{
  "best_metric": 1.5257015228271484,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.4550625711035267,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0022753128555176336,
      "grad_norm": 0.22611123323440552,
      "learning_rate": 1e-05,
      "loss": 1.0903,
      "step": 1
    },
    {
      "epoch": 0.0022753128555176336,
      "eval_loss": 1.913366675376892,
      "eval_runtime": 51.9448,
      "eval_samples_per_second": 14.246,
      "eval_steps_per_second": 3.561,
      "step": 1
    },
    {
      "epoch": 0.004550625711035267,
      "grad_norm": 0.2882371246814728,
      "learning_rate": 2e-05,
      "loss": 1.3957,
      "step": 2
    },
    {
      "epoch": 0.006825938566552901,
      "grad_norm": 0.34990158677101135,
      "learning_rate": 3e-05,
      "loss": 1.358,
      "step": 3
    },
    {
      "epoch": 0.009101251422070534,
      "grad_norm": 0.35712549090385437,
      "learning_rate": 4e-05,
      "loss": 1.7096,
      "step": 4
    },
    {
      "epoch": 0.011376564277588168,
      "grad_norm": 0.49686571955680847,
      "learning_rate": 5e-05,
      "loss": 1.6619,
      "step": 5
    },
    {
      "epoch": 0.013651877133105802,
      "grad_norm": 0.37168917059898376,
      "learning_rate": 6e-05,
      "loss": 1.4996,
      "step": 6
    },
    {
      "epoch": 0.015927189988623434,
      "grad_norm": 0.35566818714141846,
      "learning_rate": 7e-05,
      "loss": 1.7117,
      "step": 7
    },
    {
      "epoch": 0.01820250284414107,
      "grad_norm": 0.3652589023113251,
      "learning_rate": 8e-05,
      "loss": 1.7461,
      "step": 8
    },
    {
      "epoch": 0.020477815699658702,
      "grad_norm": 0.3730931580066681,
      "learning_rate": 9e-05,
      "loss": 1.6422,
      "step": 9
    },
    {
      "epoch": 0.022753128555176336,
      "grad_norm": 0.34812870621681213,
      "learning_rate": 0.0001,
      "loss": 1.5677,
      "step": 10
    },
    {
      "epoch": 0.02502844141069397,
      "grad_norm": 0.3891918659210205,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7328,
      "step": 11
    },
    {
      "epoch": 0.027303754266211604,
      "grad_norm": 0.4011567533016205,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.5787,
      "step": 12
    },
    {
      "epoch": 0.02957906712172924,
      "grad_norm": 0.40960559248924255,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.6034,
      "step": 13
    },
    {
      "epoch": 0.03185437997724687,
      "grad_norm": 0.3591533303260803,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5651,
      "step": 14
    },
    {
      "epoch": 0.034129692832764506,
      "grad_norm": 0.3521420359611511,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4885,
      "step": 15
    },
    {
      "epoch": 0.03640500568828214,
      "grad_norm": 0.3475079834461212,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5165,
      "step": 16
    },
    {
      "epoch": 0.038680318543799774,
      "grad_norm": 0.34967538714408875,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.7764,
      "step": 17
    },
    {
      "epoch": 0.040955631399317405,
      "grad_norm": 0.32981762290000916,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.5613,
      "step": 18
    },
    {
      "epoch": 0.04323094425483504,
      "grad_norm": 0.3880794644355774,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.427,
      "step": 19
    },
    {
      "epoch": 0.04550625711035267,
      "grad_norm": 0.3884115219116211,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6436,
      "step": 20
    },
    {
      "epoch": 0.04778156996587031,
      "grad_norm": 0.41450196504592896,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.7233,
      "step": 21
    },
    {
      "epoch": 0.05005688282138794,
      "grad_norm": 0.4754553735256195,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4519,
      "step": 22
    },
    {
      "epoch": 0.05233219567690557,
      "grad_norm": 0.4060521423816681,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5302,
      "step": 23
    },
    {
      "epoch": 0.05460750853242321,
      "grad_norm": 0.4019545018672943,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.6562,
      "step": 24
    },
    {
      "epoch": 0.05688282138794084,
      "grad_norm": 0.4410861134529114,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4272,
      "step": 25
    },
    {
      "epoch": 0.05915813424345848,
      "grad_norm": 0.443133682012558,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5312,
      "step": 26
    },
    {
      "epoch": 0.06143344709897611,
      "grad_norm": 0.5422303080558777,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5247,
      "step": 27
    },
    {
      "epoch": 0.06370875995449374,
      "grad_norm": 0.5324600338935852,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.535,
      "step": 28
    },
    {
      "epoch": 0.06598407281001138,
      "grad_norm": 0.48739585280418396,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2958,
      "step": 29
    },
    {
      "epoch": 0.06825938566552901,
      "grad_norm": 0.5466403961181641,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.4252,
      "step": 30
    },
    {
      "epoch": 0.07053469852104664,
      "grad_norm": 0.5209930539131165,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.6149,
      "step": 31
    },
    {
      "epoch": 0.07281001137656427,
      "grad_norm": 0.5467900037765503,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3175,
      "step": 32
    },
    {
      "epoch": 0.07508532423208192,
      "grad_norm": 0.5256214141845703,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.4573,
      "step": 33
    },
    {
      "epoch": 0.07736063708759955,
      "grad_norm": 0.6561576128005981,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4363,
      "step": 34
    },
    {
      "epoch": 0.07963594994311718,
      "grad_norm": 0.6951440572738647,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4992,
      "step": 35
    },
    {
      "epoch": 0.08191126279863481,
      "grad_norm": 0.710189938545227,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.4516,
      "step": 36
    },
    {
      "epoch": 0.08418657565415244,
      "grad_norm": 1.0408570766448975,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.6305,
      "step": 37
    },
    {
      "epoch": 0.08646188850967008,
      "grad_norm": 1.059044361114502,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.6068,
      "step": 38
    },
    {
      "epoch": 0.08873720136518772,
      "grad_norm": 0.9533563256263733,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4625,
      "step": 39
    },
    {
      "epoch": 0.09101251422070535,
      "grad_norm": 1.3289941549301147,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.5506,
      "step": 40
    },
    {
      "epoch": 0.09328782707622298,
      "grad_norm": 1.0026781558990479,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.6201,
      "step": 41
    },
    {
      "epoch": 0.09556313993174062,
      "grad_norm": 1.0984512567520142,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.7105,
      "step": 42
    },
    {
      "epoch": 0.09783845278725825,
      "grad_norm": 0.9385897517204285,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.5593,
      "step": 43
    },
    {
      "epoch": 0.10011376564277588,
      "grad_norm": 1.5114777088165283,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.6515,
      "step": 44
    },
    {
      "epoch": 0.10238907849829351,
      "grad_norm": 1.1186617612838745,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.6139,
      "step": 45
    },
    {
      "epoch": 0.10466439135381114,
      "grad_norm": 1.2790113687515259,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.574,
      "step": 46
    },
    {
      "epoch": 0.10693970420932879,
      "grad_norm": 1.5633999109268188,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.7048,
      "step": 47
    },
    {
      "epoch": 0.10921501706484642,
      "grad_norm": 1.653621792793274,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.6545,
      "step": 48
    },
    {
      "epoch": 0.11149032992036405,
      "grad_norm": 2.5716707706451416,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.3276,
      "step": 49
    },
    {
      "epoch": 0.11376564277588168,
      "grad_norm": 2.4737744331359863,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.203,
      "step": 50
    },
    {
      "epoch": 0.11376564277588168,
      "eval_loss": 1.7738934755325317,
      "eval_runtime": 52.7688,
      "eval_samples_per_second": 14.023,
      "eval_steps_per_second": 3.506,
      "step": 50
    },
    {
      "epoch": 0.11604095563139932,
      "grad_norm": 1.0260125398635864,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.3562,
      "step": 51
    },
    {
      "epoch": 0.11831626848691695,
      "grad_norm": 0.7305225133895874,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2388,
      "step": 52
    },
    {
      "epoch": 0.12059158134243458,
      "grad_norm": 0.6264262795448303,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.3545,
      "step": 53
    },
    {
      "epoch": 0.12286689419795221,
      "grad_norm": 0.5451095104217529,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.641,
      "step": 54
    },
    {
      "epoch": 0.12514220705346984,
      "grad_norm": 0.4437565505504608,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.6389,
      "step": 55
    },
    {
      "epoch": 0.12741751990898748,
      "grad_norm": 0.32434388995170593,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5354,
      "step": 56
    },
    {
      "epoch": 0.1296928327645051,
      "grad_norm": 0.2733739912509918,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4884,
      "step": 57
    },
    {
      "epoch": 0.13196814562002276,
      "grad_norm": 0.2972055971622467,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4195,
      "step": 58
    },
    {
      "epoch": 0.1342434584755404,
      "grad_norm": 0.28056544065475464,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4049,
      "step": 59
    },
    {
      "epoch": 0.13651877133105803,
      "grad_norm": 0.3651324510574341,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7733,
      "step": 60
    },
    {
      "epoch": 0.13879408418657566,
      "grad_norm": 0.3437943458557129,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.549,
      "step": 61
    },
    {
      "epoch": 0.1410693970420933,
      "grad_norm": 0.32947683334350586,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.7126,
      "step": 62
    },
    {
      "epoch": 0.14334470989761092,
      "grad_norm": 0.32842299342155457,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.619,
      "step": 63
    },
    {
      "epoch": 0.14562002275312855,
      "grad_norm": 0.314636766910553,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.6833,
      "step": 64
    },
    {
      "epoch": 0.14789533560864618,
      "grad_norm": 0.3103666305541992,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.5699,
      "step": 65
    },
    {
      "epoch": 0.15017064846416384,
      "grad_norm": 0.3245433568954468,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.8205,
      "step": 66
    },
    {
      "epoch": 0.15244596131968147,
      "grad_norm": 0.280921995639801,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.6793,
      "step": 67
    },
    {
      "epoch": 0.1547212741751991,
      "grad_norm": 0.31751391291618347,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.4389,
      "step": 68
    },
    {
      "epoch": 0.15699658703071673,
      "grad_norm": 0.3016919791698456,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5417,
      "step": 69
    },
    {
      "epoch": 0.15927189988623436,
      "grad_norm": 0.3409747779369354,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4803,
      "step": 70
    },
    {
      "epoch": 0.161547212741752,
      "grad_norm": 0.3046075403690338,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5447,
      "step": 71
    },
    {
      "epoch": 0.16382252559726962,
      "grad_norm": 0.304971843957901,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.5174,
      "step": 72
    },
    {
      "epoch": 0.16609783845278725,
      "grad_norm": 0.2970457375049591,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.7005,
      "step": 73
    },
    {
      "epoch": 0.16837315130830488,
      "grad_norm": 0.3276102542877197,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.6057,
      "step": 74
    },
    {
      "epoch": 0.17064846416382254,
      "grad_norm": 0.43218204379081726,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.6703,
      "step": 75
    },
    {
      "epoch": 0.17292377701934017,
      "grad_norm": 0.36013495922088623,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6124,
      "step": 76
    },
    {
      "epoch": 0.1751990898748578,
      "grad_norm": 0.4015674293041229,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4525,
      "step": 77
    },
    {
      "epoch": 0.17747440273037543,
      "grad_norm": 0.3538857102394104,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3639,
      "step": 78
    },
    {
      "epoch": 0.17974971558589306,
      "grad_norm": 0.3585698902606964,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3993,
      "step": 79
    },
    {
      "epoch": 0.1820250284414107,
      "grad_norm": 0.398530513048172,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.5836,
      "step": 80
    },
    {
      "epoch": 0.18430034129692832,
      "grad_norm": 0.6290268301963806,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.4226,
      "step": 81
    },
    {
      "epoch": 0.18657565415244595,
      "grad_norm": 0.5585540533065796,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.4187,
      "step": 82
    },
    {
      "epoch": 0.18885096700796358,
      "grad_norm": 0.6777237057685852,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.4208,
      "step": 83
    },
    {
      "epoch": 0.19112627986348124,
      "grad_norm": 0.6730694770812988,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4802,
      "step": 84
    },
    {
      "epoch": 0.19340159271899887,
      "grad_norm": 0.6737896203994751,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.438,
      "step": 85
    },
    {
      "epoch": 0.1956769055745165,
      "grad_norm": 0.5455721020698547,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4361,
      "step": 86
    },
    {
      "epoch": 0.19795221843003413,
      "grad_norm": 0.630711019039154,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.4103,
      "step": 87
    },
    {
      "epoch": 0.20022753128555176,
      "grad_norm": 0.6227461695671082,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.5053,
      "step": 88
    },
    {
      "epoch": 0.2025028441410694,
      "grad_norm": 0.6068131923675537,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.6091,
      "step": 89
    },
    {
      "epoch": 0.20477815699658702,
      "grad_norm": 0.6199155449867249,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.4767,
      "step": 90
    },
    {
      "epoch": 0.20705346985210465,
      "grad_norm": 0.7213968634605408,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.5521,
      "step": 91
    },
    {
      "epoch": 0.20932878270762229,
      "grad_norm": 0.9123151302337646,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4679,
      "step": 92
    },
    {
      "epoch": 0.21160409556313994,
      "grad_norm": 0.9512717127799988,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.567,
      "step": 93
    },
    {
      "epoch": 0.21387940841865757,
      "grad_norm": 0.7897658944129944,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3992,
      "step": 94
    },
    {
      "epoch": 0.2161547212741752,
      "grad_norm": 0.9762223958969116,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.2692,
      "step": 95
    },
    {
      "epoch": 0.21843003412969283,
      "grad_norm": 0.9920843243598938,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.054,
      "step": 96
    },
    {
      "epoch": 0.22070534698521047,
      "grad_norm": 1.2355668544769287,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.4947,
      "step": 97
    },
    {
      "epoch": 0.2229806598407281,
      "grad_norm": 1.5962979793548584,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.6507,
      "step": 98
    },
    {
      "epoch": 0.22525597269624573,
      "grad_norm": 2.07619571685791,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.7452,
      "step": 99
    },
    {
      "epoch": 0.22753128555176336,
      "grad_norm": 4.199832439422607,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.644,
      "step": 100
    },
    {
      "epoch": 0.22753128555176336,
      "eval_loss": 1.6227988004684448,
      "eval_runtime": 52.7896,
      "eval_samples_per_second": 14.018,
      "eval_steps_per_second": 3.504,
      "step": 100
    },
    {
      "epoch": 0.229806598407281,
      "grad_norm": 0.6693715453147888,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2433,
      "step": 101
    },
    {
      "epoch": 0.23208191126279865,
      "grad_norm": 0.7670778036117554,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2632,
      "step": 102
    },
    {
      "epoch": 0.23435722411831628,
      "grad_norm": 0.8197275996208191,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.6592,
      "step": 103
    },
    {
      "epoch": 0.2366325369738339,
      "grad_norm": 0.6911060810089111,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.6658,
      "step": 104
    },
    {
      "epoch": 0.23890784982935154,
      "grad_norm": 0.5780044794082642,
      "learning_rate": 5e-05,
      "loss": 1.4384,
      "step": 105
    },
    {
      "epoch": 0.24118316268486917,
      "grad_norm": 0.46383464336395264,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3921,
      "step": 106
    },
    {
      "epoch": 0.2434584755403868,
      "grad_norm": 0.34697556495666504,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5743,
      "step": 107
    },
    {
      "epoch": 0.24573378839590443,
      "grad_norm": 0.3593616783618927,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4955,
      "step": 108
    },
    {
      "epoch": 0.24800910125142206,
      "grad_norm": 0.3376767039299011,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.7184,
      "step": 109
    },
    {
      "epoch": 0.2502844141069397,
      "grad_norm": 0.30205029249191284,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5361,
      "step": 110
    },
    {
      "epoch": 0.2525597269624573,
      "grad_norm": 0.2967416048049927,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.5408,
      "step": 111
    },
    {
      "epoch": 0.25483503981797495,
      "grad_norm": 0.32019612193107605,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.5839,
      "step": 112
    },
    {
      "epoch": 0.2571103526734926,
      "grad_norm": 0.3251321315765381,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.7196,
      "step": 113
    },
    {
      "epoch": 0.2593856655290102,
      "grad_norm": 0.3136993646621704,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.5447,
      "step": 114
    },
    {
      "epoch": 0.2616609783845279,
      "grad_norm": 0.32139745354652405,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.7732,
      "step": 115
    },
    {
      "epoch": 0.26393629124004553,
      "grad_norm": 0.31381091475486755,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.4551,
      "step": 116
    },
    {
      "epoch": 0.26621160409556316,
      "grad_norm": 0.3346835672855377,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.5466,
      "step": 117
    },
    {
      "epoch": 0.2684869169510808,
      "grad_norm": 0.3558894097805023,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.5781,
      "step": 118
    },
    {
      "epoch": 0.2707622298065984,
      "grad_norm": 0.33181649446487427,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.5005,
      "step": 119
    },
    {
      "epoch": 0.27303754266211605,
      "grad_norm": 0.3452757000923157,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.556,
      "step": 120
    },
    {
      "epoch": 0.2753128555176337,
      "grad_norm": 0.32899919152259827,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5361,
      "step": 121
    },
    {
      "epoch": 0.2775881683731513,
      "grad_norm": 0.3074376881122589,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.6418,
      "step": 122
    },
    {
      "epoch": 0.27986348122866894,
      "grad_norm": 0.35091540217399597,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.7146,
      "step": 123
    },
    {
      "epoch": 0.2821387940841866,
      "grad_norm": 0.35279640555381775,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.5709,
      "step": 124
    },
    {
      "epoch": 0.2844141069397042,
      "grad_norm": 0.3835878074169159,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.6661,
      "step": 125
    },
    {
      "epoch": 0.28668941979522183,
      "grad_norm": 0.36722052097320557,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.5732,
      "step": 126
    },
    {
      "epoch": 0.28896473265073946,
      "grad_norm": 0.3579557240009308,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.5109,
      "step": 127
    },
    {
      "epoch": 0.2912400455062571,
      "grad_norm": 0.3704659640789032,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.3831,
      "step": 128
    },
    {
      "epoch": 0.2935153583617747,
      "grad_norm": 0.37062570452690125,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4435,
      "step": 129
    },
    {
      "epoch": 0.29579067121729236,
      "grad_norm": 0.3710121512413025,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.4904,
      "step": 130
    },
    {
      "epoch": 0.29806598407281,
      "grad_norm": 0.4146701395511627,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3529,
      "step": 131
    },
    {
      "epoch": 0.3003412969283277,
      "grad_norm": 0.4873427748680115,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3432,
      "step": 132
    },
    {
      "epoch": 0.3026166097838453,
      "grad_norm": 0.5232824683189392,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4961,
      "step": 133
    },
    {
      "epoch": 0.30489192263936293,
      "grad_norm": 0.5080462694168091,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.5651,
      "step": 134
    },
    {
      "epoch": 0.30716723549488056,
      "grad_norm": 0.5162842273712158,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4832,
      "step": 135
    },
    {
      "epoch": 0.3094425483503982,
      "grad_norm": 0.5426189303398132,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.3464,
      "step": 136
    },
    {
      "epoch": 0.3117178612059158,
      "grad_norm": 0.6192271113395691,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.5124,
      "step": 137
    },
    {
      "epoch": 0.31399317406143346,
      "grad_norm": 0.7104066610336304,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.6479,
      "step": 138
    },
    {
      "epoch": 0.3162684869169511,
      "grad_norm": 0.6071523427963257,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.363,
      "step": 139
    },
    {
      "epoch": 0.3185437997724687,
      "grad_norm": 0.6854862570762634,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.5242,
      "step": 140
    },
    {
      "epoch": 0.32081911262798635,
      "grad_norm": 0.6915633678436279,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.4792,
      "step": 141
    },
    {
      "epoch": 0.323094425483504,
      "grad_norm": 0.7244867086410522,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.4209,
      "step": 142
    },
    {
      "epoch": 0.3253697383390216,
      "grad_norm": 0.92000812292099,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4743,
      "step": 143
    },
    {
      "epoch": 0.32764505119453924,
      "grad_norm": 0.775058388710022,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.249,
      "step": 144
    },
    {
      "epoch": 0.32992036405005687,
      "grad_norm": 0.7781512141227722,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.0711,
      "step": 145
    },
    {
      "epoch": 0.3321956769055745,
      "grad_norm": 0.8318851590156555,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.852,
      "step": 146
    },
    {
      "epoch": 0.33447098976109213,
      "grad_norm": 0.8757054805755615,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.0634,
      "step": 147
    },
    {
      "epoch": 0.33674630261660976,
      "grad_norm": 1.2129353284835815,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.6328,
      "step": 148
    },
    {
      "epoch": 0.3390216154721274,
      "grad_norm": 1.4702993631362915,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.0972,
      "step": 149
    },
    {
      "epoch": 0.3412969283276451,
      "grad_norm": 1.9301505088806152,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.8096,
      "step": 150
    },
    {
      "epoch": 0.3412969283276451,
      "eval_loss": 1.5260435342788696,
      "eval_runtime": 52.7507,
      "eval_samples_per_second": 14.028,
      "eval_steps_per_second": 3.507,
      "step": 150
    },
    {
      "epoch": 0.3435722411831627,
      "grad_norm": 0.22408261895179749,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.3613,
      "step": 151
    },
    {
      "epoch": 0.34584755403868034,
      "grad_norm": 0.231350377202034,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.0711,
      "step": 152
    },
    {
      "epoch": 0.34812286689419797,
      "grad_norm": 0.326104074716568,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.4257,
      "step": 153
    },
    {
      "epoch": 0.3503981797497156,
      "grad_norm": 0.31789645552635193,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.528,
      "step": 154
    },
    {
      "epoch": 0.35267349260523323,
      "grad_norm": 0.3739003837108612,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.5169,
      "step": 155
    },
    {
      "epoch": 0.35494880546075086,
      "grad_norm": 0.3648185133934021,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5665,
      "step": 156
    },
    {
      "epoch": 0.3572241183162685,
      "grad_norm": 0.3743276000022888,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.7729,
      "step": 157
    },
    {
      "epoch": 0.3594994311717861,
      "grad_norm": 0.40119045972824097,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.6942,
      "step": 158
    },
    {
      "epoch": 0.36177474402730375,
      "grad_norm": 0.4109828770160675,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.622,
      "step": 159
    },
    {
      "epoch": 0.3640500568828214,
      "grad_norm": 0.3626691401004791,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.5766,
      "step": 160
    },
    {
      "epoch": 0.366325369738339,
      "grad_norm": 0.3739473819732666,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.6747,
      "step": 161
    },
    {
      "epoch": 0.36860068259385664,
      "grad_norm": 0.40398624539375305,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.3581,
      "step": 162
    },
    {
      "epoch": 0.3708759954493743,
      "grad_norm": 0.41251248121261597,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.635,
      "step": 163
    },
    {
      "epoch": 0.3731513083048919,
      "grad_norm": 0.4264377951622009,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.6063,
      "step": 164
    },
    {
      "epoch": 0.37542662116040953,
      "grad_norm": 0.3966737687587738,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.6201,
      "step": 165
    },
    {
      "epoch": 0.37770193401592717,
      "grad_norm": 0.4379703104496002,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.5862,
      "step": 166
    },
    {
      "epoch": 0.3799772468714448,
      "grad_norm": 0.4034595489501953,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.6002,
      "step": 167
    },
    {
      "epoch": 0.3822525597269625,
      "grad_norm": 0.4096910357475281,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.4808,
      "step": 168
    },
    {
      "epoch": 0.3845278725824801,
      "grad_norm": 0.40079763531684875,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.6244,
      "step": 169
    },
    {
      "epoch": 0.38680318543799774,
      "grad_norm": 0.42421388626098633,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.6158,
      "step": 170
    },
    {
      "epoch": 0.3890784982935154,
      "grad_norm": 0.4529460370540619,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.6335,
      "step": 171
    },
    {
      "epoch": 0.391353811149033,
      "grad_norm": 0.45348235964775085,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.4969,
      "step": 172
    },
    {
      "epoch": 0.39362912400455063,
      "grad_norm": 0.4705301523208618,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.7246,
      "step": 173
    },
    {
      "epoch": 0.39590443686006827,
      "grad_norm": 0.47413575649261475,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.6055,
      "step": 174
    },
    {
      "epoch": 0.3981797497155859,
      "grad_norm": 0.48788559436798096,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.4845,
      "step": 175
    },
    {
      "epoch": 0.4004550625711035,
      "grad_norm": 0.48533132672309875,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.5219,
      "step": 176
    },
    {
      "epoch": 0.40273037542662116,
      "grad_norm": 0.4807284474372864,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.4249,
      "step": 177
    },
    {
      "epoch": 0.4050056882821388,
      "grad_norm": 0.5407995581626892,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.3492,
      "step": 178
    },
    {
      "epoch": 0.4072810011376564,
      "grad_norm": 0.47076892852783203,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.3692,
      "step": 179
    },
    {
      "epoch": 0.40955631399317405,
      "grad_norm": 0.509828507900238,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.5629,
      "step": 180
    },
    {
      "epoch": 0.4118316268486917,
      "grad_norm": 0.4772973358631134,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.5048,
      "step": 181
    },
    {
      "epoch": 0.4141069397042093,
      "grad_norm": 0.603643536567688,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.514,
      "step": 182
    },
    {
      "epoch": 0.41638225255972694,
      "grad_norm": 0.4952646791934967,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.3583,
      "step": 183
    },
    {
      "epoch": 0.41865756541524457,
      "grad_norm": 0.6499698162078857,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.25,
      "step": 184
    },
    {
      "epoch": 0.42093287827076226,
      "grad_norm": 0.5622658729553223,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.3969,
      "step": 185
    },
    {
      "epoch": 0.4232081911262799,
      "grad_norm": 0.556089460849762,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4778,
      "step": 186
    },
    {
      "epoch": 0.4254835039817975,
      "grad_norm": 0.5632639527320862,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.5124,
      "step": 187
    },
    {
      "epoch": 0.42775881683731515,
      "grad_norm": 0.6200251579284668,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.6596,
      "step": 188
    },
    {
      "epoch": 0.4300341296928328,
      "grad_norm": 0.6467673778533936,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.5901,
      "step": 189
    },
    {
      "epoch": 0.4323094425483504,
      "grad_norm": 0.6679887175559998,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.5117,
      "step": 190
    },
    {
      "epoch": 0.43458475540386804,
      "grad_norm": 0.6394217014312744,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.513,
      "step": 191
    },
    {
      "epoch": 0.43686006825938567,
      "grad_norm": 0.7728143930435181,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3762,
      "step": 192
    },
    {
      "epoch": 0.4391353811149033,
      "grad_norm": 0.6505006551742554,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.6086,
      "step": 193
    },
    {
      "epoch": 0.44141069397042093,
      "grad_norm": 0.725277841091156,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.1553,
      "step": 194
    },
    {
      "epoch": 0.44368600682593856,
      "grad_norm": 0.7936424612998962,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.0847,
      "step": 195
    },
    {
      "epoch": 0.4459613196814562,
      "grad_norm": 0.8912354707717896,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.277,
      "step": 196
    },
    {
      "epoch": 0.4482366325369738,
      "grad_norm": 0.8243184089660645,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.3681,
      "step": 197
    },
    {
      "epoch": 0.45051194539249145,
      "grad_norm": 1.2224607467651367,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.7822,
      "step": 198
    },
    {
      "epoch": 0.4527872582480091,
      "grad_norm": 1.5814566612243652,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.6599,
      "step": 199
    },
    {
      "epoch": 0.4550625711035267,
      "grad_norm": 2.2363340854644775,
      "learning_rate": 0.0,
      "loss": 2.1814,
      "step": 200
    },
    {
      "epoch": 0.4550625711035267,
      "eval_loss": 1.5257015228271484,
      "eval_runtime": 52.7844,
      "eval_samples_per_second": 14.019,
      "eval_steps_per_second": 3.505,
      "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": 2.6743224049179034e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}