{
  "best_metric": 1.1142469644546509,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.16266775111834078,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008133387555917039,
      "grad_norm": 1.6022670269012451,
      "learning_rate": 1e-05,
      "loss": 1.2811,
      "step": 1
    },
    {
      "epoch": 0.0008133387555917039,
      "eval_loss": 1.8573664426803589,
      "eval_runtime": 105.0117,
      "eval_samples_per_second": 19.722,
      "eval_steps_per_second": 4.933,
      "step": 1
    },
    {
      "epoch": 0.0016266775111834079,
      "grad_norm": 1.5892877578735352,
      "learning_rate": 2e-05,
      "loss": 1.3231,
      "step": 2
    },
    {
      "epoch": 0.0024400162667751117,
      "grad_norm": 1.585437297821045,
      "learning_rate": 3e-05,
      "loss": 1.3782,
      "step": 3
    },
    {
      "epoch": 0.0032533550223668157,
      "grad_norm": 1.4400299787521362,
      "learning_rate": 4e-05,
      "loss": 1.2963,
      "step": 4
    },
    {
      "epoch": 0.004066693777958519,
      "grad_norm": 1.5858532190322876,
      "learning_rate": 5e-05,
      "loss": 1.3276,
      "step": 5
    },
    {
      "epoch": 0.004880032533550223,
      "grad_norm": 1.3891819715499878,
      "learning_rate": 6e-05,
      "loss": 1.1632,
      "step": 6
    },
    {
      "epoch": 0.0056933712891419274,
      "grad_norm": 1.312585473060608,
      "learning_rate": 7e-05,
      "loss": 1.2927,
      "step": 7
    },
    {
      "epoch": 0.0065067100447336315,
      "grad_norm": 1.4606459140777588,
      "learning_rate": 8e-05,
      "loss": 1.3241,
      "step": 8
    },
    {
      "epoch": 0.0073200488003253355,
      "grad_norm": 1.3136050701141357,
      "learning_rate": 9e-05,
      "loss": 1.1779,
      "step": 9
    },
    {
      "epoch": 0.008133387555917039,
      "grad_norm": 1.3164485692977905,
      "learning_rate": 0.0001,
      "loss": 1.1954,
      "step": 10
    },
    {
      "epoch": 0.008946726311508743,
      "grad_norm": 1.427384376525879,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.3077,
      "step": 11
    },
    {
      "epoch": 0.009760065067100447,
      "grad_norm": 1.1989771127700806,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.1843,
      "step": 12
    },
    {
      "epoch": 0.01057340382269215,
      "grad_norm": 1.13768470287323,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.1632,
      "step": 13
    },
    {
      "epoch": 0.011386742578283855,
      "grad_norm": 1.193123459815979,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.2145,
      "step": 14
    },
    {
      "epoch": 0.012200081333875559,
      "grad_norm": 1.1354650259017944,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.2205,
      "step": 15
    },
    {
      "epoch": 0.013013420089467263,
      "grad_norm": 1.2305574417114258,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.2329,
      "step": 16
    },
    {
      "epoch": 0.013826758845058967,
      "grad_norm": 1.382487416267395,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.2438,
      "step": 17
    },
    {
      "epoch": 0.014640097600650671,
      "grad_norm": 1.2229613065719604,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.2904,
      "step": 18
    },
    {
      "epoch": 0.015453436356242375,
      "grad_norm": 1.3983185291290283,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2182,
      "step": 19
    },
    {
      "epoch": 0.016266775111834077,
      "grad_norm": 1.3911596536636353,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.0935,
      "step": 20
    },
    {
      "epoch": 0.01708011386742578,
      "grad_norm": 1.3146138191223145,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1708,
      "step": 21
    },
    {
      "epoch": 0.017893452623017485,
      "grad_norm": 1.3501468896865845,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.2914,
      "step": 22
    },
    {
      "epoch": 0.01870679137860919,
      "grad_norm": 1.383859395980835,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1796,
      "step": 23
    },
    {
      "epoch": 0.019520130134200894,
      "grad_norm": 1.4043275117874146,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.2198,
      "step": 24
    },
    {
      "epoch": 0.020333468889792598,
      "grad_norm": 1.4049819707870483,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.185,
      "step": 25
    },
    {
      "epoch": 0.0211468076453843,
      "grad_norm": 1.7388041019439697,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1239,
      "step": 26
    },
    {
      "epoch": 0.021960146400976006,
      "grad_norm": 1.7563539743423462,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.3009,
      "step": 27
    },
    {
      "epoch": 0.02277348515656771,
      "grad_norm": 1.9792587757110596,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.2487,
      "step": 28
    },
    {
      "epoch": 0.023586823912159414,
      "grad_norm": 1.8014628887176514,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.1514,
      "step": 29
    },
    {
      "epoch": 0.024400162667751118,
      "grad_norm": 2.1420493125915527,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.4041,
      "step": 30
    },
    {
      "epoch": 0.025213501423342822,
      "grad_norm": 2.027292251586914,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.2724,
      "step": 31
    },
    {
      "epoch": 0.026026840178934526,
      "grad_norm": 2.3917717933654785,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3868,
      "step": 32
    },
    {
      "epoch": 0.02684017893452623,
      "grad_norm": 2.1205084323883057,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1567,
      "step": 33
    },
    {
      "epoch": 0.027653517690117934,
      "grad_norm": 2.3205273151397705,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.1934,
      "step": 34
    },
    {
      "epoch": 0.028466856445709638,
      "grad_norm": 2.4539246559143066,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.126,
      "step": 35
    },
    {
      "epoch": 0.029280195201301342,
      "grad_norm": 2.833463191986084,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.1508,
      "step": 36
    },
    {
      "epoch": 0.030093533956893046,
      "grad_norm": 3.2831156253814697,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3324,
      "step": 37
    },
    {
      "epoch": 0.03090687271248475,
      "grad_norm": 2.897948741912842,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1003,
      "step": 38
    },
    {
      "epoch": 0.03172021146807645,
      "grad_norm": 4.020732879638672,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.1433,
      "step": 39
    },
    {
      "epoch": 0.032533550223668155,
      "grad_norm": 3.94914174079895,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.1331,
      "step": 40
    },
    {
      "epoch": 0.03334688897925986,
      "grad_norm": 3.390803813934326,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1329,
      "step": 41
    },
    {
      "epoch": 0.03416022773485156,
      "grad_norm": 4.285490036010742,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.1166,
      "step": 42
    },
    {
      "epoch": 0.03497356649044327,
      "grad_norm": 3.830636978149414,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.8445,
      "step": 43
    },
    {
      "epoch": 0.03578690524603497,
      "grad_norm": 4.412781238555908,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.1155,
      "step": 44
    },
    {
      "epoch": 0.036600244001626675,
      "grad_norm": 3.7231521606445312,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.1511,
      "step": 45
    },
    {
      "epoch": 0.03741358275721838,
      "grad_norm": 4.253747463226318,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1276,
      "step": 46
    },
    {
      "epoch": 0.03822692151281008,
      "grad_norm": 4.689858436584473,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.0605,
      "step": 47
    },
    {
      "epoch": 0.03904026026840179,
      "grad_norm": 7.302494525909424,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.0956,
      "step": 48
    },
    {
      "epoch": 0.03985359902399349,
      "grad_norm": 5.599308490753174,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.0195,
      "step": 49
    },
    {
      "epoch": 0.040666937779585195,
      "grad_norm": 7.710896968841553,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.5138,
      "step": 50
    },
    {
      "epoch": 0.040666937779585195,
      "eval_loss": 1.3900275230407715,
      "eval_runtime": 106.3135,
      "eval_samples_per_second": 19.48,
      "eval_steps_per_second": 4.872,
      "step": 50
    },
    {
      "epoch": 0.0414802765351769,
      "grad_norm": 2.869823694229126,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.3522,
      "step": 51
    },
    {
      "epoch": 0.0422936152907686,
      "grad_norm": 2.149834632873535,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.329,
      "step": 52
    },
    {
      "epoch": 0.04310695404636031,
      "grad_norm": 1.260952115058899,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.2092,
      "step": 53
    },
    {
      "epoch": 0.04392029280195201,
      "grad_norm": 1.0235779285430908,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.2115,
      "step": 54
    },
    {
      "epoch": 0.044733631557543715,
      "grad_norm": 0.9350866675376892,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.1941,
      "step": 55
    },
    {
      "epoch": 0.04554697031313542,
      "grad_norm": 1.0205262899398804,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.2809,
      "step": 56
    },
    {
      "epoch": 0.046360309068727124,
      "grad_norm": 0.9916782975196838,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.1713,
      "step": 57
    },
    {
      "epoch": 0.04717364782431883,
      "grad_norm": 0.928591251373291,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.1614,
      "step": 58
    },
    {
      "epoch": 0.04798698657991053,
      "grad_norm": 0.8215128779411316,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.1461,
      "step": 59
    },
    {
      "epoch": 0.048800325335502236,
      "grad_norm": 0.8294538259506226,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.1496,
      "step": 60
    },
    {
      "epoch": 0.04961366409109394,
      "grad_norm": 0.8535639047622681,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.1594,
      "step": 61
    },
    {
      "epoch": 0.050427002846685644,
      "grad_norm": 0.8942161798477173,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1212,
      "step": 62
    },
    {
      "epoch": 0.05124034160227735,
      "grad_norm": 0.9346653819084167,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.2919,
      "step": 63
    },
    {
      "epoch": 0.05205368035786905,
      "grad_norm": 0.9574184417724609,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.2509,
      "step": 64
    },
    {
      "epoch": 0.052867019113460756,
      "grad_norm": 1.0795056819915771,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.2063,
      "step": 65
    },
    {
      "epoch": 0.05368035786905246,
      "grad_norm": 0.9987140893936157,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.2246,
      "step": 66
    },
    {
      "epoch": 0.054493696624644164,
      "grad_norm": 0.9882590174674988,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.2133,
      "step": 67
    },
    {
      "epoch": 0.05530703538023587,
      "grad_norm": 1.1103553771972656,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.1717,
      "step": 68
    },
    {
      "epoch": 0.05612037413582757,
      "grad_norm": 1.0271129608154297,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.1553,
      "step": 69
    },
    {
      "epoch": 0.056933712891419276,
      "grad_norm": 1.105254888534546,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2352,
      "step": 70
    },
    {
      "epoch": 0.05774705164701098,
      "grad_norm": 1.1660130023956299,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1516,
      "step": 71
    },
    {
      "epoch": 0.058560390402602684,
      "grad_norm": 1.1904737949371338,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.192,
      "step": 72
    },
    {
      "epoch": 0.05937372915819439,
      "grad_norm": 1.187439203262329,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2234,
      "step": 73
    },
    {
      "epoch": 0.06018706791378609,
      "grad_norm": 1.2708885669708252,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.2796,
      "step": 74
    },
    {
      "epoch": 0.061000406669377796,
      "grad_norm": 1.3695472478866577,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.2334,
      "step": 75
    },
    {
      "epoch": 0.0618137454249695,
      "grad_norm": 1.479674220085144,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.2525,
      "step": 76
    },
    {
      "epoch": 0.0626270841805612,
      "grad_norm": 1.6190848350524902,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.1842,
      "step": 77
    },
    {
      "epoch": 0.0634404229361529,
      "grad_norm": 1.574918508529663,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.324,
      "step": 78
    },
    {
      "epoch": 0.06425376169174461,
      "grad_norm": 1.6462424993515015,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1988,
      "step": 79
    },
    {
      "epoch": 0.06506710044733631,
      "grad_norm": 1.9923938512802124,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.2768,
      "step": 80
    },
    {
      "epoch": 0.06588043920292802,
      "grad_norm": 1.8923838138580322,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3076,
      "step": 81
    },
    {
      "epoch": 0.06669377795851972,
      "grad_norm": 2.2907886505126953,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2806,
      "step": 82
    },
    {
      "epoch": 0.06750711671411143,
      "grad_norm": 2.1847496032714844,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.259,
      "step": 83
    },
    {
      "epoch": 0.06832045546970313,
      "grad_norm": 2.1230287551879883,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1264,
      "step": 84
    },
    {
      "epoch": 0.06913379422529484,
      "grad_norm": 2.4889023303985596,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.1212,
      "step": 85
    },
    {
      "epoch": 0.06994713298088653,
      "grad_norm": 2.336205244064331,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1926,
      "step": 86
    },
    {
      "epoch": 0.07076047173647824,
      "grad_norm": 2.931776523590088,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2133,
      "step": 87
    },
    {
      "epoch": 0.07157381049206994,
      "grad_norm": 2.3583552837371826,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.024,
      "step": 88
    },
    {
      "epoch": 0.07238714924766165,
      "grad_norm": 2.620591163635254,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.012,
      "step": 89
    },
    {
      "epoch": 0.07320048800325335,
      "grad_norm": 2.955986499786377,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0753,
      "step": 90
    },
    {
      "epoch": 0.07401382675884506,
      "grad_norm": 2.8933234214782715,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.9869,
      "step": 91
    },
    {
      "epoch": 0.07482716551443676,
      "grad_norm": 2.7858355045318604,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.8879,
      "step": 92
    },
    {
      "epoch": 0.07564050427002847,
      "grad_norm": 3.634690523147583,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2163,
      "step": 93
    },
    {
      "epoch": 0.07645384302562017,
      "grad_norm": 4.090825080871582,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.0691,
      "step": 94
    },
    {
      "epoch": 0.07726718178121188,
      "grad_norm": 3.452148914337158,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.0211,
      "step": 95
    },
    {
      "epoch": 0.07808052053680357,
      "grad_norm": 3.6106181144714355,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.0255,
      "step": 96
    },
    {
      "epoch": 0.07889385929239529,
      "grad_norm": 4.434029579162598,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.9293,
      "step": 97
    },
    {
      "epoch": 0.07970719804798698,
      "grad_norm": 3.298231601715088,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.8045,
      "step": 98
    },
    {
      "epoch": 0.0805205368035787,
      "grad_norm": 4.638715744018555,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.0016,
      "step": 99
    },
    {
      "epoch": 0.08133387555917039,
      "grad_norm": 6.909847259521484,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.1923,
      "step": 100
    },
    {
      "epoch": 0.08133387555917039,
      "eval_loss": 1.3159615993499756,
      "eval_runtime": 106.3453,
      "eval_samples_per_second": 19.474,
      "eval_steps_per_second": 4.871,
      "step": 100
    },
    {
      "epoch": 0.0821472143147621,
      "grad_norm": 2.0550525188446045,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2755,
      "step": 101
    },
    {
      "epoch": 0.0829605530703538,
      "grad_norm": 2.024679183959961,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.332,
      "step": 102
    },
    {
      "epoch": 0.08377389182594551,
      "grad_norm": 1.420579433441162,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.2512,
      "step": 103
    },
    {
      "epoch": 0.0845872305815372,
      "grad_norm": 1.018290400505066,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.2379,
      "step": 104
    },
    {
      "epoch": 0.08540056933712892,
      "grad_norm": 0.8214701414108276,
      "learning_rate": 5e-05,
      "loss": 1.2182,
      "step": 105
    },
    {
      "epoch": 0.08621390809272061,
      "grad_norm": 0.7688771486282349,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.1202,
      "step": 106
    },
    {
      "epoch": 0.08702724684831233,
      "grad_norm": 0.7829757928848267,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.1586,
      "step": 107
    },
    {
      "epoch": 0.08784058560390402,
      "grad_norm": 0.7908482551574707,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.1513,
      "step": 108
    },
    {
      "epoch": 0.08865392435949573,
      "grad_norm": 0.8245127201080322,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.1873,
      "step": 109
    },
    {
      "epoch": 0.08946726311508743,
      "grad_norm": 0.8461537957191467,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1589,
      "step": 110
    },
    {
      "epoch": 0.09028060187067914,
      "grad_norm": 0.8592161536216736,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2259,
      "step": 111
    },
    {
      "epoch": 0.09109394062627084,
      "grad_norm": 0.8966860175132751,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.168,
      "step": 112
    },
    {
      "epoch": 0.09190727938186255,
      "grad_norm": 0.9335092306137085,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1877,
      "step": 113
    },
    {
      "epoch": 0.09272061813745425,
      "grad_norm": 0.9056230187416077,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.1673,
      "step": 114
    },
    {
      "epoch": 0.09353395689304596,
      "grad_norm": 0.8870036005973816,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1803,
      "step": 115
    },
    {
      "epoch": 0.09434729564863766,
      "grad_norm": 0.8796521425247192,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1435,
      "step": 116
    },
    {
      "epoch": 0.09516063440422937,
      "grad_norm": 0.9620208144187927,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.247,
      "step": 117
    },
    {
      "epoch": 0.09597397315982106,
      "grad_norm": 0.982358455657959,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2598,
      "step": 118
    },
    {
      "epoch": 0.09678731191541277,
      "grad_norm": 1.008872151374817,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.1617,
      "step": 119
    },
    {
      "epoch": 0.09760065067100447,
      "grad_norm": 1.0747501850128174,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3292,
      "step": 120
    },
    {
      "epoch": 0.09841398942659618,
      "grad_norm": 1.1409488916397095,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.2098,
      "step": 121
    },
    {
      "epoch": 0.09922732818218788,
      "grad_norm": 1.0723323822021484,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.1448,
      "step": 122
    },
    {
      "epoch": 0.10004066693777959,
      "grad_norm": 1.1127489805221558,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.3292,
      "step": 123
    },
    {
      "epoch": 0.10085400569337129,
      "grad_norm": 1.2123587131500244,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1752,
      "step": 124
    },
    {
      "epoch": 0.101667344448963,
      "grad_norm": 1.2362475395202637,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.2633,
      "step": 125
    },
    {
      "epoch": 0.1024806832045547,
      "grad_norm": 1.2689555883407593,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.3501,
      "step": 126
    },
    {
      "epoch": 0.1032940219601464,
      "grad_norm": 1.2432509660720825,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.1881,
      "step": 127
    },
    {
      "epoch": 0.1041073607157381,
      "grad_norm": 1.3342244625091553,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.2224,
      "step": 128
    },
    {
      "epoch": 0.10492069947132981,
      "grad_norm": 1.5508168935775757,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.1327,
      "step": 129
    },
    {
      "epoch": 0.10573403822692151,
      "grad_norm": 1.4915046691894531,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.1749,
      "step": 130
    },
    {
      "epoch": 0.10654737698251322,
      "grad_norm": 1.7136361598968506,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1905,
      "step": 131
    },
    {
      "epoch": 0.10736071573810492,
      "grad_norm": 1.7495901584625244,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1947,
      "step": 132
    },
    {
      "epoch": 0.10817405449369663,
      "grad_norm": 1.9899603128433228,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1413,
      "step": 133
    },
    {
      "epoch": 0.10898739324928833,
      "grad_norm": 2.017275810241699,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.2073,
      "step": 134
    },
    {
      "epoch": 0.10980073200488004,
      "grad_norm": 2.167989492416382,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.1697,
      "step": 135
    },
    {
      "epoch": 0.11061407076047174,
      "grad_norm": 2.149290084838867,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0847,
      "step": 136
    },
    {
      "epoch": 0.11142740951606345,
      "grad_norm": 2.4723289012908936,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.264,
      "step": 137
    },
    {
      "epoch": 0.11224074827165514,
      "grad_norm": 2.4806630611419678,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1523,
      "step": 138
    },
    {
      "epoch": 0.11305408702724686,
      "grad_norm": 2.612274169921875,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.0612,
      "step": 139
    },
    {
      "epoch": 0.11386742578283855,
      "grad_norm": 3.0523922443389893,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.0515,
      "step": 140
    },
    {
      "epoch": 0.11468076453843025,
      "grad_norm": 2.8779609203338623,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0776,
      "step": 141
    },
    {
      "epoch": 0.11549410329402196,
      "grad_norm": 2.810849666595459,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.0472,
      "step": 142
    },
    {
      "epoch": 0.11630744204961366,
      "grad_norm": 3.260906457901001,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0201,
      "step": 143
    },
    {
      "epoch": 0.11712078080520537,
      "grad_norm": 3.4329018592834473,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.7933,
      "step": 144
    },
    {
      "epoch": 0.11793411956079707,
      "grad_norm": 3.347562789916992,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.0614,
      "step": 145
    },
    {
      "epoch": 0.11874745831638878,
      "grad_norm": 4.097909450531006,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.0916,
      "step": 146
    },
    {
      "epoch": 0.11956079707198047,
      "grad_norm": 3.529541492462158,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.8505,
      "step": 147
    },
    {
      "epoch": 0.12037413582757218,
      "grad_norm": 4.165427207946777,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9878,
      "step": 148
    },
    {
      "epoch": 0.12118747458316388,
      "grad_norm": 4.062654495239258,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.8192,
      "step": 149
    },
    {
      "epoch": 0.12200081333875559,
      "grad_norm": 7.141842842102051,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.5491,
      "step": 150
    },
    {
      "epoch": 0.12200081333875559,
      "eval_loss": 1.1605161428451538,
      "eval_runtime": 106.3455,
      "eval_samples_per_second": 19.474,
      "eval_steps_per_second": 4.871,
      "step": 150
    },
    {
      "epoch": 0.12281415209434729,
      "grad_norm": 1.223806381225586,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.1328,
      "step": 151
    },
    {
      "epoch": 0.123627490849939,
      "grad_norm": 1.3332300186157227,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.2235,
      "step": 152
    },
    {
      "epoch": 0.1244408296055307,
      "grad_norm": 1.1319531202316284,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.2651,
      "step": 153
    },
    {
      "epoch": 0.1252541683611224,
      "grad_norm": 1.0426899194717407,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.2096,
      "step": 154
    },
    {
      "epoch": 0.1260675071167141,
      "grad_norm": 0.9765622615814209,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1382,
      "step": 155
    },
    {
      "epoch": 0.1268808458723058,
      "grad_norm": 0.9610655903816223,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.2507,
      "step": 156
    },
    {
      "epoch": 0.12769418462789753,
      "grad_norm": 0.8247324824333191,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.1435,
      "step": 157
    },
    {
      "epoch": 0.12850752338348922,
      "grad_norm": 0.8507121205329895,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1703,
      "step": 158
    },
    {
      "epoch": 0.12932086213908092,
      "grad_norm": 0.7899272441864014,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.1104,
      "step": 159
    },
    {
      "epoch": 0.13013420089467262,
      "grad_norm": 0.8426223993301392,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2192,
      "step": 160
    },
    {
      "epoch": 0.13094753965026434,
      "grad_norm": 0.8691783547401428,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2016,
      "step": 161
    },
    {
      "epoch": 0.13176087840585604,
      "grad_norm": 0.807234525680542,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2204,
      "step": 162
    },
    {
      "epoch": 0.13257421716144774,
      "grad_norm": 0.8328830003738403,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.1286,
      "step": 163
    },
    {
      "epoch": 0.13338755591703944,
      "grad_norm": 0.8518633246421814,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.1186,
      "step": 164
    },
    {
      "epoch": 0.13420089467263116,
      "grad_norm": 0.8813118934631348,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.1599,
      "step": 165
    },
    {
      "epoch": 0.13501423342822286,
      "grad_norm": 0.9315224885940552,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.2147,
      "step": 166
    },
    {
      "epoch": 0.13582757218381455,
      "grad_norm": 0.8761865496635437,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.2306,
      "step": 167
    },
    {
      "epoch": 0.13664091093940625,
      "grad_norm": 0.904680609703064,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.1391,
      "step": 168
    },
    {
      "epoch": 0.13745424969499798,
      "grad_norm": 0.9789652228355408,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.243,
      "step": 169
    },
    {
      "epoch": 0.13826758845058967,
      "grad_norm": 0.9535621404647827,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.1471,
      "step": 170
    },
    {
      "epoch": 0.13908092720618137,
      "grad_norm": 1.0461153984069824,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.3118,
      "step": 171
    },
    {
      "epoch": 0.13989426596177307,
      "grad_norm": 1.0268018245697021,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.3124,
      "step": 172
    },
    {
      "epoch": 0.1407076047173648,
      "grad_norm": 1.0548676252365112,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.2052,
      "step": 173
    },
    {
      "epoch": 0.1415209434729565,
      "grad_norm": 1.0981262922286987,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.3233,
      "step": 174
    },
    {
      "epoch": 0.1423342822285482,
      "grad_norm": 1.104276180267334,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.2218,
      "step": 175
    },
    {
      "epoch": 0.14314762098413988,
      "grad_norm": 1.2572919130325317,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.1604,
      "step": 176
    },
    {
      "epoch": 0.1439609597397316,
      "grad_norm": 1.2412906885147095,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.2462,
      "step": 177
    },
    {
      "epoch": 0.1447742984953233,
      "grad_norm": 1.2660815715789795,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.2587,
      "step": 178
    },
    {
      "epoch": 0.145587637250915,
      "grad_norm": 1.39919114112854,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.251,
      "step": 179
    },
    {
      "epoch": 0.1464009760065067,
      "grad_norm": 1.464840292930603,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.1706,
      "step": 180
    },
    {
      "epoch": 0.14721431476209842,
      "grad_norm": 1.571134328842163,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2754,
      "step": 181
    },
    {
      "epoch": 0.14802765351769012,
      "grad_norm": 2.198638677597046,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1012,
      "step": 182
    },
    {
      "epoch": 0.14884099227328182,
      "grad_norm": 1.6922011375427246,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.1154,
      "step": 183
    },
    {
      "epoch": 0.14965433102887352,
      "grad_norm": 1.8634734153747559,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.073,
      "step": 184
    },
    {
      "epoch": 0.15046766978446524,
      "grad_norm": 2.5235533714294434,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.0514,
      "step": 185
    },
    {
      "epoch": 0.15128100854005694,
      "grad_norm": 2.239865779876709,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.2084,
      "step": 186
    },
    {
      "epoch": 0.15209434729564864,
      "grad_norm": 2.519874334335327,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0341,
      "step": 187
    },
    {
      "epoch": 0.15290768605124033,
      "grad_norm": 2.7177417278289795,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1573,
      "step": 188
    },
    {
      "epoch": 0.15372102480683206,
      "grad_norm": 2.7432591915130615,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.0424,
      "step": 189
    },
    {
      "epoch": 0.15453436356242375,
      "grad_norm": 2.7542333602905273,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.9666,
      "step": 190
    },
    {
      "epoch": 0.15534770231801545,
      "grad_norm": 2.9724550247192383,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.128,
      "step": 191
    },
    {
      "epoch": 0.15616104107360715,
      "grad_norm": 2.983351707458496,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.0697,
      "step": 192
    },
    {
      "epoch": 0.15697437982919887,
      "grad_norm": 3.316290855407715,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1585,
      "step": 193
    },
    {
      "epoch": 0.15778771858479057,
      "grad_norm": 3.6211719512939453,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.0002,
      "step": 194
    },
    {
      "epoch": 0.15860105734038227,
      "grad_norm": 4.084677219390869,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.083,
      "step": 195
    },
    {
      "epoch": 0.15941439609597396,
      "grad_norm": 3.7958171367645264,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2549,
      "step": 196
    },
    {
      "epoch": 0.1602277348515657,
      "grad_norm": 4.163863658905029,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.8249,
      "step": 197
    },
    {
      "epoch": 0.1610410736071574,
      "grad_norm": 4.53021764755249,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.8578,
      "step": 198
    },
    {
      "epoch": 0.16185441236274908,
      "grad_norm": 4.715237140655518,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.8474,
      "step": 199
    },
    {
      "epoch": 0.16266775111834078,
      "grad_norm": 5.807979583740234,
      "learning_rate": 0.0,
      "loss": 1.0151,
      "step": 200
    },
    {
      "epoch": 0.16266775111834078,
      "eval_loss": 1.1142469644546509,
      "eval_runtime": 106.348,
      "eval_samples_per_second": 19.474,
      "eval_steps_per_second": 4.871,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.252687475638272e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}