{
  "best_metric": 0.6220175623893738,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7490636704119851,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003745318352059925,
      "grad_norm": 8.697524070739746,
      "learning_rate": 1e-05,
      "loss": 14.0732,
      "step": 1
    },
    {
      "epoch": 0.003745318352059925,
      "eval_loss": 2.651507616043091,
      "eval_runtime": 49.6228,
      "eval_samples_per_second": 9.068,
      "eval_steps_per_second": 2.277,
      "step": 1
    },
    {
      "epoch": 0.00749063670411985,
      "grad_norm": 7.445273399353027,
      "learning_rate": 2e-05,
      "loss": 11.2916,
      "step": 2
    },
    {
      "epoch": 0.011235955056179775,
      "grad_norm": 6.031841278076172,
      "learning_rate": 3e-05,
      "loss": 10.6298,
      "step": 3
    },
    {
      "epoch": 0.0149812734082397,
      "grad_norm": 5.524653911590576,
      "learning_rate": 4e-05,
      "loss": 9.208,
      "step": 4
    },
    {
      "epoch": 0.018726591760299626,
      "grad_norm": 5.561168670654297,
      "learning_rate": 5e-05,
      "loss": 8.982,
      "step": 5
    },
    {
      "epoch": 0.02247191011235955,
      "grad_norm": 4.357397079467773,
      "learning_rate": 6e-05,
      "loss": 8.9965,
      "step": 6
    },
    {
      "epoch": 0.026217228464419477,
      "grad_norm": 3.560116767883301,
      "learning_rate": 7e-05,
      "loss": 8.6785,
      "step": 7
    },
    {
      "epoch": 0.0299625468164794,
      "grad_norm": 3.265683174133301,
      "learning_rate": 8e-05,
      "loss": 8.0189,
      "step": 8
    },
    {
      "epoch": 0.033707865168539325,
      "grad_norm": 3.1961605548858643,
      "learning_rate": 9e-05,
      "loss": 7.1049,
      "step": 9
    },
    {
      "epoch": 0.03745318352059925,
      "grad_norm": 2.969959259033203,
      "learning_rate": 0.0001,
      "loss": 7.2922,
      "step": 10
    },
    {
      "epoch": 0.04119850187265917,
      "grad_norm": 2.6740808486938477,
      "learning_rate": 9.999316524962345e-05,
      "loss": 7.2505,
      "step": 11
    },
    {
      "epoch": 0.0449438202247191,
      "grad_norm": 3.099911689758301,
      "learning_rate": 9.997266286704631e-05,
      "loss": 7.1424,
      "step": 12
    },
    {
      "epoch": 0.04868913857677903,
      "grad_norm": 3.156374216079712,
      "learning_rate": 9.993849845741524e-05,
      "loss": 5.9513,
      "step": 13
    },
    {
      "epoch": 0.052434456928838954,
      "grad_norm": 3.512418031692505,
      "learning_rate": 9.989068136093873e-05,
      "loss": 6.387,
      "step": 14
    },
    {
      "epoch": 0.056179775280898875,
      "grad_norm": 2.694615125656128,
      "learning_rate": 9.98292246503335e-05,
      "loss": 5.8994,
      "step": 15
    },
    {
      "epoch": 0.0599250936329588,
      "grad_norm": 2.4905474185943604,
      "learning_rate": 9.975414512725057e-05,
      "loss": 5.7442,
      "step": 16
    },
    {
      "epoch": 0.06367041198501873,
      "grad_norm": 2.8079071044921875,
      "learning_rate": 9.966546331768191e-05,
      "loss": 6.0073,
      "step": 17
    },
    {
      "epoch": 0.06741573033707865,
      "grad_norm": 2.9026036262512207,
      "learning_rate": 9.956320346634876e-05,
      "loss": 5.0291,
      "step": 18
    },
    {
      "epoch": 0.07116104868913857,
      "grad_norm": 2.983949661254883,
      "learning_rate": 9.944739353007344e-05,
      "loss": 5.736,
      "step": 19
    },
    {
      "epoch": 0.0749063670411985,
      "grad_norm": 3.0429649353027344,
      "learning_rate": 9.931806517013612e-05,
      "loss": 5.3121,
      "step": 20
    },
    {
      "epoch": 0.07865168539325842,
      "grad_norm": 3.181403636932373,
      "learning_rate": 9.917525374361912e-05,
      "loss": 6.0976,
      "step": 21
    },
    {
      "epoch": 0.08239700374531835,
      "grad_norm": 3.222288131713867,
      "learning_rate": 9.901899829374047e-05,
      "loss": 5.933,
      "step": 22
    },
    {
      "epoch": 0.08614232209737828,
      "grad_norm": 3.0801918506622314,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.9227,
      "step": 23
    },
    {
      "epoch": 0.0898876404494382,
      "grad_norm": 2.950737237930298,
      "learning_rate": 9.86663298624003e-05,
      "loss": 5.5375,
      "step": 24
    },
    {
      "epoch": 0.09363295880149813,
      "grad_norm": 2.900657892227173,
      "learning_rate": 9.847001329696653e-05,
      "loss": 5.0897,
      "step": 25
    },
    {
      "epoch": 0.09737827715355805,
      "grad_norm": 2.8506317138671875,
      "learning_rate": 9.826044551386744e-05,
      "loss": 4.9127,
      "step": 26
    },
    {
      "epoch": 0.10112359550561797,
      "grad_norm": 3.210683584213257,
      "learning_rate": 9.803768380684242e-05,
      "loss": 5.4676,
      "step": 27
    },
    {
      "epoch": 0.10486891385767791,
      "grad_norm": 3.3829996585845947,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.9145,
      "step": 28
    },
    {
      "epoch": 0.10861423220973783,
      "grad_norm": 3.208481550216675,
      "learning_rate": 9.755282581475769e-05,
      "loss": 5.6933,
      "step": 29
    },
    {
      "epoch": 0.11235955056179775,
      "grad_norm": 3.634627342224121,
      "learning_rate": 9.729086208503174e-05,
      "loss": 5.9477,
      "step": 30
    },
    {
      "epoch": 0.11610486891385768,
      "grad_norm": 3.572483539581299,
      "learning_rate": 9.701596950580806e-05,
      "loss": 5.9688,
      "step": 31
    },
    {
      "epoch": 0.1198501872659176,
      "grad_norm": 3.8514907360076904,
      "learning_rate": 9.672822322997305e-05,
      "loss": 5.0967,
      "step": 32
    },
    {
      "epoch": 0.12359550561797752,
      "grad_norm": 3.564493179321289,
      "learning_rate": 9.642770192448536e-05,
      "loss": 5.0059,
      "step": 33
    },
    {
      "epoch": 0.12734082397003746,
      "grad_norm": 3.7451579570770264,
      "learning_rate": 9.611448774886924e-05,
      "loss": 5.196,
      "step": 34
    },
    {
      "epoch": 0.13108614232209737,
      "grad_norm": 3.6254284381866455,
      "learning_rate": 9.578866633275288e-05,
      "loss": 4.4033,
      "step": 35
    },
    {
      "epoch": 0.1348314606741573,
      "grad_norm": 3.5597217082977295,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.5404,
      "step": 36
    },
    {
      "epoch": 0.13857677902621723,
      "grad_norm": 3.6213245391845703,
      "learning_rate": 9.509956150664796e-05,
      "loss": 4.9233,
      "step": 37
    },
    {
      "epoch": 0.14232209737827714,
      "grad_norm": 3.888190984725952,
      "learning_rate": 9.473646649103818e-05,
      "loss": 5.6054,
      "step": 38
    },
    {
      "epoch": 0.14606741573033707,
      "grad_norm": 3.6787173748016357,
      "learning_rate": 9.43611409721806e-05,
      "loss": 5.1674,
      "step": 39
    },
    {
      "epoch": 0.149812734082397,
      "grad_norm": 4.078162670135498,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.5943,
      "step": 40
    },
    {
      "epoch": 0.15355805243445692,
      "grad_norm": 3.706515312194824,
      "learning_rate": 9.357421218136386e-05,
      "loss": 4.8016,
      "step": 41
    },
    {
      "epoch": 0.15730337078651685,
      "grad_norm": 3.4145514965057373,
      "learning_rate": 9.316282404787871e-05,
      "loss": 4.4477,
      "step": 42
    },
    {
      "epoch": 0.16104868913857678,
      "grad_norm": 4.094817638397217,
      "learning_rate": 9.273963562927695e-05,
      "loss": 5.1288,
      "step": 43
    },
    {
      "epoch": 0.1647940074906367,
      "grad_norm": 4.14848518371582,
      "learning_rate": 9.230476262104677e-05,
      "loss": 4.9813,
      "step": 44
    },
    {
      "epoch": 0.16853932584269662,
      "grad_norm": 4.413974761962891,
      "learning_rate": 9.185832391312644e-05,
      "loss": 5.9782,
      "step": 45
    },
    {
      "epoch": 0.17228464419475656,
      "grad_norm": 4.017249584197998,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.7721,
      "step": 46
    },
    {
      "epoch": 0.1760299625468165,
      "grad_norm": 4.1360087394714355,
      "learning_rate": 9.093124073433463e-05,
      "loss": 4.3844,
      "step": 47
    },
    {
      "epoch": 0.1797752808988764,
      "grad_norm": 4.620957851409912,
      "learning_rate": 9.045084971874738e-05,
      "loss": 5.6895,
      "step": 48
    },
    {
      "epoch": 0.18352059925093633,
      "grad_norm": 5.1671037673950195,
      "learning_rate": 8.995939984474624e-05,
      "loss": 5.8143,
      "step": 49
    },
    {
      "epoch": 0.18726591760299627,
      "grad_norm": 6.653675556182861,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.6729,
      "step": 50
    },
    {
      "epoch": 0.18726591760299627,
      "eval_loss": 1.087544560432434,
      "eval_runtime": 50.3565,
      "eval_samples_per_second": 8.936,
      "eval_steps_per_second": 2.244,
      "step": 50
    },
    {
      "epoch": 0.19101123595505617,
      "grad_norm": 5.07515811920166,
      "learning_rate": 8.894386393810563e-05,
      "loss": 5.8781,
      "step": 51
    },
    {
      "epoch": 0.1947565543071161,
      "grad_norm": 4.434817314147949,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.0416,
      "step": 52
    },
    {
      "epoch": 0.19850187265917604,
      "grad_norm": 3.447234869003296,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.9345,
      "step": 53
    },
    {
      "epoch": 0.20224719101123595,
      "grad_norm": 3.0183892250061035,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.459,
      "step": 54
    },
    {
      "epoch": 0.20599250936329588,
      "grad_norm": 3.0148396492004395,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.624,
      "step": 55
    },
    {
      "epoch": 0.20973782771535582,
      "grad_norm": 3.411334276199341,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.4328,
      "step": 56
    },
    {
      "epoch": 0.21348314606741572,
      "grad_norm": 3.321749210357666,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.867,
      "step": 57
    },
    {
      "epoch": 0.21722846441947566,
      "grad_norm": 3.3974671363830566,
      "learning_rate": 8.506183921362443e-05,
      "loss": 4.0606,
      "step": 58
    },
    {
      "epoch": 0.2209737827715356,
      "grad_norm": 3.2373404502868652,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.5911,
      "step": 59
    },
    {
      "epoch": 0.2247191011235955,
      "grad_norm": 3.658618927001953,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.7907,
      "step": 60
    },
    {
      "epoch": 0.22846441947565543,
      "grad_norm": 3.3463518619537354,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.6921,
      "step": 61
    },
    {
      "epoch": 0.23220973782771537,
      "grad_norm": 3.0975143909454346,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.3162,
      "step": 62
    },
    {
      "epoch": 0.23595505617977527,
      "grad_norm": 3.098482847213745,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.6433,
      "step": 63
    },
    {
      "epoch": 0.2397003745318352,
      "grad_norm": 13.961381912231445,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.59,
      "step": 64
    },
    {
      "epoch": 0.24344569288389514,
      "grad_norm": 2.997326135635376,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.1784,
      "step": 65
    },
    {
      "epoch": 0.24719101123595505,
      "grad_norm": 2.758126735687256,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.7323,
      "step": 66
    },
    {
      "epoch": 0.250936329588015,
      "grad_norm": 3.1518728733062744,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.8775,
      "step": 67
    },
    {
      "epoch": 0.2546816479400749,
      "grad_norm": 3.343369245529175,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.5882,
      "step": 68
    },
    {
      "epoch": 0.25842696629213485,
      "grad_norm": 3.42604923248291,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.5047,
      "step": 69
    },
    {
      "epoch": 0.26217228464419473,
      "grad_norm": 3.407392740249634,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.5206,
      "step": 70
    },
    {
      "epoch": 0.26591760299625467,
      "grad_norm": 3.537952184677124,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.4102,
      "step": 71
    },
    {
      "epoch": 0.2696629213483146,
      "grad_norm": 3.030775547027588,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.7486,
      "step": 72
    },
    {
      "epoch": 0.27340823970037453,
      "grad_norm": 3.636820077896118,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.6567,
      "step": 73
    },
    {
      "epoch": 0.27715355805243447,
      "grad_norm": 3.5621426105499268,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.5698,
      "step": 74
    },
    {
      "epoch": 0.2808988764044944,
      "grad_norm": 3.928917169570923,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.847,
      "step": 75
    },
    {
      "epoch": 0.2846441947565543,
      "grad_norm": 4.136871814727783,
      "learning_rate": 7.30670581489344e-05,
      "loss": 4.0049,
      "step": 76
    },
    {
      "epoch": 0.2883895131086142,
      "grad_norm": 3.693014621734619,
      "learning_rate": 7.233044034264034e-05,
      "loss": 3.3662,
      "step": 77
    },
    {
      "epoch": 0.29213483146067415,
      "grad_norm": 3.9544732570648193,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.406,
      "step": 78
    },
    {
      "epoch": 0.2958801498127341,
      "grad_norm": 4.548867702484131,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.3458,
      "step": 79
    },
    {
      "epoch": 0.299625468164794,
      "grad_norm": 3.996755361557007,
      "learning_rate": 7.008477123264848e-05,
      "loss": 4.2286,
      "step": 80
    },
    {
      "epoch": 0.30337078651685395,
      "grad_norm": 4.559913158416748,
      "learning_rate": 6.932495846462261e-05,
      "loss": 4.0957,
      "step": 81
    },
    {
      "epoch": 0.30711610486891383,
      "grad_norm": 3.9077858924865723,
      "learning_rate": 6.855986244591104e-05,
      "loss": 3.9683,
      "step": 82
    },
    {
      "epoch": 0.31086142322097376,
      "grad_norm": 4.147164821624756,
      "learning_rate": 6.778969234612584e-05,
      "loss": 4.361,
      "step": 83
    },
    {
      "epoch": 0.3146067415730337,
      "grad_norm": 3.4458258152008057,
      "learning_rate": 6.701465872208216e-05,
      "loss": 3.8915,
      "step": 84
    },
    {
      "epoch": 0.31835205992509363,
      "grad_norm": 3.4497954845428467,
      "learning_rate": 6.623497346023418e-05,
      "loss": 3.393,
      "step": 85
    },
    {
      "epoch": 0.32209737827715357,
      "grad_norm": 3.8256044387817383,
      "learning_rate": 6.545084971874738e-05,
      "loss": 3.653,
      "step": 86
    },
    {
      "epoch": 0.3258426966292135,
      "grad_norm": 3.2304205894470215,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.5072,
      "step": 87
    },
    {
      "epoch": 0.3295880149812734,
      "grad_norm": 3.710836410522461,
      "learning_rate": 6.387014543809223e-05,
      "loss": 3.564,
      "step": 88
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 3.6595144271850586,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.4068,
      "step": 89
    },
    {
      "epoch": 0.33707865168539325,
      "grad_norm": 4.20240592956543,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.707,
      "step": 90
    },
    {
      "epoch": 0.3408239700374532,
      "grad_norm": 4.653726577758789,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.2001,
      "step": 91
    },
    {
      "epoch": 0.3445692883895131,
      "grad_norm": 4.151357173919678,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.5171,
      "step": 92
    },
    {
      "epoch": 0.34831460674157305,
      "grad_norm": 4.105644702911377,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.005,
      "step": 93
    },
    {
      "epoch": 0.352059925093633,
      "grad_norm": 4.723206520080566,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.6565,
      "step": 94
    },
    {
      "epoch": 0.35580524344569286,
      "grad_norm": 4.155903339385986,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.8526,
      "step": 95
    },
    {
      "epoch": 0.3595505617977528,
      "grad_norm": 4.217616558074951,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.4657,
      "step": 96
    },
    {
      "epoch": 0.36329588014981273,
      "grad_norm": 5.101412773132324,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.0445,
      "step": 97
    },
    {
      "epoch": 0.36704119850187267,
      "grad_norm": 5.136566638946533,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.7391,
      "step": 98
    },
    {
      "epoch": 0.3707865168539326,
      "grad_norm": 4.774531841278076,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.079,
      "step": 99
    },
    {
      "epoch": 0.37453183520599254,
      "grad_norm": 6.721903324127197,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.4328,
      "step": 100
    },
    {
      "epoch": 0.37453183520599254,
      "eval_loss": 0.7906140089035034,
      "eval_runtime": 50.35,
      "eval_samples_per_second": 8.937,
      "eval_steps_per_second": 2.244,
      "step": 100
    },
    {
      "epoch": 0.3782771535580524,
      "grad_norm": 5.149799346923828,
      "learning_rate": 5.330452921628497e-05,
      "loss": 5.393,
      "step": 101
    },
    {
      "epoch": 0.38202247191011235,
      "grad_norm": 4.442603588104248,
      "learning_rate": 5.247918773366112e-05,
      "loss": 4.1819,
      "step": 102
    },
    {
      "epoch": 0.3857677902621723,
      "grad_norm": 3.451070547103882,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.0308,
      "step": 103
    },
    {
      "epoch": 0.3895131086142322,
      "grad_norm": 3.622688055038452,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.1119,
      "step": 104
    },
    {
      "epoch": 0.39325842696629215,
      "grad_norm": 3.53656268119812,
      "learning_rate": 5e-05,
      "loss": 3.5242,
      "step": 105
    },
    {
      "epoch": 0.3970037453183521,
      "grad_norm": 3.480729341506958,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.6205,
      "step": 106
    },
    {
      "epoch": 0.40074906367041196,
      "grad_norm": 3.4890174865722656,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.0373,
      "step": 107
    },
    {
      "epoch": 0.4044943820224719,
      "grad_norm": 3.736863613128662,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.123,
      "step": 108
    },
    {
      "epoch": 0.40823970037453183,
      "grad_norm": 3.0392906665802,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.3677,
      "step": 109
    },
    {
      "epoch": 0.41198501872659177,
      "grad_norm": 3.2078053951263428,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.5606,
      "step": 110
    },
    {
      "epoch": 0.4157303370786517,
      "grad_norm": 3.771505832672119,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.7557,
      "step": 111
    },
    {
      "epoch": 0.41947565543071164,
      "grad_norm": 3.1648523807525635,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.2478,
      "step": 112
    },
    {
      "epoch": 0.4232209737827715,
      "grad_norm": 3.3156416416168213,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.4345,
      "step": 113
    },
    {
      "epoch": 0.42696629213483145,
      "grad_norm": 3.7513787746429443,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.1586,
      "step": 114
    },
    {
      "epoch": 0.4307116104868914,
      "grad_norm": 3.2219436168670654,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.6671,
      "step": 115
    },
    {
      "epoch": 0.4344569288389513,
      "grad_norm": 3.214806318283081,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.1667,
      "step": 116
    },
    {
      "epoch": 0.43820224719101125,
      "grad_norm": 3.6060116291046143,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.8972,
      "step": 117
    },
    {
      "epoch": 0.4419475655430712,
      "grad_norm": 3.3740694522857666,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.2201,
      "step": 118
    },
    {
      "epoch": 0.44569288389513106,
      "grad_norm": 3.1603713035583496,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.3407,
      "step": 119
    },
    {
      "epoch": 0.449438202247191,
      "grad_norm": 3.669546365737915,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.8301,
      "step": 120
    },
    {
      "epoch": 0.45318352059925093,
      "grad_norm": 3.3857712745666504,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.1229,
      "step": 121
    },
    {
      "epoch": 0.45692883895131087,
      "grad_norm": 3.687445640563965,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.3564,
      "step": 122
    },
    {
      "epoch": 0.4606741573033708,
      "grad_norm": 3.683568000793457,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.5398,
      "step": 123
    },
    {
      "epoch": 0.46441947565543074,
      "grad_norm": 3.4279611110687256,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.5579,
      "step": 124
    },
    {
      "epoch": 0.4681647940074906,
      "grad_norm": 4.093925952911377,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.7926,
      "step": 125
    },
    {
      "epoch": 0.47191011235955055,
      "grad_norm": 3.813743829727173,
      "learning_rate": 3.298534127791785e-05,
      "loss": 3.0301,
      "step": 126
    },
    {
      "epoch": 0.4756554307116105,
      "grad_norm": 4.181626796722412,
      "learning_rate": 3.221030765387417e-05,
      "loss": 3.0427,
      "step": 127
    },
    {
      "epoch": 0.4794007490636704,
      "grad_norm": 3.6809024810791016,
      "learning_rate": 3.144013755408895e-05,
      "loss": 3.0097,
      "step": 128
    },
    {
      "epoch": 0.48314606741573035,
      "grad_norm": 3.9218976497650146,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.836,
      "step": 129
    },
    {
      "epoch": 0.4868913857677903,
      "grad_norm": 4.02546501159668,
      "learning_rate": 2.991522876735154e-05,
      "loss": 3.0594,
      "step": 130
    },
    {
      "epoch": 0.49063670411985016,
      "grad_norm": 4.353425979614258,
      "learning_rate": 2.916090697523549e-05,
      "loss": 4.0435,
      "step": 131
    },
    {
      "epoch": 0.4943820224719101,
      "grad_norm": 3.7416892051696777,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.1662,
      "step": 132
    },
    {
      "epoch": 0.49812734082397003,
      "grad_norm": 4.098910808563232,
      "learning_rate": 2.766955965735968e-05,
      "loss": 3.5148,
      "step": 133
    },
    {
      "epoch": 0.50187265917603,
      "grad_norm": 4.342579364776611,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.5424,
      "step": 134
    },
    {
      "epoch": 0.5056179775280899,
      "grad_norm": 3.517526149749756,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.7625,
      "step": 135
    },
    {
      "epoch": 0.5093632958801498,
      "grad_norm": 3.701322555541992,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.9186,
      "step": 136
    },
    {
      "epoch": 0.5131086142322098,
      "grad_norm": 3.577916145324707,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.2649,
      "step": 137
    },
    {
      "epoch": 0.5168539325842697,
      "grad_norm": 4.3659186363220215,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.7109,
      "step": 138
    },
    {
      "epoch": 0.5205992509363296,
      "grad_norm": 4.333901405334473,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.8619,
      "step": 139
    },
    {
      "epoch": 0.5243445692883895,
      "grad_norm": 4.306787014007568,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.8832,
      "step": 140
    },
    {
      "epoch": 0.5280898876404494,
      "grad_norm": 4.086246013641357,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.4503,
      "step": 141
    },
    {
      "epoch": 0.5318352059925093,
      "grad_norm": 3.8691341876983643,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.7464,
      "step": 142
    },
    {
      "epoch": 0.5355805243445693,
      "grad_norm": 4.125598430633545,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.6973,
      "step": 143
    },
    {
      "epoch": 0.5393258426966292,
      "grad_norm": 5.1870012283325195,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.2336,
      "step": 144
    },
    {
      "epoch": 0.5430711610486891,
      "grad_norm": 4.458269119262695,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.8274,
      "step": 145
    },
    {
      "epoch": 0.5468164794007491,
      "grad_norm": 4.78695821762085,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.5594,
      "step": 146
    },
    {
      "epoch": 0.550561797752809,
      "grad_norm": 5.0786237716674805,
      "learning_rate": 1.800157297483417e-05,
      "loss": 3.3043,
      "step": 147
    },
    {
      "epoch": 0.5543071161048689,
      "grad_norm": 4.59190034866333,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.7787,
      "step": 148
    },
    {
      "epoch": 0.5580524344569289,
      "grad_norm": 5.277339935302734,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.9477,
      "step": 149
    },
    {
      "epoch": 0.5617977528089888,
      "grad_norm": 6.670591354370117,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 3.526,
      "step": 150
    },
    {
      "epoch": 0.5617977528089888,
      "eval_loss": 0.654658854007721,
      "eval_runtime": 50.4118,
      "eval_samples_per_second": 8.926,
      "eval_steps_per_second": 2.242,
      "step": 150
    },
    {
      "epoch": 0.5655430711610487,
      "grad_norm": 4.878920078277588,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.8769,
      "step": 151
    },
    {
      "epoch": 0.5692883895131086,
      "grad_norm": 4.279634475708008,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.862,
      "step": 152
    },
    {
      "epoch": 0.5730337078651685,
      "grad_norm": 4.4864068031311035,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.1875,
      "step": 153
    },
    {
      "epoch": 0.5767790262172284,
      "grad_norm": 3.6915926933288574,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.159,
      "step": 154
    },
    {
      "epoch": 0.5805243445692884,
      "grad_norm": 3.60880708694458,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.901,
      "step": 155
    },
    {
      "epoch": 0.5842696629213483,
      "grad_norm": 3.513319492340088,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.0896,
      "step": 156
    },
    {
      "epoch": 0.5880149812734082,
      "grad_norm": 3.632713556289673,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.3247,
      "step": 157
    },
    {
      "epoch": 0.5917602996254682,
      "grad_norm": 3.6750118732452393,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.954,
      "step": 158
    },
    {
      "epoch": 0.5955056179775281,
      "grad_norm": 4.3678460121154785,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.7798,
      "step": 159
    },
    {
      "epoch": 0.599250936329588,
      "grad_norm": 3.3913145065307617,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.6748,
      "step": 160
    },
    {
      "epoch": 0.602996254681648,
      "grad_norm": 3.237217426300049,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.0347,
      "step": 161
    },
    {
      "epoch": 0.6067415730337079,
      "grad_norm": 3.258819580078125,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.037,
      "step": 162
    },
    {
      "epoch": 0.6104868913857678,
      "grad_norm": 3.54426908493042,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.1275,
      "step": 163
    },
    {
      "epoch": 0.6142322097378277,
      "grad_norm": 3.9705655574798584,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.7727,
      "step": 164
    },
    {
      "epoch": 0.6179775280898876,
      "grad_norm": 3.568998098373413,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.1715,
      "step": 165
    },
    {
      "epoch": 0.6217228464419475,
      "grad_norm": 3.6433839797973633,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.8851,
      "step": 166
    },
    {
      "epoch": 0.6254681647940075,
      "grad_norm": 3.701789379119873,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.5972,
      "step": 167
    },
    {
      "epoch": 0.6292134831460674,
      "grad_norm": 3.8505699634552,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.2329,
      "step": 168
    },
    {
      "epoch": 0.6329588014981273,
      "grad_norm": 4.603046417236328,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.3493,
      "step": 169
    },
    {
      "epoch": 0.6367041198501873,
      "grad_norm": 3.5844812393188477,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.6272,
      "step": 170
    },
    {
      "epoch": 0.6404494382022472,
      "grad_norm": 4.309713363647461,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.1115,
      "step": 171
    },
    {
      "epoch": 0.6441947565543071,
      "grad_norm": 3.8912484645843506,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.8015,
      "step": 172
    },
    {
      "epoch": 0.6479400749063671,
      "grad_norm": 3.424530029296875,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.9437,
      "step": 173
    },
    {
      "epoch": 0.651685393258427,
      "grad_norm": 3.795205593109131,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.5414,
      "step": 174
    },
    {
      "epoch": 0.6554307116104869,
      "grad_norm": 4.100245475769043,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.7784,
      "step": 175
    },
    {
      "epoch": 0.6591760299625468,
      "grad_norm": 3.8642256259918213,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.7422,
      "step": 176
    },
    {
      "epoch": 0.6629213483146067,
      "grad_norm": 4.408370018005371,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.6459,
      "step": 177
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 3.917318820953369,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.6922,
      "step": 178
    },
    {
      "epoch": 0.6704119850187266,
      "grad_norm": 4.09055233001709,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.4396,
      "step": 179
    },
    {
      "epoch": 0.6741573033707865,
      "grad_norm": 3.988201856613159,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.2676,
      "step": 180
    },
    {
      "epoch": 0.6779026217228464,
      "grad_norm": 4.362141132354736,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.6572,
      "step": 181
    },
    {
      "epoch": 0.6816479400749064,
      "grad_norm": 4.314810752868652,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.6495,
      "step": 182
    },
    {
      "epoch": 0.6853932584269663,
      "grad_norm": 4.31600284576416,
      "learning_rate": 1.962316193157593e-06,
      "loss": 3.0478,
      "step": 183
    },
    {
      "epoch": 0.6891385767790262,
      "grad_norm": 4.994217395782471,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 3.0878,
      "step": 184
    },
    {
      "epoch": 0.6928838951310862,
      "grad_norm": 4.858488082885742,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.7153,
      "step": 185
    },
    {
      "epoch": 0.6966292134831461,
      "grad_norm": 4.0731282234191895,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.3293,
      "step": 186
    },
    {
      "epoch": 0.700374531835206,
      "grad_norm": 4.403578281402588,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.5422,
      "step": 187
    },
    {
      "epoch": 0.704119850187266,
      "grad_norm": 3.9899237155914307,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.5385,
      "step": 188
    },
    {
      "epoch": 0.7078651685393258,
      "grad_norm": 4.005816459655762,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.7145,
      "step": 189
    },
    {
      "epoch": 0.7116104868913857,
      "grad_norm": 4.402023792266846,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.4821,
      "step": 190
    },
    {
      "epoch": 0.7153558052434457,
      "grad_norm": 4.434578895568848,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.6539,
      "step": 191
    },
    {
      "epoch": 0.7191011235955056,
      "grad_norm": 4.9623799324035645,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.2877,
      "step": 192
    },
    {
      "epoch": 0.7228464419475655,
      "grad_norm": 4.215279579162598,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.3972,
      "step": 193
    },
    {
      "epoch": 0.7265917602996255,
      "grad_norm": 3.851797342300415,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.8241,
      "step": 194
    },
    {
      "epoch": 0.7303370786516854,
      "grad_norm": 4.335139751434326,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.2999,
      "step": 195
    },
    {
      "epoch": 0.7340823970037453,
      "grad_norm": 5.122757911682129,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.759,
      "step": 196
    },
    {
      "epoch": 0.7378277153558053,
      "grad_norm": 4.752044200897217,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.521,
      "step": 197
    },
    {
      "epoch": 0.7415730337078652,
      "grad_norm": 5.508137226104736,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.6003,
      "step": 198
    },
    {
      "epoch": 0.7453183520599251,
      "grad_norm": 5.384343147277832,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.5857,
      "step": 199
    },
    {
      "epoch": 0.7490636704119851,
      "grad_norm": 6.057183742523193,
      "learning_rate": 0.0,
      "loss": 3.038,
      "step": 200
    },
    {
      "epoch": 0.7490636704119851,
      "eval_loss": 0.6220175623893738,
      "eval_runtime": 50.3078,
      "eval_samples_per_second": 8.945,
      "eval_steps_per_second": 2.246,
      "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": 5.610833556013056e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}