{
  "best_metric": 1.5319730043411255,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 0.5819592628516004,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0038797284190106693,
      "grad_norm": 5.651600360870361,
      "learning_rate": 1e-05,
      "loss": 2.4589,
      "step": 1
    },
    {
      "epoch": 0.0038797284190106693,
      "eval_loss": 2.5541110038757324,
      "eval_runtime": 43.4534,
      "eval_samples_per_second": 9.988,
      "eval_steps_per_second": 2.508,
      "step": 1
    },
    {
      "epoch": 0.007759456838021339,
      "grad_norm": 5.713309288024902,
      "learning_rate": 2e-05,
      "loss": 2.489,
      "step": 2
    },
    {
      "epoch": 0.011639185257032008,
      "grad_norm": 4.894677639007568,
      "learning_rate": 3e-05,
      "loss": 2.3998,
      "step": 3
    },
    {
      "epoch": 0.015518913676042677,
      "grad_norm": 2.3635458946228027,
      "learning_rate": 4e-05,
      "loss": 2.237,
      "step": 4
    },
    {
      "epoch": 0.019398642095053348,
      "grad_norm": 3.661370277404785,
      "learning_rate": 5e-05,
      "loss": 2.186,
      "step": 5
    },
    {
      "epoch": 0.023278370514064017,
      "grad_norm": 2.6354176998138428,
      "learning_rate": 6e-05,
      "loss": 2.1989,
      "step": 6
    },
    {
      "epoch": 0.027158098933074686,
      "grad_norm": 1.7957730293273926,
      "learning_rate": 7e-05,
      "loss": 2.0822,
      "step": 7
    },
    {
      "epoch": 0.031037827352085354,
      "grad_norm": 1.3487071990966797,
      "learning_rate": 8e-05,
      "loss": 2.0277,
      "step": 8
    },
    {
      "epoch": 0.03491755577109602,
      "grad_norm": 1.0389519929885864,
      "learning_rate": 9e-05,
      "loss": 2.0022,
      "step": 9
    },
    {
      "epoch": 0.038797284190106696,
      "grad_norm": 1.1190440654754639,
      "learning_rate": 0.0001,
      "loss": 2.0215,
      "step": 10
    },
    {
      "epoch": 0.04267701260911736,
      "grad_norm": 1.0835639238357544,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.006,
      "step": 11
    },
    {
      "epoch": 0.04655674102812803,
      "grad_norm": 0.8974905610084534,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.9841,
      "step": 12
    },
    {
      "epoch": 0.0504364694471387,
      "grad_norm": 1.0189319849014282,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.9822,
      "step": 13
    },
    {
      "epoch": 0.05431619786614937,
      "grad_norm": 0.9930517673492432,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.9002,
      "step": 14
    },
    {
      "epoch": 0.058195926285160036,
      "grad_norm": 0.8625919222831726,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.8385,
      "step": 15
    },
    {
      "epoch": 0.06207565470417071,
      "grad_norm": 0.8136669993400574,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.8936,
      "step": 16
    },
    {
      "epoch": 0.06595538312318137,
      "grad_norm": 0.851311981678009,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.8567,
      "step": 17
    },
    {
      "epoch": 0.06983511154219205,
      "grad_norm": 0.8693639636039734,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.778,
      "step": 18
    },
    {
      "epoch": 0.07371483996120272,
      "grad_norm": 0.9974572062492371,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.8108,
      "step": 19
    },
    {
      "epoch": 0.07759456838021339,
      "grad_norm": 0.8566578030586243,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6716,
      "step": 20
    },
    {
      "epoch": 0.08147429679922405,
      "grad_norm": 0.9674069881439209,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.8065,
      "step": 21
    },
    {
      "epoch": 0.08535402521823472,
      "grad_norm": 0.855084240436554,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.7205,
      "step": 22
    },
    {
      "epoch": 0.0892337536372454,
      "grad_norm": 1.0447219610214233,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.6963,
      "step": 23
    },
    {
      "epoch": 0.09311348205625607,
      "grad_norm": 0.9063634276390076,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.6128,
      "step": 24
    },
    {
      "epoch": 0.09699321047526673,
      "grad_norm": 1.1183236837387085,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6235,
      "step": 25
    },
    {
      "epoch": 0.1008729388942774,
      "grad_norm": 0.9686721563339233,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.65,
      "step": 26
    },
    {
      "epoch": 0.10475266731328807,
      "grad_norm": 1.0887641906738281,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.648,
      "step": 27
    },
    {
      "epoch": 0.10863239573229874,
      "grad_norm": 1.1189512014389038,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.6199,
      "step": 28
    },
    {
      "epoch": 0.11251212415130941,
      "grad_norm": 0.991513729095459,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.6207,
      "step": 29
    },
    {
      "epoch": 0.11639185257032007,
      "grad_norm": 1.048596739768982,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.5783,
      "step": 30
    },
    {
      "epoch": 0.12027158098933075,
      "grad_norm": 1.0062381029129028,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.5489,
      "step": 31
    },
    {
      "epoch": 0.12415130940834142,
      "grad_norm": 1.0828354358673096,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.6415,
      "step": 32
    },
    {
      "epoch": 0.1280310378273521,
      "grad_norm": 1.0989104509353638,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.5606,
      "step": 33
    },
    {
      "epoch": 0.13191076624636275,
      "grad_norm": 1.047166109085083,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5962,
      "step": 34
    },
    {
      "epoch": 0.13579049466537343,
      "grad_norm": 1.0102217197418213,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.6675,
      "step": 35
    },
    {
      "epoch": 0.1396702230843841,
      "grad_norm": 1.0507794618606567,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5584,
      "step": 36
    },
    {
      "epoch": 0.14354995150339475,
      "grad_norm": 1.0515727996826172,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.5455,
      "step": 37
    },
    {
      "epoch": 0.14742967992240544,
      "grad_norm": 1.0033230781555176,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.4866,
      "step": 38
    },
    {
      "epoch": 0.1513094083414161,
      "grad_norm": 1.0882160663604736,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.3921,
      "step": 39
    },
    {
      "epoch": 0.15518913676042678,
      "grad_norm": 1.2042813301086426,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4956,
      "step": 40
    },
    {
      "epoch": 0.15906886517943744,
      "grad_norm": 1.1519536972045898,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.4994,
      "step": 41
    },
    {
      "epoch": 0.1629485935984481,
      "grad_norm": 1.1808193922042847,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.4839,
      "step": 42
    },
    {
      "epoch": 0.16682832201745879,
      "grad_norm": 1.3587406873703003,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.4351,
      "step": 43
    },
    {
      "epoch": 0.17070805043646944,
      "grad_norm": 1.372000813484192,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5146,
      "step": 44
    },
    {
      "epoch": 0.17458777885548013,
      "grad_norm": 1.3638637065887451,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.5361,
      "step": 45
    },
    {
      "epoch": 0.1784675072744908,
      "grad_norm": 1.4246609210968018,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.514,
      "step": 46
    },
    {
      "epoch": 0.18234723569350145,
      "grad_norm": 1.4770151376724243,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5714,
      "step": 47
    },
    {
      "epoch": 0.18622696411251213,
      "grad_norm": 1.3508543968200684,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5362,
      "step": 48
    },
    {
      "epoch": 0.1901066925315228,
      "grad_norm": 1.4587279558181763,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.449,
      "step": 49
    },
    {
      "epoch": 0.19398642095053345,
      "grad_norm": 2.42429256439209,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.3435,
      "step": 50
    },
    {
      "epoch": 0.19398642095053345,
      "eval_loss": 1.9609380960464478,
      "eval_runtime": 44.0176,
      "eval_samples_per_second": 9.86,
      "eval_steps_per_second": 2.476,
      "step": 50
    },
    {
      "epoch": 0.19786614936954414,
      "grad_norm": 8.905970573425293,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.2064,
      "step": 51
    },
    {
      "epoch": 0.2017458777885548,
      "grad_norm": 2.941683769226074,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.9969,
      "step": 52
    },
    {
      "epoch": 0.20562560620756548,
      "grad_norm": 1.3455346822738647,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.8349,
      "step": 53
    },
    {
      "epoch": 0.20950533462657614,
      "grad_norm": 0.8416768908500671,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.746,
      "step": 54
    },
    {
      "epoch": 0.2133850630455868,
      "grad_norm": 0.6563642024993896,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7586,
      "step": 55
    },
    {
      "epoch": 0.21726479146459748,
      "grad_norm": 0.5815387964248657,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7285,
      "step": 56
    },
    {
      "epoch": 0.22114451988360814,
      "grad_norm": 0.5797914266586304,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.743,
      "step": 57
    },
    {
      "epoch": 0.22502424830261883,
      "grad_norm": 0.5917764902114868,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.7455,
      "step": 58
    },
    {
      "epoch": 0.2289039767216295,
      "grad_norm": 0.6019516587257385,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.7175,
      "step": 59
    },
    {
      "epoch": 0.23278370514064015,
      "grad_norm": 0.5157177448272705,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7389,
      "step": 60
    },
    {
      "epoch": 0.23666343355965083,
      "grad_norm": 0.5754866600036621,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.8373,
      "step": 61
    },
    {
      "epoch": 0.2405431619786615,
      "grad_norm": 0.6096776723861694,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.7706,
      "step": 62
    },
    {
      "epoch": 0.24442289039767218,
      "grad_norm": 0.5166305899620056,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.7316,
      "step": 63
    },
    {
      "epoch": 0.24830261881668284,
      "grad_norm": 0.5555219650268555,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.66,
      "step": 64
    },
    {
      "epoch": 0.2521823472356935,
      "grad_norm": 0.5139245390892029,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6954,
      "step": 65
    },
    {
      "epoch": 0.2560620756547042,
      "grad_norm": 0.5670962333679199,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.7405,
      "step": 66
    },
    {
      "epoch": 0.25994180407371487,
      "grad_norm": 0.5841243863105774,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.7606,
      "step": 67
    },
    {
      "epoch": 0.2638215324927255,
      "grad_norm": 0.5985312461853027,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.6409,
      "step": 68
    },
    {
      "epoch": 0.2677012609117362,
      "grad_norm": 0.622398853302002,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.6712,
      "step": 69
    },
    {
      "epoch": 0.27158098933074687,
      "grad_norm": 0.6333388090133667,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6546,
      "step": 70
    },
    {
      "epoch": 0.2754607177497575,
      "grad_norm": 0.6799418926239014,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5804,
      "step": 71
    },
    {
      "epoch": 0.2793404461687682,
      "grad_norm": 0.7029693722724915,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6598,
      "step": 72
    },
    {
      "epoch": 0.2832201745877789,
      "grad_norm": 0.7402039170265198,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5663,
      "step": 73
    },
    {
      "epoch": 0.2870999030067895,
      "grad_norm": 0.7699214220046997,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.4906,
      "step": 74
    },
    {
      "epoch": 0.2909796314258002,
      "grad_norm": 0.7594525814056396,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4451,
      "step": 75
    },
    {
      "epoch": 0.2948593598448109,
      "grad_norm": 0.8141601085662842,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.5073,
      "step": 76
    },
    {
      "epoch": 0.2987390882638215,
      "grad_norm": 0.7709934711456299,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.5256,
      "step": 77
    },
    {
      "epoch": 0.3026188166828322,
      "grad_norm": 0.7749450206756592,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.5704,
      "step": 78
    },
    {
      "epoch": 0.3064985451018429,
      "grad_norm": 0.8578307032585144,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.5245,
      "step": 79
    },
    {
      "epoch": 0.31037827352085356,
      "grad_norm": 0.8243958950042725,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3822,
      "step": 80
    },
    {
      "epoch": 0.3142580019398642,
      "grad_norm": 0.9947974681854248,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.6031,
      "step": 81
    },
    {
      "epoch": 0.3181377303588749,
      "grad_norm": 0.9279785752296448,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.4928,
      "step": 82
    },
    {
      "epoch": 0.32201745877788557,
      "grad_norm": 0.9952326416969299,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.5339,
      "step": 83
    },
    {
      "epoch": 0.3258971871968962,
      "grad_norm": 1.0020768642425537,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4438,
      "step": 84
    },
    {
      "epoch": 0.3297769156159069,
      "grad_norm": 0.9847713112831116,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.4096,
      "step": 85
    },
    {
      "epoch": 0.33365664403491757,
      "grad_norm": 0.9898356795310974,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4761,
      "step": 86
    },
    {
      "epoch": 0.3375363724539282,
      "grad_norm": 1.0013384819030762,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.3546,
      "step": 87
    },
    {
      "epoch": 0.3414161008729389,
      "grad_norm": 1.0115240812301636,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.3231,
      "step": 88
    },
    {
      "epoch": 0.3452958292919496,
      "grad_norm": 1.1733163595199585,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.4004,
      "step": 89
    },
    {
      "epoch": 0.34917555771096026,
      "grad_norm": 1.0035802125930786,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3748,
      "step": 90
    },
    {
      "epoch": 0.3530552861299709,
      "grad_norm": 1.1230354309082031,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3844,
      "step": 91
    },
    {
      "epoch": 0.3569350145489816,
      "grad_norm": 1.0277193784713745,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.3921,
      "step": 92
    },
    {
      "epoch": 0.36081474296799226,
      "grad_norm": 1.1185137033462524,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3628,
      "step": 93
    },
    {
      "epoch": 0.3646944713870029,
      "grad_norm": 1.0993831157684326,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.4835,
      "step": 94
    },
    {
      "epoch": 0.3685741998060136,
      "grad_norm": 1.1529768705368042,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.4675,
      "step": 95
    },
    {
      "epoch": 0.37245392822502427,
      "grad_norm": 1.12255859375,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.5015,
      "step": 96
    },
    {
      "epoch": 0.3763336566440349,
      "grad_norm": 1.1144906282424927,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.4869,
      "step": 97
    },
    {
      "epoch": 0.3802133850630456,
      "grad_norm": 1.1677404642105103,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.4598,
      "step": 98
    },
    {
      "epoch": 0.38409311348205627,
      "grad_norm": 1.366492748260498,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.3516,
      "step": 99
    },
    {
      "epoch": 0.3879728419010669,
      "grad_norm": 1.6263788938522339,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.1719,
      "step": 100
    },
    {
      "epoch": 0.3879728419010669,
      "eval_loss": 1.6853090524673462,
      "eval_runtime": 44.3683,
      "eval_samples_per_second": 9.782,
      "eval_steps_per_second": 2.457,
      "step": 100
    },
    {
      "epoch": 0.3918525703200776,
      "grad_norm": 2.0852928161621094,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.8501,
      "step": 101
    },
    {
      "epoch": 0.3957322987390883,
      "grad_norm": 1.4939178228378296,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.8719,
      "step": 102
    },
    {
      "epoch": 0.39961202715809896,
      "grad_norm": 1.0582674741744995,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.8006,
      "step": 103
    },
    {
      "epoch": 0.4034917555771096,
      "grad_norm": 0.8404370546340942,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.6485,
      "step": 104
    },
    {
      "epoch": 0.4073714839961203,
      "grad_norm": 0.5674321055412292,
      "learning_rate": 5e-05,
      "loss": 1.6834,
      "step": 105
    },
    {
      "epoch": 0.41125121241513096,
      "grad_norm": 0.5925096273422241,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.642,
      "step": 106
    },
    {
      "epoch": 0.4151309408341416,
      "grad_norm": 0.5303599834442139,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.7012,
      "step": 107
    },
    {
      "epoch": 0.4190106692531523,
      "grad_norm": 0.4726005494594574,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.6368,
      "step": 108
    },
    {
      "epoch": 0.42289039767216297,
      "grad_norm": 0.4614112377166748,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.682,
      "step": 109
    },
    {
      "epoch": 0.4267701260911736,
      "grad_norm": 0.4816155731678009,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.7424,
      "step": 110
    },
    {
      "epoch": 0.4306498545101843,
      "grad_norm": 0.5574136972427368,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.7514,
      "step": 111
    },
    {
      "epoch": 0.43452958292919497,
      "grad_norm": 0.47104960680007935,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.6623,
      "step": 112
    },
    {
      "epoch": 0.4384093113482056,
      "grad_norm": 0.5246789455413818,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.6477,
      "step": 113
    },
    {
      "epoch": 0.4422890397672163,
      "grad_norm": 0.5577223300933838,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6598,
      "step": 114
    },
    {
      "epoch": 0.44616876818622697,
      "grad_norm": 0.5409120321273804,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.5918,
      "step": 115
    },
    {
      "epoch": 0.45004849660523766,
      "grad_norm": 0.5998454689979553,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.7141,
      "step": 116
    },
    {
      "epoch": 0.4539282250242483,
      "grad_norm": 0.5691230893135071,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.7016,
      "step": 117
    },
    {
      "epoch": 0.457807953443259,
      "grad_norm": 0.5979486703872681,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.5888,
      "step": 118
    },
    {
      "epoch": 0.46168768186226966,
      "grad_norm": 0.5919355750083923,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.6118,
      "step": 119
    },
    {
      "epoch": 0.4655674102812803,
      "grad_norm": 0.6297187805175781,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.5175,
      "step": 120
    },
    {
      "epoch": 0.469447138700291,
      "grad_norm": 0.6464974880218506,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5499,
      "step": 121
    },
    {
      "epoch": 0.47332686711930166,
      "grad_norm": 0.6574711799621582,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.6358,
      "step": 122
    },
    {
      "epoch": 0.4772065955383123,
      "grad_norm": 0.6681584119796753,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.5807,
      "step": 123
    },
    {
      "epoch": 0.481086323957323,
      "grad_norm": 0.6784456968307495,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.5395,
      "step": 124
    },
    {
      "epoch": 0.48496605237633367,
      "grad_norm": 0.7298809289932251,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.4993,
      "step": 125
    },
    {
      "epoch": 0.48884578079534435,
      "grad_norm": 0.7544025778770447,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.4906,
      "step": 126
    },
    {
      "epoch": 0.492725509214355,
      "grad_norm": 0.7846170663833618,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.4911,
      "step": 127
    },
    {
      "epoch": 0.49660523763336567,
      "grad_norm": 0.7994524836540222,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.5301,
      "step": 128
    },
    {
      "epoch": 0.5004849660523764,
      "grad_norm": 0.8499495387077332,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4388,
      "step": 129
    },
    {
      "epoch": 0.504364694471387,
      "grad_norm": 0.8359389305114746,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.4059,
      "step": 130
    },
    {
      "epoch": 0.5082444228903976,
      "grad_norm": 0.8449875712394714,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.5015,
      "step": 131
    },
    {
      "epoch": 0.5121241513094084,
      "grad_norm": 0.8801817297935486,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.6149,
      "step": 132
    },
    {
      "epoch": 0.516003879728419,
      "grad_norm": 0.9136694073677063,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4561,
      "step": 133
    },
    {
      "epoch": 0.5198836081474297,
      "grad_norm": 0.9364612102508545,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4704,
      "step": 134
    },
    {
      "epoch": 0.5237633365664404,
      "grad_norm": 0.9270147681236267,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4447,
      "step": 135
    },
    {
      "epoch": 0.527643064985451,
      "grad_norm": 0.9426939487457275,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.4243,
      "step": 136
    },
    {
      "epoch": 0.5315227934044617,
      "grad_norm": 0.9100015163421631,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.4111,
      "step": 137
    },
    {
      "epoch": 0.5354025218234724,
      "grad_norm": 0.9527736306190491,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.3173,
      "step": 138
    },
    {
      "epoch": 0.539282250242483,
      "grad_norm": 1.048341155052185,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3918,
      "step": 139
    },
    {
      "epoch": 0.5431619786614937,
      "grad_norm": 1.0683382749557495,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.3444,
      "step": 140
    },
    {
      "epoch": 0.5470417070805044,
      "grad_norm": 0.990234911441803,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2979,
      "step": 141
    },
    {
      "epoch": 0.550921435499515,
      "grad_norm": 1.0605175495147705,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.3256,
      "step": 142
    },
    {
      "epoch": 0.5548011639185257,
      "grad_norm": 1.0509885549545288,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3868,
      "step": 143
    },
    {
      "epoch": 0.5586808923375364,
      "grad_norm": 1.065555453300476,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.3605,
      "step": 144
    },
    {
      "epoch": 0.562560620756547,
      "grad_norm": 1.0612016916275024,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.3952,
      "step": 145
    },
    {
      "epoch": 0.5664403491755577,
      "grad_norm": 1.1851342916488647,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4192,
      "step": 146
    },
    {
      "epoch": 0.5703200775945684,
      "grad_norm": 1.1040699481964111,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.419,
      "step": 147
    },
    {
      "epoch": 0.574199806013579,
      "grad_norm": 1.1693785190582275,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.4213,
      "step": 148
    },
    {
      "epoch": 0.5780795344325897,
      "grad_norm": 1.3049074411392212,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3296,
      "step": 149
    },
    {
      "epoch": 0.5819592628516004,
      "grad_norm": 1.4876773357391357,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1673,
      "step": 150
    },
    {
      "epoch": 0.5819592628516004,
      "eval_loss": 1.5319730043411255,
      "eval_runtime": 44.307,
      "eval_samples_per_second": 9.795,
      "eval_steps_per_second": 2.46,
      "step": 150
    }
  ],
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.6771197010116608e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}