607885115777342e-06,
      "loss": 0.0021,
      "step": 72020
    },
    {
      "epoch": 1.17859772559928,
      "grad_norm": 0.10650560259819031,
      "learning_rate": 7.60707286924153e-06,
      "loss": 0.0018,
      "step": 72030
    },
    {
      "epoch": 1.1787613515503559,
      "grad_norm": 0.1354352831840515,
      "learning_rate": 7.606260528205589e-06,
      "loss": 0.0017,
      "step": 72040
    },
    {
      "epoch": 1.1789249775014317,
      "grad_norm": 0.019955743104219437,
      "learning_rate": 7.6054480926989695e-06,
      "loss": 0.002,
      "step": 72050
    },
    {
      "epoch": 1.1790886034525077,
      "grad_norm": 0.10579939931631088,
      "learning_rate": 7.604635562751118e-06,
      "loss": 0.0028,
      "step": 72060
    },
    {
      "epoch": 1.1792522294035834,
      "grad_norm": 0.05810854583978653,
      "learning_rate": 7.6038229383914875e-06,
      "loss": 0.0024,
      "step": 72070
    },
    {
      "epoch": 1.1794158553546592,
      "grad_norm": 0.19434931874275208,
      "learning_rate": 7.6030102196495335e-06,
      "loss": 0.0053,
      "step": 72080
    },
    {
      "epoch": 1.1795794813057352,
      "grad_norm": 0.030727526172995567,
      "learning_rate": 7.602197406554714e-06,
      "loss": 0.0021,
      "step": 72090
    },
    {
      "epoch": 1.179743107256811,
      "grad_norm": 0.2397388070821762,
      "learning_rate": 7.601384499136495e-06,
      "loss": 0.0028,
      "step": 72100
    },
    {
      "epoch": 1.1799067332078867,
      "grad_norm": 0.05750928446650505,
      "learning_rate": 7.60057149742434e-06,
      "loss": 0.0019,
      "step": 72110
    },
    {
      "epoch": 1.1800703591589625,
      "grad_norm": 0.23812896013259888,
      "learning_rate": 7.599758401447717e-06,
      "loss": 0.0031,
      "step": 72120
    },
    {
      "epoch": 1.1802339851100385,
      "grad_norm": 0.1328461468219757,
      "learning_rate": 7.598945211236102e-06,
      "loss": 0.0021,
      "step": 72130
    },
    {
      "epoch": 1.1803976110611143,
      "grad_norm": 0.1501968502998352,
      "learning_rate": 7.598131926818968e-06,
      "loss": 0.0014,
      "step": 72140
    },
    {
      "epoch": 1.18056123701219,
      "grad_norm": 0.03078266978263855,
      "learning_rate": 7.597318548225796e-06,
      "loss": 0.0018,
      "step": 72150
    },
    {
      "epoch": 1.180724862963266,
      "grad_norm": 0.16636280715465546,
      "learning_rate": 7.596505075486069e-06,
      "loss": 0.0041,
      "step": 72160
    },
    {
      "epoch": 1.1808884889143418,
      "grad_norm": 0.09739551693201065,
      "learning_rate": 7.595691508629275e-06,
      "loss": 0.0018,
      "step": 72170
    },
    {
      "epoch": 1.1810521148654176,
      "grad_norm": 0.046920355409383774,
      "learning_rate": 7.594877847684903e-06,
      "loss": 0.002,
      "step": 72180
    },
    {
      "epoch": 1.1812157408164934,
      "grad_norm": 0.3047250509262085,
      "learning_rate": 7.594064092682443e-06,
      "loss": 0.002,
      "step": 72190
    },
    {
      "epoch": 1.1813793667675694,
      "grad_norm": 0.11077884584665298,
      "learning_rate": 7.593250243651395e-06,
      "loss": 0.0027,
      "step": 72200
    },
    {
      "epoch": 1.1815429927186452,
      "grad_norm": 0.05937687307596207,
      "learning_rate": 7.59243630062126e-06,
      "loss": 0.0022,
      "step": 72210
    },
    {
      "epoch": 1.181706618669721,
      "grad_norm": 0.04560388624668121,
      "learning_rate": 7.591622263621539e-06,
      "loss": 0.0021,
      "step": 72220
    },
    {
      "epoch": 1.181870244620797,
      "grad_norm": 0.047350239008665085,
      "learning_rate": 7.590808132681739e-06,
      "loss": 0.002,
      "step": 72230
    },
    {
      "epoch": 1.1820338705718727,
      "grad_norm": 0.07642187923192978,
      "learning_rate": 7.589993907831372e-06,
      "loss": 0.0026,
      "step": 72240
    },
    {
      "epoch": 1.1821974965229485,
      "grad_norm": 0.0686226636171341,
      "learning_rate": 7.5891795890999495e-06,
      "loss": 0.0015,
      "step": 72250
    },
    {
      "epoch": 1.1823611224740245,
      "grad_norm": 0.03936338424682617,
      "learning_rate": 7.588365176516992e-06,
      "loss": 0.002,
      "step": 72260
    },
    {
      "epoch": 1.1825247484251002,
      "grad_norm": 0.1994786560535431,
      "learning_rate": 7.587550670112016e-06,
      "loss": 0.0031,
      "step": 72270
    },
    {
      "epoch": 1.182688374376176,
      "grad_norm": 0.1494719237089157,
      "learning_rate": 7.586736069914549e-06,
      "loss": 0.0021,
      "step": 72280
    },
    {
      "epoch": 1.182852000327252,
      "grad_norm": 0.0997762605547905,
      "learning_rate": 7.585921375954115e-06,
      "loss": 0.0012,
      "step": 72290
    },
    {
      "epoch": 1.1830156262783278,
      "grad_norm": 0.1472926288843155,
      "learning_rate": 7.585106588260245e-06,
      "loss": 0.0021,
      "step": 72300
    },
    {
      "epoch": 1.1831792522294036,
      "grad_norm": 0.06689479947090149,
      "learning_rate": 7.584291706862476e-06,
      "loss": 0.0011,
      "step": 72310
    },
    {
      "epoch": 1.1833428781804793,
      "grad_norm": 0.06242343783378601,
      "learning_rate": 7.583476731790343e-06,
      "loss": 0.0021,
      "step": 72320
    },
    {
      "epoch": 1.1835065041315553,
      "grad_norm": 0.13247735798358917,
      "learning_rate": 7.582661663073389e-06,
      "loss": 0.0036,
      "step": 72330
    },
    {
      "epoch": 1.1836701300826311,
      "grad_norm": 0.1880936622619629,
      "learning_rate": 7.581846500741157e-06,
      "loss": 0.0018,
      "step": 72340
    },
    {
      "epoch": 1.1838337560337069,
      "grad_norm": 0.1525871455669403,
      "learning_rate": 7.581031244823193e-06,
      "loss": 0.0021,
      "step": 72350
    },
    {
      "epoch": 1.1839973819847829,
      "grad_norm": 0.08362074941396713,
      "learning_rate": 7.58021589534905e-06,
      "loss": 0.0016,
      "step": 72360
    },
    {
      "epoch": 1.1841610079358587,
      "grad_norm": 0.03664948791265488,
      "learning_rate": 7.579400452348285e-06,
      "loss": 0.0018,
      "step": 72370
    },
    {
      "epoch": 1.1843246338869344,
      "grad_norm": 0.08488146215677261,
      "learning_rate": 7.578584915850448e-06,
      "loss": 0.0013,
      "step": 72380
    },
    {
      "epoch": 1.1844882598380102,
      "grad_norm": 0.10582694411277771,
      "learning_rate": 7.57776928588511e-06,
      "loss": 0.002,
      "step": 72390
    },
    {
      "epoch": 1.1846518857890862,
      "grad_norm": 0.05558272823691368,
      "learning_rate": 7.576953562481828e-06,
      "loss": 0.002,
      "step": 72400
    },
    {
      "epoch": 1.184815511740162,
      "grad_norm": 0.13190887868404388,
      "learning_rate": 7.576137745670174e-06,
      "loss": 0.0026,
      "step": 72410
    },
    {
      "epoch": 1.1849791376912377,
      "grad_norm": 0.1701160967350006,
      "learning_rate": 7.575321835479719e-06,
      "loss": 0.0024,
      "step": 72420
    },
    {
      "epoch": 1.1851427636423137,
      "grad_norm": 0.12425373494625092,
      "learning_rate": 7.574505831940037e-06,
      "loss": 0.0011,
      "step": 72430
    },
    {
      "epoch": 1.1853063895933895,
      "grad_norm": 0.10949856787919998,
      "learning_rate": 7.573689735080705e-06,
      "loss": 0.0017,
      "step": 72440
    },
    {
      "epoch": 1.1854700155444653,
      "grad_norm": 0.08130192756652832,
      "learning_rate": 7.572873544931305e-06,
      "loss": 0.0035,
      "step": 72450
    },
    {
      "epoch": 1.1856336414955413,
      "grad_norm": 0.13593576848506927,
      "learning_rate": 7.572057261521424e-06,
      "loss": 0.0014,
      "step": 72460
    },
    {
      "epoch": 1.185797267446617,
      "grad_norm": 0.10031068325042725,
      "learning_rate": 7.571240884880649e-06,
      "loss": 0.0013,
      "step": 72470
    },
    {
      "epoch": 1.1859608933976928,
      "grad_norm": 0.16531318426132202,
      "learning_rate": 7.570424415038573e-06,
      "loss": 0.0029,
      "step": 72480
    },
    {
      "epoch": 1.1861245193487688,
      "grad_norm": 0.23652106523513794,
      "learning_rate": 7.56960785202479e-06,
      "loss": 0.0018,
      "step": 72490
    },
    {
      "epoch": 1.1862881452998446,
      "grad_norm": 0.09523310512304306,
      "learning_rate": 7.568791195868896e-06,
      "loss": 0.0015,
      "step": 72500
    },
    {
      "epoch": 1.1864517712509204,
      "grad_norm": 0.2035551369190216,
      "learning_rate": 7.567974446600496e-06,
      "loss": 0.0018,
      "step": 72510
    },
    {
      "epoch": 1.1866153972019962,
      "grad_norm": 0.05382750555872917,
      "learning_rate": 7.567157604249194e-06,
      "loss": 0.0014,
      "step": 72520
    },
    {
      "epoch": 1.1867790231530722,
      "grad_norm": 0.09363462030887604,
      "learning_rate": 7.566340668844598e-06,
      "loss": 0.0018,
      "step": 72530
    },
    {
      "epoch": 1.186942649104148,
      "grad_norm": 0.0806182399392128,
      "learning_rate": 7.5655236404163235e-06,
      "loss": 0.0017,
      "step": 72540
    },
    {
      "epoch": 1.1871062750552237,
      "grad_norm": 0.06348676234483719,
      "learning_rate": 7.56470651899398e-06,
      "loss": 0.0029,
      "step": 72550
    },
    {
      "epoch": 1.1872699010062995,
      "grad_norm": 0.08105562627315521,
      "learning_rate": 7.563889304607193e-06,
      "loss": 0.0022,
      "step": 72560
    },
    {
      "epoch": 1.1874335269573755,
      "grad_norm": 0.1868240237236023,
      "learning_rate": 7.5630719972855784e-06,
      "loss": 0.0017,
      "step": 72570
    },
    {
      "epoch": 1.1875971529084512,
      "grad_norm": 0.029343510046601295,
      "learning_rate": 7.562254597058764e-06,
      "loss": 0.0025,
      "step": 72580
    },
    {
      "epoch": 1.187760778859527,
      "grad_norm": 0.06690794974565506,
      "learning_rate": 7.561437103956379e-06,
      "loss": 0.0032,
      "step": 72590
    },
    {
      "epoch": 1.187924404810603,
      "grad_norm": 0.07641276717185974,
      "learning_rate": 7.560619518008054e-06,
      "loss": 0.0017,
      "step": 72600
    },
    {
      "epoch": 1.1880880307616788,
      "grad_norm": 0.24838106334209442,
      "learning_rate": 7.559801839243428e-06,
      "loss": 0.0031,
      "step": 72610
    },
    {
      "epoch": 1.1882516567127546,
      "grad_norm": 0.14931818842887878,
      "learning_rate": 7.558984067692137e-06,
      "loss": 0.0019,
      "step": 72620
    },
    {
      "epoch": 1.1884152826638306,
      "grad_norm": 0.013912595808506012,
      "learning_rate": 7.558166203383823e-06,
      "loss": 0.0014,
      "step": 72630
    },
    {
      "epoch": 1.1885789086149063,
      "grad_norm": 0.19385413825511932,
      "learning_rate": 7.557348246348133e-06,
      "loss": 0.0025,
      "step": 72640
    },
    {
      "epoch": 1.1887425345659821,
      "grad_norm": 0.17417462170124054,
      "learning_rate": 7.556530196614717e-06,
      "loss": 0.0028,
      "step": 72650
    },
    {
      "epoch": 1.188906160517058,
      "grad_norm": 0.008293598890304565,
      "learning_rate": 7.555712054213222e-06,
      "loss": 0.002,
      "step": 72660
    },
    {
      "epoch": 1.1890697864681339,
      "grad_norm": 0.1257864534854889,
      "learning_rate": 7.55489381917331e-06,
      "loss": 0.0018,
      "step": 72670
    },
    {
      "epoch": 1.1892334124192097,
      "grad_norm": 0.02947196178138256,
      "learning_rate": 7.55407549152464e-06,
      "loss": 0.0019,
      "step": 72680
    },
    {
      "epoch": 1.1893970383702854,
      "grad_norm": 0.12592129409313202,
      "learning_rate": 7.553257071296869e-06,
      "loss": 0.0021,
      "step": 72690
    },
    {
      "epoch": 1.1895606643213614,
      "grad_norm": 0.25786641240119934,
      "learning_rate": 7.552438558519666e-06,
      "loss": 0.0024,
      "step": 72700
    },
    {
      "epoch": 1.1897242902724372,
      "grad_norm": 0.07107283920049667,
      "learning_rate": 7.5516199532227e-06,
      "loss": 0.0025,
      "step": 72710
    },
    {
      "epoch": 1.189887916223513,
      "grad_norm": 0.12497085332870483,
      "learning_rate": 7.550801255435642e-06,
      "loss": 0.0033,
      "step": 72720
    },
    {
      "epoch": 1.190051542174589,
      "grad_norm": 0.22226111590862274,
      "learning_rate": 7.5499824651881726e-06,
      "loss": 0.0025,
      "step": 72730
    },
    {
      "epoch": 1.1902151681256647,
      "grad_norm": 0.1252511888742447,
      "learning_rate": 7.549163582509964e-06,
      "loss": 0.0016,
      "step": 72740
    },
    {
      "epoch": 1.1903787940767405,
      "grad_norm": 0.06868183612823486,
      "learning_rate": 7.548344607430704e-06,
      "loss": 0.0036,
      "step": 72750
    },
    {
      "epoch": 1.1905424200278163,
      "grad_norm": 0.024801086634397507,
      "learning_rate": 7.547525539980077e-06,
      "loss": 0.0015,
      "step": 72760
    },
    {
      "epoch": 1.1907060459788923,
      "grad_norm": 0.19810540974140167,
      "learning_rate": 7.546706380187772e-06,
      "loss": 0.0019,
      "step": 72770
    },
    {
      "epoch": 1.190869671929968,
      "grad_norm": 0.11733773350715637,
      "learning_rate": 7.545887128083482e-06,
      "loss": 0.0026,
      "step": 72780
    },
    {
      "epoch": 1.1910332978810438,
      "grad_norm": 0.25038450956344604,
      "learning_rate": 7.545067783696901e-06,
      "loss": 0.0014,
      "step": 72790
    },
    {
      "epoch": 1.1911969238321198,
      "grad_norm": 0.06976586580276489,
      "learning_rate": 7.54424834705773e-06,
      "loss": 0.0043,
      "step": 72800
    },
    {
      "epoch": 1.1913605497831956,
      "grad_norm": 0.26963064074516296,
      "learning_rate": 7.543428818195673e-06,
      "loss": 0.002,
      "step": 72810
    },
    {
      "epoch": 1.1915241757342714,
      "grad_norm": 0.07638239860534668,
      "learning_rate": 7.542609197140433e-06,
      "loss": 0.0016,
      "step": 72820
    },
    {
      "epoch": 1.1916878016853474,
      "grad_norm": 0.028073985129594803,
      "learning_rate": 7.541789483921721e-06,
      "loss": 0.0028,
      "step": 72830
    },
    {
      "epoch": 1.1918514276364232,
      "grad_norm": 0.11851628869771957,
      "learning_rate": 7.540969678569249e-06,
      "loss": 0.0018,
      "step": 72840
    },
    {
      "epoch": 1.192015053587499,
      "grad_norm": 0.09475129842758179,
      "learning_rate": 7.540149781112733e-06,
      "loss": 0.0018,
      "step": 72850
    },
    {
      "epoch": 1.192178679538575,
      "grad_norm": 0.12802912294864655,
      "learning_rate": 7.539329791581893e-06,
      "loss": 0.0018,
      "step": 72860
    },
    {
      "epoch": 1.1923423054896507,
      "grad_norm": 0.20128066837787628,
      "learning_rate": 7.53850971000645e-06,
      "loss": 0.0019,
      "step": 72870
    },
    {
      "epoch": 1.1925059314407265,
      "grad_norm": 0.1378030627965927,
      "learning_rate": 7.537689536416133e-06,
      "loss": 0.001,
      "step": 72880
    },
    {
      "epoch": 1.1926695573918022,
      "grad_norm": 0.21198512613773346,
      "learning_rate": 7.536869270840668e-06,
      "loss": 0.0026,
      "step": 72890
    },
    {
      "epoch": 1.1928331833428782,
      "grad_norm": 0.11297036707401276,
      "learning_rate": 7.536048913309791e-06,
      "loss": 0.0022,
      "step": 72900
    },
    {
      "epoch": 1.192996809293954,
      "grad_norm": 0.18209007382392883,
      "learning_rate": 7.535228463853234e-06,
      "loss": 0.0016,
      "step": 72910
    },
    {
      "epoch": 1.1931604352450298,
      "grad_norm": 0.09069020301103592,
      "learning_rate": 7.53440792250074e-06,
      "loss": 0.0014,
      "step": 72920
    },
    {
      "epoch": 1.1933240611961058,
      "grad_norm": 0.16172488033771515,
      "learning_rate": 7.5335872892820485e-06,
      "loss": 0.0026,
      "step": 72930
    },
    {
      "epoch": 1.1934876871471816,
      "grad_norm": 0.21352505683898926,
      "learning_rate": 7.532766564226908e-06,
      "loss": 0.002,
      "step": 72940
    },
    {
      "epoch": 1.1936513130982573,
      "grad_norm": 0.08566544950008392,
      "learning_rate": 7.531945747365066e-06,
      "loss": 0.0055,
      "step": 72950
    },
    {
      "epoch": 1.1938149390493331,
      "grad_norm": 0.11239166557788849,
      "learning_rate": 7.531124838726277e-06,
      "loss": 0.0016,
      "step": 72960
    },
    {
      "epoch": 1.193978565000409,
      "grad_norm": 0.04482845216989517,
      "learning_rate": 7.530303838340296e-06,
      "loss": 0.0014,
      "step": 72970
    },
    {
      "epoch": 1.1941421909514849,
      "grad_norm": 0.06593041867017746,
      "learning_rate": 7.5294827462368825e-06,
      "loss": 0.0018,
      "step": 72980
    },
    {
      "epoch": 1.1943058169025607,
      "grad_norm": 0.07472705096006393,
      "learning_rate": 7.528661562445798e-06,
      "loss": 0.0012,
      "step": 72990
    },
    {
      "epoch": 1.1944694428536367,
      "grad_norm": 0.21327978372573853,
      "learning_rate": 7.52784028699681e-06,
      "loss": 0.0027,
      "step": 73000
    },
    {
      "epoch": 1.1946330688047124,
      "grad_norm": 0.14941608905792236,
      "learning_rate": 7.527018919919686e-06,
      "loss": 0.0026,
      "step": 73010
    },
    {
      "epoch": 1.1947966947557882,
      "grad_norm": 0.24984896183013916,
      "learning_rate": 7.526197461244201e-06,
      "loss": 0.004,
      "step": 73020
    },
    {
      "epoch": 1.1949603207068642,
      "grad_norm": 0.2016327977180481,
      "learning_rate": 7.52537591100013e-06,
      "loss": 0.0018,
      "step": 73030
    },
    {
      "epoch": 1.19512394665794,
      "grad_norm": 0.04638541862368584,
      "learning_rate": 7.524554269217252e-06,
      "loss": 0.002,
      "step": 73040
    },
    {
      "epoch": 1.1952875726090157,
      "grad_norm": 0.09789019078016281,
      "learning_rate": 7.5237325359253475e-06,
      "loss": 0.0021,
      "step": 73050
    },
    {
      "epoch": 1.1954511985600917,
      "grad_norm": 0.14114952087402344,
      "learning_rate": 7.522910711154206e-06,
      "loss": 0.003,
      "step": 73060
    },
    {
      "epoch": 1.1956148245111675,
      "grad_norm": 0.26666074991226196,
      "learning_rate": 7.522088794933613e-06,
      "loss": 0.0017,
      "step": 73070
    },
    {
      "epoch": 1.1957784504622433,
      "grad_norm": 0.07841170579195023,
      "learning_rate": 7.521266787293364e-06,
      "loss": 0.0008,
      "step": 73080
    },
    {
      "epoch": 1.195942076413319,
      "grad_norm": 0.050705742090940475,
      "learning_rate": 7.520444688263253e-06,
      "loss": 0.0018,
      "step": 73090
    },
    {
      "epoch": 1.196105702364395,
      "grad_norm": 0.06191769242286682,
      "learning_rate": 7.519622497873081e-06,
      "loss": 0.0031,
      "step": 73100
    },
    {
      "epoch": 1.1962693283154708,
      "grad_norm": 0.03205306455492973,
      "learning_rate": 7.518800216152649e-06,
      "loss": 0.0017,
      "step": 73110
    },
    {
      "epoch": 1.1964329542665466,
      "grad_norm": 0.06465652585029602,
      "learning_rate": 7.517977843131762e-06,
      "loss": 0.0022,
      "step": 73120
    },
    {
      "epoch": 1.1965965802176226,
      "grad_norm": 0.08968547731637955,
      "learning_rate": 7.517155378840231e-06,
      "loss": 0.0034,
      "step": 73130
    },
    {
      "epoch": 1.1967602061686984,
      "grad_norm": 0.0811348631978035,
      "learning_rate": 7.516332823307867e-06,
      "loss": 0.0018,
      "step": 73140
    },
    {
      "epoch": 1.1969238321197742,
      "grad_norm": 0.0599542073905468,
      "learning_rate": 7.515510176564485e-06,
      "loss": 0.0021,
      "step": 73150
    },
    {
      "epoch": 1.19708745807085,
      "grad_norm": 0.16323351860046387,
      "learning_rate": 7.514687438639905e-06,
      "loss": 0.0025,
      "step": 73160
    },
    {
      "epoch": 1.197251084021926,
      "grad_norm": 0.18347062170505524,
      "learning_rate": 7.51386460956395e-06,
      "loss": 0.0024,
      "step": 73170
    },
    {
      "epoch": 1.1974147099730017,
      "grad_norm": 0.04154191538691521,
      "learning_rate": 7.513041689366446e-06,
      "loss": 0.0023,
      "step": 73180
    },
    {
      "epoch": 1.1975783359240775,
      "grad_norm": 0.26237234473228455,
      "learning_rate": 7.512218678077219e-06,
      "loss": 0.002,
      "step": 73190
    },
    {
      "epoch": 1.1977419618751535,
      "grad_norm": 0.24974653124809265,
      "learning_rate": 7.511395575726104e-06,
      "loss": 0.0036,
      "step": 73200
    },
    {
      "epoch": 1.1979055878262292,
      "grad_norm": 0.19067102670669556,
      "learning_rate": 7.5105723823429345e-06,
      "loss": 0.0027,
      "step": 73210
    },
    {
      "epoch": 1.198069213777305,
      "grad_norm": 0.13213880360126495,
      "learning_rate": 7.50974909795755e-06,
      "loss": 0.002,
      "step": 73220
    },
    {
      "epoch": 1.198232839728381,
      "grad_norm": 0.12766993045806885,
      "learning_rate": 7.5089257225997935e-06,
      "loss": 0.0019,
      "step": 73230
    },
    {
      "epoch": 1.1983964656794568,
      "grad_norm": 0.16532419621944427,
      "learning_rate": 7.50810225629951e-06,
      "loss": 0.0023,
      "step": 73240
    },
    {
      "epoch": 1.1985600916305326,
      "grad_norm": 0.13638997077941895,
      "learning_rate": 7.5072786990865465e-06,
      "loss": 0.0017,
      "step": 73250
    },
    {
      "epoch": 1.1987237175816086,
      "grad_norm": 0.26773399114608765,
      "learning_rate": 7.506455050990756e-06,
      "loss": 0.0018,
      "step": 73260
    },
    {
      "epoch": 1.1988873435326843,
      "grad_norm": 0.06717701256275177,
      "learning_rate": 7.505631312041994e-06,
      "loss": 0.0015,
      "step": 73270
    },
    {
      "epoch": 1.19905096948376,
      "grad_norm": 0.09393195807933807,
      "learning_rate": 7.5048074822701226e-06,
      "loss": 0.0014,
      "step": 73280
    },
    {
      "epoch": 1.1992145954348359,
      "grad_norm": 0.13148963451385498,
      "learning_rate": 7.503983561704995e-06,
      "loss": 0.0023,
      "step": 73290
    },
    {
      "epoch": 1.1993782213859119,
      "grad_norm": 0.13892988860607147,
      "learning_rate": 7.5031595503764846e-06,
      "loss": 0.0037,
      "step": 73300
    },
    {
      "epoch": 1.1995418473369877,
      "grad_norm": 0.15006650984287262,
      "learning_rate": 7.502335448314455e-06,
      "loss": 0.0021,
      "step": 73310
    },
    {
      "epoch": 1.1997054732880634,
      "grad_norm": 0.07400722801685333,
      "learning_rate": 7.501511255548782e-06,
      "loss": 0.0018,
      "step": 73320
    },
    {
      "epoch": 1.1998690992391392,
      "grad_norm": 0.1496189534664154,
      "learning_rate": 7.5006869721093365e-06,
      "loss": 0.003,
      "step": 73330
    },
    {
      "epoch": 1.2000327251902152,
      "grad_norm": 0.03369363397359848,
      "learning_rate": 7.499862598025999e-06,
      "loss": 0.0026,
      "step": 73340
    },
    {
      "epoch": 1.200196351141291,
      "grad_norm": 0.04446745291352272,
      "learning_rate": 7.499038133328649e-06,
      "loss": 0.0017,
      "step": 73350
    },
    {
      "epoch": 1.2003599770923667,
      "grad_norm": 0.13312967121601105,
      "learning_rate": 7.498213578047174e-06,
      "loss": 0.0028,
      "step": 73360
    },
    {
      "epoch": 1.2005236030434427,
      "grad_norm": 0.037850793451070786,
      "learning_rate": 7.4973889322114614e-06,
      "loss": 0.0015,
      "step": 73370
    },
    {
      "epoch": 1.2006872289945185,
      "grad_norm": 0.12466956675052643,
      "learning_rate": 7.496564195851401e-06,
      "loss": 0.0032,
      "step": 73380
    },
    {
      "epoch": 1.2008508549455943,
      "grad_norm": 0.16296936571598053,
      "learning_rate": 7.4957393689968914e-06,
      "loss": 0.0021,
      "step": 73390
    },
    {
      "epoch": 1.2010144808966703,
      "grad_norm": 0.056699302047491074,
      "learning_rate": 7.494914451677827e-06,
      "loss": 0.0022,
      "step": 73400
    },
    {
      "epoch": 1.201178106847746,
      "grad_norm": 0.058173660188913345,
      "learning_rate": 7.4940894439241094e-06,
      "loss": 0.0023,
      "step": 73410
    },
    {
      "epoch": 1.2013417327988218,
      "grad_norm": 0.18045341968536377,
      "learning_rate": 7.493264345765644e-06,
      "loss": 0.0073,
      "step": 73420
    },
    {
      "epoch": 1.2015053587498978,
      "grad_norm": 0.08167928457260132,
      "learning_rate": 7.492439157232339e-06,
      "loss": 0.0027,
      "step": 73430
    },
    {
      "epoch": 1.2016689847009736,
      "grad_norm": 0.1054738461971283,
      "learning_rate": 7.491613878354105e-06,
      "loss": 0.0018,
      "step": 73440
    },
    {
      "epoch": 1.2018326106520494,
      "grad_norm": 0.3511328101158142,
      "learning_rate": 7.490788509160856e-06,
      "loss": 0.0014,
      "step": 73450
    },
    {
      "epoch": 1.2019962366031254,
      "grad_norm": 0.05170730873942375,
      "learning_rate": 7.489963049682509e-06,
      "loss": 0.0011,
      "step": 73460
    },
    {
      "epoch": 1.2021598625542012,
      "grad_norm": 0.1402779370546341,
      "learning_rate": 7.4891374999489875e-06,
      "loss": 0.002,
      "step": 73470
    },
    {
      "epoch": 1.202323488505277,
      "grad_norm": 0.7360239028930664,
      "learning_rate": 7.4883118599902136e-06,
      "loss": 0.0029,
      "step": 73480
    },
    {
      "epoch": 1.2024871144563527,
      "grad_norm": 0.12634000182151794,
      "learning_rate": 7.487486129836115e-06,
      "loss": 0.0042,
      "step": 73490
    },
    {
      "epoch": 1.2026507404074287,
      "grad_norm": 0.2138158529996872,
      "learning_rate": 7.486660309516624e-06,
      "loss": 0.0025,
      "step": 73500
    },
    {
      "epoch": 1.2028143663585045,
      "grad_norm": 0.04578448086977005,
      "learning_rate": 7.485834399061671e-06,
      "loss": 0.0017,
      "step": 73510
    },
    {
      "epoch": 1.2029779923095802,
      "grad_norm": 0.07813996076583862,
      "learning_rate": 7.485008398501197e-06,
      "loss": 0.002,
      "step": 73520
    },
    {
      "epoch": 1.203141618260656,
      "grad_norm": 0.18239513039588928,
      "learning_rate": 7.48418230786514e-06,
      "loss": 0.0017,
      "step": 73530
    },
    {
      "epoch": 1.203305244211732,
      "grad_norm": 0.038591306656599045,
      "learning_rate": 7.483356127183448e-06,
      "loss": 0.0016,
      "step": 73540
    },
    {
      "epoch": 1.2034688701628078,
      "grad_norm": 0.10440230369567871,
      "learning_rate": 7.482529856486063e-06,
      "loss": 0.003,
      "step": 73550
    },
    {
      "epoch": 1.2036324961138836,
      "grad_norm": 0.0993897020816803,
      "learning_rate": 7.481703495802937e-06,
      "loss": 0.0015,
      "step": 73560
    },
    {
      "epoch": 1.2037961220649596,
      "grad_norm": 0.0966823622584343,
      "learning_rate": 7.480877045164023e-06,
      "loss": 0.0028,
      "step": 73570
    },
    {
      "epoch": 1.2039597480160353,
      "grad_norm": 0.026725709438323975,
      "learning_rate": 7.48005050459928e-06,
      "loss": 0.002,
      "step": 73580
    },
    {
      "epoch": 1.204123373967111,
      "grad_norm": 0.13750910758972168,
      "learning_rate": 7.4792238741386655e-06,
      "loss": 0.003,
      "step": 73590
    },
    {
      "epoch": 1.204286999918187,
      "grad_norm": 0.1044209897518158,
      "learning_rate": 7.478397153812146e-06,
      "loss": 0.002,
      "step": 73600
    },
    {
      "epoch": 1.2044506258692629,
      "grad_norm": 0.3290788531303406,
      "learning_rate": 7.477570343649686e-06,
      "loss": 0.0021,
      "step": 73610
    },
    {
      "epoch": 1.2046142518203387,
      "grad_norm": 0.0903046503663063,
      "learning_rate": 7.476743443681255e-06,
      "loss": 0.0024,
      "step": 73620
    },
    {
      "epoch": 1.2047778777714147,
      "grad_norm": 0.010570233687758446,
      "learning_rate": 7.475916453936827e-06,
      "loss": 0.0014,
      "step": 73630
    },
    {
      "epoch": 1.2049415037224904,
      "grad_norm": 0.00955671351402998,
      "learning_rate": 7.47508937444638e-06,
      "loss": 0.0027,
      "step": 73640
    },
    {
      "epoch": 1.2051051296735662,
      "grad_norm": 0.06452976167201996,
      "learning_rate": 7.47426220523989e-06,
      "loss": 0.0007,
      "step": 73650
    },
    {
      "epoch": 1.205268755624642,
      "grad_norm": 0.1388501673936844,
      "learning_rate": 7.473434946347341e-06,
      "loss": 0.0025,
      "step": 73660
    },
    {
      "epoch": 1.205432381575718,
      "grad_norm": 0.03420530632138252,
      "learning_rate": 7.472607597798721e-06,
      "loss": 0.0018,
      "step": 73670
    },
    {
      "epoch": 1.2055960075267937,
      "grad_norm": 0.13640691339969635,
      "learning_rate": 7.471780159624019e-06,
      "loss": 0.0026,
      "step": 73680
    },
    {
      "epoch": 1.2057596334778695,
      "grad_norm": 0.13650503754615784,
      "learning_rate": 7.470952631853228e-06,
      "loss": 0.0038,
      "step": 73690
    },
    {
      "epoch": 1.2059232594289455,
      "grad_norm": 0.042516257613897324,
      "learning_rate": 7.4701250145163404e-06,
      "loss": 0.0023,
      "step": 73700
    },
    {
      "epoch": 1.2060868853800213,
      "grad_norm": 0.01118846982717514,
      "learning_rate": 7.469297307643359e-06,
      "loss": 0.0019,
      "step": 73710
    },
    {
      "epoch": 1.206250511331097,
      "grad_norm": 0.2110971212387085,
      "learning_rate": 7.468469511264285e-06,
      "loss": 0.0032,
      "step": 73720
    },
    {
      "epoch": 1.2064141372821728,
      "grad_norm": 0.18284577131271362,
      "learning_rate": 7.467641625409122e-06,
      "loss": 0.0027,
      "step": 73730
    },
    {
      "epoch": 1.2065777632332488,
      "grad_norm": 0.04361890256404877,
      "learning_rate": 7.466813650107884e-06,
      "loss": 0.0017,
      "step": 73740
    },
    {
      "epoch": 1.2067413891843246,
      "grad_norm": 0.0691043883562088,
      "learning_rate": 7.465985585390579e-06,
      "loss": 0.0017,
      "step": 73750
    },
    {
      "epoch": 1.2069050151354004,
      "grad_norm": 0.15436670184135437,
      "learning_rate": 7.4651574312872225e-06,
      "loss": 0.0016,
      "step": 73760
    },
    {
      "epoch": 1.2070686410864764,
      "grad_norm": 0.08111707866191864,
      "learning_rate": 7.464329187827835e-06,
      "loss": 0.0022,
      "step": 73770
    },
    {
      "epoch": 1.2072322670375522,
      "grad_norm": 0.06875109672546387,
      "learning_rate": 7.4635008550424355e-06,
      "loss": 0.0012,
      "step": 73780
    },
    {
      "epoch": 1.207395892988628,
      "grad_norm": 0.18053790926933289,
      "learning_rate": 7.462672432961053e-06,
      "loss": 0.0024,
      "step": 73790
    },
    {
      "epoch": 1.207559518939704,
      "grad_norm": 0.028427856042981148,
      "learning_rate": 7.4618439216137125e-06,
      "loss": 0.0009,
      "step": 73800
    },
    {
      "epoch": 1.2077231448907797,
      "grad_norm": 0.04011492803692818,
      "learning_rate": 7.461015321030447e-06,
      "loss": 0.0016,
      "step": 73810
    },
    {
      "epoch": 1.2078867708418555,
      "grad_norm": 0.30109745264053345,
      "learning_rate": 7.4601866312412905e-06,
      "loss": 0.0026,
      "step": 73820
    },
    {
      "epoch": 1.2080503967929315,
      "grad_norm": 0.22535495460033417,
      "learning_rate": 7.459357852276284e-06,
      "loss": 0.0034,
      "step": 73830
    },
    {
      "epoch": 1.2082140227440072,
      "grad_norm": 0.1572130024433136,
      "learning_rate": 7.458528984165464e-06,
      "loss": 0.0032,
      "step": 73840
    },
    {
      "epoch": 1.208377648695083,
      "grad_norm": 0.15236243605613708,
      "learning_rate": 7.457700026938878e-06,
      "loss": 0.0022,
      "step": 73850
    },
    {
      "epoch": 1.2085412746461588,
      "grad_norm": 0.13043826818466187,
      "learning_rate": 7.456870980626573e-06,
      "loss": 0.0013,
      "step": 73860
    },
    {
      "epoch": 1.2087049005972348,
      "grad_norm": 0.09628401696681976,
      "learning_rate": 7.456041845258598e-06,
      "loss": 0.0013,
      "step": 73870
    },
    {
      "epoch": 1.2088685265483106,
      "grad_norm": 0.040267400443553925,
      "learning_rate": 7.45521262086501e-06,
      "loss": 0.0017,
      "step": 73880
    },
    {
      "epoch": 1.2090321524993863,
      "grad_norm": 0.008964836597442627,
      "learning_rate": 7.454383307475867e-06,
      "loss": 0.0029,
      "step": 73890
    },
    {
      "epoch": 1.2091957784504623,
      "grad_norm": 0.030985837802290916,
      "learning_rate": 7.4535539051212266e-06,
      "loss": 0.0013,
      "step": 73900
    },
    {
      "epoch": 1.209359404401538,
      "grad_norm": 0.13246433436870575,
      "learning_rate": 7.452724413831155e-06,
      "loss": 0.0038,
      "step": 73910
    },
    {
      "epoch": 1.2095230303526139,
      "grad_norm": 0.08268111944198608,
      "learning_rate": 7.451894833635718e-06,
      "loss": 0.0021,
      "step": 73920
    },
    {
      "epoch": 1.2096866563036897,
      "grad_norm": 0.08752504736185074,
      "learning_rate": 7.451065164564985e-06,
      "loss": 0.0025,
      "step": 73930
    },
    {
      "epoch": 1.2098502822547657,
      "grad_norm": 0.34026047587394714,
      "learning_rate": 7.450235406649033e-06,
      "loss": 0.0025,
      "step": 73940
    },
    {
      "epoch": 1.2100139082058414,
      "grad_norm": 0.10241920500993729,
      "learning_rate": 7.449405559917935e-06,
      "loss": 0.002,
      "step": 73950
    },
    {
      "epoch": 1.2101775341569172,
      "grad_norm": 0.1253937929868698,
      "learning_rate": 7.448575624401773e-06,
      "loss": 0.0022,
      "step": 73960
    },
    {
      "epoch": 1.2103411601079932,
      "grad_norm": 0.040853723883628845,
      "learning_rate": 7.4477456001306295e-06,
      "loss": 0.0017,
      "step": 73970
    },
    {
      "epoch": 1.210504786059069,
      "grad_norm": 0.07868856191635132,
      "learning_rate": 7.446915487134591e-06,
      "loss": 0.0038,
      "step": 73980
    },
    {
      "epoch": 1.2106684120101447,
      "grad_norm": 0.09584248811006546,
      "learning_rate": 7.446085285443748e-06,
      "loss": 0.0032,
      "step": 73990
    },
    {
      "epoch": 1.2108320379612207,
      "grad_norm": 0.048996828496456146,
      "learning_rate": 7.4452549950881905e-06,
      "loss": 0.0019,
      "step": 74000
    },
    {
      "epoch": 1.2109956639122965,
      "grad_norm": 0.1800202876329422,
      "learning_rate": 7.4444246160980176e-06,
      "loss": 0.0021,
      "step": 74010
    },
    {
      "epoch": 1.2111592898633723,
      "grad_norm": 0.09470319747924805,
      "learning_rate": 7.443594148503327e-06,
      "loss": 0.004,
      "step": 74020
    },
    {
      "epoch": 1.2113229158144483,
      "grad_norm": 0.23442859947681427,
      "learning_rate": 7.442763592334222e-06,
      "loss": 0.002,
      "step": 74030
    },
    {
      "epoch": 1.211486541765524,
      "grad_norm": 0.07005157321691513,
      "learning_rate": 7.441932947620808e-06,
      "loss": 0.0021,
      "step": 74040
    },
    {
      "epoch": 1.2116501677165998,
      "grad_norm": 0.09264736622571945,
      "learning_rate": 7.441102214393193e-06,
      "loss": 0.0021,
      "step": 74050
    },
    {
      "epoch": 1.2118137936676756,
      "grad_norm": 0.31985676288604736,
      "learning_rate": 7.440271392681491e-06,
      "loss": 0.0032,
      "step": 74060
    },
    {
      "epoch": 1.2119774196187516,
      "grad_norm": 0.18059299886226654,
      "learning_rate": 7.439440482515814e-06,
      "loss": 0.0017,
      "step": 74070
    },
    {
      "epoch": 1.2121410455698274,
      "grad_norm": 0.16675914824008942,
      "learning_rate": 7.438609483926282e-06,
      "loss": 0.0015,
      "step": 74080
    },
    {
      "epoch": 1.2123046715209032,
      "grad_norm": 0.24970807135105133,
      "learning_rate": 7.4377783969430206e-06,
      "loss": 0.0042,
      "step": 74090
    },
    {
      "epoch": 1.2124682974719792,
      "grad_norm": 0.11183233559131622,
      "learning_rate": 7.436947221596149e-06,
      "loss": 0.0025,
      "step": 74100
    },
    {
      "epoch": 1.212631923423055,
      "grad_norm": 0.09275684505701065,
      "learning_rate": 7.436115957915799e-06,
      "loss": 0.0019,
      "step": 74110
    },
    {
      "epoch": 1.2127955493741307,
      "grad_norm": 0.15110190212726593,
      "learning_rate": 7.435284605932099e-06,
      "loss": 0.0022,
      "step": 74120
    },
    {
      "epoch": 1.2129591753252065,
      "grad_norm": 0.028079111129045486,
      "learning_rate": 7.434453165675186e-06,
      "loss": 0.0023,
      "step": 74130
    },
    {
      "epoch": 1.2131228012762825,
      "grad_norm": 0.15616779029369354,
      "learning_rate": 7.433621637175197e-06,
      "loss": 0.0013,
      "step": 74140
    },
    {
      "epoch": 1.2132864272273582,
      "grad_norm": 0.1304924041032791,
      "learning_rate": 7.4327900204622725e-06,
      "loss": 0.0017,
      "step": 74150
    },
    {
      "epoch": 1.213450053178434,
      "grad_norm": 0.1529300957918167,
      "learning_rate": 7.4319583155665565e-06,
      "loss": 0.0024,
      "step": 74160
    },
    {
      "epoch": 1.21361367912951,
      "grad_norm": 0.12559092044830322,
      "learning_rate": 7.431126522518197e-06,
      "loss": 0.0044,
      "step": 74170
    },
    {
      "epoch": 1.2137773050805858,
      "grad_norm": 0.0652841329574585,
      "learning_rate": 7.430294641347344e-06,
      "loss": 0.0012,
      "step": 74180
    },
    {
      "epoch": 1.2139409310316616,
      "grad_norm": 0.0622672438621521,
      "learning_rate": 7.429462672084153e-06,
      "loss": 0.0016,
      "step": 74190
    },
    {
      "epoch": 1.2141045569827376,
      "grad_norm": 0.02094794623553753,
      "learning_rate": 7.428630614758776e-06,
      "loss": 0.0017,
      "step": 74200
    },
    {
      "epoch": 1.2142681829338133,
      "grad_norm": 0.18284811079502106,
      "learning_rate": 7.427798469401378e-06,
      "loss": 0.0022,
      "step": 74210
    },
    {
      "epoch": 1.214431808884889,
      "grad_norm": 0.1217878982424736,
      "learning_rate": 7.426966236042119e-06,
      "loss": 0.002,
      "step": 74220
    },
    {
      "epoch": 1.214595434835965,
      "grad_norm": 0.07922865450382233,
      "learning_rate": 7.426133914711169e-06,
      "loss": 0.0018,
      "step": 74230
    },
    {
      "epoch": 1.2147590607870409,
      "grad_norm": 0.29961922764778137,
      "learning_rate": 7.425301505438696e-06,
      "loss": 0.0048,
      "step": 74240
    },
    {
      "epoch": 1.2149226867381167,
      "grad_norm": 0.041613925248384476,
      "learning_rate": 7.4244690082548706e-06,
      "loss": 0.0041,
      "step": 74250
    },
    {
      "epoch": 1.2150863126891924,
      "grad_norm": 0.17574548721313477,
      "learning_rate": 7.423636423189873e-06,
      "loss": 0.0022,
      "step": 74260
    },
    {
      "epoch": 1.2152499386402684,
      "grad_norm": 0.0806005522608757,
      "learning_rate": 7.4228037502738785e-06,
      "loss": 0.0026,
      "step": 74270
    },
    {
      "epoch": 1.2154135645913442,
      "grad_norm": 0.09811805188655853,
      "learning_rate": 7.4219709895370705e-06,
      "loss": 0.0019,
      "step": 74280
    },
    {
      "epoch": 1.21557719054242,
      "grad_norm": 0.3578907549381256,
      "learning_rate": 7.421138141009636e-06,
      "loss": 0.0024,
      "step": 74290
    },
    {
      "epoch": 1.2157408164934957,
      "grad_norm": 0.1324716955423355,
      "learning_rate": 7.4203052047217624e-06,
      "loss": 0.0017,
      "step": 74300
    },
    {
      "epoch": 1.2159044424445717,
      "grad_norm": 0.1546030193567276,
      "learning_rate": 7.419472180703643e-06,
      "loss": 0.0023,
      "step": 74310
    },
    {
      "epoch": 1.2160680683956475,
      "grad_norm": 0.05751853808760643,
      "learning_rate": 7.4186390689854714e-06,
      "loss": 0.0024,
      "step": 74320
    },
    {
      "epoch": 1.2162316943467233,
      "grad_norm": 0.21121624112129211,
      "learning_rate": 7.4178058695974474e-06,
      "loss": 0.0048,
      "step": 74330
    },
    {
      "epoch": 1.2163953202977993,
      "grad_norm": 0.2034398317337036,
      "learning_rate": 7.416972582569772e-06,
      "loss": 0.0017,
      "step": 74340
    },
    {
      "epoch": 1.216558946248875,
      "grad_norm": 0.0504169799387455,
      "learning_rate": 7.416139207932649e-06,
      "loss": 0.0015,
      "step": 74350
    },
    {
      "epoch": 1.2167225721999508,
      "grad_norm": 0.17262530326843262,
      "learning_rate": 7.415305745716286e-06,
      "loss": 0.0017,
      "step": 74360
    },
    {
      "epoch": 1.2168861981510268,
      "grad_norm": 0.2116984874010086,
      "learning_rate": 7.414472195950894e-06,
      "loss": 0.0021,
      "step": 74370
    },
    {
      "epoch": 1.2170498241021026,
      "grad_norm": 0.022609880194067955,
      "learning_rate": 7.413638558666689e-06,
      "loss": 0.0018,
      "step": 74380
    },
    {
      "epoch": 1.2172134500531784,
      "grad_norm": 0.14700868725776672,
      "learning_rate": 7.41280483389389e-06,
      "loss": 0.0015,
      "step": 74390
    },
    {
      "epoch": 1.2173770760042544,
      "grad_norm": 0.09180580079555511,
      "learning_rate": 7.411971021662712e-06,
      "loss": 0.0016,
      "step": 74400
    },
    {
      "epoch": 1.2175407019553302,
      "grad_norm": 0.061221521347761154,
      "learning_rate": 7.411137122003381e-06,
      "loss": 0.0022,
      "step": 74410
    },
    {
      "epoch": 1.217704327906406,
      "grad_norm": 0.20302820205688477,
      "learning_rate": 7.410303134946125e-06,
      "loss": 0.0027,
      "step": 74420
    },
    {
      "epoch": 1.2178679538574817,
      "grad_norm": 0.08845239132642746,
      "learning_rate": 7.409469060521174e-06,
      "loss": 0.0019,
      "step": 74430
    },
    {
      "epoch": 1.2180315798085577,
      "grad_norm": 0.1340804249048233,
      "learning_rate": 7.40863489875876e-06,
      "loss": 0.0018,
      "step": 74440
    },
    {
      "epoch": 1.2181952057596335,
      "grad_norm": 0.2781793475151062,
      "learning_rate": 7.4078006496891195e-06,
      "loss": 0.0028,
      "step": 74450
    },
    {
      "epoch": 1.2183588317107092,
      "grad_norm": 0.0992671549320221,
      "learning_rate": 7.406966313342493e-06,
      "loss": 0.0018,
      "step": 74460
    },
    {
      "epoch": 1.2185224576617852,
      "grad_norm": 0.05967795103788376,
      "learning_rate": 7.406131889749122e-06,
      "loss": 0.0019,
      "step": 74470
    },
    {
      "epoch": 1.218686083612861,
      "grad_norm": 0.07916160672903061,
      "learning_rate": 7.405297378939253e-06,
      "loss": 0.002,
      "step": 74480
    },
    {
      "epoch": 1.2188497095639368,
      "grad_norm": 0.07955983281135559,
      "learning_rate": 7.404462780943138e-06,
      "loss": 0.0019,
      "step": 74490
    },
    {
      "epoch": 1.2190133355150126,
      "grad_norm": 0.08532337844371796,
      "learning_rate": 7.4036280957910226e-06,
      "loss": 0.0034,
      "step": 74500
    },
    {
      "epoch": 1.2191769614660886,
      "grad_norm": 0.08964477479457855,
      "learning_rate": 7.4027933235131666e-06,
      "loss": 0.0019,
      "step": 74510
    },
    {
      "epoch": 1.2193405874171643,
      "grad_norm": 0.09393186867237091,
      "learning_rate": 7.401958464139827e-06,
      "loss": 0.0014,
      "step": 74520
    },
    {
      "epoch": 1.21950421336824,
      "grad_norm": 0.04157751798629761,
      "learning_rate": 7.401123517701268e-06,
      "loss": 0.004,
      "step": 74530
    },
    {
      "epoch": 1.219667839319316,
      "grad_norm": 0.15870150923728943,
      "learning_rate": 7.400288484227751e-06,
      "loss": 0.0018,
      "step": 74540
    },
    {
      "epoch": 1.2198314652703919,
      "grad_norm": 0.1219710111618042,
      "learning_rate": 7.399453363749546e-06,
      "loss": 0.0019,
      "step": 74550
    },
    {
      "epoch": 1.2199950912214677,
      "grad_norm": 0.03308141976594925,
      "learning_rate": 7.398618156296923e-06,
      "loss": 0.0018,
      "step": 74560
    },
    {
      "epoch": 1.2201587171725436,
      "grad_norm": 0.22446565330028534,
      "learning_rate": 7.3977828619001564e-06,
      "loss": 0.0018,
      "step": 74570
    },
    {
      "epoch": 1.2203223431236194,
      "grad_norm": 0.14658498764038086,
      "learning_rate": 7.396947480589524e-06,
      "loss": 0.0028,
      "step": 74580
    },
    {
      "epoch": 1.2204859690746952,
      "grad_norm": 0.07871846854686737,
      "learning_rate": 7.396112012395308e-06,
      "loss": 0.0034,
      "step": 74590
    },
    {
      "epoch": 1.2206495950257712,
      "grad_norm": 0.22183069586753845,
      "learning_rate": 7.395276457347789e-06,
      "loss": 0.0024,
      "step": 74600
    },
    {
      "epoch": 1.220813220976847,
      "grad_norm": 0.12578792870044708,
      "learning_rate": 7.394440815477255e-06,
      "loss": 0.0021,
      "step": 74610
    },
    {
      "epoch": 1.2209768469279227,
      "grad_norm": 0.24926680326461792,
      "learning_rate": 7.393605086813998e-06,
      "loss": 0.0035,
      "step": 74620
    },
    {
      "epoch": 1.2211404728789985,
      "grad_norm": 0.07707841694355011,
      "learning_rate": 7.392769271388307e-06,
      "loss": 0.0027,
      "step": 74630
    },
    {
      "epoch": 1.2213040988300745,
      "grad_norm": 0.10797322541475296,
      "learning_rate": 7.3919333692304825e-06,
      "loss": 0.0019,
      "step": 74640
    },
    {
      "epoch": 1.2214677247811503,
      "grad_norm": 0.3985747694969177,
      "learning_rate": 7.391097380370822e-06,
      "loss": 0.0025,
      "step": 74650
    },
    {
      "epoch": 1.221631350732226,
      "grad_norm": 0.2966378927230835,
      "learning_rate": 7.390261304839628e-06,
      "loss": 0.0025,
      "step": 74660
    },
    {
      "epoch": 1.221794976683302,
      "grad_norm": 0.11711665242910385,
      "learning_rate": 7.389425142667207e-06,
      "loss": 0.0012,
      "step": 74670
    },
    {
      "epoch": 1.2219586026343778,
      "grad_norm": 0.25657913088798523,
      "learning_rate": 7.388588893883867e-06,
      "loss": 0.0038,
      "step": 74680
    },
    {
      "epoch": 1.2221222285854536,
      "grad_norm": 0.1111886128783226,
      "learning_rate": 7.3877525585199206e-06,
      "loss": 0.0024,
      "step": 74690
    },
    {
      "epoch": 1.2222858545365294,
      "grad_norm": 0.07965485751628876,
      "learning_rate": 7.386916136605683e-06,
      "loss": 0.0011,
      "step": 74700
    },
    {
      "epoch": 1.2224494804876054,
      "grad_norm": 0.12249523401260376,
      "learning_rate": 7.386079628171472e-06,
      "loss": 0.0037,
      "step": 74710
    },
    {
      "epoch": 1.2226131064386812,
      "grad_norm": 0.1074841096997261,
      "learning_rate": 7.385243033247609e-06,
      "loss": 0.0016,
      "step": 74720
    },
    {
      "epoch": 1.222776732389757,
      "grad_norm": 0.09446748346090317,
      "learning_rate": 7.384406351864418e-06,
      "loss": 0.0012,
      "step": 74730
    },
    {
      "epoch": 1.222940358340833,
      "grad_norm": 0.05598645284771919,
      "learning_rate": 7.38356958405223e-06,
      "loss": 0.0022,
      "step": 74740
    },
    {
      "epoch": 1.2231039842919087,
      "grad_norm": 0.22682826220989227,
      "learning_rate": 7.382732729841372e-06,
      "loss": 0.0022,
      "step": 74750
    },
    {
      "epoch": 1.2232676102429845,
      "grad_norm": 0.2344180941581726,
      "learning_rate": 7.381895789262179e-06,
      "loss": 0.0044,
      "step": 74760
    },
    {
      "epoch": 1.2234312361940605,
      "grad_norm": 0.21609066426753998,
      "learning_rate": 7.381058762344988e-06,
      "loss": 0.0032,
      "step": 74770
    },
    {
      "epoch": 1.2235948621451362,
      "grad_norm": 0.09093273431062698,
      "learning_rate": 7.380221649120141e-06,
      "loss": 0.0021,
      "step": 74780
    },
    {
      "epoch": 1.223758488096212,
      "grad_norm": 0.031923212110996246,
      "learning_rate": 7.379384449617979e-06,
      "loss": 0.0027,
      "step": 74790
    },
    {
      "epoch": 1.223922114047288,
      "grad_norm": 0.1188010424375534,
      "learning_rate": 7.37854716386885e-06,
      "loss": 0.0016,
      "step": 74800
    },
    {
      "epoch": 1.2240857399983638,
      "grad_norm": 0.14683733880519867,
      "learning_rate": 7.377709791903102e-06,
      "loss": 0.0014,
      "step": 74810
    },
    {
      "epoch": 1.2242493659494396,
      "grad_norm": 0.050854261964559555,
      "learning_rate": 7.376872333751089e-06,
      "loss": 0.0017,
      "step": 74820
    },
    {
      "epoch": 1.2244129919005153,
      "grad_norm": 0.08790922164916992,
      "learning_rate": 7.376034789443167e-06,
      "loss": 0.0017,
      "step": 74830
    },
    {
      "epoch": 1.2245766178515913,
      "grad_norm": 0.01875845342874527,
      "learning_rate": 7.375197159009695e-06,
      "loss": 0.001,
      "step": 74840
    },
    {
      "epoch": 1.224740243802667,
      "grad_norm": 0.24824963510036469,
      "learning_rate": 7.3743594424810335e-06,
      "loss": 0.0018,
      "step": 74850
    },
    {
      "epoch": 1.2249038697537429,
      "grad_norm": 0.13142342865467072,
      "learning_rate": 7.373521639887548e-06,
      "loss": 0.0016,
      "step": 74860
    },
    {
      "epoch": 1.2250674957048189,
      "grad_norm": 0.07221851497888565,
      "learning_rate": 7.3726837512596094e-06,
      "loss": 0.0021,
      "step": 74870
    },
    {
      "epoch": 1.2252311216558947,
      "grad_norm": 0.06531517207622528,
      "learning_rate": 7.3718457766275865e-06,
      "loss": 0.0027,
      "step": 74880
    },
    {
      "epoch": 1.2253947476069704,
      "grad_norm": 0.26921960711479187,
      "learning_rate": 7.371007716021857e-06,
      "loss": 0.0027,
      "step": 74890
    },
    {
      "epoch": 1.2255583735580462,
      "grad_norm": 0.11885719746351242,
      "learning_rate": 7.370169569472794e-06,
      "loss": 0.0016,
      "step": 74900
    },
    {
      "epoch": 1.2257219995091222,
      "grad_norm": 0.13144849240779877,
      "learning_rate": 7.3693313370107814e-06,
      "loss": 0.0023,
      "step": 74910
    },
    {
      "epoch": 1.225885625460198,
      "grad_norm": 0.0871674045920372,
      "learning_rate": 7.368493018666202e-06,
      "loss": 0.0023,
      "step": 74920
    },
    {
      "epoch": 1.2260492514112737,
      "grad_norm": 0.14775460958480835,
      "learning_rate": 7.367654614469444e-06,
      "loss": 0.002,
      "step": 74930
    },
    {
      "epoch": 1.2262128773623497,
      "grad_norm": 0.03863407298922539,
      "learning_rate": 7.366816124450895e-06,
      "loss": 0.002,
      "step": 74940
    },
    {
      "epoch": 1.2263765033134255,
      "grad_norm": 0.08186143636703491,
      "learning_rate": 7.365977548640952e-06,
      "loss": 0.0013,
      "step": 74950
    },
    {
      "epoch": 1.2265401292645013,
      "grad_norm": 0.1456160843372345,
      "learning_rate": 7.365138887070008e-06,
      "loss": 0.0025,
      "step": 74960
    },
    {
      "epoch": 1.2267037552155773,
      "grad_norm": 0.032390229403972626,
      "learning_rate": 7.364300139768464e-06,
      "loss": 0.0011,
      "step": 74970
    },
    {
      "epoch": 1.226867381166653,
      "grad_norm": 0.2192324697971344,
      "learning_rate": 7.363461306766722e-06,
      "loss": 0.0029,
      "step": 74980
    },
    {
      "epoch": 1.2270310071177288,
      "grad_norm": 0.1215648353099823,
      "learning_rate": 7.362622388095191e-06,
      "loss": 0.003,
      "step": 74990
    },
    {
      "epoch": 1.2271946330688048,
      "grad_norm": 0.11336185783147812,
      "learning_rate": 7.361783383784274e-06,
      "loss": 0.0025,
      "step": 75000
    },
    {
      "epoch": 1.2271946330688048,
      "eval_loss": 0.0017955370713025331,
      "eval_runtime": 3.0975,
      "eval_samples_per_second": 64.568,
      "eval_steps_per_second": 16.142,
      "step": 75000
    },
    {
      "epoch": 1.2273582590198806,
      "grad_norm": 0.15772020816802979,
      "learning_rate": 7.360944293864385e-06,
      "loss": 0.0016,
      "step": 75010
    },
    {
      "epoch": 1.2275218849709564,
      "grad_norm": 0.13088387250900269,
      "learning_rate": 7.360105118365941e-06,
      "loss": 0.0023,
      "step": 75020
    },
    {
      "epoch": 1.2276855109220322,
      "grad_norm": 0.045769598335027695,
      "learning_rate": 7.3592658573193595e-06,
      "loss": 0.0019,
      "step": 75030
    },
    {
      "epoch": 1.2278491368731081,
      "grad_norm": 0.16903211176395416,
      "learning_rate": 7.358426510755061e-06,
      "loss": 0.0035,
      "step": 75040
    },
    {
      "epoch": 1.228012762824184,
      "grad_norm": 0.03118967078626156,
      "learning_rate": 7.35758707870347e-06,
      "loss": 0.0019,
      "step": 75050
    },
    {
      "epoch": 1.2281763887752597,
      "grad_norm": 0.051891084760427475,
      "learning_rate": 7.356747561195013e-06,
      "loss": 0.0026,
      "step": 75060
    },
    {
      "epoch": 1.2283400147263357,
      "grad_norm": 0.059333398938179016,
      "learning_rate": 7.35590795826012e-06,
      "loss": 0.0019,
      "step": 75070
    },
    {
      "epoch": 1.2285036406774115,
      "grad_norm": 0.12364913523197174,
      "learning_rate": 7.3550682699292266e-06,
      "loss": 0.0024,
      "step": 75080
    },
    {
      "epoch": 1.2286672666284872,
      "grad_norm": 0.2238989621400833,
      "learning_rate": 7.35422849623277e-06,
      "loss": 0.0017,
      "step": 75090
    },
    {
      "epoch": 1.228830892579563,
      "grad_norm": 0.04785197973251343,
      "learning_rate": 7.353388637201186e-06,
      "loss": 0.0015,
      "step": 75100
    },
    {
      "epoch": 1.228994518530639,
      "grad_norm": 0.15885503590106964,
      "learning_rate": 7.352548692864923e-06,
      "loss": 0.0018,
      "step": 75110
    },
    {
      "epoch": 1.2291581444817148,
      "grad_norm": 0.05512414500117302,
      "learning_rate": 7.3517086632544225e-06,
      "loss": 0.0022,
      "step": 75120
    },
    {
      "epoch": 1.2293217704327906,
      "grad_norm": 0.08134091645479202,
      "learning_rate": 7.350868548400135e-06,
      "loss": 0.0031,
      "step": 75130
    },
    {
      "epoch": 1.2294853963838666,
      "grad_norm": 0.06621944159269333,
      "learning_rate": 7.350028348332513e-06,
      "loss": 0.0029,
      "step": 75140
    },
    {
      "epoch": 1.2296490223349423,
      "grad_norm": 0.07733746618032455,
      "learning_rate": 7.3491880630820126e-06,
      "loss": 0.0018,
      "step": 75150
    },
    {
      "epoch": 1.229812648286018,
      "grad_norm": 0.052341047674417496,
      "learning_rate": 7.348347692679089e-06,
      "loss": 0.0026,
      "step": 75160
    },
    {
      "epoch": 1.229976274237094,
      "grad_norm": 0.04901240020990372,
      "learning_rate": 7.347507237154208e-06,
      "loss": 0.0015,
      "step": 75170
    },
    {
      "epoch": 1.2301399001881699,
      "grad_norm": 0.19879671931266785,
      "learning_rate": 7.34666669653783e-06,
      "loss": 0.0018,
      "step": 75180
    },
    {
      "epoch": 1.2303035261392457,
      "grad_norm": 0.19959190487861633,
      "learning_rate": 7.345826070860425e-06,
      "loss": 0.0013,
      "step": 75190
    },
    {
      "epoch": 1.2304671520903216,
      "grad_norm": 0.2508465349674225,
      "learning_rate": 7.344985360152464e-06,
      "loss": 0.0023,
      "step": 75200
    },
    {
      "epoch": 1.2306307780413974,
      "grad_norm": 0.12279736250638962,
      "learning_rate": 7.344144564444418e-06,
      "loss": 0.0015,
      "step": 75210
    },
    {
      "epoch": 1.2307944039924732,
      "grad_norm": 0.04754587262868881,
      "learning_rate": 7.343303683766765e-06,
      "loss": 0.0025,
      "step": 75220
    },
    {
      "epoch": 1.230958029943549,
      "grad_norm": 0.0939144566655159,
      "learning_rate": 7.342462718149988e-06,
      "loss": 0.0023,
      "step": 75230
    },
    {
      "epoch": 1.231121655894625,
      "grad_norm": 0.07351992279291153,
      "learning_rate": 7.341621667624566e-06,
      "loss": 0.0013,
      "step": 75240
    },
    {
      "epoch": 1.2312852818457007,
      "grad_norm": 0.05443692207336426,
      "learning_rate": 7.340780532220986e-06,
      "loss": 0.0012,
      "step": 75250
    },
    {
      "epoch": 1.2314489077967765,
      "grad_norm": 0.25558826327323914,
      "learning_rate": 7.339939311969738e-06,
      "loss": 0.0033,
      "step": 75260
    },
    {
      "epoch": 1.2316125337478523,
      "grad_norm": 0.07000099122524261,
      "learning_rate": 7.339098006901314e-06,
      "loss": 0.0032,
      "step": 75270
    },
    {
      "epoch": 1.2317761596989283,
      "grad_norm": 0.11121514439582825,
      "learning_rate": 7.338256617046209e-06,
      "loss": 0.0021,
      "step": 75280
    },
    {
      "epoch": 1.231939785650004,
      "grad_norm": 0.18084879219532013,
      "learning_rate": 7.337415142434921e-06,
      "loss": 0.003,
      "step": 75290
    },
    {
      "epoch": 1.2321034116010798,
      "grad_norm": 0.06975261867046356,
      "learning_rate": 7.336573583097953e-06,
      "loss": 0.0017,
      "step": 75300
    },
    {
      "epoch": 1.2322670375521558,
      "grad_norm": 0.15908686816692352,
      "learning_rate": 7.3357319390658075e-06,
      "loss": 0.0013,
      "step": 75310
    },
    {
      "epoch": 1.2324306635032316,
      "grad_norm": 0.1834021806716919,
      "learning_rate": 7.334890210368994e-06,
      "loss": 0.0044,
      "step": 75320
    },
    {
      "epoch": 1.2325942894543074,
      "grad_norm": 0.1268003284931183,
      "learning_rate": 7.334048397038021e-06,
      "loss": 0.0023,
      "step": 75330
    },
    {
      "epoch": 1.2327579154053834,
      "grad_norm": 0.12502337992191315,
      "learning_rate": 7.333206499103404e-06,
      "loss": 0.0016,
      "step": 75340
    },
    {
      "epoch": 1.2329215413564591,
      "grad_norm": 0.36029180884361267,
      "learning_rate": 7.332364516595659e-06,
      "loss": 0.0021,
      "step": 75350
    },
    {
      "epoch": 1.233085167307535,
      "grad_norm": 0.04359174519777298,
      "learning_rate": 7.331522449545307e-06,
      "loss": 0.0013,
      "step": 75360
    },
    {
      "epoch": 1.233248793258611,
      "grad_norm": 0.1549016833305359,
      "learning_rate": 7.330680297982868e-06,
      "loss": 0.0019,
      "step": 75370
    },
    {
      "epoch": 1.2334124192096867,
      "grad_norm": 0.14654278755187988,
      "learning_rate": 7.3298380619388695e-06,
      "loss": 0.0016,
      "step": 75380
    },
    {
      "epoch": 1.2335760451607625,
      "grad_norm": 0.10543999820947647,
      "learning_rate": 7.328995741443843e-06,
      "loss": 0.0083,
      "step": 75390
    },
    {
      "epoch": 1.2337396711118382,
      "grad_norm": 0.1945066750049591,
      "learning_rate": 7.328153336528319e-06,
      "loss": 0.0018,
      "step": 75400
    },
    {
      "epoch": 1.2339032970629142,
      "grad_norm": 0.07839474827051163,
      "learning_rate": 7.327310847222831e-06,
      "loss": 0.0019,
      "step": 75410
    },
    {
      "epoch": 1.23406692301399,
      "grad_norm": 0.03426682949066162,
      "learning_rate": 7.3264682735579165e-06,
      "loss": 0.0014,
      "step": 75420
    },
    {
      "epoch": 1.2342305489650658,
      "grad_norm": 0.081778384745121,
      "learning_rate": 7.325625615564119e-06,
      "loss": 0.0014,
      "step": 75430
    },
    {
      "epoch": 1.2343941749161418,
      "grad_norm": 0.11724001169204712,
      "learning_rate": 7.324782873271984e-06,
      "loss": 0.0018,
      "step": 75440
    },
    {
      "epoch": 1.2345578008672176,
      "grad_norm": 0.04757022485136986,
      "learning_rate": 7.323940046712057e-06,
      "loss": 0.0014,
      "step": 75450
    },
    {
      "epoch": 1.2347214268182933,
      "grad_norm": 0.057816553860902786,
      "learning_rate": 7.323097135914889e-06,
      "loss": 0.0017,
      "step": 75460
    },
    {
      "epoch": 1.234885052769369,
      "grad_norm": 0.559280276298523,
      "learning_rate": 7.3222541409110325e-06,
      "loss": 0.0034,
      "step": 75470
    },
    {
      "epoch": 1.235048678720445,
      "grad_norm": 0.5130457878112793,
      "learning_rate": 7.3214110617310454e-06,
      "loss": 0.0026,
      "step": 75480
    },
    {
      "epoch": 1.2352123046715209,
      "grad_norm": 0.02552800253033638,
      "learning_rate": 7.320567898405487e-06,
      "loss": 0.0016,
      "step": 75490
    },
    {
      "epoch": 1.2353759306225967,
      "grad_norm": 0.04089820384979248,
      "learning_rate": 7.319724650964918e-06,
      "loss": 0.0025,
      "step": 75500
    },
    {
      "epoch": 1.2355395565736726,
      "grad_norm": 0.08133328706026077,
      "learning_rate": 7.318881319439907e-06,
      "loss": 0.003,
      "step": 75510
    },
    {
      "epoch": 1.2357031825247484,
      "grad_norm": 0.184864804148674,
      "learning_rate": 7.318037903861021e-06,
      "loss": 0.0032,
      "step": 75520
    },
    {
      "epoch": 1.2358668084758242,
      "grad_norm": 0.1661335974931717,
      "learning_rate": 7.317194404258832e-06,
      "loss": 0.002,
      "step": 75530
    },
    {
      "epoch": 1.2360304344269002,
      "grad_norm": 0.11695144325494766,
      "learning_rate": 7.316350820663915e-06,
      "loss": 0.0017,
      "step": 75540
    },
    {
      "epoch": 1.236194060377976,
      "grad_norm": 0.2725054621696472,
      "learning_rate": 7.31550715310685e-06,
      "loss": 0.0021,
      "step": 75550
    },
    {
      "epoch": 1.2363576863290517,
      "grad_norm": 0.22499892115592957,
      "learning_rate": 7.314663401618214e-06,
      "loss": 0.0015,
      "step": 75560
    },
    {
      "epoch": 1.2365213122801277,
      "grad_norm": 0.07645678520202637,
      "learning_rate": 7.313819566228592e-06,
      "loss": 0.0017,
      "step": 75570
    },
    {
      "epoch": 1.2366849382312035,
      "grad_norm": 0.11282423883676529,
      "learning_rate": 7.312975646968573e-06,
      "loss": 0.0026,
      "step": 75580
    },
    {
      "epoch": 1.2368485641822793,
      "grad_norm": 0.03659214451909065,
      "learning_rate": 7.312131643868746e-06,
      "loss": 0.0019,
      "step": 75590
    },
    {
      "epoch": 1.237012190133355,
      "grad_norm": 0.2823604345321655,
      "learning_rate": 7.311287556959705e-06,
      "loss": 0.0018,
      "step": 75600
    },
    {
      "epoch": 1.237175816084431,
      "grad_norm": 0.012864679098129272,
      "learning_rate": 7.310443386272043e-06,
      "loss": 0.0012,
      "step": 75610
    },
    {
      "epoch": 1.2373394420355068,
      "grad_norm": 0.16263914108276367,
      "learning_rate": 7.309599131836363e-06,
      "loss": 0.002,
      "step": 75620
    },
    {
      "epoch": 1.2375030679865826,
      "grad_norm": 0.18927639722824097,
      "learning_rate": 7.308754793683265e-06,
      "loss": 0.002,
      "step": 75630
    },
    {
      "epoch": 1.2376666939376586,
      "grad_norm": 0.07541917264461517,
      "learning_rate": 7.307910371843354e-06,
      "loss": 0.0017,
      "step": 75640
    },
    {
      "epoch": 1.2378303198887344,
      "grad_norm": 0.06526405364274979,
      "learning_rate": 7.307065866347238e-06,
      "loss": 0.002,
      "step": 75650
    },
    {
      "epoch": 1.2379939458398102,
      "grad_norm": 0.22764533758163452,
      "learning_rate": 7.3062212772255314e-06,
      "loss": 0.0041,
      "step": 75660
    },
    {
      "epoch": 1.238157571790886,
      "grad_norm": 0.06345629692077637,
      "learning_rate": 7.305376604508845e-06,
      "loss": 0.0015,
      "step": 75670
    },
    {
      "epoch": 1.238321197741962,
      "grad_norm": 0.05479049310088158,
      "learning_rate": 7.304531848227798e-06,
      "loss": 0.0017,
      "step": 75680
    },
    {
      "epoch": 1.2384848236930377,
      "grad_norm": 0.04212285950779915,
      "learning_rate": 7.303687008413011e-06,
      "loss": 0.0015,
      "step": 75690
    },
    {
      "epoch": 1.2386484496441135,
      "grad_norm": 0.21634085476398468,
      "learning_rate": 7.302842085095106e-06,
      "loss": 0.0029,
      "step": 75700
    },
    {
      "epoch": 1.2388120755951895,
      "grad_norm": 0.06691079586744308,
      "learning_rate": 7.30199707830471e-06,
      "loss": 0.002,
      "step": 75710
    },
    {
      "epoch": 1.2389757015462652,
      "grad_norm": 0.061119671911001205,
      "learning_rate": 7.301151988072452e-06,
      "loss": 0.0013,
      "step": 75720
    },
    {
      "epoch": 1.239139327497341,
      "grad_norm": 0.1389298290014267,
      "learning_rate": 7.300306814428966e-06,
      "loss": 0.0022,
      "step": 75730
    },
    {
      "epoch": 1.239302953448417,
      "grad_norm": 0.15854953229427338,
      "learning_rate": 7.299461557404886e-06,
      "loss": 0.0017,
      "step": 75740
    },
    {
      "epoch": 1.2394665793994928,
      "grad_norm": 0.02639460749924183,
      "learning_rate": 7.298616217030853e-06,
      "loss": 0.0019,
      "step": 75750
    },
    {
      "epoch": 1.2396302053505686,
      "grad_norm": 0.26457294821739197,
      "learning_rate": 7.2977707933375054e-06,
      "loss": 0.0025,
      "step": 75760
    },
    {
      "epoch": 1.2397938313016446,
      "grad_norm": 0.15434350073337555,
      "learning_rate": 7.296925286355489e-06,
      "loss": 0.0032,
      "step": 75770
    },
    {
      "epoch": 1.2399574572527203,
      "grad_norm": 0.1481945812702179,
      "learning_rate": 7.296079696115451e-06,
      "loss": 0.003,
      "step": 75780
    },
    {
      "epoch": 1.240121083203796,
      "grad_norm": 0.13540132343769073,
      "learning_rate": 7.295234022648042e-06,
      "loss": 0.0028,
      "step": 75790
    },
    {
      "epoch": 1.2402847091548719,
      "grad_norm": 0.1139894425868988,
      "learning_rate": 7.294388265983919e-06,
      "loss": 0.0016,
      "step": 75800
    },
    {
      "epoch": 1.2404483351059479,
      "grad_norm": 0.422225683927536,
      "learning_rate": 7.293542426153732e-06,
      "loss": 0.0024,
      "step": 75810
    },
    {
      "epoch": 1.2406119610570236,
      "grad_norm": 0.08003806322813034,
      "learning_rate": 7.2926965031881465e-06,
      "loss": 0.0015,
      "step": 75820
    },
    {
      "epoch": 1.2407755870080994,
      "grad_norm": 0.20687858760356903,
      "learning_rate": 7.291850497117824e-06,
      "loss": 0.0023,
      "step": 75830
    },
    {
      "epoch": 1.2409392129591754,
      "grad_norm": 0.08286328613758087,
      "learning_rate": 7.291004407973428e-06,
      "loss": 0.002,
      "step": 75840
    },
    {
      "epoch": 1.2411028389102512,
      "grad_norm": 0.0753655657172203,
      "learning_rate": 7.2901582357856285e-06,
      "loss": 0.0016,
      "step": 75850
    },
    {
      "epoch": 1.241266464861327,
      "grad_norm": 0.22551557421684265,
      "learning_rate": 7.289311980585098e-06,
      "loss": 0.0014,
      "step": 75860
    },
    {
      "epoch": 1.2414300908124027,
      "grad_norm": 0.08038531988859177,
      "learning_rate": 7.2884656424025095e-06,
      "loss": 0.0018,
      "step": 75870
    },
    {
      "epoch": 1.2415937167634787,
      "grad_norm": 0.16861200332641602,
      "learning_rate": 7.287619221268541e-06,
      "loss": 0.0055,
      "step": 75880
    },
    {
      "epoch": 1.2417573427145545,
      "grad_norm": 0.10394301265478134,
      "learning_rate": 7.286772717213874e-06,
      "loss": 0.003,
      "step": 75890
    },
    {
      "epoch": 1.2419209686656303,
      "grad_norm": 0.09122875332832336,
      "learning_rate": 7.285926130269195e-06,
      "loss": 0.0046,
      "step": 75900
    },
    {
      "epoch": 1.2420845946167063,
      "grad_norm": 0.2115568220615387,
      "learning_rate": 7.285079460465185e-06,
      "loss": 0.0023,
      "step": 75910
    },
    {
      "epoch": 1.242248220567782,
      "grad_norm": 0.1815410554409027,
      "learning_rate": 7.284232707832538e-06,
      "loss": 0.0027,
      "step": 75920
    },
    {
      "epoch": 1.2424118465188578,
      "grad_norm": 0.15625382959842682,
      "learning_rate": 7.283385872401944e-06,
      "loss": 0.0018,
      "step": 75930
    },
    {
      "epoch": 1.2425754724699338,
      "grad_norm": 0.058230139315128326,
      "learning_rate": 7.282538954204102e-06,
      "loss": 0.0018,
      "step": 75940
    },
    {
      "epoch": 1.2427390984210096,
      "grad_norm": 0.037460390478372574,
      "learning_rate": 7.281691953269707e-06,
      "loss": 0.0066,
      "step": 75950
    },
    {
      "epoch": 1.2429027243720854,
      "grad_norm": 0.020542221143841743,
      "learning_rate": 7.280844869629464e-06,
      "loss": 0.0015,
      "step": 75960
    },
    {
      "epoch": 1.2430663503231614,
      "grad_norm": 0.09572634845972061,
      "learning_rate": 7.279997703314076e-06,
      "loss": 0.0014,
      "step": 75970
    },
    {
      "epoch": 1.2432299762742371,
      "grad_norm": 0.030780604109168053,
      "learning_rate": 7.27915045435425e-06,
      "loss": 0.0026,
      "step": 75980
    },
    {
      "epoch": 1.243393602225313,
      "grad_norm": 0.36935174465179443,
      "learning_rate": 7.278303122780699e-06,
      "loss": 0.0027,
      "step": 75990
    },
    {
      "epoch": 1.2435572281763887,
      "grad_norm": 0.08313564956188202,
      "learning_rate": 7.277455708624135e-06,
      "loss": 0.0017,
      "step": 76000
    },
    {
      "epoch": 1.2437208541274647,
      "grad_norm": 0.03930523991584778,
      "learning_rate": 7.276608211915275e-06,
      "loss": 0.0016,
      "step": 76010
    },
    {
      "epoch": 1.2438844800785405,
      "grad_norm": 0.037010665982961655,
      "learning_rate": 7.27576063268484e-06,
      "loss": 0.002,
      "step": 76020
    },
    {
      "epoch": 1.2440481060296162,
      "grad_norm": 0.028221316635608673,
      "learning_rate": 7.27491297096355e-06,
      "loss": 0.0018,
      "step": 76030
    },
    {
      "epoch": 1.244211731980692,
      "grad_norm": 0.01731671579182148,
      "learning_rate": 7.274065226782134e-06,
      "loss": 0.0023,
      "step": 76040
    },
    {
      "epoch": 1.244375357931768,
      "grad_norm": 0.15924237668514252,
      "learning_rate": 7.27321740017132e-06,
      "loss": 0.0023,
      "step": 76050
    },
    {
      "epoch": 1.2445389838828438,
      "grad_norm": 0.06417131423950195,
      "learning_rate": 7.272369491161837e-06,
      "loss": 0.0017,
      "step": 76060
    },
    {
      "epoch": 1.2447026098339196,
      "grad_norm": 0.07908331602811813,
      "learning_rate": 7.271521499784421e-06,
      "loss": 0.002,
      "step": 76070
    },
    {
      "epoch": 1.2448662357849956,
      "grad_norm": 0.15662676095962524,
      "learning_rate": 7.270673426069811e-06,
      "loss": 0.0021,
      "step": 76080
    },
    {
      "epoch": 1.2450298617360713,
      "grad_norm": 0.026165012270212173,
      "learning_rate": 7.269825270048747e-06,
      "loss": 0.0021,
      "step": 76090
    },
    {
      "epoch": 1.245193487687147,
      "grad_norm": 0.19637195765972137,
      "learning_rate": 7.2689770317519726e-06,
      "loss": 0.002,
      "step": 76100
    },
    {
      "epoch": 1.245357113638223,
      "grad_norm": 0.1039731353521347,
      "learning_rate": 7.268128711210234e-06,
      "loss": 0.0025,
      "step": 76110
    },
    {
      "epoch": 1.2455207395892989,
      "grad_norm": 0.10456263273954391,
      "learning_rate": 7.26728030845428e-06,
      "loss": 0.0017,
      "step": 76120
    },
    {
      "epoch": 1.2456843655403746,
      "grad_norm": 0.17909160256385803,
      "learning_rate": 7.266431823514864e-06,
      "loss": 0.0013,
      "step": 76130
    },
    {
      "epoch": 1.2458479914914506,
      "grad_norm": 0.1260414570569992,
      "learning_rate": 7.265583256422742e-06,
      "loss": 0.0016,
      "step": 76140
    },
    {
      "epoch": 1.2460116174425264,
      "grad_norm": 0.12704215943813324,
      "learning_rate": 7.264734607208673e-06,
      "loss": 0.0015,
      "step": 76150
    },
    {
      "epoch": 1.2461752433936022,
      "grad_norm": 0.0274752639234066,
      "learning_rate": 7.263885875903417e-06,
      "loss": 0.0011,
      "step": 76160
    },
    {
      "epoch": 1.2463388693446782,
      "grad_norm": 0.21337048709392548,
      "learning_rate": 7.263037062537739e-06,
      "loss": 0.0023,
      "step": 76170
    },
    {
      "epoch": 1.246502495295754,
      "grad_norm": 0.11034653335809708,
      "learning_rate": 7.262188167142406e-06,
      "loss": 0.0025,
      "step": 76180
    },
    {
      "epoch": 1.2466661212468297,
      "grad_norm": 0.06820209324359894,
      "learning_rate": 7.26133918974819e-06,
      "loss": 0.0041,
      "step": 76190
    },
    {
      "epoch": 1.2468297471979055,
      "grad_norm": 0.1598266065120697,
      "learning_rate": 7.260490130385863e-06,
      "loss": 0.002,
      "step": 76200
    },
    {
      "epoch": 1.2469933731489815,
      "grad_norm": 0.11024352163076401,
      "learning_rate": 7.259640989086202e-06,
      "loss": 0.002,
      "step": 76210
    },
    {
      "epoch": 1.2471569991000573,
      "grad_norm": 0.12314579635858536,
      "learning_rate": 7.2587917658799854e-06,
      "loss": 0.0014,
      "step": 76220
    },
    {
      "epoch": 1.247320625051133,
      "grad_norm": 0.14845508337020874,
      "learning_rate": 7.257942460797996e-06,
      "loss": 0.0017,
      "step": 76230
    },
    {
      "epoch": 1.2474842510022088,
      "grad_norm": 0.11476452648639679,
      "learning_rate": 7.25709307387102e-06,
      "loss": 0.0014,
      "step": 76240
    },
    {
      "epoch": 1.2476478769532848,
      "grad_norm": 0.0909375324845314,
      "learning_rate": 7.256243605129846e-06,
      "loss": 0.0035,
      "step": 76250
    },
    {
      "epoch": 1.2478115029043606,
      "grad_norm": 0.1726021021604538,
      "learning_rate": 7.2553940546052625e-06,
      "loss": 0.0021,
      "step": 76260
    },
    {
      "epoch": 1.2479751288554364,
      "grad_norm": 0.012697853147983551,
      "learning_rate": 7.2545444223280645e-06,
      "loss": 0.0033,
      "step": 76270
    },
    {
      "epoch": 1.2481387548065124,
      "grad_norm": 0.2170550674200058,
      "learning_rate": 7.253694708329051e-06,
      "loss": 0.0017,
      "step": 76280
    },
    {
      "epoch": 1.2483023807575881,
      "grad_norm": 0.09155912697315216,
      "learning_rate": 7.252844912639019e-06,
      "loss": 0.0024,
      "step": 76290
    },
    {
      "epoch": 1.248466006708664,
      "grad_norm": 0.21381379663944244,
      "learning_rate": 7.251995035288777e-06,
      "loss": 0.0023,
      "step": 76300
    },
    {
      "epoch": 1.24862963265974,
      "grad_norm": 0.13857270777225494,
      "learning_rate": 7.251145076309125e-06,
      "loss": 0.0028,
      "step": 76310
    },
    {
      "epoch": 1.2487932586108157,
      "grad_norm": 0.12481576949357986,
      "learning_rate": 7.250295035730874e-06,
      "loss": 0.0036,
      "step": 76320
    },
    {
      "epoch": 1.2489568845618915,
      "grad_norm": 0.03987983986735344,
      "learning_rate": 7.249444913584839e-06,
      "loss": 0.0024,
      "step": 76330
    },
    {
      "epoch": 1.2491205105129675,
      "grad_norm": 0.10182177275419235,
      "learning_rate": 7.248594709901831e-06,
      "loss": 0.0017,
      "step": 76340
    },
    {
      "epoch": 1.2492841364640432,
      "grad_norm": 0.08626046031713486,
      "learning_rate": 7.247744424712668e-06,
      "loss": 0.002,
      "step": 76350
    },
    {
      "epoch": 1.249447762415119,
      "grad_norm": 0.05633806809782982,
      "learning_rate": 7.246894058048173e-06,
      "loss": 0.0017,
      "step": 76360
    },
    {
      "epoch": 1.2496113883661948,
      "grad_norm": 0.09064376354217529,
      "learning_rate": 7.246043609939169e-06,
      "loss": 0.0019,
      "step": 76370
    },
    {
      "epoch": 1.2497750143172708,
      "grad_norm": 0.19761058688163757,
      "learning_rate": 7.24519308041648e-06,
      "loss": 0.0042,
      "step": 76380
    },
    {
      "epoch": 1.2499386402683466,
      "grad_norm": 0.06989564746618271,
      "learning_rate": 7.244342469510941e-06,
      "loss": 0.0015,
      "step": 76390
    },
    {
      "epoch": 1.2501022662194223,
      "grad_norm": 0.2068031132221222,
      "learning_rate": 7.243491777253381e-06,
      "loss": 0.0017,
      "step": 76400
    },
    {
      "epoch": 1.250265892170498,
      "grad_norm": 0.21270398795604706,
      "learning_rate": 7.242641003674635e-06,
      "loss": 0.0017,
      "step": 76410
    },
    {
      "epoch": 1.250429518121574,
      "grad_norm": 0.24101294577121735,
      "learning_rate": 7.241790148805543e-06,
      "loss": 0.0016,
      "step": 76420
    },
    {
      "epoch": 1.2505931440726499,
      "grad_norm": 0.07772132754325867,
      "learning_rate": 7.240939212676946e-06,
      "loss": 0.0016,
      "step": 76430
    },
    {
      "epoch": 1.2507567700237257,
      "grad_norm": 0.09166130423545837,
      "learning_rate": 7.240088195319689e-06,
      "loss": 0.0031,
      "step": 76440
    },
    {
      "epoch": 1.2509203959748016,
      "grad_norm": 0.08283478766679764,
      "learning_rate": 7.23923709676462e-06,
      "loss": 0.0018,
      "step": 76450
    },
    {
      "epoch": 1.2510840219258774,
      "grad_norm": 0.09896039962768555,
      "learning_rate": 7.238385917042586e-06,
      "loss": 0.0011,
      "step": 76460
    },
    {
      "epoch": 1.2512476478769532,
      "grad_norm": 0.07560428977012634,
      "learning_rate": 7.237534656184443e-06,
      "loss": 0.0011,
      "step": 76470
    },
    {
      "epoch": 1.2514112738280292,
      "grad_norm": 0.1504267007112503,
      "learning_rate": 7.236683314221047e-06,
      "loss": 0.0018,
      "step": 76480
    },
    {
      "epoch": 1.251574899779105,
      "grad_norm": 0.09247831255197525,
      "learning_rate": 7.2358318911832556e-06,
      "loss": 0.002,
      "step": 76490
    },
    {
      "epoch": 1.2517385257301807,
      "grad_norm": 0.023884030058979988,
      "learning_rate": 7.234980387101933e-06,
      "loss": 0.0013,
      "step": 76500
    },
    {
      "epoch": 1.2519021516812567,
      "grad_norm": 0.009047040715813637,
      "learning_rate": 7.234128802007943e-06,
      "loss": 0.0009,
      "step": 76510
    },
    {
      "epoch": 1.2520657776323325,
      "grad_norm": 0.1714857965707779,
      "learning_rate": 7.233277135932152e-06,
      "loss": 0.0023,
      "step": 76520
    },
    {
      "epoch": 1.2522294035834083,
      "grad_norm": 0.20701821148395538,
      "learning_rate": 7.232425388905433e-06,
      "loss": 0.0019,
      "step": 76530
    },
    {
      "epoch": 1.2523930295344843,
      "grad_norm": 0.1728150099515915,
      "learning_rate": 7.23157356095866e-06,
      "loss": 0.0024,
      "step": 76540
    },
    {
      "epoch": 1.25255665548556,
      "grad_norm": 0.12822218239307404,
      "learning_rate": 7.230721652122708e-06,
      "loss": 0.0021,
      "step": 76550
    },
    {
      "epoch": 1.2527202814366358,
      "grad_norm": 0.09107663482427597,
      "learning_rate": 7.229869662428457e-06,
      "loss": 0.0021,
      "step": 76560
    },
    {
      "epoch": 1.2528839073877118,
      "grad_norm": 0.1213102787733078,
      "learning_rate": 7.229017591906792e-06,
      "loss": 0.002,
      "step": 76570
    },
    {
      "epoch": 1.2530475333387876,
      "grad_norm": 0.10051631927490234,
      "learning_rate": 7.228165440588594e-06,
      "loss": 0.0015,
      "step": 76580
    },
    {
      "epoch": 1.2532111592898634,
      "grad_norm": 0.21822141110897064,
      "learning_rate": 7.227313208504756e-06,
      "loss": 0.0018,
      "step": 76590
    },
    {
      "epoch": 1.2533747852409391,
      "grad_norm": 0.37389299273490906,
      "learning_rate": 7.226460895686168e-06,
      "loss": 0.0026,
      "step": 76600
    },
    {
      "epoch": 1.253538411192015,
      "grad_norm": 0.20237046480178833,
      "learning_rate": 7.225608502163723e-06,
      "loss": 0.0024,
      "step": 76610
    },
    {
      "epoch": 1.253702037143091,
      "grad_norm": 0.17468029260635376,
      "learning_rate": 7.2247560279683184e-06,
      "loss": 0.0029,
      "step": 76620
    },
    {
      "epoch": 1.2538656630941667,
      "grad_norm": 0.09932902455329895,
      "learning_rate": 7.2239034731308545e-06,
      "loss": 0.0022,
      "step": 76630
    },
    {
      "epoch": 1.2540292890452425,
      "grad_norm": 0.18450143933296204,
      "learning_rate": 7.223050837682235e-06,
      "loss": 0.0032,
      "step": 76640
    },
    {
      "epoch": 1.2541929149963185,
      "grad_norm": 0.1467326283454895,
      "learning_rate": 7.222198121653365e-06,
      "loss": 0.0028,
      "step": 76650
    },
    {
      "epoch": 1.2543565409473942,
      "grad_norm": 0.13686597347259521,
      "learning_rate": 7.221345325075155e-06,
      "loss": 0.0021,
      "step": 76660
    },
    {
      "epoch": 1.25452016689847,
      "grad_norm": 0.17153418064117432,
      "learning_rate": 7.220492447978516e-06,
      "loss": 0.003,
      "step": 76670
    },
    {
      "epoch": 1.254683792849546,
      "grad_norm": 0.23260289430618286,
      "learning_rate": 7.219639490394363e-06,
      "loss": 0.0035,
      "step": 76680
    },
    {
      "epoch": 1.2548474188006218,
      "grad_norm": 0.05313664302229881,
      "learning_rate": 7.218786452353612e-06,
      "loss": 0.0018,
      "step": 76690
    },
    {
      "epoch": 1.2550110447516976,
      "grad_norm": 0.17395347356796265,
      "learning_rate": 7.2179333338871845e-06,
      "loss": 0.0021,
      "step": 76700
    },
    {
      "epoch": 1.2551746707027736,
      "grad_norm": 0.15894368290901184,
      "learning_rate": 7.217080135026006e-06,
      "loss": 0.0017,
      "step": 76710
    },
    {
      "epoch": 1.2553382966538493,
      "grad_norm": 0.017329467460513115,
      "learning_rate": 7.216226855800999e-06,
      "loss": 0.003,
      "step": 76720
    },
    {
      "epoch": 1.255501922604925,
      "grad_norm": 0.1391407549381256,
      "learning_rate": 7.215373496243097e-06,
      "loss": 0.0024,
      "step": 76730
    },
    {
      "epoch": 1.255665548556001,
      "grad_norm": 0.190855011343956,
      "learning_rate": 7.21452005638323e-06,
      "loss": 0.0024,
      "step": 76740
    },
    {
      "epoch": 1.2558291745070769,
      "grad_norm": 0.09568745642900467,
      "learning_rate": 7.213666536252333e-06,
      "loss": 0.0033,
      "step": 76750
    },
    {
      "epoch": 1.2559928004581526,
      "grad_norm": 0.06416328251361847,
      "learning_rate": 7.212812935881345e-06,
      "loss": 0.0019,
      "step": 76760
    },
    {
      "epoch": 1.2561564264092284,
      "grad_norm": 0.09854545444250107,
      "learning_rate": 7.2119592553012064e-06,
      "loss": 0.004,
      "step": 76770
    },
    {
      "epoch": 1.2563200523603044,
      "grad_norm": 0.2991683781147003,
      "learning_rate": 7.2111054945428584e-06,
      "loss": 0.0017,
      "step": 76780
    },
    {
      "epoch": 1.2564836783113802,
      "grad_norm": 0.06790527701377869,
      "learning_rate": 7.210251653637253e-06,
      "loss": 0.0022,
      "step": 76790
    },
    {
      "epoch": 1.256647304262456,
      "grad_norm": 0.13237342238426208,
      "learning_rate": 7.209397732615338e-06,
      "loss": 0.0023,
      "step": 76800
    },
    {
      "epoch": 1.2568109302135317,
      "grad_norm": 0.050308067351579666,
      "learning_rate": 7.208543731508064e-06,
      "loss": 0.0019,
      "step": 76810
    },
    {
      "epoch": 1.2569745561646077,
      "grad_norm": 0.04378840699791908,
      "learning_rate": 7.207689650346388e-06,
      "loss": 0.0012,
      "step": 76820
    },
    {
      "epoch": 1.2571381821156835,
      "grad_norm": 0.07707502692937851,
      "learning_rate": 7.206835489161269e-06,
      "loss": 0.0015,
      "step": 76830
    },
    {
      "epoch": 1.2573018080667593,
      "grad_norm": 0.10681436210870743,
      "learning_rate": 7.2059812479836665e-06,
      "loss": 0.0026,
      "step": 76840
    },
    {
      "epoch": 1.2574654340178353,
      "grad_norm": 0.05072954297065735,
      "learning_rate": 7.2051269268445455e-06,
      "loss": 0.0026,
      "step": 76850
    },
    {
      "epoch": 1.257629059968911,
      "grad_norm": 0.04281345382332802,
      "learning_rate": 7.204272525774875e-06,
      "loss": 0.0047,
      "step": 76860
    },
    {
      "epoch": 1.2577926859199868,
      "grad_norm": 0.04866739735007286,
      "learning_rate": 7.203418044805621e-06,
      "loss": 0.0017,
      "step": 76870
    },
    {
      "epoch": 1.2579563118710628,
      "grad_norm": 0.11990425735712051,
      "learning_rate": 7.202563483967761e-06,
      "loss": 0.0024,
      "step": 76880
    },
    {
      "epoch": 1.2581199378221386,
      "grad_norm": 0.08831249922513962,
      "learning_rate": 7.201708843292266e-06,
      "loss": 0.0023,
      "step": 76890
    },
    {
      "epoch": 1.2582835637732144,
      "grad_norm": 0.11786916851997375,
      "learning_rate": 7.2008541228101194e-06,
      "loss": 0.0031,
      "step": 76900
    },
    {
      "epoch": 1.2584471897242904,
      "grad_norm": 0.2058493047952652,
      "learning_rate": 7.1999993225523e-06,
      "loss": 0.0018,
      "step": 76910
    },
    {
      "epoch": 1.2586108156753661,
      "grad_norm": 0.17707034945487976,
      "learning_rate": 7.19914444254979e-06,
      "loss": 0.0013,
      "step": 76920
    },
    {
      "epoch": 1.258774441626442,
      "grad_norm": 0.09533452242612839,
      "learning_rate": 7.198289482833583e-06,
      "loss": 0.0026,
      "step": 76930
    },
    {
      "epoch": 1.258938067577518,
      "grad_norm": 0.16579443216323853,
      "learning_rate": 7.197434443434663e-06,
      "loss": 0.0023,
      "step": 76940
    },
    {
      "epoch": 1.2591016935285937,
      "grad_norm": 0.20056231319904327,
      "learning_rate": 7.1965793243840284e-06,
      "loss": 0.0029,
      "step": 76950
    },
    {
      "epoch": 1.2592653194796695,
      "grad_norm": 0.040070660412311554,
      "learning_rate": 7.195724125712671e-06,
      "loss": 0.0014,
      "step": 76960
    },
    {
      "epoch": 1.2594289454307452,
      "grad_norm": 0.08155545592308044,
      "learning_rate": 7.194868847451593e-06,
      "loss": 0.0019,
      "step": 76970
    },
    {
      "epoch": 1.2595925713818212,
      "grad_norm": 0.3704371750354767,
      "learning_rate": 7.194013489631793e-06,
      "loss": 0.0017,
      "step": 76980
    },
    {
      "epoch": 1.259756197332897,
      "grad_norm": 0.0917673259973526,
      "learning_rate": 7.193158052284277e-06,
      "loss": 0.0039,
      "step": 76990
    },
    {
      "epoch": 1.2599198232839728,
      "grad_norm": 0.14519546926021576,
      "learning_rate": 7.192302535440052e-06,
      "loss": 0.0038,
      "step": 77000
    },
    {
      "epoch": 1.2600834492350486,
      "grad_norm": 0.29657483100891113,
      "learning_rate": 7.191446939130131e-06,
      "loss": 0.0014,
      "step": 77010
    },
    {
      "epoch": 1.2602470751861246,
      "grad_norm": 0.25027918815612793,
      "learning_rate": 7.190591263385525e-06,
      "loss": 0.0021,
      "step": 77020
    },
    {
      "epoch": 1.2604107011372003,
      "grad_norm": 0.1431395560503006,
      "learning_rate": 7.18973550823725e-06,
      "loss": 0.0026,
      "step": 77030
    },
    {
      "epoch": 1.260574327088276,
      "grad_norm": 0.1428649127483368,
      "learning_rate": 7.188879673716326e-06,
      "loss": 0.0023,
      "step": 77040
    },
    {
      "epoch": 1.260737953039352,
      "grad_norm": 0.07966932654380798,
      "learning_rate": 7.188023759853775e-06,
      "loss": 0.0028,
      "step": 77050
    },
    {
      "epoch": 1.2609015789904279,
      "grad_norm": 0.11174963414669037,
      "learning_rate": 7.18716776668062e-06,
      "loss": 0.0025,
      "step": 77060
    },
    {
      "epoch": 1.2610652049415036,
      "grad_norm": 0.18909013271331787,
      "learning_rate": 7.186311694227892e-06,
      "loss": 0.0021,
      "step": 77070
    },
    {
      "epoch": 1.2612288308925796,
      "grad_norm": 0.14445863664150238,
      "learning_rate": 7.185455542526618e-06,
      "loss": 0.002,
      "step": 77080
    },
    {
      "epoch": 1.2613924568436554,
      "grad_norm": 0.01052671018987894,
      "learning_rate": 7.184599311607835e-06,
      "loss": 0.0017,
      "step": 77090
    },
    {
      "epoch": 1.2615560827947312,
      "grad_norm": 0.034729667007923126,
      "learning_rate": 7.183743001502578e-06,
      "loss": 0.0015,
      "step": 77100
    },
    {
      "epoch": 1.2617197087458072,
      "grad_norm": 0.19659213721752167,
      "learning_rate": 7.182886612241884e-06,
      "loss": 0.0035,
      "step": 77110
    },
    {
      "epoch": 1.261883334696883,
      "grad_norm": 0.18396835029125214,
      "learning_rate": 7.182030143856797e-06,
      "loss": 0.0025,
      "step": 77120
    },
    {
      "epoch": 1.2620469606479587,
      "grad_norm": 0.10773087292909622,
      "learning_rate": 7.181173596378362e-06,
      "loss": 0.0028,
      "step": 77130
    },
    {
      "epoch": 1.2622105865990347,
      "grad_norm": 0.027716603130102158,
      "learning_rate": 7.1803169698376255e-06,
      "loss": 0.0025,
      "step": 77140
    },
    {
      "epoch": 1.2623742125501105,
      "grad_norm": 0.08923885226249695,
      "learning_rate": 7.179460264265641e-06,
      "loss": 0.0019,
      "step": 77150
    },
    {
      "epoch": 1.2625378385011863,
      "grad_norm": 0.2653767466545105,
      "learning_rate": 7.178603479693459e-06,
      "loss": 0.0026,
      "step": 77160
    },
    {
      "epoch": 1.262701464452262,
      "grad_norm": 0.04988548159599304,
      "learning_rate": 7.177746616152137e-06,
      "loss": 0.0013,
      "step": 77170
    },
    {
      "epoch": 1.2628650904033378,
      "grad_norm": 0.1218423992395401,
      "learning_rate": 7.176889673672733e-06,
      "loss": 0.0025,
      "step": 77180
    },
    {
      "epoch": 1.2630287163544138,
      "grad_norm": 0.20130939781665802,
      "learning_rate": 7.176032652286312e-06,
      "loss": 0.0015,
      "step": 77190
    },
    {
      "epoch": 1.2631923423054896,
      "grad_norm": 0.15818671882152557,
      "learning_rate": 7.175175552023938e-06,
      "loss": 0.0047,
      "step": 77200
    },
    {
      "epoch": 1.2633559682565654,
      "grad_norm": 0.22315305471420288,
      "learning_rate": 7.174318372916676e-06,
      "loss": 0.0016,
      "step": 77210
    },
    {
      "epoch": 1.2635195942076414,
      "grad_norm": 0.09150251746177673,
      "learning_rate": 7.1734611149956e-06,
      "loss": 0.0027,
      "step": 77220
    },
    {
      "epoch": 1.2636832201587171,
      "grad_norm": 0.15513551235198975,
      "learning_rate": 7.172603778291782e-06,
      "loss": 0.0038,
      "step": 77230
    },
    {
      "epoch": 1.263846846109793,
      "grad_norm": 0.1001538634300232,
      "learning_rate": 7.171746362836299e-06,
      "loss": 0.0023,
      "step": 77240
    },
    {
      "epoch": 1.264010472060869,
      "grad_norm": 0.13093596696853638,
      "learning_rate": 7.170888868660231e-06,
      "loss": 0.0016,
      "step": 77250
    },
    {
      "epoch": 1.2641740980119447,
      "grad_norm": 0.2395472675561905,
      "learning_rate": 7.170031295794658e-06,
      "loss": 0.0033,
      "step": 77260
    },
    {
      "epoch": 1.2643377239630205,
      "grad_norm": 0.16735276579856873,
      "learning_rate": 7.169173644270666e-06,
      "loss": 0.0027,
      "step": 77270
    },
    {
      "epoch": 1.2645013499140965,
      "grad_norm": 0.10939212143421173,
      "learning_rate": 7.168315914119342e-06,
      "loss": 0.0012,
      "step": 77280
    },
    {
      "epoch": 1.2646649758651722,
      "grad_norm": 0.12859097123146057,
      "learning_rate": 7.167458105371778e-06,
      "loss": 0.0012,
      "step": 77290
    },
    {
      "epoch": 1.264828601816248,
      "grad_norm": 0.26028716564178467,
      "learning_rate": 7.1666002180590686e-06,
      "loss": 0.0083,
      "step": 77300
    },
    {
      "epoch": 1.264992227767324,
      "grad_norm": 0.0801466554403305,
      "learning_rate": 7.165742252212307e-06,
      "loss": 0.0017,
      "step": 77310
    },
    {
      "epoch": 1.2651558537183998,
      "grad_norm": 0.013302541337907314,
      "learning_rate": 7.164884207862596e-06,
      "loss": 0.0017,
      "step": 77320
    },
    {
      "epoch": 1.2653194796694756,
      "grad_norm": 0.26006826758384705,
      "learning_rate": 7.164026085041034e-06,
      "loss": 0.0021,
      "step": 77330
    },
    {
      "epoch": 1.2654831056205516,
      "grad_norm": 0.12897484004497528,
      "learning_rate": 7.163167883778728e-06,
      "loss": 0.0015,
      "step": 77340
    },
    {
      "epoch": 1.2656467315716273,
      "grad_norm": 0.04795488342642784,
      "learning_rate": 7.162309604106784e-06,
      "loss": 0.0013,
      "step": 77350
    },
    {
      "epoch": 1.265810357522703,
      "grad_norm": 0.19672997295856476,
      "learning_rate": 7.1614512460563145e-06,
      "loss": 0.0022,
      "step": 77360
    },
    {
      "epoch": 1.2659739834737789,
      "grad_norm": 0.07301844656467438,
      "learning_rate": 7.160592809658433e-06,
      "loss": 0.0024,
      "step": 77370
    },
    {
      "epoch": 1.2661376094248546,
      "grad_norm": 0.03807608783245087,
      "learning_rate": 7.159734294944254e-06,
      "loss": 0.0043,
      "step": 77380
    },
    {
      "epoch": 1.2663012353759306,
      "grad_norm": 0.196418896317482,
      "learning_rate": 7.158875701944898e-06,
      "loss": 0.0026,
      "step": 77390
    },
    {
      "epoch": 1.2664648613270064,
      "grad_norm": 0.37065741419792175,
      "learning_rate": 7.158017030691487e-06,
      "loss": 0.0024,
      "step": 77400
    },
    {
      "epoch": 1.2666284872780822,
      "grad_norm": 0.25617995858192444,
      "learning_rate": 7.157158281215144e-06,
      "loss": 0.0026,
      "step": 77410
    },
    {
      "epoch": 1.2667921132291582,
      "grad_norm": 0.01631050743162632,
      "learning_rate": 7.1562994535469995e-06,
      "loss": 0.0008,
      "step": 77420
    },
    {
      "epoch": 1.266955739180234,
      "grad_norm": 0.10586243867874146,
      "learning_rate": 7.155440547718182e-06,
      "loss": 0.002,
      "step": 77430
    },
    {
      "epoch": 1.2671193651313097,
      "grad_norm": 0.13382920622825623,
      "learning_rate": 7.1545815637598235e-06,
      "loss": 0.0021,
      "step": 77440
    },
    {
      "epoch": 1.2672829910823857,
      "grad_norm": 0.12447115778923035,
      "learning_rate": 7.1537225017030644e-06,
      "loss": 0.002,
      "step": 77450
    },
    {
      "epoch": 1.2674466170334615,
      "grad_norm": 0.07894103229045868,
      "learning_rate": 7.15286336157904e-06,
      "loss": 0.0014,
      "step": 77460
    },
    {
      "epoch": 1.2676102429845373,
      "grad_norm": 0.14126263558864594,
      "learning_rate": 7.152004143418892e-06,
      "loss": 0.0015,
      "step": 77470
    },
    {
      "epoch": 1.2677738689356133,
      "grad_norm": 0.16852203011512756,
      "learning_rate": 7.151144847253767e-06,
      "loss": 0.0046,
      "step": 77480
    },
    {
      "epoch": 1.267937494886689,
      "grad_norm": 0.1813594102859497,
      "learning_rate": 7.1502854731148095e-06,
      "loss": 0.0023,
      "step": 77490
    },
    {
      "epoch": 1.2681011208377648,
      "grad_norm": 0.09075183421373367,
      "learning_rate": 7.149426021033173e-06,
      "loss": 0.0036,
      "step": 77500
    },
    {
      "epoch": 1.2682647467888408,
      "grad_norm": 0.14291784167289734,
      "learning_rate": 7.14856649104001e-06,
      "loss": 0.0024,
      "step": 77510
    },
    {
      "epoch": 1.2684283727399166,
      "grad_norm": 0.08343230932950974,
      "learning_rate": 7.147706883166476e-06,
      "loss": 0.001,
      "step": 77520
    },
    {
      "epoch": 1.2685919986909924,
      "grad_norm": 0.05725209042429924,
      "learning_rate": 7.146847197443728e-06,
      "loss": 0.0056,
      "step": 77530
    },
    {
      "epoch": 1.2687556246420684,
      "grad_norm": 0.05695614963769913,
      "learning_rate": 7.145987433902929e-06,
      "loss": 0.0021,
      "step": 77540
    },
    {
      "epoch": 1.2689192505931441,
      "grad_norm": 0.1129378080368042,
      "learning_rate": 7.1451275925752435e-06,
      "loss": 0.002,
      "step": 77550
    },
    {
      "epoch": 1.26908287654422,
      "grad_norm": 0.21865597367286682,
      "learning_rate": 7.1442676734918385e-06,
      "loss": 0.0035,
      "step": 77560
    },
    {
      "epoch": 1.2692465024952957,
      "grad_norm": 0.15413720905780792,
      "learning_rate": 7.143407676683885e-06,
      "loss": 0.0011,
      "step": 77570
    },
    {
      "epoch": 1.2694101284463715,
      "grad_norm": 0.04753812402486801,
      "learning_rate": 7.142547602182552e-06,
      "loss": 0.0015,
      "step": 77580
    },
    {
      "epoch": 1.2695737543974475,
      "grad_norm": 0.6218154430389404,
      "learning_rate": 7.141687450019021e-06,
      "loss": 0.0016,
      "step": 77590
    },
    {
      "epoch": 1.2697373803485232,
      "grad_norm": 0.07290442287921906,
      "learning_rate": 7.140827220224466e-06,
      "loss": 0.0025,
      "step": 77600
    },
    {
      "epoch": 1.269901006299599,
      "grad_norm": 0.07512924820184708,
      "learning_rate": 7.13996691283007e-06,
      "loss": 0.0012,
      "step": 77610
    },
    {
      "epoch": 1.270064632250675,
      "grad_norm": 0.11224841326475143,
      "learning_rate": 7.139106527867016e-06,
      "loss": 0.0009,
      "step": 77620
    },
    {
      "epoch": 1.2702282582017508,
      "grad_norm": 0.1324429214000702,
      "learning_rate": 7.138246065366491e-06,
      "loss": 0.0021,
      "step": 77630
    },
    {
      "epoch": 1.2703918841528266,
      "grad_norm": 0.07526404410600662,
      "learning_rate": 7.137385525359686e-06,
      "loss": 0.001,
      "step": 77640
    },
    {
      "epoch": 1.2705555101039026,
      "grad_norm": 0.24098370969295502,
      "learning_rate": 7.136524907877793e-06,
      "loss": 0.0016,
      "step": 77650
    },
    {
      "epoch": 1.2707191360549783,
      "grad_norm": 0.021831529214978218,
      "learning_rate": 7.135664212952007e-06,
      "loss": 0.0019,
      "step": 77660
    },
    {
      "epoch": 1.270882762006054,
      "grad_norm": 0.04822388291358948,
      "learning_rate": 7.134803440613527e-06,
      "loss": 0.0015,
      "step": 77670
    },
    {
      "epoch": 1.27104638795713,
      "grad_norm": 0.01008241344243288,
      "learning_rate": 7.133942590893552e-06,
      "loss": 0.0011,
      "step": 77680
    },
    {
      "epoch": 1.2712100139082059,
      "grad_norm": 0.1856769174337387,
      "learning_rate": 7.133081663823287e-06,
      "loss": 0.0021,
      "step": 77690
    },
    {
      "epoch": 1.2713736398592816,
      "grad_norm": 0.11940106004476547,
      "learning_rate": 7.132220659433938e-06,
      "loss": 0.0025,
      "step": 77700
    },
    {
      "epoch": 1.2715372658103576,
      "grad_norm": 0.157598614692688,
      "learning_rate": 7.131359577756716e-06,
      "loss": 0.001,
      "step": 77710
    },
    {
      "epoch": 1.2717008917614334,
      "grad_norm": 0.1254163533449173,
      "learning_rate": 7.130498418822831e-06,
      "loss": 0.0028,
      "step": 77720
    },
    {
      "epoch": 1.2718645177125092,
      "grad_norm": 0.1596977859735489,
      "learning_rate": 7.129637182663499e-06,
      "loss": 0.002,
      "step": 77730
    },
    {
      "epoch": 1.272028143663585,
      "grad_norm": 0.07138001918792725,
      "learning_rate": 7.1287758693099385e-06,
      "loss": 0.0016,
      "step": 77740
    },
    {
      "epoch": 1.272191769614661,
      "grad_norm": 0.17522409558296204,
      "learning_rate": 7.127914478793368e-06,
      "loss": 0.002,
      "step": 77750
    },
    {
      "epoch": 1.2723553955657367,
      "grad_norm": 0.16013853251934052,
      "learning_rate": 7.1270530111450136e-06,
      "loss": 0.0037,
      "step": 77760
    },
    {
      "epoch": 1.2725190215168125,
      "grad_norm": 0.07808253169059753,
      "learning_rate": 7.126191466396096e-06,
      "loss": 0.0011,
      "step": 77770
    },
    {
      "epoch": 1.2726826474678883,
      "grad_norm": 0.19619536399841309,
      "learning_rate": 7.125329844577851e-06,
      "loss": 0.0014,
      "step": 77780
    },
    {
      "epoch": 1.2728462734189643,
      "grad_norm": 0.32044342160224915,
      "learning_rate": 7.124468145721506e-06,
      "loss": 0.002,
      "step": 77790
    },
    {
      "epoch": 1.27300989937004,
      "grad_norm": 0.23559348285198212,
      "learning_rate": 7.123606369858298e-06,
      "loss": 0.0023,
      "step": 77800
    },
    {
      "epoch": 1.2731735253211158,
      "grad_norm": 0.06079702451825142,
      "learning_rate": 7.122744517019463e-06,
      "loss": 0.0028,
      "step": 77810
    },
    {
      "epoch": 1.2733371512721918,
      "grad_norm": 0.18395693600177765,
      "learning_rate": 7.121882587236241e-06,
      "loss": 0.0015,
      "step": 77820
    },
    {
      "epoch": 1.2735007772232676,
      "grad_norm": 0.032289862632751465,
      "learning_rate": 7.121020580539873e-06,
      "loss": 0.0014,
      "step": 77830
    },
    {
      "epoch": 1.2736644031743434,
      "grad_norm": 0.0709395557641983,
      "learning_rate": 7.120158496961609e-06,
      "loss": 0.0034,
      "step": 77840
    },
    {
      "epoch": 1.2738280291254194,
      "grad_norm": 0.11847332864999771,
      "learning_rate": 7.119296336532693e-06,
      "loss": 0.0015,
      "step": 77850
    },
    {
      "epoch": 1.2739916550764951,
      "grad_norm": 0.13741576671600342,
      "learning_rate": 7.118434099284382e-06,
      "loss": 0.0021,
      "step": 77860
    },
    {
      "epoch": 1.274155281027571,
      "grad_norm": 0.056646984070539474,
      "learning_rate": 7.117571785247923e-06,
      "loss": 0.0013,
      "step": 77870
    },
    {
      "epoch": 1.274318906978647,
      "grad_norm": 0.12713302671909332,
      "learning_rate": 7.116709394454577e-06,
      "loss": 0.0014,
      "step": 77880
    },
    {
      "epoch": 1.2744825329297227,
      "grad_norm": 0.15320396423339844,
      "learning_rate": 7.115846926935604e-06,
      "loss": 0.003,
      "step": 77890
    },
    {
      "epoch": 1.2746461588807985,
      "grad_norm": 0.24882537126541138,
      "learning_rate": 7.114984382722264e-06,
      "loss": 0.0028,
      "step": 77900
    },
    {
      "epoch": 1.2748097848318745,
      "grad_norm": 0.3457034230232239,
      "learning_rate": 7.114121761845823e-06,
      "loss": 0.0019,
      "step": 77910
    },
    {
      "epoch": 1.2749734107829502,
      "grad_norm": 0.009374169632792473,
      "learning_rate": 7.113259064337551e-06,
      "loss": 0.001,
      "step": 77920
    },
    {
      "epoch": 1.275137036734026,
      "grad_norm": 0.06365344673395157,
      "learning_rate": 7.112396290228716e-06,
      "loss": 0.0014,
      "step": 77930
    },
    {
      "epoch": 1.2753006626851018,
      "grad_norm": 0.11823870986700058,
      "learning_rate": 7.111533439550593e-06,
      "loss": 0.0019,
      "step": 77940
    },
    {
      "epoch": 1.2754642886361778,
      "grad_norm": 0.20753897726535797,
      "learning_rate": 7.110670512334456e-06,
      "loss": 0.0023,
      "step": 77950
    },
    {
      "epoch": 1.2756279145872536,
      "grad_norm": 0.15599587559700012,
      "learning_rate": 7.109807508611589e-06,
      "loss": 0.0022,
      "step": 77960
    },
    {
      "epoch": 1.2757915405383293,
      "grad_norm": 0.06066054850816727,
      "learning_rate": 7.1089444284132695e-06,
      "loss": 0.0024,
      "step": 77970
    },
    {
      "epoch": 1.275955166489405,
      "grad_norm": 0.11398950964212418,
      "learning_rate": 7.1080812717707815e-06,
      "loss": 0.0014,
      "step": 77980
    },
    {
      "epoch": 1.276118792440481,
      "grad_norm": 0.29165568947792053,
      "learning_rate": 7.107218038715414e-06,
      "loss": 0.0029,
      "step": 77990
    },
    {
      "epoch": 1.2762824183915569,
      "grad_norm": 0.051056746393442154,
      "learning_rate": 7.1063547292784585e-06,
      "loss": 0.0016,
      "step": 78000
    },
    {
      "epoch": 1.2764460443426326,
      "grad_norm": 0.04217154532670975,
      "learning_rate": 7.105491343491207e-06,
      "loss": 0.0014,
      "step": 78010
    },
    {
      "epoch": 1.2766096702937086,
      "grad_norm": 0.15325215458869934,
      "learning_rate": 7.104627881384953e-06,
      "loss": 0.005,
      "step": 78020
    },
    {
      "epoch": 1.2767732962447844,
      "grad_norm": 0.05750011280179024,
      "learning_rate": 7.103764342990997e-06,
      "loss": 0.0017,
      "step": 78030
    },
    {
      "epoch": 1.2769369221958602,
      "grad_norm": 0.08452239632606506,
      "learning_rate": 7.102900728340641e-06,
      "loss": 0.002,
      "step": 78040
    },
    {
      "epoch": 1.2771005481469362,
      "grad_norm": 0.07005751878023148,
      "learning_rate": 7.102037037465188e-06,
      "loss": 0.0019,
      "step": 78050
    },
    {
      "epoch": 1.277264174098012,
      "grad_norm": 0.15740787982940674,
      "learning_rate": 7.101173270395943e-06,
      "loss": 0.0016,
      "step": 78060
    },
    {
      "epoch": 1.2774278000490877,
      "grad_norm": 0.301065593957901,
      "learning_rate": 7.1003094271642185e-06,
      "loss": 0.0023,
      "step": 78070
    },
    {
      "epoch": 1.2775914260001637,
      "grad_norm": 0.1475955843925476,
      "learning_rate": 7.099445507801324e-06,
      "loss": 0.0016,
      "step": 78080
    },
    {
      "epoch": 1.2777550519512395,
      "grad_norm": 0.04055211693048477,
      "learning_rate": 7.098581512338576e-06,
      "loss": 0.0021,
      "step": 78090
    },
    {
      "epoch": 1.2779186779023153,
      "grad_norm": 0.08093906939029694,
      "learning_rate": 7.097717440807291e-06,
      "loss": 0.0022,
      "step": 78100
    },
    {
      "epoch": 1.2780823038533913,
      "grad_norm": 0.10586875677108765,
      "learning_rate": 7.0968532932387935e-06,
      "loss": 0.0029,
      "step": 78110
    },
    {
      "epoch": 1.278245929804467,
      "grad_norm": 0.06545928865671158,
      "learning_rate": 7.095989069664402e-06,
      "loss": 0.0037,
      "step": 78120
    },
    {
      "epoch": 1.2784095557555428,
      "grad_norm": 0.21746321022510529,
      "learning_rate": 7.095124770115442e-06,
      "loss": 0.0025,
      "step": 78130
    },
    {
      "epoch": 1.2785731817066186,
      "grad_norm": 0.04639434069395065,
      "learning_rate": 7.094260394623247e-06,
      "loss": 0.0018,
      "step": 78140
    },
    {
      "epoch": 1.2787368076576944,
      "grad_norm": 0.10664045810699463,
      "learning_rate": 7.093395943219146e-06,
      "loss": 0.0018,
      "step": 78150
    },
    {
      "epoch": 1.2789004336087704,
      "grad_norm": 0.05785437300801277,
      "learning_rate": 7.092531415934473e-06,
      "loss": 0.0021,
      "step": 78160
    },
    {
      "epoch": 1.2790640595598461,
      "grad_norm": 0.28598418831825256,
      "learning_rate": 7.091666812800567e-06,
      "loss": 0.0026,
      "step": 78170
    },
    {
      "epoch": 1.279227685510922,
      "grad_norm": 0.10265430808067322,
      "learning_rate": 7.090802133848764e-06,
      "loss": 0.0011,
      "step": 78180
    },
    {
      "epoch": 1.279391311461998,
      "grad_norm": 0.18367788195610046,
      "learning_rate": 7.08993737911041e-06,
      "loss": 0.0019,
      "step": 78190
    },
    {
      "epoch": 1.2795549374130737,
      "grad_norm": 0.09943254292011261,
      "learning_rate": 7.089072548616849e-06,
      "loss": 0.0012,
      "step": 78200
    },
    {
      "epoch": 1.2797185633641495,
      "grad_norm": 0.036855001002550125,
      "learning_rate": 7.088207642399429e-06,
      "loss": 0.0011,
      "step": 78210
    },
    {
      "epoch": 1.2798821893152255,
      "grad_norm": 0.07823864370584488,
      "learning_rate": 7.0873426604895e-06,
      "loss": 0.0013,
      "step": 78220
    },
    {
      "epoch": 1.2800458152663012,
      "grad_norm": 0.23223811388015747,
      "learning_rate": 7.086477602918418e-06,
      "loss": 0.0023,
      "step": 78230
    },
    {
      "epoch": 1.280209441217377,
      "grad_norm": 0.10750643163919449,
      "learning_rate": 7.085612469717536e-06,
      "loss": 0.0015,
      "step": 78240
    },
    {
      "epoch": 1.280373067168453,
      "grad_norm": 0.09745393693447113,
      "learning_rate": 7.084747260918214e-06,
      "loss": 0.0018,
      "step": 78250
    },
    {
      "epoch": 1.2805366931195288,
      "grad_norm": 0.0866905227303505,
      "learning_rate": 7.083881976551817e-06,
      "loss": 0.0009,
      "step": 78260
    },
    {
      "epoch": 1.2807003190706046,
      "grad_norm": 0.4757748246192932,
      "learning_rate": 7.083016616649705e-06,
      "loss": 0.0017,
      "step": 78270
    },
    {
      "epoch": 1.2808639450216805,
      "grad_norm": 0.26839548349380493,
      "learning_rate": 7.082151181243248e-06,
      "loss": 0.0024,
      "step": 78280
    },
    {
      "epoch": 1.2810275709727563,
      "grad_norm": 0.1678846776485443,
      "learning_rate": 7.0812856703638135e-06,
      "loss": 0.0032,
      "step": 78290
    },
    {
      "epoch": 1.281191196923832,
      "grad_norm": 0.08280716091394424,
      "learning_rate": 7.0804200840427765e-06,
      "loss": 0.0018,
      "step": 78300
    },
    {
      "epoch": 1.281354822874908,
      "grad_norm": 0.17985455691814423,
      "learning_rate": 7.079554422311513e-06,
      "loss": 0.008,
      "step": 78310
    },
    {
      "epoch": 1.2815184488259839,
      "grad_norm": 0.0738675519824028,
      "learning_rate": 7.078688685201398e-06,
      "loss": 0.0022,
      "step": 78320
    },
    {
      "epoch": 1.2816820747770596,
      "grad_norm": 0.5514221787452698,
      "learning_rate": 7.0778228727438146e-06,
      "loss": 0.0033,
      "step": 78330
    },
    {
      "epoch": 1.2818457007281354,
      "grad_norm": 0.07533999532461166,
      "learning_rate": 7.076956984970147e-06,
      "loss": 0.0025,
      "step": 78340
    },
    {
      "epoch": 1.2820093266792112,
      "grad_norm": 0.1644175499677658,
      "learning_rate": 7.076091021911777e-06,
      "loss": 0.0025,
      "step": 78350
    },
    {
      "epoch": 1.2821729526302872,
      "grad_norm": 0.09712538868188858,
      "learning_rate": 7.075224983600102e-06,
      "loss": 0.0017,
      "step": 78360
    },
    {
      "epoch": 1.282336578581363,
      "grad_norm": 0.04440287500619888,
      "learning_rate": 7.074358870066507e-06,
      "loss": 0.0029,
      "step": 78370
    },
    {
      "epoch": 1.2825002045324387,
      "grad_norm": 0.02812427282333374,
      "learning_rate": 7.073492681342388e-06,
      "loss": 0.0015,
      "step": 78380
    },
    {
      "epoch": 1.2826638304835147,
      "grad_norm": 0.050098370760679245,
      "learning_rate": 7.072626417459143e-06,
      "loss": 0.002,
      "step": 78390
    },
    {
      "epoch": 1.2828274564345905,
      "grad_norm": 0.0541042797267437,
      "learning_rate": 7.071760078448171e-06,
      "loss": 0.0013,
      "step": 78400
    },
    {
      "epoch": 1.2829910823856663,
      "grad_norm": 0.10522080212831497,
      "learning_rate": 7.070893664340877e-06,
      "loss": 0.0021,
      "step": 78410
    },
    {
      "epoch": 1.2831547083367423,
      "grad_norm": 0.04674161970615387,
      "learning_rate": 7.070027175168665e-06,
      "loss": 0.001,
      "step": 78420
    },
    {
      "epoch": 1.283318334287818,
      "grad_norm": 0.09381629526615143,
      "learning_rate": 7.069160610962942e-06,
      "loss": 0.0014,
      "step": 78430
    },
    {
      "epoch": 1.2834819602388938,
      "grad_norm": 0.01699410378932953,
      "learning_rate": 7.0682939717551205e-06,
      "loss": 0.0023,
      "step": 78440
    },
    {
      "epoch": 1.2836455861899698,
      "grad_norm": 0.195709228515625,
      "learning_rate": 7.067427257576613e-06,
      "loss": 0.0026,
      "step": 78450
    },
    {
      "epoch": 1.2838092121410456,
      "grad_norm": 0.1978006511926651,
      "learning_rate": 7.066560468458837e-06,
      "loss": 0.0024,
      "step": 78460
    },
    {
      "epoch": 1.2839728380921214,
      "grad_norm": 0.07010656595230103,
      "learning_rate": 7.065693604433212e-06,
      "loss": 0.0036,
      "step": 78470
    },
    {
      "epoch": 1.2841364640431974,
      "grad_norm": 0.10471019893884659,
      "learning_rate": 7.064826665531157e-06,
      "loss": 0.0025,
      "step": 78480
    },
    {
      "epoch": 1.2843000899942731,
      "grad_norm": 0.12300293892621994,
      "learning_rate": 7.063959651784097e-06,
      "loss": 0.002,
      "step": 78490
    },
    {
      "epoch": 1.284463715945349,
      "grad_norm": 0.16696114838123322,
      "learning_rate": 7.063092563223461e-06,
      "loss": 0.0021,
      "step": 78500
    },
    {
      "epoch": 1.2846273418964247,
      "grad_norm": 0.08266822248697281,
      "learning_rate": 7.06222539988068e-06,
      "loss": 0.0017,
      "step": 78510
    },
    {
      "epoch": 1.2847909678475007,
      "grad_norm": 0.05953631550073624,
      "learning_rate": 7.061358161787183e-06,
      "loss": 0.0033,
      "step": 78520
    },
    {
      "epoch": 1.2849545937985765,
      "grad_norm": 0.08135845512151718,
      "learning_rate": 7.0604908489744075e-06,
      "loss": 0.0013,
      "step": 78530
    },
    {
      "epoch": 1.2851182197496522,
      "grad_norm": 0.050916410982608795,
      "learning_rate": 7.05962346147379e-06,
      "loss": 0.0029,
      "step": 78540
    },
    {
      "epoch": 1.285281845700728,
      "grad_norm": 0.06050095707178116,
      "learning_rate": 7.058755999316773e-06,
      "loss": 0.0016,
      "step": 78550
    },
    {
      "epoch": 1.285445471651804,
      "grad_norm": 0.045107923448085785,
      "learning_rate": 7.0578884625347985e-06,
      "loss": 0.0016,
      "step": 78560
    },
    {
      "epoch": 1.2856090976028798,
      "grad_norm": 0.06090124696493149,
      "learning_rate": 7.057020851159315e-06,
      "loss": 0.0013,
      "step": 78570
    },
    {
      "epoch": 1.2857727235539556,
      "grad_norm": 0.15376600623130798,
      "learning_rate": 7.056153165221768e-06,
      "loss": 0.0016,
      "step": 78580
    },
    {
      "epoch": 1.2859363495050316,
      "grad_norm": 0.16558267176151276,
      "learning_rate": 7.055285404753611e-06,
      "loss": 0.0017,
      "step": 78590
    },
    {
      "epoch": 1.2860999754561073,
      "grad_norm": 0.16581164300441742,
      "learning_rate": 7.054417569786298e-06,
      "loss": 0.0022,
      "step": 78600
    },
    {
      "epoch": 1.286263601407183,
      "grad_norm": 0.12265679240226746,
      "learning_rate": 7.053549660351286e-06,
      "loss": 0.0021,
      "step": 78610
    },
    {
      "epoch": 1.286427227358259,
      "grad_norm": 0.26179322600364685,
      "learning_rate": 7.052681676480033e-06,
      "loss": 0.0024,
      "step": 78620
    },
    {
      "epoch": 1.2865908533093349,
      "grad_norm": 0.034944869577884674,
      "learning_rate": 7.0518136182040045e-06,
      "loss": 0.0022,
      "step": 78630
    },
    {
      "epoch": 1.2867544792604106,
      "grad_norm": 0.061872292309999466,
      "learning_rate": 7.050945485554662e-06,
      "loss": 0.0026,
      "step": 78640
    },
    {
      "epoch": 1.2869181052114866,
      "grad_norm": 0.04613882675766945,
      "learning_rate": 7.050077278563476e-06,
      "loss": 0.0016,
      "step": 78650
    },
    {
      "epoch": 1.2870817311625624,
      "grad_norm": 0.09044929593801498,
      "learning_rate": 7.049208997261916e-06,
      "loss": 0.002,
      "step": 78660
    },
    {
      "epoch": 1.2872453571136382,
      "grad_norm": 0.18111932277679443,
      "learning_rate": 7.048340641681455e-06,
      "loss": 0.0025,
      "step": 78670
    },
    {
      "epoch": 1.2874089830647142,
      "grad_norm": 0.054907627403736115,
      "learning_rate": 7.047472211853569e-06,
      "loss": 0.0036,
      "step": 78680
    },
    {
      "epoch": 1.28757260901579,
      "grad_norm": 0.14432856440544128,
      "learning_rate": 7.046603707809736e-06,
      "loss": 0.0029,
      "step": 78690
    },
    {
      "epoch": 1.2877362349668657,
      "grad_norm": 0.1697741448879242,
      "learning_rate": 7.045735129581436e-06,
      "loss": 0.0026,
      "step": 78700
    },
    {
      "epoch": 1.2878998609179415,
      "grad_norm": 0.12358502298593521,
      "learning_rate": 7.044866477200156e-06,
      "loss": 0.0031,
      "step": 78710
    },
    {
      "epoch": 1.2880634868690175,
      "grad_norm": 0.1889401525259018,
      "learning_rate": 7.043997750697381e-06,
      "loss": 0.004,
      "step": 78720
    },
    {
      "epoch": 1.2882271128200933,
      "grad_norm": 0.14887337386608124,
      "learning_rate": 7.0431289501046e-06,
      "loss": 0.0045,
      "step": 78730
    },
    {
      "epoch": 1.288390738771169,
      "grad_norm": 0.08990081399679184,
      "learning_rate": 7.0422600754533045e-06,
      "loss": 0.0022,
      "step": 78740
    },
    {
      "epoch": 1.2885543647222448,
      "grad_norm": 0.06419666111469269,
      "learning_rate": 7.0413911267749905e-06,
      "loss": 0.0018,
      "step": 78750
    },
    {
      "epoch": 1.2887179906733208,
      "grad_norm": 0.030298173427581787,
      "learning_rate": 7.040522104101153e-06,
      "loss": 0.0023,
      "step": 78760
    },
    {
      "epoch": 1.2888816166243966,
      "grad_norm": 0.0547197200357914,
      "learning_rate": 7.039653007463294e-06,
      "loss": 0.0019,
      "step": 78770
    },
    {
      "epoch": 1.2890452425754724,
      "grad_norm": 0.20454275608062744,
      "learning_rate": 7.038783836892917e-06,
      "loss": 0.002,
      "step": 78780
    },
    {
      "epoch": 1.2892088685265484,
      "grad_norm": 0.0036465420853346586,
      "learning_rate": 7.0379145924215255e-06,
      "loss": 0.0026,
      "step": 78790
    },
    {
      "epoch": 1.2893724944776241,
      "grad_norm": 0.18773530423641205,
      "learning_rate": 7.037045274080628e-06,
      "loss": 0.003,
      "step": 78800
    },
    {
      "epoch": 1.2895361204287,
      "grad_norm": 0.6376838684082031,
      "learning_rate": 7.036175881901735e-06,
      "loss": 0.0014,
      "step": 78810
    },
    {
      "epoch": 1.289699746379776,
      "grad_norm": 0.02673187293112278,
      "learning_rate": 7.0353064159163595e-06,
      "loss": 0.0026,
      "step": 78820
    },
    {
      "epoch": 1.2898633723308517,
      "grad_norm": 0.16669833660125732,
      "learning_rate": 7.034436876156019e-06,
      "loss": 0.0027,
      "step": 78830
    },
    {
      "epoch": 1.2900269982819275,
      "grad_norm": 0.13926498591899872,
      "learning_rate": 7.03356726265223e-06,
      "loss": 0.0052,
      "step": 78840
    },
    {
      "epoch": 1.2901906242330035,
      "grad_norm": 0.054456811398267746,
      "learning_rate": 7.0326975754365165e-06,
      "loss": 0.0017,
      "step": 78850
    },
    {
      "epoch": 1.2903542501840792,
      "grad_norm": 0.072306789457798,
      "learning_rate": 7.0318278145404016e-06,
      "loss": 0.002,
      "step": 78860
    },
    {
      "epoch": 1.290517876135155,
      "grad_norm": 0.3277745842933655,
      "learning_rate": 7.030957979995412e-06,
      "loss": 0.0023,
      "step": 78870
    },
    {
      "epoch": 1.290681502086231,
      "grad_norm": 0.11229134351015091,
      "learning_rate": 7.030088071833077e-06,
      "loss": 0.0032,
      "step": 78880
    },
    {
      "epoch": 1.2908451280373068,
      "grad_norm": 0.15741701424121857,
      "learning_rate": 7.0292180900849275e-06,
      "loss": 0.0024,
      "step": 78890
    },
    {
      "epoch": 1.2910087539883826,
      "grad_norm": 0.09956037998199463,
      "learning_rate": 7.0283480347825e-06,
      "loss": 0.0023,
      "step": 78900
    },
    {
      "epoch": 1.2911723799394583,
      "grad_norm": 0.044183846563100815,
      "learning_rate": 7.027477905957329e-06,
      "loss": 0.0017,
      "step": 78910
    },
    {
      "epoch": 1.2913360058905343,
      "grad_norm": 0.05782116577029228,
      "learning_rate": 7.0266077036409594e-06,
      "loss": 0.0018,
      "step": 78920
    },
    {
      "epoch": 1.29149963184161,
      "grad_norm": 0.2579874098300934,
      "learning_rate": 7.0257374278649314e-06,
      "loss": 0.0022,
      "step": 78930
    },
    {
      "epoch": 1.2916632577926859,
      "grad_norm": 0.05683138966560364,
      "learning_rate": 7.024867078660789e-06,
      "loss": 0.0026,
      "step": 78940
    },
    {
      "epoch": 1.2918268837437616,
      "grad_norm": 0.2643553912639618,
      "learning_rate": 7.023996656060082e-06,
      "loss": 0.0027,
      "step": 78950
    },
    {
      "epoch": 1.2919905096948376,
      "grad_norm": 0.07228142023086548,
      "learning_rate": 7.023126160094363e-06,
      "loss": 0.0017,
      "step": 78960
    },
    {
      "epoch": 1.2921541356459134,
      "grad_norm": 0.021867431700229645,
      "learning_rate": 7.022255590795181e-06,
      "loss": 0.0016,
      "step": 78970
    },
    {
      "epoch": 1.2923177615969892,
      "grad_norm": 0.3119870126247406,
      "learning_rate": 7.021384948194094e-06,
      "loss": 0.0032,
      "step": 78980
    },
    {
      "epoch": 1.2924813875480652,
      "grad_norm": 0.12576338648796082,
      "learning_rate": 7.02051423232266e-06,
      "loss": 0.0024,
      "step": 78990
    },
    {
      "epoch": 1.292645013499141,
      "grad_norm": 0.03334770351648331,
      "learning_rate": 7.0196434432124425e-06,
      "loss": 0.0015,
      "step": 79000
    },
    {
      "epoch": 1.2928086394502167,
      "grad_norm": 0.1753448247909546,
      "learning_rate": 7.018772580895006e-06,
      "loss": 0.002,
      "step": 79010
    },
    {
      "epoch": 1.2929722654012927,
      "grad_norm": 0.08526654541492462,
      "learning_rate": 7.017901645401915e-06,
      "loss": 0.0015,
      "step": 79020
    },
    {
      "epoch": 1.2931358913523685,
      "grad_norm": 0.26862430572509766,
      "learning_rate": 7.0170306367647366e-06,
      "loss": 0.003,
      "step": 79030
    },
    {
      "epoch": 1.2932995173034443,
      "grad_norm": 0.09505033493041992,
      "learning_rate": 7.016159555015047e-06,
      "loss": 0.0024,
      "step": 79040
    },
    {
      "epoch": 1.2934631432545203,
      "grad_norm": 0.013281266205012798,
      "learning_rate": 7.015288400184419e-06,
      "loss": 0.0024,
      "step": 79050
    },
    {
      "epoch": 1.293626769205596,
      "grad_norm": 0.10048270970582962,
      "learning_rate": 7.014417172304431e-06,
      "loss": 0.0018,
      "step": 79060
    },
    {
      "epoch": 1.2937903951566718,
      "grad_norm": 0.14620321989059448,
      "learning_rate": 7.0135458714066604e-06,
      "loss": 0.0024,
      "step": 79070
    },
    {
      "epoch": 1.2939540211077478,
      "grad_norm": 0.21103666722774506,
      "learning_rate": 7.012674497522692e-06,
      "loss": 0.0022,
      "step": 79080
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 0.06079899147152901,
      "learning_rate": 7.0118030506841114e-06,
      "loss": 0.0015,
      "step": 79090
    },
    {
      "epoch": 1.2942812730098994,
      "grad_norm": 0.12934254109859467,
      "learning_rate": 7.010931530922504e-06,
      "loss": 0.0021,
      "step": 79100
    },
    {
      "epoch": 1.2944448989609751,
      "grad_norm": 0.07510894536972046,
      "learning_rate": 7.010059938269463e-06,
      "loss": 0.0019,
      "step": 79110
    },
    {
      "epoch": 1.294608524912051,
      "grad_norm": 0.027073414996266365,
      "learning_rate": 7.0091882727565795e-06,
      "loss": 0.0019,
      "step": 79120
    },
    {
      "epoch": 1.294772150863127,
      "grad_norm": 0.034520912915468216,
      "learning_rate": 7.0083165344154505e-06,
      "loss": 0.002,
      "step": 79130
    },
    {
      "epoch": 1.2949357768142027,
      "grad_norm": 0.20583203434944153,
      "learning_rate": 7.007444723277673e-06,
      "loss": 0.003,
      "step": 79140
    },
    {
      "epoch": 1.2950994027652785,
      "grad_norm": 0.2449686974287033,
      "learning_rate": 7.00657283937485e-06,
      "loss": 0.0039,
      "step": 79150
    },
    {
      "epoch": 1.2952630287163545,
      "grad_norm": 0.13770616054534912,
      "learning_rate": 7.005700882738585e-06,
      "loss": 0.0023,
      "step": 79160
    },
    {
      "epoch": 1.2954266546674302,
      "grad_norm": 0.02674509398639202,
      "learning_rate": 7.004828853400483e-06,
      "loss": 0.0014,
      "step": 79170
    },
    {
      "epoch": 1.295590280618506,
      "grad_norm": 0.06522715091705322,
      "learning_rate": 7.003956751392154e-06,
      "loss": 0.0037,
      "step": 79180
    },
    {
      "epoch": 1.295753906569582,
      "grad_norm": 0.06211570277810097,
      "learning_rate": 7.003084576745207e-06,
      "loss": 0.0023,
      "step": 79190
    },
    {
      "epoch": 1.2959175325206578,
      "grad_norm": 0.08482442796230316,
      "learning_rate": 7.002212329491259e-06,
      "loss": 0.002,
      "step": 79200
    },
    {
      "epoch": 1.2960811584717336,
      "grad_norm": 0.10277386754751205,
      "learning_rate": 7.001340009661927e-06,
      "loss": 0.0028,
      "step": 79210
    },
    {
      "epoch": 1.2962447844228095,
      "grad_norm": 0.07776857167482376,
      "learning_rate": 7.0004676172888305e-06,
      "loss": 0.0012,
      "step": 79220
    },
    {
      "epoch": 1.2964084103738853,
      "grad_norm": 0.05071259289979935,
      "learning_rate": 6.999595152403591e-06,
      "loss": 0.0037,
      "step": 79230
    },
    {
      "epoch": 1.296572036324961,
      "grad_norm": 0.1432274878025055,
      "learning_rate": 6.998722615037832e-06,
      "loss": 0.0022,
      "step": 79240
    },
    {
      "epoch": 1.296735662276037,
      "grad_norm": 0.16380329430103302,
      "learning_rate": 6.997850005223182e-06,
      "loss": 0.0012,
      "step": 79250
    },
    {
      "epoch": 1.2968992882271129,
      "grad_norm": 0.26686760783195496,
      "learning_rate": 6.996977322991272e-06,
      "loss": 0.0043,
      "step": 79260
    },
    {
      "epoch": 1.2970629141781886,
      "grad_norm": 0.04556310921907425,
      "learning_rate": 6.9961045683737316e-06,
      "loss": 0.0013,
      "step": 79270
    },
    {
      "epoch": 1.2972265401292646,
      "grad_norm": 0.22555766999721527,
      "learning_rate": 6.995231741402199e-06,
      "loss": 0.0027,
      "step": 79280
    },
    {
      "epoch": 1.2973901660803404,
      "grad_norm": 0.3267669081687927,
      "learning_rate": 6.994358842108313e-06,
      "loss": 0.0013,
      "step": 79290
    },
    {
      "epoch": 1.2975537920314162,
      "grad_norm": 0.07013623416423798,
      "learning_rate": 6.993485870523711e-06,
      "loss": 0.0018,
      "step": 79300
    },
    {
      "epoch": 1.297717417982492,
      "grad_norm": 0.19409142434597015,
      "learning_rate": 6.992612826680037e-06,
      "loss": 0.0023,
      "step": 79310
    },
    {
      "epoch": 1.2978810439335677,
      "grad_norm": 0.032443542033433914,
      "learning_rate": 6.991739710608938e-06,
      "loss": 0.0027,
      "step": 79320
    },
    {
      "epoch": 1.2980446698846437,
      "grad_norm": 0.17591342329978943,
      "learning_rate": 6.990866522342061e-06,
      "loss": 0.002,
      "step": 79330
    },
    {
      "epoch": 1.2982082958357195,
      "grad_norm": 0.16668100655078888,
      "learning_rate": 6.989993261911057e-06,
      "loss": 0.0021,
      "step": 79340
    },
    {
      "epoch": 1.2983719217867953,
      "grad_norm": 0.0907452404499054,
      "learning_rate": 6.9891199293475794e-06,
      "loss": 0.0036,
      "step": 79350
    },
    {
      "epoch": 1.2985355477378713,
      "grad_norm": 0.02006194181740284,
      "learning_rate": 6.988246524683288e-06,
      "loss": 0.0017,
      "step": 79360
    },
    {
      "epoch": 1.298699173688947,
      "grad_norm": 0.15572962164878845,
      "learning_rate": 6.987373047949837e-06,
      "loss": 0.002,
      "step": 79370
    },
    {
      "epoch": 1.2988627996400228,
      "grad_norm": 0.26907023787498474,
      "learning_rate": 6.986499499178889e-06,
      "loss": 0.0029,
      "step": 79380
    },
    {
      "epoch": 1.2990264255910988,
      "grad_norm": 0.10807106643915176,
      "learning_rate": 6.985625878402109e-06,
      "loss": 0.0023,
      "step": 79390
    },
    {
      "epoch": 1.2991900515421746,
      "grad_norm": 0.05270524322986603,
      "learning_rate": 6.9847521856511635e-06,
      "loss": 0.0021,
      "step": 79400
    },
    {
      "epoch": 1.2993536774932504,
      "grad_norm": 0.3921222686767578,
      "learning_rate": 6.983878420957722e-06,
      "loss": 0.0036,
      "step": 79410
    },
    {
      "epoch": 1.2995173034443264,
      "grad_norm": 0.1369101107120514,
      "learning_rate": 6.9830045843534546e-06,
      "loss": 0.0027,
      "step": 79420
    },
    {
      "epoch": 1.2996809293954021,
      "grad_norm": 0.23730507493019104,
      "learning_rate": 6.982130675870037e-06,
      "loss": 0.0021,
      "step": 79430
    },
    {
      "epoch": 1.299844555346478,
      "grad_norm": 0.038298897445201874,
      "learning_rate": 6.981256695539147e-06,
      "loss": 0.0018,
      "step": 79440
    },
    {
      "epoch": 1.300008181297554,
      "grad_norm": 0.06326301395893097,
      "learning_rate": 6.980382643392463e-06,
      "loss": 0.0012,
      "step": 79450
    },
    {
      "epoch": 1.3001718072486297,
      "grad_norm": 0.039917342364788055,
      "learning_rate": 6.9795085194616685e-06,
      "loss": 0.0014,
      "step": 79460
    },
    {
      "epoch": 1.3003354331997055,
      "grad_norm": 0.25611624121665955,
      "learning_rate": 6.978634323778449e-06,
      "loss": 0.0022,
      "step": 79470
    },
    {
      "epoch": 1.3004990591507812,
      "grad_norm": 0.14208780229091644,
      "learning_rate": 6.977760056374489e-06,
      "loss": 0.0014,
      "step": 79480
    },
    {
      "epoch": 1.3006626851018572,
      "grad_norm": 0.1369677037000656,
      "learning_rate": 6.976885717281481e-06,
      "loss": 0.004,
      "step": 79490
    },
    {
      "epoch": 1.300826311052933,
      "grad_norm": 0.052475377917289734,
      "learning_rate": 6.976011306531117e-06,
      "loss": 0.0027,
      "step": 79500
    },
    {
      "epoch": 1.3009899370040088,
      "grad_norm": 0.11846672743558884,
      "learning_rate": 6.975136824155091e-06,
      "loss": 0.0016,
      "step": 79510
    },
    {
      "epoch": 1.3011535629550846,
      "grad_norm": 0.10087010264396667,
      "learning_rate": 6.974262270185105e-06,
      "loss": 0.002,
      "step": 79520
    },
    {
      "epoch": 1.3013171889061605,
      "grad_norm": 0.1653912514448166,
      "learning_rate": 6.973387644652856e-06,
      "loss": 0.0023,
      "step": 79530
    },
    {
      "epoch": 1.3014808148572363,
      "grad_norm": 0.08504143357276917,
      "learning_rate": 6.972512947590047e-06,
      "loss": 0.0023,
      "step": 79540
    },
    {
      "epoch": 1.301644440808312,
      "grad_norm": 0.08199617266654968,
      "learning_rate": 6.971638179028384e-06,
      "loss": 0.002,
      "step": 79550
    },
    {
      "epoch": 1.301808066759388,
      "grad_norm": 0.07083486765623093,
      "learning_rate": 6.970763338999575e-06,
      "loss": 0.0008,
      "step": 79560
    },
    {
      "epoch": 1.3019716927104639,
      "grad_norm": 0.09887305647134781,
      "learning_rate": 6.969888427535334e-06,
      "loss": 0.0081,
      "step": 79570
    },
    {
      "epoch": 1.3021353186615396,
      "grad_norm": 0.19988499581813812,
      "learning_rate": 6.969013444667371e-06,
      "loss": 0.0013,
      "step": 79580
    },
    {
      "epoch": 1.3022989446126156,
      "grad_norm": 0.025854721665382385,
      "learning_rate": 6.968138390427403e-06,
      "loss": 0.0022,
      "step": 79590
    },
    {
      "epoch": 1.3024625705636914,
      "grad_norm": 0.2710879147052765,
      "learning_rate": 6.967263264847148e-06,
      "loss": 0.0032,
      "step": 79600
    },
    {
      "epoch": 1.3026261965147672,
      "grad_norm": 0.06273536384105682,
      "learning_rate": 6.966388067958329e-06,
      "loss": 0.0019,
      "step": 79610
    },
    {
      "epoch": 1.3027898224658432,
      "grad_norm": 0.1396106630563736,
      "learning_rate": 6.965512799792666e-06,
      "loss": 0.0028,
      "step": 79620
    },
    {
      "epoch": 1.302953448416919,
      "grad_norm": 0.3284461498260498,
      "learning_rate": 6.964637460381891e-06,
      "loss": 0.002,
      "step": 79630
    },
    {
      "epoch": 1.3031170743679947,
      "grad_norm": 0.24664875864982605,
      "learning_rate": 6.963762049757729e-06,
      "loss": 0.0017,
      "step": 79640
    },
    {
      "epoch": 1.3032807003190707,
      "grad_norm": 0.03524404764175415,
      "learning_rate": 6.962886567951912e-06,
      "loss": 0.0014,
      "step": 79650
    },
    {
      "epoch": 1.3034443262701465,
      "grad_norm": 0.359945684671402,
      "learning_rate": 6.962011014996175e-06,
      "loss": 0.0032,
      "step": 79660
    },
    {
      "epoch": 1.3036079522212223,
      "grad_norm": 0.5091138482093811,
      "learning_rate": 6.961135390922255e-06,
      "loss": 0.0035,
      "step": 79670
    },
    {
      "epoch": 1.303771578172298,
      "grad_norm": 0.19365240633487701,
      "learning_rate": 6.960259695761889e-06,
      "loss": 0.0023,
      "step": 79680
    },
    {
      "epoch": 1.303935204123374,
      "grad_norm": 0.0849161446094513,
      "learning_rate": 6.959383929546823e-06,
      "loss": 0.0015,
      "step": 79690
    },
    {
      "epoch": 1.3040988300744498,
      "grad_norm": 0.13367627561092377,
      "learning_rate": 6.958508092308796e-06,
      "loss": 0.0043,
      "step": 79700
    },
    {
      "epoch": 1.3042624560255256,
      "grad_norm": 0.10489772260189056,
      "learning_rate": 6.957632184079558e-06,
      "loss": 0.0026,
      "step": 79710
    },
    {
      "epoch": 1.3044260819766014,
      "grad_norm": 0.08480428904294968,
      "learning_rate": 6.95675620489086e-06,
      "loss": 0.0013,
      "step": 79720
    },
    {
      "epoch": 1.3045897079276774,
      "grad_norm": 0.18104498088359833,
      "learning_rate": 6.955880154774451e-06,
      "loss": 0.0025,
      "step": 79730
    },
    {
      "epoch": 1.3047533338787531,
      "grad_norm": 0.07868177443742752,
      "learning_rate": 6.955004033762088e-06,
      "loss": 0.0016,
      "step": 79740
    },
    {
      "epoch": 1.304916959829829,
      "grad_norm": 0.09211232513189316,
      "learning_rate": 6.954127841885528e-06,
      "loss": 0.0018,
      "step": 79750
    },
    {
      "epoch": 1.305080585780905,
      "grad_norm": 0.12492793053388596,
      "learning_rate": 6.95325157917653e-06,
      "loss": 0.002,
      "step": 79760
    },
    {
      "epoch": 1.3052442117319807,
      "grad_norm": 0.1060870811343193,
      "learning_rate": 6.952375245666856e-06,
      "loss": 0.0023,
      "step": 79770
    },
    {
      "epoch": 1.3054078376830565,
      "grad_norm": 0.07007330656051636,
      "learning_rate": 6.951498841388272e-06,
      "loss": 0.0028,
      "step": 79780
    },
    {
      "epoch": 1.3055714636341325,
      "grad_norm": 0.2370176464319229,
      "learning_rate": 6.9506223663725454e-06,
      "loss": 0.0023,
      "step": 79790
    },
    {
      "epoch": 1.3057350895852082,
      "grad_norm": 0.04554685950279236,
      "learning_rate": 6.949745820651447e-06,
      "loss": 0.0025,
      "step": 79800
    },
    {
      "epoch": 1.305898715536284,
      "grad_norm": 0.06224415451288223,
      "learning_rate": 6.948869204256748e-06,
      "loss": 0.0019,
      "step": 79810
    },
    {
      "epoch": 1.30606234148736,
      "grad_norm": 0.16847485303878784,
      "learning_rate": 6.947992517220224e-06,
      "loss": 0.0023,
      "step": 79820
    },
    {
      "epoch": 1.3062259674384358,
      "grad_norm": 0.05676880106329918,
      "learning_rate": 6.947115759573653e-06,
      "loss": 0.0015,
      "step": 79830
    },
    {
      "epoch": 1.3063895933895115,
      "grad_norm": 0.11860613524913788,
      "learning_rate": 6.946238931348816e-06,
      "loss": 0.0027,
      "step": 79840
    },
    {
      "epoch": 1.3065532193405875,
      "grad_norm": 0.5313612222671509,
      "learning_rate": 6.945362032577496e-06,
      "loss": 0.0013,
      "step": 79850
    },
    {
      "epoch": 1.3067168452916633,
      "grad_norm": 0.22232674062252045,
      "learning_rate": 6.9444850632914775e-06,
      "loss": 0.0015,
      "step": 79860
    },
    {
      "epoch": 1.306880471242739,
      "grad_norm": 0.048142462968826294,
      "learning_rate": 6.943608023522549e-06,
      "loss": 0.0034,
      "step": 79870
    },
    {
      "epoch": 1.3070440971938149,
      "grad_norm": 0.24687138199806213,
      "learning_rate": 6.942730913302501e-06,
      "loss": 0.0026,
      "step": 79880
    },
    {
      "epoch": 1.3072077231448906,
      "grad_norm": 0.19159138202667236,
      "learning_rate": 6.941853732663127e-06,
      "loss": 0.0033,
      "step": 79890
    },
    {
      "epoch": 1.3073713490959666,
      "grad_norm": 0.08600754290819168,
      "learning_rate": 6.9409764816362215e-06,
      "loss": 0.0022,
      "step": 79900
    },
    {
      "epoch": 1.3075349750470424,
      "grad_norm": 0.1998615264892578,
      "learning_rate": 6.9400991602535815e-06,
      "loss": 0.0018,
      "step": 79910
    },
    {
      "epoch": 1.3076986009981182,
      "grad_norm": 0.10417447239160538,
      "learning_rate": 6.939221768547013e-06,
      "loss": 0.0022,
      "step": 79920
    },
    {
      "epoch": 1.3078622269491942,
      "grad_norm": 0.1524556428194046,
      "learning_rate": 6.9383443065483155e-06,
      "loss": 0.0015,
      "step": 79930
    },
    {
      "epoch": 1.30802585290027,
      "grad_norm": 0.04746182635426521,
      "learning_rate": 6.937466774289296e-06,
      "loss": 0.0017,
      "step": 79940
    },
    {
      "epoch": 1.3081894788513457,
      "grad_norm": 0.2966523766517639,
      "learning_rate": 6.936589171801761e-06,
      "loss": 0.0017,
      "step": 79950
    },
    {
      "epoch": 1.3083531048024217,
      "grad_norm": 0.1536683440208435,
      "learning_rate": 6.935711499117523e-06,
      "loss": 0.0017,
      "step": 79960
    },
    {
      "epoch": 1.3085167307534975,
      "grad_norm": 0.055951982736587524,
      "learning_rate": 6.934833756268398e-06,
      "loss": 0.0012,
      "step": 79970
    },
    {
      "epoch": 1.3086803567045733,
      "grad_norm": 0.16837568581104279,
      "learning_rate": 6.933955943286197e-06,
      "loss": 0.0028,
      "step": 79980
    },
    {
      "epoch": 1.3088439826556493,
      "grad_norm": 0.0725744217634201,
      "learning_rate": 6.933078060202742e-06,
      "loss": 0.0038,
      "step": 79990
    },
    {
      "epoch": 1.309007608606725,
      "grad_norm": 0.07453279197216034,
      "learning_rate": 6.932200107049854e-06,
      "loss": 0.0023,
      "step": 80000
    },
    {
      "epoch": 1.3091712345578008,
      "grad_norm": 0.2862430810928345,
      "learning_rate": 6.931322083859355e-06,
      "loss": 0.0057,
      "step": 80010
    },
    {
      "epoch": 1.3093348605088768,
      "grad_norm": 0.08960889279842377,
      "learning_rate": 6.930443990663074e-06,
      "loss": 0.0022,
      "step": 80020
    },
    {
      "epoch": 1.3094984864599526,
      "grad_norm": 0.24223047494888306,
      "learning_rate": 6.929565827492837e-06,
      "loss": 0.002,
      "step": 80030
    },
    {
      "epoch": 1.3096621124110284,
      "grad_norm": 0.20139755308628082,
      "learning_rate": 6.928687594380475e-06,
      "loss": 0.0029,
      "step": 80040
    },
    {
      "epoch": 1.3098257383621044,
      "grad_norm": 0.04352078586816788,
      "learning_rate": 6.927809291357823e-06,
      "loss": 0.0021,
      "step": 80050
    },
    {
      "epoch": 1.3099893643131801,
      "grad_norm": 0.08494716137647629,
      "learning_rate": 6.926930918456719e-06,
      "loss": 0.0043,
      "step": 80060
    },
    {
      "epoch": 1.310152990264256,
      "grad_norm": 0.25197190046310425,
      "learning_rate": 6.926052475709002e-06,
      "loss": 0.0021,
      "step": 80070
    },
    {
      "epoch": 1.3103166162153317,
      "grad_norm": 0.2362338751554489,
      "learning_rate": 6.9251739631465095e-06,
      "loss": 0.0031,
      "step": 80080
    },
    {
      "epoch": 1.3104802421664075,
      "grad_norm": 0.1120762825012207,
      "learning_rate": 6.9242953808010896e-06,
      "loss": 0.0018,
      "step": 80090
    },
    {
      "epoch": 1.3106438681174835,
      "grad_norm": 0.17322848737239838,
      "learning_rate": 6.923416728704585e-06,
      "loss": 0.0013,
      "step": 80100
    },
    {
      "epoch": 1.3108074940685592,
      "grad_norm": 0.05941097438335419,
      "learning_rate": 6.922538006888848e-06,
      "loss": 0.0015,
      "step": 80110
    },
    {
      "epoch": 1.310971120019635,
      "grad_norm": 0.14604979753494263,
      "learning_rate": 6.921659215385729e-06,
      "loss": 0.0019,
      "step": 80120
    },
    {
      "epoch": 1.311134745970711,
      "grad_norm": 0.08937562257051468,
      "learning_rate": 6.9207803542270815e-06,
      "loss": 0.0016,
      "step": 80130
    },
    {
      "epoch": 1.3112983719217868,
      "grad_norm": 0.1113855242729187,
      "learning_rate": 6.919901423444764e-06,
      "loss": 0.0024,
      "step": 80140
    },
    {
      "epoch": 1.3114619978728626,
      "grad_norm": 0.026293085888028145,
      "learning_rate": 6.919022423070632e-06,
      "loss": 0.0013,
      "step": 80150
    },
    {
      "epoch": 1.3116256238239385,
      "grad_norm": 0.16834822297096252,
      "learning_rate": 6.91814335313655e-06,
      "loss": 0.0021,
      "step": 80160
    },
    {
      "epoch": 1.3117892497750143,
      "grad_norm": 0.17871636152267456,
      "learning_rate": 6.917264213674384e-06,
      "loss": 0.0018,
      "step": 80170
    },
    {
      "epoch": 1.31195287572609,
      "grad_norm": 0.08431093394756317,
      "learning_rate": 6.916385004715997e-06,
      "loss": 0.0014,
      "step": 80180
    },
    {
      "epoch": 1.312116501677166,
      "grad_norm": 0.0246613547205925,
      "learning_rate": 6.915505726293258e-06,
      "loss": 0.002,
      "step": 80190
    },
    {
      "epoch": 1.3122801276282419,
      "grad_norm": 0.045339860022068024,
      "learning_rate": 6.914626378438041e-06,
      "loss": 0.0025,
      "step": 80200
    },
    {
      "epoch": 1.3124437535793176,
      "grad_norm": 0.049601420760154724,
      "learning_rate": 6.91374696118222e-06,
      "loss": 0.0019,
      "step": 80210
    },
    {
      "epoch": 1.3126073795303936,
      "grad_norm": 0.14835095405578613,
      "learning_rate": 6.912867474557672e-06,
      "loss": 0.0022,
      "step": 80220
    },
    {
      "epoch": 1.3127710054814694,
      "grad_norm": 0.029804671183228493,
      "learning_rate": 6.9119879185962736e-06,
      "loss": 0.0028,
      "step": 80230
    },
    {
      "epoch": 1.3129346314325452,
      "grad_norm": 0.12879392504692078,
      "learning_rate": 6.911108293329909e-06,
      "loss": 0.0016,
      "step": 80240
    },
    {
      "epoch": 1.313098257383621,
      "grad_norm": 0.06669005751609802,
      "learning_rate": 6.910228598790463e-06,
      "loss": 0.0026,
      "step": 80250
    },
    {
      "epoch": 1.313261883334697,
      "grad_norm": 0.12902788817882538,
      "learning_rate": 6.90934883500982e-06,
      "loss": 0.0018,
      "step": 80260
    },
    {
      "epoch": 1.3134255092857727,
      "grad_norm": 0.05246992036700249,
      "learning_rate": 6.908469002019872e-06,
      "loss": 0.0011,
      "step": 80270
    },
    {
      "epoch": 1.3135891352368485,
      "grad_norm": 0.12925085425376892,
      "learning_rate": 6.907589099852508e-06,
      "loss": 0.0019,
      "step": 80280
    },
    {
      "epoch": 1.3137527611879243,
      "grad_norm": 0.07201463729143143,
      "learning_rate": 6.906709128539626e-06,
      "loss": 0.0012,
      "step": 80290
    },
    {
      "epoch": 1.3139163871390003,
      "grad_norm": 0.13973310589790344,
      "learning_rate": 6.905829088113118e-06,
      "loss": 0.0018,
      "step": 80300
    },
    {
      "epoch": 1.314080013090076,
      "grad_norm": 0.10512196272611618,
      "learning_rate": 6.904948978604888e-06,
      "loss": 0.0018,
      "step": 80310
    },
    {
      "epoch": 1.3142436390411518,
      "grad_norm": 0.1256009340286255,
      "learning_rate": 6.904068800046837e-06,
      "loss": 0.0023,
      "step": 80320
    },
    {
      "epoch": 1.3144072649922278,
      "grad_norm": 0.13715624809265137,
      "learning_rate": 6.903188552470867e-06,
      "loss": 0.0019,
      "step": 80330
    },
    {
      "epoch": 1.3145708909433036,
      "grad_norm": 0.15711772441864014,
      "learning_rate": 6.902308235908885e-06,
      "loss": 0.0022,
      "step": 80340
    },
    {
      "epoch": 1.3147345168943794,
      "grad_norm": 0.09801596403121948,
      "learning_rate": 6.901427850392804e-06,
      "loss": 0.0026,
      "step": 80350
    },
    {
      "epoch": 1.3148981428454554,
      "grad_norm": 0.14305567741394043,
      "learning_rate": 6.900547395954531e-06,
      "loss": 0.0017,
      "step": 80360
    },
    {
      "epoch": 1.3150617687965311,
      "grad_norm": 0.18697737157344818,
      "learning_rate": 6.899666872625986e-06,
      "loss": 0.0026,
      "step": 80370
    },
    {
      "epoch": 1.315225394747607,
      "grad_norm": 0.06310805678367615,
      "learning_rate": 6.898786280439079e-06,
      "loss": 0.0026,
      "step": 80380
    },
    {
      "epoch": 1.315389020698683,
      "grad_norm": 0.1619984656572342,
      "learning_rate": 6.897905619425735e-06,
      "loss": 0.0023,
      "step": 80390
    },
    {
      "epoch": 1.3155526466497587,
      "grad_norm": 0.05076758936047554,
      "learning_rate": 6.897024889617873e-06,
      "loss": 0.0025,
      "step": 80400
    },
    {
      "epoch": 1.3157162726008345,
      "grad_norm": 0.17684154212474823,
      "learning_rate": 6.896144091047417e-06,
      "loss": 0.0037,
      "step": 80410
    },
    {
      "epoch": 1.3158798985519105,
      "grad_norm": 0.13992469012737274,
      "learning_rate": 6.895263223746297e-06,
      "loss": 0.0031,
      "step": 80420
    },
    {
      "epoch": 1.3160435245029862,
      "grad_norm": 0.0648011714220047,
      "learning_rate": 6.8943822877464395e-06,
      "loss": 0.0011,
      "step": 80430
    },
    {
      "epoch": 1.316207150454062,
      "grad_norm": 0.0791841596364975,
      "learning_rate": 6.893501283079778e-06,
      "loss": 0.0016,
      "step": 80440
    },
    {
      "epoch": 1.3163707764051378,
      "grad_norm": 0.17516595125198364,
      "learning_rate": 6.892620209778244e-06,
      "loss": 0.0021,
      "step": 80450
    },
    {
      "epoch": 1.3165344023562138,
      "grad_norm": 0.041413962841033936,
      "learning_rate": 6.891739067873777e-06,
      "loss": 0.0012,
      "step": 80460
    },
    {
      "epoch": 1.3166980283072895,
      "grad_norm": 0.02108311466872692,
      "learning_rate": 6.890857857398317e-06,
      "loss": 0.0019,
      "step": 80470
    },
    {
      "epoch": 1.3168616542583653,
      "grad_norm": 0.11779098957777023,
      "learning_rate": 6.889976578383801e-06,
      "loss": 0.0017,
      "step": 80480
    },
    {
      "epoch": 1.317025280209441,
      "grad_norm": 0.05293414369225502,
      "learning_rate": 6.889095230862179e-06,
      "loss": 0.0016,
      "step": 80490
    },
    {
      "epoch": 1.317188906160517,
      "grad_norm": 0.07809446007013321,
      "learning_rate": 6.888213814865394e-06,
      "loss": 0.0026,
      "step": 80500
    },
    {
      "epoch": 1.3173525321115929,
      "grad_norm": 0.22116778790950775,
      "learning_rate": 6.887332330425398e-06,
      "loss": 0.0026,
      "step": 80510
    },
    {
      "epoch": 1.3175161580626686,
      "grad_norm": 0.07223636656999588,
      "learning_rate": 6.886450777574141e-06,
      "loss": 0.0011,
      "step": 80520
    },
    {
      "epoch": 1.3176797840137446,
      "grad_norm": 0.4159006178379059,
      "learning_rate": 6.885569156343577e-06,
      "loss": 0.0019,
      "step": 80530
    },
    {
      "epoch": 1.3178434099648204,
      "grad_norm": 0.16960126161575317,
      "learning_rate": 6.8846874667656614e-06,
      "loss": 0.0015,
      "step": 80540
    },
    {
      "epoch": 1.3180070359158962,
      "grad_norm": 0.18952810764312744,
      "learning_rate": 6.883805708872355e-06,
      "loss": 0.0021,
      "step": 80550
    },
    {
      "epoch": 1.3181706618669722,
      "grad_norm": 0.10406055301427841,
      "learning_rate": 6.8829238826956205e-06,
      "loss": 0.0038,
      "step": 80560
    },
    {
      "epoch": 1.318334287818048,
      "grad_norm": 0.05601206421852112,
      "learning_rate": 6.882041988267421e-06,
      "loss": 0.0016,
      "step": 80570
    },
    {
      "epoch": 1.3184979137691237,
      "grad_norm": 0.040606867522001266,
      "learning_rate": 6.881160025619722e-06,
      "loss": 0.0022,
      "step": 80580
    },
    {
      "epoch": 1.3186615397201997,
      "grad_norm": 0.1328790932893753,
      "learning_rate": 6.880277994784492e-06,
      "loss": 0.0029,
      "step": 80590
    },
    {
      "epoch": 1.3188251656712755,
      "grad_norm": 0.061259590089321136,
      "learning_rate": 6.879395895793706e-06,
      "loss": 0.0016,
      "step": 80600
    },
    {
      "epoch": 1.3189887916223513,
      "grad_norm": 0.08032367378473282,
      "learning_rate": 6.878513728679335e-06,
      "loss": 0.001,
      "step": 80610
    },
    {
      "epoch": 1.3191524175734273,
      "grad_norm": 0.1469169408082962,
      "learning_rate": 6.877631493473356e-06,
      "loss": 0.002,
      "step": 80620
    },
    {
      "epoch": 1.319316043524503,
      "grad_norm": 0.033294882625341415,
      "learning_rate": 6.876749190207749e-06,
      "loss": 0.0014,
      "step": 80630
    },
    {
      "epoch": 1.3194796694755788,
      "grad_norm": 0.2902836799621582,
      "learning_rate": 6.875866818914494e-06,
      "loss": 0.0021,
      "step": 80640
    },
    {
      "epoch": 1.3196432954266546,
      "grad_norm": 0.08702889829874039,
      "learning_rate": 6.874984379625574e-06,
      "loss": 0.0018,
      "step": 80650
    },
    {
      "epoch": 1.3198069213777306,
      "grad_norm": 0.1536441296339035,
      "learning_rate": 6.874101872372978e-06,
      "loss": 0.0028,
      "step": 80660
    },
    {
      "epoch": 1.3199705473288064,
      "grad_norm": 0.0942351371049881,
      "learning_rate": 6.873219297188695e-06,
      "loss": 0.0024,
      "step": 80670
    },
    {
      "epoch": 1.3201341732798821,
      "grad_norm": 0.038800500333309174,
      "learning_rate": 6.872336654104712e-06,
      "loss": 0.0036,
      "step": 80680
    },
    {
      "epoch": 1.320297799230958,
      "grad_norm": 0.11567391455173492,
      "learning_rate": 6.8714539431530255e-06,
      "loss": 0.0021,
      "step": 80690
    },
    {
      "epoch": 1.320461425182034,
      "grad_norm": 0.1451907455921173,
      "learning_rate": 6.870571164365631e-06,
      "loss": 0.0026,
      "step": 80700
    },
    {
      "epoch": 1.3206250511331097,
      "grad_norm": 0.09195049852132797,
      "learning_rate": 6.86968831777453e-06,
      "loss": 0.0017,
      "step": 80710
    },
    {
      "epoch": 1.3207886770841855,
      "grad_norm": 0.048784442245960236,
      "learning_rate": 6.86880540341172e-06,
      "loss": 0.0018,
      "step": 80720
    },
    {
      "epoch": 1.3209523030352615,
      "grad_norm": 0.012231039814651012,
      "learning_rate": 6.867922421309207e-06,
      "loss": 0.0032,
      "step": 80730
    },
    {
      "epoch": 1.3211159289863372,
      "grad_norm": 0.15991760790348053,
      "learning_rate": 6.867039371498993e-06,
      "loss": 0.0016,
      "step": 80740
    },
    {
      "epoch": 1.321279554937413,
      "grad_norm": 0.2143750935792923,
      "learning_rate": 6.866156254013091e-06,
      "loss": 0.0036,
      "step": 80750
    },
    {
      "epoch": 1.321443180888489,
      "grad_norm": 0.18175247311592102,
      "learning_rate": 6.865273068883509e-06,
      "loss": 0.004,
      "step": 80760
    },
    {
      "epoch": 1.3216068068395648,
      "grad_norm": 0.1462283432483673,
      "learning_rate": 6.864389816142262e-06,
      "loss": 0.0015,
      "step": 80770
    },
    {
      "epoch": 1.3217704327906405,
      "grad_norm": 0.2172851413488388,
      "learning_rate": 6.863506495821365e-06,
      "loss": 0.0016,
      "step": 80780
    },
    {
      "epoch": 1.3219340587417165,
      "grad_norm": 0.3600696623325348,
      "learning_rate": 6.862623107952836e-06,
      "loss": 0.0019,
      "step": 80790
    },
    {
      "epoch": 1.3220976846927923,
      "grad_norm": 0.11137879639863968,
      "learning_rate": 6.8617396525686955e-06,
      "loss": 0.0017,
      "step": 80800
    },
    {
      "epoch": 1.322261310643868,
      "grad_norm": 0.08805038779973984,
      "learning_rate": 6.860856129700968e-06,
      "loss": 0.0022,
      "step": 80810
    },
    {
      "epoch": 1.322424936594944,
      "grad_norm": 0.08044935762882233,
      "learning_rate": 6.85997253938168e-06,
      "loss": 0.0023,
      "step": 80820
    },
    {
      "epoch": 1.3225885625460199,
      "grad_norm": 0.07507777959108353,
      "learning_rate": 6.859088881642855e-06,
      "loss": 0.0018,
      "step": 80830
    },
    {
      "epoch": 1.3227521884970956,
      "grad_norm": 0.27787670493125916,
      "learning_rate": 6.858205156516526e-06,
      "loss": 0.0032,
      "step": 80840
    },
    {
      "epoch": 1.3229158144481714,
      "grad_norm": 0.04928312078118324,
      "learning_rate": 6.857321364034727e-06,
      "loss": 0.0019,
      "step": 80850
    },
    {
      "epoch": 1.3230794403992472,
      "grad_norm": 0.23112504184246063,
      "learning_rate": 6.856437504229493e-06,
      "loss": 0.0028,
      "step": 80860
    },
    {
      "epoch": 1.3232430663503232,
      "grad_norm": 0.20398372411727905,
      "learning_rate": 6.855553577132862e-06,
      "loss": 0.0013,
      "step": 80870
    },
    {
      "epoch": 1.323406692301399,
      "grad_norm": 0.08332142978906631,
      "learning_rate": 6.854669582776873e-06,
      "loss": 0.0014,
      "step": 80880
    },
    {
      "epoch": 1.3235703182524747,
      "grad_norm": 0.07513080537319183,
      "learning_rate": 6.853785521193568e-06,
      "loss": 0.0026,
      "step": 80890
    },
    {
      "epoch": 1.3237339442035507,
      "grad_norm": 0.07339072972536087,
      "learning_rate": 6.8529013924149945e-06,
      "loss": 0.0015,
      "step": 80900
    },
    {
      "epoch": 1.3238975701546265,
      "grad_norm": 0.22547726333141327,
      "learning_rate": 6.852017196473197e-06,
      "loss": 0.0021,
      "step": 80910
    },
    {
      "epoch": 1.3240611961057023,
      "grad_norm": 0.1536349654197693,
      "learning_rate": 6.851132933400231e-06,
      "loss": 0.0014,
      "step": 80920
    },
    {
      "epoch": 1.3242248220567783,
      "grad_norm": 0.2098226547241211,
      "learning_rate": 6.850248603228143e-06,
      "loss": 0.0015,
      "step": 80930
    },
    {
      "epoch": 1.324388448007854,
      "grad_norm": 0.046367328613996506,
      "learning_rate": 6.8493642059889886e-06,
      "loss": 0.0023,
      "step": 80940
    },
    {
      "epoch": 1.3245520739589298,
      "grad_norm": 0.17950865626335144,
      "learning_rate": 6.848479741714829e-06,
      "loss": 0.0021,
      "step": 80950
    },
    {
      "epoch": 1.3247156999100058,
      "grad_norm": 0.19667834043502808,
      "learning_rate": 6.847595210437721e-06,
      "loss": 0.0019,
      "step": 80960
    },
    {
      "epoch": 1.3248793258610816,
      "grad_norm": 0.08681071549654007,
      "learning_rate": 6.846710612189726e-06,
      "loss": 0.0012,
      "step": 80970
    },
    {
      "epoch": 1.3250429518121574,
      "grad_norm": 0.14749613404273987,
      "learning_rate": 6.845825947002911e-06,
      "loss": 0.0023,
      "step": 80980
    },
    {
      "epoch": 1.3252065777632334,
      "grad_norm": 0.13464826345443726,
      "learning_rate": 6.844941214909341e-06,
      "loss": 0.0016,
      "step": 80990
    },
    {
      "epoch": 1.3253702037143091,
      "grad_norm": 0.05621561035513878,
      "learning_rate": 6.844056415941086e-06,
      "loss": 0.0015,
      "step": 81000
    },
    {
      "epoch": 1.325533829665385,
      "grad_norm": 0.040989041328430176,
      "learning_rate": 6.843171550130219e-06,
      "loss": 0.002,
      "step": 81010
    },
    {
      "epoch": 1.325697455616461,
      "grad_norm": 0.07023930549621582,
      "learning_rate": 6.842286617508813e-06,
      "loss": 0.0039,
      "step": 81020
    },
    {
      "epoch": 1.3258610815675367,
      "grad_norm": 0.09771239757537842,
      "learning_rate": 6.8414016181089445e-06,
      "loss": 0.0027,
      "step": 81030
    },
    {
      "epoch": 1.3260247075186125,
      "grad_norm": 0.34062427282333374,
      "learning_rate": 6.840516551962692e-06,
      "loss": 0.0046,
      "step": 81040
    },
    {
      "epoch": 1.3261883334696882,
      "grad_norm": 0.07065514475107193,
      "learning_rate": 6.83963141910214e-06,
      "loss": 0.0023,
      "step": 81050
    },
    {
      "epoch": 1.326351959420764,
      "grad_norm": 0.09008131921291351,
      "learning_rate": 6.838746219559369e-06,
      "loss": 0.0014,
      "step": 81060
    },
    {
      "epoch": 1.32651558537184,
      "grad_norm": 0.028675584122538567,
      "learning_rate": 6.837860953366468e-06,
      "loss": 0.0019,
      "step": 81070
    },
    {
      "epoch": 1.3266792113229158,
      "grad_norm": 0.27796679735183716,
      "learning_rate": 6.836975620555525e-06,
      "loss": 0.0017,
      "step": 81080
    },
    {
      "epoch": 1.3268428372739915,
      "grad_norm": 0.1774597316980362,
      "learning_rate": 6.83609022115863e-06,
      "loss": 0.0026,
      "step": 81090
    },
    {
      "epoch": 1.3270064632250675,
      "grad_norm": 0.15093399584293365,
      "learning_rate": 6.835204755207877e-06,
      "loss": 0.0034,
      "step": 81100
    },
    {
      "epoch": 1.3271700891761433,
      "grad_norm": 0.17880648374557495,
      "learning_rate": 6.834319222735363e-06,
      "loss": 0.0014,
      "step": 81110
    },
    {
      "epoch": 1.327333715127219,
      "grad_norm": 0.08786498755216599,
      "learning_rate": 6.833433623773185e-06,
      "loss": 0.0011,
      "step": 81120
    },
    {
      "epoch": 1.327497341078295,
      "grad_norm": 0.17943353950977325,
      "learning_rate": 6.832547958353446e-06,
      "loss": 0.0018,
      "step": 81130
    },
    {
      "epoch": 1.3276609670293709,
      "grad_norm": 0.08343411982059479,
      "learning_rate": 6.831662226508247e-06,
      "loss": 0.0024,
      "step": 81140
    },
    {
      "epoch": 1.3278245929804466,
      "grad_norm": 0.16493742167949677,
      "learning_rate": 6.830776428269696e-06,
      "loss": 0.0011,
      "step": 81150
    },
    {
      "epoch": 1.3279882189315226,
      "grad_norm": 0.11954963207244873,
      "learning_rate": 6.829890563669897e-06,
      "loss": 0.0015,
      "step": 81160
    },
    {
      "epoch": 1.3281518448825984,
      "grad_norm": 0.028271540999412537,
      "learning_rate": 6.829004632740967e-06,
      "loss": 0.0027,
      "step": 81170
    },
    {
      "epoch": 1.3283154708336742,
      "grad_norm": 0.14057444036006927,
      "learning_rate": 6.828118635515011e-06,
      "loss": 0.0019,
      "step": 81180
    },
    {
      "epoch": 1.3284790967847502,
      "grad_norm": 0.012790342792868614,
      "learning_rate": 6.827232572024149e-06,
      "loss": 0.0019,
      "step": 81190
    },
    {
      "epoch": 1.328642722735826,
      "grad_norm": 0.11902596056461334,
      "learning_rate": 6.826346442300499e-06,
      "loss": 0.0018,
      "step": 81200
    },
    {
      "epoch": 1.3288063486869017,
      "grad_norm": 0.07261660695075989,
      "learning_rate": 6.82546024637618e-06,
      "loss": 0.002,
      "step": 81210
    },
    {
      "epoch": 1.3289699746379775,
      "grad_norm": 0.024522479623556137,
      "learning_rate": 6.824573984283315e-06,
      "loss": 0.0017,
      "step": 81220
    },
    {
      "epoch": 1.3291336005890535,
      "grad_norm": 0.08588574826717377,
      "learning_rate": 6.823687656054027e-06,
      "loss": 0.0015,
      "step": 81230
    },
    {
      "epoch": 1.3292972265401293,
      "grad_norm": 0.2700470983982086,
      "learning_rate": 6.822801261720444e-06,
      "loss": 0.0031,
      "step": 81240
    },
    {
      "epoch": 1.329460852491205,
      "grad_norm": 0.2070290446281433,
      "learning_rate": 6.821914801314698e-06,
      "loss": 0.0022,
      "step": 81250
    },
    {
      "epoch": 1.3296244784422808,
      "grad_norm": 0.05021479353308678,
      "learning_rate": 6.821028274868917e-06,
      "loss": 0.0029,
      "step": 81260
    },
    {
      "epoch": 1.3297881043933568,
      "grad_norm": 0.0749955102801323,
      "learning_rate": 6.820141682415241e-06,
      "loss": 0.0017,
      "step": 81270
    },
    {
      "epoch": 1.3299517303444326,
      "grad_norm": 0.3230012357234955,
      "learning_rate": 6.8192550239858024e-06,
      "loss": 0.003,
      "step": 81280
    },
    {
      "epoch": 1.3301153562955084,
      "grad_norm": 0.185124933719635,
      "learning_rate": 6.818368299612741e-06,
      "loss": 0.0044,
      "step": 81290
    },
    {
      "epoch": 1.3302789822465844,
      "grad_norm": 0.07615358382463455,
      "learning_rate": 6.817481509328199e-06,
      "loss": 0.0015,
      "step": 81300
    },
    {
      "epoch": 1.3304426081976601,
      "grad_norm": 0.11890163272619247,
      "learning_rate": 6.81659465316432e-06,
      "loss": 0.0022,
      "step": 81310
    },
    {
      "epoch": 1.330606234148736,
      "grad_norm": 0.09652549028396606,
      "learning_rate": 6.815707731153252e-06,
      "loss": 0.0019,
      "step": 81320
    },
    {
      "epoch": 1.330769860099812,
      "grad_norm": 0.011425946839153767,
      "learning_rate": 6.814820743327142e-06,
      "loss": 0.0022,
      "step": 81330
    },
    {
      "epoch": 1.3309334860508877,
      "grad_norm": 0.16947855055332184,
      "learning_rate": 6.8139336897181414e-06,
      "loss": 0.0015,
      "step": 81340
    },
    {
      "epoch": 1.3310971120019635,
      "grad_norm": 0.21875901520252228,
      "learning_rate": 6.813046570358406e-06,
      "loss": 0.0019,
      "step": 81350
    },
    {
      "epoch": 1.3312607379530395,
      "grad_norm": 0.3321792781352997,
      "learning_rate": 6.8121593852800885e-06,
      "loss": 0.0021,
      "step": 81360
    },
    {
      "epoch": 1.3314243639041152,
      "grad_norm": 0.12055980414152145,
      "learning_rate": 6.8112721345153495e-06,
      "loss": 0.0024,
      "step": 81370
    },
    {
      "epoch": 1.331587989855191,
      "grad_norm": 0.08639483153820038,
      "learning_rate": 6.81038481809635e-06,
      "loss": 0.0017,
      "step": 81380
    },
    {
      "epoch": 1.331751615806267,
      "grad_norm": 0.18138200044631958,
      "learning_rate": 6.8094974360552505e-06,
      "loss": 0.0021,
      "step": 81390
    },
    {
      "epoch": 1.3319152417573428,
      "grad_norm": 0.09246304631233215,
      "learning_rate": 6.808609988424217e-06,
      "loss": 0.0015,
      "step": 81400
    },
    {
      "epoch": 1.3320788677084185,
      "grad_norm": 0.01652436889708042,
      "learning_rate": 6.807722475235419e-06,
      "loss": 0.0021,
      "step": 81410
    },
    {
      "epoch": 1.3322424936594943,
      "grad_norm": 0.10271818935871124,
      "learning_rate": 6.806834896521027e-06,
      "loss": 0.0011,
      "step": 81420
    },
    {
      "epoch": 1.3324061196105703,
      "grad_norm": 0.14635935425758362,
      "learning_rate": 6.805947252313212e-06,
      "loss": 0.0015,
      "step": 81430
    },
    {
      "epoch": 1.332569745561646,
      "grad_norm": 0.015574569813907146,
      "learning_rate": 6.805059542644149e-06,
      "loss": 0.001,
      "step": 81440
    },
    {
      "epoch": 1.3327333715127219,
      "grad_norm": 0.20389361679553986,
      "learning_rate": 6.8041717675460175e-06,
      "loss": 0.002,
      "step": 81450
    },
    {
      "epoch": 1.3328969974637976,
      "grad_norm": 0.03448345139622688,
      "learning_rate": 6.803283927050994e-06,
      "loss": 0.0015,
      "step": 81460
    },
    {
      "epoch": 1.3330606234148736,
      "grad_norm": 0.164043590426445,
      "learning_rate": 6.802396021191262e-06,
      "loss": 0.0021,
      "step": 81470
    },
    {
      "epoch": 1.3332242493659494,
      "grad_norm": 0.0645124539732933,
      "learning_rate": 6.8015080499990075e-06,
      "loss": 0.0019,
      "step": 81480
    },
    {
      "epoch": 1.3333878753170252,
      "grad_norm": 0.03765098378062248,
      "learning_rate": 6.800620013506416e-06,
      "loss": 0.0013,
      "step": 81490
    },
    {
      "epoch": 1.3335515012681012,
      "grad_norm": 0.044347334653139114,
      "learning_rate": 6.799731911745675e-06,
      "loss": 0.0025,
      "step": 81500
    },
    {
      "epoch": 1.333715127219177,
      "grad_norm": 0.25207430124282837,
      "learning_rate": 6.79884374474898e-06,
      "loss": 0.0035,
      "step": 81510
    },
    {
      "epoch": 1.3338787531702527,
      "grad_norm": 0.0751473531126976,
      "learning_rate": 6.797955512548521e-06,
      "loss": 0.003,
      "step": 81520
    },
    {
      "epoch": 1.3340423791213287,
      "grad_norm": 0.1638592928647995,
      "learning_rate": 6.797067215176497e-06,
      "loss": 0.0022,
      "step": 81530
    },
    {
      "epoch": 1.3342060050724045,
      "grad_norm": 0.0020773319993168116,
      "learning_rate": 6.796178852665104e-06,
      "loss": 0.0022,
      "step": 81540
    },
    {
      "epoch": 1.3343696310234803,
      "grad_norm": 0.16474537551403046,
      "learning_rate": 6.795290425046546e-06,
      "loss": 0.002,
      "step": 81550
    },
    {
      "epoch": 1.3345332569745563,
      "grad_norm": 0.20047356188297272,
      "learning_rate": 6.794401932353025e-06,
      "loss": 0.0018,
      "step": 81560
    },
    {
      "epoch": 1.334696882925632,
      "grad_norm": 0.10200228542089462,
      "learning_rate": 6.793513374616746e-06,
      "loss": 0.0028,
      "step": 81570
    },
    {
      "epoch": 1.3348605088767078,
      "grad_norm": 0.19652168452739716,
      "learning_rate": 6.792624751869919e-06,
      "loss": 0.0021,
      "step": 81580
    },
    {
      "epoch": 1.3350241348277838,
      "grad_norm": 0.15968818962574005,
      "learning_rate": 6.791736064144753e-06,
      "loss": 0.0015,
      "step": 81590
    },
    {
      "epoch": 1.3351877607788596,
      "grad_norm": 0.19664816558361053,
      "learning_rate": 6.79084731147346e-06,
      "loss": 0.0028,
      "step": 81600
    },
    {
      "epoch": 1.3353513867299354,
      "grad_norm": 0.5528716444969177,
      "learning_rate": 6.789958493888256e-06,
      "loss": 0.0027,
      "step": 81610
    },
    {
      "epoch": 1.3355150126810111,
      "grad_norm": 0.03361961990594864,
      "learning_rate": 6.789069611421358e-06,
      "loss": 0.0014,
      "step": 81620
    },
    {
      "epoch": 1.335678638632087,
      "grad_norm": 0.26299917697906494,
      "learning_rate": 6.788180664104989e-06,
      "loss": 0.003,
      "step": 81630
    },
    {
      "epoch": 1.335842264583163,
      "grad_norm": 0.07342177629470825,
      "learning_rate": 6.787291651971367e-06,
      "loss": 0.0016,
      "step": 81640
    },
    {
      "epoch": 1.3360058905342387,
      "grad_norm": 0.08538855612277985,
      "learning_rate": 6.786402575052718e-06,
      "loss": 0.0021,
      "step": 81650
    },
    {
      "epoch": 1.3361695164853145,
      "grad_norm": 0.09256561845541,
      "learning_rate": 6.78551343338127e-06,
      "loss": 0.004,
      "step": 81660
    },
    {
      "epoch": 1.3363331424363905,
      "grad_norm": 0.0368439182639122,
      "learning_rate": 6.784624226989251e-06,
      "loss": 0.0018,
      "step": 81670
    },
    {
      "epoch": 1.3364967683874662,
      "grad_norm": 0.0727786123752594,
      "learning_rate": 6.783734955908894e-06,
      "loss": 0.0011,
      "step": 81680
    },
    {
      "epoch": 1.336660394338542,
      "grad_norm": 0.04618093743920326,
      "learning_rate": 6.78284562017243e-06,
      "loss": 0.0029,
      "step": 81690
    },
    {
      "epoch": 1.336824020289618,
      "grad_norm": 0.05337835103273392,
      "learning_rate": 6.781956219812098e-06,
      "loss": 0.0011,
      "step": 81700
    },
    {
      "epoch": 1.3369876462406938,
      "grad_norm": 0.09037453681230545,
      "learning_rate": 6.781066754860137e-06,
      "loss": 0.0023,
      "step": 81710
    },
    {
      "epoch": 1.3371512721917695,
      "grad_norm": 0.29447680711746216,
      "learning_rate": 6.780177225348785e-06,
      "loss": 0.002,
      "step": 81720
    },
    {
      "epoch": 1.3373148981428455,
      "grad_norm": 0.39623990654945374,
      "learning_rate": 6.7792876313102895e-06,
      "loss": 0.0022,
      "step": 81730
    },
    {
      "epoch": 1.3374785240939213,
      "grad_norm": 0.2530408501625061,
      "learning_rate": 6.778397972776892e-06,
      "loss": 0.0041,
      "step": 81740
    },
    {
      "epoch": 1.337642150044997,
      "grad_norm": 0.29657208919525146,
      "learning_rate": 6.777508249780843e-06,
      "loss": 0.0024,
      "step": 81750
    },
    {
      "epoch": 1.337805775996073,
      "grad_norm": 0.28433340787887573,
      "learning_rate": 6.776618462354391e-06,
      "loss": 0.0019,
      "step": 81760
    },
    {
      "epoch": 1.3379694019471489,
      "grad_norm": 0.08459485322237015,
      "learning_rate": 6.775728610529791e-06,
      "loss": 0.0027,
      "step": 81770
    },
    {
      "epoch": 1.3381330278982246,
      "grad_norm": 0.012726878747344017,
      "learning_rate": 6.774838694339297e-06,
      "loss": 0.0019,
      "step": 81780
    },
    {
      "epoch": 1.3382966538493006,
      "grad_norm": 0.08615007251501083,
      "learning_rate": 6.773948713815164e-06,
      "loss": 0.0012,
      "step": 81790
    },
    {
      "epoch": 1.3384602798003764,
      "grad_norm": 0.030178209766745567,
      "learning_rate": 6.773058668989655e-06,
      "loss": 0.0016,
      "step": 81800
    },
    {
      "epoch": 1.3386239057514522,
      "grad_norm": 0.08457730710506439,
      "learning_rate": 6.77216855989503e-06,
      "loss": 0.0011,
      "step": 81810
    },
    {
      "epoch": 1.338787531702528,
      "grad_norm": 0.34581729769706726,
      "learning_rate": 6.771278386563556e-06,
      "loss": 0.0033,
      "step": 81820
    },
    {
      "epoch": 1.3389511576536037,
      "grad_norm": 0.06941339373588562,
      "learning_rate": 6.770388149027496e-06,
      "loss": 0.0022,
      "step": 81830
    },
    {
      "epoch": 1.3391147836046797,
      "grad_norm": 0.0813087597489357,
      "learning_rate": 6.76949784731912e-06,
      "loss": 0.002,
      "step": 81840
    },
    {
      "epoch": 1.3392784095557555,
      "grad_norm": 0.0886281281709671,
      "learning_rate": 6.7686074814707015e-06,
      "loss": 0.001,
      "step": 81850
    },
    {
      "epoch": 1.3394420355068313,
      "grad_norm": 0.12007439136505127,
      "learning_rate": 6.76771705151451e-06,
      "loss": 0.0018,
      "step": 81860
    },
    {
      "epoch": 1.3396056614579073,
      "grad_norm": 0.018145669251680374,
      "learning_rate": 6.766826557482826e-06,
      "loss": 0.0017,
      "step": 81870
    },
    {
      "epoch": 1.339769287408983,
      "grad_norm": 0.20330819487571716,
      "learning_rate": 6.765935999407927e-06,
      "loss": 0.0016,
      "step": 81880
    },
    {
      "epoch": 1.3399329133600588,
      "grad_norm": 0.018880534917116165,
      "learning_rate": 6.765045377322091e-06,
      "loss": 0.0025,
      "step": 81890
    },
    {
      "epoch": 1.3400965393111348,
      "grad_norm": 0.039413075894117355,
      "learning_rate": 6.764154691257601e-06,
      "loss": 0.0011,
      "step": 81900
    },
    {
      "epoch": 1.3402601652622106,
      "grad_norm": 0.08342961221933365,
      "learning_rate": 6.763263941246745e-06,
      "loss": 0.001,
      "step": 81910
    },
    {
      "epoch": 1.3404237912132864,
      "grad_norm": 0.17808179557323456,
      "learning_rate": 6.7623731273218084e-06,
      "loss": 0.0035,
      "step": 81920
    },
    {
      "epoch": 1.3405874171643624,
      "grad_norm": 0.355609267950058,
      "learning_rate": 6.761482249515082e-06,
      "loss": 0.0031,
      "step": 81930
    },
    {
      "epoch": 1.3407510431154381,
      "grad_norm": 0.1452360302209854,
      "learning_rate": 6.760591307858858e-06,
      "loss": 0.0039,
      "step": 81940
    },
    {
      "epoch": 1.340914669066514,
      "grad_norm": 0.12057671695947647,
      "learning_rate": 6.75970030238543e-06,
      "loss": 0.0021,
      "step": 81950
    },
    {
      "epoch": 1.34107829501759,
      "grad_norm": 0.08325853943824768,
      "learning_rate": 6.758809233127095e-06,
      "loss": 0.0033,
      "step": 81960
    },
    {
      "epoch": 1.3412419209686657,
      "grad_norm": 0.3102765381336212,
      "learning_rate": 6.757918100116153e-06,
      "loss": 0.0031,
      "step": 81970
    },
    {
      "epoch": 1.3414055469197415,
      "grad_norm": 0.09912633895874023,
      "learning_rate": 6.757026903384906e-06,
      "loss": 0.0015,
      "step": 81980
    },
    {
      "epoch": 1.3415691728708175,
      "grad_norm": 0.021710427477955818,
      "learning_rate": 6.756135642965655e-06,
      "loss": 0.0021,
      "step": 81990
    },
    {
      "epoch": 1.3417327988218932,
      "grad_norm": 0.08131515979766846,
      "learning_rate": 6.755244318890708e-06,
      "loss": 0.0031,
      "step": 82000
    },
    {
      "epoch": 1.341896424772969,
      "grad_norm": 0.45112860202789307,
      "learning_rate": 6.754352931192373e-06,
      "loss": 0.0019,
      "step": 82010
    },
    {
      "epoch": 1.3420600507240448,
      "grad_norm": 0.19981126487255096,
      "learning_rate": 6.75346147990296e-06,
      "loss": 0.0024,
      "step": 82020
    },
    {
      "epoch": 1.3422236766751205,
      "grad_norm": 0.09313564002513885,
      "learning_rate": 6.752569965054786e-06,
      "loss": 0.0011,
      "step": 82030
    },
    {
      "epoch": 1.3423873026261965,
      "grad_norm": 0.22114932537078857,
      "learning_rate": 6.751678386680159e-06,
      "loss": 0.0035,
      "step": 82040
    },
    {
      "epoch": 1.3425509285772723,
      "grad_norm": 0.1432260274887085,
      "learning_rate": 6.7507867448114025e-06,
      "loss": 0.003,
      "step": 82050
    },
    {
      "epoch": 1.342714554528348,
      "grad_norm": 0.04274013265967369,
      "learning_rate": 6.749895039480834e-06,
      "loss": 0.0019,
      "step": 82060
    },
    {
      "epoch": 1.342878180479424,
      "grad_norm": 0.14999033510684967,
      "learning_rate": 6.749003270720776e-06,
      "loss": 0.0022,
      "step": 82070
    },
    {
      "epoch": 1.3430418064304999,
      "grad_norm": 0.09435473382472992,
      "learning_rate": 6.748111438563554e-06,
      "loss": 0.0013,
      "step": 82080
    },
    {
      "epoch": 1.3432054323815756,
      "grad_norm": 0.1602046638727188,
      "learning_rate": 6.747219543041493e-06,
      "loss": 0.0056,
      "step": 82090
    },
    {
      "epoch": 1.3433690583326516,
      "grad_norm": 0.1247342899441719,
      "learning_rate": 6.7463275841869234e-06,
      "loss": 0.0019,
      "step": 82100
    },
    {
      "epoch": 1.3435326842837274,
      "grad_norm": 0.07622984796762466,
      "learning_rate": 6.745435562032175e-06,
      "loss": 0.002,
      "step": 82110
    },
    {
      "epoch": 1.3436963102348032,
      "grad_norm": 0.06252642720937729,
      "learning_rate": 6.744543476609582e-06,
      "loss": 0.0029,
      "step": 82120
    },
    {
      "epoch": 1.3438599361858792,
      "grad_norm": 0.08780164271593094,
      "learning_rate": 6.743651327951484e-06,
      "loss": 0.0029,
      "step": 82130
    },
    {
      "epoch": 1.344023562136955,
      "grad_norm": 0.10929341614246368,
      "learning_rate": 6.742759116090214e-06,
      "loss": 0.0014,
      "step": 82140
    },
    {
      "epoch": 1.3441871880880307,
      "grad_norm": 0.16372643411159515,
      "learning_rate": 6.741866841058114e-06,
      "loss": 0.0039,
      "step": 82150
    },
    {
      "epoch": 1.3443508140391067,
      "grad_norm": 0.2781262695789337,
      "learning_rate": 6.740974502887528e-06,
      "loss": 0.0016,
      "step": 82160
    },
    {
      "epoch": 1.3445144399901825,
      "grad_norm": 0.20046466588974,
      "learning_rate": 6.740082101610801e-06,
      "loss": 0.0027,
      "step": 82170
    },
    {
      "epoch": 1.3446780659412583,
      "grad_norm": 0.07419824600219727,
      "learning_rate": 6.739189637260279e-06,
      "loss": 0.0019,
      "step": 82180
    },
    {
      "epoch": 1.344841691892334,
      "grad_norm": 0.02324909158051014,
      "learning_rate": 6.7382971098683146e-06,
      "loss": 0.0028,
      "step": 82190
    },
    {
      "epoch": 1.34500531784341,
      "grad_norm": 0.003428959520533681,
      "learning_rate": 6.7374045194672565e-06,
      "loss": 0.0025,
      "step": 82200
    },
    {
      "epoch": 1.3451689437944858,
      "grad_norm": 0.04728710278868675,
      "learning_rate": 6.736511866089461e-06,
      "loss": 0.0016,
      "step": 82210
    },
    {
      "epoch": 1.3453325697455616,
      "grad_norm": 0.16150616109371185,
      "learning_rate": 6.7356191497672825e-06,
      "loss": 0.0026,
      "step": 82220
    },
    {
      "epoch": 1.3454961956966374,
      "grad_norm": 0.19331979751586914,
      "learning_rate": 6.734726370533084e-06,
      "loss": 0.007,
      "step": 82230
    },
    {
      "epoch": 1.3456598216477134,
      "grad_norm": 0.10543633252382278,
      "learning_rate": 6.733833528419222e-06,
      "loss": 0.0026,
      "step": 82240
    },
    {
      "epoch": 1.3458234475987891,
      "grad_norm": 0.04915414750576019,
      "learning_rate": 6.732940623458061e-06,
      "loss": 0.0024,
      "step": 82250
    },
    {
      "epoch": 1.345987073549865,
      "grad_norm": 0.12134712189435959,
      "learning_rate": 6.732047655681968e-06,
      "loss": 0.0015,
      "step": 82260
    },
    {
      "epoch": 1.346150699500941,
      "grad_norm": 0.03422749787569046,
      "learning_rate": 6.73115462512331e-06,
      "loss": 0.0019,
      "step": 82270
    },
    {
      "epoch": 1.3463143254520167,
      "grad_norm": 0.14769293367862701,
      "learning_rate": 6.73026153181446e-06,
      "loss": 0.0019,
      "step": 82280
    },
    {
      "epoch": 1.3464779514030925,
      "grad_norm": 0.08648399263620377,
      "learning_rate": 6.729368375787786e-06,
      "loss": 0.0014,
      "step": 82290
    },
    {
      "epoch": 1.3466415773541685,
      "grad_norm": 0.0378507636487484,
      "learning_rate": 6.728475157075666e-06,
      "loss": 0.003,
      "step": 82300
    },
    {
      "epoch": 1.3468052033052442,
      "grad_norm": 0.04808375984430313,
      "learning_rate": 6.727581875710474e-06,
      "loss": 0.0035,
      "step": 82310
    },
    {
      "epoch": 1.34696882925632,
      "grad_norm": 0.08152385801076889,
      "learning_rate": 6.726688531724592e-06,
      "loss": 0.0026,
      "step": 82320
    },
    {
      "epoch": 1.347132455207396,
      "grad_norm": 0.14797881245613098,
      "learning_rate": 6.725795125150401e-06,
      "loss": 0.0037,
      "step": 82330
    },
    {
      "epoch": 1.3472960811584718,
      "grad_norm": 0.11176542192697525,
      "learning_rate": 6.724901656020284e-06,
      "loss": 0.0023,
      "step": 82340
    },
    {
      "epoch": 1.3474597071095475,
      "grad_norm": 0.08064939081668854,
      "learning_rate": 6.724008124366628e-06,
      "loss": 0.0022,
      "step": 82350
    },
    {
      "epoch": 1.3476233330606235,
      "grad_norm": 0.11147861927747726,
      "learning_rate": 6.723114530221822e-06,
      "loss": 0.0016,
      "step": 82360
    },
    {
      "epoch": 1.3477869590116993,
      "grad_norm": 0.09922857582569122,
      "learning_rate": 6.722220873618254e-06,
      "loss": 0.0021,
      "step": 82370
    },
    {
      "epoch": 1.347950584962775,
      "grad_norm": 0.08916383236646652,
      "learning_rate": 6.72132715458832e-06,
      "loss": 0.0018,
      "step": 82380
    },
    {
      "epoch": 1.3481142109138509,
      "grad_norm": 0.17162224650382996,
      "learning_rate": 6.720433373164412e-06,
      "loss": 0.0035,
      "step": 82390
    },
    {
      "epoch": 1.3482778368649269,
      "grad_norm": 0.11682993918657303,
      "learning_rate": 6.71953952937893e-06,
      "loss": 0.0024,
      "step": 82400
    },
    {
      "epoch": 1.3484414628160026,
      "grad_norm": 0.05537468567490578,
      "learning_rate": 6.718645623264273e-06,
      "loss": 0.002,
      "step": 82410
    },
    {
      "epoch": 1.3486050887670784,
      "grad_norm": 0.09403576701879501,
      "learning_rate": 6.717751654852843e-06,
      "loss": 0.0016,
      "step": 82420
    },
    {
      "epoch": 1.3487687147181542,
      "grad_norm": 0.19460827112197876,
      "learning_rate": 6.716857624177045e-06,
      "loss": 0.0032,
      "step": 82430
    },
    {
      "epoch": 1.3489323406692302,
      "grad_norm": 0.26909196376800537,
      "learning_rate": 6.7159635312692825e-06,
      "loss": 0.0012,
      "step": 82440
    },
    {
      "epoch": 1.349095966620306,
      "grad_norm": 0.10852129012346268,
      "learning_rate": 6.715069376161968e-06,
      "loss": 0.002,
      "step": 82450
    },
    {
      "epoch": 1.3492595925713817,
      "grad_norm": 0.29933688044548035,
      "learning_rate": 6.71417515888751e-06,
      "loss": 0.0031,
      "step": 82460
    },
    {
      "epoch": 1.3494232185224577,
      "grad_norm": 0.16139064729213715,
      "learning_rate": 6.713280879478321e-06,
      "loss": 0.0021,
      "step": 82470
    },
    {
      "epoch": 1.3495868444735335,
      "grad_norm": 0.07797890901565552,
      "learning_rate": 6.7123865379668196e-06,
      "loss": 0.0015,
      "step": 82480
    },
    {
      "epoch": 1.3497504704246093,
      "grad_norm": 0.1679505854845047,
      "learning_rate": 6.711492134385421e-06,
      "loss": 0.0022,
      "step": 82490
    },
    {
      "epoch": 1.3499140963756853,
      "grad_norm": 0.03788674622774124,
      "learning_rate": 6.710597668766546e-06,
      "loss": 0.0025,
      "step": 82500
    },
    {
      "epoch": 1.350077722326761,
      "grad_norm": 0.2341817021369934,
      "learning_rate": 6.709703141142618e-06,
      "loss": 0.0016,
      "step": 82510
    },
    {
      "epoch": 1.3502413482778368,
      "grad_norm": 0.04366068169474602,
      "learning_rate": 6.708808551546059e-06,
      "loss": 0.0026,
      "step": 82520
    },
    {
      "epoch": 1.3504049742289128,
      "grad_norm": 0.08851949870586395,
      "learning_rate": 6.707913900009298e-06,
      "loss": 0.0028,
      "step": 82530
    },
    {
      "epoch": 1.3505686001799886,
      "grad_norm": 0.4423024356365204,
      "learning_rate": 6.707019186564761e-06,
      "loss": 0.0046,
      "step": 82540
    },
    {
      "epoch": 1.3507322261310644,
      "grad_norm": 0.05465300753712654,
      "learning_rate": 6.706124411244883e-06,
      "loss": 0.0021,
      "step": 82550
    },
    {
      "epoch": 1.3508958520821404,
      "grad_norm": 0.14524762332439423,
      "learning_rate": 6.705229574082095e-06,
      "loss": 0.0013,
      "step": 82560
    },
    {
      "epoch": 1.3510594780332161,
      "grad_norm": 0.0880657285451889,
      "learning_rate": 6.7043346751088344e-06,
      "loss": 0.0028,
      "step": 82570
    },
    {
      "epoch": 1.351223103984292,
      "grad_norm": 0.08138912171125412,
      "learning_rate": 6.7034397143575384e-06,
      "loss": 0.0028,
      "step": 82580
    },
    {
      "epoch": 1.3513867299353677,
      "grad_norm": 0.13508175313472748,
      "learning_rate": 6.702544691860645e-06,
      "loss": 0.002,
      "step": 82590
    },
    {
      "epoch": 1.3515503558864435,
      "grad_norm": 0.013995610177516937,
      "learning_rate": 6.7016496076506e-06,
      "loss": 0.0025,
      "step": 82600
    },
    {
      "epoch": 1.3517139818375195,
      "grad_norm": 0.4418068528175354,
      "learning_rate": 6.700754461759844e-06,
      "loss": 0.0022,
      "step": 82610
    },
    {
      "epoch": 1.3518776077885952,
      "grad_norm": 0.02073877491056919,
      "learning_rate": 6.699859254220829e-06,
      "loss": 0.0031,
      "step": 82620
    },
    {
      "epoch": 1.352041233739671,
      "grad_norm": 0.06183016300201416,
      "learning_rate": 6.698963985066001e-06,
      "loss": 0.0011,
      "step": 82630
    },
    {
      "epoch": 1.352204859690747,
      "grad_norm": 0.0949455201625824,
      "learning_rate": 6.698068654327811e-06,
      "loss": 0.003,
      "step": 82640
    },
    {
      "epoch": 1.3523684856418228,
      "grad_norm": 0.15206284821033478,
      "learning_rate": 6.697173262038714e-06,
      "loss": 0.0027,
      "step": 82650
    },
    {
      "epoch": 1.3525321115928985,
      "grad_norm": 0.1209939643740654,
      "learning_rate": 6.696277808231166e-06,
      "loss": 0.0021,
      "step": 82660
    },
    {
      "epoch": 1.3526957375439745,
      "grad_norm": 0.10925428569316864,
      "learning_rate": 6.695382292937623e-06,
      "loss": 0.0015,
      "step": 82670
    },
    {
      "epoch": 1.3528593634950503,
      "grad_norm": 0.20161554217338562,
      "learning_rate": 6.694486716190548e-06,
      "loss": 0.002,
      "step": 82680
    },
    {
      "epoch": 1.353022989446126,
      "grad_norm": 0.0996408611536026,
      "learning_rate": 6.693591078022401e-06,
      "loss": 0.0062,
      "step": 82690
    },
    {
      "epoch": 1.353186615397202,
      "grad_norm": 0.020687879994511604,
      "learning_rate": 6.692695378465648e-06,
      "loss": 0.0043,
      "step": 82700
    },
    {
      "epoch": 1.3533502413482779,
      "grad_norm": 0.48303523659706116,
      "learning_rate": 6.6917996175527566e-06,
      "loss": 0.002,
      "step": 82710
    },
    {
      "epoch": 1.3535138672993536,
      "grad_norm": 0.048049282282590866,
      "learning_rate": 6.690903795316195e-06,
      "loss": 0.002,
      "step": 82720
    },
    {
      "epoch": 1.3536774932504296,
      "grad_norm": 0.05348822847008705,
      "learning_rate": 6.690007911788435e-06,
      "loss": 0.0035,
      "step": 82730
    },
    {
      "epoch": 1.3538411192015054,
      "grad_norm": 0.1339612603187561,
      "learning_rate": 6.68911196700195e-06,
      "loss": 0.002,
      "step": 82740
    },
    {
      "epoch": 1.3540047451525812,
      "grad_norm": 0.10389325022697449,
      "learning_rate": 6.688215960989214e-06,
      "loss": 0.0025,
      "step": 82750
    },
    {
      "epoch": 1.3541683711036572,
      "grad_norm": 0.2246437817811966,
      "learning_rate": 6.687319893782709e-06,
      "loss": 0.0023,
      "step": 82760
    },
    {
      "epoch": 1.354331997054733,
      "grad_norm": 0.04397423937916756,
      "learning_rate": 6.686423765414912e-06,
      "loss": 0.0015,
      "step": 82770
    },
    {
      "epoch": 1.3544956230058087,
      "grad_norm": 0.2279076874256134,
      "learning_rate": 6.685527575918308e-06,
      "loss": 0.0019,
      "step": 82780
    },
    {
      "epoch": 1.3546592489568845,
      "grad_norm": 0.2660253643989563,
      "learning_rate": 6.6846313253253804e-06,
      "loss": 0.0018,
      "step": 82790
    },
    {
      "epoch": 1.3548228749079603,
      "grad_norm": 0.0699646845459938,
      "learning_rate": 6.6837350136686155e-06,
      "loss": 0.0012,
      "step": 82800
    },
    {
      "epoch": 1.3549865008590363,
      "grad_norm": 0.09754594415426254,
      "learning_rate": 6.682838640980503e-06,
      "loss": 0.0031,
      "step": 82810
    },
    {
      "epoch": 1.355150126810112,
      "grad_norm": 0.08128782361745834,
      "learning_rate": 6.681942207293535e-06,
      "loss": 0.0024,
      "step": 82820
    },
    {
      "epoch": 1.3553137527611878,
      "grad_norm": 0.011833651922643185,
      "learning_rate": 6.681045712640203e-06,
      "loss": 0.0035,
      "step": 82830
    },
    {
      "epoch": 1.3554773787122638,
      "grad_norm": 0.07097772508859634,
      "learning_rate": 6.680149157053005e-06,
      "loss": 0.0017,
      "step": 82840
    },
    {
      "epoch": 1.3556410046633396,
      "grad_norm": 0.08166410773992538,
      "learning_rate": 6.679252540564438e-06,
      "loss": 0.0018,
      "step": 82850
    },
    {
      "epoch": 1.3558046306144154,
      "grad_norm": 0.09992627054452896,
      "learning_rate": 6.678355863207003e-06,
      "loss": 0.0021,
      "step": 82860
    },
    {
      "epoch": 1.3559682565654914,
      "grad_norm": 0.0964866578578949,
      "learning_rate": 6.677459125013202e-06,
      "loss": 0.0014,
      "step": 82870
    },
    {
      "epoch": 1.3561318825165671,
      "grad_norm": 0.0698394924402237,
      "learning_rate": 6.6765623260155385e-06,
      "loss": 0.0017,
      "step": 82880
    },
    {
      "epoch": 1.356295508467643,
      "grad_norm": 0.14303836226463318,
      "learning_rate": 6.675665466246521e-06,
      "loss": 0.0019,
      "step": 82890
    },
    {
      "epoch": 1.356459134418719,
      "grad_norm": 0.13235551118850708,
      "learning_rate": 6.6747685457386565e-06,
      "loss": 0.0023,
      "step": 82900
    },
    {
      "epoch": 1.3566227603697947,
      "grad_norm": 0.2126377820968628,
      "learning_rate": 6.673871564524458e-06,
      "loss": 0.0036,
      "step": 82910
    },
    {
      "epoch": 1.3567863863208705,
      "grad_norm": 0.11085109412670135,
      "learning_rate": 6.6729745226364385e-06,
      "loss": 0.0013,
      "step": 82920
    },
    {
      "epoch": 1.3569500122719464,
      "grad_norm": 0.18328532576560974,
      "learning_rate": 6.672077420107115e-06,
      "loss": 0.0017,
      "step": 82930
    },
    {
      "epoch": 1.3571136382230222,
      "grad_norm": 0.06743686646223068,
      "learning_rate": 6.671180256969001e-06,
      "loss": 0.0019,
      "step": 82940
    },
    {
      "epoch": 1.357277264174098,
      "grad_norm": 0.09188753366470337,
      "learning_rate": 6.670283033254621e-06,
      "loss": 0.0018,
      "step": 82950
    },
    {
      "epoch": 1.3574408901251738,
      "grad_norm": 0.18873248994350433,
      "learning_rate": 6.669385748996495e-06,
      "loss": 0.0017,
      "step": 82960
    },
    {
      "epoch": 1.3576045160762498,
      "grad_norm": 0.15171048045158386,
      "learning_rate": 6.668488404227147e-06,
      "loss": 0.0016,
      "step": 82970
    },
    {
      "epoch": 1.3577681420273255,
      "grad_norm": 0.40069663524627686,
      "learning_rate": 6.667590998979106e-06,
      "loss": 0.0023,
      "step": 82980
    },
    {
      "epoch": 1.3579317679784013,
      "grad_norm": 0.09955428540706635,
      "learning_rate": 6.666693533284898e-06,
      "loss": 0.0029,
      "step": 82990
    },
    {
      "epoch": 1.358095393929477,
      "grad_norm": 0.3799716532230377,
      "learning_rate": 6.6657960071770555e-06,
      "loss": 0.0044,
      "step": 83000
    },
    {
      "epoch": 1.358259019880553,
      "grad_norm": 0.04664906486868858,
      "learning_rate": 6.664898420688112e-06,
      "loss": 0.0015,
      "step": 83010
    },
    {
      "epoch": 1.3584226458316289,
      "grad_norm": 0.011510207317769527,
      "learning_rate": 6.664000773850602e-06,
      "loss": 0.0023,
      "step": 83020
    },
    {
      "epoch": 1.3585862717827046,
      "grad_norm": 0.218876913189888,
      "learning_rate": 6.663103066697063e-06,
      "loss": 0.0011,
      "step": 83030
    },
    {
      "epoch": 1.3587498977337806,
      "grad_norm": 0.04186856746673584,
      "learning_rate": 6.662205299260034e-06,
      "loss": 0.0019,
      "step": 83040
    },
    {
      "epoch": 1.3589135236848564,
      "grad_norm": 0.12495332211256027,
      "learning_rate": 6.6613074715720595e-06,
      "loss": 0.0028,
      "step": 83050
    },
    {
      "epoch": 1.3590771496359322,
      "grad_norm": 0.08146946132183075,
      "learning_rate": 6.660409583665681e-06,
      "loss": 0.0021,
      "step": 83060
    },
    {
      "epoch": 1.3592407755870082,
      "grad_norm": 0.16242024302482605,
      "learning_rate": 6.659511635573447e-06,
      "loss": 0.0036,
      "step": 83070
    },
    {
      "epoch": 1.359404401538084,
      "grad_norm": 0.23032371699810028,
      "learning_rate": 6.658613627327905e-06,
      "loss": 0.0014,
      "step": 83080
    },
    {
      "epoch": 1.3595680274891597,
      "grad_norm": 0.20575572550296783,
      "learning_rate": 6.657715558961604e-06,
      "loss": 0.0018,
      "step": 83090
    },
    {
      "epoch": 1.3597316534402357,
      "grad_norm": 0.08109265565872192,
      "learning_rate": 6.656817430507099e-06,
      "loss": 0.0017,
      "step": 83100
    },
    {
      "epoch": 1.3598952793913115,
      "grad_norm": 0.21918833255767822,
      "learning_rate": 6.6559192419969424e-06,
      "loss": 0.0022,
      "step": 83110
    },
    {
      "epoch": 1.3600589053423873,
      "grad_norm": 0.05995751544833183,
      "learning_rate": 6.655020993463695e-06,
      "loss": 0.0028,
      "step": 83120
    },
    {
      "epoch": 1.3602225312934633,
      "grad_norm": 0.042985714972019196,
      "learning_rate": 6.654122684939914e-06,
      "loss": 0.0021,
      "step": 83130
    },
    {
      "epoch": 1.360386157244539,
      "grad_norm": 0.158111572265625,
      "learning_rate": 6.653224316458161e-06,
      "loss": 0.0027,
      "step": 83140
    },
    {
      "epoch": 1.3605497831956148,
      "grad_norm": 0.07086026668548584,
      "learning_rate": 6.652325888051e-06,
      "loss": 0.0016,
      "step": 83150
    },
    {
      "epoch": 1.3607134091466906,
      "grad_norm": 0.14250591397285461,
      "learning_rate": 6.651427399750997e-06,
      "loss": 0.0022,
      "step": 83160
    },
    {
      "epoch": 1.3608770350977666,
      "grad_norm": 0.07695437967777252,
      "learning_rate": 6.65052885159072e-06,
      "loss": 0.001,
      "step": 83170
    },
    {
      "epoch": 1.3610406610488424,
      "grad_norm": 0.06912094354629517,
      "learning_rate": 6.649630243602738e-06,
      "loss": 0.0019,
      "step": 83180
    },
    {
      "epoch": 1.3612042869999181,
      "grad_norm": 0.050239890813827515,
      "learning_rate": 6.648731575819626e-06,
      "loss": 0.0016,
      "step": 83190
    },
    {
      "epoch": 1.361367912950994,
      "grad_norm": 0.05261654034256935,
      "learning_rate": 6.6478328482739566e-06,
      "loss": 0.002,
      "step": 83200
    },
    {
      "epoch": 1.36153153890207,
      "grad_norm": 0.08113102614879608,
      "learning_rate": 6.6469340609983056e-06,
      "loss": 0.0033,
      "step": 83210
    },
    {
      "epoch": 1.3616951648531457,
      "grad_norm": 0.09291556477546692,
      "learning_rate": 6.646035214025255e-06,
      "loss": 0.0022,
      "step": 83220
    },
    {
      "epoch": 1.3618587908042215,
      "grad_norm": 0.026314903050661087,
      "learning_rate": 6.6451363073873835e-06,
      "loss": 0.0015,
      "step": 83230
    },
    {
      "epoch": 1.3620224167552974,
      "grad_norm": 0.29929569363594055,
      "learning_rate": 6.644237341117273e-06,
      "loss": 0.0019,
      "step": 83240
    },
    {
      "epoch": 1.3621860427063732,
      "grad_norm": 0.19190314412117004,
      "learning_rate": 6.643338315247513e-06,
      "loss": 0.0033,
      "step": 83250
    },
    {
      "epoch": 1.362349668657449,
      "grad_norm": 0.20633620023727417,
      "learning_rate": 6.642439229810688e-06,
      "loss": 0.0046,
      "step": 83260
    },
    {
      "epoch": 1.362513294608525,
      "grad_norm": 0.03409310057759285,
      "learning_rate": 6.6415400848393885e-06,
      "loss": 0.0022,
      "step": 83270
    },
    {
      "epoch": 1.3626769205596008,
      "grad_norm": 0.02324938215315342,
      "learning_rate": 6.640640880366207e-06,
      "loss": 0.0019,
      "step": 83280
    },
    {
      "epoch": 1.3628405465106765,
      "grad_norm": 0.024398524314165115,
      "learning_rate": 6.639741616423736e-06,
      "loss": 0.0019,
      "step": 83290
    },
    {
      "epoch": 1.3630041724617525,
      "grad_norm": 0.11157342046499252,
      "learning_rate": 6.638842293044572e-06,
      "loss": 0.0021,
      "step": 83300
    },
    {
      "epoch": 1.3631677984128283,
      "grad_norm": 0.07478858530521393,
      "learning_rate": 6.6379429102613135e-06,
      "loss": 0.0024,
      "step": 83310
    },
    {
      "epoch": 1.363331424363904,
      "grad_norm": 0.2838660180568695,
      "learning_rate": 6.637043468106562e-06,
      "loss": 0.0022,
      "step": 83320
    },
    {
      "epoch": 1.36349505031498,
      "grad_norm": 0.15137562155723572,
      "learning_rate": 6.636143966612919e-06,
      "loss": 0.0035,
      "step": 83330
    },
    {
      "epoch": 1.3636586762660559,
      "grad_norm": 0.15146780014038086,
      "learning_rate": 6.6352444058129895e-06,
      "loss": 0.0027,
      "step": 83340
    },
    {
      "epoch": 1.3638223022171316,
      "grad_norm": 0.2696821987628937,
      "learning_rate": 6.6343447857393804e-06,
      "loss": 0.0038,
      "step": 83350
    },
    {
      "epoch": 1.3639859281682074,
      "grad_norm": 0.02089490182697773,
      "learning_rate": 6.6334451064247004e-06,
      "loss": 0.0014,
      "step": 83360
    },
    {
      "epoch": 1.3641495541192832,
      "grad_norm": 0.2376302182674408,
      "learning_rate": 6.632545367901563e-06,
      "loss": 0.0017,
      "step": 83370
    },
    {
      "epoch": 1.3643131800703592,
      "grad_norm": 0.0406130887567997,
      "learning_rate": 6.631645570202578e-06,
      "loss": 0.0017,
      "step": 83380
    },
    {
      "epoch": 1.364476806021435,
      "grad_norm": 0.24158768355846405,
      "learning_rate": 6.630745713360363e-06,
      "loss": 0.0015,
      "step": 83390
    },
    {
      "epoch": 1.3646404319725107,
      "grad_norm": 0.04271834343671799,
      "learning_rate": 6.629845797407536e-06,
      "loss": 0.0017,
      "step": 83400
    },
    {
      "epoch": 1.3648040579235867,
      "grad_norm": 0.07265377789735794,
      "learning_rate": 6.628945822376716e-06,
      "loss": 0.0021,
      "step": 83410
    },
    {
      "epoch": 1.3649676838746625,
      "grad_norm": 0.02887498401105404,
      "learning_rate": 6.628045788300524e-06,
      "loss": 0.0014,
      "step": 83420
    },
    {
      "epoch": 1.3651313098257383,
      "grad_norm": 0.1427624374628067,
      "learning_rate": 6.627145695211586e-06,
      "loss": 0.0021,
      "step": 83430
    },
    {
      "epoch": 1.3652949357768143,
      "grad_norm": 0.10718031972646713,
      "learning_rate": 6.626245543142528e-06,
      "loss": 0.0015,
      "step": 83440
    },
    {
      "epoch": 1.36545856172789,
      "grad_norm": 0.008648457936942577,
      "learning_rate": 6.625345332125976e-06,
      "loss": 0.0016,
      "step": 83450
    },
    {
      "epoch": 1.3656221876789658,
      "grad_norm": 0.05365251004695892,
      "learning_rate": 6.6244450621945624e-06,
      "loss": 0.0024,
      "step": 83460
    },
    {
      "epoch": 1.3657858136300418,
      "grad_norm": 0.2586226165294647,
      "learning_rate": 6.62354473338092e-06,
      "loss": 0.0014,
      "step": 83470
    },
    {
      "epoch": 1.3659494395811176,
      "grad_norm": 0.06785815209150314,
      "learning_rate": 6.6226443457176836e-06,
      "loss": 0.0019,
      "step": 83480
    },
    {
      "epoch": 1.3661130655321934,
      "grad_norm": 0.13367609679698944,
      "learning_rate": 6.621743899237489e-06,
      "loss": 0.0016,
      "step": 83490
    },
    {
      "epoch": 1.3662766914832694,
      "grad_norm": 0.0460641123354435,
      "learning_rate": 6.620843393972975e-06,
      "loss": 0.0019,
      "step": 83500
    },
    {
      "epoch": 1.3664403174343451,
      "grad_norm": 0.05655849352478981,
      "learning_rate": 6.619942829956785e-06,
      "loss": 0.0019,
      "step": 83510
    },
    {
      "epoch": 1.366603943385421,
      "grad_norm": 0.045292120426893234,
      "learning_rate": 6.61904220722156e-06,
      "loss": 0.0023,
      "step": 83520
    },
    {
      "epoch": 1.366767569336497,
      "grad_norm": 0.3453798294067383,
      "learning_rate": 6.618141525799946e-06,
      "loss": 0.0023,
      "step": 83530
    },
    {
      "epoch": 1.3669311952875727,
      "grad_norm": 0.06676346063613892,
      "learning_rate": 6.617240785724593e-06,
      "loss": 0.0027,
      "step": 83540
    },
    {
      "epoch": 1.3670948212386484,
      "grad_norm": 0.13619641959667206,
      "learning_rate": 6.616339987028145e-06,
      "loss": 0.0023,
      "step": 83550
    },
    {
      "epoch": 1.3672584471897242,
      "grad_norm": 0.14426226913928986,
      "learning_rate": 6.61543912974326e-06,
      "loss": 0.0033,
      "step": 83560
    },
    {
      "epoch": 1.3674220731408,
      "grad_norm": 0.12246865779161453,
      "learning_rate": 6.614538213902587e-06,
      "loss": 0.002,
      "step": 83570
    },
    {
      "epoch": 1.367585699091876,
      "grad_norm": 0.19099079072475433,
      "learning_rate": 6.613637239538786e-06,
      "loss": 0.0023,
      "step": 83580
    },
    {
      "epoch": 1.3677493250429518,
      "grad_norm": 0.020889585837721825,
      "learning_rate": 6.612736206684513e-06,
      "loss": 0.001,
      "step": 83590
    },
    {
      "epoch": 1.3679129509940275,
      "grad_norm": 0.25056174397468567,
      "learning_rate": 6.611835115372427e-06,
      "loss": 0.0029,
      "step": 83600
    },
    {
      "epoch": 1.3680765769451035,
      "grad_norm": 0.09843003004789352,
      "learning_rate": 6.610933965635193e-06,
      "loss": 0.0016,
      "step": 83610
    },
    {
      "epoch": 1.3682402028961793,
      "grad_norm": 0.09536533057689667,
      "learning_rate": 6.610032757505476e-06,
      "loss": 0.0012,
      "step": 83620
    },
    {
      "epoch": 1.368403828847255,
      "grad_norm": 0.08397623896598816,
      "learning_rate": 6.6091314910159386e-06,
      "loss": 0.0018,
      "step": 83630
    },
    {
      "epoch": 1.368567454798331,
      "grad_norm": 0.15769927203655243,
      "learning_rate": 6.608230166199255e-06,
      "loss": 0.0033,
      "step": 83640
    },
    {
      "epoch": 1.3687310807494069,
      "grad_norm": 0.05273592099547386,
      "learning_rate": 6.607328783088092e-06,
      "loss": 0.0017,
      "step": 83650
    },
    {
      "epoch": 1.3688947067004826,
      "grad_norm": 0.06668082624673843,
      "learning_rate": 6.606427341715123e-06,
      "loss": 0.0013,
      "step": 83660
    },
    {
      "epoch": 1.3690583326515586,
      "grad_norm": 0.23718303442001343,
      "learning_rate": 6.605525842113024e-06,
      "loss": 0.0024,
      "step": 83670
    },
    {
      "epoch": 1.3692219586026344,
      "grad_norm": 0.04254599288105965,
      "learning_rate": 6.604624284314472e-06,
      "loss": 0.0024,
      "step": 83680
    },
    {
      "epoch": 1.3693855845537102,
      "grad_norm": 0.03069309890270233,
      "learning_rate": 6.603722668352146e-06,
      "loss": 0.0025,
      "step": 83690
    },
    {
      "epoch": 1.3695492105047862,
      "grad_norm": 0.11927279084920883,
      "learning_rate": 6.602820994258728e-06,
      "loss": 0.0019,
      "step": 83700
    },
    {
      "epoch": 1.369712836455862,
      "grad_norm": 0.1013031005859375,
      "learning_rate": 6.601919262066903e-06,
      "loss": 0.0013,
      "step": 83710
    },
    {
      "epoch": 1.3698764624069377,
      "grad_norm": 0.2055090218782425,
      "learning_rate": 6.601017471809352e-06,
      "loss": 0.002,
      "step": 83720
    },
    {
      "epoch": 1.3700400883580137,
      "grad_norm": 0.05567789822816849,
      "learning_rate": 6.600115623518767e-06,
      "loss": 0.0027,
      "step": 83730
    },
    {
      "epoch": 1.3702037143090895,
      "grad_norm": 0.06563231348991394,
      "learning_rate": 6.599213717227836e-06,
      "loss": 0.0017,
      "step": 83740
    },
    {
      "epoch": 1.3703673402601653,
      "grad_norm": 0.10842576622962952,
      "learning_rate": 6.598311752969252e-06,
      "loss": 0.0026,
      "step": 83750
    },
    {
      "epoch": 1.370530966211241,
      "grad_norm": 0.1533299833536148,
      "learning_rate": 6.597409730775708e-06,
      "loss": 0.0016,
      "step": 83760
    },
    {
      "epoch": 1.3706945921623168,
      "grad_norm": 0.16760112345218658,
      "learning_rate": 6.5965076506799e-06,
      "loss": 0.003,
      "step": 83770
    },
    {
      "epoch": 1.3708582181133928,
      "grad_norm": 0.12350402027368546,
      "learning_rate": 6.595605512714527e-06,
      "loss": 0.0032,
      "step": 83780
    },
    {
      "epoch": 1.3710218440644686,
      "grad_norm": 0.035299547016620636,
      "learning_rate": 6.594703316912291e-06,
      "loss": 0.0032,
      "step": 83790
    },
    {
      "epoch": 1.3711854700155444,
      "grad_norm": 0.1186697781085968,
      "learning_rate": 6.593801063305891e-06,
      "loss": 0.0023,
      "step": 83800
    },
    {
      "epoch": 1.3713490959666204,
      "grad_norm": 0.3366985619068146,
      "learning_rate": 6.592898751928033e-06,
      "loss": 0.0054,
      "step": 83810
    },
    {
      "epoch": 1.3715127219176961,
      "grad_norm": 0.24717161059379578,
      "learning_rate": 6.591996382811423e-06,
      "loss": 0.0019,
      "step": 83820
    },
    {
      "epoch": 1.371676347868772,
      "grad_norm": 0.055966999381780624,
      "learning_rate": 6.59109395598877e-06,
      "loss": 0.0016,
      "step": 83830
    },
    {
      "epoch": 1.371839973819848,
      "grad_norm": 0.08399087935686111,
      "learning_rate": 6.590191471492789e-06,
      "loss": 0.002,
      "step": 83840
    },
    {
      "epoch": 1.3720035997709237,
      "grad_norm": 0.11979738622903824,
      "learning_rate": 6.589288929356185e-06,
      "loss": 0.0013,
      "step": 83850
    },
    {
      "epoch": 1.3721672257219995,
      "grad_norm": 0.10720144957304001,
      "learning_rate": 6.588386329611678e-06,
      "loss": 0.002,
      "step": 83860
    },
    {
      "epoch": 1.3723308516730754,
      "grad_norm": 0.06949020177125931,
      "learning_rate": 6.587483672291986e-06,
      "loss": 0.0012,
      "step": 83870
    },
    {
      "epoch": 1.3724944776241512,
      "grad_norm": 0.19956162571907043,
      "learning_rate": 6.5865809574298235e-06,
      "loss": 0.0027,
      "step": 83880
    },
    {
      "epoch": 1.372658103575227,
      "grad_norm": 0.1763794720172882,
      "learning_rate": 6.585678185057915e-06,
      "loss": 0.0028,
      "step": 83890
    },
    {
      "epoch": 1.372821729526303,
      "grad_norm": 0.07271043956279755,
      "learning_rate": 6.584775355208986e-06,
      "loss": 0.0016,
      "step": 83900
    },
    {
      "epoch": 1.3729853554773788,
      "grad_norm": 0.0488058403134346,
      "learning_rate": 6.583872467915757e-06,
      "loss": 0.0016,
      "step": 83910
    },
    {
      "epoch": 1.3731489814284545,
      "grad_norm": 0.3219110667705536,
      "learning_rate": 6.582969523210958e-06,
      "loss": 0.0036,
      "step": 83920
    },
    {
      "epoch": 1.3733126073795303,
      "grad_norm": 0.09001462906599045,
      "learning_rate": 6.582066521127318e-06,
      "loss": 0.0029,
      "step": 83930
    },
    {
      "epoch": 1.3734762333306063,
      "grad_norm": 0.06409922242164612,
      "learning_rate": 6.581163461697571e-06,
      "loss": 0.0045,
      "step": 83940
    },
    {
      "epoch": 1.373639859281682,
      "grad_norm": 0.1760329157114029,
      "learning_rate": 6.580260344954447e-06,
      "loss": 0.0016,
      "step": 83950
    },
    {
      "epoch": 1.3738034852327579,
      "grad_norm": 0.07624134421348572,
      "learning_rate": 6.579357170930684e-06,
      "loss": 0.0029,
      "step": 83960
    },
    {
      "epoch": 1.3739671111838336,
      "grad_norm": 0.07200030237436295,
      "learning_rate": 6.578453939659018e-06,
      "loss": 0.0016,
      "step": 83970
    },
    {
      "epoch": 1.3741307371349096,
      "grad_norm": 0.0995061844587326,
      "learning_rate": 6.5775506511721935e-06,
      "loss": 0.0016,
      "step": 83980
    },
    {
      "epoch": 1.3742943630859854,
      "grad_norm": 0.33647817373275757,
      "learning_rate": 6.576647305502948e-06,
      "loss": 0.0025,
      "step": 83990
    },
    {
      "epoch": 1.3744579890370612,
      "grad_norm": 0.263393372297287,
      "learning_rate": 6.575743902684026e-06,
      "loss": 0.0027,
      "step": 84000
    },
    {
      "epoch": 1.3746216149881372,
      "grad_norm": 0.044238802045583725,
      "learning_rate": 6.574840442748176e-06,
      "loss": 0.0023,
      "step": 84010
    },
    {
      "epoch": 1.374785240939213,
      "grad_norm": 0.13966703414916992,
      "learning_rate": 6.573936925728144e-06,
      "loss": 0.0011,
      "step": 84020
    },
    {
      "epoch": 1.3749488668902887,
      "grad_norm": 0.045412782579660416,
      "learning_rate": 6.573033351656683e-06,
      "loss": 0.0015,
      "step": 84030
    },
    {
      "epoch": 1.3751124928413647,
      "grad_norm": 0.10076606273651123,
      "learning_rate": 6.572129720566541e-06,
      "loss": 0.005,
      "step": 84040
    },
    {
      "epoch": 1.3752761187924405,
      "grad_norm": 0.08290413022041321,
      "learning_rate": 6.571226032490477e-06,
      "loss": 0.0026,
      "step": 84050
    },
    {
      "epoch": 1.3754397447435163,
      "grad_norm": 0.2463889867067337,
      "learning_rate": 6.570322287461246e-06,
      "loss": 0.0023,
      "step": 84060
    },
    {
      "epoch": 1.3756033706945923,
      "grad_norm": 0.14944596588611603,
      "learning_rate": 6.569418485511605e-06,
      "loss": 0.0014,
      "step": 84070
    },
    {
      "epoch": 1.375766996645668,
      "grad_norm": 0.1984856128692627,
      "learning_rate": 6.568514626674316e-06,
      "loss": 0.0024,
      "step": 84080
    },
    {
      "epoch": 1.3759306225967438,
      "grad_norm": 0.2240912914276123,
      "learning_rate": 6.567610710982143e-06,
      "loss": 0.002,
      "step": 84090
    },
    {
      "epoch": 1.3760942485478198,
      "grad_norm": 0.3564155399799347,
      "learning_rate": 6.5667067384678475e-06,
      "loss": 0.0033,
      "step": 84100
    },
    {
      "epoch": 1.3762578744988956,
      "grad_norm": 0.12061890959739685,
      "learning_rate": 6.565802709164199e-06,
      "loss": 0.002,
      "step": 84110
    },
    {
      "epoch": 1.3764215004499714,
      "grad_norm": 0.3658267557621002,
      "learning_rate": 6.564898623103966e-06,
      "loss": 0.0015,
      "step": 84120
    },
    {
      "epoch": 1.3765851264010471,
      "grad_norm": 0.4962645173072815,
      "learning_rate": 6.563994480319919e-06,
      "loss": 0.0034,
      "step": 84130
    },
    {
      "epoch": 1.3767487523521231,
      "grad_norm": 0.24085886776447296,
      "learning_rate": 6.563090280844832e-06,
      "loss": 0.0023,
      "step": 84140
    },
    {
      "epoch": 1.376912378303199,
      "grad_norm": 0.10248955339193344,
      "learning_rate": 6.562186024711478e-06,
      "loss": 0.0027,
      "step": 84150
    },
    {
      "epoch": 1.3770760042542747,
      "grad_norm": 0.2887658476829529,
      "learning_rate": 6.561281711952634e-06,
      "loss": 0.0052,
      "step": 84160
    },
    {
      "epoch": 1.3772396302053505,
      "grad_norm": 0.16747045516967773,
      "learning_rate": 6.560377342601082e-06,
      "loss": 0.0023,
      "step": 84170
    },
    {
      "epoch": 1.3774032561564264,
      "grad_norm": 0.09652359038591385,
      "learning_rate": 6.5594729166896e-06,
      "loss": 0.0019,
      "step": 84180
    },
    {
      "epoch": 1.3775668821075022,
      "grad_norm": 0.07700984179973602,
      "learning_rate": 6.558568434250975e-06,
      "loss": 0.0011,
      "step": 84190
    },
    {
      "epoch": 1.377730508058578,
      "grad_norm": 0.05332626402378082,
      "learning_rate": 6.55766389531799e-06,
      "loss": 0.0022,
      "step": 84200
    },
    {
      "epoch": 1.377894134009654,
      "grad_norm": 0.021167131140828133,
      "learning_rate": 6.5567592999234306e-06,
      "loss": 0.0021,
      "step": 84210
    },
    {
      "epoch": 1.3780577599607298,
      "grad_norm": 0.1650628000497818,
      "learning_rate": 6.55585464810009e-06,
      "loss": 0.0035,
      "step": 84220
    },
    {
      "epoch": 1.3782213859118055,
      "grad_norm": 0.11689629405736923,
      "learning_rate": 6.554949939880757e-06,
      "loss": 0.0012,
      "step": 84230
    },
    {
      "epoch": 1.3783850118628815,
      "grad_norm": 0.10608311742544174,
      "learning_rate": 6.554045175298226e-06,
      "loss": 0.0014,
      "step": 84240
    },
    {
      "epoch": 1.3785486378139573,
      "grad_norm": 0.04725388064980507,
      "learning_rate": 6.553140354385292e-06,
      "loss": 0.0012,
      "step": 84250
    },
    {
      "epoch": 1.378712263765033,
      "grad_norm": 0.2436954379081726,
      "learning_rate": 6.5522354771747535e-06,
      "loss": 0.0023,
      "step": 84260
    },
    {
      "epoch": 1.378875889716109,
      "grad_norm": 0.2977786958217621,
      "learning_rate": 6.551330543699409e-06,
      "loss": 0.0014,
      "step": 84270
    },
    {
      "epoch": 1.3790395156671849,
      "grad_norm": 0.05094189569354057,
      "learning_rate": 6.550425553992062e-06,
      "loss": 0.002,
      "step": 84280
    },
    {
      "epoch": 1.3792031416182606,
      "grad_norm": 0.09922634065151215,
      "learning_rate": 6.549520508085515e-06,
      "loss": 0.0016,
      "step": 84290
    },
    {
      "epoch": 1.3793667675693366,
      "grad_norm": 0.04487872123718262,
      "learning_rate": 6.548615406012573e-06,
      "loss": 0.0019,
      "step": 84300
    },
    {
      "epoch": 1.3795303935204124,
      "grad_norm": 0.06067148968577385,
      "learning_rate": 6.547710247806045e-06,
      "loss": 0.0018,
      "step": 84310
    },
    {
      "epoch": 1.3796940194714882,
      "grad_norm": 0.12283045053482056,
      "learning_rate": 6.546805033498739e-06,
      "loss": 0.0016,
      "step": 84320
    },
    {
      "epoch": 1.379857645422564,
      "grad_norm": 0.09044544398784637,
      "learning_rate": 6.545899763123469e-06,
      "loss": 0.0018,
      "step": 84330
    },
    {
      "epoch": 1.3800212713736397,
      "grad_norm": 0.09511348605155945,
      "learning_rate": 6.54499443671305e-06,
      "loss": 0.0028,
      "step": 84340
    },
    {
      "epoch": 1.3801848973247157,
      "grad_norm": 0.23826463520526886,
      "learning_rate": 6.5440890543002936e-06,
      "loss": 0.0033,
      "step": 84350
    },
    {
      "epoch": 1.3803485232757915,
      "grad_norm": 0.008653919212520123,
      "learning_rate": 6.543183615918021e-06,
      "loss": 0.0015,
      "step": 84360
    },
    {
      "epoch": 1.3805121492268673,
      "grad_norm": 0.021614257246255875,
      "learning_rate": 6.54227812159905e-06,
      "loss": 0.0014,
      "step": 84370
    },
    {
      "epoch": 1.3806757751779433,
      "grad_norm": 0.16154079139232635,
      "learning_rate": 6.541372571376206e-06,
      "loss": 0.0028,
      "step": 84380
    },
    {
      "epoch": 1.380839401129019,
      "grad_norm": 0.13914792239665985,
      "learning_rate": 6.540466965282309e-06,
      "loss": 0.0017,
      "step": 84390
    },
    {
      "epoch": 1.3810030270800948,
      "grad_norm": 0.06671799719333649,
      "learning_rate": 6.539561303350188e-06,
      "loss": 0.0019,
      "step": 84400
    },
    {
      "epoch": 1.3811666530311708,
      "grad_norm": 0.08875478804111481,
      "learning_rate": 6.53865558561267e-06,
      "loss": 0.0019,
      "step": 84410
    },
    {
      "epoch": 1.3813302789822466,
      "grad_norm": 0.1481930911540985,
      "learning_rate": 6.537749812102585e-06,
      "loss": 0.0031,
      "step": 84420
    },
    {
      "epoch": 1.3814939049333224,
      "grad_norm": 0.1867065280675888,
      "learning_rate": 6.536843982852765e-06,
      "loss": 0.0031,
      "step": 84430
    },
    {
      "epoch": 1.3816575308843984,
      "grad_norm": 0.0756026953458786,
      "learning_rate": 6.535938097896045e-06,
      "loss": 0.0024,
      "step": 84440
    },
    {
      "epoch": 1.3818211568354741,
      "grad_norm": 0.06768887490034103,
      "learning_rate": 6.535032157265258e-06,
      "loss": 0.0023,
      "step": 84450
    },
    {
      "epoch": 1.38198478278655,
      "grad_norm": 0.16543477773666382,
      "learning_rate": 6.534126160993247e-06,
      "loss": 0.0023,
      "step": 84460
    },
    {
      "epoch": 1.382148408737626,
      "grad_norm": 0.12259591370820999,
      "learning_rate": 6.533220109112849e-06,
      "loss": 0.0031,
      "step": 84470
    },
    {
      "epoch": 1.3823120346887017,
      "grad_norm": 0.06564922630786896,
      "learning_rate": 6.5323140016569075e-06,
      "loss": 0.0013,
      "step": 84480
    },
    {
      "epoch": 1.3824756606397774,
      "grad_norm": 0.04190611466765404,
      "learning_rate": 6.531407838658267e-06,
      "loss": 0.0017,
      "step": 84490
    },
    {
      "epoch": 1.3826392865908534,
      "grad_norm": 0.13495048880577087,
      "learning_rate": 6.5305016201497715e-06,
      "loss": 0.0017,
      "step": 84500
    },
    {
      "epoch": 1.3828029125419292,
      "grad_norm": 0.1354617476463318,
      "learning_rate": 6.529595346164273e-06,
      "loss": 0.002,
      "step": 84510
    },
    {
      "epoch": 1.382966538493005,
      "grad_norm": 0.14890646934509277,
      "learning_rate": 6.528689016734617e-06,
      "loss": 0.002,
      "step": 84520
    },
    {
      "epoch": 1.3831301644440808,
      "grad_norm": 0.19110403954982758,
      "learning_rate": 6.527782631893658e-06,
      "loss": 0.002,
      "step": 84530
    },
    {
      "epoch": 1.3832937903951565,
      "grad_norm": 0.07025747001171112,
      "learning_rate": 6.526876191674252e-06,
      "loss": 0.0021,
      "step": 84540
    },
    {
      "epoch": 1.3834574163462325,
      "grad_norm": 0.07577703893184662,
      "learning_rate": 6.525969696109254e-06,
      "loss": 0.0027,
      "step": 84550
    },
    {
      "epoch": 1.3836210422973083,
      "grad_norm": 0.13240648806095123,
      "learning_rate": 6.52506314523152e-06,
      "loss": 0.002,
      "step": 84560
    },
    {
      "epoch": 1.383784668248384,
      "grad_norm": 0.2762661576271057,
      "learning_rate": 6.524156539073915e-06,
      "loss": 0.0021,
      "step": 84570
    },
    {
      "epoch": 1.38394829419946,
      "grad_norm": 0.04597948491573334,
      "learning_rate": 6.5232498776692975e-06,
      "loss": 0.0016,
      "step": 84580
    },
    {
      "epoch": 1.3841119201505359,
      "grad_norm": 0.02019342966377735,
      "learning_rate": 6.522343161050533e-06,
      "loss": 0.0013,
      "step": 84590
    },
    {
      "epoch": 1.3842755461016116,
      "grad_norm": 0.18653501570224762,
      "learning_rate": 6.521436389250488e-06,
      "loss": 0.0014,
      "step": 84600
    },
    {
      "epoch": 1.3844391720526876,
      "grad_norm": 0.09549687802791595,
      "learning_rate": 6.520529562302029e-06,
      "loss": 0.0017,
      "step": 84610
    },
    {
      "epoch": 1.3846027980037634,
      "grad_norm": 0.005977577064186335,
      "learning_rate": 6.519622680238029e-06,
      "loss": 0.002,
      "step": 84620
    },
    {
      "epoch": 1.3847664239548392,
      "grad_norm": 0.02042592316865921,
      "learning_rate": 6.518715743091359e-06,
      "loss": 0.002,
      "step": 84630
    },
    {
      "epoch": 1.3849300499059152,
      "grad_norm": 0.16510340571403503,
      "learning_rate": 6.5178087508948965e-06,
      "loss": 0.002,
      "step": 84640
    },
    {
      "epoch": 1.385093675856991,
      "grad_norm": 0.09785642474889755,
      "learning_rate": 6.516901703681511e-06,
      "loss": 0.003,
      "step": 84650
    },
    {
      "epoch": 1.3852573018080667,
      "grad_norm": 0.035534076392650604,
      "learning_rate": 6.5159946014840854e-06,
      "loss": 0.0018,
      "step": 84660
    },
    {
      "epoch": 1.3854209277591427,
      "grad_norm": 0.0675075426697731,
      "learning_rate": 6.515087444335498e-06,
      "loss": 0.0013,
      "step": 84670
    },
    {
      "epoch": 1.3855845537102185,
      "grad_norm": 0.22779537737369537,
      "learning_rate": 6.5141802322686345e-06,
      "loss": 0.0026,
      "step": 84680
    },
    {
      "epoch": 1.3857481796612943,
      "grad_norm": 0.04339180886745453,
      "learning_rate": 6.513272965316376e-06,
      "loss": 0.0013,
      "step": 84690
    },
    {
      "epoch": 1.38591180561237,
      "grad_norm": 0.055955853313207626,
      "learning_rate": 6.512365643511609e-06,
      "loss": 0.006,
      "step": 84700
    },
    {
      "epoch": 1.386075431563446,
      "grad_norm": 0.12068263441324234,
      "learning_rate": 6.511458266887223e-06,
      "loss": 0.0029,
      "step": 84710
    },
    {
      "epoch": 1.3862390575145218,
      "grad_norm": 0.1273995041847229,
      "learning_rate": 6.5105508354761075e-06,
      "loss": 0.0024,
      "step": 84720
    },
    {
      "epoch": 1.3864026834655976,
      "grad_norm": 0.3073548674583435,
      "learning_rate": 6.509643349311153e-06,
      "loss": 0.0019,
      "step": 84730
    },
    {
      "epoch": 1.3865663094166734,
      "grad_norm": 0.0018799295648932457,
      "learning_rate": 6.5087358084252575e-06,
      "loss": 0.0014,
      "step": 84740
    },
    {
      "epoch": 1.3867299353677494,
      "grad_norm": 0.5400633811950684,
      "learning_rate": 6.507828212851315e-06,
      "loss": 0.0031,
      "step": 84750
    },
    {
      "epoch": 1.3868935613188251,
      "grad_norm": 0.0834333673119545,
      "learning_rate": 6.506920562622222e-06,
      "loss": 0.0017,
      "step": 84760
    },
    {
      "epoch": 1.387057187269901,
      "grad_norm": 0.06015801057219505,
      "learning_rate": 6.506012857770882e-06,
      "loss": 0.0016,
      "step": 84770
    },
    {
      "epoch": 1.387220813220977,
      "grad_norm": 0.25534600019454956,
      "learning_rate": 6.505105098330194e-06,
      "loss": 0.0025,
      "step": 84780
    },
    {
      "epoch": 1.3873844391720527,
      "grad_norm": 0.10330186784267426,
      "learning_rate": 6.5041972843330654e-06,
      "loss": 0.0014,
      "step": 84790
    },
    {
      "epoch": 1.3875480651231284,
      "grad_norm": 0.37720584869384766,
      "learning_rate": 6.503289415812401e-06,
      "loss": 0.002,
      "step": 84800
    },
    {
      "epoch": 1.3877116910742044,
      "grad_norm": 0.16098958253860474,
      "learning_rate": 6.502381492801105e-06,
      "loss": 0.0013,
      "step": 84810
    },
    {
      "epoch": 1.3878753170252802,
      "grad_norm": 0.0783856213092804,
      "learning_rate": 6.501473515332093e-06,
      "loss": 0.0019,
      "step": 84820
    },
    {
      "epoch": 1.388038942976356,
      "grad_norm": 0.013071945868432522,
      "learning_rate": 6.500565483438275e-06,
      "loss": 0.0014,
      "step": 84830
    },
    {
      "epoch": 1.388202568927432,
      "grad_norm": 0.00263294018805027,
      "learning_rate": 6.499657397152566e-06,
      "loss": 0.0028,
      "step": 84840
    },
    {
      "epoch": 1.3883661948785078,
      "grad_norm": 0.04995952174067497,
      "learning_rate": 6.498749256507878e-06,
      "loss": 0.0012,
      "step": 84850
    },
    {
      "epoch": 1.3885298208295835,
      "grad_norm": 0.010463164187967777,
      "learning_rate": 6.497841061537132e-06,
      "loss": 0.0013,
      "step": 84860
    },
    {
      "epoch": 1.3886934467806595,
      "grad_norm": 0.19178052246570587,
      "learning_rate": 6.496932812273249e-06,
      "loss": 0.0041,
      "step": 84870
    },
    {
      "epoch": 1.3888570727317353,
      "grad_norm": 0.1179526299238205,
      "learning_rate": 6.496024508749148e-06,
      "loss": 0.0017,
      "step": 84880
    },
    {
      "epoch": 1.389020698682811,
      "grad_norm": 0.358823299407959,
      "learning_rate": 6.495116150997754e-06,
      "loss": 0.0032,
      "step": 84890
    },
    {
      "epoch": 1.3891843246338869,
      "grad_norm": 0.09603816270828247,
      "learning_rate": 6.494207739051994e-06,
      "loss": 0.0016,
      "step": 84900
    },
    {
      "epoch": 1.3893479505849629,
      "grad_norm": 0.10268957912921906,
      "learning_rate": 6.493299272944794e-06,
      "loss": 0.0018,
      "step": 84910
    },
    {
      "epoch": 1.3895115765360386,
      "grad_norm": 0.02950035035610199,
      "learning_rate": 6.4923907527090845e-06,
      "loss": 0.0009,
      "step": 84920
    },
    {
      "epoch": 1.3896752024871144,
      "grad_norm": 0.10302207618951797,
      "learning_rate": 6.491482178377797e-06,
      "loss": 0.0017,
      "step": 84930
    },
    {
      "epoch": 1.3898388284381902,
      "grad_norm": 0.040904585272073746,
      "learning_rate": 6.490573549983866e-06,
      "loss": 0.0012,
      "step": 84940
    },
    {
      "epoch": 1.3900024543892662,
      "grad_norm": 0.04564441367983818,
      "learning_rate": 6.4896648675602255e-06,
      "loss": 0.0021,
      "step": 84950
    },
    {
      "epoch": 1.390166080340342,
      "grad_norm": 0.06184501200914383,
      "learning_rate": 6.488756131139814e-06,
      "loss": 0.0015,
      "step": 84960
    },
    {
      "epoch": 1.3903297062914177,
      "grad_norm": 0.056682560592889786,
      "learning_rate": 6.4878473407555705e-06,
      "loss": 0.0016,
      "step": 84970
    },
    {
      "epoch": 1.3904933322424937,
      "grad_norm": 0.17566584050655365,
      "learning_rate": 6.486938496440438e-06,
      "loss": 0.0019,
      "step": 84980
    },
    {
      "epoch": 1.3906569581935695,
      "grad_norm": 0.09449051320552826,
      "learning_rate": 6.486029598227359e-06,
      "loss": 0.0025,
      "step": 84990
    },
    {
      "epoch": 1.3908205841446453,
      "grad_norm": 0.028499802574515343,
      "learning_rate": 6.4851206461492765e-06,
      "loss": 0.0018,
      "step": 85000
    },
    {
      "epoch": 1.3909842100957213,
      "grad_norm": 0.046537838876247406,
      "learning_rate": 6.484211640239142e-06,
      "loss": 0.0033,
      "step": 85010
    },
    {
      "epoch": 1.391147836046797,
      "grad_norm": 0.016465824097394943,
      "learning_rate": 6.483302580529902e-06,
      "loss": 0.0014,
      "step": 85020
    },
    {
      "epoch": 1.3913114619978728,
      "grad_norm": 0.09119275957345963,
      "learning_rate": 6.4823934670545075e-06,
      "loss": 0.0014,
      "step": 85030
    },
    {
      "epoch": 1.3914750879489488,
      "grad_norm": 0.17053423821926117,
      "learning_rate": 6.481484299845914e-06,
      "loss": 0.0027,
      "step": 85040
    },
    {
      "epoch": 1.3916387139000246,
      "grad_norm": 0.12290740013122559,
      "learning_rate": 6.480575078937075e-06,
      "loss": 0.0016,
      "step": 85050
    },
    {
      "epoch": 1.3918023398511004,
      "grad_norm": 0.03296781703829765,
      "learning_rate": 6.479665804360947e-06,
      "loss": 0.0018,
      "step": 85060
    },
    {
      "epoch": 1.3919659658021764,
      "grad_norm": 0.08438185602426529,
      "learning_rate": 6.478756476150492e-06,
      "loss": 0.0013,
      "step": 85070
    },
    {
      "epoch": 1.3921295917532521,
      "grad_norm": 0.1113714650273323,
      "learning_rate": 6.477847094338666e-06,
      "loss": 0.0014,
      "step": 85080
    },
    {
      "epoch": 1.392293217704328,
      "grad_norm": 0.12829799950122833,
      "learning_rate": 6.476937658958436e-06,
      "loss": 0.0017,
      "step": 85090
    },
    {
      "epoch": 1.3924568436554037,
      "grad_norm": 0.06273569166660309,
      "learning_rate": 6.4760281700427665e-06,
      "loss": 0.0018,
      "step": 85100
    },
    {
      "epoch": 1.3926204696064794,
      "grad_norm": 0.014807458966970444,
      "learning_rate": 6.475118627624622e-06,
      "loss": 0.0021,
      "step": 85110
    },
    {
      "epoch": 1.3927840955575554,
      "grad_norm": 0.1321907341480255,
      "learning_rate": 6.474209031736973e-06,
      "loss": 0.0012,
      "step": 85120
    },
    {
      "epoch": 1.3929477215086312,
      "grad_norm": 0.15450501441955566,
      "learning_rate": 6.473299382412791e-06,
      "loss": 0.0033,
      "step": 85130
    },
    {
      "epoch": 1.393111347459707,
      "grad_norm": 0.09167923778295517,
      "learning_rate": 6.472389679685047e-06,
      "loss": 0.0015,
      "step": 85140
    },
    {
      "epoch": 1.393274973410783,
      "grad_norm": 0.1072862520813942,
      "learning_rate": 6.471479923586715e-06,
      "loss": 0.0019,
      "step": 85150
    },
    {
      "epoch": 1.3934385993618588,
      "grad_norm": 0.172815203666687,
      "learning_rate": 6.470570114150773e-06,
      "loss": 0.0018,
      "step": 85160
    },
    {
      "epoch": 1.3936022253129345,
      "grad_norm": 0.003767084563151002,
      "learning_rate": 6.469660251410197e-06,
      "loss": 0.0008,
      "step": 85170
    },
    {
      "epoch": 1.3937658512640105,
      "grad_norm": 0.18949241936206818,
      "learning_rate": 6.4687503353979705e-06,
      "loss": 0.002,
      "step": 85180
    },
    {
      "epoch": 1.3939294772150863,
      "grad_norm": 0.018428917974233627,
      "learning_rate": 6.467840366147076e-06,
      "loss": 0.0024,
      "step": 85190
    },
    {
      "epoch": 1.394093103166162,
      "grad_norm": 0.054442957043647766,
      "learning_rate": 6.466930343690494e-06,
      "loss": 0.0019,
      "step": 85200
    },
    {
      "epoch": 1.394256729117238,
      "grad_norm": 0.00616827979683876,
      "learning_rate": 6.466020268061214e-06,
      "loss": 0.0017,
      "step": 85210
    },
    {
      "epoch": 1.3944203550683139,
      "grad_norm": 0.10218977928161621,
      "learning_rate": 6.465110139292221e-06,
      "loss": 0.0029,
      "step": 85220
    },
    {
      "epoch": 1.3945839810193896,
      "grad_norm": 0.08989010751247406,
      "learning_rate": 6.464199957416509e-06,
      "loss": 0.002,
      "step": 85230
    },
    {
      "epoch": 1.3947476069704656,
      "grad_norm": 0.08905472606420517,
      "learning_rate": 6.463289722467065e-06,
      "loss": 0.0011,
      "step": 85240
    },
    {
      "epoch": 1.3949112329215414,
      "grad_norm": 0.047487299889326096,
      "learning_rate": 6.462379434476888e-06,
      "loss": 0.0012,
      "step": 85250
    },
    {
      "epoch": 1.3950748588726172,
      "grad_norm": 0.1420343816280365,
      "learning_rate": 6.4614690934789695e-06,
      "loss": 0.0011,
      "step": 85260
    },
    {
      "epoch": 1.3952384848236932,
      "grad_norm": 0.04268716648221016,
      "learning_rate": 6.4605586995063095e-06,
      "loss": 0.0013,
      "step": 85270
    },
    {
      "epoch": 1.395402110774769,
      "grad_norm": 0.021871600300073624,
      "learning_rate": 6.459648252591907e-06,
      "loss": 0.002,
      "step": 85280
    },
    {
      "epoch": 1.3955657367258447,
      "grad_norm": 0.02697080932557583,
      "learning_rate": 6.458737752768765e-06,
      "loss": 0.0021,
      "step": 85290
    },
    {
      "epoch": 1.3957293626769205,
      "grad_norm": 0.004370892886072397,
      "learning_rate": 6.457827200069884e-06,
      "loss": 0.0021,
      "step": 85300
    },
    {
      "epoch": 1.3958929886279963,
      "grad_norm": 0.1204993799328804,
      "learning_rate": 6.45691659452827e-06,
      "loss": 0.0014,
      "step": 85310
    },
    {
      "epoch": 1.3960566145790723,
      "grad_norm": 0.03712393715977669,
      "learning_rate": 6.456005936176931e-06,
      "loss": 0.0017,
      "step": 85320
    },
    {
      "epoch": 1.396220240530148,
      "grad_norm": 0.1656939834356308,
      "learning_rate": 6.455095225048877e-06,
      "loss": 0.0017,
      "step": 85330
    },
    {
      "epoch": 1.3963838664812238,
      "grad_norm": 0.04335532709956169,
      "learning_rate": 6.45418446117712e-06,
      "loss": 0.0013,
      "step": 85340
    },
    {
      "epoch": 1.3965474924322998,
      "grad_norm": 0.021557774394750595,
      "learning_rate": 6.45327364459467e-06,
      "loss": 0.0019,
      "step": 85350
    },
    {
      "epoch": 1.3967111183833756,
      "grad_norm": 0.0697641521692276,
      "learning_rate": 6.452362775334542e-06,
      "loss": 0.002,
      "step": 85360
    },
    {
      "epoch": 1.3968747443344514,
      "grad_norm": 0.11676472425460815,
      "learning_rate": 6.4514518534297554e-06,
      "loss": 0.0046,
      "step": 85370
    },
    {
      "epoch": 1.3970383702855274,
      "grad_norm": 0.10342272371053696,
      "learning_rate": 6.450540878913327e-06,
      "loss": 0.0019,
      "step": 85380
    },
    {
      "epoch": 1.3972019962366031,
      "grad_norm": 0.07428403943777084,
      "learning_rate": 6.449629851818279e-06,
      "loss": 0.0012,
      "step": 85390
    },
    {
      "epoch": 1.397365622187679,
      "grad_norm": 0.04550778120756149,
      "learning_rate": 6.448718772177631e-06,
      "loss": 0.0016,
      "step": 85400
    },
    {
      "epoch": 1.397529248138755,
      "grad_norm": 0.11181758344173431,
      "learning_rate": 6.447807640024412e-06,
      "loss": 0.0013,
      "step": 85410
    },
    {
      "epoch": 1.3976928740898307,
      "grad_norm": 0.05142239108681679,
      "learning_rate": 6.446896455391644e-06,
      "loss": 0.0015,
      "step": 85420
    },
    {
      "epoch": 1.3978565000409064,
      "grad_norm": 0.0831603929400444,
      "learning_rate": 6.445985218312357e-06,
      "loss": 0.001,
      "step": 85430
    },
    {
      "epoch": 1.3980201259919824,
      "grad_norm": 0.03458372503519058,
      "learning_rate": 6.445073928819583e-06,
      "loss": 0.001,
      "step": 85440
    },
    {
      "epoch": 1.3981837519430582,
      "grad_norm": 0.07006905972957611,
      "learning_rate": 6.44416258694635e-06,
      "loss": 0.0026,
      "step": 85450
    },
    {
      "epoch": 1.398347377894134,
      "grad_norm": 0.08403483778238297,
      "learning_rate": 6.443251192725694e-06,
      "loss": 0.0023,
      "step": 85460
    },
    {
      "epoch": 1.39851100384521,
      "grad_norm": 0.049667708575725555,
      "learning_rate": 6.442339746190652e-06,
      "loss": 0.0009,
      "step": 85470
    },
    {
      "epoch": 1.3986746297962858,
      "grad_norm": 0.02640850841999054,
      "learning_rate": 6.441428247374261e-06,
      "loss": 0.0017,
      "step": 85480
    },
    {
      "epoch": 1.3988382557473615,
      "grad_norm": 0.19479133188724518,
      "learning_rate": 6.440516696309561e-06,
      "loss": 0.0023,
      "step": 85490
    },
    {
      "epoch": 1.3990018816984373,
      "grad_norm": 0.3405801057815552,
      "learning_rate": 6.439605093029593e-06,
      "loss": 0.002,
      "step": 85500
    },
    {
      "epoch": 1.399165507649513,
      "grad_norm": 0.09277898073196411,
      "learning_rate": 6.438693437567398e-06,
      "loss": 0.0016,
      "step": 85510
    },
    {
      "epoch": 1.399329133600589,
      "grad_norm": 0.17509441077709198,
      "learning_rate": 6.437781729956026e-06,
      "loss": 0.0021,
      "step": 85520
    },
    {
      "epoch": 1.3994927595516649,
      "grad_norm": 0.2545210123062134,
      "learning_rate": 6.43686997022852e-06,
      "loss": 0.0018,
      "step": 85530
    },
    {
      "epoch": 1.3996563855027406,
      "grad_norm": 0.2908412218093872,
      "learning_rate": 6.435958158417933e-06,
      "loss": 0.0024,
      "step": 85540
    },
    {
      "epoch": 1.3998200114538166,
      "grad_norm": 0.11098383367061615,
      "learning_rate": 6.435046294557314e-06,
      "loss": 0.0015,
      "step": 85550
    },
    {
      "epoch": 1.3999836374048924,
      "grad_norm": 0.04553581029176712,
      "learning_rate": 6.434134378679714e-06,
      "loss": 0.001,
      "step": 85560
    },
    {
      "epoch": 1.4001472633559682,
      "grad_norm": 0.05689772963523865,
      "learning_rate": 6.433222410818191e-06,
      "loss": 0.0013,
      "step": 85570
    },
    {
      "epoch": 1.4003108893070442,
      "grad_norm": 0.34929540753364563,
      "learning_rate": 6.432310391005799e-06,
      "loss": 0.0018,
      "step": 85580
    },
    {
      "epoch": 1.40047451525812,
      "grad_norm": 0.14437046647071838,
      "learning_rate": 6.431398319275599e-06,
      "loss": 0.0029,
      "step": 85590
    },
    {
      "epoch": 1.4006381412091957,
      "grad_norm": 0.08320649713277817,
      "learning_rate": 6.43048619566065e-06,
      "loss": 0.0035,
      "step": 85600
    },
    {
      "epoch": 1.4008017671602717,
      "grad_norm": 0.14735233783721924,
      "learning_rate": 6.429574020194014e-06,
      "loss": 0.0011,
      "step": 85610
    },
    {
      "epoch": 1.4009653931113475,
      "grad_norm": 0.13174661993980408,
      "learning_rate": 6.4286617929087565e-06,
      "loss": 0.0024,
      "step": 85620
    },
    {
      "epoch": 1.4011290190624233,
      "grad_norm": 0.09346847236156464,
      "learning_rate": 6.427749513837942e-06,
      "loss": 0.0019,
      "step": 85630
    },
    {
      "epoch": 1.4012926450134993,
      "grad_norm": 0.04263807833194733,
      "learning_rate": 6.426837183014639e-06,
      "loss": 0.0014,
      "step": 85640
    },
    {
      "epoch": 1.401456270964575,
      "grad_norm": 0.13263411819934845,
      "learning_rate": 6.425924800471918e-06,
      "loss": 0.0021,
      "step": 85650
    },
    {
      "epoch": 1.4016198969156508,
      "grad_norm": 0.05086040124297142,
      "learning_rate": 6.42501236624285e-06,
      "loss": 0.0012,
      "step": 85660
    },
    {
      "epoch": 1.4017835228667266,
      "grad_norm": 0.23759202659130096,
      "learning_rate": 6.424099880360508e-06,
      "loss": 0.0019,
      "step": 85670
    },
    {
      "epoch": 1.4019471488178026,
      "grad_norm": 0.08066391944885254,
      "learning_rate": 6.42318734285797e-06,
      "loss": 0.0018,
      "step": 85680
    },
    {
      "epoch": 1.4021107747688784,
      "grad_norm": 0.2468746453523636,
      "learning_rate": 6.422274753768309e-06,
      "loss": 0.005,
      "step": 85690
    },
    {
      "epoch": 1.4022744007199541,
      "grad_norm": 0.09473896771669388,
      "learning_rate": 6.421362113124609e-06,
      "loss": 0.0019,
      "step": 85700
    },
    {
      "epoch": 1.40243802667103,
      "grad_norm": 0.1353735625743866,
      "learning_rate": 6.420449420959946e-06,
      "loss": 0.0041,
      "step": 85710
    },
    {
      "epoch": 1.402601652622106,
      "grad_norm": 0.11348866671323776,
      "learning_rate": 6.419536677307408e-06,
      "loss": 0.0024,
      "step": 85720
    },
    {
      "epoch": 1.4027652785731817,
      "grad_norm": 0.02876884676516056,
      "learning_rate": 6.418623882200075e-06,
      "loss": 0.0022,
      "step": 85730
    },
    {
      "epoch": 1.4029289045242574,
      "grad_norm": 0.02298186533153057,
      "learning_rate": 6.417711035671035e-06,
      "loss": 0.0015,
      "step": 85740
    },
    {
      "epoch": 1.4030925304753334,
      "grad_norm": 0.07512074708938599,
      "learning_rate": 6.4167981377533795e-06,
      "loss": 0.0013,
      "step": 85750
    },
    {
      "epoch": 1.4032561564264092,
      "grad_norm": 0.12532132863998413,
      "learning_rate": 6.4158851884801964e-06,
      "loss": 0.0022,
      "step": 85760
    },
    {
      "epoch": 1.403419782377485,
      "grad_norm": 0.19724524021148682,
      "learning_rate": 6.414972187884578e-06,
      "loss": 0.0015,
      "step": 85770
    },
    {
      "epoch": 1.403583408328561,
      "grad_norm": 0.15418198704719543,
      "learning_rate": 6.414059135999618e-06,
      "loss": 0.0026,
      "step": 85780
    },
    {
      "epoch": 1.4037470342796368,
      "grad_norm": 0.1800295114517212,
      "learning_rate": 6.413146032858414e-06,
      "loss": 0.0022,
      "step": 85790
    },
    {
      "epoch": 1.4039106602307125,
      "grad_norm": 0.14737734198570251,
      "learning_rate": 6.412232878494063e-06,
      "loss": 0.0015,
      "step": 85800
    },
    {
      "epoch": 1.4040742861817885,
      "grad_norm": 0.02988053299486637,
      "learning_rate": 6.411319672939662e-06,
      "loss": 0.0037,
      "step": 85810
    },
    {
      "epoch": 1.4042379121328643,
      "grad_norm": 0.2550038993358612,
      "learning_rate": 6.410406416228316e-06,
      "loss": 0.0023,
      "step": 85820
    },
    {
      "epoch": 1.40440153808394,
      "grad_norm": 0.057182129472494125,
      "learning_rate": 6.409493108393128e-06,
      "loss": 0.0013,
      "step": 85830
    },
    {
      "epoch": 1.404565164035016,
      "grad_norm": 0.17738579213619232,
      "learning_rate": 6.408579749467202e-06,
      "loss": 0.0021,
      "step": 85840
    },
    {
      "epoch": 1.4047287899860919,
      "grad_norm": 0.23171603679656982,
      "learning_rate": 6.407666339483647e-06,
      "loss": 0.0026,
      "step": 85850
    },
    {
      "epoch": 1.4048924159371676,
      "grad_norm": 0.03303425386548042,
      "learning_rate": 6.406752878475568e-06,
      "loss": 0.0017,
      "step": 85860
    },
    {
      "epoch": 1.4050560418882434,
      "grad_norm": 0.2983608543872833,
      "learning_rate": 6.4058393664760795e-06,
      "loss": 0.0028,
      "step": 85870
    },
    {
      "epoch": 1.4052196678393194,
      "grad_norm": 0.2720623016357422,
      "learning_rate": 6.404925803518292e-06,
      "loss": 0.0021,
      "step": 85880
    },
    {
      "epoch": 1.4053832937903952,
      "grad_norm": 0.021536650136113167,
      "learning_rate": 6.404012189635321e-06,
      "loss": 0.0011,
      "step": 85890
    },
    {
      "epoch": 1.405546919741471,
      "grad_norm": 0.07657308131456375,
      "learning_rate": 6.403098524860284e-06,
      "loss": 0.0023,
      "step": 85900
    },
    {
      "epoch": 1.4057105456925467,
      "grad_norm": 0.21242614090442657,
      "learning_rate": 6.4021848092262964e-06,
      "loss": 0.0017,
      "step": 85910
    },
    {
      "epoch": 1.4058741716436227,
      "grad_norm": 0.07514762878417969,
      "learning_rate": 6.401271042766479e-06,
      "loss": 0.0017,
      "step": 85920
    },
    {
      "epoch": 1.4060377975946985,
      "grad_norm": 0.2019110769033432,
      "learning_rate": 6.400357225513956e-06,
      "loss": 0.0017,
      "step": 85930
    },
    {
      "epoch": 1.4062014235457743,
      "grad_norm": 0.017901379615068436,
      "learning_rate": 6.399443357501848e-06,
      "loss": 0.0017,
      "step": 85940
    },
    {
      "epoch": 1.4063650494968503,
      "grad_norm": 0.1320720762014389,
      "learning_rate": 6.398529438763281e-06,
      "loss": 0.0012,
      "step": 85950
    },
    {
      "epoch": 1.406528675447926,
      "grad_norm": 0.09165991842746735,
      "learning_rate": 6.397615469331384e-06,
      "loss": 0.003,
      "step": 85960
    },
    {
      "epoch": 1.4066923013990018,
      "grad_norm": 0.0652407631278038,
      "learning_rate": 6.396701449239286e-06,
      "loss": 0.0024,
      "step": 85970
    },
    {
      "epoch": 1.4068559273500778,
      "grad_norm": 0.10655125975608826,
      "learning_rate": 6.395787378520117e-06,
      "loss": 0.0041,
      "step": 85980
    },
    {
      "epoch": 1.4070195533011536,
      "grad_norm": 0.16529837250709534,
      "learning_rate": 6.394873257207009e-06,
      "loss": 0.0018,
      "step": 85990
    },
    {
      "epoch": 1.4071831792522294,
      "grad_norm": 0.12448734790086746,
      "learning_rate": 6.3939590853331e-06,
      "loss": 0.0028,
      "step": 86000
    },
    {
      "epoch": 1.4073468052033054,
      "grad_norm": 0.12860512733459473,
      "learning_rate": 6.3930448629315236e-06,
      "loss": 0.0023,
      "step": 86010
    },
    {
      "epoch": 1.4075104311543811,
      "grad_norm": 0.06594104319810867,
      "learning_rate": 6.3921305900354165e-06,
      "loss": 0.0016,
      "step": 86020
    },
    {
      "epoch": 1.407674057105457,
      "grad_norm": 0.13318806886672974,
      "learning_rate": 6.391216266677924e-06,
      "loss": 0.0024,
      "step": 86030
    },
    {
      "epoch": 1.407837683056533,
      "grad_norm": 0.10796231031417847,
      "learning_rate": 6.390301892892186e-06,
      "loss": 0.0023,
      "step": 86040
    },
    {
      "epoch": 1.4080013090076087,
      "grad_norm": 0.27822428941726685,
      "learning_rate": 6.389387468711346e-06,
      "loss": 0.0045,
      "step": 86050
    },
    {
      "epoch": 1.4081649349586844,
      "grad_norm": 0.16741658747196198,
      "learning_rate": 6.388472994168548e-06,
      "loss": 0.0015,
      "step": 86060
    },
    {
      "epoch": 1.4083285609097602,
      "grad_norm": 0.007848259061574936,
      "learning_rate": 6.3875584692969414e-06,
      "loss": 0.0016,
      "step": 86070
    },
    {
      "epoch": 1.408492186860836,
      "grad_norm": 0.03564431518316269,
      "learning_rate": 6.386643894129676e-06,
      "loss": 0.0015,
      "step": 86080
    },
    {
      "epoch": 1.408655812811912,
      "grad_norm": 0.03646448627114296,
      "learning_rate": 6.3857292686999e-06,
      "loss": 0.0014,
      "step": 86090
    },
    {
      "epoch": 1.4088194387629878,
      "grad_norm": 0.10099207609891891,
      "learning_rate": 6.384814593040771e-06,
      "loss": 0.0015,
      "step": 86100
    },
    {
      "epoch": 1.4089830647140635,
      "grad_norm": 0.10978300124406815,
      "learning_rate": 6.3838998671854405e-06,
      "loss": 0.0018,
      "step": 86110
    },
    {
      "epoch": 1.4091466906651395,
      "grad_norm": 0.055727116763591766,
      "learning_rate": 6.382985091167065e-06,
      "loss": 0.0016,
      "step": 86120
    },
    {
      "epoch": 1.4093103166162153,
      "grad_norm": 0.13373956084251404,
      "learning_rate": 6.382070265018805e-06,
      "loss": 0.0018,
      "step": 86130
    },
    {
      "epoch": 1.409473942567291,
      "grad_norm": 0.008454692550003529,
      "learning_rate": 6.38115538877382e-06,
      "loss": 0.0021,
      "step": 86140
    },
    {
      "epoch": 1.409637568518367,
      "grad_norm": 0.12276609987020493,
      "learning_rate": 6.380240462465272e-06,
      "loss": 0.0018,
      "step": 86150
    },
    {
      "epoch": 1.4098011944694429,
      "grad_norm": 0.3033892512321472,
      "learning_rate": 6.379325486126322e-06,
      "loss": 0.0024,
      "step": 86160
    },
    {
      "epoch": 1.4099648204205186,
      "grad_norm": 0.03300086036324501,
      "learning_rate": 6.37841045979014e-06,
      "loss": 0.0017,
      "step": 86170
    },
    {
      "epoch": 1.4101284463715946,
      "grad_norm": 0.05865491181612015,
      "learning_rate": 6.377495383489891e-06,
      "loss": 0.0019,
      "step": 86180
    },
    {
      "epoch": 1.4102920723226704,
      "grad_norm": 0.04635308310389519,
      "learning_rate": 6.376580257258746e-06,
      "loss": 0.0015,
      "step": 86190
    },
    {
      "epoch": 1.4104556982737462,
      "grad_norm": 0.07225222140550613,
      "learning_rate": 6.375665081129876e-06,
      "loss": 0.0014,
      "step": 86200
    },
    {
      "epoch": 1.4106193242248222,
      "grad_norm": 0.07321282476186752,
      "learning_rate": 6.374749855136451e-06,
      "loss": 0.0035,
      "step": 86210
    },
    {
      "epoch": 1.410782950175898,
      "grad_norm": 0.13845479488372803,
      "learning_rate": 6.37383457931165e-06,
      "loss": 0.0015,
      "step": 86220
    },
    {
      "epoch": 1.4109465761269737,
      "grad_norm": 0.24736085534095764,
      "learning_rate": 6.372919253688645e-06,
      "loss": 0.0025,
      "step": 86230
    },
    {
      "epoch": 1.4111102020780497,
      "grad_norm": 0.014483810402452946,
      "learning_rate": 6.372003878300618e-06,
      "loss": 0.0019,
      "step": 86240
    },
    {
      "epoch": 1.4112738280291255,
      "grad_norm": 0.07591627538204193,
      "learning_rate": 6.371088453180749e-06,
      "loss": 0.0014,
      "step": 86250
    },
    {
      "epoch": 1.4114374539802013,
      "grad_norm": 0.06420943140983582,
      "learning_rate": 6.370172978362217e-06,
      "loss": 0.0028,
      "step": 86260
    },
    {
      "epoch": 1.411601079931277,
      "grad_norm": 0.08503778278827667,
      "learning_rate": 6.369257453878208e-06,
      "loss": 0.0021,
      "step": 86270
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 0.021942317485809326,
      "learning_rate": 6.3683418797619075e-06,
      "loss": 0.0014,
      "step": 86280
    },
    {
      "epoch": 1.4119283318334288,
      "grad_norm": 0.06504148989915848,
      "learning_rate": 6.3674262560465036e-06,
      "loss": 0.0021,
      "step": 86290
    },
    {
      "epoch": 1.4120919577845046,
      "grad_norm": 0.06938780844211578,
      "learning_rate": 6.366510582765184e-06,
      "loss": 0.0018,
      "step": 86300
    },
    {
      "epoch": 1.4122555837355804,
      "grad_norm": 0.06081559509038925,
      "learning_rate": 6.36559485995114e-06,
      "loss": 0.0021,
      "step": 86310
    },
    {
      "epoch": 1.4124192096866564,
      "grad_norm": 0.16031400859355927,
      "learning_rate": 6.3646790876375645e-06,
      "loss": 0.0021,
      "step": 86320
    },
    {
      "epoch": 1.4125828356377321,
      "grad_norm": 0.1319561004638672,
      "learning_rate": 6.3637632658576505e-06,
      "loss": 0.0016,
      "step": 86330
    },
    {
      "epoch": 1.412746461588808,
      "grad_norm": 0.09470713138580322,
      "learning_rate": 6.362847394644598e-06,
      "loss": 0.0019,
      "step": 86340
    },
    {
      "epoch": 1.412910087539884,
      "grad_norm": 0.020638514310121536,
      "learning_rate": 6.361931474031602e-06,
      "loss": 0.0055,
      "step": 86350
    },
    {
      "epoch": 1.4130737134909597,
      "grad_norm": 0.08732564002275467,
      "learning_rate": 6.361015504051864e-06,
      "loss": 0.0035,
      "step": 86360
    },
    {
      "epoch": 1.4132373394420354,
      "grad_norm": 0.0567031092941761,
      "learning_rate": 6.3600994847385845e-06,
      "loss": 0.0018,
      "step": 86370
    },
    {
      "epoch": 1.4134009653931114,
      "grad_norm": 0.17103177309036255,
      "learning_rate": 6.359183416124965e-06,
      "loss": 0.0019,
      "step": 86380
    },
    {
      "epoch": 1.4135645913441872,
      "grad_norm": 0.06520421802997589,
      "learning_rate": 6.358267298244217e-06,
      "loss": 0.0026,
      "step": 86390
    },
    {
      "epoch": 1.413728217295263,
      "grad_norm": 0.09191549569368362,
      "learning_rate": 6.357351131129543e-06,
      "loss": 0.0013,
      "step": 86400
    },
    {
      "epoch": 1.413891843246339,
      "grad_norm": 0.10752929747104645,
      "learning_rate": 6.356434914814154e-06,
      "loss": 0.0014,
      "step": 86410
    },
    {
      "epoch": 1.4140554691974148,
      "grad_norm": 0.10741899907588959,
      "learning_rate": 6.355518649331258e-06,
      "loss": 0.0019,
      "step": 86420
    },
    {
      "epoch": 1.4142190951484905,
      "grad_norm": 0.15231560170650482,
      "learning_rate": 6.3546023347140686e-06,
      "loss": 0.002,
      "step": 86430
    },
    {
      "epoch": 1.4143827210995663,
      "grad_norm": 0.10231545567512512,
      "learning_rate": 6.353685970995801e-06,
      "loss": 0.0018,
      "step": 86440
    },
    {
      "epoch": 1.4145463470506423,
      "grad_norm": 0.08341783285140991,
      "learning_rate": 6.35276955820967e-06,
      "loss": 0.0037,
      "step": 86450
    },
    {
      "epoch": 1.414709973001718,
      "grad_norm": 0.03726617619395256,
      "learning_rate": 6.351853096388893e-06,
      "loss": 0.0013,
      "step": 86460
    },
    {
      "epoch": 1.4148735989527939,
      "grad_norm": 0.10797476023435593,
      "learning_rate": 6.350936585566692e-06,
      "loss": 0.0018,
      "step": 86470
    },
    {
      "epoch": 1.4150372249038696,
      "grad_norm": 0.06698980182409286,
      "learning_rate": 6.350020025776286e-06,
      "loss": 0.002,
      "step": 86480
    },
    {
      "epoch": 1.4152008508549456,
      "grad_norm": 0.026316193863749504,
      "learning_rate": 6.349103417050898e-06,
      "loss": 0.0022,
      "step": 86490
    },
    {
      "epoch": 1.4153644768060214,
      "grad_norm": 0.08767364919185638,
      "learning_rate": 6.348186759423756e-06,
      "loss": 0.0024,
      "step": 86500
    },
    {
      "epoch": 1.4155281027570972,
      "grad_norm": 0.28113704919815063,
      "learning_rate": 6.3472700529280825e-06,
      "loss": 0.0014,
      "step": 86510
    },
    {
      "epoch": 1.4156917287081732,
      "grad_norm": 0.07719475775957108,
      "learning_rate": 6.346353297597107e-06,
      "loss": 0.0015,
      "step": 86520
    },
    {
      "epoch": 1.415855354659249,
      "grad_norm": 0.0954866036772728,
      "learning_rate": 6.345436493464061e-06,
      "loss": 0.0023,
      "step": 86530
    },
    {
      "epoch": 1.4160189806103247,
      "grad_norm": 0.13685014843940735,
      "learning_rate": 6.344519640562176e-06,
      "loss": 0.0028,
      "step": 86540
    },
    {
      "epoch": 1.4161826065614007,
      "grad_norm": 0.06707563251256943,
      "learning_rate": 6.343602738924686e-06,
      "loss": 0.0019,
      "step": 86550
    },
    {
      "epoch": 1.4163462325124765,
      "grad_norm": 0.16344474256038666,
      "learning_rate": 6.3426857885848245e-06,
      "loss": 0.0035,
      "step": 86560
    },
    {
      "epoch": 1.4165098584635523,
      "grad_norm": 0.3190053105354309,
      "learning_rate": 6.341768789575832e-06,
      "loss": 0.0025,
      "step": 86570
    },
    {
      "epoch": 1.4166734844146283,
      "grad_norm": 0.08888473361730576,
      "learning_rate": 6.340851741930944e-06,
      "loss": 0.0035,
      "step": 86580
    },
    {
      "epoch": 1.416837110365704,
      "grad_norm": 0.164824977517128,
      "learning_rate": 6.339934645683403e-06,
      "loss": 0.0026,
      "step": 86590
    },
    {
      "epoch": 1.4170007363167798,
      "grad_norm": 0.06405527144670486,
      "learning_rate": 6.339017500866453e-06,
      "loss": 0.0035,
      "step": 86600
    },
    {
      "epoch": 1.4171643622678558,
      "grad_norm": 0.0634603276848793,
      "learning_rate": 6.3381003075133366e-06,
      "loss": 0.0015,
      "step": 86610
    },
    {
      "epoch": 1.4173279882189316,
      "grad_norm": 0.13260386884212494,
      "learning_rate": 6.3371830656573e-06,
      "loss": 0.0013,
      "step": 86620
    },
    {
      "epoch": 1.4174916141700074,
      "grad_norm": 0.025781145319342613,
      "learning_rate": 6.33626577533159e-06,
      "loss": 0.001,
      "step": 86630
    },
    {
      "epoch": 1.4176552401210831,
      "grad_norm": 0.04053422436118126,
      "learning_rate": 6.335348436569459e-06,
      "loss": 0.0021,
      "step": 86640
    },
    {
      "epoch": 1.4178188660721591,
      "grad_norm": 0.24975523352622986,
      "learning_rate": 6.334431049404156e-06,
      "loss": 0.0018,
      "step": 86650
    },
    {
      "epoch": 1.417982492023235,
      "grad_norm": 0.20761322975158691,
      "learning_rate": 6.3335136138689336e-06,
      "loss": 0.0019,
      "step": 86660
    },
    {
      "epoch": 1.4181461179743107,
      "grad_norm": 0.11526196449995041,
      "learning_rate": 6.3325961299970475e-06,
      "loss": 0.002,
      "step": 86670
    },
    {
      "epoch": 1.4183097439253864,
      "grad_norm": 0.06664963066577911,
      "learning_rate": 6.331678597821756e-06,
      "loss": 0.0017,
      "step": 86680
    },
    {
      "epoch": 1.4184733698764624,
      "grad_norm": 0.321092426776886,
      "learning_rate": 6.330761017376315e-06,
      "loss": 0.0039,
      "step": 86690
    },
    {
      "epoch": 1.4186369958275382,
      "grad_norm": 0.04938043653964996,
      "learning_rate": 6.329843388693987e-06,
      "loss": 0.0013,
      "step": 86700
    },
    {
      "epoch": 1.418800621778614,
      "grad_norm": 0.07609985768795013,
      "learning_rate": 6.328925711808032e-06,
      "loss": 0.0025,
      "step": 86710
    },
    {
      "epoch": 1.41896424772969,
      "grad_norm": 0.024673007428646088,
      "learning_rate": 6.328007986751712e-06,
      "loss": 0.0017,
      "step": 86720
    },
    {
      "epoch": 1.4191278736807658,
      "grad_norm": 0.20874156057834625,
      "learning_rate": 6.3270902135582954e-06,
      "loss": 0.0023,
      "step": 86730
    },
    {
      "epoch": 1.4192914996318415,
      "grad_norm": 0.025386065244674683,
      "learning_rate": 6.326172392261047e-06,
      "loss": 0.0016,
      "step": 86740
    },
    {
      "epoch": 1.4194551255829175,
      "grad_norm": 0.18945609033107758,
      "learning_rate": 6.32525452289324e-06,
      "loss": 0.0025,
      "step": 86750
    },
    {
      "epoch": 1.4196187515339933,
      "grad_norm": 0.0067061251029372215,
      "learning_rate": 6.3243366054881394e-06,
      "loss": 0.0022,
      "step": 86760
    },
    {
      "epoch": 1.419782377485069,
      "grad_norm": 0.1727306842803955,
      "learning_rate": 6.32341864007902e-06,
      "loss": 0.0016,
      "step": 86770
    },
    {
      "epoch": 1.419946003436145,
      "grad_norm": 0.09030259400606155,
      "learning_rate": 6.322500626699155e-06,
      "loss": 0.0016,
      "step": 86780
    },
    {
      "epoch": 1.4201096293872209,
      "grad_norm": 0.06210469827055931,
      "learning_rate": 6.321582565381822e-06,
      "loss": 0.0018,
      "step": 86790
    },
    {
      "epoch": 1.4202732553382966,
      "grad_norm": 0.08547190576791763,
      "learning_rate": 6.320664456160297e-06,
      "loss": 0.0023,
      "step": 86800
    },
    {
      "epoch": 1.4204368812893726,
      "grad_norm": 0.07036415487527847,
      "learning_rate": 6.319746299067859e-06,
      "loss": 0.0024,
      "step": 86810
    },
    {
      "epoch": 1.4206005072404484,
      "grad_norm": 0.1708243042230606,
      "learning_rate": 6.31882809413779e-06,
      "loss": 0.0028,
      "step": 86820
    },
    {
      "epoch": 1.4207641331915242,
      "grad_norm": 0.10920480638742447,
      "learning_rate": 6.317909841403372e-06,
      "loss": 0.0019,
      "step": 86830
    },
    {
      "epoch": 1.4209277591426,
      "grad_norm": 0.10143332928419113,
      "learning_rate": 6.31699154089789e-06,
      "loss": 0.0013,
      "step": 86840
    },
    {
      "epoch": 1.4210913850936757,
      "grad_norm": 0.048016954213380814,
      "learning_rate": 6.316073192654632e-06,
      "loss": 0.0031,
      "step": 86850
    },
    {
      "epoch": 1.4212550110447517,
      "grad_norm": 0.09750109910964966,
      "learning_rate": 6.315154796706881e-06,
      "loss": 0.0035,
      "step": 86860
    },
    {
      "epoch": 1.4214186369958275,
      "grad_norm": 0.01112271286547184,
      "learning_rate": 6.314236353087929e-06,
      "loss": 0.0019,
      "step": 86870
    },
    {
      "epoch": 1.4215822629469033,
      "grad_norm": 0.14084471762180328,
      "learning_rate": 6.313317861831069e-06,
      "loss": 0.0017,
      "step": 86880
    },
    {
      "epoch": 1.4217458888979793,
      "grad_norm": 0.1658242791891098,
      "learning_rate": 6.312399322969593e-06,
      "loss": 0.0022,
      "step": 86890
    },
    {
      "epoch": 1.421909514849055,
      "grad_norm": 0.05576367676258087,
      "learning_rate": 6.311480736536798e-06,
      "loss": 0.0017,
      "step": 86900
    },
    {
      "epoch": 1.4220731408001308,
      "grad_norm": 0.1433534026145935,
      "learning_rate": 6.3105621025659766e-06,
      "loss": 0.0019,
      "step": 86910
    },
    {
      "epoch": 1.4222367667512068,
      "grad_norm": 0.03527117148041725,
      "learning_rate": 6.309643421090429e-06,
      "loss": 0.0017,
      "step": 86920
    },
    {
      "epoch": 1.4224003927022826,
      "grad_norm": 0.2578357458114624,
      "learning_rate": 6.308724692143454e-06,
      "loss": 0.002,
      "step": 86930
    },
    {
      "epoch": 1.4225640186533584,
      "grad_norm": 0.009160785004496574,
      "learning_rate": 6.307805915758356e-06,
      "loss": 0.0028,
      "step": 86940
    },
    {
      "epoch": 1.4227276446044343,
      "grad_norm": 0.0328209213912487,
      "learning_rate": 6.306887091968435e-06,
      "loss": 0.0017,
      "step": 86950
    },
    {
      "epoch": 1.4228912705555101,
      "grad_norm": 0.029015813022851944,
      "learning_rate": 6.305968220806999e-06,
      "loss": 0.0017,
      "step": 86960
    },
    {
      "epoch": 1.423054896506586,
      "grad_norm": 0.08222070336341858,
      "learning_rate": 6.305049302307354e-06,
      "loss": 0.0013,
      "step": 86970
    },
    {
      "epoch": 1.423218522457662,
      "grad_norm": 0.09447557479143143,
      "learning_rate": 6.3041303365028075e-06,
      "loss": 0.0016,
      "step": 86980
    },
    {
      "epoch": 1.4233821484087377,
      "grad_norm": 0.18126505613327026,
      "learning_rate": 6.303211323426671e-06,
      "loss": 0.0016,
      "step": 86990
    },
    {
      "epoch": 1.4235457743598134,
      "grad_norm": 0.05435970053076744,
      "learning_rate": 6.3022922631122575e-06,
      "loss": 0.0017,
      "step": 87000
    },
    {
      "epoch": 1.4237094003108894,
      "grad_norm": 0.055926378816366196,
      "learning_rate": 6.301373155592878e-06,
      "loss": 0.0023,
      "step": 87010
    },
    {
      "epoch": 1.4238730262619652,
      "grad_norm": 0.1094655767083168,
      "learning_rate": 6.30045400090185e-06,
      "loss": 0.0025,
      "step": 87020
    },
    {
      "epoch": 1.424036652213041,
      "grad_norm": 0.08583476394414902,
      "learning_rate": 6.299534799072488e-06,
      "loss": 0.0024,
      "step": 87030
    },
    {
      "epoch": 1.4242002781641168,
      "grad_norm": 0.10022076219320297,
      "learning_rate": 6.298615550138116e-06,
      "loss": 0.0031,
      "step": 87040
    },
    {
      "epoch": 1.4243639041151925,
      "grad_norm": 0.08233474940061569,
      "learning_rate": 6.297696254132051e-06,
      "loss": 0.002,
      "step": 87050
    },
    {
      "epoch": 1.4245275300662685,
      "grad_norm": 0.050768155604600906,
      "learning_rate": 6.296776911087614e-06,
      "loss": 0.0017,
      "step": 87060
    },
    {
      "epoch": 1.4246911560173443,
      "grad_norm": 0.10270874202251434,
      "learning_rate": 6.295857521038132e-06,
      "loss": 0.0009,
      "step": 87070
    },
    {
      "epoch": 1.42485478196842,
      "grad_norm": 0.10369263589382172,
      "learning_rate": 6.294938084016928e-06,
      "loss": 0.0014,
      "step": 87080
    },
    {
      "epoch": 1.425018407919496,
      "grad_norm": 0.16410450637340546,
      "learning_rate": 6.294018600057331e-06,
      "loss": 0.0028,
      "step": 87090
    },
    {
      "epoch": 1.4251820338705719,
      "grad_norm": 0.024053268134593964,
      "learning_rate": 6.293099069192672e-06,
      "loss": 0.0014,
      "step": 87100
    },
    {
      "epoch": 1.4253456598216476,
      "grad_norm": 0.1633210927248001,
      "learning_rate": 6.292179491456279e-06,
      "loss": 0.0015,
      "step": 87110
    },
    {
      "epoch": 1.4255092857727236,
      "grad_norm": 0.15724772214889526,
      "learning_rate": 6.291259866881483e-06,
      "loss": 0.0027,
      "step": 87120
    },
    {
      "epoch": 1.4256729117237994,
      "grad_norm": 0.19977770745754242,
      "learning_rate": 6.290340195501622e-06,
      "loss": 0.002,
      "step": 87130
    },
    {
      "epoch": 1.4258365376748752,
      "grad_norm": 0.08609337359666824,
      "learning_rate": 6.28942047735003e-06,
      "loss": 0.0021,
      "step": 87140
    },
    {
      "epoch": 1.4260001636259512,
      "grad_norm": 0.17799293994903564,
      "learning_rate": 6.288500712460044e-06,
      "loss": 0.0021,
      "step": 87150
    },
    {
      "epoch": 1.426163789577027,
      "grad_norm": 0.017902495339512825,
      "learning_rate": 6.287580900865005e-06,
      "loss": 0.0008,
      "step": 87160
    },
    {
      "epoch": 1.4263274155281027,
      "grad_norm": 0.08376562595367432,
      "learning_rate": 6.286661042598251e-06,
      "loss": 0.0019,
      "step": 87170
    },
    {
      "epoch": 1.4264910414791787,
      "grad_norm": 0.016464361920952797,
      "learning_rate": 6.285741137693129e-06,
      "loss": 0.0007,
      "step": 87180
    },
    {
      "epoch": 1.4266546674302545,
      "grad_norm": 0.061246857047080994,
      "learning_rate": 6.2848211861829785e-06,
      "loss": 0.0013,
      "step": 87190
    },
    {
      "epoch": 1.4268182933813303,
      "grad_norm": 0.015438602305948734,
      "learning_rate": 6.2839011881011516e-06,
      "loss": 0.0025,
      "step": 87200
    },
    {
      "epoch": 1.4269819193324063,
      "grad_norm": 0.10207697749137878,
      "learning_rate": 6.28298114348099e-06,
      "loss": 0.0021,
      "step": 87210
    },
    {
      "epoch": 1.427145545283482,
      "grad_norm": 0.13249380886554718,
      "learning_rate": 6.2820610523558454e-06,
      "loss": 0.0016,
      "step": 87220
    },
    {
      "epoch": 1.4273091712345578,
      "grad_norm": 0.1997976154088974,
      "learning_rate": 6.2811409147590684e-06,
      "loss": 0.0023,
      "step": 87230
    },
    {
      "epoch": 1.4274727971856336,
      "grad_norm": 0.35590627789497375,
      "learning_rate": 6.280220730724013e-06,
      "loss": 0.0024,
      "step": 87240
    },
    {
      "epoch": 1.4276364231367094,
      "grad_norm": 0.10247053951025009,
      "learning_rate": 6.279300500284035e-06,
      "loss": 0.0016,
      "step": 87250
    },
    {
      "epoch": 1.4278000490877854,
      "grad_norm": 0.12032989412546158,
      "learning_rate": 6.278380223472487e-06,
      "loss": 0.0022,
      "step": 87260
    },
    {
      "epoch": 1.4279636750388611,
      "grad_norm": 0.04637526720762253,
      "learning_rate": 6.2774599003227274e-06,
      "loss": 0.002,
      "step": 87270
    },
    {
      "epoch": 1.428127300989937,
      "grad_norm": 0.06521841883659363,
      "learning_rate": 6.276539530868117e-06,
      "loss": 0.0026,
      "step": 87280
    },
    {
      "epoch": 1.428290926941013,
      "grad_norm": 0.09272162616252899,
      "learning_rate": 6.275619115142018e-06,
      "loss": 0.0013,
      "step": 87290
    },
    {
      "epoch": 1.4284545528920887,
      "grad_norm": 0.011339899152517319,
      "learning_rate": 6.274698653177792e-06,
      "loss": 0.0018,
      "step": 87300
    },
    {
      "epoch": 1.4286181788431644,
      "grad_norm": 0.03778223320841789,
      "learning_rate": 6.273778145008802e-06,
      "loss": 0.0018,
      "step": 87310
    },
    {
      "epoch": 1.4287818047942404,
      "grad_norm": 0.10297144949436188,
      "learning_rate": 6.272857590668418e-06,
      "loss": 0.0013,
      "step": 87320
    },
    {
      "epoch": 1.4289454307453162,
      "grad_norm": 0.14527471363544464,
      "learning_rate": 6.271936990190003e-06,
      "loss": 0.0015,
      "step": 87330
    },
    {
      "epoch": 1.429109056696392,
      "grad_norm": 0.0076146963983774185,
      "learning_rate": 6.271016343606931e-06,
      "loss": 0.001,
      "step": 87340
    },
    {
      "epoch": 1.429272682647468,
      "grad_norm": 0.052294228225946426,
      "learning_rate": 6.27009565095257e-06,
      "loss": 0.0017,
      "step": 87350
    },
    {
      "epoch": 1.4294363085985438,
      "grad_norm": 0.09691692143678665,
      "learning_rate": 6.2691749122602954e-06,
      "loss": 0.0031,
      "step": 87360
    },
    {
      "epoch": 1.4295999345496195,
      "grad_norm": 0.03427308797836304,
      "learning_rate": 6.268254127563479e-06,
      "loss": 0.0021,
      "step": 87370
    },
    {
      "epoch": 1.4297635605006955,
      "grad_norm": 0.10378725826740265,
      "learning_rate": 6.267333296895499e-06,
      "loss": 0.0021,
      "step": 87380
    },
    {
      "epoch": 1.4299271864517713,
      "grad_norm": 0.032112445682287216,
      "learning_rate": 6.266412420289734e-06,
      "loss": 0.0022,
      "step": 87390
    },
    {
      "epoch": 1.430090812402847,
      "grad_norm": 0.0702488124370575,
      "learning_rate": 6.265491497779564e-06,
      "loss": 0.0014,
      "step": 87400
    },
    {
      "epoch": 1.4302544383539229,
      "grad_norm": 0.04237963259220123,
      "learning_rate": 6.264570529398366e-06,
      "loss": 0.0013,
      "step": 87410
    },
    {
      "epoch": 1.4304180643049988,
      "grad_norm": 0.016835927963256836,
      "learning_rate": 6.263649515179526e-06,
      "loss": 0.0012,
      "step": 87420
    },
    {
      "epoch": 1.4305816902560746,
      "grad_norm": 0.136257141828537,
      "learning_rate": 6.262728455156428e-06,
      "loss": 0.0028,
      "step": 87430
    },
    {
      "epoch": 1.4307453162071504,
      "grad_norm": 0.06842660903930664,
      "learning_rate": 6.26180734936246e-06,
      "loss": 0.0013,
      "step": 87440
    },
    {
      "epoch": 1.4309089421582262,
      "grad_norm": 0.031677670776844025,
      "learning_rate": 6.2608861978310065e-06,
      "loss": 0.0029,
      "step": 87450
    },
    {
      "epoch": 1.4310725681093022,
      "grad_norm": 0.0027826251462101936,
      "learning_rate": 6.25996500059546e-06,
      "loss": 0.0017,
      "step": 87460
    },
    {
      "epoch": 1.431236194060378,
      "grad_norm": 0.252929151058197,
      "learning_rate": 6.2590437576892095e-06,
      "loss": 0.0019,
      "step": 87470
    },
    {
      "epoch": 1.4313998200114537,
      "grad_norm": 0.2734964191913605,
      "learning_rate": 6.25812246914565e-06,
      "loss": 0.002,
      "step": 87480
    },
    {
      "epoch": 1.4315634459625297,
      "grad_norm": 0.10104244202375412,
      "learning_rate": 6.257201134998174e-06,
      "loss": 0.0016,
      "step": 87490
    },
    {
      "epoch": 1.4317270719136055,
      "grad_norm": 0.11319293826818466,
      "learning_rate": 6.256279755280179e-06,
      "loss": 0.0016,
      "step": 87500
    },
    {
      "epoch": 1.4318906978646813,
      "grad_norm": 0.06297991424798965,
      "learning_rate": 6.255358330025061e-06,
      "loss": 0.0017,
      "step": 87510
    },
    {
      "epoch": 1.4320543238157573,
      "grad_norm": 0.07003606110811234,
      "learning_rate": 6.254436859266222e-06,
      "loss": 0.0011,
      "step": 87520
    },
    {
      "epoch": 1.432217949766833,
      "grad_norm": 0.047473471611738205,
      "learning_rate": 6.253515343037061e-06,
      "loss": 0.0014,
      "step": 87530
    },
    {
      "epoch": 1.4323815757179088,
      "grad_norm": 0.003380363341420889,
      "learning_rate": 6.252593781370981e-06,
      "loss": 0.0018,
      "step": 87540
    },
    {
      "epoch": 1.4325452016689848,
      "grad_norm": 0.09378622472286224,
      "learning_rate": 6.2516721743013894e-06,
      "loss": 0.0014,
      "step": 87550
    },
    {
      "epoch": 1.4327088276200606,
      "grad_norm": 0.12010186910629272,
      "learning_rate": 6.250750521861688e-06,
      "loss": 0.0014,
      "step": 87560
    },
    {
      "epoch": 1.4328724535711364,
      "grad_norm": 0.08450304716825485,
      "learning_rate": 6.249828824085288e-06,
      "loss": 0.0013,
      "step": 87570
    },
    {
      "epoch": 1.4330360795222123,
      "grad_norm": 0.08225636929273605,
      "learning_rate": 6.248907081005595e-06,
      "loss": 0.0027,
      "step": 87580
    },
    {
      "epoch": 1.4331997054732881,
      "grad_norm": 0.06786559522151947,
      "learning_rate": 6.247985292656022e-06,
      "loss": 0.0016,
      "step": 87590
    },
    {
      "epoch": 1.433363331424364,
      "grad_norm": 0.011361506767570972,
      "learning_rate": 6.247063459069982e-06,
      "loss": 0.0011,
      "step": 87600
    },
    {
      "epoch": 1.4335269573754397,
      "grad_norm": 0.16560013592243195,
      "learning_rate": 6.24614158028089e-06,
      "loss": 0.002,
      "step": 87610
    },
    {
      "epoch": 1.4336905833265157,
      "grad_norm": 0.16421037912368774,
      "learning_rate": 6.245219656322159e-06,
      "loss": 0.0018,
      "step": 87620
    },
    {
      "epoch": 1.4338542092775914,
      "grad_norm": 0.03722596913576126,
      "learning_rate": 6.24429768722721e-06,
      "loss": 0.0012,
      "step": 87630
    },
    {
      "epoch": 1.4340178352286672,
      "grad_norm": 0.2592485547065735,
      "learning_rate": 6.243375673029459e-06,
      "loss": 0.0036,
      "step": 87640
    },
    {
      "epoch": 1.434181461179743,
      "grad_norm": 0.08613810688257217,
      "learning_rate": 6.242453613762328e-06,
      "loss": 0.0008,
      "step": 87650
    },
    {
      "epoch": 1.434345087130819,
      "grad_norm": 0.21068191528320312,
      "learning_rate": 6.241531509459241e-06,
      "loss": 0.0021,
      "step": 87660
    },
    {
      "epoch": 1.4345087130818948,
      "grad_norm": 0.17119361460208893,
      "learning_rate": 6.24060936015362e-06,
      "loss": 0.0016,
      "step": 87670
    },
    {
      "epoch": 1.4346723390329705,
      "grad_norm": 0.07850255817174911,
      "learning_rate": 6.239687165878891e-06,
      "loss": 0.0011,
      "step": 87680
    },
    {
      "epoch": 1.4348359649840465,
      "grad_norm": 0.08151416480541229,
      "learning_rate": 6.238764926668482e-06,
      "loss": 0.0021,
      "step": 87690
    },
    {
      "epoch": 1.4349995909351223,
      "grad_norm": 0.17128393054008484,
      "learning_rate": 6.237842642555822e-06,
      "loss": 0.0022,
      "step": 87700
    },
    {
      "epoch": 1.435163216886198,
      "grad_norm": 0.16158713400363922,
      "learning_rate": 6.236920313574342e-06,
      "loss": 0.002,
      "step": 87710
    },
    {
      "epoch": 1.435326842837274,
      "grad_norm": 0.01578509248793125,
      "learning_rate": 6.235997939757473e-06,
      "loss": 0.0017,
      "step": 87720
    },
    {
      "epoch": 1.4354904687883498,
      "grad_norm": 0.1902141124010086,
      "learning_rate": 6.2350755211386486e-06,
      "loss": 0.0032,
      "step": 87730
    },
    {
      "epoch": 1.4356540947394256,
      "grad_norm": 0.1205715611577034,
      "learning_rate": 6.2341530577513045e-06,
      "loss": 0.0034,
      "step": 87740
    },
    {
      "epoch": 1.4358177206905016,
      "grad_norm": 0.054978013038635254,
      "learning_rate": 6.233230549628879e-06,
      "loss": 0.0015,
      "step": 87750
    },
    {
      "epoch": 1.4359813466415774,
      "grad_norm": 0.14329639077186584,
      "learning_rate": 6.232307996804812e-06,
      "loss": 0.0027,
      "step": 87760
    },
    {
      "epoch": 1.4361449725926532,
      "grad_norm": 0.005561621394008398,
      "learning_rate": 6.231385399312539e-06,
      "loss": 0.0027,
      "step": 87770
    },
    {
      "epoch": 1.4363085985437292,
      "grad_norm": 0.09101153165102005,
      "learning_rate": 6.230462757185507e-06,
      "loss": 0.002,
      "step": 87780
    },
    {
      "epoch": 1.436472224494805,
      "grad_norm": 0.17430588603019714,
      "learning_rate": 6.2295400704571565e-06,
      "loss": 0.0022,
      "step": 87790
    },
    {
      "epoch": 1.4366358504458807,
      "grad_norm": 0.13493849337100983,
      "learning_rate": 6.228617339160933e-06,
      "loss": 0.0018,
      "step": 87800
    },
    {
      "epoch": 1.4367994763969565,
      "grad_norm": 0.04612865298986435,
      "learning_rate": 6.227694563330284e-06,
      "loss": 0.0013,
      "step": 87810
    },
    {
      "epoch": 1.4369631023480323,
      "grad_norm": 0.0956103727221489,
      "learning_rate": 6.226771742998658e-06,
      "loss": 0.0017,
      "step": 87820
    },
    {
      "epoch": 1.4371267282991083,
      "grad_norm": 0.08610368520021439,
      "learning_rate": 6.225848878199505e-06,
      "loss": 0.0011,
      "step": 87830
    },
    {
      "epoch": 1.437290354250184,
      "grad_norm": 0.07028482109308243,
      "learning_rate": 6.224925968966277e-06,
      "loss": 0.0029,
      "step": 87840
    },
    {
      "epoch": 1.4374539802012598,
      "grad_norm": 0.10654260963201523,
      "learning_rate": 6.2240030153324245e-06,
      "loss": 0.0024,
      "step": 87850
    },
    {
      "epoch": 1.4376176061523358,
      "grad_norm": 0.07870994508266449,
      "learning_rate": 6.223080017331408e-06,
      "loss": 0.0018,
      "step": 87860
    },
    {
      "epoch": 1.4377812321034116,
      "grad_norm": 0.22111137211322784,
      "learning_rate": 6.222156974996677e-06,
      "loss": 0.0014,
      "step": 87870
    },
    {
      "epoch": 1.4379448580544874,
      "grad_norm": 0.026112055405974388,
      "learning_rate": 6.221233888361694e-06,
      "loss": 0.0014,
      "step": 87880
    },
    {
      "epoch": 1.4381084840055633,
      "grad_norm": 0.04195261374115944,
      "learning_rate": 6.220310757459919e-06,
      "loss": 0.001,
      "step": 87890
    },
    {
      "epoch": 1.4382721099566391,
      "grad_norm": 0.010488742962479591,
      "learning_rate": 6.219387582324811e-06,
      "loss": 0.0012,
      "step": 87900
    },
    {
      "epoch": 1.438435735907715,
      "grad_norm": 0.2689409852027893,
      "learning_rate": 6.218464362989834e-06,
      "loss": 0.0024,
      "step": 87910
    },
    {
      "epoch": 1.438599361858791,
      "grad_norm": 0.11243392527103424,
      "learning_rate": 6.217541099488454e-06,
      "loss": 0.0016,
      "step": 87920
    },
    {
      "epoch": 1.4387629878098667,
      "grad_norm": 0.05361480638384819,
      "learning_rate": 6.2166177918541325e-06,
      "loss": 0.0014,
      "step": 87930
    },
    {
      "epoch": 1.4389266137609424,
      "grad_norm": 0.18623264133930206,
      "learning_rate": 6.215694440120341e-06,
      "loss": 0.0022,
      "step": 87940
    },
    {
      "epoch": 1.4390902397120184,
      "grad_norm": 0.14766648411750793,
      "learning_rate": 6.2147710443205475e-06,
      "loss": 0.0027,
      "step": 87950
    },
    {
      "epoch": 1.4392538656630942,
      "grad_norm": 0.15501351654529572,
      "learning_rate": 6.213847604488225e-06,
      "loss": 0.0025,
      "step": 87960
    },
    {
      "epoch": 1.43941749161417,
      "grad_norm": 0.26560544967651367,
      "learning_rate": 6.212924120656843e-06,
      "loss": 0.0024,
      "step": 87970
    },
    {
      "epoch": 1.439581117565246,
      "grad_norm": 0.045336879789829254,
      "learning_rate": 6.212000592859877e-06,
      "loss": 0.0008,
      "step": 87980
    },
    {
      "epoch": 1.4397447435163218,
      "grad_norm": 0.28163769841194153,
      "learning_rate": 6.211077021130802e-06,
      "loss": 0.0018,
      "step": 87990
    },
    {
      "epoch": 1.4399083694673975,
      "grad_norm": 0.08188337087631226,
      "learning_rate": 6.210153405503097e-06,
      "loss": 0.0019,
      "step": 88000
    },
    {
      "epoch": 1.4400719954184733,
      "grad_norm": 0.265301376581192,
      "learning_rate": 6.209229746010239e-06,
      "loss": 0.0021,
      "step": 88010
    },
    {
      "epoch": 1.440235621369549,
      "grad_norm": 0.016715556383132935,
      "learning_rate": 6.208306042685709e-06,
      "loss": 0.002,
      "step": 88020
    },
    {
      "epoch": 1.440399247320625,
      "grad_norm": 0.23580516874790192,
      "learning_rate": 6.207382295562988e-06,
      "loss": 0.0019,
      "step": 88030
    },
    {
      "epoch": 1.4405628732717008,
      "grad_norm": 0.6500301957130432,
      "learning_rate": 6.206458504675562e-06,
      "loss": 0.0025,
      "step": 88040
    },
    {
      "epoch": 1.4407264992227766,
      "grad_norm": 0.16514134407043457,
      "learning_rate": 6.205534670056914e-06,
      "loss": 0.0015,
      "step": 88050
    },
    {
      "epoch": 1.4408901251738526,
      "grad_norm": 0.027558708563447,
      "learning_rate": 6.204610791740533e-06,
      "loss": 0.0021,
      "step": 88060
    },
    {
      "epoch": 1.4410537511249284,
      "grad_norm": 0.07553763687610626,
      "learning_rate": 6.203686869759905e-06,
      "loss": 0.0032,
      "step": 88070
    },
    {
      "epoch": 1.4412173770760042,
      "grad_norm": 0.015969766303896904,
      "learning_rate": 6.202762904148521e-06,
      "loss": 0.002,
      "step": 88080
    },
    {
      "epoch": 1.4413810030270802,
      "grad_norm": 0.16341017186641693,
      "learning_rate": 6.201838894939872e-06,
      "loss": 0.0018,
      "step": 88090
    },
    {
      "epoch": 1.441544628978156,
      "grad_norm": 0.05430678650736809,
      "learning_rate": 6.200914842167451e-06,
      "loss": 0.0013,
      "step": 88100
    },
    {
      "epoch": 1.4417082549292317,
      "grad_norm": 0.044062547385692596,
      "learning_rate": 6.199990745864757e-06,
      "loss": 0.0024,
      "step": 88110
    },
    {
      "epoch": 1.4418718808803077,
      "grad_norm": 0.20913057029247284,
      "learning_rate": 6.199066606065279e-06,
      "loss": 0.0023,
      "step": 88120
    },
    {
      "epoch": 1.4420355068313835,
      "grad_norm": 0.0816015675663948,
      "learning_rate": 6.19814242280252e-06,
      "loss": 0.0023,
      "step": 88130
    },
    {
      "epoch": 1.4421991327824593,
      "grad_norm": 0.10608971863985062,
      "learning_rate": 6.197218196109976e-06,
      "loss": 0.002,
      "step": 88140
    },
    {
      "epoch": 1.4423627587335353,
      "grad_norm": 0.05991344898939133,
      "learning_rate": 6.196293926021152e-06,
      "loss": 0.0025,
      "step": 88150
    },
    {
      "epoch": 1.442526384684611,
      "grad_norm": 0.1403913050889969,
      "learning_rate": 6.195369612569547e-06,
      "loss": 0.0017,
      "step": 88160
    },
    {
      "epoch": 1.4426900106356868,
      "grad_norm": 0.15912337601184845,
      "learning_rate": 6.194445255788667e-06,
      "loss": 0.0016,
      "step": 88170
    },
    {
      "epoch": 1.4428536365867626,
      "grad_norm": 0.1175880879163742,
      "learning_rate": 6.193520855712018e-06,
      "loss": 0.005,
      "step": 88180
    },
    {
      "epoch": 1.4430172625378386,
      "grad_norm": 0.2903042733669281,
      "learning_rate": 6.192596412373105e-06,
      "loss": 0.0033,
      "step": 88190
    },
    {
      "epoch": 1.4431808884889143,
      "grad_norm": 0.046630214899778366,
      "learning_rate": 6.1916719258054395e-06,
      "loss": 0.0014,
      "step": 88200
    },
    {
      "epoch": 1.4433445144399901,
      "grad_norm": 0.03290325403213501,
      "learning_rate": 6.1907473960425315e-06,
      "loss": 0.001,
      "step": 88210
    },
    {
      "epoch": 1.443508140391066,
      "grad_norm": 0.1491260826587677,
      "learning_rate": 6.18982282311789e-06,
      "loss": 0.0023,
      "step": 88220
    },
    {
      "epoch": 1.443671766342142,
      "grad_norm": 0.27651795744895935,
      "learning_rate": 6.188898207065032e-06,
      "loss": 0.0022,
      "step": 88230
    },
    {
      "epoch": 1.4438353922932177,
      "grad_norm": 0.1683707982301712,
      "learning_rate": 6.187973547917471e-06,
      "loss": 0.0037,
      "step": 88240
    },
    {
      "epoch": 1.4439990182442934,
      "grad_norm": 0.07415008544921875,
      "learning_rate": 6.1870488457087254e-06,
      "loss": 0.0026,
      "step": 88250
    },
    {
      "epoch": 1.4441626441953694,
      "grad_norm": 0.19869717955589294,
      "learning_rate": 6.186124100472311e-06,
      "loss": 0.0015,
      "step": 88260
    },
    {
      "epoch": 1.4443262701464452,
      "grad_norm": 0.1601708084344864,
      "learning_rate": 6.18519931224175e-06,
      "loss": 0.0022,
      "step": 88270
    },
    {
      "epoch": 1.444489896097521,
      "grad_norm": 0.22000622749328613,
      "learning_rate": 6.1842744810505606e-06,
      "loss": 0.0016,
      "step": 88280
    },
    {
      "epoch": 1.444653522048597,
      "grad_norm": 0.12108389288187027,
      "learning_rate": 6.183349606932269e-06,
      "loss": 0.003,
      "step": 88290
    },
    {
      "epoch": 1.4448171479996728,
      "grad_norm": 0.21934489905834198,
      "learning_rate": 6.1824246899203966e-06,
      "loss": 0.0036,
      "step": 88300
    },
    {
      "epoch": 1.4449807739507485,
      "grad_norm": 0.07911412417888641,
      "learning_rate": 6.181499730048474e-06,
      "loss": 0.0009,
      "step": 88310
    },
    {
      "epoch": 1.4451443999018245,
      "grad_norm": 0.13370271027088165,
      "learning_rate": 6.180574727350024e-06,
      "loss": 0.0026,
      "step": 88320
    },
    {
      "epoch": 1.4453080258529003,
      "grad_norm": 0.012059704400599003,
      "learning_rate": 6.179649681858577e-06,
      "loss": 0.0011,
      "step": 88330
    },
    {
      "epoch": 1.445471651803976,
      "grad_norm": 0.27589085698127747,
      "learning_rate": 6.178724593607664e-06,
      "loss": 0.0014,
      "step": 88340
    },
    {
      "epoch": 1.445635277755052,
      "grad_norm": 0.22645272314548492,
      "learning_rate": 6.177799462630819e-06,
      "loss": 0.0022,
      "step": 88350
    },
    {
      "epoch": 1.4457989037061278,
      "grad_norm": 0.010753162205219269,
      "learning_rate": 6.176874288961574e-06,
      "loss": 0.0019,
      "step": 88360
    },
    {
      "epoch": 1.4459625296572036,
      "grad_norm": 0.07627929002046585,
      "learning_rate": 6.1759490726334635e-06,
      "loss": 0.0019,
      "step": 88370
    },
    {
      "epoch": 1.4461261556082794,
      "grad_norm": 0.2940148413181305,
      "learning_rate": 6.175023813680027e-06,
      "loss": 0.0029,
      "step": 88380
    },
    {
      "epoch": 1.4462897815593554,
      "grad_norm": 0.10072994232177734,
      "learning_rate": 6.174098512134801e-06,
      "loss": 0.0023,
      "step": 88390
    },
    {
      "epoch": 1.4464534075104312,
      "grad_norm": 0.06951545923948288,
      "learning_rate": 6.173173168031325e-06,
      "loss": 0.0018,
      "step": 88400
    },
    {
      "epoch": 1.446617033461507,
      "grad_norm": 0.050156425684690475,
      "learning_rate": 6.172247781403142e-06,
      "loss": 0.0024,
      "step": 88410
    },
    {
      "epoch": 1.4467806594125827,
      "grad_norm": 0.057347770780324936,
      "learning_rate": 6.171322352283794e-06,
      "loss": 0.002,
      "step": 88420
    },
    {
      "epoch": 1.4469442853636587,
      "grad_norm": 0.052968766540288925,
      "learning_rate": 6.170396880706826e-06,
      "loss": 0.002,
      "step": 88430
    },
    {
      "epoch": 1.4471079113147345,
      "grad_norm": 0.19407209753990173,
      "learning_rate": 6.169471366705782e-06,
      "loss": 0.0031,
      "step": 88440
    },
    {
      "epoch": 1.4472715372658103,
      "grad_norm": 0.19769182801246643,
      "learning_rate": 6.168545810314214e-06,
      "loss": 0.0016,
      "step": 88450
    },
    {
      "epoch": 1.4474351632168863,
      "grad_norm": 0.07108977437019348,
      "learning_rate": 6.16762021156567e-06,
      "loss": 0.0016,
      "step": 88460
    },
    {
      "epoch": 1.447598789167962,
      "grad_norm": 0.10954627394676208,
      "learning_rate": 6.1666945704936976e-06,
      "loss": 0.0021,
      "step": 88470
    },
    {
      "epoch": 1.4477624151190378,
      "grad_norm": 0.02270127460360527,
      "learning_rate": 6.165768887131852e-06,
      "loss": 0.0027,
      "step": 88480
    },
    {
      "epoch": 1.4479260410701138,
      "grad_norm": 0.10544577240943909,
      "learning_rate": 6.1648431615136865e-06,
      "loss": 0.0022,
      "step": 88490
    },
    {
      "epoch": 1.4480896670211896,
      "grad_norm": 0.10242760926485062,
      "learning_rate": 6.1639173936727545e-06,
      "loss": 0.0019,
      "step": 88500
    },
    {
      "epoch": 1.4482532929722653,
      "grad_norm": 0.09470520913600922,
      "learning_rate": 6.162991583642615e-06,
      "loss": 0.002,
      "step": 88510
    },
    {
      "epoch": 1.4484169189233413,
      "grad_norm": 0.17790016531944275,
      "learning_rate": 6.162065731456825e-06,
      "loss": 0.0024,
      "step": 88520
    },
    {
      "epoch": 1.4485805448744171,
      "grad_norm": 0.07417348772287369,
      "learning_rate": 6.161139837148947e-06,
      "loss": 0.0031,
      "step": 88530
    },
    {
      "epoch": 1.448744170825493,
      "grad_norm": 0.03617400676012039,
      "learning_rate": 6.1602139007525396e-06,
      "loss": 0.0015,
      "step": 88540
    },
    {
      "epoch": 1.448907796776569,
      "grad_norm": 0.11755359917879105,
      "learning_rate": 6.159287922301167e-06,
      "loss": 0.0016,
      "step": 88550
    },
    {
      "epoch": 1.4490714227276447,
      "grad_norm": 0.09294844418764114,
      "learning_rate": 6.158361901828395e-06,
      "loss": 0.0014,
      "step": 88560
    },
    {
      "epoch": 1.4492350486787204,
      "grad_norm": 0.10152885317802429,
      "learning_rate": 6.157435839367786e-06,
      "loss": 0.0022,
      "step": 88570
    },
    {
      "epoch": 1.4493986746297962,
      "grad_norm": 0.14733688533306122,
      "learning_rate": 6.156509734952909e-06,
      "loss": 0.0019,
      "step": 88580
    },
    {
      "epoch": 1.4495623005808722,
      "grad_norm": 0.06413263082504272,
      "learning_rate": 6.1555835886173356e-06,
      "loss": 0.0014,
      "step": 88590
    },
    {
      "epoch": 1.449725926531948,
      "grad_norm": 0.004010655917227268,
      "learning_rate": 6.154657400394633e-06,
      "loss": 0.0017,
      "step": 88600
    },
    {
      "epoch": 1.4498895524830238,
      "grad_norm": 0.04507434740662575,
      "learning_rate": 6.153731170318376e-06,
      "loss": 0.0027,
      "step": 88610
    },
    {
      "epoch": 1.4500531784340995,
      "grad_norm": 0.06232910230755806,
      "learning_rate": 6.152804898422138e-06,
      "loss": 0.0011,
      "step": 88620
    },
    {
      "epoch": 1.4502168043851755,
      "grad_norm": 0.18974187970161438,
      "learning_rate": 6.151878584739492e-06,
      "loss": 0.0056,
      "step": 88630
    },
    {
      "epoch": 1.4503804303362513,
      "grad_norm": 0.32222968339920044,
      "learning_rate": 6.150952229304015e-06,
      "loss": 0.0024,
      "step": 88640
    },
    {
      "epoch": 1.450544056287327,
      "grad_norm": 0.20506054162979126,
      "learning_rate": 6.150025832149285e-06,
      "loss": 0.0022,
      "step": 88650
    },
    {
      "epoch": 1.450707682238403,
      "grad_norm": 0.1331210881471634,
      "learning_rate": 6.149099393308885e-06,
      "loss": 0.0019,
      "step": 88660
    },
    {
      "epoch": 1.4508713081894788,
      "grad_norm": 0.026927372440695763,
      "learning_rate": 6.148172912816392e-06,
      "loss": 0.0019,
      "step": 88670
    },
    {
      "epoch": 1.4510349341405546,
      "grad_norm": 0.04572971910238266,
      "learning_rate": 6.147246390705391e-06,
      "loss": 0.0021,
      "step": 88680
    },
    {
      "epoch": 1.4511985600916306,
      "grad_norm": 0.06819307804107666,
      "learning_rate": 6.1463198270094655e-06,
      "loss": 0.0015,
      "step": 88690
    },
    {
      "epoch": 1.4513621860427064,
      "grad_norm": 0.09001713246107101,
      "learning_rate": 6.145393221762203e-06,
      "loss": 0.0021,
      "step": 88700
    },
    {
      "epoch": 1.4515258119937822,
      "grad_norm": 0.05180295556783676,
      "learning_rate": 6.1444665749971885e-06,
      "loss": 0.0011,
      "step": 88710
    },
    {
      "epoch": 1.4516894379448582,
      "grad_norm": 0.14307278394699097,
      "learning_rate": 6.143539886748009e-06,
      "loss": 0.0016,
      "step": 88720
    },
    {
      "epoch": 1.451853063895934,
      "grad_norm": 0.17013002932071686,
      "learning_rate": 6.142613157048259e-06,
      "loss": 0.0027,
      "step": 88730
    },
    {
      "epoch": 1.4520166898470097,
      "grad_norm": 0.09118823707103729,
      "learning_rate": 6.141686385931528e-06,
      "loss": 0.0023,
      "step": 88740
    },
    {
      "epoch": 1.4521803157980857,
      "grad_norm": 0.18551497161388397,
      "learning_rate": 6.1407595734314095e-06,
      "loss": 0.0019,
      "step": 88750
    },
    {
      "epoch": 1.4523439417491615,
      "grad_norm": 0.018743224442005157,
      "learning_rate": 6.139832719581499e-06,
      "loss": 0.0025,
      "step": 88760
    },
    {
      "epoch": 1.4525075677002373,
      "grad_norm": 0.08347221463918686,
      "learning_rate": 6.138905824415391e-06,
      "loss": 0.0014,
      "step": 88770
    },
    {
      "epoch": 1.452671193651313,
      "grad_norm": 0.21199621260166168,
      "learning_rate": 6.1379788879666845e-06,
      "loss": 0.0014,
      "step": 88780
    },
    {
      "epoch": 1.4528348196023888,
      "grad_norm": 0.09008124470710754,
      "learning_rate": 6.137051910268978e-06,
      "loss": 0.0014,
      "step": 88790
    },
    {
      "epoch": 1.4529984455534648,
      "grad_norm": 0.16531740128993988,
      "learning_rate": 6.136124891355871e-06,
      "loss": 0.0023,
      "step": 88800
    },
    {
      "epoch": 1.4531620715045406,
      "grad_norm": 0.13263924419879913,
      "learning_rate": 6.1351978312609704e-06,
      "loss": 0.0009,
      "step": 88810
    },
    {
      "epoch": 1.4533256974556163,
      "grad_norm": 0.10232087224721909,
      "learning_rate": 6.134270730017876e-06,
      "loss": 0.0021,
      "step": 88820
    },
    {
      "epoch": 1.4534893234066923,
      "grad_norm": 0.12131325155496597,
      "learning_rate": 6.133343587660194e-06,
      "loss": 0.0021,
      "step": 88830
    },
    {
      "epoch": 1.4536529493577681,
      "grad_norm": 0.05863536521792412,
      "learning_rate": 6.1324164042215285e-06,
      "loss": 0.0028,
      "step": 88840
    },
    {
      "epoch": 1.453816575308844,
      "grad_norm": 0.059465281665325165,
      "learning_rate": 6.131489179735492e-06,
      "loss": 0.0028,
      "step": 88850
    },
    {
      "epoch": 1.45398020125992,
      "grad_norm": 0.08583907037973404,
      "learning_rate": 6.130561914235692e-06,
      "loss": 0.0023,
      "step": 88860
    },
    {
      "epoch": 1.4541438272109957,
      "grad_norm": 0.12189940363168716,
      "learning_rate": 6.129634607755739e-06,
      "loss": 0.0018,
      "step": 88870
    },
    {
      "epoch": 1.4543074531620714,
      "grad_norm": 0.1641402393579483,
      "learning_rate": 6.128707260329248e-06,
      "loss": 0.0021,
      "step": 88880
    },
    {
      "epoch": 1.4544710791131474,
      "grad_norm": 0.15608078241348267,
      "learning_rate": 6.1277798719898295e-06,
      "loss": 0.0026,
      "step": 88890
    },
    {
      "epoch": 1.4546347050642232,
      "grad_norm": 0.10500966757535934,
      "learning_rate": 6.126852442771103e-06,
      "loss": 0.002,
      "step": 88900
    },
    {
      "epoch": 1.454798331015299,
      "grad_norm": 0.09700223803520203,
      "learning_rate": 6.1259249727066834e-06,
      "loss": 0.0021,
      "step": 88910
    },
    {
      "epoch": 1.454961956966375,
      "grad_norm": 0.18499355018138885,
      "learning_rate": 6.12499746183019e-06,
      "loss": 0.0018,
      "step": 88920
    },
    {
      "epoch": 1.4551255829174508,
      "grad_norm": 0.17238733172416687,
      "learning_rate": 6.124069910175241e-06,
      "loss": 0.0022,
      "step": 88930
    },
    {
      "epoch": 1.4552892088685265,
      "grad_norm": 0.02622097171843052,
      "learning_rate": 6.12314231777546e-06,
      "loss": 0.0022,
      "step": 88940
    },
    {
      "epoch": 1.4554528348196025,
      "grad_norm": 0.1282316893339157,
      "learning_rate": 6.122214684664469e-06,
      "loss": 0.0023,
      "step": 88950
    },
    {
      "epoch": 1.4556164607706783,
      "grad_norm": 0.144830122590065,
      "learning_rate": 6.121287010875895e-06,
      "loss": 0.0018,
      "step": 88960
    },
    {
      "epoch": 1.455780086721754,
      "grad_norm": 0.0503481850028038,
      "learning_rate": 6.12035929644336e-06,
      "loss": 0.0016,
      "step": 88970
    },
    {
      "epoch": 1.4559437126728298,
      "grad_norm": 0.12261451780796051,
      "learning_rate": 6.1194315414004935e-06,
      "loss": 0.0025,
      "step": 88980
    },
    {
      "epoch": 1.4561073386239056,
      "grad_norm": 0.06970880180597305,
      "learning_rate": 6.118503745780924e-06,
      "loss": 0.0023,
      "step": 88990
    },
    {
      "epoch": 1.4562709645749816,
      "grad_norm": 0.140335813164711,
      "learning_rate": 6.117575909618282e-06,
      "loss": 0.0017,
      "step": 89000
    },
    {
      "epoch": 1.4564345905260574,
      "grad_norm": 0.09891358017921448,
      "learning_rate": 6.1166480329462e-06,
      "loss": 0.0011,
      "step": 89010
    },
    {
      "epoch": 1.4565982164771332,
      "grad_norm": 0.06568451970815659,
      "learning_rate": 6.115720115798309e-06,
      "loss": 0.0018,
      "step": 89020
    },
    {
      "epoch": 1.4567618424282092,
      "grad_norm": 0.03837354853749275,
      "learning_rate": 6.114792158208247e-06,
      "loss": 0.0009,
      "step": 89030
    },
    {
      "epoch": 1.456925468379285,
      "grad_norm": 0.06534527242183685,
      "learning_rate": 6.113864160209647e-06,
      "loss": 0.0029,
      "step": 89040
    },
    {
      "epoch": 1.4570890943303607,
      "grad_norm": 0.16001906991004944,
      "learning_rate": 6.112936121836149e-06,
      "loss": 0.0016,
      "step": 89050
    },
    {
      "epoch": 1.4572527202814367,
      "grad_norm": 0.1633518636226654,
      "learning_rate": 6.112008043121393e-06,
      "loss": 0.0071,
      "step": 89060
    },
    {
      "epoch": 1.4574163462325125,
      "grad_norm": 0.09686394035816193,
      "learning_rate": 6.111079924099017e-06,
      "loss": 0.0015,
      "step": 89070
    },
    {
      "epoch": 1.4575799721835883,
      "grad_norm": 0.08456028252840042,
      "learning_rate": 6.110151764802663e-06,
      "loss": 0.0023,
      "step": 89080
    },
    {
      "epoch": 1.4577435981346643,
      "grad_norm": 0.12298714369535446,
      "learning_rate": 6.109223565265976e-06,
      "loss": 0.0025,
      "step": 89090
    },
    {
      "epoch": 1.45790722408574,
      "grad_norm": 0.07047034054994583,
      "learning_rate": 6.108295325522602e-06,
      "loss": 0.0019,
      "step": 89100
    },
    {
      "epoch": 1.4580708500368158,
      "grad_norm": 0.34442999958992004,
      "learning_rate": 6.107367045606186e-06,
      "loss": 0.0024,
      "step": 89110
    },
    {
      "epoch": 1.4582344759878918,
      "grad_norm": 0.09606515616178513,
      "learning_rate": 6.106438725550375e-06,
      "loss": 0.002,
      "step": 89120
    },
    {
      "epoch": 1.4583981019389676,
      "grad_norm": 0.08594342321157455,
      "learning_rate": 6.105510365388821e-06,
      "loss": 0.0015,
      "step": 89130
    },
    {
      "epoch": 1.4585617278900433,
      "grad_norm": 0.11370536684989929,
      "learning_rate": 6.104581965155172e-06,
      "loss": 0.0012,
      "step": 89140
    },
    {
      "epoch": 1.4587253538411191,
      "grad_norm": 0.04058707505464554,
      "learning_rate": 6.103653524883082e-06,
      "loss": 0.0014,
      "step": 89150
    },
    {
      "epoch": 1.4588889797921951,
      "grad_norm": 0.29345694184303284,
      "learning_rate": 6.1027250446062035e-06,
      "loss": 0.0029,
      "step": 89160
    },
    {
      "epoch": 1.459052605743271,
      "grad_norm": 0.1731075793504715,
      "learning_rate": 6.1017965243581936e-06,
      "loss": 0.0036,
      "step": 89170
    },
    {
      "epoch": 1.4592162316943467,
      "grad_norm": 0.10479287058115005,
      "learning_rate": 6.100867964172706e-06,
      "loss": 0.0017,
      "step": 89180
    },
    {
      "epoch": 1.4593798576454224,
      "grad_norm": 0.1564171463251114,
      "learning_rate": 6.099939364083403e-06,
      "loss": 0.003,
      "step": 89190
    },
    {
      "epoch": 1.4595434835964984,
      "grad_norm": 0.13803674280643463,
      "learning_rate": 6.09901072412394e-06,
      "loss": 0.0019,
      "step": 89200
    },
    {
      "epoch": 1.4597071095475742,
      "grad_norm": 0.07581519335508347,
      "learning_rate": 6.098082044327979e-06,
      "loss": 0.0014,
      "step": 89210
    },
    {
      "epoch": 1.45987073549865,
      "grad_norm": 0.048461902886629105,
      "learning_rate": 6.097153324729186e-06,
      "loss": 0.0029,
      "step": 89220
    },
    {
      "epoch": 1.460034361449726,
      "grad_norm": 0.1506853997707367,
      "learning_rate": 6.096224565361219e-06,
      "loss": 0.0014,
      "step": 89230
    },
    {
      "epoch": 1.4601979874008018,
      "grad_norm": 0.2848055064678192,
      "learning_rate": 6.095295766257749e-06,
      "loss": 0.0011,
      "step": 89240
    },
    {
      "epoch": 1.4603616133518775,
      "grad_norm": 0.26289716362953186,
      "learning_rate": 6.094366927452437e-06,
      "loss": 0.0023,
      "step": 89250
    },
    {
      "epoch": 1.4605252393029535,
      "grad_norm": 0.08146172761917114,
      "learning_rate": 6.0934380489789566e-06,
      "loss": 0.0017,
      "step": 89260
    },
    {
      "epoch": 1.4606888652540293,
      "grad_norm": 0.03863754868507385,
      "learning_rate": 6.092509130870974e-06,
      "loss": 0.0028,
      "step": 89270
    },
    {
      "epoch": 1.460852491205105,
      "grad_norm": 0.1781115084886551,
      "learning_rate": 6.091580173162161e-06,
      "loss": 0.0016,
      "step": 89280
    },
    {
      "epoch": 1.461016117156181,
      "grad_norm": 0.3425150513648987,
      "learning_rate": 6.090651175886188e-06,
      "loss": 0.0028,
      "step": 89290
    },
    {
      "epoch": 1.4611797431072568,
      "grad_norm": 0.336593359708786,
      "learning_rate": 6.089722139076735e-06,
      "loss": 0.0014,
      "step": 89300
    },
    {
      "epoch": 1.4613433690583326,
      "grad_norm": 0.1394587755203247,
      "learning_rate": 6.088793062767473e-06,
      "loss": 0.0024,
      "step": 89310
    },
    {
      "epoch": 1.4615069950094086,
      "grad_norm": 0.0684904158115387,
      "learning_rate": 6.087863946992077e-06,
      "loss": 0.0018,
      "step": 89320
    },
    {
      "epoch": 1.4616706209604844,
      "grad_norm": 0.08376423269510269,
      "learning_rate": 6.0869347917842295e-06,
      "loss": 0.001,
      "step": 89330
    },
    {
      "epoch": 1.4618342469115602,
      "grad_norm": 0.10150188207626343,
      "learning_rate": 6.086005597177608e-06,
      "loss": 0.0013,
      "step": 89340
    },
    {
      "epoch": 1.461997872862636,
      "grad_norm": 0.1953052282333374,
      "learning_rate": 6.085076363205893e-06,
      "loss": 0.0024,
      "step": 89350
    },
    {
      "epoch": 1.462161498813712,
      "grad_norm": 0.07185788452625275,
      "learning_rate": 6.084147089902768e-06,
      "loss": 0.0029,
      "step": 89360
    },
    {
      "epoch": 1.4623251247647877,
      "grad_norm": 0.07225426286458969,
      "learning_rate": 6.0832177773019165e-06,
      "loss": 0.0007,
      "step": 89370
    },
    {
      "epoch": 1.4624887507158635,
      "grad_norm": 0.12425591051578522,
      "learning_rate": 6.082288425437024e-06,
      "loss": 0.0016,
      "step": 89380
    },
    {
      "epoch": 1.4626523766669393,
      "grad_norm": 0.05503705143928528,
      "learning_rate": 6.081359034341777e-06,
      "loss": 0.0036,
      "step": 89390
    },
    {
      "epoch": 1.4628160026180153,
      "grad_norm": 0.19168508052825928,
      "learning_rate": 6.0804296040498645e-06,
      "loss": 0.0017,
      "step": 89400
    },
    {
      "epoch": 1.462979628569091,
      "grad_norm": 0.5246029496192932,
      "learning_rate": 6.079500134594977e-06,
      "loss": 0.0024,
      "step": 89410
    },
    {
      "epoch": 1.4631432545201668,
      "grad_norm": 0.09161466360092163,
      "learning_rate": 6.078570626010803e-06,
      "loss": 0.0021,
      "step": 89420
    },
    {
      "epoch": 1.4633068804712428,
      "grad_norm": 0.10353600978851318,
      "learning_rate": 6.077641078331033e-06,
      "loss": 0.0024,
      "step": 89430
    },
    {
      "epoch": 1.4634705064223186,
      "grad_norm": 0.031692489981651306,
      "learning_rate": 6.076711491589367e-06,
      "loss": 0.0018,
      "step": 89440
    },
    {
      "epoch": 1.4636341323733943,
      "grad_norm": 0.004756220616400242,
      "learning_rate": 6.075781865819498e-06,
      "loss": 0.003,
      "step": 89450
    },
    {
      "epoch": 1.4637977583244703,
      "grad_norm": 0.20435446500778198,
      "learning_rate": 6.074852201055121e-06,
      "loss": 0.0021,
      "step": 89460
    },
    {
      "epoch": 1.4639613842755461,
      "grad_norm": 0.3000190556049347,
      "learning_rate": 6.073922497329936e-06,
      "loss": 0.003,
      "step": 89470
    },
    {
      "epoch": 1.464125010226622,
      "grad_norm": 0.011369393207132816,
      "learning_rate": 6.07299275467764e-06,
      "loss": 0.003,
      "step": 89480
    },
    {
      "epoch": 1.4642886361776979,
      "grad_norm": 0.12313947081565857,
      "learning_rate": 6.072062973131936e-06,
      "loss": 0.0027,
      "step": 89490
    },
    {
      "epoch": 1.4644522621287737,
      "grad_norm": 0.07110333442687988,
      "learning_rate": 6.071133152726526e-06,
      "loss": 0.0009,
      "step": 89500
    },
    {
      "epoch": 1.4646158880798494,
      "grad_norm": 0.13085950911045074,
      "learning_rate": 6.0702032934951135e-06,
      "loss": 0.0014,
      "step": 89510
    },
    {
      "epoch": 1.4647795140309254,
      "grad_norm": 0.04204509034752846,
      "learning_rate": 6.069273395471403e-06,
      "loss": 0.0011,
      "step": 89520
    },
    {
      "epoch": 1.4649431399820012,
      "grad_norm": 0.04049510136246681,
      "learning_rate": 6.0683434586891035e-06,
      "loss": 0.0012,
      "step": 89530
    },
    {
      "epoch": 1.465106765933077,
      "grad_norm": 0.23977002501487732,
      "learning_rate": 6.067413483181919e-06,
      "loss": 0.0016,
      "step": 89540
    },
    {
      "epoch": 1.4652703918841528,
      "grad_norm": 0.11275684833526611,
      "learning_rate": 6.066483468983564e-06,
      "loss": 0.0011,
      "step": 89550
    },
    {
      "epoch": 1.4654340178352285,
      "grad_norm": 0.1879366785287857,
      "learning_rate": 6.065553416127745e-06,
      "loss": 0.0028,
      "step": 89560
    },
    {
      "epoch": 1.4655976437863045,
      "grad_norm": 0.2041204571723938,
      "learning_rate": 6.064623324648176e-06,
      "loss": 0.0014,
      "step": 89570
    },
    {
      "epoch": 1.4657612697373803,
      "grad_norm": 0.2825053632259369,
      "learning_rate": 6.063693194578569e-06,
      "loss": 0.0013,
      "step": 89580
    },
    {
      "epoch": 1.465924895688456,
      "grad_norm": 0.06186201423406601,
      "learning_rate": 6.062763025952642e-06,
      "loss": 0.0015,
      "step": 89590
    },
    {
      "epoch": 1.466088521639532,
      "grad_norm": 0.025388730689883232,
      "learning_rate": 6.061832818804109e-06,
      "loss": 0.0012,
      "step": 89600
    },
    {
      "epoch": 1.4662521475906078,
      "grad_norm": 0.03444620221853256,
      "learning_rate": 6.060902573166688e-06,
      "loss": 0.0022,
      "step": 89610
    },
    {
      "epoch": 1.4664157735416836,
      "grad_norm": 0.14855274558067322,
      "learning_rate": 6.059972289074099e-06,
      "loss": 0.0018,
      "step": 89620
    },
    {
      "epoch": 1.4665793994927596,
      "grad_norm": 0.0637306272983551,
      "learning_rate": 6.059041966560062e-06,
      "loss": 0.0024,
      "step": 89630
    },
    {
      "epoch": 1.4667430254438354,
      "grad_norm": 0.0553729347884655,
      "learning_rate": 6.058111605658299e-06,
      "loss": 0.0011,
      "step": 89640
    },
    {
      "epoch": 1.4669066513949112,
      "grad_norm": 0.04310569167137146,
      "learning_rate": 6.057181206402531e-06,
      "loss": 0.0008,
      "step": 89650
    },
    {
      "epoch": 1.4670702773459872,
      "grad_norm": 0.15886901319026947,
      "learning_rate": 6.056250768826488e-06,
      "loss": 0.0017,
      "step": 89660
    },
    {
      "epoch": 1.467233903297063,
      "grad_norm": 0.035845134407281876,
      "learning_rate": 6.0553202929638945e-06,
      "loss": 0.0018,
      "step": 89670
    },
    {
      "epoch": 1.4673975292481387,
      "grad_norm": 0.1825220137834549,
      "learning_rate": 6.054389778848474e-06,
      "loss": 0.0034,
      "step": 89680
    },
    {
      "epoch": 1.4675611551992147,
      "grad_norm": 0.09203601628541946,
      "learning_rate": 6.053459226513959e-06,
      "loss": 0.0018,
      "step": 89690
    },
    {
      "epoch": 1.4677247811502905,
      "grad_norm": 0.04955407604575157,
      "learning_rate": 6.0525286359940794e-06,
      "loss": 0.0019,
      "step": 89700
    },
    {
      "epoch": 1.4678884071013663,
      "grad_norm": 0.08668463677167892,
      "learning_rate": 6.051598007322565e-06,
      "loss": 0.0026,
      "step": 89710
    },
    {
      "epoch": 1.4680520330524423,
      "grad_norm": 0.10305390506982803,
      "learning_rate": 6.050667340533153e-06,
      "loss": 0.0018,
      "step": 89720
    },
    {
      "epoch": 1.468215659003518,
      "grad_norm": 0.07880593836307526,
      "learning_rate": 6.049736635659573e-06,
      "loss": 0.0016,
      "step": 89730
    },
    {
      "epoch": 1.4683792849545938,
      "grad_norm": 0.2380165159702301,
      "learning_rate": 6.048805892735563e-06,
      "loss": 0.0019,
      "step": 89740
    },
    {
      "epoch": 1.4685429109056696,
      "grad_norm": 0.22985823452472687,
      "learning_rate": 6.047875111794861e-06,
      "loss": 0.0019,
      "step": 89750
    },
    {
      "epoch": 1.4687065368567453,
      "grad_norm": 0.08305205404758453,
      "learning_rate": 6.046944292871204e-06,
      "loss": 0.0016,
      "step": 89760
    },
    {
      "epoch": 1.4688701628078213,
      "grad_norm": 0.12747210264205933,
      "learning_rate": 6.046013435998335e-06,
      "loss": 0.0022,
      "step": 89770
    },
    {
      "epoch": 1.4690337887588971,
      "grad_norm": 0.2798585295677185,
      "learning_rate": 6.04508254120999e-06,
      "loss": 0.0029,
      "step": 89780
    },
    {
      "epoch": 1.469197414709973,
      "grad_norm": 0.11919490993022919,
      "learning_rate": 6.044151608539914e-06,
      "loss": 0.0025,
      "step": 89790
    },
    {
      "epoch": 1.469361040661049,
      "grad_norm": 0.1528950184583664,
      "learning_rate": 6.043220638021854e-06,
      "loss": 0.0071,
      "step": 89800
    },
    {
      "epoch": 1.4695246666121247,
      "grad_norm": 0.2320098876953125,
      "learning_rate": 6.042289629689552e-06,
      "loss": 0.0018,
      "step": 89810
    },
    {
      "epoch": 1.4696882925632004,
      "grad_norm": 0.133858785033226,
      "learning_rate": 6.041358583576757e-06,
      "loss": 0.002,
      "step": 89820
    },
    {
      "epoch": 1.4698519185142764,
      "grad_norm": 0.16765472292900085,
      "learning_rate": 6.040427499717216e-06,
      "loss": 0.001,
      "step": 89830
    },
    {
      "epoch": 1.4700155444653522,
      "grad_norm": 0.08123244345188141,
      "learning_rate": 6.039496378144678e-06,
      "loss": 0.0011,
      "step": 89840
    },
    {
      "epoch": 1.470179170416428,
      "grad_norm": 0.05696301907300949,
      "learning_rate": 6.0385652188928955e-06,
      "loss": 0.0013,
      "step": 89850
    },
    {
      "epoch": 1.470342796367504,
      "grad_norm": 0.09900623559951782,
      "learning_rate": 6.0376340219956195e-06,
      "loss": 0.0029,
      "step": 89860
    },
    {
      "epoch": 1.4705064223185798,
      "grad_norm": 0.08991098403930664,
      "learning_rate": 6.036702787486604e-06,
      "loss": 0.0017,
      "step": 89870
    },
    {
      "epoch": 1.4706700482696555,
      "grad_norm": 0.28116747736930847,
      "learning_rate": 6.035771515399605e-06,
      "loss": 0.0024,
      "step": 89880
    },
    {
      "epoch": 1.4708336742207315,
      "grad_norm": 0.07289552688598633,
      "learning_rate": 6.034840205768376e-06,
      "loss": 0.0018,
      "step": 89890
    },
    {
      "epoch": 1.4709973001718073,
      "grad_norm": 0.009397052228450775,
      "learning_rate": 6.033908858626678e-06,
      "loss": 0.0024,
      "step": 89900
    },
    {
      "epoch": 1.471160926122883,
      "grad_norm": 0.18024064600467682,
      "learning_rate": 6.0329774740082695e-06,
      "loss": 0.0018,
      "step": 89910
    },
    {
      "epoch": 1.4713245520739588,
      "grad_norm": 0.15219052135944366,
      "learning_rate": 6.03204605194691e-06,
      "loss": 0.0019,
      "step": 89920
    },
    {
      "epoch": 1.4714881780250348,
      "grad_norm": 0.04910378158092499,
      "learning_rate": 6.0311145924763605e-06,
      "loss": 0.0014,
      "step": 89930
    },
    {
      "epoch": 1.4716518039761106,
      "grad_norm": 0.08571364730596542,
      "learning_rate": 6.030183095630385e-06,
      "loss": 0.0033,
      "step": 89940
    },
    {
      "epoch": 1.4718154299271864,
      "grad_norm": 0.1076994240283966,
      "learning_rate": 6.029251561442749e-06,
      "loss": 0.0026,
      "step": 89950
    },
    {
      "epoch": 1.4719790558782622,
      "grad_norm": 0.16011343896389008,
      "learning_rate": 6.028319989947217e-06,
      "loss": 0.0012,
      "step": 89960
    },
    {
      "epoch": 1.4721426818293382,
      "grad_norm": 0.20377467572689056,
      "learning_rate": 6.0273883811775595e-06,
      "loss": 0.0025,
      "step": 89970
    },
    {
      "epoch": 1.472306307780414,
      "grad_norm": 0.14934121072292328,
      "learning_rate": 6.026456735167541e-06,
      "loss": 0.0016,
      "step": 89980
    },
    {
      "epoch": 1.4724699337314897,
      "grad_norm": 0.3843739628791809,
      "learning_rate": 6.025525051950931e-06,
      "loss": 0.0017,
      "step": 89990
    },
    {
      "epoch": 1.4726335596825657,
      "grad_norm": 0.10053126513957977,
      "learning_rate": 6.0245933315615044e-06,
      "loss": 0.0013,
      "step": 90000
    },
    {
      "epoch": 1.4727971856336415,
      "grad_norm": 0.06712966412305832,
      "learning_rate": 6.02366157403303e-06,
      "loss": 0.0029,
      "step": 90010
    },
    {
      "epoch": 1.4729608115847173,
      "grad_norm": 0.15196646749973297,
      "learning_rate": 6.022729779399288e-06,
      "loss": 0.0022,
      "step": 90020
    },
    {
      "epoch": 1.4731244375357933,
      "grad_norm": 0.14024144411087036,
      "learning_rate": 6.021797947694047e-06,
      "loss": 0.003,
      "step": 90030
    },
    {
      "epoch": 1.473288063486869,
      "grad_norm": 0.0713011622428894,
      "learning_rate": 6.020866078951085e-06,
      "loss": 0.0013,
      "step": 90040
    },
    {
      "epoch": 1.4734516894379448,
      "grad_norm": 0.013995494693517685,
      "learning_rate": 6.019934173204182e-06,
      "loss": 0.0013,
      "step": 90050
    },
    {
      "epoch": 1.4736153153890208,
      "grad_norm": 0.05694727227091789,
      "learning_rate": 6.019002230487119e-06,
      "loss": 0.0025,
      "step": 90060
    },
    {
      "epoch": 1.4737789413400966,
      "grad_norm": 0.10397981852293015,
      "learning_rate": 6.018070250833672e-06,
      "loss": 0.004,
      "step": 90070
    },
    {
      "epoch": 1.4739425672911723,
      "grad_norm": 0.06639304757118225,
      "learning_rate": 6.017138234277625e-06,
      "loss": 0.0016,
      "step": 90080
    },
    {
      "epoch": 1.4741061932422483,
      "grad_norm": 0.19975242018699646,
      "learning_rate": 6.016206180852762e-06,
      "loss": 0.003,
      "step": 90090
    },
    {
      "epoch": 1.4742698191933241,
      "grad_norm": 0.17541468143463135,
      "learning_rate": 6.0152740905928685e-06,
      "loss": 0.0028,
      "step": 90100
    },
    {
      "epoch": 1.4744334451444,
      "grad_norm": 0.02349376678466797,
      "learning_rate": 6.014341963531728e-06,
      "loss": 0.0009,
      "step": 90110
    },
    {
      "epoch": 1.4745970710954757,
      "grad_norm": 0.14489562809467316,
      "learning_rate": 6.01340979970313e-06,
      "loss": 0.0025,
      "step": 90120
    },
    {
      "epoch": 1.4747606970465517,
      "grad_norm": 0.11375927180051804,
      "learning_rate": 6.012477599140863e-06,
      "loss": 0.0017,
      "step": 90130
    },
    {
      "epoch": 1.4749243229976274,
      "grad_norm": 0.10609984397888184,
      "learning_rate": 6.011545361878716e-06,
      "loss": 0.0026,
      "step": 90140
    },
    {
      "epoch": 1.4750879489487032,
      "grad_norm": 0.1411619633436203,
      "learning_rate": 6.010613087950479e-06,
      "loss": 0.0025,
      "step": 90150
    },
    {
      "epoch": 1.475251574899779,
      "grad_norm": 0.2597443163394928,
      "learning_rate": 6.009680777389948e-06,
      "loss": 0.0018,
      "step": 90160
    },
    {
      "epoch": 1.475415200850855,
      "grad_norm": 0.0626143217086792,
      "learning_rate": 6.008748430230916e-06,
      "loss": 0.0014,
      "step": 90170
    },
    {
      "epoch": 1.4755788268019308,
      "grad_norm": 0.09211093187332153,
      "learning_rate": 6.007816046507177e-06,
      "loss": 0.0017,
      "step": 90180
    },
    {
      "epoch": 1.4757424527530065,
      "grad_norm": 0.2062859684228897,
      "learning_rate": 6.006883626252529e-06,
      "loss": 0.0014,
      "step": 90190
    },
    {
      "epoch": 1.4759060787040825,
      "grad_norm": 0.023430490866303444,
      "learning_rate": 6.00595116950077e-06,
      "loss": 0.0016,
      "step": 90200
    },
    {
      "epoch": 1.4760697046551583,
      "grad_norm": 0.02907814458012581,
      "learning_rate": 6.005018676285698e-06,
      "loss": 0.0019,
      "step": 90210
    },
    {
      "epoch": 1.476233330606234,
      "grad_norm": 0.07631395757198334,
      "learning_rate": 6.004086146641115e-06,
      "loss": 0.0018,
      "step": 90220
    },
    {
      "epoch": 1.47639695655731,
      "grad_norm": 0.13297241926193237,
      "learning_rate": 6.003153580600821e-06,
      "loss": 0.0023,
      "step": 90230
    },
    {
      "epoch": 1.4765605825083858,
      "grad_norm": 0.25503307580947876,
      "learning_rate": 6.002220978198623e-06,
      "loss": 0.0029,
      "step": 90240
    },
    {
      "epoch": 1.4767242084594616,
      "grad_norm": 0.12940753996372223,
      "learning_rate": 6.001288339468322e-06,
      "loss": 0.0024,
      "step": 90250
    },
    {
      "epoch": 1.4768878344105376,
      "grad_norm": 0.09497050940990448,
      "learning_rate": 6.0003556644437244e-06,
      "loss": 0.0023,
      "step": 90260
    },
    {
      "epoch": 1.4770514603616134,
      "grad_norm": 0.04867600277066231,
      "learning_rate": 5.999422953158641e-06,
      "loss": 0.0027,
      "step": 90270
    },
    {
      "epoch": 1.4772150863126892,
      "grad_norm": 0.06594760715961456,
      "learning_rate": 5.998490205646874e-06,
      "loss": 0.0029,
      "step": 90280
    },
    {
      "epoch": 1.4773787122637652,
      "grad_norm": 0.15169459581375122,
      "learning_rate": 5.997557421942239e-06,
      "loss": 0.0018,
      "step": 90290
    },
    {
      "epoch": 1.477542338214841,
      "grad_norm": 0.20777268707752228,
      "learning_rate": 5.996624602078545e-06,
      "loss": 0.0013,
      "step": 90300
    },
    {
      "epoch": 1.4777059641659167,
      "grad_norm": 0.09152727574110031,
      "learning_rate": 5.995691746089604e-06,
      "loss": 0.0016,
      "step": 90310
    },
    {
      "epoch": 1.4778695901169925,
      "grad_norm": 0.07250243425369263,
      "learning_rate": 5.994758854009232e-06,
      "loss": 0.0013,
      "step": 90320
    },
    {
      "epoch": 1.4780332160680685,
      "grad_norm": 0.32375043630599976,
      "learning_rate": 5.99382592587124e-06,
      "loss": 0.0017,
      "step": 90330
    },
    {
      "epoch": 1.4781968420191443,
      "grad_norm": 0.1271114945411682,
      "learning_rate": 5.992892961709448e-06,
      "loss": 0.0021,
      "step": 90340
    },
    {
      "epoch": 1.47836046797022,
      "grad_norm": 0.08686201274394989,
      "learning_rate": 5.9919599615576716e-06,
      "loss": 0.002,
      "step": 90350
    },
    {
      "epoch": 1.4785240939212958,
      "grad_norm": 0.13541172444820404,
      "learning_rate": 5.991026925449732e-06,
      "loss": 0.001,
      "step": 90360
    },
    {
      "epoch": 1.4786877198723718,
      "grad_norm": 0.05715705081820488,
      "learning_rate": 5.990093853419447e-06,
      "loss": 0.0043,
      "step": 90370
    },
    {
      "epoch": 1.4788513458234476,
      "grad_norm": 0.11732552200555801,
      "learning_rate": 5.989160745500641e-06,
      "loss": 0.0017,
      "step": 90380
    },
    {
      "epoch": 1.4790149717745233,
      "grad_norm": 0.039383791387081146,
      "learning_rate": 5.988227601727133e-06,
      "loss": 0.0012,
      "step": 90390
    },
    {
      "epoch": 1.4791785977255993,
      "grad_norm": 0.10291915386915207,
      "learning_rate": 5.987294422132751e-06,
      "loss": 0.0022,
      "step": 90400
    },
    {
      "epoch": 1.4793422236766751,
      "grad_norm": 0.07636168599128723,
      "learning_rate": 5.986361206751319e-06,
      "loss": 0.0015,
      "step": 90410
    },
    {
      "epoch": 1.479505849627751,
      "grad_norm": 0.233656108379364,
      "learning_rate": 5.985427955616664e-06,
      "loss": 0.0031,
      "step": 90420
    },
    {
      "epoch": 1.4796694755788269,
      "grad_norm": 0.09308003634214401,
      "learning_rate": 5.984494668762614e-06,
      "loss": 0.0025,
      "step": 90430
    },
    {
      "epoch": 1.4798331015299027,
      "grad_norm": 0.15329548716545105,
      "learning_rate": 5.983561346222998e-06,
      "loss": 0.0016,
      "step": 90440
    },
    {
      "epoch": 1.4799967274809784,
      "grad_norm": 0.06337698549032211,
      "learning_rate": 5.982627988031647e-06,
      "loss": 0.0029,
      "step": 90450
    },
    {
      "epoch": 1.4801603534320544,
      "grad_norm": 0.15393149852752686,
      "learning_rate": 5.981694594222395e-06,
      "loss": 0.002,
      "step": 90460
    },
    {
      "epoch": 1.4803239793831302,
      "grad_norm": 0.02823197841644287,
      "learning_rate": 5.980761164829072e-06,
      "loss": 0.0012,
      "step": 90470
    },
    {
      "epoch": 1.480487605334206,
      "grad_norm": 0.1599329710006714,
      "learning_rate": 5.979827699885514e-06,
      "loss": 0.0022,
      "step": 90480
    },
    {
      "epoch": 1.480651231285282,
      "grad_norm": 0.0665646642446518,
      "learning_rate": 5.978894199425556e-06,
      "loss": 0.0036,
      "step": 90490
    },
    {
      "epoch": 1.4808148572363578,
      "grad_norm": 0.1860765665769577,
      "learning_rate": 5.977960663483037e-06,
      "loss": 0.0025,
      "step": 90500
    },
    {
      "epoch": 1.4809784831874335,
      "grad_norm": 0.0776752158999443,
      "learning_rate": 5.977027092091794e-06,
      "loss": 0.0012,
      "step": 90510
    },
    {
      "epoch": 1.4811421091385093,
      "grad_norm": 0.012308900244534016,
      "learning_rate": 5.976093485285668e-06,
      "loss": 0.0012,
      "step": 90520
    },
    {
      "epoch": 1.481305735089585,
      "grad_norm": 0.030911168083548546,
      "learning_rate": 5.975159843098499e-06,
      "loss": 0.0018,
      "step": 90530
    },
    {
      "epoch": 1.481469361040661,
      "grad_norm": 0.05663762986660004,
      "learning_rate": 5.9742261655641284e-06,
      "loss": 0.0014,
      "step": 90540
    },
    {
      "epoch": 1.4816329869917368,
      "grad_norm": 0.22436107695102692,
      "learning_rate": 5.973292452716401e-06,
      "loss": 0.0027,
      "step": 90550
    },
    {
      "epoch": 1.4817966129428126,
      "grad_norm": 0.14220155775547028,
      "learning_rate": 5.972358704589162e-06,
      "loss": 0.0012,
      "step": 90560
    },
    {
      "epoch": 1.4819602388938886,
      "grad_norm": 0.21445974707603455,
      "learning_rate": 5.971424921216257e-06,
      "loss": 0.0022,
      "step": 90570
    },
    {
      "epoch": 1.4821238648449644,
      "grad_norm": 0.08079153299331665,
      "learning_rate": 5.9704911026315325e-06,
      "loss": 0.0031,
      "step": 90580
    },
    {
      "epoch": 1.4822874907960402,
      "grad_norm": 0.1684669852256775,
      "learning_rate": 5.969557248868839e-06,
      "loss": 0.0019,
      "step": 90590
    },
    {
      "epoch": 1.4824511167471162,
      "grad_norm": 0.18776816129684448,
      "learning_rate": 5.968623359962025e-06,
      "loss": 0.0018,
      "step": 90600
    },
    {
      "epoch": 1.482614742698192,
      "grad_norm": 0.09689784795045853,
      "learning_rate": 5.967689435944942e-06,
      "loss": 0.0011,
      "step": 90610
    },
    {
      "epoch": 1.4827783686492677,
      "grad_norm": 0.07097337394952774,
      "learning_rate": 5.966755476851444e-06,
      "loss": 0.0019,
      "step": 90620
    },
    {
      "epoch": 1.4829419946003437,
      "grad_norm": 0.055122386664152145,
      "learning_rate": 5.965821482715382e-06,
      "loss": 0.002,
      "step": 90630
    },
    {
      "epoch": 1.4831056205514195,
      "grad_norm": 0.045418158173561096,
      "learning_rate": 5.964887453570613e-06,
      "loss": 0.0027,
      "step": 90640
    },
    {
      "epoch": 1.4832692465024953,
      "grad_norm": 0.03818480297923088,
      "learning_rate": 5.963953389450993e-06,
      "loss": 0.0014,
      "step": 90650
    },
    {
      "epoch": 1.4834328724535712,
      "grad_norm": 0.3111845850944519,
      "learning_rate": 5.9630192903903795e-06,
      "loss": 0.0021,
      "step": 90660
    },
    {
      "epoch": 1.483596498404647,
      "grad_norm": 0.1479066014289856,
      "learning_rate": 5.962085156422631e-06,
      "loss": 0.001,
      "step": 90670
    },
    {
      "epoch": 1.4837601243557228,
      "grad_norm": 0.05225803703069687,
      "learning_rate": 5.961150987581608e-06,
      "loss": 0.0015,
      "step": 90680
    },
    {
      "epoch": 1.4839237503067988,
      "grad_norm": 0.06672590225934982,
      "learning_rate": 5.9602167839011725e-06,
      "loss": 0.0014,
      "step": 90690
    },
    {
      "epoch": 1.4840873762578746,
      "grad_norm": 0.06295999139547348,
      "learning_rate": 5.959282545415186e-06,
      "loss": 0.0011,
      "step": 90700
    },
    {
      "epoch": 1.4842510022089503,
      "grad_norm": 0.04728035628795624,
      "learning_rate": 5.9583482721575125e-06,
      "loss": 0.0034,
      "step": 90710
    },
    {
      "epoch": 1.4844146281600261,
      "grad_norm": 0.1677972823381424,
      "learning_rate": 5.957413964162016e-06,
      "loss": 0.0012,
      "step": 90720
    },
    {
      "epoch": 1.484578254111102,
      "grad_norm": 0.19957977533340454,
      "learning_rate": 5.9564796214625665e-06,
      "loss": 0.0018,
      "step": 90730
    },
    {
      "epoch": 1.4847418800621779,
      "grad_norm": 0.08272489905357361,
      "learning_rate": 5.95554524409303e-06,
      "loss": 0.0017,
      "step": 90740
    },
    {
      "epoch": 1.4849055060132537,
      "grad_norm": 0.03452751785516739,
      "learning_rate": 5.954610832087273e-06,
      "loss": 0.0011,
      "step": 90750
    },
    {
      "epoch": 1.4850691319643294,
      "grad_norm": 0.08366342633962631,
      "learning_rate": 5.953676385479168e-06,
      "loss": 0.0019,
      "step": 90760
    },
    {
      "epoch": 1.4852327579154054,
      "grad_norm": 0.037623144686222076,
      "learning_rate": 5.952741904302588e-06,
      "loss": 0.0022,
      "step": 90770
    },
    {
      "epoch": 1.4853963838664812,
      "grad_norm": 0.1549617499113083,
      "learning_rate": 5.951807388591402e-06,
      "loss": 0.0015,
      "step": 90780
    },
    {
      "epoch": 1.485560009817557,
      "grad_norm": 0.014540654607117176,
      "learning_rate": 5.950872838379486e-06,
      "loss": 0.0033,
      "step": 90790
    },
    {
      "epoch": 1.485723635768633,
      "grad_norm": 0.1081242635846138,
      "learning_rate": 5.949938253700716e-06,
      "loss": 0.0036,
      "step": 90800
    },
    {
      "epoch": 1.4858872617197088,
      "grad_norm": 0.11421678215265274,
      "learning_rate": 5.949003634588965e-06,
      "loss": 0.0028,
      "step": 90810
    },
    {
      "epoch": 1.4860508876707845,
      "grad_norm": 0.09907367825508118,
      "learning_rate": 5.948068981078117e-06,
      "loss": 0.0024,
      "step": 90820
    },
    {
      "epoch": 1.4862145136218605,
      "grad_norm": 0.06756455451250076,
      "learning_rate": 5.9471342932020444e-06,
      "loss": 0.0011,
      "step": 90830
    },
    {
      "epoch": 1.4863781395729363,
      "grad_norm": 0.12372926622629166,
      "learning_rate": 5.946199570994631e-06,
      "loss": 0.0026,
      "step": 90840
    },
    {
      "epoch": 1.486541765524012,
      "grad_norm": 0.05814385786652565,
      "learning_rate": 5.945264814489756e-06,
      "loss": 0.0012,
      "step": 90850
    },
    {
      "epoch": 1.486705391475088,
      "grad_norm": 0.10990256071090698,
      "learning_rate": 5.944330023721303e-06,
      "loss": 0.0017,
      "step": 90860
    },
    {
      "epoch": 1.4868690174261638,
      "grad_norm": 0.11669552326202393,
      "learning_rate": 5.943395198723159e-06,
      "loss": 0.0015,
      "step": 90870
    },
    {
      "epoch": 1.4870326433772396,
      "grad_norm": 0.027970630675554276,
      "learning_rate": 5.942460339529205e-06,
      "loss": 0.0013,
      "step": 90880
    },
    {
      "epoch": 1.4871962693283154,
      "grad_norm": 0.1698177307844162,
      "learning_rate": 5.941525446173328e-06,
      "loss": 0.0016,
      "step": 90890
    },
    {
      "epoch": 1.4873598952793914,
      "grad_norm": 0.10908200591802597,
      "learning_rate": 5.940590518689417e-06,
      "loss": 0.0011,
      "step": 90900
    },
    {
      "epoch": 1.4875235212304672,
      "grad_norm": 0.07497163116931915,
      "learning_rate": 5.939655557111361e-06,
      "loss": 0.0012,
      "step": 90910
    },
    {
      "epoch": 1.487687147181543,
      "grad_norm": 0.14659839868545532,
      "learning_rate": 5.938720561473049e-06,
      "loss": 0.0026,
      "step": 90920
    },
    {
      "epoch": 1.4878507731326187,
      "grad_norm": 0.1053742915391922,
      "learning_rate": 5.937785531808373e-06,
      "loss": 0.0023,
      "step": 90930
    },
    {
      "epoch": 1.4880143990836947,
      "grad_norm": 0.19887208938598633,
      "learning_rate": 5.936850468151225e-06,
      "loss": 0.0013,
      "step": 90940
    },
    {
      "epoch": 1.4881780250347705,
      "grad_norm": 0.23559197783470154,
      "learning_rate": 5.9359153705355e-06,
      "loss": 0.0015,
      "step": 90950
    },
    {
      "epoch": 1.4883416509858463,
      "grad_norm": 0.48884499073028564,
      "learning_rate": 5.934980238995091e-06,
      "loss": 0.0018,
      "step": 90960
    },
    {
      "epoch": 1.4885052769369223,
      "grad_norm": 0.02871726267039776,
      "learning_rate": 5.9340450735638976e-06,
      "loss": 0.002,
      "step": 90970
    },
    {
      "epoch": 1.488668902887998,
      "grad_norm": 0.2396923303604126,
      "learning_rate": 5.9331098742758134e-06,
      "loss": 0.0011,
      "step": 90980
    },
    {
      "epoch": 1.4888325288390738,
      "grad_norm": 0.06204919144511223,
      "learning_rate": 5.932174641164738e-06,
      "loss": 0.0015,
      "step": 90990
    },
    {
      "epoch": 1.4889961547901498,
      "grad_norm": 0.13123126327991486,
      "learning_rate": 5.931239374264573e-06,
      "loss": 0.001,
      "step": 91000
    },
    {
      "epoch": 1.4891597807412256,
      "grad_norm": 0.14632417261600494,
      "learning_rate": 5.93030407360922e-06,
      "loss": 0.0011,
      "step": 91010
    },
    {
      "epoch": 1.4893234066923013,
      "grad_norm": 0.07743164151906967,
      "learning_rate": 5.9293687392325806e-06,
      "loss": 0.0021,
      "step": 91020
    },
    {
      "epoch": 1.4894870326433773,
      "grad_norm": 0.05706700682640076,
      "learning_rate": 5.928433371168557e-06,
      "loss": 0.0016,
      "step": 91030
    },
    {
      "epoch": 1.4896506585944531,
      "grad_norm": 0.25472885370254517,
      "learning_rate": 5.9274979694510545e-06,
      "loss": 0.0025,
      "step": 91040
    },
    {
      "epoch": 1.4898142845455289,
      "grad_norm": 0.17433689534664154,
      "learning_rate": 5.926562534113981e-06,
      "loss": 0.0012,
      "step": 91050
    },
    {
      "epoch": 1.4899779104966049,
      "grad_norm": 0.059186335653066635,
      "learning_rate": 5.925627065191242e-06,
      "loss": 0.0016,
      "step": 91060
    },
    {
      "epoch": 1.4901415364476807,
      "grad_norm": 0.10192203521728516,
      "learning_rate": 5.9246915627167465e-06,
      "loss": 0.0019,
      "step": 91070
    },
    {
      "epoch": 1.4903051623987564,
      "grad_norm": 0.06516320258378983,
      "learning_rate": 5.923756026724406e-06,
      "loss": 0.0011,
      "step": 91080
    },
    {
      "epoch": 1.4904687883498322,
      "grad_norm": 0.19599901139736176,
      "learning_rate": 5.9228204572481276e-06,
      "loss": 0.0023,
      "step": 91090
    },
    {
      "epoch": 1.4906324143009082,
      "grad_norm": 0.1105550080537796,
      "learning_rate": 5.921884854321825e-06,
      "loss": 0.0008,
      "step": 91100
    },
    {
      "epoch": 1.490796040251984,
      "grad_norm": 0.07540274411439896,
      "learning_rate": 5.920949217979414e-06,
      "loss": 0.0014,
      "step": 91110
    },
    {
      "epoch": 1.4909596662030598,
      "grad_norm": 0.06494259834289551,
      "learning_rate": 5.9200135482548075e-06,
      "loss": 0.001,
      "step": 91120
    },
    {
      "epoch": 1.4911232921541355,
      "grad_norm": 0.2632906138896942,
      "learning_rate": 5.91907784518192e-06,
      "loss": 0.0023,
      "step": 91130
    },
    {
      "epoch": 1.4912869181052115,
      "grad_norm": 0.24756161868572235,
      "learning_rate": 5.9181421087946685e-06,
      "loss": 0.003,
      "step": 91140
    },
    {
      "epoch": 1.4914505440562873,
      "grad_norm": 0.012240666896104813,
      "learning_rate": 5.917206339126973e-06,
      "loss": 0.0038,
      "step": 91150
    },
    {
      "epoch": 1.491614170007363,
      "grad_norm": 0.22262974083423615,
      "learning_rate": 5.916270536212753e-06,
      "loss": 0.0027,
      "step": 91160
    },
    {
      "epoch": 1.491777795958439,
      "grad_norm": 0.49610939621925354,
      "learning_rate": 5.915334700085928e-06,
      "loss": 0.0039,
      "step": 91170
    },
    {
      "epoch": 1.4919414219095148,
      "grad_norm": 0.0776975154876709,
      "learning_rate": 5.914398830780421e-06,
      "loss": 0.0019,
      "step": 91180
    },
    {
      "epoch": 1.4921050478605906,
      "grad_norm": 0.07331117242574692,
      "learning_rate": 5.913462928330153e-06,
      "loss": 0.0017,
      "step": 91190
    },
    {
      "epoch": 1.4922686738116666,
      "grad_norm": 0.1427258998155594,
      "learning_rate": 5.9125269927690485e-06,
      "loss": 0.0016,
      "step": 91200
    },
    {
      "epoch": 1.4924322997627424,
      "grad_norm": 0.056492023169994354,
      "learning_rate": 5.911591024131035e-06,
      "loss": 0.0022,
      "step": 91210
    },
    {
      "epoch": 1.4925959257138182,
      "grad_norm": 0.053930215537548065,
      "learning_rate": 5.910655022450037e-06,
      "loss": 0.0013,
      "step": 91220
    },
    {
      "epoch": 1.4927595516648942,
      "grad_norm": 0.061126112937927246,
      "learning_rate": 5.909718987759983e-06,
      "loss": 0.0021,
      "step": 91230
    },
    {
      "epoch": 1.49292317761597,
      "grad_norm": 0.05530744045972824,
      "learning_rate": 5.908782920094803e-06,
      "loss": 0.0018,
      "step": 91240
    },
    {
      "epoch": 1.4930868035670457,
      "grad_norm": 0.08054271340370178,
      "learning_rate": 5.907846819488426e-06,
      "loss": 0.0013,
      "step": 91250
    },
    {
      "epoch": 1.4932504295181217,
      "grad_norm": 0.16302962601184845,
      "learning_rate": 5.906910685974784e-06,
      "loss": 0.0017,
      "step": 91260
    },
    {
      "epoch": 1.4934140554691975,
      "grad_norm": 0.344541072845459,
      "learning_rate": 5.905974519587811e-06,
      "loss": 0.0026,
      "step": 91270
    },
    {
      "epoch": 1.4935776814202733,
      "grad_norm": 0.027134334668517113,
      "learning_rate": 5.905038320361436e-06,
      "loss": 0.0021,
      "step": 91280
    },
    {
      "epoch": 1.493741307371349,
      "grad_norm": 0.04441569745540619,
      "learning_rate": 5.904102088329597e-06,
      "loss": 0.002,
      "step": 91290
    },
    {
      "epoch": 1.4939049333224248,
      "grad_norm": 0.21158558130264282,
      "learning_rate": 5.903165823526233e-06,
      "loss": 0.0017,
      "step": 91300
    },
    {
      "epoch": 1.4940685592735008,
      "grad_norm": 0.16600491106510162,
      "learning_rate": 5.902229525985276e-06,
      "loss": 0.0018,
      "step": 91310
    },
    {
      "epoch": 1.4942321852245766,
      "grad_norm": 0.09154591709375381,
      "learning_rate": 5.901293195740669e-06,
      "loss": 0.0006,
      "step": 91320
    },
    {
      "epoch": 1.4943958111756523,
      "grad_norm": 0.12867434322834015,
      "learning_rate": 5.90035683282635e-06,
      "loss": 0.0019,
      "step": 91330
    },
    {
      "epoch": 1.4945594371267283,
      "grad_norm": 0.1908167153596878,
      "learning_rate": 5.899420437276259e-06,
      "loss": 0.0033,
      "step": 91340
    },
    {
      "epoch": 1.4947230630778041,
      "grad_norm": 0.09817846119403839,
      "learning_rate": 5.898484009124338e-06,
      "loss": 0.0011,
      "step": 91350
    },
    {
      "epoch": 1.49488668902888,
      "grad_norm": 0.12647955119609833,
      "learning_rate": 5.897547548404531e-06,
      "loss": 0.0009,
      "step": 91360
    },
    {
      "epoch": 1.4950503149799559,
      "grad_norm": 0.46595045924186707,
      "learning_rate": 5.896611055150784e-06,
      "loss": 0.0016,
      "step": 91370
    },
    {
      "epoch": 1.4952139409310317,
      "grad_norm": 0.1935940831899643,
      "learning_rate": 5.895674529397041e-06,
      "loss": 0.0018,
      "step": 91380
    },
    {
      "epoch": 1.4953775668821074,
      "grad_norm": 0.0651322677731514,
      "learning_rate": 5.894737971177249e-06,
      "loss": 0.0016,
      "step": 91390
    },
    {
      "epoch": 1.4955411928331834,
      "grad_norm": 0.030681701377034187,
      "learning_rate": 5.893801380525355e-06,
      "loss": 0.0017,
      "step": 91400
    },
    {
      "epoch": 1.4957048187842592,
      "grad_norm": 0.07809016108512878,
      "learning_rate": 5.89286475747531e-06,
      "loss": 0.0019,
      "step": 91410
    },
    {
      "epoch": 1.495868444735335,
      "grad_norm": 0.10390011966228485,
      "learning_rate": 5.891928102061063e-06,
      "loss": 0.0013,
      "step": 91420
    },
    {
      "epoch": 1.496032070686411,
      "grad_norm": 0.08651623129844666,
      "learning_rate": 5.8909914143165654e-06,
      "loss": 0.0016,
      "step": 91430
    },
    {
      "epoch": 1.4961956966374867,
      "grad_norm": 0.3666991591453552,
      "learning_rate": 5.890054694275772e-06,
      "loss": 0.003,
      "step": 91440
    },
    {
      "epoch": 1.4963593225885625,
      "grad_norm": 0.11857790499925613,
      "learning_rate": 5.889117941972634e-06,
      "loss": 0.0013,
      "step": 91450
    },
    {
      "epoch": 1.4965229485396385,
      "grad_norm": 0.04910532012581825,
      "learning_rate": 5.8881811574411085e-06,
      "loss": 0.002,
      "step": 91460
    },
    {
      "epoch": 1.4966865744907143,
      "grad_norm": 0.14415289461612701,
      "learning_rate": 5.887244340715151e-06,
      "loss": 0.0016,
      "step": 91470
    },
    {
      "epoch": 1.49685020044179,
      "grad_norm": 0.04828951507806778,
      "learning_rate": 5.886307491828718e-06,
      "loss": 0.0011,
      "step": 91480
    },
    {
      "epoch": 1.4970138263928658,
      "grad_norm": 0.14078818261623383,
      "learning_rate": 5.885370610815767e-06,
      "loss": 0.0022,
      "step": 91490
    },
    {
      "epoch": 1.4971774523439416,
      "grad_norm": 0.2517523169517517,
      "learning_rate": 5.884433697710261e-06,
      "loss": 0.0032,
      "step": 91500
    },
    {
      "epoch": 1.4973410782950176,
      "grad_norm": 0.07893718779087067,
      "learning_rate": 5.883496752546158e-06,
      "loss": 0.0019,
      "step": 91510
    },
    {
      "epoch": 1.4975047042460934,
      "grad_norm": 0.13495363295078278,
      "learning_rate": 5.882559775357424e-06,
      "loss": 0.0028,
      "step": 91520
    },
    {
      "epoch": 1.4976683301971692,
      "grad_norm": 0.12683121860027313,
      "learning_rate": 5.881622766178017e-06,
      "loss": 0.002,
      "step": 91530
    },
    {
      "epoch": 1.4978319561482452,
      "grad_norm": 0.011476846411824226,
      "learning_rate": 5.880685725041904e-06,
      "loss": 0.002,
      "step": 91540
    },
    {
      "epoch": 1.497995582099321,
      "grad_norm": 0.11071216315031052,
      "learning_rate": 5.8797486519830505e-06,
      "loss": 0.0024,
      "step": 91550
    },
    {
      "epoch": 1.4981592080503967,
      "grad_norm": 0.12032093852758408,
      "learning_rate": 5.878811547035421e-06,
      "loss": 0.0016,
      "step": 91560
    },
    {
      "epoch": 1.4983228340014727,
      "grad_norm": 0.16129246354103088,
      "learning_rate": 5.877874410232987e-06,
      "loss": 0.002,
      "step": 91570
    },
    {
      "epoch": 1.4984864599525485,
      "grad_norm": 0.06416504085063934,
      "learning_rate": 5.876937241609715e-06,
      "loss": 0.0018,
      "step": 91580
    },
    {
      "epoch": 1.4986500859036243,
      "grad_norm": 0.30934450030326843,
      "learning_rate": 5.876000041199575e-06,
      "loss": 0.0028,
      "step": 91590
    },
    {
      "epoch": 1.4988137118547002,
      "grad_norm": 0.06825904548168182,
      "learning_rate": 5.87506280903654e-06,
      "loss": 0.0012,
      "step": 91600
    },
    {
      "epoch": 1.498977337805776,
      "grad_norm": 0.08883596956729889,
      "learning_rate": 5.874125545154581e-06,
      "loss": 0.0022,
      "step": 91610
    },
    {
      "epoch": 1.4991409637568518,
      "grad_norm": 0.11655041575431824,
      "learning_rate": 5.873188249587673e-06,
      "loss": 0.0014,
      "step": 91620
    },
    {
      "epoch": 1.4993045897079278,
      "grad_norm": 0.10432343184947968,
      "learning_rate": 5.872250922369788e-06,
      "loss": 0.0012,
      "step": 91630
    },
    {
      "epoch": 1.4994682156590036,
      "grad_norm": 0.12425997853279114,
      "learning_rate": 5.871313563534903e-06,
      "loss": 0.0021,
      "step": 91640
    },
    {
      "epoch": 1.4996318416100793,
      "grad_norm": 0.23453903198242188,
      "learning_rate": 5.870376173116996e-06,
      "loss": 0.0013,
      "step": 91650
    },
    {
      "epoch": 1.4997954675611553,
      "grad_norm": 0.08189664036035538,
      "learning_rate": 5.869438751150045e-06,
      "loss": 0.0018,
      "step": 91660
    },
    {
      "epoch": 1.4999590935122311,
      "grad_norm": 0.153335303068161,
      "learning_rate": 5.86850129766803e-06,
      "loss": 0.0016,
      "step": 91670
    },
    {
      "epoch": 1.5001227194633069,
      "grad_norm": 0.14467045664787292,
      "learning_rate": 5.86756381270493e-06,
      "loss": 0.0021,
      "step": 91680
    },
    {
      "epoch": 1.5002863454143829,
      "grad_norm": 0.11179284006357193,
      "learning_rate": 5.8666262962947255e-06,
      "loss": 0.002,
      "step": 91690
    },
    {
      "epoch": 1.5004499713654584,
      "grad_norm": 0.027374764904379845,
      "learning_rate": 5.865688748471401e-06,
      "loss": 0.0014,
      "step": 91700
    },
    {
      "epoch": 1.5006135973165344,
      "grad_norm": 0.02600018121302128,
      "learning_rate": 5.864751169268939e-06,
      "loss": 0.0018,
      "step": 91710
    },
    {
      "epoch": 1.5007772232676102,
      "grad_norm": 0.019345024600625038,
      "learning_rate": 5.863813558721327e-06,
      "loss": 0.0013,
      "step": 91720
    },
    {
      "epoch": 1.500940849218686,
      "grad_norm": 0.1246350035071373,
      "learning_rate": 5.86287591686255e-06,
      "loss": 0.0016,
      "step": 91730
    },
    {
      "epoch": 1.501104475169762,
      "grad_norm": 0.11646867543458939,
      "learning_rate": 5.861938243726595e-06,
      "loss": 0.0012,
      "step": 91740
    },
    {
      "epoch": 1.5012681011208378,
      "grad_norm": 0.09142351895570755,
      "learning_rate": 5.861000539347449e-06,
      "loss": 0.0023,
      "step": 91750
    },
    {
      "epoch": 1.5014317270719135,
      "grad_norm": 0.15168513357639313,
      "learning_rate": 5.860062803759103e-06,
      "loss": 0.0023,
      "step": 91760
    },
    {
      "epoch": 1.5015953530229895,
      "grad_norm": 0.04419459402561188,
      "learning_rate": 5.8591250369955484e-06,
      "loss": 0.0042,
      "step": 91770
    },
    {
      "epoch": 1.5017589789740653,
      "grad_norm": 0.11802008748054504,
      "learning_rate": 5.858187239090776e-06,
      "loss": 0.0015,
      "step": 91780
    },
    {
      "epoch": 1.501922604925141,
      "grad_norm": 0.13296104967594147,
      "learning_rate": 5.857249410078778e-06,
      "loss": 0.0029,
      "step": 91790
    },
    {
      "epoch": 1.502086230876217,
      "grad_norm": 0.07423113286495209,
      "learning_rate": 5.85631154999355e-06,
      "loss": 0.0009,
      "step": 91800
    },
    {
      "epoch": 1.5022498568272928,
      "grad_norm": 0.21428649127483368,
      "learning_rate": 5.8553736588690844e-06,
      "loss": 0.0013,
      "step": 91810
    },
    {
      "epoch": 1.5024134827783686,
      "grad_norm": 0.12769107520580292,
      "learning_rate": 5.854435736739381e-06,
      "loss": 0.0034,
      "step": 91820
    },
    {
      "epoch": 1.5025771087294446,
      "grad_norm": 0.08731834590435028,
      "learning_rate": 5.853497783638438e-06,
      "loss": 0.0017,
      "step": 91830
    },
    {
      "epoch": 1.5027407346805204,
      "grad_norm": 0.15267683565616608,
      "learning_rate": 5.852559799600248e-06,
      "loss": 0.0024,
      "step": 91840
    },
    {
      "epoch": 1.5029043606315962,
      "grad_norm": 0.06569907814264297,
      "learning_rate": 5.851621784658815e-06,
      "loss": 0.0019,
      "step": 91850
    },
    {
      "epoch": 1.5030679865826722,
      "grad_norm": 0.01709694229066372,
      "learning_rate": 5.85068373884814e-06,
      "loss": 0.0015,
      "step": 91860
    },
    {
      "epoch": 1.5032316125337477,
      "grad_norm": 0.048211030662059784,
      "learning_rate": 5.849745662202224e-06,
      "loss": 0.0014,
      "step": 91870
    },
    {
      "epoch": 1.5033952384848237,
      "grad_norm": 0.10423155128955841,
      "learning_rate": 5.848807554755071e-06,
      "loss": 0.0018,
      "step": 91880
    },
    {
      "epoch": 1.5035588644358997,
      "grad_norm": 0.09340240806341171,
      "learning_rate": 5.8478694165406835e-06,
      "loss": 0.0011,
      "step": 91890
    },
    {
      "epoch": 1.5037224903869753,
      "grad_norm": 0.08209457248449326,
      "learning_rate": 5.8469312475930665e-06,
      "loss": 0.0019,
      "step": 91900
    },
    {
      "epoch": 1.5038861163380512,
      "grad_norm": 0.07592350989580154,
      "learning_rate": 5.845993047946228e-06,
      "loss": 0.0013,
      "step": 91910
    },
    {
      "epoch": 1.504049742289127,
      "grad_norm": 0.14570800960063934,
      "learning_rate": 5.8450548176341745e-06,
      "loss": 0.0025,
      "step": 91920
    },
    {
      "epoch": 1.5042133682402028,
      "grad_norm": 0.05754445493221283,
      "learning_rate": 5.8441165566909155e-06,
      "loss": 0.0017,
      "step": 91930
    },
    {
      "epoch": 1.5043769941912788,
      "grad_norm": 0.006399228237569332,
      "learning_rate": 5.84317826515046e-06,
      "loss": 0.0017,
      "step": 91940
    },
    {
      "epoch": 1.5045406201423546,
      "grad_norm": 0.11717326194047928,
      "learning_rate": 5.842239943046818e-06,
      "loss": 0.0018,
      "step": 91950
    },
    {
      "epoch": 1.5047042460934303,
      "grad_norm": 0.066010482609272,
      "learning_rate": 5.841301590414004e-06,
      "loss": 0.0011,
      "step": 91960
    },
    {
      "epoch": 1.5048678720445063,
      "grad_norm": 0.04655023664236069,
      "learning_rate": 5.840363207286027e-06,
      "loss": 0.0013,
      "step": 91970
    },
    {
      "epoch": 1.5050314979955821,
      "grad_norm": 0.25716346502304077,
      "learning_rate": 5.8394247936969065e-06,
      "loss": 0.0015,
      "step": 91980
    },
    {
      "epoch": 1.5051951239466579,
      "grad_norm": 0.1548953652381897,
      "learning_rate": 5.838486349680652e-06,
      "loss": 0.0019,
      "step": 91990
    },
    {
      "epoch": 1.5053587498977339,
      "grad_norm": 0.11267152428627014,
      "learning_rate": 5.837547875271283e-06,
      "loss": 0.0028,
      "step": 92000
    },
    {
      "epoch": 1.5055223758488097,
      "grad_norm": 0.13756878674030304,
      "learning_rate": 5.8366093705028174e-06,
      "loss": 0.0016,
      "step": 92010
    },
    {
      "epoch": 1.5056860017998854,
      "grad_norm": 0.025240028277039528,
      "learning_rate": 5.835670835409273e-06,
      "loss": 0.0012,
      "step": 92020
    },
    {
      "epoch": 1.5058496277509614,
      "grad_norm": 0.15717455744743347,
      "learning_rate": 5.8347322700246704e-06,
      "loss": 0.0016,
      "step": 92030
    },
    {
      "epoch": 1.506013253702037,
      "grad_norm": 0.04991165176033974,
      "learning_rate": 5.833793674383027e-06,
      "loss": 0.0033,
      "step": 92040
    },
    {
      "epoch": 1.506176879653113,
      "grad_norm": 0.02723216451704502,
      "learning_rate": 5.832855048518367e-06,
      "loss": 0.0011,
      "step": 92050
    },
    {
      "epoch": 1.506340505604189,
      "grad_norm": 0.12921293079853058,
      "learning_rate": 5.831916392464714e-06,
      "loss": 0.0017,
      "step": 92060
    },
    {
      "epoch": 1.5065041315552645,
      "grad_norm": 0.01611558347940445,
      "learning_rate": 5.83097770625609e-06,
      "loss": 0.0012,
      "step": 92070
    },
    {
      "epoch": 1.5066677575063405,
      "grad_norm": 0.04227316752076149,
      "learning_rate": 5.830038989926523e-06,
      "loss": 0.0017,
      "step": 92080
    },
    {
      "epoch": 1.5068313834574163,
      "grad_norm": 0.35677310824394226,
      "learning_rate": 5.8291002435100374e-06,
      "loss": 0.0035,
      "step": 92090
    },
    {
      "epoch": 1.506995009408492,
      "grad_norm": 0.05789892002940178,
      "learning_rate": 5.828161467040661e-06,
      "loss": 0.0028,
      "step": 92100
    },
    {
      "epoch": 1.507158635359568,
      "grad_norm": 0.0786251351237297,
      "learning_rate": 5.827222660552421e-06,
      "loss": 0.0017,
      "step": 92110
    },
    {
      "epoch": 1.5073222613106438,
      "grad_norm": 0.06178510561585426,
      "learning_rate": 5.826283824079348e-06,
      "loss": 0.0012,
      "step": 92120
    },
    {
      "epoch": 1.5074858872617196,
      "grad_norm": 0.11979813128709793,
      "learning_rate": 5.825344957655471e-06,
      "loss": 0.0019,
      "step": 92130
    },
    {
      "epoch": 1.5076495132127956,
      "grad_norm": 0.046463653445243835,
      "learning_rate": 5.824406061314825e-06,
      "loss": 0.0016,
      "step": 92140
    },
    {
      "epoch": 1.5078131391638714,
      "grad_norm": 0.132389634847641,
      "learning_rate": 5.823467135091438e-06,
      "loss": 0.0021,
      "step": 92150
    },
    {
      "epoch": 1.5079767651149472,
      "grad_norm": 0.08071785420179367,
      "learning_rate": 5.822528179019349e-06,
      "loss": 0.0014,
      "step": 92160
    },
    {
      "epoch": 1.5081403910660232,
      "grad_norm": 0.014393768273293972,
      "learning_rate": 5.82158919313259e-06,
      "loss": 0.0015,
      "step": 92170
    },
    {
      "epoch": 1.508304017017099,
      "grad_norm": 0.06573475897312164,
      "learning_rate": 5.820650177465198e-06,
      "loss": 0.002,
      "step": 92180
    },
    {
      "epoch": 1.5084676429681747,
      "grad_norm": 0.27615317702293396,
      "learning_rate": 5.819711132051208e-06,
      "loss": 0.0013,
      "step": 92190
    },
    {
      "epoch": 1.5086312689192507,
      "grad_norm": 0.012685788795351982,
      "learning_rate": 5.81877205692466e-06,
      "loss": 0.0015,
      "step": 92200
    },
    {
      "epoch": 1.5087948948703265,
      "grad_norm": 0.08624280989170074,
      "learning_rate": 5.817832952119592e-06,
      "loss": 0.0017,
      "step": 92210
    },
    {
      "epoch": 1.5089585208214022,
      "grad_norm": 0.2544707953929901,
      "learning_rate": 5.816893817670046e-06,
      "loss": 0.0022,
      "step": 92220
    },
    {
      "epoch": 1.5091221467724782,
      "grad_norm": 0.15449848771095276,
      "learning_rate": 5.815954653610063e-06,
      "loss": 0.002,
      "step": 92230
    },
    {
      "epoch": 1.5092857727235538,
      "grad_norm": 0.07188665121793747,
      "learning_rate": 5.815015459973685e-06,
      "loss": 0.0011,
      "step": 92240
    },
    {
      "epoch": 1.5094493986746298,
      "grad_norm": 0.08724378794431686,
      "learning_rate": 5.8140762367949545e-06,
      "loss": 0.0037,
      "step": 92250
    },
    {
      "epoch": 1.5096130246257058,
      "grad_norm": 0.1033218577504158,
      "learning_rate": 5.813136984107917e-06,
      "loss": 0.0022,
      "step": 92260
    },
    {
      "epoch": 1.5097766505767813,
      "grad_norm": 0.20820222795009613,
      "learning_rate": 5.812197701946618e-06,
      "loss": 0.0039,
      "step": 92270
    },
    {
      "epoch": 1.5099402765278573,
      "grad_norm": 0.12309335917234421,
      "learning_rate": 5.811258390345105e-06,
      "loss": 0.0023,
      "step": 92280
    },
    {
      "epoch": 1.5101039024789331,
      "grad_norm": 0.11384641379117966,
      "learning_rate": 5.810319049337425e-06,
      "loss": 0.0017,
      "step": 92290
    },
    {
      "epoch": 1.5102675284300089,
      "grad_norm": 0.0959995836019516,
      "learning_rate": 5.8093796789576275e-06,
      "loss": 0.0011,
      "step": 92300
    },
    {
      "epoch": 1.5104311543810849,
      "grad_norm": 0.2018422782421112,
      "learning_rate": 5.8084402792397606e-06,
      "loss": 0.0026,
      "step": 92310
    },
    {
      "epoch": 1.5105947803321607,
      "grad_norm": 0.05143508315086365,
      "learning_rate": 5.807500850217877e-06,
      "loss": 0.003,
      "step": 92320
    },
    {
      "epoch": 1.5107584062832364,
      "grad_norm": 0.02562854066491127,
      "learning_rate": 5.80656139192603e-06,
      "loss": 0.0012,
      "step": 92330
    },
    {
      "epoch": 1.5109220322343124,
      "grad_norm": 0.3012652099132538,
      "learning_rate": 5.80562190439827e-06,
      "loss": 0.0019,
      "step": 92340
    },
    {
      "epoch": 1.5110856581853882,
      "grad_norm": 0.1167297214269638,
      "learning_rate": 5.804682387668651e-06,
      "loss": 0.0022,
      "step": 92350
    },
    {
      "epoch": 1.511249284136464,
      "grad_norm": 0.07637261599302292,
      "learning_rate": 5.80374284177123e-06,
      "loss": 0.0013,
      "step": 92360
    },
    {
      "epoch": 1.51141291008754,
      "grad_norm": 0.06420022994279861,
      "learning_rate": 5.802803266740063e-06,
      "loss": 0.0014,
      "step": 92370
    },
    {
      "epoch": 1.5115765360386157,
      "grad_norm": 0.10261160135269165,
      "learning_rate": 5.801863662609208e-06,
      "loss": 0.0018,
      "step": 92380
    },
    {
      "epoch": 1.5117401619896915,
      "grad_norm": 0.08156854659318924,
      "learning_rate": 5.800924029412722e-06,
      "loss": 0.0014,
      "step": 92390
    },
    {
      "epoch": 1.5119037879407675,
      "grad_norm": 0.11222469806671143,
      "learning_rate": 5.799984367184666e-06,
      "loss": 0.0017,
      "step": 92400
    },
    {
      "epoch": 1.5120674138918433,
      "grad_norm": 0.11377032846212387,
      "learning_rate": 5.7990446759590965e-06,
      "loss": 0.0025,
      "step": 92410
    },
    {
      "epoch": 1.512231039842919,
      "grad_norm": 0.03881581500172615,
      "learning_rate": 5.79810495577008e-06,
      "loss": 0.001,
      "step": 92420
    },
    {
      "epoch": 1.512394665793995,
      "grad_norm": 0.0993439257144928,
      "learning_rate": 5.797165206651677e-06,
      "loss": 0.0014,
      "step": 92430
    },
    {
      "epoch": 1.5125582917450706,
      "grad_norm": 0.07430081814527512,
      "learning_rate": 5.7962254286379505e-06,
      "loss": 0.0034,
      "step": 92440
    },
    {
      "epoch": 1.5127219176961466,
      "grad_norm": 0.15329165756702423,
      "learning_rate": 5.795285621762966e-06,
      "loss": 0.0015,
      "step": 92450
    },
    {
      "epoch": 1.5128855436472226,
      "grad_norm": 0.10792247205972672,
      "learning_rate": 5.794345786060789e-06,
      "loss": 0.0012,
      "step": 92460
    },
    {
      "epoch": 1.5130491695982982,
      "grad_norm": 0.11069736629724503,
      "learning_rate": 5.793405921565487e-06,
      "loss": 0.002,
      "step": 92470
    },
    {
      "epoch": 1.5132127955493742,
      "grad_norm": 0.1862097531557083,
      "learning_rate": 5.792466028311128e-06,
      "loss": 0.0017,
      "step": 92480
    },
    {
      "epoch": 1.51337642150045,
      "grad_norm": 0.04604573920369148,
      "learning_rate": 5.791526106331778e-06,
      "loss": 0.0019,
      "step": 92490
    },
    {
      "epoch": 1.5135400474515257,
      "grad_norm": 0.07079262286424637,
      "learning_rate": 5.79058615566151e-06,
      "loss": 0.0032,
      "step": 92500
    },
    {
      "epoch": 1.5137036734026017,
      "grad_norm": 0.0942351296544075,
      "learning_rate": 5.789646176334394e-06,
      "loss": 0.0017,
      "step": 92510
    },
    {
      "epoch": 1.5138672993536775,
      "grad_norm": 0.10990624874830246,
      "learning_rate": 5.788706168384502e-06,
      "loss": 0.0025,
      "step": 92520
    },
    {
      "epoch": 1.5140309253047533,
      "grad_norm": 0.09396950900554657,
      "learning_rate": 5.787766131845908e-06,
      "loss": 0.0009,
      "step": 92530
    },
    {
      "epoch": 1.5141945512558292,
      "grad_norm": 0.07850679755210876,
      "learning_rate": 5.786826066752683e-06,
      "loss": 0.0077,
      "step": 92540
    },
    {
      "epoch": 1.514358177206905,
      "grad_norm": 0.13065828382968903,
      "learning_rate": 5.785885973138906e-06,
      "loss": 0.0013,
      "step": 92550
    },
    {
      "epoch": 1.5145218031579808,
      "grad_norm": 0.05356348678469658,
      "learning_rate": 5.78494585103865e-06,
      "loss": 0.0134,
      "step": 92560
    },
    {
      "epoch": 1.5146854291090568,
      "grad_norm": 0.11250830441713333,
      "learning_rate": 5.784005700485993e-06,
      "loss": 0.0022,
      "step": 92570
    },
    {
      "epoch": 1.5148490550601326,
      "grad_norm": 0.11656971275806427,
      "learning_rate": 5.783065521515016e-06,
      "loss": 0.0035,
      "step": 92580
    },
    {
      "epoch": 1.5150126810112083,
      "grad_norm": 0.09869369119405746,
      "learning_rate": 5.782125314159795e-06,
      "loss": 0.0016,
      "step": 92590
    },
    {
      "epoch": 1.5151763069622843,
      "grad_norm": 0.053420618176460266,
      "learning_rate": 5.7811850784544096e-06,
      "loss": 0.0011,
      "step": 92600
    },
    {
      "epoch": 1.51533993291336,
      "grad_norm": 0.06559205055236816,
      "learning_rate": 5.780244814432945e-06,
      "loss": 0.0023,
      "step": 92610
    },
    {
      "epoch": 1.5155035588644359,
      "grad_norm": 0.09503054618835449,
      "learning_rate": 5.779304522129478e-06,
      "loss": 0.002,
      "step": 92620
    },
    {
      "epoch": 1.5156671848155119,
      "grad_norm": 0.0468287393450737,
      "learning_rate": 5.778364201578098e-06,
      "loss": 0.0011,
      "step": 92630
    },
    {
      "epoch": 1.5158308107665874,
      "grad_norm": 0.240324467420578,
      "learning_rate": 5.777423852812885e-06,
      "loss": 0.002,
      "step": 92640
    },
    {
      "epoch": 1.5159944367176634,
      "grad_norm": 0.0628838762640953,
      "learning_rate": 5.776483475867927e-06,
      "loss": 0.0018,
      "step": 92650
    },
    {
      "epoch": 1.5161580626687394,
      "grad_norm": 0.11935097724199295,
      "learning_rate": 5.775543070777308e-06,
      "loss": 0.0022,
      "step": 92660
    },
    {
      "epoch": 1.516321688619815,
      "grad_norm": 0.2892530560493469,
      "learning_rate": 5.7746026375751175e-06,
      "loss": 0.0029,
      "step": 92670
    },
    {
      "epoch": 1.516485314570891,
      "grad_norm": 0.07304031401872635,
      "learning_rate": 5.773662176295444e-06,
      "loss": 0.0011,
      "step": 92680
    },
    {
      "epoch": 1.5166489405219667,
      "grad_norm": 0.12155020236968994,
      "learning_rate": 5.772721686972374e-06,
      "loss": 0.0017,
      "step": 92690
    },
    {
      "epoch": 1.5168125664730425,
      "grad_norm": 0.06272289156913757,
      "learning_rate": 5.7717811696400004e-06,
      "loss": 0.0012,
      "step": 92700
    },
    {
      "epoch": 1.5169761924241185,
      "grad_norm": 0.2657904624938965,
      "learning_rate": 5.7708406243324145e-06,
      "loss": 0.0029,
      "step": 92710
    },
    {
      "epoch": 1.5171398183751943,
      "grad_norm": 0.08259091526269913,
      "learning_rate": 5.769900051083709e-06,
      "loss": 0.0014,
      "step": 92720
    },
    {
      "epoch": 1.51730344432627,
      "grad_norm": 0.3218785226345062,
      "learning_rate": 5.768959449927978e-06,
      "loss": 0.0037,
      "step": 92730
    },
    {
      "epoch": 1.517467070277346,
      "grad_norm": 0.17818595468997955,
      "learning_rate": 5.768018820899314e-06,
      "loss": 0.0017,
      "step": 92740
    },
    {
      "epoch": 1.5176306962284218,
      "grad_norm": 0.07660868018865585,
      "learning_rate": 5.767078164031813e-06,
      "loss": 0.0013,
      "step": 92750
    },
    {
      "epoch": 1.5177943221794976,
      "grad_norm": 0.023576997220516205,
      "learning_rate": 5.766137479359573e-06,
      "loss": 0.0027,
      "step": 92760
    },
    {
      "epoch": 1.5179579481305736,
      "grad_norm": 0.0293793436139822,
      "learning_rate": 5.765196766916691e-06,
      "loss": 0.0013,
      "step": 92770
    },
    {
      "epoch": 1.5181215740816494,
      "grad_norm": 0.08452704548835754,
      "learning_rate": 5.764256026737265e-06,
      "loss": 0.0014,
      "step": 92780
    },
    {
      "epoch": 1.5182852000327252,
      "grad_norm": 0.10538189113140106,
      "learning_rate": 5.7633152588553945e-06,
      "loss": 0.0012,
      "step": 92790
    },
    {
      "epoch": 1.5184488259838012,
      "grad_norm": 0.04788101464509964,
      "learning_rate": 5.76237446330518e-06,
      "loss": 0.0019,
      "step": 92800
    },
    {
      "epoch": 1.5186124519348767,
      "grad_norm": 0.3791896104812622,
      "learning_rate": 5.761433640120724e-06,
      "loss": 0.0022,
      "step": 92810
    },
    {
      "epoch": 1.5187760778859527,
      "grad_norm": 0.07501193881034851,
      "learning_rate": 5.76049278933613e-06,
      "loss": 0.0013,
      "step": 92820
    },
    {
      "epoch": 1.5189397038370287,
      "grad_norm": 0.12513427436351776,
      "learning_rate": 5.759551910985499e-06,
      "loss": 0.0017,
      "step": 92830
    },
    {
      "epoch": 1.5191033297881043,
      "grad_norm": 0.3205004930496216,
      "learning_rate": 5.758611005102935e-06,
      "loss": 0.0021,
      "step": 92840
    },
    {
      "epoch": 1.5192669557391802,
      "grad_norm": 0.1463504284620285,
      "learning_rate": 5.757670071722547e-06,
      "loss": 0.0019,
      "step": 92850
    },
    {
      "epoch": 1.519430581690256,
      "grad_norm": 0.040811534970998764,
      "learning_rate": 5.75672911087844e-06,
      "loss": 0.0031,
      "step": 92860
    },
    {
      "epoch": 1.5195942076413318,
      "grad_norm": 0.10794855654239655,
      "learning_rate": 5.755788122604722e-06,
      "loss": 0.0009,
      "step": 92870
    },
    {
      "epoch": 1.5197578335924078,
      "grad_norm": 0.13340447843074799,
      "learning_rate": 5.754847106935501e-06,
      "loss": 0.0012,
      "step": 92880
    },
    {
      "epoch": 1.5199214595434836,
      "grad_norm": 0.22035622596740723,
      "learning_rate": 5.753906063904887e-06,
      "loss": 0.002,
      "step": 92890
    },
    {
      "epoch": 1.5200850854945593,
      "grad_norm": 0.09209461510181427,
      "learning_rate": 5.752964993546989e-06,
      "loss": 0.0017,
      "step": 92900
    },
    {
      "epoch": 1.5202487114456353,
      "grad_norm": 0.017345735803246498,
      "learning_rate": 5.7520238958959196e-06,
      "loss": 0.0012,
      "step": 92910
    },
    {
      "epoch": 1.5204123373967111,
      "grad_norm": 0.040596071630716324,
      "learning_rate": 5.751082770985791e-06,
      "loss": 0.0019,
      "step": 92920
    },
    {
      "epoch": 1.5205759633477869,
      "grad_norm": 0.074709951877594,
      "learning_rate": 5.7501416188507194e-06,
      "loss": 0.0016,
      "step": 92930
    },
    {
      "epoch": 1.5207395892988629,
      "grad_norm": 0.02531144767999649,
      "learning_rate": 5.749200439524816e-06,
      "loss": 0.0025,
      "step": 92940
    },
    {
      "epoch": 1.5209032152499387,
      "grad_norm": 0.0254683755338192,
      "learning_rate": 5.748259233042196e-06,
      "loss": 0.0026,
      "step": 92950
    },
    {
      "epoch": 1.5210668412010144,
      "grad_norm": 0.12783615291118622,
      "learning_rate": 5.747317999436979e-06,
      "loss": 0.0032,
      "step": 92960
    },
    {
      "epoch": 1.5212304671520904,
      "grad_norm": 0.06539308279752731,
      "learning_rate": 5.746376738743279e-06,
      "loss": 0.0013,
      "step": 92970
    },
    {
      "epoch": 1.5213940931031662,
      "grad_norm": 0.2535451054573059,
      "learning_rate": 5.745435450995217e-06,
      "loss": 0.0017,
      "step": 92980
    },
    {
      "epoch": 1.521557719054242,
      "grad_norm": 0.27654707431793213,
      "learning_rate": 5.744494136226911e-06,
      "loss": 0.0026,
      "step": 92990
    },
    {
      "epoch": 1.521721345005318,
      "grad_norm": 0.008708052337169647,
      "learning_rate": 5.743552794472483e-06,
      "loss": 0.0016,
      "step": 93000
    },
    {
      "epoch": 1.5218849709563935,
      "grad_norm": 0.15754805505275726,
      "learning_rate": 5.742611425766052e-06,
      "loss": 0.0023,
      "step": 93010
    },
    {
      "epoch": 1.5220485969074695,
      "grad_norm": 0.08248157799243927,
      "learning_rate": 5.741670030141742e-06,
      "loss": 0.0017,
      "step": 93020
    },
    {
      "epoch": 1.5222122228585455,
      "grad_norm": 0.11970769613981247,
      "learning_rate": 5.740728607633676e-06,
      "loss": 0.0025,
      "step": 93030
    },
    {
      "epoch": 1.522375848809621,
      "grad_norm": 0.06609880924224854,
      "learning_rate": 5.739787158275979e-06,
      "loss": 0.0019,
      "step": 93040
    },
    {
      "epoch": 1.522539474760697,
      "grad_norm": 0.28861188888549805,
      "learning_rate": 5.738845682102775e-06,
      "loss": 0.0035,
      "step": 93050
    },
    {
      "epoch": 1.5227031007117728,
      "grad_norm": 0.07973862439393997,
      "learning_rate": 5.737904179148189e-06,
      "loss": 0.002,
      "step": 93060
    },
    {
      "epoch": 1.5228667266628486,
      "grad_norm": 0.1383950263261795,
      "learning_rate": 5.736962649446352e-06,
      "loss": 0.0014,
      "step": 93070
    },
    {
      "epoch": 1.5230303526139246,
      "grad_norm": 0.13354241847991943,
      "learning_rate": 5.736021093031391e-06,
      "loss": 0.0027,
      "step": 93080
    },
    {
      "epoch": 1.5231939785650004,
      "grad_norm": 0.1700313240289688,
      "learning_rate": 5.735079509937434e-06,
      "loss": 0.0021,
      "step": 93090
    },
    {
      "epoch": 1.5233576045160762,
      "grad_norm": 0.13881738483905792,
      "learning_rate": 5.734137900198611e-06,
      "loss": 0.0028,
      "step": 93100
    },
    {
      "epoch": 1.5235212304671522,
      "grad_norm": 0.05037563666701317,
      "learning_rate": 5.733196263849052e-06,
      "loss": 0.0021,
      "step": 93110
    },
    {
      "epoch": 1.523684856418228,
      "grad_norm": 0.1296752691268921,
      "learning_rate": 5.732254600922893e-06,
      "loss": 0.0017,
      "step": 93120
    },
    {
      "epoch": 1.5238484823693037,
      "grad_norm": 0.18967126309871674,
      "learning_rate": 5.731312911454263e-06,
      "loss": 0.0017,
      "step": 93130
    },
    {
      "epoch": 1.5240121083203797,
      "grad_norm": 0.0447850376367569,
      "learning_rate": 5.730371195477298e-06,
      "loss": 0.0007,
      "step": 93140
    },
    {
      "epoch": 1.5241757342714555,
      "grad_norm": 0.09801087528467178,
      "learning_rate": 5.729429453026133e-06,
      "loss": 0.0037,
      "step": 93150
    },
    {
      "epoch": 1.5243393602225312,
      "grad_norm": 0.008364412933588028,
      "learning_rate": 5.728487684134904e-06,
      "loss": 0.0015,
      "step": 93160
    },
    {
      "epoch": 1.5245029861736072,
      "grad_norm": 0.09155111759901047,
      "learning_rate": 5.7275458888377455e-06,
      "loss": 0.0019,
      "step": 93170
    },
    {
      "epoch": 1.524666612124683,
      "grad_norm": 0.12245266884565353,
      "learning_rate": 5.726604067168799e-06,
      "loss": 0.0018,
      "step": 93180
    },
    {
      "epoch": 1.5248302380757588,
      "grad_norm": 0.033675309270620346,
      "learning_rate": 5.725662219162201e-06,
      "loss": 0.0021,
      "step": 93190
    },
    {
      "epoch": 1.5249938640268348,
      "grad_norm": 0.05388784036040306,
      "learning_rate": 5.72472034485209e-06,
      "loss": 0.0022,
      "step": 93200
    },
    {
      "epoch": 1.5251574899779103,
      "grad_norm": 0.150056853890419,
      "learning_rate": 5.723778444272609e-06,
      "loss": 0.0024,
      "step": 93210
    },
    {
      "epoch": 1.5253211159289863,
      "grad_norm": 0.11732979863882065,
      "learning_rate": 5.7228365174579e-06,
      "loss": 0.0018,
      "step": 93220
    },
    {
      "epoch": 1.5254847418800623,
      "grad_norm": 0.09717657417058945,
      "learning_rate": 5.721894564442105e-06,
      "loss": 0.0015,
      "step": 93230
    },
    {
      "epoch": 1.5256483678311379,
      "grad_norm": 0.0715896487236023,
      "learning_rate": 5.720952585259368e-06,
      "loss": 0.0025,
      "step": 93240
    },
    {
      "epoch": 1.5258119937822139,
      "grad_norm": 0.07088734209537506,
      "learning_rate": 5.720010579943829e-06,
      "loss": 0.001,
      "step": 93250
    },
    {
      "epoch": 1.5259756197332897,
      "grad_norm": 0.08547870814800262,
      "learning_rate": 5.719068548529639e-06,
      "loss": 0.0016,
      "step": 93260
    },
    {
      "epoch": 1.5261392456843654,
      "grad_norm": 0.02438625507056713,
      "learning_rate": 5.718126491050941e-06,
      "loss": 0.0011,
      "step": 93270
    },
    {
      "epoch": 1.5263028716354414,
      "grad_norm": 0.11504413932561874,
      "learning_rate": 5.717184407541886e-06,
      "loss": 0.0013,
      "step": 93280
    },
    {
      "epoch": 1.5264664975865172,
      "grad_norm": 0.061093397438526154,
      "learning_rate": 5.716242298036618e-06,
      "loss": 0.0012,
      "step": 93290
    },
    {
      "epoch": 1.526630123537593,
      "grad_norm": 0.13431446254253387,
      "learning_rate": 5.715300162569288e-06,
      "loss": 0.0023,
      "step": 93300
    },
    {
      "epoch": 1.526793749488669,
      "grad_norm": 0.27459803223609924,
      "learning_rate": 5.714358001174048e-06,
      "loss": 0.0034,
      "step": 93310
    },
    {
      "epoch": 1.5269573754397447,
      "grad_norm": 0.04602712392807007,
      "learning_rate": 5.713415813885046e-06,
      "loss": 0.0032,
      "step": 93320
    },
    {
      "epoch": 1.5271210013908205,
      "grad_norm": 0.16966798901557922,
      "learning_rate": 5.712473600736437e-06,
      "loss": 0.0078,
      "step": 93330
    },
    {
      "epoch": 1.5272846273418965,
      "grad_norm": 0.0796501561999321,
      "learning_rate": 5.71153136176237e-06,
      "loss": 0.0021,
      "step": 93340
    },
    {
      "epoch": 1.5274482532929723,
      "grad_norm": 0.21493178606033325,
      "learning_rate": 5.710589096997002e-06,
      "loss": 0.0021,
      "step": 93350
    },
    {
      "epoch": 1.527611879244048,
      "grad_norm": 0.03788410872220993,
      "learning_rate": 5.7096468064744885e-06,
      "loss": 0.0014,
      "step": 93360
    },
    {
      "epoch": 1.527775505195124,
      "grad_norm": 0.04838072508573532,
      "learning_rate": 5.708704490228983e-06,
      "loss": 0.0019,
      "step": 93370
    },
    {
      "epoch": 1.5279391311461998,
      "grad_norm": 0.06535517424345016,
      "learning_rate": 5.707762148294643e-06,
      "loss": 0.0019,
      "step": 93380
    },
    {
      "epoch": 1.5281027570972756,
      "grad_norm": 0.03749818727374077,
      "learning_rate": 5.706819780705627e-06,
      "loss": 0.0024,
      "step": 93390
    },
    {
      "epoch": 1.5282663830483516,
      "grad_norm": 0.22494404017925262,
      "learning_rate": 5.705877387496091e-06,
      "loss": 0.0015,
      "step": 93400
    },
    {
      "epoch": 1.5284300089994272,
      "grad_norm": 0.07843749970197678,
      "learning_rate": 5.704934968700197e-06,
      "loss": 0.0024,
      "step": 93410
    },
    {
      "epoch": 1.5285936349505032,
      "grad_norm": 0.07453209906816483,
      "learning_rate": 5.703992524352104e-06,
      "loss": 0.0022,
      "step": 93420
    },
    {
      "epoch": 1.5287572609015792,
      "grad_norm": 0.04251955822110176,
      "learning_rate": 5.703050054485976e-06,
      "loss": 0.0009,
      "step": 93430
    },
    {
      "epoch": 1.5289208868526547,
      "grad_norm": 0.06735698878765106,
      "learning_rate": 5.702107559135972e-06,
      "loss": 0.0016,
      "step": 93440
    },
    {
      "epoch": 1.5290845128037307,
      "grad_norm": 0.11309357732534409,
      "learning_rate": 5.701165038336256e-06,
      "loss": 0.0018,
      "step": 93450
    },
    {
      "epoch": 1.5292481387548065,
      "grad_norm": 0.16005238890647888,
      "learning_rate": 5.7002224921209935e-06,
      "loss": 0.003,
      "step": 93460
    },
    {
      "epoch": 1.5294117647058822,
      "grad_norm": 0.06462351977825165,
      "learning_rate": 5.699279920524348e-06,
      "loss": 0.0013,
      "step": 93470
    },
    {
      "epoch": 1.5295753906569582,
      "grad_norm": 0.012961739674210548,
      "learning_rate": 5.698337323580486e-06,
      "loss": 0.0015,
      "step": 93480
    },
    {
      "epoch": 1.529739016608034,
      "grad_norm": 0.025400031358003616,
      "learning_rate": 5.697394701323574e-06,
      "loss": 0.0026,
      "step": 93490
    },
    {
      "epoch": 1.5299026425591098,
      "grad_norm": 0.1740194410085678,
      "learning_rate": 5.696452053787781e-06,
      "loss": 0.0023,
      "step": 93500
    },
    {
      "epoch": 1.5300662685101858,
      "grad_norm": 0.12747587263584137,
      "learning_rate": 5.695509381007275e-06,
      "loss": 0.0012,
      "step": 93510
    },
    {
      "epoch": 1.5302298944612616,
      "grad_norm": 0.09946295619010925,
      "learning_rate": 5.694566683016225e-06,
      "loss": 0.0017,
      "step": 93520
    },
    {
      "epoch": 1.5303935204123373,
      "grad_norm": 0.17157401144504547,
      "learning_rate": 5.693623959848804e-06,
      "loss": 0.0022,
      "step": 93530
    },
    {
      "epoch": 1.5305571463634133,
      "grad_norm": 0.32592588663101196,
      "learning_rate": 5.692681211539179e-06,
      "loss": 0.0011,
      "step": 93540
    },
    {
      "epoch": 1.530720772314489,
      "grad_norm": 0.5229214429855347,
      "learning_rate": 5.691738438121527e-06,
      "loss": 0.0031,
      "step": 93550
    },
    {
      "epoch": 1.5308843982655649,
      "grad_norm": 0.09958718717098236,
      "learning_rate": 5.690795639630017e-06,
      "loss": 0.0032,
      "step": 93560
    },
    {
      "epoch": 1.5310480242166409,
      "grad_norm": 0.22845932841300964,
      "learning_rate": 5.689852816098825e-06,
      "loss": 0.0016,
      "step": 93570
    },
    {
      "epoch": 1.5312116501677167,
      "grad_norm": 0.12248405069112778,
      "learning_rate": 5.688909967562129e-06,
      "loss": 0.0013,
      "step": 93580
    },
    {
      "epoch": 1.5313752761187924,
      "grad_norm": 0.07610813528299332,
      "learning_rate": 5.687967094054102e-06,
      "loss": 0.0013,
      "step": 93590
    },
    {
      "epoch": 1.5315389020698684,
      "grad_norm": 0.20003165304660797,
      "learning_rate": 5.68702419560892e-06,
      "loss": 0.0025,
      "step": 93600
    },
    {
      "epoch": 1.531702528020944,
      "grad_norm": 0.05017508938908577,
      "learning_rate": 5.686081272260763e-06,
      "loss": 0.0029,
      "step": 93610
    },
    {
      "epoch": 1.53186615397202,
      "grad_norm": 0.14710979163646698,
      "learning_rate": 5.685138324043808e-06,
      "loss": 0.0013,
      "step": 93620
    },
    {
      "epoch": 1.532029779923096,
      "grad_norm": 0.6115508675575256,
      "learning_rate": 5.684195350992236e-06,
      "loss": 0.0025,
      "step": 93630
    },
    {
      "epoch": 1.5321934058741715,
      "grad_norm": 0.049535464495420456,
      "learning_rate": 5.683252353140227e-06,
      "loss": 0.0018,
      "step": 93640
    },
    {
      "epoch": 1.5323570318252475,
      "grad_norm": 0.1612945944070816,
      "learning_rate": 5.682309330521962e-06,
      "loss": 0.0016,
      "step": 93650
    },
    {
      "epoch": 1.5325206577763233,
      "grad_norm": 0.10504493117332458,
      "learning_rate": 5.681366283171624e-06,
      "loss": 0.0017,
      "step": 93660
    },
    {
      "epoch": 1.532684283727399,
      "grad_norm": 0.05319498851895332,
      "learning_rate": 5.680423211123396e-06,
      "loss": 0.0025,
      "step": 93670
    },
    {
      "epoch": 1.532847909678475,
      "grad_norm": 0.05620495229959488,
      "learning_rate": 5.679480114411463e-06,
      "loss": 0.0009,
      "step": 93680
    },
    {
      "epoch": 1.5330115356295508,
      "grad_norm": 0.05144511163234711,
      "learning_rate": 5.678536993070007e-06,
      "loss": 0.0016,
      "step": 93690
    },
    {
      "epoch": 1.5331751615806266,
      "grad_norm": 0.02680237591266632,
      "learning_rate": 5.677593847133216e-06,
      "loss": 0.0019,
      "step": 93700
    },
    {
      "epoch": 1.5333387875317026,
      "grad_norm": 0.07639974355697632,
      "learning_rate": 5.676650676635277e-06,
      "loss": 0.0014,
      "step": 93710
    },
    {
      "epoch": 1.5335024134827784,
      "grad_norm": 0.10937551409006119,
      "learning_rate": 5.675707481610377e-06,
      "loss": 0.0018,
      "step": 93720
    },
    {
      "epoch": 1.5336660394338542,
      "grad_norm": 0.050971146672964096,
      "learning_rate": 5.674764262092705e-06,
      "loss": 0.0027,
      "step": 93730
    },
    {
      "epoch": 1.5338296653849302,
      "grad_norm": 0.028386056423187256,
      "learning_rate": 5.673821018116452e-06,
      "loss": 0.0015,
      "step": 93740
    },
    {
      "epoch": 1.533993291336006,
      "grad_norm": 0.1227363646030426,
      "learning_rate": 5.672877749715804e-06,
      "loss": 0.0017,
      "step": 93750
    },
    {
      "epoch": 1.5341569172870817,
      "grad_norm": 0.1913052499294281,
      "learning_rate": 5.671934456924956e-06,
      "loss": 0.003,
      "step": 93760
    },
    {
      "epoch": 1.5343205432381577,
      "grad_norm": 0.11627969890832901,
      "learning_rate": 5.6709911397780994e-06,
      "loss": 0.0014,
      "step": 93770
    },
    {
      "epoch": 1.5344841691892332,
      "grad_norm": 0.19431929290294647,
      "learning_rate": 5.6700477983094236e-06,
      "loss": 0.0017,
      "step": 93780
    },
    {
      "epoch": 1.5346477951403092,
      "grad_norm": 0.15582622587680817,
      "learning_rate": 5.66910443255313e-06,
      "loss": 0.0019,
      "step": 93790
    },
    {
      "epoch": 1.5348114210913852,
      "grad_norm": 0.06606801599264145,
      "learning_rate": 5.6681610425434065e-06,
      "loss": 0.0015,
      "step": 93800
    },
    {
      "epoch": 1.5349750470424608,
      "grad_norm": 0.06744344532489777,
      "learning_rate": 5.6672176283144505e-06,
      "loss": 0.0029,
      "step": 93810
    },
    {
      "epoch": 1.5351386729935368,
      "grad_norm": 0.09633451700210571,
      "learning_rate": 5.66627418990046e-06,
      "loss": 0.0022,
      "step": 93820
    },
    {
      "epoch": 1.5353022989446126,
      "grad_norm": 0.09357360750436783,
      "learning_rate": 5.665330727335632e-06,
      "loss": 0.0023,
      "step": 93830
    },
    {
      "epoch": 1.5354659248956883,
      "grad_norm": 0.09555008262395859,
      "learning_rate": 5.664387240654163e-06,
      "loss": 0.0013,
      "step": 93840
    },
    {
      "epoch": 1.5356295508467643,
      "grad_norm": 0.3494846224784851,
      "learning_rate": 5.663443729890255e-06,
      "loss": 0.0017,
      "step": 93850
    },
    {
      "epoch": 1.53579317679784,
      "grad_norm": 0.10591018944978714,
      "learning_rate": 5.6625001950781044e-06,
      "loss": 0.0036,
      "step": 93860
    },
    {
      "epoch": 1.5359568027489159,
      "grad_norm": 0.017858639359474182,
      "learning_rate": 5.661556636251915e-06,
      "loss": 0.0013,
      "step": 93870
    },
    {
      "epoch": 1.5361204286999919,
      "grad_norm": 0.055446069687604904,
      "learning_rate": 5.660613053445887e-06,
      "loss": 0.001,
      "step": 93880
    },
    {
      "epoch": 1.5362840546510677,
      "grad_norm": 0.04519607126712799,
      "learning_rate": 5.6596694466942245e-06,
      "loss": 0.0033,
      "step": 93890
    },
    {
      "epoch": 1.5364476806021434,
      "grad_norm": 0.03166491538286209,
      "learning_rate": 5.6587258160311295e-06,
      "loss": 0.0014,
      "step": 93900
    },
    {
      "epoch": 1.5366113065532194,
      "grad_norm": 0.09406087547540665,
      "learning_rate": 5.657782161490806e-06,
      "loss": 0.0017,
      "step": 93910
    },
    {
      "epoch": 1.5367749325042952,
      "grad_norm": 0.07638242840766907,
      "learning_rate": 5.656838483107461e-06,
      "loss": 0.0015,
      "step": 93920
    },
    {
      "epoch": 1.536938558455371,
      "grad_norm": 0.13121093809604645,
      "learning_rate": 5.6558947809153e-06,
      "loss": 0.0117,
      "step": 93930
    },
    {
      "epoch": 1.537102184406447,
      "grad_norm": 0.1247217059135437,
      "learning_rate": 5.65495105494853e-06,
      "loss": 0.0009,
      "step": 93940
    },
    {
      "epoch": 1.5372658103575227,
      "grad_norm": 0.009813793003559113,
      "learning_rate": 5.654007305241358e-06,
      "loss": 0.0015,
      "step": 93950
    },
    {
      "epoch": 1.5374294363085985,
      "grad_norm": 0.11754433065652847,
      "learning_rate": 5.653063531827993e-06,
      "loss": 0.0012,
      "step": 93960
    },
    {
      "epoch": 1.5375930622596745,
      "grad_norm": 0.055549997836351395,
      "learning_rate": 5.652119734742642e-06,
      "loss": 0.001,
      "step": 93970
    },
    {
      "epoch": 1.53775668821075,
      "grad_norm": 0.12981769442558289,
      "learning_rate": 5.6511759140195214e-06,
      "loss": 0.002,
      "step": 93980
    },
    {
      "epoch": 1.537920314161826,
      "grad_norm": 0.08106281608343124,
      "learning_rate": 5.650232069692838e-06,
      "loss": 0.0023,
      "step": 93990
    },
    {
      "epoch": 1.538083940112902,
      "grad_norm": 0.33574244379997253,
      "learning_rate": 5.649288201796804e-06,
      "loss": 0.0026,
      "step": 94000
    },
    {
      "epoch": 1.5382475660639776,
      "grad_norm": 0.1332085132598877,
      "learning_rate": 5.648344310365634e-06,
      "loss": 0.0026,
      "step": 94010
    },
    {
      "epoch": 1.5384111920150536,
      "grad_norm": 0.16106055676937103,
      "learning_rate": 5.647400395433541e-06,
      "loss": 0.0033,
      "step": 94020
    },
    {
      "epoch": 1.5385748179661294,
      "grad_norm": 0.13626396656036377,
      "learning_rate": 5.646456457034739e-06,
      "loss": 0.003,
      "step": 94030
    },
    {
      "epoch": 1.5387384439172052,
      "grad_norm": 0.10682516545057297,
      "learning_rate": 5.645512495203446e-06,
      "loss": 0.002,
      "step": 94040
    },
    {
      "epoch": 1.5389020698682812,
      "grad_norm": 0.06553791463375092,
      "learning_rate": 5.644568509973874e-06,
      "loss": 0.0014,
      "step": 94050
    },
    {
      "epoch": 1.539065695819357,
      "grad_norm": 0.16887657344341278,
      "learning_rate": 5.6436245013802436e-06,
      "loss": 0.0038,
      "step": 94060
    },
    {
      "epoch": 1.5392293217704327,
      "grad_norm": 0.10290270298719406,
      "learning_rate": 5.642680469456771e-06,
      "loss": 0.002,
      "step": 94070
    },
    {
      "epoch": 1.5393929477215087,
      "grad_norm": 0.09996610134840012,
      "learning_rate": 5.641736414237678e-06,
      "loss": 0.002,
      "step": 94080
    },
    {
      "epoch": 1.5395565736725845,
      "grad_norm": 0.04237337037920952,
      "learning_rate": 5.640792335757182e-06,
      "loss": 0.0028,
      "step": 94090
    },
    {
      "epoch": 1.5397201996236602,
      "grad_norm": 0.18140679597854614,
      "learning_rate": 5.639848234049504e-06,
      "loss": 0.0021,
      "step": 94100
    },
    {
      "epoch": 1.5398838255747362,
      "grad_norm": 0.24077308177947998,
      "learning_rate": 5.638904109148865e-06,
      "loss": 0.002,
      "step": 94110
    },
    {
      "epoch": 1.540047451525812,
      "grad_norm": 0.07921037077903748,
      "learning_rate": 5.637959961089487e-06,
      "loss": 0.0017,
      "step": 94120
    },
    {
      "epoch": 1.5402110774768878,
      "grad_norm": 0.07384908944368362,
      "learning_rate": 5.637015789905593e-06,
      "loss": 0.0017,
      "step": 94130
    },
    {
      "epoch": 1.5403747034279638,
      "grad_norm": 0.2995862364768982,
      "learning_rate": 5.636071595631409e-06,
      "loss": 0.0024,
      "step": 94140
    },
    {
      "epoch": 1.5405383293790396,
      "grad_norm": 0.06803367286920547,
      "learning_rate": 5.63512737830116e-06,
      "loss": 0.001,
      "step": 94150
    },
    {
      "epoch": 1.5407019553301153,
      "grad_norm": 0.08918260037899017,
      "learning_rate": 5.634183137949068e-06,
      "loss": 0.003,
      "step": 94160
    },
    {
      "epoch": 1.5408655812811913,
      "grad_norm": 0.10834448784589767,
      "learning_rate": 5.633238874609361e-06,
      "loss": 0.0018,
      "step": 94170
    },
    {
      "epoch": 1.5410292072322669,
      "grad_norm": 0.057486191391944885,
      "learning_rate": 5.632294588316267e-06,
      "loss": 0.0025,
      "step": 94180
    },
    {
      "epoch": 1.5411928331833429,
      "grad_norm": 0.10096415132284164,
      "learning_rate": 5.6313502791040145e-06,
      "loss": 0.0028,
      "step": 94190
    },
    {
      "epoch": 1.5413564591344189,
      "grad_norm": 0.11893796920776367,
      "learning_rate": 5.630405947006832e-06,
      "loss": 0.0015,
      "step": 94200
    },
    {
      "epoch": 1.5415200850854944,
      "grad_norm": 0.06593409180641174,
      "learning_rate": 5.629461592058948e-06,
      "loss": 0.0014,
      "step": 94210
    },
    {
      "epoch": 1.5416837110365704,
      "grad_norm": 0.021921513602137566,
      "learning_rate": 5.628517214294594e-06,
      "loss": 0.0016,
      "step": 94220
    },
    {
      "epoch": 1.5418473369876462,
      "grad_norm": 0.05870698392391205,
      "learning_rate": 5.6275728137480034e-06,
      "loss": 0.001,
      "step": 94230
    },
    {
      "epoch": 1.542010962938722,
      "grad_norm": 0.13184301555156708,
      "learning_rate": 5.626628390453406e-06,
      "loss": 0.0015,
      "step": 94240
    },
    {
      "epoch": 1.542174588889798,
      "grad_norm": 0.06875904649496078,
      "learning_rate": 5.625683944445035e-06,
      "loss": 0.0012,
      "step": 94250
    },
    {
      "epoch": 1.5423382148408737,
      "grad_norm": 0.06648250669240952,
      "learning_rate": 5.624739475757125e-06,
      "loss": 0.0021,
      "step": 94260
    },
    {
      "epoch": 1.5425018407919495,
      "grad_norm": 0.06631302088499069,
      "learning_rate": 5.623794984423909e-06,
      "loss": 0.0017,
      "step": 94270
    },
    {
      "epoch": 1.5426654667430255,
      "grad_norm": 0.06341740489006042,
      "learning_rate": 5.622850470479626e-06,
      "loss": 0.002,
      "step": 94280
    },
    {
      "epoch": 1.5428290926941013,
      "grad_norm": 0.06753361225128174,
      "learning_rate": 5.6219059339585115e-06,
      "loss": 0.002,
      "step": 94290
    },
    {
      "epoch": 1.542992718645177,
      "grad_norm": 0.06312259286642075,
      "learning_rate": 5.6209613748948e-06,
      "loss": 0.0023,
      "step": 94300
    },
    {
      "epoch": 1.543156344596253,
      "grad_norm": 0.08323318511247635,
      "learning_rate": 5.62001679332273e-06,
      "loss": 0.0016,
      "step": 94310
    },
    {
      "epoch": 1.5433199705473288,
      "grad_norm": 0.24576202034950256,
      "learning_rate": 5.6190721892765435e-06,
      "loss": 0.0015,
      "step": 94320
    },
    {
      "epoch": 1.5434835964984046,
      "grad_norm": 0.14893104135990143,
      "learning_rate": 5.618127562790476e-06,
      "loss": 0.0017,
      "step": 94330
    },
    {
      "epoch": 1.5436472224494806,
      "grad_norm": 0.06032591685652733,
      "learning_rate": 5.617182913898772e-06,
      "loss": 0.0016,
      "step": 94340
    },
    {
      "epoch": 1.5438108484005564,
      "grad_norm": 0.1430543214082718,
      "learning_rate": 5.6162382426356696e-06,
      "loss": 0.0016,
      "step": 94350
    },
    {
      "epoch": 1.5439744743516322,
      "grad_norm": 0.01067027822136879,
      "learning_rate": 5.615293549035412e-06,
      "loss": 0.0013,
      "step": 94360
    },
    {
      "epoch": 1.5441381003027081,
      "grad_norm": 0.07077572494745255,
      "learning_rate": 5.614348833132242e-06,
      "loss": 0.002,
      "step": 94370
    },
    {
      "epoch": 1.5443017262537837,
      "grad_norm": 0.11045695841312408,
      "learning_rate": 5.613404094960404e-06,
      "loss": 0.0023,
      "step": 94380
    },
    {
      "epoch": 1.5444653522048597,
      "grad_norm": 0.02371237799525261,
      "learning_rate": 5.612459334554143e-06,
      "loss": 0.0009,
      "step": 94390
    },
    {
      "epoch": 1.5446289781559357,
      "grad_norm": 0.1202223002910614,
      "learning_rate": 5.611514551947701e-06,
      "loss": 0.0023,
      "step": 94400
    },
    {
      "epoch": 1.5447926041070112,
      "grad_norm": 0.08319107443094254,
      "learning_rate": 5.610569747175325e-06,
      "loss": 0.0032,
      "step": 94410
    },
    {
      "epoch": 1.5449562300580872,
      "grad_norm": 0.06446091085672379,
      "learning_rate": 5.609624920271265e-06,
      "loss": 0.0013,
      "step": 94420
    },
    {
      "epoch": 1.545119856009163,
      "grad_norm": 0.10032033920288086,
      "learning_rate": 5.608680071269768e-06,
      "loss": 0.0026,
      "step": 94430
    },
    {
      "epoch": 1.5452834819602388,
      "grad_norm": 0.19656500220298767,
      "learning_rate": 5.60773520020508e-06,
      "loss": 0.0024,
      "step": 94440
    },
    {
      "epoch": 1.5454471079113148,
      "grad_norm": 0.04258721321821213,
      "learning_rate": 5.606790307111452e-06,
      "loss": 0.0023,
      "step": 94450
    },
    {
      "epoch": 1.5456107338623906,
      "grad_norm": 0.4815067946910858,
      "learning_rate": 5.605845392023133e-06,
      "loss": 0.0021,
      "step": 94460
    },
    {
      "epoch": 1.5457743598134663,
      "grad_norm": 0.10441657155752182,
      "learning_rate": 5.6049004549743755e-06,
      "loss": 0.0022,
      "step": 94470
    },
    {
      "epoch": 1.5459379857645423,
      "grad_norm": 0.2149830162525177,
      "learning_rate": 5.60395549599943e-06,
      "loss": 0.0025,
      "step": 94480
    },
    {
      "epoch": 1.546101611715618,
      "grad_norm": 0.11567191779613495,
      "learning_rate": 5.6030105151325485e-06,
      "loss": 0.0022,
      "step": 94490
    },
    {
      "epoch": 1.5462652376666939,
      "grad_norm": 0.25759270787239075,
      "learning_rate": 5.602065512407987e-06,
      "loss": 0.0027,
      "step": 94500
    },
    {
      "epoch": 1.5464288636177699,
      "grad_norm": 0.13471858203411102,
      "learning_rate": 5.601120487859997e-06,
      "loss": 0.0027,
      "step": 94510
    },
    {
      "epoch": 1.5465924895688457,
      "grad_norm": 0.06561803817749023,
      "learning_rate": 5.600175441522832e-06,
      "loss": 0.0016,
      "step": 94520
    },
    {
      "epoch": 1.5467561155199214,
      "grad_norm": 0.0779205858707428,
      "learning_rate": 5.5992303734307515e-06,
      "loss": 0.0008,
      "step": 94530
    },
    {
      "epoch": 1.5469197414709974,
      "grad_norm": 0.1218620166182518,
      "learning_rate": 5.5982852836180125e-06,
      "loss": 0.001,
      "step": 94540
    },
    {
      "epoch": 1.547083367422073,
      "grad_norm": 0.13607439398765564,
      "learning_rate": 5.597340172118866e-06,
      "loss": 0.0017,
      "step": 94550
    },
    {
      "epoch": 1.547246993373149,
      "grad_norm": 0.10613955557346344,
      "learning_rate": 5.596395038967576e-06,
      "loss": 0.0021,
      "step": 94560
    },
    {
      "epoch": 1.547410619324225,
      "grad_norm": 0.028159035369753838,
      "learning_rate": 5.595449884198399e-06,
      "loss": 0.0023,
      "step": 94570
    },
    {
      "epoch": 1.5475742452753005,
      "grad_norm": 0.15041640400886536,
      "learning_rate": 5.594504707845594e-06,
      "loss": 0.0015,
      "step": 94580
    },
    {
      "epoch": 1.5477378712263765,
      "grad_norm": 0.12473514676094055,
      "learning_rate": 5.593559509943423e-06,
      "loss": 0.0016,
      "step": 94590
    },
    {
      "epoch": 1.5479014971774523,
      "grad_norm": 0.08697626739740372,
      "learning_rate": 5.592614290526146e-06,
      "loss": 0.0015,
      "step": 94600
    },
    {
      "epoch": 1.548065123128528,
      "grad_norm": 0.3288215398788452,
      "learning_rate": 5.591669049628026e-06,
      "loss": 0.0025,
      "step": 94610
    },
    {
      "epoch": 1.548228749079604,
      "grad_norm": 0.18107466399669647,
      "learning_rate": 5.5907237872833246e-06,
      "loss": 0.0016,
      "step": 94620
    },
    {
      "epoch": 1.5483923750306798,
      "grad_norm": 0.3711811900138855,
      "learning_rate": 5.589778503526304e-06,
      "loss": 0.0044,
      "step": 94630
    },
    {
      "epoch": 1.5485560009817556,
      "grad_norm": 0.061258625239133835,
      "learning_rate": 5.588833198391233e-06,
      "loss": 0.0013,
      "step": 94640
    },
    {
      "epoch": 1.5487196269328316,
      "grad_norm": 0.06625762581825256,
      "learning_rate": 5.587887871912372e-06,
      "loss": 0.0014,
      "step": 94650
    },
    {
      "epoch": 1.5488832528839074,
      "grad_norm": 0.14573733508586884,
      "learning_rate": 5.586942524123988e-06,
      "loss": 0.0021,
      "step": 94660
    },
    {
      "epoch": 1.5490468788349832,
      "grad_norm": 0.057698752731084824,
      "learning_rate": 5.58599715506035e-06,
      "loss": 0.0007,
      "step": 94670
    },
    {
      "epoch": 1.5492105047860592,
      "grad_norm": 0.018713563680648804,
      "learning_rate": 5.585051764755722e-06,
      "loss": 0.0014,
      "step": 94680
    },
    {
      "epoch": 1.549374130737135,
      "grad_norm": 0.04081779718399048,
      "learning_rate": 5.584106353244374e-06,
      "loss": 0.0018,
      "step": 94690
    },
    {
      "epoch": 1.5495377566882107,
      "grad_norm": 0.09372851252555847,
      "learning_rate": 5.583160920560576e-06,
      "loss": 0.0018,
      "step": 94700
    },
    {
      "epoch": 1.5497013826392867,
      "grad_norm": 0.08448544144630432,
      "learning_rate": 5.582215466738594e-06,
      "loss": 0.0008,
      "step": 94710
    },
    {
      "epoch": 1.5498650085903625,
      "grad_norm": 0.06745804101228714,
      "learning_rate": 5.581269991812702e-06,
      "loss": 0.0011,
      "step": 94720
    },
    {
      "epoch": 1.5500286345414382,
      "grad_norm": 0.01718783751130104,
      "learning_rate": 5.58032449581717e-06,
      "loss": 0.0015,
      "step": 94730
    },
    {
      "epoch": 1.5501922604925142,
      "grad_norm": 0.10950352251529694,
      "learning_rate": 5.5793789787862694e-06,
      "loss": 0.001,
      "step": 94740
    },
    {
      "epoch": 1.5503558864435898,
      "grad_norm": 0.08597629517316818,
      "learning_rate": 5.578433440754274e-06,
      "loss": 0.0018,
      "step": 94750
    },
    {
      "epoch": 1.5505195123946658,
      "grad_norm": 0.042926348745822906,
      "learning_rate": 5.577487881755456e-06,
      "loss": 0.0011,
      "step": 94760
    },
    {
      "epoch": 1.5506831383457418,
      "grad_norm": 0.11701756715774536,
      "learning_rate": 5.5765423018240884e-06,
      "loss": 0.0012,
      "step": 94770
    },
    {
      "epoch": 1.5508467642968173,
      "grad_norm": 0.19784662127494812,
      "learning_rate": 5.57559670099445e-06,
      "loss": 0.0026,
      "step": 94780
    },
    {
      "epoch": 1.5510103902478933,
      "grad_norm": 0.22977997362613678,
      "learning_rate": 5.5746510793008156e-06,
      "loss": 0.0015,
      "step": 94790
    },
    {
      "epoch": 1.551174016198969,
      "grad_norm": 0.09358534216880798,
      "learning_rate": 5.573705436777459e-06,
      "loss": 0.0028,
      "step": 94800
    },
    {
      "epoch": 1.5513376421500449,
      "grad_norm": 0.17664788663387299,
      "learning_rate": 5.572759773458661e-06,
      "loss": 0.0016,
      "step": 94810
    },
    {
      "epoch": 1.5515012681011209,
      "grad_norm": 0.06913845986127853,
      "learning_rate": 5.571814089378695e-06,
      "loss": 0.0011,
      "step": 94820
    },
    {
      "epoch": 1.5516648940521967,
      "grad_norm": 0.07980062067508698,
      "learning_rate": 5.570868384571845e-06,
      "loss": 0.002,
      "step": 94830
    },
    {
      "epoch": 1.5518285200032724,
      "grad_norm": 0.011418584734201431,
      "learning_rate": 5.5699226590723865e-06,
      "loss": 0.0017,
      "step": 94840
    },
    {
      "epoch": 1.5519921459543484,
      "grad_norm": 0.10197842866182327,
      "learning_rate": 5.568976912914602e-06,
      "loss": 0.002,
      "step": 94850
    },
    {
      "epoch": 1.5521557719054242,
      "grad_norm": 0.048731040209531784,
      "learning_rate": 5.5680311461327716e-06,
      "loss": 0.0018,
      "step": 94860
    },
    {
      "epoch": 1.5523193978565,
      "grad_norm": 0.10829249769449234,
      "learning_rate": 5.567085358761179e-06,
      "loss": 0.0025,
      "step": 94870
    },
    {
      "epoch": 1.552483023807576,
      "grad_norm": 0.05648889020085335,
      "learning_rate": 5.5661395508341035e-06,
      "loss": 0.002,
      "step": 94880
    },
    {
      "epoch": 1.5526466497586517,
      "grad_norm": 0.07775375247001648,
      "learning_rate": 5.565193722385831e-06,
      "loss": 0.0011,
      "step": 94890
    },
    {
      "epoch": 1.5528102757097275,
      "grad_norm": 0.2735661268234253,
      "learning_rate": 5.564247873450643e-06,
      "loss": 0.0034,
      "step": 94900
    },
    {
      "epoch": 1.5529739016608035,
      "grad_norm": 0.11515632271766663,
      "learning_rate": 5.563302004062826e-06,
      "loss": 0.0023,
      "step": 94910
    },
    {
      "epoch": 1.5531375276118793,
      "grad_norm": 0.2695034146308899,
      "learning_rate": 5.562356114256666e-06,
      "loss": 0.0023,
      "step": 94920
    },
    {
      "epoch": 1.553301153562955,
      "grad_norm": 0.13368797302246094,
      "learning_rate": 5.561410204066447e-06,
      "loss": 0.0012,
      "step": 94930
    },
    {
      "epoch": 1.553464779514031,
      "grad_norm": 0.07765199989080429,
      "learning_rate": 5.5604642735264605e-06,
      "loss": 0.001,
      "step": 94940
    },
    {
      "epoch": 1.5536284054651066,
      "grad_norm": 0.040681757032871246,
      "learning_rate": 5.559518322670989e-06,
      "loss": 0.0017,
      "step": 94950
    },
    {
      "epoch": 1.5537920314161826,
      "grad_norm": 0.09414543956518173,
      "learning_rate": 5.558572351534323e-06,
      "loss": 0.0013,
      "step": 94960
    },
    {
      "epoch": 1.5539556573672586,
      "grad_norm": 0.13122491538524628,
      "learning_rate": 5.557626360150752e-06,
      "loss": 0.0013,
      "step": 94970
    },
    {
      "epoch": 1.5541192833183342,
      "grad_norm": 0.030454393476247787,
      "learning_rate": 5.556680348554566e-06,
      "loss": 0.0017,
      "step": 94980
    },
    {
      "epoch": 1.5542829092694102,
      "grad_norm": 0.06736773252487183,
      "learning_rate": 5.555734316780055e-06,
      "loss": 0.0027,
      "step": 94990
    },
    {
      "epoch": 1.554446535220486,
      "grad_norm": 0.05552549287676811,
      "learning_rate": 5.55478826486151e-06,
      "loss": 0.0014,
      "step": 95000
    },
    {
      "epoch": 1.5546101611715617,
      "grad_norm": 0.05441640689969063,
      "learning_rate": 5.553842192833225e-06,
      "loss": 0.0021,
      "step": 95010
    },
    {
      "epoch": 1.5547737871226377,
      "grad_norm": 0.13764795660972595,
      "learning_rate": 5.5528961007294915e-06,
      "loss": 0.0014,
      "step": 95020
    },
    {
      "epoch": 1.5549374130737135,
      "grad_norm": 0.1312810778617859,
      "learning_rate": 5.5519499885846025e-06,
      "loss": 0.0019,
      "step": 95030
    },
    {
      "epoch": 1.5551010390247892,
      "grad_norm": 0.12424931675195694,
      "learning_rate": 5.551003856432853e-06,
      "loss": 0.0011,
      "step": 95040
    },
    {
      "epoch": 1.5552646649758652,
      "grad_norm": 0.3694095015525818,
      "learning_rate": 5.550057704308538e-06,
      "loss": 0.0021,
      "step": 95050
    },
    {
      "epoch": 1.555428290926941,
      "grad_norm": 0.758162260055542,
      "learning_rate": 5.5491115322459545e-06,
      "loss": 0.0015,
      "step": 95060
    },
    {
      "epoch": 1.5555919168780168,
      "grad_norm": 0.07636237889528275,
      "learning_rate": 5.548165340279397e-06,
      "loss": 0.0014,
      "step": 95070
    },
    {
      "epoch": 1.5557555428290928,
      "grad_norm": 0.41262513399124146,
      "learning_rate": 5.547219128443162e-06,
      "loss": 0.0037,
      "step": 95080
    },
    {
      "epoch": 1.5559191687801686,
      "grad_norm": 0.2304043471813202,
      "learning_rate": 5.546272896771552e-06,
      "loss": 0.0016,
      "step": 95090
    },
    {
      "epoch": 1.5560827947312443,
      "grad_norm": 0.05563835799694061,
      "learning_rate": 5.5453266452988596e-06,
      "loss": 0.0019,
      "step": 95100
    },
    {
      "epoch": 1.5562464206823203,
      "grad_norm": 0.07931830734014511,
      "learning_rate": 5.544380374059388e-06,
      "loss": 0.0017,
      "step": 95110
    },
    {
      "epoch": 1.556410046633396,
      "grad_norm": 0.09851199388504028,
      "learning_rate": 5.543434083087434e-06,
      "loss": 0.0018,
      "step": 95120
    },
    {
      "epoch": 1.5565736725844719,
      "grad_norm": 0.2715357840061188,
      "learning_rate": 5.5424877724173035e-06,
      "loss": 0.0018,
      "step": 95130
    },
    {
      "epoch": 1.5567372985355479,
      "grad_norm": 0.18828217685222626,
      "learning_rate": 5.541541442083295e-06,
      "loss": 0.0006,
      "step": 95140
    },
    {
      "epoch": 1.5569009244866234,
      "grad_norm": 0.06957677006721497,
      "learning_rate": 5.540595092119709e-06,
      "loss": 0.0011,
      "step": 95150
    },
    {
      "epoch": 1.5570645504376994,
      "grad_norm": 0.400738000869751,
      "learning_rate": 5.539648722560852e-06,
      "loss": 0.0015,
      "step": 95160
    },
    {
      "epoch": 1.5572281763887754,
      "grad_norm": 0.2115340381860733,
      "learning_rate": 5.538702333441025e-06,
      "loss": 0.0026,
      "step": 95170
    },
    {
      "epoch": 1.557391802339851,
      "grad_norm": 0.020447999238967896,
      "learning_rate": 5.537755924794533e-06,
      "loss": 0.0017,
      "step": 95180
    },
    {
      "epoch": 1.557555428290927,
      "grad_norm": 0.06256935000419617,
      "learning_rate": 5.53680949665568e-06,
      "loss": 0.0014,
      "step": 95190
    },
    {
      "epoch": 1.5577190542420027,
      "grad_norm": 0.14920572936534882,
      "learning_rate": 5.535863049058775e-06,
      "loss": 0.0013,
      "step": 95200
    },
    {
      "epoch": 1.5578826801930785,
      "grad_norm": 0.060673076659440994,
      "learning_rate": 5.534916582038122e-06,
      "loss": 0.0017,
      "step": 95210
    },
    {
      "epoch": 1.5580463061441545,
      "grad_norm": 0.08463137596845627,
      "learning_rate": 5.533970095628029e-06,
      "loss": 0.002,
      "step": 95220
    },
    {
      "epoch": 1.5582099320952303,
      "grad_norm": 0.0693800151348114,
      "learning_rate": 5.5330235898628014e-06,
      "loss": 0.0017,
      "step": 95230
    },
    {
      "epoch": 1.558373558046306,
      "grad_norm": 0.1274455040693283,
      "learning_rate": 5.532077064776752e-06,
      "loss": 0.0018,
      "step": 95240
    },
    {
      "epoch": 1.558537183997382,
      "grad_norm": 0.05534921959042549,
      "learning_rate": 5.531130520404186e-06,
      "loss": 0.001,
      "step": 95250
    },
    {
      "epoch": 1.5587008099484578,
      "grad_norm": 0.13551075756549835,
      "learning_rate": 5.530183956779416e-06,
      "loss": 0.0015,
      "step": 95260
    },
    {
      "epoch": 1.5588644358995336,
      "grad_norm": 0.07281338423490524,
      "learning_rate": 5.529237373936751e-06,
      "loss": 0.0019,
      "step": 95270
    },
    {
      "epoch": 1.5590280618506096,
      "grad_norm": 0.15223409235477448,
      "learning_rate": 5.528290771910504e-06,
      "loss": 0.002,
      "step": 95280
    },
    {
      "epoch": 1.5591916878016854,
      "grad_norm": 0.14520815014839172,
      "learning_rate": 5.527344150734988e-06,
      "loss": 0.0011,
      "step": 95290
    },
    {
      "epoch": 1.5593553137527612,
      "grad_norm": 0.06700566411018372,
      "learning_rate": 5.526397510444511e-06,
      "loss": 0.0021,
      "step": 95300
    },
    {
      "epoch": 1.5595189397038371,
      "grad_norm": 0.07925879210233688,
      "learning_rate": 5.525450851073391e-06,
      "loss": 0.0018,
      "step": 95310
    },
    {
      "epoch": 1.559682565654913,
      "grad_norm": 0.10962279140949249,
      "learning_rate": 5.52450417265594e-06,
      "loss": 0.0012,
      "step": 95320
    },
    {
      "epoch": 1.5598461916059887,
      "grad_norm": 0.19689267873764038,
      "learning_rate": 5.523557475226473e-06,
      "loss": 0.0022,
      "step": 95330
    },
    {
      "epoch": 1.5600098175570647,
      "grad_norm": 0.09281846135854721,
      "learning_rate": 5.522610758819307e-06,
      "loss": 0.0019,
      "step": 95340
    },
    {
      "epoch": 1.5601734435081402,
      "grad_norm": 0.09342874586582184,
      "learning_rate": 5.521664023468756e-06,
      "loss": 0.0034,
      "step": 95350
    },
    {
      "epoch": 1.5603370694592162,
      "grad_norm": 0.279384970664978,
      "learning_rate": 5.5207172692091386e-06,
      "loss": 0.0019,
      "step": 95360
    },
    {
      "epoch": 1.5605006954102922,
      "grad_norm": 0.01354216132313013,
      "learning_rate": 5.519770496074771e-06,
      "loss": 0.0019,
      "step": 95370
    },
    {
      "epoch": 1.5606643213613678,
      "grad_norm": 0.08631009608507156,
      "learning_rate": 5.518823704099973e-06,
      "loss": 0.0021,
      "step": 95380
    },
    {
      "epoch": 1.5608279473124438,
      "grad_norm": 0.04090086370706558,
      "learning_rate": 5.517876893319063e-06,
      "loss": 0.0008,
      "step": 95390
    },
    {
      "epoch": 1.5609915732635196,
      "grad_norm": 0.0927191749215126,
      "learning_rate": 5.516930063766358e-06,
      "loss": 0.0011,
      "step": 95400
    },
    {
      "epoch": 1.5611551992145953,
      "grad_norm": 0.027318453416228294,
      "learning_rate": 5.515983215476182e-06,
      "loss": 0.0024,
      "step": 95410
    },
    {
      "epoch": 1.5613188251656713,
      "grad_norm": 0.15403422713279724,
      "learning_rate": 5.515036348482855e-06,
      "loss": 0.002,
      "step": 95420
    },
    {
      "epoch": 1.561482451116747,
      "grad_norm": 0.17846344411373138,
      "learning_rate": 5.514089462820697e-06,
      "loss": 0.0018,
      "step": 95430
    },
    {
      "epoch": 1.5616460770678229,
      "grad_norm": 0.21921469271183014,
      "learning_rate": 5.513142558524034e-06,
      "loss": 0.0018,
      "step": 95440
    },
    {
      "epoch": 1.5618097030188989,
      "grad_norm": 0.016745688393712044,
      "learning_rate": 5.512195635627185e-06,
      "loss": 0.0015,
      "step": 95450
    },
    {
      "epoch": 1.5619733289699747,
      "grad_norm": 0.017428111284971237,
      "learning_rate": 5.511248694164475e-06,
      "loss": 0.0012,
      "step": 95460
    },
    {
      "epoch": 1.5621369549210504,
      "grad_norm": 0.0534135177731514,
      "learning_rate": 5.510301734170229e-06,
      "loss": 0.0012,
      "step": 95470
    },
    {
      "epoch": 1.5623005808721264,
      "grad_norm": 0.10955236107110977,
      "learning_rate": 5.509354755678771e-06,
      "loss": 0.0015,
      "step": 95480
    },
    {
      "epoch": 1.5624642068232022,
      "grad_norm": 0.13564962148666382,
      "learning_rate": 5.5084077587244275e-06,
      "loss": 0.002,
      "step": 95490
    },
    {
      "epoch": 1.562627832774278,
      "grad_norm": 0.031119294464588165,
      "learning_rate": 5.507460743341526e-06,
      "loss": 0.0021,
      "step": 95500
    },
    {
      "epoch": 1.562791458725354,
      "grad_norm": 0.09961502254009247,
      "learning_rate": 5.506513709564391e-06,
      "loss": 0.0028,
      "step": 95510
    },
    {
      "epoch": 1.5629550846764295,
      "grad_norm": 0.009616827592253685,
      "learning_rate": 5.505566657427352e-06,
      "loss": 0.0013,
      "step": 95520
    },
    {
      "epoch": 1.5631187106275055,
      "grad_norm": 0.015020889230072498,
      "learning_rate": 5.504619586964735e-06,
      "loss": 0.0046,
      "step": 95530
    },
    {
      "epoch": 1.5632823365785815,
      "grad_norm": 0.09533273428678513,
      "learning_rate": 5.5036724982108726e-06,
      "loss": 0.0018,
      "step": 95540
    },
    {
      "epoch": 1.563445962529657,
      "grad_norm": 0.0701201781630516,
      "learning_rate": 5.502725391200092e-06,
      "loss": 0.0015,
      "step": 95550
    },
    {
      "epoch": 1.563609588480733,
      "grad_norm": 0.1311163306236267,
      "learning_rate": 5.5017782659667244e-06,
      "loss": 0.0019,
      "step": 95560
    },
    {
      "epoch": 1.5637732144318088,
      "grad_norm": 0.13136912882328033,
      "learning_rate": 5.5008311225451e-06,
      "loss": 0.0011,
      "step": 95570
    },
    {
      "epoch": 1.5639368403828846,
      "grad_norm": 0.08102806657552719,
      "learning_rate": 5.499883960969552e-06,
      "loss": 0.0009,
      "step": 95580
    },
    {
      "epoch": 1.5641004663339606,
      "grad_norm": 0.04429985582828522,
      "learning_rate": 5.498936781274414e-06,
      "loss": 0.0015,
      "step": 95590
    },
    {
      "epoch": 1.5642640922850364,
      "grad_norm": 0.062323760241270065,
      "learning_rate": 5.497989583494013e-06,
      "loss": 0.0026,
      "step": 95600
    },
    {
      "epoch": 1.5644277182361122,
      "grad_norm": 0.12206123024225235,
      "learning_rate": 5.497042367662688e-06,
      "loss": 0.0009,
      "step": 95610
    },
    {
      "epoch": 1.5645913441871881,
      "grad_norm": 0.11871977895498276,
      "learning_rate": 5.496095133814772e-06,
      "loss": 0.0014,
      "step": 95620
    },
    {
      "epoch": 1.564754970138264,
      "grad_norm": 0.04354385659098625,
      "learning_rate": 5.495147881984599e-06,
      "loss": 0.0017,
      "step": 95630
    },
    {
      "epoch": 1.5649185960893397,
      "grad_norm": 0.1686330884695053,
      "learning_rate": 5.4942006122065074e-06,
      "loss": 0.0017,
      "step": 95640
    },
    {
      "epoch": 1.5650822220404157,
      "grad_norm": 0.09138898551464081,
      "learning_rate": 5.493253324514829e-06,
      "loss": 0.0021,
      "step": 95650
    },
    {
      "epoch": 1.5652458479914915,
      "grad_norm": 0.02709484100341797,
      "learning_rate": 5.492306018943903e-06,
      "loss": 0.0026,
      "step": 95660
    },
    {
      "epoch": 1.5654094739425672,
      "grad_norm": 0.10914696753025055,
      "learning_rate": 5.4913586955280685e-06,
      "loss": 0.0014,
      "step": 95670
    },
    {
      "epoch": 1.5655730998936432,
      "grad_norm": 0.019028667360544205,
      "learning_rate": 5.490411354301663e-06,
      "loss": 0.0017,
      "step": 95680
    },
    {
      "epoch": 1.565736725844719,
      "grad_norm": 0.28978875279426575,
      "learning_rate": 5.489463995299023e-06,
      "loss": 0.0028,
      "step": 95690
    },
    {
      "epoch": 1.5659003517957948,
      "grad_norm": 0.061479244381189346,
      "learning_rate": 5.488516618554489e-06,
      "loss": 0.0012,
      "step": 95700
    },
    {
      "epoch": 1.5660639777468708,
      "grad_norm": 0.14110243320465088,
      "learning_rate": 5.487569224102403e-06,
      "loss": 0.0017,
      "step": 95710
    },
    {
      "epoch": 1.5662276036979463,
      "grad_norm": 0.03730341047048569,
      "learning_rate": 5.486621811977103e-06,
      "loss": 0.0022,
      "step": 95720
    },
    {
      "epoch": 1.5663912296490223,
      "grad_norm": 0.15115883946418762,
      "learning_rate": 5.485674382212933e-06,
      "loss": 0.0025,
      "step": 95730
    },
    {
      "epoch": 1.5665548556000983,
      "grad_norm": 0.03132474049925804,
      "learning_rate": 5.484726934844236e-06,
      "loss": 0.001,
      "step": 95740
    },
    {
      "epoch": 1.5667184815511739,
      "grad_norm": 0.0572274886071682,
      "learning_rate": 5.483779469905349e-06,
      "loss": 0.0014,
      "step": 95750
    },
    {
      "epoch": 1.5668821075022499,
      "grad_norm": 0.07731965184211731,
      "learning_rate": 5.48283198743062e-06,
      "loss": 0.0013,
      "step": 95760
    },
    {
      "epoch": 1.5670457334533257,
      "grad_norm": 0.026288717985153198,
      "learning_rate": 5.481884487454392e-06,
      "loss": 0.0015,
      "step": 95770
    },
    {
      "epoch": 1.5672093594044014,
      "grad_norm": 0.16183966398239136,
      "learning_rate": 5.48093697001101e-06,
      "loss": 0.0025,
      "step": 95780
    },
    {
      "epoch": 1.5673729853554774,
      "grad_norm": 0.068525031208992,
      "learning_rate": 5.479989435134818e-06,
      "loss": 0.0022,
      "step": 95790
    },
    {
      "epoch": 1.5675366113065532,
      "grad_norm": 0.10348375886678696,
      "learning_rate": 5.479041882860164e-06,
      "loss": 0.0013,
      "step": 95800
    },
    {
      "epoch": 1.567700237257629,
      "grad_norm": 0.05617411434650421,
      "learning_rate": 5.4780943132213925e-06,
      "loss": 0.0017,
      "step": 95810
    },
    {
      "epoch": 1.567863863208705,
      "grad_norm": 0.3189500570297241,
      "learning_rate": 5.477146726252851e-06,
      "loss": 0.0019,
      "step": 95820
    },
    {
      "epoch": 1.5680274891597807,
      "grad_norm": 0.011146112345159054,
      "learning_rate": 5.476199121988888e-06,
      "loss": 0.0021,
      "step": 95830
    },
    {
      "epoch": 1.5681911151108565,
      "grad_norm": 0.06527738273143768,
      "learning_rate": 5.475251500463849e-06,
      "loss": 0.0014,
      "step": 95840
    },
    {
      "epoch": 1.5683547410619325,
      "grad_norm": 0.12300140410661697,
      "learning_rate": 5.474303861712088e-06,
      "loss": 0.0031,
      "step": 95850
    },
    {
      "epoch": 1.5685183670130083,
      "grad_norm": 0.08339247107505798,
      "learning_rate": 5.473356205767952e-06,
      "loss": 0.0025,
      "step": 95860
    },
    {
      "epoch": 1.568681992964084,
      "grad_norm": 0.05605313554406166,
      "learning_rate": 5.4724085326657894e-06,
      "loss": 0.0012,
      "step": 95870
    },
    {
      "epoch": 1.56884561891516,
      "grad_norm": 0.053953222930431366,
      "learning_rate": 5.471460842439954e-06,
      "loss": 0.0024,
      "step": 95880
    },
    {
      "epoch": 1.5690092448662358,
      "grad_norm": 0.07839611172676086,
      "learning_rate": 5.470513135124796e-06,
      "loss": 0.0012,
      "step": 95890
    },
    {
      "epoch": 1.5691728708173116,
      "grad_norm": 0.09375961869955063,
      "learning_rate": 5.469565410754668e-06,
      "loss": 0.0018,
      "step": 95900
    },
    {
      "epoch": 1.5693364967683876,
      "grad_norm": 0.10395301133394241,
      "learning_rate": 5.468617669363922e-06,
      "loss": 0.002,
      "step": 95910
    },
    {
      "epoch": 1.5695001227194632,
      "grad_norm": 0.18192291259765625,
      "learning_rate": 5.467669910986912e-06,
      "loss": 0.0016,
      "step": 95920
    },
    {
      "epoch": 1.5696637486705391,
      "grad_norm": 0.13954170048236847,
      "learning_rate": 5.466722135657991e-06,
      "loss": 0.002,
      "step": 95930
    },
    {
      "epoch": 1.5698273746216151,
      "grad_norm": 0.0689585879445076,
      "learning_rate": 5.465774343411515e-06,
      "loss": 0.002,
      "step": 95940
    },
    {
      "epoch": 1.5699910005726907,
      "grad_norm": 0.054865192621946335,
      "learning_rate": 5.46482653428184e-06,
      "loss": 0.0013,
      "step": 95950
    },
    {
      "epoch": 1.5701546265237667,
      "grad_norm": 0.22509394586086273,
      "learning_rate": 5.463878708303319e-06,
      "loss": 0.0021,
      "step": 95960
    },
    {
      "epoch": 1.5703182524748425,
      "grad_norm": 0.005309760104864836,
      "learning_rate": 5.462930865510309e-06,
      "loss": 0.0028,
      "step": 95970
    },
    {
      "epoch": 1.5704818784259182,
      "grad_norm": 0.11050279438495636,
      "learning_rate": 5.461983005937167e-06,
      "loss": 0.0021,
      "step": 95980
    },
    {
      "epoch": 1.5706455043769942,
      "grad_norm": 0.14140237867832184,
      "learning_rate": 5.4610351296182525e-06,
      "loss": 0.0008,
      "step": 95990
    },
    {
      "epoch": 1.57080913032807,
      "grad_norm": 0.1602981686592102,
      "learning_rate": 5.460087236587924e-06,
      "loss": 0.002,
      "step": 96000
    },
    {
      "epoch": 1.5709727562791458,
      "grad_norm": 0.8511959314346313,
      "learning_rate": 5.459139326880538e-06,
      "loss": 0.0032,
      "step": 96010
    },
    {
      "epoch": 1.5711363822302218,
      "grad_norm": 0.31806570291519165,
      "learning_rate": 5.458191400530453e-06,
      "loss": 0.0027,
      "step": 96020
    },
    {
      "epoch": 1.5713000081812976,
      "grad_norm": 0.05419665202498436,
      "learning_rate": 5.457243457572033e-06,
      "loss": 0.004,
      "step": 96030
    },
    {
      "epoch": 1.5714636341323733,
      "grad_norm": 0.01707945205271244,
      "learning_rate": 5.456295498039636e-06,
      "loss": 0.0022,
      "step": 96040
    },
    {
      "epoch": 1.5716272600834493,
      "grad_norm": 0.04216013103723526,
      "learning_rate": 5.455347521967624e-06,
      "loss": 0.001,
      "step": 96050
    },
    {
      "epoch": 1.571790886034525,
      "grad_norm": 0.09771130979061127,
      "learning_rate": 5.4543995293903595e-06,
      "loss": 0.001,
      "step": 96060
    },
    {
      "epoch": 1.5719545119856009,
      "grad_norm": 0.03456299379467964,
      "learning_rate": 5.453451520342204e-06,
      "loss": 0.0014,
      "step": 96070
    },
    {
      "epoch": 1.5721181379366769,
      "grad_norm": 0.22416889667510986,
      "learning_rate": 5.452503494857519e-06,
      "loss": 0.0033,
      "step": 96080
    },
    {
      "epoch": 1.5722817638877526,
      "grad_norm": 0.05665203183889389,
      "learning_rate": 5.451555452970671e-06,
      "loss": 0.0012,
      "step": 96090
    },
    {
      "epoch": 1.5724453898388284,
      "grad_norm": 0.018297085538506508,
      "learning_rate": 5.450607394716025e-06,
      "loss": 0.0009,
      "step": 96100
    },
    {
      "epoch": 1.5726090157899044,
      "grad_norm": 0.07933646440505981,
      "learning_rate": 5.449659320127942e-06,
      "loss": 0.0016,
      "step": 96110
    },
    {
      "epoch": 1.57277264174098,
      "grad_norm": 0.03373287245631218,
      "learning_rate": 5.448711229240789e-06,
      "loss": 0.0011,
      "step": 96120
    },
    {
      "epoch": 1.572936267692056,
      "grad_norm": 0.13468582928180695,
      "learning_rate": 5.447763122088933e-06,
      "loss": 0.003,
      "step": 96130
    },
    {
      "epoch": 1.573099893643132,
      "grad_norm": 0.09387189894914627,
      "learning_rate": 5.446814998706739e-06,
      "loss": 0.0016,
      "step": 96140
    },
    {
      "epoch": 1.5732635195942075,
      "grad_norm": 0.05871766805648804,
      "learning_rate": 5.445866859128578e-06,
      "loss": 0.0009,
      "step": 96150
    },
    {
      "epoch": 1.5734271455452835,
      "grad_norm": 0.035517919808626175,
      "learning_rate": 5.444918703388813e-06,
      "loss": 0.0025,
      "step": 96160
    },
    {
      "epoch": 1.5735907714963593,
      "grad_norm": 0.23661324381828308,
      "learning_rate": 5.443970531521814e-06,
      "loss": 0.0019,
      "step": 96170
    },
    {
      "epoch": 1.573754397447435,
      "grad_norm": 0.1736181229352951,
      "learning_rate": 5.443022343561951e-06,
      "loss": 0.0021,
      "step": 96180
    },
    {
      "epoch": 1.573918023398511,
      "grad_norm": 0.02907741256058216,
      "learning_rate": 5.442074139543593e-06,
      "loss": 0.0021,
      "step": 96190
    },
    {
      "epoch": 1.5740816493495868,
      "grad_norm": 0.1024564579129219,
      "learning_rate": 5.441125919501109e-06,
      "loss": 0.0025,
      "step": 96200
    },
    {
      "epoch": 1.5742452753006626,
      "grad_norm": 0.13733956217765808,
      "learning_rate": 5.440177683468871e-06,
      "loss": 0.0018,
      "step": 96210
    },
    {
      "epoch": 1.5744089012517386,
      "grad_norm": 0.0663224309682846,
      "learning_rate": 5.43922943148125e-06,
      "loss": 0.002,
      "step": 96220
    },
    {
      "epoch": 1.5745725272028144,
      "grad_norm": 0.07897910475730896,
      "learning_rate": 5.438281163572618e-06,
      "loss": 0.0015,
      "step": 96230
    },
    {
      "epoch": 1.5747361531538902,
      "grad_norm": 0.08347785472869873,
      "learning_rate": 5.437332879777346e-06,
      "loss": 0.0017,
      "step": 96240
    },
    {
      "epoch": 1.5748997791049661,
      "grad_norm": 0.2307664155960083,
      "learning_rate": 5.43638458012981e-06,
      "loss": 0.0018,
      "step": 96250
    },
    {
      "epoch": 1.575063405056042,
      "grad_norm": 0.009943345561623573,
      "learning_rate": 5.435436264664381e-06,
      "loss": 0.0016,
      "step": 96260
    },
    {
      "epoch": 1.5752270310071177,
      "grad_norm": 0.09883014112710953,
      "learning_rate": 5.434487933415434e-06,
      "loss": 0.0014,
      "step": 96270
    },
    {
      "epoch": 1.5753906569581937,
      "grad_norm": 0.25964921712875366,
      "learning_rate": 5.433539586417343e-06,
      "loss": 0.0017,
      "step": 96280
    },
    {
      "epoch": 1.5755542829092692,
      "grad_norm": 0.14060312509536743,
      "learning_rate": 5.432591223704484e-06,
      "loss": 0.0024,
      "step": 96290
    },
    {
      "epoch": 1.5757179088603452,
      "grad_norm": 0.13465207815170288,
      "learning_rate": 5.431642845311235e-06,
      "loss": 0.0024,
      "step": 96300
    },
    {
      "epoch": 1.5758815348114212,
      "grad_norm": 0.08285928517580032,
      "learning_rate": 5.4306944512719686e-06,
      "loss": 0.0014,
      "step": 96310
    },
    {
      "epoch": 1.5760451607624968,
      "grad_norm": 0.05067675560712814,
      "learning_rate": 5.4297460416210635e-06,
      "loss": 0.0045,
      "step": 96320
    },
    {
      "epoch": 1.5762087867135728,
      "grad_norm": 0.09514006227254868,
      "learning_rate": 5.4287976163928965e-06,
      "loss": 0.0008,
      "step": 96330
    },
    {
      "epoch": 1.5763724126646486,
      "grad_norm": 0.04533170163631439,
      "learning_rate": 5.427849175621848e-06,
      "loss": 0.0016,
      "step": 96340
    },
    {
      "epoch": 1.5765360386157243,
      "grad_norm": 0.03759074956178665,
      "learning_rate": 5.426900719342295e-06,
      "loss": 0.0016,
      "step": 96350
    },
    {
      "epoch": 1.5766996645668003,
      "grad_norm": 0.07338988035917282,
      "learning_rate": 5.425952247588616e-06,
      "loss": 0.0012,
      "step": 96360
    },
    {
      "epoch": 1.576863290517876,
      "grad_norm": 0.08183891326189041,
      "learning_rate": 5.425003760395192e-06,
      "loss": 0.0027,
      "step": 96370
    },
    {
      "epoch": 1.5770269164689519,
      "grad_norm": 0.1686449944972992,
      "learning_rate": 5.424055257796404e-06,
      "loss": 0.0035,
      "step": 96380
    },
    {
      "epoch": 1.5771905424200279,
      "grad_norm": 0.14980323612689972,
      "learning_rate": 5.423106739826631e-06,
      "loss": 0.0012,
      "step": 96390
    },
    {
      "epoch": 1.5773541683711036,
      "grad_norm": 0.1996951550245285,
      "learning_rate": 5.4221582065202575e-06,
      "loss": 0.0021,
      "step": 96400
    },
    {
      "epoch": 1.5775177943221794,
      "grad_norm": 0.29163432121276855,
      "learning_rate": 5.421209657911661e-06,
      "loss": 0.0024,
      "step": 96410
    },
    {
      "epoch": 1.5776814202732554,
      "grad_norm": 0.12521252036094666,
      "learning_rate": 5.420261094035228e-06,
      "loss": 0.0026,
      "step": 96420
    },
    {
      "epoch": 1.5778450462243312,
      "grad_norm": 0.013463077135384083,
      "learning_rate": 5.419312514925341e-06,
      "loss": 0.0046,
      "step": 96430
    },
    {
      "epoch": 1.578008672175407,
      "grad_norm": 0.048513565212488174,
      "learning_rate": 5.418363920616382e-06,
      "loss": 0.0021,
      "step": 96440
    },
    {
      "epoch": 1.578172298126483,
      "grad_norm": 0.00983465276658535,
      "learning_rate": 5.417415311142739e-06,
      "loss": 0.0016,
      "step": 96450
    },
    {
      "epoch": 1.5783359240775587,
      "grad_norm": 0.16053526103496552,
      "learning_rate": 5.416466686538792e-06,
      "loss": 0.0018,
      "step": 96460
    },
    {
      "epoch": 1.5784995500286345,
      "grad_norm": 0.03312075138092041,
      "learning_rate": 5.4155180468389265e-06,
      "loss": 0.0026,
      "step": 96470
    },
    {
      "epoch": 1.5786631759797105,
      "grad_norm": 0.10898065567016602,
      "learning_rate": 5.414569392077533e-06,
      "loss": 0.0016,
      "step": 96480
    },
    {
      "epoch": 1.578826801930786,
      "grad_norm": 0.1669439673423767,
      "learning_rate": 5.413620722288995e-06,
      "loss": 0.0026,
      "step": 96490
    },
    {
      "epoch": 1.578990427881862,
      "grad_norm": 0.07796142995357513,
      "learning_rate": 5.412672037507701e-06,
      "loss": 0.0011,
      "step": 96500
    },
    {
      "epoch": 1.579154053832938,
      "grad_norm": 0.031167810782790184,
      "learning_rate": 5.411723337768037e-06,
      "loss": 0.0014,
      "step": 96510
    },
    {
      "epoch": 1.5793176797840136,
      "grad_norm": 0.05776277929544449,
      "learning_rate": 5.410774623104391e-06,
      "loss": 0.0017,
      "step": 96520
    },
    {
      "epoch": 1.5794813057350896,
      "grad_norm": 0.0943605899810791,
      "learning_rate": 5.409825893551152e-06,
      "loss": 0.0016,
      "step": 96530
    },
    {
      "epoch": 1.5796449316861654,
      "grad_norm": 0.05084320157766342,
      "learning_rate": 5.40887714914271e-06,
      "loss": 0.0022,
      "step": 96540
    },
    {
      "epoch": 1.5798085576372412,
      "grad_norm": 0.13700053095817566,
      "learning_rate": 5.407928389913453e-06,
      "loss": 0.0018,
      "step": 96550
    },
    {
      "epoch": 1.5799721835883171,
      "grad_norm": 0.18050524592399597,
      "learning_rate": 5.406979615897773e-06,
      "loss": 0.0011,
      "step": 96560
    },
    {
      "epoch": 1.580135809539393,
      "grad_norm": 0.2934496998786926,
      "learning_rate": 5.40603082713006e-06,
      "loss": 0.0018,
      "step": 96570
    },
    {
      "epoch": 1.5802994354904687,
      "grad_norm": 0.056892964988946915,
      "learning_rate": 5.405082023644705e-06,
      "loss": 0.0012,
      "step": 96580
    },
    {
      "epoch": 1.5804630614415447,
      "grad_norm": 0.14180442690849304,
      "learning_rate": 5.404133205476101e-06,
      "loss": 0.0018,
      "step": 96590
    },
    {
      "epoch": 1.5806266873926205,
      "grad_norm": 0.16149818897247314,
      "learning_rate": 5.403184372658641e-06,
      "loss": 0.0008,
      "step": 96600
    },
    {
      "epoch": 1.5807903133436962,
      "grad_norm": 0.24570034444332123,
      "learning_rate": 5.402235525226713e-06,
      "loss": 0.0019,
      "step": 96610
    },
    {
      "epoch": 1.5809539392947722,
      "grad_norm": 0.07833485305309296,
      "learning_rate": 5.401286663214715e-06,
      "loss": 0.0017,
      "step": 96620
    },
    {
      "epoch": 1.581117565245848,
      "grad_norm": 0.08162520080804825,
      "learning_rate": 5.400337786657042e-06,
      "loss": 0.0011,
      "step": 96630
    },
    {
      "epoch": 1.5812811911969238,
      "grad_norm": 0.10021688044071198,
      "learning_rate": 5.399388895588084e-06,
      "loss": 0.0012,
      "step": 96640
    },
    {
      "epoch": 1.5814448171479998,
      "grad_norm": 0.0414818711578846,
      "learning_rate": 5.398439990042241e-06,
      "loss": 0.0014,
      "step": 96650
    },
    {
      "epoch": 1.5816084430990756,
      "grad_norm": 0.06889066100120544,
      "learning_rate": 5.397491070053904e-06,
      "loss": 0.0013,
      "step": 96660
    },
    {
      "epoch": 1.5817720690501513,
      "grad_norm": 0.21186897158622742,
      "learning_rate": 5.396542135657472e-06,
      "loss": 0.0016,
      "step": 96670
    },
    {
      "epoch": 1.5819356950012273,
      "grad_norm": 0.04048130661249161,
      "learning_rate": 5.39559318688734e-06,
      "loss": 0.0034,
      "step": 96680
    },
    {
      "epoch": 1.5820993209523029,
      "grad_norm": 0.17197616398334503,
      "learning_rate": 5.394644223777905e-06,
      "loss": 0.0033,
      "step": 96690
    },
    {
      "epoch": 1.5822629469033789,
      "grad_norm": 0.07203453034162521,
      "learning_rate": 5.393695246363567e-06,
      "loss": 0.0019,
      "step": 96700
    },
    {
      "epoch": 1.5824265728544549,
      "grad_norm": 0.05323415249586105,
      "learning_rate": 5.392746254678723e-06,
      "loss": 0.0015,
      "step": 96710
    },
    {
      "epoch": 1.5825901988055304,
      "grad_norm": 0.1038953885436058,
      "learning_rate": 5.39179724875777e-06,
      "loss": 0.0041,
      "step": 96720
    },
    {
      "epoch": 1.5827538247566064,
      "grad_norm": 0.04694046080112457,
      "learning_rate": 5.390848228635108e-06,
      "loss": 0.0014,
      "step": 96730
    },
    {
      "epoch": 1.5829174507076822,
      "grad_norm": 0.39353007078170776,
      "learning_rate": 5.389899194345137e-06,
      "loss": 0.0019,
      "step": 96740
    },
    {
      "epoch": 1.583081076658758,
      "grad_norm": 0.1378042846918106,
      "learning_rate": 5.388950145922258e-06,
      "loss": 0.0011,
      "step": 96750
    },
    {
      "epoch": 1.583244702609834,
      "grad_norm": 0.06501578539609909,
      "learning_rate": 5.388001083400871e-06,
      "loss": 0.0021,
      "step": 96760
    },
    {
      "epoch": 1.5834083285609097,
      "grad_norm": 0.11308283358812332,
      "learning_rate": 5.387052006815378e-06,
      "loss": 0.0013,
      "step": 96770
    },
    {
      "epoch": 1.5835719545119855,
      "grad_norm": 0.3216404318809509,
      "learning_rate": 5.386102916200178e-06,
      "loss": 0.001,
      "step": 96780
    },
    {
      "epoch": 1.5837355804630615,
      "grad_norm": 0.0115175386890769,
      "learning_rate": 5.385153811589677e-06,
      "loss": 0.0011,
      "step": 96790
    },
    {
      "epoch": 1.5838992064141373,
      "grad_norm": 0.12202366441488266,
      "learning_rate": 5.3842046930182765e-06,
      "loss": 0.0014,
      "step": 96800
    },
    {
      "epoch": 1.584062832365213,
      "grad_norm": 0.04255205765366554,
      "learning_rate": 5.3832555605203784e-06,
      "loss": 0.0014,
      "step": 96810
    },
    {
      "epoch": 1.584226458316289,
      "grad_norm": 0.44954219460487366,
      "learning_rate": 5.382306414130387e-06,
      "loss": 0.002,
      "step": 96820
    },
    {
      "epoch": 1.5843900842673648,
      "grad_norm": 0.09452875703573227,
      "learning_rate": 5.381357253882706e-06,
      "loss": 0.0008,
      "step": 96830
    },
    {
      "epoch": 1.5845537102184406,
      "grad_norm": 0.08879400044679642,
      "learning_rate": 5.380408079811742e-06,
      "loss": 0.0015,
      "step": 96840
    },
    {
      "epoch": 1.5847173361695166,
      "grad_norm": 0.02210010215640068,
      "learning_rate": 5.379458891951901e-06,
      "loss": 0.0016,
      "step": 96850
    },
    {
      "epoch": 1.5848809621205924,
      "grad_norm": 0.06366530805826187,
      "learning_rate": 5.3785096903375854e-06,
      "loss": 0.0007,
      "step": 96860
    },
    {
      "epoch": 1.5850445880716681,
      "grad_norm": 0.11315908282995224,
      "learning_rate": 5.377560475003204e-06,
      "loss": 0.0019,
      "step": 96870
    },
    {
      "epoch": 1.5852082140227441,
      "grad_norm": 0.13750480115413666,
      "learning_rate": 5.3766112459831625e-06,
      "loss": 0.001,
      "step": 96880
    },
    {
      "epoch": 1.5853718399738197,
      "grad_norm": 0.04556933417916298,
      "learning_rate": 5.375662003311869e-06,
      "loss": 0.0016,
      "step": 96890
    },
    {
      "epoch": 1.5855354659248957,
      "grad_norm": 0.0854511708021164,
      "learning_rate": 5.374712747023729e-06,
      "loss": 0.001,
      "step": 96900
    },
    {
      "epoch": 1.5856990918759717,
      "grad_norm": 0.09421778470277786,
      "learning_rate": 5.373763477153154e-06,
      "loss": 0.0019,
      "step": 96910
    },
    {
      "epoch": 1.5858627178270472,
      "grad_norm": 0.04897712916135788,
      "learning_rate": 5.372814193734551e-06,
      "loss": 0.001,
      "step": 96920
    },
    {
      "epoch": 1.5860263437781232,
      "grad_norm": 0.11712972074747086,
      "learning_rate": 5.37186489680233e-06,
      "loss": 0.0027,
      "step": 96930
    },
    {
      "epoch": 1.586189969729199,
      "grad_norm": 0.08070344477891922,
      "learning_rate": 5.3709155863909e-06,
      "loss": 0.0015,
      "step": 96940
    },
    {
      "epoch": 1.5863535956802748,
      "grad_norm": 0.13983389735221863,
      "learning_rate": 5.369966262534672e-06,
      "loss": 0.002,
      "step": 96950
    },
    {
      "epoch": 1.5865172216313508,
      "grad_norm": 0.10265609622001648,
      "learning_rate": 5.3690169252680555e-06,
      "loss": 0.001,
      "step": 96960
    },
    {
      "epoch": 1.5866808475824266,
      "grad_norm": 0.03303215280175209,
      "learning_rate": 5.368067574625461e-06,
      "loss": 0.0018,
      "step": 96970
    },
    {
      "epoch": 1.5868444735335023,
      "grad_norm": 0.05763198807835579,
      "learning_rate": 5.3671182106413035e-06,
      "loss": 0.0019,
      "step": 96980
    },
    {
      "epoch": 1.5870080994845783,
      "grad_norm": 0.3887484669685364,
      "learning_rate": 5.366168833349992e-06,
      "loss": 0.0052,
      "step": 96990
    },
    {
      "epoch": 1.587171725435654,
      "grad_norm": 0.10816752165555954,
      "learning_rate": 5.3652194427859414e-06,
      "loss": 0.0018,
      "step": 97000
    },
    {
      "epoch": 1.5873353513867299,
      "grad_norm": 0.12773962318897247,
      "learning_rate": 5.364270038983564e-06,
      "loss": 0.0014,
      "step": 97010
    },
    {
      "epoch": 1.5874989773378059,
      "grad_norm": 0.11422409862279892,
      "learning_rate": 5.363320621977273e-06,
      "loss": 0.0021,
      "step": 97020
    },
    {
      "epoch": 1.5876626032888816,
      "grad_norm": 0.17639519274234772,
      "learning_rate": 5.362371191801482e-06,
      "loss": 0.0018,
      "step": 97030
    },
    {
      "epoch": 1.5878262292399574,
      "grad_norm": 0.10923583805561066,
      "learning_rate": 5.361421748490606e-06,
      "loss": 0.0015,
      "step": 97040
    },
    {
      "epoch": 1.5879898551910334,
      "grad_norm": 0.032507553696632385,
      "learning_rate": 5.360472292079061e-06,
      "loss": 0.0013,
      "step": 97050
    },
    {
      "epoch": 1.5881534811421092,
      "grad_norm": 0.04594506695866585,
      "learning_rate": 5.3595228226012605e-06,
      "loss": 0.0019,
      "step": 97060
    },
    {
      "epoch": 1.588317107093185,
      "grad_norm": 0.10563760250806808,
      "learning_rate": 5.358573340091623e-06,
      "loss": 0.0017,
      "step": 97070
    },
    {
      "epoch": 1.588480733044261,
      "grad_norm": 0.026333678513765335,
      "learning_rate": 5.357623844584564e-06,
      "loss": 0.0018,
      "step": 97080
    },
    {
      "epoch": 1.5886443589953365,
      "grad_norm": 0.060836926102638245,
      "learning_rate": 5.356674336114499e-06,
      "loss": 0.0023,
      "step": 97090
    },
    {
      "epoch": 1.5888079849464125,
      "grad_norm": 0.05666762590408325,
      "learning_rate": 5.355724814715848e-06,
      "loss": 0.0015,
      "step": 97100
    },
    {
      "epoch": 1.5889716108974885,
      "grad_norm": 0.3019115924835205,
      "learning_rate": 5.354775280423026e-06,
      "loss": 0.0034,
      "step": 97110
    },
    {
      "epoch": 1.589135236848564,
      "grad_norm": 0.10878598690032959,
      "learning_rate": 5.353825733270454e-06,
      "loss": 0.0014,
      "step": 97120
    },
    {
      "epoch": 1.58929886279964,
      "grad_norm": 0.08829003572463989,
      "learning_rate": 5.352876173292548e-06,
      "loss": 0.0019,
      "step": 97130
    },
    {
      "epoch": 1.5894624887507158,
      "grad_norm": 0.19793982803821564,
      "learning_rate": 5.35192660052373e-06,
      "loss": 0.0022,
      "step": 97140
    },
    {
      "epoch": 1.5896261147017916,
      "grad_norm": 0.10565929859876633,
      "learning_rate": 5.35097701499842e-06,
      "loss": 0.0011,
      "step": 97150
    },
    {
      "epoch": 1.5897897406528676,
      "grad_norm": 0.10133195668458939,
      "learning_rate": 5.350027416751034e-06,
      "loss": 0.0017,
      "step": 97160
    },
    {
      "epoch": 1.5899533666039434,
      "grad_norm": 0.05131659284234047,
      "learning_rate": 5.349077805815997e-06,
      "loss": 0.0023,
      "step": 97170
    },
    {
      "epoch": 1.5901169925550191,
      "grad_norm": 0.04218674078583717,
      "learning_rate": 5.348128182227728e-06,
      "loss": 0.0019,
      "step": 97180
    },
    {
      "epoch": 1.5902806185060951,
      "grad_norm": 0.08554338663816452,
      "learning_rate": 5.347178546020648e-06,
      "loss": 0.0017,
      "step": 97190
    },
    {
      "epoch": 1.590444244457171,
      "grad_norm": 0.06378992646932602,
      "learning_rate": 5.346228897229182e-06,
      "loss": 0.0017,
      "step": 97200
    },
    {
      "epoch": 1.5906078704082467,
      "grad_norm": 0.10118988901376724,
      "learning_rate": 5.34527923588775e-06,
      "loss": 0.001,
      "step": 97210
    },
    {
      "epoch": 1.5907714963593227,
      "grad_norm": 0.060810595750808716,
      "learning_rate": 5.344329562030775e-06,
      "loss": 0.0011,
      "step": 97220
    },
    {
      "epoch": 1.5909351223103985,
      "grad_norm": 0.07332179695367813,
      "learning_rate": 5.343379875692681e-06,
      "loss": 0.0064,
      "step": 97230
    },
    {
      "epoch": 1.5910987482614742,
      "grad_norm": 0.08635635673999786,
      "learning_rate": 5.342430176907893e-06,
      "loss": 0.0017,
      "step": 97240
    },
    {
      "epoch": 1.5912623742125502,
      "grad_norm": 0.1599757820367813,
      "learning_rate": 5.341480465710834e-06,
      "loss": 0.0014,
      "step": 97250
    },
    {
      "epoch": 1.5914260001636258,
      "grad_norm": 0.12012854218482971,
      "learning_rate": 5.3405307421359275e-06,
      "loss": 0.0013,
      "step": 97260
    },
    {
      "epoch": 1.5915896261147018,
      "grad_norm": 0.08608757704496384,
      "learning_rate": 5.3395810062176e-06,
      "loss": 0.0031,
      "step": 97270
    },
    {
      "epoch": 1.5917532520657778,
      "grad_norm": 0.03763792663812637,
      "learning_rate": 5.338631257990279e-06,
      "loss": 0.0009,
      "step": 97280
    },
    {
      "epoch": 1.5919168780168533,
      "grad_norm": 0.07741168886423111,
      "learning_rate": 5.337681497488387e-06,
      "loss": 0.001,
      "step": 97290
    },
    {
      "epoch": 1.5920805039679293,
      "grad_norm": 0.09594119340181351,
      "learning_rate": 5.336731724746353e-06,
      "loss": 0.0014,
      "step": 97300
    },
    {
      "epoch": 1.592244129919005,
      "grad_norm": 0.09797827899456024,
      "learning_rate": 5.335781939798604e-06,
      "loss": 0.0016,
      "step": 97310
    },
    {
      "epoch": 1.5924077558700809,
      "grad_norm": 0.015874236822128296,
      "learning_rate": 5.3348321426795644e-06,
      "loss": 0.0024,
      "step": 97320
    },
    {
      "epoch": 1.5925713818211569,
      "grad_norm": 0.09323128312826157,
      "learning_rate": 5.3338823334236665e-06,
      "loss": 0.0019,
      "step": 97330
    },
    {
      "epoch": 1.5927350077722326,
      "grad_norm": 0.15608489513397217,
      "learning_rate": 5.3329325120653355e-06,
      "loss": 0.0025,
      "step": 97340
    },
    {
      "epoch": 1.5928986337233084,
      "grad_norm": 0.036289140582084656,
      "learning_rate": 5.3319826786390025e-06,
      "loss": 0.0014,
      "step": 97350
    },
    {
      "epoch": 1.5930622596743844,
      "grad_norm": 0.0709943175315857,
      "learning_rate": 5.331032833179095e-06,
      "loss": 0.0014,
      "step": 97360
    },
    {
      "epoch": 1.5932258856254602,
      "grad_norm": 0.06235069781541824,
      "learning_rate": 5.330082975720041e-06,
      "loss": 0.0009,
      "step": 97370
    },
    {
      "epoch": 1.593389511576536,
      "grad_norm": 0.07837652415037155,
      "learning_rate": 5.329133106296275e-06,
      "loss": 0.0014,
      "step": 97380
    },
    {
      "epoch": 1.593553137527612,
      "grad_norm": 0.05847574770450592,
      "learning_rate": 5.328183224942223e-06,
      "loss": 0.0011,
      "step": 97390
    },
    {
      "epoch": 1.5937167634786877,
      "grad_norm": 0.04407477378845215,
      "learning_rate": 5.3272333316923185e-06,
      "loss": 0.0019,
      "step": 97400
    },
    {
      "epoch": 1.5938803894297635,
      "grad_norm": 0.08566240221261978,
      "learning_rate": 5.3262834265809926e-06,
      "loss": 0.0019,
      "step": 97410
    },
    {
      "epoch": 1.5940440153808395,
      "grad_norm": 0.143513485789299,
      "learning_rate": 5.325333509642676e-06,
      "loss": 0.0008,
      "step": 97420
    },
    {
      "epoch": 1.5942076413319153,
      "grad_norm": 0.1667342185974121,
      "learning_rate": 5.324383580911802e-06,
      "loss": 0.0017,
      "step": 97430
    },
    {
      "epoch": 1.594371267282991,
      "grad_norm": 0.0306320209056139,
      "learning_rate": 5.323433640422803e-06,
      "loss": 0.0041,
      "step": 97440
    },
    {
      "epoch": 1.594534893234067,
      "grad_norm": 0.05147324874997139,
      "learning_rate": 5.322483688210112e-06,
      "loss": 0.0011,
      "step": 97450
    },
    {
      "epoch": 1.5946985191851426,
      "grad_norm": 0.020326945930719376,
      "learning_rate": 5.321533724308161e-06,
      "loss": 0.0015,
      "step": 97460
    },
    {
      "epoch": 1.5948621451362186,
      "grad_norm": 0.2365170121192932,
      "learning_rate": 5.320583748751387e-06,
      "loss": 0.0025,
      "step": 97470
    },
    {
      "epoch": 1.5950257710872946,
      "grad_norm": 0.05175967887043953,
      "learning_rate": 5.319633761574221e-06,
      "loss": 0.0022,
      "step": 97480
    },
    {
      "epoch": 1.5951893970383701,
      "grad_norm": 0.03263901546597481,
      "learning_rate": 5.3186837628111e-06,
      "loss": 0.0011,
      "step": 97490
    },
    {
      "epoch": 1.5953530229894461,
      "grad_norm": 0.053251780569553375,
      "learning_rate": 5.31773375249646e-06,
      "loss": 0.0014,
      "step": 97500
    },
    {
      "epoch": 1.595516648940522,
      "grad_norm": 0.11376066505908966,
      "learning_rate": 5.316783730664734e-06,
      "loss": 0.0016,
      "step": 97510
    },
    {
      "epoch": 1.5956802748915977,
      "grad_norm": 0.10079194605350494,
      "learning_rate": 5.315833697350359e-06,
      "loss": 0.0016,
      "step": 97520
    },
    {
      "epoch": 1.5958439008426737,
      "grad_norm": 0.07104119658470154,
      "learning_rate": 5.3148836525877715e-06,
      "loss": 0.0034,
      "step": 97530
    },
    {
      "epoch": 1.5960075267937495,
      "grad_norm": 0.06889428198337555,
      "learning_rate": 5.313933596411407e-06,
      "loss": 0.0013,
      "step": 97540
    },
    {
      "epoch": 1.5961711527448252,
      "grad_norm": 0.1272384375333786,
      "learning_rate": 5.312983528855708e-06,
      "loss": 0.0033,
      "step": 97550
    },
    {
      "epoch": 1.5963347786959012,
      "grad_norm": 0.1345529407262802,
      "learning_rate": 5.312033449955105e-06,
      "loss": 0.0018,
      "step": 97560
    },
    {
      "epoch": 1.596498404646977,
      "grad_norm": 0.06928007304668427,
      "learning_rate": 5.311083359744041e-06,
      "loss": 0.0013,
      "step": 97570
    },
    {
      "epoch": 1.5966620305980528,
      "grad_norm": 0.15723025798797607,
      "learning_rate": 5.310133258256951e-06,
      "loss": 0.0017,
      "step": 97580
    },
    {
      "epoch": 1.5968256565491288,
      "grad_norm": 0.06681761890649796,
      "learning_rate": 5.309183145528278e-06,
      "loss": 0.0023,
      "step": 97590
    },
    {
      "epoch": 1.5969892825002046,
      "grad_norm": 0.07914423197507858,
      "learning_rate": 5.308233021592457e-06,
      "loss": 0.0015,
      "step": 97600
    },
    {
      "epoch": 1.5971529084512803,
      "grad_norm": 0.20144113898277283,
      "learning_rate": 5.307282886483931e-06,
      "loss": 0.0031,
      "step": 97610
    },
    {
      "epoch": 1.5973165344023563,
      "grad_norm": 0.09355407953262329,
      "learning_rate": 5.306332740237139e-06,
      "loss": 0.0035,
      "step": 97620
    },
    {
      "epoch": 1.597480160353432,
      "grad_norm": 0.20603951811790466,
      "learning_rate": 5.305382582886521e-06,
      "loss": 0.0016,
      "step": 97630
    },
    {
      "epoch": 1.5976437863045079,
      "grad_norm": 0.14873671531677246,
      "learning_rate": 5.304432414466518e-06,
      "loss": 0.0019,
      "step": 97640
    },
    {
      "epoch": 1.5978074122555839,
      "grad_norm": 0.051463715732097626,
      "learning_rate": 5.303482235011573e-06,
      "loss": 0.002,
      "step": 97650
    },
    {
      "epoch": 1.5979710382066594,
      "grad_norm": 0.13742105662822723,
      "learning_rate": 5.3025320445561265e-06,
      "loss": 0.0025,
      "step": 97660
    },
    {
      "epoch": 1.5981346641577354,
      "grad_norm": 0.09530379623174667,
      "learning_rate": 5.301581843134619e-06,
      "loss": 0.002,
      "step": 97670
    },
    {
      "epoch": 1.5982982901088114,
      "grad_norm": 0.0777374729514122,
      "learning_rate": 5.300631630781494e-06,
      "loss": 0.0019,
      "step": 97680
    },
    {
      "epoch": 1.598461916059887,
      "grad_norm": 0.031285736709833145,
      "learning_rate": 5.299681407531197e-06,
      "loss": 0.0014,
      "step": 97690
    },
    {
      "epoch": 1.598625542010963,
      "grad_norm": 0.0172182135283947,
      "learning_rate": 5.298731173418169e-06,
      "loss": 0.0013,
      "step": 97700
    },
    {
      "epoch": 1.5987891679620387,
      "grad_norm": 0.11044405400753021,
      "learning_rate": 5.297780928476852e-06,
      "loss": 0.0026,
      "step": 97710
    },
    {
      "epoch": 1.5989527939131145,
      "grad_norm": 0.06067486107349396,
      "learning_rate": 5.296830672741694e-06,
      "loss": 0.0007,
      "step": 97720
    },
    {
      "epoch": 1.5991164198641905,
      "grad_norm": 0.1673189103603363,
      "learning_rate": 5.295880406247137e-06,
      "loss": 0.0016,
      "step": 97730
    },
    {
      "epoch": 1.5992800458152663,
      "grad_norm": 0.04932381212711334,
      "learning_rate": 5.294930129027625e-06,
      "loss": 0.0018,
      "step": 97740
    },
    {
      "epoch": 1.599443671766342,
      "grad_norm": 0.07004158943891525,
      "learning_rate": 5.293979841117607e-06,
      "loss": 0.0022,
      "step": 97750
    },
    {
      "epoch": 1.599607297717418,
      "grad_norm": 0.08543721586465836,
      "learning_rate": 5.293029542551524e-06,
      "loss": 0.002,
      "step": 97760
    },
    {
      "epoch": 1.5997709236684938,
      "grad_norm": 0.3352496325969696,
      "learning_rate": 5.292079233363826e-06,
      "loss": 0.0025,
      "step": 97770
    },
    {
      "epoch": 1.5999345496195696,
      "grad_norm": 0.0744246244430542,
      "learning_rate": 5.2911289135889575e-06,
      "loss": 0.0028,
      "step": 97780
    },
    {
      "epoch": 1.6000981755706456,
      "grad_norm": 0.11093030869960785,
      "learning_rate": 5.290178583261365e-06,
      "loss": 0.0018,
      "step": 97790
    },
    {
      "epoch": 1.6002618015217214,
      "grad_norm": 0.052570831030607224,
      "learning_rate": 5.2892282424154975e-06,
      "loss": 0.0026,
      "step": 97800
    },
    {
      "epoch": 1.6004254274727971,
      "grad_norm": 0.24708853662014008,
      "learning_rate": 5.288277891085801e-06,
      "loss": 0.0019,
      "step": 97810
    },
    {
      "epoch": 1.6005890534238731,
      "grad_norm": 0.08212660253047943,
      "learning_rate": 5.287327529306722e-06,
      "loss": 0.0014,
      "step": 97820
    },
    {
      "epoch": 1.600752679374949,
      "grad_norm": 0.06644292920827866,
      "learning_rate": 5.286377157112712e-06,
      "loss": 0.0007,
      "step": 97830
    },
    {
      "epoch": 1.6009163053260247,
      "grad_norm": 0.04395301640033722,
      "learning_rate": 5.2854267745382185e-06,
      "loss": 0.0019,
      "step": 97840
    },
    {
      "epoch": 1.6010799312771007,
      "grad_norm": 0.04277435317635536,
      "learning_rate": 5.28447638161769e-06,
      "loss": 0.0018,
      "step": 97850
    },
    {
      "epoch": 1.6012435572281762,
      "grad_norm": 0.1711159348487854,
      "learning_rate": 5.283525978385577e-06,
      "loss": 0.0016,
      "step": 97860
    },
    {
      "epoch": 1.6014071831792522,
      "grad_norm": 0.1063678115606308,
      "learning_rate": 5.282575564876329e-06,
      "loss": 0.0013,
      "step": 97870
    },
    {
      "epoch": 1.6015708091303282,
      "grad_norm": 0.035489022731781006,
      "learning_rate": 5.281625141124396e-06,
      "loss": 0.0018,
      "step": 97880
    },
    {
      "epoch": 1.6017344350814038,
      "grad_norm": 0.15812666714191437,
      "learning_rate": 5.280674707164227e-06,
      "loss": 0.0024,
      "step": 97890
    },
    {
      "epoch": 1.6018980610324798,
      "grad_norm": 0.028400007635354996,
      "learning_rate": 5.279724263030274e-06,
      "loss": 0.0029,
      "step": 97900
    },
    {
      "epoch": 1.6020616869835556,
      "grad_norm": 0.08711693435907364,
      "learning_rate": 5.278773808756992e-06,
      "loss": 0.0012,
      "step": 97910
    },
    {
      "epoch": 1.6022253129346313,
      "grad_norm": 0.016361359506845474,
      "learning_rate": 5.277823344378827e-06,
      "loss": 0.002,
      "step": 97920
    },
    {
      "epoch": 1.6023889388857073,
      "grad_norm": 0.14891168475151062,
      "learning_rate": 5.2768728699302344e-06,
      "loss": 0.0029,
      "step": 97930
    },
    {
      "epoch": 1.602552564836783,
      "grad_norm": 0.2883024513721466,
      "learning_rate": 5.275922385445665e-06,
      "loss": 0.002,
      "step": 97940
    },
    {
      "epoch": 1.6027161907878589,
      "grad_norm": 0.06108860298991203,
      "learning_rate": 5.274971890959571e-06,
      "loss": 0.0011,
      "step": 97950
    },
    {
      "epoch": 1.6028798167389349,
      "grad_norm": 0.0763351172208786,
      "learning_rate": 5.27402138650641e-06,
      "loss": 0.0018,
      "step": 97960
    },
    {
      "epoch": 1.6030434426900106,
      "grad_norm": 0.14666365087032318,
      "learning_rate": 5.27307087212063e-06,
      "loss": 0.0015,
      "step": 97970
    },
    {
      "epoch": 1.6032070686410864,
      "grad_norm": 0.11977433413267136,
      "learning_rate": 5.272120347836687e-06,
      "loss": 0.0019,
      "step": 97980
    },
    {
      "epoch": 1.6033706945921624,
      "grad_norm": 0.1182369738817215,
      "learning_rate": 5.271169813689037e-06,
      "loss": 0.0044,
      "step": 97990
    },
    {
      "epoch": 1.6035343205432382,
      "grad_norm": 0.18548201024532318,
      "learning_rate": 5.270219269712132e-06,
      "loss": 0.0021,
      "step": 98000
    },
    {
      "epoch": 1.603697946494314,
      "grad_norm": 0.03501434251666069,
      "learning_rate": 5.26926871594043e-06,
      "loss": 0.001,
      "step": 98010
    },
    {
      "epoch": 1.60386157244539,
      "grad_norm": 0.1672307252883911,
      "learning_rate": 5.2683181524083815e-06,
      "loss": 0.0032,
      "step": 98020
    },
    {
      "epoch": 1.6040251983964657,
      "grad_norm": 0.09246539324522018,
      "learning_rate": 5.267367579150447e-06,
      "loss": 0.0012,
      "step": 98030
    },
    {
      "epoch": 1.6041888243475415,
      "grad_norm": 0.2178044617176056,
      "learning_rate": 5.266416996201077e-06,
      "loss": 0.0019,
      "step": 98040
    },
    {
      "epoch": 1.6043524502986175,
      "grad_norm": 0.038778070360422134,
      "learning_rate": 5.265466403594733e-06,
      "loss": 0.0011,
      "step": 98050
    },
    {
      "epoch": 1.604516076249693,
      "grad_norm": 0.17926302552223206,
      "learning_rate": 5.264515801365872e-06,
      "loss": 0.0012,
      "step": 98060
    },
    {
      "epoch": 1.604679702200769,
      "grad_norm": 0.17083346843719482,
      "learning_rate": 5.263565189548947e-06,
      "loss": 0.0022,
      "step": 98070
    },
    {
      "epoch": 1.6048433281518448,
      "grad_norm": 0.03759386017918587,
      "learning_rate": 5.262614568178418e-06,
      "loss": 0.0009,
      "step": 98080
    },
    {
      "epoch": 1.6050069541029206,
      "grad_norm": 0.11836827546358109,
      "learning_rate": 5.2616639372887416e-06,
      "loss": 0.0025,
      "step": 98090
    },
    {
      "epoch": 1.6051705800539966,
      "grad_norm": 0.013447455130517483,
      "learning_rate": 5.260713296914376e-06,
      "loss": 0.0016,
      "step": 98100
    },
    {
      "epoch": 1.6053342060050724,
      "grad_norm": 0.11086362600326538,
      "learning_rate": 5.25976264708978e-06,
      "loss": 0.0019,
      "step": 98110
    },
    {
      "epoch": 1.6054978319561481,
      "grad_norm": 0.0646359845995903,
      "learning_rate": 5.258811987849413e-06,
      "loss": 0.0017,
      "step": 98120
    },
    {
      "epoch": 1.6056614579072241,
      "grad_norm": 0.24520373344421387,
      "learning_rate": 5.2578613192277325e-06,
      "loss": 0.0021,
      "step": 98130
    },
    {
      "epoch": 1.6058250838583,
      "grad_norm": 0.09542601555585861,
      "learning_rate": 5.256910641259201e-06,
      "loss": 0.0016,
      "step": 98140
    },
    {
      "epoch": 1.6059887098093757,
      "grad_norm": 0.472331166267395,
      "learning_rate": 5.255959953978275e-06,
      "loss": 0.0022,
      "step": 98150
    },
    {
      "epoch": 1.6061523357604517,
      "grad_norm": 0.06958793103694916,
      "learning_rate": 5.2550092574194165e-06,
      "loss": 0.0018,
      "step": 98160
    },
    {
      "epoch": 1.6063159617115275,
      "grad_norm": 0.09929090738296509,
      "learning_rate": 5.254058551617084e-06,
      "loss": 0.0016,
      "step": 98170
    },
    {
      "epoch": 1.6064795876626032,
      "grad_norm": 0.07176434248685837,
      "learning_rate": 5.25310783660574e-06,
      "loss": 0.0014,
      "step": 98180
    },
    {
      "epoch": 1.6066432136136792,
      "grad_norm": 0.0863189771771431,
      "learning_rate": 5.252157112419844e-06,
      "loss": 0.0014,
      "step": 98190
    },
    {
      "epoch": 1.606806839564755,
      "grad_norm": 0.05574188381433487,
      "learning_rate": 5.251206379093861e-06,
      "loss": 0.0018,
      "step": 98200
    },
    {
      "epoch": 1.6069704655158308,
      "grad_norm": 0.3285781443119049,
      "learning_rate": 5.25025563666225e-06,
      "loss": 0.0033,
      "step": 98210
    },
    {
      "epoch": 1.6071340914669068,
      "grad_norm": 0.1631193906068802,
      "learning_rate": 5.249304885159474e-06,
      "loss": 0.0006,
      "step": 98220
    },
    {
      "epoch": 1.6072977174179823,
      "grad_norm": 0.2060345858335495,
      "learning_rate": 5.248354124619994e-06,
      "loss": 0.0016,
      "step": 98230
    },
    {
      "epoch": 1.6074613433690583,
      "grad_norm": 0.09031964093446732,
      "learning_rate": 5.247403355078275e-06,
      "loss": 0.0012,
      "step": 98240
    },
    {
      "epoch": 1.6076249693201343,
      "grad_norm": 0.13665316998958588,
      "learning_rate": 5.246452576568778e-06,
      "loss": 0.0019,
      "step": 98250
    },
    {
      "epoch": 1.6077885952712099,
      "grad_norm": 0.08412686735391617,
      "learning_rate": 5.245501789125968e-06,
      "loss": 0.0022,
      "step": 98260
    },
    {
      "epoch": 1.6079522212222859,
      "grad_norm": 0.03635793179273605,
      "learning_rate": 5.2445509927843074e-06,
      "loss": 0.0013,
      "step": 98270
    },
    {
      "epoch": 1.6081158471733616,
      "grad_norm": 0.04021189734339714,
      "learning_rate": 5.243600187578262e-06,
      "loss": 0.003,
      "step": 98280
    },
    {
      "epoch": 1.6082794731244374,
      "grad_norm": 0.04579615592956543,
      "learning_rate": 5.242649373542295e-06,
      "loss": 0.001,
      "step": 98290
    },
    {
      "epoch": 1.6084430990755134,
      "grad_norm": 0.03251798078417778,
      "learning_rate": 5.241698550710871e-06,
      "loss": 0.002,
      "step": 98300
    },
    {
      "epoch": 1.6086067250265892,
      "grad_norm": 0.08444109559059143,
      "learning_rate": 5.240747719118456e-06,
      "loss": 0.0019,
      "step": 98310
    },
    {
      "epoch": 1.608770350977665,
      "grad_norm": 0.08320547640323639,
      "learning_rate": 5.239796878799514e-06,
      "loss": 0.0016,
      "step": 98320
    },
    {
      "epoch": 1.608933976928741,
      "grad_norm": 0.0728529691696167,
      "learning_rate": 5.238846029788511e-06,
      "loss": 0.001,
      "step": 98330
    },
    {
      "epoch": 1.6090976028798167,
      "grad_norm": 0.0988699272274971,
      "learning_rate": 5.2378951721199154e-06,
      "loss": 0.0026,
      "step": 98340
    },
    {
      "epoch": 1.6092612288308925,
      "grad_norm": 0.1827646940946579,
      "learning_rate": 5.23694430582819e-06,
      "loss": 0.0015,
      "step": 98350
    },
    {
      "epoch": 1.6094248547819685,
      "grad_norm": 0.12196511030197144,
      "learning_rate": 5.235993430947806e-06,
      "loss": 0.0026,
      "step": 98360
    },
    {
      "epoch": 1.6095884807330443,
      "grad_norm": 0.07533494383096695,
      "learning_rate": 5.235042547513224e-06,
      "loss": 0.0013,
      "step": 98370
    },
    {
      "epoch": 1.60975210668412,
      "grad_norm": 0.14361101388931274,
      "learning_rate": 5.234091655558915e-06,
      "loss": 0.0013,
      "step": 98380
    },
    {
      "epoch": 1.609915732635196,
      "grad_norm": 0.0888654962182045,
      "learning_rate": 5.233140755119347e-06,
      "loss": 0.0012,
      "step": 98390
    },
    {
      "epoch": 1.6100793585862718,
      "grad_norm": 0.13821378350257874,
      "learning_rate": 5.2321898462289855e-06,
      "loss": 0.0024,
      "step": 98400
    },
    {
      "epoch": 1.6102429845373476,
      "grad_norm": 0.17549480497837067,
      "learning_rate": 5.231238928922302e-06,
      "loss": 0.0026,
      "step": 98410
    },
    {
      "epoch": 1.6104066104884236,
      "grad_norm": 0.34210479259490967,
      "learning_rate": 5.230288003233764e-06,
      "loss": 0.0011,
      "step": 98420
    },
    {
      "epoch": 1.6105702364394991,
      "grad_norm": 0.12255807220935822,
      "learning_rate": 5.229337069197836e-06,
      "loss": 0.0015,
      "step": 98430
    },
    {
      "epoch": 1.6107338623905751,
      "grad_norm": 0.1388000100851059,
      "learning_rate": 5.228386126848992e-06,
      "loss": 0.0025,
      "step": 98440
    },
    {
      "epoch": 1.6108974883416511,
      "grad_norm": 0.17837822437286377,
      "learning_rate": 5.2274351762217005e-06,
      "loss": 0.0014,
      "step": 98450
    },
    {
      "epoch": 1.6110611142927267,
      "grad_norm": 0.057697638869285583,
      "learning_rate": 5.226484217350429e-06,
      "loss": 0.0014,
      "step": 98460
    },
    {
      "epoch": 1.6112247402438027,
      "grad_norm": 0.03137467801570892,
      "learning_rate": 5.225533250269651e-06,
      "loss": 0.0012,
      "step": 98470
    },
    {
      "epoch": 1.6113883661948785,
      "grad_norm": 0.07953894883394241,
      "learning_rate": 5.224582275013833e-06,
      "loss": 0.0021,
      "step": 98480
    },
    {
      "epoch": 1.6115519921459542,
      "grad_norm": 0.17380009591579437,
      "learning_rate": 5.2236312916174484e-06,
      "loss": 0.0016,
      "step": 98490
    },
    {
      "epoch": 1.6117156180970302,
      "grad_norm": 0.24816903471946716,
      "learning_rate": 5.222680300114966e-06,
      "loss": 0.0017,
      "step": 98500
    },
    {
      "epoch": 1.611879244048106,
      "grad_norm": 0.08738347887992859,
      "learning_rate": 5.22172930054086e-06,
      "loss": 0.0032,
      "step": 98510
    },
    {
      "epoch": 1.6120428699991818,
      "grad_norm": 0.24474535882472992,
      "learning_rate": 5.220778292929598e-06,
      "loss": 0.0016,
      "step": 98520
    },
    {
      "epoch": 1.6122064959502578,
      "grad_norm": 0.06241137534379959,
      "learning_rate": 5.219827277315652e-06,
      "loss": 0.0014,
      "step": 98530
    },
    {
      "epoch": 1.6123701219013336,
      "grad_norm": 0.0731671005487442,
      "learning_rate": 5.218876253733498e-06,
      "loss": 0.0015,
      "step": 98540
    },
    {
      "epoch": 1.6125337478524093,
      "grad_norm": 0.036975424736738205,
      "learning_rate": 5.217925222217604e-06,
      "loss": 0.001,
      "step": 98550
    },
    {
      "epoch": 1.6126973738034853,
      "grad_norm": 0.09705603867769241,
      "learning_rate": 5.216974182802448e-06,
      "loss": 0.0017,
      "step": 98560
    },
    {
      "epoch": 1.612860999754561,
      "grad_norm": 0.4200478792190552,
      "learning_rate": 5.216023135522497e-06,
      "loss": 0.0016,
      "step": 98570
    },
    {
      "epoch": 1.6130246257056369,
      "grad_norm": 0.07903515547513962,
      "learning_rate": 5.215072080412225e-06,
      "loss": 0.0012,
      "step": 98580
    },
    {
      "epoch": 1.6131882516567129,
      "grad_norm": 0.17032834887504578,
      "learning_rate": 5.214121017506108e-06,
      "loss": 0.0012,
      "step": 98590
    },
    {
      "epoch": 1.6133518776077886,
      "grad_norm": 0.11135027557611465,
      "learning_rate": 5.21316994683862e-06,
      "loss": 0.0018,
      "step": 98600
    },
    {
      "epoch": 1.6135155035588644,
      "grad_norm": 0.03420327976346016,
      "learning_rate": 5.212218868444233e-06,
      "loss": 0.0018,
      "step": 98610
    },
    {
      "epoch": 1.6136791295099404,
      "grad_norm": 0.08032601326704025,
      "learning_rate": 5.211267782357422e-06,
      "loss": 0.0019,
      "step": 98620
    },
    {
      "epoch": 1.613842755461016,
      "grad_norm": 0.056985605508089066,
      "learning_rate": 5.210316688612662e-06,
      "loss": 0.0014,
      "step": 98630
    },
    {
      "epoch": 1.614006381412092,
      "grad_norm": 0.1564008891582489,
      "learning_rate": 5.209365587244427e-06,
      "loss": 0.0014,
      "step": 98640
    },
    {
      "epoch": 1.614170007363168,
      "grad_norm": 0.021098554134368896,
      "learning_rate": 5.208414478287193e-06,
      "loss": 0.0021,
      "step": 98650
    },
    {
      "epoch": 1.6143336333142435,
      "grad_norm": 0.057873841375112534,
      "learning_rate": 5.207463361775436e-06,
      "loss": 0.002,
      "step": 98660
    },
    {
      "epoch": 1.6144972592653195,
      "grad_norm": 0.17769621312618256,
      "learning_rate": 5.206512237743628e-06,
      "loss": 0.0017,
      "step": 98670
    },
    {
      "epoch": 1.6146608852163953,
      "grad_norm": 0.1033492311835289,
      "learning_rate": 5.20556110622625e-06,
      "loss": 0.0018,
      "step": 98680
    },
    {
      "epoch": 1.614824511167471,
      "grad_norm": 0.014122343622148037,
      "learning_rate": 5.204609967257775e-06,
      "loss": 0.0015,
      "step": 98690
    },
    {
      "epoch": 1.614988137118547,
      "grad_norm": 0.008405575528740883,
      "learning_rate": 5.203658820872681e-06,
      "loss": 0.0031,
      "step": 98700
    },
    {
      "epoch": 1.6151517630696228,
      "grad_norm": 0.05221950635313988,
      "learning_rate": 5.202707667105445e-06,
      "loss": 0.0015,
      "step": 98710
    },
    {
      "epoch": 1.6153153890206986,
      "grad_norm": 0.06928742676973343,
      "learning_rate": 5.2017565059905415e-06,
      "loss": 0.0026,
      "step": 98720
    },
    {
      "epoch": 1.6154790149717746,
      "grad_norm": 0.056128133088350296,
      "learning_rate": 5.2008053375624515e-06,
      "loss": 0.0013,
      "step": 98730
    },
    {
      "epoch": 1.6156426409228504,
      "grad_norm": 0.5218814015388489,
      "learning_rate": 5.199854161855649e-06,
      "loss": 0.0026,
      "step": 98740
    },
    {
      "epoch": 1.6158062668739261,
      "grad_norm": 0.08757639676332474,
      "learning_rate": 5.198902978904613e-06,
      "loss": 0.0018,
      "step": 98750
    },
    {
      "epoch": 1.6159698928250021,
      "grad_norm": 0.23714497685432434,
      "learning_rate": 5.197951788743824e-06,
      "loss": 0.002,
      "step": 98760
    },
    {
      "epoch": 1.616133518776078,
      "grad_norm": 0.0703224167227745,
      "learning_rate": 5.197000591407757e-06,
      "loss": 0.0016,
      "step": 98770
    },
    {
      "epoch": 1.6162971447271537,
      "grad_norm": 0.05482254922389984,
      "learning_rate": 5.196049386930893e-06,
      "loss": 0.0009,
      "step": 98780
    },
    {
      "epoch": 1.6164607706782297,
      "grad_norm": 0.05517350509762764,
      "learning_rate": 5.195098175347709e-06,
      "loss": 0.0013,
      "step": 98790
    },
    {
      "epoch": 1.6166243966293055,
      "grad_norm": 0.35550621151924133,
      "learning_rate": 5.194146956692686e-06,
      "loss": 0.0019,
      "step": 98800
    },
    {
      "epoch": 1.6167880225803812,
      "grad_norm": 0.01884710229933262,
      "learning_rate": 5.193195731000301e-06,
      "loss": 0.0009,
      "step": 98810
    },
    {
      "epoch": 1.6169516485314572,
      "grad_norm": 0.20061872899532318,
      "learning_rate": 5.192244498305037e-06,
      "loss": 0.0018,
      "step": 98820
    },
    {
      "epoch": 1.6171152744825328,
      "grad_norm": 0.2248377650976181,
      "learning_rate": 5.19129325864137e-06,
      "loss": 0.0014,
      "step": 98830
    },
    {
      "epoch": 1.6172789004336088,
      "grad_norm": 0.19466260075569153,
      "learning_rate": 5.190342012043783e-06,
      "loss": 0.0013,
      "step": 98840
    },
    {
      "epoch": 1.6174425263846848,
      "grad_norm": 0.32903891801834106,
      "learning_rate": 5.189390758546756e-06,
      "loss": 0.002,
      "step": 98850
    },
    {
      "epoch": 1.6176061523357603,
      "grad_norm": 0.08313871920108795,
      "learning_rate": 5.188439498184769e-06,
      "loss": 0.0024,
      "step": 98860
    },
    {
      "epoch": 1.6177697782868363,
      "grad_norm": 0.06497074663639069,
      "learning_rate": 5.187488230992304e-06,
      "loss": 0.0037,
      "step": 98870
    },
    {
      "epoch": 1.617933404237912,
      "grad_norm": 0.02705974504351616,
      "learning_rate": 5.186536957003841e-06,
      "loss": 0.0011,
      "step": 98880
    },
    {
      "epoch": 1.6180970301889879,
      "grad_norm": 0.13264022767543793,
      "learning_rate": 5.185585676253859e-06,
      "loss": 0.007,
      "step": 98890
    },
    {
      "epoch": 1.6182606561400639,
      "grad_norm": 0.12600697576999664,
      "learning_rate": 5.184634388776844e-06,
      "loss": 0.0021,
      "step": 98900
    },
    {
      "epoch": 1.6184242820911396,
      "grad_norm": 0.08227627724409103,
      "learning_rate": 5.183683094607277e-06,
      "loss": 0.0015,
      "step": 98910
    },
    {
      "epoch": 1.6185879080422154,
      "grad_norm": 0.1436682790517807,
      "learning_rate": 5.182731793779639e-06,
      "loss": 0.0021,
      "step": 98920
    },
    {
      "epoch": 1.6187515339932914,
      "grad_norm": 0.09100278466939926,
      "learning_rate": 5.1817804863284114e-06,
      "loss": 0.0016,
      "step": 98930
    },
    {
      "epoch": 1.6189151599443672,
      "grad_norm": 0.07239263504743576,
      "learning_rate": 5.18082917228808e-06,
      "loss": 0.001,
      "step": 98940
    },
    {
      "epoch": 1.619078785895443,
      "grad_norm": 0.001040601171553135,
      "learning_rate": 5.179877851693123e-06,
      "loss": 0.0013,
      "step": 98950
    },
    {
      "epoch": 1.619242411846519,
      "grad_norm": 0.11623147130012512,
      "learning_rate": 5.1789265245780286e-06,
      "loss": 0.0018,
      "step": 98960
    },
    {
      "epoch": 1.6194060377975947,
      "grad_norm": 0.1853215992450714,
      "learning_rate": 5.177975190977277e-06,
      "loss": 0.0016,
      "step": 98970
    },
    {
      "epoch": 1.6195696637486705,
      "grad_norm": 0.042178887873888016,
      "learning_rate": 5.177023850925353e-06,
      "loss": 0.0016,
      "step": 98980
    },
    {
      "epoch": 1.6197332896997465,
      "grad_norm": 0.09309615939855576,
      "learning_rate": 5.1760725044567385e-06,
      "loss": 0.0012,
      "step": 98990
    },
    {
      "epoch": 1.619896915650822,
      "grad_norm": 0.08934682607650757,
      "learning_rate": 5.17512115160592e-06,
      "loss": 0.0028,
      "step": 99000
    },
    {
      "epoch": 1.620060541601898,
      "grad_norm": 0.17292897403240204,
      "learning_rate": 5.1741697924073806e-06,
      "loss": 0.0013,
      "step": 99010
    },
    {
      "epoch": 1.620224167552974,
      "grad_norm": 0.22815579175949097,
      "learning_rate": 5.173218426895606e-06,
      "loss": 0.0039,
      "step": 99020
    },
    {
      "epoch": 1.6203877935040496,
      "grad_norm": 0.2680216431617737,
      "learning_rate": 5.172267055105076e-06,
      "loss": 0.0024,
      "step": 99030
    },
    {
      "epoch": 1.6205514194551256,
      "grad_norm": 0.02104036509990692,
      "learning_rate": 5.1713156770702825e-06,
      "loss": 0.0012,
      "step": 99040
    },
    {
      "epoch": 1.6207150454062014,
      "grad_norm": 0.08003778010606766,
      "learning_rate": 5.170364292825706e-06,
      "loss": 0.0024,
      "step": 99050
    },
    {
      "epoch": 1.6208786713572771,
      "grad_norm": 0.15551789104938507,
      "learning_rate": 5.169412902405835e-06,
      "loss": 0.0027,
      "step": 99060
    },
    {
      "epoch": 1.6210422973083531,
      "grad_norm": 0.04570987820625305,
      "learning_rate": 5.168461505845153e-06,
      "loss": 0.0009,
      "step": 99070
    },
    {
      "epoch": 1.621205923259429,
      "grad_norm": 0.08118777722120285,
      "learning_rate": 5.1675101031781446e-06,
      "loss": 0.0017,
      "step": 99080
    },
    {
      "epoch": 1.6213695492105047,
      "grad_norm": 0.09732763469219208,
      "learning_rate": 5.1665586944392985e-06,
      "loss": 0.0017,
      "step": 99090
    },
    {
      "epoch": 1.6215331751615807,
      "grad_norm": 0.06535319983959198,
      "learning_rate": 5.1656072796631006e-06,
      "loss": 0.0017,
      "step": 99100
    },
    {
      "epoch": 1.6216968011126565,
      "grad_norm": 0.06291642785072327,
      "learning_rate": 5.164655858884036e-06,
      "loss": 0.0021,
      "step": 99110
    },
    {
      "epoch": 1.6218604270637322,
      "grad_norm": 0.24212847650051117,
      "learning_rate": 5.163704432136593e-06,
      "loss": 0.0017,
      "step": 99120
    },
    {
      "epoch": 1.6220240530148082,
      "grad_norm": 0.0826900377869606,
      "learning_rate": 5.1627529994552576e-06,
      "loss": 0.0015,
      "step": 99130
    },
    {
      "epoch": 1.622187678965884,
      "grad_norm": 0.08897511661052704,
      "learning_rate": 5.161801560874517e-06,
      "loss": 0.0014,
      "step": 99140
    },
    {
      "epoch": 1.6223513049169598,
      "grad_norm": 0.005243068560957909,
      "learning_rate": 5.160850116428859e-06,
      "loss": 0.0013,
      "step": 99150
    },
    {
      "epoch": 1.6225149308680358,
      "grad_norm": 0.03812631964683533,
      "learning_rate": 5.15989866615277e-06,
      "loss": 0.0015,
      "step": 99160
    },
    {
      "epoch": 1.6226785568191116,
      "grad_norm": 0.13536326587200165,
      "learning_rate": 5.158947210080738e-06,
      "loss": 0.0013,
      "step": 99170
    },
    {
      "epoch": 1.6228421827701873,
      "grad_norm": 0.12537388503551483,
      "learning_rate": 5.157995748247253e-06,
      "loss": 0.002,
      "step": 99180
    },
    {
      "epoch": 1.6230058087212633,
      "grad_norm": 0.3827470541000366,
      "learning_rate": 5.157044280686802e-06,
      "loss": 0.0019,
      "step": 99190
    },
    {
      "epoch": 1.6231694346723389,
      "grad_norm": 0.014011899009346962,
      "learning_rate": 5.1560928074338735e-06,
      "loss": 0.0013,
      "step": 99200
    },
    {
      "epoch": 1.6233330606234149,
      "grad_norm": 0.0309713464230299,
      "learning_rate": 5.155141328522956e-06,
      "loss": 0.0015,
      "step": 99210
    },
    {
      "epoch": 1.6234966865744909,
      "grad_norm": 0.07933453470468521,
      "learning_rate": 5.154189843988538e-06,
      "loss": 0.0015,
      "step": 99220
    },
    {
      "epoch": 1.6236603125255664,
      "grad_norm": 0.06462419778108597,
      "learning_rate": 5.1532383538651075e-06,
      "loss": 0.0018,
      "step": 99230
    },
    {
      "epoch": 1.6238239384766424,
      "grad_norm": 0.03575167804956436,
      "learning_rate": 5.1522868581871555e-06,
      "loss": 0.0008,
      "step": 99240
    },
    {
      "epoch": 1.6239875644277182,
      "grad_norm": 0.05680159106850624,
      "learning_rate": 5.15133535698917e-06,
      "loss": 0.001,
      "step": 99250
    },
    {
      "epoch": 1.624151190378794,
      "grad_norm": 0.059106118977069855,
      "learning_rate": 5.1503838503056446e-06,
      "loss": 0.003,
      "step": 99260
    },
    {
      "epoch": 1.62431481632987,
      "grad_norm": 0.08148624747991562,
      "learning_rate": 5.149432338171065e-06,
      "loss": 0.0011,
      "step": 99270
    },
    {
      "epoch": 1.6244784422809457,
      "grad_norm": 0.07005994766950607,
      "learning_rate": 5.14848082061992e-06,
      "loss": 0.0027,
      "step": 99280
    },
    {
      "epoch": 1.6246420682320215,
      "grad_norm": 0.04762598127126694,
      "learning_rate": 5.147529297686704e-06,
      "loss": 0.0012,
      "step": 99290
    },
    {
      "epoch": 1.6248056941830975,
      "grad_norm": 0.2658834457397461,
      "learning_rate": 5.146577769405905e-06,
      "loss": 0.0011,
      "step": 99300
    },
    {
      "epoch": 1.6249693201341733,
      "grad_norm": 0.11333608627319336,
      "learning_rate": 5.145626235812015e-06,
      "loss": 0.0021,
      "step": 99310
    },
    {
      "epoch": 1.625132946085249,
      "grad_norm": 0.12355748564004898,
      "learning_rate": 5.144674696939522e-06,
      "loss": 0.0015,
      "step": 99320
    },
    {
      "epoch": 1.625296572036325,
      "grad_norm": 0.20626477897167206,
      "learning_rate": 5.14372315282292e-06,
      "loss": 0.0019,
      "step": 99330
    },
    {
      "epoch": 1.6254601979874008,
      "grad_norm": 0.04642678424715996,
      "learning_rate": 5.1427716034967e-06,
      "loss": 0.0009,
      "step": 99340
    },
    {
      "epoch": 1.6256238239384766,
      "grad_norm": 0.042989879846572876,
      "learning_rate": 5.141820048995352e-06,
      "loss": 0.0026,
      "step": 99350
    },
    {
      "epoch": 1.6257874498895526,
      "grad_norm": 0.2868697941303253,
      "learning_rate": 5.140868489353369e-06,
      "loss": 0.0017,
      "step": 99360
    },
    {
      "epoch": 1.6259510758406284,
      "grad_norm": 0.12098990380764008,
      "learning_rate": 5.13991692460524e-06,
      "loss": 0.0018,
      "step": 99370
    },
    {
      "epoch": 1.6261147017917041,
      "grad_norm": 0.01503510121256113,
      "learning_rate": 5.13896535478546e-06,
      "loss": 0.0009,
      "step": 99380
    },
    {
      "epoch": 1.6262783277427801,
      "grad_norm": 0.08358880877494812,
      "learning_rate": 5.138013779928518e-06,
      "loss": 0.0012,
      "step": 99390
    },
    {
      "epoch": 1.6264419536938557,
      "grad_norm": 0.028315600007772446,
      "learning_rate": 5.13706220006891e-06,
      "loss": 0.0021,
      "step": 99400
    },
    {
      "epoch": 1.6266055796449317,
      "grad_norm": 0.04941651225090027,
      "learning_rate": 5.1361106152411265e-06,
      "loss": 0.0018,
      "step": 99410
    },
    {
      "epoch": 1.6267692055960077,
      "grad_norm": 0.07402452081441879,
      "learning_rate": 5.13515902547966e-06,
      "loss": 0.0028,
      "step": 99420
    },
    {
      "epoch": 1.6269328315470832,
      "grad_norm": 0.3740224540233612,
      "learning_rate": 5.134207430819004e-06,
      "loss": 0.0017,
      "step": 99430
    },
    {
      "epoch": 1.6270964574981592,
      "grad_norm": 0.14909420907497406,
      "learning_rate": 5.133255831293651e-06,
      "loss": 0.0017,
      "step": 99440
    },
    {
      "epoch": 1.627260083449235,
      "grad_norm": 0.12497895210981369,
      "learning_rate": 5.132304226938094e-06,
      "loss": 0.0032,
      "step": 99450
    },
    {
      "epoch": 1.6274237094003108,
      "grad_norm": 0.12497039139270782,
      "learning_rate": 5.131352617786828e-06,
      "loss": 0.0022,
      "step": 99460
    },
    {
      "epoch": 1.6275873353513868,
      "grad_norm": 0.055445168167352676,
      "learning_rate": 5.130401003874343e-06,
      "loss": 0.0016,
      "step": 99470
    },
    {
      "epoch": 1.6277509613024626,
      "grad_norm": 0.09017663449048996,
      "learning_rate": 5.129449385235137e-06,
      "loss": 0.002,
      "step": 99480
    },
    {
      "epoch": 1.6279145872535383,
      "grad_norm": 0.05772421509027481,
      "learning_rate": 5.128497761903701e-06,
      "loss": 0.0017,
      "step": 99490
    },
    {
      "epoch": 1.6280782132046143,
      "grad_norm": 0.18963901698589325,
      "learning_rate": 5.1275461339145304e-06,
      "loss": 0.0019,
      "step": 99500
    },
    {
      "epoch": 1.62824183915569,
      "grad_norm": 0.0779237151145935,
      "learning_rate": 5.12659450130212e-06,
      "loss": 0.0019,
      "step": 99510
    },
    {
      "epoch": 1.6284054651067659,
      "grad_norm": 0.12682569026947021,
      "learning_rate": 5.125642864100962e-06,
      "loss": 0.0019,
      "step": 99520
    },
    {
      "epoch": 1.6285690910578419,
      "grad_norm": 0.06850654631853104,
      "learning_rate": 5.124691222345551e-06,
      "loss": 0.0011,
      "step": 99530
    },
    {
      "epoch": 1.6287327170089176,
      "grad_norm": 0.13229204714298248,
      "learning_rate": 5.123739576070384e-06,
      "loss": 0.0042,
      "step": 99540
    },
    {
      "epoch": 1.6288963429599934,
      "grad_norm": 0.10641125589609146,
      "learning_rate": 5.122787925309955e-06,
      "loss": 0.0017,
      "step": 99550
    },
    {
      "epoch": 1.6290599689110694,
      "grad_norm": 0.16319550573825836,
      "learning_rate": 5.12183627009876e-06,
      "loss": 0.0018,
      "step": 99560
    },
    {
      "epoch": 1.6292235948621452,
      "grad_norm": 0.004445069935172796,
      "learning_rate": 5.120884610471292e-06,
      "loss": 0.0009,
      "step": 99570
    },
    {
      "epoch": 1.629387220813221,
      "grad_norm": 0.016375571489334106,
      "learning_rate": 5.119932946462047e-06,
      "loss": 0.0009,
      "step": 99580
    },
    {
      "epoch": 1.629550846764297,
      "grad_norm": 0.1307780146598816,
      "learning_rate": 5.118981278105521e-06,
      "loss": 0.0016,
      "step": 99590
    },
    {
      "epoch": 1.6297144727153725,
      "grad_norm": 0.05227362737059593,
      "learning_rate": 5.11802960543621e-06,
      "loss": 0.0019,
      "step": 99600
    },
    {
      "epoch": 1.6298780986664485,
      "grad_norm": 0.11242035031318665,
      "learning_rate": 5.1170779284886085e-06,
      "loss": 0.0021,
      "step": 99610
    },
    {
      "epoch": 1.6300417246175245,
      "grad_norm": 0.08303198963403702,
      "learning_rate": 5.116126247297214e-06,
      "loss": 0.0017,
      "step": 99620
    },
    {
      "epoch": 1.6302053505686,
      "grad_norm": 0.008203918114304543,
      "learning_rate": 5.1151745618965235e-06,
      "loss": 0.0016,
      "step": 99630
    },
    {
      "epoch": 1.630368976519676,
      "grad_norm": 0.060079704970121384,
      "learning_rate": 5.11422287232103e-06,
      "loss": 0.0021,
      "step": 99640
    },
    {
      "epoch": 1.6305326024707518,
      "grad_norm": 0.19092924892902374,
      "learning_rate": 5.113271178605232e-06,
      "loss": 0.0014,
      "step": 99650
    },
    {
      "epoch": 1.6306962284218276,
      "grad_norm": 0.31950604915618896,
      "learning_rate": 5.112319480783627e-06,
      "loss": 0.0058,
      "step": 99660
    },
    {
      "epoch": 1.6308598543729036,
      "grad_norm": 0.07352284342050552,
      "learning_rate": 5.11136777889071e-06,
      "loss": 0.0012,
      "step": 99670
    },
    {
      "epoch": 1.6310234803239794,
      "grad_norm": 0.2863670885562897,
      "learning_rate": 5.11041607296098e-06,
      "loss": 0.0043,
      "step": 99680
    },
    {
      "epoch": 1.6311871062750551,
      "grad_norm": 0.02335083857178688,
      "learning_rate": 5.109464363028932e-06,
      "loss": 0.0013,
      "step": 99690
    },
    {
      "epoch": 1.6313507322261311,
      "grad_norm": 0.07093873620033264,
      "learning_rate": 5.108512649129064e-06,
      "loss": 0.0011,
      "step": 99700
    },
    {
      "epoch": 1.631514358177207,
      "grad_norm": 0.0666184350848198,
      "learning_rate": 5.107560931295873e-06,
      "loss": 0.0018,
      "step": 99710
    },
    {
      "epoch": 1.6316779841282827,
      "grad_norm": 0.07024268805980682,
      "learning_rate": 5.106609209563857e-06,
      "loss": 0.0015,
      "step": 99720
    },
    {
      "epoch": 1.6318416100793587,
      "grad_norm": 0.10046453773975372,
      "learning_rate": 5.105657483967513e-06,
      "loss": 0.0016,
      "step": 99730
    },
    {
      "epoch": 1.6320052360304345,
      "grad_norm": 0.1354244500398636,
      "learning_rate": 5.104705754541338e-06,
      "loss": 0.0019,
      "step": 99740
    },
    {
      "epoch": 1.6321688619815102,
      "grad_norm": 0.00312987738288939,
      "learning_rate": 5.103754021319833e-06,
      "loss": 0.0019,
      "step": 99750
    },
    {
      "epoch": 1.6323324879325862,
      "grad_norm": 0.1013323962688446,
      "learning_rate": 5.102802284337495e-06,
      "loss": 0.0027,
      "step": 99760
    },
    {
      "epoch": 1.632496113883662,
      "grad_norm": 0.08389976620674133,
      "learning_rate": 5.101850543628819e-06,
      "loss": 0.0023,
      "step": 99770
    },
    {
      "epoch": 1.6326597398347378,
      "grad_norm": 0.07518161833286285,
      "learning_rate": 5.100898799228307e-06,
      "loss": 0.0011,
      "step": 99780
    },
    {
      "epoch": 1.6328233657858138,
      "grad_norm": 0.16343435645103455,
      "learning_rate": 5.099947051170455e-06,
      "loss": 0.0011,
      "step": 99790
    },
    {
      "epoch": 1.6329869917368893,
      "grad_norm": 0.04742646962404251,
      "learning_rate": 5.098995299489763e-06,
      "loss": 0.0018,
      "step": 99800
    },
    {
      "epoch": 1.6331506176879653,
      "grad_norm": 0.02837369218468666,
      "learning_rate": 5.098043544220729e-06,
      "loss": 0.0022,
      "step": 99810
    },
    {
      "epoch": 1.633314243639041,
      "grad_norm": 0.014262927696108818,
      "learning_rate": 5.097091785397853e-06,
      "loss": 0.001,
      "step": 99820
    },
    {
      "epoch": 1.6334778695901169,
      "grad_norm": 0.2345682978630066,
      "learning_rate": 5.0961400230556325e-06,
      "loss": 0.0023,
      "step": 99830
    },
    {
      "epoch": 1.6336414955411929,
      "grad_norm": 0.10950388014316559,
      "learning_rate": 5.095188257228567e-06,
      "loss": 0.0022,
      "step": 99840
    },
    {
      "epoch": 1.6338051214922686,
      "grad_norm": 0.06905578076839447,
      "learning_rate": 5.094236487951156e-06,
      "loss": 0.0012,
      "step": 99850
    },
    {
      "epoch": 1.6339687474433444,
      "grad_norm": 0.16081662476062775,
      "learning_rate": 5.093284715257899e-06,
      "loss": 0.0022,
      "step": 99860
    },
    {
      "epoch": 1.6341323733944204,
      "grad_norm": 0.019619664177298546,
      "learning_rate": 5.092332939183297e-06,
      "loss": 0.0017,
      "step": 99870
    },
    {
      "epoch": 1.6342959993454962,
      "grad_norm": 0.07543018460273743,
      "learning_rate": 5.091381159761844e-06,
      "loss": 0.0016,
      "step": 99880
    },
    {
      "epoch": 1.634459625296572,
      "grad_norm": 0.034987207502126694,
      "learning_rate": 5.090429377028047e-06,
      "loss": 0.001,
      "step": 99890
    },
    {
      "epoch": 1.634623251247648,
      "grad_norm": 0.09814410656690598,
      "learning_rate": 5.089477591016401e-06,
      "loss": 0.0017,
      "step": 99900
    },
    {
      "epoch": 1.6347868771987237,
      "grad_norm": 0.17710573971271515,
      "learning_rate": 5.0885258017614085e-06,
      "loss": 0.0022,
      "step": 99910
    },
    {
      "epoch": 1.6349505031497995,
      "grad_norm": 0.1105768233537674,
      "learning_rate": 5.087574009297568e-06,
      "loss": 0.0017,
      "step": 99920
    },
    {
      "epoch": 1.6351141291008755,
      "grad_norm": 0.13785147666931152,
      "learning_rate": 5.086622213659379e-06,
      "loss": 0.0016,
      "step": 99930
    },
    {
      "epoch": 1.6352777550519513,
      "grad_norm": 0.05918983742594719,
      "learning_rate": 5.0856704148813445e-06,
      "loss": 0.0013,
      "step": 99940
    },
    {
      "epoch": 1.635441381003027,
      "grad_norm": 0.45161759853363037,
      "learning_rate": 5.084718612997962e-06,
      "loss": 0.0029,
      "step": 99950
    },
    {
      "epoch": 1.635605006954103,
      "grad_norm": 0.021274369210004807,
      "learning_rate": 5.083766808043733e-06,
      "loss": 0.001,
      "step": 99960
    },
    {
      "epoch": 1.6357686329051786,
      "grad_norm": 0.11046069115400314,
      "learning_rate": 5.082815000053161e-06,
      "loss": 0.0042,
      "step": 99970
    },
    {
      "epoch": 1.6359322588562546,
      "grad_norm": 0.19795791804790497,
      "learning_rate": 5.081863189060741e-06,
      "loss": 0.0021,
      "step": 99980
    },
    {
      "epoch": 1.6360958848073306,
      "grad_norm": 0.16111892461776733,
      "learning_rate": 5.08091137510098e-06,
      "loss": 0.002,
      "step": 99990
    },
    {
      "epoch": 1.6362595107584061,
      "grad_norm": 0.5397040247917175,
      "learning_rate": 5.079959558208375e-06,
      "loss": 0.0009,
      "step": 100000
    },
    {
      "epoch": 1.6362595107584061,
      "eval_loss": 0.0012626892421394587,
      "eval_runtime": 3.0965,
      "eval_samples_per_second": 64.59,
      "eval_steps_per_second": 16.147,
      "step": 100000
    },
    {
      "epoch": 1.6364231367094821,
      "grad_norm": 0.05758707970380783,
      "learning_rate": 5.079007738417429e-06,
      "loss": 0.0018,
      "step": 100010
    },
    {
      "epoch": 1.636586762660558,
      "grad_norm": 0.2286084145307541,
      "learning_rate": 5.0780559157626405e-06,
      "loss": 0.0014,
      "step": 100020
    },
    {
      "epoch": 1.6367503886116337,
      "grad_norm": 0.04949363321065903,
      "learning_rate": 5.077104090278514e-06,
      "loss": 0.0015,
      "step": 100030
    },
    {
      "epoch": 1.6369140145627097,
      "grad_norm": 0.05404341220855713,
      "learning_rate": 5.07615226199955e-06,
      "loss": 0.0015,
      "step": 100040
    },
    {
      "epoch": 1.6370776405137855,
      "grad_norm": 0.1697544902563095,
      "learning_rate": 5.075200430960248e-06,
      "loss": 0.0018,
      "step": 100050
    },
    {
      "epoch": 1.6372412664648612,
      "grad_norm": 0.1308528035879135,
      "learning_rate": 5.074248597195112e-06,
      "loss": 0.002,
      "step": 100060
    },
    {
      "epoch": 1.6374048924159372,
      "grad_norm": 0.19643068313598633,
      "learning_rate": 5.073296760738644e-06,
      "loss": 0.0024,
      "step": 100070
    },
    {
      "epoch": 1.637568518367013,
      "grad_norm": 0.0887424498796463,
      "learning_rate": 5.072344921625343e-06,
      "loss": 0.0015,
      "step": 100080
    },
    {
      "epoch": 1.6377321443180888,
      "grad_norm": 0.16684751212596893,
      "learning_rate": 5.071393079889713e-06,
      "loss": 0.0011,
      "step": 100090
    },
    {
      "epoch": 1.6378957702691648,
      "grad_norm": 0.07709401845932007,
      "learning_rate": 5.070441235566254e-06,
      "loss": 0.0019,
      "step": 100100
    },
    {
      "epoch": 1.6380593962202405,
      "grad_norm": 0.06361225992441177,
      "learning_rate": 5.06948938868947e-06,
      "loss": 0.0029,
      "step": 100110
    },
    {
      "epoch": 1.6382230221713163,
      "grad_norm": 0.1230153813958168,
      "learning_rate": 5.068537539293865e-06,
      "loss": 0.0034,
      "step": 100120
    },
    {
      "epoch": 1.6383866481223923,
      "grad_norm": 0.12074223905801773,
      "learning_rate": 5.067585687413937e-06,
      "loss": 0.0022,
      "step": 100130
    },
    {
      "epoch": 1.638550274073468,
      "grad_norm": 0.03241406008601189,
      "learning_rate": 5.066633833084191e-06,
      "loss": 0.0011,
      "step": 100140
    },
    {
      "epoch": 1.6387139000245439,
      "grad_norm": 0.594880998134613,
      "learning_rate": 5.065681976339129e-06,
      "loss": 0.0039,
      "step": 100150
    },
    {
      "epoch": 1.6388775259756199,
      "grad_norm": 0.057453226298093796,
      "learning_rate": 5.064730117213252e-06,
      "loss": 0.0014,
      "step": 100160
    },
    {
      "epoch": 1.6390411519266954,
      "grad_norm": 0.11164303123950958,
      "learning_rate": 5.063778255741064e-06,
      "loss": 0.0015,
      "step": 100170
    },
    {
      "epoch": 1.6392047778777714,
      "grad_norm": 0.09474091231822968,
      "learning_rate": 5.062826391957068e-06,
      "loss": 0.0011,
      "step": 100180
    },
    {
      "epoch": 1.6393684038288474,
      "grad_norm": 0.005235459189862013,
      "learning_rate": 5.0618745258957666e-06,
      "loss": 0.0024,
      "step": 100190
    },
    {
      "epoch": 1.639532029779923,
      "grad_norm": 0.039148926734924316,
      "learning_rate": 5.0609226575916614e-06,
      "loss": 0.0023,
      "step": 100200
    },
    {
      "epoch": 1.639695655730999,
      "grad_norm": 0.08107282221317291,
      "learning_rate": 5.059970787079257e-06,
      "loss": 0.0012,
      "step": 100210
    },
    {
      "epoch": 1.6398592816820747,
      "grad_norm": 0.13695508241653442,
      "learning_rate": 5.059018914393057e-06,
      "loss": 0.0017,
      "step": 100220
    },
    {
      "epoch": 1.6400229076331505,
      "grad_norm": 0.14939555525779724,
      "learning_rate": 5.0580670395675615e-06,
      "loss": 0.0016,
      "step": 100230
    },
    {
      "epoch": 1.6401865335842265,
      "grad_norm": 0.030955970287322998,
      "learning_rate": 5.057115162637274e-06,
      "loss": 0.0014,
      "step": 100240
    },
    {
      "epoch": 1.6403501595353023,
      "grad_norm": 0.08128859102725983,
      "learning_rate": 5.056163283636701e-06,
      "loss": 0.0013,
      "step": 100250
    },
    {
      "epoch": 1.640513785486378,
      "grad_norm": 0.035780806094408035,
      "learning_rate": 5.055211402600344e-06,
      "loss": 0.0021,
      "step": 100260
    },
    {
      "epoch": 1.640677411437454,
      "grad_norm": 0.13452747464179993,
      "learning_rate": 5.054259519562706e-06,
      "loss": 0.0015,
      "step": 100270
    },
    {
      "epoch": 1.6408410373885298,
      "grad_norm": 0.08999369293451309,
      "learning_rate": 5.053307634558291e-06,
      "loss": 0.0019,
      "step": 100280
    },
    {
      "epoch": 1.6410046633396056,
      "grad_norm": 0.04026127606630325,
      "learning_rate": 5.0523557476216025e-06,
      "loss": 0.0011,
      "step": 100290
    },
    {
      "epoch": 1.6411682892906816,
      "grad_norm": 0.07283289730548859,
      "learning_rate": 5.051403858787144e-06,
      "loss": 0.0024,
      "step": 100300
    },
    {
      "epoch": 1.6413319152417574,
      "grad_norm": 0.018817242234945297,
      "learning_rate": 5.050451968089418e-06,
      "loss": 0.0022,
      "step": 100310
    },
    {
      "epoch": 1.6414955411928331,
      "grad_norm": 0.021105105057358742,
      "learning_rate": 5.0495000755629305e-06,
      "loss": 0.0015,
      "step": 100320
    },
    {
      "epoch": 1.6416591671439091,
      "grad_norm": 0.06151590123772621,
      "learning_rate": 5.048548181242185e-06,
      "loss": 0.0013,
      "step": 100330
    },
    {
      "epoch": 1.641822793094985,
      "grad_norm": 0.09511099010705948,
      "learning_rate": 5.047596285161683e-06,
      "loss": 0.0018,
      "step": 100340
    },
    {
      "epoch": 1.6419864190460607,
      "grad_norm": 0.1867058277130127,
      "learning_rate": 5.046644387355931e-06,
      "loss": 0.0024,
      "step": 100350
    },
    {
      "epoch": 1.6421500449971367,
      "grad_norm": 0.12276323139667511,
      "learning_rate": 5.045692487859431e-06,
      "loss": 0.0021,
      "step": 100360
    },
    {
      "epoch": 1.6423136709482122,
      "grad_norm": 0.09928416460752487,
      "learning_rate": 5.04474058670669e-06,
      "loss": 0.0015,
      "step": 100370
    },
    {
      "epoch": 1.6424772968992882,
      "grad_norm": 0.1165539100766182,
      "learning_rate": 5.0437886839322085e-06,
      "loss": 0.0017,
      "step": 100380
    },
    {
      "epoch": 1.6426409228503642,
      "grad_norm": 0.049185093492269516,
      "learning_rate": 5.042836779570493e-06,
      "loss": 0.0011,
      "step": 100390
    },
    {
      "epoch": 1.6428045488014398,
      "grad_norm": 0.06552588194608688,
      "learning_rate": 5.041884873656048e-06,
      "loss": 0.0018,
      "step": 100400
    },
    {
      "epoch": 1.6429681747525158,
      "grad_norm": 0.05554191768169403,
      "learning_rate": 5.040932966223375e-06,
      "loss": 0.0021,
      "step": 100410
    },
    {
      "epoch": 1.6431318007035915,
      "grad_norm": 0.19864696264266968,
      "learning_rate": 5.039981057306983e-06,
      "loss": 0.0028,
      "step": 100420
    },
    {
      "epoch": 1.6432954266546673,
      "grad_norm": 0.2563404440879822,
      "learning_rate": 5.039029146941372e-06,
      "loss": 0.0019,
      "step": 100430
    },
    {
      "epoch": 1.6434590526057433,
      "grad_norm": 0.04238563030958176,
      "learning_rate": 5.0380772351610495e-06,
      "loss": 0.0022,
      "step": 100440
    },
    {
      "epoch": 1.643622678556819,
      "grad_norm": 0.19535382091999054,
      "learning_rate": 5.037125322000518e-06,
      "loss": 0.003,
      "step": 100450
    },
    {
      "epoch": 1.6437863045078949,
      "grad_norm": 0.06673549115657806,
      "learning_rate": 5.036173407494282e-06,
      "loss": 0.0011,
      "step": 100460
    },
    {
      "epoch": 1.6439499304589709,
      "grad_norm": 0.08663325756788254,
      "learning_rate": 5.035221491676849e-06,
      "loss": 0.0018,
      "step": 100470
    },
    {
      "epoch": 1.6441135564100466,
      "grad_norm": 0.08847000449895859,
      "learning_rate": 5.034269574582718e-06,
      "loss": 0.002,
      "step": 100480
    },
    {
      "epoch": 1.6442771823611224,
      "grad_norm": 0.017689118161797523,
      "learning_rate": 5.0333176562464004e-06,
      "loss": 0.0012,
      "step": 100490
    },
    {
      "epoch": 1.6444408083121984,
      "grad_norm": 0.10258142650127411,
      "learning_rate": 5.032365736702397e-06,
      "loss": 0.0018,
      "step": 100500
    },
    {
      "epoch": 1.6446044342632742,
      "grad_norm": 0.1761052906513214,
      "learning_rate": 5.031413815985213e-06,
      "loss": 0.0018,
      "step": 100510
    },
    {
      "epoch": 1.64476806021435,
      "grad_norm": 0.02352570742368698,
      "learning_rate": 5.030461894129353e-06,
      "loss": 0.0008,
      "step": 100520
    },
    {
      "epoch": 1.644931686165426,
      "grad_norm": 0.32751473784446716,
      "learning_rate": 5.029509971169325e-06,
      "loss": 0.0021,
      "step": 100530
    },
    {
      "epoch": 1.6450953121165017,
      "grad_norm": 0.17443785071372986,
      "learning_rate": 5.02855804713963e-06,
      "loss": 0.002,
      "step": 100540
    },
    {
      "epoch": 1.6452589380675775,
      "grad_norm": 0.11152517795562744,
      "learning_rate": 5.027606122074774e-06,
      "loss": 0.003,
      "step": 100550
    },
    {
      "epoch": 1.6454225640186535,
      "grad_norm": 0.04345415160059929,
      "learning_rate": 5.026654196009263e-06,
      "loss": 0.0014,
      "step": 100560
    },
    {
      "epoch": 1.645586189969729,
      "grad_norm": 0.32868242263793945,
      "learning_rate": 5.025702268977602e-06,
      "loss": 0.0014,
      "step": 100570
    },
    {
      "epoch": 1.645749815920805,
      "grad_norm": 0.03208612650632858,
      "learning_rate": 5.024750341014295e-06,
      "loss": 0.0006,
      "step": 100580
    },
    {
      "epoch": 1.645913441871881,
      "grad_norm": 0.07968027144670486,
      "learning_rate": 5.023798412153847e-06,
      "loss": 0.0027,
      "step": 100590
    },
    {
      "epoch": 1.6460770678229566,
      "grad_norm": 0.11156504601240158,
      "learning_rate": 5.022846482430763e-06,
      "loss": 0.0014,
      "step": 100600
    },
    {
      "epoch": 1.6462406937740326,
      "grad_norm": 0.03609035536646843,
      "learning_rate": 5.021894551879552e-06,
      "loss": 0.0018,
      "step": 100610
    },
    {
      "epoch": 1.6464043197251084,
      "grad_norm": 0.07954364269971848,
      "learning_rate": 5.020942620534715e-06,
      "loss": 0.0015,
      "step": 100620
    },
    {
      "epoch": 1.6465679456761841,
      "grad_norm": 0.03949743136763573,
      "learning_rate": 5.019990688430757e-06,
      "loss": 0.0013,
      "step": 100630
    },
    {
      "epoch": 1.6467315716272601,
      "grad_norm": 0.32489776611328125,
      "learning_rate": 5.019038755602184e-06,
      "loss": 0.0021,
      "step": 100640
    },
    {
      "epoch": 1.646895197578336,
      "grad_norm": 0.0025795656256377697,
      "learning_rate": 5.018086822083504e-06,
      "loss": 0.0016,
      "step": 100650
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 0.13780875504016876,
      "learning_rate": 5.017134887909219e-06,
      "loss": 0.0013,
      "step": 100660
    },
    {
      "epoch": 1.6472224494804877,
      "grad_norm": 0.017130527645349503,
      "learning_rate": 5.016182953113836e-06,
      "loss": 0.002,
      "step": 100670
    },
    {
      "epoch": 1.6473860754315635,
      "grad_norm": 0.14699749648571014,
      "learning_rate": 5.015231017731859e-06,
      "loss": 0.0031,
      "step": 100680
    },
    {
      "epoch": 1.6475497013826392,
      "grad_norm": 0.16536717116832733,
      "learning_rate": 5.014279081797795e-06,
      "loss": 0.0011,
      "step": 100690
    },
    {
      "epoch": 1.6477133273337152,
      "grad_norm": 0.0576251819729805,
      "learning_rate": 5.013327145346148e-06,
      "loss": 0.0025,
      "step": 100700
    },
    {
      "epoch": 1.647876953284791,
      "grad_norm": 0.05248529464006424,
      "learning_rate": 5.0123752084114254e-06,
      "loss": 0.0019,
      "step": 100710
    },
    {
      "epoch": 1.6480405792358668,
      "grad_norm": 0.25557661056518555,
      "learning_rate": 5.011423271028131e-06,
      "loss": 0.0024,
      "step": 100720
    },
    {
      "epoch": 1.6482042051869428,
      "grad_norm": 0.24020500481128693,
      "learning_rate": 5.01047133323077e-06,
      "loss": 0.0025,
      "step": 100730
    },
    {
      "epoch": 1.6483678311380183,
      "grad_norm": 0.06769675016403198,
      "learning_rate": 5.009519395053847e-06,
      "loss": 0.0029,
      "step": 100740
    },
    {
      "epoch": 1.6485314570890943,
      "grad_norm": 0.09917227178812027,
      "learning_rate": 5.00856745653187e-06,
      "loss": 0.0007,
      "step": 100750
    },
    {
      "epoch": 1.6486950830401703,
      "grad_norm": 0.051329903304576874,
      "learning_rate": 5.007615517699342e-06,
      "loss": 0.0024,
      "step": 100760
    },
    {
      "epoch": 1.6488587089912459,
      "grad_norm": 0.12798112630844116,
      "learning_rate": 5.006663578590772e-06,
      "loss": 0.0014,
      "step": 100770
    },
    {
      "epoch": 1.6490223349423219,
      "grad_norm": 0.14152054488658905,
      "learning_rate": 5.005711639240664e-06,
      "loss": 0.0017,
      "step": 100780
    },
    {
      "epoch": 1.6491859608933976,
      "grad_norm": 0.05302917957305908,
      "learning_rate": 5.00475969968352e-06,
      "loss": 0.001,
      "step": 100790
    },
    {
      "epoch": 1.6493495868444734,
      "grad_norm": 0.13204002380371094,
      "learning_rate": 5.003807759953849e-06,
      "loss": 0.0011,
      "step": 100800
    },
    {
      "epoch": 1.6495132127955494,
      "grad_norm": 0.0041569992899894714,
      "learning_rate": 5.002855820086156e-06,
      "loss": 0.0021,
      "step": 100810
    },
    {
      "epoch": 1.6496768387466252,
      "grad_norm": 0.15092462301254272,
      "learning_rate": 5.0019038801149475e-06,
      "loss": 0.0018,
      "step": 100820
    },
    {
      "epoch": 1.649840464697701,
      "grad_norm": 0.08255352824926376,
      "learning_rate": 5.000951940074727e-06,
      "loss": 0.0017,
      "step": 100830
    },
    {
      "epoch": 1.650004090648777,
      "grad_norm": 0.05293919891119003,
      "learning_rate": 5e-06,
      "loss": 0.0014,
      "step": 100840
    },
    {
      "epoch": 1.6501677165998527,
      "grad_norm": 0.03797246143221855,
      "learning_rate": 4.999048059925274e-06,
      "loss": 0.0015,
      "step": 100850
    },
    {
      "epoch": 1.6503313425509285,
      "grad_norm": 0.04087596386671066,
      "learning_rate": 4.998096119885054e-06,
      "loss": 0.0011,
      "step": 100860
    },
    {
      "epoch": 1.6504949685020045,
      "grad_norm": 0.07037179172039032,
      "learning_rate": 4.997144179913845e-06,
      "loss": 0.0017,
      "step": 100870
    },
    {
      "epoch": 1.6506585944530803,
      "grad_norm": 0.13170303404331207,
      "learning_rate": 4.996192240046152e-06,
      "loss": 0.002,
      "step": 100880
    },
    {
      "epoch": 1.650822220404156,
      "grad_norm": 0.06456456333398819,
      "learning_rate": 4.995240300316481e-06,
      "loss": 0.0015,
      "step": 100890
    },
    {
      "epoch": 1.650985846355232,
      "grad_norm": 0.16258928179740906,
      "learning_rate": 4.9942883607593395e-06,
      "loss": 0.003,
      "step": 100900
    },
    {
      "epoch": 1.6511494723063078,
      "grad_norm": 0.05388384312391281,
      "learning_rate": 4.99333642140923e-06,
      "loss": 0.0026,
      "step": 100910
    },
    {
      "epoch": 1.6513130982573836,
      "grad_norm": 0.3144051432609558,
      "learning_rate": 4.992384482300659e-06,
      "loss": 0.0026,
      "step": 100920
    },
    {
      "epoch": 1.6514767242084596,
      "grad_norm": 0.10867591202259064,
      "learning_rate": 4.991432543468132e-06,
      "loss": 0.0011,
      "step": 100930
    },
    {
      "epoch": 1.6516403501595351,
      "grad_norm": 0.20788510143756866,
      "learning_rate": 4.990480604946155e-06,
      "loss": 0.0033,
      "step": 100940
    },
    {
      "epoch": 1.6518039761106111,
      "grad_norm": 0.07863085716962814,
      "learning_rate": 4.989528666769232e-06,
      "loss": 0.0015,
      "step": 100950
    },
    {
      "epoch": 1.6519676020616871,
      "grad_norm": 0.07487991452217102,
      "learning_rate": 4.988576728971871e-06,
      "loss": 0.0016,
      "step": 100960
    },
    {
      "epoch": 1.6521312280127627,
      "grad_norm": 0.11730311810970306,
      "learning_rate": 4.987624791588575e-06,
      "loss": 0.0016,
      "step": 100970
    },
    {
      "epoch": 1.6522948539638387,
      "grad_norm": 0.07989366352558136,
      "learning_rate": 4.986672854653852e-06,
      "loss": 0.0019,
      "step": 100980
    },
    {
      "epoch": 1.6524584799149145,
      "grad_norm": 0.03052920289337635,
      "learning_rate": 4.985720918202205e-06,
      "loss": 0.0018,
      "step": 100990
    },
    {
      "epoch": 1.6526221058659902,
      "grad_norm": 0.060879986733198166,
      "learning_rate": 4.984768982268143e-06,
      "loss": 0.0011,
      "step": 101000
    },
    {
      "epoch": 1.6527857318170662,
      "grad_norm": 0.22433774173259735,
      "learning_rate": 4.983817046886165e-06,
      "loss": 0.0015,
      "step": 101010
    },
    {
      "epoch": 1.652949357768142,
      "grad_norm": 0.08787201344966888,
      "learning_rate": 4.982865112090782e-06,
      "loss": 0.0018,
      "step": 101020
    },
    {
      "epoch": 1.6531129837192178,
      "grad_norm": 0.051112495362758636,
      "learning_rate": 4.981913177916498e-06,
      "loss": 0.0015,
      "step": 101030
    },
    {
      "epoch": 1.6532766096702938,
      "grad_norm": 0.24164657294750214,
      "learning_rate": 4.9809612443978165e-06,
      "loss": 0.0021,
      "step": 101040
    },
    {
      "epoch": 1.6534402356213695,
      "grad_norm": 0.010544318705797195,
      "learning_rate": 4.9800093115692444e-06,
      "loss": 0.0016,
      "step": 101050
    },
    {
      "epoch": 1.6536038615724453,
      "grad_norm": 0.12615899741649628,
      "learning_rate": 4.979057379465287e-06,
      "loss": 0.0013,
      "step": 101060
    },
    {
      "epoch": 1.6537674875235213,
      "grad_norm": 0.02685977704823017,
      "learning_rate": 4.978105448120449e-06,
      "loss": 0.0014,
      "step": 101070
    },
    {
      "epoch": 1.653931113474597,
      "grad_norm": 0.022763237357139587,
      "learning_rate": 4.977153517569237e-06,
      "loss": 0.0005,
      "step": 101080
    },
    {
      "epoch": 1.6540947394256729,
      "grad_norm": 0.4035637080669403,
      "learning_rate": 4.9762015878461536e-06,
      "loss": 0.0013,
      "step": 101090
    },
    {
      "epoch": 1.6542583653767489,
      "grad_norm": 0.020210569724440575,
      "learning_rate": 4.975249658985707e-06,
      "loss": 0.0024,
      "step": 101100
    },
    {
      "epoch": 1.6544219913278246,
      "grad_norm": 0.006763089913874865,
      "learning_rate": 4.9742977310224e-06,
      "loss": 0.0015,
      "step": 101110
    },
    {
      "epoch": 1.6545856172789004,
      "grad_norm": 0.23854990303516388,
      "learning_rate": 4.9733458039907386e-06,
      "loss": 0.0026,
      "step": 101120
    },
    {
      "epoch": 1.6547492432299764,
      "grad_norm": 0.12342479079961777,
      "learning_rate": 4.972393877925227e-06,
      "loss": 0.0015,
      "step": 101130
    },
    {
      "epoch": 1.654912869181052,
      "grad_norm": 0.05465370789170265,
      "learning_rate": 4.971441952860372e-06,
      "loss": 0.0023,
      "step": 101140
    },
    {
      "epoch": 1.655076495132128,
      "grad_norm": 0.10789911448955536,
      "learning_rate": 4.970490028830677e-06,
      "loss": 0.0014,
      "step": 101150
    },
    {
      "epoch": 1.655240121083204,
      "grad_norm": 0.02612781710922718,
      "learning_rate": 4.9695381058706475e-06,
      "loss": 0.0012,
      "step": 101160
    },
    {
      "epoch": 1.6554037470342795,
      "grad_norm": 0.1010490208864212,
      "learning_rate": 4.968586184014787e-06,
      "loss": 0.0027,
      "step": 101170
    },
    {
      "epoch": 1.6555673729853555,
      "grad_norm": 0.018832700327038765,
      "learning_rate": 4.967634263297603e-06,
      "loss": 0.0012,
      "step": 101180
    },
    {
      "epoch": 1.6557309989364313,
      "grad_norm": 0.13586290180683136,
      "learning_rate": 4.9666823437535995e-06,
      "loss": 0.0019,
      "step": 101190
    },
    {
      "epoch": 1.655894624887507,
      "grad_norm": 0.10224128514528275,
      "learning_rate": 4.965730425417284e-06,
      "loss": 0.0012,
      "step": 101200
    },
    {
      "epoch": 1.656058250838583,
      "grad_norm": 0.02834460139274597,
      "learning_rate": 4.964778508323155e-06,
      "loss": 0.0009,
      "step": 101210
    },
    {
      "epoch": 1.6562218767896588,
      "grad_norm": 0.1569192260503769,
      "learning_rate": 4.963826592505721e-06,
      "loss": 0.0011,
      "step": 101220
    },
    {
      "epoch": 1.6563855027407346,
      "grad_norm": 0.1372176855802536,
      "learning_rate": 4.962874677999484e-06,
      "loss": 0.0016,
      "step": 101230
    },
    {
      "epoch": 1.6565491286918106,
      "grad_norm": 0.09356169402599335,
      "learning_rate": 4.961922764838952e-06,
      "loss": 0.0017,
      "step": 101240
    },
    {
      "epoch": 1.6567127546428864,
      "grad_norm": 0.12333190441131592,
      "learning_rate": 4.960970853058629e-06,
      "loss": 0.0018,
      "step": 101250
    },
    {
      "epoch": 1.6568763805939621,
      "grad_norm": 0.046497467905282974,
      "learning_rate": 4.960018942693018e-06,
      "loss": 0.0015,
      "step": 101260
    },
    {
      "epoch": 1.6570400065450381,
      "grad_norm": 0.3702888488769531,
      "learning_rate": 4.959067033776625e-06,
      "loss": 0.0021,
      "step": 101270
    },
    {
      "epoch": 1.657203632496114,
      "grad_norm": 0.10438147932291031,
      "learning_rate": 4.958115126343953e-06,
      "loss": 0.002,
      "step": 101280
    },
    {
      "epoch": 1.6573672584471897,
      "grad_norm": 0.11183849722146988,
      "learning_rate": 4.957163220429507e-06,
      "loss": 0.0023,
      "step": 101290
    },
    {
      "epoch": 1.6575308843982657,
      "grad_norm": 0.08231933414936066,
      "learning_rate": 4.956211316067793e-06,
      "loss": 0.0011,
      "step": 101300
    },
    {
      "epoch": 1.6576945103493415,
      "grad_norm": 0.03424077108502388,
      "learning_rate": 4.955259413293312e-06,
      "loss": 0.0013,
      "step": 101310
    },
    {
      "epoch": 1.6578581363004172,
      "grad_norm": 0.14053568243980408,
      "learning_rate": 4.95430751214057e-06,
      "loss": 0.0007,
      "step": 101320
    },
    {
      "epoch": 1.6580217622514932,
      "grad_norm": 0.09763520210981369,
      "learning_rate": 4.9533556126440705e-06,
      "loss": 0.004,
      "step": 101330
    },
    {
      "epoch": 1.6581853882025688,
      "grad_norm": 0.1472080796957016,
      "learning_rate": 4.952403714838319e-06,
      "loss": 0.0013,
      "step": 101340
    },
    {
      "epoch": 1.6583490141536448,
      "grad_norm": 0.19369348883628845,
      "learning_rate": 4.951451818757817e-06,
      "loss": 0.0019,
      "step": 101350
    },
    {
      "epoch": 1.6585126401047208,
      "grad_norm": 0.1374535858631134,
      "learning_rate": 4.95049992443707e-06,
      "loss": 0.0016,
      "step": 101360
    },
    {
      "epoch": 1.6586762660557963,
      "grad_norm": 0.03457542136311531,
      "learning_rate": 4.9495480319105825e-06,
      "loss": 0.0017,
      "step": 101370
    },
    {
      "epoch": 1.6588398920068723,
      "grad_norm": 0.11844110488891602,
      "learning_rate": 4.9485961412128575e-06,
      "loss": 0.0018,
      "step": 101380
    },
    {
      "epoch": 1.659003517957948,
      "grad_norm": 0.06468743085861206,
      "learning_rate": 4.9476442523783974e-06,
      "loss": 0.0014,
      "step": 101390
    },
    {
      "epoch": 1.6591671439090239,
      "grad_norm": 0.15154115855693817,
      "learning_rate": 4.946692365441711e-06,
      "loss": 0.0017,
      "step": 101400
    },
    {
      "epoch": 1.6593307698600999,
      "grad_norm": 0.08270050585269928,
      "learning_rate": 4.945740480437296e-06,
      "loss": 0.0015,
      "step": 101410
    },
    {
      "epoch": 1.6594943958111756,
      "grad_norm": 0.07217680662870407,
      "learning_rate": 4.944788597399658e-06,
      "loss": 0.0017,
      "step": 101420
    },
    {
      "epoch": 1.6596580217622514,
      "grad_norm": 0.13497240841388702,
      "learning_rate": 4.943836716363301e-06,
      "loss": 0.0027,
      "step": 101430
    },
    {
      "epoch": 1.6598216477133274,
      "grad_norm": 0.014526392333209515,
      "learning_rate": 4.942884837362727e-06,
      "loss": 0.0025,
      "step": 101440
    },
    {
      "epoch": 1.6599852736644032,
      "grad_norm": 0.1114325225353241,
      "learning_rate": 4.94193296043244e-06,
      "loss": 0.0084,
      "step": 101450
    },
    {
      "epoch": 1.660148899615479,
      "grad_norm": 0.28763630986213684,
      "learning_rate": 4.9409810856069445e-06,
      "loss": 0.0015,
      "step": 101460
    },
    {
      "epoch": 1.660312525566555,
      "grad_norm": 0.1795928180217743,
      "learning_rate": 4.940029212920743e-06,
      "loss": 0.0009,
      "step": 101470
    },
    {
      "epoch": 1.6604761515176307,
      "grad_norm": 0.03637954220175743,
      "learning_rate": 4.9390773424083385e-06,
      "loss": 0.0011,
      "step": 101480
    },
    {
      "epoch": 1.6606397774687065,
      "grad_norm": 0.06823773682117462,
      "learning_rate": 4.938125474104234e-06,
      "loss": 0.0018,
      "step": 101490
    },
    {
      "epoch": 1.6608034034197825,
      "grad_norm": 0.1163569837808609,
      "learning_rate": 4.937173608042934e-06,
      "loss": 0.0023,
      "step": 101500
    },
    {
      "epoch": 1.6609670293708583,
      "grad_norm": 0.027382541447877884,
      "learning_rate": 4.9362217442589375e-06,
      "loss": 0.0009,
      "step": 101510
    },
    {
      "epoch": 1.661130655321934,
      "grad_norm": 0.1631830930709839,
      "learning_rate": 4.935269882786749e-06,
      "loss": 0.0013,
      "step": 101520
    },
    {
      "epoch": 1.66129428127301,
      "grad_norm": 0.14096766710281372,
      "learning_rate": 4.934318023660873e-06,
      "loss": 0.0018,
      "step": 101530
    },
    {
      "epoch": 1.6614579072240856,
      "grad_norm": 0.12342914193868637,
      "learning_rate": 4.93336616691581e-06,
      "loss": 0.002,
      "step": 101540
    },
    {
      "epoch": 1.6616215331751616,
      "grad_norm": 0.06792972981929779,
      "learning_rate": 4.9324143125860645e-06,
      "loss": 0.0019,
      "step": 101550
    },
    {
      "epoch": 1.6617851591262376,
      "grad_norm": 0.17113278806209564,
      "learning_rate": 4.9314624607061365e-06,
      "loss": 0.0024,
      "step": 101560
    },
    {
      "epoch": 1.6619487850773131,
      "grad_norm": 0.18928660452365875,
      "learning_rate": 4.930510611310529e-06,
      "loss": 0.0018,
      "step": 101570
    },
    {
      "epoch": 1.6621124110283891,
      "grad_norm": 0.021074140444397926,
      "learning_rate": 4.929558764433746e-06,
      "loss": 0.0012,
      "step": 101580
    },
    {
      "epoch": 1.662276036979465,
      "grad_norm": 0.0746341347694397,
      "learning_rate": 4.928606920110288e-06,
      "loss": 0.0009,
      "step": 101590
    },
    {
      "epoch": 1.6624396629305407,
      "grad_norm": 0.24103547632694244,
      "learning_rate": 4.927655078374659e-06,
      "loss": 0.0016,
      "step": 101600
    },
    {
      "epoch": 1.6626032888816167,
      "grad_norm": 0.08127669245004654,
      "learning_rate": 4.926703239261358e-06,
      "loss": 0.0014,
      "step": 101610
    },
    {
      "epoch": 1.6627669148326925,
      "grad_norm": 0.10390322655439377,
      "learning_rate": 4.925751402804889e-06,
      "loss": 0.0012,
      "step": 101620
    },
    {
      "epoch": 1.6629305407837682,
      "grad_norm": 0.06363886594772339,
      "learning_rate": 4.924799569039753e-06,
      "loss": 0.0012,
      "step": 101630
    },
    {
      "epoch": 1.6630941667348442,
      "grad_norm": 0.06226643547415733,
      "learning_rate": 4.923847738000452e-06,
      "loss": 0.0013,
      "step": 101640
    },
    {
      "epoch": 1.66325779268592,
      "grad_norm": 0.2445535510778427,
      "learning_rate": 4.922895909721488e-06,
      "loss": 0.0016,
      "step": 101650
    },
    {
      "epoch": 1.6634214186369958,
      "grad_norm": 0.2508658170700073,
      "learning_rate": 4.92194408423736e-06,
      "loss": 0.0018,
      "step": 101660
    },
    {
      "epoch": 1.6635850445880718,
      "grad_norm": 0.0561600998044014,
      "learning_rate": 4.920992261582572e-06,
      "loss": 0.0015,
      "step": 101670
    },
    {
      "epoch": 1.6637486705391475,
      "grad_norm": 0.04056508466601372,
      "learning_rate": 4.920040441791625e-06,
      "loss": 0.0015,
      "step": 101680
    },
    {
      "epoch": 1.6639122964902233,
      "grad_norm": 0.14251777529716492,
      "learning_rate": 4.91908862489902e-06,
      "loss": 0.0026,
      "step": 101690
    },
    {
      "epoch": 1.6640759224412993,
      "grad_norm": 0.07927603274583817,
      "learning_rate": 4.91813681093926e-06,
      "loss": 0.001,
      "step": 101700
    },
    {
      "epoch": 1.6642395483923749,
      "grad_norm": 0.16103298962116241,
      "learning_rate": 4.917184999946842e-06,
      "loss": 0.0017,
      "step": 101710
    },
    {
      "epoch": 1.6644031743434509,
      "grad_norm": 0.06844595819711685,
      "learning_rate": 4.916233191956268e-06,
      "loss": 0.0028,
      "step": 101720
    },
    {
      "epoch": 1.6645668002945269,
      "grad_norm": 0.05104893445968628,
      "learning_rate": 4.915281387002039e-06,
      "loss": 0.0025,
      "step": 101730
    },
    {
      "epoch": 1.6647304262456024,
      "grad_norm": 0.09660768508911133,
      "learning_rate": 4.914329585118657e-06,
      "loss": 0.0014,
      "step": 101740
    },
    {
      "epoch": 1.6648940521966784,
      "grad_norm": 0.10084972530603409,
      "learning_rate": 4.9133777863406215e-06,
      "loss": 0.0021,
      "step": 101750
    },
    {
      "epoch": 1.6650576781477542,
      "grad_norm": 0.017485352233052254,
      "learning_rate": 4.912425990702434e-06,
      "loss": 0.0012,
      "step": 101760
    },
    {
      "epoch": 1.66522130409883,
      "grad_norm": 0.08894899487495422,
      "learning_rate": 4.911474198238592e-06,
      "loss": 0.0017,
      "step": 101770
    },
    {
      "epoch": 1.665384930049906,
      "grad_norm": 0.007512619718909264,
      "learning_rate": 4.9105224089836e-06,
      "loss": 0.0015,
      "step": 101780
    },
    {
      "epoch": 1.6655485560009817,
      "grad_norm": 0.024128632619976997,
      "learning_rate": 4.909570622971953e-06,
      "loss": 0.0026,
      "step": 101790
    },
    {
      "epoch": 1.6657121819520575,
      "grad_norm": 0.07698648422956467,
      "learning_rate": 4.9086188402381565e-06,
      "loss": 0.0011,
      "step": 101800
    },
    {
      "epoch": 1.6658758079031335,
      "grad_norm": 0.07614482939243317,
      "learning_rate": 4.907667060816706e-06,
      "loss": 0.0018,
      "step": 101810
    },
    {
      "epoch": 1.6660394338542093,
      "grad_norm": 0.05847081169486046,
      "learning_rate": 4.906715284742102e-06,
      "loss": 0.0013,
      "step": 101820
    },
    {
      "epoch": 1.666203059805285,
      "grad_norm": 0.12756794691085815,
      "learning_rate": 4.905763512048845e-06,
      "loss": 0.0012,
      "step": 101830
    },
    {
      "epoch": 1.666366685756361,
      "grad_norm": 0.18289168179035187,
      "learning_rate": 4.904811742771434e-06,
      "loss": 0.0015,
      "step": 101840
    },
    {
      "epoch": 1.6665303117074368,
      "grad_norm": 0.2453628033399582,
      "learning_rate": 4.903859976944369e-06,
      "loss": 0.0013,
      "step": 101850
    },
    {
      "epoch": 1.6666939376585126,
      "grad_norm": 0.1711772233247757,
      "learning_rate": 4.9029082146021486e-06,
      "loss": 0.0011,
      "step": 101860
    },
    {
      "epoch": 1.6668575636095886,
      "grad_norm": 0.047105997800827026,
      "learning_rate": 4.901956455779272e-06,
      "loss": 0.0008,
      "step": 101870
    },
    {
      "epoch": 1.6670211895606644,
      "grad_norm": 0.12165622413158417,
      "learning_rate": 4.901004700510238e-06,
      "loss": 0.0019,
      "step": 101880
    },
    {
      "epoch": 1.6671848155117401,
      "grad_norm": 0.06262020766735077,
      "learning_rate": 4.900052948829546e-06,
      "loss": 0.0011,
      "step": 101890
    },
    {
      "epoch": 1.6673484414628161,
      "grad_norm": 0.08097384870052338,
      "learning_rate": 4.899101200771696e-06,
      "loss": 0.0018,
      "step": 101900
    },
    {
      "epoch": 1.6675120674138917,
      "grad_norm": 0.35216423869132996,
      "learning_rate": 4.898149456371184e-06,
      "loss": 0.0017,
      "step": 101910
    },
    {
      "epoch": 1.6676756933649677,
      "grad_norm": 0.15078285336494446,
      "learning_rate": 4.897197715662507e-06,
      "loss": 0.001,
      "step": 101920
    },
    {
      "epoch": 1.6678393193160437,
      "grad_norm": 0.3330957293510437,
      "learning_rate": 4.896245978680169e-06,
      "loss": 0.001,
      "step": 101930
    },
    {
      "epoch": 1.6680029452671192,
      "grad_norm": 0.10701004415750504,
      "learning_rate": 4.895294245458663e-06,
      "loss": 0.0015,
      "step": 101940
    },
    {
      "epoch": 1.6681665712181952,
      "grad_norm": 0.13928097486495972,
      "learning_rate": 4.894342516032488e-06,
      "loss": 0.0034,
      "step": 101950
    },
    {
      "epoch": 1.668330197169271,
      "grad_norm": 0.041342947632074356,
      "learning_rate": 4.893390790436144e-06,
      "loss": 0.0012,
      "step": 101960
    },
    {
      "epoch": 1.6684938231203468,
      "grad_norm": 0.04353650286793709,
      "learning_rate": 4.892439068704128e-06,
      "loss": 0.0011,
      "step": 101970
    },
    {
      "epoch": 1.6686574490714228,
      "grad_norm": 0.37439417839050293,
      "learning_rate": 4.891487350870937e-06,
      "loss": 0.002,
      "step": 101980
    },
    {
      "epoch": 1.6688210750224985,
      "grad_norm": 0.13891646265983582,
      "learning_rate": 4.890535636971069e-06,
      "loss": 0.0023,
      "step": 101990
    },
    {
      "epoch": 1.6689847009735743,
      "grad_norm": 0.06375404447317123,
      "learning_rate": 4.8895839270390215e-06,
      "loss": 0.001,
      "step": 102000
    },
    {
      "epoch": 1.6691483269246503,
      "grad_norm": 0.054322827607393265,
      "learning_rate": 4.888632221109291e-06,
      "loss": 0.0015,
      "step": 102010
    },
    {
      "epoch": 1.669311952875726,
      "grad_norm": 0.15197350084781647,
      "learning_rate": 4.887680519216374e-06,
      "loss": 0.0031,
      "step": 102020
    },
    {
      "epoch": 1.6694755788268019,
      "grad_norm": 0.0697273313999176,
      "learning_rate": 4.886728821394769e-06,
      "loss": 0.0021,
      "step": 102030
    },
    {
      "epoch": 1.6696392047778779,
      "grad_norm": 0.13675615191459656,
      "learning_rate": 4.885777127678972e-06,
      "loss": 0.0007,
      "step": 102040
    },
    {
      "epoch": 1.6698028307289536,
      "grad_norm": 0.34040510654449463,
      "learning_rate": 4.884825438103479e-06,
      "loss": 0.0028,
      "step": 102050
    },
    {
      "epoch": 1.6699664566800294,
      "grad_norm": 0.23215000331401825,
      "learning_rate": 4.883873752702787e-06,
      "loss": 0.0016,
      "step": 102060
    },
    {
      "epoch": 1.6701300826311054,
      "grad_norm": 0.1076582595705986,
      "learning_rate": 4.882922071511393e-06,
      "loss": 0.0012,
      "step": 102070
    },
    {
      "epoch": 1.6702937085821812,
      "grad_norm": 0.1291218400001526,
      "learning_rate": 4.881970394563792e-06,
      "loss": 0.0016,
      "step": 102080
    },
    {
      "epoch": 1.670457334533257,
      "grad_norm": 0.19146296381950378,
      "learning_rate": 4.88101872189448e-06,
      "loss": 0.0023,
      "step": 102090
    },
    {
      "epoch": 1.670620960484333,
      "grad_norm": 0.054804664105176926,
      "learning_rate": 4.880067053537956e-06,
      "loss": 0.0014,
      "step": 102100
    },
    {
      "epoch": 1.6707845864354085,
      "grad_norm": 0.04826841130852699,
      "learning_rate": 4.879115389528711e-06,
      "loss": 0.0016,
      "step": 102110
    },
    {
      "epoch": 1.6709482123864845,
      "grad_norm": 0.04021270573139191,
      "learning_rate": 4.878163729901242e-06,
      "loss": 0.0015,
      "step": 102120
    },
    {
      "epoch": 1.6711118383375605,
      "grad_norm": 0.14630895853042603,
      "learning_rate": 4.877212074690046e-06,
      "loss": 0.0021,
      "step": 102130
    },
    {
      "epoch": 1.671275464288636,
      "grad_norm": 0.042796749621629715,
      "learning_rate": 4.8762604239296165e-06,
      "loss": 0.0016,
      "step": 102140
    },
    {
      "epoch": 1.671439090239712,
      "grad_norm": 0.03952167555689812,
      "learning_rate": 4.87530877765445e-06,
      "loss": 0.0013,
      "step": 102150
    },
    {
      "epoch": 1.6716027161907878,
      "grad_norm": 0.42436733841896057,
      "learning_rate": 4.8743571358990386e-06,
      "loss": 0.0023,
      "step": 102160
    },
    {
      "epoch": 1.6717663421418636,
      "grad_norm": 0.06887306272983551,
      "learning_rate": 4.8734054986978815e-06,
      "loss": 0.0014,
      "step": 102170
    },
    {
      "epoch": 1.6719299680929396,
      "grad_norm": 0.019923284649848938,
      "learning_rate": 4.8724538660854695e-06,
      "loss": 0.0014,
      "step": 102180
    },
    {
      "epoch": 1.6720935940440154,
      "grad_norm": 0.04699859023094177,
      "learning_rate": 4.871502238096299e-06,
      "loss": 0.0012,
      "step": 102190
    },
    {
      "epoch": 1.6722572199950911,
      "grad_norm": 0.04753861203789711,
      "learning_rate": 4.870550614764865e-06,
      "loss": 0.001,
      "step": 102200
    },
    {
      "epoch": 1.6724208459461671,
      "grad_norm": 0.06307676434516907,
      "learning_rate": 4.869598996125658e-06,
      "loss": 0.002,
      "step": 102210
    },
    {
      "epoch": 1.672584471897243,
      "grad_norm": 0.08829619735479355,
      "learning_rate": 4.868647382213174e-06,
      "loss": 0.0027,
      "step": 102220
    },
    {
      "epoch": 1.6727480978483187,
      "grad_norm": 0.05037305876612663,
      "learning_rate": 4.867695773061907e-06,
      "loss": 0.0011,
      "step": 102230
    },
    {
      "epoch": 1.6729117237993947,
      "grad_norm": 0.22658604383468628,
      "learning_rate": 4.8667441687063506e-06,
      "loss": 0.0018,
      "step": 102240
    },
    {
      "epoch": 1.6730753497504705,
      "grad_norm": 0.19167353212833405,
      "learning_rate": 4.865792569180997e-06,
      "loss": 0.0011,
      "step": 102250
    },
    {
      "epoch": 1.6732389757015462,
      "grad_norm": 0.011489463970065117,
      "learning_rate": 4.864840974520341e-06,
      "loss": 0.0008,
      "step": 102260
    },
    {
      "epoch": 1.6734026016526222,
      "grad_norm": 0.1869586706161499,
      "learning_rate": 4.863889384758874e-06,
      "loss": 0.0008,
      "step": 102270
    },
    {
      "epoch": 1.673566227603698,
      "grad_norm": 0.07528888434171677,
      "learning_rate": 4.862937799931091e-06,
      "loss": 0.0018,
      "step": 102280
    },
    {
      "epoch": 1.6737298535547738,
      "grad_norm": 0.09467701613903046,
      "learning_rate": 4.8619862200714826e-06,
      "loss": 0.0012,
      "step": 102290
    },
    {
      "epoch": 1.6738934795058498,
      "grad_norm": 0.1533868908882141,
      "learning_rate": 4.861034645214543e-06,
      "loss": 0.0014,
      "step": 102300
    },
    {
      "epoch": 1.6740571054569253,
      "grad_norm": 0.08616022765636444,
      "learning_rate": 4.860083075394762e-06,
      "loss": 0.0013,
      "step": 102310
    },
    {
      "epoch": 1.6742207314080013,
      "grad_norm": 0.08038025349378586,
      "learning_rate": 4.859131510646634e-06,
      "loss": 0.0019,
      "step": 102320
    },
    {
      "epoch": 1.6743843573590773,
      "grad_norm": 0.19674372673034668,
      "learning_rate": 4.8581799510046495e-06,
      "loss": 0.0013,
      "step": 102330
    },
    {
      "epoch": 1.6745479833101529,
      "grad_norm": 0.07996311783790588,
      "learning_rate": 4.857228396503302e-06,
      "loss": 0.0011,
      "step": 102340
    },
    {
      "epoch": 1.6747116092612289,
      "grad_norm": 0.09866899251937866,
      "learning_rate": 4.856276847177081e-06,
      "loss": 0.0013,
      "step": 102350
    },
    {
      "epoch": 1.6748752352123046,
      "grad_norm": 0.08916238695383072,
      "learning_rate": 4.855325303060479e-06,
      "loss": 0.0014,
      "step": 102360
    },
    {
      "epoch": 1.6750388611633804,
      "grad_norm": 0.01661216840147972,
      "learning_rate": 4.854373764187988e-06,
      "loss": 0.003,
      "step": 102370
    },
    {
      "epoch": 1.6752024871144564,
      "grad_norm": 0.029480399563908577,
      "learning_rate": 4.853422230594096e-06,
      "loss": 0.0013,
      "step": 102380
    },
    {
      "epoch": 1.6753661130655322,
      "grad_norm": 0.07779080420732498,
      "learning_rate": 4.852470702313297e-06,
      "loss": 0.0023,
      "step": 102390
    },
    {
      "epoch": 1.675529739016608,
      "grad_norm": 0.12956365942955017,
      "learning_rate": 4.851519179380082e-06,
      "loss": 0.0017,
      "step": 102400
    },
    {
      "epoch": 1.675693364967684,
      "grad_norm": 0.010274041444063187,
      "learning_rate": 4.850567661828939e-06,
      "loss": 0.0011,
      "step": 102410
    },
    {
      "epoch": 1.6758569909187597,
      "grad_norm": 0.09707976132631302,
      "learning_rate": 4.849616149694358e-06,
      "loss": 0.0019,
      "step": 102420
    },
    {
      "epoch": 1.6760206168698355,
      "grad_norm": 0.11230769008398056,
      "learning_rate": 4.848664643010831e-06,
      "loss": 0.0015,
      "step": 102430
    },
    {
      "epoch": 1.6761842428209115,
      "grad_norm": 0.007492445409297943,
      "learning_rate": 4.847713141812845e-06,
      "loss": 0.0009,
      "step": 102440
    },
    {
      "epoch": 1.6763478687719873,
      "grad_norm": 0.08946806192398071,
      "learning_rate": 4.846761646134894e-06,
      "loss": 0.0013,
      "step": 102450
    },
    {
      "epoch": 1.676511494723063,
      "grad_norm": 0.08170616626739502,
      "learning_rate": 4.845810156011464e-06,
      "loss": 0.0019,
      "step": 102460
    },
    {
      "epoch": 1.676675120674139,
      "grad_norm": 0.11584620177745819,
      "learning_rate": 4.844858671477046e-06,
      "loss": 0.0016,
      "step": 102470
    },
    {
      "epoch": 1.6768387466252146,
      "grad_norm": 0.29062244296073914,
      "learning_rate": 4.843907192566127e-06,
      "loss": 0.0016,
      "step": 102480
    },
    {
      "epoch": 1.6770023725762906,
      "grad_norm": 0.058468639850616455,
      "learning_rate": 4.842955719313198e-06,
      "loss": 0.0028,
      "step": 102490
    },
    {
      "epoch": 1.6771659985273666,
      "grad_norm": 0.19412654638290405,
      "learning_rate": 4.842004251752748e-06,
      "loss": 0.0021,
      "step": 102500
    },
    {
      "epoch": 1.6773296244784421,
      "grad_norm": 0.017403732985258102,
      "learning_rate": 4.841052789919263e-06,
      "loss": 0.0006,
      "step": 102510
    },
    {
      "epoch": 1.6774932504295181,
      "grad_norm": 0.06674040853977203,
      "learning_rate": 4.840101333847232e-06,
      "loss": 0.0012,
      "step": 102520
    },
    {
      "epoch": 1.677656876380594,
      "grad_norm": 0.07726369798183441,
      "learning_rate": 4.839149883571144e-06,
      "loss": 0.0017,
      "step": 102530
    },
    {
      "epoch": 1.6778205023316697,
      "grad_norm": 0.02296977862715721,
      "learning_rate": 4.838198439125485e-06,
      "loss": 0.0009,
      "step": 102540
    },
    {
      "epoch": 1.6779841282827457,
      "grad_norm": 0.1686161309480667,
      "learning_rate": 4.837247000544745e-06,
      "loss": 0.0012,
      "step": 102550
    },
    {
      "epoch": 1.6781477542338215,
      "grad_norm": 0.050670076161623,
      "learning_rate": 4.836295567863409e-06,
      "loss": 0.001,
      "step": 102560
    },
    {
      "epoch": 1.6783113801848972,
      "grad_norm": 0.006474161520600319,
      "learning_rate": 4.835344141115965e-06,
      "loss": 0.0022,
      "step": 102570
    },
    {
      "epoch": 1.6784750061359732,
      "grad_norm": 0.050367601215839386,
      "learning_rate": 4.834392720336901e-06,
      "loss": 0.0017,
      "step": 102580
    },
    {
      "epoch": 1.678638632087049,
      "grad_norm": 0.11805913597345352,
      "learning_rate": 4.833441305560702e-06,
      "loss": 0.0023,
      "step": 102590
    },
    {
      "epoch": 1.6788022580381248,
      "grad_norm": 0.015265713445842266,
      "learning_rate": 4.832489896821858e-06,
      "loss": 0.0013,
      "step": 102600
    },
    {
      "epoch": 1.6789658839892008,
      "grad_norm": 0.13472580909729004,
      "learning_rate": 4.831538494154851e-06,
      "loss": 0.0078,
      "step": 102610
    },
    {
      "epoch": 1.6791295099402765,
      "grad_norm": 0.18250545859336853,
      "learning_rate": 4.830587097594167e-06,
      "loss": 0.0015,
      "step": 102620
    },
    {
      "epoch": 1.6792931358913523,
      "grad_norm": 0.010597911663353443,
      "learning_rate": 4.829635707174295e-06,
      "loss": 0.0015,
      "step": 102630
    },
    {
      "epoch": 1.6794567618424283,
      "grad_norm": 0.3909345269203186,
      "learning_rate": 4.82868432292972e-06,
      "loss": 0.0015,
      "step": 102640
    },
    {
      "epoch": 1.679620387793504,
      "grad_norm": 0.05626668408513069,
      "learning_rate": 4.827732944894925e-06,
      "loss": 0.0019,
      "step": 102650
    },
    {
      "epoch": 1.6797840137445799,
      "grad_norm": 0.10206018388271332,
      "learning_rate": 4.826781573104397e-06,
      "loss": 0.0014,
      "step": 102660
    },
    {
      "epoch": 1.6799476396956559,
      "grad_norm": 0.18538853526115417,
      "learning_rate": 4.82583020759262e-06,
      "loss": 0.0009,
      "step": 102670
    },
    {
      "epoch": 1.6801112656467314,
      "grad_norm": 0.07171491533517838,
      "learning_rate": 4.824878848394081e-06,
      "loss": 0.0022,
      "step": 102680
    },
    {
      "epoch": 1.6802748915978074,
      "grad_norm": 0.07217099517583847,
      "learning_rate": 4.8239274955432615e-06,
      "loss": 0.0012,
      "step": 102690
    },
    {
      "epoch": 1.6804385175488834,
      "grad_norm": 0.013641337864100933,
      "learning_rate": 4.82297614907465e-06,
      "loss": 0.0011,
      "step": 102700
    },
    {
      "epoch": 1.680602143499959,
      "grad_norm": 0.10386582463979721,
      "learning_rate": 4.822024809022725e-06,
      "loss": 0.0015,
      "step": 102710
    },
    {
      "epoch": 1.680765769451035,
      "grad_norm": 0.012440638616681099,
      "learning_rate": 4.821073475421972e-06,
      "loss": 0.0019,
      "step": 102720
    },
    {
      "epoch": 1.6809293954021107,
      "grad_norm": 0.10142288357019424,
      "learning_rate": 4.820122148306878e-06,
      "loss": 0.0014,
      "step": 102730
    },
    {
      "epoch": 1.6810930213531865,
      "grad_norm": 0.06657785922288895,
      "learning_rate": 4.819170827711922e-06,
      "loss": 0.0017,
      "step": 102740
    },
    {
      "epoch": 1.6812566473042625,
      "grad_norm": 0.1728191375732422,
      "learning_rate": 4.818219513671589e-06,
      "loss": 0.0025,
      "step": 102750
    },
    {
      "epoch": 1.6814202732553383,
      "grad_norm": 0.34341153502464294,
      "learning_rate": 4.817268206220362e-06,
      "loss": 0.0013,
      "step": 102760
    },
    {
      "epoch": 1.681583899206414,
      "grad_norm": 0.07850497215986252,
      "learning_rate": 4.816316905392724e-06,
      "loss": 0.0013,
      "step": 102770
    },
    {
      "epoch": 1.68174752515749,
      "grad_norm": 0.1214834600687027,
      "learning_rate": 4.815365611223156e-06,
      "loss": 0.002,
      "step": 102780
    },
    {
      "epoch": 1.6819111511085658,
      "grad_norm": 0.09038961678743362,
      "learning_rate": 4.8144143237461416e-06,
      "loss": 0.0019,
      "step": 102790
    },
    {
      "epoch": 1.6820747770596416,
      "grad_norm": 0.04943402484059334,
      "learning_rate": 4.813463042996162e-06,
      "loss": 0.0022,
      "step": 102800
    },
    {
      "epoch": 1.6822384030107176,
      "grad_norm": 0.40753304958343506,
      "learning_rate": 4.812511769007699e-06,
      "loss": 0.0013,
      "step": 102810
    },
    {
      "epoch": 1.6824020289617934,
      "grad_norm": 0.023161349818110466,
      "learning_rate": 4.811560501815232e-06,
      "loss": 0.0016,
      "step": 102820
    },
    {
      "epoch": 1.6825656549128691,
      "grad_norm": 0.14704051613807678,
      "learning_rate": 4.810609241453245e-06,
      "loss": 0.0013,
      "step": 102830
    },
    {
      "epoch": 1.6827292808639451,
      "grad_norm": 0.14873899519443512,
      "learning_rate": 4.809657987956218e-06,
      "loss": 0.0011,
      "step": 102840
    },
    {
      "epoch": 1.682892906815021,
      "grad_norm": 0.15760017931461334,
      "learning_rate": 4.808706741358631e-06,
      "loss": 0.0012,
      "step": 102850
    },
    {
      "epoch": 1.6830565327660967,
      "grad_norm": 0.10897698253393173,
      "learning_rate": 4.807755501694965e-06,
      "loss": 0.0014,
      "step": 102860
    },
    {
      "epoch": 1.6832201587171727,
      "grad_norm": 0.014204490929841995,
      "learning_rate": 4.8068042689997e-06,
      "loss": 0.001,
      "step": 102870
    },
    {
      "epoch": 1.6833837846682482,
      "grad_norm": 0.20954525470733643,
      "learning_rate": 4.805853043307315e-06,
      "loss": 0.0015,
      "step": 102880
    },
    {
      "epoch": 1.6835474106193242,
      "grad_norm": 0.02612285688519478,
      "learning_rate": 4.804901824652291e-06,
      "loss": 0.0017,
      "step": 102890
    },
    {
      "epoch": 1.6837110365704002,
      "grad_norm": 0.05482814833521843,
      "learning_rate": 4.803950613069107e-06,
      "loss": 0.001,
      "step": 102900
    },
    {
      "epoch": 1.6838746625214758,
      "grad_norm": 0.10835659503936768,
      "learning_rate": 4.8029994085922446e-06,
      "loss": 0.0013,
      "step": 102910
    },
    {
      "epoch": 1.6840382884725518,
      "grad_norm": 0.03488203510642052,
      "learning_rate": 4.802048211256179e-06,
      "loss": 0.0009,
      "step": 102920
    },
    {
      "epoch": 1.6842019144236275,
      "grad_norm": 0.09130261838436127,
      "learning_rate": 4.801097021095388e-06,
      "loss": 0.0016,
      "step": 102930
    },
    {
      "epoch": 1.6843655403747033,
      "grad_norm": 0.06012992560863495,
      "learning_rate": 4.800145838144352e-06,
      "loss": 0.007,
      "step": 102940
    },
    {
      "epoch": 1.6845291663257793,
      "grad_norm": 0.006967310328036547,
      "learning_rate": 4.79919466243755e-06,
      "loss": 0.002,
      "step": 102950
    },
    {
      "epoch": 1.684692792276855,
      "grad_norm": 0.16076301038265228,
      "learning_rate": 4.798243494009459e-06,
      "loss": 0.0019,
      "step": 102960
    },
    {
      "epoch": 1.6848564182279309,
      "grad_norm": 0.03123948909342289,
      "learning_rate": 4.797292332894556e-06,
      "loss": 0.0012,
      "step": 102970
    },
    {
      "epoch": 1.6850200441790069,
      "grad_norm": 0.4742262065410614,
      "learning_rate": 4.79634117912732e-06,
      "loss": 0.0028,
      "step": 102980
    },
    {
      "epoch": 1.6851836701300826,
      "grad_norm": 0.06803572922945023,
      "learning_rate": 4.7953900327422254e-06,
      "loss": 0.0027,
      "step": 102990
    },
    {
      "epoch": 1.6853472960811584,
      "grad_norm": 0.10734101384878159,
      "learning_rate": 4.794438893773751e-06,
      "loss": 0.0021,
      "step": 103000
    },
    {
      "epoch": 1.6855109220322344,
      "grad_norm": 0.08367973566055298,
      "learning_rate": 4.793487762256373e-06,
      "loss": 0.0013,
      "step": 103010
    },
    {
      "epoch": 1.6856745479833102,
      "grad_norm": 0.13024799525737762,
      "learning_rate": 4.792536638224567e-06,
      "loss": 0.002,
      "step": 103020
    },
    {
      "epoch": 1.685838173934386,
      "grad_norm": 0.05504586920142174,
      "learning_rate": 4.791585521712809e-06,
      "loss": 0.0017,
      "step": 103030
    },
    {
      "epoch": 1.686001799885462,
      "grad_norm": 0.13645358383655548,
      "learning_rate": 4.790634412755575e-06,
      "loss": 0.002,
      "step": 103040
    },
    {
      "epoch": 1.6861654258365377,
      "grad_norm": 0.09810586273670197,
      "learning_rate": 4.78968331138734e-06,
      "loss": 0.0016,
      "step": 103050
    },
    {
      "epoch": 1.6863290517876135,
      "grad_norm": 0.19095927476882935,
      "learning_rate": 4.788732217642579e-06,
      "loss": 0.0019,
      "step": 103060
    },
    {
      "epoch": 1.6864926777386895,
      "grad_norm": 0.13595440983772278,
      "learning_rate": 4.787781131555768e-06,
      "loss": 0.0015,
      "step": 103070
    },
    {
      "epoch": 1.686656303689765,
      "grad_norm": 0.004306282382458448,
      "learning_rate": 4.786830053161381e-06,
      "loss": 0.0027,
      "step": 103080
    },
    {
      "epoch": 1.686819929640841,
      "grad_norm": 0.09551531076431274,
      "learning_rate": 4.785878982493893e-06,
      "loss": 0.0028,
      "step": 103090
    },
    {
      "epoch": 1.686983555591917,
      "grad_norm": 0.015349028632044792,
      "learning_rate": 4.784927919587775e-06,
      "loss": 0.0008,
      "step": 103100
    },
    {
      "epoch": 1.6871471815429926,
      "grad_norm": 0.16596737504005432,
      "learning_rate": 4.7839768644775066e-06,
      "loss": 0.0022,
      "step": 103110
    },
    {
      "epoch": 1.6873108074940686,
      "grad_norm": 0.058846212923526764,
      "learning_rate": 4.783025817197555e-06,
      "loss": 0.0023,
      "step": 103120
    },
    {
      "epoch": 1.6874744334451444,
      "grad_norm": 0.02900262549519539,
      "learning_rate": 4.7820747777823965e-06,
      "loss": 0.0018,
      "step": 103130
    },
    {
      "epoch": 1.6876380593962201,
      "grad_norm": 0.056173913180828094,
      "learning_rate": 4.781123746266504e-06,
      "loss": 0.0019,
      "step": 103140
    },
    {
      "epoch": 1.6878016853472961,
      "grad_norm": 0.05425018072128296,
      "learning_rate": 4.780172722684349e-06,
      "loss": 0.0014,
      "step": 103150
    },
    {
      "epoch": 1.687965311298372,
      "grad_norm": 0.08125453442335129,
      "learning_rate": 4.779221707070404e-06,
      "loss": 0.0017,
      "step": 103160
    },
    {
      "epoch": 1.6881289372494477,
      "grad_norm": 0.16048917174339294,
      "learning_rate": 4.778270699459141e-06,
      "loss": 0.0023,
      "step": 103170
    },
    {
      "epoch": 1.6882925632005237,
      "grad_norm": 0.11048179119825363,
      "learning_rate": 4.777319699885034e-06,
      "loss": 0.0016,
      "step": 103180
    },
    {
      "epoch": 1.6884561891515995,
      "grad_norm": 0.034016139805316925,
      "learning_rate": 4.776368708382552e-06,
      "loss": 0.0023,
      "step": 103190
    },
    {
      "epoch": 1.6886198151026752,
      "grad_norm": 0.020225824788212776,
      "learning_rate": 4.775417724986167e-06,
      "loss": 0.0011,
      "step": 103200
    },
    {
      "epoch": 1.6887834410537512,
      "grad_norm": 0.13484328985214233,
      "learning_rate": 4.774466749730351e-06,
      "loss": 0.0017,
      "step": 103210
    },
    {
      "epoch": 1.688947067004827,
      "grad_norm": 0.03486631065607071,
      "learning_rate": 4.773515782649572e-06,
      "loss": 0.0025,
      "step": 103220
    },
    {
      "epoch": 1.6891106929559028,
      "grad_norm": 0.0689317137002945,
      "learning_rate": 4.772564823778301e-06,
      "loss": 0.0016,
      "step": 103230
    },
    {
      "epoch": 1.6892743189069788,
      "grad_norm": 0.08258197456598282,
      "learning_rate": 4.771613873151009e-06,
      "loss": 0.002,
      "step": 103240
    },
    {
      "epoch": 1.6894379448580545,
      "grad_norm": 0.0489678680896759,
      "learning_rate": 4.7706629308021655e-06,
      "loss": 0.0018,
      "step": 103250
    },
    {
      "epoch": 1.6896015708091303,
      "grad_norm": 0.06331710517406464,
      "learning_rate": 4.769711996766239e-06,
      "loss": 0.001,
      "step": 103260
    },
    {
      "epoch": 1.6897651967602063,
      "grad_norm": 0.10209188610315323,
      "learning_rate": 4.768761071077699e-06,
      "loss": 0.001,
      "step": 103270
    },
    {
      "epoch": 1.6899288227112819,
      "grad_norm": 0.048321615904569626,
      "learning_rate": 4.7678101537710145e-06,
      "loss": 0.001,
      "step": 103280
    },
    {
      "epoch": 1.6900924486623579,
      "grad_norm": 0.12960122525691986,
      "learning_rate": 4.766859244880654e-06,
      "loss": 0.003,
      "step": 103290
    },
    {
      "epoch": 1.6902560746134339,
      "grad_norm": 0.05311160162091255,
      "learning_rate": 4.765908344441086e-06,
      "loss": 0.0015,
      "step": 103300
    },
    {
      "epoch": 1.6904197005645094,
      "grad_norm": 0.07875974476337433,
      "learning_rate": 4.764957452486778e-06,
      "loss": 0.0011,
      "step": 103310
    },
    {
      "epoch": 1.6905833265155854,
      "grad_norm": 0.06443706899881363,
      "learning_rate": 4.764006569052198e-06,
      "loss": 0.0012,
      "step": 103320
    },
    {
      "epoch": 1.6907469524666612,
      "grad_norm": 0.04119015485048294,
      "learning_rate": 4.763055694171811e-06,
      "loss": 0.0009,
      "step": 103330
    },
    {
      "epoch": 1.690910578417737,
      "grad_norm": 0.39334192872047424,
      "learning_rate": 4.762104827880086e-06,
      "loss": 0.0029,
      "step": 103340
    },
    {
      "epoch": 1.691074204368813,
      "grad_norm": 0.10048554092645645,
      "learning_rate": 4.76115397021149e-06,
      "loss": 0.0012,
      "step": 103350
    },
    {
      "epoch": 1.6912378303198887,
      "grad_norm": 0.06044529378414154,
      "learning_rate": 4.760203121200488e-06,
      "loss": 0.0009,
      "step": 103360
    },
    {
      "epoch": 1.6914014562709645,
      "grad_norm": 0.07305005192756653,
      "learning_rate": 4.759252280881545e-06,
      "loss": 0.0022,
      "step": 103370
    },
    {
      "epoch": 1.6915650822220405,
      "grad_norm": 0.28686127066612244,
      "learning_rate": 4.75830144928913e-06,
      "loss": 0.0021,
      "step": 103380
    },
    {
      "epoch": 1.6917287081731163,
      "grad_norm": 0.10054086893796921,
      "learning_rate": 4.757350626457706e-06,
      "loss": 0.0022,
      "step": 103390
    },
    {
      "epoch": 1.691892334124192,
      "grad_norm": 0.1459931582212448,
      "learning_rate": 4.756399812421739e-06,
      "loss": 0.0017,
      "step": 103400
    },
    {
      "epoch": 1.692055960075268,
      "grad_norm": 0.06654481589794159,
      "learning_rate": 4.755449007215695e-06,
      "loss": 0.0008,
      "step": 103410
    },
    {
      "epoch": 1.6922195860263438,
      "grad_norm": 0.10007412731647491,
      "learning_rate": 4.754498210874035e-06,
      "loss": 0.0022,
      "step": 103420
    },
    {
      "epoch": 1.6923832119774196,
      "grad_norm": 0.04054652899503708,
      "learning_rate": 4.753547423431224e-06,
      "loss": 0.003,
      "step": 103430
    },
    {
      "epoch": 1.6925468379284956,
      "grad_norm": 0.15338733792304993,
      "learning_rate": 4.752596644921727e-06,
      "loss": 0.0015,
      "step": 103440
    },
    {
      "epoch": 1.6927104638795711,
      "grad_norm": 0.008783353492617607,
      "learning_rate": 4.751645875380007e-06,
      "loss": 0.0012,
      "step": 103450
    },
    {
      "epoch": 1.6928740898306471,
      "grad_norm": 0.04697680473327637,
      "learning_rate": 4.750695114840528e-06,
      "loss": 0.0012,
      "step": 103460
    },
    {
      "epoch": 1.6930377157817231,
      "grad_norm": 0.03068576194345951,
      "learning_rate": 4.749744363337751e-06,
      "loss": 0.0012,
      "step": 103470
    },
    {
      "epoch": 1.6932013417327987,
      "grad_norm": 0.11015503108501434,
      "learning_rate": 4.74879362090614e-06,
      "loss": 0.002,
      "step": 103480
    },
    {
      "epoch": 1.6933649676838747,
      "grad_norm": 0.1084451824426651,
      "learning_rate": 4.747842887580156e-06,
      "loss": 0.0019,
      "step": 103490
    },
    {
      "epoch": 1.6935285936349505,
      "grad_norm": 0.052836913615465164,
      "learning_rate": 4.746892163394261e-06,
      "loss": 0.001,
      "step": 103500
    },
    {
      "epoch": 1.6936922195860262,
      "grad_norm": 0.1374897062778473,
      "learning_rate": 4.745941448382918e-06,
      "loss": 0.0016,
      "step": 103510
    },
    {
      "epoch": 1.6938558455371022,
      "grad_norm": 0.19534315168857574,
      "learning_rate": 4.744990742580586e-06,
      "loss": 0.0013,
      "step": 103520
    },
    {
      "epoch": 1.694019471488178,
      "grad_norm": 0.007172324229031801,
      "learning_rate": 4.744040046021728e-06,
      "loss": 0.0008,
      "step": 103530
    },
    {
      "epoch": 1.6941830974392538,
      "grad_norm": 0.04822874441742897,
      "learning_rate": 4.743089358740801e-06,
      "loss": 0.0014,
      "step": 103540
    },
    {
      "epoch": 1.6943467233903298,
      "grad_norm": 0.0776359960436821,
      "learning_rate": 4.742138680772268e-06,
      "loss": 0.0009,
      "step": 103550
    },
    {
      "epoch": 1.6945103493414055,
      "grad_norm": 0.021961791440844536,
      "learning_rate": 4.741188012150588e-06,
      "loss": 0.0011,
      "step": 103560
    },
    {
      "epoch": 1.6946739752924813,
      "grad_norm": 0.08490357547998428,
      "learning_rate": 4.740237352910221e-06,
      "loss": 0.0022,
      "step": 103570
    },
    {
      "epoch": 1.6948376012435573,
      "grad_norm": 0.08773563802242279,
      "learning_rate": 4.739286703085625e-06,
      "loss": 0.0021,
      "step": 103580
    },
    {
      "epoch": 1.695001227194633,
      "grad_norm": 0.11180084943771362,
      "learning_rate": 4.738336062711259e-06,
      "loss": 0.0029,
      "step": 103590
    },
    {
      "epoch": 1.6951648531457089,
      "grad_norm": 0.22186006605625153,
      "learning_rate": 4.737385431821583e-06,
      "loss": 0.0021,
      "step": 103600
    },
    {
      "epoch": 1.6953284790967849,
      "grad_norm": 0.07085279375314713,
      "learning_rate": 4.736434810451055e-06,
      "loss": 0.0025,
      "step": 103610
    },
    {
      "epoch": 1.6954921050478606,
      "grad_norm": 0.04977083578705788,
      "learning_rate": 4.735484198634131e-06,
      "loss": 0.0012,
      "step": 103620
    },
    {
      "epoch": 1.6956557309989364,
      "grad_norm": 0.08676992356777191,
      "learning_rate": 4.734533596405268e-06,
      "loss": 0.0017,
      "step": 103630
    },
    {
      "epoch": 1.6958193569500124,
      "grad_norm": 0.16149508953094482,
      "learning_rate": 4.733583003798924e-06,
      "loss": 0.0013,
      "step": 103640
    },
    {
      "epoch": 1.695982982901088,
      "grad_norm": 0.022713854908943176,
      "learning_rate": 4.732632420849555e-06,
      "loss": 0.0015,
      "step": 103650
    },
    {
      "epoch": 1.696146608852164,
      "grad_norm": 0.1551360934972763,
      "learning_rate": 4.731681847591619e-06,
      "loss": 0.0024,
      "step": 103660
    },
    {
      "epoch": 1.69631023480324,
      "grad_norm": 0.0020840680226683617,
      "learning_rate": 4.730731284059572e-06,
      "loss": 0.0011,
      "step": 103670
    },
    {
      "epoch": 1.6964738607543155,
      "grad_norm": 0.22082167863845825,
      "learning_rate": 4.729780730287868e-06,
      "loss": 0.002,
      "step": 103680
    },
    {
      "epoch": 1.6966374867053915,
      "grad_norm": 0.05507946014404297,
      "learning_rate": 4.728830186310963e-06,
      "loss": 0.0014,
      "step": 103690
    },
    {
      "epoch": 1.6968011126564673,
      "grad_norm": 0.048816781491041183,
      "learning_rate": 4.7278796521633126e-06,
      "loss": 0.0021,
      "step": 103700
    },
    {
      "epoch": 1.696964738607543,
      "grad_norm": 0.08371330052614212,
      "learning_rate": 4.726929127879372e-06,
      "loss": 0.0019,
      "step": 103710
    },
    {
      "epoch": 1.697128364558619,
      "grad_norm": 0.029956074431538582,
      "learning_rate": 4.725978613493593e-06,
      "loss": 0.0021,
      "step": 103720
    },
    {
      "epoch": 1.6972919905096948,
      "grad_norm": 0.14128124713897705,
      "learning_rate": 4.72502810904043e-06,
      "loss": 0.0019,
      "step": 103730
    },
    {
      "epoch": 1.6974556164607706,
      "grad_norm": 0.15012232959270477,
      "learning_rate": 4.724077614554337e-06,
      "loss": 0.0032,
      "step": 103740
    },
    {
      "epoch": 1.6976192424118466,
      "grad_norm": 0.05193810537457466,
      "learning_rate": 4.723127130069768e-06,
      "loss": 0.0019,
      "step": 103750
    },
    {
      "epoch": 1.6977828683629224,
      "grad_norm": 0.11085796356201172,
      "learning_rate": 4.722176655621174e-06,
      "loss": 0.0016,
      "step": 103760
    },
    {
      "epoch": 1.6979464943139981,
      "grad_norm": 0.2096051573753357,
      "learning_rate": 4.72122619124301e-06,
      "loss": 0.0015,
      "step": 103770
    },
    {
      "epoch": 1.6981101202650741,
      "grad_norm": 0.03283541277050972,
      "learning_rate": 4.720275736969726e-06,
      "loss": 0.0015,
      "step": 103780
    },
    {
      "epoch": 1.69827374621615,
      "grad_norm": 0.01805386133491993,
      "learning_rate": 4.719325292835774e-06,
      "loss": 0.0009,
      "step": 103790
    },
    {
      "epoch": 1.6984373721672257,
      "grad_norm": 0.1322762668132782,
      "learning_rate": 4.718374858875606e-06,
      "loss": 0.0022,
      "step": 103800
    },
    {
      "epoch": 1.6986009981183017,
      "grad_norm": 0.4298863410949707,
      "learning_rate": 4.717424435123673e-06,
      "loss": 0.0026,
      "step": 103810
    },
    {
      "epoch": 1.6987646240693774,
      "grad_norm": 0.008327971212565899,
      "learning_rate": 4.716474021614425e-06,
      "loss": 0.0017,
      "step": 103820
    },
    {
      "epoch": 1.6989282500204532,
      "grad_norm": 0.031137490645051003,
      "learning_rate": 4.715523618382312e-06,
      "loss": 0.0008,
      "step": 103830
    },
    {
      "epoch": 1.6990918759715292,
      "grad_norm": 0.1273033618927002,
      "learning_rate": 4.714573225461783e-06,
      "loss": 0.0015,
      "step": 103840
    },
    {
      "epoch": 1.6992555019226048,
      "grad_norm": 0.10375382751226425,
      "learning_rate": 4.71362284288729e-06,
      "loss": 0.002,
      "step": 103850
    },
    {
      "epoch": 1.6994191278736808,
      "grad_norm": 0.016301585361361504,
      "learning_rate": 4.712672470693279e-06,
      "loss": 0.0009,
      "step": 103860
    },
    {
      "epoch": 1.6995827538247568,
      "grad_norm": 0.23067854344844818,
      "learning_rate": 4.7117221089142e-06,
      "loss": 0.0023,
      "step": 103870
    },
    {
      "epoch": 1.6997463797758323,
      "grad_norm": 0.3417958915233612,
      "learning_rate": 4.710771757584503e-06,
      "loss": 0.0016,
      "step": 103880
    },
    {
      "epoch": 1.6999100057269083,
      "grad_norm": 0.1388198435306549,
      "learning_rate": 4.7098214167386356e-06,
      "loss": 0.0018,
      "step": 103890
    },
    {
      "epoch": 1.700073631677984,
      "grad_norm": 0.11060667037963867,
      "learning_rate": 4.708871086411043e-06,
      "loss": 0.0013,
      "step": 103900
    },
    {
      "epoch": 1.7002372576290599,
      "grad_norm": 0.04264014959335327,
      "learning_rate": 4.707920766636176e-06,
      "loss": 0.0025,
      "step": 103910
    },
    {
      "epoch": 1.7004008835801359,
      "grad_norm": 0.06201804056763649,
      "learning_rate": 4.706970457448478e-06,
      "loss": 0.0006,
      "step": 103920
    },
    {
      "epoch": 1.7005645095312116,
      "grad_norm": 0.09567633271217346,
      "learning_rate": 4.706020158882395e-06,
      "loss": 0.0017,
      "step": 103930
    },
    {
      "epoch": 1.7007281354822874,
      "grad_norm": 0.02220143750309944,
      "learning_rate": 4.7050698709723754e-06,
      "loss": 0.0015,
      "step": 103940
    },
    {
      "epoch": 1.7008917614333634,
      "grad_norm": 0.1430577039718628,
      "learning_rate": 4.704119593752865e-06,
      "loss": 0.0043,
      "step": 103950
    },
    {
      "epoch": 1.7010553873844392,
      "grad_norm": 0.23501190543174744,
      "learning_rate": 4.703169327258307e-06,
      "loss": 0.002,
      "step": 103960
    },
    {
      "epoch": 1.701219013335515,
      "grad_norm": 0.0639282688498497,
      "learning_rate": 4.702219071523149e-06,
      "loss": 0.0022,
      "step": 103970
    },
    {
      "epoch": 1.701382639286591,
      "grad_norm": 0.10013594478368759,
      "learning_rate": 4.701268826581833e-06,
      "loss": 0.0014,
      "step": 103980
    },
    {
      "epoch": 1.7015462652376667,
      "grad_norm": 0.14735466241836548,
      "learning_rate": 4.700318592468804e-06,
      "loss": 0.0018,
      "step": 103990
    },
    {
      "epoch": 1.7017098911887425,
      "grad_norm": 0.04981229081749916,
      "learning_rate": 4.699368369218506e-06,
      "loss": 0.0008,
      "step": 104000
    },
    {
      "epoch": 1.7018735171398185,
      "grad_norm": 0.03957488387823105,
      "learning_rate": 4.698418156865383e-06,
      "loss": 0.0036,
      "step": 104010
    },
    {
      "epoch": 1.7020371430908943,
      "grad_norm": 0.017070377245545387,
      "learning_rate": 4.697467955443876e-06,
      "loss": 0.0017,
      "step": 104020
    },
    {
      "epoch": 1.70220076904197,
      "grad_norm": 0.005074461922049522,
      "learning_rate": 4.6965177649884285e-06,
      "loss": 0.0009,
      "step": 104030
    },
    {
      "epoch": 1.702364394993046,
      "grad_norm": 0.034012116491794586,
      "learning_rate": 4.695567585533483e-06,
      "loss": 0.0013,
      "step": 104040
    },
    {
      "epoch": 1.7025280209441216,
      "grad_norm": 0.055761151015758514,
      "learning_rate": 4.69461741711348e-06,
      "loss": 0.0016,
      "step": 104050
    },
    {
      "epoch": 1.7026916468951976,
      "grad_norm": 0.06119237095117569,
      "learning_rate": 4.693667259762863e-06,
      "loss": 0.0015,
      "step": 104060
    },
    {
      "epoch": 1.7028552728462736,
      "grad_norm": 0.18447674810886383,
      "learning_rate": 4.692717113516071e-06,
      "loss": 0.0006,
      "step": 104070
    },
    {
      "epoch": 1.7030188987973491,
      "grad_norm": 0.1918679177761078,
      "learning_rate": 4.691766978407544e-06,
      "loss": 0.0027,
      "step": 104080
    },
    {
      "epoch": 1.7031825247484251,
      "grad_norm": 0.12115088105201721,
      "learning_rate": 4.690816854471723e-06,
      "loss": 0.0017,
      "step": 104090
    },
    {
      "epoch": 1.703346150699501,
      "grad_norm": 0.010279589332640171,
      "learning_rate": 4.689866741743049e-06,
      "loss": 0.001,
      "step": 104100
    },
    {
      "epoch": 1.7035097766505767,
      "grad_norm": 0.031147800385951996,
      "learning_rate": 4.688916640255962e-06,
      "loss": 0.0013,
      "step": 104110
    },
    {
      "epoch": 1.7036734026016527,
      "grad_norm": 0.1230681762099266,
      "learning_rate": 4.687966550044898e-06,
      "loss": 0.0019,
      "step": 104120
    },
    {
      "epoch": 1.7038370285527284,
      "grad_norm": 0.17274823784828186,
      "learning_rate": 4.687016471144296e-06,
      "loss": 0.001,
      "step": 104130
    },
    {
      "epoch": 1.7040006545038042,
      "grad_norm": 0.04568912833929062,
      "learning_rate": 4.686066403588594e-06,
      "loss": 0.0015,
      "step": 104140
    },
    {
      "epoch": 1.7041642804548802,
      "grad_norm": 0.008773451671004295,
      "learning_rate": 4.68511634741223e-06,
      "loss": 0.001,
      "step": 104150
    },
    {
      "epoch": 1.704327906405956,
      "grad_norm": 0.11072470247745514,
      "learning_rate": 4.684166302649643e-06,
      "loss": 0.0026,
      "step": 104160
    },
    {
      "epoch": 1.7044915323570318,
      "grad_norm": 0.049063943326473236,
      "learning_rate": 4.6832162693352674e-06,
      "loss": 0.0018,
      "step": 104170
    },
    {
      "epoch": 1.7046551583081078,
      "grad_norm": 0.018669793382287025,
      "learning_rate": 4.682266247503542e-06,
      "loss": 0.0011,
      "step": 104180
    },
    {
      "epoch": 1.7048187842591835,
      "grad_norm": 0.09930496662855148,
      "learning_rate": 4.6813162371889e-06,
      "loss": 0.001,
      "step": 104190
    },
    {
      "epoch": 1.7049824102102593,
      "grad_norm": 0.044762734323740005,
      "learning_rate": 4.6803662384257796e-06,
      "loss": 0.0015,
      "step": 104200
    },
    {
      "epoch": 1.7051460361613353,
      "grad_norm": 0.017309872433543205,
      "learning_rate": 4.679416251248615e-06,
      "loss": 0.0019,
      "step": 104210
    },
    {
      "epoch": 1.7053096621124109,
      "grad_norm": 0.04318152368068695,
      "learning_rate": 4.67846627569184e-06,
      "loss": 0.0022,
      "step": 104220
    },
    {
      "epoch": 1.7054732880634869,
      "grad_norm": 0.05442947894334793,
      "learning_rate": 4.6775163117898905e-06,
      "loss": 0.0005,
      "step": 104230
    },
    {
      "epoch": 1.7056369140145629,
      "grad_norm": 0.04039434716105461,
      "learning_rate": 4.6765663595771995e-06,
      "loss": 0.0011,
      "step": 104240
    },
    {
      "epoch": 1.7058005399656384,
      "grad_norm": 0.051713377237319946,
      "learning_rate": 4.6756164190882e-06,
      "loss": 0.0026,
      "step": 104250
    },
    {
      "epoch": 1.7059641659167144,
      "grad_norm": 0.06726450473070145,
      "learning_rate": 4.674666490357326e-06,
      "loss": 0.0019,
      "step": 104260
    },
    {
      "epoch": 1.7061277918677902,
      "grad_norm": 0.01772114261984825,
      "learning_rate": 4.673716573419009e-06,
      "loss": 0.0016,
      "step": 104270
    },
    {
      "epoch": 1.706291417818866,
      "grad_norm": 0.07602676749229431,
      "learning_rate": 4.672766668307682e-06,
      "loss": 0.0016,
      "step": 104280
    },
    {
      "epoch": 1.706455043769942,
      "grad_norm": 0.0698457732796669,
      "learning_rate": 4.671816775057779e-06,
      "loss": 0.0019,
      "step": 104290
    },
    {
      "epoch": 1.7066186697210177,
      "grad_norm": 0.03086455538868904,
      "learning_rate": 4.670866893703727e-06,
      "loss": 0.0017,
      "step": 104300
    },
    {
      "epoch": 1.7067822956720935,
      "grad_norm": 0.09401411563158035,
      "learning_rate": 4.66991702427996e-06,
      "loss": 0.0025,
      "step": 104310
    },
    {
      "epoch": 1.7069459216231695,
      "grad_norm": 0.023286331444978714,
      "learning_rate": 4.668967166820908e-06,
      "loss": 0.0023,
      "step": 104320
    },
    {
      "epoch": 1.7071095475742453,
      "grad_norm": 0.08056782186031342,
      "learning_rate": 4.668017321361e-06,
      "loss": 0.0011,
      "step": 104330
    },
    {
      "epoch": 1.707273173525321,
      "grad_norm": 0.03559701517224312,
      "learning_rate": 4.667067487934665e-06,
      "loss": 0.0014,
      "step": 104340
    },
    {
      "epoch": 1.707436799476397,
      "grad_norm": 0.0799616128206253,
      "learning_rate": 4.666117666576335e-06,
      "loss": 0.0014,
      "step": 104350
    },
    {
      "epoch": 1.7076004254274728,
      "grad_norm": 0.08515532314777374,
      "learning_rate": 4.665167857320436e-06,
      "loss": 0.0013,
      "step": 104360
    },
    {
      "epoch": 1.7077640513785486,
      "grad_norm": 0.05431477725505829,
      "learning_rate": 4.664218060201397e-06,
      "loss": 0.0011,
      "step": 104370
    },
    {
      "epoch": 1.7079276773296246,
      "grad_norm": 0.10838386416435242,
      "learning_rate": 4.6632682752536475e-06,
      "loss": 0.0022,
      "step": 104380
    },
    {
      "epoch": 1.7080913032807004,
      "grad_norm": 0.26230278611183167,
      "learning_rate": 4.662318502511614e-06,
      "loss": 0.0014,
      "step": 104390
    },
    {
      "epoch": 1.7082549292317761,
      "grad_norm": 0.19241011142730713,
      "learning_rate": 4.661368742009722e-06,
      "loss": 0.002,
      "step": 104400
    },
    {
      "epoch": 1.7084185551828521,
      "grad_norm": 0.06803303211927414,
      "learning_rate": 4.660418993782402e-06,
      "loss": 0.0013,
      "step": 104410
    },
    {
      "epoch": 1.7085821811339277,
      "grad_norm": 0.04411003366112709,
      "learning_rate": 4.659469257864074e-06,
      "loss": 0.0018,
      "step": 104420
    },
    {
      "epoch": 1.7087458070850037,
      "grad_norm": 0.034814633429050446,
      "learning_rate": 4.658519534289168e-06,
      "loss": 0.0014,
      "step": 104430
    },
    {
      "epoch": 1.7089094330360797,
      "grad_norm": 0.06919162720441818,
      "learning_rate": 4.6575698230921085e-06,
      "loss": 0.0013,
      "step": 104440
    },
    {
      "epoch": 1.7090730589871552,
      "grad_norm": 0.11741115152835846,
      "learning_rate": 4.65662012430732e-06,
      "loss": 0.0018,
      "step": 104450
    },
    {
      "epoch": 1.7092366849382312,
      "grad_norm": 0.05263363942503929,
      "learning_rate": 4.6556704379692255e-06,
      "loss": 0.0008,
      "step": 104460
    },
    {
      "epoch": 1.709400310889307,
      "grad_norm": 0.051880110055208206,
      "learning_rate": 4.654720764112251e-06,
      "loss": 0.0014,
      "step": 104470
    },
    {
      "epoch": 1.7095639368403828,
      "grad_norm": 0.07953453063964844,
      "learning_rate": 4.653771102770819e-06,
      "loss": 0.0021,
      "step": 104480
    },
    {
      "epoch": 1.7097275627914588,
      "grad_norm": 0.19696493446826935,
      "learning_rate": 4.6528214539793526e-06,
      "loss": 0.0022,
      "step": 104490
    },
    {
      "epoch": 1.7098911887425345,
      "grad_norm": 0.028127476572990417,
      "learning_rate": 4.651871817772273e-06,
      "loss": 0.0011,
      "step": 104500
    },
    {
      "epoch": 1.7100548146936103,
      "grad_norm": 0.07425235956907272,
      "learning_rate": 4.650922194184006e-06,
      "loss": 0.0011,
      "step": 104510
    },
    {
      "epoch": 1.7102184406446863,
      "grad_norm": 0.11308459937572479,
      "learning_rate": 4.649972583248968e-06,
      "loss": 0.0016,
      "step": 104520
    },
    {
      "epoch": 1.710382066595762,
      "grad_norm": 0.12562865018844604,
      "learning_rate": 4.649022985001584e-06,
      "loss": 0.0018,
      "step": 104530
    },
    {
      "epoch": 1.7105456925468379,
      "grad_norm": 0.068203866481781,
      "learning_rate": 4.6480733994762716e-06,
      "loss": 0.001,
      "step": 104540
    },
    {
      "epoch": 1.7107093184979139,
      "grad_norm": 0.18430009484291077,
      "learning_rate": 4.647123826707453e-06,
      "loss": 0.0012,
      "step": 104550
    },
    {
      "epoch": 1.7108729444489896,
      "grad_norm": 0.05046533793210983,
      "learning_rate": 4.646174266729549e-06,
      "loss": 0.0014,
      "step": 104560
    },
    {
      "epoch": 1.7110365704000654,
      "grad_norm": 0.19980613887310028,
      "learning_rate": 4.6452247195769755e-06,
      "loss": 0.0018,
      "step": 104570
    },
    {
      "epoch": 1.7112001963511414,
      "grad_norm": 0.10487794131040573,
      "learning_rate": 4.644275185284153e-06,
      "loss": 0.0013,
      "step": 104580
    },
    {
      "epoch": 1.7113638223022172,
      "grad_norm": 0.06061362847685814,
      "learning_rate": 4.6433256638855015e-06,
      "loss": 0.0012,
      "step": 104590
    },
    {
      "epoch": 1.711527448253293,
      "grad_norm": 0.18957191705703735,
      "learning_rate": 4.6423761554154364e-06,
      "loss": 0.0023,
      "step": 104600
    },
    {
      "epoch": 1.711691074204369,
      "grad_norm": 0.09567522257566452,
      "learning_rate": 4.641426659908379e-06,
      "loss": 0.0013,
      "step": 104610
    },
    {
      "epoch": 1.7118547001554445,
      "grad_norm": 0.07723551988601685,
      "learning_rate": 4.640477177398742e-06,
      "loss": 0.0014,
      "step": 104620
    },
    {
      "epoch": 1.7120183261065205,
      "grad_norm": 0.09386233985424042,
      "learning_rate": 4.6395277079209425e-06,
      "loss": 0.0022,
      "step": 104630
    },
    {
      "epoch": 1.7121819520575965,
      "grad_norm": 0.06522838026285172,
      "learning_rate": 4.638578251509395e-06,
      "loss": 0.0011,
      "step": 104640
    },
    {
      "epoch": 1.712345578008672,
      "grad_norm": 0.028610533103346825,
      "learning_rate": 4.63762880819852e-06,
      "loss": 0.0027,
      "step": 104650
    },
    {
      "epoch": 1.712509203959748,
      "grad_norm": 0.17405836284160614,
      "learning_rate": 4.63667937802273e-06,
      "loss": 0.0017,
      "step": 104660
    },
    {
      "epoch": 1.7126728299108238,
      "grad_norm": 0.06394306570291519,
      "learning_rate": 4.635729961016438e-06,
      "loss": 0.0019,
      "step": 104670
    },
    {
      "epoch": 1.7128364558618996,
      "grad_norm": 0.043279122561216354,
      "learning_rate": 4.634780557214059e-06,
      "loss": 0.0017,
      "step": 104680
    },
    {
      "epoch": 1.7130000818129756,
      "grad_norm": 0.15762358903884888,
      "learning_rate": 4.633831166650008e-06,
      "loss": 0.0019,
      "step": 104690
    },
    {
      "epoch": 1.7131637077640514,
      "grad_norm": 0.09532841295003891,
      "learning_rate": 4.632881789358697e-06,
      "loss": 0.0013,
      "step": 104700
    },
    {
      "epoch": 1.7133273337151271,
      "grad_norm": 0.24289748072624207,
      "learning_rate": 4.63193242537454e-06,
      "loss": 0.0017,
      "step": 104710
    },
    {
      "epoch": 1.7134909596662031,
      "grad_norm": 0.020615093410015106,
      "learning_rate": 4.630983074731947e-06,
      "loss": 0.002,
      "step": 104720
    },
    {
      "epoch": 1.713654585617279,
      "grad_norm": 0.035493504256010056,
      "learning_rate": 4.630033737465331e-06,
      "loss": 0.0008,
      "step": 104730
    },
    {
      "epoch": 1.7138182115683547,
      "grad_norm": 0.15942108631134033,
      "learning_rate": 4.629084413609102e-06,
      "loss": 0.0014,
      "step": 104740
    },
    {
      "epoch": 1.7139818375194307,
      "grad_norm": 0.031997308135032654,
      "learning_rate": 4.628135103197671e-06,
      "loss": 0.0017,
      "step": 104750
    },
    {
      "epoch": 1.7141454634705064,
      "grad_norm": 0.11948850750923157,
      "learning_rate": 4.62718580626545e-06,
      "loss": 0.0012,
      "step": 104760
    },
    {
      "epoch": 1.7143090894215822,
      "grad_norm": 0.2451469451189041,
      "learning_rate": 4.626236522846847e-06,
      "loss": 0.0019,
      "step": 104770
    },
    {
      "epoch": 1.7144727153726582,
      "grad_norm": 0.04973438009619713,
      "learning_rate": 4.625287252976272e-06,
      "loss": 0.0022,
      "step": 104780
    },
    {
      "epoch": 1.714636341323734,
      "grad_norm": 0.1627313792705536,
      "learning_rate": 4.624337996688133e-06,
      "loss": 0.0012,
      "step": 104790
    },
    {
      "epoch": 1.7147999672748098,
      "grad_norm": 0.06593550741672516,
      "learning_rate": 4.6233887540168375e-06,
      "loss": 0.0011,
      "step": 104800
    },
    {
      "epoch": 1.7149635932258858,
      "grad_norm": 0.09836692363023758,
      "learning_rate": 4.622439524996798e-06,
      "loss": 0.002,
      "step": 104810
    },
    {
      "epoch": 1.7151272191769613,
      "grad_norm": 0.11044531315565109,
      "learning_rate": 4.621490309662417e-06,
      "loss": 0.0016,
      "step": 104820
    },
    {
      "epoch": 1.7152908451280373,
      "grad_norm": 0.11051111668348312,
      "learning_rate": 4.620541108048102e-06,
      "loss": 0.001,
      "step": 104830
    },
    {
      "epoch": 1.7154544710791133,
      "grad_norm": 0.013667666353285313,
      "learning_rate": 4.619591920188259e-06,
      "loss": 0.0034,
      "step": 104840
    },
    {
      "epoch": 1.7156180970301889,
      "grad_norm": 0.13319715857505798,
      "learning_rate": 4.618642746117295e-06,
      "loss": 0.0017,
      "step": 104850
    },
    {
      "epoch": 1.7157817229812649,
      "grad_norm": 0.06539462506771088,
      "learning_rate": 4.617693585869615e-06,
      "loss": 0.0014,
      "step": 104860
    },
    {
      "epoch": 1.7159453489323406,
      "grad_norm": 0.08072780072689056,
      "learning_rate": 4.616744439479623e-06,
      "loss": 0.0015,
      "step": 104870
    },
    {
      "epoch": 1.7161089748834164,
      "grad_norm": 0.13563817739486694,
      "learning_rate": 4.615795306981725e-06,
      "loss": 0.0013,
      "step": 104880
    },
    {
      "epoch": 1.7162726008344924,
      "grad_norm": 0.01548470463603735,
      "learning_rate": 4.614846188410324e-06,
      "loss": 0.0018,
      "step": 104890
    },
    {
      "epoch": 1.7164362267855682,
      "grad_norm": 0.08023739606142044,
      "learning_rate": 4.613897083799822e-06,
      "loss": 0.0022,
      "step": 104900
    },
    {
      "epoch": 1.716599852736644,
      "grad_norm": 0.07746725529432297,
      "learning_rate": 4.6129479931846255e-06,
      "loss": 0.0014,
      "step": 104910
    },
    {
      "epoch": 1.71676347868772,
      "grad_norm": 0.008192102424800396,
      "learning_rate": 4.61199891659913e-06,
      "loss": 0.0015,
      "step": 104920
    },
    {
      "epoch": 1.7169271046387957,
      "grad_norm": 0.03445400297641754,
      "learning_rate": 4.611049854077743e-06,
      "loss": 0.0009,
      "step": 104930
    },
    {
      "epoch": 1.7170907305898715,
      "grad_norm": 0.2858472764492035,
      "learning_rate": 4.6101008056548644e-06,
      "loss": 0.0021,
      "step": 104940
    },
    {
      "epoch": 1.7172543565409475,
      "grad_norm": 0.045578714460134506,
      "learning_rate": 4.609151771364894e-06,
      "loss": 0.0004,
      "step": 104950
    },
    {
      "epoch": 1.7174179824920233,
      "grad_norm": 0.2471114993095398,
      "learning_rate": 4.6082027512422325e-06,
      "loss": 0.0018,
      "step": 104960
    },
    {
      "epoch": 1.717581608443099,
      "grad_norm": 0.054288070648908615,
      "learning_rate": 4.607253745321279e-06,
      "loss": 0.0012,
      "step": 104970
    },
    {
      "epoch": 1.717745234394175,
      "grad_norm": 0.10607331991195679,
      "learning_rate": 4.606304753636434e-06,
      "loss": 0.0011,
      "step": 104980
    },
    {
      "epoch": 1.7179088603452508,
      "grad_norm": 0.04377147927880287,
      "learning_rate": 4.605355776222095e-06,
      "loss": 0.001,
      "step": 104990
    },
    {
      "epoch": 1.7180724862963266,
      "grad_norm": 0.07112236320972443,
      "learning_rate": 4.604406813112661e-06,
      "loss": 0.0024,
      "step": 105000
    },
    {
      "epoch": 1.7182361122474026,
      "grad_norm": 0.05970732867717743,
      "learning_rate": 4.60345786434253e-06,
      "loss": 0.0019,
      "step": 105010
    },
    {
      "epoch": 1.7183997381984781,
      "grad_norm": 0.17530439794063568,
      "learning_rate": 4.602508929946098e-06,
      "loss": 0.0031,
      "step": 105020
    },
    {
      "epoch": 1.7185633641495541,
      "grad_norm": 0.11928589642047882,
      "learning_rate": 4.6015600099577615e-06,
      "loss": 0.0012,
      "step": 105030
    },
    {
      "epoch": 1.7187269901006301,
      "grad_norm": 0.06392933428287506,
      "learning_rate": 4.600611104411917e-06,
      "loss": 0.0011,
      "step": 105040
    },
    {
      "epoch": 1.7188906160517057,
      "grad_norm": 0.2600875794887543,
      "learning_rate": 4.5996622133429605e-06,
      "loss": 0.0021,
      "step": 105050
    },
    {
      "epoch": 1.7190542420027817,
      "grad_norm": 0.0062109315767884254,
      "learning_rate": 4.598713336785286e-06,
      "loss": 0.0019,
      "step": 105060
    },
    {
      "epoch": 1.7192178679538574,
      "grad_norm": 0.10749581456184387,
      "learning_rate": 4.5977644747732885e-06,
      "loss": 0.0022,
      "step": 105070
    },
    {
      "epoch": 1.7193814939049332,
      "grad_norm": 0.09237188845872879,
      "learning_rate": 4.596815627341361e-06,
      "loss": 0.0024,
      "step": 105080
    },
    {
      "epoch": 1.7195451198560092,
      "grad_norm": 0.1016964465379715,
      "learning_rate": 4.595866794523899e-06,
      "loss": 0.0016,
      "step": 105090
    },
    {
      "epoch": 1.719708745807085,
      "grad_norm": 0.016306094825267792,
      "learning_rate": 4.5949179763552946e-06,
      "loss": 0.0037,
      "step": 105100
    },
    {
      "epoch": 1.7198723717581608,
      "grad_norm": 0.08354363590478897,
      "learning_rate": 4.59396917286994e-06,
      "loss": 0.0025,
      "step": 105110
    },
    {
      "epoch": 1.7200359977092368,
      "grad_norm": 0.04717186838388443,
      "learning_rate": 4.593020384102229e-06,
      "loss": 0.0012,
      "step": 105120
    },
    {
      "epoch": 1.7201996236603125,
      "grad_norm": 0.004931902978569269,
      "learning_rate": 4.59207161008655e-06,
      "loss": 0.0019,
      "step": 105130
    },
    {
      "epoch": 1.7203632496113883,
      "grad_norm": 0.43936723470687866,
      "learning_rate": 4.591122850857292e-06,
      "loss": 0.0025,
      "step": 105140
    },
    {
      "epoch": 1.7205268755624643,
      "grad_norm": 0.42821720242500305,
      "learning_rate": 4.590174106448849e-06,
      "loss": 0.0034,
      "step": 105150
    },
    {
      "epoch": 1.72069050151354,
      "grad_norm": 0.1212688684463501,
      "learning_rate": 4.589225376895611e-06,
      "loss": 0.0015,
      "step": 105160
    },
    {
      "epoch": 1.7208541274646159,
      "grad_norm": 0.00839544553309679,
      "learning_rate": 4.588276662231965e-06,
      "loss": 0.0015,
      "step": 105170
    },
    {
      "epoch": 1.7210177534156919,
      "grad_norm": 0.06713610887527466,
      "learning_rate": 4.5873279624923e-06,
      "loss": 0.0016,
      "step": 105180
    },
    {
      "epoch": 1.7211813793667674,
      "grad_norm": 0.1132173091173172,
      "learning_rate": 4.586379277711005e-06,
      "loss": 0.0009,
      "step": 105190
    },
    {
      "epoch": 1.7213450053178434,
      "grad_norm": 0.08310399204492569,
      "learning_rate": 4.585430607922467e-06,
      "loss": 0.0019,
      "step": 105200
    },
    {
      "epoch": 1.7215086312689194,
      "grad_norm": 0.04769390448927879,
      "learning_rate": 4.5844819531610735e-06,
      "loss": 0.0021,
      "step": 105210
    },
    {
      "epoch": 1.721672257219995,
      "grad_norm": 0.029055854305624962,
      "learning_rate": 4.5835333134612106e-06,
      "loss": 0.0036,
      "step": 105220
    },
    {
      "epoch": 1.721835883171071,
      "grad_norm": 0.20655041933059692,
      "learning_rate": 4.582584688857265e-06,
      "loss": 0.0018,
      "step": 105230
    },
    {
      "epoch": 1.7219995091221467,
      "grad_norm": 0.12657411396503448,
      "learning_rate": 4.581636079383619e-06,
      "loss": 0.002,
      "step": 105240
    },
    {
      "epoch": 1.7221631350732225,
      "grad_norm": 0.06903214752674103,
      "learning_rate": 4.580687485074661e-06,
      "loss": 0.0011,
      "step": 105250
    },
    {
      "epoch": 1.7223267610242985,
      "grad_norm": 0.0782797634601593,
      "learning_rate": 4.579738905964773e-06,
      "loss": 0.0016,
      "step": 105260
    },
    {
      "epoch": 1.7224903869753743,
      "grad_norm": 0.07491712272167206,
      "learning_rate": 4.5787903420883405e-06,
      "loss": 0.0019,
      "step": 105270
    },
    {
      "epoch": 1.72265401292645,
      "grad_norm": 0.08151070773601532,
      "learning_rate": 4.577841793479745e-06,
      "loss": 0.0009,
      "step": 105280
    },
    {
      "epoch": 1.722817638877526,
      "grad_norm": 0.031249873340129852,
      "learning_rate": 4.57689326017337e-06,
      "loss": 0.0012,
      "step": 105290
    },
    {
      "epoch": 1.7229812648286018,
      "grad_norm": 0.12277843058109283,
      "learning_rate": 4.575944742203597e-06,
      "loss": 0.0011,
      "step": 105300
    },
    {
      "epoch": 1.7231448907796776,
      "grad_norm": 0.06664171814918518,
      "learning_rate": 4.574996239604808e-06,
      "loss": 0.0014,
      "step": 105310
    },
    {
      "epoch": 1.7233085167307536,
      "grad_norm": 0.12797382473945618,
      "learning_rate": 4.574047752411386e-06,
      "loss": 0.0021,
      "step": 105320
    },
    {
      "epoch": 1.7234721426818294,
      "grad_norm": 0.008203708566725254,
      "learning_rate": 4.573099280657708e-06,
      "loss": 0.0021,
      "step": 105330
    },
    {
      "epoch": 1.7236357686329051,
      "grad_norm": 0.047096915543079376,
      "learning_rate": 4.572150824378154e-06,
      "loss": 0.0034,
      "step": 105340
    },
    {
      "epoch": 1.7237993945839811,
      "grad_norm": 0.04287709295749664,
      "learning_rate": 4.571202383607105e-06,
      "loss": 0.0025,
      "step": 105350
    },
    {
      "epoch": 1.723963020535057,
      "grad_norm": 0.1158096045255661,
      "learning_rate": 4.570253958378937e-06,
      "loss": 0.0013,
      "step": 105360
    },
    {
      "epoch": 1.7241266464861327,
      "grad_norm": 0.08027780055999756,
      "learning_rate": 4.569305548728032e-06,
      "loss": 0.0011,
      "step": 105370
    },
    {
      "epoch": 1.7242902724372087,
      "grad_norm": 0.15397228300571442,
      "learning_rate": 4.568357154688766e-06,
      "loss": 0.0024,
      "step": 105380
    },
    {
      "epoch": 1.7244538983882842,
      "grad_norm": 0.04861374944448471,
      "learning_rate": 4.5674087762955156e-06,
      "loss": 0.0013,
      "step": 105390
    },
    {
      "epoch": 1.7246175243393602,
      "grad_norm": 0.03549257293343544,
      "learning_rate": 4.5664604135826576e-06,
      "loss": 0.0015,
      "step": 105400
    },
    {
      "epoch": 1.7247811502904362,
      "grad_norm": 0.0403401181101799,
      "learning_rate": 4.565512066584567e-06,
      "loss": 0.0019,
      "step": 105410
    },
    {
      "epoch": 1.7249447762415118,
      "grad_norm": 0.11930004507303238,
      "learning_rate": 4.564563735335621e-06,
      "loss": 0.0014,
      "step": 105420
    },
    {
      "epoch": 1.7251084021925878,
      "grad_norm": 0.037674449384212494,
      "learning_rate": 4.5636154198701925e-06,
      "loss": 0.0007,
      "step": 105430
    },
    {
      "epoch": 1.7252720281436635,
      "grad_norm": 0.04420173168182373,
      "learning_rate": 4.562667120222655e-06,
      "loss": 0.0016,
      "step": 105440
    },
    {
      "epoch": 1.7254356540947393,
      "grad_norm": 0.1562848538160324,
      "learning_rate": 4.561718836427384e-06,
      "loss": 0.0015,
      "step": 105450
    },
    {
      "epoch": 1.7255992800458153,
      "grad_norm": 0.10649880021810532,
      "learning_rate": 4.560770568518751e-06,
      "loss": 0.0008,
      "step": 105460
    },
    {
      "epoch": 1.725762905996891,
      "grad_norm": 0.043602969497442245,
      "learning_rate": 4.55982231653113e-06,
      "loss": 0.0014,
      "step": 105470
    },
    {
      "epoch": 1.7259265319479669,
      "grad_norm": 0.11116377264261246,
      "learning_rate": 4.558874080498893e-06,
      "loss": 0.0023,
      "step": 105480
    },
    {
      "epoch": 1.7260901578990429,
      "grad_norm": 0.2922625243663788,
      "learning_rate": 4.5579258604564095e-06,
      "loss": 0.0026,
      "step": 105490
    },
    {
      "epoch": 1.7262537838501186,
      "grad_norm": 0.07799088209867477,
      "learning_rate": 4.556977656438051e-06,
      "loss": 0.001,
      "step": 105500
    },
    {
      "epoch": 1.7264174098011944,
      "grad_norm": 0.0895681381225586,
      "learning_rate": 4.556029468478187e-06,
      "loss": 0.0023,
      "step": 105510
    },
    {
      "epoch": 1.7265810357522704,
      "grad_norm": 0.14441503584384918,
      "learning_rate": 4.55508129661119e-06,
      "loss": 0.0012,
      "step": 105520
    },
    {
      "epoch": 1.7267446617033462,
      "grad_norm": 0.06283687800168991,
      "learning_rate": 4.5541331408714255e-06,
      "loss": 0.0019,
      "step": 105530
    },
    {
      "epoch": 1.726908287654422,
      "grad_norm": 0.06532728672027588,
      "learning_rate": 4.553185001293262e-06,
      "loss": 0.0011,
      "step": 105540
    },
    {
      "epoch": 1.727071913605498,
      "grad_norm": 0.1469145119190216,
      "learning_rate": 4.552236877911069e-06,
      "loss": 0.0024,
      "step": 105550
    },
    {
      "epoch": 1.7272355395565737,
      "grad_norm": 0.08351492136716843,
      "learning_rate": 4.551288770759212e-06,
      "loss": 0.0008,
      "step": 105560
    },
    {
      "epoch": 1.7273991655076495,
      "grad_norm": 0.18137399852275848,
      "learning_rate": 4.550340679872059e-06,
      "loss": 0.0019,
      "step": 105570
    },
    {
      "epoch": 1.7275627914587255,
      "grad_norm": 0.22894568741321564,
      "learning_rate": 4.549392605283976e-06,
      "loss": 0.0015,
      "step": 105580
    },
    {
      "epoch": 1.727726417409801,
      "grad_norm": 0.05140344426035881,
      "learning_rate": 4.548444547029328e-06,
      "loss": 0.0012,
      "step": 105590
    },
    {
      "epoch": 1.727890043360877,
      "grad_norm": 0.3428930640220642,
      "learning_rate": 4.54749650514248e-06,
      "loss": 0.0018,
      "step": 105600
    },
    {
      "epoch": 1.728053669311953,
      "grad_norm": 0.12339276075363159,
      "learning_rate": 4.546548479657796e-06,
      "loss": 0.0021,
      "step": 105610
    },
    {
      "epoch": 1.7282172952630286,
      "grad_norm": 0.2664509117603302,
      "learning_rate": 4.545600470609643e-06,
      "loss": 0.0019,
      "step": 105620
    },
    {
      "epoch": 1.7283809212141046,
      "grad_norm": 0.0845964327454567,
      "learning_rate": 4.544652478032377e-06,
      "loss": 0.0015,
      "step": 105630
    },
    {
      "epoch": 1.7285445471651804,
      "grad_norm": 0.10369815677404404,
      "learning_rate": 4.543704501960365e-06,
      "loss": 0.0016,
      "step": 105640
    },
    {
      "epoch": 1.7287081731162561,
      "grad_norm": 0.07420734316110611,
      "learning_rate": 4.542756542427968e-06,
      "loss": 0.0011,
      "step": 105650
    },
    {
      "epoch": 1.7288717990673321,
      "grad_norm": 0.2443924844264984,
      "learning_rate": 4.5418085994695475e-06,
      "loss": 0.0013,
      "step": 105660
    },
    {
      "epoch": 1.729035425018408,
      "grad_norm": 0.0880240648984909,
      "learning_rate": 4.540860673119464e-06,
      "loss": 0.0015,
      "step": 105670
    },
    {
      "epoch": 1.7291990509694837,
      "grad_norm": 0.4116462767124176,
      "learning_rate": 4.5399127634120775e-06,
      "loss": 0.0022,
      "step": 105680
    },
    {
      "epoch": 1.7293626769205597,
      "grad_norm": 0.19862475991249084,
      "learning_rate": 4.538964870381748e-06,
      "loss": 0.0022,
      "step": 105690
    },
    {
      "epoch": 1.7295263028716354,
      "grad_norm": 0.1152542307972908,
      "learning_rate": 4.538016994062833e-06,
      "loss": 0.0012,
      "step": 105700
    },
    {
      "epoch": 1.7296899288227112,
      "grad_norm": 0.1057204082608223,
      "learning_rate": 4.537069134489693e-06,
      "loss": 0.0019,
      "step": 105710
    },
    {
      "epoch": 1.7298535547737872,
      "grad_norm": 0.09458913654088974,
      "learning_rate": 4.536121291696684e-06,
      "loss": 0.0017,
      "step": 105720
    },
    {
      "epoch": 1.730017180724863,
      "grad_norm": 0.07179639488458633,
      "learning_rate": 4.535173465718163e-06,
      "loss": 0.0017,
      "step": 105730
    },
    {
      "epoch": 1.7301808066759388,
      "grad_norm": 0.31437626481056213,
      "learning_rate": 4.534225656588486e-06,
      "loss": 0.0009,
      "step": 105740
    },
    {
      "epoch": 1.7303444326270148,
      "grad_norm": 0.0046170419082045555,
      "learning_rate": 4.53327786434201e-06,
      "loss": 0.0011,
      "step": 105750
    },
    {
      "epoch": 1.7305080585780905,
      "grad_norm": 0.019300058484077454,
      "learning_rate": 4.53233008901309e-06,
      "loss": 0.0019,
      "step": 105760
    },
    {
      "epoch": 1.7306716845291663,
      "grad_norm": 0.020240964367985725,
      "learning_rate": 4.53138233063608e-06,
      "loss": 0.0011,
      "step": 105770
    },
    {
      "epoch": 1.7308353104802423,
      "grad_norm": 0.07901404052972794,
      "learning_rate": 4.530434589245334e-06,
      "loss": 0.0015,
      "step": 105780
    },
    {
      "epoch": 1.7309989364313179,
      "grad_norm": 0.026835916563868523,
      "learning_rate": 4.529486864875205e-06,
      "loss": 0.0009,
      "step": 105790
    },
    {
      "epoch": 1.7311625623823939,
      "grad_norm": 0.031669583171606064,
      "learning_rate": 4.528539157560046e-06,
      "loss": 0.0009,
      "step": 105800
    },
    {
      "epoch": 1.7313261883334699,
      "grad_norm": 0.038425758481025696,
      "learning_rate": 4.5275914673342105e-06,
      "loss": 0.0013,
      "step": 105810
    },
    {
      "epoch": 1.7314898142845454,
      "grad_norm": 0.07629675418138504,
      "learning_rate": 4.526643794232051e-06,
      "loss": 0.0012,
      "step": 105820
    },
    {
      "epoch": 1.7316534402356214,
      "grad_norm": 0.07075434923171997,
      "learning_rate": 4.525696138287914e-06,
      "loss": 0.0021,
      "step": 105830
    },
    {
      "epoch": 1.7318170661866972,
      "grad_norm": 0.11859335005283356,
      "learning_rate": 4.524748499536152e-06,
      "loss": 0.0013,
      "step": 105840
    },
    {
      "epoch": 1.731980692137773,
      "grad_norm": 0.12713493406772614,
      "learning_rate": 4.523800878011114e-06,
      "loss": 0.0018,
      "step": 105850
    },
    {
      "epoch": 1.732144318088849,
      "grad_norm": 0.1701129525899887,
      "learning_rate": 4.5228532737471505e-06,
      "loss": 0.0014,
      "step": 105860
    },
    {
      "epoch": 1.7323079440399247,
      "grad_norm": 0.17722174525260925,
      "learning_rate": 4.521905686778609e-06,
      "loss": 0.0014,
      "step": 105870
    },
    {
      "epoch": 1.7324715699910005,
      "grad_norm": 0.0571046806871891,
      "learning_rate": 4.520958117139837e-06,
      "loss": 0.0015,
      "step": 105880
    },
    {
      "epoch": 1.7326351959420765,
      "grad_norm": 0.07202465087175369,
      "learning_rate": 4.520010564865182e-06,
      "loss": 0.0015,
      "step": 105890
    },
    {
      "epoch": 1.7327988218931523,
      "grad_norm": 0.03762871026992798,
      "learning_rate": 4.519063029988991e-06,
      "loss": 0.0018,
      "step": 105900
    },
    {
      "epoch": 1.732962447844228,
      "grad_norm": 0.0380859375,
      "learning_rate": 4.518115512545608e-06,
      "loss": 0.0014,
      "step": 105910
    },
    {
      "epoch": 1.733126073795304,
      "grad_norm": 0.270017147064209,
      "learning_rate": 4.517168012569382e-06,
      "loss": 0.0011,
      "step": 105920
    },
    {
      "epoch": 1.7332896997463798,
      "grad_norm": 0.08873353898525238,
      "learning_rate": 4.516220530094652e-06,
      "loss": 0.0013,
      "step": 105930
    },
    {
      "epoch": 1.7334533256974556,
      "grad_norm": 0.13098040223121643,
      "learning_rate": 4.5152730651557675e-06,
      "loss": 0.0016,
      "step": 105940
    },
    {
      "epoch": 1.7336169516485316,
      "grad_norm": 0.25841355323791504,
      "learning_rate": 4.514325617787068e-06,
      "loss": 0.0011,
      "step": 105950
    },
    {
      "epoch": 1.7337805775996074,
      "grad_norm": 0.06951875239610672,
      "learning_rate": 4.513378188022898e-06,
      "loss": 0.0018,
      "step": 105960
    },
    {
      "epoch": 1.7339442035506831,
      "grad_norm": 0.0480615571141243,
      "learning_rate": 4.512430775897599e-06,
      "loss": 0.0017,
      "step": 105970
    },
    {
      "epoch": 1.7341078295017591,
      "grad_norm": 0.08204066753387451,
      "learning_rate": 4.511483381445512e-06,
      "loss": 0.001,
      "step": 105980
    },
    {
      "epoch": 1.7342714554528347,
      "grad_norm": 0.06344208121299744,
      "learning_rate": 4.510536004700979e-06,
      "loss": 0.001,
      "step": 105990
    },
    {
      "epoch": 1.7344350814039107,
      "grad_norm": 0.033863600343465805,
      "learning_rate": 4.50958864569834e-06,
      "loss": 0.001,
      "step": 106000
    },
    {
      "epoch": 1.7345987073549864,
      "grad_norm": 0.0188735518604517,
      "learning_rate": 4.5086413044719314e-06,
      "loss": 0.0014,
      "step": 106010
    },
    {
      "epoch": 1.7347623333060622,
      "grad_norm": 0.0985097885131836,
      "learning_rate": 4.507693981056098e-06,
      "loss": 0.002,
      "step": 106020
    },
    {
      "epoch": 1.7349259592571382,
      "grad_norm": 0.005792176350951195,
      "learning_rate": 4.506746675485173e-06,
      "loss": 0.0016,
      "step": 106030
    },
    {
      "epoch": 1.735089585208214,
      "grad_norm": 0.2639869749546051,
      "learning_rate": 4.505799387793496e-06,
      "loss": 0.0011,
      "step": 106040
    },
    {
      "epoch": 1.7352532111592898,
      "grad_norm": 0.09465857595205307,
      "learning_rate": 4.5048521180154025e-06,
      "loss": 0.0022,
      "step": 106050
    },
    {
      "epoch": 1.7354168371103658,
      "grad_norm": 0.027954036369919777,
      "learning_rate": 4.50390486618523e-06,
      "loss": 0.0011,
      "step": 106060
    },
    {
      "epoch": 1.7355804630614415,
      "grad_norm": 0.20050562918186188,
      "learning_rate": 4.502957632337313e-06,
      "loss": 0.0015,
      "step": 106070
    },
    {
      "epoch": 1.7357440890125173,
      "grad_norm": 0.2531886398792267,
      "learning_rate": 4.502010416505988e-06,
      "loss": 0.0013,
      "step": 106080
    },
    {
      "epoch": 1.7359077149635933,
      "grad_norm": 0.0523017980158329,
      "learning_rate": 4.501063218725588e-06,
      "loss": 0.0013,
      "step": 106090
    },
    {
      "epoch": 1.736071340914669,
      "grad_norm": 0.02355448715388775,
      "learning_rate": 4.500116039030448e-06,
      "loss": 0.0008,
      "step": 106100
    },
    {
      "epoch": 1.7362349668657449,
      "grad_norm": 0.03599315136671066,
      "learning_rate": 4.4991688774549e-06,
      "loss": 0.0009,
      "step": 106110
    },
    {
      "epoch": 1.7363985928168209,
      "grad_norm": 0.056937552988529205,
      "learning_rate": 4.498221734033278e-06,
      "loss": 0.0011,
      "step": 106120
    },
    {
      "epoch": 1.7365622187678966,
      "grad_norm": 0.20131470263004303,
      "learning_rate": 4.4972746087999095e-06,
      "loss": 0.003,
      "step": 106130
    },
    {
      "epoch": 1.7367258447189724,
      "grad_norm": 0.1308128982782364,
      "learning_rate": 4.496327501789129e-06,
      "loss": 0.0011,
      "step": 106140
    },
    {
      "epoch": 1.7368894706700484,
      "grad_norm": 0.11488918960094452,
      "learning_rate": 4.4953804130352656e-06,
      "loss": 0.0012,
      "step": 106150
    },
    {
      "epoch": 1.737053096621124,
      "grad_norm": 0.06370162218809128,
      "learning_rate": 4.49443334257265e-06,
      "loss": 0.002,
      "step": 106160
    },
    {
      "epoch": 1.7372167225722,
      "grad_norm": 0.010542163625359535,
      "learning_rate": 4.493486290435611e-06,
      "loss": 0.0016,
      "step": 106170
    },
    {
      "epoch": 1.737380348523276,
      "grad_norm": 0.1069006696343422,
      "learning_rate": 4.492539256658476e-06,
      "loss": 0.001,
      "step": 106180
    },
    {
      "epoch": 1.7375439744743515,
      "grad_norm": 0.3381604850292206,
      "learning_rate": 4.491592241275573e-06,
      "loss": 0.0018,
      "step": 106190
    },
    {
      "epoch": 1.7377076004254275,
      "grad_norm": 0.036023881286382675,
      "learning_rate": 4.49064524432123e-06,
      "loss": 0.0014,
      "step": 106200
    },
    {
      "epoch": 1.7378712263765033,
      "grad_norm": 0.06754820048809052,
      "learning_rate": 4.4896982658297725e-06,
      "loss": 0.0008,
      "step": 106210
    },
    {
      "epoch": 1.738034852327579,
      "grad_norm": 0.20016489923000336,
      "learning_rate": 4.488751305835527e-06,
      "loss": 0.0027,
      "step": 106220
    },
    {
      "epoch": 1.738198478278655,
      "grad_norm": 0.05603393167257309,
      "learning_rate": 4.487804364372818e-06,
      "loss": 0.0009,
      "step": 106230
    },
    {
      "epoch": 1.7383621042297308,
      "grad_norm": 0.2260708063840866,
      "learning_rate": 4.4868574414759684e-06,
      "loss": 0.0036,
      "step": 106240
    },
    {
      "epoch": 1.7385257301808066,
      "grad_norm": 0.050879478454589844,
      "learning_rate": 4.4859105371793034e-06,
      "loss": 0.0012,
      "step": 106250
    },
    {
      "epoch": 1.7386893561318826,
      "grad_norm": 0.14517945051193237,
      "learning_rate": 4.4849636515171464e-06,
      "loss": 0.0018,
      "step": 106260
    },
    {
      "epoch": 1.7388529820829584,
      "grad_norm": 0.14041414856910706,
      "learning_rate": 4.484016784523819e-06,
      "loss": 0.0014,
      "step": 106270
    },
    {
      "epoch": 1.7390166080340341,
      "grad_norm": 0.1947072595357895,
      "learning_rate": 4.4830699362336436e-06,
      "loss": 0.0013,
      "step": 106280
    },
    {
      "epoch": 1.7391802339851101,
      "grad_norm": 0.25436705350875854,
      "learning_rate": 4.482123106680939e-06,
      "loss": 0.0014,
      "step": 106290
    },
    {
      "epoch": 1.739343859936186,
      "grad_norm": 0.231724813580513,
      "learning_rate": 4.481176295900028e-06,
      "loss": 0.0025,
      "step": 106300
    },
    {
      "epoch": 1.7395074858872617,
      "grad_norm": 0.053413812071084976,
      "learning_rate": 4.48022950392523e-06,
      "loss": 0.0009,
      "step": 106310
    },
    {
      "epoch": 1.7396711118383377,
      "grad_norm": 0.04543972760438919,
      "learning_rate": 4.479282730790864e-06,
      "loss": 0.0011,
      "step": 106320
    },
    {
      "epoch": 1.7398347377894134,
      "grad_norm": 0.05133870616555214,
      "learning_rate": 4.478335976531246e-06,
      "loss": 0.0012,
      "step": 106330
    },
    {
      "epoch": 1.7399983637404892,
      "grad_norm": 0.03417579457163811,
      "learning_rate": 4.477389241180696e-06,
      "loss": 0.0014,
      "step": 106340
    },
    {
      "epoch": 1.7401619896915652,
      "grad_norm": 0.09423999488353729,
      "learning_rate": 4.476442524773528e-06,
      "loss": 0.0017,
      "step": 106350
    },
    {
      "epoch": 1.7403256156426408,
      "grad_norm": 0.06281811743974686,
      "learning_rate": 4.475495827344061e-06,
      "loss": 0.0008,
      "step": 106360
    },
    {
      "epoch": 1.7404892415937168,
      "grad_norm": 0.07891196012496948,
      "learning_rate": 4.47454914892661e-06,
      "loss": 0.0006,
      "step": 106370
    },
    {
      "epoch": 1.7406528675447928,
      "grad_norm": 0.042788852006196976,
      "learning_rate": 4.47360248955549e-06,
      "loss": 0.0015,
      "step": 106380
    },
    {
      "epoch": 1.7408164934958683,
      "grad_norm": 0.14284440875053406,
      "learning_rate": 4.472655849265014e-06,
      "loss": 0.0019,
      "step": 106390
    },
    {
      "epoch": 1.7409801194469443,
      "grad_norm": 0.12304006516933441,
      "learning_rate": 4.4717092280894966e-06,
      "loss": 0.0013,
      "step": 106400
    },
    {
      "epoch": 1.74114374539802,
      "grad_norm": 0.14403140544891357,
      "learning_rate": 4.470762626063249e-06,
      "loss": 0.0018,
      "step": 106410
    },
    {
      "epoch": 1.7413073713490959,
      "grad_norm": 0.05266826972365379,
      "learning_rate": 4.469816043220586e-06,
      "loss": 0.0022,
      "step": 106420
    },
    {
      "epoch": 1.7414709973001719,
      "grad_norm": 0.12585389614105225,
      "learning_rate": 4.468869479595815e-06,
      "loss": 0.0022,
      "step": 106430
    },
    {
      "epoch": 1.7416346232512476,
      "grad_norm": 0.13879172503948212,
      "learning_rate": 4.467922935223251e-06,
      "loss": 0.0014,
      "step": 106440
    },
    {
      "epoch": 1.7417982492023234,
      "grad_norm": 0.043695759028196335,
      "learning_rate": 4.4669764101372e-06,
      "loss": 0.0017,
      "step": 106450
    },
    {
      "epoch": 1.7419618751533994,
      "grad_norm": 0.09431126713752747,
      "learning_rate": 4.466029904371974e-06,
      "loss": 0.002,
      "step": 106460
    },
    {
      "epoch": 1.7421255011044752,
      "grad_norm": 0.1239037811756134,
      "learning_rate": 4.46508341796188e-06,
      "loss": 0.002,
      "step": 106470
    },
    {
      "epoch": 1.742289127055551,
      "grad_norm": 0.13240762054920197,
      "learning_rate": 4.464136950941226e-06,
      "loss": 0.0016,
      "step": 106480
    },
    {
      "epoch": 1.742452753006627,
      "grad_norm": 0.03952857851982117,
      "learning_rate": 4.4631905033443204e-06,
      "loss": 0.0018,
      "step": 106490
    },
    {
      "epoch": 1.7426163789577027,
      "grad_norm": 0.2009357362985611,
      "learning_rate": 4.462244075205468e-06,
      "loss": 0.0021,
      "step": 106500
    },
    {
      "epoch": 1.7427800049087785,
      "grad_norm": 0.07433266937732697,
      "learning_rate": 4.4612976665589754e-06,
      "loss": 0.0008,
      "step": 106510
    },
    {
      "epoch": 1.7429436308598545,
      "grad_norm": 0.02268165908753872,
      "learning_rate": 4.460351277439151e-06,
      "loss": 0.0008,
      "step": 106520
    },
    {
      "epoch": 1.7431072568109303,
      "grad_norm": 0.017054162919521332,
      "learning_rate": 4.459404907880293e-06,
      "loss": 0.0016,
      "step": 106530
    },
    {
      "epoch": 1.743270882762006,
      "grad_norm": 0.10768765956163406,
      "learning_rate": 4.458458557916707e-06,
      "loss": 0.0018,
      "step": 106540
    },
    {
      "epoch": 1.743434508713082,
      "grad_norm": 0.10166916996240616,
      "learning_rate": 4.457512227582698e-06,
      "loss": 0.0005,
      "step": 106550
    },
    {
      "epoch": 1.7435981346641576,
      "grad_norm": 0.3123462200164795,
      "learning_rate": 4.456565916912567e-06,
      "loss": 0.0013,
      "step": 106560
    },
    {
      "epoch": 1.7437617606152336,
      "grad_norm": 0.03125937283039093,
      "learning_rate": 4.455619625940614e-06,
      "loss": 0.001,
      "step": 106570
    },
    {
      "epoch": 1.7439253865663096,
      "grad_norm": 0.2657507658004761,
      "learning_rate": 4.454673354701142e-06,
      "loss": 0.0026,
      "step": 106580
    },
    {
      "epoch": 1.7440890125173851,
      "grad_norm": 0.09327944368124008,
      "learning_rate": 4.45372710322845e-06,
      "loss": 0.0058,
      "step": 106590
    },
    {
      "epoch": 1.7442526384684611,
      "grad_norm": 0.09993879497051239,
      "learning_rate": 4.452780871556838e-06,
      "loss": 0.0015,
      "step": 106600
    },
    {
      "epoch": 1.744416264419537,
      "grad_norm": 0.09217282384634018,
      "learning_rate": 4.451834659720604e-06,
      "loss": 0.0016,
      "step": 106610
    },
    {
      "epoch": 1.7445798903706127,
      "grad_norm": 0.030844299122691154,
      "learning_rate": 4.450888467754048e-06,
      "loss": 0.0013,
      "step": 106620
    },
    {
      "epoch": 1.7447435163216887,
      "grad_norm": 0.023357996717095375,
      "learning_rate": 4.449942295691463e-06,
      "loss": 0.0011,
      "step": 106630
    },
    {
      "epoch": 1.7449071422727644,
      "grad_norm": 0.10645625740289688,
      "learning_rate": 4.448996143567148e-06,
      "loss": 0.0014,
      "step": 106640
    },
    {
      "epoch": 1.7450707682238402,
      "grad_norm": 0.15672515332698822,
      "learning_rate": 4.448050011415399e-06,
      "loss": 0.0025,
      "step": 106650
    },
    {
      "epoch": 1.7452343941749162,
      "grad_norm": 0.04402748495340347,
      "learning_rate": 4.44710389927051e-06,
      "loss": 0.0014,
      "step": 106660
    },
    {
      "epoch": 1.745398020125992,
      "grad_norm": 0.06813716143369675,
      "learning_rate": 4.446157807166776e-06,
      "loss": 0.0019,
      "step": 106670
    },
    {
      "epoch": 1.7455616460770678,
      "grad_norm": 0.08055794984102249,
      "learning_rate": 4.445211735138491e-06,
      "loss": 0.002,
      "step": 106680
    },
    {
      "epoch": 1.7457252720281438,
      "grad_norm": 0.14186498522758484,
      "learning_rate": 4.444265683219947e-06,
      "loss": 0.0018,
      "step": 106690
    },
    {
      "epoch": 1.7458888979792195,
      "grad_norm": 0.05113886669278145,
      "learning_rate": 4.443319651445436e-06,
      "loss": 0.0008,
      "step": 106700
    },
    {
      "epoch": 1.7460525239302953,
      "grad_norm": 0.12206219136714935,
      "learning_rate": 4.442373639849249e-06,
      "loss": 0.0016,
      "step": 106710
    },
    {
      "epoch": 1.7462161498813713,
      "grad_norm": 0.1278335154056549,
      "learning_rate": 4.441427648465679e-06,
      "loss": 0.0017,
      "step": 106720
    },
    {
      "epoch": 1.746379775832447,
      "grad_norm": 0.08471362292766571,
      "learning_rate": 4.4404816773290134e-06,
      "loss": 0.0017,
      "step": 106730
    },
    {
      "epoch": 1.7465434017835229,
      "grad_norm": 0.051166124641895294,
      "learning_rate": 4.439535726473542e-06,
      "loss": 0.0028,
      "step": 106740
    },
    {
      "epoch": 1.7467070277345988,
      "grad_norm": 0.41442620754241943,
      "learning_rate": 4.438589795933554e-06,
      "loss": 0.0018,
      "step": 106750
    },
    {
      "epoch": 1.7468706536856744,
      "grad_norm": 0.1367095708847046,
      "learning_rate": 4.437643885743335e-06,
      "loss": 0.0013,
      "step": 106760
    },
    {
      "epoch": 1.7470342796367504,
      "grad_norm": 0.0769708976149559,
      "learning_rate": 4.436697995937176e-06,
      "loss": 0.0015,
      "step": 106770
    },
    {
      "epoch": 1.7471979055878264,
      "grad_norm": 0.07013755291700363,
      "learning_rate": 4.435752126549358e-06,
      "loss": 0.0015,
      "step": 106780
    },
    {
      "epoch": 1.747361531538902,
      "grad_norm": 0.06018676608800888,
      "learning_rate": 4.43480627761417e-06,
      "loss": 0.002,
      "step": 106790
    },
    {
      "epoch": 1.747525157489978,
      "grad_norm": 0.16071408987045288,
      "learning_rate": 4.433860449165897e-06,
      "loss": 0.0015,
      "step": 106800
    },
    {
      "epoch": 1.7476887834410537,
      "grad_norm": 0.03976573795080185,
      "learning_rate": 4.432914641238822e-06,
      "loss": 0.0011,
      "step": 106810
    },
    {
      "epoch": 1.7478524093921295,
      "grad_norm": 0.04503847658634186,
      "learning_rate": 4.43196885386723e-06,
      "loss": 0.001,
      "step": 106820
    },
    {
      "epoch": 1.7480160353432055,
      "grad_norm": 0.06381500512361526,
      "learning_rate": 4.4310230870854e-06,
      "loss": 0.0015,
      "step": 106830
    },
    {
      "epoch": 1.7481796612942813,
      "grad_norm": 0.0932886004447937,
      "learning_rate": 4.430077340927615e-06,
      "loss": 0.0024,
      "step": 106840
    },
    {
      "epoch": 1.748343287245357,
      "grad_norm": 0.06423021852970123,
      "learning_rate": 4.429131615428156e-06,
      "loss": 0.0015,
      "step": 106850
    },
    {
      "epoch": 1.748506913196433,
      "grad_norm": 0.14009442925453186,
      "learning_rate": 4.428185910621306e-06,
      "loss": 0.0027,
      "step": 106860
    },
    {
      "epoch": 1.7486705391475088,
      "grad_norm": 0.28321385383605957,
      "learning_rate": 4.427240226541341e-06,
      "loss": 0.0013,
      "step": 106870
    },
    {
      "epoch": 1.7488341650985846,
      "grad_norm": 0.06612227112054825,
      "learning_rate": 4.426294563222542e-06,
      "loss": 0.0018,
      "step": 106880
    },
    {
      "epoch": 1.7489977910496606,
      "grad_norm": 0.0508941188454628,
      "learning_rate": 4.425348920699185e-06,
      "loss": 0.0008,
      "step": 106890
    },
    {
      "epoch": 1.7491614170007364,
      "grad_norm": 0.12185681611299515,
      "learning_rate": 4.42440329900555e-06,
      "loss": 0.0005,
      "step": 106900
    },
    {
      "epoch": 1.7493250429518121,
      "grad_norm": 0.16195951402187347,
      "learning_rate": 4.4234576981759115e-06,
      "loss": 0.0023,
      "step": 106910
    },
    {
      "epoch": 1.7494886689028881,
      "grad_norm": 0.059162188321352005,
      "learning_rate": 4.422512118244547e-06,
      "loss": 0.001,
      "step": 106920
    },
    {
      "epoch": 1.7496522948539637,
      "grad_norm": 0.25423464179039,
      "learning_rate": 4.421566559245728e-06,
      "loss": 0.0018,
      "step": 106930
    },
    {
      "epoch": 1.7498159208050397,
      "grad_norm": 0.2411469668149948,
      "learning_rate": 4.420621021213732e-06,
      "loss": 0.0018,
      "step": 106940
    },
    {
      "epoch": 1.7499795467561157,
      "grad_norm": 0.02898610197007656,
      "learning_rate": 4.419675504182832e-06,
      "loss": 0.0016,
      "step": 106950
    },
    {
      "epoch": 1.7501431727071912,
      "grad_norm": 0.07759073376655579,
      "learning_rate": 4.4187300081872995e-06,
      "loss": 0.0015,
      "step": 106960
    },
    {
      "epoch": 1.7503067986582672,
      "grad_norm": 0.1778598576784134,
      "learning_rate": 4.417784533261407e-06,
      "loss": 0.0017,
      "step": 106970
    },
    {
      "epoch": 1.750470424609343,
      "grad_norm": 0.07045693695545197,
      "learning_rate": 4.416839079439426e-06,
      "loss": 0.0017,
      "step": 106980
    },
    {
      "epoch": 1.7506340505604188,
      "grad_norm": 0.0670531764626503,
      "learning_rate": 4.415893646755627e-06,
      "loss": 0.0008,
      "step": 106990
    },
    {
      "epoch": 1.7507976765114948,
      "grad_norm": 0.0903734341263771,
      "learning_rate": 4.414948235244278e-06,
      "loss": 0.002,
      "step": 107000
    },
    {
      "epoch": 1.7509613024625705,
      "grad_norm": 0.05755678936839104,
      "learning_rate": 4.414002844939651e-06,
      "loss": 0.0015,
      "step": 107010
    },
    {
      "epoch": 1.7511249284136463,
      "grad_norm": 0.004980756435543299,
      "learning_rate": 4.413057475876012e-06,
      "loss": 0.0016,
      "step": 107020
    },
    {
      "epoch": 1.7512885543647223,
      "grad_norm": 0.09295650571584702,
      "learning_rate": 4.412112128087631e-06,
      "loss": 0.0014,
      "step": 107030
    },
    {
      "epoch": 1.751452180315798,
      "grad_norm": 0.025870684534311295,
      "learning_rate": 4.411166801608771e-06,
      "loss": 0.0016,
      "step": 107040
    },
    {
      "epoch": 1.7516158062668739,
      "grad_norm": 0.01784415915608406,
      "learning_rate": 4.410221496473698e-06,
      "loss": 0.0008,
      "step": 107050
    },
    {
      "epoch": 1.7517794322179499,
      "grad_norm": 0.0039959256537258625,
      "learning_rate": 4.409276212716678e-06,
      "loss": 0.0027,
      "step": 107060
    },
    {
      "epoch": 1.7519430581690256,
      "grad_norm": 0.125201016664505,
      "learning_rate": 4.408330950371976e-06,
      "loss": 0.0018,
      "step": 107070
    },
    {
      "epoch": 1.7521066841201014,
      "grad_norm": 0.09336341917514801,
      "learning_rate": 4.407385709473855e-06,
      "loss": 0.0021,
      "step": 107080
    },
    {
      "epoch": 1.7522703100711774,
      "grad_norm": 0.22945904731750488,
      "learning_rate": 4.406440490056578e-06,
      "loss": 0.0017,
      "step": 107090
    },
    {
      "epoch": 1.7524339360222532,
      "grad_norm": 0.023744212463498116,
      "learning_rate": 4.405495292154407e-06,
      "loss": 0.0012,
      "step": 107100
    },
    {
      "epoch": 1.752597561973329,
      "grad_norm": 0.04172080010175705,
      "learning_rate": 4.404550115801602e-06,
      "loss": 0.0013,
      "step": 107110
    },
    {
      "epoch": 1.752761187924405,
      "grad_norm": 0.09865215420722961,
      "learning_rate": 4.403604961032425e-06,
      "loss": 0.002,
      "step": 107120
    },
    {
      "epoch": 1.7529248138754805,
      "grad_norm": 0.0782988891005516,
      "learning_rate": 4.402659827881136e-06,
      "loss": 0.0014,
      "step": 107130
    },
    {
      "epoch": 1.7530884398265565,
      "grad_norm": 0.09502945840358734,
      "learning_rate": 4.401714716381991e-06,
      "loss": 0.0044,
      "step": 107140
    },
    {
      "epoch": 1.7532520657776325,
      "grad_norm": 0.05701727047562599,
      "learning_rate": 4.400769626569249e-06,
      "loss": 0.0017,
      "step": 107150
    },
    {
      "epoch": 1.753415691728708,
      "grad_norm": 0.1252390444278717,
      "learning_rate": 4.3998245584771685e-06,
      "loss": 0.0015,
      "step": 107160
    },
    {
      "epoch": 1.753579317679784,
      "grad_norm": 0.2256326526403427,
      "learning_rate": 4.398879512140005e-06,
      "loss": 0.0014,
      "step": 107170
    },
    {
      "epoch": 1.7537429436308598,
      "grad_norm": 0.29919615387916565,
      "learning_rate": 4.397934487592015e-06,
      "loss": 0.0014,
      "step": 107180
    },
    {
      "epoch": 1.7539065695819356,
      "grad_norm": 0.02892584167420864,
      "learning_rate": 4.396989484867452e-06,
      "loss": 0.0024,
      "step": 107190
    },
    {
      "epoch": 1.7540701955330116,
      "grad_norm": 0.013569826260209084,
      "learning_rate": 4.396044504000571e-06,
      "loss": 0.0008,
      "step": 107200
    },
    {
      "epoch": 1.7542338214840874,
      "grad_norm": 0.038387250155210495,
      "learning_rate": 4.395099545025626e-06,
      "loss": 0.0026,
      "step": 107210
    },
    {
      "epoch": 1.7543974474351631,
      "grad_norm": 0.3043919503688812,
      "learning_rate": 4.394154607976867e-06,
      "loss": 0.0024,
      "step": 107220
    },
    {
      "epoch": 1.7545610733862391,
      "grad_norm": 0.1005256325006485,
      "learning_rate": 4.39320969288855e-06,
      "loss": 0.0023,
      "step": 107230
    },
    {
      "epoch": 1.754724699337315,
      "grad_norm": 0.0739796981215477,
      "learning_rate": 4.392264799794923e-06,
      "loss": 0.002,
      "step": 107240
    },
    {
      "epoch": 1.7548883252883907,
      "grad_norm": 0.0268404521048069,
      "learning_rate": 4.391319928730235e-06,
      "loss": 0.0009,
      "step": 107250
    },
    {
      "epoch": 1.7550519512394667,
      "grad_norm": 0.05511759594082832,
      "learning_rate": 4.390375079728736e-06,
      "loss": 0.0023,
      "step": 107260
    },
    {
      "epoch": 1.7552155771905424,
      "grad_norm": 0.030853334814310074,
      "learning_rate": 4.389430252824675e-06,
      "loss": 0.0012,
      "step": 107270
    },
    {
      "epoch": 1.7553792031416182,
      "grad_norm": 0.3306729793548584,
      "learning_rate": 4.388485448052301e-06,
      "loss": 0.0022,
      "step": 107280
    },
    {
      "epoch": 1.7555428290926942,
      "grad_norm": 0.06889146566390991,
      "learning_rate": 4.387540665445859e-06,
      "loss": 0.002,
      "step": 107290
    },
    {
      "epoch": 1.75570645504377,
      "grad_norm": 0.0657404437661171,
      "learning_rate": 4.386595905039597e-06,
      "loss": 0.001,
      "step": 107300
    },
    {
      "epoch": 1.7558700809948458,
      "grad_norm": 0.10056816041469574,
      "learning_rate": 4.3856511668677585e-06,
      "loss": 0.0015,
      "step": 107310
    },
    {
      "epoch": 1.7560337069459218,
      "grad_norm": 0.16094477474689484,
      "learning_rate": 4.3847064509645884e-06,
      "loss": 0.0017,
      "step": 107320
    },
    {
      "epoch": 1.7561973328969973,
      "grad_norm": 0.06102989614009857,
      "learning_rate": 4.383761757364333e-06,
      "loss": 0.0013,
      "step": 107330
    },
    {
      "epoch": 1.7563609588480733,
      "grad_norm": 0.2393198162317276,
      "learning_rate": 4.38281708610123e-06,
      "loss": 0.0016,
      "step": 107340
    },
    {
      "epoch": 1.7565245847991493,
      "grad_norm": 0.03686806187033653,
      "learning_rate": 4.381872437209525e-06,
      "loss": 0.0007,
      "step": 107350
    },
    {
      "epoch": 1.7566882107502249,
      "grad_norm": 0.05576544627547264,
      "learning_rate": 4.380927810723458e-06,
      "loss": 0.0012,
      "step": 107360
    },
    {
      "epoch": 1.7568518367013009,
      "grad_norm": 0.10331454873085022,
      "learning_rate": 4.3799832066772705e-06,
      "loss": 0.0013,
      "step": 107370
    },
    {
      "epoch": 1.7570154626523766,
      "grad_norm": 0.10350741446018219,
      "learning_rate": 4.379038625105202e-06,
      "loss": 0.0022,
      "step": 107380
    },
    {
      "epoch": 1.7571790886034524,
      "grad_norm": 0.07576325535774231,
      "learning_rate": 4.378094066041491e-06,
      "loss": 0.0016,
      "step": 107390
    },
    {
      "epoch": 1.7573427145545284,
      "grad_norm": 0.08821357786655426,
      "learning_rate": 4.377149529520374e-06,
      "loss": 0.0011,
      "step": 107400
    },
    {
      "epoch": 1.7575063405056042,
      "grad_norm": 0.12737809121608734,
      "learning_rate": 4.376205015576091e-06,
      "loss": 0.0015,
      "step": 107410
    },
    {
      "epoch": 1.75766996645668,
      "grad_norm": 0.06821626424789429,
      "learning_rate": 4.375260524242876e-06,
      "loss": 0.0014,
      "step": 107420
    },
    {
      "epoch": 1.757833592407756,
      "grad_norm": 0.042156461626291275,
      "learning_rate": 4.374316055554967e-06,
      "loss": 0.0008,
      "step": 107430
    },
    {
      "epoch": 1.7579972183588317,
      "grad_norm": 0.26228514313697815,
      "learning_rate": 4.3733716095465965e-06,
      "loss": 0.0023,
      "step": 107440
    },
    {
      "epoch": 1.7581608443099075,
      "grad_norm": 0.03440668806433678,
      "learning_rate": 4.372427186251998e-06,
      "loss": 0.0014,
      "step": 107450
    },
    {
      "epoch": 1.7583244702609835,
      "grad_norm": 0.10256185382604599,
      "learning_rate": 4.371482785705407e-06,
      "loss": 0.002,
      "step": 107460
    },
    {
      "epoch": 1.7584880962120593,
      "grad_norm": 0.06891493499279022,
      "learning_rate": 4.370538407941054e-06,
      "loss": 0.0008,
      "step": 107470
    },
    {
      "epoch": 1.758651722163135,
      "grad_norm": 0.17362768948078156,
      "learning_rate": 4.36959405299317e-06,
      "loss": 0.0038,
      "step": 107480
    },
    {
      "epoch": 1.758815348114211,
      "grad_norm": 0.004728842992335558,
      "learning_rate": 4.368649720895987e-06,
      "loss": 0.0013,
      "step": 107490
    },
    {
      "epoch": 1.7589789740652868,
      "grad_norm": 0.0966043695807457,
      "learning_rate": 4.367705411683733e-06,
      "loss": 0.0023,
      "step": 107500
    },
    {
      "epoch": 1.7591426000163626,
      "grad_norm": 0.1799917370080948,
      "learning_rate": 4.3667611253906395e-06,
      "loss": 0.0009,
      "step": 107510
    },
    {
      "epoch": 1.7593062259674386,
      "grad_norm": 0.06199290230870247,
      "learning_rate": 4.365816862050933e-06,
      "loss": 0.0016,
      "step": 107520
    },
    {
      "epoch": 1.7594698519185141,
      "grad_norm": 0.005445132963359356,
      "learning_rate": 4.364872621698844e-06,
      "loss": 0.0014,
      "step": 107530
    },
    {
      "epoch": 1.7596334778695901,
      "grad_norm": 0.11272435635328293,
      "learning_rate": 4.363928404368592e-06,
      "loss": 0.0015,
      "step": 107540
    },
    {
      "epoch": 1.7597971038206661,
      "grad_norm": 0.09983003884553909,
      "learning_rate": 4.362984210094408e-06,
      "loss": 0.002,
      "step": 107550
    },
    {
      "epoch": 1.7599607297717417,
      "grad_norm": 0.09683094173669815,
      "learning_rate": 4.362040038910514e-06,
      "loss": 0.0018,
      "step": 107560
    },
    {
      "epoch": 1.7601243557228177,
      "grad_norm": 0.023452797904610634,
      "learning_rate": 4.361095890851137e-06,
      "loss": 0.0013,
      "step": 107570
    },
    {
      "epoch": 1.7602879816738934,
      "grad_norm": 0.012471520341932774,
      "learning_rate": 4.360151765950498e-06,
      "loss": 0.0016,
      "step": 107580
    },
    {
      "epoch": 1.7604516076249692,
      "grad_norm": 0.06669203191995621,
      "learning_rate": 4.3592076642428185e-06,
      "loss": 0.0012,
      "step": 107590
    },
    {
      "epoch": 1.7606152335760452,
      "grad_norm": 0.2633495032787323,
      "learning_rate": 4.358263585762323e-06,
      "loss": 0.0019,
      "step": 107600
    },
    {
      "epoch": 1.760778859527121,
      "grad_norm": 0.08729929476976395,
      "learning_rate": 4.357319530543228e-06,
      "loss": 0.0012,
      "step": 107610
    },
    {
      "epoch": 1.7609424854781968,
      "grad_norm": 0.20309489965438843,
      "learning_rate": 4.356375498619757e-06,
      "loss": 0.0014,
      "step": 107620
    },
    {
      "epoch": 1.7611061114292728,
      "grad_norm": 0.029592743143439293,
      "learning_rate": 4.355431490026127e-06,
      "loss": 0.0011,
      "step": 107630
    },
    {
      "epoch": 1.7612697373803485,
      "grad_norm": 0.16835619509220123,
      "learning_rate": 4.3544875047965565e-06,
      "loss": 0.0018,
      "step": 107640
    },
    {
      "epoch": 1.7614333633314243,
      "grad_norm": 0.20504416525363922,
      "learning_rate": 4.353543542965262e-06,
      "loss": 0.0028,
      "step": 107650
    },
    {
      "epoch": 1.7615969892825003,
      "grad_norm": 0.08531437814235687,
      "learning_rate": 4.352599604566461e-06,
      "loss": 0.0016,
      "step": 107660
    },
    {
      "epoch": 1.761760615233576,
      "grad_norm": 0.12506519258022308,
      "learning_rate": 4.351655689634367e-06,
      "loss": 0.0017,
      "step": 107670
    },
    {
      "epoch": 1.7619242411846519,
      "grad_norm": 0.06132843717932701,
      "learning_rate": 4.350711798203197e-06,
      "loss": 0.002,
      "step": 107680
    },
    {
      "epoch": 1.7620878671357278,
      "grad_norm": 0.19101710617542267,
      "learning_rate": 4.3497679303071636e-06,
      "loss": 0.001,
      "step": 107690
    },
    {
      "epoch": 1.7622514930868036,
      "grad_norm": 0.0844285637140274,
      "learning_rate": 4.34882408598048e-06,
      "loss": 0.0015,
      "step": 107700
    },
    {
      "epoch": 1.7624151190378794,
      "grad_norm": 0.12264546006917953,
      "learning_rate": 4.3478802652573585e-06,
      "loss": 0.0021,
      "step": 107710
    },
    {
      "epoch": 1.7625787449889554,
      "grad_norm": 0.1508146971464157,
      "learning_rate": 4.346936468172008e-06,
      "loss": 0.0016,
      "step": 107720
    },
    {
      "epoch": 1.762742370940031,
      "grad_norm": 0.04992277920246124,
      "learning_rate": 4.3459926947586454e-06,
      "loss": 0.0023,
      "step": 107730
    },
    {
      "epoch": 1.762905996891107,
      "grad_norm": 0.05946950986981392,
      "learning_rate": 4.345048945051472e-06,
      "loss": 0.0027,
      "step": 107740
    },
    {
      "epoch": 1.7630696228421827,
      "grad_norm": 0.02963419444859028,
      "learning_rate": 4.344105219084702e-06,
      "loss": 0.0013,
      "step": 107750
    },
    {
      "epoch": 1.7632332487932585,
      "grad_norm": 0.13194815814495087,
      "learning_rate": 4.34316151689254e-06,
      "loss": 0.001,
      "step": 107760
    },
    {
      "epoch": 1.7633968747443345,
      "grad_norm": 0.04725028946995735,
      "learning_rate": 4.342217838509195e-06,
      "loss": 0.0015,
      "step": 107770
    },
    {
      "epoch": 1.7635605006954103,
      "grad_norm": 0.09281488507986069,
      "learning_rate": 4.341274183968871e-06,
      "loss": 0.0011,
      "step": 107780
    },
    {
      "epoch": 1.763724126646486,
      "grad_norm": 0.07120165973901749,
      "learning_rate": 4.340330553305776e-06,
      "loss": 0.0016,
      "step": 107790
    },
    {
      "epoch": 1.763887752597562,
      "grad_norm": 0.07696498930454254,
      "learning_rate": 4.339386946554113e-06,
      "loss": 0.001,
      "step": 107800
    },
    {
      "epoch": 1.7640513785486378,
      "grad_norm": 0.07577105611562729,
      "learning_rate": 4.338443363748086e-06,
      "loss": 0.0012,
      "step": 107810
    },
    {
      "epoch": 1.7642150044997136,
      "grad_norm": 0.09016141295433044,
      "learning_rate": 4.337499804921896e-06,
      "loss": 0.0019,
      "step": 107820
    },
    {
      "epoch": 1.7643786304507896,
      "grad_norm": 0.04631270840764046,
      "learning_rate": 4.336556270109748e-06,
      "loss": 0.0011,
      "step": 107830
    },
    {
      "epoch": 1.7645422564018653,
      "grad_norm": 0.002991677960380912,
      "learning_rate": 4.335612759345838e-06,
      "loss": 0.0036,
      "step": 107840
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.286753386259079,
      "learning_rate": 4.33466927266437e-06,
      "loss": 0.0033,
      "step": 107850
    },
    {
      "epoch": 1.7648695083040171,
      "grad_norm": 0.1092371866106987,
      "learning_rate": 4.333725810099541e-06,
      "loss": 0.0034,
      "step": 107860
    },
    {
      "epoch": 1.765033134255093,
      "grad_norm": 0.022794928401708603,
      "learning_rate": 4.33278237168555e-06,
      "loss": 0.0007,
      "step": 107870
    },
    {
      "epoch": 1.7651967602061687,
      "grad_norm": 0.07644696533679962,
      "learning_rate": 4.331838957456595e-06,
      "loss": 0.0014,
      "step": 107880
    },
    {
      "epoch": 1.7653603861572447,
      "grad_norm": 0.08925812691450119,
      "learning_rate": 4.330895567446873e-06,
      "loss": 0.0011,
      "step": 107890
    },
    {
      "epoch": 1.7655240121083202,
      "grad_norm": 0.17215490341186523,
      "learning_rate": 4.3299522016905756e-06,
      "loss": 0.0016,
      "step": 107900
    },
    {
      "epoch": 1.7656876380593962,
      "grad_norm": 0.1430642157793045,
      "learning_rate": 4.329008860221903e-06,
      "loss": 0.0021,
      "step": 107910
    },
    {
      "epoch": 1.7658512640104722,
      "grad_norm": 0.09571950882673264,
      "learning_rate": 4.328065543075045e-06,
      "loss": 0.003,
      "step": 107920
    },
    {
      "epoch": 1.7660148899615478,
      "grad_norm": 0.07642622292041779,
      "learning_rate": 4.327122250284198e-06,
      "loss": 0.0018,
      "step": 107930
    },
    {
      "epoch": 1.7661785159126238,
      "grad_norm": 0.07825654000043869,
      "learning_rate": 4.326178981883551e-06,
      "loss": 0.0019,
      "step": 107940
    },
    {
      "epoch": 1.7663421418636995,
      "grad_norm": 0.008431176654994488,
      "learning_rate": 4.325235737907296e-06,
      "loss": 0.0017,
      "step": 107950
    },
    {
      "epoch": 1.7665057678147753,
      "grad_norm": 0.0462493970990181,
      "learning_rate": 4.3242925183896246e-06,
      "loss": 0.0017,
      "step": 107960
    },
    {
      "epoch": 1.7666693937658513,
      "grad_norm": 0.016918199136853218,
      "learning_rate": 4.323349323364725e-06,
      "loss": 0.0048,
      "step": 107970
    },
    {
      "epoch": 1.766833019716927,
      "grad_norm": 0.2721311151981354,
      "learning_rate": 4.322406152866786e-06,
      "loss": 0.0018,
      "step": 107980
    },
    {
      "epoch": 1.7669966456680029,
      "grad_norm": 0.07073357701301575,
      "learning_rate": 4.321463006929994e-06,
      "loss": 0.0014,
      "step": 107990
    },
    {
      "epoch": 1.7671602716190788,
      "grad_norm": 0.05635339021682739,
      "learning_rate": 4.320519885588539e-06,
      "loss": 0.0009,
      "step": 108000
    },
    {
      "epoch": 1.7673238975701546,
      "grad_norm": 0.1365121603012085,
      "learning_rate": 4.319576788876605e-06,
      "loss": 0.0029,
      "step": 108010
    },
    {
      "epoch": 1.7674875235212304,
      "grad_norm": 0.09400023519992828,
      "learning_rate": 4.318633716828376e-06,
      "loss": 0.0014,
      "step": 108020
    },
    {
      "epoch": 1.7676511494723064,
      "grad_norm": 0.09861313551664352,
      "learning_rate": 4.31769066947804e-06,
      "loss": 0.0018,
      "step": 108030
    },
    {
      "epoch": 1.7678147754233822,
      "grad_norm": 0.0662696436047554,
      "learning_rate": 4.316747646859775e-06,
      "loss": 0.0019,
      "step": 108040
    },
    {
      "epoch": 1.767978401374458,
      "grad_norm": 0.1007886677980423,
      "learning_rate": 4.315804649007765e-06,
      "loss": 0.0013,
      "step": 108050
    },
    {
      "epoch": 1.768142027325534,
      "grad_norm": 0.09016755223274231,
      "learning_rate": 4.314861675956193e-06,
      "loss": 0.001,
      "step": 108060
    },
    {
      "epoch": 1.7683056532766097,
      "grad_norm": 0.13750973343849182,
      "learning_rate": 4.313918727739239e-06,
      "loss": 0.0015,
      "step": 108070
    },
    {
      "epoch": 1.7684692792276855,
      "grad_norm": 0.06927937269210815,
      "learning_rate": 4.312975804391081e-06,
      "loss": 0.0018,
      "step": 108080
    },
    {
      "epoch": 1.7686329051787615,
      "grad_norm": 0.05693129077553749,
      "learning_rate": 4.3120329059459e-06,
      "loss": 0.0006,
      "step": 108090
    },
    {
      "epoch": 1.768796531129837,
      "grad_norm": 0.25312113761901855,
      "learning_rate": 4.3110900324378715e-06,
      "loss": 0.0014,
      "step": 108100
    },
    {
      "epoch": 1.768960157080913,
      "grad_norm": 0.04375087842345238,
      "learning_rate": 4.310147183901175e-06,
      "loss": 0.0017,
      "step": 108110
    },
    {
      "epoch": 1.769123783031989,
      "grad_norm": 0.07191846519708633,
      "learning_rate": 4.309204360369984e-06,
      "loss": 0.0016,
      "step": 108120
    },
    {
      "epoch": 1.7692874089830646,
      "grad_norm": 0.15051211416721344,
      "learning_rate": 4.308261561878476e-06,
      "loss": 0.0016,
      "step": 108130
    },
    {
      "epoch": 1.7694510349341406,
      "grad_norm": 0.06570404022932053,
      "learning_rate": 4.3073187884608226e-06,
      "loss": 0.0008,
      "step": 108140
    },
    {
      "epoch": 1.7696146608852164,
      "grad_norm": 0.18022041022777557,
      "learning_rate": 4.306376040151199e-06,
      "loss": 0.002,
      "step": 108150
    },
    {
      "epoch": 1.7697782868362921,
      "grad_norm": 0.10511868447065353,
      "learning_rate": 4.305433316983776e-06,
      "loss": 0.0016,
      "step": 108160
    },
    {
      "epoch": 1.7699419127873681,
      "grad_norm": 0.11730948835611343,
      "learning_rate": 4.304490618992726e-06,
      "loss": 0.0016,
      "step": 108170
    },
    {
      "epoch": 1.770105538738444,
      "grad_norm": 0.007314768619835377,
      "learning_rate": 4.3035479462122195e-06,
      "loss": 0.0011,
      "step": 108180
    },
    {
      "epoch": 1.7702691646895197,
      "grad_norm": 0.4316589832305908,
      "learning_rate": 4.302605298676427e-06,
      "loss": 0.0014,
      "step": 108190
    },
    {
      "epoch": 1.7704327906405957,
      "grad_norm": 0.03252164646983147,
      "learning_rate": 4.3016626764195155e-06,
      "loss": 0.001,
      "step": 108200
    },
    {
      "epoch": 1.7705964165916714,
      "grad_norm": 0.07453364133834839,
      "learning_rate": 4.300720079475653e-06,
      "loss": 0.001,
      "step": 108210
    },
    {
      "epoch": 1.7707600425427472,
      "grad_norm": 0.007671711500734091,
      "learning_rate": 4.299777507879007e-06,
      "loss": 0.0012,
      "step": 108220
    },
    {
      "epoch": 1.7709236684938232,
      "grad_norm": 0.1688695251941681,
      "learning_rate": 4.2988349616637455e-06,
      "loss": 0.002,
      "step": 108230
    },
    {
      "epoch": 1.771087294444899,
      "grad_norm": 0.02427785098552704,
      "learning_rate": 4.297892440864031e-06,
      "loss": 0.0018,
      "step": 108240
    },
    {
      "epoch": 1.7712509203959748,
      "grad_norm": 0.15369784832000732,
      "learning_rate": 4.296949945514027e-06,
      "loss": 0.0024,
      "step": 108250
    },
    {
      "epoch": 1.7714145463470508,
      "grad_norm": 0.06314118951559067,
      "learning_rate": 4.296007475647897e-06,
      "loss": 0.0012,
      "step": 108260
    },
    {
      "epoch": 1.7715781722981265,
      "grad_norm": 0.16762550175189972,
      "learning_rate": 4.2950650312998036e-06,
      "loss": 0.0017,
      "step": 108270
    },
    {
      "epoch": 1.7717417982492023,
      "grad_norm": 0.03174692392349243,
      "learning_rate": 4.2941226125039105e-06,
      "loss": 0.0011,
      "step": 108280
    },
    {
      "epoch": 1.7719054242002783,
      "grad_norm": 0.376834899187088,
      "learning_rate": 4.293180219294375e-06,
      "loss": 0.0023,
      "step": 108290
    },
    {
      "epoch": 1.7720690501513539,
      "grad_norm": 0.09652141481637955,
      "learning_rate": 4.292237851705358e-06,
      "loss": 0.0019,
      "step": 108300
    },
    {
      "epoch": 1.7722326761024298,
      "grad_norm": 0.15823204815387726,
      "learning_rate": 4.291295509771018e-06,
      "loss": 0.0019,
      "step": 108310
    },
    {
      "epoch": 1.7723963020535058,
      "grad_norm": 0.15063823759555817,
      "learning_rate": 4.290353193525512e-06,
      "loss": 0.0018,
      "step": 108320
    },
    {
      "epoch": 1.7725599280045814,
      "grad_norm": 0.08814850449562073,
      "learning_rate": 4.289410903002999e-06,
      "loss": 0.0013,
      "step": 108330
    },
    {
      "epoch": 1.7727235539556574,
      "grad_norm": 0.08576839417219162,
      "learning_rate": 4.288468638237631e-06,
      "loss": 0.0034,
      "step": 108340
    },
    {
      "epoch": 1.7728871799067332,
      "grad_norm": 0.08166787773370743,
      "learning_rate": 4.287526399263565e-06,
      "loss": 0.0013,
      "step": 108350
    },
    {
      "epoch": 1.773050805857809,
      "grad_norm": 0.11101599782705307,
      "learning_rate": 4.286584186114955e-06,
      "loss": 0.0014,
      "step": 108360
    },
    {
      "epoch": 1.773214431808885,
      "grad_norm": 0.039324697107076645,
      "learning_rate": 4.285641998825953e-06,
      "loss": 0.0009,
      "step": 108370
    },
    {
      "epoch": 1.7733780577599607,
      "grad_norm": 0.07791756838560104,
      "learning_rate": 4.284699837430713e-06,
      "loss": 0.0022,
      "step": 108380
    },
    {
      "epoch": 1.7735416837110365,
      "grad_norm": 0.16867142915725708,
      "learning_rate": 4.283757701963383e-06,
      "loss": 0.0013,
      "step": 108390
    },
    {
      "epoch": 1.7737053096621125,
      "grad_norm": 0.04311840608716011,
      "learning_rate": 4.282815592458116e-06,
      "loss": 0.0015,
      "step": 108400
    },
    {
      "epoch": 1.7738689356131883,
      "grad_norm": 0.008796016685664654,
      "learning_rate": 4.281873508949059e-06,
      "loss": 0.0033,
      "step": 108410
    },
    {
      "epoch": 1.774032561564264,
      "grad_norm": 0.034651562571525574,
      "learning_rate": 4.280931451470362e-06,
      "loss": 0.0016,
      "step": 108420
    },
    {
      "epoch": 1.77419618751534,
      "grad_norm": 0.16715854406356812,
      "learning_rate": 4.279989420056173e-06,
      "loss": 0.0014,
      "step": 108430
    },
    {
      "epoch": 1.7743598134664158,
      "grad_norm": 0.19988031685352325,
      "learning_rate": 4.279047414740636e-06,
      "loss": 0.0019,
      "step": 108440
    },
    {
      "epoch": 1.7745234394174916,
      "grad_norm": 0.11294061690568924,
      "learning_rate": 4.2781054355578975e-06,
      "loss": 0.0011,
      "step": 108450
    },
    {
      "epoch": 1.7746870653685676,
      "grad_norm": 0.10826808214187622,
      "learning_rate": 4.277163482542101e-06,
      "loss": 0.0017,
      "step": 108460
    },
    {
      "epoch": 1.7748506913196433,
      "grad_norm": 0.053027503192424774,
      "learning_rate": 4.2762215557273914e-06,
      "loss": 0.0015,
      "step": 108470
    },
    {
      "epoch": 1.7750143172707191,
      "grad_norm": 0.009415642358362675,
      "learning_rate": 4.275279655147911e-06,
      "loss": 0.0014,
      "step": 108480
    },
    {
      "epoch": 1.7751779432217951,
      "grad_norm": 0.13150624930858612,
      "learning_rate": 4.2743377808378004e-06,
      "loss": 0.0013,
      "step": 108490
    },
    {
      "epoch": 1.7753415691728707,
      "grad_norm": 0.08152973651885986,
      "learning_rate": 4.2733959328312016e-06,
      "loss": 0.0007,
      "step": 108500
    },
    {
      "epoch": 1.7755051951239467,
      "grad_norm": 0.05660593509674072,
      "learning_rate": 4.2724541111622545e-06,
      "loss": 0.0037,
      "step": 108510
    },
    {
      "epoch": 1.7756688210750227,
      "grad_norm": 0.052778106182813644,
      "learning_rate": 4.271512315865097e-06,
      "loss": 0.0024,
      "step": 108520
    },
    {
      "epoch": 1.7758324470260982,
      "grad_norm": 0.0690077692270279,
      "learning_rate": 4.270570546973869e-06,
      "loss": 0.0024,
      "step": 108530
    },
    {
      "epoch": 1.7759960729771742,
      "grad_norm": 0.25202247500419617,
      "learning_rate": 4.2696288045227045e-06,
      "loss": 0.0017,
      "step": 108540
    },
    {
      "epoch": 1.77615969892825,
      "grad_norm": 0.00635350588709116,
      "learning_rate": 4.268687088545738e-06,
      "loss": 0.0007,
      "step": 108550
    },
    {
      "epoch": 1.7763233248793258,
      "grad_norm": 0.13210521638393402,
      "learning_rate": 4.267745399077109e-06,
      "loss": 0.0015,
      "step": 108560
    },
    {
      "epoch": 1.7764869508304018,
      "grad_norm": 0.019530724734067917,
      "learning_rate": 4.266803736150949e-06,
      "loss": 0.0012,
      "step": 108570
    },
    {
      "epoch": 1.7766505767814775,
      "grad_norm": 0.1282965987920761,
      "learning_rate": 4.265862099801391e-06,
      "loss": 0.0015,
      "step": 108580
    },
    {
      "epoch": 1.7768142027325533,
      "grad_norm": 0.14573904871940613,
      "learning_rate": 4.264920490062568e-06,
      "loss": 0.0019,
      "step": 108590
    },
    {
      "epoch": 1.7769778286836293,
      "grad_norm": 0.07039479911327362,
      "learning_rate": 4.26397890696861e-06,
      "loss": 0.0011,
      "step": 108600
    },
    {
      "epoch": 1.777141454634705,
      "grad_norm": 0.08231132477521896,
      "learning_rate": 4.2630373505536485e-06,
      "loss": 0.001,
      "step": 108610
    },
    {
      "epoch": 1.7773050805857808,
      "grad_norm": 0.23946666717529297,
      "learning_rate": 4.26209582085181e-06,
      "loss": 0.002,
      "step": 108620
    },
    {
      "epoch": 1.7774687065368568,
      "grad_norm": 0.07521241903305054,
      "learning_rate": 4.2611543178972274e-06,
      "loss": 0.0011,
      "step": 108630
    },
    {
      "epoch": 1.7776323324879326,
      "grad_norm": 0.06746865063905716,
      "learning_rate": 4.260212841724023e-06,
      "loss": 0.001,
      "step": 108640
    },
    {
      "epoch": 1.7777959584390084,
      "grad_norm": 0.05463298410177231,
      "learning_rate": 4.259271392366325e-06,
      "loss": 0.0014,
      "step": 108650
    },
    {
      "epoch": 1.7779595843900844,
      "grad_norm": 0.08373922854661942,
      "learning_rate": 4.25832996985826e-06,
      "loss": 0.0011,
      "step": 108660
    },
    {
      "epoch": 1.77812321034116,
      "grad_norm": 0.06922430545091629,
      "learning_rate": 4.25738857423395e-06,
      "loss": 0.0009,
      "step": 108670
    },
    {
      "epoch": 1.778286836292236,
      "grad_norm": 0.09196797758340836,
      "learning_rate": 4.25644720552752e-06,
      "loss": 0.0017,
      "step": 108680
    },
    {
      "epoch": 1.778450462243312,
      "grad_norm": 0.2511034607887268,
      "learning_rate": 4.25550586377309e-06,
      "loss": 0.0014,
      "step": 108690
    },
    {
      "epoch": 1.7786140881943875,
      "grad_norm": 0.08643194288015366,
      "learning_rate": 4.254564549004785e-06,
      "loss": 0.0012,
      "step": 108700
    },
    {
      "epoch": 1.7787777141454635,
      "grad_norm": 0.0703665018081665,
      "learning_rate": 4.253623261256721e-06,
      "loss": 0.0022,
      "step": 108710
    },
    {
      "epoch": 1.7789413400965393,
      "grad_norm": 0.13588392734527588,
      "learning_rate": 4.252682000563022e-06,
      "loss": 0.0019,
      "step": 108720
    },
    {
      "epoch": 1.779104966047615,
      "grad_norm": 0.0583539679646492,
      "learning_rate": 4.251740766957806e-06,
      "loss": 0.0017,
      "step": 108730
    },
    {
      "epoch": 1.779268591998691,
      "grad_norm": 0.05728354677557945,
      "learning_rate": 4.250799560475187e-06,
      "loss": 0.0014,
      "step": 108740
    },
    {
      "epoch": 1.7794322179497668,
      "grad_norm": 0.14650148153305054,
      "learning_rate": 4.249858381149283e-06,
      "loss": 0.0012,
      "step": 108750
    },
    {
      "epoch": 1.7795958439008426,
      "grad_norm": 0.012672608718276024,
      "learning_rate": 4.24891722901421e-06,
      "loss": 0.0038,
      "step": 108760
    },
    {
      "epoch": 1.7797594698519186,
      "grad_norm": 0.07040093839168549,
      "learning_rate": 4.247976104104081e-06,
      "loss": 0.0009,
      "step": 108770
    },
    {
      "epoch": 1.7799230958029943,
      "grad_norm": 0.03387663885951042,
      "learning_rate": 4.247035006453012e-06,
      "loss": 0.0016,
      "step": 108780
    },
    {
      "epoch": 1.7800867217540701,
      "grad_norm": 0.16323858499526978,
      "learning_rate": 4.246093936095115e-06,
      "loss": 0.0015,
      "step": 108790
    },
    {
      "epoch": 1.7802503477051461,
      "grad_norm": 0.05390274524688721,
      "learning_rate": 4.2451528930645e-06,
      "loss": 0.0014,
      "step": 108800
    },
    {
      "epoch": 1.780413973656222,
      "grad_norm": 0.070576012134552,
      "learning_rate": 4.244211877395278e-06,
      "loss": 0.0019,
      "step": 108810
    },
    {
      "epoch": 1.7805775996072977,
      "grad_norm": 0.01758529618382454,
      "learning_rate": 4.24327088912156e-06,
      "loss": 0.0008,
      "step": 108820
    },
    {
      "epoch": 1.7807412255583737,
      "grad_norm": 0.03800707682967186,
      "learning_rate": 4.2423299282774534e-06,
      "loss": 0.0013,
      "step": 108830
    },
    {
      "epoch": 1.7809048515094494,
      "grad_norm": 0.09394492954015732,
      "learning_rate": 4.241388994897066e-06,
      "loss": 0.001,
      "step": 108840
    },
    {
      "epoch": 1.7810684774605252,
      "grad_norm": 0.09371686726808548,
      "learning_rate": 4.240448089014503e-06,
      "loss": 0.0024,
      "step": 108850
    },
    {
      "epoch": 1.7812321034116012,
      "grad_norm": 0.1738329529762268,
      "learning_rate": 4.239507210663873e-06,
      "loss": 0.0008,
      "step": 108860
    },
    {
      "epoch": 1.7813957293626768,
      "grad_norm": 0.027741970494389534,
      "learning_rate": 4.238566359879277e-06,
      "loss": 0.0015,
      "step": 108870
    },
    {
      "epoch": 1.7815593553137528,
      "grad_norm": 0.21437999606132507,
      "learning_rate": 4.237625536694821e-06,
      "loss": 0.0018,
      "step": 108880
    },
    {
      "epoch": 1.7817229812648288,
      "grad_norm": 0.07579521089792252,
      "learning_rate": 4.236684741144608e-06,
      "loss": 0.0017,
      "step": 108890
    },
    {
      "epoch": 1.7818866072159043,
      "grad_norm": 0.0707777738571167,
      "learning_rate": 4.235743973262737e-06,
      "loss": 0.0053,
      "step": 108900
    },
    {
      "epoch": 1.7820502331669803,
      "grad_norm": 0.06048990413546562,
      "learning_rate": 4.2348032330833115e-06,
      "loss": 0.0016,
      "step": 108910
    },
    {
      "epoch": 1.782213859118056,
      "grad_norm": 0.3154433071613312,
      "learning_rate": 4.233862520640428e-06,
      "loss": 0.0018,
      "step": 108920
    },
    {
      "epoch": 1.7823774850691319,
      "grad_norm": 0.07315990328788757,
      "learning_rate": 4.232921835968189e-06,
      "loss": 0.0014,
      "step": 108930
    },
    {
      "epoch": 1.7825411110202078,
      "grad_norm": 0.1767452359199524,
      "learning_rate": 4.231981179100688e-06,
      "loss": 0.0018,
      "step": 108940
    },
    {
      "epoch": 1.7827047369712836,
      "grad_norm": 0.3045351207256317,
      "learning_rate": 4.231040550072025e-06,
      "loss": 0.0024,
      "step": 108950
    },
    {
      "epoch": 1.7828683629223594,
      "grad_norm": 0.06885327398777008,
      "learning_rate": 4.230099948916293e-06,
      "loss": 0.0016,
      "step": 108960
    },
    {
      "epoch": 1.7830319888734354,
      "grad_norm": 0.09926348179578781,
      "learning_rate": 4.229159375667587e-06,
      "loss": 0.0013,
      "step": 108970
    },
    {
      "epoch": 1.7831956148245112,
      "grad_norm": 0.05821095407009125,
      "learning_rate": 4.228218830360001e-06,
      "loss": 0.001,
      "step": 108980
    },
    {
      "epoch": 1.783359240775587,
      "grad_norm": 0.07888317853212357,
      "learning_rate": 4.227278313027627e-06,
      "loss": 0.0023,
      "step": 108990
    },
    {
      "epoch": 1.783522866726663,
      "grad_norm": 0.07661629468202591,
      "learning_rate": 4.226337823704558e-06,
      "loss": 0.002,
      "step": 109000
    },
    {
      "epoch": 1.7836864926777387,
      "grad_norm": 0.04742414504289627,
      "learning_rate": 4.225397362424883e-06,
      "loss": 0.0008,
      "step": 109010
    },
    {
      "epoch": 1.7838501186288145,
      "grad_norm": 0.15801192820072174,
      "learning_rate": 4.2244569292226925e-06,
      "loss": 0.0053,
      "step": 109020
    },
    {
      "epoch": 1.7840137445798905,
      "grad_norm": 0.03451507166028023,
      "learning_rate": 4.223516524132073e-06,
      "loss": 0.0011,
      "step": 109030
    },
    {
      "epoch": 1.7841773705309663,
      "grad_norm": 0.20218102633953094,
      "learning_rate": 4.2225761471871155e-06,
      "loss": 0.0022,
      "step": 109040
    },
    {
      "epoch": 1.784340996482042,
      "grad_norm": 0.29750487208366394,
      "learning_rate": 4.221635798421904e-06,
      "loss": 0.0019,
      "step": 109050
    },
    {
      "epoch": 1.784504622433118,
      "grad_norm": 0.21108809113502502,
      "learning_rate": 4.2206954778705225e-06,
      "loss": 0.0013,
      "step": 109060
    },
    {
      "epoch": 1.7846682483841936,
      "grad_norm": 0.04725620523095131,
      "learning_rate": 4.219755185567057e-06,
      "loss": 0.0007,
      "step": 109070
    },
    {
      "epoch": 1.7848318743352696,
      "grad_norm": 0.10169366747140884,
      "learning_rate": 4.218814921545591e-06,
      "loss": 0.0019,
      "step": 109080
    },
    {
      "epoch": 1.7849955002863456,
      "grad_norm": 0.13469062745571136,
      "learning_rate": 4.217874685840207e-06,
      "loss": 0.0015,
      "step": 109090
    },
    {
      "epoch": 1.7851591262374211,
      "grad_norm": 0.10235961526632309,
      "learning_rate": 4.216934478484985e-06,
      "loss": 0.0012,
      "step": 109100
    },
    {
      "epoch": 1.7853227521884971,
      "grad_norm": 0.30775031447410583,
      "learning_rate": 4.215994299514007e-06,
      "loss": 0.0025,
      "step": 109110
    },
    {
      "epoch": 1.785486378139573,
      "grad_norm": 0.13852088153362274,
      "learning_rate": 4.215054148961351e-06,
      "loss": 0.0023,
      "step": 109120
    },
    {
      "epoch": 1.7856500040906487,
      "grad_norm": 0.11138801276683807,
      "learning_rate": 4.214114026861096e-06,
      "loss": 0.0011,
      "step": 109130
    },
    {
      "epoch": 1.7858136300417247,
      "grad_norm": 0.23532314598560333,
      "learning_rate": 4.213173933247318e-06,
      "loss": 0.0019,
      "step": 109140
    },
    {
      "epoch": 1.7859772559928004,
      "grad_norm": 0.03958771750330925,
      "learning_rate": 4.2122338681540945e-06,
      "loss": 0.0013,
      "step": 109150
    },
    {
      "epoch": 1.7861408819438762,
      "grad_norm": 0.06906077265739441,
      "learning_rate": 4.2112938316155e-06,
      "loss": 0.0016,
      "step": 109160
    },
    {
      "epoch": 1.7863045078949522,
      "grad_norm": 0.0847359374165535,
      "learning_rate": 4.210353823665608e-06,
      "loss": 0.0016,
      "step": 109170
    },
    {
      "epoch": 1.786468133846028,
      "grad_norm": 0.04323013499379158,
      "learning_rate": 4.209413844338492e-06,
      "loss": 0.0014,
      "step": 109180
    },
    {
      "epoch": 1.7866317597971038,
      "grad_norm": 0.2102014422416687,
      "learning_rate": 4.208473893668224e-06,
      "loss": 0.0018,
      "step": 109190
    },
    {
      "epoch": 1.7867953857481798,
      "grad_norm": 0.09043881297111511,
      "learning_rate": 4.207533971688874e-06,
      "loss": 0.0015,
      "step": 109200
    },
    {
      "epoch": 1.7869590116992555,
      "grad_norm": 0.04318220913410187,
      "learning_rate": 4.206594078434514e-06,
      "loss": 0.0007,
      "step": 109210
    },
    {
      "epoch": 1.7871226376503313,
      "grad_norm": 0.024060461670160294,
      "learning_rate": 4.2056542139392105e-06,
      "loss": 0.0031,
      "step": 109220
    },
    {
      "epoch": 1.7872862636014073,
      "grad_norm": 0.03796510398387909,
      "learning_rate": 4.2047143782370346e-06,
      "loss": 0.0016,
      "step": 109230
    },
    {
      "epoch": 1.787449889552483,
      "grad_norm": 0.03802335262298584,
      "learning_rate": 4.203774571362052e-06,
      "loss": 0.0023,
      "step": 109240
    },
    {
      "epoch": 1.7876135155035588,
      "grad_norm": 0.026080958545207977,
      "learning_rate": 4.202834793348326e-06,
      "loss": 0.0012,
      "step": 109250
    },
    {
      "epoch": 1.7877771414546348,
      "grad_norm": 0.2294594943523407,
      "learning_rate": 4.201895044229921e-06,
      "loss": 0.0007,
      "step": 109260
    },
    {
      "epoch": 1.7879407674057104,
      "grad_norm": 0.17704853415489197,
      "learning_rate": 4.200955324040904e-06,
      "loss": 0.003,
      "step": 109270
    },
    {
      "epoch": 1.7881043933567864,
      "grad_norm": 0.2035708725452423,
      "learning_rate": 4.2000156328153365e-06,
      "loss": 0.001,
      "step": 109280
    },
    {
      "epoch": 1.7882680193078624,
      "grad_norm": 0.1668217033147812,
      "learning_rate": 4.1990759705872785e-06,
      "loss": 0.0032,
      "step": 109290
    },
    {
      "epoch": 1.788431645258938,
      "grad_norm": 0.1366310715675354,
      "learning_rate": 4.198136337390793e-06,
      "loss": 0.0013,
      "step": 109300
    },
    {
      "epoch": 1.788595271210014,
      "grad_norm": 0.010214103385806084,
      "learning_rate": 4.197196733259937e-06,
      "loss": 0.0021,
      "step": 109310
    },
    {
      "epoch": 1.7887588971610897,
      "grad_norm": 0.02637605555355549,
      "learning_rate": 4.19625715822877e-06,
      "loss": 0.0017,
      "step": 109320
    },
    {
      "epoch": 1.7889225231121655,
      "grad_norm": 0.05430176481604576,
      "learning_rate": 4.195317612331349e-06,
      "loss": 0.001,
      "step": 109330
    },
    {
      "epoch": 1.7890861490632415,
      "grad_norm": 0.17617493867874146,
      "learning_rate": 4.194378095601732e-06,
      "loss": 0.0021,
      "step": 109340
    },
    {
      "epoch": 1.7892497750143173,
      "grad_norm": 0.0738600417971611,
      "learning_rate": 4.193438608073972e-06,
      "loss": 0.0013,
      "step": 109350
    },
    {
      "epoch": 1.789413400965393,
      "grad_norm": 0.054634593427181244,
      "learning_rate": 4.192499149782123e-06,
      "loss": 0.0014,
      "step": 109360
    },
    {
      "epoch": 1.789577026916469,
      "grad_norm": 0.16539928317070007,
      "learning_rate": 4.191559720760241e-06,
      "loss": 0.0013,
      "step": 109370
    },
    {
      "epoch": 1.7897406528675448,
      "grad_norm": 0.1007479578256607,
      "learning_rate": 4.190620321042374e-06,
      "loss": 0.0014,
      "step": 109380
    },
    {
      "epoch": 1.7899042788186206,
      "grad_norm": 0.07088002562522888,
      "learning_rate": 4.189680950662576e-06,
      "loss": 0.0014,
      "step": 109390
    },
    {
      "epoch": 1.7900679047696966,
      "grad_norm": 0.13595464825630188,
      "learning_rate": 4.188741609654896e-06,
      "loss": 0.0021,
      "step": 109400
    },
    {
      "epoch": 1.7902315307207723,
      "grad_norm": 0.3133603632450104,
      "learning_rate": 4.187802298053383e-06,
      "loss": 0.0022,
      "step": 109410
    },
    {
      "epoch": 1.7903951566718481,
      "grad_norm": 0.012337069027125835,
      "learning_rate": 4.186863015892083e-06,
      "loss": 0.0017,
      "step": 109420
    },
    {
      "epoch": 1.7905587826229241,
      "grad_norm": 0.1178058385848999,
      "learning_rate": 4.185923763205046e-06,
      "loss": 0.0024,
      "step": 109430
    },
    {
      "epoch": 1.790722408574,
      "grad_norm": 0.07045537233352661,
      "learning_rate": 4.184984540026318e-06,
      "loss": 0.0015,
      "step": 109440
    },
    {
      "epoch": 1.7908860345250757,
      "grad_norm": 0.11354552209377289,
      "learning_rate": 4.184045346389939e-06,
      "loss": 0.0014,
      "step": 109450
    },
    {
      "epoch": 1.7910496604761517,
      "grad_norm": 0.008082418702542782,
      "learning_rate": 4.183106182329955e-06,
      "loss": 0.0014,
      "step": 109460
    },
    {
      "epoch": 1.7912132864272272,
      "grad_norm": 0.07397264987230301,
      "learning_rate": 4.182167047880409e-06,
      "loss": 0.0011,
      "step": 109470
    },
    {
      "epoch": 1.7913769123783032,
      "grad_norm": 0.17240633070468903,
      "learning_rate": 4.181227943075341e-06,
      "loss": 0.0031,
      "step": 109480
    },
    {
      "epoch": 1.791540538329379,
      "grad_norm": 0.042109161615371704,
      "learning_rate": 4.180288867948793e-06,
      "loss": 0.0009,
      "step": 109490
    },
    {
      "epoch": 1.7917041642804548,
      "grad_norm": 0.2711697518825531,
      "learning_rate": 4.1793498225348035e-06,
      "loss": 0.0025,
      "step": 109500
    },
    {
      "epoch": 1.7918677902315308,
      "grad_norm": 0.10704971104860306,
      "learning_rate": 4.178410806867411e-06,
      "loss": 0.0016,
      "step": 109510
    },
    {
      "epoch": 1.7920314161826065,
      "grad_norm": 0.0459461510181427,
      "learning_rate": 4.177471820980652e-06,
      "loss": 0.0019,
      "step": 109520
    },
    {
      "epoch": 1.7921950421336823,
      "grad_norm": 0.22187310457229614,
      "learning_rate": 4.176532864908561e-06,
      "loss": 0.0017,
      "step": 109530
    },
    {
      "epoch": 1.7923586680847583,
      "grad_norm": 0.03752073645591736,
      "learning_rate": 4.175593938685178e-06,
      "loss": 0.0026,
      "step": 109540
    },
    {
      "epoch": 1.792522294035834,
      "grad_norm": 0.025677448138594627,
      "learning_rate": 4.17465504234453e-06,
      "loss": 0.0022,
      "step": 109550
    },
    {
      "epoch": 1.7926859199869098,
      "grad_norm": 0.06776012480258942,
      "learning_rate": 4.173716175920655e-06,
      "loss": 0.0047,
      "step": 109560
    },
    {
      "epoch": 1.7928495459379858,
      "grad_norm": 0.04825940728187561,
      "learning_rate": 4.172777339447581e-06,
      "loss": 0.0009,
      "step": 109570
    },
    {
      "epoch": 1.7930131718890616,
      "grad_norm": 0.09853023290634155,
      "learning_rate": 4.171838532959341e-06,
      "loss": 0.0017,
      "step": 109580
    },
    {
      "epoch": 1.7931767978401374,
      "grad_norm": 0.028595373034477234,
      "learning_rate": 4.170899756489964e-06,
      "loss": 0.001,
      "step": 109590
    },
    {
      "epoch": 1.7933404237912134,
      "grad_norm": 0.34742894768714905,
      "learning_rate": 4.169961010073478e-06,
      "loss": 0.0034,
      "step": 109600
    },
    {
      "epoch": 1.7935040497422892,
      "grad_norm": 0.045857951045036316,
      "learning_rate": 4.169022293743911e-06,
      "loss": 0.0022,
      "step": 109610
    },
    {
      "epoch": 1.793667675693365,
      "grad_norm": 0.14168353378772736,
      "learning_rate": 4.168083607535287e-06,
      "loss": 0.0014,
      "step": 109620
    },
    {
      "epoch": 1.793831301644441,
      "grad_norm": 0.11010447144508362,
      "learning_rate": 4.167144951481634e-06,
      "loss": 0.0014,
      "step": 109630
    },
    {
      "epoch": 1.7939949275955165,
      "grad_norm": 0.3340127468109131,
      "learning_rate": 4.166206325616976e-06,
      "loss": 0.0014,
      "step": 109640
    },
    {
      "epoch": 1.7941585535465925,
      "grad_norm": 0.01691707782447338,
      "learning_rate": 4.165267729975333e-06,
      "loss": 0.0012,
      "step": 109650
    },
    {
      "epoch": 1.7943221794976685,
      "grad_norm": 0.015904491767287254,
      "learning_rate": 4.164329164590729e-06,
      "loss": 0.0045,
      "step": 109660
    },
    {
      "epoch": 1.794485805448744,
      "grad_norm": 0.04675578698515892,
      "learning_rate": 4.163390629497184e-06,
      "loss": 0.002,
      "step": 109670
    },
    {
      "epoch": 1.79464943139982,
      "grad_norm": 0.04986276105046272,
      "learning_rate": 4.1624521247287185e-06,
      "loss": 0.0014,
      "step": 109680
    },
    {
      "epoch": 1.7948130573508958,
      "grad_norm": 0.045936889946460724,
      "learning_rate": 4.16151365031935e-06,
      "loss": 0.0013,
      "step": 109690
    },
    {
      "epoch": 1.7949766833019716,
      "grad_norm": 0.1072971522808075,
      "learning_rate": 4.160575206303095e-06,
      "loss": 0.0016,
      "step": 109700
    },
    {
      "epoch": 1.7951403092530476,
      "grad_norm": 0.23212066292762756,
      "learning_rate": 4.159636792713973e-06,
      "loss": 0.0017,
      "step": 109710
    },
    {
      "epoch": 1.7953039352041233,
      "grad_norm": 0.03734473139047623,
      "learning_rate": 4.158698409585997e-06,
      "loss": 0.0011,
      "step": 109720
    },
    {
      "epoch": 1.7954675611551991,
      "grad_norm": 0.032329753041267395,
      "learning_rate": 4.1577600569531825e-06,
      "loss": 0.0021,
      "step": 109730
    },
    {
      "epoch": 1.7956311871062751,
      "grad_norm": 0.07856018096208572,
      "learning_rate": 4.1568217348495436e-06,
      "loss": 0.0012,
      "step": 109740
    },
    {
      "epoch": 1.795794813057351,
      "grad_norm": 0.06815605610609055,
      "learning_rate": 4.155883443309087e-06,
      "loss": 0.0012,
      "step": 109750
    },
    {
      "epoch": 1.7959584390084267,
      "grad_norm": 0.4105449616909027,
      "learning_rate": 4.154945182365826e-06,
      "loss": 0.0008,
      "step": 109760
    },
    {
      "epoch": 1.7961220649595027,
      "grad_norm": 0.04402492195367813,
      "learning_rate": 4.154006952053773e-06,
      "loss": 0.0011,
      "step": 109770
    },
    {
      "epoch": 1.7962856909105784,
      "grad_norm": 0.0037064552307128906,
      "learning_rate": 4.153068752406934e-06,
      "loss": 0.0014,
      "step": 109780
    },
    {
      "epoch": 1.7964493168616542,
      "grad_norm": 0.08393971621990204,
      "learning_rate": 4.152130583459318e-06,
      "loss": 0.0013,
      "step": 109790
    },
    {
      "epoch": 1.7966129428127302,
      "grad_norm": 0.032570336014032364,
      "learning_rate": 4.15119244524493e-06,
      "loss": 0.0013,
      "step": 109800
    },
    {
      "epoch": 1.796776568763806,
      "grad_norm": 0.06312573701143265,
      "learning_rate": 4.150254337797777e-06,
      "loss": 0.0015,
      "step": 109810
    },
    {
      "epoch": 1.7969401947148818,
      "grad_norm": 0.08007284253835678,
      "learning_rate": 4.14931626115186e-06,
      "loss": 0.0031,
      "step": 109820
    },
    {
      "epoch": 1.7971038206659578,
      "grad_norm": 0.255429744720459,
      "learning_rate": 4.148378215341185e-06,
      "loss": 0.0042,
      "step": 109830
    },
    {
      "epoch": 1.7972674466170333,
      "grad_norm": 0.07173515111207962,
      "learning_rate": 4.147440200399753e-06,
      "loss": 0.0025,
      "step": 109840
    },
    {
      "epoch": 1.7974310725681093,
      "grad_norm": 0.05971740931272507,
      "learning_rate": 4.146502216361566e-06,
      "loss": 0.0018,
      "step": 109850
    },
    {
      "epoch": 1.7975946985191853,
      "grad_norm": 0.07941849529743195,
      "learning_rate": 4.1455642632606196e-06,
      "loss": 0.0013,
      "step": 109860
    },
    {
      "epoch": 1.7977583244702608,
      "grad_norm": 0.03401295840740204,
      "learning_rate": 4.144626341130916e-06,
      "loss": 0.0006,
      "step": 109870
    },
    {
      "epoch": 1.7979219504213368,
      "grad_norm": 0.07793786376714706,
      "learning_rate": 4.143688450006452e-06,
      "loss": 0.0012,
      "step": 109880
    },
    {
      "epoch": 1.7980855763724126,
      "grad_norm": 0.04069236293435097,
      "learning_rate": 4.142750589921224e-06,
      "loss": 0.0011,
      "step": 109890
    },
    {
      "epoch": 1.7982492023234884,
      "grad_norm": 0.13962702453136444,
      "learning_rate": 4.141812760909226e-06,
      "loss": 0.0026,
      "step": 109900
    },
    {
      "epoch": 1.7984128282745644,
      "grad_norm": 0.13767190277576447,
      "learning_rate": 4.140874963004453e-06,
      "loss": 0.0012,
      "step": 109910
    },
    {
      "epoch": 1.7985764542256402,
      "grad_norm": 0.038861233741045,
      "learning_rate": 4.139937196240897e-06,
      "loss": 0.0015,
      "step": 109920
    },
    {
      "epoch": 1.798740080176716,
      "grad_norm": 0.04253111407160759,
      "learning_rate": 4.138999460652551e-06,
      "loss": 0.0013,
      "step": 109930
    },
    {
      "epoch": 1.798903706127792,
      "grad_norm": 0.12363880127668381,
      "learning_rate": 4.138061756273408e-06,
      "loss": 0.0018,
      "step": 109940
    },
    {
      "epoch": 1.7990673320788677,
      "grad_norm": 0.13247255980968475,
      "learning_rate": 4.137124083137452e-06,
      "loss": 0.0017,
      "step": 109950
    },
    {
      "epoch": 1.7992309580299435,
      "grad_norm": 0.12151532620191574,
      "learning_rate": 4.136186441278674e-06,
      "loss": 0.0017,
      "step": 109960
    },
    {
      "epoch": 1.7993945839810195,
      "grad_norm": 0.07101260870695114,
      "learning_rate": 4.135248830731062e-06,
      "loss": 0.0015,
      "step": 109970
    },
    {
      "epoch": 1.7995582099320953,
      "grad_norm": 0.0801393985748291,
      "learning_rate": 4.1343112515286e-06,
      "loss": 0.0015,
      "step": 109980
    },
    {
      "epoch": 1.799721835883171,
      "grad_norm": 0.09624263644218445,
      "learning_rate": 4.133373703705275e-06,
      "loss": 0.0019,
      "step": 109990
    },
    {
      "epoch": 1.799885461834247,
      "grad_norm": 0.174933522939682,
      "learning_rate": 4.132436187295072e-06,
      "loss": 0.0007,
      "step": 110000
    },
    {
      "epoch": 1.8000490877853228,
      "grad_norm": 0.13256537914276123,
      "learning_rate": 4.131498702331972e-06,
      "loss": 0.0011,
      "step": 110010
    },
    {
      "epoch": 1.8002127137363986,
      "grad_norm": 0.24644337594509125,
      "learning_rate": 4.130561248849954e-06,
      "loss": 0.0022,
      "step": 110020
    },
    {
      "epoch": 1.8003763396874746,
      "grad_norm": 0.0620858408510685,
      "learning_rate": 4.129623826883004e-06,
      "loss": 0.001,
      "step": 110030
    },
    {
      "epoch": 1.8005399656385501,
      "grad_norm": 0.04734351485967636,
      "learning_rate": 4.128686436465098e-06,
      "loss": 0.0014,
      "step": 110040
    },
    {
      "epoch": 1.8007035915896261,
      "grad_norm": 0.09699749946594238,
      "learning_rate": 4.127749077630214e-06,
      "loss": 0.003,
      "step": 110050
    },
    {
      "epoch": 1.8008672175407021,
      "grad_norm": 0.16596907377243042,
      "learning_rate": 4.126811750412329e-06,
      "loss": 0.0037,
      "step": 110060
    },
    {
      "epoch": 1.8010308434917777,
      "grad_norm": 0.03903839737176895,
      "learning_rate": 4.125874454845421e-06,
      "loss": 0.002,
      "step": 110070
    },
    {
      "epoch": 1.8011944694428537,
      "grad_norm": 0.06339141726493835,
      "learning_rate": 4.124937190963461e-06,
      "loss": 0.0009,
      "step": 110080
    },
    {
      "epoch": 1.8013580953939294,
      "grad_norm": 0.12396344542503357,
      "learning_rate": 4.123999958800426e-06,
      "loss": 0.0014,
      "step": 110090
    },
    {
      "epoch": 1.8015217213450052,
      "grad_norm": 0.0903758704662323,
      "learning_rate": 4.123062758390286e-06,
      "loss": 0.001,
      "step": 110100
    },
    {
      "epoch": 1.8016853472960812,
      "grad_norm": 0.1341772973537445,
      "learning_rate": 4.122125589767014e-06,
      "loss": 0.0017,
      "step": 110110
    },
    {
      "epoch": 1.801848973247157,
      "grad_norm": 0.005815631244331598,
      "learning_rate": 4.12118845296458e-06,
      "loss": 0.0021,
      "step": 110120
    },
    {
      "epoch": 1.8020125991982328,
      "grad_norm": 0.36865493655204773,
      "learning_rate": 4.120251348016951e-06,
      "loss": 0.0016,
      "step": 110130
    },
    {
      "epoch": 1.8021762251493088,
      "grad_norm": 0.021256666630506516,
      "learning_rate": 4.119314274958099e-06,
      "loss": 0.0015,
      "step": 110140
    },
    {
      "epoch": 1.8023398511003845,
      "grad_norm": 0.03181765228509903,
      "learning_rate": 4.1183772338219855e-06,
      "loss": 0.001,
      "step": 110150
    },
    {
      "epoch": 1.8025034770514603,
      "grad_norm": 0.11860894411802292,
      "learning_rate": 4.117440224642579e-06,
      "loss": 0.0071,
      "step": 110160
    },
    {
      "epoch": 1.8026671030025363,
      "grad_norm": 0.08703216910362244,
      "learning_rate": 4.116503247453843e-06,
      "loss": 0.0011,
      "step": 110170
    },
    {
      "epoch": 1.802830728953612,
      "grad_norm": 0.03114217333495617,
      "learning_rate": 4.115566302289741e-06,
      "loss": 0.001,
      "step": 110180
    },
    {
      "epoch": 1.8029943549046878,
      "grad_norm": 0.06639087945222855,
      "learning_rate": 4.114629389184234e-06,
      "loss": 0.0012,
      "step": 110190
    },
    {
      "epoch": 1.8031579808557638,
      "grad_norm": 0.03908311948180199,
      "learning_rate": 4.113692508171283e-06,
      "loss": 0.0019,
      "step": 110200
    },
    {
      "epoch": 1.8033216068068396,
      "grad_norm": 0.053245507180690765,
      "learning_rate": 4.11275565928485e-06,
      "loss": 0.0012,
      "step": 110210
    },
    {
      "epoch": 1.8034852327579154,
      "grad_norm": 0.254879891872406,
      "learning_rate": 4.1118188425588915e-06,
      "loss": 0.0012,
      "step": 110220
    },
    {
      "epoch": 1.8036488587089914,
      "grad_norm": 0.09775219857692719,
      "learning_rate": 4.110882058027366e-06,
      "loss": 0.0021,
      "step": 110230
    },
    {
      "epoch": 1.803812484660067,
      "grad_norm": 0.014985921792685986,
      "learning_rate": 4.109945305724231e-06,
      "loss": 0.002,
      "step": 110240
    },
    {
      "epoch": 1.803976110611143,
      "grad_norm": 0.02410202845931053,
      "learning_rate": 4.109008585683435e-06,
      "loss": 0.0021,
      "step": 110250
    },
    {
      "epoch": 1.804139736562219,
      "grad_norm": 0.04714832827448845,
      "learning_rate": 4.108071897938938e-06,
      "loss": 0.0011,
      "step": 110260
    },
    {
      "epoch": 1.8043033625132945,
      "grad_norm": 0.03473523259162903,
      "learning_rate": 4.107135242524692e-06,
      "loss": 0.0011,
      "step": 110270
    },
    {
      "epoch": 1.8044669884643705,
      "grad_norm": 0.057406578212976456,
      "learning_rate": 4.106198619474646e-06,
      "loss": 0.0016,
      "step": 110280
    },
    {
      "epoch": 1.8046306144154463,
      "grad_norm": 0.12736418843269348,
      "learning_rate": 4.105262028822753e-06,
      "loss": 0.0026,
      "step": 110290
    },
    {
      "epoch": 1.804794240366522,
      "grad_norm": 0.04663657024502754,
      "learning_rate": 4.1043254706029605e-06,
      "loss": 0.0009,
      "step": 110300
    },
    {
      "epoch": 1.804957866317598,
      "grad_norm": 0.020401667803525925,
      "learning_rate": 4.103388944849217e-06,
      "loss": 0.0022,
      "step": 110310
    },
    {
      "epoch": 1.8051214922686738,
      "grad_norm": 0.02002941444516182,
      "learning_rate": 4.10245245159547e-06,
      "loss": 0.003,
      "step": 110320
    },
    {
      "epoch": 1.8052851182197496,
      "grad_norm": 0.08543750643730164,
      "learning_rate": 4.101515990875663e-06,
      "loss": 0.0012,
      "step": 110330
    },
    {
      "epoch": 1.8054487441708256,
      "grad_norm": 0.05389794334769249,
      "learning_rate": 4.100579562723744e-06,
      "loss": 0.001,
      "step": 110340
    },
    {
      "epoch": 1.8056123701219013,
      "grad_norm": 0.07623659819364548,
      "learning_rate": 4.099643167173653e-06,
      "loss": 0.0008,
      "step": 110350
    },
    {
      "epoch": 1.8057759960729771,
      "grad_norm": 0.11723193526268005,
      "learning_rate": 4.0987068042593335e-06,
      "loss": 0.001,
      "step": 110360
    },
    {
      "epoch": 1.8059396220240531,
      "grad_norm": 0.14318618178367615,
      "learning_rate": 4.097770474014725e-06,
      "loss": 0.0013,
      "step": 110370
    },
    {
      "epoch": 1.806103247975129,
      "grad_norm": 0.02823631465435028,
      "learning_rate": 4.09683417647377e-06,
      "loss": 0.0012,
      "step": 110380
    },
    {
      "epoch": 1.8062668739262047,
      "grad_norm": 0.13594180345535278,
      "learning_rate": 4.095897911670404e-06,
      "loss": 0.0025,
      "step": 110390
    },
    {
      "epoch": 1.8064304998772807,
      "grad_norm": 0.08361697196960449,
      "learning_rate": 4.0949616796385665e-06,
      "loss": 0.0021,
      "step": 110400
    },
    {
      "epoch": 1.8065941258283562,
      "grad_norm": 0.03334399312734604,
      "learning_rate": 4.094025480412191e-06,
      "loss": 0.0014,
      "step": 110410
    },
    {
      "epoch": 1.8067577517794322,
      "grad_norm": 0.047940444201231,
      "learning_rate": 4.093089314025216e-06,
      "loss": 0.0007,
      "step": 110420
    },
    {
      "epoch": 1.8069213777305082,
      "grad_norm": 0.10455339401960373,
      "learning_rate": 4.0921531805115736e-06,
      "loss": 0.0018,
      "step": 110430
    },
    {
      "epoch": 1.8070850036815838,
      "grad_norm": 0.03962060064077377,
      "learning_rate": 4.091217079905198e-06,
      "loss": 0.0021,
      "step": 110440
    },
    {
      "epoch": 1.8072486296326598,
      "grad_norm": 0.013049505650997162,
      "learning_rate": 4.090281012240018e-06,
      "loss": 0.0018,
      "step": 110450
    },
    {
      "epoch": 1.8074122555837355,
      "grad_norm": 0.29218167066574097,
      "learning_rate": 4.089344977549965e-06,
      "loss": 0.0019,
      "step": 110460
    },
    {
      "epoch": 1.8075758815348113,
      "grad_norm": 0.046039290726184845,
      "learning_rate": 4.088408975868966e-06,
      "loss": 0.0009,
      "step": 110470
    },
    {
      "epoch": 1.8077395074858873,
      "grad_norm": 0.029266629368066788,
      "learning_rate": 4.087473007230952e-06,
      "loss": 0.001,
      "step": 110480
    },
    {
      "epoch": 1.807903133436963,
      "grad_norm": 0.06409554928541183,
      "learning_rate": 4.086537071669848e-06,
      "loss": 0.0017,
      "step": 110490
    },
    {
      "epoch": 1.8080667593880388,
      "grad_norm": 0.0726979449391365,
      "learning_rate": 4.08560116921958e-06,
      "loss": 0.0015,
      "step": 110500
    },
    {
      "epoch": 1.8082303853391148,
      "grad_norm": 0.048297226428985596,
      "learning_rate": 4.084665299914073e-06,
      "loss": 0.0016,
      "step": 110510
    },
    {
      "epoch": 1.8083940112901906,
      "grad_norm": 0.04781966283917427,
      "learning_rate": 4.083729463787247e-06,
      "loss": 0.0007,
      "step": 110520
    },
    {
      "epoch": 1.8085576372412664,
      "grad_norm": 0.0702255517244339,
      "learning_rate": 4.082793660873027e-06,
      "loss": 0.0011,
      "step": 110530
    },
    {
      "epoch": 1.8087212631923424,
      "grad_norm": 0.04958431050181389,
      "learning_rate": 4.081857891205333e-06,
      "loss": 0.0017,
      "step": 110540
    },
    {
      "epoch": 1.8088848891434182,
      "grad_norm": 0.04033922776579857,
      "learning_rate": 4.080922154818082e-06,
      "loss": 0.0009,
      "step": 110550
    },
    {
      "epoch": 1.809048515094494,
      "grad_norm": 0.15074045956134796,
      "learning_rate": 4.079986451745195e-06,
      "loss": 0.0015,
      "step": 110560
    },
    {
      "epoch": 1.80921214104557,
      "grad_norm": 0.13608914613723755,
      "learning_rate": 4.079050782020588e-06,
      "loss": 0.0022,
      "step": 110570
    },
    {
      "epoch": 1.8093757669966457,
      "grad_norm": 0.08144427835941315,
      "learning_rate": 4.078115145678176e-06,
      "loss": 0.0014,
      "step": 110580
    },
    {
      "epoch": 1.8095393929477215,
      "grad_norm": 0.10574265569448471,
      "learning_rate": 4.077179542751875e-06,
      "loss": 0.0026,
      "step": 110590
    },
    {
      "epoch": 1.8097030188987975,
      "grad_norm": 0.09092884510755539,
      "learning_rate": 4.076243973275597e-06,
      "loss": 0.0011,
      "step": 110600
    },
    {
      "epoch": 1.809866644849873,
      "grad_norm": 0.013703079894185066,
      "learning_rate": 4.075308437283254e-06,
      "loss": 0.0021,
      "step": 110610
    },
    {
      "epoch": 1.810030270800949,
      "grad_norm": 0.09637783467769623,
      "learning_rate": 4.074372934808759e-06,
      "loss": 0.0015,
      "step": 110620
    },
    {
      "epoch": 1.810193896752025,
      "grad_norm": 0.1178259551525116,
      "learning_rate": 4.073437465886019e-06,
      "loss": 0.0015,
      "step": 110630
    },
    {
      "epoch": 1.8103575227031006,
      "grad_norm": 0.05465967208147049,
      "learning_rate": 4.072502030548947e-06,
      "loss": 0.0012,
      "step": 110640
    },
    {
      "epoch": 1.8105211486541766,
      "grad_norm": 0.05660577118396759,
      "learning_rate": 4.071566628831446e-06,
      "loss": 0.0012,
      "step": 110650
    },
    {
      "epoch": 1.8106847746052523,
      "grad_norm": 0.005766091402620077,
      "learning_rate": 4.070631260767422e-06,
      "loss": 0.0015,
      "step": 110660
    },
    {
      "epoch": 1.8108484005563281,
      "grad_norm": 0.06887154281139374,
      "learning_rate": 4.069695926390781e-06,
      "loss": 0.0025,
      "step": 110670
    },
    {
      "epoch": 1.8110120265074041,
      "grad_norm": 0.07911459356546402,
      "learning_rate": 4.068760625735428e-06,
      "loss": 0.0019,
      "step": 110680
    },
    {
      "epoch": 1.81117565245848,
      "grad_norm": 0.0156877338886261,
      "learning_rate": 4.067825358835263e-06,
      "loss": 0.0016,
      "step": 110690
    },
    {
      "epoch": 1.8113392784095557,
      "grad_norm": 0.06569913774728775,
      "learning_rate": 4.066890125724188e-06,
      "loss": 0.0017,
      "step": 110700
    },
    {
      "epoch": 1.8115029043606317,
      "grad_norm": 0.05675230920314789,
      "learning_rate": 4.065954926436104e-06,
      "loss": 0.0024,
      "step": 110710
    },
    {
      "epoch": 1.8116665303117074,
      "grad_norm": 0.131346195936203,
      "learning_rate": 4.0650197610049094e-06,
      "loss": 0.0014,
      "step": 110720
    },
    {
      "epoch": 1.8118301562627832,
      "grad_norm": 0.06230497732758522,
      "learning_rate": 4.064084629464501e-06,
      "loss": 0.0013,
      "step": 110730
    },
    {
      "epoch": 1.8119937822138592,
      "grad_norm": 0.07833479344844818,
      "learning_rate": 4.063149531848777e-06,
      "loss": 0.0026,
      "step": 110740
    },
    {
      "epoch": 1.812157408164935,
      "grad_norm": 0.08271912485361099,
      "learning_rate": 4.062214468191629e-06,
      "loss": 0.0028,
      "step": 110750
    },
    {
      "epoch": 1.8123210341160108,
      "grad_norm": 0.0654798299074173,
      "learning_rate": 4.0612794385269525e-06,
      "loss": 0.003,
      "step": 110760
    },
    {
      "epoch": 1.8124846600670868,
      "grad_norm": 0.1459072083234787,
      "learning_rate": 4.06034444288864e-06,
      "loss": 0.001,
      "step": 110770
    },
    {
      "epoch": 1.8126482860181625,
      "grad_norm": 0.07602879405021667,
      "learning_rate": 4.059409481310584e-06,
      "loss": 0.0017,
      "step": 110780
    },
    {
      "epoch": 1.8128119119692383,
      "grad_norm": 0.05197446420788765,
      "learning_rate": 4.0584745538266725e-06,
      "loss": 0.0007,
      "step": 110790
    },
    {
      "epoch": 1.8129755379203143,
      "grad_norm": 0.02462603896856308,
      "learning_rate": 4.057539660470796e-06,
      "loss": 0.0013,
      "step": 110800
    },
    {
      "epoch": 1.8131391638713898,
      "grad_norm": 0.0999525636434555,
      "learning_rate": 4.056604801276842e-06,
      "loss": 0.0019,
      "step": 110810
    },
    {
      "epoch": 1.8133027898224658,
      "grad_norm": 0.05077390372753143,
      "learning_rate": 4.055669976278697e-06,
      "loss": 0.0017,
      "step": 110820
    },
    {
      "epoch": 1.8134664157735418,
      "grad_norm": 0.024046001955866814,
      "learning_rate": 4.0547351855102446e-06,
      "loss": 0.0011,
      "step": 110830
    },
    {
      "epoch": 1.8136300417246174,
      "grad_norm": 0.015562029555439949,
      "learning_rate": 4.053800429005372e-06,
      "loss": 0.0006,
      "step": 110840
    },
    {
      "epoch": 1.8137936676756934,
      "grad_norm": 0.05708250403404236,
      "learning_rate": 4.052865706797958e-06,
      "loss": 0.0009,
      "step": 110850
    },
    {
      "epoch": 1.8139572936267692,
      "grad_norm": 0.13831275701522827,
      "learning_rate": 4.051931018921886e-06,
      "loss": 0.0014,
      "step": 110860
    },
    {
      "epoch": 1.814120919577845,
      "grad_norm": 0.06047903373837471,
      "learning_rate": 4.0509963654110355e-06,
      "loss": 0.0028,
      "step": 110870
    },
    {
      "epoch": 1.814284545528921,
      "grad_norm": 0.1734655648469925,
      "learning_rate": 4.050061746299287e-06,
      "loss": 0.0016,
      "step": 110880
    },
    {
      "epoch": 1.8144481714799967,
      "grad_norm": 0.04613037407398224,
      "learning_rate": 4.049127161620515e-06,
      "loss": 0.0014,
      "step": 110890
    },
    {
      "epoch": 1.8146117974310725,
      "grad_norm": 0.018489491194486618,
      "learning_rate": 4.048192611408599e-06,
      "loss": 0.0007,
      "step": 110900
    },
    {
      "epoch": 1.8147754233821485,
      "grad_norm": 0.011923164129257202,
      "learning_rate": 4.0472580956974125e-06,
      "loss": 0.001,
      "step": 110910
    },
    {
      "epoch": 1.8149390493332243,
      "grad_norm": 0.11047337204217911,
      "learning_rate": 4.046323614520832e-06,
      "loss": 0.0021,
      "step": 110920
    },
    {
      "epoch": 1.8151026752843,
      "grad_norm": 0.12533338367938995,
      "learning_rate": 4.045389167912728e-06,
      "loss": 0.002,
      "step": 110930
    },
    {
      "epoch": 1.815266301235376,
      "grad_norm": 0.02617049589753151,
      "learning_rate": 4.0444547559069735e-06,
      "loss": 0.0009,
      "step": 110940
    },
    {
      "epoch": 1.8154299271864518,
      "grad_norm": 0.07797876745462418,
      "learning_rate": 4.043520378537435e-06,
      "loss": 0.0016,
      "step": 110950
    },
    {
      "epoch": 1.8155935531375276,
      "grad_norm": 0.06467451900243759,
      "learning_rate": 4.042586035837985e-06,
      "loss": 0.0009,
      "step": 110960
    },
    {
      "epoch": 1.8157571790886036,
      "grad_norm": 0.07913364470005035,
      "learning_rate": 4.041651727842489e-06,
      "loss": 0.002,
      "step": 110970
    },
    {
      "epoch": 1.8159208050396793,
      "grad_norm": 0.037827394902706146,
      "learning_rate": 4.040717454584816e-06,
      "loss": 0.0011,
      "step": 110980
    },
    {
      "epoch": 1.8160844309907551,
      "grad_norm": 0.06379867345094681,
      "learning_rate": 4.039783216098829e-06,
      "loss": 0.002,
      "step": 110990
    },
    {
      "epoch": 1.8162480569418311,
      "grad_norm": 0.061186887323856354,
      "learning_rate": 4.0388490124183925e-06,
      "loss": 0.0011,
      "step": 111000
    },
    {
      "epoch": 1.8164116828929067,
      "grad_norm": 0.11079651117324829,
      "learning_rate": 4.0379148435773695e-06,
      "loss": 0.0015,
      "step": 111010
    },
    {
      "epoch": 1.8165753088439827,
      "grad_norm": 0.013405365869402885,
      "learning_rate": 4.036980709609621e-06,
      "loss": 0.0012,
      "step": 111020
    },
    {
      "epoch": 1.8167389347950587,
      "grad_norm": 0.18405316770076752,
      "learning_rate": 4.036046610549007e-06,
      "loss": 0.0016,
      "step": 111030
    },
    {
      "epoch": 1.8169025607461342,
      "grad_norm": 0.09061921387910843,
      "learning_rate": 4.035112546429387e-06,
      "loss": 0.0019,
      "step": 111040
    },
    {
      "epoch": 1.8170661866972102,
      "grad_norm": 0.05544005334377289,
      "learning_rate": 4.03417851728462e-06,
      "loss": 0.0015,
      "step": 111050
    },
    {
      "epoch": 1.817229812648286,
      "grad_norm": 0.11394292116165161,
      "learning_rate": 4.033244523148558e-06,
      "loss": 0.0011,
      "step": 111060
    },
    {
      "epoch": 1.8173934385993618,
      "grad_norm": 0.27714091539382935,
      "learning_rate": 4.032310564055059e-06,
      "loss": 0.0017,
      "step": 111070
    },
    {
      "epoch": 1.8175570645504378,
      "grad_norm": 0.08347877860069275,
      "learning_rate": 4.031376640037977e-06,
      "loss": 0.0011,
      "step": 111080
    },
    {
      "epoch": 1.8177206905015135,
      "grad_norm": 0.009594296105206013,
      "learning_rate": 4.030442751131162e-06,
      "loss": 0.0015,
      "step": 111090
    },
    {
      "epoch": 1.8178843164525893,
      "grad_norm": 0.0414123460650444,
      "learning_rate": 4.029508897368468e-06,
      "loss": 0.0008,
      "step": 111100
    },
    {
      "epoch": 1.8180479424036653,
      "grad_norm": 0.21520625054836273,
      "learning_rate": 4.028575078783745e-06,
      "loss": 0.002,
      "step": 111110
    },
    {
      "epoch": 1.818211568354741,
      "grad_norm": 0.06374149024486542,
      "learning_rate": 4.027641295410839e-06,
      "loss": 0.0013,
      "step": 111120
    },
    {
      "epoch": 1.8183751943058168,
      "grad_norm": 0.042976684868335724,
      "learning_rate": 4.0267075472836e-06,
      "loss": 0.0008,
      "step": 111130
    },
    {
      "epoch": 1.8185388202568928,
      "grad_norm": 0.15336090326309204,
      "learning_rate": 4.0257738344358715e-06,
      "loss": 0.0015,
      "step": 111140
    },
    {
      "epoch": 1.8187024462079686,
      "grad_norm": 0.19176100194454193,
      "learning_rate": 4.0248401569015035e-06,
      "loss": 0.0009,
      "step": 111150
    },
    {
      "epoch": 1.8188660721590444,
      "grad_norm": 0.1569773256778717,
      "learning_rate": 4.0239065147143345e-06,
      "loss": 0.0009,
      "step": 111160
    },
    {
      "epoch": 1.8190296981101204,
      "grad_norm": 0.3488730490207672,
      "learning_rate": 4.022972907908208e-06,
      "loss": 0.001,
      "step": 111170
    },
    {
      "epoch": 1.8191933240611962,
      "grad_norm": 0.055071260780096054,
      "learning_rate": 4.022039336516965e-06,
      "loss": 0.003,
      "step": 111180
    },
    {
      "epoch": 1.819356950012272,
      "grad_norm": 0.07290855050086975,
      "learning_rate": 4.021105800574444e-06,
      "loss": 0.0022,
      "step": 111190
    },
    {
      "epoch": 1.819520575963348,
      "grad_norm": 0.03136163204908371,
      "learning_rate": 4.0201723001144875e-06,
      "loss": 0.0024,
      "step": 111200
    },
    {
      "epoch": 1.8196842019144235,
      "grad_norm": 0.10398129373788834,
      "learning_rate": 4.019238835170929e-06,
      "loss": 0.002,
      "step": 111210
    },
    {
      "epoch": 1.8198478278654995,
      "grad_norm": 0.19065433740615845,
      "learning_rate": 4.018305405777606e-06,
      "loss": 0.002,
      "step": 111220
    },
    {
      "epoch": 1.8200114538165755,
      "grad_norm": 0.01684756949543953,
      "learning_rate": 4.017372011968353e-06,
      "loss": 0.0014,
      "step": 111230
    },
    {
      "epoch": 1.820175079767651,
      "grad_norm": 0.09546512365341187,
      "learning_rate": 4.016438653777002e-06,
      "loss": 0.0012,
      "step": 111240
    },
    {
      "epoch": 1.820338705718727,
      "grad_norm": 0.08042707294225693,
      "learning_rate": 4.015505331237387e-06,
      "loss": 0.0011,
      "step": 111250
    },
    {
      "epoch": 1.8205023316698028,
      "grad_norm": 0.15508301556110382,
      "learning_rate": 4.014572044383337e-06,
      "loss": 0.0021,
      "step": 111260
    },
    {
      "epoch": 1.8206659576208786,
      "grad_norm": 0.1055852472782135,
      "learning_rate": 4.013638793248682e-06,
      "loss": 0.0019,
      "step": 111270
    },
    {
      "epoch": 1.8208295835719546,
      "grad_norm": 0.03518375754356384,
      "learning_rate": 4.01270557786725e-06,
      "loss": 0.0011,
      "step": 111280
    },
    {
      "epoch": 1.8209932095230303,
      "grad_norm": 0.025266781449317932,
      "learning_rate": 4.011772398272868e-06,
      "loss": 0.0011,
      "step": 111290
    },
    {
      "epoch": 1.8211568354741061,
      "grad_norm": 0.16470734775066376,
      "learning_rate": 4.010839254499362e-06,
      "loss": 0.0018,
      "step": 111300
    },
    {
      "epoch": 1.8213204614251821,
      "grad_norm": 0.08801986277103424,
      "learning_rate": 4.009906146580554e-06,
      "loss": 0.0019,
      "step": 111310
    },
    {
      "epoch": 1.8214840873762579,
      "grad_norm": 0.06811599433422089,
      "learning_rate": 4.0089730745502695e-06,
      "loss": 0.0021,
      "step": 111320
    },
    {
      "epoch": 1.8216477133273337,
      "grad_norm": 0.1471439003944397,
      "learning_rate": 4.008040038442329e-06,
      "loss": 0.0018,
      "step": 111330
    },
    {
      "epoch": 1.8218113392784097,
      "grad_norm": 0.05991533026099205,
      "learning_rate": 4.007107038290553e-06,
      "loss": 0.0019,
      "step": 111340
    },
    {
      "epoch": 1.8219749652294854,
      "grad_norm": 0.04979366809129715,
      "learning_rate": 4.006174074128762e-06,
      "loss": 0.0021,
      "step": 111350
    },
    {
      "epoch": 1.8221385911805612,
      "grad_norm": 0.3377133905887604,
      "learning_rate": 4.005241145990771e-06,
      "loss": 0.0015,
      "step": 111360
    },
    {
      "epoch": 1.8223022171316372,
      "grad_norm": 0.08466890454292297,
      "learning_rate": 4.004308253910397e-06,
      "loss": 0.0007,
      "step": 111370
    },
    {
      "epoch": 1.8224658430827128,
      "grad_norm": 0.028911596164107323,
      "learning_rate": 4.003375397921457e-06,
      "loss": 0.0025,
      "step": 111380
    },
    {
      "epoch": 1.8226294690337888,
      "grad_norm": 0.12721887230873108,
      "learning_rate": 4.002442578057763e-06,
      "loss": 0.002,
      "step": 111390
    },
    {
      "epoch": 1.8227930949848647,
      "grad_norm": 0.0052746557630598545,
      "learning_rate": 4.001509794353127e-06,
      "loss": 0.0027,
      "step": 111400
    },
    {
      "epoch": 1.8229567209359403,
      "grad_norm": 0.10063627362251282,
      "learning_rate": 4.000577046841361e-06,
      "loss": 0.0006,
      "step": 111410
    },
    {
      "epoch": 1.8231203468870163,
      "grad_norm": 0.21845905482769012,
      "learning_rate": 3.9996443355562755e-06,
      "loss": 0.0015,
      "step": 111420
    },
    {
      "epoch": 1.823283972838092,
      "grad_norm": 0.16975466907024384,
      "learning_rate": 3.998711660531679e-06,
      "loss": 0.0019,
      "step": 111430
    },
    {
      "epoch": 1.8234475987891678,
      "grad_norm": 0.03261583298444748,
      "learning_rate": 3.997779021801378e-06,
      "loss": 0.0018,
      "step": 111440
    },
    {
      "epoch": 1.8236112247402438,
      "grad_norm": 0.049272313714027405,
      "learning_rate": 3.996846419399181e-06,
      "loss": 0.0015,
      "step": 111450
    },
    {
      "epoch": 1.8237748506913196,
      "grad_norm": 0.02194603905081749,
      "learning_rate": 3.995913853358887e-06,
      "loss": 0.0021,
      "step": 111460
    },
    {
      "epoch": 1.8239384766423954,
      "grad_norm": 0.09400305896997452,
      "learning_rate": 3.994981323714304e-06,
      "loss": 0.0008,
      "step": 111470
    },
    {
      "epoch": 1.8241021025934714,
      "grad_norm": 0.1379026621580124,
      "learning_rate": 3.994048830499232e-06,
      "loss": 0.0022,
      "step": 111480
    },
    {
      "epoch": 1.8242657285445472,
      "grad_norm": 0.08610210567712784,
      "learning_rate": 3.993116373747472e-06,
      "loss": 0.0014,
      "step": 111490
    },
    {
      "epoch": 1.824429354495623,
      "grad_norm": 0.05595541000366211,
      "learning_rate": 3.992183953492824e-06,
      "loss": 0.0008,
      "step": 111500
    },
    {
      "epoch": 1.824592980446699,
      "grad_norm": 0.06122478470206261,
      "learning_rate": 3.991251569769085e-06,
      "loss": 0.0018,
      "step": 111510
    },
    {
      "epoch": 1.8247566063977747,
      "grad_norm": 0.08093955367803574,
      "learning_rate": 3.990319222610053e-06,
      "loss": 0.0013,
      "step": 111520
    },
    {
      "epoch": 1.8249202323488505,
      "grad_norm": 0.11188928782939911,
      "learning_rate": 3.989386912049521e-06,
      "loss": 0.0012,
      "step": 111530
    },
    {
      "epoch": 1.8250838582999265,
      "grad_norm": 0.09362109750509262,
      "learning_rate": 3.988454638121286e-06,
      "loss": 0.0015,
      "step": 111540
    },
    {
      "epoch": 1.8252474842510023,
      "grad_norm": 0.047657836228609085,
      "learning_rate": 3.98752240085914e-06,
      "loss": 0.0016,
      "step": 111550
    },
    {
      "epoch": 1.825411110202078,
      "grad_norm": 0.06392499059438705,
      "learning_rate": 3.986590200296872e-06,
      "loss": 0.0013,
      "step": 111560
    },
    {
      "epoch": 1.825574736153154,
      "grad_norm": 0.021842509508132935,
      "learning_rate": 3.985658036468273e-06,
      "loss": 0.0009,
      "step": 111570
    },
    {
      "epoch": 1.8257383621042296,
      "grad_norm": 0.06912890821695328,
      "learning_rate": 3.984725909407133e-06,
      "loss": 0.0014,
      "step": 111580
    },
    {
      "epoch": 1.8259019880553056,
      "grad_norm": 0.062236666679382324,
      "learning_rate": 3.983793819147239e-06,
      "loss": 0.0014,
      "step": 111590
    },
    {
      "epoch": 1.8260656140063816,
      "grad_norm": 0.051780059933662415,
      "learning_rate": 3.9828617657223765e-06,
      "loss": 0.0018,
      "step": 111600
    },
    {
      "epoch": 1.8262292399574571,
      "grad_norm": 0.37786462903022766,
      "learning_rate": 3.98192974916633e-06,
      "loss": 0.0019,
      "step": 111610
    },
    {
      "epoch": 1.8263928659085331,
      "grad_norm": 0.07932049036026001,
      "learning_rate": 3.980997769512882e-06,
      "loss": 0.0009,
      "step": 111620
    },
    {
      "epoch": 1.8265564918596089,
      "grad_norm": 0.08780483901500702,
      "learning_rate": 3.980065826795817e-06,
      "loss": 0.0024,
      "step": 111630
    },
    {
      "epoch": 1.8267201178106847,
      "grad_norm": 0.14719057083129883,
      "learning_rate": 3.979133921048915e-06,
      "loss": 0.0014,
      "step": 111640
    },
    {
      "epoch": 1.8268837437617607,
      "grad_norm": 0.024466460570693016,
      "learning_rate": 3.9782020523059565e-06,
      "loss": 0.0017,
      "step": 111650
    },
    {
      "epoch": 1.8270473697128364,
      "grad_norm": 0.0958031713962555,
      "learning_rate": 3.977270220600715e-06,
      "loss": 0.0024,
      "step": 111660
    },
    {
      "epoch": 1.8272109956639122,
      "grad_norm": 0.14239314198493958,
      "learning_rate": 3.976338425966971e-06,
      "loss": 0.0017,
      "step": 111670
    },
    {
      "epoch": 1.8273746216149882,
      "grad_norm": 0.13566423952579498,
      "learning_rate": 3.975406668438496e-06,
      "loss": 0.0032,
      "step": 111680
    },
    {
      "epoch": 1.827538247566064,
      "grad_norm": 0.07592587172985077,
      "learning_rate": 3.9744749480490695e-06,
      "loss": 0.0011,
      "step": 111690
    },
    {
      "epoch": 1.8277018735171398,
      "grad_norm": 0.11134257912635803,
      "learning_rate": 3.973543264832461e-06,
      "loss": 0.0009,
      "step": 111700
    },
    {
      "epoch": 1.8278654994682157,
      "grad_norm": 0.12962402403354645,
      "learning_rate": 3.972611618822441e-06,
      "loss": 0.0013,
      "step": 111710
    },
    {
      "epoch": 1.8280291254192915,
      "grad_norm": 0.025920966640114784,
      "learning_rate": 3.9716800100527826e-06,
      "loss": 0.0012,
      "step": 111720
    },
    {
      "epoch": 1.8281927513703673,
      "grad_norm": 0.04941943287849426,
      "learning_rate": 3.970748438557252e-06,
      "loss": 0.0015,
      "step": 111730
    },
    {
      "epoch": 1.8283563773214433,
      "grad_norm": 0.0675387755036354,
      "learning_rate": 3.9698169043696155e-06,
      "loss": 0.0016,
      "step": 111740
    },
    {
      "epoch": 1.828520003272519,
      "grad_norm": 0.17346258461475372,
      "learning_rate": 3.968885407523641e-06,
      "loss": 0.0017,
      "step": 111750
    },
    {
      "epoch": 1.8286836292235948,
      "grad_norm": 0.3269699811935425,
      "learning_rate": 3.967953948053093e-06,
      "loss": 0.0027,
      "step": 111760
    },
    {
      "epoch": 1.8288472551746708,
      "grad_norm": 0.21526245772838593,
      "learning_rate": 3.967022525991733e-06,
      "loss": 0.002,
      "step": 111770
    },
    {
      "epoch": 1.8290108811257464,
      "grad_norm": 0.12804865837097168,
      "learning_rate": 3.966091141373323e-06,
      "loss": 0.0019,
      "step": 111780
    },
    {
      "epoch": 1.8291745070768224,
      "grad_norm": 0.06192389875650406,
      "learning_rate": 3.965159794231625e-06,
      "loss": 0.0017,
      "step": 111790
    },
    {
      "epoch": 1.8293381330278984,
      "grad_norm": 0.06430929154157639,
      "learning_rate": 3.964228484600397e-06,
      "loss": 0.0014,
      "step": 111800
    },
    {
      "epoch": 1.829501758978974,
      "grad_norm": 0.03203820809721947,
      "learning_rate": 3.963297212513397e-06,
      "loss": 0.0011,
      "step": 111810
    },
    {
      "epoch": 1.82966538493005,
      "grad_norm": 0.19966986775398254,
      "learning_rate": 3.962365978004381e-06,
      "loss": 0.0016,
      "step": 111820
    },
    {
      "epoch": 1.8298290108811257,
      "grad_norm": 0.05639839172363281,
      "learning_rate": 3.961434781107105e-06,
      "loss": 0.0013,
      "step": 111830
    },
    {
      "epoch": 1.8299926368322015,
      "grad_norm": 0.04205803573131561,
      "learning_rate": 3.960503621855322e-06,
      "loss": 0.0023,
      "step": 111840
    },
    {
      "epoch": 1.8301562627832775,
      "grad_norm": 0.1549540013074875,
      "learning_rate": 3.959572500282786e-06,
      "loss": 0.0013,
      "step": 111850
    },
    {
      "epoch": 1.8303198887343533,
      "grad_norm": 0.022245367988944054,
      "learning_rate": 3.958641416423245e-06,
      "loss": 0.0018,
      "step": 111860
    },
    {
      "epoch": 1.830483514685429,
      "grad_norm": 0.10699019581079483,
      "learning_rate": 3.957710370310449e-06,
      "loss": 0.0013,
      "step": 111870
    },
    {
      "epoch": 1.830647140636505,
      "grad_norm": 0.05376095697283745,
      "learning_rate": 3.956779361978148e-06,
      "loss": 0.0013,
      "step": 111880
    },
    {
      "epoch": 1.8308107665875808,
      "grad_norm": 0.09463055431842804,
      "learning_rate": 3.955848391460087e-06,
      "loss": 0.0014,
      "step": 111890
    },
    {
      "epoch": 1.8309743925386566,
      "grad_norm": 0.023398658260703087,
      "learning_rate": 3.954917458790011e-06,
      "loss": 0.0021,
      "step": 111900
    },
    {
      "epoch": 1.8311380184897326,
      "grad_norm": 0.16249677538871765,
      "learning_rate": 3.9539865640016675e-06,
      "loss": 0.0012,
      "step": 111910
    },
    {
      "epoch": 1.8313016444408083,
      "grad_norm": 0.08558373898267746,
      "learning_rate": 3.953055707128796e-06,
      "loss": 0.0014,
      "step": 111920
    },
    {
      "epoch": 1.8314652703918841,
      "grad_norm": 0.04150815308094025,
      "learning_rate": 3.952124888205139e-06,
      "loss": 0.001,
      "step": 111930
    },
    {
      "epoch": 1.8316288963429601,
      "grad_norm": 0.039937574416399,
      "learning_rate": 3.951194107264436e-06,
      "loss": 0.0016,
      "step": 111940
    },
    {
      "epoch": 1.8317925222940359,
      "grad_norm": 0.18946243822574615,
      "learning_rate": 3.95026336434043e-06,
      "loss": 0.0025,
      "step": 111950
    },
    {
      "epoch": 1.8319561482451117,
      "grad_norm": 0.023021534085273743,
      "learning_rate": 3.9493326594668494e-06,
      "loss": 0.0018,
      "step": 111960
    },
    {
      "epoch": 1.8321197741961877,
      "grad_norm": 0.22383436560630798,
      "learning_rate": 3.9484019926774356e-06,
      "loss": 0.0016,
      "step": 111970
    },
    {
      "epoch": 1.8322834001472632,
      "grad_norm": 0.07325699180364609,
      "learning_rate": 3.947471364005922e-06,
      "loss": 0.0018,
      "step": 111980
    },
    {
      "epoch": 1.8324470260983392,
      "grad_norm": 0.09470678120851517,
      "learning_rate": 3.946540773486042e-06,
      "loss": 0.0017,
      "step": 111990
    },
    {
      "epoch": 1.8326106520494152,
      "grad_norm": 0.07208456844091415,
      "learning_rate": 3.945610221151527e-06,
      "loss": 0.0011,
      "step": 112000
    },
    {
      "epoch": 1.8327742780004908,
      "grad_norm": 0.22049634158611298,
      "learning_rate": 3.944679707036107e-06,
      "loss": 0.0031,
      "step": 112010
    },
    {
      "epoch": 1.8329379039515667,
      "grad_norm": 0.04909619316458702,
      "learning_rate": 3.943749231173512e-06,
      "loss": 0.0012,
      "step": 112020
    },
    {
      "epoch": 1.8331015299026425,
      "grad_norm": 0.07673922926187515,
      "learning_rate": 3.942818793597468e-06,
      "loss": 0.0019,
      "step": 112030
    },
    {
      "epoch": 1.8332651558537183,
      "grad_norm": 0.14284048974514008,
      "learning_rate": 3.941888394341703e-06,
      "loss": 0.0018,
      "step": 112040
    },
    {
      "epoch": 1.8334287818047943,
      "grad_norm": 0.046910837292671204,
      "learning_rate": 3.94095803343994e-06,
      "loss": 0.0013,
      "step": 112050
    },
    {
      "epoch": 1.83359240775587,
      "grad_norm": 0.08056025952100754,
      "learning_rate": 3.940027710925903e-06,
      "loss": 0.0013,
      "step": 112060
    },
    {
      "epoch": 1.8337560337069458,
      "grad_norm": 0.06476296484470367,
      "learning_rate": 3.9390974268333134e-06,
      "loss": 0.0021,
      "step": 112070
    },
    {
      "epoch": 1.8339196596580218,
      "grad_norm": 0.14994403719902039,
      "learning_rate": 3.938167181195893e-06,
      "loss": 0.0014,
      "step": 112080
    },
    {
      "epoch": 1.8340832856090976,
      "grad_norm": 0.09715993702411652,
      "learning_rate": 3.93723697404736e-06,
      "loss": 0.0017,
      "step": 112090
    },
    {
      "epoch": 1.8342469115601734,
      "grad_norm": 0.12101032584905624,
      "learning_rate": 3.936306805421432e-06,
      "loss": 0.0009,
      "step": 112100
    },
    {
      "epoch": 1.8344105375112494,
      "grad_norm": 0.08441198617219925,
      "learning_rate": 3.935376675351826e-06,
      "loss": 0.0013,
      "step": 112110
    },
    {
      "epoch": 1.8345741634623252,
      "grad_norm": 0.13522681593894958,
      "learning_rate": 3.934446583872256e-06,
      "loss": 0.0029,
      "step": 112120
    },
    {
      "epoch": 1.834737789413401,
      "grad_norm": 0.34163525700569153,
      "learning_rate": 3.933516531016437e-06,
      "loss": 0.002,
      "step": 112130
    },
    {
      "epoch": 1.834901415364477,
      "grad_norm": 0.09370183944702148,
      "learning_rate": 3.9325865168180806e-06,
      "loss": 0.0011,
      "step": 112140
    },
    {
      "epoch": 1.8350650413155525,
      "grad_norm": 0.05973367393016815,
      "learning_rate": 3.9316565413109e-06,
      "loss": 0.0016,
      "step": 112150
    },
    {
      "epoch": 1.8352286672666285,
      "grad_norm": 0.047129448503255844,
      "learning_rate": 3.930726604528599e-06,
      "loss": 0.0009,
      "step": 112160
    },
    {
      "epoch": 1.8353922932177045,
      "grad_norm": 0.06735831499099731,
      "learning_rate": 3.929796706504889e-06,
      "loss": 0.0019,
      "step": 112170
    },
    {
      "epoch": 1.83555591916878,
      "grad_norm": 0.03659762814640999,
      "learning_rate": 3.928866847273476e-06,
      "loss": 0.0015,
      "step": 112180
    },
    {
      "epoch": 1.835719545119856,
      "grad_norm": 0.1317969709634781,
      "learning_rate": 3.927937026868065e-06,
      "loss": 0.0016,
      "step": 112190
    },
    {
      "epoch": 1.8358831710709318,
      "grad_norm": 0.08199972659349442,
      "learning_rate": 3.927007245322362e-06,
      "loss": 0.0008,
      "step": 112200
    },
    {
      "epoch": 1.8360467970220076,
      "grad_norm": 0.24865856766700745,
      "learning_rate": 3.926077502670066e-06,
      "loss": 0.0019,
      "step": 112210
    },
    {
      "epoch": 1.8362104229730836,
      "grad_norm": 0.01620643213391304,
      "learning_rate": 3.92514779894488e-06,
      "loss": 0.001,
      "step": 112220
    },
    {
      "epoch": 1.8363740489241593,
      "grad_norm": 0.06472311168909073,
      "learning_rate": 3.924218134180503e-06,
      "loss": 0.0014,
      "step": 112230
    },
    {
      "epoch": 1.8365376748752351,
      "grad_norm": 0.04937063530087471,
      "learning_rate": 3.923288508410633e-06,
      "loss": 0.0019,
      "step": 112240
    },
    {
      "epoch": 1.8367013008263111,
      "grad_norm": 0.06868140399456024,
      "learning_rate": 3.9223589216689674e-06,
      "loss": 0.0009,
      "step": 112250
    },
    {
      "epoch": 1.8368649267773869,
      "grad_norm": 0.04069497808814049,
      "learning_rate": 3.9214293739892e-06,
      "loss": 0.0014,
      "step": 112260
    },
    {
      "epoch": 1.8370285527284627,
      "grad_norm": 0.11491206288337708,
      "learning_rate": 3.9204998654050256e-06,
      "loss": 0.0009,
      "step": 112270
    },
    {
      "epoch": 1.8371921786795387,
      "grad_norm": 0.06248701736330986,
      "learning_rate": 3.919570395950137e-06,
      "loss": 0.0029,
      "step": 112280
    },
    {
      "epoch": 1.8373558046306144,
      "grad_norm": 0.07207425683736801,
      "learning_rate": 3.9186409656582245e-06,
      "loss": 0.0015,
      "step": 112290
    },
    {
      "epoch": 1.8375194305816902,
      "grad_norm": 0.06867527216672897,
      "learning_rate": 3.9177115745629775e-06,
      "loss": 0.0012,
      "step": 112300
    },
    {
      "epoch": 1.8376830565327662,
      "grad_norm": 0.38233765959739685,
      "learning_rate": 3.916782222698085e-06,
      "loss": 0.0018,
      "step": 112310
    },
    {
      "epoch": 1.837846682483842,
      "grad_norm": 0.07279100269079208,
      "learning_rate": 3.915852910097234e-06,
      "loss": 0.0015,
      "step": 112320
    },
    {
      "epoch": 1.8380103084349178,
      "grad_norm": 0.1465032696723938,
      "learning_rate": 3.914923636794109e-06,
      "loss": 0.0012,
      "step": 112330
    },
    {
      "epoch": 1.8381739343859937,
      "grad_norm": 0.08444641530513763,
      "learning_rate": 3.913994402822393e-06,
      "loss": 0.0012,
      "step": 112340
    },
    {
      "epoch": 1.8383375603370693,
      "grad_norm": 0.11615827679634094,
      "learning_rate": 3.913065208215773e-06,
      "loss": 0.001,
      "step": 112350
    },
    {
      "epoch": 1.8385011862881453,
      "grad_norm": 0.1292015016078949,
      "learning_rate": 3.912136053007925e-06,
      "loss": 0.0015,
      "step": 112360
    },
    {
      "epoch": 1.8386648122392213,
      "grad_norm": 0.05482890456914902,
      "learning_rate": 3.91120693723253e-06,
      "loss": 0.0011,
      "step": 112370
    },
    {
      "epoch": 1.8388284381902968,
      "grad_norm": 0.05896975100040436,
      "learning_rate": 3.910277860923267e-06,
      "loss": 0.0014,
      "step": 112380
    },
    {
      "epoch": 1.8389920641413728,
      "grad_norm": 0.0504162423312664,
      "learning_rate": 3.9093488241138125e-06,
      "loss": 0.0008,
      "step": 112390
    },
    {
      "epoch": 1.8391556900924486,
      "grad_norm": 0.19282062351703644,
      "learning_rate": 3.90841982683784e-06,
      "loss": 0.0018,
      "step": 112400
    },
    {
      "epoch": 1.8393193160435244,
      "grad_norm": 0.030025051906704903,
      "learning_rate": 3.907490869129028e-06,
      "loss": 0.0011,
      "step": 112410
    },
    {
      "epoch": 1.8394829419946004,
      "grad_norm": 0.051199305802583694,
      "learning_rate": 3.906561951021044e-06,
      "loss": 0.0014,
      "step": 112420
    },
    {
      "epoch": 1.8396465679456762,
      "grad_norm": 0.00431101955473423,
      "learning_rate": 3.905633072547564e-06,
      "loss": 0.0021,
      "step": 112430
    },
    {
      "epoch": 1.839810193896752,
      "grad_norm": 0.07762044668197632,
      "learning_rate": 3.904704233742252e-06,
      "loss": 0.0019,
      "step": 112440
    },
    {
      "epoch": 1.839973819847828,
      "grad_norm": 0.07841220498085022,
      "learning_rate": 3.903775434638782e-06,
      "loss": 0.001,
      "step": 112450
    },
    {
      "epoch": 1.8401374457989037,
      "grad_norm": 0.025246400386095047,
      "learning_rate": 3.902846675270816e-06,
      "loss": 0.0012,
      "step": 112460
    },
    {
      "epoch": 1.8403010717499795,
      "grad_norm": 0.03706952929496765,
      "learning_rate": 3.901917955672021e-06,
      "loss": 0.0006,
      "step": 112470
    },
    {
      "epoch": 1.8404646977010555,
      "grad_norm": 0.18933212757110596,
      "learning_rate": 3.900989275876061e-06,
      "loss": 0.0018,
      "step": 112480
    },
    {
      "epoch": 1.8406283236521312,
      "grad_norm": 0.02917795069515705,
      "learning_rate": 3.900060635916599e-06,
      "loss": 0.0014,
      "step": 112490
    },
    {
      "epoch": 1.840791949603207,
      "grad_norm": 0.1295352280139923,
      "learning_rate": 3.8991320358272945e-06,
      "loss": 0.0013,
      "step": 112500
    },
    {
      "epoch": 1.840955575554283,
      "grad_norm": 0.05583084374666214,
      "learning_rate": 3.898203475641808e-06,
      "loss": 0.0009,
      "step": 112510
    },
    {
      "epoch": 1.8411192015053588,
      "grad_norm": 0.09758180379867554,
      "learning_rate": 3.897274955393798e-06,
      "loss": 0.0023,
      "step": 112520
    },
    {
      "epoch": 1.8412828274564346,
      "grad_norm": 0.07284069806337357,
      "learning_rate": 3.89634647511692e-06,
      "loss": 0.0012,
      "step": 112530
    },
    {
      "epoch": 1.8414464534075106,
      "grad_norm": 0.012880626134574413,
      "learning_rate": 3.89541803484483e-06,
      "loss": 0.0029,
      "step": 112540
    },
    {
      "epoch": 1.8416100793585861,
      "grad_norm": 0.010092221200466156,
      "learning_rate": 3.894489634611182e-06,
      "loss": 0.0014,
      "step": 112550
    },
    {
      "epoch": 1.8417737053096621,
      "grad_norm": 0.0824425220489502,
      "learning_rate": 3.893561274449627e-06,
      "loss": 0.0012,
      "step": 112560
    },
    {
      "epoch": 1.841937331260738,
      "grad_norm": 0.14740385115146637,
      "learning_rate": 3.892632954393817e-06,
      "loss": 0.0012,
      "step": 112570
    },
    {
      "epoch": 1.8421009572118137,
      "grad_norm": 0.11079197376966476,
      "learning_rate": 3.8917046744774e-06,
      "loss": 0.0006,
      "step": 112580
    },
    {
      "epoch": 1.8422645831628897,
      "grad_norm": 0.05790552496910095,
      "learning_rate": 3.890776434734025e-06,
      "loss": 0.0026,
      "step": 112590
    },
    {
      "epoch": 1.8424282091139654,
      "grad_norm": 0.03640960156917572,
      "learning_rate": 3.8898482351973385e-06,
      "loss": 0.0016,
      "step": 112600
    },
    {
      "epoch": 1.8425918350650412,
      "grad_norm": 0.15015466511249542,
      "learning_rate": 3.888920075900985e-06,
      "loss": 0.0021,
      "step": 112610
    },
    {
      "epoch": 1.8427554610161172,
      "grad_norm": 0.16366706788539886,
      "learning_rate": 3.8879919568786086e-06,
      "loss": 0.0015,
      "step": 112620
    },
    {
      "epoch": 1.842919086967193,
      "grad_norm": 0.07634207606315613,
      "learning_rate": 3.887063878163851e-06,
      "loss": 0.0012,
      "step": 112630
    },
    {
      "epoch": 1.8430827129182688,
      "grad_norm": 0.1957336962223053,
      "learning_rate": 3.886135839790353e-06,
      "loss": 0.0009,
      "step": 112640
    },
    {
      "epoch": 1.8432463388693447,
      "grad_norm": 0.04755368083715439,
      "learning_rate": 3.885207841791756e-06,
      "loss": 0.0016,
      "step": 112650
    },
    {
      "epoch": 1.8434099648204205,
      "grad_norm": 0.02451455406844616,
      "learning_rate": 3.8842798842016935e-06,
      "loss": 0.0017,
      "step": 112660
    },
    {
      "epoch": 1.8435735907714963,
      "grad_norm": 0.10452144593000412,
      "learning_rate": 3.8833519670538026e-06,
      "loss": 0.0014,
      "step": 112670
    },
    {
      "epoch": 1.8437372167225723,
      "grad_norm": 0.1028767079114914,
      "learning_rate": 3.88242409038172e-06,
      "loss": 0.0025,
      "step": 112680
    },
    {
      "epoch": 1.843900842673648,
      "grad_norm": 0.01705436035990715,
      "learning_rate": 3.881496254219078e-06,
      "loss": 0.0017,
      "step": 112690
    },
    {
      "epoch": 1.8440644686247238,
      "grad_norm": 0.1025237888097763,
      "learning_rate": 3.880568458599508e-06,
      "loss": 0.0023,
      "step": 112700
    },
    {
      "epoch": 1.8442280945757998,
      "grad_norm": 0.044633958488702774,
      "learning_rate": 3.879640703556641e-06,
      "loss": 0.0015,
      "step": 112710
    },
    {
      "epoch": 1.8443917205268756,
      "grad_norm": 0.08035072684288025,
      "learning_rate": 3.878712989124107e-06,
      "loss": 0.0005,
      "step": 112720
    },
    {
      "epoch": 1.8445553464779514,
      "grad_norm": 0.013839014805853367,
      "learning_rate": 3.877785315335531e-06,
      "loss": 0.0013,
      "step": 112730
    },
    {
      "epoch": 1.8447189724290274,
      "grad_norm": 0.043414004147052765,
      "learning_rate": 3.8768576822245405e-06,
      "loss": 0.0018,
      "step": 112740
    },
    {
      "epoch": 1.844882598380103,
      "grad_norm": 0.08317239582538605,
      "learning_rate": 3.875930089824761e-06,
      "loss": 0.0013,
      "step": 112750
    },
    {
      "epoch": 1.845046224331179,
      "grad_norm": 0.16005107760429382,
      "learning_rate": 3.875002538169812e-06,
      "loss": 0.0013,
      "step": 112760
    },
    {
      "epoch": 1.845209850282255,
      "grad_norm": 0.06475251913070679,
      "learning_rate": 3.874075027293318e-06,
      "loss": 0.0015,
      "step": 112770
    },
    {
      "epoch": 1.8453734762333305,
      "grad_norm": 0.13961884379386902,
      "learning_rate": 3.873147557228899e-06,
      "loss": 0.0008,
      "step": 112780
    },
    {
      "epoch": 1.8455371021844065,
      "grad_norm": 0.14066645503044128,
      "learning_rate": 3.872220128010171e-06,
      "loss": 0.0014,
      "step": 112790
    },
    {
      "epoch": 1.8457007281354822,
      "grad_norm": 0.0874023362994194,
      "learning_rate": 3.871292739670754e-06,
      "loss": 0.0022,
      "step": 112800
    },
    {
      "epoch": 1.845864354086558,
      "grad_norm": 0.14040616154670715,
      "learning_rate": 3.8703653922442615e-06,
      "loss": 0.0009,
      "step": 112810
    },
    {
      "epoch": 1.846027980037634,
      "grad_norm": 0.38856273889541626,
      "learning_rate": 3.86943808576431e-06,
      "loss": 0.0014,
      "step": 112820
    },
    {
      "epoch": 1.8461916059887098,
      "grad_norm": 0.01496563758701086,
      "learning_rate": 3.868510820264508e-06,
      "loss": 0.0011,
      "step": 112830
    },
    {
      "epoch": 1.8463552319397856,
      "grad_norm": 0.068240687251091,
      "learning_rate": 3.8675835957784715e-06,
      "loss": 0.0021,
      "step": 112840
    },
    {
      "epoch": 1.8465188578908616,
      "grad_norm": 0.07026727497577667,
      "learning_rate": 3.86665641233981e-06,
      "loss": 0.0014,
      "step": 112850
    },
    {
      "epoch": 1.8466824838419373,
      "grad_norm": 0.14649271965026855,
      "learning_rate": 3.8657292699821265e-06,
      "loss": 0.0009,
      "step": 112860
    },
    {
      "epoch": 1.8468461097930131,
      "grad_norm": 0.2210388481616974,
      "learning_rate": 3.864802168739032e-06,
      "loss": 0.0021,
      "step": 112870
    },
    {
      "epoch": 1.847009735744089,
      "grad_norm": 0.21292605996131897,
      "learning_rate": 3.86387510864413e-06,
      "loss": 0.0012,
      "step": 112880
    },
    {
      "epoch": 1.8471733616951649,
      "grad_norm": 0.06951511651277542,
      "learning_rate": 3.8629480897310235e-06,
      "loss": 0.0012,
      "step": 112890
    },
    {
      "epoch": 1.8473369876462407,
      "grad_norm": 0.06008486449718475,
      "learning_rate": 3.862021112033317e-06,
      "loss": 0.0017,
      "step": 112900
    },
    {
      "epoch": 1.8475006135973167,
      "grad_norm": 0.08109413832426071,
      "learning_rate": 3.86109417558461e-06,
      "loss": 0.0022,
      "step": 112910
    },
    {
      "epoch": 1.8476642395483924,
      "grad_norm": 0.04310496151447296,
      "learning_rate": 3.860167280418502e-06,
      "loss": 0.0015,
      "step": 112920
    },
    {
      "epoch": 1.8478278654994682,
      "grad_norm": 0.048203244805336,
      "learning_rate": 3.8592404265685905e-06,
      "loss": 0.0017,
      "step": 112930
    },
    {
      "epoch": 1.8479914914505442,
      "grad_norm": 0.10649536550045013,
      "learning_rate": 3.858313614068472e-06,
      "loss": 0.0013,
      "step": 112940
    },
    {
      "epoch": 1.8481551174016198,
      "grad_norm": 0.005922969430685043,
      "learning_rate": 3.857386842951742e-06,
      "loss": 0.0006,
      "step": 112950
    },
    {
      "epoch": 1.8483187433526957,
      "grad_norm": 0.15154141187667847,
      "learning_rate": 3.856460113251992e-06,
      "loss": 0.0017,
      "step": 112960
    },
    {
      "epoch": 1.8484823693037717,
      "grad_norm": 0.15933385491371155,
      "learning_rate": 3.855533425002814e-06,
      "loss": 0.001,
      "step": 112970
    },
    {
      "epoch": 1.8486459952548473,
      "grad_norm": 0.23300373554229736,
      "learning_rate": 3.854606778237799e-06,
      "loss": 0.0023,
      "step": 112980
    },
    {
      "epoch": 1.8488096212059233,
      "grad_norm": 0.2681264877319336,
      "learning_rate": 3.853680172990535e-06,
      "loss": 0.0011,
      "step": 112990
    },
    {
      "epoch": 1.848973247156999,
      "grad_norm": 0.06967673450708389,
      "learning_rate": 3.85275360929461e-06,
      "loss": 0.0011,
      "step": 113000
    },
    {
      "epoch": 1.8491368731080748,
      "grad_norm": 0.12202316522598267,
      "learning_rate": 3.8518270871836094e-06,
      "loss": 0.0025,
      "step": 113010
    },
    {
      "epoch": 1.8493004990591508,
      "grad_norm": 0.06977274268865585,
      "learning_rate": 3.850900606691117e-06,
      "loss": 0.0016,
      "step": 113020
    },
    {
      "epoch": 1.8494641250102266,
      "grad_norm": 0.13219143450260162,
      "learning_rate": 3.849974167850715e-06,
      "loss": 0.0014,
      "step": 113030
    },
    {
      "epoch": 1.8496277509613024,
      "grad_norm": 0.005337636452168226,
      "learning_rate": 3.849047770695987e-06,
      "loss": 0.0008,
      "step": 113040
    },
    {
      "epoch": 1.8497913769123784,
      "grad_norm": 0.05751505866646767,
      "learning_rate": 3.848121415260512e-06,
      "loss": 0.001,
      "step": 113050
    },
    {
      "epoch": 1.8499550028634542,
      "grad_norm": 0.04354952648282051,
      "learning_rate": 3.8471951015778655e-06,
      "loss": 0.0019,
      "step": 113060
    },
    {
      "epoch": 1.85011862881453,
      "grad_norm": 0.05468391999602318,
      "learning_rate": 3.8462688296816254e-06,
      "loss": 0.0012,
      "step": 113070
    },
    {
      "epoch": 1.850282254765606,
      "grad_norm": 0.08674248307943344,
      "learning_rate": 3.845342599605368e-06,
      "loss": 0.0014,
      "step": 113080
    },
    {
      "epoch": 1.8504458807166817,
      "grad_norm": 0.12483728677034378,
      "learning_rate": 3.844416411382667e-06,
      "loss": 0.0021,
      "step": 113090
    },
    {
      "epoch": 1.8506095066677575,
      "grad_norm": 0.013608505949378014,
      "learning_rate": 3.843490265047092e-06,
      "loss": 0.0021,
      "step": 113100
    },
    {
      "epoch": 1.8507731326188335,
      "grad_norm": 0.11857075989246368,
      "learning_rate": 3.842564160632216e-06,
      "loss": 0.0019,
      "step": 113110
    },
    {
      "epoch": 1.850936758569909,
      "grad_norm": 0.18585699796676636,
      "learning_rate": 3.841638098171607e-06,
      "loss": 0.0019,
      "step": 113120
    },
    {
      "epoch": 1.851100384520985,
      "grad_norm": 0.15069429576396942,
      "learning_rate": 3.8407120776988335e-06,
      "loss": 0.0016,
      "step": 113130
    },
    {
      "epoch": 1.851264010472061,
      "grad_norm": 0.015933746472001076,
      "learning_rate": 3.83978609924746e-06,
      "loss": 0.0014,
      "step": 113140
    },
    {
      "epoch": 1.8514276364231366,
      "grad_norm": 0.03865732252597809,
      "learning_rate": 3.838860162851054e-06,
      "loss": 0.001,
      "step": 113150
    },
    {
      "epoch": 1.8515912623742126,
      "grad_norm": 0.05710527300834656,
      "learning_rate": 3.837934268543176e-06,
      "loss": 0.0006,
      "step": 113160
    },
    {
      "epoch": 1.8517548883252883,
      "grad_norm": 0.09011007845401764,
      "learning_rate": 3.837008416357386e-06,
      "loss": 0.0015,
      "step": 113170
    },
    {
      "epoch": 1.8519185142763641,
      "grad_norm": 0.4016435444355011,
      "learning_rate": 3.836082606327247e-06,
      "loss": 0.0024,
      "step": 113180
    },
    {
      "epoch": 1.85208214022744,
      "grad_norm": 0.10283300280570984,
      "learning_rate": 3.835156838486316e-06,
      "loss": 0.001,
      "step": 113190
    },
    {
      "epoch": 1.8522457661785159,
      "grad_norm": 0.15302906930446625,
      "learning_rate": 3.83423111286815e-06,
      "loss": 0.001,
      "step": 113200
    },
    {
      "epoch": 1.8524093921295917,
      "grad_norm": 0.045707181096076965,
      "learning_rate": 3.833305429506303e-06,
      "loss": 0.0013,
      "step": 113210
    },
    {
      "epoch": 1.8525730180806677,
      "grad_norm": 0.06647147238254547,
      "learning_rate": 3.832379788434331e-06,
      "loss": 0.0011,
      "step": 113220
    },
    {
      "epoch": 1.8527366440317434,
      "grad_norm": 0.2728255093097687,
      "learning_rate": 3.831454189685786e-06,
      "loss": 0.0026,
      "step": 113230
    },
    {
      "epoch": 1.8529002699828192,
      "grad_norm": 0.10586028546094894,
      "learning_rate": 3.830528633294218e-06,
      "loss": 0.0019,
      "step": 113240
    },
    {
      "epoch": 1.8530638959338952,
      "grad_norm": 0.12289302051067352,
      "learning_rate": 3.829603119293176e-06,
      "loss": 0.002,
      "step": 113250
    },
    {
      "epoch": 1.853227521884971,
      "grad_norm": 0.025495300069451332,
      "learning_rate": 3.828677647716209e-06,
      "loss": 0.0015,
      "step": 113260
    },
    {
      "epoch": 1.8533911478360467,
      "grad_norm": 0.083602674305439,
      "learning_rate": 3.827752218596861e-06,
      "loss": 0.0014,
      "step": 113270
    },
    {
      "epoch": 1.8535547737871227,
      "grad_norm": 0.14927873015403748,
      "learning_rate": 3.826826831968677e-06,
      "loss": 0.002,
      "step": 113280
    },
    {
      "epoch": 1.8537183997381985,
      "grad_norm": 0.10858125239610672,
      "learning_rate": 3.825901487865201e-06,
      "loss": 0.0013,
      "step": 113290
    },
    {
      "epoch": 1.8538820256892743,
      "grad_norm": 0.24482282996177673,
      "learning_rate": 3.824976186319975e-06,
      "loss": 0.0013,
      "step": 113300
    },
    {
      "epoch": 1.8540456516403503,
      "grad_norm": 0.10449431091547012,
      "learning_rate": 3.824050927366537e-06,
      "loss": 0.0017,
      "step": 113310
    },
    {
      "epoch": 1.8542092775914258,
      "grad_norm": 0.053672026842832565,
      "learning_rate": 3.823125711038427e-06,
      "loss": 0.0011,
      "step": 113320
    },
    {
      "epoch": 1.8543729035425018,
      "grad_norm": 0.05485406890511513,
      "learning_rate": 3.822200537369181e-06,
      "loss": 0.0019,
      "step": 113330
    },
    {
      "epoch": 1.8545365294935778,
      "grad_norm": 0.1184929832816124,
      "learning_rate": 3.821275406392336e-06,
      "loss": 0.0013,
      "step": 113340
    },
    {
      "epoch": 1.8547001554446534,
      "grad_norm": 0.08440351486206055,
      "learning_rate": 3.8203503181414234e-06,
      "loss": 0.0015,
      "step": 113350
    },
    {
      "epoch": 1.8548637813957294,
      "grad_norm": 0.3196423649787903,
      "learning_rate": 3.819425272649979e-06,
      "loss": 0.0013,
      "step": 113360
    },
    {
      "epoch": 1.8550274073468052,
      "grad_norm": 0.1023973599076271,
      "learning_rate": 3.81850026995153e-06,
      "loss": 0.0012,
      "step": 113370
    },
    {
      "epoch": 1.855191033297881,
      "grad_norm": 0.019896909594535828,
      "learning_rate": 3.817575310079605e-06,
      "loss": 0.0007,
      "step": 113380
    },
    {
      "epoch": 1.855354659248957,
      "grad_norm": 0.021599577739834785,
      "learning_rate": 3.816650393067732e-06,
      "loss": 0.0011,
      "step": 113390
    },
    {
      "epoch": 1.8555182852000327,
      "grad_norm": 0.10887371748685837,
      "learning_rate": 3.81572551894944e-06,
      "loss": 0.0009,
      "step": 113400
    },
    {
      "epoch": 1.8556819111511085,
      "grad_norm": 0.028830768540501595,
      "learning_rate": 3.814800687758252e-06,
      "loss": 0.0006,
      "step": 113410
    },
    {
      "epoch": 1.8558455371021845,
      "grad_norm": 0.03024921752512455,
      "learning_rate": 3.8138758995276896e-06,
      "loss": 0.0024,
      "step": 113420
    },
    {
      "epoch": 1.8560091630532602,
      "grad_norm": 0.024848848581314087,
      "learning_rate": 3.812951154291276e-06,
      "loss": 0.0012,
      "step": 113430
    },
    {
      "epoch": 1.856172789004336,
      "grad_norm": 0.06721098721027374,
      "learning_rate": 3.812026452082529e-06,
      "loss": 0.0019,
      "step": 113440
    },
    {
      "epoch": 1.856336414955412,
      "grad_norm": 0.09745202958583832,
      "learning_rate": 3.8111017929349685e-06,
      "loss": 0.0016,
      "step": 113450
    },
    {
      "epoch": 1.8565000409064878,
      "grad_norm": 0.027345167472958565,
      "learning_rate": 3.8101771768821117e-06,
      "loss": 0.0008,
      "step": 113460
    },
    {
      "epoch": 1.8566636668575636,
      "grad_norm": 0.12136715650558472,
      "learning_rate": 3.809252603957472e-06,
      "loss": 0.0015,
      "step": 113470
    },
    {
      "epoch": 1.8568272928086396,
      "grad_norm": 0.03409392014145851,
      "learning_rate": 3.808328074194562e-06,
      "loss": 0.0018,
      "step": 113480
    },
    {
      "epoch": 1.8569909187597153,
      "grad_norm": 0.04928814619779587,
      "learning_rate": 3.807403587626897e-06,
      "loss": 0.0018,
      "step": 113490
    },
    {
      "epoch": 1.857154544710791,
      "grad_norm": 0.0720706507563591,
      "learning_rate": 3.806479144287984e-06,
      "loss": 0.0015,
      "step": 113500
    },
    {
      "epoch": 1.857318170661867,
      "grad_norm": 0.048940613865852356,
      "learning_rate": 3.8055547442113345e-06,
      "loss": 0.001,
      "step": 113510
    },
    {
      "epoch": 1.8574817966129427,
      "grad_norm": 0.019793879240751266,
      "learning_rate": 3.8046303874304543e-06,
      "loss": 0.0015,
      "step": 113520
    },
    {
      "epoch": 1.8576454225640187,
      "grad_norm": 0.10107264667749405,
      "learning_rate": 3.8037060739788495e-06,
      "loss": 0.001,
      "step": 113530
    },
    {
      "epoch": 1.8578090485150947,
      "grad_norm": 0.06543976068496704,
      "learning_rate": 3.8027818038900245e-06,
      "loss": 0.0008,
      "step": 113540
    },
    {
      "epoch": 1.8579726744661702,
      "grad_norm": 0.04555689916014671,
      "learning_rate": 3.8018575771974808e-06,
      "loss": 0.0013,
      "step": 113550
    },
    {
      "epoch": 1.8581363004172462,
      "grad_norm": 0.031223345547914505,
      "learning_rate": 3.800933393934723e-06,
      "loss": 0.0016,
      "step": 113560
    },
    {
      "epoch": 1.858299926368322,
      "grad_norm": 0.07470271736383438,
      "learning_rate": 3.800009254135246e-06,
      "loss": 0.001,
      "step": 113570
    },
    {
      "epoch": 1.8584635523193977,
      "grad_norm": 0.0019687185995280743,
      "learning_rate": 3.7990851578325496e-06,
      "loss": 0.0017,
      "step": 113580
    },
    {
      "epoch": 1.8586271782704737,
      "grad_norm": 0.1276399940252304,
      "learning_rate": 3.7981611050601298e-06,
      "loss": 0.0008,
      "step": 113590
    },
    {
      "epoch": 1.8587908042215495,
      "grad_norm": 0.18228983879089355,
      "learning_rate": 3.797237095851481e-06,
      "loss": 0.0017,
      "step": 113600
    },
    {
      "epoch": 1.8589544301726253,
      "grad_norm": 0.08215165138244629,
      "learning_rate": 3.7963131302400957e-06,
      "loss": 0.0018,
      "step": 113610
    },
    {
      "epoch": 1.8591180561237013,
      "grad_norm": 0.04695281758904457,
      "learning_rate": 3.7953892082594677e-06,
      "loss": 0.0016,
      "step": 113620
    },
    {
      "epoch": 1.859281682074777,
      "grad_norm": 0.19269028306007385,
      "learning_rate": 3.7944653299430854e-06,
      "loss": 0.0012,
      "step": 113630
    },
    {
      "epoch": 1.8594453080258528,
      "grad_norm": 0.03540990501642227,
      "learning_rate": 3.7935414953244382e-06,
      "loss": 0.0014,
      "step": 113640
    },
    {
      "epoch": 1.8596089339769288,
      "grad_norm": 0.0897429808974266,
      "learning_rate": 3.7926177044370116e-06,
      "loss": 0.0008,
      "step": 113650
    },
    {
      "epoch": 1.8597725599280046,
      "grad_norm": 0.17248032987117767,
      "learning_rate": 3.791693957314293e-06,
      "loss": 0.002,
      "step": 113660
    },
    {
      "epoch": 1.8599361858790804,
      "grad_norm": 0.09790355712175369,
      "learning_rate": 3.790770253989763e-06,
      "loss": 0.0009,
      "step": 113670
    },
    {
      "epoch": 1.8600998118301564,
      "grad_norm": 0.08103250712156296,
      "learning_rate": 3.789846594496904e-06,
      "loss": 0.0013,
      "step": 113680
    },
    {
      "epoch": 1.8602634377812322,
      "grad_norm": 0.08564095944166183,
      "learning_rate": 3.7889229788691985e-06,
      "loss": 0.0015,
      "step": 113690
    },
    {
      "epoch": 1.860427063732308,
      "grad_norm": 0.07448884099721909,
      "learning_rate": 3.787999407140124e-06,
      "loss": 0.0012,
      "step": 113700
    },
    {
      "epoch": 1.860590689683384,
      "grad_norm": 0.026719890534877777,
      "learning_rate": 3.7870758793431584e-06,
      "loss": 0.0014,
      "step": 113710
    },
    {
      "epoch": 1.8607543156344595,
      "grad_norm": 0.04840150102972984,
      "learning_rate": 3.786152395511776e-06,
      "loss": 0.0017,
      "step": 113720
    },
    {
      "epoch": 1.8609179415855355,
      "grad_norm": 0.05359083414077759,
      "learning_rate": 3.785228955679453e-06,
      "loss": 0.0011,
      "step": 113730
    },
    {
      "epoch": 1.8610815675366115,
      "grad_norm": 0.30011770129203796,
      "learning_rate": 3.78430555987966e-06,
      "loss": 0.0016,
      "step": 113740
    },
    {
      "epoch": 1.861245193487687,
      "grad_norm": 0.08568993210792542,
      "learning_rate": 3.7833822081458687e-06,
      "loss": 0.0017,
      "step": 113750
    },
    {
      "epoch": 1.861408819438763,
      "grad_norm": 0.06905089318752289,
      "learning_rate": 3.7824589005115497e-06,
      "loss": 0.0007,
      "step": 113760
    },
    {
      "epoch": 1.8615724453898388,
      "grad_norm": 0.10211402177810669,
      "learning_rate": 3.781535637010168e-06,
      "loss": 0.0016,
      "step": 113770
    },
    {
      "epoch": 1.8617360713409146,
      "grad_norm": 0.1789311319589615,
      "learning_rate": 3.7806124176751906e-06,
      "loss": 0.0013,
      "step": 113780
    },
    {
      "epoch": 1.8618996972919906,
      "grad_norm": 0.04372971132397652,
      "learning_rate": 3.779689242540083e-06,
      "loss": 0.0018,
      "step": 113790
    },
    {
      "epoch": 1.8620633232430663,
      "grad_norm": 0.10091143846511841,
      "learning_rate": 3.778766111638307e-06,
      "loss": 0.0034,
      "step": 113800
    },
    {
      "epoch": 1.8622269491941421,
      "grad_norm": 0.008842662908136845,
      "learning_rate": 3.7778430250033244e-06,
      "loss": 0.001,
      "step": 113810
    },
    {
      "epoch": 1.862390575145218,
      "grad_norm": 0.1413860321044922,
      "learning_rate": 3.7769199826685937e-06,
      "loss": 0.0013,
      "step": 113820
    },
    {
      "epoch": 1.8625542010962939,
      "grad_norm": 0.10416331887245178,
      "learning_rate": 3.7759969846675746e-06,
      "loss": 0.001,
      "step": 113830
    },
    {
      "epoch": 1.8627178270473697,
      "grad_norm": 0.1822335422039032,
      "learning_rate": 3.775074031033724e-06,
      "loss": 0.0018,
      "step": 113840
    },
    {
      "epoch": 1.8628814529984457,
      "grad_norm": 0.051955800503492355,
      "learning_rate": 3.774151121800495e-06,
      "loss": 0.0013,
      "step": 113850
    },
    {
      "epoch": 1.8630450789495214,
      "grad_norm": 0.07419634610414505,
      "learning_rate": 3.773228257001344e-06,
      "loss": 0.0009,
      "step": 113860
    },
    {
      "epoch": 1.8632087049005972,
      "grad_norm": 0.08996762335300446,
      "learning_rate": 3.772305436669718e-06,
      "loss": 0.0011,
      "step": 113870
    },
    {
      "epoch": 1.8633723308516732,
      "grad_norm": 0.008551595732569695,
      "learning_rate": 3.771382660839068e-06,
      "loss": 0.001,
      "step": 113880
    },
    {
      "epoch": 1.8635359568027487,
      "grad_norm": 0.15793809294700623,
      "learning_rate": 3.770459929542845e-06,
      "loss": 0.0011,
      "step": 113890
    },
    {
      "epoch": 1.8636995827538247,
      "grad_norm": 0.07449346780776978,
      "learning_rate": 3.7695372428144945e-06,
      "loss": 0.0023,
      "step": 113900
    },
    {
      "epoch": 1.8638632087049007,
      "grad_norm": 0.05696192756295204,
      "learning_rate": 3.7686146006874615e-06,
      "loss": 0.0011,
      "step": 113910
    },
    {
      "epoch": 1.8640268346559763,
      "grad_norm": 0.04130572825670242,
      "learning_rate": 3.7676920031951897e-06,
      "loss": 0.0026,
      "step": 113920
    },
    {
      "epoch": 1.8641904606070523,
      "grad_norm": 0.16821983456611633,
      "learning_rate": 3.7667694503711207e-06,
      "loss": 0.0024,
      "step": 113930
    },
    {
      "epoch": 1.864354086558128,
      "grad_norm": 0.028291620314121246,
      "learning_rate": 3.7658469422486955e-06,
      "loss": 0.0011,
      "step": 113940
    },
    {
      "epoch": 1.8645177125092038,
      "grad_norm": 0.023886337876319885,
      "learning_rate": 3.764924478861352e-06,
      "loss": 0.0019,
      "step": 113950
    },
    {
      "epoch": 1.8646813384602798,
      "grad_norm": 0.12927983701229095,
      "learning_rate": 3.7640020602425294e-06,
      "loss": 0.0009,
      "step": 113960
    },
    {
      "epoch": 1.8648449644113556,
      "grad_norm": 0.23467424511909485,
      "learning_rate": 3.76307968642566e-06,
      "loss": 0.0012,
      "step": 113970
    },
    {
      "epoch": 1.8650085903624314,
      "grad_norm": 0.062484048306941986,
      "learning_rate": 3.7621573574441794e-06,
      "loss": 0.0013,
      "step": 113980
    },
    {
      "epoch": 1.8651722163135074,
      "grad_norm": 0.0071717011742293835,
      "learning_rate": 3.761235073331519e-06,
      "loss": 0.0014,
      "step": 113990
    },
    {
      "epoch": 1.8653358422645832,
      "grad_norm": 0.10824783891439438,
      "learning_rate": 3.7603128341211105e-06,
      "loss": 0.0013,
      "step": 114000
    },
    {
      "epoch": 1.865499468215659,
      "grad_norm": 0.10666628926992416,
      "learning_rate": 3.759390639846382e-06,
      "loss": 0.0018,
      "step": 114010
    },
    {
      "epoch": 1.865663094166735,
      "grad_norm": 0.10495135933160782,
      "learning_rate": 3.758468490540761e-06,
      "loss": 0.0012,
      "step": 114020
    },
    {
      "epoch": 1.8658267201178107,
      "grad_norm": 0.0713539868593216,
      "learning_rate": 3.757546386237673e-06,
      "loss": 0.0025,
      "step": 114030
    },
    {
      "epoch": 1.8659903460688865,
      "grad_norm": 0.05254041403532028,
      "learning_rate": 3.7566243269705415e-06,
      "loss": 0.0009,
      "step": 114040
    },
    {
      "epoch": 1.8661539720199625,
      "grad_norm": 0.1140858605504036,
      "learning_rate": 3.755702312772791e-06,
      "loss": 0.0013,
      "step": 114050
    },
    {
      "epoch": 1.8663175979710382,
      "grad_norm": 0.00451473006978631,
      "learning_rate": 3.7547803436778428e-06,
      "loss": 0.002,
      "step": 114060
    },
    {
      "epoch": 1.866481223922114,
      "grad_norm": 0.062445018440485,
      "learning_rate": 3.753858419719113e-06,
      "loss": 0.0016,
      "step": 114070
    },
    {
      "epoch": 1.86664484987319,
      "grad_norm": 0.20512571930885315,
      "learning_rate": 3.75293654093002e-06,
      "loss": 0.0012,
      "step": 114080
    },
    {
      "epoch": 1.8668084758242656,
      "grad_norm": 0.05938255414366722,
      "learning_rate": 3.75201470734398e-06,
      "loss": 0.0019,
      "step": 114090
    },
    {
      "epoch": 1.8669721017753416,
      "grad_norm": 0.027770772576332092,
      "learning_rate": 3.751092918994407e-06,
      "loss": 0.0022,
      "step": 114100
    },
    {
      "epoch": 1.8671357277264176,
      "grad_norm": 0.1386348456144333,
      "learning_rate": 3.750171175914714e-06,
      "loss": 0.0016,
      "step": 114110
    },
    {
      "epoch": 1.8672993536774931,
      "grad_norm": 0.1404191255569458,
      "learning_rate": 3.7492494781383126e-06,
      "loss": 0.0013,
      "step": 114120
    },
    {
      "epoch": 1.867462979628569,
      "grad_norm": 0.15915922820568085,
      "learning_rate": 3.748327825698611e-06,
      "loss": 0.0017,
      "step": 114130
    },
    {
      "epoch": 1.8676266055796449,
      "grad_norm": 0.049886420369148254,
      "learning_rate": 3.7474062186290183e-06,
      "loss": 0.0021,
      "step": 114140
    },
    {
      "epoch": 1.8677902315307207,
      "grad_norm": 0.08974398672580719,
      "learning_rate": 3.746484656962939e-06,
      "loss": 0.0013,
      "step": 114150
    },
    {
      "epoch": 1.8679538574817967,
      "grad_norm": 0.09072622656822205,
      "learning_rate": 3.7455631407337794e-06,
      "loss": 0.0012,
      "step": 114160
    },
    {
      "epoch": 1.8681174834328724,
      "grad_norm": 0.06740576028823853,
      "learning_rate": 3.74464166997494e-06,
      "loss": 0.0026,
      "step": 114170
    },
    {
      "epoch": 1.8682811093839482,
      "grad_norm": 0.045059073716402054,
      "learning_rate": 3.743720244719823e-06,
      "loss": 0.0009,
      "step": 114180
    },
    {
      "epoch": 1.8684447353350242,
      "grad_norm": 0.10823206603527069,
      "learning_rate": 3.7427988650018275e-06,
      "loss": 0.0012,
      "step": 114190
    },
    {
      "epoch": 1.8686083612861,
      "grad_norm": 0.03656313195824623,
      "learning_rate": 3.741877530854352e-06,
      "loss": 0.0018,
      "step": 114200
    },
    {
      "epoch": 1.8687719872371757,
      "grad_norm": 0.06545853614807129,
      "learning_rate": 3.7409562423107914e-06,
      "loss": 0.0017,
      "step": 114210
    },
    {
      "epoch": 1.8689356131882517,
      "grad_norm": 0.07835962623357773,
      "learning_rate": 3.7400349994045414e-06,
      "loss": 0.0021,
      "step": 114220
    },
    {
      "epoch": 1.8690992391393275,
      "grad_norm": 0.410089910030365,
      "learning_rate": 3.7391138021689943e-06,
      "loss": 0.0014,
      "step": 114230
    },
    {
      "epoch": 1.8692628650904033,
      "grad_norm": 0.1445467323064804,
      "learning_rate": 3.7381926506375417e-06,
      "loss": 0.0017,
      "step": 114240
    },
    {
      "epoch": 1.8694264910414793,
      "grad_norm": 0.08186060935258865,
      "learning_rate": 3.7372715448435727e-06,
      "loss": 0.001,
      "step": 114250
    },
    {
      "epoch": 1.869590116992555,
      "grad_norm": 0.08139314502477646,
      "learning_rate": 3.7363504848204762e-06,
      "loss": 0.001,
      "step": 114260
    },
    {
      "epoch": 1.8697537429436308,
      "grad_norm": 0.2628469467163086,
      "learning_rate": 3.735429470601637e-06,
      "loss": 0.0018,
      "step": 114270
    },
    {
      "epoch": 1.8699173688947068,
      "grad_norm": 0.004719309974461794,
      "learning_rate": 3.7345085022204396e-06,
      "loss": 0.0006,
      "step": 114280
    },
    {
      "epoch": 1.8700809948457824,
      "grad_norm": 0.1432575285434723,
      "learning_rate": 3.7335875797102673e-06,
      "loss": 0.0032,
      "step": 114290
    },
    {
      "epoch": 1.8702446207968584,
      "grad_norm": 0.004074563272297382,
      "learning_rate": 3.7326667031045017e-06,
      "loss": 0.0014,
      "step": 114300
    },
    {
      "epoch": 1.8704082467479344,
      "grad_norm": 0.005161906126886606,
      "learning_rate": 3.731745872436523e-06,
      "loss": 0.001,
      "step": 114310
    },
    {
      "epoch": 1.87057187269901,
      "grad_norm": 0.10010579973459244,
      "learning_rate": 3.730825087739706e-06,
      "loss": 0.0037,
      "step": 114320
    },
    {
      "epoch": 1.870735498650086,
      "grad_norm": 0.0828857570886612,
      "learning_rate": 3.7299043490474305e-06,
      "loss": 0.0015,
      "step": 114330
    },
    {
      "epoch": 1.8708991246011617,
      "grad_norm": 0.09110900014638901,
      "learning_rate": 3.7289836563930705e-06,
      "loss": 0.0018,
      "step": 114340
    },
    {
      "epoch": 1.8710627505522375,
      "grad_norm": 0.20596274733543396,
      "learning_rate": 3.728063009809997e-06,
      "loss": 0.0012,
      "step": 114350
    },
    {
      "epoch": 1.8712263765033135,
      "grad_norm": 0.05929534137248993,
      "learning_rate": 3.7271424093315857e-06,
      "loss": 0.0009,
      "step": 114360
    },
    {
      "epoch": 1.8713900024543892,
      "grad_norm": 0.04741894453763962,
      "learning_rate": 3.7262218549912e-06,
      "loss": 0.0019,
      "step": 114370
    },
    {
      "epoch": 1.871553628405465,
      "grad_norm": 0.10395877063274384,
      "learning_rate": 3.72530134682221e-06,
      "loss": 0.0016,
      "step": 114380
    },
    {
      "epoch": 1.871717254356541,
      "grad_norm": 0.1305025815963745,
      "learning_rate": 3.7243808848579834e-06,
      "loss": 0.0017,
      "step": 114390
    },
    {
      "epoch": 1.8718808803076168,
      "grad_norm": 0.0972873792052269,
      "learning_rate": 3.723460469131883e-06,
      "loss": 0.0012,
      "step": 114400
    },
    {
      "epoch": 1.8720445062586926,
      "grad_norm": 0.29801979660987854,
      "learning_rate": 3.722540099677273e-06,
      "loss": 0.002,
      "step": 114410
    },
    {
      "epoch": 1.8722081322097686,
      "grad_norm": 0.122939832508564,
      "learning_rate": 3.7216197765275146e-06,
      "loss": 0.0019,
      "step": 114420
    },
    {
      "epoch": 1.8723717581608443,
      "grad_norm": 0.1650342494249344,
      "learning_rate": 3.7206994997159667e-06,
      "loss": 0.0017,
      "step": 114430
    },
    {
      "epoch": 1.87253538411192,
      "grad_norm": 0.01189844124019146,
      "learning_rate": 3.719779269275987e-06,
      "loss": 0.0007,
      "step": 114440
    },
    {
      "epoch": 1.872699010062996,
      "grad_norm": 0.049906808882951736,
      "learning_rate": 3.7188590852409324e-06,
      "loss": 0.0052,
      "step": 114450
    },
    {
      "epoch": 1.8728626360140719,
      "grad_norm": 0.191539004445076,
      "learning_rate": 3.717938947644157e-06,
      "loss": 0.0015,
      "step": 114460
    },
    {
      "epoch": 1.8730262619651477,
      "grad_norm": 0.08642210066318512,
      "learning_rate": 3.717018856519012e-06,
      "loss": 0.001,
      "step": 114470
    },
    {
      "epoch": 1.8731898879162237,
      "grad_norm": 0.07195541262626648,
      "learning_rate": 3.716098811898851e-06,
      "loss": 0.0029,
      "step": 114480
    },
    {
      "epoch": 1.8733535138672992,
      "grad_norm": 0.1362007111310959,
      "learning_rate": 3.715178813817022e-06,
      "loss": 0.0015,
      "step": 114490
    },
    {
      "epoch": 1.8735171398183752,
      "grad_norm": 0.09035487473011017,
      "learning_rate": 3.714258862306873e-06,
      "loss": 0.0062,
      "step": 114500
    },
    {
      "epoch": 1.8736807657694512,
      "grad_norm": 0.07487788051366806,
      "learning_rate": 3.7133389574017497e-06,
      "loss": 0.0009,
      "step": 114510
    },
    {
      "epoch": 1.8738443917205267,
      "grad_norm": 0.046876538544893265,
      "learning_rate": 3.7124190991349973e-06,
      "loss": 0.0011,
      "step": 114520
    },
    {
      "epoch": 1.8740080176716027,
      "grad_norm": 0.008197380229830742,
      "learning_rate": 3.711499287539958e-06,
      "loss": 0.0011,
      "step": 114530
    },
    {
      "epoch": 1.8741716436226785,
      "grad_norm": 0.15889716148376465,
      "learning_rate": 3.710579522649971e-06,
      "loss": 0.0034,
      "step": 114540
    },
    {
      "epoch": 1.8743352695737543,
      "grad_norm": 0.016329629346728325,
      "learning_rate": 3.709659804498379e-06,
      "loss": 0.0029,
      "step": 114550
    },
    {
      "epoch": 1.8744988955248303,
      "grad_norm": 0.11683212220668793,
      "learning_rate": 3.708740133118519e-06,
      "loss": 0.0014,
      "step": 114560
    },
    {
      "epoch": 1.874662521475906,
      "grad_norm": 0.02318652905523777,
      "learning_rate": 3.7078205085437246e-06,
      "loss": 0.0014,
      "step": 114570
    },
    {
      "epoch": 1.8748261474269818,
      "grad_norm": 0.01382275577634573,
      "learning_rate": 3.7069009308073306e-06,
      "loss": 0.0014,
      "step": 114580
    },
    {
      "epoch": 1.8749897733780578,
      "grad_norm": 0.05621064826846123,
      "learning_rate": 3.70598139994267e-06,
      "loss": 0.0015,
      "step": 114590
    },
    {
      "epoch": 1.8751533993291336,
      "grad_norm": 0.0998888686299324,
      "learning_rate": 3.7050619159830727e-06,
      "loss": 0.0014,
      "step": 114600
    },
    {
      "epoch": 1.8753170252802094,
      "grad_norm": 0.10172632336616516,
      "learning_rate": 3.70414247896187e-06,
      "loss": 0.0015,
      "step": 114610
    },
    {
      "epoch": 1.8754806512312854,
      "grad_norm": 0.08705252408981323,
      "learning_rate": 3.7032230889123865e-06,
      "loss": 0.0013,
      "step": 114620
    },
    {
      "epoch": 1.8756442771823612,
      "grad_norm": 0.03748011961579323,
      "learning_rate": 3.7023037458679505e-06,
      "loss": 0.0017,
      "step": 114630
    },
    {
      "epoch": 1.875807903133437,
      "grad_norm": 0.1621546745300293,
      "learning_rate": 3.7013844498618845e-06,
      "loss": 0.0012,
      "step": 114640
    },
    {
      "epoch": 1.875971529084513,
      "grad_norm": 0.07121383398771286,
      "learning_rate": 3.700465200927511e-06,
      "loss": 0.0011,
      "step": 114650
    },
    {
      "epoch": 1.8761351550355887,
      "grad_norm": 0.058320626616477966,
      "learning_rate": 3.699545999098152e-06,
      "loss": 0.0008,
      "step": 114660
    },
    {
      "epoch": 1.8762987809866645,
      "grad_norm": 0.09125392138957977,
      "learning_rate": 3.698626844407124e-06,
      "loss": 0.0008,
      "step": 114670
    },
    {
      "epoch": 1.8764624069377405,
      "grad_norm": 0.1257014125585556,
      "learning_rate": 3.6977077368877446e-06,
      "loss": 0.0014,
      "step": 114680
    },
    {
      "epoch": 1.876626032888816,
      "grad_norm": 0.1659439206123352,
      "learning_rate": 3.69678867657333e-06,
      "loss": 0.0015,
      "step": 114690
    },
    {
      "epoch": 1.876789658839892,
      "grad_norm": 0.03649391233921051,
      "learning_rate": 3.6958696634971937e-06,
      "loss": 0.0017,
      "step": 114700
    },
    {
      "epoch": 1.876953284790968,
      "grad_norm": 0.15149487555027008,
      "learning_rate": 3.6949506976926476e-06,
      "loss": 0.0015,
      "step": 114710
    },
    {
      "epoch": 1.8771169107420436,
      "grad_norm": 0.04018774628639221,
      "learning_rate": 3.6940317791930027e-06,
      "loss": 0.0011,
      "step": 114720
    },
    {
      "epoch": 1.8772805366931196,
      "grad_norm": 0.08447539806365967,
      "learning_rate": 3.6931129080315665e-06,
      "loss": 0.001,
      "step": 114730
    },
    {
      "epoch": 1.8774441626441953,
      "grad_norm": 0.06348925828933716,
      "learning_rate": 3.692194084241646e-06,
      "loss": 0.0013,
      "step": 114740
    },
    {
      "epoch": 1.877607788595271,
      "grad_norm": 0.045159969478845596,
      "learning_rate": 3.691275307856547e-06,
      "loss": 0.0012,
      "step": 114750
    },
    {
      "epoch": 1.877771414546347,
      "grad_norm": 0.02692187763750553,
      "learning_rate": 3.690356578909574e-06,
      "loss": 0.0006,
      "step": 114760
    },
    {
      "epoch": 1.8779350404974229,
      "grad_norm": 0.15412628650665283,
      "learning_rate": 3.689437897434026e-06,
      "loss": 0.0026,
      "step": 114770
    },
    {
      "epoch": 1.8780986664484987,
      "grad_norm": 0.5776950716972351,
      "learning_rate": 3.688519263463204e-06,
      "loss": 0.0013,
      "step": 114780
    },
    {
      "epoch": 1.8782622923995747,
      "grad_norm": 0.07777441293001175,
      "learning_rate": 3.6876006770304075e-06,
      "loss": 0.0013,
      "step": 114790
    },
    {
      "epoch": 1.8784259183506504,
      "grad_norm": 0.13046696782112122,
      "learning_rate": 3.6866821381689314e-06,
      "loss": 0.0016,
      "step": 114800
    },
    {
      "epoch": 1.8785895443017262,
      "grad_norm": 0.2156224399805069,
      "learning_rate": 3.6857636469120715e-06,
      "loss": 0.0019,
      "step": 114810
    },
    {
      "epoch": 1.8787531702528022,
      "grad_norm": 0.07816867530345917,
      "learning_rate": 3.6848452032931204e-06,
      "loss": 0.0013,
      "step": 114820
    },
    {
      "epoch": 1.878916796203878,
      "grad_norm": 0.027057237923145294,
      "learning_rate": 3.6839268073453704e-06,
      "loss": 0.0012,
      "step": 114830
    },
    {
      "epoch": 1.8790804221549537,
      "grad_norm": 0.016057878732681274,
      "learning_rate": 3.68300845910211e-06,
      "loss": 0.0017,
      "step": 114840
    },
    {
      "epoch": 1.8792440481060297,
      "grad_norm": 0.07097262144088745,
      "learning_rate": 3.6820901585966284e-06,
      "loss": 0.002,
      "step": 114850
    },
    {
      "epoch": 1.8794076740571053,
      "grad_norm": 0.13552403450012207,
      "learning_rate": 3.681171905862213e-06,
      "loss": 0.0012,
      "step": 114860
    },
    {
      "epoch": 1.8795713000081813,
      "grad_norm": 0.45568346977233887,
      "learning_rate": 3.6802537009321427e-06,
      "loss": 0.003,
      "step": 114870
    },
    {
      "epoch": 1.8797349259592573,
      "grad_norm": 0.08118440955877304,
      "learning_rate": 3.6793355438397048e-06,
      "loss": 0.0016,
      "step": 114880
    },
    {
      "epoch": 1.8798985519103328,
      "grad_norm": 0.08286190032958984,
      "learning_rate": 3.67841743461818e-06,
      "loss": 0.0012,
      "step": 114890
    },
    {
      "epoch": 1.8800621778614088,
      "grad_norm": 0.005519147962331772,
      "learning_rate": 3.6774993733008463e-06,
      "loss": 0.001,
      "step": 114900
    },
    {
      "epoch": 1.8802258038124846,
      "grad_norm": 0.005147336516529322,
      "learning_rate": 3.6765813599209822e-06,
      "loss": 0.001,
      "step": 114910
    },
    {
      "epoch": 1.8803894297635604,
      "grad_norm": 0.03634466975927353,
      "learning_rate": 3.6756633945118626e-06,
      "loss": 0.0007,
      "step": 114920
    },
    {
      "epoch": 1.8805530557146364,
      "grad_norm": 0.14535662531852722,
      "learning_rate": 3.6747454771067624e-06,
      "loss": 0.0016,
      "step": 114930
    },
    {
      "epoch": 1.8807166816657122,
      "grad_norm": 0.11837170273065567,
      "learning_rate": 3.6738276077389534e-06,
      "loss": 0.0025,
      "step": 114940
    },
    {
      "epoch": 1.880880307616788,
      "grad_norm": 0.04300757870078087,
      "learning_rate": 3.672909786441706e-06,
      "loss": 0.0012,
      "step": 114950
    },
    {
      "epoch": 1.881043933567864,
      "grad_norm": 0.052255939692258835,
      "learning_rate": 3.6719920132482902e-06,
      "loss": 0.0007,
      "step": 114960
    },
    {
      "epoch": 1.8812075595189397,
      "grad_norm": 0.13350211083889008,
      "learning_rate": 3.6710742881919713e-06,
      "loss": 0.001,
      "step": 114970
    },
    {
      "epoch": 1.8813711854700155,
      "grad_norm": 0.04611939191818237,
      "learning_rate": 3.6701566113060155e-06,
      "loss": 0.0013,
      "step": 114980
    },
    {
      "epoch": 1.8815348114210915,
      "grad_norm": 0.017334578558802605,
      "learning_rate": 3.669238982623686e-06,
      "loss": 0.0011,
      "step": 114990
    },
    {
      "epoch": 1.8816984373721672,
      "grad_norm": 0.17697244882583618,
      "learning_rate": 3.6683214021782454e-06,
      "loss": 0.0011,
      "step": 115000
    },
    {
      "epoch": 1.881862063323243,
      "grad_norm": 0.07738719880580902,
      "learning_rate": 3.6674038700029537e-06,
      "loss": 0.0009,
      "step": 115010
    },
    {
      "epoch": 1.882025689274319,
      "grad_norm": 0.10444359481334686,
      "learning_rate": 3.6664863861310685e-06,
      "loss": 0.0027,
      "step": 115020
    },
    {
      "epoch": 1.8821893152253948,
      "grad_norm": 0.08220594376325607,
      "learning_rate": 3.6655689505958457e-06,
      "loss": 0.0012,
      "step": 115030
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 0.08716414123773575,
      "learning_rate": 3.6646515634305426e-06,
      "loss": 0.0019,
      "step": 115040
    },
    {
      "epoch": 1.8825165671275466,
      "grad_norm": 0.03188503533601761,
      "learning_rate": 3.6637342246684104e-06,
      "loss": 0.0008,
      "step": 115050
    },
    {
      "epoch": 1.882680193078622,
      "grad_norm": 0.0638166293501854,
      "learning_rate": 3.6628169343427033e-06,
      "loss": 0.002,
      "step": 115060
    },
    {
      "epoch": 1.882843819029698,
      "grad_norm": 0.13622552156448364,
      "learning_rate": 3.6618996924866664e-06,
      "loss": 0.003,
      "step": 115070
    },
    {
      "epoch": 1.883007444980774,
      "grad_norm": 0.03605085238814354,
      "learning_rate": 3.6609824991335495e-06,
      "loss": 0.0006,
      "step": 115080
    },
    {
      "epoch": 1.8831710709318497,
      "grad_norm": 0.06659548729658127,
      "learning_rate": 3.6600653543165974e-06,
      "loss": 0.0008,
      "step": 115090
    },
    {
      "epoch": 1.8833346968829257,
      "grad_norm": 0.02931560017168522,
      "learning_rate": 3.6591482580690575e-06,
      "loss": 0.0016,
      "step": 115100
    },
    {
      "epoch": 1.8834983228340014,
      "grad_norm": 0.1813018023967743,
      "learning_rate": 3.6582312104241703e-06,
      "loss": 0.0023,
      "step": 115110
    },
    {
      "epoch": 1.8836619487850772,
      "grad_norm": 0.021614031866192818,
      "learning_rate": 3.6573142114151767e-06,
      "loss": 0.0015,
      "step": 115120
    },
    {
      "epoch": 1.8838255747361532,
      "grad_norm": 0.011086588725447655,
      "learning_rate": 3.6563972610753156e-06,
      "loss": 0.0014,
      "step": 115130
    },
    {
      "epoch": 1.883989200687229,
      "grad_norm": 0.11071640253067017,
      "learning_rate": 3.6554803594378254e-06,
      "loss": 0.0021,
      "step": 115140
    },
    {
      "epoch": 1.8841528266383047,
      "grad_norm": 0.10829813778400421,
      "learning_rate": 3.65456350653594e-06,
      "loss": 0.0012,
      "step": 115150
    },
    {
      "epoch": 1.8843164525893807,
      "grad_norm": 0.04370303824543953,
      "learning_rate": 3.6536467024028945e-06,
      "loss": 0.0015,
      "step": 115160
    },
    {
      "epoch": 1.8844800785404565,
      "grad_norm": 0.101749949157238,
      "learning_rate": 3.6527299470719204e-06,
      "loss": 0.0015,
      "step": 115170
    },
    {
      "epoch": 1.8846437044915323,
      "grad_norm": 0.04757365584373474,
      "learning_rate": 3.6518132405762463e-06,
      "loss": 0.0005,
      "step": 115180
    },
    {
      "epoch": 1.8848073304426083,
      "grad_norm": 0.10030307620763779,
      "learning_rate": 3.6508965829491027e-06,
      "loss": 0.0008,
      "step": 115190
    },
    {
      "epoch": 1.884970956393684,
      "grad_norm": 0.16621261835098267,
      "learning_rate": 3.6499799742237156e-06,
      "loss": 0.0012,
      "step": 115200
    },
    {
      "epoch": 1.8851345823447598,
      "grad_norm": 0.08705494552850723,
      "learning_rate": 3.6490634144333094e-06,
      "loss": 0.0022,
      "step": 115210
    },
    {
      "epoch": 1.8852982082958358,
      "grad_norm": 0.08569933474063873,
      "learning_rate": 3.648146903611108e-06,
      "loss": 0.003,
      "step": 115220
    },
    {
      "epoch": 1.8854618342469116,
      "grad_norm": 0.22078154981136322,
      "learning_rate": 3.6472304417903315e-06,
      "loss": 0.0019,
      "step": 115230
    },
    {
      "epoch": 1.8856254601979874,
      "grad_norm": 0.06548316031694412,
      "learning_rate": 3.646314029004201e-06,
      "loss": 0.0017,
      "step": 115240
    },
    {
      "epoch": 1.8857890861490634,
      "grad_norm": 0.1948816180229187,
      "learning_rate": 3.6453976652859314e-06,
      "loss": 0.0023,
      "step": 115250
    },
    {
      "epoch": 1.885952712100139,
      "grad_norm": 0.029058460146188736,
      "learning_rate": 3.6444813506687426e-06,
      "loss": 0.0017,
      "step": 115260
    },
    {
      "epoch": 1.886116338051215,
      "grad_norm": 0.1261594444513321,
      "learning_rate": 3.643565085185849e-06,
      "loss": 0.0018,
      "step": 115270
    },
    {
      "epoch": 1.886279964002291,
      "grad_norm": 0.0823182463645935,
      "learning_rate": 3.642648868870459e-06,
      "loss": 0.0013,
      "step": 115280
    },
    {
      "epoch": 1.8864435899533665,
      "grad_norm": 0.04459149390459061,
      "learning_rate": 3.6417327017557848e-06,
      "loss": 0.0027,
      "step": 115290
    },
    {
      "epoch": 1.8866072159044425,
      "grad_norm": 0.008105945773422718,
      "learning_rate": 3.6408165838750355e-06,
      "loss": 0.0011,
      "step": 115300
    },
    {
      "epoch": 1.8867708418555182,
      "grad_norm": 0.05859200283885002,
      "learning_rate": 3.6399005152614176e-06,
      "loss": 0.0009,
      "step": 115310
    },
    {
      "epoch": 1.886934467806594,
      "grad_norm": 0.06281672418117523,
      "learning_rate": 3.6389844959481378e-06,
      "loss": 0.0014,
      "step": 115320
    },
    {
      "epoch": 1.88709809375767,
      "grad_norm": 0.028158392757177353,
      "learning_rate": 3.638068525968399e-06,
      "loss": 0.0012,
      "step": 115330
    },
    {
      "epoch": 1.8872617197087458,
      "grad_norm": 0.004508281126618385,
      "learning_rate": 3.6371526053554028e-06,
      "loss": 0.0011,
      "step": 115340
    },
    {
      "epoch": 1.8874253456598216,
      "grad_norm": 0.030568337067961693,
      "learning_rate": 3.6362367341423495e-06,
      "loss": 0.0011,
      "step": 115350
    },
    {
      "epoch": 1.8875889716108976,
      "grad_norm": 0.029163209721446037,
      "learning_rate": 3.6353209123624367e-06,
      "loss": 0.0008,
      "step": 115360
    },
    {
      "epoch": 1.8877525975619733,
      "grad_norm": 0.07090315222740173,
      "learning_rate": 3.6344051400488623e-06,
      "loss": 0.0012,
      "step": 115370
    },
    {
      "epoch": 1.887916223513049,
      "grad_norm": 0.1078358069062233,
      "learning_rate": 3.633489417234818e-06,
      "loss": 0.0008,
      "step": 115380
    },
    {
      "epoch": 1.888079849464125,
      "grad_norm": 0.09491295367479324,
      "learning_rate": 3.6325737439534977e-06,
      "loss": 0.0013,
      "step": 115390
    },
    {
      "epoch": 1.8882434754152009,
      "grad_norm": 0.2326422929763794,
      "learning_rate": 3.631658120238093e-06,
      "loss": 0.0023,
      "step": 115400
    },
    {
      "epoch": 1.8884071013662767,
      "grad_norm": 0.09236391633749008,
      "learning_rate": 3.6307425461217925e-06,
      "loss": 0.0017,
      "step": 115410
    },
    {
      "epoch": 1.8885707273173526,
      "grad_norm": 0.12886789441108704,
      "learning_rate": 3.629827021637784e-06,
      "loss": 0.001,
      "step": 115420
    },
    {
      "epoch": 1.8887343532684284,
      "grad_norm": 0.21762512624263763,
      "learning_rate": 3.6289115468192527e-06,
      "loss": 0.0016,
      "step": 115430
    },
    {
      "epoch": 1.8888979792195042,
      "grad_norm": 0.23086415231227875,
      "learning_rate": 3.6279961216993833e-06,
      "loss": 0.0021,
      "step": 115440
    },
    {
      "epoch": 1.8890616051705802,
      "grad_norm": 0.008762970566749573,
      "learning_rate": 3.6270807463113556e-06,
      "loss": 0.0006,
      "step": 115450
    },
    {
      "epoch": 1.8892252311216557,
      "grad_norm": 0.22121384739875793,
      "learning_rate": 3.6261654206883524e-06,
      "loss": 0.0014,
      "step": 115460
    },
    {
      "epoch": 1.8893888570727317,
      "grad_norm": 0.08388222754001617,
      "learning_rate": 3.6252501448635508e-06,
      "loss": 0.001,
      "step": 115470
    },
    {
      "epoch": 1.8895524830238077,
      "grad_norm": 0.12055619806051254,
      "learning_rate": 3.6243349188701273e-06,
      "loss": 0.0022,
      "step": 115480
    },
    {
      "epoch": 1.8897161089748833,
      "grad_norm": 0.07749462127685547,
      "learning_rate": 3.623419742741255e-06,
      "loss": 0.003,
      "step": 115490
    },
    {
      "epoch": 1.8898797349259593,
      "grad_norm": 0.057838600128889084,
      "learning_rate": 3.62250461651011e-06,
      "loss": 0.0013,
      "step": 115500
    },
    {
      "epoch": 1.890043360877035,
      "grad_norm": 0.004064687993377447,
      "learning_rate": 3.6215895402098617e-06,
      "loss": 0.0013,
      "step": 115510
    },
    {
      "epoch": 1.8902069868281108,
      "grad_norm": 0.06768014281988144,
      "learning_rate": 3.6206745138736794e-06,
      "loss": 0.0011,
      "step": 115520
    },
    {
      "epoch": 1.8903706127791868,
      "grad_norm": 0.08197381347417831,
      "learning_rate": 3.61975953753473e-06,
      "loss": 0.0011,
      "step": 115530
    },
    {
      "epoch": 1.8905342387302626,
      "grad_norm": 0.07014795392751694,
      "learning_rate": 3.6188446112261812e-06,
      "loss": 0.0014,
      "step": 115540
    },
    {
      "epoch": 1.8906978646813384,
      "grad_norm": 0.11544433981180191,
      "learning_rate": 3.6179297349811947e-06,
      "loss": 0.0014,
      "step": 115550
    },
    {
      "epoch": 1.8908614906324144,
      "grad_norm": 0.060647331178188324,
      "learning_rate": 3.6170149088329346e-06,
      "loss": 0.0025,
      "step": 115560
    },
    {
      "epoch": 1.8910251165834902,
      "grad_norm": 0.03720762953162193,
      "learning_rate": 3.6161001328145624e-06,
      "loss": 0.0011,
      "step": 115570
    },
    {
      "epoch": 1.891188742534566,
      "grad_norm": 0.07355286180973053,
      "learning_rate": 3.6151854069592317e-06,
      "loss": 0.0008,
      "step": 115580
    },
    {
      "epoch": 1.891352368485642,
      "grad_norm": 0.11242623627185822,
      "learning_rate": 3.6142707313001e-06,
      "loss": 0.0028,
      "step": 115590
    },
    {
      "epoch": 1.8915159944367177,
      "grad_norm": 0.12988241016864777,
      "learning_rate": 3.613356105870326e-06,
      "loss": 0.0017,
      "step": 115600
    },
    {
      "epoch": 1.8916796203877935,
      "grad_norm": 0.07360272854566574,
      "learning_rate": 3.6124415307030602e-06,
      "loss": 0.0012,
      "step": 115610
    },
    {
      "epoch": 1.8918432463388695,
      "grad_norm": 0.12990425527095795,
      "learning_rate": 3.611527005831453e-06,
      "loss": 0.0018,
      "step": 115620
    },
    {
      "epoch": 1.8920068722899452,
      "grad_norm": 0.11665112525224686,
      "learning_rate": 3.6106125312886566e-06,
      "loss": 0.0015,
      "step": 115630
    },
    {
      "epoch": 1.892170498241021,
      "grad_norm": 0.1733156144618988,
      "learning_rate": 3.6096981071078154e-06,
      "loss": 0.0023,
      "step": 115640
    },
    {
      "epoch": 1.892334124192097,
      "grad_norm": 0.13508737087249756,
      "learning_rate": 3.6087837333220765e-06,
      "loss": 0.0015,
      "step": 115650
    },
    {
      "epoch": 1.8924977501431726,
      "grad_norm": 0.15719018876552582,
      "learning_rate": 3.6078694099645834e-06,
      "loss": 0.002,
      "step": 115660
    },
    {
      "epoch": 1.8926613760942486,
      "grad_norm": 0.06254814565181732,
      "learning_rate": 3.60695513706848e-06,
      "loss": 0.0007,
      "step": 115670
    },
    {
      "epoch": 1.8928250020453243,
      "grad_norm": 0.22196121513843536,
      "learning_rate": 3.6060409146669028e-06,
      "loss": 0.0018,
      "step": 115680
    },
    {
      "epoch": 1.8929886279964,
      "grad_norm": 0.12035829573869705,
      "learning_rate": 3.6051267427929925e-06,
      "loss": 0.0012,
      "step": 115690
    },
    {
      "epoch": 1.893152253947476,
      "grad_norm": 0.05168607085943222,
      "learning_rate": 3.604212621479885e-06,
      "loss": 0.0008,
      "step": 115700
    },
    {
      "epoch": 1.8933158798985519,
      "grad_norm": 0.002085619606077671,
      "learning_rate": 3.6032985507607155e-06,
      "loss": 0.0013,
      "step": 115710
    },
    {
      "epoch": 1.8934795058496277,
      "grad_norm": 0.029281770810484886,
      "learning_rate": 3.602384530668617e-06,
      "loss": 0.0022,
      "step": 115720
    },
    {
      "epoch": 1.8936431318007036,
      "grad_norm": 0.44876983761787415,
      "learning_rate": 3.6014705612367203e-06,
      "loss": 0.0023,
      "step": 115730
    },
    {
      "epoch": 1.8938067577517794,
      "grad_norm": 0.024062059819698334,
      "learning_rate": 3.600556642498154e-06,
      "loss": 0.0014,
      "step": 115740
    },
    {
      "epoch": 1.8939703837028552,
      "grad_norm": 0.06086837127804756,
      "learning_rate": 3.599642774486045e-06,
      "loss": 0.001,
      "step": 115750
    },
    {
      "epoch": 1.8941340096539312,
      "grad_norm": 0.058834198862314224,
      "learning_rate": 3.5987289572335205e-06,
      "loss": 0.0013,
      "step": 115760
    },
    {
      "epoch": 1.894297635605007,
      "grad_norm": 0.07535523921251297,
      "learning_rate": 3.5978151907737057e-06,
      "loss": 0.0015,
      "step": 115770
    },
    {
      "epoch": 1.8944612615560827,
      "grad_norm": 0.09088030457496643,
      "learning_rate": 3.5969014751397187e-06,
      "loss": 0.0008,
      "step": 115780
    },
    {
      "epoch": 1.8946248875071587,
      "grad_norm": 0.09978930652141571,
      "learning_rate": 3.5959878103646807e-06,
      "loss": 0.0017,
      "step": 115790
    },
    {
      "epoch": 1.8947885134582345,
      "grad_norm": 0.06307248771190643,
      "learning_rate": 3.5950741964817094e-06,
      "loss": 0.0011,
      "step": 115800
    },
    {
      "epoch": 1.8949521394093103,
      "grad_norm": 0.05008325353264809,
      "learning_rate": 3.5941606335239217e-06,
      "loss": 0.0014,
      "step": 115810
    },
    {
      "epoch": 1.8951157653603863,
      "grad_norm": 0.04925980791449547,
      "learning_rate": 3.5932471215244323e-06,
      "loss": 0.0011,
      "step": 115820
    },
    {
      "epoch": 1.8952793913114618,
      "grad_norm": 0.0952456146478653,
      "learning_rate": 3.592333660516355e-06,
      "loss": 0.0009,
      "step": 115830
    },
    {
      "epoch": 1.8954430172625378,
      "grad_norm": 0.052398569881916046,
      "learning_rate": 3.5914202505327977e-06,
      "loss": 0.0012,
      "step": 115840
    },
    {
      "epoch": 1.8956066432136138,
      "grad_norm": 0.07709585875272751,
      "learning_rate": 3.590506891606872e-06,
      "loss": 0.001,
      "step": 115850
    },
    {
      "epoch": 1.8957702691646894,
      "grad_norm": 0.19080610573291779,
      "learning_rate": 3.5895935837716835e-06,
      "loss": 0.0022,
      "step": 115860
    },
    {
      "epoch": 1.8959338951157654,
      "grad_norm": 0.10824272036552429,
      "learning_rate": 3.5886803270603388e-06,
      "loss": 0.0016,
      "step": 115870
    },
    {
      "epoch": 1.8960975210668412,
      "grad_norm": 0.1646525114774704,
      "learning_rate": 3.5877671215059394e-06,
      "loss": 0.0011,
      "step": 115880
    },
    {
      "epoch": 1.896261147017917,
      "grad_norm": 0.07289395481348038,
      "learning_rate": 3.5868539671415874e-06,
      "loss": 0.0012,
      "step": 115890
    },
    {
      "epoch": 1.896424772968993,
      "grad_norm": 0.08073840290307999,
      "learning_rate": 3.5859408640003827e-06,
      "loss": 0.0019,
      "step": 115900
    },
    {
      "epoch": 1.8965883989200687,
      "grad_norm": 0.018072491511702538,
      "learning_rate": 3.5850278121154236e-06,
      "loss": 0.0016,
      "step": 115910
    },
    {
      "epoch": 1.8967520248711445,
      "grad_norm": 0.2929452955722809,
      "learning_rate": 3.584114811519805e-06,
      "loss": 0.0038,
      "step": 115920
    },
    {
      "epoch": 1.8969156508222205,
      "grad_norm": 0.03789547085762024,
      "learning_rate": 3.5832018622466213e-06,
      "loss": 0.0016,
      "step": 115930
    },
    {
      "epoch": 1.8970792767732962,
      "grad_norm": 0.04028784856200218,
      "learning_rate": 3.582288964328965e-06,
      "loss": 0.0011,
      "step": 115940
    },
    {
      "epoch": 1.897242902724372,
      "grad_norm": 0.18984396755695343,
      "learning_rate": 3.5813761177999267e-06,
      "loss": 0.0014,
      "step": 115950
    },
    {
      "epoch": 1.897406528675448,
      "grad_norm": 0.04057304561138153,
      "learning_rate": 3.5804633226925943e-06,
      "loss": 0.0012,
      "step": 115960
    },
    {
      "epoch": 1.8975701546265238,
      "grad_norm": 0.0909348577260971,
      "learning_rate": 3.5795505790400557e-06,
      "loss": 0.0008,
      "step": 115970
    },
    {
      "epoch": 1.8977337805775996,
      "grad_norm": 0.020928556099534035,
      "learning_rate": 3.578637886875394e-06,
      "loss": 0.0014,
      "step": 115980
    },
    {
      "epoch": 1.8978974065286756,
      "grad_norm": 0.048062387853860855,
      "learning_rate": 3.5777252462316927e-06,
      "loss": 0.001,
      "step": 115990
    },
    {
      "epoch": 1.8980610324797513,
      "grad_norm": 0.10723450779914856,
      "learning_rate": 3.5768126571420325e-06,
      "loss": 0.0017,
      "step": 116000
    },
    {
      "epoch": 1.898224658430827,
      "grad_norm": 0.019371047616004944,
      "learning_rate": 3.5759001196394926e-06,
      "loss": 0.0039,
      "step": 116010
    },
    {
      "epoch": 1.898388284381903,
      "grad_norm": 0.07669398933649063,
      "learning_rate": 3.5749876337571513e-06,
      "loss": 0.002,
      "step": 116020
    },
    {
      "epoch": 1.8985519103329787,
      "grad_norm": 0.04043760895729065,
      "learning_rate": 3.574075199528082e-06,
      "loss": 0.0014,
      "step": 116030
    },
    {
      "epoch": 1.8987155362840547,
      "grad_norm": 0.09259256720542908,
      "learning_rate": 3.573162816985361e-06,
      "loss": 0.0016,
      "step": 116040
    },
    {
      "epoch": 1.8988791622351306,
      "grad_norm": 0.04458944872021675,
      "learning_rate": 3.5722504861620584e-06,
      "loss": 0.0034,
      "step": 116050
    },
    {
      "epoch": 1.8990427881862062,
      "grad_norm": 0.07175411283969879,
      "learning_rate": 3.571338207091244e-06,
      "loss": 0.0036,
      "step": 116060
    },
    {
      "epoch": 1.8992064141372822,
      "grad_norm": 0.16684776544570923,
      "learning_rate": 3.5704259798059883e-06,
      "loss": 0.0023,
      "step": 116070
    },
    {
      "epoch": 1.899370040088358,
      "grad_norm": 0.06403648853302002,
      "learning_rate": 3.5695138043393514e-06,
      "loss": 0.0031,
      "step": 116080
    },
    {
      "epoch": 1.8995336660394337,
      "grad_norm": 0.29816925525665283,
      "learning_rate": 3.5686016807244022e-06,
      "loss": 0.0021,
      "step": 116090
    },
    {
      "epoch": 1.8996972919905097,
      "grad_norm": 0.09877566993236542,
      "learning_rate": 3.5676896089942013e-06,
      "loss": 0.0023,
      "step": 116100
    },
    {
      "epoch": 1.8998609179415855,
      "grad_norm": 0.07043410837650299,
      "learning_rate": 3.5667775891818108e-06,
      "loss": 0.0014,
      "step": 116110
    },
    {
      "epoch": 1.9000245438926613,
      "grad_norm": 0.10018762946128845,
      "learning_rate": 3.565865621320287e-06,
      "loss": 0.0016,
      "step": 116120
    },
    {
      "epoch": 1.9001881698437373,
      "grad_norm": 0.11627231538295746,
      "learning_rate": 3.5649537054426885e-06,
      "loss": 0.0017,
      "step": 116130
    },
    {
      "epoch": 1.900351795794813,
      "grad_norm": 0.04538247734308243,
      "learning_rate": 3.564041841582068e-06,
      "loss": 0.0009,
      "step": 116140
    },
    {
      "epoch": 1.9005154217458888,
      "grad_norm": 0.0025490394327789545,
      "learning_rate": 3.5631300297714804e-06,
      "loss": 0.0018,
      "step": 116150
    },
    {
      "epoch": 1.9006790476969648,
      "grad_norm": 0.14060045778751373,
      "learning_rate": 3.562218270043976e-06,
      "loss": 0.0015,
      "step": 116160
    },
    {
      "epoch": 1.9008426736480406,
      "grad_norm": 0.02201317623257637,
      "learning_rate": 3.561306562432604e-06,
      "loss": 0.0015,
      "step": 116170
    },
    {
      "epoch": 1.9010062995991164,
      "grad_norm": 0.11906857043504715,
      "learning_rate": 3.5603949069704103e-06,
      "loss": 0.0014,
      "step": 116180
    },
    {
      "epoch": 1.9011699255501924,
      "grad_norm": 0.1848725825548172,
      "learning_rate": 3.559483303690441e-06,
      "loss": 0.0012,
      "step": 116190
    },
    {
      "epoch": 1.9013335515012681,
      "grad_norm": 0.09553230553865433,
      "learning_rate": 3.5585717526257402e-06,
      "loss": 0.0009,
      "step": 116200
    },
    {
      "epoch": 1.901497177452344,
      "grad_norm": 0.14421942830085754,
      "learning_rate": 3.5576602538093493e-06,
      "loss": 0.0029,
      "step": 116210
    },
    {
      "epoch": 1.90166080340342,
      "grad_norm": 0.009725402109324932,
      "learning_rate": 3.5567488072743073e-06,
      "loss": 0.0007,
      "step": 116220
    },
    {
      "epoch": 1.9018244293544955,
      "grad_norm": 0.20262233912944794,
      "learning_rate": 3.5558374130536523e-06,
      "loss": 0.0015,
      "step": 116230
    },
    {
      "epoch": 1.9019880553055715,
      "grad_norm": 0.1150551363825798,
      "learning_rate": 3.554926071180419e-06,
      "loss": 0.0067,
      "step": 116240
    },
    {
      "epoch": 1.9021516812566475,
      "grad_norm": 0.03977881371974945,
      "learning_rate": 3.5540147816876437e-06,
      "loss": 0.0011,
      "step": 116250
    },
    {
      "epoch": 1.902315307207723,
      "grad_norm": 0.06885017454624176,
      "learning_rate": 3.553103544608357e-06,
      "loss": 0.0015,
      "step": 116260
    },
    {
      "epoch": 1.902478933158799,
      "grad_norm": 0.033356595784425735,
      "learning_rate": 3.5521923599755913e-06,
      "loss": 0.0008,
      "step": 116270
    },
    {
      "epoch": 1.9026425591098748,
      "grad_norm": 0.01597157120704651,
      "learning_rate": 3.5512812278223707e-06,
      "loss": 0.0008,
      "step": 116280
    },
    {
      "epoch": 1.9028061850609506,
      "grad_norm": 0.12661388516426086,
      "learning_rate": 3.550370148181724e-06,
      "loss": 0.0019,
      "step": 116290
    },
    {
      "epoch": 1.9029698110120266,
      "grad_norm": 0.11293073743581772,
      "learning_rate": 3.5494591210866742e-06,
      "loss": 0.001,
      "step": 116300
    },
    {
      "epoch": 1.9031334369631023,
      "grad_norm": 0.11791744083166122,
      "learning_rate": 3.548548146570246e-06,
      "loss": 0.002,
      "step": 116310
    },
    {
      "epoch": 1.903297062914178,
      "grad_norm": 0.01586984470486641,
      "learning_rate": 3.547637224665459e-06,
      "loss": 0.0013,
      "step": 116320
    },
    {
      "epoch": 1.903460688865254,
      "grad_norm": 0.16123881936073303,
      "learning_rate": 3.5467263554053322e-06,
      "loss": 0.0028,
      "step": 116330
    },
    {
      "epoch": 1.9036243148163299,
      "grad_norm": 0.011089896783232689,
      "learning_rate": 3.5458155388228817e-06,
      "loss": 0.0011,
      "step": 116340
    },
    {
      "epoch": 1.9037879407674057,
      "grad_norm": 0.07788226753473282,
      "learning_rate": 3.5449047749511234e-06,
      "loss": 0.001,
      "step": 116350
    },
    {
      "epoch": 1.9039515667184816,
      "grad_norm": 0.0635981559753418,
      "learning_rate": 3.543994063823069e-06,
      "loss": 0.0009,
      "step": 116360
    },
    {
      "epoch": 1.9041151926695574,
      "grad_norm": 0.03099675104022026,
      "learning_rate": 3.5430834054717324e-06,
      "loss": 0.0008,
      "step": 116370
    },
    {
      "epoch": 1.9042788186206332,
      "grad_norm": 0.08618183434009552,
      "learning_rate": 3.5421727999301187e-06,
      "loss": 0.0007,
      "step": 116380
    },
    {
      "epoch": 1.9044424445717092,
      "grad_norm": 0.00502800801768899,
      "learning_rate": 3.5412622472312373e-06,
      "loss": 0.001,
      "step": 116390
    },
    {
      "epoch": 1.904606070522785,
      "grad_norm": 0.03535205498337746,
      "learning_rate": 3.540351747408094e-06,
      "loss": 0.0023,
      "step": 116400
    },
    {
      "epoch": 1.9047696964738607,
      "grad_norm": 0.0633096769452095,
      "learning_rate": 3.5394413004936917e-06,
      "loss": 0.0011,
      "step": 116410
    },
    {
      "epoch": 1.9049333224249367,
      "grad_norm": 0.05483037233352661,
      "learning_rate": 3.5385309065210317e-06,
      "loss": 0.0011,
      "step": 116420
    },
    {
      "epoch": 1.9050969483760123,
      "grad_norm": 0.1297806203365326,
      "learning_rate": 3.5376205655231138e-06,
      "loss": 0.0018,
      "step": 116430
    },
    {
      "epoch": 1.9052605743270883,
      "grad_norm": 0.03207574039697647,
      "learning_rate": 3.536710277532936e-06,
      "loss": 0.0013,
      "step": 116440
    },
    {
      "epoch": 1.9054242002781643,
      "grad_norm": 0.052945345640182495,
      "learning_rate": 3.535800042583493e-06,
      "loss": 0.0025,
      "step": 116450
    },
    {
      "epoch": 1.9055878262292398,
      "grad_norm": 0.09225403517484665,
      "learning_rate": 3.5348898607077785e-06,
      "loss": 0.0019,
      "step": 116460
    },
    {
      "epoch": 1.9057514521803158,
      "grad_norm": 0.14315487444400787,
      "learning_rate": 3.5339797319387893e-06,
      "loss": 0.0018,
      "step": 116470
    },
    {
      "epoch": 1.9059150781313916,
      "grad_norm": 0.16204451024532318,
      "learning_rate": 3.533069656309508e-06,
      "loss": 0.0033,
      "step": 116480
    },
    {
      "epoch": 1.9060787040824674,
      "grad_norm": 0.06964494287967682,
      "learning_rate": 3.532159633852927e-06,
      "loss": 0.0015,
      "step": 116490
    },
    {
      "epoch": 1.9062423300335434,
      "grad_norm": 0.12034778296947479,
      "learning_rate": 3.5312496646020307e-06,
      "loss": 0.0017,
      "step": 116500
    },
    {
      "epoch": 1.9064059559846191,
      "grad_norm": 0.11584047973155975,
      "learning_rate": 3.530339748589804e-06,
      "loss": 0.0013,
      "step": 116510
    },
    {
      "epoch": 1.906569581935695,
      "grad_norm": 0.045869264751672745,
      "learning_rate": 3.5294298858492284e-06,
      "loss": 0.0006,
      "step": 116520
    },
    {
      "epoch": 1.906733207886771,
      "grad_norm": 0.15335826575756073,
      "learning_rate": 3.5285200764132866e-06,
      "loss": 0.0014,
      "step": 116530
    },
    {
      "epoch": 1.9068968338378467,
      "grad_norm": 0.14240601658821106,
      "learning_rate": 3.527610320314955e-06,
      "loss": 0.0009,
      "step": 116540
    },
    {
      "epoch": 1.9070604597889225,
      "grad_norm": 0.07842226326465607,
      "learning_rate": 3.52670061758721e-06,
      "loss": 0.0024,
      "step": 116550
    },
    {
      "epoch": 1.9072240857399985,
      "grad_norm": 0.03699515014886856,
      "learning_rate": 3.5257909682630266e-06,
      "loss": 0.0014,
      "step": 116560
    },
    {
      "epoch": 1.9073877116910742,
      "grad_norm": 0.06172734871506691,
      "learning_rate": 3.52488137237538e-06,
      "loss": 0.0005,
      "step": 116570
    },
    {
      "epoch": 1.90755133764215,
      "grad_norm": 0.004461097065359354,
      "learning_rate": 3.523971829957235e-06,
      "loss": 0.0012,
      "step": 116580
    },
    {
      "epoch": 1.907714963593226,
      "grad_norm": 0.05300891399383545,
      "learning_rate": 3.523062341041565e-06,
      "loss": 0.0015,
      "step": 116590
    },
    {
      "epoch": 1.9078785895443016,
      "grad_norm": 0.05722389370203018,
      "learning_rate": 3.5221529056613346e-06,
      "loss": 0.0016,
      "step": 116600
    },
    {
      "epoch": 1.9080422154953776,
      "grad_norm": 0.16748911142349243,
      "learning_rate": 3.5212435238495104e-06,
      "loss": 0.0016,
      "step": 116610
    },
    {
      "epoch": 1.9082058414464536,
      "grad_norm": 0.3486364185810089,
      "learning_rate": 3.5203341956390534e-06,
      "loss": 0.0012,
      "step": 116620
    },
    {
      "epoch": 1.908369467397529,
      "grad_norm": 0.2486005425453186,
      "learning_rate": 3.5194249210629263e-06,
      "loss": 0.0023,
      "step": 116630
    },
    {
      "epoch": 1.908533093348605,
      "grad_norm": 0.06751266121864319,
      "learning_rate": 3.518515700154087e-06,
      "loss": 0.0016,
      "step": 116640
    },
    {
      "epoch": 1.9086967192996809,
      "grad_norm": 0.20044063031673431,
      "learning_rate": 3.5176065329454933e-06,
      "loss": 0.0011,
      "step": 116650
    },
    {
      "epoch": 1.9088603452507567,
      "grad_norm": 0.0877494141459465,
      "learning_rate": 3.5166974194700997e-06,
      "loss": 0.0013,
      "step": 116660
    },
    {
      "epoch": 1.9090239712018326,
      "grad_norm": 0.08999335020780563,
      "learning_rate": 3.5157883597608606e-06,
      "loss": 0.0017,
      "step": 116670
    },
    {
      "epoch": 1.9091875971529084,
      "grad_norm": 0.14329926669597626,
      "learning_rate": 3.514879353850725e-06,
      "loss": 0.0013,
      "step": 116680
    },
    {
      "epoch": 1.9093512231039842,
      "grad_norm": 0.2274893969297409,
      "learning_rate": 3.513970401772644e-06,
      "loss": 0.002,
      "step": 116690
    },
    {
      "epoch": 1.9095148490550602,
      "grad_norm": 0.024614742025732994,
      "learning_rate": 3.513061503559564e-06,
      "loss": 0.0009,
      "step": 116700
    },
    {
      "epoch": 1.909678475006136,
      "grad_norm": 0.0431903600692749,
      "learning_rate": 3.512152659244431e-06,
      "loss": 0.0011,
      "step": 116710
    },
    {
      "epoch": 1.9098421009572117,
      "grad_norm": 0.1141614317893982,
      "learning_rate": 3.511243868860188e-06,
      "loss": 0.0017,
      "step": 116720
    },
    {
      "epoch": 1.9100057269082877,
      "grad_norm": 0.051052387803792953,
      "learning_rate": 3.510335132439776e-06,
      "loss": 0.0013,
      "step": 116730
    },
    {
      "epoch": 1.9101693528593635,
      "grad_norm": 0.09419436007738113,
      "learning_rate": 3.5094264500161345e-06,
      "loss": 0.0029,
      "step": 116740
    },
    {
      "epoch": 1.9103329788104393,
      "grad_norm": 0.06832196563482285,
      "learning_rate": 3.5085178216222034e-06,
      "loss": 0.0008,
      "step": 116750
    },
    {
      "epoch": 1.9104966047615153,
      "grad_norm": 0.12940138578414917,
      "learning_rate": 3.5076092472909163e-06,
      "loss": 0.001,
      "step": 116760
    },
    {
      "epoch": 1.910660230712591,
      "grad_norm": 0.11513227969408035,
      "learning_rate": 3.5067007270552084e-06,
      "loss": 0.0011,
      "step": 116770
    },
    {
      "epoch": 1.9108238566636668,
      "grad_norm": 0.048090241849422455,
      "learning_rate": 3.505792260948009e-06,
      "loss": 0.0013,
      "step": 116780
    },
    {
      "epoch": 1.9109874826147428,
      "grad_norm": 0.03949848562479019,
      "learning_rate": 3.5048838490022487e-06,
      "loss": 0.0019,
      "step": 116790
    },
    {
      "epoch": 1.9111511085658184,
      "grad_norm": 0.050484053790569305,
      "learning_rate": 3.503975491250854e-06,
      "loss": 0.0018,
      "step": 116800
    },
    {
      "epoch": 1.9113147345168944,
      "grad_norm": 0.03547117859125137,
      "learning_rate": 3.503067187726753e-06,
      "loss": 0.0018,
      "step": 116810
    },
    {
      "epoch": 1.9114783604679704,
      "grad_norm": 0.06423182040452957,
      "learning_rate": 3.5021589384628692e-06,
      "loss": 0.0009,
      "step": 116820
    },
    {
      "epoch": 1.911641986419046,
      "grad_norm": 0.032825957983732224,
      "learning_rate": 3.501250743492124e-06,
      "loss": 0.0017,
      "step": 116830
    },
    {
      "epoch": 1.911805612370122,
      "grad_norm": 0.1677391529083252,
      "learning_rate": 3.5003426028474365e-06,
      "loss": 0.0016,
      "step": 116840
    },
    {
      "epoch": 1.9119692383211977,
      "grad_norm": 0.20630770921707153,
      "learning_rate": 3.499434516561726e-06,
      "loss": 0.001,
      "step": 116850
    },
    {
      "epoch": 1.9121328642722735,
      "grad_norm": 0.04731212183833122,
      "learning_rate": 3.4985264846679075e-06,
      "loss": 0.0021,
      "step": 116860
    },
    {
      "epoch": 1.9122964902233495,
      "grad_norm": 0.13591893017292023,
      "learning_rate": 3.497618507198896e-06,
      "loss": 0.0023,
      "step": 116870
    },
    {
      "epoch": 1.9124601161744252,
      "grad_norm": 0.028104446828365326,
      "learning_rate": 3.496710584187602e-06,
      "loss": 0.0011,
      "step": 116880
    },
    {
      "epoch": 1.912623742125501,
      "grad_norm": 0.039137497544288635,
      "learning_rate": 3.4958027156669362e-06,
      "loss": 0.0016,
      "step": 116890
    },
    {
      "epoch": 1.912787368076577,
      "grad_norm": 0.03131349757313728,
      "learning_rate": 3.4948949016698068e-06,
      "loss": 0.0013,
      "step": 116900
    },
    {
      "epoch": 1.9129509940276528,
      "grad_norm": 0.062079157680273056,
      "learning_rate": 3.4939871422291195e-06,
      "loss": 0.0013,
      "step": 116910
    },
    {
      "epoch": 1.9131146199787286,
      "grad_norm": 0.08098719269037247,
      "learning_rate": 3.493079437377779e-06,
      "loss": 0.0011,
      "step": 116920
    },
    {
      "epoch": 1.9132782459298046,
      "grad_norm": 0.032798703759908676,
      "learning_rate": 3.492171787148687e-06,
      "loss": 0.0016,
      "step": 116930
    },
    {
      "epoch": 1.9134418718808803,
      "grad_norm": 0.00832432322204113,
      "learning_rate": 3.491264191574744e-06,
      "loss": 0.0008,
      "step": 116940
    },
    {
      "epoch": 1.913605497831956,
      "grad_norm": 0.07371308654546738,
      "learning_rate": 3.4903566506888476e-06,
      "loss": 0.0009,
      "step": 116950
    },
    {
      "epoch": 1.913769123783032,
      "grad_norm": 0.06439722329378128,
      "learning_rate": 3.4894491645238937e-06,
      "loss": 0.0009,
      "step": 116960
    },
    {
      "epoch": 1.9139327497341079,
      "grad_norm": 0.08005999773740768,
      "learning_rate": 3.4885417331127797e-06,
      "loss": 0.0009,
      "step": 116970
    },
    {
      "epoch": 1.9140963756851836,
      "grad_norm": 0.09819091856479645,
      "learning_rate": 3.487634356488393e-06,
      "loss": 0.0013,
      "step": 116980
    },
    {
      "epoch": 1.9142600016362596,
      "grad_norm": 0.06988803297281265,
      "learning_rate": 3.4867270346836263e-06,
      "loss": 0.0013,
      "step": 116990
    },
    {
      "epoch": 1.9144236275873352,
      "grad_norm": 0.022273756563663483,
      "learning_rate": 3.4858197677313676e-06,
      "loss": 0.001,
      "step": 117000
    },
    {
      "epoch": 1.9145872535384112,
      "grad_norm": 0.07267189770936966,
      "learning_rate": 3.4849125556645026e-06,
      "loss": 0.0016,
      "step": 117010
    },
    {
      "epoch": 1.9147508794894872,
      "grad_norm": 0.07633133977651596,
      "learning_rate": 3.4840053985159154e-06,
      "loss": 0.0012,
      "step": 117020
    },
    {
      "epoch": 1.9149145054405627,
      "grad_norm": 0.0614902563393116,
      "learning_rate": 3.4830982963184902e-06,
      "loss": 0.0011,
      "step": 117030
    },
    {
      "epoch": 1.9150781313916387,
      "grad_norm": 0.046877384185791016,
      "learning_rate": 3.4821912491051056e-06,
      "loss": 0.0013,
      "step": 117040
    },
    {
      "epoch": 1.9152417573427145,
      "grad_norm": 0.03472324460744858,
      "learning_rate": 3.4812842569086403e-06,
      "loss": 0.0024,
      "step": 117050
    },
    {
      "epoch": 1.9154053832937903,
      "grad_norm": 0.12010905891656876,
      "learning_rate": 3.4803773197619707e-06,
      "loss": 0.0018,
      "step": 117060
    },
    {
      "epoch": 1.9155690092448663,
      "grad_norm": 0.18073269724845886,
      "learning_rate": 3.4794704376979714e-06,
      "loss": 0.0035,
      "step": 117070
    },
    {
      "epoch": 1.915732635195942,
      "grad_norm": 0.12548808753490448,
      "learning_rate": 3.478563610749514e-06,
      "loss": 0.0026,
      "step": 117080
    },
    {
      "epoch": 1.9158962611470178,
      "grad_norm": 0.05737784877419472,
      "learning_rate": 3.477656838949469e-06,
      "loss": 0.0019,
      "step": 117090
    },
    {
      "epoch": 1.9160598870980938,
      "grad_norm": 0.04905271530151367,
      "learning_rate": 3.476750122330704e-06,
      "loss": 0.0012,
      "step": 117100
    },
    {
      "epoch": 1.9162235130491696,
      "grad_norm": 0.10038364678621292,
      "learning_rate": 3.4758434609260865e-06,
      "loss": 0.0015,
      "step": 117110
    },
    {
      "epoch": 1.9163871390002454,
      "grad_norm": 0.03029683418571949,
      "learning_rate": 3.4749368547684804e-06,
      "loss": 0.0013,
      "step": 117120
    },
    {
      "epoch": 1.9165507649513214,
      "grad_norm": 0.16878052055835724,
      "learning_rate": 3.474030303890748e-06,
      "loss": 0.0009,
      "step": 117130
    },
    {
      "epoch": 1.9167143909023971,
      "grad_norm": 0.0847473070025444,
      "learning_rate": 3.473123808325749e-06,
      "loss": 0.0013,
      "step": 117140
    },
    {
      "epoch": 1.916878016853473,
      "grad_norm": 0.0683387890458107,
      "learning_rate": 3.472217368106343e-06,
      "loss": 0.0009,
      "step": 117150
    },
    {
      "epoch": 1.917041642804549,
      "grad_norm": 0.0396602600812912,
      "learning_rate": 3.4713109832653846e-06,
      "loss": 0.001,
      "step": 117160
    },
    {
      "epoch": 1.9172052687556247,
      "grad_norm": 0.18097415566444397,
      "learning_rate": 3.4704046538357294e-06,
      "loss": 0.0016,
      "step": 117170
    },
    {
      "epoch": 1.9173688947067005,
      "grad_norm": 0.13552826642990112,
      "learning_rate": 3.4694983798502302e-06,
      "loss": 0.0009,
      "step": 117180
    },
    {
      "epoch": 1.9175325206577765,
      "grad_norm": 0.013414356857538223,
      "learning_rate": 3.468592161341735e-06,
      "loss": 0.0016,
      "step": 117190
    },
    {
      "epoch": 1.917696146608852,
      "grad_norm": 0.051815301179885864,
      "learning_rate": 3.4676859983430938e-06,
      "loss": 0.0012,
      "step": 117200
    },
    {
      "epoch": 1.917859772559928,
      "grad_norm": 0.02268894761800766,
      "learning_rate": 3.466779890887152e-06,
      "loss": 0.0015,
      "step": 117210
    },
    {
      "epoch": 1.918023398511004,
      "grad_norm": 0.029339473694562912,
      "learning_rate": 3.465873839006755e-06,
      "loss": 0.0009,
      "step": 117220
    },
    {
      "epoch": 1.9181870244620796,
      "grad_norm": 0.34612491726875305,
      "learning_rate": 3.464967842734743e-06,
      "loss": 0.002,
      "step": 117230
    },
    {
      "epoch": 1.9183506504131556,
      "grad_norm": 0.14712712168693542,
      "learning_rate": 3.464061902103957e-06,
      "loss": 0.0015,
      "step": 117240
    },
    {
      "epoch": 1.9185142763642313,
      "grad_norm": 0.1953924298286438,
      "learning_rate": 3.4631560171472357e-06,
      "loss": 0.0026,
      "step": 117250
    },
    {
      "epoch": 1.918677902315307,
      "grad_norm": 0.02603316865861416,
      "learning_rate": 3.462250187897416e-06,
      "loss": 0.0012,
      "step": 117260
    },
    {
      "epoch": 1.918841528266383,
      "grad_norm": 0.0692998468875885,
      "learning_rate": 3.4613444143873306e-06,
      "loss": 0.0012,
      "step": 117270
    },
    {
      "epoch": 1.9190051542174589,
      "grad_norm": 0.056674063205718994,
      "learning_rate": 3.460438696649814e-06,
      "loss": 0.0019,
      "step": 117280
    },
    {
      "epoch": 1.9191687801685346,
      "grad_norm": 0.05686596408486366,
      "learning_rate": 3.459533034717692e-06,
      "loss": 0.0015,
      "step": 117290
    },
    {
      "epoch": 1.9193324061196106,
      "grad_norm": 0.030991658568382263,
      "learning_rate": 3.4586274286237955e-06,
      "loss": 0.0007,
      "step": 117300
    },
    {
      "epoch": 1.9194960320706864,
      "grad_norm": 0.030475223436951637,
      "learning_rate": 3.45772187840095e-06,
      "loss": 0.001,
      "step": 117310
    },
    {
      "epoch": 1.9196596580217622,
      "grad_norm": 0.08460620045661926,
      "learning_rate": 3.4568163840819803e-06,
      "loss": 0.0014,
      "step": 117320
    },
    {
      "epoch": 1.9198232839728382,
      "grad_norm": 0.00464920699596405,
      "learning_rate": 3.4559109456997077e-06,
      "loss": 0.0012,
      "step": 117330
    },
    {
      "epoch": 1.919986909923914,
      "grad_norm": 0.061519816517829895,
      "learning_rate": 3.455005563286952e-06,
      "loss": 0.0019,
      "step": 117340
    },
    {
      "epoch": 1.9201505358749897,
      "grad_norm": 0.04516362026333809,
      "learning_rate": 3.454100236876531e-06,
      "loss": 0.0011,
      "step": 117350
    },
    {
      "epoch": 1.9203141618260657,
      "grad_norm": 0.018166685476899147,
      "learning_rate": 3.453194966501261e-06,
      "loss": 0.0008,
      "step": 117360
    },
    {
      "epoch": 1.9204777877771415,
      "grad_norm": 0.014799079857766628,
      "learning_rate": 3.452289752193956e-06,
      "loss": 0.0012,
      "step": 117370
    },
    {
      "epoch": 1.9206414137282173,
      "grad_norm": 0.5212557315826416,
      "learning_rate": 3.451384593987429e-06,
      "loss": 0.0077,
      "step": 117380
    },
    {
      "epoch": 1.9208050396792933,
      "grad_norm": 0.053820863366127014,
      "learning_rate": 3.450479491914487e-06,
      "loss": 0.0005,
      "step": 117390
    },
    {
      "epoch": 1.9209686656303688,
      "grad_norm": 0.05057581514120102,
      "learning_rate": 3.4495744460079395e-06,
      "loss": 0.0012,
      "step": 117400
    },
    {
      "epoch": 1.9211322915814448,
      "grad_norm": 0.17194992303848267,
      "learning_rate": 3.448669456300592e-06,
      "loss": 0.0015,
      "step": 117410
    },
    {
      "epoch": 1.9212959175325206,
      "grad_norm": 0.0869787335395813,
      "learning_rate": 3.447764522825248e-06,
      "loss": 0.0019,
      "step": 117420
    },
    {
      "epoch": 1.9214595434835964,
      "grad_norm": 0.07992877811193466,
      "learning_rate": 3.4468596456147096e-06,
      "loss": 0.001,
      "step": 117430
    },
    {
      "epoch": 1.9216231694346724,
      "grad_norm": 0.10534562915563583,
      "learning_rate": 3.445954824701776e-06,
      "loss": 0.0011,
      "step": 117440
    },
    {
      "epoch": 1.9217867953857481,
      "grad_norm": 0.2773364782333374,
      "learning_rate": 3.445050060119244e-06,
      "loss": 0.0023,
      "step": 117450
    },
    {
      "epoch": 1.921950421336824,
      "grad_norm": 0.1982908993959427,
      "learning_rate": 3.4441453518999107e-06,
      "loss": 0.0016,
      "step": 117460
    },
    {
      "epoch": 1.9221140472879,
      "grad_norm": 0.027084222063422203,
      "learning_rate": 3.4432407000765694e-06,
      "loss": 0.0011,
      "step": 117470
    },
    {
      "epoch": 1.9222776732389757,
      "grad_norm": 0.13308580219745636,
      "learning_rate": 3.4423361046820137e-06,
      "loss": 0.0014,
      "step": 117480
    },
    {
      "epoch": 1.9224412991900515,
      "grad_norm": 0.09366597980260849,
      "learning_rate": 3.4414315657490273e-06,
      "loss": 0.0012,
      "step": 117490
    },
    {
      "epoch": 1.9226049251411275,
      "grad_norm": 0.013624492101371288,
      "learning_rate": 3.440527083310401e-06,
      "loss": 0.0013,
      "step": 117500
    },
    {
      "epoch": 1.9227685510922032,
      "grad_norm": 0.3883621394634247,
      "learning_rate": 3.4396226573989195e-06,
      "loss": 0.0015,
      "step": 117510
    },
    {
      "epoch": 1.922932177043279,
      "grad_norm": 0.005440632347017527,
      "learning_rate": 3.438718288047367e-06,
      "loss": 0.001,
      "step": 117520
    },
    {
      "epoch": 1.923095802994355,
      "grad_norm": 0.2455538958311081,
      "learning_rate": 3.4378139752885243e-06,
      "loss": 0.0015,
      "step": 117530
    },
    {
      "epoch": 1.9232594289454308,
      "grad_norm": 0.0019826283678412437,
      "learning_rate": 3.43690971915517e-06,
      "loss": 0.0008,
      "step": 117540
    },
    {
      "epoch": 1.9234230548965066,
      "grad_norm": 0.11283384263515472,
      "learning_rate": 3.436005519680082e-06,
      "loss": 0.0019,
      "step": 117550
    },
    {
      "epoch": 1.9235866808475826,
      "grad_norm": 0.08573871105909348,
      "learning_rate": 3.4351013768960345e-06,
      "loss": 0.0018,
      "step": 117560
    },
    {
      "epoch": 1.923750306798658,
      "grad_norm": 0.07458851486444473,
      "learning_rate": 3.434197290835801e-06,
      "loss": 0.0012,
      "step": 117570
    },
    {
      "epoch": 1.923913932749734,
      "grad_norm": 0.03993265703320503,
      "learning_rate": 3.4332932615321534e-06,
      "loss": 0.0018,
      "step": 117580
    },
    {
      "epoch": 1.92407755870081,
      "grad_norm": 0.006638629361987114,
      "learning_rate": 3.432389289017859e-06,
      "loss": 0.0011,
      "step": 117590
    },
    {
      "epoch": 1.9242411846518856,
      "grad_norm": 0.1080513596534729,
      "learning_rate": 3.431485373325685e-06,
      "loss": 0.0012,
      "step": 117600
    },
    {
      "epoch": 1.9244048106029616,
      "grad_norm": 0.04473069682717323,
      "learning_rate": 3.4305815144883965e-06,
      "loss": 0.0011,
      "step": 117610
    },
    {
      "epoch": 1.9245684365540374,
      "grad_norm": 0.013832427561283112,
      "learning_rate": 3.4296777125387557e-06,
      "loss": 0.0024,
      "step": 117620
    },
    {
      "epoch": 1.9247320625051132,
      "grad_norm": 0.05827764421701431,
      "learning_rate": 3.4287739675095244e-06,
      "loss": 0.0029,
      "step": 117630
    },
    {
      "epoch": 1.9248956884561892,
      "grad_norm": 0.26668936014175415,
      "learning_rate": 3.4278702794334595e-06,
      "loss": 0.0014,
      "step": 117640
    },
    {
      "epoch": 1.925059314407265,
      "grad_norm": 0.029696675017476082,
      "learning_rate": 3.4269666483433195e-06,
      "loss": 0.0013,
      "step": 117650
    },
    {
      "epoch": 1.9252229403583407,
      "grad_norm": 0.1839888095855713,
      "learning_rate": 3.426063074271857e-06,
      "loss": 0.001,
      "step": 117660
    },
    {
      "epoch": 1.9253865663094167,
      "grad_norm": 0.025584794580936432,
      "learning_rate": 3.425159557251824e-06,
      "loss": 0.0016,
      "step": 117670
    },
    {
      "epoch": 1.9255501922604925,
      "grad_norm": 0.20758064091205597,
      "learning_rate": 3.424256097315976e-06,
      "loss": 0.0022,
      "step": 117680
    },
    {
      "epoch": 1.9257138182115683,
      "grad_norm": 0.022229362279176712,
      "learning_rate": 3.423352694497055e-06,
      "loss": 0.001,
      "step": 117690
    },
    {
      "epoch": 1.9258774441626443,
      "grad_norm": 0.05995320901274681,
      "learning_rate": 3.4224493488278086e-06,
      "loss": 0.0011,
      "step": 117700
    },
    {
      "epoch": 1.92604107011372,
      "grad_norm": 0.006995433010160923,
      "learning_rate": 3.4215460603409824e-06,
      "loss": 0.0016,
      "step": 117710
    },
    {
      "epoch": 1.9262046960647958,
      "grad_norm": 0.047567665576934814,
      "learning_rate": 3.4206428290693184e-06,
      "loss": 0.004,
      "step": 117720
    },
    {
      "epoch": 1.9263683220158718,
      "grad_norm": 0.024939458817243576,
      "learning_rate": 3.4197396550455543e-06,
      "loss": 0.0015,
      "step": 117730
    },
    {
      "epoch": 1.9265319479669476,
      "grad_norm": 0.047782547771930695,
      "learning_rate": 3.41883653830243e-06,
      "loss": 0.0011,
      "step": 117740
    },
    {
      "epoch": 1.9266955739180234,
      "grad_norm": 0.13636085391044617,
      "learning_rate": 3.4179334788726816e-06,
      "loss": 0.0029,
      "step": 117750
    },
    {
      "epoch": 1.9268591998690994,
      "grad_norm": 0.10191311687231064,
      "learning_rate": 3.4170304767890426e-06,
      "loss": 0.0015,
      "step": 117760
    },
    {
      "epoch": 1.927022825820175,
      "grad_norm": 0.1681540161371231,
      "learning_rate": 3.4161275320842435e-06,
      "loss": 0.0013,
      "step": 117770
    },
    {
      "epoch": 1.927186451771251,
      "grad_norm": 0.07280343770980835,
      "learning_rate": 3.4152246447910174e-06,
      "loss": 0.0016,
      "step": 117780
    },
    {
      "epoch": 1.927350077722327,
      "grad_norm": 0.05919857323169708,
      "learning_rate": 3.4143218149420855e-06,
      "loss": 0.0021,
      "step": 117790
    },
    {
      "epoch": 1.9275137036734025,
      "grad_norm": 0.06087833642959595,
      "learning_rate": 3.4134190425701773e-06,
      "loss": 0.0017,
      "step": 117800
    },
    {
      "epoch": 1.9276773296244785,
      "grad_norm": 0.09941631555557251,
      "learning_rate": 3.4125163277080166e-06,
      "loss": 0.0015,
      "step": 117810
    },
    {
      "epoch": 1.9278409555755542,
      "grad_norm": 0.19936014711856842,
      "learning_rate": 3.411613670388323e-06,
      "loss": 0.0012,
      "step": 117820
    },
    {
      "epoch": 1.92800458152663,
      "grad_norm": 0.04386596381664276,
      "learning_rate": 3.410711070643816e-06,
      "loss": 0.0008,
      "step": 117830
    },
    {
      "epoch": 1.928168207477706,
      "grad_norm": 0.052457183599472046,
      "learning_rate": 3.4098085285072135e-06,
      "loss": 0.0017,
      "step": 117840
    },
    {
      "epoch": 1.9283318334287818,
      "grad_norm": 0.15446393191814423,
      "learning_rate": 3.40890604401123e-06,
      "loss": 0.0016,
      "step": 117850
    },
    {
      "epoch": 1.9284954593798576,
      "grad_norm": 0.05146511644124985,
      "learning_rate": 3.4080036171885776e-06,
      "loss": 0.0016,
      "step": 117860
    },
    {
      "epoch": 1.9286590853309336,
      "grad_norm": 0.03194218873977661,
      "learning_rate": 3.407101248071969e-06,
      "loss": 0.0006,
      "step": 117870
    },
    {
      "epoch": 1.9288227112820093,
      "grad_norm": 0.0396483913064003,
      "learning_rate": 3.406198936694112e-06,
      "loss": 0.0015,
      "step": 117880
    },
    {
      "epoch": 1.928986337233085,
      "grad_norm": 0.08825863897800446,
      "learning_rate": 3.4052966830877117e-06,
      "loss": 0.0016,
      "step": 117890
    },
    {
      "epoch": 1.929149963184161,
      "grad_norm": 0.2015662044286728,
      "learning_rate": 3.404394487285474e-06,
      "loss": 0.0022,
      "step": 117900
    },
    {
      "epoch": 1.9293135891352369,
      "grad_norm": 0.005432970356196165,
      "learning_rate": 3.403492349320101e-06,
      "loss": 0.0004,
      "step": 117910
    },
    {
      "epoch": 1.9294772150863126,
      "grad_norm": 0.05115284398198128,
      "learning_rate": 3.402590269224294e-06,
      "loss": 0.0012,
      "step": 117920
    },
    {
      "epoch": 1.9296408410373886,
      "grad_norm": 0.09749193489551544,
      "learning_rate": 3.40168824703075e-06,
      "loss": 0.001,
      "step": 117930
    },
    {
      "epoch": 1.9298044669884644,
      "grad_norm": 0.01844898983836174,
      "learning_rate": 3.400786282772165e-06,
      "loss": 0.001,
      "step": 117940
    },
    {
      "epoch": 1.9299680929395402,
      "grad_norm": 0.1313873678445816,
      "learning_rate": 3.3998843764812332e-06,
      "loss": 0.0021,
      "step": 117950
    },
    {
      "epoch": 1.9301317188906162,
      "grad_norm": 0.11527419090270996,
      "learning_rate": 3.3989825281906486e-06,
      "loss": 0.0021,
      "step": 117960
    },
    {
      "epoch": 1.9302953448416917,
      "grad_norm": 0.17308522760868073,
      "learning_rate": 3.3980807379330983e-06,
      "loss": 0.0014,
      "step": 117970
    },
    {
      "epoch": 1.9304589707927677,
      "grad_norm": 0.35439544916152954,
      "learning_rate": 3.3971790057412735e-06,
      "loss": 0.0015,
      "step": 117980
    },
    {
      "epoch": 1.9306225967438437,
      "grad_norm": 0.11949679255485535,
      "learning_rate": 3.3962773316478557e-06,
      "loss": 0.0011,
      "step": 117990
    },
    {
      "epoch": 1.9307862226949193,
      "grad_norm": 0.03896412253379822,
      "learning_rate": 3.3953757156855307e-06,
      "loss": 0.0012,
      "step": 118000
    },
    {
      "epoch": 1.9309498486459953,
      "grad_norm": 0.08541835099458694,
      "learning_rate": 3.3944741578869778e-06,
      "loss": 0.0016,
      "step": 118010
    },
    {
      "epoch": 1.931113474597071,
      "grad_norm": 0.14984484016895294,
      "learning_rate": 3.393572658284878e-06,
      "loss": 0.0014,
      "step": 118020
    },
    {
      "epoch": 1.9312771005481468,
      "grad_norm": 0.11960729211568832,
      "learning_rate": 3.3926712169119097e-06,
      "loss": 0.0016,
      "step": 118030
    },
    {
      "epoch": 1.9314407264992228,
      "grad_norm": 0.035255882889032364,
      "learning_rate": 3.3917698338007464e-06,
      "loss": 0.0013,
      "step": 118040
    },
    {
      "epoch": 1.9316043524502986,
      "grad_norm": 0.01587798073887825,
      "learning_rate": 3.390868508984061e-06,
      "loss": 0.001,
      "step": 118050
    },
    {
      "epoch": 1.9317679784013744,
      "grad_norm": 0.11635394394397736,
      "learning_rate": 3.389967242494525e-06,
      "loss": 0.0012,
      "step": 118060
    },
    {
      "epoch": 1.9319316043524504,
      "grad_norm": 0.13837963342666626,
      "learning_rate": 3.389066034364806e-06,
      "loss": 0.001,
      "step": 118070
    },
    {
      "epoch": 1.9320952303035261,
      "grad_norm": 0.09984228760004044,
      "learning_rate": 3.388164884627574e-06,
      "loss": 0.0016,
      "step": 118080
    },
    {
      "epoch": 1.932258856254602,
      "grad_norm": 0.05568477511405945,
      "learning_rate": 3.3872637933154893e-06,
      "loss": 0.0021,
      "step": 118090
    },
    {
      "epoch": 1.932422482205678,
      "grad_norm": 0.052926015108823776,
      "learning_rate": 3.386362760461216e-06,
      "loss": 0.0038,
      "step": 118100
    },
    {
      "epoch": 1.9325861081567537,
      "grad_norm": 0.065251424908638,
      "learning_rate": 3.3854617860974136e-06,
      "loss": 0.0027,
      "step": 118110
    },
    {
      "epoch": 1.9327497341078295,
      "grad_norm": 0.019251657649874687,
      "learning_rate": 3.3845608702567424e-06,
      "loss": 0.0008,
      "step": 118120
    },
    {
      "epoch": 1.9329133600589055,
      "grad_norm": 0.03404831886291504,
      "learning_rate": 3.383660012971856e-06,
      "loss": 0.0009,
      "step": 118130
    },
    {
      "epoch": 1.9330769860099812,
      "grad_norm": 0.08136383444070816,
      "learning_rate": 3.3827592142754095e-06,
      "loss": 0.0015,
      "step": 118140
    },
    {
      "epoch": 1.933240611961057,
      "grad_norm": 0.15428341925144196,
      "learning_rate": 3.381858474200055e-06,
      "loss": 0.0021,
      "step": 118150
    },
    {
      "epoch": 1.933404237912133,
      "grad_norm": 0.0905461311340332,
      "learning_rate": 3.3809577927784407e-06,
      "loss": 0.0014,
      "step": 118160
    },
    {
      "epoch": 1.9335678638632086,
      "grad_norm": 0.1536383479833603,
      "learning_rate": 3.380057170043215e-06,
      "loss": 0.0018,
      "step": 118170
    },
    {
      "epoch": 1.9337314898142846,
      "grad_norm": 0.06010723114013672,
      "learning_rate": 3.3791566060270264e-06,
      "loss": 0.0016,
      "step": 118180
    },
    {
      "epoch": 1.9338951157653606,
      "grad_norm": 0.05428522452712059,
      "learning_rate": 3.3782561007625138e-06,
      "loss": 0.0014,
      "step": 118190
    },
    {
      "epoch": 1.934058741716436,
      "grad_norm": 0.013938963413238525,
      "learning_rate": 3.377355654282319e-06,
      "loss": 0.0007,
      "step": 118200
    },
    {
      "epoch": 1.934222367667512,
      "grad_norm": 0.07637760043144226,
      "learning_rate": 3.376455266619082e-06,
      "loss": 0.0011,
      "step": 118210
    },
    {
      "epoch": 1.9343859936185879,
      "grad_norm": 0.09367643296718597,
      "learning_rate": 3.3755549378054396e-06,
      "loss": 0.0014,
      "step": 118220
    },
    {
      "epoch": 1.9345496195696636,
      "grad_norm": 0.08403017371892929,
      "learning_rate": 3.374654667874025e-06,
      "loss": 0.0009,
      "step": 118230
    },
    {
      "epoch": 1.9347132455207396,
      "grad_norm": 0.06915469467639923,
      "learning_rate": 3.3737544568574736e-06,
      "loss": 0.0005,
      "step": 118240
    },
    {
      "epoch": 1.9348768714718154,
      "grad_norm": 0.2075105905532837,
      "learning_rate": 3.3728543047884142e-06,
      "loss": 0.0014,
      "step": 118250
    },
    {
      "epoch": 1.9350404974228912,
      "grad_norm": 0.05022398754954338,
      "learning_rate": 3.3719542116994765e-06,
      "loss": 0.0018,
      "step": 118260
    },
    {
      "epoch": 1.9352041233739672,
      "grad_norm": 0.04598352313041687,
      "learning_rate": 3.371054177623285e-06,
      "loss": 0.0018,
      "step": 118270
    },
    {
      "epoch": 1.935367749325043,
      "grad_norm": 0.05838308855891228,
      "learning_rate": 3.3701542025924657e-06,
      "loss": 0.0011,
      "step": 118280
    },
    {
      "epoch": 1.9355313752761187,
      "grad_norm": 0.07422434538602829,
      "learning_rate": 3.3692542866396383e-06,
      "loss": 0.0032,
      "step": 118290
    },
    {
      "epoch": 1.9356950012271947,
      "grad_norm": 0.10354265570640564,
      "learning_rate": 3.368354429797423e-06,
      "loss": 0.002,
      "step": 118300
    },
    {
      "epoch": 1.9358586271782705,
      "grad_norm": 0.006993905175477266,
      "learning_rate": 3.3674546320984385e-06,
      "loss": 0.0012,
      "step": 118310
    },
    {
      "epoch": 1.9360222531293463,
      "grad_norm": 0.048503126949071884,
      "learning_rate": 3.3665548935753e-06,
      "loss": 0.0018,
      "step": 118320
    },
    {
      "epoch": 1.9361858790804223,
      "grad_norm": 0.05226905271410942,
      "learning_rate": 3.365655214260621e-06,
      "loss": 0.0019,
      "step": 118330
    },
    {
      "epoch": 1.9363495050314978,
      "grad_norm": 0.023771366104483604,
      "learning_rate": 3.3647555941870118e-06,
      "loss": 0.0031,
      "step": 118340
    },
    {
      "epoch": 1.9365131309825738,
      "grad_norm": 0.05646058917045593,
      "learning_rate": 3.3638560333870824e-06,
      "loss": 0.0015,
      "step": 118350
    },
    {
      "epoch": 1.9366767569336498,
      "grad_norm": 0.020113365724682808,
      "learning_rate": 3.36295653189344e-06,
      "loss": 0.0017,
      "step": 118360
    },
    {
      "epoch": 1.9368403828847254,
      "grad_norm": 0.018712610006332397,
      "learning_rate": 3.3620570897386877e-06,
      "loss": 0.0016,
      "step": 118370
    },
    {
      "epoch": 1.9370040088358014,
      "grad_norm": 0.041349492967128754,
      "learning_rate": 3.361157706955431e-06,
      "loss": 0.0011,
      "step": 118380
    },
    {
      "epoch": 1.9371676347868771,
      "grad_norm": 0.03684849664568901,
      "learning_rate": 3.360258383576267e-06,
      "loss": 0.0012,
      "step": 118390
    },
    {
      "epoch": 1.937331260737953,
      "grad_norm": 0.019560372456908226,
      "learning_rate": 3.3593591196337962e-06,
      "loss": 0.0008,
      "step": 118400
    },
    {
      "epoch": 1.937494886689029,
      "grad_norm": 0.13348115980625153,
      "learning_rate": 3.3584599151606136e-06,
      "loss": 0.0015,
      "step": 118410
    },
    {
      "epoch": 1.9376585126401047,
      "grad_norm": 0.07879369705915451,
      "learning_rate": 3.3575607701893144e-06,
      "loss": 0.0012,
      "step": 118420
    },
    {
      "epoch": 1.9378221385911805,
      "grad_norm": 0.11280323565006256,
      "learning_rate": 3.356661684752489e-06,
      "loss": 0.0023,
      "step": 118430
    },
    {
      "epoch": 1.9379857645422565,
      "grad_norm": 0.04920584708452225,
      "learning_rate": 3.355762658882728e-06,
      "loss": 0.0021,
      "step": 118440
    },
    {
      "epoch": 1.9381493904933322,
      "grad_norm": 0.08949918299913406,
      "learning_rate": 3.3548636926126177e-06,
      "loss": 0.0027,
      "step": 118450
    },
    {
      "epoch": 1.938313016444408,
      "grad_norm": 0.10668892413377762,
      "learning_rate": 3.353964785974746e-06,
      "loss": 0.0015,
      "step": 118460
    },
    {
      "epoch": 1.938476642395484,
      "grad_norm": 0.05627458170056343,
      "learning_rate": 3.3530659390016944e-06,
      "loss": 0.0016,
      "step": 118470
    },
    {
      "epoch": 1.9386402683465598,
      "grad_norm": 0.26361411809921265,
      "learning_rate": 3.352167151726047e-06,
      "loss": 0.0016,
      "step": 118480
    },
    {
      "epoch": 1.9388038942976356,
      "grad_norm": 0.1131122037768364,
      "learning_rate": 3.351268424180377e-06,
      "loss": 0.0012,
      "step": 118490
    },
    {
      "epoch": 1.9389675202487116,
      "grad_norm": 0.02040417492389679,
      "learning_rate": 3.3503697563972632e-06,
      "loss": 0.0015,
      "step": 118500
    },
    {
      "epoch": 1.9391311461997873,
      "grad_norm": 0.28824949264526367,
      "learning_rate": 3.349471148409282e-06,
      "loss": 0.0016,
      "step": 118510
    },
    {
      "epoch": 1.939294772150863,
      "grad_norm": 0.01501093152910471,
      "learning_rate": 3.348572600249005e-06,
      "loss": 0.0014,
      "step": 118520
    },
    {
      "epoch": 1.939458398101939,
      "grad_norm": 0.01099783182144165,
      "learning_rate": 3.3476741119490013e-06,
      "loss": 0.0011,
      "step": 118530
    },
    {
      "epoch": 1.9396220240530146,
      "grad_norm": 0.07206486165523529,
      "learning_rate": 3.34677568354184e-06,
      "loss": 0.0009,
      "step": 118540
    },
    {
      "epoch": 1.9397856500040906,
      "grad_norm": 0.09977972507476807,
      "learning_rate": 3.3458773150600877e-06,
      "loss": 0.0013,
      "step": 118550
    },
    {
      "epoch": 1.9399492759551666,
      "grad_norm": 0.2926282584667206,
      "learning_rate": 3.3449790065363066e-06,
      "loss": 0.0022,
      "step": 118560
    },
    {
      "epoch": 1.9401129019062422,
      "grad_norm": 0.08862805366516113,
      "learning_rate": 3.3440807580030584e-06,
      "loss": 0.0014,
      "step": 118570
    },
    {
      "epoch": 1.9402765278573182,
      "grad_norm": 0.18451426923274994,
      "learning_rate": 3.3431825694929044e-06,
      "loss": 0.0012,
      "step": 118580
    },
    {
      "epoch": 1.940440153808394,
      "grad_norm": 0.07498830556869507,
      "learning_rate": 3.3422844410383985e-06,
      "loss": 0.001,
      "step": 118590
    },
    {
      "epoch": 1.9406037797594697,
      "grad_norm": 0.061219144612550735,
      "learning_rate": 3.3413863726720978e-06,
      "loss": 0.0012,
      "step": 118600
    },
    {
      "epoch": 1.9407674057105457,
      "grad_norm": 0.04234624654054642,
      "learning_rate": 3.340488364426555e-06,
      "loss": 0.0029,
      "step": 118610
    },
    {
      "epoch": 1.9409310316616215,
      "grad_norm": 0.06528057157993317,
      "learning_rate": 3.3395904163343202e-06,
      "loss": 0.0011,
      "step": 118620
    },
    {
      "epoch": 1.9410946576126973,
      "grad_norm": 0.13006280362606049,
      "learning_rate": 3.338692528427942e-06,
      "loss": 0.0021,
      "step": 118630
    },
    {
      "epoch": 1.9412582835637733,
      "grad_norm": 0.005967382341623306,
      "learning_rate": 3.337794700739967e-06,
      "loss": 0.0021,
      "step": 118640
    },
    {
      "epoch": 1.941421909514849,
      "grad_norm": 0.02839636616408825,
      "learning_rate": 3.3368969333029388e-06,
      "loss": 0.0013,
      "step": 118650
    },
    {
      "epoch": 1.9415855354659248,
      "grad_norm": 0.09886398166418076,
      "learning_rate": 3.335999226149399e-06,
      "loss": 0.0018,
      "step": 118660
    },
    {
      "epoch": 1.9417491614170008,
      "grad_norm": 0.06083366647362709,
      "learning_rate": 3.3351015793118885e-06,
      "loss": 0.0021,
      "step": 118670
    },
    {
      "epoch": 1.9419127873680766,
      "grad_norm": 0.07727126777172089,
      "learning_rate": 3.3342039928229466e-06,
      "loss": 0.0015,
      "step": 118680
    },
    {
      "epoch": 1.9420764133191524,
      "grad_norm": 0.09327799081802368,
      "learning_rate": 3.3333064667151045e-06,
      "loss": 0.0009,
      "step": 118690
    },
    {
      "epoch": 1.9422400392702284,
      "grad_norm": 0.06905252486467361,
      "learning_rate": 3.3324090010208963e-06,
      "loss": 0.0022,
      "step": 118700
    },
    {
      "epoch": 1.9424036652213041,
      "grad_norm": 0.04499583691358566,
      "learning_rate": 3.331511595772855e-06,
      "loss": 0.0019,
      "step": 118710
    },
    {
      "epoch": 1.94256729117238,
      "grad_norm": 0.07709557563066483,
      "learning_rate": 3.3306142510035068e-06,
      "loss": 0.0012,
      "step": 118720
    },
    {
      "epoch": 1.942730917123456,
      "grad_norm": 0.032073136419057846,
      "learning_rate": 3.3297169667453805e-06,
      "loss": 0.0006,
      "step": 118730
    },
    {
      "epoch": 1.9428945430745315,
      "grad_norm": 0.05796315521001816,
      "learning_rate": 3.3288197430309997e-06,
      "loss": 0.0009,
      "step": 118740
    },
    {
      "epoch": 1.9430581690256075,
      "grad_norm": 0.0746682807803154,
      "learning_rate": 3.3279225798928873e-06,
      "loss": 0.001,
      "step": 118750
    },
    {
      "epoch": 1.9432217949766835,
      "grad_norm": 0.20048604905605316,
      "learning_rate": 3.327025477363562e-06,
      "loss": 0.0009,
      "step": 118760
    },
    {
      "epoch": 1.943385420927759,
      "grad_norm": 0.040062688291072845,
      "learning_rate": 3.3261284354755422e-06,
      "loss": 0.0013,
      "step": 118770
    },
    {
      "epoch": 1.943549046878835,
      "grad_norm": 0.0030569257214665413,
      "learning_rate": 3.325231454261345e-06,
      "loss": 0.0011,
      "step": 118780
    },
    {
      "epoch": 1.9437126728299108,
      "grad_norm": 0.020612314343452454,
      "learning_rate": 3.3243345337534814e-06,
      "loss": 0.0012,
      "step": 118790
    },
    {
      "epoch": 1.9438762987809866,
      "grad_norm": 0.09768067300319672,
      "learning_rate": 3.323437673984463e-06,
      "loss": 0.0011,
      "step": 118800
    },
    {
      "epoch": 1.9440399247320626,
      "grad_norm": 0.04377545416355133,
      "learning_rate": 3.3225408749868e-06,
      "loss": 0.0013,
      "step": 118810
    },
    {
      "epoch": 1.9442035506831383,
      "grad_norm": 0.07871692627668381,
      "learning_rate": 3.3216441367929987e-06,
      "loss": 0.0013,
      "step": 118820
    },
    {
      "epoch": 1.944367176634214,
      "grad_norm": 0.04945913329720497,
      "learning_rate": 3.3207474594355627e-06,
      "loss": 0.0013,
      "step": 118830
    },
    {
      "epoch": 1.94453080258529,
      "grad_norm": 0.3979766368865967,
      "learning_rate": 3.3198508429469968e-06,
      "loss": 0.0017,
      "step": 118840
    },
    {
      "epoch": 1.9446944285363659,
      "grad_norm": 0.10912535339593887,
      "learning_rate": 3.3189542873597978e-06,
      "loss": 0.0009,
      "step": 118850
    },
    {
      "epoch": 1.9448580544874416,
      "grad_norm": 0.029575081542134285,
      "learning_rate": 3.3180577927064672e-06,
      "loss": 0.0015,
      "step": 118860
    },
    {
      "epoch": 1.9450216804385176,
      "grad_norm": 0.0829453319311142,
      "learning_rate": 3.3171613590194986e-06,
      "loss": 0.0023,
      "step": 118870
    },
    {
      "epoch": 1.9451853063895934,
      "grad_norm": 0.21266023814678192,
      "learning_rate": 3.3162649863313874e-06,
      "loss": 0.0028,
      "step": 118880
    },
    {
      "epoch": 1.9453489323406692,
      "grad_norm": 0.17821168899536133,
      "learning_rate": 3.315368674674622e-06,
      "loss": 0.0013,
      "step": 118890
    },
    {
      "epoch": 1.9455125582917452,
      "grad_norm": 0.11173222213983536,
      "learning_rate": 3.314472424081694e-06,
      "loss": 0.0013,
      "step": 118900
    },
    {
      "epoch": 1.945676184242821,
      "grad_norm": 0.05157876014709473,
      "learning_rate": 3.3135762345850896e-06,
      "loss": 0.0012,
      "step": 118910
    },
    {
      "epoch": 1.9458398101938967,
      "grad_norm": 0.1706482619047165,
      "learning_rate": 3.3126801062172933e-06,
      "loss": 0.0018,
      "step": 118920
    },
    {
      "epoch": 1.9460034361449727,
      "grad_norm": 0.05622512474656105,
      "learning_rate": 3.311784039010787e-06,
      "loss": 0.0013,
      "step": 118930
    },
    {
      "epoch": 1.9461670620960483,
      "grad_norm": 0.3344621956348419,
      "learning_rate": 3.310888032998052e-06,
      "loss": 0.0017,
      "step": 118940
    },
    {
      "epoch": 1.9463306880471243,
      "grad_norm": 0.04745124280452728,
      "learning_rate": 3.309992088211566e-06,
      "loss": 0.0011,
      "step": 118950
    },
    {
      "epoch": 1.9464943139982003,
      "grad_norm": 0.07338631898164749,
      "learning_rate": 3.309096204683806e-06,
      "loss": 0.0011,
      "step": 118960
    },
    {
      "epoch": 1.9466579399492758,
      "grad_norm": 0.170085608959198,
      "learning_rate": 3.308200382447244e-06,
      "loss": 0.0015,
      "step": 118970
    },
    {
      "epoch": 1.9468215659003518,
      "grad_norm": 0.27082160115242004,
      "learning_rate": 3.307304621534354e-06,
      "loss": 0.0012,
      "step": 118980
    },
    {
      "epoch": 1.9469851918514276,
      "grad_norm": 0.08283862471580505,
      "learning_rate": 3.3064089219776013e-06,
      "loss": 0.0013,
      "step": 118990
    },
    {
      "epoch": 1.9471488178025034,
      "grad_norm": 0.06743039190769196,
      "learning_rate": 3.305513283809454e-06,
      "loss": 0.0018,
      "step": 119000
    },
    {
      "epoch": 1.9473124437535794,
      "grad_norm": 0.20719702541828156,
      "learning_rate": 3.304617707062378e-06,
      "loss": 0.0042,
      "step": 119010
    },
    {
      "epoch": 1.9474760697046551,
      "grad_norm": 0.07183507829904556,
      "learning_rate": 3.3037221917688357e-06,
      "loss": 0.0009,
      "step": 119020
    },
    {
      "epoch": 1.947639695655731,
      "grad_norm": 0.013016562908887863,
      "learning_rate": 3.3028267379612865e-06,
      "loss": 0.0006,
      "step": 119030
    },
    {
      "epoch": 1.947803321606807,
      "grad_norm": 0.058981407433748245,
      "learning_rate": 3.3019313456721903e-06,
      "loss": 0.0022,
      "step": 119040
    },
    {
      "epoch": 1.9479669475578827,
      "grad_norm": 0.03922748193144798,
      "learning_rate": 3.3010360149340006e-06,
      "loss": 0.0016,
      "step": 119050
    },
    {
      "epoch": 1.9481305735089585,
      "grad_norm": 0.03358004614710808,
      "learning_rate": 3.3001407457791723e-06,
      "loss": 0.0014,
      "step": 119060
    },
    {
      "epoch": 1.9482941994600345,
      "grad_norm": 0.09242702275514603,
      "learning_rate": 3.2992455382401566e-06,
      "loss": 0.0013,
      "step": 119070
    },
    {
      "epoch": 1.9484578254111102,
      "grad_norm": 0.049873560667037964,
      "learning_rate": 3.2983503923494033e-06,
      "loss": 0.0016,
      "step": 119080
    },
    {
      "epoch": 1.948621451362186,
      "grad_norm": 0.00814768485724926,
      "learning_rate": 3.2974553081393575e-06,
      "loss": 0.0011,
      "step": 119090
    },
    {
      "epoch": 1.948785077313262,
      "grad_norm": 0.12519040703773499,
      "learning_rate": 3.2965602856424645e-06,
      "loss": 0.0023,
      "step": 119100
    },
    {
      "epoch": 1.9489487032643378,
      "grad_norm": 0.016813622787594795,
      "learning_rate": 3.2956653248911672e-06,
      "loss": 0.0024,
      "step": 119110
    },
    {
      "epoch": 1.9491123292154136,
      "grad_norm": 0.058385998010635376,
      "learning_rate": 3.294770425917906e-06,
      "loss": 0.0013,
      "step": 119120
    },
    {
      "epoch": 1.9492759551664895,
      "grad_norm": 0.04116176441311836,
      "learning_rate": 3.293875588755118e-06,
      "loss": 0.0014,
      "step": 119130
    },
    {
      "epoch": 1.949439581117565,
      "grad_norm": 0.12463755160570145,
      "learning_rate": 3.2929808134352392e-06,
      "loss": 0.0019,
      "step": 119140
    },
    {
      "epoch": 1.949603207068641,
      "grad_norm": 0.04978432506322861,
      "learning_rate": 3.292086099990704e-06,
      "loss": 0.0007,
      "step": 119150
    },
    {
      "epoch": 1.9497668330197169,
      "grad_norm": 0.047517552971839905,
      "learning_rate": 3.291191448453941e-06,
      "loss": 0.0022,
      "step": 119160
    },
    {
      "epoch": 1.9499304589707926,
      "grad_norm": 0.1562863290309906,
      "learning_rate": 3.2902968588573823e-06,
      "loss": 0.0013,
      "step": 119170
    },
    {
      "epoch": 1.9500940849218686,
      "grad_norm": 0.037684086710214615,
      "learning_rate": 3.289402331233453e-06,
      "loss": 0.0012,
      "step": 119180
    },
    {
      "epoch": 1.9502577108729444,
      "grad_norm": 0.028352729976177216,
      "learning_rate": 3.288507865614581e-06,
      "loss": 0.0019,
      "step": 119190
    },
    {
      "epoch": 1.9504213368240202,
      "grad_norm": 0.024281544610857964,
      "learning_rate": 3.2876134620331825e-06,
      "loss": 0.0027,
      "step": 119200
    },
    {
      "epoch": 1.9505849627750962,
      "grad_norm": 0.08222485333681107,
      "learning_rate": 3.286719120521681e-06,
      "loss": 0.0012,
      "step": 119210
    },
    {
      "epoch": 1.950748588726172,
      "grad_norm": 0.09528180956840515,
      "learning_rate": 3.2858248411124916e-06,
      "loss": 0.0014,
      "step": 119220
    },
    {
      "epoch": 1.9509122146772477,
      "grad_norm": 0.09575015306472778,
      "learning_rate": 3.2849306238380335e-06,
      "loss": 0.003,
      "step": 119230
    },
    {
      "epoch": 1.9510758406283237,
      "grad_norm": 0.05369120091199875,
      "learning_rate": 3.284036468730718e-06,
      "loss": 0.0025,
      "step": 119240
    },
    {
      "epoch": 1.9512394665793995,
      "grad_norm": 0.05860856920480728,
      "learning_rate": 3.2831423758229565e-06,
      "loss": 0.0015,
      "step": 119250
    },
    {
      "epoch": 1.9514030925304753,
      "grad_norm": 0.06878620386123657,
      "learning_rate": 3.282248345147157e-06,
      "loss": 0.0017,
      "step": 119260
    },
    {
      "epoch": 1.9515667184815513,
      "grad_norm": 0.020936544984579086,
      "learning_rate": 3.281354376735727e-06,
      "loss": 0.0016,
      "step": 119270
    },
    {
      "epoch": 1.951730344432627,
      "grad_norm": 0.16774356365203857,
      "learning_rate": 3.2804604706210697e-06,
      "loss": 0.0013,
      "step": 119280
    },
    {
      "epoch": 1.9518939703837028,
      "grad_norm": 0.004553603008389473,
      "learning_rate": 3.2795666268355893e-06,
      "loss": 0.0008,
      "step": 119290
    },
    {
      "epoch": 1.9520575963347788,
      "grad_norm": 0.10118407011032104,
      "learning_rate": 3.2786728454116823e-06,
      "loss": 0.0015,
      "step": 119300
    },
    {
      "epoch": 1.9522212222858544,
      "grad_norm": 0.08749271929264069,
      "learning_rate": 3.2777791263817472e-06,
      "loss": 0.001,
      "step": 119310
    },
    {
      "epoch": 1.9523848482369304,
      "grad_norm": 0.01847228966653347,
      "learning_rate": 3.2768854697781806e-06,
      "loss": 0.0005,
      "step": 119320
    },
    {
      "epoch": 1.9525484741880064,
      "grad_norm": 0.0919080600142479,
      "learning_rate": 3.275991875633373e-06,
      "loss": 0.0008,
      "step": 119330
    },
    {
      "epoch": 1.952712100139082,
      "grad_norm": 0.07413464039564133,
      "learning_rate": 3.2750983439797167e-06,
      "loss": 0.0039,
      "step": 119340
    },
    {
      "epoch": 1.952875726090158,
      "grad_norm": 0.11682228744029999,
      "learning_rate": 3.2742048748496003e-06,
      "loss": 0.002,
      "step": 119350
    },
    {
      "epoch": 1.9530393520412337,
      "grad_norm": 0.07065902650356293,
      "learning_rate": 3.273311468275409e-06,
      "loss": 0.0009,
      "step": 119360
    },
    {
      "epoch": 1.9532029779923095,
      "grad_norm": 0.1162300705909729,
      "learning_rate": 3.272418124289527e-06,
      "loss": 0.0011,
      "step": 119370
    },
    {
      "epoch": 1.9533666039433855,
      "grad_norm": 0.19947503507137299,
      "learning_rate": 3.2715248429243353e-06,
      "loss": 0.0086,
      "step": 119380
    },
    {
      "epoch": 1.9535302298944612,
      "grad_norm": 0.18376384675502777,
      "learning_rate": 3.270631624212216e-06,
      "loss": 0.0024,
      "step": 119390
    },
    {
      "epoch": 1.953693855845537,
      "grad_norm": 0.03526970371603966,
      "learning_rate": 3.2697384681855426e-06,
      "loss": 0.0011,
      "step": 119400
    },
    {
      "epoch": 1.953857481796613,
      "grad_norm": 0.032023217529058456,
      "learning_rate": 3.2688453748766905e-06,
      "loss": 0.0014,
      "step": 119410
    },
    {
      "epoch": 1.9540211077476888,
      "grad_norm": 0.01755298301577568,
      "learning_rate": 3.267952344318033e-06,
      "loss": 0.001,
      "step": 119420
    },
    {
      "epoch": 1.9541847336987646,
      "grad_norm": 0.07165593653917313,
      "learning_rate": 3.2670593765419406e-06,
      "loss": 0.0018,
      "step": 119430
    },
    {
      "epoch": 1.9543483596498405,
      "grad_norm": 0.17985360324382782,
      "learning_rate": 3.2661664715807794e-06,
      "loss": 0.0012,
      "step": 119440
    },
    {
      "epoch": 1.9545119856009163,
      "grad_norm": 0.12182191014289856,
      "learning_rate": 3.2652736294669174e-06,
      "loss": 0.0011,
      "step": 119450
    },
    {
      "epoch": 1.954675611551992,
      "grad_norm": 0.04350307211279869,
      "learning_rate": 3.2643808502327175e-06,
      "loss": 0.0011,
      "step": 119460
    },
    {
      "epoch": 1.954839237503068,
      "grad_norm": 0.060890890657901764,
      "learning_rate": 3.2634881339105395e-06,
      "loss": 0.0025,
      "step": 119470
    },
    {
      "epoch": 1.9550028634541439,
      "grad_norm": 0.035087425261735916,
      "learning_rate": 3.2625954805327435e-06,
      "loss": 0.0011,
      "step": 119480
    },
    {
      "epoch": 1.9551664894052196,
      "grad_norm": 0.07437703758478165,
      "learning_rate": 3.261702890131687e-06,
      "loss": 0.0008,
      "step": 119490
    },
    {
      "epoch": 1.9553301153562956,
      "grad_norm": 0.04648261144757271,
      "learning_rate": 3.2608103627397214e-06,
      "loss": 0.002,
      "step": 119500
    },
    {
      "epoch": 1.9554937413073712,
      "grad_norm": 0.10275488346815109,
      "learning_rate": 3.2599178983892e-06,
      "loss": 0.0015,
      "step": 119510
    },
    {
      "epoch": 1.9556573672584472,
      "grad_norm": 0.0663665384054184,
      "learning_rate": 3.259025497112473e-06,
      "loss": 0.0027,
      "step": 119520
    },
    {
      "epoch": 1.9558209932095232,
      "grad_norm": 0.050408367067575455,
      "learning_rate": 3.2581331589418875e-06,
      "loss": 0.0017,
      "step": 119530
    },
    {
      "epoch": 1.9559846191605987,
      "grad_norm": 0.08994525671005249,
      "learning_rate": 3.257240883909788e-06,
      "loss": 0.0014,
      "step": 119540
    },
    {
      "epoch": 1.9561482451116747,
      "grad_norm": 0.07794841378927231,
      "learning_rate": 3.256348672048518e-06,
      "loss": 0.0016,
      "step": 119550
    },
    {
      "epoch": 1.9563118710627505,
      "grad_norm": 0.22953827679157257,
      "learning_rate": 3.2554565233904178e-06,
      "loss": 0.0017,
      "step": 119560
    },
    {
      "epoch": 1.9564754970138263,
      "grad_norm": 0.04843238741159439,
      "learning_rate": 3.254564437967826e-06,
      "loss": 0.0018,
      "step": 119570
    },
    {
      "epoch": 1.9566391229649023,
      "grad_norm": 0.09702663123607635,
      "learning_rate": 3.2536724158130782e-06,
      "loss": 0.0008,
      "step": 119580
    },
    {
      "epoch": 1.956802748915978,
      "grad_norm": 0.04243415221571922,
      "learning_rate": 3.252780456958509e-06,
      "loss": 0.0009,
      "step": 119590
    },
    {
      "epoch": 1.9569663748670538,
      "grad_norm": 0.3596123456954956,
      "learning_rate": 3.251888561436448e-06,
      "loss": 0.0015,
      "step": 119600
    },
    {
      "epoch": 1.9571300008181298,
      "grad_norm": 0.12265444546937943,
      "learning_rate": 3.2509967292792254e-06,
      "loss": 0.0029,
      "step": 119610
    },
    {
      "epoch": 1.9572936267692056,
      "grad_norm": 0.08800535649061203,
      "learning_rate": 3.250104960519167e-06,
      "loss": 0.0015,
      "step": 119620
    },
    {
      "epoch": 1.9574572527202814,
      "grad_norm": 0.027957957237958908,
      "learning_rate": 3.249213255188599e-06,
      "loss": 0.0009,
      "step": 119630
    },
    {
      "epoch": 1.9576208786713574,
      "grad_norm": 0.12052654474973679,
      "learning_rate": 3.248321613319842e-06,
      "loss": 0.0008,
      "step": 119640
    },
    {
      "epoch": 1.9577845046224331,
      "grad_norm": 0.06927067786455154,
      "learning_rate": 3.247430034945216e-06,
      "loss": 0.0022,
      "step": 119650
    },
    {
      "epoch": 1.957948130573509,
      "grad_norm": 0.01786019094288349,
      "learning_rate": 3.2465385200970388e-06,
      "loss": 0.0014,
      "step": 119660
    },
    {
      "epoch": 1.958111756524585,
      "grad_norm": 0.018616029992699623,
      "learning_rate": 3.245647068807627e-06,
      "loss": 0.0036,
      "step": 119670
    },
    {
      "epoch": 1.9582753824756607,
      "grad_norm": 0.061355095356702805,
      "learning_rate": 3.244755681109293e-06,
      "loss": 0.0015,
      "step": 119680
    },
    {
      "epoch": 1.9584390084267365,
      "grad_norm": 0.07513242959976196,
      "learning_rate": 3.2438643570343474e-06,
      "loss": 0.0004,
      "step": 119690
    },
    {
      "epoch": 1.9586026343778125,
      "grad_norm": 0.07908494770526886,
      "learning_rate": 3.2429730966150975e-06,
      "loss": 0.0014,
      "step": 119700
    },
    {
      "epoch": 1.958766260328888,
      "grad_norm": 0.08945582807064056,
      "learning_rate": 3.242081899883848e-06,
      "loss": 0.0015,
      "step": 119710
    },
    {
      "epoch": 1.958929886279964,
      "grad_norm": 0.12176748365163803,
      "learning_rate": 3.2411907668729057e-06,
      "loss": 0.001,
      "step": 119720
    },
    {
      "epoch": 1.95909351223104,
      "grad_norm": 0.005745396483689547,
      "learning_rate": 3.2402996976145713e-06,
      "loss": 0.0012,
      "step": 119730
    },
    {
      "epoch": 1.9592571381821156,
      "grad_norm": 0.07633748650550842,
      "learning_rate": 3.2394086921411433e-06,
      "loss": 0.001,
      "step": 119740
    },
    {
      "epoch": 1.9594207641331916,
      "grad_norm": 0.0820692852139473,
      "learning_rate": 3.2385177504849185e-06,
      "loss": 0.0012,
      "step": 119750
    },
    {
      "epoch": 1.9595843900842673,
      "grad_norm": 0.20230506360530853,
      "learning_rate": 3.2376268726781924e-06,
      "loss": 0.0012,
      "step": 119760
    },
    {
      "epoch": 1.959748016035343,
      "grad_norm": 0.07433783262968063,
      "learning_rate": 3.236736058753256e-06,
      "loss": 0.0009,
      "step": 119770
    },
    {
      "epoch": 1.959911641986419,
      "grad_norm": 0.044084712862968445,
      "learning_rate": 3.235845308742399e-06,
      "loss": 0.0014,
      "step": 119780
    },
    {
      "epoch": 1.9600752679374949,
      "grad_norm": 0.04565664008259773,
      "learning_rate": 3.2349546226779117e-06,
      "loss": 0.0012,
      "step": 119790
    },
    {
      "epoch": 1.9602388938885706,
      "grad_norm": 0.03038705885410309,
      "learning_rate": 3.2340640005920755e-06,
      "loss": 0.0015,
      "step": 119800
    },
    {
      "epoch": 1.9604025198396466,
      "grad_norm": 0.05366574227809906,
      "learning_rate": 3.2331734425171744e-06,
      "loss": 0.002,
      "step": 119810
    },
    {
      "epoch": 1.9605661457907224,
      "grad_norm": 0.0558791346848011,
      "learning_rate": 3.23228294848549e-06,
      "loss": 0.0008,
      "step": 119820
    },
    {
      "epoch": 1.9607297717417982,
      "grad_norm": 0.03427383676171303,
      "learning_rate": 3.2313925185293006e-06,
      "loss": 0.0012,
      "step": 119830
    },
    {
      "epoch": 1.9608933976928742,
      "grad_norm": 0.2161935418844223,
      "learning_rate": 3.230502152680881e-06,
      "loss": 0.0019,
      "step": 119840
    },
    {
      "epoch": 1.96105702364395,
      "grad_norm": 0.1068362295627594,
      "learning_rate": 3.229611850972506e-06,
      "loss": 0.0008,
      "step": 119850
    },
    {
      "epoch": 1.9612206495950257,
      "grad_norm": 0.09048112481832504,
      "learning_rate": 3.2287216134364465e-06,
      "loss": 0.0012,
      "step": 119860
    },
    {
      "epoch": 1.9613842755461017,
      "grad_norm": 0.006110770627856255,
      "learning_rate": 3.2278314401049694e-06,
      "loss": 0.0011,
      "step": 119870
    },
    {
      "epoch": 1.9615479014971775,
      "grad_norm": 0.01368734147399664,
      "learning_rate": 3.226941331010345e-06,
      "loss": 0.0011,
      "step": 119880
    },
    {
      "epoch": 1.9617115274482533,
      "grad_norm": 0.12932072579860687,
      "learning_rate": 3.2260512861848383e-06,
      "loss": 0.001,
      "step": 119890
    },
    {
      "epoch": 1.9618751533993293,
      "grad_norm": 0.024174556136131287,
      "learning_rate": 3.2251613056607062e-06,
      "loss": 0.0016,
      "step": 119900
    },
    {
      "epoch": 1.9620387793504048,
      "grad_norm": 0.0598149299621582,
      "learning_rate": 3.2242713894702118e-06,
      "loss": 0.0014,
      "step": 119910
    },
    {
      "epoch": 1.9622024053014808,
      "grad_norm": 0.025891564786434174,
      "learning_rate": 3.223381537645611e-06,
      "loss": 0.0006,
      "step": 119920
    },
    {
      "epoch": 1.9623660312525568,
      "grad_norm": 0.09268096834421158,
      "learning_rate": 3.2224917502191583e-06,
      "loss": 0.0007,
      "step": 119930
    },
    {
      "epoch": 1.9625296572036324,
      "grad_norm": 0.037256430834531784,
      "learning_rate": 3.2216020272231085e-06,
      "loss": 0.0014,
      "step": 119940
    },
    {
      "epoch": 1.9626932831547084,
      "grad_norm": 0.09555631130933762,
      "learning_rate": 3.2207123686897117e-06,
      "loss": 0.0019,
      "step": 119950
    },
    {
      "epoch": 1.9628569091057841,
      "grad_norm": 0.06838426738977432,
      "learning_rate": 3.2198227746512145e-06,
      "loss": 0.0016,
      "step": 119960
    },
    {
      "epoch": 1.96302053505686,
      "grad_norm": 0.05205368250608444,
      "learning_rate": 3.2189332451398638e-06,
      "loss": 0.0012,
      "step": 119970
    },
    {
      "epoch": 1.963184161007936,
      "grad_norm": 0.04782318323850632,
      "learning_rate": 3.2180437801879015e-06,
      "loss": 0.0011,
      "step": 119980
    },
    {
      "epoch": 1.9633477869590117,
      "grad_norm": 0.07910872250795364,
      "learning_rate": 3.217154379827571e-06,
      "loss": 0.0016,
      "step": 119990
    },
    {
      "epoch": 1.9635114129100875,
      "grad_norm": 0.22480346262454987,
      "learning_rate": 3.216265044091108e-06,
      "loss": 0.0014,
      "step": 120000
    },
    {
      "epoch": 1.9636750388611635,
      "grad_norm": 0.005954029969871044,
      "learning_rate": 3.2153757730107504e-06,
      "loss": 0.0008,
      "step": 120010
    },
    {
      "epoch": 1.9638386648122392,
      "grad_norm": 0.11898721754550934,
      "learning_rate": 3.2144865666187314e-06,
      "loss": 0.002,
      "step": 120020
    },
    {
      "epoch": 1.964002290763315,
      "grad_norm": 0.0021562569309026003,
      "learning_rate": 3.2135974249472827e-06,
      "loss": 0.0006,
      "step": 120030
    },
    {
      "epoch": 1.964165916714391,
      "grad_norm": 0.034869857132434845,
      "learning_rate": 3.2127083480286348e-06,
      "loss": 0.0005,
      "step": 120040
    },
    {
      "epoch": 1.9643295426654668,
      "grad_norm": 0.18674185872077942,
      "learning_rate": 3.211819335895013e-06,
      "loss": 0.0014,
      "step": 120050
    },
    {
      "epoch": 1.9644931686165426,
      "grad_norm": 0.09326320886611938,
      "learning_rate": 3.2109303885786427e-06,
      "loss": 0.0009,
      "step": 120060
    },
    {
      "epoch": 1.9646567945676185,
      "grad_norm": 0.03501841053366661,
      "learning_rate": 3.210041506111745e-06,
      "loss": 0.0009,
      "step": 120070
    },
    {
      "epoch": 1.964820420518694,
      "grad_norm": 0.07621312141418457,
      "learning_rate": 3.2091526885265415e-06,
      "loss": 0.0014,
      "step": 120080
    },
    {
      "epoch": 1.96498404646977,
      "grad_norm": 0.07037162780761719,
      "learning_rate": 3.2082639358552504e-06,
      "loss": 0.0026,
      "step": 120090
    },
    {
      "epoch": 1.965147672420846,
      "grad_norm": 0.11594855785369873,
      "learning_rate": 3.2073752481300835e-06,
      "loss": 0.001,
      "step": 120100
    },
    {
      "epoch": 1.9653112983719216,
      "grad_norm": 0.19071663916110992,
      "learning_rate": 3.2064866253832555e-06,
      "loss": 0.0013,
      "step": 120110
    },
    {
      "epoch": 1.9654749243229976,
      "grad_norm": 0.22957459092140198,
      "learning_rate": 3.2055980676469766e-06,
      "loss": 0.0019,
      "step": 120120
    },
    {
      "epoch": 1.9656385502740734,
      "grad_norm": 0.06698928773403168,
      "learning_rate": 3.2047095749534552e-06,
      "loss": 0.0022,
      "step": 120130
    },
    {
      "epoch": 1.9658021762251492,
      "grad_norm": 0.23116269707679749,
      "learning_rate": 3.203821147334897e-06,
      "loss": 0.0018,
      "step": 120140
    },
    {
      "epoch": 1.9659658021762252,
      "grad_norm": 0.060072775930166245,
      "learning_rate": 3.202932784823504e-06,
      "loss": 0.0021,
      "step": 120150
    },
    {
      "epoch": 1.966129428127301,
      "grad_norm": 0.004396048840135336,
      "learning_rate": 3.2020444874514793e-06,
      "loss": 0.0015,
      "step": 120160
    },
    {
      "epoch": 1.9662930540783767,
      "grad_norm": 0.10115991532802582,
      "learning_rate": 3.201156255251021e-06,
      "loss": 0.0014,
      "step": 120170
    },
    {
      "epoch": 1.9664566800294527,
      "grad_norm": 0.021288610994815826,
      "learning_rate": 3.2002680882543246e-06,
      "loss": 0.0008,
      "step": 120180
    },
    {
      "epoch": 1.9666203059805285,
      "grad_norm": 0.04749828577041626,
      "learning_rate": 3.199379986493587e-06,
      "loss": 0.001,
      "step": 120190
    },
    {
      "epoch": 1.9667839319316043,
      "grad_norm": 0.11034960299730301,
      "learning_rate": 3.1984919500009946e-06,
      "loss": 0.0019,
      "step": 120200
    },
    {
      "epoch": 1.9669475578826803,
      "grad_norm": 0.17236697673797607,
      "learning_rate": 3.197603978808739e-06,
      "loss": 0.0018,
      "step": 120210
    },
    {
      "epoch": 1.967111183833756,
      "grad_norm": 0.3318023383617401,
      "learning_rate": 3.1967160729490078e-06,
      "loss": 0.0014,
      "step": 120220
    },
    {
      "epoch": 1.9672748097848318,
      "grad_norm": 0.005125746130943298,
      "learning_rate": 3.1958282324539846e-06,
      "loss": 0.0011,
      "step": 120230
    },
    {
      "epoch": 1.9674384357359078,
      "grad_norm": 0.04570695012807846,
      "learning_rate": 3.194940457355852e-06,
      "loss": 0.0013,
      "step": 120240
    },
    {
      "epoch": 1.9676020616869836,
      "grad_norm": 0.13440853357315063,
      "learning_rate": 3.19405274768679e-06,
      "loss": 0.0011,
      "step": 120250
    },
    {
      "epoch": 1.9677656876380594,
      "grad_norm": 0.3376067876815796,
      "learning_rate": 3.1931651034789746e-06,
      "loss": 0.0025,
      "step": 120260
    },
    {
      "epoch": 1.9679293135891354,
      "grad_norm": 0.14046721160411835,
      "learning_rate": 3.192277524764582e-06,
      "loss": 0.0012,
      "step": 120270
    },
    {
      "epoch": 1.968092939540211,
      "grad_norm": 0.02051129750907421,
      "learning_rate": 3.191390011575784e-06,
      "loss": 0.0006,
      "step": 120280
    },
    {
      "epoch": 1.968256565491287,
      "grad_norm": 0.0698816180229187,
      "learning_rate": 3.1905025639447525e-06,
      "loss": 0.0012,
      "step": 120290
    },
    {
      "epoch": 1.968420191442363,
      "grad_norm": 0.09853628277778625,
      "learning_rate": 3.189615181903653e-06,
      "loss": 0.0007,
      "step": 120300
    },
    {
      "epoch": 1.9685838173934385,
      "grad_norm": 0.04063701629638672,
      "learning_rate": 3.188727865484652e-06,
      "loss": 0.0011,
      "step": 120310
    },
    {
      "epoch": 1.9687474433445145,
      "grad_norm": 0.04318047687411308,
      "learning_rate": 3.1878406147199127e-06,
      "loss": 0.001,
      "step": 120320
    },
    {
      "epoch": 1.9689110692955902,
      "grad_norm": 0.12426009774208069,
      "learning_rate": 3.1869534296415954e-06,
      "loss": 0.0012,
      "step": 120330
    },
    {
      "epoch": 1.969074695246666,
      "grad_norm": 0.08373285084962845,
      "learning_rate": 3.186066310281859e-06,
      "loss": 0.0006,
      "step": 120340
    },
    {
      "epoch": 1.969238321197742,
      "grad_norm": 0.0494953952729702,
      "learning_rate": 3.185179256672859e-06,
      "loss": 0.0021,
      "step": 120350
    },
    {
      "epoch": 1.9694019471488178,
      "grad_norm": 0.2756549119949341,
      "learning_rate": 3.1842922688467493e-06,
      "loss": 0.0022,
      "step": 120360
    },
    {
      "epoch": 1.9695655730998936,
      "grad_norm": 0.1246170699596405,
      "learning_rate": 3.1834053468356797e-06,
      "loss": 0.0017,
      "step": 120370
    },
    {
      "epoch": 1.9697291990509695,
      "grad_norm": 0.026286819949746132,
      "learning_rate": 3.182518490671802e-06,
      "loss": 0.0023,
      "step": 120380
    },
    {
      "epoch": 1.9698928250020453,
      "grad_norm": 0.17359715700149536,
      "learning_rate": 3.181631700387262e-06,
      "loss": 0.0015,
      "step": 120390
    },
    {
      "epoch": 1.970056450953121,
      "grad_norm": 0.06552699953317642,
      "learning_rate": 3.1807449760142005e-06,
      "loss": 0.002,
      "step": 120400
    },
    {
      "epoch": 1.970220076904197,
      "grad_norm": 0.0627862960100174,
      "learning_rate": 3.179858317584762e-06,
      "loss": 0.0015,
      "step": 120410
    },
    {
      "epoch": 1.9703837028552729,
      "grad_norm": 0.00994772370904684,
      "learning_rate": 3.1789717251310836e-06,
      "loss": 0.0014,
      "step": 120420
    },
    {
      "epoch": 1.9705473288063486,
      "grad_norm": 0.07194051146507263,
      "learning_rate": 3.1780851986853034e-06,
      "loss": 0.0011,
      "step": 120430
    },
    {
      "epoch": 1.9707109547574246,
      "grad_norm": 0.08550158143043518,
      "learning_rate": 3.1771987382795566e-06,
      "loss": 0.0015,
      "step": 120440
    },
    {
      "epoch": 1.9708745807085004,
      "grad_norm": 0.044907134026288986,
      "learning_rate": 3.1763123439459744e-06,
      "loss": 0.0023,
      "step": 120450
    },
    {
      "epoch": 1.9710382066595762,
      "grad_norm": 0.21220263838768005,
      "learning_rate": 3.175426015716687e-06,
      "loss": 0.0014,
      "step": 120460
    },
    {
      "epoch": 1.9712018326106522,
      "grad_norm": 0.11774542927742004,
      "learning_rate": 3.174539753623821e-06,
      "loss": 0.0018,
      "step": 120470
    },
    {
      "epoch": 1.9713654585617277,
      "grad_norm": 0.07862376421689987,
      "learning_rate": 3.1736535576995016e-06,
      "loss": 0.0009,
      "step": 120480
    },
    {
      "epoch": 1.9715290845128037,
      "grad_norm": 0.03766781836748123,
      "learning_rate": 3.1727674279758523e-06,
      "loss": 0.0009,
      "step": 120490
    },
    {
      "epoch": 1.9716927104638797,
      "grad_norm": 0.05893450602889061,
      "learning_rate": 3.171881364484991e-06,
      "loss": 0.0016,
      "step": 120500
    },
    {
      "epoch": 1.9718563364149553,
      "grad_norm": 0.03269325569272041,
      "learning_rate": 3.1709953672590367e-06,
      "loss": 0.0015,
      "step": 120510
    },
    {
      "epoch": 1.9720199623660313,
      "grad_norm": 0.03670454025268555,
      "learning_rate": 3.170109436330104e-06,
      "loss": 0.0014,
      "step": 120520
    },
    {
      "epoch": 1.972183588317107,
      "grad_norm": 0.04544687643647194,
      "learning_rate": 3.1692235717303064e-06,
      "loss": 0.0018,
      "step": 120530
    },
    {
      "epoch": 1.9723472142681828,
      "grad_norm": 0.11783545464277267,
      "learning_rate": 3.1683377734917542e-06,
      "loss": 0.001,
      "step": 120540
    },
    {
      "epoch": 1.9725108402192588,
      "grad_norm": 0.04126141965389252,
      "learning_rate": 3.1674520416465552e-06,
      "loss": 0.0009,
      "step": 120550
    },
    {
      "epoch": 1.9726744661703346,
      "grad_norm": 0.010780500248074532,
      "learning_rate": 3.166566376226816e-06,
      "loss": 0.0014,
      "step": 120560
    },
    {
      "epoch": 1.9728380921214104,
      "grad_norm": 0.2720125615596771,
      "learning_rate": 3.165680777264638e-06,
      "loss": 0.0021,
      "step": 120570
    },
    {
      "epoch": 1.9730017180724864,
      "grad_norm": 0.07383672893047333,
      "learning_rate": 3.164795244792124e-06,
      "loss": 0.0011,
      "step": 120580
    },
    {
      "epoch": 1.9731653440235621,
      "grad_norm": 0.1570451408624649,
      "learning_rate": 3.1639097788413727e-06,
      "loss": 0.0018,
      "step": 120590
    },
    {
      "epoch": 1.973328969974638,
      "grad_norm": 0.059065233916044235,
      "learning_rate": 3.1630243794444772e-06,
      "loss": 0.0026,
      "step": 120600
    },
    {
      "epoch": 1.973492595925714,
      "grad_norm": 0.06240195780992508,
      "learning_rate": 3.1621390466335334e-06,
      "loss": 0.0016,
      "step": 120610
    },
    {
      "epoch": 1.9736562218767897,
      "grad_norm": 0.016657231375575066,
      "learning_rate": 3.161253780440632e-06,
      "loss": 0.0005,
      "step": 120620
    },
    {
      "epoch": 1.9738198478278655,
      "grad_norm": 0.1470792591571808,
      "learning_rate": 3.160368580897862e-06,
      "loss": 0.0021,
      "step": 120630
    },
    {
      "epoch": 1.9739834737789415,
      "grad_norm": 0.14704667031764984,
      "learning_rate": 3.1594834480373093e-06,
      "loss": 0.0012,
      "step": 120640
    },
    {
      "epoch": 1.9741470997300172,
      "grad_norm": 0.19293953478336334,
      "learning_rate": 3.1585983818910563e-06,
      "loss": 0.0016,
      "step": 120650
    },
    {
      "epoch": 1.974310725681093,
      "grad_norm": 0.07996378093957901,
      "learning_rate": 3.1577133824911884e-06,
      "loss": 0.0014,
      "step": 120660
    },
    {
      "epoch": 1.974474351632169,
      "grad_norm": 0.056830473244190216,
      "learning_rate": 3.156828449869782e-06,
      "loss": 0.0011,
      "step": 120670
    },
    {
      "epoch": 1.9746379775832446,
      "grad_norm": 0.051241327077150345,
      "learning_rate": 3.1559435840589144e-06,
      "loss": 0.0024,
      "step": 120680
    },
    {
      "epoch": 1.9748016035343205,
      "grad_norm": 0.04359995946288109,
      "learning_rate": 3.1550587850906617e-06,
      "loss": 0.0022,
      "step": 120690
    },
    {
      "epoch": 1.9749652294853965,
      "grad_norm": 0.17184430360794067,
      "learning_rate": 3.154174052997091e-06,
      "loss": 0.0014,
      "step": 120700
    },
    {
      "epoch": 1.975128855436472,
      "grad_norm": 0.18006198108196259,
      "learning_rate": 3.1532893878102756e-06,
      "loss": 0.0017,
      "step": 120710
    },
    {
      "epoch": 1.975292481387548,
      "grad_norm": 0.03541097790002823,
      "learning_rate": 3.1524047895622812e-06,
      "loss": 0.003,
      "step": 120720
    },
    {
      "epoch": 1.9754561073386239,
      "grad_norm": 0.06656794250011444,
      "learning_rate": 3.151520258285172e-06,
      "loss": 0.0006,
      "step": 120730
    },
    {
      "epoch": 1.9756197332896996,
      "grad_norm": 0.03277677670121193,
      "learning_rate": 3.1506357940110123e-06,
      "loss": 0.0015,
      "step": 120740
    },
    {
      "epoch": 1.9757833592407756,
      "grad_norm": 0.04108794033527374,
      "learning_rate": 3.1497513967718594e-06,
      "loss": 0.0018,
      "step": 120750
    },
    {
      "epoch": 1.9759469851918514,
      "grad_norm": 0.06622787564992905,
      "learning_rate": 3.148867066599771e-06,
      "loss": 0.0009,
      "step": 120760
    },
    {
      "epoch": 1.9761106111429272,
      "grad_norm": 0.08014035224914551,
      "learning_rate": 3.147982803526803e-06,
      "loss": 0.0006,
      "step": 120770
    },
    {
      "epoch": 1.9762742370940032,
      "grad_norm": 0.12275131791830063,
      "learning_rate": 3.1470986075850067e-06,
      "loss": 0.0011,
      "step": 120780
    },
    {
      "epoch": 1.976437863045079,
      "grad_norm": 0.13255465030670166,
      "learning_rate": 3.146214478806434e-06,
      "loss": 0.0009,
      "step": 120790
    },
    {
      "epoch": 1.9766014889961547,
      "grad_norm": 0.047800809144973755,
      "learning_rate": 3.1453304172231295e-06,
      "loss": 0.0012,
      "step": 120800
    },
    {
      "epoch": 1.9767651149472307,
      "grad_norm": 0.06365638226270676,
      "learning_rate": 3.14444642286714e-06,
      "loss": 0.0021,
      "step": 120810
    },
    {
      "epoch": 1.9769287408983065,
      "grad_norm": 0.14298172295093536,
      "learning_rate": 3.143562495770508e-06,
      "loss": 0.0012,
      "step": 120820
    },
    {
      "epoch": 1.9770923668493823,
      "grad_norm": 0.10809532552957535,
      "learning_rate": 3.1426786359652738e-06,
      "loss": 0.0021,
      "step": 120830
    },
    {
      "epoch": 1.9772559928004583,
      "grad_norm": 0.0827135443687439,
      "learning_rate": 3.1417948434834756e-06,
      "loss": 0.0018,
      "step": 120840
    },
    {
      "epoch": 1.977419618751534,
      "grad_norm": 0.06294337660074234,
      "learning_rate": 3.1409111183571474e-06,
      "loss": 0.0014,
      "step": 120850
    },
    {
      "epoch": 1.9775832447026098,
      "grad_norm": 0.11825191229581833,
      "learning_rate": 3.1400274606183223e-06,
      "loss": 0.0029,
      "step": 120860
    },
    {
      "epoch": 1.9777468706536858,
      "grad_norm": 0.07278622686862946,
      "learning_rate": 3.1391438702990325e-06,
      "loss": 0.0019,
      "step": 120870
    },
    {
      "epoch": 1.9779104966047614,
      "grad_norm": 0.03915053978562355,
      "learning_rate": 3.1382603474313044e-06,
      "loss": 0.0011,
      "step": 120880
    },
    {
      "epoch": 1.9780741225558374,
      "grad_norm": 0.13825000822544098,
      "learning_rate": 3.137376892047167e-06,
      "loss": 0.0017,
      "step": 120890
    },
    {
      "epoch": 1.9782377485069134,
      "grad_norm": 0.15757323801517487,
      "learning_rate": 3.136493504178638e-06,
      "loss": 0.0009,
      "step": 120900
    },
    {
      "epoch": 1.978401374457989,
      "grad_norm": 0.1990974396467209,
      "learning_rate": 3.13561018385774e-06,
      "loss": 0.0015,
      "step": 120910
    },
    {
      "epoch": 1.978565000409065,
      "grad_norm": 0.04972890019416809,
      "learning_rate": 3.134726931116492e-06,
      "loss": 0.001,
      "step": 120920
    },
    {
      "epoch": 1.9787286263601407,
      "grad_norm": 0.053142745047807693,
      "learning_rate": 3.133843745986911e-06,
      "loss": 0.0018,
      "step": 120930
    },
    {
      "epoch": 1.9788922523112165,
      "grad_norm": 0.01802622526884079,
      "learning_rate": 3.132960628501008e-06,
      "loss": 0.0013,
      "step": 120940
    },
    {
      "epoch": 1.9790558782622925,
      "grad_norm": 0.07915715128183365,
      "learning_rate": 3.1320775786907954e-06,
      "loss": 0.0008,
      "step": 120950
    },
    {
      "epoch": 1.9792195042133682,
      "grad_norm": 0.06082886829972267,
      "learning_rate": 3.1311945965882805e-06,
      "loss": 0.0017,
      "step": 120960
    },
    {
      "epoch": 1.979383130164444,
      "grad_norm": 0.12227126955986023,
      "learning_rate": 3.130311682225471e-06,
      "loss": 0.0023,
      "step": 120970
    },
    {
      "epoch": 1.97954675611552,
      "grad_norm": 0.15932504832744598,
      "learning_rate": 3.129428835634368e-06,
      "loss": 0.0019,
      "step": 120980
    },
    {
      "epoch": 1.9797103820665958,
      "grad_norm": 0.049096036702394485,
      "learning_rate": 3.128546056846976e-06,
      "loss": 0.0009,
      "step": 120990
    },
    {
      "epoch": 1.9798740080176715,
      "grad_norm": 0.05467770993709564,
      "learning_rate": 3.1276633458952897e-06,
      "loss": 0.0023,
      "step": 121000
    },
    {
      "epoch": 1.9800376339687475,
      "grad_norm": 0.16764862835407257,
      "learning_rate": 3.1267807028113084e-06,
      "loss": 0.0018,
      "step": 121010
    },
    {
      "epoch": 1.9802012599198233,
      "grad_norm": 0.2800367772579193,
      "learning_rate": 3.125898127627023e-06,
      "loss": 0.0023,
      "step": 121020
    },
    {
      "epoch": 1.980364885870899,
      "grad_norm": 0.03497200459241867,
      "learning_rate": 3.125015620374427e-06,
      "loss": 0.0007,
      "step": 121030
    },
    {
      "epoch": 1.980528511821975,
      "grad_norm": 0.03715066611766815,
      "learning_rate": 3.124133181085508e-06,
      "loss": 0.0008,
      "step": 121040
    },
    {
      "epoch": 1.9806921377730506,
      "grad_norm": 0.06147247552871704,
      "learning_rate": 3.123250809792253e-06,
      "loss": 0.0008,
      "step": 121050
    },
    {
      "epoch": 1.9808557637241266,
      "grad_norm": 0.08657212555408478,
      "learning_rate": 3.122368506526645e-06,
      "loss": 0.002,
      "step": 121060
    },
    {
      "epoch": 1.9810193896752026,
      "grad_norm": 0.07349998503923416,
      "learning_rate": 3.1214862713206663e-06,
      "loss": 0.0014,
      "step": 121070
    },
    {
      "epoch": 1.9811830156262782,
      "grad_norm": 0.08907697349786758,
      "learning_rate": 3.120604104206294e-06,
      "loss": 0.0008,
      "step": 121080
    },
    {
      "epoch": 1.9813466415773542,
      "grad_norm": 0.0578862726688385,
      "learning_rate": 3.11972200521551e-06,
      "loss": 0.0018,
      "step": 121090
    },
    {
      "epoch": 1.98151026752843,
      "grad_norm": 0.12811973690986633,
      "learning_rate": 3.118839974380281e-06,
      "loss": 0.0011,
      "step": 121100
    },
    {
      "epoch": 1.9816738934795057,
      "grad_norm": 0.0376611165702343,
      "learning_rate": 3.1179580117325817e-06,
      "loss": 0.0007,
      "step": 121110
    },
    {
      "epoch": 1.9818375194305817,
      "grad_norm": 0.03496914356946945,
      "learning_rate": 3.1170761173043816e-06,
      "loss": 0.001,
      "step": 121120
    },
    {
      "epoch": 1.9820011453816575,
      "grad_norm": 0.1055348739027977,
      "learning_rate": 3.1161942911276467e-06,
      "loss": 0.0011,
      "step": 121130
    },
    {
      "epoch": 1.9821647713327333,
      "grad_norm": 0.040344804525375366,
      "learning_rate": 3.11531253323434e-06,
      "loss": 0.0011,
      "step": 121140
    },
    {
      "epoch": 1.9823283972838093,
      "grad_norm": 0.02520809881389141,
      "learning_rate": 3.114430843656425e-06,
      "loss": 0.001,
      "step": 121150
    },
    {
      "epoch": 1.982492023234885,
      "grad_norm": 0.11748852580785751,
      "learning_rate": 3.11354922242586e-06,
      "loss": 0.0012,
      "step": 121160
    },
    {
      "epoch": 1.9826556491859608,
      "grad_norm": 0.23024903237819672,
      "learning_rate": 3.112667669574603e-06,
      "loss": 0.0016,
      "step": 121170
    },
    {
      "epoch": 1.9828192751370368,
      "grad_norm": 0.006447488442063332,
      "learning_rate": 3.1117861851346056e-06,
      "loss": 0.0007,
      "step": 121180
    },
    {
      "epoch": 1.9829829010881126,
      "grad_norm": 0.07881323248147964,
      "learning_rate": 3.110904769137823e-06,
      "loss": 0.0021,
      "step": 121190
    },
    {
      "epoch": 1.9831465270391884,
      "grad_norm": 0.12419156730175018,
      "learning_rate": 3.1100234216161995e-06,
      "loss": 0.0025,
      "step": 121200
    },
    {
      "epoch": 1.9833101529902644,
      "grad_norm": 0.015714025124907494,
      "learning_rate": 3.1091421426016854e-06,
      "loss": 0.0015,
      "step": 121210
    },
    {
      "epoch": 1.9834737789413401,
      "grad_norm": 0.06415874511003494,
      "learning_rate": 3.1082609321262237e-06,
      "loss": 0.0015,
      "step": 121220
    },
    {
      "epoch": 1.983637404892416,
      "grad_norm": 0.00596196623519063,
      "learning_rate": 3.1073797902217573e-06,
      "loss": 0.0021,
      "step": 121230
    },
    {
      "epoch": 1.983801030843492,
      "grad_norm": 0.08138204365968704,
      "learning_rate": 3.1064987169202245e-06,
      "loss": 0.0013,
      "step": 121240
    },
    {
      "epoch": 1.9839646567945675,
      "grad_norm": 0.10327907651662827,
      "learning_rate": 3.1056177122535613e-06,
      "loss": 0.0011,
      "step": 121250
    },
    {
      "epoch": 1.9841282827456435,
      "grad_norm": 0.09449491649866104,
      "learning_rate": 3.104736776253704e-06,
      "loss": 0.0016,
      "step": 121260
    },
    {
      "epoch": 1.9842919086967195,
      "grad_norm": 0.4243868887424469,
      "learning_rate": 3.103855908952583e-06,
      "loss": 0.003,
      "step": 121270
    },
    {
      "epoch": 1.984455534647795,
      "grad_norm": 0.10233750939369202,
      "learning_rate": 3.1029751103821286e-06,
      "loss": 0.0031,
      "step": 121280
    },
    {
      "epoch": 1.984619160598871,
      "grad_norm": 0.01858137920498848,
      "learning_rate": 3.1020943805742666e-06,
      "loss": 0.0011,
      "step": 121290
    },
    {
      "epoch": 1.9847827865499468,
      "grad_norm": 0.15394708514213562,
      "learning_rate": 3.1012137195609234e-06,
      "loss": 0.0027,
      "step": 121300
    },
    {
      "epoch": 1.9849464125010226,
      "grad_norm": 0.01425251830369234,
      "learning_rate": 3.1003331273740178e-06,
      "loss": 0.0007,
      "step": 121310
    },
    {
      "epoch": 1.9851100384520985,
      "grad_norm": 0.37392958998680115,
      "learning_rate": 3.09945260404547e-06,
      "loss": 0.0014,
      "step": 121320
    },
    {
      "epoch": 1.9852736644031743,
      "grad_norm": 0.1045062318444252,
      "learning_rate": 3.0985721496071985e-06,
      "loss": 0.0025,
      "step": 121330
    },
    {
      "epoch": 1.98543729035425,
      "grad_norm": 0.07641514390707016,
      "learning_rate": 3.097691764091116e-06,
      "loss": 0.001,
      "step": 121340
    },
    {
      "epoch": 1.985600916305326,
      "grad_norm": 0.08478570729494095,
      "learning_rate": 3.096811447529135e-06,
      "loss": 0.0009,
      "step": 121350
    },
    {
      "epoch": 1.9857645422564019,
      "grad_norm": 0.39634406566619873,
      "learning_rate": 3.0959311999531637e-06,
      "loss": 0.0022,
      "step": 121360
    },
    {
      "epoch": 1.9859281682074776,
      "grad_norm": 0.06185883283615112,
      "learning_rate": 3.095051021395111e-06,
      "loss": 0.001,
      "step": 121370
    },
    {
      "epoch": 1.9860917941585536,
      "grad_norm": 0.08347039669752121,
      "learning_rate": 3.094170911886881e-06,
      "loss": 0.0012,
      "step": 121380
    },
    {
      "epoch": 1.9862554201096294,
      "grad_norm": 0.031031904742121696,
      "learning_rate": 3.093290871460375e-06,
      "loss": 0.0009,
      "step": 121390
    },
    {
      "epoch": 1.9864190460607052,
      "grad_norm": 0.08313092589378357,
      "learning_rate": 3.092410900147494e-06,
      "loss": 0.0014,
      "step": 121400
    },
    {
      "epoch": 1.9865826720117812,
      "grad_norm": 0.0425654873251915,
      "learning_rate": 3.091530997980131e-06,
      "loss": 0.001,
      "step": 121410
    },
    {
      "epoch": 1.986746297962857,
      "grad_norm": 0.11481112986803055,
      "learning_rate": 3.0906511649901817e-06,
      "loss": 0.0017,
      "step": 121420
    },
    {
      "epoch": 1.9869099239139327,
      "grad_norm": 0.14661121368408203,
      "learning_rate": 3.0897714012095394e-06,
      "loss": 0.0021,
      "step": 121430
    },
    {
      "epoch": 1.9870735498650087,
      "grad_norm": 0.01511545293033123,
      "learning_rate": 3.088891706670092e-06,
      "loss": 0.0012,
      "step": 121440
    },
    {
      "epoch": 1.9872371758160843,
      "grad_norm": 0.007139799650758505,
      "learning_rate": 3.0880120814037277e-06,
      "loss": 0.0015,
      "step": 121450
    },
    {
      "epoch": 1.9874008017671603,
      "grad_norm": 0.10353057086467743,
      "learning_rate": 3.08713252544233e-06,
      "loss": 0.001,
      "step": 121460
    },
    {
      "epoch": 1.9875644277182363,
      "grad_norm": 0.04498282074928284,
      "learning_rate": 3.086253038817781e-06,
      "loss": 0.0006,
      "step": 121470
    },
    {
      "epoch": 1.9877280536693118,
      "grad_norm": 0.11789406836032867,
      "learning_rate": 3.0853736215619594e-06,
      "loss": 0.0016,
      "step": 121480
    },
    {
      "epoch": 1.9878916796203878,
      "grad_norm": 0.03144877776503563,
      "learning_rate": 3.084494273706743e-06,
      "loss": 0.0012,
      "step": 121490
    },
    {
      "epoch": 1.9880553055714636,
      "grad_norm": 0.055184394121170044,
      "learning_rate": 3.0836149952840054e-06,
      "loss": 0.0011,
      "step": 121500
    },
    {
      "epoch": 1.9882189315225394,
      "grad_norm": 0.10970151424407959,
      "learning_rate": 3.082735786325618e-06,
      "loss": 0.0013,
      "step": 121510
    },
    {
      "epoch": 1.9883825574736154,
      "grad_norm": 0.11161845177412033,
      "learning_rate": 3.0818566468634503e-06,
      "loss": 0.0013,
      "step": 121520
    },
    {
      "epoch": 1.9885461834246911,
      "grad_norm": 0.15732745826244354,
      "learning_rate": 3.0809775769293694e-06,
      "loss": 0.0007,
      "step": 121530
    },
    {
      "epoch": 1.988709809375767,
      "grad_norm": 0.07671280950307846,
      "learning_rate": 3.0800985765552384e-06,
      "loss": 0.0011,
      "step": 121540
    },
    {
      "epoch": 1.988873435326843,
      "grad_norm": 0.16346527636051178,
      "learning_rate": 3.0792196457729194e-06,
      "loss": 0.0007,
      "step": 121550
    },
    {
      "epoch": 1.9890370612779187,
      "grad_norm": 0.051224809139966965,
      "learning_rate": 3.078340784614272e-06,
      "loss": 0.0014,
      "step": 121560
    },
    {
      "epoch": 1.9892006872289945,
      "grad_norm": 0.03920578584074974,
      "learning_rate": 3.0774619931111534e-06,
      "loss": 0.0024,
      "step": 121570
    },
    {
      "epoch": 1.9893643131800705,
      "grad_norm": 0.14436499774456024,
      "learning_rate": 3.0765832712954145e-06,
      "loss": 0.0012,
      "step": 121580
    },
    {
      "epoch": 1.9895279391311462,
      "grad_norm": 0.06951668113470078,
      "learning_rate": 3.0757046191989117e-06,
      "loss": 0.0009,
      "step": 121590
    },
    {
      "epoch": 1.989691565082222,
      "grad_norm": 0.08532749861478806,
      "learning_rate": 3.0748260368534926e-06,
      "loss": 0.0019,
      "step": 121600
    },
    {
      "epoch": 1.989855191033298,
      "grad_norm": 0.04310256987810135,
      "learning_rate": 3.073947524291001e-06,
      "loss": 0.0014,
      "step": 121610
    },
    {
      "epoch": 1.9900188169843738,
      "grad_norm": 0.11547352373600006,
      "learning_rate": 3.0730690815432817e-06,
      "loss": 0.0013,
      "step": 121620
    },
    {
      "epoch": 1.9901824429354495,
      "grad_norm": 0.035640716552734375,
      "learning_rate": 3.0721907086421778e-06,
      "loss": 0.0018,
      "step": 121630
    },
    {
      "epoch": 1.9903460688865255,
      "grad_norm": 0.02316875010728836,
      "learning_rate": 3.0713124056195264e-06,
      "loss": 0.0022,
      "step": 121640
    },
    {
      "epoch": 1.990509694837601,
      "grad_norm": 0.025657478719949722,
      "learning_rate": 3.070434172507165e-06,
      "loss": 0.001,
      "step": 121650
    },
    {
      "epoch": 1.990673320788677,
      "grad_norm": 0.13552971184253693,
      "learning_rate": 3.069556009336928e-06,
      "loss": 0.0016,
      "step": 121660
    },
    {
      "epoch": 1.990836946739753,
      "grad_norm": 0.08910717815160751,
      "learning_rate": 3.0686779161406455e-06,
      "loss": 0.0008,
      "step": 121670
    },
    {
      "epoch": 1.9910005726908286,
      "grad_norm": 0.1187802329659462,
      "learning_rate": 3.0677998929501475e-06,
      "loss": 0.0015,
      "step": 121680
    },
    {
      "epoch": 1.9911641986419046,
      "grad_norm": 0.11421894282102585,
      "learning_rate": 3.066921939797258e-06,
      "loss": 0.0018,
      "step": 121690
    },
    {
      "epoch": 1.9913278245929804,
      "grad_norm": 0.0626273825764656,
      "learning_rate": 3.066044056713805e-06,
      "loss": 0.0009,
      "step": 121700
    },
    {
      "epoch": 1.9914914505440562,
      "grad_norm": 0.03288870304822922,
      "learning_rate": 3.0651662437316043e-06,
      "loss": 0.0013,
      "step": 121710
    },
    {
      "epoch": 1.9916550764951322,
      "grad_norm": 0.019862784072756767,
      "learning_rate": 3.0642885008824773e-06,
      "loss": 0.0013,
      "step": 121720
    },
    {
      "epoch": 1.991818702446208,
      "grad_norm": 0.21904274821281433,
      "learning_rate": 3.0634108281982405e-06,
      "loss": 0.0014,
      "step": 121730
    },
    {
      "epoch": 1.9919823283972837,
      "grad_norm": 0.07169432193040848,
      "learning_rate": 3.0625332257107064e-06,
      "loss": 0.0009,
      "step": 121740
    },
    {
      "epoch": 1.9921459543483597,
      "grad_norm": 0.09033267199993134,
      "learning_rate": 3.061655693451686e-06,
      "loss": 0.0024,
      "step": 121750
    },
    {
      "epoch": 1.9923095802994355,
      "grad_norm": 0.12846073508262634,
      "learning_rate": 3.060778231452988e-06,
      "loss": 0.0016,
      "step": 121760
    },
    {
      "epoch": 1.9924732062505113,
      "grad_norm": 0.05491194128990173,
      "learning_rate": 3.0599008397464185e-06,
      "loss": 0.0018,
      "step": 121770
    },
    {
      "epoch": 1.9926368322015873,
      "grad_norm": 0.11106289178133011,
      "learning_rate": 3.0590235183637806e-06,
      "loss": 0.0018,
      "step": 121780
    },
    {
      "epoch": 1.992800458152663,
      "grad_norm": 0.023505032062530518,
      "learning_rate": 3.0581462673368754e-06,
      "loss": 0.0014,
      "step": 121790
    },
    {
      "epoch": 1.9929640841037388,
      "grad_norm": 0.011300702579319477,
      "learning_rate": 3.057269086697502e-06,
      "loss": 0.001,
      "step": 121800
    },
    {
      "epoch": 1.9931277100548148,
      "grad_norm": 0.11059733480215073,
      "learning_rate": 3.056391976477453e-06,
      "loss": 0.0018,
      "step": 121810
    },
    {
      "epoch": 1.9932913360058904,
      "grad_norm": 0.11768092960119247,
      "learning_rate": 3.0555149367085246e-06,
      "loss": 0.0017,
      "step": 121820
    },
    {
      "epoch": 1.9934549619569664,
      "grad_norm": 0.1224488690495491,
      "learning_rate": 3.0546379674225057e-06,
      "loss": 0.0011,
      "step": 121830
    },
    {
      "epoch": 1.9936185879080424,
      "grad_norm": 0.051735639572143555,
      "learning_rate": 3.0537610686511855e-06,
      "loss": 0.0005,
      "step": 121840
    },
    {
      "epoch": 1.993782213859118,
      "grad_norm": 0.04027004912495613,
      "learning_rate": 3.052884240426348e-06,
      "loss": 0.0024,
      "step": 121850
    },
    {
      "epoch": 1.993945839810194,
      "grad_norm": 0.03983096405863762,
      "learning_rate": 3.0520074827797764e-06,
      "loss": 0.0011,
      "step": 121860
    },
    {
      "epoch": 1.9941094657612697,
      "grad_norm": 0.08986416459083557,
      "learning_rate": 3.0511307957432527e-06,
      "loss": 0.0011,
      "step": 121870
    },
    {
      "epoch": 1.9942730917123455,
      "grad_norm": 0.05776599422097206,
      "learning_rate": 3.0502541793485542e-06,
      "loss": 0.0013,
      "step": 121880
    },
    {
      "epoch": 1.9944367176634215,
      "grad_norm": 0.09268534928560257,
      "learning_rate": 3.049377633627455e-06,
      "loss": 0.0016,
      "step": 121890
    },
    {
      "epoch": 1.9946003436144972,
      "grad_norm": 0.13744103908538818,
      "learning_rate": 3.048501158611731e-06,
      "loss": 0.0019,
      "step": 121900
    },
    {
      "epoch": 1.994763969565573,
      "grad_norm": 0.11338288336992264,
      "learning_rate": 3.0476247543331453e-06,
      "loss": 0.0017,
      "step": 121910
    },
    {
      "epoch": 1.994927595516649,
      "grad_norm": 0.054671287536621094,
      "learning_rate": 3.046748420823472e-06,
      "loss": 0.0011,
      "step": 121920
    },
    {
      "epoch": 1.9950912214677248,
      "grad_norm": 0.17806117236614227,
      "learning_rate": 3.0458721581144735e-06,
      "loss": 0.0023,
      "step": 121930
    },
    {
      "epoch": 1.9952548474188005,
      "grad_norm": 0.15047022700309753,
      "learning_rate": 3.044995966237913e-06,
      "loss": 0.001,
      "step": 121940
    },
    {
      "epoch": 1.9954184733698765,
      "grad_norm": 0.13910718262195587,
      "learning_rate": 3.0441198452255497e-06,
      "loss": 0.001,
      "step": 121950
    },
    {
      "epoch": 1.9955820993209523,
      "grad_norm": 0.23290413618087769,
      "learning_rate": 3.043243795109142e-06,
      "loss": 0.002,
      "step": 121960
    },
    {
      "epoch": 1.995745725272028,
      "grad_norm": 0.01668221317231655,
      "learning_rate": 3.042367815920443e-06,
      "loss": 0.0004,
      "step": 121970
    },
    {
      "epoch": 1.995909351223104,
      "grad_norm": 0.0994178056716919,
      "learning_rate": 3.041491907691205e-06,
      "loss": 0.0011,
      "step": 121980
    },
    {
      "epoch": 1.9960729771741799,
      "grad_norm": 0.07157222926616669,
      "learning_rate": 3.0406160704531794e-06,
      "loss": 0.0006,
      "step": 121990
    },
    {
      "epoch": 1.9962366031252556,
      "grad_norm": 0.31447213888168335,
      "learning_rate": 3.0397403042381124e-06,
      "loss": 0.0012,
      "step": 122000
    },
    {
      "epoch": 1.9964002290763316,
      "grad_norm": 0.12848053872585297,
      "learning_rate": 3.0388646090777475e-06,
      "loss": 0.0015,
      "step": 122010
    },
    {
      "epoch": 1.9965638550274072,
      "grad_norm": 0.09544669836759567,
      "learning_rate": 3.037988985003826e-06,
      "loss": 0.0013,
      "step": 122020
    },
    {
      "epoch": 1.9967274809784832,
      "grad_norm": 0.09430322051048279,
      "learning_rate": 3.0371134320480896e-06,
      "loss": 0.0016,
      "step": 122030
    },
    {
      "epoch": 1.9968911069295592,
      "grad_norm": 0.13307379186153412,
      "learning_rate": 3.0362379502422722e-06,
      "loss": 0.0015,
      "step": 122040
    },
    {
      "epoch": 1.9970547328806347,
      "grad_norm": 0.14317603409290314,
      "learning_rate": 3.035362539618111e-06,
      "loss": 0.0019,
      "step": 122050
    },
    {
      "epoch": 1.9972183588317107,
      "grad_norm": 0.006431240122765303,
      "learning_rate": 3.0344872002073343e-06,
      "loss": 0.0008,
      "step": 122060
    },
    {
      "epoch": 1.9973819847827865,
      "grad_norm": 0.1079738661646843,
      "learning_rate": 3.0336119320416725e-06,
      "loss": 0.0028,
      "step": 122070
    },
    {
      "epoch": 1.9975456107338623,
      "grad_norm": 0.059948522597551346,
      "learning_rate": 3.0327367351528523e-06,
      "loss": 0.0015,
      "step": 122080
    },
    {
      "epoch": 1.9977092366849383,
      "grad_norm": 0.09824782609939575,
      "learning_rate": 3.031861609572598e-06,
      "loss": 0.0014,
      "step": 122090
    },
    {
      "epoch": 1.997872862636014,
      "grad_norm": 0.050781819969415665,
      "learning_rate": 3.0309865553326323e-06,
      "loss": 0.0011,
      "step": 122100
    },
    {
      "epoch": 1.9980364885870898,
      "grad_norm": 0.1278054267168045,
      "learning_rate": 3.030111572464669e-06,
      "loss": 0.0008,
      "step": 122110
    },
    {
      "epoch": 1.9982001145381658,
      "grad_norm": 0.08743195235729218,
      "learning_rate": 3.0292366610004265e-06,
      "loss": 0.0012,
      "step": 122120
    },
    {
      "epoch": 1.9983637404892416,
      "grad_norm": 0.03167599439620972,
      "learning_rate": 3.0283618209716176e-06,
      "loss": 0.0013,
      "step": 122130
    },
    {
      "epoch": 1.9985273664403174,
      "grad_norm": 0.1520116925239563,
      "learning_rate": 3.0274870524099553e-06,
      "loss": 0.0014,
      "step": 122140
    },
    {
      "epoch": 1.9986909923913934,
      "grad_norm": 0.044764839112758636,
      "learning_rate": 3.026612355347146e-06,
      "loss": 0.0012,
      "step": 122150
    },
    {
      "epoch": 1.9988546183424691,
      "grad_norm": 0.10647802799940109,
      "learning_rate": 3.025737729814896e-06,
      "loss": 0.0017,
      "step": 122160
    },
    {
      "epoch": 1.999018244293545,
      "grad_norm": 0.022934338077902794,
      "learning_rate": 3.0248631758449086e-06,
      "loss": 0.0014,
      "step": 122170
    },
    {
      "epoch": 1.999181870244621,
      "grad_norm": 0.05098085105419159,
      "learning_rate": 3.0239886934688845e-06,
      "loss": 0.0026,
      "step": 122180
    },
    {
      "epoch": 1.9993454961956967,
      "grad_norm": 0.031086700037121773,
      "learning_rate": 3.02311428271852e-06,
      "loss": 0.0011,
      "step": 122190
    },
    {
      "epoch": 1.9995091221467725,
      "grad_norm": 0.1163347065448761,
      "learning_rate": 3.022239943625513e-06,
      "loss": 0.0011,
      "step": 122200
    },
    {
      "epoch": 1.9996727480978485,
      "grad_norm": 0.07484007626771927,
      "learning_rate": 3.0213656762215537e-06,
      "loss": 0.0013,
      "step": 122210
    },
    {
      "epoch": 1.999836374048924,
      "grad_norm": 0.05343592166900635,
      "learning_rate": 3.0204914805383324e-06,
      "loss": 0.0007,
      "step": 122220
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.007802479900419712,
      "learning_rate": 3.0196173566075383e-06,
      "loss": 0.0007,
      "step": 122230
    },
    {
      "epoch": 2.000163625951076,
      "grad_norm": 0.039337046444416046,
      "learning_rate": 3.0187433044608545e-06,
      "loss": 0.0008,
      "step": 122240
    },
    {
      "epoch": 2.0003272519021515,
      "grad_norm": 0.07899314165115356,
      "learning_rate": 3.017869324129964e-06,
      "loss": 0.0011,
      "step": 122250
    },
    {
      "epoch": 2.0004908778532275,
      "grad_norm": 0.12674546241760254,
      "learning_rate": 3.0169954156465475e-06,
      "loss": 0.0009,
      "step": 122260
    },
    {
      "epoch": 2.0006545038043035,
      "grad_norm": 0.032715678215026855,
      "learning_rate": 3.0161215790422803e-06,
      "loss": 0.0012,
      "step": 122270
    },
    {
      "epoch": 2.000818129755379,
      "grad_norm": 0.10841579735279083,
      "learning_rate": 3.015247814348838e-06,
      "loss": 0.0016,
      "step": 122280
    },
    {
      "epoch": 2.000981755706455,
      "grad_norm": 0.07675003260374069,
      "learning_rate": 3.0143741215978917e-06,
      "loss": 0.001,
      "step": 122290
    },
    {
      "epoch": 2.001145381657531,
      "grad_norm": 0.004280860535800457,
      "learning_rate": 3.013500500821114e-06,
      "loss": 0.0014,
      "step": 122300
    },
    {
      "epoch": 2.0013090076086066,
      "grad_norm": 0.06504587084054947,
      "learning_rate": 3.012626952050166e-06,
      "loss": 0.0011,
      "step": 122310
    },
    {
      "epoch": 2.0014726335596826,
      "grad_norm": 0.033733267337083817,
      "learning_rate": 3.0117534753167145e-06,
      "loss": 0.0009,
      "step": 122320
    },
    {
      "epoch": 2.001636259510758,
      "grad_norm": 0.03392750024795532,
      "learning_rate": 3.010880070652422e-06,
      "loss": 0.0031,
      "step": 122330
    },
    {
      "epoch": 2.001799885461834,
      "grad_norm": 0.06573561578989029,
      "learning_rate": 3.0100067380889453e-06,
      "loss": 0.0009,
      "step": 122340
    },
    {
      "epoch": 2.00196351141291,
      "grad_norm": 0.10168496519327164,
      "learning_rate": 3.009133477657941e-06,
      "loss": 0.0011,
      "step": 122350
    },
    {
      "epoch": 2.0021271373639857,
      "grad_norm": 0.0023911609314382076,
      "learning_rate": 3.0082602893910636e-06,
      "loss": 0.0008,
      "step": 122360
    },
    {
      "epoch": 2.0022907633150617,
      "grad_norm": 0.12138720601797104,
      "learning_rate": 3.007387173319964e-06,
      "loss": 0.0013,
      "step": 122370
    },
    {
      "epoch": 2.0024543892661377,
      "grad_norm": 0.09041742235422134,
      "learning_rate": 3.0065141294762903e-06,
      "loss": 0.0006,
      "step": 122380
    },
    {
      "epoch": 2.0026180152172133,
      "grad_norm": 0.12582673132419586,
      "learning_rate": 3.005641157891688e-06,
      "loss": 0.001,
      "step": 122390
    },
    {
      "epoch": 2.0027816411682893,
      "grad_norm": 0.03776866942644119,
      "learning_rate": 3.0047682585978023e-06,
      "loss": 0.0013,
      "step": 122400
    },
    {
      "epoch": 2.0029452671193653,
      "grad_norm": 0.1255517154932022,
      "learning_rate": 3.0038954316262693e-06,
      "loss": 0.0008,
      "step": 122410
    },
    {
      "epoch": 2.003108893070441,
      "grad_norm": 0.0556458905339241,
      "learning_rate": 3.0030226770087302e-06,
      "loss": 0.0005,
      "step": 122420
    },
    {
      "epoch": 2.003272519021517,
      "grad_norm": 0.045946069061756134,
      "learning_rate": 3.0021499947768197e-06,
      "loss": 0.0008,
      "step": 122430
    },
    {
      "epoch": 2.003436144972593,
      "grad_norm": 0.11650247871875763,
      "learning_rate": 3.0012773849621694e-06,
      "loss": 0.0007,
      "step": 122440
    },
    {
      "epoch": 2.0035997709236684,
      "grad_norm": 0.029856454581022263,
      "learning_rate": 3.0004048475964107e-06,
      "loss": 0.0011,
      "step": 122450
    },
    {
      "epoch": 2.0037633968747444,
      "grad_norm": 0.11715023964643478,
      "learning_rate": 2.99953238271117e-06,
      "loss": 0.0018,
      "step": 122460
    },
    {
      "epoch": 2.0039270228258204,
      "grad_norm": 0.0032879426144063473,
      "learning_rate": 2.998659990338073e-06,
      "loss": 0.001,
      "step": 122470
    },
    {
      "epoch": 2.004090648776896,
      "grad_norm": 0.026977650821208954,
      "learning_rate": 2.997787670508741e-06,
      "loss": 0.0016,
      "step": 122480
    },
    {
      "epoch": 2.004254274727972,
      "grad_norm": 0.029142964631319046,
      "learning_rate": 2.9969154232547936e-06,
      "loss": 0.0006,
      "step": 122490
    },
    {
      "epoch": 2.004417900679048,
      "grad_norm": 0.04164513945579529,
      "learning_rate": 2.9960432486078494e-06,
      "loss": 0.0008,
      "step": 122500
    },
    {
      "epoch": 2.0045815266301235,
      "grad_norm": 0.08730745315551758,
      "learning_rate": 2.995171146599519e-06,
      "loss": 0.002,
      "step": 122510
    },
    {
      "epoch": 2.0047451525811995,
      "grad_norm": 0.01673181727528572,
      "learning_rate": 2.9942991172614168e-06,
      "loss": 0.0006,
      "step": 122520
    },
    {
      "epoch": 2.004908778532275,
      "grad_norm": 0.016155635938048363,
      "learning_rate": 2.993427160625151e-06,
      "loss": 0.0016,
      "step": 122530
    },
    {
      "epoch": 2.005072404483351,
      "grad_norm": 0.011071273125708103,
      "learning_rate": 2.9925552767223276e-06,
      "loss": 0.0008,
      "step": 122540
    },
    {
      "epoch": 2.005236030434427,
      "grad_norm": 0.023952504619956017,
      "learning_rate": 2.991683465584551e-06,
      "loss": 0.0012,
      "step": 122550
    },
    {
      "epoch": 2.0053996563855025,
      "grad_norm": 0.10510613769292831,
      "learning_rate": 2.990811727243422e-06,
      "loss": 0.0006,
      "step": 122560
    },
    {
      "epoch": 2.0055632823365785,
      "grad_norm": 0.16675055027008057,
      "learning_rate": 2.9899400617305374e-06,
      "loss": 0.0007,
      "step": 122570
    },
    {
      "epoch": 2.0057269082876545,
      "grad_norm": 0.05483875051140785,
      "learning_rate": 2.989068469077496e-06,
      "loss": 0.001,
      "step": 122580
    },
    {
      "epoch": 2.00589053423873,
      "grad_norm": 0.05115516111254692,
      "learning_rate": 2.988196949315889e-06,
      "loss": 0.0024,
      "step": 122590
    },
    {
      "epoch": 2.006054160189806,
      "grad_norm": 0.0855933129787445,
      "learning_rate": 2.9873255024773096e-06,
      "loss": 0.0012,
      "step": 122600
    },
    {
      "epoch": 2.006217786140882,
      "grad_norm": 0.10746615380048752,
      "learning_rate": 2.9864541285933417e-06,
      "loss": 0.0013,
      "step": 122610
    },
    {
      "epoch": 2.0063814120919576,
      "grad_norm": 0.0017163899028673768,
      "learning_rate": 2.9855828276955724e-06,
      "loss": 0.0006,
      "step": 122620
    },
    {
      "epoch": 2.0065450380430336,
      "grad_norm": 0.10305550694465637,
      "learning_rate": 2.9847115998155824e-06,
      "loss": 0.0006,
      "step": 122630
    },
    {
      "epoch": 2.0067086639941096,
      "grad_norm": 0.04796452447772026,
      "learning_rate": 2.9838404449849544e-06,
      "loss": 0.0006,
      "step": 122640
    },
    {
      "epoch": 2.006872289945185,
      "grad_norm": 0.120184026658535,
      "learning_rate": 2.9829693632352647e-06,
      "loss": 0.0008,
      "step": 122650
    },
    {
      "epoch": 2.007035915896261,
      "grad_norm": 0.00197650701738894,
      "learning_rate": 2.9820983545980874e-06,
      "loss": 0.0007,
      "step": 122660
    },
    {
      "epoch": 2.007199541847337,
      "grad_norm": 0.18119893968105316,
      "learning_rate": 2.9812274191049952e-06,
      "loss": 0.0007,
      "step": 122670
    },
    {
      "epoch": 2.0073631677984127,
      "grad_norm": 0.3269008696079254,
      "learning_rate": 2.980356556787557e-06,
      "loss": 0.0017,
      "step": 122680
    },
    {
      "epoch": 2.0075267937494887,
      "grad_norm": 0.10589903593063354,
      "learning_rate": 2.97948576767734e-06,
      "loss": 0.0007,
      "step": 122690
    },
    {
      "epoch": 2.0076904197005647,
      "grad_norm": 0.0075447214767336845,
      "learning_rate": 2.978615051805908e-06,
      "loss": 0.0008,
      "step": 122700
    },
    {
      "epoch": 2.0078540456516403,
      "grad_norm": 0.16812700033187866,
      "learning_rate": 2.9777444092048214e-06,
      "loss": 0.0009,
      "step": 122710
    },
    {
      "epoch": 2.0080176716027163,
      "grad_norm": 0.4088684320449829,
      "learning_rate": 2.9768738399056395e-06,
      "loss": 0.0011,
      "step": 122720
    },
    {
      "epoch": 2.008181297553792,
      "grad_norm": 0.044109005481004715,
      "learning_rate": 2.976003343939919e-06,
      "loss": 0.0006,
      "step": 122730
    },
    {
      "epoch": 2.008344923504868,
      "grad_norm": 0.07573593407869339,
      "learning_rate": 2.975132921339212e-06,
      "loss": 0.0008,
      "step": 122740
    },
    {
      "epoch": 2.008508549455944,
      "grad_norm": 0.10288336127996445,
      "learning_rate": 2.9742625721350702e-06,
      "loss": 0.001,
      "step": 122750
    },
    {
      "epoch": 2.0086721754070194,
      "grad_norm": 0.05118601769208908,
      "learning_rate": 2.973392296359042e-06,
      "loss": 0.0013,
      "step": 122760
    },
    {
      "epoch": 2.0088358013580954,
      "grad_norm": 0.010381937958300114,
      "learning_rate": 2.972522094042671e-06,
      "loss": 0.0005,
      "step": 122770
    },
    {
      "epoch": 2.0089994273091714,
      "grad_norm": 0.05149473249912262,
      "learning_rate": 2.9716519652175024e-06,
      "loss": 0.0007,
      "step": 122780
    },
    {
      "epoch": 2.009163053260247,
      "grad_norm": 0.1264180690050125,
      "learning_rate": 2.9707819099150737e-06,
      "loss": 0.0024,
      "step": 122790
    },
    {
      "epoch": 2.009326679211323,
      "grad_norm": 0.03264115750789642,
      "learning_rate": 2.9699119281669263e-06,
      "loss": 0.0007,
      "step": 122800
    },
    {
      "epoch": 2.009490305162399,
      "grad_norm": 0.03596828505396843,
      "learning_rate": 2.9690420200045903e-06,
      "loss": 0.0018,
      "step": 122810
    },
    {
      "epoch": 2.0096539311134745,
      "grad_norm": 0.030557796359062195,
      "learning_rate": 2.9681721854596e-06,
      "loss": 0.0006,
      "step": 122820
    },
    {
      "epoch": 2.0098175570645505,
      "grad_norm": 0.028343254700303078,
      "learning_rate": 2.967302424563485e-06,
      "loss": 0.0007,
      "step": 122830
    },
    {
      "epoch": 2.0099811830156264,
      "grad_norm": 0.004581385292112827,
      "learning_rate": 2.9664327373477707e-06,
      "loss": 0.0008,
      "step": 122840
    },
    {
      "epoch": 2.010144808966702,
      "grad_norm": 0.049201253801584244,
      "learning_rate": 2.965563123843982e-06,
      "loss": 0.0005,
      "step": 122850
    },
    {
      "epoch": 2.010308434917778,
      "grad_norm": 0.11184683442115784,
      "learning_rate": 2.964693584083641e-06,
      "loss": 0.0006,
      "step": 122860
    },
    {
      "epoch": 2.010472060868854,
      "grad_norm": 0.11916273087263107,
      "learning_rate": 2.9638241180982663e-06,
      "loss": 0.0009,
      "step": 122870
    },
    {
      "epoch": 2.0106356868199295,
      "grad_norm": 0.045228488743305206,
      "learning_rate": 2.9629547259193736e-06,
      "loss": 0.0006,
      "step": 122880
    },
    {
      "epoch": 2.0107993127710055,
      "grad_norm": 0.11494050174951553,
      "learning_rate": 2.9620854075784753e-06,
      "loss": 0.0015,
      "step": 122890
    },
    {
      "epoch": 2.0109629387220815,
      "grad_norm": 0.01775909960269928,
      "learning_rate": 2.9612161631070857e-06,
      "loss": 0.0009,
      "step": 122900
    },
    {
      "epoch": 2.011126564673157,
      "grad_norm": 0.06603578478097916,
      "learning_rate": 2.960346992536707e-06,
      "loss": 0.001,
      "step": 122910
    },
    {
      "epoch": 2.011290190624233,
      "grad_norm": 0.1089678555727005,
      "learning_rate": 2.959477895898848e-06,
      "loss": 0.0016,
      "step": 122920
    },
    {
      "epoch": 2.0114538165753086,
      "grad_norm": 0.6310131549835205,
      "learning_rate": 2.9586088732250116e-06,
      "loss": 0.0175,
      "step": 122930
    },
    {
      "epoch": 2.0116174425263846,
      "grad_norm": 0.0440681166946888,
      "learning_rate": 2.957739924546697e-06,
      "loss": 0.0011,
      "step": 122940
    },
    {
      "epoch": 2.0117810684774606,
      "grad_norm": 0.017576299607753754,
      "learning_rate": 2.956871049895402e-06,
      "loss": 0.0031,
      "step": 122950
    },
    {
      "epoch": 2.011944694428536,
      "grad_norm": 0.0920783281326294,
      "learning_rate": 2.956002249302621e-06,
      "loss": 0.0005,
      "step": 122960
    },
    {
      "epoch": 2.012108320379612,
      "grad_norm": 0.2482539713382721,
      "learning_rate": 2.9551335227998453e-06,
      "loss": 0.0008,
      "step": 122970
    },
    {
      "epoch": 2.012271946330688,
      "grad_norm": 0.11471476405858994,
      "learning_rate": 2.9542648704185652e-06,
      "loss": 0.0009,
      "step": 122980
    },
    {
      "epoch": 2.0124355722817637,
      "grad_norm": 0.1749299168586731,
      "learning_rate": 2.953396292190266e-06,
      "loss": 0.0017,
      "step": 122990
    },
    {
      "epoch": 2.0125991982328397,
      "grad_norm": 0.06950230151414871,
      "learning_rate": 2.952527788146434e-06,
      "loss": 0.0006,
      "step": 123000
    },
    {
      "epoch": 2.0127628241839157,
      "grad_norm": 0.014681133441627026,
      "learning_rate": 2.9516593583185473e-06,
      "loss": 0.0013,
      "step": 123010
    },
    {
      "epoch": 2.0129264501349913,
      "grad_norm": 0.04077136144042015,
      "learning_rate": 2.950791002738086e-06,
      "loss": 0.0006,
      "step": 123020
    },
    {
      "epoch": 2.0130900760860673,
      "grad_norm": 0.04269719496369362,
      "learning_rate": 2.9499227214365257e-06,
      "loss": 0.0006,
      "step": 123030
    },
    {
      "epoch": 2.0132537020371433,
      "grad_norm": 0.08492803573608398,
      "learning_rate": 2.9490545144453397e-06,
      "loss": 0.0006,
      "step": 123040
    },
    {
      "epoch": 2.013417327988219,
      "grad_norm": 0.05451168119907379,
      "learning_rate": 2.9481863817959976e-06,
      "loss": 0.0011,
      "step": 123050
    },
    {
      "epoch": 2.013580953939295,
      "grad_norm": 0.039799947291612625,
      "learning_rate": 2.9473183235199688e-06,
      "loss": 0.0013,
      "step": 123060
    },
    {
      "epoch": 2.013744579890371,
      "grad_norm": 0.08183470368385315,
      "learning_rate": 2.9464503396487155e-06,
      "loss": 0.0007,
      "step": 123070
    },
    {
      "epoch": 2.0139082058414464,
      "grad_norm": 0.10906807333230972,
      "learning_rate": 2.9455824302137024e-06,
      "loss": 0.0008,
      "step": 123080
    },
    {
      "epoch": 2.0140718317925224,
      "grad_norm": 0.0852169394493103,
      "learning_rate": 2.9447145952463897e-06,
      "loss": 0.0014,
      "step": 123090
    },
    {
      "epoch": 2.014235457743598,
      "grad_norm": 0.07407552003860474,
      "learning_rate": 2.9438468347782343e-06,
      "loss": 0.0008,
      "step": 123100
    },
    {
      "epoch": 2.014399083694674,
      "grad_norm": 0.0034440841991454363,
      "learning_rate": 2.9429791488406876e-06,
      "loss": 0.0012,
      "step": 123110
    },
    {
      "epoch": 2.01456270964575,
      "grad_norm": 0.02655049040913582,
      "learning_rate": 2.942111537465202e-06,
      "loss": 0.0008,
      "step": 123120
    },
    {
      "epoch": 2.0147263355968255,
      "grad_norm": 0.0492183193564415,
      "learning_rate": 2.9412440006832276e-06,
      "loss": 0.0006,
      "step": 123130
    },
    {
      "epoch": 2.0148899615479015,
      "grad_norm": 0.1544003188610077,
      "learning_rate": 2.940376538526211e-06,
      "loss": 0.0003,
      "step": 123140
    },
    {
      "epoch": 2.0150535874989774,
      "grad_norm": 0.04998074844479561,
      "learning_rate": 2.939509151025594e-06,
      "loss": 0.0006,
      "step": 123150
    },
    {
      "epoch": 2.015217213450053,
      "grad_norm": 0.0605974905192852,
      "learning_rate": 2.9386418382128184e-06,
      "loss": 0.0007,
      "step": 123160
    },
    {
      "epoch": 2.015380839401129,
      "grad_norm": 0.05646251514554024,
      "learning_rate": 2.9377746001193218e-06,
      "loss": 0.001,
      "step": 123170
    },
    {
      "epoch": 2.015544465352205,
      "grad_norm": 0.1739867627620697,
      "learning_rate": 2.936907436776539e-06,
      "loss": 0.0016,
      "step": 123180
    },
    {
      "epoch": 2.0157080913032805,
      "grad_norm": 0.17260171473026276,
      "learning_rate": 2.936040348215904e-06,
      "loss": 0.0023,
      "step": 123190
    },
    {
      "epoch": 2.0158717172543565,
      "grad_norm": 0.022806884720921516,
      "learning_rate": 2.935173334468846e-06,
      "loss": 0.0006,
      "step": 123200
    },
    {
      "epoch": 2.0160353432054325,
      "grad_norm": 0.0752556324005127,
      "learning_rate": 2.9343063955667913e-06,
      "loss": 0.0013,
      "step": 123210
    },
    {
      "epoch": 2.016198969156508,
      "grad_norm": 0.2098541110754013,
      "learning_rate": 2.9334395315411644e-06,
      "loss": 0.0008,
      "step": 123220
    },
    {
      "epoch": 2.016362595107584,
      "grad_norm": 0.05458023026585579,
      "learning_rate": 2.9325727424233887e-06,
      "loss": 0.0011,
      "step": 123230
    },
    {
      "epoch": 2.01652622105866,
      "grad_norm": 0.02899891696870327,
      "learning_rate": 2.9317060282448815e-06,
      "loss": 0.0006,
      "step": 123240
    },
    {
      "epoch": 2.0166898470097356,
      "grad_norm": 0.044367577880620956,
      "learning_rate": 2.9308393890370602e-06,
      "loss": 0.0009,
      "step": 123250
    },
    {
      "epoch": 2.0168534729608116,
      "grad_norm": 0.05011260136961937,
      "learning_rate": 2.9299728248313374e-06,
      "loss": 0.0005,
      "step": 123260
    },
    {
      "epoch": 2.0170170989118876,
      "grad_norm": 0.0861293151974678,
      "learning_rate": 2.9291063356591245e-06,
      "loss": 0.0009,
      "step": 123270
    },
    {
      "epoch": 2.017180724862963,
      "grad_norm": 0.07861240208148956,
      "learning_rate": 2.928239921551829e-06,
      "loss": 0.0013,
      "step": 123280
    },
    {
      "epoch": 2.017344350814039,
      "grad_norm": 0.12659820914268494,
      "learning_rate": 2.927373582540858e-06,
      "loss": 0.0013,
      "step": 123290
    },
    {
      "epoch": 2.0175079767651147,
      "grad_norm": 0.11719324439764023,
      "learning_rate": 2.926507318657613e-06,
      "loss": 0.0024,
      "step": 123300
    },
    {
      "epoch": 2.0176716027161907,
      "grad_norm": 0.013598169200122356,
      "learning_rate": 2.9256411299334964e-06,
      "loss": 0.0006,
      "step": 123310
    },
    {
      "epoch": 2.0178352286672667,
      "grad_norm": 0.00677691213786602,
      "learning_rate": 2.924775016399901e-06,
      "loss": 0.0009,
      "step": 123320
    },
    {
      "epoch": 2.0179988546183423,
      "grad_norm": 0.07832401990890503,
      "learning_rate": 2.9239089780882234e-06,
      "loss": 0.0008,
      "step": 123330
    },
    {
      "epoch": 2.0181624805694183,
      "grad_norm": 0.11921148002147675,
      "learning_rate": 2.923043015029855e-06,
      "loss": 0.0009,
      "step": 123340
    },
    {
      "epoch": 2.0183261065204943,
      "grad_norm": 0.10726704448461533,
      "learning_rate": 2.9221771272561863e-06,
      "loss": 0.0013,
      "step": 123350
    },
    {
      "epoch": 2.01848973247157,
      "grad_norm": 0.32264629006385803,
      "learning_rate": 2.9213113147986026e-06,
      "loss": 0.0012,
      "step": 123360
    },
    {
      "epoch": 2.018653358422646,
      "grad_norm": 0.3090819716453552,
      "learning_rate": 2.9204455776884888e-06,
      "loss": 0.0007,
      "step": 123370
    },
    {
      "epoch": 2.018816984373722,
      "grad_norm": 0.010182442143559456,
      "learning_rate": 2.9195799159572235e-06,
      "loss": 0.0007,
      "step": 123380
    },
    {
      "epoch": 2.0189806103247974,
      "grad_norm": 0.09297315031290054,
      "learning_rate": 2.9187143296361864e-06,
      "loss": 0.0009,
      "step": 123390
    },
    {
      "epoch": 2.0191442362758734,
      "grad_norm": 0.12436174601316452,
      "learning_rate": 2.917848818756752e-06,
      "loss": 0.0016,
      "step": 123400
    },
    {
      "epoch": 2.0193078622269494,
      "grad_norm": 0.04419776052236557,
      "learning_rate": 2.9169833833502974e-06,
      "loss": 0.001,
      "step": 123410
    },
    {
      "epoch": 2.019471488178025,
      "grad_norm": 0.11890144646167755,
      "learning_rate": 2.9161180234481844e-06,
      "loss": 0.0018,
      "step": 123420
    },
    {
      "epoch": 2.019635114129101,
      "grad_norm": 0.19173476099967957,
      "learning_rate": 2.9152527390817874e-06,
      "loss": 0.0011,
      "step": 123430
    },
    {
      "epoch": 2.019798740080177,
      "grad_norm": 0.03202402964234352,
      "learning_rate": 2.9143875302824655e-06,
      "loss": 0.0008,
      "step": 123440
    },
    {
      "epoch": 2.0199623660312525,
      "grad_norm": 0.45920664072036743,
      "learning_rate": 2.913522397081583e-06,
      "loss": 0.001,
      "step": 123450
    },
    {
      "epoch": 2.0201259919823285,
      "grad_norm": 0.09136476367712021,
      "learning_rate": 2.912657339510501e-06,
      "loss": 0.0009,
      "step": 123460
    },
    {
      "epoch": 2.0202896179334044,
      "grad_norm": 0.2168736308813095,
      "learning_rate": 2.9117923576005713e-06,
      "loss": 0.0017,
      "step": 123470
    },
    {
      "epoch": 2.02045324388448,
      "grad_norm": 0.023497063666582108,
      "learning_rate": 2.910927451383152e-06,
      "loss": 0.0011,
      "step": 123480
    },
    {
      "epoch": 2.020616869835556,
      "grad_norm": 0.05596434324979782,
      "learning_rate": 2.91006262088959e-06,
      "loss": 0.0012,
      "step": 123490
    },
    {
      "epoch": 2.0207804957866315,
      "grad_norm": 0.10114511847496033,
      "learning_rate": 2.9091978661512366e-06,
      "loss": 0.0015,
      "step": 123500
    },
    {
      "epoch": 2.0209441217377075,
      "grad_norm": 0.03261394053697586,
      "learning_rate": 2.908333187199436e-06,
      "loss": 0.0011,
      "step": 123510
    },
    {
      "epoch": 2.0211077476887835,
      "grad_norm": 0.0462544783949852,
      "learning_rate": 2.907468584065527e-06,
      "loss": 0.0014,
      "step": 123520
    },
    {
      "epoch": 2.021271373639859,
      "grad_norm": 0.5521852970123291,
      "learning_rate": 2.9066040567808553e-06,
      "loss": 0.0005,
      "step": 123530
    },
    {
      "epoch": 2.021434999590935,
      "grad_norm": 0.06335219740867615,
      "learning_rate": 2.905739605376753e-06,
      "loss": 0.0013,
      "step": 123540
    },
    {
      "epoch": 2.021598625542011,
      "grad_norm": 0.043145306408405304,
      "learning_rate": 2.904875229884559e-06,
      "loss": 0.0006,
      "step": 123550
    },
    {
      "epoch": 2.0217622514930866,
      "grad_norm": 0.07603517919778824,
      "learning_rate": 2.9040109303356e-06,
      "loss": 0.0008,
      "step": 123560
    },
    {
      "epoch": 2.0219258774441626,
      "grad_norm": 0.06110682338476181,
      "learning_rate": 2.9031467067612095e-06,
      "loss": 0.0007,
      "step": 123570
    },
    {
      "epoch": 2.0220895033952386,
      "grad_norm": 0.04667156934738159,
      "learning_rate": 2.9022825591927086e-06,
      "loss": 0.0008,
      "step": 123580
    },
    {
      "epoch": 2.022253129346314,
      "grad_norm": 0.06395115703344345,
      "learning_rate": 2.901418487661426e-06,
      "loss": 0.0009,
      "step": 123590
    },
    {
      "epoch": 2.02241675529739,
      "grad_norm": 0.050273597240448,
      "learning_rate": 2.9005544921986774e-06,
      "loss": 0.0009,
      "step": 123600
    },
    {
      "epoch": 2.022580381248466,
      "grad_norm": 0.06157859414815903,
      "learning_rate": 2.899690572835784e-06,
      "loss": 0.0004,
      "step": 123610
    },
    {
      "epoch": 2.0227440071995417,
      "grad_norm": 0.11917011439800262,
      "learning_rate": 2.8988267296040594e-06,
      "loss": 0.0013,
      "step": 123620
    },
    {
      "epoch": 2.0229076331506177,
      "grad_norm": 0.011408847756683826,
      "learning_rate": 2.8979629625348138e-06,
      "loss": 0.0005,
      "step": 123630
    },
    {
      "epoch": 2.0230712591016937,
      "grad_norm": 0.22224397957324982,
      "learning_rate": 2.897099271659361e-06,
      "loss": 0.0012,
      "step": 123640
    },
    {
      "epoch": 2.0232348850527693,
      "grad_norm": 0.0042869048193097115,
      "learning_rate": 2.8962356570090035e-06,
      "loss": 0.0011,
      "step": 123650
    },
    {
      "epoch": 2.0233985110038453,
      "grad_norm": 0.059125710278749466,
      "learning_rate": 2.8953721186150485e-06,
      "loss": 0.0011,
      "step": 123660
    },
    {
      "epoch": 2.0235621369549213,
      "grad_norm": 0.01862114667892456,
      "learning_rate": 2.894508656508794e-06,
      "loss": 0.0008,
      "step": 123670
    },
    {
      "epoch": 2.023725762905997,
      "grad_norm": 0.08622175455093384,
      "learning_rate": 2.893645270721541e-06,
      "loss": 0.0018,
      "step": 123680
    },
    {
      "epoch": 2.023889388857073,
      "grad_norm": 0.01131894439458847,
      "learning_rate": 2.8927819612845863e-06,
      "loss": 0.001,
      "step": 123690
    },
    {
      "epoch": 2.0240530148081484,
      "grad_norm": 0.10598060488700867,
      "learning_rate": 2.891918728229219e-06,
      "loss": 0.0007,
      "step": 123700
    },
    {
      "epoch": 2.0242166407592244,
      "grad_norm": 0.08070161193609238,
      "learning_rate": 2.891055571586735e-06,
      "loss": 0.0006,
      "step": 123710
    },
    {
      "epoch": 2.0243802667103004,
      "grad_norm": 0.05542897433042526,
      "learning_rate": 2.890192491388413e-06,
      "loss": 0.0012,
      "step": 123720
    },
    {
      "epoch": 2.024543892661376,
      "grad_norm": 0.0634867325425148,
      "learning_rate": 2.889329487665543e-06,
      "loss": 0.0007,
      "step": 123730
    },
    {
      "epoch": 2.024707518612452,
      "grad_norm": 0.02125907689332962,
      "learning_rate": 2.8884665604494088e-06,
      "loss": 0.0012,
      "step": 123740
    },
    {
      "epoch": 2.024871144563528,
      "grad_norm": 0.07873150706291199,
      "learning_rate": 2.887603709771284e-06,
      "loss": 0.0007,
      "step": 123750
    },
    {
      "epoch": 2.0250347705146035,
      "grad_norm": 0.133035808801651,
      "learning_rate": 2.8867409356624503e-06,
      "loss": 0.0004,
      "step": 123760
    },
    {
      "epoch": 2.0251983964656795,
      "grad_norm": 0.05324627831578255,
      "learning_rate": 2.8858782381541763e-06,
      "loss": 0.0006,
      "step": 123770
    },
    {
      "epoch": 2.0253620224167554,
      "grad_norm": 0.10550583153963089,
      "learning_rate": 2.885015617277737e-06,
      "loss": 0.0009,
      "step": 123780
    },
    {
      "epoch": 2.025525648367831,
      "grad_norm": 0.10863450914621353,
      "learning_rate": 2.8841530730643963e-06,
      "loss": 0.0008,
      "step": 123790
    },
    {
      "epoch": 2.025689274318907,
      "grad_norm": 0.10028284043073654,
      "learning_rate": 2.8832906055454237e-06,
      "loss": 0.0012,
      "step": 123800
    },
    {
      "epoch": 2.025852900269983,
      "grad_norm": 0.17284740507602692,
      "learning_rate": 2.8824282147520786e-06,
      "loss": 0.0007,
      "step": 123810
    },
    {
      "epoch": 2.0260165262210585,
      "grad_norm": 0.06500721722841263,
      "learning_rate": 2.8815659007156205e-06,
      "loss": 0.001,
      "step": 123820
    },
    {
      "epoch": 2.0261801521721345,
      "grad_norm": 0.05808574706315994,
      "learning_rate": 2.880703663467308e-06,
      "loss": 0.0004,
      "step": 123830
    },
    {
      "epoch": 2.0263437781232105,
      "grad_norm": 0.09927265346050262,
      "learning_rate": 2.879841503038392e-06,
      "loss": 0.0006,
      "step": 123840
    },
    {
      "epoch": 2.026507404074286,
      "grad_norm": 0.024597322568297386,
      "learning_rate": 2.8789794194601285e-06,
      "loss": 0.0019,
      "step": 123850
    },
    {
      "epoch": 2.026671030025362,
      "grad_norm": 0.058953169733285904,
      "learning_rate": 2.8781174127637612e-06,
      "loss": 0.0008,
      "step": 123860
    },
    {
      "epoch": 2.026834655976438,
      "grad_norm": 0.09125494956970215,
      "learning_rate": 2.8772554829805394e-06,
      "loss": 0.0011,
      "step": 123870
    },
    {
      "epoch": 2.0269982819275136,
      "grad_norm": 0.06706090271472931,
      "learning_rate": 2.8763936301417028e-06,
      "loss": 0.0009,
      "step": 123880
    },
    {
      "epoch": 2.0271619078785896,
      "grad_norm": 0.11271265149116516,
      "learning_rate": 2.8755318542784925e-06,
      "loss": 0.001,
      "step": 123890
    },
    {
      "epoch": 2.027325533829665,
      "grad_norm": 0.05396962910890579,
      "learning_rate": 2.8746701554221494e-06,
      "loss": 0.0007,
      "step": 123900
    },
    {
      "epoch": 2.027489159780741,
      "grad_norm": 0.07635393738746643,
      "learning_rate": 2.8738085336039047e-06,
      "loss": 0.0012,
      "step": 123910
    },
    {
      "epoch": 2.027652785731817,
      "grad_norm": 0.1010383889079094,
      "learning_rate": 2.8729469888549906e-06,
      "loss": 0.0006,
      "step": 123920
    },
    {
      "epoch": 2.0278164116828927,
      "grad_norm": 0.04615742713212967,
      "learning_rate": 2.8720855212066336e-06,
      "loss": 0.0009,
      "step": 123930
    },
    {
      "epoch": 2.0279800376339687,
      "grad_norm": 0.07685250788927078,
      "learning_rate": 2.8712241306900644e-06,
      "loss": 0.001,
      "step": 123940
    },
    {
      "epoch": 2.0281436635850447,
      "grad_norm": 0.056674107909202576,
      "learning_rate": 2.8703628173365022e-06,
      "loss": 0.0015,
      "step": 123950
    },
    {
      "epoch": 2.0283072895361203,
      "grad_norm": 0.043701887130737305,
      "learning_rate": 2.869501581177169e-06,
      "loss": 0.0008,
      "step": 123960
    },
    {
      "epoch": 2.0284709154871963,
      "grad_norm": 0.06434541195631027,
      "learning_rate": 2.868640422243285e-06,
      "loss": 0.0017,
      "step": 123970
    },
    {
      "epoch": 2.0286345414382723,
      "grad_norm": 0.15282042324543,
      "learning_rate": 2.8677793405660614e-06,
      "loss": 0.0006,
      "step": 123980
    },
    {
      "epoch": 2.028798167389348,
      "grad_norm": 0.08339986950159073,
      "learning_rate": 2.866918336176714e-06,
      "loss": 0.0016,
      "step": 123990
    },
    {
      "epoch": 2.028961793340424,
      "grad_norm": 0.009265325032174587,
      "learning_rate": 2.8660574091064487e-06,
      "loss": 0.0007,
      "step": 124000
    },
    {
      "epoch": 2.0291254192915,
      "grad_norm": 0.006840995047241449,
      "learning_rate": 2.865196559386475e-06,
      "loss": 0.0009,
      "step": 124010
    },
    {
      "epoch": 2.0292890452425754,
      "grad_norm": 0.00866103358566761,
      "learning_rate": 2.8643357870479947e-06,
      "loss": 0.0004,
      "step": 124020
    },
    {
      "epoch": 2.0294526711936514,
      "grad_norm": 0.08059342205524445,
      "learning_rate": 2.863475092122208e-06,
      "loss": 0.0014,
      "step": 124030
    },
    {
      "epoch": 2.0296162971447274,
      "grad_norm": 0.05488971620798111,
      "learning_rate": 2.8626144746403155e-06,
      "loss": 0.0019,
      "step": 124040
    },
    {
      "epoch": 2.029779923095803,
      "grad_norm": 0.13141994178295135,
      "learning_rate": 2.8617539346335097e-06,
      "loss": 0.0011,
      "step": 124050
    },
    {
      "epoch": 2.029943549046879,
      "grad_norm": 0.09598705172538757,
      "learning_rate": 2.8608934721329863e-06,
      "loss": 0.0006,
      "step": 124060
    },
    {
      "epoch": 2.0301071749979545,
      "grad_norm": 0.0666499063372612,
      "learning_rate": 2.8600330871699317e-06,
      "loss": 0.0008,
      "step": 124070
    },
    {
      "epoch": 2.0302708009490305,
      "grad_norm": 0.12923429906368256,
      "learning_rate": 2.8591727797755363e-06,
      "loss": 0.001,
      "step": 124080
    },
    {
      "epoch": 2.0304344269001064,
      "grad_norm": 0.08453816920518875,
      "learning_rate": 2.8583125499809805e-06,
      "loss": 0.0008,
      "step": 124090
    },
    {
      "epoch": 2.030598052851182,
      "grad_norm": 0.051464490592479706,
      "learning_rate": 2.8574523978174484e-06,
      "loss": 0.0007,
      "step": 124100
    },
    {
      "epoch": 2.030761678802258,
      "grad_norm": 0.13468940556049347,
      "learning_rate": 2.8565923233161187e-06,
      "loss": 0.0006,
      "step": 124110
    },
    {
      "epoch": 2.030925304753334,
      "grad_norm": 0.060485709458589554,
      "learning_rate": 2.855732326508163e-06,
      "loss": 0.0009,
      "step": 124120
    },
    {
      "epoch": 2.0310889307044095,
      "grad_norm": 0.0032297370489686728,
      "learning_rate": 2.854872407424759e-06,
      "loss": 0.0011,
      "step": 124130
    },
    {
      "epoch": 2.0312525566554855,
      "grad_norm": 0.06342637538909912,
      "learning_rate": 2.854012566097072e-06,
      "loss": 0.0008,
      "step": 124140
    },
    {
      "epoch": 2.0314161826065615,
      "grad_norm": 0.06412553042173386,
      "learning_rate": 2.8531528025562742e-06,
      "loss": 0.0011,
      "step": 124150
    },
    {
      "epoch": 2.031579808557637,
      "grad_norm": 0.12065336853265762,
      "learning_rate": 2.8522931168335255e-06,
      "loss": 0.0008,
      "step": 124160
    },
    {
      "epoch": 2.031743434508713,
      "grad_norm": 0.09229922294616699,
      "learning_rate": 2.85143350895999e-06,
      "loss": 0.0006,
      "step": 124170
    },
    {
      "epoch": 2.031907060459789,
      "grad_norm": 0.5208779573440552,
      "learning_rate": 2.8505739789668272e-06,
      "loss": 0.0011,
      "step": 124180
    },
    {
      "epoch": 2.0320706864108646,
      "grad_norm": 0.02819540910422802,
      "learning_rate": 2.84971452688519e-06,
      "loss": 0.0011,
      "step": 124190
    },
    {
      "epoch": 2.0322343123619406,
      "grad_norm": 0.056050196290016174,
      "learning_rate": 2.848855152746235e-06,
      "loss": 0.0007,
      "step": 124200
    },
    {
      "epoch": 2.0323979383130166,
      "grad_norm": 0.050121258944272995,
      "learning_rate": 2.8479958565811106e-06,
      "loss": 0.0008,
      "step": 124210
    },
    {
      "epoch": 2.032561564264092,
      "grad_norm": 0.04265778511762619,
      "learning_rate": 2.847136638420962e-06,
      "loss": 0.0008,
      "step": 124220
    },
    {
      "epoch": 2.032725190215168,
      "grad_norm": 0.011792679317295551,
      "learning_rate": 2.8462774982969376e-06,
      "loss": 0.0008,
      "step": 124230
    },
    {
      "epoch": 2.032888816166244,
      "grad_norm": 0.4496041238307953,
      "learning_rate": 2.845418436240176e-06,
      "loss": 0.0011,
      "step": 124240
    },
    {
      "epoch": 2.0330524421173197,
      "grad_norm": 0.2413029968738556,
      "learning_rate": 2.84455945228182e-06,
      "loss": 0.001,
      "step": 124250
    },
    {
      "epoch": 2.0332160680683957,
      "grad_norm": 0.09965922683477402,
      "learning_rate": 2.843700546453001e-06,
      "loss": 0.0007,
      "step": 124260
    },
    {
      "epoch": 2.0333796940194713,
      "grad_norm": 0.04634123295545578,
      "learning_rate": 2.8428417187848565e-06,
      "loss": 0.0007,
      "step": 124270
    },
    {
      "epoch": 2.0335433199705473,
      "grad_norm": 0.061617203056812286,
      "learning_rate": 2.841982969308513e-06,
      "loss": 0.0013,
      "step": 124280
    },
    {
      "epoch": 2.0337069459216233,
      "grad_norm": 0.038158830255270004,
      "learning_rate": 2.8411242980551034e-06,
      "loss": 0.0019,
      "step": 124290
    },
    {
      "epoch": 2.033870571872699,
      "grad_norm": 0.038674451410770416,
      "learning_rate": 2.840265705055746e-06,
      "loss": 0.0011,
      "step": 124300
    },
    {
      "epoch": 2.034034197823775,
      "grad_norm": 0.00625960249453783,
      "learning_rate": 2.839407190341569e-06,
      "loss": 0.001,
      "step": 124310
    },
    {
      "epoch": 2.034197823774851,
      "grad_norm": 0.04486432299017906,
      "learning_rate": 2.838548753943687e-06,
      "loss": 0.001,
      "step": 124320
    },
    {
      "epoch": 2.0343614497259264,
      "grad_norm": 0.02593972347676754,
      "learning_rate": 2.8376903958932174e-06,
      "loss": 0.0009,
      "step": 124330
    },
    {
      "epoch": 2.0345250756770024,
      "grad_norm": 0.019660795107483864,
      "learning_rate": 2.836832116221275e-06,
      "loss": 0.0008,
      "step": 124340
    },
    {
      "epoch": 2.0346887016280784,
      "grad_norm": 0.29560962319374084,
      "learning_rate": 2.8359739149589676e-06,
      "loss": 0.0012,
      "step": 124350
    },
    {
      "epoch": 2.034852327579154,
      "grad_norm": 0.07079388946294785,
      "learning_rate": 2.835115792137407e-06,
      "loss": 0.0011,
      "step": 124360
    },
    {
      "epoch": 2.03501595353023,
      "grad_norm": 0.039573509246110916,
      "learning_rate": 2.834257747787693e-06,
      "loss": 0.0009,
      "step": 124370
    },
    {
      "epoch": 2.035179579481306,
      "grad_norm": 0.022018538787961006,
      "learning_rate": 2.833399781940931e-06,
      "loss": 0.0007,
      "step": 124380
    },
    {
      "epoch": 2.0353432054323815,
      "grad_norm": 0.03453710675239563,
      "learning_rate": 2.832541894628222e-06,
      "loss": 0.001,
      "step": 124390
    },
    {
      "epoch": 2.0355068313834574,
      "grad_norm": 0.032292742282152176,
      "learning_rate": 2.8316840858806573e-06,
      "loss": 0.0008,
      "step": 124400
    },
    {
      "epoch": 2.0356704573345334,
      "grad_norm": 0.07762788981199265,
      "learning_rate": 2.830826355729337e-06,
      "loss": 0.0004,
      "step": 124410
    },
    {
      "epoch": 2.035834083285609,
      "grad_norm": 0.0215966384857893,
      "learning_rate": 2.829968704205344e-06,
      "loss": 0.002,
      "step": 124420
    },
    {
      "epoch": 2.035997709236685,
      "grad_norm": 0.06726696342229843,
      "learning_rate": 2.8291111313397724e-06,
      "loss": 0.001,
      "step": 124430
    },
    {
      "epoch": 2.036161335187761,
      "grad_norm": 0.02724245935678482,
      "learning_rate": 2.8282536371637026e-06,
      "loss": 0.0012,
      "step": 124440
    },
    {
      "epoch": 2.0363249611388365,
      "grad_norm": 0.1215810626745224,
      "learning_rate": 2.8273962217082186e-06,
      "loss": 0.001,
      "step": 124450
    },
    {
      "epoch": 2.0364885870899125,
      "grad_norm": 0.00971477571874857,
      "learning_rate": 2.826538885004401e-06,
      "loss": 0.0004,
      "step": 124460
    },
    {
      "epoch": 2.036652213040988,
      "grad_norm": 0.09301253408193588,
      "learning_rate": 2.825681627083324e-06,
      "loss": 0.0013,
      "step": 124470
    },
    {
      "epoch": 2.036815838992064,
      "grad_norm": 0.03361900895833969,
      "learning_rate": 2.8248244479760643e-06,
      "loss": 0.001,
      "step": 124480
    },
    {
      "epoch": 2.03697946494314,
      "grad_norm": 0.024573149159550667,
      "learning_rate": 2.823967347713688e-06,
      "loss": 0.0012,
      "step": 124490
    },
    {
      "epoch": 2.0371430908942156,
      "grad_norm": 0.005056385882198811,
      "learning_rate": 2.8231103263272674e-06,
      "loss": 0.001,
      "step": 124500
    },
    {
      "epoch": 2.0373067168452916,
      "grad_norm": 0.08200830221176147,
      "learning_rate": 2.822253383847866e-06,
      "loss": 0.0005,
      "step": 124510
    },
    {
      "epoch": 2.0374703427963676,
      "grad_norm": 0.0070111858658492565,
      "learning_rate": 2.8213965203065428e-06,
      "loss": 0.001,
      "step": 124520
    },
    {
      "epoch": 2.037633968747443,
      "grad_norm": 0.06848406046628952,
      "learning_rate": 2.8205397357343617e-06,
      "loss": 0.0011,
      "step": 124530
    },
    {
      "epoch": 2.037797594698519,
      "grad_norm": 0.08236326277256012,
      "learning_rate": 2.8196830301623753e-06,
      "loss": 0.0008,
      "step": 124540
    },
    {
      "epoch": 2.037961220649595,
      "grad_norm": 0.06222716346383095,
      "learning_rate": 2.8188264036216406e-06,
      "loss": 0.0008,
      "step": 124550
    },
    {
      "epoch": 2.0381248466006707,
      "grad_norm": 0.032447900623083115,
      "learning_rate": 2.8179698561432043e-06,
      "loss": 0.0009,
      "step": 124560
    },
    {
      "epoch": 2.0382884725517467,
      "grad_norm": 0.08052060008049011,
      "learning_rate": 2.817113387758118e-06,
      "loss": 0.0006,
      "step": 124570
    },
    {
      "epoch": 2.0384520985028227,
      "grad_norm": 0.030338529497385025,
      "learning_rate": 2.8162569984974236e-06,
      "loss": 0.0008,
      "step": 124580
    },
    {
      "epoch": 2.0386157244538983,
      "grad_norm": 0.03316605091094971,
      "learning_rate": 2.8154006883921662e-06,
      "loss": 0.0007,
      "step": 124590
    },
    {
      "epoch": 2.0387793504049743,
      "grad_norm": 0.18145349621772766,
      "learning_rate": 2.8145444574733814e-06,
      "loss": 0.0024,
      "step": 124600
    },
    {
      "epoch": 2.0389429763560503,
      "grad_norm": 0.053525932133197784,
      "learning_rate": 2.81368830577211e-06,
      "loss": 0.0008,
      "step": 124610
    },
    {
      "epoch": 2.039106602307126,
      "grad_norm": 0.005097508430480957,
      "learning_rate": 2.8128322333193817e-06,
      "loss": 0.0006,
      "step": 124620
    },
    {
      "epoch": 2.039270228258202,
      "grad_norm": 0.038973547518253326,
      "learning_rate": 2.811976240146227e-06,
      "loss": 0.0006,
      "step": 124630
    },
    {
      "epoch": 2.039433854209278,
      "grad_norm": 0.015489310026168823,
      "learning_rate": 2.8111203262836767e-06,
      "loss": 0.0006,
      "step": 124640
    },
    {
      "epoch": 2.0395974801603534,
      "grad_norm": 0.09316594898700714,
      "learning_rate": 2.810264491762752e-06,
      "loss": 0.0018,
      "step": 124650
    },
    {
      "epoch": 2.0397611061114294,
      "grad_norm": 0.08134543150663376,
      "learning_rate": 2.8094087366144763e-06,
      "loss": 0.0008,
      "step": 124660
    },
    {
      "epoch": 2.039924732062505,
      "grad_norm": 0.06554632633924484,
      "learning_rate": 2.8085530608698707e-06,
      "loss": 0.0008,
      "step": 124670
    },
    {
      "epoch": 2.040088358013581,
      "grad_norm": 0.21216492354869843,
      "learning_rate": 2.8076974645599476e-06,
      "loss": 0.0011,
      "step": 124680
    },
    {
      "epoch": 2.040251983964657,
      "grad_norm": 0.1355913281440735,
      "learning_rate": 2.806841947715725e-06,
      "loss": 0.0005,
      "step": 124690
    },
    {
      "epoch": 2.0404156099157325,
      "grad_norm": 0.23601964116096497,
      "learning_rate": 2.805986510368208e-06,
      "loss": 0.0011,
      "step": 124700
    },
    {
      "epoch": 2.0405792358668084,
      "grad_norm": 0.005074649583548307,
      "learning_rate": 2.8051311525484115e-06,
      "loss": 0.0004,
      "step": 124710
    },
    {
      "epoch": 2.0407428618178844,
      "grad_norm": 0.21899673342704773,
      "learning_rate": 2.804275874287331e-06,
      "loss": 0.0005,
      "step": 124720
    },
    {
      "epoch": 2.04090648776896,
      "grad_norm": 0.012752222828567028,
      "learning_rate": 2.803420675615973e-06,
      "loss": 0.0008,
      "step": 124730
    },
    {
      "epoch": 2.041070113720036,
      "grad_norm": 0.24455346167087555,
      "learning_rate": 2.8025655565653374e-06,
      "loss": 0.0006,
      "step": 124740
    },
    {
      "epoch": 2.041233739671112,
      "grad_norm": 0.048273444175720215,
      "learning_rate": 2.8017105171664184e-06,
      "loss": 0.0004,
      "step": 124750
    },
    {
      "epoch": 2.0413973656221875,
      "grad_norm": 0.09782805293798447,
      "learning_rate": 2.8008555574502105e-06,
      "loss": 0.0017,
      "step": 124760
    },
    {
      "epoch": 2.0415609915732635,
      "grad_norm": 0.0613371916115284,
      "learning_rate": 2.800000677447702e-06,
      "loss": 0.001,
      "step": 124770
    },
    {
      "epoch": 2.0417246175243395,
      "grad_norm": 0.1340452879667282,
      "learning_rate": 2.7991458771898827e-06,
      "loss": 0.0013,
      "step": 124780
    },
    {
      "epoch": 2.041888243475415,
      "grad_norm": 0.1598801463842392,
      "learning_rate": 2.798291156707734e-06,
      "loss": 0.0014,
      "step": 124790
    },
    {
      "epoch": 2.042051869426491,
      "grad_norm": 0.27418652176856995,
      "learning_rate": 2.797436516032241e-06,
      "loss": 0.001,
      "step": 124800
    },
    {
      "epoch": 2.042215495377567,
      "grad_norm": 0.07217760384082794,
      "learning_rate": 2.796581955194381e-06,
      "loss": 0.002,
      "step": 124810
    },
    {
      "epoch": 2.0423791213286426,
      "grad_norm": 0.005071448162198067,
      "learning_rate": 2.795727474225127e-06,
      "loss": 0.0004,
      "step": 124820
    },
    {
      "epoch": 2.0425427472797186,
      "grad_norm": 0.22226819396018982,
      "learning_rate": 2.794873073155456e-06,
      "loss": 0.0008,
      "step": 124830
    },
    {
      "epoch": 2.0427063732307946,
      "grad_norm": 0.05380718410015106,
      "learning_rate": 2.7940187520163347e-06,
      "loss": 0.001,
      "step": 124840
    },
    {
      "epoch": 2.04286999918187,
      "grad_norm": 0.10900881886482239,
      "learning_rate": 2.793164510838734e-06,
      "loss": 0.0011,
      "step": 124850
    },
    {
      "epoch": 2.043033625132946,
      "grad_norm": 0.04940652847290039,
      "learning_rate": 2.792310349653613e-06,
      "loss": 0.0009,
      "step": 124860
    },
    {
      "epoch": 2.0431972510840217,
      "grad_norm": 0.08405929058790207,
      "learning_rate": 2.7914562684919377e-06,
      "loss": 0.0005,
      "step": 124870
    },
    {
      "epoch": 2.0433608770350977,
      "grad_norm": 0.274469792842865,
      "learning_rate": 2.7906022673846634e-06,
      "loss": 0.0011,
      "step": 124880
    },
    {
      "epoch": 2.0435245029861737,
      "grad_norm": 0.07355121523141861,
      "learning_rate": 2.789748346362746e-06,
      "loss": 0.0009,
      "step": 124890
    },
    {
      "epoch": 2.0436881289372493,
      "grad_norm": 0.07836034148931503,
      "learning_rate": 2.7888945054571415e-06,
      "loss": 0.0013,
      "step": 124900
    },
    {
      "epoch": 2.0438517548883253,
      "grad_norm": 0.05816254764795303,
      "learning_rate": 2.788040744698797e-06,
      "loss": 0.0006,
      "step": 124910
    },
    {
      "epoch": 2.0440153808394013,
      "grad_norm": 0.037988368421792984,
      "learning_rate": 2.787187064118658e-06,
      "loss": 0.001,
      "step": 124920
    },
    {
      "epoch": 2.044179006790477,
      "grad_norm": 0.0506337471306324,
      "learning_rate": 2.786333463747669e-06,
      "loss": 0.0007,
      "step": 124930
    },
    {
      "epoch": 2.044342632741553,
      "grad_norm": 0.13877160847187042,
      "learning_rate": 2.7854799436167713e-06,
      "loss": 0.0004,
      "step": 124940
    },
    {
      "epoch": 2.044506258692629,
      "grad_norm": 0.11030317842960358,
      "learning_rate": 2.784626503756904e-06,
      "loss": 0.0004,
      "step": 124950
    },
    {
      "epoch": 2.0446698846437044,
      "grad_norm": 0.1030447781085968,
      "learning_rate": 2.7837731441990006e-06,
      "loss": 0.0007,
      "step": 124960
    },
    {
      "epoch": 2.0448335105947804,
      "grad_norm": 0.04685341566801071,
      "learning_rate": 2.7829198649739957e-06,
      "loss": 0.0007,
      "step": 124970
    },
    {
      "epoch": 2.0449971365458564,
      "grad_norm": 0.02271329239010811,
      "learning_rate": 2.7820666661128155e-06,
      "loss": 0.0006,
      "step": 124980
    },
    {
      "epoch": 2.045160762496932,
      "grad_norm": 0.07728131860494614,
      "learning_rate": 2.7812135476463898e-06,
      "loss": 0.0008,
      "step": 124990
    },
    {
      "epoch": 2.045324388448008,
      "grad_norm": 0.0070898751728236675,
      "learning_rate": 2.7803605096056378e-06,
      "loss": 0.0013,
      "step": 125000
    },
    {
      "epoch": 2.045324388448008,
      "eval_loss": 0.0010078081395477057,
      "eval_runtime": 3.0959,
      "eval_samples_per_second": 64.601,
      "eval_steps_per_second": 16.15,
      "step": 125000
    },
    {
      "epoch": 2.045488014399084,
      "grad_norm": 0.1685505360364914,
      "learning_rate": 2.7795075520214855e-06,
      "loss": 0.0013,
      "step": 125010
    },
    {
      "epoch": 2.0456516403501595,
      "grad_norm": 0.17462106049060822,
      "learning_rate": 2.7786546749248468e-06,
      "loss": 0.0015,
      "step": 125020
    },
    {
      "epoch": 2.0458152663012354,
      "grad_norm": 0.0633067786693573,
      "learning_rate": 2.7778018783466353e-06,
      "loss": 0.0012,
      "step": 125030
    },
    {
      "epoch": 2.045978892252311,
      "grad_norm": 0.013548222370445728,
      "learning_rate": 2.776949162317767e-06,
      "loss": 0.0009,
      "step": 125040
    },
    {
      "epoch": 2.046142518203387,
      "grad_norm": 0.051946550607681274,
      "learning_rate": 2.7760965268691463e-06,
      "loss": 0.0012,
      "step": 125050
    },
    {
      "epoch": 2.046306144154463,
      "grad_norm": 0.0045087323524057865,
      "learning_rate": 2.775243972031684e-06,
      "loss": 0.0014,
      "step": 125060
    },
    {
      "epoch": 2.0464697701055385,
      "grad_norm": 0.12369585037231445,
      "learning_rate": 2.7743914978362786e-06,
      "loss": 0.0008,
      "step": 125070
    },
    {
      "epoch": 2.0466333960566145,
      "grad_norm": 0.028464335948228836,
      "learning_rate": 2.773539104313834e-06,
      "loss": 0.0006,
      "step": 125080
    },
    {
      "epoch": 2.0467970220076905,
      "grad_norm": 0.10350577533245087,
      "learning_rate": 2.772686791495244e-06,
      "loss": 0.0006,
      "step": 125090
    },
    {
      "epoch": 2.046960647958766,
      "grad_norm": 0.019794771447777748,
      "learning_rate": 2.7718345594114047e-06,
      "loss": 0.0009,
      "step": 125100
    },
    {
      "epoch": 2.047124273909842,
      "grad_norm": 0.0605911947786808,
      "learning_rate": 2.7709824080932113e-06,
      "loss": 0.001,
      "step": 125110
    },
    {
      "epoch": 2.047287899860918,
      "grad_norm": 0.15976080298423767,
      "learning_rate": 2.7701303375715437e-06,
      "loss": 0.0011,
      "step": 125120
    },
    {
      "epoch": 2.0474515258119936,
      "grad_norm": 0.09896986931562424,
      "learning_rate": 2.7692783478772943e-06,
      "loss": 0.0009,
      "step": 125130
    },
    {
      "epoch": 2.0476151517630696,
      "grad_norm": 0.16207146644592285,
      "learning_rate": 2.768426439041342e-06,
      "loss": 0.0013,
      "step": 125140
    },
    {
      "epoch": 2.0477787777141456,
      "grad_norm": 0.03074829652905464,
      "learning_rate": 2.767574611094569e-06,
      "loss": 0.0009,
      "step": 125150
    },
    {
      "epoch": 2.047942403665221,
      "grad_norm": 0.05201895907521248,
      "learning_rate": 2.7667228640678495e-06,
      "loss": 0.0008,
      "step": 125160
    },
    {
      "epoch": 2.048106029616297,
      "grad_norm": 0.009670531377196312,
      "learning_rate": 2.765871197992058e-06,
      "loss": 0.0024,
      "step": 125170
    },
    {
      "epoch": 2.048269655567373,
      "grad_norm": 0.07450035959482193,
      "learning_rate": 2.765019612898069e-06,
      "loss": 0.0011,
      "step": 125180
    },
    {
      "epoch": 2.0484332815184487,
      "grad_norm": 0.0816681906580925,
      "learning_rate": 2.764168108816745e-06,
      "loss": 0.0006,
      "step": 125190
    },
    {
      "epoch": 2.0485969074695247,
      "grad_norm": 0.06683312356472015,
      "learning_rate": 2.7633166857789545e-06,
      "loss": 0.0004,
      "step": 125200
    },
    {
      "epoch": 2.0487605334206007,
      "grad_norm": 0.030104627832770348,
      "learning_rate": 2.762465343815559e-06,
      "loss": 0.0013,
      "step": 125210
    },
    {
      "epoch": 2.0489241593716763,
      "grad_norm": 0.010404893197119236,
      "learning_rate": 2.7616140829574157e-06,
      "loss": 0.0011,
      "step": 125220
    },
    {
      "epoch": 2.0490877853227523,
      "grad_norm": 0.003294219495728612,
      "learning_rate": 2.7607629032353834e-06,
      "loss": 0.0005,
      "step": 125230
    },
    {
      "epoch": 2.049251411273828,
      "grad_norm": 0.08630453050136566,
      "learning_rate": 2.759911804680312e-06,
      "loss": 0.0009,
      "step": 125240
    },
    {
      "epoch": 2.049415037224904,
      "grad_norm": 0.1154833510518074,
      "learning_rate": 2.7590607873230555e-06,
      "loss": 0.001,
      "step": 125250
    },
    {
      "epoch": 2.04957866317598,
      "grad_norm": 0.09941628575325012,
      "learning_rate": 2.7582098511944576e-06,
      "loss": 0.0008,
      "step": 125260
    },
    {
      "epoch": 2.0497422891270554,
      "grad_norm": 0.03716711699962616,
      "learning_rate": 2.7573589963253665e-06,
      "loss": 0.0006,
      "step": 125270
    },
    {
      "epoch": 2.0499059150781314,
      "grad_norm": 0.0713198184967041,
      "learning_rate": 2.7565082227466208e-06,
      "loss": 0.0008,
      "step": 125280
    },
    {
      "epoch": 2.0500695410292074,
      "grad_norm": 0.13239486515522003,
      "learning_rate": 2.7556575304890608e-06,
      "loss": 0.0009,
      "step": 125290
    },
    {
      "epoch": 2.050233166980283,
      "grad_norm": 0.039633095264434814,
      "learning_rate": 2.75480691958352e-06,
      "loss": 0.0005,
      "step": 125300
    },
    {
      "epoch": 2.050396792931359,
      "grad_norm": 0.01314575970172882,
      "learning_rate": 2.7539563900608333e-06,
      "loss": 0.0007,
      "step": 125310
    },
    {
      "epoch": 2.050560418882435,
      "grad_norm": 0.0521257258951664,
      "learning_rate": 2.7531059419518298e-06,
      "loss": 0.0012,
      "step": 125320
    },
    {
      "epoch": 2.0507240448335105,
      "grad_norm": 0.03675038740038872,
      "learning_rate": 2.7522555752873337e-06,
      "loss": 0.0007,
      "step": 125330
    },
    {
      "epoch": 2.0508876707845864,
      "grad_norm": 0.026663633063435555,
      "learning_rate": 2.7514052900981723e-06,
      "loss": 0.001,
      "step": 125340
    },
    {
      "epoch": 2.0510512967356624,
      "grad_norm": 0.06920617818832397,
      "learning_rate": 2.750555086415163e-06,
      "loss": 0.001,
      "step": 125350
    },
    {
      "epoch": 2.051214922686738,
      "grad_norm": 0.09731853753328323,
      "learning_rate": 2.7497049642691277e-06,
      "loss": 0.0017,
      "step": 125360
    },
    {
      "epoch": 2.051378548637814,
      "grad_norm": 0.2518601417541504,
      "learning_rate": 2.748854923690876e-06,
      "loss": 0.0023,
      "step": 125370
    },
    {
      "epoch": 2.05154217458889,
      "grad_norm": 0.07137306779623032,
      "learning_rate": 2.748004964711224e-06,
      "loss": 0.0009,
      "step": 125380
    },
    {
      "epoch": 2.0517058005399655,
      "grad_norm": 0.11210113763809204,
      "learning_rate": 2.747155087360981e-06,
      "loss": 0.0009,
      "step": 125390
    },
    {
      "epoch": 2.0518694264910415,
      "grad_norm": 0.024057498201727867,
      "learning_rate": 2.7463052916709497e-06,
      "loss": 0.0007,
      "step": 125400
    },
    {
      "epoch": 2.0520330524421175,
      "grad_norm": 0.15057770907878876,
      "learning_rate": 2.7454555776719367e-06,
      "loss": 0.0015,
      "step": 125410
    },
    {
      "epoch": 2.052196678393193,
      "grad_norm": 0.08948255330324173,
      "learning_rate": 2.7446059453947405e-06,
      "loss": 0.0005,
      "step": 125420
    },
    {
      "epoch": 2.052360304344269,
      "grad_norm": 0.027236690744757652,
      "learning_rate": 2.7437563948701564e-06,
      "loss": 0.0006,
      "step": 125430
    },
    {
      "epoch": 2.0525239302953446,
      "grad_norm": 0.10346370190382004,
      "learning_rate": 2.7429069261289815e-06,
      "loss": 0.0008,
      "step": 125440
    },
    {
      "epoch": 2.0526875562464206,
      "grad_norm": 0.10105740278959274,
      "learning_rate": 2.7420575392020044e-06,
      "loss": 0.0006,
      "step": 125450
    },
    {
      "epoch": 2.0528511821974966,
      "grad_norm": 0.05941835790872574,
      "learning_rate": 2.7412082341200162e-06,
      "loss": 0.0038,
      "step": 125460
    },
    {
      "epoch": 2.053014808148572,
      "grad_norm": 0.12703943252563477,
      "learning_rate": 2.740359010913799e-06,
      "loss": 0.001,
      "step": 125470
    },
    {
      "epoch": 2.053178434099648,
      "grad_norm": 0.06517039239406586,
      "learning_rate": 2.7395098696141386e-06,
      "loss": 0.0013,
      "step": 125480
    },
    {
      "epoch": 2.053342060050724,
      "grad_norm": 0.08573034405708313,
      "learning_rate": 2.7386608102518103e-06,
      "loss": 0.001,
      "step": 125490
    },
    {
      "epoch": 2.0535056860017997,
      "grad_norm": 0.01614839769899845,
      "learning_rate": 2.7378118328575946e-06,
      "loss": 0.0007,
      "step": 125500
    },
    {
      "epoch": 2.0536693119528757,
      "grad_norm": 0.02678837440907955,
      "learning_rate": 2.7369629374622614e-06,
      "loss": 0.0009,
      "step": 125510
    },
    {
      "epoch": 2.0538329379039517,
      "grad_norm": 0.08163011819124222,
      "learning_rate": 2.7361141240965846e-06,
      "loss": 0.0008,
      "step": 125520
    },
    {
      "epoch": 2.0539965638550273,
      "grad_norm": 0.061010152101516724,
      "learning_rate": 2.7352653927913296e-06,
      "loss": 0.0012,
      "step": 125530
    },
    {
      "epoch": 2.0541601898061033,
      "grad_norm": 0.11241861432790756,
      "learning_rate": 2.7344167435772583e-06,
      "loss": 0.0008,
      "step": 125540
    },
    {
      "epoch": 2.0543238157571793,
      "grad_norm": 0.11194446682929993,
      "learning_rate": 2.733568176485138e-06,
      "loss": 0.0017,
      "step": 125550
    },
    {
      "epoch": 2.054487441708255,
      "grad_norm": 0.042952071875333786,
      "learning_rate": 2.732719691545721e-06,
      "loss": 0.0007,
      "step": 125560
    },
    {
      "epoch": 2.054651067659331,
      "grad_norm": 0.16491369903087616,
      "learning_rate": 2.731871288789769e-06,
      "loss": 0.0008,
      "step": 125570
    },
    {
      "epoch": 2.054814693610407,
      "grad_norm": 0.05632215738296509,
      "learning_rate": 2.7310229682480287e-06,
      "loss": 0.0017,
      "step": 125580
    },
    {
      "epoch": 2.0549783195614824,
      "grad_norm": 0.06487197428941727,
      "learning_rate": 2.7301747299512526e-06,
      "loss": 0.0005,
      "step": 125590
    },
    {
      "epoch": 2.0551419455125584,
      "grad_norm": 0.0015484405448660254,
      "learning_rate": 2.7293265739301896e-06,
      "loss": 0.0011,
      "step": 125600
    },
    {
      "epoch": 2.0553055714636344,
      "grad_norm": 0.046187739819288254,
      "learning_rate": 2.7284785002155785e-06,
      "loss": 0.001,
      "step": 125610
    },
    {
      "epoch": 2.05546919741471,
      "grad_norm": 0.101535864174366,
      "learning_rate": 2.727630508838166e-06,
      "loss": 0.0009,
      "step": 125620
    },
    {
      "epoch": 2.055632823365786,
      "grad_norm": 0.02309855818748474,
      "learning_rate": 2.7267825998286824e-06,
      "loss": 0.0009,
      "step": 125630
    },
    {
      "epoch": 2.0557964493168615,
      "grad_norm": 0.004552806727588177,
      "learning_rate": 2.725934773217868e-06,
      "loss": 0.0007,
      "step": 125640
    },
    {
      "epoch": 2.0559600752679374,
      "grad_norm": 0.05617595463991165,
      "learning_rate": 2.7250870290364506e-06,
      "loss": 0.0006,
      "step": 125650
    },
    {
      "epoch": 2.0561237012190134,
      "grad_norm": 0.1635713279247284,
      "learning_rate": 2.7242393673151605e-06,
      "loss": 0.001,
      "step": 125660
    },
    {
      "epoch": 2.056287327170089,
      "grad_norm": 0.05797175318002701,
      "learning_rate": 2.723391788084726e-06,
      "loss": 0.0009,
      "step": 125670
    },
    {
      "epoch": 2.056450953121165,
      "grad_norm": 0.12851861119270325,
      "learning_rate": 2.7225442913758654e-06,
      "loss": 0.0013,
      "step": 125680
    },
    {
      "epoch": 2.056614579072241,
      "grad_norm": 0.13630680739879608,
      "learning_rate": 2.7216968772193025e-06,
      "loss": 0.0015,
      "step": 125690
    },
    {
      "epoch": 2.0567782050233165,
      "grad_norm": 0.03139360621571541,
      "learning_rate": 2.7208495456457503e-06,
      "loss": 0.0008,
      "step": 125700
    },
    {
      "epoch": 2.0569418309743925,
      "grad_norm": 0.040974587202072144,
      "learning_rate": 2.7200022966859257e-06,
      "loss": 0.0012,
      "step": 125710
    },
    {
      "epoch": 2.0571054569254685,
      "grad_norm": 0.05311542749404907,
      "learning_rate": 2.7191551303705386e-06,
      "loss": 0.0012,
      "step": 125720
    },
    {
      "epoch": 2.057269082876544,
      "grad_norm": 0.08876026421785355,
      "learning_rate": 2.7183080467302936e-06,
      "loss": 0.0009,
      "step": 125730
    },
    {
      "epoch": 2.05743270882762,
      "grad_norm": 0.09702279418706894,
      "learning_rate": 2.7174610457959004e-06,
      "loss": 0.0007,
      "step": 125740
    },
    {
      "epoch": 2.057596334778696,
      "grad_norm": 0.02597414143383503,
      "learning_rate": 2.716614127598056e-06,
      "loss": 0.0006,
      "step": 125750
    },
    {
      "epoch": 2.0577599607297716,
      "grad_norm": 0.0771472305059433,
      "learning_rate": 2.7157672921674638e-06,
      "loss": 0.0007,
      "step": 125760
    },
    {
      "epoch": 2.0579235866808476,
      "grad_norm": 0.006495010107755661,
      "learning_rate": 2.7149205395348154e-06,
      "loss": 0.0008,
      "step": 125770
    },
    {
      "epoch": 2.0580872126319236,
      "grad_norm": 0.05171903595328331,
      "learning_rate": 2.7140738697308073e-06,
      "loss": 0.0008,
      "step": 125780
    },
    {
      "epoch": 2.058250838582999,
      "grad_norm": 0.15237563848495483,
      "learning_rate": 2.713227282786125e-06,
      "loss": 0.0005,
      "step": 125790
    },
    {
      "epoch": 2.058414464534075,
      "grad_norm": 0.06719356030225754,
      "learning_rate": 2.71238077873146e-06,
      "loss": 0.0018,
      "step": 125800
    },
    {
      "epoch": 2.0585780904851507,
      "grad_norm": 0.14036859571933746,
      "learning_rate": 2.7115343575974918e-06,
      "loss": 0.0011,
      "step": 125810
    },
    {
      "epoch": 2.0587417164362267,
      "grad_norm": 0.008981617167592049,
      "learning_rate": 2.7106880194149043e-06,
      "loss": 0.0016,
      "step": 125820
    },
    {
      "epoch": 2.0589053423873027,
      "grad_norm": 0.04011741653084755,
      "learning_rate": 2.7098417642143736e-06,
      "loss": 0.0008,
      "step": 125830
    },
    {
      "epoch": 2.0590689683383783,
      "grad_norm": 0.02799740433692932,
      "learning_rate": 2.708995592026574e-06,
      "loss": 0.0007,
      "step": 125840
    },
    {
      "epoch": 2.0592325942894543,
      "grad_norm": 0.004476170055568218,
      "learning_rate": 2.708149502882179e-06,
      "loss": 0.0015,
      "step": 125850
    },
    {
      "epoch": 2.0593962202405303,
      "grad_norm": 0.07621454447507858,
      "learning_rate": 2.707303496811854e-06,
      "loss": 0.0011,
      "step": 125860
    },
    {
      "epoch": 2.059559846191606,
      "grad_norm": 0.08320765197277069,
      "learning_rate": 2.7064575738462672e-06,
      "loss": 0.0009,
      "step": 125870
    },
    {
      "epoch": 2.059723472142682,
      "grad_norm": 0.04357193410396576,
      "learning_rate": 2.705611734016084e-06,
      "loss": 0.0006,
      "step": 125880
    },
    {
      "epoch": 2.059887098093758,
      "grad_norm": 0.10944069176912308,
      "learning_rate": 2.7047659773519575e-06,
      "loss": 0.0011,
      "step": 125890
    },
    {
      "epoch": 2.0600507240448334,
      "grad_norm": 0.07267940789461136,
      "learning_rate": 2.7039203038845507e-06,
      "loss": 0.0012,
      "step": 125900
    },
    {
      "epoch": 2.0602143499959094,
      "grad_norm": 0.04635965824127197,
      "learning_rate": 2.7030747136445117e-06,
      "loss": 0.0009,
      "step": 125910
    },
    {
      "epoch": 2.0603779759469854,
      "grad_norm": 0.0933053120970726,
      "learning_rate": 2.7022292066624984e-06,
      "loss": 0.0007,
      "step": 125920
    },
    {
      "epoch": 2.060541601898061,
      "grad_norm": 0.01606355980038643,
      "learning_rate": 2.7013837829691496e-06,
      "loss": 0.0006,
      "step": 125930
    },
    {
      "epoch": 2.060705227849137,
      "grad_norm": 0.0085771344602108,
      "learning_rate": 2.700538442595114e-06,
      "loss": 0.0008,
      "step": 125940
    },
    {
      "epoch": 2.060868853800213,
      "grad_norm": 0.06753838807344437,
      "learning_rate": 2.6996931855710356e-06,
      "loss": 0.0006,
      "step": 125950
    },
    {
      "epoch": 2.0610324797512884,
      "grad_norm": 0.13619694113731384,
      "learning_rate": 2.6988480119275486e-06,
      "loss": 0.0015,
      "step": 125960
    },
    {
      "epoch": 2.0611961057023644,
      "grad_norm": 0.10229191184043884,
      "learning_rate": 2.698002921695292e-06,
      "loss": 0.0016,
      "step": 125970
    },
    {
      "epoch": 2.0613597316534404,
      "grad_norm": 0.09704141318798065,
      "learning_rate": 2.6971579149048952e-06,
      "loss": 0.0008,
      "step": 125980
    },
    {
      "epoch": 2.061523357604516,
      "grad_norm": 0.10450167208909988,
      "learning_rate": 2.6963129915869914e-06,
      "loss": 0.0007,
      "step": 125990
    },
    {
      "epoch": 2.061686983555592,
      "grad_norm": 0.07419240474700928,
      "learning_rate": 2.695468151772202e-06,
      "loss": 0.0013,
      "step": 126000
    },
    {
      "epoch": 2.0618506095066675,
      "grad_norm": 0.0417998768389225,
      "learning_rate": 2.694623395491156e-06,
      "loss": 0.0014,
      "step": 126010
    },
    {
      "epoch": 2.0620142354577435,
      "grad_norm": 0.07510271668434143,
      "learning_rate": 2.693778722774471e-06,
      "loss": 0.001,
      "step": 126020
    },
    {
      "epoch": 2.0621778614088195,
      "grad_norm": 0.09251377731561661,
      "learning_rate": 2.6929341336527627e-06,
      "loss": 0.001,
      "step": 126030
    },
    {
      "epoch": 2.062341487359895,
      "grad_norm": 0.12444275617599487,
      "learning_rate": 2.6920896281566487e-06,
      "loss": 0.0007,
      "step": 126040
    },
    {
      "epoch": 2.062505113310971,
      "grad_norm": 0.1669503003358841,
      "learning_rate": 2.691245206316737e-06,
      "loss": 0.0011,
      "step": 126050
    },
    {
      "epoch": 2.062668739262047,
      "grad_norm": 0.07519892603158951,
      "learning_rate": 2.6904008681636394e-06,
      "loss": 0.0016,
      "step": 126060
    },
    {
      "epoch": 2.0628323652131226,
      "grad_norm": 0.17180950939655304,
      "learning_rate": 2.689556613727957e-06,
      "loss": 0.0007,
      "step": 126070
    },
    {
      "epoch": 2.0629959911641986,
      "grad_norm": 0.05516451224684715,
      "learning_rate": 2.6887124430402977e-06,
      "loss": 0.0006,
      "step": 126080
    },
    {
      "epoch": 2.0631596171152746,
      "grad_norm": 0.1077236458659172,
      "learning_rate": 2.6878683561312546e-06,
      "loss": 0.001,
      "step": 126090
    },
    {
      "epoch": 2.06332324306635,
      "grad_norm": 0.1515728086233139,
      "learning_rate": 2.687024353031426e-06,
      "loss": 0.0015,
      "step": 126100
    },
    {
      "epoch": 2.063486869017426,
      "grad_norm": 0.05036976560950279,
      "learning_rate": 2.686180433771408e-06,
      "loss": 0.0014,
      "step": 126110
    },
    {
      "epoch": 2.063650494968502,
      "grad_norm": 0.055411651730537415,
      "learning_rate": 2.685336598381788e-06,
      "loss": 0.0011,
      "step": 126120
    },
    {
      "epoch": 2.0638141209195777,
      "grad_norm": 0.15393434464931488,
      "learning_rate": 2.684492846893153e-06,
      "loss": 0.0015,
      "step": 126130
    },
    {
      "epoch": 2.0639777468706537,
      "grad_norm": 0.04202623292803764,
      "learning_rate": 2.683649179336085e-06,
      "loss": 0.0015,
      "step": 126140
    },
    {
      "epoch": 2.0641413728217297,
      "grad_norm": 0.0740785077214241,
      "learning_rate": 2.6828055957411683e-06,
      "loss": 0.001,
      "step": 126150
    },
    {
      "epoch": 2.0643049987728053,
      "grad_norm": 0.08951706439256668,
      "learning_rate": 2.6819620961389804e-06,
      "loss": 0.0006,
      "step": 126160
    },
    {
      "epoch": 2.0644686247238813,
      "grad_norm": 0.11555913090705872,
      "learning_rate": 2.681118680560094e-06,
      "loss": 0.0004,
      "step": 126170
    },
    {
      "epoch": 2.0646322506749573,
      "grad_norm": 0.033050332218408585,
      "learning_rate": 2.680275349035083e-06,
      "loss": 0.0007,
      "step": 126180
    },
    {
      "epoch": 2.064795876626033,
      "grad_norm": 0.056626204401254654,
      "learning_rate": 2.6794321015945142e-06,
      "loss": 0.0011,
      "step": 126190
    },
    {
      "epoch": 2.064959502577109,
      "grad_norm": 0.12083358317613602,
      "learning_rate": 2.678588938268956e-06,
      "loss": 0.0002,
      "step": 126200
    },
    {
      "epoch": 2.0651231285281844,
      "grad_norm": 0.0798579603433609,
      "learning_rate": 2.6777458590889675e-06,
      "loss": 0.0015,
      "step": 126210
    },
    {
      "epoch": 2.0652867544792604,
      "grad_norm": 0.005720012821257114,
      "learning_rate": 2.6769028640851124e-06,
      "loss": 0.0015,
      "step": 126220
    },
    {
      "epoch": 2.0654503804303364,
      "grad_norm": 0.006193923763930798,
      "learning_rate": 2.676059953287945e-06,
      "loss": 0.0005,
      "step": 126230
    },
    {
      "epoch": 2.065614006381412,
      "grad_norm": 0.004612925928086042,
      "learning_rate": 2.675217126728017e-06,
      "loss": 0.0009,
      "step": 126240
    },
    {
      "epoch": 2.065777632332488,
      "grad_norm": 0.077668696641922,
      "learning_rate": 2.6743743844358815e-06,
      "loss": 0.0015,
      "step": 126250
    },
    {
      "epoch": 2.065941258283564,
      "grad_norm": 0.0745009183883667,
      "learning_rate": 2.6735317264420847e-06,
      "loss": 0.0009,
      "step": 126260
    },
    {
      "epoch": 2.0661048842346394,
      "grad_norm": 0.03949791565537453,
      "learning_rate": 2.6726891527771725e-06,
      "loss": 0.0005,
      "step": 126270
    },
    {
      "epoch": 2.0662685101857154,
      "grad_norm": 0.020940886810421944,
      "learning_rate": 2.6718466634716834e-06,
      "loss": 0.0044,
      "step": 126280
    },
    {
      "epoch": 2.0664321361367914,
      "grad_norm": 0.005787865724414587,
      "learning_rate": 2.6710042585561584e-06,
      "loss": 0.0012,
      "step": 126290
    },
    {
      "epoch": 2.066595762087867,
      "grad_norm": 0.009020130150020123,
      "learning_rate": 2.67016193806113e-06,
      "loss": 0.0009,
      "step": 126300
    },
    {
      "epoch": 2.066759388038943,
      "grad_norm": 0.03346995636820793,
      "learning_rate": 2.6693197020171314e-06,
      "loss": 0.0012,
      "step": 126310
    },
    {
      "epoch": 2.066923013990019,
      "grad_norm": 0.08424467593431473,
      "learning_rate": 2.6684775504546966e-06,
      "loss": 0.0016,
      "step": 126320
    },
    {
      "epoch": 2.0670866399410945,
      "grad_norm": 0.05418844893574715,
      "learning_rate": 2.6676354834043424e-06,
      "loss": 0.001,
      "step": 126330
    },
    {
      "epoch": 2.0672502658921705,
      "grad_norm": 0.001397250802256167,
      "learning_rate": 2.6667935008965985e-06,
      "loss": 0.0012,
      "step": 126340
    },
    {
      "epoch": 2.0674138918432465,
      "grad_norm": 0.03630855306982994,
      "learning_rate": 2.6659516029619804e-06,
      "loss": 0.0008,
      "step": 126350
    },
    {
      "epoch": 2.067577517794322,
      "grad_norm": 0.09149971604347229,
      "learning_rate": 2.6651097896310086e-06,
      "loss": 0.0014,
      "step": 126360
    },
    {
      "epoch": 2.067741143745398,
      "grad_norm": 0.06094851717352867,
      "learning_rate": 2.6642680609341933e-06,
      "loss": 0.0028,
      "step": 126370
    },
    {
      "epoch": 2.067904769696474,
      "grad_norm": 0.06838110089302063,
      "learning_rate": 2.663426416902047e-06,
      "loss": 0.0007,
      "step": 126380
    },
    {
      "epoch": 2.0680683956475496,
      "grad_norm": 0.06788390129804611,
      "learning_rate": 2.66258485756508e-06,
      "loss": 0.0005,
      "step": 126390
    },
    {
      "epoch": 2.0682320215986256,
      "grad_norm": 0.01420030277222395,
      "learning_rate": 2.6617433829537908e-06,
      "loss": 0.0007,
      "step": 126400
    },
    {
      "epoch": 2.068395647549701,
      "grad_norm": 0.21180304884910583,
      "learning_rate": 2.6609019930986867e-06,
      "loss": 0.0009,
      "step": 126410
    },
    {
      "epoch": 2.068559273500777,
      "grad_norm": 0.0568615160882473,
      "learning_rate": 2.660060688030264e-06,
      "loss": 0.0007,
      "step": 126420
    },
    {
      "epoch": 2.068722899451853,
      "grad_norm": 0.007821504957973957,
      "learning_rate": 2.6592194677790152e-06,
      "loss": 0.0012,
      "step": 126430
    },
    {
      "epoch": 2.0688865254029287,
      "grad_norm": 0.002589412033557892,
      "learning_rate": 2.6583783323754363e-06,
      "loss": 0.0004,
      "step": 126440
    },
    {
      "epoch": 2.0690501513540047,
      "grad_norm": 0.28683483600616455,
      "learning_rate": 2.6575372818500132e-06,
      "loss": 0.0018,
      "step": 126450
    },
    {
      "epoch": 2.0692137773050807,
      "grad_norm": 0.03341438248753548,
      "learning_rate": 2.656696316233236e-06,
      "loss": 0.0011,
      "step": 126460
    },
    {
      "epoch": 2.0693774032561563,
      "grad_norm": 0.12024161964654922,
      "learning_rate": 2.6558554355555834e-06,
      "loss": 0.0007,
      "step": 126470
    },
    {
      "epoch": 2.0695410292072323,
      "grad_norm": 0.029168833047151566,
      "learning_rate": 2.6550146398475385e-06,
      "loss": 0.001,
      "step": 126480
    },
    {
      "epoch": 2.0697046551583083,
      "grad_norm": 0.02467380091547966,
      "learning_rate": 2.654173929139575e-06,
      "loss": 0.0013,
      "step": 126490
    },
    {
      "epoch": 2.069868281109384,
      "grad_norm": 0.1980421543121338,
      "learning_rate": 2.653333303462171e-06,
      "loss": 0.001,
      "step": 126500
    },
    {
      "epoch": 2.07003190706046,
      "grad_norm": 0.058259110897779465,
      "learning_rate": 2.652492762845793e-06,
      "loss": 0.0011,
      "step": 126510
    },
    {
      "epoch": 2.070195533011536,
      "grad_norm": 0.11114665120840073,
      "learning_rate": 2.6516523073209114e-06,
      "loss": 0.0008,
      "step": 126520
    },
    {
      "epoch": 2.0703591589626114,
      "grad_norm": 0.07873139530420303,
      "learning_rate": 2.6508119369179904e-06,
      "loss": 0.0017,
      "step": 126530
    },
    {
      "epoch": 2.0705227849136874,
      "grad_norm": 0.06262929737567902,
      "learning_rate": 2.649971651667488e-06,
      "loss": 0.0008,
      "step": 126540
    },
    {
      "epoch": 2.0706864108647633,
      "grad_norm": 0.08900629729032516,
      "learning_rate": 2.6491314515998666e-06,
      "loss": 0.0008,
      "step": 126550
    },
    {
      "epoch": 2.070850036815839,
      "grad_norm": 0.07310552895069122,
      "learning_rate": 2.6482913367455788e-06,
      "loss": 0.001,
      "step": 126560
    },
    {
      "epoch": 2.071013662766915,
      "grad_norm": 0.10595718026161194,
      "learning_rate": 2.64745130713508e-06,
      "loss": 0.0006,
      "step": 126570
    },
    {
      "epoch": 2.0711772887179905,
      "grad_norm": 0.06159178540110588,
      "learning_rate": 2.6466113627988145e-06,
      "loss": 0.0009,
      "step": 126580
    },
    {
      "epoch": 2.0713409146690664,
      "grad_norm": 0.032799605280160904,
      "learning_rate": 2.6457715037672312e-06,
      "loss": 0.0006,
      "step": 126590
    },
    {
      "epoch": 2.0715045406201424,
      "grad_norm": 0.08961030095815659,
      "learning_rate": 2.6449317300707742e-06,
      "loss": 0.001,
      "step": 126600
    },
    {
      "epoch": 2.071668166571218,
      "grad_norm": 0.007548108231276274,
      "learning_rate": 2.6440920417398804e-06,
      "loss": 0.0011,
      "step": 126610
    },
    {
      "epoch": 2.071831792522294,
      "grad_norm": 0.022765422239899635,
      "learning_rate": 2.643252438804991e-06,
      "loss": 0.0005,
      "step": 126620
    },
    {
      "epoch": 2.07199541847337,
      "grad_norm": 0.030695730820298195,
      "learning_rate": 2.642412921296533e-06,
      "loss": 0.0004,
      "step": 126630
    },
    {
      "epoch": 2.0721590444244455,
      "grad_norm": 0.006426703650504351,
      "learning_rate": 2.64157348924494e-06,
      "loss": 0.0009,
      "step": 126640
    },
    {
      "epoch": 2.0723226703755215,
      "grad_norm": 0.06154328212141991,
      "learning_rate": 2.640734142680642e-06,
      "loss": 0.001,
      "step": 126650
    },
    {
      "epoch": 2.0724862963265975,
      "grad_norm": 0.0014200330479070544,
      "learning_rate": 2.6398948816340587e-06,
      "loss": 0.0004,
      "step": 126660
    },
    {
      "epoch": 2.072649922277673,
      "grad_norm": 0.008701635524630547,
      "learning_rate": 2.639055706135616e-06,
      "loss": 0.0008,
      "step": 126670
    },
    {
      "epoch": 2.072813548228749,
      "grad_norm": 0.05183007940649986,
      "learning_rate": 2.6382166162157276e-06,
      "loss": 0.0005,
      "step": 126680
    },
    {
      "epoch": 2.072977174179825,
      "grad_norm": 0.24049779772758484,
      "learning_rate": 2.6373776119048118e-06,
      "loss": 0.002,
      "step": 126690
    },
    {
      "epoch": 2.0731408001309006,
      "grad_norm": 0.025092674419283867,
      "learning_rate": 2.636538693233277e-06,
      "loss": 0.001,
      "step": 126700
    },
    {
      "epoch": 2.0733044260819766,
      "grad_norm": 0.0905616506934166,
      "learning_rate": 2.635699860231537e-06,
      "loss": 0.001,
      "step": 126710
    },
    {
      "epoch": 2.0734680520330526,
      "grad_norm": 0.03912216052412987,
      "learning_rate": 2.634861112929994e-06,
      "loss": 0.001,
      "step": 126720
    },
    {
      "epoch": 2.073631677984128,
      "grad_norm": 0.13487021625041962,
      "learning_rate": 2.6340224513590496e-06,
      "loss": 0.0012,
      "step": 126730
    },
    {
      "epoch": 2.073795303935204,
      "grad_norm": 0.053416233509778976,
      "learning_rate": 2.633183875549107e-06,
      "loss": 0.0014,
      "step": 126740
    },
    {
      "epoch": 2.07395892988628,
      "grad_norm": 0.004410147201269865,
      "learning_rate": 2.632345385530558e-06,
      "loss": 0.0005,
      "step": 126750
    },
    {
      "epoch": 2.0741225558373557,
      "grad_norm": 0.007632214110344648,
      "learning_rate": 2.6315069813338e-06,
      "loss": 0.0035,
      "step": 126760
    },
    {
      "epoch": 2.0742861817884317,
      "grad_norm": 0.007893110625445843,
      "learning_rate": 2.63066866298922e-06,
      "loss": 0.0006,
      "step": 126770
    },
    {
      "epoch": 2.0744498077395077,
      "grad_norm": 0.0865241214632988,
      "learning_rate": 2.6298304305272084e-06,
      "loss": 0.0009,
      "step": 126780
    },
    {
      "epoch": 2.0746134336905833,
      "grad_norm": 0.05783380940556526,
      "learning_rate": 2.6289922839781445e-06,
      "loss": 0.0012,
      "step": 126790
    },
    {
      "epoch": 2.0747770596416593,
      "grad_norm": 0.08939733356237411,
      "learning_rate": 2.628154223372412e-06,
      "loss": 0.0008,
      "step": 126800
    },
    {
      "epoch": 2.074940685592735,
      "grad_norm": 0.048028670251369476,
      "learning_rate": 2.627316248740391e-06,
      "loss": 0.0005,
      "step": 126810
    },
    {
      "epoch": 2.075104311543811,
      "grad_norm": 0.14464986324310303,
      "learning_rate": 2.6264783601124525e-06,
      "loss": 0.0015,
      "step": 126820
    },
    {
      "epoch": 2.075267937494887,
      "grad_norm": 0.15395614504814148,
      "learning_rate": 2.625640557518969e-06,
      "loss": 0.001,
      "step": 126830
    },
    {
      "epoch": 2.0754315634459624,
      "grad_norm": 0.20120304822921753,
      "learning_rate": 2.624802840990307e-06,
      "loss": 0.0009,
      "step": 126840
    },
    {
      "epoch": 2.0755951893970384,
      "grad_norm": 0.03794405236840248,
      "learning_rate": 2.6239652105568356e-06,
      "loss": 0.0009,
      "step": 126850
    },
    {
      "epoch": 2.0757588153481144,
      "grad_norm": 0.09057488292455673,
      "learning_rate": 2.6231276662489126e-06,
      "loss": 0.0014,
      "step": 126860
    },
    {
      "epoch": 2.07592244129919,
      "grad_norm": 0.007605287246406078,
      "learning_rate": 2.622290208096899e-06,
      "loss": 0.0031,
      "step": 126870
    },
    {
      "epoch": 2.076086067250266,
      "grad_norm": 0.013623950071632862,
      "learning_rate": 2.621452836131152e-06,
      "loss": 0.0006,
      "step": 126880
    },
    {
      "epoch": 2.076249693201342,
      "grad_norm": 0.007958965376019478,
      "learning_rate": 2.620615550382022e-06,
      "loss": 0.0008,
      "step": 126890
    },
    {
      "epoch": 2.0764133191524174,
      "grad_norm": 0.099567711353302,
      "learning_rate": 2.61977835087986e-06,
      "loss": 0.0008,
      "step": 126900
    },
    {
      "epoch": 2.0765769451034934,
      "grad_norm": 0.08914471417665482,
      "learning_rate": 2.6189412376550114e-06,
      "loss": 0.0016,
      "step": 126910
    },
    {
      "epoch": 2.0767405710545694,
      "grad_norm": 0.14258307218551636,
      "learning_rate": 2.6181042107378222e-06,
      "loss": 0.0012,
      "step": 126920
    },
    {
      "epoch": 2.076904197005645,
      "grad_norm": 0.09684636443853378,
      "learning_rate": 2.6172672701586306e-06,
      "loss": 0.0006,
      "step": 126930
    },
    {
      "epoch": 2.077067822956721,
      "grad_norm": 0.30909478664398193,
      "learning_rate": 2.6164304159477714e-06,
      "loss": 0.0008,
      "step": 126940
    },
    {
      "epoch": 2.077231448907797,
      "grad_norm": 0.1499127447605133,
      "learning_rate": 2.6155936481355827e-06,
      "loss": 0.0007,
      "step": 126950
    },
    {
      "epoch": 2.0773950748588725,
      "grad_norm": 0.11543216556310654,
      "learning_rate": 2.614756966752392e-06,
      "loss": 0.0006,
      "step": 126960
    },
    {
      "epoch": 2.0775587008099485,
      "grad_norm": 0.07330917567014694,
      "learning_rate": 2.61392037182853e-06,
      "loss": 0.0009,
      "step": 126970
    },
    {
      "epoch": 2.077722326761024,
      "grad_norm": 0.026456883177161217,
      "learning_rate": 2.613083863394318e-06,
      "loss": 0.0008,
      "step": 126980
    },
    {
      "epoch": 2.0778859527121,
      "grad_norm": 0.17785537242889404,
      "learning_rate": 2.6122474414800815e-06,
      "loss": 0.001,
      "step": 126990
    },
    {
      "epoch": 2.078049578663176,
      "grad_norm": 0.02467922866344452,
      "learning_rate": 2.611411106116134e-06,
      "loss": 0.0006,
      "step": 127000
    },
    {
      "epoch": 2.0782132046142516,
      "grad_norm": 0.033409010618925095,
      "learning_rate": 2.610574857332795e-06,
      "loss": 0.0007,
      "step": 127010
    },
    {
      "epoch": 2.0783768305653276,
      "grad_norm": 0.015222851186990738,
      "learning_rate": 2.6097386951603743e-06,
      "loss": 0.0009,
      "step": 127020
    },
    {
      "epoch": 2.0785404565164036,
      "grad_norm": 0.10475429892539978,
      "learning_rate": 2.6089026196291795e-06,
      "loss": 0.0007,
      "step": 127030
    },
    {
      "epoch": 2.078704082467479,
      "grad_norm": 0.09920498728752136,
      "learning_rate": 2.60806663076952e-06,
      "loss": 0.001,
      "step": 127040
    },
    {
      "epoch": 2.078867708418555,
      "grad_norm": 0.1791248768568039,
      "learning_rate": 2.6072307286116936e-06,
      "loss": 0.0009,
      "step": 127050
    },
    {
      "epoch": 2.079031334369631,
      "grad_norm": 0.012286944314837456,
      "learning_rate": 2.6063949131860057e-06,
      "loss": 0.0007,
      "step": 127060
    },
    {
      "epoch": 2.0791949603207067,
      "grad_norm": 0.03031480312347412,
      "learning_rate": 2.6055591845227462e-06,
      "loss": 0.0006,
      "step": 127070
    },
    {
      "epoch": 2.0793585862717827,
      "grad_norm": 0.27137839794158936,
      "learning_rate": 2.6047235426522114e-06,
      "loss": 0.001,
      "step": 127080
    },
    {
      "epoch": 2.0795222122228587,
      "grad_norm": 0.01968301273882389,
      "learning_rate": 2.603887987604694e-06,
      "loss": 0.0004,
      "step": 127090
    },
    {
      "epoch": 2.0796858381739343,
      "grad_norm": 0.09470977634191513,
      "learning_rate": 2.6030525194104755e-06,
      "loss": 0.0009,
      "step": 127100
    },
    {
      "epoch": 2.0798494641250103,
      "grad_norm": 0.11560133099555969,
      "learning_rate": 2.6022171380998444e-06,
      "loss": 0.0006,
      "step": 127110
    },
    {
      "epoch": 2.0800130900760863,
      "grad_norm": 0.11293934285640717,
      "learning_rate": 2.601381843703079e-06,
      "loss": 0.0015,
      "step": 127120
    },
    {
      "epoch": 2.080176716027162,
      "grad_norm": 0.07455451786518097,
      "learning_rate": 2.600546636250457e-06,
      "loss": 0.0013,
      "step": 127130
    },
    {
      "epoch": 2.080340341978238,
      "grad_norm": 0.02396007999777794,
      "learning_rate": 2.5997115157722508e-06,
      "loss": 0.0011,
      "step": 127140
    },
    {
      "epoch": 2.080503967929314,
      "grad_norm": 0.10201536864042282,
      "learning_rate": 2.5988764822987335e-06,
      "loss": 0.0004,
      "step": 127150
    },
    {
      "epoch": 2.0806675938803894,
      "grad_norm": 0.09642160683870316,
      "learning_rate": 2.598041535860174e-06,
      "loss": 0.0007,
      "step": 127160
    },
    {
      "epoch": 2.0808312198314654,
      "grad_norm": 0.15044547617435455,
      "learning_rate": 2.5972066764868343e-06,
      "loss": 0.0007,
      "step": 127170
    },
    {
      "epoch": 2.080994845782541,
      "grad_norm": 0.17772674560546875,
      "learning_rate": 2.5963719042089795e-06,
      "loss": 0.0011,
      "step": 127180
    },
    {
      "epoch": 2.081158471733617,
      "grad_norm": 0.06801004707813263,
      "learning_rate": 2.595537219056864e-06,
      "loss": 0.0007,
      "step": 127190
    },
    {
      "epoch": 2.081322097684693,
      "grad_norm": 0.052550725638866425,
      "learning_rate": 2.5947026210607475e-06,
      "loss": 0.0009,
      "step": 127200
    },
    {
      "epoch": 2.0814857236357684,
      "grad_norm": 0.0493013821542263,
      "learning_rate": 2.5938681102508776e-06,
      "loss": 0.0024,
      "step": 127210
    },
    {
      "epoch": 2.0816493495868444,
      "grad_norm": 0.0904151052236557,
      "learning_rate": 2.5930336866575085e-06,
      "loss": 0.0009,
      "step": 127220
    },
    {
      "epoch": 2.0818129755379204,
      "grad_norm": 0.2083015739917755,
      "learning_rate": 2.5921993503108826e-06,
      "loss": 0.0012,
      "step": 127230
    },
    {
      "epoch": 2.081976601488996,
      "grad_norm": 0.002322424203157425,
      "learning_rate": 2.5913651012412416e-06,
      "loss": 0.0013,
      "step": 127240
    },
    {
      "epoch": 2.082140227440072,
      "grad_norm": 0.040781036019325256,
      "learning_rate": 2.5905309394788287e-06,
      "loss": 0.0005,
      "step": 127250
    },
    {
      "epoch": 2.082303853391148,
      "grad_norm": 0.20012660324573517,
      "learning_rate": 2.5896968650538763e-06,
      "loss": 0.0013,
      "step": 127260
    },
    {
      "epoch": 2.0824674793422235,
      "grad_norm": 0.08569946885108948,
      "learning_rate": 2.588862877996621e-06,
      "loss": 0.0016,
      "step": 127270
    },
    {
      "epoch": 2.0826311052932995,
      "grad_norm": 0.06452517211437225,
      "learning_rate": 2.58802897833729e-06,
      "loss": 0.001,
      "step": 127280
    },
    {
      "epoch": 2.0827947312443755,
      "grad_norm": 0.03474804759025574,
      "learning_rate": 2.5871951661061133e-06,
      "loss": 0.0005,
      "step": 127290
    },
    {
      "epoch": 2.082958357195451,
      "grad_norm": 0.013522247783839703,
      "learning_rate": 2.5863614413333106e-06,
      "loss": 0.0012,
      "step": 127300
    },
    {
      "epoch": 2.083121983146527,
      "grad_norm": 0.22929659485816956,
      "learning_rate": 2.585527804049105e-06,
      "loss": 0.0009,
      "step": 127310
    },
    {
      "epoch": 2.083285609097603,
      "grad_norm": 0.05873755365610123,
      "learning_rate": 2.584694254283717e-06,
      "loss": 0.0011,
      "step": 127320
    },
    {
      "epoch": 2.0834492350486786,
      "grad_norm": 0.05695571377873421,
      "learning_rate": 2.583860792067353e-06,
      "loss": 0.0009,
      "step": 127330
    },
    {
      "epoch": 2.0836128609997546,
      "grad_norm": 0.04229049012064934,
      "learning_rate": 2.5830274174302312e-06,
      "loss": 0.0011,
      "step": 127340
    },
    {
      "epoch": 2.08377648695083,
      "grad_norm": 0.03599122539162636,
      "learning_rate": 2.5821941304025542e-06,
      "loss": 0.0007,
      "step": 127350
    },
    {
      "epoch": 2.083940112901906,
      "grad_norm": 0.08846523612737656,
      "learning_rate": 2.581360931014528e-06,
      "loss": 0.001,
      "step": 127360
    },
    {
      "epoch": 2.084103738852982,
      "grad_norm": 0.0830601379275322,
      "learning_rate": 2.5805278192963577e-06,
      "loss": 0.0011,
      "step": 127370
    },
    {
      "epoch": 2.0842673648040577,
      "grad_norm": 0.010371796786785126,
      "learning_rate": 2.5796947952782375e-06,
      "loss": 0.001,
      "step": 127380
    },
    {
      "epoch": 2.0844309907551337,
      "grad_norm": 0.16246408224105835,
      "learning_rate": 2.5788618589903657e-06,
      "loss": 0.0012,
      "step": 127390
    },
    {
      "epoch": 2.0845946167062097,
      "grad_norm": 0.006537714973092079,
      "learning_rate": 2.5780290104629303e-06,
      "loss": 0.0007,
      "step": 127400
    },
    {
      "epoch": 2.0847582426572853,
      "grad_norm": 0.020766111090779305,
      "learning_rate": 2.577196249726124e-06,
      "loss": 0.0026,
      "step": 127410
    },
    {
      "epoch": 2.0849218686083613,
      "grad_norm": 0.06766985356807709,
      "learning_rate": 2.576363576810128e-06,
      "loss": 0.0008,
      "step": 127420
    },
    {
      "epoch": 2.0850854945594373,
      "grad_norm": 0.19978763163089752,
      "learning_rate": 2.57553099174513e-06,
      "loss": 0.0011,
      "step": 127430
    },
    {
      "epoch": 2.085249120510513,
      "grad_norm": 0.03641851618885994,
      "learning_rate": 2.5746984945613062e-06,
      "loss": 0.0004,
      "step": 127440
    },
    {
      "epoch": 2.085412746461589,
      "grad_norm": 0.06487105041742325,
      "learning_rate": 2.5738660852888316e-06,
      "loss": 0.001,
      "step": 127450
    },
    {
      "epoch": 2.085576372412665,
      "grad_norm": 0.09939587861299515,
      "learning_rate": 2.5730337639578816e-06,
      "loss": 0.0008,
      "step": 127460
    },
    {
      "epoch": 2.0857399983637404,
      "grad_norm": 0.12017474323511124,
      "learning_rate": 2.572201530598623e-06,
      "loss": 0.0005,
      "step": 127470
    },
    {
      "epoch": 2.0859036243148164,
      "grad_norm": 0.08546924591064453,
      "learning_rate": 2.571369385241226e-06,
      "loss": 0.0013,
      "step": 127480
    },
    {
      "epoch": 2.0860672502658923,
      "grad_norm": 0.019498176872730255,
      "learning_rate": 2.5705373279158495e-06,
      "loss": 0.0011,
      "step": 127490
    },
    {
      "epoch": 2.086230876216968,
      "grad_norm": 0.008147344924509525,
      "learning_rate": 2.569705358652658e-06,
      "loss": 0.0005,
      "step": 127500
    },
    {
      "epoch": 2.086394502168044,
      "grad_norm": 0.0732748731970787,
      "learning_rate": 2.5688734774818037e-06,
      "loss": 0.002,
      "step": 127510
    },
    {
      "epoch": 2.08655812811912,
      "grad_norm": 0.17362816631793976,
      "learning_rate": 2.5680416844334427e-06,
      "loss": 0.0009,
      "step": 127520
    },
    {
      "epoch": 2.0867217540701954,
      "grad_norm": 0.09583607316017151,
      "learning_rate": 2.5672099795377304e-06,
      "loss": 0.0007,
      "step": 127530
    },
    {
      "epoch": 2.0868853800212714,
      "grad_norm": 0.04557041451334953,
      "learning_rate": 2.566378362824804e-06,
      "loss": 0.001,
      "step": 127540
    },
    {
      "epoch": 2.0870490059723474,
      "grad_norm": 0.07319483160972595,
      "learning_rate": 2.5655468343248156e-06,
      "loss": 0.0005,
      "step": 127550
    },
    {
      "epoch": 2.087212631923423,
      "grad_norm": 0.05781925842165947,
      "learning_rate": 2.564715394067902e-06,
      "loss": 0.0005,
      "step": 127560
    },
    {
      "epoch": 2.087376257874499,
      "grad_norm": 0.0935058519244194,
      "learning_rate": 2.5638840420842036e-06,
      "loss": 0.0008,
      "step": 127570
    },
    {
      "epoch": 2.0875398838255745,
      "grad_norm": 0.00953678973019123,
      "learning_rate": 2.563052778403852e-06,
      "loss": 0.0011,
      "step": 127580
    },
    {
      "epoch": 2.0877035097766505,
      "grad_norm": 0.03904193639755249,
      "learning_rate": 2.5622216030569802e-06,
      "loss": 0.0012,
      "step": 127590
    },
    {
      "epoch": 2.0878671357277265,
      "grad_norm": 0.09811195731163025,
      "learning_rate": 2.5613905160737173e-06,
      "loss": 0.0008,
      "step": 127600
    },
    {
      "epoch": 2.088030761678802,
      "grad_norm": 0.05469155311584473,
      "learning_rate": 2.560559517484186e-06,
      "loss": 0.0011,
      "step": 127610
    },
    {
      "epoch": 2.088194387629878,
      "grad_norm": 0.0048719062469899654,
      "learning_rate": 2.5597286073185114e-06,
      "loss": 0.0006,
      "step": 127620
    },
    {
      "epoch": 2.088358013580954,
      "grad_norm": 0.003756213467568159,
      "learning_rate": 2.5588977856068094e-06,
      "loss": 0.0006,
      "step": 127630
    },
    {
      "epoch": 2.0885216395320296,
      "grad_norm": 0.04063684120774269,
      "learning_rate": 2.5580670523791934e-06,
      "loss": 0.0005,
      "step": 127640
    },
    {
      "epoch": 2.0886852654831056,
      "grad_norm": 0.07121266424655914,
      "learning_rate": 2.5572364076657796e-06,
      "loss": 0.0016,
      "step": 127650
    },
    {
      "epoch": 2.0888488914341816,
      "grad_norm": 0.03202417120337486,
      "learning_rate": 2.556405851496674e-06,
      "loss": 0.0011,
      "step": 127660
    },
    {
      "epoch": 2.089012517385257,
      "grad_norm": 0.0037659406661987305,
      "learning_rate": 2.555575383901984e-06,
      "loss": 0.0004,
      "step": 127670
    },
    {
      "epoch": 2.089176143336333,
      "grad_norm": 0.0045356908813118935,
      "learning_rate": 2.55474500491181e-06,
      "loss": 0.0012,
      "step": 127680
    },
    {
      "epoch": 2.089339769287409,
      "grad_norm": 0.05127200856804848,
      "learning_rate": 2.553914714556255e-06,
      "loss": 0.0008,
      "step": 127690
    },
    {
      "epoch": 2.0895033952384847,
      "grad_norm": 0.10988567769527435,
      "learning_rate": 2.5530845128654092e-06,
      "loss": 0.0015,
      "step": 127700
    },
    {
      "epoch": 2.0896670211895607,
      "grad_norm": 0.06253285706043243,
      "learning_rate": 2.5522543998693717e-06,
      "loss": 0.0007,
      "step": 127710
    },
    {
      "epoch": 2.0898306471406367,
      "grad_norm": 0.03700726851820946,
      "learning_rate": 2.551424375598227e-06,
      "loss": 0.001,
      "step": 127720
    },
    {
      "epoch": 2.0899942730917123,
      "grad_norm": 0.08548401296138763,
      "learning_rate": 2.550594440082066e-06,
      "loss": 0.0008,
      "step": 127730
    },
    {
      "epoch": 2.0901578990427883,
      "grad_norm": 0.0011281209299340844,
      "learning_rate": 2.5497645933509697e-06,
      "loss": 0.0009,
      "step": 127740
    },
    {
      "epoch": 2.090321524993864,
      "grad_norm": 0.1124374121427536,
      "learning_rate": 2.548934835435015e-06,
      "loss": 0.0008,
      "step": 127750
    },
    {
      "epoch": 2.09048515094494,
      "grad_norm": 0.047349508851766586,
      "learning_rate": 2.548105166364284e-06,
      "loss": 0.0016,
      "step": 127760
    },
    {
      "epoch": 2.090648776896016,
      "grad_norm": 0.1386995166540146,
      "learning_rate": 2.5472755861688466e-06,
      "loss": 0.0012,
      "step": 127770
    },
    {
      "epoch": 2.0908124028470914,
      "grad_norm": 0.33263498544692993,
      "learning_rate": 2.546446094878775e-06,
      "loss": 0.0016,
      "step": 127780
    },
    {
      "epoch": 2.0909760287981674,
      "grad_norm": 0.05350246652960777,
      "learning_rate": 2.5456166925241345e-06,
      "loss": 0.0009,
      "step": 127790
    },
    {
      "epoch": 2.0911396547492433,
      "grad_norm": 0.01669466681778431,
      "learning_rate": 2.5447873791349893e-06,
      "loss": 0.0009,
      "step": 127800
    },
    {
      "epoch": 2.091303280700319,
      "grad_norm": 0.048967618495225906,
      "learning_rate": 2.5439581547414026e-06,
      "loss": 0.0007,
      "step": 127810
    },
    {
      "epoch": 2.091466906651395,
      "grad_norm": 0.5545778274536133,
      "learning_rate": 2.543129019373428e-06,
      "loss": 0.0011,
      "step": 127820
    },
    {
      "epoch": 2.091630532602471,
      "grad_norm": 0.05185588076710701,
      "learning_rate": 2.5422999730611253e-06,
      "loss": 0.0006,
      "step": 127830
    },
    {
      "epoch": 2.0917941585535464,
      "grad_norm": 0.0487663559615612,
      "learning_rate": 2.541471015834538e-06,
      "loss": 0.0008,
      "step": 127840
    },
    {
      "epoch": 2.0919577845046224,
      "grad_norm": 0.12212276458740234,
      "learning_rate": 2.5406421477237176e-06,
      "loss": 0.0016,
      "step": 127850
    },
    {
      "epoch": 2.0921214104556984,
      "grad_norm": 0.06837128102779388,
      "learning_rate": 2.5398133687587103e-06,
      "loss": 0.0013,
      "step": 127860
    },
    {
      "epoch": 2.092285036406774,
      "grad_norm": 0.04178814962506294,
      "learning_rate": 2.538984678969553e-06,
      "loss": 0.0011,
      "step": 127870
    },
    {
      "epoch": 2.09244866235785,
      "grad_norm": 0.20282943546772003,
      "learning_rate": 2.5381560783862883e-06,
      "loss": 0.0013,
      "step": 127880
    },
    {
      "epoch": 2.092612288308926,
      "grad_norm": 0.28980764746665955,
      "learning_rate": 2.5373275670389474e-06,
      "loss": 0.001,
      "step": 127890
    },
    {
      "epoch": 2.0927759142600015,
      "grad_norm": 0.029055584222078323,
      "learning_rate": 2.536499144957565e-06,
      "loss": 0.0011,
      "step": 127900
    },
    {
      "epoch": 2.0929395402110775,
      "grad_norm": 0.0028315105009824038,
      "learning_rate": 2.5356708121721664e-06,
      "loss": 0.0007,
      "step": 127910
    },
    {
      "epoch": 2.0931031661621535,
      "grad_norm": 0.009406630881130695,
      "learning_rate": 2.5348425687127788e-06,
      "loss": 0.0004,
      "step": 127920
    },
    {
      "epoch": 2.093266792113229,
      "grad_norm": 0.01671595871448517,
      "learning_rate": 2.5340144146094235e-06,
      "loss": 0.0005,
      "step": 127930
    },
    {
      "epoch": 2.093430418064305,
      "grad_norm": 0.0425165519118309,
      "learning_rate": 2.533186349892117e-06,
      "loss": 0.001,
      "step": 127940
    },
    {
      "epoch": 2.0935940440153806,
      "grad_norm": 0.1454932689666748,
      "learning_rate": 2.5323583745908793e-06,
      "loss": 0.0009,
      "step": 127950
    },
    {
      "epoch": 2.0937576699664566,
      "grad_norm": 0.05067780241370201,
      "learning_rate": 2.5315304887357174e-06,
      "loss": 0.0007,
      "step": 127960
    },
    {
      "epoch": 2.0939212959175326,
      "grad_norm": 0.09026180952787399,
      "learning_rate": 2.5307026923566434e-06,
      "loss": 0.0034,
      "step": 127970
    },
    {
      "epoch": 2.094084921868608,
      "grad_norm": 0.05677011236548424,
      "learning_rate": 2.529874985483661e-06,
      "loss": 0.001,
      "step": 127980
    },
    {
      "epoch": 2.094248547819684,
      "grad_norm": 0.06247275695204735,
      "learning_rate": 2.5290473681467753e-06,
      "loss": 0.0005,
      "step": 127990
    },
    {
      "epoch": 2.09441217377076,
      "grad_norm": 0.08413197100162506,
      "learning_rate": 2.5282198403759816e-06,
      "loss": 0.0004,
      "step": 128000
    },
    {
      "epoch": 2.0945757997218357,
      "grad_norm": 0.013439448550343513,
      "learning_rate": 2.5273924022012782e-06,
      "loss": 0.0008,
      "step": 128010
    },
    {
      "epoch": 2.0947394256729117,
      "grad_norm": 0.008650384843349457,
      "learning_rate": 2.5265650536526586e-06,
      "loss": 0.001,
      "step": 128020
    },
    {
      "epoch": 2.0949030516239877,
      "grad_norm": 0.0435977466404438,
      "learning_rate": 2.5257377947601124e-06,
      "loss": 0.0004,
      "step": 128030
    },
    {
      "epoch": 2.0950666775750633,
      "grad_norm": 0.06966936588287354,
      "learning_rate": 2.5249106255536238e-06,
      "loss": 0.0006,
      "step": 128040
    },
    {
      "epoch": 2.0952303035261393,
      "grad_norm": 0.020104793831706047,
      "learning_rate": 2.524083546063174e-06,
      "loss": 0.001,
      "step": 128050
    },
    {
      "epoch": 2.0953939294772153,
      "grad_norm": 0.053379885852336884,
      "learning_rate": 2.5232565563187474e-06,
      "loss": 0.0014,
      "step": 128060
    },
    {
      "epoch": 2.095557555428291,
      "grad_norm": 0.08038292080163956,
      "learning_rate": 2.522429656350316e-06,
      "loss": 0.0012,
      "step": 128070
    },
    {
      "epoch": 2.095721181379367,
      "grad_norm": 0.11014243215322495,
      "learning_rate": 2.5216028461878538e-06,
      "loss": 0.0018,
      "step": 128080
    },
    {
      "epoch": 2.095884807330443,
      "grad_norm": 0.05124003440141678,
      "learning_rate": 2.520776125861335e-06,
      "loss": 0.0012,
      "step": 128090
    },
    {
      "epoch": 2.0960484332815184,
      "grad_norm": 0.09936625510454178,
      "learning_rate": 2.519949495400721e-06,
      "loss": 0.0009,
      "step": 128100
    },
    {
      "epoch": 2.0962120592325943,
      "grad_norm": 0.027762778103351593,
      "learning_rate": 2.5191229548359785e-06,
      "loss": 0.0005,
      "step": 128110
    },
    {
      "epoch": 2.09637568518367,
      "grad_norm": 0.08110368996858597,
      "learning_rate": 2.518296504197064e-06,
      "loss": 0.0006,
      "step": 128120
    },
    {
      "epoch": 2.096539311134746,
      "grad_norm": 0.006522516254335642,
      "learning_rate": 2.517470143513939e-06,
      "loss": 0.0015,
      "step": 128130
    },
    {
      "epoch": 2.096702937085822,
      "grad_norm": 0.07181324064731598,
      "learning_rate": 2.5166438728165555e-06,
      "loss": 0.0009,
      "step": 128140
    },
    {
      "epoch": 2.0968665630368974,
      "grad_norm": 0.07044582813978195,
      "learning_rate": 2.51581769213486e-06,
      "loss": 0.0006,
      "step": 128150
    },
    {
      "epoch": 2.0970301889879734,
      "grad_norm": 0.13840921223163605,
      "learning_rate": 2.5149916014988046e-06,
      "loss": 0.0011,
      "step": 128160
    },
    {
      "epoch": 2.0971938149390494,
      "grad_norm": 0.13683614134788513,
      "learning_rate": 2.5141656009383297e-06,
      "loss": 0.0013,
      "step": 128170
    },
    {
      "epoch": 2.097357440890125,
      "grad_norm": 0.1456993669271469,
      "learning_rate": 2.5133396904833783e-06,
      "loss": 0.0008,
      "step": 128180
    },
    {
      "epoch": 2.097521066841201,
      "grad_norm": 0.05824455991387367,
      "learning_rate": 2.512513870163885e-06,
      "loss": 0.0006,
      "step": 128190
    },
    {
      "epoch": 2.097684692792277,
      "grad_norm": 0.05801787227392197,
      "learning_rate": 2.511688140009788e-06,
      "loss": 0.001,
      "step": 128200
    },
    {
      "epoch": 2.0978483187433525,
      "grad_norm": 0.04467809945344925,
      "learning_rate": 2.5108625000510133e-06,
      "loss": 0.0008,
      "step": 128210
    },
    {
      "epoch": 2.0980119446944285,
      "grad_norm": 0.061349913477897644,
      "learning_rate": 2.5100369503174913e-06,
      "loss": 0.0012,
      "step": 128220
    },
    {
      "epoch": 2.0981755706455045,
      "grad_norm": 0.23603478074073792,
      "learning_rate": 2.5092114908391467e-06,
      "loss": 0.0012,
      "step": 128230
    },
    {
      "epoch": 2.09833919659658,
      "grad_norm": 0.09558813273906708,
      "learning_rate": 2.5083861216458973e-06,
      "loss": 0.0009,
      "step": 128240
    },
    {
      "epoch": 2.098502822547656,
      "grad_norm": 0.039386648684740067,
      "learning_rate": 2.5075608427676634e-06,
      "loss": 0.0011,
      "step": 128250
    },
    {
      "epoch": 2.098666448498732,
      "grad_norm": 0.00835203193128109,
      "learning_rate": 2.5067356542343567e-06,
      "loss": 0.0007,
      "step": 128260
    },
    {
      "epoch": 2.0988300744498076,
      "grad_norm": 0.04446261748671532,
      "learning_rate": 2.5059105560758926e-06,
      "loss": 0.0009,
      "step": 128270
    },
    {
      "epoch": 2.0989937004008836,
      "grad_norm": 0.05806954577565193,
      "learning_rate": 2.505085548322175e-06,
      "loss": 0.0008,
      "step": 128280
    },
    {
      "epoch": 2.0991573263519596,
      "grad_norm": 0.2028346210718155,
      "learning_rate": 2.5042606310031085e-06,
      "loss": 0.0018,
      "step": 128290
    },
    {
      "epoch": 2.099320952303035,
      "grad_norm": 0.0670069083571434,
      "learning_rate": 2.5034358041485983e-06,
      "loss": 0.0011,
      "step": 128300
    },
    {
      "epoch": 2.099484578254111,
      "grad_norm": 0.006675695534795523,
      "learning_rate": 2.5026110677885385e-06,
      "loss": 0.0004,
      "step": 128310
    },
    {
      "epoch": 2.099648204205187,
      "grad_norm": 0.07480159401893616,
      "learning_rate": 2.5017864219528266e-06,
      "loss": 0.0011,
      "step": 128320
    },
    {
      "epoch": 2.0998118301562627,
      "grad_norm": 0.18533605337142944,
      "learning_rate": 2.500961866671353e-06,
      "loss": 0.0006,
      "step": 128330
    },
    {
      "epoch": 2.0999754561073387,
      "grad_norm": 0.13750103116035461,
      "learning_rate": 2.500137401974005e-06,
      "loss": 0.0009,
      "step": 128340
    },
    {
      "epoch": 2.1001390820584143,
      "grad_norm": 0.1057664230465889,
      "learning_rate": 2.4993130278906664e-06,
      "loss": 0.0009,
      "step": 128350
    },
    {
      "epoch": 2.1003027080094903,
      "grad_norm": 0.09257295727729797,
      "learning_rate": 2.4984887444512197e-06,
      "loss": 0.0009,
      "step": 128360
    },
    {
      "epoch": 2.1004663339605663,
      "grad_norm": 0.04528724029660225,
      "learning_rate": 2.497664551685546e-06,
      "loss": 0.0006,
      "step": 128370
    },
    {
      "epoch": 2.100629959911642,
      "grad_norm": 0.08403719216585159,
      "learning_rate": 2.4968404496235167e-06,
      "loss": 0.001,
      "step": 128380
    },
    {
      "epoch": 2.100793585862718,
      "grad_norm": 0.008571864105761051,
      "learning_rate": 2.496016438295006e-06,
      "loss": 0.0006,
      "step": 128390
    },
    {
      "epoch": 2.100957211813794,
      "grad_norm": 0.013630916364490986,
      "learning_rate": 2.4951925177298795e-06,
      "loss": 0.0012,
      "step": 128400
    },
    {
      "epoch": 2.1011208377648694,
      "grad_norm": 0.05910114943981171,
      "learning_rate": 2.4943686879580063e-06,
      "loss": 0.0007,
      "step": 128410
    },
    {
      "epoch": 2.1012844637159453,
      "grad_norm": 0.05891280248761177,
      "learning_rate": 2.4935449490092443e-06,
      "loss": 0.0019,
      "step": 128420
    },
    {
      "epoch": 2.1014480896670213,
      "grad_norm": 0.13174556195735931,
      "learning_rate": 2.4927213009134556e-06,
      "loss": 0.0007,
      "step": 128430
    },
    {
      "epoch": 2.101611715618097,
      "grad_norm": 0.054453738033771515,
      "learning_rate": 2.491897743700493e-06,
      "loss": 0.0007,
      "step": 128440
    },
    {
      "epoch": 2.101775341569173,
      "grad_norm": 0.0880228653550148,
      "learning_rate": 2.4910742774002077e-06,
      "loss": 0.0007,
      "step": 128450
    },
    {
      "epoch": 2.101938967520249,
      "grad_norm": 0.05741582810878754,
      "learning_rate": 2.490250902042452e-06,
      "loss": 0.001,
      "step": 128460
    },
    {
      "epoch": 2.1021025934713244,
      "grad_norm": 0.06323856860399246,
      "learning_rate": 2.489427617657067e-06,
      "loss": 0.0004,
      "step": 128470
    },
    {
      "epoch": 2.1022662194224004,
      "grad_norm": 0.0194765105843544,
      "learning_rate": 2.4886044242738984e-06,
      "loss": 0.001,
      "step": 128480
    },
    {
      "epoch": 2.1024298453734764,
      "grad_norm": 0.14218570291996002,
      "learning_rate": 2.487781321922782e-06,
      "loss": 0.0015,
      "step": 128490
    },
    {
      "epoch": 2.102593471324552,
      "grad_norm": 0.06330674886703491,
      "learning_rate": 2.4869583106335563e-06,
      "loss": 0.001,
      "step": 128500
    },
    {
      "epoch": 2.102757097275628,
      "grad_norm": 0.12341947853565216,
      "learning_rate": 2.48613539043605e-06,
      "loss": 0.0006,
      "step": 128510
    },
    {
      "epoch": 2.1029207232267035,
      "grad_norm": 0.05440869927406311,
      "learning_rate": 2.4853125613600943e-06,
      "loss": 0.0008,
      "step": 128520
    },
    {
      "epoch": 2.1030843491777795,
      "grad_norm": 0.02906016632914543,
      "learning_rate": 2.4844898234355174e-06,
      "loss": 0.0011,
      "step": 128530
    },
    {
      "epoch": 2.1032479751288555,
      "grad_norm": 0.037199489772319794,
      "learning_rate": 2.4836671766921356e-06,
      "loss": 0.0022,
      "step": 128540
    },
    {
      "epoch": 2.103411601079931,
      "grad_norm": 0.10032356530427933,
      "learning_rate": 2.482844621159772e-06,
      "loss": 0.0015,
      "step": 128550
    },
    {
      "epoch": 2.103575227031007,
      "grad_norm": 0.015970641747117043,
      "learning_rate": 2.4820221568682397e-06,
      "loss": 0.0009,
      "step": 128560
    },
    {
      "epoch": 2.103738852982083,
      "grad_norm": 0.07271319627761841,
      "learning_rate": 2.4811997838473517e-06,
      "loss": 0.0016,
      "step": 128570
    },
    {
      "epoch": 2.1039024789331586,
      "grad_norm": 0.09008684009313583,
      "learning_rate": 2.4803775021269206e-06,
      "loss": 0.0013,
      "step": 128580
    },
    {
      "epoch": 2.1040661048842346,
      "grad_norm": 0.04694205895066261,
      "learning_rate": 2.4795553117367464e-06,
      "loss": 0.0007,
      "step": 128590
    },
    {
      "epoch": 2.1042297308353106,
      "grad_norm": 0.14489251375198364,
      "learning_rate": 2.478733212706637e-06,
      "loss": 0.0013,
      "step": 128600
    },
    {
      "epoch": 2.104393356786386,
      "grad_norm": 0.002187573816627264,
      "learning_rate": 2.4779112050663865e-06,
      "loss": 0.0009,
      "step": 128610
    },
    {
      "epoch": 2.104556982737462,
      "grad_norm": 0.20876021683216095,
      "learning_rate": 2.477089288845796e-06,
      "loss": 0.0008,
      "step": 128620
    },
    {
      "epoch": 2.104720608688538,
      "grad_norm": 0.11453013122081757,
      "learning_rate": 2.476267464074654e-06,
      "loss": 0.0009,
      "step": 128630
    },
    {
      "epoch": 2.1048842346396137,
      "grad_norm": 0.04934345930814743,
      "learning_rate": 2.4754457307827496e-06,
      "loss": 0.0007,
      "step": 128640
    },
    {
      "epoch": 2.1050478605906897,
      "grad_norm": 0.12709276378154755,
      "learning_rate": 2.4746240889998717e-06,
      "loss": 0.0009,
      "step": 128650
    },
    {
      "epoch": 2.1052114865417657,
      "grad_norm": 0.04703143611550331,
      "learning_rate": 2.473802538755799e-06,
      "loss": 0.0014,
      "step": 128660
    },
    {
      "epoch": 2.1053751124928413,
      "grad_norm": 0.048432085663080215,
      "learning_rate": 2.4729810800803147e-06,
      "loss": 0.0015,
      "step": 128670
    },
    {
      "epoch": 2.1055387384439173,
      "grad_norm": 0.15259037911891937,
      "learning_rate": 2.4721597130031904e-06,
      "loss": 0.001,
      "step": 128680
    },
    {
      "epoch": 2.1057023643949933,
      "grad_norm": 0.16169042885303497,
      "learning_rate": 2.471338437554203e-06,
      "loss": 0.0013,
      "step": 128690
    },
    {
      "epoch": 2.105865990346069,
      "grad_norm": 0.02376764826476574,
      "learning_rate": 2.4705172537631184e-06,
      "loss": 0.0005,
      "step": 128700
    },
    {
      "epoch": 2.106029616297145,
      "grad_norm": 0.016686461865901947,
      "learning_rate": 2.4696961616597055e-06,
      "loss": 0.0008,
      "step": 128710
    },
    {
      "epoch": 2.1061932422482204,
      "grad_norm": 0.11294475197792053,
      "learning_rate": 2.468875161273723e-06,
      "loss": 0.0012,
      "step": 128720
    },
    {
      "epoch": 2.1063568681992964,
      "grad_norm": 0.26991215348243713,
      "learning_rate": 2.468054252634935e-06,
      "loss": 0.0014,
      "step": 128730
    },
    {
      "epoch": 2.1065204941503723,
      "grad_norm": 0.03612404689192772,
      "learning_rate": 2.4672334357730947e-06,
      "loss": 0.0005,
      "step": 128740
    },
    {
      "epoch": 2.106684120101448,
      "grad_norm": 0.02044064551591873,
      "learning_rate": 2.466412710717953e-06,
      "loss": 0.0009,
      "step": 128750
    },
    {
      "epoch": 2.106847746052524,
      "grad_norm": 0.2583995461463928,
      "learning_rate": 2.4655920774992638e-06,
      "loss": 0.0012,
      "step": 128760
    },
    {
      "epoch": 2.1070113720036,
      "grad_norm": 0.18512903153896332,
      "learning_rate": 2.4647715361467677e-06,
      "loss": 0.0012,
      "step": 128770
    },
    {
      "epoch": 2.1071749979546754,
      "grad_norm": 0.14752094447612762,
      "learning_rate": 2.4639510866902123e-06,
      "loss": 0.0011,
      "step": 128780
    },
    {
      "epoch": 2.1073386239057514,
      "grad_norm": 0.10565897822380066,
      "learning_rate": 2.463130729159333e-06,
      "loss": 0.0015,
      "step": 128790
    },
    {
      "epoch": 2.1075022498568274,
      "grad_norm": 0.013931378722190857,
      "learning_rate": 2.462310463583868e-06,
      "loss": 0.0007,
      "step": 128800
    },
    {
      "epoch": 2.107665875807903,
      "grad_norm": 0.009323307313024998,
      "learning_rate": 2.4614902899935505e-06,
      "loss": 0.0019,
      "step": 128810
    },
    {
      "epoch": 2.107829501758979,
      "grad_norm": 0.05106600001454353,
      "learning_rate": 2.4606702084181077e-06,
      "loss": 0.0008,
      "step": 128820
    },
    {
      "epoch": 2.107993127710055,
      "grad_norm": 0.012281034141778946,
      "learning_rate": 2.4598502188872704e-06,
      "loss": 0.0008,
      "step": 128830
    },
    {
      "epoch": 2.1081567536611305,
      "grad_norm": 0.08368691802024841,
      "learning_rate": 2.459030321430753e-06,
      "loss": 0.0008,
      "step": 128840
    },
    {
      "epoch": 2.1083203796122065,
      "grad_norm": 0.08300020545721054,
      "learning_rate": 2.4582105160782803e-06,
      "loss": 0.0007,
      "step": 128850
    },
    {
      "epoch": 2.1084840055632825,
      "grad_norm": 0.013460628688335419,
      "learning_rate": 2.457390802859569e-06,
      "loss": 0.0012,
      "step": 128860
    },
    {
      "epoch": 2.108647631514358,
      "grad_norm": 0.13130810856819153,
      "learning_rate": 2.456571181804328e-06,
      "loss": 0.0011,
      "step": 128870
    },
    {
      "epoch": 2.108811257465434,
      "grad_norm": 0.092060387134552,
      "learning_rate": 2.455751652942271e-06,
      "loss": 0.0009,
      "step": 128880
    },
    {
      "epoch": 2.10897488341651,
      "grad_norm": 0.24746465682983398,
      "learning_rate": 2.4549322163030997e-06,
      "loss": 0.0009,
      "step": 128890
    },
    {
      "epoch": 2.1091385093675856,
      "grad_norm": 0.10120374709367752,
      "learning_rate": 2.4541128719165203e-06,
      "loss": 0.0005,
      "step": 128900
    },
    {
      "epoch": 2.1093021353186616,
      "grad_norm": 0.09808824956417084,
      "learning_rate": 2.453293619812228e-06,
      "loss": 0.0008,
      "step": 128910
    },
    {
      "epoch": 2.109465761269737,
      "grad_norm": 0.08791407197713852,
      "learning_rate": 2.4524744600199242e-06,
      "loss": 0.0014,
      "step": 128920
    },
    {
      "epoch": 2.109629387220813,
      "grad_norm": 0.10329283028841019,
      "learning_rate": 2.4516553925692978e-06,
      "loss": 0.002,
      "step": 128930
    },
    {
      "epoch": 2.109793013171889,
      "grad_norm": 0.012688706628978252,
      "learning_rate": 2.4508364174900367e-06,
      "loss": 0.0006,
      "step": 128940
    },
    {
      "epoch": 2.1099566391229647,
      "grad_norm": 0.15681971609592438,
      "learning_rate": 2.4500175348118304e-06,
      "loss": 0.0014,
      "step": 128950
    },
    {
      "epoch": 2.1101202650740407,
      "grad_norm": 0.09324906021356583,
      "learning_rate": 2.4491987445643584e-06,
      "loss": 0.0006,
      "step": 128960
    },
    {
      "epoch": 2.1102838910251167,
      "grad_norm": 0.047888562083244324,
      "learning_rate": 2.4483800467773026e-06,
      "loss": 0.0006,
      "step": 128970
    },
    {
      "epoch": 2.1104475169761923,
      "grad_norm": 0.05987181514501572,
      "learning_rate": 2.4475614414803357e-06,
      "loss": 0.0008,
      "step": 128980
    },
    {
      "epoch": 2.1106111429272683,
      "grad_norm": 0.16377393901348114,
      "learning_rate": 2.446742928703134e-06,
      "loss": 0.0011,
      "step": 128990
    },
    {
      "epoch": 2.1107747688783443,
      "grad_norm": 0.08190133422613144,
      "learning_rate": 2.4459245084753626e-06,
      "loss": 0.0008,
      "step": 129000
    },
    {
      "epoch": 2.11093839482942,
      "grad_norm": 0.18243969976902008,
      "learning_rate": 2.445106180826689e-06,
      "loss": 0.0011,
      "step": 129010
    },
    {
      "epoch": 2.111102020780496,
      "grad_norm": 0.06374434381723404,
      "learning_rate": 2.444287945786778e-06,
      "loss": 0.0017,
      "step": 129020
    },
    {
      "epoch": 2.111265646731572,
      "grad_norm": 0.12796828150749207,
      "learning_rate": 2.443469803385286e-06,
      "loss": 0.0012,
      "step": 129030
    },
    {
      "epoch": 2.1114292726826474,
      "grad_norm": 0.009844370186328888,
      "learning_rate": 2.4426517536518695e-06,
      "loss": 0.0004,
      "step": 129040
    },
    {
      "epoch": 2.1115928986337233,
      "grad_norm": 0.02215762436389923,
      "learning_rate": 2.4418337966161786e-06,
      "loss": 0.0007,
      "step": 129050
    },
    {
      "epoch": 2.1117565245847993,
      "grad_norm": 0.03647473827004433,
      "learning_rate": 2.441015932307864e-06,
      "loss": 0.0008,
      "step": 129060
    },
    {
      "epoch": 2.111920150535875,
      "grad_norm": 0.1903500258922577,
      "learning_rate": 2.4401981607565737e-06,
      "loss": 0.0008,
      "step": 129070
    },
    {
      "epoch": 2.112083776486951,
      "grad_norm": 0.1311829835176468,
      "learning_rate": 2.439380481991946e-06,
      "loss": 0.0004,
      "step": 129080
    },
    {
      "epoch": 2.112247402438027,
      "grad_norm": 0.15377609431743622,
      "learning_rate": 2.438562896043623e-06,
      "loss": 0.0014,
      "step": 129090
    },
    {
      "epoch": 2.1124110283891024,
      "grad_norm": 0.08897221088409424,
      "learning_rate": 2.437745402941237e-06,
      "loss": 0.0008,
      "step": 129100
    },
    {
      "epoch": 2.1125746543401784,
      "grad_norm": 0.05890337750315666,
      "learning_rate": 2.4369280027144236e-06,
      "loss": 0.0011,
      "step": 129110
    },
    {
      "epoch": 2.112738280291254,
      "grad_norm": 0.07072455435991287,
      "learning_rate": 2.4361106953928088e-06,
      "loss": 0.0012,
      "step": 129120
    },
    {
      "epoch": 2.11290190624233,
      "grad_norm": 0.04468684270977974,
      "learning_rate": 2.4352934810060205e-06,
      "loss": 0.0008,
      "step": 129130
    },
    {
      "epoch": 2.113065532193406,
      "grad_norm": 0.05255168676376343,
      "learning_rate": 2.434476359583679e-06,
      "loss": 0.002,
      "step": 129140
    },
    {
      "epoch": 2.1132291581444815,
      "grad_norm": 0.05946776643395424,
      "learning_rate": 2.4336593311554022e-06,
      "loss": 0.0007,
      "step": 129150
    },
    {
      "epoch": 2.1133927840955575,
      "grad_norm": 0.13737957179546356,
      "learning_rate": 2.432842395750808e-06,
      "loss": 0.001,
      "step": 129160
    },
    {
      "epoch": 2.1135564100466335,
      "grad_norm": 0.091910719871521,
      "learning_rate": 2.4320255533995056e-06,
      "loss": 0.0009,
      "step": 129170
    },
    {
      "epoch": 2.113720035997709,
      "grad_norm": 0.014578170143067837,
      "learning_rate": 2.4312088041311066e-06,
      "loss": 0.0008,
      "step": 129180
    },
    {
      "epoch": 2.113883661948785,
      "grad_norm": 0.06120060756802559,
      "learning_rate": 2.4303921479752124e-06,
      "loss": 0.0008,
      "step": 129190
    },
    {
      "epoch": 2.114047287899861,
      "grad_norm": 0.1434168666601181,
      "learning_rate": 2.429575584961429e-06,
      "loss": 0.001,
      "step": 129200
    },
    {
      "epoch": 2.1142109138509366,
      "grad_norm": 0.013018888421356678,
      "learning_rate": 2.4287591151193507e-06,
      "loss": 0.0006,
      "step": 129210
    },
    {
      "epoch": 2.1143745398020126,
      "grad_norm": 0.0070012821815907955,
      "learning_rate": 2.4279427384785748e-06,
      "loss": 0.001,
      "step": 129220
    },
    {
      "epoch": 2.1145381657530886,
      "grad_norm": 0.10873915255069733,
      "learning_rate": 2.427126455068697e-06,
      "loss": 0.0008,
      "step": 129230
    },
    {
      "epoch": 2.114701791704164,
      "grad_norm": 0.0031114406883716583,
      "learning_rate": 2.426310264919297e-06,
      "loss": 0.0008,
      "step": 129240
    },
    {
      "epoch": 2.11486541765524,
      "grad_norm": 0.05338197946548462,
      "learning_rate": 2.4254941680599665e-06,
      "loss": 0.0004,
      "step": 129250
    },
    {
      "epoch": 2.115029043606316,
      "grad_norm": 0.0195078756660223,
      "learning_rate": 2.4246781645202826e-06,
      "loss": 0.0008,
      "step": 129260
    },
    {
      "epoch": 2.1151926695573917,
      "grad_norm": 0.08901358395814896,
      "learning_rate": 2.4238622543298277e-06,
      "loss": 0.0006,
      "step": 129270
    },
    {
      "epoch": 2.1153562955084677,
      "grad_norm": 0.05972006171941757,
      "learning_rate": 2.423046437518173e-06,
      "loss": 0.0011,
      "step": 129280
    },
    {
      "epoch": 2.1155199214595433,
      "grad_norm": 0.05445694923400879,
      "learning_rate": 2.422230714114891e-06,
      "loss": 0.001,
      "step": 129290
    },
    {
      "epoch": 2.1156835474106193,
      "grad_norm": 0.018343938514590263,
      "learning_rate": 2.421415084149552e-06,
      "loss": 0.0007,
      "step": 129300
    },
    {
      "epoch": 2.1158471733616953,
      "grad_norm": 0.020434966310858727,
      "learning_rate": 2.420599547651717e-06,
      "loss": 0.0009,
      "step": 129310
    },
    {
      "epoch": 2.116010799312771,
      "grad_norm": 0.014766097068786621,
      "learning_rate": 2.41978410465095e-06,
      "loss": 0.0008,
      "step": 129320
    },
    {
      "epoch": 2.116174425263847,
      "grad_norm": 0.031918562948703766,
      "learning_rate": 2.4189687551768087e-06,
      "loss": 0.0013,
      "step": 129330
    },
    {
      "epoch": 2.116338051214923,
      "grad_norm": 0.01247891690582037,
      "learning_rate": 2.418153499258845e-06,
      "loss": 0.0007,
      "step": 129340
    },
    {
      "epoch": 2.1165016771659984,
      "grad_norm": 0.0784144252538681,
      "learning_rate": 2.417338336926613e-06,
      "loss": 0.0008,
      "step": 129350
    },
    {
      "epoch": 2.1166653031170743,
      "grad_norm": 0.007132540922611952,
      "learning_rate": 2.4165232682096574e-06,
      "loss": 0.0012,
      "step": 129360
    },
    {
      "epoch": 2.1168289290681503,
      "grad_norm": 0.13466258347034454,
      "learning_rate": 2.415708293137526e-06,
      "loss": 0.0009,
      "step": 129370
    },
    {
      "epoch": 2.116992555019226,
      "grad_norm": 0.13869674503803253,
      "learning_rate": 2.4148934117397553e-06,
      "loss": 0.0008,
      "step": 129380
    },
    {
      "epoch": 2.117156180970302,
      "grad_norm": 0.0726885050535202,
      "learning_rate": 2.414078624045888e-06,
      "loss": 0.0009,
      "step": 129390
    },
    {
      "epoch": 2.117319806921378,
      "grad_norm": 0.13166485726833344,
      "learning_rate": 2.4132639300854534e-06,
      "loss": 0.0007,
      "step": 129400
    },
    {
      "epoch": 2.1174834328724534,
      "grad_norm": 0.10797033458948135,
      "learning_rate": 2.412449329887986e-06,
      "loss": 0.001,
      "step": 129410
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 0.058146655559539795,
      "learning_rate": 2.4116348234830097e-06,
      "loss": 0.0014,
      "step": 129420
    },
    {
      "epoch": 2.1178106847746054,
      "grad_norm": 0.16427108645439148,
      "learning_rate": 2.4108204109000517e-06,
      "loss": 0.0011,
      "step": 129430
    },
    {
      "epoch": 2.117974310725681,
      "grad_norm": 0.15037886798381805,
      "learning_rate": 2.410006092168631e-06,
      "loss": 0.0011,
      "step": 129440
    },
    {
      "epoch": 2.118137936676757,
      "grad_norm": 0.02971005067229271,
      "learning_rate": 2.4091918673182625e-06,
      "loss": 0.0005,
      "step": 129450
    },
    {
      "epoch": 2.118301562627833,
      "grad_norm": 0.03783925622701645,
      "learning_rate": 2.408377736378464e-06,
      "loss": 0.0004,
      "step": 129460
    },
    {
      "epoch": 2.1184651885789085,
      "grad_norm": 0.14881537854671478,
      "learning_rate": 2.4075636993787416e-06,
      "loss": 0.0014,
      "step": 129470
    },
    {
      "epoch": 2.1186288145299845,
      "grad_norm": 0.14918789267539978,
      "learning_rate": 2.4067497563486064e-06,
      "loss": 0.0009,
      "step": 129480
    },
    {
      "epoch": 2.1187924404810605,
      "grad_norm": 0.007797111291438341,
      "learning_rate": 2.405935907317558e-06,
      "loss": 0.0006,
      "step": 129490
    },
    {
      "epoch": 2.118956066432136,
      "grad_norm": 0.10536335408687592,
      "learning_rate": 2.405122152315098e-06,
      "loss": 0.0009,
      "step": 129500
    },
    {
      "epoch": 2.119119692383212,
      "grad_norm": 0.0683324858546257,
      "learning_rate": 2.404308491370726e-06,
      "loss": 0.0005,
      "step": 129510
    },
    {
      "epoch": 2.1192833183342876,
      "grad_norm": 0.05309787392616272,
      "learning_rate": 2.4034949245139298e-06,
      "loss": 0.001,
      "step": 129520
    },
    {
      "epoch": 2.1194469442853636,
      "grad_norm": 0.07505122572183609,
      "learning_rate": 2.4026814517742047e-06,
      "loss": 0.0007,
      "step": 129530
    },
    {
      "epoch": 2.1196105702364396,
      "grad_norm": 0.08722759038209915,
      "learning_rate": 2.401868073181034e-06,
      "loss": 0.0007,
      "step": 129540
    },
    {
      "epoch": 2.119774196187515,
      "grad_norm": 0.002473121974617243,
      "learning_rate": 2.4010547887639014e-06,
      "loss": 0.0013,
      "step": 129550
    },
    {
      "epoch": 2.119937822138591,
      "grad_norm": 0.4195724129676819,
      "learning_rate": 2.4002415985522848e-06,
      "loss": 0.0014,
      "step": 129560
    },
    {
      "epoch": 2.120101448089667,
      "grad_norm": 0.04630810767412186,
      "learning_rate": 2.3994285025756615e-06,
      "loss": 0.0004,
      "step": 129570
    },
    {
      "epoch": 2.1202650740407427,
      "grad_norm": 0.13098259270191193,
      "learning_rate": 2.3986155008635066e-06,
      "loss": 0.001,
      "step": 129580
    },
    {
      "epoch": 2.1204286999918187,
      "grad_norm": 0.05063588544726372,
      "learning_rate": 2.3978025934452854e-06,
      "loss": 0.0005,
      "step": 129590
    },
    {
      "epoch": 2.1205923259428947,
      "grad_norm": 0.055887069553136826,
      "learning_rate": 2.396989780350468e-06,
      "loss": 0.0011,
      "step": 129600
    },
    {
      "epoch": 2.1207559518939703,
      "grad_norm": 0.22654558718204498,
      "learning_rate": 2.396177061608513e-06,
      "loss": 0.0005,
      "step": 129610
    },
    {
      "epoch": 2.1209195778450463,
      "grad_norm": 0.009620541706681252,
      "learning_rate": 2.3953644372488833e-06,
      "loss": 0.0017,
      "step": 129620
    },
    {
      "epoch": 2.1210832037961223,
      "grad_norm": 0.04028141126036644,
      "learning_rate": 2.394551907301031e-06,
      "loss": 0.0011,
      "step": 129630
    },
    {
      "epoch": 2.121246829747198,
      "grad_norm": 0.012738639488816261,
      "learning_rate": 2.3937394717944116e-06,
      "loss": 0.0006,
      "step": 129640
    },
    {
      "epoch": 2.121410455698274,
      "grad_norm": 0.05119181051850319,
      "learning_rate": 2.3929271307584726e-06,
      "loss": 0.0012,
      "step": 129650
    },
    {
      "epoch": 2.12157408164935,
      "grad_norm": 0.08494886010885239,
      "learning_rate": 2.392114884222657e-06,
      "loss": 0.0006,
      "step": 129660
    },
    {
      "epoch": 2.1217377076004253,
      "grad_norm": 0.033152852207422256,
      "learning_rate": 2.391302732216411e-06,
      "loss": 0.0009,
      "step": 129670
    },
    {
      "epoch": 2.1219013335515013,
      "grad_norm": 0.06638175994157791,
      "learning_rate": 2.39049067476917e-06,
      "loss": 0.0008,
      "step": 129680
    },
    {
      "epoch": 2.122064959502577,
      "grad_norm": 0.024622250348329544,
      "learning_rate": 2.389678711910371e-06,
      "loss": 0.0012,
      "step": 129690
    },
    {
      "epoch": 2.122228585453653,
      "grad_norm": 0.08107002824544907,
      "learning_rate": 2.3888668436694434e-06,
      "loss": 0.001,
      "step": 129700
    },
    {
      "epoch": 2.122392211404729,
      "grad_norm": 0.06753237545490265,
      "learning_rate": 2.3880550700758187e-06,
      "loss": 0.0006,
      "step": 129710
    },
    {
      "epoch": 2.1225558373558044,
      "grad_norm": 0.08426906168460846,
      "learning_rate": 2.3872433911589188e-06,
      "loss": 0.0011,
      "step": 129720
    },
    {
      "epoch": 2.1227194633068804,
      "grad_norm": 0.07835567742586136,
      "learning_rate": 2.3864318069481662e-06,
      "loss": 0.0008,
      "step": 129730
    },
    {
      "epoch": 2.1228830892579564,
      "grad_norm": 0.0586898997426033,
      "learning_rate": 2.3856203174729823e-06,
      "loss": 0.0008,
      "step": 129740
    },
    {
      "epoch": 2.123046715209032,
      "grad_norm": 0.0021694970782846212,
      "learning_rate": 2.384808922762775e-06,
      "loss": 0.0003,
      "step": 129750
    },
    {
      "epoch": 2.123210341160108,
      "grad_norm": 0.10634828358888626,
      "learning_rate": 2.383997622846961e-06,
      "loss": 0.0014,
      "step": 129760
    },
    {
      "epoch": 2.123373967111184,
      "grad_norm": 0.06046171858906746,
      "learning_rate": 2.3831864177549436e-06,
      "loss": 0.0005,
      "step": 129770
    },
    {
      "epoch": 2.1235375930622595,
      "grad_norm": 0.021502366289496422,
      "learning_rate": 2.3823753075161287e-06,
      "loss": 0.0012,
      "step": 129780
    },
    {
      "epoch": 2.1237012190133355,
      "grad_norm": 0.035366907715797424,
      "learning_rate": 2.38156429215992e-06,
      "loss": 0.001,
      "step": 129790
    },
    {
      "epoch": 2.1238648449644115,
      "grad_norm": 0.10414451360702515,
      "learning_rate": 2.380753371715711e-06,
      "loss": 0.0005,
      "step": 129800
    },
    {
      "epoch": 2.124028470915487,
      "grad_norm": 0.059287555515766144,
      "learning_rate": 2.379942546212898e-06,
      "loss": 0.0008,
      "step": 129810
    },
    {
      "epoch": 2.124192096866563,
      "grad_norm": 0.06971880048513412,
      "learning_rate": 2.3791318156808698e-06,
      "loss": 0.0011,
      "step": 129820
    },
    {
      "epoch": 2.124355722817639,
      "grad_norm": 0.04357230290770531,
      "learning_rate": 2.3783211801490155e-06,
      "loss": 0.0008,
      "step": 129830
    },
    {
      "epoch": 2.1245193487687146,
      "grad_norm": 0.002090972615405917,
      "learning_rate": 2.3775106396467167e-06,
      "loss": 0.0008,
      "step": 129840
    },
    {
      "epoch": 2.1246829747197906,
      "grad_norm": 0.020743712782859802,
      "learning_rate": 2.3767001942033537e-06,
      "loss": 0.0004,
      "step": 129850
    },
    {
      "epoch": 2.1248466006708666,
      "grad_norm": 0.056417014449834824,
      "learning_rate": 2.375889843848305e-06,
      "loss": 0.0006,
      "step": 129860
    },
    {
      "epoch": 2.125010226621942,
      "grad_norm": 0.04254251345992088,
      "learning_rate": 2.3750795886109403e-06,
      "loss": 0.0007,
      "step": 129870
    },
    {
      "epoch": 2.125173852573018,
      "grad_norm": 0.05097575485706329,
      "learning_rate": 2.3742694285206334e-06,
      "loss": 0.0012,
      "step": 129880
    },
    {
      "epoch": 2.1253374785240937,
      "grad_norm": 0.06107296049594879,
      "learning_rate": 2.373459363606748e-06,
      "loss": 0.0011,
      "step": 129890
    },
    {
      "epoch": 2.1255011044751697,
      "grad_norm": 0.06233549490571022,
      "learning_rate": 2.372649393898649e-06,
      "loss": 0.001,
      "step": 129900
    },
    {
      "epoch": 2.1256647304262457,
      "grad_norm": 0.07916392385959625,
      "learning_rate": 2.371839519425693e-06,
      "loss": 0.0011,
      "step": 129910
    },
    {
      "epoch": 2.1258283563773213,
      "grad_norm": 0.0730438157916069,
      "learning_rate": 2.3710297402172404e-06,
      "loss": 0.0007,
      "step": 129920
    },
    {
      "epoch": 2.1259919823283973,
      "grad_norm": 0.11279507726430893,
      "learning_rate": 2.370220056302639e-06,
      "loss": 0.0008,
      "step": 129930
    },
    {
      "epoch": 2.1261556082794733,
      "grad_norm": 0.04357591271400452,
      "learning_rate": 2.369410467711242e-06,
      "loss": 0.0007,
      "step": 129940
    },
    {
      "epoch": 2.126319234230549,
      "grad_norm": 0.09587182849645615,
      "learning_rate": 2.3686009744723924e-06,
      "loss": 0.0013,
      "step": 129950
    },
    {
      "epoch": 2.126482860181625,
      "grad_norm": 0.005223568994551897,
      "learning_rate": 2.3677915766154312e-06,
      "loss": 0.0009,
      "step": 129960
    },
    {
      "epoch": 2.126646486132701,
      "grad_norm": 0.07404561340808868,
      "learning_rate": 2.366982274169702e-06,
      "loss": 0.0011,
      "step": 129970
    },
    {
      "epoch": 2.1268101120837763,
      "grad_norm": 0.12952719628810883,
      "learning_rate": 2.3661730671645346e-06,
      "loss": 0.0012,
      "step": 129980
    },
    {
      "epoch": 2.1269737380348523,
      "grad_norm": 0.07653428614139557,
      "learning_rate": 2.3653639556292646e-06,
      "loss": 0.0004,
      "step": 129990
    },
    {
      "epoch": 2.1271373639859283,
      "grad_norm": 0.010169204324483871,
      "learning_rate": 2.3645549395932177e-06,
      "loss": 0.001,
      "step": 130000
    },
    {
      "epoch": 2.127300989937004,
      "grad_norm": 0.04137658700346947,
      "learning_rate": 2.36374601908572e-06,
      "loss": 0.0005,
      "step": 130010
    },
    {
      "epoch": 2.12746461588808,
      "grad_norm": 0.09358082711696625,
      "learning_rate": 2.3629371941360952e-06,
      "loss": 0.0004,
      "step": 130020
    },
    {
      "epoch": 2.127628241839156,
      "grad_norm": 0.07212058454751968,
      "learning_rate": 2.3621284647736563e-06,
      "loss": 0.0009,
      "step": 130030
    },
    {
      "epoch": 2.1277918677902314,
      "grad_norm": 0.07064887136220932,
      "learning_rate": 2.3613198310277253e-06,
      "loss": 0.001,
      "step": 130040
    },
    {
      "epoch": 2.1279554937413074,
      "grad_norm": 0.08241944015026093,
      "learning_rate": 2.360511292927605e-06,
      "loss": 0.0009,
      "step": 130050
    },
    {
      "epoch": 2.128119119692383,
      "grad_norm": 0.13032805919647217,
      "learning_rate": 2.3597028505026052e-06,
      "loss": 0.001,
      "step": 130060
    },
    {
      "epoch": 2.128282745643459,
      "grad_norm": 0.038160644471645355,
      "learning_rate": 2.3588945037820342e-06,
      "loss": 0.0009,
      "step": 130070
    },
    {
      "epoch": 2.128446371594535,
      "grad_norm": 0.016643980517983437,
      "learning_rate": 2.3580862527951874e-06,
      "loss": 0.0008,
      "step": 130080
    },
    {
      "epoch": 2.1286099975456105,
      "grad_norm": 0.0335405059158802,
      "learning_rate": 2.3572780975713664e-06,
      "loss": 0.0014,
      "step": 130090
    },
    {
      "epoch": 2.1287736234966865,
      "grad_norm": 0.15138103067874908,
      "learning_rate": 2.356470038139861e-06,
      "loss": 0.0011,
      "step": 130100
    },
    {
      "epoch": 2.1289372494477625,
      "grad_norm": 0.04020567983388901,
      "learning_rate": 2.355662074529965e-06,
      "loss": 0.0009,
      "step": 130110
    },
    {
      "epoch": 2.129100875398838,
      "grad_norm": 0.04866993427276611,
      "learning_rate": 2.3548542067709607e-06,
      "loss": 0.0008,
      "step": 130120
    },
    {
      "epoch": 2.129264501349914,
      "grad_norm": 0.044186897575855255,
      "learning_rate": 2.3540464348921365e-06,
      "loss": 0.0005,
      "step": 130130
    },
    {
      "epoch": 2.12942812730099,
      "grad_norm": 0.0282368715852499,
      "learning_rate": 2.3532387589227695e-06,
      "loss": 0.0006,
      "step": 130140
    },
    {
      "epoch": 2.1295917532520656,
      "grad_norm": 0.07377664744853973,
      "learning_rate": 2.3524311788921344e-06,
      "loss": 0.0011,
      "step": 130150
    },
    {
      "epoch": 2.1297553792031416,
      "grad_norm": 0.014379864558577538,
      "learning_rate": 2.3516236948295072e-06,
      "loss": 0.0011,
      "step": 130160
    },
    {
      "epoch": 2.1299190051542176,
      "grad_norm": 0.03418883681297302,
      "learning_rate": 2.3508163067641547e-06,
      "loss": 0.0004,
      "step": 130170
    },
    {
      "epoch": 2.130082631105293,
      "grad_norm": 0.06331291794776917,
      "learning_rate": 2.350009014725346e-06,
      "loss": 0.0007,
      "step": 130180
    },
    {
      "epoch": 2.130246257056369,
      "grad_norm": 0.19612273573875427,
      "learning_rate": 2.3492018187423387e-06,
      "loss": 0.0007,
      "step": 130190
    },
    {
      "epoch": 2.130409883007445,
      "grad_norm": 0.012460467405617237,
      "learning_rate": 2.348394718844397e-06,
      "loss": 0.0014,
      "step": 130200
    },
    {
      "epoch": 2.1305735089585207,
      "grad_norm": 0.06946510076522827,
      "learning_rate": 2.347587715060771e-06,
      "loss": 0.0006,
      "step": 130210
    },
    {
      "epoch": 2.1307371349095967,
      "grad_norm": 0.07056104391813278,
      "learning_rate": 2.346780807420716e-06,
      "loss": 0.0011,
      "step": 130220
    },
    {
      "epoch": 2.1309007608606727,
      "grad_norm": 0.1000114232301712,
      "learning_rate": 2.3459739959534816e-06,
      "loss": 0.0013,
      "step": 130230
    },
    {
      "epoch": 2.1310643868117483,
      "grad_norm": 0.06651310622692108,
      "learning_rate": 2.345167280688311e-06,
      "loss": 0.0009,
      "step": 130240
    },
    {
      "epoch": 2.1312280127628243,
      "grad_norm": 0.08983628451824188,
      "learning_rate": 2.3443606616544463e-06,
      "loss": 0.0006,
      "step": 130250
    },
    {
      "epoch": 2.1313916387139002,
      "grad_norm": 0.03396475315093994,
      "learning_rate": 2.343554138881123e-06,
      "loss": 0.0005,
      "step": 130260
    },
    {
      "epoch": 2.131555264664976,
      "grad_norm": 0.051050715148448944,
      "learning_rate": 2.3427477123975773e-06,
      "loss": 0.0007,
      "step": 130270
    },
    {
      "epoch": 2.131718890616052,
      "grad_norm": 0.045728445053100586,
      "learning_rate": 2.3419413822330424e-06,
      "loss": 0.0005,
      "step": 130280
    },
    {
      "epoch": 2.1318825165671274,
      "grad_norm": 0.07560687512159348,
      "learning_rate": 2.3411351484167414e-06,
      "loss": 0.0016,
      "step": 130290
    },
    {
      "epoch": 2.1320461425182033,
      "grad_norm": 0.025737281888723373,
      "learning_rate": 2.3403290109779036e-06,
      "loss": 0.0009,
      "step": 130300
    },
    {
      "epoch": 2.1322097684692793,
      "grad_norm": 0.09985017776489258,
      "learning_rate": 2.339522969945745e-06,
      "loss": 0.0006,
      "step": 130310
    },
    {
      "epoch": 2.132373394420355,
      "grad_norm": 0.039045870304107666,
      "learning_rate": 2.3387170253494858e-06,
      "loss": 0.0007,
      "step": 130320
    },
    {
      "epoch": 2.132537020371431,
      "grad_norm": 0.010081917978823185,
      "learning_rate": 2.3379111772183366e-06,
      "loss": 0.0008,
      "step": 130330
    },
    {
      "epoch": 2.132700646322507,
      "grad_norm": 0.29969823360443115,
      "learning_rate": 2.3371054255815107e-06,
      "loss": 0.0016,
      "step": 130340
    },
    {
      "epoch": 2.1328642722735824,
      "grad_norm": 0.023552823811769485,
      "learning_rate": 2.3362997704682133e-06,
      "loss": 0.0012,
      "step": 130350
    },
    {
      "epoch": 2.1330278982246584,
      "grad_norm": 0.1577552706003189,
      "learning_rate": 2.335494211907645e-06,
      "loss": 0.0011,
      "step": 130360
    },
    {
      "epoch": 2.1331915241757344,
      "grad_norm": 0.1033872663974762,
      "learning_rate": 2.33468874992901e-06,
      "loss": 0.0007,
      "step": 130370
    },
    {
      "epoch": 2.13335515012681,
      "grad_norm": 0.06565041840076447,
      "learning_rate": 2.3338833845614995e-06,
      "loss": 0.0007,
      "step": 130380
    },
    {
      "epoch": 2.133518776077886,
      "grad_norm": 0.11987122148275375,
      "learning_rate": 2.33307811583431e-06,
      "loss": 0.0015,
      "step": 130390
    },
    {
      "epoch": 2.133682402028962,
      "grad_norm": 0.06925756484270096,
      "learning_rate": 2.332272943776628e-06,
      "loss": 0.0012,
      "step": 130400
    },
    {
      "epoch": 2.1338460279800375,
      "grad_norm": 0.0964903011918068,
      "learning_rate": 2.331467868417642e-06,
      "loss": 0.0009,
      "step": 130410
    },
    {
      "epoch": 2.1340096539311135,
      "grad_norm": 0.08841458708047867,
      "learning_rate": 2.3306628897865297e-06,
      "loss": 0.0011,
      "step": 130420
    },
    {
      "epoch": 2.1341732798821895,
      "grad_norm": 0.03820227086544037,
      "learning_rate": 2.3298580079124723e-06,
      "loss": 0.0006,
      "step": 130430
    },
    {
      "epoch": 2.134336905833265,
      "grad_norm": 0.006761993281543255,
      "learning_rate": 2.329053222824648e-06,
      "loss": 0.0019,
      "step": 130440
    },
    {
      "epoch": 2.134500531784341,
      "grad_norm": 0.11034632474184036,
      "learning_rate": 2.3282485345522215e-06,
      "loss": 0.0012,
      "step": 130450
    },
    {
      "epoch": 2.1346641577354166,
      "grad_norm": 0.08911069482564926,
      "learning_rate": 2.327443943124366e-06,
      "loss": 0.0013,
      "step": 130460
    },
    {
      "epoch": 2.1348277836864926,
      "grad_norm": 0.04564070701599121,
      "learning_rate": 2.326639448570242e-06,
      "loss": 0.0005,
      "step": 130470
    },
    {
      "epoch": 2.1349914096375686,
      "grad_norm": 0.023495858535170555,
      "learning_rate": 2.325835050919015e-06,
      "loss": 0.0004,
      "step": 130480
    },
    {
      "epoch": 2.135155035588644,
      "grad_norm": 0.05203722044825554,
      "learning_rate": 2.3250307501998385e-06,
      "loss": 0.0005,
      "step": 130490
    },
    {
      "epoch": 2.13531866153972,
      "grad_norm": 0.11222314834594727,
      "learning_rate": 2.3242265464418684e-06,
      "loss": 0.0007,
      "step": 130500
    },
    {
      "epoch": 2.135482287490796,
      "grad_norm": 0.08663816750049591,
      "learning_rate": 2.3234224396742563e-06,
      "loss": 0.0007,
      "step": 130510
    },
    {
      "epoch": 2.1356459134418717,
      "grad_norm": 0.1006455048918724,
      "learning_rate": 2.322618429926146e-06,
      "loss": 0.0011,
      "step": 130520
    },
    {
      "epoch": 2.1358095393929477,
      "grad_norm": 0.23484046757221222,
      "learning_rate": 2.3218145172266846e-06,
      "loss": 0.0006,
      "step": 130530
    },
    {
      "epoch": 2.1359731653440237,
      "grad_norm": 0.08672276884317398,
      "learning_rate": 2.32101070160501e-06,
      "loss": 0.0014,
      "step": 130540
    },
    {
      "epoch": 2.1361367912950993,
      "grad_norm": 0.052922364324331284,
      "learning_rate": 2.3202069830902567e-06,
      "loss": 0.0009,
      "step": 130550
    },
    {
      "epoch": 2.1363004172461753,
      "grad_norm": 0.06208193674683571,
      "learning_rate": 2.319403361711562e-06,
      "loss": 0.001,
      "step": 130560
    },
    {
      "epoch": 2.1364640431972513,
      "grad_norm": 0.13968831300735474,
      "learning_rate": 2.3185998374980506e-06,
      "loss": 0.001,
      "step": 130570
    },
    {
      "epoch": 2.136627669148327,
      "grad_norm": 0.051334232091903687,
      "learning_rate": 2.3177964104788527e-06,
      "loss": 0.001,
      "step": 130580
    },
    {
      "epoch": 2.136791295099403,
      "grad_norm": 0.008811775594949722,
      "learning_rate": 2.316993080683086e-06,
      "loss": 0.0017,
      "step": 130590
    },
    {
      "epoch": 2.136954921050479,
      "grad_norm": 0.07008931040763855,
      "learning_rate": 2.3161898481398735e-06,
      "loss": 0.0009,
      "step": 130600
    },
    {
      "epoch": 2.1371185470015543,
      "grad_norm": 0.12066005170345306,
      "learning_rate": 2.315386712878327e-06,
      "loss": 0.0009,
      "step": 130610
    },
    {
      "epoch": 2.1372821729526303,
      "grad_norm": 0.15701177716255188,
      "learning_rate": 2.3145836749275617e-06,
      "loss": 0.0012,
      "step": 130620
    },
    {
      "epoch": 2.1374457989037063,
      "grad_norm": 0.1830628365278244,
      "learning_rate": 2.3137807343166823e-06,
      "loss": 0.0011,
      "step": 130630
    },
    {
      "epoch": 2.137609424854782,
      "grad_norm": 0.09408551454544067,
      "learning_rate": 2.312977891074797e-06,
      "loss": 0.0006,
      "step": 130640
    },
    {
      "epoch": 2.137773050805858,
      "grad_norm": 0.16284410655498505,
      "learning_rate": 2.312175145231005e-06,
      "loss": 0.0013,
      "step": 130650
    },
    {
      "epoch": 2.137936676756934,
      "grad_norm": 0.06853299587965012,
      "learning_rate": 2.311372496814402e-06,
      "loss": 0.0008,
      "step": 130660
    },
    {
      "epoch": 2.1381003027080094,
      "grad_norm": 0.03175092115998268,
      "learning_rate": 2.310569945854086e-06,
      "loss": 0.0006,
      "step": 130670
    },
    {
      "epoch": 2.1382639286590854,
      "grad_norm": 0.060112301260232925,
      "learning_rate": 2.3097674923791435e-06,
      "loss": 0.0008,
      "step": 130680
    },
    {
      "epoch": 2.138427554610161,
      "grad_norm": 0.08591391891241074,
      "learning_rate": 2.3089651364186655e-06,
      "loss": 0.0009,
      "step": 130690
    },
    {
      "epoch": 2.138591180561237,
      "grad_norm": 0.1547025591135025,
      "learning_rate": 2.3081628780017318e-06,
      "loss": 0.0007,
      "step": 130700
    },
    {
      "epoch": 2.138754806512313,
      "grad_norm": 0.034328050911426544,
      "learning_rate": 2.3073607171574243e-06,
      "loss": 0.0013,
      "step": 130710
    },
    {
      "epoch": 2.1389184324633885,
      "grad_norm": 0.2480616718530655,
      "learning_rate": 2.3065586539148205e-06,
      "loss": 0.0012,
      "step": 130720
    },
    {
      "epoch": 2.1390820584144645,
      "grad_norm": 0.1626928150653839,
      "learning_rate": 2.305756688302991e-06,
      "loss": 0.0008,
      "step": 130730
    },
    {
      "epoch": 2.1392456843655405,
      "grad_norm": 0.06996849179267883,
      "learning_rate": 2.304954820351009e-06,
      "loss": 0.0017,
      "step": 130740
    },
    {
      "epoch": 2.139409310316616,
      "grad_norm": 0.05595912039279938,
      "learning_rate": 2.304153050087934e-06,
      "loss": 0.0009,
      "step": 130750
    },
    {
      "epoch": 2.139572936267692,
      "grad_norm": 0.08856242895126343,
      "learning_rate": 2.3033513775428333e-06,
      "loss": 0.0006,
      "step": 130760
    },
    {
      "epoch": 2.139736562218768,
      "grad_norm": 0.04912002757191658,
      "learning_rate": 2.302549802744762e-06,
      "loss": 0.0006,
      "step": 130770
    },
    {
      "epoch": 2.1399001881698436,
      "grad_norm": 0.046053074300289154,
      "learning_rate": 2.301748325722778e-06,
      "loss": 0.0005,
      "step": 130780
    },
    {
      "epoch": 2.1400638141209196,
      "grad_norm": 0.025236330926418304,
      "learning_rate": 2.3009469465059337e-06,
      "loss": 0.0007,
      "step": 130790
    },
    {
      "epoch": 2.1402274400719956,
      "grad_norm": 0.05236167460680008,
      "learning_rate": 2.3001456651232735e-06,
      "loss": 0.001,
      "step": 130800
    },
    {
      "epoch": 2.140391066023071,
      "grad_norm": 0.08195017278194427,
      "learning_rate": 2.2993444816038456e-06,
      "loss": 0.0005,
      "step": 130810
    },
    {
      "epoch": 2.140554691974147,
      "grad_norm": 0.06903974711894989,
      "learning_rate": 2.298543395976688e-06,
      "loss": 0.0007,
      "step": 130820
    },
    {
      "epoch": 2.1407183179252227,
      "grad_norm": 0.03243299946188927,
      "learning_rate": 2.297742408270841e-06,
      "loss": 0.0006,
      "step": 130830
    },
    {
      "epoch": 2.1408819438762987,
      "grad_norm": 0.008946661837399006,
      "learning_rate": 2.296941518515337e-06,
      "loss": 0.0004,
      "step": 130840
    },
    {
      "epoch": 2.1410455698273747,
      "grad_norm": 0.05911422148346901,
      "learning_rate": 2.296140726739205e-06,
      "loss": 0.0017,
      "step": 130850
    },
    {
      "epoch": 2.1412091957784503,
      "grad_norm": 0.030707096680998802,
      "learning_rate": 2.295340032971474e-06,
      "loss": 0.0008,
      "step": 130860
    },
    {
      "epoch": 2.1413728217295263,
      "grad_norm": 0.038526423275470734,
      "learning_rate": 2.2945394372411656e-06,
      "loss": 0.0008,
      "step": 130870
    },
    {
      "epoch": 2.1415364476806023,
      "grad_norm": 0.03181355819106102,
      "learning_rate": 2.2937389395773013e-06,
      "loss": 0.0005,
      "step": 130880
    },
    {
      "epoch": 2.141700073631678,
      "grad_norm": 0.11275369673967361,
      "learning_rate": 2.2929385400088944e-06,
      "loss": 0.0004,
      "step": 130890
    },
    {
      "epoch": 2.141863699582754,
      "grad_norm": 0.014903356321156025,
      "learning_rate": 2.292138238564961e-06,
      "loss": 0.0015,
      "step": 130900
    },
    {
      "epoch": 2.14202732553383,
      "grad_norm": 0.0010434207506477833,
      "learning_rate": 2.291338035274507e-06,
      "loss": 0.0003,
      "step": 130910
    },
    {
      "epoch": 2.1421909514849053,
      "grad_norm": 0.1800602823495865,
      "learning_rate": 2.2905379301665403e-06,
      "loss": 0.0008,
      "step": 130920
    },
    {
      "epoch": 2.1423545774359813,
      "grad_norm": 0.054574571549892426,
      "learning_rate": 2.2897379232700603e-06,
      "loss": 0.0013,
      "step": 130930
    },
    {
      "epoch": 2.1425182033870573,
      "grad_norm": 0.045944079756736755,
      "learning_rate": 2.288938014614068e-06,
      "loss": 0.0005,
      "step": 130940
    },
    {
      "epoch": 2.142681829338133,
      "grad_norm": 0.035358306020498276,
      "learning_rate": 2.288138204227557e-06,
      "loss": 0.0008,
      "step": 130950
    },
    {
      "epoch": 2.142845455289209,
      "grad_norm": 0.097017303109169,
      "learning_rate": 2.287338492139516e-06,
      "loss": 0.001,
      "step": 130960
    },
    {
      "epoch": 2.143009081240285,
      "grad_norm": 0.06005150079727173,
      "learning_rate": 2.2865388783789374e-06,
      "loss": 0.0006,
      "step": 130970
    },
    {
      "epoch": 2.1431727071913604,
      "grad_norm": 0.22261619567871094,
      "learning_rate": 2.2857393629748005e-06,
      "loss": 0.001,
      "step": 130980
    },
    {
      "epoch": 2.1433363331424364,
      "grad_norm": 0.1900072991847992,
      "learning_rate": 2.2849399459560883e-06,
      "loss": 0.0012,
      "step": 130990
    },
    {
      "epoch": 2.1434999590935124,
      "grad_norm": 0.04064079001545906,
      "learning_rate": 2.2841406273517795e-06,
      "loss": 0.0012,
      "step": 131000
    },
    {
      "epoch": 2.143663585044588,
      "grad_norm": 0.22620823979377747,
      "learning_rate": 2.283341407190844e-06,
      "loss": 0.0014,
      "step": 131010
    },
    {
      "epoch": 2.143827210995664,
      "grad_norm": 0.020436285063624382,
      "learning_rate": 2.282542285502255e-06,
      "loss": 0.0017,
      "step": 131020
    },
    {
      "epoch": 2.14399083694674,
      "grad_norm": 0.15270817279815674,
      "learning_rate": 2.281743262314975e-06,
      "loss": 0.0008,
      "step": 131030
    },
    {
      "epoch": 2.1441544628978155,
      "grad_norm": 0.24394603073596954,
      "learning_rate": 2.280944337657972e-06,
      "loss": 0.001,
      "step": 131040
    },
    {
      "epoch": 2.1443180888488915,
      "grad_norm": 0.03933184593915939,
      "learning_rate": 2.280145511560198e-06,
      "loss": 0.0006,
      "step": 131050
    },
    {
      "epoch": 2.144481714799967,
      "grad_norm": 0.02958499826490879,
      "learning_rate": 2.279346784050612e-06,
      "loss": 0.0006,
      "step": 131060
    },
    {
      "epoch": 2.144645340751043,
      "grad_norm": 0.05615805834531784,
      "learning_rate": 2.2785481551581683e-06,
      "loss": 0.0007,
      "step": 131070
    },
    {
      "epoch": 2.144808966702119,
      "grad_norm": 0.25226667523384094,
      "learning_rate": 2.2777496249118113e-06,
      "loss": 0.0007,
      "step": 131080
    },
    {
      "epoch": 2.1449725926531946,
      "grad_norm": 0.06918755918741226,
      "learning_rate": 2.2769511933404897e-06,
      "loss": 0.0008,
      "step": 131090
    },
    {
      "epoch": 2.1451362186042706,
      "grad_norm": 0.09297676384449005,
      "learning_rate": 2.2761528604731408e-06,
      "loss": 0.0015,
      "step": 131100
    },
    {
      "epoch": 2.1452998445553466,
      "grad_norm": 0.0052992356941103935,
      "learning_rate": 2.2753546263387054e-06,
      "loss": 0.0012,
      "step": 131110
    },
    {
      "epoch": 2.145463470506422,
      "grad_norm": 0.14361236989498138,
      "learning_rate": 2.274556490966115e-06,
      "loss": 0.0008,
      "step": 131120
    },
    {
      "epoch": 2.145627096457498,
      "grad_norm": 0.03398771584033966,
      "learning_rate": 2.2737584543843035e-06,
      "loss": 0.0007,
      "step": 131130
    },
    {
      "epoch": 2.145790722408574,
      "grad_norm": 0.02347593382000923,
      "learning_rate": 2.272960516622195e-06,
      "loss": 0.0013,
      "step": 131140
    },
    {
      "epoch": 2.1459543483596497,
      "grad_norm": 0.005705694667994976,
      "learning_rate": 2.272162677708712e-06,
      "loss": 0.0011,
      "step": 131150
    },
    {
      "epoch": 2.1461179743107257,
      "grad_norm": 0.07995748519897461,
      "learning_rate": 2.271364937672777e-06,
      "loss": 0.0007,
      "step": 131160
    },
    {
      "epoch": 2.1462816002618017,
      "grad_norm": 0.030589068308472633,
      "learning_rate": 2.270567296543304e-06,
      "loss": 0.0007,
      "step": 131170
    },
    {
      "epoch": 2.1464452262128773,
      "grad_norm": 0.06749261170625687,
      "learning_rate": 2.2697697543492076e-06,
      "loss": 0.0015,
      "step": 131180
    },
    {
      "epoch": 2.1466088521639533,
      "grad_norm": 0.0033417295198887587,
      "learning_rate": 2.2689723111193938e-06,
      "loss": 0.0009,
      "step": 131190
    },
    {
      "epoch": 2.1467724781150292,
      "grad_norm": 0.07669030874967575,
      "learning_rate": 2.2681749668827724e-06,
      "loss": 0.0012,
      "step": 131200
    },
    {
      "epoch": 2.146936104066105,
      "grad_norm": 0.0583362951874733,
      "learning_rate": 2.26737772166824e-06,
      "loss": 0.0007,
      "step": 131210
    },
    {
      "epoch": 2.147099730017181,
      "grad_norm": 0.06939921528100967,
      "learning_rate": 2.266580575504698e-06,
      "loss": 0.0006,
      "step": 131220
    },
    {
      "epoch": 2.1472633559682563,
      "grad_norm": 0.006901615299284458,
      "learning_rate": 2.2657835284210427e-06,
      "loss": 0.0005,
      "step": 131230
    },
    {
      "epoch": 2.1474269819193323,
      "grad_norm": 0.06196759268641472,
      "learning_rate": 2.2649865804461623e-06,
      "loss": 0.0005,
      "step": 131240
    },
    {
      "epoch": 2.1475906078704083,
      "grad_norm": 0.028714854270219803,
      "learning_rate": 2.2641897316089455e-06,
      "loss": 0.001,
      "step": 131250
    },
    {
      "epoch": 2.147754233821484,
      "grad_norm": 0.07110073417425156,
      "learning_rate": 2.2633929819382733e-06,
      "loss": 0.0007,
      "step": 131260
    },
    {
      "epoch": 2.14791785977256,
      "grad_norm": 0.05138629302382469,
      "learning_rate": 2.2625963314630285e-06,
      "loss": 0.0006,
      "step": 131270
    },
    {
      "epoch": 2.148081485723636,
      "grad_norm": 0.1584242582321167,
      "learning_rate": 2.2617997802120896e-06,
      "loss": 0.0013,
      "step": 131280
    },
    {
      "epoch": 2.1482451116747114,
      "grad_norm": 0.13005654513835907,
      "learning_rate": 2.261003328214325e-06,
      "loss": 0.0016,
      "step": 131290
    },
    {
      "epoch": 2.1484087376257874,
      "grad_norm": 0.05873752385377884,
      "learning_rate": 2.260206975498609e-06,
      "loss": 0.0007,
      "step": 131300
    },
    {
      "epoch": 2.1485723635768634,
      "grad_norm": 0.34052443504333496,
      "learning_rate": 2.2594107220938035e-06,
      "loss": 0.001,
      "step": 131310
    },
    {
      "epoch": 2.148735989527939,
      "grad_norm": 0.07647483795881271,
      "learning_rate": 2.258614568028774e-06,
      "loss": 0.0013,
      "step": 131320
    },
    {
      "epoch": 2.148899615479015,
      "grad_norm": 0.0016928694676607847,
      "learning_rate": 2.257818513332376e-06,
      "loss": 0.0006,
      "step": 131330
    },
    {
      "epoch": 2.149063241430091,
      "grad_norm": 0.15298034250736237,
      "learning_rate": 2.257022558033468e-06,
      "loss": 0.0006,
      "step": 131340
    },
    {
      "epoch": 2.1492268673811665,
      "grad_norm": 0.013507881201803684,
      "learning_rate": 2.2562267021609e-06,
      "loss": 0.0006,
      "step": 131350
    },
    {
      "epoch": 2.1493904933322425,
      "grad_norm": 0.06394945085048676,
      "learning_rate": 2.2554309457435174e-06,
      "loss": 0.0007,
      "step": 131360
    },
    {
      "epoch": 2.1495541192833185,
      "grad_norm": 0.005180387292057276,
      "learning_rate": 2.2546352888101685e-06,
      "loss": 0.0009,
      "step": 131370
    },
    {
      "epoch": 2.149717745234394,
      "grad_norm": 0.04113948345184326,
      "learning_rate": 2.2538397313896904e-06,
      "loss": 0.0008,
      "step": 131380
    },
    {
      "epoch": 2.14988137118547,
      "grad_norm": 0.013941937126219273,
      "learning_rate": 2.2530442735109238e-06,
      "loss": 0.0006,
      "step": 131390
    },
    {
      "epoch": 2.150044997136546,
      "grad_norm": 0.06706994026899338,
      "learning_rate": 2.252248915202698e-06,
      "loss": 0.0006,
      "step": 131400
    },
    {
      "epoch": 2.1502086230876216,
      "grad_norm": 0.02786465361714363,
      "learning_rate": 2.2514536564938464e-06,
      "loss": 0.001,
      "step": 131410
    },
    {
      "epoch": 2.1503722490386976,
      "grad_norm": 0.060484081506729126,
      "learning_rate": 2.250658497413193e-06,
      "loss": 0.0006,
      "step": 131420
    },
    {
      "epoch": 2.1505358749897736,
      "grad_norm": 0.08929912000894547,
      "learning_rate": 2.2498634379895607e-06,
      "loss": 0.0008,
      "step": 131430
    },
    {
      "epoch": 2.150699500940849,
      "grad_norm": 0.004630949813872576,
      "learning_rate": 2.2490684782517704e-06,
      "loss": 0.0009,
      "step": 131440
    },
    {
      "epoch": 2.150863126891925,
      "grad_norm": 0.07582361251115799,
      "learning_rate": 2.248273618228636e-06,
      "loss": 0.0014,
      "step": 131450
    },
    {
      "epoch": 2.1510267528430007,
      "grad_norm": 0.15050379931926727,
      "learning_rate": 2.2474788579489703e-06,
      "loss": 0.001,
      "step": 131460
    },
    {
      "epoch": 2.1511903787940767,
      "grad_norm": 0.1495954990386963,
      "learning_rate": 2.2466841974415784e-06,
      "loss": 0.0008,
      "step": 131470
    },
    {
      "epoch": 2.1513540047451527,
      "grad_norm": 0.0027888508047908545,
      "learning_rate": 2.245889636735266e-06,
      "loss": 0.0006,
      "step": 131480
    },
    {
      "epoch": 2.1515176306962283,
      "grad_norm": 0.09791406989097595,
      "learning_rate": 2.245095175858838e-06,
      "loss": 0.0013,
      "step": 131490
    },
    {
      "epoch": 2.1516812566473043,
      "grad_norm": 0.019554410129785538,
      "learning_rate": 2.244300814841086e-06,
      "loss": 0.0011,
      "step": 131500
    },
    {
      "epoch": 2.1518448825983802,
      "grad_norm": 0.06986770033836365,
      "learning_rate": 2.2435065537108086e-06,
      "loss": 0.0009,
      "step": 131510
    },
    {
      "epoch": 2.152008508549456,
      "grad_norm": 0.004764092620462179,
      "learning_rate": 2.2427123924967914e-06,
      "loss": 0.0012,
      "step": 131520
    },
    {
      "epoch": 2.152172134500532,
      "grad_norm": 0.00810950342565775,
      "learning_rate": 2.241918331227825e-06,
      "loss": 0.0009,
      "step": 131530
    },
    {
      "epoch": 2.152335760451608,
      "grad_norm": 0.03504367172718048,
      "learning_rate": 2.2411243699326877e-06,
      "loss": 0.0009,
      "step": 131540
    },
    {
      "epoch": 2.1524993864026833,
      "grad_norm": 0.029177607968449593,
      "learning_rate": 2.240330508640164e-06,
      "loss": 0.0007,
      "step": 131550
    },
    {
      "epoch": 2.1526630123537593,
      "grad_norm": 0.1350104957818985,
      "learning_rate": 2.2395367473790264e-06,
      "loss": 0.0007,
      "step": 131560
    },
    {
      "epoch": 2.1528266383048353,
      "grad_norm": 0.061625488102436066,
      "learning_rate": 2.238743086178045e-06,
      "loss": 0.0013,
      "step": 131570
    },
    {
      "epoch": 2.152990264255911,
      "grad_norm": 0.07101751118898392,
      "learning_rate": 2.2379495250659925e-06,
      "loss": 0.0014,
      "step": 131580
    },
    {
      "epoch": 2.153153890206987,
      "grad_norm": 0.2722167670726776,
      "learning_rate": 2.2371560640716296e-06,
      "loss": 0.0017,
      "step": 131590
    },
    {
      "epoch": 2.1533175161580624,
      "grad_norm": 0.04820204898715019,
      "learning_rate": 2.2363627032237203e-06,
      "loss": 0.0008,
      "step": 131600
    },
    {
      "epoch": 2.1534811421091384,
      "grad_norm": 0.0014235300477594137,
      "learning_rate": 2.2355694425510192e-06,
      "loss": 0.0011,
      "step": 131610
    },
    {
      "epoch": 2.1536447680602144,
      "grad_norm": 0.20849184691905975,
      "learning_rate": 2.2347762820822832e-06,
      "loss": 0.0009,
      "step": 131620
    },
    {
      "epoch": 2.15380839401129,
      "grad_norm": 0.06146630272269249,
      "learning_rate": 2.2339832218462597e-06,
      "loss": 0.0008,
      "step": 131630
    },
    {
      "epoch": 2.153972019962366,
      "grad_norm": 0.227002814412117,
      "learning_rate": 2.2331902618716962e-06,
      "loss": 0.0006,
      "step": 131640
    },
    {
      "epoch": 2.154135645913442,
      "grad_norm": 0.004443961661309004,
      "learning_rate": 2.2323974021873397e-06,
      "loss": 0.0014,
      "step": 131650
    },
    {
      "epoch": 2.1542992718645175,
      "grad_norm": 0.06790367513895035,
      "learning_rate": 2.2316046428219214e-06,
      "loss": 0.001,
      "step": 131660
    },
    {
      "epoch": 2.1544628978155935,
      "grad_norm": 0.12898868322372437,
      "learning_rate": 2.230811983804183e-06,
      "loss": 0.0006,
      "step": 131670
    },
    {
      "epoch": 2.1546265237666695,
      "grad_norm": 0.024522315710783005,
      "learning_rate": 2.2300194251628534e-06,
      "loss": 0.0004,
      "step": 131680
    },
    {
      "epoch": 2.154790149717745,
      "grad_norm": 0.01787228137254715,
      "learning_rate": 2.2292269669266637e-06,
      "loss": 0.001,
      "step": 131690
    },
    {
      "epoch": 2.154953775668821,
      "grad_norm": 0.005904348101466894,
      "learning_rate": 2.2284346091243353e-06,
      "loss": 0.0004,
      "step": 131700
    },
    {
      "epoch": 2.155117401619897,
      "grad_norm": 0.058560553938150406,
      "learning_rate": 2.2276423517845906e-06,
      "loss": 0.0007,
      "step": 131710
    },
    {
      "epoch": 2.1552810275709726,
      "grad_norm": 0.03541845828294754,
      "learning_rate": 2.22685019493615e-06,
      "loss": 0.0007,
      "step": 131720
    },
    {
      "epoch": 2.1554446535220486,
      "grad_norm": 0.35124671459198,
      "learning_rate": 2.226058138607722e-06,
      "loss": 0.0011,
      "step": 131730
    },
    {
      "epoch": 2.1556082794731246,
      "grad_norm": 0.05504801869392395,
      "learning_rate": 2.225266182828022e-06,
      "loss": 0.0009,
      "step": 131740
    },
    {
      "epoch": 2.1557719054242,
      "grad_norm": 0.05280764773488045,
      "learning_rate": 2.2244743276257543e-06,
      "loss": 0.0012,
      "step": 131750
    },
    {
      "epoch": 2.155935531375276,
      "grad_norm": 0.22598738968372345,
      "learning_rate": 2.2236825730296192e-06,
      "loss": 0.001,
      "step": 131760
    },
    {
      "epoch": 2.156099157326352,
      "grad_norm": 0.06541605293750763,
      "learning_rate": 2.2228909190683197e-06,
      "loss": 0.0006,
      "step": 131770
    },
    {
      "epoch": 2.1562627832774277,
      "grad_norm": 0.0897442102432251,
      "learning_rate": 2.222099365770548e-06,
      "loss": 0.0006,
      "step": 131780
    },
    {
      "epoch": 2.1564264092285037,
      "grad_norm": 0.08767735958099365,
      "learning_rate": 2.2213079131649995e-06,
      "loss": 0.001,
      "step": 131790
    },
    {
      "epoch": 2.1565900351795797,
      "grad_norm": 0.06079059839248657,
      "learning_rate": 2.220516561280359e-06,
      "loss": 0.0006,
      "step": 131800
    },
    {
      "epoch": 2.1567536611306553,
      "grad_norm": 0.03120972216129303,
      "learning_rate": 2.2197253101453155e-06,
      "loss": 0.0006,
      "step": 131810
    },
    {
      "epoch": 2.1569172870817312,
      "grad_norm": 0.09705932438373566,
      "learning_rate": 2.2189341597885445e-06,
      "loss": 0.0008,
      "step": 131820
    },
    {
      "epoch": 2.157080913032807,
      "grad_norm": 0.09551506489515305,
      "learning_rate": 2.218143110238728e-06,
      "loss": 0.0009,
      "step": 131830
    },
    {
      "epoch": 2.157244538983883,
      "grad_norm": 0.053732823580503464,
      "learning_rate": 2.2173521615245363e-06,
      "loss": 0.0008,
      "step": 131840
    },
    {
      "epoch": 2.157408164934959,
      "grad_norm": 0.13685588538646698,
      "learning_rate": 2.216561313674642e-06,
      "loss": 0.0006,
      "step": 131850
    },
    {
      "epoch": 2.1575717908860343,
      "grad_norm": 0.19548673927783966,
      "learning_rate": 2.215770566717711e-06,
      "loss": 0.0011,
      "step": 131860
    },
    {
      "epoch": 2.1577354168371103,
      "grad_norm": 0.014646853320300579,
      "learning_rate": 2.2149799206824036e-06,
      "loss": 0.0003,
      "step": 131870
    },
    {
      "epoch": 2.1578990427881863,
      "grad_norm": 0.1305849254131317,
      "learning_rate": 2.2141893755973814e-06,
      "loss": 0.0007,
      "step": 131880
    },
    {
      "epoch": 2.158062668739262,
      "grad_norm": 0.007354858331382275,
      "learning_rate": 2.213398931491297e-06,
      "loss": 0.0008,
      "step": 131890
    },
    {
      "epoch": 2.158226294690338,
      "grad_norm": 0.046017296612262726,
      "learning_rate": 2.2126085883928065e-06,
      "loss": 0.001,
      "step": 131900
    },
    {
      "epoch": 2.158389920641414,
      "grad_norm": 0.0065455688163638115,
      "learning_rate": 2.2118183463305533e-06,
      "loss": 0.0004,
      "step": 131910
    },
    {
      "epoch": 2.1585535465924894,
      "grad_norm": 0.19899258017539978,
      "learning_rate": 2.211028205333184e-06,
      "loss": 0.0008,
      "step": 131920
    },
    {
      "epoch": 2.1587171725435654,
      "grad_norm": 0.08297527581453323,
      "learning_rate": 2.21023816542934e-06,
      "loss": 0.0005,
      "step": 131930
    },
    {
      "epoch": 2.1588807984946414,
      "grad_norm": 0.290002703666687,
      "learning_rate": 2.209448226647657e-06,
      "loss": 0.0012,
      "step": 131940
    },
    {
      "epoch": 2.159044424445717,
      "grad_norm": 0.008366294205188751,
      "learning_rate": 2.208658389016772e-06,
      "loss": 0.0011,
      "step": 131950
    },
    {
      "epoch": 2.159208050396793,
      "grad_norm": 0.13878047466278076,
      "learning_rate": 2.2078686525653075e-06,
      "loss": 0.0012,
      "step": 131960
    },
    {
      "epoch": 2.159371676347869,
      "grad_norm": 0.48104971647262573,
      "learning_rate": 2.207079017321896e-06,
      "loss": 0.0017,
      "step": 131970
    },
    {
      "epoch": 2.1595353022989445,
      "grad_norm": 0.045549146831035614,
      "learning_rate": 2.2062894833151547e-06,
      "loss": 0.0023,
      "step": 131980
    },
    {
      "epoch": 2.1596989282500205,
      "grad_norm": 0.31518876552581787,
      "learning_rate": 2.2055000505737063e-06,
      "loss": 0.0008,
      "step": 131990
    },
    {
      "epoch": 2.159862554201096,
      "grad_norm": 0.08737169206142426,
      "learning_rate": 2.204710719126166e-06,
      "loss": 0.0007,
      "step": 132000
    },
    {
      "epoch": 2.160026180152172,
      "grad_norm": 0.05547555536031723,
      "learning_rate": 2.2039214890011423e-06,
      "loss": 0.0018,
      "step": 132010
    },
    {
      "epoch": 2.160189806103248,
      "grad_norm": 0.07971277832984924,
      "learning_rate": 2.203132360227246e-06,
      "loss": 0.0006,
      "step": 132020
    },
    {
      "epoch": 2.1603534320543236,
      "grad_norm": 0.06319615244865417,
      "learning_rate": 2.2023433328330774e-06,
      "loss": 0.001,
      "step": 132030
    },
    {
      "epoch": 2.1605170580053996,
      "grad_norm": 0.09186331927776337,
      "learning_rate": 2.201554406847241e-06,
      "loss": 0.0011,
      "step": 132040
    },
    {
      "epoch": 2.1606806839564756,
      "grad_norm": 0.15043455362319946,
      "learning_rate": 2.200765582298332e-06,
      "loss": 0.0008,
      "step": 132050
    },
    {
      "epoch": 2.160844309907551,
      "grad_norm": 0.002870648168027401,
      "learning_rate": 2.1999768592149407e-06,
      "loss": 0.0008,
      "step": 132060
    },
    {
      "epoch": 2.161007935858627,
      "grad_norm": 0.13395456969738007,
      "learning_rate": 2.19918823762566e-06,
      "loss": 0.0007,
      "step": 132070
    },
    {
      "epoch": 2.161171561809703,
      "grad_norm": 0.07483165711164474,
      "learning_rate": 2.198399717559073e-06,
      "loss": 0.0013,
      "step": 132080
    },
    {
      "epoch": 2.1613351877607787,
      "grad_norm": 0.08251338452100754,
      "learning_rate": 2.1976112990437644e-06,
      "loss": 0.0008,
      "step": 132090
    },
    {
      "epoch": 2.1614988137118547,
      "grad_norm": 0.14885979890823364,
      "learning_rate": 2.196822982108308e-06,
      "loss": 0.0012,
      "step": 132100
    },
    {
      "epoch": 2.1616624396629307,
      "grad_norm": 0.008364741690456867,
      "learning_rate": 2.196034766781284e-06,
      "loss": 0.0007,
      "step": 132110
    },
    {
      "epoch": 2.1618260656140063,
      "grad_norm": 0.03293644264340401,
      "learning_rate": 2.1952466530912592e-06,
      "loss": 0.0008,
      "step": 132120
    },
    {
      "epoch": 2.1619896915650822,
      "grad_norm": 0.04341413080692291,
      "learning_rate": 2.1944586410668035e-06,
      "loss": 0.0006,
      "step": 132130
    },
    {
      "epoch": 2.1621533175161582,
      "grad_norm": 0.16618528962135315,
      "learning_rate": 2.1936707307364767e-06,
      "loss": 0.0012,
      "step": 132140
    },
    {
      "epoch": 2.162316943467234,
      "grad_norm": 0.11989282071590424,
      "learning_rate": 2.1928829221288438e-06,
      "loss": 0.001,
      "step": 132150
    },
    {
      "epoch": 2.16248056941831,
      "grad_norm": 0.04387267306447029,
      "learning_rate": 2.1920952152724577e-06,
      "loss": 0.001,
      "step": 132160
    },
    {
      "epoch": 2.162644195369386,
      "grad_norm": 0.04715392366051674,
      "learning_rate": 2.1913076101958696e-06,
      "loss": 0.0005,
      "step": 132170
    },
    {
      "epoch": 2.1628078213204613,
      "grad_norm": 0.10019509494304657,
      "learning_rate": 2.1905201069276323e-06,
      "loss": 0.0012,
      "step": 132180
    },
    {
      "epoch": 2.1629714472715373,
      "grad_norm": 0.010975530371069908,
      "learning_rate": 2.189732705496287e-06,
      "loss": 0.0007,
      "step": 132190
    },
    {
      "epoch": 2.1631350732226133,
      "grad_norm": 0.060079388320446014,
      "learning_rate": 2.188945405930376e-06,
      "loss": 0.0015,
      "step": 132200
    },
    {
      "epoch": 2.163298699173689,
      "grad_norm": 0.070175401866436,
      "learning_rate": 2.188158208258441e-06,
      "loss": 0.0007,
      "step": 132210
    },
    {
      "epoch": 2.163462325124765,
      "grad_norm": 0.0512833297252655,
      "learning_rate": 2.18737111250901e-06,
      "loss": 0.0007,
      "step": 132220
    },
    {
      "epoch": 2.1636259510758404,
      "grad_norm": 0.04308553412556648,
      "learning_rate": 2.186584118710618e-06,
      "loss": 0.0008,
      "step": 132230
    },
    {
      "epoch": 2.1637895770269164,
      "grad_norm": 0.07017209380865097,
      "learning_rate": 2.1857972268917883e-06,
      "loss": 0.0015,
      "step": 132240
    },
    {
      "epoch": 2.1639532029779924,
      "grad_norm": 0.22650322318077087,
      "learning_rate": 2.1850104370810487e-06,
      "loss": 0.0014,
      "step": 132250
    },
    {
      "epoch": 2.164116828929068,
      "grad_norm": 0.07379472255706787,
      "learning_rate": 2.1842237493069116e-06,
      "loss": 0.0008,
      "step": 132260
    },
    {
      "epoch": 2.164280454880144,
      "grad_norm": 0.12927652895450592,
      "learning_rate": 2.183437163597897e-06,
      "loss": 0.001,
      "step": 132270
    },
    {
      "epoch": 2.16444408083122,
      "grad_norm": 0.11865617334842682,
      "learning_rate": 2.1826506799825165e-06,
      "loss": 0.0008,
      "step": 132280
    },
    {
      "epoch": 2.1646077067822955,
      "grad_norm": 0.0906696766614914,
      "learning_rate": 2.1818642984892756e-06,
      "loss": 0.0007,
      "step": 132290
    },
    {
      "epoch": 2.1647713327333715,
      "grad_norm": 0.007964309304952621,
      "learning_rate": 2.181078019146683e-06,
      "loss": 0.0005,
      "step": 132300
    },
    {
      "epoch": 2.1649349586844475,
      "grad_norm": 0.08176179975271225,
      "learning_rate": 2.1802918419832355e-06,
      "loss": 0.001,
      "step": 132310
    },
    {
      "epoch": 2.165098584635523,
      "grad_norm": 0.1684895157814026,
      "learning_rate": 2.179505767027433e-06,
      "loss": 0.0006,
      "step": 132320
    },
    {
      "epoch": 2.165262210586599,
      "grad_norm": 0.050997935235500336,
      "learning_rate": 2.178719794307766e-06,
      "loss": 0.001,
      "step": 132330
    },
    {
      "epoch": 2.165425836537675,
      "grad_norm": 0.005915137007832527,
      "learning_rate": 2.1779339238527274e-06,
      "loss": 0.0007,
      "step": 132340
    },
    {
      "epoch": 2.1655894624887506,
      "grad_norm": 0.04680750519037247,
      "learning_rate": 2.177148155690802e-06,
      "loss": 0.0009,
      "step": 132350
    },
    {
      "epoch": 2.1657530884398266,
      "grad_norm": 0.08636245131492615,
      "learning_rate": 2.1763624898504687e-06,
      "loss": 0.0011,
      "step": 132360
    },
    {
      "epoch": 2.165916714390902,
      "grad_norm": 0.03715136647224426,
      "learning_rate": 2.1755769263602113e-06,
      "loss": 0.0007,
      "step": 132370
    },
    {
      "epoch": 2.166080340341978,
      "grad_norm": 0.021826880052685738,
      "learning_rate": 2.1747914652484997e-06,
      "loss": 0.0006,
      "step": 132380
    },
    {
      "epoch": 2.166243966293054,
      "grad_norm": 0.030047036707401276,
      "learning_rate": 2.174006106543809e-06,
      "loss": 0.0016,
      "step": 132390
    },
    {
      "epoch": 2.1664075922441297,
      "grad_norm": 0.19203877449035645,
      "learning_rate": 2.1732208502746033e-06,
      "loss": 0.0007,
      "step": 132400
    },
    {
      "epoch": 2.1665712181952057,
      "grad_norm": 0.08410325646400452,
      "learning_rate": 2.1724356964693495e-06,
      "loss": 0.0008,
      "step": 132410
    },
    {
      "epoch": 2.1667348441462817,
      "grad_norm": 0.04899807274341583,
      "learning_rate": 2.171650645156504e-06,
      "loss": 0.001,
      "step": 132420
    },
    {
      "epoch": 2.1668984700973573,
      "grad_norm": 0.026054365560412407,
      "learning_rate": 2.1708656963645248e-06,
      "loss": 0.0008,
      "step": 132430
    },
    {
      "epoch": 2.1670620960484333,
      "grad_norm": 0.03904905542731285,
      "learning_rate": 2.170080850121866e-06,
      "loss": 0.0005,
      "step": 132440
    },
    {
      "epoch": 2.1672257219995092,
      "grad_norm": 0.05899961292743683,
      "learning_rate": 2.1692961064569757e-06,
      "loss": 0.0004,
      "step": 132450
    },
    {
      "epoch": 2.167389347950585,
      "grad_norm": 0.13511371612548828,
      "learning_rate": 2.1685114653982974e-06,
      "loss": 0.001,
      "step": 132460
    },
    {
      "epoch": 2.167552973901661,
      "grad_norm": 0.034165531396865845,
      "learning_rate": 2.1677269269742718e-06,
      "loss": 0.0011,
      "step": 132470
    },
    {
      "epoch": 2.167716599852737,
      "grad_norm": 0.2054072618484497,
      "learning_rate": 2.1669424912133376e-06,
      "loss": 0.0008,
      "step": 132480
    },
    {
      "epoch": 2.1678802258038123,
      "grad_norm": 0.002577691338956356,
      "learning_rate": 2.166158158143931e-06,
      "loss": 0.0006,
      "step": 132490
    },
    {
      "epoch": 2.1680438517548883,
      "grad_norm": 0.09754899144172668,
      "learning_rate": 2.1653739277944784e-06,
      "loss": 0.0006,
      "step": 132500
    },
    {
      "epoch": 2.1682074777059643,
      "grad_norm": 0.038507331162691116,
      "learning_rate": 2.1645898001934096e-06,
      "loss": 0.0007,
      "step": 132510
    },
    {
      "epoch": 2.16837110365704,
      "grad_norm": 0.08859949558973312,
      "learning_rate": 2.1638057753691444e-06,
      "loss": 0.001,
      "step": 132520
    },
    {
      "epoch": 2.168534729608116,
      "grad_norm": 0.12233471870422363,
      "learning_rate": 2.1630218533501052e-06,
      "loss": 0.0005,
      "step": 132530
    },
    {
      "epoch": 2.168698355559192,
      "grad_norm": 0.1635713428258896,
      "learning_rate": 2.162238034164703e-06,
      "loss": 0.0012,
      "step": 132540
    },
    {
      "epoch": 2.1688619815102674,
      "grad_norm": 0.0937272235751152,
      "learning_rate": 2.1614543178413533e-06,
      "loss": 0.0011,
      "step": 132550
    },
    {
      "epoch": 2.1690256074613434,
      "grad_norm": 0.1224890798330307,
      "learning_rate": 2.160670704408463e-06,
      "loss": 0.0006,
      "step": 132560
    },
    {
      "epoch": 2.1691892334124194,
      "grad_norm": 0.023271478712558746,
      "learning_rate": 2.159887193894433e-06,
      "loss": 0.0007,
      "step": 132570
    },
    {
      "epoch": 2.169352859363495,
      "grad_norm": 0.07572520524263382,
      "learning_rate": 2.1591037863276685e-06,
      "loss": 0.0006,
      "step": 132580
    },
    {
      "epoch": 2.169516485314571,
      "grad_norm": 0.13142816722393036,
      "learning_rate": 2.1583204817365623e-06,
      "loss": 0.0005,
      "step": 132590
    },
    {
      "epoch": 2.1696801112656465,
      "grad_norm": 0.09191661328077316,
      "learning_rate": 2.15753728014951e-06,
      "loss": 0.0008,
      "step": 132600
    },
    {
      "epoch": 2.1698437372167225,
      "grad_norm": 0.0891927033662796,
      "learning_rate": 2.1567541815948983e-06,
      "loss": 0.0008,
      "step": 132610
    },
    {
      "epoch": 2.1700073631677985,
      "grad_norm": 0.03349055349826813,
      "learning_rate": 2.155971186101115e-06,
      "loss": 0.0004,
      "step": 132620
    },
    {
      "epoch": 2.170170989118874,
      "grad_norm": 0.07418906688690186,
      "learning_rate": 2.1551882936965397e-06,
      "loss": 0.0017,
      "step": 132630
    },
    {
      "epoch": 2.17033461506995,
      "grad_norm": 0.2738913297653198,
      "learning_rate": 2.154405504409551e-06,
      "loss": 0.0006,
      "step": 132640
    },
    {
      "epoch": 2.170498241021026,
      "grad_norm": 0.11192761361598969,
      "learning_rate": 2.1536228182685276e-06,
      "loss": 0.0014,
      "step": 132650
    },
    {
      "epoch": 2.1706618669721016,
      "grad_norm": 0.04489920288324356,
      "learning_rate": 2.1528402353018325e-06,
      "loss": 0.0008,
      "step": 132660
    },
    {
      "epoch": 2.1708254929231776,
      "grad_norm": 0.096799835562706,
      "learning_rate": 2.152057755537837e-06,
      "loss": 0.0005,
      "step": 132670
    },
    {
      "epoch": 2.1709891188742536,
      "grad_norm": 0.060996923595666885,
      "learning_rate": 2.151275379004902e-06,
      "loss": 0.0011,
      "step": 132680
    },
    {
      "epoch": 2.171152744825329,
      "grad_norm": 0.05241464450955391,
      "learning_rate": 2.150493105731387e-06,
      "loss": 0.0004,
      "step": 132690
    },
    {
      "epoch": 2.171316370776405,
      "grad_norm": 0.0061174314469099045,
      "learning_rate": 2.14971093574565e-06,
      "loss": 0.0009,
      "step": 132700
    },
    {
      "epoch": 2.171479996727481,
      "grad_norm": 0.03174373134970665,
      "learning_rate": 2.14892886907604e-06,
      "loss": 0.0006,
      "step": 132710
    },
    {
      "epoch": 2.1716436226785567,
      "grad_norm": 0.06321625411510468,
      "learning_rate": 2.1481469057509074e-06,
      "loss": 0.0008,
      "step": 132720
    },
    {
      "epoch": 2.1718072486296327,
      "grad_norm": 0.0803573802113533,
      "learning_rate": 2.147365045798594e-06,
      "loss": 0.0007,
      "step": 132730
    },
    {
      "epoch": 2.1719708745807087,
      "grad_norm": 0.12570713460445404,
      "learning_rate": 2.146583289247443e-06,
      "loss": 0.001,
      "step": 132740
    },
    {
      "epoch": 2.1721345005317843,
      "grad_norm": 0.03365262970328331,
      "learning_rate": 2.1458016361257898e-06,
      "loss": 0.0011,
      "step": 132750
    },
    {
      "epoch": 2.1722981264828602,
      "grad_norm": 0.16729222238063812,
      "learning_rate": 2.145020086461966e-06,
      "loss": 0.0009,
      "step": 132760
    },
    {
      "epoch": 2.172461752433936,
      "grad_norm": 0.10984192788600922,
      "learning_rate": 2.1442386402843034e-06,
      "loss": 0.0006,
      "step": 132770
    },
    {
      "epoch": 2.172625378385012,
      "grad_norm": 0.012926003895699978,
      "learning_rate": 2.143457297621125e-06,
      "loss": 0.0008,
      "step": 132780
    },
    {
      "epoch": 2.172789004336088,
      "grad_norm": 0.042420223355293274,
      "learning_rate": 2.1426760585007557e-06,
      "loss": 0.0008,
      "step": 132790
    },
    {
      "epoch": 2.1729526302871633,
      "grad_norm": 0.20108425617218018,
      "learning_rate": 2.14189492295151e-06,
      "loss": 0.0011,
      "step": 132800
    },
    {
      "epoch": 2.1731162562382393,
      "grad_norm": 0.12637944519519806,
      "learning_rate": 2.141113891001706e-06,
      "loss": 0.0009,
      "step": 132810
    },
    {
      "epoch": 2.1732798821893153,
      "grad_norm": 0.010441829450428486,
      "learning_rate": 2.1403329626796503e-06,
      "loss": 0.0004,
      "step": 132820
    },
    {
      "epoch": 2.173443508140391,
      "grad_norm": 0.06600575149059296,
      "learning_rate": 2.1395521380136542e-06,
      "loss": 0.0013,
      "step": 132830
    },
    {
      "epoch": 2.173607134091467,
      "grad_norm": 0.13038602471351624,
      "learning_rate": 2.1387714170320154e-06,
      "loss": 0.0011,
      "step": 132840
    },
    {
      "epoch": 2.173770760042543,
      "grad_norm": 0.053065553307533264,
      "learning_rate": 2.1379907997630385e-06,
      "loss": 0.0007,
      "step": 132850
    },
    {
      "epoch": 2.1739343859936184,
      "grad_norm": 0.03505024313926697,
      "learning_rate": 2.1372102862350164e-06,
      "loss": 0.0005,
      "step": 132860
    },
    {
      "epoch": 2.1740980119446944,
      "grad_norm": 0.030295664444565773,
      "learning_rate": 2.1364298764762387e-06,
      "loss": 0.0003,
      "step": 132870
    },
    {
      "epoch": 2.1742616378957704,
      "grad_norm": 0.026615168899297714,
      "learning_rate": 2.1356495705149977e-06,
      "loss": 0.0004,
      "step": 132880
    },
    {
      "epoch": 2.174425263846846,
      "grad_norm": 0.08507993817329407,
      "learning_rate": 2.1348693683795736e-06,
      "loss": 0.0008,
      "step": 132890
    },
    {
      "epoch": 2.174588889797922,
      "grad_norm": 0.04860775172710419,
      "learning_rate": 2.13408927009825e-06,
      "loss": 0.0005,
      "step": 132900
    },
    {
      "epoch": 2.174752515748998,
      "grad_norm": 0.02069712057709694,
      "learning_rate": 2.1333092756993013e-06,
      "loss": 0.0007,
      "step": 132910
    },
    {
      "epoch": 2.1749161417000735,
      "grad_norm": 0.040829841047525406,
      "learning_rate": 2.132529385211001e-06,
      "loss": 0.0007,
      "step": 132920
    },
    {
      "epoch": 2.1750797676511495,
      "grad_norm": 0.0293737780302763,
      "learning_rate": 2.1317495986616204e-06,
      "loss": 0.0009,
      "step": 132930
    },
    {
      "epoch": 2.1752433936022255,
      "grad_norm": 0.03710537403821945,
      "learning_rate": 2.1309699160794218e-06,
      "loss": 0.0012,
      "step": 132940
    },
    {
      "epoch": 2.175407019553301,
      "grad_norm": 0.05738138034939766,
      "learning_rate": 2.130190337492672e-06,
      "loss": 0.0013,
      "step": 132950
    },
    {
      "epoch": 2.175570645504377,
      "grad_norm": 0.14531312882900238,
      "learning_rate": 2.1294108629296205e-06,
      "loss": 0.001,
      "step": 132960
    },
    {
      "epoch": 2.175734271455453,
      "grad_norm": 0.054931506514549255,
      "learning_rate": 2.128631492418527e-06,
      "loss": 0.0016,
      "step": 132970
    },
    {
      "epoch": 2.1758978974065286,
      "grad_norm": 0.11332526057958603,
      "learning_rate": 2.1278522259876423e-06,
      "loss": 0.0007,
      "step": 132980
    },
    {
      "epoch": 2.1760615233576046,
      "grad_norm": 0.04289693385362625,
      "learning_rate": 2.1270730636652094e-06,
      "loss": 0.0006,
      "step": 132990
    },
    {
      "epoch": 2.17622514930868,
      "grad_norm": 0.06438548117876053,
      "learning_rate": 2.1262940054794745e-06,
      "loss": 0.0006,
      "step": 133000
    },
    {
      "epoch": 2.176388775259756,
      "grad_norm": 0.005645431112498045,
      "learning_rate": 2.1255150514586737e-06,
      "loss": 0.0011,
      "step": 133010
    },
    {
      "epoch": 2.176552401210832,
      "grad_norm": 0.04957915470004082,
      "learning_rate": 2.124736201631045e-06,
      "loss": 0.0005,
      "step": 133020
    },
    {
      "epoch": 2.1767160271619077,
      "grad_norm": 0.08441191911697388,
      "learning_rate": 2.1239574560248174e-06,
      "loss": 0.0006,
      "step": 133030
    },
    {
      "epoch": 2.1768796531129837,
      "grad_norm": 0.018327437341213226,
      "learning_rate": 2.1231788146682208e-06,
      "loss": 0.0011,
      "step": 133040
    },
    {
      "epoch": 2.1770432790640597,
      "grad_norm": 0.006724250037223101,
      "learning_rate": 2.1224002775894784e-06,
      "loss": 0.0006,
      "step": 133050
    },
    {
      "epoch": 2.1772069050151353,
      "grad_norm": 0.028384385630488396,
      "learning_rate": 2.121621844816808e-06,
      "loss": 0.0009,
      "step": 133060
    },
    {
      "epoch": 2.1773705309662112,
      "grad_norm": 0.17846648395061493,
      "learning_rate": 2.1208435163784293e-06,
      "loss": 0.0007,
      "step": 133070
    },
    {
      "epoch": 2.1775341569172872,
      "grad_norm": 0.004224427510052919,
      "learning_rate": 2.120065292302552e-06,
      "loss": 0.0009,
      "step": 133080
    },
    {
      "epoch": 2.177697782868363,
      "grad_norm": 0.03458670154213905,
      "learning_rate": 2.119287172617387e-06,
      "loss": 0.0011,
      "step": 133090
    },
    {
      "epoch": 2.177861408819439,
      "grad_norm": 0.1660071760416031,
      "learning_rate": 2.118509157351137e-06,
      "loss": 0.0005,
      "step": 133100
    },
    {
      "epoch": 2.178025034770515,
      "grad_norm": 0.11826536059379578,
      "learning_rate": 2.117731246532006e-06,
      "loss": 0.0005,
      "step": 133110
    },
    {
      "epoch": 2.1781886607215903,
      "grad_norm": 0.05635539069771767,
      "learning_rate": 2.1169534401881886e-06,
      "loss": 0.0013,
      "step": 133120
    },
    {
      "epoch": 2.1783522866726663,
      "grad_norm": 0.1616489589214325,
      "learning_rate": 2.11617573834788e-06,
      "loss": 0.001,
      "step": 133130
    },
    {
      "epoch": 2.178515912623742,
      "grad_norm": 0.07777858525514603,
      "learning_rate": 2.115398141039271e-06,
      "loss": 0.0007,
      "step": 133140
    },
    {
      "epoch": 2.178679538574818,
      "grad_norm": 0.055220961570739746,
      "learning_rate": 2.1146206482905467e-06,
      "loss": 0.0011,
      "step": 133150
    },
    {
      "epoch": 2.178843164525894,
      "grad_norm": 0.10929786413908005,
      "learning_rate": 2.1138432601298893e-06,
      "loss": 0.0013,
      "step": 133160
    },
    {
      "epoch": 2.1790067904769694,
      "grad_norm": 0.051200803369283676,
      "learning_rate": 2.113065976585475e-06,
      "loss": 0.0009,
      "step": 133170
    },
    {
      "epoch": 2.1791704164280454,
      "grad_norm": 0.14246031641960144,
      "learning_rate": 2.112288797685482e-06,
      "loss": 0.0013,
      "step": 133180
    },
    {
      "epoch": 2.1793340423791214,
      "grad_norm": 0.49097469449043274,
      "learning_rate": 2.1115117234580783e-06,
      "loss": 0.001,
      "step": 133190
    },
    {
      "epoch": 2.179497668330197,
      "grad_norm": 0.0888969674706459,
      "learning_rate": 2.1107347539314323e-06,
      "loss": 0.0009,
      "step": 133200
    },
    {
      "epoch": 2.179661294281273,
      "grad_norm": 0.017145998775959015,
      "learning_rate": 2.109957889133709e-06,
      "loss": 0.001,
      "step": 133210
    },
    {
      "epoch": 2.179824920232349,
      "grad_norm": 0.12172749638557434,
      "learning_rate": 2.109181129093064e-06,
      "loss": 0.0045,
      "step": 133220
    },
    {
      "epoch": 2.1799885461834245,
      "grad_norm": 0.03835159167647362,
      "learning_rate": 2.1084044738376576e-06,
      "loss": 0.0007,
      "step": 133230
    },
    {
      "epoch": 2.1801521721345005,
      "grad_norm": 0.15464426577091217,
      "learning_rate": 2.107627923395637e-06,
      "loss": 0.0006,
      "step": 133240
    },
    {
      "epoch": 2.1803157980855765,
      "grad_norm": 0.00905674323439598,
      "learning_rate": 2.106851477795155e-06,
      "loss": 0.0012,
      "step": 133250
    },
    {
      "epoch": 2.180479424036652,
      "grad_norm": 0.1142902597784996,
      "learning_rate": 2.106075137064353e-06,
      "loss": 0.0008,
      "step": 133260
    },
    {
      "epoch": 2.180643049987728,
      "grad_norm": 0.04101067781448364,
      "learning_rate": 2.10529890123137e-06,
      "loss": 0.0014,
      "step": 133270
    },
    {
      "epoch": 2.180806675938804,
      "grad_norm": 0.002335965633392334,
      "learning_rate": 2.1045227703243464e-06,
      "loss": 0.0012,
      "step": 133280
    },
    {
      "epoch": 2.1809703018898796,
      "grad_norm": 0.11783561110496521,
      "learning_rate": 2.1037467443714117e-06,
      "loss": 0.0012,
      "step": 133290
    },
    {
      "epoch": 2.1811339278409556,
      "grad_norm": 0.1293431669473648,
      "learning_rate": 2.102970823400698e-06,
      "loss": 0.0006,
      "step": 133300
    },
    {
      "epoch": 2.1812975537920316,
      "grad_norm": 0.15829302370548248,
      "learning_rate": 2.1021950074403273e-06,
      "loss": 0.0008,
      "step": 133310
    },
    {
      "epoch": 2.181461179743107,
      "grad_norm": 0.0488809309899807,
      "learning_rate": 2.1014192965184253e-06,
      "loss": 0.0007,
      "step": 133320
    },
    {
      "epoch": 2.181624805694183,
      "grad_norm": 0.013442493975162506,
      "learning_rate": 2.100643690663105e-06,
      "loss": 0.0026,
      "step": 133330
    },
    {
      "epoch": 2.181788431645259,
      "grad_norm": 0.03188556432723999,
      "learning_rate": 2.099868189902484e-06,
      "loss": 0.001,
      "step": 133340
    },
    {
      "epoch": 2.1819520575963347,
      "grad_norm": 0.05730525776743889,
      "learning_rate": 2.099092794264672e-06,
      "loss": 0.0008,
      "step": 133350
    },
    {
      "epoch": 2.1821156835474107,
      "grad_norm": 0.17060551047325134,
      "learning_rate": 2.098317503777771e-06,
      "loss": 0.0016,
      "step": 133360
    },
    {
      "epoch": 2.1822793094984863,
      "grad_norm": 0.14788545668125153,
      "learning_rate": 2.097542318469889e-06,
      "loss": 0.0012,
      "step": 133370
    },
    {
      "epoch": 2.1824429354495622,
      "grad_norm": 0.03542299196124077,
      "learning_rate": 2.09676723836912e-06,
      "loss": 0.0006,
      "step": 133380
    },
    {
      "epoch": 2.1826065614006382,
      "grad_norm": 0.19183960556983948,
      "learning_rate": 2.0959922635035623e-06,
      "loss": 0.0013,
      "step": 133390
    },
    {
      "epoch": 2.182770187351714,
      "grad_norm": 0.03962555155158043,
      "learning_rate": 2.095217393901304e-06,
      "loss": 0.0023,
      "step": 133400
    },
    {
      "epoch": 2.18293381330279,
      "grad_norm": 0.0568387545645237,
      "learning_rate": 2.0944426295904342e-06,
      "loss": 0.0009,
      "step": 133410
    },
    {
      "epoch": 2.183097439253866,
      "grad_norm": 0.005415895022451878,
      "learning_rate": 2.093667970599037e-06,
      "loss": 0.0005,
      "step": 133420
    },
    {
      "epoch": 2.1832610652049413,
      "grad_norm": 0.07867998629808426,
      "learning_rate": 2.092893416955189e-06,
      "loss": 0.001,
      "step": 133430
    },
    {
      "epoch": 2.1834246911560173,
      "grad_norm": 0.06450201570987701,
      "learning_rate": 2.092118968686969e-06,
      "loss": 0.0007,
      "step": 133440
    },
    {
      "epoch": 2.1835883171070933,
      "grad_norm": 0.1559048295021057,
      "learning_rate": 2.0913446258224463e-06,
      "loss": 0.0011,
      "step": 133450
    },
    {
      "epoch": 2.183751943058169,
      "grad_norm": 0.12856632471084595,
      "learning_rate": 2.0905703883896937e-06,
      "loss": 0.0011,
      "step": 133460
    },
    {
      "epoch": 2.183915569009245,
      "grad_norm": 0.07965055853128433,
      "learning_rate": 2.089796256416768e-06,
      "loss": 0.0008,
      "step": 133470
    },
    {
      "epoch": 2.184079194960321,
      "grad_norm": 0.08476488292217255,
      "learning_rate": 2.0890222299317336e-06,
      "loss": 0.0009,
      "step": 133480
    },
    {
      "epoch": 2.1842428209113964,
      "grad_norm": 0.014255587011575699,
      "learning_rate": 2.088248308962649e-06,
      "loss": 0.0005,
      "step": 133490
    },
    {
      "epoch": 2.1844064468624724,
      "grad_norm": 0.14668671786785126,
      "learning_rate": 2.0874744935375627e-06,
      "loss": 0.0009,
      "step": 133500
    },
    {
      "epoch": 2.1845700728135484,
      "grad_norm": 0.15126439929008484,
      "learning_rate": 2.0867007836845282e-06,
      "loss": 0.001,
      "step": 133510
    },
    {
      "epoch": 2.184733698764624,
      "grad_norm": 0.06401389092206955,
      "learning_rate": 2.085927179431587e-06,
      "loss": 0.0008,
      "step": 133520
    },
    {
      "epoch": 2.1848973247157,
      "grad_norm": 0.06215459480881691,
      "learning_rate": 2.0851536808067824e-06,
      "loss": 0.0004,
      "step": 133530
    },
    {
      "epoch": 2.1850609506667755,
      "grad_norm": 0.0100501524284482,
      "learning_rate": 2.08438028783815e-06,
      "loss": 0.0008,
      "step": 133540
    },
    {
      "epoch": 2.1852245766178515,
      "grad_norm": 0.015297452919185162,
      "learning_rate": 2.0836070005537264e-06,
      "loss": 0.0008,
      "step": 133550
    },
    {
      "epoch": 2.1853882025689275,
      "grad_norm": 0.10489283502101898,
      "learning_rate": 2.08283381898154e-06,
      "loss": 0.0014,
      "step": 133560
    },
    {
      "epoch": 2.185551828520003,
      "grad_norm": 0.12617242336273193,
      "learning_rate": 2.0820607431496143e-06,
      "loss": 0.0009,
      "step": 133570
    },
    {
      "epoch": 2.185715454471079,
      "grad_norm": 0.061616938561201096,
      "learning_rate": 2.0812877730859754e-06,
      "loss": 0.0007,
      "step": 133580
    },
    {
      "epoch": 2.185879080422155,
      "grad_norm": 0.05792645737528801,
      "learning_rate": 2.0805149088186375e-06,
      "loss": 0.0006,
      "step": 133590
    },
    {
      "epoch": 2.1860427063732306,
      "grad_norm": 0.37530988454818726,
      "learning_rate": 2.0797421503756197e-06,
      "loss": 0.0011,
      "step": 133600
    },
    {
      "epoch": 2.1862063323243066,
      "grad_norm": 0.0029510220047086477,
      "learning_rate": 2.078969497784929e-06,
      "loss": 0.0012,
      "step": 133610
    },
    {
      "epoch": 2.1863699582753826,
      "grad_norm": 0.09248299896717072,
      "learning_rate": 2.078196951074575e-06,
      "loss": 0.0009,
      "step": 133620
    },
    {
      "epoch": 2.186533584226458,
      "grad_norm": 0.05774078145623207,
      "learning_rate": 2.0774245102725565e-06,
      "loss": 0.0003,
      "step": 133630
    },
    {
      "epoch": 2.186697210177534,
      "grad_norm": 0.0465075820684433,
      "learning_rate": 2.076652175406876e-06,
      "loss": 0.001,
      "step": 133640
    },
    {
      "epoch": 2.18686083612861,
      "grad_norm": 0.06741433590650558,
      "learning_rate": 2.07587994650553e-06,
      "loss": 0.0009,
      "step": 133650
    },
    {
      "epoch": 2.1870244620796857,
      "grad_norm": 0.004792368970811367,
      "learning_rate": 2.0751078235965076e-06,
      "loss": 0.0006,
      "step": 133660
    },
    {
      "epoch": 2.1871880880307617,
      "grad_norm": 0.1143391877412796,
      "learning_rate": 2.0743358067077967e-06,
      "loss": 0.001,
      "step": 133670
    },
    {
      "epoch": 2.1873517139818377,
      "grad_norm": 0.02038666047155857,
      "learning_rate": 2.07356389586738e-06,
      "loss": 0.0006,
      "step": 133680
    },
    {
      "epoch": 2.1875153399329132,
      "grad_norm": 0.29164034128189087,
      "learning_rate": 2.0727920911032377e-06,
      "loss": 0.0008,
      "step": 133690
    },
    {
      "epoch": 2.1876789658839892,
      "grad_norm": 0.016913043335080147,
      "learning_rate": 2.072020392443349e-06,
      "loss": 0.0008,
      "step": 133700
    },
    {
      "epoch": 2.1878425918350652,
      "grad_norm": 0.05663401260972023,
      "learning_rate": 2.0712487999156823e-06,
      "loss": 0.0008,
      "step": 133710
    },
    {
      "epoch": 2.188006217786141,
      "grad_norm": 0.13354496657848358,
      "learning_rate": 2.070477313548209e-06,
      "loss": 0.0005,
      "step": 133720
    },
    {
      "epoch": 2.188169843737217,
      "grad_norm": 0.09139169752597809,
      "learning_rate": 2.0697059333688906e-06,
      "loss": 0.0006,
      "step": 133730
    },
    {
      "epoch": 2.188333469688293,
      "grad_norm": 0.18714956939220428,
      "learning_rate": 2.0689346594056914e-06,
      "loss": 0.0017,
      "step": 133740
    },
    {
      "epoch": 2.1884970956393683,
      "grad_norm": 0.14113494753837585,
      "learning_rate": 2.068163491686564e-06,
      "loss": 0.0011,
      "step": 133750
    },
    {
      "epoch": 2.1886607215904443,
      "grad_norm": 0.017769034951925278,
      "learning_rate": 2.067392430239465e-06,
      "loss": 0.0007,
      "step": 133760
    },
    {
      "epoch": 2.18882434754152,
      "grad_norm": 0.08506893366575241,
      "learning_rate": 2.0666214750923426e-06,
      "loss": 0.0009,
      "step": 133770
    },
    {
      "epoch": 2.188987973492596,
      "grad_norm": 0.10716760903596878,
      "learning_rate": 2.065850626273139e-06,
      "loss": 0.0008,
      "step": 133780
    },
    {
      "epoch": 2.189151599443672,
      "grad_norm": 0.005107826087623835,
      "learning_rate": 2.0650798838098007e-06,
      "loss": 0.0004,
      "step": 133790
    },
    {
      "epoch": 2.1893152253947474,
      "grad_norm": 0.09522493928670883,
      "learning_rate": 2.0643092477302606e-06,
      "loss": 0.0005,
      "step": 133800
    },
    {
      "epoch": 2.1894788513458234,
      "grad_norm": 0.09006105363368988,
      "learning_rate": 2.0635387180624567e-06,
      "loss": 0.002,
      "step": 133810
    },
    {
      "epoch": 2.1896424772968994,
      "grad_norm": 0.04400699958205223,
      "learning_rate": 2.0627682948343147e-06,
      "loss": 0.0004,
      "step": 133820
    },
    {
      "epoch": 2.189806103247975,
      "grad_norm": 0.11281992495059967,
      "learning_rate": 2.0619979780737643e-06,
      "loss": 0.0008,
      "step": 133830
    },
    {
      "epoch": 2.189969729199051,
      "grad_norm": 0.04164297506213188,
      "learning_rate": 2.061227767808724e-06,
      "loss": 0.0006,
      "step": 133840
    },
    {
      "epoch": 2.190133355150127,
      "grad_norm": 0.05682015046477318,
      "learning_rate": 2.0604576640671136e-06,
      "loss": 0.0013,
      "step": 133850
    },
    {
      "epoch": 2.1902969811012025,
      "grad_norm": 0.03694860637187958,
      "learning_rate": 2.059687666876852e-06,
      "loss": 0.0007,
      "step": 133860
    },
    {
      "epoch": 2.1904606070522785,
      "grad_norm": 0.08465483039617538,
      "learning_rate": 2.0589177762658417e-06,
      "loss": 0.0008,
      "step": 133870
    },
    {
      "epoch": 2.1906242330033545,
      "grad_norm": 0.05325772613286972,
      "learning_rate": 2.0581479922619956e-06,
      "loss": 0.0014,
      "step": 133880
    },
    {
      "epoch": 2.19078785895443,
      "grad_norm": 0.04683837294578552,
      "learning_rate": 2.0573783148932126e-06,
      "loss": 0.0012,
      "step": 133890
    },
    {
      "epoch": 2.190951484905506,
      "grad_norm": 0.05582217872142792,
      "learning_rate": 2.0566087441873924e-06,
      "loss": 0.0006,
      "step": 133900
    },
    {
      "epoch": 2.191115110856582,
      "grad_norm": 0.02832895517349243,
      "learning_rate": 2.055839280172433e-06,
      "loss": 0.0006,
      "step": 133910
    },
    {
      "epoch": 2.1912787368076576,
      "grad_norm": 0.050031501799821854,
      "learning_rate": 2.055069922876222e-06,
      "loss": 0.0006,
      "step": 133920
    },
    {
      "epoch": 2.1914423627587336,
      "grad_norm": 0.1772473156452179,
      "learning_rate": 2.05430067232665e-06,
      "loss": 0.0012,
      "step": 133930
    },
    {
      "epoch": 2.191605988709809,
      "grad_norm": 0.009561151266098022,
      "learning_rate": 2.053531528551598e-06,
      "loss": 0.0005,
      "step": 133940
    },
    {
      "epoch": 2.191769614660885,
      "grad_norm": 0.060883939266204834,
      "learning_rate": 2.0527624915789467e-06,
      "loss": 0.0018,
      "step": 133950
    },
    {
      "epoch": 2.191933240611961,
      "grad_norm": 0.04925726354122162,
      "learning_rate": 2.051993561436573e-06,
      "loss": 0.0008,
      "step": 133960
    },
    {
      "epoch": 2.1920968665630367,
      "grad_norm": 0.07513914257287979,
      "learning_rate": 2.051224738152344e-06,
      "loss": 0.0006,
      "step": 133970
    },
    {
      "epoch": 2.1922604925141127,
      "grad_norm": 0.14928783476352692,
      "learning_rate": 2.0504560217541343e-06,
      "loss": 0.0008,
      "step": 133980
    },
    {
      "epoch": 2.1924241184651887,
      "grad_norm": 0.028079692274332047,
      "learning_rate": 2.0496874122698017e-06,
      "loss": 0.0007,
      "step": 133990
    },
    {
      "epoch": 2.1925877444162643,
      "grad_norm": 0.0803908258676529,
      "learning_rate": 2.0489189097272124e-06,
      "loss": 0.0008,
      "step": 134000
    },
    {
      "epoch": 2.1927513703673402,
      "grad_norm": 0.04568302631378174,
      "learning_rate": 2.0481505141542176e-06,
      "loss": 0.0004,
      "step": 134010
    },
    {
      "epoch": 2.1929149963184162,
      "grad_norm": 0.057581063359975815,
      "learning_rate": 2.0473822255786733e-06,
      "loss": 0.0009,
      "step": 134020
    },
    {
      "epoch": 2.193078622269492,
      "grad_norm": 0.035008154809474945,
      "learning_rate": 2.046614044028426e-06,
      "loss": 0.0005,
      "step": 134030
    },
    {
      "epoch": 2.193242248220568,
      "grad_norm": 0.04905608296394348,
      "learning_rate": 2.045845969531323e-06,
      "loss": 0.0011,
      "step": 134040
    },
    {
      "epoch": 2.193405874171644,
      "grad_norm": 0.0941646546125412,
      "learning_rate": 2.0450780021152017e-06,
      "loss": 0.0014,
      "step": 134050
    },
    {
      "epoch": 2.1935695001227193,
      "grad_norm": 0.07459456473588943,
      "learning_rate": 2.0443101418079025e-06,
      "loss": 0.0005,
      "step": 134060
    },
    {
      "epoch": 2.1937331260737953,
      "grad_norm": 0.00786400306969881,
      "learning_rate": 2.043542388637257e-06,
      "loss": 0.0009,
      "step": 134070
    },
    {
      "epoch": 2.1938967520248713,
      "grad_norm": 0.34217965602874756,
      "learning_rate": 2.042774742631092e-06,
      "loss": 0.0017,
      "step": 134080
    },
    {
      "epoch": 2.194060377975947,
      "grad_norm": 0.12149641662836075,
      "learning_rate": 2.0420072038172374e-06,
      "loss": 0.0016,
      "step": 134090
    },
    {
      "epoch": 2.194224003927023,
      "grad_norm": 0.04716979339718819,
      "learning_rate": 2.04123977222351e-06,
      "loss": 0.0005,
      "step": 134100
    },
    {
      "epoch": 2.194387629878099,
      "grad_norm": 0.02315519005060196,
      "learning_rate": 2.0404724478777314e-06,
      "loss": 0.0007,
      "step": 134110
    },
    {
      "epoch": 2.1945512558291744,
      "grad_norm": 0.05076749250292778,
      "learning_rate": 2.039705230807712e-06,
      "loss": 0.0007,
      "step": 134120
    },
    {
      "epoch": 2.1947148817802504,
      "grad_norm": 0.05808860436081886,
      "learning_rate": 2.0389381210412634e-06,
      "loss": 0.0007,
      "step": 134130
    },
    {
      "epoch": 2.1948785077313264,
      "grad_norm": 0.03759097680449486,
      "learning_rate": 2.0381711186061925e-06,
      "loss": 0.0009,
      "step": 134140
    },
    {
      "epoch": 2.195042133682402,
      "grad_norm": 0.0022665902506560087,
      "learning_rate": 2.037404223530299e-06,
      "loss": 0.0018,
      "step": 134150
    },
    {
      "epoch": 2.195205759633478,
      "grad_norm": 0.06863267719745636,
      "learning_rate": 2.036637435841385e-06,
      "loss": 0.0013,
      "step": 134160
    },
    {
      "epoch": 2.1953693855845535,
      "grad_norm": 0.16734172403812408,
      "learning_rate": 2.035870755567238e-06,
      "loss": 0.0007,
      "step": 134170
    },
    {
      "epoch": 2.1955330115356295,
      "grad_norm": 0.06679178029298782,
      "learning_rate": 2.0351041827356523e-06,
      "loss": 0.0008,
      "step": 134180
    },
    {
      "epoch": 2.1956966374867055,
      "grad_norm": 0.10849496722221375,
      "learning_rate": 2.034337717374415e-06,
      "loss": 0.0006,
      "step": 134190
    },
    {
      "epoch": 2.195860263437781,
      "grad_norm": 0.05884971469640732,
      "learning_rate": 2.0335713595113066e-06,
      "loss": 0.0006,
      "step": 134200
    },
    {
      "epoch": 2.196023889388857,
      "grad_norm": 0.1716867834329605,
      "learning_rate": 2.0328051091741085e-06,
      "loss": 0.0008,
      "step": 134210
    },
    {
      "epoch": 2.196187515339933,
      "grad_norm": 0.0182280782610178,
      "learning_rate": 2.032038966390592e-06,
      "loss": 0.0004,
      "step": 134220
    },
    {
      "epoch": 2.1963511412910086,
      "grad_norm": 0.07228348404169083,
      "learning_rate": 2.0312729311885314e-06,
      "loss": 0.0018,
      "step": 134230
    },
    {
      "epoch": 2.1965147672420846,
      "grad_norm": 0.02363823726773262,
      "learning_rate": 2.0305070035956903e-06,
      "loss": 0.0002,
      "step": 134240
    },
    {
      "epoch": 2.1966783931931606,
      "grad_norm": 0.10083484649658203,
      "learning_rate": 2.029741183639835e-06,
      "loss": 0.001,
      "step": 134250
    },
    {
      "epoch": 2.196842019144236,
      "grad_norm": 0.07068591564893723,
      "learning_rate": 2.0289754713487227e-06,
      "loss": 0.002,
      "step": 134260
    },
    {
      "epoch": 2.197005645095312,
      "grad_norm": 0.026403365656733513,
      "learning_rate": 2.0282098667501073e-06,
      "loss": 0.0007,
      "step": 134270
    },
    {
      "epoch": 2.197169271046388,
      "grad_norm": 0.10789033025503159,
      "learning_rate": 2.027444369871743e-06,
      "loss": 0.0007,
      "step": 134280
    },
    {
      "epoch": 2.1973328969974637,
      "grad_norm": 0.06284467875957489,
      "learning_rate": 2.026678980741375e-06,
      "loss": 0.0006,
      "step": 134290
    },
    {
      "epoch": 2.1974965229485397,
      "grad_norm": 0.032083287835121155,
      "learning_rate": 2.0259136993867495e-06,
      "loss": 0.0006,
      "step": 134300
    },
    {
      "epoch": 2.1976601488996153,
      "grad_norm": 0.08283887803554535,
      "learning_rate": 2.0251485258356023e-06,
      "loss": 0.0008,
      "step": 134310
    },
    {
      "epoch": 2.1978237748506912,
      "grad_norm": 0.005516393110156059,
      "learning_rate": 2.0243834601156727e-06,
      "loss": 0.0003,
      "step": 134320
    },
    {
      "epoch": 2.1979874008017672,
      "grad_norm": 0.07858219742774963,
      "learning_rate": 2.0236185022546897e-06,
      "loss": 0.0008,
      "step": 134330
    },
    {
      "epoch": 2.198151026752843,
      "grad_norm": 0.09823130816221237,
      "learning_rate": 2.022853652280382e-06,
      "loss": 0.001,
      "step": 134340
    },
    {
      "epoch": 2.198314652703919,
      "grad_norm": 0.036159295588731766,
      "learning_rate": 2.022088910220476e-06,
      "loss": 0.0009,
      "step": 134350
    },
    {
      "epoch": 2.198478278654995,
      "grad_norm": 0.06728903949260712,
      "learning_rate": 2.0213242761026896e-06,
      "loss": 0.0011,
      "step": 134360
    },
    {
      "epoch": 2.1986419046060703,
      "grad_norm": 0.12728987634181976,
      "learning_rate": 2.0205597499547393e-06,
      "loss": 0.0008,
      "step": 134370
    },
    {
      "epoch": 2.1988055305571463,
      "grad_norm": 0.028947558254003525,
      "learning_rate": 2.0197953318043355e-06,
      "loss": 0.0007,
      "step": 134380
    },
    {
      "epoch": 2.1989691565082223,
      "grad_norm": 0.04294012114405632,
      "learning_rate": 2.01903102167919e-06,
      "loss": 0.001,
      "step": 134390
    },
    {
      "epoch": 2.199132782459298,
      "grad_norm": 0.28890275955200195,
      "learning_rate": 2.018266819607003e-06,
      "loss": 0.0011,
      "step": 134400
    },
    {
      "epoch": 2.199296408410374,
      "grad_norm": 0.02271469682455063,
      "learning_rate": 2.0175027256154785e-06,
      "loss": 0.0004,
      "step": 134410
    },
    {
      "epoch": 2.19946003436145,
      "grad_norm": 0.10078608244657516,
      "learning_rate": 2.016738739732313e-06,
      "loss": 0.001,
      "step": 134420
    },
    {
      "epoch": 2.1996236603125254,
      "grad_norm": 0.04654580354690552,
      "learning_rate": 2.015974861985196e-06,
      "loss": 0.001,
      "step": 134430
    },
    {
      "epoch": 2.1997872862636014,
      "grad_norm": 0.0663909912109375,
      "learning_rate": 2.0152110924018204e-06,
      "loss": 0.0005,
      "step": 134440
    },
    {
      "epoch": 2.1999509122146774,
      "grad_norm": 0.07165167480707169,
      "learning_rate": 2.014447431009868e-06,
      "loss": 0.0006,
      "step": 134450
    },
    {
      "epoch": 2.200114538165753,
      "grad_norm": 0.08370284736156464,
      "learning_rate": 2.013683877837021e-06,
      "loss": 0.0009,
      "step": 134460
    },
    {
      "epoch": 2.200278164116829,
      "grad_norm": 0.1733408421278,
      "learning_rate": 2.012920432910957e-06,
      "loss": 0.0008,
      "step": 134470
    },
    {
      "epoch": 2.200441790067905,
      "grad_norm": 0.07748666405677795,
      "learning_rate": 2.0121570962593456e-06,
      "loss": 0.0004,
      "step": 134480
    },
    {
      "epoch": 2.2006054160189805,
      "grad_norm": 0.08153659850358963,
      "learning_rate": 2.011393867909861e-06,
      "loss": 0.001,
      "step": 134490
    },
    {
      "epoch": 2.2007690419700565,
      "grad_norm": 0.0690116211771965,
      "learning_rate": 2.010630747890163e-06,
      "loss": 0.0013,
      "step": 134500
    },
    {
      "epoch": 2.2009326679211325,
      "grad_norm": 0.07460810989141464,
      "learning_rate": 2.0098677362279182e-06,
      "loss": 0.0007,
      "step": 134510
    },
    {
      "epoch": 2.201096293872208,
      "grad_norm": 0.043203577399253845,
      "learning_rate": 2.0091048329507794e-06,
      "loss": 0.0008,
      "step": 134520
    },
    {
      "epoch": 2.201259919823284,
      "grad_norm": 0.0757097378373146,
      "learning_rate": 2.008342038086404e-06,
      "loss": 0.0005,
      "step": 134530
    },
    {
      "epoch": 2.2014235457743596,
      "grad_norm": 0.01177325751632452,
      "learning_rate": 2.007579351662437e-06,
      "loss": 0.001,
      "step": 134540
    },
    {
      "epoch": 2.2015871717254356,
      "grad_norm": 0.012817207723855972,
      "learning_rate": 2.0068167737065286e-06,
      "loss": 0.0006,
      "step": 134550
    },
    {
      "epoch": 2.2017507976765116,
      "grad_norm": 0.15084563195705414,
      "learning_rate": 2.006054304246318e-06,
      "loss": 0.0008,
      "step": 134560
    },
    {
      "epoch": 2.201914423627587,
      "grad_norm": 0.09428867697715759,
      "learning_rate": 2.0052919433094412e-06,
      "loss": 0.0005,
      "step": 134570
    },
    {
      "epoch": 2.202078049578663,
      "grad_norm": 0.02012704871594906,
      "learning_rate": 2.004529690923536e-06,
      "loss": 0.0005,
      "step": 134580
    },
    {
      "epoch": 2.202241675529739,
      "grad_norm": 0.10638763755559921,
      "learning_rate": 2.003767547116229e-06,
      "loss": 0.0009,
      "step": 134590
    },
    {
      "epoch": 2.2024053014808147,
      "grad_norm": 0.013622605241835117,
      "learning_rate": 2.0030055119151474e-06,
      "loss": 0.0015,
      "step": 134600
    },
    {
      "epoch": 2.2025689274318907,
      "grad_norm": 0.08915100991725922,
      "learning_rate": 2.002243585347912e-06,
      "loss": 0.0007,
      "step": 134610
    },
    {
      "epoch": 2.2027325533829667,
      "grad_norm": 0.08205676823854446,
      "learning_rate": 2.0014817674421417e-06,
      "loss": 0.0009,
      "step": 134620
    },
    {
      "epoch": 2.2028961793340422,
      "grad_norm": 0.08206752687692642,
      "learning_rate": 2.000720058225452e-06,
      "loss": 0.001,
      "step": 134630
    },
    {
      "epoch": 2.2030598052851182,
      "grad_norm": 0.0041562942788004875,
      "learning_rate": 1.9999584577254506e-06,
      "loss": 0.0009,
      "step": 134640
    },
    {
      "epoch": 2.2032234312361942,
      "grad_norm": 0.020313045009970665,
      "learning_rate": 1.9991969659697457e-06,
      "loss": 0.0014,
      "step": 134650
    },
    {
      "epoch": 2.20338705718727,
      "grad_norm": 0.04158085957169533,
      "learning_rate": 1.998435582985939e-06,
      "loss": 0.0006,
      "step": 134660
    },
    {
      "epoch": 2.203550683138346,
      "grad_norm": 0.13851560652256012,
      "learning_rate": 1.9976743088016264e-06,
      "loss": 0.0021,
      "step": 134670
    },
    {
      "epoch": 2.203714309089422,
      "grad_norm": 0.03624722734093666,
      "learning_rate": 1.9969131434444062e-06,
      "loss": 0.0006,
      "step": 134680
    },
    {
      "epoch": 2.2038779350404973,
      "grad_norm": 0.09091798961162567,
      "learning_rate": 1.9961520869418643e-06,
      "loss": 0.0008,
      "step": 134690
    },
    {
      "epoch": 2.2040415609915733,
      "grad_norm": 0.014389803633093834,
      "learning_rate": 1.9953911393215923e-06,
      "loss": 0.0017,
      "step": 134700
    },
    {
      "epoch": 2.204205186942649,
      "grad_norm": 0.031772494316101074,
      "learning_rate": 1.994630300611168e-06,
      "loss": 0.001,
      "step": 134710
    },
    {
      "epoch": 2.204368812893725,
      "grad_norm": 0.11197585612535477,
      "learning_rate": 1.993869570838174e-06,
      "loss": 0.001,
      "step": 134720
    },
    {
      "epoch": 2.204532438844801,
      "grad_norm": 0.05739670991897583,
      "learning_rate": 1.9931089500301807e-06,
      "loss": 0.0006,
      "step": 134730
    },
    {
      "epoch": 2.2046960647958764,
      "grad_norm": 0.03058839589357376,
      "learning_rate": 1.9923484382147632e-06,
      "loss": 0.0007,
      "step": 134740
    },
    {
      "epoch": 2.2048596907469524,
      "grad_norm": 0.03984062373638153,
      "learning_rate": 1.991588035419485e-06,
      "loss": 0.0009,
      "step": 134750
    },
    {
      "epoch": 2.2050233166980284,
      "grad_norm": 0.1270948052406311,
      "learning_rate": 1.9908277416719112e-06,
      "loss": 0.0009,
      "step": 134760
    },
    {
      "epoch": 2.205186942649104,
      "grad_norm": 0.04314705356955528,
      "learning_rate": 1.990067556999599e-06,
      "loss": 0.0003,
      "step": 134770
    },
    {
      "epoch": 2.20535056860018,
      "grad_norm": 0.03575983643531799,
      "learning_rate": 1.9893074814301027e-06,
      "loss": 0.0005,
      "step": 134780
    },
    {
      "epoch": 2.205514194551256,
      "grad_norm": 0.05156363919377327,
      "learning_rate": 1.9885475149909755e-06,
      "loss": 0.0009,
      "step": 134790
    },
    {
      "epoch": 2.2056778205023315,
      "grad_norm": 0.009673473425209522,
      "learning_rate": 1.9877876577097614e-06,
      "loss": 0.0011,
      "step": 134800
    },
    {
      "epoch": 2.2058414464534075,
      "grad_norm": 0.05272473767399788,
      "learning_rate": 1.9870279096140067e-06,
      "loss": 0.0007,
      "step": 134810
    },
    {
      "epoch": 2.2060050724044835,
      "grad_norm": 0.05631604418158531,
      "learning_rate": 1.9862682707312468e-06,
      "loss": 0.0008,
      "step": 134820
    },
    {
      "epoch": 2.206168698355559,
      "grad_norm": 0.09839834272861481,
      "learning_rate": 1.985508741089021e-06,
      "loss": 0.0008,
      "step": 134830
    },
    {
      "epoch": 2.206332324306635,
      "grad_norm": 0.06707464158535004,
      "learning_rate": 1.984749320714856e-06,
      "loss": 0.0006,
      "step": 134840
    },
    {
      "epoch": 2.206495950257711,
      "grad_norm": 0.02126784808933735,
      "learning_rate": 1.983990009636282e-06,
      "loss": 0.0011,
      "step": 134850
    },
    {
      "epoch": 2.2066595762087866,
      "grad_norm": 0.034013696014881134,
      "learning_rate": 1.9832308078808236e-06,
      "loss": 0.0007,
      "step": 134860
    },
    {
      "epoch": 2.2068232021598626,
      "grad_norm": 0.005292203743010759,
      "learning_rate": 1.9824717154759953e-06,
      "loss": 0.0009,
      "step": 134870
    },
    {
      "epoch": 2.2069868281109386,
      "grad_norm": 0.14281445741653442,
      "learning_rate": 1.9817127324493163e-06,
      "loss": 0.0018,
      "step": 134880
    },
    {
      "epoch": 2.207150454062014,
      "grad_norm": 0.0043332139030098915,
      "learning_rate": 1.980953858828295e-06,
      "loss": 0.0003,
      "step": 134890
    },
    {
      "epoch": 2.20731408001309,
      "grad_norm": 0.005699648056179285,
      "learning_rate": 1.98019509464044e-06,
      "loss": 0.0008,
      "step": 134900
    },
    {
      "epoch": 2.207477705964166,
      "grad_norm": 0.05777039751410484,
      "learning_rate": 1.9794364399132563e-06,
      "loss": 0.0007,
      "step": 134910
    },
    {
      "epoch": 2.2076413319152417,
      "grad_norm": 0.047504715621471405,
      "learning_rate": 1.9786778946742406e-06,
      "loss": 0.0008,
      "step": 134920
    },
    {
      "epoch": 2.2078049578663177,
      "grad_norm": 0.32164159417152405,
      "learning_rate": 1.9779194589508917e-06,
      "loss": 0.0008,
      "step": 134930
    },
    {
      "epoch": 2.2079685838173932,
      "grad_norm": 0.15752612054347992,
      "learning_rate": 1.977161132770696e-06,
      "loss": 0.0007,
      "step": 134940
    },
    {
      "epoch": 2.2081322097684692,
      "grad_norm": 0.026226697489619255,
      "learning_rate": 1.976402916161147e-06,
      "loss": 0.0009,
      "step": 134950
    },
    {
      "epoch": 2.2082958357195452,
      "grad_norm": 0.05336420238018036,
      "learning_rate": 1.975644809149725e-06,
      "loss": 0.0011,
      "step": 134960
    },
    {
      "epoch": 2.208459461670621,
      "grad_norm": 0.0060340347699820995,
      "learning_rate": 1.974886811763908e-06,
      "loss": 0.0039,
      "step": 134970
    },
    {
      "epoch": 2.208623087621697,
      "grad_norm": 0.03583291172981262,
      "learning_rate": 1.9741289240311757e-06,
      "loss": 0.0005,
      "step": 134980
    },
    {
      "epoch": 2.208786713572773,
      "grad_norm": 0.0591684952378273,
      "learning_rate": 1.973371145978995e-06,
      "loss": 0.0008,
      "step": 134990
    },
    {
      "epoch": 2.2089503395238483,
      "grad_norm": 0.02983226254582405,
      "learning_rate": 1.9726134776348377e-06,
      "loss": 0.0003,
      "step": 135000
    },
    {
      "epoch": 2.2091139654749243,
      "grad_norm": 0.08417952805757523,
      "learning_rate": 1.971855919026164e-06,
      "loss": 0.0008,
      "step": 135010
    },
    {
      "epoch": 2.2092775914260003,
      "grad_norm": 0.06742607057094574,
      "learning_rate": 1.971098470180437e-06,
      "loss": 0.0004,
      "step": 135020
    },
    {
      "epoch": 2.209441217377076,
      "grad_norm": 0.01580294594168663,
      "learning_rate": 1.970341131125109e-06,
      "loss": 0.001,
      "step": 135030
    },
    {
      "epoch": 2.209604843328152,
      "grad_norm": 0.01390895526856184,
      "learning_rate": 1.9695839018876356e-06,
      "loss": 0.0005,
      "step": 135040
    },
    {
      "epoch": 2.209768469279228,
      "grad_norm": 0.22131311893463135,
      "learning_rate": 1.9688267824954597e-06,
      "loss": 0.0004,
      "step": 135050
    },
    {
      "epoch": 2.2099320952303034,
      "grad_norm": 0.07881221920251846,
      "learning_rate": 1.96806977297603e-06,
      "loss": 0.0008,
      "step": 135060
    },
    {
      "epoch": 2.2100957211813794,
      "grad_norm": 0.010887800715863705,
      "learning_rate": 1.967312873356784e-06,
      "loss": 0.0003,
      "step": 135070
    },
    {
      "epoch": 2.210259347132455,
      "grad_norm": 0.0820031687617302,
      "learning_rate": 1.966556083665156e-06,
      "loss": 0.0006,
      "step": 135080
    },
    {
      "epoch": 2.210422973083531,
      "grad_norm": 0.21759140491485596,
      "learning_rate": 1.9657994039285806e-06,
      "loss": 0.0008,
      "step": 135090
    },
    {
      "epoch": 2.210586599034607,
      "grad_norm": 0.12937645614147186,
      "learning_rate": 1.9650428341744824e-06,
      "loss": 0.0016,
      "step": 135100
    },
    {
      "epoch": 2.2107502249856825,
      "grad_norm": 0.038025375455617905,
      "learning_rate": 1.9642863744302875e-06,
      "loss": 0.0008,
      "step": 135110
    },
    {
      "epoch": 2.2109138509367585,
      "grad_norm": 0.000715558766387403,
      "learning_rate": 1.963530024723417e-06,
      "loss": 0.001,
      "step": 135120
    },
    {
      "epoch": 2.2110774768878345,
      "grad_norm": 0.003580195363610983,
      "learning_rate": 1.962773785081283e-06,
      "loss": 0.0008,
      "step": 135130
    },
    {
      "epoch": 2.21124110283891,
      "grad_norm": 0.11361295729875565,
      "learning_rate": 1.962017655531302e-06,
      "loss": 0.0004,
      "step": 135140
    },
    {
      "epoch": 2.211404728789986,
      "grad_norm": 0.10403428971767426,
      "learning_rate": 1.9612616361008773e-06,
      "loss": 0.001,
      "step": 135150
    },
    {
      "epoch": 2.211568354741062,
      "grad_norm": 0.11772803217172623,
      "learning_rate": 1.960505726817418e-06,
      "loss": 0.0007,
      "step": 135160
    },
    {
      "epoch": 2.2117319806921376,
      "grad_norm": 0.02060648612678051,
      "learning_rate": 1.959749927708318e-06,
      "loss": 0.0005,
      "step": 135170
    },
    {
      "epoch": 2.2118956066432136,
      "grad_norm": 0.037243980914354324,
      "learning_rate": 1.958994238800976e-06,
      "loss": 0.0015,
      "step": 135180
    },
    {
      "epoch": 2.2120592325942896,
      "grad_norm": 0.058939263224601746,
      "learning_rate": 1.9582386601227853e-06,
      "loss": 0.0009,
      "step": 135190
    },
    {
      "epoch": 2.212222858545365,
      "grad_norm": 0.0410596989095211,
      "learning_rate": 1.957483191701131e-06,
      "loss": 0.0006,
      "step": 135200
    },
    {
      "epoch": 2.212386484496441,
      "grad_norm": 0.08171757310628891,
      "learning_rate": 1.9567278335634e-06,
      "loss": 0.0005,
      "step": 135210
    },
    {
      "epoch": 2.212550110447517,
      "grad_norm": 0.1396220326423645,
      "learning_rate": 1.9559725857369695e-06,
      "loss": 0.0012,
      "step": 135220
    },
    {
      "epoch": 2.2127137363985927,
      "grad_norm": 0.10604510456323624,
      "learning_rate": 1.955217448249218e-06,
      "loss": 0.0006,
      "step": 135230
    },
    {
      "epoch": 2.2128773623496687,
      "grad_norm": 0.0802365094423294,
      "learning_rate": 1.954462421127514e-06,
      "loss": 0.0011,
      "step": 135240
    },
    {
      "epoch": 2.2130409883007447,
      "grad_norm": 0.0205355454236269,
      "learning_rate": 1.9537075043992296e-06,
      "loss": 0.0007,
      "step": 135250
    },
    {
      "epoch": 2.2132046142518202,
      "grad_norm": 0.0017985167214646935,
      "learning_rate": 1.952952698091727e-06,
      "loss": 0.0008,
      "step": 135260
    },
    {
      "epoch": 2.2133682402028962,
      "grad_norm": 0.07845429331064224,
      "learning_rate": 1.9521980022323635e-06,
      "loss": 0.0006,
      "step": 135270
    },
    {
      "epoch": 2.2135318661539722,
      "grad_norm": 0.04080011323094368,
      "learning_rate": 1.951443416848499e-06,
      "loss": 0.0008,
      "step": 135280
    },
    {
      "epoch": 2.213695492105048,
      "grad_norm": 0.007538751699030399,
      "learning_rate": 1.9506889419674817e-06,
      "loss": 0.0008,
      "step": 135290
    },
    {
      "epoch": 2.213859118056124,
      "grad_norm": 0.06548520177602768,
      "learning_rate": 1.9499345776166634e-06,
      "loss": 0.0007,
      "step": 135300
    },
    {
      "epoch": 2.2140227440071993,
      "grad_norm": 0.177206352353096,
      "learning_rate": 1.949180323823385e-06,
      "loss": 0.001,
      "step": 135310
    },
    {
      "epoch": 2.2141863699582753,
      "grad_norm": 0.004906008020043373,
      "learning_rate": 1.948426180614988e-06,
      "loss": 0.0014,
      "step": 135320
    },
    {
      "epoch": 2.2143499959093513,
      "grad_norm": 0.5798944234848022,
      "learning_rate": 1.947672148018806e-06,
      "loss": 0.0015,
      "step": 135330
    },
    {
      "epoch": 2.214513621860427,
      "grad_norm": 0.031253550201654434,
      "learning_rate": 1.9469182260621733e-06,
      "loss": 0.0025,
      "step": 135340
    },
    {
      "epoch": 2.214677247811503,
      "grad_norm": 0.14731977880001068,
      "learning_rate": 1.9461644147724186e-06,
      "loss": 0.0023,
      "step": 135350
    },
    {
      "epoch": 2.214840873762579,
      "grad_norm": 0.05601713806390762,
      "learning_rate": 1.9454107141768636e-06,
      "loss": 0.001,
      "step": 135360
    },
    {
      "epoch": 2.2150044997136544,
      "grad_norm": 0.04318203032016754,
      "learning_rate": 1.9446571243028288e-06,
      "loss": 0.0006,
      "step": 135370
    },
    {
      "epoch": 2.2151681256647304,
      "grad_norm": 0.05038510262966156,
      "learning_rate": 1.943903645177629e-06,
      "loss": 0.0008,
      "step": 135380
    },
    {
      "epoch": 2.2153317516158064,
      "grad_norm": 0.23002773523330688,
      "learning_rate": 1.943150276828576e-06,
      "loss": 0.0011,
      "step": 135390
    },
    {
      "epoch": 2.215495377566882,
      "grad_norm": 0.13838393986225128,
      "learning_rate": 1.9423970192829807e-06,
      "loss": 0.0007,
      "step": 135400
    },
    {
      "epoch": 2.215659003517958,
      "grad_norm": 0.03591505065560341,
      "learning_rate": 1.941643872568143e-06,
      "loss": 0.0004,
      "step": 135410
    },
    {
      "epoch": 2.215822629469034,
      "grad_norm": 0.02029172144830227,
      "learning_rate": 1.9408908367113655e-06,
      "loss": 0.001,
      "step": 135420
    },
    {
      "epoch": 2.2159862554201095,
      "grad_norm": 0.08574935793876648,
      "learning_rate": 1.9401379117399416e-06,
      "loss": 0.0004,
      "step": 135430
    },
    {
      "epoch": 2.2161498813711855,
      "grad_norm": 0.03263247385621071,
      "learning_rate": 1.939385097681166e-06,
      "loss": 0.0006,
      "step": 135440
    },
    {
      "epoch": 2.2163135073222615,
      "grad_norm": 0.2616935968399048,
      "learning_rate": 1.9386323945623227e-06,
      "loss": 0.0014,
      "step": 135450
    },
    {
      "epoch": 2.216477133273337,
      "grad_norm": 0.010783706791698933,
      "learning_rate": 1.937879802410699e-06,
      "loss": 0.0025,
      "step": 135460
    },
    {
      "epoch": 2.216640759224413,
      "grad_norm": 0.15725213289260864,
      "learning_rate": 1.937127321253573e-06,
      "loss": 0.001,
      "step": 135470
    },
    {
      "epoch": 2.2168043851754886,
      "grad_norm": 0.31683480739593506,
      "learning_rate": 1.9363749511182184e-06,
      "loss": 0.0006,
      "step": 135480
    },
    {
      "epoch": 2.2169680111265646,
      "grad_norm": 0.053549326956272125,
      "learning_rate": 1.9356226920319106e-06,
      "loss": 0.0011,
      "step": 135490
    },
    {
      "epoch": 2.2171316370776406,
      "grad_norm": 0.10388994216918945,
      "learning_rate": 1.934870544021913e-06,
      "loss": 0.0011,
      "step": 135500
    },
    {
      "epoch": 2.217295263028716,
      "grad_norm": 0.08462473750114441,
      "learning_rate": 1.9341185071154935e-06,
      "loss": 0.0013,
      "step": 135510
    },
    {
      "epoch": 2.217458888979792,
      "grad_norm": 0.004310683347284794,
      "learning_rate": 1.9333665813399073e-06,
      "loss": 0.0007,
      "step": 135520
    },
    {
      "epoch": 2.217622514930868,
      "grad_norm": 0.03336222469806671,
      "learning_rate": 1.9326147667224136e-06,
      "loss": 0.001,
      "step": 135530
    },
    {
      "epoch": 2.2177861408819437,
      "grad_norm": 0.0442003533244133,
      "learning_rate": 1.931863063290261e-06,
      "loss": 0.0004,
      "step": 135540
    },
    {
      "epoch": 2.2179497668330197,
      "grad_norm": 0.14304393529891968,
      "learning_rate": 1.931111471070698e-06,
      "loss": 0.0013,
      "step": 135550
    },
    {
      "epoch": 2.2181133927840957,
      "grad_norm": 0.12799952924251556,
      "learning_rate": 1.93035999009097e-06,
      "loss": 0.0007,
      "step": 135560
    },
    {
      "epoch": 2.2182770187351712,
      "grad_norm": 0.06458614766597748,
      "learning_rate": 1.9296086203783155e-06,
      "loss": 0.0003,
      "step": 135570
    },
    {
      "epoch": 2.2184406446862472,
      "grad_norm": 0.03923512250185013,
      "learning_rate": 1.9288573619599684e-06,
      "loss": 0.001,
      "step": 135580
    },
    {
      "epoch": 2.2186042706373232,
      "grad_norm": 0.012834126129746437,
      "learning_rate": 1.928106214863159e-06,
      "loss": 0.001,
      "step": 135590
    },
    {
      "epoch": 2.218767896588399,
      "grad_norm": 0.07307049632072449,
      "learning_rate": 1.927355179115118e-06,
      "loss": 0.0016,
      "step": 135600
    },
    {
      "epoch": 2.218931522539475,
      "grad_norm": 0.06783021241426468,
      "learning_rate": 1.9266042547430646e-06,
      "loss": 0.0009,
      "step": 135610
    },
    {
      "epoch": 2.219095148490551,
      "grad_norm": 0.04411931708455086,
      "learning_rate": 1.9258534417742208e-06,
      "loss": 0.0008,
      "step": 135620
    },
    {
      "epoch": 2.2192587744416263,
      "grad_norm": 0.12660111486911774,
      "learning_rate": 1.925102740235803e-06,
      "loss": 0.0009,
      "step": 135630
    },
    {
      "epoch": 2.2194224003927023,
      "grad_norm": 0.03758559376001358,
      "learning_rate": 1.9243521501550185e-06,
      "loss": 0.0008,
      "step": 135640
    },
    {
      "epoch": 2.2195860263437783,
      "grad_norm": 0.048410918563604355,
      "learning_rate": 1.923601671559078e-06,
      "loss": 0.0006,
      "step": 135650
    },
    {
      "epoch": 2.219749652294854,
      "grad_norm": 0.11889099329710007,
      "learning_rate": 1.922851304475182e-06,
      "loss": 0.001,
      "step": 135660
    },
    {
      "epoch": 2.21991327824593,
      "grad_norm": 0.17633835971355438,
      "learning_rate": 1.9221010489305315e-06,
      "loss": 0.0008,
      "step": 135670
    },
    {
      "epoch": 2.220076904197006,
      "grad_norm": 0.030358219519257545,
      "learning_rate": 1.921350904952321e-06,
      "loss": 0.0007,
      "step": 135680
    },
    {
      "epoch": 2.2202405301480814,
      "grad_norm": 0.04289516061544418,
      "learning_rate": 1.9206008725677383e-06,
      "loss": 0.0007,
      "step": 135690
    },
    {
      "epoch": 2.2204041560991574,
      "grad_norm": 0.08639387041330338,
      "learning_rate": 1.9198509518039754e-06,
      "loss": 0.0005,
      "step": 135700
    },
    {
      "epoch": 2.220567782050233,
      "grad_norm": 0.005752048455178738,
      "learning_rate": 1.9191011426882105e-06,
      "loss": 0.0006,
      "step": 135710
    },
    {
      "epoch": 2.220731408001309,
      "grad_norm": 0.04405929520726204,
      "learning_rate": 1.918351445247626e-06,
      "loss": 0.0052,
      "step": 135720
    },
    {
      "epoch": 2.220895033952385,
      "grad_norm": 0.1026475802063942,
      "learning_rate": 1.917601859509394e-06,
      "loss": 0.0008,
      "step": 135730
    },
    {
      "epoch": 2.2210586599034605,
      "grad_norm": 0.01991293951869011,
      "learning_rate": 1.916852385500687e-06,
      "loss": 0.0026,
      "step": 135740
    },
    {
      "epoch": 2.2212222858545365,
      "grad_norm": 0.04852480813860893,
      "learning_rate": 1.916103023248669e-06,
      "loss": 0.002,
      "step": 135750
    },
    {
      "epoch": 2.2213859118056125,
      "grad_norm": 0.03531540557742119,
      "learning_rate": 1.9153537727805073e-06,
      "loss": 0.0012,
      "step": 135760
    },
    {
      "epoch": 2.221549537756688,
      "grad_norm": 0.12029346823692322,
      "learning_rate": 1.914604634123357e-06,
      "loss": 0.0015,
      "step": 135770
    },
    {
      "epoch": 2.221713163707764,
      "grad_norm": 0.05186917632818222,
      "learning_rate": 1.913855607304372e-06,
      "loss": 0.0013,
      "step": 135780
    },
    {
      "epoch": 2.22187678965884,
      "grad_norm": 0.14210397005081177,
      "learning_rate": 1.9131066923507048e-06,
      "loss": 0.001,
      "step": 135790
    },
    {
      "epoch": 2.2220404156099156,
      "grad_norm": 0.0436009056866169,
      "learning_rate": 1.9123578892894995e-06,
      "loss": 0.0005,
      "step": 135800
    },
    {
      "epoch": 2.2222040415609916,
      "grad_norm": 0.006005428731441498,
      "learning_rate": 1.9116091981479014e-06,
      "loss": 0.0007,
      "step": 135810
    },
    {
      "epoch": 2.2223676675120676,
      "grad_norm": 0.04222996160387993,
      "learning_rate": 1.9108606189530453e-06,
      "loss": 0.0005,
      "step": 135820
    },
    {
      "epoch": 2.222531293463143,
      "grad_norm": 0.02853001467883587,
      "learning_rate": 1.9101121517320676e-06,
      "loss": 0.001,
      "step": 135830
    },
    {
      "epoch": 2.222694919414219,
      "grad_norm": 0.027031652629375458,
      "learning_rate": 1.9093637965120994e-06,
      "loss": 0.0006,
      "step": 135840
    },
    {
      "epoch": 2.2228585453652947,
      "grad_norm": 0.04568779841065407,
      "learning_rate": 1.908615553320264e-06,
      "loss": 0.0007,
      "step": 135850
    },
    {
      "epoch": 2.2230221713163707,
      "grad_norm": 0.02065465785562992,
      "learning_rate": 1.9078674221836865e-06,
      "loss": 0.001,
      "step": 135860
    },
    {
      "epoch": 2.2231857972674467,
      "grad_norm": 0.13224564492702484,
      "learning_rate": 1.9071194031294826e-06,
      "loss": 0.001,
      "step": 135870
    },
    {
      "epoch": 2.2233494232185222,
      "grad_norm": 0.12640483677387238,
      "learning_rate": 1.9063714961847656e-06,
      "loss": 0.0011,
      "step": 135880
    },
    {
      "epoch": 2.2235130491695982,
      "grad_norm": 0.060399483889341354,
      "learning_rate": 1.9056237013766481e-06,
      "loss": 0.0006,
      "step": 135890
    },
    {
      "epoch": 2.2236766751206742,
      "grad_norm": 0.005786554887890816,
      "learning_rate": 1.9048760187322329e-06,
      "loss": 0.001,
      "step": 135900
    },
    {
      "epoch": 2.22384030107175,
      "grad_norm": 0.05425728112459183,
      "learning_rate": 1.9041284482786243e-06,
      "loss": 0.001,
      "step": 135910
    },
    {
      "epoch": 2.224003927022826,
      "grad_norm": 0.06975378096103668,
      "learning_rate": 1.9033809900429174e-06,
      "loss": 0.002,
      "step": 135920
    },
    {
      "epoch": 2.224167552973902,
      "grad_norm": 0.1109786257147789,
      "learning_rate": 1.9026336440522086e-06,
      "loss": 0.0009,
      "step": 135930
    },
    {
      "epoch": 2.2243311789249773,
      "grad_norm": 0.3379153609275818,
      "learning_rate": 1.9018864103335844e-06,
      "loss": 0.0015,
      "step": 135940
    },
    {
      "epoch": 2.2244948048760533,
      "grad_norm": 0.01581559143960476,
      "learning_rate": 1.9011392889141333e-06,
      "loss": 0.0003,
      "step": 135950
    },
    {
      "epoch": 2.2246584308271293,
      "grad_norm": 0.17589594423770905,
      "learning_rate": 1.9003922798209334e-06,
      "loss": 0.0019,
      "step": 135960
    },
    {
      "epoch": 2.224822056778205,
      "grad_norm": 0.10207583755254745,
      "learning_rate": 1.8996453830810653e-06,
      "loss": 0.0007,
      "step": 135970
    },
    {
      "epoch": 2.224985682729281,
      "grad_norm": 0.09917905181646347,
      "learning_rate": 1.8988985987216002e-06,
      "loss": 0.0007,
      "step": 135980
    },
    {
      "epoch": 2.225149308680357,
      "grad_norm": 0.31906333565711975,
      "learning_rate": 1.8981519267696063e-06,
      "loss": 0.0011,
      "step": 135990
    },
    {
      "epoch": 2.2253129346314324,
      "grad_norm": 0.06105310842394829,
      "learning_rate": 1.8974053672521515e-06,
      "loss": 0.0015,
      "step": 136000
    },
    {
      "epoch": 2.2254765605825084,
      "grad_norm": 0.10157790780067444,
      "learning_rate": 1.8966589201962938e-06,
      "loss": 0.0011,
      "step": 136010
    },
    {
      "epoch": 2.2256401865335844,
      "grad_norm": 0.05836651846766472,
      "learning_rate": 1.8959125856290928e-06,
      "loss": 0.001,
      "step": 136020
    },
    {
      "epoch": 2.22580381248466,
      "grad_norm": 0.06173566356301308,
      "learning_rate": 1.8951663635775985e-06,
      "loss": 0.0005,
      "step": 136030
    },
    {
      "epoch": 2.225967438435736,
      "grad_norm": 0.040976330637931824,
      "learning_rate": 1.8944202540688634e-06,
      "loss": 0.0006,
      "step": 136040
    },
    {
      "epoch": 2.226131064386812,
      "grad_norm": 0.09951294958591461,
      "learning_rate": 1.8936742571299282e-06,
      "loss": 0.001,
      "step": 136050
    },
    {
      "epoch": 2.2262946903378875,
      "grad_norm": 0.04871763661503792,
      "learning_rate": 1.892928372787835e-06,
      "loss": 0.0009,
      "step": 136060
    },
    {
      "epoch": 2.2264583162889635,
      "grad_norm": 0.19348010420799255,
      "learning_rate": 1.8921826010696242e-06,
      "loss": 0.001,
      "step": 136070
    },
    {
      "epoch": 2.226621942240039,
      "grad_norm": 0.09974869340658188,
      "learning_rate": 1.8914369420023215e-06,
      "loss": 0.0011,
      "step": 136080
    },
    {
      "epoch": 2.226785568191115,
      "grad_norm": 0.21305035054683685,
      "learning_rate": 1.8906913956129596e-06,
      "loss": 0.0011,
      "step": 136090
    },
    {
      "epoch": 2.226949194142191,
      "grad_norm": 0.0825585201382637,
      "learning_rate": 1.8899459619285604e-06,
      "loss": 0.001,
      "step": 136100
    },
    {
      "epoch": 2.2271128200932666,
      "grad_norm": 0.02103789523243904,
      "learning_rate": 1.889200640976145e-06,
      "loss": 0.0007,
      "step": 136110
    },
    {
      "epoch": 2.2272764460443426,
      "grad_norm": 0.09959770739078522,
      "learning_rate": 1.8884554327827314e-06,
      "loss": 0.0009,
      "step": 136120
    },
    {
      "epoch": 2.2274400719954186,
      "grad_norm": 0.07241784036159515,
      "learning_rate": 1.887710337375328e-06,
      "loss": 0.001,
      "step": 136130
    },
    {
      "epoch": 2.227603697946494,
      "grad_norm": 0.06889087706804276,
      "learning_rate": 1.8869653547809468e-06,
      "loss": 0.0016,
      "step": 136140
    },
    {
      "epoch": 2.22776732389757,
      "grad_norm": 0.03271825984120369,
      "learning_rate": 1.886220485026588e-06,
      "loss": 0.001,
      "step": 136150
    },
    {
      "epoch": 2.227930949848646,
      "grad_norm": 0.18051211535930634,
      "learning_rate": 1.8854757281392544e-06,
      "loss": 0.0013,
      "step": 136160
    },
    {
      "epoch": 2.2280945757997217,
      "grad_norm": 0.032112330198287964,
      "learning_rate": 1.8847310841459398e-06,
      "loss": 0.0007,
      "step": 136170
    },
    {
      "epoch": 2.2282582017507977,
      "grad_norm": 0.04270172864198685,
      "learning_rate": 1.8839865530736346e-06,
      "loss": 0.001,
      "step": 136180
    },
    {
      "epoch": 2.2284218277018737,
      "grad_norm": 0.11945780366659164,
      "learning_rate": 1.883242134949329e-06,
      "loss": 0.0009,
      "step": 136190
    },
    {
      "epoch": 2.2285854536529492,
      "grad_norm": 0.0691685602068901,
      "learning_rate": 1.882497829800004e-06,
      "loss": 0.0003,
      "step": 136200
    },
    {
      "epoch": 2.2287490796040252,
      "grad_norm": 0.003541781799867749,
      "learning_rate": 1.8817536376526412e-06,
      "loss": 0.0007,
      "step": 136210
    },
    {
      "epoch": 2.2289127055551012,
      "grad_norm": 0.04141651839017868,
      "learning_rate": 1.8810095585342131e-06,
      "loss": 0.0034,
      "step": 136220
    },
    {
      "epoch": 2.229076331506177,
      "grad_norm": 0.09079122543334961,
      "learning_rate": 1.880265592471694e-06,
      "loss": 0.0018,
      "step": 136230
    },
    {
      "epoch": 2.229239957457253,
      "grad_norm": 0.05387713387608528,
      "learning_rate": 1.8795217394920467e-06,
      "loss": 0.0006,
      "step": 136240
    },
    {
      "epoch": 2.2294035834083283,
      "grad_norm": 0.09211508929729462,
      "learning_rate": 1.878777999622239e-06,
      "loss": 0.001,
      "step": 136250
    },
    {
      "epoch": 2.2295672093594043,
      "grad_norm": 0.07431850582361221,
      "learning_rate": 1.8780343728892247e-06,
      "loss": 0.0007,
      "step": 136260
    },
    {
      "epoch": 2.2297308353104803,
      "grad_norm": 0.033514849841594696,
      "learning_rate": 1.8772908593199635e-06,
      "loss": 0.001,
      "step": 136270
    },
    {
      "epoch": 2.229894461261556,
      "grad_norm": 0.13017304241657257,
      "learning_rate": 1.876547458941403e-06,
      "loss": 0.0006,
      "step": 136280
    },
    {
      "epoch": 2.230058087212632,
      "grad_norm": 0.07834970951080322,
      "learning_rate": 1.8758041717804882e-06,
      "loss": 0.001,
      "step": 136290
    },
    {
      "epoch": 2.230221713163708,
      "grad_norm": 0.007600969169288874,
      "learning_rate": 1.8750609978641647e-06,
      "loss": 0.0005,
      "step": 136300
    },
    {
      "epoch": 2.2303853391147834,
      "grad_norm": 0.13221855461597443,
      "learning_rate": 1.8743179372193688e-06,
      "loss": 0.0008,
      "step": 136310
    },
    {
      "epoch": 2.2305489650658594,
      "grad_norm": 0.21034346520900726,
      "learning_rate": 1.8735749898730343e-06,
      "loss": 0.0006,
      "step": 136320
    },
    {
      "epoch": 2.2307125910169354,
      "grad_norm": 0.012454049661755562,
      "learning_rate": 1.8728321558520946e-06,
      "loss": 0.0009,
      "step": 136330
    },
    {
      "epoch": 2.230876216968011,
      "grad_norm": 0.10688856989145279,
      "learning_rate": 1.8720894351834713e-06,
      "loss": 0.0007,
      "step": 136340
    },
    {
      "epoch": 2.231039842919087,
      "grad_norm": 0.22139160335063934,
      "learning_rate": 1.87134682789409e-06,
      "loss": 0.0015,
      "step": 136350
    },
    {
      "epoch": 2.231203468870163,
      "grad_norm": 0.04291486367583275,
      "learning_rate": 1.8706043340108649e-06,
      "loss": 0.0013,
      "step": 136360
    },
    {
      "epoch": 2.2313670948212385,
      "grad_norm": 0.2205728441476822,
      "learning_rate": 1.8698619535607149e-06,
      "loss": 0.0008,
      "step": 136370
    },
    {
      "epoch": 2.2315307207723145,
      "grad_norm": 0.027403313666582108,
      "learning_rate": 1.869119686570543e-06,
      "loss": 0.0008,
      "step": 136380
    },
    {
      "epoch": 2.2316943467233905,
      "grad_norm": 0.15601393580436707,
      "learning_rate": 1.868377533067257e-06,
      "loss": 0.0006,
      "step": 136390
    },
    {
      "epoch": 2.231857972674466,
      "grad_norm": 0.04582241177558899,
      "learning_rate": 1.8676354930777606e-06,
      "loss": 0.0005,
      "step": 136400
    },
    {
      "epoch": 2.232021598625542,
      "grad_norm": 0.044447582215070724,
      "learning_rate": 1.8668935666289472e-06,
      "loss": 0.0009,
      "step": 136410
    },
    {
      "epoch": 2.232185224576618,
      "grad_norm": 0.07942578941583633,
      "learning_rate": 1.8661517537477141e-06,
      "loss": 0.0011,
      "step": 136420
    },
    {
      "epoch": 2.2323488505276936,
      "grad_norm": 0.0710284560918808,
      "learning_rate": 1.865410054460946e-06,
      "loss": 0.0005,
      "step": 136430
    },
    {
      "epoch": 2.2325124764787696,
      "grad_norm": 0.09684465825557709,
      "learning_rate": 1.8646684687955313e-06,
      "loss": 0.0005,
      "step": 136440
    },
    {
      "epoch": 2.2326761024298456,
      "grad_norm": 0.024918200448155403,
      "learning_rate": 1.8639269967783468e-06,
      "loss": 0.0012,
      "step": 136450
    },
    {
      "epoch": 2.232839728380921,
      "grad_norm": 0.2589452564716339,
      "learning_rate": 1.8631856384362734e-06,
      "loss": 0.0012,
      "step": 136460
    },
    {
      "epoch": 2.233003354331997,
      "grad_norm": 0.05960176885128021,
      "learning_rate": 1.8624443937961817e-06,
      "loss": 0.0007,
      "step": 136470
    },
    {
      "epoch": 2.2331669802830727,
      "grad_norm": 0.03820612281560898,
      "learning_rate": 1.861703262884938e-06,
      "loss": 0.0011,
      "step": 136480
    },
    {
      "epoch": 2.2333306062341487,
      "grad_norm": 0.048957861959934235,
      "learning_rate": 1.8609622457294101e-06,
      "loss": 0.0007,
      "step": 136490
    },
    {
      "epoch": 2.2334942321852247,
      "grad_norm": 0.1657371073961258,
      "learning_rate": 1.8602213423564547e-06,
      "loss": 0.001,
      "step": 136500
    },
    {
      "epoch": 2.2336578581363002,
      "grad_norm": 0.030870649963617325,
      "learning_rate": 1.8594805527929305e-06,
      "loss": 0.001,
      "step": 136510
    },
    {
      "epoch": 2.2338214840873762,
      "grad_norm": 0.10805943608283997,
      "learning_rate": 1.8587398770656868e-06,
      "loss": 0.0006,
      "step": 136520
    },
    {
      "epoch": 2.2339851100384522,
      "grad_norm": 0.0062743667513132095,
      "learning_rate": 1.8579993152015746e-06,
      "loss": 0.0004,
      "step": 136530
    },
    {
      "epoch": 2.234148735989528,
      "grad_norm": 0.0220036618411541,
      "learning_rate": 1.8572588672274338e-06,
      "loss": 0.0003,
      "step": 136540
    },
    {
      "epoch": 2.234312361940604,
      "grad_norm": 0.19887731969356537,
      "learning_rate": 1.8565185331701057e-06,
      "loss": 0.0006,
      "step": 136550
    },
    {
      "epoch": 2.23447598789168,
      "grad_norm": 0.039455074816942215,
      "learning_rate": 1.8557783130564277e-06,
      "loss": 0.0012,
      "step": 136560
    },
    {
      "epoch": 2.2346396138427553,
      "grad_norm": 0.015283849090337753,
      "learning_rate": 1.8550382069132283e-06,
      "loss": 0.0005,
      "step": 136570
    },
    {
      "epoch": 2.2348032397938313,
      "grad_norm": 0.004910776391625404,
      "learning_rate": 1.8542982147673355e-06,
      "loss": 0.0004,
      "step": 136580
    },
    {
      "epoch": 2.2349668657449073,
      "grad_norm": 0.007749320473521948,
      "learning_rate": 1.85355833664557e-06,
      "loss": 0.0014,
      "step": 136590
    },
    {
      "epoch": 2.235130491695983,
      "grad_norm": 0.00988097582012415,
      "learning_rate": 1.8528185725747528e-06,
      "loss": 0.0013,
      "step": 136600
    },
    {
      "epoch": 2.235294117647059,
      "grad_norm": 0.11493630707263947,
      "learning_rate": 1.8520789225816999e-06,
      "loss": 0.0006,
      "step": 136610
    },
    {
      "epoch": 2.235457743598135,
      "grad_norm": 0.04362785443663597,
      "learning_rate": 1.8513393866932184e-06,
      "loss": 0.0007,
      "step": 136620
    },
    {
      "epoch": 2.2356213695492104,
      "grad_norm": 0.05324506387114525,
      "learning_rate": 1.8505999649361184e-06,
      "loss": 0.0012,
      "step": 136630
    },
    {
      "epoch": 2.2357849955002864,
      "grad_norm": 0.011651734821498394,
      "learning_rate": 1.8498606573371985e-06,
      "loss": 0.0007,
      "step": 136640
    },
    {
      "epoch": 2.235948621451362,
      "grad_norm": 0.047895416617393494,
      "learning_rate": 1.8491214639232602e-06,
      "loss": 0.001,
      "step": 136650
    },
    {
      "epoch": 2.236112247402438,
      "grad_norm": 0.018472813069820404,
      "learning_rate": 1.8483823847210946e-06,
      "loss": 0.0007,
      "step": 136660
    },
    {
      "epoch": 2.236275873353514,
      "grad_norm": 0.040261320769786835,
      "learning_rate": 1.8476434197574943e-06,
      "loss": 0.0005,
      "step": 136670
    },
    {
      "epoch": 2.2364394993045895,
      "grad_norm": 0.006628134753555059,
      "learning_rate": 1.8469045690592436e-06,
      "loss": 0.0005,
      "step": 136680
    },
    {
      "epoch": 2.2366031252556655,
      "grad_norm": 0.03279632329940796,
      "learning_rate": 1.8461658326531218e-06,
      "loss": 0.001,
      "step": 136690
    },
    {
      "epoch": 2.2367667512067415,
      "grad_norm": 0.06808466464281082,
      "learning_rate": 1.845427210565911e-06,
      "loss": 0.0005,
      "step": 136700
    },
    {
      "epoch": 2.236930377157817,
      "grad_norm": 0.13237743079662323,
      "learning_rate": 1.8446887028243799e-06,
      "loss": 0.0011,
      "step": 136710
    },
    {
      "epoch": 2.237094003108893,
      "grad_norm": 0.04189366474747658,
      "learning_rate": 1.8439503094553013e-06,
      "loss": 0.0006,
      "step": 136720
    },
    {
      "epoch": 2.237257629059969,
      "grad_norm": 0.15231986343860626,
      "learning_rate": 1.8432120304854373e-06,
      "loss": 0.0011,
      "step": 136730
    },
    {
      "epoch": 2.2374212550110446,
      "grad_norm": 0.05270028114318848,
      "learning_rate": 1.8424738659415513e-06,
      "loss": 0.0006,
      "step": 136740
    },
    {
      "epoch": 2.2375848809621206,
      "grad_norm": 0.22483132779598236,
      "learning_rate": 1.8417358158503972e-06,
      "loss": 0.0011,
      "step": 136750
    },
    {
      "epoch": 2.2377485069131966,
      "grad_norm": 0.2633575201034546,
      "learning_rate": 1.8409978802387296e-06,
      "loss": 0.0017,
      "step": 136760
    },
    {
      "epoch": 2.237912132864272,
      "grad_norm": 0.15843403339385986,
      "learning_rate": 1.840260059133299e-06,
      "loss": 0.001,
      "step": 136770
    },
    {
      "epoch": 2.238075758815348,
      "grad_norm": 0.11026959121227264,
      "learning_rate": 1.8395223525608441e-06,
      "loss": 0.0016,
      "step": 136780
    },
    {
      "epoch": 2.238239384766424,
      "grad_norm": 0.14783233404159546,
      "learning_rate": 1.8387847605481096e-06,
      "loss": 0.0013,
      "step": 136790
    },
    {
      "epoch": 2.2384030107174997,
      "grad_norm": 0.08157098293304443,
      "learning_rate": 1.8380472831218276e-06,
      "loss": 0.0011,
      "step": 136800
    },
    {
      "epoch": 2.2385666366685757,
      "grad_norm": 0.05505836382508278,
      "learning_rate": 1.8373099203087341e-06,
      "loss": 0.0009,
      "step": 136810
    },
    {
      "epoch": 2.2387302626196517,
      "grad_norm": 0.03696466609835625,
      "learning_rate": 1.8365726721355531e-06,
      "loss": 0.0023,
      "step": 136820
    },
    {
      "epoch": 2.2388938885707272,
      "grad_norm": 0.02781374379992485,
      "learning_rate": 1.8358355386290094e-06,
      "loss": 0.0009,
      "step": 136830
    },
    {
      "epoch": 2.2390575145218032,
      "grad_norm": 0.12648823857307434,
      "learning_rate": 1.8350985198158245e-06,
      "loss": 0.0009,
      "step": 136840
    },
    {
      "epoch": 2.239221140472879,
      "grad_norm": 0.036604978144168854,
      "learning_rate": 1.8343616157227096e-06,
      "loss": 0.0005,
      "step": 136850
    },
    {
      "epoch": 2.239384766423955,
      "grad_norm": 0.1295877993106842,
      "learning_rate": 1.8336248263763795e-06,
      "loss": 0.0008,
      "step": 136860
    },
    {
      "epoch": 2.239548392375031,
      "grad_norm": 0.10521502047777176,
      "learning_rate": 1.8328881518035397e-06,
      "loss": 0.0007,
      "step": 136870
    },
    {
      "epoch": 2.2397120183261063,
      "grad_norm": 0.058176614344120026,
      "learning_rate": 1.8321515920308902e-06,
      "loss": 0.0006,
      "step": 136880
    },
    {
      "epoch": 2.2398756442771823,
      "grad_norm": 0.06203823164105415,
      "learning_rate": 1.8314151470851337e-06,
      "loss": 0.0008,
      "step": 136890
    },
    {
      "epoch": 2.2400392702282583,
      "grad_norm": 0.159650519490242,
      "learning_rate": 1.8306788169929608e-06,
      "loss": 0.0007,
      "step": 136900
    },
    {
      "epoch": 2.240202896179334,
      "grad_norm": 0.10077209025621414,
      "learning_rate": 1.829942601781065e-06,
      "loss": 0.0007,
      "step": 136910
    },
    {
      "epoch": 2.24036652213041,
      "grad_norm": 0.03476560860872269,
      "learning_rate": 1.8292065014761295e-06,
      "loss": 0.0004,
      "step": 136920
    },
    {
      "epoch": 2.240530148081486,
      "grad_norm": 0.047771159559488297,
      "learning_rate": 1.8284705161048393e-06,
      "loss": 0.0009,
      "step": 136930
    },
    {
      "epoch": 2.2406937740325614,
      "grad_norm": 0.18116824328899384,
      "learning_rate": 1.8277346456938683e-06,
      "loss": 0.0007,
      "step": 136940
    },
    {
      "epoch": 2.2408573999836374,
      "grad_norm": 0.04550543054938316,
      "learning_rate": 1.8269988902698944e-06,
      "loss": 0.0009,
      "step": 136950
    },
    {
      "epoch": 2.2410210259347134,
      "grad_norm": 0.03038889914751053,
      "learning_rate": 1.8262632498595824e-06,
      "loss": 0.0006,
      "step": 136960
    },
    {
      "epoch": 2.241184651885789,
      "grad_norm": 0.17002473771572113,
      "learning_rate": 1.8255277244896013e-06,
      "loss": 0.0008,
      "step": 136970
    },
    {
      "epoch": 2.241348277836865,
      "grad_norm": 0.06227805092930794,
      "learning_rate": 1.8247923141866108e-06,
      "loss": 0.0047,
      "step": 136980
    },
    {
      "epoch": 2.241511903787941,
      "grad_norm": 0.12031940370798111,
      "learning_rate": 1.8240570189772655e-06,
      "loss": 0.0012,
      "step": 136990
    },
    {
      "epoch": 2.2416755297390165,
      "grad_norm": 0.12806041538715363,
      "learning_rate": 1.8233218388882218e-06,
      "loss": 0.0005,
      "step": 137000
    },
    {
      "epoch": 2.2418391556900925,
      "grad_norm": 0.028910886496305466,
      "learning_rate": 1.8225867739461251e-06,
      "loss": 0.0007,
      "step": 137010
    },
    {
      "epoch": 2.242002781641168,
      "grad_norm": 0.10452350974082947,
      "learning_rate": 1.8218518241776228e-06,
      "loss": 0.0005,
      "step": 137020
    },
    {
      "epoch": 2.242166407592244,
      "grad_norm": 0.12235300242900848,
      "learning_rate": 1.8211169896093516e-06,
      "loss": 0.0004,
      "step": 137030
    },
    {
      "epoch": 2.24233003354332,
      "grad_norm": 0.11636658757925034,
      "learning_rate": 1.8203822702679492e-06,
      "loss": 0.0016,
      "step": 137040
    },
    {
      "epoch": 2.2424936594943956,
      "grad_norm": 0.04234936460852623,
      "learning_rate": 1.8196476661800494e-06,
      "loss": 0.0005,
      "step": 137050
    },
    {
      "epoch": 2.2426572854454716,
      "grad_norm": 0.09420938044786453,
      "learning_rate": 1.8189131773722762e-06,
      "loss": 0.0006,
      "step": 137060
    },
    {
      "epoch": 2.2428209113965476,
      "grad_norm": 0.01809314824640751,
      "learning_rate": 1.8181788038712578e-06,
      "loss": 0.0006,
      "step": 137070
    },
    {
      "epoch": 2.242984537347623,
      "grad_norm": 0.009729955345392227,
      "learning_rate": 1.8174445457036077e-06,
      "loss": 0.0007,
      "step": 137080
    },
    {
      "epoch": 2.243148163298699,
      "grad_norm": 0.09742939472198486,
      "learning_rate": 1.8167104028959431e-06,
      "loss": 0.0007,
      "step": 137090
    },
    {
      "epoch": 2.243311789249775,
      "grad_norm": 0.06619960069656372,
      "learning_rate": 1.8159763754748776e-06,
      "loss": 0.0007,
      "step": 137100
    },
    {
      "epoch": 2.2434754152008507,
      "grad_norm": 0.052535660564899445,
      "learning_rate": 1.8152424634670141e-06,
      "loss": 0.0009,
      "step": 137110
    },
    {
      "epoch": 2.2436390411519267,
      "grad_norm": 0.08759599179029465,
      "learning_rate": 1.8145086668989587e-06,
      "loss": 0.0006,
      "step": 137120
    },
    {
      "epoch": 2.2438026671030027,
      "grad_norm": 0.1907338798046112,
      "learning_rate": 1.8137749857973063e-06,
      "loss": 0.0006,
      "step": 137130
    },
    {
      "epoch": 2.2439662930540782,
      "grad_norm": 0.05970700457692146,
      "learning_rate": 1.8130414201886543e-06,
      "loss": 0.0008,
      "step": 137140
    },
    {
      "epoch": 2.2441299190051542,
      "grad_norm": 0.08070128411054611,
      "learning_rate": 1.81230797009959e-06,
      "loss": 0.0011,
      "step": 137150
    },
    {
      "epoch": 2.2442935449562302,
      "grad_norm": 0.05773957073688507,
      "learning_rate": 1.8115746355567015e-06,
      "loss": 0.0009,
      "step": 137160
    },
    {
      "epoch": 2.244457170907306,
      "grad_norm": 0.10264132171869278,
      "learning_rate": 1.8108414165865695e-06,
      "loss": 0.0008,
      "step": 137170
    },
    {
      "epoch": 2.244620796858382,
      "grad_norm": 0.06791161000728607,
      "learning_rate": 1.8101083132157694e-06,
      "loss": 0.0004,
      "step": 137180
    },
    {
      "epoch": 2.244784422809458,
      "grad_norm": 0.14877314865589142,
      "learning_rate": 1.8093753254708778e-06,
      "loss": 0.0023,
      "step": 137190
    },
    {
      "epoch": 2.2449480487605333,
      "grad_norm": 0.03966068476438522,
      "learning_rate": 1.8086424533784608e-06,
      "loss": 0.0007,
      "step": 137200
    },
    {
      "epoch": 2.2451116747116093,
      "grad_norm": 0.14917223155498505,
      "learning_rate": 1.8079096969650866e-06,
      "loss": 0.0009,
      "step": 137210
    },
    {
      "epoch": 2.2452753006626853,
      "grad_norm": 0.07329724729061127,
      "learning_rate": 1.8071770562573115e-06,
      "loss": 0.0005,
      "step": 137220
    },
    {
      "epoch": 2.245438926613761,
      "grad_norm": 0.004673945251852274,
      "learning_rate": 1.8064445312816963e-06,
      "loss": 0.0011,
      "step": 137230
    },
    {
      "epoch": 2.245602552564837,
      "grad_norm": 0.0570252500474453,
      "learning_rate": 1.80571212206479e-06,
      "loss": 0.0006,
      "step": 137240
    },
    {
      "epoch": 2.2457661785159124,
      "grad_norm": 0.17214785516262054,
      "learning_rate": 1.8049798286331416e-06,
      "loss": 0.0008,
      "step": 137250
    },
    {
      "epoch": 2.2459298044669884,
      "grad_norm": 0.12331068515777588,
      "learning_rate": 1.8042476510132973e-06,
      "loss": 0.0014,
      "step": 137260
    },
    {
      "epoch": 2.2460934304180644,
      "grad_norm": 0.060974083840847015,
      "learning_rate": 1.8035155892317947e-06,
      "loss": 0.0015,
      "step": 137270
    },
    {
      "epoch": 2.24625705636914,
      "grad_norm": 0.16413341462612152,
      "learning_rate": 1.8027836433151697e-06,
      "loss": 0.0011,
      "step": 137280
    },
    {
      "epoch": 2.246420682320216,
      "grad_norm": 0.03981379047036171,
      "learning_rate": 1.8020518132899517e-06,
      "loss": 0.0012,
      "step": 137290
    },
    {
      "epoch": 2.246584308271292,
      "grad_norm": 0.08881141990423203,
      "learning_rate": 1.8013200991826707e-06,
      "loss": 0.001,
      "step": 137300
    },
    {
      "epoch": 2.2467479342223675,
      "grad_norm": 0.1925244927406311,
      "learning_rate": 1.8005885010198465e-06,
      "loss": 0.0016,
      "step": 137310
    },
    {
      "epoch": 2.2469115601734435,
      "grad_norm": 0.05137971043586731,
      "learning_rate": 1.7998570188280002e-06,
      "loss": 0.0008,
      "step": 137320
    },
    {
      "epoch": 2.2470751861245195,
      "grad_norm": 0.02664925530552864,
      "learning_rate": 1.7991256526336464e-06,
      "loss": 0.0006,
      "step": 137330
    },
    {
      "epoch": 2.247238812075595,
      "grad_norm": 0.10559311509132385,
      "learning_rate": 1.7983944024632938e-06,
      "loss": 0.0011,
      "step": 137340
    },
    {
      "epoch": 2.247402438026671,
      "grad_norm": 0.09940566122531891,
      "learning_rate": 1.7976632683434504e-06,
      "loss": 0.0007,
      "step": 137350
    },
    {
      "epoch": 2.247566063977747,
      "grad_norm": 0.058757755905389786,
      "learning_rate": 1.7969322503006154e-06,
      "loss": 0.0007,
      "step": 137360
    },
    {
      "epoch": 2.2477296899288226,
      "grad_norm": 0.04622402414679527,
      "learning_rate": 1.7962013483612895e-06,
      "loss": 0.0004,
      "step": 137370
    },
    {
      "epoch": 2.2478933158798986,
      "grad_norm": 0.1606961339712143,
      "learning_rate": 1.7954705625519647e-06,
      "loss": 0.0006,
      "step": 137380
    },
    {
      "epoch": 2.2480569418309746,
      "grad_norm": 0.10025844722986221,
      "learning_rate": 1.7947398928991282e-06,
      "loss": 0.0012,
      "step": 137390
    },
    {
      "epoch": 2.24822056778205,
      "grad_norm": 0.03751726821064949,
      "learning_rate": 1.794009339429269e-06,
      "loss": 0.0007,
      "step": 137400
    },
    {
      "epoch": 2.248384193733126,
      "grad_norm": 0.02809935249388218,
      "learning_rate": 1.7932789021688647e-06,
      "loss": 0.0007,
      "step": 137410
    },
    {
      "epoch": 2.2485478196842017,
      "grad_norm": 0.13995084166526794,
      "learning_rate": 1.7925485811443938e-06,
      "loss": 0.0011,
      "step": 137420
    },
    {
      "epoch": 2.2487114456352777,
      "grad_norm": 0.1931743174791336,
      "learning_rate": 1.7918183763823266e-06,
      "loss": 0.0008,
      "step": 137430
    },
    {
      "epoch": 2.2488750715863537,
      "grad_norm": 0.11885688453912735,
      "learning_rate": 1.791088287909134e-06,
      "loss": 0.0006,
      "step": 137440
    },
    {
      "epoch": 2.2490386975374292,
      "grad_norm": 0.028767570853233337,
      "learning_rate": 1.790358315751277e-06,
      "loss": 0.0008,
      "step": 137450
    },
    {
      "epoch": 2.2492023234885052,
      "grad_norm": 0.13071975111961365,
      "learning_rate": 1.7896284599352188e-06,
      "loss": 0.0007,
      "step": 137460
    },
    {
      "epoch": 2.2493659494395812,
      "grad_norm": 0.03357565030455589,
      "learning_rate": 1.7888987204874126e-06,
      "loss": 0.0006,
      "step": 137470
    },
    {
      "epoch": 2.249529575390657,
      "grad_norm": 0.06460310518741608,
      "learning_rate": 1.7881690974343086e-06,
      "loss": 0.0007,
      "step": 137480
    },
    {
      "epoch": 2.249693201341733,
      "grad_norm": 0.002815136918798089,
      "learning_rate": 1.7874395908023573e-06,
      "loss": 0.0015,
      "step": 137490
    },
    {
      "epoch": 2.249856827292809,
      "grad_norm": 0.09775934368371964,
      "learning_rate": 1.7867102006179977e-06,
      "loss": 0.0008,
      "step": 137500
    },
    {
      "epoch": 2.2500204532438843,
      "grad_norm": 0.1800021231174469,
      "learning_rate": 1.7859809269076722e-06,
      "loss": 0.0023,
      "step": 137510
    },
    {
      "epoch": 2.2501840791949603,
      "grad_norm": 0.04702913388609886,
      "learning_rate": 1.7852517696978117e-06,
      "loss": 0.0016,
      "step": 137520
    },
    {
      "epoch": 2.2503477051460363,
      "grad_norm": 0.31794503331184387,
      "learning_rate": 1.784522729014848e-06,
      "loss": 0.0007,
      "step": 137530
    },
    {
      "epoch": 2.250511331097112,
      "grad_norm": 0.005553154274821281,
      "learning_rate": 1.7837938048852089e-06,
      "loss": 0.0003,
      "step": 137540
    },
    {
      "epoch": 2.250674957048188,
      "grad_norm": 0.033034540712833405,
      "learning_rate": 1.7830649973353131e-06,
      "loss": 0.0013,
      "step": 137550
    },
    {
      "epoch": 2.250838582999264,
      "grad_norm": 0.2753106951713562,
      "learning_rate": 1.7823363063915804e-06,
      "loss": 0.0017,
      "step": 137560
    },
    {
      "epoch": 2.2510022089503394,
      "grad_norm": 0.05271976441144943,
      "learning_rate": 1.781607732080422e-06,
      "loss": 0.0006,
      "step": 137570
    },
    {
      "epoch": 2.2511658349014154,
      "grad_norm": 0.041430339217185974,
      "learning_rate": 1.7808792744282516e-06,
      "loss": 0.0009,
      "step": 137580
    },
    {
      "epoch": 2.2513294608524914,
      "grad_norm": 0.09326998889446259,
      "learning_rate": 1.7801509334614668e-06,
      "loss": 0.0018,
      "step": 137590
    },
    {
      "epoch": 2.251493086803567,
      "grad_norm": 0.6315110325813293,
      "learning_rate": 1.779422709206472e-06,
      "loss": 0.0009,
      "step": 137600
    },
    {
      "epoch": 2.251656712754643,
      "grad_norm": 0.03473074361681938,
      "learning_rate": 1.7786946016896655e-06,
      "loss": 0.0007,
      "step": 137610
    },
    {
      "epoch": 2.251820338705719,
      "grad_norm": 0.12610793113708496,
      "learning_rate": 1.7779666109374355e-06,
      "loss": 0.0009,
      "step": 137620
    },
    {
      "epoch": 2.2519839646567945,
      "grad_norm": 0.11989783495664597,
      "learning_rate": 1.7772387369761735e-06,
      "loss": 0.0007,
      "step": 137630
    },
    {
      "epoch": 2.2521475906078705,
      "grad_norm": 0.09480896592140198,
      "learning_rate": 1.7765109798322604e-06,
      "loss": 0.0008,
      "step": 137640
    },
    {
      "epoch": 2.252311216558946,
      "grad_norm": 0.023414844647049904,
      "learning_rate": 1.7757833395320784e-06,
      "loss": 0.0011,
      "step": 137650
    },
    {
      "epoch": 2.252474842510022,
      "grad_norm": 0.08589056134223938,
      "learning_rate": 1.7750558161019986e-06,
      "loss": 0.0007,
      "step": 137660
    },
    {
      "epoch": 2.252638468461098,
      "grad_norm": 0.14193852245807648,
      "learning_rate": 1.7743284095683972e-06,
      "loss": 0.0011,
      "step": 137670
    },
    {
      "epoch": 2.2528020944121736,
      "grad_norm": 0.0688340961933136,
      "learning_rate": 1.773601119957638e-06,
      "loss": 0.0009,
      "step": 137680
    },
    {
      "epoch": 2.2529657203632496,
      "grad_norm": 0.10896433144807816,
      "learning_rate": 1.772873947296082e-06,
      "loss": 0.0012,
      "step": 137690
    },
    {
      "epoch": 2.2531293463143256,
      "grad_norm": 0.04505003243684769,
      "learning_rate": 1.7721468916100908e-06,
      "loss": 0.0007,
      "step": 137700
    },
    {
      "epoch": 2.253292972265401,
      "grad_norm": 0.17665976285934448,
      "learning_rate": 1.771419952926015e-06,
      "loss": 0.0016,
      "step": 137710
    },
    {
      "epoch": 2.253456598216477,
      "grad_norm": 0.14921338856220245,
      "learning_rate": 1.7706931312702086e-06,
      "loss": 0.0009,
      "step": 137720
    },
    {
      "epoch": 2.253620224167553,
      "grad_norm": 0.16425618529319763,
      "learning_rate": 1.7699664266690124e-06,
      "loss": 0.001,
      "step": 137730
    },
    {
      "epoch": 2.2537838501186287,
      "grad_norm": 0.06294117867946625,
      "learning_rate": 1.769239839148772e-06,
      "loss": 0.0008,
      "step": 137740
    },
    {
      "epoch": 2.2539474760697047,
      "grad_norm": 0.03120293840765953,
      "learning_rate": 1.7685133687358208e-06,
      "loss": 0.0008,
      "step": 137750
    },
    {
      "epoch": 2.2541111020207807,
      "grad_norm": 0.07448040693998337,
      "learning_rate": 1.7677870154564936e-06,
      "loss": 0.0007,
      "step": 137760
    },
    {
      "epoch": 2.2542747279718562,
      "grad_norm": 0.12275935709476471,
      "learning_rate": 1.7670607793371197e-06,
      "loss": 0.0008,
      "step": 137770
    },
    {
      "epoch": 2.2544383539229322,
      "grad_norm": 0.04093034192919731,
      "learning_rate": 1.7663346604040227e-06,
      "loss": 0.0009,
      "step": 137780
    },
    {
      "epoch": 2.254601979874008,
      "grad_norm": 0.13303150236606598,
      "learning_rate": 1.7656086586835225e-06,
      "loss": 0.0008,
      "step": 137790
    },
    {
      "epoch": 2.254765605825084,
      "grad_norm": 0.0024219504557549953,
      "learning_rate": 1.764882774201933e-06,
      "loss": 0.0004,
      "step": 137800
    },
    {
      "epoch": 2.25492923177616,
      "grad_norm": 0.23065805435180664,
      "learning_rate": 1.7641570069855674e-06,
      "loss": 0.0013,
      "step": 137810
    },
    {
      "epoch": 2.2550928577272353,
      "grad_norm": 0.057689767330884933,
      "learning_rate": 1.7634313570607349e-06,
      "loss": 0.0008,
      "step": 137820
    },
    {
      "epoch": 2.2552564836783113,
      "grad_norm": 0.1388271450996399,
      "learning_rate": 1.7627058244537355e-06,
      "loss": 0.001,
      "step": 137830
    },
    {
      "epoch": 2.2554201096293873,
      "grad_norm": 0.05789140984416008,
      "learning_rate": 1.7619804091908704e-06,
      "loss": 0.0008,
      "step": 137840
    },
    {
      "epoch": 2.255583735580463,
      "grad_norm": 0.10220491141080856,
      "learning_rate": 1.7612551112984316e-06,
      "loss": 0.0005,
      "step": 137850
    },
    {
      "epoch": 2.255747361531539,
      "grad_norm": 0.35700955986976624,
      "learning_rate": 1.7605299308027129e-06,
      "loss": 0.0015,
      "step": 137860
    },
    {
      "epoch": 2.255910987482615,
      "grad_norm": 0.13218556344509125,
      "learning_rate": 1.7598048677299961e-06,
      "loss": 0.0043,
      "step": 137870
    },
    {
      "epoch": 2.2560746134336904,
      "grad_norm": 0.1223706528544426,
      "learning_rate": 1.7590799221065674e-06,
      "loss": 0.0015,
      "step": 137880
    },
    {
      "epoch": 2.2562382393847664,
      "grad_norm": 0.04382982477545738,
      "learning_rate": 1.7583550939587018e-06,
      "loss": 0.0007,
      "step": 137890
    },
    {
      "epoch": 2.2564018653358424,
      "grad_norm": 0.3298861086368561,
      "learning_rate": 1.7576303833126717e-06,
      "loss": 0.0014,
      "step": 137900
    },
    {
      "epoch": 2.256565491286918,
      "grad_norm": 0.029039116576313972,
      "learning_rate": 1.7569057901947489e-06,
      "loss": 0.0014,
      "step": 137910
    },
    {
      "epoch": 2.256729117237994,
      "grad_norm": 0.021003151312470436,
      "learning_rate": 1.7561813146311952e-06,
      "loss": 0.001,
      "step": 137920
    },
    {
      "epoch": 2.25689274318907,
      "grad_norm": 0.04633378982543945,
      "learning_rate": 1.7554569566482738e-06,
      "loss": 0.0005,
      "step": 137930
    },
    {
      "epoch": 2.2570563691401455,
      "grad_norm": 0.07034675031900406,
      "learning_rate": 1.7547327162722383e-06,
      "loss": 0.001,
      "step": 137940
    },
    {
      "epoch": 2.2572199950912215,
      "grad_norm": 0.04566957429051399,
      "learning_rate": 1.7540085935293433e-06,
      "loss": 0.001,
      "step": 137950
    },
    {
      "epoch": 2.2573836210422975,
      "grad_norm": 0.03136362135410309,
      "learning_rate": 1.7532845884458339e-06,
      "loss": 0.0013,
      "step": 137960
    },
    {
      "epoch": 2.257547246993373,
      "grad_norm": 0.09383810311555862,
      "learning_rate": 1.7525607010479546e-06,
      "loss": 0.0014,
      "step": 137970
    },
    {
      "epoch": 2.257710872944449,
      "grad_norm": 0.11463681608438492,
      "learning_rate": 1.7518369313619488e-06,
      "loss": 0.0008,
      "step": 137980
    },
    {
      "epoch": 2.257874498895525,
      "grad_norm": 0.08502574265003204,
      "learning_rate": 1.7511132794140433e-06,
      "loss": 0.0008,
      "step": 137990
    },
    {
      "epoch": 2.2580381248466006,
      "grad_norm": 0.09654149413108826,
      "learning_rate": 1.7503897452304753e-06,
      "loss": 0.0008,
      "step": 138000
    },
    {
      "epoch": 2.2582017507976766,
      "grad_norm": 0.08157924562692642,
      "learning_rate": 1.7496663288374665e-06,
      "loss": 0.001,
      "step": 138010
    },
    {
      "epoch": 2.2583653767487526,
      "grad_norm": 0.056903015822172165,
      "learning_rate": 1.7489430302612437e-06,
      "loss": 0.001,
      "step": 138020
    },
    {
      "epoch": 2.258529002699828,
      "grad_norm": 0.05095800384879112,
      "learning_rate": 1.7482198495280206e-06,
      "loss": 0.0005,
      "step": 138030
    },
    {
      "epoch": 2.258692628650904,
      "grad_norm": 0.04462926462292671,
      "learning_rate": 1.747496786664012e-06,
      "loss": 0.0009,
      "step": 138040
    },
    {
      "epoch": 2.2588562546019797,
      "grad_norm": 0.17156077921390533,
      "learning_rate": 1.7467738416954293e-06,
      "loss": 0.0008,
      "step": 138050
    },
    {
      "epoch": 2.2590198805530557,
      "grad_norm": 0.11664771288633347,
      "learning_rate": 1.746051014648475e-06,
      "loss": 0.0012,
      "step": 138060
    },
    {
      "epoch": 2.2591835065041317,
      "grad_norm": 0.11774548143148422,
      "learning_rate": 1.745328305549352e-06,
      "loss": 0.0005,
      "step": 138070
    },
    {
      "epoch": 2.2593471324552072,
      "grad_norm": 0.042643509805202484,
      "learning_rate": 1.7446057144242557e-06,
      "loss": 0.001,
      "step": 138080
    },
    {
      "epoch": 2.2595107584062832,
      "grad_norm": 0.07703953981399536,
      "learning_rate": 1.7438832412993762e-06,
      "loss": 0.0009,
      "step": 138090
    },
    {
      "epoch": 2.2596743843573592,
      "grad_norm": 0.09358423203229904,
      "learning_rate": 1.7431608862009047e-06,
      "loss": 0.0006,
      "step": 138100
    },
    {
      "epoch": 2.259838010308435,
      "grad_norm": 0.1952233910560608,
      "learning_rate": 1.7424386491550222e-06,
      "loss": 0.0011,
      "step": 138110
    },
    {
      "epoch": 2.260001636259511,
      "grad_norm": 0.21610145270824432,
      "learning_rate": 1.7417165301879107e-06,
      "loss": 0.0009,
      "step": 138120
    },
    {
      "epoch": 2.260165262210587,
      "grad_norm": 0.035237573087215424,
      "learning_rate": 1.740994529325743e-06,
      "loss": 0.0007,
      "step": 138130
    },
    {
      "epoch": 2.2603288881616623,
      "grad_norm": 0.01159909088164568,
      "learning_rate": 1.7402726465946918e-06,
      "loss": 0.001,
      "step": 138140
    },
    {
      "epoch": 2.2604925141127383,
      "grad_norm": 0.008366485126316547,
      "learning_rate": 1.7395508820209212e-06,
      "loss": 0.0008,
      "step": 138150
    },
    {
      "epoch": 2.260656140063814,
      "grad_norm": 0.0494733490049839,
      "learning_rate": 1.7388292356305964e-06,
      "loss": 0.0008,
      "step": 138160
    },
    {
      "epoch": 2.26081976601489,
      "grad_norm": 0.03611666336655617,
      "learning_rate": 1.738107707449872e-06,
      "loss": 0.0007,
      "step": 138170
    },
    {
      "epoch": 2.260983391965966,
      "grad_norm": 0.06630255281925201,
      "learning_rate": 1.7373862975049056e-06,
      "loss": 0.001,
      "step": 138180
    },
    {
      "epoch": 2.2611470179170414,
      "grad_norm": 0.101160429418087,
      "learning_rate": 1.7366650058218437e-06,
      "loss": 0.001,
      "step": 138190
    },
    {
      "epoch": 2.2613106438681174,
      "grad_norm": 0.21658700704574585,
      "learning_rate": 1.7359438324268306e-06,
      "loss": 0.0007,
      "step": 138200
    },
    {
      "epoch": 2.2614742698191934,
      "grad_norm": 0.008043624460697174,
      "learning_rate": 1.7352227773460107e-06,
      "loss": 0.0006,
      "step": 138210
    },
    {
      "epoch": 2.261637895770269,
      "grad_norm": 0.002044062362983823,
      "learning_rate": 1.734501840605517e-06,
      "loss": 0.0006,
      "step": 138220
    },
    {
      "epoch": 2.261801521721345,
      "grad_norm": 0.005898504052311182,
      "learning_rate": 1.7337810222314844e-06,
      "loss": 0.0005,
      "step": 138230
    },
    {
      "epoch": 2.261965147672421,
      "grad_norm": 0.0536799281835556,
      "learning_rate": 1.7330603222500376e-06,
      "loss": 0.0008,
      "step": 138240
    },
    {
      "epoch": 2.2621287736234965,
      "grad_norm": 0.08331379294395447,
      "learning_rate": 1.7323397406873027e-06,
      "loss": 0.0007,
      "step": 138250
    },
    {
      "epoch": 2.2622923995745725,
      "grad_norm": 0.04971930757164955,
      "learning_rate": 1.7316192775694001e-06,
      "loss": 0.0006,
      "step": 138260
    },
    {
      "epoch": 2.2624560255256485,
      "grad_norm": 0.15873540937900543,
      "learning_rate": 1.730898932922442e-06,
      "loss": 0.0013,
      "step": 138270
    },
    {
      "epoch": 2.262619651476724,
      "grad_norm": 0.09229025989770889,
      "learning_rate": 1.7301787067725434e-06,
      "loss": 0.0006,
      "step": 138280
    },
    {
      "epoch": 2.2627832774278,
      "grad_norm": 0.05827879533171654,
      "learning_rate": 1.7294585991458046e-06,
      "loss": 0.0009,
      "step": 138290
    },
    {
      "epoch": 2.262946903378876,
      "grad_norm": 0.10686063021421432,
      "learning_rate": 1.7287386100683311e-06,
      "loss": 0.0009,
      "step": 138300
    },
    {
      "epoch": 2.2631105293299516,
      "grad_norm": 0.05689430609345436,
      "learning_rate": 1.7280187395662224e-06,
      "loss": 0.0004,
      "step": 138310
    },
    {
      "epoch": 2.2632741552810276,
      "grad_norm": 0.17898309230804443,
      "learning_rate": 1.7272989876655688e-06,
      "loss": 0.0011,
      "step": 138320
    },
    {
      "epoch": 2.2634377812321036,
      "grad_norm": 0.18826553225517273,
      "learning_rate": 1.726579354392463e-06,
      "loss": 0.0012,
      "step": 138330
    },
    {
      "epoch": 2.263601407183179,
      "grad_norm": 0.05082348734140396,
      "learning_rate": 1.725859839772986e-06,
      "loss": 0.0007,
      "step": 138340
    },
    {
      "epoch": 2.263765033134255,
      "grad_norm": 0.09817861020565033,
      "learning_rate": 1.7251404438332226e-06,
      "loss": 0.0011,
      "step": 138350
    },
    {
      "epoch": 2.263928659085331,
      "grad_norm": 0.004967184737324715,
      "learning_rate": 1.7244211665992456e-06,
      "loss": 0.0007,
      "step": 138360
    },
    {
      "epoch": 2.2640922850364067,
      "grad_norm": 0.15865665674209595,
      "learning_rate": 1.72370200809713e-06,
      "loss": 0.0012,
      "step": 138370
    },
    {
      "epoch": 2.2642559109874827,
      "grad_norm": 0.014201820828020573,
      "learning_rate": 1.7229829683529424e-06,
      "loss": 0.001,
      "step": 138380
    },
    {
      "epoch": 2.2644195369385587,
      "grad_norm": 0.1520019769668579,
      "learning_rate": 1.7222640473927443e-06,
      "loss": 0.0009,
      "step": 138390
    },
    {
      "epoch": 2.2645831628896342,
      "grad_norm": 0.06845227628946304,
      "learning_rate": 1.7215452452425986e-06,
      "loss": 0.0013,
      "step": 138400
    },
    {
      "epoch": 2.2647467888407102,
      "grad_norm": 0.05643242970108986,
      "learning_rate": 1.7208265619285563e-06,
      "loss": 0.001,
      "step": 138410
    },
    {
      "epoch": 2.264910414791786,
      "grad_norm": 0.019576242193579674,
      "learning_rate": 1.7201079974766716e-06,
      "loss": 0.0009,
      "step": 138420
    },
    {
      "epoch": 2.265074040742862,
      "grad_norm": 0.1266564279794693,
      "learning_rate": 1.7193895519129877e-06,
      "loss": 0.0011,
      "step": 138430
    },
    {
      "epoch": 2.265237666693938,
      "grad_norm": 0.04488179087638855,
      "learning_rate": 1.718671225263549e-06,
      "loss": 0.0009,
      "step": 138440
    },
    {
      "epoch": 2.2654012926450133,
      "grad_norm": 0.06462817639112473,
      "learning_rate": 1.71795301755439e-06,
      "loss": 0.0007,
      "step": 138450
    },
    {
      "epoch": 2.2655649185960893,
      "grad_norm": 0.04263461008667946,
      "learning_rate": 1.7172349288115464e-06,
      "loss": 0.0009,
      "step": 138460
    },
    {
      "epoch": 2.2657285445471653,
      "grad_norm": 0.023121269419789314,
      "learning_rate": 1.7165169590610486e-06,
      "loss": 0.0003,
      "step": 138470
    },
    {
      "epoch": 2.265892170498241,
      "grad_norm": 0.0011576018296182156,
      "learning_rate": 1.7157991083289189e-06,
      "loss": 0.0011,
      "step": 138480
    },
    {
      "epoch": 2.266055796449317,
      "grad_norm": 0.022658947855234146,
      "learning_rate": 1.7150813766411784e-06,
      "loss": 0.001,
      "step": 138490
    },
    {
      "epoch": 2.266219422400393,
      "grad_norm": 0.022222619503736496,
      "learning_rate": 1.714363764023841e-06,
      "loss": 0.001,
      "step": 138500
    },
    {
      "epoch": 2.2663830483514684,
      "grad_norm": 0.01836101897060871,
      "learning_rate": 1.7136462705029221e-06,
      "loss": 0.001,
      "step": 138510
    },
    {
      "epoch": 2.2665466743025444,
      "grad_norm": 0.30837783217430115,
      "learning_rate": 1.7129288961044255e-06,
      "loss": 0.0014,
      "step": 138520
    },
    {
      "epoch": 2.2667103002536204,
      "grad_norm": 0.04001494124531746,
      "learning_rate": 1.7122116408543566e-06,
      "loss": 0.0011,
      "step": 138530
    },
    {
      "epoch": 2.266873926204696,
      "grad_norm": 0.15171189606189728,
      "learning_rate": 1.711494504778715e-06,
      "loss": 0.0013,
      "step": 138540
    },
    {
      "epoch": 2.267037552155772,
      "grad_norm": 0.11166616529226303,
      "learning_rate": 1.7107774879034927e-06,
      "loss": 0.0012,
      "step": 138550
    },
    {
      "epoch": 2.2672011781068475,
      "grad_norm": 0.04553128033876419,
      "learning_rate": 1.7100605902546824e-06,
      "loss": 0.0006,
      "step": 138560
    },
    {
      "epoch": 2.2673648040579235,
      "grad_norm": 0.05317135155200958,
      "learning_rate": 1.7093438118582666e-06,
      "loss": 0.0007,
      "step": 138570
    },
    {
      "epoch": 2.2675284300089995,
      "grad_norm": 0.21715816855430603,
      "learning_rate": 1.7086271527402308e-06,
      "loss": 0.0016,
      "step": 138580
    },
    {
      "epoch": 2.267692055960075,
      "grad_norm": 0.07029976695775986,
      "learning_rate": 1.70791061292655e-06,
      "loss": 0.0008,
      "step": 138590
    },
    {
      "epoch": 2.267855681911151,
      "grad_norm": 0.084548220038414,
      "learning_rate": 1.707194192443195e-06,
      "loss": 0.0006,
      "step": 138600
    },
    {
      "epoch": 2.268019307862227,
      "grad_norm": 0.002569173462688923,
      "learning_rate": 1.706477891316139e-06,
      "loss": 0.0006,
      "step": 138610
    },
    {
      "epoch": 2.2681829338133026,
      "grad_norm": 0.04806216433644295,
      "learning_rate": 1.7057617095713412e-06,
      "loss": 0.0008,
      "step": 138620
    },
    {
      "epoch": 2.2683465597643786,
      "grad_norm": 0.16387756168842316,
      "learning_rate": 1.705045647234766e-06,
      "loss": 0.0007,
      "step": 138630
    },
    {
      "epoch": 2.2685101857154546,
      "grad_norm": 0.023508461192250252,
      "learning_rate": 1.7043297043323652e-06,
      "loss": 0.0011,
      "step": 138640
    },
    {
      "epoch": 2.26867381166653,
      "grad_norm": 0.1013295128941536,
      "learning_rate": 1.7036138808900938e-06,
      "loss": 0.0008,
      "step": 138650
    },
    {
      "epoch": 2.268837437617606,
      "grad_norm": 0.059112463146448135,
      "learning_rate": 1.7028981769338943e-06,
      "loss": 0.0013,
      "step": 138660
    },
    {
      "epoch": 2.269001063568682,
      "grad_norm": 0.11990657448768616,
      "learning_rate": 1.7021825924897133e-06,
      "loss": 0.0013,
      "step": 138670
    },
    {
      "epoch": 2.2691646895197577,
      "grad_norm": 0.03965405747294426,
      "learning_rate": 1.7014671275834871e-06,
      "loss": 0.0006,
      "step": 138680
    },
    {
      "epoch": 2.2693283154708337,
      "grad_norm": 0.09039021283388138,
      "learning_rate": 1.7007517822411485e-06,
      "loss": 0.0012,
      "step": 138690
    },
    {
      "epoch": 2.2694919414219097,
      "grad_norm": 0.05818749591708183,
      "learning_rate": 1.7000365564886295e-06,
      "loss": 0.0014,
      "step": 138700
    },
    {
      "epoch": 2.2696555673729852,
      "grad_norm": 0.09912757575511932,
      "learning_rate": 1.6993214503518523e-06,
      "loss": 0.0005,
      "step": 138710
    },
    {
      "epoch": 2.2698191933240612,
      "grad_norm": 0.06012880802154541,
      "learning_rate": 1.6986064638567412e-06,
      "loss": 0.0005,
      "step": 138720
    },
    {
      "epoch": 2.2699828192751372,
      "grad_norm": 0.07397420704364777,
      "learning_rate": 1.6978915970292097e-06,
      "loss": 0.001,
      "step": 138730
    },
    {
      "epoch": 2.270146445226213,
      "grad_norm": 0.0503038614988327,
      "learning_rate": 1.6971768498951717e-06,
      "loss": 0.0008,
      "step": 138740
    },
    {
      "epoch": 2.270310071177289,
      "grad_norm": 0.02471921779215336,
      "learning_rate": 1.6964622224805361e-06,
      "loss": 0.0005,
      "step": 138750
    },
    {
      "epoch": 2.2704736971283648,
      "grad_norm": 0.03774268925189972,
      "learning_rate": 1.6957477148112034e-06,
      "loss": 0.0004,
      "step": 138760
    },
    {
      "epoch": 2.2706373230794403,
      "grad_norm": 0.3109720051288605,
      "learning_rate": 1.695033326913076e-06,
      "loss": 0.0006,
      "step": 138770
    },
    {
      "epoch": 2.2708009490305163,
      "grad_norm": 0.0879649743437767,
      "learning_rate": 1.6943190588120472e-06,
      "loss": 0.0005,
      "step": 138780
    },
    {
      "epoch": 2.2709645749815923,
      "grad_norm": 0.11692360043525696,
      "learning_rate": 1.6936049105340079e-06,
      "loss": 0.0013,
      "step": 138790
    },
    {
      "epoch": 2.271128200932668,
      "grad_norm": 0.03319636359810829,
      "learning_rate": 1.6928908821048422e-06,
      "loss": 0.0009,
      "step": 138800
    },
    {
      "epoch": 2.271291826883744,
      "grad_norm": 0.03742299973964691,
      "learning_rate": 1.6921769735504334e-06,
      "loss": 0.0008,
      "step": 138810
    },
    {
      "epoch": 2.2714554528348194,
      "grad_norm": 0.04606965184211731,
      "learning_rate": 1.6914631848966607e-06,
      "loss": 0.0006,
      "step": 138820
    },
    {
      "epoch": 2.2716190787858954,
      "grad_norm": 0.05139930173754692,
      "learning_rate": 1.6907495161693943e-06,
      "loss": 0.0005,
      "step": 138830
    },
    {
      "epoch": 2.2717827047369714,
      "grad_norm": 0.007837047800421715,
      "learning_rate": 1.690035967394506e-06,
      "loss": 0.0011,
      "step": 138840
    },
    {
      "epoch": 2.271946330688047,
      "grad_norm": 0.07360134273767471,
      "learning_rate": 1.6893225385978567e-06,
      "loss": 0.0005,
      "step": 138850
    },
    {
      "epoch": 2.272109956639123,
      "grad_norm": 0.023278450593352318,
      "learning_rate": 1.6886092298053104e-06,
      "loss": 0.0012,
      "step": 138860
    },
    {
      "epoch": 2.272273582590199,
      "grad_norm": 0.03744323551654816,
      "learning_rate": 1.6878960410427187e-06,
      "loss": 0.0012,
      "step": 138870
    },
    {
      "epoch": 2.2724372085412745,
      "grad_norm": 0.028468498960137367,
      "learning_rate": 1.6871829723359368e-06,
      "loss": 0.0009,
      "step": 138880
    },
    {
      "epoch": 2.2726008344923505,
      "grad_norm": 0.13218821585178375,
      "learning_rate": 1.68647002371081e-06,
      "loss": 0.0012,
      "step": 138890
    },
    {
      "epoch": 2.2727644604434265,
      "grad_norm": 0.00504994997754693,
      "learning_rate": 1.6857571951931795e-06,
      "loss": 0.0012,
      "step": 138900
    },
    {
      "epoch": 2.272928086394502,
      "grad_norm": 0.16253262758255005,
      "learning_rate": 1.6850444868088866e-06,
      "loss": 0.001,
      "step": 138910
    },
    {
      "epoch": 2.273091712345578,
      "grad_norm": 0.05473746359348297,
      "learning_rate": 1.6843318985837626e-06,
      "loss": 0.0005,
      "step": 138920
    },
    {
      "epoch": 2.2732553382966536,
      "grad_norm": 0.20815257728099823,
      "learning_rate": 1.6836194305436393e-06,
      "loss": 0.0015,
      "step": 138930
    },
    {
      "epoch": 2.2734189642477296,
      "grad_norm": 0.0493653267621994,
      "learning_rate": 1.6829070827143396e-06,
      "loss": 0.001,
      "step": 138940
    },
    {
      "epoch": 2.2735825901988056,
      "grad_norm": 0.03902864083647728,
      "learning_rate": 1.6821948551216878e-06,
      "loss": 0.001,
      "step": 138950
    },
    {
      "epoch": 2.273746216149881,
      "grad_norm": 0.08249147981405258,
      "learning_rate": 1.6814827477914959e-06,
      "loss": 0.0009,
      "step": 138960
    },
    {
      "epoch": 2.273909842100957,
      "grad_norm": 0.01999511383473873,
      "learning_rate": 1.6807707607495787e-06,
      "loss": 0.0007,
      "step": 138970
    },
    {
      "epoch": 2.274073468052033,
      "grad_norm": 0.05212975665926933,
      "learning_rate": 1.6800588940217478e-06,
      "loss": 0.0005,
      "step": 138980
    },
    {
      "epoch": 2.2742370940031087,
      "grad_norm": 0.14709383249282837,
      "learning_rate": 1.6793471476337985e-06,
      "loss": 0.0012,
      "step": 138990
    },
    {
      "epoch": 2.2744007199541847,
      "grad_norm": 0.09429161995649338,
      "learning_rate": 1.6786355216115363e-06,
      "loss": 0.0008,
      "step": 139000
    },
    {
      "epoch": 2.2745643459052607,
      "grad_norm": 0.22234313189983368,
      "learning_rate": 1.6779240159807526e-06,
      "loss": 0.0014,
      "step": 139010
    },
    {
      "epoch": 2.2747279718563362,
      "grad_norm": 0.05838272348046303,
      "learning_rate": 1.677212630767238e-06,
      "loss": 0.0009,
      "step": 139020
    },
    {
      "epoch": 2.2748915978074122,
      "grad_norm": 0.07106290012598038,
      "learning_rate": 1.6765013659967815e-06,
      "loss": 0.0007,
      "step": 139030
    },
    {
      "epoch": 2.2750552237584882,
      "grad_norm": 0.04249867796897888,
      "learning_rate": 1.6757902216951611e-06,
      "loss": 0.0014,
      "step": 139040
    },
    {
      "epoch": 2.275218849709564,
      "grad_norm": 0.028932945802807808,
      "learning_rate": 1.675079197888157e-06,
      "loss": 0.0022,
      "step": 139050
    },
    {
      "epoch": 2.27538247566064,
      "grad_norm": 0.15324606001377106,
      "learning_rate": 1.6743682946015393e-06,
      "loss": 0.0006,
      "step": 139060
    },
    {
      "epoch": 2.2755461016117158,
      "grad_norm": 0.15140491724014282,
      "learning_rate": 1.6736575118610792e-06,
      "loss": 0.0011,
      "step": 139070
    },
    {
      "epoch": 2.2757097275627913,
      "grad_norm": 0.08114051073789597,
      "learning_rate": 1.6729468496925406e-06,
      "loss": 0.0005,
      "step": 139080
    },
    {
      "epoch": 2.2758733535138673,
      "grad_norm": 0.018584074452519417,
      "learning_rate": 1.6722363081216797e-06,
      "loss": 0.0008,
      "step": 139090
    },
    {
      "epoch": 2.2760369794649433,
      "grad_norm": 0.057265494018793106,
      "learning_rate": 1.6715258871742567e-06,
      "loss": 0.0011,
      "step": 139100
    },
    {
      "epoch": 2.276200605416019,
      "grad_norm": 0.004021097905933857,
      "learning_rate": 1.670815586876019e-06,
      "loss": 0.0011,
      "step": 139110
    },
    {
      "epoch": 2.276364231367095,
      "grad_norm": 0.04205913841724396,
      "learning_rate": 1.6701054072527162e-06,
      "loss": 0.0011,
      "step": 139120
    },
    {
      "epoch": 2.276527857318171,
      "grad_norm": 0.03988330066204071,
      "learning_rate": 1.6693953483300874e-06,
      "loss": 0.0013,
      "step": 139130
    },
    {
      "epoch": 2.2766914832692464,
      "grad_norm": 0.011853029951453209,
      "learning_rate": 1.668685410133874e-06,
      "loss": 0.0004,
      "step": 139140
    },
    {
      "epoch": 2.2768551092203224,
      "grad_norm": 0.12218745797872543,
      "learning_rate": 1.6679755926898066e-06,
      "loss": 0.001,
      "step": 139150
    },
    {
      "epoch": 2.2770187351713984,
      "grad_norm": 0.06630395352840424,
      "learning_rate": 1.667265896023617e-06,
      "loss": 0.0011,
      "step": 139160
    },
    {
      "epoch": 2.277182361122474,
      "grad_norm": 0.02177361026406288,
      "learning_rate": 1.666556320161027e-06,
      "loss": 0.0007,
      "step": 139170
    },
    {
      "epoch": 2.27734598707355,
      "grad_norm": 0.0012743271654471755,
      "learning_rate": 1.6658468651277604e-06,
      "loss": 0.0009,
      "step": 139180
    },
    {
      "epoch": 2.2775096130246255,
      "grad_norm": 0.058784399181604385,
      "learning_rate": 1.6651375309495315e-06,
      "loss": 0.001,
      "step": 139190
    },
    {
      "epoch": 2.2776732389757015,
      "grad_norm": 0.11113272607326508,
      "learning_rate": 1.6644283176520504e-06,
      "loss": 0.0012,
      "step": 139200
    },
    {
      "epoch": 2.2778368649267775,
      "grad_norm": 0.10286493599414825,
      "learning_rate": 1.6637192252610274e-06,
      "loss": 0.0008,
      "step": 139210
    },
    {
      "epoch": 2.278000490877853,
      "grad_norm": 0.09020040929317474,
      "learning_rate": 1.6630102538021625e-06,
      "loss": 0.001,
      "step": 139220
    },
    {
      "epoch": 2.278164116828929,
      "grad_norm": 0.11142237484455109,
      "learning_rate": 1.662301403301157e-06,
      "loss": 0.0012,
      "step": 139230
    },
    {
      "epoch": 2.278327742780005,
      "grad_norm": 0.08888515084981918,
      "learning_rate": 1.6615926737837023e-06,
      "loss": 0.0005,
      "step": 139240
    },
    {
      "epoch": 2.2784913687310806,
      "grad_norm": 0.01639641635119915,
      "learning_rate": 1.6608840652754898e-06,
      "loss": 0.0004,
      "step": 139250
    },
    {
      "epoch": 2.2786549946821566,
      "grad_norm": 0.07680242508649826,
      "learning_rate": 1.6601755778022055e-06,
      "loss": 0.0013,
      "step": 139260
    },
    {
      "epoch": 2.2788186206332326,
      "grad_norm": 0.08850005269050598,
      "learning_rate": 1.6594672113895283e-06,
      "loss": 0.0006,
      "step": 139270
    },
    {
      "epoch": 2.278982246584308,
      "grad_norm": 0.082381471991539,
      "learning_rate": 1.6587589660631392e-06,
      "loss": 0.0008,
      "step": 139280
    },
    {
      "epoch": 2.279145872535384,
      "grad_norm": 0.08754458278417587,
      "learning_rate": 1.658050841848704e-06,
      "loss": 0.0007,
      "step": 139290
    },
    {
      "epoch": 2.27930949848646,
      "grad_norm": 0.06146247312426567,
      "learning_rate": 1.6573428387718933e-06,
      "loss": 0.0019,
      "step": 139300
    },
    {
      "epoch": 2.2794731244375357,
      "grad_norm": 0.0855596661567688,
      "learning_rate": 1.6566349568583723e-06,
      "loss": 0.0006,
      "step": 139310
    },
    {
      "epoch": 2.2796367503886117,
      "grad_norm": 0.01028352975845337,
      "learning_rate": 1.6559271961337974e-06,
      "loss": 0.0004,
      "step": 139320
    },
    {
      "epoch": 2.2798003763396872,
      "grad_norm": 0.007872547022998333,
      "learning_rate": 1.6552195566238254e-06,
      "loss": 0.0006,
      "step": 139330
    },
    {
      "epoch": 2.2799640022907632,
      "grad_norm": 0.044140662997961044,
      "learning_rate": 1.6545120383541047e-06,
      "loss": 0.0005,
      "step": 139340
    },
    {
      "epoch": 2.2801276282418392,
      "grad_norm": 0.013871453702449799,
      "learning_rate": 1.6538046413502828e-06,
      "loss": 0.0004,
      "step": 139350
    },
    {
      "epoch": 2.280291254192915,
      "grad_norm": 0.08327380567789078,
      "learning_rate": 1.653097365637999e-06,
      "loss": 0.0012,
      "step": 139360
    },
    {
      "epoch": 2.280454880143991,
      "grad_norm": 0.12830397486686707,
      "learning_rate": 1.6523902112428935e-06,
      "loss": 0.001,
      "step": 139370
    },
    {
      "epoch": 2.2806185060950668,
      "grad_norm": 0.03175719827413559,
      "learning_rate": 1.6516831781905968e-06,
      "loss": 0.0011,
      "step": 139380
    },
    {
      "epoch": 2.2807821320461423,
      "grad_norm": 0.015279852785170078,
      "learning_rate": 1.6509762665067358e-06,
      "loss": 0.0004,
      "step": 139390
    },
    {
      "epoch": 2.2809457579972183,
      "grad_norm": 0.011449161916971207,
      "learning_rate": 1.6502694762169376e-06,
      "loss": 0.0007,
      "step": 139400
    },
    {
      "epoch": 2.2811093839482943,
      "grad_norm": 0.05544215068221092,
      "learning_rate": 1.6495628073468184e-06,
      "loss": 0.0018,
      "step": 139410
    },
    {
      "epoch": 2.28127300989937,
      "grad_norm": 0.007027834188193083,
      "learning_rate": 1.6488562599219965e-06,
      "loss": 0.0007,
      "step": 139420
    },
    {
      "epoch": 2.281436635850446,
      "grad_norm": 0.035232458263635635,
      "learning_rate": 1.648149833968079e-06,
      "loss": 0.0016,
      "step": 139430
    },
    {
      "epoch": 2.281600261801522,
      "grad_norm": 0.15208080410957336,
      "learning_rate": 1.647443529510676e-06,
      "loss": 0.0018,
      "step": 139440
    },
    {
      "epoch": 2.2817638877525974,
      "grad_norm": 0.035637788474559784,
      "learning_rate": 1.6467373465753855e-06,
      "loss": 0.0007,
      "step": 139450
    },
    {
      "epoch": 2.2819275137036734,
      "grad_norm": 0.01628064177930355,
      "learning_rate": 1.6460312851878074e-06,
      "loss": 0.0004,
      "step": 139460
    },
    {
      "epoch": 2.2820911396547494,
      "grad_norm": 0.05448537692427635,
      "learning_rate": 1.6453253453735353e-06,
      "loss": 0.0012,
      "step": 139470
    },
    {
      "epoch": 2.282254765605825,
      "grad_norm": 0.13321532309055328,
      "learning_rate": 1.6446195271581566e-06,
      "loss": 0.0008,
      "step": 139480
    },
    {
      "epoch": 2.282418391556901,
      "grad_norm": 0.05506069213151932,
      "learning_rate": 1.6439138305672559e-06,
      "loss": 0.0006,
      "step": 139490
    },
    {
      "epoch": 2.282582017507977,
      "grad_norm": 0.19812196493148804,
      "learning_rate": 1.6432082556264111e-06,
      "loss": 0.0005,
      "step": 139500
    },
    {
      "epoch": 2.2827456434590525,
      "grad_norm": 0.03946609050035477,
      "learning_rate": 1.6425028023611994e-06,
      "loss": 0.0007,
      "step": 139510
    },
    {
      "epoch": 2.2829092694101285,
      "grad_norm": 0.30005836486816406,
      "learning_rate": 1.641797470797193e-06,
      "loss": 0.0013,
      "step": 139520
    },
    {
      "epoch": 2.2830728953612045,
      "grad_norm": 0.3092999756336212,
      "learning_rate": 1.641092260959955e-06,
      "loss": 0.0018,
      "step": 139530
    },
    {
      "epoch": 2.28323652131228,
      "grad_norm": 0.10823874175548553,
      "learning_rate": 1.6403871728750514e-06,
      "loss": 0.0011,
      "step": 139540
    },
    {
      "epoch": 2.283400147263356,
      "grad_norm": 0.02327662892639637,
      "learning_rate": 1.6396822065680368e-06,
      "loss": 0.0009,
      "step": 139550
    },
    {
      "epoch": 2.283563773214432,
      "grad_norm": 0.06493327766656876,
      "learning_rate": 1.6389773620644678e-06,
      "loss": 0.0003,
      "step": 139560
    },
    {
      "epoch": 2.2837273991655076,
      "grad_norm": 0.06941473484039307,
      "learning_rate": 1.638272639389889e-06,
      "loss": 0.0003,
      "step": 139570
    },
    {
      "epoch": 2.2838910251165836,
      "grad_norm": 0.02352878823876381,
      "learning_rate": 1.6375680385698496e-06,
      "loss": 0.0006,
      "step": 139580
    },
    {
      "epoch": 2.284054651067659,
      "grad_norm": 0.03207981958985329,
      "learning_rate": 1.6368635596298872e-06,
      "loss": 0.0007,
      "step": 139590
    },
    {
      "epoch": 2.284218277018735,
      "grad_norm": 0.049908850342035294,
      "learning_rate": 1.6361592025955359e-06,
      "loss": 0.0007,
      "step": 139600
    },
    {
      "epoch": 2.284381902969811,
      "grad_norm": 0.019871102645993233,
      "learning_rate": 1.6354549674923303e-06,
      "loss": 0.001,
      "step": 139610
    },
    {
      "epoch": 2.2845455289208867,
      "grad_norm": 0.007271668408066034,
      "learning_rate": 1.6347508543457947e-06,
      "loss": 0.0019,
      "step": 139620
    },
    {
      "epoch": 2.2847091548719627,
      "grad_norm": 0.031424205750226974,
      "learning_rate": 1.6340468631814533e-06,
      "loss": 0.0004,
      "step": 139630
    },
    {
      "epoch": 2.2848727808230387,
      "grad_norm": 0.046694621443748474,
      "learning_rate": 1.6333429940248219e-06,
      "loss": 0.0009,
      "step": 139640
    },
    {
      "epoch": 2.2850364067741142,
      "grad_norm": 0.04434395954012871,
      "learning_rate": 1.6326392469014174e-06,
      "loss": 0.0006,
      "step": 139650
    },
    {
      "epoch": 2.2852000327251902,
      "grad_norm": 0.05857578665018082,
      "learning_rate": 1.6319356218367448e-06,
      "loss": 0.0013,
      "step": 139660
    },
    {
      "epoch": 2.2853636586762662,
      "grad_norm": 0.0665457472205162,
      "learning_rate": 1.6312321188563113e-06,
      "loss": 0.001,
      "step": 139670
    },
    {
      "epoch": 2.285527284627342,
      "grad_norm": 0.06540035456418991,
      "learning_rate": 1.6305287379856182e-06,
      "loss": 0.0005,
      "step": 139680
    },
    {
      "epoch": 2.285690910578418,
      "grad_norm": 0.06826451420783997,
      "learning_rate": 1.6298254792501606e-06,
      "loss": 0.0007,
      "step": 139690
    },
    {
      "epoch": 2.2858545365294933,
      "grad_norm": 0.05183140188455582,
      "learning_rate": 1.6291223426754289e-06,
      "loss": 0.0008,
      "step": 139700
    },
    {
      "epoch": 2.2860181624805693,
      "grad_norm": 0.04997354373335838,
      "learning_rate": 1.6284193282869093e-06,
      "loss": 0.0014,
      "step": 139710
    },
    {
      "epoch": 2.2861817884316453,
      "grad_norm": 0.030869508162140846,
      "learning_rate": 1.6277164361100868e-06,
      "loss": 0.001,
      "step": 139720
    },
    {
      "epoch": 2.286345414382721,
      "grad_norm": 0.11118374764919281,
      "learning_rate": 1.6270136661704373e-06,
      "loss": 0.001,
      "step": 139730
    },
    {
      "epoch": 2.286509040333797,
      "grad_norm": 0.22236043214797974,
      "learning_rate": 1.6263110184934356e-06,
      "loss": 0.0009,
      "step": 139740
    },
    {
      "epoch": 2.286672666284873,
      "grad_norm": 0.030773678794503212,
      "learning_rate": 1.6256084931045524e-06,
      "loss": 0.0008,
      "step": 139750
    },
    {
      "epoch": 2.2868362922359484,
      "grad_norm": 0.0017438497161492705,
      "learning_rate": 1.6249060900292495e-06,
      "loss": 0.0012,
      "step": 139760
    },
    {
      "epoch": 2.2869999181870244,
      "grad_norm": 0.12379049509763718,
      "learning_rate": 1.6242038092929908e-06,
      "loss": 0.0008,
      "step": 139770
    },
    {
      "epoch": 2.2871635441381004,
      "grad_norm": 0.09350680559873581,
      "learning_rate": 1.6235016509212286e-06,
      "loss": 0.0006,
      "step": 139780
    },
    {
      "epoch": 2.287327170089176,
      "grad_norm": 0.11177900433540344,
      "learning_rate": 1.6227996149394182e-06,
      "loss": 0.0008,
      "step": 139790
    },
    {
      "epoch": 2.287490796040252,
      "grad_norm": 0.04311496764421463,
      "learning_rate": 1.6220977013730049e-06,
      "loss": 0.0008,
      "step": 139800
    },
    {
      "epoch": 2.287654421991328,
      "grad_norm": 0.09959407895803452,
      "learning_rate": 1.6213959102474298e-06,
      "loss": 0.0006,
      "step": 139810
    },
    {
      "epoch": 2.2878180479424035,
      "grad_norm": 0.03599425405263901,
      "learning_rate": 1.6206942415881343e-06,
      "loss": 0.0002,
      "step": 139820
    },
    {
      "epoch": 2.2879816738934795,
      "grad_norm": 0.1324683129787445,
      "learning_rate": 1.6199926954205491e-06,
      "loss": 0.0011,
      "step": 139830
    },
    {
      "epoch": 2.2881452998445555,
      "grad_norm": 0.05062973499298096,
      "learning_rate": 1.6192912717701066e-06,
      "loss": 0.0003,
      "step": 139840
    },
    {
      "epoch": 2.288308925795631,
      "grad_norm": 0.25937387347221375,
      "learning_rate": 1.6185899706622282e-06,
      "loss": 0.0029,
      "step": 139850
    },
    {
      "epoch": 2.288472551746707,
      "grad_norm": 0.0058353738859295845,
      "learning_rate": 1.617888792122338e-06,
      "loss": 0.0007,
      "step": 139860
    },
    {
      "epoch": 2.288636177697783,
      "grad_norm": 0.050435684621334076,
      "learning_rate": 1.6171877361758493e-06,
      "loss": 0.0004,
      "step": 139870
    },
    {
      "epoch": 2.2887998036488586,
      "grad_norm": 0.09273172914981842,
      "learning_rate": 1.6164868028481761e-06,
      "loss": 0.0017,
      "step": 139880
    },
    {
      "epoch": 2.2889634295999346,
      "grad_norm": 0.09553926438093185,
      "learning_rate": 1.615785992164724e-06,
      "loss": 0.0005,
      "step": 139890
    },
    {
      "epoch": 2.2891270555510106,
      "grad_norm": 0.038811590522527695,
      "learning_rate": 1.6150853041508945e-06,
      "loss": 0.0003,
      "step": 139900
    },
    {
      "epoch": 2.289290681502086,
      "grad_norm": 0.0106173912063241,
      "learning_rate": 1.6143847388320894e-06,
      "loss": 0.0013,
      "step": 139910
    },
    {
      "epoch": 2.289454307453162,
      "grad_norm": 0.07198160886764526,
      "learning_rate": 1.613684296233698e-06,
      "loss": 0.0005,
      "step": 139920
    },
    {
      "epoch": 2.289617933404238,
      "grad_norm": 0.05207657814025879,
      "learning_rate": 1.612983976381114e-06,
      "loss": 0.0006,
      "step": 139930
    },
    {
      "epoch": 2.2897815593553137,
      "grad_norm": 0.017597423866391182,
      "learning_rate": 1.6122837792997191e-06,
      "loss": 0.0008,
      "step": 139940
    },
    {
      "epoch": 2.2899451853063897,
      "grad_norm": 0.0697166845202446,
      "learning_rate": 1.6115837050148947e-06,
      "loss": 0.0011,
      "step": 139950
    },
    {
      "epoch": 2.2901088112574657,
      "grad_norm": 0.024002980440855026,
      "learning_rate": 1.610883753552019e-06,
      "loss": 0.0009,
      "step": 139960
    },
    {
      "epoch": 2.2902724372085412,
      "grad_norm": 0.06490520387887955,
      "learning_rate": 1.61018392493646e-06,
      "loss": 0.0007,
      "step": 139970
    },
    {
      "epoch": 2.2904360631596172,
      "grad_norm": 0.14310207962989807,
      "learning_rate": 1.6094842191935877e-06,
      "loss": 0.001,
      "step": 139980
    },
    {
      "epoch": 2.290599689110693,
      "grad_norm": 0.1240270584821701,
      "learning_rate": 1.608784636348764e-06,
      "loss": 0.001,
      "step": 139990
    },
    {
      "epoch": 2.290763315061769,
      "grad_norm": 0.1480078399181366,
      "learning_rate": 1.6080851764273465e-06,
      "loss": 0.0007,
      "step": 140000
    },
    {
      "epoch": 2.2909269410128448,
      "grad_norm": 0.19509311020374298,
      "learning_rate": 1.6073858394546875e-06,
      "loss": 0.0011,
      "step": 140010
    },
    {
      "epoch": 2.2910905669639203,
      "grad_norm": 0.0041755251586437225,
      "learning_rate": 1.6066866254561381e-06,
      "loss": 0.0012,
      "step": 140020
    },
    {
      "epoch": 2.2912541929149963,
      "grad_norm": 0.13380256295204163,
      "learning_rate": 1.6059875344570442e-06,
      "loss": 0.0007,
      "step": 140030
    },
    {
      "epoch": 2.2914178188660723,
      "grad_norm": 0.14776718616485596,
      "learning_rate": 1.6052885664827433e-06,
      "loss": 0.001,
      "step": 140040
    },
    {
      "epoch": 2.291581444817148,
      "grad_norm": 0.307634562253952,
      "learning_rate": 1.6045897215585748e-06,
      "loss": 0.001,
      "step": 140050
    },
    {
      "epoch": 2.291745070768224,
      "grad_norm": 0.12230852246284485,
      "learning_rate": 1.6038909997098662e-06,
      "loss": 0.0009,
      "step": 140060
    },
    {
      "epoch": 2.2919086967193,
      "grad_norm": 0.005413628648966551,
      "learning_rate": 1.6031924009619488e-06,
      "loss": 0.001,
      "step": 140070
    },
    {
      "epoch": 2.2920723226703754,
      "grad_norm": 0.026682093739509583,
      "learning_rate": 1.6024939253401407e-06,
      "loss": 0.0005,
      "step": 140080
    },
    {
      "epoch": 2.2922359486214514,
      "grad_norm": 0.009006984531879425,
      "learning_rate": 1.6017955728697638e-06,
      "loss": 0.0007,
      "step": 140090
    },
    {
      "epoch": 2.292399574572527,
      "grad_norm": 0.1327577829360962,
      "learning_rate": 1.6010973435761302e-06,
      "loss": 0.0009,
      "step": 140100
    },
    {
      "epoch": 2.292563200523603,
      "grad_norm": 0.04897262528538704,
      "learning_rate": 1.6003992374845467e-06,
      "loss": 0.0006,
      "step": 140110
    },
    {
      "epoch": 2.292726826474679,
      "grad_norm": 0.08505528420209885,
      "learning_rate": 1.5997012546203223e-06,
      "loss": 0.0006,
      "step": 140120
    },
    {
      "epoch": 2.2928904524257545,
      "grad_norm": 0.023017607629299164,
      "learning_rate": 1.5990033950087537e-06,
      "loss": 0.0009,
      "step": 140130
    },
    {
      "epoch": 2.2930540783768305,
      "grad_norm": 0.1936735212802887,
      "learning_rate": 1.5983056586751388e-06,
      "loss": 0.0008,
      "step": 140140
    },
    {
      "epoch": 2.2932177043279065,
      "grad_norm": 0.011501717381179333,
      "learning_rate": 1.5976080456447668e-06,
      "loss": 0.0008,
      "step": 140150
    },
    {
      "epoch": 2.293381330278982,
      "grad_norm": 0.04580685496330261,
      "learning_rate": 1.596910555942927e-06,
      "loss": 0.0006,
      "step": 140160
    },
    {
      "epoch": 2.293544956230058,
      "grad_norm": 0.09653496742248535,
      "learning_rate": 1.596213189594899e-06,
      "loss": 0.0007,
      "step": 140170
    },
    {
      "epoch": 2.293708582181134,
      "grad_norm": 0.043103598058223724,
      "learning_rate": 1.5955159466259623e-06,
      "loss": 0.001,
      "step": 140180
    },
    {
      "epoch": 2.2938722081322096,
      "grad_norm": 0.021309738978743553,
      "learning_rate": 1.5948188270613935e-06,
      "loss": 0.0007,
      "step": 140190
    },
    {
      "epoch": 2.2940358340832856,
      "grad_norm": 0.19352178275585175,
      "learning_rate": 1.5941218309264545e-06,
      "loss": 0.0008,
      "step": 140200
    },
    {
      "epoch": 2.2941994600343616,
      "grad_norm": 0.03913641721010208,
      "learning_rate": 1.5934249582464146e-06,
      "loss": 0.001,
      "step": 140210
    },
    {
      "epoch": 2.294363085985437,
      "grad_norm": 0.06472831219434738,
      "learning_rate": 1.5927282090465313e-06,
      "loss": 0.0011,
      "step": 140220
    },
    {
      "epoch": 2.294526711936513,
      "grad_norm": 0.12262832373380661,
      "learning_rate": 1.592031583352061e-06,
      "loss": 0.0007,
      "step": 140230
    },
    {
      "epoch": 2.294690337887589,
      "grad_norm": 0.08961786329746246,
      "learning_rate": 1.5913350811882567e-06,
      "loss": 0.0006,
      "step": 140240
    },
    {
      "epoch": 2.2948539638386647,
      "grad_norm": 0.013164161704480648,
      "learning_rate": 1.5906387025803616e-06,
      "loss": 0.0007,
      "step": 140250
    },
    {
      "epoch": 2.2950175897897407,
      "grad_norm": 0.0371859073638916,
      "learning_rate": 1.5899424475536213e-06,
      "loss": 0.0008,
      "step": 140260
    },
    {
      "epoch": 2.2951812157408167,
      "grad_norm": 0.10145356506109238,
      "learning_rate": 1.5892463161332695e-06,
      "loss": 0.0008,
      "step": 140270
    },
    {
      "epoch": 2.2953448416918922,
      "grad_norm": 0.026181021705269814,
      "learning_rate": 1.5885503083445426e-06,
      "loss": 0.0008,
      "step": 140280
    },
    {
      "epoch": 2.2955084676429682,
      "grad_norm": 0.08042646944522858,
      "learning_rate": 1.587854424212668e-06,
      "loss": 0.001,
      "step": 140290
    },
    {
      "epoch": 2.2956720935940442,
      "grad_norm": 0.19835972785949707,
      "learning_rate": 1.5871586637628683e-06,
      "loss": 0.0005,
      "step": 140300
    },
    {
      "epoch": 2.29583571954512,
      "grad_norm": 0.0745401605963707,
      "learning_rate": 1.5864630270203656e-06,
      "loss": 0.001,
      "step": 140310
    },
    {
      "epoch": 2.2959993454961958,
      "grad_norm": 0.014865288510918617,
      "learning_rate": 1.5857675140103728e-06,
      "loss": 0.0005,
      "step": 140320
    },
    {
      "epoch": 2.2961629714472718,
      "grad_norm": 0.10254581272602081,
      "learning_rate": 1.585072124758103e-06,
      "loss": 0.0011,
      "step": 140330
    },
    {
      "epoch": 2.2963265973983473,
      "grad_norm": 0.007817289792001247,
      "learning_rate": 1.58437685928876e-06,
      "loss": 0.0007,
      "step": 140340
    },
    {
      "epoch": 2.2964902233494233,
      "grad_norm": 0.12360544502735138,
      "learning_rate": 1.5836817176275477e-06,
      "loss": 0.0011,
      "step": 140350
    },
    {
      "epoch": 2.296653849300499,
      "grad_norm": 0.08883540332317352,
      "learning_rate": 1.5829866997996608e-06,
      "loss": 0.0009,
      "step": 140360
    },
    {
      "epoch": 2.296817475251575,
      "grad_norm": 0.01839541085064411,
      "learning_rate": 1.582291805830295e-06,
      "loss": 0.001,
      "step": 140370
    },
    {
      "epoch": 2.296981101202651,
      "grad_norm": 0.12035586684942245,
      "learning_rate": 1.5815970357446353e-06,
      "loss": 0.0013,
      "step": 140380
    },
    {
      "epoch": 2.2971447271537264,
      "grad_norm": 0.14982575178146362,
      "learning_rate": 1.5809023895678687e-06,
      "loss": 0.001,
      "step": 140390
    },
    {
      "epoch": 2.2973083531048024,
      "grad_norm": 0.05199159309267998,
      "learning_rate": 1.5802078673251736e-06,
      "loss": 0.0007,
      "step": 140400
    },
    {
      "epoch": 2.2974719790558784,
      "grad_norm": 0.089327372610569,
      "learning_rate": 1.5795134690417218e-06,
      "loss": 0.0009,
      "step": 140410
    },
    {
      "epoch": 2.297635605006954,
      "grad_norm": 0.08943206816911697,
      "learning_rate": 1.5788191947426874e-06,
      "loss": 0.001,
      "step": 140420
    },
    {
      "epoch": 2.29779923095803,
      "grad_norm": 0.15100742876529694,
      "learning_rate": 1.5781250444532331e-06,
      "loss": 0.0008,
      "step": 140430
    },
    {
      "epoch": 2.297962856909106,
      "grad_norm": 0.3235109746456146,
      "learning_rate": 1.5774310181985236e-06,
      "loss": 0.0014,
      "step": 140440
    },
    {
      "epoch": 2.2981264828601815,
      "grad_norm": 0.08636219054460526,
      "learning_rate": 1.576737116003712e-06,
      "loss": 0.0006,
      "step": 140450
    },
    {
      "epoch": 2.2982901088112575,
      "grad_norm": 0.1043621376156807,
      "learning_rate": 1.5760433378939522e-06,
      "loss": 0.001,
      "step": 140460
    },
    {
      "epoch": 2.2984537347623335,
      "grad_norm": 0.060790009796619415,
      "learning_rate": 1.5753496838943938e-06,
      "loss": 0.0006,
      "step": 140470
    },
    {
      "epoch": 2.298617360713409,
      "grad_norm": 0.03214503824710846,
      "learning_rate": 1.5746561540301763e-06,
      "loss": 0.0006,
      "step": 140480
    },
    {
      "epoch": 2.298780986664485,
      "grad_norm": 0.05269099026918411,
      "learning_rate": 1.5739627483264447e-06,
      "loss": 0.0007,
      "step": 140490
    },
    {
      "epoch": 2.2989446126155606,
      "grad_norm": 0.06363150477409363,
      "learning_rate": 1.5732694668083255e-06,
      "loss": 0.0006,
      "step": 140500
    },
    {
      "epoch": 2.2991082385666366,
      "grad_norm": 0.07094758003950119,
      "learning_rate": 1.5725763095009521e-06,
      "loss": 0.0007,
      "step": 140510
    },
    {
      "epoch": 2.2992718645177126,
      "grad_norm": 0.055907078087329865,
      "learning_rate": 1.5718832764294523e-06,
      "loss": 0.0006,
      "step": 140520
    },
    {
      "epoch": 2.299435490468788,
      "grad_norm": 0.003993616439402103,
      "learning_rate": 1.5711903676189422e-06,
      "loss": 0.0007,
      "step": 140530
    },
    {
      "epoch": 2.299599116419864,
      "grad_norm": 0.007719043642282486,
      "learning_rate": 1.5704975830945419e-06,
      "loss": 0.0015,
      "step": 140540
    },
    {
      "epoch": 2.29976274237094,
      "grad_norm": 0.1341666728258133,
      "learning_rate": 1.56980492288136e-06,
      "loss": 0.0011,
      "step": 140550
    },
    {
      "epoch": 2.2999263683220157,
      "grad_norm": 0.06888158619403839,
      "learning_rate": 1.5691123870045071e-06,
      "loss": 0.0006,
      "step": 140560
    },
    {
      "epoch": 2.3000899942730917,
      "grad_norm": 0.09794234484434128,
      "learning_rate": 1.5684199754890828e-06,
      "loss": 0.0008,
      "step": 140570
    },
    {
      "epoch": 2.3002536202241677,
      "grad_norm": 0.03282199427485466,
      "learning_rate": 1.5677276883601884e-06,
      "loss": 0.0014,
      "step": 140580
    },
    {
      "epoch": 2.3004172461752432,
      "grad_norm": 0.11884036660194397,
      "learning_rate": 1.567035525642916e-06,
      "loss": 0.0009,
      "step": 140590
    },
    {
      "epoch": 2.3005808721263192,
      "grad_norm": 0.09278003871440887,
      "learning_rate": 1.5663434873623535e-06,
      "loss": 0.001,
      "step": 140600
    },
    {
      "epoch": 2.3007444980773952,
      "grad_norm": 0.17077024281024933,
      "learning_rate": 1.565651573543588e-06,
      "loss": 0.0006,
      "step": 140610
    },
    {
      "epoch": 2.300908124028471,
      "grad_norm": 0.07796622812747955,
      "learning_rate": 1.5649597842116975e-06,
      "loss": 0.0011,
      "step": 140620
    },
    {
      "epoch": 2.3010717499795468,
      "grad_norm": 0.004719328600913286,
      "learning_rate": 1.5642681193917602e-06,
      "loss": 0.001,
      "step": 140630
    },
    {
      "epoch": 2.3012353759306228,
      "grad_norm": 0.002486445941030979,
      "learning_rate": 1.5635765791088448e-06,
      "loss": 0.0005,
      "step": 140640
    },
    {
      "epoch": 2.3013990018816983,
      "grad_norm": 0.023250866681337357,
      "learning_rate": 1.5628851633880209e-06,
      "loss": 0.0006,
      "step": 140650
    },
    {
      "epoch": 2.3015626278327743,
      "grad_norm": 0.007309789769351482,
      "learning_rate": 1.5621938722543473e-06,
      "loss": 0.0008,
      "step": 140660
    },
    {
      "epoch": 2.3017262537838503,
      "grad_norm": 0.04784340783953667,
      "learning_rate": 1.561502705732883e-06,
      "loss": 0.0014,
      "step": 140670
    },
    {
      "epoch": 2.301889879734926,
      "grad_norm": 0.025625161826610565,
      "learning_rate": 1.5608116638486832e-06,
      "loss": 0.0018,
      "step": 140680
    },
    {
      "epoch": 2.302053505686002,
      "grad_norm": 0.12948989868164062,
      "learning_rate": 1.5601207466267942e-06,
      "loss": 0.0012,
      "step": 140690
    },
    {
      "epoch": 2.302217131637078,
      "grad_norm": 0.03865404427051544,
      "learning_rate": 1.5594299540922614e-06,
      "loss": 0.0011,
      "step": 140700
    },
    {
      "epoch": 2.3023807575881534,
      "grad_norm": 0.023492984473705292,
      "learning_rate": 1.5587392862701211e-06,
      "loss": 0.0014,
      "step": 140710
    },
    {
      "epoch": 2.3025443835392294,
      "grad_norm": 0.08059046417474747,
      "learning_rate": 1.5580487431854113e-06,
      "loss": 0.0006,
      "step": 140720
    },
    {
      "epoch": 2.3027080094903054,
      "grad_norm": 0.038865942507982254,
      "learning_rate": 1.557358324863164e-06,
      "loss": 0.0014,
      "step": 140730
    },
    {
      "epoch": 2.302871635441381,
      "grad_norm": 0.0996570736169815,
      "learning_rate": 1.556668031328401e-06,
      "loss": 0.0008,
      "step": 140740
    },
    {
      "epoch": 2.303035261392457,
      "grad_norm": 0.08180870115756989,
      "learning_rate": 1.5559778626061484e-06,
      "loss": 0.0008,
      "step": 140750
    },
    {
      "epoch": 2.3031988873435325,
      "grad_norm": 0.013720731250941753,
      "learning_rate": 1.5552878187214182e-06,
      "loss": 0.0005,
      "step": 140760
    },
    {
      "epoch": 2.3033625132946085,
      "grad_norm": 0.007137408945709467,
      "learning_rate": 1.5545978996992278e-06,
      "loss": 0.0004,
      "step": 140770
    },
    {
      "epoch": 2.3035261392456845,
      "grad_norm": 0.08784053474664688,
      "learning_rate": 1.553908105564581e-06,
      "loss": 0.0014,
      "step": 140780
    },
    {
      "epoch": 2.30368976519676,
      "grad_norm": 0.03646000102162361,
      "learning_rate": 1.553218436342484e-06,
      "loss": 0.0007,
      "step": 140790
    },
    {
      "epoch": 2.303853391147836,
      "grad_norm": 0.09041323512792587,
      "learning_rate": 1.552528892057935e-06,
      "loss": 0.001,
      "step": 140800
    },
    {
      "epoch": 2.304017017098912,
      "grad_norm": 0.020745286718010902,
      "learning_rate": 1.5518394727359265e-06,
      "loss": 0.0017,
      "step": 140810
    },
    {
      "epoch": 2.3041806430499876,
      "grad_norm": 0.0675344169139862,
      "learning_rate": 1.551150178401451e-06,
      "loss": 0.0012,
      "step": 140820
    },
    {
      "epoch": 2.3043442690010636,
      "grad_norm": 0.07974598556756973,
      "learning_rate": 1.5504610090794908e-06,
      "loss": 0.0013,
      "step": 140830
    },
    {
      "epoch": 2.3045078949521396,
      "grad_norm": 0.0454944409430027,
      "learning_rate": 1.54977196479503e-06,
      "loss": 0.0007,
      "step": 140840
    },
    {
      "epoch": 2.304671520903215,
      "grad_norm": 0.03727050870656967,
      "learning_rate": 1.5490830455730416e-06,
      "loss": 0.0008,
      "step": 140850
    },
    {
      "epoch": 2.304835146854291,
      "grad_norm": 0.1320389360189438,
      "learning_rate": 1.5483942514384997e-06,
      "loss": 0.0006,
      "step": 140860
    },
    {
      "epoch": 2.3049987728053667,
      "grad_norm": 0.2437596619129181,
      "learning_rate": 1.5477055824163696e-06,
      "loss": 0.0007,
      "step": 140870
    },
    {
      "epoch": 2.3051623987564427,
      "grad_norm": 0.01806657575070858,
      "learning_rate": 1.5470170385316141e-06,
      "loss": 0.0003,
      "step": 140880
    },
    {
      "epoch": 2.3053260247075187,
      "grad_norm": 0.3036212921142578,
      "learning_rate": 1.5463286198091958e-06,
      "loss": 0.001,
      "step": 140890
    },
    {
      "epoch": 2.3054896506585942,
      "grad_norm": 0.10354020446538925,
      "learning_rate": 1.545640326274061e-06,
      "loss": 0.0008,
      "step": 140900
    },
    {
      "epoch": 2.3056532766096702,
      "grad_norm": 0.05205728858709335,
      "learning_rate": 1.5449521579511634e-06,
      "loss": 0.0007,
      "step": 140910
    },
    {
      "epoch": 2.3058169025607462,
      "grad_norm": 0.10246003419160843,
      "learning_rate": 1.5442641148654447e-06,
      "loss": 0.0008,
      "step": 140920
    },
    {
      "epoch": 2.305980528511822,
      "grad_norm": 0.01056346669793129,
      "learning_rate": 1.5435761970418473e-06,
      "loss": 0.0013,
      "step": 140930
    },
    {
      "epoch": 2.3061441544628978,
      "grad_norm": 0.04447948932647705,
      "learning_rate": 1.5428884045053044e-06,
      "loss": 0.0005,
      "step": 140940
    },
    {
      "epoch": 2.3063077804139738,
      "grad_norm": 0.19907037913799286,
      "learning_rate": 1.5422007372807479e-06,
      "loss": 0.0019,
      "step": 140950
    },
    {
      "epoch": 2.3064714063650493,
      "grad_norm": 0.17131435871124268,
      "learning_rate": 1.5415131953931051e-06,
      "loss": 0.0009,
      "step": 140960
    },
    {
      "epoch": 2.3066350323161253,
      "grad_norm": 0.04842172563076019,
      "learning_rate": 1.540825778867296e-06,
      "loss": 0.0012,
      "step": 140970
    },
    {
      "epoch": 2.3067986582672013,
      "grad_norm": 0.03325434774160385,
      "learning_rate": 1.5401384877282394e-06,
      "loss": 0.0002,
      "step": 140980
    },
    {
      "epoch": 2.306962284218277,
      "grad_norm": 0.014830000698566437,
      "learning_rate": 1.5394513220008473e-06,
      "loss": 0.0005,
      "step": 140990
    },
    {
      "epoch": 2.307125910169353,
      "grad_norm": 0.11610081791877747,
      "learning_rate": 1.5387642817100257e-06,
      "loss": 0.001,
      "step": 141000
    },
    {
      "epoch": 2.307289536120429,
      "grad_norm": 0.04810063913464546,
      "learning_rate": 1.538077366880682e-06,
      "loss": 0.0005,
      "step": 141010
    },
    {
      "epoch": 2.3074531620715044,
      "grad_norm": 0.08294409513473511,
      "learning_rate": 1.5373905775377112e-06,
      "loss": 0.0012,
      "step": 141020
    },
    {
      "epoch": 2.3076167880225804,
      "grad_norm": 0.06494452059268951,
      "learning_rate": 1.5367039137060119e-06,
      "loss": 0.0007,
      "step": 141030
    },
    {
      "epoch": 2.3077804139736564,
      "grad_norm": 0.02179872803390026,
      "learning_rate": 1.53601737541047e-06,
      "loss": 0.0005,
      "step": 141040
    },
    {
      "epoch": 2.307944039924732,
      "grad_norm": 0.06045236811041832,
      "learning_rate": 1.5353309626759745e-06,
      "loss": 0.0012,
      "step": 141050
    },
    {
      "epoch": 2.308107665875808,
      "grad_norm": 0.05607156828045845,
      "learning_rate": 1.534644675527403e-06,
      "loss": 0.0005,
      "step": 141060
    },
    {
      "epoch": 2.308271291826884,
      "grad_norm": 0.1847607046365738,
      "learning_rate": 1.5339585139896346e-06,
      "loss": 0.0012,
      "step": 141070
    },
    {
      "epoch": 2.3084349177779595,
      "grad_norm": 0.059711333364248276,
      "learning_rate": 1.5332724780875386e-06,
      "loss": 0.0005,
      "step": 141080
    },
    {
      "epoch": 2.3085985437290355,
      "grad_norm": 0.028678933158516884,
      "learning_rate": 1.532586567845984e-06,
      "loss": 0.0005,
      "step": 141090
    },
    {
      "epoch": 2.3087621696801115,
      "grad_norm": 0.3262660801410675,
      "learning_rate": 1.531900783289833e-06,
      "loss": 0.0006,
      "step": 141100
    },
    {
      "epoch": 2.308925795631187,
      "grad_norm": 0.025376588106155396,
      "learning_rate": 1.5312151244439417e-06,
      "loss": 0.0008,
      "step": 141110
    },
    {
      "epoch": 2.309089421582263,
      "grad_norm": 0.0805472657084465,
      "learning_rate": 1.530529591333167e-06,
      "loss": 0.0008,
      "step": 141120
    },
    {
      "epoch": 2.3092530475333386,
      "grad_norm": 0.08390944451093674,
      "learning_rate": 1.5298441839823541e-06,
      "loss": 0.0007,
      "step": 141130
    },
    {
      "epoch": 2.3094166734844146,
      "grad_norm": 0.03424423560500145,
      "learning_rate": 1.5291589024163505e-06,
      "loss": 0.0004,
      "step": 141140
    },
    {
      "epoch": 2.3095802994354906,
      "grad_norm": 0.08039483428001404,
      "learning_rate": 1.5284737466599936e-06,
      "loss": 0.0011,
      "step": 141150
    },
    {
      "epoch": 2.309743925386566,
      "grad_norm": 0.09530168771743774,
      "learning_rate": 1.5277887167381195e-06,
      "loss": 0.0011,
      "step": 141160
    },
    {
      "epoch": 2.309907551337642,
      "grad_norm": 0.13639923930168152,
      "learning_rate": 1.527103812675561e-06,
      "loss": 0.0009,
      "step": 141170
    },
    {
      "epoch": 2.310071177288718,
      "grad_norm": 0.011712112464010715,
      "learning_rate": 1.5264190344971407e-06,
      "loss": 0.0006,
      "step": 141180
    },
    {
      "epoch": 2.3102348032397937,
      "grad_norm": 0.05489076301455498,
      "learning_rate": 1.5257343822276848e-06,
      "loss": 0.0005,
      "step": 141190
    },
    {
      "epoch": 2.3103984291908697,
      "grad_norm": 0.06974972784519196,
      "learning_rate": 1.5250498558920046e-06,
      "loss": 0.0006,
      "step": 141200
    },
    {
      "epoch": 2.3105620551419457,
      "grad_norm": 0.03023819997906685,
      "learning_rate": 1.5243654555149167e-06,
      "loss": 0.0007,
      "step": 141210
    },
    {
      "epoch": 2.3107256810930212,
      "grad_norm": 0.0018115267157554626,
      "learning_rate": 1.523681181121226e-06,
      "loss": 0.0005,
      "step": 141220
    },
    {
      "epoch": 2.3108893070440972,
      "grad_norm": 0.010107466019690037,
      "learning_rate": 1.5229970327357374e-06,
      "loss": 0.0006,
      "step": 141230
    },
    {
      "epoch": 2.3110529329951732,
      "grad_norm": 0.1055271327495575,
      "learning_rate": 1.5223130103832513e-06,
      "loss": 0.0005,
      "step": 141240
    },
    {
      "epoch": 2.311216558946249,
      "grad_norm": 0.19064393639564514,
      "learning_rate": 1.521629114088558e-06,
      "loss": 0.0007,
      "step": 141250
    },
    {
      "epoch": 2.3113801848973248,
      "grad_norm": 0.02189820259809494,
      "learning_rate": 1.5209453438764516e-06,
      "loss": 0.0006,
      "step": 141260
    },
    {
      "epoch": 2.3115438108484003,
      "grad_norm": 0.056310828775167465,
      "learning_rate": 1.5202616997717128e-06,
      "loss": 0.0013,
      "step": 141270
    },
    {
      "epoch": 2.3117074367994763,
      "grad_norm": 0.02012699283659458,
      "learning_rate": 1.5195781817991256e-06,
      "loss": 0.0007,
      "step": 141280
    },
    {
      "epoch": 2.3118710627505523,
      "grad_norm": 0.015752756968140602,
      "learning_rate": 1.518894789983465e-06,
      "loss": 0.0012,
      "step": 141290
    },
    {
      "epoch": 2.312034688701628,
      "grad_norm": 0.0858929455280304,
      "learning_rate": 1.5182115243494994e-06,
      "loss": 0.0013,
      "step": 141300
    },
    {
      "epoch": 2.312198314652704,
      "grad_norm": 0.046227049082517624,
      "learning_rate": 1.5175283849219995e-06,
      "loss": 0.0007,
      "step": 141310
    },
    {
      "epoch": 2.31236194060378,
      "grad_norm": 0.12042857706546783,
      "learning_rate": 1.5168453717257242e-06,
      "loss": 0.0005,
      "step": 141320
    },
    {
      "epoch": 2.3125255665548554,
      "grad_norm": 0.005126805044710636,
      "learning_rate": 1.516162484785434e-06,
      "loss": 0.0017,
      "step": 141330
    },
    {
      "epoch": 2.3126891925059314,
      "grad_norm": 0.005924693308770657,
      "learning_rate": 1.5154797241258795e-06,
      "loss": 0.0019,
      "step": 141340
    },
    {
      "epoch": 2.3128528184570074,
      "grad_norm": 0.014651631005108356,
      "learning_rate": 1.5147970897718111e-06,
      "loss": 0.0011,
      "step": 141350
    },
    {
      "epoch": 2.313016444408083,
      "grad_norm": 0.012786723673343658,
      "learning_rate": 1.5141145817479708e-06,
      "loss": 0.0006,
      "step": 141360
    },
    {
      "epoch": 2.313180070359159,
      "grad_norm": 0.12540513277053833,
      "learning_rate": 1.5134322000790996e-06,
      "loss": 0.0009,
      "step": 141370
    },
    {
      "epoch": 2.313343696310235,
      "grad_norm": 0.22066181898117065,
      "learning_rate": 1.5127499447899307e-06,
      "loss": 0.0021,
      "step": 141380
    },
    {
      "epoch": 2.3135073222613105,
      "grad_norm": 0.22398459911346436,
      "learning_rate": 1.512067815905196e-06,
      "loss": 0.0009,
      "step": 141390
    },
    {
      "epoch": 2.3136709482123865,
      "grad_norm": 0.1565244346857071,
      "learning_rate": 1.51138581344962e-06,
      "loss": 0.0008,
      "step": 141400
    },
    {
      "epoch": 2.3138345741634625,
      "grad_norm": 0.36510229110717773,
      "learning_rate": 1.510703937447922e-06,
      "loss": 0.0012,
      "step": 141410
    },
    {
      "epoch": 2.313998200114538,
      "grad_norm": 0.006878407672047615,
      "learning_rate": 1.5100221879248218e-06,
      "loss": 0.0014,
      "step": 141420
    },
    {
      "epoch": 2.314161826065614,
      "grad_norm": 0.029306985437870026,
      "learning_rate": 1.5093405649050275e-06,
      "loss": 0.0006,
      "step": 141430
    },
    {
      "epoch": 2.31432545201669,
      "grad_norm": 0.04922221228480339,
      "learning_rate": 1.508659068413248e-06,
      "loss": 0.0008,
      "step": 141440
    },
    {
      "epoch": 2.3144890779677656,
      "grad_norm": 0.12085670977830887,
      "learning_rate": 1.5079776984741879e-06,
      "loss": 0.0009,
      "step": 141450
    },
    {
      "epoch": 2.3146527039188416,
      "grad_norm": 0.12174364179372787,
      "learning_rate": 1.5072964551125418e-06,
      "loss": 0.0006,
      "step": 141460
    },
    {
      "epoch": 2.3148163298699176,
      "grad_norm": 0.05523371696472168,
      "learning_rate": 1.5066153383530063e-06,
      "loss": 0.0006,
      "step": 141470
    },
    {
      "epoch": 2.314979955820993,
      "grad_norm": 0.07125862687826157,
      "learning_rate": 1.505934348220267e-06,
      "loss": 0.0011,
      "step": 141480
    },
    {
      "epoch": 2.315143581772069,
      "grad_norm": 0.14378949999809265,
      "learning_rate": 1.5052534847390126e-06,
      "loss": 0.0011,
      "step": 141490
    },
    {
      "epoch": 2.315307207723145,
      "grad_norm": 0.055636826902627945,
      "learning_rate": 1.5045727479339172e-06,
      "loss": 0.0006,
      "step": 141500
    },
    {
      "epoch": 2.3154708336742207,
      "grad_norm": 0.024866728112101555,
      "learning_rate": 1.503892137829659e-06,
      "loss": 0.0006,
      "step": 141510
    },
    {
      "epoch": 2.3156344596252967,
      "grad_norm": 0.058483924716711044,
      "learning_rate": 1.503211654450909e-06,
      "loss": 0.0012,
      "step": 141520
    },
    {
      "epoch": 2.3157980855763722,
      "grad_norm": 0.06118519976735115,
      "learning_rate": 1.502531297822331e-06,
      "loss": 0.0012,
      "step": 141530
    },
    {
      "epoch": 2.3159617115274482,
      "grad_norm": 0.10276416689157486,
      "learning_rate": 1.5018510679685894e-06,
      "loss": 0.0005,
      "step": 141540
    },
    {
      "epoch": 2.3161253374785242,
      "grad_norm": 0.009653984569013119,
      "learning_rate": 1.501170964914337e-06,
      "loss": 0.0009,
      "step": 141550
    },
    {
      "epoch": 2.3162889634296,
      "grad_norm": 0.008758802898228168,
      "learning_rate": 1.5004909886842294e-06,
      "loss": 0.0005,
      "step": 141560
    },
    {
      "epoch": 2.3164525893806758,
      "grad_norm": 0.04171372577548027,
      "learning_rate": 1.4998111393029114e-06,
      "loss": 0.0006,
      "step": 141570
    },
    {
      "epoch": 2.3166162153317518,
      "grad_norm": 0.051459841430187225,
      "learning_rate": 1.4991314167950283e-06,
      "loss": 0.0007,
      "step": 141580
    },
    {
      "epoch": 2.3167798412828273,
      "grad_norm": 0.0077985436655581,
      "learning_rate": 1.4984518211852162e-06,
      "loss": 0.0007,
      "step": 141590
    },
    {
      "epoch": 2.3169434672339033,
      "grad_norm": 0.0700778141617775,
      "learning_rate": 1.4977723524981108e-06,
      "loss": 0.0007,
      "step": 141600
    },
    {
      "epoch": 2.3171070931849793,
      "grad_norm": 0.08564010262489319,
      "learning_rate": 1.4970930107583408e-06,
      "loss": 0.0032,
      "step": 141610
    },
    {
      "epoch": 2.317270719136055,
      "grad_norm": 0.027825219556689262,
      "learning_rate": 1.4964137959905283e-06,
      "loss": 0.0009,
      "step": 141620
    },
    {
      "epoch": 2.317434345087131,
      "grad_norm": 0.01738942787051201,
      "learning_rate": 1.4957347082192962e-06,
      "loss": 0.0006,
      "step": 141630
    },
    {
      "epoch": 2.3175979710382064,
      "grad_norm": 0.05528319254517555,
      "learning_rate": 1.495055747469258e-06,
      "loss": 0.0006,
      "step": 141640
    },
    {
      "epoch": 2.3177615969892824,
      "grad_norm": 0.06310944259166718,
      "learning_rate": 1.494376913765026e-06,
      "loss": 0.0006,
      "step": 141650
    },
    {
      "epoch": 2.3179252229403584,
      "grad_norm": 0.08646228909492493,
      "learning_rate": 1.4936982071312034e-06,
      "loss": 0.0005,
      "step": 141660
    },
    {
      "epoch": 2.318088848891434,
      "grad_norm": 0.11556138843297958,
      "learning_rate": 1.4930196275923942e-06,
      "loss": 0.0014,
      "step": 141670
    },
    {
      "epoch": 2.31825247484251,
      "grad_norm": 0.04201691970229149,
      "learning_rate": 1.4923411751731965e-06,
      "loss": 0.0011,
      "step": 141680
    },
    {
      "epoch": 2.318416100793586,
      "grad_norm": 0.017121024429798126,
      "learning_rate": 1.4916628498981983e-06,
      "loss": 0.0007,
      "step": 141690
    },
    {
      "epoch": 2.3185797267446615,
      "grad_norm": 0.017644086852669716,
      "learning_rate": 1.4909846517919935e-06,
      "loss": 0.0007,
      "step": 141700
    },
    {
      "epoch": 2.3187433526957375,
      "grad_norm": 0.15564754605293274,
      "learning_rate": 1.4903065808791585e-06,
      "loss": 0.0011,
      "step": 141710
    },
    {
      "epoch": 2.3189069786468135,
      "grad_norm": 0.014784197323024273,
      "learning_rate": 1.4896286371842738e-06,
      "loss": 0.0007,
      "step": 141720
    },
    {
      "epoch": 2.319070604597889,
      "grad_norm": 0.08888337016105652,
      "learning_rate": 1.4889508207319165e-06,
      "loss": 0.0006,
      "step": 141730
    },
    {
      "epoch": 2.319234230548965,
      "grad_norm": 0.024108178913593292,
      "learning_rate": 1.488273131546651e-06,
      "loss": 0.0009,
      "step": 141740
    },
    {
      "epoch": 2.319397856500041,
      "grad_norm": 0.041571635752916336,
      "learning_rate": 1.4875955696530464e-06,
      "loss": 0.0006,
      "step": 141750
    },
    {
      "epoch": 2.3195614824511166,
      "grad_norm": 0.11115611344575882,
      "learning_rate": 1.4869181350756584e-06,
      "loss": 0.001,
      "step": 141760
    },
    {
      "epoch": 2.3197251084021926,
      "grad_norm": 0.06924387067556381,
      "learning_rate": 1.4862408278390467e-06,
      "loss": 0.0006,
      "step": 141770
    },
    {
      "epoch": 2.3198887343532686,
      "grad_norm": 0.06451443582773209,
      "learning_rate": 1.4855636479677582e-06,
      "loss": 0.0013,
      "step": 141780
    },
    {
      "epoch": 2.320052360304344,
      "grad_norm": 0.07121411710977554,
      "learning_rate": 1.4848865954863423e-06,
      "loss": 0.0008,
      "step": 141790
    },
    {
      "epoch": 2.32021598625542,
      "grad_norm": 0.049608126282691956,
      "learning_rate": 1.4842096704193387e-06,
      "loss": 0.0009,
      "step": 141800
    },
    {
      "epoch": 2.320379612206496,
      "grad_norm": 0.03642949089407921,
      "learning_rate": 1.4835328727912834e-06,
      "loss": 0.0006,
      "step": 141810
    },
    {
      "epoch": 2.3205432381575717,
      "grad_norm": 0.006874953396618366,
      "learning_rate": 1.4828562026267112e-06,
      "loss": 0.0006,
      "step": 141820
    },
    {
      "epoch": 2.3207068641086477,
      "grad_norm": 0.11241057515144348,
      "learning_rate": 1.4821796599501465e-06,
      "loss": 0.0008,
      "step": 141830
    },
    {
      "epoch": 2.3208704900597237,
      "grad_norm": 0.04891185462474823,
      "learning_rate": 1.481503244786116e-06,
      "loss": 0.0009,
      "step": 141840
    },
    {
      "epoch": 2.3210341160107992,
      "grad_norm": 0.04376582056283951,
      "learning_rate": 1.4808269571591354e-06,
      "loss": 0.0012,
      "step": 141850
    },
    {
      "epoch": 2.3211977419618752,
      "grad_norm": 0.006816682871431112,
      "learning_rate": 1.4801507970937208e-06,
      "loss": 0.0007,
      "step": 141860
    },
    {
      "epoch": 2.3213613679129512,
      "grad_norm": 0.029662000015378,
      "learning_rate": 1.4794747646143781e-06,
      "loss": 0.0007,
      "step": 141870
    },
    {
      "epoch": 2.3215249938640268,
      "grad_norm": 0.03641365095973015,
      "learning_rate": 1.4787988597456142e-06,
      "loss": 0.0011,
      "step": 141880
    },
    {
      "epoch": 2.3216886198151028,
      "grad_norm": 0.026273412629961967,
      "learning_rate": 1.4781230825119304e-06,
      "loss": 0.0009,
      "step": 141890
    },
    {
      "epoch": 2.3218522457661783,
      "grad_norm": 0.017448076978325844,
      "learning_rate": 1.4774474329378196e-06,
      "loss": 0.0006,
      "step": 141900
    },
    {
      "epoch": 2.3220158717172543,
      "grad_norm": 0.024807600304484367,
      "learning_rate": 1.4767719110477736e-06,
      "loss": 0.0011,
      "step": 141910
    },
    {
      "epoch": 2.3221794976683303,
      "grad_norm": 0.11860799789428711,
      "learning_rate": 1.476096516866276e-06,
      "loss": 0.0015,
      "step": 141920
    },
    {
      "epoch": 2.322343123619406,
      "grad_norm": 0.2275388538837433,
      "learning_rate": 1.4754212504178105e-06,
      "loss": 0.0008,
      "step": 141930
    },
    {
      "epoch": 2.322506749570482,
      "grad_norm": 0.022442590445280075,
      "learning_rate": 1.4747461117268552e-06,
      "loss": 0.0006,
      "step": 141940
    },
    {
      "epoch": 2.322670375521558,
      "grad_norm": 0.15710876882076263,
      "learning_rate": 1.4740711008178793e-06,
      "loss": 0.0008,
      "step": 141950
    },
    {
      "epoch": 2.3228340014726334,
      "grad_norm": 0.098350889980793,
      "learning_rate": 1.4733962177153527e-06,
      "loss": 0.0009,
      "step": 141960
    },
    {
      "epoch": 2.3229976274237094,
      "grad_norm": 0.007286694832146168,
      "learning_rate": 1.4727214624437358e-06,
      "loss": 0.0003,
      "step": 141970
    },
    {
      "epoch": 2.3231612533747854,
      "grad_norm": 0.2774623930454254,
      "learning_rate": 1.4720468350274896e-06,
      "loss": 0.0008,
      "step": 141980
    },
    {
      "epoch": 2.323324879325861,
      "grad_norm": 0.04916456714272499,
      "learning_rate": 1.4713723354910654e-06,
      "loss": 0.0005,
      "step": 141990
    },
    {
      "epoch": 2.323488505276937,
      "grad_norm": 0.10238541662693024,
      "learning_rate": 1.4706979638589148e-06,
      "loss": 0.0008,
      "step": 142000
    },
    {
      "epoch": 2.323652131228013,
      "grad_norm": 0.012358625419437885,
      "learning_rate": 1.47002372015548e-06,
      "loss": 0.0005,
      "step": 142010
    },
    {
      "epoch": 2.3238157571790885,
      "grad_norm": 0.11864018440246582,
      "learning_rate": 1.4693496044052002e-06,
      "loss": 0.0009,
      "step": 142020
    },
    {
      "epoch": 2.3239793831301645,
      "grad_norm": 0.3684505224227905,
      "learning_rate": 1.4686756166325128e-06,
      "loss": 0.0011,
      "step": 142030
    },
    {
      "epoch": 2.32414300908124,
      "grad_norm": 0.032025329768657684,
      "learning_rate": 1.4680017568618454e-06,
      "loss": 0.0009,
      "step": 142040
    },
    {
      "epoch": 2.324306635032316,
      "grad_norm": 0.025022050365805626,
      "learning_rate": 1.4673280251176269e-06,
      "loss": 0.0006,
      "step": 142050
    },
    {
      "epoch": 2.324470260983392,
      "grad_norm": 0.04085569456219673,
      "learning_rate": 1.4666544214242756e-06,
      "loss": 0.0005,
      "step": 142060
    },
    {
      "epoch": 2.3246338869344676,
      "grad_norm": 0.04850098118185997,
      "learning_rate": 1.4659809458062113e-06,
      "loss": 0.0011,
      "step": 142070
    },
    {
      "epoch": 2.3247975128855436,
      "grad_norm": 0.1168837621808052,
      "learning_rate": 1.4653075982878417e-06,
      "loss": 0.0009,
      "step": 142080
    },
    {
      "epoch": 2.3249611388366196,
      "grad_norm": 0.04493911191821098,
      "learning_rate": 1.4646343788935762e-06,
      "loss": 0.0011,
      "step": 142090
    },
    {
      "epoch": 2.325124764787695,
      "grad_norm": 0.06869116425514221,
      "learning_rate": 1.463961287647821e-06,
      "loss": 0.0006,
      "step": 142100
    },
    {
      "epoch": 2.325288390738771,
      "grad_norm": 0.06428693234920502,
      "learning_rate": 1.4632883245749675e-06,
      "loss": 0.0013,
      "step": 142110
    },
    {
      "epoch": 2.325452016689847,
      "grad_norm": 0.08442873507738113,
      "learning_rate": 1.4626154896994133e-06,
      "loss": 0.0007,
      "step": 142120
    },
    {
      "epoch": 2.3256156426409227,
      "grad_norm": 0.002236641477793455,
      "learning_rate": 1.4619427830455441e-06,
      "loss": 0.0003,
      "step": 142130
    },
    {
      "epoch": 2.3257792685919987,
      "grad_norm": 0.0899309292435646,
      "learning_rate": 1.461270204637747e-06,
      "loss": 0.0009,
      "step": 142140
    },
    {
      "epoch": 2.3259428945430747,
      "grad_norm": 0.12744580209255219,
      "learning_rate": 1.460597754500398e-06,
      "loss": 0.0005,
      "step": 142150
    },
    {
      "epoch": 2.3261065204941502,
      "grad_norm": 0.0532243438065052,
      "learning_rate": 1.4599254326578743e-06,
      "loss": 0.0008,
      "step": 142160
    },
    {
      "epoch": 2.3262701464452262,
      "grad_norm": 0.06452793627977371,
      "learning_rate": 1.4592532391345466e-06,
      "loss": 0.0006,
      "step": 142170
    },
    {
      "epoch": 2.3264337723963022,
      "grad_norm": 0.05454513058066368,
      "learning_rate": 1.4585811739547773e-06,
      "loss": 0.0008,
      "step": 142180
    },
    {
      "epoch": 2.3265973983473778,
      "grad_norm": 0.03910773620009422,
      "learning_rate": 1.4579092371429305e-06,
      "loss": 0.0004,
      "step": 142190
    },
    {
      "epoch": 2.3267610242984538,
      "grad_norm": 0.04444894567131996,
      "learning_rate": 1.4572374287233609e-06,
      "loss": 0.0006,
      "step": 142200
    },
    {
      "epoch": 2.3269246502495298,
      "grad_norm": 0.03417925536632538,
      "learning_rate": 1.4565657487204183e-06,
      "loss": 0.0006,
      "step": 142210
    },
    {
      "epoch": 2.3270882762006053,
      "grad_norm": 0.24629062414169312,
      "learning_rate": 1.4558941971584516e-06,
      "loss": 0.0008,
      "step": 142220
    },
    {
      "epoch": 2.3272519021516813,
      "grad_norm": 0.11486601084470749,
      "learning_rate": 1.4552227740618013e-06,
      "loss": 0.0008,
      "step": 142230
    },
    {
      "epoch": 2.3274155281027573,
      "grad_norm": 0.12311146408319473,
      "learning_rate": 1.4545514794548076e-06,
      "loss": 0.0007,
      "step": 142240
    },
    {
      "epoch": 2.327579154053833,
      "grad_norm": 0.06289301067590714,
      "learning_rate": 1.4538803133617996e-06,
      "loss": 0.0007,
      "step": 142250
    },
    {
      "epoch": 2.327742780004909,
      "grad_norm": 0.1327168047428131,
      "learning_rate": 1.4532092758071087e-06,
      "loss": 0.001,
      "step": 142260
    },
    {
      "epoch": 2.327906405955985,
      "grad_norm": 0.11817660927772522,
      "learning_rate": 1.452538366815056e-06,
      "loss": 0.0011,
      "step": 142270
    },
    {
      "epoch": 2.3280700319070604,
      "grad_norm": 0.08783981949090958,
      "learning_rate": 1.4518675864099624e-06,
      "loss": 0.0004,
      "step": 142280
    },
    {
      "epoch": 2.3282336578581364,
      "grad_norm": 0.07068328559398651,
      "learning_rate": 1.4511969346161397e-06,
      "loss": 0.0006,
      "step": 142290
    },
    {
      "epoch": 2.328397283809212,
      "grad_norm": 0.03534681722521782,
      "learning_rate": 1.450526411457901e-06,
      "loss": 0.0008,
      "step": 142300
    },
    {
      "epoch": 2.328560909760288,
      "grad_norm": 0.19301646947860718,
      "learning_rate": 1.4498560169595483e-06,
      "loss": 0.0008,
      "step": 142310
    },
    {
      "epoch": 2.328724535711364,
      "grad_norm": 0.11231374740600586,
      "learning_rate": 1.449185751145381e-06,
      "loss": 0.0008,
      "step": 142320
    },
    {
      "epoch": 2.3288881616624395,
      "grad_norm": 0.002907052868977189,
      "learning_rate": 1.4485156140396978e-06,
      "loss": 0.0005,
      "step": 142330
    },
    {
      "epoch": 2.3290517876135155,
      "grad_norm": 0.037952326238155365,
      "learning_rate": 1.4478456056667861e-06,
      "loss": 0.0011,
      "step": 142340
    },
    {
      "epoch": 2.3292154135645915,
      "grad_norm": 0.07551920413970947,
      "learning_rate": 1.447175726050935e-06,
      "loss": 0.0007,
      "step": 142350
    },
    {
      "epoch": 2.329379039515667,
      "grad_norm": 0.032292623072862625,
      "learning_rate": 1.446505975216424e-06,
      "loss": 0.0009,
      "step": 142360
    },
    {
      "epoch": 2.329542665466743,
      "grad_norm": 0.05330345779657364,
      "learning_rate": 1.4458363531875302e-06,
      "loss": 0.0009,
      "step": 142370
    },
    {
      "epoch": 2.329706291417819,
      "grad_norm": 0.06061384454369545,
      "learning_rate": 1.4451668599885282e-06,
      "loss": 0.0005,
      "step": 142380
    },
    {
      "epoch": 2.3298699173688946,
      "grad_norm": 0.011113053187727928,
      "learning_rate": 1.4444974956436824e-06,
      "loss": 0.0007,
      "step": 142390
    },
    {
      "epoch": 2.3300335433199706,
      "grad_norm": 0.18374551832675934,
      "learning_rate": 1.4438282601772597e-06,
      "loss": 0.0006,
      "step": 142400
    },
    {
      "epoch": 2.330197169271046,
      "grad_norm": 0.07437953352928162,
      "learning_rate": 1.4431591536135125e-06,
      "loss": 0.0012,
      "step": 142410
    },
    {
      "epoch": 2.330360795222122,
      "grad_norm": 0.09924048185348511,
      "learning_rate": 1.442490175976699e-06,
      "loss": 0.0012,
      "step": 142420
    },
    {
      "epoch": 2.330524421173198,
      "grad_norm": 0.12255416065454483,
      "learning_rate": 1.4418213272910647e-06,
      "loss": 0.0013,
      "step": 142430
    },
    {
      "epoch": 2.3306880471242737,
      "grad_norm": 0.021334750577807426,
      "learning_rate": 1.4411526075808558e-06,
      "loss": 0.0004,
      "step": 142440
    },
    {
      "epoch": 2.3308516730753497,
      "grad_norm": 0.06889131665229797,
      "learning_rate": 1.4404840168703122e-06,
      "loss": 0.0004,
      "step": 142450
    },
    {
      "epoch": 2.3310152990264257,
      "grad_norm": 0.05617127567529678,
      "learning_rate": 1.4398155551836668e-06,
      "loss": 0.0009,
      "step": 142460
    },
    {
      "epoch": 2.3311789249775012,
      "grad_norm": 0.0668850690126419,
      "learning_rate": 1.4391472225451524e-06,
      "loss": 0.0011,
      "step": 142470
    },
    {
      "epoch": 2.3313425509285772,
      "grad_norm": 0.24449512362480164,
      "learning_rate": 1.4384790189789916e-06,
      "loss": 0.0007,
      "step": 142480
    },
    {
      "epoch": 2.3315061768796532,
      "grad_norm": 0.0331275649368763,
      "learning_rate": 1.4378109445094079e-06,
      "loss": 0.0005,
      "step": 142490
    },
    {
      "epoch": 2.3316698028307288,
      "grad_norm": 0.14289069175720215,
      "learning_rate": 1.4371429991606156e-06,
      "loss": 0.0013,
      "step": 142500
    },
    {
      "epoch": 2.3318334287818048,
      "grad_norm": 0.046746205538511276,
      "learning_rate": 1.4364751829568253e-06,
      "loss": 0.0008,
      "step": 142510
    },
    {
      "epoch": 2.3319970547328808,
      "grad_norm": 0.009582074359059334,
      "learning_rate": 1.4358074959222461e-06,
      "loss": 0.0007,
      "step": 142520
    },
    {
      "epoch": 2.3321606806839563,
      "grad_norm": 0.10607443749904633,
      "learning_rate": 1.4351399380810776e-06,
      "loss": 0.0006,
      "step": 142530
    },
    {
      "epoch": 2.3323243066350323,
      "grad_norm": 0.051514916121959686,
      "learning_rate": 1.43447250945752e-06,
      "loss": 0.0007,
      "step": 142540
    },
    {
      "epoch": 2.3324879325861083,
      "grad_norm": 0.3152959644794464,
      "learning_rate": 1.4338052100757626e-06,
      "loss": 0.0007,
      "step": 142550
    },
    {
      "epoch": 2.332651558537184,
      "grad_norm": 0.16308702528476715,
      "learning_rate": 1.433138039959997e-06,
      "loss": 0.0013,
      "step": 142560
    },
    {
      "epoch": 2.33281518448826,
      "grad_norm": 0.0715632513165474,
      "learning_rate": 1.4324709991344032e-06,
      "loss": 0.001,
      "step": 142570
    },
    {
      "epoch": 2.332978810439336,
      "grad_norm": 0.038515858352184296,
      "learning_rate": 1.431804087623163e-06,
      "loss": 0.0008,
      "step": 142580
    },
    {
      "epoch": 2.3331424363904114,
      "grad_norm": 0.07718722522258759,
      "learning_rate": 1.4311373054504469e-06,
      "loss": 0.0015,
      "step": 142590
    },
    {
      "epoch": 2.3333060623414874,
      "grad_norm": 0.05153379589319229,
      "learning_rate": 1.4304706526404278e-06,
      "loss": 0.0005,
      "step": 142600
    },
    {
      "epoch": 2.3334696882925634,
      "grad_norm": 0.018288012593984604,
      "learning_rate": 1.4298041292172683e-06,
      "loss": 0.0005,
      "step": 142610
    },
    {
      "epoch": 2.333633314243639,
      "grad_norm": 0.03371068462729454,
      "learning_rate": 1.4291377352051266e-06,
      "loss": 0.0007,
      "step": 142620
    },
    {
      "epoch": 2.333796940194715,
      "grad_norm": 0.05810781940817833,
      "learning_rate": 1.4284714706281611e-06,
      "loss": 0.0005,
      "step": 142630
    },
    {
      "epoch": 2.333960566145791,
      "grad_norm": 0.018913494423031807,
      "learning_rate": 1.4278053355105198e-06,
      "loss": 0.0011,
      "step": 142640
    },
    {
      "epoch": 2.3341241920968665,
      "grad_norm": 0.01816447637975216,
      "learning_rate": 1.427139329876349e-06,
      "loss": 0.0009,
      "step": 142650
    },
    {
      "epoch": 2.3342878180479425,
      "grad_norm": 0.08949961513280869,
      "learning_rate": 1.426473453749792e-06,
      "loss": 0.0005,
      "step": 142660
    },
    {
      "epoch": 2.3344514439990185,
      "grad_norm": 0.058733511716127396,
      "learning_rate": 1.4258077071549825e-06,
      "loss": 0.0014,
      "step": 142670
    },
    {
      "epoch": 2.334615069950094,
      "grad_norm": 0.02654833346605301,
      "learning_rate": 1.425142090116054e-06,
      "loss": 0.0006,
      "step": 142680
    },
    {
      "epoch": 2.33477869590117,
      "grad_norm": 0.18389055132865906,
      "learning_rate": 1.424476602657132e-06,
      "loss": 0.001,
      "step": 142690
    },
    {
      "epoch": 2.3349423218522456,
      "grad_norm": 0.02986842952668667,
      "learning_rate": 1.4238112448023422e-06,
      "loss": 0.0009,
      "step": 142700
    },
    {
      "epoch": 2.3351059478033216,
      "grad_norm": 0.09791618585586548,
      "learning_rate": 1.4231460165757972e-06,
      "loss": 0.0008,
      "step": 142710
    },
    {
      "epoch": 2.3352695737543976,
      "grad_norm": 0.012414880096912384,
      "learning_rate": 1.4224809180016114e-06,
      "loss": 0.0006,
      "step": 142720
    },
    {
      "epoch": 2.335433199705473,
      "grad_norm": 0.007660441100597382,
      "learning_rate": 1.4218159491038962e-06,
      "loss": 0.0011,
      "step": 142730
    },
    {
      "epoch": 2.335596825656549,
      "grad_norm": 0.024664273485541344,
      "learning_rate": 1.421151109906751e-06,
      "loss": 0.0003,
      "step": 142740
    },
    {
      "epoch": 2.335760451607625,
      "grad_norm": 0.1829763650894165,
      "learning_rate": 1.4204864004342782e-06,
      "loss": 0.0005,
      "step": 142750
    },
    {
      "epoch": 2.3359240775587007,
      "grad_norm": 0.03625491261482239,
      "learning_rate": 1.419821820710569e-06,
      "loss": 0.0012,
      "step": 142760
    },
    {
      "epoch": 2.3360877035097767,
      "grad_norm": 0.13838981091976166,
      "learning_rate": 1.4191573707597155e-06,
      "loss": 0.0008,
      "step": 142770
    },
    {
      "epoch": 2.3362513294608527,
      "grad_norm": 0.04847600683569908,
      "learning_rate": 1.4184930506057993e-06,
      "loss": 0.001,
      "step": 142780
    },
    {
      "epoch": 2.3364149554119282,
      "grad_norm": 0.14510351419448853,
      "learning_rate": 1.4178288602729035e-06,
      "loss": 0.0028,
      "step": 142790
    },
    {
      "epoch": 2.3365785813630042,
      "grad_norm": 0.04510261490941048,
      "learning_rate": 1.4171647997851024e-06,
      "loss": 0.0006,
      "step": 142800
    },
    {
      "epoch": 2.3367422073140798,
      "grad_norm": 0.04008515551686287,
      "learning_rate": 1.4165008691664644e-06,
      "loss": 0.0003,
      "step": 142810
    },
    {
      "epoch": 2.3369058332651558,
      "grad_norm": 0.06763363629579544,
      "learning_rate": 1.4158370684410588e-06,
      "loss": 0.0008,
      "step": 142820
    },
    {
      "epoch": 2.3370694592162318,
      "grad_norm": 0.09663360565900803,
      "learning_rate": 1.4151733976329434e-06,
      "loss": 0.0003,
      "step": 142830
    },
    {
      "epoch": 2.3372330851673073,
      "grad_norm": 0.06752414256334305,
      "learning_rate": 1.414509856766178e-06,
      "loss": 0.0004,
      "step": 142840
    },
    {
      "epoch": 2.3373967111183833,
      "grad_norm": 0.08504626154899597,
      "learning_rate": 1.4138464458648115e-06,
      "loss": 0.0017,
      "step": 142850
    },
    {
      "epoch": 2.3375603370694593,
      "grad_norm": 0.03862268850207329,
      "learning_rate": 1.4131831649528938e-06,
      "loss": 0.0006,
      "step": 142860
    },
    {
      "epoch": 2.337723963020535,
      "grad_norm": 0.11114795506000519,
      "learning_rate": 1.4125200140544638e-06,
      "loss": 0.0025,
      "step": 142870
    },
    {
      "epoch": 2.337887588971611,
      "grad_norm": 0.021131668239831924,
      "learning_rate": 1.4118569931935612e-06,
      "loss": 0.0012,
      "step": 142880
    },
    {
      "epoch": 2.338051214922687,
      "grad_norm": 0.14534032344818115,
      "learning_rate": 1.4111941023942199e-06,
      "loss": 0.0012,
      "step": 142890
    },
    {
      "epoch": 2.3382148408737624,
      "grad_norm": 0.02748599648475647,
      "learning_rate": 1.4105313416804677e-06,
      "loss": 0.0005,
      "step": 142900
    },
    {
      "epoch": 2.3383784668248384,
      "grad_norm": 0.16241908073425293,
      "learning_rate": 1.4098687110763265e-06,
      "loss": 0.0006,
      "step": 142910
    },
    {
      "epoch": 2.3385420927759144,
      "grad_norm": 0.05146666616201401,
      "learning_rate": 1.4092062106058152e-06,
      "loss": 0.0008,
      "step": 142920
    },
    {
      "epoch": 2.33870571872699,
      "grad_norm": 0.06983890384435654,
      "learning_rate": 1.408543840292948e-06,
      "loss": 0.0009,
      "step": 142930
    },
    {
      "epoch": 2.338869344678066,
      "grad_norm": 0.05474630743265152,
      "learning_rate": 1.4078816001617363e-06,
      "loss": 0.0007,
      "step": 142940
    },
    {
      "epoch": 2.339032970629142,
      "grad_norm": 0.03730502352118492,
      "learning_rate": 1.4072194902361814e-06,
      "loss": 0.0025,
      "step": 142950
    },
    {
      "epoch": 2.3391965965802175,
      "grad_norm": 0.033826302736997604,
      "learning_rate": 1.4065575105402868e-06,
      "loss": 0.0008,
      "step": 142960
    },
    {
      "epoch": 2.3393602225312935,
      "grad_norm": 0.1366487741470337,
      "learning_rate": 1.405895661098044e-06,
      "loss": 0.002,
      "step": 142970
    },
    {
      "epoch": 2.3395238484823695,
      "grad_norm": 0.050285596400499344,
      "learning_rate": 1.4052339419334471e-06,
      "loss": 0.0005,
      "step": 142980
    },
    {
      "epoch": 2.339687474433445,
      "grad_norm": 0.02242635004222393,
      "learning_rate": 1.4045723530704785e-06,
      "loss": 0.0006,
      "step": 142990
    },
    {
      "epoch": 2.339851100384521,
      "grad_norm": 0.04862105846405029,
      "learning_rate": 1.4039108945331225e-06,
      "loss": 0.0008,
      "step": 143000
    },
    {
      "epoch": 2.340014726335597,
      "grad_norm": 0.006551207974553108,
      "learning_rate": 1.4032495663453533e-06,
      "loss": 0.0009,
      "step": 143010
    },
    {
      "epoch": 2.3401783522866726,
      "grad_norm": 0.10910077393054962,
      "learning_rate": 1.402588368531141e-06,
      "loss": 0.0009,
      "step": 143020
    },
    {
      "epoch": 2.3403419782377486,
      "grad_norm": 0.002840406494215131,
      "learning_rate": 1.4019273011144557e-06,
      "loss": 0.0007,
      "step": 143030
    },
    {
      "epoch": 2.3405056041888246,
      "grad_norm": 0.11268679052591324,
      "learning_rate": 1.401266364119257e-06,
      "loss": 0.0007,
      "step": 143040
    },
    {
      "epoch": 2.3406692301399,
      "grad_norm": 0.12169431149959564,
      "learning_rate": 1.4006055575695044e-06,
      "loss": 0.0009,
      "step": 143050
    },
    {
      "epoch": 2.340832856090976,
      "grad_norm": 0.028649700805544853,
      "learning_rate": 1.3999448814891482e-06,
      "loss": 0.0007,
      "step": 143060
    },
    {
      "epoch": 2.3409964820420517,
      "grad_norm": 0.08962417393922806,
      "learning_rate": 1.3992843359021385e-06,
      "loss": 0.0009,
      "step": 143070
    },
    {
      "epoch": 2.3411601079931277,
      "grad_norm": 0.03773658722639084,
      "learning_rate": 1.3986239208324165e-06,
      "loss": 0.0013,
      "step": 143080
    },
    {
      "epoch": 2.3413237339442037,
      "grad_norm": 0.02422274462878704,
      "learning_rate": 1.3979636363039218e-06,
      "loss": 0.0007,
      "step": 143090
    },
    {
      "epoch": 2.3414873598952792,
      "grad_norm": 0.05366696044802666,
      "learning_rate": 1.3973034823405908e-06,
      "loss": 0.0007,
      "step": 143100
    },
    {
      "epoch": 2.3416509858463552,
      "grad_norm": 0.08595477044582367,
      "learning_rate": 1.3966434589663464e-06,
      "loss": 0.0008,
      "step": 143110
    },
    {
      "epoch": 2.3418146117974312,
      "grad_norm": 0.06815607100725174,
      "learning_rate": 1.395983566205118e-06,
      "loss": 0.0011,
      "step": 143120
    },
    {
      "epoch": 2.3419782377485068,
      "grad_norm": 0.03770340606570244,
      "learning_rate": 1.395323804080822e-06,
      "loss": 0.0007,
      "step": 143130
    },
    {
      "epoch": 2.3421418636995828,
      "grad_norm": 0.20763979852199554,
      "learning_rate": 1.3946641726173744e-06,
      "loss": 0.0014,
      "step": 143140
    },
    {
      "epoch": 2.3423054896506588,
      "grad_norm": 0.061255186796188354,
      "learning_rate": 1.3940046718386863e-06,
      "loss": 0.0005,
      "step": 143150
    },
    {
      "epoch": 2.3424691156017343,
      "grad_norm": 0.0543268658220768,
      "learning_rate": 1.393345301768661e-06,
      "loss": 0.0007,
      "step": 143160
    },
    {
      "epoch": 2.3426327415528103,
      "grad_norm": 0.08834155648946762,
      "learning_rate": 1.3926860624312017e-06,
      "loss": 0.0005,
      "step": 143170
    },
    {
      "epoch": 2.342796367503886,
      "grad_norm": 0.06812989711761475,
      "learning_rate": 1.3920269538502013e-06,
      "loss": 0.001,
      "step": 143180
    },
    {
      "epoch": 2.342959993454962,
      "grad_norm": 0.10159070044755936,
      "learning_rate": 1.3913679760495536e-06,
      "loss": 0.0008,
      "step": 143190
    },
    {
      "epoch": 2.343123619406038,
      "grad_norm": 0.17776423692703247,
      "learning_rate": 1.3907091290531433e-06,
      "loss": 0.0007,
      "step": 143200
    },
    {
      "epoch": 2.3432872453571134,
      "grad_norm": 0.052238188683986664,
      "learning_rate": 1.3900504128848512e-06,
      "loss": 0.0007,
      "step": 143210
    },
    {
      "epoch": 2.3434508713081894,
      "grad_norm": 0.007336069364100695,
      "learning_rate": 1.389391827568557e-06,
      "loss": 0.0004,
      "step": 143220
    },
    {
      "epoch": 2.3436144972592654,
      "grad_norm": 0.08986391127109528,
      "learning_rate": 1.3887333731281294e-06,
      "loss": 0.001,
      "step": 143230
    },
    {
      "epoch": 2.343778123210341,
      "grad_norm": 0.06561437249183655,
      "learning_rate": 1.3880750495874395e-06,
      "loss": 0.0006,
      "step": 143240
    },
    {
      "epoch": 2.343941749161417,
      "grad_norm": 0.017843183130025864,
      "learning_rate": 1.3874168569703461e-06,
      "loss": 0.0008,
      "step": 143250
    },
    {
      "epoch": 2.344105375112493,
      "grad_norm": 0.24887534976005554,
      "learning_rate": 1.3867587953007106e-06,
      "loss": 0.0006,
      "step": 143260
    },
    {
      "epoch": 2.3442690010635685,
      "grad_norm": 0.0790000706911087,
      "learning_rate": 1.3861008646023833e-06,
      "loss": 0.0014,
      "step": 143270
    },
    {
      "epoch": 2.3444326270146445,
      "grad_norm": 0.07997293025255203,
      "learning_rate": 1.385443064899215e-06,
      "loss": 0.0007,
      "step": 143280
    },
    {
      "epoch": 2.3445962529657205,
      "grad_norm": 0.10728421062231064,
      "learning_rate": 1.384785396215047e-06,
      "loss": 0.001,
      "step": 143290
    },
    {
      "epoch": 2.344759878916796,
      "grad_norm": 0.02984154038131237,
      "learning_rate": 1.384127858573721e-06,
      "loss": 0.001,
      "step": 143300
    },
    {
      "epoch": 2.344923504867872,
      "grad_norm": 0.13059760630130768,
      "learning_rate": 1.3834704519990695e-06,
      "loss": 0.0006,
      "step": 143310
    },
    {
      "epoch": 2.345087130818948,
      "grad_norm": 0.07913115620613098,
      "learning_rate": 1.3828131765149205e-06,
      "loss": 0.0007,
      "step": 143320
    },
    {
      "epoch": 2.3452507567700236,
      "grad_norm": 0.20202594995498657,
      "learning_rate": 1.3821560321451017e-06,
      "loss": 0.0008,
      "step": 143330
    },
    {
      "epoch": 2.3454143827210996,
      "grad_norm": 0.003404019633308053,
      "learning_rate": 1.3814990189134298e-06,
      "loss": 0.0008,
      "step": 143340
    },
    {
      "epoch": 2.3455780086721756,
      "grad_norm": 0.1013917624950409,
      "learning_rate": 1.3808421368437235e-06,
      "loss": 0.0009,
      "step": 143350
    },
    {
      "epoch": 2.345741634623251,
      "grad_norm": 0.01398392952978611,
      "learning_rate": 1.3801853859597897e-06,
      "loss": 0.0008,
      "step": 143360
    },
    {
      "epoch": 2.345905260574327,
      "grad_norm": 0.11806602030992508,
      "learning_rate": 1.3795287662854351e-06,
      "loss": 0.0004,
      "step": 143370
    },
    {
      "epoch": 2.346068886525403,
      "grad_norm": 0.0037658826913684607,
      "learning_rate": 1.3788722778444634e-06,
      "loss": 0.0008,
      "step": 143380
    },
    {
      "epoch": 2.3462325124764787,
      "grad_norm": 0.03889146447181702,
      "learning_rate": 1.3782159206606666e-06,
      "loss": 0.0017,
      "step": 143390
    },
    {
      "epoch": 2.3463961384275547,
      "grad_norm": 0.07284333556890488,
      "learning_rate": 1.3775596947578413e-06,
      "loss": 0.0007,
      "step": 143400
    },
    {
      "epoch": 2.3465597643786307,
      "grad_norm": 0.03644486516714096,
      "learning_rate": 1.3769036001597674e-06,
      "loss": 0.0005,
      "step": 143410
    },
    {
      "epoch": 2.3467233903297062,
      "grad_norm": 0.07116071879863739,
      "learning_rate": 1.37624763689023e-06,
      "loss": 0.0008,
      "step": 143420
    },
    {
      "epoch": 2.3468870162807822,
      "grad_norm": 0.03153816983103752,
      "learning_rate": 1.3755918049730083e-06,
      "loss": 0.0006,
      "step": 143430
    },
    {
      "epoch": 2.347050642231858,
      "grad_norm": 0.08310026675462723,
      "learning_rate": 1.3749361044318704e-06,
      "loss": 0.0007,
      "step": 143440
    },
    {
      "epoch": 2.3472142681829338,
      "grad_norm": 0.09923568367958069,
      "learning_rate": 1.3742805352905886e-06,
      "loss": 0.0008,
      "step": 143450
    },
    {
      "epoch": 2.3473778941340098,
      "grad_norm": 0.21217654645442963,
      "learning_rate": 1.373625097572921e-06,
      "loss": 0.0009,
      "step": 143460
    },
    {
      "epoch": 2.3475415200850853,
      "grad_norm": 0.022235210984945297,
      "learning_rate": 1.3729697913026297e-06,
      "loss": 0.0006,
      "step": 143470
    },
    {
      "epoch": 2.3477051460361613,
      "grad_norm": 0.10887635499238968,
      "learning_rate": 1.372314616503464e-06,
      "loss": 0.001,
      "step": 143480
    },
    {
      "epoch": 2.3478687719872373,
      "grad_norm": 0.008988043293356895,
      "learning_rate": 1.3716595731991765e-06,
      "loss": 0.0009,
      "step": 143490
    },
    {
      "epoch": 2.348032397938313,
      "grad_norm": 0.11141078174114227,
      "learning_rate": 1.3710046614135092e-06,
      "loss": 0.0006,
      "step": 143500
    },
    {
      "epoch": 2.348196023889389,
      "grad_norm": 0.030218739062547684,
      "learning_rate": 1.3703498811701992e-06,
      "loss": 0.0009,
      "step": 143510
    },
    {
      "epoch": 2.348359649840465,
      "grad_norm": 0.10009713470935822,
      "learning_rate": 1.3696952324929836e-06,
      "loss": 0.0005,
      "step": 143520
    },
    {
      "epoch": 2.3485232757915404,
      "grad_norm": 0.15290938317775726,
      "learning_rate": 1.3690407154055896e-06,
      "loss": 0.0007,
      "step": 143530
    },
    {
      "epoch": 2.3486869017426164,
      "grad_norm": 0.12174730002880096,
      "learning_rate": 1.368386329931744e-06,
      "loss": 0.0007,
      "step": 143540
    },
    {
      "epoch": 2.3488505276936924,
      "grad_norm": 0.13056187331676483,
      "learning_rate": 1.367732076095164e-06,
      "loss": 0.0012,
      "step": 143550
    },
    {
      "epoch": 2.349014153644768,
      "grad_norm": 0.09802248328924179,
      "learning_rate": 1.3670779539195677e-06,
      "loss": 0.001,
      "step": 143560
    },
    {
      "epoch": 2.349177779595844,
      "grad_norm": 0.12196998298168182,
      "learning_rate": 1.366423963428663e-06,
      "loss": 0.0011,
      "step": 143570
    },
    {
      "epoch": 2.3493414055469195,
      "grad_norm": 0.011269372887909412,
      "learning_rate": 1.3657701046461564e-06,
      "loss": 0.0007,
      "step": 143580
    },
    {
      "epoch": 2.3495050314979955,
      "grad_norm": 0.12164096534252167,
      "learning_rate": 1.3651163775957505e-06,
      "loss": 0.0009,
      "step": 143590
    },
    {
      "epoch": 2.3496686574490715,
      "grad_norm": 0.09820759296417236,
      "learning_rate": 1.3644627823011398e-06,
      "loss": 0.0013,
      "step": 143600
    },
    {
      "epoch": 2.349832283400147,
      "grad_norm": 0.08753691613674164,
      "learning_rate": 1.3638093187860151e-06,
      "loss": 0.0019,
      "step": 143610
    },
    {
      "epoch": 2.349995909351223,
      "grad_norm": 0.05304019898176193,
      "learning_rate": 1.3631559870740618e-06,
      "loss": 0.0015,
      "step": 143620
    },
    {
      "epoch": 2.350159535302299,
      "grad_norm": 0.10311562567949295,
      "learning_rate": 1.3625027871889645e-06,
      "loss": 0.0007,
      "step": 143630
    },
    {
      "epoch": 2.3503231612533746,
      "grad_norm": 0.010541926138103008,
      "learning_rate": 1.3618497191543979e-06,
      "loss": 0.0012,
      "step": 143640
    },
    {
      "epoch": 2.3504867872044506,
      "grad_norm": 0.09467317909002304,
      "learning_rate": 1.3611967829940342e-06,
      "loss": 0.0007,
      "step": 143650
    },
    {
      "epoch": 2.3506504131555266,
      "grad_norm": 0.0015669726999476552,
      "learning_rate": 1.3605439787315428e-06,
      "loss": 0.0011,
      "step": 143660
    },
    {
      "epoch": 2.350814039106602,
      "grad_norm": 0.1442929357290268,
      "learning_rate": 1.359891306390584e-06,
      "loss": 0.0012,
      "step": 143670
    },
    {
      "epoch": 2.350977665057678,
      "grad_norm": 0.043991636484861374,
      "learning_rate": 1.3592387659948175e-06,
      "loss": 0.0005,
      "step": 143680
    },
    {
      "epoch": 2.351141291008754,
      "grad_norm": 0.011225903406739235,
      "learning_rate": 1.3585863575678943e-06,
      "loss": 0.0009,
      "step": 143690
    },
    {
      "epoch": 2.3513049169598297,
      "grad_norm": 0.20907096564769745,
      "learning_rate": 1.357934081133465e-06,
      "loss": 0.0008,
      "step": 143700
    },
    {
      "epoch": 2.3514685429109057,
      "grad_norm": 0.049303993582725525,
      "learning_rate": 1.3572819367151724e-06,
      "loss": 0.001,
      "step": 143710
    },
    {
      "epoch": 2.3516321688619817,
      "grad_norm": 0.05887862294912338,
      "learning_rate": 1.3566299243366526e-06,
      "loss": 0.0004,
      "step": 143720
    },
    {
      "epoch": 2.3517957948130572,
      "grad_norm": 0.21594934165477753,
      "learning_rate": 1.355978044021543e-06,
      "loss": 0.0009,
      "step": 143730
    },
    {
      "epoch": 2.3519594207641332,
      "grad_norm": 0.058085665106773376,
      "learning_rate": 1.3553262957934693e-06,
      "loss": 0.0003,
      "step": 143740
    },
    {
      "epoch": 2.352123046715209,
      "grad_norm": 0.07392590492963791,
      "learning_rate": 1.3546746796760596e-06,
      "loss": 0.0008,
      "step": 143750
    },
    {
      "epoch": 2.3522866726662848,
      "grad_norm": 0.02047325111925602,
      "learning_rate": 1.3540231956929295e-06,
      "loss": 0.0009,
      "step": 143760
    },
    {
      "epoch": 2.3524502986173608,
      "grad_norm": 0.05070842429995537,
      "learning_rate": 1.3533718438676974e-06,
      "loss": 0.0006,
      "step": 143770
    },
    {
      "epoch": 2.3526139245684368,
      "grad_norm": 0.019670866429805756,
      "learning_rate": 1.3527206242239704e-06,
      "loss": 0.0005,
      "step": 143780
    },
    {
      "epoch": 2.3527775505195123,
      "grad_norm": 0.1090909093618393,
      "learning_rate": 1.352069536785356e-06,
      "loss": 0.0013,
      "step": 143790
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 0.08506675064563751,
      "learning_rate": 1.3514185815754516e-06,
      "loss": 0.0006,
      "step": 143800
    },
    {
      "epoch": 2.3531048024216643,
      "grad_norm": 0.060148466378450394,
      "learning_rate": 1.3507677586178559e-06,
      "loss": 0.0009,
      "step": 143810
    },
    {
      "epoch": 2.35326842837274,
      "grad_norm": 0.00532922288402915,
      "learning_rate": 1.3501170679361586e-06,
      "loss": 0.0008,
      "step": 143820
    },
    {
      "epoch": 2.353432054323816,
      "grad_norm": 0.45534756779670715,
      "learning_rate": 1.3494665095539438e-06,
      "loss": 0.001,
      "step": 143830
    },
    {
      "epoch": 2.3535956802748914,
      "grad_norm": 0.1013123095035553,
      "learning_rate": 1.3488160834947955e-06,
      "loss": 0.0008,
      "step": 143840
    },
    {
      "epoch": 2.3537593062259674,
      "grad_norm": 0.060462888330221176,
      "learning_rate": 1.3481657897822876e-06,
      "loss": 0.0009,
      "step": 143850
    },
    {
      "epoch": 2.3539229321770434,
      "grad_norm": 0.11193085461854935,
      "learning_rate": 1.3475156284399927e-06,
      "loss": 0.0008,
      "step": 143860
    },
    {
      "epoch": 2.354086558128119,
      "grad_norm": 0.007070471066981554,
      "learning_rate": 1.3468655994914792e-06,
      "loss": 0.0021,
      "step": 143870
    },
    {
      "epoch": 2.354250184079195,
      "grad_norm": 0.0891304761171341,
      "learning_rate": 1.3462157029603057e-06,
      "loss": 0.0011,
      "step": 143880
    },
    {
      "epoch": 2.354413810030271,
      "grad_norm": 0.325884073972702,
      "learning_rate": 1.3455659388700337e-06,
      "loss": 0.0017,
      "step": 143890
    },
    {
      "epoch": 2.3545774359813465,
      "grad_norm": 0.06715056300163269,
      "learning_rate": 1.3449163072442112e-06,
      "loss": 0.0004,
      "step": 143900
    },
    {
      "epoch": 2.3547410619324225,
      "grad_norm": 0.039136070758104324,
      "learning_rate": 1.3442668081063908e-06,
      "loss": 0.001,
      "step": 143910
    },
    {
      "epoch": 2.3549046878834985,
      "grad_norm": 0.007674661464989185,
      "learning_rate": 1.3436174414801095e-06,
      "loss": 0.0004,
      "step": 143920
    },
    {
      "epoch": 2.355068313834574,
      "grad_norm": 0.03370317444205284,
      "learning_rate": 1.3429682073889078e-06,
      "loss": 0.0005,
      "step": 143930
    },
    {
      "epoch": 2.35523193978565,
      "grad_norm": 0.021954510360956192,
      "learning_rate": 1.3423191058563207e-06,
      "loss": 0.0011,
      "step": 143940
    },
    {
      "epoch": 2.355395565736726,
      "grad_norm": 0.01823018118739128,
      "learning_rate": 1.3416701369058739e-06,
      "loss": 0.0013,
      "step": 143950
    },
    {
      "epoch": 2.3555591916878016,
      "grad_norm": 0.04143868014216423,
      "learning_rate": 1.3410213005610928e-06,
      "loss": 0.0005,
      "step": 143960
    },
    {
      "epoch": 2.3557228176388776,
      "grad_norm": 0.07386468350887299,
      "learning_rate": 1.3403725968454945e-06,
      "loss": 0.0008,
      "step": 143970
    },
    {
      "epoch": 2.355886443589953,
      "grad_norm": 0.03366195783019066,
      "learning_rate": 1.3397240257825956e-06,
      "loss": 0.0014,
      "step": 143980
    },
    {
      "epoch": 2.356050069541029,
      "grad_norm": 0.007476123981177807,
      "learning_rate": 1.3390755873959016e-06,
      "loss": 0.0008,
      "step": 143990
    },
    {
      "epoch": 2.356213695492105,
      "grad_norm": 0.03846132382750511,
      "learning_rate": 1.33842728170892e-06,
      "loss": 0.0007,
      "step": 144000
    },
    {
      "epoch": 2.3563773214431807,
      "grad_norm": 0.038211554288864136,
      "learning_rate": 1.3377791087451497e-06,
      "loss": 0.0009,
      "step": 144010
    },
    {
      "epoch": 2.3565409473942567,
      "grad_norm": 0.06402503699064255,
      "learning_rate": 1.337131068528083e-06,
      "loss": 0.0012,
      "step": 144020
    },
    {
      "epoch": 2.3567045733453327,
      "grad_norm": 0.009781122207641602,
      "learning_rate": 1.336483161081213e-06,
      "loss": 0.0002,
      "step": 144030
    },
    {
      "epoch": 2.3568681992964082,
      "grad_norm": 0.03583794832229614,
      "learning_rate": 1.335835386428022e-06,
      "loss": 0.0007,
      "step": 144040
    },
    {
      "epoch": 2.3570318252474842,
      "grad_norm": 0.3033040463924408,
      "learning_rate": 1.335187744591993e-06,
      "loss": 0.0007,
      "step": 144050
    },
    {
      "epoch": 2.35719545119856,
      "grad_norm": 0.13907165825366974,
      "learning_rate": 1.3345402355965992e-06,
      "loss": 0.0009,
      "step": 144060
    },
    {
      "epoch": 2.3573590771496358,
      "grad_norm": 0.16955947875976562,
      "learning_rate": 1.3338928594653128e-06,
      "loss": 0.0009,
      "step": 144070
    },
    {
      "epoch": 2.3575227031007118,
      "grad_norm": 0.08944796025753021,
      "learning_rate": 1.3332456162215979e-06,
      "loss": 0.001,
      "step": 144080
    },
    {
      "epoch": 2.3576863290517878,
      "grad_norm": 0.11608299612998962,
      "learning_rate": 1.3325985058889168e-06,
      "loss": 0.0006,
      "step": 144090
    },
    {
      "epoch": 2.3578499550028633,
      "grad_norm": 0.0029833803419023752,
      "learning_rate": 1.3319515284907269e-06,
      "loss": 0.0004,
      "step": 144100
    },
    {
      "epoch": 2.3580135809539393,
      "grad_norm": 0.04761983081698418,
      "learning_rate": 1.3313046840504778e-06,
      "loss": 0.0005,
      "step": 144110
    },
    {
      "epoch": 2.3581772069050153,
      "grad_norm": 0.023272233083844185,
      "learning_rate": 1.330657972591617e-06,
      "loss": 0.0007,
      "step": 144120
    },
    {
      "epoch": 2.358340832856091,
      "grad_norm": 0.17743664979934692,
      "learning_rate": 1.3300113941375836e-06,
      "loss": 0.0007,
      "step": 144130
    },
    {
      "epoch": 2.358504458807167,
      "grad_norm": 0.06324001401662827,
      "learning_rate": 1.329364948711817e-06,
      "loss": 0.0011,
      "step": 144140
    },
    {
      "epoch": 2.358668084758243,
      "grad_norm": 0.11405488103628159,
      "learning_rate": 1.3287186363377503e-06,
      "loss": 0.0006,
      "step": 144150
    },
    {
      "epoch": 2.3588317107093184,
      "grad_norm": 0.11704226583242416,
      "learning_rate": 1.3280724570388076e-06,
      "loss": 0.0008,
      "step": 144160
    },
    {
      "epoch": 2.3589953366603944,
      "grad_norm": 0.15636445581912994,
      "learning_rate": 1.3274264108384149e-06,
      "loss": 0.0007,
      "step": 144170
    },
    {
      "epoch": 2.3591589626114704,
      "grad_norm": 0.10471849143505096,
      "learning_rate": 1.3267804977599864e-06,
      "loss": 0.001,
      "step": 144180
    },
    {
      "epoch": 2.359322588562546,
      "grad_norm": 0.011177187785506248,
      "learning_rate": 1.3261347178269379e-06,
      "loss": 0.0007,
      "step": 144190
    },
    {
      "epoch": 2.359486214513622,
      "grad_norm": 0.08316013962030411,
      "learning_rate": 1.3254890710626744e-06,
      "loss": 0.0009,
      "step": 144200
    },
    {
      "epoch": 2.359649840464698,
      "grad_norm": 0.17139707505702972,
      "learning_rate": 1.324843557490602e-06,
      "loss": 0.0007,
      "step": 144210
    },
    {
      "epoch": 2.3598134664157735,
      "grad_norm": 0.07555080950260162,
      "learning_rate": 1.3241981771341177e-06,
      "loss": 0.001,
      "step": 144220
    },
    {
      "epoch": 2.3599770923668495,
      "grad_norm": 0.09622034430503845,
      "learning_rate": 1.323552930016614e-06,
      "loss": 0.0005,
      "step": 144230
    },
    {
      "epoch": 2.360140718317925,
      "grad_norm": 0.0039297244511544704,
      "learning_rate": 1.3229078161614816e-06,
      "loss": 0.0007,
      "step": 144240
    },
    {
      "epoch": 2.360304344269001,
      "grad_norm": 0.05190461501479149,
      "learning_rate": 1.3222628355921014e-06,
      "loss": 0.0022,
      "step": 144250
    },
    {
      "epoch": 2.360467970220077,
      "grad_norm": 0.11169778555631638,
      "learning_rate": 1.3216179883318563e-06,
      "loss": 0.0012,
      "step": 144260
    },
    {
      "epoch": 2.3606315961711526,
      "grad_norm": 0.06501086056232452,
      "learning_rate": 1.3209732744041164e-06,
      "loss": 0.0008,
      "step": 144270
    },
    {
      "epoch": 2.3607952221222286,
      "grad_norm": 0.030380522832274437,
      "learning_rate": 1.3203286938322546e-06,
      "loss": 0.0005,
      "step": 144280
    },
    {
      "epoch": 2.3609588480733046,
      "grad_norm": 0.19924689829349518,
      "learning_rate": 1.3196842466396326e-06,
      "loss": 0.0011,
      "step": 144290
    },
    {
      "epoch": 2.36112247402438,
      "grad_norm": 0.003139902837574482,
      "learning_rate": 1.3190399328496107e-06,
      "loss": 0.0005,
      "step": 144300
    },
    {
      "epoch": 2.361286099975456,
      "grad_norm": 0.11221468448638916,
      "learning_rate": 1.3183957524855479e-06,
      "loss": 0.0014,
      "step": 144310
    },
    {
      "epoch": 2.361449725926532,
      "grad_norm": 0.05913063511252403,
      "learning_rate": 1.3177517055707866e-06,
      "loss": 0.0005,
      "step": 144320
    },
    {
      "epoch": 2.3616133518776077,
      "grad_norm": 0.10795065760612488,
      "learning_rate": 1.3171077921286784e-06,
      "loss": 0.0006,
      "step": 144330
    },
    {
      "epoch": 2.3617769778286837,
      "grad_norm": 0.07169833034276962,
      "learning_rate": 1.3164640121825594e-06,
      "loss": 0.0004,
      "step": 144340
    },
    {
      "epoch": 2.3619406037797592,
      "grad_norm": 0.0510670505464077,
      "learning_rate": 1.3158203657557667e-06,
      "loss": 0.0006,
      "step": 144350
    },
    {
      "epoch": 2.3621042297308352,
      "grad_norm": 0.07229570299386978,
      "learning_rate": 1.3151768528716325e-06,
      "loss": 0.0009,
      "step": 144360
    },
    {
      "epoch": 2.362267855681911,
      "grad_norm": 0.19994962215423584,
      "learning_rate": 1.3145334735534798e-06,
      "loss": 0.0011,
      "step": 144370
    },
    {
      "epoch": 2.3624314816329868,
      "grad_norm": 0.31083160638809204,
      "learning_rate": 1.3138902278246324e-06,
      "loss": 0.0013,
      "step": 144380
    },
    {
      "epoch": 2.3625951075840628,
      "grad_norm": 0.18726523220539093,
      "learning_rate": 1.3132471157084032e-06,
      "loss": 0.0007,
      "step": 144390
    },
    {
      "epoch": 2.3627587335351388,
      "grad_norm": 0.06729884445667267,
      "learning_rate": 1.3126041372281068e-06,
      "loss": 0.0009,
      "step": 144400
    },
    {
      "epoch": 2.3629223594862143,
      "grad_norm": 0.004335499834269285,
      "learning_rate": 1.311961292407048e-06,
      "loss": 0.0009,
      "step": 144410
    },
    {
      "epoch": 2.3630859854372903,
      "grad_norm": 0.214744433760643,
      "learning_rate": 1.311318581268527e-06,
      "loss": 0.0007,
      "step": 144420
    },
    {
      "epoch": 2.3632496113883663,
      "grad_norm": 0.05723525956273079,
      "learning_rate": 1.3106760038358429e-06,
      "loss": 0.0008,
      "step": 144430
    },
    {
      "epoch": 2.363413237339442,
      "grad_norm": 0.06948518753051758,
      "learning_rate": 1.3100335601322855e-06,
      "loss": 0.0005,
      "step": 144440
    },
    {
      "epoch": 2.363576863290518,
      "grad_norm": 0.005343471188098192,
      "learning_rate": 1.3093912501811445e-06,
      "loss": 0.0006,
      "step": 144450
    },
    {
      "epoch": 2.363740489241594,
      "grad_norm": 0.18003512918949127,
      "learning_rate": 1.308749074005699e-06,
      "loss": 0.0005,
      "step": 144460
    },
    {
      "epoch": 2.3639041151926694,
      "grad_norm": 0.10720117390155792,
      "learning_rate": 1.3081070316292294e-06,
      "loss": 0.0007,
      "step": 144470
    },
    {
      "epoch": 2.3640677411437454,
      "grad_norm": 0.018040727823972702,
      "learning_rate": 1.3074651230750052e-06,
      "loss": 0.0003,
      "step": 144480
    },
    {
      "epoch": 2.3642313670948214,
      "grad_norm": 0.009498990140855312,
      "learning_rate": 1.3068233483662968e-06,
      "loss": 0.0002,
      "step": 144490
    },
    {
      "epoch": 2.364394993045897,
      "grad_norm": 0.0802672803401947,
      "learning_rate": 1.3061817075263644e-06,
      "loss": 0.0011,
      "step": 144500
    },
    {
      "epoch": 2.364558618996973,
      "grad_norm": 0.010741434060037136,
      "learning_rate": 1.3055402005784684e-06,
      "loss": 0.0006,
      "step": 144510
    },
    {
      "epoch": 2.364722244948049,
      "grad_norm": 0.07852223515510559,
      "learning_rate": 1.3048988275458614e-06,
      "loss": 0.0015,
      "step": 144520
    },
    {
      "epoch": 2.3648858708991245,
      "grad_norm": 0.03402560576796532,
      "learning_rate": 1.3042575884517888e-06,
      "loss": 0.0007,
      "step": 144530
    },
    {
      "epoch": 2.3650494968502005,
      "grad_norm": 0.07025851309299469,
      "learning_rate": 1.3036164833194982e-06,
      "loss": 0.0011,
      "step": 144540
    },
    {
      "epoch": 2.3652131228012765,
      "grad_norm": 0.11702341586351395,
      "learning_rate": 1.3029755121722242e-06,
      "loss": 0.0008,
      "step": 144550
    },
    {
      "epoch": 2.365376748752352,
      "grad_norm": 0.03939018025994301,
      "learning_rate": 1.3023346750332038e-06,
      "loss": 0.0005,
      "step": 144560
    },
    {
      "epoch": 2.365540374703428,
      "grad_norm": 0.07736250758171082,
      "learning_rate": 1.3016939719256628e-06,
      "loss": 0.0003,
      "step": 144570
    },
    {
      "epoch": 2.365704000654504,
      "grad_norm": 0.09059028327465057,
      "learning_rate": 1.3010534028728267e-06,
      "loss": 0.0014,
      "step": 144580
    },
    {
      "epoch": 2.3658676266055796,
      "grad_norm": 0.10941189527511597,
      "learning_rate": 1.3004129678979161e-06,
      "loss": 0.0008,
      "step": 144590
    },
    {
      "epoch": 2.3660312525566556,
      "grad_norm": 0.04468689113855362,
      "learning_rate": 1.2997726670241422e-06,
      "loss": 0.0009,
      "step": 144600
    },
    {
      "epoch": 2.366194878507731,
      "grad_norm": 0.2643345892429352,
      "learning_rate": 1.299132500274719e-06,
      "loss": 0.0012,
      "step": 144610
    },
    {
      "epoch": 2.366358504458807,
      "grad_norm": 0.1810191571712494,
      "learning_rate": 1.2984924676728449e-06,
      "loss": 0.0008,
      "step": 144620
    },
    {
      "epoch": 2.366522130409883,
      "grad_norm": 0.06795274466276169,
      "learning_rate": 1.2978525692417221e-06,
      "loss": 0.0005,
      "step": 144630
    },
    {
      "epoch": 2.3666857563609587,
      "grad_norm": 0.09930314123630524,
      "learning_rate": 1.297212805004548e-06,
      "loss": 0.0006,
      "step": 144640
    },
    {
      "epoch": 2.3668493823120347,
      "grad_norm": 0.0886782556772232,
      "learning_rate": 1.2965731749845078e-06,
      "loss": 0.0009,
      "step": 144650
    },
    {
      "epoch": 2.3670130082631107,
      "grad_norm": 0.053361061960458755,
      "learning_rate": 1.2959336792047916e-06,
      "loss": 0.0004,
      "step": 144660
    },
    {
      "epoch": 2.3671766342141862,
      "grad_norm": 0.013816573657095432,
      "learning_rate": 1.2952943176885746e-06,
      "loss": 0.0004,
      "step": 144670
    },
    {
      "epoch": 2.3673402601652622,
      "grad_norm": 0.06315802037715912,
      "learning_rate": 1.2946550904590366e-06,
      "loss": 0.001,
      "step": 144680
    },
    {
      "epoch": 2.367503886116338,
      "grad_norm": 0.023802505806088448,
      "learning_rate": 1.2940159975393446e-06,
      "loss": 0.0003,
      "step": 144690
    },
    {
      "epoch": 2.3676675120674138,
      "grad_norm": 0.07392996549606323,
      "learning_rate": 1.2933770389526668e-06,
      "loss": 0.0008,
      "step": 144700
    },
    {
      "epoch": 2.3678311380184898,
      "grad_norm": 0.0754457488656044,
      "learning_rate": 1.292738214722163e-06,
      "loss": 0.001,
      "step": 144710
    },
    {
      "epoch": 2.3679947639695658,
      "grad_norm": 0.033821530640125275,
      "learning_rate": 1.2920995248709873e-06,
      "loss": 0.0009,
      "step": 144720
    },
    {
      "epoch": 2.3681583899206413,
      "grad_norm": 0.10110820829868317,
      "learning_rate": 1.291460969422293e-06,
      "loss": 0.0007,
      "step": 144730
    },
    {
      "epoch": 2.3683220158717173,
      "grad_norm": 0.042355962097644806,
      "learning_rate": 1.2908225483992242e-06,
      "loss": 0.0007,
      "step": 144740
    },
    {
      "epoch": 2.368485641822793,
      "grad_norm": 0.03476399928331375,
      "learning_rate": 1.2901842618249249e-06,
      "loss": 0.0017,
      "step": 144750
    },
    {
      "epoch": 2.368649267773869,
      "grad_norm": 0.1271200031042099,
      "learning_rate": 1.2895461097225282e-06,
      "loss": 0.0004,
      "step": 144760
    },
    {
      "epoch": 2.368812893724945,
      "grad_norm": 0.057854343205690384,
      "learning_rate": 1.2889080921151687e-06,
      "loss": 0.0008,
      "step": 144770
    },
    {
      "epoch": 2.3689765196760204,
      "grad_norm": 0.031231729313731194,
      "learning_rate": 1.2882702090259697e-06,
      "loss": 0.0007,
      "step": 144780
    },
    {
      "epoch": 2.3691401456270964,
      "grad_norm": 0.09763874858617783,
      "learning_rate": 1.287632460478055e-06,
      "loss": 0.0007,
      "step": 144790
    },
    {
      "epoch": 2.3693037715781724,
      "grad_norm": 0.0897286906838417,
      "learning_rate": 1.2869948464945426e-06,
      "loss": 0.0008,
      "step": 144800
    },
    {
      "epoch": 2.369467397529248,
      "grad_norm": 0.09197789430618286,
      "learning_rate": 1.2863573670985429e-06,
      "loss": 0.0006,
      "step": 144810
    },
    {
      "epoch": 2.369631023480324,
      "grad_norm": 0.02861623279750347,
      "learning_rate": 1.2857200223131634e-06,
      "loss": 0.0009,
      "step": 144820
    },
    {
      "epoch": 2.3697946494314,
      "grad_norm": 0.055444080382585526,
      "learning_rate": 1.2850828121615045e-06,
      "loss": 0.0006,
      "step": 144830
    },
    {
      "epoch": 2.3699582753824755,
      "grad_norm": 0.0253133587539196,
      "learning_rate": 1.2844457366666662e-06,
      "loss": 0.0003,
      "step": 144840
    },
    {
      "epoch": 2.3701219013335515,
      "grad_norm": 0.0453915111720562,
      "learning_rate": 1.2838087958517388e-06,
      "loss": 0.0009,
      "step": 144850
    },
    {
      "epoch": 2.3702855272846275,
      "grad_norm": 0.18144530057907104,
      "learning_rate": 1.2831719897398108e-06,
      "loss": 0.0006,
      "step": 144860
    },
    {
      "epoch": 2.370449153235703,
      "grad_norm": 0.06157657131552696,
      "learning_rate": 1.2825353183539662e-06,
      "loss": 0.0012,
      "step": 144870
    },
    {
      "epoch": 2.370612779186779,
      "grad_norm": 0.10179923474788666,
      "learning_rate": 1.2818987817172807e-06,
      "loss": 0.0009,
      "step": 144880
    },
    {
      "epoch": 2.370776405137855,
      "grad_norm": 0.12163715809583664,
      "learning_rate": 1.2812623798528295e-06,
      "loss": 0.0004,
      "step": 144890
    },
    {
      "epoch": 2.3709400310889306,
      "grad_norm": 0.06507319211959839,
      "learning_rate": 1.2806261127836778e-06,
      "loss": 0.0006,
      "step": 144900
    },
    {
      "epoch": 2.3711036570400066,
      "grad_norm": 0.10177026689052582,
      "learning_rate": 1.2799899805328914e-06,
      "loss": 0.0008,
      "step": 144910
    },
    {
      "epoch": 2.3712672829910826,
      "grad_norm": 0.09454219788312912,
      "learning_rate": 1.2793539831235275e-06,
      "loss": 0.0006,
      "step": 144920
    },
    {
      "epoch": 2.371430908942158,
      "grad_norm": 0.08566269278526306,
      "learning_rate": 1.278718120578638e-06,
      "loss": 0.0009,
      "step": 144930
    },
    {
      "epoch": 2.371594534893234,
      "grad_norm": 0.04897194355726242,
      "learning_rate": 1.2780823929212743e-06,
      "loss": 0.001,
      "step": 144940
    },
    {
      "epoch": 2.37175816084431,
      "grad_norm": 0.10678929835557938,
      "learning_rate": 1.2774468001744767e-06,
      "loss": 0.0007,
      "step": 144950
    },
    {
      "epoch": 2.3719217867953857,
      "grad_norm": 0.11534004658460617,
      "learning_rate": 1.2768113423612871e-06,
      "loss": 0.0007,
      "step": 144960
    },
    {
      "epoch": 2.3720854127464617,
      "grad_norm": 0.026602929458022118,
      "learning_rate": 1.2761760195047369e-06,
      "loss": 0.0011,
      "step": 144970
    },
    {
      "epoch": 2.3722490386975377,
      "grad_norm": 0.049671001732349396,
      "learning_rate": 1.2755408316278573e-06,
      "loss": 0.001,
      "step": 144980
    },
    {
      "epoch": 2.3724126646486132,
      "grad_norm": 0.05801291763782501,
      "learning_rate": 1.2749057787536694e-06,
      "loss": 0.0007,
      "step": 144990
    },
    {
      "epoch": 2.372576290599689,
      "grad_norm": 0.12459327280521393,
      "learning_rate": 1.2742708609051957e-06,
      "loss": 0.0007,
      "step": 145000
    },
    {
      "epoch": 2.3727399165507648,
      "grad_norm": 0.11902379989624023,
      "learning_rate": 1.2736360781054486e-06,
      "loss": 0.0007,
      "step": 145010
    },
    {
      "epoch": 2.3729035425018408,
      "grad_norm": 0.10373744368553162,
      "learning_rate": 1.2730014303774358e-06,
      "loss": 0.001,
      "step": 145020
    },
    {
      "epoch": 2.3730671684529168,
      "grad_norm": 0.0296646561473608,
      "learning_rate": 1.2723669177441655e-06,
      "loss": 0.0009,
      "step": 145030
    },
    {
      "epoch": 2.3732307944039923,
      "grad_norm": 0.010697804391384125,
      "learning_rate": 1.2717325402286334e-06,
      "loss": 0.0007,
      "step": 145040
    },
    {
      "epoch": 2.3733944203550683,
      "grad_norm": 0.007616598159074783,
      "learning_rate": 1.2710982978538383e-06,
      "loss": 0.0007,
      "step": 145050
    },
    {
      "epoch": 2.3735580463061443,
      "grad_norm": 0.037158578634262085,
      "learning_rate": 1.2704641906427655e-06,
      "loss": 0.0014,
      "step": 145060
    },
    {
      "epoch": 2.37372167225722,
      "grad_norm": 0.019848110154271126,
      "learning_rate": 1.2698302186184025e-06,
      "loss": 0.0006,
      "step": 145070
    },
    {
      "epoch": 2.373885298208296,
      "grad_norm": 0.04519448056817055,
      "learning_rate": 1.26919638180373e-06,
      "loss": 0.001,
      "step": 145080
    },
    {
      "epoch": 2.374048924159372,
      "grad_norm": 0.05504337698221207,
      "learning_rate": 1.2685626802217206e-06,
      "loss": 0.0008,
      "step": 145090
    },
    {
      "epoch": 2.3742125501104474,
      "grad_norm": 0.10191631317138672,
      "learning_rate": 1.2679291138953475e-06,
      "loss": 0.0008,
      "step": 145100
    },
    {
      "epoch": 2.3743761760615234,
      "grad_norm": 0.11442823708057404,
      "learning_rate": 1.2672956828475735e-06,
      "loss": 0.0016,
      "step": 145110
    },
    {
      "epoch": 2.374539802012599,
      "grad_norm": 0.03787439689040184,
      "learning_rate": 1.2666623871013605e-06,
      "loss": 0.0006,
      "step": 145120
    },
    {
      "epoch": 2.374703427963675,
      "grad_norm": 0.015325605869293213,
      "learning_rate": 1.2660292266796615e-06,
      "loss": 0.0006,
      "step": 145130
    },
    {
      "epoch": 2.374867053914751,
      "grad_norm": 0.054970066994428635,
      "learning_rate": 1.2653962016054283e-06,
      "loss": 0.001,
      "step": 145140
    },
    {
      "epoch": 2.3750306798658265,
      "grad_norm": 0.1488233059644699,
      "learning_rate": 1.2647633119016085e-06,
      "loss": 0.0007,
      "step": 145150
    },
    {
      "epoch": 2.3751943058169025,
      "grad_norm": 0.08479508012533188,
      "learning_rate": 1.26413055759114e-06,
      "loss": 0.0008,
      "step": 145160
    },
    {
      "epoch": 2.3753579317679785,
      "grad_norm": 0.009266995824873447,
      "learning_rate": 1.263497938696961e-06,
      "loss": 0.0004,
      "step": 145170
    },
    {
      "epoch": 2.375521557719054,
      "grad_norm": 0.061846502125263214,
      "learning_rate": 1.262865455242e-06,
      "loss": 0.0006,
      "step": 145180
    },
    {
      "epoch": 2.37568518367013,
      "grad_norm": 0.026510871946811676,
      "learning_rate": 1.2622331072491855e-06,
      "loss": 0.0006,
      "step": 145190
    },
    {
      "epoch": 2.375848809621206,
      "grad_norm": 0.06797093152999878,
      "learning_rate": 1.261600894741436e-06,
      "loss": 0.0005,
      "step": 145200
    },
    {
      "epoch": 2.3760124355722816,
      "grad_norm": 0.06298942863941193,
      "learning_rate": 1.2609688177416707e-06,
      "loss": 0.0005,
      "step": 145210
    },
    {
      "epoch": 2.3761760615233576,
      "grad_norm": 0.1795349419116974,
      "learning_rate": 1.2603368762727992e-06,
      "loss": 0.0008,
      "step": 145220
    },
    {
      "epoch": 2.3763396874744336,
      "grad_norm": 0.030167920514941216,
      "learning_rate": 1.2597050703577263e-06,
      "loss": 0.0008,
      "step": 145230
    },
    {
      "epoch": 2.376503313425509,
      "grad_norm": 0.026989977806806564,
      "learning_rate": 1.2590734000193566e-06,
      "loss": 0.0005,
      "step": 145240
    },
    {
      "epoch": 2.376666939376585,
      "grad_norm": 0.11359591782093048,
      "learning_rate": 1.2584418652805835e-06,
      "loss": 0.0016,
      "step": 145250
    },
    {
      "epoch": 2.376830565327661,
      "grad_norm": 0.07843035459518433,
      "learning_rate": 1.257810466164302e-06,
      "loss": 0.0007,
      "step": 145260
    },
    {
      "epoch": 2.3769941912787367,
      "grad_norm": 0.051659245043992996,
      "learning_rate": 1.2571792026933955e-06,
      "loss": 0.0004,
      "step": 145270
    },
    {
      "epoch": 2.3771578172298127,
      "grad_norm": 0.016711194068193436,
      "learning_rate": 1.2565480748907482e-06,
      "loss": 0.0012,
      "step": 145280
    },
    {
      "epoch": 2.3773214431808887,
      "grad_norm": 0.01970168948173523,
      "learning_rate": 1.255917082779235e-06,
      "loss": 0.0006,
      "step": 145290
    },
    {
      "epoch": 2.3774850691319642,
      "grad_norm": 0.005773472134023905,
      "learning_rate": 1.2552862263817288e-06,
      "loss": 0.0006,
      "step": 145300
    },
    {
      "epoch": 2.37764869508304,
      "grad_norm": 0.14136260747909546,
      "learning_rate": 1.2546555057210996e-06,
      "loss": 0.0009,
      "step": 145310
    },
    {
      "epoch": 2.377812321034116,
      "grad_norm": 0.06797131896018982,
      "learning_rate": 1.2540249208202038e-06,
      "loss": 0.0011,
      "step": 145320
    },
    {
      "epoch": 2.3779759469851918,
      "grad_norm": 0.13964861631393433,
      "learning_rate": 1.2533944717019026e-06,
      "loss": 0.001,
      "step": 145330
    },
    {
      "epoch": 2.3781395729362678,
      "grad_norm": 0.15940672159194946,
      "learning_rate": 1.252764158389046e-06,
      "loss": 0.0013,
      "step": 145340
    },
    {
      "epoch": 2.3783031988873438,
      "grad_norm": 0.005445521324872971,
      "learning_rate": 1.252133980904482e-06,
      "loss": 0.0011,
      "step": 145350
    },
    {
      "epoch": 2.3784668248384193,
      "grad_norm": 0.0383513979613781,
      "learning_rate": 1.2515039392710548e-06,
      "loss": 0.0014,
      "step": 145360
    },
    {
      "epoch": 2.3786304507894953,
      "grad_norm": 0.06214066594839096,
      "learning_rate": 1.2508740335116e-06,
      "loss": 0.0006,
      "step": 145370
    },
    {
      "epoch": 2.378794076740571,
      "grad_norm": 0.0362640880048275,
      "learning_rate": 1.250244263648951e-06,
      "loss": 0.0004,
      "step": 145380
    },
    {
      "epoch": 2.378957702691647,
      "grad_norm": 0.08192306011915207,
      "learning_rate": 1.249614629705934e-06,
      "loss": 0.0004,
      "step": 145390
    },
    {
      "epoch": 2.379121328642723,
      "grad_norm": 0.0460490882396698,
      "learning_rate": 1.2489851317053747e-06,
      "loss": 0.0005,
      "step": 145400
    },
    {
      "epoch": 2.3792849545937984,
      "grad_norm": 0.006301357876509428,
      "learning_rate": 1.2483557696700888e-06,
      "loss": 0.0006,
      "step": 145410
    },
    {
      "epoch": 2.3794485805448744,
      "grad_norm": 0.1867118626832962,
      "learning_rate": 1.247726543622888e-06,
      "loss": 0.0012,
      "step": 145420
    },
    {
      "epoch": 2.3796122064959504,
      "grad_norm": 0.04853400960564613,
      "learning_rate": 1.247097453586583e-06,
      "loss": 0.0006,
      "step": 145430
    },
    {
      "epoch": 2.379775832447026,
      "grad_norm": 0.08408501744270325,
      "learning_rate": 1.2464684995839744e-06,
      "loss": 0.0004,
      "step": 145440
    },
    {
      "epoch": 2.379939458398102,
      "grad_norm": 0.032083846628665924,
      "learning_rate": 1.2458396816378627e-06,
      "loss": 0.0007,
      "step": 145450
    },
    {
      "epoch": 2.380103084349178,
      "grad_norm": 0.07761266082525253,
      "learning_rate": 1.2452109997710377e-06,
      "loss": 0.0012,
      "step": 145460
    },
    {
      "epoch": 2.3802667103002535,
      "grad_norm": 0.09874489903450012,
      "learning_rate": 1.2445824540062911e-06,
      "loss": 0.001,
      "step": 145470
    },
    {
      "epoch": 2.3804303362513295,
      "grad_norm": 0.026025863364338875,
      "learning_rate": 1.2439540443664034e-06,
      "loss": 0.0006,
      "step": 145480
    },
    {
      "epoch": 2.3805939622024055,
      "grad_norm": 0.05554347485303879,
      "learning_rate": 1.243325770874156e-06,
      "loss": 0.001,
      "step": 145490
    },
    {
      "epoch": 2.380757588153481,
      "grad_norm": 0.0900362953543663,
      "learning_rate": 1.2426976335523184e-06,
      "loss": 0.001,
      "step": 145500
    },
    {
      "epoch": 2.380921214104557,
      "grad_norm": 0.06566480547189713,
      "learning_rate": 1.2420696324236625e-06,
      "loss": 0.0011,
      "step": 145510
    },
    {
      "epoch": 2.3810848400556326,
      "grad_norm": 0.07317785918712616,
      "learning_rate": 1.241441767510951e-06,
      "loss": 0.001,
      "step": 145520
    },
    {
      "epoch": 2.3812484660067086,
      "grad_norm": 0.08004292845726013,
      "learning_rate": 1.2408140388369405e-06,
      "loss": 0.0027,
      "step": 145530
    },
    {
      "epoch": 2.3814120919577846,
      "grad_norm": 0.04221239686012268,
      "learning_rate": 1.2401864464243874e-06,
      "loss": 0.0007,
      "step": 145540
    },
    {
      "epoch": 2.38157571790886,
      "grad_norm": 0.14445243775844574,
      "learning_rate": 1.2395589902960377e-06,
      "loss": 0.0012,
      "step": 145550
    },
    {
      "epoch": 2.381739343859936,
      "grad_norm": 0.040830545127391815,
      "learning_rate": 1.2389316704746362e-06,
      "loss": 0.0006,
      "step": 145560
    },
    {
      "epoch": 2.381902969811012,
      "grad_norm": 0.24122272431850433,
      "learning_rate": 1.2383044869829242e-06,
      "loss": 0.0006,
      "step": 145570
    },
    {
      "epoch": 2.3820665957620877,
      "grad_norm": 0.10114267468452454,
      "learning_rate": 1.237677439843632e-06,
      "loss": 0.0007,
      "step": 145580
    },
    {
      "epoch": 2.3822302217131637,
      "grad_norm": 0.02469678781926632,
      "learning_rate": 1.237050529079491e-06,
      "loss": 0.0008,
      "step": 145590
    },
    {
      "epoch": 2.3823938476642397,
      "grad_norm": 0.070464126765728,
      "learning_rate": 1.2364237547132235e-06,
      "loss": 0.0006,
      "step": 145600
    },
    {
      "epoch": 2.3825574736153152,
      "grad_norm": 0.06091037020087242,
      "learning_rate": 1.235797116767552e-06,
      "loss": 0.0015,
      "step": 145610
    },
    {
      "epoch": 2.382721099566391,
      "grad_norm": 0.004935974720865488,
      "learning_rate": 1.2351706152651856e-06,
      "loss": 0.0005,
      "step": 145620
    },
    {
      "epoch": 2.382884725517467,
      "grad_norm": 0.033449091017246246,
      "learning_rate": 1.2345442502288353e-06,
      "loss": 0.0022,
      "step": 145630
    },
    {
      "epoch": 2.3830483514685428,
      "grad_norm": 0.003820921527221799,
      "learning_rate": 1.2339180216812074e-06,
      "loss": 0.0004,
      "step": 145640
    },
    {
      "epoch": 2.3832119774196188,
      "grad_norm": 0.027925308793783188,
      "learning_rate": 1.233291929644998e-06,
      "loss": 0.0011,
      "step": 145650
    },
    {
      "epoch": 2.3833756033706948,
      "grad_norm": 0.13035884499549866,
      "learning_rate": 1.2326659741429054e-06,
      "loss": 0.0009,
      "step": 145660
    },
    {
      "epoch": 2.3835392293217703,
      "grad_norm": 0.007219295017421246,
      "learning_rate": 1.2320401551976146e-06,
      "loss": 0.0008,
      "step": 145670
    },
    {
      "epoch": 2.3837028552728463,
      "grad_norm": 0.07360933721065521,
      "learning_rate": 1.2314144728318134e-06,
      "loss": 0.0005,
      "step": 145680
    },
    {
      "epoch": 2.3838664812239223,
      "grad_norm": 0.007033075205981731,
      "learning_rate": 1.2307889270681788e-06,
      "loss": 0.0013,
      "step": 145690
    },
    {
      "epoch": 2.384030107174998,
      "grad_norm": 0.07547487318515778,
      "learning_rate": 1.230163517929388e-06,
      "loss": 0.0005,
      "step": 145700
    },
    {
      "epoch": 2.384193733126074,
      "grad_norm": 0.036217715591192245,
      "learning_rate": 1.2295382454381076e-06,
      "loss": 0.0008,
      "step": 145710
    },
    {
      "epoch": 2.38435735907715,
      "grad_norm": 0.02261698804795742,
      "learning_rate": 1.2289131096170053e-06,
      "loss": 0.0005,
      "step": 145720
    },
    {
      "epoch": 2.3845209850282254,
      "grad_norm": 0.08924450725317001,
      "learning_rate": 1.2282881104887395e-06,
      "loss": 0.0009,
      "step": 145730
    },
    {
      "epoch": 2.3846846109793014,
      "grad_norm": 0.01222652941942215,
      "learning_rate": 1.2276632480759625e-06,
      "loss": 0.0015,
      "step": 145740
    },
    {
      "epoch": 2.3848482369303774,
      "grad_norm": 0.018236972391605377,
      "learning_rate": 1.227038522401328e-06,
      "loss": 0.0012,
      "step": 145750
    },
    {
      "epoch": 2.385011862881453,
      "grad_norm": 0.017342614009976387,
      "learning_rate": 1.2264139334874775e-06,
      "loss": 0.0017,
      "step": 145760
    },
    {
      "epoch": 2.385175488832529,
      "grad_norm": 0.07415437698364258,
      "learning_rate": 1.2257894813570537e-06,
      "loss": 0.0008,
      "step": 145770
    },
    {
      "epoch": 2.3853391147836045,
      "grad_norm": 0.06055690720677376,
      "learning_rate": 1.2251651660326891e-06,
      "loss": 0.0007,
      "step": 145780
    },
    {
      "epoch": 2.3855027407346805,
      "grad_norm": 0.002141202799975872,
      "learning_rate": 1.224540987537014e-06,
      "loss": 0.0005,
      "step": 145790
    },
    {
      "epoch": 2.3856663666857565,
      "grad_norm": 0.15238164365291595,
      "learning_rate": 1.2239169458926558e-06,
      "loss": 0.0009,
      "step": 145800
    },
    {
      "epoch": 2.385829992636832,
      "grad_norm": 0.07647334784269333,
      "learning_rate": 1.2232930411222316e-06,
      "loss": 0.0007,
      "step": 145810
    },
    {
      "epoch": 2.385993618587908,
      "grad_norm": 0.04357259348034859,
      "learning_rate": 1.2226692732483602e-06,
      "loss": 0.0005,
      "step": 145820
    },
    {
      "epoch": 2.386157244538984,
      "grad_norm": 0.04395972564816475,
      "learning_rate": 1.222045642293646e-06,
      "loss": 0.0007,
      "step": 145830
    },
    {
      "epoch": 2.3863208704900596,
      "grad_norm": 0.001686961273662746,
      "learning_rate": 1.2214221482806976e-06,
      "loss": 0.0009,
      "step": 145840
    },
    {
      "epoch": 2.3864844964411356,
      "grad_norm": 0.01953599415719509,
      "learning_rate": 1.2207987912321156e-06,
      "loss": 0.0006,
      "step": 145850
    },
    {
      "epoch": 2.3866481223922116,
      "grad_norm": 0.02389693073928356,
      "learning_rate": 1.2201755711704933e-06,
      "loss": 0.0016,
      "step": 145860
    },
    {
      "epoch": 2.386811748343287,
      "grad_norm": 0.04619196057319641,
      "learning_rate": 1.2195524881184234e-06,
      "loss": 0.0009,
      "step": 145870
    },
    {
      "epoch": 2.386975374294363,
      "grad_norm": 0.11164496093988419,
      "learning_rate": 1.2189295420984882e-06,
      "loss": 0.0006,
      "step": 145880
    },
    {
      "epoch": 2.3871390002454387,
      "grad_norm": 0.17305664718151093,
      "learning_rate": 1.218306733133271e-06,
      "loss": 0.0012,
      "step": 145890
    },
    {
      "epoch": 2.3873026261965147,
      "grad_norm": 0.08470435440540314,
      "learning_rate": 1.2176840612453445e-06,
      "loss": 0.0013,
      "step": 145900
    },
    {
      "epoch": 2.3874662521475907,
      "grad_norm": 0.07801004499197006,
      "learning_rate": 1.217061526457281e-06,
      "loss": 0.0006,
      "step": 145910
    },
    {
      "epoch": 2.3876298780986662,
      "grad_norm": 0.16480372846126556,
      "learning_rate": 1.2164391287916454e-06,
      "loss": 0.0009,
      "step": 145920
    },
    {
      "epoch": 2.387793504049742,
      "grad_norm": 0.08581792563199997,
      "learning_rate": 1.215816868270997e-06,
      "loss": 0.0006,
      "step": 145930
    },
    {
      "epoch": 2.387957130000818,
      "grad_norm": 0.05874474346637726,
      "learning_rate": 1.215194744917893e-06,
      "loss": 0.0008,
      "step": 145940
    },
    {
      "epoch": 2.3881207559518938,
      "grad_norm": 0.0018676667241379619,
      "learning_rate": 1.2145727587548817e-06,
      "loss": 0.0008,
      "step": 145950
    },
    {
      "epoch": 2.3882843819029698,
      "grad_norm": 0.13485835492610931,
      "learning_rate": 1.213950909804511e-06,
      "loss": 0.0013,
      "step": 145960
    },
    {
      "epoch": 2.3884480078540458,
      "grad_norm": 0.05608774721622467,
      "learning_rate": 1.2133291980893191e-06,
      "loss": 0.0007,
      "step": 145970
    },
    {
      "epoch": 2.3886116338051213,
      "grad_norm": 0.11544715613126755,
      "learning_rate": 1.212707623631844e-06,
      "loss": 0.001,
      "step": 145980
    },
    {
      "epoch": 2.3887752597561973,
      "grad_norm": 0.0441889762878418,
      "learning_rate": 1.2120861864546135e-06,
      "loss": 0.0007,
      "step": 145990
    },
    {
      "epoch": 2.3889388857072733,
      "grad_norm": 0.045418061316013336,
      "learning_rate": 1.2114648865801547e-06,
      "loss": 0.0006,
      "step": 146000
    },
    {
      "epoch": 2.389102511658349,
      "grad_norm": 0.04495017230510712,
      "learning_rate": 1.21084372403099e-06,
      "loss": 0.0008,
      "step": 146010
    },
    {
      "epoch": 2.389266137609425,
      "grad_norm": 0.00138551602140069,
      "learning_rate": 1.2102226988296334e-06,
      "loss": 0.0009,
      "step": 146020
    },
    {
      "epoch": 2.389429763560501,
      "grad_norm": 0.07855869829654694,
      "learning_rate": 1.2096018109985951e-06,
      "loss": 0.0006,
      "step": 146030
    },
    {
      "epoch": 2.3895933895115764,
      "grad_norm": 0.08308794349431992,
      "learning_rate": 1.2089810605603802e-06,
      "loss": 0.0004,
      "step": 146040
    },
    {
      "epoch": 2.3897570154626524,
      "grad_norm": 0.09332466125488281,
      "learning_rate": 1.2083604475374917e-06,
      "loss": 0.0005,
      "step": 146050
    },
    {
      "epoch": 2.3899206414137284,
      "grad_norm": 0.12089268118143082,
      "learning_rate": 1.2077399719524218e-06,
      "loss": 0.001,
      "step": 146060
    },
    {
      "epoch": 2.390084267364804,
      "grad_norm": 0.13675138354301453,
      "learning_rate": 1.207119633827664e-06,
      "loss": 0.0005,
      "step": 146070
    },
    {
      "epoch": 2.39024789331588,
      "grad_norm": 0.10944545269012451,
      "learning_rate": 1.2064994331857056e-06,
      "loss": 0.0006,
      "step": 146080
    },
    {
      "epoch": 2.390411519266956,
      "grad_norm": 0.012564612552523613,
      "learning_rate": 1.205879370049023e-06,
      "loss": 0.0009,
      "step": 146090
    },
    {
      "epoch": 2.3905751452180315,
      "grad_norm": 0.03477565944194794,
      "learning_rate": 1.205259444440096e-06,
      "loss": 0.0006,
      "step": 146100
    },
    {
      "epoch": 2.3907387711691075,
      "grad_norm": 0.11118346452713013,
      "learning_rate": 1.2046396563813922e-06,
      "loss": 0.0007,
      "step": 146110
    },
    {
      "epoch": 2.3909023971201835,
      "grad_norm": 0.01566425710916519,
      "learning_rate": 1.204020005895381e-06,
      "loss": 0.0004,
      "step": 146120
    },
    {
      "epoch": 2.391066023071259,
      "grad_norm": 0.06493952125310898,
      "learning_rate": 1.2034004930045206e-06,
      "loss": 0.0009,
      "step": 146130
    },
    {
      "epoch": 2.391229649022335,
      "grad_norm": 0.12279953807592392,
      "learning_rate": 1.202781117731266e-06,
      "loss": 0.0024,
      "step": 146140
    },
    {
      "epoch": 2.391393274973411,
      "grad_norm": 0.010475469753146172,
      "learning_rate": 1.202161880098071e-06,
      "loss": 0.0009,
      "step": 146150
    },
    {
      "epoch": 2.3915569009244866,
      "grad_norm": 0.02797282487154007,
      "learning_rate": 1.2015427801273787e-06,
      "loss": 0.0007,
      "step": 146160
    },
    {
      "epoch": 2.3917205268755626,
      "grad_norm": 0.09259378910064697,
      "learning_rate": 1.2009238178416326e-06,
      "loss": 0.0009,
      "step": 146170
    },
    {
      "epoch": 2.391884152826638,
      "grad_norm": 0.04144535958766937,
      "learning_rate": 1.200304993263266e-06,
      "loss": 0.0006,
      "step": 146180
    },
    {
      "epoch": 2.392047778777714,
      "grad_norm": 0.028234334662556648,
      "learning_rate": 1.1996863064147118e-06,
      "loss": 0.0008,
      "step": 146190
    },
    {
      "epoch": 2.39221140472879,
      "grad_norm": 0.16661839187145233,
      "learning_rate": 1.1990677573183945e-06,
      "loss": 0.0021,
      "step": 146200
    },
    {
      "epoch": 2.3923750306798657,
      "grad_norm": 0.06529664993286133,
      "learning_rate": 1.1984493459967367e-06,
      "loss": 0.0011,
      "step": 146210
    },
    {
      "epoch": 2.3925386566309417,
      "grad_norm": 0.026793505996465683,
      "learning_rate": 1.1978310724721538e-06,
      "loss": 0.001,
      "step": 146220
    },
    {
      "epoch": 2.3927022825820177,
      "grad_norm": 0.042733270674943924,
      "learning_rate": 1.1972129367670543e-06,
      "loss": 0.0006,
      "step": 146230
    },
    {
      "epoch": 2.3928659085330932,
      "grad_norm": 0.02896728552877903,
      "learning_rate": 1.1965949389038479e-06,
      "loss": 0.0008,
      "step": 146240
    },
    {
      "epoch": 2.393029534484169,
      "grad_norm": 0.047306910157203674,
      "learning_rate": 1.1959770789049324e-06,
      "loss": 0.0008,
      "step": 146250
    },
    {
      "epoch": 2.393193160435245,
      "grad_norm": 0.030137227848172188,
      "learning_rate": 1.1953593567927063e-06,
      "loss": 0.0005,
      "step": 146260
    },
    {
      "epoch": 2.3933567863863208,
      "grad_norm": 0.12347330152988434,
      "learning_rate": 1.1947417725895577e-06,
      "loss": 0.0007,
      "step": 146270
    },
    {
      "epoch": 2.3935204123373968,
      "grad_norm": 0.020726656541228294,
      "learning_rate": 1.1941243263178742e-06,
      "loss": 0.001,
      "step": 146280
    },
    {
      "epoch": 2.3936840382884723,
      "grad_norm": 0.007426445838063955,
      "learning_rate": 1.193507018000038e-06,
      "loss": 0.0005,
      "step": 146290
    },
    {
      "epoch": 2.3938476642395483,
      "grad_norm": 0.021827425807714462,
      "learning_rate": 1.1928898476584228e-06,
      "loss": 0.0007,
      "step": 146300
    },
    {
      "epoch": 2.3940112901906243,
      "grad_norm": 0.05956770479679108,
      "learning_rate": 1.1922728153154012e-06,
      "loss": 0.0014,
      "step": 146310
    },
    {
      "epoch": 2.3941749161417,
      "grad_norm": 0.08083844929933548,
      "learning_rate": 1.1916559209933392e-06,
      "loss": 0.0011,
      "step": 146320
    },
    {
      "epoch": 2.394338542092776,
      "grad_norm": 0.04298021271824837,
      "learning_rate": 1.1910391647145963e-06,
      "loss": 0.0007,
      "step": 146330
    },
    {
      "epoch": 2.394502168043852,
      "grad_norm": 0.0014561654534190893,
      "learning_rate": 1.190422546501528e-06,
      "loss": 0.0006,
      "step": 146340
    },
    {
      "epoch": 2.3946657939949274,
      "grad_norm": 0.06279488652944565,
      "learning_rate": 1.189806066376486e-06,
      "loss": 0.0004,
      "step": 146350
    },
    {
      "epoch": 2.3948294199460034,
      "grad_norm": 0.04115691035985947,
      "learning_rate": 1.1891897243618184e-06,
      "loss": 0.0007,
      "step": 146360
    },
    {
      "epoch": 2.3949930458970794,
      "grad_norm": 0.12842579185962677,
      "learning_rate": 1.1885735204798626e-06,
      "loss": 0.0009,
      "step": 146370
    },
    {
      "epoch": 2.395156671848155,
      "grad_norm": 0.008940541185438633,
      "learning_rate": 1.1879574547529576e-06,
      "loss": 0.001,
      "step": 146380
    },
    {
      "epoch": 2.395320297799231,
      "grad_norm": 0.4329315721988678,
      "learning_rate": 1.187341527203431e-06,
      "loss": 0.0006,
      "step": 146390
    },
    {
      "epoch": 2.395483923750307,
      "grad_norm": 0.11849091202020645,
      "learning_rate": 1.1867257378536128e-06,
      "loss": 0.0007,
      "step": 146400
    },
    {
      "epoch": 2.3956475497013825,
      "grad_norm": 0.10733088850975037,
      "learning_rate": 1.1861100867258197e-06,
      "loss": 0.0011,
      "step": 146410
    },
    {
      "epoch": 2.3958111756524585,
      "grad_norm": 0.12135057896375656,
      "learning_rate": 1.1854945738423711e-06,
      "loss": 0.0009,
      "step": 146420
    },
    {
      "epoch": 2.3959748016035345,
      "grad_norm": 0.06494513899087906,
      "learning_rate": 1.1848791992255764e-06,
      "loss": 0.001,
      "step": 146430
    },
    {
      "epoch": 2.39613842755461,
      "grad_norm": 0.044759757816791534,
      "learning_rate": 1.18426396289774e-06,
      "loss": 0.0009,
      "step": 146440
    },
    {
      "epoch": 2.396302053505686,
      "grad_norm": 0.08611297607421875,
      "learning_rate": 1.183648864881165e-06,
      "loss": 0.0007,
      "step": 146450
    },
    {
      "epoch": 2.396465679456762,
      "grad_norm": 0.16296802461147308,
      "learning_rate": 1.183033905198145e-06,
      "loss": 0.0008,
      "step": 146460
    },
    {
      "epoch": 2.3966293054078376,
      "grad_norm": 0.03838413208723068,
      "learning_rate": 1.1824190838709737e-06,
      "loss": 0.0008,
      "step": 146470
    },
    {
      "epoch": 2.3967929313589136,
      "grad_norm": 0.03129525110125542,
      "learning_rate": 1.1818044009219337e-06,
      "loss": 0.0009,
      "step": 146480
    },
    {
      "epoch": 2.3969565573099896,
      "grad_norm": 0.03448084369301796,
      "learning_rate": 1.1811898563733087e-06,
      "loss": 0.0003,
      "step": 146490
    },
    {
      "epoch": 2.397120183261065,
      "grad_norm": 0.06899776309728622,
      "learning_rate": 1.180575450247372e-06,
      "loss": 0.0007,
      "step": 146500
    },
    {
      "epoch": 2.397283809212141,
      "grad_norm": 0.02704421617090702,
      "learning_rate": 1.1799611825663954e-06,
      "loss": 0.0004,
      "step": 146510
    },
    {
      "epoch": 2.397447435163217,
      "grad_norm": 0.06607653200626373,
      "learning_rate": 1.1793470533526473e-06,
      "loss": 0.0011,
      "step": 146520
    },
    {
      "epoch": 2.3976110611142927,
      "grad_norm": 0.028309181332588196,
      "learning_rate": 1.1787330626283834e-06,
      "loss": 0.001,
      "step": 146530
    },
    {
      "epoch": 2.3977746870653687,
      "grad_norm": 0.27555280923843384,
      "learning_rate": 1.178119210415863e-06,
      "loss": 0.0006,
      "step": 146540
    },
    {
      "epoch": 2.3979383130164442,
      "grad_norm": 0.20501168072223663,
      "learning_rate": 1.177505496737334e-06,
      "loss": 0.0014,
      "step": 146550
    },
    {
      "epoch": 2.39810193896752,
      "grad_norm": 0.06874565780162811,
      "learning_rate": 1.1768919216150437e-06,
      "loss": 0.0009,
      "step": 146560
    },
    {
      "epoch": 2.398265564918596,
      "grad_norm": 0.042318861931562424,
      "learning_rate": 1.1762784850712345e-06,
      "loss": 0.0006,
      "step": 146570
    },
    {
      "epoch": 2.3984291908696718,
      "grad_norm": 0.14016929268836975,
      "learning_rate": 1.1756651871281384e-06,
      "loss": 0.0014,
      "step": 146580
    },
    {
      "epoch": 2.3985928168207478,
      "grad_norm": 0.08457785099744797,
      "learning_rate": 1.175052027807989e-06,
      "loss": 0.0003,
      "step": 146590
    },
    {
      "epoch": 2.3987564427718238,
      "grad_norm": 0.16232720017433167,
      "learning_rate": 1.1744390071330092e-06,
      "loss": 0.0015,
      "step": 146600
    },
    {
      "epoch": 2.3989200687228993,
      "grad_norm": 0.026947375386953354,
      "learning_rate": 1.1738261251254224e-06,
      "loss": 0.0016,
      "step": 146610
    },
    {
      "epoch": 2.3990836946739753,
      "grad_norm": 0.15962563455104828,
      "learning_rate": 1.1732133818074426e-06,
      "loss": 0.0015,
      "step": 146620
    },
    {
      "epoch": 2.3992473206250513,
      "grad_norm": 0.02198173850774765,
      "learning_rate": 1.1726007772012793e-06,
      "loss": 0.0009,
      "step": 146630
    },
    {
      "epoch": 2.399410946576127,
      "grad_norm": 0.04337415099143982,
      "learning_rate": 1.1719883113291397e-06,
      "loss": 0.0004,
      "step": 146640
    },
    {
      "epoch": 2.399574572527203,
      "grad_norm": 0.03358631581068039,
      "learning_rate": 1.1713759842132222e-06,
      "loss": 0.0006,
      "step": 146650
    },
    {
      "epoch": 2.3997381984782784,
      "grad_norm": 0.017577163875102997,
      "learning_rate": 1.1707637958757244e-06,
      "loss": 0.0007,
      "step": 146660
    },
    {
      "epoch": 2.3999018244293544,
      "grad_norm": 0.007853773422539234,
      "learning_rate": 1.1701517463388351e-06,
      "loss": 0.0003,
      "step": 146670
    },
    {
      "epoch": 2.4000654503804304,
      "grad_norm": 0.10636381059885025,
      "learning_rate": 1.1695398356247412e-06,
      "loss": 0.0008,
      "step": 146680
    },
    {
      "epoch": 2.400229076331506,
      "grad_norm": 0.0726991817355156,
      "learning_rate": 1.1689280637556205e-06,
      "loss": 0.001,
      "step": 146690
    },
    {
      "epoch": 2.400392702282582,
      "grad_norm": 0.21138660609722137,
      "learning_rate": 1.168316430753651e-06,
      "loss": 0.0007,
      "step": 146700
    },
    {
      "epoch": 2.400556328233658,
      "grad_norm": 0.18228918313980103,
      "learning_rate": 1.1677049366410008e-06,
      "loss": 0.0009,
      "step": 146710
    },
    {
      "epoch": 2.4007199541847335,
      "grad_norm": 0.0521659292280674,
      "learning_rate": 1.167093581439837e-06,
      "loss": 0.0007,
      "step": 146720
    },
    {
      "epoch": 2.4008835801358095,
      "grad_norm": 0.000778741086833179,
      "learning_rate": 1.166482365172319e-06,
      "loss": 0.0012,
      "step": 146730
    },
    {
      "epoch": 2.4010472060868855,
      "grad_norm": 0.11970508843660355,
      "learning_rate": 1.1658712878606e-06,
      "loss": 0.0006,
      "step": 146740
    },
    {
      "epoch": 2.401210832037961,
      "grad_norm": 0.006852451246231794,
      "learning_rate": 1.1652603495268333e-06,
      "loss": 0.0009,
      "step": 146750
    },
    {
      "epoch": 2.401374457989037,
      "grad_norm": 0.0973595604300499,
      "learning_rate": 1.1646495501931604e-06,
      "loss": 0.0006,
      "step": 146760
    },
    {
      "epoch": 2.401538083940113,
      "grad_norm": 0.0892665907740593,
      "learning_rate": 1.164038889881724e-06,
      "loss": 0.0012,
      "step": 146770
    },
    {
      "epoch": 2.4017017098911886,
      "grad_norm": 0.09378153085708618,
      "learning_rate": 1.1634283686146591e-06,
      "loss": 0.0007,
      "step": 146780
    },
    {
      "epoch": 2.4018653358422646,
      "grad_norm": 0.048388123512268066,
      "learning_rate": 1.1628179864140942e-06,
      "loss": 0.0006,
      "step": 146790
    },
    {
      "epoch": 2.4020289617933406,
      "grad_norm": 0.07055846601724625,
      "learning_rate": 1.1622077433021562e-06,
      "loss": 0.0006,
      "step": 146800
    },
    {
      "epoch": 2.402192587744416,
      "grad_norm": 0.07691455632448196,
      "learning_rate": 1.1615976393009614e-06,
      "loss": 0.0013,
      "step": 146810
    },
    {
      "epoch": 2.402356213695492,
      "grad_norm": 0.06037082150578499,
      "learning_rate": 1.1609876744326303e-06,
      "loss": 0.0006,
      "step": 146820
    },
    {
      "epoch": 2.402519839646568,
      "grad_norm": 0.0030878460966050625,
      "learning_rate": 1.1603778487192663e-06,
      "loss": 0.0008,
      "step": 146830
    },
    {
      "epoch": 2.4026834655976437,
      "grad_norm": 0.0932287722826004,
      "learning_rate": 1.1597681621829771e-06,
      "loss": 0.0009,
      "step": 146840
    },
    {
      "epoch": 2.4028470915487197,
      "grad_norm": 0.14290370047092438,
      "learning_rate": 1.1591586148458633e-06,
      "loss": 0.0006,
      "step": 146850
    },
    {
      "epoch": 2.4030107174997957,
      "grad_norm": 0.04808291792869568,
      "learning_rate": 1.1585492067300174e-06,
      "loss": 0.0006,
      "step": 146860
    },
    {
      "epoch": 2.403174343450871,
      "grad_norm": 0.07518628984689713,
      "learning_rate": 1.1579399378575312e-06,
      "loss": 0.0004,
      "step": 146870
    },
    {
      "epoch": 2.403337969401947,
      "grad_norm": 0.07845789194107056,
      "learning_rate": 1.157330808250487e-06,
      "loss": 0.0009,
      "step": 146880
    },
    {
      "epoch": 2.403501595353023,
      "grad_norm": 0.07071030884981155,
      "learning_rate": 1.1567218179309668e-06,
      "loss": 0.0009,
      "step": 146890
    },
    {
      "epoch": 2.4036652213040988,
      "grad_norm": 0.026469646021723747,
      "learning_rate": 1.1561129669210419e-06,
      "loss": 0.0009,
      "step": 146900
    },
    {
      "epoch": 2.4038288472551748,
      "grad_norm": 0.12547431886196136,
      "learning_rate": 1.1555042552427853e-06,
      "loss": 0.0011,
      "step": 146910
    },
    {
      "epoch": 2.4039924732062508,
      "grad_norm": 0.010703513398766518,
      "learning_rate": 1.1548956829182588e-06,
      "loss": 0.0007,
      "step": 146920
    },
    {
      "epoch": 2.4041560991573263,
      "grad_norm": 0.13315097987651825,
      "learning_rate": 1.1542872499695212e-06,
      "loss": 0.001,
      "step": 146930
    },
    {
      "epoch": 2.4043197251084023,
      "grad_norm": 0.04090576246380806,
      "learning_rate": 1.1536789564186297e-06,
      "loss": 0.001,
      "step": 146940
    },
    {
      "epoch": 2.404483351059478,
      "grad_norm": 0.10873720049858093,
      "learning_rate": 1.1530708022876297e-06,
      "loss": 0.0008,
      "step": 146950
    },
    {
      "epoch": 2.404646977010554,
      "grad_norm": 0.038747694343328476,
      "learning_rate": 1.152462787598569e-06,
      "loss": 0.0005,
      "step": 146960
    },
    {
      "epoch": 2.40481060296163,
      "grad_norm": 0.056009504944086075,
      "learning_rate": 1.1518549123734834e-06,
      "loss": 0.0026,
      "step": 146970
    },
    {
      "epoch": 2.4049742289127054,
      "grad_norm": 0.0809401348233223,
      "learning_rate": 1.1512471766344096e-06,
      "loss": 0.001,
      "step": 146980
    },
    {
      "epoch": 2.4051378548637814,
      "grad_norm": 0.05070461705327034,
      "learning_rate": 1.1506395804033743e-06,
      "loss": 0.0003,
      "step": 146990
    },
    {
      "epoch": 2.4053014808148574,
      "grad_norm": 0.016141260042786598,
      "learning_rate": 1.1500321237024027e-06,
      "loss": 0.0006,
      "step": 147000
    },
    {
      "epoch": 2.405465106765933,
      "grad_norm": 0.06742151826620102,
      "learning_rate": 1.1494248065535146e-06,
      "loss": 0.0009,
      "step": 147010
    },
    {
      "epoch": 2.405628732717009,
      "grad_norm": 0.05144858360290527,
      "learning_rate": 1.1488176289787223e-06,
      "loss": 0.0008,
      "step": 147020
    },
    {
      "epoch": 2.405792358668085,
      "grad_norm": 0.27428361773490906,
      "learning_rate": 1.1482105910000352e-06,
      "loss": 0.001,
      "step": 147030
    },
    {
      "epoch": 2.4059559846191605,
      "grad_norm": 0.0735877975821495,
      "learning_rate": 1.1476036926394557e-06,
      "loss": 0.0009,
      "step": 147040
    },
    {
      "epoch": 2.4061196105702365,
      "grad_norm": 0.027983354404568672,
      "learning_rate": 1.1469969339189829e-06,
      "loss": 0.0011,
      "step": 147050
    },
    {
      "epoch": 2.406283236521312,
      "grad_norm": 0.08877575397491455,
      "learning_rate": 1.1463903148606121e-06,
      "loss": 0.001,
      "step": 147060
    },
    {
      "epoch": 2.406446862472388,
      "grad_norm": 0.07310646772384644,
      "learning_rate": 1.1457838354863293e-06,
      "loss": 0.0013,
      "step": 147070
    },
    {
      "epoch": 2.406610488423464,
      "grad_norm": 0.042939409613609314,
      "learning_rate": 1.1451774958181206e-06,
      "loss": 0.0012,
      "step": 147080
    },
    {
      "epoch": 2.4067741143745396,
      "grad_norm": 0.06409641355276108,
      "learning_rate": 1.1445712958779608e-06,
      "loss": 0.0012,
      "step": 147090
    },
    {
      "epoch": 2.4069377403256156,
      "grad_norm": 0.07361728698015213,
      "learning_rate": 1.1439652356878271e-06,
      "loss": 0.0008,
      "step": 147100
    },
    {
      "epoch": 2.4071013662766916,
      "grad_norm": 0.14623193442821503,
      "learning_rate": 1.1433593152696847e-06,
      "loss": 0.0006,
      "step": 147110
    },
    {
      "epoch": 2.407264992227767,
      "grad_norm": 0.15490563213825226,
      "learning_rate": 1.1427535346454993e-06,
      "loss": 0.0006,
      "step": 147120
    },
    {
      "epoch": 2.407428618178843,
      "grad_norm": 0.012476619333028793,
      "learning_rate": 1.142147893837227e-06,
      "loss": 0.0006,
      "step": 147130
    },
    {
      "epoch": 2.407592244129919,
      "grad_norm": 0.03630430996417999,
      "learning_rate": 1.1415423928668207e-06,
      "loss": 0.0004,
      "step": 147140
    },
    {
      "epoch": 2.4077558700809947,
      "grad_norm": 0.0012697308557108045,
      "learning_rate": 1.14093703175623e-06,
      "loss": 0.0009,
      "step": 147150
    },
    {
      "epoch": 2.4079194960320707,
      "grad_norm": 0.013975553214550018,
      "learning_rate": 1.1403318105273958e-06,
      "loss": 0.0008,
      "step": 147160
    },
    {
      "epoch": 2.4080831219831467,
      "grad_norm": 0.08526013791561127,
      "learning_rate": 1.1397267292022585e-06,
      "loss": 0.0007,
      "step": 147170
    },
    {
      "epoch": 2.408246747934222,
      "grad_norm": 0.008147203363478184,
      "learning_rate": 1.1391217878027477e-06,
      "loss": 0.0009,
      "step": 147180
    },
    {
      "epoch": 2.408410373885298,
      "grad_norm": 0.024288080632686615,
      "learning_rate": 1.1385169863507944e-06,
      "loss": 0.0019,
      "step": 147190
    },
    {
      "epoch": 2.408573999836374,
      "grad_norm": 0.14177322387695312,
      "learning_rate": 1.1379123248683178e-06,
      "loss": 0.0009,
      "step": 147200
    },
    {
      "epoch": 2.4087376257874498,
      "grad_norm": 0.06706675887107849,
      "learning_rate": 1.1373078033772373e-06,
      "loss": 0.0008,
      "step": 147210
    },
    {
      "epoch": 2.4089012517385258,
      "grad_norm": 0.029751280322670937,
      "learning_rate": 1.1367034218994682e-06,
      "loss": 0.0007,
      "step": 147220
    },
    {
      "epoch": 2.4090648776896018,
      "grad_norm": 0.06997836381196976,
      "learning_rate": 1.1360991804569116e-06,
      "loss": 0.0003,
      "step": 147230
    },
    {
      "epoch": 2.4092285036406773,
      "grad_norm": 0.02214796468615532,
      "learning_rate": 1.1354950790714748e-06,
      "loss": 0.0013,
      "step": 147240
    },
    {
      "epoch": 2.4093921295917533,
      "grad_norm": 0.047099754214286804,
      "learning_rate": 1.1348911177650518e-06,
      "loss": 0.0006,
      "step": 147250
    },
    {
      "epoch": 2.4095557555428293,
      "grad_norm": 0.04467353597283363,
      "learning_rate": 1.1342872965595374e-06,
      "loss": 0.0004,
      "step": 147260
    },
    {
      "epoch": 2.409719381493905,
      "grad_norm": 0.1443975865840912,
      "learning_rate": 1.1336836154768167e-06,
      "loss": 0.0017,
      "step": 147270
    },
    {
      "epoch": 2.409883007444981,
      "grad_norm": 0.013578618876636028,
      "learning_rate": 1.1330800745387722e-06,
      "loss": 0.0006,
      "step": 147280
    },
    {
      "epoch": 2.410046633396057,
      "grad_norm": 0.025736160576343536,
      "learning_rate": 1.1324766737672822e-06,
      "loss": 0.0008,
      "step": 147290
    },
    {
      "epoch": 2.4102102593471324,
      "grad_norm": 0.009556686505675316,
      "learning_rate": 1.1318734131842162e-06,
      "loss": 0.0008,
      "step": 147300
    },
    {
      "epoch": 2.4103738852982084,
      "grad_norm": 0.0320097953081131,
      "learning_rate": 1.131270292811444e-06,
      "loss": 0.0014,
      "step": 147310
    },
    {
      "epoch": 2.410537511249284,
      "grad_norm": 0.11312876641750336,
      "learning_rate": 1.1306673126708245e-06,
      "loss": 0.0008,
      "step": 147320
    },
    {
      "epoch": 2.41070113720036,
      "grad_norm": 0.0021174512803554535,
      "learning_rate": 1.1300644727842147e-06,
      "loss": 0.0008,
      "step": 147330
    },
    {
      "epoch": 2.410864763151436,
      "grad_norm": 0.05402659252285957,
      "learning_rate": 1.1294617731734676e-06,
      "loss": 0.0008,
      "step": 147340
    },
    {
      "epoch": 2.4110283891025115,
      "grad_norm": 0.0137928556650877,
      "learning_rate": 1.1288592138604276e-06,
      "loss": 0.0009,
      "step": 147350
    },
    {
      "epoch": 2.4111920150535875,
      "grad_norm": 0.06480717658996582,
      "learning_rate": 1.1282567948669383e-06,
      "loss": 0.0005,
      "step": 147360
    },
    {
      "epoch": 2.4113556410046635,
      "grad_norm": 0.006785198114812374,
      "learning_rate": 1.1276545162148327e-06,
      "loss": 0.0009,
      "step": 147370
    },
    {
      "epoch": 2.411519266955739,
      "grad_norm": 0.08239125460386276,
      "learning_rate": 1.1270523779259463e-06,
      "loss": 0.0008,
      "step": 147380
    },
    {
      "epoch": 2.411682892906815,
      "grad_norm": 0.06658472865819931,
      "learning_rate": 1.1264503800221005e-06,
      "loss": 0.0005,
      "step": 147390
    },
    {
      "epoch": 2.411846518857891,
      "grad_norm": 0.062044478952884674,
      "learning_rate": 1.1258485225251203e-06,
      "loss": 0.0006,
      "step": 147400
    },
    {
      "epoch": 2.4120101448089666,
      "grad_norm": 0.11041797697544098,
      "learning_rate": 1.125246805456819e-06,
      "loss": 0.0008,
      "step": 147410
    },
    {
      "epoch": 2.4121737707600426,
      "grad_norm": 0.06895242631435394,
      "learning_rate": 1.1246452288390092e-06,
      "loss": 0.0017,
      "step": 147420
    },
    {
      "epoch": 2.4123373967111186,
      "grad_norm": 0.0976947769522667,
      "learning_rate": 1.1240437926934961e-06,
      "loss": 0.0008,
      "step": 147430
    },
    {
      "epoch": 2.412501022662194,
      "grad_norm": 0.011171468533575535,
      "learning_rate": 1.1234424970420776e-06,
      "loss": 0.0006,
      "step": 147440
    },
    {
      "epoch": 2.41266464861327,
      "grad_norm": 0.01117353979498148,
      "learning_rate": 1.1228413419065537e-06,
      "loss": 0.0005,
      "step": 147450
    },
    {
      "epoch": 2.4128282745643457,
      "grad_norm": 0.1373976618051529,
      "learning_rate": 1.1222403273087112e-06,
      "loss": 0.0009,
      "step": 147460
    },
    {
      "epoch": 2.4129919005154217,
      "grad_norm": 0.11558207124471664,
      "learning_rate": 1.1216394532703378e-06,
      "loss": 0.0004,
      "step": 147470
    },
    {
      "epoch": 2.4131555264664977,
      "grad_norm": 0.045516468584537506,
      "learning_rate": 1.1210387198132123e-06,
      "loss": 0.0009,
      "step": 147480
    },
    {
      "epoch": 2.413319152417573,
      "grad_norm": 0.07894159108400345,
      "learning_rate": 1.1204381269591097e-06,
      "loss": 0.001,
      "step": 147490
    },
    {
      "epoch": 2.413482778368649,
      "grad_norm": 0.10689076781272888,
      "learning_rate": 1.1198376747298022e-06,
      "loss": 0.0007,
      "step": 147500
    },
    {
      "epoch": 2.413646404319725,
      "grad_norm": 0.03708473592996597,
      "learning_rate": 1.1192373631470516e-06,
      "loss": 0.0005,
      "step": 147510
    },
    {
      "epoch": 2.4138100302708008,
      "grad_norm": 0.01977674663066864,
      "learning_rate": 1.1186371922326228e-06,
      "loss": 0.0008,
      "step": 147520
    },
    {
      "epoch": 2.4139736562218768,
      "grad_norm": 0.005746379029005766,
      "learning_rate": 1.1180371620082641e-06,
      "loss": 0.0017,
      "step": 147530
    },
    {
      "epoch": 2.4141372821729528,
      "grad_norm": 0.08772705495357513,
      "learning_rate": 1.1174372724957294e-06,
      "loss": 0.0011,
      "step": 147540
    },
    {
      "epoch": 2.4143009081240283,
      "grad_norm": 0.1947985142469406,
      "learning_rate": 1.116837523716761e-06,
      "loss": 0.0012,
      "step": 147550
    },
    {
      "epoch": 2.4144645340751043,
      "grad_norm": 0.07773470133543015,
      "learning_rate": 1.1162379156930997e-06,
      "loss": 0.0005,
      "step": 147560
    },
    {
      "epoch": 2.4146281600261803,
      "grad_norm": 0.017453191801905632,
      "learning_rate": 1.1156384484464805e-06,
      "loss": 0.0005,
      "step": 147570
    },
    {
      "epoch": 2.414791785977256,
      "grad_norm": 0.096875861287117,
      "learning_rate": 1.11503912199863e-06,
      "loss": 0.0004,
      "step": 147580
    },
    {
      "epoch": 2.414955411928332,
      "grad_norm": 0.04409999027848244,
      "learning_rate": 1.114439936371276e-06,
      "loss": 0.0006,
      "step": 147590
    },
    {
      "epoch": 2.415119037879408,
      "grad_norm": 0.008241796866059303,
      "learning_rate": 1.1138408915861343e-06,
      "loss": 0.0009,
      "step": 147600
    },
    {
      "epoch": 2.4152826638304834,
      "grad_norm": 0.013936730101704597,
      "learning_rate": 1.1132419876649214e-06,
      "loss": 0.0006,
      "step": 147610
    },
    {
      "epoch": 2.4154462897815594,
      "grad_norm": 0.05870835483074188,
      "learning_rate": 1.112643224629345e-06,
      "loss": 0.0007,
      "step": 147620
    },
    {
      "epoch": 2.4156099157326354,
      "grad_norm": 0.20783457159996033,
      "learning_rate": 1.1120446025011071e-06,
      "loss": 0.0016,
      "step": 147630
    },
    {
      "epoch": 2.415773541683711,
      "grad_norm": 0.14202994108200073,
      "learning_rate": 1.111446121301909e-06,
      "loss": 0.0008,
      "step": 147640
    },
    {
      "epoch": 2.415937167634787,
      "grad_norm": 0.027852363884449005,
      "learning_rate": 1.1108477810534423e-06,
      "loss": 0.0006,
      "step": 147650
    },
    {
      "epoch": 2.416100793585863,
      "grad_norm": 0.02043231576681137,
      "learning_rate": 1.1102495817773973e-06,
      "loss": 0.0004,
      "step": 147660
    },
    {
      "epoch": 2.4162644195369385,
      "grad_norm": 0.12623021006584167,
      "learning_rate": 1.1096515234954547e-06,
      "loss": 0.0008,
      "step": 147670
    },
    {
      "epoch": 2.4164280454880145,
      "grad_norm": 0.12809127569198608,
      "learning_rate": 1.1090536062292956e-06,
      "loss": 0.0008,
      "step": 147680
    },
    {
      "epoch": 2.4165916714390905,
      "grad_norm": 0.013466783799231052,
      "learning_rate": 1.1084558300005905e-06,
      "loss": 0.0013,
      "step": 147690
    },
    {
      "epoch": 2.416755297390166,
      "grad_norm": 0.04727175831794739,
      "learning_rate": 1.1078581948310096e-06,
      "loss": 0.0006,
      "step": 147700
    },
    {
      "epoch": 2.416918923341242,
      "grad_norm": 0.12849493324756622,
      "learning_rate": 1.1072607007422136e-06,
      "loss": 0.0013,
      "step": 147710
    },
    {
      "epoch": 2.4170825492923176,
      "grad_norm": 0.12960892915725708,
      "learning_rate": 1.106663347755861e-06,
      "loss": 0.0007,
      "step": 147720
    },
    {
      "epoch": 2.4172461752433936,
      "grad_norm": 0.06974505633115768,
      "learning_rate": 1.1060661358936076e-06,
      "loss": 0.0009,
      "step": 147730
    },
    {
      "epoch": 2.4174098011944696,
      "grad_norm": 0.040147941559553146,
      "learning_rate": 1.105469065177095e-06,
      "loss": 0.0008,
      "step": 147740
    },
    {
      "epoch": 2.417573427145545,
      "grad_norm": 0.08561286330223083,
      "learning_rate": 1.1048721356279695e-06,
      "loss": 0.0009,
      "step": 147750
    },
    {
      "epoch": 2.417737053096621,
      "grad_norm": 0.05234234780073166,
      "learning_rate": 1.1042753472678663e-06,
      "loss": 0.0008,
      "step": 147760
    },
    {
      "epoch": 2.417900679047697,
      "grad_norm": 0.11896946281194687,
      "learning_rate": 1.103678700118419e-06,
      "loss": 0.0008,
      "step": 147770
    },
    {
      "epoch": 2.4180643049987727,
      "grad_norm": 0.06305025517940521,
      "learning_rate": 1.1030821942012553e-06,
      "loss": 0.0008,
      "step": 147780
    },
    {
      "epoch": 2.4182279309498487,
      "grad_norm": 0.0925489142537117,
      "learning_rate": 1.1024858295379949e-06,
      "loss": 0.0008,
      "step": 147790
    },
    {
      "epoch": 2.4183915569009247,
      "grad_norm": 0.009487279690802097,
      "learning_rate": 1.101889606150257e-06,
      "loss": 0.0008,
      "step": 147800
    },
    {
      "epoch": 2.418555182852,
      "grad_norm": 0.31293049454689026,
      "learning_rate": 1.1012935240596506e-06,
      "loss": 0.0011,
      "step": 147810
    },
    {
      "epoch": 2.418718808803076,
      "grad_norm": 0.06797828525304794,
      "learning_rate": 1.1006975832877853e-06,
      "loss": 0.0008,
      "step": 147820
    },
    {
      "epoch": 2.4188824347541518,
      "grad_norm": 0.073615662753582,
      "learning_rate": 1.1001017838562606e-06,
      "loss": 0.0007,
      "step": 147830
    },
    {
      "epoch": 2.4190460607052278,
      "grad_norm": 0.09402614086866379,
      "learning_rate": 1.0995061257866718e-06,
      "loss": 0.0012,
      "step": 147840
    },
    {
      "epoch": 2.4192096866563038,
      "grad_norm": 0.09767060726881027,
      "learning_rate": 1.0989106091006125e-06,
      "loss": 0.0008,
      "step": 147850
    },
    {
      "epoch": 2.4193733126073793,
      "grad_norm": 0.02674877643585205,
      "learning_rate": 1.0983152338196662e-06,
      "loss": 0.001,
      "step": 147860
    },
    {
      "epoch": 2.4195369385584553,
      "grad_norm": 0.022399865090847015,
      "learning_rate": 1.097719999965417e-06,
      "loss": 0.001,
      "step": 147870
    },
    {
      "epoch": 2.4197005645095313,
      "grad_norm": 0.1608082801103592,
      "learning_rate": 1.0971249075594376e-06,
      "loss": 0.001,
      "step": 147880
    },
    {
      "epoch": 2.419864190460607,
      "grad_norm": 0.0173958633095026,
      "learning_rate": 1.0965299566233007e-06,
      "loss": 0.0007,
      "step": 147890
    },
    {
      "epoch": 2.420027816411683,
      "grad_norm": 0.05245708301663399,
      "learning_rate": 1.09593514717857e-06,
      "loss": 0.0007,
      "step": 147900
    },
    {
      "epoch": 2.420191442362759,
      "grad_norm": 0.07192543148994446,
      "learning_rate": 1.0953404792468086e-06,
      "loss": 0.0004,
      "step": 147910
    },
    {
      "epoch": 2.4203550683138344,
      "grad_norm": 0.07834618538618088,
      "learning_rate": 1.094745952849569e-06,
      "loss": 0.0004,
      "step": 147920
    },
    {
      "epoch": 2.4205186942649104,
      "grad_norm": 0.08509614318609238,
      "learning_rate": 1.0941515680084035e-06,
      "loss": 0.0007,
      "step": 147930
    },
    {
      "epoch": 2.4206823202159864,
      "grad_norm": 0.06275741755962372,
      "learning_rate": 1.0935573247448566e-06,
      "loss": 0.0012,
      "step": 147940
    },
    {
      "epoch": 2.420845946167062,
      "grad_norm": 0.13832004368305206,
      "learning_rate": 1.0929632230804665e-06,
      "loss": 0.0011,
      "step": 147950
    },
    {
      "epoch": 2.421009572118138,
      "grad_norm": 0.050170887261629105,
      "learning_rate": 1.0923692630367704e-06,
      "loss": 0.0005,
      "step": 147960
    },
    {
      "epoch": 2.421173198069214,
      "grad_norm": 0.05667343735694885,
      "learning_rate": 1.0917754446352958e-06,
      "loss": 0.0005,
      "step": 147970
    },
    {
      "epoch": 2.4213368240202895,
      "grad_norm": 0.07946646958589554,
      "learning_rate": 1.0911817678975672e-06,
      "loss": 0.0008,
      "step": 147980
    },
    {
      "epoch": 2.4215004499713655,
      "grad_norm": 0.021370291709899902,
      "learning_rate": 1.0905882328451073e-06,
      "loss": 0.001,
      "step": 147990
    },
    {
      "epoch": 2.4216640759224415,
      "grad_norm": 0.033564623445272446,
      "learning_rate": 1.0899948394994258e-06,
      "loss": 0.0012,
      "step": 148000
    },
    {
      "epoch": 2.421827701873517,
      "grad_norm": 0.07143114507198334,
      "learning_rate": 1.0894015878820358e-06,
      "loss": 0.0019,
      "step": 148010
    },
    {
      "epoch": 2.421991327824593,
      "grad_norm": 0.16682365536689758,
      "learning_rate": 1.0888084780144376e-06,
      "loss": 0.0009,
      "step": 148020
    },
    {
      "epoch": 2.422154953775669,
      "grad_norm": 0.06324465572834015,
      "learning_rate": 1.0882155099181353e-06,
      "loss": 0.0007,
      "step": 148030
    },
    {
      "epoch": 2.4223185797267446,
      "grad_norm": 0.05568674951791763,
      "learning_rate": 1.0876226836146154e-06,
      "loss": 0.0006,
      "step": 148040
    },
    {
      "epoch": 2.4224822056778206,
      "grad_norm": 0.02703133225440979,
      "learning_rate": 1.0870299991253702e-06,
      "loss": 0.0017,
      "step": 148050
    },
    {
      "epoch": 2.4226458316288966,
      "grad_norm": 0.07188435643911362,
      "learning_rate": 1.0864374564718848e-06,
      "loss": 0.0011,
      "step": 148060
    },
    {
      "epoch": 2.422809457579972,
      "grad_norm": 0.00939931906759739,
      "learning_rate": 1.0858450556756338e-06,
      "loss": 0.0006,
      "step": 148070
    },
    {
      "epoch": 2.422973083531048,
      "grad_norm": 0.005172068253159523,
      "learning_rate": 1.0852527967580933e-06,
      "loss": 0.0004,
      "step": 148080
    },
    {
      "epoch": 2.4231367094821237,
      "grad_norm": 0.11073308438062668,
      "learning_rate": 1.0846606797407294e-06,
      "loss": 0.0013,
      "step": 148090
    },
    {
      "epoch": 2.4233003354331997,
      "grad_norm": 0.07497929036617279,
      "learning_rate": 1.0840687046450066e-06,
      "loss": 0.0008,
      "step": 148100
    },
    {
      "epoch": 2.4234639613842757,
      "grad_norm": 0.07553837448358536,
      "learning_rate": 1.08347687149238e-06,
      "loss": 0.0005,
      "step": 148110
    },
    {
      "epoch": 2.423627587335351,
      "grad_norm": 0.11249102652072906,
      "learning_rate": 1.0828851803043055e-06,
      "loss": 0.0006,
      "step": 148120
    },
    {
      "epoch": 2.423791213286427,
      "grad_norm": 0.009983002208173275,
      "learning_rate": 1.082293631102228e-06,
      "loss": 0.0003,
      "step": 148130
    },
    {
      "epoch": 2.423954839237503,
      "grad_norm": 0.10860998928546906,
      "learning_rate": 1.0817022239075897e-06,
      "loss": 0.0007,
      "step": 148140
    },
    {
      "epoch": 2.4241184651885788,
      "grad_norm": 0.07956936210393906,
      "learning_rate": 1.0811109587418294e-06,
      "loss": 0.0035,
      "step": 148150
    },
    {
      "epoch": 2.4242820911396548,
      "grad_norm": 0.055588968098163605,
      "learning_rate": 1.0805198356263768e-06,
      "loss": 0.0011,
      "step": 148160
    },
    {
      "epoch": 2.4244457170907308,
      "grad_norm": 0.03204049542546272,
      "learning_rate": 1.0799288545826613e-06,
      "loss": 0.001,
      "step": 148170
    },
    {
      "epoch": 2.4246093430418063,
      "grad_norm": 0.039452750235795975,
      "learning_rate": 1.079338015632102e-06,
      "loss": 0.0005,
      "step": 148180
    },
    {
      "epoch": 2.4247729689928823,
      "grad_norm": 0.026575636118650436,
      "learning_rate": 1.0787473187961179e-06,
      "loss": 0.001,
      "step": 148190
    },
    {
      "epoch": 2.4249365949439583,
      "grad_norm": 0.02265419065952301,
      "learning_rate": 1.078156764096117e-06,
      "loss": 0.0005,
      "step": 148200
    },
    {
      "epoch": 2.425100220895034,
      "grad_norm": 0.3368068039417267,
      "learning_rate": 1.0775663515535084e-06,
      "loss": 0.0011,
      "step": 148210
    },
    {
      "epoch": 2.42526384684611,
      "grad_norm": 0.1913364976644516,
      "learning_rate": 1.0769760811896934e-06,
      "loss": 0.001,
      "step": 148220
    },
    {
      "epoch": 2.4254274727971854,
      "grad_norm": 0.07805593311786652,
      "learning_rate": 1.0763859530260661e-06,
      "loss": 0.0026,
      "step": 148230
    },
    {
      "epoch": 2.4255910987482614,
      "grad_norm": 0.09813379496335983,
      "learning_rate": 1.0757959670840185e-06,
      "loss": 0.0014,
      "step": 148240
    },
    {
      "epoch": 2.4257547246993374,
      "grad_norm": 0.08269993960857391,
      "learning_rate": 1.0752061233849337e-06,
      "loss": 0.0007,
      "step": 148250
    },
    {
      "epoch": 2.425918350650413,
      "grad_norm": 0.03976655378937721,
      "learning_rate": 1.0746164219501942e-06,
      "loss": 0.0006,
      "step": 148260
    },
    {
      "epoch": 2.426081976601489,
      "grad_norm": 0.005758312996476889,
      "learning_rate": 1.0740268628011763e-06,
      "loss": 0.0009,
      "step": 148270
    },
    {
      "epoch": 2.426245602552565,
      "grad_norm": 0.061509136110544205,
      "learning_rate": 1.0734374459592478e-06,
      "loss": 0.0013,
      "step": 148280
    },
    {
      "epoch": 2.4264092285036405,
      "grad_norm": 0.01987549662590027,
      "learning_rate": 1.0728481714457755e-06,
      "loss": 0.0006,
      "step": 148290
    },
    {
      "epoch": 2.4265728544547165,
      "grad_norm": 0.07388292998075485,
      "learning_rate": 1.0722590392821169e-06,
      "loss": 0.0013,
      "step": 148300
    },
    {
      "epoch": 2.4267364804057925,
      "grad_norm": 0.0431431420147419,
      "learning_rate": 1.0716700494896299e-06,
      "loss": 0.0003,
      "step": 148310
    },
    {
      "epoch": 2.426900106356868,
      "grad_norm": 0.27885037660598755,
      "learning_rate": 1.0710812020896606e-06,
      "loss": 0.0013,
      "step": 148320
    },
    {
      "epoch": 2.427063732307944,
      "grad_norm": 0.058474019169807434,
      "learning_rate": 1.0704924971035557e-06,
      "loss": 0.0009,
      "step": 148330
    },
    {
      "epoch": 2.42722735825902,
      "grad_norm": 0.30061018466949463,
      "learning_rate": 1.069903934552654e-06,
      "loss": 0.0008,
      "step": 148340
    },
    {
      "epoch": 2.4273909842100956,
      "grad_norm": 0.050697799772024155,
      "learning_rate": 1.0693155144582872e-06,
      "loss": 0.0008,
      "step": 148350
    },
    {
      "epoch": 2.4275546101611716,
      "grad_norm": 0.13990169763565063,
      "learning_rate": 1.0687272368417872e-06,
      "loss": 0.0009,
      "step": 148360
    },
    {
      "epoch": 2.4277182361122476,
      "grad_norm": 0.021898178383708,
      "learning_rate": 1.0681391017244752e-06,
      "loss": 0.0008,
      "step": 148370
    },
    {
      "epoch": 2.427881862063323,
      "grad_norm": 0.017092812806367874,
      "learning_rate": 1.0675511091276724e-06,
      "loss": 0.0006,
      "step": 148380
    },
    {
      "epoch": 2.428045488014399,
      "grad_norm": 0.10358000546693802,
      "learning_rate": 1.0669632590726887e-06,
      "loss": 0.0006,
      "step": 148390
    },
    {
      "epoch": 2.428209113965475,
      "grad_norm": 0.026845429092645645,
      "learning_rate": 1.0663755515808355e-06,
      "loss": 0.0015,
      "step": 148400
    },
    {
      "epoch": 2.4283727399165507,
      "grad_norm": 0.028029434382915497,
      "learning_rate": 1.0657879866734129e-06,
      "loss": 0.0005,
      "step": 148410
    },
    {
      "epoch": 2.4285363658676267,
      "grad_norm": 0.06742192804813385,
      "learning_rate": 1.0652005643717201e-06,
      "loss": 0.0011,
      "step": 148420
    },
    {
      "epoch": 2.4286999918187027,
      "grad_norm": 0.08109088987112045,
      "learning_rate": 1.0646132846970531e-06,
      "loss": 0.0012,
      "step": 148430
    },
    {
      "epoch": 2.428863617769778,
      "grad_norm": 0.01835411973297596,
      "learning_rate": 1.0640261476706925e-06,
      "loss": 0.0006,
      "step": 148440
    },
    {
      "epoch": 2.429027243720854,
      "grad_norm": 0.09666718542575836,
      "learning_rate": 1.063439153313926e-06,
      "loss": 0.001,
      "step": 148450
    },
    {
      "epoch": 2.42919086967193,
      "grad_norm": 0.026973022148013115,
      "learning_rate": 1.0628523016480275e-06,
      "loss": 0.0007,
      "step": 148460
    },
    {
      "epoch": 2.4293544956230058,
      "grad_norm": 0.0076769450679421425,
      "learning_rate": 1.0622655926942721e-06,
      "loss": 0.0013,
      "step": 148470
    },
    {
      "epoch": 2.4295181215740818,
      "grad_norm": 0.10681433975696564,
      "learning_rate": 1.061679026473923e-06,
      "loss": 0.0005,
      "step": 148480
    },
    {
      "epoch": 2.4296817475251573,
      "grad_norm": 0.15214432775974274,
      "learning_rate": 1.0610926030082442e-06,
      "loss": 0.0007,
      "step": 148490
    },
    {
      "epoch": 2.4298453734762333,
      "grad_norm": 0.08576693385839462,
      "learning_rate": 1.0605063223184926e-06,
      "loss": 0.0005,
      "step": 148500
    },
    {
      "epoch": 2.4300089994273093,
      "grad_norm": 0.0789518728852272,
      "learning_rate": 1.0599201844259171e-06,
      "loss": 0.0005,
      "step": 148510
    },
    {
      "epoch": 2.430172625378385,
      "grad_norm": 0.13777755200862885,
      "learning_rate": 1.0593341893517666e-06,
      "loss": 0.001,
      "step": 148520
    },
    {
      "epoch": 2.430336251329461,
      "grad_norm": 0.11252978444099426,
      "learning_rate": 1.0587483371172802e-06,
      "loss": 0.0015,
      "step": 148530
    },
    {
      "epoch": 2.430499877280537,
      "grad_norm": 0.12324199080467224,
      "learning_rate": 1.0581626277436936e-06,
      "loss": 0.0004,
      "step": 148540
    },
    {
      "epoch": 2.4306635032316124,
      "grad_norm": 0.07565046846866608,
      "learning_rate": 1.0575770612522385e-06,
      "loss": 0.0006,
      "step": 148550
    },
    {
      "epoch": 2.4308271291826884,
      "grad_norm": 0.01156596653163433,
      "learning_rate": 1.056991637664138e-06,
      "loss": 0.0008,
      "step": 148560
    },
    {
      "epoch": 2.4309907551337644,
      "grad_norm": 0.00789470225572586,
      "learning_rate": 1.0564063570006156e-06,
      "loss": 0.002,
      "step": 148570
    },
    {
      "epoch": 2.43115438108484,
      "grad_norm": 0.002704763552173972,
      "learning_rate": 1.055821219282883e-06,
      "loss": 0.0013,
      "step": 148580
    },
    {
      "epoch": 2.431318007035916,
      "grad_norm": 0.0011895995121449232,
      "learning_rate": 1.0552362245321535e-06,
      "loss": 0.0012,
      "step": 148590
    },
    {
      "epoch": 2.4314816329869915,
      "grad_norm": 0.06007545441389084,
      "learning_rate": 1.0546513727696284e-06,
      "loss": 0.0004,
      "step": 148600
    },
    {
      "epoch": 2.4316452589380675,
      "grad_norm": 0.02888408862054348,
      "learning_rate": 1.0540666640165097e-06,
      "loss": 0.0005,
      "step": 148610
    },
    {
      "epoch": 2.4318088848891435,
      "grad_norm": 0.08068275451660156,
      "learning_rate": 1.0534820982939893e-06,
      "loss": 0.0013,
      "step": 148620
    },
    {
      "epoch": 2.431972510840219,
      "grad_norm": 0.05243679881095886,
      "learning_rate": 1.052897675623259e-06,
      "loss": 0.0009,
      "step": 148630
    },
    {
      "epoch": 2.432136136791295,
      "grad_norm": 0.013122976757586002,
      "learning_rate": 1.0523133960255016e-06,
      "loss": 0.0008,
      "step": 148640
    },
    {
      "epoch": 2.432299762742371,
      "grad_norm": 0.037479501217603683,
      "learning_rate": 1.0517292595218936e-06,
      "loss": 0.0009,
      "step": 148650
    },
    {
      "epoch": 2.4324633886934466,
      "grad_norm": 0.05510633438825607,
      "learning_rate": 1.0511452661336125e-06,
      "loss": 0.0007,
      "step": 148660
    },
    {
      "epoch": 2.4326270146445226,
      "grad_norm": 0.09597143530845642,
      "learning_rate": 1.0505614158818228e-06,
      "loss": 0.0009,
      "step": 148670
    },
    {
      "epoch": 2.4327906405955986,
      "grad_norm": 0.07137784361839294,
      "learning_rate": 1.0499777087876917e-06,
      "loss": 0.0009,
      "step": 148680
    },
    {
      "epoch": 2.432954266546674,
      "grad_norm": 0.048212554305791855,
      "learning_rate": 1.0493941448723726e-06,
      "loss": 0.001,
      "step": 148690
    },
    {
      "epoch": 2.43311789249775,
      "grad_norm": 0.044151097536087036,
      "learning_rate": 1.0488107241570217e-06,
      "loss": 0.0011,
      "step": 148700
    },
    {
      "epoch": 2.433281518448826,
      "grad_norm": 0.025755858048796654,
      "learning_rate": 1.0482274466627868e-06,
      "loss": 0.0009,
      "step": 148710
    },
    {
      "epoch": 2.4334451443999017,
      "grad_norm": 0.0961119681596756,
      "learning_rate": 1.0476443124108071e-06,
      "loss": 0.0004,
      "step": 148720
    },
    {
      "epoch": 2.4336087703509777,
      "grad_norm": 0.02293434739112854,
      "learning_rate": 1.0470613214222253e-06,
      "loss": 0.0006,
      "step": 148730
    },
    {
      "epoch": 2.4337723963020537,
      "grad_norm": 0.03176530823111534,
      "learning_rate": 1.0464784737181666e-06,
      "loss": 0.0007,
      "step": 148740
    },
    {
      "epoch": 2.433936022253129,
      "grad_norm": 0.07863471657037735,
      "learning_rate": 1.045895769319762e-06,
      "loss": 0.0005,
      "step": 148750
    },
    {
      "epoch": 2.434099648204205,
      "grad_norm": 0.1033850759267807,
      "learning_rate": 1.0453132082481326e-06,
      "loss": 0.0012,
      "step": 148760
    },
    {
      "epoch": 2.434263274155281,
      "grad_norm": 0.1333463042974472,
      "learning_rate": 1.044730790524394e-06,
      "loss": 0.0007,
      "step": 148770
    },
    {
      "epoch": 2.4344269001063568,
      "grad_norm": 0.04183622822165489,
      "learning_rate": 1.0441485161696591e-06,
      "loss": 0.0007,
      "step": 148780
    },
    {
      "epoch": 2.4345905260574328,
      "grad_norm": 0.15752290189266205,
      "learning_rate": 1.0435663852050315e-06,
      "loss": 0.0008,
      "step": 148790
    },
    {
      "epoch": 2.4347541520085088,
      "grad_norm": 0.023099087178707123,
      "learning_rate": 1.042984397651615e-06,
      "loss": 0.0012,
      "step": 148800
    },
    {
      "epoch": 2.4349177779595843,
      "grad_norm": 0.08653181046247482,
      "learning_rate": 1.042402553530502e-06,
      "loss": 0.0011,
      "step": 148810
    },
    {
      "epoch": 2.4350814039106603,
      "grad_norm": 0.057959381490945816,
      "learning_rate": 1.041820852862786e-06,
      "loss": 0.0005,
      "step": 148820
    },
    {
      "epoch": 2.4352450298617363,
      "grad_norm": 0.06256592273712158,
      "learning_rate": 1.041239295669551e-06,
      "loss": 0.0005,
      "step": 148830
    },
    {
      "epoch": 2.435408655812812,
      "grad_norm": 0.022995127364993095,
      "learning_rate": 1.0406578819718755e-06,
      "loss": 0.0006,
      "step": 148840
    },
    {
      "epoch": 2.435572281763888,
      "grad_norm": 0.06696300953626633,
      "learning_rate": 1.0400766117908374e-06,
      "loss": 0.0013,
      "step": 148850
    },
    {
      "epoch": 2.4357359077149634,
      "grad_norm": 0.0834878534078598,
      "learning_rate": 1.0394954851475025e-06,
      "loss": 0.0005,
      "step": 148860
    },
    {
      "epoch": 2.4358995336660394,
      "grad_norm": 0.032019905745983124,
      "learning_rate": 1.0389145020629398e-06,
      "loss": 0.0007,
      "step": 148870
    },
    {
      "epoch": 2.4360631596171154,
      "grad_norm": 0.1997094303369522,
      "learning_rate": 1.038333662558204e-06,
      "loss": 0.0013,
      "step": 148880
    },
    {
      "epoch": 2.436226785568191,
      "grad_norm": 0.03288447856903076,
      "learning_rate": 1.037752966654353e-06,
      "loss": 0.0004,
      "step": 148890
    },
    {
      "epoch": 2.436390411519267,
      "grad_norm": 0.03944979980587959,
      "learning_rate": 1.0371724143724326e-06,
      "loss": 0.0007,
      "step": 148900
    },
    {
      "epoch": 2.436554037470343,
      "grad_norm": 0.028130657970905304,
      "learning_rate": 1.0365920057334893e-06,
      "loss": 0.0007,
      "step": 148910
    },
    {
      "epoch": 2.4367176634214185,
      "grad_norm": 0.07924254238605499,
      "learning_rate": 1.0360117407585584e-06,
      "loss": 0.0012,
      "step": 148920
    },
    {
      "epoch": 2.4368812893724945,
      "grad_norm": 0.002409644890576601,
      "learning_rate": 1.0354316194686764e-06,
      "loss": 0.0006,
      "step": 148930
    },
    {
      "epoch": 2.4370449153235705,
      "grad_norm": 0.008279784582555294,
      "learning_rate": 1.0348516418848691e-06,
      "loss": 0.0003,
      "step": 148940
    },
    {
      "epoch": 2.437208541274646,
      "grad_norm": 0.051741521805524826,
      "learning_rate": 1.0342718080281588e-06,
      "loss": 0.0007,
      "step": 148950
    },
    {
      "epoch": 2.437372167225722,
      "grad_norm": 0.2284378558397293,
      "learning_rate": 1.0336921179195653e-06,
      "loss": 0.0007,
      "step": 148960
    },
    {
      "epoch": 2.437535793176798,
      "grad_norm": 0.038089849054813385,
      "learning_rate": 1.0331125715800982e-06,
      "loss": 0.0005,
      "step": 148970
    },
    {
      "epoch": 2.4376994191278736,
      "grad_norm": 0.09312382340431213,
      "learning_rate": 1.0325331690307666e-06,
      "loss": 0.0009,
      "step": 148980
    },
    {
      "epoch": 2.4378630450789496,
      "grad_norm": 0.14112703502178192,
      "learning_rate": 1.031953910292573e-06,
      "loss": 0.0004,
      "step": 148990
    },
    {
      "epoch": 2.438026671030025,
      "grad_norm": 0.03761272504925728,
      "learning_rate": 1.0313747953865123e-06,
      "loss": 0.0006,
      "step": 149000
    },
    {
      "epoch": 2.438190296981101,
      "grad_norm": 0.03520971164107323,
      "learning_rate": 1.0307958243335781e-06,
      "loss": 0.0008,
      "step": 149010
    },
    {
      "epoch": 2.438353922932177,
      "grad_norm": 0.016446659341454506,
      "learning_rate": 1.0302169971547542e-06,
      "loss": 0.0007,
      "step": 149020
    },
    {
      "epoch": 2.4385175488832527,
      "grad_norm": 0.04374958574771881,
      "learning_rate": 1.0296383138710241e-06,
      "loss": 0.0011,
      "step": 149030
    },
    {
      "epoch": 2.4386811748343287,
      "grad_norm": 0.07221683114767075,
      "learning_rate": 1.029059774503363e-06,
      "loss": 0.0005,
      "step": 149040
    },
    {
      "epoch": 2.4388448007854047,
      "grad_norm": 0.11075487732887268,
      "learning_rate": 1.0284813790727395e-06,
      "loss": 0.0008,
      "step": 149050
    },
    {
      "epoch": 2.43900842673648,
      "grad_norm": 0.11402343213558197,
      "learning_rate": 1.0279031276001222e-06,
      "loss": 0.0008,
      "step": 149060
    },
    {
      "epoch": 2.439172052687556,
      "grad_norm": 0.1286894679069519,
      "learning_rate": 1.0273250201064689e-06,
      "loss": 0.0009,
      "step": 149070
    },
    {
      "epoch": 2.439335678638632,
      "grad_norm": 0.015726864337921143,
      "learning_rate": 1.0267470566127363e-06,
      "loss": 0.0006,
      "step": 149080
    },
    {
      "epoch": 2.4394993045897078,
      "grad_norm": 0.0982898697257042,
      "learning_rate": 1.0261692371398724e-06,
      "loss": 0.0006,
      "step": 149090
    },
    {
      "epoch": 2.4396629305407838,
      "grad_norm": 0.08235268294811249,
      "learning_rate": 1.025591561708824e-06,
      "loss": 0.0005,
      "step": 149100
    },
    {
      "epoch": 2.4398265564918598,
      "grad_norm": 0.00945932324975729,
      "learning_rate": 1.0250140303405282e-06,
      "loss": 0.0005,
      "step": 149110
    },
    {
      "epoch": 2.4399901824429353,
      "grad_norm": 0.40362250804901123,
      "learning_rate": 1.024436643055921e-06,
      "loss": 0.0016,
      "step": 149120
    },
    {
      "epoch": 2.4401538083940113,
      "grad_norm": 0.07312536239624023,
      "learning_rate": 1.023859399875931e-06,
      "loss": 0.0006,
      "step": 149130
    },
    {
      "epoch": 2.4403174343450873,
      "grad_norm": 0.06973206996917725,
      "learning_rate": 1.0232823008214798e-06,
      "loss": 0.0003,
      "step": 149140
    },
    {
      "epoch": 2.440481060296163,
      "grad_norm": 0.06767765432596207,
      "learning_rate": 1.0227053459134884e-06,
      "loss": 0.0006,
      "step": 149150
    },
    {
      "epoch": 2.440644686247239,
      "grad_norm": 0.10995814204216003,
      "learning_rate": 1.0221285351728682e-06,
      "loss": 0.0009,
      "step": 149160
    },
    {
      "epoch": 2.440808312198315,
      "grad_norm": 0.022389668971300125,
      "learning_rate": 1.021551868620529e-06,
      "loss": 0.0011,
      "step": 149170
    },
    {
      "epoch": 2.4409719381493904,
      "grad_norm": 0.009879834949970245,
      "learning_rate": 1.0209753462773713e-06,
      "loss": 0.0006,
      "step": 149180
    },
    {
      "epoch": 2.4411355641004664,
      "grad_norm": 0.039834361523389816,
      "learning_rate": 1.0203989681642946e-06,
      "loss": 0.001,
      "step": 149190
    },
    {
      "epoch": 2.4412991900515424,
      "grad_norm": 0.061837732791900635,
      "learning_rate": 1.0198227343021911e-06,
      "loss": 0.0008,
      "step": 149200
    },
    {
      "epoch": 2.441462816002618,
      "grad_norm": 0.13374507427215576,
      "learning_rate": 1.0192466447119464e-06,
      "loss": 0.0007,
      "step": 149210
    },
    {
      "epoch": 2.441626441953694,
      "grad_norm": 0.05544538050889969,
      "learning_rate": 1.0186706994144445e-06,
      "loss": 0.0008,
      "step": 149220
    },
    {
      "epoch": 2.44179006790477,
      "grad_norm": 0.0643562600016594,
      "learning_rate": 1.018094898430561e-06,
      "loss": 0.0009,
      "step": 149230
    },
    {
      "epoch": 2.4419536938558455,
      "grad_norm": 0.12697699666023254,
      "learning_rate": 1.0175192417811675e-06,
      "loss": 0.0013,
      "step": 149240
    },
    {
      "epoch": 2.4421173198069215,
      "grad_norm": 0.020932389423251152,
      "learning_rate": 1.016943729487128e-06,
      "loss": 0.0009,
      "step": 149250
    },
    {
      "epoch": 2.442280945757997,
      "grad_norm": 0.025386499240994453,
      "learning_rate": 1.016368361569306e-06,
      "loss": 0.0008,
      "step": 149260
    },
    {
      "epoch": 2.442444571709073,
      "grad_norm": 0.0721283107995987,
      "learning_rate": 1.015793138048557e-06,
      "loss": 0.0006,
      "step": 149270
    },
    {
      "epoch": 2.442608197660149,
      "grad_norm": 0.0813249722123146,
      "learning_rate": 1.01521805894573e-06,
      "loss": 0.0006,
      "step": 149280
    },
    {
      "epoch": 2.4427718236112246,
      "grad_norm": 0.03963208571076393,
      "learning_rate": 1.0146431242816723e-06,
      "loss": 0.0006,
      "step": 149290
    },
    {
      "epoch": 2.4429354495623006,
      "grad_norm": 0.045182742178440094,
      "learning_rate": 1.0140683340772217e-06,
      "loss": 0.0006,
      "step": 149300
    },
    {
      "epoch": 2.4430990755133766,
      "grad_norm": 0.0777864158153534,
      "learning_rate": 1.0134936883532154e-06,
      "loss": 0.0006,
      "step": 149310
    },
    {
      "epoch": 2.443262701464452,
      "grad_norm": 0.07508112490177155,
      "learning_rate": 1.01291918713048e-06,
      "loss": 0.001,
      "step": 149320
    },
    {
      "epoch": 2.443426327415528,
      "grad_norm": 0.041357167065143585,
      "learning_rate": 1.0123448304298428e-06,
      "loss": 0.0007,
      "step": 149330
    },
    {
      "epoch": 2.443589953366604,
      "grad_norm": 0.03501289337873459,
      "learning_rate": 1.0117706182721215e-06,
      "loss": 0.0006,
      "step": 149340
    },
    {
      "epoch": 2.4437535793176797,
      "grad_norm": 0.042439885437488556,
      "learning_rate": 1.0111965506781286e-06,
      "loss": 0.0008,
      "step": 149350
    },
    {
      "epoch": 2.4439172052687557,
      "grad_norm": 0.042123135179281235,
      "learning_rate": 1.0106226276686753e-06,
      "loss": 0.0003,
      "step": 149360
    },
    {
      "epoch": 2.444080831219831,
      "grad_norm": 0.030775081366300583,
      "learning_rate": 1.0100488492645622e-06,
      "loss": 0.0009,
      "step": 149370
    },
    {
      "epoch": 2.444244457170907,
      "grad_norm": 0.04664922133088112,
      "learning_rate": 1.0094752154865895e-06,
      "loss": 0.0004,
      "step": 149380
    },
    {
      "epoch": 2.444408083121983,
      "grad_norm": 0.06221944838762283,
      "learning_rate": 1.0089017263555483e-06,
      "loss": 0.0004,
      "step": 149390
    },
    {
      "epoch": 2.4445717090730588,
      "grad_norm": 0.07560790330171585,
      "learning_rate": 1.0083283818922284e-06,
      "loss": 0.0007,
      "step": 149400
    },
    {
      "epoch": 2.4447353350241348,
      "grad_norm": 0.04249478131532669,
      "learning_rate": 1.0077551821174097e-06,
      "loss": 0.0016,
      "step": 149410
    },
    {
      "epoch": 2.4448989609752108,
      "grad_norm": 0.09271867573261261,
      "learning_rate": 1.0071821270518705e-06,
      "loss": 0.0008,
      "step": 149420
    },
    {
      "epoch": 2.4450625869262863,
      "grad_norm": 0.028555242344737053,
      "learning_rate": 1.0066092167163855e-06,
      "loss": 0.0008,
      "step": 149430
    },
    {
      "epoch": 2.4452262128773623,
      "grad_norm": 0.14769424498081207,
      "learning_rate": 1.0060364511317161e-06,
      "loss": 0.0007,
      "step": 149440
    },
    {
      "epoch": 2.4453898388284383,
      "grad_norm": 0.056885331869125366,
      "learning_rate": 1.0054638303186271e-06,
      "loss": 0.0005,
      "step": 149450
    },
    {
      "epoch": 2.445553464779514,
      "grad_norm": 0.0065961177460849285,
      "learning_rate": 1.0048913542978721e-06,
      "loss": 0.0003,
      "step": 149460
    },
    {
      "epoch": 2.44571709073059,
      "grad_norm": 0.3480302095413208,
      "learning_rate": 1.0043190230902039e-06,
      "loss": 0.0016,
      "step": 149470
    },
    {
      "epoch": 2.445880716681666,
      "grad_norm": 0.004454368259757757,
      "learning_rate": 1.0037468367163689e-06,
      "loss": 0.0007,
      "step": 149480
    },
    {
      "epoch": 2.4460443426327414,
      "grad_norm": 0.11310214549303055,
      "learning_rate": 1.003174795197105e-06,
      "loss": 0.0005,
      "step": 149490
    },
    {
      "epoch": 2.4462079685838174,
      "grad_norm": 0.049962688237428665,
      "learning_rate": 1.0026028985531499e-06,
      "loss": 0.0013,
      "step": 149500
    },
    {
      "epoch": 2.4463715945348934,
      "grad_norm": 0.07255863398313522,
      "learning_rate": 1.0020311468052312e-06,
      "loss": 0.0021,
      "step": 149510
    },
    {
      "epoch": 2.446535220485969,
      "grad_norm": 0.07148192077875137,
      "learning_rate": 1.0014595399740755e-06,
      "loss": 0.0008,
      "step": 149520
    },
    {
      "epoch": 2.446698846437045,
      "grad_norm": 0.10815980285406113,
      "learning_rate": 1.0008880780804014e-06,
      "loss": 0.0011,
      "step": 149530
    },
    {
      "epoch": 2.446862472388121,
      "grad_norm": 0.17929501831531525,
      "learning_rate": 1.0003167611449215e-06,
      "loss": 0.0005,
      "step": 149540
    },
    {
      "epoch": 2.4470260983391965,
      "grad_norm": 0.006514658220112324,
      "learning_rate": 9.997455891883473e-07,
      "loss": 0.0008,
      "step": 149550
    },
    {
      "epoch": 2.4471897242902725,
      "grad_norm": 0.08479084074497223,
      "learning_rate": 9.991745622313803e-07,
      "loss": 0.0006,
      "step": 149560
    },
    {
      "epoch": 2.4473533502413485,
      "grad_norm": 0.18950589001178741,
      "learning_rate": 9.986036802947207e-07,
      "loss": 0.0009,
      "step": 149570
    },
    {
      "epoch": 2.447516976192424,
      "grad_norm": 0.07851356267929077,
      "learning_rate": 9.980329433990588e-07,
      "loss": 0.0007,
      "step": 149580
    },
    {
      "epoch": 2.4476806021435,
      "grad_norm": 0.06302106380462646,
      "learning_rate": 9.97462351565086e-07,
      "loss": 0.0003,
      "step": 149590
    },
    {
      "epoch": 2.447844228094576,
      "grad_norm": 0.1629902571439743,
      "learning_rate": 9.968919048134818e-07,
      "loss": 0.0013,
      "step": 149600
    },
    {
      "epoch": 2.4480078540456516,
      "grad_norm": 0.03731869161128998,
      "learning_rate": 9.963216031649264e-07,
      "loss": 0.001,
      "step": 149610
    },
    {
      "epoch": 2.4481714799967276,
      "grad_norm": 0.06414326280355453,
      "learning_rate": 9.957514466400885e-07,
      "loss": 0.0007,
      "step": 149620
    },
    {
      "epoch": 2.4483351059478036,
      "grad_norm": 0.01729876548051834,
      "learning_rate": 9.951814352596384e-07,
      "loss": 0.0006,
      "step": 149630
    },
    {
      "epoch": 2.448498731898879,
      "grad_norm": 0.20005588233470917,
      "learning_rate": 9.946115690442354e-07,
      "loss": 0.0013,
      "step": 149640
    },
    {
      "epoch": 2.448662357849955,
      "grad_norm": 0.02489263191819191,
      "learning_rate": 9.940418480145352e-07,
      "loss": 0.0006,
      "step": 149650
    },
    {
      "epoch": 2.4488259838010307,
      "grad_norm": 0.017807386815547943,
      "learning_rate": 9.934722721911915e-07,
      "loss": 0.0008,
      "step": 149660
    },
    {
      "epoch": 2.4489896097521067,
      "grad_norm": 0.07528525590896606,
      "learning_rate": 9.929028415948472e-07,
      "loss": 0.0007,
      "step": 149670
    },
    {
      "epoch": 2.4491532357031827,
      "grad_norm": 0.0757269486784935,
      "learning_rate": 9.92333556246145e-07,
      "loss": 0.0009,
      "step": 149680
    },
    {
      "epoch": 2.449316861654258,
      "grad_norm": 0.05382451415061951,
      "learning_rate": 9.917644161657187e-07,
      "loss": 0.0006,
      "step": 149690
    },
    {
      "epoch": 2.449480487605334,
      "grad_norm": 0.04170555621385574,
      "learning_rate": 9.911954213741981e-07,
      "loss": 0.0007,
      "step": 149700
    },
    {
      "epoch": 2.44964411355641,
      "grad_norm": 0.004949782509356737,
      "learning_rate": 9.906265718922097e-07,
      "loss": 0.0018,
      "step": 149710
    },
    {
      "epoch": 2.4498077395074858,
      "grad_norm": 0.14138126373291016,
      "learning_rate": 9.90057867740371e-07,
      "loss": 0.0013,
      "step": 149720
    },
    {
      "epoch": 2.4499713654585618,
      "grad_norm": 0.0234735868871212,
      "learning_rate": 9.894893089392994e-07,
      "loss": 0.0014,
      "step": 149730
    },
    {
      "epoch": 2.4501349914096378,
      "grad_norm": 0.1300615519285202,
      "learning_rate": 9.889208955095986e-07,
      "loss": 0.0007,
      "step": 149740
    },
    {
      "epoch": 2.4502986173607133,
      "grad_norm": 0.045777950435876846,
      "learning_rate": 9.883526274718752e-07,
      "loss": 0.0008,
      "step": 149750
    },
    {
      "epoch": 2.4504622433117893,
      "grad_norm": 0.046468086540699005,
      "learning_rate": 9.877845048467282e-07,
      "loss": 0.0003,
      "step": 149760
    },
    {
      "epoch": 2.450625869262865,
      "grad_norm": 0.050322871655225754,
      "learning_rate": 9.87216527654749e-07,
      "loss": 0.0007,
      "step": 149770
    },
    {
      "epoch": 2.450789495213941,
      "grad_norm": 0.03755843639373779,
      "learning_rate": 9.866486959165273e-07,
      "loss": 0.0007,
      "step": 149780
    },
    {
      "epoch": 2.450953121165017,
      "grad_norm": 0.08474308252334595,
      "learning_rate": 9.86081009652643e-07,
      "loss": 0.0005,
      "step": 149790
    },
    {
      "epoch": 2.4511167471160924,
      "grad_norm": 0.07001464068889618,
      "learning_rate": 9.855134688836765e-07,
      "loss": 0.0005,
      "step": 149800
    },
    {
      "epoch": 2.4512803730671684,
      "grad_norm": 0.06313111633062363,
      "learning_rate": 9.84946073630197e-07,
      "loss": 0.0004,
      "step": 149810
    },
    {
      "epoch": 2.4514439990182444,
      "grad_norm": 0.05203631520271301,
      "learning_rate": 9.843788239127733e-07,
      "loss": 0.0006,
      "step": 149820
    },
    {
      "epoch": 2.45160762496932,
      "grad_norm": 0.021992292255163193,
      "learning_rate": 9.838117197519648e-07,
      "loss": 0.0011,
      "step": 149830
    },
    {
      "epoch": 2.451771250920396,
      "grad_norm": 0.005559555254876614,
      "learning_rate": 9.832447611683305e-07,
      "loss": 0.0003,
      "step": 149840
    },
    {
      "epoch": 2.451934876871472,
      "grad_norm": 0.04270913824439049,
      "learning_rate": 9.82677948182419e-07,
      "loss": 0.0008,
      "step": 149850
    },
    {
      "epoch": 2.4520985028225475,
      "grad_norm": 0.05839170888066292,
      "learning_rate": 9.821112808147753e-07,
      "loss": 0.0024,
      "step": 149860
    },
    {
      "epoch": 2.4522621287736235,
      "grad_norm": 0.06364672631025314,
      "learning_rate": 9.815447590859428e-07,
      "loss": 0.0007,
      "step": 149870
    },
    {
      "epoch": 2.4524257547246995,
      "grad_norm": 0.07619593292474747,
      "learning_rate": 9.809783830164526e-07,
      "loss": 0.0011,
      "step": 149880
    },
    {
      "epoch": 2.452589380675775,
      "grad_norm": 0.009764869697391987,
      "learning_rate": 9.804121526268385e-07,
      "loss": 0.0012,
      "step": 149890
    },
    {
      "epoch": 2.452753006626851,
      "grad_norm": 0.004904359579086304,
      "learning_rate": 9.798460679376215e-07,
      "loss": 0.0012,
      "step": 149900
    },
    {
      "epoch": 2.452916632577927,
      "grad_norm": 0.07469763606786728,
      "learning_rate": 9.79280128969322e-07,
      "loss": 0.001,
      "step": 149910
    },
    {
      "epoch": 2.4530802585290026,
      "grad_norm": 0.007057026494294405,
      "learning_rate": 9.787143357424562e-07,
      "loss": 0.0007,
      "step": 149920
    },
    {
      "epoch": 2.4532438844800786,
      "grad_norm": 0.123782679438591,
      "learning_rate": 9.781486882775293e-07,
      "loss": 0.0012,
      "step": 149930
    },
    {
      "epoch": 2.4534075104311546,
      "grad_norm": 0.004463258199393749,
      "learning_rate": 9.775831865950487e-07,
      "loss": 0.0006,
      "step": 149940
    },
    {
      "epoch": 2.45357113638223,
      "grad_norm": 0.022445421665906906,
      "learning_rate": 9.77017830715507e-07,
      "loss": 0.0007,
      "step": 149950
    },
    {
      "epoch": 2.453734762333306,
      "grad_norm": 0.11141089349985123,
      "learning_rate": 9.764526206594005e-07,
      "loss": 0.001,
      "step": 149960
    },
    {
      "epoch": 2.453898388284382,
      "grad_norm": 0.07449063658714294,
      "learning_rate": 9.75887556447217e-07,
      "loss": 0.0009,
      "step": 149970
    },
    {
      "epoch": 2.4540620142354577,
      "grad_norm": 0.03904491290450096,
      "learning_rate": 9.753226380994368e-07,
      "loss": 0.0005,
      "step": 149980
    },
    {
      "epoch": 2.4542256401865337,
      "grad_norm": 0.020189976319670677,
      "learning_rate": 9.747578656365397e-07,
      "loss": 0.0008,
      "step": 149990
    },
    {
      "epoch": 2.4543892661376097,
      "grad_norm": 0.0536421574652195,
      "learning_rate": 9.741932390789937e-07,
      "loss": 0.0012,
      "step": 150000
    },
    {
      "epoch": 2.4543892661376097,
      "eval_loss": 0.0009711539605632424,
      "eval_runtime": 3.0967,
      "eval_samples_per_second": 64.584,
      "eval_steps_per_second": 16.146,
      "step": 150000
    },
    {
      "epoch": 2.454552892088685,
      "grad_norm": 0.16856323182582855,
      "learning_rate": 9.736287584472681e-07,
      "loss": 0.0017,
      "step": 150010
    },
    {
      "epoch": 2.454716518039761,
      "grad_norm": 0.038104940205812454,
      "learning_rate": 9.73064423761822e-07,
      "loss": 0.0007,
      "step": 150020
    },
    {
      "epoch": 2.4548801439908368,
      "grad_norm": 0.06241224706172943,
      "learning_rate": 9.725002350431135e-07,
      "loss": 0.0008,
      "step": 150030
    },
    {
      "epoch": 2.4550437699419128,
      "grad_norm": 0.057114940136671066,
      "learning_rate": 9.719361923115912e-07,
      "loss": 0.0007,
      "step": 150040
    },
    {
      "epoch": 2.4552073958929888,
      "grad_norm": 0.09591123461723328,
      "learning_rate": 9.713722955876997e-07,
      "loss": 0.0006,
      "step": 150050
    },
    {
      "epoch": 2.4553710218440643,
      "grad_norm": 0.02816822938621044,
      "learning_rate": 9.708085448918814e-07,
      "loss": 0.001,
      "step": 150060
    },
    {
      "epoch": 2.4555346477951403,
      "grad_norm": 0.11524377763271332,
      "learning_rate": 9.702449402445674e-07,
      "loss": 0.0008,
      "step": 150070
    },
    {
      "epoch": 2.4556982737462163,
      "grad_norm": 0.08896174281835556,
      "learning_rate": 9.696814816661908e-07,
      "loss": 0.0019,
      "step": 150080
    },
    {
      "epoch": 2.455861899697292,
      "grad_norm": 0.006541398353874683,
      "learning_rate": 9.691181691771727e-07,
      "loss": 0.0009,
      "step": 150090
    },
    {
      "epoch": 2.456025525648368,
      "grad_norm": 0.067821204662323,
      "learning_rate": 9.685550027979346e-07,
      "loss": 0.0014,
      "step": 150100
    },
    {
      "epoch": 2.456189151599444,
      "grad_norm": 0.08969834446907043,
      "learning_rate": 9.679919825488865e-07,
      "loss": 0.0008,
      "step": 150110
    },
    {
      "epoch": 2.4563527775505194,
      "grad_norm": 0.08168666809797287,
      "learning_rate": 9.6742910845044e-07,
      "loss": 0.0011,
      "step": 150120
    },
    {
      "epoch": 2.4565164035015954,
      "grad_norm": 0.0837826058268547,
      "learning_rate": 9.668663805229944e-07,
      "loss": 0.0009,
      "step": 150130
    },
    {
      "epoch": 2.4566800294526714,
      "grad_norm": 0.04135086387395859,
      "learning_rate": 9.663037987869512e-07,
      "loss": 0.0009,
      "step": 150140
    },
    {
      "epoch": 2.456843655403747,
      "grad_norm": 0.0828266441822052,
      "learning_rate": 9.657413632627e-07,
      "loss": 0.0005,
      "step": 150150
    },
    {
      "epoch": 2.457007281354823,
      "grad_norm": 0.14552417397499084,
      "learning_rate": 9.651790739706274e-07,
      "loss": 0.0011,
      "step": 150160
    },
    {
      "epoch": 2.4571709073058985,
      "grad_norm": 0.021406356245279312,
      "learning_rate": 9.646169309311165e-07,
      "loss": 0.0009,
      "step": 150170
    },
    {
      "epoch": 2.4573345332569745,
      "grad_norm": 0.08791796863079071,
      "learning_rate": 9.640549341645423e-07,
      "loss": 0.0014,
      "step": 150180
    },
    {
      "epoch": 2.4574981592080505,
      "grad_norm": 0.04786737635731697,
      "learning_rate": 9.634930836912765e-07,
      "loss": 0.0008,
      "step": 150190
    },
    {
      "epoch": 2.457661785159126,
      "grad_norm": 0.0012093438999727368,
      "learning_rate": 9.629313795316863e-07,
      "loss": 0.0006,
      "step": 150200
    },
    {
      "epoch": 2.457825411110202,
      "grad_norm": 0.09320314973592758,
      "learning_rate": 9.623698217061294e-07,
      "loss": 0.0007,
      "step": 150210
    },
    {
      "epoch": 2.457989037061278,
      "grad_norm": 0.0720137283205986,
      "learning_rate": 9.61808410234964e-07,
      "loss": 0.0006,
      "step": 150220
    },
    {
      "epoch": 2.4581526630123536,
      "grad_norm": 0.006113241892307997,
      "learning_rate": 9.612471451385363e-07,
      "loss": 0.0009,
      "step": 150230
    },
    {
      "epoch": 2.4583162889634296,
      "grad_norm": 0.04886533319950104,
      "learning_rate": 9.606860264371942e-07,
      "loss": 0.0006,
      "step": 150240
    },
    {
      "epoch": 2.4584799149145056,
      "grad_norm": 0.059853337705135345,
      "learning_rate": 9.601250541512758e-07,
      "loss": 0.0005,
      "step": 150250
    },
    {
      "epoch": 2.458643540865581,
      "grad_norm": 0.053843386471271515,
      "learning_rate": 9.595642283011125e-07,
      "loss": 0.0006,
      "step": 150260
    },
    {
      "epoch": 2.458807166816657,
      "grad_norm": 0.1918380856513977,
      "learning_rate": 9.59003548907037e-07,
      "loss": 0.0018,
      "step": 150270
    },
    {
      "epoch": 2.458970792767733,
      "grad_norm": 0.11621823906898499,
      "learning_rate": 9.584430159893687e-07,
      "loss": 0.0007,
      "step": 150280
    },
    {
      "epoch": 2.4591344187188087,
      "grad_norm": 0.06775251030921936,
      "learning_rate": 9.57882629568429e-07,
      "loss": 0.0009,
      "step": 150290
    },
    {
      "epoch": 2.4592980446698847,
      "grad_norm": 0.06714984774589539,
      "learning_rate": 9.573223896645278e-07,
      "loss": 0.0008,
      "step": 150300
    },
    {
      "epoch": 2.4594616706209607,
      "grad_norm": 0.04675167426466942,
      "learning_rate": 9.56762296297975e-07,
      "loss": 0.0004,
      "step": 150310
    },
    {
      "epoch": 2.459625296572036,
      "grad_norm": 0.07848186790943146,
      "learning_rate": 9.562023494890698e-07,
      "loss": 0.0012,
      "step": 150320
    },
    {
      "epoch": 2.459788922523112,
      "grad_norm": 0.3382268249988556,
      "learning_rate": 9.556425492581118e-07,
      "loss": 0.0023,
      "step": 150330
    },
    {
      "epoch": 2.459952548474188,
      "grad_norm": 0.002287573181092739,
      "learning_rate": 9.55082895625391e-07,
      "loss": 0.0006,
      "step": 150340
    },
    {
      "epoch": 2.4601161744252638,
      "grad_norm": 0.007926524616777897,
      "learning_rate": 9.545233886111926e-07,
      "loss": 0.0007,
      "step": 150350
    },
    {
      "epoch": 2.4602798003763398,
      "grad_norm": 0.06791617721319199,
      "learning_rate": 9.539640282357992e-07,
      "loss": 0.0007,
      "step": 150360
    },
    {
      "epoch": 2.4604434263274158,
      "grad_norm": 0.047069791704416275,
      "learning_rate": 9.534048145194841e-07,
      "loss": 0.001,
      "step": 150370
    },
    {
      "epoch": 2.4606070522784913,
      "grad_norm": 0.007581259123980999,
      "learning_rate": 9.528457474825203e-07,
      "loss": 0.0005,
      "step": 150380
    },
    {
      "epoch": 2.4607706782295673,
      "grad_norm": 0.13201960921287537,
      "learning_rate": 9.522868271451696e-07,
      "loss": 0.0011,
      "step": 150390
    },
    {
      "epoch": 2.4609343041806433,
      "grad_norm": 0.09165289998054504,
      "learning_rate": 9.517280535276929e-07,
      "loss": 0.0013,
      "step": 150400
    },
    {
      "epoch": 2.461097930131719,
      "grad_norm": 0.02847292274236679,
      "learning_rate": 9.511694266503457e-07,
      "loss": 0.0006,
      "step": 150410
    },
    {
      "epoch": 2.461261556082795,
      "grad_norm": 0.08580125868320465,
      "learning_rate": 9.506109465333746e-07,
      "loss": 0.0008,
      "step": 150420
    },
    {
      "epoch": 2.4614251820338704,
      "grad_norm": 0.0063904342241585255,
      "learning_rate": 9.500526131970255e-07,
      "loss": 0.0009,
      "step": 150430
    },
    {
      "epoch": 2.4615888079849464,
      "grad_norm": 0.15242663025856018,
      "learning_rate": 9.494944266615352e-07,
      "loss": 0.0005,
      "step": 150440
    },
    {
      "epoch": 2.4617524339360224,
      "grad_norm": 0.0859651044011116,
      "learning_rate": 9.489363869471369e-07,
      "loss": 0.0009,
      "step": 150450
    },
    {
      "epoch": 2.461916059887098,
      "grad_norm": 0.31555071473121643,
      "learning_rate": 9.483784940740571e-07,
      "loss": 0.0007,
      "step": 150460
    },
    {
      "epoch": 2.462079685838174,
      "grad_norm": 0.06957252323627472,
      "learning_rate": 9.47820748062519e-07,
      "loss": 0.0011,
      "step": 150470
    },
    {
      "epoch": 2.46224331178925,
      "grad_norm": 0.0669672042131424,
      "learning_rate": 9.472631489327406e-07,
      "loss": 0.0007,
      "step": 150480
    },
    {
      "epoch": 2.4624069377403255,
      "grad_norm": 0.04413294047117233,
      "learning_rate": 9.467056967049315e-07,
      "loss": 0.0005,
      "step": 150490
    },
    {
      "epoch": 2.4625705636914015,
      "grad_norm": 0.058431029319763184,
      "learning_rate": 9.461483913993003e-07,
      "loss": 0.0006,
      "step": 150500
    },
    {
      "epoch": 2.4627341896424775,
      "grad_norm": 0.07261022180318832,
      "learning_rate": 9.45591233036045e-07,
      "loss": 0.0014,
      "step": 150510
    },
    {
      "epoch": 2.462897815593553,
      "grad_norm": 0.02955297753214836,
      "learning_rate": 9.450342216353647e-07,
      "loss": 0.0005,
      "step": 150520
    },
    {
      "epoch": 2.463061441544629,
      "grad_norm": 0.05156468600034714,
      "learning_rate": 9.444773572174471e-07,
      "loss": 0.0005,
      "step": 150530
    },
    {
      "epoch": 2.4632250674957046,
      "grad_norm": 0.027956686913967133,
      "learning_rate": 9.439206398024786e-07,
      "loss": 0.0017,
      "step": 150540
    },
    {
      "epoch": 2.4633886934467806,
      "grad_norm": 0.03266816586256027,
      "learning_rate": 9.433640694106389e-07,
      "loss": 0.0011,
      "step": 150550
    },
    {
      "epoch": 2.4635523193978566,
      "grad_norm": 0.06012602522969246,
      "learning_rate": 9.428076460621005e-07,
      "loss": 0.0007,
      "step": 150560
    },
    {
      "epoch": 2.463715945348932,
      "grad_norm": 0.052492085844278336,
      "learning_rate": 9.422513697770353e-07,
      "loss": 0.0005,
      "step": 150570
    },
    {
      "epoch": 2.463879571300008,
      "grad_norm": 0.07562370598316193,
      "learning_rate": 9.416952405756036e-07,
      "loss": 0.0011,
      "step": 150580
    },
    {
      "epoch": 2.464043197251084,
      "grad_norm": 0.1032552719116211,
      "learning_rate": 9.411392584779666e-07,
      "loss": 0.0006,
      "step": 150590
    },
    {
      "epoch": 2.4642068232021597,
      "grad_norm": 0.0015117744915187359,
      "learning_rate": 9.405834235042754e-07,
      "loss": 0.0004,
      "step": 150600
    },
    {
      "epoch": 2.4643704491532357,
      "grad_norm": 0.06402118504047394,
      "learning_rate": 9.400277356746801e-07,
      "loss": 0.0004,
      "step": 150610
    },
    {
      "epoch": 2.4645340751043117,
      "grad_norm": 0.012864670716226101,
      "learning_rate": 9.3947219500932e-07,
      "loss": 0.0005,
      "step": 150620
    },
    {
      "epoch": 2.464697701055387,
      "grad_norm": 0.07661350816488266,
      "learning_rate": 9.389168015283334e-07,
      "loss": 0.0007,
      "step": 150630
    },
    {
      "epoch": 2.464861327006463,
      "grad_norm": 0.09371276944875717,
      "learning_rate": 9.383615552518549e-07,
      "loss": 0.0005,
      "step": 150640
    },
    {
      "epoch": 2.465024952957539,
      "grad_norm": 0.02587028406560421,
      "learning_rate": 9.378064562000055e-07,
      "loss": 0.0009,
      "step": 150650
    },
    {
      "epoch": 2.4651885789086148,
      "grad_norm": 0.04956110194325447,
      "learning_rate": 9.372515043929098e-07,
      "loss": 0.001,
      "step": 150660
    },
    {
      "epoch": 2.4653522048596908,
      "grad_norm": 0.12506696581840515,
      "learning_rate": 9.366966998506816e-07,
      "loss": 0.001,
      "step": 150670
    },
    {
      "epoch": 2.4655158308107668,
      "grad_norm": 0.11542143672704697,
      "learning_rate": 9.361420425934325e-07,
      "loss": 0.0007,
      "step": 150680
    },
    {
      "epoch": 2.4656794567618423,
      "grad_norm": 0.04239243268966675,
      "learning_rate": 9.355875326412678e-07,
      "loss": 0.002,
      "step": 150690
    },
    {
      "epoch": 2.4658430827129183,
      "grad_norm": 0.1125103160738945,
      "learning_rate": 9.350331700142856e-07,
      "loss": 0.0005,
      "step": 150700
    },
    {
      "epoch": 2.4660067086639943,
      "grad_norm": 0.011683939956128597,
      "learning_rate": 9.344789547325822e-07,
      "loss": 0.0009,
      "step": 150710
    },
    {
      "epoch": 2.46617033461507,
      "grad_norm": 0.10891806334257126,
      "learning_rate": 9.339248868162443e-07,
      "loss": 0.0008,
      "step": 150720
    },
    {
      "epoch": 2.466333960566146,
      "grad_norm": 0.04900682345032692,
      "learning_rate": 9.333709662853579e-07,
      "loss": 0.0004,
      "step": 150730
    },
    {
      "epoch": 2.466497586517222,
      "grad_norm": 0.020068299025297165,
      "learning_rate": 9.3281719316e-07,
      "loss": 0.0009,
      "step": 150740
    },
    {
      "epoch": 2.4666612124682974,
      "grad_norm": 0.00489389430731535,
      "learning_rate": 9.322635674602421e-07,
      "loss": 0.0004,
      "step": 150750
    },
    {
      "epoch": 2.4668248384193734,
      "grad_norm": 0.04533260688185692,
      "learning_rate": 9.317100892061548e-07,
      "loss": 0.0012,
      "step": 150760
    },
    {
      "epoch": 2.4669884643704494,
      "grad_norm": 0.0490998700261116,
      "learning_rate": 9.311567584177978e-07,
      "loss": 0.0006,
      "step": 150770
    },
    {
      "epoch": 2.467152090321525,
      "grad_norm": 0.08508940041065216,
      "learning_rate": 9.306035751152303e-07,
      "loss": 0.0005,
      "step": 150780
    },
    {
      "epoch": 2.467315716272601,
      "grad_norm": 0.1055036187171936,
      "learning_rate": 9.300505393185017e-07,
      "loss": 0.0009,
      "step": 150790
    },
    {
      "epoch": 2.4674793422236765,
      "grad_norm": 0.009024235419929028,
      "learning_rate": 9.2949765104766e-07,
      "loss": 0.0002,
      "step": 150800
    },
    {
      "epoch": 2.4676429681747525,
      "grad_norm": 0.029518017545342445,
      "learning_rate": 9.289449103227444e-07,
      "loss": 0.0005,
      "step": 150810
    },
    {
      "epoch": 2.4678065941258285,
      "grad_norm": 0.00842087808996439,
      "learning_rate": 9.283923171637921e-07,
      "loss": 0.0004,
      "step": 150820
    },
    {
      "epoch": 2.467970220076904,
      "grad_norm": 0.1646094024181366,
      "learning_rate": 9.278398715908316e-07,
      "loss": 0.0009,
      "step": 150830
    },
    {
      "epoch": 2.46813384602798,
      "grad_norm": 0.061742912977933884,
      "learning_rate": 9.272875736238896e-07,
      "loss": 0.0007,
      "step": 150840
    },
    {
      "epoch": 2.468297471979056,
      "grad_norm": 0.07695599645376205,
      "learning_rate": 9.267354232829845e-07,
      "loss": 0.0009,
      "step": 150850
    },
    {
      "epoch": 2.4684610979301316,
      "grad_norm": 0.032638199627399445,
      "learning_rate": 9.261834205881293e-07,
      "loss": 0.0007,
      "step": 150860
    },
    {
      "epoch": 2.4686247238812076,
      "grad_norm": 0.056871455162763596,
      "learning_rate": 9.256315655593351e-07,
      "loss": 0.0011,
      "step": 150870
    },
    {
      "epoch": 2.4687883498322836,
      "grad_norm": 0.002389301545917988,
      "learning_rate": 9.250798582166027e-07,
      "loss": 0.0005,
      "step": 150880
    },
    {
      "epoch": 2.468951975783359,
      "grad_norm": 0.059549376368522644,
      "learning_rate": 9.245282985799331e-07,
      "loss": 0.0008,
      "step": 150890
    },
    {
      "epoch": 2.469115601734435,
      "grad_norm": 0.11891381442546844,
      "learning_rate": 9.239768866693166e-07,
      "loss": 0.0004,
      "step": 150900
    },
    {
      "epoch": 2.469279227685511,
      "grad_norm": 0.05941632017493248,
      "learning_rate": 9.234256225047411e-07,
      "loss": 0.0006,
      "step": 150910
    },
    {
      "epoch": 2.4694428536365867,
      "grad_norm": 0.027064897119998932,
      "learning_rate": 9.228745061061906e-07,
      "loss": 0.0003,
      "step": 150920
    },
    {
      "epoch": 2.4696064795876627,
      "grad_norm": 0.13816119730472565,
      "learning_rate": 9.223235374936385e-07,
      "loss": 0.0008,
      "step": 150930
    },
    {
      "epoch": 2.469770105538738,
      "grad_norm": 0.2471037358045578,
      "learning_rate": 9.217727166870604e-07,
      "loss": 0.0011,
      "step": 150940
    },
    {
      "epoch": 2.469933731489814,
      "grad_norm": 0.04939854517579079,
      "learning_rate": 9.212220437064173e-07,
      "loss": 0.0006,
      "step": 150950
    },
    {
      "epoch": 2.47009735744089,
      "grad_norm": 0.0747624933719635,
      "learning_rate": 9.206715185716714e-07,
      "loss": 0.0005,
      "step": 150960
    },
    {
      "epoch": 2.4702609833919658,
      "grad_norm": 0.43480560183525085,
      "learning_rate": 9.201211413027805e-07,
      "loss": 0.0019,
      "step": 150970
    },
    {
      "epoch": 2.4704246093430418,
      "grad_norm": 0.051545508205890656,
      "learning_rate": 9.195709119196911e-07,
      "loss": 0.0006,
      "step": 150980
    },
    {
      "epoch": 2.4705882352941178,
      "grad_norm": 0.07411473989486694,
      "learning_rate": 9.190208304423498e-07,
      "loss": 0.0005,
      "step": 150990
    },
    {
      "epoch": 2.4707518612451933,
      "grad_norm": 0.19455954432487488,
      "learning_rate": 9.184708968906942e-07,
      "loss": 0.0012,
      "step": 151000
    },
    {
      "epoch": 2.4709154871962693,
      "grad_norm": 0.0461120642721653,
      "learning_rate": 9.179211112846603e-07,
      "loss": 0.0006,
      "step": 151010
    },
    {
      "epoch": 2.4710791131473453,
      "grad_norm": 0.07214047759771347,
      "learning_rate": 9.173714736441736e-07,
      "loss": 0.0004,
      "step": 151020
    },
    {
      "epoch": 2.471242739098421,
      "grad_norm": 0.03791683539748192,
      "learning_rate": 9.168219839891595e-07,
      "loss": 0.0006,
      "step": 151030
    },
    {
      "epoch": 2.471406365049497,
      "grad_norm": 0.03463110700249672,
      "learning_rate": 9.162726423395357e-07,
      "loss": 0.0006,
      "step": 151040
    },
    {
      "epoch": 2.471569991000573,
      "grad_norm": 0.0014401538064703345,
      "learning_rate": 9.157234487152117e-07,
      "loss": 0.0006,
      "step": 151050
    },
    {
      "epoch": 2.4717336169516484,
      "grad_norm": 0.027242766693234444,
      "learning_rate": 9.151744031360982e-07,
      "loss": 0.0009,
      "step": 151060
    },
    {
      "epoch": 2.4718972429027244,
      "grad_norm": 0.03567284718155861,
      "learning_rate": 9.146255056220932e-07,
      "loss": 0.0005,
      "step": 151070
    },
    {
      "epoch": 2.4720608688538004,
      "grad_norm": 0.06301916390657425,
      "learning_rate": 9.140767561930958e-07,
      "loss": 0.001,
      "step": 151080
    },
    {
      "epoch": 2.472224494804876,
      "grad_norm": 0.028930868953466415,
      "learning_rate": 9.13528154868995e-07,
      "loss": 0.0006,
      "step": 151090
    },
    {
      "epoch": 2.472388120755952,
      "grad_norm": 0.05450453981757164,
      "learning_rate": 9.129797016696778e-07,
      "loss": 0.0005,
      "step": 151100
    },
    {
      "epoch": 2.472551746707028,
      "grad_norm": 0.1395966112613678,
      "learning_rate": 9.124313966150228e-07,
      "loss": 0.0013,
      "step": 151110
    },
    {
      "epoch": 2.4727153726581035,
      "grad_norm": 0.04936983436346054,
      "learning_rate": 9.11883239724905e-07,
      "loss": 0.0008,
      "step": 151120
    },
    {
      "epoch": 2.4728789986091795,
      "grad_norm": 0.026668457314372063,
      "learning_rate": 9.113352310191958e-07,
      "loss": 0.0018,
      "step": 151130
    },
    {
      "epoch": 2.4730426245602555,
      "grad_norm": 0.03131609410047531,
      "learning_rate": 9.107873705177572e-07,
      "loss": 0.0006,
      "step": 151140
    },
    {
      "epoch": 2.473206250511331,
      "grad_norm": 0.04698535427451134,
      "learning_rate": 9.102396582404488e-07,
      "loss": 0.0008,
      "step": 151150
    },
    {
      "epoch": 2.473369876462407,
      "grad_norm": 0.07373373955488205,
      "learning_rate": 9.096920942071219e-07,
      "loss": 0.0006,
      "step": 151160
    },
    {
      "epoch": 2.473533502413483,
      "grad_norm": 0.035534121096134186,
      "learning_rate": 9.091446784376257e-07,
      "loss": 0.0005,
      "step": 151170
    },
    {
      "epoch": 2.4736971283645586,
      "grad_norm": 0.12654584646224976,
      "learning_rate": 9.085974109518043e-07,
      "loss": 0.0006,
      "step": 151180
    },
    {
      "epoch": 2.4738607543156346,
      "grad_norm": 0.11844980716705322,
      "learning_rate": 9.080502917694917e-07,
      "loss": 0.0011,
      "step": 151190
    },
    {
      "epoch": 2.47402438026671,
      "grad_norm": 0.03693728148937225,
      "learning_rate": 9.075033209105222e-07,
      "loss": 0.0004,
      "step": 151200
    },
    {
      "epoch": 2.474188006217786,
      "grad_norm": 0.058446429669857025,
      "learning_rate": 9.069564983947204e-07,
      "loss": 0.0006,
      "step": 151210
    },
    {
      "epoch": 2.474351632168862,
      "grad_norm": 0.09844362735748291,
      "learning_rate": 9.064098242419095e-07,
      "loss": 0.0005,
      "step": 151220
    },
    {
      "epoch": 2.4745152581199377,
      "grad_norm": 0.06673082709312439,
      "learning_rate": 9.05863298471903e-07,
      "loss": 0.0008,
      "step": 151230
    },
    {
      "epoch": 2.4746788840710137,
      "grad_norm": 0.06816031038761139,
      "learning_rate": 9.053169211045127e-07,
      "loss": 0.0005,
      "step": 151240
    },
    {
      "epoch": 2.4748425100220897,
      "grad_norm": 0.004089050460606813,
      "learning_rate": 9.047706921595422e-07,
      "loss": 0.0004,
      "step": 151250
    },
    {
      "epoch": 2.475006135973165,
      "grad_norm": 0.13913549482822418,
      "learning_rate": 9.04224611656791e-07,
      "loss": 0.0008,
      "step": 151260
    },
    {
      "epoch": 2.475169761924241,
      "grad_norm": 0.15670907497406006,
      "learning_rate": 9.036786796160546e-07,
      "loss": 0.0004,
      "step": 151270
    },
    {
      "epoch": 2.475333387875317,
      "grad_norm": 0.00462714210152626,
      "learning_rate": 9.031328960571195e-07,
      "loss": 0.0007,
      "step": 151280
    },
    {
      "epoch": 2.4754970138263928,
      "grad_norm": 0.11180935800075531,
      "learning_rate": 9.025872609997715e-07,
      "loss": 0.0011,
      "step": 151290
    },
    {
      "epoch": 2.4756606397774688,
      "grad_norm": 0.13851602375507355,
      "learning_rate": 9.020417744637866e-07,
      "loss": 0.0006,
      "step": 151300
    },
    {
      "epoch": 2.4758242657285443,
      "grad_norm": 0.24512170255184174,
      "learning_rate": 9.014964364689393e-07,
      "loss": 0.0013,
      "step": 151310
    },
    {
      "epoch": 2.4759878916796203,
      "grad_norm": 0.009779210202395916,
      "learning_rate": 9.009512470349946e-07,
      "loss": 0.0008,
      "step": 151320
    },
    {
      "epoch": 2.4761515176306963,
      "grad_norm": 0.007476090453565121,
      "learning_rate": 9.004062061817148e-07,
      "loss": 0.0008,
      "step": 151330
    },
    {
      "epoch": 2.476315143581772,
      "grad_norm": 0.011915738694369793,
      "learning_rate": 8.998613139288597e-07,
      "loss": 0.001,
      "step": 151340
    },
    {
      "epoch": 2.476478769532848,
      "grad_norm": 0.13148239254951477,
      "learning_rate": 8.993165702961753e-07,
      "loss": 0.0011,
      "step": 151350
    },
    {
      "epoch": 2.476642395483924,
      "grad_norm": 0.006391483824700117,
      "learning_rate": 8.987719753034108e-07,
      "loss": 0.0007,
      "step": 151360
    },
    {
      "epoch": 2.4768060214349994,
      "grad_norm": 0.14566324651241302,
      "learning_rate": 8.982275289703035e-07,
      "loss": 0.0007,
      "step": 151370
    },
    {
      "epoch": 2.4769696473860754,
      "grad_norm": 0.039603304117918015,
      "learning_rate": 8.976832313165912e-07,
      "loss": 0.0005,
      "step": 151380
    },
    {
      "epoch": 2.4771332733371514,
      "grad_norm": 0.008537969551980495,
      "learning_rate": 8.971390823620013e-07,
      "loss": 0.0008,
      "step": 151390
    },
    {
      "epoch": 2.477296899288227,
      "grad_norm": 0.1054685190320015,
      "learning_rate": 8.965950821262582e-07,
      "loss": 0.0011,
      "step": 151400
    },
    {
      "epoch": 2.477460525239303,
      "grad_norm": 0.023423688486218452,
      "learning_rate": 8.960512306290825e-07,
      "loss": 0.0003,
      "step": 151410
    },
    {
      "epoch": 2.477624151190379,
      "grad_norm": 0.10136052966117859,
      "learning_rate": 8.955075278901848e-07,
      "loss": 0.001,
      "step": 151420
    },
    {
      "epoch": 2.4777877771414545,
      "grad_norm": 0.14078889787197113,
      "learning_rate": 8.949639739292759e-07,
      "loss": 0.0011,
      "step": 151430
    },
    {
      "epoch": 2.4779514030925305,
      "grad_norm": 0.03205703943967819,
      "learning_rate": 8.944205687660562e-07,
      "loss": 0.0006,
      "step": 151440
    },
    {
      "epoch": 2.4781150290436065,
      "grad_norm": 0.20053310692310333,
      "learning_rate": 8.938773124202227e-07,
      "loss": 0.0007,
      "step": 151450
    },
    {
      "epoch": 2.478278654994682,
      "grad_norm": 0.05311646685004234,
      "learning_rate": 8.93334204911469e-07,
      "loss": 0.0006,
      "step": 151460
    },
    {
      "epoch": 2.478442280945758,
      "grad_norm": 0.025086546316742897,
      "learning_rate": 8.927912462594784e-07,
      "loss": 0.0007,
      "step": 151470
    },
    {
      "epoch": 2.478605906896834,
      "grad_norm": 0.1363171488046646,
      "learning_rate": 8.922484364839357e-07,
      "loss": 0.0007,
      "step": 151480
    },
    {
      "epoch": 2.4787695328479096,
      "grad_norm": 0.1924663633108139,
      "learning_rate": 8.917057756045128e-07,
      "loss": 0.0009,
      "step": 151490
    },
    {
      "epoch": 2.4789331587989856,
      "grad_norm": 0.06424161046743393,
      "learning_rate": 8.911632636408829e-07,
      "loss": 0.0016,
      "step": 151500
    },
    {
      "epoch": 2.4790967847500616,
      "grad_norm": 0.04472389444708824,
      "learning_rate": 8.906209006127082e-07,
      "loss": 0.001,
      "step": 151510
    },
    {
      "epoch": 2.479260410701137,
      "grad_norm": 0.032719388604164124,
      "learning_rate": 8.900786865396504e-07,
      "loss": 0.0002,
      "step": 151520
    },
    {
      "epoch": 2.479424036652213,
      "grad_norm": 0.04274709150195122,
      "learning_rate": 8.895366214413609e-07,
      "loss": 0.0016,
      "step": 151530
    },
    {
      "epoch": 2.479587662603289,
      "grad_norm": 0.11197847872972488,
      "learning_rate": 8.889947053374909e-07,
      "loss": 0.0013,
      "step": 151540
    },
    {
      "epoch": 2.4797512885543647,
      "grad_norm": 0.013463164679706097,
      "learning_rate": 8.884529382476825e-07,
      "loss": 0.0007,
      "step": 151550
    },
    {
      "epoch": 2.4799149145054407,
      "grad_norm": 0.14517927169799805,
      "learning_rate": 8.879113201915717e-07,
      "loss": 0.0013,
      "step": 151560
    },
    {
      "epoch": 2.480078540456516,
      "grad_norm": 0.10182119160890579,
      "learning_rate": 8.873698511887935e-07,
      "loss": 0.0011,
      "step": 151570
    },
    {
      "epoch": 2.480242166407592,
      "grad_norm": 0.21422433853149414,
      "learning_rate": 8.868285312589731e-07,
      "loss": 0.0004,
      "step": 151580
    },
    {
      "epoch": 2.480405792358668,
      "grad_norm": 0.04126915708184242,
      "learning_rate": 8.862873604217337e-07,
      "loss": 0.0007,
      "step": 151590
    },
    {
      "epoch": 2.4805694183097438,
      "grad_norm": 0.0857510194182396,
      "learning_rate": 8.857463386966892e-07,
      "loss": 0.0007,
      "step": 151600
    },
    {
      "epoch": 2.4807330442608198,
      "grad_norm": 0.013697189278900623,
      "learning_rate": 8.852054661034509e-07,
      "loss": 0.0009,
      "step": 151610
    },
    {
      "epoch": 2.4808966702118957,
      "grad_norm": 0.07937291264533997,
      "learning_rate": 8.846647426616267e-07,
      "loss": 0.001,
      "step": 151620
    },
    {
      "epoch": 2.4810602961629713,
      "grad_norm": 0.08973510563373566,
      "learning_rate": 8.841241683908136e-07,
      "loss": 0.0009,
      "step": 151630
    },
    {
      "epoch": 2.4812239221140473,
      "grad_norm": 0.1182357668876648,
      "learning_rate": 8.835837433106092e-07,
      "loss": 0.0009,
      "step": 151640
    },
    {
      "epoch": 2.4813875480651233,
      "grad_norm": 0.08104902505874634,
      "learning_rate": 8.830434674405986e-07,
      "loss": 0.001,
      "step": 151650
    },
    {
      "epoch": 2.481551174016199,
      "grad_norm": 0.04527677595615387,
      "learning_rate": 8.825033408003681e-07,
      "loss": 0.0007,
      "step": 151660
    },
    {
      "epoch": 2.481714799967275,
      "grad_norm": 0.12170953303575516,
      "learning_rate": 8.819633634094943e-07,
      "loss": 0.0012,
      "step": 151670
    },
    {
      "epoch": 2.481878425918351,
      "grad_norm": 0.06582707166671753,
      "learning_rate": 8.814235352875511e-07,
      "loss": 0.0011,
      "step": 151680
    },
    {
      "epoch": 2.4820420518694264,
      "grad_norm": 0.038412194699048996,
      "learning_rate": 8.808838564541073e-07,
      "loss": 0.0009,
      "step": 151690
    },
    {
      "epoch": 2.4822056778205024,
      "grad_norm": 0.48581215739250183,
      "learning_rate": 8.803443269287226e-07,
      "loss": 0.0008,
      "step": 151700
    },
    {
      "epoch": 2.482369303771578,
      "grad_norm": 0.07216094434261322,
      "learning_rate": 8.798049467309561e-07,
      "loss": 0.0011,
      "step": 151710
    },
    {
      "epoch": 2.482532929722654,
      "grad_norm": 0.12745238840579987,
      "learning_rate": 8.792657158803558e-07,
      "loss": 0.0008,
      "step": 151720
    },
    {
      "epoch": 2.48269655567373,
      "grad_norm": 0.031186701729893684,
      "learning_rate": 8.787266343964712e-07,
      "loss": 0.0008,
      "step": 151730
    },
    {
      "epoch": 2.4828601816248055,
      "grad_norm": 0.1181822195649147,
      "learning_rate": 8.781877022988405e-07,
      "loss": 0.0006,
      "step": 151740
    },
    {
      "epoch": 2.4830238075758815,
      "grad_norm": 0.3319505453109741,
      "learning_rate": 8.776489196069982e-07,
      "loss": 0.0008,
      "step": 151750
    },
    {
      "epoch": 2.4831874335269575,
      "grad_norm": 0.39207926392555237,
      "learning_rate": 8.771102863404757e-07,
      "loss": 0.001,
      "step": 151760
    },
    {
      "epoch": 2.483351059478033,
      "grad_norm": 0.04138851538300514,
      "learning_rate": 8.765718025187946e-07,
      "loss": 0.0015,
      "step": 151770
    },
    {
      "epoch": 2.483514685429109,
      "grad_norm": 0.04688561335206032,
      "learning_rate": 8.760334681614774e-07,
      "loss": 0.001,
      "step": 151780
    },
    {
      "epoch": 2.483678311380185,
      "grad_norm": 0.048823513090610504,
      "learning_rate": 8.754952832880331e-07,
      "loss": 0.0009,
      "step": 151790
    },
    {
      "epoch": 2.4838419373312606,
      "grad_norm": 0.034230902791023254,
      "learning_rate": 8.749572479179736e-07,
      "loss": 0.0003,
      "step": 151800
    },
    {
      "epoch": 2.4840055632823366,
      "grad_norm": 0.0801633670926094,
      "learning_rate": 8.744193620707986e-07,
      "loss": 0.0017,
      "step": 151810
    },
    {
      "epoch": 2.4841691892334126,
      "grad_norm": 0.09538417309522629,
      "learning_rate": 8.738816257660071e-07,
      "loss": 0.001,
      "step": 151820
    },
    {
      "epoch": 2.484332815184488,
      "grad_norm": 0.1383579671382904,
      "learning_rate": 8.733440390230885e-07,
      "loss": 0.0004,
      "step": 151830
    },
    {
      "epoch": 2.484496441135564,
      "grad_norm": 0.06353263556957245,
      "learning_rate": 8.7280660186153e-07,
      "loss": 0.0006,
      "step": 151840
    },
    {
      "epoch": 2.48466006708664,
      "grad_norm": 0.007243297062814236,
      "learning_rate": 8.72269314300816e-07,
      "loss": 0.0004,
      "step": 151850
    },
    {
      "epoch": 2.4848236930377157,
      "grad_norm": 0.0869915783405304,
      "learning_rate": 8.717321763604153e-07,
      "loss": 0.0007,
      "step": 151860
    },
    {
      "epoch": 2.4849873189887917,
      "grad_norm": 0.016138210892677307,
      "learning_rate": 8.71195188059803e-07,
      "loss": 0.0006,
      "step": 151870
    },
    {
      "epoch": 2.4851509449398677,
      "grad_norm": 0.03762826323509216,
      "learning_rate": 8.706583494184401e-07,
      "loss": 0.0012,
      "step": 151880
    },
    {
      "epoch": 2.485314570890943,
      "grad_norm": 0.0773472711443901,
      "learning_rate": 8.701216604557877e-07,
      "loss": 0.0006,
      "step": 151890
    },
    {
      "epoch": 2.485478196842019,
      "grad_norm": 0.03130606561899185,
      "learning_rate": 8.695851211913009e-07,
      "loss": 0.0007,
      "step": 151900
    },
    {
      "epoch": 2.485641822793095,
      "grad_norm": 0.05759549140930176,
      "learning_rate": 8.690487316444241e-07,
      "loss": 0.0008,
      "step": 151910
    },
    {
      "epoch": 2.4858054487441708,
      "grad_norm": 0.05057528242468834,
      "learning_rate": 8.685124918346043e-07,
      "loss": 0.001,
      "step": 151920
    },
    {
      "epoch": 2.4859690746952467,
      "grad_norm": 0.03764611855149269,
      "learning_rate": 8.679764017812753e-07,
      "loss": 0.0016,
      "step": 151930
    },
    {
      "epoch": 2.4861327006463227,
      "grad_norm": 0.5055283308029175,
      "learning_rate": 8.674404615038717e-07,
      "loss": 0.0008,
      "step": 151940
    },
    {
      "epoch": 2.4862963265973983,
      "grad_norm": 0.19533708691596985,
      "learning_rate": 8.669046710218199e-07,
      "loss": 0.0007,
      "step": 151950
    },
    {
      "epoch": 2.4864599525484743,
      "grad_norm": 0.07553951442241669,
      "learning_rate": 8.66369030354538e-07,
      "loss": 0.0006,
      "step": 151960
    },
    {
      "epoch": 2.48662357849955,
      "grad_norm": 0.11274247616529465,
      "learning_rate": 8.658335395214457e-07,
      "loss": 0.0006,
      "step": 151970
    },
    {
      "epoch": 2.486787204450626,
      "grad_norm": 0.0959426611661911,
      "learning_rate": 8.652981985419495e-07,
      "loss": 0.001,
      "step": 151980
    },
    {
      "epoch": 2.486950830401702,
      "grad_norm": 0.07797201722860336,
      "learning_rate": 8.64763007435458e-07,
      "loss": 0.0006,
      "step": 151990
    },
    {
      "epoch": 2.4871144563527774,
      "grad_norm": 0.01973983645439148,
      "learning_rate": 8.64227966221367e-07,
      "loss": 0.0006,
      "step": 152000
    },
    {
      "epoch": 2.4872780823038534,
      "grad_norm": 0.13674390316009521,
      "learning_rate": 8.636930749190736e-07,
      "loss": 0.0011,
      "step": 152010
    },
    {
      "epoch": 2.4874417082549294,
      "grad_norm": 0.05802439898252487,
      "learning_rate": 8.631583335479632e-07,
      "loss": 0.0005,
      "step": 152020
    },
    {
      "epoch": 2.487605334206005,
      "grad_norm": 0.125584214925766,
      "learning_rate": 8.626237421274225e-07,
      "loss": 0.001,
      "step": 152030
    },
    {
      "epoch": 2.487768960157081,
      "grad_norm": 0.18783192336559296,
      "learning_rate": 8.620893006768255e-07,
      "loss": 0.0008,
      "step": 152040
    },
    {
      "epoch": 2.487932586108157,
      "grad_norm": 0.050240159034729004,
      "learning_rate": 8.615550092155478e-07,
      "loss": 0.0007,
      "step": 152050
    },
    {
      "epoch": 2.4880962120592325,
      "grad_norm": 0.023553142324090004,
      "learning_rate": 8.610208677629539e-07,
      "loss": 0.0003,
      "step": 152060
    },
    {
      "epoch": 2.4882598380103085,
      "grad_norm": 0.08275075256824493,
      "learning_rate": 8.60486876338405e-07,
      "loss": 0.0011,
      "step": 152070
    },
    {
      "epoch": 2.488423463961384,
      "grad_norm": 0.09971792250871658,
      "learning_rate": 8.59953034961259e-07,
      "loss": 0.0006,
      "step": 152080
    },
    {
      "epoch": 2.48858708991246,
      "grad_norm": 0.08894790709018707,
      "learning_rate": 8.594193436508641e-07,
      "loss": 0.001,
      "step": 152090
    },
    {
      "epoch": 2.488750715863536,
      "grad_norm": 0.32146868109703064,
      "learning_rate": 8.588858024265673e-07,
      "loss": 0.0008,
      "step": 152100
    },
    {
      "epoch": 2.4889143418146116,
      "grad_norm": 0.07286940515041351,
      "learning_rate": 8.583524113077058e-07,
      "loss": 0.0014,
      "step": 152110
    },
    {
      "epoch": 2.4890779677656876,
      "grad_norm": 0.0459439717233181,
      "learning_rate": 8.578191703136157e-07,
      "loss": 0.0008,
      "step": 152120
    },
    {
      "epoch": 2.4892415937167636,
      "grad_norm": 0.048590224236249924,
      "learning_rate": 8.572860794636262e-07,
      "loss": 0.0013,
      "step": 152130
    },
    {
      "epoch": 2.489405219667839,
      "grad_norm": 0.05707553029060364,
      "learning_rate": 8.567531387770583e-07,
      "loss": 0.0009,
      "step": 152140
    },
    {
      "epoch": 2.489568845618915,
      "grad_norm": 0.08448565006256104,
      "learning_rate": 8.562203482732334e-07,
      "loss": 0.0007,
      "step": 152150
    },
    {
      "epoch": 2.489732471569991,
      "grad_norm": 0.06321943551301956,
      "learning_rate": 8.556877079714592e-07,
      "loss": 0.0007,
      "step": 152160
    },
    {
      "epoch": 2.4898960975210667,
      "grad_norm": 0.0648920014500618,
      "learning_rate": 8.551552178910455e-07,
      "loss": 0.0007,
      "step": 152170
    },
    {
      "epoch": 2.4900597234721427,
      "grad_norm": 0.05136335268616676,
      "learning_rate": 8.546228780512938e-07,
      "loss": 0.0009,
      "step": 152180
    },
    {
      "epoch": 2.4902233494232187,
      "grad_norm": 0.06561985611915588,
      "learning_rate": 8.540906884714989e-07,
      "loss": 0.0008,
      "step": 152190
    },
    {
      "epoch": 2.490386975374294,
      "grad_norm": 0.044398918747901917,
      "learning_rate": 8.53558649170953e-07,
      "loss": 0.0007,
      "step": 152200
    },
    {
      "epoch": 2.49055060132537,
      "grad_norm": 0.07872509956359863,
      "learning_rate": 8.53026760168939e-07,
      "loss": 0.0007,
      "step": 152210
    },
    {
      "epoch": 2.490714227276446,
      "grad_norm": 0.03529883921146393,
      "learning_rate": 8.524950214847399e-07,
      "loss": 0.0008,
      "step": 152220
    },
    {
      "epoch": 2.4908778532275218,
      "grad_norm": 0.27992355823516846,
      "learning_rate": 8.519634331376264e-07,
      "loss": 0.0012,
      "step": 152230
    },
    {
      "epoch": 2.4910414791785978,
      "grad_norm": 0.008263720199465752,
      "learning_rate": 8.514319951468697e-07,
      "loss": 0.0003,
      "step": 152240
    },
    {
      "epoch": 2.4912051051296737,
      "grad_norm": 0.1407913714647293,
      "learning_rate": 8.509007075317327e-07,
      "loss": 0.0004,
      "step": 152250
    },
    {
      "epoch": 2.4913687310807493,
      "grad_norm": 0.044623058289289474,
      "learning_rate": 8.503695703114717e-07,
      "loss": 0.0009,
      "step": 152260
    },
    {
      "epoch": 2.4915323570318253,
      "grad_norm": 0.04765020310878754,
      "learning_rate": 8.498385835053419e-07,
      "loss": 0.0009,
      "step": 152270
    },
    {
      "epoch": 2.4916959829829013,
      "grad_norm": 0.15740187466144562,
      "learning_rate": 8.493077471325872e-07,
      "loss": 0.001,
      "step": 152280
    },
    {
      "epoch": 2.491859608933977,
      "grad_norm": 0.08389198035001755,
      "learning_rate": 8.487770612124524e-07,
      "loss": 0.0005,
      "step": 152290
    },
    {
      "epoch": 2.492023234885053,
      "grad_norm": 0.06761686503887177,
      "learning_rate": 8.482465257641703e-07,
      "loss": 0.0009,
      "step": 152300
    },
    {
      "epoch": 2.492186860836129,
      "grad_norm": 0.03844596445560455,
      "learning_rate": 8.477161408069745e-07,
      "loss": 0.0006,
      "step": 152310
    },
    {
      "epoch": 2.4923504867872044,
      "grad_norm": 0.04087105765938759,
      "learning_rate": 8.471859063600874e-07,
      "loss": 0.0008,
      "step": 152320
    },
    {
      "epoch": 2.4925141127382804,
      "grad_norm": 0.11108768731355667,
      "learning_rate": 8.466558224427302e-07,
      "loss": 0.0006,
      "step": 152330
    },
    {
      "epoch": 2.4926777386893564,
      "grad_norm": 0.1178639605641365,
      "learning_rate": 8.461258890741186e-07,
      "loss": 0.0006,
      "step": 152340
    },
    {
      "epoch": 2.492841364640432,
      "grad_norm": 0.12647679448127747,
      "learning_rate": 8.455961062734602e-07,
      "loss": 0.0012,
      "step": 152350
    },
    {
      "epoch": 2.493004990591508,
      "grad_norm": 0.1367761641740799,
      "learning_rate": 8.450664740599579e-07,
      "loss": 0.0009,
      "step": 152360
    },
    {
      "epoch": 2.4931686165425835,
      "grad_norm": 0.03589765727519989,
      "learning_rate": 8.44536992452808e-07,
      "loss": 0.0004,
      "step": 152370
    },
    {
      "epoch": 2.4933322424936595,
      "grad_norm": 0.017452800646424294,
      "learning_rate": 8.440076614712056e-07,
      "loss": 0.0005,
      "step": 152380
    },
    {
      "epoch": 2.4934958684447355,
      "grad_norm": 0.09697717428207397,
      "learning_rate": 8.434784811343371e-07,
      "loss": 0.0007,
      "step": 152390
    },
    {
      "epoch": 2.493659494395811,
      "grad_norm": 0.04965473338961601,
      "learning_rate": 8.429494514613834e-07,
      "loss": 0.0005,
      "step": 152400
    },
    {
      "epoch": 2.493823120346887,
      "grad_norm": 0.056667428463697433,
      "learning_rate": 8.424205724715218e-07,
      "loss": 0.0007,
      "step": 152410
    },
    {
      "epoch": 2.493986746297963,
      "grad_norm": 0.006122634280472994,
      "learning_rate": 8.418918441839202e-07,
      "loss": 0.0008,
      "step": 152420
    },
    {
      "epoch": 2.4941503722490386,
      "grad_norm": 0.11279958486557007,
      "learning_rate": 8.413632666177473e-07,
      "loss": 0.0006,
      "step": 152430
    },
    {
      "epoch": 2.4943139982001146,
      "grad_norm": 0.03761591762304306,
      "learning_rate": 8.408348397921601e-07,
      "loss": 0.001,
      "step": 152440
    },
    {
      "epoch": 2.4944776241511906,
      "grad_norm": 0.05883323401212692,
      "learning_rate": 8.403065637263141e-07,
      "loss": 0.001,
      "step": 152450
    },
    {
      "epoch": 2.494641250102266,
      "grad_norm": 0.03394457697868347,
      "learning_rate": 8.397784384393581e-07,
      "loss": 0.0006,
      "step": 152460
    },
    {
      "epoch": 2.494804876053342,
      "grad_norm": 0.08883547782897949,
      "learning_rate": 8.392504639504334e-07,
      "loss": 0.0016,
      "step": 152470
    },
    {
      "epoch": 2.4949685020044177,
      "grad_norm": 0.003349594073370099,
      "learning_rate": 8.387226402786808e-07,
      "loss": 0.0008,
      "step": 152480
    },
    {
      "epoch": 2.4951321279554937,
      "grad_norm": 0.06985002756118774,
      "learning_rate": 8.381949674432294e-07,
      "loss": 0.0009,
      "step": 152490
    },
    {
      "epoch": 2.4952957539065697,
      "grad_norm": 0.019473988562822342,
      "learning_rate": 8.376674454632094e-07,
      "loss": 0.0004,
      "step": 152500
    },
    {
      "epoch": 2.495459379857645,
      "grad_norm": 0.0562312975525856,
      "learning_rate": 8.371400743577391e-07,
      "loss": 0.0008,
      "step": 152510
    },
    {
      "epoch": 2.495623005808721,
      "grad_norm": 0.015812261030077934,
      "learning_rate": 8.366128541459373e-07,
      "loss": 0.0005,
      "step": 152520
    },
    {
      "epoch": 2.495786631759797,
      "grad_norm": 0.1904226541519165,
      "learning_rate": 8.360857848469123e-07,
      "loss": 0.0009,
      "step": 152530
    },
    {
      "epoch": 2.4959502577108728,
      "grad_norm": 0.020058510825037956,
      "learning_rate": 8.355588664797693e-07,
      "loss": 0.0004,
      "step": 152540
    },
    {
      "epoch": 2.4961138836619488,
      "grad_norm": 0.14645639061927795,
      "learning_rate": 8.350320990636112e-07,
      "loss": 0.0008,
      "step": 152550
    },
    {
      "epoch": 2.4962775096130247,
      "grad_norm": 0.2425813525915146,
      "learning_rate": 8.345054826175264e-07,
      "loss": 0.0011,
      "step": 152560
    },
    {
      "epoch": 2.4964411355641003,
      "grad_norm": 0.060374826192855835,
      "learning_rate": 8.339790171606082e-07,
      "loss": 0.0006,
      "step": 152570
    },
    {
      "epoch": 2.4966047615151763,
      "grad_norm": 0.05970359221100807,
      "learning_rate": 8.334527027119361e-07,
      "loss": 0.0006,
      "step": 152580
    },
    {
      "epoch": 2.4967683874662523,
      "grad_norm": 0.03606754541397095,
      "learning_rate": 8.329265392905911e-07,
      "loss": 0.0008,
      "step": 152590
    },
    {
      "epoch": 2.496932013417328,
      "grad_norm": 0.016530266031622887,
      "learning_rate": 8.32400526915642e-07,
      "loss": 0.0005,
      "step": 152600
    },
    {
      "epoch": 2.497095639368404,
      "grad_norm": 0.153111532330513,
      "learning_rate": 8.318746656061577e-07,
      "loss": 0.0009,
      "step": 152610
    },
    {
      "epoch": 2.49725926531948,
      "grad_norm": 0.13251042366027832,
      "learning_rate": 8.313489553811999e-07,
      "loss": 0.0015,
      "step": 152620
    },
    {
      "epoch": 2.4974228912705554,
      "grad_norm": 0.06636282056570053,
      "learning_rate": 8.308233962598228e-07,
      "loss": 0.0011,
      "step": 152630
    },
    {
      "epoch": 2.4975865172216314,
      "grad_norm": 0.0069928402081131935,
      "learning_rate": 8.302979882610779e-07,
      "loss": 0.0007,
      "step": 152640
    },
    {
      "epoch": 2.4977501431727074,
      "grad_norm": 0.0771067813038826,
      "learning_rate": 8.297727314040094e-07,
      "loss": 0.0006,
      "step": 152650
    },
    {
      "epoch": 2.497913769123783,
      "grad_norm": 0.1012401208281517,
      "learning_rate": 8.292476257076548e-07,
      "loss": 0.0007,
      "step": 152660
    },
    {
      "epoch": 2.498077395074859,
      "grad_norm": 0.09480570256710052,
      "learning_rate": 8.287226711910517e-07,
      "loss": 0.0007,
      "step": 152670
    },
    {
      "epoch": 2.498241021025935,
      "grad_norm": 0.04493458941578865,
      "learning_rate": 8.281978678732244e-07,
      "loss": 0.0007,
      "step": 152680
    },
    {
      "epoch": 2.4984046469770105,
      "grad_norm": 0.00846731849014759,
      "learning_rate": 8.276732157731998e-07,
      "loss": 0.0006,
      "step": 152690
    },
    {
      "epoch": 2.4985682729280865,
      "grad_norm": 0.009564354084432125,
      "learning_rate": 8.271487149099911e-07,
      "loss": 0.0004,
      "step": 152700
    },
    {
      "epoch": 2.4987318988791625,
      "grad_norm": 0.07969698309898376,
      "learning_rate": 8.26624365302614e-07,
      "loss": 0.001,
      "step": 152710
    },
    {
      "epoch": 2.498895524830238,
      "grad_norm": 0.008087425492703915,
      "learning_rate": 8.261001669700719e-07,
      "loss": 0.0004,
      "step": 152720
    },
    {
      "epoch": 2.499059150781314,
      "grad_norm": 0.15744855999946594,
      "learning_rate": 8.255761199313683e-07,
      "loss": 0.0023,
      "step": 152730
    },
    {
      "epoch": 2.4992227767323896,
      "grad_norm": 0.30521252751350403,
      "learning_rate": 8.250522242054965e-07,
      "loss": 0.0018,
      "step": 152740
    },
    {
      "epoch": 2.4993864026834656,
      "grad_norm": 0.005865911953151226,
      "learning_rate": 8.245284798114478e-07,
      "loss": 0.0005,
      "step": 152750
    },
    {
      "epoch": 2.4995500286345416,
      "grad_norm": 0.0023570482153445482,
      "learning_rate": 8.240048867682066e-07,
      "loss": 0.0011,
      "step": 152760
    },
    {
      "epoch": 2.499713654585617,
      "grad_norm": 0.0015459408750757575,
      "learning_rate": 8.234814450947503e-07,
      "loss": 0.0006,
      "step": 152770
    },
    {
      "epoch": 2.499877280536693,
      "grad_norm": 0.017839189618825912,
      "learning_rate": 8.229581548100551e-07,
      "loss": 0.0006,
      "step": 152780
    },
    {
      "epoch": 2.500040906487769,
      "grad_norm": 0.21259450912475586,
      "learning_rate": 8.224350159330863e-07,
      "loss": 0.0016,
      "step": 152790
    },
    {
      "epoch": 2.5002045324388447,
      "grad_norm": 0.03174861893057823,
      "learning_rate": 8.219120284828086e-07,
      "loss": 0.0014,
      "step": 152800
    },
    {
      "epoch": 2.5003681583899207,
      "grad_norm": 0.29108726978302,
      "learning_rate": 8.213891924781769e-07,
      "loss": 0.0013,
      "step": 152810
    },
    {
      "epoch": 2.500531784340996,
      "grad_norm": 0.15005546808242798,
      "learning_rate": 8.208665079381439e-07,
      "loss": 0.0006,
      "step": 152820
    },
    {
      "epoch": 2.500695410292072,
      "grad_norm": 0.07798822969198227,
      "learning_rate": 8.203439748816572e-07,
      "loss": 0.0007,
      "step": 152830
    },
    {
      "epoch": 2.500859036243148,
      "grad_norm": 0.08350639790296555,
      "learning_rate": 8.198215933276543e-07,
      "loss": 0.0007,
      "step": 152840
    },
    {
      "epoch": 2.5010226621942238,
      "grad_norm": 0.20817913115024567,
      "learning_rate": 8.192993632950747e-07,
      "loss": 0.003,
      "step": 152850
    },
    {
      "epoch": 2.5011862881452998,
      "grad_norm": 0.04662749543786049,
      "learning_rate": 8.187772848028419e-07,
      "loss": 0.0019,
      "step": 152860
    },
    {
      "epoch": 2.5013499140963757,
      "grad_norm": 0.14220450818538666,
      "learning_rate": 8.182553578698854e-07,
      "loss": 0.0009,
      "step": 152870
    },
    {
      "epoch": 2.5015135400474513,
      "grad_norm": 0.08429339528083801,
      "learning_rate": 8.1773358251512e-07,
      "loss": 0.0011,
      "step": 152880
    },
    {
      "epoch": 2.5016771659985273,
      "grad_norm": 0.11262070387601852,
      "learning_rate": 8.172119587574601e-07,
      "loss": 0.0006,
      "step": 152890
    },
    {
      "epoch": 2.5018407919496033,
      "grad_norm": 0.07514145225286484,
      "learning_rate": 8.166904866158154e-07,
      "loss": 0.0007,
      "step": 152900
    },
    {
      "epoch": 2.502004417900679,
      "grad_norm": 0.0766606330871582,
      "learning_rate": 8.161691661090848e-07,
      "loss": 0.0011,
      "step": 152910
    },
    {
      "epoch": 2.502168043851755,
      "grad_norm": 0.05160791799426079,
      "learning_rate": 8.156479972561677e-07,
      "loss": 0.0009,
      "step": 152920
    },
    {
      "epoch": 2.502331669802831,
      "grad_norm": 0.10994260758161545,
      "learning_rate": 8.15126980075952e-07,
      "loss": 0.0005,
      "step": 152930
    },
    {
      "epoch": 2.5024952957539064,
      "grad_norm": 0.0421556755900383,
      "learning_rate": 8.146061145873269e-07,
      "loss": 0.0008,
      "step": 152940
    },
    {
      "epoch": 2.5026589217049824,
      "grad_norm": 0.05681430548429489,
      "learning_rate": 8.140854008091708e-07,
      "loss": 0.0008,
      "step": 152950
    },
    {
      "epoch": 2.5028225476560584,
      "grad_norm": 0.004639185965061188,
      "learning_rate": 8.135648387603573e-07,
      "loss": 0.0005,
      "step": 152960
    },
    {
      "epoch": 2.502986173607134,
      "grad_norm": 0.06267756968736649,
      "learning_rate": 8.130444284597577e-07,
      "loss": 0.0008,
      "step": 152970
    },
    {
      "epoch": 2.50314979955821,
      "grad_norm": 0.051748666912317276,
      "learning_rate": 8.125241699262331e-07,
      "loss": 0.0005,
      "step": 152980
    },
    {
      "epoch": 2.503313425509286,
      "grad_norm": 0.2513732612133026,
      "learning_rate": 8.120040631786447e-07,
      "loss": 0.0014,
      "step": 152990
    },
    {
      "epoch": 2.5034770514603615,
      "grad_norm": 0.007329247891902924,
      "learning_rate": 8.114841082358415e-07,
      "loss": 0.0006,
      "step": 153000
    },
    {
      "epoch": 2.5036406774114375,
      "grad_norm": 0.050909966230392456,
      "learning_rate": 8.109643051166743e-07,
      "loss": 0.0013,
      "step": 153010
    },
    {
      "epoch": 2.5038043033625135,
      "grad_norm": 0.060371242463588715,
      "learning_rate": 8.104446538399813e-07,
      "loss": 0.0006,
      "step": 153020
    },
    {
      "epoch": 2.503967929313589,
      "grad_norm": 0.16189496219158173,
      "learning_rate": 8.099251544246018e-07,
      "loss": 0.0012,
      "step": 153030
    },
    {
      "epoch": 2.504131555264665,
      "grad_norm": 0.027333498001098633,
      "learning_rate": 8.094058068893634e-07,
      "loss": 0.0005,
      "step": 153040
    },
    {
      "epoch": 2.504295181215741,
      "grad_norm": 0.01574457250535488,
      "learning_rate": 8.088866112530941e-07,
      "loss": 0.0005,
      "step": 153050
    },
    {
      "epoch": 2.5044588071668166,
      "grad_norm": 0.08019930124282837,
      "learning_rate": 8.08367567534612e-07,
      "loss": 0.0009,
      "step": 153060
    },
    {
      "epoch": 2.5046224331178926,
      "grad_norm": 0.10814481228590012,
      "learning_rate": 8.078486757527304e-07,
      "loss": 0.0006,
      "step": 153070
    },
    {
      "epoch": 2.5047860590689686,
      "grad_norm": 0.006461639888584614,
      "learning_rate": 8.073299359262593e-07,
      "loss": 0.0008,
      "step": 153080
    },
    {
      "epoch": 2.504949685020044,
      "grad_norm": 0.013817775063216686,
      "learning_rate": 8.068113480740003e-07,
      "loss": 0.0005,
      "step": 153090
    },
    {
      "epoch": 2.50511331097112,
      "grad_norm": 0.06163875758647919,
      "learning_rate": 8.062929122147522e-07,
      "loss": 0.0007,
      "step": 153100
    },
    {
      "epoch": 2.505276936922196,
      "grad_norm": 0.13409602642059326,
      "learning_rate": 8.057746283673073e-07,
      "loss": 0.0009,
      "step": 153110
    },
    {
      "epoch": 2.5054405628732717,
      "grad_norm": 0.03343839943408966,
      "learning_rate": 8.052564965504506e-07,
      "loss": 0.0005,
      "step": 153120
    },
    {
      "epoch": 2.5056041888243477,
      "grad_norm": 0.020452139899134636,
      "learning_rate": 8.047385167829658e-07,
      "loss": 0.0009,
      "step": 153130
    },
    {
      "epoch": 2.5057678147754237,
      "grad_norm": 0.06642342358827591,
      "learning_rate": 8.042206890836252e-07,
      "loss": 0.0007,
      "step": 153140
    },
    {
      "epoch": 2.505931440726499,
      "grad_norm": 0.11334613710641861,
      "learning_rate": 8.037030134712026e-07,
      "loss": 0.0011,
      "step": 153150
    },
    {
      "epoch": 2.506095066677575,
      "grad_norm": 0.08409946411848068,
      "learning_rate": 8.031854899644575e-07,
      "loss": 0.0005,
      "step": 153160
    },
    {
      "epoch": 2.5062586926286508,
      "grad_norm": 0.07164382189512253,
      "learning_rate": 8.026681185821522e-07,
      "loss": 0.0007,
      "step": 153170
    },
    {
      "epoch": 2.5064223185797267,
      "grad_norm": 0.028023742139339447,
      "learning_rate": 8.021508993430405e-07,
      "loss": 0.0012,
      "step": 153180
    },
    {
      "epoch": 2.5065859445308027,
      "grad_norm": 0.06099424883723259,
      "learning_rate": 8.016338322658679e-07,
      "loss": 0.0005,
      "step": 153190
    },
    {
      "epoch": 2.5067495704818783,
      "grad_norm": 0.10231123864650726,
      "learning_rate": 8.011169173693801e-07,
      "loss": 0.0005,
      "step": 153200
    },
    {
      "epoch": 2.5069131964329543,
      "grad_norm": 0.12664766609668732,
      "learning_rate": 8.006001546723107e-07,
      "loss": 0.001,
      "step": 153210
    },
    {
      "epoch": 2.50707682238403,
      "grad_norm": 0.16597691178321838,
      "learning_rate": 8.000835441933947e-07,
      "loss": 0.0007,
      "step": 153220
    },
    {
      "epoch": 2.507240448335106,
      "grad_norm": 0.12338769435882568,
      "learning_rate": 7.995670859513538e-07,
      "loss": 0.0009,
      "step": 153230
    },
    {
      "epoch": 2.507404074286182,
      "grad_norm": 0.03035760670900345,
      "learning_rate": 7.990507799649122e-07,
      "loss": 0.0006,
      "step": 153240
    },
    {
      "epoch": 2.5075677002372574,
      "grad_norm": 0.06313779205083847,
      "learning_rate": 7.985346262527832e-07,
      "loss": 0.0004,
      "step": 153250
    },
    {
      "epoch": 2.5077313261883334,
      "grad_norm": 0.10467544943094254,
      "learning_rate": 7.980186248336746e-07,
      "loss": 0.0011,
      "step": 153260
    },
    {
      "epoch": 2.5078949521394094,
      "grad_norm": 0.01759118027985096,
      "learning_rate": 7.975027757262932e-07,
      "loss": 0.0005,
      "step": 153270
    },
    {
      "epoch": 2.508058578090485,
      "grad_norm": 0.0060456921346485615,
      "learning_rate": 7.969870789493345e-07,
      "loss": 0.0005,
      "step": 153280
    },
    {
      "epoch": 2.508222204041561,
      "grad_norm": 0.014970232732594013,
      "learning_rate": 7.964715345214935e-07,
      "loss": 0.0006,
      "step": 153290
    },
    {
      "epoch": 2.508385829992637,
      "grad_norm": 0.07907494902610779,
      "learning_rate": 7.959561424614553e-07,
      "loss": 0.0008,
      "step": 153300
    },
    {
      "epoch": 2.5085494559437125,
      "grad_norm": 0.13713517785072327,
      "learning_rate": 7.95440902787904e-07,
      "loss": 0.0006,
      "step": 153310
    },
    {
      "epoch": 2.5087130818947885,
      "grad_norm": 0.05684197321534157,
      "learning_rate": 7.949258155195133e-07,
      "loss": 0.0004,
      "step": 153320
    },
    {
      "epoch": 2.5088767078458645,
      "grad_norm": 0.05368838831782341,
      "learning_rate": 7.944108806749551e-07,
      "loss": 0.0007,
      "step": 153330
    },
    {
      "epoch": 2.50904033379694,
      "grad_norm": 0.04662998020648956,
      "learning_rate": 7.938960982728961e-07,
      "loss": 0.0006,
      "step": 153340
    },
    {
      "epoch": 2.509203959748016,
      "grad_norm": 0.04324856027960777,
      "learning_rate": 7.93381468331994e-07,
      "loss": 0.0005,
      "step": 153350
    },
    {
      "epoch": 2.509367585699092,
      "grad_norm": 0.002995358081534505,
      "learning_rate": 7.928669908709041e-07,
      "loss": 0.0008,
      "step": 153360
    },
    {
      "epoch": 2.5095312116501676,
      "grad_norm": 0.0844714492559433,
      "learning_rate": 7.923526659082725e-07,
      "loss": 0.0008,
      "step": 153370
    },
    {
      "epoch": 2.5096948376012436,
      "grad_norm": 0.06279635429382324,
      "learning_rate": 7.918384934627438e-07,
      "loss": 0.0006,
      "step": 153380
    },
    {
      "epoch": 2.5098584635523196,
      "grad_norm": 0.1071927472949028,
      "learning_rate": 7.913244735529569e-07,
      "loss": 0.0008,
      "step": 153390
    },
    {
      "epoch": 2.510022089503395,
      "grad_norm": 0.22083964943885803,
      "learning_rate": 7.908106061975418e-07,
      "loss": 0.0007,
      "step": 153400
    },
    {
      "epoch": 2.510185715454471,
      "grad_norm": 0.11697695404291153,
      "learning_rate": 7.902968914151265e-07,
      "loss": 0.001,
      "step": 153410
    },
    {
      "epoch": 2.510349341405547,
      "grad_norm": 0.0012254059547558427,
      "learning_rate": 7.897833292243295e-07,
      "loss": 0.0005,
      "step": 153420
    },
    {
      "epoch": 2.5105129673566227,
      "grad_norm": 0.0795721560716629,
      "learning_rate": 7.892699196437698e-07,
      "loss": 0.0042,
      "step": 153430
    },
    {
      "epoch": 2.5106765933076987,
      "grad_norm": 0.10146130621433258,
      "learning_rate": 7.88756662692054e-07,
      "loss": 0.0006,
      "step": 153440
    },
    {
      "epoch": 2.5108402192587747,
      "grad_norm": 0.0592157244682312,
      "learning_rate": 7.882435583877885e-07,
      "loss": 0.0007,
      "step": 153450
    },
    {
      "epoch": 2.51100384520985,
      "grad_norm": 0.012290414422750473,
      "learning_rate": 7.877306067495716e-07,
      "loss": 0.001,
      "step": 153460
    },
    {
      "epoch": 2.511167471160926,
      "grad_norm": 0.03896334022283554,
      "learning_rate": 7.872178077959953e-07,
      "loss": 0.0009,
      "step": 153470
    },
    {
      "epoch": 2.511331097112002,
      "grad_norm": 0.006612281780689955,
      "learning_rate": 7.867051615456489e-07,
      "loss": 0.0009,
      "step": 153480
    },
    {
      "epoch": 2.5114947230630777,
      "grad_norm": 0.07035757601261139,
      "learning_rate": 7.861926680171133e-07,
      "loss": 0.0011,
      "step": 153490
    },
    {
      "epoch": 2.5116583490141537,
      "grad_norm": 0.023695295676589012,
      "learning_rate": 7.85680327228967e-07,
      "loss": 0.0005,
      "step": 153500
    },
    {
      "epoch": 2.5118219749652297,
      "grad_norm": 0.10984907299280167,
      "learning_rate": 7.851681391997784e-07,
      "loss": 0.001,
      "step": 153510
    },
    {
      "epoch": 2.5119856009163053,
      "grad_norm": 0.08776549249887466,
      "learning_rate": 7.846561039481166e-07,
      "loss": 0.0009,
      "step": 153520
    },
    {
      "epoch": 2.5121492268673813,
      "grad_norm": 0.08424146473407745,
      "learning_rate": 7.841442214925376e-07,
      "loss": 0.0004,
      "step": 153530
    },
    {
      "epoch": 2.512312852818457,
      "grad_norm": 0.07765801250934601,
      "learning_rate": 7.836324918515986e-07,
      "loss": 0.0012,
      "step": 153540
    },
    {
      "epoch": 2.512476478769533,
      "grad_norm": 0.08758638799190521,
      "learning_rate": 7.831209150438507e-07,
      "loss": 0.0014,
      "step": 153550
    },
    {
      "epoch": 2.512640104720609,
      "grad_norm": 0.1425127536058426,
      "learning_rate": 7.826094910878318e-07,
      "loss": 0.0004,
      "step": 153560
    },
    {
      "epoch": 2.5128037306716844,
      "grad_norm": 0.13368819653987885,
      "learning_rate": 7.820982200020838e-07,
      "loss": 0.0008,
      "step": 153570
    },
    {
      "epoch": 2.5129673566227604,
      "grad_norm": 0.15390488505363464,
      "learning_rate": 7.815871018051369e-07,
      "loss": 0.0009,
      "step": 153580
    },
    {
      "epoch": 2.513130982573836,
      "grad_norm": 0.07005128264427185,
      "learning_rate": 7.810761365155184e-07,
      "loss": 0.0008,
      "step": 153590
    },
    {
      "epoch": 2.513294608524912,
      "grad_norm": 0.01793326810002327,
      "learning_rate": 7.805653241517519e-07,
      "loss": 0.0006,
      "step": 153600
    },
    {
      "epoch": 2.513458234475988,
      "grad_norm": 0.01182679831981659,
      "learning_rate": 7.800546647323493e-07,
      "loss": 0.0002,
      "step": 153610
    },
    {
      "epoch": 2.5136218604270635,
      "grad_norm": 0.07243483513593674,
      "learning_rate": 7.795441582758245e-07,
      "loss": 0.0007,
      "step": 153620
    },
    {
      "epoch": 2.5137854863781395,
      "grad_norm": 0.11097566783428192,
      "learning_rate": 7.790338048006791e-07,
      "loss": 0.0006,
      "step": 153630
    },
    {
      "epoch": 2.5139491123292155,
      "grad_norm": 0.16654329001903534,
      "learning_rate": 7.785236043254146e-07,
      "loss": 0.0008,
      "step": 153640
    },
    {
      "epoch": 2.514112738280291,
      "grad_norm": 0.10942531377077103,
      "learning_rate": 7.780135568685232e-07,
      "loss": 0.0008,
      "step": 153650
    },
    {
      "epoch": 2.514276364231367,
      "grad_norm": 0.009948287159204483,
      "learning_rate": 7.775036624484922e-07,
      "loss": 0.0004,
      "step": 153660
    },
    {
      "epoch": 2.514439990182443,
      "grad_norm": 0.13962295651435852,
      "learning_rate": 7.769939210838057e-07,
      "loss": 0.0013,
      "step": 153670
    },
    {
      "epoch": 2.5146036161335186,
      "grad_norm": 0.05111374706029892,
      "learning_rate": 7.764843327929389e-07,
      "loss": 0.0004,
      "step": 153680
    },
    {
      "epoch": 2.5147672420845946,
      "grad_norm": 0.0080978162586689,
      "learning_rate": 7.759748975943654e-07,
      "loss": 0.0008,
      "step": 153690
    },
    {
      "epoch": 2.5149308680356706,
      "grad_norm": 0.0877828449010849,
      "learning_rate": 7.754656155065482e-07,
      "loss": 0.0006,
      "step": 153700
    },
    {
      "epoch": 2.515094493986746,
      "grad_norm": 0.08099866658449173,
      "learning_rate": 7.749564865479503e-07,
      "loss": 0.0009,
      "step": 153710
    },
    {
      "epoch": 2.515258119937822,
      "grad_norm": 0.06333150714635849,
      "learning_rate": 7.744475107370241e-07,
      "loss": 0.0007,
      "step": 153720
    },
    {
      "epoch": 2.515421745888898,
      "grad_norm": 0.2652546167373657,
      "learning_rate": 7.739386880922212e-07,
      "loss": 0.0014,
      "step": 153730
    },
    {
      "epoch": 2.5155853718399737,
      "grad_norm": 0.00912480615079403,
      "learning_rate": 7.734300186319821e-07,
      "loss": 0.0006,
      "step": 153740
    },
    {
      "epoch": 2.5157489977910497,
      "grad_norm": 0.010948944836854935,
      "learning_rate": 7.729215023747477e-07,
      "loss": 0.0008,
      "step": 153750
    },
    {
      "epoch": 2.5159126237421257,
      "grad_norm": 0.013310256414115429,
      "learning_rate": 7.7241313933895e-07,
      "loss": 0.0006,
      "step": 153760
    },
    {
      "epoch": 2.516076249693201,
      "grad_norm": 0.00677884416654706,
      "learning_rate": 7.719049295430131e-07,
      "loss": 0.0008,
      "step": 153770
    },
    {
      "epoch": 2.516239875644277,
      "grad_norm": 0.1461547166109085,
      "learning_rate": 7.713968730053623e-07,
      "loss": 0.0013,
      "step": 153780
    },
    {
      "epoch": 2.516403501595353,
      "grad_norm": 0.2367231249809265,
      "learning_rate": 7.708889697444106e-07,
      "loss": 0.0008,
      "step": 153790
    },
    {
      "epoch": 2.5165671275464288,
      "grad_norm": 0.14580337703227997,
      "learning_rate": 7.703812197785704e-07,
      "loss": 0.0009,
      "step": 153800
    },
    {
      "epoch": 2.5167307534975047,
      "grad_norm": 0.08432180434465408,
      "learning_rate": 7.698736231262444e-07,
      "loss": 0.0024,
      "step": 153810
    },
    {
      "epoch": 2.5168943794485807,
      "grad_norm": 0.06188319995999336,
      "learning_rate": 7.693661798058321e-07,
      "loss": 0.0008,
      "step": 153820
    },
    {
      "epoch": 2.5170580053996563,
      "grad_norm": 0.06611020863056183,
      "learning_rate": 7.688588898357296e-07,
      "loss": 0.0007,
      "step": 153830
    },
    {
      "epoch": 2.5172216313507323,
      "grad_norm": 0.04763653501868248,
      "learning_rate": 7.683517532343215e-07,
      "loss": 0.0005,
      "step": 153840
    },
    {
      "epoch": 2.5173852573018083,
      "grad_norm": 0.002139064949005842,
      "learning_rate": 7.678447700199931e-07,
      "loss": 0.0003,
      "step": 153850
    },
    {
      "epoch": 2.517548883252884,
      "grad_norm": 0.0751110166311264,
      "learning_rate": 7.673379402111197e-07,
      "loss": 0.0005,
      "step": 153860
    },
    {
      "epoch": 2.51771250920396,
      "grad_norm": 0.042201071977615356,
      "learning_rate": 7.668312638260722e-07,
      "loss": 0.0004,
      "step": 153870
    },
    {
      "epoch": 2.517876135155036,
      "grad_norm": 0.007482283748686314,
      "learning_rate": 7.663247408832187e-07,
      "loss": 0.0008,
      "step": 153880
    },
    {
      "epoch": 2.5180397611061114,
      "grad_norm": 0.09299102425575256,
      "learning_rate": 7.658183714009166e-07,
      "loss": 0.0008,
      "step": 153890
    },
    {
      "epoch": 2.5182033870571874,
      "grad_norm": 0.051814738661050797,
      "learning_rate": 7.653121553975229e-07,
      "loss": 0.0009,
      "step": 153900
    },
    {
      "epoch": 2.5183670130082634,
      "grad_norm": 0.09576883167028427,
      "learning_rate": 7.648060928913847e-07,
      "loss": 0.0009,
      "step": 153910
    },
    {
      "epoch": 2.518530638959339,
      "grad_norm": 0.017623549327254295,
      "learning_rate": 7.643001839008479e-07,
      "loss": 0.0004,
      "step": 153920
    },
    {
      "epoch": 2.518694264910415,
      "grad_norm": 0.14092279970645905,
      "learning_rate": 7.637944284442478e-07,
      "loss": 0.0012,
      "step": 153930
    },
    {
      "epoch": 2.5188578908614905,
      "grad_norm": 0.13379254937171936,
      "learning_rate": 7.632888265399196e-07,
      "loss": 0.0006,
      "step": 153940
    },
    {
      "epoch": 2.5190215168125665,
      "grad_norm": 0.12886199355125427,
      "learning_rate": 7.627833782061877e-07,
      "loss": 0.0017,
      "step": 153950
    },
    {
      "epoch": 2.5191851427636425,
      "grad_norm": 0.05948249250650406,
      "learning_rate": 7.622780834613752e-07,
      "loss": 0.0007,
      "step": 153960
    },
    {
      "epoch": 2.519348768714718,
      "grad_norm": 0.12851360440254211,
      "learning_rate": 7.617729423237974e-07,
      "loss": 0.0008,
      "step": 153970
    },
    {
      "epoch": 2.519512394665794,
      "grad_norm": 0.09617947787046432,
      "learning_rate": 7.61267954811763e-07,
      "loss": 0.0005,
      "step": 153980
    },
    {
      "epoch": 2.5196760206168696,
      "grad_norm": 0.019781513139605522,
      "learning_rate": 7.60763120943579e-07,
      "loss": 0.0006,
      "step": 153990
    },
    {
      "epoch": 2.5198396465679456,
      "grad_norm": 0.1499721109867096,
      "learning_rate": 7.602584407375413e-07,
      "loss": 0.0008,
      "step": 154000
    },
    {
      "epoch": 2.5200032725190216,
      "grad_norm": 0.07409385591745377,
      "learning_rate": 7.59753914211947e-07,
      "loss": 0.0005,
      "step": 154010
    },
    {
      "epoch": 2.520166898470097,
      "grad_norm": 0.019088326022028923,
      "learning_rate": 7.592495413850809e-07,
      "loss": 0.0004,
      "step": 154020
    },
    {
      "epoch": 2.520330524421173,
      "grad_norm": 0.051016125828027725,
      "learning_rate": 7.587453222752261e-07,
      "loss": 0.0005,
      "step": 154030
    },
    {
      "epoch": 2.520494150372249,
      "grad_norm": 0.08586464077234268,
      "learning_rate": 7.582412569006614e-07,
      "loss": 0.0005,
      "step": 154040
    },
    {
      "epoch": 2.5206577763233247,
      "grad_norm": 0.050012800842523575,
      "learning_rate": 7.577373452796549e-07,
      "loss": 0.0007,
      "step": 154050
    },
    {
      "epoch": 2.5208214022744007,
      "grad_norm": 0.4315398335456848,
      "learning_rate": 7.572335874304759e-07,
      "loss": 0.0006,
      "step": 154060
    },
    {
      "epoch": 2.5209850282254767,
      "grad_norm": 0.17931661009788513,
      "learning_rate": 7.567299833713803e-07,
      "loss": 0.0013,
      "step": 154070
    },
    {
      "epoch": 2.521148654176552,
      "grad_norm": 0.09433405101299286,
      "learning_rate": 7.562265331206253e-07,
      "loss": 0.0012,
      "step": 154080
    },
    {
      "epoch": 2.521312280127628,
      "grad_norm": 0.010839276947081089,
      "learning_rate": 7.557232366964573e-07,
      "loss": 0.0006,
      "step": 154090
    },
    {
      "epoch": 2.521475906078704,
      "grad_norm": 0.06988219916820526,
      "learning_rate": 7.552200941171217e-07,
      "loss": 0.0014,
      "step": 154100
    },
    {
      "epoch": 2.5216395320297798,
      "grad_norm": 0.11719708144664764,
      "learning_rate": 7.547171054008562e-07,
      "loss": 0.0017,
      "step": 154110
    },
    {
      "epoch": 2.5218031579808557,
      "grad_norm": 0.08835003525018692,
      "learning_rate": 7.542142705658917e-07,
      "loss": 0.0007,
      "step": 154120
    },
    {
      "epoch": 2.5219667839319317,
      "grad_norm": 0.08042863011360168,
      "learning_rate": 7.537115896304565e-07,
      "loss": 0.0007,
      "step": 154130
    },
    {
      "epoch": 2.5221304098830073,
      "grad_norm": 0.07623101025819778,
      "learning_rate": 7.532090626127691e-07,
      "loss": 0.0012,
      "step": 154140
    },
    {
      "epoch": 2.5222940358340833,
      "grad_norm": 0.07541704922914505,
      "learning_rate": 7.527066895310475e-07,
      "loss": 0.001,
      "step": 154150
    },
    {
      "epoch": 2.5224576617851593,
      "grad_norm": 0.00587494857609272,
      "learning_rate": 7.522044704035003e-07,
      "loss": 0.0005,
      "step": 154160
    },
    {
      "epoch": 2.522621287736235,
      "grad_norm": 0.2748318016529083,
      "learning_rate": 7.517024052483307e-07,
      "loss": 0.001,
      "step": 154170
    },
    {
      "epoch": 2.522784913687311,
      "grad_norm": 0.1162383109331131,
      "learning_rate": 7.512004940837392e-07,
      "loss": 0.0005,
      "step": 154180
    },
    {
      "epoch": 2.522948539638387,
      "grad_norm": 0.08337512612342834,
      "learning_rate": 7.506987369279173e-07,
      "loss": 0.0005,
      "step": 154190
    },
    {
      "epoch": 2.5231121655894624,
      "grad_norm": 0.0536191388964653,
      "learning_rate": 7.501971337990544e-07,
      "loss": 0.0006,
      "step": 154200
    },
    {
      "epoch": 2.5232757915405384,
      "grad_norm": 0.03441673144698143,
      "learning_rate": 7.496956847153297e-07,
      "loss": 0.0007,
      "step": 154210
    },
    {
      "epoch": 2.5234394174916144,
      "grad_norm": 0.020692232996225357,
      "learning_rate": 7.491943896949227e-07,
      "loss": 0.0006,
      "step": 154220
    },
    {
      "epoch": 2.52360304344269,
      "grad_norm": 0.12605896592140198,
      "learning_rate": 7.486932487560011e-07,
      "loss": 0.0007,
      "step": 154230
    },
    {
      "epoch": 2.523766669393766,
      "grad_norm": 0.06521426141262054,
      "learning_rate": 7.481922619167331e-07,
      "loss": 0.0008,
      "step": 154240
    },
    {
      "epoch": 2.523930295344842,
      "grad_norm": 0.08876814693212509,
      "learning_rate": 7.476914291952753e-07,
      "loss": 0.0007,
      "step": 154250
    },
    {
      "epoch": 2.5240939212959175,
      "grad_norm": 0.034678563475608826,
      "learning_rate": 7.471907506097842e-07,
      "loss": 0.0011,
      "step": 154260
    },
    {
      "epoch": 2.5242575472469935,
      "grad_norm": 0.0986066609621048,
      "learning_rate": 7.466902261784065e-07,
      "loss": 0.0007,
      "step": 154270
    },
    {
      "epoch": 2.5244211731980695,
      "grad_norm": 0.03450404107570648,
      "learning_rate": 7.461898559192848e-07,
      "loss": 0.0006,
      "step": 154280
    },
    {
      "epoch": 2.524584799149145,
      "grad_norm": 0.048597030341625214,
      "learning_rate": 7.456896398505581e-07,
      "loss": 0.0004,
      "step": 154290
    },
    {
      "epoch": 2.524748425100221,
      "grad_norm": 0.04881170764565468,
      "learning_rate": 7.451895779903562e-07,
      "loss": 0.0012,
      "step": 154300
    },
    {
      "epoch": 2.524912051051297,
      "grad_norm": 0.19125165045261383,
      "learning_rate": 7.446896703568057e-07,
      "loss": 0.0005,
      "step": 154310
    },
    {
      "epoch": 2.5250756770023726,
      "grad_norm": 0.014020061120390892,
      "learning_rate": 7.441899169680284e-07,
      "loss": 0.0009,
      "step": 154320
    },
    {
      "epoch": 2.5252393029534486,
      "grad_norm": 0.027234815061092377,
      "learning_rate": 7.436903178421368e-07,
      "loss": 0.0005,
      "step": 154330
    },
    {
      "epoch": 2.525402928904524,
      "grad_norm": 0.019534893333911896,
      "learning_rate": 7.431908729972426e-07,
      "loss": 0.0008,
      "step": 154340
    },
    {
      "epoch": 2.5255665548556,
      "grad_norm": 0.030101966112852097,
      "learning_rate": 7.426915824514474e-07,
      "loss": 0.0004,
      "step": 154350
    },
    {
      "epoch": 2.5257301808066757,
      "grad_norm": 0.1376025378704071,
      "learning_rate": 7.421924462228519e-07,
      "loss": 0.001,
      "step": 154360
    },
    {
      "epoch": 2.5258938067577517,
      "grad_norm": 0.13926704227924347,
      "learning_rate": 7.416934643295454e-07,
      "loss": 0.0013,
      "step": 154370
    },
    {
      "epoch": 2.5260574327088277,
      "grad_norm": 0.0640011802315712,
      "learning_rate": 7.411946367896156e-07,
      "loss": 0.0006,
      "step": 154380
    },
    {
      "epoch": 2.526221058659903,
      "grad_norm": 0.04125287011265755,
      "learning_rate": 7.406959636211458e-07,
      "loss": 0.0008,
      "step": 154390
    },
    {
      "epoch": 2.526384684610979,
      "grad_norm": 0.021210316568613052,
      "learning_rate": 7.40197444842209e-07,
      "loss": 0.0012,
      "step": 154400
    },
    {
      "epoch": 2.526548310562055,
      "grad_norm": 0.0033365320414304733,
      "learning_rate": 7.396990804708781e-07,
      "loss": 0.0005,
      "step": 154410
    },
    {
      "epoch": 2.5267119365131308,
      "grad_norm": 0.22003218531608582,
      "learning_rate": 7.392008705252146e-07,
      "loss": 0.0013,
      "step": 154420
    },
    {
      "epoch": 2.5268755624642067,
      "grad_norm": 0.05437715724110603,
      "learning_rate": 7.387028150232806e-07,
      "loss": 0.0012,
      "step": 154430
    },
    {
      "epoch": 2.5270391884152827,
      "grad_norm": 0.07756680995225906,
      "learning_rate": 7.382049139831266e-07,
      "loss": 0.001,
      "step": 154440
    },
    {
      "epoch": 2.5272028143663583,
      "grad_norm": 0.11223863810300827,
      "learning_rate": 7.377071674228025e-07,
      "loss": 0.0008,
      "step": 154450
    },
    {
      "epoch": 2.5273664403174343,
      "grad_norm": 0.046417899429798126,
      "learning_rate": 7.372095753603492e-07,
      "loss": 0.001,
      "step": 154460
    },
    {
      "epoch": 2.5275300662685103,
      "grad_norm": 0.0523376427590847,
      "learning_rate": 7.367121378138025e-07,
      "loss": 0.0007,
      "step": 154470
    },
    {
      "epoch": 2.527693692219586,
      "grad_norm": 0.012180913239717484,
      "learning_rate": 7.362148548011955e-07,
      "loss": 0.0015,
      "step": 154480
    },
    {
      "epoch": 2.527857318170662,
      "grad_norm": 0.11514504253864288,
      "learning_rate": 7.357177263405513e-07,
      "loss": 0.0006,
      "step": 154490
    },
    {
      "epoch": 2.528020944121738,
      "grad_norm": 0.03242136165499687,
      "learning_rate": 7.352207524498917e-07,
      "loss": 0.0004,
      "step": 154500
    },
    {
      "epoch": 2.5281845700728134,
      "grad_norm": 0.02298622950911522,
      "learning_rate": 7.347239331472278e-07,
      "loss": 0.0007,
      "step": 154510
    },
    {
      "epoch": 2.5283481960238894,
      "grad_norm": 0.016909729689359665,
      "learning_rate": 7.342272684505719e-07,
      "loss": 0.0009,
      "step": 154520
    },
    {
      "epoch": 2.5285118219749654,
      "grad_norm": 0.08366072922945023,
      "learning_rate": 7.337307583779235e-07,
      "loss": 0.0009,
      "step": 154530
    },
    {
      "epoch": 2.528675447926041,
      "grad_norm": 0.07496605068445206,
      "learning_rate": 7.332344029472815e-07,
      "loss": 0.0004,
      "step": 154540
    },
    {
      "epoch": 2.528839073877117,
      "grad_norm": 0.006733448710292578,
      "learning_rate": 7.327382021766388e-07,
      "loss": 0.0005,
      "step": 154550
    },
    {
      "epoch": 2.529002699828193,
      "grad_norm": 0.04287673532962799,
      "learning_rate": 7.322421560839804e-07,
      "loss": 0.0006,
      "step": 154560
    },
    {
      "epoch": 2.5291663257792685,
      "grad_norm": 0.07711534202098846,
      "learning_rate": 7.317462646872864e-07,
      "loss": 0.0007,
      "step": 154570
    },
    {
      "epoch": 2.5293299517303445,
      "grad_norm": 0.28493911027908325,
      "learning_rate": 7.312505280045301e-07,
      "loss": 0.0006,
      "step": 154580
    },
    {
      "epoch": 2.5294935776814205,
      "grad_norm": 0.050935205072164536,
      "learning_rate": 7.307549460536834e-07,
      "loss": 0.0004,
      "step": 154590
    },
    {
      "epoch": 2.529657203632496,
      "grad_norm": 0.026742778718471527,
      "learning_rate": 7.302595188527095e-07,
      "loss": 0.0006,
      "step": 154600
    },
    {
      "epoch": 2.529820829583572,
      "grad_norm": 0.0814829170703888,
      "learning_rate": 7.297642464195647e-07,
      "loss": 0.0004,
      "step": 154610
    },
    {
      "epoch": 2.529984455534648,
      "grad_norm": 0.1479707509279251,
      "learning_rate": 7.292691287722042e-07,
      "loss": 0.0007,
      "step": 154620
    },
    {
      "epoch": 2.5301480814857236,
      "grad_norm": 0.11793030798435211,
      "learning_rate": 7.287741659285725e-07,
      "loss": 0.0022,
      "step": 154630
    },
    {
      "epoch": 2.5303117074367996,
      "grad_norm": 0.013708998449146748,
      "learning_rate": 7.282793579066128e-07,
      "loss": 0.0009,
      "step": 154640
    },
    {
      "epoch": 2.5304753333878756,
      "grad_norm": 0.05208819732069969,
      "learning_rate": 7.277847047242581e-07,
      "loss": 0.0018,
      "step": 154650
    },
    {
      "epoch": 2.530638959338951,
      "grad_norm": 0.040433481335639954,
      "learning_rate": 7.272902063994408e-07,
      "loss": 0.001,
      "step": 154660
    },
    {
      "epoch": 2.530802585290027,
      "grad_norm": 0.07328460365533829,
      "learning_rate": 7.267958629500849e-07,
      "loss": 0.0008,
      "step": 154670
    },
    {
      "epoch": 2.530966211241103,
      "grad_norm": 0.14707572758197784,
      "learning_rate": 7.26301674394107e-07,
      "loss": 0.002,
      "step": 154680
    },
    {
      "epoch": 2.5311298371921787,
      "grad_norm": 0.05553380027413368,
      "learning_rate": 7.25807640749423e-07,
      "loss": 0.0011,
      "step": 154690
    },
    {
      "epoch": 2.5312934631432547,
      "grad_norm": 0.07082612812519073,
      "learning_rate": 7.253137620339384e-07,
      "loss": 0.0005,
      "step": 154700
    },
    {
      "epoch": 2.53145708909433,
      "grad_norm": 0.013697193004190922,
      "learning_rate": 7.248200382655568e-07,
      "loss": 0.0004,
      "step": 154710
    },
    {
      "epoch": 2.531620715045406,
      "grad_norm": 0.07054273784160614,
      "learning_rate": 7.243264694621727e-07,
      "loss": 0.0007,
      "step": 154720
    },
    {
      "epoch": 2.531784340996482,
      "grad_norm": 0.057278189808130264,
      "learning_rate": 7.238330556416789e-07,
      "loss": 0.0005,
      "step": 154730
    },
    {
      "epoch": 2.5319479669475577,
      "grad_norm": 0.06561379134654999,
      "learning_rate": 7.233397968219585e-07,
      "loss": 0.0017,
      "step": 154740
    },
    {
      "epoch": 2.5321115928986337,
      "grad_norm": 0.1253882348537445,
      "learning_rate": 7.228466930208916e-07,
      "loss": 0.0005,
      "step": 154750
    },
    {
      "epoch": 2.5322752188497093,
      "grad_norm": 0.03089512698352337,
      "learning_rate": 7.223537442563544e-07,
      "loss": 0.0005,
      "step": 154760
    },
    {
      "epoch": 2.5324388448007853,
      "grad_norm": 0.01702384650707245,
      "learning_rate": 7.218609505462104e-07,
      "loss": 0.0007,
      "step": 154770
    },
    {
      "epoch": 2.5326024707518613,
      "grad_norm": 0.18375924229621887,
      "learning_rate": 7.213683119083259e-07,
      "loss": 0.0039,
      "step": 154780
    },
    {
      "epoch": 2.532766096702937,
      "grad_norm": 0.3570653200149536,
      "learning_rate": 7.20875828360556e-07,
      "loss": 0.0009,
      "step": 154790
    },
    {
      "epoch": 2.532929722654013,
      "grad_norm": 0.09150499105453491,
      "learning_rate": 7.203834999207521e-07,
      "loss": 0.0004,
      "step": 154800
    },
    {
      "epoch": 2.533093348605089,
      "grad_norm": 0.03373308852314949,
      "learning_rate": 7.198913266067625e-07,
      "loss": 0.0003,
      "step": 154810
    },
    {
      "epoch": 2.5332569745561644,
      "grad_norm": 0.015074292197823524,
      "learning_rate": 7.193993084364236e-07,
      "loss": 0.0006,
      "step": 154820
    },
    {
      "epoch": 2.5334206005072404,
      "grad_norm": 0.0695846900343895,
      "learning_rate": 7.189074454275729e-07,
      "loss": 0.0008,
      "step": 154830
    },
    {
      "epoch": 2.5335842264583164,
      "grad_norm": 0.09850388020277023,
      "learning_rate": 7.184157375980366e-07,
      "loss": 0.0006,
      "step": 154840
    },
    {
      "epoch": 2.533747852409392,
      "grad_norm": 0.14277240633964539,
      "learning_rate": 7.17924184965641e-07,
      "loss": 0.0005,
      "step": 154850
    },
    {
      "epoch": 2.533911478360468,
      "grad_norm": 0.05869103968143463,
      "learning_rate": 7.174327875482012e-07,
      "loss": 0.0009,
      "step": 154860
    },
    {
      "epoch": 2.534075104311544,
      "grad_norm": 0.0034374427050352097,
      "learning_rate": 7.169415453635292e-07,
      "loss": 0.0004,
      "step": 154870
    },
    {
      "epoch": 2.5342387302626195,
      "grad_norm": 0.10387668013572693,
      "learning_rate": 7.164504584294335e-07,
      "loss": 0.0009,
      "step": 154880
    },
    {
      "epoch": 2.5344023562136955,
      "grad_norm": 0.04663649946451187,
      "learning_rate": 7.15959526763712e-07,
      "loss": 0.0003,
      "step": 154890
    },
    {
      "epoch": 2.5345659821647715,
      "grad_norm": 0.13063158094882965,
      "learning_rate": 7.154687503841629e-07,
      "loss": 0.0012,
      "step": 154900
    },
    {
      "epoch": 2.534729608115847,
      "grad_norm": 0.022338690236210823,
      "learning_rate": 7.149781293085722e-07,
      "loss": 0.0013,
      "step": 154910
    },
    {
      "epoch": 2.534893234066923,
      "grad_norm": 0.040566470474004745,
      "learning_rate": 7.144876635547271e-07,
      "loss": 0.0008,
      "step": 154920
    },
    {
      "epoch": 2.535056860017999,
      "grad_norm": 0.05255088955163956,
      "learning_rate": 7.139973531404026e-07,
      "loss": 0.0007,
      "step": 154930
    },
    {
      "epoch": 2.5352204859690746,
      "grad_norm": 0.37699568271636963,
      "learning_rate": 7.135071980833746e-07,
      "loss": 0.001,
      "step": 154940
    },
    {
      "epoch": 2.5353841119201506,
      "grad_norm": 0.09932459145784378,
      "learning_rate": 7.130171984014072e-07,
      "loss": 0.0005,
      "step": 154950
    },
    {
      "epoch": 2.5355477378712266,
      "grad_norm": 0.03639541193842888,
      "learning_rate": 7.125273541122635e-07,
      "loss": 0.0009,
      "step": 154960
    },
    {
      "epoch": 2.535711363822302,
      "grad_norm": 0.0035783532075583935,
      "learning_rate": 7.120376652336991e-07,
      "loss": 0.0013,
      "step": 154970
    },
    {
      "epoch": 2.535874989773378,
      "grad_norm": 0.12814374268054962,
      "learning_rate": 7.115481317834622e-07,
      "loss": 0.001,
      "step": 154980
    },
    {
      "epoch": 2.536038615724454,
      "grad_norm": 0.0740990862250328,
      "learning_rate": 7.110587537792996e-07,
      "loss": 0.0008,
      "step": 154990
    },
    {
      "epoch": 2.5362022416755297,
      "grad_norm": 0.07794848829507828,
      "learning_rate": 7.105695312389477e-07,
      "loss": 0.0012,
      "step": 155000
    },
    {
      "epoch": 2.5363658676266057,
      "grad_norm": 0.02584184892475605,
      "learning_rate": 7.100804641801417e-07,
      "loss": 0.0002,
      "step": 155010
    },
    {
      "epoch": 2.5365294935776816,
      "grad_norm": 0.037902913987636566,
      "learning_rate": 7.095915526206076e-07,
      "loss": 0.0007,
      "step": 155020
    },
    {
      "epoch": 2.536693119528757,
      "grad_norm": 0.2638149559497833,
      "learning_rate": 7.091027965780683e-07,
      "loss": 0.0008,
      "step": 155030
    },
    {
      "epoch": 2.536856745479833,
      "grad_norm": 0.058093030005693436,
      "learning_rate": 7.086141960702403e-07,
      "loss": 0.0006,
      "step": 155040
    },
    {
      "epoch": 2.537020371430909,
      "grad_norm": 0.09543897211551666,
      "learning_rate": 7.081257511148332e-07,
      "loss": 0.0006,
      "step": 155050
    },
    {
      "epoch": 2.5371839973819847,
      "grad_norm": 0.12779735028743744,
      "learning_rate": 7.07637461729554e-07,
      "loss": 0.0013,
      "step": 155060
    },
    {
      "epoch": 2.5373476233330607,
      "grad_norm": 0.054012127220630646,
      "learning_rate": 7.071493279320984e-07,
      "loss": 0.0007,
      "step": 155070
    },
    {
      "epoch": 2.5375112492841367,
      "grad_norm": 0.07275006920099258,
      "learning_rate": 7.06661349740162e-07,
      "loss": 0.0007,
      "step": 155080
    },
    {
      "epoch": 2.5376748752352123,
      "grad_norm": 0.04402687028050423,
      "learning_rate": 7.061735271714338e-07,
      "loss": 0.0007,
      "step": 155090
    },
    {
      "epoch": 2.5378385011862883,
      "grad_norm": 0.22682365775108337,
      "learning_rate": 7.056858602435945e-07,
      "loss": 0.0011,
      "step": 155100
    },
    {
      "epoch": 2.538002127137364,
      "grad_norm": 0.1548054814338684,
      "learning_rate": 7.051983489743225e-07,
      "loss": 0.0007,
      "step": 155110
    },
    {
      "epoch": 2.53816575308844,
      "grad_norm": 0.11934594810009003,
      "learning_rate": 7.047109933812873e-07,
      "loss": 0.0004,
      "step": 155120
    },
    {
      "epoch": 2.538329379039516,
      "grad_norm": 0.022091029211878777,
      "learning_rate": 7.042237934821561e-07,
      "loss": 0.0004,
      "step": 155130
    },
    {
      "epoch": 2.5384930049905914,
      "grad_norm": 0.01121938694268465,
      "learning_rate": 7.037367492945868e-07,
      "loss": 0.0004,
      "step": 155140
    },
    {
      "epoch": 2.5386566309416674,
      "grad_norm": 0.19864557683467865,
      "learning_rate": 7.032498608362354e-07,
      "loss": 0.0007,
      "step": 155150
    },
    {
      "epoch": 2.538820256892743,
      "grad_norm": 0.008831099607050419,
      "learning_rate": 7.027631281247494e-07,
      "loss": 0.0004,
      "step": 155160
    },
    {
      "epoch": 2.538983882843819,
      "grad_norm": 0.08325351774692535,
      "learning_rate": 7.022765511777712e-07,
      "loss": 0.0016,
      "step": 155170
    },
    {
      "epoch": 2.539147508794895,
      "grad_norm": 0.03260786086320877,
      "learning_rate": 7.017901300129398e-07,
      "loss": 0.0005,
      "step": 155180
    },
    {
      "epoch": 2.5393111347459705,
      "grad_norm": 0.0021589382085949183,
      "learning_rate": 7.013038646478843e-07,
      "loss": 0.0004,
      "step": 155190
    },
    {
      "epoch": 2.5394747606970465,
      "grad_norm": 0.02630332112312317,
      "learning_rate": 7.008177551002332e-07,
      "loss": 0.0008,
      "step": 155200
    },
    {
      "epoch": 2.5396383866481225,
      "grad_norm": 0.010716327466070652,
      "learning_rate": 7.003318013876049e-07,
      "loss": 0.0007,
      "step": 155210
    },
    {
      "epoch": 2.539802012599198,
      "grad_norm": 0.1302867829799652,
      "learning_rate": 6.998460035276155e-07,
      "loss": 0.0014,
      "step": 155220
    },
    {
      "epoch": 2.539965638550274,
      "grad_norm": 0.08095908910036087,
      "learning_rate": 6.993603615378725e-07,
      "loss": 0.0006,
      "step": 155230
    },
    {
      "epoch": 2.54012926450135,
      "grad_norm": 0.023711491376161575,
      "learning_rate": 6.988748754359797e-07,
      "loss": 0.0005,
      "step": 155240
    },
    {
      "epoch": 2.5402928904524256,
      "grad_norm": 0.004862272180616856,
      "learning_rate": 6.983895452395367e-07,
      "loss": 0.0007,
      "step": 155250
    },
    {
      "epoch": 2.5404565164035016,
      "grad_norm": 0.019349688664078712,
      "learning_rate": 6.979043709661337e-07,
      "loss": 0.0007,
      "step": 155260
    },
    {
      "epoch": 2.5406201423545776,
      "grad_norm": 0.010107055306434631,
      "learning_rate": 6.974193526333578e-07,
      "loss": 0.0018,
      "step": 155270
    },
    {
      "epoch": 2.540783768305653,
      "grad_norm": 0.011530616320669651,
      "learning_rate": 6.969344902587883e-07,
      "loss": 0.001,
      "step": 155280
    },
    {
      "epoch": 2.540947394256729,
      "grad_norm": 0.131178081035614,
      "learning_rate": 6.964497838600021e-07,
      "loss": 0.0014,
      "step": 155290
    },
    {
      "epoch": 2.541111020207805,
      "grad_norm": 0.034970127046108246,
      "learning_rate": 6.959652334545669e-07,
      "loss": 0.0003,
      "step": 155300
    },
    {
      "epoch": 2.5412746461588807,
      "grad_norm": 0.019622420892119408,
      "learning_rate": 6.954808390600475e-07,
      "loss": 0.0004,
      "step": 155310
    },
    {
      "epoch": 2.5414382721099567,
      "grad_norm": 0.05171030014753342,
      "learning_rate": 6.949966006940034e-07,
      "loss": 0.0007,
      "step": 155320
    },
    {
      "epoch": 2.5416018980610326,
      "grad_norm": 0.007063876371830702,
      "learning_rate": 6.945125183739848e-07,
      "loss": 0.0007,
      "step": 155330
    },
    {
      "epoch": 2.541765524012108,
      "grad_norm": 0.15509851276874542,
      "learning_rate": 6.940285921175404e-07,
      "loss": 0.0008,
      "step": 155340
    },
    {
      "epoch": 2.541929149963184,
      "grad_norm": 0.05866532027721405,
      "learning_rate": 6.935448219422091e-07,
      "loss": 0.0006,
      "step": 155350
    },
    {
      "epoch": 2.54209277591426,
      "grad_norm": 0.1022576168179512,
      "learning_rate": 6.930612078655285e-07,
      "loss": 0.0007,
      "step": 155360
    },
    {
      "epoch": 2.5422564018653357,
      "grad_norm": 0.12192277610301971,
      "learning_rate": 6.925777499050285e-07,
      "loss": 0.0009,
      "step": 155370
    },
    {
      "epoch": 2.5424200278164117,
      "grad_norm": 0.020920803770422935,
      "learning_rate": 6.920944480782305e-07,
      "loss": 0.0005,
      "step": 155380
    },
    {
      "epoch": 2.5425836537674877,
      "grad_norm": 0.158793643116951,
      "learning_rate": 6.916113024026566e-07,
      "loss": 0.0007,
      "step": 155390
    },
    {
      "epoch": 2.5427472797185633,
      "grad_norm": 0.051074981689453125,
      "learning_rate": 6.911283128958168e-07,
      "loss": 0.0006,
      "step": 155400
    },
    {
      "epoch": 2.5429109056696393,
      "grad_norm": 0.03006868250668049,
      "learning_rate": 6.906454795752204e-07,
      "loss": 0.0011,
      "step": 155410
    },
    {
      "epoch": 2.5430745316207153,
      "grad_norm": 0.0280043575912714,
      "learning_rate": 6.901628024583674e-07,
      "loss": 0.0011,
      "step": 155420
    },
    {
      "epoch": 2.543238157571791,
      "grad_norm": 0.0024522743187844753,
      "learning_rate": 6.896802815627556e-07,
      "loss": 0.0014,
      "step": 155430
    },
    {
      "epoch": 2.543401783522867,
      "grad_norm": 0.11928123980760574,
      "learning_rate": 6.891979169058727e-07,
      "loss": 0.0009,
      "step": 155440
    },
    {
      "epoch": 2.543565409473943,
      "grad_norm": 0.0108345290645957,
      "learning_rate": 6.887157085052054e-07,
      "loss": 0.0015,
      "step": 155450
    },
    {
      "epoch": 2.5437290354250184,
      "grad_norm": 0.015453699044883251,
      "learning_rate": 6.882336563782321e-07,
      "loss": 0.0005,
      "step": 155460
    },
    {
      "epoch": 2.5438926613760944,
      "grad_norm": 0.07692184299230576,
      "learning_rate": 6.877517605424245e-07,
      "loss": 0.0005,
      "step": 155470
    },
    {
      "epoch": 2.54405628732717,
      "grad_norm": 0.01467913668602705,
      "learning_rate": 6.872700210152528e-07,
      "loss": 0.0006,
      "step": 155480
    },
    {
      "epoch": 2.544219913278246,
      "grad_norm": 0.22888760268688202,
      "learning_rate": 6.867884378141759e-07,
      "loss": 0.001,
      "step": 155490
    },
    {
      "epoch": 2.544383539229322,
      "grad_norm": 0.12037533521652222,
      "learning_rate": 6.863070109566528e-07,
      "loss": 0.0007,
      "step": 155500
    },
    {
      "epoch": 2.5445471651803975,
      "grad_norm": 0.0068886131048202515,
      "learning_rate": 6.858257404601315e-07,
      "loss": 0.0008,
      "step": 155510
    },
    {
      "epoch": 2.5447107911314735,
      "grad_norm": 0.027898428961634636,
      "learning_rate": 6.853446263420588e-07,
      "loss": 0.0006,
      "step": 155520
    },
    {
      "epoch": 2.544874417082549,
      "grad_norm": 0.0355159267783165,
      "learning_rate": 6.84863668619874e-07,
      "loss": 0.0009,
      "step": 155530
    },
    {
      "epoch": 2.545038043033625,
      "grad_norm": 0.010494722984731197,
      "learning_rate": 6.843828673110092e-07,
      "loss": 0.0004,
      "step": 155540
    },
    {
      "epoch": 2.545201668984701,
      "grad_norm": 0.04081439971923828,
      "learning_rate": 6.839022224328945e-07,
      "loss": 0.0007,
      "step": 155550
    },
    {
      "epoch": 2.5453652949357766,
      "grad_norm": 0.029591375961899757,
      "learning_rate": 6.834217340029503e-07,
      "loss": 0.0005,
      "step": 155560
    },
    {
      "epoch": 2.5455289208868526,
      "grad_norm": 0.08713199198246002,
      "learning_rate": 6.829414020385938e-07,
      "loss": 0.0008,
      "step": 155570
    },
    {
      "epoch": 2.5456925468379286,
      "grad_norm": 0.01554613746702671,
      "learning_rate": 6.82461226557235e-07,
      "loss": 0.0003,
      "step": 155580
    },
    {
      "epoch": 2.545856172789004,
      "grad_norm": 0.169443279504776,
      "learning_rate": 6.819812075762794e-07,
      "loss": 0.0006,
      "step": 155590
    },
    {
      "epoch": 2.54601979874008,
      "grad_norm": 0.0005545279709622264,
      "learning_rate": 6.815013451131281e-07,
      "loss": 0.0006,
      "step": 155600
    },
    {
      "epoch": 2.546183424691156,
      "grad_norm": 0.019296841695904732,
      "learning_rate": 6.810216391851726e-07,
      "loss": 0.0006,
      "step": 155610
    },
    {
      "epoch": 2.5463470506422317,
      "grad_norm": 0.00811464712023735,
      "learning_rate": 6.805420898098036e-07,
      "loss": 0.0006,
      "step": 155620
    },
    {
      "epoch": 2.5465106765933077,
      "grad_norm": 0.044037241488695145,
      "learning_rate": 6.800626970044011e-07,
      "loss": 0.0011,
      "step": 155630
    },
    {
      "epoch": 2.5466743025443837,
      "grad_norm": 0.2193140834569931,
      "learning_rate": 6.795834607863444e-07,
      "loss": 0.0011,
      "step": 155640
    },
    {
      "epoch": 2.546837928495459,
      "grad_norm": 0.04491673782467842,
      "learning_rate": 6.791043811730019e-07,
      "loss": 0.0008,
      "step": 155650
    },
    {
      "epoch": 2.547001554446535,
      "grad_norm": 0.09423117339611053,
      "learning_rate": 6.78625458181742e-07,
      "loss": 0.0007,
      "step": 155660
    },
    {
      "epoch": 2.547165180397611,
      "grad_norm": 0.07031507045030594,
      "learning_rate": 6.781466918299229e-07,
      "loss": 0.0004,
      "step": 155670
    },
    {
      "epoch": 2.5473288063486867,
      "grad_norm": 0.09583548456430435,
      "learning_rate": 6.77668082134898e-07,
      "loss": 0.0007,
      "step": 155680
    },
    {
      "epoch": 2.5474924322997627,
      "grad_norm": 0.12722645699977875,
      "learning_rate": 6.771896291140173e-07,
      "loss": 0.0005,
      "step": 155690
    },
    {
      "epoch": 2.5476560582508387,
      "grad_norm": 0.037257205694913864,
      "learning_rate": 6.767113327846225e-07,
      "loss": 0.0009,
      "step": 155700
    },
    {
      "epoch": 2.5478196842019143,
      "grad_norm": 0.1378713995218277,
      "learning_rate": 6.76233193164052e-07,
      "loss": 0.0007,
      "step": 155710
    },
    {
      "epoch": 2.5479833101529903,
      "grad_norm": 0.002491287887096405,
      "learning_rate": 6.757552102696352e-07,
      "loss": 0.0004,
      "step": 155720
    },
    {
      "epoch": 2.5481469361040663,
      "grad_norm": 0.07962657511234283,
      "learning_rate": 6.752773841187e-07,
      "loss": 0.0009,
      "step": 155730
    },
    {
      "epoch": 2.548310562055142,
      "grad_norm": 0.22309452295303345,
      "learning_rate": 6.747997147285645e-07,
      "loss": 0.0008,
      "step": 155740
    },
    {
      "epoch": 2.548474188006218,
      "grad_norm": 0.08748647570610046,
      "learning_rate": 6.743222021165441e-07,
      "loss": 0.0013,
      "step": 155750
    },
    {
      "epoch": 2.548637813957294,
      "grad_norm": 0.038402657955884933,
      "learning_rate": 6.738448462999491e-07,
      "loss": 0.0005,
      "step": 155760
    },
    {
      "epoch": 2.5488014399083694,
      "grad_norm": 0.08433476835489273,
      "learning_rate": 6.733676472960787e-07,
      "loss": 0.0007,
      "step": 155770
    },
    {
      "epoch": 2.5489650658594454,
      "grad_norm": 0.10695762932300568,
      "learning_rate": 6.728906051222339e-07,
      "loss": 0.0008,
      "step": 155780
    },
    {
      "epoch": 2.5491286918105214,
      "grad_norm": 0.11408143490552902,
      "learning_rate": 6.724137197957031e-07,
      "loss": 0.0009,
      "step": 155790
    },
    {
      "epoch": 2.549292317761597,
      "grad_norm": 0.007140147965401411,
      "learning_rate": 6.719369913337742e-07,
      "loss": 0.0005,
      "step": 155800
    },
    {
      "epoch": 2.549455943712673,
      "grad_norm": 0.0691949725151062,
      "learning_rate": 6.714604197537283e-07,
      "loss": 0.0007,
      "step": 155810
    },
    {
      "epoch": 2.549619569663749,
      "grad_norm": 0.10916735231876373,
      "learning_rate": 6.709840050728372e-07,
      "loss": 0.0004,
      "step": 155820
    },
    {
      "epoch": 2.5497831956148245,
      "grad_norm": 0.06625304371118546,
      "learning_rate": 6.705077473083732e-07,
      "loss": 0.0005,
      "step": 155830
    },
    {
      "epoch": 2.5499468215659005,
      "grad_norm": 0.026830971240997314,
      "learning_rate": 6.700316464775958e-07,
      "loss": 0.0012,
      "step": 155840
    },
    {
      "epoch": 2.5501104475169765,
      "grad_norm": 0.10253141820430756,
      "learning_rate": 6.695557025977661e-07,
      "loss": 0.0006,
      "step": 155850
    },
    {
      "epoch": 2.550274073468052,
      "grad_norm": 0.028377173468470573,
      "learning_rate": 6.690799156861333e-07,
      "loss": 0.0009,
      "step": 155860
    },
    {
      "epoch": 2.550437699419128,
      "grad_norm": 0.08253173530101776,
      "learning_rate": 6.686042857599451e-07,
      "loss": 0.0009,
      "step": 155870
    },
    {
      "epoch": 2.5506013253702036,
      "grad_norm": 0.11206453293561935,
      "learning_rate": 6.681288128364416e-07,
      "loss": 0.0006,
      "step": 155880
    },
    {
      "epoch": 2.5507649513212796,
      "grad_norm": 0.004035826772451401,
      "learning_rate": 6.676534969328558e-07,
      "loss": 0.0003,
      "step": 155890
    },
    {
      "epoch": 2.5509285772723556,
      "grad_norm": 0.013340151868760586,
      "learning_rate": 6.671783380664193e-07,
      "loss": 0.0006,
      "step": 155900
    },
    {
      "epoch": 2.551092203223431,
      "grad_norm": 0.11304190754890442,
      "learning_rate": 6.667033362543534e-07,
      "loss": 0.0007,
      "step": 155910
    },
    {
      "epoch": 2.551255829174507,
      "grad_norm": 0.02534390054643154,
      "learning_rate": 6.662284915138784e-07,
      "loss": 0.0008,
      "step": 155920
    },
    {
      "epoch": 2.5514194551255827,
      "grad_norm": 0.0638873279094696,
      "learning_rate": 6.65753803862203e-07,
      "loss": 0.0003,
      "step": 155930
    },
    {
      "epoch": 2.5515830810766587,
      "grad_norm": 0.05399458110332489,
      "learning_rate": 6.652792733165364e-07,
      "loss": 0.0008,
      "step": 155940
    },
    {
      "epoch": 2.5517467070277347,
      "grad_norm": 0.05092838034033775,
      "learning_rate": 6.648048998940765e-07,
      "loss": 0.0006,
      "step": 155950
    },
    {
      "epoch": 2.55191033297881,
      "grad_norm": 0.299388587474823,
      "learning_rate": 6.6433068361202e-07,
      "loss": 0.002,
      "step": 155960
    },
    {
      "epoch": 2.552073958929886,
      "grad_norm": 0.01573743298649788,
      "learning_rate": 6.638566244875577e-07,
      "loss": 0.0005,
      "step": 155970
    },
    {
      "epoch": 2.552237584880962,
      "grad_norm": 0.0637495219707489,
      "learning_rate": 6.633827225378692e-07,
      "loss": 0.0007,
      "step": 155980
    },
    {
      "epoch": 2.5524012108320377,
      "grad_norm": 0.020792296156287193,
      "learning_rate": 6.629089777801351e-07,
      "loss": 0.0009,
      "step": 155990
    },
    {
      "epoch": 2.5525648367831137,
      "grad_norm": 0.07451620697975159,
      "learning_rate": 6.624353902315251e-07,
      "loss": 0.001,
      "step": 156000
    },
    {
      "epoch": 2.5527284627341897,
      "grad_norm": 0.21134567260742188,
      "learning_rate": 6.619619599092075e-07,
      "loss": 0.0015,
      "step": 156010
    },
    {
      "epoch": 2.5528920886852653,
      "grad_norm": 0.05754583701491356,
      "learning_rate": 6.614886868303438e-07,
      "loss": 0.0005,
      "step": 156020
    },
    {
      "epoch": 2.5530557146363413,
      "grad_norm": 0.07743552327156067,
      "learning_rate": 6.610155710120869e-07,
      "loss": 0.0005,
      "step": 156030
    },
    {
      "epoch": 2.5532193405874173,
      "grad_norm": 0.007690403610467911,
      "learning_rate": 6.60542612471588e-07,
      "loss": 0.0005,
      "step": 156040
    },
    {
      "epoch": 2.553382966538493,
      "grad_norm": 0.13462114334106445,
      "learning_rate": 6.60069811225989e-07,
      "loss": 0.0016,
      "step": 156050
    },
    {
      "epoch": 2.553546592489569,
      "grad_norm": 0.08158423006534576,
      "learning_rate": 6.595971672924295e-07,
      "loss": 0.0008,
      "step": 156060
    },
    {
      "epoch": 2.553710218440645,
      "grad_norm": 0.10743445158004761,
      "learning_rate": 6.591246806880408e-07,
      "loss": 0.0012,
      "step": 156070
    },
    {
      "epoch": 2.5538738443917204,
      "grad_norm": 0.12791578471660614,
      "learning_rate": 6.586523514299481e-07,
      "loss": 0.0017,
      "step": 156080
    },
    {
      "epoch": 2.5540374703427964,
      "grad_norm": 0.04703845456242561,
      "learning_rate": 6.58180179535275e-07,
      "loss": 0.0008,
      "step": 156090
    },
    {
      "epoch": 2.5542010962938724,
      "grad_norm": 0.10353540629148483,
      "learning_rate": 6.577081650211336e-07,
      "loss": 0.001,
      "step": 156100
    },
    {
      "epoch": 2.554364722244948,
      "grad_norm": 0.009049639105796814,
      "learning_rate": 6.572363079046357e-07,
      "loss": 0.0006,
      "step": 156110
    },
    {
      "epoch": 2.554528348196024,
      "grad_norm": 0.023534933105111122,
      "learning_rate": 6.567646082028834e-07,
      "loss": 0.0007,
      "step": 156120
    },
    {
      "epoch": 2.5546919741471,
      "grad_norm": 0.06436805427074432,
      "learning_rate": 6.562930659329764e-07,
      "loss": 0.0008,
      "step": 156130
    },
    {
      "epoch": 2.5548556000981755,
      "grad_norm": 0.005399988032877445,
      "learning_rate": 6.55821681112005e-07,
      "loss": 0.0003,
      "step": 156140
    },
    {
      "epoch": 2.5550192260492515,
      "grad_norm": 0.07851773500442505,
      "learning_rate": 6.553504537570577e-07,
      "loss": 0.0009,
      "step": 156150
    },
    {
      "epoch": 2.5551828520003275,
      "grad_norm": 0.08111679553985596,
      "learning_rate": 6.54879383885213e-07,
      "loss": 0.001,
      "step": 156160
    },
    {
      "epoch": 2.555346477951403,
      "grad_norm": 0.08843546360731125,
      "learning_rate": 6.54408471513549e-07,
      "loss": 0.0024,
      "step": 156170
    },
    {
      "epoch": 2.555510103902479,
      "grad_norm": 0.006365268956869841,
      "learning_rate": 6.539377166591332e-07,
      "loss": 0.0011,
      "step": 156180
    },
    {
      "epoch": 2.555673729853555,
      "grad_norm": 0.10550446808338165,
      "learning_rate": 6.534671193390286e-07,
      "loss": 0.0006,
      "step": 156190
    },
    {
      "epoch": 2.5558373558046306,
      "grad_norm": 0.03861622512340546,
      "learning_rate": 6.529966795702953e-07,
      "loss": 0.0011,
      "step": 156200
    },
    {
      "epoch": 2.5560009817557066,
      "grad_norm": 0.12708547711372375,
      "learning_rate": 6.525263973699836e-07,
      "loss": 0.0007,
      "step": 156210
    },
    {
      "epoch": 2.5561646077067826,
      "grad_norm": 0.015784651041030884,
      "learning_rate": 6.520562727551422e-07,
      "loss": 0.0005,
      "step": 156220
    },
    {
      "epoch": 2.556328233657858,
      "grad_norm": 0.005091626662760973,
      "learning_rate": 6.515863057428101e-07,
      "loss": 0.0009,
      "step": 156230
    },
    {
      "epoch": 2.556491859608934,
      "grad_norm": 0.044475264847278595,
      "learning_rate": 6.511164963500233e-07,
      "loss": 0.0008,
      "step": 156240
    },
    {
      "epoch": 2.5566554855600097,
      "grad_norm": 0.12598271667957306,
      "learning_rate": 6.506468445938119e-07,
      "loss": 0.0019,
      "step": 156250
    },
    {
      "epoch": 2.5568191115110857,
      "grad_norm": 0.06293056905269623,
      "learning_rate": 6.501773504911979e-07,
      "loss": 0.0006,
      "step": 156260
    },
    {
      "epoch": 2.5569827374621616,
      "grad_norm": 0.139814555644989,
      "learning_rate": 6.497080140592027e-07,
      "loss": 0.0007,
      "step": 156270
    },
    {
      "epoch": 2.557146363413237,
      "grad_norm": 0.04048897698521614,
      "learning_rate": 6.492388353148343e-07,
      "loss": 0.0005,
      "step": 156280
    },
    {
      "epoch": 2.557309989364313,
      "grad_norm": 0.23403649032115936,
      "learning_rate": 6.487698142751014e-07,
      "loss": 0.0009,
      "step": 156290
    },
    {
      "epoch": 2.5574736153153887,
      "grad_norm": 0.043612074106931686,
      "learning_rate": 6.483009509570059e-07,
      "loss": 0.001,
      "step": 156300
    },
    {
      "epoch": 2.5576372412664647,
      "grad_norm": 0.0636618435382843,
      "learning_rate": 6.478322453775409e-07,
      "loss": 0.0004,
      "step": 156310
    },
    {
      "epoch": 2.5578008672175407,
      "grad_norm": 0.05871919170022011,
      "learning_rate": 6.473636975536979e-07,
      "loss": 0.0009,
      "step": 156320
    },
    {
      "epoch": 2.5579644931686163,
      "grad_norm": 0.023224275559186935,
      "learning_rate": 6.468953075024581e-07,
      "loss": 0.0005,
      "step": 156330
    },
    {
      "epoch": 2.5581281191196923,
      "grad_norm": 0.1109846830368042,
      "learning_rate": 6.464270752408025e-07,
      "loss": 0.0009,
      "step": 156340
    },
    {
      "epoch": 2.5582917450707683,
      "grad_norm": 0.04998786747455597,
      "learning_rate": 6.459590007857003e-07,
      "loss": 0.0004,
      "step": 156350
    },
    {
      "epoch": 2.558455371021844,
      "grad_norm": 0.004235696978867054,
      "learning_rate": 6.454910841541212e-07,
      "loss": 0.0006,
      "step": 156360
    },
    {
      "epoch": 2.55861899697292,
      "grad_norm": 0.07944649457931519,
      "learning_rate": 6.450233253630244e-07,
      "loss": 0.0004,
      "step": 156370
    },
    {
      "epoch": 2.558782622923996,
      "grad_norm": 0.06845125555992126,
      "learning_rate": 6.445557244293638e-07,
      "loss": 0.0003,
      "step": 156380
    },
    {
      "epoch": 2.5589462488750714,
      "grad_norm": 0.007240402512252331,
      "learning_rate": 6.440882813700916e-07,
      "loss": 0.0006,
      "step": 156390
    },
    {
      "epoch": 2.5591098748261474,
      "grad_norm": 0.06601081788539886,
      "learning_rate": 6.436209962021489e-07,
      "loss": 0.0005,
      "step": 156400
    },
    {
      "epoch": 2.5592735007772234,
      "grad_norm": 0.08220779150724411,
      "learning_rate": 6.43153868942476e-07,
      "loss": 0.0005,
      "step": 156410
    },
    {
      "epoch": 2.559437126728299,
      "grad_norm": 0.015207826159894466,
      "learning_rate": 6.426868996080027e-07,
      "loss": 0.0009,
      "step": 156420
    },
    {
      "epoch": 2.559600752679375,
      "grad_norm": 0.16617946326732635,
      "learning_rate": 6.422200882156577e-07,
      "loss": 0.0013,
      "step": 156430
    },
    {
      "epoch": 2.559764378630451,
      "grad_norm": 0.0746876522898674,
      "learning_rate": 6.417534347823601e-07,
      "loss": 0.0005,
      "step": 156440
    },
    {
      "epoch": 2.5599280045815265,
      "grad_norm": 0.24322889745235443,
      "learning_rate": 6.412869393250253e-07,
      "loss": 0.0011,
      "step": 156450
    },
    {
      "epoch": 2.5600916305326025,
      "grad_norm": 0.07840018719434738,
      "learning_rate": 6.408206018605645e-07,
      "loss": 0.0011,
      "step": 156460
    },
    {
      "epoch": 2.5602552564836785,
      "grad_norm": 0.17892731726169586,
      "learning_rate": 6.403544224058794e-07,
      "loss": 0.0008,
      "step": 156470
    },
    {
      "epoch": 2.560418882434754,
      "grad_norm": 0.14070363342761993,
      "learning_rate": 6.398884009778689e-07,
      "loss": 0.0004,
      "step": 156480
    },
    {
      "epoch": 2.56058250838583,
      "grad_norm": 0.07920346409082413,
      "learning_rate": 6.394225375934232e-07,
      "loss": 0.0004,
      "step": 156490
    },
    {
      "epoch": 2.560746134336906,
      "grad_norm": 0.11343622952699661,
      "learning_rate": 6.389568322694317e-07,
      "loss": 0.0008,
      "step": 156500
    },
    {
      "epoch": 2.5609097602879816,
      "grad_norm": 0.0613960400223732,
      "learning_rate": 6.38491285022772e-07,
      "loss": 0.0005,
      "step": 156510
    },
    {
      "epoch": 2.5610733862390576,
      "grad_norm": 0.04808366298675537,
      "learning_rate": 6.380258958703206e-07,
      "loss": 0.0008,
      "step": 156520
    },
    {
      "epoch": 2.5612370121901336,
      "grad_norm": 0.0035386946983635426,
      "learning_rate": 6.375606648289478e-07,
      "loss": 0.0008,
      "step": 156530
    },
    {
      "epoch": 2.561400638141209,
      "grad_norm": 0.09408704936504364,
      "learning_rate": 6.370955919155152e-07,
      "loss": 0.0011,
      "step": 156540
    },
    {
      "epoch": 2.561564264092285,
      "grad_norm": 0.0018898221896961331,
      "learning_rate": 6.366306771468822e-07,
      "loss": 0.0014,
      "step": 156550
    },
    {
      "epoch": 2.561727890043361,
      "grad_norm": 0.14769554138183594,
      "learning_rate": 6.361659205398995e-07,
      "loss": 0.0004,
      "step": 156560
    },
    {
      "epoch": 2.5618915159944367,
      "grad_norm": 0.0028517222963273525,
      "learning_rate": 6.35701322111415e-07,
      "loss": 0.0008,
      "step": 156570
    },
    {
      "epoch": 2.5620551419455126,
      "grad_norm": 0.04328923672437668,
      "learning_rate": 6.352368818782678e-07,
      "loss": 0.0008,
      "step": 156580
    },
    {
      "epoch": 2.5622187678965886,
      "grad_norm": 0.056838445365428925,
      "learning_rate": 6.347725998572924e-07,
      "loss": 0.0007,
      "step": 156590
    },
    {
      "epoch": 2.562382393847664,
      "grad_norm": 0.017132185399532318,
      "learning_rate": 6.343084760653195e-07,
      "loss": 0.0004,
      "step": 156600
    },
    {
      "epoch": 2.56254601979874,
      "grad_norm": 0.07038655877113342,
      "learning_rate": 6.338445105191709e-07,
      "loss": 0.0013,
      "step": 156610
    },
    {
      "epoch": 2.562709645749816,
      "grad_norm": 0.11558183282613754,
      "learning_rate": 6.333807032356659e-07,
      "loss": 0.0003,
      "step": 156620
    },
    {
      "epoch": 2.5628732717008917,
      "grad_norm": 0.14009438455104828,
      "learning_rate": 6.329170542316148e-07,
      "loss": 0.0008,
      "step": 156630
    },
    {
      "epoch": 2.5630368976519677,
      "grad_norm": 0.1374894678592682,
      "learning_rate": 6.324535635238255e-07,
      "loss": 0.0009,
      "step": 156640
    },
    {
      "epoch": 2.5632005236030433,
      "grad_norm": 0.04323675483465195,
      "learning_rate": 6.31990231129096e-07,
      "loss": 0.0006,
      "step": 156650
    },
    {
      "epoch": 2.5633641495541193,
      "grad_norm": 0.004057818092405796,
      "learning_rate": 6.315270570642234e-07,
      "loss": 0.0006,
      "step": 156660
    },
    {
      "epoch": 2.5635277755051953,
      "grad_norm": 0.022212103009223938,
      "learning_rate": 6.310640413459962e-07,
      "loss": 0.0006,
      "step": 156670
    },
    {
      "epoch": 2.563691401456271,
      "grad_norm": 0.0089713204652071,
      "learning_rate": 6.306011839911952e-07,
      "loss": 0.0008,
      "step": 156680
    },
    {
      "epoch": 2.563855027407347,
      "grad_norm": 0.08577723056077957,
      "learning_rate": 6.301384850166015e-07,
      "loss": 0.0005,
      "step": 156690
    },
    {
      "epoch": 2.5640186533584224,
      "grad_norm": 0.09871995449066162,
      "learning_rate": 6.296759444389833e-07,
      "loss": 0.0008,
      "step": 156700
    },
    {
      "epoch": 2.5641822793094984,
      "grad_norm": 0.046393319964408875,
      "learning_rate": 6.2921356227511e-07,
      "loss": 0.0009,
      "step": 156710
    },
    {
      "epoch": 2.5643459052605744,
      "grad_norm": 0.019942471757531166,
      "learning_rate": 6.287513385417388e-07,
      "loss": 0.0006,
      "step": 156720
    },
    {
      "epoch": 2.56450953121165,
      "grad_norm": 0.02362806536257267,
      "learning_rate": 6.282892732556256e-07,
      "loss": 0.0011,
      "step": 156730
    },
    {
      "epoch": 2.564673157162726,
      "grad_norm": 0.10480473190546036,
      "learning_rate": 6.278273664335199e-07,
      "loss": 0.001,
      "step": 156740
    },
    {
      "epoch": 2.564836783113802,
      "grad_norm": 0.02225787192583084,
      "learning_rate": 6.273656180921634e-07,
      "loss": 0.0011,
      "step": 156750
    },
    {
      "epoch": 2.5650004090648775,
      "grad_norm": 0.07879892736673355,
      "learning_rate": 6.269040282482947e-07,
      "loss": 0.0007,
      "step": 156760
    },
    {
      "epoch": 2.5651640350159535,
      "grad_norm": 0.13206292688846588,
      "learning_rate": 6.264425969186444e-07,
      "loss": 0.0009,
      "step": 156770
    },
    {
      "epoch": 2.5653276609670295,
      "grad_norm": 0.028706125915050507,
      "learning_rate": 6.259813241199386e-07,
      "loss": 0.0006,
      "step": 156780
    },
    {
      "epoch": 2.565491286918105,
      "grad_norm": 0.02318262681365013,
      "learning_rate": 6.255202098688956e-07,
      "loss": 0.0006,
      "step": 156790
    },
    {
      "epoch": 2.565654912869181,
      "grad_norm": 0.04380005970597267,
      "learning_rate": 6.25059254182232e-07,
      "loss": 0.0005,
      "step": 156800
    },
    {
      "epoch": 2.565818538820257,
      "grad_norm": 0.04346119612455368,
      "learning_rate": 6.245984570766561e-07,
      "loss": 0.0008,
      "step": 156810
    },
    {
      "epoch": 2.5659821647713326,
      "grad_norm": 0.05704323202371597,
      "learning_rate": 6.24137818568869e-07,
      "loss": 0.0011,
      "step": 156820
    },
    {
      "epoch": 2.5661457907224086,
      "grad_norm": 0.0700591504573822,
      "learning_rate": 6.236773386755701e-07,
      "loss": 0.0005,
      "step": 156830
    },
    {
      "epoch": 2.5663094166734846,
      "grad_norm": 0.09425564855337143,
      "learning_rate": 6.232170174134483e-07,
      "loss": 0.0009,
      "step": 156840
    },
    {
      "epoch": 2.56647304262456,
      "grad_norm": 0.10713460296392441,
      "learning_rate": 6.227568547991913e-07,
      "loss": 0.0006,
      "step": 156850
    },
    {
      "epoch": 2.566636668575636,
      "grad_norm": 0.022373080253601074,
      "learning_rate": 6.22296850849477e-07,
      "loss": 0.0002,
      "step": 156860
    },
    {
      "epoch": 2.566800294526712,
      "grad_norm": 0.047354452311992645,
      "learning_rate": 6.218370055809814e-07,
      "loss": 0.0004,
      "step": 156870
    },
    {
      "epoch": 2.5669639204777877,
      "grad_norm": 0.1322072148323059,
      "learning_rate": 6.213773190103717e-07,
      "loss": 0.0015,
      "step": 156880
    },
    {
      "epoch": 2.5671275464288636,
      "grad_norm": 0.052421461790800095,
      "learning_rate": 6.209177911543091e-07,
      "loss": 0.0006,
      "step": 156890
    },
    {
      "epoch": 2.5672911723799396,
      "grad_norm": 0.09074535965919495,
      "learning_rate": 6.204584220294529e-07,
      "loss": 0.0007,
      "step": 156900
    },
    {
      "epoch": 2.567454798331015,
      "grad_norm": 0.005451531615108252,
      "learning_rate": 6.199992116524522e-07,
      "loss": 0.0005,
      "step": 156910
    },
    {
      "epoch": 2.567618424282091,
      "grad_norm": 0.1349174529314041,
      "learning_rate": 6.195401600399536e-07,
      "loss": 0.001,
      "step": 156920
    },
    {
      "epoch": 2.567782050233167,
      "grad_norm": 0.06312589347362518,
      "learning_rate": 6.190812672085955e-07,
      "loss": 0.0008,
      "step": 156930
    },
    {
      "epoch": 2.5679456761842427,
      "grad_norm": 0.04968241974711418,
      "learning_rate": 6.186225331750129e-07,
      "loss": 0.0008,
      "step": 156940
    },
    {
      "epoch": 2.5681093021353187,
      "grad_norm": 0.07449182122945786,
      "learning_rate": 6.181639579558324e-07,
      "loss": 0.0009,
      "step": 156950
    },
    {
      "epoch": 2.5682729280863947,
      "grad_norm": 0.041133783757686615,
      "learning_rate": 6.177055415676769e-07,
      "loss": 0.0009,
      "step": 156960
    },
    {
      "epoch": 2.5684365540374703,
      "grad_norm": 0.028303591534495354,
      "learning_rate": 6.172472840271649e-07,
      "loss": 0.0004,
      "step": 156970
    },
    {
      "epoch": 2.5686001799885463,
      "grad_norm": 0.0668925791978836,
      "learning_rate": 6.167891853509028e-07,
      "loss": 0.0006,
      "step": 156980
    },
    {
      "epoch": 2.5687638059396223,
      "grad_norm": 0.06688759475946426,
      "learning_rate": 6.163312455554993e-07,
      "loss": 0.0009,
      "step": 156990
    },
    {
      "epoch": 2.568927431890698,
      "grad_norm": 0.1382204294204712,
      "learning_rate": 6.158734646575515e-07,
      "loss": 0.0005,
      "step": 157000
    },
    {
      "epoch": 2.569091057841774,
      "grad_norm": 0.014015398919582367,
      "learning_rate": 6.154158426736529e-07,
      "loss": 0.0004,
      "step": 157010
    },
    {
      "epoch": 2.5692546837928494,
      "grad_norm": 0.013954179361462593,
      "learning_rate": 6.149583796203934e-07,
      "loss": 0.0007,
      "step": 157020
    },
    {
      "epoch": 2.5694183097439254,
      "grad_norm": 0.11087533831596375,
      "learning_rate": 6.145010755143527e-07,
      "loss": 0.0007,
      "step": 157030
    },
    {
      "epoch": 2.5695819356950014,
      "grad_norm": 0.044837746769189835,
      "learning_rate": 6.140439303721085e-07,
      "loss": 0.0005,
      "step": 157040
    },
    {
      "epoch": 2.569745561646077,
      "grad_norm": 0.057757701724767685,
      "learning_rate": 6.135869442102294e-07,
      "loss": 0.0007,
      "step": 157050
    },
    {
      "epoch": 2.569909187597153,
      "grad_norm": 0.05135273188352585,
      "learning_rate": 6.131301170452825e-07,
      "loss": 0.0008,
      "step": 157060
    },
    {
      "epoch": 2.5700728135482285,
      "grad_norm": 0.13656525313854218,
      "learning_rate": 6.126734488938246e-07,
      "loss": 0.0004,
      "step": 157070
    },
    {
      "epoch": 2.5702364394993045,
      "grad_norm": 0.01967809908092022,
      "learning_rate": 6.122169397724087e-07,
      "loss": 0.0003,
      "step": 157080
    },
    {
      "epoch": 2.5704000654503805,
      "grad_norm": 0.027018357068300247,
      "learning_rate": 6.117605896975842e-07,
      "loss": 0.001,
      "step": 157090
    },
    {
      "epoch": 2.570563691401456,
      "grad_norm": 0.12258586287498474,
      "learning_rate": 6.113043986858908e-07,
      "loss": 0.0007,
      "step": 157100
    },
    {
      "epoch": 2.570727317352532,
      "grad_norm": 0.03481500595808029,
      "learning_rate": 6.108483667538656e-07,
      "loss": 0.0005,
      "step": 157110
    },
    {
      "epoch": 2.570890943303608,
      "grad_norm": 0.1512630730867386,
      "learning_rate": 6.103924939180366e-07,
      "loss": 0.0015,
      "step": 157120
    },
    {
      "epoch": 2.5710545692546836,
      "grad_norm": 0.03388206288218498,
      "learning_rate": 6.099367801949313e-07,
      "loss": 0.0004,
      "step": 157130
    },
    {
      "epoch": 2.5712181952057596,
      "grad_norm": 0.0012828354956582189,
      "learning_rate": 6.094812256010652e-07,
      "loss": 0.0007,
      "step": 157140
    },
    {
      "epoch": 2.5713818211568356,
      "grad_norm": 0.1583210676908493,
      "learning_rate": 6.090258301529534e-07,
      "loss": 0.0009,
      "step": 157150
    },
    {
      "epoch": 2.571545447107911,
      "grad_norm": 0.06827402859926224,
      "learning_rate": 6.085705938671005e-07,
      "loss": 0.0006,
      "step": 157160
    },
    {
      "epoch": 2.571709073058987,
      "grad_norm": 0.035827767103910446,
      "learning_rate": 6.081155167600106e-07,
      "loss": 0.0004,
      "step": 157170
    },
    {
      "epoch": 2.571872699010063,
      "grad_norm": 0.04035130888223648,
      "learning_rate": 6.076605988481771e-07,
      "loss": 0.0007,
      "step": 157180
    },
    {
      "epoch": 2.5720363249611387,
      "grad_norm": 0.10407208651304245,
      "learning_rate": 6.072058401480896e-07,
      "loss": 0.0006,
      "step": 157190
    },
    {
      "epoch": 2.5721999509122146,
      "grad_norm": 0.051525264978408813,
      "learning_rate": 6.067512406762333e-07,
      "loss": 0.0011,
      "step": 157200
    },
    {
      "epoch": 2.5723635768632906,
      "grad_norm": 0.15741367638111115,
      "learning_rate": 6.062968004490849e-07,
      "loss": 0.0005,
      "step": 157210
    },
    {
      "epoch": 2.572527202814366,
      "grad_norm": 0.012779264710843563,
      "learning_rate": 6.058425194831175e-07,
      "loss": 0.0004,
      "step": 157220
    },
    {
      "epoch": 2.572690828765442,
      "grad_norm": 0.016114961355924606,
      "learning_rate": 6.053883977947983e-07,
      "loss": 0.0011,
      "step": 157230
    },
    {
      "epoch": 2.572854454716518,
      "grad_norm": 0.13488760590553284,
      "learning_rate": 6.049344354005864e-07,
      "loss": 0.0007,
      "step": 157240
    },
    {
      "epoch": 2.5730180806675937,
      "grad_norm": 0.31687450408935547,
      "learning_rate": 6.044806323169394e-07,
      "loss": 0.0026,
      "step": 157250
    },
    {
      "epoch": 2.5731817066186697,
      "grad_norm": 0.03509585186839104,
      "learning_rate": 6.04026988560304e-07,
      "loss": 0.0012,
      "step": 157260
    },
    {
      "epoch": 2.5733453325697457,
      "grad_norm": 0.036281418055295944,
      "learning_rate": 6.035735041471269e-07,
      "loss": 0.0007,
      "step": 157270
    },
    {
      "epoch": 2.5735089585208213,
      "grad_norm": 0.03477535396814346,
      "learning_rate": 6.031201790938418e-07,
      "loss": 0.0005,
      "step": 157280
    },
    {
      "epoch": 2.5736725844718973,
      "grad_norm": 0.13433320820331573,
      "learning_rate": 6.026670134168821e-07,
      "loss": 0.0013,
      "step": 157290
    },
    {
      "epoch": 2.5738362104229733,
      "grad_norm": 0.02297845110297203,
      "learning_rate": 6.02214007132676e-07,
      "loss": 0.0006,
      "step": 157300
    },
    {
      "epoch": 2.573999836374049,
      "grad_norm": 0.15281973779201508,
      "learning_rate": 6.017611602576412e-07,
      "loss": 0.0008,
      "step": 157310
    },
    {
      "epoch": 2.574163462325125,
      "grad_norm": 0.06785403937101364,
      "learning_rate": 6.013084728081941e-07,
      "loss": 0.0006,
      "step": 157320
    },
    {
      "epoch": 2.574327088276201,
      "grad_norm": 0.10274849086999893,
      "learning_rate": 6.008559448007423e-07,
      "loss": 0.0009,
      "step": 157330
    },
    {
      "epoch": 2.5744907142272764,
      "grad_norm": 0.055550601333379745,
      "learning_rate": 6.004035762516908e-07,
      "loss": 0.0008,
      "step": 157340
    },
    {
      "epoch": 2.5746543401783524,
      "grad_norm": 0.04932958632707596,
      "learning_rate": 5.999513671774338e-07,
      "loss": 0.0008,
      "step": 157350
    },
    {
      "epoch": 2.5748179661294284,
      "grad_norm": 0.1966402232646942,
      "learning_rate": 5.994993175943659e-07,
      "loss": 0.001,
      "step": 157360
    },
    {
      "epoch": 2.574981592080504,
      "grad_norm": 0.03227045387029648,
      "learning_rate": 5.990474275188712e-07,
      "loss": 0.0004,
      "step": 157370
    },
    {
      "epoch": 2.57514521803158,
      "grad_norm": 0.02620607055723667,
      "learning_rate": 5.985956969673285e-07,
      "loss": 0.0013,
      "step": 157380
    },
    {
      "epoch": 2.575308843982656,
      "grad_norm": 0.0728297084569931,
      "learning_rate": 5.981441259561149e-07,
      "loss": 0.0005,
      "step": 157390
    },
    {
      "epoch": 2.5754724699337315,
      "grad_norm": 0.04984407126903534,
      "learning_rate": 5.976927145015959e-07,
      "loss": 0.0008,
      "step": 157400
    },
    {
      "epoch": 2.5756360958848075,
      "grad_norm": 0.04106207564473152,
      "learning_rate": 5.972414626201361e-07,
      "loss": 0.0004,
      "step": 157410
    },
    {
      "epoch": 2.575799721835883,
      "grad_norm": 0.06779506057500839,
      "learning_rate": 5.967903703280909e-07,
      "loss": 0.0006,
      "step": 157420
    },
    {
      "epoch": 2.575963347786959,
      "grad_norm": 0.015237879939377308,
      "learning_rate": 5.96339437641813e-07,
      "loss": 0.0005,
      "step": 157430
    },
    {
      "epoch": 2.576126973738035,
      "grad_norm": 0.18523386120796204,
      "learning_rate": 5.958886645776452e-07,
      "loss": 0.0007,
      "step": 157440
    },
    {
      "epoch": 2.5762905996891106,
      "grad_norm": 0.07157806307077408,
      "learning_rate": 5.954380511519286e-07,
      "loss": 0.0003,
      "step": 157450
    },
    {
      "epoch": 2.5764542256401866,
      "grad_norm": 0.25537362694740295,
      "learning_rate": 5.949875973809977e-07,
      "loss": 0.0007,
      "step": 157460
    },
    {
      "epoch": 2.576617851591262,
      "grad_norm": 0.028988340869545937,
      "learning_rate": 5.945373032811785e-07,
      "loss": 0.0003,
      "step": 157470
    },
    {
      "epoch": 2.576781477542338,
      "grad_norm": 0.05501402169466019,
      "learning_rate": 5.940871688687944e-07,
      "loss": 0.001,
      "step": 157480
    },
    {
      "epoch": 2.576945103493414,
      "grad_norm": 0.005681911017745733,
      "learning_rate": 5.936371941601599e-07,
      "loss": 0.0008,
      "step": 157490
    },
    {
      "epoch": 2.5771087294444897,
      "grad_norm": 0.06160204857587814,
      "learning_rate": 5.93187379171587e-07,
      "loss": 0.0008,
      "step": 157500
    },
    {
      "epoch": 2.5772723553955657,
      "grad_norm": 0.051627062261104584,
      "learning_rate": 5.92737723919381e-07,
      "loss": 0.0004,
      "step": 157510
    },
    {
      "epoch": 2.5774359813466416,
      "grad_norm": 0.09027938544750214,
      "learning_rate": 5.922882284198383e-07,
      "loss": 0.0014,
      "step": 157520
    },
    {
      "epoch": 2.577599607297717,
      "grad_norm": 0.03917877748608589,
      "learning_rate": 5.918388926892554e-07,
      "loss": 0.0005,
      "step": 157530
    },
    {
      "epoch": 2.577763233248793,
      "grad_norm": 0.0567542165517807,
      "learning_rate": 5.91389716743917e-07,
      "loss": 0.0005,
      "step": 157540
    },
    {
      "epoch": 2.577926859199869,
      "grad_norm": 0.056425791233778,
      "learning_rate": 5.909407006001061e-07,
      "loss": 0.0006,
      "step": 157550
    },
    {
      "epoch": 2.5780904851509447,
      "grad_norm": 0.03405746817588806,
      "learning_rate": 5.90491844274097e-07,
      "loss": 0.0007,
      "step": 157560
    },
    {
      "epoch": 2.5782541111020207,
      "grad_norm": 0.038308098912239075,
      "learning_rate": 5.900431477821616e-07,
      "loss": 0.0004,
      "step": 157570
    },
    {
      "epoch": 2.5784177370530967,
      "grad_norm": 0.08552277088165283,
      "learning_rate": 5.895946111405632e-07,
      "loss": 0.0011,
      "step": 157580
    },
    {
      "epoch": 2.5785813630041723,
      "grad_norm": 0.04410054534673691,
      "learning_rate": 5.891462343655586e-07,
      "loss": 0.0005,
      "step": 157590
    },
    {
      "epoch": 2.5787449889552483,
      "grad_norm": 0.040066108107566833,
      "learning_rate": 5.886980174734031e-07,
      "loss": 0.0007,
      "step": 157600
    },
    {
      "epoch": 2.5789086149063243,
      "grad_norm": 0.05465305596590042,
      "learning_rate": 5.88249960480341e-07,
      "loss": 0.0011,
      "step": 157610
    },
    {
      "epoch": 2.5790722408574,
      "grad_norm": 0.23041120171546936,
      "learning_rate": 5.87802063402616e-07,
      "loss": 0.0015,
      "step": 157620
    },
    {
      "epoch": 2.579235866808476,
      "grad_norm": 0.14901487529277802,
      "learning_rate": 5.873543262564602e-07,
      "loss": 0.0011,
      "step": 157630
    },
    {
      "epoch": 2.579399492759552,
      "grad_norm": 0.015259046107530594,
      "learning_rate": 5.86906749058106e-07,
      "loss": 0.0002,
      "step": 157640
    },
    {
      "epoch": 2.5795631187106274,
      "grad_norm": 0.20521098375320435,
      "learning_rate": 5.864593318237738e-07,
      "loss": 0.0004,
      "step": 157650
    },
    {
      "epoch": 2.5797267446617034,
      "grad_norm": 0.02627727948129177,
      "learning_rate": 5.86012074569684e-07,
      "loss": 0.0004,
      "step": 157660
    },
    {
      "epoch": 2.5798903706127794,
      "grad_norm": 0.054305899888277054,
      "learning_rate": 5.855649773120492e-07,
      "loss": 0.0009,
      "step": 157670
    },
    {
      "epoch": 2.580053996563855,
      "grad_norm": 0.06675571948289871,
      "learning_rate": 5.851180400670725e-07,
      "loss": 0.0007,
      "step": 157680
    },
    {
      "epoch": 2.580217622514931,
      "grad_norm": 0.01838107220828533,
      "learning_rate": 5.846712628509565e-07,
      "loss": 0.0009,
      "step": 157690
    },
    {
      "epoch": 2.580381248466007,
      "grad_norm": 0.05560867488384247,
      "learning_rate": 5.84224645679895e-07,
      "loss": 0.0008,
      "step": 157700
    },
    {
      "epoch": 2.5805448744170825,
      "grad_norm": 0.0031373733654618263,
      "learning_rate": 5.837781885700777e-07,
      "loss": 0.0004,
      "step": 157710
    },
    {
      "epoch": 2.5807085003681585,
      "grad_norm": 0.0669962614774704,
      "learning_rate": 5.833318915376857e-07,
      "loss": 0.0006,
      "step": 157720
    },
    {
      "epoch": 2.5808721263192345,
      "grad_norm": 0.023115383461117744,
      "learning_rate": 5.828857545988976e-07,
      "loss": 0.0006,
      "step": 157730
    },
    {
      "epoch": 2.58103575227031,
      "grad_norm": 0.04747607931494713,
      "learning_rate": 5.824397777698859e-07,
      "loss": 0.0005,
      "step": 157740
    },
    {
      "epoch": 2.581199378221386,
      "grad_norm": 0.0429086871445179,
      "learning_rate": 5.819939610668135e-07,
      "loss": 0.0007,
      "step": 157750
    },
    {
      "epoch": 2.581363004172462,
      "grad_norm": 0.13357284665107727,
      "learning_rate": 5.815483045058428e-07,
      "loss": 0.0014,
      "step": 157760
    },
    {
      "epoch": 2.5815266301235376,
      "grad_norm": 0.09011368453502655,
      "learning_rate": 5.811028081031267e-07,
      "loss": 0.0004,
      "step": 157770
    },
    {
      "epoch": 2.5816902560746136,
      "grad_norm": 0.09679222106933594,
      "learning_rate": 5.80657471874812e-07,
      "loss": 0.0005,
      "step": 157780
    },
    {
      "epoch": 2.5818538820256896,
      "grad_norm": 0.08673311024904251,
      "learning_rate": 5.802122958370438e-07,
      "loss": 0.0007,
      "step": 157790
    },
    {
      "epoch": 2.582017507976765,
      "grad_norm": 0.09451400488615036,
      "learning_rate": 5.797672800059556e-07,
      "loss": 0.0007,
      "step": 157800
    },
    {
      "epoch": 2.582181133927841,
      "grad_norm": 0.11670426279306412,
      "learning_rate": 5.793224243976814e-07,
      "loss": 0.0007,
      "step": 157810
    },
    {
      "epoch": 2.5823447598789167,
      "grad_norm": 0.031171889975667,
      "learning_rate": 5.788777290283432e-07,
      "loss": 0.0006,
      "step": 157820
    },
    {
      "epoch": 2.5825083858299926,
      "grad_norm": 0.09375748783349991,
      "learning_rate": 5.784331939140619e-07,
      "loss": 0.001,
      "step": 157830
    },
    {
      "epoch": 2.5826720117810686,
      "grad_norm": 0.037660080939531326,
      "learning_rate": 5.779888190709498e-07,
      "loss": 0.0018,
      "step": 157840
    },
    {
      "epoch": 2.582835637732144,
      "grad_norm": 0.07823069393634796,
      "learning_rate": 5.775446045151162e-07,
      "loss": 0.0014,
      "step": 157850
    },
    {
      "epoch": 2.58299926368322,
      "grad_norm": 0.0292635727673769,
      "learning_rate": 5.771005502626603e-07,
      "loss": 0.0006,
      "step": 157860
    },
    {
      "epoch": 2.5831628896342957,
      "grad_norm": 0.04550391063094139,
      "learning_rate": 5.766566563296805e-07,
      "loss": 0.0007,
      "step": 157870
    },
    {
      "epoch": 2.5833265155853717,
      "grad_norm": 0.08595076948404312,
      "learning_rate": 5.762129227322661e-07,
      "loss": 0.0008,
      "step": 157880
    },
    {
      "epoch": 2.5834901415364477,
      "grad_norm": 0.05614157021045685,
      "learning_rate": 5.757693494864991e-07,
      "loss": 0.0012,
      "step": 157890
    },
    {
      "epoch": 2.5836537674875233,
      "grad_norm": 0.04381489008665085,
      "learning_rate": 5.753259366084612e-07,
      "loss": 0.0004,
      "step": 157900
    },
    {
      "epoch": 2.5838173934385993,
      "grad_norm": 0.05486591160297394,
      "learning_rate": 5.74882684114223e-07,
      "loss": 0.0005,
      "step": 157910
    },
    {
      "epoch": 2.5839810193896753,
      "grad_norm": 0.049609098583459854,
      "learning_rate": 5.744395920198525e-07,
      "loss": 0.0005,
      "step": 157920
    },
    {
      "epoch": 2.584144645340751,
      "grad_norm": 0.18027809262275696,
      "learning_rate": 5.739966603414093e-07,
      "loss": 0.0008,
      "step": 157930
    },
    {
      "epoch": 2.584308271291827,
      "grad_norm": 0.13170255720615387,
      "learning_rate": 5.735538890949499e-07,
      "loss": 0.0004,
      "step": 157940
    },
    {
      "epoch": 2.584471897242903,
      "grad_norm": 0.039940591901540756,
      "learning_rate": 5.731112782965237e-07,
      "loss": 0.0005,
      "step": 157950
    },
    {
      "epoch": 2.5846355231939784,
      "grad_norm": 0.31801334023475647,
      "learning_rate": 5.726688279621734e-07,
      "loss": 0.0013,
      "step": 157960
    },
    {
      "epoch": 2.5847991491450544,
      "grad_norm": 0.05857295170426369,
      "learning_rate": 5.722265381079384e-07,
      "loss": 0.0005,
      "step": 157970
    },
    {
      "epoch": 2.5849627750961304,
      "grad_norm": 0.01864403858780861,
      "learning_rate": 5.717844087498492e-07,
      "loss": 0.0006,
      "step": 157980
    },
    {
      "epoch": 2.585126401047206,
      "grad_norm": 0.12821301817893982,
      "learning_rate": 5.713424399039324e-07,
      "loss": 0.0005,
      "step": 157990
    },
    {
      "epoch": 2.585290026998282,
      "grad_norm": 0.046080347150564194,
      "learning_rate": 5.70900631586207e-07,
      "loss": 0.0005,
      "step": 158000
    },
    {
      "epoch": 2.585453652949358,
      "grad_norm": 0.0629124641418457,
      "learning_rate": 5.704589838126889e-07,
      "loss": 0.0008,
      "step": 158010
    },
    {
      "epoch": 2.5856172789004335,
      "grad_norm": 0.0994546040892601,
      "learning_rate": 5.700174965993871e-07,
      "loss": 0.0007,
      "step": 158020
    },
    {
      "epoch": 2.5857809048515095,
      "grad_norm": 0.21401220560073853,
      "learning_rate": 5.695761699623031e-07,
      "loss": 0.0014,
      "step": 158030
    },
    {
      "epoch": 2.5859445308025855,
      "grad_norm": 0.13591761887073517,
      "learning_rate": 5.691350039174359e-07,
      "loss": 0.001,
      "step": 158040
    },
    {
      "epoch": 2.586108156753661,
      "grad_norm": 0.07070714980363846,
      "learning_rate": 5.686939984807743e-07,
      "loss": 0.0005,
      "step": 158050
    },
    {
      "epoch": 2.586271782704737,
      "grad_norm": 0.015184435062110424,
      "learning_rate": 5.682531536683062e-07,
      "loss": 0.0014,
      "step": 158060
    },
    {
      "epoch": 2.586435408655813,
      "grad_norm": 0.10118429362773895,
      "learning_rate": 5.678124694960085e-07,
      "loss": 0.0012,
      "step": 158070
    },
    {
      "epoch": 2.5865990346068886,
      "grad_norm": 0.04829082265496254,
      "learning_rate": 5.673719459798571e-07,
      "loss": 0.0003,
      "step": 158080
    },
    {
      "epoch": 2.5867626605579646,
      "grad_norm": 0.03125864639878273,
      "learning_rate": 5.669315831358196e-07,
      "loss": 0.0007,
      "step": 158090
    },
    {
      "epoch": 2.5869262865090406,
      "grad_norm": 0.10676240921020508,
      "learning_rate": 5.664913809798561e-07,
      "loss": 0.0007,
      "step": 158100
    },
    {
      "epoch": 2.587089912460116,
      "grad_norm": 0.1163443773984909,
      "learning_rate": 5.66051339527926e-07,
      "loss": 0.0004,
      "step": 158110
    },
    {
      "epoch": 2.587253538411192,
      "grad_norm": 0.06596343219280243,
      "learning_rate": 5.656114587959765e-07,
      "loss": 0.0007,
      "step": 158120
    },
    {
      "epoch": 2.587417164362268,
      "grad_norm": 0.05356109142303467,
      "learning_rate": 5.651717387999556e-07,
      "loss": 0.0003,
      "step": 158130
    },
    {
      "epoch": 2.5875807903133436,
      "grad_norm": 0.00801369734108448,
      "learning_rate": 5.647321795557986e-07,
      "loss": 0.0007,
      "step": 158140
    },
    {
      "epoch": 2.5877444162644196,
      "grad_norm": 0.02087208814918995,
      "learning_rate": 5.642927810794413e-07,
      "loss": 0.0006,
      "step": 158150
    },
    {
      "epoch": 2.5879080422154956,
      "grad_norm": 0.07800925523042679,
      "learning_rate": 5.638535433868092e-07,
      "loss": 0.0007,
      "step": 158160
    },
    {
      "epoch": 2.588071668166571,
      "grad_norm": 0.0074714599177241325,
      "learning_rate": 5.634144664938241e-07,
      "loss": 0.0005,
      "step": 158170
    },
    {
      "epoch": 2.588235294117647,
      "grad_norm": 0.07059358060359955,
      "learning_rate": 5.629755504164031e-07,
      "loss": 0.0006,
      "step": 158180
    },
    {
      "epoch": 2.5883989200687227,
      "grad_norm": 0.06601470708847046,
      "learning_rate": 5.625367951704524e-07,
      "loss": 0.0007,
      "step": 158190
    },
    {
      "epoch": 2.5885625460197987,
      "grad_norm": 0.017410963773727417,
      "learning_rate": 5.620982007718789e-07,
      "loss": 0.0005,
      "step": 158200
    },
    {
      "epoch": 2.5887261719708747,
      "grad_norm": 0.022253625094890594,
      "learning_rate": 5.616597672365786e-07,
      "loss": 0.0007,
      "step": 158210
    },
    {
      "epoch": 2.5888897979219503,
      "grad_norm": 0.1345239281654358,
      "learning_rate": 5.612214945804439e-07,
      "loss": 0.0004,
      "step": 158220
    },
    {
      "epoch": 2.5890534238730263,
      "grad_norm": 0.005066594574600458,
      "learning_rate": 5.607833828193632e-07,
      "loss": 0.0005,
      "step": 158230
    },
    {
      "epoch": 2.589217049824102,
      "grad_norm": 0.04432334378361702,
      "learning_rate": 5.603454319692142e-07,
      "loss": 0.0006,
      "step": 158240
    },
    {
      "epoch": 2.589380675775178,
      "grad_norm": 0.047843486070632935,
      "learning_rate": 5.599076420458744e-07,
      "loss": 0.0006,
      "step": 158250
    },
    {
      "epoch": 2.589544301726254,
      "grad_norm": 0.06755217909812927,
      "learning_rate": 5.594700130652092e-07,
      "loss": 0.0003,
      "step": 158260
    },
    {
      "epoch": 2.5897079276773294,
      "grad_norm": 0.09173483401536942,
      "learning_rate": 5.59032545043085e-07,
      "loss": 0.0009,
      "step": 158270
    },
    {
      "epoch": 2.5898715536284054,
      "grad_norm": 0.03438813239336014,
      "learning_rate": 5.585952379953574e-07,
      "loss": 0.0006,
      "step": 158280
    },
    {
      "epoch": 2.5900351795794814,
      "grad_norm": 0.04588679224252701,
      "learning_rate": 5.58158091937877e-07,
      "loss": 0.0006,
      "step": 158290
    },
    {
      "epoch": 2.590198805530557,
      "grad_norm": 0.03274879232048988,
      "learning_rate": 5.577211068864907e-07,
      "loss": 0.0004,
      "step": 158300
    },
    {
      "epoch": 2.590362431481633,
      "grad_norm": 0.011022274382412434,
      "learning_rate": 5.572842828570368e-07,
      "loss": 0.0006,
      "step": 158310
    },
    {
      "epoch": 2.590526057432709,
      "grad_norm": 0.07231555134057999,
      "learning_rate": 5.56847619865351e-07,
      "loss": 0.0006,
      "step": 158320
    },
    {
      "epoch": 2.5906896833837845,
      "grad_norm": 0.041771817952394485,
      "learning_rate": 5.564111179272586e-07,
      "loss": 0.0005,
      "step": 158330
    },
    {
      "epoch": 2.5908533093348605,
      "grad_norm": 0.06861957907676697,
      "learning_rate": 5.55974777058585e-07,
      "loss": 0.0004,
      "step": 158340
    },
    {
      "epoch": 2.5910169352859365,
      "grad_norm": 0.11759281903505325,
      "learning_rate": 5.555385972751431e-07,
      "loss": 0.001,
      "step": 158350
    },
    {
      "epoch": 2.591180561237012,
      "grad_norm": 0.134982630610466,
      "learning_rate": 5.551025785927466e-07,
      "loss": 0.0005,
      "step": 158360
    },
    {
      "epoch": 2.591344187188088,
      "grad_norm": 0.05459032207727432,
      "learning_rate": 5.546667210271972e-07,
      "loss": 0.0007,
      "step": 158370
    },
    {
      "epoch": 2.591507813139164,
      "grad_norm": 0.09381499886512756,
      "learning_rate": 5.542310245942961e-07,
      "loss": 0.0005,
      "step": 158380
    },
    {
      "epoch": 2.5916714390902396,
      "grad_norm": 0.03078806772828102,
      "learning_rate": 5.537954893098357e-07,
      "loss": 0.0007,
      "step": 158390
    },
    {
      "epoch": 2.5918350650413156,
      "grad_norm": 0.0036064907908439636,
      "learning_rate": 5.533601151896013e-07,
      "loss": 0.0005,
      "step": 158400
    },
    {
      "epoch": 2.5919986909923916,
      "grad_norm": 0.1519811898469925,
      "learning_rate": 5.529249022493766e-07,
      "loss": 0.0013,
      "step": 158410
    },
    {
      "epoch": 2.592162316943467,
      "grad_norm": 0.039427027106285095,
      "learning_rate": 5.524898505049348e-07,
      "loss": 0.0003,
      "step": 158420
    },
    {
      "epoch": 2.592325942894543,
      "grad_norm": 0.1695336550474167,
      "learning_rate": 5.520549599720465e-07,
      "loss": 0.0005,
      "step": 158430
    },
    {
      "epoch": 2.592489568845619,
      "grad_norm": 0.058742981404066086,
      "learning_rate": 5.516202306664769e-07,
      "loss": 0.0004,
      "step": 158440
    },
    {
      "epoch": 2.5926531947966946,
      "grad_norm": 0.02838270366191864,
      "learning_rate": 5.511856626039813e-07,
      "loss": 0.0018,
      "step": 158450
    },
    {
      "epoch": 2.5928168207477706,
      "grad_norm": 0.043018046766519547,
      "learning_rate": 5.507512558003142e-07,
      "loss": 0.001,
      "step": 158460
    },
    {
      "epoch": 2.5929804466988466,
      "grad_norm": 0.02285918965935707,
      "learning_rate": 5.503170102712197e-07,
      "loss": 0.0007,
      "step": 158470
    },
    {
      "epoch": 2.593144072649922,
      "grad_norm": 0.1558447778224945,
      "learning_rate": 5.498829260324406e-07,
      "loss": 0.0012,
      "step": 158480
    },
    {
      "epoch": 2.593307698600998,
      "grad_norm": 0.03746763989329338,
      "learning_rate": 5.494490030997079e-07,
      "loss": 0.0004,
      "step": 158490
    },
    {
      "epoch": 2.593471324552074,
      "grad_norm": 0.0841994658112526,
      "learning_rate": 5.490152414887523e-07,
      "loss": 0.0011,
      "step": 158500
    },
    {
      "epoch": 2.5936349505031497,
      "grad_norm": 0.11033559590578079,
      "learning_rate": 5.485816412152978e-07,
      "loss": 0.0004,
      "step": 158510
    },
    {
      "epoch": 2.5937985764542257,
      "grad_norm": 0.005358075723052025,
      "learning_rate": 5.481482022950591e-07,
      "loss": 0.0003,
      "step": 158520
    },
    {
      "epoch": 2.5939622024053017,
      "grad_norm": 0.0492313951253891,
      "learning_rate": 5.477149247437491e-07,
      "loss": 0.0003,
      "step": 158530
    },
    {
      "epoch": 2.5941258283563773,
      "grad_norm": 0.10793673992156982,
      "learning_rate": 5.47281808577071e-07,
      "loss": 0.0012,
      "step": 158540
    },
    {
      "epoch": 2.5942894543074533,
      "grad_norm": 0.02911180444061756,
      "learning_rate": 5.46848853810727e-07,
      "loss": 0.0005,
      "step": 158550
    },
    {
      "epoch": 2.5944530802585293,
      "grad_norm": 0.15029145777225494,
      "learning_rate": 5.464160604604079e-07,
      "loss": 0.0008,
      "step": 158560
    },
    {
      "epoch": 2.594616706209605,
      "grad_norm": 0.14947408437728882,
      "learning_rate": 5.459834285418036e-07,
      "loss": 0.0008,
      "step": 158570
    },
    {
      "epoch": 2.594780332160681,
      "grad_norm": 0.07210278511047363,
      "learning_rate": 5.455509580705953e-07,
      "loss": 0.0005,
      "step": 158580
    },
    {
      "epoch": 2.5949439581117564,
      "grad_norm": 0.12886103987693787,
      "learning_rate": 5.451186490624583e-07,
      "loss": 0.0006,
      "step": 158590
    },
    {
      "epoch": 2.5951075840628324,
      "grad_norm": 0.22724021971225739,
      "learning_rate": 5.446865015330632e-07,
      "loss": 0.0008,
      "step": 158600
    },
    {
      "epoch": 2.5952712100139084,
      "grad_norm": 0.0335865318775177,
      "learning_rate": 5.442545154980744e-07,
      "loss": 0.0008,
      "step": 158610
    },
    {
      "epoch": 2.595434835964984,
      "grad_norm": 0.11760455369949341,
      "learning_rate": 5.438226909731504e-07,
      "loss": 0.0008,
      "step": 158620
    },
    {
      "epoch": 2.59559846191606,
      "grad_norm": 0.26422008872032166,
      "learning_rate": 5.433910279739435e-07,
      "loss": 0.0012,
      "step": 158630
    },
    {
      "epoch": 2.5957620878671355,
      "grad_norm": 0.06604434549808502,
      "learning_rate": 5.429595265161008e-07,
      "loss": 0.0004,
      "step": 158640
    },
    {
      "epoch": 2.5959257138182115,
      "grad_norm": 0.05335859954357147,
      "learning_rate": 5.425281866152627e-07,
      "loss": 0.0004,
      "step": 158650
    },
    {
      "epoch": 2.5960893397692875,
      "grad_norm": 0.011076969094574451,
      "learning_rate": 5.420970082870641e-07,
      "loss": 0.0013,
      "step": 158660
    },
    {
      "epoch": 2.596252965720363,
      "grad_norm": 0.02541961707174778,
      "learning_rate": 5.416659915471362e-07,
      "loss": 0.0008,
      "step": 158670
    },
    {
      "epoch": 2.596416591671439,
      "grad_norm": 0.3922705054283142,
      "learning_rate": 5.412351364111007e-07,
      "loss": 0.0018,
      "step": 158680
    },
    {
      "epoch": 2.596580217622515,
      "grad_norm": 0.22217248380184174,
      "learning_rate": 5.408044428945747e-07,
      "loss": 0.001,
      "step": 158690
    },
    {
      "epoch": 2.5967438435735906,
      "grad_norm": 0.0994834452867508,
      "learning_rate": 5.403739110131695e-07,
      "loss": 0.0008,
      "step": 158700
    },
    {
      "epoch": 2.5969074695246666,
      "grad_norm": 0.051326099783182144,
      "learning_rate": 5.399435407824916e-07,
      "loss": 0.0007,
      "step": 158710
    },
    {
      "epoch": 2.5970710954757426,
      "grad_norm": 0.07883021235466003,
      "learning_rate": 5.395133322181423e-07,
      "loss": 0.0011,
      "step": 158720
    },
    {
      "epoch": 2.597234721426818,
      "grad_norm": 0.0007013934082351625,
      "learning_rate": 5.390832853357125e-07,
      "loss": 0.0016,
      "step": 158730
    },
    {
      "epoch": 2.597398347377894,
      "grad_norm": 0.0654192566871643,
      "learning_rate": 5.386534001507937e-07,
      "loss": 0.0006,
      "step": 158740
    },
    {
      "epoch": 2.59756197332897,
      "grad_norm": 0.08958490192890167,
      "learning_rate": 5.38223676678965e-07,
      "loss": 0.0011,
      "step": 158750
    },
    {
      "epoch": 2.5977255992800456,
      "grad_norm": 0.011869450099766254,
      "learning_rate": 5.377941149358057e-07,
      "loss": 0.0006,
      "step": 158760
    },
    {
      "epoch": 2.5978892252311216,
      "grad_norm": 0.025177253410220146,
      "learning_rate": 5.373647149368843e-07,
      "loss": 0.0004,
      "step": 158770
    },
    {
      "epoch": 2.5980528511821976,
      "grad_norm": 0.027847208082675934,
      "learning_rate": 5.369354766977669e-07,
      "loss": 0.0005,
      "step": 158780
    },
    {
      "epoch": 2.598216477133273,
      "grad_norm": 0.08823072910308838,
      "learning_rate": 5.36506400234012e-07,
      "loss": 0.0013,
      "step": 158790
    },
    {
      "epoch": 2.598380103084349,
      "grad_norm": 0.007788459304720163,
      "learning_rate": 5.360774855611717e-07,
      "loss": 0.0008,
      "step": 158800
    },
    {
      "epoch": 2.598543729035425,
      "grad_norm": 0.024371150881052017,
      "learning_rate": 5.35648732694794e-07,
      "loss": 0.0005,
      "step": 158810
    },
    {
      "epoch": 2.5987073549865007,
      "grad_norm": 0.03180588781833649,
      "learning_rate": 5.352201416504193e-07,
      "loss": 0.0008,
      "step": 158820
    },
    {
      "epoch": 2.5988709809375767,
      "grad_norm": 0.048278626054525375,
      "learning_rate": 5.347917124435853e-07,
      "loss": 0.0006,
      "step": 158830
    },
    {
      "epoch": 2.5990346068886527,
      "grad_norm": 0.03488343581557274,
      "learning_rate": 5.343634450898183e-07,
      "loss": 0.0004,
      "step": 158840
    },
    {
      "epoch": 2.5991982328397283,
      "grad_norm": 0.36903172731399536,
      "learning_rate": 5.339353396046443e-07,
      "loss": 0.0012,
      "step": 158850
    },
    {
      "epoch": 2.5993618587908043,
      "grad_norm": 0.025290217250585556,
      "learning_rate": 5.335073960035797e-07,
      "loss": 0.0008,
      "step": 158860
    },
    {
      "epoch": 2.5995254847418803,
      "grad_norm": 0.13256970047950745,
      "learning_rate": 5.330796143021372e-07,
      "loss": 0.0014,
      "step": 158870
    },
    {
      "epoch": 2.599689110692956,
      "grad_norm": 0.06488094478845596,
      "learning_rate": 5.326519945158242e-07,
      "loss": 0.0005,
      "step": 158880
    },
    {
      "epoch": 2.599852736644032,
      "grad_norm": 0.3628421425819397,
      "learning_rate": 5.322245366601381e-07,
      "loss": 0.0015,
      "step": 158890
    },
    {
      "epoch": 2.600016362595108,
      "grad_norm": 0.10871259868144989,
      "learning_rate": 5.31797240750575e-07,
      "loss": 0.0012,
      "step": 158900
    },
    {
      "epoch": 2.6001799885461834,
      "grad_norm": 0.08409851789474487,
      "learning_rate": 5.313701068026217e-07,
      "loss": 0.0006,
      "step": 158910
    },
    {
      "epoch": 2.6003436144972594,
      "grad_norm": 0.10313976556062698,
      "learning_rate": 5.309431348317634e-07,
      "loss": 0.0007,
      "step": 158920
    },
    {
      "epoch": 2.6005072404483354,
      "grad_norm": 0.04597606509923935,
      "learning_rate": 5.305163248534745e-07,
      "loss": 0.0006,
      "step": 158930
    },
    {
      "epoch": 2.600670866399411,
      "grad_norm": 0.041362982243299484,
      "learning_rate": 5.300896768832264e-07,
      "loss": 0.0006,
      "step": 158940
    },
    {
      "epoch": 2.600834492350487,
      "grad_norm": 0.051144372671842575,
      "learning_rate": 5.296631909364852e-07,
      "loss": 0.001,
      "step": 158950
    },
    {
      "epoch": 2.6009981183015625,
      "grad_norm": 0.07852887362241745,
      "learning_rate": 5.292368670287084e-07,
      "loss": 0.0004,
      "step": 158960
    },
    {
      "epoch": 2.6011617442526385,
      "grad_norm": 0.0826437845826149,
      "learning_rate": 5.288107051753511e-07,
      "loss": 0.0011,
      "step": 158970
    },
    {
      "epoch": 2.6013253702037145,
      "grad_norm": 0.034881506115198135,
      "learning_rate": 5.283847053918595e-07,
      "loss": 0.0008,
      "step": 158980
    },
    {
      "epoch": 2.60148899615479,
      "grad_norm": 0.04568963870406151,
      "learning_rate": 5.279588676936737e-07,
      "loss": 0.0005,
      "step": 158990
    },
    {
      "epoch": 2.601652622105866,
      "grad_norm": 0.15884359180927277,
      "learning_rate": 5.275331920962318e-07,
      "loss": 0.0009,
      "step": 159000
    },
    {
      "epoch": 2.6018162480569416,
      "grad_norm": 0.10657093673944473,
      "learning_rate": 5.271076786149615e-07,
      "loss": 0.0003,
      "step": 159010
    },
    {
      "epoch": 2.6019798740080176,
      "grad_norm": 0.061083387583494186,
      "learning_rate": 5.266823272652887e-07,
      "loss": 0.001,
      "step": 159020
    },
    {
      "epoch": 2.6021434999590936,
      "grad_norm": 0.014530250802636147,
      "learning_rate": 5.262571380626291e-07,
      "loss": 0.0006,
      "step": 159030
    },
    {
      "epoch": 2.602307125910169,
      "grad_norm": 0.028397101908922195,
      "learning_rate": 5.258321110223968e-07,
      "loss": 0.0006,
      "step": 159040
    },
    {
      "epoch": 2.602470751861245,
      "grad_norm": 0.17988136410713196,
      "learning_rate": 5.254072461599963e-07,
      "loss": 0.0007,
      "step": 159050
    },
    {
      "epoch": 2.602634377812321,
      "grad_norm": 0.09252893179655075,
      "learning_rate": 5.249825434908296e-07,
      "loss": 0.0006,
      "step": 159060
    },
    {
      "epoch": 2.6027980037633967,
      "grad_norm": 0.02128220535814762,
      "learning_rate": 5.245580030302888e-07,
      "loss": 0.0015,
      "step": 159070
    },
    {
      "epoch": 2.6029616297144726,
      "grad_norm": 0.0709434300661087,
      "learning_rate": 5.241336247937656e-07,
      "loss": 0.0005,
      "step": 159080
    },
    {
      "epoch": 2.6031252556655486,
      "grad_norm": 0.020560117438435555,
      "learning_rate": 5.23709408796641e-07,
      "loss": 0.0008,
      "step": 159090
    },
    {
      "epoch": 2.603288881616624,
      "grad_norm": 0.17943745851516724,
      "learning_rate": 5.232853550542904e-07,
      "loss": 0.0014,
      "step": 159100
    },
    {
      "epoch": 2.6034525075677,
      "grad_norm": 0.053974755108356476,
      "learning_rate": 5.228614635820878e-07,
      "loss": 0.0009,
      "step": 159110
    },
    {
      "epoch": 2.603616133518776,
      "grad_norm": 0.10413036495447159,
      "learning_rate": 5.224377343953951e-07,
      "loss": 0.0012,
      "step": 159120
    },
    {
      "epoch": 2.6037797594698517,
      "grad_norm": 0.12252624332904816,
      "learning_rate": 5.220141675095741e-07,
      "loss": 0.0009,
      "step": 159130
    },
    {
      "epoch": 2.6039433854209277,
      "grad_norm": 0.0013508291449397802,
      "learning_rate": 5.215907629399763e-07,
      "loss": 0.0006,
      "step": 159140
    },
    {
      "epoch": 2.6041070113720037,
      "grad_norm": 0.05760389566421509,
      "learning_rate": 5.211675207019495e-07,
      "loss": 0.001,
      "step": 159150
    },
    {
      "epoch": 2.6042706373230793,
      "grad_norm": 0.0725826546549797,
      "learning_rate": 5.207444408108364e-07,
      "loss": 0.001,
      "step": 159160
    },
    {
      "epoch": 2.6044342632741553,
      "grad_norm": 0.07261897623538971,
      "learning_rate": 5.203215232819709e-07,
      "loss": 0.0006,
      "step": 159170
    },
    {
      "epoch": 2.6045978892252313,
      "grad_norm": 0.11361802369356155,
      "learning_rate": 5.198987681306855e-07,
      "loss": 0.0008,
      "step": 159180
    },
    {
      "epoch": 2.604761515176307,
      "grad_norm": 0.19185525178909302,
      "learning_rate": 5.194761753722999e-07,
      "loss": 0.0008,
      "step": 159190
    },
    {
      "epoch": 2.604925141127383,
      "grad_norm": 0.058302342891693115,
      "learning_rate": 5.190537450221356e-07,
      "loss": 0.0005,
      "step": 159200
    },
    {
      "epoch": 2.605088767078459,
      "grad_norm": 0.018518047407269478,
      "learning_rate": 5.18631477095502e-07,
      "loss": 0.0006,
      "step": 159210
    },
    {
      "epoch": 2.6052523930295344,
      "grad_norm": 0.11293692141771317,
      "learning_rate": 5.182093716077069e-07,
      "loss": 0.001,
      "step": 159220
    },
    {
      "epoch": 2.6054160189806104,
      "grad_norm": 0.07691781967878342,
      "learning_rate": 5.177874285740513e-07,
      "loss": 0.0008,
      "step": 159230
    },
    {
      "epoch": 2.6055796449316864,
      "grad_norm": 0.07937619835138321,
      "learning_rate": 5.173656480098283e-07,
      "loss": 0.0004,
      "step": 159240
    },
    {
      "epoch": 2.605743270882762,
      "grad_norm": 0.182184636592865,
      "learning_rate": 5.169440299303269e-07,
      "loss": 0.001,
      "step": 159250
    },
    {
      "epoch": 2.605906896833838,
      "grad_norm": 0.013033591210842133,
      "learning_rate": 5.16522574350829e-07,
      "loss": 0.0006,
      "step": 159260
    },
    {
      "epoch": 2.606070522784914,
      "grad_norm": 0.11774258315563202,
      "learning_rate": 5.161012812866134e-07,
      "loss": 0.0007,
      "step": 159270
    },
    {
      "epoch": 2.6062341487359895,
      "grad_norm": 0.08002663403749466,
      "learning_rate": 5.156801507529491e-07,
      "loss": 0.0011,
      "step": 159280
    },
    {
      "epoch": 2.6063977746870655,
      "grad_norm": 0.2496061474084854,
      "learning_rate": 5.152591827651011e-07,
      "loss": 0.001,
      "step": 159290
    },
    {
      "epoch": 2.6065614006381415,
      "grad_norm": 0.11888951063156128,
      "learning_rate": 5.148383773383292e-07,
      "loss": 0.0009,
      "step": 159300
    },
    {
      "epoch": 2.606725026589217,
      "grad_norm": 0.08274941891431808,
      "learning_rate": 5.144177344878859e-07,
      "loss": 0.0003,
      "step": 159310
    },
    {
      "epoch": 2.606888652540293,
      "grad_norm": 0.18723848462104797,
      "learning_rate": 5.1399725422902e-07,
      "loss": 0.0011,
      "step": 159320
    },
    {
      "epoch": 2.607052278491369,
      "grad_norm": 0.08062460273504257,
      "learning_rate": 5.135769365769705e-07,
      "loss": 0.0007,
      "step": 159330
    },
    {
      "epoch": 2.6072159044424446,
      "grad_norm": 0.1098063737154007,
      "learning_rate": 5.131567815469752e-07,
      "loss": 0.0008,
      "step": 159340
    },
    {
      "epoch": 2.6073795303935206,
      "grad_norm": 0.057517293840646744,
      "learning_rate": 5.127367891542618e-07,
      "loss": 0.0007,
      "step": 159350
    },
    {
      "epoch": 2.607543156344596,
      "grad_norm": 0.0426187664270401,
      "learning_rate": 5.123169594140559e-07,
      "loss": 0.0004,
      "step": 159360
    },
    {
      "epoch": 2.607706782295672,
      "grad_norm": 0.0022553694434463978,
      "learning_rate": 5.118972923415733e-07,
      "loss": 0.0006,
      "step": 159370
    },
    {
      "epoch": 2.607870408246748,
      "grad_norm": 0.010330184362828732,
      "learning_rate": 5.114777879520278e-07,
      "loss": 0.0005,
      "step": 159380
    },
    {
      "epoch": 2.6080340341978236,
      "grad_norm": 0.03572759032249451,
      "learning_rate": 5.110584462606249e-07,
      "loss": 0.0007,
      "step": 159390
    },
    {
      "epoch": 2.6081976601488996,
      "grad_norm": 0.08679718524217606,
      "learning_rate": 5.106392672825639e-07,
      "loss": 0.0009,
      "step": 159400
    },
    {
      "epoch": 2.608361286099975,
      "grad_norm": 0.09565508365631104,
      "learning_rate": 5.102202510330396e-07,
      "loss": 0.0015,
      "step": 159410
    },
    {
      "epoch": 2.608524912051051,
      "grad_norm": 0.12138701975345612,
      "learning_rate": 5.098013975272397e-07,
      "loss": 0.0007,
      "step": 159420
    },
    {
      "epoch": 2.608688538002127,
      "grad_norm": 0.0751548707485199,
      "learning_rate": 5.093827067803475e-07,
      "loss": 0.0006,
      "step": 159430
    },
    {
      "epoch": 2.6088521639532027,
      "grad_norm": 0.037758518010377884,
      "learning_rate": 5.0896417880754e-07,
      "loss": 0.0012,
      "step": 159440
    },
    {
      "epoch": 2.6090157899042787,
      "grad_norm": 0.0408528707921505,
      "learning_rate": 5.085458136239862e-07,
      "loss": 0.0003,
      "step": 159450
    },
    {
      "epoch": 2.6091794158553547,
      "grad_norm": 0.07202189415693283,
      "learning_rate": 5.08127611244853e-07,
      "loss": 0.0007,
      "step": 159460
    },
    {
      "epoch": 2.6093430418064303,
      "grad_norm": 0.03407467156648636,
      "learning_rate": 5.077095716852965e-07,
      "loss": 0.0005,
      "step": 159470
    },
    {
      "epoch": 2.6095066677575063,
      "grad_norm": 0.13512128591537476,
      "learning_rate": 5.072916949604723e-07,
      "loss": 0.0006,
      "step": 159480
    },
    {
      "epoch": 2.6096702937085823,
      "grad_norm": 0.13897289335727692,
      "learning_rate": 5.068739810855261e-07,
      "loss": 0.0009,
      "step": 159490
    },
    {
      "epoch": 2.609833919659658,
      "grad_norm": 0.23511986434459686,
      "learning_rate": 5.064564300755981e-07,
      "loss": 0.001,
      "step": 159500
    },
    {
      "epoch": 2.609997545610734,
      "grad_norm": 0.03486602380871773,
      "learning_rate": 5.060390419458255e-07,
      "loss": 0.0011,
      "step": 159510
    },
    {
      "epoch": 2.61016117156181,
      "grad_norm": 0.04421933740377426,
      "learning_rate": 5.056218167113359e-07,
      "loss": 0.0007,
      "step": 159520
    },
    {
      "epoch": 2.6103247975128854,
      "grad_norm": 0.03591364622116089,
      "learning_rate": 5.052047543872546e-07,
      "loss": 0.0011,
      "step": 159530
    },
    {
      "epoch": 2.6104884234639614,
      "grad_norm": 0.1253529191017151,
      "learning_rate": 5.04787854988697e-07,
      "loss": 0.0011,
      "step": 159540
    },
    {
      "epoch": 2.6106520494150374,
      "grad_norm": 0.0727643072605133,
      "learning_rate": 5.043711185307765e-07,
      "loss": 0.0006,
      "step": 159550
    },
    {
      "epoch": 2.610815675366113,
      "grad_norm": 0.020934706553816795,
      "learning_rate": 5.039545450285971e-07,
      "loss": 0.0006,
      "step": 159560
    },
    {
      "epoch": 2.610979301317189,
      "grad_norm": 0.02424517087638378,
      "learning_rate": 5.035381344972601e-07,
      "loss": 0.0005,
      "step": 159570
    },
    {
      "epoch": 2.611142927268265,
      "grad_norm": 0.029768170788884163,
      "learning_rate": 5.03121886951859e-07,
      "loss": 0.0003,
      "step": 159580
    },
    {
      "epoch": 2.6113065532193405,
      "grad_norm": 0.045855093747377396,
      "learning_rate": 5.027058024074805e-07,
      "loss": 0.0006,
      "step": 159590
    },
    {
      "epoch": 2.6114701791704165,
      "grad_norm": 0.13867069780826569,
      "learning_rate": 5.022898808792093e-07,
      "loss": 0.0005,
      "step": 159600
    },
    {
      "epoch": 2.6116338051214925,
      "grad_norm": 0.038604188710451126,
      "learning_rate": 5.018741223821183e-07,
      "loss": 0.0011,
      "step": 159610
    },
    {
      "epoch": 2.611797431072568,
      "grad_norm": 0.03313662111759186,
      "learning_rate": 5.014585269312805e-07,
      "loss": 0.0009,
      "step": 159620
    },
    {
      "epoch": 2.611961057023644,
      "grad_norm": 0.06124007701873779,
      "learning_rate": 5.010430945417582e-07,
      "loss": 0.0008,
      "step": 159630
    },
    {
      "epoch": 2.61212468297472,
      "grad_norm": 0.009094329550862312,
      "learning_rate": 5.006278252286106e-07,
      "loss": 0.0005,
      "step": 159640
    },
    {
      "epoch": 2.6122883089257956,
      "grad_norm": 0.033455125987529755,
      "learning_rate": 5.002127190068917e-07,
      "loss": 0.0002,
      "step": 159650
    },
    {
      "epoch": 2.6124519348768716,
      "grad_norm": 0.02971513755619526,
      "learning_rate": 4.997977758916456e-07,
      "loss": 0.0009,
      "step": 159660
    },
    {
      "epoch": 2.6126155608279475,
      "grad_norm": 0.15464822947978973,
      "learning_rate": 4.993829958979151e-07,
      "loss": 0.0007,
      "step": 159670
    },
    {
      "epoch": 2.612779186779023,
      "grad_norm": 0.0738198310136795,
      "learning_rate": 4.98968379040734e-07,
      "loss": 0.0008,
      "step": 159680
    },
    {
      "epoch": 2.612942812730099,
      "grad_norm": 0.1407453864812851,
      "learning_rate": 4.985539253351318e-07,
      "loss": 0.0011,
      "step": 159690
    },
    {
      "epoch": 2.613106438681175,
      "grad_norm": 0.026515763252973557,
      "learning_rate": 4.981396347961293e-07,
      "loss": 0.0007,
      "step": 159700
    },
    {
      "epoch": 2.6132700646322506,
      "grad_norm": 0.0684126541018486,
      "learning_rate": 4.97725507438745e-07,
      "loss": 0.0003,
      "step": 159710
    },
    {
      "epoch": 2.6134336905833266,
      "grad_norm": 0.021793697029352188,
      "learning_rate": 4.973115432779918e-07,
      "loss": 0.0013,
      "step": 159720
    },
    {
      "epoch": 2.613597316534402,
      "grad_norm": 0.013136429712176323,
      "learning_rate": 4.968977423288718e-07,
      "loss": 0.0009,
      "step": 159730
    },
    {
      "epoch": 2.613760942485478,
      "grad_norm": 0.04625855013728142,
      "learning_rate": 4.964841046063862e-07,
      "loss": 0.0004,
      "step": 159740
    },
    {
      "epoch": 2.613924568436554,
      "grad_norm": 0.0081694470718503,
      "learning_rate": 4.960706301255275e-07,
      "loss": 0.0005,
      "step": 159750
    },
    {
      "epoch": 2.6140881943876297,
      "grad_norm": 0.05213101953268051,
      "learning_rate": 4.956573189012848e-07,
      "loss": 0.0006,
      "step": 159760
    },
    {
      "epoch": 2.6142518203387057,
      "grad_norm": 0.06317899376153946,
      "learning_rate": 4.952441709486366e-07,
      "loss": 0.0008,
      "step": 159770
    },
    {
      "epoch": 2.6144154462897813,
      "grad_norm": 0.13428868353366852,
      "learning_rate": 4.948311862825622e-07,
      "loss": 0.0006,
      "step": 159780
    },
    {
      "epoch": 2.6145790722408573,
      "grad_norm": 0.04951729625463486,
      "learning_rate": 4.944183649180284e-07,
      "loss": 0.0005,
      "step": 159790
    },
    {
      "epoch": 2.6147426981919333,
      "grad_norm": 0.06658707559108734,
      "learning_rate": 4.940057068699999e-07,
      "loss": 0.0005,
      "step": 159800
    },
    {
      "epoch": 2.614906324143009,
      "grad_norm": 0.05197855830192566,
      "learning_rate": 4.935932121534353e-07,
      "loss": 0.0017,
      "step": 159810
    },
    {
      "epoch": 2.615069950094085,
      "grad_norm": 0.20846471190452576,
      "learning_rate": 4.931808807832849e-07,
      "loss": 0.0007,
      "step": 159820
    },
    {
      "epoch": 2.615233576045161,
      "grad_norm": 0.2562103271484375,
      "learning_rate": 4.927687127744968e-07,
      "loss": 0.0016,
      "step": 159830
    },
    {
      "epoch": 2.6153972019962364,
      "grad_norm": 0.05031110346317291,
      "learning_rate": 4.923567081420089e-07,
      "loss": 0.0007,
      "step": 159840
    },
    {
      "epoch": 2.6155608279473124,
      "grad_norm": 0.11691921204328537,
      "learning_rate": 4.919448669007576e-07,
      "loss": 0.0004,
      "step": 159850
    },
    {
      "epoch": 2.6157244538983884,
      "grad_norm": 0.20276382565498352,
      "learning_rate": 4.915331890656689e-07,
      "loss": 0.0009,
      "step": 159860
    },
    {
      "epoch": 2.615888079849464,
      "grad_norm": 0.09058766067028046,
      "learning_rate": 4.911216746516662e-07,
      "loss": 0.0013,
      "step": 159870
    },
    {
      "epoch": 2.61605170580054,
      "grad_norm": 0.1431046426296234,
      "learning_rate": 4.907103236736683e-07,
      "loss": 0.0006,
      "step": 159880
    },
    {
      "epoch": 2.616215331751616,
      "grad_norm": 0.16631601750850677,
      "learning_rate": 4.902991361465809e-07,
      "loss": 0.0006,
      "step": 159890
    },
    {
      "epoch": 2.6163789577026915,
      "grad_norm": 0.04936761036515236,
      "learning_rate": 4.898881120853121e-07,
      "loss": 0.0004,
      "step": 159900
    },
    {
      "epoch": 2.6165425836537675,
      "grad_norm": 0.04305139183998108,
      "learning_rate": 4.894772515047591e-07,
      "loss": 0.0006,
      "step": 159910
    },
    {
      "epoch": 2.6167062096048435,
      "grad_norm": 0.004726100247353315,
      "learning_rate": 4.890665544198147e-07,
      "loss": 0.0005,
      "step": 159920
    },
    {
      "epoch": 2.616869835555919,
      "grad_norm": 0.034349896013736725,
      "learning_rate": 4.886560208453667e-07,
      "loss": 0.0008,
      "step": 159930
    },
    {
      "epoch": 2.617033461506995,
      "grad_norm": 0.028120718896389008,
      "learning_rate": 4.882456507962946e-07,
      "loss": 0.0009,
      "step": 159940
    },
    {
      "epoch": 2.617197087458071,
      "grad_norm": 0.16208238899707794,
      "learning_rate": 4.878354442874745e-07,
      "loss": 0.0008,
      "step": 159950
    },
    {
      "epoch": 2.6173607134091466,
      "grad_norm": 0.058615487068891525,
      "learning_rate": 4.874254013337743e-07,
      "loss": 0.0006,
      "step": 159960
    },
    {
      "epoch": 2.6175243393602226,
      "grad_norm": 0.03521759435534477,
      "learning_rate": 4.870155219500583e-07,
      "loss": 0.0003,
      "step": 159970
    },
    {
      "epoch": 2.6176879653112985,
      "grad_norm": 0.05476340651512146,
      "learning_rate": 4.86605806151182e-07,
      "loss": 0.0007,
      "step": 159980
    },
    {
      "epoch": 2.617851591262374,
      "grad_norm": 0.0009394604712724686,
      "learning_rate": 4.861962539519982e-07,
      "loss": 0.0008,
      "step": 159990
    },
    {
      "epoch": 2.61801521721345,
      "grad_norm": 0.13531015813350677,
      "learning_rate": 4.85786865367352e-07,
      "loss": 0.0006,
      "step": 160000
    },
    {
      "epoch": 2.618178843164526,
      "grad_norm": 0.03410707414150238,
      "learning_rate": 4.853776404120814e-07,
      "loss": 0.0009,
      "step": 160010
    },
    {
      "epoch": 2.6183424691156016,
      "grad_norm": 0.0123378811404109,
      "learning_rate": 4.849685791010217e-07,
      "loss": 0.0008,
      "step": 160020
    },
    {
      "epoch": 2.6185060950666776,
      "grad_norm": 0.03474989905953407,
      "learning_rate": 4.845596814489983e-07,
      "loss": 0.0006,
      "step": 160030
    },
    {
      "epoch": 2.6186697210177536,
      "grad_norm": 0.1863306313753128,
      "learning_rate": 4.841509474708345e-07,
      "loss": 0.0005,
      "step": 160040
    },
    {
      "epoch": 2.618833346968829,
      "grad_norm": 0.015147080644965172,
      "learning_rate": 4.837423771813449e-07,
      "loss": 0.0008,
      "step": 160050
    },
    {
      "epoch": 2.618996972919905,
      "grad_norm": 0.061684221029281616,
      "learning_rate": 4.833339705953405e-07,
      "loss": 0.0005,
      "step": 160060
    },
    {
      "epoch": 2.619160598870981,
      "grad_norm": 0.02952273003757,
      "learning_rate": 4.829257277276228e-07,
      "loss": 0.0006,
      "step": 160070
    },
    {
      "epoch": 2.6193242248220567,
      "grad_norm": 0.1860480159521103,
      "learning_rate": 4.825176485929911e-07,
      "loss": 0.0011,
      "step": 160080
    },
    {
      "epoch": 2.6194878507731327,
      "grad_norm": 0.08580615371465683,
      "learning_rate": 4.821097332062391e-07,
      "loss": 0.0011,
      "step": 160090
    },
    {
      "epoch": 2.6196514767242087,
      "grad_norm": 0.12832696735858917,
      "learning_rate": 4.817019815821489e-07,
      "loss": 0.0005,
      "step": 160100
    },
    {
      "epoch": 2.6198151026752843,
      "grad_norm": 0.00863903108984232,
      "learning_rate": 4.812943937355035e-07,
      "loss": 0.0011,
      "step": 160110
    },
    {
      "epoch": 2.6199787286263603,
      "grad_norm": 0.11040623486042023,
      "learning_rate": 4.808869696810747e-07,
      "loss": 0.0027,
      "step": 160120
    },
    {
      "epoch": 2.620142354577436,
      "grad_norm": 0.05967620015144348,
      "learning_rate": 4.804797094336333e-07,
      "loss": 0.0004,
      "step": 160130
    },
    {
      "epoch": 2.620305980528512,
      "grad_norm": 0.12195799499750137,
      "learning_rate": 4.800726130079386e-07,
      "loss": 0.0011,
      "step": 160140
    },
    {
      "epoch": 2.620469606479588,
      "grad_norm": 0.07639157027006149,
      "learning_rate": 4.796656804187488e-07,
      "loss": 0.0007,
      "step": 160150
    },
    {
      "epoch": 2.6206332324306634,
      "grad_norm": 0.03131287544965744,
      "learning_rate": 4.792589116808144e-07,
      "loss": 0.0008,
      "step": 160160
    },
    {
      "epoch": 2.6207968583817394,
      "grad_norm": 0.12282074242830276,
      "learning_rate": 4.788523068088779e-07,
      "loss": 0.001,
      "step": 160170
    },
    {
      "epoch": 2.620960484332815,
      "grad_norm": 0.04232249781489372,
      "learning_rate": 4.784458658176805e-07,
      "loss": 0.0004,
      "step": 160180
    },
    {
      "epoch": 2.621124110283891,
      "grad_norm": 0.007921084761619568,
      "learning_rate": 4.78039588721953e-07,
      "loss": 0.0007,
      "step": 160190
    },
    {
      "epoch": 2.621287736234967,
      "grad_norm": 0.01960846409201622,
      "learning_rate": 4.77633475536421e-07,
      "loss": 0.0007,
      "step": 160200
    },
    {
      "epoch": 2.6214513621860425,
      "grad_norm": 0.07180566340684891,
      "learning_rate": 4.772275262758069e-07,
      "loss": 0.0008,
      "step": 160210
    },
    {
      "epoch": 2.6216149881371185,
      "grad_norm": 0.06493759900331497,
      "learning_rate": 4.768217409548248e-07,
      "loss": 0.0005,
      "step": 160220
    },
    {
      "epoch": 2.6217786140881945,
      "grad_norm": 0.08901230245828629,
      "learning_rate": 4.764161195881839e-07,
      "loss": 0.0005,
      "step": 160230
    },
    {
      "epoch": 2.62194224003927,
      "grad_norm": 0.07273291051387787,
      "learning_rate": 4.760106621905852e-07,
      "loss": 0.0005,
      "step": 160240
    },
    {
      "epoch": 2.622105865990346,
      "grad_norm": 0.030528923496603966,
      "learning_rate": 4.7560536877672814e-07,
      "loss": 0.0008,
      "step": 160250
    },
    {
      "epoch": 2.622269491941422,
      "grad_norm": 0.057882267981767654,
      "learning_rate": 4.7520023936130147e-07,
      "loss": 0.0004,
      "step": 160260
    },
    {
      "epoch": 2.6224331178924976,
      "grad_norm": 0.04587721824645996,
      "learning_rate": 4.7479527395899175e-07,
      "loss": 0.0005,
      "step": 160270
    },
    {
      "epoch": 2.6225967438435736,
      "grad_norm": 0.02117150090634823,
      "learning_rate": 4.743904725844767e-07,
      "loss": 0.0006,
      "step": 160280
    },
    {
      "epoch": 2.6227603697946495,
      "grad_norm": 0.05879410356283188,
      "learning_rate": 4.739858352524307e-07,
      "loss": 0.0003,
      "step": 160290
    },
    {
      "epoch": 2.622923995745725,
      "grad_norm": 0.05598529428243637,
      "learning_rate": 4.735813619775204e-07,
      "loss": 0.0008,
      "step": 160300
    },
    {
      "epoch": 2.623087621696801,
      "grad_norm": 0.0191376693546772,
      "learning_rate": 4.731770527744056e-07,
      "loss": 0.0011,
      "step": 160310
    },
    {
      "epoch": 2.623251247647877,
      "grad_norm": 0.02689223363995552,
      "learning_rate": 4.727729076577442e-07,
      "loss": 0.001,
      "step": 160320
    },
    {
      "epoch": 2.6234148735989526,
      "grad_norm": 0.1103043407201767,
      "learning_rate": 4.7236892664218217e-07,
      "loss": 0.0005,
      "step": 160330
    },
    {
      "epoch": 2.6235784995500286,
      "grad_norm": 0.01456078328192234,
      "learning_rate": 4.719651097423661e-07,
      "loss": 0.0009,
      "step": 160340
    },
    {
      "epoch": 2.6237421255011046,
      "grad_norm": 0.05939066782593727,
      "learning_rate": 4.71561456972931e-07,
      "loss": 0.0008,
      "step": 160350
    },
    {
      "epoch": 2.62390575145218,
      "grad_norm": 0.22322286665439606,
      "learning_rate": 4.7115796834850957e-07,
      "loss": 0.0008,
      "step": 160360
    },
    {
      "epoch": 2.624069377403256,
      "grad_norm": 0.02628941461443901,
      "learning_rate": 4.707546438837274e-07,
      "loss": 0.0003,
      "step": 160370
    },
    {
      "epoch": 2.624233003354332,
      "grad_norm": 0.07713694125413895,
      "learning_rate": 4.703514835932027e-07,
      "loss": 0.0006,
      "step": 160380
    },
    {
      "epoch": 2.6243966293054077,
      "grad_norm": 0.1457454115152359,
      "learning_rate": 4.699484874915522e-07,
      "loss": 0.0008,
      "step": 160390
    },
    {
      "epoch": 2.6245602552564837,
      "grad_norm": 0.04235680028796196,
      "learning_rate": 4.6954565559337917e-07,
      "loss": 0.0007,
      "step": 160400
    },
    {
      "epoch": 2.6247238812075597,
      "grad_norm": 0.0984048843383789,
      "learning_rate": 4.691429879132886e-07,
      "loss": 0.0021,
      "step": 160410
    },
    {
      "epoch": 2.6248875071586353,
      "grad_norm": 0.10287055373191833,
      "learning_rate": 4.687404844658744e-07,
      "loss": 0.0007,
      "step": 160420
    },
    {
      "epoch": 2.6250511331097113,
      "grad_norm": 0.02886793576180935,
      "learning_rate": 4.6833814526572654e-07,
      "loss": 0.0008,
      "step": 160430
    },
    {
      "epoch": 2.6252147590607873,
      "grad_norm": 0.019250407814979553,
      "learning_rate": 4.679359703274305e-07,
      "loss": 0.0004,
      "step": 160440
    },
    {
      "epoch": 2.625378385011863,
      "grad_norm": 0.059348076581954956,
      "learning_rate": 4.6753395966556203e-07,
      "loss": 0.0008,
      "step": 160450
    },
    {
      "epoch": 2.625542010962939,
      "grad_norm": 0.17093642055988312,
      "learning_rate": 4.6713211329469543e-07,
      "loss": 0.0007,
      "step": 160460
    },
    {
      "epoch": 2.625705636914015,
      "grad_norm": 0.11317221075296402,
      "learning_rate": 4.6673043122939354e-07,
      "loss": 0.0007,
      "step": 160470
    },
    {
      "epoch": 2.6258692628650904,
      "grad_norm": 0.021574676036834717,
      "learning_rate": 4.663289134842197e-07,
      "loss": 0.0005,
      "step": 160480
    },
    {
      "epoch": 2.6260328888161664,
      "grad_norm": 0.011897759512066841,
      "learning_rate": 4.659275600737262e-07,
      "loss": 0.0005,
      "step": 160490
    },
    {
      "epoch": 2.626196514767242,
      "grad_norm": 0.014802451245486736,
      "learning_rate": 4.6552637101246026e-07,
      "loss": 0.0011,
      "step": 160500
    },
    {
      "epoch": 2.626360140718318,
      "grad_norm": 0.0313987135887146,
      "learning_rate": 4.6512534631496575e-07,
      "loss": 0.0008,
      "step": 160510
    },
    {
      "epoch": 2.626523766669394,
      "grad_norm": 0.054005250334739685,
      "learning_rate": 4.6472448599577724e-07,
      "loss": 0.0004,
      "step": 160520
    },
    {
      "epoch": 2.6266873926204695,
      "grad_norm": 0.09609302878379822,
      "learning_rate": 4.6432379006942695e-07,
      "loss": 0.0024,
      "step": 160530
    },
    {
      "epoch": 2.6268510185715455,
      "grad_norm": 0.17505386471748352,
      "learning_rate": 4.6392325855043716e-07,
      "loss": 0.0005,
      "step": 160540
    },
    {
      "epoch": 2.627014644522621,
      "grad_norm": 0.050218816846609116,
      "learning_rate": 4.6352289145332797e-07,
      "loss": 0.0007,
      "step": 160550
    },
    {
      "epoch": 2.627178270473697,
      "grad_norm": 0.10288070887327194,
      "learning_rate": 4.631226887926099e-07,
      "loss": 0.0009,
      "step": 160560
    },
    {
      "epoch": 2.627341896424773,
      "grad_norm": 0.01783854514360428,
      "learning_rate": 4.627226505827914e-07,
      "loss": 0.0006,
      "step": 160570
    },
    {
      "epoch": 2.6275055223758486,
      "grad_norm": 0.019573703408241272,
      "learning_rate": 4.623227768383703e-07,
      "loss": 0.0003,
      "step": 160580
    },
    {
      "epoch": 2.6276691483269246,
      "grad_norm": 0.09200419485569,
      "learning_rate": 4.6192306757384397e-07,
      "loss": 0.0006,
      "step": 160590
    },
    {
      "epoch": 2.6278327742780005,
      "grad_norm": 0.25513818860054016,
      "learning_rate": 4.6152352280369904e-07,
      "loss": 0.0008,
      "step": 160600
    },
    {
      "epoch": 2.627996400229076,
      "grad_norm": 0.08952823281288147,
      "learning_rate": 4.611241425424179e-07,
      "loss": 0.0006,
      "step": 160610
    },
    {
      "epoch": 2.628160026180152,
      "grad_norm": 0.02707403525710106,
      "learning_rate": 4.607249268044789e-07,
      "loss": 0.0008,
      "step": 160620
    },
    {
      "epoch": 2.628323652131228,
      "grad_norm": 0.06539997458457947,
      "learning_rate": 4.603258756043499e-07,
      "loss": 0.0011,
      "step": 160630
    },
    {
      "epoch": 2.6284872780823036,
      "grad_norm": 0.24186575412750244,
      "learning_rate": 4.5992698895649723e-07,
      "loss": 0.0017,
      "step": 160640
    },
    {
      "epoch": 2.6286509040333796,
      "grad_norm": 0.08850037306547165,
      "learning_rate": 4.5952826687538033e-07,
      "loss": 0.001,
      "step": 160650
    },
    {
      "epoch": 2.6288145299844556,
      "grad_norm": 0.14397820830345154,
      "learning_rate": 4.5912970937545044e-07,
      "loss": 0.0008,
      "step": 160660
    },
    {
      "epoch": 2.628978155935531,
      "grad_norm": 0.10474473237991333,
      "learning_rate": 4.5873131647115545e-07,
      "loss": 0.0006,
      "step": 160670
    },
    {
      "epoch": 2.629141781886607,
      "grad_norm": 0.07800954580307007,
      "learning_rate": 4.583330881769349e-07,
      "loss": 0.0008,
      "step": 160680
    },
    {
      "epoch": 2.629305407837683,
      "grad_norm": 0.14406746625900269,
      "learning_rate": 4.57935024507225e-07,
      "loss": 0.0012,
      "step": 160690
    },
    {
      "epoch": 2.6294690337887587,
      "grad_norm": 0.267629474401474,
      "learning_rate": 4.575371254764538e-07,
      "loss": 0.0008,
      "step": 160700
    },
    {
      "epoch": 2.6296326597398347,
      "grad_norm": 0.06929265707731247,
      "learning_rate": 4.571393910990435e-07,
      "loss": 0.0008,
      "step": 160710
    },
    {
      "epoch": 2.6297962856909107,
      "grad_norm": 0.28277674317359924,
      "learning_rate": 4.5674182138941317e-07,
      "loss": 0.0008,
      "step": 160720
    },
    {
      "epoch": 2.6299599116419863,
      "grad_norm": 0.11098241060972214,
      "learning_rate": 4.563444163619707e-07,
      "loss": 0.0009,
      "step": 160730
    },
    {
      "epoch": 2.6301235375930623,
      "grad_norm": 0.0017275033751502633,
      "learning_rate": 4.5594717603112404e-07,
      "loss": 0.0005,
      "step": 160740
    },
    {
      "epoch": 2.6302871635441383,
      "grad_norm": 0.11607204377651215,
      "learning_rate": 4.5555010041127004e-07,
      "loss": 0.001,
      "step": 160750
    },
    {
      "epoch": 2.630450789495214,
      "grad_norm": 0.034611113369464874,
      "learning_rate": 4.551531895168032e-07,
      "loss": 0.0008,
      "step": 160760
    },
    {
      "epoch": 2.63061441544629,
      "grad_norm": 0.009676181711256504,
      "learning_rate": 4.5475644336210933e-07,
      "loss": 0.0004,
      "step": 160770
    },
    {
      "epoch": 2.630778041397366,
      "grad_norm": 0.0855923593044281,
      "learning_rate": 4.5435986196157136e-07,
      "loss": 0.0009,
      "step": 160780
    },
    {
      "epoch": 2.6309416673484414,
      "grad_norm": 0.04754136502742767,
      "learning_rate": 4.539634453295627e-07,
      "loss": 0.001,
      "step": 160790
    },
    {
      "epoch": 2.6311052932995174,
      "grad_norm": 0.15470431745052338,
      "learning_rate": 4.5356719348045253e-07,
      "loss": 0.001,
      "step": 160800
    },
    {
      "epoch": 2.6312689192505934,
      "grad_norm": 0.09109357744455338,
      "learning_rate": 4.531711064286054e-07,
      "loss": 0.0008,
      "step": 160810
    },
    {
      "epoch": 2.631432545201669,
      "grad_norm": 0.07098764181137085,
      "learning_rate": 4.5277518418837654e-07,
      "loss": 0.0021,
      "step": 160820
    },
    {
      "epoch": 2.631596171152745,
      "grad_norm": 0.001638766610994935,
      "learning_rate": 4.523794267741194e-07,
      "loss": 0.0008,
      "step": 160830
    },
    {
      "epoch": 2.631759797103821,
      "grad_norm": 0.0016671274788677692,
      "learning_rate": 4.51983834200177e-07,
      "loss": 0.0009,
      "step": 160840
    },
    {
      "epoch": 2.6319234230548965,
      "grad_norm": 0.05554373562335968,
      "learning_rate": 4.5158840648089006e-07,
      "loss": 0.0011,
      "step": 160850
    },
    {
      "epoch": 2.6320870490059725,
      "grad_norm": 0.06322480738162994,
      "learning_rate": 4.511931436305922e-07,
      "loss": 0.0006,
      "step": 160860
    },
    {
      "epoch": 2.6322506749570485,
      "grad_norm": 0.1607675850391388,
      "learning_rate": 4.5079804566360906e-07,
      "loss": 0.001,
      "step": 160870
    },
    {
      "epoch": 2.632414300908124,
      "grad_norm": 0.15252161026000977,
      "learning_rate": 4.5040311259426426e-07,
      "loss": 0.0009,
      "step": 160880
    },
    {
      "epoch": 2.6325779268592,
      "grad_norm": 0.0909084603190422,
      "learning_rate": 4.500083444368719e-07,
      "loss": 0.0005,
      "step": 160890
    },
    {
      "epoch": 2.6327415528102756,
      "grad_norm": 0.19390174746513367,
      "learning_rate": 4.496137412057411e-07,
      "loss": 0.0012,
      "step": 160900
    },
    {
      "epoch": 2.6329051787613516,
      "grad_norm": 0.010709865018725395,
      "learning_rate": 4.492193029151753e-07,
      "loss": 0.0007,
      "step": 160910
    },
    {
      "epoch": 2.6330688047124275,
      "grad_norm": 0.0480482317507267,
      "learning_rate": 4.488250295794716e-07,
      "loss": 0.0008,
      "step": 160920
    },
    {
      "epoch": 2.633232430663503,
      "grad_norm": 0.014920058660209179,
      "learning_rate": 4.4843092121292344e-07,
      "loss": 0.0007,
      "step": 160930
    },
    {
      "epoch": 2.633396056614579,
      "grad_norm": 0.014779975637793541,
      "learning_rate": 4.480369778298144e-07,
      "loss": 0.0004,
      "step": 160940
    },
    {
      "epoch": 2.6335596825656546,
      "grad_norm": 0.03539162874221802,
      "learning_rate": 4.476431994444247e-07,
      "loss": 0.0007,
      "step": 160950
    },
    {
      "epoch": 2.6337233085167306,
      "grad_norm": 0.014194296672940254,
      "learning_rate": 4.4724958607102744e-07,
      "loss": 0.0007,
      "step": 160960
    },
    {
      "epoch": 2.6338869344678066,
      "grad_norm": 0.02027301676571369,
      "learning_rate": 4.468561377238911e-07,
      "loss": 0.0005,
      "step": 160970
    },
    {
      "epoch": 2.634050560418882,
      "grad_norm": 0.09363052248954773,
      "learning_rate": 4.4646285441727555e-07,
      "loss": 0.0008,
      "step": 160980
    },
    {
      "epoch": 2.634214186369958,
      "grad_norm": 0.003301275661215186,
      "learning_rate": 4.460697361654387e-07,
      "loss": 0.0009,
      "step": 160990
    },
    {
      "epoch": 2.634377812321034,
      "grad_norm": 0.12710069119930267,
      "learning_rate": 4.456767829826281e-07,
      "loss": 0.0005,
      "step": 161000
    },
    {
      "epoch": 2.6345414382721097,
      "grad_norm": 0.03920842334628105,
      "learning_rate": 4.452839948830878e-07,
      "loss": 0.0014,
      "step": 161010
    },
    {
      "epoch": 2.6347050642231857,
      "grad_norm": 0.10991855710744858,
      "learning_rate": 4.4489137188105657e-07,
      "loss": 0.0008,
      "step": 161020
    },
    {
      "epoch": 2.6348686901742617,
      "grad_norm": 0.126530721783638,
      "learning_rate": 4.444989139907641e-07,
      "loss": 0.001,
      "step": 161030
    },
    {
      "epoch": 2.6350323161253373,
      "grad_norm": 0.0752364844083786,
      "learning_rate": 4.441066212264378e-07,
      "loss": 0.0016,
      "step": 161040
    },
    {
      "epoch": 2.6351959420764133,
      "grad_norm": 0.09029126167297363,
      "learning_rate": 4.4371449360229637e-07,
      "loss": 0.0005,
      "step": 161050
    },
    {
      "epoch": 2.6353595680274893,
      "grad_norm": 0.13464708626270294,
      "learning_rate": 4.433225311325545e-07,
      "loss": 0.0006,
      "step": 161060
    },
    {
      "epoch": 2.635523193978565,
      "grad_norm": 0.10298698395490646,
      "learning_rate": 4.429307338314182e-07,
      "loss": 0.001,
      "step": 161070
    },
    {
      "epoch": 2.635686819929641,
      "grad_norm": 0.03460034355521202,
      "learning_rate": 4.425391017130898e-07,
      "loss": 0.0011,
      "step": 161080
    },
    {
      "epoch": 2.635850445880717,
      "grad_norm": 0.016474898904561996,
      "learning_rate": 4.4214763479176804e-07,
      "loss": 0.0002,
      "step": 161090
    },
    {
      "epoch": 2.6360140718317924,
      "grad_norm": 0.011465839110314846,
      "learning_rate": 4.4175633308163766e-07,
      "loss": 0.0008,
      "step": 161100
    },
    {
      "epoch": 2.6361776977828684,
      "grad_norm": 0.2227979302406311,
      "learning_rate": 4.413651965968857e-07,
      "loss": 0.0012,
      "step": 161110
    },
    {
      "epoch": 2.6363413237339444,
      "grad_norm": 0.051704708486795425,
      "learning_rate": 4.40974225351688e-07,
      "loss": 0.0004,
      "step": 161120
    },
    {
      "epoch": 2.63650494968502,
      "grad_norm": 0.04620182514190674,
      "learning_rate": 4.4058341936021707e-07,
      "loss": 0.0007,
      "step": 161130
    },
    {
      "epoch": 2.636668575636096,
      "grad_norm": 0.036082345992326736,
      "learning_rate": 4.4019277863664054e-07,
      "loss": 0.0004,
      "step": 161140
    },
    {
      "epoch": 2.636832201587172,
      "grad_norm": 0.009952438995242119,
      "learning_rate": 4.3980230319511483e-07,
      "loss": 0.0005,
      "step": 161150
    },
    {
      "epoch": 2.6369958275382475,
      "grad_norm": 0.026040753349661827,
      "learning_rate": 4.39411993049797e-07,
      "loss": 0.0008,
      "step": 161160
    },
    {
      "epoch": 2.6371594534893235,
      "grad_norm": 0.07812991738319397,
      "learning_rate": 4.390218482148323e-07,
      "loss": 0.0006,
      "step": 161170
    },
    {
      "epoch": 2.6373230794403995,
      "grad_norm": 0.07248272001743317,
      "learning_rate": 4.386318687043645e-07,
      "loss": 0.0007,
      "step": 161180
    },
    {
      "epoch": 2.637486705391475,
      "grad_norm": 0.02122293971478939,
      "learning_rate": 4.382420545325283e-07,
      "loss": 0.0004,
      "step": 161190
    },
    {
      "epoch": 2.637650331342551,
      "grad_norm": 0.045053280889987946,
      "learning_rate": 4.378524057134531e-07,
      "loss": 0.0004,
      "step": 161200
    },
    {
      "epoch": 2.637813957293627,
      "grad_norm": 0.07399696856737137,
      "learning_rate": 4.374629222612642e-07,
      "loss": 0.0008,
      "step": 161210
    },
    {
      "epoch": 2.6379775832447026,
      "grad_norm": 0.04394274204969406,
      "learning_rate": 4.370736041900775e-07,
      "loss": 0.0004,
      "step": 161220
    },
    {
      "epoch": 2.6381412091957785,
      "grad_norm": 0.07947829365730286,
      "learning_rate": 4.366844515140073e-07,
      "loss": 0.0006,
      "step": 161230
    },
    {
      "epoch": 2.6383048351468545,
      "grad_norm": 0.002457377966493368,
      "learning_rate": 4.362954642471567e-07,
      "loss": 0.0004,
      "step": 161240
    },
    {
      "epoch": 2.63846846109793,
      "grad_norm": 0.1295071840286255,
      "learning_rate": 4.359066424036279e-07,
      "loss": 0.0005,
      "step": 161250
    },
    {
      "epoch": 2.638632087049006,
      "grad_norm": 0.08755620568990707,
      "learning_rate": 4.355179859975134e-07,
      "loss": 0.0006,
      "step": 161260
    },
    {
      "epoch": 2.638795713000082,
      "grad_norm": 0.057457584887742996,
      "learning_rate": 4.3512949504290193e-07,
      "loss": 0.0004,
      "step": 161270
    },
    {
      "epoch": 2.6389593389511576,
      "grad_norm": 0.03893290087580681,
      "learning_rate": 4.347411695538739e-07,
      "loss": 0.0008,
      "step": 161280
    },
    {
      "epoch": 2.6391229649022336,
      "grad_norm": 0.2828652262687683,
      "learning_rate": 4.34353009544507e-07,
      "loss": 0.0014,
      "step": 161290
    },
    {
      "epoch": 2.639286590853309,
      "grad_norm": 0.24717114865779877,
      "learning_rate": 4.3396501502887047e-07,
      "loss": 0.0013,
      "step": 161300
    },
    {
      "epoch": 2.639450216804385,
      "grad_norm": 0.0561048686504364,
      "learning_rate": 4.335771860210275e-07,
      "loss": 0.0003,
      "step": 161310
    },
    {
      "epoch": 2.639613842755461,
      "grad_norm": 0.11476080119609833,
      "learning_rate": 4.331895225350369e-07,
      "loss": 0.0007,
      "step": 161320
    },
    {
      "epoch": 2.6397774687065367,
      "grad_norm": 0.13832363486289978,
      "learning_rate": 4.3280202458494904e-07,
      "loss": 0.0025,
      "step": 161330
    },
    {
      "epoch": 2.6399410946576127,
      "grad_norm": 0.0995819941163063,
      "learning_rate": 4.324146921848116e-07,
      "loss": 0.0013,
      "step": 161340
    },
    {
      "epoch": 2.6401047206086883,
      "grad_norm": 0.05939680337905884,
      "learning_rate": 4.3202752534866333e-07,
      "loss": 0.001,
      "step": 161350
    },
    {
      "epoch": 2.6402683465597643,
      "grad_norm": 0.019526952877640724,
      "learning_rate": 4.3164052409053804e-07,
      "loss": 0.001,
      "step": 161360
    },
    {
      "epoch": 2.6404319725108403,
      "grad_norm": 0.008612528443336487,
      "learning_rate": 4.3125368842446503e-07,
      "loss": 0.0009,
      "step": 161370
    },
    {
      "epoch": 2.640595598461916,
      "grad_norm": 0.08374768495559692,
      "learning_rate": 4.308670183644642e-07,
      "loss": 0.0009,
      "step": 161380
    },
    {
      "epoch": 2.640759224412992,
      "grad_norm": 0.11215036362409592,
      "learning_rate": 4.304805139245538e-07,
      "loss": 0.0008,
      "step": 161390
    },
    {
      "epoch": 2.640922850364068,
      "grad_norm": 0.23724402487277985,
      "learning_rate": 4.3009417511874096e-07,
      "loss": 0.0008,
      "step": 161400
    },
    {
      "epoch": 2.6410864763151434,
      "grad_norm": 0.1612662822008133,
      "learning_rate": 4.2970800196103056e-07,
      "loss": 0.0008,
      "step": 161410
    },
    {
      "epoch": 2.6412501022662194,
      "grad_norm": 0.19655001163482666,
      "learning_rate": 4.2932199446542143e-07,
      "loss": 0.0014,
      "step": 161420
    },
    {
      "epoch": 2.6414137282172954,
      "grad_norm": 0.008041562512516975,
      "learning_rate": 4.289361526459035e-07,
      "loss": 0.0005,
      "step": 161430
    },
    {
      "epoch": 2.641577354168371,
      "grad_norm": 0.029825009405612946,
      "learning_rate": 4.285504765164655e-07,
      "loss": 0.0004,
      "step": 161440
    },
    {
      "epoch": 2.641740980119447,
      "grad_norm": 0.13402365148067474,
      "learning_rate": 4.2816496609108416e-07,
      "loss": 0.0011,
      "step": 161450
    },
    {
      "epoch": 2.641904606070523,
      "grad_norm": 0.004117707721889019,
      "learning_rate": 4.277796213837354e-07,
      "loss": 0.0004,
      "step": 161460
    },
    {
      "epoch": 2.6420682320215985,
      "grad_norm": 0.14433643221855164,
      "learning_rate": 4.273944424083859e-07,
      "loss": 0.0009,
      "step": 161470
    },
    {
      "epoch": 2.6422318579726745,
      "grad_norm": 0.062206171452999115,
      "learning_rate": 4.2700942917899834e-07,
      "loss": 0.0004,
      "step": 161480
    },
    {
      "epoch": 2.6423954839237505,
      "grad_norm": 0.054306913167238235,
      "learning_rate": 4.2662458170952825e-07,
      "loss": 0.0006,
      "step": 161490
    },
    {
      "epoch": 2.642559109874826,
      "grad_norm": 0.009300809353590012,
      "learning_rate": 4.2623990001392447e-07,
      "loss": 0.0003,
      "step": 161500
    },
    {
      "epoch": 2.642722735825902,
      "grad_norm": 0.007989195175468922,
      "learning_rate": 4.2585538410613245e-07,
      "loss": 0.0007,
      "step": 161510
    },
    {
      "epoch": 2.642886361776978,
      "grad_norm": 0.05148007348179817,
      "learning_rate": 4.2547103400008783e-07,
      "loss": 0.0008,
      "step": 161520
    },
    {
      "epoch": 2.6430499877280536,
      "grad_norm": 0.018591446802020073,
      "learning_rate": 4.250868497097249e-07,
      "loss": 0.0004,
      "step": 161530
    },
    {
      "epoch": 2.6432136136791295,
      "grad_norm": 0.018516890704631805,
      "learning_rate": 4.2470283124896705e-07,
      "loss": 0.0004,
      "step": 161540
    },
    {
      "epoch": 2.6433772396302055,
      "grad_norm": 0.1172805055975914,
      "learning_rate": 4.243189786317364e-07,
      "loss": 0.001,
      "step": 161550
    },
    {
      "epoch": 2.643540865581281,
      "grad_norm": 0.05946112796664238,
      "learning_rate": 4.239352918719442e-07,
      "loss": 0.0004,
      "step": 161560
    },
    {
      "epoch": 2.643704491532357,
      "grad_norm": 0.0028226361609995365,
      "learning_rate": 4.2355177098349966e-07,
      "loss": 0.0004,
      "step": 161570
    },
    {
      "epoch": 2.643868117483433,
      "grad_norm": 0.06161634251475334,
      "learning_rate": 4.2316841598030455e-07,
      "loss": 0.0004,
      "step": 161580
    },
    {
      "epoch": 2.6440317434345086,
      "grad_norm": 0.005187698174268007,
      "learning_rate": 4.22785226876255e-07,
      "loss": 0.0004,
      "step": 161590
    },
    {
      "epoch": 2.6441953693855846,
      "grad_norm": 0.02173246443271637,
      "learning_rate": 4.224022036852393e-07,
      "loss": 0.0003,
      "step": 161600
    },
    {
      "epoch": 2.6443589953366606,
      "grad_norm": 0.006877606734633446,
      "learning_rate": 4.220193464211414e-07,
      "loss": 0.0008,
      "step": 161610
    },
    {
      "epoch": 2.644522621287736,
      "grad_norm": 0.047592807561159134,
      "learning_rate": 4.2163665509783956e-07,
      "loss": 0.0006,
      "step": 161620
    },
    {
      "epoch": 2.644686247238812,
      "grad_norm": 0.11689148098230362,
      "learning_rate": 4.21254129729205e-07,
      "loss": 0.0006,
      "step": 161630
    },
    {
      "epoch": 2.644849873189888,
      "grad_norm": 0.07109644263982773,
      "learning_rate": 4.2087177032910263e-07,
      "loss": 0.0005,
      "step": 161640
    },
    {
      "epoch": 2.6450134991409637,
      "grad_norm": 0.039288464933633804,
      "learning_rate": 4.204895769113948e-07,
      "loss": 0.0008,
      "step": 161650
    },
    {
      "epoch": 2.6451771250920397,
      "grad_norm": 0.059210970997810364,
      "learning_rate": 4.201075494899315e-07,
      "loss": 0.0008,
      "step": 161660
    },
    {
      "epoch": 2.6453407510431153,
      "grad_norm": 0.11788593232631683,
      "learning_rate": 4.197256880785633e-07,
      "loss": 0.0011,
      "step": 161670
    },
    {
      "epoch": 2.6455043769941913,
      "grad_norm": 0.17615070939064026,
      "learning_rate": 4.193439926911297e-07,
      "loss": 0.0017,
      "step": 161680
    },
    {
      "epoch": 2.6456680029452673,
      "grad_norm": 0.053512394428253174,
      "learning_rate": 4.1896246334146797e-07,
      "loss": 0.0014,
      "step": 161690
    },
    {
      "epoch": 2.645831628896343,
      "grad_norm": 0.02111648954451084,
      "learning_rate": 4.1858110004340647e-07,
      "loss": 0.0003,
      "step": 161700
    },
    {
      "epoch": 2.645995254847419,
      "grad_norm": 0.026846205815672874,
      "learning_rate": 4.1819990281076807e-07,
      "loss": 0.0008,
      "step": 161710
    },
    {
      "epoch": 2.6461588807984944,
      "grad_norm": 0.0030224171932786703,
      "learning_rate": 4.178188716573722e-07,
      "loss": 0.0003,
      "step": 161720
    },
    {
      "epoch": 2.6463225067495704,
      "grad_norm": 0.12042593955993652,
      "learning_rate": 4.174380065970285e-07,
      "loss": 0.0005,
      "step": 161730
    },
    {
      "epoch": 2.6464861327006464,
      "grad_norm": 0.05651644617319107,
      "learning_rate": 4.170573076435436e-07,
      "loss": 0.0014,
      "step": 161740
    },
    {
      "epoch": 2.646649758651722,
      "grad_norm": 0.07848593592643738,
      "learning_rate": 4.166767748107159e-07,
      "loss": 0.0011,
      "step": 161750
    },
    {
      "epoch": 2.646813384602798,
      "grad_norm": 0.028205307200551033,
      "learning_rate": 4.1629640811234053e-07,
      "loss": 0.0004,
      "step": 161760
    },
    {
      "epoch": 2.646977010553874,
      "grad_norm": 0.03122643381357193,
      "learning_rate": 4.1591620756220254e-07,
      "loss": 0.0007,
      "step": 161770
    },
    {
      "epoch": 2.6471406365049495,
      "grad_norm": 0.09759751707315445,
      "learning_rate": 4.1553617317408535e-07,
      "loss": 0.0009,
      "step": 161780
    },
    {
      "epoch": 2.6473042624560255,
      "grad_norm": 0.08184006810188293,
      "learning_rate": 4.1515630496176293e-07,
      "loss": 0.0012,
      "step": 161790
    },
    {
      "epoch": 2.6474678884071015,
      "grad_norm": 0.08442173153162003,
      "learning_rate": 4.147766029390049e-07,
      "loss": 0.0006,
      "step": 161800
    },
    {
      "epoch": 2.647631514358177,
      "grad_norm": 0.02862313576042652,
      "learning_rate": 4.1439706711957517e-07,
      "loss": 0.001,
      "step": 161810
    },
    {
      "epoch": 2.647795140309253,
      "grad_norm": 0.006327215116471052,
      "learning_rate": 4.1401769751723e-07,
      "loss": 0.0005,
      "step": 161820
    },
    {
      "epoch": 2.647958766260329,
      "grad_norm": 0.07080142199993134,
      "learning_rate": 4.136384941457222e-07,
      "loss": 0.0008,
      "step": 161830
    },
    {
      "epoch": 2.6481223922114046,
      "grad_norm": 0.044419772922992706,
      "learning_rate": 4.132594570187948e-07,
      "loss": 0.001,
      "step": 161840
    },
    {
      "epoch": 2.6482860181624805,
      "grad_norm": 0.015733499079942703,
      "learning_rate": 4.1288058615018777e-07,
      "loss": 0.0012,
      "step": 161850
    },
    {
      "epoch": 2.6484496441135565,
      "grad_norm": 0.03354400023818016,
      "learning_rate": 4.125018815536358e-07,
      "loss": 0.0007,
      "step": 161860
    },
    {
      "epoch": 2.648613270064632,
      "grad_norm": 0.16005954146385193,
      "learning_rate": 4.1212334324286453e-07,
      "loss": 0.0009,
      "step": 161870
    },
    {
      "epoch": 2.648776896015708,
      "grad_norm": 0.0617651529610157,
      "learning_rate": 4.1174497123159573e-07,
      "loss": 0.0004,
      "step": 161880
    },
    {
      "epoch": 2.648940521966784,
      "grad_norm": 0.14231514930725098,
      "learning_rate": 4.1136676553354407e-07,
      "loss": 0.0004,
      "step": 161890
    },
    {
      "epoch": 2.6491041479178596,
      "grad_norm": 0.03389750048518181,
      "learning_rate": 4.109887261624179e-07,
      "loss": 0.0011,
      "step": 161900
    },
    {
      "epoch": 2.6492677738689356,
      "grad_norm": 0.08259738236665726,
      "learning_rate": 4.1061085313192196e-07,
      "loss": 0.0008,
      "step": 161910
    },
    {
      "epoch": 2.6494313998200116,
      "grad_norm": 0.09009433537721634,
      "learning_rate": 4.102331464557513e-07,
      "loss": 0.0009,
      "step": 161920
    },
    {
      "epoch": 2.649595025771087,
      "grad_norm": 0.1344250589609146,
      "learning_rate": 4.098556061475989e-07,
      "loss": 0.0008,
      "step": 161930
    },
    {
      "epoch": 2.649758651722163,
      "grad_norm": 0.024069903418421745,
      "learning_rate": 4.094782322211477e-07,
      "loss": 0.0008,
      "step": 161940
    },
    {
      "epoch": 2.649922277673239,
      "grad_norm": 0.07617423683404922,
      "learning_rate": 4.091010246900784e-07,
      "loss": 0.0005,
      "step": 161950
    },
    {
      "epoch": 2.6500859036243147,
      "grad_norm": 0.11341037601232529,
      "learning_rate": 4.087239835680623e-07,
      "loss": 0.0007,
      "step": 161960
    },
    {
      "epoch": 2.6502495295753907,
      "grad_norm": 0.06256689876317978,
      "learning_rate": 4.0834710886876794e-07,
      "loss": 0.0005,
      "step": 161970
    },
    {
      "epoch": 2.6504131555264667,
      "grad_norm": 0.06798188388347626,
      "learning_rate": 4.079704006058538e-07,
      "loss": 0.0004,
      "step": 161980
    },
    {
      "epoch": 2.6505767814775423,
      "grad_norm": 0.009177806787192822,
      "learning_rate": 4.075938587929773e-07,
      "loss": 0.0006,
      "step": 161990
    },
    {
      "epoch": 2.6507404074286183,
      "grad_norm": 0.14366771280765533,
      "learning_rate": 4.0721748344378533e-07,
      "loss": 0.0012,
      "step": 162000
    },
    {
      "epoch": 2.6509040333796943,
      "grad_norm": 0.008799301460385323,
      "learning_rate": 4.068412745719208e-07,
      "loss": 0.0006,
      "step": 162010
    },
    {
      "epoch": 2.65106765933077,
      "grad_norm": 0.005362886004149914,
      "learning_rate": 4.064652321910212e-07,
      "loss": 0.0007,
      "step": 162020
    },
    {
      "epoch": 2.651231285281846,
      "grad_norm": 0.025467636063694954,
      "learning_rate": 4.0608935631471615e-07,
      "loss": 0.001,
      "step": 162030
    },
    {
      "epoch": 2.651394911232922,
      "grad_norm": 0.05472426488995552,
      "learning_rate": 4.057136469566314e-07,
      "loss": 0.0011,
      "step": 162040
    },
    {
      "epoch": 2.6515585371839974,
      "grad_norm": 0.14054687321186066,
      "learning_rate": 4.053381041303839e-07,
      "loss": 0.0007,
      "step": 162050
    },
    {
      "epoch": 2.6517221631350734,
      "grad_norm": 0.06658267229795456,
      "learning_rate": 4.049627278495871e-07,
      "loss": 0.0006,
      "step": 162060
    },
    {
      "epoch": 2.651885789086149,
      "grad_norm": 0.0328206904232502,
      "learning_rate": 4.0458751812784855e-07,
      "loss": 0.0009,
      "step": 162070
    },
    {
      "epoch": 2.652049415037225,
      "grad_norm": 0.0794779509305954,
      "learning_rate": 4.042124749787668e-07,
      "loss": 0.0018,
      "step": 162080
    },
    {
      "epoch": 2.652213040988301,
      "grad_norm": 0.028756428509950638,
      "learning_rate": 4.038375984159382e-07,
      "loss": 0.0007,
      "step": 162090
    },
    {
      "epoch": 2.6523766669393765,
      "grad_norm": 0.027016280218958855,
      "learning_rate": 4.034628884529501e-07,
      "loss": 0.001,
      "step": 162100
    },
    {
      "epoch": 2.6525402928904525,
      "grad_norm": 0.05401135981082916,
      "learning_rate": 4.030883451033846e-07,
      "loss": 0.0008,
      "step": 162110
    },
    {
      "epoch": 2.652703918841528,
      "grad_norm": 0.2249971628189087,
      "learning_rate": 4.027139683808173e-07,
      "loss": 0.001,
      "step": 162120
    },
    {
      "epoch": 2.652867544792604,
      "grad_norm": 0.10034111887216568,
      "learning_rate": 4.023397582988192e-07,
      "loss": 0.0008,
      "step": 162130
    },
    {
      "epoch": 2.65303117074368,
      "grad_norm": 0.02204626426100731,
      "learning_rate": 4.01965714870956e-07,
      "loss": 0.0009,
      "step": 162140
    },
    {
      "epoch": 2.6531947966947556,
      "grad_norm": 0.04960593208670616,
      "learning_rate": 4.0159183811078305e-07,
      "loss": 0.0008,
      "step": 162150
    },
    {
      "epoch": 2.6533584226458315,
      "grad_norm": 0.05953105911612511,
      "learning_rate": 4.0121812803185556e-07,
      "loss": 0.0007,
      "step": 162160
    },
    {
      "epoch": 2.6535220485969075,
      "grad_norm": 0.024794332683086395,
      "learning_rate": 4.008445846477166e-07,
      "loss": 0.0006,
      "step": 162170
    },
    {
      "epoch": 2.653685674547983,
      "grad_norm": 0.09195355325937271,
      "learning_rate": 4.004712079719092e-07,
      "loss": 0.0009,
      "step": 162180
    },
    {
      "epoch": 2.653849300499059,
      "grad_norm": 0.04879765585064888,
      "learning_rate": 4.000979980179642e-07,
      "loss": 0.0006,
      "step": 162190
    },
    {
      "epoch": 2.654012926450135,
      "grad_norm": 0.10736338794231415,
      "learning_rate": 3.9972495479941255e-07,
      "loss": 0.0003,
      "step": 162200
    },
    {
      "epoch": 2.6541765524012106,
      "grad_norm": 0.10588818043470383,
      "learning_rate": 3.9935207832977497e-07,
      "loss": 0.0007,
      "step": 162210
    },
    {
      "epoch": 2.6543401783522866,
      "grad_norm": 0.07515053451061249,
      "learning_rate": 3.989793686225657e-07,
      "loss": 0.0012,
      "step": 162220
    },
    {
      "epoch": 2.6545038043033626,
      "grad_norm": 0.04193764925003052,
      "learning_rate": 3.986068256912973e-07,
      "loss": 0.0005,
      "step": 162230
    },
    {
      "epoch": 2.654667430254438,
      "grad_norm": 0.028369365260004997,
      "learning_rate": 3.982344495494717e-07,
      "loss": 0.0006,
      "step": 162240
    },
    {
      "epoch": 2.654831056205514,
      "grad_norm": 0.05866362899541855,
      "learning_rate": 3.978622402105875e-07,
      "loss": 0.0006,
      "step": 162250
    },
    {
      "epoch": 2.65499468215659,
      "grad_norm": 0.03806088864803314,
      "learning_rate": 3.974901976881357e-07,
      "loss": 0.0007,
      "step": 162260
    },
    {
      "epoch": 2.6551583081076657,
      "grad_norm": 0.02487987093627453,
      "learning_rate": 3.971183219956032e-07,
      "loss": 0.0003,
      "step": 162270
    },
    {
      "epoch": 2.6553219340587417,
      "grad_norm": 0.08773603290319443,
      "learning_rate": 3.9674661314646814e-07,
      "loss": 0.0011,
      "step": 162280
    },
    {
      "epoch": 2.6554855600098177,
      "grad_norm": 0.20498532056808472,
      "learning_rate": 3.9637507115420417e-07,
      "loss": 0.0006,
      "step": 162290
    },
    {
      "epoch": 2.6556491859608933,
      "grad_norm": 0.11393287032842636,
      "learning_rate": 3.960036960322816e-07,
      "loss": 0.0011,
      "step": 162300
    },
    {
      "epoch": 2.6558128119119693,
      "grad_norm": 0.005545942112803459,
      "learning_rate": 3.956324877941575e-07,
      "loss": 0.0006,
      "step": 162310
    },
    {
      "epoch": 2.6559764378630453,
      "grad_norm": 0.05825549364089966,
      "learning_rate": 3.9526144645329054e-07,
      "loss": 0.0008,
      "step": 162320
    },
    {
      "epoch": 2.656140063814121,
      "grad_norm": 0.08447104692459106,
      "learning_rate": 3.948905720231283e-07,
      "loss": 0.001,
      "step": 162330
    },
    {
      "epoch": 2.656303689765197,
      "grad_norm": 0.003699325956404209,
      "learning_rate": 3.9451986451711446e-07,
      "loss": 0.0009,
      "step": 162340
    },
    {
      "epoch": 2.656467315716273,
      "grad_norm": 0.008734593167901039,
      "learning_rate": 3.941493239486871e-07,
      "loss": 0.0009,
      "step": 162350
    },
    {
      "epoch": 2.6566309416673484,
      "grad_norm": 0.056675247848033905,
      "learning_rate": 3.937789503312767e-07,
      "loss": 0.0004,
      "step": 162360
    },
    {
      "epoch": 2.6567945676184244,
      "grad_norm": 0.013299272395670414,
      "learning_rate": 3.9340874367830903e-07,
      "loss": 0.0003,
      "step": 162370
    },
    {
      "epoch": 2.6569581935695004,
      "grad_norm": 0.077885702252388,
      "learning_rate": 3.930387040032019e-07,
      "loss": 0.0005,
      "step": 162380
    },
    {
      "epoch": 2.657121819520576,
      "grad_norm": 0.10574847459793091,
      "learning_rate": 3.9266883131936994e-07,
      "loss": 0.0011,
      "step": 162390
    },
    {
      "epoch": 2.657285445471652,
      "grad_norm": 0.08272770792245865,
      "learning_rate": 3.922991256402198e-07,
      "loss": 0.0009,
      "step": 162400
    },
    {
      "epoch": 2.657449071422728,
      "grad_norm": 0.05639660730957985,
      "learning_rate": 3.9192958697915063e-07,
      "loss": 0.0007,
      "step": 162410
    },
    {
      "epoch": 2.6576126973738035,
      "grad_norm": 0.007139271590858698,
      "learning_rate": 3.915602153495601e-07,
      "loss": 0.0014,
      "step": 162420
    },
    {
      "epoch": 2.6577763233248795,
      "grad_norm": 0.1234506294131279,
      "learning_rate": 3.911910107648348e-07,
      "loss": 0.0007,
      "step": 162430
    },
    {
      "epoch": 2.657939949275955,
      "grad_norm": 0.09703536331653595,
      "learning_rate": 3.908219732383589e-07,
      "loss": 0.0007,
      "step": 162440
    },
    {
      "epoch": 2.658103575227031,
      "grad_norm": 0.08111412823200226,
      "learning_rate": 3.904531027835079e-07,
      "loss": 0.0006,
      "step": 162450
    },
    {
      "epoch": 2.658267201178107,
      "grad_norm": 0.0030443049035966396,
      "learning_rate": 3.900843994136538e-07,
      "loss": 0.0007,
      "step": 162460
    },
    {
      "epoch": 2.6584308271291825,
      "grad_norm": 0.14771559834480286,
      "learning_rate": 3.897158631421605e-07,
      "loss": 0.0011,
      "step": 162470
    },
    {
      "epoch": 2.6585944530802585,
      "grad_norm": 0.1352393627166748,
      "learning_rate": 3.893474939823866e-07,
      "loss": 0.0007,
      "step": 162480
    },
    {
      "epoch": 2.658758079031334,
      "grad_norm": 0.015554625540971756,
      "learning_rate": 3.8897929194768423e-07,
      "loss": 0.001,
      "step": 162490
    },
    {
      "epoch": 2.65892170498241,
      "grad_norm": 0.07386418431997299,
      "learning_rate": 3.886112570514011e-07,
      "loss": 0.0007,
      "step": 162500
    },
    {
      "epoch": 2.659085330933486,
      "grad_norm": 0.1800573319196701,
      "learning_rate": 3.8824338930687654e-07,
      "loss": 0.0006,
      "step": 162510
    },
    {
      "epoch": 2.6592489568845616,
      "grad_norm": 0.006648830138146877,
      "learning_rate": 3.878756887274443e-07,
      "loss": 0.0004,
      "step": 162520
    },
    {
      "epoch": 2.6594125828356376,
      "grad_norm": 0.03640538454055786,
      "learning_rate": 3.8750815532643425e-07,
      "loss": 0.0003,
      "step": 162530
    },
    {
      "epoch": 2.6595762087867136,
      "grad_norm": 0.19106590747833252,
      "learning_rate": 3.871407891171669e-07,
      "loss": 0.0008,
      "step": 162540
    },
    {
      "epoch": 2.659739834737789,
      "grad_norm": 0.016939766705036163,
      "learning_rate": 3.867735901129599e-07,
      "loss": 0.0003,
      "step": 162550
    },
    {
      "epoch": 2.659903460688865,
      "grad_norm": 0.004801356699317694,
      "learning_rate": 3.8640655832712204e-07,
      "loss": 0.0006,
      "step": 162560
    },
    {
      "epoch": 2.660067086639941,
      "grad_norm": 0.05608263984322548,
      "learning_rate": 3.8603969377295767e-07,
      "loss": 0.0007,
      "step": 162570
    },
    {
      "epoch": 2.6602307125910167,
      "grad_norm": 0.06587318331003189,
      "learning_rate": 3.856729964637662e-07,
      "loss": 0.0007,
      "step": 162580
    },
    {
      "epoch": 2.6603943385420927,
      "grad_norm": 0.04565051943063736,
      "learning_rate": 3.8530646641283755e-07,
      "loss": 0.0006,
      "step": 162590
    },
    {
      "epoch": 2.6605579644931687,
      "grad_norm": 0.11791886389255524,
      "learning_rate": 3.849401036334599e-07,
      "loss": 0.001,
      "step": 162600
    },
    {
      "epoch": 2.6607215904442443,
      "grad_norm": 0.5547646880149841,
      "learning_rate": 3.845739081389094e-07,
      "loss": 0.0004,
      "step": 162610
    },
    {
      "epoch": 2.6608852163953203,
      "grad_norm": 0.07615247368812561,
      "learning_rate": 3.8420787994246254e-07,
      "loss": 0.0006,
      "step": 162620
    },
    {
      "epoch": 2.6610488423463963,
      "grad_norm": 0.05546354502439499,
      "learning_rate": 3.838420190573866e-07,
      "loss": 0.0008,
      "step": 162630
    },
    {
      "epoch": 2.661212468297472,
      "grad_norm": 0.009736438281834126,
      "learning_rate": 3.83476325496942e-07,
      "loss": 0.0009,
      "step": 162640
    },
    {
      "epoch": 2.661376094248548,
      "grad_norm": 0.1731843650341034,
      "learning_rate": 3.8311079927438656e-07,
      "loss": 0.0009,
      "step": 162650
    },
    {
      "epoch": 2.661539720199624,
      "grad_norm": 0.1371946781873703,
      "learning_rate": 3.827454404029668e-07,
      "loss": 0.001,
      "step": 162660
    },
    {
      "epoch": 2.6617033461506994,
      "grad_norm": 0.0035603567957878113,
      "learning_rate": 3.8238024889592884e-07,
      "loss": 0.0005,
      "step": 162670
    },
    {
      "epoch": 2.6618669721017754,
      "grad_norm": 0.05154554545879364,
      "learning_rate": 3.820152247665071e-07,
      "loss": 0.0014,
      "step": 162680
    },
    {
      "epoch": 2.6620305980528514,
      "grad_norm": 0.004191881977021694,
      "learning_rate": 3.81650368027936e-07,
      "loss": 0.0008,
      "step": 162690
    },
    {
      "epoch": 2.662194224003927,
      "grad_norm": 0.022267114371061325,
      "learning_rate": 3.812856786934388e-07,
      "loss": 0.0007,
      "step": 162700
    },
    {
      "epoch": 2.662357849955003,
      "grad_norm": 0.07482533901929855,
      "learning_rate": 3.8092115677623445e-07,
      "loss": 0.0006,
      "step": 162710
    },
    {
      "epoch": 2.662521475906079,
      "grad_norm": 0.08196624368429184,
      "learning_rate": 3.8055680228953673e-07,
      "loss": 0.0011,
      "step": 162720
    },
    {
      "epoch": 2.6626851018571545,
      "grad_norm": 0.12013554573059082,
      "learning_rate": 3.8019261524655184e-07,
      "loss": 0.0005,
      "step": 162730
    },
    {
      "epoch": 2.6628487278082305,
      "grad_norm": 0.005785492714494467,
      "learning_rate": 3.798285956604825e-07,
      "loss": 0.001,
      "step": 162740
    },
    {
      "epoch": 2.6630123537593064,
      "grad_norm": 0.34245383739471436,
      "learning_rate": 3.794647435445209e-07,
      "loss": 0.0006,
      "step": 162750
    },
    {
      "epoch": 2.663175979710382,
      "grad_norm": 0.027923041954636574,
      "learning_rate": 3.791010589118582e-07,
      "loss": 0.001,
      "step": 162760
    },
    {
      "epoch": 2.663339605661458,
      "grad_norm": 0.08423155546188354,
      "learning_rate": 3.7873754177567555e-07,
      "loss": 0.0007,
      "step": 162770
    },
    {
      "epoch": 2.663503231612534,
      "grad_norm": 0.027789434418082237,
      "learning_rate": 3.783741921491496e-07,
      "loss": 0.0007,
      "step": 162780
    },
    {
      "epoch": 2.6636668575636095,
      "grad_norm": 0.003839026903733611,
      "learning_rate": 3.780110100454526e-07,
      "loss": 0.001,
      "step": 162790
    },
    {
      "epoch": 2.6638304835146855,
      "grad_norm": 0.06512545794248581,
      "learning_rate": 3.776479954777473e-07,
      "loss": 0.0007,
      "step": 162800
    },
    {
      "epoch": 2.6639941094657615,
      "grad_norm": 0.09297475963830948,
      "learning_rate": 3.772851484591933e-07,
      "loss": 0.0005,
      "step": 162810
    },
    {
      "epoch": 2.664157735416837,
      "grad_norm": 0.026656849309802055,
      "learning_rate": 3.7692246900294103e-07,
      "loss": 0.0006,
      "step": 162820
    },
    {
      "epoch": 2.664321361367913,
      "grad_norm": 0.11279799789190292,
      "learning_rate": 3.765599571221379e-07,
      "loss": 0.0009,
      "step": 162830
    },
    {
      "epoch": 2.6644849873189886,
      "grad_norm": 0.013835698366165161,
      "learning_rate": 3.761976128299255e-07,
      "loss": 0.0005,
      "step": 162840
    },
    {
      "epoch": 2.6646486132700646,
      "grad_norm": 0.16747301816940308,
      "learning_rate": 3.758354361394351e-07,
      "loss": 0.0013,
      "step": 162850
    },
    {
      "epoch": 2.6648122392211406,
      "grad_norm": 0.1294667273759842,
      "learning_rate": 3.7547342706379773e-07,
      "loss": 0.0005,
      "step": 162860
    },
    {
      "epoch": 2.664975865172216,
      "grad_norm": 0.021930944174528122,
      "learning_rate": 3.7511158561613246e-07,
      "loss": 0.0011,
      "step": 162870
    },
    {
      "epoch": 2.665139491123292,
      "grad_norm": 0.033442918211221695,
      "learning_rate": 3.7474991180955824e-07,
      "loss": 0.0006,
      "step": 162880
    },
    {
      "epoch": 2.6653031170743677,
      "grad_norm": 0.15374310314655304,
      "learning_rate": 3.743884056571817e-07,
      "loss": 0.0009,
      "step": 162890
    },
    {
      "epoch": 2.6654667430254437,
      "grad_norm": 0.1310236155986786,
      "learning_rate": 3.740270671721091e-07,
      "loss": 0.001,
      "step": 162900
    },
    {
      "epoch": 2.6656303689765197,
      "grad_norm": 0.0040460145100951195,
      "learning_rate": 3.736658963674372e-07,
      "loss": 0.0005,
      "step": 162910
    },
    {
      "epoch": 2.6657939949275953,
      "grad_norm": 0.017761170864105225,
      "learning_rate": 3.7330489325625606e-07,
      "loss": 0.0006,
      "step": 162920
    },
    {
      "epoch": 2.6659576208786713,
      "grad_norm": 0.12663203477859497,
      "learning_rate": 3.729440578516541e-07,
      "loss": 0.0022,
      "step": 162930
    },
    {
      "epoch": 2.6661212468297473,
      "grad_norm": 0.08255751430988312,
      "learning_rate": 3.725833901667081e-07,
      "loss": 0.0005,
      "step": 162940
    },
    {
      "epoch": 2.666284872780823,
      "grad_norm": 0.023657580837607384,
      "learning_rate": 3.722228902144931e-07,
      "loss": 0.0005,
      "step": 162950
    },
    {
      "epoch": 2.666448498731899,
      "grad_norm": 0.07764837145805359,
      "learning_rate": 3.7186255800807536e-07,
      "loss": 0.0007,
      "step": 162960
    },
    {
      "epoch": 2.666612124682975,
      "grad_norm": 0.07613088935613632,
      "learning_rate": 3.715023935605172e-07,
      "loss": 0.0009,
      "step": 162970
    },
    {
      "epoch": 2.6667757506340504,
      "grad_norm": 0.02817651443183422,
      "learning_rate": 3.7114239688487206e-07,
      "loss": 0.0005,
      "step": 162980
    },
    {
      "epoch": 2.6669393765851264,
      "grad_norm": 0.05277569219470024,
      "learning_rate": 3.7078256799419066e-07,
      "loss": 0.0006,
      "step": 162990
    },
    {
      "epoch": 2.6671030025362024,
      "grad_norm": 0.07029453665018082,
      "learning_rate": 3.704229069015153e-07,
      "loss": 0.0004,
      "step": 163000
    },
    {
      "epoch": 2.667266628487278,
      "grad_norm": 0.030672775581479073,
      "learning_rate": 3.7006341361988165e-07,
      "loss": 0.0002,
      "step": 163010
    },
    {
      "epoch": 2.667430254438354,
      "grad_norm": 0.00485110841691494,
      "learning_rate": 3.6970408816232205e-07,
      "loss": 0.0008,
      "step": 163020
    },
    {
      "epoch": 2.66759388038943,
      "grad_norm": 0.008853784762322903,
      "learning_rate": 3.693449305418606e-07,
      "loss": 0.0003,
      "step": 163030
    },
    {
      "epoch": 2.6677575063405055,
      "grad_norm": 0.0329727828502655,
      "learning_rate": 3.6898594077151627e-07,
      "loss": 0.0009,
      "step": 163040
    },
    {
      "epoch": 2.6679211322915815,
      "grad_norm": 0.027727501466870308,
      "learning_rate": 3.6862711886430036e-07,
      "loss": 0.0004,
      "step": 163050
    },
    {
      "epoch": 2.6680847582426575,
      "grad_norm": 0.05496814101934433,
      "learning_rate": 3.682684648332202e-07,
      "loss": 0.0007,
      "step": 163060
    },
    {
      "epoch": 2.668248384193733,
      "grad_norm": 0.012437095865607262,
      "learning_rate": 3.679099786912771e-07,
      "loss": 0.0004,
      "step": 163070
    },
    {
      "epoch": 2.668412010144809,
      "grad_norm": 0.028985943645238876,
      "learning_rate": 3.675516604514634e-07,
      "loss": 0.0008,
      "step": 163080
    },
    {
      "epoch": 2.668575636095885,
      "grad_norm": 0.03409262374043465,
      "learning_rate": 3.671935101267687e-07,
      "loss": 0.0006,
      "step": 163090
    },
    {
      "epoch": 2.6687392620469605,
      "grad_norm": 0.025195077061653137,
      "learning_rate": 3.6683552773017495e-07,
      "loss": 0.0003,
      "step": 163100
    },
    {
      "epoch": 2.6689028879980365,
      "grad_norm": 0.05438625067472458,
      "learning_rate": 3.664777132746572e-07,
      "loss": 0.0008,
      "step": 163110
    },
    {
      "epoch": 2.6690665139491125,
      "grad_norm": 0.044088296592235565,
      "learning_rate": 3.6612006677318626e-07,
      "loss": 0.0006,
      "step": 163120
    },
    {
      "epoch": 2.669230139900188,
      "grad_norm": 0.07317398488521576,
      "learning_rate": 3.657625882387256e-07,
      "loss": 0.0009,
      "step": 163130
    },
    {
      "epoch": 2.669393765851264,
      "grad_norm": 0.05124928057193756,
      "learning_rate": 3.6540527768423317e-07,
      "loss": 0.0004,
      "step": 163140
    },
    {
      "epoch": 2.66955739180234,
      "grad_norm": 0.11896055191755295,
      "learning_rate": 3.650481351226598e-07,
      "loss": 0.0008,
      "step": 163150
    },
    {
      "epoch": 2.6697210177534156,
      "grad_norm": 0.06174270436167717,
      "learning_rate": 3.6469116056695287e-07,
      "loss": 0.0005,
      "step": 163160
    },
    {
      "epoch": 2.6698846437044916,
      "grad_norm": 0.02606838010251522,
      "learning_rate": 3.6433435403004923e-07,
      "loss": 0.001,
      "step": 163170
    },
    {
      "epoch": 2.6700482696555676,
      "grad_norm": 0.020135827362537384,
      "learning_rate": 3.639777155248853e-07,
      "loss": 0.0005,
      "step": 163180
    },
    {
      "epoch": 2.670211895606643,
      "grad_norm": 0.03413727879524231,
      "learning_rate": 3.6362124506438555e-07,
      "loss": 0.0011,
      "step": 163190
    },
    {
      "epoch": 2.670375521557719,
      "grad_norm": 0.155034601688385,
      "learning_rate": 3.6326494266147314e-07,
      "loss": 0.0008,
      "step": 163200
    },
    {
      "epoch": 2.6705391475087947,
      "grad_norm": 0.0019010301912203431,
      "learning_rate": 3.6290880832906275e-07,
      "loss": 0.0006,
      "step": 163210
    },
    {
      "epoch": 2.6707027734598707,
      "grad_norm": 0.1596842110157013,
      "learning_rate": 3.6255284208006226e-07,
      "loss": 0.0007,
      "step": 163220
    },
    {
      "epoch": 2.6708663994109467,
      "grad_norm": 0.05480702966451645,
      "learning_rate": 3.6219704392737584e-07,
      "loss": 0.0008,
      "step": 163230
    },
    {
      "epoch": 2.6710300253620223,
      "grad_norm": 0.08250179886817932,
      "learning_rate": 3.618414138838994e-07,
      "loss": 0.0004,
      "step": 163240
    },
    {
      "epoch": 2.6711936513130983,
      "grad_norm": 0.07675474882125854,
      "learning_rate": 3.614859519625252e-07,
      "loss": 0.0008,
      "step": 163250
    },
    {
      "epoch": 2.671357277264174,
      "grad_norm": 0.01846032775938511,
      "learning_rate": 3.611306581761359e-07,
      "loss": 0.0004,
      "step": 163260
    },
    {
      "epoch": 2.67152090321525,
      "grad_norm": 0.10128302127122879,
      "learning_rate": 3.6077553253761056e-07,
      "loss": 0.0017,
      "step": 163270
    },
    {
      "epoch": 2.671684529166326,
      "grad_norm": 0.051420170813798904,
      "learning_rate": 3.604205750598233e-07,
      "loss": 0.0008,
      "step": 163280
    },
    {
      "epoch": 2.6718481551174014,
      "grad_norm": 0.023811087012290955,
      "learning_rate": 3.6006578575563833e-07,
      "loss": 0.0004,
      "step": 163290
    },
    {
      "epoch": 2.6720117810684774,
      "grad_norm": 0.10541754961013794,
      "learning_rate": 3.5971116463791864e-07,
      "loss": 0.0005,
      "step": 163300
    },
    {
      "epoch": 2.6721754070195534,
      "grad_norm": 0.7095283269882202,
      "learning_rate": 3.5935671171951514e-07,
      "loss": 0.0006,
      "step": 163310
    },
    {
      "epoch": 2.672339032970629,
      "grad_norm": 0.1180049404501915,
      "learning_rate": 3.59002427013278e-07,
      "loss": 0.0005,
      "step": 163320
    },
    {
      "epoch": 2.672502658921705,
      "grad_norm": 0.0037099390756338835,
      "learning_rate": 3.586483105320476e-07,
      "loss": 0.0009,
      "step": 163330
    },
    {
      "epoch": 2.672666284872781,
      "grad_norm": 0.05947168916463852,
      "learning_rate": 3.582943622886609e-07,
      "loss": 0.0007,
      "step": 163340
    },
    {
      "epoch": 2.6728299108238565,
      "grad_norm": 0.008914670906960964,
      "learning_rate": 3.579405822959481e-07,
      "loss": 0.0005,
      "step": 163350
    },
    {
      "epoch": 2.6729935367749325,
      "grad_norm": 0.03897206112742424,
      "learning_rate": 3.5758697056673173e-07,
      "loss": 0.0005,
      "step": 163360
    },
    {
      "epoch": 2.6731571627260085,
      "grad_norm": 0.16469772160053253,
      "learning_rate": 3.5723352711383055e-07,
      "loss": 0.0007,
      "step": 163370
    },
    {
      "epoch": 2.673320788677084,
      "grad_norm": 0.021919772028923035,
      "learning_rate": 3.568802519500547e-07,
      "loss": 0.0007,
      "step": 163380
    },
    {
      "epoch": 2.67348441462816,
      "grad_norm": 0.11993957310914993,
      "learning_rate": 3.5652714508821075e-07,
      "loss": 0.0008,
      "step": 163390
    },
    {
      "epoch": 2.673648040579236,
      "grad_norm": 0.02776331827044487,
      "learning_rate": 3.561742065410978e-07,
      "loss": 0.0003,
      "step": 163400
    },
    {
      "epoch": 2.6738116665303115,
      "grad_norm": 0.09428710490465164,
      "learning_rate": 3.558214363215079e-07,
      "loss": 0.0013,
      "step": 163410
    },
    {
      "epoch": 2.6739752924813875,
      "grad_norm": 0.023654617369174957,
      "learning_rate": 3.554688344422297e-07,
      "loss": 0.0008,
      "step": 163420
    },
    {
      "epoch": 2.6741389184324635,
      "grad_norm": 0.0558219775557518,
      "learning_rate": 3.5511640091604293e-07,
      "loss": 0.002,
      "step": 163430
    },
    {
      "epoch": 2.674302544383539,
      "grad_norm": 0.15546190738677979,
      "learning_rate": 3.5476413575572354e-07,
      "loss": 0.0005,
      "step": 163440
    },
    {
      "epoch": 2.674466170334615,
      "grad_norm": 0.148259237408638,
      "learning_rate": 3.5441203897403854e-07,
      "loss": 0.001,
      "step": 163450
    },
    {
      "epoch": 2.674629796285691,
      "grad_norm": 0.052620917558670044,
      "learning_rate": 3.540601105837532e-07,
      "loss": 0.0005,
      "step": 163460
    },
    {
      "epoch": 2.6747934222367666,
      "grad_norm": 0.056272465735673904,
      "learning_rate": 3.537083505976213e-07,
      "loss": 0.0005,
      "step": 163470
    },
    {
      "epoch": 2.6749570481878426,
      "grad_norm": 0.04841306805610657,
      "learning_rate": 3.5335675902839596e-07,
      "loss": 0.0008,
      "step": 163480
    },
    {
      "epoch": 2.6751206741389186,
      "grad_norm": 0.037078436464071274,
      "learning_rate": 3.530053358888186e-07,
      "loss": 0.0006,
      "step": 163490
    },
    {
      "epoch": 2.675284300089994,
      "grad_norm": 0.08532621711492538,
      "learning_rate": 3.526540811916307e-07,
      "loss": 0.0004,
      "step": 163500
    },
    {
      "epoch": 2.67544792604107,
      "grad_norm": 0.00392522756010294,
      "learning_rate": 3.523029949495621e-07,
      "loss": 0.0004,
      "step": 163510
    },
    {
      "epoch": 2.675611551992146,
      "grad_norm": 0.2651732861995697,
      "learning_rate": 3.519520771753387e-07,
      "loss": 0.0018,
      "step": 163520
    },
    {
      "epoch": 2.6757751779432217,
      "grad_norm": 0.041324179619550705,
      "learning_rate": 3.516013278816827e-07,
      "loss": 0.0004,
      "step": 163530
    },
    {
      "epoch": 2.6759388038942977,
      "grad_norm": 0.06897354871034622,
      "learning_rate": 3.512507470813048e-07,
      "loss": 0.0003,
      "step": 163540
    },
    {
      "epoch": 2.6761024298453737,
      "grad_norm": 0.03266638517379761,
      "learning_rate": 3.50900334786915e-07,
      "loss": 0.0014,
      "step": 163550
    },
    {
      "epoch": 2.6762660557964493,
      "grad_norm": 0.04949835315346718,
      "learning_rate": 3.5055009101121475e-07,
      "loss": 0.0006,
      "step": 163560
    },
    {
      "epoch": 2.6764296817475253,
      "grad_norm": 0.0038154867943376303,
      "learning_rate": 3.5020001576689834e-07,
      "loss": 0.0008,
      "step": 163570
    },
    {
      "epoch": 2.6765933076986013,
      "grad_norm": 0.3559225797653198,
      "learning_rate": 3.498501090666567e-07,
      "loss": 0.001,
      "step": 163580
    },
    {
      "epoch": 2.676756933649677,
      "grad_norm": 0.0827450081706047,
      "learning_rate": 3.495003709231709e-07,
      "loss": 0.0007,
      "step": 163590
    },
    {
      "epoch": 2.676920559600753,
      "grad_norm": 0.005671032704412937,
      "learning_rate": 3.491508013491218e-07,
      "loss": 0.0012,
      "step": 163600
    },
    {
      "epoch": 2.6770841855518284,
      "grad_norm": 0.030812224373221397,
      "learning_rate": 3.48801400357176e-07,
      "loss": 0.001,
      "step": 163610
    },
    {
      "epoch": 2.6772478115029044,
      "grad_norm": 0.10496310889720917,
      "learning_rate": 3.4845216796000115e-07,
      "loss": 0.0008,
      "step": 163620
    },
    {
      "epoch": 2.6774114374539804,
      "grad_norm": 0.06562676280736923,
      "learning_rate": 3.4810310417025594e-07,
      "loss": 0.0007,
      "step": 163630
    },
    {
      "epoch": 2.677575063405056,
      "grad_norm": 0.028749961405992508,
      "learning_rate": 3.4775420900059197e-07,
      "loss": 0.0009,
      "step": 163640
    },
    {
      "epoch": 2.677738689356132,
      "grad_norm": 0.060133472084999084,
      "learning_rate": 3.4740548246365745e-07,
      "loss": 0.0011,
      "step": 163650
    },
    {
      "epoch": 2.6779023153072075,
      "grad_norm": 0.07799925655126572,
      "learning_rate": 3.470569245720906e-07,
      "loss": 0.0004,
      "step": 163660
    },
    {
      "epoch": 2.6780659412582835,
      "grad_norm": 0.0006128168315626681,
      "learning_rate": 3.4670853533852857e-07,
      "loss": 0.0007,
      "step": 163670
    },
    {
      "epoch": 2.6782295672093595,
      "grad_norm": 0.07360197603702545,
      "learning_rate": 3.4636031477559727e-07,
      "loss": 0.0005,
      "step": 163680
    },
    {
      "epoch": 2.678393193160435,
      "grad_norm": 0.05974465608596802,
      "learning_rate": 3.460122628959206e-07,
      "loss": 0.0005,
      "step": 163690
    },
    {
      "epoch": 2.678556819111511,
      "grad_norm": 0.04297952726483345,
      "learning_rate": 3.45664379712114e-07,
      "loss": 0.0006,
      "step": 163700
    },
    {
      "epoch": 2.678720445062587,
      "grad_norm": 0.05019129812717438,
      "learning_rate": 3.453166652367862e-07,
      "loss": 0.0008,
      "step": 163710
    },
    {
      "epoch": 2.6788840710136625,
      "grad_norm": 0.09492860734462738,
      "learning_rate": 3.4496911948254274e-07,
      "loss": 0.0014,
      "step": 163720
    },
    {
      "epoch": 2.6790476969647385,
      "grad_norm": 0.05131504684686661,
      "learning_rate": 3.4462174246198023e-07,
      "loss": 0.0007,
      "step": 163730
    },
    {
      "epoch": 2.6792113229158145,
      "grad_norm": 0.1537952721118927,
      "learning_rate": 3.442745341876913e-07,
      "loss": 0.0005,
      "step": 163740
    },
    {
      "epoch": 2.67937494886689,
      "grad_norm": 0.09487973153591156,
      "learning_rate": 3.439274946722598e-07,
      "loss": 0.0006,
      "step": 163750
    },
    {
      "epoch": 2.679538574817966,
      "grad_norm": 0.032783087342977524,
      "learning_rate": 3.4358062392826684e-07,
      "loss": 0.0006,
      "step": 163760
    },
    {
      "epoch": 2.679702200769042,
      "grad_norm": 0.09356853365898132,
      "learning_rate": 3.432339219682845e-07,
      "loss": 0.0009,
      "step": 163770
    },
    {
      "epoch": 2.6798658267201176,
      "grad_norm": 0.07034482806921005,
      "learning_rate": 3.428873888048795e-07,
      "loss": 0.0005,
      "step": 163780
    },
    {
      "epoch": 2.6800294526711936,
      "grad_norm": 0.0939268171787262,
      "learning_rate": 3.4254102445061497e-07,
      "loss": 0.0007,
      "step": 163790
    },
    {
      "epoch": 2.6801930786222696,
      "grad_norm": 0.09289959073066711,
      "learning_rate": 3.4219482891804434e-07,
      "loss": 0.0004,
      "step": 163800
    },
    {
      "epoch": 2.680356704573345,
      "grad_norm": 0.06619736552238464,
      "learning_rate": 3.418488022197164e-07,
      "loss": 0.0014,
      "step": 163810
    },
    {
      "epoch": 2.680520330524421,
      "grad_norm": 0.014307755045592785,
      "learning_rate": 3.415029443681728e-07,
      "loss": 0.0003,
      "step": 163820
    },
    {
      "epoch": 2.680683956475497,
      "grad_norm": 0.04305620864033699,
      "learning_rate": 3.4115725537595125e-07,
      "loss": 0.0008,
      "step": 163830
    },
    {
      "epoch": 2.6808475824265727,
      "grad_norm": 0.07943262904882431,
      "learning_rate": 3.4081173525558233e-07,
      "loss": 0.0007,
      "step": 163840
    },
    {
      "epoch": 2.6810112083776487,
      "grad_norm": 0.04751698672771454,
      "learning_rate": 3.4046638401958985e-07,
      "loss": 0.0003,
      "step": 163850
    },
    {
      "epoch": 2.6811748343287247,
      "grad_norm": 0.04202309623360634,
      "learning_rate": 3.401212016804922e-07,
      "loss": 0.0006,
      "step": 163860
    },
    {
      "epoch": 2.6813384602798003,
      "grad_norm": 0.1558055579662323,
      "learning_rate": 3.3977618825080103e-07,
      "loss": 0.0008,
      "step": 163870
    },
    {
      "epoch": 2.6815020862308763,
      "grad_norm": 0.045074786990880966,
      "learning_rate": 3.39431343743023e-07,
      "loss": 0.0007,
      "step": 163880
    },
    {
      "epoch": 2.6816657121819523,
      "grad_norm": 0.09941434115171432,
      "learning_rate": 3.390866681696564e-07,
      "loss": 0.0006,
      "step": 163890
    },
    {
      "epoch": 2.681829338133028,
      "grad_norm": 0.03489271551370621,
      "learning_rate": 3.3874216154319686e-07,
      "loss": 0.0007,
      "step": 163900
    },
    {
      "epoch": 2.681992964084104,
      "grad_norm": 0.0354074165225029,
      "learning_rate": 3.3839782387613097e-07,
      "loss": 0.001,
      "step": 163910
    },
    {
      "epoch": 2.68215659003518,
      "grad_norm": 0.035373836755752563,
      "learning_rate": 3.380536551809394e-07,
      "loss": 0.0007,
      "step": 163920
    },
    {
      "epoch": 2.6823202159862554,
      "grad_norm": 0.056779470294713974,
      "learning_rate": 3.377096554700987e-07,
      "loss": 0.0008,
      "step": 163930
    },
    {
      "epoch": 2.6824838419373314,
      "grad_norm": 0.06307129561901093,
      "learning_rate": 3.373658247560768e-07,
      "loss": 0.0006,
      "step": 163940
    },
    {
      "epoch": 2.6826474678884074,
      "grad_norm": 0.053183939307928085,
      "learning_rate": 3.370221630513382e-07,
      "loss": 0.0004,
      "step": 163950
    },
    {
      "epoch": 2.682811093839483,
      "grad_norm": 0.031169995665550232,
      "learning_rate": 3.366786703683383e-07,
      "loss": 0.0005,
      "step": 163960
    },
    {
      "epoch": 2.682974719790559,
      "grad_norm": 0.21075117588043213,
      "learning_rate": 3.3633534671952904e-07,
      "loss": 0.001,
      "step": 163970
    },
    {
      "epoch": 2.683138345741635,
      "grad_norm": 0.0974084883928299,
      "learning_rate": 3.3599219211735423e-07,
      "loss": 0.0006,
      "step": 163980
    },
    {
      "epoch": 2.6833019716927105,
      "grad_norm": 0.05762771889567375,
      "learning_rate": 3.356492065742528e-07,
      "loss": 0.0006,
      "step": 163990
    },
    {
      "epoch": 2.6834655976437864,
      "grad_norm": 0.04072760418057442,
      "learning_rate": 3.353063901026576e-07,
      "loss": 0.0007,
      "step": 164000
    },
    {
      "epoch": 2.683629223594862,
      "grad_norm": 0.0807693675160408,
      "learning_rate": 3.3496374271499486e-07,
      "loss": 0.0004,
      "step": 164010
    },
    {
      "epoch": 2.683792849545938,
      "grad_norm": 0.009045381098985672,
      "learning_rate": 3.34621264423684e-07,
      "loss": 0.0003,
      "step": 164020
    },
    {
      "epoch": 2.6839564754970135,
      "grad_norm": 0.14643540978431702,
      "learning_rate": 3.342789552411385e-07,
      "loss": 0.0012,
      "step": 164030
    },
    {
      "epoch": 2.6841201014480895,
      "grad_norm": 0.005377517081797123,
      "learning_rate": 3.3393681517976784e-07,
      "loss": 0.0008,
      "step": 164040
    },
    {
      "epoch": 2.6842837273991655,
      "grad_norm": 0.08317583054304123,
      "learning_rate": 3.335948442519732e-07,
      "loss": 0.0005,
      "step": 164050
    },
    {
      "epoch": 2.684447353350241,
      "grad_norm": 0.11639754474163055,
      "learning_rate": 3.332530424701491e-07,
      "loss": 0.0015,
      "step": 164060
    },
    {
      "epoch": 2.684610979301317,
      "grad_norm": 0.07140519469976425,
      "learning_rate": 3.3291140984668737e-07,
      "loss": 0.0007,
      "step": 164070
    },
    {
      "epoch": 2.684774605252393,
      "grad_norm": 0.021656684577465057,
      "learning_rate": 3.3256994639396856e-07,
      "loss": 0.0006,
      "step": 164080
    },
    {
      "epoch": 2.6849382312034686,
      "grad_norm": 0.04261285439133644,
      "learning_rate": 3.3222865212437227e-07,
      "loss": 0.0006,
      "step": 164090
    },
    {
      "epoch": 2.6851018571545446,
      "grad_norm": 0.15076954662799835,
      "learning_rate": 3.3188752705026804e-07,
      "loss": 0.0004,
      "step": 164100
    },
    {
      "epoch": 2.6852654831056206,
      "grad_norm": 0.012875348329544067,
      "learning_rate": 3.3154657118402265e-07,
      "loss": 0.0009,
      "step": 164110
    },
    {
      "epoch": 2.685429109056696,
      "grad_norm": 0.07233261317014694,
      "learning_rate": 3.312057845379929e-07,
      "loss": 0.0006,
      "step": 164120
    },
    {
      "epoch": 2.685592735007772,
      "grad_norm": 0.001244951388798654,
      "learning_rate": 3.308651671245322e-07,
      "loss": 0.0013,
      "step": 164130
    },
    {
      "epoch": 2.685756360958848,
      "grad_norm": 0.03029131516814232,
      "learning_rate": 3.3052471895598736e-07,
      "loss": 0.001,
      "step": 164140
    },
    {
      "epoch": 2.6859199869099237,
      "grad_norm": 0.062144286930561066,
      "learning_rate": 3.3018444004469853e-07,
      "loss": 0.0007,
      "step": 164150
    },
    {
      "epoch": 2.6860836128609997,
      "grad_norm": 0.036022696644067764,
      "learning_rate": 3.2984433040300026e-07,
      "loss": 0.0008,
      "step": 164160
    },
    {
      "epoch": 2.6862472388120757,
      "grad_norm": 0.15065248310565948,
      "learning_rate": 3.2950439004322046e-07,
      "loss": 0.0004,
      "step": 164170
    },
    {
      "epoch": 2.6864108647631513,
      "grad_norm": 0.16965873539447784,
      "learning_rate": 3.2916461897768147e-07,
      "loss": 0.0009,
      "step": 164180
    },
    {
      "epoch": 2.6865744907142273,
      "grad_norm": 0.08845871686935425,
      "learning_rate": 3.2882501721869855e-07,
      "loss": 0.0011,
      "step": 164190
    },
    {
      "epoch": 2.6867381166653033,
      "grad_norm": 0.03509719297289848,
      "learning_rate": 3.2848558477858286e-07,
      "loss": 0.0017,
      "step": 164200
    },
    {
      "epoch": 2.686901742616379,
      "grad_norm": 0.09602032601833344,
      "learning_rate": 3.281463216696368e-07,
      "loss": 0.0009,
      "step": 164210
    },
    {
      "epoch": 2.687065368567455,
      "grad_norm": 0.06660565733909607,
      "learning_rate": 3.278072279041572e-07,
      "loss": 0.0004,
      "step": 164220
    },
    {
      "epoch": 2.687228994518531,
      "grad_norm": 0.016373775899410248,
      "learning_rate": 3.2746830349443705e-07,
      "loss": 0.0006,
      "step": 164230
    },
    {
      "epoch": 2.6873926204696064,
      "grad_norm": 0.0728883147239685,
      "learning_rate": 3.2712954845276033e-07,
      "loss": 0.0005,
      "step": 164240
    },
    {
      "epoch": 2.6875562464206824,
      "grad_norm": 0.01966550201177597,
      "learning_rate": 3.267909627914068e-07,
      "loss": 0.0008,
      "step": 164250
    },
    {
      "epoch": 2.6877198723717584,
      "grad_norm": 0.03364823758602142,
      "learning_rate": 3.264525465226487e-07,
      "loss": 0.0004,
      "step": 164260
    },
    {
      "epoch": 2.687883498322834,
      "grad_norm": 0.0681309923529625,
      "learning_rate": 3.2611429965875296e-07,
      "loss": 0.0007,
      "step": 164270
    },
    {
      "epoch": 2.68804712427391,
      "grad_norm": 0.12694673240184784,
      "learning_rate": 3.257762222119815e-07,
      "loss": 0.0011,
      "step": 164280
    },
    {
      "epoch": 2.688210750224986,
      "grad_norm": 0.02490386739373207,
      "learning_rate": 3.254383141945866e-07,
      "loss": 0.0004,
      "step": 164290
    },
    {
      "epoch": 2.6883743761760615,
      "grad_norm": 0.020304597914218903,
      "learning_rate": 3.251005756188191e-07,
      "loss": 0.0011,
      "step": 164300
    },
    {
      "epoch": 2.6885380021271374,
      "grad_norm": 0.012983457185328007,
      "learning_rate": 3.247630064969193e-07,
      "loss": 0.0005,
      "step": 164310
    },
    {
      "epoch": 2.6887016280782134,
      "grad_norm": 0.0048445831052958965,
      "learning_rate": 3.244256068411239e-07,
      "loss": 0.0004,
      "step": 164320
    },
    {
      "epoch": 2.688865254029289,
      "grad_norm": 0.011834460310637951,
      "learning_rate": 3.2408837666366267e-07,
      "loss": 0.0013,
      "step": 164330
    },
    {
      "epoch": 2.689028879980365,
      "grad_norm": 0.026330221444368362,
      "learning_rate": 3.2375131597675967e-07,
      "loss": 0.0005,
      "step": 164340
    },
    {
      "epoch": 2.689192505931441,
      "grad_norm": 0.09093364328145981,
      "learning_rate": 3.2341442479263243e-07,
      "loss": 0.0006,
      "step": 164350
    },
    {
      "epoch": 2.6893561318825165,
      "grad_norm": 0.05825169384479523,
      "learning_rate": 3.230777031234922e-07,
      "loss": 0.0004,
      "step": 164360
    },
    {
      "epoch": 2.6895197578335925,
      "grad_norm": 0.035214781761169434,
      "learning_rate": 3.227411509815448e-07,
      "loss": 0.0005,
      "step": 164370
    },
    {
      "epoch": 2.689683383784668,
      "grad_norm": 0.04588169604539871,
      "learning_rate": 3.224047683789888e-07,
      "loss": 0.0003,
      "step": 164380
    },
    {
      "epoch": 2.689847009735744,
      "grad_norm": 0.04934491962194443,
      "learning_rate": 3.220685553280184e-07,
      "loss": 0.0004,
      "step": 164390
    },
    {
      "epoch": 2.69001063568682,
      "grad_norm": 0.10355619341135025,
      "learning_rate": 3.2173251184081876e-07,
      "loss": 0.0004,
      "step": 164400
    },
    {
      "epoch": 2.6901742616378956,
      "grad_norm": 0.05426101014018059,
      "learning_rate": 3.213966379295724e-07,
      "loss": 0.0008,
      "step": 164410
    },
    {
      "epoch": 2.6903378875889716,
      "grad_norm": 0.0355713777244091,
      "learning_rate": 3.210609336064535e-07,
      "loss": 0.0007,
      "step": 164420
    },
    {
      "epoch": 2.690501513540047,
      "grad_norm": 0.0870102271437645,
      "learning_rate": 3.20725398883629e-07,
      "loss": 0.0004,
      "step": 164430
    },
    {
      "epoch": 2.690665139491123,
      "grad_norm": 0.1024090051651001,
      "learning_rate": 3.203900337732635e-07,
      "loss": 0.0007,
      "step": 164440
    },
    {
      "epoch": 2.690828765442199,
      "grad_norm": 0.020521871745586395,
      "learning_rate": 3.200548382875113e-07,
      "loss": 0.0006,
      "step": 164450
    },
    {
      "epoch": 2.6909923913932747,
      "grad_norm": 0.14649441838264465,
      "learning_rate": 3.197198124385242e-07,
      "loss": 0.0007,
      "step": 164460
    },
    {
      "epoch": 2.6911560173443507,
      "grad_norm": 0.059924088418483734,
      "learning_rate": 3.1938495623844435e-07,
      "loss": 0.0004,
      "step": 164470
    },
    {
      "epoch": 2.6913196432954267,
      "grad_norm": 0.026155566796660423,
      "learning_rate": 3.190502696994102e-07,
      "loss": 0.0005,
      "step": 164480
    },
    {
      "epoch": 2.6914832692465023,
      "grad_norm": 0.12354004383087158,
      "learning_rate": 3.1871575283355383e-07,
      "loss": 0.0015,
      "step": 164490
    },
    {
      "epoch": 2.6916468951975783,
      "grad_norm": 0.11201626062393188,
      "learning_rate": 3.183814056529999e-07,
      "loss": 0.0011,
      "step": 164500
    },
    {
      "epoch": 2.6918105211486543,
      "grad_norm": 0.06511817127466202,
      "learning_rate": 3.180472281698693e-07,
      "loss": 0.0005,
      "step": 164510
    },
    {
      "epoch": 2.69197414709973,
      "grad_norm": 0.06703072041273117,
      "learning_rate": 3.177132203962724e-07,
      "loss": 0.0003,
      "step": 164520
    },
    {
      "epoch": 2.692137773050806,
      "grad_norm": 0.06826429814100266,
      "learning_rate": 3.173793823443189e-07,
      "loss": 0.0006,
      "step": 164530
    },
    {
      "epoch": 2.692301399001882,
      "grad_norm": 0.09261713176965714,
      "learning_rate": 3.170457140261074e-07,
      "loss": 0.0004,
      "step": 164540
    },
    {
      "epoch": 2.6924650249529574,
      "grad_norm": 0.10847348719835281,
      "learning_rate": 3.1671221545373334e-07,
      "loss": 0.0003,
      "step": 164550
    },
    {
      "epoch": 2.6926286509040334,
      "grad_norm": 0.3239886462688446,
      "learning_rate": 3.163788866392864e-07,
      "loss": 0.0008,
      "step": 164560
    },
    {
      "epoch": 2.6927922768551094,
      "grad_norm": 0.025863589718937874,
      "learning_rate": 3.160457275948475e-07,
      "loss": 0.0005,
      "step": 164570
    },
    {
      "epoch": 2.692955902806185,
      "grad_norm": 0.02969971112906933,
      "learning_rate": 3.157127383324937e-07,
      "loss": 0.0005,
      "step": 164580
    },
    {
      "epoch": 2.693119528757261,
      "grad_norm": 0.11970417946577072,
      "learning_rate": 3.153799188642942e-07,
      "loss": 0.0007,
      "step": 164590
    },
    {
      "epoch": 2.693283154708337,
      "grad_norm": 0.09122994542121887,
      "learning_rate": 3.1504726920231435e-07,
      "loss": 0.0011,
      "step": 164600
    },
    {
      "epoch": 2.6934467806594125,
      "grad_norm": 0.0669901967048645,
      "learning_rate": 3.1471478935861056e-07,
      "loss": 0.0006,
      "step": 164610
    },
    {
      "epoch": 2.6936104066104885,
      "grad_norm": 0.0223726574331522,
      "learning_rate": 3.1438247934523494e-07,
      "loss": 0.0009,
      "step": 164620
    },
    {
      "epoch": 2.6937740325615644,
      "grad_norm": 0.0420125387609005,
      "learning_rate": 3.1405033917423277e-07,
      "loss": 0.0005,
      "step": 164630
    },
    {
      "epoch": 2.69393765851264,
      "grad_norm": 0.10832918435335159,
      "learning_rate": 3.137183688576428e-07,
      "loss": 0.0005,
      "step": 164640
    },
    {
      "epoch": 2.694101284463716,
      "grad_norm": 0.08849113434553146,
      "learning_rate": 3.1338656840749995e-07,
      "loss": 0.001,
      "step": 164650
    },
    {
      "epoch": 2.694264910414792,
      "grad_norm": 0.14301711320877075,
      "learning_rate": 3.1305493783582885e-07,
      "loss": 0.0007,
      "step": 164660
    },
    {
      "epoch": 2.6944285363658675,
      "grad_norm": 0.008534056134521961,
      "learning_rate": 3.127234771546528e-07,
      "loss": 0.0004,
      "step": 164670
    },
    {
      "epoch": 2.6945921623169435,
      "grad_norm": 0.027357056736946106,
      "learning_rate": 3.123921863759838e-07,
      "loss": 0.0006,
      "step": 164680
    },
    {
      "epoch": 2.6947557882680195,
      "grad_norm": 0.04958884045481682,
      "learning_rate": 3.120610655118328e-07,
      "loss": 0.0004,
      "step": 164690
    },
    {
      "epoch": 2.694919414219095,
      "grad_norm": 0.1362619400024414,
      "learning_rate": 3.1173011457420024e-07,
      "loss": 0.0009,
      "step": 164700
    },
    {
      "epoch": 2.695083040170171,
      "grad_norm": 0.003076396184042096,
      "learning_rate": 3.1139933357508313e-07,
      "loss": 0.0005,
      "step": 164710
    },
    {
      "epoch": 2.695246666121247,
      "grad_norm": 0.012932802550494671,
      "learning_rate": 3.1106872252647244e-07,
      "loss": 0.0003,
      "step": 164720
    },
    {
      "epoch": 2.6954102920723226,
      "grad_norm": 0.03946397826075554,
      "learning_rate": 3.1073828144034977e-07,
      "loss": 0.0005,
      "step": 164730
    },
    {
      "epoch": 2.6955739180233986,
      "grad_norm": 0.11020883917808533,
      "learning_rate": 3.104080103286944e-07,
      "loss": 0.0006,
      "step": 164740
    },
    {
      "epoch": 2.6957375439744746,
      "grad_norm": 0.11737523972988129,
      "learning_rate": 3.1007790920347724e-07,
      "loss": 0.0006,
      "step": 164750
    },
    {
      "epoch": 2.69590116992555,
      "grad_norm": 0.008979442529380322,
      "learning_rate": 3.097479780766638e-07,
      "loss": 0.0008,
      "step": 164760
    },
    {
      "epoch": 2.696064795876626,
      "grad_norm": 0.1808817833662033,
      "learning_rate": 3.0941821696021446e-07,
      "loss": 0.0011,
      "step": 164770
    },
    {
      "epoch": 2.6962284218277017,
      "grad_norm": 0.1960621476173401,
      "learning_rate": 3.0908862586608024e-07,
      "loss": 0.0007,
      "step": 164780
    },
    {
      "epoch": 2.6963920477787777,
      "grad_norm": 0.0022592798341065645,
      "learning_rate": 3.087592048062099e-07,
      "loss": 0.0007,
      "step": 164790
    },
    {
      "epoch": 2.6965556737298537,
      "grad_norm": 0.03683222457766533,
      "learning_rate": 3.0842995379254226e-07,
      "loss": 0.0005,
      "step": 164800
    },
    {
      "epoch": 2.6967192996809293,
      "grad_norm": 0.015587952919304371,
      "learning_rate": 3.0810087283701493e-07,
      "loss": 0.0012,
      "step": 164810
    },
    {
      "epoch": 2.6968829256320053,
      "grad_norm": 0.015529097989201546,
      "learning_rate": 3.0777196195155234e-07,
      "loss": 0.0005,
      "step": 164820
    },
    {
      "epoch": 2.697046551583081,
      "grad_norm": 0.06242026016116142,
      "learning_rate": 3.0744322114807877e-07,
      "loss": 0.0011,
      "step": 164830
    },
    {
      "epoch": 2.697210177534157,
      "grad_norm": 0.20920215547084808,
      "learning_rate": 3.0711465043851084e-07,
      "loss": 0.0008,
      "step": 164840
    },
    {
      "epoch": 2.697373803485233,
      "grad_norm": 0.0022285813465714455,
      "learning_rate": 3.0678624983475733e-07,
      "loss": 0.0003,
      "step": 164850
    },
    {
      "epoch": 2.6975374294363084,
      "grad_norm": 0.1550416499376297,
      "learning_rate": 3.0645801934872264e-07,
      "loss": 0.0009,
      "step": 164860
    },
    {
      "epoch": 2.6977010553873844,
      "grad_norm": 0.039119429886341095,
      "learning_rate": 3.061299589923039e-07,
      "loss": 0.0006,
      "step": 164870
    },
    {
      "epoch": 2.6978646813384604,
      "grad_norm": 0.038074083626270294,
      "learning_rate": 3.0580206877739327e-07,
      "loss": 0.0004,
      "step": 164880
    },
    {
      "epoch": 2.698028307289536,
      "grad_norm": 0.18133632838726044,
      "learning_rate": 3.0547434871587455e-07,
      "loss": 0.0013,
      "step": 164890
    },
    {
      "epoch": 2.698191933240612,
      "grad_norm": 0.08124921470880508,
      "learning_rate": 3.051467988196283e-07,
      "loss": 0.0006,
      "step": 164900
    },
    {
      "epoch": 2.698355559191688,
      "grad_norm": 0.04892897978425026,
      "learning_rate": 3.0481941910052715e-07,
      "loss": 0.0005,
      "step": 164910
    },
    {
      "epoch": 2.6985191851427635,
      "grad_norm": 0.009849685244262218,
      "learning_rate": 3.044922095704367e-07,
      "loss": 0.0005,
      "step": 164920
    },
    {
      "epoch": 2.6986828110938395,
      "grad_norm": 0.15147916972637177,
      "learning_rate": 3.0416517024121905e-07,
      "loss": 0.0015,
      "step": 164930
    },
    {
      "epoch": 2.6988464370449154,
      "grad_norm": 0.025554845109581947,
      "learning_rate": 3.03838301124727e-07,
      "loss": 0.0008,
      "step": 164940
    },
    {
      "epoch": 2.699010062995991,
      "grad_norm": 0.10244035720825195,
      "learning_rate": 3.035116022328105e-07,
      "loss": 0.0008,
      "step": 164950
    },
    {
      "epoch": 2.699173688947067,
      "grad_norm": 0.08899954706430435,
      "learning_rate": 3.031850735773101e-07,
      "loss": 0.0006,
      "step": 164960
    },
    {
      "epoch": 2.699337314898143,
      "grad_norm": 0.08565444499254227,
      "learning_rate": 3.028587151700624e-07,
      "loss": 0.0009,
      "step": 164970
    },
    {
      "epoch": 2.6995009408492185,
      "grad_norm": 0.06779224425554276,
      "learning_rate": 3.0253252702289684e-07,
      "loss": 0.0003,
      "step": 164980
    },
    {
      "epoch": 2.6996645668002945,
      "grad_norm": 0.13577570021152496,
      "learning_rate": 3.022065091476367e-07,
      "loss": 0.0007,
      "step": 164990
    },
    {
      "epoch": 2.6998281927513705,
      "grad_norm": 0.05621223524212837,
      "learning_rate": 3.0188066155610097e-07,
      "loss": 0.0006,
      "step": 165000
    },
    {
      "epoch": 2.699991818702446,
      "grad_norm": 0.19461823999881744,
      "learning_rate": 3.01554984260099e-07,
      "loss": 0.0006,
      "step": 165010
    },
    {
      "epoch": 2.700155444653522,
      "grad_norm": 0.02049155905842781,
      "learning_rate": 3.012294772714369e-07,
      "loss": 0.0006,
      "step": 165020
    },
    {
      "epoch": 2.700319070604598,
      "grad_norm": 0.11019431054592133,
      "learning_rate": 3.0090414060191255e-07,
      "loss": 0.0009,
      "step": 165030
    },
    {
      "epoch": 2.7004826965556736,
      "grad_norm": 0.14438244700431824,
      "learning_rate": 3.005789742633186e-07,
      "loss": 0.0006,
      "step": 165040
    },
    {
      "epoch": 2.7006463225067496,
      "grad_norm": 0.01577475666999817,
      "learning_rate": 3.0025397826744296e-07,
      "loss": 0.0005,
      "step": 165050
    },
    {
      "epoch": 2.7008099484578256,
      "grad_norm": 0.057898905128240585,
      "learning_rate": 2.999291526260645e-07,
      "loss": 0.0011,
      "step": 165060
    },
    {
      "epoch": 2.700973574408901,
      "grad_norm": 0.050362300127744675,
      "learning_rate": 2.996044973509582e-07,
      "loss": 0.0009,
      "step": 165070
    },
    {
      "epoch": 2.701137200359977,
      "grad_norm": 0.08215784281492233,
      "learning_rate": 2.992800124538914e-07,
      "loss": 0.0011,
      "step": 165080
    },
    {
      "epoch": 2.701300826311053,
      "grad_norm": 0.06541654467582703,
      "learning_rate": 2.989556979466268e-07,
      "loss": 0.0005,
      "step": 165090
    },
    {
      "epoch": 2.7014644522621287,
      "grad_norm": 0.13151216506958008,
      "learning_rate": 2.98631553840919e-07,
      "loss": 0.0005,
      "step": 165100
    },
    {
      "epoch": 2.7016280782132047,
      "grad_norm": 0.03314230218529701,
      "learning_rate": 2.983075801485186e-07,
      "loss": 0.0006,
      "step": 165110
    },
    {
      "epoch": 2.7017917041642807,
      "grad_norm": 0.18590034544467926,
      "learning_rate": 2.9798377688116833e-07,
      "loss": 0.0014,
      "step": 165120
    },
    {
      "epoch": 2.7019553301153563,
      "grad_norm": 0.05290968343615532,
      "learning_rate": 2.9766014405060437e-07,
      "loss": 0.0008,
      "step": 165130
    },
    {
      "epoch": 2.7021189560664323,
      "grad_norm": 0.11153509467840195,
      "learning_rate": 2.973366816685591e-07,
      "loss": 0.0014,
      "step": 165140
    },
    {
      "epoch": 2.702282582017508,
      "grad_norm": 0.05995293706655502,
      "learning_rate": 2.970133897467559e-07,
      "loss": 0.0003,
      "step": 165150
    },
    {
      "epoch": 2.702446207968584,
      "grad_norm": 0.11263551563024521,
      "learning_rate": 2.966902682969142e-07,
      "loss": 0.0007,
      "step": 165160
    },
    {
      "epoch": 2.70260983391966,
      "grad_norm": 0.02554144337773323,
      "learning_rate": 2.9636731733074573e-07,
      "loss": 0.0005,
      "step": 165170
    },
    {
      "epoch": 2.7027734598707354,
      "grad_norm": 0.443741112947464,
      "learning_rate": 2.960445368599579e-07,
      "loss": 0.0012,
      "step": 165180
    },
    {
      "epoch": 2.7029370858218114,
      "grad_norm": 0.026707736775279045,
      "learning_rate": 2.957219268962497e-07,
      "loss": 0.0005,
      "step": 165190
    },
    {
      "epoch": 2.703100711772887,
      "grad_norm": 0.10003632307052612,
      "learning_rate": 2.953994874513144e-07,
      "loss": 0.0005,
      "step": 165200
    },
    {
      "epoch": 2.703264337723963,
      "grad_norm": 0.035505056381225586,
      "learning_rate": 2.950772185368428e-07,
      "loss": 0.0014,
      "step": 165210
    },
    {
      "epoch": 2.703427963675039,
      "grad_norm": 0.012073550373315811,
      "learning_rate": 2.947551201645116e-07,
      "loss": 0.0009,
      "step": 165220
    },
    {
      "epoch": 2.7035915896261145,
      "grad_norm": 0.06280562281608582,
      "learning_rate": 2.944331923459998e-07,
      "loss": 0.0009,
      "step": 165230
    },
    {
      "epoch": 2.7037552155771905,
      "grad_norm": 0.015164192765951157,
      "learning_rate": 2.9411143509297467e-07,
      "loss": 0.0007,
      "step": 165240
    },
    {
      "epoch": 2.7039188415282664,
      "grad_norm": 0.03793412446975708,
      "learning_rate": 2.937898484170998e-07,
      "loss": 0.0009,
      "step": 165250
    },
    {
      "epoch": 2.704082467479342,
      "grad_norm": 0.06300512701272964,
      "learning_rate": 2.9346843233003185e-07,
      "loss": 0.0006,
      "step": 165260
    },
    {
      "epoch": 2.704246093430418,
      "grad_norm": 0.09818796068429947,
      "learning_rate": 2.9314718684342105e-07,
      "loss": 0.0007,
      "step": 165270
    },
    {
      "epoch": 2.704409719381494,
      "grad_norm": 0.037996191531419754,
      "learning_rate": 2.92826111968913e-07,
      "loss": 0.0006,
      "step": 165280
    },
    {
      "epoch": 2.7045733453325695,
      "grad_norm": 0.1250123679637909,
      "learning_rate": 2.925052077181445e-07,
      "loss": 0.0007,
      "step": 165290
    },
    {
      "epoch": 2.7047369712836455,
      "grad_norm": 0.018413212150335312,
      "learning_rate": 2.92184474102748e-07,
      "loss": 0.0011,
      "step": 165300
    },
    {
      "epoch": 2.7049005972347215,
      "grad_norm": 0.03687401860952377,
      "learning_rate": 2.918639111343502e-07,
      "loss": 0.0004,
      "step": 165310
    },
    {
      "epoch": 2.705064223185797,
      "grad_norm": 0.0521615706384182,
      "learning_rate": 2.9154351882456855e-07,
      "loss": 0.0005,
      "step": 165320
    },
    {
      "epoch": 2.705227849136873,
      "grad_norm": 0.042304541915655136,
      "learning_rate": 2.9122329718501876e-07,
      "loss": 0.0006,
      "step": 165330
    },
    {
      "epoch": 2.705391475087949,
      "grad_norm": 0.1874370276927948,
      "learning_rate": 2.909032462273065e-07,
      "loss": 0.0005,
      "step": 165340
    },
    {
      "epoch": 2.7055551010390246,
      "grad_norm": 0.10817678272724152,
      "learning_rate": 2.905833659630347e-07,
      "loss": 0.0017,
      "step": 165350
    },
    {
      "epoch": 2.7057187269901006,
      "grad_norm": 0.015270596370100975,
      "learning_rate": 2.9026365640379583e-07,
      "loss": 0.0009,
      "step": 165360
    },
    {
      "epoch": 2.7058823529411766,
      "grad_norm": 0.016826342791318893,
      "learning_rate": 2.899441175611806e-07,
      "loss": 0.0004,
      "step": 165370
    },
    {
      "epoch": 2.706045978892252,
      "grad_norm": 0.01082519069314003,
      "learning_rate": 2.896247494467702e-07,
      "loss": 0.0011,
      "step": 165380
    },
    {
      "epoch": 2.706209604843328,
      "grad_norm": 0.03310023620724678,
      "learning_rate": 2.8930555207214217e-07,
      "loss": 0.0011,
      "step": 165390
    },
    {
      "epoch": 2.706373230794404,
      "grad_norm": 0.21069121360778809,
      "learning_rate": 2.889865254488655e-07,
      "loss": 0.0005,
      "step": 165400
    },
    {
      "epoch": 2.7065368567454797,
      "grad_norm": 0.06574288010597229,
      "learning_rate": 2.8866766958850543e-07,
      "loss": 0.0004,
      "step": 165410
    },
    {
      "epoch": 2.7067004826965557,
      "grad_norm": 0.07977749407291412,
      "learning_rate": 2.8834898450261885e-07,
      "loss": 0.0009,
      "step": 165420
    },
    {
      "epoch": 2.7068641086476317,
      "grad_norm": 0.11978202313184738,
      "learning_rate": 2.880304702027564e-07,
      "loss": 0.0005,
      "step": 165430
    },
    {
      "epoch": 2.7070277345987073,
      "grad_norm": 0.02209595777094364,
      "learning_rate": 2.8771212670046566e-07,
      "loss": 0.0008,
      "step": 165440
    },
    {
      "epoch": 2.7071913605497833,
      "grad_norm": 0.11386047303676605,
      "learning_rate": 2.87393954007284e-07,
      "loss": 0.0018,
      "step": 165450
    },
    {
      "epoch": 2.7073549865008593,
      "grad_norm": 0.05324776843190193,
      "learning_rate": 2.870759521347455e-07,
      "loss": 0.0015,
      "step": 165460
    },
    {
      "epoch": 2.707518612451935,
      "grad_norm": 0.0474587082862854,
      "learning_rate": 2.8675812109437537e-07,
      "loss": 0.0007,
      "step": 165470
    },
    {
      "epoch": 2.707682238403011,
      "grad_norm": 0.13699106872081757,
      "learning_rate": 2.8644046089769615e-07,
      "loss": 0.001,
      "step": 165480
    },
    {
      "epoch": 2.707845864354087,
      "grad_norm": 0.07346556335687637,
      "learning_rate": 2.861229715562219e-07,
      "loss": 0.0007,
      "step": 165490
    },
    {
      "epoch": 2.7080094903051624,
      "grad_norm": 0.07224927842617035,
      "learning_rate": 2.858056530814596e-07,
      "loss": 0.0007,
      "step": 165500
    },
    {
      "epoch": 2.7081731162562384,
      "grad_norm": 0.09595628827810287,
      "learning_rate": 2.854885054849138e-07,
      "loss": 0.0008,
      "step": 165510
    },
    {
      "epoch": 2.7083367422073144,
      "grad_norm": 0.038459885865449905,
      "learning_rate": 2.851715287780765e-07,
      "loss": 0.0007,
      "step": 165520
    },
    {
      "epoch": 2.70850036815839,
      "grad_norm": 0.01125528197735548,
      "learning_rate": 2.848547229724402e-07,
      "loss": 0.0008,
      "step": 165530
    },
    {
      "epoch": 2.708663994109466,
      "grad_norm": 0.008230326697230339,
      "learning_rate": 2.8453808807948846e-07,
      "loss": 0.0017,
      "step": 165540
    },
    {
      "epoch": 2.7088276200605415,
      "grad_norm": 0.0632973462343216,
      "learning_rate": 2.842216241106965e-07,
      "loss": 0.0004,
      "step": 165550
    },
    {
      "epoch": 2.7089912460116174,
      "grad_norm": 0.023056013509631157,
      "learning_rate": 2.8390533107753746e-07,
      "loss": 0.0005,
      "step": 165560
    },
    {
      "epoch": 2.7091548719626934,
      "grad_norm": 0.008053119294345379,
      "learning_rate": 2.835892089914749e-07,
      "loss": 0.0003,
      "step": 165570
    },
    {
      "epoch": 2.709318497913769,
      "grad_norm": 0.007786785252392292,
      "learning_rate": 2.8327325786396797e-07,
      "loss": 0.0007,
      "step": 165580
    },
    {
      "epoch": 2.709482123864845,
      "grad_norm": 0.18835268914699554,
      "learning_rate": 2.829574777064692e-07,
      "loss": 0.0008,
      "step": 165590
    },
    {
      "epoch": 2.7096457498159205,
      "grad_norm": 0.04131743311882019,
      "learning_rate": 2.8264186853042494e-07,
      "loss": 0.0003,
      "step": 165600
    },
    {
      "epoch": 2.7098093757669965,
      "grad_norm": 0.09654130786657333,
      "learning_rate": 2.82326430347275e-07,
      "loss": 0.001,
      "step": 165610
    },
    {
      "epoch": 2.7099730017180725,
      "grad_norm": 0.08916892856359482,
      "learning_rate": 2.82011163168453e-07,
      "loss": 0.0008,
      "step": 165620
    },
    {
      "epoch": 2.710136627669148,
      "grad_norm": 0.04151221364736557,
      "learning_rate": 2.816960670053875e-07,
      "loss": 0.0004,
      "step": 165630
    },
    {
      "epoch": 2.710300253620224,
      "grad_norm": 0.07366225868463516,
      "learning_rate": 2.813811418694984e-07,
      "loss": 0.0013,
      "step": 165640
    },
    {
      "epoch": 2.7104638795713,
      "grad_norm": 0.15956096351146698,
      "learning_rate": 2.810663877722031e-07,
      "loss": 0.0006,
      "step": 165650
    },
    {
      "epoch": 2.7106275055223756,
      "grad_norm": 0.015970144420862198,
      "learning_rate": 2.8075180472490917e-07,
      "loss": 0.0008,
      "step": 165660
    },
    {
      "epoch": 2.7107911314734516,
      "grad_norm": 0.06413602083921432,
      "learning_rate": 2.8043739273902026e-07,
      "loss": 0.0005,
      "step": 165670
    },
    {
      "epoch": 2.7109547574245276,
      "grad_norm": 0.06689391285181046,
      "learning_rate": 2.8012315182593176e-07,
      "loss": 0.0013,
      "step": 165680
    },
    {
      "epoch": 2.711118383375603,
      "grad_norm": 0.08121802657842636,
      "learning_rate": 2.7980908199703503e-07,
      "loss": 0.0006,
      "step": 165690
    },
    {
      "epoch": 2.711282009326679,
      "grad_norm": 0.03946439549326897,
      "learning_rate": 2.794951832637155e-07,
      "loss": 0.0008,
      "step": 165700
    },
    {
      "epoch": 2.711445635277755,
      "grad_norm": 0.03253056854009628,
      "learning_rate": 2.791814556373501e-07,
      "loss": 0.0006,
      "step": 165710
    },
    {
      "epoch": 2.7116092612288307,
      "grad_norm": 0.03238653764128685,
      "learning_rate": 2.7886789912931034e-07,
      "loss": 0.0007,
      "step": 165720
    },
    {
      "epoch": 2.7117728871799067,
      "grad_norm": 0.0024456228129565716,
      "learning_rate": 2.785545137509621e-07,
      "loss": 0.0005,
      "step": 165730
    },
    {
      "epoch": 2.7119365131309827,
      "grad_norm": 0.11946140229701996,
      "learning_rate": 2.7824129951366575e-07,
      "loss": 0.0007,
      "step": 165740
    },
    {
      "epoch": 2.7121001390820583,
      "grad_norm": 0.0031162812374532223,
      "learning_rate": 2.7792825642877276e-07,
      "loss": 0.0007,
      "step": 165750
    },
    {
      "epoch": 2.7122637650331343,
      "grad_norm": 0.0009196795872412622,
      "learning_rate": 2.7761538450763124e-07,
      "loss": 0.0016,
      "step": 165760
    },
    {
      "epoch": 2.7124273909842103,
      "grad_norm": 0.042442865669727325,
      "learning_rate": 2.7730268376158333e-07,
      "loss": 0.0008,
      "step": 165770
    },
    {
      "epoch": 2.712591016935286,
      "grad_norm": 0.09980204701423645,
      "learning_rate": 2.7699015420196095e-07,
      "loss": 0.0004,
      "step": 165780
    },
    {
      "epoch": 2.712754642886362,
      "grad_norm": 0.20933552086353302,
      "learning_rate": 2.766777958400951e-07,
      "loss": 0.0012,
      "step": 165790
    },
    {
      "epoch": 2.712918268837438,
      "grad_norm": 0.044055577367544174,
      "learning_rate": 2.7636560868730676e-07,
      "loss": 0.0011,
      "step": 165800
    },
    {
      "epoch": 2.7130818947885134,
      "grad_norm": 0.062321409583091736,
      "learning_rate": 2.760535927549124e-07,
      "loss": 0.001,
      "step": 165810
    },
    {
      "epoch": 2.7132455207395894,
      "grad_norm": 0.0921667143702507,
      "learning_rate": 2.7574174805422127e-07,
      "loss": 0.0003,
      "step": 165820
    },
    {
      "epoch": 2.7134091466906654,
      "grad_norm": 0.06988582015037537,
      "learning_rate": 2.7543007459653717e-07,
      "loss": 0.001,
      "step": 165830
    },
    {
      "epoch": 2.713572772641741,
      "grad_norm": 0.018518909811973572,
      "learning_rate": 2.751185723931582e-07,
      "loss": 0.0006,
      "step": 165840
    },
    {
      "epoch": 2.713736398592817,
      "grad_norm": 0.0658596009016037,
      "learning_rate": 2.7480724145537376e-07,
      "loss": 0.0004,
      "step": 165850
    },
    {
      "epoch": 2.713900024543893,
      "grad_norm": 0.020756209269165993,
      "learning_rate": 2.744960817944714e-07,
      "loss": 0.0025,
      "step": 165860
    },
    {
      "epoch": 2.7140636504949684,
      "grad_norm": 0.003743681823834777,
      "learning_rate": 2.7418509342172773e-07,
      "loss": 0.0007,
      "step": 165870
    },
    {
      "epoch": 2.7142272764460444,
      "grad_norm": 0.13827617466449738,
      "learning_rate": 2.73874276348417e-07,
      "loss": 0.0016,
      "step": 165880
    },
    {
      "epoch": 2.7143909023971204,
      "grad_norm": 0.0017665864434093237,
      "learning_rate": 2.7356363058580413e-07,
      "loss": 0.0005,
      "step": 165890
    },
    {
      "epoch": 2.714554528348196,
      "grad_norm": 0.04928507283329964,
      "learning_rate": 2.732531561451507e-07,
      "loss": 0.0008,
      "step": 165900
    },
    {
      "epoch": 2.714718154299272,
      "grad_norm": 0.007752994075417519,
      "learning_rate": 2.7294285303770983e-07,
      "loss": 0.0007,
      "step": 165910
    },
    {
      "epoch": 2.7148817802503475,
      "grad_norm": 0.06837140768766403,
      "learning_rate": 2.7263272127472873e-07,
      "loss": 0.0008,
      "step": 165920
    },
    {
      "epoch": 2.7150454062014235,
      "grad_norm": 0.007732836063951254,
      "learning_rate": 2.7232276086745057e-07,
      "loss": 0.0008,
      "step": 165930
    },
    {
      "epoch": 2.7152090321524995,
      "grad_norm": 0.06131186708807945,
      "learning_rate": 2.720129718271086e-07,
      "loss": 0.0005,
      "step": 165940
    },
    {
      "epoch": 2.715372658103575,
      "grad_norm": 0.024128209799528122,
      "learning_rate": 2.717033541649333e-07,
      "loss": 0.0004,
      "step": 165950
    },
    {
      "epoch": 2.715536284054651,
      "grad_norm": 0.059776730835437775,
      "learning_rate": 2.713939078921474e-07,
      "loss": 0.0006,
      "step": 165960
    },
    {
      "epoch": 2.7156999100057266,
      "grad_norm": 0.14890922605991364,
      "learning_rate": 2.710846330199668e-07,
      "loss": 0.0006,
      "step": 165970
    },
    {
      "epoch": 2.7158635359568026,
      "grad_norm": 0.2550153434276581,
      "learning_rate": 2.707755295596037e-07,
      "loss": 0.0007,
      "step": 165980
    },
    {
      "epoch": 2.7160271619078786,
      "grad_norm": 0.15260539948940277,
      "learning_rate": 2.7046659752226035e-07,
      "loss": 0.0006,
      "step": 165990
    },
    {
      "epoch": 2.716190787858954,
      "grad_norm": 0.04986754432320595,
      "learning_rate": 2.70157836919136e-07,
      "loss": 0.0006,
      "step": 166000
    },
    {
      "epoch": 2.71635441381003,
      "grad_norm": 0.060021527111530304,
      "learning_rate": 2.6984924776142287e-07,
      "loss": 0.0011,
      "step": 166010
    },
    {
      "epoch": 2.716518039761106,
      "grad_norm": 0.1671852469444275,
      "learning_rate": 2.695408300603053e-07,
      "loss": 0.0004,
      "step": 166020
    },
    {
      "epoch": 2.7166816657121817,
      "grad_norm": 0.024218380451202393,
      "learning_rate": 2.692325838269633e-07,
      "loss": 0.0003,
      "step": 166030
    },
    {
      "epoch": 2.7168452916632577,
      "grad_norm": 0.04165758565068245,
      "learning_rate": 2.689245090725695e-07,
      "loss": 0.0005,
      "step": 166040
    },
    {
      "epoch": 2.7170089176143337,
      "grad_norm": 0.021144501864910126,
      "learning_rate": 2.686166058082923e-07,
      "loss": 0.0008,
      "step": 166050
    },
    {
      "epoch": 2.7171725435654093,
      "grad_norm": 0.06020462140440941,
      "learning_rate": 2.6830887404529047e-07,
      "loss": 0.0007,
      "step": 166060
    },
    {
      "epoch": 2.7173361695164853,
      "grad_norm": 0.13731226325035095,
      "learning_rate": 2.6800131379472074e-07,
      "loss": 0.0008,
      "step": 166070
    },
    {
      "epoch": 2.7174997954675613,
      "grad_norm": 0.12679561972618103,
      "learning_rate": 2.6769392506772964e-07,
      "loss": 0.0004,
      "step": 166080
    },
    {
      "epoch": 2.717663421418637,
      "grad_norm": 0.18874065577983856,
      "learning_rate": 2.673867078754605e-07,
      "loss": 0.0011,
      "step": 166090
    },
    {
      "epoch": 2.717827047369713,
      "grad_norm": 0.03484984487295151,
      "learning_rate": 2.670796622290478e-07,
      "loss": 0.0004,
      "step": 166100
    },
    {
      "epoch": 2.717990673320789,
      "grad_norm": 0.025047173723578453,
      "learning_rate": 2.6677278813962317e-07,
      "loss": 0.0007,
      "step": 166110
    },
    {
      "epoch": 2.7181542992718644,
      "grad_norm": 0.02176237292587757,
      "learning_rate": 2.6646608561830876e-07,
      "loss": 0.0006,
      "step": 166120
    },
    {
      "epoch": 2.7183179252229404,
      "grad_norm": 0.06795595586299896,
      "learning_rate": 2.6615955467622136e-07,
      "loss": 0.0008,
      "step": 166130
    },
    {
      "epoch": 2.7184815511740164,
      "grad_norm": 0.04296782240271568,
      "learning_rate": 2.658531953244731e-07,
      "loss": 0.0009,
      "step": 166140
    },
    {
      "epoch": 2.718645177125092,
      "grad_norm": 0.021120456978678703,
      "learning_rate": 2.6554700757416784e-07,
      "loss": 0.0006,
      "step": 166150
    },
    {
      "epoch": 2.718808803076168,
      "grad_norm": 0.172512024641037,
      "learning_rate": 2.6524099143640515e-07,
      "loss": 0.0011,
      "step": 166160
    },
    {
      "epoch": 2.718972429027244,
      "grad_norm": 0.11277034133672714,
      "learning_rate": 2.6493514692227664e-07,
      "loss": 0.0006,
      "step": 166170
    },
    {
      "epoch": 2.7191360549783195,
      "grad_norm": 0.14692123234272003,
      "learning_rate": 2.6462947404286845e-07,
      "loss": 0.001,
      "step": 166180
    },
    {
      "epoch": 2.7192996809293954,
      "grad_norm": 0.057102132588624954,
      "learning_rate": 2.643239728092606e-07,
      "loss": 0.0004,
      "step": 166190
    },
    {
      "epoch": 2.7194633068804714,
      "grad_norm": 0.05140982195734978,
      "learning_rate": 2.640186432325265e-07,
      "loss": 0.0006,
      "step": 166200
    },
    {
      "epoch": 2.719626932831547,
      "grad_norm": 0.038949571549892426,
      "learning_rate": 2.63713485323735e-07,
      "loss": 0.0005,
      "step": 166210
    },
    {
      "epoch": 2.719790558782623,
      "grad_norm": 0.032647065818309784,
      "learning_rate": 2.634084990939462e-07,
      "loss": 0.0004,
      "step": 166220
    },
    {
      "epoch": 2.719954184733699,
      "grad_norm": 0.15789948403835297,
      "learning_rate": 2.631036845542151e-07,
      "loss": 0.0014,
      "step": 166230
    },
    {
      "epoch": 2.7201178106847745,
      "grad_norm": 0.062313031405210495,
      "learning_rate": 2.6279904171558954e-07,
      "loss": 0.0008,
      "step": 166240
    },
    {
      "epoch": 2.7202814366358505,
      "grad_norm": 0.15636670589447021,
      "learning_rate": 2.6249457058911356e-07,
      "loss": 0.001,
      "step": 166250
    },
    {
      "epoch": 2.7204450625869265,
      "grad_norm": 0.0162058062851429,
      "learning_rate": 2.6219027118582375e-07,
      "loss": 0.0005,
      "step": 166260
    },
    {
      "epoch": 2.720608688538002,
      "grad_norm": 0.005950163584202528,
      "learning_rate": 2.6188614351674913e-07,
      "loss": 0.0009,
      "step": 166270
    },
    {
      "epoch": 2.720772314489078,
      "grad_norm": 0.05566615238785744,
      "learning_rate": 2.615821875929142e-07,
      "loss": 0.0006,
      "step": 166280
    },
    {
      "epoch": 2.720935940440154,
      "grad_norm": 0.0025238730013370514,
      "learning_rate": 2.6127840342533627e-07,
      "loss": 0.0003,
      "step": 166290
    },
    {
      "epoch": 2.7210995663912296,
      "grad_norm": 0.03258737549185753,
      "learning_rate": 2.6097479102502707e-07,
      "loss": 0.0024,
      "step": 166300
    },
    {
      "epoch": 2.7212631923423056,
      "grad_norm": 0.0484694167971611,
      "learning_rate": 2.6067135040299174e-07,
      "loss": 0.0005,
      "step": 166310
    },
    {
      "epoch": 2.721426818293381,
      "grad_norm": 0.03294214978814125,
      "learning_rate": 2.6036808157022975e-07,
      "loss": 0.0011,
      "step": 166320
    },
    {
      "epoch": 2.721590444244457,
      "grad_norm": 0.04648750275373459,
      "learning_rate": 2.6006498453773345e-07,
      "loss": 0.0009,
      "step": 166330
    },
    {
      "epoch": 2.721754070195533,
      "grad_norm": 0.005018678493797779,
      "learning_rate": 2.5976205931648846e-07,
      "loss": 0.0007,
      "step": 166340
    },
    {
      "epoch": 2.7219176961466087,
      "grad_norm": 0.5051056742668152,
      "learning_rate": 2.594593059174766e-07,
      "loss": 0.0013,
      "step": 166350
    },
    {
      "epoch": 2.7220813220976847,
      "grad_norm": 0.08327613770961761,
      "learning_rate": 2.5915672435167074e-07,
      "loss": 0.0006,
      "step": 166360
    },
    {
      "epoch": 2.7222449480487603,
      "grad_norm": 0.00232758023776114,
      "learning_rate": 2.588543146300404e-07,
      "loss": 0.0005,
      "step": 166370
    },
    {
      "epoch": 2.7224085739998363,
      "grad_norm": 0.06786621361970901,
      "learning_rate": 2.585520767635447e-07,
      "loss": 0.0009,
      "step": 166380
    },
    {
      "epoch": 2.7225721999509123,
      "grad_norm": 0.0009498855215497315,
      "learning_rate": 2.582500107631414e-07,
      "loss": 0.001,
      "step": 166390
    },
    {
      "epoch": 2.722735825901988,
      "grad_norm": 0.07119929790496826,
      "learning_rate": 2.579481166397785e-07,
      "loss": 0.0009,
      "step": 166400
    },
    {
      "epoch": 2.722899451853064,
      "grad_norm": 0.028171630576252937,
      "learning_rate": 2.5764639440439833e-07,
      "loss": 0.0012,
      "step": 166410
    },
    {
      "epoch": 2.72306307780414,
      "grad_norm": 0.15306329727172852,
      "learning_rate": 2.5734484406794046e-07,
      "loss": 0.0009,
      "step": 166420
    },
    {
      "epoch": 2.7232267037552154,
      "grad_norm": 0.09083644300699234,
      "learning_rate": 2.5704346564133174e-07,
      "loss": 0.0008,
      "step": 166430
    },
    {
      "epoch": 2.7233903297062914,
      "grad_norm": 0.01764063350856304,
      "learning_rate": 2.5674225913549843e-07,
      "loss": 0.0004,
      "step": 166440
    },
    {
      "epoch": 2.7235539556573674,
      "grad_norm": 0.006110843271017075,
      "learning_rate": 2.564412245613579e-07,
      "loss": 0.0013,
      "step": 166450
    },
    {
      "epoch": 2.723717581608443,
      "grad_norm": 0.003444591537117958,
      "learning_rate": 2.5614036192982195e-07,
      "loss": 0.0005,
      "step": 166460
    },
    {
      "epoch": 2.723881207559519,
      "grad_norm": 0.05287844315171242,
      "learning_rate": 2.5583967125179687e-07,
      "loss": 0.001,
      "step": 166470
    },
    {
      "epoch": 2.724044833510595,
      "grad_norm": 0.019583825021982193,
      "learning_rate": 2.555391525381806e-07,
      "loss": 0.0005,
      "step": 166480
    },
    {
      "epoch": 2.7242084594616705,
      "grad_norm": 0.005608867853879929,
      "learning_rate": 2.5523880579986783e-07,
      "loss": 0.0009,
      "step": 166490
    },
    {
      "epoch": 2.7243720854127464,
      "grad_norm": 0.02322128787636757,
      "learning_rate": 2.5493863104774417e-07,
      "loss": 0.0006,
      "step": 166500
    },
    {
      "epoch": 2.7245357113638224,
      "grad_norm": 0.04924976825714111,
      "learning_rate": 2.54638628292691e-07,
      "loss": 0.0005,
      "step": 166510
    },
    {
      "epoch": 2.724699337314898,
      "grad_norm": 0.0036141378805041313,
      "learning_rate": 2.5433879754558246e-07,
      "loss": 0.0011,
      "step": 166520
    },
    {
      "epoch": 2.724862963265974,
      "grad_norm": 0.004840285051614046,
      "learning_rate": 2.540391388172864e-07,
      "loss": 0.0011,
      "step": 166530
    },
    {
      "epoch": 2.72502658921705,
      "grad_norm": 0.1532498002052307,
      "learning_rate": 2.5373965211866524e-07,
      "loss": 0.0011,
      "step": 166540
    },
    {
      "epoch": 2.7251902151681255,
      "grad_norm": 0.06085013225674629,
      "learning_rate": 2.534403374605737e-07,
      "loss": 0.0015,
      "step": 166550
    },
    {
      "epoch": 2.7253538411192015,
      "grad_norm": 0.0757409930229187,
      "learning_rate": 2.5314119485386257e-07,
      "loss": 0.001,
      "step": 166560
    },
    {
      "epoch": 2.7255174670702775,
      "grad_norm": 0.0154356574639678,
      "learning_rate": 2.5284222430937365e-07,
      "loss": 0.0003,
      "step": 166570
    },
    {
      "epoch": 2.725681093021353,
      "grad_norm": 0.004803037736564875,
      "learning_rate": 2.52543425837945e-07,
      "loss": 0.0005,
      "step": 166580
    },
    {
      "epoch": 2.725844718972429,
      "grad_norm": 0.13299861550331116,
      "learning_rate": 2.522447994504068e-07,
      "loss": 0.001,
      "step": 166590
    },
    {
      "epoch": 2.726008344923505,
      "grad_norm": 0.10318078845739365,
      "learning_rate": 2.5194634515758376e-07,
      "loss": 0.0006,
      "step": 166600
    },
    {
      "epoch": 2.7261719708745806,
      "grad_norm": 0.055688053369522095,
      "learning_rate": 2.516480629702939e-07,
      "loss": 0.0013,
      "step": 166610
    },
    {
      "epoch": 2.7263355968256566,
      "grad_norm": 0.02145429700613022,
      "learning_rate": 2.5134995289935017e-07,
      "loss": 0.0003,
      "step": 166620
    },
    {
      "epoch": 2.7264992227767326,
      "grad_norm": 0.007415490690618753,
      "learning_rate": 2.510520149555573e-07,
      "loss": 0.0006,
      "step": 166630
    },
    {
      "epoch": 2.726662848727808,
      "grad_norm": 0.0563824288547039,
      "learning_rate": 2.5075424914971445e-07,
      "loss": 0.0004,
      "step": 166640
    },
    {
      "epoch": 2.726826474678884,
      "grad_norm": 0.11017031967639923,
      "learning_rate": 2.5045665549261633e-07,
      "loss": 0.0005,
      "step": 166650
    },
    {
      "epoch": 2.72699010062996,
      "grad_norm": 0.09914813190698624,
      "learning_rate": 2.501592339950482e-07,
      "loss": 0.0004,
      "step": 166660
    },
    {
      "epoch": 2.7271537265810357,
      "grad_norm": 0.04692218452692032,
      "learning_rate": 2.49861984667793e-07,
      "loss": 0.001,
      "step": 166670
    },
    {
      "epoch": 2.7273173525321117,
      "grad_norm": 0.028819413855671883,
      "learning_rate": 2.495649075216233e-07,
      "loss": 0.0006,
      "step": 166680
    },
    {
      "epoch": 2.7274809784831873,
      "grad_norm": 0.212903693318367,
      "learning_rate": 2.492680025673083e-07,
      "loss": 0.0018,
      "step": 166690
    },
    {
      "epoch": 2.7276446044342633,
      "grad_norm": 0.06485790014266968,
      "learning_rate": 2.489712698156105e-07,
      "loss": 0.0009,
      "step": 166700
    },
    {
      "epoch": 2.7278082303853393,
      "grad_norm": 0.09056265652179718,
      "learning_rate": 2.486747092772851e-07,
      "loss": 0.0011,
      "step": 166710
    },
    {
      "epoch": 2.727971856336415,
      "grad_norm": 0.01741917058825493,
      "learning_rate": 2.4837832096308355e-07,
      "loss": 0.0003,
      "step": 166720
    },
    {
      "epoch": 2.728135482287491,
      "grad_norm": 0.03494131192564964,
      "learning_rate": 2.4808210488374615e-07,
      "loss": 0.0008,
      "step": 166730
    },
    {
      "epoch": 2.7282991082385664,
      "grad_norm": 0.11381885409355164,
      "learning_rate": 2.47786061050011e-07,
      "loss": 0.0007,
      "step": 166740
    },
    {
      "epoch": 2.7284627341896424,
      "grad_norm": 0.09058448672294617,
      "learning_rate": 2.474901894726106e-07,
      "loss": 0.0006,
      "step": 166750
    },
    {
      "epoch": 2.7286263601407184,
      "grad_norm": 0.17616605758666992,
      "learning_rate": 2.47194490162268e-07,
      "loss": 0.0027,
      "step": 166760
    },
    {
      "epoch": 2.728789986091794,
      "grad_norm": 0.09523951262235641,
      "learning_rate": 2.468989631297025e-07,
      "loss": 0.001,
      "step": 166770
    },
    {
      "epoch": 2.72895361204287,
      "grad_norm": 0.05731816962361336,
      "learning_rate": 2.4660360838562494e-07,
      "loss": 0.0003,
      "step": 166780
    },
    {
      "epoch": 2.729117237993946,
      "grad_norm": 0.09204509109258652,
      "learning_rate": 2.463084259407428e-07,
      "loss": 0.0006,
      "step": 166790
    },
    {
      "epoch": 2.7292808639450215,
      "grad_norm": 0.006113148294389248,
      "learning_rate": 2.4601341580575477e-07,
      "loss": 0.0008,
      "step": 166800
    },
    {
      "epoch": 2.7294444898960974,
      "grad_norm": 0.2661568522453308,
      "learning_rate": 2.457185779913551e-07,
      "loss": 0.0008,
      "step": 166810
    },
    {
      "epoch": 2.7296081158471734,
      "grad_norm": 0.026791594922542572,
      "learning_rate": 2.454239125082297e-07,
      "loss": 0.0006,
      "step": 166820
    },
    {
      "epoch": 2.729771741798249,
      "grad_norm": 0.06673616915941238,
      "learning_rate": 2.4512941936706e-07,
      "loss": 0.0004,
      "step": 166830
    },
    {
      "epoch": 2.729935367749325,
      "grad_norm": 0.03314259648323059,
      "learning_rate": 2.4483509857852193e-07,
      "loss": 0.0008,
      "step": 166840
    },
    {
      "epoch": 2.730098993700401,
      "grad_norm": 0.004142835736274719,
      "learning_rate": 2.445409501532814e-07,
      "loss": 0.0005,
      "step": 166850
    },
    {
      "epoch": 2.7302626196514765,
      "grad_norm": 0.06205581873655319,
      "learning_rate": 2.4424697410200316e-07,
      "loss": 0.0008,
      "step": 166860
    },
    {
      "epoch": 2.7304262456025525,
      "grad_norm": 0.1340777575969696,
      "learning_rate": 2.4395317043534097e-07,
      "loss": 0.0009,
      "step": 166870
    },
    {
      "epoch": 2.7305898715536285,
      "grad_norm": 0.02982744574546814,
      "learning_rate": 2.4365953916394627e-07,
      "loss": 0.0004,
      "step": 166880
    },
    {
      "epoch": 2.730753497504704,
      "grad_norm": 0.10288508236408234,
      "learning_rate": 2.4336608029846054e-07,
      "loss": 0.0013,
      "step": 166890
    },
    {
      "epoch": 2.73091712345578,
      "grad_norm": 0.10716257244348526,
      "learning_rate": 2.430727938495225e-07,
      "loss": 0.0007,
      "step": 166900
    },
    {
      "epoch": 2.731080749406856,
      "grad_norm": 0.08042197674512863,
      "learning_rate": 2.4277967982776375e-07,
      "loss": 0.0012,
      "step": 166910
    },
    {
      "epoch": 2.7312443753579316,
      "grad_norm": 0.0348656103014946,
      "learning_rate": 2.4248673824380787e-07,
      "loss": 0.0006,
      "step": 166920
    },
    {
      "epoch": 2.7314080013090076,
      "grad_norm": 0.06871537119150162,
      "learning_rate": 2.4219396910827307e-07,
      "loss": 0.0007,
      "step": 166930
    },
    {
      "epoch": 2.7315716272600836,
      "grad_norm": 0.05744808539748192,
      "learning_rate": 2.4190137243177093e-07,
      "loss": 0.0013,
      "step": 166940
    },
    {
      "epoch": 2.731735253211159,
      "grad_norm": 0.08951016515493393,
      "learning_rate": 2.41608948224909e-07,
      "loss": 0.0008,
      "step": 166950
    },
    {
      "epoch": 2.731898879162235,
      "grad_norm": 0.1780056655406952,
      "learning_rate": 2.4131669649828606e-07,
      "loss": 0.0006,
      "step": 166960
    },
    {
      "epoch": 2.732062505113311,
      "grad_norm": 0.05167065188288689,
      "learning_rate": 2.410246172624953e-07,
      "loss": 0.0006,
      "step": 166970
    },
    {
      "epoch": 2.7322261310643867,
      "grad_norm": 0.0031203618273139,
      "learning_rate": 2.407327105281249e-07,
      "loss": 0.001,
      "step": 166980
    },
    {
      "epoch": 2.7323897570154627,
      "grad_norm": 0.009707254357635975,
      "learning_rate": 2.404409763057541e-07,
      "loss": 0.0005,
      "step": 166990
    },
    {
      "epoch": 2.7325533829665387,
      "grad_norm": 0.03431081771850586,
      "learning_rate": 2.401494146059602e-07,
      "loss": 0.0007,
      "step": 167000
    },
    {
      "epoch": 2.7327170089176143,
      "grad_norm": 0.04879627749323845,
      "learning_rate": 2.3985802543930846e-07,
      "loss": 0.0005,
      "step": 167010
    },
    {
      "epoch": 2.7328806348686903,
      "grad_norm": 0.04363630712032318,
      "learning_rate": 2.395668088163638e-07,
      "loss": 0.0008,
      "step": 167020
    },
    {
      "epoch": 2.7330442608197663,
      "grad_norm": 0.028646238148212433,
      "learning_rate": 2.3927576474768064e-07,
      "loss": 0.0006,
      "step": 167030
    },
    {
      "epoch": 2.733207886770842,
      "grad_norm": 0.028668992221355438,
      "learning_rate": 2.389848932438088e-07,
      "loss": 0.0006,
      "step": 167040
    },
    {
      "epoch": 2.733371512721918,
      "grad_norm": 0.07995645701885223,
      "learning_rate": 2.386941943152921e-07,
      "loss": 0.0005,
      "step": 167050
    },
    {
      "epoch": 2.733535138672994,
      "grad_norm": 0.037354037165641785,
      "learning_rate": 2.3840366797266644e-07,
      "loss": 0.0004,
      "step": 167060
    },
    {
      "epoch": 2.7336987646240694,
      "grad_norm": 0.05522950366139412,
      "learning_rate": 2.3811331422646467e-07,
      "loss": 0.0009,
      "step": 167070
    },
    {
      "epoch": 2.7338623905751454,
      "grad_norm": 0.005512866657227278,
      "learning_rate": 2.3782313308720938e-07,
      "loss": 0.0009,
      "step": 167080
    },
    {
      "epoch": 2.734026016526221,
      "grad_norm": 0.028287706896662712,
      "learning_rate": 2.3753312456542055e-07,
      "loss": 0.0009,
      "step": 167090
    },
    {
      "epoch": 2.734189642477297,
      "grad_norm": 0.034400373697280884,
      "learning_rate": 2.3724328867160918e-07,
      "loss": 0.0015,
      "step": 167100
    },
    {
      "epoch": 2.734353268428373,
      "grad_norm": 0.007700387388467789,
      "learning_rate": 2.3695362541628242e-07,
      "loss": 0.0005,
      "step": 167110
    },
    {
      "epoch": 2.7345168943794484,
      "grad_norm": 0.05857130140066147,
      "learning_rate": 2.3666413480993856e-07,
      "loss": 0.0003,
      "step": 167120
    },
    {
      "epoch": 2.7346805203305244,
      "grad_norm": 0.029179135337471962,
      "learning_rate": 2.363748168630714e-07,
      "loss": 0.0033,
      "step": 167130
    },
    {
      "epoch": 2.7348441462816,
      "grad_norm": 0.10953580588102341,
      "learning_rate": 2.3608567158616812e-07,
      "loss": 0.0007,
      "step": 167140
    },
    {
      "epoch": 2.735007772232676,
      "grad_norm": 0.0232537928968668,
      "learning_rate": 2.3579669898970926e-07,
      "loss": 0.0004,
      "step": 167150
    },
    {
      "epoch": 2.735171398183752,
      "grad_norm": 0.02705479972064495,
      "learning_rate": 2.3550789908417027e-07,
      "loss": 0.0008,
      "step": 167160
    },
    {
      "epoch": 2.7353350241348275,
      "grad_norm": 0.047809701412916183,
      "learning_rate": 2.352192718800178e-07,
      "loss": 0.0007,
      "step": 167170
    },
    {
      "epoch": 2.7354986500859035,
      "grad_norm": 0.06296701729297638,
      "learning_rate": 2.3493081738771517e-07,
      "loss": 0.0006,
      "step": 167180
    },
    {
      "epoch": 2.7356622760369795,
      "grad_norm": 0.013241423293948174,
      "learning_rate": 2.3464253561771844e-07,
      "loss": 0.0004,
      "step": 167190
    },
    {
      "epoch": 2.735825901988055,
      "grad_norm": 0.025256283581256866,
      "learning_rate": 2.343544265804759e-07,
      "loss": 0.0008,
      "step": 167200
    },
    {
      "epoch": 2.735989527939131,
      "grad_norm": 0.006607535295188427,
      "learning_rate": 2.3406649028643203e-07,
      "loss": 0.0004,
      "step": 167210
    },
    {
      "epoch": 2.736153153890207,
      "grad_norm": 0.0622825063765049,
      "learning_rate": 2.3377872674602343e-07,
      "loss": 0.0007,
      "step": 167220
    },
    {
      "epoch": 2.7363167798412826,
      "grad_norm": 0.05698620155453682,
      "learning_rate": 2.3349113596968064e-07,
      "loss": 0.0003,
      "step": 167230
    },
    {
      "epoch": 2.7364804057923586,
      "grad_norm": 0.0025128894485533237,
      "learning_rate": 2.3320371796782758e-07,
      "loss": 0.001,
      "step": 167240
    },
    {
      "epoch": 2.7366440317434346,
      "grad_norm": 0.1656695008277893,
      "learning_rate": 2.329164727508826e-07,
      "loss": 0.0008,
      "step": 167250
    },
    {
      "epoch": 2.73680765769451,
      "grad_norm": 0.08089619874954224,
      "learning_rate": 2.326294003292595e-07,
      "loss": 0.001,
      "step": 167260
    },
    {
      "epoch": 2.736971283645586,
      "grad_norm": 0.06426557153463364,
      "learning_rate": 2.3234250071336119e-07,
      "loss": 0.0012,
      "step": 167270
    },
    {
      "epoch": 2.737134909596662,
      "grad_norm": 0.06766103953123093,
      "learning_rate": 2.3205577391358924e-07,
      "loss": 0.0008,
      "step": 167280
    },
    {
      "epoch": 2.7372985355477377,
      "grad_norm": 0.10296272486448288,
      "learning_rate": 2.3176921994033597e-07,
      "loss": 0.0009,
      "step": 167290
    },
    {
      "epoch": 2.7374621614988137,
      "grad_norm": 0.04676128551363945,
      "learning_rate": 2.314828388039886e-07,
      "loss": 0.0012,
      "step": 167300
    },
    {
      "epoch": 2.7376257874498897,
      "grad_norm": 0.008685381151735783,
      "learning_rate": 2.3119663051492659e-07,
      "loss": 0.0005,
      "step": 167310
    },
    {
      "epoch": 2.7377894134009653,
      "grad_norm": 0.13609428703784943,
      "learning_rate": 2.3091059508352609e-07,
      "loss": 0.0009,
      "step": 167320
    },
    {
      "epoch": 2.7379530393520413,
      "grad_norm": 0.04204880818724632,
      "learning_rate": 2.3062473252015438e-07,
      "loss": 0.0005,
      "step": 167330
    },
    {
      "epoch": 2.7381166653031173,
      "grad_norm": 0.31170615553855896,
      "learning_rate": 2.303390428351726e-07,
      "loss": 0.0012,
      "step": 167340
    },
    {
      "epoch": 2.738280291254193,
      "grad_norm": 0.06603459268808365,
      "learning_rate": 2.300535260389375e-07,
      "loss": 0.0003,
      "step": 167350
    },
    {
      "epoch": 2.738443917205269,
      "grad_norm": 0.11828278750181198,
      "learning_rate": 2.2976818214179686e-07,
      "loss": 0.0006,
      "step": 167360
    },
    {
      "epoch": 2.738607543156345,
      "grad_norm": 0.21026110649108887,
      "learning_rate": 2.2948301115409576e-07,
      "loss": 0.0007,
      "step": 167370
    },
    {
      "epoch": 2.7387711691074204,
      "grad_norm": 0.23438793420791626,
      "learning_rate": 2.291980130861693e-07,
      "loss": 0.0011,
      "step": 167380
    },
    {
      "epoch": 2.7389347950584964,
      "grad_norm": 0.044866740703582764,
      "learning_rate": 2.289131879483486e-07,
      "loss": 0.0006,
      "step": 167390
    },
    {
      "epoch": 2.7390984210095723,
      "grad_norm": 0.08250736445188522,
      "learning_rate": 2.2862853575095766e-07,
      "loss": 0.0003,
      "step": 167400
    },
    {
      "epoch": 2.739262046960648,
      "grad_norm": 0.05013498291373253,
      "learning_rate": 2.2834405650431435e-07,
      "loss": 0.0007,
      "step": 167410
    },
    {
      "epoch": 2.739425672911724,
      "grad_norm": 0.030337626114487648,
      "learning_rate": 2.2805975021873206e-07,
      "loss": 0.0005,
      "step": 167420
    },
    {
      "epoch": 2.7395892988628,
      "grad_norm": 0.03338876739144325,
      "learning_rate": 2.2777561690451366e-07,
      "loss": 0.0009,
      "step": 167430
    },
    {
      "epoch": 2.7397529248138754,
      "grad_norm": 0.008874011225998402,
      "learning_rate": 2.2749165657195982e-07,
      "loss": 0.0007,
      "step": 167440
    },
    {
      "epoch": 2.7399165507649514,
      "grad_norm": 0.039966460317373276,
      "learning_rate": 2.272078692313623e-07,
      "loss": 0.0005,
      "step": 167450
    },
    {
      "epoch": 2.7400801767160274,
      "grad_norm": 0.10308028012514114,
      "learning_rate": 2.2692425489300897e-07,
      "loss": 0.0008,
      "step": 167460
    },
    {
      "epoch": 2.740243802667103,
      "grad_norm": 0.09634079784154892,
      "learning_rate": 2.2664081356717938e-07,
      "loss": 0.0003,
      "step": 167470
    },
    {
      "epoch": 2.740407428618179,
      "grad_norm": 0.04774674400687218,
      "learning_rate": 2.263575452641481e-07,
      "loss": 0.0015,
      "step": 167480
    },
    {
      "epoch": 2.7405710545692545,
      "grad_norm": 0.07143358141183853,
      "learning_rate": 2.2607444999418248e-07,
      "loss": 0.0011,
      "step": 167490
    },
    {
      "epoch": 2.7407346805203305,
      "grad_norm": 0.013922899961471558,
      "learning_rate": 2.2579152776754377e-07,
      "loss": 0.0007,
      "step": 167500
    },
    {
      "epoch": 2.7408983064714065,
      "grad_norm": 0.07386036217212677,
      "learning_rate": 2.2550877859448872e-07,
      "loss": 0.0005,
      "step": 167510
    },
    {
      "epoch": 2.741061932422482,
      "grad_norm": 0.05879054218530655,
      "learning_rate": 2.2522620248526528e-07,
      "loss": 0.0006,
      "step": 167520
    },
    {
      "epoch": 2.741225558373558,
      "grad_norm": 0.0702357366681099,
      "learning_rate": 2.2494379945011523e-07,
      "loss": 0.0008,
      "step": 167530
    },
    {
      "epoch": 2.7413891843246336,
      "grad_norm": 0.03725006431341171,
      "learning_rate": 2.2466156949927653e-07,
      "loss": 0.0007,
      "step": 167540
    },
    {
      "epoch": 2.7415528102757096,
      "grad_norm": 0.0011913494672626257,
      "learning_rate": 2.2437951264297763e-07,
      "loss": 0.0008,
      "step": 167550
    },
    {
      "epoch": 2.7417164362267856,
      "grad_norm": 0.06734905391931534,
      "learning_rate": 2.2409762889144482e-07,
      "loss": 0.0006,
      "step": 167560
    },
    {
      "epoch": 2.741880062177861,
      "grad_norm": 0.10126426815986633,
      "learning_rate": 2.2381591825489325e-07,
      "loss": 0.0007,
      "step": 167570
    },
    {
      "epoch": 2.742043688128937,
      "grad_norm": 0.11646626889705658,
      "learning_rate": 2.235343807435364e-07,
      "loss": 0.0009,
      "step": 167580
    },
    {
      "epoch": 2.742207314080013,
      "grad_norm": 0.030360354110598564,
      "learning_rate": 2.2325301636757778e-07,
      "loss": 0.0009,
      "step": 167590
    },
    {
      "epoch": 2.7423709400310887,
      "grad_norm": 0.02005661092698574,
      "learning_rate": 2.2297182513721705e-07,
      "loss": 0.0008,
      "step": 167600
    },
    {
      "epoch": 2.7425345659821647,
      "grad_norm": 0.05383145436644554,
      "learning_rate": 2.22690807062646e-07,
      "loss": 0.0007,
      "step": 167610
    },
    {
      "epoch": 2.7426981919332407,
      "grad_norm": 0.14692895114421844,
      "learning_rate": 2.2240996215405154e-07,
      "loss": 0.0009,
      "step": 167620
    },
    {
      "epoch": 2.7428618178843163,
      "grad_norm": 0.05291958898305893,
      "learning_rate": 2.2212929042161325e-07,
      "loss": 0.0009,
      "step": 167630
    },
    {
      "epoch": 2.7430254438353923,
      "grad_norm": 0.07483596354722977,
      "learning_rate": 2.218487918755047e-07,
      "loss": 0.0006,
      "step": 167640
    },
    {
      "epoch": 2.7431890697864683,
      "grad_norm": 0.10124974697828293,
      "learning_rate": 2.215684665258938e-07,
      "loss": 0.0011,
      "step": 167650
    },
    {
      "epoch": 2.743352695737544,
      "grad_norm": 0.06880058348178864,
      "learning_rate": 2.2128831438294086e-07,
      "loss": 0.0008,
      "step": 167660
    },
    {
      "epoch": 2.74351632168862,
      "grad_norm": 0.0684800073504448,
      "learning_rate": 2.21008335456801e-07,
      "loss": 0.0004,
      "step": 167670
    },
    {
      "epoch": 2.743679947639696,
      "grad_norm": 0.01002908032387495,
      "learning_rate": 2.2072852975762338e-07,
      "loss": 0.0005,
      "step": 167680
    },
    {
      "epoch": 2.7438435735907714,
      "grad_norm": 0.112838514149189,
      "learning_rate": 2.2044889729554985e-07,
      "loss": 0.0006,
      "step": 167690
    },
    {
      "epoch": 2.7440071995418474,
      "grad_norm": 0.0022344677709043026,
      "learning_rate": 2.2016943808071678e-07,
      "loss": 0.0005,
      "step": 167700
    },
    {
      "epoch": 2.7441708254929233,
      "grad_norm": 0.018456408753991127,
      "learning_rate": 2.198901521232527e-07,
      "loss": 0.0008,
      "step": 167710
    },
    {
      "epoch": 2.744334451443999,
      "grad_norm": 0.08982288837432861,
      "learning_rate": 2.1961103943328343e-07,
      "loss": 0.0008,
      "step": 167720
    },
    {
      "epoch": 2.744498077395075,
      "grad_norm": 0.10520012676715851,
      "learning_rate": 2.1933210002092308e-07,
      "loss": 0.0009,
      "step": 167730
    },
    {
      "epoch": 2.744661703346151,
      "grad_norm": 0.12554024159908295,
      "learning_rate": 2.190533338962847e-07,
      "loss": 0.0007,
      "step": 167740
    },
    {
      "epoch": 2.7448253292972264,
      "grad_norm": 0.08705730736255646,
      "learning_rate": 2.1877474106947238e-07,
      "loss": 0.0004,
      "step": 167750
    },
    {
      "epoch": 2.7449889552483024,
      "grad_norm": 0.05151456966996193,
      "learning_rate": 2.1849632155058364e-07,
      "loss": 0.0008,
      "step": 167760
    },
    {
      "epoch": 2.7451525811993784,
      "grad_norm": 0.16612239181995392,
      "learning_rate": 2.1821807534971207e-07,
      "loss": 0.0008,
      "step": 167770
    },
    {
      "epoch": 2.745316207150454,
      "grad_norm": 0.08895605802536011,
      "learning_rate": 2.179400024769418e-07,
      "loss": 0.0008,
      "step": 167780
    },
    {
      "epoch": 2.74547983310153,
      "grad_norm": 0.05931415408849716,
      "learning_rate": 2.1766210294235367e-07,
      "loss": 0.0007,
      "step": 167790
    },
    {
      "epoch": 2.745643459052606,
      "grad_norm": 0.06572817265987396,
      "learning_rate": 2.1738437675601965e-07,
      "loss": 0.0005,
      "step": 167800
    },
    {
      "epoch": 2.7458070850036815,
      "grad_norm": 0.12826888263225555,
      "learning_rate": 2.171068239280083e-07,
      "loss": 0.0008,
      "step": 167810
    },
    {
      "epoch": 2.7459707109547575,
      "grad_norm": 0.09133157879114151,
      "learning_rate": 2.1682944446837884e-07,
      "loss": 0.0008,
      "step": 167820
    },
    {
      "epoch": 2.7461343369058335,
      "grad_norm": 0.03975919261574745,
      "learning_rate": 2.165522383871854e-07,
      "loss": 0.0006,
      "step": 167830
    },
    {
      "epoch": 2.746297962856909,
      "grad_norm": 0.2132592797279358,
      "learning_rate": 2.1627520569447723e-07,
      "loss": 0.0006,
      "step": 167840
    },
    {
      "epoch": 2.746461588807985,
      "grad_norm": 0.060637082904577255,
      "learning_rate": 2.159983464002946e-07,
      "loss": 0.0004,
      "step": 167850
    },
    {
      "epoch": 2.7466252147590606,
      "grad_norm": 0.037224747240543365,
      "learning_rate": 2.1572166051467502e-07,
      "loss": 0.0005,
      "step": 167860
    },
    {
      "epoch": 2.7467888407101366,
      "grad_norm": 0.05405132472515106,
      "learning_rate": 2.1544514804764549e-07,
      "loss": 0.0013,
      "step": 167870
    },
    {
      "epoch": 2.7469524666612126,
      "grad_norm": 0.12484581768512726,
      "learning_rate": 2.1516880900923076e-07,
      "loss": 0.0006,
      "step": 167880
    },
    {
      "epoch": 2.747116092612288,
      "grad_norm": 0.17253288626670837,
      "learning_rate": 2.148926434094456e-07,
      "loss": 0.0008,
      "step": 167890
    },
    {
      "epoch": 2.747279718563364,
      "grad_norm": 0.03848373889923096,
      "learning_rate": 2.14616651258302e-07,
      "loss": 0.0009,
      "step": 167900
    },
    {
      "epoch": 2.7474433445144397,
      "grad_norm": 0.10382080078125,
      "learning_rate": 2.1434083256580364e-07,
      "loss": 0.0008,
      "step": 167910
    },
    {
      "epoch": 2.7476069704655157,
      "grad_norm": 0.02144470252096653,
      "learning_rate": 2.140651873419486e-07,
      "loss": 0.0003,
      "step": 167920
    },
    {
      "epoch": 2.7477705964165917,
      "grad_norm": 0.037637434899806976,
      "learning_rate": 2.1378971559672723e-07,
      "loss": 0.0005,
      "step": 167930
    },
    {
      "epoch": 2.7479342223676673,
      "grad_norm": 0.05315548554062843,
      "learning_rate": 2.135144173401249e-07,
      "loss": 0.0013,
      "step": 167940
    },
    {
      "epoch": 2.7480978483187433,
      "grad_norm": 0.0408809520304203,
      "learning_rate": 2.132392925821214e-07,
      "loss": 0.0007,
      "step": 167950
    },
    {
      "epoch": 2.7482614742698193,
      "grad_norm": 0.0859406515955925,
      "learning_rate": 2.1296434133268873e-07,
      "loss": 0.0008,
      "step": 167960
    },
    {
      "epoch": 2.748425100220895,
      "grad_norm": 0.0014430563896894455,
      "learning_rate": 2.126895636017934e-07,
      "loss": 0.0005,
      "step": 167970
    },
    {
      "epoch": 2.748588726171971,
      "grad_norm": 0.16755545139312744,
      "learning_rate": 2.1241495939939572e-07,
      "loss": 0.0011,
      "step": 167980
    },
    {
      "epoch": 2.748752352123047,
      "grad_norm": 0.021886218339204788,
      "learning_rate": 2.1214052873544832e-07,
      "loss": 0.0005,
      "step": 167990
    },
    {
      "epoch": 2.7489159780741224,
      "grad_norm": 0.020729944109916687,
      "learning_rate": 2.1186627161989993e-07,
      "loss": 0.0008,
      "step": 168000
    },
    {
      "epoch": 2.7490796040251984,
      "grad_norm": 0.0311489999294281,
      "learning_rate": 2.115921880626909e-07,
      "loss": 0.0009,
      "step": 168010
    },
    {
      "epoch": 2.7492432299762743,
      "grad_norm": 0.15849840641021729,
      "learning_rate": 2.1131827807375716e-07,
      "loss": 0.0009,
      "step": 168020
    },
    {
      "epoch": 2.74940685592735,
      "grad_norm": 0.10417167842388153,
      "learning_rate": 2.1104454166302634e-07,
      "loss": 0.001,
      "step": 168030
    },
    {
      "epoch": 2.749570481878426,
      "grad_norm": 0.043993063271045685,
      "learning_rate": 2.1077097884042053e-07,
      "loss": 0.001,
      "step": 168040
    },
    {
      "epoch": 2.749734107829502,
      "grad_norm": 0.005415800493210554,
      "learning_rate": 2.104975896158562e-07,
      "loss": 0.0005,
      "step": 168050
    },
    {
      "epoch": 2.7498977337805774,
      "grad_norm": 0.052111536264419556,
      "learning_rate": 2.1022437399924267e-07,
      "loss": 0.001,
      "step": 168060
    },
    {
      "epoch": 2.7500613597316534,
      "grad_norm": 0.07010488957166672,
      "learning_rate": 2.0995133200048478e-07,
      "loss": 0.0006,
      "step": 168070
    },
    {
      "epoch": 2.7502249856827294,
      "grad_norm": 0.033969417214393616,
      "learning_rate": 2.0967846362947742e-07,
      "loss": 0.0007,
      "step": 168080
    },
    {
      "epoch": 2.750388611633805,
      "grad_norm": 0.1209041103720665,
      "learning_rate": 2.0940576889611318e-07,
      "loss": 0.0005,
      "step": 168090
    },
    {
      "epoch": 2.750552237584881,
      "grad_norm": 0.0452299602329731,
      "learning_rate": 2.0913324781027534e-07,
      "loss": 0.0011,
      "step": 168100
    },
    {
      "epoch": 2.750715863535957,
      "grad_norm": 0.001307224971242249,
      "learning_rate": 2.088609003818426e-07,
      "loss": 0.0006,
      "step": 168110
    },
    {
      "epoch": 2.7508794894870325,
      "grad_norm": 0.09175059199333191,
      "learning_rate": 2.0858872662068764e-07,
      "loss": 0.0006,
      "step": 168120
    },
    {
      "epoch": 2.7510431154381085,
      "grad_norm": 0.033883705735206604,
      "learning_rate": 2.0831672653667535e-07,
      "loss": 0.0007,
      "step": 168130
    },
    {
      "epoch": 2.7512067413891845,
      "grad_norm": 0.08595077693462372,
      "learning_rate": 2.080449001396656e-07,
      "loss": 0.0005,
      "step": 168140
    },
    {
      "epoch": 2.75137036734026,
      "grad_norm": 0.06489706039428711,
      "learning_rate": 2.0777324743951056e-07,
      "loss": 0.0007,
      "step": 168150
    },
    {
      "epoch": 2.751533993291336,
      "grad_norm": 0.050071485340595245,
      "learning_rate": 2.075017684460573e-07,
      "loss": 0.0015,
      "step": 168160
    },
    {
      "epoch": 2.751697619242412,
      "grad_norm": 0.05935819447040558,
      "learning_rate": 2.0723046316914686e-07,
      "loss": 0.0006,
      "step": 168170
    },
    {
      "epoch": 2.7518612451934876,
      "grad_norm": 0.04035269469022751,
      "learning_rate": 2.069593316186125e-07,
      "loss": 0.0005,
      "step": 168180
    },
    {
      "epoch": 2.7520248711445636,
      "grad_norm": 0.04774356260895729,
      "learning_rate": 2.0668837380428296e-07,
      "loss": 0.0005,
      "step": 168190
    },
    {
      "epoch": 2.7521884970956396,
      "grad_norm": 0.06154807284474373,
      "learning_rate": 2.0641758973597936e-07,
      "loss": 0.0004,
      "step": 168200
    },
    {
      "epoch": 2.752352123046715,
      "grad_norm": 0.06879732757806778,
      "learning_rate": 2.0614697942351714e-07,
      "loss": 0.0005,
      "step": 168210
    },
    {
      "epoch": 2.752515748997791,
      "grad_norm": 0.051487430930137634,
      "learning_rate": 2.058765428767051e-07,
      "loss": 0.0007,
      "step": 168220
    },
    {
      "epoch": 2.752679374948867,
      "grad_norm": 0.09697844088077545,
      "learning_rate": 2.0560628010534655e-07,
      "loss": 0.0003,
      "step": 168230
    },
    {
      "epoch": 2.7528430008999427,
      "grad_norm": 0.13867394626140594,
      "learning_rate": 2.0533619111923697e-07,
      "loss": 0.0007,
      "step": 168240
    },
    {
      "epoch": 2.7530066268510187,
      "grad_norm": 0.04300633445382118,
      "learning_rate": 2.050662759281663e-07,
      "loss": 0.0004,
      "step": 168250
    },
    {
      "epoch": 2.7531702528020943,
      "grad_norm": 0.1252574324607849,
      "learning_rate": 2.047965345419195e-07,
      "loss": 0.0007,
      "step": 168260
    },
    {
      "epoch": 2.7533338787531703,
      "grad_norm": 0.003122778609395027,
      "learning_rate": 2.0452696697027264e-07,
      "loss": 0.0006,
      "step": 168270
    },
    {
      "epoch": 2.7534975047042463,
      "grad_norm": 0.04841751605272293,
      "learning_rate": 2.0425757322299843e-07,
      "loss": 0.0004,
      "step": 168280
    },
    {
      "epoch": 2.753661130655322,
      "grad_norm": 0.03142056614160538,
      "learning_rate": 2.039883533098602e-07,
      "loss": 0.0007,
      "step": 168290
    },
    {
      "epoch": 2.753824756606398,
      "grad_norm": 0.13306310772895813,
      "learning_rate": 2.0371930724061795e-07,
      "loss": 0.0009,
      "step": 168300
    },
    {
      "epoch": 2.7539883825574734,
      "grad_norm": 0.11865317821502686,
      "learning_rate": 2.034504350250227e-07,
      "loss": 0.0008,
      "step": 168310
    },
    {
      "epoch": 2.7541520085085494,
      "grad_norm": 0.01555688213557005,
      "learning_rate": 2.0318173667282116e-07,
      "loss": 0.0007,
      "step": 168320
    },
    {
      "epoch": 2.7543156344596254,
      "grad_norm": 0.06026741862297058,
      "learning_rate": 2.0291321219375327e-07,
      "loss": 0.0005,
      "step": 168330
    },
    {
      "epoch": 2.754479260410701,
      "grad_norm": 0.11783447861671448,
      "learning_rate": 2.0264486159755125e-07,
      "loss": 0.0005,
      "step": 168340
    },
    {
      "epoch": 2.754642886361777,
      "grad_norm": 0.039642784744501114,
      "learning_rate": 2.0237668489394346e-07,
      "loss": 0.0004,
      "step": 168350
    },
    {
      "epoch": 2.754806512312853,
      "grad_norm": 0.07264643162488937,
      "learning_rate": 2.0210868209264932e-07,
      "loss": 0.0004,
      "step": 168360
    },
    {
      "epoch": 2.7549701382639284,
      "grad_norm": 0.10343144834041595,
      "learning_rate": 2.0184085320338443e-07,
      "loss": 0.001,
      "step": 168370
    },
    {
      "epoch": 2.7551337642150044,
      "grad_norm": 0.08168608695268631,
      "learning_rate": 2.0157319823585651e-07,
      "loss": 0.0003,
      "step": 168380
    },
    {
      "epoch": 2.7552973901660804,
      "grad_norm": 0.16937950253486633,
      "learning_rate": 2.0130571719976676e-07,
      "loss": 0.0009,
      "step": 168390
    },
    {
      "epoch": 2.755461016117156,
      "grad_norm": 0.09547882527112961,
      "learning_rate": 2.010384101048124e-07,
      "loss": 0.0006,
      "step": 168400
    },
    {
      "epoch": 2.755624642068232,
      "grad_norm": 0.006628206465393305,
      "learning_rate": 2.0077127696068122e-07,
      "loss": 0.0005,
      "step": 168410
    },
    {
      "epoch": 2.755788268019308,
      "grad_norm": 0.1632036715745926,
      "learning_rate": 2.0050431777705714e-07,
      "loss": 0.0008,
      "step": 168420
    },
    {
      "epoch": 2.7559518939703835,
      "grad_norm": 0.08565066754817963,
      "learning_rate": 2.0023753256361577e-07,
      "loss": 0.0008,
      "step": 168430
    },
    {
      "epoch": 2.7561155199214595,
      "grad_norm": 0.26326531171798706,
      "learning_rate": 1.9997092133002827e-07,
      "loss": 0.0013,
      "step": 168440
    },
    {
      "epoch": 2.7562791458725355,
      "grad_norm": 0.08266288042068481,
      "learning_rate": 1.99704484085958e-07,
      "loss": 0.0005,
      "step": 168450
    },
    {
      "epoch": 2.756442771823611,
      "grad_norm": 0.023020917549729347,
      "learning_rate": 1.994382208410628e-07,
      "loss": 0.0003,
      "step": 168460
    },
    {
      "epoch": 2.756606397774687,
      "grad_norm": 0.03823775053024292,
      "learning_rate": 1.9917213160499437e-07,
      "loss": 0.0006,
      "step": 168470
    },
    {
      "epoch": 2.756770023725763,
      "grad_norm": 0.16163672506809235,
      "learning_rate": 1.9890621638739783e-07,
      "loss": 0.0014,
      "step": 168480
    },
    {
      "epoch": 2.7569336496768386,
      "grad_norm": 0.002065907930955291,
      "learning_rate": 1.9864047519791207e-07,
      "loss": 0.0004,
      "step": 168490
    },
    {
      "epoch": 2.7570972756279146,
      "grad_norm": 0.06376825273036957,
      "learning_rate": 1.9837490804616834e-07,
      "loss": 0.0008,
      "step": 168500
    },
    {
      "epoch": 2.7572609015789906,
      "grad_norm": 0.002479027258232236,
      "learning_rate": 1.9810951494179497e-07,
      "loss": 0.0006,
      "step": 168510
    },
    {
      "epoch": 2.757424527530066,
      "grad_norm": 0.01967574656009674,
      "learning_rate": 1.978442958944099e-07,
      "loss": 0.0005,
      "step": 168520
    },
    {
      "epoch": 2.757588153481142,
      "grad_norm": 0.0020061624236404896,
      "learning_rate": 1.975792509136276e-07,
      "loss": 0.0008,
      "step": 168530
    },
    {
      "epoch": 2.757751779432218,
      "grad_norm": 0.14201155304908752,
      "learning_rate": 1.9731438000905602e-07,
      "loss": 0.0004,
      "step": 168540
    },
    {
      "epoch": 2.7579154053832937,
      "grad_norm": 0.09914567321538925,
      "learning_rate": 1.9704968319029405e-07,
      "loss": 0.0005,
      "step": 168550
    },
    {
      "epoch": 2.7580790313343697,
      "grad_norm": 0.15743231773376465,
      "learning_rate": 1.9678516046693797e-07,
      "loss": 0.001,
      "step": 168560
    },
    {
      "epoch": 2.7582426572854457,
      "grad_norm": 0.012206529267132282,
      "learning_rate": 1.9652081184857507e-07,
      "loss": 0.0003,
      "step": 168570
    },
    {
      "epoch": 2.7584062832365213,
      "grad_norm": 0.21962665021419525,
      "learning_rate": 1.9625663734478883e-07,
      "loss": 0.0008,
      "step": 168580
    },
    {
      "epoch": 2.7585699091875973,
      "grad_norm": 0.017736637964844704,
      "learning_rate": 1.959926369651527e-07,
      "loss": 0.0003,
      "step": 168590
    },
    {
      "epoch": 2.7587335351386733,
      "grad_norm": 0.10496079176664352,
      "learning_rate": 1.9572881071923843e-07,
      "loss": 0.0008,
      "step": 168600
    },
    {
      "epoch": 2.758897161089749,
      "grad_norm": 0.03617104887962341,
      "learning_rate": 1.954651586166073e-07,
      "loss": 0.0005,
      "step": 168610
    },
    {
      "epoch": 2.759060787040825,
      "grad_norm": 0.01400375459343195,
      "learning_rate": 1.9520168066681722e-07,
      "loss": 0.0005,
      "step": 168620
    },
    {
      "epoch": 2.7592244129919004,
      "grad_norm": 0.07204016298055649,
      "learning_rate": 1.9493837687941886e-07,
      "loss": 0.0005,
      "step": 168630
    },
    {
      "epoch": 2.7593880389429764,
      "grad_norm": 0.11820922791957855,
      "learning_rate": 1.946752472639546e-07,
      "loss": 0.0013,
      "step": 168640
    },
    {
      "epoch": 2.7595516648940523,
      "grad_norm": 0.031934287399053574,
      "learning_rate": 1.9441229182996402e-07,
      "loss": 0.001,
      "step": 168650
    },
    {
      "epoch": 2.759715290845128,
      "grad_norm": 0.05650338530540466,
      "learning_rate": 1.9414951058697729e-07,
      "loss": 0.0011,
      "step": 168660
    },
    {
      "epoch": 2.759878916796204,
      "grad_norm": 0.007053412031382322,
      "learning_rate": 1.9388690354452012e-07,
      "loss": 0.0005,
      "step": 168670
    },
    {
      "epoch": 2.7600425427472794,
      "grad_norm": 0.08474508672952652,
      "learning_rate": 1.936244707121121e-07,
      "loss": 0.0007,
      "step": 168680
    },
    {
      "epoch": 2.7602061686983554,
      "grad_norm": 0.023448863998055458,
      "learning_rate": 1.9336221209926454e-07,
      "loss": 0.001,
      "step": 168690
    },
    {
      "epoch": 2.7603697946494314,
      "grad_norm": 0.00744744623079896,
      "learning_rate": 1.9310012771548536e-07,
      "loss": 0.0002,
      "step": 168700
    },
    {
      "epoch": 2.760533420600507,
      "grad_norm": 0.054840464144945145,
      "learning_rate": 1.9283821757027254e-07,
      "loss": 0.0005,
      "step": 168710
    },
    {
      "epoch": 2.760697046551583,
      "grad_norm": 0.10844000428915024,
      "learning_rate": 1.9257648167312127e-07,
      "loss": 0.0008,
      "step": 168720
    },
    {
      "epoch": 2.760860672502659,
      "grad_norm": 0.03459271416068077,
      "learning_rate": 1.9231492003351838e-07,
      "loss": 0.0008,
      "step": 168730
    },
    {
      "epoch": 2.7610242984537345,
      "grad_norm": 0.039451319724321365,
      "learning_rate": 1.9205353266094352e-07,
      "loss": 0.0009,
      "step": 168740
    },
    {
      "epoch": 2.7611879244048105,
      "grad_norm": 0.11334728449583054,
      "learning_rate": 1.9179231956487353e-07,
      "loss": 0.0004,
      "step": 168750
    },
    {
      "epoch": 2.7613515503558865,
      "grad_norm": 0.06936286389827728,
      "learning_rate": 1.9153128075477534e-07,
      "loss": 0.0006,
      "step": 168760
    },
    {
      "epoch": 2.761515176306962,
      "grad_norm": 0.09018911421298981,
      "learning_rate": 1.912704162401119e-07,
      "loss": 0.0008,
      "step": 168770
    },
    {
      "epoch": 2.761678802258038,
      "grad_norm": 0.0995088741183281,
      "learning_rate": 1.910097260303373e-07,
      "loss": 0.0002,
      "step": 168780
    },
    {
      "epoch": 2.761842428209114,
      "grad_norm": 0.12075003981590271,
      "learning_rate": 1.907492101349029e-07,
      "loss": 0.0006,
      "step": 168790
    },
    {
      "epoch": 2.7620060541601896,
      "grad_norm": 0.002675691619515419,
      "learning_rate": 1.9048886856325055e-07,
      "loss": 0.0005,
      "step": 168800
    },
    {
      "epoch": 2.7621696801112656,
      "grad_norm": 0.08235309273004532,
      "learning_rate": 1.9022870132481774e-07,
      "loss": 0.0006,
      "step": 168810
    },
    {
      "epoch": 2.7623333060623416,
      "grad_norm": 0.002063245279714465,
      "learning_rate": 1.8996870842903415e-07,
      "loss": 0.0013,
      "step": 168820
    },
    {
      "epoch": 2.762496932013417,
      "grad_norm": 0.003093044040724635,
      "learning_rate": 1.8970888988532444e-07,
      "loss": 0.0005,
      "step": 168830
    },
    {
      "epoch": 2.762660557964493,
      "grad_norm": 0.05535441264510155,
      "learning_rate": 1.8944924570310662e-07,
      "loss": 0.0004,
      "step": 168840
    },
    {
      "epoch": 2.762824183915569,
      "grad_norm": 0.06835031509399414,
      "learning_rate": 1.8918977589179098e-07,
      "loss": 0.0017,
      "step": 168850
    },
    {
      "epoch": 2.7629878098666447,
      "grad_norm": 0.05116264522075653,
      "learning_rate": 1.889304804607839e-07,
      "loss": 0.0004,
      "step": 168860
    },
    {
      "epoch": 2.7631514358177207,
      "grad_norm": 0.08292914927005768,
      "learning_rate": 1.886713594194839e-07,
      "loss": 0.0007,
      "step": 168870
    },
    {
      "epoch": 2.7633150617687967,
      "grad_norm": 0.03854930028319359,
      "learning_rate": 1.8841241277728295e-07,
      "loss": 0.0006,
      "step": 168880
    },
    {
      "epoch": 2.7634786877198723,
      "grad_norm": 0.15408547222614288,
      "learning_rate": 1.8815364054356798e-07,
      "loss": 0.0015,
      "step": 168890
    },
    {
      "epoch": 2.7636423136709483,
      "grad_norm": 0.03105132281780243,
      "learning_rate": 1.878950427277182e-07,
      "loss": 0.0004,
      "step": 168900
    },
    {
      "epoch": 2.7638059396220243,
      "grad_norm": 0.07394374907016754,
      "learning_rate": 1.8763661933910826e-07,
      "loss": 0.001,
      "step": 168910
    },
    {
      "epoch": 2.7639695655731,
      "grad_norm": 0.08863889425992966,
      "learning_rate": 1.8737837038710405e-07,
      "loss": 0.001,
      "step": 168920
    },
    {
      "epoch": 2.764133191524176,
      "grad_norm": 0.10919832438230515,
      "learning_rate": 1.8712029588106862e-07,
      "loss": 0.0008,
      "step": 168930
    },
    {
      "epoch": 2.764296817475252,
      "grad_norm": 0.11833513528108597,
      "learning_rate": 1.8686239583035336e-07,
      "loss": 0.0007,
      "step": 168940
    },
    {
      "epoch": 2.7644604434263274,
      "grad_norm": 0.02775750495493412,
      "learning_rate": 1.8660467024430806e-07,
      "loss": 0.0005,
      "step": 168950
    },
    {
      "epoch": 2.7646240693774033,
      "grad_norm": 0.06226540356874466,
      "learning_rate": 1.8634711913227576e-07,
      "loss": 0.0007,
      "step": 168960
    },
    {
      "epoch": 2.7647876953284793,
      "grad_norm": 0.01621055230498314,
      "learning_rate": 1.8608974250359014e-07,
      "loss": 0.0007,
      "step": 168970
    },
    {
      "epoch": 2.764951321279555,
      "grad_norm": 0.051388658583164215,
      "learning_rate": 1.8583254036758204e-07,
      "loss": 0.0004,
      "step": 168980
    },
    {
      "epoch": 2.765114947230631,
      "grad_norm": 0.1612025797367096,
      "learning_rate": 1.8557551273357344e-07,
      "loss": 0.0013,
      "step": 168990
    },
    {
      "epoch": 2.765278573181707,
      "grad_norm": 0.10645221173763275,
      "learning_rate": 1.853186596108819e-07,
      "loss": 0.0009,
      "step": 169000
    },
    {
      "epoch": 2.7654421991327824,
      "grad_norm": 0.0673551931977272,
      "learning_rate": 1.8506198100881666e-07,
      "loss": 0.0017,
      "step": 169010
    },
    {
      "epoch": 2.7656058250838584,
      "grad_norm": 0.09177505224943161,
      "learning_rate": 1.848054769366825e-07,
      "loss": 0.0005,
      "step": 169020
    },
    {
      "epoch": 2.765769451034934,
      "grad_norm": 0.04661207273602486,
      "learning_rate": 1.8454914740377695e-07,
      "loss": 0.0005,
      "step": 169030
    },
    {
      "epoch": 2.76593307698601,
      "grad_norm": 0.07712335139513016,
      "learning_rate": 1.8429299241939036e-07,
      "loss": 0.0012,
      "step": 169040
    },
    {
      "epoch": 2.766096702937086,
      "grad_norm": 0.06612737476825714,
      "learning_rate": 1.8403701199280976e-07,
      "loss": 0.0008,
      "step": 169050
    },
    {
      "epoch": 2.7662603288881615,
      "grad_norm": 0.06156130135059357,
      "learning_rate": 1.8378120613331108e-07,
      "loss": 0.0006,
      "step": 169060
    },
    {
      "epoch": 2.7664239548392375,
      "grad_norm": 0.05858384072780609,
      "learning_rate": 1.8352557485016963e-07,
      "loss": 0.0003,
      "step": 169070
    },
    {
      "epoch": 2.766587580790313,
      "grad_norm": 0.09774133563041687,
      "learning_rate": 1.8327011815264916e-07,
      "loss": 0.0007,
      "step": 169080
    },
    {
      "epoch": 2.766751206741389,
      "grad_norm": 0.12427332997322083,
      "learning_rate": 1.830148360500106e-07,
      "loss": 0.0011,
      "step": 169090
    },
    {
      "epoch": 2.766914832692465,
      "grad_norm": 0.13921351730823517,
      "learning_rate": 1.8275972855150648e-07,
      "loss": 0.0007,
      "step": 169100
    },
    {
      "epoch": 2.7670784586435406,
      "grad_norm": 0.029401782900094986,
      "learning_rate": 1.8250479566638446e-07,
      "loss": 0.001,
      "step": 169110
    },
    {
      "epoch": 2.7672420845946166,
      "grad_norm": 0.004431414883583784,
      "learning_rate": 1.8225003740388546e-07,
      "loss": 0.001,
      "step": 169120
    },
    {
      "epoch": 2.7674057105456926,
      "grad_norm": 0.0358114130795002,
      "learning_rate": 1.8199545377324323e-07,
      "loss": 0.0004,
      "step": 169130
    },
    {
      "epoch": 2.767569336496768,
      "grad_norm": 0.07633160054683685,
      "learning_rate": 1.8174104478368593e-07,
      "loss": 0.0005,
      "step": 169140
    },
    {
      "epoch": 2.767732962447844,
      "grad_norm": 0.06541363149881363,
      "learning_rate": 1.8148681044443505e-07,
      "loss": 0.0048,
      "step": 169150
    },
    {
      "epoch": 2.76789658839892,
      "grad_norm": 0.13870662450790405,
      "learning_rate": 1.812327507647066e-07,
      "loss": 0.0009,
      "step": 169160
    },
    {
      "epoch": 2.7680602143499957,
      "grad_norm": 0.16139960289001465,
      "learning_rate": 1.8097886575370983e-07,
      "loss": 0.0014,
      "step": 169170
    },
    {
      "epoch": 2.7682238403010717,
      "grad_norm": 0.18176931142807007,
      "learning_rate": 1.8072515542064573e-07,
      "loss": 0.001,
      "step": 169180
    },
    {
      "epoch": 2.7683874662521477,
      "grad_norm": 0.07228479534387589,
      "learning_rate": 1.8047161977471306e-07,
      "loss": 0.0003,
      "step": 169190
    },
    {
      "epoch": 2.7685510922032233,
      "grad_norm": 0.01882397197186947,
      "learning_rate": 1.8021825882510002e-07,
      "loss": 0.0006,
      "step": 169200
    },
    {
      "epoch": 2.7687147181542993,
      "grad_norm": 0.05472211912274361,
      "learning_rate": 1.7996507258099206e-07,
      "loss": 0.0009,
      "step": 169210
    },
    {
      "epoch": 2.7688783441053753,
      "grad_norm": 0.06442705541849136,
      "learning_rate": 1.7971206105156458e-07,
      "loss": 0.0005,
      "step": 169220
    },
    {
      "epoch": 2.769041970056451,
      "grad_norm": 0.12161727994680405,
      "learning_rate": 1.7945922424599026e-07,
      "loss": 0.0009,
      "step": 169230
    },
    {
      "epoch": 2.769205596007527,
      "grad_norm": 0.10433565080165863,
      "learning_rate": 1.792065621734329e-07,
      "loss": 0.0006,
      "step": 169240
    },
    {
      "epoch": 2.769369221958603,
      "grad_norm": 0.02848692052066326,
      "learning_rate": 1.7895407484305126e-07,
      "loss": 0.0008,
      "step": 169250
    },
    {
      "epoch": 2.7695328479096784,
      "grad_norm": 0.06888028979301453,
      "learning_rate": 1.787017622639975e-07,
      "loss": 0.0012,
      "step": 169260
    },
    {
      "epoch": 2.7696964738607543,
      "grad_norm": 0.06015644967556,
      "learning_rate": 1.7844962444541646e-07,
      "loss": 0.0005,
      "step": 169270
    },
    {
      "epoch": 2.7698600998118303,
      "grad_norm": 0.04998675361275673,
      "learning_rate": 1.7819766139644922e-07,
      "loss": 0.0013,
      "step": 169280
    },
    {
      "epoch": 2.770023725762906,
      "grad_norm": 0.032340746372938156,
      "learning_rate": 1.779458731262268e-07,
      "loss": 0.0007,
      "step": 169290
    },
    {
      "epoch": 2.770187351713982,
      "grad_norm": 0.1105969026684761,
      "learning_rate": 1.77694259643878e-07,
      "loss": 0.0004,
      "step": 169300
    },
    {
      "epoch": 2.770350977665058,
      "grad_norm": 0.004225532524287701,
      "learning_rate": 1.7744282095852106e-07,
      "loss": 0.0006,
      "step": 169310
    },
    {
      "epoch": 2.7705146036161334,
      "grad_norm": 0.04925353825092316,
      "learning_rate": 1.7719155707927148e-07,
      "loss": 0.0004,
      "step": 169320
    },
    {
      "epoch": 2.7706782295672094,
      "grad_norm": 0.04826878756284714,
      "learning_rate": 1.7694046801523812e-07,
      "loss": 0.0006,
      "step": 169330
    },
    {
      "epoch": 2.7708418555182854,
      "grad_norm": 0.029839392751455307,
      "learning_rate": 1.7668955377551922e-07,
      "loss": 0.0004,
      "step": 169340
    },
    {
      "epoch": 2.771005481469361,
      "grad_norm": 0.01830497942864895,
      "learning_rate": 1.764388143692125e-07,
      "loss": 0.0008,
      "step": 169350
    },
    {
      "epoch": 2.771169107420437,
      "grad_norm": 0.0425875186920166,
      "learning_rate": 1.761882498054046e-07,
      "loss": 0.0005,
      "step": 169360
    },
    {
      "epoch": 2.771332733371513,
      "grad_norm": 0.005086121615022421,
      "learning_rate": 1.7593786009317992e-07,
      "loss": 0.0014,
      "step": 169370
    },
    {
      "epoch": 2.7714963593225885,
      "grad_norm": 0.01563134975731373,
      "learning_rate": 1.756876452416123e-07,
      "loss": 0.0004,
      "step": 169380
    },
    {
      "epoch": 2.7716599852736645,
      "grad_norm": 0.003388570621609688,
      "learning_rate": 1.754376052597734e-07,
      "loss": 0.0009,
      "step": 169390
    },
    {
      "epoch": 2.77182361122474,
      "grad_norm": 0.03311295062303543,
      "learning_rate": 1.751877401567259e-07,
      "loss": 0.0009,
      "step": 169400
    },
    {
      "epoch": 2.771987237175816,
      "grad_norm": 0.02959679253399372,
      "learning_rate": 1.7493804994152598e-07,
      "loss": 0.0005,
      "step": 169410
    },
    {
      "epoch": 2.772150863126892,
      "grad_norm": 0.563259482383728,
      "learning_rate": 1.7468853462322576e-07,
      "loss": 0.001,
      "step": 169420
    },
    {
      "epoch": 2.7723144890779676,
      "grad_norm": 0.02752726711332798,
      "learning_rate": 1.7443919421086865e-07,
      "loss": 0.0008,
      "step": 169430
    },
    {
      "epoch": 2.7724781150290436,
      "grad_norm": 0.004192272666841745,
      "learning_rate": 1.7419002871349288e-07,
      "loss": 0.0002,
      "step": 169440
    },
    {
      "epoch": 2.772641740980119,
      "grad_norm": 0.10158471018075943,
      "learning_rate": 1.7394103814013018e-07,
      "loss": 0.0007,
      "step": 169450
    },
    {
      "epoch": 2.772805366931195,
      "grad_norm": 0.043052881956100464,
      "learning_rate": 1.73692222499805e-07,
      "loss": 0.0012,
      "step": 169460
    },
    {
      "epoch": 2.772968992882271,
      "grad_norm": 0.02363894321024418,
      "learning_rate": 1.7344358180153787e-07,
      "loss": 0.0005,
      "step": 169470
    },
    {
      "epoch": 2.7731326188333467,
      "grad_norm": 0.062009889632463455,
      "learning_rate": 1.7319511605433992e-07,
      "loss": 0.0005,
      "step": 169480
    },
    {
      "epoch": 2.7732962447844227,
      "grad_norm": 0.0013674815418198705,
      "learning_rate": 1.729468252672184e-07,
      "loss": 0.0006,
      "step": 169490
    },
    {
      "epoch": 2.7734598707354987,
      "grad_norm": 0.0437176488339901,
      "learning_rate": 1.7269870944917276e-07,
      "loss": 0.0009,
      "step": 169500
    },
    {
      "epoch": 2.7736234966865743,
      "grad_norm": 0.02126944065093994,
      "learning_rate": 1.7245076860919697e-07,
      "loss": 0.0005,
      "step": 169510
    },
    {
      "epoch": 2.7737871226376503,
      "grad_norm": 0.0443657822906971,
      "learning_rate": 1.7220300275627766e-07,
      "loss": 0.0004,
      "step": 169520
    },
    {
      "epoch": 2.7739507485887263,
      "grad_norm": 0.04853919520974159,
      "learning_rate": 1.7195541189939712e-07,
      "loss": 0.0011,
      "step": 169530
    },
    {
      "epoch": 2.774114374539802,
      "grad_norm": 0.1551184505224228,
      "learning_rate": 1.7170799604752874e-07,
      "loss": 0.0007,
      "step": 169540
    },
    {
      "epoch": 2.774278000490878,
      "grad_norm": 0.07881522923707962,
      "learning_rate": 1.7146075520964035e-07,
      "loss": 0.0004,
      "step": 169550
    },
    {
      "epoch": 2.774441626441954,
      "grad_norm": 0.08043744415044785,
      "learning_rate": 1.712136893946953e-07,
      "loss": 0.0012,
      "step": 169560
    },
    {
      "epoch": 2.7746052523930294,
      "grad_norm": 0.08677130937576294,
      "learning_rate": 1.7096679861164755e-07,
      "loss": 0.0006,
      "step": 169570
    },
    {
      "epoch": 2.7747688783441053,
      "grad_norm": 0.1499408483505249,
      "learning_rate": 1.7072008286944776e-07,
      "loss": 0.0004,
      "step": 169580
    },
    {
      "epoch": 2.7749325042951813,
      "grad_norm": 0.05759814754128456,
      "learning_rate": 1.704735421770376e-07,
      "loss": 0.0022,
      "step": 169590
    },
    {
      "epoch": 2.775096130246257,
      "grad_norm": 0.029853580519557,
      "learning_rate": 1.7022717654335386e-07,
      "loss": 0.0007,
      "step": 169600
    },
    {
      "epoch": 2.775259756197333,
      "grad_norm": 0.014423384331166744,
      "learning_rate": 1.699809859773277e-07,
      "loss": 0.0013,
      "step": 169610
    },
    {
      "epoch": 2.775423382148409,
      "grad_norm": 0.14842648804187775,
      "learning_rate": 1.6973497048788145e-07,
      "loss": 0.0002,
      "step": 169620
    },
    {
      "epoch": 2.7755870080994844,
      "grad_norm": 0.008933762088418007,
      "learning_rate": 1.694891300839341e-07,
      "loss": 0.0005,
      "step": 169630
    },
    {
      "epoch": 2.7757506340505604,
      "grad_norm": 0.16842113435268402,
      "learning_rate": 1.6924346477439513e-07,
      "loss": 0.0021,
      "step": 169640
    },
    {
      "epoch": 2.7759142600016364,
      "grad_norm": 0.05917133018374443,
      "learning_rate": 1.689979745681708e-07,
      "loss": 0.0004,
      "step": 169650
    },
    {
      "epoch": 2.776077885952712,
      "grad_norm": 0.05606184899806976,
      "learning_rate": 1.687526594741584e-07,
      "loss": 0.0007,
      "step": 169660
    },
    {
      "epoch": 2.776241511903788,
      "grad_norm": 0.09376686066389084,
      "learning_rate": 1.6850751950125032e-07,
      "loss": 0.0005,
      "step": 169670
    },
    {
      "epoch": 2.776405137854864,
      "grad_norm": 0.04740943759679794,
      "learning_rate": 1.682625546583333e-07,
      "loss": 0.0006,
      "step": 169680
    },
    {
      "epoch": 2.7765687638059395,
      "grad_norm": 0.09572571516036987,
      "learning_rate": 1.680177649542847e-07,
      "loss": 0.0002,
      "step": 169690
    },
    {
      "epoch": 2.7767323897570155,
      "grad_norm": 0.060865387320518494,
      "learning_rate": 1.6777315039798014e-07,
      "loss": 0.0005,
      "step": 169700
    },
    {
      "epoch": 2.7768960157080915,
      "grad_norm": 0.012650974094867706,
      "learning_rate": 1.6752871099828373e-07,
      "loss": 0.0005,
      "step": 169710
    },
    {
      "epoch": 2.777059641659167,
      "grad_norm": 0.12507717311382294,
      "learning_rate": 1.6728444676405774e-07,
      "loss": 0.0006,
      "step": 169720
    },
    {
      "epoch": 2.777223267610243,
      "grad_norm": 0.10915827751159668,
      "learning_rate": 1.670403577041557e-07,
      "loss": 0.0015,
      "step": 169730
    },
    {
      "epoch": 2.777386893561319,
      "grad_norm": 0.16892381012439728,
      "learning_rate": 1.6679644382742387e-07,
      "loss": 0.0008,
      "step": 169740
    },
    {
      "epoch": 2.7775505195123946,
      "grad_norm": 0.03212336078286171,
      "learning_rate": 1.6655270514270571e-07,
      "loss": 0.0004,
      "step": 169750
    },
    {
      "epoch": 2.7777141454634706,
      "grad_norm": 0.0148609085008502,
      "learning_rate": 1.6630914165883416e-07,
      "loss": 0.0006,
      "step": 169760
    },
    {
      "epoch": 2.7778777714145466,
      "grad_norm": 0.04622737318277359,
      "learning_rate": 1.6606575338463991e-07,
      "loss": 0.0008,
      "step": 169770
    },
    {
      "epoch": 2.778041397365622,
      "grad_norm": 0.017943989485502243,
      "learning_rate": 1.6582254032894264e-07,
      "loss": 0.0016,
      "step": 169780
    },
    {
      "epoch": 2.778205023316698,
      "grad_norm": 0.008822825737297535,
      "learning_rate": 1.6557950250056076e-07,
      "loss": 0.0012,
      "step": 169790
    },
    {
      "epoch": 2.7783686492677737,
      "grad_norm": 0.10040675848722458,
      "learning_rate": 1.653366399083023e-07,
      "loss": 0.0005,
      "step": 169800
    },
    {
      "epoch": 2.7785322752188497,
      "grad_norm": 0.0025606669951230288,
      "learning_rate": 1.6509395256097128e-07,
      "loss": 0.0007,
      "step": 169810
    },
    {
      "epoch": 2.7786959011699257,
      "grad_norm": 0.015258150175213814,
      "learning_rate": 1.648514404673629e-07,
      "loss": 0.0006,
      "step": 169820
    },
    {
      "epoch": 2.7788595271210013,
      "grad_norm": 0.13851045072078705,
      "learning_rate": 1.6460910363627015e-07,
      "loss": 0.0007,
      "step": 169830
    },
    {
      "epoch": 2.7790231530720773,
      "grad_norm": 0.04888838529586792,
      "learning_rate": 1.643669420764754e-07,
      "loss": 0.0005,
      "step": 169840
    },
    {
      "epoch": 2.779186779023153,
      "grad_norm": 0.07551861554384232,
      "learning_rate": 1.641249557967567e-07,
      "loss": 0.0006,
      "step": 169850
    },
    {
      "epoch": 2.779350404974229,
      "grad_norm": 0.06484108418226242,
      "learning_rate": 1.6388314480588584e-07,
      "loss": 0.0012,
      "step": 169860
    },
    {
      "epoch": 2.779514030925305,
      "grad_norm": 0.016625039279460907,
      "learning_rate": 1.63641509112627e-07,
      "loss": 0.0008,
      "step": 169870
    },
    {
      "epoch": 2.7796776568763804,
      "grad_norm": 0.05826619267463684,
      "learning_rate": 1.6340004872573977e-07,
      "loss": 0.0008,
      "step": 169880
    },
    {
      "epoch": 2.7798412828274564,
      "grad_norm": 0.021110009402036667,
      "learning_rate": 1.631587636539772e-07,
      "loss": 0.0005,
      "step": 169890
    },
    {
      "epoch": 2.7800049087785323,
      "grad_norm": 0.02184729464352131,
      "learning_rate": 1.629176539060834e-07,
      "loss": 0.0006,
      "step": 169900
    },
    {
      "epoch": 2.780168534729608,
      "grad_norm": 0.04591450095176697,
      "learning_rate": 1.6267671949079967e-07,
      "loss": 0.0003,
      "step": 169910
    },
    {
      "epoch": 2.780332160680684,
      "grad_norm": 0.06512957066297531,
      "learning_rate": 1.624359604168585e-07,
      "loss": 0.0005,
      "step": 169920
    },
    {
      "epoch": 2.78049578663176,
      "grad_norm": 0.021881897002458572,
      "learning_rate": 1.621953766929868e-07,
      "loss": 0.0009,
      "step": 169930
    },
    {
      "epoch": 2.7806594125828354,
      "grad_norm": 0.024408766999840736,
      "learning_rate": 1.6195496832790592e-07,
      "loss": 0.0004,
      "step": 169940
    },
    {
      "epoch": 2.7808230385339114,
      "grad_norm": 0.040458228439092636,
      "learning_rate": 1.6171473533032888e-07,
      "loss": 0.0003,
      "step": 169950
    },
    {
      "epoch": 2.7809866644849874,
      "grad_norm": 0.12420674413442612,
      "learning_rate": 1.614746777089643e-07,
      "loss": 0.0003,
      "step": 169960
    },
    {
      "epoch": 2.781150290436063,
      "grad_norm": 0.06038803979754448,
      "learning_rate": 1.6123479547251352e-07,
      "loss": 0.0005,
      "step": 169970
    },
    {
      "epoch": 2.781313916387139,
      "grad_norm": 0.0764453336596489,
      "learning_rate": 1.609950886296724e-07,
      "loss": 0.0006,
      "step": 169980
    },
    {
      "epoch": 2.781477542338215,
      "grad_norm": 0.0035014047753065825,
      "learning_rate": 1.6075555718912784e-07,
      "loss": 0.0006,
      "step": 169990
    },
    {
      "epoch": 2.7816411682892905,
      "grad_norm": 0.03599059581756592,
      "learning_rate": 1.6051620115956456e-07,
      "loss": 0.0005,
      "step": 170000
    },
    {
      "epoch": 2.7818047942403665,
      "grad_norm": 0.06846936792135239,
      "learning_rate": 1.6027702054965677e-07,
      "loss": 0.0024,
      "step": 170010
    },
    {
      "epoch": 2.7819684201914425,
      "grad_norm": 0.04152410849928856,
      "learning_rate": 1.6003801536807583e-07,
      "loss": 0.0007,
      "step": 170020
    },
    {
      "epoch": 2.782132046142518,
      "grad_norm": 0.04233233258128166,
      "learning_rate": 1.597991856234843e-07,
      "loss": 0.0013,
      "step": 170030
    },
    {
      "epoch": 2.782295672093594,
      "grad_norm": 0.05166567116975784,
      "learning_rate": 1.5956053132453853e-07,
      "loss": 0.0004,
      "step": 170040
    },
    {
      "epoch": 2.78245929804467,
      "grad_norm": 0.23059377074241638,
      "learning_rate": 1.5932205247989052e-07,
      "loss": 0.0004,
      "step": 170050
    },
    {
      "epoch": 2.7826229239957456,
      "grad_norm": 0.0020703545305877924,
      "learning_rate": 1.590837490981828e-07,
      "loss": 0.0004,
      "step": 170060
    },
    {
      "epoch": 2.7827865499468216,
      "grad_norm": 0.08792093396186829,
      "learning_rate": 1.5884562118805512e-07,
      "loss": 0.0002,
      "step": 170070
    },
    {
      "epoch": 2.7829501758978976,
      "grad_norm": 0.12561355531215668,
      "learning_rate": 1.5860766875813782e-07,
      "loss": 0.0013,
      "step": 170080
    },
    {
      "epoch": 2.783113801848973,
      "grad_norm": 0.05214882269501686,
      "learning_rate": 1.5836989181705675e-07,
      "loss": 0.0006,
      "step": 170090
    },
    {
      "epoch": 2.783277427800049,
      "grad_norm": 0.047570452094078064,
      "learning_rate": 1.581322903734306e-07,
      "loss": 0.0014,
      "step": 170100
    },
    {
      "epoch": 2.783441053751125,
      "grad_norm": 0.06508895754814148,
      "learning_rate": 1.578948644358719e-07,
      "loss": 0.0005,
      "step": 170110
    },
    {
      "epoch": 2.7836046797022007,
      "grad_norm": 0.09429115802049637,
      "learning_rate": 1.5765761401298652e-07,
      "loss": 0.0012,
      "step": 170120
    },
    {
      "epoch": 2.7837683056532767,
      "grad_norm": 0.08601327240467072,
      "learning_rate": 1.574205391133743e-07,
      "loss": 0.0014,
      "step": 170130
    },
    {
      "epoch": 2.7839319316043527,
      "grad_norm": 0.12766635417938232,
      "learning_rate": 1.5718363974562945e-07,
      "loss": 0.0005,
      "step": 170140
    },
    {
      "epoch": 2.7840955575554283,
      "grad_norm": 0.1010679379105568,
      "learning_rate": 1.5694691591833734e-07,
      "loss": 0.0003,
      "step": 170150
    },
    {
      "epoch": 2.7842591835065043,
      "grad_norm": 0.018694084137678146,
      "learning_rate": 1.5671036764007997e-07,
      "loss": 0.001,
      "step": 170160
    },
    {
      "epoch": 2.78442280945758,
      "grad_norm": 0.014573877677321434,
      "learning_rate": 1.5647399491943106e-07,
      "loss": 0.0004,
      "step": 170170
    },
    {
      "epoch": 2.784586435408656,
      "grad_norm": 0.00832452904433012,
      "learning_rate": 1.5623779776495874e-07,
      "loss": 0.0004,
      "step": 170180
    },
    {
      "epoch": 2.784750061359732,
      "grad_norm": 0.06192980334162712,
      "learning_rate": 1.5600177618522504e-07,
      "loss": 0.0011,
      "step": 170190
    },
    {
      "epoch": 2.7849136873108074,
      "grad_norm": 0.034070562571287155,
      "learning_rate": 1.5576593018878372e-07,
      "loss": 0.0008,
      "step": 170200
    },
    {
      "epoch": 2.7850773132618833,
      "grad_norm": 0.030460629612207413,
      "learning_rate": 1.5553025978418567e-07,
      "loss": 0.0006,
      "step": 170210
    },
    {
      "epoch": 2.785240939212959,
      "grad_norm": 0.01990927755832672,
      "learning_rate": 1.552947649799713e-07,
      "loss": 0.0008,
      "step": 170220
    },
    {
      "epoch": 2.785404565164035,
      "grad_norm": 0.014857985079288483,
      "learning_rate": 1.550594457846788e-07,
      "loss": 0.0005,
      "step": 170230
    },
    {
      "epoch": 2.785568191115111,
      "grad_norm": 0.013553816825151443,
      "learning_rate": 1.548243022068363e-07,
      "loss": 0.0009,
      "step": 170240
    },
    {
      "epoch": 2.7857318170661864,
      "grad_norm": 0.10838443785905838,
      "learning_rate": 1.545893342549676e-07,
      "loss": 0.0007,
      "step": 170250
    },
    {
      "epoch": 2.7858954430172624,
      "grad_norm": 0.01778852567076683,
      "learning_rate": 1.543545419375908e-07,
      "loss": 0.0009,
      "step": 170260
    },
    {
      "epoch": 2.7860590689683384,
      "grad_norm": 0.10566195100545883,
      "learning_rate": 1.5411992526321472e-07,
      "loss": 0.0007,
      "step": 170270
    },
    {
      "epoch": 2.786222694919414,
      "grad_norm": 0.034300994127988815,
      "learning_rate": 1.5388548424034476e-07,
      "loss": 0.0008,
      "step": 170280
    },
    {
      "epoch": 2.78638632087049,
      "grad_norm": 0.004707236774265766,
      "learning_rate": 1.5365121887747857e-07,
      "loss": 0.0007,
      "step": 170290
    },
    {
      "epoch": 2.786549946821566,
      "grad_norm": 0.012610087171196938,
      "learning_rate": 1.5341712918310824e-07,
      "loss": 0.0004,
      "step": 170300
    },
    {
      "epoch": 2.7867135727726415,
      "grad_norm": 0.07268116623163223,
      "learning_rate": 1.531832151657181e-07,
      "loss": 0.0016,
      "step": 170310
    },
    {
      "epoch": 2.7868771987237175,
      "grad_norm": 0.013467668555676937,
      "learning_rate": 1.5294947683378748e-07,
      "loss": 0.0009,
      "step": 170320
    },
    {
      "epoch": 2.7870408246747935,
      "grad_norm": 0.09897198528051376,
      "learning_rate": 1.5271591419578903e-07,
      "loss": 0.0006,
      "step": 170330
    },
    {
      "epoch": 2.787204450625869,
      "grad_norm": 0.16411009430885315,
      "learning_rate": 1.5248252726018876e-07,
      "loss": 0.0008,
      "step": 170340
    },
    {
      "epoch": 2.787368076576945,
      "grad_norm": 0.026736533269286156,
      "learning_rate": 1.52249316035446e-07,
      "loss": 0.001,
      "step": 170350
    },
    {
      "epoch": 2.787531702528021,
      "grad_norm": 0.042290136218070984,
      "learning_rate": 1.520162805300135e-07,
      "loss": 0.0006,
      "step": 170360
    },
    {
      "epoch": 2.7876953284790966,
      "grad_norm": 0.02710956707596779,
      "learning_rate": 1.5178342075233942e-07,
      "loss": 0.0015,
      "step": 170370
    },
    {
      "epoch": 2.7878589544301726,
      "grad_norm": 0.36796876788139343,
      "learning_rate": 1.5155073671086427e-07,
      "loss": 0.0013,
      "step": 170380
    },
    {
      "epoch": 2.7880225803812486,
      "grad_norm": 0.17700661718845367,
      "learning_rate": 1.5131822841402133e-07,
      "loss": 0.0008,
      "step": 170390
    },
    {
      "epoch": 2.788186206332324,
      "grad_norm": 0.17054308950901031,
      "learning_rate": 1.5108589587023993e-07,
      "loss": 0.0009,
      "step": 170400
    },
    {
      "epoch": 2.7883498322834,
      "grad_norm": 0.03665393590927124,
      "learning_rate": 1.5085373908793998e-07,
      "loss": 0.0004,
      "step": 170410
    },
    {
      "epoch": 2.788513458234476,
      "grad_norm": 0.06940967589616776,
      "learning_rate": 1.5062175807553814e-07,
      "loss": 0.0004,
      "step": 170420
    },
    {
      "epoch": 2.7886770841855517,
      "grad_norm": 0.02967650257050991,
      "learning_rate": 1.5038995284144208e-07,
      "loss": 0.0005,
      "step": 170430
    },
    {
      "epoch": 2.7888407101366277,
      "grad_norm": 0.12153536826372147,
      "learning_rate": 1.5015832339405457e-07,
      "loss": 0.0008,
      "step": 170440
    },
    {
      "epoch": 2.7890043360877037,
      "grad_norm": 0.03611701726913452,
      "learning_rate": 1.4992686974177162e-07,
      "loss": 0.0004,
      "step": 170450
    },
    {
      "epoch": 2.7891679620387793,
      "grad_norm": 0.07035545259714127,
      "learning_rate": 1.4969559189298265e-07,
      "loss": 0.0005,
      "step": 170460
    },
    {
      "epoch": 2.7893315879898553,
      "grad_norm": 0.1179262101650238,
      "learning_rate": 1.49464489856071e-07,
      "loss": 0.0009,
      "step": 170470
    },
    {
      "epoch": 2.7894952139409313,
      "grad_norm": 0.07383953034877777,
      "learning_rate": 1.4923356363941376e-07,
      "loss": 0.0004,
      "step": 170480
    },
    {
      "epoch": 2.789658839892007,
      "grad_norm": 0.03154398128390312,
      "learning_rate": 1.49002813251381e-07,
      "loss": 0.0006,
      "step": 170490
    },
    {
      "epoch": 2.789822465843083,
      "grad_norm": 0.07447236031293869,
      "learning_rate": 1.4877223870033763e-07,
      "loss": 0.0006,
      "step": 170500
    },
    {
      "epoch": 2.789986091794159,
      "grad_norm": 0.00967530906200409,
      "learning_rate": 1.4854183999464088e-07,
      "loss": 0.0006,
      "step": 170510
    },
    {
      "epoch": 2.7901497177452343,
      "grad_norm": 0.02058189921081066,
      "learning_rate": 1.4831161714264186e-07,
      "loss": 0.0009,
      "step": 170520
    },
    {
      "epoch": 2.7903133436963103,
      "grad_norm": 0.006499041803181171,
      "learning_rate": 1.480815701526861e-07,
      "loss": 0.0011,
      "step": 170530
    },
    {
      "epoch": 2.7904769696473863,
      "grad_norm": 0.03343840315937996,
      "learning_rate": 1.4785169903311302e-07,
      "loss": 0.0014,
      "step": 170540
    },
    {
      "epoch": 2.790640595598462,
      "grad_norm": 0.05925329402089119,
      "learning_rate": 1.4762200379225322e-07,
      "loss": 0.0005,
      "step": 170550
    },
    {
      "epoch": 2.790804221549538,
      "grad_norm": 0.09993096441030502,
      "learning_rate": 1.473924844384339e-07,
      "loss": 0.0014,
      "step": 170560
    },
    {
      "epoch": 2.7909678475006134,
      "grad_norm": 0.039635904133319855,
      "learning_rate": 1.4716314097997343e-07,
      "loss": 0.0017,
      "step": 170570
    },
    {
      "epoch": 2.7911314734516894,
      "grad_norm": 0.15017932653427124,
      "learning_rate": 1.4693397342518623e-07,
      "loss": 0.0008,
      "step": 170580
    },
    {
      "epoch": 2.7912950994027654,
      "grad_norm": 0.16104936599731445,
      "learning_rate": 1.4670498178237736e-07,
      "loss": 0.0006,
      "step": 170590
    },
    {
      "epoch": 2.791458725353841,
      "grad_norm": 0.0601748451590538,
      "learning_rate": 1.4647616605984903e-07,
      "loss": 0.0004,
      "step": 170600
    },
    {
      "epoch": 2.791622351304917,
      "grad_norm": 0.0385313406586647,
      "learning_rate": 1.4624752626589466e-07,
      "loss": 0.0005,
      "step": 170610
    },
    {
      "epoch": 2.7917859772559925,
      "grad_norm": 0.00401941267773509,
      "learning_rate": 1.460190624088015e-07,
      "loss": 0.0003,
      "step": 170620
    },
    {
      "epoch": 2.7919496032070685,
      "grad_norm": 0.019108308479189873,
      "learning_rate": 1.4579077449685175e-07,
      "loss": 0.0006,
      "step": 170630
    },
    {
      "epoch": 2.7921132291581445,
      "grad_norm": 0.014254197478294373,
      "learning_rate": 1.455626625383194e-07,
      "loss": 0.0014,
      "step": 170640
    },
    {
      "epoch": 2.79227685510922,
      "grad_norm": 0.06113189831376076,
      "learning_rate": 1.4533472654147284e-07,
      "loss": 0.0007,
      "step": 170650
    },
    {
      "epoch": 2.792440481060296,
      "grad_norm": 0.10673297196626663,
      "learning_rate": 1.451069665145749e-07,
      "loss": 0.0012,
      "step": 170660
    },
    {
      "epoch": 2.792604107011372,
      "grad_norm": 0.10668063908815384,
      "learning_rate": 1.4487938246588063e-07,
      "loss": 0.0004,
      "step": 170670
    },
    {
      "epoch": 2.7927677329624476,
      "grad_norm": 0.15307560563087463,
      "learning_rate": 1.446519744036401e-07,
      "loss": 0.0008,
      "step": 170680
    },
    {
      "epoch": 2.7929313589135236,
      "grad_norm": 0.06924367696046829,
      "learning_rate": 1.444247423360956e-07,
      "loss": 0.0008,
      "step": 170690
    },
    {
      "epoch": 2.7930949848645996,
      "grad_norm": 0.033333614468574524,
      "learning_rate": 1.4419768627148444e-07,
      "loss": 0.0007,
      "step": 170700
    },
    {
      "epoch": 2.793258610815675,
      "grad_norm": 0.07976774126291275,
      "learning_rate": 1.4397080621803615e-07,
      "loss": 0.0006,
      "step": 170710
    },
    {
      "epoch": 2.793422236766751,
      "grad_norm": 0.025536082684993744,
      "learning_rate": 1.4374410218397528e-07,
      "loss": 0.0009,
      "step": 170720
    },
    {
      "epoch": 2.793585862717827,
      "grad_norm": 0.07201147824525833,
      "learning_rate": 1.435175741775191e-07,
      "loss": 0.0003,
      "step": 170730
    },
    {
      "epoch": 2.7937494886689027,
      "grad_norm": 0.045723602175712585,
      "learning_rate": 1.4329122220687884e-07,
      "loss": 0.0008,
      "step": 170740
    },
    {
      "epoch": 2.7939131146199787,
      "grad_norm": 0.01806153543293476,
      "learning_rate": 1.4306504628025853e-07,
      "loss": 0.0006,
      "step": 170750
    },
    {
      "epoch": 2.7940767405710547,
      "grad_norm": 0.09223387390375137,
      "learning_rate": 1.4283904640585711e-07,
      "loss": 0.0013,
      "step": 170760
    },
    {
      "epoch": 2.7942403665221303,
      "grad_norm": 0.018166862428188324,
      "learning_rate": 1.4261322259186638e-07,
      "loss": 0.0006,
      "step": 170770
    },
    {
      "epoch": 2.7944039924732063,
      "grad_norm": 0.13295575976371765,
      "learning_rate": 1.4238757484647148e-07,
      "loss": 0.0007,
      "step": 170780
    },
    {
      "epoch": 2.7945676184242823,
      "grad_norm": 0.022350525483489037,
      "learning_rate": 1.4216210317785307e-07,
      "loss": 0.0007,
      "step": 170790
    },
    {
      "epoch": 2.794731244375358,
      "grad_norm": 0.022321805357933044,
      "learning_rate": 1.4193680759418183e-07,
      "loss": 0.0006,
      "step": 170800
    },
    {
      "epoch": 2.794894870326434,
      "grad_norm": 0.02440972998738289,
      "learning_rate": 1.417116881036251e-07,
      "loss": 0.0007,
      "step": 170810
    },
    {
      "epoch": 2.79505849627751,
      "grad_norm": 0.02165321446955204,
      "learning_rate": 1.414867447143442e-07,
      "loss": 0.0004,
      "step": 170820
    },
    {
      "epoch": 2.7952221222285853,
      "grad_norm": 0.0063326614908874035,
      "learning_rate": 1.4126197743449088e-07,
      "loss": 0.0008,
      "step": 170830
    },
    {
      "epoch": 2.7953857481796613,
      "grad_norm": 0.07477544993162155,
      "learning_rate": 1.410373862722142e-07,
      "loss": 0.0006,
      "step": 170840
    },
    {
      "epoch": 2.7955493741307373,
      "grad_norm": 0.09436183422803879,
      "learning_rate": 1.4081297123565317e-07,
      "loss": 0.0009,
      "step": 170850
    },
    {
      "epoch": 2.795713000081813,
      "grad_norm": 0.02822510525584221,
      "learning_rate": 1.4058873233294355e-07,
      "loss": 0.0011,
      "step": 170860
    },
    {
      "epoch": 2.795876626032889,
      "grad_norm": 0.04970188066363335,
      "learning_rate": 1.4036466957221273e-07,
      "loss": 0.0007,
      "step": 170870
    },
    {
      "epoch": 2.796040251983965,
      "grad_norm": 0.06209738925099373,
      "learning_rate": 1.4014078296158252e-07,
      "loss": 0.0007,
      "step": 170880
    },
    {
      "epoch": 2.7962038779350404,
      "grad_norm": 0.14559872448444366,
      "learning_rate": 1.399170725091692e-07,
      "loss": 0.0007,
      "step": 170890
    },
    {
      "epoch": 2.7963675038861164,
      "grad_norm": 0.07818220555782318,
      "learning_rate": 1.3969353822308074e-07,
      "loss": 0.0009,
      "step": 170900
    },
    {
      "epoch": 2.7965311298371924,
      "grad_norm": 0.07494895905256271,
      "learning_rate": 1.3947018011142065e-07,
      "loss": 0.0009,
      "step": 170910
    },
    {
      "epoch": 2.796694755788268,
      "grad_norm": 0.02316666580736637,
      "learning_rate": 1.392469981822836e-07,
      "loss": 0.0007,
      "step": 170920
    },
    {
      "epoch": 2.796858381739344,
      "grad_norm": 0.002424146980047226,
      "learning_rate": 1.3902399244376142e-07,
      "loss": 0.0004,
      "step": 170930
    },
    {
      "epoch": 2.79702200769042,
      "grad_norm": 0.022954905405640602,
      "learning_rate": 1.3880116290393596e-07,
      "loss": 0.0003,
      "step": 170940
    },
    {
      "epoch": 2.7971856336414955,
      "grad_norm": 0.17274229228496552,
      "learning_rate": 1.385785095708847e-07,
      "loss": 0.0007,
      "step": 170950
    },
    {
      "epoch": 2.7973492595925715,
      "grad_norm": 0.021951640024781227,
      "learning_rate": 1.3835603245267893e-07,
      "loss": 0.0007,
      "step": 170960
    },
    {
      "epoch": 2.797512885543647,
      "grad_norm": 0.011379717849195004,
      "learning_rate": 1.3813373155738107e-07,
      "loss": 0.0013,
      "step": 170970
    },
    {
      "epoch": 2.797676511494723,
      "grad_norm": 0.09212465584278107,
      "learning_rate": 1.379116068930514e-07,
      "loss": 0.0007,
      "step": 170980
    },
    {
      "epoch": 2.797840137445799,
      "grad_norm": 0.07407135516405106,
      "learning_rate": 1.3768965846773952e-07,
      "loss": 0.0011,
      "step": 170990
    },
    {
      "epoch": 2.7980037633968746,
      "grad_norm": 0.007346815429627895,
      "learning_rate": 1.374678862894918e-07,
      "loss": 0.0004,
      "step": 171000
    },
    {
      "epoch": 2.7981673893479506,
      "grad_norm": 0.02472073957324028,
      "learning_rate": 1.3724629036634628e-07,
      "loss": 0.001,
      "step": 171010
    },
    {
      "epoch": 2.798331015299026,
      "grad_norm": 0.07992814481258392,
      "learning_rate": 1.370248707063354e-07,
      "loss": 0.001,
      "step": 171020
    },
    {
      "epoch": 2.798494641250102,
      "grad_norm": 0.0026793377473950386,
      "learning_rate": 1.3680362731748498e-07,
      "loss": 0.0005,
      "step": 171030
    },
    {
      "epoch": 2.798658267201178,
      "grad_norm": 0.008786062709987164,
      "learning_rate": 1.3658256020781468e-07,
      "loss": 0.0007,
      "step": 171040
    },
    {
      "epoch": 2.7988218931522537,
      "grad_norm": 0.03457338735461235,
      "learning_rate": 1.3636166938533814e-07,
      "loss": 0.0006,
      "step": 171050
    },
    {
      "epoch": 2.7989855191033297,
      "grad_norm": 0.02264559082686901,
      "learning_rate": 1.3614095485806112e-07,
      "loss": 0.0005,
      "step": 171060
    },
    {
      "epoch": 2.7991491450544057,
      "grad_norm": 0.09252757579088211,
      "learning_rate": 1.3592041663398504e-07,
      "loss": 0.0006,
      "step": 171070
    },
    {
      "epoch": 2.7993127710054813,
      "grad_norm": 0.05027490481734276,
      "learning_rate": 1.3570005472110238e-07,
      "loss": 0.0004,
      "step": 171080
    },
    {
      "epoch": 2.7994763969565573,
      "grad_norm": 0.10141106694936752,
      "learning_rate": 1.354798691274023e-07,
      "loss": 0.0007,
      "step": 171090
    },
    {
      "epoch": 2.7996400229076333,
      "grad_norm": 0.014897673390805721,
      "learning_rate": 1.352598598608651e-07,
      "loss": 0.0019,
      "step": 171100
    },
    {
      "epoch": 2.799803648858709,
      "grad_norm": 0.09295172244310379,
      "learning_rate": 1.3504002692946605e-07,
      "loss": 0.0011,
      "step": 171110
    },
    {
      "epoch": 2.799967274809785,
      "grad_norm": 0.22135546803474426,
      "learning_rate": 1.3482037034117378e-07,
      "loss": 0.0014,
      "step": 171120
    },
    {
      "epoch": 2.800130900760861,
      "grad_norm": 0.12304525077342987,
      "learning_rate": 1.3460089010394915e-07,
      "loss": 0.0007,
      "step": 171130
    },
    {
      "epoch": 2.8002945267119363,
      "grad_norm": 0.10921672731637955,
      "learning_rate": 1.343815862257497e-07,
      "loss": 0.001,
      "step": 171140
    },
    {
      "epoch": 2.8004581526630123,
      "grad_norm": 0.014127643778920174,
      "learning_rate": 1.3416245871452293e-07,
      "loss": 0.0005,
      "step": 171150
    },
    {
      "epoch": 2.8006217786140883,
      "grad_norm": 0.05219822749495506,
      "learning_rate": 1.3394350757821196e-07,
      "loss": 0.0006,
      "step": 171160
    },
    {
      "epoch": 2.800785404565164,
      "grad_norm": 0.12080032378435135,
      "learning_rate": 1.3372473282475428e-07,
      "loss": 0.0008,
      "step": 171170
    },
    {
      "epoch": 2.80094903051624,
      "grad_norm": 0.07122016698122025,
      "learning_rate": 1.3350613446207862e-07,
      "loss": 0.0005,
      "step": 171180
    },
    {
      "epoch": 2.801112656467316,
      "grad_norm": 0.08172375708818436,
      "learning_rate": 1.3328771249810967e-07,
      "loss": 0.0004,
      "step": 171190
    },
    {
      "epoch": 2.8012762824183914,
      "grad_norm": 0.04016002267599106,
      "learning_rate": 1.3306946694076395e-07,
      "loss": 0.0004,
      "step": 171200
    },
    {
      "epoch": 2.8014399083694674,
      "grad_norm": 0.12552621960639954,
      "learning_rate": 1.3285139779795285e-07,
      "loss": 0.0005,
      "step": 171210
    },
    {
      "epoch": 2.8016035343205434,
      "grad_norm": 0.054361145943403244,
      "learning_rate": 1.3263350507758065e-07,
      "loss": 0.0007,
      "step": 171220
    },
    {
      "epoch": 2.801767160271619,
      "grad_norm": 0.18638886511325836,
      "learning_rate": 1.3241578878754602e-07,
      "loss": 0.0009,
      "step": 171230
    },
    {
      "epoch": 2.801930786222695,
      "grad_norm": 0.06557103991508484,
      "learning_rate": 1.3219824893573985e-07,
      "loss": 0.0002,
      "step": 171240
    },
    {
      "epoch": 2.802094412173771,
      "grad_norm": 0.04394310712814331,
      "learning_rate": 1.3198088553004752e-07,
      "loss": 0.0007,
      "step": 171250
    },
    {
      "epoch": 2.8022580381248465,
      "grad_norm": 0.03255559504032135,
      "learning_rate": 1.3176369857834825e-07,
      "loss": 0.0004,
      "step": 171260
    },
    {
      "epoch": 2.8024216640759225,
      "grad_norm": 0.10194858908653259,
      "learning_rate": 1.315466880885141e-07,
      "loss": 0.0007,
      "step": 171270
    },
    {
      "epoch": 2.8025852900269985,
      "grad_norm": 0.015548547729849815,
      "learning_rate": 1.313298540684116e-07,
      "loss": 0.0007,
      "step": 171280
    },
    {
      "epoch": 2.802748915978074,
      "grad_norm": 0.2654331624507904,
      "learning_rate": 1.3111319652590048e-07,
      "loss": 0.0007,
      "step": 171290
    },
    {
      "epoch": 2.80291254192915,
      "grad_norm": 0.007964244112372398,
      "learning_rate": 1.3089671546883343e-07,
      "loss": 0.0007,
      "step": 171300
    },
    {
      "epoch": 2.803076167880226,
      "grad_norm": 0.037779804319143295,
      "learning_rate": 1.3068041090505858e-07,
      "loss": 0.0006,
      "step": 171310
    },
    {
      "epoch": 2.8032397938313016,
      "grad_norm": 0.05188390240073204,
      "learning_rate": 1.304642828424152e-07,
      "loss": 0.0005,
      "step": 171320
    },
    {
      "epoch": 2.8034034197823776,
      "grad_norm": 0.037394631654024124,
      "learning_rate": 1.3024833128873815e-07,
      "loss": 0.001,
      "step": 171330
    },
    {
      "epoch": 2.803567045733453,
      "grad_norm": 0.09160242974758148,
      "learning_rate": 1.3003255625185563e-07,
      "loss": 0.0007,
      "step": 171340
    },
    {
      "epoch": 2.803730671684529,
      "grad_norm": 0.08693581819534302,
      "learning_rate": 1.2981695773958746e-07,
      "loss": 0.0005,
      "step": 171350
    },
    {
      "epoch": 2.803894297635605,
      "grad_norm": 0.20058026909828186,
      "learning_rate": 1.2960153575974965e-07,
      "loss": 0.0006,
      "step": 171360
    },
    {
      "epoch": 2.8040579235866807,
      "grad_norm": 0.07765739411115646,
      "learning_rate": 1.2938629032014982e-07,
      "loss": 0.0006,
      "step": 171370
    },
    {
      "epoch": 2.8042215495377567,
      "grad_norm": 0.1717786341905594,
      "learning_rate": 1.2917122142859173e-07,
      "loss": 0.0006,
      "step": 171380
    },
    {
      "epoch": 2.8043851754888323,
      "grad_norm": 0.003905646735802293,
      "learning_rate": 1.289563290928697e-07,
      "loss": 0.0015,
      "step": 171390
    },
    {
      "epoch": 2.8045488014399083,
      "grad_norm": 0.027350837364792824,
      "learning_rate": 1.2874161332077363e-07,
      "loss": 0.0004,
      "step": 171400
    },
    {
      "epoch": 2.8047124273909843,
      "grad_norm": 0.07118779420852661,
      "learning_rate": 1.2852707412008614e-07,
      "loss": 0.0009,
      "step": 171410
    },
    {
      "epoch": 2.80487605334206,
      "grad_norm": 0.07049226760864258,
      "learning_rate": 1.2831271149858382e-07,
      "loss": 0.0006,
      "step": 171420
    },
    {
      "epoch": 2.805039679293136,
      "grad_norm": 0.0034819552674889565,
      "learning_rate": 1.280985254640371e-07,
      "loss": 0.0005,
      "step": 171430
    },
    {
      "epoch": 2.805203305244212,
      "grad_norm": 0.1797843724489212,
      "learning_rate": 1.2788451602420982e-07,
      "loss": 0.0013,
      "step": 171440
    },
    {
      "epoch": 2.8053669311952873,
      "grad_norm": 0.1690874993801117,
      "learning_rate": 1.2767068318685904e-07,
      "loss": 0.0008,
      "step": 171450
    },
    {
      "epoch": 2.8055305571463633,
      "grad_norm": 0.008725961670279503,
      "learning_rate": 1.2745702695973528e-07,
      "loss": 0.0003,
      "step": 171460
    },
    {
      "epoch": 2.8056941830974393,
      "grad_norm": 0.027935713529586792,
      "learning_rate": 1.272435473505834e-07,
      "loss": 0.0007,
      "step": 171470
    },
    {
      "epoch": 2.805857809048515,
      "grad_norm": 0.13406161963939667,
      "learning_rate": 1.2703024436714172e-07,
      "loss": 0.0006,
      "step": 171480
    },
    {
      "epoch": 2.806021434999591,
      "grad_norm": 0.11690158396959305,
      "learning_rate": 1.268171180171418e-07,
      "loss": 0.0004,
      "step": 171490
    },
    {
      "epoch": 2.806185060950667,
      "grad_norm": 0.0568169429898262,
      "learning_rate": 1.2660416830830857e-07,
      "loss": 0.0006,
      "step": 171500
    },
    {
      "epoch": 2.8063486869017424,
      "grad_norm": 0.025959106162190437,
      "learning_rate": 1.2639139524836196e-07,
      "loss": 0.0004,
      "step": 171510
    },
    {
      "epoch": 2.8065123128528184,
      "grad_norm": 0.14221347868442535,
      "learning_rate": 1.261787988450136e-07,
      "loss": 0.0008,
      "step": 171520
    },
    {
      "epoch": 2.8066759388038944,
      "grad_norm": 0.07157721370458603,
      "learning_rate": 1.2596637910596954e-07,
      "loss": 0.0004,
      "step": 171530
    },
    {
      "epoch": 2.80683956475497,
      "grad_norm": 0.09572681784629822,
      "learning_rate": 1.2575413603893082e-07,
      "loss": 0.0003,
      "step": 171540
    },
    {
      "epoch": 2.807003190706046,
      "grad_norm": 0.0512634739279747,
      "learning_rate": 1.2554206965158855e-07,
      "loss": 0.0006,
      "step": 171550
    },
    {
      "epoch": 2.807166816657122,
      "grad_norm": 0.15759795904159546,
      "learning_rate": 1.2533017995163155e-07,
      "loss": 0.0006,
      "step": 171560
    },
    {
      "epoch": 2.8073304426081975,
      "grad_norm": 0.013139360584318638,
      "learning_rate": 1.2511846694673923e-07,
      "loss": 0.0007,
      "step": 171570
    },
    {
      "epoch": 2.8074940685592735,
      "grad_norm": 0.04685650393366814,
      "learning_rate": 1.2490693064458604e-07,
      "loss": 0.0008,
      "step": 171580
    },
    {
      "epoch": 2.8076576945103495,
      "grad_norm": 0.31251591444015503,
      "learning_rate": 1.246955710528397e-07,
      "loss": 0.0009,
      "step": 171590
    },
    {
      "epoch": 2.807821320461425,
      "grad_norm": 0.07122431695461273,
      "learning_rate": 1.244843881791613e-07,
      "loss": 0.0003,
      "step": 171600
    },
    {
      "epoch": 2.807984946412501,
      "grad_norm": 0.03874616324901581,
      "learning_rate": 1.2427338203120641e-07,
      "loss": 0.0009,
      "step": 171610
    },
    {
      "epoch": 2.808148572363577,
      "grad_norm": 0.09619405120611191,
      "learning_rate": 1.240625526166217e-07,
      "loss": 0.0006,
      "step": 171620
    },
    {
      "epoch": 2.8083121983146526,
      "grad_norm": 0.03502504900097847,
      "learning_rate": 1.238518999430516e-07,
      "loss": 0.0005,
      "step": 171630
    },
    {
      "epoch": 2.8084758242657286,
      "grad_norm": 0.015436643734574318,
      "learning_rate": 1.2364142401813052e-07,
      "loss": 0.001,
      "step": 171640
    },
    {
      "epoch": 2.8086394502168046,
      "grad_norm": 0.015409058891236782,
      "learning_rate": 1.2343112484948737e-07,
      "loss": 0.0008,
      "step": 171650
    },
    {
      "epoch": 2.80880307616788,
      "grad_norm": 0.37831759452819824,
      "learning_rate": 1.2322100244474555e-07,
      "loss": 0.0012,
      "step": 171660
    },
    {
      "epoch": 2.808966702118956,
      "grad_norm": 0.1155666708946228,
      "learning_rate": 1.2301105681152115e-07,
      "loss": 0.0007,
      "step": 171670
    },
    {
      "epoch": 2.809130328070032,
      "grad_norm": 0.0317828431725502,
      "learning_rate": 1.228012879574253e-07,
      "loss": 0.0011,
      "step": 171680
    },
    {
      "epoch": 2.8092939540211077,
      "grad_norm": 0.04493634030222893,
      "learning_rate": 1.225916958900597e-07,
      "loss": 0.0004,
      "step": 171690
    },
    {
      "epoch": 2.8094575799721837,
      "grad_norm": 0.03831320255994797,
      "learning_rate": 1.2238228061702274e-07,
      "loss": 0.0007,
      "step": 171700
    },
    {
      "epoch": 2.8096212059232597,
      "grad_norm": 0.02901962399482727,
      "learning_rate": 1.2217304214590498e-07,
      "loss": 0.0005,
      "step": 171710
    },
    {
      "epoch": 2.8097848318743353,
      "grad_norm": 0.052062779664993286,
      "learning_rate": 1.2196398048429148e-07,
      "loss": 0.0009,
      "step": 171720
    },
    {
      "epoch": 2.8099484578254112,
      "grad_norm": 0.07571718841791153,
      "learning_rate": 1.217550956397595e-07,
      "loss": 0.0008,
      "step": 171730
    },
    {
      "epoch": 2.810112083776487,
      "grad_norm": 0.04107063263654709,
      "learning_rate": 1.215463876198808e-07,
      "loss": 0.0004,
      "step": 171740
    },
    {
      "epoch": 2.810275709727563,
      "grad_norm": 0.1320413500070572,
      "learning_rate": 1.213378564322204e-07,
      "loss": 0.0008,
      "step": 171750
    },
    {
      "epoch": 2.810439335678639,
      "grad_norm": 0.0882088765501976,
      "learning_rate": 1.211295020843367e-07,
      "loss": 0.0006,
      "step": 171760
    },
    {
      "epoch": 2.8106029616297143,
      "grad_norm": 0.06519163399934769,
      "learning_rate": 1.2092132458378314e-07,
      "loss": 0.0005,
      "step": 171770
    },
    {
      "epoch": 2.8107665875807903,
      "grad_norm": 0.13670633733272552,
      "learning_rate": 1.207133239381042e-07,
      "loss": 0.0008,
      "step": 171780
    },
    {
      "epoch": 2.810930213531866,
      "grad_norm": 0.12694181501865387,
      "learning_rate": 1.2050550015484108e-07,
      "loss": 0.0011,
      "step": 171790
    },
    {
      "epoch": 2.811093839482942,
      "grad_norm": 0.08233991265296936,
      "learning_rate": 1.20297853241525e-07,
      "loss": 0.0007,
      "step": 171800
    },
    {
      "epoch": 2.811257465434018,
      "grad_norm": 0.06695646047592163,
      "learning_rate": 1.2009038320568435e-07,
      "loss": 0.0004,
      "step": 171810
    },
    {
      "epoch": 2.8114210913850934,
      "grad_norm": 0.07969870418310165,
      "learning_rate": 1.1988309005483866e-07,
      "loss": 0.0008,
      "step": 171820
    },
    {
      "epoch": 2.8115847173361694,
      "grad_norm": 0.15012790262699127,
      "learning_rate": 1.1967597379650198e-07,
      "loss": 0.0009,
      "step": 171830
    },
    {
      "epoch": 2.8117483432872454,
      "grad_norm": 0.0005089350161142647,
      "learning_rate": 1.1946903443818215e-07,
      "loss": 0.0006,
      "step": 171840
    },
    {
      "epoch": 2.811911969238321,
      "grad_norm": 0.05697668343782425,
      "learning_rate": 1.1926227198737927e-07,
      "loss": 0.0008,
      "step": 171850
    },
    {
      "epoch": 2.812075595189397,
      "grad_norm": 0.1401207596063614,
      "learning_rate": 1.190556864515885e-07,
      "loss": 0.0004,
      "step": 171860
    },
    {
      "epoch": 2.812239221140473,
      "grad_norm": 0.05562201514840126,
      "learning_rate": 1.1884927783829824e-07,
      "loss": 0.0009,
      "step": 171870
    },
    {
      "epoch": 2.8124028470915485,
      "grad_norm": 0.11965223401784897,
      "learning_rate": 1.1864304615498978e-07,
      "loss": 0.0008,
      "step": 171880
    },
    {
      "epoch": 2.8125664730426245,
      "grad_norm": 0.05904148146510124,
      "learning_rate": 1.1843699140913878e-07,
      "loss": 0.0004,
      "step": 171890
    },
    {
      "epoch": 2.8127300989937005,
      "grad_norm": 0.019149212166666985,
      "learning_rate": 1.1823111360821482e-07,
      "loss": 0.0013,
      "step": 171900
    },
    {
      "epoch": 2.812893724944776,
      "grad_norm": 0.0012291534803807735,
      "learning_rate": 1.1802541275967971e-07,
      "loss": 0.0004,
      "step": 171910
    },
    {
      "epoch": 2.813057350895852,
      "grad_norm": 0.002934809308499098,
      "learning_rate": 1.1781988887098972e-07,
      "loss": 0.0008,
      "step": 171920
    },
    {
      "epoch": 2.813220976846928,
      "grad_norm": 0.11563082784414291,
      "learning_rate": 1.1761454194959498e-07,
      "loss": 0.0006,
      "step": 171930
    },
    {
      "epoch": 2.8133846027980036,
      "grad_norm": 0.10422677546739578,
      "learning_rate": 1.1740937200293901e-07,
      "loss": 0.0008,
      "step": 171940
    },
    {
      "epoch": 2.8135482287490796,
      "grad_norm": 0.06244797632098198,
      "learning_rate": 1.172043790384575e-07,
      "loss": 0.0006,
      "step": 171950
    },
    {
      "epoch": 2.8137118547001556,
      "grad_norm": 0.15237131714820862,
      "learning_rate": 1.169995630635823e-07,
      "loss": 0.0007,
      "step": 171960
    },
    {
      "epoch": 2.813875480651231,
      "grad_norm": 0.06148587167263031,
      "learning_rate": 1.1679492408573634e-07,
      "loss": 0.0007,
      "step": 171970
    },
    {
      "epoch": 2.814039106602307,
      "grad_norm": 0.003219089936465025,
      "learning_rate": 1.165904621123387e-07,
      "loss": 0.0007,
      "step": 171980
    },
    {
      "epoch": 2.814202732553383,
      "grad_norm": 0.10074108839035034,
      "learning_rate": 1.1638617715079902e-07,
      "loss": 0.0005,
      "step": 171990
    },
    {
      "epoch": 2.8143663585044587,
      "grad_norm": 0.08588927984237671,
      "learning_rate": 1.1618206920852359e-07,
      "loss": 0.0008,
      "step": 172000
    },
    {
      "epoch": 2.8145299844555347,
      "grad_norm": 0.05853542312979698,
      "learning_rate": 1.1597813829290982e-07,
      "loss": 0.0015,
      "step": 172010
    },
    {
      "epoch": 2.8146936104066107,
      "grad_norm": 0.02257656306028366,
      "learning_rate": 1.1577438441135013e-07,
      "loss": 0.0003,
      "step": 172020
    },
    {
      "epoch": 2.8148572363576863,
      "grad_norm": 0.10484442859888077,
      "learning_rate": 1.155708075712303e-07,
      "loss": 0.0006,
      "step": 172030
    },
    {
      "epoch": 2.8150208623087623,
      "grad_norm": 0.05636831372976303,
      "learning_rate": 1.1536740777992938e-07,
      "loss": 0.0004,
      "step": 172040
    },
    {
      "epoch": 2.8151844882598382,
      "grad_norm": 0.2545546591281891,
      "learning_rate": 1.1516418504482041e-07,
      "loss": 0.001,
      "step": 172050
    },
    {
      "epoch": 2.815348114210914,
      "grad_norm": 0.08680564910173416,
      "learning_rate": 1.1496113937326858e-07,
      "loss": 0.0007,
      "step": 172060
    },
    {
      "epoch": 2.81551174016199,
      "grad_norm": 0.0426112599670887,
      "learning_rate": 1.1475827077263468e-07,
      "loss": 0.0024,
      "step": 172070
    },
    {
      "epoch": 2.815675366113066,
      "grad_norm": 0.00658803666010499,
      "learning_rate": 1.1455557925027172e-07,
      "loss": 0.0004,
      "step": 172080
    },
    {
      "epoch": 2.8158389920641413,
      "grad_norm": 0.07226915657520294,
      "learning_rate": 1.143530648135277e-07,
      "loss": 0.0007,
      "step": 172090
    },
    {
      "epoch": 2.8160026180152173,
      "grad_norm": 0.026515459641814232,
      "learning_rate": 1.1415072746974288e-07,
      "loss": 0.0004,
      "step": 172100
    },
    {
      "epoch": 2.816166243966293,
      "grad_norm": 0.11451422423124313,
      "learning_rate": 1.139485672262508e-07,
      "loss": 0.0007,
      "step": 172110
    },
    {
      "epoch": 2.816329869917369,
      "grad_norm": 0.06423421949148178,
      "learning_rate": 1.1374658409038064e-07,
      "loss": 0.0011,
      "step": 172120
    },
    {
      "epoch": 2.816493495868445,
      "grad_norm": 0.03168177977204323,
      "learning_rate": 1.1354477806945207e-07,
      "loss": 0.0007,
      "step": 172130
    },
    {
      "epoch": 2.8166571218195204,
      "grad_norm": 0.065270334482193,
      "learning_rate": 1.1334314917078149e-07,
      "loss": 0.0016,
      "step": 172140
    },
    {
      "epoch": 2.8168207477705964,
      "grad_norm": 0.061595842242240906,
      "learning_rate": 1.1314169740167747e-07,
      "loss": 0.0009,
      "step": 172150
    },
    {
      "epoch": 2.816984373721672,
      "grad_norm": 0.04262515902519226,
      "learning_rate": 1.1294042276944084e-07,
      "loss": 0.0007,
      "step": 172160
    },
    {
      "epoch": 2.817147999672748,
      "grad_norm": 0.08238794654607773,
      "learning_rate": 1.1273932528136854e-07,
      "loss": 0.0005,
      "step": 172170
    },
    {
      "epoch": 2.817311625623824,
      "grad_norm": 0.0027544735930860043,
      "learning_rate": 1.1253840494474922e-07,
      "loss": 0.0003,
      "step": 172180
    },
    {
      "epoch": 2.8174752515748995,
      "grad_norm": 0.2192983329296112,
      "learning_rate": 1.1233766176686645e-07,
      "loss": 0.0007,
      "step": 172190
    },
    {
      "epoch": 2.8176388775259755,
      "grad_norm": 0.0644710510969162,
      "learning_rate": 1.1213709575499555e-07,
      "loss": 0.0005,
      "step": 172200
    },
    {
      "epoch": 2.8178025034770515,
      "grad_norm": 0.020667342469096184,
      "learning_rate": 1.119367069164079e-07,
      "loss": 0.0005,
      "step": 172210
    },
    {
      "epoch": 2.817966129428127,
      "grad_norm": 0.4020586609840393,
      "learning_rate": 1.1173649525836605e-07,
      "loss": 0.001,
      "step": 172220
    },
    {
      "epoch": 2.818129755379203,
      "grad_norm": 0.05030730366706848,
      "learning_rate": 1.1153646078812808e-07,
      "loss": 0.0009,
      "step": 172230
    },
    {
      "epoch": 2.818293381330279,
      "grad_norm": 0.022579913958907127,
      "learning_rate": 1.1133660351294428e-07,
      "loss": 0.0005,
      "step": 172240
    },
    {
      "epoch": 2.8184570072813546,
      "grad_norm": 0.028644638136029243,
      "learning_rate": 1.1113692344005889e-07,
      "loss": 0.0006,
      "step": 172250
    },
    {
      "epoch": 2.8186206332324306,
      "grad_norm": 0.0095040462911129,
      "learning_rate": 1.1093742057670998e-07,
      "loss": 0.0006,
      "step": 172260
    },
    {
      "epoch": 2.8187842591835066,
      "grad_norm": 0.10112432390451431,
      "learning_rate": 1.1073809493012843e-07,
      "loss": 0.0007,
      "step": 172270
    },
    {
      "epoch": 2.818947885134582,
      "grad_norm": 0.0946236327290535,
      "learning_rate": 1.1053894650754071e-07,
      "loss": 0.0005,
      "step": 172280
    },
    {
      "epoch": 2.819111511085658,
      "grad_norm": 0.02809935249388218,
      "learning_rate": 1.1033997531616436e-07,
      "loss": 0.0008,
      "step": 172290
    },
    {
      "epoch": 2.819275137036734,
      "grad_norm": 0.04630189388990402,
      "learning_rate": 1.101411813632114e-07,
      "loss": 0.0006,
      "step": 172300
    },
    {
      "epoch": 2.8194387629878097,
      "grad_norm": 0.0056246803142130375,
      "learning_rate": 1.0994256465588882e-07,
      "loss": 0.0004,
      "step": 172310
    },
    {
      "epoch": 2.8196023889388857,
      "grad_norm": 0.09475484490394592,
      "learning_rate": 1.0974412520139533e-07,
      "loss": 0.0006,
      "step": 172320
    },
    {
      "epoch": 2.8197660148899617,
      "grad_norm": 0.039400387555360794,
      "learning_rate": 1.0954586300692349e-07,
      "loss": 0.0004,
      "step": 172330
    },
    {
      "epoch": 2.8199296408410373,
      "grad_norm": 0.11696458607912064,
      "learning_rate": 1.0934777807966035e-07,
      "loss": 0.0005,
      "step": 172340
    },
    {
      "epoch": 2.8200932667921133,
      "grad_norm": 0.3796849548816681,
      "learning_rate": 1.0914987042678626e-07,
      "loss": 0.0016,
      "step": 172350
    },
    {
      "epoch": 2.8202568927431892,
      "grad_norm": 0.08543279021978378,
      "learning_rate": 1.0895214005547438e-07,
      "loss": 0.001,
      "step": 172360
    },
    {
      "epoch": 2.820420518694265,
      "grad_norm": 0.2826046645641327,
      "learning_rate": 1.0875458697289176e-07,
      "loss": 0.001,
      "step": 172370
    },
    {
      "epoch": 2.820584144645341,
      "grad_norm": 0.07091300189495087,
      "learning_rate": 1.0855721118619989e-07,
      "loss": 0.0009,
      "step": 172380
    },
    {
      "epoch": 2.820747770596417,
      "grad_norm": 0.026521019637584686,
      "learning_rate": 1.0836001270255248e-07,
      "loss": 0.0008,
      "step": 172390
    },
    {
      "epoch": 2.8209113965474923,
      "grad_norm": 0.11292688548564911,
      "learning_rate": 1.0816299152909826e-07,
      "loss": 0.0008,
      "step": 172400
    },
    {
      "epoch": 2.8210750224985683,
      "grad_norm": 0.010016994550824165,
      "learning_rate": 1.0796614767297764e-07,
      "loss": 0.0007,
      "step": 172410
    },
    {
      "epoch": 2.8212386484496443,
      "grad_norm": 0.034603919833898544,
      "learning_rate": 1.0776948114132769e-07,
      "loss": 0.0032,
      "step": 172420
    },
    {
      "epoch": 2.82140227440072,
      "grad_norm": 0.010988828726112843,
      "learning_rate": 1.075729919412749e-07,
      "loss": 0.0008,
      "step": 172430
    },
    {
      "epoch": 2.821565900351796,
      "grad_norm": 0.006998078897595406,
      "learning_rate": 1.0737668007994306e-07,
      "loss": 0.0008,
      "step": 172440
    },
    {
      "epoch": 2.821729526302872,
      "grad_norm": 0.03496389091014862,
      "learning_rate": 1.0718054556444756e-07,
      "loss": 0.0011,
      "step": 172450
    },
    {
      "epoch": 2.8218931522539474,
      "grad_norm": 0.053340282291173935,
      "learning_rate": 1.0698458840189662e-07,
      "loss": 0.0017,
      "step": 172460
    },
    {
      "epoch": 2.8220567782050234,
      "grad_norm": 0.0879424586892128,
      "learning_rate": 1.0678880859939566e-07,
      "loss": 0.0005,
      "step": 172470
    },
    {
      "epoch": 2.8222204041560994,
      "grad_norm": 0.09867893159389496,
      "learning_rate": 1.0659320616403901e-07,
      "loss": 0.0007,
      "step": 172480
    },
    {
      "epoch": 2.822384030107175,
      "grad_norm": 0.027121204882860184,
      "learning_rate": 1.0639778110291765e-07,
      "loss": 0.0006,
      "step": 172490
    },
    {
      "epoch": 2.822547656058251,
      "grad_norm": 0.01210506446659565,
      "learning_rate": 1.0620253342311538e-07,
      "loss": 0.0011,
      "step": 172500
    },
    {
      "epoch": 2.8227112820093265,
      "grad_norm": 0.10004500299692154,
      "learning_rate": 1.0600746313170929e-07,
      "loss": 0.0009,
      "step": 172510
    },
    {
      "epoch": 2.8228749079604025,
      "grad_norm": 0.10850732773542404,
      "learning_rate": 1.0581257023577096e-07,
      "loss": 0.0006,
      "step": 172520
    },
    {
      "epoch": 2.8230385339114785,
      "grad_norm": 0.030207976698875427,
      "learning_rate": 1.0561785474236307e-07,
      "loss": 0.0007,
      "step": 172530
    },
    {
      "epoch": 2.823202159862554,
      "grad_norm": 0.027631904929876328,
      "learning_rate": 1.0542331665854555e-07,
      "loss": 0.0005,
      "step": 172540
    },
    {
      "epoch": 2.82336578581363,
      "grad_norm": 0.07763148844242096,
      "learning_rate": 1.0522895599136829e-07,
      "loss": 0.0003,
      "step": 172550
    },
    {
      "epoch": 2.8235294117647056,
      "grad_norm": 0.04144243896007538,
      "learning_rate": 1.0503477274787787e-07,
      "loss": 0.0004,
      "step": 172560
    },
    {
      "epoch": 2.8236930377157816,
      "grad_norm": 0.18461155891418457,
      "learning_rate": 1.0484076693511147e-07,
      "loss": 0.0007,
      "step": 172570
    },
    {
      "epoch": 2.8238566636668576,
      "grad_norm": 0.07364876568317413,
      "learning_rate": 1.0464693856010178e-07,
      "loss": 0.0012,
      "step": 172580
    },
    {
      "epoch": 2.824020289617933,
      "grad_norm": 0.03745228424668312,
      "learning_rate": 1.0445328762987595e-07,
      "loss": 0.0006,
      "step": 172590
    },
    {
      "epoch": 2.824183915569009,
      "grad_norm": 0.06604311615228653,
      "learning_rate": 1.0425981415145114e-07,
      "loss": 0.0006,
      "step": 172600
    },
    {
      "epoch": 2.824347541520085,
      "grad_norm": 0.014754494652152061,
      "learning_rate": 1.0406651813184232e-07,
      "loss": 0.0004,
      "step": 172610
    },
    {
      "epoch": 2.8245111674711607,
      "grad_norm": 0.1038784608244896,
      "learning_rate": 1.0387339957805497e-07,
      "loss": 0.0008,
      "step": 172620
    },
    {
      "epoch": 2.8246747934222367,
      "grad_norm": 0.02769569493830204,
      "learning_rate": 1.0368045849708907e-07,
      "loss": 0.0009,
      "step": 172630
    },
    {
      "epoch": 2.8248384193733127,
      "grad_norm": 0.0018812255002558231,
      "learning_rate": 1.0348769489593847e-07,
      "loss": 0.0005,
      "step": 172640
    },
    {
      "epoch": 2.8250020453243883,
      "grad_norm": 0.02080894634127617,
      "learning_rate": 1.0329510878159144e-07,
      "loss": 0.0008,
      "step": 172650
    },
    {
      "epoch": 2.8251656712754643,
      "grad_norm": 0.13385115563869476,
      "learning_rate": 1.0310270016102686e-07,
      "loss": 0.0007,
      "step": 172660
    },
    {
      "epoch": 2.8253292972265402,
      "grad_norm": 0.0864618569612503,
      "learning_rate": 1.0291046904122026e-07,
      "loss": 0.0021,
      "step": 172670
    },
    {
      "epoch": 2.825492923177616,
      "grad_norm": 0.03795851767063141,
      "learning_rate": 1.0271841542913941e-07,
      "loss": 0.0007,
      "step": 172680
    },
    {
      "epoch": 2.825656549128692,
      "grad_norm": 0.03524649888277054,
      "learning_rate": 1.0252653933174539e-07,
      "loss": 0.0005,
      "step": 172690
    },
    {
      "epoch": 2.825820175079768,
      "grad_norm": 0.025355640798807144,
      "learning_rate": 1.0233484075599431e-07,
      "loss": 0.0005,
      "step": 172700
    },
    {
      "epoch": 2.8259838010308433,
      "grad_norm": 0.021268676966428757,
      "learning_rate": 1.0214331970883285e-07,
      "loss": 0.0003,
      "step": 172710
    },
    {
      "epoch": 2.8261474269819193,
      "grad_norm": 0.29022401571273804,
      "learning_rate": 1.0195197619720542e-07,
      "loss": 0.0009,
      "step": 172720
    },
    {
      "epoch": 2.8263110529329953,
      "grad_norm": 0.03801962360739708,
      "learning_rate": 1.0176081022804596e-07,
      "loss": 0.0005,
      "step": 172730
    },
    {
      "epoch": 2.826474678884071,
      "grad_norm": 0.08470553159713745,
      "learning_rate": 1.01569821808285e-07,
      "loss": 0.0007,
      "step": 172740
    },
    {
      "epoch": 2.826638304835147,
      "grad_norm": 0.10154219716787338,
      "learning_rate": 1.0137901094484537e-07,
      "loss": 0.0006,
      "step": 172750
    },
    {
      "epoch": 2.826801930786223,
      "grad_norm": 0.09925787895917892,
      "learning_rate": 1.0118837764464207e-07,
      "loss": 0.0008,
      "step": 172760
    },
    {
      "epoch": 2.8269655567372984,
      "grad_norm": 0.2239476591348648,
      "learning_rate": 1.009979219145868e-07,
      "loss": 0.0012,
      "step": 172770
    },
    {
      "epoch": 2.8271291826883744,
      "grad_norm": 0.42145049571990967,
      "learning_rate": 1.008076437615818e-07,
      "loss": 0.0016,
      "step": 172780
    },
    {
      "epoch": 2.8272928086394504,
      "grad_norm": 0.020358722656965256,
      "learning_rate": 1.0061754319252493e-07,
      "loss": 0.0008,
      "step": 172790
    },
    {
      "epoch": 2.827456434590526,
      "grad_norm": 0.012594140134751797,
      "learning_rate": 1.0042762021430729e-07,
      "loss": 0.0014,
      "step": 172800
    },
    {
      "epoch": 2.827620060541602,
      "grad_norm": 0.05987085402011871,
      "learning_rate": 1.0023787483381175e-07,
      "loss": 0.0007,
      "step": 172810
    },
    {
      "epoch": 2.827783686492678,
      "grad_norm": 0.10527680069208145,
      "learning_rate": 1.000483070579178e-07,
      "loss": 0.0007,
      "step": 172820
    },
    {
      "epoch": 2.8279473124437535,
      "grad_norm": 0.08468912541866302,
      "learning_rate": 9.985891689349548e-08,
      "loss": 0.0008,
      "step": 172830
    },
    {
      "epoch": 2.8281109383948295,
      "grad_norm": 0.07809275388717651,
      "learning_rate": 9.966970434741096e-08,
      "loss": 0.0008,
      "step": 172840
    },
    {
      "epoch": 2.8282745643459055,
      "grad_norm": 0.11739545315504074,
      "learning_rate": 9.948066942652157e-08,
      "loss": 0.0017,
      "step": 172850
    },
    {
      "epoch": 2.828438190296981,
      "grad_norm": 0.03874126449227333,
      "learning_rate": 9.929181213767957e-08,
      "loss": 0.0005,
      "step": 172860
    },
    {
      "epoch": 2.828601816248057,
      "grad_norm": 0.03518207371234894,
      "learning_rate": 9.910313248773118e-08,
      "loss": 0.0003,
      "step": 172870
    },
    {
      "epoch": 2.8287654421991326,
      "grad_norm": 0.06403477489948273,
      "learning_rate": 9.89146304835148e-08,
      "loss": 0.0003,
      "step": 172880
    },
    {
      "epoch": 2.8289290681502086,
      "grad_norm": 0.025648249313235283,
      "learning_rate": 9.87263061318644e-08,
      "loss": 0.0004,
      "step": 172890
    },
    {
      "epoch": 2.8290926941012846,
      "grad_norm": 0.07102353125810623,
      "learning_rate": 9.853815943960454e-08,
      "loss": 0.0005,
      "step": 172900
    },
    {
      "epoch": 2.82925632005236,
      "grad_norm": 0.007533930242061615,
      "learning_rate": 9.835019041355697e-08,
      "loss": 0.0011,
      "step": 172910
    },
    {
      "epoch": 2.829419946003436,
      "grad_norm": 0.19549287855625153,
      "learning_rate": 9.816239906053348e-08,
      "loss": 0.0004,
      "step": 172920
    },
    {
      "epoch": 2.8295835719545117,
      "grad_norm": 0.02645001746714115,
      "learning_rate": 9.79747853873425e-08,
      "loss": 0.0005,
      "step": 172930
    },
    {
      "epoch": 2.8297471979055877,
      "grad_norm": 0.06650792062282562,
      "learning_rate": 9.778734940078305e-08,
      "loss": 0.0012,
      "step": 172940
    },
    {
      "epoch": 2.8299108238566637,
      "grad_norm": 0.006504692602902651,
      "learning_rate": 9.760009110765022e-08,
      "loss": 0.0005,
      "step": 172950
    },
    {
      "epoch": 2.8300744498077393,
      "grad_norm": 0.14010770618915558,
      "learning_rate": 9.741301051473195e-08,
      "loss": 0.001,
      "step": 172960
    },
    {
      "epoch": 2.8302380757588153,
      "grad_norm": 0.009898989461362362,
      "learning_rate": 9.722610762880836e-08,
      "loss": 0.0006,
      "step": 172970
    },
    {
      "epoch": 2.8304017017098912,
      "grad_norm": 0.06638094782829285,
      "learning_rate": 9.703938245665567e-08,
      "loss": 0.0008,
      "step": 172980
    },
    {
      "epoch": 2.830565327660967,
      "grad_norm": 0.13438630104064941,
      "learning_rate": 9.685283500504073e-08,
      "loss": 0.0008,
      "step": 172990
    },
    {
      "epoch": 2.830728953612043,
      "grad_norm": 0.006189462263137102,
      "learning_rate": 9.666646528072643e-08,
      "loss": 0.0009,
      "step": 173000
    },
    {
      "epoch": 2.830892579563119,
      "grad_norm": 0.12859684228897095,
      "learning_rate": 9.648027329046738e-08,
      "loss": 0.0005,
      "step": 173010
    },
    {
      "epoch": 2.8310562055141943,
      "grad_norm": 0.2224467396736145,
      "learning_rate": 9.629425904101319e-08,
      "loss": 0.0005,
      "step": 173020
    },
    {
      "epoch": 2.8312198314652703,
      "grad_norm": 0.018172206357121468,
      "learning_rate": 9.610842253910735e-08,
      "loss": 0.001,
      "step": 173030
    },
    {
      "epoch": 2.8313834574163463,
      "grad_norm": 0.03721041977405548,
      "learning_rate": 9.592276379148391e-08,
      "loss": 0.0006,
      "step": 173040
    },
    {
      "epoch": 2.831547083367422,
      "grad_norm": 0.08655502647161484,
      "learning_rate": 9.573728280487471e-08,
      "loss": 0.0011,
      "step": 173050
    },
    {
      "epoch": 2.831710709318498,
      "grad_norm": 0.0030999192968010902,
      "learning_rate": 9.555197958600104e-08,
      "loss": 0.0007,
      "step": 173060
    },
    {
      "epoch": 2.831874335269574,
      "grad_norm": 0.08408474177122116,
      "learning_rate": 9.536685414158086e-08,
      "loss": 0.0005,
      "step": 173070
    },
    {
      "epoch": 2.8320379612206494,
      "grad_norm": 0.09025789797306061,
      "learning_rate": 9.518190647832438e-08,
      "loss": 0.0005,
      "step": 173080
    },
    {
      "epoch": 2.8322015871717254,
      "grad_norm": 0.012574669905006886,
      "learning_rate": 9.499713660293508e-08,
      "loss": 0.0007,
      "step": 173090
    },
    {
      "epoch": 2.8323652131228014,
      "grad_norm": 0.008571373298764229,
      "learning_rate": 9.481254452211097e-08,
      "loss": 0.0003,
      "step": 173100
    },
    {
      "epoch": 2.832528839073877,
      "grad_norm": 0.0567251555621624,
      "learning_rate": 9.462813024254225e-08,
      "loss": 0.0005,
      "step": 173110
    },
    {
      "epoch": 2.832692465024953,
      "grad_norm": 0.029907390475273132,
      "learning_rate": 9.444389377091468e-08,
      "loss": 0.0009,
      "step": 173120
    },
    {
      "epoch": 2.832856090976029,
      "grad_norm": 0.0502646267414093,
      "learning_rate": 9.425983511390569e-08,
      "loss": 0.0009,
      "step": 173130
    },
    {
      "epoch": 2.8330197169271045,
      "grad_norm": 0.035456735640764236,
      "learning_rate": 9.40759542781866e-08,
      "loss": 0.0005,
      "step": 173140
    },
    {
      "epoch": 2.8331833428781805,
      "grad_norm": 0.0010037024039775133,
      "learning_rate": 9.389225127042378e-08,
      "loss": 0.0003,
      "step": 173150
    },
    {
      "epoch": 2.8333469688292565,
      "grad_norm": 0.03924174606800079,
      "learning_rate": 9.370872609727466e-08,
      "loss": 0.0003,
      "step": 173160
    },
    {
      "epoch": 2.833510594780332,
      "grad_norm": 0.02810162864625454,
      "learning_rate": 9.352537876539281e-08,
      "loss": 0.0007,
      "step": 173170
    },
    {
      "epoch": 2.833674220731408,
      "grad_norm": 0.021958060562610626,
      "learning_rate": 9.334220928142346e-08,
      "loss": 0.0005,
      "step": 173180
    },
    {
      "epoch": 2.833837846682484,
      "grad_norm": 0.016990140080451965,
      "learning_rate": 9.315921765200631e-08,
      "loss": 0.0005,
      "step": 173190
    },
    {
      "epoch": 2.8340014726335596,
      "grad_norm": 0.018822545185685158,
      "learning_rate": 9.297640388377383e-08,
      "loss": 0.0004,
      "step": 173200
    },
    {
      "epoch": 2.8341650985846356,
      "grad_norm": 0.24558059871196747,
      "learning_rate": 9.279376798335294e-08,
      "loss": 0.0008,
      "step": 173210
    },
    {
      "epoch": 2.8343287245357116,
      "grad_norm": 0.10267245024442673,
      "learning_rate": 9.261130995736389e-08,
      "loss": 0.0006,
      "step": 173220
    },
    {
      "epoch": 2.834492350486787,
      "grad_norm": 0.15842097997665405,
      "learning_rate": 9.24290298124203e-08,
      "loss": 0.0005,
      "step": 173230
    },
    {
      "epoch": 2.834655976437863,
      "grad_norm": 0.16673073172569275,
      "learning_rate": 9.224692755512965e-08,
      "loss": 0.0009,
      "step": 173240
    },
    {
      "epoch": 2.834819602388939,
      "grad_norm": 0.040787916630506516,
      "learning_rate": 9.206500319209222e-08,
      "loss": 0.0006,
      "step": 173250
    },
    {
      "epoch": 2.8349832283400147,
      "grad_norm": 0.08588576316833496,
      "learning_rate": 9.188325672990272e-08,
      "loss": 0.0005,
      "step": 173260
    },
    {
      "epoch": 2.8351468542910907,
      "grad_norm": 0.18132197856903076,
      "learning_rate": 9.170168817514813e-08,
      "loss": 0.0009,
      "step": 173270
    },
    {
      "epoch": 2.8353104802421663,
      "grad_norm": 0.02750786580145359,
      "learning_rate": 9.152029753441149e-08,
      "loss": 0.0006,
      "step": 173280
    },
    {
      "epoch": 2.8354741061932422,
      "grad_norm": 0.004450025036931038,
      "learning_rate": 9.13390848142659e-08,
      "loss": 0.0006,
      "step": 173290
    },
    {
      "epoch": 2.8356377321443182,
      "grad_norm": 0.008194906637072563,
      "learning_rate": 9.115805002128109e-08,
      "loss": 0.0005,
      "step": 173300
    },
    {
      "epoch": 2.835801358095394,
      "grad_norm": 0.048089105635881424,
      "learning_rate": 9.097719316201903e-08,
      "loss": 0.0003,
      "step": 173310
    },
    {
      "epoch": 2.83596498404647,
      "grad_norm": 0.028552373871207237,
      "learning_rate": 9.079651424303503e-08,
      "loss": 0.0006,
      "step": 173320
    },
    {
      "epoch": 2.8361286099975453,
      "grad_norm": 0.0013884111540392041,
      "learning_rate": 9.061601327087888e-08,
      "loss": 0.0006,
      "step": 173330
    },
    {
      "epoch": 2.8362922359486213,
      "grad_norm": 0.056187860667705536,
      "learning_rate": 9.043569025209253e-08,
      "loss": 0.0007,
      "step": 173340
    },
    {
      "epoch": 2.8364558618996973,
      "grad_norm": 0.011024200357496738,
      "learning_rate": 9.0255545193213e-08,
      "loss": 0.0006,
      "step": 173350
    },
    {
      "epoch": 2.836619487850773,
      "grad_norm": 0.18066450953483582,
      "learning_rate": 9.00755781007695e-08,
      "loss": 0.0005,
      "step": 173360
    },
    {
      "epoch": 2.836783113801849,
      "grad_norm": 0.06157389283180237,
      "learning_rate": 8.989578898128515e-08,
      "loss": 0.0005,
      "step": 173370
    },
    {
      "epoch": 2.836946739752925,
      "grad_norm": 0.07567913830280304,
      "learning_rate": 8.97161778412775e-08,
      "loss": 0.0004,
      "step": 173380
    },
    {
      "epoch": 2.8371103657040004,
      "grad_norm": 0.012874470092356205,
      "learning_rate": 8.953674468725692e-08,
      "loss": 0.0005,
      "step": 173390
    },
    {
      "epoch": 2.8372739916550764,
      "grad_norm": 0.09604622423648834,
      "learning_rate": 8.935748952572765e-08,
      "loss": 0.0003,
      "step": 173400
    },
    {
      "epoch": 2.8374376176061524,
      "grad_norm": 0.06352297961711884,
      "learning_rate": 8.917841236318669e-08,
      "loss": 0.0008,
      "step": 173410
    },
    {
      "epoch": 2.837601243557228,
      "grad_norm": 0.06465206295251846,
      "learning_rate": 8.899951320612554e-08,
      "loss": 0.0008,
      "step": 173420
    },
    {
      "epoch": 2.837764869508304,
      "grad_norm": 0.032496921718120575,
      "learning_rate": 8.882079206102844e-08,
      "loss": 0.0008,
      "step": 173430
    },
    {
      "epoch": 2.83792849545938,
      "grad_norm": 0.060703475028276443,
      "learning_rate": 8.864224893437412e-08,
      "loss": 0.0014,
      "step": 173440
    },
    {
      "epoch": 2.8380921214104555,
      "grad_norm": 0.05830821767449379,
      "learning_rate": 8.84638838326346e-08,
      "loss": 0.0005,
      "step": 173450
    },
    {
      "epoch": 2.8382557473615315,
      "grad_norm": 0.05228746309876442,
      "learning_rate": 8.828569676227417e-08,
      "loss": 0.001,
      "step": 173460
    },
    {
      "epoch": 2.8384193733126075,
      "grad_norm": 0.08254444599151611,
      "learning_rate": 8.810768772975265e-08,
      "loss": 0.0004,
      "step": 173470
    },
    {
      "epoch": 2.838582999263683,
      "grad_norm": 0.03338150680065155,
      "learning_rate": 8.792985674152099e-08,
      "loss": 0.0004,
      "step": 173480
    },
    {
      "epoch": 2.838746625214759,
      "grad_norm": 0.03274044767022133,
      "learning_rate": 8.775220380402738e-08,
      "loss": 0.0008,
      "step": 173490
    },
    {
      "epoch": 2.838910251165835,
      "grad_norm": 0.041836876422166824,
      "learning_rate": 8.757472892370943e-08,
      "loss": 0.0006,
      "step": 173500
    },
    {
      "epoch": 2.8390738771169106,
      "grad_norm": 0.02346048690378666,
      "learning_rate": 8.739743210700091e-08,
      "loss": 0.0003,
      "step": 173510
    },
    {
      "epoch": 2.8392375030679866,
      "grad_norm": 0.07030380517244339,
      "learning_rate": 8.722031336032888e-08,
      "loss": 0.0005,
      "step": 173520
    },
    {
      "epoch": 2.8394011290190626,
      "grad_norm": 0.056974031031131744,
      "learning_rate": 8.704337269011209e-08,
      "loss": 0.0015,
      "step": 173530
    },
    {
      "epoch": 2.839564754970138,
      "grad_norm": 0.02623937651515007,
      "learning_rate": 8.686661010276598e-08,
      "loss": 0.0007,
      "step": 173540
    },
    {
      "epoch": 2.839728380921214,
      "grad_norm": 0.04807814583182335,
      "learning_rate": 8.669002560469653e-08,
      "loss": 0.0005,
      "step": 173550
    },
    {
      "epoch": 2.83989200687229,
      "grad_norm": 0.037363871932029724,
      "learning_rate": 8.651361920230417e-08,
      "loss": 0.0005,
      "step": 173560
    },
    {
      "epoch": 2.8400556328233657,
      "grad_norm": 0.05650179833173752,
      "learning_rate": 8.63373909019849e-08,
      "loss": 0.0008,
      "step": 173570
    },
    {
      "epoch": 2.8402192587744417,
      "grad_norm": 0.01332570519298315,
      "learning_rate": 8.616134071012472e-08,
      "loss": 0.0005,
      "step": 173580
    },
    {
      "epoch": 2.8403828847255177,
      "grad_norm": 0.0773705393075943,
      "learning_rate": 8.598546863310631e-08,
      "loss": 0.0004,
      "step": 173590
    },
    {
      "epoch": 2.8405465106765933,
      "grad_norm": 0.010461600497364998,
      "learning_rate": 8.580977467730345e-08,
      "loss": 0.0006,
      "step": 173600
    },
    {
      "epoch": 2.8407101366276692,
      "grad_norm": 0.045217934995889664,
      "learning_rate": 8.563425884908604e-08,
      "loss": 0.0005,
      "step": 173610
    },
    {
      "epoch": 2.8408737625787452,
      "grad_norm": 0.060924991965293884,
      "learning_rate": 8.545892115481514e-08,
      "loss": 0.001,
      "step": 173620
    },
    {
      "epoch": 2.841037388529821,
      "grad_norm": 0.023583533242344856,
      "learning_rate": 8.528376160084673e-08,
      "loss": 0.0007,
      "step": 173630
    },
    {
      "epoch": 2.841201014480897,
      "grad_norm": 0.10948897898197174,
      "learning_rate": 8.510878019352965e-08,
      "loss": 0.0016,
      "step": 173640
    },
    {
      "epoch": 2.841364640431973,
      "grad_norm": 0.2389497011899948,
      "learning_rate": 8.493397693920657e-08,
      "loss": 0.0013,
      "step": 173650
    },
    {
      "epoch": 2.8415282663830483,
      "grad_norm": 0.027986939996480942,
      "learning_rate": 8.475935184421413e-08,
      "loss": 0.0004,
      "step": 173660
    },
    {
      "epoch": 2.8416918923341243,
      "grad_norm": 0.11178158968687057,
      "learning_rate": 8.458490491488114e-08,
      "loss": 0.0007,
      "step": 173670
    },
    {
      "epoch": 2.8418555182852,
      "grad_norm": 0.200485959649086,
      "learning_rate": 8.441063615753197e-08,
      "loss": 0.001,
      "step": 173680
    },
    {
      "epoch": 2.842019144236276,
      "grad_norm": 0.03016781434416771,
      "learning_rate": 8.423654557848215e-08,
      "loss": 0.0011,
      "step": 173690
    },
    {
      "epoch": 2.8421827701873514,
      "grad_norm": 0.05747764930129051,
      "learning_rate": 8.406263318404384e-08,
      "loss": 0.0005,
      "step": 173700
    },
    {
      "epoch": 2.8423463961384274,
      "grad_norm": 0.08647391200065613,
      "learning_rate": 8.388889898051922e-08,
      "loss": 0.0005,
      "step": 173710
    },
    {
      "epoch": 2.8425100220895034,
      "grad_norm": 0.04320474714040756,
      "learning_rate": 8.371534297420658e-08,
      "loss": 0.0005,
      "step": 173720
    },
    {
      "epoch": 2.842673648040579,
      "grad_norm": 0.11320873349905014,
      "learning_rate": 8.354196517139701e-08,
      "loss": 0.0006,
      "step": 173730
    },
    {
      "epoch": 2.842837273991655,
      "grad_norm": 0.07384967803955078,
      "learning_rate": 8.336876557837436e-08,
      "loss": 0.0007,
      "step": 173740
    },
    {
      "epoch": 2.843000899942731,
      "grad_norm": 0.21078170835971832,
      "learning_rate": 8.319574420141807e-08,
      "loss": 0.0012,
      "step": 173750
    },
    {
      "epoch": 2.8431645258938065,
      "grad_norm": 0.04817560687661171,
      "learning_rate": 8.302290104679811e-08,
      "loss": 0.0006,
      "step": 173760
    },
    {
      "epoch": 2.8433281518448825,
      "grad_norm": 0.058492567390203476,
      "learning_rate": 8.285023612078059e-08,
      "loss": 0.0011,
      "step": 173770
    },
    {
      "epoch": 2.8434917777959585,
      "grad_norm": 0.022522741928696632,
      "learning_rate": 8.267774942962382e-08,
      "loss": 0.0004,
      "step": 173780
    },
    {
      "epoch": 2.843655403747034,
      "grad_norm": 0.21324092149734497,
      "learning_rate": 8.250544097957946e-08,
      "loss": 0.0011,
      "step": 173790
    },
    {
      "epoch": 2.84381902969811,
      "grad_norm": 0.08894387632608414,
      "learning_rate": 8.233331077689532e-08,
      "loss": 0.0006,
      "step": 173800
    },
    {
      "epoch": 2.843982655649186,
      "grad_norm": 0.0018750797025859356,
      "learning_rate": 8.216135882780807e-08,
      "loss": 0.0006,
      "step": 173810
    },
    {
      "epoch": 2.8441462816002616,
      "grad_norm": 0.058654479682445526,
      "learning_rate": 8.198958513855271e-08,
      "loss": 0.0003,
      "step": 173820
    },
    {
      "epoch": 2.8443099075513376,
      "grad_norm": 0.07440018653869629,
      "learning_rate": 8.181798971535482e-08,
      "loss": 0.0007,
      "step": 173830
    },
    {
      "epoch": 2.8444735335024136,
      "grad_norm": 0.029554273933172226,
      "learning_rate": 8.164657256443386e-08,
      "loss": 0.0013,
      "step": 173840
    },
    {
      "epoch": 2.844637159453489,
      "grad_norm": 0.13609381020069122,
      "learning_rate": 8.147533369200433e-08,
      "loss": 0.0006,
      "step": 173850
    },
    {
      "epoch": 2.844800785404565,
      "grad_norm": 0.026124535128474236,
      "learning_rate": 8.13042731042718e-08,
      "loss": 0.0002,
      "step": 173860
    },
    {
      "epoch": 2.844964411355641,
      "grad_norm": 0.07678091526031494,
      "learning_rate": 8.113339080743798e-08,
      "loss": 0.0004,
      "step": 173870
    },
    {
      "epoch": 2.8451280373067167,
      "grad_norm": 0.1092134565114975,
      "learning_rate": 8.096268680769626e-08,
      "loss": 0.0006,
      "step": 173880
    },
    {
      "epoch": 2.8452916632577927,
      "grad_norm": 0.06291822344064713,
      "learning_rate": 8.0792161111235e-08,
      "loss": 0.0015,
      "step": 173890
    },
    {
      "epoch": 2.8454552892088687,
      "grad_norm": 0.15437263250350952,
      "learning_rate": 8.062181372423428e-08,
      "loss": 0.0006,
      "step": 173900
    },
    {
      "epoch": 2.8456189151599443,
      "grad_norm": 0.0075996145606040955,
      "learning_rate": 8.045164465286969e-08,
      "loss": 0.0005,
      "step": 173910
    },
    {
      "epoch": 2.8457825411110202,
      "grad_norm": 0.33961060643196106,
      "learning_rate": 8.02816539033091e-08,
      "loss": 0.0008,
      "step": 173920
    },
    {
      "epoch": 2.8459461670620962,
      "grad_norm": 0.5353030562400818,
      "learning_rate": 8.01118414817148e-08,
      "loss": 0.0009,
      "step": 173930
    },
    {
      "epoch": 2.846109793013172,
      "grad_norm": 0.06713154166936874,
      "learning_rate": 7.994220739424075e-08,
      "loss": 0.0005,
      "step": 173940
    },
    {
      "epoch": 2.846273418964248,
      "grad_norm": 0.02468140609562397,
      "learning_rate": 7.977275164703757e-08,
      "loss": 0.0005,
      "step": 173950
    },
    {
      "epoch": 2.846437044915324,
      "grad_norm": 0.34537604451179504,
      "learning_rate": 7.960347424624648e-08,
      "loss": 0.0011,
      "step": 173960
    },
    {
      "epoch": 2.8466006708663993,
      "grad_norm": 0.06103261560201645,
      "learning_rate": 7.943437519800312e-08,
      "loss": 0.0004,
      "step": 173970
    },
    {
      "epoch": 2.8467642968174753,
      "grad_norm": 0.05063708499073982,
      "learning_rate": 7.926545450843758e-08,
      "loss": 0.0004,
      "step": 173980
    },
    {
      "epoch": 2.8469279227685513,
      "grad_norm": 0.10059316456317902,
      "learning_rate": 7.909671218367277e-08,
      "loss": 0.0016,
      "step": 173990
    },
    {
      "epoch": 2.847091548719627,
      "grad_norm": 0.10250764340162277,
      "learning_rate": 7.892814822982487e-08,
      "loss": 0.0005,
      "step": 174000
    },
    {
      "epoch": 2.847255174670703,
      "grad_norm": 0.16875368356704712,
      "learning_rate": 7.875976265300456e-08,
      "loss": 0.0009,
      "step": 174010
    },
    {
      "epoch": 2.847418800621779,
      "grad_norm": 0.131410151720047,
      "learning_rate": 7.859155545931418e-08,
      "loss": 0.0009,
      "step": 174020
    },
    {
      "epoch": 2.8475824265728544,
      "grad_norm": 0.16458827257156372,
      "learning_rate": 7.84235266548522e-08,
      "loss": 0.0014,
      "step": 174030
    },
    {
      "epoch": 2.8477460525239304,
      "grad_norm": 0.08364089578390121,
      "learning_rate": 7.825567624570818e-08,
      "loss": 0.0005,
      "step": 174040
    },
    {
      "epoch": 2.847909678475006,
      "grad_norm": 0.016370099037885666,
      "learning_rate": 7.808800423796781e-08,
      "loss": 0.0004,
      "step": 174050
    },
    {
      "epoch": 2.848073304426082,
      "grad_norm": 0.009846849367022514,
      "learning_rate": 7.792051063770733e-08,
      "loss": 0.0008,
      "step": 174060
    },
    {
      "epoch": 2.848236930377158,
      "grad_norm": 0.037272486835718155,
      "learning_rate": 7.775319545099802e-08,
      "loss": 0.0016,
      "step": 174070
    },
    {
      "epoch": 2.8484005563282335,
      "grad_norm": 0.12668658792972565,
      "learning_rate": 7.7586058683905e-08,
      "loss": 0.0009,
      "step": 174080
    },
    {
      "epoch": 2.8485641822793095,
      "grad_norm": 0.019610220566391945,
      "learning_rate": 7.741910034248678e-08,
      "loss": 0.0008,
      "step": 174090
    },
    {
      "epoch": 2.848727808230385,
      "grad_norm": 0.0810527428984642,
      "learning_rate": 7.725232043279574e-08,
      "loss": 0.0006,
      "step": 174100
    },
    {
      "epoch": 2.848891434181461,
      "grad_norm": 0.0777733325958252,
      "learning_rate": 7.708571896087591e-08,
      "loss": 0.0005,
      "step": 174110
    },
    {
      "epoch": 2.849055060132537,
      "grad_norm": 0.04647945985198021,
      "learning_rate": 7.691929593276692e-08,
      "loss": 0.0006,
      "step": 174120
    },
    {
      "epoch": 2.8492186860836126,
      "grad_norm": 0.052896466106176376,
      "learning_rate": 7.675305135450117e-08,
      "loss": 0.0013,
      "step": 174130
    },
    {
      "epoch": 2.8493823120346886,
      "grad_norm": 0.09321468323469162,
      "learning_rate": 7.658698523210496e-08,
      "loss": 0.0003,
      "step": 174140
    },
    {
      "epoch": 2.8495459379857646,
      "grad_norm": 0.031998150050640106,
      "learning_rate": 7.642109757159733e-08,
      "loss": 0.0004,
      "step": 174150
    },
    {
      "epoch": 2.84970956393684,
      "grad_norm": 0.14246892929077148,
      "learning_rate": 7.625538837899071e-08,
      "loss": 0.0011,
      "step": 174160
    },
    {
      "epoch": 2.849873189887916,
      "grad_norm": 0.05914086848497391,
      "learning_rate": 7.608985766029309e-08,
      "loss": 0.0008,
      "step": 174170
    },
    {
      "epoch": 2.850036815838992,
      "grad_norm": 0.08203393965959549,
      "learning_rate": 7.592450542150297e-08,
      "loss": 0.001,
      "step": 174180
    },
    {
      "epoch": 2.8502004417900677,
      "grad_norm": 0.07545647025108337,
      "learning_rate": 7.57593316686156e-08,
      "loss": 0.0005,
      "step": 174190
    },
    {
      "epoch": 2.8503640677411437,
      "grad_norm": 0.011526244692504406,
      "learning_rate": 7.559433640761726e-08,
      "loss": 0.0004,
      "step": 174200
    },
    {
      "epoch": 2.8505276936922197,
      "grad_norm": 0.13257025182247162,
      "learning_rate": 7.542951964448874e-08,
      "loss": 0.0006,
      "step": 174210
    },
    {
      "epoch": 2.8506913196432953,
      "grad_norm": 0.045285262167453766,
      "learning_rate": 7.526488138520415e-08,
      "loss": 0.0004,
      "step": 174220
    },
    {
      "epoch": 2.8508549455943712,
      "grad_norm": 0.009759592823684216,
      "learning_rate": 7.510042163573095e-08,
      "loss": 0.0003,
      "step": 174230
    },
    {
      "epoch": 2.8510185715454472,
      "grad_norm": 0.2923644781112671,
      "learning_rate": 7.493614040203101e-08,
      "loss": 0.0016,
      "step": 174240
    },
    {
      "epoch": 2.851182197496523,
      "grad_norm": 0.1614042967557907,
      "learning_rate": 7.477203769005903e-08,
      "loss": 0.0008,
      "step": 174250
    },
    {
      "epoch": 2.851345823447599,
      "grad_norm": 0.0468735545873642,
      "learning_rate": 7.460811350576359e-08,
      "loss": 0.0003,
      "step": 174260
    },
    {
      "epoch": 2.851509449398675,
      "grad_norm": 0.32878729701042175,
      "learning_rate": 7.444436785508602e-08,
      "loss": 0.0012,
      "step": 174270
    },
    {
      "epoch": 2.8516730753497503,
      "grad_norm": 0.08398101478815079,
      "learning_rate": 7.428080074396105e-08,
      "loss": 0.0006,
      "step": 174280
    },
    {
      "epoch": 2.8518367013008263,
      "grad_norm": 0.051134463399648666,
      "learning_rate": 7.411741217831947e-08,
      "loss": 0.0009,
      "step": 174290
    },
    {
      "epoch": 2.8520003272519023,
      "grad_norm": 0.07836651057004929,
      "learning_rate": 7.395420216408155e-08,
      "loss": 0.0007,
      "step": 174300
    },
    {
      "epoch": 2.852163953202978,
      "grad_norm": 0.05573960021138191,
      "learning_rate": 7.379117070716534e-08,
      "loss": 0.0007,
      "step": 174310
    },
    {
      "epoch": 2.852327579154054,
      "grad_norm": 0.05728914588689804,
      "learning_rate": 7.36283178134789e-08,
      "loss": 0.0003,
      "step": 174320
    },
    {
      "epoch": 2.85249120510513,
      "grad_norm": 0.04238661751151085,
      "learning_rate": 7.346564348892582e-08,
      "loss": 0.0008,
      "step": 174330
    },
    {
      "epoch": 2.8526548310562054,
      "grad_norm": 0.028013156726956367,
      "learning_rate": 7.33031477394025e-08,
      "loss": 0.0005,
      "step": 174340
    },
    {
      "epoch": 2.8528184570072814,
      "grad_norm": 0.15349432826042175,
      "learning_rate": 7.314083057079924e-08,
      "loss": 0.0011,
      "step": 174350
    },
    {
      "epoch": 2.8529820829583574,
      "grad_norm": 0.001606814912520349,
      "learning_rate": 7.29786919889991e-08,
      "loss": 0.0005,
      "step": 174360
    },
    {
      "epoch": 2.853145708909433,
      "grad_norm": 0.01952027715742588,
      "learning_rate": 7.281673199987959e-08,
      "loss": 0.0005,
      "step": 174370
    },
    {
      "epoch": 2.853309334860509,
      "grad_norm": 0.007583265658468008,
      "learning_rate": 7.265495060931216e-08,
      "loss": 0.0011,
      "step": 174380
    },
    {
      "epoch": 2.853472960811585,
      "grad_norm": 0.12129753082990646,
      "learning_rate": 7.249334782315931e-08,
      "loss": 0.0006,
      "step": 174390
    },
    {
      "epoch": 2.8536365867626605,
      "grad_norm": 0.21874582767486572,
      "learning_rate": 7.233192364728025e-08,
      "loss": 0.001,
      "step": 174400
    },
    {
      "epoch": 2.8538002127137365,
      "grad_norm": 0.06910939514636993,
      "learning_rate": 7.217067808752532e-08,
      "loss": 0.0005,
      "step": 174410
    },
    {
      "epoch": 2.8539638386648125,
      "grad_norm": 0.08435803651809692,
      "learning_rate": 7.200961114973925e-08,
      "loss": 0.0008,
      "step": 174420
    },
    {
      "epoch": 2.854127464615888,
      "grad_norm": 0.03137621656060219,
      "learning_rate": 7.18487228397613e-08,
      "loss": 0.0004,
      "step": 174430
    },
    {
      "epoch": 2.854291090566964,
      "grad_norm": 0.011698718182742596,
      "learning_rate": 7.168801316342178e-08,
      "loss": 0.0007,
      "step": 174440
    },
    {
      "epoch": 2.8544547165180396,
      "grad_norm": 0.05233995243906975,
      "learning_rate": 7.152748212654769e-08,
      "loss": 0.0005,
      "step": 174450
    },
    {
      "epoch": 2.8546183424691156,
      "grad_norm": 0.01946445368230343,
      "learning_rate": 7.136712973495719e-08,
      "loss": 0.0005,
      "step": 174460
    },
    {
      "epoch": 2.8547819684201916,
      "grad_norm": 0.13896925747394562,
      "learning_rate": 7.120695599446281e-08,
      "loss": 0.0008,
      "step": 174470
    },
    {
      "epoch": 2.854945594371267,
      "grad_norm": 0.008294851519167423,
      "learning_rate": 7.104696091086937e-08,
      "loss": 0.0006,
      "step": 174480
    },
    {
      "epoch": 2.855109220322343,
      "grad_norm": 0.031204285100102425,
      "learning_rate": 7.08871444899778e-08,
      "loss": 0.0007,
      "step": 174490
    },
    {
      "epoch": 2.8552728462734187,
      "grad_norm": 0.014123676344752312,
      "learning_rate": 7.072750673757955e-08,
      "loss": 0.0016,
      "step": 174500
    },
    {
      "epoch": 2.8554364722244947,
      "grad_norm": 0.1242511197924614,
      "learning_rate": 7.05680476594628e-08,
      "loss": 0.0009,
      "step": 174510
    },
    {
      "epoch": 2.8556000981755707,
      "grad_norm": 0.034279245883226395,
      "learning_rate": 7.040876726140678e-08,
      "loss": 0.0004,
      "step": 174520
    },
    {
      "epoch": 2.8557637241266463,
      "grad_norm": 0.08257488906383514,
      "learning_rate": 7.024966554918411e-08,
      "loss": 0.0007,
      "step": 174530
    },
    {
      "epoch": 2.8559273500777222,
      "grad_norm": 0.06265582889318466,
      "learning_rate": 7.009074252856352e-08,
      "loss": 0.0004,
      "step": 174540
    },
    {
      "epoch": 2.8560909760287982,
      "grad_norm": 0.011068779975175858,
      "learning_rate": 6.993199820530482e-08,
      "loss": 0.0008,
      "step": 174550
    },
    {
      "epoch": 2.856254601979874,
      "grad_norm": 0.005089957267045975,
      "learning_rate": 6.977343258516178e-08,
      "loss": 0.0008,
      "step": 174560
    },
    {
      "epoch": 2.85641822793095,
      "grad_norm": 0.0039634425193071365,
      "learning_rate": 6.961504567388255e-08,
      "loss": 0.0011,
      "step": 174570
    },
    {
      "epoch": 2.856581853882026,
      "grad_norm": 0.0073012858629226685,
      "learning_rate": 6.945683747720755e-08,
      "loss": 0.0004,
      "step": 174580
    },
    {
      "epoch": 2.8567454798331013,
      "grad_norm": 0.02657819725573063,
      "learning_rate": 6.929880800087274e-08,
      "loss": 0.001,
      "step": 174590
    },
    {
      "epoch": 2.8569091057841773,
      "grad_norm": 0.020564179867506027,
      "learning_rate": 6.914095725060466e-08,
      "loss": 0.0008,
      "step": 174600
    },
    {
      "epoch": 2.8570727317352533,
      "grad_norm": 0.06299027800559998,
      "learning_rate": 6.89832852321265e-08,
      "loss": 0.0009,
      "step": 174610
    },
    {
      "epoch": 2.857236357686329,
      "grad_norm": 0.02254130132496357,
      "learning_rate": 6.882579195115257e-08,
      "loss": 0.0004,
      "step": 174620
    },
    {
      "epoch": 2.857399983637405,
      "grad_norm": 0.004705629777163267,
      "learning_rate": 6.866847741339222e-08,
      "loss": 0.0004,
      "step": 174630
    },
    {
      "epoch": 2.857563609588481,
      "grad_norm": 0.049303796142339706,
      "learning_rate": 6.851134162454698e-08,
      "loss": 0.0005,
      "step": 174640
    },
    {
      "epoch": 2.8577272355395564,
      "grad_norm": 0.0052415053360164165,
      "learning_rate": 6.83543845903134e-08,
      "loss": 0.0009,
      "step": 174650
    },
    {
      "epoch": 2.8578908614906324,
      "grad_norm": 0.04257628694176674,
      "learning_rate": 6.819760631638083e-08,
      "loss": 0.0008,
      "step": 174660
    },
    {
      "epoch": 2.8580544874417084,
      "grad_norm": 0.15036487579345703,
      "learning_rate": 6.804100680843084e-08,
      "loss": 0.001,
      "step": 174670
    },
    {
      "epoch": 2.858218113392784,
      "grad_norm": 0.03437655046582222,
      "learning_rate": 6.788458607214166e-08,
      "loss": 0.0006,
      "step": 174680
    },
    {
      "epoch": 2.85838173934386,
      "grad_norm": 0.06352607905864716,
      "learning_rate": 6.772834411318152e-08,
      "loss": 0.0008,
      "step": 174690
    },
    {
      "epoch": 2.858545365294936,
      "grad_norm": 0.06638634949922562,
      "learning_rate": 6.75722809372148e-08,
      "loss": 0.0005,
      "step": 174700
    },
    {
      "epoch": 2.8587089912460115,
      "grad_norm": 0.0277895238250494,
      "learning_rate": 6.741639654989752e-08,
      "loss": 0.0006,
      "step": 174710
    },
    {
      "epoch": 2.8588726171970875,
      "grad_norm": 0.06302885711193085,
      "learning_rate": 6.726069095688126e-08,
      "loss": 0.0012,
      "step": 174720
    },
    {
      "epoch": 2.8590362431481635,
      "grad_norm": 0.005250515416264534,
      "learning_rate": 6.71051641638093e-08,
      "loss": 0.0007,
      "step": 174730
    },
    {
      "epoch": 2.859199869099239,
      "grad_norm": 0.077210932970047,
      "learning_rate": 6.694981617631935e-08,
      "loss": 0.0003,
      "step": 174740
    },
    {
      "epoch": 2.859363495050315,
      "grad_norm": 0.0019042406929656863,
      "learning_rate": 6.679464700004246e-08,
      "loss": 0.0005,
      "step": 174750
    },
    {
      "epoch": 2.859527121001391,
      "grad_norm": 0.06501390039920807,
      "learning_rate": 6.663965664060301e-08,
      "loss": 0.0007,
      "step": 174760
    },
    {
      "epoch": 2.8596907469524666,
      "grad_norm": 0.03229280188679695,
      "learning_rate": 6.648484510361819e-08,
      "loss": 0.0011,
      "step": 174770
    },
    {
      "epoch": 2.8598543729035426,
      "grad_norm": 0.01788502186536789,
      "learning_rate": 6.633021239470127e-08,
      "loss": 0.0004,
      "step": 174780
    },
    {
      "epoch": 2.8600179988546186,
      "grad_norm": 0.01595594547688961,
      "learning_rate": 6.617575851945557e-08,
      "loss": 0.0005,
      "step": 174790
    },
    {
      "epoch": 2.860181624805694,
      "grad_norm": 0.06669995188713074,
      "learning_rate": 6.602148348348159e-08,
      "loss": 0.0004,
      "step": 174800
    },
    {
      "epoch": 2.86034525075677,
      "grad_norm": 0.14184415340423584,
      "learning_rate": 6.58673872923693e-08,
      "loss": 0.0007,
      "step": 174810
    },
    {
      "epoch": 2.8605088767078457,
      "grad_norm": 0.4916815757751465,
      "learning_rate": 6.571346995170591e-08,
      "loss": 0.0044,
      "step": 174820
    },
    {
      "epoch": 2.8606725026589217,
      "grad_norm": 0.03106622025370598,
      "learning_rate": 6.555973146706973e-08,
      "loss": 0.0004,
      "step": 174830
    },
    {
      "epoch": 2.8608361286099977,
      "grad_norm": 0.0013145022094249725,
      "learning_rate": 6.540617184403353e-08,
      "loss": 0.0002,
      "step": 174840
    },
    {
      "epoch": 2.8609997545610732,
      "grad_norm": 0.13275207579135895,
      "learning_rate": 6.525279108816396e-08,
      "loss": 0.0006,
      "step": 174850
    },
    {
      "epoch": 2.8611633805121492,
      "grad_norm": 0.01137633714824915,
      "learning_rate": 6.509958920502046e-08,
      "loss": 0.0004,
      "step": 174860
    },
    {
      "epoch": 2.861327006463225,
      "grad_norm": 0.04059248045086861,
      "learning_rate": 6.494656620015582e-08,
      "loss": 0.0005,
      "step": 174870
    },
    {
      "epoch": 2.861490632414301,
      "grad_norm": 0.0016524163074791431,
      "learning_rate": 6.479372207911672e-08,
      "loss": 0.0004,
      "step": 174880
    },
    {
      "epoch": 2.861654258365377,
      "grad_norm": 0.0027647637762129307,
      "learning_rate": 6.464105684744426e-08,
      "loss": 0.0007,
      "step": 174890
    },
    {
      "epoch": 2.8618178843164523,
      "grad_norm": 0.05818874016404152,
      "learning_rate": 6.448857051067181e-08,
      "loss": 0.0015,
      "step": 174900
    },
    {
      "epoch": 2.8619815102675283,
      "grad_norm": 0.012284564785659313,
      "learning_rate": 6.433626307432606e-08,
      "loss": 0.0004,
      "step": 174910
    },
    {
      "epoch": 2.8621451362186043,
      "grad_norm": 0.021488336846232414,
      "learning_rate": 6.418413454392813e-08,
      "loss": 0.0007,
      "step": 174920
    },
    {
      "epoch": 2.86230876216968,
      "grad_norm": 0.09011509269475937,
      "learning_rate": 6.403218492499253e-08,
      "loss": 0.0006,
      "step": 174930
    },
    {
      "epoch": 2.862472388120756,
      "grad_norm": 0.005234993528574705,
      "learning_rate": 6.388041422302759e-08,
      "loss": 0.0009,
      "step": 174940
    },
    {
      "epoch": 2.862636014071832,
      "grad_norm": 0.07900192588567734,
      "learning_rate": 6.37288224435334e-08,
      "loss": 0.0008,
      "step": 174950
    },
    {
      "epoch": 2.8627996400229074,
      "grad_norm": 0.02275443635880947,
      "learning_rate": 6.357740959200553e-08,
      "loss": 0.0005,
      "step": 174960
    },
    {
      "epoch": 2.8629632659739834,
      "grad_norm": 0.05092756822705269,
      "learning_rate": 6.342617567393184e-08,
      "loss": 0.0005,
      "step": 174970
    },
    {
      "epoch": 2.8631268919250594,
      "grad_norm": 0.036253590136766434,
      "learning_rate": 6.327512069479513e-08,
      "loss": 0.0008,
      "step": 174980
    },
    {
      "epoch": 2.863290517876135,
      "grad_norm": 0.017339251935482025,
      "learning_rate": 6.312424466006995e-08,
      "loss": 0.0003,
      "step": 174990
    },
    {
      "epoch": 2.863454143827211,
      "grad_norm": 0.11925626546144485,
      "learning_rate": 6.297354757522523e-08,
      "loss": 0.0003,
      "step": 175000
    },
    {
      "epoch": 2.863454143827211,
      "eval_loss": 0.0009074246045202017,
      "eval_runtime": 3.0981,
      "eval_samples_per_second": 64.556,
      "eval_steps_per_second": 16.139,
      "step": 175000
    },
    {
      "epoch": 2.863617769778287,
      "grad_norm": 0.05676291137933731,
      "learning_rate": 6.282302944572383e-08,
      "loss": 0.0007,
      "step": 175010
    },
    {
      "epoch": 2.8637813957293625,
      "grad_norm": 0.02615579403936863,
      "learning_rate": 6.267269027702084e-08,
      "loss": 0.0007,
      "step": 175020
    },
    {
      "epoch": 2.8639450216804385,
      "grad_norm": 0.10852567851543427,
      "learning_rate": 6.252253007456688e-08,
      "loss": 0.0005,
      "step": 175030
    },
    {
      "epoch": 2.8641086476315145,
      "grad_norm": 0.04380883648991585,
      "learning_rate": 6.237254884380373e-08,
      "loss": 0.0007,
      "step": 175040
    },
    {
      "epoch": 2.86427227358259,
      "grad_norm": 0.2027987241744995,
      "learning_rate": 6.222274659016925e-08,
      "loss": 0.0007,
      "step": 175050
    },
    {
      "epoch": 2.864435899533666,
      "grad_norm": 0.061378706246614456,
      "learning_rate": 6.207312331909244e-08,
      "loss": 0.0013,
      "step": 175060
    },
    {
      "epoch": 2.864599525484742,
      "grad_norm": 0.12418750673532486,
      "learning_rate": 6.192367903599617e-08,
      "loss": 0.001,
      "step": 175070
    },
    {
      "epoch": 2.8647631514358176,
      "grad_norm": 0.005105985328555107,
      "learning_rate": 6.177441374629889e-08,
      "loss": 0.0006,
      "step": 175080
    },
    {
      "epoch": 2.8649267773868936,
      "grad_norm": 0.007131421472877264,
      "learning_rate": 6.162532745541017e-08,
      "loss": 0.0007,
      "step": 175090
    },
    {
      "epoch": 2.8650904033379696,
      "grad_norm": 0.04102492332458496,
      "learning_rate": 6.147642016873401e-08,
      "loss": 0.0011,
      "step": 175100
    },
    {
      "epoch": 2.865254029289045,
      "grad_norm": 1.4314508438110352,
      "learning_rate": 6.132769189166832e-08,
      "loss": 0.0012,
      "step": 175110
    },
    {
      "epoch": 2.865417655240121,
      "grad_norm": 0.014503179118037224,
      "learning_rate": 6.117914262960433e-08,
      "loss": 0.0009,
      "step": 175120
    },
    {
      "epoch": 2.865581281191197,
      "grad_norm": 0.09569036960601807,
      "learning_rate": 6.103077238792554e-08,
      "loss": 0.0008,
      "step": 175130
    },
    {
      "epoch": 2.8657449071422727,
      "grad_norm": 0.03811396285891533,
      "learning_rate": 6.088258117201151e-08,
      "loss": 0.0008,
      "step": 175140
    },
    {
      "epoch": 2.8659085330933487,
      "grad_norm": 0.09336081147193909,
      "learning_rate": 6.073456898723296e-08,
      "loss": 0.0008,
      "step": 175150
    },
    {
      "epoch": 2.8660721590444247,
      "grad_norm": 0.06616807729005814,
      "learning_rate": 6.058673583895502e-08,
      "loss": 0.0014,
      "step": 175160
    },
    {
      "epoch": 2.8662357849955002,
      "grad_norm": 0.06602644920349121,
      "learning_rate": 6.043908173253621e-08,
      "loss": 0.001,
      "step": 175170
    },
    {
      "epoch": 2.8663994109465762,
      "grad_norm": 0.0740407407283783,
      "learning_rate": 6.029160667332834e-08,
      "loss": 0.0005,
      "step": 175180
    },
    {
      "epoch": 2.8665630368976522,
      "grad_norm": 0.08181922137737274,
      "learning_rate": 6.014431066667826e-08,
      "loss": 0.001,
      "step": 175190
    },
    {
      "epoch": 2.866726662848728,
      "grad_norm": 0.020462464541196823,
      "learning_rate": 5.99971937179239e-08,
      "loss": 0.0005,
      "step": 175200
    },
    {
      "epoch": 2.866890288799804,
      "grad_norm": 0.08246901631355286,
      "learning_rate": 5.985025583239767e-08,
      "loss": 0.002,
      "step": 175210
    },
    {
      "epoch": 2.8670539147508793,
      "grad_norm": 0.12530817091464996,
      "learning_rate": 5.970349701542755e-08,
      "loss": 0.0009,
      "step": 175220
    },
    {
      "epoch": 2.8672175407019553,
      "grad_norm": 0.018198858946561813,
      "learning_rate": 5.9556917272330925e-08,
      "loss": 0.0005,
      "step": 175230
    },
    {
      "epoch": 2.8673811666530313,
      "grad_norm": 0.006078992038965225,
      "learning_rate": 5.9410516608422455e-08,
      "loss": 0.0013,
      "step": 175240
    },
    {
      "epoch": 2.867544792604107,
      "grad_norm": 0.06168961152434349,
      "learning_rate": 5.9264295029008435e-08,
      "loss": 0.0008,
      "step": 175250
    },
    {
      "epoch": 2.867708418555183,
      "grad_norm": 0.028056951239705086,
      "learning_rate": 5.911825253938963e-08,
      "loss": 0.0005,
      "step": 175260
    },
    {
      "epoch": 2.8678720445062584,
      "grad_norm": 0.07744128257036209,
      "learning_rate": 5.897238914485792e-08,
      "loss": 0.0005,
      "step": 175270
    },
    {
      "epoch": 2.8680356704573344,
      "grad_norm": 0.10157930105924606,
      "learning_rate": 5.882670485070186e-08,
      "loss": 0.0007,
      "step": 175280
    },
    {
      "epoch": 2.8681992964084104,
      "grad_norm": 0.7099071741104126,
      "learning_rate": 5.868119966220221e-08,
      "loss": 0.0015,
      "step": 175290
    },
    {
      "epoch": 2.868362922359486,
      "grad_norm": 0.13857851922512054,
      "learning_rate": 5.8535873584633084e-08,
      "loss": 0.0009,
      "step": 175300
    },
    {
      "epoch": 2.868526548310562,
      "grad_norm": 0.20338502526283264,
      "learning_rate": 5.839072662326195e-08,
      "loss": 0.0009,
      "step": 175310
    },
    {
      "epoch": 2.868690174261638,
      "grad_norm": 0.10245632380247116,
      "learning_rate": 5.8245758783349596e-08,
      "loss": 0.0011,
      "step": 175320
    },
    {
      "epoch": 2.8688538002127135,
      "grad_norm": 0.010015835054218769,
      "learning_rate": 5.810097007015181e-08,
      "loss": 0.0006,
      "step": 175330
    },
    {
      "epoch": 2.8690174261637895,
      "grad_norm": 0.10068521648645401,
      "learning_rate": 5.795636048891606e-08,
      "loss": 0.0012,
      "step": 175340
    },
    {
      "epoch": 2.8691810521148655,
      "grad_norm": 0.11369694024324417,
      "learning_rate": 5.781193004488483e-08,
      "loss": 0.0014,
      "step": 175350
    },
    {
      "epoch": 2.869344678065941,
      "grad_norm": 0.08066345006227493,
      "learning_rate": 5.766767874329282e-08,
      "loss": 0.0006,
      "step": 175360
    },
    {
      "epoch": 2.869508304017017,
      "grad_norm": 0.07779251039028168,
      "learning_rate": 5.752360658936862e-08,
      "loss": 0.001,
      "step": 175370
    },
    {
      "epoch": 2.869671929968093,
      "grad_norm": 0.01802058145403862,
      "learning_rate": 5.737971358833472e-08,
      "loss": 0.0004,
      "step": 175380
    },
    {
      "epoch": 2.8698355559191686,
      "grad_norm": 0.05836491659283638,
      "learning_rate": 5.723599974540695e-08,
      "loss": 0.001,
      "step": 175390
    },
    {
      "epoch": 2.8699991818702446,
      "grad_norm": 0.016642700880765915,
      "learning_rate": 5.709246506579447e-08,
      "loss": 0.0004,
      "step": 175400
    },
    {
      "epoch": 2.8701628078213206,
      "grad_norm": 0.03695647791028023,
      "learning_rate": 5.6949109554700345e-08,
      "loss": 0.0008,
      "step": 175410
    },
    {
      "epoch": 2.870326433772396,
      "grad_norm": 0.03990011289715767,
      "learning_rate": 5.6805933217320975e-08,
      "loss": 0.0007,
      "step": 175420
    },
    {
      "epoch": 2.870490059723472,
      "grad_norm": 0.058529507368803024,
      "learning_rate": 5.666293605884499e-08,
      "loss": 0.0005,
      "step": 175430
    },
    {
      "epoch": 2.870653685674548,
      "grad_norm": 0.15740081667900085,
      "learning_rate": 5.652011808445712e-08,
      "loss": 0.0009,
      "step": 175440
    },
    {
      "epoch": 2.8708173116256237,
      "grad_norm": 0.11523908376693726,
      "learning_rate": 5.6377479299333793e-08,
      "loss": 0.0007,
      "step": 175450
    },
    {
      "epoch": 2.8709809375766997,
      "grad_norm": 0.05271280184388161,
      "learning_rate": 5.623501970864531e-08,
      "loss": 0.0007,
      "step": 175460
    },
    {
      "epoch": 2.8711445635277757,
      "grad_norm": 0.05872455984354019,
      "learning_rate": 5.609273931755532e-08,
      "loss": 0.0006,
      "step": 175470
    },
    {
      "epoch": 2.8713081894788512,
      "grad_norm": 0.13023623824119568,
      "learning_rate": 5.595063813122026e-08,
      "loss": 0.001,
      "step": 175480
    },
    {
      "epoch": 2.8714718154299272,
      "grad_norm": 0.06581814587116241,
      "learning_rate": 5.5808716154792666e-08,
      "loss": 0.0007,
      "step": 175490
    },
    {
      "epoch": 2.8716354413810032,
      "grad_norm": 0.5319797992706299,
      "learning_rate": 5.566697339341565e-08,
      "loss": 0.0007,
      "step": 175500
    },
    {
      "epoch": 2.871799067332079,
      "grad_norm": 0.047922633588314056,
      "learning_rate": 5.552540985222788e-08,
      "loss": 0.0005,
      "step": 175510
    },
    {
      "epoch": 2.871962693283155,
      "grad_norm": 0.06529483944177628,
      "learning_rate": 5.538402553636024e-08,
      "loss": 0.0007,
      "step": 175520
    },
    {
      "epoch": 2.872126319234231,
      "grad_norm": 0.1498195230960846,
      "learning_rate": 5.5242820450936984e-08,
      "loss": 0.002,
      "step": 175530
    },
    {
      "epoch": 2.8722899451853063,
      "grad_norm": 0.005223225802183151,
      "learning_rate": 5.51017946010779e-08,
      "loss": 0.0004,
      "step": 175540
    },
    {
      "epoch": 2.8724535711363823,
      "grad_norm": 0.020728344097733498,
      "learning_rate": 5.4960947991893334e-08,
      "loss": 0.0002,
      "step": 175550
    },
    {
      "epoch": 2.8726171970874583,
      "grad_norm": 0.013324741274118423,
      "learning_rate": 5.4820280628489765e-08,
      "loss": 0.0002,
      "step": 175560
    },
    {
      "epoch": 2.872780823038534,
      "grad_norm": 0.02622302435338497,
      "learning_rate": 5.467979251596533e-08,
      "loss": 0.0005,
      "step": 175570
    },
    {
      "epoch": 2.87294444898961,
      "grad_norm": 0.07516628503799438,
      "learning_rate": 5.453948365941264e-08,
      "loss": 0.0011,
      "step": 175580
    },
    {
      "epoch": 2.8731080749406854,
      "grad_norm": 0.05655301362276077,
      "learning_rate": 5.439935406391816e-08,
      "loss": 0.0004,
      "step": 175590
    },
    {
      "epoch": 2.8732717008917614,
      "grad_norm": 0.10728980600833893,
      "learning_rate": 5.425940373456007e-08,
      "loss": 0.0006,
      "step": 175600
    },
    {
      "epoch": 2.8734353268428374,
      "grad_norm": 0.1915968358516693,
      "learning_rate": 5.411963267641207e-08,
      "loss": 0.0008,
      "step": 175610
    },
    {
      "epoch": 2.873598952793913,
      "grad_norm": 0.14025180041790009,
      "learning_rate": 5.398004089453957e-08,
      "loss": 0.0006,
      "step": 175620
    },
    {
      "epoch": 2.873762578744989,
      "grad_norm": 0.002755478722974658,
      "learning_rate": 5.384062839400406e-08,
      "loss": 0.0008,
      "step": 175630
    },
    {
      "epoch": 2.8739262046960645,
      "grad_norm": 0.0776972845196724,
      "learning_rate": 5.3701395179857615e-08,
      "loss": 0.001,
      "step": 175640
    },
    {
      "epoch": 2.8740898306471405,
      "grad_norm": 0.05088486149907112,
      "learning_rate": 5.3562341257147324e-08,
      "loss": 0.0008,
      "step": 175650
    },
    {
      "epoch": 2.8742534565982165,
      "grad_norm": 0.2233416885137558,
      "learning_rate": 5.3423466630914136e-08,
      "loss": 0.0005,
      "step": 175660
    },
    {
      "epoch": 2.874417082549292,
      "grad_norm": 0.038037072867155075,
      "learning_rate": 5.3284771306191254e-08,
      "loss": 0.0009,
      "step": 175670
    },
    {
      "epoch": 2.874580708500368,
      "grad_norm": 0.02905154787003994,
      "learning_rate": 5.314625528800632e-08,
      "loss": 0.0004,
      "step": 175680
    },
    {
      "epoch": 2.874744334451444,
      "grad_norm": 0.027582917362451553,
      "learning_rate": 5.300791858137977e-08,
      "loss": 0.0011,
      "step": 175690
    },
    {
      "epoch": 2.8749079604025196,
      "grad_norm": 0.0049156672321259975,
      "learning_rate": 5.286976119132703e-08,
      "loss": 0.0005,
      "step": 175700
    },
    {
      "epoch": 2.8750715863535956,
      "grad_norm": 0.06905953586101532,
      "learning_rate": 5.2731783122855206e-08,
      "loss": 0.0006,
      "step": 175710
    },
    {
      "epoch": 2.8752352123046716,
      "grad_norm": 0.032916389405727386,
      "learning_rate": 5.25939843809653e-08,
      "loss": 0.0003,
      "step": 175720
    },
    {
      "epoch": 2.875398838255747,
      "grad_norm": 0.06201181933283806,
      "learning_rate": 5.2456364970653317e-08,
      "loss": 0.0006,
      "step": 175730
    },
    {
      "epoch": 2.875562464206823,
      "grad_norm": 0.21089965105056763,
      "learning_rate": 5.2318924896906375e-08,
      "loss": 0.0012,
      "step": 175740
    },
    {
      "epoch": 2.875726090157899,
      "grad_norm": 0.03820250183343887,
      "learning_rate": 5.218166416470771e-08,
      "loss": 0.0008,
      "step": 175750
    },
    {
      "epoch": 2.8758897161089747,
      "grad_norm": 0.028971681371331215,
      "learning_rate": 5.204458277903168e-08,
      "loss": 0.0005,
      "step": 175760
    },
    {
      "epoch": 2.8760533420600507,
      "grad_norm": 0.07568809390068054,
      "learning_rate": 5.1907680744847086e-08,
      "loss": 0.0007,
      "step": 175770
    },
    {
      "epoch": 2.8762169680111267,
      "grad_norm": 0.28838589787483215,
      "learning_rate": 5.1770958067117184e-08,
      "loss": 0.0011,
      "step": 175780
    },
    {
      "epoch": 2.8763805939622022,
      "grad_norm": 0.07241639494895935,
      "learning_rate": 5.163441475079634e-08,
      "loss": 0.0005,
      "step": 175790
    },
    {
      "epoch": 2.8765442199132782,
      "grad_norm": 0.06779029220342636,
      "learning_rate": 5.1498050800835583e-08,
      "loss": 0.0009,
      "step": 175800
    },
    {
      "epoch": 2.8767078458643542,
      "grad_norm": 0.08997996151447296,
      "learning_rate": 5.136186622217709e-08,
      "loss": 0.0004,
      "step": 175810
    },
    {
      "epoch": 2.87687147181543,
      "grad_norm": 0.07818464189767838,
      "learning_rate": 5.1225861019757464e-08,
      "loss": 0.0007,
      "step": 175820
    },
    {
      "epoch": 2.877035097766506,
      "grad_norm": 0.08132357150316238,
      "learning_rate": 5.109003519850553e-08,
      "loss": 0.001,
      "step": 175830
    },
    {
      "epoch": 2.877198723717582,
      "grad_norm": 0.2596634328365326,
      "learning_rate": 5.095438876334624e-08,
      "loss": 0.0009,
      "step": 175840
    },
    {
      "epoch": 2.8773623496686573,
      "grad_norm": 0.07025530189275742,
      "learning_rate": 5.081892171919511e-08,
      "loss": 0.0006,
      "step": 175850
    },
    {
      "epoch": 2.8775259756197333,
      "grad_norm": 0.01663915440440178,
      "learning_rate": 5.068363407096322e-08,
      "loss": 0.0005,
      "step": 175860
    },
    {
      "epoch": 2.8776896015708093,
      "grad_norm": 0.04181954264640808,
      "learning_rate": 5.054852582355441e-08,
      "loss": 0.0011,
      "step": 175870
    },
    {
      "epoch": 2.877853227521885,
      "grad_norm": 0.0700729638338089,
      "learning_rate": 5.041359698186532e-08,
      "loss": 0.0005,
      "step": 175880
    },
    {
      "epoch": 2.878016853472961,
      "grad_norm": 0.11738396435976028,
      "learning_rate": 5.0278847550787605e-08,
      "loss": 0.0003,
      "step": 175890
    },
    {
      "epoch": 2.878180479424037,
      "grad_norm": 0.14627595245838165,
      "learning_rate": 5.014427753520512e-08,
      "loss": 0.0013,
      "step": 175900
    },
    {
      "epoch": 2.8783441053751124,
      "grad_norm": 0.05835794284939766,
      "learning_rate": 5.00098869399962e-08,
      "loss": 0.0005,
      "step": 175910
    },
    {
      "epoch": 2.8785077313261884,
      "grad_norm": 0.03929300233721733,
      "learning_rate": 4.9875675770031937e-08,
      "loss": 0.0008,
      "step": 175920
    },
    {
      "epoch": 2.8786713572772644,
      "grad_norm": 0.13615332543849945,
      "learning_rate": 4.974164403017678e-08,
      "loss": 0.0007,
      "step": 175930
    },
    {
      "epoch": 2.87883498322834,
      "grad_norm": 0.05815568193793297,
      "learning_rate": 4.960779172528962e-08,
      "loss": 0.0006,
      "step": 175940
    },
    {
      "epoch": 2.878998609179416,
      "grad_norm": 0.08804910629987717,
      "learning_rate": 4.947411886022213e-08,
      "loss": 0.0005,
      "step": 175950
    },
    {
      "epoch": 2.879162235130492,
      "grad_norm": 0.005042679142206907,
      "learning_rate": 4.934062543981988e-08,
      "loss": 0.0009,
      "step": 175960
    },
    {
      "epoch": 2.8793258610815675,
      "grad_norm": 0.06677313894033432,
      "learning_rate": 4.920731146892066e-08,
      "loss": 0.0005,
      "step": 175970
    },
    {
      "epoch": 2.8794894870326435,
      "grad_norm": 0.16244550049304962,
      "learning_rate": 4.907417695235783e-08,
      "loss": 0.001,
      "step": 175980
    },
    {
      "epoch": 2.879653112983719,
      "grad_norm": 0.11529611051082611,
      "learning_rate": 4.894122189495698e-08,
      "loss": 0.0011,
      "step": 175990
    },
    {
      "epoch": 2.879816738934795,
      "grad_norm": 0.027447983622550964,
      "learning_rate": 4.880844630153647e-08,
      "loss": 0.0007,
      "step": 176000
    },
    {
      "epoch": 2.879980364885871,
      "grad_norm": 0.07845830172300339,
      "learning_rate": 4.867585017691079e-08,
      "loss": 0.0007,
      "step": 176010
    },
    {
      "epoch": 2.8801439908369466,
      "grad_norm": 0.03748183697462082,
      "learning_rate": 4.8543433525884974e-08,
      "loss": 0.0004,
      "step": 176020
    },
    {
      "epoch": 2.8803076167880226,
      "grad_norm": 0.26329025626182556,
      "learning_rate": 4.8411196353259636e-08,
      "loss": 0.0012,
      "step": 176030
    },
    {
      "epoch": 2.880471242739098,
      "grad_norm": 0.07574775069952011,
      "learning_rate": 4.827913866382705e-08,
      "loss": 0.0015,
      "step": 176040
    },
    {
      "epoch": 2.880634868690174,
      "grad_norm": 0.04075975716114044,
      "learning_rate": 4.8147260462375054e-08,
      "loss": 0.0004,
      "step": 176050
    },
    {
      "epoch": 2.88079849464125,
      "grad_norm": 0.00475145922973752,
      "learning_rate": 4.801556175368316e-08,
      "loss": 0.0007,
      "step": 176060
    },
    {
      "epoch": 2.8809621205923257,
      "grad_norm": 0.04693582281470299,
      "learning_rate": 4.788404254252477e-08,
      "loss": 0.0005,
      "step": 176070
    },
    {
      "epoch": 2.8811257465434017,
      "grad_norm": 0.11720842868089676,
      "learning_rate": 4.775270283366884e-08,
      "loss": 0.001,
      "step": 176080
    },
    {
      "epoch": 2.8812893724944777,
      "grad_norm": 0.1877339780330658,
      "learning_rate": 4.762154263187435e-08,
      "loss": 0.0006,
      "step": 176090
    },
    {
      "epoch": 2.8814529984455532,
      "grad_norm": 0.059261370450258255,
      "learning_rate": 4.749056194189638e-08,
      "loss": 0.0006,
      "step": 176100
    },
    {
      "epoch": 2.8816166243966292,
      "grad_norm": 0.05121181905269623,
      "learning_rate": 4.735976076848225e-08,
      "loss": 0.0016,
      "step": 176110
    },
    {
      "epoch": 2.8817802503477052,
      "grad_norm": 0.09191296249628067,
      "learning_rate": 4.7229139116373725e-08,
      "loss": 0.0004,
      "step": 176120
    },
    {
      "epoch": 2.881943876298781,
      "grad_norm": 0.15236833691596985,
      "learning_rate": 4.7098696990304785e-08,
      "loss": 0.0004,
      "step": 176130
    },
    {
      "epoch": 2.882107502249857,
      "grad_norm": 0.046635061502456665,
      "learning_rate": 4.6968434395004425e-08,
      "loss": 0.0009,
      "step": 176140
    },
    {
      "epoch": 2.882271128200933,
      "grad_norm": 0.08995907008647919,
      "learning_rate": 4.6838351335193875e-08,
      "loss": 0.0008,
      "step": 176150
    },
    {
      "epoch": 2.8824347541520083,
      "grad_norm": 0.22900702059268951,
      "learning_rate": 4.670844781558881e-08,
      "loss": 0.0008,
      "step": 176160
    },
    {
      "epoch": 2.8825983801030843,
      "grad_norm": 0.07879273593425751,
      "learning_rate": 4.657872384089768e-08,
      "loss": 0.0008,
      "step": 176170
    },
    {
      "epoch": 2.8827620060541603,
      "grad_norm": 0.06937658786773682,
      "learning_rate": 4.644917941582227e-08,
      "loss": 0.0007,
      "step": 176180
    },
    {
      "epoch": 2.882925632005236,
      "grad_norm": 0.05095773935317993,
      "learning_rate": 4.6319814545058847e-08,
      "loss": 0.0004,
      "step": 176190
    },
    {
      "epoch": 2.883089257956312,
      "grad_norm": 0.009493554942309856,
      "learning_rate": 4.6190629233295867e-08,
      "loss": 0.0013,
      "step": 176200
    },
    {
      "epoch": 2.883252883907388,
      "grad_norm": 0.06941714882850647,
      "learning_rate": 4.6061623485216254e-08,
      "loss": 0.0003,
      "step": 176210
    },
    {
      "epoch": 2.8834165098584634,
      "grad_norm": 0.008341421373188496,
      "learning_rate": 4.593279730549682e-08,
      "loss": 0.0006,
      "step": 176220
    },
    {
      "epoch": 2.8835801358095394,
      "grad_norm": 0.08923845738172531,
      "learning_rate": 4.580415069880606e-08,
      "loss": 0.001,
      "step": 176230
    },
    {
      "epoch": 2.8837437617606154,
      "grad_norm": 0.03379319980740547,
      "learning_rate": 4.567568366980857e-08,
      "loss": 0.0005,
      "step": 176240
    },
    {
      "epoch": 2.883907387711691,
      "grad_norm": 0.06666112691164017,
      "learning_rate": 4.554739622315951e-08,
      "loss": 0.0009,
      "step": 176250
    },
    {
      "epoch": 2.884071013662767,
      "grad_norm": 0.32875609397888184,
      "learning_rate": 4.5419288363510175e-08,
      "loss": 0.0007,
      "step": 176260
    },
    {
      "epoch": 2.884234639613843,
      "grad_norm": 0.07230079919099808,
      "learning_rate": 4.52913600955035e-08,
      "loss": 0.0008,
      "step": 176270
    },
    {
      "epoch": 2.8843982655649185,
      "grad_norm": 0.14306780695915222,
      "learning_rate": 4.5163611423776346e-08,
      "loss": 0.0013,
      "step": 176280
    },
    {
      "epoch": 2.8845618915159945,
      "grad_norm": 0.09511516243219376,
      "learning_rate": 4.503604235296e-08,
      "loss": 0.0006,
      "step": 176290
    },
    {
      "epoch": 2.8847255174670705,
      "grad_norm": 0.04371592029929161,
      "learning_rate": 4.490865288767743e-08,
      "loss": 0.0008,
      "step": 176300
    },
    {
      "epoch": 2.884889143418146,
      "grad_norm": 0.2356683611869812,
      "learning_rate": 4.4781443032547725e-08,
      "loss": 0.0007,
      "step": 176310
    },
    {
      "epoch": 2.885052769369222,
      "grad_norm": 0.2509373724460602,
      "learning_rate": 4.465441279218108e-08,
      "loss": 0.0007,
      "step": 176320
    },
    {
      "epoch": 2.885216395320298,
      "grad_norm": 0.02567668817937374,
      "learning_rate": 4.452756217118215e-08,
      "loss": 0.0003,
      "step": 176330
    },
    {
      "epoch": 2.8853800212713736,
      "grad_norm": 0.0780859887599945,
      "learning_rate": 4.440089117414836e-08,
      "loss": 0.0005,
      "step": 176340
    },
    {
      "epoch": 2.8855436472224496,
      "grad_norm": 0.008311559446156025,
      "learning_rate": 4.4274399805672164e-08,
      "loss": 0.0005,
      "step": 176350
    },
    {
      "epoch": 2.885707273173525,
      "grad_norm": 0.09923581779003143,
      "learning_rate": 4.414808807033821e-08,
      "loss": 0.0013,
      "step": 176360
    },
    {
      "epoch": 2.885870899124601,
      "grad_norm": 0.06938090175390244,
      "learning_rate": 4.402195597272507e-08,
      "loss": 0.0004,
      "step": 176370
    },
    {
      "epoch": 2.886034525075677,
      "grad_norm": 0.06059519946575165,
      "learning_rate": 4.389600351740464e-08,
      "loss": 0.0004,
      "step": 176380
    },
    {
      "epoch": 2.8861981510267527,
      "grad_norm": 0.05628759786486626,
      "learning_rate": 4.377023070894215e-08,
      "loss": 0.0005,
      "step": 176390
    },
    {
      "epoch": 2.8863617769778287,
      "grad_norm": 0.043270789086818695,
      "learning_rate": 4.36446375518973e-08,
      "loss": 0.0008,
      "step": 176400
    },
    {
      "epoch": 2.8865254029289042,
      "grad_norm": 0.05593279376626015,
      "learning_rate": 4.351922405082143e-08,
      "loss": 0.0004,
      "step": 176410
    },
    {
      "epoch": 2.8866890288799802,
      "grad_norm": 0.015211150981485844,
      "learning_rate": 4.339399021026147e-08,
      "loss": 0.0011,
      "step": 176420
    },
    {
      "epoch": 2.8868526548310562,
      "grad_norm": 0.12670786678791046,
      "learning_rate": 4.326893603475657e-08,
      "loss": 0.0009,
      "step": 176430
    },
    {
      "epoch": 2.887016280782132,
      "grad_norm": 0.07527585327625275,
      "learning_rate": 4.314406152883921e-08,
      "loss": 0.0005,
      "step": 176440
    },
    {
      "epoch": 2.887179906733208,
      "grad_norm": 0.013026727363467216,
      "learning_rate": 4.3019366697036323e-08,
      "loss": 0.0008,
      "step": 176450
    },
    {
      "epoch": 2.887343532684284,
      "grad_norm": 0.17186667025089264,
      "learning_rate": 4.289485154386763e-08,
      "loss": 0.0007,
      "step": 176460
    },
    {
      "epoch": 2.8875071586353593,
      "grad_norm": 0.11464191228151321,
      "learning_rate": 4.2770516073846745e-08,
      "loss": 0.0009,
      "step": 176470
    },
    {
      "epoch": 2.8876707845864353,
      "grad_norm": 0.1785428375005722,
      "learning_rate": 4.264636029148006e-08,
      "loss": 0.001,
      "step": 176480
    },
    {
      "epoch": 2.8878344105375113,
      "grad_norm": 0.020543817430734634,
      "learning_rate": 4.2522384201267865e-08,
      "loss": 0.0007,
      "step": 176490
    },
    {
      "epoch": 2.887998036488587,
      "grad_norm": 0.10109040886163712,
      "learning_rate": 4.239858780770434e-08,
      "loss": 0.0005,
      "step": 176500
    },
    {
      "epoch": 2.888161662439663,
      "grad_norm": 0.1169227659702301,
      "learning_rate": 4.227497111527645e-08,
      "loss": 0.0007,
      "step": 176510
    },
    {
      "epoch": 2.888325288390739,
      "grad_norm": 0.1300945281982422,
      "learning_rate": 4.2151534128466175e-08,
      "loss": 0.001,
      "step": 176520
    },
    {
      "epoch": 2.8884889143418144,
      "grad_norm": 0.09155488014221191,
      "learning_rate": 4.202827685174604e-08,
      "loss": 0.0007,
      "step": 176530
    },
    {
      "epoch": 2.8886525402928904,
      "grad_norm": 0.13801400363445282,
      "learning_rate": 4.190519928958525e-08,
      "loss": 0.0005,
      "step": 176540
    },
    {
      "epoch": 2.8888161662439664,
      "grad_norm": 0.053257450461387634,
      "learning_rate": 4.178230144644413e-08,
      "loss": 0.0006,
      "step": 176550
    },
    {
      "epoch": 2.888979792195042,
      "grad_norm": 0.018202928826212883,
      "learning_rate": 4.1659583326778e-08,
      "loss": 0.0006,
      "step": 176560
    },
    {
      "epoch": 2.889143418146118,
      "grad_norm": 0.026498859748244286,
      "learning_rate": 4.153704493503496e-08,
      "loss": 0.0014,
      "step": 176570
    },
    {
      "epoch": 2.889307044097194,
      "grad_norm": 0.17384694516658783,
      "learning_rate": 4.141468627565648e-08,
      "loss": 0.0007,
      "step": 176580
    },
    {
      "epoch": 2.8894706700482695,
      "grad_norm": 0.019504738971590996,
      "learning_rate": 4.1292507353078434e-08,
      "loss": 0.0011,
      "step": 176590
    },
    {
      "epoch": 2.8896342959993455,
      "grad_norm": 0.0477384589612484,
      "learning_rate": 4.1170508171728407e-08,
      "loss": 0.0005,
      "step": 176600
    },
    {
      "epoch": 2.8897979219504215,
      "grad_norm": 0.13389673829078674,
      "learning_rate": 4.104868873603007e-08,
      "loss": 0.0008,
      "step": 176610
    },
    {
      "epoch": 2.889961547901497,
      "grad_norm": 0.10623396933078766,
      "learning_rate": 4.0927049050397685e-08,
      "loss": 0.0005,
      "step": 176620
    },
    {
      "epoch": 2.890125173852573,
      "grad_norm": 0.06848946958780289,
      "learning_rate": 4.080558911924104e-08,
      "loss": 0.0005,
      "step": 176630
    },
    {
      "epoch": 2.890288799803649,
      "grad_norm": 0.03842702507972717,
      "learning_rate": 4.068430894696218e-08,
      "loss": 0.0006,
      "step": 176640
    },
    {
      "epoch": 2.8904524257547246,
      "grad_norm": 0.024883877485990524,
      "learning_rate": 4.0563208537958146e-08,
      "loss": 0.0006,
      "step": 176650
    },
    {
      "epoch": 2.8906160517058006,
      "grad_norm": 0.18561606109142303,
      "learning_rate": 4.044228789661819e-08,
      "loss": 0.0008,
      "step": 176660
    },
    {
      "epoch": 2.8907796776568766,
      "grad_norm": 0.09274275600910187,
      "learning_rate": 4.0321547027325495e-08,
      "loss": 0.0005,
      "step": 176670
    },
    {
      "epoch": 2.890943303607952,
      "grad_norm": 0.11740662902593613,
      "learning_rate": 4.020098593445654e-08,
      "loss": 0.0007,
      "step": 176680
    },
    {
      "epoch": 2.891106929559028,
      "grad_norm": 0.013774445280432701,
      "learning_rate": 4.008060462238061e-08,
      "loss": 0.0009,
      "step": 176690
    },
    {
      "epoch": 2.891270555510104,
      "grad_norm": 0.03898048773407936,
      "learning_rate": 3.9960403095462567e-08,
      "loss": 0.0005,
      "step": 176700
    },
    {
      "epoch": 2.8914341814611797,
      "grad_norm": 0.0460006482899189,
      "learning_rate": 3.984038135805835e-08,
      "loss": 0.0006,
      "step": 176710
    },
    {
      "epoch": 2.8915978074122557,
      "grad_norm": 0.12517037987709045,
      "learning_rate": 3.972053941451892e-08,
      "loss": 0.0003,
      "step": 176720
    },
    {
      "epoch": 2.8917614333633317,
      "grad_norm": 0.0437329038977623,
      "learning_rate": 3.9600877269188595e-08,
      "loss": 0.0005,
      "step": 176730
    },
    {
      "epoch": 2.8919250593144072,
      "grad_norm": 0.144351065158844,
      "learning_rate": 3.9481394926403906e-08,
      "loss": 0.0008,
      "step": 176740
    },
    {
      "epoch": 2.8920886852654832,
      "grad_norm": 0.14014112949371338,
      "learning_rate": 3.9362092390496934e-08,
      "loss": 0.0013,
      "step": 176750
    },
    {
      "epoch": 2.892252311216559,
      "grad_norm": 0.025197744369506836,
      "learning_rate": 3.924296966579089e-08,
      "loss": 0.001,
      "step": 176760
    },
    {
      "epoch": 2.892415937167635,
      "grad_norm": 0.03783867135643959,
      "learning_rate": 3.912402675660454e-08,
      "loss": 0.0011,
      "step": 176770
    },
    {
      "epoch": 2.892579563118711,
      "grad_norm": 0.10282964259386063,
      "learning_rate": 3.9005263667249435e-08,
      "loss": 0.0006,
      "step": 176780
    },
    {
      "epoch": 2.8927431890697863,
      "grad_norm": 0.023211972787976265,
      "learning_rate": 3.888668040202992e-08,
      "loss": 0.0005,
      "step": 176790
    },
    {
      "epoch": 2.8929068150208623,
      "grad_norm": 0.057641055434942245,
      "learning_rate": 3.876827696524421e-08,
      "loss": 0.0004,
      "step": 176800
    },
    {
      "epoch": 2.893070440971938,
      "grad_norm": 0.08486399054527283,
      "learning_rate": 3.865005336118499e-08,
      "loss": 0.0008,
      "step": 176810
    },
    {
      "epoch": 2.893234066923014,
      "grad_norm": 0.031157856807112694,
      "learning_rate": 3.853200959413661e-08,
      "loss": 0.0005,
      "step": 176820
    },
    {
      "epoch": 2.89339769287409,
      "grad_norm": 0.005268704146146774,
      "learning_rate": 3.8414145668378425e-08,
      "loss": 0.0004,
      "step": 176830
    },
    {
      "epoch": 2.8935613188251654,
      "grad_norm": 0.16689620912075043,
      "learning_rate": 3.829646158818256e-08,
      "loss": 0.0008,
      "step": 176840
    },
    {
      "epoch": 2.8937249447762414,
      "grad_norm": 0.13241204619407654,
      "learning_rate": 3.817895735781507e-08,
      "loss": 0.0005,
      "step": 176850
    },
    {
      "epoch": 2.8938885707273174,
      "grad_norm": 0.15282315015792847,
      "learning_rate": 3.8061632981534204e-08,
      "loss": 0.0008,
      "step": 176860
    },
    {
      "epoch": 2.894052196678393,
      "grad_norm": 0.2884654402732849,
      "learning_rate": 3.794448846359433e-08,
      "loss": 0.0017,
      "step": 176870
    },
    {
      "epoch": 2.894215822629469,
      "grad_norm": 0.03419988229870796,
      "learning_rate": 3.7827523808240396e-08,
      "loss": 0.0006,
      "step": 176880
    },
    {
      "epoch": 2.894379448580545,
      "grad_norm": 0.05453098192811012,
      "learning_rate": 3.77107390197129e-08,
      "loss": 0.0022,
      "step": 176890
    },
    {
      "epoch": 2.8945430745316205,
      "grad_norm": 0.002680734498426318,
      "learning_rate": 3.759413410224399e-08,
      "loss": 0.0004,
      "step": 176900
    },
    {
      "epoch": 2.8947067004826965,
      "grad_norm": 0.048047974705696106,
      "learning_rate": 3.747770906006087e-08,
      "loss": 0.0004,
      "step": 176910
    },
    {
      "epoch": 2.8948703264337725,
      "grad_norm": 0.005502645391970873,
      "learning_rate": 3.7361463897384045e-08,
      "loss": 0.0005,
      "step": 176920
    },
    {
      "epoch": 2.895033952384848,
      "grad_norm": 0.006893428508192301,
      "learning_rate": 3.724539861842624e-08,
      "loss": 0.0006,
      "step": 176930
    },
    {
      "epoch": 2.895197578335924,
      "grad_norm": 0.01869441196322441,
      "learning_rate": 3.712951322739578e-08,
      "loss": 0.0012,
      "step": 176940
    },
    {
      "epoch": 2.895361204287,
      "grad_norm": 0.04699801653623581,
      "learning_rate": 3.7013807728492054e-08,
      "loss": 0.0007,
      "step": 176950
    },
    {
      "epoch": 2.8955248302380756,
      "grad_norm": 0.04201870411634445,
      "learning_rate": 3.689828212590951e-08,
      "loss": 0.0005,
      "step": 176960
    },
    {
      "epoch": 2.8956884561891516,
      "grad_norm": 0.047326136380434036,
      "learning_rate": 3.67829364238359e-08,
      "loss": 0.0006,
      "step": 176970
    },
    {
      "epoch": 2.8958520821402276,
      "grad_norm": 0.02309175580739975,
      "learning_rate": 3.666777062645233e-08,
      "loss": 0.0008,
      "step": 176980
    },
    {
      "epoch": 2.896015708091303,
      "grad_norm": 0.06949126720428467,
      "learning_rate": 3.655278473793267e-08,
      "loss": 0.0007,
      "step": 176990
    },
    {
      "epoch": 2.896179334042379,
      "grad_norm": 0.1710657924413681,
      "learning_rate": 3.643797876244526e-08,
      "loss": 0.0009,
      "step": 177000
    },
    {
      "epoch": 2.896342959993455,
      "grad_norm": 0.1285581886768341,
      "learning_rate": 3.632335270415177e-08,
      "loss": 0.0006,
      "step": 177010
    },
    {
      "epoch": 2.8965065859445307,
      "grad_norm": 0.0462639145553112,
      "learning_rate": 3.620890656720666e-08,
      "loss": 0.0007,
      "step": 177020
    },
    {
      "epoch": 2.8966702118956067,
      "grad_norm": 0.06266423314809799,
      "learning_rate": 3.609464035575827e-08,
      "loss": 0.0006,
      "step": 177030
    },
    {
      "epoch": 2.8968338378466827,
      "grad_norm": 0.2848026752471924,
      "learning_rate": 3.598055407394885e-08,
      "loss": 0.0007,
      "step": 177040
    },
    {
      "epoch": 2.8969974637977582,
      "grad_norm": 0.12045657634735107,
      "learning_rate": 3.5866647725913975e-08,
      "loss": 0.0005,
      "step": 177050
    },
    {
      "epoch": 2.8971610897488342,
      "grad_norm": 0.06083511933684349,
      "learning_rate": 3.575292131578201e-08,
      "loss": 0.0008,
      "step": 177060
    },
    {
      "epoch": 2.8973247156999102,
      "grad_norm": 0.14872334897518158,
      "learning_rate": 3.563937484767521e-08,
      "loss": 0.0012,
      "step": 177070
    },
    {
      "epoch": 2.897488341650986,
      "grad_norm": 0.05737492814660072,
      "learning_rate": 3.552600832570974e-08,
      "loss": 0.0006,
      "step": 177080
    },
    {
      "epoch": 2.897651967602062,
      "grad_norm": 0.12255825847387314,
      "learning_rate": 3.541282175399452e-08,
      "loss": 0.0012,
      "step": 177090
    },
    {
      "epoch": 2.897815593553138,
      "grad_norm": 0.0018714523175731301,
      "learning_rate": 3.529981513663294e-08,
      "loss": 0.0005,
      "step": 177100
    },
    {
      "epoch": 2.8979792195042133,
      "grad_norm": 0.0896361842751503,
      "learning_rate": 3.518698847772006e-08,
      "loss": 0.0006,
      "step": 177110
    },
    {
      "epoch": 2.8981428454552893,
      "grad_norm": 0.017417022958397865,
      "learning_rate": 3.50743417813465e-08,
      "loss": 0.0007,
      "step": 177120
    },
    {
      "epoch": 2.8983064714063653,
      "grad_norm": 0.05408822372555733,
      "learning_rate": 3.4961875051595095e-08,
      "loss": 0.001,
      "step": 177130
    },
    {
      "epoch": 2.898470097357441,
      "grad_norm": 0.11233150213956833,
      "learning_rate": 3.4849588292542593e-08,
      "loss": 0.001,
      "step": 177140
    },
    {
      "epoch": 2.898633723308517,
      "grad_norm": 0.0716085135936737,
      "learning_rate": 3.473748150825906e-08,
      "loss": 0.0007,
      "step": 177150
    },
    {
      "epoch": 2.8987973492595924,
      "grad_norm": 0.0025160168297588825,
      "learning_rate": 3.462555470280793e-08,
      "loss": 0.0005,
      "step": 177160
    },
    {
      "epoch": 2.8989609752106684,
      "grad_norm": 0.010875099338591099,
      "learning_rate": 3.451380788024705e-08,
      "loss": 0.0005,
      "step": 177170
    },
    {
      "epoch": 2.8991246011617444,
      "grad_norm": 0.09221598505973816,
      "learning_rate": 3.440224104462597e-08,
      "loss": 0.0003,
      "step": 177180
    },
    {
      "epoch": 2.89928822711282,
      "grad_norm": 0.1053747832775116,
      "learning_rate": 3.429085419998923e-08,
      "loss": 0.0007,
      "step": 177190
    },
    {
      "epoch": 2.899451853063896,
      "grad_norm": 0.06489173322916031,
      "learning_rate": 3.417964735037471e-08,
      "loss": 0.001,
      "step": 177200
    },
    {
      "epoch": 2.8996154790149715,
      "grad_norm": 0.07743917405605316,
      "learning_rate": 3.4068620499811964e-08,
      "loss": 0.001,
      "step": 177210
    },
    {
      "epoch": 2.8997791049660475,
      "grad_norm": 0.0979137271642685,
      "learning_rate": 3.395777365232722e-08,
      "loss": 0.0006,
      "step": 177220
    },
    {
      "epoch": 2.8999427309171235,
      "grad_norm": 0.11220269650220871,
      "learning_rate": 3.384710681193726e-08,
      "loss": 0.0008,
      "step": 177230
    },
    {
      "epoch": 2.900106356868199,
      "grad_norm": 0.06027612462639809,
      "learning_rate": 3.3736619982653875e-08,
      "loss": 0.0005,
      "step": 177240
    },
    {
      "epoch": 2.900269982819275,
      "grad_norm": 0.06976976245641708,
      "learning_rate": 3.3626313168482195e-08,
      "loss": 0.0005,
      "step": 177250
    },
    {
      "epoch": 2.900433608770351,
      "grad_norm": 0.1064118966460228,
      "learning_rate": 3.351618637342013e-08,
      "loss": 0.0006,
      "step": 177260
    },
    {
      "epoch": 2.9005972347214266,
      "grad_norm": 0.05113185942173004,
      "learning_rate": 3.34062396014595e-08,
      "loss": 0.0008,
      "step": 177270
    },
    {
      "epoch": 2.9007608606725026,
      "grad_norm": 0.04026055335998535,
      "learning_rate": 3.329647285658599e-08,
      "loss": 0.0003,
      "step": 177280
    },
    {
      "epoch": 2.9009244866235786,
      "grad_norm": 0.17570102214813232,
      "learning_rate": 3.31868861427781e-08,
      "loss": 0.0006,
      "step": 177290
    },
    {
      "epoch": 2.901088112574654,
      "grad_norm": 0.07997167855501175,
      "learning_rate": 3.307747946400819e-08,
      "loss": 0.0011,
      "step": 177300
    },
    {
      "epoch": 2.90125173852573,
      "grad_norm": 0.031603582203388214,
      "learning_rate": 3.2968252824241986e-08,
      "loss": 0.0006,
      "step": 177310
    },
    {
      "epoch": 2.901415364476806,
      "grad_norm": 0.018800415098667145,
      "learning_rate": 3.2859206227438547e-08,
      "loss": 0.0004,
      "step": 177320
    },
    {
      "epoch": 2.9015789904278817,
      "grad_norm": 0.06749694049358368,
      "learning_rate": 3.275033967755081e-08,
      "loss": 0.0006,
      "step": 177330
    },
    {
      "epoch": 2.9017426163789577,
      "grad_norm": 0.15386687219142914,
      "learning_rate": 3.264165317852508e-08,
      "loss": 0.0037,
      "step": 177340
    },
    {
      "epoch": 2.9019062423300337,
      "grad_norm": 0.05814754217863083,
      "learning_rate": 3.253314673429986e-08,
      "loss": 0.0008,
      "step": 177350
    },
    {
      "epoch": 2.9020698682811092,
      "grad_norm": 0.030864736065268517,
      "learning_rate": 3.2424820348809785e-08,
      "loss": 0.0007,
      "step": 177360
    },
    {
      "epoch": 2.9022334942321852,
      "grad_norm": 0.10367145389318466,
      "learning_rate": 3.23166740259806e-08,
      "loss": 0.001,
      "step": 177370
    },
    {
      "epoch": 2.9023971201832612,
      "grad_norm": 0.07243552803993225,
      "learning_rate": 3.2208707769732505e-08,
      "loss": 0.0007,
      "step": 177380
    },
    {
      "epoch": 2.902560746134337,
      "grad_norm": 0.012534837238490582,
      "learning_rate": 3.210092158397904e-08,
      "loss": 0.0008,
      "step": 177390
    },
    {
      "epoch": 2.902724372085413,
      "grad_norm": 0.06368628889322281,
      "learning_rate": 3.199331547262707e-08,
      "loss": 0.0008,
      "step": 177400
    },
    {
      "epoch": 2.902887998036489,
      "grad_norm": 0.010164530947804451,
      "learning_rate": 3.188588943957682e-08,
      "loss": 0.001,
      "step": 177410
    },
    {
      "epoch": 2.9030516239875643,
      "grad_norm": 0.035111453384160995,
      "learning_rate": 3.177864348872239e-08,
      "loss": 0.0006,
      "step": 177420
    },
    {
      "epoch": 2.9032152499386403,
      "grad_norm": 0.03125438094139099,
      "learning_rate": 3.167157762395179e-08,
      "loss": 0.0007,
      "step": 177430
    },
    {
      "epoch": 2.9033788758897163,
      "grad_norm": 0.04680990055203438,
      "learning_rate": 3.156469184914523e-08,
      "loss": 0.0005,
      "step": 177440
    },
    {
      "epoch": 2.903542501840792,
      "grad_norm": 0.06495669484138489,
      "learning_rate": 3.145798616817741e-08,
      "loss": 0.0008,
      "step": 177450
    },
    {
      "epoch": 2.903706127791868,
      "grad_norm": 0.003417240222916007,
      "learning_rate": 3.1351460584915784e-08,
      "loss": 0.0003,
      "step": 177460
    },
    {
      "epoch": 2.903869753742944,
      "grad_norm": 0.03974118083715439,
      "learning_rate": 3.1245115103222254e-08,
      "loss": 0.0008,
      "step": 177470
    },
    {
      "epoch": 2.9040333796940194,
      "grad_norm": 0.0995679646730423,
      "learning_rate": 3.113894972695097e-08,
      "loss": 0.0008,
      "step": 177480
    },
    {
      "epoch": 2.9041970056450954,
      "grad_norm": 0.2517743706703186,
      "learning_rate": 3.103296445995052e-08,
      "loss": 0.0008,
      "step": 177490
    },
    {
      "epoch": 2.9043606315961714,
      "grad_norm": 0.01019982434809208,
      "learning_rate": 3.092715930606227e-08,
      "loss": 0.0003,
      "step": 177500
    },
    {
      "epoch": 2.904524257547247,
      "grad_norm": 0.020068984478712082,
      "learning_rate": 3.082153426912204e-08,
      "loss": 0.0007,
      "step": 177510
    },
    {
      "epoch": 2.904687883498323,
      "grad_norm": 0.08207608014345169,
      "learning_rate": 3.071608935295789e-08,
      "loss": 0.0004,
      "step": 177520
    },
    {
      "epoch": 2.9048515094493985,
      "grad_norm": 0.08420080691576004,
      "learning_rate": 3.061082456139175e-08,
      "loss": 0.001,
      "step": 177530
    },
    {
      "epoch": 2.9050151354004745,
      "grad_norm": 0.07505779713392258,
      "learning_rate": 3.050573989824057e-08,
      "loss": 0.0004,
      "step": 177540
    },
    {
      "epoch": 2.9051787613515505,
      "grad_norm": 0.09097757190465927,
      "learning_rate": 3.040083536731131e-08,
      "loss": 0.0005,
      "step": 177550
    },
    {
      "epoch": 2.905342387302626,
      "grad_norm": 0.08892517536878586,
      "learning_rate": 3.02961109724087e-08,
      "loss": 0.0007,
      "step": 177560
    },
    {
      "epoch": 2.905506013253702,
      "grad_norm": 0.041635241359472275,
      "learning_rate": 3.019156671732693e-08,
      "loss": 0.0004,
      "step": 177570
    },
    {
      "epoch": 2.9056696392047776,
      "grad_norm": 0.012195191346108913,
      "learning_rate": 3.008720260585629e-08,
      "loss": 0.0006,
      "step": 177580
    },
    {
      "epoch": 2.9058332651558536,
      "grad_norm": 0.026278575882315636,
      "learning_rate": 2.9983018641780437e-08,
      "loss": 0.0004,
      "step": 177590
    },
    {
      "epoch": 2.9059968911069296,
      "grad_norm": 0.13457244634628296,
      "learning_rate": 2.987901482887412e-08,
      "loss": 0.0033,
      "step": 177600
    },
    {
      "epoch": 2.906160517058005,
      "grad_norm": 0.04808173328638077,
      "learning_rate": 2.9775191170908768e-08,
      "loss": 0.0005,
      "step": 177610
    },
    {
      "epoch": 2.906324143009081,
      "grad_norm": 0.08267733454704285,
      "learning_rate": 2.9671547671646928e-08,
      "loss": 0.0007,
      "step": 177620
    },
    {
      "epoch": 2.906487768960157,
      "grad_norm": 0.01851843111217022,
      "learning_rate": 2.956808433484504e-08,
      "loss": 0.0004,
      "step": 177630
    },
    {
      "epoch": 2.9066513949112327,
      "grad_norm": 0.06739016622304916,
      "learning_rate": 2.946480116425454e-08,
      "loss": 0.0011,
      "step": 177640
    },
    {
      "epoch": 2.9068150208623087,
      "grad_norm": 0.023466920480132103,
      "learning_rate": 2.9361698163618557e-08,
      "loss": 0.0006,
      "step": 177650
    },
    {
      "epoch": 2.9069786468133847,
      "grad_norm": 0.15963415801525116,
      "learning_rate": 2.925877533667465e-08,
      "loss": 0.0012,
      "step": 177660
    },
    {
      "epoch": 2.9071422727644602,
      "grad_norm": 0.09810908883810043,
      "learning_rate": 2.915603268715317e-08,
      "loss": 0.0008,
      "step": 177670
    },
    {
      "epoch": 2.9073058987155362,
      "grad_norm": 0.11275790631771088,
      "learning_rate": 2.9053470218778356e-08,
      "loss": 0.0008,
      "step": 177680
    },
    {
      "epoch": 2.9074695246666122,
      "grad_norm": 0.0028494063299149275,
      "learning_rate": 2.8951087935267797e-08,
      "loss": 0.0006,
      "step": 177690
    },
    {
      "epoch": 2.907633150617688,
      "grad_norm": 0.0122525654733181,
      "learning_rate": 2.884888584033241e-08,
      "loss": 0.0006,
      "step": 177700
    },
    {
      "epoch": 2.907796776568764,
      "grad_norm": 0.020344428718090057,
      "learning_rate": 2.874686393767756e-08,
      "loss": 0.0008,
      "step": 177710
    },
    {
      "epoch": 2.90796040251984,
      "grad_norm": 0.007619696203619242,
      "learning_rate": 2.8645022231000853e-08,
      "loss": 0.0007,
      "step": 177720
    },
    {
      "epoch": 2.9081240284709153,
      "grad_norm": 0.015370062552392483,
      "learning_rate": 2.8543360723993218e-08,
      "loss": 0.0016,
      "step": 177730
    },
    {
      "epoch": 2.9082876544219913,
      "grad_norm": 0.031640708446502686,
      "learning_rate": 2.8441879420340602e-08,
      "loss": 0.0006,
      "step": 177740
    },
    {
      "epoch": 2.9084512803730673,
      "grad_norm": 0.08246639370918274,
      "learning_rate": 2.834057832372117e-08,
      "loss": 0.0012,
      "step": 177750
    },
    {
      "epoch": 2.908614906324143,
      "grad_norm": 0.06570059806108475,
      "learning_rate": 2.8239457437805872e-08,
      "loss": 0.0005,
      "step": 177760
    },
    {
      "epoch": 2.908778532275219,
      "grad_norm": 0.034638144075870514,
      "learning_rate": 2.813851676626178e-08,
      "loss": 0.0004,
      "step": 177770
    },
    {
      "epoch": 2.908942158226295,
      "grad_norm": 0.11941083520650864,
      "learning_rate": 2.8037756312746522e-08,
      "loss": 0.001,
      "step": 177780
    },
    {
      "epoch": 2.9091057841773704,
      "grad_norm": 0.11816394329071045,
      "learning_rate": 2.793717608091273e-08,
      "loss": 0.0006,
      "step": 177790
    },
    {
      "epoch": 2.9092694101284464,
      "grad_norm": 0.003514394862577319,
      "learning_rate": 2.7836776074406935e-08,
      "loss": 0.0006,
      "step": 177800
    },
    {
      "epoch": 2.9094330360795224,
      "grad_norm": 0.005191583652049303,
      "learning_rate": 2.773655629686678e-08,
      "loss": 0.0005,
      "step": 177810
    },
    {
      "epoch": 2.909596662030598,
      "grad_norm": 0.12702132761478424,
      "learning_rate": 2.7636516751926578e-08,
      "loss": 0.0008,
      "step": 177820
    },
    {
      "epoch": 2.909760287981674,
      "grad_norm": 0.040010128170251846,
      "learning_rate": 2.7536657443211767e-08,
      "loss": 0.0026,
      "step": 177830
    },
    {
      "epoch": 2.90992391393275,
      "grad_norm": 0.07131540030241013,
      "learning_rate": 2.7436978374342226e-08,
      "loss": 0.001,
      "step": 177840
    },
    {
      "epoch": 2.9100875398838255,
      "grad_norm": 0.024162158370018005,
      "learning_rate": 2.7337479548930623e-08,
      "loss": 0.0014,
      "step": 177850
    },
    {
      "epoch": 2.9102511658349015,
      "grad_norm": 0.0013242284767329693,
      "learning_rate": 2.723816097058407e-08,
      "loss": 0.0007,
      "step": 177860
    },
    {
      "epoch": 2.9104147917859775,
      "grad_norm": 0.0698835477232933,
      "learning_rate": 2.713902264290247e-08,
      "loss": 0.0011,
      "step": 177870
    },
    {
      "epoch": 2.910578417737053,
      "grad_norm": 0.0022747977636754513,
      "learning_rate": 2.7040064569479606e-08,
      "loss": 0.0005,
      "step": 177880
    },
    {
      "epoch": 2.910742043688129,
      "grad_norm": 0.13248001039028168,
      "learning_rate": 2.6941286753901507e-08,
      "loss": 0.0004,
      "step": 177890
    },
    {
      "epoch": 2.910905669639205,
      "grad_norm": 0.016011444851756096,
      "learning_rate": 2.684268919974975e-08,
      "loss": 0.0009,
      "step": 177900
    },
    {
      "epoch": 2.9110692955902806,
      "grad_norm": 0.189195916056633,
      "learning_rate": 2.674427191059814e-08,
      "loss": 0.0004,
      "step": 177910
    },
    {
      "epoch": 2.9112329215413566,
      "grad_norm": 0.12591543793678284,
      "learning_rate": 2.6646034890013272e-08,
      "loss": 0.0013,
      "step": 177920
    },
    {
      "epoch": 2.911396547492432,
      "grad_norm": 0.007671535015106201,
      "learning_rate": 2.6547978141556742e-08,
      "loss": 0.0006,
      "step": 177930
    },
    {
      "epoch": 2.911560173443508,
      "grad_norm": 0.0503133125603199,
      "learning_rate": 2.645010166878237e-08,
      "loss": 0.0004,
      "step": 177940
    },
    {
      "epoch": 2.911723799394584,
      "grad_norm": 0.04913368821144104,
      "learning_rate": 2.635240547523843e-08,
      "loss": 0.0012,
      "step": 177950
    },
    {
      "epoch": 2.9118874253456597,
      "grad_norm": 0.0596780888736248,
      "learning_rate": 2.625488956446598e-08,
      "loss": 0.0004,
      "step": 177960
    },
    {
      "epoch": 2.9120510512967357,
      "grad_norm": 0.007770646829158068,
      "learning_rate": 2.615755393999997e-08,
      "loss": 0.0005,
      "step": 177970
    },
    {
      "epoch": 2.9122146772478112,
      "grad_norm": 0.07564467191696167,
      "learning_rate": 2.6060398605368132e-08,
      "loss": 0.0023,
      "step": 177980
    },
    {
      "epoch": 2.9123783031988872,
      "grad_norm": 0.0484570637345314,
      "learning_rate": 2.5963423564092095e-08,
      "loss": 0.0013,
      "step": 177990
    },
    {
      "epoch": 2.9125419291499632,
      "grad_norm": 0.06047716736793518,
      "learning_rate": 2.5866628819686823e-08,
      "loss": 0.0011,
      "step": 178000
    },
    {
      "epoch": 2.912705555101039,
      "grad_norm": 0.15049371123313904,
      "learning_rate": 2.5770014375662288e-08,
      "loss": 0.0012,
      "step": 178010
    },
    {
      "epoch": 2.912869181052115,
      "grad_norm": 0.0027358862571418285,
      "learning_rate": 2.5673580235518468e-08,
      "loss": 0.0008,
      "step": 178020
    },
    {
      "epoch": 2.913032807003191,
      "grad_norm": 0.0036435327492654324,
      "learning_rate": 2.5577326402752565e-08,
      "loss": 0.0005,
      "step": 178030
    },
    {
      "epoch": 2.9131964329542663,
      "grad_norm": 0.09912025183439255,
      "learning_rate": 2.5481252880852348e-08,
      "loss": 0.0004,
      "step": 178040
    },
    {
      "epoch": 2.9133600589053423,
      "grad_norm": 0.09612133353948593,
      "learning_rate": 2.5385359673301137e-08,
      "loss": 0.0008,
      "step": 178050
    },
    {
      "epoch": 2.9135236848564183,
      "grad_norm": 0.06519845128059387,
      "learning_rate": 2.528964678357393e-08,
      "loss": 0.0009,
      "step": 178060
    },
    {
      "epoch": 2.913687310807494,
      "grad_norm": 0.039203792810440063,
      "learning_rate": 2.5194114215141287e-08,
      "loss": 0.0004,
      "step": 178070
    },
    {
      "epoch": 2.91385093675857,
      "grad_norm": 0.08433068543672562,
      "learning_rate": 2.5098761971465434e-08,
      "loss": 0.0005,
      "step": 178080
    },
    {
      "epoch": 2.914014562709646,
      "grad_norm": 0.048395153135061264,
      "learning_rate": 2.5003590056001946e-08,
      "loss": 0.0004,
      "step": 178090
    },
    {
      "epoch": 2.9141781886607214,
      "grad_norm": 0.05308118090033531,
      "learning_rate": 2.490859847220195e-08,
      "loss": 0.0004,
      "step": 178100
    },
    {
      "epoch": 2.9143418146117974,
      "grad_norm": 0.09533210098743439,
      "learning_rate": 2.4813787223507134e-08,
      "loss": 0.0002,
      "step": 178110
    },
    {
      "epoch": 2.9145054405628734,
      "grad_norm": 0.007121017202734947,
      "learning_rate": 2.4719156313355307e-08,
      "loss": 0.0006,
      "step": 178120
    },
    {
      "epoch": 2.914669066513949,
      "grad_norm": 0.10885131359100342,
      "learning_rate": 2.4624705745176503e-08,
      "loss": 0.0007,
      "step": 178130
    },
    {
      "epoch": 2.914832692465025,
      "grad_norm": 0.07306266576051712,
      "learning_rate": 2.4530435522394092e-08,
      "loss": 0.0005,
      "step": 178140
    },
    {
      "epoch": 2.914996318416101,
      "grad_norm": 0.025600915774703026,
      "learning_rate": 2.4436345648424786e-08,
      "loss": 0.0009,
      "step": 178150
    },
    {
      "epoch": 2.9151599443671765,
      "grad_norm": 0.12582968175411224,
      "learning_rate": 2.4342436126679747e-08,
      "loss": 0.0014,
      "step": 178160
    },
    {
      "epoch": 2.9153235703182525,
      "grad_norm": 0.008355293422937393,
      "learning_rate": 2.4248706960563473e-08,
      "loss": 0.0004,
      "step": 178170
    },
    {
      "epoch": 2.9154871962693285,
      "grad_norm": 0.05410925671458244,
      "learning_rate": 2.415515815347158e-08,
      "loss": 0.0008,
      "step": 178180
    },
    {
      "epoch": 2.915650822220404,
      "grad_norm": 0.06869533658027649,
      "learning_rate": 2.4061789708796356e-08,
      "loss": 0.0006,
      "step": 178190
    },
    {
      "epoch": 2.91581444817148,
      "grad_norm": 0.09714231640100479,
      "learning_rate": 2.3968601629922315e-08,
      "loss": 0.0006,
      "step": 178200
    },
    {
      "epoch": 2.915978074122556,
      "grad_norm": 0.09477607905864716,
      "learning_rate": 2.3875593920226203e-08,
      "loss": 0.0004,
      "step": 178210
    },
    {
      "epoch": 2.9161417000736316,
      "grad_norm": 0.0013071083230897784,
      "learning_rate": 2.378276658308032e-08,
      "loss": 0.0005,
      "step": 178220
    },
    {
      "epoch": 2.9163053260247076,
      "grad_norm": 0.141518235206604,
      "learning_rate": 2.3690119621849196e-08,
      "loss": 0.0015,
      "step": 178230
    },
    {
      "epoch": 2.9164689519757836,
      "grad_norm": 0.0389409214258194,
      "learning_rate": 2.3597653039891254e-08,
      "loss": 0.0009,
      "step": 178240
    },
    {
      "epoch": 2.916632577926859,
      "grad_norm": 0.07559758424758911,
      "learning_rate": 2.350536684055771e-08,
      "loss": 0.0007,
      "step": 178250
    },
    {
      "epoch": 2.916796203877935,
      "grad_norm": 0.07326913625001907,
      "learning_rate": 2.3413261027193655e-08,
      "loss": 0.0008,
      "step": 178260
    },
    {
      "epoch": 2.916959829829011,
      "grad_norm": 0.08001751452684402,
      "learning_rate": 2.3321335603138097e-08,
      "loss": 0.0005,
      "step": 178270
    },
    {
      "epoch": 2.9171234557800867,
      "grad_norm": 0.0018589055398479104,
      "learning_rate": 2.322959057172336e-08,
      "loss": 0.0005,
      "step": 178280
    },
    {
      "epoch": 2.9172870817311627,
      "grad_norm": 0.09099394828081131,
      "learning_rate": 2.3138025936274565e-08,
      "loss": 0.0007,
      "step": 178290
    },
    {
      "epoch": 2.9174507076822382,
      "grad_norm": 0.030106818303465843,
      "learning_rate": 2.304664170011017e-08,
      "loss": 0.0005,
      "step": 178300
    },
    {
      "epoch": 2.9176143336333142,
      "grad_norm": 0.15790459513664246,
      "learning_rate": 2.295543786654364e-08,
      "loss": 0.005,
      "step": 178310
    },
    {
      "epoch": 2.9177779595843902,
      "grad_norm": 0.049048371613025665,
      "learning_rate": 2.28644144388801e-08,
      "loss": 0.0006,
      "step": 178320
    },
    {
      "epoch": 2.917941585535466,
      "grad_norm": 0.041185732930898666,
      "learning_rate": 2.277357142042025e-08,
      "loss": 0.0011,
      "step": 178330
    },
    {
      "epoch": 2.918105211486542,
      "grad_norm": 0.08303149044513702,
      "learning_rate": 2.2682908814454785e-08,
      "loss": 0.0009,
      "step": 178340
    },
    {
      "epoch": 2.9182688374376173,
      "grad_norm": 0.04346705600619316,
      "learning_rate": 2.2592426624271635e-08,
      "loss": 0.0003,
      "step": 178350
    },
    {
      "epoch": 2.9184324633886933,
      "grad_norm": 0.016193971037864685,
      "learning_rate": 2.2502124853150396e-08,
      "loss": 0.0006,
      "step": 178360
    },
    {
      "epoch": 2.9185960893397693,
      "grad_norm": 0.09852645546197891,
      "learning_rate": 2.241200350436401e-08,
      "loss": 0.0005,
      "step": 178370
    },
    {
      "epoch": 2.918759715290845,
      "grad_norm": 0.054551128298044205,
      "learning_rate": 2.2322062581179305e-08,
      "loss": 0.0004,
      "step": 178380
    },
    {
      "epoch": 2.918923341241921,
      "grad_norm": 0.1080770269036293,
      "learning_rate": 2.2232302086855894e-08,
      "loss": 0.0008,
      "step": 178390
    },
    {
      "epoch": 2.919086967192997,
      "grad_norm": 0.07357295602560043,
      "learning_rate": 2.2142722024647844e-08,
      "loss": 0.0008,
      "step": 178400
    },
    {
      "epoch": 2.9192505931440724,
      "grad_norm": 0.1444116085767746,
      "learning_rate": 2.2053322397802558e-08,
      "loss": 0.0014,
      "step": 178410
    },
    {
      "epoch": 2.9194142190951484,
      "grad_norm": 0.03502218797802925,
      "learning_rate": 2.196410320955966e-08,
      "loss": 0.0004,
      "step": 178420
    },
    {
      "epoch": 2.9195778450462244,
      "grad_norm": 0.09382104128599167,
      "learning_rate": 2.1875064463153794e-08,
      "loss": 0.0005,
      "step": 178430
    },
    {
      "epoch": 2.9197414709973,
      "grad_norm": 0.022338813170790672,
      "learning_rate": 2.1786206161812373e-08,
      "loss": 0.0006,
      "step": 178440
    },
    {
      "epoch": 2.919905096948376,
      "grad_norm": 0.1384303718805313,
      "learning_rate": 2.1697528308756154e-08,
      "loss": 0.0011,
      "step": 178450
    },
    {
      "epoch": 2.920068722899452,
      "grad_norm": 0.02841377817094326,
      "learning_rate": 2.1609030907199234e-08,
      "loss": 0.0004,
      "step": 178460
    },
    {
      "epoch": 2.9202323488505275,
      "grad_norm": 0.023619532585144043,
      "learning_rate": 2.152071396035016e-08,
      "loss": 0.0003,
      "step": 178470
    },
    {
      "epoch": 2.9203959748016035,
      "grad_norm": 0.038409534841775894,
      "learning_rate": 2.14325774714097e-08,
      "loss": 0.0005,
      "step": 178480
    },
    {
      "epoch": 2.9205596007526795,
      "grad_norm": 0.011579222045838833,
      "learning_rate": 2.1344621443571966e-08,
      "loss": 0.0003,
      "step": 178490
    },
    {
      "epoch": 2.920723226703755,
      "grad_norm": 0.012659910134971142,
      "learning_rate": 2.1256845880026634e-08,
      "loss": 0.0007,
      "step": 178500
    },
    {
      "epoch": 2.920886852654831,
      "grad_norm": 0.056106604635715485,
      "learning_rate": 2.1169250783953933e-08,
      "loss": 0.0008,
      "step": 178510
    },
    {
      "epoch": 2.921050478605907,
      "grad_norm": 0.024914825335144997,
      "learning_rate": 2.1081836158529655e-08,
      "loss": 0.0008,
      "step": 178520
    },
    {
      "epoch": 2.9212141045569826,
      "grad_norm": 0.08566151559352875,
      "learning_rate": 2.0994602006922383e-08,
      "loss": 0.0006,
      "step": 178530
    },
    {
      "epoch": 2.9213777305080586,
      "grad_norm": 0.0793798640370369,
      "learning_rate": 2.0907548332294025e-08,
      "loss": 0.0006,
      "step": 178540
    },
    {
      "epoch": 2.9215413564591346,
      "grad_norm": 0.06904827803373337,
      "learning_rate": 2.0820675137800396e-08,
      "loss": 0.0005,
      "step": 178550
    },
    {
      "epoch": 2.92170498241021,
      "grad_norm": 0.02474076673388481,
      "learning_rate": 2.073398242659008e-08,
      "loss": 0.0014,
      "step": 178560
    },
    {
      "epoch": 2.921868608361286,
      "grad_norm": 0.21148435771465302,
      "learning_rate": 2.064747020180502e-08,
      "loss": 0.0011,
      "step": 178570
    },
    {
      "epoch": 2.922032234312362,
      "grad_norm": 0.018140675500035286,
      "learning_rate": 2.0561138466582142e-08,
      "loss": 0.0006,
      "step": 178580
    },
    {
      "epoch": 2.9221958602634377,
      "grad_norm": 0.0843304768204689,
      "learning_rate": 2.0474987224050613e-08,
      "loss": 0.001,
      "step": 178590
    },
    {
      "epoch": 2.9223594862145137,
      "grad_norm": 0.06330302357673645,
      "learning_rate": 2.0389016477331825e-08,
      "loss": 0.0008,
      "step": 178600
    },
    {
      "epoch": 2.9225231121655897,
      "grad_norm": 0.09981589764356613,
      "learning_rate": 2.0303226229543837e-08,
      "loss": 0.0006,
      "step": 178610
    },
    {
      "epoch": 2.9226867381166652,
      "grad_norm": 0.1229078397154808,
      "learning_rate": 2.0217616483795278e-08,
      "loss": 0.0005,
      "step": 178620
    },
    {
      "epoch": 2.9228503640677412,
      "grad_norm": 0.21442505717277527,
      "learning_rate": 2.0132187243189215e-08,
      "loss": 0.0007,
      "step": 178630
    },
    {
      "epoch": 2.9230139900188172,
      "grad_norm": 0.06810841709375381,
      "learning_rate": 2.0046938510822622e-08,
      "loss": 0.001,
      "step": 178640
    },
    {
      "epoch": 2.923177615969893,
      "grad_norm": 0.11991714686155319,
      "learning_rate": 1.9961870289785802e-08,
      "loss": 0.0006,
      "step": 178650
    },
    {
      "epoch": 2.923341241920969,
      "grad_norm": 0.05042572319507599,
      "learning_rate": 1.987698258316184e-08,
      "loss": 0.0007,
      "step": 178660
    },
    {
      "epoch": 2.9235048678720448,
      "grad_norm": 0.008573350496590137,
      "learning_rate": 1.9792275394027727e-08,
      "loss": 0.0009,
      "step": 178670
    },
    {
      "epoch": 2.9236684938231203,
      "grad_norm": 0.1576259881258011,
      "learning_rate": 1.9707748725454334e-08,
      "loss": 0.0011,
      "step": 178680
    },
    {
      "epoch": 2.9238321197741963,
      "grad_norm": 0.018002888187766075,
      "learning_rate": 1.9623402580505324e-08,
      "loss": 0.0005,
      "step": 178690
    },
    {
      "epoch": 2.923995745725272,
      "grad_norm": 0.09167953580617905,
      "learning_rate": 1.953923696223714e-08,
      "loss": 0.0006,
      "step": 178700
    },
    {
      "epoch": 2.924159371676348,
      "grad_norm": 0.011544295586645603,
      "learning_rate": 1.945525187370234e-08,
      "loss": 0.001,
      "step": 178710
    },
    {
      "epoch": 2.924322997627424,
      "grad_norm": 0.06612348556518555,
      "learning_rate": 1.9371447317943494e-08,
      "loss": 0.0006,
      "step": 178720
    },
    {
      "epoch": 2.9244866235784994,
      "grad_norm": 0.03486032038927078,
      "learning_rate": 1.9287823297999274e-08,
      "loss": 0.0004,
      "step": 178730
    },
    {
      "epoch": 2.9246502495295754,
      "grad_norm": 0.03081473894417286,
      "learning_rate": 1.9204379816901153e-08,
      "loss": 0.0005,
      "step": 178740
    },
    {
      "epoch": 2.924813875480651,
      "grad_norm": 0.06880506873130798,
      "learning_rate": 1.912111687767282e-08,
      "loss": 0.0005,
      "step": 178750
    },
    {
      "epoch": 2.924977501431727,
      "grad_norm": 0.20954085886478424,
      "learning_rate": 1.9038034483332414e-08,
      "loss": 0.0006,
      "step": 178760
    },
    {
      "epoch": 2.925141127382803,
      "grad_norm": 0.015317626297473907,
      "learning_rate": 1.8955132636892526e-08,
      "loss": 0.0004,
      "step": 178770
    },
    {
      "epoch": 2.9253047533338785,
      "grad_norm": 0.09438851475715637,
      "learning_rate": 1.8872411341356866e-08,
      "loss": 0.0004,
      "step": 178780
    },
    {
      "epoch": 2.9254683792849545,
      "grad_norm": 0.03295215591788292,
      "learning_rate": 1.87898705997247e-08,
      "loss": 0.0009,
      "step": 178790
    },
    {
      "epoch": 2.9256320052360305,
      "grad_norm": 0.06873004883527756,
      "learning_rate": 1.8707510414988083e-08,
      "loss": 0.0013,
      "step": 178800
    },
    {
      "epoch": 2.925795631187106,
      "grad_norm": 0.1080973744392395,
      "learning_rate": 1.862533079013129e-08,
      "loss": 0.0011,
      "step": 178810
    },
    {
      "epoch": 2.925959257138182,
      "grad_norm": 0.023874951526522636,
      "learning_rate": 1.854333172813416e-08,
      "loss": 0.0005,
      "step": 178820
    },
    {
      "epoch": 2.926122883089258,
      "grad_norm": 0.06175655126571655,
      "learning_rate": 1.8461513231968764e-08,
      "loss": 0.0007,
      "step": 178830
    },
    {
      "epoch": 2.9262865090403336,
      "grad_norm": 0.05289186164736748,
      "learning_rate": 1.8379875304600502e-08,
      "loss": 0.0002,
      "step": 178840
    },
    {
      "epoch": 2.9264501349914096,
      "grad_norm": 0.11557146161794662,
      "learning_rate": 1.829841794898868e-08,
      "loss": 0.0009,
      "step": 178850
    },
    {
      "epoch": 2.9266137609424856,
      "grad_norm": 0.07466714084148407,
      "learning_rate": 1.8217141168086484e-08,
      "loss": 0.0004,
      "step": 178860
    },
    {
      "epoch": 2.926777386893561,
      "grad_norm": 0.06818711757659912,
      "learning_rate": 1.813604496483934e-08,
      "loss": 0.0006,
      "step": 178870
    },
    {
      "epoch": 2.926941012844637,
      "grad_norm": 0.106024831533432,
      "learning_rate": 1.805512934218656e-08,
      "loss": 0.0014,
      "step": 178880
    },
    {
      "epoch": 2.927104638795713,
      "grad_norm": 0.13435648381710052,
      "learning_rate": 1.7974394303061916e-08,
      "loss": 0.0007,
      "step": 178890
    },
    {
      "epoch": 2.9272682647467887,
      "grad_norm": 0.06102341040968895,
      "learning_rate": 1.7893839850391393e-08,
      "loss": 0.0007,
      "step": 178900
    },
    {
      "epoch": 2.9274318906978647,
      "grad_norm": 0.005345731507986784,
      "learning_rate": 1.781346598709488e-08,
      "loss": 0.0006,
      "step": 178910
    },
    {
      "epoch": 2.9275955166489407,
      "grad_norm": 0.07079720497131348,
      "learning_rate": 1.7733272716086158e-08,
      "loss": 0.0011,
      "step": 178920
    },
    {
      "epoch": 2.9277591426000162,
      "grad_norm": 0.0029305056668817997,
      "learning_rate": 1.7653260040271235e-08,
      "loss": 0.0039,
      "step": 178930
    },
    {
      "epoch": 2.9279227685510922,
      "grad_norm": 0.05053407698869705,
      "learning_rate": 1.7573427962551683e-08,
      "loss": 0.0015,
      "step": 178940
    },
    {
      "epoch": 2.9280863945021682,
      "grad_norm": 0.024717232212424278,
      "learning_rate": 1.7493776485819624e-08,
      "loss": 0.0008,
      "step": 178950
    },
    {
      "epoch": 2.928250020453244,
      "grad_norm": 0.1315728724002838,
      "learning_rate": 1.741430561296331e-08,
      "loss": 0.0008,
      "step": 178960
    },
    {
      "epoch": 2.92841364640432,
      "grad_norm": 0.058604784309864044,
      "learning_rate": 1.7335015346863215e-08,
      "loss": 0.0005,
      "step": 178970
    },
    {
      "epoch": 2.9285772723553958,
      "grad_norm": 0.054452478885650635,
      "learning_rate": 1.7255905690393148e-08,
      "loss": 0.0005,
      "step": 178980
    },
    {
      "epoch": 2.9287408983064713,
      "grad_norm": 0.01984330266714096,
      "learning_rate": 1.7176976646420818e-08,
      "loss": 0.0003,
      "step": 178990
    },
    {
      "epoch": 2.9289045242575473,
      "grad_norm": 0.03282059729099274,
      "learning_rate": 1.7098228217807268e-08,
      "loss": 0.0004,
      "step": 179000
    },
    {
      "epoch": 2.9290681502086233,
      "grad_norm": 0.08029294013977051,
      "learning_rate": 1.701966040740688e-08,
      "loss": 0.0006,
      "step": 179010
    },
    {
      "epoch": 2.929231776159699,
      "grad_norm": 0.034790992736816406,
      "learning_rate": 1.694127321806738e-08,
      "loss": 0.0012,
      "step": 179020
    },
    {
      "epoch": 2.929395402110775,
      "grad_norm": 0.009209046140313148,
      "learning_rate": 1.6863066652630377e-08,
      "loss": 0.0012,
      "step": 179030
    },
    {
      "epoch": 2.929559028061851,
      "grad_norm": 0.14171026647090912,
      "learning_rate": 1.678504071393028e-08,
      "loss": 0.0007,
      "step": 179040
    },
    {
      "epoch": 2.9297226540129264,
      "grad_norm": 0.03202229365706444,
      "learning_rate": 1.670719540479593e-08,
      "loss": 0.0004,
      "step": 179050
    },
    {
      "epoch": 2.9298862799640024,
      "grad_norm": 0.09599992632865906,
      "learning_rate": 1.662953072804896e-08,
      "loss": 0.0006,
      "step": 179060
    },
    {
      "epoch": 2.930049905915078,
      "grad_norm": 0.041298095136880875,
      "learning_rate": 1.6552046686503786e-08,
      "loss": 0.0005,
      "step": 179070
    },
    {
      "epoch": 2.930213531866154,
      "grad_norm": 0.09955354779958725,
      "learning_rate": 1.647474328296983e-08,
      "loss": 0.0006,
      "step": 179080
    },
    {
      "epoch": 2.93037715781723,
      "grad_norm": 0.0810522511601448,
      "learning_rate": 1.6397620520248735e-08,
      "loss": 0.0011,
      "step": 179090
    },
    {
      "epoch": 2.9305407837683055,
      "grad_norm": 0.10365840047597885,
      "learning_rate": 1.6320678401136048e-08,
      "loss": 0.0015,
      "step": 179100
    },
    {
      "epoch": 2.9307044097193815,
      "grad_norm": 0.09869789332151413,
      "learning_rate": 1.6243916928420643e-08,
      "loss": 0.0005,
      "step": 179110
    },
    {
      "epoch": 2.930868035670457,
      "grad_norm": 0.056728944182395935,
      "learning_rate": 1.61673361048853e-08,
      "loss": 0.0005,
      "step": 179120
    },
    {
      "epoch": 2.931031661621533,
      "grad_norm": 0.07496386021375656,
      "learning_rate": 1.6090935933305573e-08,
      "loss": 0.001,
      "step": 179130
    },
    {
      "epoch": 2.931195287572609,
      "grad_norm": 0.16725483536720276,
      "learning_rate": 1.6014716416450915e-08,
      "loss": 0.0007,
      "step": 179140
    },
    {
      "epoch": 2.9313589135236846,
      "grad_norm": 0.07106459140777588,
      "learning_rate": 1.593867755708467e-08,
      "loss": 0.0017,
      "step": 179150
    },
    {
      "epoch": 2.9315225394747606,
      "grad_norm": 0.09330414235591888,
      "learning_rate": 1.5862819357961857e-08,
      "loss": 0.0006,
      "step": 179160
    },
    {
      "epoch": 2.9316861654258366,
      "grad_norm": 0.054158855229616165,
      "learning_rate": 1.578714182183305e-08,
      "loss": 0.0007,
      "step": 179170
    },
    {
      "epoch": 2.931849791376912,
      "grad_norm": 0.03244738653302193,
      "learning_rate": 1.571164495144106e-08,
      "loss": 0.0019,
      "step": 179180
    },
    {
      "epoch": 2.932013417327988,
      "grad_norm": 0.07225721329450607,
      "learning_rate": 1.5636328749522588e-08,
      "loss": 0.0011,
      "step": 179190
    },
    {
      "epoch": 2.932177043279064,
      "grad_norm": 0.12369494885206223,
      "learning_rate": 1.5561193218807114e-08,
      "loss": 0.0006,
      "step": 179200
    },
    {
      "epoch": 2.9323406692301397,
      "grad_norm": 0.032483603805303574,
      "learning_rate": 1.5486238362019122e-08,
      "loss": 0.0005,
      "step": 179210
    },
    {
      "epoch": 2.9325042951812157,
      "grad_norm": 0.1872226595878601,
      "learning_rate": 1.541146418187478e-08,
      "loss": 0.0009,
      "step": 179220
    },
    {
      "epoch": 2.9326679211322917,
      "grad_norm": 0.043272245675325394,
      "learning_rate": 1.5336870681084693e-08,
      "loss": 0.0006,
      "step": 179230
    },
    {
      "epoch": 2.9328315470833672,
      "grad_norm": 0.06240785866975784,
      "learning_rate": 1.5262457862353363e-08,
      "loss": 0.0007,
      "step": 179240
    },
    {
      "epoch": 2.9329951730344432,
      "grad_norm": 0.08079081773757935,
      "learning_rate": 1.5188225728376415e-08,
      "loss": 0.001,
      "step": 179250
    },
    {
      "epoch": 2.9331587989855192,
      "grad_norm": 0.012990414164960384,
      "learning_rate": 1.5114174281846138e-08,
      "loss": 0.0004,
      "step": 179260
    },
    {
      "epoch": 2.933322424936595,
      "grad_norm": 0.05128638818860054,
      "learning_rate": 1.5040303525446498e-08,
      "loss": 0.0003,
      "step": 179270
    },
    {
      "epoch": 2.933486050887671,
      "grad_norm": 0.055838439613580704,
      "learning_rate": 1.496661346185424e-08,
      "loss": 0.0005,
      "step": 179280
    },
    {
      "epoch": 2.933649676838747,
      "grad_norm": 0.15984193980693817,
      "learning_rate": 1.4893104093741672e-08,
      "loss": 0.0008,
      "step": 179290
    },
    {
      "epoch": 2.9338133027898223,
      "grad_norm": 0.010083814151585102,
      "learning_rate": 1.481977542377222e-08,
      "loss": 0.0003,
      "step": 179300
    },
    {
      "epoch": 2.9339769287408983,
      "grad_norm": 0.04718276485800743,
      "learning_rate": 1.4746627454604311e-08,
      "loss": 0.0007,
      "step": 179310
    },
    {
      "epoch": 2.9341405546919743,
      "grad_norm": 0.030455535277724266,
      "learning_rate": 1.467366018888916e-08,
      "loss": 0.0008,
      "step": 179320
    },
    {
      "epoch": 2.93430418064305,
      "grad_norm": 0.03171857073903084,
      "learning_rate": 1.4600873629272427e-08,
      "loss": 0.0008,
      "step": 179330
    },
    {
      "epoch": 2.934467806594126,
      "grad_norm": 0.05610997974872589,
      "learning_rate": 1.4528267778392003e-08,
      "loss": 0.0009,
      "step": 179340
    },
    {
      "epoch": 2.934631432545202,
      "grad_norm": 0.0778215080499649,
      "learning_rate": 1.4455842638879113e-08,
      "loss": 0.001,
      "step": 179350
    },
    {
      "epoch": 2.9347950584962774,
      "grad_norm": 0.07766593992710114,
      "learning_rate": 1.4383598213359995e-08,
      "loss": 0.0007,
      "step": 179360
    },
    {
      "epoch": 2.9349586844473534,
      "grad_norm": 0.09563033282756805,
      "learning_rate": 1.4311534504452551e-08,
      "loss": 0.0007,
      "step": 179370
    },
    {
      "epoch": 2.9351223103984294,
      "grad_norm": 0.006421692669391632,
      "learning_rate": 1.4239651514769138e-08,
      "loss": 0.0007,
      "step": 179380
    },
    {
      "epoch": 2.935285936349505,
      "grad_norm": 0.22701066732406616,
      "learning_rate": 1.4167949246915446e-08,
      "loss": 0.0011,
      "step": 179390
    },
    {
      "epoch": 2.935449562300581,
      "grad_norm": 0.09389105439186096,
      "learning_rate": 1.4096427703490512e-08,
      "loss": 0.0006,
      "step": 179400
    },
    {
      "epoch": 2.935613188251657,
      "grad_norm": 0.04223419353365898,
      "learning_rate": 1.4025086887086703e-08,
      "loss": 0.0007,
      "step": 179410
    },
    {
      "epoch": 2.9357768142027325,
      "grad_norm": 0.23792944848537445,
      "learning_rate": 1.3953926800290284e-08,
      "loss": 0.0009,
      "step": 179420
    },
    {
      "epoch": 2.9359404401538085,
      "grad_norm": 0.043693047016859055,
      "learning_rate": 1.3882947445680306e-08,
      "loss": 0.0005,
      "step": 179430
    },
    {
      "epoch": 2.9361040661048845,
      "grad_norm": 0.010036669671535492,
      "learning_rate": 1.3812148825829152e-08,
      "loss": 0.0003,
      "step": 179440
    },
    {
      "epoch": 2.93626769205596,
      "grad_norm": 0.14526362717151642,
      "learning_rate": 1.3741530943304216e-08,
      "loss": 0.0009,
      "step": 179450
    },
    {
      "epoch": 2.936431318007036,
      "grad_norm": 0.05847657844424248,
      "learning_rate": 1.3671093800664003e-08,
      "loss": 0.0003,
      "step": 179460
    },
    {
      "epoch": 2.9365949439581116,
      "grad_norm": 0.06222475692629814,
      "learning_rate": 1.360083740046314e-08,
      "loss": 0.0011,
      "step": 179470
    },
    {
      "epoch": 2.9367585699091876,
      "grad_norm": 0.003975050989538431,
      "learning_rate": 1.3530761745246812e-08,
      "loss": 0.0005,
      "step": 179480
    },
    {
      "epoch": 2.9369221958602636,
      "grad_norm": 0.03169276565313339,
      "learning_rate": 1.3460866837555764e-08,
      "loss": 0.0004,
      "step": 179490
    },
    {
      "epoch": 2.937085821811339,
      "grad_norm": 0.37221333384513855,
      "learning_rate": 1.339115267992408e-08,
      "loss": 0.0018,
      "step": 179500
    },
    {
      "epoch": 2.937249447762415,
      "grad_norm": 0.07852032035589218,
      "learning_rate": 1.3321619274877517e-08,
      "loss": 0.0011,
      "step": 179510
    },
    {
      "epoch": 2.9374130737134907,
      "grad_norm": 0.23444631695747375,
      "learning_rate": 1.3252266624936838e-08,
      "loss": 0.0008,
      "step": 179520
    },
    {
      "epoch": 2.9375766996645667,
      "grad_norm": 0.055987413972616196,
      "learning_rate": 1.3183094732616696e-08,
      "loss": 0.0004,
      "step": 179530
    },
    {
      "epoch": 2.9377403256156427,
      "grad_norm": 0.08950019627809525,
      "learning_rate": 1.3114103600423977e-08,
      "loss": 0.0005,
      "step": 179540
    },
    {
      "epoch": 2.9379039515667182,
      "grad_norm": 0.0325082391500473,
      "learning_rate": 1.3045293230858902e-08,
      "loss": 0.0021,
      "step": 179550
    },
    {
      "epoch": 2.9380675775177942,
      "grad_norm": 0.034914035350084305,
      "learning_rate": 1.2976663626416142e-08,
      "loss": 0.001,
      "step": 179560
    },
    {
      "epoch": 2.9382312034688702,
      "grad_norm": 0.001426022034138441,
      "learning_rate": 1.2908214789583706e-08,
      "loss": 0.0013,
      "step": 179570
    },
    {
      "epoch": 2.938394829419946,
      "grad_norm": 0.07312200963497162,
      "learning_rate": 1.2839946722841834e-08,
      "loss": 0.0004,
      "step": 179580
    },
    {
      "epoch": 2.938558455371022,
      "grad_norm": 0.01221251580864191,
      "learning_rate": 1.2771859428665767e-08,
      "loss": 0.0007,
      "step": 179590
    },
    {
      "epoch": 2.938722081322098,
      "grad_norm": 0.06478913128376007,
      "learning_rate": 1.2703952909523531e-08,
      "loss": 0.0005,
      "step": 179600
    },
    {
      "epoch": 2.9388857072731733,
      "grad_norm": 0.15559005737304688,
      "learning_rate": 1.2636227167875936e-08,
      "loss": 0.0005,
      "step": 179610
    },
    {
      "epoch": 2.9390493332242493,
      "grad_norm": 0.046208590269088745,
      "learning_rate": 1.2568682206178796e-08,
      "loss": 0.0003,
      "step": 179620
    },
    {
      "epoch": 2.9392129591753253,
      "grad_norm": 0.016566820442676544,
      "learning_rate": 1.2501318026879594e-08,
      "loss": 0.0012,
      "step": 179630
    },
    {
      "epoch": 2.939376585126401,
      "grad_norm": 0.03489551693201065,
      "learning_rate": 1.243413463242027e-08,
      "loss": 0.0009,
      "step": 179640
    },
    {
      "epoch": 2.939540211077477,
      "grad_norm": 0.04836392030119896,
      "learning_rate": 1.236713202523665e-08,
      "loss": 0.0011,
      "step": 179650
    },
    {
      "epoch": 2.939703837028553,
      "grad_norm": 0.044863633811473846,
      "learning_rate": 1.2300310207757348e-08,
      "loss": 0.0007,
      "step": 179660
    },
    {
      "epoch": 2.9398674629796284,
      "grad_norm": 0.08851360529661179,
      "learning_rate": 1.2233669182403207e-08,
      "loss": 0.0008,
      "step": 179670
    },
    {
      "epoch": 2.9400310889307044,
      "grad_norm": 0.01972980983555317,
      "learning_rate": 1.2167208951591736e-08,
      "loss": 0.0004,
      "step": 179680
    },
    {
      "epoch": 2.9401947148817804,
      "grad_norm": 0.012178403325378895,
      "learning_rate": 1.2100929517730454e-08,
      "loss": 0.0008,
      "step": 179690
    },
    {
      "epoch": 2.940358340832856,
      "grad_norm": 0.09655093401670456,
      "learning_rate": 1.2034830883222437e-08,
      "loss": 0.0004,
      "step": 179700
    },
    {
      "epoch": 2.940521966783932,
      "grad_norm": 0.002753537381067872,
      "learning_rate": 1.1968913050464104e-08,
      "loss": 0.0004,
      "step": 179710
    },
    {
      "epoch": 2.940685592735008,
      "grad_norm": 0.00786948949098587,
      "learning_rate": 1.1903176021843543e-08,
      "loss": 0.0004,
      "step": 179720
    },
    {
      "epoch": 2.9408492186860835,
      "grad_norm": 0.04288225248456001,
      "learning_rate": 1.1837619799744404e-08,
      "loss": 0.0004,
      "step": 179730
    },
    {
      "epoch": 2.9410128446371595,
      "grad_norm": 0.06836715340614319,
      "learning_rate": 1.177224438654312e-08,
      "loss": 0.0006,
      "step": 179740
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 0.039772454649209976,
      "learning_rate": 1.1707049784608903e-08,
      "loss": 0.0005,
      "step": 179750
    },
    {
      "epoch": 2.941340096539311,
      "grad_norm": 0.003927233163267374,
      "learning_rate": 1.1642035996304868e-08,
      "loss": 0.0005,
      "step": 179760
    },
    {
      "epoch": 2.941503722490387,
      "grad_norm": 0.07734093070030212,
      "learning_rate": 1.1577203023988015e-08,
      "loss": 0.0006,
      "step": 179770
    },
    {
      "epoch": 2.941667348441463,
      "grad_norm": 0.018544917926192284,
      "learning_rate": 1.1512550870008687e-08,
      "loss": 0.0008,
      "step": 179780
    },
    {
      "epoch": 2.9418309743925386,
      "grad_norm": 0.07599605619907379,
      "learning_rate": 1.14480795367089e-08,
      "loss": 0.0005,
      "step": 179790
    },
    {
      "epoch": 2.9419946003436146,
      "grad_norm": 0.051710207015275955,
      "learning_rate": 1.1383789026427339e-08,
      "loss": 0.0007,
      "step": 179800
    },
    {
      "epoch": 2.9421582262946906,
      "grad_norm": 0.001478518359363079,
      "learning_rate": 1.131967934149325e-08,
      "loss": 0.0007,
      "step": 179810
    },
    {
      "epoch": 2.942321852245766,
      "grad_norm": 0.04656429588794708,
      "learning_rate": 1.1255750484230887e-08,
      "loss": 0.001,
      "step": 179820
    },
    {
      "epoch": 2.942485478196842,
      "grad_norm": 0.06156375631690025,
      "learning_rate": 1.1192002456957286e-08,
      "loss": 0.001,
      "step": 179830
    },
    {
      "epoch": 2.9426491041479177,
      "grad_norm": 0.02736838534474373,
      "learning_rate": 1.1128435261983373e-08,
      "loss": 0.0007,
      "step": 179840
    },
    {
      "epoch": 2.9428127300989937,
      "grad_norm": 0.04875340312719345,
      "learning_rate": 1.1065048901613418e-08,
      "loss": 0.0007,
      "step": 179850
    },
    {
      "epoch": 2.9429763560500697,
      "grad_norm": 0.029956664890050888,
      "learning_rate": 1.1001843378145028e-08,
      "loss": 0.0022,
      "step": 179860
    },
    {
      "epoch": 2.9431399820011452,
      "grad_norm": 0.03844644874334335,
      "learning_rate": 1.0938818693868591e-08,
      "loss": 0.0005,
      "step": 179870
    },
    {
      "epoch": 2.9433036079522212,
      "grad_norm": 0.0018806824227795005,
      "learning_rate": 1.0875974851069504e-08,
      "loss": 0.0003,
      "step": 179880
    },
    {
      "epoch": 2.943467233903297,
      "grad_norm": 0.040779080241918564,
      "learning_rate": 1.0813311852024832e-08,
      "loss": 0.0004,
      "step": 179890
    },
    {
      "epoch": 2.943630859854373,
      "grad_norm": 0.09237122535705566,
      "learning_rate": 1.0750829699006649e-08,
      "loss": 0.0013,
      "step": 179900
    },
    {
      "epoch": 2.943794485805449,
      "grad_norm": 0.0704556405544281,
      "learning_rate": 1.0688528394279807e-08,
      "loss": 0.0011,
      "step": 179910
    },
    {
      "epoch": 2.9439581117565243,
      "grad_norm": 0.04864276573061943,
      "learning_rate": 1.0626407940101946e-08,
      "loss": 0.0004,
      "step": 179920
    },
    {
      "epoch": 2.9441217377076003,
      "grad_norm": 0.5201018452644348,
      "learning_rate": 1.0564468338725153e-08,
      "loss": 0.0009,
      "step": 179930
    },
    {
      "epoch": 2.9442853636586763,
      "grad_norm": 0.1330273449420929,
      "learning_rate": 1.0502709592394855e-08,
      "loss": 0.0007,
      "step": 179940
    },
    {
      "epoch": 2.944448989609752,
      "grad_norm": 0.18576321005821228,
      "learning_rate": 1.044113170334926e-08,
      "loss": 0.0007,
      "step": 179950
    },
    {
      "epoch": 2.944612615560828,
      "grad_norm": 0.026141202077269554,
      "learning_rate": 1.0379734673820474e-08,
      "loss": 0.0011,
      "step": 179960
    },
    {
      "epoch": 2.944776241511904,
      "grad_norm": 0.06412602961063385,
      "learning_rate": 1.0318518506034491e-08,
      "loss": 0.0007,
      "step": 179970
    },
    {
      "epoch": 2.9449398674629794,
      "grad_norm": 0.006818361114710569,
      "learning_rate": 1.0257483202209539e-08,
      "loss": 0.0009,
      "step": 179980
    },
    {
      "epoch": 2.9451034934140554,
      "grad_norm": 0.0350944884121418,
      "learning_rate": 1.0196628764558291e-08,
      "loss": 0.0012,
      "step": 179990
    },
    {
      "epoch": 2.9452671193651314,
      "grad_norm": 0.004240986425429583,
      "learning_rate": 1.0135955195286207e-08,
      "loss": 0.0013,
      "step": 180000
    },
    {
      "epoch": 2.945430745316207,
      "grad_norm": 0.01066744513809681,
      "learning_rate": 1.0075462496593192e-08,
      "loss": 0.0003,
      "step": 180010
    },
    {
      "epoch": 2.945594371267283,
      "grad_norm": 0.05837064981460571,
      "learning_rate": 1.0015150670671937e-08,
      "loss": 0.0004,
      "step": 180020
    },
    {
      "epoch": 2.945757997218359,
      "grad_norm": 0.032356880605220795,
      "learning_rate": 9.955019719707914e-09,
      "loss": 0.0007,
      "step": 180030
    },
    {
      "epoch": 2.9459216231694345,
      "grad_norm": 0.004852637182921171,
      "learning_rate": 9.895069645881605e-09,
      "loss": 0.0005,
      "step": 180040
    },
    {
      "epoch": 2.9460852491205105,
      "grad_norm": 0.15991511940956116,
      "learning_rate": 9.83530045136516e-09,
      "loss": 0.0005,
      "step": 180050
    },
    {
      "epoch": 2.9462488750715865,
      "grad_norm": 0.026175465434789658,
      "learning_rate": 9.775712138325733e-09,
      "loss": 0.0005,
      "step": 180060
    },
    {
      "epoch": 2.946412501022662,
      "grad_norm": 0.0007901117787696421,
      "learning_rate": 9.716304708923262e-09,
      "loss": 0.0008,
      "step": 180070
    },
    {
      "epoch": 2.946576126973738,
      "grad_norm": 0.0336458720266819,
      "learning_rate": 9.65707816531103e-09,
      "loss": 0.0012,
      "step": 180080
    },
    {
      "epoch": 2.946739752924814,
      "grad_norm": 0.024761490523815155,
      "learning_rate": 9.59803250963509e-09,
      "loss": 0.0008,
      "step": 180090
    },
    {
      "epoch": 2.9469033788758896,
      "grad_norm": 0.0074148597195744514,
      "learning_rate": 9.53916774403707e-09,
      "loss": 0.0008,
      "step": 180100
    },
    {
      "epoch": 2.9470670048269656,
      "grad_norm": 0.0588478222489357,
      "learning_rate": 9.480483870649704e-09,
      "loss": 0.001,
      "step": 180110
    },
    {
      "epoch": 2.9472306307780416,
      "grad_norm": 0.05119823291897774,
      "learning_rate": 9.42198089160018e-09,
      "loss": 0.0008,
      "step": 180120
    },
    {
      "epoch": 2.947394256729117,
      "grad_norm": 0.10345159471035004,
      "learning_rate": 9.36365880900958e-09,
      "loss": 0.0007,
      "step": 180130
    },
    {
      "epoch": 2.947557882680193,
      "grad_norm": 0.22478891909122467,
      "learning_rate": 9.305517624991212e-09,
      "loss": 0.0008,
      "step": 180140
    },
    {
      "epoch": 2.947721508631269,
      "grad_norm": 0.0670064240694046,
      "learning_rate": 9.24755734165339e-09,
      "loss": 0.0016,
      "step": 180150
    },
    {
      "epoch": 2.9478851345823447,
      "grad_norm": 0.08296709507703781,
      "learning_rate": 9.189777961096657e-09,
      "loss": 0.0006,
      "step": 180160
    },
    {
      "epoch": 2.9480487605334207,
      "grad_norm": 0.034969232976436615,
      "learning_rate": 9.132179485415449e-09,
      "loss": 0.0009,
      "step": 180170
    },
    {
      "epoch": 2.9482123864844967,
      "grad_norm": 0.00866776704788208,
      "learning_rate": 9.07476191669754e-09,
      "loss": 0.0011,
      "step": 180180
    },
    {
      "epoch": 2.9483760124355722,
      "grad_norm": 0.2486938238143921,
      "learning_rate": 9.017525257023485e-09,
      "loss": 0.001,
      "step": 180190
    },
    {
      "epoch": 2.9485396383866482,
      "grad_norm": 0.09883379191160202,
      "learning_rate": 8.960469508469404e-09,
      "loss": 0.0003,
      "step": 180200
    },
    {
      "epoch": 2.9487032643377242,
      "grad_norm": 0.030171388760209084,
      "learning_rate": 8.903594673102533e-09,
      "loss": 0.0009,
      "step": 180210
    },
    {
      "epoch": 2.9488668902888,
      "grad_norm": 0.07790575176477432,
      "learning_rate": 8.846900752984e-09,
      "loss": 0.0003,
      "step": 180220
    },
    {
      "epoch": 2.9490305162398758,
      "grad_norm": 0.07743711769580841,
      "learning_rate": 8.790387750169938e-09,
      "loss": 0.0008,
      "step": 180230
    },
    {
      "epoch": 2.9491941421909513,
      "grad_norm": 0.0037441018503159285,
      "learning_rate": 8.734055666708153e-09,
      "loss": 0.0005,
      "step": 180240
    },
    {
      "epoch": 2.9493577681420273,
      "grad_norm": 0.02908901497721672,
      "learning_rate": 8.6779045046409e-09,
      "loss": 0.0007,
      "step": 180250
    },
    {
      "epoch": 2.9495213940931033,
      "grad_norm": 0.018190663307905197,
      "learning_rate": 8.621934266002663e-09,
      "loss": 0.0007,
      "step": 180260
    },
    {
      "epoch": 2.949685020044179,
      "grad_norm": 0.1992783546447754,
      "learning_rate": 8.566144952823485e-09,
      "loss": 0.0008,
      "step": 180270
    },
    {
      "epoch": 2.949848645995255,
      "grad_norm": 0.028692010790109634,
      "learning_rate": 8.510536567124528e-09,
      "loss": 0.0004,
      "step": 180280
    },
    {
      "epoch": 2.9500122719463304,
      "grad_norm": 0.07820817828178406,
      "learning_rate": 8.455109110921956e-09,
      "loss": 0.0007,
      "step": 180290
    },
    {
      "epoch": 2.9501758978974064,
      "grad_norm": 0.05783466249704361,
      "learning_rate": 8.399862586224716e-09,
      "loss": 0.0004,
      "step": 180300
    },
    {
      "epoch": 2.9503395238484824,
      "grad_norm": 0.012966896407306194,
      "learning_rate": 8.344796995036208e-09,
      "loss": 0.0009,
      "step": 180310
    },
    {
      "epoch": 2.950503149799558,
      "grad_norm": 0.02000482752919197,
      "learning_rate": 8.289912339350947e-09,
      "loss": 0.0011,
      "step": 180320
    },
    {
      "epoch": 2.950666775750634,
      "grad_norm": 0.08531814813613892,
      "learning_rate": 8.235208621159008e-09,
      "loss": 0.0008,
      "step": 180330
    },
    {
      "epoch": 2.95083040170171,
      "grad_norm": 0.008870822377502918,
      "learning_rate": 8.180685842443803e-09,
      "loss": 0.0032,
      "step": 180340
    },
    {
      "epoch": 2.9509940276527855,
      "grad_norm": 0.03043711557984352,
      "learning_rate": 8.126344005180975e-09,
      "loss": 0.0006,
      "step": 180350
    },
    {
      "epoch": 2.9511576536038615,
      "grad_norm": 0.04350176081061363,
      "learning_rate": 8.072183111340615e-09,
      "loss": 0.0006,
      "step": 180360
    },
    {
      "epoch": 2.9513212795549375,
      "grad_norm": 0.07302550971508026,
      "learning_rate": 8.018203162885597e-09,
      "loss": 0.0012,
      "step": 180370
    },
    {
      "epoch": 2.951484905506013,
      "grad_norm": 0.10953707247972488,
      "learning_rate": 7.964404161773242e-09,
      "loss": 0.0004,
      "step": 180380
    },
    {
      "epoch": 2.951648531457089,
      "grad_norm": 0.11266885697841644,
      "learning_rate": 7.910786109953105e-09,
      "loss": 0.0008,
      "step": 180390
    },
    {
      "epoch": 2.951812157408165,
      "grad_norm": 0.20524853467941284,
      "learning_rate": 7.85734900936863e-09,
      "loss": 0.0008,
      "step": 180400
    },
    {
      "epoch": 2.9519757833592406,
      "grad_norm": 0.018620336428284645,
      "learning_rate": 7.804092861957158e-09,
      "loss": 0.0005,
      "step": 180410
    },
    {
      "epoch": 2.9521394093103166,
      "grad_norm": 0.09038258343935013,
      "learning_rate": 7.751017669648254e-09,
      "loss": 0.0007,
      "step": 180420
    },
    {
      "epoch": 2.9523030352613926,
      "grad_norm": 0.10112320631742477,
      "learning_rate": 7.698123434367043e-09,
      "loss": 0.0005,
      "step": 180430
    },
    {
      "epoch": 2.952466661212468,
      "grad_norm": 0.006774208042770624,
      "learning_rate": 7.645410158029776e-09,
      "loss": 0.0007,
      "step": 180440
    },
    {
      "epoch": 2.952630287163544,
      "grad_norm": 0.09673327952623367,
      "learning_rate": 7.592877842548251e-09,
      "loss": 0.0007,
      "step": 180450
    },
    {
      "epoch": 2.95279391311462,
      "grad_norm": 0.3230985999107361,
      "learning_rate": 7.540526489825395e-09,
      "loss": 0.0004,
      "step": 180460
    },
    {
      "epoch": 2.9529575390656957,
      "grad_norm": 0.020928246900439262,
      "learning_rate": 7.488356101759686e-09,
      "loss": 0.0005,
      "step": 180470
    },
    {
      "epoch": 2.9531211650167717,
      "grad_norm": 0.10235882550477982,
      "learning_rate": 7.4363666802412804e-09,
      "loss": 0.0011,
      "step": 180480
    },
    {
      "epoch": 2.9532847909678477,
      "grad_norm": 0.03449095040559769,
      "learning_rate": 7.3845582271558914e-09,
      "loss": 0.0007,
      "step": 180490
    },
    {
      "epoch": 2.9534484169189232,
      "grad_norm": 0.08394557237625122,
      "learning_rate": 7.332930744380906e-09,
      "loss": 0.0005,
      "step": 180500
    },
    {
      "epoch": 2.9536120428699992,
      "grad_norm": 0.11212869733572006,
      "learning_rate": 7.28148423378705e-09,
      "loss": 0.0008,
      "step": 180510
    },
    {
      "epoch": 2.9537756688210752,
      "grad_norm": 0.10503393411636353,
      "learning_rate": 7.230218697240054e-09,
      "loss": 0.0006,
      "step": 180520
    },
    {
      "epoch": 2.953939294772151,
      "grad_norm": 0.102867491543293,
      "learning_rate": 7.1791341365978765e-09,
      "loss": 0.0004,
      "step": 180530
    },
    {
      "epoch": 2.9541029207232268,
      "grad_norm": 0.1311519891023636,
      "learning_rate": 7.128230553711813e-09,
      "loss": 0.0007,
      "step": 180540
    },
    {
      "epoch": 2.9542665466743028,
      "grad_norm": 0.0568668358027935,
      "learning_rate": 7.0775079504281645e-09,
      "loss": 0.0008,
      "step": 180550
    },
    {
      "epoch": 2.9544301726253783,
      "grad_norm": 0.06415213644504547,
      "learning_rate": 7.026966328583795e-09,
      "loss": 0.0011,
      "step": 180560
    },
    {
      "epoch": 2.9545937985764543,
      "grad_norm": 0.03379752114415169,
      "learning_rate": 6.976605690012239e-09,
      "loss": 0.0006,
      "step": 180570
    },
    {
      "epoch": 2.9547574245275303,
      "grad_norm": 0.04067510366439819,
      "learning_rate": 6.926426036537592e-09,
      "loss": 0.0004,
      "step": 180580
    },
    {
      "epoch": 2.954921050478606,
      "grad_norm": 0.06238815188407898,
      "learning_rate": 6.876427369980066e-09,
      "loss": 0.0009,
      "step": 180590
    },
    {
      "epoch": 2.955084676429682,
      "grad_norm": 0.04814431443810463,
      "learning_rate": 6.8266096921509876e-09,
      "loss": 0.0007,
      "step": 180600
    },
    {
      "epoch": 2.955248302380758,
      "grad_norm": 0.003641321789473295,
      "learning_rate": 6.776973004856691e-09,
      "loss": 0.0015,
      "step": 180610
    },
    {
      "epoch": 2.9554119283318334,
      "grad_norm": 0.07586206495761871,
      "learning_rate": 6.727517309896292e-09,
      "loss": 0.0004,
      "step": 180620
    },
    {
      "epoch": 2.9555755542829094,
      "grad_norm": 0.03522724658250809,
      "learning_rate": 6.678242609062802e-09,
      "loss": 0.0006,
      "step": 180630
    },
    {
      "epoch": 2.955739180233985,
      "grad_norm": 0.04871169105172157,
      "learning_rate": 6.629148904141458e-09,
      "loss": 0.0009,
      "step": 180640
    },
    {
      "epoch": 2.955902806185061,
      "grad_norm": 0.04193451255559921,
      "learning_rate": 6.580236196911949e-09,
      "loss": 0.0005,
      "step": 180650
    },
    {
      "epoch": 2.956066432136137,
      "grad_norm": 0.036512892693281174,
      "learning_rate": 6.5315044891478555e-09,
      "loss": 0.0005,
      "step": 180660
    },
    {
      "epoch": 2.9562300580872125,
      "grad_norm": 0.04258807748556137,
      "learning_rate": 6.4829537826149865e-09,
      "loss": 0.0007,
      "step": 180670
    },
    {
      "epoch": 2.9563936840382885,
      "grad_norm": 0.13228893280029297,
      "learning_rate": 6.434584079073603e-09,
      "loss": 0.0008,
      "step": 180680
    },
    {
      "epoch": 2.956557309989364,
      "grad_norm": 0.011979234404861927,
      "learning_rate": 6.386395380276744e-09,
      "loss": 0.0006,
      "step": 180690
    },
    {
      "epoch": 2.95672093594044,
      "grad_norm": 0.003308415412902832,
      "learning_rate": 6.338387687971348e-09,
      "loss": 0.0014,
      "step": 180700
    },
    {
      "epoch": 2.956884561891516,
      "grad_norm": 0.1035497784614563,
      "learning_rate": 6.290561003897688e-09,
      "loss": 0.0006,
      "step": 180710
    },
    {
      "epoch": 2.9570481878425916,
      "grad_norm": 0.177953839302063,
      "learning_rate": 6.242915329788823e-09,
      "loss": 0.0007,
      "step": 180720
    },
    {
      "epoch": 2.9572118137936676,
      "grad_norm": 0.004573258105665445,
      "learning_rate": 6.19545066737226e-09,
      "loss": 0.0004,
      "step": 180730
    },
    {
      "epoch": 2.9573754397447436,
      "grad_norm": 0.05622275173664093,
      "learning_rate": 6.148167018367734e-09,
      "loss": 0.0009,
      "step": 180740
    },
    {
      "epoch": 2.957539065695819,
      "grad_norm": 0.04541414603590965,
      "learning_rate": 6.101064384490541e-09,
      "loss": 0.0011,
      "step": 180750
    },
    {
      "epoch": 2.957702691646895,
      "grad_norm": 0.04543953016400337,
      "learning_rate": 6.054142767446536e-09,
      "loss": 0.0009,
      "step": 180760
    },
    {
      "epoch": 2.957866317597971,
      "grad_norm": 0.07907293736934662,
      "learning_rate": 6.0074021689376926e-09,
      "loss": 0.0008,
      "step": 180770
    },
    {
      "epoch": 2.9580299435490467,
      "grad_norm": 0.00527379522100091,
      "learning_rate": 5.960842590657656e-09,
      "loss": 0.0006,
      "step": 180780
    },
    {
      "epoch": 2.9581935695001227,
      "grad_norm": 0.09399823844432831,
      "learning_rate": 5.914464034293965e-09,
      "loss": 0.0009,
      "step": 180790
    },
    {
      "epoch": 2.9583571954511987,
      "grad_norm": 0.009637588635087013,
      "learning_rate": 5.868266501528608e-09,
      "loss": 0.0005,
      "step": 180800
    },
    {
      "epoch": 2.9585208214022742,
      "grad_norm": 0.09165926277637482,
      "learning_rate": 5.822249994035245e-09,
      "loss": 0.0007,
      "step": 180810
    },
    {
      "epoch": 2.9586844473533502,
      "grad_norm": 0.05676725506782532,
      "learning_rate": 5.776414513481987e-09,
      "loss": 0.0003,
      "step": 180820
    },
    {
      "epoch": 2.9588480733044262,
      "grad_norm": 0.047069769352674484,
      "learning_rate": 5.730760061530283e-09,
      "loss": 0.0004,
      "step": 180830
    },
    {
      "epoch": 2.959011699255502,
      "grad_norm": 0.035100530833005905,
      "learning_rate": 5.685286639834919e-09,
      "loss": 0.0011,
      "step": 180840
    },
    {
      "epoch": 2.9591753252065778,
      "grad_norm": 0.08382165431976318,
      "learning_rate": 5.639994250045133e-09,
      "loss": 0.0006,
      "step": 180850
    },
    {
      "epoch": 2.9593389511576538,
      "grad_norm": 0.02935672551393509,
      "learning_rate": 5.594882893801279e-09,
      "loss": 0.0008,
      "step": 180860
    },
    {
      "epoch": 2.9595025771087293,
      "grad_norm": 0.07367635518312454,
      "learning_rate": 5.549952572739825e-09,
      "loss": 0.0011,
      "step": 180870
    },
    {
      "epoch": 2.9596662030598053,
      "grad_norm": 0.018671972677111626,
      "learning_rate": 5.505203288488359e-09,
      "loss": 0.0007,
      "step": 180880
    },
    {
      "epoch": 2.9598298290108813,
      "grad_norm": 0.058731503784656525,
      "learning_rate": 5.460635042668916e-09,
      "loss": 0.0007,
      "step": 180890
    },
    {
      "epoch": 2.959993454961957,
      "grad_norm": 0.09905584901571274,
      "learning_rate": 5.416247836897981e-09,
      "loss": 0.0006,
      "step": 180900
    },
    {
      "epoch": 2.960157080913033,
      "grad_norm": 0.08817479014396667,
      "learning_rate": 5.372041672783712e-09,
      "loss": 0.0005,
      "step": 180910
    },
    {
      "epoch": 2.960320706864109,
      "grad_norm": 0.03982220217585564,
      "learning_rate": 5.3280165519287165e-09,
      "loss": 0.001,
      "step": 180920
    },
    {
      "epoch": 2.9604843328151844,
      "grad_norm": 0.08821292966604233,
      "learning_rate": 5.284172475928384e-09,
      "loss": 0.001,
      "step": 180930
    },
    {
      "epoch": 2.9606479587662604,
      "grad_norm": 0.0493648424744606,
      "learning_rate": 5.2405094463731095e-09,
      "loss": 0.0005,
      "step": 180940
    },
    {
      "epoch": 2.9608115847173364,
      "grad_norm": 0.10805744677782059,
      "learning_rate": 5.197027464843851e-09,
      "loss": 0.0007,
      "step": 180950
    },
    {
      "epoch": 2.960975210668412,
      "grad_norm": 0.003662973642349243,
      "learning_rate": 5.1537265329182355e-09,
      "loss": 0.0004,
      "step": 180960
    },
    {
      "epoch": 2.961138836619488,
      "grad_norm": 0.23632536828517914,
      "learning_rate": 5.110606652165007e-09,
      "loss": 0.0013,
      "step": 180970
    },
    {
      "epoch": 2.961302462570564,
      "grad_norm": 0.009528382681310177,
      "learning_rate": 5.0676678241473595e-09,
      "loss": 0.0007,
      "step": 180980
    },
    {
      "epoch": 2.9614660885216395,
      "grad_norm": 0.0876237154006958,
      "learning_rate": 5.024910050422382e-09,
      "loss": 0.0006,
      "step": 180990
    },
    {
      "epoch": 2.9616297144727155,
      "grad_norm": 0.03398312255740166,
      "learning_rate": 4.982333332538836e-09,
      "loss": 0.0008,
      "step": 181000
    },
    {
      "epoch": 2.961793340423791,
      "grad_norm": 0.05097678676247597,
      "learning_rate": 4.9399376720404845e-09,
      "loss": 0.0011,
      "step": 181010
    },
    {
      "epoch": 2.961956966374867,
      "grad_norm": 0.03587082773447037,
      "learning_rate": 4.897723070463878e-09,
      "loss": 0.0005,
      "step": 181020
    },
    {
      "epoch": 2.962120592325943,
      "grad_norm": 0.08462606370449066,
      "learning_rate": 4.855689529339458e-09,
      "loss": 0.0009,
      "step": 181030
    },
    {
      "epoch": 2.9622842182770186,
      "grad_norm": 0.002155303256586194,
      "learning_rate": 4.813837050191006e-09,
      "loss": 0.0009,
      "step": 181040
    },
    {
      "epoch": 2.9624478442280946,
      "grad_norm": 0.13236530125141144,
      "learning_rate": 4.772165634535086e-09,
      "loss": 0.0004,
      "step": 181050
    },
    {
      "epoch": 2.96261147017917,
      "grad_norm": 0.09248964488506317,
      "learning_rate": 4.730675283882713e-09,
      "loss": 0.0005,
      "step": 181060
    },
    {
      "epoch": 2.962775096130246,
      "grad_norm": 0.07607980817556381,
      "learning_rate": 4.689365999737128e-09,
      "loss": 0.0009,
      "step": 181070
    },
    {
      "epoch": 2.962938722081322,
      "grad_norm": 0.013493630103766918,
      "learning_rate": 4.648237783596576e-09,
      "loss": 0.0013,
      "step": 181080
    },
    {
      "epoch": 2.9631023480323977,
      "grad_norm": 0.00715222954750061,
      "learning_rate": 4.607290636951534e-09,
      "loss": 0.0003,
      "step": 181090
    },
    {
      "epoch": 2.9632659739834737,
      "grad_norm": 0.15975996851921082,
      "learning_rate": 4.566524561285812e-09,
      "loss": 0.0012,
      "step": 181100
    },
    {
      "epoch": 2.9634295999345497,
      "grad_norm": 0.11691099405288696,
      "learning_rate": 4.525939558077119e-09,
      "loss": 0.0005,
      "step": 181110
    },
    {
      "epoch": 2.9635932258856252,
      "grad_norm": 0.032300256192684174,
      "learning_rate": 4.485535628797055e-09,
      "loss": 0.0002,
      "step": 181120
    },
    {
      "epoch": 2.9637568518367012,
      "grad_norm": 0.03603142499923706,
      "learning_rate": 4.445312774910005e-09,
      "loss": 0.0009,
      "step": 181130
    },
    {
      "epoch": 2.9639204777877772,
      "grad_norm": 0.06174484267830849,
      "learning_rate": 4.40527099787369e-09,
      "loss": 0.0008,
      "step": 181140
    },
    {
      "epoch": 2.964084103738853,
      "grad_norm": 0.034598637372255325,
      "learning_rate": 4.365410299139727e-09,
      "loss": 0.0007,
      "step": 181150
    },
    {
      "epoch": 2.964247729689929,
      "grad_norm": 0.008270945399999619,
      "learning_rate": 4.325730680153073e-09,
      "loss": 0.0006,
      "step": 181160
    },
    {
      "epoch": 2.9644113556410048,
      "grad_norm": 0.11949373036623001,
      "learning_rate": 4.28623214235202e-09,
      "loss": 0.0005,
      "step": 181170
    },
    {
      "epoch": 2.9645749815920803,
      "grad_norm": 0.09367172420024872,
      "learning_rate": 4.246914687167647e-09,
      "loss": 0.0008,
      "step": 181180
    },
    {
      "epoch": 2.9647386075431563,
      "grad_norm": 0.1346801370382309,
      "learning_rate": 4.2077783160260346e-09,
      "loss": 0.0004,
      "step": 181190
    },
    {
      "epoch": 2.9649022334942323,
      "grad_norm": 0.07340075820684433,
      "learning_rate": 4.168823030345493e-09,
      "loss": 0.0007,
      "step": 181200
    },
    {
      "epoch": 2.965065859445308,
      "grad_norm": 0.007032521069049835,
      "learning_rate": 4.13004883153767e-09,
      "loss": 0.0015,
      "step": 181210
    },
    {
      "epoch": 2.965229485396384,
      "grad_norm": 0.047295257449150085,
      "learning_rate": 4.0914557210086634e-09,
      "loss": 0.0014,
      "step": 181220
    },
    {
      "epoch": 2.96539311134746,
      "grad_norm": 0.20061364769935608,
      "learning_rate": 4.053043700156245e-09,
      "loss": 0.0012,
      "step": 181230
    },
    {
      "epoch": 2.9655567372985354,
      "grad_norm": 0.037586621940135956,
      "learning_rate": 4.014812770374299e-09,
      "loss": 0.0006,
      "step": 181240
    },
    {
      "epoch": 2.9657203632496114,
      "grad_norm": 0.0829203873872757,
      "learning_rate": 3.976762933047829e-09,
      "loss": 0.001,
      "step": 181250
    },
    {
      "epoch": 2.9658839892006874,
      "grad_norm": 0.023354124277830124,
      "learning_rate": 3.9388941895557305e-09,
      "loss": 0.0013,
      "step": 181260
    },
    {
      "epoch": 2.966047615151763,
      "grad_norm": 0.05273908004164696,
      "learning_rate": 3.901206541271352e-09,
      "loss": 0.0012,
      "step": 181270
    },
    {
      "epoch": 2.966211241102839,
      "grad_norm": 0.06305355578660965,
      "learning_rate": 3.863699989560265e-09,
      "loss": 0.0004,
      "step": 181280
    },
    {
      "epoch": 2.966374867053915,
      "grad_norm": 0.05551360920071602,
      "learning_rate": 3.826374535781941e-09,
      "loss": 0.0007,
      "step": 181290
    },
    {
      "epoch": 2.9665384930049905,
      "grad_norm": 0.054925598204135895,
      "learning_rate": 3.789230181289738e-09,
      "loss": 0.0009,
      "step": 181300
    },
    {
      "epoch": 2.9667021189560665,
      "grad_norm": 0.08407167345285416,
      "learning_rate": 3.752266927430359e-09,
      "loss": 0.0008,
      "step": 181310
    },
    {
      "epoch": 2.9668657449071425,
      "grad_norm": 0.0009622799698263407,
      "learning_rate": 3.7154847755421775e-09,
      "loss": 0.0003,
      "step": 181320
    },
    {
      "epoch": 2.967029370858218,
      "grad_norm": 0.03719404712319374,
      "learning_rate": 3.678883726960236e-09,
      "loss": 0.0011,
      "step": 181330
    },
    {
      "epoch": 2.967192996809294,
      "grad_norm": 0.031055787578225136,
      "learning_rate": 3.642463783010142e-09,
      "loss": 0.0009,
      "step": 181340
    },
    {
      "epoch": 2.96735662276037,
      "grad_norm": 0.10827036947011948,
      "learning_rate": 3.606224945011949e-09,
      "loss": 0.0008,
      "step": 181350
    },
    {
      "epoch": 2.9675202487114456,
      "grad_norm": 0.03840309754014015,
      "learning_rate": 3.5701672142801626e-09,
      "loss": 0.0004,
      "step": 181360
    },
    {
      "epoch": 2.9676838746625216,
      "grad_norm": 0.12112395465373993,
      "learning_rate": 3.534290592120959e-09,
      "loss": 0.0006,
      "step": 181370
    },
    {
      "epoch": 2.9678475006135976,
      "grad_norm": 0.17998169362545013,
      "learning_rate": 3.4985950798349654e-09,
      "loss": 0.0011,
      "step": 181380
    },
    {
      "epoch": 2.968011126564673,
      "grad_norm": 0.0540277473628521,
      "learning_rate": 3.4630806787161463e-09,
      "loss": 0.001,
      "step": 181390
    },
    {
      "epoch": 2.968174752515749,
      "grad_norm": 0.05453220009803772,
      "learning_rate": 3.4277473900518055e-09,
      "loss": 0.0005,
      "step": 181400
    },
    {
      "epoch": 2.9683383784668247,
      "grad_norm": 0.11262725293636322,
      "learning_rate": 3.3925952151225848e-09,
      "loss": 0.0004,
      "step": 181410
    },
    {
      "epoch": 2.9685020044179007,
      "grad_norm": 0.07697691768407822,
      "learning_rate": 3.357624155203021e-09,
      "loss": 0.0006,
      "step": 181420
    },
    {
      "epoch": 2.9686656303689767,
      "grad_norm": 0.03737746551632881,
      "learning_rate": 3.322834211560433e-09,
      "loss": 0.0009,
      "step": 181430
    },
    {
      "epoch": 2.9688292563200522,
      "grad_norm": 0.07274530082941055,
      "learning_rate": 3.288225385455479e-09,
      "loss": 0.0009,
      "step": 181440
    },
    {
      "epoch": 2.9689928822711282,
      "grad_norm": 0.16672612726688385,
      "learning_rate": 3.2537976781432666e-09,
      "loss": 0.0008,
      "step": 181450
    },
    {
      "epoch": 2.969156508222204,
      "grad_norm": 0.06296700239181519,
      "learning_rate": 3.2195510908716865e-09,
      "loss": 0.0007,
      "step": 181460
    },
    {
      "epoch": 2.96932013417328,
      "grad_norm": 0.02773393504321575,
      "learning_rate": 3.1854856248819677e-09,
      "loss": 0.0007,
      "step": 181470
    },
    {
      "epoch": 2.9694837601243558,
      "grad_norm": 0.07432061433792114,
      "learning_rate": 3.1516012814086783e-09,
      "loss": 0.0006,
      "step": 181480
    },
    {
      "epoch": 2.9696473860754313,
      "grad_norm": 0.040386345237493515,
      "learning_rate": 3.11789806168028e-09,
      "loss": 0.0002,
      "step": 181490
    },
    {
      "epoch": 2.9698110120265073,
      "grad_norm": 0.04588755592703819,
      "learning_rate": 3.0843759669185736e-09,
      "loss": 0.0005,
      "step": 181500
    },
    {
      "epoch": 2.9699746379775833,
      "grad_norm": 0.005032803863286972,
      "learning_rate": 3.0510349983381428e-09,
      "loss": 0.0004,
      "step": 181510
    },
    {
      "epoch": 2.970138263928659,
      "grad_norm": 0.07821472734212875,
      "learning_rate": 3.0178751571485755e-09,
      "loss": 0.0008,
      "step": 181520
    },
    {
      "epoch": 2.970301889879735,
      "grad_norm": 0.27122142910957336,
      "learning_rate": 2.984896444550578e-09,
      "loss": 0.0008,
      "step": 181530
    },
    {
      "epoch": 2.970465515830811,
      "grad_norm": 0.08411692827939987,
      "learning_rate": 2.9520988617398603e-09,
      "loss": 0.0005,
      "step": 181540
    },
    {
      "epoch": 2.9706291417818864,
      "grad_norm": 0.056563347578048706,
      "learning_rate": 2.919482409905472e-09,
      "loss": 0.0022,
      "step": 181550
    },
    {
      "epoch": 2.9707927677329624,
      "grad_norm": 0.012228360399603844,
      "learning_rate": 2.8870470902298e-09,
      "loss": 0.0005,
      "step": 181560
    },
    {
      "epoch": 2.9709563936840384,
      "grad_norm": 0.08094001561403275,
      "learning_rate": 2.85479290388857e-09,
      "loss": 0.0011,
      "step": 181570
    },
    {
      "epoch": 2.971120019635114,
      "grad_norm": 0.15044502913951874,
      "learning_rate": 2.8227198520508482e-09,
      "loss": 0.0007,
      "step": 181580
    },
    {
      "epoch": 2.97128364558619,
      "grad_norm": 0.0014699194580316544,
      "learning_rate": 2.790827935879037e-09,
      "loss": 0.0004,
      "step": 181590
    },
    {
      "epoch": 2.971447271537266,
      "grad_norm": 0.24049873650074005,
      "learning_rate": 2.759117156529434e-09,
      "loss": 0.0008,
      "step": 181600
    },
    {
      "epoch": 2.9716108974883415,
      "grad_norm": 0.08181938529014587,
      "learning_rate": 2.727587515150565e-09,
      "loss": 0.0009,
      "step": 181610
    },
    {
      "epoch": 2.9717745234394175,
      "grad_norm": 0.05511162057518959,
      "learning_rate": 2.6962390128865147e-09,
      "loss": 0.0017,
      "step": 181620
    },
    {
      "epoch": 2.9719381493904935,
      "grad_norm": 0.027295531705021858,
      "learning_rate": 2.665071650873041e-09,
      "loss": 0.001,
      "step": 181630
    },
    {
      "epoch": 2.972101775341569,
      "grad_norm": 0.07224902510643005,
      "learning_rate": 2.634085430240352e-09,
      "loss": 0.0008,
      "step": 181640
    },
    {
      "epoch": 2.972265401292645,
      "grad_norm": 0.09082729369401932,
      "learning_rate": 2.6032803521108816e-09,
      "loss": 0.0009,
      "step": 181650
    },
    {
      "epoch": 2.972429027243721,
      "grad_norm": 0.0989830419421196,
      "learning_rate": 2.5726564176015156e-09,
      "loss": 0.0015,
      "step": 181660
    },
    {
      "epoch": 2.9725926531947966,
      "grad_norm": 0.005387819372117519,
      "learning_rate": 2.5422136278219213e-09,
      "loss": 0.0007,
      "step": 181670
    },
    {
      "epoch": 2.9727562791458726,
      "grad_norm": 0.07024528086185455,
      "learning_rate": 2.5119519838767705e-09,
      "loss": 0.0008,
      "step": 181680
    },
    {
      "epoch": 2.9729199050969486,
      "grad_norm": 0.058167263865470886,
      "learning_rate": 2.481871486861853e-09,
      "loss": 0.0005,
      "step": 181690
    },
    {
      "epoch": 2.973083531048024,
      "grad_norm": 0.0681462436914444,
      "learning_rate": 2.451972137867964e-09,
      "loss": 0.0006,
      "step": 181700
    },
    {
      "epoch": 2.9732471569991,
      "grad_norm": 0.010845375247299671,
      "learning_rate": 2.42225393797868e-09,
      "loss": 0.001,
      "step": 181710
    },
    {
      "epoch": 2.973410782950176,
      "grad_norm": 0.03632626309990883,
      "learning_rate": 2.3927168882714733e-09,
      "loss": 0.0003,
      "step": 181720
    },
    {
      "epoch": 2.9735744089012517,
      "grad_norm": 0.03113074228167534,
      "learning_rate": 2.363360989816599e-09,
      "loss": 0.0003,
      "step": 181730
    },
    {
      "epoch": 2.9737380348523277,
      "grad_norm": 0.009494964964687824,
      "learning_rate": 2.33418624367876e-09,
      "loss": 0.0007,
      "step": 181740
    },
    {
      "epoch": 2.9739016608034037,
      "grad_norm": 0.02749892883002758,
      "learning_rate": 2.305192650914889e-09,
      "loss": 0.0015,
      "step": 181750
    },
    {
      "epoch": 2.9740652867544792,
      "grad_norm": 0.03593337908387184,
      "learning_rate": 2.276380212576368e-09,
      "loss": 0.0004,
      "step": 181760
    },
    {
      "epoch": 2.9742289127055552,
      "grad_norm": 0.11148542910814285,
      "learning_rate": 2.2477489297068056e-09,
      "loss": 0.0008,
      "step": 181770
    },
    {
      "epoch": 2.974392538656631,
      "grad_norm": 0.051847826689481735,
      "learning_rate": 2.219298803344816e-09,
      "loss": 0.0006,
      "step": 181780
    },
    {
      "epoch": 2.9745561646077068,
      "grad_norm": 0.05268994718790054,
      "learning_rate": 2.1910298345217964e-09,
      "loss": 0.0006,
      "step": 181790
    },
    {
      "epoch": 2.9747197905587828,
      "grad_norm": 0.13268661499023438,
      "learning_rate": 2.1629420242613718e-09,
      "loss": 0.0006,
      "step": 181800
    },
    {
      "epoch": 2.9748834165098583,
      "grad_norm": 0.08490833640098572,
      "learning_rate": 2.1350353735827276e-09,
      "loss": 0.0015,
      "step": 181810
    },
    {
      "epoch": 2.9750470424609343,
      "grad_norm": 0.0894186943769455,
      "learning_rate": 2.107309883496722e-09,
      "loss": 0.0012,
      "step": 181820
    },
    {
      "epoch": 2.97521066841201,
      "grad_norm": 0.013229744508862495,
      "learning_rate": 2.0797655550092168e-09,
      "loss": 0.0012,
      "step": 181830
    },
    {
      "epoch": 2.975374294363086,
      "grad_norm": 0.07367993146181107,
      "learning_rate": 2.052402389117747e-09,
      "loss": 0.0006,
      "step": 181840
    },
    {
      "epoch": 2.975537920314162,
      "grad_norm": 0.02634665183722973,
      "learning_rate": 2.025220386814297e-09,
      "loss": 0.0011,
      "step": 181850
    },
    {
      "epoch": 2.9757015462652374,
      "grad_norm": 0.08619394898414612,
      "learning_rate": 1.9982195490841903e-09,
      "loss": 0.0006,
      "step": 181860
    },
    {
      "epoch": 2.9758651722163134,
      "grad_norm": 0.005855135153979063,
      "learning_rate": 1.9713998769060884e-09,
      "loss": 0.0005,
      "step": 181870
    },
    {
      "epoch": 2.9760287981673894,
      "grad_norm": 0.054620079696178436,
      "learning_rate": 1.9447613712525457e-09,
      "loss": 0.0009,
      "step": 181880
    },
    {
      "epoch": 2.976192424118465,
      "grad_norm": 0.045846857130527496,
      "learning_rate": 1.9183040330889024e-09,
      "loss": 0.0011,
      "step": 181890
    },
    {
      "epoch": 2.976356050069541,
      "grad_norm": 0.08647055178880692,
      "learning_rate": 1.8920278633743904e-09,
      "loss": 0.0006,
      "step": 181900
    },
    {
      "epoch": 2.976519676020617,
      "grad_norm": 0.0648687556385994,
      "learning_rate": 1.8659328630610264e-09,
      "loss": 0.0005,
      "step": 181910
    },
    {
      "epoch": 2.9766833019716925,
      "grad_norm": 0.09652049094438553,
      "learning_rate": 1.84001903309472e-09,
      "loss": 0.001,
      "step": 181920
    },
    {
      "epoch": 2.9768469279227685,
      "grad_norm": 0.14204098284244537,
      "learning_rate": 1.814286374415275e-09,
      "loss": 0.0006,
      "step": 181930
    },
    {
      "epoch": 2.9770105538738445,
      "grad_norm": 0.04247429966926575,
      "learning_rate": 1.788734887954724e-09,
      "loss": 0.001,
      "step": 181940
    },
    {
      "epoch": 2.97717417982492,
      "grad_norm": 0.058686546981334686,
      "learning_rate": 1.7633645746401028e-09,
      "loss": 0.001,
      "step": 181950
    },
    {
      "epoch": 2.977337805775996,
      "grad_norm": 0.05955192446708679,
      "learning_rate": 1.7381754353906765e-09,
      "loss": 0.002,
      "step": 181960
    },
    {
      "epoch": 2.977501431727072,
      "grad_norm": 0.026648730039596558,
      "learning_rate": 1.7131674711190483e-09,
      "loss": 0.0005,
      "step": 181970
    },
    {
      "epoch": 2.9776650576781476,
      "grad_norm": 0.3306833505630493,
      "learning_rate": 1.688340682731715e-09,
      "loss": 0.0008,
      "step": 181980
    },
    {
      "epoch": 2.9778286836292236,
      "grad_norm": 0.11923018097877502,
      "learning_rate": 1.663695071129623e-09,
      "loss": 0.0008,
      "step": 181990
    },
    {
      "epoch": 2.9779923095802996,
      "grad_norm": 0.003203323343768716,
      "learning_rate": 1.639230637204836e-09,
      "loss": 0.0004,
      "step": 182000
    },
    {
      "epoch": 2.978155935531375,
      "grad_norm": 0.06617830693721771,
      "learning_rate": 1.614947381844978e-09,
      "loss": 0.0007,
      "step": 182010
    },
    {
      "epoch": 2.978319561482451,
      "grad_norm": 0.01506265252828598,
      "learning_rate": 1.5908453059299001e-09,
      "loss": 0.0009,
      "step": 182020
    },
    {
      "epoch": 2.978483187433527,
      "grad_norm": 0.03575095906853676,
      "learning_rate": 1.566924410333348e-09,
      "loss": 0.0014,
      "step": 182030
    },
    {
      "epoch": 2.9786468133846027,
      "grad_norm": 0.18627874553203583,
      "learning_rate": 1.543184695921851e-09,
      "loss": 0.0009,
      "step": 182040
    },
    {
      "epoch": 2.9788104393356787,
      "grad_norm": 0.03441288694739342,
      "learning_rate": 1.519626163556942e-09,
      "loss": 0.0005,
      "step": 182050
    },
    {
      "epoch": 2.9789740652867547,
      "grad_norm": 0.23827986419200897,
      "learning_rate": 1.4962488140923826e-09,
      "loss": 0.0005,
      "step": 182060
    },
    {
      "epoch": 2.9791376912378302,
      "grad_norm": 0.021551622077822685,
      "learning_rate": 1.4730526483747177e-09,
      "loss": 0.0006,
      "step": 182070
    },
    {
      "epoch": 2.9793013171889062,
      "grad_norm": 0.09389097988605499,
      "learning_rate": 1.4500376672454963e-09,
      "loss": 0.0006,
      "step": 182080
    },
    {
      "epoch": 2.9794649431399822,
      "grad_norm": 0.02499030902981758,
      "learning_rate": 1.4272038715384962e-09,
      "loss": 0.0003,
      "step": 182090
    },
    {
      "epoch": 2.9796285690910578,
      "grad_norm": 0.01591278612613678,
      "learning_rate": 1.4045512620819434e-09,
      "loss": 0.0005,
      "step": 182100
    },
    {
      "epoch": 2.9797921950421338,
      "grad_norm": 0.08829034864902496,
      "learning_rate": 1.3820798396962932e-09,
      "loss": 0.0007,
      "step": 182110
    },
    {
      "epoch": 2.9799558209932098,
      "grad_norm": 0.029104532673954964,
      "learning_rate": 1.3597896051964487e-09,
      "loss": 0.0005,
      "step": 182120
    },
    {
      "epoch": 2.9801194469442853,
      "grad_norm": 0.05610715597867966,
      "learning_rate": 1.3376805593900977e-09,
      "loss": 0.0006,
      "step": 182130
    },
    {
      "epoch": 2.9802830728953613,
      "grad_norm": 0.04652120918035507,
      "learning_rate": 1.315752703079376e-09,
      "loss": 0.0004,
      "step": 182140
    },
    {
      "epoch": 2.9804466988464373,
      "grad_norm": 0.1480550915002823,
      "learning_rate": 1.2940060370580932e-09,
      "loss": 0.0013,
      "step": 182150
    },
    {
      "epoch": 2.980610324797513,
      "grad_norm": 0.03793784976005554,
      "learning_rate": 1.2724405621150626e-09,
      "loss": 0.0008,
      "step": 182160
    },
    {
      "epoch": 2.980773950748589,
      "grad_norm": 0.03641463816165924,
      "learning_rate": 1.2510562790324365e-09,
      "loss": 0.001,
      "step": 182170
    },
    {
      "epoch": 2.9809375766996644,
      "grad_norm": 0.044259123504161835,
      "learning_rate": 1.2298531885840403e-09,
      "loss": 0.0004,
      "step": 182180
    },
    {
      "epoch": 2.9811012026507404,
      "grad_norm": 0.021113863214850426,
      "learning_rate": 1.2088312915398137e-09,
      "loss": 0.0006,
      "step": 182190
    },
    {
      "epoch": 2.9812648286018164,
      "grad_norm": 0.06147037446498871,
      "learning_rate": 1.1879905886613696e-09,
      "loss": 0.0007,
      "step": 182200
    },
    {
      "epoch": 2.981428454552892,
      "grad_norm": 0.046186938881874084,
      "learning_rate": 1.1673310807036598e-09,
      "loss": 0.0004,
      "step": 182210
    },
    {
      "epoch": 2.981592080503968,
      "grad_norm": 0.038473181426525116,
      "learning_rate": 1.1468527684155295e-09,
      "loss": 0.0003,
      "step": 182220
    },
    {
      "epoch": 2.9817557064550435,
      "grad_norm": 0.05932663753628731,
      "learning_rate": 1.126555652539718e-09,
      "loss": 0.0004,
      "step": 182230
    },
    {
      "epoch": 2.9819193324061195,
      "grad_norm": 0.06103590503334999,
      "learning_rate": 1.1064397338123034e-09,
      "loss": 0.0007,
      "step": 182240
    },
    {
      "epoch": 2.9820829583571955,
      "grad_norm": 0.09790539741516113,
      "learning_rate": 1.0865050129610366e-09,
      "loss": 0.0006,
      "step": 182250
    },
    {
      "epoch": 2.982246584308271,
      "grad_norm": 0.14284656941890717,
      "learning_rate": 1.0667514907103383e-09,
      "loss": 0.0011,
      "step": 182260
    },
    {
      "epoch": 2.982410210259347,
      "grad_norm": 0.08526279777288437,
      "learning_rate": 1.0471791677746367e-09,
      "loss": 0.0007,
      "step": 182270
    },
    {
      "epoch": 2.982573836210423,
      "grad_norm": 0.026263464242219925,
      "learning_rate": 1.02778804486392e-09,
      "loss": 0.0006,
      "step": 182280
    },
    {
      "epoch": 2.9827374621614986,
      "grad_norm": 0.04552841931581497,
      "learning_rate": 1.008578122681514e-09,
      "loss": 0.0025,
      "step": 182290
    },
    {
      "epoch": 2.9829010881125746,
      "grad_norm": 0.047282617539167404,
      "learning_rate": 9.895494019235285e-10,
      "loss": 0.0009,
      "step": 182300
    },
    {
      "epoch": 2.9830647140636506,
      "grad_norm": 0.08310357481241226,
      "learning_rate": 9.707018832794124e-10,
      "loss": 0.0005,
      "step": 182310
    },
    {
      "epoch": 2.983228340014726,
      "grad_norm": 0.08298409730195999,
      "learning_rate": 9.520355674325078e-10,
      "loss": 0.0003,
      "step": 182320
    },
    {
      "epoch": 2.983391965965802,
      "grad_norm": 0.031720537692308426,
      "learning_rate": 9.335504550594954e-10,
      "loss": 0.0005,
      "step": 182330
    },
    {
      "epoch": 2.983555591916878,
      "grad_norm": 0.1097896620631218,
      "learning_rate": 9.1524654682984e-10,
      "loss": 0.0008,
      "step": 182340
    },
    {
      "epoch": 2.9837192178679537,
      "grad_norm": 0.11241824179887772,
      "learning_rate": 8.971238434080098e-10,
      "loss": 0.0006,
      "step": 182350
    },
    {
      "epoch": 2.9838828438190297,
      "grad_norm": 0.030461788177490234,
      "learning_rate": 8.791823454501469e-10,
      "loss": 0.0002,
      "step": 182360
    },
    {
      "epoch": 2.9840464697701057,
      "grad_norm": 0.09207808971405029,
      "learning_rate": 8.614220536073969e-10,
      "loss": 0.0006,
      "step": 182370
    },
    {
      "epoch": 2.9842100957211812,
      "grad_norm": 0.05260929837822914,
      "learning_rate": 8.438429685225791e-10,
      "loss": 0.0008,
      "step": 182380
    },
    {
      "epoch": 2.9843737216722572,
      "grad_norm": 0.12333384156227112,
      "learning_rate": 8.264450908329613e-10,
      "loss": 0.0008,
      "step": 182390
    },
    {
      "epoch": 2.9845373476233332,
      "grad_norm": 0.04292740672826767,
      "learning_rate": 8.092284211697055e-10,
      "loss": 0.0004,
      "step": 182400
    },
    {
      "epoch": 2.9847009735744088,
      "grad_norm": 0.08580801635980606,
      "learning_rate": 7.92192960157312e-10,
      "loss": 0.0006,
      "step": 182410
    },
    {
      "epoch": 2.9848645995254848,
      "grad_norm": 0.12693290412425995,
      "learning_rate": 7.753387084119545e-10,
      "loss": 0.0008,
      "step": 182420
    },
    {
      "epoch": 2.9850282254765608,
      "grad_norm": 0.07805383205413818,
      "learning_rate": 7.586656665453662e-10,
      "loss": 0.0006,
      "step": 182430
    },
    {
      "epoch": 2.9851918514276363,
      "grad_norm": 0.021498166024684906,
      "learning_rate": 7.421738351620634e-10,
      "loss": 0.0007,
      "step": 182440
    },
    {
      "epoch": 2.9853554773787123,
      "grad_norm": 0.009048529900610447,
      "learning_rate": 7.258632148593459e-10,
      "loss": 0.0009,
      "step": 182450
    },
    {
      "epoch": 2.9855191033297883,
      "grad_norm": 0.06429654359817505,
      "learning_rate": 7.097338062284075e-10,
      "loss": 0.0042,
      "step": 182460
    },
    {
      "epoch": 2.985682729280864,
      "grad_norm": 0.1947738230228424,
      "learning_rate": 6.937856098543361e-10,
      "loss": 0.0012,
      "step": 182470
    },
    {
      "epoch": 2.98584635523194,
      "grad_norm": 0.01577065698802471,
      "learning_rate": 6.780186263155574e-10,
      "loss": 0.0003,
      "step": 182480
    },
    {
      "epoch": 2.986009981183016,
      "grad_norm": 0.07304892688989639,
      "learning_rate": 6.624328561821713e-10,
      "loss": 0.0009,
      "step": 182490
    },
    {
      "epoch": 2.9861736071340914,
      "grad_norm": 0.04181336984038353,
      "learning_rate": 6.470283000203914e-10,
      "loss": 0.0004,
      "step": 182500
    },
    {
      "epoch": 2.9863372330851674,
      "grad_norm": 0.017054041847586632,
      "learning_rate": 6.318049583886598e-10,
      "loss": 0.0003,
      "step": 182510
    },
    {
      "epoch": 2.9865008590362434,
      "grad_norm": 0.016465414315462112,
      "learning_rate": 6.167628318376473e-10,
      "loss": 0.0006,
      "step": 182520
    },
    {
      "epoch": 2.986664484987319,
      "grad_norm": 0.01617741771042347,
      "learning_rate": 6.019019209135835e-10,
      "loss": 0.0005,
      "step": 182530
    },
    {
      "epoch": 2.986828110938395,
      "grad_norm": 0.048710986971855164,
      "learning_rate": 5.872222261549265e-10,
      "loss": 0.0004,
      "step": 182540
    },
    {
      "epoch": 2.9869917368894705,
      "grad_norm": 0.16679561138153076,
      "learning_rate": 5.727237480934733e-10,
      "loss": 0.0008,
      "step": 182550
    },
    {
      "epoch": 2.9871553628405465,
      "grad_norm": 0.14510859549045563,
      "learning_rate": 5.584064872549144e-10,
      "loss": 0.0015,
      "step": 182560
    },
    {
      "epoch": 2.9873189887916225,
      "grad_norm": 0.11583229899406433,
      "learning_rate": 5.442704441582791e-10,
      "loss": 0.0013,
      "step": 182570
    },
    {
      "epoch": 2.987482614742698,
      "grad_norm": 0.029368869960308075,
      "learning_rate": 5.303156193164904e-10,
      "loss": 0.0006,
      "step": 182580
    },
    {
      "epoch": 2.987646240693774,
      "grad_norm": 0.0253090038895607,
      "learning_rate": 5.165420132341448e-10,
      "loss": 0.0008,
      "step": 182590
    },
    {
      "epoch": 2.9878098666448496,
      "grad_norm": 0.0328531488776207,
      "learning_rate": 5.029496264119526e-10,
      "loss": 0.0005,
      "step": 182600
    },
    {
      "epoch": 2.9879734925959256,
      "grad_norm": 0.09279884397983551,
      "learning_rate": 4.895384593411878e-10,
      "loss": 0.0007,
      "step": 182610
    },
    {
      "epoch": 2.9881371185470016,
      "grad_norm": 0.05588304623961449,
      "learning_rate": 4.76308512508683e-10,
      "loss": 0.0008,
      "step": 182620
    },
    {
      "epoch": 2.988300744498077,
      "grad_norm": 0.08876816183328629,
      "learning_rate": 4.632597863946098e-10,
      "loss": 0.0007,
      "step": 182630
    },
    {
      "epoch": 2.988464370449153,
      "grad_norm": 0.08923585712909698,
      "learning_rate": 4.503922814708128e-10,
      "loss": 0.001,
      "step": 182640
    },
    {
      "epoch": 2.988627996400229,
      "grad_norm": 0.15535946190357208,
      "learning_rate": 4.3770599820414095e-10,
      "loss": 0.0006,
      "step": 182650
    },
    {
      "epoch": 2.9887916223513047,
      "grad_norm": 0.08236636966466904,
      "learning_rate": 4.252009370547816e-10,
      "loss": 0.0005,
      "step": 182660
    },
    {
      "epoch": 2.9889552483023807,
      "grad_norm": 0.034059662371873856,
      "learning_rate": 4.128770984751507e-10,
      "loss": 0.0004,
      "step": 182670
    },
    {
      "epoch": 2.9891188742534567,
      "grad_norm": 0.1988143026828766,
      "learning_rate": 4.007344829132231e-10,
      "loss": 0.0006,
      "step": 182680
    },
    {
      "epoch": 2.9892825002045322,
      "grad_norm": 0.29730433225631714,
      "learning_rate": 3.8877309080809224e-10,
      "loss": 0.0009,
      "step": 182690
    },
    {
      "epoch": 2.9894461261556082,
      "grad_norm": 0.011211794801056385,
      "learning_rate": 3.769929225938551e-10,
      "loss": 0.0005,
      "step": 182700
    },
    {
      "epoch": 2.9896097521066842,
      "grad_norm": 0.07937619835138321,
      "learning_rate": 3.6539397869683745e-10,
      "loss": 0.0008,
      "step": 182710
    },
    {
      "epoch": 2.98977337805776,
      "grad_norm": 0.06813499331474304,
      "learning_rate": 3.539762595383689e-10,
      "loss": 0.0013,
      "step": 182720
    },
    {
      "epoch": 2.9899370040088358,
      "grad_norm": 0.02483954094350338,
      "learning_rate": 3.427397655320075e-10,
      "loss": 0.0004,
      "step": 182730
    },
    {
      "epoch": 2.9901006299599118,
      "grad_norm": 0.05213353410363197,
      "learning_rate": 3.316844970852051e-10,
      "loss": 0.0006,
      "step": 182740
    },
    {
      "epoch": 2.9902642559109873,
      "grad_norm": 0.1355007290840149,
      "learning_rate": 3.2081045459819713e-10,
      "loss": 0.0002,
      "step": 182750
    },
    {
      "epoch": 2.9904278818620633,
      "grad_norm": 0.04843417927622795,
      "learning_rate": 3.101176384651128e-10,
      "loss": 0.0007,
      "step": 182760
    },
    {
      "epoch": 2.9905915078131393,
      "grad_norm": 0.030777323991060257,
      "learning_rate": 2.9960604907397497e-10,
      "loss": 0.0003,
      "step": 182770
    },
    {
      "epoch": 2.990755133764215,
      "grad_norm": 0.20406000316143036,
      "learning_rate": 2.892756868055901e-10,
      "loss": 0.0011,
      "step": 182780
    },
    {
      "epoch": 2.990918759715291,
      "grad_norm": 0.07328087091445923,
      "learning_rate": 2.791265520346587e-10,
      "loss": 0.0013,
      "step": 182790
    },
    {
      "epoch": 2.991082385666367,
      "grad_norm": 0.05528191849589348,
      "learning_rate": 2.691586451292194e-10,
      "loss": 0.0004,
      "step": 182800
    },
    {
      "epoch": 2.9912460116174424,
      "grad_norm": 0.06187997758388519,
      "learning_rate": 2.5937196644953975e-10,
      "loss": 0.0005,
      "step": 182810
    },
    {
      "epoch": 2.9914096375685184,
      "grad_norm": 0.06434860825538635,
      "learning_rate": 2.4976651635144623e-10,
      "loss": 0.001,
      "step": 182820
    },
    {
      "epoch": 2.9915732635195944,
      "grad_norm": 0.10044960677623749,
      "learning_rate": 2.4034229518299366e-10,
      "loss": 0.001,
      "step": 182830
    },
    {
      "epoch": 2.99173688947067,
      "grad_norm": 0.15940852463245392,
      "learning_rate": 2.3109930328502062e-10,
      "loss": 0.0006,
      "step": 182840
    },
    {
      "epoch": 2.991900515421746,
      "grad_norm": 0.011512136086821556,
      "learning_rate": 2.2203754099336949e-10,
      "loss": 0.0007,
      "step": 182850
    },
    {
      "epoch": 2.992064141372822,
      "grad_norm": 0.07312177866697311,
      "learning_rate": 2.1315700863611122e-10,
      "loss": 0.0004,
      "step": 182860
    },
    {
      "epoch": 2.9922277673238975,
      "grad_norm": 0.09036201983690262,
      "learning_rate": 2.0445770653521046e-10,
      "loss": 0.0005,
      "step": 182870
    },
    {
      "epoch": 2.9923913932749735,
      "grad_norm": 0.011012058705091476,
      "learning_rate": 1.9593963500597056e-10,
      "loss": 0.0005,
      "step": 182880
    },
    {
      "epoch": 2.9925550192260495,
      "grad_norm": 0.04315563663840294,
      "learning_rate": 1.8760279435703354e-10,
      "loss": 0.001,
      "step": 182890
    },
    {
      "epoch": 2.992718645177125,
      "grad_norm": 0.011310326866805553,
      "learning_rate": 1.7944718489149027e-10,
      "loss": 0.0002,
      "step": 182900
    },
    {
      "epoch": 2.992882271128201,
      "grad_norm": 0.04459337517619133,
      "learning_rate": 1.7147280690354984e-10,
      "loss": 0.0003,
      "step": 182910
    },
    {
      "epoch": 2.993045897079277,
      "grad_norm": 0.055467452853918076,
      "learning_rate": 1.6367966068298047e-10,
      "loss": 0.0004,
      "step": 182920
    },
    {
      "epoch": 2.9932095230303526,
      "grad_norm": 0.13603198528289795,
      "learning_rate": 1.5606774651233392e-10,
      "loss": 0.0005,
      "step": 182930
    },
    {
      "epoch": 2.9933731489814286,
      "grad_norm": 0.015258897095918655,
      "learning_rate": 1.486370646675006e-10,
      "loss": 0.0002,
      "step": 182940
    },
    {
      "epoch": 2.993536774932504,
      "grad_norm": 0.04858879745006561,
      "learning_rate": 1.4138761541770963e-10,
      "loss": 0.0004,
      "step": 182950
    },
    {
      "epoch": 2.99370040088358,
      "grad_norm": 0.0620223730802536,
      "learning_rate": 1.3431939902608383e-10,
      "loss": 0.0007,
      "step": 182960
    },
    {
      "epoch": 2.993864026834656,
      "grad_norm": 0.052294228225946426,
      "learning_rate": 1.274324157479745e-10,
      "loss": 0.001,
      "step": 182970
    },
    {
      "epoch": 2.9940276527857317,
      "grad_norm": 0.04354868084192276,
      "learning_rate": 1.2072666583429205e-10,
      "loss": 0.0006,
      "step": 182980
    },
    {
      "epoch": 2.9941912787368077,
      "grad_norm": 0.2174597680568695,
      "learning_rate": 1.1420214952650998e-10,
      "loss": 0.0006,
      "step": 182990
    },
    {
      "epoch": 2.9943549046878832,
      "grad_norm": 0.0406862236559391,
      "learning_rate": 1.0785886706277116e-10,
      "loss": 0.002,
      "step": 183000
    },
    {
      "epoch": 2.9945185306389592,
      "grad_norm": 0.09220794588327408,
      "learning_rate": 1.0169681867178149e-10,
      "loss": 0.0005,
      "step": 183010
    },
    {
      "epoch": 2.9946821565900352,
      "grad_norm": 0.20663510262966156,
      "learning_rate": 9.571600457725094e-11,
      "loss": 0.0014,
      "step": 183020
    },
    {
      "epoch": 2.994845782541111,
      "grad_norm": 0.08783263713121414,
      "learning_rate": 8.99164249962281e-11,
      "loss": 0.0005,
      "step": 183030
    },
    {
      "epoch": 2.9950094084921868,
      "grad_norm": 0.0773741602897644,
      "learning_rate": 8.429808013854513e-11,
      "loss": 0.0003,
      "step": 183040
    },
    {
      "epoch": 2.9951730344432628,
      "grad_norm": 0.03865004703402519,
      "learning_rate": 7.886097020848304e-11,
      "loss": 0.0006,
      "step": 183050
    },
    {
      "epoch": 2.9953366603943383,
      "grad_norm": 0.04568766802549362,
      "learning_rate": 7.360509540255135e-11,
      "loss": 0.0007,
      "step": 183060
    },
    {
      "epoch": 2.9955002863454143,
      "grad_norm": 0.0493677519261837,
      "learning_rate": 6.853045591115325e-11,
      "loss": 0.0007,
      "step": 183070
    },
    {
      "epoch": 2.9956639122964903,
      "grad_norm": 0.11075688898563385,
      "learning_rate": 6.363705191914094e-11,
      "loss": 0.0006,
      "step": 183080
    },
    {
      "epoch": 2.995827538247566,
      "grad_norm": 0.044846996665000916,
      "learning_rate": 5.892488360248471e-11,
      "loss": 0.0011,
      "step": 183090
    },
    {
      "epoch": 2.995991164198642,
      "grad_norm": 0.12038176506757736,
      "learning_rate": 5.4393951133269174e-11,
      "loss": 0.0006,
      "step": 183100
    },
    {
      "epoch": 2.996154790149718,
      "grad_norm": 0.007127447985112667,
      "learning_rate": 5.0044254675252203e-11,
      "loss": 0.0004,
      "step": 183110
    },
    {
      "epoch": 2.9963184161007934,
      "grad_norm": 0.12169210612773895,
      "learning_rate": 4.587579438553036e-11,
      "loss": 0.0008,
      "step": 183120
    },
    {
      "epoch": 2.9964820420518694,
      "grad_norm": 0.20229792594909668,
      "learning_rate": 4.18885704162042e-11,
      "loss": 0.0006,
      "step": 183130
    },
    {
      "epoch": 2.9966456680029454,
      "grad_norm": 0.09927067160606384,
      "learning_rate": 3.8082582911602715e-11,
      "loss": 0.0007,
      "step": 183140
    },
    {
      "epoch": 2.996809293954021,
      "grad_norm": 0.08279476314783096,
      "learning_rate": 3.4457832008838456e-11,
      "loss": 0.0005,
      "step": 183150
    },
    {
      "epoch": 2.996972919905097,
      "grad_norm": 0.11548232287168503,
      "learning_rate": 3.101431784002795e-11,
      "loss": 0.0008,
      "step": 183160
    },
    {
      "epoch": 2.997136545856173,
      "grad_norm": 0.01916341483592987,
      "learning_rate": 2.77520405300713e-11,
      "loss": 0.0006,
      "step": 183170
    },
    {
      "epoch": 2.9973001718072485,
      "grad_norm": 0.09040320664644241,
      "learning_rate": 2.467100019665214e-11,
      "loss": 0.0002,
      "step": 183180
    },
    {
      "epoch": 2.9974637977583245,
      "grad_norm": 0.013019285164773464,
      "learning_rate": 2.1771196951902994e-11,
      "loss": 0.0005,
      "step": 183190
    },
    {
      "epoch": 2.9976274237094005,
      "grad_norm": 0.037323277443647385,
      "learning_rate": 1.9052630900739944e-11,
      "loss": 0.0008,
      "step": 183200
    },
    {
      "epoch": 2.997791049660476,
      "grad_norm": 0.16036146879196167,
      "learning_rate": 1.6515302141972835e-11,
      "loss": 0.001,
      "step": 183210
    },
    {
      "epoch": 2.997954675611552,
      "grad_norm": 0.06244174391031265,
      "learning_rate": 1.4159210767750176e-11,
      "loss": 0.0012,
      "step": 183220
    },
    {
      "epoch": 2.998118301562628,
      "grad_norm": 0.10027268528938293,
      "learning_rate": 1.1984356862448921e-11,
      "loss": 0.0005,
      "step": 183230
    },
    {
      "epoch": 2.9982819275137036,
      "grad_norm": 0.03646072745323181,
      "learning_rate": 9.990740506005126e-12,
      "loss": 0.0007,
      "step": 183240
    },
    {
      "epoch": 2.9984455534647796,
      "grad_norm": 0.035138629376888275,
      "learning_rate": 8.178361770028176e-12,
      "loss": 0.0006,
      "step": 183250
    },
    {
      "epoch": 2.9986091794158556,
      "grad_norm": 0.03447127342224121,
      "learning_rate": 6.54722072002123e-12,
      "loss": 0.0007,
      "step": 183260
    },
    {
      "epoch": 2.998772805366931,
      "grad_norm": 0.08323480188846588,
      "learning_rate": 5.097317415936331e-12,
      "loss": 0.0009,
      "step": 183270
    },
    {
      "epoch": 2.998936431318007,
      "grad_norm": 0.21281728148460388,
      "learning_rate": 3.828651909953962e-12,
      "loss": 0.0005,
      "step": 183280
    },
    {
      "epoch": 2.999100057269083,
      "grad_norm": 0.03264172002673149,
      "learning_rate": 2.7412242481483775e-12,
      "loss": 0.0009,
      "step": 183290
    },
    {
      "epoch": 2.9992636832201587,
      "grad_norm": 0.014256869442760944,
      "learning_rate": 1.8350344693773834e-12,
      "loss": 0.0007,
      "step": 183300
    },
    {
      "epoch": 2.9994273091712347,
      "grad_norm": 0.006932792719453573,
      "learning_rate": 1.110082606947671e-12,
      "loss": 0.0006,
      "step": 183310
    },
    {
      "epoch": 2.9995909351223107,
      "grad_norm": 0.10003849864006042,
      "learning_rate": 5.663686875045926e-13,
      "loss": 0.0008,
      "step": 183320
    },
    {
      "epoch": 2.9997545610733862,
      "grad_norm": 0.04948035627603531,
      "learning_rate": 2.0389272992193953e-13,
      "loss": 0.0005,
      "step": 183330
    },
    {
      "epoch": 2.9999181870244622,
      "grad_norm": 0.09036850929260254,
      "learning_rate": 2.2654748077499677e-14,
      "loss": 0.0006,
      "step": 183340
    },
    {
      "epoch": 3.0,
      "step": 183345,
      "total_flos": 5.837921389771751e+18,
      "train_loss": 0.004886563129512852,
      "train_runtime": 102730.9024,
      "train_samples_per_second": 14.278,
      "train_steps_per_second": 1.785
    }
  ],
  "logging_steps": 10,
  "max_steps": 183345,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.837921389771751e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}