diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10125 @@
+{
+  "best_metric": 0.039273809641599655,
+  "best_model_checkpoint": "./data/EuroSAT_output/checkpoint-14345",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 14345,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003485535029627048,
+      "grad_norm": 2.583953380584717,
+      "learning_rate": 1.9986057859881492e-05,
+      "loss": 2.3035,
+      "step": 10
+    },
+    {
+      "epoch": 0.006971070059254096,
+      "grad_norm": 3.04152774810791,
+      "learning_rate": 1.9972115719762987e-05,
+      "loss": 2.2087,
+      "step": 20
+    },
+    {
+      "epoch": 0.010456605088881143,
+      "grad_norm": 2.454313039779663,
+      "learning_rate": 1.9958173579644477e-05,
+      "loss": 2.179,
+      "step": 30
+    },
+    {
+      "epoch": 0.013942140118508192,
+      "grad_norm": 2.8994503021240234,
+      "learning_rate": 1.9944231439525968e-05,
+      "loss": 2.1153,
+      "step": 40
+    },
+    {
+      "epoch": 0.01742767514813524,
+      "grad_norm": 2.569394111633301,
+      "learning_rate": 1.9930289299407462e-05,
+      "loss": 2.0797,
+      "step": 50
+    },
+    {
+      "epoch": 0.020913210177762286,
+      "grad_norm": 2.860028028488159,
+      "learning_rate": 1.9916347159288953e-05,
+      "loss": 1.9448,
+      "step": 60
+    },
+    {
+      "epoch": 0.024398745207389334,
+      "grad_norm": 3.2729475498199463,
+      "learning_rate": 1.9902405019170444e-05,
+      "loss": 1.8354,
+      "step": 70
+    },
+    {
+      "epoch": 0.027884280237016383,
+      "grad_norm": 3.2359659671783447,
+      "learning_rate": 1.9888462879051935e-05,
+      "loss": 1.8398,
+      "step": 80
+    },
+    {
+      "epoch": 0.03136981526664343,
+      "grad_norm": 3.4146273136138916,
+      "learning_rate": 1.9874520738933426e-05,
+      "loss": 1.7559,
+      "step": 90
+    },
+    {
+      "epoch": 0.03485535029627048,
+      "grad_norm": 3.6460695266723633,
+      "learning_rate": 1.986057859881492e-05,
+      "loss": 1.7249,
+      "step": 100
+    },
+    {
+      "epoch": 0.03834088532589752,
+      "grad_norm": 3.8710196018218994,
+      "learning_rate": 1.984663645869641e-05,
+      "loss": 1.6526,
+      "step": 110
+    },
+    {
+      "epoch": 0.04182642035552457,
+      "grad_norm": 3.0699777603149414,
+      "learning_rate": 1.98326943185779e-05,
+      "loss": 1.4757,
+      "step": 120
+    },
+    {
+      "epoch": 0.04531195538515162,
+      "grad_norm": 3.8974356651306152,
+      "learning_rate": 1.9818752178459396e-05,
+      "loss": 1.5169,
+      "step": 130
+    },
+    {
+      "epoch": 0.04879749041477867,
+      "grad_norm": 4.513424873352051,
+      "learning_rate": 1.9804810038340887e-05,
+      "loss": 1.4487,
+      "step": 140
+    },
+    {
+      "epoch": 0.05228302544440572,
+      "grad_norm": 3.8483262062072754,
+      "learning_rate": 1.9790867898222377e-05,
+      "loss": 1.4174,
+      "step": 150
+    },
+    {
+      "epoch": 0.055768560474032766,
+      "grad_norm": 3.938114643096924,
+      "learning_rate": 1.977692575810387e-05,
+      "loss": 1.2989,
+      "step": 160
+    },
+    {
+      "epoch": 0.059254095503659815,
+      "grad_norm": 4.86665678024292,
+      "learning_rate": 1.9762983617985362e-05,
+      "loss": 1.3303,
+      "step": 170
+    },
+    {
+      "epoch": 0.06273963053328686,
+      "grad_norm": 2.8868749141693115,
+      "learning_rate": 1.9749041477866853e-05,
+      "loss": 1.2444,
+      "step": 180
+    },
+    {
+      "epoch": 0.06622516556291391,
+      "grad_norm": 2.838812828063965,
+      "learning_rate": 1.9735099337748347e-05,
+      "loss": 1.0791,
+      "step": 190
+    },
+    {
+      "epoch": 0.06971070059254096,
+      "grad_norm": 3.111001968383789,
+      "learning_rate": 1.9721157197629838e-05,
+      "loss": 1.1166,
+      "step": 200
+    },
+    {
+      "epoch": 0.073196235622168,
+      "grad_norm": 4.862301826477051,
+      "learning_rate": 1.970721505751133e-05,
+      "loss": 1.053,
+      "step": 210
+    },
+    {
+      "epoch": 0.07668177065179504,
+      "grad_norm": 3.7676494121551514,
+      "learning_rate": 1.969327291739282e-05,
+      "loss": 1.0223,
+      "step": 220
+    },
+    {
+      "epoch": 0.0801673056814221,
+      "grad_norm": 4.827447414398193,
+      "learning_rate": 1.9679330777274314e-05,
+      "loss": 1.0277,
+      "step": 230
+    },
+    {
+      "epoch": 0.08365284071104914,
+      "grad_norm": 2.027529716491699,
+      "learning_rate": 1.9665388637155805e-05,
+      "loss": 0.8187,
+      "step": 240
+    },
+    {
+      "epoch": 0.08713837574067619,
+      "grad_norm": 5.508512020111084,
+      "learning_rate": 1.9651446497037296e-05,
+      "loss": 0.9514,
+      "step": 250
+    },
+    {
+      "epoch": 0.09062391077030324,
+      "grad_norm": 4.481660842895508,
+      "learning_rate": 1.963750435691879e-05,
+      "loss": 0.9503,
+      "step": 260
+    },
+    {
+      "epoch": 0.09410944579993029,
+      "grad_norm": 2.511476993560791,
+      "learning_rate": 1.962356221680028e-05,
+      "loss": 0.9108,
+      "step": 270
+    },
+    {
+      "epoch": 0.09759498082955734,
+      "grad_norm": 5.287069320678711,
+      "learning_rate": 1.960962007668177e-05,
+      "loss": 0.8772,
+      "step": 280
+    },
+    {
+      "epoch": 0.10108051585918439,
+      "grad_norm": 3.6010959148406982,
+      "learning_rate": 1.9595677936563266e-05,
+      "loss": 0.793,
+      "step": 290
+    },
+    {
+      "epoch": 0.10456605088881143,
+      "grad_norm": 4.5751776695251465,
+      "learning_rate": 1.9581735796444757e-05,
+      "loss": 0.7934,
+      "step": 300
+    },
+    {
+      "epoch": 0.10805158591843848,
+      "grad_norm": 2.073023796081543,
+      "learning_rate": 1.9567793656326247e-05,
+      "loss": 0.7081,
+      "step": 310
+    },
+    {
+      "epoch": 0.11153712094806553,
+      "grad_norm": 5.257015228271484,
+      "learning_rate": 1.955385151620774e-05,
+      "loss": 0.7023,
+      "step": 320
+    },
+    {
+      "epoch": 0.11502265597769258,
+      "grad_norm": 5.037456035614014,
+      "learning_rate": 1.9539909376089232e-05,
+      "loss": 0.705,
+      "step": 330
+    },
+    {
+      "epoch": 0.11850819100731963,
+      "grad_norm": 2.995671510696411,
+      "learning_rate": 1.9525967235970723e-05,
+      "loss": 0.654,
+      "step": 340
+    },
+    {
+      "epoch": 0.12199372603694666,
+      "grad_norm": 4.856020450592041,
+      "learning_rate": 1.9512025095852214e-05,
+      "loss": 0.6384,
+      "step": 350
+    },
+    {
+      "epoch": 0.12547926106657373,
+      "grad_norm": 4.719180107116699,
+      "learning_rate": 1.9498082955733705e-05,
+      "loss": 0.6655,
+      "step": 360
+    },
+    {
+      "epoch": 0.12896479609620076,
+      "grad_norm": 7.83165979385376,
+      "learning_rate": 1.94841408156152e-05,
+      "loss": 0.5098,
+      "step": 370
+    },
+    {
+      "epoch": 0.13245033112582782,
+      "grad_norm": 3.3765158653259277,
+      "learning_rate": 1.947019867549669e-05,
+      "loss": 0.6758,
+      "step": 380
+    },
+    {
+      "epoch": 0.13593586615545486,
+      "grad_norm": 4.360907554626465,
+      "learning_rate": 1.945625653537818e-05,
+      "loss": 0.6017,
+      "step": 390
+    },
+    {
+      "epoch": 0.13942140118508192,
+      "grad_norm": 3.162569522857666,
+      "learning_rate": 1.9442314395259675e-05,
+      "loss": 0.6965,
+      "step": 400
+    },
+    {
+      "epoch": 0.14290693621470896,
+      "grad_norm": 5.929446697235107,
+      "learning_rate": 1.9428372255141166e-05,
+      "loss": 0.6226,
+      "step": 410
+    },
+    {
+      "epoch": 0.146392471244336,
+      "grad_norm": 2.6196141242980957,
+      "learning_rate": 1.9414430115022657e-05,
+      "loss": 0.5276,
+      "step": 420
+    },
+    {
+      "epoch": 0.14987800627396305,
+      "grad_norm": 4.234184265136719,
+      "learning_rate": 1.940048797490415e-05,
+      "loss": 0.7085,
+      "step": 430
+    },
+    {
+      "epoch": 0.1533635413035901,
+      "grad_norm": 3.8775856494903564,
+      "learning_rate": 1.938654583478564e-05,
+      "loss": 0.7142,
+      "step": 440
+    },
+    {
+      "epoch": 0.15684907633321715,
+      "grad_norm": 2.744117498397827,
+      "learning_rate": 1.9372603694667132e-05,
+      "loss": 0.606,
+      "step": 450
+    },
+    {
+      "epoch": 0.1603346113628442,
+      "grad_norm": 4.7066969871521,
+      "learning_rate": 1.9358661554548627e-05,
+      "loss": 0.5027,
+      "step": 460
+    },
+    {
+      "epoch": 0.16382014639247125,
+      "grad_norm": 6.463034152984619,
+      "learning_rate": 1.9344719414430117e-05,
+      "loss": 0.5223,
+      "step": 470
+    },
+    {
+      "epoch": 0.16730568142209828,
+      "grad_norm": 9.271512985229492,
+      "learning_rate": 1.9330777274311608e-05,
+      "loss": 0.6037,
+      "step": 480
+    },
+    {
+      "epoch": 0.17079121645172535,
+      "grad_norm": 14.535713195800781,
+      "learning_rate": 1.93168351341931e-05,
+      "loss": 0.5479,
+      "step": 490
+    },
+    {
+      "epoch": 0.17427675148135238,
+      "grad_norm": 8.693368911743164,
+      "learning_rate": 1.930289299407459e-05,
+      "loss": 0.6084,
+      "step": 500
+    },
+    {
+      "epoch": 0.17776228651097944,
+      "grad_norm": 5.175466537475586,
+      "learning_rate": 1.9288950853956084e-05,
+      "loss": 0.6587,
+      "step": 510
+    },
+    {
+      "epoch": 0.18124782154060648,
+      "grad_norm": 2.11287522315979,
+      "learning_rate": 1.9275008713837575e-05,
+      "loss": 0.49,
+      "step": 520
+    },
+    {
+      "epoch": 0.18473335657023354,
+      "grad_norm": 2.2307288646698,
+      "learning_rate": 1.9261066573719066e-05,
+      "loss": 0.6132,
+      "step": 530
+    },
+    {
+      "epoch": 0.18821889159986058,
+      "grad_norm": 11.079753875732422,
+      "learning_rate": 1.924712443360056e-05,
+      "loss": 0.572,
+      "step": 540
+    },
+    {
+      "epoch": 0.19170442662948764,
+      "grad_norm": 2.487985610961914,
+      "learning_rate": 1.923318229348205e-05,
+      "loss": 0.3897,
+      "step": 550
+    },
+    {
+      "epoch": 0.19518996165911467,
+      "grad_norm": 3.330732822418213,
+      "learning_rate": 1.921924015336354e-05,
+      "loss": 0.3884,
+      "step": 560
+    },
+    {
+      "epoch": 0.1986754966887417,
+      "grad_norm": 9.687525749206543,
+      "learning_rate": 1.9205298013245036e-05,
+      "loss": 0.4155,
+      "step": 570
+    },
+    {
+      "epoch": 0.20216103171836877,
+      "grad_norm": 5.554359436035156,
+      "learning_rate": 1.9191355873126526e-05,
+      "loss": 0.5435,
+      "step": 580
+    },
+    {
+      "epoch": 0.2056465667479958,
+      "grad_norm": 2.3810067176818848,
+      "learning_rate": 1.917741373300802e-05,
+      "loss": 0.4487,
+      "step": 590
+    },
+    {
+      "epoch": 0.20913210177762287,
+      "grad_norm": 2.6795718669891357,
+      "learning_rate": 1.916347159288951e-05,
+      "loss": 0.5523,
+      "step": 600
+    },
+    {
+      "epoch": 0.2126176368072499,
+      "grad_norm": 11.402462005615234,
+      "learning_rate": 1.9149529452771002e-05,
+      "loss": 0.6798,
+      "step": 610
+    },
+    {
+      "epoch": 0.21610317183687697,
+      "grad_norm": 4.038872241973877,
+      "learning_rate": 1.9135587312652493e-05,
+      "loss": 0.3321,
+      "step": 620
+    },
+    {
+      "epoch": 0.219588706866504,
+      "grad_norm": 0.8252215385437012,
+      "learning_rate": 1.9121645172533984e-05,
+      "loss": 0.4267,
+      "step": 630
+    },
+    {
+      "epoch": 0.22307424189613106,
+      "grad_norm": 3.365762233734131,
+      "learning_rate": 1.9107703032415478e-05,
+      "loss": 0.5648,
+      "step": 640
+    },
+    {
+      "epoch": 0.2265597769257581,
+      "grad_norm": 3.1780312061309814,
+      "learning_rate": 1.909376089229697e-05,
+      "loss": 0.4703,
+      "step": 650
+    },
+    {
+      "epoch": 0.23004531195538516,
+      "grad_norm": 1.3519302606582642,
+      "learning_rate": 1.907981875217846e-05,
+      "loss": 0.4887,
+      "step": 660
+    },
+    {
+      "epoch": 0.2335308469850122,
+      "grad_norm": 2.861978054046631,
+      "learning_rate": 1.9065876612059954e-05,
+      "loss": 0.5227,
+      "step": 670
+    },
+    {
+      "epoch": 0.23701638201463926,
+      "grad_norm": 9.564464569091797,
+      "learning_rate": 1.9051934471941445e-05,
+      "loss": 0.3791,
+      "step": 680
+    },
+    {
+      "epoch": 0.2405019170442663,
+      "grad_norm": 6.82305383682251,
+      "learning_rate": 1.9037992331822936e-05,
+      "loss": 0.5024,
+      "step": 690
+    },
+    {
+      "epoch": 0.24398745207389333,
+      "grad_norm": 0.8181214332580566,
+      "learning_rate": 1.902405019170443e-05,
+      "loss": 0.4091,
+      "step": 700
+    },
+    {
+      "epoch": 0.2474729871035204,
+      "grad_norm": 11.525835990905762,
+      "learning_rate": 1.901010805158592e-05,
+      "loss": 0.3477,
+      "step": 710
+    },
+    {
+      "epoch": 0.25095852213314745,
+      "grad_norm": 2.1437933444976807,
+      "learning_rate": 1.899616591146741e-05,
+      "loss": 0.3997,
+      "step": 720
+    },
+    {
+      "epoch": 0.25444405716277446,
+      "grad_norm": 14.871402740478516,
+      "learning_rate": 1.8982223771348906e-05,
+      "loss": 0.4577,
+      "step": 730
+    },
+    {
+      "epoch": 0.2579295921924015,
+      "grad_norm": 1.370771050453186,
+      "learning_rate": 1.8968281631230396e-05,
+      "loss": 0.2808,
+      "step": 740
+    },
+    {
+      "epoch": 0.2614151272220286,
+      "grad_norm": 14.198845863342285,
+      "learning_rate": 1.8954339491111887e-05,
+      "loss": 0.4222,
+      "step": 750
+    },
+    {
+      "epoch": 0.26490066225165565,
+      "grad_norm": 0.7197427749633789,
+      "learning_rate": 1.8940397350993378e-05,
+      "loss": 0.3704,
+      "step": 760
+    },
+    {
+      "epoch": 0.26838619728128266,
+      "grad_norm": 8.696059226989746,
+      "learning_rate": 1.892645521087487e-05,
+      "loss": 0.3291,
+      "step": 770
+    },
+    {
+      "epoch": 0.2718717323109097,
+      "grad_norm": 0.5741353034973145,
+      "learning_rate": 1.8912513070756363e-05,
+      "loss": 0.2361,
+      "step": 780
+    },
+    {
+      "epoch": 0.2753572673405368,
+      "grad_norm": 3.235818386077881,
+      "learning_rate": 1.8898570930637854e-05,
+      "loss": 0.4824,
+      "step": 790
+    },
+    {
+      "epoch": 0.27884280237016384,
+      "grad_norm": 0.8503827452659607,
+      "learning_rate": 1.8884628790519345e-05,
+      "loss": 0.3376,
+      "step": 800
+    },
+    {
+      "epoch": 0.28232833739979085,
+      "grad_norm": 1.6630173921585083,
+      "learning_rate": 1.887068665040084e-05,
+      "loss": 0.3883,
+      "step": 810
+    },
+    {
+      "epoch": 0.2858138724294179,
+      "grad_norm": 10.400406837463379,
+      "learning_rate": 1.885674451028233e-05,
+      "loss": 0.4094,
+      "step": 820
+    },
+    {
+      "epoch": 0.289299407459045,
+      "grad_norm": 1.3901340961456299,
+      "learning_rate": 1.884280237016382e-05,
+      "loss": 0.5164,
+      "step": 830
+    },
+    {
+      "epoch": 0.292784942488672,
+      "grad_norm": 5.530642509460449,
+      "learning_rate": 1.8828860230045315e-05,
+      "loss": 0.3125,
+      "step": 840
+    },
+    {
+      "epoch": 0.29627047751829905,
+      "grad_norm": 6.685891151428223,
+      "learning_rate": 1.8814918089926806e-05,
+      "loss": 0.3943,
+      "step": 850
+    },
+    {
+      "epoch": 0.2997560125479261,
+      "grad_norm": 15.618851661682129,
+      "learning_rate": 1.8800975949808296e-05,
+      "loss": 0.3619,
+      "step": 860
+    },
+    {
+      "epoch": 0.30324154757755317,
+      "grad_norm": 9.351723670959473,
+      "learning_rate": 1.878703380968979e-05,
+      "loss": 0.4237,
+      "step": 870
+    },
+    {
+      "epoch": 0.3067270826071802,
+      "grad_norm": 3.25188946723938,
+      "learning_rate": 1.877309166957128e-05,
+      "loss": 0.3106,
+      "step": 880
+    },
+    {
+      "epoch": 0.31021261763680724,
+      "grad_norm": 3.0205211639404297,
+      "learning_rate": 1.8759149529452772e-05,
+      "loss": 0.3932,
+      "step": 890
+    },
+    {
+      "epoch": 0.3136981526664343,
+      "grad_norm": 0.8111696243286133,
+      "learning_rate": 1.8745207389334263e-05,
+      "loss": 0.3497,
+      "step": 900
+    },
+    {
+      "epoch": 0.31718368769606137,
+      "grad_norm": 1.2595362663269043,
+      "learning_rate": 1.8731265249215754e-05,
+      "loss": 0.4002,
+      "step": 910
+    },
+    {
+      "epoch": 0.3206692227256884,
+      "grad_norm": 0.6795431971549988,
+      "learning_rate": 1.8717323109097248e-05,
+      "loss": 0.3498,
+      "step": 920
+    },
+    {
+      "epoch": 0.32415475775531544,
+      "grad_norm": 1.0169798135757446,
+      "learning_rate": 1.870338096897874e-05,
+      "loss": 0.2839,
+      "step": 930
+    },
+    {
+      "epoch": 0.3276402927849425,
+      "grad_norm": 9.539275169372559,
+      "learning_rate": 1.868943882886023e-05,
+      "loss": 0.3325,
+      "step": 940
+    },
+    {
+      "epoch": 0.33112582781456956,
+      "grad_norm": 8.384815216064453,
+      "learning_rate": 1.8675496688741724e-05,
+      "loss": 0.3253,
+      "step": 950
+    },
+    {
+      "epoch": 0.33461136284419657,
+      "grad_norm": 4.2581281661987305,
+      "learning_rate": 1.8661554548623215e-05,
+      "loss": 0.3612,
+      "step": 960
+    },
+    {
+      "epoch": 0.33809689787382363,
+      "grad_norm": 1.6423273086547852,
+      "learning_rate": 1.8647612408504706e-05,
+      "loss": 0.3876,
+      "step": 970
+    },
+    {
+      "epoch": 0.3415824329034507,
+      "grad_norm": 3.153898000717163,
+      "learning_rate": 1.86336702683862e-05,
+      "loss": 0.4319,
+      "step": 980
+    },
+    {
+      "epoch": 0.3450679679330777,
+      "grad_norm": 0.48236197233200073,
+      "learning_rate": 1.861972812826769e-05,
+      "loss": 0.1993,
+      "step": 990
+    },
+    {
+      "epoch": 0.34855350296270476,
+      "grad_norm": 0.4741288423538208,
+      "learning_rate": 1.8605785988149185e-05,
+      "loss": 0.3256,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3520390379923318,
+      "grad_norm": 10.781346321105957,
+      "learning_rate": 1.8591843848030676e-05,
+      "loss": 0.4149,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3555245730219589,
+      "grad_norm": 0.4621415436267853,
+      "learning_rate": 1.8577901707912166e-05,
+      "loss": 0.3206,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3590101080515859,
+      "grad_norm": 6.963232040405273,
+      "learning_rate": 1.8563959567793657e-05,
+      "loss": 0.3789,
+      "step": 1030
+    },
+    {
+      "epoch": 0.36249564308121296,
+      "grad_norm": 6.472696781158447,
+      "learning_rate": 1.8550017427675148e-05,
+      "loss": 0.2999,
+      "step": 1040
+    },
+    {
+      "epoch": 0.36598117811084,
+      "grad_norm": 8.689462661743164,
+      "learning_rate": 1.8536075287556642e-05,
+      "loss": 0.2934,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3694667131404671,
+      "grad_norm": 0.33935561776161194,
+      "learning_rate": 1.8522133147438133e-05,
+      "loss": 0.2774,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3729522481700941,
+      "grad_norm": 10.953472137451172,
+      "learning_rate": 1.8508191007319624e-05,
+      "loss": 0.3193,
+      "step": 1070
+    },
+    {
+      "epoch": 0.37643778319972115,
+      "grad_norm": 1.503441333770752,
+      "learning_rate": 1.8494248867201118e-05,
+      "loss": 0.3257,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3799233182293482,
+      "grad_norm": 0.4625272750854492,
+      "learning_rate": 1.848030672708261e-05,
+      "loss": 0.2634,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3834088532589753,
+      "grad_norm": 11.566400527954102,
+      "learning_rate": 1.84663645869641e-05,
+      "loss": 0.2718,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3868943882886023,
+      "grad_norm": 12.47866439819336,
+      "learning_rate": 1.8452422446845594e-05,
+      "loss": 0.3186,
+      "step": 1110
+    },
+    {
+      "epoch": 0.39037992331822935,
+      "grad_norm": 8.613419532775879,
+      "learning_rate": 1.8438480306727085e-05,
+      "loss": 0.3122,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3938654583478564,
+      "grad_norm": 0.35111886262893677,
+      "learning_rate": 1.8424538166608576e-05,
+      "loss": 0.2495,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3973509933774834,
+      "grad_norm": 0.34094560146331787,
+      "learning_rate": 1.841059602649007e-05,
+      "loss": 0.2337,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4008365284071105,
+      "grad_norm": 4.428099155426025,
+      "learning_rate": 1.839665388637156e-05,
+      "loss": 0.1903,
+      "step": 1150
+    },
+    {
+      "epoch": 0.40432206343673754,
+      "grad_norm": 0.3854966461658478,
+      "learning_rate": 1.838271174625305e-05,
+      "loss": 0.3,
+      "step": 1160
+    },
+    {
+      "epoch": 0.4078075984663646,
+      "grad_norm": 38.84294509887695,
+      "learning_rate": 1.8368769606134542e-05,
+      "loss": 0.2134,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4112931334959916,
+      "grad_norm": 1.8577314615249634,
+      "learning_rate": 1.8354827466016033e-05,
+      "loss": 0.2276,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4147786685256187,
+      "grad_norm": 0.9145134091377258,
+      "learning_rate": 1.8340885325897527e-05,
+      "loss": 0.2989,
+      "step": 1190
+    },
+    {
+      "epoch": 0.41826420355524574,
+      "grad_norm": 13.17152214050293,
+      "learning_rate": 1.8326943185779018e-05,
+      "loss": 0.2938,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4217497385848728,
+      "grad_norm": 0.34265998005867004,
+      "learning_rate": 1.831300104566051e-05,
+      "loss": 0.2257,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4252352736144998,
+      "grad_norm": 10.514459609985352,
+      "learning_rate": 1.8299058905542003e-05,
+      "loss": 0.382,
+      "step": 1220
+    },
+    {
+      "epoch": 0.42872080864412687,
+      "grad_norm": 16.24938201904297,
+      "learning_rate": 1.8285116765423494e-05,
+      "loss": 0.3638,
+      "step": 1230
+    },
+    {
+      "epoch": 0.43220634367375393,
+      "grad_norm": 2.323594331741333,
+      "learning_rate": 1.8271174625304985e-05,
+      "loss": 0.3058,
+      "step": 1240
+    },
+    {
+      "epoch": 0.43569187870338094,
+      "grad_norm": 0.8403987288475037,
+      "learning_rate": 1.825723248518648e-05,
+      "loss": 0.1748,
+      "step": 1250
+    },
+    {
+      "epoch": 0.439177413733008,
+      "grad_norm": 0.3924444913864136,
+      "learning_rate": 1.824329034506797e-05,
+      "loss": 0.2051,
+      "step": 1260
+    },
+    {
+      "epoch": 0.44266294876263507,
+      "grad_norm": 1.6590919494628906,
+      "learning_rate": 1.822934820494946e-05,
+      "loss": 0.1713,
+      "step": 1270
+    },
+    {
+      "epoch": 0.44614848379226213,
+      "grad_norm": 10.351805686950684,
+      "learning_rate": 1.8215406064830955e-05,
+      "loss": 0.2202,
+      "step": 1280
+    },
+    {
+      "epoch": 0.44963401882188914,
+      "grad_norm": 13.177727699279785,
+      "learning_rate": 1.8201463924712445e-05,
+      "loss": 0.2005,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4531195538515162,
+      "grad_norm": 3.9433481693267822,
+      "learning_rate": 1.8187521784593936e-05,
+      "loss": 0.2948,
+      "step": 1300
+    },
+    {
+      "epoch": 0.45660508888114326,
+      "grad_norm": 0.3618289530277252,
+      "learning_rate": 1.8173579644475427e-05,
+      "loss": 0.148,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4600906239107703,
+      "grad_norm": 10.462213516235352,
+      "learning_rate": 1.8159637504356918e-05,
+      "loss": 0.2468,
+      "step": 1320
+    },
+    {
+      "epoch": 0.46357615894039733,
+      "grad_norm": 0.270712286233902,
+      "learning_rate": 1.8145695364238412e-05,
+      "loss": 0.2283,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4670616939700244,
+      "grad_norm": 0.30840155482292175,
+      "learning_rate": 1.8131753224119903e-05,
+      "loss": 0.1058,
+      "step": 1340
+    },
+    {
+      "epoch": 0.47054722899965146,
+      "grad_norm": 7.213718414306641,
+      "learning_rate": 1.8117811084001394e-05,
+      "loss": 0.2548,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4740327640292785,
+      "grad_norm": 4.371262550354004,
+      "learning_rate": 1.8103868943882888e-05,
+      "loss": 0.2346,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4775182990589055,
+      "grad_norm": 5.258789539337158,
+      "learning_rate": 1.808992680376438e-05,
+      "loss": 0.3612,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4810038340885326,
+      "grad_norm": 0.29574891924858093,
+      "learning_rate": 1.8075984663645873e-05,
+      "loss": 0.2976,
+      "step": 1380
+    },
+    {
+      "epoch": 0.48448936911815965,
+      "grad_norm": 0.7350137233734131,
+      "learning_rate": 1.8062042523527364e-05,
+      "loss": 0.1397,
+      "step": 1390
+    },
+    {
+      "epoch": 0.48797490414778666,
+      "grad_norm": 0.22566929459571838,
+      "learning_rate": 1.8048100383408855e-05,
+      "loss": 0.2482,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4914604391774137,
+      "grad_norm": 4.66607141494751,
+      "learning_rate": 1.803415824329035e-05,
+      "loss": 0.2287,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4949459742070408,
+      "grad_norm": 1.6220591068267822,
+      "learning_rate": 1.802021610317184e-05,
+      "loss": 0.2499,
+      "step": 1420
+    },
+    {
+      "epoch": 0.49843150923666785,
+      "grad_norm": 9.882462501525879,
+      "learning_rate": 1.800627396305333e-05,
+      "loss": 0.1941,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5019170442662949,
+      "grad_norm": 9.781709671020508,
+      "learning_rate": 1.799233182293482e-05,
+      "loss": 0.2008,
+      "step": 1440
+    },
+    {
+      "epoch": 0.5054025792959219,
+      "grad_norm": 20.839242935180664,
+      "learning_rate": 1.7978389682816312e-05,
+      "loss": 0.4964,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5088881143255489,
+      "grad_norm": 14.371743202209473,
+      "learning_rate": 1.7964447542697806e-05,
+      "loss": 0.3222,
+      "step": 1460
+    },
+    {
+      "epoch": 0.512373649355176,
+      "grad_norm": 0.32212069630622864,
+      "learning_rate": 1.7950505402579297e-05,
+      "loss": 0.2794,
+      "step": 1470
+    },
+    {
+      "epoch": 0.515859184384803,
+      "grad_norm": 0.4930780827999115,
+      "learning_rate": 1.7936563262460788e-05,
+      "loss": 0.0921,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5193447194144302,
+      "grad_norm": 2.0369691848754883,
+      "learning_rate": 1.7922621122342282e-05,
+      "loss": 0.2109,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5228302544440572,
+      "grad_norm": 1.3328112363815308,
+      "learning_rate": 1.7908678982223773e-05,
+      "loss": 0.2685,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.5157321691513062,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.343,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5298013245033113,
+      "grad_norm": 24.59377670288086,
+      "learning_rate": 1.7880794701986758e-05,
+      "loss": 0.257,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5332868595329383,
+      "grad_norm": 0.25167328119277954,
+      "learning_rate": 1.786685256186825e-05,
+      "loss": 0.2256,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5367723945625653,
+      "grad_norm": 0.31769871711730957,
+      "learning_rate": 1.785291042174974e-05,
+      "loss": 0.2461,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5402579295921924,
+      "grad_norm": 4.568899631500244,
+      "learning_rate": 1.7838968281631234e-05,
+      "loss": 0.1961,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5437434646218194,
+      "grad_norm": 3.1680665016174316,
+      "learning_rate": 1.7825026141512725e-05,
+      "loss": 0.2128,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5472289996514464,
+      "grad_norm": 4.516887187957764,
+      "learning_rate": 1.7811084001394215e-05,
+      "loss": 0.2493,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5507145346810736,
+      "grad_norm": 9.322068214416504,
+      "learning_rate": 1.7797141861275706e-05,
+      "loss": 0.1234,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5542000697107006,
+      "grad_norm": 0.5057626366615295,
+      "learning_rate": 1.7783199721157197e-05,
+      "loss": 0.1955,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5576856047403277,
+      "grad_norm": 3.936608076095581,
+      "learning_rate": 1.776925758103869e-05,
+      "loss": 0.4223,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5611711397699547,
+      "grad_norm": 1.4608641862869263,
+      "learning_rate": 1.7755315440920182e-05,
+      "loss": 0.2389,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5646566747995817,
+      "grad_norm": 0.2644886076450348,
+      "learning_rate": 1.7741373300801673e-05,
+      "loss": 0.2196,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5681422098292088,
+      "grad_norm": 8.523164749145508,
+      "learning_rate": 1.7727431160683167e-05,
+      "loss": 0.4851,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5716277448588358,
+      "grad_norm": 0.24561679363250732,
+      "learning_rate": 1.7713489020564658e-05,
+      "loss": 0.1291,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5751132798884628,
+      "grad_norm": 3.906648874282837,
+      "learning_rate": 1.769954688044615e-05,
+      "loss": 0.1317,
+      "step": 1650
+    },
+    {
+      "epoch": 0.57859881491809,
+      "grad_norm": 12.78685474395752,
+      "learning_rate": 1.7685604740327643e-05,
+      "loss": 0.2986,
+      "step": 1660
+    },
+    {
+      "epoch": 0.582084349947717,
+      "grad_norm": 0.1987033486366272,
+      "learning_rate": 1.7671662600209134e-05,
+      "loss": 0.165,
+      "step": 1670
+    },
+    {
+      "epoch": 0.585569884977344,
+      "grad_norm": 29.639982223510742,
+      "learning_rate": 1.7657720460090625e-05,
+      "loss": 0.1937,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5890554200069711,
+      "grad_norm": 3.289419174194336,
+      "learning_rate": 1.764377831997212e-05,
+      "loss": 0.0802,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5925409550365981,
+      "grad_norm": 0.2210395336151123,
+      "learning_rate": 1.762983617985361e-05,
+      "loss": 0.2471,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5960264900662252,
+      "grad_norm": 23.282852172851562,
+      "learning_rate": 1.76158940397351e-05,
+      "loss": 0.2112,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5995120250958522,
+      "grad_norm": 0.22058001160621643,
+      "learning_rate": 1.760195189961659e-05,
+      "loss": 0.0758,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6029975601254792,
+      "grad_norm": 7.556725025177002,
+      "learning_rate": 1.7588009759498082e-05,
+      "loss": 0.1733,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6064830951551063,
+      "grad_norm": 13.092087745666504,
+      "learning_rate": 1.7574067619379576e-05,
+      "loss": 0.118,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6099686301847334,
+      "grad_norm": 2.1389198303222656,
+      "learning_rate": 1.7560125479261067e-05,
+      "loss": 0.1971,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6134541652143604,
+      "grad_norm": 2.126339912414551,
+      "learning_rate": 1.754618333914256e-05,
+      "loss": 0.3391,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6169397002439875,
+      "grad_norm": 0.7699007987976074,
+      "learning_rate": 1.7532241199024052e-05,
+      "loss": 0.299,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6204252352736145,
+      "grad_norm": 0.1900128424167633,
+      "learning_rate": 1.7518299058905543e-05,
+      "loss": 0.1021,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6239107703032416,
+      "grad_norm": 20.255779266357422,
+      "learning_rate": 1.7504356918787037e-05,
+      "loss": 0.158,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6273963053328686,
+      "grad_norm": 0.18125827610492706,
+      "learning_rate": 1.7490414778668528e-05,
+      "loss": 0.2431,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6308818403624956,
+      "grad_norm": 0.5560604929924011,
+      "learning_rate": 1.747647263855002e-05,
+      "loss": 0.281,
+      "step": 1810
+    },
+    {
+      "epoch": 0.6343673753921227,
+      "grad_norm": 0.6502192616462708,
+      "learning_rate": 1.7462530498431513e-05,
+      "loss": 0.1709,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6378529104217497,
+      "grad_norm": 3.676496982574463,
+      "learning_rate": 1.7448588358313004e-05,
+      "loss": 0.3918,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6413384454513767,
+      "grad_norm": 8.176740646362305,
+      "learning_rate": 1.7434646218194494e-05,
+      "loss": 0.1222,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6448239804810039,
+      "grad_norm": 1.6892282962799072,
+      "learning_rate": 1.7420704078075985e-05,
+      "loss": 0.3701,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6483095155106309,
+      "grad_norm": 3.25469708442688,
+      "learning_rate": 1.7406761937957476e-05,
+      "loss": 0.2383,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6517950505402579,
+      "grad_norm": 1.433481216430664,
+      "learning_rate": 1.739281979783897e-05,
+      "loss": 0.1936,
+      "step": 1870
+    },
+    {
+      "epoch": 0.655280585569885,
+      "grad_norm": 0.20236587524414062,
+      "learning_rate": 1.737887765772046e-05,
+      "loss": 0.4097,
+      "step": 1880
+    },
+    {
+      "epoch": 0.658766120599512,
+      "grad_norm": 3.3700122833251953,
+      "learning_rate": 1.7364935517601952e-05,
+      "loss": 0.2351,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6622516556291391,
+      "grad_norm": 10.565374374389648,
+      "learning_rate": 1.7350993377483446e-05,
+      "loss": 0.3335,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6657371906587661,
+      "grad_norm": 0.19991923868656158,
+      "learning_rate": 1.7337051237364937e-05,
+      "loss": 0.4573,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6692227256883931,
+      "grad_norm": 2.1597394943237305,
+      "learning_rate": 1.7323109097246428e-05,
+      "loss": 0.1396,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6727082607180203,
+      "grad_norm": 23.80716896057129,
+      "learning_rate": 1.7309166957127922e-05,
+      "loss": 0.2434,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6761937957476473,
+      "grad_norm": 0.18005388975143433,
+      "learning_rate": 1.7295224817009413e-05,
+      "loss": 0.2398,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6796793307772743,
+      "grad_norm": 0.24591070413589478,
+      "learning_rate": 1.7281282676890904e-05,
+      "loss": 0.1673,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6831648658069014,
+      "grad_norm": 0.35577160120010376,
+      "learning_rate": 1.7267340536772398e-05,
+      "loss": 0.1134,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6866504008365284,
+      "grad_norm": 0.14578354358673096,
+      "learning_rate": 1.725339839665389e-05,
+      "loss": 0.1394,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6901359358661554,
+      "grad_norm": 0.15032856166362762,
+      "learning_rate": 1.723945625653538e-05,
+      "loss": 0.2795,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6936214708957825,
+      "grad_norm": 0.15948134660720825,
+      "learning_rate": 1.722551411641687e-05,
+      "loss": 0.1617,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6971070059254095,
+      "grad_norm": 1.1098570823669434,
+      "learning_rate": 1.721157197629836e-05,
+      "loss": 0.1358,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7005925409550366,
+      "grad_norm": 14.691607475280762,
+      "learning_rate": 1.7197629836179855e-05,
+      "loss": 0.2749,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7040780759846637,
+      "grad_norm": 9.642597198486328,
+      "learning_rate": 1.7183687696061346e-05,
+      "loss": 0.2413,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7075636110142907,
+      "grad_norm": 0.25982531905174255,
+      "learning_rate": 1.7169745555942837e-05,
+      "loss": 0.2381,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7110491460439178,
+      "grad_norm": 11.684551239013672,
+      "learning_rate": 1.715580341582433e-05,
+      "loss": 0.2735,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7145346810735448,
+      "grad_norm": 13.072251319885254,
+      "learning_rate": 1.7141861275705822e-05,
+      "loss": 0.1853,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7180202161031718,
+      "grad_norm": 4.4599504470825195,
+      "learning_rate": 1.7127919135587313e-05,
+      "loss": 0.0791,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7215057511327989,
+      "grad_norm": 11.390928268432617,
+      "learning_rate": 1.7113976995468807e-05,
+      "loss": 0.363,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7249912861624259,
+      "grad_norm": 24.80426788330078,
+      "learning_rate": 1.7100034855350298e-05,
+      "loss": 0.1271,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7284768211920529,
+      "grad_norm": 4.6373724937438965,
+      "learning_rate": 1.708609271523179e-05,
+      "loss": 0.2046,
+      "step": 2090
+    },
+    {
+      "epoch": 0.73196235622168,
+      "grad_norm": 18.688011169433594,
+      "learning_rate": 1.7072150575113283e-05,
+      "loss": 0.3764,
+      "step": 2100
+    },
+    {
+      "epoch": 0.735447891251307,
+      "grad_norm": 0.1868332177400589,
+      "learning_rate": 1.7058208434994774e-05,
+      "loss": 0.2338,
+      "step": 2110
+    },
+    {
+      "epoch": 0.7389334262809342,
+      "grad_norm": 0.9303812980651855,
+      "learning_rate": 1.7044266294876264e-05,
+      "loss": 0.1955,
+      "step": 2120
+    },
+    {
+      "epoch": 0.7424189613105612,
+      "grad_norm": 20.024219512939453,
+      "learning_rate": 1.7030324154757755e-05,
+      "loss": 0.2066,
+      "step": 2130
+    },
+    {
+      "epoch": 0.7459044963401882,
+      "grad_norm": 0.41837865114212036,
+      "learning_rate": 1.701638201463925e-05,
+      "loss": 0.43,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7493900313698153,
+      "grad_norm": 0.2002245932817459,
+      "learning_rate": 1.700243987452074e-05,
+      "loss": 0.2395,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7528755663994423,
+      "grad_norm": 16.88410186767578,
+      "learning_rate": 1.698849773440223e-05,
+      "loss": 0.3061,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7563611014290693,
+      "grad_norm": 13.094812393188477,
+      "learning_rate": 1.6974555594283725e-05,
+      "loss": 0.2037,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7598466364586964,
+      "grad_norm": 0.489964097738266,
+      "learning_rate": 1.6960613454165216e-05,
+      "loss": 0.1176,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7633321714883234,
+      "grad_norm": 0.36262962222099304,
+      "learning_rate": 1.6946671314046707e-05,
+      "loss": 0.1996,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7668177065179506,
+      "grad_norm": 13.340808868408203,
+      "learning_rate": 1.69327291739282e-05,
+      "loss": 0.3046,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7703032415475776,
+      "grad_norm": 4.607888221740723,
+      "learning_rate": 1.6918787033809692e-05,
+      "loss": 0.2871,
+      "step": 2210
+    },
+    {
+      "epoch": 0.7737887765772046,
+      "grad_norm": 0.15974846482276917,
+      "learning_rate": 1.6904844893691183e-05,
+      "loss": 0.2577,
+      "step": 2220
+    },
+    {
+      "epoch": 0.7772743116068317,
+      "grad_norm": 0.5607108473777771,
+      "learning_rate": 1.6890902753572677e-05,
+      "loss": 0.1506,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7807598466364587,
+      "grad_norm": 0.22983232140541077,
+      "learning_rate": 1.6876960613454168e-05,
+      "loss": 0.0976,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7842453816660857,
+      "grad_norm": 0.15347087383270264,
+      "learning_rate": 1.686301847333566e-05,
+      "loss": 0.1101,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7877309166957128,
+      "grad_norm": 0.15394413471221924,
+      "learning_rate": 1.684907633321715e-05,
+      "loss": 0.2706,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7912164517253398,
+      "grad_norm": 16.60938262939453,
+      "learning_rate": 1.683513419309864e-05,
+      "loss": 0.3779,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7947019867549668,
+      "grad_norm": 19.347156524658203,
+      "learning_rate": 1.6821192052980134e-05,
+      "loss": 0.2697,
+      "step": 2280
+    },
+    {
+      "epoch": 0.798187521784594,
+      "grad_norm": 18.826566696166992,
+      "learning_rate": 1.6807249912861625e-05,
+      "loss": 0.1772,
+      "step": 2290
+    },
+    {
+      "epoch": 0.801673056814221,
+      "grad_norm": 6.404747009277344,
+      "learning_rate": 1.6793307772743116e-05,
+      "loss": 0.1967,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8051585918438481,
+      "grad_norm": 2.8832345008850098,
+      "learning_rate": 1.677936563262461e-05,
+      "loss": 0.2426,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8086441268734751,
+      "grad_norm": 8.400115966796875,
+      "learning_rate": 1.67654234925061e-05,
+      "loss": 0.1401,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8121296619031021,
+      "grad_norm": 3.5098154544830322,
+      "learning_rate": 1.6751481352387592e-05,
+      "loss": 0.1521,
+      "step": 2330
+    },
+    {
+      "epoch": 0.8156151969327292,
+      "grad_norm": 7.560535907745361,
+      "learning_rate": 1.6737539212269086e-05,
+      "loss": 0.2052,
+      "step": 2340
+    },
+    {
+      "epoch": 0.8191007319623562,
+      "grad_norm": 20.31093978881836,
+      "learning_rate": 1.6723597072150577e-05,
+      "loss": 0.2422,
+      "step": 2350
+    },
+    {
+      "epoch": 0.8225862669919832,
+      "grad_norm": 0.14886406064033508,
+      "learning_rate": 1.6709654932032068e-05,
+      "loss": 0.3423,
+      "step": 2360
+    },
+    {
+      "epoch": 0.8260718020216103,
+      "grad_norm": 0.16942627727985382,
+      "learning_rate": 1.6695712791913562e-05,
+      "loss": 0.1422,
+      "step": 2370
+    },
+    {
+      "epoch": 0.8295573370512374,
+      "grad_norm": 2.075434446334839,
+      "learning_rate": 1.6681770651795053e-05,
+      "loss": 0.2055,
+      "step": 2380
+    },
+    {
+      "epoch": 0.8330428720808644,
+      "grad_norm": 12.085251808166504,
+      "learning_rate": 1.6667828511676544e-05,
+      "loss": 0.1778,
+      "step": 2390
+    },
+    {
+      "epoch": 0.8365284071104915,
+      "grad_norm": 9.108433723449707,
+      "learning_rate": 1.6653886371558034e-05,
+      "loss": 0.2636,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8400139421401185,
+      "grad_norm": 0.22255125641822815,
+      "learning_rate": 1.6639944231439525e-05,
+      "loss": 0.1163,
+      "step": 2410
+    },
+    {
+      "epoch": 0.8434994771697456,
+      "grad_norm": 0.34382200241088867,
+      "learning_rate": 1.662600209132102e-05,
+      "loss": 0.3501,
+      "step": 2420
+    },
+    {
+      "epoch": 0.8469850121993726,
+      "grad_norm": 3.8618369102478027,
+      "learning_rate": 1.661205995120251e-05,
+      "loss": 0.3013,
+      "step": 2430
+    },
+    {
+      "epoch": 0.8504705472289996,
+      "grad_norm": 3.9153378009796143,
+      "learning_rate": 1.6598117811084e-05,
+      "loss": 0.2558,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8539560822586267,
+      "grad_norm": 6.622893810272217,
+      "learning_rate": 1.6584175670965495e-05,
+      "loss": 0.1465,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8574416172882537,
+      "grad_norm": 0.25614631175994873,
+      "learning_rate": 1.6570233530846986e-05,
+      "loss": 0.3136,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8609271523178808,
+      "grad_norm": 1.1627339124679565,
+      "learning_rate": 1.6556291390728477e-05,
+      "loss": 0.1084,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8644126873475079,
+      "grad_norm": 3.3835268020629883,
+      "learning_rate": 1.654234925060997e-05,
+      "loss": 0.4099,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8678982223771349,
+      "grad_norm": 15.521218299865723,
+      "learning_rate": 1.6528407110491462e-05,
+      "loss": 0.1028,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8713837574067619,
+      "grad_norm": 0.2716628313064575,
+      "learning_rate": 1.6514464970372953e-05,
+      "loss": 0.1371,
+      "step": 2500
+    },
+    {
+      "epoch": 0.874869292436389,
+      "grad_norm": 0.11692991107702255,
+      "learning_rate": 1.6500522830254447e-05,
+      "loss": 0.2051,
+      "step": 2510
+    },
+    {
+      "epoch": 0.878354827466016,
+      "grad_norm": 5.96846866607666,
+      "learning_rate": 1.6486580690135938e-05,
+      "loss": 0.1758,
+      "step": 2520
+    },
+    {
+      "epoch": 0.8818403624956431,
+      "grad_norm": 3.1128947734832764,
+      "learning_rate": 1.647263855001743e-05,
+      "loss": 0.3391,
+      "step": 2530
+    },
+    {
+      "epoch": 0.8853258975252701,
+      "grad_norm": 0.7964233756065369,
+      "learning_rate": 1.645869640989892e-05,
+      "loss": 0.3088,
+      "step": 2540
+    },
+    {
+      "epoch": 0.8888114325548971,
+      "grad_norm": 0.2802188992500305,
+      "learning_rate": 1.6444754269780413e-05,
+      "loss": 0.2082,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8922969675845243,
+      "grad_norm": 0.3661327362060547,
+      "learning_rate": 1.6430812129661904e-05,
+      "loss": 0.231,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8957825026141513,
+      "grad_norm": 6.788111686706543,
+      "learning_rate": 1.6416869989543395e-05,
+      "loss": 0.183,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8992680376437783,
+      "grad_norm": 17.159854888916016,
+      "learning_rate": 1.640292784942489e-05,
+      "loss": 0.2526,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9027535726734054,
+      "grad_norm": 23.243045806884766,
+      "learning_rate": 1.638898570930638e-05,
+      "loss": 0.2203,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9062391077030324,
+      "grad_norm": 0.10591837018728256,
+      "learning_rate": 1.637504356918787e-05,
+      "loss": 0.2705,
+      "step": 2600
+    },
+    {
+      "epoch": 0.9097246427326595,
+      "grad_norm": 8.592724800109863,
+      "learning_rate": 1.6361101429069365e-05,
+      "loss": 0.1686,
+      "step": 2610
+    },
+    {
+      "epoch": 0.9132101777622865,
+      "grad_norm": 0.5851829051971436,
+      "learning_rate": 1.6347159288950856e-05,
+      "loss": 0.1452,
+      "step": 2620
+    },
+    {
+      "epoch": 0.9166957127919135,
+      "grad_norm": 0.5782294869422913,
+      "learning_rate": 1.6333217148832347e-05,
+      "loss": 0.1843,
+      "step": 2630
+    },
+    {
+      "epoch": 0.9201812478215406,
+      "grad_norm": 27.61756134033203,
+      "learning_rate": 1.631927500871384e-05,
+      "loss": 0.1058,
+      "step": 2640
+    },
+    {
+      "epoch": 0.9236667828511677,
+      "grad_norm": 0.1436556726694107,
+      "learning_rate": 1.6305332868595332e-05,
+      "loss": 0.1947,
+      "step": 2650
+    },
+    {
+      "epoch": 0.9271523178807947,
+      "grad_norm": 0.11741270869970322,
+      "learning_rate": 1.6291390728476823e-05,
+      "loss": 0.2313,
+      "step": 2660
+    },
+    {
+      "epoch": 0.9306378529104218,
+      "grad_norm": 6.36224889755249,
+      "learning_rate": 1.6277448588358313e-05,
+      "loss": 0.291,
+      "step": 2670
+    },
+    {
+      "epoch": 0.9341233879400488,
+      "grad_norm": 0.2954885959625244,
+      "learning_rate": 1.6263506448239804e-05,
+      "loss": 0.3318,
+      "step": 2680
+    },
+    {
+      "epoch": 0.9376089229696758,
+      "grad_norm": 13.240195274353027,
+      "learning_rate": 1.62495643081213e-05,
+      "loss": 0.3679,
+      "step": 2690
+    },
+    {
+      "epoch": 0.9410944579993029,
+      "grad_norm": 24.815946578979492,
+      "learning_rate": 1.623562216800279e-05,
+      "loss": 0.3279,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9445799930289299,
+      "grad_norm": 0.11210108548402786,
+      "learning_rate": 1.622168002788428e-05,
+      "loss": 0.2753,
+      "step": 2710
+    },
+    {
+      "epoch": 0.948065528058557,
+      "grad_norm": 3.065356969833374,
+      "learning_rate": 1.6207737887765774e-05,
+      "loss": 0.1844,
+      "step": 2720
+    },
+    {
+      "epoch": 0.951551063088184,
+      "grad_norm": 1.7035739421844482,
+      "learning_rate": 1.6193795747647265e-05,
+      "loss": 0.2426,
+      "step": 2730
+    },
+    {
+      "epoch": 0.955036598117811,
+      "grad_norm": 23.689104080200195,
+      "learning_rate": 1.6179853607528756e-05,
+      "loss": 0.3324,
+      "step": 2740
+    },
+    {
+      "epoch": 0.9585221331474382,
+      "grad_norm": 8.64907169342041,
+      "learning_rate": 1.616591146741025e-05,
+      "loss": 0.1349,
+      "step": 2750
+    },
+    {
+      "epoch": 0.9620076681770652,
+      "grad_norm": 0.21376356482505798,
+      "learning_rate": 1.615196932729174e-05,
+      "loss": 0.1536,
+      "step": 2760
+    },
+    {
+      "epoch": 0.9654932032066922,
+      "grad_norm": 21.229475021362305,
+      "learning_rate": 1.6138027187173232e-05,
+      "loss": 0.3432,
+      "step": 2770
+    },
+    {
+      "epoch": 0.9689787382363193,
+      "grad_norm": 0.1763678640127182,
+      "learning_rate": 1.6124085047054726e-05,
+      "loss": 0.121,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9724642732659463,
+      "grad_norm": 0.35627982020378113,
+      "learning_rate": 1.6110142906936217e-05,
+      "loss": 0.2114,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9759498082955733,
+      "grad_norm": 0.675436794757843,
+      "learning_rate": 1.6096200766817708e-05,
+      "loss": 0.2136,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9794353433252004,
+      "grad_norm": 0.7453556656837463,
+      "learning_rate": 1.60822586266992e-05,
+      "loss": 0.1687,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9829208783548274,
+      "grad_norm": 1.4391916990280151,
+      "learning_rate": 1.606831648658069e-05,
+      "loss": 0.1633,
+      "step": 2820
+    },
+    {
+      "epoch": 0.9864064133844546,
+      "grad_norm": 0.10471702367067337,
+      "learning_rate": 1.6054374346462183e-05,
+      "loss": 0.1795,
+      "step": 2830
+    },
+    {
+      "epoch": 0.9898919484140816,
+      "grad_norm": 2.418611764907837,
+      "learning_rate": 1.6040432206343674e-05,
+      "loss": 0.1899,
+      "step": 2840
+    },
+    {
+      "epoch": 0.9933774834437086,
+      "grad_norm": 6.218674659729004,
+      "learning_rate": 1.6026490066225165e-05,
+      "loss": 0.1613,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9968630184733357,
+      "grad_norm": 23.257627487182617,
+      "learning_rate": 1.601254792610666e-05,
+      "loss": 0.1829,
+      "step": 2860
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9686419753086419,
+      "eval_loss": 0.131916344165802,
+      "eval_runtime": 20.8607,
+      "eval_samples_per_second": 194.145,
+      "eval_steps_per_second": 24.304,
+      "step": 2869
+    },
+    {
+      "epoch": 1.0003485535029628,
+      "grad_norm": 13.868711471557617,
+      "learning_rate": 1.599860578598815e-05,
+      "loss": 0.3279,
+      "step": 2870
+    },
+    {
+      "epoch": 1.0038340885325898,
+      "grad_norm": 0.0889001339673996,
+      "learning_rate": 1.598466364586964e-05,
+      "loss": 0.1179,
+      "step": 2880
+    },
+    {
+      "epoch": 1.0073196235622168,
+      "grad_norm": 7.238311767578125,
+      "learning_rate": 1.5970721505751135e-05,
+      "loss": 0.1339,
+      "step": 2890
+    },
+    {
+      "epoch": 1.0108051585918438,
+      "grad_norm": 9.563148498535156,
+      "learning_rate": 1.5956779365632626e-05,
+      "loss": 0.2212,
+      "step": 2900
+    },
+    {
+      "epoch": 1.0142906936214708,
+      "grad_norm": 10.648134231567383,
+      "learning_rate": 1.5942837225514117e-05,
+      "loss": 0.232,
+      "step": 2910
+    },
+    {
+      "epoch": 1.0177762286510978,
+      "grad_norm": 0.2760317325592041,
+      "learning_rate": 1.592889508539561e-05,
+      "loss": 0.1669,
+      "step": 2920
+    },
+    {
+      "epoch": 1.021261763680725,
+      "grad_norm": 0.15449553728103638,
+      "learning_rate": 1.5914952945277102e-05,
+      "loss": 0.2226,
+      "step": 2930
+    },
+    {
+      "epoch": 1.024747298710352,
+      "grad_norm": 0.09482862800359726,
+      "learning_rate": 1.5901010805158593e-05,
+      "loss": 0.1987,
+      "step": 2940
+    },
+    {
+      "epoch": 1.028232833739979,
+      "grad_norm": 1.185239553451538,
+      "learning_rate": 1.5887068665040083e-05,
+      "loss": 0.2861,
+      "step": 2950
+    },
+    {
+      "epoch": 1.031718368769606,
+      "grad_norm": 0.9025907516479492,
+      "learning_rate": 1.5873126524921578e-05,
+      "loss": 0.1047,
+      "step": 2960
+    },
+    {
+      "epoch": 1.035203903799233,
+      "grad_norm": 21.41613006591797,
+      "learning_rate": 1.585918438480307e-05,
+      "loss": 0.2394,
+      "step": 2970
+    },
+    {
+      "epoch": 1.0386894388288603,
+      "grad_norm": 25.966737747192383,
+      "learning_rate": 1.584524224468456e-05,
+      "loss": 0.1353,
+      "step": 2980
+    },
+    {
+      "epoch": 1.0421749738584873,
+      "grad_norm": 0.21152754127979279,
+      "learning_rate": 1.5831300104566053e-05,
+      "loss": 0.3361,
+      "step": 2990
+    },
+    {
+      "epoch": 1.0456605088881143,
+      "grad_norm": 0.37001797556877136,
+      "learning_rate": 1.5817357964447544e-05,
+      "loss": 0.2875,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0491460439177414,
+      "grad_norm": 0.16642731428146362,
+      "learning_rate": 1.5803415824329035e-05,
+      "loss": 0.2697,
+      "step": 3010
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.22467225790023804,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.1819,
+      "step": 3020
+    },
+    {
+      "epoch": 1.0561171139769954,
+      "grad_norm": 0.6170486807823181,
+      "learning_rate": 1.577553154409202e-05,
+      "loss": 0.0742,
+      "step": 3030
+    },
+    {
+      "epoch": 1.0596026490066226,
+      "grad_norm": 7.1253862380981445,
+      "learning_rate": 1.576158940397351e-05,
+      "loss": 0.2141,
+      "step": 3040
+    },
+    {
+      "epoch": 1.0630881840362496,
+      "grad_norm": 0.6570123434066772,
+      "learning_rate": 1.5747647263855005e-05,
+      "loss": 0.1353,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0665737190658766,
+      "grad_norm": 1.323370337486267,
+      "learning_rate": 1.5733705123736496e-05,
+      "loss": 0.2516,
+      "step": 3060
+    },
+    {
+      "epoch": 1.0700592540955036,
+      "grad_norm": 0.3584739863872528,
+      "learning_rate": 1.5719762983617987e-05,
+      "loss": 0.1877,
+      "step": 3070
+    },
+    {
+      "epoch": 1.0735447891251306,
+      "grad_norm": 26.268049240112305,
+      "learning_rate": 1.5705820843499477e-05,
+      "loss": 0.1618,
+      "step": 3080
+    },
+    {
+      "epoch": 1.0770303241547579,
+      "grad_norm": 1.0968873500823975,
+      "learning_rate": 1.5691878703380968e-05,
+      "loss": 0.2546,
+      "step": 3090
+    },
+    {
+      "epoch": 1.0805158591843849,
+      "grad_norm": 0.08166905492544174,
+      "learning_rate": 1.5677936563262463e-05,
+      "loss": 0.1949,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0840013942140119,
+      "grad_norm": 0.12049412727355957,
+      "learning_rate": 1.5663994423143953e-05,
+      "loss": 0.2507,
+      "step": 3110
+    },
+    {
+      "epoch": 1.0874869292436389,
+      "grad_norm": 14.46548080444336,
+      "learning_rate": 1.5650052283025444e-05,
+      "loss": 0.1774,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0909724642732659,
+      "grad_norm": 3.6209402084350586,
+      "learning_rate": 1.563611014290694e-05,
+      "loss": 0.162,
+      "step": 3130
+    },
+    {
+      "epoch": 1.094457999302893,
+      "grad_norm": 0.8061731457710266,
+      "learning_rate": 1.562216800278843e-05,
+      "loss": 0.117,
+      "step": 3140
+    },
+    {
+      "epoch": 1.0979435343325201,
+      "grad_norm": 0.17544050514698029,
+      "learning_rate": 1.560822586266992e-05,
+      "loss": 0.1057,
+      "step": 3150
+    },
+    {
+      "epoch": 1.1014290693621471,
+      "grad_norm": 0.06735656410455704,
+      "learning_rate": 1.5594283722551414e-05,
+      "loss": 0.2237,
+      "step": 3160
+    },
+    {
+      "epoch": 1.1049146043917741,
+      "grad_norm": 12.56591510772705,
+      "learning_rate": 1.5580341582432905e-05,
+      "loss": 0.1499,
+      "step": 3170
+    },
+    {
+      "epoch": 1.1084001394214011,
+      "grad_norm": 1.4435107707977295,
+      "learning_rate": 1.5566399442314396e-05,
+      "loss": 0.3691,
+      "step": 3180
+    },
+    {
+      "epoch": 1.1118856744510281,
+      "grad_norm": 0.12776677310466766,
+      "learning_rate": 1.555245730219589e-05,
+      "loss": 0.1918,
+      "step": 3190
+    },
+    {
+      "epoch": 1.1153712094806554,
+      "grad_norm": 0.09562991559505463,
+      "learning_rate": 1.553851516207738e-05,
+      "loss": 0.4167,
+      "step": 3200
+    },
+    {
+      "epoch": 1.1188567445102824,
+      "grad_norm": 1.6694416999816895,
+      "learning_rate": 1.552457302195887e-05,
+      "loss": 0.1662,
+      "step": 3210
+    },
+    {
+      "epoch": 1.1223422795399094,
+      "grad_norm": 16.271982192993164,
+      "learning_rate": 1.5510630881840362e-05,
+      "loss": 0.238,
+      "step": 3220
+    },
+    {
+      "epoch": 1.1258278145695364,
+      "grad_norm": 10.23219108581543,
+      "learning_rate": 1.5496688741721853e-05,
+      "loss": 0.2752,
+      "step": 3230
+    },
+    {
+      "epoch": 1.1293133495991634,
+      "grad_norm": 2.7908616065979004,
+      "learning_rate": 1.5482746601603347e-05,
+      "loss": 0.058,
+      "step": 3240
+    },
+    {
+      "epoch": 1.1327988846287904,
+      "grad_norm": 0.17575271427631378,
+      "learning_rate": 1.5468804461484838e-05,
+      "loss": 0.2244,
+      "step": 3250
+    },
+    {
+      "epoch": 1.1362844196584176,
+      "grad_norm": 1.6930599212646484,
+      "learning_rate": 1.545486232136633e-05,
+      "loss": 0.2232,
+      "step": 3260
+    },
+    {
+      "epoch": 1.1397699546880447,
+      "grad_norm": 1.9705103635787964,
+      "learning_rate": 1.5440920181247823e-05,
+      "loss": 0.1543,
+      "step": 3270
+    },
+    {
+      "epoch": 1.1432554897176717,
+      "grad_norm": 11.21917724609375,
+      "learning_rate": 1.5426978041129314e-05,
+      "loss": 0.3028,
+      "step": 3280
+    },
+    {
+      "epoch": 1.1467410247472987,
+      "grad_norm": 4.643821716308594,
+      "learning_rate": 1.5413035901010805e-05,
+      "loss": 0.303,
+      "step": 3290
+    },
+    {
+      "epoch": 1.1502265597769257,
+      "grad_norm": 19.168472290039062,
+      "learning_rate": 1.53990937608923e-05,
+      "loss": 0.2017,
+      "step": 3300
+    },
+    {
+      "epoch": 1.153712094806553,
+      "grad_norm": 0.08442903310060501,
+      "learning_rate": 1.538515162077379e-05,
+      "loss": 0.0612,
+      "step": 3310
+    },
+    {
+      "epoch": 1.15719762983618,
+      "grad_norm": 0.07782625406980515,
+      "learning_rate": 1.5371209480655284e-05,
+      "loss": 0.1009,
+      "step": 3320
+    },
+    {
+      "epoch": 1.160683164865807,
+      "grad_norm": 17.379108428955078,
+      "learning_rate": 1.5357267340536775e-05,
+      "loss": 0.205,
+      "step": 3330
+    },
+    {
+      "epoch": 1.164168699895434,
+      "grad_norm": 0.07793194055557251,
+      "learning_rate": 1.5343325200418266e-05,
+      "loss": 0.261,
+      "step": 3340
+    },
+    {
+      "epoch": 1.167654234925061,
+      "grad_norm": 0.10120818763971329,
+      "learning_rate": 1.5329383060299757e-05,
+      "loss": 0.1319,
+      "step": 3350
+    },
+    {
+      "epoch": 1.1711397699546882,
+      "grad_norm": 13.039447784423828,
+      "learning_rate": 1.5315440920181247e-05,
+      "loss": 0.2302,
+      "step": 3360
+    },
+    {
+      "epoch": 1.1746253049843152,
+      "grad_norm": 0.0781761035323143,
+      "learning_rate": 1.530149878006274e-05,
+      "loss": 0.1687,
+      "step": 3370
+    },
+    {
+      "epoch": 1.1781108400139422,
+      "grad_norm": 5.389220237731934,
+      "learning_rate": 1.5287556639944232e-05,
+      "loss": 0.1653,
+      "step": 3380
+    },
+    {
+      "epoch": 1.1815963750435692,
+      "grad_norm": 16.386268615722656,
+      "learning_rate": 1.5273614499825723e-05,
+      "loss": 0.2369,
+      "step": 3390
+    },
+    {
+      "epoch": 1.1850819100731962,
+      "grad_norm": 0.0885128378868103,
+      "learning_rate": 1.5259672359707217e-05,
+      "loss": 0.1952,
+      "step": 3400
+    },
+    {
+      "epoch": 1.1885674451028232,
+      "grad_norm": 0.17366395890712738,
+      "learning_rate": 1.5245730219588708e-05,
+      "loss": 0.2546,
+      "step": 3410
+    },
+    {
+      "epoch": 1.1920529801324504,
+      "grad_norm": 0.10797934234142303,
+      "learning_rate": 1.52317880794702e-05,
+      "loss": 0.1879,
+      "step": 3420
+    },
+    {
+      "epoch": 1.1955385151620774,
+      "grad_norm": 5.789648532867432,
+      "learning_rate": 1.5217845939351692e-05,
+      "loss": 0.2316,
+      "step": 3430
+    },
+    {
+      "epoch": 1.1990240501917044,
+      "grad_norm": 0.19041916728019714,
+      "learning_rate": 1.5203903799233184e-05,
+      "loss": 0.3066,
+      "step": 3440
+    },
+    {
+      "epoch": 1.2025095852213314,
+      "grad_norm": 0.12138816714286804,
+      "learning_rate": 1.5189961659114677e-05,
+      "loss": 0.0508,
+      "step": 3450
+    },
+    {
+      "epoch": 1.2059951202509585,
+      "grad_norm": 0.9047268629074097,
+      "learning_rate": 1.5176019518996167e-05,
+      "loss": 0.2256,
+      "step": 3460
+    },
+    {
+      "epoch": 1.2094806552805855,
+      "grad_norm": 15.816200256347656,
+      "learning_rate": 1.516207737887766e-05,
+      "loss": 0.2817,
+      "step": 3470
+    },
+    {
+      "epoch": 1.2129661903102127,
+      "grad_norm": 0.109500952064991,
+      "learning_rate": 1.5148135238759152e-05,
+      "loss": 0.0286,
+      "step": 3480
+    },
+    {
+      "epoch": 1.2164517253398397,
+      "grad_norm": 1.6199902296066284,
+      "learning_rate": 1.5134193098640642e-05,
+      "loss": 0.124,
+      "step": 3490
+    },
+    {
+      "epoch": 1.2199372603694667,
+      "grad_norm": 0.5839070677757263,
+      "learning_rate": 1.5120250958522134e-05,
+      "loss": 0.1435,
+      "step": 3500
+    },
+    {
+      "epoch": 1.2234227953990937,
+      "grad_norm": 0.088262178003788,
+      "learning_rate": 1.5106308818403625e-05,
+      "loss": 0.1329,
+      "step": 3510
+    },
+    {
+      "epoch": 1.2269083304287207,
+      "grad_norm": 19.781274795532227,
+      "learning_rate": 1.5092366678285117e-05,
+      "loss": 0.105,
+      "step": 3520
+    },
+    {
+      "epoch": 1.230393865458348,
+      "grad_norm": 0.21191132068634033,
+      "learning_rate": 1.507842453816661e-05,
+      "loss": 0.2493,
+      "step": 3530
+    },
+    {
+      "epoch": 1.233879400487975,
+      "grad_norm": 12.934003829956055,
+      "learning_rate": 1.50644823980481e-05,
+      "loss": 0.185,
+      "step": 3540
+    },
+    {
+      "epoch": 1.237364935517602,
+      "grad_norm": 8.322224617004395,
+      "learning_rate": 1.5050540257929593e-05,
+      "loss": 0.2953,
+      "step": 3550
+    },
+    {
+      "epoch": 1.240850470547229,
+      "grad_norm": 0.5193243622779846,
+      "learning_rate": 1.5036598117811086e-05,
+      "loss": 0.2215,
+      "step": 3560
+    },
+    {
+      "epoch": 1.244336005576856,
+      "grad_norm": 12.667407989501953,
+      "learning_rate": 1.5022655977692577e-05,
+      "loss": 0.1893,
+      "step": 3570
+    },
+    {
+      "epoch": 1.2478215406064832,
+      "grad_norm": 0.07419061660766602,
+      "learning_rate": 1.5008713837574069e-05,
+      "loss": 0.1897,
+      "step": 3580
+    },
+    {
+      "epoch": 1.2513070756361102,
+      "grad_norm": 14.40754222869873,
+      "learning_rate": 1.4994771697455562e-05,
+      "loss": 0.4052,
+      "step": 3590
+    },
+    {
+      "epoch": 1.2547926106657372,
+      "grad_norm": 3.1872828006744385,
+      "learning_rate": 1.4980829557337054e-05,
+      "loss": 0.1727,
+      "step": 3600
+    },
+    {
+      "epoch": 1.2582781456953642,
+      "grad_norm": 28.886524200439453,
+      "learning_rate": 1.4966887417218545e-05,
+      "loss": 0.5003,
+      "step": 3610
+    },
+    {
+      "epoch": 1.2617636807249912,
+      "grad_norm": 0.11840394884347916,
+      "learning_rate": 1.4952945277100037e-05,
+      "loss": 0.1676,
+      "step": 3620
+    },
+    {
+      "epoch": 1.2652492157546185,
+      "grad_norm": 13.540458679199219,
+      "learning_rate": 1.4939003136981527e-05,
+      "loss": 0.2395,
+      "step": 3630
+    },
+    {
+      "epoch": 1.2687347507842452,
+      "grad_norm": 0.23518075048923492,
+      "learning_rate": 1.4925060996863019e-05,
+      "loss": 0.3769,
+      "step": 3640
+    },
+    {
+      "epoch": 1.2722202858138725,
+      "grad_norm": 0.5583080053329468,
+      "learning_rate": 1.4911118856744512e-05,
+      "loss": 0.2068,
+      "step": 3650
+    },
+    {
+      "epoch": 1.2757058208434995,
+      "grad_norm": 3.620421886444092,
+      "learning_rate": 1.4897176716626002e-05,
+      "loss": 0.2443,
+      "step": 3660
+    },
+    {
+      "epoch": 1.2791913558731265,
+      "grad_norm": 14.638289451599121,
+      "learning_rate": 1.4883234576507495e-05,
+      "loss": 0.0804,
+      "step": 3670
+    },
+    {
+      "epoch": 1.2826768909027535,
+      "grad_norm": 0.17217248678207397,
+      "learning_rate": 1.4869292436388987e-05,
+      "loss": 0.1282,
+      "step": 3680
+    },
+    {
+      "epoch": 1.2861624259323805,
+      "grad_norm": 0.0552227608859539,
+      "learning_rate": 1.4855350296270478e-05,
+      "loss": 0.0728,
+      "step": 3690
+    },
+    {
+      "epoch": 1.2896479609620077,
+      "grad_norm": 0.18412978947162628,
+      "learning_rate": 1.484140815615197e-05,
+      "loss": 0.3107,
+      "step": 3700
+    },
+    {
+      "epoch": 1.2931334959916347,
+      "grad_norm": 1.4496484994888306,
+      "learning_rate": 1.4827466016033463e-05,
+      "loss": 0.2406,
+      "step": 3710
+    },
+    {
+      "epoch": 1.2966190310212617,
+      "grad_norm": 16.918373107910156,
+      "learning_rate": 1.4813523875914954e-05,
+      "loss": 0.3134,
+      "step": 3720
+    },
+    {
+      "epoch": 1.3001045660508888,
+      "grad_norm": 5.249573230743408,
+      "learning_rate": 1.4799581735796447e-05,
+      "loss": 0.1284,
+      "step": 3730
+    },
+    {
+      "epoch": 1.3035901010805158,
+      "grad_norm": 19.409761428833008,
+      "learning_rate": 1.4785639595677939e-05,
+      "loss": 0.0873,
+      "step": 3740
+    },
+    {
+      "epoch": 1.307075636110143,
+      "grad_norm": 0.08998361974954605,
+      "learning_rate": 1.477169745555943e-05,
+      "loss": 0.2249,
+      "step": 3750
+    },
+    {
+      "epoch": 1.31056117113977,
+      "grad_norm": 4.676770210266113,
+      "learning_rate": 1.475775531544092e-05,
+      "loss": 0.2761,
+      "step": 3760
+    },
+    {
+      "epoch": 1.314046706169397,
+      "grad_norm": 4.773134708404541,
+      "learning_rate": 1.4743813175322411e-05,
+      "loss": 0.1475,
+      "step": 3770
+    },
+    {
+      "epoch": 1.317532241199024,
+      "grad_norm": 23.77858543395996,
+      "learning_rate": 1.4729871035203904e-05,
+      "loss": 0.1701,
+      "step": 3780
+    },
+    {
+      "epoch": 1.321017776228651,
+      "grad_norm": 6.024458408355713,
+      "learning_rate": 1.4715928895085396e-05,
+      "loss": 0.245,
+      "step": 3790
+    },
+    {
+      "epoch": 1.3245033112582782,
+      "grad_norm": 26.570066452026367,
+      "learning_rate": 1.4701986754966889e-05,
+      "loss": 0.2245,
+      "step": 3800
+    },
+    {
+      "epoch": 1.3279888462879053,
+      "grad_norm": 0.06447375565767288,
+      "learning_rate": 1.468804461484838e-05,
+      "loss": 0.101,
+      "step": 3810
+    },
+    {
+      "epoch": 1.3314743813175323,
+      "grad_norm": 11.744895935058594,
+      "learning_rate": 1.4674102474729872e-05,
+      "loss": 0.2009,
+      "step": 3820
+    },
+    {
+      "epoch": 1.3349599163471593,
+      "grad_norm": 0.07097235321998596,
+      "learning_rate": 1.4660160334611365e-05,
+      "loss": 0.0224,
+      "step": 3830
+    },
+    {
+      "epoch": 1.3384454513767863,
+      "grad_norm": 0.11342521011829376,
+      "learning_rate": 1.4646218194492856e-05,
+      "loss": 0.0885,
+      "step": 3840
+    },
+    {
+      "epoch": 1.3419309864064135,
+      "grad_norm": 0.06430939584970474,
+      "learning_rate": 1.4632276054374348e-05,
+      "loss": 0.2388,
+      "step": 3850
+    },
+    {
+      "epoch": 1.3454165214360405,
+      "grad_norm": 3.096365451812744,
+      "learning_rate": 1.461833391425584e-05,
+      "loss": 0.1814,
+      "step": 3860
+    },
+    {
+      "epoch": 1.3489020564656675,
+      "grad_norm": 0.2073626071214676,
+      "learning_rate": 1.4604391774137331e-05,
+      "loss": 0.1304,
+      "step": 3870
+    },
+    {
+      "epoch": 1.3523875914952945,
+      "grad_norm": 0.8284794688224792,
+      "learning_rate": 1.4590449634018824e-05,
+      "loss": 0.2057,
+      "step": 3880
+    },
+    {
+      "epoch": 1.3558731265249215,
+      "grad_norm": 23.129026412963867,
+      "learning_rate": 1.4576507493900316e-05,
+      "loss": 0.175,
+      "step": 3890
+    },
+    {
+      "epoch": 1.3593586615545485,
+      "grad_norm": 0.11513429880142212,
+      "learning_rate": 1.4562565353781806e-05,
+      "loss": 0.2399,
+      "step": 3900
+    },
+    {
+      "epoch": 1.3628441965841755,
+      "grad_norm": 5.663066864013672,
+      "learning_rate": 1.4548623213663298e-05,
+      "loss": 0.1074,
+      "step": 3910
+    },
+    {
+      "epoch": 1.3663297316138028,
+      "grad_norm": 21.3768310546875,
+      "learning_rate": 1.4534681073544789e-05,
+      "loss": 0.1827,
+      "step": 3920
+    },
+    {
+      "epoch": 1.3698152666434298,
+      "grad_norm": 9.777505874633789,
+      "learning_rate": 1.4520738933426281e-05,
+      "loss": 0.2197,
+      "step": 3930
+    },
+    {
+      "epoch": 1.3733008016730568,
+      "grad_norm": 0.8618378043174744,
+      "learning_rate": 1.4506796793307774e-05,
+      "loss": 0.2736,
+      "step": 3940
+    },
+    {
+      "epoch": 1.3767863367026838,
+      "grad_norm": 23.52062225341797,
+      "learning_rate": 1.4492854653189265e-05,
+      "loss": 0.2393,
+      "step": 3950
+    },
+    {
+      "epoch": 1.3802718717323108,
+      "grad_norm": 7.636297225952148,
+      "learning_rate": 1.4478912513070757e-05,
+      "loss": 0.16,
+      "step": 3960
+    },
+    {
+      "epoch": 1.383757406761938,
+      "grad_norm": 3.086662530899048,
+      "learning_rate": 1.446497037295225e-05,
+      "loss": 0.2637,
+      "step": 3970
+    },
+    {
+      "epoch": 1.387242941791565,
+      "grad_norm": 6.053905487060547,
+      "learning_rate": 1.4451028232833742e-05,
+      "loss": 0.1277,
+      "step": 3980
+    },
+    {
+      "epoch": 1.390728476821192,
+      "grad_norm": 0.09486166387796402,
+      "learning_rate": 1.4437086092715233e-05,
+      "loss": 0.0652,
+      "step": 3990
+    },
+    {
+      "epoch": 1.394214011850819,
+      "grad_norm": 9.314697265625,
+      "learning_rate": 1.4423143952596726e-05,
+      "loss": 0.194,
+      "step": 4000
+    },
+    {
+      "epoch": 1.397699546880446,
+      "grad_norm": 5.727607727050781,
+      "learning_rate": 1.4409201812478218e-05,
+      "loss": 0.1755,
+      "step": 4010
+    },
+    {
+      "epoch": 1.4011850819100733,
+      "grad_norm": 10.165273666381836,
+      "learning_rate": 1.4395259672359709e-05,
+      "loss": 0.3493,
+      "step": 4020
+    },
+    {
+      "epoch": 1.4046706169397003,
+      "grad_norm": 0.11753875017166138,
+      "learning_rate": 1.4381317532241201e-05,
+      "loss": 0.1553,
+      "step": 4030
+    },
+    {
+      "epoch": 1.4081561519693273,
+      "grad_norm": 0.0666135773062706,
+      "learning_rate": 1.436737539212269e-05,
+      "loss": 0.3826,
+      "step": 4040
+    },
+    {
+      "epoch": 1.4116416869989543,
+      "grad_norm": 0.06710106134414673,
+      "learning_rate": 1.4353433252004183e-05,
+      "loss": 0.1867,
+      "step": 4050
+    },
+    {
+      "epoch": 1.4151272220285813,
+      "grad_norm": 18.661643981933594,
+      "learning_rate": 1.4339491111885676e-05,
+      "loss": 0.2184,
+      "step": 4060
+    },
+    {
+      "epoch": 1.4186127570582086,
+      "grad_norm": 0.0907972976565361,
+      "learning_rate": 1.4325548971767166e-05,
+      "loss": 0.2187,
+      "step": 4070
+    },
+    {
+      "epoch": 1.4220982920878356,
+      "grad_norm": 1.6616415977478027,
+      "learning_rate": 1.4311606831648659e-05,
+      "loss": 0.1063,
+      "step": 4080
+    },
+    {
+      "epoch": 1.4255838271174626,
+      "grad_norm": 0.11328552663326263,
+      "learning_rate": 1.4297664691530151e-05,
+      "loss": 0.1416,
+      "step": 4090
+    },
+    {
+      "epoch": 1.4290693621470896,
+      "grad_norm": 0.36197254061698914,
+      "learning_rate": 1.4283722551411642e-05,
+      "loss": 0.3247,
+      "step": 4100
+    },
+    {
+      "epoch": 1.4325548971767166,
+      "grad_norm": 2.550544500350952,
+      "learning_rate": 1.4269780411293135e-05,
+      "loss": 0.2398,
+      "step": 4110
+    },
+    {
+      "epoch": 1.4360404322063438,
+      "grad_norm": 0.25212812423706055,
+      "learning_rate": 1.4255838271174627e-05,
+      "loss": 0.0808,
+      "step": 4120
+    },
+    {
+      "epoch": 1.4395259672359706,
+      "grad_norm": 0.5252062678337097,
+      "learning_rate": 1.4241896131056118e-05,
+      "loss": 0.0474,
+      "step": 4130
+    },
+    {
+      "epoch": 1.4430115022655978,
+      "grad_norm": 1.4656167030334473,
+      "learning_rate": 1.422795399093761e-05,
+      "loss": 0.2964,
+      "step": 4140
+    },
+    {
+      "epoch": 1.4464970372952248,
+      "grad_norm": 0.06279879063367844,
+      "learning_rate": 1.4214011850819103e-05,
+      "loss": 0.2259,
+      "step": 4150
+    },
+    {
+      "epoch": 1.4499825723248518,
+      "grad_norm": 0.2818046510219574,
+      "learning_rate": 1.4200069710700594e-05,
+      "loss": 0.2609,
+      "step": 4160
+    },
+    {
+      "epoch": 1.4534681073544788,
+      "grad_norm": 3.4003350734710693,
+      "learning_rate": 1.4186127570582085e-05,
+      "loss": 0.1227,
+      "step": 4170
+    },
+    {
+      "epoch": 1.4569536423841059,
+      "grad_norm": 0.0695640966296196,
+      "learning_rate": 1.4172185430463577e-05,
+      "loss": 0.0424,
+      "step": 4180
+    },
+    {
+      "epoch": 1.460439177413733,
+      "grad_norm": 0.06560744345188141,
+      "learning_rate": 1.4158243290345068e-05,
+      "loss": 0.2265,
+      "step": 4190
+    },
+    {
+      "epoch": 1.46392471244336,
+      "grad_norm": 0.46215468645095825,
+      "learning_rate": 1.414430115022656e-05,
+      "loss": 0.1508,
+      "step": 4200
+    },
+    {
+      "epoch": 1.467410247472987,
+      "grad_norm": 0.21034209430217743,
+      "learning_rate": 1.4130359010108053e-05,
+      "loss": 0.1693,
+      "step": 4210
+    },
+    {
+      "epoch": 1.470895782502614,
+      "grad_norm": 24.144786834716797,
+      "learning_rate": 1.4116416869989544e-05,
+      "loss": 0.1907,
+      "step": 4220
+    },
+    {
+      "epoch": 1.474381317532241,
+      "grad_norm": 0.0716620460152626,
+      "learning_rate": 1.4102474729871036e-05,
+      "loss": 0.1234,
+      "step": 4230
+    },
+    {
+      "epoch": 1.4778668525618683,
+      "grad_norm": 5.373108863830566,
+      "learning_rate": 1.4088532589752529e-05,
+      "loss": 0.0568,
+      "step": 4240
+    },
+    {
+      "epoch": 1.4813523875914953,
+      "grad_norm": 0.06202000752091408,
+      "learning_rate": 1.407459044963402e-05,
+      "loss": 0.2224,
+      "step": 4250
+    },
+    {
+      "epoch": 1.4848379226211224,
+      "grad_norm": 0.06813399493694305,
+      "learning_rate": 1.4060648309515512e-05,
+      "loss": 0.2171,
+      "step": 4260
+    },
+    {
+      "epoch": 1.4883234576507494,
+      "grad_norm": 0.05181474983692169,
+      "learning_rate": 1.4046706169397005e-05,
+      "loss": 0.1021,
+      "step": 4270
+    },
+    {
+      "epoch": 1.4918089926803764,
+      "grad_norm": 1.736260175704956,
+      "learning_rate": 1.4032764029278496e-05,
+      "loss": 0.0671,
+      "step": 4280
+    },
+    {
+      "epoch": 1.4952945277100036,
+      "grad_norm": 0.14703615009784698,
+      "learning_rate": 1.4018821889159988e-05,
+      "loss": 0.073,
+      "step": 4290
+    },
+    {
+      "epoch": 1.4987800627396306,
+      "grad_norm": 0.08682423084974289,
+      "learning_rate": 1.400487974904148e-05,
+      "loss": 0.1046,
+      "step": 4300
+    },
+    {
+      "epoch": 1.5022655977692576,
+      "grad_norm": 0.08216214179992676,
+      "learning_rate": 1.399093760892297e-05,
+      "loss": 0.1728,
+      "step": 4310
+    },
+    {
+      "epoch": 1.5057511327988846,
+      "grad_norm": 4.310842037200928,
+      "learning_rate": 1.3976995468804462e-05,
+      "loss": 0.0201,
+      "step": 4320
+    },
+    {
+      "epoch": 1.5092366678285116,
+      "grad_norm": 0.6964895129203796,
+      "learning_rate": 1.3963053328685953e-05,
+      "loss": 0.2253,
+      "step": 4330
+    },
+    {
+      "epoch": 1.5127222028581389,
+      "grad_norm": 0.06760009378194809,
+      "learning_rate": 1.3949111188567445e-05,
+      "loss": 0.0558,
+      "step": 4340
+    },
+    {
+      "epoch": 1.5162077378877656,
+      "grad_norm": 1.6589277982711792,
+      "learning_rate": 1.3935169048448938e-05,
+      "loss": 0.286,
+      "step": 4350
+    },
+    {
+      "epoch": 1.5196932729173929,
+      "grad_norm": 2.4069876670837402,
+      "learning_rate": 1.3921226908330429e-05,
+      "loss": 0.2241,
+      "step": 4360
+    },
+    {
+      "epoch": 1.5231788079470199,
+      "grad_norm": 0.12570062279701233,
+      "learning_rate": 1.3907284768211921e-05,
+      "loss": 0.1356,
+      "step": 4370
+    },
+    {
+      "epoch": 1.5266643429766469,
+      "grad_norm": 0.06179194152355194,
+      "learning_rate": 1.3893342628093414e-05,
+      "loss": 0.1527,
+      "step": 4380
+    },
+    {
+      "epoch": 1.5301498780062741,
+      "grad_norm": 0.05693870410323143,
+      "learning_rate": 1.3879400487974906e-05,
+      "loss": 0.1209,
+      "step": 4390
+    },
+    {
+      "epoch": 1.533635413035901,
+      "grad_norm": 9.671772003173828,
+      "learning_rate": 1.3865458347856397e-05,
+      "loss": 0.1119,
+      "step": 4400
+    },
+    {
+      "epoch": 1.5371209480655281,
+      "grad_norm": 0.3705396354198456,
+      "learning_rate": 1.385151620773789e-05,
+      "loss": 0.1369,
+      "step": 4410
+    },
+    {
+      "epoch": 1.5406064830951551,
+      "grad_norm": 0.1401708871126175,
+      "learning_rate": 1.3837574067619382e-05,
+      "loss": 0.1132,
+      "step": 4420
+    },
+    {
+      "epoch": 1.5440920181247821,
+      "grad_norm": 2.265974521636963,
+      "learning_rate": 1.3823631927500873e-05,
+      "loss": 0.3808,
+      "step": 4430
+    },
+    {
+      "epoch": 1.5475775531544091,
+      "grad_norm": 3.8402411937713623,
+      "learning_rate": 1.3809689787382366e-05,
+      "loss": 0.2986,
+      "step": 4440
+    },
+    {
+      "epoch": 1.5510630881840362,
+      "grad_norm": 1.6255053281784058,
+      "learning_rate": 1.3795747647263855e-05,
+      "loss": 0.042,
+      "step": 4450
+    },
+    {
+      "epoch": 1.5545486232136634,
+      "grad_norm": 0.05299604684114456,
+      "learning_rate": 1.3781805507145347e-05,
+      "loss": 0.0497,
+      "step": 4460
+    },
+    {
+      "epoch": 1.5580341582432904,
+      "grad_norm": 0.05339618772268295,
+      "learning_rate": 1.376786336702684e-05,
+      "loss": 0.1915,
+      "step": 4470
+    },
+    {
+      "epoch": 1.5615196932729174,
+      "grad_norm": 30.932308197021484,
+      "learning_rate": 1.375392122690833e-05,
+      "loss": 0.1334,
+      "step": 4480
+    },
+    {
+      "epoch": 1.5650052283025444,
+      "grad_norm": 18.293027877807617,
+      "learning_rate": 1.3739979086789823e-05,
+      "loss": 0.2085,
+      "step": 4490
+    },
+    {
+      "epoch": 1.5684907633321714,
+      "grad_norm": 5.334358215332031,
+      "learning_rate": 1.3726036946671315e-05,
+      "loss": 0.1781,
+      "step": 4500
+    },
+    {
+      "epoch": 1.5719762983617986,
+      "grad_norm": 0.09247801452875137,
+      "learning_rate": 1.3712094806552806e-05,
+      "loss": 0.1584,
+      "step": 4510
+    },
+    {
+      "epoch": 1.5754618333914254,
+      "grad_norm": 0.10990609228610992,
+      "learning_rate": 1.3698152666434299e-05,
+      "loss": 0.0539,
+      "step": 4520
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.12785302102565765,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.2398,
+      "step": 4530
+    },
+    {
+      "epoch": 1.5824329034506797,
+      "grad_norm": 0.24163594841957092,
+      "learning_rate": 1.3670268386197282e-05,
+      "loss": 0.227,
+      "step": 4540
+    },
+    {
+      "epoch": 1.5859184384803067,
+      "grad_norm": 0.06425183266401291,
+      "learning_rate": 1.3656326246078775e-05,
+      "loss": 0.0495,
+      "step": 4550
+    },
+    {
+      "epoch": 1.589403973509934,
+      "grad_norm": 15.79089641571045,
+      "learning_rate": 1.3642384105960267e-05,
+      "loss": 0.1335,
+      "step": 4560
+    },
+    {
+      "epoch": 1.5928895085395607,
+      "grad_norm": 8.039728164672852,
+      "learning_rate": 1.362844196584176e-05,
+      "loss": 0.1999,
+      "step": 4570
+    },
+    {
+      "epoch": 1.596375043569188,
+      "grad_norm": 0.10453764349222183,
+      "learning_rate": 1.3614499825723249e-05,
+      "loss": 0.0428,
+      "step": 4580
+    },
+    {
+      "epoch": 1.599860578598815,
+      "grad_norm": 0.09684242308139801,
+      "learning_rate": 1.3600557685604741e-05,
+      "loss": 0.0348,
+      "step": 4590
+    },
+    {
+      "epoch": 1.603346113628442,
+      "grad_norm": 2.0049784183502197,
+      "learning_rate": 1.3586615545486232e-05,
+      "loss": 0.0227,
+      "step": 4600
+    },
+    {
+      "epoch": 1.6068316486580692,
+      "grad_norm": 0.05621238425374031,
+      "learning_rate": 1.3572673405367725e-05,
+      "loss": 0.2134,
+      "step": 4610
+    },
+    {
+      "epoch": 1.610317183687696,
+      "grad_norm": 24.956762313842773,
+      "learning_rate": 1.3558731265249217e-05,
+      "loss": 0.0569,
+      "step": 4620
+    },
+    {
+      "epoch": 1.6138027187173232,
+      "grad_norm": 0.04639355093240738,
+      "learning_rate": 1.3544789125130708e-05,
+      "loss": 0.1955,
+      "step": 4630
+    },
+    {
+      "epoch": 1.6172882537469502,
+      "grad_norm": 21.884952545166016,
+      "learning_rate": 1.35308469850122e-05,
+      "loss": 0.2995,
+      "step": 4640
+    },
+    {
+      "epoch": 1.6207737887765772,
+      "grad_norm": 0.07225365191698074,
+      "learning_rate": 1.3516904844893693e-05,
+      "loss": 0.1419,
+      "step": 4650
+    },
+    {
+      "epoch": 1.6242593238062044,
+      "grad_norm": 0.06399868428707123,
+      "learning_rate": 1.3502962704775184e-05,
+      "loss": 0.0805,
+      "step": 4660
+    },
+    {
+      "epoch": 1.6277448588358312,
+      "grad_norm": 0.06368258595466614,
+      "learning_rate": 1.3489020564656676e-05,
+      "loss": 0.3106,
+      "step": 4670
+    },
+    {
+      "epoch": 1.6312303938654584,
+      "grad_norm": 0.08223626762628555,
+      "learning_rate": 1.3475078424538169e-05,
+      "loss": 0.1631,
+      "step": 4680
+    },
+    {
+      "epoch": 1.6347159288950854,
+      "grad_norm": 6.698259353637695,
+      "learning_rate": 1.346113628441966e-05,
+      "loss": 0.2371,
+      "step": 4690
+    },
+    {
+      "epoch": 1.6382014639247124,
+      "grad_norm": 0.10331364721059799,
+      "learning_rate": 1.3447194144301152e-05,
+      "loss": 0.1428,
+      "step": 4700
+    },
+    {
+      "epoch": 1.6416869989543394,
+      "grad_norm": 0.18476131558418274,
+      "learning_rate": 1.3433252004182645e-05,
+      "loss": 0.1106,
+      "step": 4710
+    },
+    {
+      "epoch": 1.6451725339839665,
+      "grad_norm": 5.083899021148682,
+      "learning_rate": 1.3419309864064134e-05,
+      "loss": 0.1615,
+      "step": 4720
+    },
+    {
+      "epoch": 1.6486580690135937,
+      "grad_norm": 0.05285824462771416,
+      "learning_rate": 1.3405367723945626e-05,
+      "loss": 0.368,
+      "step": 4730
+    },
+    {
+      "epoch": 1.6521436040432205,
+      "grad_norm": 4.834890365600586,
+      "learning_rate": 1.3391425583827117e-05,
+      "loss": 0.2775,
+      "step": 4740
+    },
+    {
+      "epoch": 1.6556291390728477,
+      "grad_norm": 0.3817296624183655,
+      "learning_rate": 1.337748344370861e-05,
+      "loss": 0.0175,
+      "step": 4750
+    },
+    {
+      "epoch": 1.6591146741024747,
+      "grad_norm": 21.72384262084961,
+      "learning_rate": 1.3363541303590102e-05,
+      "loss": 0.181,
+      "step": 4760
+    },
+    {
+      "epoch": 1.6626002091321017,
+      "grad_norm": 6.993847846984863,
+      "learning_rate": 1.3349599163471595e-05,
+      "loss": 0.1962,
+      "step": 4770
+    },
+    {
+      "epoch": 1.666085744161729,
+      "grad_norm": 0.06561506539583206,
+      "learning_rate": 1.3335657023353085e-05,
+      "loss": 0.1334,
+      "step": 4780
+    },
+    {
+      "epoch": 1.6695712791913557,
+      "grad_norm": 10.643206596374512,
+      "learning_rate": 1.3321714883234578e-05,
+      "loss": 0.183,
+      "step": 4790
+    },
+    {
+      "epoch": 1.673056814220983,
+      "grad_norm": 10.353646278381348,
+      "learning_rate": 1.330777274311607e-05,
+      "loss": 0.2462,
+      "step": 4800
+    },
+    {
+      "epoch": 1.67654234925061,
+      "grad_norm": 14.645241737365723,
+      "learning_rate": 1.3293830602997561e-05,
+      "loss": 0.1927,
+      "step": 4810
+    },
+    {
+      "epoch": 1.680027884280237,
+      "grad_norm": 0.07404288649559021,
+      "learning_rate": 1.3279888462879054e-05,
+      "loss": 0.1398,
+      "step": 4820
+    },
+    {
+      "epoch": 1.6835134193098642,
+      "grad_norm": 11.214003562927246,
+      "learning_rate": 1.3265946322760546e-05,
+      "loss": 0.1837,
+      "step": 4830
+    },
+    {
+      "epoch": 1.686998954339491,
+      "grad_norm": 0.11196375638246536,
+      "learning_rate": 1.3252004182642037e-05,
+      "loss": 0.1121,
+      "step": 4840
+    },
+    {
+      "epoch": 1.6904844893691182,
+      "grad_norm": 9.329215049743652,
+      "learning_rate": 1.3238062042523528e-05,
+      "loss": 0.1331,
+      "step": 4850
+    },
+    {
+      "epoch": 1.6939700243987452,
+      "grad_norm": 0.13003626465797424,
+      "learning_rate": 1.3224119902405019e-05,
+      "loss": 0.1499,
+      "step": 4860
+    },
+    {
+      "epoch": 1.6974555594283722,
+      "grad_norm": 0.055021703243255615,
+      "learning_rate": 1.3210177762286511e-05,
+      "loss": 0.1088,
+      "step": 4870
+    },
+    {
+      "epoch": 1.7009410944579995,
+      "grad_norm": 0.05173968896269798,
+      "learning_rate": 1.3196235622168004e-05,
+      "loss": 0.0519,
+      "step": 4880
+    },
+    {
+      "epoch": 1.7044266294876262,
+      "grad_norm": 1.1130493879318237,
+      "learning_rate": 1.3182293482049495e-05,
+      "loss": 0.215,
+      "step": 4890
+    },
+    {
+      "epoch": 1.7079121645172535,
+      "grad_norm": 36.07246017456055,
+      "learning_rate": 1.3168351341930987e-05,
+      "loss": 0.26,
+      "step": 4900
+    },
+    {
+      "epoch": 1.7113976995468805,
+      "grad_norm": 0.09109174460172653,
+      "learning_rate": 1.315440920181248e-05,
+      "loss": 0.018,
+      "step": 4910
+    },
+    {
+      "epoch": 1.7148832345765075,
+      "grad_norm": 2.289961814880371,
+      "learning_rate": 1.314046706169397e-05,
+      "loss": 0.1378,
+      "step": 4920
+    },
+    {
+      "epoch": 1.7183687696061345,
+      "grad_norm": 0.04753530025482178,
+      "learning_rate": 1.3126524921575463e-05,
+      "loss": 0.1949,
+      "step": 4930
+    },
+    {
+      "epoch": 1.7218543046357615,
+      "grad_norm": 6.459181785583496,
+      "learning_rate": 1.3112582781456955e-05,
+      "loss": 0.288,
+      "step": 4940
+    },
+    {
+      "epoch": 1.7253398396653887,
+      "grad_norm": 11.19622802734375,
+      "learning_rate": 1.3098640641338448e-05,
+      "loss": 0.1554,
+      "step": 4950
+    },
+    {
+      "epoch": 1.7288253746950157,
+      "grad_norm": 15.069147109985352,
+      "learning_rate": 1.3084698501219939e-05,
+      "loss": 0.2112,
+      "step": 4960
+    },
+    {
+      "epoch": 1.7323109097246427,
+      "grad_norm": 17.245620727539062,
+      "learning_rate": 1.3070756361101431e-05,
+      "loss": 0.2637,
+      "step": 4970
+    },
+    {
+      "epoch": 1.7357964447542698,
+      "grad_norm": 25.261966705322266,
+      "learning_rate": 1.3056814220982924e-05,
+      "loss": 0.1142,
+      "step": 4980
+    },
+    {
+      "epoch": 1.7392819797838968,
+      "grad_norm": 0.08086353540420532,
+      "learning_rate": 1.3042872080864413e-05,
+      "loss": 0.2082,
+      "step": 4990
+    },
+    {
+      "epoch": 1.742767514813524,
+      "grad_norm": 33.71073532104492,
+      "learning_rate": 1.3028929940745905e-05,
+      "loss": 0.0617,
+      "step": 5000
+    },
+    {
+      "epoch": 1.7462530498431508,
+      "grad_norm": 8.686712265014648,
+      "learning_rate": 1.3014987800627396e-05,
+      "loss": 0.1367,
+      "step": 5010
+    },
+    {
+      "epoch": 1.749738584872778,
+      "grad_norm": 2.0827035903930664,
+      "learning_rate": 1.3001045660508889e-05,
+      "loss": 0.1099,
+      "step": 5020
+    },
+    {
+      "epoch": 1.753224119902405,
+      "grad_norm": 7.724724769592285,
+      "learning_rate": 1.2987103520390381e-05,
+      "loss": 0.132,
+      "step": 5030
+    },
+    {
+      "epoch": 1.756709654932032,
+      "grad_norm": 7.940341949462891,
+      "learning_rate": 1.2973161380271872e-05,
+      "loss": 0.3053,
+      "step": 5040
+    },
+    {
+      "epoch": 1.7601951899616592,
+      "grad_norm": 0.10044675320386887,
+      "learning_rate": 1.2959219240153364e-05,
+      "loss": 0.1138,
+      "step": 5050
+    },
+    {
+      "epoch": 1.763680724991286,
+      "grad_norm": 0.045176051557064056,
+      "learning_rate": 1.2945277100034857e-05,
+      "loss": 0.212,
+      "step": 5060
+    },
+    {
+      "epoch": 1.7671662600209133,
+      "grad_norm": 3.353107452392578,
+      "learning_rate": 1.2931334959916348e-05,
+      "loss": 0.2066,
+      "step": 5070
+    },
+    {
+      "epoch": 1.7706517950505403,
+      "grad_norm": 0.062912218272686,
+      "learning_rate": 1.291739281979784e-05,
+      "loss": 0.2011,
+      "step": 5080
+    },
+    {
+      "epoch": 1.7741373300801673,
+      "grad_norm": 7.203383922576904,
+      "learning_rate": 1.2903450679679333e-05,
+      "loss": 0.4203,
+      "step": 5090
+    },
+    {
+      "epoch": 1.7776228651097945,
+      "grad_norm": 0.4804973602294922,
+      "learning_rate": 1.2889508539560824e-05,
+      "loss": 0.2008,
+      "step": 5100
+    },
+    {
+      "epoch": 1.7811084001394213,
+      "grad_norm": 0.10035104304552078,
+      "learning_rate": 1.2875566399442316e-05,
+      "loss": 0.1394,
+      "step": 5110
+    },
+    {
+      "epoch": 1.7845939351690485,
+      "grad_norm": 18.202369689941406,
+      "learning_rate": 1.2861624259323809e-05,
+      "loss": 0.0973,
+      "step": 5120
+    },
+    {
+      "epoch": 1.7880794701986755,
+      "grad_norm": 27.99166488647461,
+      "learning_rate": 1.2847682119205298e-05,
+      "loss": 0.3659,
+      "step": 5130
+    },
+    {
+      "epoch": 1.7915650052283025,
+      "grad_norm": 5.776055812835693,
+      "learning_rate": 1.283373997908679e-05,
+      "loss": 0.0682,
+      "step": 5140
+    },
+    {
+      "epoch": 1.7950505402579295,
+      "grad_norm": 6.58804178237915,
+      "learning_rate": 1.2819797838968283e-05,
+      "loss": 0.0869,
+      "step": 5150
+    },
+    {
+      "epoch": 1.7985360752875565,
+      "grad_norm": 2.2123751640319824,
+      "learning_rate": 1.2805855698849774e-05,
+      "loss": 0.1033,
+      "step": 5160
+    },
+    {
+      "epoch": 1.8020216103171838,
+      "grad_norm": 5.483887195587158,
+      "learning_rate": 1.2791913558731266e-05,
+      "loss": 0.21,
+      "step": 5170
+    },
+    {
+      "epoch": 1.8055071453468108,
+      "grad_norm": 35.22694778442383,
+      "learning_rate": 1.2777971418612759e-05,
+      "loss": 0.0469,
+      "step": 5180
+    },
+    {
+      "epoch": 1.8089926803764378,
+      "grad_norm": 16.363428115844727,
+      "learning_rate": 1.276402927849425e-05,
+      "loss": 0.2203,
+      "step": 5190
+    },
+    {
+      "epoch": 1.8124782154060648,
+      "grad_norm": 3.4047915935516357,
+      "learning_rate": 1.2750087138375742e-05,
+      "loss": 0.0254,
+      "step": 5200
+    },
+    {
+      "epoch": 1.8159637504356918,
+      "grad_norm": 0.04784770309925079,
+      "learning_rate": 1.2736144998257234e-05,
+      "loss": 0.1671,
+      "step": 5210
+    },
+    {
+      "epoch": 1.819449285465319,
+      "grad_norm": 0.18466421961784363,
+      "learning_rate": 1.2722202858138725e-05,
+      "loss": 0.2333,
+      "step": 5220
+    },
+    {
+      "epoch": 1.8229348204949458,
+      "grad_norm": 0.04754915460944176,
+      "learning_rate": 1.2708260718020218e-05,
+      "loss": 0.109,
+      "step": 5230
+    },
+    {
+      "epoch": 1.826420355524573,
+      "grad_norm": 12.06218147277832,
+      "learning_rate": 1.269431857790171e-05,
+      "loss": 0.1115,
+      "step": 5240
+    },
+    {
+      "epoch": 1.8299058905542,
+      "grad_norm": 12.35152816772461,
+      "learning_rate": 1.2680376437783201e-05,
+      "loss": 0.1294,
+      "step": 5250
+    },
+    {
+      "epoch": 1.833391425583827,
+      "grad_norm": 0.5232785940170288,
+      "learning_rate": 1.2666434297664692e-05,
+      "loss": 0.1716,
+      "step": 5260
+    },
+    {
+      "epoch": 1.8368769606134543,
+      "grad_norm": 0.065973199903965,
+      "learning_rate": 1.2652492157546183e-05,
+      "loss": 0.2806,
+      "step": 5270
+    },
+    {
+      "epoch": 1.840362495643081,
+      "grad_norm": 0.05121095851063728,
+      "learning_rate": 1.2638550017427675e-05,
+      "loss": 0.3057,
+      "step": 5280
+    },
+    {
+      "epoch": 1.8438480306727083,
+      "grad_norm": 0.04605868458747864,
+      "learning_rate": 1.2624607877309168e-05,
+      "loss": 0.182,
+      "step": 5290
+    },
+    {
+      "epoch": 1.8473335657023353,
+      "grad_norm": 3.7501471042633057,
+      "learning_rate": 1.2610665737190659e-05,
+      "loss": 0.0176,
+      "step": 5300
+    },
+    {
+      "epoch": 1.8508191007319623,
+      "grad_norm": 21.904052734375,
+      "learning_rate": 1.2596723597072151e-05,
+      "loss": 0.2681,
+      "step": 5310
+    },
+    {
+      "epoch": 1.8543046357615895,
+      "grad_norm": 4.048603534698486,
+      "learning_rate": 1.2582781456953644e-05,
+      "loss": 0.1016,
+      "step": 5320
+    },
+    {
+      "epoch": 1.8577901707912163,
+      "grad_norm": 17.176145553588867,
+      "learning_rate": 1.2568839316835134e-05,
+      "loss": 0.344,
+      "step": 5330
+    },
+    {
+      "epoch": 1.8612757058208436,
+      "grad_norm": 0.05373512953519821,
+      "learning_rate": 1.2554897176716627e-05,
+      "loss": 0.0364,
+      "step": 5340
+    },
+    {
+      "epoch": 1.8647612408504706,
+      "grad_norm": 0.4178902506828308,
+      "learning_rate": 1.254095503659812e-05,
+      "loss": 0.2752,
+      "step": 5350
+    },
+    {
+      "epoch": 1.8682467758800976,
+      "grad_norm": 0.11804527789354324,
+      "learning_rate": 1.2527012896479612e-05,
+      "loss": 0.2173,
+      "step": 5360
+    },
+    {
+      "epoch": 1.8717323109097248,
+      "grad_norm": 17.27764892578125,
+      "learning_rate": 1.2513070756361103e-05,
+      "loss": 0.1362,
+      "step": 5370
+    },
+    {
+      "epoch": 1.8752178459393516,
+      "grad_norm": 20.017900466918945,
+      "learning_rate": 1.2499128616242595e-05,
+      "loss": 0.1478,
+      "step": 5380
+    },
+    {
+      "epoch": 1.8787033809689788,
+      "grad_norm": 34.95002746582031,
+      "learning_rate": 1.2485186476124088e-05,
+      "loss": 0.1725,
+      "step": 5390
+    },
+    {
+      "epoch": 1.8821889159986058,
+      "grad_norm": 0.23638759553432465,
+      "learning_rate": 1.2471244336005577e-05,
+      "loss": 0.0903,
+      "step": 5400
+    },
+    {
+      "epoch": 1.8856744510282328,
+      "grad_norm": 8.639537811279297,
+      "learning_rate": 1.245730219588707e-05,
+      "loss": 0.1225,
+      "step": 5410
+    },
+    {
+      "epoch": 1.8891599860578598,
+      "grad_norm": 0.10233496874570847,
+      "learning_rate": 1.244336005576856e-05,
+      "loss": 0.2227,
+      "step": 5420
+    },
+    {
+      "epoch": 1.8926455210874868,
+      "grad_norm": 4.306946754455566,
+      "learning_rate": 1.2429417915650053e-05,
+      "loss": 0.1647,
+      "step": 5430
+    },
+    {
+      "epoch": 1.896131056117114,
+      "grad_norm": 18.103878021240234,
+      "learning_rate": 1.2415475775531545e-05,
+      "loss": 0.231,
+      "step": 5440
+    },
+    {
+      "epoch": 1.8996165911467409,
+      "grad_norm": 20.41852569580078,
+      "learning_rate": 1.2401533635413036e-05,
+      "loss": 0.2714,
+      "step": 5450
+    },
+    {
+      "epoch": 1.903102126176368,
+      "grad_norm": 2.2409658432006836,
+      "learning_rate": 1.2387591495294529e-05,
+      "loss": 0.1807,
+      "step": 5460
+    },
+    {
+      "epoch": 1.906587661205995,
+      "grad_norm": 0.10098310559988022,
+      "learning_rate": 1.2373649355176021e-05,
+      "loss": 0.2041,
+      "step": 5470
+    },
+    {
+      "epoch": 1.910073196235622,
+      "grad_norm": 2.7435004711151123,
+      "learning_rate": 1.2359707215057512e-05,
+      "loss": 0.1135,
+      "step": 5480
+    },
+    {
+      "epoch": 1.9135587312652493,
+      "grad_norm": 0.11639636754989624,
+      "learning_rate": 1.2345765074939004e-05,
+      "loss": 0.1849,
+      "step": 5490
+    },
+    {
+      "epoch": 1.9170442662948761,
+      "grad_norm": 13.145666122436523,
+      "learning_rate": 1.2331822934820497e-05,
+      "loss": 0.2934,
+      "step": 5500
+    },
+    {
+      "epoch": 1.9205298013245033,
+      "grad_norm": 8.368687629699707,
+      "learning_rate": 1.2317880794701988e-05,
+      "loss": 0.1166,
+      "step": 5510
+    },
+    {
+      "epoch": 1.9240153363541304,
+      "grad_norm": 0.16210561990737915,
+      "learning_rate": 1.230393865458348e-05,
+      "loss": 0.133,
+      "step": 5520
+    },
+    {
+      "epoch": 1.9275008713837574,
+      "grad_norm": 0.6329889893531799,
+      "learning_rate": 1.2289996514464973e-05,
+      "loss": 0.1613,
+      "step": 5530
+    },
+    {
+      "epoch": 1.9309864064133846,
+      "grad_norm": 4.6153693199157715,
+      "learning_rate": 1.2276054374346462e-05,
+      "loss": 0.09,
+      "step": 5540
+    },
+    {
+      "epoch": 1.9344719414430114,
+      "grad_norm": 0.1312878131866455,
+      "learning_rate": 1.2262112234227954e-05,
+      "loss": 0.1287,
+      "step": 5550
+    },
+    {
+      "epoch": 1.9379574764726386,
+      "grad_norm": 0.05980315059423447,
+      "learning_rate": 1.2248170094109447e-05,
+      "loss": 0.0959,
+      "step": 5560
+    },
+    {
+      "epoch": 1.9414430115022656,
+      "grad_norm": 19.431610107421875,
+      "learning_rate": 1.2234227953990938e-05,
+      "loss": 0.1181,
+      "step": 5570
+    },
+    {
+      "epoch": 1.9449285465318926,
+      "grad_norm": 0.44517797231674194,
+      "learning_rate": 1.222028581387243e-05,
+      "loss": 0.0802,
+      "step": 5580
+    },
+    {
+      "epoch": 1.9484140815615199,
+      "grad_norm": 0.04525260254740715,
+      "learning_rate": 1.2206343673753923e-05,
+      "loss": 0.2204,
+      "step": 5590
+    },
+    {
+      "epoch": 1.9518996165911466,
+      "grad_norm": 0.1254422813653946,
+      "learning_rate": 1.2192401533635414e-05,
+      "loss": 0.0652,
+      "step": 5600
+    },
+    {
+      "epoch": 1.9553851516207739,
+      "grad_norm": 1.6299126148223877,
+      "learning_rate": 1.2178459393516906e-05,
+      "loss": 0.2201,
+      "step": 5610
+    },
+    {
+      "epoch": 1.9588706866504009,
+      "grad_norm": 27.747234344482422,
+      "learning_rate": 1.2164517253398399e-05,
+      "loss": 0.0627,
+      "step": 5620
+    },
+    {
+      "epoch": 1.9623562216800279,
+      "grad_norm": 2.0174477100372314,
+      "learning_rate": 1.215057511327989e-05,
+      "loss": 0.2007,
+      "step": 5630
+    },
+    {
+      "epoch": 1.9658417567096549,
+      "grad_norm": 7.153161525726318,
+      "learning_rate": 1.2136632973161382e-05,
+      "loss": 0.2419,
+      "step": 5640
+    },
+    {
+      "epoch": 1.969327291739282,
+      "grad_norm": 0.46940529346466064,
+      "learning_rate": 1.2122690833042874e-05,
+      "loss": 0.0573,
+      "step": 5650
+    },
+    {
+      "epoch": 1.9728128267689091,
+      "grad_norm": 4.334830284118652,
+      "learning_rate": 1.2108748692924365e-05,
+      "loss": 0.2491,
+      "step": 5660
+    },
+    {
+      "epoch": 1.9762983617985361,
+      "grad_norm": 0.04316120222210884,
+      "learning_rate": 1.2094806552805856e-05,
+      "loss": 0.1859,
+      "step": 5670
+    },
+    {
+      "epoch": 1.9797838968281631,
+      "grad_norm": 0.04128291830420494,
+      "learning_rate": 1.2080864412687347e-05,
+      "loss": 0.1061,
+      "step": 5680
+    },
+    {
+      "epoch": 1.9832694318577901,
+      "grad_norm": 0.06560923904180527,
+      "learning_rate": 1.206692227256884e-05,
+      "loss": 0.054,
+      "step": 5690
+    },
+    {
+      "epoch": 1.9867549668874172,
+      "grad_norm": 0.05319322645664215,
+      "learning_rate": 1.2052980132450332e-05,
+      "loss": 0.1104,
+      "step": 5700
+    },
+    {
+      "epoch": 1.9902405019170444,
+      "grad_norm": 0.04641329124569893,
+      "learning_rate": 1.2039037992331823e-05,
+      "loss": 0.184,
+      "step": 5710
+    },
+    {
+      "epoch": 1.9937260369466712,
+      "grad_norm": 6.287529468536377,
+      "learning_rate": 1.2025095852213315e-05,
+      "loss": 0.0912,
+      "step": 5720
+    },
+    {
+      "epoch": 1.9972115719762984,
+      "grad_norm": 0.6517956852912903,
+      "learning_rate": 1.2011153712094808e-05,
+      "loss": 0.1706,
+      "step": 5730
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9795061728395061,
+      "eval_loss": 0.08464130759239197,
+      "eval_runtime": 18.9519,
+      "eval_samples_per_second": 213.699,
+      "eval_steps_per_second": 26.752,
+      "step": 5738
+    },
+    {
+      "epoch": 2.0006971070059256,
+      "grad_norm": 9.767017364501953,
+      "learning_rate": 1.19972115719763e-05,
+      "loss": 0.0665,
+      "step": 5740
+    },
+    {
+      "epoch": 2.0041826420355524,
+      "grad_norm": 9.686302185058594,
+      "learning_rate": 1.1983269431857791e-05,
+      "loss": 0.3084,
+      "step": 5750
+    },
+    {
+      "epoch": 2.0076681770651796,
+      "grad_norm": 34.309242248535156,
+      "learning_rate": 1.1969327291739283e-05,
+      "loss": 0.1616,
+      "step": 5760
+    },
+    {
+      "epoch": 2.0111537120948064,
+      "grad_norm": 0.03740281984210014,
+      "learning_rate": 1.1955385151620776e-05,
+      "loss": 0.0595,
+      "step": 5770
+    },
+    {
+      "epoch": 2.0146392471244337,
+      "grad_norm": 10.491089820861816,
+      "learning_rate": 1.1941443011502267e-05,
+      "loss": 0.2387,
+      "step": 5780
+    },
+    {
+      "epoch": 2.0181247821540604,
+      "grad_norm": 3.4705872535705566,
+      "learning_rate": 1.192750087138376e-05,
+      "loss": 0.0817,
+      "step": 5790
+    },
+    {
+      "epoch": 2.0216103171836877,
+      "grad_norm": 0.04854891821742058,
+      "learning_rate": 1.1913558731265252e-05,
+      "loss": 0.1739,
+      "step": 5800
+    },
+    {
+      "epoch": 2.025095852213315,
+      "grad_norm": 3.5387609004974365,
+      "learning_rate": 1.1899616591146741e-05,
+      "loss": 0.3629,
+      "step": 5810
+    },
+    {
+      "epoch": 2.0285813872429417,
+      "grad_norm": 0.45299866795539856,
+      "learning_rate": 1.1885674451028233e-05,
+      "loss": 0.2141,
+      "step": 5820
+    },
+    {
+      "epoch": 2.032066922272569,
+      "grad_norm": 0.03833978250622749,
+      "learning_rate": 1.1871732310909724e-05,
+      "loss": 0.0408,
+      "step": 5830
+    },
+    {
+      "epoch": 2.0355524573021957,
+      "grad_norm": 9.347251892089844,
+      "learning_rate": 1.1857790170791217e-05,
+      "loss": 0.2311,
+      "step": 5840
+    },
+    {
+      "epoch": 2.039037992331823,
+      "grad_norm": 0.11472854763269424,
+      "learning_rate": 1.184384803067271e-05,
+      "loss": 0.0753,
+      "step": 5850
+    },
+    {
+      "epoch": 2.04252352736145,
+      "grad_norm": 0.03732588514685631,
+      "learning_rate": 1.18299058905542e-05,
+      "loss": 0.0874,
+      "step": 5860
+    },
+    {
+      "epoch": 2.046009062391077,
+      "grad_norm": 9.975323677062988,
+      "learning_rate": 1.1815963750435693e-05,
+      "loss": 0.1819,
+      "step": 5870
+    },
+    {
+      "epoch": 2.049494597420704,
+      "grad_norm": 0.06995538622140884,
+      "learning_rate": 1.1802021610317185e-05,
+      "loss": 0.1342,
+      "step": 5880
+    },
+    {
+      "epoch": 2.052980132450331,
+      "grad_norm": 12.45875072479248,
+      "learning_rate": 1.1788079470198676e-05,
+      "loss": 0.0675,
+      "step": 5890
+    },
+    {
+      "epoch": 2.056465667479958,
+      "grad_norm": 15.665087699890137,
+      "learning_rate": 1.1774137330080168e-05,
+      "loss": 0.3064,
+      "step": 5900
+    },
+    {
+      "epoch": 2.0599512025095854,
+      "grad_norm": 6.563640594482422,
+      "learning_rate": 1.1760195189961661e-05,
+      "loss": 0.115,
+      "step": 5910
+    },
+    {
+      "epoch": 2.063436737539212,
+      "grad_norm": 0.057756856083869934,
+      "learning_rate": 1.1746253049843153e-05,
+      "loss": 0.1534,
+      "step": 5920
+    },
+    {
+      "epoch": 2.0669222725688394,
+      "grad_norm": 1.6861991882324219,
+      "learning_rate": 1.1732310909724644e-05,
+      "loss": 0.0366,
+      "step": 5930
+    },
+    {
+      "epoch": 2.070407807598466,
+      "grad_norm": 7.513923645019531,
+      "learning_rate": 1.1718368769606137e-05,
+      "loss": 0.3493,
+      "step": 5940
+    },
+    {
+      "epoch": 2.0738933426280934,
+      "grad_norm": 0.05242444574832916,
+      "learning_rate": 1.1704426629487626e-05,
+      "loss": 0.117,
+      "step": 5950
+    },
+    {
+      "epoch": 2.0773788776577207,
+      "grad_norm": 8.048820495605469,
+      "learning_rate": 1.1690484489369118e-05,
+      "loss": 0.1696,
+      "step": 5960
+    },
+    {
+      "epoch": 2.0808644126873475,
+      "grad_norm": 6.724079608917236,
+      "learning_rate": 1.1676542349250611e-05,
+      "loss": 0.3178,
+      "step": 5970
+    },
+    {
+      "epoch": 2.0843499477169747,
+      "grad_norm": 20.328981399536133,
+      "learning_rate": 1.1662600209132102e-05,
+      "loss": 0.0367,
+      "step": 5980
+    },
+    {
+      "epoch": 2.0878354827466015,
+      "grad_norm": 0.051835183054208755,
+      "learning_rate": 1.1648658069013594e-05,
+      "loss": 0.1535,
+      "step": 5990
+    },
+    {
+      "epoch": 2.0913210177762287,
+      "grad_norm": 1.784183382987976,
+      "learning_rate": 1.1634715928895087e-05,
+      "loss": 0.0967,
+      "step": 6000
+    },
+    {
+      "epoch": 2.0948065528058555,
+      "grad_norm": 0.08854708820581436,
+      "learning_rate": 1.1620773788776578e-05,
+      "loss": 0.2557,
+      "step": 6010
+    },
+    {
+      "epoch": 2.0982920878354827,
+      "grad_norm": 0.035098157823085785,
+      "learning_rate": 1.160683164865807e-05,
+      "loss": 0.0805,
+      "step": 6020
+    },
+    {
+      "epoch": 2.10177762286511,
+      "grad_norm": 2.4362149238586426,
+      "learning_rate": 1.1592889508539563e-05,
+      "loss": 0.1533,
+      "step": 6030
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.06450653076171875,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.0535,
+      "step": 6040
+    },
+    {
+      "epoch": 2.108748692924364,
+      "grad_norm": 0.1979568749666214,
+      "learning_rate": 1.1565005228302546e-05,
+      "loss": 0.1879,
+      "step": 6050
+    },
+    {
+      "epoch": 2.1122342279539907,
+      "grad_norm": 0.039692219346761703,
+      "learning_rate": 1.1551063088184038e-05,
+      "loss": 0.2937,
+      "step": 6060
+    },
+    {
+      "epoch": 2.115719762983618,
+      "grad_norm": 0.07281683385372162,
+      "learning_rate": 1.153712094806553e-05,
+      "loss": 0.2128,
+      "step": 6070
+    },
+    {
+      "epoch": 2.119205298013245,
+      "grad_norm": 0.1039639413356781,
+      "learning_rate": 1.152317880794702e-05,
+      "loss": 0.0335,
+      "step": 6080
+    },
+    {
+      "epoch": 2.122690833042872,
+      "grad_norm": 6.56930685043335,
+      "learning_rate": 1.1509236667828511e-05,
+      "loss": 0.1515,
+      "step": 6090
+    },
+    {
+      "epoch": 2.126176368072499,
+      "grad_norm": 0.03588613122701645,
+      "learning_rate": 1.1495294527710003e-05,
+      "loss": 0.2306,
+      "step": 6100
+    },
+    {
+      "epoch": 2.129661903102126,
+      "grad_norm": 11.106224060058594,
+      "learning_rate": 1.1481352387591496e-05,
+      "loss": 0.1734,
+      "step": 6110
+    },
+    {
+      "epoch": 2.1331474381317532,
+      "grad_norm": 26.07666778564453,
+      "learning_rate": 1.1467410247472988e-05,
+      "loss": 0.2769,
+      "step": 6120
+    },
+    {
+      "epoch": 2.1366329731613805,
+      "grad_norm": 10.249650955200195,
+      "learning_rate": 1.145346810735448e-05,
+      "loss": 0.1453,
+      "step": 6130
+    },
+    {
+      "epoch": 2.1401185081910072,
+      "grad_norm": 2.9616682529449463,
+      "learning_rate": 1.1439525967235972e-05,
+      "loss": 0.1438,
+      "step": 6140
+    },
+    {
+      "epoch": 2.1436040432206345,
+      "grad_norm": 0.1858883649110794,
+      "learning_rate": 1.1425583827117464e-05,
+      "loss": 0.0802,
+      "step": 6150
+    },
+    {
+      "epoch": 2.1470895782502613,
+      "grad_norm": 0.5032067894935608,
+      "learning_rate": 1.1411641686998955e-05,
+      "loss": 0.1924,
+      "step": 6160
+    },
+    {
+      "epoch": 2.1505751132798885,
+      "grad_norm": 0.04852207750082016,
+      "learning_rate": 1.1397699546880448e-05,
+      "loss": 0.0768,
+      "step": 6170
+    },
+    {
+      "epoch": 2.1540606483095157,
+      "grad_norm": 0.9909989833831787,
+      "learning_rate": 1.138375740676194e-05,
+      "loss": 0.0138,
+      "step": 6180
+    },
+    {
+      "epoch": 2.1575461833391425,
+      "grad_norm": 0.036640316247940063,
+      "learning_rate": 1.1369815266643431e-05,
+      "loss": 0.0621,
+      "step": 6190
+    },
+    {
+      "epoch": 2.1610317183687697,
+      "grad_norm": 11.941831588745117,
+      "learning_rate": 1.1355873126524923e-05,
+      "loss": 0.1628,
+      "step": 6200
+    },
+    {
+      "epoch": 2.1645172533983965,
+      "grad_norm": 0.07637592405080795,
+      "learning_rate": 1.1341930986406416e-05,
+      "loss": 0.1821,
+      "step": 6210
+    },
+    {
+      "epoch": 2.1680027884280237,
+      "grad_norm": 0.07637934386730194,
+      "learning_rate": 1.1327988846287905e-05,
+      "loss": 0.1699,
+      "step": 6220
+    },
+    {
+      "epoch": 2.171488323457651,
+      "grad_norm": 0.0320703499019146,
+      "learning_rate": 1.1314046706169398e-05,
+      "loss": 0.136,
+      "step": 6230
+    },
+    {
+      "epoch": 2.1749738584872778,
+      "grad_norm": 0.06301452219486237,
+      "learning_rate": 1.1300104566050888e-05,
+      "loss": 0.1863,
+      "step": 6240
+    },
+    {
+      "epoch": 2.178459393516905,
+      "grad_norm": 0.03842555731534958,
+      "learning_rate": 1.128616242593238e-05,
+      "loss": 0.0453,
+      "step": 6250
+    },
+    {
+      "epoch": 2.1819449285465318,
+      "grad_norm": 0.03716614842414856,
+      "learning_rate": 1.1272220285813873e-05,
+      "loss": 0.1863,
+      "step": 6260
+    },
+    {
+      "epoch": 2.185430463576159,
+      "grad_norm": 0.043157655745744705,
+      "learning_rate": 1.1258278145695364e-05,
+      "loss": 0.2721,
+      "step": 6270
+    },
+    {
+      "epoch": 2.188915998605786,
+      "grad_norm": 0.07806668430566788,
+      "learning_rate": 1.1244336005576857e-05,
+      "loss": 0.1212,
+      "step": 6280
+    },
+    {
+      "epoch": 2.192401533635413,
+      "grad_norm": 0.025718241930007935,
+      "learning_rate": 1.123039386545835e-05,
+      "loss": 0.0891,
+      "step": 6290
+    },
+    {
+      "epoch": 2.1958870686650402,
+      "grad_norm": 1.4488409757614136,
+      "learning_rate": 1.121645172533984e-05,
+      "loss": 0.1036,
+      "step": 6300
+    },
+    {
+      "epoch": 2.199372603694667,
+      "grad_norm": 0.10433078557252884,
+      "learning_rate": 1.1202509585221333e-05,
+      "loss": 0.2275,
+      "step": 6310
+    },
+    {
+      "epoch": 2.2028581387242943,
+      "grad_norm": 0.049195412546396255,
+      "learning_rate": 1.1188567445102825e-05,
+      "loss": 0.01,
+      "step": 6320
+    },
+    {
+      "epoch": 2.206343673753921,
+      "grad_norm": 1.253458857536316,
+      "learning_rate": 1.1174625304984318e-05,
+      "loss": 0.0472,
+      "step": 6330
+    },
+    {
+      "epoch": 2.2098292087835483,
+      "grad_norm": 0.04194582253694534,
+      "learning_rate": 1.1160683164865808e-05,
+      "loss": 0.2199,
+      "step": 6340
+    },
+    {
+      "epoch": 2.2133147438131755,
+      "grad_norm": 0.22297650575637817,
+      "learning_rate": 1.11467410247473e-05,
+      "loss": 0.0768,
+      "step": 6350
+    },
+    {
+      "epoch": 2.2168002788428023,
+      "grad_norm": 0.03070775978267193,
+      "learning_rate": 1.113279888462879e-05,
+      "loss": 0.2436,
+      "step": 6360
+    },
+    {
+      "epoch": 2.2202858138724295,
+      "grad_norm": 0.04577281326055527,
+      "learning_rate": 1.1118856744510282e-05,
+      "loss": 0.0474,
+      "step": 6370
+    },
+    {
+      "epoch": 2.2237713489020563,
+      "grad_norm": 0.07843345403671265,
+      "learning_rate": 1.1104914604391775e-05,
+      "loss": 0.0127,
+      "step": 6380
+    },
+    {
+      "epoch": 2.2272568839316835,
+      "grad_norm": 0.08817867934703827,
+      "learning_rate": 1.1090972464273266e-05,
+      "loss": 0.2568,
+      "step": 6390
+    },
+    {
+      "epoch": 2.2307424189613108,
+      "grad_norm": 0.16615049540996552,
+      "learning_rate": 1.1077030324154758e-05,
+      "loss": 0.0813,
+      "step": 6400
+    },
+    {
+      "epoch": 2.2342279539909375,
+      "grad_norm": 0.0826653391122818,
+      "learning_rate": 1.106308818403625e-05,
+      "loss": 0.0563,
+      "step": 6410
+    },
+    {
+      "epoch": 2.2377134890205648,
+      "grad_norm": 0.05890136957168579,
+      "learning_rate": 1.1049146043917742e-05,
+      "loss": 0.1036,
+      "step": 6420
+    },
+    {
+      "epoch": 2.2411990240501916,
+      "grad_norm": 0.02518044412136078,
+      "learning_rate": 1.1035203903799234e-05,
+      "loss": 0.3812,
+      "step": 6430
+    },
+    {
+      "epoch": 2.244684559079819,
+      "grad_norm": 10.567590713500977,
+      "learning_rate": 1.1021261763680727e-05,
+      "loss": 0.1296,
+      "step": 6440
+    },
+    {
+      "epoch": 2.2481700941094456,
+      "grad_norm": 34.6755485534668,
+      "learning_rate": 1.1007319623562217e-05,
+      "loss": 0.1012,
+      "step": 6450
+    },
+    {
+      "epoch": 2.251655629139073,
+      "grad_norm": 15.21702766418457,
+      "learning_rate": 1.099337748344371e-05,
+      "loss": 0.104,
+      "step": 6460
+    },
+    {
+      "epoch": 2.2551411641687,
+      "grad_norm": 0.03763876110315323,
+      "learning_rate": 1.0979435343325202e-05,
+      "loss": 0.1439,
+      "step": 6470
+    },
+    {
+      "epoch": 2.258626699198327,
+      "grad_norm": 1.3450989723205566,
+      "learning_rate": 1.0965493203206693e-05,
+      "loss": 0.1072,
+      "step": 6480
+    },
+    {
+      "epoch": 2.262112234227954,
+      "grad_norm": 0.040909543633461,
+      "learning_rate": 1.0951551063088184e-05,
+      "loss": 0.1536,
+      "step": 6490
+    },
+    {
+      "epoch": 2.265597769257581,
+      "grad_norm": 11.683106422424316,
+      "learning_rate": 1.0937608922969677e-05,
+      "loss": 0.1513,
+      "step": 6500
+    },
+    {
+      "epoch": 2.269083304287208,
+      "grad_norm": 17.460403442382812,
+      "learning_rate": 1.0923666782851167e-05,
+      "loss": 0.2438,
+      "step": 6510
+    },
+    {
+      "epoch": 2.2725688393168353,
+      "grad_norm": 0.03460359200835228,
+      "learning_rate": 1.090972464273266e-05,
+      "loss": 0.0227,
+      "step": 6520
+    },
+    {
+      "epoch": 2.276054374346462,
+      "grad_norm": 0.37530404329299927,
+      "learning_rate": 1.0895782502614152e-05,
+      "loss": 0.3024,
+      "step": 6530
+    },
+    {
+      "epoch": 2.2795399093760893,
+      "grad_norm": 2.3782732486724854,
+      "learning_rate": 1.0881840362495643e-05,
+      "loss": 0.0873,
+      "step": 6540
+    },
+    {
+      "epoch": 2.283025444405716,
+      "grad_norm": 0.10322298854589462,
+      "learning_rate": 1.0867898222377136e-05,
+      "loss": 0.0503,
+      "step": 6550
+    },
+    {
+      "epoch": 2.2865109794353433,
+      "grad_norm": 0.08421849459409714,
+      "learning_rate": 1.0853956082258628e-05,
+      "loss": 0.2299,
+      "step": 6560
+    },
+    {
+      "epoch": 2.2899965144649705,
+      "grad_norm": 0.02602728269994259,
+      "learning_rate": 1.0840013942140119e-05,
+      "loss": 0.0616,
+      "step": 6570
+    },
+    {
+      "epoch": 2.2934820494945973,
+      "grad_norm": 0.14224806427955627,
+      "learning_rate": 1.0826071802021612e-05,
+      "loss": 0.1129,
+      "step": 6580
+    },
+    {
+      "epoch": 2.2969675845242246,
+      "grad_norm": 0.02626832202076912,
+      "learning_rate": 1.0812129661903104e-05,
+      "loss": 0.1778,
+      "step": 6590
+    },
+    {
+      "epoch": 2.3004531195538513,
+      "grad_norm": 0.044093526899814606,
+      "learning_rate": 1.0798187521784595e-05,
+      "loss": 0.3261,
+      "step": 6600
+    },
+    {
+      "epoch": 2.3039386545834786,
+      "grad_norm": 10.756707191467285,
+      "learning_rate": 1.0784245381666087e-05,
+      "loss": 0.0347,
+      "step": 6610
+    },
+    {
+      "epoch": 2.307424189613106,
+      "grad_norm": 21.886564254760742,
+      "learning_rate": 1.077030324154758e-05,
+      "loss": 0.0578,
+      "step": 6620
+    },
+    {
+      "epoch": 2.3109097246427326,
+      "grad_norm": 0.18429026007652283,
+      "learning_rate": 1.0756361101429069e-05,
+      "loss": 0.2541,
+      "step": 6630
+    },
+    {
+      "epoch": 2.31439525967236,
+      "grad_norm": 0.07464715838432312,
+      "learning_rate": 1.0742418961310562e-05,
+      "loss": 0.1547,
+      "step": 6640
+    },
+    {
+      "epoch": 2.3178807947019866,
+      "grad_norm": 3.714336395263672,
+      "learning_rate": 1.0728476821192052e-05,
+      "loss": 0.2902,
+      "step": 6650
+    },
+    {
+      "epoch": 2.321366329731614,
+      "grad_norm": 0.04718979448080063,
+      "learning_rate": 1.0714534681073545e-05,
+      "loss": 0.1189,
+      "step": 6660
+    },
+    {
+      "epoch": 2.324851864761241,
+      "grad_norm": 0.9812682867050171,
+      "learning_rate": 1.0700592540955037e-05,
+      "loss": 0.2747,
+      "step": 6670
+    },
+    {
+      "epoch": 2.328337399790868,
+      "grad_norm": 5.4444684982299805,
+      "learning_rate": 1.0686650400836528e-05,
+      "loss": 0.1446,
+      "step": 6680
+    },
+    {
+      "epoch": 2.331822934820495,
+      "grad_norm": 59.1495475769043,
+      "learning_rate": 1.067270826071802e-05,
+      "loss": 0.0812,
+      "step": 6690
+    },
+    {
+      "epoch": 2.335308469850122,
+      "grad_norm": 0.24086332321166992,
+      "learning_rate": 1.0658766120599513e-05,
+      "loss": 0.1042,
+      "step": 6700
+    },
+    {
+      "epoch": 2.338794004879749,
+      "grad_norm": 0.045006413012742996,
+      "learning_rate": 1.0644823980481006e-05,
+      "loss": 0.0306,
+      "step": 6710
+    },
+    {
+      "epoch": 2.3422795399093763,
+      "grad_norm": 0.03548438847064972,
+      "learning_rate": 1.0630881840362497e-05,
+      "loss": 0.0339,
+      "step": 6720
+    },
+    {
+      "epoch": 2.345765074939003,
+      "grad_norm": 0.043661557137966156,
+      "learning_rate": 1.0616939700243989e-05,
+      "loss": 0.1046,
+      "step": 6730
+    },
+    {
+      "epoch": 2.3492506099686303,
+      "grad_norm": 0.03901302441954613,
+      "learning_rate": 1.0602997560125482e-05,
+      "loss": 0.1696,
+      "step": 6740
+    },
+    {
+      "epoch": 2.352736144998257,
+      "grad_norm": 31.03610610961914,
+      "learning_rate": 1.0589055420006972e-05,
+      "loss": 0.464,
+      "step": 6750
+    },
+    {
+      "epoch": 2.3562216800278843,
+      "grad_norm": 1.4271265268325806,
+      "learning_rate": 1.0575113279888465e-05,
+      "loss": 0.0679,
+      "step": 6760
+    },
+    {
+      "epoch": 2.3597072150575116,
+      "grad_norm": 15.831421852111816,
+      "learning_rate": 1.0561171139769954e-05,
+      "loss": 0.1207,
+      "step": 6770
+    },
+    {
+      "epoch": 2.3631927500871384,
+      "grad_norm": 0.0829152911901474,
+      "learning_rate": 1.0547228999651447e-05,
+      "loss": 0.1525,
+      "step": 6780
+    },
+    {
+      "epoch": 2.3666782851167656,
+      "grad_norm": 0.04117709770798683,
+      "learning_rate": 1.0533286859532939e-05,
+      "loss": 0.1828,
+      "step": 6790
+    },
+    {
+      "epoch": 2.3701638201463924,
+      "grad_norm": 0.49705639481544495,
+      "learning_rate": 1.051934471941443e-05,
+      "loss": 0.2104,
+      "step": 6800
+    },
+    {
+      "epoch": 2.3736493551760196,
+      "grad_norm": 24.12613296508789,
+      "learning_rate": 1.0505402579295922e-05,
+      "loss": 0.2642,
+      "step": 6810
+    },
+    {
+      "epoch": 2.3771348902056464,
+      "grad_norm": 9.280982971191406,
+      "learning_rate": 1.0491460439177415e-05,
+      "loss": 0.1206,
+      "step": 6820
+    },
+    {
+      "epoch": 2.3806204252352736,
+      "grad_norm": 0.5508664846420288,
+      "learning_rate": 1.0477518299058906e-05,
+      "loss": 0.1583,
+      "step": 6830
+    },
+    {
+      "epoch": 2.384105960264901,
+      "grad_norm": 9.97512149810791,
+      "learning_rate": 1.0463576158940398e-05,
+      "loss": 0.112,
+      "step": 6840
+    },
+    {
+      "epoch": 2.3875914952945276,
+      "grad_norm": 0.03401152417063713,
+      "learning_rate": 1.044963401882189e-05,
+      "loss": 0.0891,
+      "step": 6850
+    },
+    {
+      "epoch": 2.391077030324155,
+      "grad_norm": 9.554537773132324,
+      "learning_rate": 1.0435691878703382e-05,
+      "loss": 0.1256,
+      "step": 6860
+    },
+    {
+      "epoch": 2.3945625653537816,
+      "grad_norm": 10.09185791015625,
+      "learning_rate": 1.0421749738584874e-05,
+      "loss": 0.1756,
+      "step": 6870
+    },
+    {
+      "epoch": 2.398048100383409,
+      "grad_norm": 0.05580438673496246,
+      "learning_rate": 1.0407807598466367e-05,
+      "loss": 0.3326,
+      "step": 6880
+    },
+    {
+      "epoch": 2.4015336354130357,
+      "grad_norm": 0.7927899360656738,
+      "learning_rate": 1.0393865458347859e-05,
+      "loss": 0.2091,
+      "step": 6890
+    },
+    {
+      "epoch": 2.405019170442663,
+      "grad_norm": 4.687802791595459,
+      "learning_rate": 1.0379923318229348e-05,
+      "loss": 0.1468,
+      "step": 6900
+    },
+    {
+      "epoch": 2.40850470547229,
+      "grad_norm": 0.03195074200630188,
+      "learning_rate": 1.036598117811084e-05,
+      "loss": 0.0722,
+      "step": 6910
+    },
+    {
+      "epoch": 2.411990240501917,
+      "grad_norm": 7.4297590255737305,
+      "learning_rate": 1.0352039037992331e-05,
+      "loss": 0.1048,
+      "step": 6920
+    },
+    {
+      "epoch": 2.415475775531544,
+      "grad_norm": 0.04244573786854744,
+      "learning_rate": 1.0338096897873824e-05,
+      "loss": 0.1695,
+      "step": 6930
+    },
+    {
+      "epoch": 2.418961310561171,
+      "grad_norm": 0.28662291169166565,
+      "learning_rate": 1.0324154757755317e-05,
+      "loss": 0.0572,
+      "step": 6940
+    },
+    {
+      "epoch": 2.422446845590798,
+      "grad_norm": 8.817971229553223,
+      "learning_rate": 1.0310212617636807e-05,
+      "loss": 0.169,
+      "step": 6950
+    },
+    {
+      "epoch": 2.4259323806204254,
+      "grad_norm": 0.06383698433637619,
+      "learning_rate": 1.02962704775183e-05,
+      "loss": 0.0905,
+      "step": 6960
+    },
+    {
+      "epoch": 2.429417915650052,
+      "grad_norm": 0.18662381172180176,
+      "learning_rate": 1.0282328337399792e-05,
+      "loss": 0.0093,
+      "step": 6970
+    },
+    {
+      "epoch": 2.4329034506796794,
+      "grad_norm": 23.655628204345703,
+      "learning_rate": 1.0268386197281283e-05,
+      "loss": 0.1657,
+      "step": 6980
+    },
+    {
+      "epoch": 2.436388985709306,
+      "grad_norm": 19.459394454956055,
+      "learning_rate": 1.0254444057162776e-05,
+      "loss": 0.1519,
+      "step": 6990
+    },
+    {
+      "epoch": 2.4398745207389334,
+      "grad_norm": 9.555098533630371,
+      "learning_rate": 1.0240501917044268e-05,
+      "loss": 0.1095,
+      "step": 7000
+    },
+    {
+      "epoch": 2.4433600557685606,
+      "grad_norm": 0.10387060046195984,
+      "learning_rate": 1.0226559776925759e-05,
+      "loss": 0.1536,
+      "step": 7010
+    },
+    {
+      "epoch": 2.4468455907981874,
+      "grad_norm": 0.0778062492609024,
+      "learning_rate": 1.0212617636807251e-05,
+      "loss": 0.0229,
+      "step": 7020
+    },
+    {
+      "epoch": 2.4503311258278146,
+      "grad_norm": 15.077468872070312,
+      "learning_rate": 1.0198675496688744e-05,
+      "loss": 0.2126,
+      "step": 7030
+    },
+    {
+      "epoch": 2.4538166608574414,
+      "grad_norm": 0.02709888108074665,
+      "learning_rate": 1.0184733356570233e-05,
+      "loss": 0.0632,
+      "step": 7040
+    },
+    {
+      "epoch": 2.4573021958870687,
+      "grad_norm": 0.04719749838113785,
+      "learning_rate": 1.0170791216451726e-05,
+      "loss": 0.2281,
+      "step": 7050
+    },
+    {
+      "epoch": 2.460787730916696,
+      "grad_norm": 12.81986141204834,
+      "learning_rate": 1.0156849076333216e-05,
+      "loss": 0.2326,
+      "step": 7060
+    },
+    {
+      "epoch": 2.4642732659463227,
+      "grad_norm": 0.12026797980070114,
+      "learning_rate": 1.0142906936214709e-05,
+      "loss": 0.1655,
+      "step": 7070
+    },
+    {
+      "epoch": 2.46775880097595,
+      "grad_norm": 0.03786276653409004,
+      "learning_rate": 1.0128964796096201e-05,
+      "loss": 0.1009,
+      "step": 7080
+    },
+    {
+      "epoch": 2.4712443360055767,
+      "grad_norm": 0.12191443145275116,
+      "learning_rate": 1.0115022655977694e-05,
+      "loss": 0.1566,
+      "step": 7090
+    },
+    {
+      "epoch": 2.474729871035204,
+      "grad_norm": 3.893367290496826,
+      "learning_rate": 1.0101080515859185e-05,
+      "loss": 0.152,
+      "step": 7100
+    },
+    {
+      "epoch": 2.478215406064831,
+      "grad_norm": 0.02769341878592968,
+      "learning_rate": 1.0087138375740677e-05,
+      "loss": 0.1226,
+      "step": 7110
+    },
+    {
+      "epoch": 2.481700941094458,
+      "grad_norm": 14.743918418884277,
+      "learning_rate": 1.007319623562217e-05,
+      "loss": 0.0629,
+      "step": 7120
+    },
+    {
+      "epoch": 2.485186476124085,
+      "grad_norm": 0.024167189374566078,
+      "learning_rate": 1.005925409550366e-05,
+      "loss": 0.1118,
+      "step": 7130
+    },
+    {
+      "epoch": 2.488672011153712,
+      "grad_norm": 0.027763212099671364,
+      "learning_rate": 1.0045311955385153e-05,
+      "loss": 0.1814,
+      "step": 7140
+    },
+    {
+      "epoch": 2.492157546183339,
+      "grad_norm": 0.03643191605806351,
+      "learning_rate": 1.0031369815266646e-05,
+      "loss": 0.1689,
+      "step": 7150
+    },
+    {
+      "epoch": 2.4956430812129664,
+      "grad_norm": 9.347540855407715,
+      "learning_rate": 1.0017427675148136e-05,
+      "loss": 0.1229,
+      "step": 7160
+    },
+    {
+      "epoch": 2.499128616242593,
+      "grad_norm": 0.03700919449329376,
+      "learning_rate": 1.0003485535029629e-05,
+      "loss": 0.1254,
+      "step": 7170
+    },
+    {
+      "epoch": 2.5026141512722204,
+      "grad_norm": 0.03421923145651817,
+      "learning_rate": 9.98954339491112e-06,
+      "loss": 0.086,
+      "step": 7180
+    },
+    {
+      "epoch": 2.506099686301847,
+      "grad_norm": 9.573734283447266,
+      "learning_rate": 9.975601254792612e-06,
+      "loss": 0.2747,
+      "step": 7190
+    },
+    {
+      "epoch": 2.5095852213314744,
+      "grad_norm": 0.4348721504211426,
+      "learning_rate": 9.961659114674103e-06,
+      "loss": 0.1462,
+      "step": 7200
+    },
+    {
+      "epoch": 2.5130707563611017,
+      "grad_norm": 0.0570727176964283,
+      "learning_rate": 9.947716974555594e-06,
+      "loss": 0.3057,
+      "step": 7210
+    },
+    {
+      "epoch": 2.5165562913907285,
+      "grad_norm": 8.233874320983887,
+      "learning_rate": 9.933774834437086e-06,
+      "loss": 0.1181,
+      "step": 7220
+    },
+    {
+      "epoch": 2.5200418264203557,
+      "grad_norm": 0.04349507763981819,
+      "learning_rate": 9.919832694318579e-06,
+      "loss": 0.0152,
+      "step": 7230
+    },
+    {
+      "epoch": 2.5235273614499825,
+      "grad_norm": 0.12035762518644333,
+      "learning_rate": 9.90589055420007e-06,
+      "loss": 0.2576,
+      "step": 7240
+    },
+    {
+      "epoch": 2.5270128964796097,
+      "grad_norm": 1.1068867444992065,
+      "learning_rate": 9.891948414081562e-06,
+      "loss": 0.3256,
+      "step": 7250
+    },
+    {
+      "epoch": 2.530498431509237,
+      "grad_norm": 0.03243900090456009,
+      "learning_rate": 9.878006273963055e-06,
+      "loss": 0.1656,
+      "step": 7260
+    },
+    {
+      "epoch": 2.5339839665388637,
+      "grad_norm": 0.02715582214295864,
+      "learning_rate": 9.864064133844546e-06,
+      "loss": 0.053,
+      "step": 7270
+    },
+    {
+      "epoch": 2.5374695015684905,
+      "grad_norm": 1.750795602798462,
+      "learning_rate": 9.850121993726038e-06,
+      "loss": 0.1305,
+      "step": 7280
+    },
+    {
+      "epoch": 2.5409550365981177,
+      "grad_norm": 2.8734793663024902,
+      "learning_rate": 9.836179853607529e-06,
+      "loss": 0.019,
+      "step": 7290
+    },
+    {
+      "epoch": 2.544440571627745,
+      "grad_norm": 0.04098232463002205,
+      "learning_rate": 9.822237713489021e-06,
+      "loss": 0.0792,
+      "step": 7300
+    },
+    {
+      "epoch": 2.547926106657372,
+      "grad_norm": 0.94338059425354,
+      "learning_rate": 9.808295573370514e-06,
+      "loss": 0.0443,
+      "step": 7310
+    },
+    {
+      "epoch": 2.551411641686999,
+      "grad_norm": 0.20982791483402252,
+      "learning_rate": 9.794353433252005e-06,
+      "loss": 0.0906,
+      "step": 7320
+    },
+    {
+      "epoch": 2.5548971767166258,
+      "grad_norm": 15.172703742980957,
+      "learning_rate": 9.780411293133497e-06,
+      "loss": 0.3467,
+      "step": 7330
+    },
+    {
+      "epoch": 2.558382711746253,
+      "grad_norm": 0.26785117387771606,
+      "learning_rate": 9.76646915301499e-06,
+      "loss": 0.0437,
+      "step": 7340
+    },
+    {
+      "epoch": 2.56186824677588,
+      "grad_norm": 12.561800003051758,
+      "learning_rate": 9.75252701289648e-06,
+      "loss": 0.1028,
+      "step": 7350
+    },
+    {
+      "epoch": 2.565353781805507,
+      "grad_norm": 20.95243263244629,
+      "learning_rate": 9.738584872777971e-06,
+      "loss": 0.1875,
+      "step": 7360
+    },
+    {
+      "epoch": 2.5688393168351342,
+      "grad_norm": 0.034853193908929825,
+      "learning_rate": 9.724642732659464e-06,
+      "loss": 0.0107,
+      "step": 7370
+    },
+    {
+      "epoch": 2.572324851864761,
+      "grad_norm": 3.5008201599121094,
+      "learning_rate": 9.710700592540956e-06,
+      "loss": 0.1182,
+      "step": 7380
+    },
+    {
+      "epoch": 2.5758103868943882,
+      "grad_norm": 0.5372085571289062,
+      "learning_rate": 9.696758452422447e-06,
+      "loss": 0.0679,
+      "step": 7390
+    },
+    {
+      "epoch": 2.5792959219240155,
+      "grad_norm": 0.05770620331168175,
+      "learning_rate": 9.68281631230394e-06,
+      "loss": 0.0422,
+      "step": 7400
+    },
+    {
+      "epoch": 2.5827814569536423,
+      "grad_norm": 0.0414259098470211,
+      "learning_rate": 9.668874172185432e-06,
+      "loss": 0.0136,
+      "step": 7410
+    },
+    {
+      "epoch": 2.5862669919832695,
+      "grad_norm": 0.03852393478155136,
+      "learning_rate": 9.654932032066923e-06,
+      "loss": 0.2012,
+      "step": 7420
+    },
+    {
+      "epoch": 2.5897525270128963,
+      "grad_norm": 13.666277885437012,
+      "learning_rate": 9.640989891948414e-06,
+      "loss": 0.0595,
+      "step": 7430
+    },
+    {
+      "epoch": 2.5932380620425235,
+      "grad_norm": 0.4676874279975891,
+      "learning_rate": 9.627047751829906e-06,
+      "loss": 0.2056,
+      "step": 7440
+    },
+    {
+      "epoch": 2.5967235970721507,
+      "grad_norm": 2.7470552921295166,
+      "learning_rate": 9.613105611711399e-06,
+      "loss": 0.0593,
+      "step": 7450
+    },
+    {
+      "epoch": 2.6002091321017775,
+      "grad_norm": 0.027122966945171356,
+      "learning_rate": 9.59916347159289e-06,
+      "loss": 0.1028,
+      "step": 7460
+    },
+    {
+      "epoch": 2.6036946671314047,
+      "grad_norm": 0.020191051065921783,
+      "learning_rate": 9.585221331474382e-06,
+      "loss": 0.0137,
+      "step": 7470
+    },
+    {
+      "epoch": 2.6071802021610315,
+      "grad_norm": 5.530102729797363,
+      "learning_rate": 9.571279191355873e-06,
+      "loss": 0.0952,
+      "step": 7480
+    },
+    {
+      "epoch": 2.6106657371906588,
+      "grad_norm": 30.30436897277832,
+      "learning_rate": 9.557337051237366e-06,
+      "loss": 0.1527,
+      "step": 7490
+    },
+    {
+      "epoch": 2.614151272220286,
+      "grad_norm": 0.10788305848836899,
+      "learning_rate": 9.543394911118858e-06,
+      "loss": 0.1038,
+      "step": 7500
+    },
+    {
+      "epoch": 2.6176368072499128,
+      "grad_norm": 4.27375602722168,
+      "learning_rate": 9.529452771000349e-06,
+      "loss": 0.1816,
+      "step": 7510
+    },
+    {
+      "epoch": 2.62112234227954,
+      "grad_norm": 0.06542309373617172,
+      "learning_rate": 9.515510630881841e-06,
+      "loss": 0.1523,
+      "step": 7520
+    },
+    {
+      "epoch": 2.624607877309167,
+      "grad_norm": 13.549959182739258,
+      "learning_rate": 9.501568490763334e-06,
+      "loss": 0.208,
+      "step": 7530
+    },
+    {
+      "epoch": 2.628093412338794,
+      "grad_norm": 15.343768119812012,
+      "learning_rate": 9.487626350644825e-06,
+      "loss": 0.2428,
+      "step": 7540
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.05019600689411163,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.249,
+      "step": 7550
+    },
+    {
+      "epoch": 2.635064482398048,
+      "grad_norm": 13.942766189575195,
+      "learning_rate": 9.459742070407808e-06,
+      "loss": 0.215,
+      "step": 7560
+    },
+    {
+      "epoch": 2.6385500174276753,
+      "grad_norm": 0.032675858587026596,
+      "learning_rate": 9.4457999302893e-06,
+      "loss": 0.133,
+      "step": 7570
+    },
+    {
+      "epoch": 2.642035552457302,
+      "grad_norm": 0.05308655649423599,
+      "learning_rate": 9.431857790170791e-06,
+      "loss": 0.218,
+      "step": 7580
+    },
+    {
+      "epoch": 2.6455210874869293,
+      "grad_norm": 0.03882508724927902,
+      "learning_rate": 9.417915650052284e-06,
+      "loss": 0.1013,
+      "step": 7590
+    },
+    {
+      "epoch": 2.6490066225165565,
+      "grad_norm": 0.08975423872470856,
+      "learning_rate": 9.403973509933776e-06,
+      "loss": 0.3583,
+      "step": 7600
+    },
+    {
+      "epoch": 2.6524921575461833,
+      "grad_norm": 5.848064422607422,
+      "learning_rate": 9.390031369815267e-06,
+      "loss": 0.117,
+      "step": 7610
+    },
+    {
+      "epoch": 2.6559776925758105,
+      "grad_norm": 9.604795455932617,
+      "learning_rate": 9.376089229696758e-06,
+      "loss": 0.2501,
+      "step": 7620
+    },
+    {
+      "epoch": 2.6594632276054373,
+      "grad_norm": 0.13165131211280823,
+      "learning_rate": 9.36214708957825e-06,
+      "loss": 0.1344,
+      "step": 7630
+    },
+    {
+      "epoch": 2.6629487626350645,
+      "grad_norm": 0.2645893692970276,
+      "learning_rate": 9.348204949459743e-06,
+      "loss": 0.0126,
+      "step": 7640
+    },
+    {
+      "epoch": 2.6664342976646918,
+      "grad_norm": 0.19416283071041107,
+      "learning_rate": 9.334262809341234e-06,
+      "loss": 0.0361,
+      "step": 7650
+    },
+    {
+      "epoch": 2.6699198326943185,
+      "grad_norm": 0.15585561096668243,
+      "learning_rate": 9.320320669222726e-06,
+      "loss": 0.0768,
+      "step": 7660
+    },
+    {
+      "epoch": 2.6734053677239458,
+      "grad_norm": 0.03315681591629982,
+      "learning_rate": 9.306378529104219e-06,
+      "loss": 0.1021,
+      "step": 7670
+    },
+    {
+      "epoch": 2.6768909027535726,
+      "grad_norm": 0.26877591013908386,
+      "learning_rate": 9.292436388985711e-06,
+      "loss": 0.1073,
+      "step": 7680
+    },
+    {
+      "epoch": 2.6803764377832,
+      "grad_norm": 0.08659540116786957,
+      "learning_rate": 9.278494248867202e-06,
+      "loss": 0.1891,
+      "step": 7690
+    },
+    {
+      "epoch": 2.683861972812827,
+      "grad_norm": 21.811100006103516,
+      "learning_rate": 9.264552108748693e-06,
+      "loss": 0.2154,
+      "step": 7700
+    },
+    {
+      "epoch": 2.687347507842454,
+      "grad_norm": 0.033906009048223495,
+      "learning_rate": 9.250609968630185e-06,
+      "loss": 0.2231,
+      "step": 7710
+    },
+    {
+      "epoch": 2.690833042872081,
+      "grad_norm": 0.03380822762846947,
+      "learning_rate": 9.236667828511678e-06,
+      "loss": 0.2515,
+      "step": 7720
+    },
+    {
+      "epoch": 2.694318577901708,
+      "grad_norm": 0.03534507378935814,
+      "learning_rate": 9.222725688393169e-06,
+      "loss": 0.1303,
+      "step": 7730
+    },
+    {
+      "epoch": 2.697804112931335,
+      "grad_norm": 0.03269047290086746,
+      "learning_rate": 9.208783548274661e-06,
+      "loss": 0.0908,
+      "step": 7740
+    },
+    {
+      "epoch": 2.7012896479609623,
+      "grad_norm": 5.287631988525391,
+      "learning_rate": 9.194841408156154e-06,
+      "loss": 0.1089,
+      "step": 7750
+    },
+    {
+      "epoch": 2.704775182990589,
+      "grad_norm": 0.03315020725131035,
+      "learning_rate": 9.180899268037645e-06,
+      "loss": 0.0785,
+      "step": 7760
+    },
+    {
+      "epoch": 2.708260718020216,
+      "grad_norm": 0.04839996621012688,
+      "learning_rate": 9.166957127919135e-06,
+      "loss": 0.1913,
+      "step": 7770
+    },
+    {
+      "epoch": 2.711746253049843,
+      "grad_norm": 17.74064826965332,
+      "learning_rate": 9.153014987800628e-06,
+      "loss": 0.3599,
+      "step": 7780
+    },
+    {
+      "epoch": 2.7152317880794703,
+      "grad_norm": 0.03409993648529053,
+      "learning_rate": 9.13907284768212e-06,
+      "loss": 0.0807,
+      "step": 7790
+    },
+    {
+      "epoch": 2.718717323109097,
+      "grad_norm": 15.558609008789062,
+      "learning_rate": 9.125130707563611e-06,
+      "loss": 0.2329,
+      "step": 7800
+    },
+    {
+      "epoch": 2.7222028581387243,
+      "grad_norm": 0.7072465419769287,
+      "learning_rate": 9.111188567445104e-06,
+      "loss": 0.2571,
+      "step": 7810
+    },
+    {
+      "epoch": 2.725688393168351,
+      "grad_norm": 0.0745309516787529,
+      "learning_rate": 9.097246427326596e-06,
+      "loss": 0.072,
+      "step": 7820
+    },
+    {
+      "epoch": 2.7291739281979783,
+      "grad_norm": 0.05214143171906471,
+      "learning_rate": 9.083304287208087e-06,
+      "loss": 0.3442,
+      "step": 7830
+    },
+    {
+      "epoch": 2.7326594632276056,
+      "grad_norm": 0.5700211524963379,
+      "learning_rate": 9.069362147089578e-06,
+      "loss": 0.1222,
+      "step": 7840
+    },
+    {
+      "epoch": 2.7361449982572323,
+      "grad_norm": 0.032385822385549545,
+      "learning_rate": 9.05542000697107e-06,
+      "loss": 0.0603,
+      "step": 7850
+    },
+    {
+      "epoch": 2.7396305332868596,
+      "grad_norm": 1.3482292890548706,
+      "learning_rate": 9.041477866852563e-06,
+      "loss": 0.018,
+      "step": 7860
+    },
+    {
+      "epoch": 2.7431160683164864,
+      "grad_norm": 0.031450305134058,
+      "learning_rate": 9.027535726734055e-06,
+      "loss": 0.2208,
+      "step": 7870
+    },
+    {
+      "epoch": 2.7466016033461136,
+      "grad_norm": 17.35287857055664,
+      "learning_rate": 9.013593586615546e-06,
+      "loss": 0.1161,
+      "step": 7880
+    },
+    {
+      "epoch": 2.750087138375741,
+      "grad_norm": 0.044247403740882874,
+      "learning_rate": 8.999651446497037e-06,
+      "loss": 0.1503,
+      "step": 7890
+    },
+    {
+      "epoch": 2.7535726734053676,
+      "grad_norm": 0.03640054911375046,
+      "learning_rate": 8.98570930637853e-06,
+      "loss": 0.1018,
+      "step": 7900
+    },
+    {
+      "epoch": 2.757058208434995,
+      "grad_norm": 2.4188730716705322,
+      "learning_rate": 8.971767166260022e-06,
+      "loss": 0.2335,
+      "step": 7910
+    },
+    {
+      "epoch": 2.7605437434646216,
+      "grad_norm": 1.5793354511260986,
+      "learning_rate": 8.957825026141513e-06,
+      "loss": 0.0794,
+      "step": 7920
+    },
+    {
+      "epoch": 2.764029278494249,
+      "grad_norm": 7.847117900848389,
+      "learning_rate": 8.943882886023005e-06,
+      "loss": 0.1338,
+      "step": 7930
+    },
+    {
+      "epoch": 2.767514813523876,
+      "grad_norm": 0.0489552803337574,
+      "learning_rate": 8.929940745904498e-06,
+      "loss": 0.2494,
+      "step": 7940
+    },
+    {
+      "epoch": 2.771000348553503,
+      "grad_norm": 0.023390300571918488,
+      "learning_rate": 8.915998605785989e-06,
+      "loss": 0.2496,
+      "step": 7950
+    },
+    {
+      "epoch": 2.77448588358313,
+      "grad_norm": 0.07795246690511703,
+      "learning_rate": 8.90205646566748e-06,
+      "loss": 0.1616,
+      "step": 7960
+    },
+    {
+      "epoch": 2.777971418612757,
+      "grad_norm": 6.65225076675415,
+      "learning_rate": 8.888114325548972e-06,
+      "loss": 0.2364,
+      "step": 7970
+    },
+    {
+      "epoch": 2.781456953642384,
+      "grad_norm": 1.1383771896362305,
+      "learning_rate": 8.874172185430465e-06,
+      "loss": 0.2328,
+      "step": 7980
+    },
+    {
+      "epoch": 2.7849424886720113,
+      "grad_norm": 0.02891557849943638,
+      "learning_rate": 8.860230045311955e-06,
+      "loss": 0.0924,
+      "step": 7990
+    },
+    {
+      "epoch": 2.788428023701638,
+      "grad_norm": 0.2302275449037552,
+      "learning_rate": 8.846287905193448e-06,
+      "loss": 0.1521,
+      "step": 8000
+    },
+    {
+      "epoch": 2.7919135587312653,
+      "grad_norm": 0.022319387644529343,
+      "learning_rate": 8.83234576507494e-06,
+      "loss": 0.2744,
+      "step": 8010
+    },
+    {
+      "epoch": 2.795399093760892,
+      "grad_norm": 10.700783729553223,
+      "learning_rate": 8.818403624956431e-06,
+      "loss": 0.113,
+      "step": 8020
+    },
+    {
+      "epoch": 2.7988846287905194,
+      "grad_norm": 8.70779800415039,
+      "learning_rate": 8.804461484837922e-06,
+      "loss": 0.0871,
+      "step": 8030
+    },
+    {
+      "epoch": 2.8023701638201466,
+      "grad_norm": 21.17486000061035,
+      "learning_rate": 8.790519344719415e-06,
+      "loss": 0.1093,
+      "step": 8040
+    },
+    {
+      "epoch": 2.8058556988497734,
+      "grad_norm": 0.03342998027801514,
+      "learning_rate": 8.776577204600907e-06,
+      "loss": 0.0455,
+      "step": 8050
+    },
+    {
+      "epoch": 2.8093412338794006,
+      "grad_norm": 0.0786479264497757,
+      "learning_rate": 8.7626350644824e-06,
+      "loss": 0.1769,
+      "step": 8060
+    },
+    {
+      "epoch": 2.8128267689090274,
+      "grad_norm": 0.03312867507338524,
+      "learning_rate": 8.74869292436389e-06,
+      "loss": 0.0294,
+      "step": 8070
+    },
+    {
+      "epoch": 2.8163123039386546,
+      "grad_norm": 0.3976738154888153,
+      "learning_rate": 8.734750784245383e-06,
+      "loss": 0.0484,
+      "step": 8080
+    },
+    {
+      "epoch": 2.819797838968282,
+      "grad_norm": 0.04097994789481163,
+      "learning_rate": 8.720808644126875e-06,
+      "loss": 0.2901,
+      "step": 8090
+    },
+    {
+      "epoch": 2.8232833739979086,
+      "grad_norm": 0.03706745803356171,
+      "learning_rate": 8.706866504008366e-06,
+      "loss": 0.1388,
+      "step": 8100
+    },
+    {
+      "epoch": 2.826768909027536,
+      "grad_norm": 2.765982151031494,
+      "learning_rate": 8.692924363889857e-06,
+      "loss": 0.2213,
+      "step": 8110
+    },
+    {
+      "epoch": 2.8302544440571626,
+      "grad_norm": 0.11387185752391815,
+      "learning_rate": 8.67898222377135e-06,
+      "loss": 0.3626,
+      "step": 8120
+    },
+    {
+      "epoch": 2.83373997908679,
+      "grad_norm": 10.69524097442627,
+      "learning_rate": 8.665040083652842e-06,
+      "loss": 0.1166,
+      "step": 8130
+    },
+    {
+      "epoch": 2.837225514116417,
+      "grad_norm": 0.020516468212008476,
+      "learning_rate": 8.651097943534333e-06,
+      "loss": 0.1093,
+      "step": 8140
+    },
+    {
+      "epoch": 2.840711049146044,
+      "grad_norm": 0.24759964644908905,
+      "learning_rate": 8.637155803415825e-06,
+      "loss": 0.2226,
+      "step": 8150
+    },
+    {
+      "epoch": 2.844196584175671,
+      "grad_norm": 0.030304880812764168,
+      "learning_rate": 8.623213663297318e-06,
+      "loss": 0.1346,
+      "step": 8160
+    },
+    {
+      "epoch": 2.847682119205298,
+      "grad_norm": 2.6233885288238525,
+      "learning_rate": 8.609271523178809e-06,
+      "loss": 0.175,
+      "step": 8170
+    },
+    {
+      "epoch": 2.851167654234925,
+      "grad_norm": 2.439249038696289,
+      "learning_rate": 8.5953293830603e-06,
+      "loss": 0.0595,
+      "step": 8180
+    },
+    {
+      "epoch": 2.8546531892645524,
+      "grad_norm": 0.10265187174081802,
+      "learning_rate": 8.581387242941792e-06,
+      "loss": 0.0761,
+      "step": 8190
+    },
+    {
+      "epoch": 2.858138724294179,
+      "grad_norm": 17.636390686035156,
+      "learning_rate": 8.567445102823285e-06,
+      "loss": 0.2557,
+      "step": 8200
+    },
+    {
+      "epoch": 2.861624259323806,
+      "grad_norm": 0.16799131035804749,
+      "learning_rate": 8.553502962704775e-06,
+      "loss": 0.0318,
+      "step": 8210
+    },
+    {
+      "epoch": 2.865109794353433,
+      "grad_norm": 0.3623441755771637,
+      "learning_rate": 8.539560822586268e-06,
+      "loss": 0.0818,
+      "step": 8220
+    },
+    {
+      "epoch": 2.8685953293830604,
+      "grad_norm": 6.526904106140137,
+      "learning_rate": 8.52561868246776e-06,
+      "loss": 0.0823,
+      "step": 8230
+    },
+    {
+      "epoch": 2.8720808644126876,
+      "grad_norm": 3.7460978031158447,
+      "learning_rate": 8.511676542349251e-06,
+      "loss": 0.2567,
+      "step": 8240
+    },
+    {
+      "epoch": 2.8755663994423144,
+      "grad_norm": 0.03773918002843857,
+      "learning_rate": 8.497734402230744e-06,
+      "loss": 0.0277,
+      "step": 8250
+    },
+    {
+      "epoch": 2.879051934471941,
+      "grad_norm": 0.03944707289338112,
+      "learning_rate": 8.483792262112234e-06,
+      "loss": 0.0547,
+      "step": 8260
+    },
+    {
+      "epoch": 2.8825374695015684,
+      "grad_norm": 0.16772694885730743,
+      "learning_rate": 8.469850121993727e-06,
+      "loss": 0.0881,
+      "step": 8270
+    },
+    {
+      "epoch": 2.8860230045311956,
+      "grad_norm": 7.099332809448242,
+      "learning_rate": 8.45590798187522e-06,
+      "loss": 0.1931,
+      "step": 8280
+    },
+    {
+      "epoch": 2.8895085395608224,
+      "grad_norm": 1.7227249145507812,
+      "learning_rate": 8.44196584175671e-06,
+      "loss": 0.1587,
+      "step": 8290
+    },
+    {
+      "epoch": 2.8929940745904497,
+      "grad_norm": 0.22683005034923553,
+      "learning_rate": 8.428023701638201e-06,
+      "loss": 0.0558,
+      "step": 8300
+    },
+    {
+      "epoch": 2.8964796096200764,
+      "grad_norm": 0.030454624444246292,
+      "learning_rate": 8.414081561519694e-06,
+      "loss": 0.1197,
+      "step": 8310
+    },
+    {
+      "epoch": 2.8999651446497037,
+      "grad_norm": 0.07277271151542664,
+      "learning_rate": 8.400139421401186e-06,
+      "loss": 0.0851,
+      "step": 8320
+    },
+    {
+      "epoch": 2.903450679679331,
+      "grad_norm": 6.2670063972473145,
+      "learning_rate": 8.386197281282677e-06,
+      "loss": 0.0101,
+      "step": 8330
+    },
+    {
+      "epoch": 2.9069362147089577,
+      "grad_norm": 25.918701171875,
+      "learning_rate": 8.37225514116417e-06,
+      "loss": 0.2227,
+      "step": 8340
+    },
+    {
+      "epoch": 2.910421749738585,
+      "grad_norm": 0.02563529834151268,
+      "learning_rate": 8.358313001045662e-06,
+      "loss": 0.2458,
+      "step": 8350
+    },
+    {
+      "epoch": 2.9139072847682117,
+      "grad_norm": 0.12701044976711273,
+      "learning_rate": 8.344370860927153e-06,
+      "loss": 0.0929,
+      "step": 8360
+    },
+    {
+      "epoch": 2.917392819797839,
+      "grad_norm": 0.0416131317615509,
+      "learning_rate": 8.330428720808644e-06,
+      "loss": 0.1182,
+      "step": 8370
+    },
+    {
+      "epoch": 2.920878354827466,
+      "grad_norm": 20.982830047607422,
+      "learning_rate": 8.316486580690136e-06,
+      "loss": 0.2912,
+      "step": 8380
+    },
+    {
+      "epoch": 2.924363889857093,
+      "grad_norm": 0.04384233430027962,
+      "learning_rate": 8.302544440571629e-06,
+      "loss": 0.0101,
+      "step": 8390
+    },
+    {
+      "epoch": 2.92784942488672,
+      "grad_norm": 0.35305264592170715,
+      "learning_rate": 8.28860230045312e-06,
+      "loss": 0.1257,
+      "step": 8400
+    },
+    {
+      "epoch": 2.931334959916347,
+      "grad_norm": 0.6124236583709717,
+      "learning_rate": 8.274660160334612e-06,
+      "loss": 0.0244,
+      "step": 8410
+    },
+    {
+      "epoch": 2.934820494945974,
+      "grad_norm": 21.486831665039062,
+      "learning_rate": 8.260718020216104e-06,
+      "loss": 0.1246,
+      "step": 8420
+    },
+    {
+      "epoch": 2.9383060299756014,
+      "grad_norm": 0.04249065741896629,
+      "learning_rate": 8.246775880097595e-06,
+      "loss": 0.1105,
+      "step": 8430
+    },
+    {
+      "epoch": 2.941791565005228,
+      "grad_norm": 22.740060806274414,
+      "learning_rate": 8.232833739979088e-06,
+      "loss": 0.3444,
+      "step": 8440
+    },
+    {
+      "epoch": 2.9452771000348554,
+      "grad_norm": 0.03325342759490013,
+      "learning_rate": 8.218891599860579e-06,
+      "loss": 0.0593,
+      "step": 8450
+    },
+    {
+      "epoch": 2.948762635064482,
+      "grad_norm": 0.06233534216880798,
+      "learning_rate": 8.204949459742071e-06,
+      "loss": 0.0664,
+      "step": 8460
+    },
+    {
+      "epoch": 2.9522481700941094,
+      "grad_norm": 0.13591769337654114,
+      "learning_rate": 8.191007319623564e-06,
+      "loss": 0.2097,
+      "step": 8470
+    },
+    {
+      "epoch": 2.9557337051237367,
+      "grad_norm": 0.034143611788749695,
+      "learning_rate": 8.177065179505054e-06,
+      "loss": 0.0226,
+      "step": 8480
+    },
+    {
+      "epoch": 2.9592192401533635,
+      "grad_norm": 0.04051746428012848,
+      "learning_rate": 8.163123039386547e-06,
+      "loss": 0.1239,
+      "step": 8490
+    },
+    {
+      "epoch": 2.9627047751829907,
+      "grad_norm": 0.17615370452404022,
+      "learning_rate": 8.14918089926804e-06,
+      "loss": 0.0881,
+      "step": 8500
+    },
+    {
+      "epoch": 2.9661903102126175,
+      "grad_norm": 10.80210018157959,
+      "learning_rate": 8.13523875914953e-06,
+      "loss": 0.1604,
+      "step": 8510
+    },
+    {
+      "epoch": 2.9696758452422447,
+      "grad_norm": 0.04547272250056267,
+      "learning_rate": 8.121296619031021e-06,
+      "loss": 0.1647,
+      "step": 8520
+    },
+    {
+      "epoch": 2.973161380271872,
+      "grad_norm": 0.026660796254873276,
+      "learning_rate": 8.107354478912514e-06,
+      "loss": 0.0634,
+      "step": 8530
+    },
+    {
+      "epoch": 2.9766469153014987,
+      "grad_norm": 0.02724742889404297,
+      "learning_rate": 8.093412338794006e-06,
+      "loss": 0.0997,
+      "step": 8540
+    },
+    {
+      "epoch": 2.980132450331126,
+      "grad_norm": 0.02272176183760166,
+      "learning_rate": 8.079470198675497e-06,
+      "loss": 0.1778,
+      "step": 8550
+    },
+    {
+      "epoch": 2.9836179853607527,
+      "grad_norm": 0.02710247039794922,
+      "learning_rate": 8.06552805855699e-06,
+      "loss": 0.1714,
+      "step": 8560
+    },
+    {
+      "epoch": 2.98710352039038,
+      "grad_norm": 3.23300838470459,
+      "learning_rate": 8.051585918438482e-06,
+      "loss": 0.1514,
+      "step": 8570
+    },
+    {
+      "epoch": 2.990589055420007,
+      "grad_norm": 19.48651123046875,
+      "learning_rate": 8.037643778319973e-06,
+      "loss": 0.101,
+      "step": 8580
+    },
+    {
+      "epoch": 2.994074590449634,
+      "grad_norm": 15.483784675598145,
+      "learning_rate": 8.023701638201464e-06,
+      "loss": 0.072,
+      "step": 8590
+    },
+    {
+      "epoch": 2.997560125479261,
+      "grad_norm": 13.562616348266602,
+      "learning_rate": 8.009759498082956e-06,
+      "loss": 0.0941,
+      "step": 8600
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9861728395061728,
+      "eval_loss": 0.05898062884807587,
+      "eval_runtime": 18.9157,
+      "eval_samples_per_second": 214.108,
+      "eval_steps_per_second": 26.803,
+      "step": 8607
+    },
+    {
+      "epoch": 3.001045660508888,
+      "grad_norm": 0.05343186855316162,
+      "learning_rate": 7.995817357964449e-06,
+      "loss": 0.1422,
+      "step": 8610
+    },
+    {
+      "epoch": 3.004531195538515,
+      "grad_norm": 24.563711166381836,
+      "learning_rate": 7.98187521784594e-06,
+      "loss": 0.0213,
+      "step": 8620
+    },
+    {
+      "epoch": 3.008016730568142,
+      "grad_norm": 0.026707723736763,
+      "learning_rate": 7.967933077727432e-06,
+      "loss": 0.031,
+      "step": 8630
+    },
+    {
+      "epoch": 3.0115022655977692,
+      "grad_norm": 0.02573820762336254,
+      "learning_rate": 7.953990937608924e-06,
+      "loss": 0.0848,
+      "step": 8640
+    },
+    {
+      "epoch": 3.0149878006273965,
+      "grad_norm": 0.020314253866672516,
+      "learning_rate": 7.940048797490415e-06,
+      "loss": 0.1811,
+      "step": 8650
+    },
+    {
+      "epoch": 3.0184733356570232,
+      "grad_norm": 0.0360274612903595,
+      "learning_rate": 7.926106657371908e-06,
+      "loss": 0.0079,
+      "step": 8660
+    },
+    {
+      "epoch": 3.0219588706866505,
+      "grad_norm": 0.34295937418937683,
+      "learning_rate": 7.912164517253399e-06,
+      "loss": 0.141,
+      "step": 8670
+    },
+    {
+      "epoch": 3.0254444057162773,
+      "grad_norm": 0.383859783411026,
+      "learning_rate": 7.898222377134891e-06,
+      "loss": 0.1501,
+      "step": 8680
+    },
+    {
+      "epoch": 3.0289299407459045,
+      "grad_norm": 0.06587964296340942,
+      "learning_rate": 7.884280237016384e-06,
+      "loss": 0.095,
+      "step": 8690
+    },
+    {
+      "epoch": 3.0324154757755317,
+      "grad_norm": 0.041697971522808075,
+      "learning_rate": 7.870338096897874e-06,
+      "loss": 0.1642,
+      "step": 8700
+    },
+    {
+      "epoch": 3.0359010108051585,
+      "grad_norm": 28.092723846435547,
+      "learning_rate": 7.856395956779365e-06,
+      "loss": 0.1282,
+      "step": 8710
+    },
+    {
+      "epoch": 3.0393865458347857,
+      "grad_norm": 1.7253819704055786,
+      "learning_rate": 7.842453816660858e-06,
+      "loss": 0.0581,
+      "step": 8720
+    },
+    {
+      "epoch": 3.0428720808644125,
+      "grad_norm": 1.214120626449585,
+      "learning_rate": 7.82851167654235e-06,
+      "loss": 0.0404,
+      "step": 8730
+    },
+    {
+      "epoch": 3.0463576158940397,
+      "grad_norm": 18.530441284179688,
+      "learning_rate": 7.814569536423841e-06,
+      "loss": 0.1388,
+      "step": 8740
+    },
+    {
+      "epoch": 3.049843150923667,
+      "grad_norm": 0.030280515551567078,
+      "learning_rate": 7.800627396305334e-06,
+      "loss": 0.0956,
+      "step": 8750
+    },
+    {
+      "epoch": 3.0533286859532938,
+      "grad_norm": 0.02037186734378338,
+      "learning_rate": 7.786685256186826e-06,
+      "loss": 0.126,
+      "step": 8760
+    },
+    {
+      "epoch": 3.056814220982921,
+      "grad_norm": 33.261940002441406,
+      "learning_rate": 7.772743116068317e-06,
+      "loss": 0.1255,
+      "step": 8770
+    },
+    {
+      "epoch": 3.060299756012548,
+      "grad_norm": 22.391355514526367,
+      "learning_rate": 7.758800975949808e-06,
+      "loss": 0.1261,
+      "step": 8780
+    },
+    {
+      "epoch": 3.063785291042175,
+      "grad_norm": 0.017767876386642456,
+      "learning_rate": 7.7448588358313e-06,
+      "loss": 0.1211,
+      "step": 8790
+    },
+    {
+      "epoch": 3.0672708260718022,
+      "grad_norm": 5.44157600402832,
+      "learning_rate": 7.730916695712793e-06,
+      "loss": 0.1787,
+      "step": 8800
+    },
+    {
+      "epoch": 3.070756361101429,
+      "grad_norm": 0.02152939699590206,
+      "learning_rate": 7.716974555594284e-06,
+      "loss": 0.0867,
+      "step": 8810
+    },
+    {
+      "epoch": 3.0742418961310563,
+      "grad_norm": 29.05510711669922,
+      "learning_rate": 7.703032415475776e-06,
+      "loss": 0.2069,
+      "step": 8820
+    },
+    {
+      "epoch": 3.077727431160683,
+      "grad_norm": 0.029673922806978226,
+      "learning_rate": 7.689090275357269e-06,
+      "loss": 0.1669,
+      "step": 8830
+    },
+    {
+      "epoch": 3.0812129661903103,
+      "grad_norm": 0.02840520814061165,
+      "learning_rate": 7.675148135238761e-06,
+      "loss": 0.2332,
+      "step": 8840
+    },
+    {
+      "epoch": 3.084698501219937,
+      "grad_norm": 0.2850615978240967,
+      "learning_rate": 7.661205995120252e-06,
+      "loss": 0.146,
+      "step": 8850
+    },
+    {
+      "epoch": 3.0881840362495643,
+      "grad_norm": 3.1782281398773193,
+      "learning_rate": 7.647263855001743e-06,
+      "loss": 0.0449,
+      "step": 8860
+    },
+    {
+      "epoch": 3.0916695712791915,
+      "grad_norm": 2.359084367752075,
+      "learning_rate": 7.633321714883235e-06,
+      "loss": 0.0291,
+      "step": 8870
+    },
+    {
+      "epoch": 3.0951551063088183,
+      "grad_norm": 0.0349331870675087,
+      "learning_rate": 7.619379574764727e-06,
+      "loss": 0.1953,
+      "step": 8880
+    },
+    {
+      "epoch": 3.0986406413384455,
+      "grad_norm": 0.03900326043367386,
+      "learning_rate": 7.605437434646219e-06,
+      "loss": 0.0101,
+      "step": 8890
+    },
+    {
+      "epoch": 3.1021261763680723,
+      "grad_norm": 0.05492232367396355,
+      "learning_rate": 7.591495294527711e-06,
+      "loss": 0.024,
+      "step": 8900
+    },
+    {
+      "epoch": 3.1056117113976995,
+      "grad_norm": 0.024204595014452934,
+      "learning_rate": 7.577553154409203e-06,
+      "loss": 0.2914,
+      "step": 8910
+    },
+    {
+      "epoch": 3.1090972464273268,
+      "grad_norm": 0.21245832741260529,
+      "learning_rate": 7.5636110142906935e-06,
+      "loss": 0.1574,
+      "step": 8920
+    },
+    {
+      "epoch": 3.1125827814569536,
+      "grad_norm": 4.7999587059021,
+      "learning_rate": 7.549668874172186e-06,
+      "loss": 0.1669,
+      "step": 8930
+    },
+    {
+      "epoch": 3.116068316486581,
+      "grad_norm": 0.0458659753203392,
+      "learning_rate": 7.535726734053678e-06,
+      "loss": 0.2493,
+      "step": 8940
+    },
+    {
+      "epoch": 3.1195538515162076,
+      "grad_norm": 0.025298912078142166,
+      "learning_rate": 7.521784593935169e-06,
+      "loss": 0.0541,
+      "step": 8950
+    },
+    {
+      "epoch": 3.123039386545835,
+      "grad_norm": 1.2849235534667969,
+      "learning_rate": 7.507842453816662e-06,
+      "loss": 0.0486,
+      "step": 8960
+    },
+    {
+      "epoch": 3.126524921575462,
+      "grad_norm": 0.03680930659174919,
+      "learning_rate": 7.4939003136981535e-06,
+      "loss": 0.1282,
+      "step": 8970
+    },
+    {
+      "epoch": 3.130010456605089,
+      "grad_norm": 13.066097259521484,
+      "learning_rate": 7.479958173579646e-06,
+      "loss": 0.2534,
+      "step": 8980
+    },
+    {
+      "epoch": 3.133495991634716,
+      "grad_norm": 0.02675449103116989,
+      "learning_rate": 7.466016033461137e-06,
+      "loss": 0.0103,
+      "step": 8990
+    },
+    {
+      "epoch": 3.136981526664343,
+      "grad_norm": 0.038677338510751724,
+      "learning_rate": 7.4520738933426285e-06,
+      "loss": 0.0742,
+      "step": 9000
+    },
+    {
+      "epoch": 3.14046706169397,
+      "grad_norm": 0.026775427162647247,
+      "learning_rate": 7.43813175322412e-06,
+      "loss": 0.0929,
+      "step": 9010
+    },
+    {
+      "epoch": 3.1439525967235973,
+      "grad_norm": 0.04959922656416893,
+      "learning_rate": 7.424189613105613e-06,
+      "loss": 0.171,
+      "step": 9020
+    },
+    {
+      "epoch": 3.147438131753224,
+      "grad_norm": 0.018528427928686142,
+      "learning_rate": 7.410247472987104e-06,
+      "loss": 0.1509,
+      "step": 9030
+    },
+    {
+      "epoch": 3.1509236667828513,
+      "grad_norm": 0.02553379535675049,
+      "learning_rate": 7.396305332868596e-06,
+      "loss": 0.0405,
+      "step": 9040
+    },
+    {
+      "epoch": 3.154409201812478,
+      "grad_norm": 0.045496754348278046,
+      "learning_rate": 7.3823631927500885e-06,
+      "loss": 0.0465,
+      "step": 9050
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 18.267221450805664,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.1788,
+      "step": 9060
+    },
+    {
+      "epoch": 3.1613802718717325,
+      "grad_norm": 0.037292297929525375,
+      "learning_rate": 7.354478912513071e-06,
+      "loss": 0.1418,
+      "step": 9070
+    },
+    {
+      "epoch": 3.1648658069013593,
+      "grad_norm": 1.9102615118026733,
+      "learning_rate": 7.3405367723945635e-06,
+      "loss": 0.0699,
+      "step": 9080
+    },
+    {
+      "epoch": 3.1683513419309866,
+      "grad_norm": 47.18280029296875,
+      "learning_rate": 7.326594632276055e-06,
+      "loss": 0.1691,
+      "step": 9090
+    },
+    {
+      "epoch": 3.1718368769606133,
+      "grad_norm": 0.13229234516620636,
+      "learning_rate": 7.312652492157547e-06,
+      "loss": 0.1772,
+      "step": 9100
+    },
+    {
+      "epoch": 3.1753224119902406,
+      "grad_norm": 2.6047537326812744,
+      "learning_rate": 7.298710352039039e-06,
+      "loss": 0.07,
+      "step": 9110
+    },
+    {
+      "epoch": 3.1788079470198674,
+      "grad_norm": 0.02106287144124508,
+      "learning_rate": 7.28476821192053e-06,
+      "loss": 0.1055,
+      "step": 9120
+    },
+    {
+      "epoch": 3.1822934820494946,
+      "grad_norm": 10.96900749206543,
+      "learning_rate": 7.270826071802022e-06,
+      "loss": 0.1797,
+      "step": 9130
+    },
+    {
+      "epoch": 3.185779017079122,
+      "grad_norm": 0.1008637398481369,
+      "learning_rate": 7.256883931683513e-06,
+      "loss": 0.0178,
+      "step": 9140
+    },
+    {
+      "epoch": 3.1892645521087486,
+      "grad_norm": 0.0249412152916193,
+      "learning_rate": 7.242941791565006e-06,
+      "loss": 0.3967,
+      "step": 9150
+    },
+    {
+      "epoch": 3.192750087138376,
+      "grad_norm": 0.025104615837335587,
+      "learning_rate": 7.228999651446498e-06,
+      "loss": 0.0881,
+      "step": 9160
+    },
+    {
+      "epoch": 3.1962356221680026,
+      "grad_norm": 0.02654813602566719,
+      "learning_rate": 7.21505751132799e-06,
+      "loss": 0.1262,
+      "step": 9170
+    },
+    {
+      "epoch": 3.19972115719763,
+      "grad_norm": 0.03677063435316086,
+      "learning_rate": 7.201115371209482e-06,
+      "loss": 0.0244,
+      "step": 9180
+    },
+    {
+      "epoch": 3.203206692227257,
+      "grad_norm": 4.220332145690918,
+      "learning_rate": 7.187173231090973e-06,
+      "loss": 0.1071,
+      "step": 9190
+    },
+    {
+      "epoch": 3.206692227256884,
+      "grad_norm": 0.019408810883760452,
+      "learning_rate": 7.173231090972464e-06,
+      "loss": 0.1627,
+      "step": 9200
+    },
+    {
+      "epoch": 3.210177762286511,
+      "grad_norm": 1.1298311948776245,
+      "learning_rate": 7.159288950853957e-06,
+      "loss": 0.1347,
+      "step": 9210
+    },
+    {
+      "epoch": 3.213663297316138,
+      "grad_norm": 0.04483231157064438,
+      "learning_rate": 7.145346810735448e-06,
+      "loss": 0.2057,
+      "step": 9220
+    },
+    {
+      "epoch": 3.217148832345765,
+      "grad_norm": 0.020672811195254326,
+      "learning_rate": 7.13140467061694e-06,
+      "loss": 0.0078,
+      "step": 9230
+    },
+    {
+      "epoch": 3.2206343673753923,
+      "grad_norm": 0.022838251665234566,
+      "learning_rate": 7.117462530498433e-06,
+      "loss": 0.1516,
+      "step": 9240
+    },
+    {
+      "epoch": 3.224119902405019,
+      "grad_norm": 0.037012387067079544,
+      "learning_rate": 7.103520390379924e-06,
+      "loss": 0.0569,
+      "step": 9250
+    },
+    {
+      "epoch": 3.2276054374346463,
+      "grad_norm": 0.2272426038980484,
+      "learning_rate": 7.089578250261415e-06,
+      "loss": 0.0739,
+      "step": 9260
+    },
+    {
+      "epoch": 3.231090972464273,
+      "grad_norm": 26.441532135009766,
+      "learning_rate": 7.0756361101429076e-06,
+      "loss": 0.0436,
+      "step": 9270
+    },
+    {
+      "epoch": 3.2345765074939004,
+      "grad_norm": 19.132356643676758,
+      "learning_rate": 7.061693970024399e-06,
+      "loss": 0.181,
+      "step": 9280
+    },
+    {
+      "epoch": 3.238062042523527,
+      "grad_norm": 0.4390275776386261,
+      "learning_rate": 7.047751829905891e-06,
+      "loss": 0.18,
+      "step": 9290
+    },
+    {
+      "epoch": 3.2415475775531544,
+      "grad_norm": 0.013922056183218956,
+      "learning_rate": 7.033809689787383e-06,
+      "loss": 0.148,
+      "step": 9300
+    },
+    {
+      "epoch": 3.2450331125827816,
+      "grad_norm": 0.01766609586775303,
+      "learning_rate": 7.019867549668875e-06,
+      "loss": 0.1277,
+      "step": 9310
+    },
+    {
+      "epoch": 3.2485186476124084,
+      "grad_norm": 0.03631417080760002,
+      "learning_rate": 7.005925409550367e-06,
+      "loss": 0.0666,
+      "step": 9320
+    },
+    {
+      "epoch": 3.2520041826420356,
+      "grad_norm": 0.022327765822410583,
+      "learning_rate": 6.9919832694318575e-06,
+      "loss": 0.1104,
+      "step": 9330
+    },
+    {
+      "epoch": 3.2554897176716624,
+      "grad_norm": 28.965906143188477,
+      "learning_rate": 6.97804112931335e-06,
+      "loss": 0.1194,
+      "step": 9340
+    },
+    {
+      "epoch": 3.2589752527012896,
+      "grad_norm": 0.016817284747958183,
+      "learning_rate": 6.964098989194842e-06,
+      "loss": 0.105,
+      "step": 9350
+    },
+    {
+      "epoch": 3.262460787730917,
+      "grad_norm": 19.65138816833496,
+      "learning_rate": 6.950156849076334e-06,
+      "loss": 0.1736,
+      "step": 9360
+    },
+    {
+      "epoch": 3.2659463227605436,
+      "grad_norm": 0.1805201917886734,
+      "learning_rate": 6.936214708957826e-06,
+      "loss": 0.2812,
+      "step": 9370
+    },
+    {
+      "epoch": 3.269431857790171,
+      "grad_norm": 3.660529613494873,
+      "learning_rate": 6.9222725688393175e-06,
+      "loss": 0.1089,
+      "step": 9380
+    },
+    {
+      "epoch": 3.2729173928197977,
+      "grad_norm": 18.988616943359375,
+      "learning_rate": 6.90833042872081e-06,
+      "loss": 0.2735,
+      "step": 9390
+    },
+    {
+      "epoch": 3.276402927849425,
+      "grad_norm": 0.05193915590643883,
+      "learning_rate": 6.894388288602301e-06,
+      "loss": 0.3342,
+      "step": 9400
+    },
+    {
+      "epoch": 3.279888462879052,
+      "grad_norm": 0.2501579225063324,
+      "learning_rate": 6.8804461484837925e-06,
+      "loss": 0.0079,
+      "step": 9410
+    },
+    {
+      "epoch": 3.283373997908679,
+      "grad_norm": 0.035554543137550354,
+      "learning_rate": 6.866504008365284e-06,
+      "loss": 0.0192,
+      "step": 9420
+    },
+    {
+      "epoch": 3.286859532938306,
+      "grad_norm": 0.020876824855804443,
+      "learning_rate": 6.852561868246777e-06,
+      "loss": 0.1817,
+      "step": 9430
+    },
+    {
+      "epoch": 3.290345067967933,
+      "grad_norm": 3.084765672683716,
+      "learning_rate": 6.838619728128268e-06,
+      "loss": 0.044,
+      "step": 9440
+    },
+    {
+      "epoch": 3.29383060299756,
+      "grad_norm": 18.532669067382812,
+      "learning_rate": 6.82467758800976e-06,
+      "loss": 0.272,
+      "step": 9450
+    },
+    {
+      "epoch": 3.2973161380271874,
+      "grad_norm": 5.328271865844727,
+      "learning_rate": 6.8107354478912525e-06,
+      "loss": 0.1708,
+      "step": 9460
+    },
+    {
+      "epoch": 3.300801673056814,
+      "grad_norm": 16.922470092773438,
+      "learning_rate": 6.796793307772743e-06,
+      "loss": 0.1577,
+      "step": 9470
+    },
+    {
+      "epoch": 3.3042872080864414,
+      "grad_norm": 20.75090980529785,
+      "learning_rate": 6.782851167654235e-06,
+      "loss": 0.2414,
+      "step": 9480
+    },
+    {
+      "epoch": 3.307772743116068,
+      "grad_norm": 0.03418176993727684,
+      "learning_rate": 6.7689090275357275e-06,
+      "loss": 0.1943,
+      "step": 9490
+    },
+    {
+      "epoch": 3.3112582781456954,
+      "grad_norm": 0.4410526156425476,
+      "learning_rate": 6.754966887417219e-06,
+      "loss": 0.0739,
+      "step": 9500
+    },
+    {
+      "epoch": 3.3147438131753226,
+      "grad_norm": 0.02994650788605213,
+      "learning_rate": 6.741024747298711e-06,
+      "loss": 0.187,
+      "step": 9510
+    },
+    {
+      "epoch": 3.3182293482049494,
+      "grad_norm": 6.901801109313965,
+      "learning_rate": 6.727082607180203e-06,
+      "loss": 0.1074,
+      "step": 9520
+    },
+    {
+      "epoch": 3.3217148832345766,
+      "grad_norm": 24.88802719116211,
+      "learning_rate": 6.713140467061694e-06,
+      "loss": 0.1535,
+      "step": 9530
+    },
+    {
+      "epoch": 3.3252004182642034,
+      "grad_norm": 0.2463180273771286,
+      "learning_rate": 6.699198326943186e-06,
+      "loss": 0.1021,
+      "step": 9540
+    },
+    {
+      "epoch": 3.3286859532938307,
+      "grad_norm": 0.09169216454029083,
+      "learning_rate": 6.685256186824678e-06,
+      "loss": 0.0811,
+      "step": 9550
+    },
+    {
+      "epoch": 3.332171488323458,
+      "grad_norm": 2.2960045337677,
+      "learning_rate": 6.67131404670617e-06,
+      "loss": 0.1068,
+      "step": 9560
+    },
+    {
+      "epoch": 3.3356570233530847,
+      "grad_norm": 0.021253207698464394,
+      "learning_rate": 6.657371906587662e-06,
+      "loss": 0.1687,
+      "step": 9570
+    },
+    {
+      "epoch": 3.339142558382712,
+      "grad_norm": 28.67542266845703,
+      "learning_rate": 6.643429766469154e-06,
+      "loss": 0.1184,
+      "step": 9580
+    },
+    {
+      "epoch": 3.3426280934123387,
+      "grad_norm": 0.02517896145582199,
+      "learning_rate": 6.629487626350646e-06,
+      "loss": 0.1933,
+      "step": 9590
+    },
+    {
+      "epoch": 3.346113628441966,
+      "grad_norm": 0.04085422307252884,
+      "learning_rate": 6.615545486232137e-06,
+      "loss": 0.0423,
+      "step": 9600
+    },
+    {
+      "epoch": 3.3495991634715927,
+      "grad_norm": 1.9252256155014038,
+      "learning_rate": 6.601603346113628e-06,
+      "loss": 0.1851,
+      "step": 9610
+    },
+    {
+      "epoch": 3.35308469850122,
+      "grad_norm": 15.485836029052734,
+      "learning_rate": 6.587661205995121e-06,
+      "loss": 0.2882,
+      "step": 9620
+    },
+    {
+      "epoch": 3.356570233530847,
+      "grad_norm": 7.36222505569458,
+      "learning_rate": 6.5737190658766125e-06,
+      "loss": 0.256,
+      "step": 9630
+    },
+    {
+      "epoch": 3.360055768560474,
+      "grad_norm": 0.07581675052642822,
+      "learning_rate": 6.559776925758104e-06,
+      "loss": 0.1117,
+      "step": 9640
+    },
+    {
+      "epoch": 3.363541303590101,
+      "grad_norm": 1.1792480945587158,
+      "learning_rate": 6.545834785639597e-06,
+      "loss": 0.2192,
+      "step": 9650
+    },
+    {
+      "epoch": 3.367026838619728,
+      "grad_norm": 0.04260709509253502,
+      "learning_rate": 6.531892645521088e-06,
+      "loss": 0.0455,
+      "step": 9660
+    },
+    {
+      "epoch": 3.370512373649355,
+      "grad_norm": 0.02434307336807251,
+      "learning_rate": 6.517950505402579e-06,
+      "loss": 0.0985,
+      "step": 9670
+    },
+    {
+      "epoch": 3.3739979086789824,
+      "grad_norm": 80.342529296875,
+      "learning_rate": 6.504008365284072e-06,
+      "loss": 0.0361,
+      "step": 9680
+    },
+    {
+      "epoch": 3.377483443708609,
+      "grad_norm": 0.01803007163107395,
+      "learning_rate": 6.490066225165563e-06,
+      "loss": 0.07,
+      "step": 9690
+    },
+    {
+      "epoch": 3.3809689787382364,
+      "grad_norm": 0.06161894649267197,
+      "learning_rate": 6.476124085047055e-06,
+      "loss": 0.1747,
+      "step": 9700
+    },
+    {
+      "epoch": 3.384454513767863,
+      "grad_norm": 0.02253262884914875,
+      "learning_rate": 6.4621819449285475e-06,
+      "loss": 0.0444,
+      "step": 9710
+    },
+    {
+      "epoch": 3.3879400487974904,
+      "grad_norm": 0.027096986770629883,
+      "learning_rate": 6.448239804810039e-06,
+      "loss": 0.1927,
+      "step": 9720
+    },
+    {
+      "epoch": 3.3914255838271172,
+      "grad_norm": 0.42232781648635864,
+      "learning_rate": 6.434297664691531e-06,
+      "loss": 0.1841,
+      "step": 9730
+    },
+    {
+      "epoch": 3.3949111188567445,
+      "grad_norm": 0.01901787333190441,
+      "learning_rate": 6.4203555245730224e-06,
+      "loss": 0.135,
+      "step": 9740
+    },
+    {
+      "epoch": 3.3983966538863717,
+      "grad_norm": 4.167501926422119,
+      "learning_rate": 6.406413384454514e-06,
+      "loss": 0.0409,
+      "step": 9750
+    },
+    {
+      "epoch": 3.4018821889159985,
+      "grad_norm": 17.263132095336914,
+      "learning_rate": 6.392471244336006e-06,
+      "loss": 0.1366,
+      "step": 9760
+    },
+    {
+      "epoch": 3.4053677239456257,
+      "grad_norm": 4.265865325927734,
+      "learning_rate": 6.378529104217498e-06,
+      "loss": 0.1339,
+      "step": 9770
+    },
+    {
+      "epoch": 3.4088532589752525,
+      "grad_norm": 0.022587908431887627,
+      "learning_rate": 6.36458696409899e-06,
+      "loss": 0.1715,
+      "step": 9780
+    },
+    {
+      "epoch": 3.4123387940048797,
+      "grad_norm": 15.038971900939941,
+      "learning_rate": 6.350644823980482e-06,
+      "loss": 0.1432,
+      "step": 9790
+    },
+    {
+      "epoch": 3.415824329034507,
+      "grad_norm": 0.056241635233163834,
+      "learning_rate": 6.336702683861974e-06,
+      "loss": 0.093,
+      "step": 9800
+    },
+    {
+      "epoch": 3.4193098640641337,
+      "grad_norm": 0.08624168485403061,
+      "learning_rate": 6.322760543743465e-06,
+      "loss": 0.139,
+      "step": 9810
+    },
+    {
+      "epoch": 3.422795399093761,
+      "grad_norm": 0.025193965062499046,
+      "learning_rate": 6.308818403624957e-06,
+      "loss": 0.0069,
+      "step": 9820
+    },
+    {
+      "epoch": 3.4262809341233877,
+      "grad_norm": 12.65317440032959,
+      "learning_rate": 6.294876263506448e-06,
+      "loss": 0.1008,
+      "step": 9830
+    },
+    {
+      "epoch": 3.429766469153015,
+      "grad_norm": 3.4288766384124756,
+      "learning_rate": 6.280934123387941e-06,
+      "loss": 0.2026,
+      "step": 9840
+    },
+    {
+      "epoch": 3.433252004182642,
+      "grad_norm": 0.01608997769653797,
+      "learning_rate": 6.266991983269432e-06,
+      "loss": 0.0358,
+      "step": 9850
+    },
+    {
+      "epoch": 3.436737539212269,
+      "grad_norm": 9.135321617126465,
+      "learning_rate": 6.253049843150925e-06,
+      "loss": 0.1489,
+      "step": 9860
+    },
+    {
+      "epoch": 3.440223074241896,
+      "grad_norm": 0.02817300148308277,
+      "learning_rate": 6.239107703032417e-06,
+      "loss": 0.0225,
+      "step": 9870
+    },
+    {
+      "epoch": 3.443708609271523,
+      "grad_norm": 0.03426109254360199,
+      "learning_rate": 6.225165562913907e-06,
+      "loss": 0.0896,
+      "step": 9880
+    },
+    {
+      "epoch": 3.4471941443011502,
+      "grad_norm": 0.029791146516799927,
+      "learning_rate": 6.211223422795399e-06,
+      "loss": 0.0733,
+      "step": 9890
+    },
+    {
+      "epoch": 3.4506796793307775,
+      "grad_norm": 5.195836067199707,
+      "learning_rate": 6.1972812826768916e-06,
+      "loss": 0.0285,
+      "step": 9900
+    },
+    {
+      "epoch": 3.4541652143604042,
+      "grad_norm": 0.046240124851465225,
+      "learning_rate": 6.183339142558383e-06,
+      "loss": 0.2035,
+      "step": 9910
+    },
+    {
+      "epoch": 3.4576507493900315,
+      "grad_norm": 14.654541969299316,
+      "learning_rate": 6.169397002439875e-06,
+      "loss": 0.115,
+      "step": 9920
+    },
+    {
+      "epoch": 3.4611362844196583,
+      "grad_norm": 0.05055411905050278,
+      "learning_rate": 6.155454862321367e-06,
+      "loss": 0.0856,
+      "step": 9930
+    },
+    {
+      "epoch": 3.4646218194492855,
+      "grad_norm": 0.04881644248962402,
+      "learning_rate": 6.141512722202858e-06,
+      "loss": 0.1484,
+      "step": 9940
+    },
+    {
+      "epoch": 3.4681073544789127,
+      "grad_norm": 0.08495481312274933,
+      "learning_rate": 6.12757058208435e-06,
+      "loss": 0.0758,
+      "step": 9950
+    },
+    {
+      "epoch": 3.4715928895085395,
+      "grad_norm": 0.028945187106728554,
+      "learning_rate": 6.113628441965842e-06,
+      "loss": 0.0965,
+      "step": 9960
+    },
+    {
+      "epoch": 3.4750784245381667,
+      "grad_norm": 8.535677909851074,
+      "learning_rate": 6.099686301847334e-06,
+      "loss": 0.2292,
+      "step": 9970
+    },
+    {
+      "epoch": 3.4785639595677935,
+      "grad_norm": 7.189770221710205,
+      "learning_rate": 6.085744161728826e-06,
+      "loss": 0.155,
+      "step": 9980
+    },
+    {
+      "epoch": 3.4820494945974207,
+      "grad_norm": 0.022676438093185425,
+      "learning_rate": 6.071802021610318e-06,
+      "loss": 0.1451,
+      "step": 9990
+    },
+    {
+      "epoch": 3.485535029627048,
+      "grad_norm": 37.57761764526367,
+      "learning_rate": 6.05785988149181e-06,
+      "loss": 0.1904,
+      "step": 10000
+    },
+    {
+      "epoch": 3.4890205646566748,
+      "grad_norm": 0.05265439301729202,
+      "learning_rate": 6.043917741373301e-06,
+      "loss": 0.0894,
+      "step": 10010
+    },
+    {
+      "epoch": 3.492506099686302,
+      "grad_norm": 0.08444428443908691,
+      "learning_rate": 6.029975601254792e-06,
+      "loss": 0.1031,
+      "step": 10020
+    },
+    {
+      "epoch": 3.4959916347159288,
+      "grad_norm": 0.3577669858932495,
+      "learning_rate": 6.016033461136285e-06,
+      "loss": 0.0539,
+      "step": 10030
+    },
+    {
+      "epoch": 3.499477169745556,
+      "grad_norm": 0.026337046176195145,
+      "learning_rate": 6.0020913210177765e-06,
+      "loss": 0.1267,
+      "step": 10040
+    },
+    {
+      "epoch": 3.5029627047751832,
+      "grad_norm": 4.85567569732666,
+      "learning_rate": 5.988149180899269e-06,
+      "loss": 0.0094,
+      "step": 10050
+    },
+    {
+      "epoch": 3.50644823980481,
+      "grad_norm": 1.0068359375,
+      "learning_rate": 5.974207040780761e-06,
+      "loss": 0.1011,
+      "step": 10060
+    },
+    {
+      "epoch": 3.5099337748344372,
+      "grad_norm": 0.03198888525366783,
+      "learning_rate": 5.960264900662252e-06,
+      "loss": 0.1203,
+      "step": 10070
+    },
+    {
+      "epoch": 3.513419309864064,
+      "grad_norm": 0.08028540015220642,
+      "learning_rate": 5.946322760543743e-06,
+      "loss": 0.073,
+      "step": 10080
+    },
+    {
+      "epoch": 3.5169048448936913,
+      "grad_norm": 6.395030975341797,
+      "learning_rate": 5.932380620425236e-06,
+      "loss": 0.1022,
+      "step": 10090
+    },
+    {
+      "epoch": 3.5203903799233185,
+      "grad_norm": 0.9497456550598145,
+      "learning_rate": 5.918438480306727e-06,
+      "loss": 0.133,
+      "step": 10100
+    },
+    {
+      "epoch": 3.5238759149529453,
+      "grad_norm": 0.02326410636305809,
+      "learning_rate": 5.904496340188219e-06,
+      "loss": 0.218,
+      "step": 10110
+    },
+    {
+      "epoch": 3.527361449982572,
+      "grad_norm": 0.12932069599628448,
+      "learning_rate": 5.8905542000697115e-06,
+      "loss": 0.0913,
+      "step": 10120
+    },
+    {
+      "epoch": 3.5308469850121993,
+      "grad_norm": 0.023471644148230553,
+      "learning_rate": 5.876612059951203e-06,
+      "loss": 0.1593,
+      "step": 10130
+    },
+    {
+      "epoch": 3.5343325200418265,
+      "grad_norm": 11.825116157531738,
+      "learning_rate": 5.862669919832696e-06,
+      "loss": 0.0478,
+      "step": 10140
+    },
+    {
+      "epoch": 3.5378180550714533,
+      "grad_norm": 4.660243034362793,
+      "learning_rate": 5.8487277797141865e-06,
+      "loss": 0.0862,
+      "step": 10150
+    },
+    {
+      "epoch": 3.5413035901010805,
+      "grad_norm": 0.017631346359848976,
+      "learning_rate": 5.834785639595678e-06,
+      "loss": 0.072,
+      "step": 10160
+    },
+    {
+      "epoch": 3.5447891251307073,
+      "grad_norm": 0.02127777598798275,
+      "learning_rate": 5.82084349947717e-06,
+      "loss": 0.0143,
+      "step": 10170
+    },
+    {
+      "epoch": 3.5482746601603345,
+      "grad_norm": 12.403838157653809,
+      "learning_rate": 5.806901359358662e-06,
+      "loss": 0.194,
+      "step": 10180
+    },
+    {
+      "epoch": 3.5517601951899618,
+      "grad_norm": 0.01695849932730198,
+      "learning_rate": 5.792959219240154e-06,
+      "loss": 0.0156,
+      "step": 10190
+    },
+    {
+      "epoch": 3.5552457302195886,
+      "grad_norm": 0.06215062364935875,
+      "learning_rate": 5.779017079121646e-06,
+      "loss": 0.3112,
+      "step": 10200
+    },
+    {
+      "epoch": 3.558731265249216,
+      "grad_norm": 0.015378969721496105,
+      "learning_rate": 5.765074939003138e-06,
+      "loss": 0.0328,
+      "step": 10210
+    },
+    {
+      "epoch": 3.5622168002788426,
+      "grad_norm": 0.5997032523155212,
+      "learning_rate": 5.751132798884629e-06,
+      "loss": 0.0973,
+      "step": 10220
+    },
+    {
+      "epoch": 3.56570233530847,
+      "grad_norm": 0.17242105305194855,
+      "learning_rate": 5.737190658766121e-06,
+      "loss": 0.0697,
+      "step": 10230
+    },
+    {
+      "epoch": 3.569187870338097,
+      "grad_norm": 0.028457053005695343,
+      "learning_rate": 5.723248518647613e-06,
+      "loss": 0.2034,
+      "step": 10240
+    },
+    {
+      "epoch": 3.572673405367724,
+      "grad_norm": 51.637733459472656,
+      "learning_rate": 5.709306378529105e-06,
+      "loss": 0.3527,
+      "step": 10250
+    },
+    {
+      "epoch": 3.576158940397351,
+      "grad_norm": 0.02154484950006008,
+      "learning_rate": 5.6953642384105965e-06,
+      "loss": 0.0558,
+      "step": 10260
+    },
+    {
+      "epoch": 3.579644475426978,
+      "grad_norm": 0.02288723737001419,
+      "learning_rate": 5.681422098292089e-06,
+      "loss": 0.1087,
+      "step": 10270
+    },
+    {
+      "epoch": 3.583130010456605,
+      "grad_norm": 0.02297695353627205,
+      "learning_rate": 5.667479958173581e-06,
+      "loss": 0.0692,
+      "step": 10280
+    },
+    {
+      "epoch": 3.5866155454862323,
+      "grad_norm": 0.0711924359202385,
+      "learning_rate": 5.6535378180550715e-06,
+      "loss": 0.065,
+      "step": 10290
+    },
+    {
+      "epoch": 3.590101080515859,
+      "grad_norm": 17.65730857849121,
+      "learning_rate": 5.639595677936563e-06,
+      "loss": 0.2072,
+      "step": 10300
+    },
+    {
+      "epoch": 3.5935866155454863,
+      "grad_norm": 6.287166118621826,
+      "learning_rate": 5.625653537818056e-06,
+      "loss": 0.3501,
+      "step": 10310
+    },
+    {
+      "epoch": 3.597072150575113,
+      "grad_norm": 0.05497564375400543,
+      "learning_rate": 5.611711397699547e-06,
+      "loss": 0.1217,
+      "step": 10320
+    },
+    {
+      "epoch": 3.6005576856047403,
+      "grad_norm": 0.2877632677555084,
+      "learning_rate": 5.59776925758104e-06,
+      "loss": 0.1098,
+      "step": 10330
+    },
+    {
+      "epoch": 3.6040432206343676,
+      "grad_norm": 0.033131808042526245,
+      "learning_rate": 5.5838271174625315e-06,
+      "loss": 0.0095,
+      "step": 10340
+    },
+    {
+      "epoch": 3.6075287556639943,
+      "grad_norm": 0.057142358273267746,
+      "learning_rate": 5.569884977344022e-06,
+      "loss": 0.1609,
+      "step": 10350
+    },
+    {
+      "epoch": 3.6110142906936216,
+      "grad_norm": 3.4275271892547607,
+      "learning_rate": 5.555942837225514e-06,
+      "loss": 0.1044,
+      "step": 10360
+    },
+    {
+      "epoch": 3.6144998257232483,
+      "grad_norm": 0.03326639533042908,
+      "learning_rate": 5.5420006971070064e-06,
+      "loss": 0.0441,
+      "step": 10370
+    },
+    {
+      "epoch": 3.6179853607528756,
+      "grad_norm": 0.018506675958633423,
+      "learning_rate": 5.528058556988498e-06,
+      "loss": 0.1586,
+      "step": 10380
+    },
+    {
+      "epoch": 3.621470895782503,
+      "grad_norm": 0.05169185996055603,
+      "learning_rate": 5.51411641686999e-06,
+      "loss": 0.0747,
+      "step": 10390
+    },
+    {
+      "epoch": 3.6249564308121296,
+      "grad_norm": 0.017710169777274132,
+      "learning_rate": 5.500174276751482e-06,
+      "loss": 0.0171,
+      "step": 10400
+    },
+    {
+      "epoch": 3.628441965841757,
+      "grad_norm": 20.366230010986328,
+      "learning_rate": 5.486232136632974e-06,
+      "loss": 0.4428,
+      "step": 10410
+    },
+    {
+      "epoch": 3.6319275008713836,
+      "grad_norm": 0.12488202750682831,
+      "learning_rate": 5.472289996514465e-06,
+      "loss": 0.2049,
+      "step": 10420
+    },
+    {
+      "epoch": 3.635413035901011,
+      "grad_norm": 0.027057647705078125,
+      "learning_rate": 5.458347856395957e-06,
+      "loss": 0.0281,
+      "step": 10430
+    },
+    {
+      "epoch": 3.638898570930638,
+      "grad_norm": 0.2535144090652466,
+      "learning_rate": 5.444405716277449e-06,
+      "loss": 0.009,
+      "step": 10440
+    },
+    {
+      "epoch": 3.642384105960265,
+      "grad_norm": 5.463354110717773,
+      "learning_rate": 5.430463576158941e-06,
+      "loss": 0.131,
+      "step": 10450
+    },
+    {
+      "epoch": 3.645869640989892,
+      "grad_norm": 0.021357407793402672,
+      "learning_rate": 5.416521436040433e-06,
+      "loss": 0.407,
+      "step": 10460
+    },
+    {
+      "epoch": 3.649355176019519,
+      "grad_norm": 30.37545394897461,
+      "learning_rate": 5.402579295921925e-06,
+      "loss": 0.231,
+      "step": 10470
+    },
+    {
+      "epoch": 3.652840711049146,
+      "grad_norm": 0.027980081737041473,
+      "learning_rate": 5.388637155803416e-06,
+      "loss": 0.2154,
+      "step": 10480
+    },
+    {
+      "epoch": 3.6563262460787733,
+      "grad_norm": 0.03117654100060463,
+      "learning_rate": 5.374695015684907e-06,
+      "loss": 0.1123,
+      "step": 10490
+    },
+    {
+      "epoch": 3.6598117811084,
+      "grad_norm": 0.9009264707565308,
+      "learning_rate": 5.3607528755664e-06,
+      "loss": 0.1333,
+      "step": 10500
+    },
+    {
+      "epoch": 3.6632973161380273,
+      "grad_norm": 5.609897613525391,
+      "learning_rate": 5.346810735447891e-06,
+      "loss": 0.1772,
+      "step": 10510
+    },
+    {
+      "epoch": 3.666782851167654,
+      "grad_norm": 0.05884939059615135,
+      "learning_rate": 5.332868595329384e-06,
+      "loss": 0.0984,
+      "step": 10520
+    },
+    {
+      "epoch": 3.6702683861972814,
+      "grad_norm": 1.9585541486740112,
+      "learning_rate": 5.318926455210876e-06,
+      "loss": 0.0611,
+      "step": 10530
+    },
+    {
+      "epoch": 3.6737539212269086,
+      "grad_norm": 1.0173242092132568,
+      "learning_rate": 5.304984315092367e-06,
+      "loss": 0.2067,
+      "step": 10540
+    },
+    {
+      "epoch": 3.6772394562565354,
+      "grad_norm": 0.15780754387378693,
+      "learning_rate": 5.29104217497386e-06,
+      "loss": 0.1282,
+      "step": 10550
+    },
+    {
+      "epoch": 3.680724991286162,
+      "grad_norm": 27.38100814819336,
+      "learning_rate": 5.2771000348553506e-06,
+      "loss": 0.2614,
+      "step": 10560
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.04858740046620369,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.197,
+      "step": 10570
+    },
+    {
+      "epoch": 3.6876960613454166,
+      "grad_norm": 0.01936270296573639,
+      "learning_rate": 5.249215754618334e-06,
+      "loss": 0.1291,
+      "step": 10580
+    },
+    {
+      "epoch": 3.691181596375044,
+      "grad_norm": 0.02714901603758335,
+      "learning_rate": 5.235273614499826e-06,
+      "loss": 0.1723,
+      "step": 10590
+    },
+    {
+      "epoch": 3.6946671314046706,
+      "grad_norm": 0.02381339855492115,
+      "learning_rate": 5.221331474381318e-06,
+      "loss": 0.0174,
+      "step": 10600
+    },
+    {
+      "epoch": 3.6981526664342974,
+      "grad_norm": 0.06992737948894501,
+      "learning_rate": 5.20738933426281e-06,
+      "loss": 0.0499,
+      "step": 10610
+    },
+    {
+      "epoch": 3.7016382014639246,
+      "grad_norm": 0.4815942943096161,
+      "learning_rate": 5.193447194144302e-06,
+      "loss": 0.0657,
+      "step": 10620
+    },
+    {
+      "epoch": 3.705123736493552,
+      "grad_norm": 4.941831588745117,
+      "learning_rate": 5.179505054025793e-06,
+      "loss": 0.1945,
+      "step": 10630
+    },
+    {
+      "epoch": 3.7086092715231787,
+      "grad_norm": 4.8328166007995605,
+      "learning_rate": 5.165562913907285e-06,
+      "loss": 0.1758,
+      "step": 10640
+    },
+    {
+      "epoch": 3.712094806552806,
+      "grad_norm": 0.24197635054588318,
+      "learning_rate": 5.151620773788777e-06,
+      "loss": 0.1795,
+      "step": 10650
+    },
+    {
+      "epoch": 3.7155803415824327,
+      "grad_norm": 22.532445907592773,
+      "learning_rate": 5.137678633670269e-06,
+      "loss": 0.1157,
+      "step": 10660
+    },
+    {
+      "epoch": 3.71906587661206,
+      "grad_norm": 0.19902735948562622,
+      "learning_rate": 5.1237364935517605e-06,
+      "loss": 0.0064,
+      "step": 10670
+    },
+    {
+      "epoch": 3.722551411641687,
+      "grad_norm": 12.738110542297363,
+      "learning_rate": 5.109794353433253e-06,
+      "loss": 0.0219,
+      "step": 10680
+    },
+    {
+      "epoch": 3.726036946671314,
+      "grad_norm": 0.07453146576881409,
+      "learning_rate": 5.095852213314745e-06,
+      "loss": 0.1586,
+      "step": 10690
+    },
+    {
+      "epoch": 3.729522481700941,
+      "grad_norm": 0.022043762728571892,
+      "learning_rate": 5.0819100731962355e-06,
+      "loss": 0.0053,
+      "step": 10700
+    },
+    {
+      "epoch": 3.733008016730568,
+      "grad_norm": 15.734837532043457,
+      "learning_rate": 5.067967933077728e-06,
+      "loss": 0.2529,
+      "step": 10710
+    },
+    {
+      "epoch": 3.736493551760195,
+      "grad_norm": 0.5030373930931091,
+      "learning_rate": 5.05402579295922e-06,
+      "loss": 0.0513,
+      "step": 10720
+    },
+    {
+      "epoch": 3.7399790867898224,
+      "grad_norm": 0.10042136162519455,
+      "learning_rate": 5.040083652840711e-06,
+      "loss": 0.2097,
+      "step": 10730
+    },
+    {
+      "epoch": 3.743464621819449,
+      "grad_norm": 0.4963725507259369,
+      "learning_rate": 5.026141512722204e-06,
+      "loss": 0.0957,
+      "step": 10740
+    },
+    {
+      "epoch": 3.7469501568490764,
+      "grad_norm": 8.767095565795898,
+      "learning_rate": 5.0121993726036955e-06,
+      "loss": 0.2483,
+      "step": 10750
+    },
+    {
+      "epoch": 3.750435691878703,
+      "grad_norm": 12.400495529174805,
+      "learning_rate": 4.998257232485187e-06,
+      "loss": 0.1802,
+      "step": 10760
+    },
+    {
+      "epoch": 3.7539212269083304,
+      "grad_norm": 5.502656936645508,
+      "learning_rate": 4.984315092366679e-06,
+      "loss": 0.1525,
+      "step": 10770
+    },
+    {
+      "epoch": 3.7574067619379576,
+      "grad_norm": 0.01613186113536358,
+      "learning_rate": 4.9703729522481705e-06,
+      "loss": 0.1123,
+      "step": 10780
+    },
+    {
+      "epoch": 3.7608922969675844,
+      "grad_norm": 0.01936766505241394,
+      "learning_rate": 4.956430812129662e-06,
+      "loss": 0.0666,
+      "step": 10790
+    },
+    {
+      "epoch": 3.7643778319972117,
+      "grad_norm": 16.322816848754883,
+      "learning_rate": 4.942488672011154e-06,
+      "loss": 0.0653,
+      "step": 10800
+    },
+    {
+      "epoch": 3.7678633670268384,
+      "grad_norm": 17.28783416748047,
+      "learning_rate": 4.9285465318926455e-06,
+      "loss": 0.0617,
+      "step": 10810
+    },
+    {
+      "epoch": 3.7713489020564657,
+      "grad_norm": 0.042436350136995316,
+      "learning_rate": 4.914604391774138e-06,
+      "loss": 0.152,
+      "step": 10820
+    },
+    {
+      "epoch": 3.774834437086093,
+      "grad_norm": 0.20540641248226166,
+      "learning_rate": 4.90066225165563e-06,
+      "loss": 0.0412,
+      "step": 10830
+    },
+    {
+      "epoch": 3.7783199721157197,
+      "grad_norm": 0.211298406124115,
+      "learning_rate": 4.886720111537121e-06,
+      "loss": 0.0181,
+      "step": 10840
+    },
+    {
+      "epoch": 3.781805507145347,
+      "grad_norm": 0.09502696245908737,
+      "learning_rate": 4.872777971418613e-06,
+      "loss": 0.0132,
+      "step": 10850
+    },
+    {
+      "epoch": 3.7852910421749737,
+      "grad_norm": 0.03848971426486969,
+      "learning_rate": 4.858835831300105e-06,
+      "loss": 0.0608,
+      "step": 10860
+    },
+    {
+      "epoch": 3.788776577204601,
+      "grad_norm": 21.687044143676758,
+      "learning_rate": 4.844893691181597e-06,
+      "loss": 0.1938,
+      "step": 10870
+    },
+    {
+      "epoch": 3.792262112234228,
+      "grad_norm": 0.02101794257760048,
+      "learning_rate": 4.830951551063088e-06,
+      "loss": 0.0113,
+      "step": 10880
+    },
+    {
+      "epoch": 3.795747647263855,
+      "grad_norm": 0.036022286862134933,
+      "learning_rate": 4.8170094109445805e-06,
+      "loss": 0.164,
+      "step": 10890
+    },
+    {
+      "epoch": 3.799233182293482,
+      "grad_norm": 0.035032596439123154,
+      "learning_rate": 4.803067270826072e-06,
+      "loss": 0.0974,
+      "step": 10900
+    },
+    {
+      "epoch": 3.802718717323109,
+      "grad_norm": 8.165773391723633,
+      "learning_rate": 4.789125130707565e-06,
+      "loss": 0.0294,
+      "step": 10910
+    },
+    {
+      "epoch": 3.806204252352736,
+      "grad_norm": 0.01464312057942152,
+      "learning_rate": 4.7751829905890555e-06,
+      "loss": 0.1462,
+      "step": 10920
+    },
+    {
+      "epoch": 3.8096897873823634,
+      "grad_norm": 0.02338375523686409,
+      "learning_rate": 4.761240850470548e-06,
+      "loss": 0.1984,
+      "step": 10930
+    },
+    {
+      "epoch": 3.81317532241199,
+      "grad_norm": 0.021365277469158173,
+      "learning_rate": 4.74729871035204e-06,
+      "loss": 0.1061,
+      "step": 10940
+    },
+    {
+      "epoch": 3.8166608574416174,
+      "grad_norm": 0.01990444026887417,
+      "learning_rate": 4.733356570233531e-06,
+      "loss": 0.0306,
+      "step": 10950
+    },
+    {
+      "epoch": 3.820146392471244,
+      "grad_norm": 0.664445161819458,
+      "learning_rate": 4.719414430115023e-06,
+      "loss": 0.0078,
+      "step": 10960
+    },
+    {
+      "epoch": 3.8236319275008714,
+      "grad_norm": 2.076838254928589,
+      "learning_rate": 4.705472289996515e-06,
+      "loss": 0.0971,
+      "step": 10970
+    },
+    {
+      "epoch": 3.8271174625304987,
+      "grad_norm": 0.01907547004520893,
+      "learning_rate": 4.691530149878007e-06,
+      "loss": 0.0404,
+      "step": 10980
+    },
+    {
+      "epoch": 3.8306029975601255,
+      "grad_norm": 0.4929560720920563,
+      "learning_rate": 4.677588009759498e-06,
+      "loss": 0.0403,
+      "step": 10990
+    },
+    {
+      "epoch": 3.8340885325897527,
+      "grad_norm": 3.4360711574554443,
+      "learning_rate": 4.6636458696409905e-06,
+      "loss": 0.2111,
+      "step": 11000
+    },
+    {
+      "epoch": 3.8375740676193795,
+      "grad_norm": 3.6092143058776855,
+      "learning_rate": 4.649703729522482e-06,
+      "loss": 0.2806,
+      "step": 11010
+    },
+    {
+      "epoch": 3.8410596026490067,
+      "grad_norm": 0.01587533950805664,
+      "learning_rate": 4.635761589403974e-06,
+      "loss": 0.0803,
+      "step": 11020
+    },
+    {
+      "epoch": 3.844545137678634,
+      "grad_norm": 17.90401840209961,
+      "learning_rate": 4.6218194492854654e-06,
+      "loss": 0.1311,
+      "step": 11030
+    },
+    {
+      "epoch": 3.8480306727082607,
+      "grad_norm": 0.02171589806675911,
+      "learning_rate": 4.607877309166958e-06,
+      "loss": 0.0515,
+      "step": 11040
+    },
+    {
+      "epoch": 3.8515162077378875,
+      "grad_norm": 0.36789968609809875,
+      "learning_rate": 4.593935169048449e-06,
+      "loss": 0.1818,
+      "step": 11050
+    },
+    {
+      "epoch": 3.8550017427675147,
+      "grad_norm": 0.017076525837183,
+      "learning_rate": 4.579993028929941e-06,
+      "loss": 0.0441,
+      "step": 11060
+    },
+    {
+      "epoch": 3.858487277797142,
+      "grad_norm": 0.03205644711852074,
+      "learning_rate": 4.566050888811433e-06,
+      "loss": 0.0985,
+      "step": 11070
+    },
+    {
+      "epoch": 3.861972812826769,
+      "grad_norm": 0.031755849719047546,
+      "learning_rate": 4.552108748692925e-06,
+      "loss": 0.1107,
+      "step": 11080
+    },
+    {
+      "epoch": 3.865458347856396,
+      "grad_norm": 0.020482953637838364,
+      "learning_rate": 4.538166608574416e-06,
+      "loss": 0.1496,
+      "step": 11090
+    },
+    {
+      "epoch": 3.8689438828860228,
+      "grad_norm": 0.047831941395998,
+      "learning_rate": 4.524224468455909e-06,
+      "loss": 0.1769,
+      "step": 11100
+    },
+    {
+      "epoch": 3.87242941791565,
+      "grad_norm": 0.02326442301273346,
+      "learning_rate": 4.5102823283374004e-06,
+      "loss": 0.1952,
+      "step": 11110
+    },
+    {
+      "epoch": 3.875914952945277,
+      "grad_norm": 0.2660370469093323,
+      "learning_rate": 4.496340188218892e-06,
+      "loss": 0.0117,
+      "step": 11120
+    },
+    {
+      "epoch": 3.879400487974904,
+      "grad_norm": 0.023432690650224686,
+      "learning_rate": 4.482398048100384e-06,
+      "loss": 0.1016,
+      "step": 11130
+    },
+    {
+      "epoch": 3.8828860230045312,
+      "grad_norm": 11.599336624145508,
+      "learning_rate": 4.468455907981875e-06,
+      "loss": 0.0267,
+      "step": 11140
+    },
+    {
+      "epoch": 3.886371558034158,
+      "grad_norm": 0.030066780745983124,
+      "learning_rate": 4.454513767863368e-06,
+      "loss": 0.0146,
+      "step": 11150
+    },
+    {
+      "epoch": 3.8898570930637852,
+      "grad_norm": 0.30465713143348694,
+      "learning_rate": 4.440571627744859e-06,
+      "loss": 0.0186,
+      "step": 11160
+    },
+    {
+      "epoch": 3.8933426280934125,
+      "grad_norm": 0.04410243034362793,
+      "learning_rate": 4.426629487626351e-06,
+      "loss": 0.0072,
+      "step": 11170
+    },
+    {
+      "epoch": 3.8968281631230393,
+      "grad_norm": 39.044795989990234,
+      "learning_rate": 4.412687347507843e-06,
+      "loss": 0.1416,
+      "step": 11180
+    },
+    {
+      "epoch": 3.9003136981526665,
+      "grad_norm": 28.497705459594727,
+      "learning_rate": 4.3987452073893346e-06,
+      "loss": 0.0732,
+      "step": 11190
+    },
+    {
+      "epoch": 3.9037992331822933,
+      "grad_norm": 0.019765986129641533,
+      "learning_rate": 4.384803067270826e-06,
+      "loss": 0.0791,
+      "step": 11200
+    },
+    {
+      "epoch": 3.9072847682119205,
+      "grad_norm": 0.32468003034591675,
+      "learning_rate": 4.370860927152319e-06,
+      "loss": 0.1067,
+      "step": 11210
+    },
+    {
+      "epoch": 3.9107703032415477,
+      "grad_norm": 0.02749641239643097,
+      "learning_rate": 4.3569187870338096e-06,
+      "loss": 0.171,
+      "step": 11220
+    },
+    {
+      "epoch": 3.9142558382711745,
+      "grad_norm": 12.454182624816895,
+      "learning_rate": 4.342976646915302e-06,
+      "loss": 0.2891,
+      "step": 11230
+    },
+    {
+      "epoch": 3.9177413733008017,
+      "grad_norm": 0.03475377708673477,
+      "learning_rate": 4.329034506796794e-06,
+      "loss": 0.2393,
+      "step": 11240
+    },
+    {
+      "epoch": 3.9212269083304285,
+      "grad_norm": 0.01971781998872757,
+      "learning_rate": 4.315092366678285e-06,
+      "loss": 0.1189,
+      "step": 11250
+    },
+    {
+      "epoch": 3.9247124433600558,
+      "grad_norm": 0.7159892916679382,
+      "learning_rate": 4.301150226559777e-06,
+      "loss": 0.1387,
+      "step": 11260
+    },
+    {
+      "epoch": 3.928197978389683,
+      "grad_norm": 10.798375129699707,
+      "learning_rate": 4.287208086441269e-06,
+      "loss": 0.2423,
+      "step": 11270
+    },
+    {
+      "epoch": 3.9316835134193098,
+      "grad_norm": 1.2279161214828491,
+      "learning_rate": 4.273265946322761e-06,
+      "loss": 0.0822,
+      "step": 11280
+    },
+    {
+      "epoch": 3.935169048448937,
+      "grad_norm": 0.023021556437015533,
+      "learning_rate": 4.259323806204252e-06,
+      "loss": 0.1275,
+      "step": 11290
+    },
+    {
+      "epoch": 3.938654583478564,
+      "grad_norm": 0.041272278875112534,
+      "learning_rate": 4.2453816660857445e-06,
+      "loss": 0.0148,
+      "step": 11300
+    },
+    {
+      "epoch": 3.942140118508191,
+      "grad_norm": 3.4936373233795166,
+      "learning_rate": 4.231439525967236e-06,
+      "loss": 0.1178,
+      "step": 11310
+    },
+    {
+      "epoch": 3.9456256535378182,
+      "grad_norm": 0.033816587179899216,
+      "learning_rate": 4.217497385848729e-06,
+      "loss": 0.2568,
+      "step": 11320
+    },
+    {
+      "epoch": 3.949111188567445,
+      "grad_norm": 0.018212752416729927,
+      "learning_rate": 4.2035552457302195e-06,
+      "loss": 0.0131,
+      "step": 11330
+    },
+    {
+      "epoch": 3.9525967235970723,
+      "grad_norm": 10.551046371459961,
+      "learning_rate": 4.189613105611712e-06,
+      "loss": 0.1611,
+      "step": 11340
+    },
+    {
+      "epoch": 3.956082258626699,
+      "grad_norm": 0.018220404163002968,
+      "learning_rate": 4.175670965493204e-06,
+      "loss": 0.223,
+      "step": 11350
+    },
+    {
+      "epoch": 3.9595677936563263,
+      "grad_norm": 0.02707715705037117,
+      "learning_rate": 4.161728825374695e-06,
+      "loss": 0.0611,
+      "step": 11360
+    },
+    {
+      "epoch": 3.9630533286859535,
+      "grad_norm": 6.247076511383057,
+      "learning_rate": 4.147786685256187e-06,
+      "loss": 0.2263,
+      "step": 11370
+    },
+    {
+      "epoch": 3.9665388637155803,
+      "grad_norm": 0.013209417462348938,
+      "learning_rate": 4.133844545137679e-06,
+      "loss": 0.0905,
+      "step": 11380
+    },
+    {
+      "epoch": 3.9700243987452075,
+      "grad_norm": 0.19896091520786285,
+      "learning_rate": 4.119902405019171e-06,
+      "loss": 0.1132,
+      "step": 11390
+    },
+    {
+      "epoch": 3.9735099337748343,
+      "grad_norm": 0.01759384572505951,
+      "learning_rate": 4.105960264900663e-06,
+      "loss": 0.0424,
+      "step": 11400
+    },
+    {
+      "epoch": 3.9769954688044615,
+      "grad_norm": 0.018540671095252037,
+      "learning_rate": 4.0920181247821545e-06,
+      "loss": 0.095,
+      "step": 11410
+    },
+    {
+      "epoch": 3.9804810038340888,
+      "grad_norm": 0.018298756331205368,
+      "learning_rate": 4.078075984663646e-06,
+      "loss": 0.0239,
+      "step": 11420
+    },
+    {
+      "epoch": 3.9839665388637155,
+      "grad_norm": 0.10425955057144165,
+      "learning_rate": 4.064133844545138e-06,
+      "loss": 0.0831,
+      "step": 11430
+    },
+    {
+      "epoch": 3.9874520738933428,
+      "grad_norm": 0.29570281505584717,
+      "learning_rate": 4.0501917044266295e-06,
+      "loss": 0.0078,
+      "step": 11440
+    },
+    {
+      "epoch": 3.9909376089229696,
+      "grad_norm": 10.413164138793945,
+      "learning_rate": 4.036249564308122e-06,
+      "loss": 0.0978,
+      "step": 11450
+    },
+    {
+      "epoch": 3.994423143952597,
+      "grad_norm": 0.021612277254462242,
+      "learning_rate": 4.022307424189613e-06,
+      "loss": 0.0971,
+      "step": 11460
+    },
+    {
+      "epoch": 3.997908678982224,
+      "grad_norm": 2.315714120864868,
+      "learning_rate": 4.008365284071105e-06,
+      "loss": 0.0977,
+      "step": 11470
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9906172839506173,
+      "eval_loss": 0.04471336677670479,
+      "eval_runtime": 18.8269,
+      "eval_samples_per_second": 215.118,
+      "eval_steps_per_second": 26.93,
+      "step": 11476
+    },
+    {
+      "epoch": 4.001394214011851,
+      "grad_norm": 0.014861468225717545,
+      "learning_rate": 3.994423143952597e-06,
+      "loss": 0.0664,
+      "step": 11480
+    },
+    {
+      "epoch": 4.004879749041478,
+      "grad_norm": 0.18193350732326508,
+      "learning_rate": 3.9804810038340895e-06,
+      "loss": 0.1891,
+      "step": 11490
+    },
+    {
+      "epoch": 4.008365284071105,
+      "grad_norm": 0.04888654500246048,
+      "learning_rate": 3.96653886371558e-06,
+      "loss": 0.054,
+      "step": 11500
+    },
+    {
+      "epoch": 4.011850819100732,
+      "grad_norm": 0.022148391231894493,
+      "learning_rate": 3.952596723597073e-06,
+      "loss": 0.0184,
+      "step": 11510
+    },
+    {
+      "epoch": 4.015336354130359,
+      "grad_norm": 3.5250911712646484,
+      "learning_rate": 3.9386545834785645e-06,
+      "loss": 0.0466,
+      "step": 11520
+    },
+    {
+      "epoch": 4.018821889159986,
+      "grad_norm": 8.17469596862793,
+      "learning_rate": 3.924712443360056e-06,
+      "loss": 0.1563,
+      "step": 11530
+    },
+    {
+      "epoch": 4.022307424189613,
+      "grad_norm": 0.12059949338436127,
+      "learning_rate": 3.910770303241548e-06,
+      "loss": 0.1135,
+      "step": 11540
+    },
+    {
+      "epoch": 4.02579295921924,
+      "grad_norm": 0.23522257804870605,
+      "learning_rate": 3.8968281631230395e-06,
+      "loss": 0.1346,
+      "step": 11550
+    },
+    {
+      "epoch": 4.029278494248867,
+      "grad_norm": 0.014988483861088753,
+      "learning_rate": 3.882886023004532e-06,
+      "loss": 0.0105,
+      "step": 11560
+    },
+    {
+      "epoch": 4.0327640292784945,
+      "grad_norm": 0.3500690460205078,
+      "learning_rate": 3.868943882886023e-06,
+      "loss": 0.0901,
+      "step": 11570
+    },
+    {
+      "epoch": 4.036249564308121,
+      "grad_norm": 0.01964680850505829,
+      "learning_rate": 3.855001742767515e-06,
+      "loss": 0.0047,
+      "step": 11580
+    },
+    {
+      "epoch": 4.039735099337748,
+      "grad_norm": 0.01860162802040577,
+      "learning_rate": 3.841059602649007e-06,
+      "loss": 0.0216,
+      "step": 11590
+    },
+    {
+      "epoch": 4.043220634367375,
+      "grad_norm": 31.001237869262695,
+      "learning_rate": 3.827117462530499e-06,
+      "loss": 0.1348,
+      "step": 11600
+    },
+    {
+      "epoch": 4.046706169397003,
+      "grad_norm": 1.8247324228286743,
+      "learning_rate": 3.8131753224119907e-06,
+      "loss": 0.1087,
+      "step": 11610
+    },
+    {
+      "epoch": 4.05019170442663,
+      "grad_norm": 0.9444401860237122,
+      "learning_rate": 3.7992331822934824e-06,
+      "loss": 0.1226,
+      "step": 11620
+    },
+    {
+      "epoch": 4.053677239456256,
+      "grad_norm": 0.05249320715665817,
+      "learning_rate": 3.785291042174974e-06,
+      "loss": 0.1094,
+      "step": 11630
+    },
+    {
+      "epoch": 4.057162774485883,
+      "grad_norm": 13.864377975463867,
+      "learning_rate": 3.7713489020564657e-06,
+      "loss": 0.1238,
+      "step": 11640
+    },
+    {
+      "epoch": 4.060648309515511,
+      "grad_norm": 0.03523989021778107,
+      "learning_rate": 3.7574067619379578e-06,
+      "loss": 0.0571,
+      "step": 11650
+    },
+    {
+      "epoch": 4.064133844545138,
+      "grad_norm": 22.960308074951172,
+      "learning_rate": 3.74346462181945e-06,
+      "loss": 0.2689,
+      "step": 11660
+    },
+    {
+      "epoch": 4.067619379574765,
+      "grad_norm": 0.023842979222536087,
+      "learning_rate": 3.729522481700941e-06,
+      "loss": 0.1566,
+      "step": 11670
+    },
+    {
+      "epoch": 4.071104914604391,
+      "grad_norm": 0.2230115532875061,
+      "learning_rate": 3.715580341582433e-06,
+      "loss": 0.0784,
+      "step": 11680
+    },
+    {
+      "epoch": 4.074590449634019,
+      "grad_norm": 1.818585753440857,
+      "learning_rate": 3.7016382014639253e-06,
+      "loss": 0.1562,
+      "step": 11690
+    },
+    {
+      "epoch": 4.078075984663646,
+      "grad_norm": 0.024207105860114098,
+      "learning_rate": 3.6876960613454165e-06,
+      "loss": 0.0896,
+      "step": 11700
+    },
+    {
+      "epoch": 4.081561519693273,
+      "grad_norm": 8.846477508544922,
+      "learning_rate": 3.6737539212269086e-06,
+      "loss": 0.1073,
+      "step": 11710
+    },
+    {
+      "epoch": 4.0850470547229,
+      "grad_norm": 4.648227214813232,
+      "learning_rate": 3.6598117811084007e-06,
+      "loss": 0.0199,
+      "step": 11720
+    },
+    {
+      "epoch": 4.088532589752527,
+      "grad_norm": 0.014286634512245655,
+      "learning_rate": 3.6458696409898923e-06,
+      "loss": 0.0208,
+      "step": 11730
+    },
+    {
+      "epoch": 4.092018124782154,
+      "grad_norm": 8.536304473876953,
+      "learning_rate": 3.631927500871384e-06,
+      "loss": 0.0569,
+      "step": 11740
+    },
+    {
+      "epoch": 4.095503659811781,
+      "grad_norm": 0.018616320565342903,
+      "learning_rate": 3.6179853607528757e-06,
+      "loss": 0.0343,
+      "step": 11750
+    },
+    {
+      "epoch": 4.098989194841408,
+      "grad_norm": 0.03940120339393616,
+      "learning_rate": 3.6040432206343678e-06,
+      "loss": 0.0799,
+      "step": 11760
+    },
+    {
+      "epoch": 4.102474729871036,
+      "grad_norm": 0.017780063673853874,
+      "learning_rate": 3.5901010805158594e-06,
+      "loss": 0.0922,
+      "step": 11770
+    },
+    {
+      "epoch": 4.105960264900662,
+      "grad_norm": 0.03300795704126358,
+      "learning_rate": 3.576158940397351e-06,
+      "loss": 0.0589,
+      "step": 11780
+    },
+    {
+      "epoch": 4.109445799930289,
+      "grad_norm": 37.71131896972656,
+      "learning_rate": 3.562216800278843e-06,
+      "loss": 0.144,
+      "step": 11790
+    },
+    {
+      "epoch": 4.112931334959916,
+      "grad_norm": 0.2066090852022171,
+      "learning_rate": 3.5482746601603352e-06,
+      "loss": 0.0068,
+      "step": 11800
+    },
+    {
+      "epoch": 4.116416869989544,
+      "grad_norm": 0.02809598483145237,
+      "learning_rate": 3.5343325200418265e-06,
+      "loss": 0.2141,
+      "step": 11810
+    },
+    {
+      "epoch": 4.119902405019171,
+      "grad_norm": 11.681623458862305,
+      "learning_rate": 3.5203903799233186e-06,
+      "loss": 0.1601,
+      "step": 11820
+    },
+    {
+      "epoch": 4.123387940048797,
+      "grad_norm": 0.021636666730046272,
+      "learning_rate": 3.5064482398048107e-06,
+      "loss": 0.1021,
+      "step": 11830
+    },
+    {
+      "epoch": 4.126873475078424,
+      "grad_norm": 0.01840069517493248,
+      "learning_rate": 3.492506099686302e-06,
+      "loss": 0.0054,
+      "step": 11840
+    },
+    {
+      "epoch": 4.130359010108052,
+      "grad_norm": 14.279787063598633,
+      "learning_rate": 3.478563959567794e-06,
+      "loss": 0.4595,
+      "step": 11850
+    },
+    {
+      "epoch": 4.133844545137679,
+      "grad_norm": 0.06327791512012482,
+      "learning_rate": 3.464621819449286e-06,
+      "loss": 0.0815,
+      "step": 11860
+    },
+    {
+      "epoch": 4.137330080167306,
+      "grad_norm": 7.440787315368652,
+      "learning_rate": 3.4506796793307773e-06,
+      "loss": 0.0566,
+      "step": 11870
+    },
+    {
+      "epoch": 4.140815615196932,
+      "grad_norm": 0.01611819863319397,
+      "learning_rate": 3.4367375392122694e-06,
+      "loss": 0.1947,
+      "step": 11880
+    },
+    {
+      "epoch": 4.14430115022656,
+      "grad_norm": 0.07511651515960693,
+      "learning_rate": 3.422795399093761e-06,
+      "loss": 0.2238,
+      "step": 11890
+    },
+    {
+      "epoch": 4.147786685256187,
+      "grad_norm": 0.05026530474424362,
+      "learning_rate": 3.408853258975253e-06,
+      "loss": 0.1576,
+      "step": 11900
+    },
+    {
+      "epoch": 4.151272220285814,
+      "grad_norm": 12.329663276672363,
+      "learning_rate": 3.394911118856745e-06,
+      "loss": 0.0859,
+      "step": 11910
+    },
+    {
+      "epoch": 4.154757755315441,
+      "grad_norm": 6.723945140838623,
+      "learning_rate": 3.3809689787382365e-06,
+      "loss": 0.1676,
+      "step": 11920
+    },
+    {
+      "epoch": 4.158243290345068,
+      "grad_norm": 0.024456940591335297,
+      "learning_rate": 3.3670268386197285e-06,
+      "loss": 0.1249,
+      "step": 11930
+    },
+    {
+      "epoch": 4.161728825374695,
+      "grad_norm": 0.017061809077858925,
+      "learning_rate": 3.3530846985012198e-06,
+      "loss": 0.1512,
+      "step": 11940
+    },
+    {
+      "epoch": 4.165214360404322,
+      "grad_norm": 0.03508226200938225,
+      "learning_rate": 3.339142558382712e-06,
+      "loss": 0.1291,
+      "step": 11950
+    },
+    {
+      "epoch": 4.168699895433949,
+      "grad_norm": 0.017415596172213554,
+      "learning_rate": 3.325200418264204e-06,
+      "loss": 0.1062,
+      "step": 11960
+    },
+    {
+      "epoch": 4.172185430463577,
+      "grad_norm": 4.462165355682373,
+      "learning_rate": 3.311258278145696e-06,
+      "loss": 0.0761,
+      "step": 11970
+    },
+    {
+      "epoch": 4.175670965493203,
+      "grad_norm": 0.017748460173606873,
+      "learning_rate": 3.2973161380271873e-06,
+      "loss": 0.1781,
+      "step": 11980
+    },
+    {
+      "epoch": 4.17915650052283,
+      "grad_norm": 0.04285305365920067,
+      "learning_rate": 3.2833739979086794e-06,
+      "loss": 0.0114,
+      "step": 11990
+    },
+    {
+      "epoch": 4.182642035552457,
+      "grad_norm": 13.827252388000488,
+      "learning_rate": 3.2694318577901714e-06,
+      "loss": 0.0203,
+      "step": 12000
+    },
+    {
+      "epoch": 4.186127570582085,
+      "grad_norm": 0.01608210802078247,
+      "learning_rate": 3.2554897176716627e-06,
+      "loss": 0.065,
+      "step": 12010
+    },
+    {
+      "epoch": 4.189613105611711,
+      "grad_norm": 22.72849464416504,
+      "learning_rate": 3.2415475775531548e-06,
+      "loss": 0.146,
+      "step": 12020
+    },
+    {
+      "epoch": 4.193098640641338,
+      "grad_norm": 0.013584673404693604,
+      "learning_rate": 3.2276054374346464e-06,
+      "loss": 0.0754,
+      "step": 12030
+    },
+    {
+      "epoch": 4.196584175670965,
+      "grad_norm": 0.02281971462070942,
+      "learning_rate": 3.213663297316138e-06,
+      "loss": 0.0626,
+      "step": 12040
+    },
+    {
+      "epoch": 4.200069710700593,
+      "grad_norm": 1.6404436826705933,
+      "learning_rate": 3.19972115719763e-06,
+      "loss": 0.2863,
+      "step": 12050
+    },
+    {
+      "epoch": 4.20355524573022,
+      "grad_norm": 6.147258281707764,
+      "learning_rate": 3.185779017079122e-06,
+      "loss": 0.1192,
+      "step": 12060
+    },
+    {
+      "epoch": 4.207040780759846,
+      "grad_norm": 19.095373153686523,
+      "learning_rate": 3.171836876960614e-06,
+      "loss": 0.2516,
+      "step": 12070
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.01998833194375038,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.0836,
+      "step": 12080
+    },
+    {
+      "epoch": 4.214011850819101,
+      "grad_norm": 0.02509804256260395,
+      "learning_rate": 3.1439525967235973e-06,
+      "loss": 0.1586,
+      "step": 12090
+    },
+    {
+      "epoch": 4.217497385848728,
+      "grad_norm": 0.02285916358232498,
+      "learning_rate": 3.1300104566050893e-06,
+      "loss": 0.1148,
+      "step": 12100
+    },
+    {
+      "epoch": 4.220982920878355,
+      "grad_norm": 0.017659736797213554,
+      "learning_rate": 3.1160683164865806e-06,
+      "loss": 0.0536,
+      "step": 12110
+    },
+    {
+      "epoch": 4.2244684559079815,
+      "grad_norm": 0.14031673967838287,
+      "learning_rate": 3.1021261763680727e-06,
+      "loss": 0.1591,
+      "step": 12120
+    },
+    {
+      "epoch": 4.227953990937609,
+      "grad_norm": 0.03257942944765091,
+      "learning_rate": 3.0881840362495647e-06,
+      "loss": 0.0295,
+      "step": 12130
+    },
+    {
+      "epoch": 4.231439525967236,
+      "grad_norm": 5.860164165496826,
+      "learning_rate": 3.0742418961310564e-06,
+      "loss": 0.0525,
+      "step": 12140
+    },
+    {
+      "epoch": 4.234925060996863,
+      "grad_norm": 0.023333774879574776,
+      "learning_rate": 3.060299756012548e-06,
+      "loss": 0.1362,
+      "step": 12150
+    },
+    {
+      "epoch": 4.23841059602649,
+      "grad_norm": 0.021630102768540382,
+      "learning_rate": 3.04635761589404e-06,
+      "loss": 0.306,
+      "step": 12160
+    },
+    {
+      "epoch": 4.241896131056117,
+      "grad_norm": 0.9809282422065735,
+      "learning_rate": 3.032415475775532e-06,
+      "loss": 0.0831,
+      "step": 12170
+    },
+    {
+      "epoch": 4.245381666085744,
+      "grad_norm": 0.17336943745613098,
+      "learning_rate": 3.0184733356570235e-06,
+      "loss": 0.0056,
+      "step": 12180
+    },
+    {
+      "epoch": 4.248867201115371,
+      "grad_norm": 0.020092690363526344,
+      "learning_rate": 3.0045311955385156e-06,
+      "loss": 0.023,
+      "step": 12190
+    },
+    {
+      "epoch": 4.252352736144998,
+      "grad_norm": 6.104828834533691,
+      "learning_rate": 2.9905890554200072e-06,
+      "loss": 0.1396,
+      "step": 12200
+    },
+    {
+      "epoch": 4.255838271174626,
+      "grad_norm": 0.06953276693820953,
+      "learning_rate": 2.9766469153014993e-06,
+      "loss": 0.0521,
+      "step": 12210
+    },
+    {
+      "epoch": 4.259323806204252,
+      "grad_norm": 0.06008859723806381,
+      "learning_rate": 2.9627047751829905e-06,
+      "loss": 0.0795,
+      "step": 12220
+    },
+    {
+      "epoch": 4.262809341233879,
+      "grad_norm": 0.06809389591217041,
+      "learning_rate": 2.9487626350644826e-06,
+      "loss": 0.0766,
+      "step": 12230
+    },
+    {
+      "epoch": 4.2662948762635065,
+      "grad_norm": 0.011282606981694698,
+      "learning_rate": 2.9348204949459747e-06,
+      "loss": 0.1127,
+      "step": 12240
+    },
+    {
+      "epoch": 4.269780411293134,
+      "grad_norm": 0.01990368217229843,
+      "learning_rate": 2.920878354827466e-06,
+      "loss": 0.0512,
+      "step": 12250
+    },
+    {
+      "epoch": 4.273265946322761,
+      "grad_norm": 2.730123996734619,
+      "learning_rate": 2.906936214708958e-06,
+      "loss": 0.133,
+      "step": 12260
+    },
+    {
+      "epoch": 4.276751481352387,
+      "grad_norm": 0.018478725105524063,
+      "learning_rate": 2.89299407459045e-06,
+      "loss": 0.3299,
+      "step": 12270
+    },
+    {
+      "epoch": 4.2802370163820145,
+      "grad_norm": 0.027384992688894272,
+      "learning_rate": 2.8790519344719414e-06,
+      "loss": 0.015,
+      "step": 12280
+    },
+    {
+      "epoch": 4.283722551411642,
+      "grad_norm": 0.38204970955848694,
+      "learning_rate": 2.8651097943534334e-06,
+      "loss": 0.0074,
+      "step": 12290
+    },
+    {
+      "epoch": 4.287208086441269,
+      "grad_norm": 6.249034881591797,
+      "learning_rate": 2.8511676542349255e-06,
+      "loss": 0.0899,
+      "step": 12300
+    },
+    {
+      "epoch": 4.290693621470896,
+      "grad_norm": 2.7420356273651123,
+      "learning_rate": 2.837225514116417e-06,
+      "loss": 0.0179,
+      "step": 12310
+    },
+    {
+      "epoch": 4.2941791565005225,
+      "grad_norm": 3.861992835998535,
+      "learning_rate": 2.823283373997909e-06,
+      "loss": 0.1254,
+      "step": 12320
+    },
+    {
+      "epoch": 4.29766469153015,
+      "grad_norm": 1.4951390027999878,
+      "learning_rate": 2.8093412338794005e-06,
+      "loss": 0.0504,
+      "step": 12330
+    },
+    {
+      "epoch": 4.301150226559777,
+      "grad_norm": 0.025097187608480453,
+      "learning_rate": 2.7953990937608926e-06,
+      "loss": 0.0815,
+      "step": 12340
+    },
+    {
+      "epoch": 4.304635761589404,
+      "grad_norm": 0.020737938582897186,
+      "learning_rate": 2.7814569536423843e-06,
+      "loss": 0.0526,
+      "step": 12350
+    },
+    {
+      "epoch": 4.308121296619031,
+      "grad_norm": 0.025950776413083076,
+      "learning_rate": 2.767514813523876e-06,
+      "loss": 0.0167,
+      "step": 12360
+    },
+    {
+      "epoch": 4.311606831648658,
+      "grad_norm": 0.4815838634967804,
+      "learning_rate": 2.753572673405368e-06,
+      "loss": 0.0061,
+      "step": 12370
+    },
+    {
+      "epoch": 4.315092366678285,
+      "grad_norm": 0.020343678072094917,
+      "learning_rate": 2.73963053328686e-06,
+      "loss": 0.1312,
+      "step": 12380
+    },
+    {
+      "epoch": 4.318577901707912,
+      "grad_norm": 6.27520227432251,
+      "learning_rate": 2.7256883931683513e-06,
+      "loss": 0.1138,
+      "step": 12390
+    },
+    {
+      "epoch": 4.3220634367375395,
+      "grad_norm": 19.571958541870117,
+      "learning_rate": 2.7117462530498434e-06,
+      "loss": 0.1826,
+      "step": 12400
+    },
+    {
+      "epoch": 4.325548971767166,
+      "grad_norm": 0.015069302171468735,
+      "learning_rate": 2.6978041129313355e-06,
+      "loss": 0.0538,
+      "step": 12410
+    },
+    {
+      "epoch": 4.329034506796793,
+      "grad_norm": 0.01998594030737877,
+      "learning_rate": 2.6838619728128267e-06,
+      "loss": 0.0339,
+      "step": 12420
+    },
+    {
+      "epoch": 4.33252004182642,
+      "grad_norm": 0.017047762870788574,
+      "learning_rate": 2.669919832694319e-06,
+      "loss": 0.0475,
+      "step": 12430
+    },
+    {
+      "epoch": 4.3360055768560475,
+      "grad_norm": 6.753593444824219,
+      "learning_rate": 2.655977692575811e-06,
+      "loss": 0.2493,
+      "step": 12440
+    },
+    {
+      "epoch": 4.339491111885675,
+      "grad_norm": 0.022272732108831406,
+      "learning_rate": 2.642035552457302e-06,
+      "loss": 0.1162,
+      "step": 12450
+    },
+    {
+      "epoch": 4.342976646915302,
+      "grad_norm": 0.01944439485669136,
+      "learning_rate": 2.6280934123387942e-06,
+      "loss": 0.2014,
+      "step": 12460
+    },
+    {
+      "epoch": 4.346462181944928,
+      "grad_norm": 0.020634399726986885,
+      "learning_rate": 2.614151272220286e-06,
+      "loss": 0.0043,
+      "step": 12470
+    },
+    {
+      "epoch": 4.3499477169745555,
+      "grad_norm": 0.12367178499698639,
+      "learning_rate": 2.600209132101778e-06,
+      "loss": 0.0248,
+      "step": 12480
+    },
+    {
+      "epoch": 4.353433252004183,
+      "grad_norm": 6.508563995361328,
+      "learning_rate": 2.5862669919832696e-06,
+      "loss": 0.1527,
+      "step": 12490
+    },
+    {
+      "epoch": 4.35691878703381,
+      "grad_norm": 0.014639653265476227,
+      "learning_rate": 2.5723248518647613e-06,
+      "loss": 0.0247,
+      "step": 12500
+    },
+    {
+      "epoch": 4.360404322063436,
+      "grad_norm": 0.128875732421875,
+      "learning_rate": 2.5583827117462534e-06,
+      "loss": 0.1266,
+      "step": 12510
+    },
+    {
+      "epoch": 4.3638898570930635,
+      "grad_norm": 0.028906451538205147,
+      "learning_rate": 2.5444405716277446e-06,
+      "loss": 0.1789,
+      "step": 12520
+    },
+    {
+      "epoch": 4.367375392122691,
+      "grad_norm": 0.0448245145380497,
+      "learning_rate": 2.5304984315092367e-06,
+      "loss": 0.1235,
+      "step": 12530
+    },
+    {
+      "epoch": 4.370860927152318,
+      "grad_norm": 0.015661604702472687,
+      "learning_rate": 2.516556291390729e-06,
+      "loss": 0.2039,
+      "step": 12540
+    },
+    {
+      "epoch": 4.374346462181945,
+      "grad_norm": 0.021742140874266624,
+      "learning_rate": 2.502614151272221e-06,
+      "loss": 0.0054,
+      "step": 12550
+    },
+    {
+      "epoch": 4.377831997211572,
+      "grad_norm": 68.89356994628906,
+      "learning_rate": 2.4886720111537126e-06,
+      "loss": 0.1393,
+      "step": 12560
+    },
+    {
+      "epoch": 4.381317532241199,
+      "grad_norm": 0.11168365180492401,
+      "learning_rate": 2.4747298710352042e-06,
+      "loss": 0.0047,
+      "step": 12570
+    },
+    {
+      "epoch": 4.384803067270826,
+      "grad_norm": 0.05092855915427208,
+      "learning_rate": 2.460787730916696e-06,
+      "loss": 0.1151,
+      "step": 12580
+    },
+    {
+      "epoch": 4.388288602300453,
+      "grad_norm": 0.025330761447548866,
+      "learning_rate": 2.446845590798188e-06,
+      "loss": 0.0385,
+      "step": 12590
+    },
+    {
+      "epoch": 4.3917741373300805,
+      "grad_norm": 0.034395404160022736,
+      "learning_rate": 2.4329034506796796e-06,
+      "loss": 0.1206,
+      "step": 12600
+    },
+    {
+      "epoch": 4.395259672359707,
+      "grad_norm": 10.358739852905273,
+      "learning_rate": 2.4189613105611713e-06,
+      "loss": 0.1857,
+      "step": 12610
+    },
+    {
+      "epoch": 4.398745207389334,
+      "grad_norm": 1.8648558855056763,
+      "learning_rate": 2.405019170442663e-06,
+      "loss": 0.0807,
+      "step": 12620
+    },
+    {
+      "epoch": 4.402230742418961,
+      "grad_norm": 3.7046706676483154,
+      "learning_rate": 2.391077030324155e-06,
+      "loss": 0.0273,
+      "step": 12630
+    },
+    {
+      "epoch": 4.4057162774485885,
+      "grad_norm": 0.2638864517211914,
+      "learning_rate": 2.3771348902056467e-06,
+      "loss": 0.052,
+      "step": 12640
+    },
+    {
+      "epoch": 4.409201812478216,
+      "grad_norm": 10.242549896240234,
+      "learning_rate": 2.3631927500871384e-06,
+      "loss": 0.25,
+      "step": 12650
+    },
+    {
+      "epoch": 4.412687347507842,
+      "grad_norm": 0.021201113238930702,
+      "learning_rate": 2.3492506099686304e-06,
+      "loss": 0.1335,
+      "step": 12660
+    },
+    {
+      "epoch": 4.416172882537469,
+      "grad_norm": 0.015749704092741013,
+      "learning_rate": 2.335308469850122e-06,
+      "loss": 0.2059,
+      "step": 12670
+    },
+    {
+      "epoch": 4.4196584175670965,
+      "grad_norm": 0.01748356781899929,
+      "learning_rate": 2.3213663297316138e-06,
+      "loss": 0.1123,
+      "step": 12680
+    },
+    {
+      "epoch": 4.423143952596724,
+      "grad_norm": 16.068105697631836,
+      "learning_rate": 2.307424189613106e-06,
+      "loss": 0.0576,
+      "step": 12690
+    },
+    {
+      "epoch": 4.426629487626351,
+      "grad_norm": 0.042768694460392,
+      "learning_rate": 2.2934820494945975e-06,
+      "loss": 0.0666,
+      "step": 12700
+    },
+    {
+      "epoch": 4.430115022655977,
+      "grad_norm": 6.316594123840332,
+      "learning_rate": 2.2795399093760896e-06,
+      "loss": 0.1412,
+      "step": 12710
+    },
+    {
+      "epoch": 4.433600557685605,
+      "grad_norm": 0.023556068539619446,
+      "learning_rate": 2.2655977692575813e-06,
+      "loss": 0.1011,
+      "step": 12720
+    },
+    {
+      "epoch": 4.437086092715232,
+      "grad_norm": 0.03524250537157059,
+      "learning_rate": 2.2516556291390733e-06,
+      "loss": 0.1937,
+      "step": 12730
+    },
+    {
+      "epoch": 4.440571627744859,
+      "grad_norm": 4.944858551025391,
+      "learning_rate": 2.237713489020565e-06,
+      "loss": 0.1197,
+      "step": 12740
+    },
+    {
+      "epoch": 4.444057162774486,
+      "grad_norm": 18.027040481567383,
+      "learning_rate": 2.2237713489020567e-06,
+      "loss": 0.1547,
+      "step": 12750
+    },
+    {
+      "epoch": 4.447542697804113,
+      "grad_norm": 0.22646109759807587,
+      "learning_rate": 2.2098292087835483e-06,
+      "loss": 0.1819,
+      "step": 12760
+    },
+    {
+      "epoch": 4.45102823283374,
+      "grad_norm": 1.5214154720306396,
+      "learning_rate": 2.1958870686650404e-06,
+      "loss": 0.0784,
+      "step": 12770
+    },
+    {
+      "epoch": 4.454513767863367,
+      "grad_norm": 9.586079597473145,
+      "learning_rate": 2.181944928546532e-06,
+      "loss": 0.1344,
+      "step": 12780
+    },
+    {
+      "epoch": 4.457999302892994,
+      "grad_norm": 0.03214803338050842,
+      "learning_rate": 2.1680027884280237e-06,
+      "loss": 0.0525,
+      "step": 12790
+    },
+    {
+      "epoch": 4.4614848379226215,
+      "grad_norm": 0.06377983093261719,
+      "learning_rate": 2.1540606483095154e-06,
+      "loss": 0.1088,
+      "step": 12800
+    },
+    {
+      "epoch": 4.464970372952248,
+      "grad_norm": 2.2348597049713135,
+      "learning_rate": 2.1401185081910075e-06,
+      "loss": 0.0789,
+      "step": 12810
+    },
+    {
+      "epoch": 4.468455907981875,
+      "grad_norm": 0.040668126195669174,
+      "learning_rate": 2.126176368072499e-06,
+      "loss": 0.033,
+      "step": 12820
+    },
+    {
+      "epoch": 4.471941443011502,
+      "grad_norm": 0.012590868398547173,
+      "learning_rate": 2.1122342279539912e-06,
+      "loss": 0.027,
+      "step": 12830
+    },
+    {
+      "epoch": 4.4754269780411295,
+      "grad_norm": 0.01730727031826973,
+      "learning_rate": 2.098292087835483e-06,
+      "loss": 0.0822,
+      "step": 12840
+    },
+    {
+      "epoch": 4.478912513070757,
+      "grad_norm": 0.05510111153125763,
+      "learning_rate": 2.084349947716975e-06,
+      "loss": 0.0093,
+      "step": 12850
+    },
+    {
+      "epoch": 4.482398048100383,
+      "grad_norm": 0.07657311856746674,
+      "learning_rate": 2.0704078075984666e-06,
+      "loss": 0.0442,
+      "step": 12860
+    },
+    {
+      "epoch": 4.48588358313001,
+      "grad_norm": 9.510165214538574,
+      "learning_rate": 2.0564656674799583e-06,
+      "loss": 0.1325,
+      "step": 12870
+    },
+    {
+      "epoch": 4.489369118159638,
+      "grad_norm": 0.020914927124977112,
+      "learning_rate": 2.0425235273614504e-06,
+      "loss": 0.0475,
+      "step": 12880
+    },
+    {
+      "epoch": 4.492854653189265,
+      "grad_norm": 41.944488525390625,
+      "learning_rate": 2.028581387242942e-06,
+      "loss": 0.1917,
+      "step": 12890
+    },
+    {
+      "epoch": 4.496340188218891,
+      "grad_norm": 25.84819984436035,
+      "learning_rate": 2.0146392471244337e-06,
+      "loss": 0.2525,
+      "step": 12900
+    },
+    {
+      "epoch": 4.499825723248518,
+      "grad_norm": 0.020068377256393433,
+      "learning_rate": 2.0006971070059254e-06,
+      "loss": 0.1163,
+      "step": 12910
+    },
+    {
+      "epoch": 4.503311258278146,
+      "grad_norm": 0.1646548956632614,
+      "learning_rate": 1.9867549668874175e-06,
+      "loss": 0.124,
+      "step": 12920
+    },
+    {
+      "epoch": 4.506796793307773,
+      "grad_norm": 0.013263706117868423,
+      "learning_rate": 1.972812826768909e-06,
+      "loss": 0.1735,
+      "step": 12930
+    },
+    {
+      "epoch": 4.5102823283374,
+      "grad_norm": 0.03346388414502144,
+      "learning_rate": 1.9588706866504008e-06,
+      "loss": 0.006,
+      "step": 12940
+    },
+    {
+      "epoch": 4.513767863367027,
+      "grad_norm": 0.04874487593770027,
+      "learning_rate": 1.944928546531893e-06,
+      "loss": 0.1541,
+      "step": 12950
+    },
+    {
+      "epoch": 4.517253398396654,
+      "grad_norm": 0.03367030993103981,
+      "learning_rate": 1.9309864064133845e-06,
+      "loss": 0.095,
+      "step": 12960
+    },
+    {
+      "epoch": 4.520738933426281,
+      "grad_norm": 0.015354972332715988,
+      "learning_rate": 1.9170442662948766e-06,
+      "loss": 0.2585,
+      "step": 12970
+    },
+    {
+      "epoch": 4.524224468455908,
+      "grad_norm": 17.290773391723633,
+      "learning_rate": 1.9031021261763683e-06,
+      "loss": 0.1559,
+      "step": 12980
+    },
+    {
+      "epoch": 4.527710003485535,
+      "grad_norm": 0.03650696575641632,
+      "learning_rate": 1.88915998605786e-06,
+      "loss": 0.0843,
+      "step": 12990
+    },
+    {
+      "epoch": 4.531195538515162,
+      "grad_norm": 0.8950408697128296,
+      "learning_rate": 1.8752178459393518e-06,
+      "loss": 0.0086,
+      "step": 13000
+    },
+    {
+      "epoch": 4.534681073544789,
+      "grad_norm": 0.047185130417346954,
+      "learning_rate": 1.8612757058208437e-06,
+      "loss": 0.1137,
+      "step": 13010
+    },
+    {
+      "epoch": 4.538166608574416,
+      "grad_norm": 19.941810607910156,
+      "learning_rate": 1.8473335657023356e-06,
+      "loss": 0.0534,
+      "step": 13020
+    },
+    {
+      "epoch": 4.541652143604043,
+      "grad_norm": 0.03801906108856201,
+      "learning_rate": 1.8333914255838272e-06,
+      "loss": 0.1251,
+      "step": 13030
+    },
+    {
+      "epoch": 4.545137678633671,
+      "grad_norm": 0.018815016373991966,
+      "learning_rate": 1.8194492854653189e-06,
+      "loss": 0.1215,
+      "step": 13040
+    },
+    {
+      "epoch": 4.548623213663298,
+      "grad_norm": 0.015067674219608307,
+      "learning_rate": 1.805507145346811e-06,
+      "loss": 0.1624,
+      "step": 13050
+    },
+    {
+      "epoch": 4.552108748692924,
+      "grad_norm": 0.05160703510046005,
+      "learning_rate": 1.7915650052283026e-06,
+      "loss": 0.0379,
+      "step": 13060
+    },
+    {
+      "epoch": 4.555594283722551,
+      "grad_norm": 0.9364856481552124,
+      "learning_rate": 1.7776228651097945e-06,
+      "loss": 0.0417,
+      "step": 13070
+    },
+    {
+      "epoch": 4.559079818752179,
+      "grad_norm": 0.06938211619853973,
+      "learning_rate": 1.7636807249912864e-06,
+      "loss": 0.0135,
+      "step": 13080
+    },
+    {
+      "epoch": 4.562565353781806,
+      "grad_norm": 0.019827021285891533,
+      "learning_rate": 1.749738584872778e-06,
+      "loss": 0.0508,
+      "step": 13090
+    },
+    {
+      "epoch": 4.566050888811432,
+      "grad_norm": 0.11981372535228729,
+      "learning_rate": 1.73579644475427e-06,
+      "loss": 0.057,
+      "step": 13100
+    },
+    {
+      "epoch": 4.569536423841059,
+      "grad_norm": 0.3380275368690491,
+      "learning_rate": 1.7218543046357616e-06,
+      "loss": 0.0149,
+      "step": 13110
+    },
+    {
+      "epoch": 4.573021958870687,
+      "grad_norm": 0.017787497490644455,
+      "learning_rate": 1.7079121645172537e-06,
+      "loss": 0.0651,
+      "step": 13120
+    },
+    {
+      "epoch": 4.576507493900314,
+      "grad_norm": 0.033968303352594376,
+      "learning_rate": 1.6939700243987453e-06,
+      "loss": 0.0363,
+      "step": 13130
+    },
+    {
+      "epoch": 4.579993028929941,
+      "grad_norm": 0.08648999035358429,
+      "learning_rate": 1.6800278842802372e-06,
+      "loss": 0.1379,
+      "step": 13140
+    },
+    {
+      "epoch": 4.583478563959567,
+      "grad_norm": 11.68699836730957,
+      "learning_rate": 1.666085744161729e-06,
+      "loss": 0.0815,
+      "step": 13150
+    },
+    {
+      "epoch": 4.586964098989195,
+      "grad_norm": 0.021847940981388092,
+      "learning_rate": 1.6521436040432207e-06,
+      "loss": 0.116,
+      "step": 13160
+    },
+    {
+      "epoch": 4.590449634018822,
+      "grad_norm": 0.018791014328598976,
+      "learning_rate": 1.6382014639247126e-06,
+      "loss": 0.0267,
+      "step": 13170
+    },
+    {
+      "epoch": 4.593935169048449,
+      "grad_norm": 18.658403396606445,
+      "learning_rate": 1.6242593238062043e-06,
+      "loss": 0.1086,
+      "step": 13180
+    },
+    {
+      "epoch": 4.597420704078076,
+      "grad_norm": 0.09876930713653564,
+      "learning_rate": 1.6103171836876963e-06,
+      "loss": 0.0523,
+      "step": 13190
+    },
+    {
+      "epoch": 4.600906239107703,
+      "grad_norm": 1.7560397386550903,
+      "learning_rate": 1.596375043569188e-06,
+      "loss": 0.0086,
+      "step": 13200
+    },
+    {
+      "epoch": 4.60439177413733,
+      "grad_norm": 0.05320524796843529,
+      "learning_rate": 1.5824329034506797e-06,
+      "loss": 0.0986,
+      "step": 13210
+    },
+    {
+      "epoch": 4.607877309166957,
+      "grad_norm": 0.028736671432852745,
+      "learning_rate": 1.5684907633321718e-06,
+      "loss": 0.1403,
+      "step": 13220
+    },
+    {
+      "epoch": 4.611362844196584,
+      "grad_norm": 0.021549373865127563,
+      "learning_rate": 1.5545486232136634e-06,
+      "loss": 0.0132,
+      "step": 13230
+    },
+    {
+      "epoch": 4.614848379226212,
+      "grad_norm": 0.7282363176345825,
+      "learning_rate": 1.5406064830951553e-06,
+      "loss": 0.0712,
+      "step": 13240
+    },
+    {
+      "epoch": 4.618333914255838,
+      "grad_norm": 0.08495603501796722,
+      "learning_rate": 1.526664342976647e-06,
+      "loss": 0.0062,
+      "step": 13250
+    },
+    {
+      "epoch": 4.621819449285465,
+      "grad_norm": 0.015024428255856037,
+      "learning_rate": 1.512722202858139e-06,
+      "loss": 0.0218,
+      "step": 13260
+    },
+    {
+      "epoch": 4.625304984315092,
+      "grad_norm": 0.04030692204833031,
+      "learning_rate": 1.4987800627396307e-06,
+      "loss": 0.0916,
+      "step": 13270
+    },
+    {
+      "epoch": 4.62879051934472,
+      "grad_norm": 0.018205171450972557,
+      "learning_rate": 1.4848379226211224e-06,
+      "loss": 0.0038,
+      "step": 13280
+    },
+    {
+      "epoch": 4.632276054374346,
+      "grad_norm": 1.867049217224121,
+      "learning_rate": 1.4708957825026142e-06,
+      "loss": 0.1129,
+      "step": 13290
+    },
+    {
+      "epoch": 4.635761589403973,
+      "grad_norm": 9.02762508392334,
+      "learning_rate": 1.456953642384106e-06,
+      "loss": 0.0186,
+      "step": 13300
+    },
+    {
+      "epoch": 4.6392471244336,
+      "grad_norm": 9.14907169342041,
+      "learning_rate": 1.443011502265598e-06,
+      "loss": 0.1519,
+      "step": 13310
+    },
+    {
+      "epoch": 4.642732659463228,
+      "grad_norm": 15.300692558288574,
+      "learning_rate": 1.4290693621470896e-06,
+      "loss": 0.0941,
+      "step": 13320
+    },
+    {
+      "epoch": 4.646218194492855,
+      "grad_norm": 0.02297472208738327,
+      "learning_rate": 1.4151272220285813e-06,
+      "loss": 0.0892,
+      "step": 13330
+    },
+    {
+      "epoch": 4.649703729522482,
+      "grad_norm": 0.013637103140354156,
+      "learning_rate": 1.4011850819100734e-06,
+      "loss": 0.1728,
+      "step": 13340
+    },
+    {
+      "epoch": 4.6531892645521085,
+      "grad_norm": 3.2457101345062256,
+      "learning_rate": 1.387242941791565e-06,
+      "loss": 0.2769,
+      "step": 13350
+    },
+    {
+      "epoch": 4.656674799581736,
+      "grad_norm": 0.019199127331376076,
+      "learning_rate": 1.373300801673057e-06,
+      "loss": 0.0128,
+      "step": 13360
+    },
+    {
+      "epoch": 4.660160334611363,
+      "grad_norm": 4.897704601287842,
+      "learning_rate": 1.3593586615545488e-06,
+      "loss": 0.0735,
+      "step": 13370
+    },
+    {
+      "epoch": 4.66364586964099,
+      "grad_norm": 3.3588168621063232,
+      "learning_rate": 1.3454165214360407e-06,
+      "loss": 0.1323,
+      "step": 13380
+    },
+    {
+      "epoch": 4.6671314046706165,
+      "grad_norm": 4.936280727386475,
+      "learning_rate": 1.3314743813175323e-06,
+      "loss": 0.0841,
+      "step": 13390
+    },
+    {
+      "epoch": 4.670616939700244,
+      "grad_norm": 3.541562080383301,
+      "learning_rate": 1.317532241199024e-06,
+      "loss": 0.0276,
+      "step": 13400
+    },
+    {
+      "epoch": 4.674102474729871,
+      "grad_norm": 0.05973378196358681,
+      "learning_rate": 1.303590101080516e-06,
+      "loss": 0.0098,
+      "step": 13410
+    },
+    {
+      "epoch": 4.677588009759498,
+      "grad_norm": 0.05543622002005577,
+      "learning_rate": 1.2896479609620077e-06,
+      "loss": 0.0983,
+      "step": 13420
+    },
+    {
+      "epoch": 4.681073544789125,
+      "grad_norm": 0.028797583654522896,
+      "learning_rate": 1.2757058208434996e-06,
+      "loss": 0.0048,
+      "step": 13430
+    },
+    {
+      "epoch": 4.684559079818753,
+      "grad_norm": 0.03675423562526703,
+      "learning_rate": 1.2617636807249915e-06,
+      "loss": 0.041,
+      "step": 13440
+    },
+    {
+      "epoch": 4.688044614848379,
+      "grad_norm": 0.04537774622440338,
+      "learning_rate": 1.2478215406064831e-06,
+      "loss": 0.0517,
+      "step": 13450
+    },
+    {
+      "epoch": 4.691530149878006,
+      "grad_norm": 0.11532427370548248,
+      "learning_rate": 1.233879400487975e-06,
+      "loss": 0.056,
+      "step": 13460
+    },
+    {
+      "epoch": 4.695015684907633,
+      "grad_norm": 0.06099528446793556,
+      "learning_rate": 1.2199372603694667e-06,
+      "loss": 0.0924,
+      "step": 13470
+    },
+    {
+      "epoch": 4.698501219937261,
+      "grad_norm": 0.050252217799425125,
+      "learning_rate": 1.2059951202509586e-06,
+      "loss": 0.1061,
+      "step": 13480
+    },
+    {
+      "epoch": 4.701986754966887,
+      "grad_norm": 0.04012449085712433,
+      "learning_rate": 1.1920529801324504e-06,
+      "loss": 0.1954,
+      "step": 13490
+    },
+    {
+      "epoch": 4.705472289996514,
+      "grad_norm": 0.016682496294379234,
+      "learning_rate": 1.1781108400139423e-06,
+      "loss": 0.1782,
+      "step": 13500
+    },
+    {
+      "epoch": 4.7089578250261415,
+      "grad_norm": 0.021084588021039963,
+      "learning_rate": 1.1641686998954342e-06,
+      "loss": 0.2397,
+      "step": 13510
+    },
+    {
+      "epoch": 4.712443360055769,
+      "grad_norm": 0.016580209136009216,
+      "learning_rate": 1.1502265597769258e-06,
+      "loss": 0.0376,
+      "step": 13520
+    },
+    {
+      "epoch": 4.715928895085396,
+      "grad_norm": 11.953326225280762,
+      "learning_rate": 1.1362844196584175e-06,
+      "loss": 0.0902,
+      "step": 13530
+    },
+    {
+      "epoch": 4.719414430115023,
+      "grad_norm": 17.264039993286133,
+      "learning_rate": 1.1223422795399094e-06,
+      "loss": 0.189,
+      "step": 13540
+    },
+    {
+      "epoch": 4.7228999651446495,
+      "grad_norm": 0.02113143354654312,
+      "learning_rate": 1.1084001394214012e-06,
+      "loss": 0.1533,
+      "step": 13550
+    },
+    {
+      "epoch": 4.726385500174277,
+      "grad_norm": 0.029918327927589417,
+      "learning_rate": 1.0944579993028931e-06,
+      "loss": 0.1908,
+      "step": 13560
+    },
+    {
+      "epoch": 4.729871035203904,
+      "grad_norm": 0.1918916404247284,
+      "learning_rate": 1.080515859184385e-06,
+      "loss": 0.1448,
+      "step": 13570
+    },
+    {
+      "epoch": 4.733356570233531,
+      "grad_norm": 3.7918004989624023,
+      "learning_rate": 1.0665737190658767e-06,
+      "loss": 0.0895,
+      "step": 13580
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 15.452225685119629,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.154,
+      "step": 13590
+    },
+    {
+      "epoch": 4.740327640292785,
+      "grad_norm": 0.013763573952019215,
+      "learning_rate": 1.0386894388288602e-06,
+      "loss": 0.0615,
+      "step": 13600
+    },
+    {
+      "epoch": 4.743813175322412,
+      "grad_norm": 2.3458456993103027,
+      "learning_rate": 1.024747298710352e-06,
+      "loss": 0.1512,
+      "step": 13610
+    },
+    {
+      "epoch": 4.747298710352039,
+      "grad_norm": 25.5869197845459,
+      "learning_rate": 1.010805158591844e-06,
+      "loss": 0.1152,
+      "step": 13620
+    },
+    {
+      "epoch": 4.750784245381666,
+      "grad_norm": 0.01689908280968666,
+      "learning_rate": 9.968630184733358e-07,
+      "loss": 0.0267,
+      "step": 13630
+    },
+    {
+      "epoch": 4.754269780411293,
+      "grad_norm": 0.054154131561517715,
+      "learning_rate": 9.829208783548277e-07,
+      "loss": 0.0818,
+      "step": 13640
+    },
+    {
+      "epoch": 4.75775531544092,
+      "grad_norm": 0.030660415068268776,
+      "learning_rate": 9.689787382363193e-07,
+      "loss": 0.0049,
+      "step": 13650
+    },
+    {
+      "epoch": 4.761240850470547,
+      "grad_norm": 0.10165391117334366,
+      "learning_rate": 9.550365981178112e-07,
+      "loss": 0.179,
+      "step": 13660
+    },
+    {
+      "epoch": 4.7647263855001745,
+      "grad_norm": 0.038559578359127045,
+      "learning_rate": 9.41094457999303e-07,
+      "loss": 0.1079,
+      "step": 13670
+    },
+    {
+      "epoch": 4.768211920529802,
+      "grad_norm": 0.322663813829422,
+      "learning_rate": 9.271523178807948e-07,
+      "loss": 0.073,
+      "step": 13680
+    },
+    {
+      "epoch": 4.771697455559428,
+      "grad_norm": 0.013543305918574333,
+      "learning_rate": 9.132101777622866e-07,
+      "loss": 0.0798,
+      "step": 13690
+    },
+    {
+      "epoch": 4.775182990589055,
+      "grad_norm": 0.022806629538536072,
+      "learning_rate": 8.992680376437785e-07,
+      "loss": 0.056,
+      "step": 13700
+    },
+    {
+      "epoch": 4.7786685256186825,
+      "grad_norm": 0.024512002244591713,
+      "learning_rate": 8.853258975252702e-07,
+      "loss": 0.1357,
+      "step": 13710
+    },
+    {
+      "epoch": 4.78215406064831,
+      "grad_norm": 0.03469526022672653,
+      "learning_rate": 8.713837574067619e-07,
+      "loss": 0.0626,
+      "step": 13720
+    },
+    {
+      "epoch": 4.785639595677937,
+      "grad_norm": 19.904218673706055,
+      "learning_rate": 8.574416172882538e-07,
+      "loss": 0.0779,
+      "step": 13730
+    },
+    {
+      "epoch": 4.789125130707563,
+      "grad_norm": 0.01593809947371483,
+      "learning_rate": 8.434994771697457e-07,
+      "loss": 0.0538,
+      "step": 13740
+    },
+    {
+      "epoch": 4.7926106657371905,
+      "grad_norm": 0.012609903700649738,
+      "learning_rate": 8.295573370512374e-07,
+      "loss": 0.162,
+      "step": 13750
+    },
+    {
+      "epoch": 4.796096200766818,
+      "grad_norm": 0.14370125532150269,
+      "learning_rate": 8.156151969327293e-07,
+      "loss": 0.0212,
+      "step": 13760
+    },
+    {
+      "epoch": 4.799581735796445,
+      "grad_norm": 0.16198384761810303,
+      "learning_rate": 8.01673056814221e-07,
+      "loss": 0.0167,
+      "step": 13770
+    },
+    {
+      "epoch": 4.803067270826071,
+      "grad_norm": 37.06806945800781,
+      "learning_rate": 7.877309166957129e-07,
+      "loss": 0.0545,
+      "step": 13780
+    },
+    {
+      "epoch": 4.8065528058556986,
+      "grad_norm": 0.036267928779125214,
+      "learning_rate": 7.737887765772046e-07,
+      "loss": 0.1962,
+      "step": 13790
+    },
+    {
+      "epoch": 4.810038340885326,
+      "grad_norm": 0.038520101457834244,
+      "learning_rate": 7.598466364586965e-07,
+      "loss": 0.0049,
+      "step": 13800
+    },
+    {
+      "epoch": 4.813523875914953,
+      "grad_norm": 0.34361743927001953,
+      "learning_rate": 7.459044963401884e-07,
+      "loss": 0.063,
+      "step": 13810
+    },
+    {
+      "epoch": 4.81700941094458,
+      "grad_norm": 1.973721981048584,
+      "learning_rate": 7.3196235622168e-07,
+      "loss": 0.1321,
+      "step": 13820
+    },
+    {
+      "epoch": 4.8204949459742075,
+      "grad_norm": 0.058219779282808304,
+      "learning_rate": 7.180202161031719e-07,
+      "loss": 0.0716,
+      "step": 13830
+    },
+    {
+      "epoch": 4.823980481003834,
+      "grad_norm": 0.029252415522933006,
+      "learning_rate": 7.040780759846637e-07,
+      "loss": 0.11,
+      "step": 13840
+    },
+    {
+      "epoch": 4.827466016033461,
+      "grad_norm": 8.357165336608887,
+      "learning_rate": 6.901359358661555e-07,
+      "loss": 0.1614,
+      "step": 13850
+    },
+    {
+      "epoch": 4.830951551063088,
+      "grad_norm": 0.27816373109817505,
+      "learning_rate": 6.761937957476473e-07,
+      "loss": 0.1628,
+      "step": 13860
+    },
+    {
+      "epoch": 4.8344370860927155,
+      "grad_norm": 0.04683591425418854,
+      "learning_rate": 6.622516556291392e-07,
+      "loss": 0.1641,
+      "step": 13870
+    },
+    {
+      "epoch": 4.837922621122342,
+      "grad_norm": 13.681280136108398,
+      "learning_rate": 6.483095155106308e-07,
+      "loss": 0.0121,
+      "step": 13880
+    },
+    {
+      "epoch": 4.841408156151969,
+      "grad_norm": 0.020022893324494362,
+      "learning_rate": 6.343673753921227e-07,
+      "loss": 0.0361,
+      "step": 13890
+    },
+    {
+      "epoch": 4.844893691181596,
+      "grad_norm": 0.016766728833317757,
+      "learning_rate": 6.204252352736145e-07,
+      "loss": 0.0044,
+      "step": 13900
+    },
+    {
+      "epoch": 4.8483792262112235,
+      "grad_norm": 0.034691717475652695,
+      "learning_rate": 6.064830951551064e-07,
+      "loss": 0.1031,
+      "step": 13910
+    },
+    {
+      "epoch": 4.851864761240851,
+      "grad_norm": 0.28107964992523193,
+      "learning_rate": 5.925409550365982e-07,
+      "loss": 0.1587,
+      "step": 13920
+    },
+    {
+      "epoch": 4.855350296270478,
+      "grad_norm": 0.032920971512794495,
+      "learning_rate": 5.7859881491809e-07,
+      "loss": 0.1591,
+      "step": 13930
+    },
+    {
+      "epoch": 4.858835831300104,
+      "grad_norm": 0.01294040773063898,
+      "learning_rate": 5.646566747995818e-07,
+      "loss": 0.0468,
+      "step": 13940
+    },
+    {
+      "epoch": 4.8623213663297316,
+      "grad_norm": 0.06196725741028786,
+      "learning_rate": 5.507145346810736e-07,
+      "loss": 0.1613,
+      "step": 13950
+    },
+    {
+      "epoch": 4.865806901359359,
+      "grad_norm": 2.9842488765716553,
+      "learning_rate": 5.367723945625654e-07,
+      "loss": 0.0059,
+      "step": 13960
+    },
+    {
+      "epoch": 4.869292436388986,
+      "grad_norm": 0.224669948220253,
+      "learning_rate": 5.228302544440572e-07,
+      "loss": 0.0523,
+      "step": 13970
+    },
+    {
+      "epoch": 4.872777971418612,
+      "grad_norm": 2.8570218086242676,
+      "learning_rate": 5.08888114325549e-07,
+      "loss": 0.088,
+      "step": 13980
+    },
+    {
+      "epoch": 4.87626350644824,
+      "grad_norm": 0.2920582592487335,
+      "learning_rate": 4.949459742070408e-07,
+      "loss": 0.0094,
+      "step": 13990
+    },
+    {
+      "epoch": 4.879749041477867,
+      "grad_norm": 0.02076049894094467,
+      "learning_rate": 4.810038340885327e-07,
+      "loss": 0.1568,
+      "step": 14000
+    },
+    {
+      "epoch": 4.883234576507494,
+      "grad_norm": 0.015274460427463055,
+      "learning_rate": 4.670616939700244e-07,
+      "loss": 0.1205,
+      "step": 14010
+    },
+    {
+      "epoch": 4.886720111537121,
+      "grad_norm": 2.485515594482422,
+      "learning_rate": 4.5311955385151623e-07,
+      "loss": 0.1998,
+      "step": 14020
+    },
+    {
+      "epoch": 4.890205646566748,
+      "grad_norm": 0.03131948783993721,
+      "learning_rate": 4.3917741373300805e-07,
+      "loss": 0.0188,
+      "step": 14030
+    },
+    {
+      "epoch": 4.893691181596375,
+      "grad_norm": 0.014717698097229004,
+      "learning_rate": 4.252352736144998e-07,
+      "loss": 0.0876,
+      "step": 14040
+    },
+    {
+      "epoch": 4.897176716626002,
+      "grad_norm": 0.023664269596338272,
+      "learning_rate": 4.1129313349599164e-07,
+      "loss": 0.1329,
+      "step": 14050
+    },
+    {
+      "epoch": 4.900662251655629,
+      "grad_norm": 0.017263269051909447,
+      "learning_rate": 3.973509933774835e-07,
+      "loss": 0.2117,
+      "step": 14060
+    },
+    {
+      "epoch": 4.9041477866852565,
+      "grad_norm": 0.2318657636642456,
+      "learning_rate": 3.834088532589753e-07,
+      "loss": 0.0779,
+      "step": 14070
+    },
+    {
+      "epoch": 4.907633321714883,
+      "grad_norm": 0.06534525007009506,
+      "learning_rate": 3.694667131404671e-07,
+      "loss": 0.0394,
+      "step": 14080
+    },
+    {
+      "epoch": 4.91111885674451,
+      "grad_norm": 14.957361221313477,
+      "learning_rate": 3.555245730219589e-07,
+      "loss": 0.0659,
+      "step": 14090
+    },
+    {
+      "epoch": 4.914604391774137,
+      "grad_norm": 11.34598445892334,
+      "learning_rate": 3.415824329034507e-07,
+      "loss": 0.0993,
+      "step": 14100
+    },
+    {
+      "epoch": 4.9180899268037646,
+      "grad_norm": 3.656956672668457,
+      "learning_rate": 3.276402927849425e-07,
+      "loss": 0.0936,
+      "step": 14110
+    },
+    {
+      "epoch": 4.921575461833392,
+      "grad_norm": 0.048377808183431625,
+      "learning_rate": 3.1369815266643433e-07,
+      "loss": 0.1247,
+      "step": 14120
+    },
+    {
+      "epoch": 4.925060996863018,
+      "grad_norm": 21.954980850219727,
+      "learning_rate": 2.9975601254792615e-07,
+      "loss": 0.1416,
+      "step": 14130
+    },
+    {
+      "epoch": 4.928546531892645,
+      "grad_norm": 10.352675437927246,
+      "learning_rate": 2.858138724294179e-07,
+      "loss": 0.0832,
+      "step": 14140
+    },
+    {
+      "epoch": 4.932032066922273,
+      "grad_norm": 3.8607187271118164,
+      "learning_rate": 2.7187173231090974e-07,
+      "loss": 0.1472,
+      "step": 14150
+    },
+    {
+      "epoch": 4.9355176019519,
+      "grad_norm": 0.041085511445999146,
+      "learning_rate": 2.5792959219240156e-07,
+      "loss": 0.1278,
+      "step": 14160
+    },
+    {
+      "epoch": 4.939003136981527,
+      "grad_norm": 0.017858577892184258,
+      "learning_rate": 2.439874520738934e-07,
+      "loss": 0.0691,
+      "step": 14170
+    },
+    {
+      "epoch": 4.942488672011153,
+      "grad_norm": 10.322635650634766,
+      "learning_rate": 2.3004531195538517e-07,
+      "loss": 0.0753,
+      "step": 14180
+    },
+    {
+      "epoch": 4.945974207040781,
+      "grad_norm": 0.03775576502084732,
+      "learning_rate": 2.16103171836877e-07,
+      "loss": 0.022,
+      "step": 14190
+    },
+    {
+      "epoch": 4.949459742070408,
+      "grad_norm": 23.317575454711914,
+      "learning_rate": 2.021610317183688e-07,
+      "loss": 0.1228,
+      "step": 14200
+    },
+    {
+      "epoch": 4.952945277100035,
+      "grad_norm": 0.027847325429320335,
+      "learning_rate": 1.8821889159986058e-07,
+      "loss": 0.0041,
+      "step": 14210
+    },
+    {
+      "epoch": 4.956430812129662,
+      "grad_norm": 7.33128023147583,
+      "learning_rate": 1.7427675148135243e-07,
+      "loss": 0.0204,
+      "step": 14220
+    },
+    {
+      "epoch": 4.959916347159289,
+      "grad_norm": 0.49565252661705017,
+      "learning_rate": 1.6033461136284422e-07,
+      "loss": 0.0091,
+      "step": 14230
+    },
+    {
+      "epoch": 4.963401882188916,
+      "grad_norm": 0.3179240822792053,
+      "learning_rate": 1.4639247124433602e-07,
+      "loss": 0.1085,
+      "step": 14240
+    },
+    {
+      "epoch": 4.966887417218543,
+      "grad_norm": 0.015233837999403477,
+      "learning_rate": 1.3245033112582784e-07,
+      "loss": 0.0079,
+      "step": 14250
+    },
+    {
+      "epoch": 4.97037295224817,
+      "grad_norm": 4.127240180969238,
+      "learning_rate": 1.1850819100731964e-07,
+      "loss": 0.0403,
+      "step": 14260
+    },
+    {
+      "epoch": 4.973858487277797,
+      "grad_norm": 8.784873962402344,
+      "learning_rate": 1.0456605088881144e-07,
+      "loss": 0.0699,
+      "step": 14270
+    },
+    {
+      "epoch": 4.977344022307424,
+      "grad_norm": 9.374968528747559,
+      "learning_rate": 9.062391077030325e-08,
+      "loss": 0.039,
+      "step": 14280
+    },
+    {
+      "epoch": 4.980829557337051,
+      "grad_norm": 0.012251177802681923,
+      "learning_rate": 7.668177065179505e-08,
+      "loss": 0.1414,
+      "step": 14290
+    },
+    {
+      "epoch": 4.984315092366678,
+      "grad_norm": 2.2470829486846924,
+      "learning_rate": 6.273963053328686e-08,
+      "loss": 0.1129,
+      "step": 14300
+    },
+    {
+      "epoch": 4.987800627396306,
+      "grad_norm": 1.4505870342254639,
+      "learning_rate": 4.8797490414778674e-08,
+      "loss": 0.0438,
+      "step": 14310
+    },
+    {
+      "epoch": 4.991286162425933,
+      "grad_norm": 0.05611838400363922,
+      "learning_rate": 3.485535029627048e-08,
+      "loss": 0.0759,
+      "step": 14320
+    },
+    {
+      "epoch": 4.994771697455559,
+      "grad_norm": 0.01623629219830036,
+      "learning_rate": 2.0913210177762286e-08,
+      "loss": 0.0602,
+      "step": 14330
+    },
+    {
+      "epoch": 4.998257232485186,
+      "grad_norm": 0.022672150284051895,
+      "learning_rate": 6.971070059254096e-09,
+      "loss": 0.1617,
+      "step": 14340
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9916049382716049,
+      "eval_loss": 0.039273809641599655,
+      "eval_runtime": 19.3797,
+      "eval_samples_per_second": 208.981,
+      "eval_steps_per_second": 26.161,
+      "step": 14345
+    },
+    {
+      "epoch": 5.0,
+      "step": 14345,
+      "total_flos": 8.892843392498688e+18,
+      "train_loss": 0.1924674165219266,
+      "train_runtime": 1341.6959,
+      "train_samples_per_second": 85.526,
+      "train_steps_per_second": 10.692
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 14345,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.892843392498688e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}