{
  "best_metric": 2.1066830158233643,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9535160905840286,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004767580452920143,
      "grad_norm": 3.17598819732666,
      "learning_rate": 1e-05,
      "loss": 2.7102,
      "step": 1
    },
    {
      "epoch": 0.004767580452920143,
      "eval_loss": 4.050970077514648,
      "eval_runtime": 24.7205,
      "eval_samples_per_second": 14.32,
      "eval_steps_per_second": 3.6,
      "step": 1
    },
    {
      "epoch": 0.009535160905840286,
      "grad_norm": 3.5215253829956055,
      "learning_rate": 2e-05,
      "loss": 2.6229,
      "step": 2
    },
    {
      "epoch": 0.014302741358760428,
      "grad_norm": 1.8396703004837036,
      "learning_rate": 3e-05,
      "loss": 2.5976,
      "step": 3
    },
    {
      "epoch": 0.01907032181168057,
      "grad_norm": 1.822470784187317,
      "learning_rate": 4e-05,
      "loss": 2.8094,
      "step": 4
    },
    {
      "epoch": 0.023837902264600714,
      "grad_norm": 1.784137487411499,
      "learning_rate": 5e-05,
      "loss": 2.7152,
      "step": 5
    },
    {
      "epoch": 0.028605482717520857,
      "grad_norm": 1.5671970844268799,
      "learning_rate": 6e-05,
      "loss": 2.8053,
      "step": 6
    },
    {
      "epoch": 0.033373063170441,
      "grad_norm": 1.2720322608947754,
      "learning_rate": 7e-05,
      "loss": 2.7655,
      "step": 7
    },
    {
      "epoch": 0.03814064362336114,
      "grad_norm": 1.3592026233673096,
      "learning_rate": 8e-05,
      "loss": 2.852,
      "step": 8
    },
    {
      "epoch": 0.04290822407628129,
      "grad_norm": 1.2666702270507812,
      "learning_rate": 9e-05,
      "loss": 2.8961,
      "step": 9
    },
    {
      "epoch": 0.04767580452920143,
      "grad_norm": 1.1434534788131714,
      "learning_rate": 0.0001,
      "loss": 2.5641,
      "step": 10
    },
    {
      "epoch": 0.052443384982121574,
      "grad_norm": 1.2260290384292603,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.7242,
      "step": 11
    },
    {
      "epoch": 0.057210965435041714,
      "grad_norm": 1.1492360830307007,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.6978,
      "step": 12
    },
    {
      "epoch": 0.06197854588796186,
      "grad_norm": 1.1313751935958862,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.4025,
      "step": 13
    },
    {
      "epoch": 0.066746126340882,
      "grad_norm": 1.216576099395752,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.5651,
      "step": 14
    },
    {
      "epoch": 0.07151370679380215,
      "grad_norm": 1.0459645986557007,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.5407,
      "step": 15
    },
    {
      "epoch": 0.07628128724672228,
      "grad_norm": 1.1250581741333008,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.5113,
      "step": 16
    },
    {
      "epoch": 0.08104886769964244,
      "grad_norm": 1.1516832113265991,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.5224,
      "step": 17
    },
    {
      "epoch": 0.08581644815256258,
      "grad_norm": 0.9984982013702393,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.5051,
      "step": 18
    },
    {
      "epoch": 0.09058402860548272,
      "grad_norm": 0.9816920161247253,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.5969,
      "step": 19
    },
    {
      "epoch": 0.09535160905840286,
      "grad_norm": 1.113448977470398,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.4411,
      "step": 20
    },
    {
      "epoch": 0.10011918951132301,
      "grad_norm": 1.2237359285354614,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.5044,
      "step": 21
    },
    {
      "epoch": 0.10488676996424315,
      "grad_norm": 1.055956482887268,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3804,
      "step": 22
    },
    {
      "epoch": 0.10965435041716329,
      "grad_norm": 1.0263848304748535,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.3218,
      "step": 23
    },
    {
      "epoch": 0.11442193087008343,
      "grad_norm": 1.0815330743789673,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.4918,
      "step": 24
    },
    {
      "epoch": 0.11918951132300358,
      "grad_norm": 1.05085289478302,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.2042,
      "step": 25
    },
    {
      "epoch": 0.12395709177592372,
      "grad_norm": 1.147608757019043,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.4462,
      "step": 26
    },
    {
      "epoch": 0.12872467222884387,
      "grad_norm": 1.1940661668777466,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.4366,
      "step": 27
    },
    {
      "epoch": 0.133492252681764,
      "grad_norm": 1.2002006769180298,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.3046,
      "step": 28
    },
    {
      "epoch": 0.13825983313468415,
      "grad_norm": 1.225758671760559,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.344,
      "step": 29
    },
    {
      "epoch": 0.1430274135876043,
      "grad_norm": 1.453913927078247,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.3262,
      "step": 30
    },
    {
      "epoch": 0.14779499404052443,
      "grad_norm": 1.2313761711120605,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2569,
      "step": 31
    },
    {
      "epoch": 0.15256257449344457,
      "grad_norm": 1.3174500465393066,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.0015,
      "step": 32
    },
    {
      "epoch": 0.1573301549463647,
      "grad_norm": 1.4254320859909058,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.2475,
      "step": 33
    },
    {
      "epoch": 0.16209773539928488,
      "grad_norm": 2.1407878398895264,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.4093,
      "step": 34
    },
    {
      "epoch": 0.16686531585220502,
      "grad_norm": 1.4871824979782104,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.2877,
      "step": 35
    },
    {
      "epoch": 0.17163289630512515,
      "grad_norm": 1.8685574531555176,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.1265,
      "step": 36
    },
    {
      "epoch": 0.1764004767580453,
      "grad_norm": 1.6461660861968994,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.4455,
      "step": 37
    },
    {
      "epoch": 0.18116805721096543,
      "grad_norm": 1.6946738958358765,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.4016,
      "step": 38
    },
    {
      "epoch": 0.18593563766388557,
      "grad_norm": 1.6008548736572266,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.5618,
      "step": 39
    },
    {
      "epoch": 0.1907032181168057,
      "grad_norm": 1.5371747016906738,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.4849,
      "step": 40
    },
    {
      "epoch": 0.19547079856972585,
      "grad_norm": 1.770545244216919,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.3975,
      "step": 41
    },
    {
      "epoch": 0.20023837902264602,
      "grad_norm": 1.924229621887207,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.7272,
      "step": 42
    },
    {
      "epoch": 0.20500595947556616,
      "grad_norm": 1.900739073753357,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.8349,
      "step": 43
    },
    {
      "epoch": 0.2097735399284863,
      "grad_norm": 2.0874862670898438,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.7109,
      "step": 44
    },
    {
      "epoch": 0.21454112038140644,
      "grad_norm": 1.8650916814804077,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.6047,
      "step": 45
    },
    {
      "epoch": 0.21930870083432658,
      "grad_norm": 2.0557563304901123,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.5962,
      "step": 46
    },
    {
      "epoch": 0.22407628128724671,
      "grad_norm": 1.7432714700698853,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.3363,
      "step": 47
    },
    {
      "epoch": 0.22884386174016685,
      "grad_norm": 2.252993583679199,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.5628,
      "step": 48
    },
    {
      "epoch": 0.23361144219308702,
      "grad_norm": 2.2323484420776367,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.4012,
      "step": 49
    },
    {
      "epoch": 0.23837902264600716,
      "grad_norm": 4.776430606842041,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.2413,
      "step": 50
    },
    {
      "epoch": 0.23837902264600716,
      "eval_loss": 2.505044460296631,
      "eval_runtime": 25.1253,
      "eval_samples_per_second": 14.089,
      "eval_steps_per_second": 3.542,
      "step": 50
    },
    {
      "epoch": 0.2431466030989273,
      "grad_norm": 1.8412961959838867,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.4755,
      "step": 51
    },
    {
      "epoch": 0.24791418355184744,
      "grad_norm": 1.689751148223877,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.2588,
      "step": 52
    },
    {
      "epoch": 0.2526817640047676,
      "grad_norm": 0.9649447202682495,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1355,
      "step": 53
    },
    {
      "epoch": 0.25744934445768775,
      "grad_norm": 0.8039404153823853,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.256,
      "step": 54
    },
    {
      "epoch": 0.2622169249106079,
      "grad_norm": 0.900518000125885,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.3378,
      "step": 55
    },
    {
      "epoch": 0.266984505363528,
      "grad_norm": 0.8476229310035706,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.2062,
      "step": 56
    },
    {
      "epoch": 0.27175208581644816,
      "grad_norm": 0.8238363862037659,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3695,
      "step": 57
    },
    {
      "epoch": 0.2765196662693683,
      "grad_norm": 0.7881808876991272,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.3642,
      "step": 58
    },
    {
      "epoch": 0.28128724672228844,
      "grad_norm": 0.7238750457763672,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.1443,
      "step": 59
    },
    {
      "epoch": 0.2860548271752086,
      "grad_norm": 0.6919417977333069,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.1517,
      "step": 60
    },
    {
      "epoch": 0.2908224076281287,
      "grad_norm": 0.6629623174667358,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.082,
      "step": 61
    },
    {
      "epoch": 0.29558998808104886,
      "grad_norm": 0.7471420764923096,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.3289,
      "step": 62
    },
    {
      "epoch": 0.300357568533969,
      "grad_norm": 0.6996567845344543,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.1893,
      "step": 63
    },
    {
      "epoch": 0.30512514898688914,
      "grad_norm": 0.811893105506897,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.4038,
      "step": 64
    },
    {
      "epoch": 0.3098927294398093,
      "grad_norm": 0.7495221495628357,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.111,
      "step": 65
    },
    {
      "epoch": 0.3146603098927294,
      "grad_norm": 0.8111178874969482,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3879,
      "step": 66
    },
    {
      "epoch": 0.31942789034564956,
      "grad_norm": 0.73891681432724,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.2142,
      "step": 67
    },
    {
      "epoch": 0.32419547079856975,
      "grad_norm": 0.7016865611076355,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.1581,
      "step": 68
    },
    {
      "epoch": 0.3289630512514899,
      "grad_norm": 0.820429265499115,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.0799,
      "step": 69
    },
    {
      "epoch": 0.33373063170441003,
      "grad_norm": 0.7954638600349426,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.0854,
      "step": 70
    },
    {
      "epoch": 0.33849821215733017,
      "grad_norm": 0.7249769568443298,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.0811,
      "step": 71
    },
    {
      "epoch": 0.3432657926102503,
      "grad_norm": 0.798295795917511,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.281,
      "step": 72
    },
    {
      "epoch": 0.34803337306317045,
      "grad_norm": 0.8393540978431702,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.2054,
      "step": 73
    },
    {
      "epoch": 0.3528009535160906,
      "grad_norm": 0.820716917514801,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.182,
      "step": 74
    },
    {
      "epoch": 0.3575685339690107,
      "grad_norm": 0.9036383628845215,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.0523,
      "step": 75
    },
    {
      "epoch": 0.36233611442193087,
      "grad_norm": 0.9055408835411072,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.3064,
      "step": 76
    },
    {
      "epoch": 0.367103694874851,
      "grad_norm": 1.0610731840133667,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.0459,
      "step": 77
    },
    {
      "epoch": 0.37187127532777114,
      "grad_norm": 0.9533140659332275,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.1942,
      "step": 78
    },
    {
      "epoch": 0.3766388557806913,
      "grad_norm": 0.9222361445426941,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.138,
      "step": 79
    },
    {
      "epoch": 0.3814064362336114,
      "grad_norm": 0.9380738139152527,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.0383,
      "step": 80
    },
    {
      "epoch": 0.38617401668653156,
      "grad_norm": 1.0802791118621826,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.0009,
      "step": 81
    },
    {
      "epoch": 0.3909415971394517,
      "grad_norm": 1.0297811031341553,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.0742,
      "step": 82
    },
    {
      "epoch": 0.3957091775923719,
      "grad_norm": 1.0209712982177734,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.2836,
      "step": 83
    },
    {
      "epoch": 0.40047675804529204,
      "grad_norm": 1.1667689085006714,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.0793,
      "step": 84
    },
    {
      "epoch": 0.4052443384982122,
      "grad_norm": 1.1458278894424438,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.1959,
      "step": 85
    },
    {
      "epoch": 0.4100119189511323,
      "grad_norm": 1.1719489097595215,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.0824,
      "step": 86
    },
    {
      "epoch": 0.41477949940405245,
      "grad_norm": 1.2250354290008545,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.5144,
      "step": 87
    },
    {
      "epoch": 0.4195470798569726,
      "grad_norm": 1.2479459047317505,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.1278,
      "step": 88
    },
    {
      "epoch": 0.42431466030989273,
      "grad_norm": 1.448906660079956,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.3074,
      "step": 89
    },
    {
      "epoch": 0.42908224076281287,
      "grad_norm": 1.4189363718032837,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.4023,
      "step": 90
    },
    {
      "epoch": 0.433849821215733,
      "grad_norm": 1.4955360889434814,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.1318,
      "step": 91
    },
    {
      "epoch": 0.43861740166865315,
      "grad_norm": 1.3838410377502441,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.0099,
      "step": 92
    },
    {
      "epoch": 0.4433849821215733,
      "grad_norm": 1.5779086351394653,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.5704,
      "step": 93
    },
    {
      "epoch": 0.44815256257449343,
      "grad_norm": 1.6373097896575928,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.3962,
      "step": 94
    },
    {
      "epoch": 0.45292014302741357,
      "grad_norm": 1.5908904075622559,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.1469,
      "step": 95
    },
    {
      "epoch": 0.4576877234803337,
      "grad_norm": 1.9041376113891602,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.6006,
      "step": 96
    },
    {
      "epoch": 0.46245530393325385,
      "grad_norm": 2.09298038482666,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.3567,
      "step": 97
    },
    {
      "epoch": 0.46722288438617404,
      "grad_norm": 1.851805329322815,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.1391,
      "step": 98
    },
    {
      "epoch": 0.4719904648390942,
      "grad_norm": 2.135078191757202,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.4175,
      "step": 99
    },
    {
      "epoch": 0.4767580452920143,
      "grad_norm": 3.424588203430176,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4322,
      "step": 100
    },
    {
      "epoch": 0.4767580452920143,
      "eval_loss": 2.311724901199341,
      "eval_runtime": 25.1495,
      "eval_samples_per_second": 14.076,
      "eval_steps_per_second": 3.539,
      "step": 100
    },
    {
      "epoch": 0.48152562574493446,
      "grad_norm": 1.1873295307159424,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.3272,
      "step": 101
    },
    {
      "epoch": 0.4862932061978546,
      "grad_norm": 0.9908881187438965,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.1446,
      "step": 102
    },
    {
      "epoch": 0.49106078665077474,
      "grad_norm": 0.9470567107200623,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.1245,
      "step": 103
    },
    {
      "epoch": 0.4958283671036949,
      "grad_norm": 0.8336878418922424,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.2065,
      "step": 104
    },
    {
      "epoch": 0.5005959475566151,
      "grad_norm": 0.682320773601532,
      "learning_rate": 5e-05,
      "loss": 1.9701,
      "step": 105
    },
    {
      "epoch": 0.5053635280095352,
      "grad_norm": 0.6770563125610352,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.0406,
      "step": 106
    },
    {
      "epoch": 0.5101311084624554,
      "grad_norm": 0.7286104559898376,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.148,
      "step": 107
    },
    {
      "epoch": 0.5148986889153755,
      "grad_norm": 0.7244910001754761,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.3225,
      "step": 108
    },
    {
      "epoch": 0.5196662693682956,
      "grad_norm": 0.7211068272590637,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.119,
      "step": 109
    },
    {
      "epoch": 0.5244338498212158,
      "grad_norm": 0.7703940272331238,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.1895,
      "step": 110
    },
    {
      "epoch": 0.5292014302741359,
      "grad_norm": 0.7164751887321472,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.1936,
      "step": 111
    },
    {
      "epoch": 0.533969010727056,
      "grad_norm": 0.7430546283721924,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.2136,
      "step": 112
    },
    {
      "epoch": 0.5387365911799762,
      "grad_norm": 0.8059442639350891,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.2511,
      "step": 113
    },
    {
      "epoch": 0.5435041716328963,
      "grad_norm": 0.7626630663871765,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.1158,
      "step": 114
    },
    {
      "epoch": 0.5482717520858165,
      "grad_norm": 0.734927237033844,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.0999,
      "step": 115
    },
    {
      "epoch": 0.5530393325387366,
      "grad_norm": 0.8071227669715881,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.2868,
      "step": 116
    },
    {
      "epoch": 0.5578069129916567,
      "grad_norm": 0.767591118812561,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.108,
      "step": 117
    },
    {
      "epoch": 0.5625744934445769,
      "grad_norm": 0.762491762638092,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.0989,
      "step": 118
    },
    {
      "epoch": 0.567342073897497,
      "grad_norm": 0.7456633448600769,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.0387,
      "step": 119
    },
    {
      "epoch": 0.5721096543504172,
      "grad_norm": 0.822797417640686,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.3575,
      "step": 120
    },
    {
      "epoch": 0.5768772348033373,
      "grad_norm": 0.8068381547927856,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.9786,
      "step": 121
    },
    {
      "epoch": 0.5816448152562574,
      "grad_norm": 0.8712824583053589,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.0851,
      "step": 122
    },
    {
      "epoch": 0.5864123957091776,
      "grad_norm": 0.8513266444206238,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.1167,
      "step": 123
    },
    {
      "epoch": 0.5911799761620977,
      "grad_norm": 0.8698190450668335,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.085,
      "step": 124
    },
    {
      "epoch": 0.5959475566150179,
      "grad_norm": 0.9339384436607361,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.0645,
      "step": 125
    },
    {
      "epoch": 0.600715137067938,
      "grad_norm": 0.9093461632728577,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.1754,
      "step": 126
    },
    {
      "epoch": 0.6054827175208581,
      "grad_norm": 0.998595654964447,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.3316,
      "step": 127
    },
    {
      "epoch": 0.6102502979737783,
      "grad_norm": 0.9968245029449463,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.0921,
      "step": 128
    },
    {
      "epoch": 0.6150178784266984,
      "grad_norm": 0.994280993938446,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.111,
      "step": 129
    },
    {
      "epoch": 0.6197854588796186,
      "grad_norm": 0.9800683856010437,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.13,
      "step": 130
    },
    {
      "epoch": 0.6245530393325387,
      "grad_norm": 1.0401815176010132,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.887,
      "step": 131
    },
    {
      "epoch": 0.6293206197854588,
      "grad_norm": 1.0081592798233032,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.0431,
      "step": 132
    },
    {
      "epoch": 0.634088200238379,
      "grad_norm": 1.05696702003479,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.9861,
      "step": 133
    },
    {
      "epoch": 0.6388557806912991,
      "grad_norm": 1.148400068283081,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.0777,
      "step": 134
    },
    {
      "epoch": 0.6436233611442194,
      "grad_norm": 1.1722474098205566,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.1075,
      "step": 135
    },
    {
      "epoch": 0.6483909415971395,
      "grad_norm": 1.175913691520691,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.9695,
      "step": 136
    },
    {
      "epoch": 0.6531585220500596,
      "grad_norm": 1.1160712242126465,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.0807,
      "step": 137
    },
    {
      "epoch": 0.6579261025029798,
      "grad_norm": 1.238513708114624,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.8404,
      "step": 138
    },
    {
      "epoch": 0.6626936829558999,
      "grad_norm": 1.2601982355117798,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.4115,
      "step": 139
    },
    {
      "epoch": 0.6674612634088201,
      "grad_norm": 1.3604203462600708,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.0889,
      "step": 140
    },
    {
      "epoch": 0.6722288438617402,
      "grad_norm": 1.4191426038742065,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.4819,
      "step": 141
    },
    {
      "epoch": 0.6769964243146603,
      "grad_norm": 1.400758147239685,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.3057,
      "step": 142
    },
    {
      "epoch": 0.6817640047675805,
      "grad_norm": 1.4623340368270874,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.3842,
      "step": 143
    },
    {
      "epoch": 0.6865315852205006,
      "grad_norm": 1.6199415922164917,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.5805,
      "step": 144
    },
    {
      "epoch": 0.6912991656734208,
      "grad_norm": 1.7270547151565552,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.6062,
      "step": 145
    },
    {
      "epoch": 0.6960667461263409,
      "grad_norm": 1.7278194427490234,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.5503,
      "step": 146
    },
    {
      "epoch": 0.700834326579261,
      "grad_norm": 1.6305183172225952,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.2978,
      "step": 147
    },
    {
      "epoch": 0.7056019070321812,
      "grad_norm": 1.9193475246429443,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0608,
      "step": 148
    },
    {
      "epoch": 0.7103694874851013,
      "grad_norm": 2.0708417892456055,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.222,
      "step": 149
    },
    {
      "epoch": 0.7151370679380215,
      "grad_norm": 2.8837761878967285,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.0448,
      "step": 150
    },
    {
      "epoch": 0.7151370679380215,
      "eval_loss": 2.1478922367095947,
      "eval_runtime": 25.163,
      "eval_samples_per_second": 14.068,
      "eval_steps_per_second": 3.537,
      "step": 150
    },
    {
      "epoch": 0.7199046483909416,
      "grad_norm": 0.5570347309112549,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.2434,
      "step": 151
    },
    {
      "epoch": 0.7246722288438617,
      "grad_norm": 0.5925551056861877,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9218,
      "step": 152
    },
    {
      "epoch": 0.7294398092967819,
      "grad_norm": 0.7982468008995056,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.9788,
      "step": 153
    },
    {
      "epoch": 0.734207389749702,
      "grad_norm": 0.8105083107948303,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.915,
      "step": 154
    },
    {
      "epoch": 0.7389749702026222,
      "grad_norm": 0.8352587223052979,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.2492,
      "step": 155
    },
    {
      "epoch": 0.7437425506555423,
      "grad_norm": 0.674999475479126,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.3595,
      "step": 156
    },
    {
      "epoch": 0.7485101311084624,
      "grad_norm": 0.7149543166160583,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.1465,
      "step": 157
    },
    {
      "epoch": 0.7532777115613826,
      "grad_norm": 0.723102331161499,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.1712,
      "step": 158
    },
    {
      "epoch": 0.7580452920143027,
      "grad_norm": 0.6700135469436646,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.2641,
      "step": 159
    },
    {
      "epoch": 0.7628128724672228,
      "grad_norm": 0.6606097221374512,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.0487,
      "step": 160
    },
    {
      "epoch": 0.767580452920143,
      "grad_norm": 0.7337142825126648,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.1306,
      "step": 161
    },
    {
      "epoch": 0.7723480333730631,
      "grad_norm": 0.6384457349777222,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.0698,
      "step": 162
    },
    {
      "epoch": 0.7771156138259833,
      "grad_norm": 0.6712755560874939,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.0717,
      "step": 163
    },
    {
      "epoch": 0.7818831942789034,
      "grad_norm": 0.7360457181930542,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.1774,
      "step": 164
    },
    {
      "epoch": 0.7866507747318237,
      "grad_norm": 0.724876344203949,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.0904,
      "step": 165
    },
    {
      "epoch": 0.7914183551847438,
      "grad_norm": 0.8232304453849792,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.2429,
      "step": 166
    },
    {
      "epoch": 0.7961859356376639,
      "grad_norm": 0.8048411011695862,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.0947,
      "step": 167
    },
    {
      "epoch": 0.8009535160905841,
      "grad_norm": 0.7875757217407227,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.9854,
      "step": 168
    },
    {
      "epoch": 0.8057210965435042,
      "grad_norm": 0.8814699649810791,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0022,
      "step": 169
    },
    {
      "epoch": 0.8104886769964244,
      "grad_norm": 0.8261225819587708,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.1227,
      "step": 170
    },
    {
      "epoch": 0.8152562574493445,
      "grad_norm": 0.8244904279708862,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.1583,
      "step": 171
    },
    {
      "epoch": 0.8200238379022646,
      "grad_norm": 0.8068302869796753,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.0933,
      "step": 172
    },
    {
      "epoch": 0.8247914183551848,
      "grad_norm": 0.8379300236701965,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.152,
      "step": 173
    },
    {
      "epoch": 0.8295589988081049,
      "grad_norm": 0.9288865923881531,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.0066,
      "step": 174
    },
    {
      "epoch": 0.834326579261025,
      "grad_norm": 0.9022939205169678,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.1387,
      "step": 175
    },
    {
      "epoch": 0.8390941597139452,
      "grad_norm": 0.9126566052436829,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.0102,
      "step": 176
    },
    {
      "epoch": 0.8438617401668653,
      "grad_norm": 0.9721250534057617,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.2065,
      "step": 177
    },
    {
      "epoch": 0.8486293206197855,
      "grad_norm": 0.9469491243362427,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.09,
      "step": 178
    },
    {
      "epoch": 0.8533969010727056,
      "grad_norm": 0.9847070574760437,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.0802,
      "step": 179
    },
    {
      "epoch": 0.8581644815256257,
      "grad_norm": 1.0105953216552734,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.9934,
      "step": 180
    },
    {
      "epoch": 0.8629320619785459,
      "grad_norm": 1.0612765550613403,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.7581,
      "step": 181
    },
    {
      "epoch": 0.867699642431466,
      "grad_norm": 1.0604398250579834,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1245,
      "step": 182
    },
    {
      "epoch": 0.8724672228843862,
      "grad_norm": 1.0303086042404175,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.1671,
      "step": 183
    },
    {
      "epoch": 0.8772348033373063,
      "grad_norm": 1.0367300510406494,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9276,
      "step": 184
    },
    {
      "epoch": 0.8820023837902264,
      "grad_norm": 1.1234383583068848,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.9505,
      "step": 185
    },
    {
      "epoch": 0.8867699642431466,
      "grad_norm": 1.2010043859481812,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.1285,
      "step": 186
    },
    {
      "epoch": 0.8915375446960667,
      "grad_norm": 1.1489535570144653,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.9923,
      "step": 187
    },
    {
      "epoch": 0.8963051251489869,
      "grad_norm": 1.2861065864562988,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.1916,
      "step": 188
    },
    {
      "epoch": 0.901072705601907,
      "grad_norm": 1.2785236835479736,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.185,
      "step": 189
    },
    {
      "epoch": 0.9058402860548271,
      "grad_norm": 1.339624285697937,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.3496,
      "step": 190
    },
    {
      "epoch": 0.9106078665077473,
      "grad_norm": 1.5450823307037354,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.4015,
      "step": 191
    },
    {
      "epoch": 0.9153754469606674,
      "grad_norm": 1.3474223613739014,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.2637,
      "step": 192
    },
    {
      "epoch": 0.9201430274135876,
      "grad_norm": 1.6195087432861328,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.4341,
      "step": 193
    },
    {
      "epoch": 0.9249106078665077,
      "grad_norm": 1.5226205587387085,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.4441,
      "step": 194
    },
    {
      "epoch": 0.929678188319428,
      "grad_norm": 1.7517061233520508,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.517,
      "step": 195
    },
    {
      "epoch": 0.9344457687723481,
      "grad_norm": 1.7285345792770386,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.4959,
      "step": 196
    },
    {
      "epoch": 0.9392133492252682,
      "grad_norm": 1.9032076597213745,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.4887,
      "step": 197
    },
    {
      "epoch": 0.9439809296781884,
      "grad_norm": 2.0340332984924316,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.1746,
      "step": 198
    },
    {
      "epoch": 0.9487485101311085,
      "grad_norm": 2.3427014350891113,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.1673,
      "step": 199
    },
    {
      "epoch": 0.9535160905840286,
      "grad_norm": 3.5722577571868896,
      "learning_rate": 0.0,
      "loss": 2.2554,
      "step": 200
    },
    {
      "epoch": 0.9535160905840286,
      "eval_loss": 2.1066830158233643,
      "eval_runtime": 25.1636,
      "eval_samples_per_second": 14.068,
      "eval_steps_per_second": 3.537,
      "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": 2.7075670040169677e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}