{
  "best_metric": 1.8008776903152466,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.05356903709655819,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00026784518548279097,
      "grad_norm": 3.333930253982544,
      "learning_rate": 1e-05,
      "loss": 1.4524,
      "step": 1
    },
    {
      "epoch": 0.00026784518548279097,
      "eval_loss": 3.339059352874756,
      "eval_runtime": 377.673,
      "eval_samples_per_second": 16.649,
      "eval_steps_per_second": 4.162,
      "step": 1
    },
    {
      "epoch": 0.0005356903709655819,
      "grad_norm": 3.44179105758667,
      "learning_rate": 2e-05,
      "loss": 1.5758,
      "step": 2
    },
    {
      "epoch": 0.0008035355564483728,
      "grad_norm": 2.6693384647369385,
      "learning_rate": 3e-05,
      "loss": 1.4645,
      "step": 3
    },
    {
      "epoch": 0.0010713807419311639,
      "grad_norm": 1.710910677909851,
      "learning_rate": 4e-05,
      "loss": 1.5269,
      "step": 4
    },
    {
      "epoch": 0.0013392259274139548,
      "grad_norm": 3.0236446857452393,
      "learning_rate": 5e-05,
      "loss": 1.516,
      "step": 5
    },
    {
      "epoch": 0.0016070711128967456,
      "grad_norm": 2.9545600414276123,
      "learning_rate": 6e-05,
      "loss": 1.7818,
      "step": 6
    },
    {
      "epoch": 0.0018749162983795367,
      "grad_norm": 1.961796522140503,
      "learning_rate": 7e-05,
      "loss": 1.8019,
      "step": 7
    },
    {
      "epoch": 0.0021427614838623278,
      "grad_norm": 1.8427914381027222,
      "learning_rate": 8e-05,
      "loss": 1.945,
      "step": 8
    },
    {
      "epoch": 0.0024106066693451184,
      "grad_norm": 1.7744075059890747,
      "learning_rate": 9e-05,
      "loss": 1.742,
      "step": 9
    },
    {
      "epoch": 0.0026784518548279095,
      "grad_norm": 1.4324676990509033,
      "learning_rate": 0.0001,
      "loss": 1.7219,
      "step": 10
    },
    {
      "epoch": 0.0029462970403107006,
      "grad_norm": 0.94417405128479,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7778,
      "step": 11
    },
    {
      "epoch": 0.0032141422257934912,
      "grad_norm": 1.920737624168396,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.7155,
      "step": 12
    },
    {
      "epoch": 0.0034819874112762823,
      "grad_norm": 1.378326416015625,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.8534,
      "step": 13
    },
    {
      "epoch": 0.0037498325967590734,
      "grad_norm": 1.459309458732605,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.8655,
      "step": 14
    },
    {
      "epoch": 0.004017677782241864,
      "grad_norm": 1.5178602933883667,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.6379,
      "step": 15
    },
    {
      "epoch": 0.0042855229677246556,
      "grad_norm": 0.7386080026626587,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7265,
      "step": 16
    },
    {
      "epoch": 0.004553368153207446,
      "grad_norm": 1.2269948720932007,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.7025,
      "step": 17
    },
    {
      "epoch": 0.004821213338690237,
      "grad_norm": 0.905255138874054,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.685,
      "step": 18
    },
    {
      "epoch": 0.005089058524173028,
      "grad_norm": 0.8691777586936951,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.7192,
      "step": 19
    },
    {
      "epoch": 0.005356903709655819,
      "grad_norm": 0.9960977435112,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.7914,
      "step": 20
    },
    {
      "epoch": 0.00562474889513861,
      "grad_norm": 0.9538270235061646,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.9184,
      "step": 21
    },
    {
      "epoch": 0.005892594080621401,
      "grad_norm": 1.0331635475158691,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.8969,
      "step": 22
    },
    {
      "epoch": 0.006160439266104192,
      "grad_norm": 0.8776848316192627,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.9563,
      "step": 23
    },
    {
      "epoch": 0.0064282844515869825,
      "grad_norm": 0.8616760969161987,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.8526,
      "step": 24
    },
    {
      "epoch": 0.006696129637069774,
      "grad_norm": 0.7754805684089661,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.8823,
      "step": 25
    },
    {
      "epoch": 0.006963974822552565,
      "grad_norm": 1.0088698863983154,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.9282,
      "step": 26
    },
    {
      "epoch": 0.007231820008035355,
      "grad_norm": 0.933375895023346,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.0393,
      "step": 27
    },
    {
      "epoch": 0.007499665193518147,
      "grad_norm": 0.9726508855819702,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.0976,
      "step": 28
    },
    {
      "epoch": 0.0077675103790009374,
      "grad_norm": 1.098980188369751,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.0853,
      "step": 29
    },
    {
      "epoch": 0.008035355564483728,
      "grad_norm": 1.2418533563613892,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.057,
      "step": 30
    },
    {
      "epoch": 0.00830320074996652,
      "grad_norm": 1.032130479812622,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2126,
      "step": 31
    },
    {
      "epoch": 0.008571045935449311,
      "grad_norm": 1.275407314300537,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.3442,
      "step": 32
    },
    {
      "epoch": 0.008838891120932101,
      "grad_norm": 0.9324800372123718,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.1617,
      "step": 33
    },
    {
      "epoch": 0.009106736306414892,
      "grad_norm": 0.9083219170570374,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.2688,
      "step": 34
    },
    {
      "epoch": 0.009374581491897684,
      "grad_norm": 1.0046746730804443,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.0271,
      "step": 35
    },
    {
      "epoch": 0.009642426677380474,
      "grad_norm": 1.07561194896698,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.0521,
      "step": 36
    },
    {
      "epoch": 0.009910271862863265,
      "grad_norm": 1.2042580842971802,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.0862,
      "step": 37
    },
    {
      "epoch": 0.010178117048346057,
      "grad_norm": 1.1449425220489502,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.1439,
      "step": 38
    },
    {
      "epoch": 0.010445962233828847,
      "grad_norm": 1.4371974468231201,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.0648,
      "step": 39
    },
    {
      "epoch": 0.010713807419311638,
      "grad_norm": 1.1394928693771362,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.1797,
      "step": 40
    },
    {
      "epoch": 0.01098165260479443,
      "grad_norm": 1.1724424362182617,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.1833,
      "step": 41
    },
    {
      "epoch": 0.01124949779027722,
      "grad_norm": 1.2767730951309204,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.1733,
      "step": 42
    },
    {
      "epoch": 0.01151734297576001,
      "grad_norm": 1.176896333694458,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.2022,
      "step": 43
    },
    {
      "epoch": 0.011785188161242802,
      "grad_norm": 2.1997969150543213,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.3014,
      "step": 44
    },
    {
      "epoch": 0.012053033346725592,
      "grad_norm": 2.0579676628112793,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.3595,
      "step": 45
    },
    {
      "epoch": 0.012320878532208384,
      "grad_norm": 2.482654571533203,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.4269,
      "step": 46
    },
    {
      "epoch": 0.012588723717691175,
      "grad_norm": 1.8847991228103638,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.31,
      "step": 47
    },
    {
      "epoch": 0.012856568903173965,
      "grad_norm": 2.7223808765411377,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.458,
      "step": 48
    },
    {
      "epoch": 0.013124414088656756,
      "grad_norm": 2.435748815536499,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.448,
      "step": 49
    },
    {
      "epoch": 0.013392259274139548,
      "grad_norm": 6.3010478019714355,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.8487,
      "step": 50
    },
    {
      "epoch": 0.013392259274139548,
      "eval_loss": 3.23032546043396,
      "eval_runtime": 378.6302,
      "eval_samples_per_second": 16.607,
      "eval_steps_per_second": 4.152,
      "step": 50
    },
    {
      "epoch": 0.013660104459622338,
      "grad_norm": 35.09332275390625,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.1518,
      "step": 51
    },
    {
      "epoch": 0.01392794964510513,
      "grad_norm": 15.837681770324707,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.2305,
      "step": 52
    },
    {
      "epoch": 0.01419579483058792,
      "grad_norm": 7.110929489135742,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5328,
      "step": 53
    },
    {
      "epoch": 0.01446364001607071,
      "grad_norm": 2.344552993774414,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3117,
      "step": 54
    },
    {
      "epoch": 0.014731485201553502,
      "grad_norm": 1.6074857711791992,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.2791,
      "step": 55
    },
    {
      "epoch": 0.014999330387036294,
      "grad_norm": 0.9663664102554321,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.2976,
      "step": 56
    },
    {
      "epoch": 0.015267175572519083,
      "grad_norm": 0.7298042178153992,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.2671,
      "step": 57
    },
    {
      "epoch": 0.015535020758001875,
      "grad_norm": 0.6781096458435059,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4487,
      "step": 58
    },
    {
      "epoch": 0.015802865943484665,
      "grad_norm": 0.6039943099021912,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4686,
      "step": 59
    },
    {
      "epoch": 0.016070711128967456,
      "grad_norm": 0.5714627504348755,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5361,
      "step": 60
    },
    {
      "epoch": 0.016338556314450248,
      "grad_norm": 0.5353637337684631,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4845,
      "step": 61
    },
    {
      "epoch": 0.01660640149993304,
      "grad_norm": 0.586633026599884,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.5548,
      "step": 62
    },
    {
      "epoch": 0.01687424668541583,
      "grad_norm": 0.5950109362602234,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.5925,
      "step": 63
    },
    {
      "epoch": 0.017142091870898622,
      "grad_norm": 0.513549268245697,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5946,
      "step": 64
    },
    {
      "epoch": 0.01740993705638141,
      "grad_norm": 0.4591107964515686,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.5853,
      "step": 65
    },
    {
      "epoch": 0.017677782241864202,
      "grad_norm": 0.5080562233924866,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5563,
      "step": 66
    },
    {
      "epoch": 0.017945627427346993,
      "grad_norm": 0.46998336911201477,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.6126,
      "step": 67
    },
    {
      "epoch": 0.018213472612829785,
      "grad_norm": 0.573527455329895,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.6408,
      "step": 68
    },
    {
      "epoch": 0.018481317798312576,
      "grad_norm": 0.543536901473999,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.9348,
      "step": 69
    },
    {
      "epoch": 0.018749162983795368,
      "grad_norm": 0.5021911859512329,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6445,
      "step": 70
    },
    {
      "epoch": 0.019017008169278156,
      "grad_norm": 0.6000838279724121,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.8059,
      "step": 71
    },
    {
      "epoch": 0.019284853354760947,
      "grad_norm": 0.645431399345398,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6823,
      "step": 72
    },
    {
      "epoch": 0.01955269854024374,
      "grad_norm": 0.6973119378089905,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.9438,
      "step": 73
    },
    {
      "epoch": 0.01982054372572653,
      "grad_norm": 0.6624037027359009,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.8866,
      "step": 74
    },
    {
      "epoch": 0.020088388911209322,
      "grad_norm": 0.7418327927589417,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.7553,
      "step": 75
    },
    {
      "epoch": 0.020356234096692113,
      "grad_norm": 0.8361924886703491,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.0092,
      "step": 76
    },
    {
      "epoch": 0.0206240792821749,
      "grad_norm": 0.72418612241745,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7896,
      "step": 77
    },
    {
      "epoch": 0.020891924467657693,
      "grad_norm": 0.713162362575531,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.1179,
      "step": 78
    },
    {
      "epoch": 0.021159769653140485,
      "grad_norm": 0.7853183150291443,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.0416,
      "step": 79
    },
    {
      "epoch": 0.021427614838623276,
      "grad_norm": 0.7384597063064575,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.097,
      "step": 80
    },
    {
      "epoch": 0.021695460024106068,
      "grad_norm": 0.7446250319480896,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.252,
      "step": 81
    },
    {
      "epoch": 0.02196330520958886,
      "grad_norm": 0.6318081021308899,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.1584,
      "step": 82
    },
    {
      "epoch": 0.022231150395071647,
      "grad_norm": 0.7878530025482178,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.0542,
      "step": 83
    },
    {
      "epoch": 0.02249899558055444,
      "grad_norm": 0.793310284614563,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.1004,
      "step": 84
    },
    {
      "epoch": 0.02276684076603723,
      "grad_norm": 1.1869111061096191,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.224,
      "step": 85
    },
    {
      "epoch": 0.02303468595152002,
      "grad_norm": 0.8399895429611206,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.1265,
      "step": 86
    },
    {
      "epoch": 0.023302531137002813,
      "grad_norm": 0.8845256567001343,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.1339,
      "step": 87
    },
    {
      "epoch": 0.023570376322485605,
      "grad_norm": 1.097377896308899,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.9944,
      "step": 88
    },
    {
      "epoch": 0.023838221507968393,
      "grad_norm": 1.1093826293945312,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.085,
      "step": 89
    },
    {
      "epoch": 0.024106066693451184,
      "grad_norm": 1.1365543603897095,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.9185,
      "step": 90
    },
    {
      "epoch": 0.024373911878933976,
      "grad_norm": 1.2219469547271729,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.134,
      "step": 91
    },
    {
      "epoch": 0.024641757064416767,
      "grad_norm": 1.3199540376663208,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.1947,
      "step": 92
    },
    {
      "epoch": 0.02490960224989956,
      "grad_norm": 0.9772177338600159,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.1418,
      "step": 93
    },
    {
      "epoch": 0.02517744743538235,
      "grad_norm": 1.0486177206039429,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.2436,
      "step": 94
    },
    {
      "epoch": 0.02544529262086514,
      "grad_norm": 1.3388665914535522,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.199,
      "step": 95
    },
    {
      "epoch": 0.02571313780634793,
      "grad_norm": 1.2399343252182007,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.3399,
      "step": 96
    },
    {
      "epoch": 0.02598098299183072,
      "grad_norm": 1.3912508487701416,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.3684,
      "step": 97
    },
    {
      "epoch": 0.026248828177313513,
      "grad_norm": 2.3723835945129395,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.4135,
      "step": 98
    },
    {
      "epoch": 0.026516673362796304,
      "grad_norm": 1.4856144189834595,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.371,
      "step": 99
    },
    {
      "epoch": 0.026784518548279096,
      "grad_norm": 1.8692500591278076,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4434,
      "step": 100
    },
    {
      "epoch": 0.026784518548279096,
      "eval_loss": 2.1365230083465576,
      "eval_runtime": 379.5769,
      "eval_samples_per_second": 16.566,
      "eval_steps_per_second": 4.141,
      "step": 100
    },
    {
      "epoch": 0.027052363733761884,
      "grad_norm": 4.3066325187683105,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2774,
      "step": 101
    },
    {
      "epoch": 0.027320208919244676,
      "grad_norm": 2.4925754070281982,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.0941,
      "step": 102
    },
    {
      "epoch": 0.027588054104727467,
      "grad_norm": 1.183905005455017,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.0709,
      "step": 103
    },
    {
      "epoch": 0.02785589929021026,
      "grad_norm": 0.8178330659866333,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.1397,
      "step": 104
    },
    {
      "epoch": 0.02812374447569305,
      "grad_norm": 0.7322461009025574,
      "learning_rate": 5e-05,
      "loss": 1.2571,
      "step": 105
    },
    {
      "epoch": 0.02839158966117584,
      "grad_norm": 0.6318075060844421,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.144,
      "step": 106
    },
    {
      "epoch": 0.02865943484665863,
      "grad_norm": 0.7195743918418884,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2815,
      "step": 107
    },
    {
      "epoch": 0.02892728003214142,
      "grad_norm": 0.7300739288330078,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4335,
      "step": 108
    },
    {
      "epoch": 0.029195125217624213,
      "grad_norm": 0.6241258382797241,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4808,
      "step": 109
    },
    {
      "epoch": 0.029462970403107004,
      "grad_norm": 0.4644920825958252,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5138,
      "step": 110
    },
    {
      "epoch": 0.029730815588589796,
      "grad_norm": 0.5784435272216797,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.4932,
      "step": 111
    },
    {
      "epoch": 0.029998660774072587,
      "grad_norm": 0.5770624279975891,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.552,
      "step": 112
    },
    {
      "epoch": 0.03026650595955538,
      "grad_norm": 0.6292299032211304,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.4303,
      "step": 113
    },
    {
      "epoch": 0.030534351145038167,
      "grad_norm": 0.6070420742034912,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6523,
      "step": 114
    },
    {
      "epoch": 0.030802196330520958,
      "grad_norm": 0.45281872153282166,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.6234,
      "step": 115
    },
    {
      "epoch": 0.03107004151600375,
      "grad_norm": 0.4802415370941162,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.7052,
      "step": 116
    },
    {
      "epoch": 0.03133788670148654,
      "grad_norm": 0.5472726821899414,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.6384,
      "step": 117
    },
    {
      "epoch": 0.03160573188696933,
      "grad_norm": 0.5683273077011108,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.688,
      "step": 118
    },
    {
      "epoch": 0.03187357707245212,
      "grad_norm": 0.5509693622589111,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7113,
      "step": 119
    },
    {
      "epoch": 0.03214142225793491,
      "grad_norm": 0.46628203988075256,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.7117,
      "step": 120
    },
    {
      "epoch": 0.032409267443417704,
      "grad_norm": 0.4309813380241394,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.8916,
      "step": 121
    },
    {
      "epoch": 0.032677112628900495,
      "grad_norm": 0.4455784857273102,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.788,
      "step": 122
    },
    {
      "epoch": 0.03294495781438329,
      "grad_norm": 0.5357702374458313,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.8375,
      "step": 123
    },
    {
      "epoch": 0.03321280299986608,
      "grad_norm": 0.693437397480011,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.8821,
      "step": 124
    },
    {
      "epoch": 0.03348064818534887,
      "grad_norm": 0.6140393018722534,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.788,
      "step": 125
    },
    {
      "epoch": 0.03374849337083166,
      "grad_norm": 0.5691301226615906,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7983,
      "step": 126
    },
    {
      "epoch": 0.03401633855631445,
      "grad_norm": 0.6249701976776123,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.8647,
      "step": 127
    },
    {
      "epoch": 0.034284183741797244,
      "grad_norm": 0.5885187983512878,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.989,
      "step": 128
    },
    {
      "epoch": 0.03455202892728003,
      "grad_norm": 0.6370279788970947,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.0277,
      "step": 129
    },
    {
      "epoch": 0.03481987411276282,
      "grad_norm": 0.7725704908370972,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.1655,
      "step": 130
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": 0.6694130897521973,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.8983,
      "step": 131
    },
    {
      "epoch": 0.035355564483728404,
      "grad_norm": 0.7064082026481628,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.1638,
      "step": 132
    },
    {
      "epoch": 0.035623409669211195,
      "grad_norm": 0.6084491610527039,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.092,
      "step": 133
    },
    {
      "epoch": 0.03589125485469399,
      "grad_norm": 0.8428956866264343,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.137,
      "step": 134
    },
    {
      "epoch": 0.03615910004017678,
      "grad_norm": 0.7592659592628479,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.1571,
      "step": 135
    },
    {
      "epoch": 0.03642694522565957,
      "grad_norm": 0.692391574382782,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.1446,
      "step": 136
    },
    {
      "epoch": 0.03669479041114236,
      "grad_norm": 0.6206789016723633,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.9945,
      "step": 137
    },
    {
      "epoch": 0.03696263559662515,
      "grad_norm": 0.9543753862380981,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.9802,
      "step": 138
    },
    {
      "epoch": 0.037230480782107944,
      "grad_norm": 0.8085548281669617,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.188,
      "step": 139
    },
    {
      "epoch": 0.037498325967590736,
      "grad_norm": 1.008013367652893,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.9366,
      "step": 140
    },
    {
      "epoch": 0.03776617115307352,
      "grad_norm": 1.034171462059021,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.1741,
      "step": 141
    },
    {
      "epoch": 0.03803401633855631,
      "grad_norm": 1.1292489767074585,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.1649,
      "step": 142
    },
    {
      "epoch": 0.0383018615240391,
      "grad_norm": 1.0804150104522705,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.0542,
      "step": 143
    },
    {
      "epoch": 0.038569706709521895,
      "grad_norm": 2.013336181640625,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.1298,
      "step": 144
    },
    {
      "epoch": 0.038837551895004686,
      "grad_norm": 1.199326753616333,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.1158,
      "step": 145
    },
    {
      "epoch": 0.03910539708048748,
      "grad_norm": 1.4698890447616577,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.279,
      "step": 146
    },
    {
      "epoch": 0.03937324226597027,
      "grad_norm": 1.5185235738754272,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.3614,
      "step": 147
    },
    {
      "epoch": 0.03964108745145306,
      "grad_norm": 1.9969290494918823,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.396,
      "step": 148
    },
    {
      "epoch": 0.03990893263693585,
      "grad_norm": 2.4296302795410156,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.3918,
      "step": 149
    },
    {
      "epoch": 0.040176777822418644,
      "grad_norm": 4.925536155700684,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.7508,
      "step": 150
    },
    {
      "epoch": 0.040176777822418644,
      "eval_loss": 1.9040244817733765,
      "eval_runtime": 378.7858,
      "eval_samples_per_second": 16.6,
      "eval_steps_per_second": 4.15,
      "step": 150
    },
    {
      "epoch": 0.040444623007901435,
      "grad_norm": 1.2821409702301025,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.0704,
      "step": 151
    },
    {
      "epoch": 0.04071246819338423,
      "grad_norm": 1.2796623706817627,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.0673,
      "step": 152
    },
    {
      "epoch": 0.04098031337886701,
      "grad_norm": 1.1115186214447021,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.9968,
      "step": 153
    },
    {
      "epoch": 0.0412481585643498,
      "grad_norm": 1.1020795106887817,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.0894,
      "step": 154
    },
    {
      "epoch": 0.041516003749832595,
      "grad_norm": 1.0832066535949707,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.2044,
      "step": 155
    },
    {
      "epoch": 0.041783848935315386,
      "grad_norm": 0.9941951036453247,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.3582,
      "step": 156
    },
    {
      "epoch": 0.04205169412079818,
      "grad_norm": 0.8040450215339661,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3239,
      "step": 157
    },
    {
      "epoch": 0.04231953930628097,
      "grad_norm": 0.53887939453125,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.2509,
      "step": 158
    },
    {
      "epoch": 0.04258738449176376,
      "grad_norm": 0.5014508366584778,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3392,
      "step": 159
    },
    {
      "epoch": 0.04285522967724655,
      "grad_norm": 0.6074042320251465,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4532,
      "step": 160
    },
    {
      "epoch": 0.043123074862729344,
      "grad_norm": 0.7176408767700195,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.4695,
      "step": 161
    },
    {
      "epoch": 0.043390920048212135,
      "grad_norm": 0.813952624797821,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.5388,
      "step": 162
    },
    {
      "epoch": 0.04365876523369493,
      "grad_norm": 0.7163006663322449,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.6173,
      "step": 163
    },
    {
      "epoch": 0.04392661041917772,
      "grad_norm": 0.6497973203659058,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.5445,
      "step": 164
    },
    {
      "epoch": 0.0441944556046605,
      "grad_norm": 0.6885555386543274,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.5574,
      "step": 165
    },
    {
      "epoch": 0.044462300790143294,
      "grad_norm": 0.5802903175354004,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.6435,
      "step": 166
    },
    {
      "epoch": 0.044730145975626086,
      "grad_norm": 0.6520971059799194,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.6832,
      "step": 167
    },
    {
      "epoch": 0.04499799116110888,
      "grad_norm": 0.689909815788269,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.6478,
      "step": 168
    },
    {
      "epoch": 0.04526583634659167,
      "grad_norm": 0.605025589466095,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.8783,
      "step": 169
    },
    {
      "epoch": 0.04553368153207446,
      "grad_norm": 0.5244045853614807,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.6479,
      "step": 170
    },
    {
      "epoch": 0.04580152671755725,
      "grad_norm": 0.5516737103462219,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.7692,
      "step": 171
    },
    {
      "epoch": 0.04606937190304004,
      "grad_norm": 0.646731436252594,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.8083,
      "step": 172
    },
    {
      "epoch": 0.046337217088522835,
      "grad_norm": 0.6894277334213257,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.8561,
      "step": 173
    },
    {
      "epoch": 0.046605062274005626,
      "grad_norm": 0.6532986760139465,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.7915,
      "step": 174
    },
    {
      "epoch": 0.04687290745948842,
      "grad_norm": 0.6645336151123047,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.0924,
      "step": 175
    },
    {
      "epoch": 0.04714075264497121,
      "grad_norm": 0.652217447757721,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.9303,
      "step": 176
    },
    {
      "epoch": 0.047408597830454,
      "grad_norm": 0.638865053653717,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.9629,
      "step": 177
    },
    {
      "epoch": 0.047676443015936786,
      "grad_norm": 0.6751264929771423,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.1087,
      "step": 178
    },
    {
      "epoch": 0.04794428820141958,
      "grad_norm": 0.6628146767616272,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.0419,
      "step": 179
    },
    {
      "epoch": 0.04821213338690237,
      "grad_norm": 0.6044755578041077,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.1322,
      "step": 180
    },
    {
      "epoch": 0.04847997857238516,
      "grad_norm": 0.6654837727546692,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.2168,
      "step": 181
    },
    {
      "epoch": 0.04874782375786795,
      "grad_norm": 0.7022812962532043,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.0709,
      "step": 182
    },
    {
      "epoch": 0.04901566894335074,
      "grad_norm": 0.7306815385818481,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0484,
      "step": 183
    },
    {
      "epoch": 0.049283514128833535,
      "grad_norm": 1.6151546239852905,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.1358,
      "step": 184
    },
    {
      "epoch": 0.049551359314316326,
      "grad_norm": 0.7555141448974609,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.9179,
      "step": 185
    },
    {
      "epoch": 0.04981920449979912,
      "grad_norm": 0.7630206346511841,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.9506,
      "step": 186
    },
    {
      "epoch": 0.05008704968528191,
      "grad_norm": 0.8862537145614624,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.1334,
      "step": 187
    },
    {
      "epoch": 0.0503548948707647,
      "grad_norm": 0.8738630414009094,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.1546,
      "step": 188
    },
    {
      "epoch": 0.05062274005624749,
      "grad_norm": 0.8641951680183411,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.8285,
      "step": 189
    },
    {
      "epoch": 0.05089058524173028,
      "grad_norm": 0.8461931347846985,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.0966,
      "step": 190
    },
    {
      "epoch": 0.05115843042721307,
      "grad_norm": 0.9750451445579529,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.0548,
      "step": 191
    },
    {
      "epoch": 0.05142627561269586,
      "grad_norm": 1.125663161277771,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.1871,
      "step": 192
    },
    {
      "epoch": 0.05169412079817865,
      "grad_norm": 1.7482781410217285,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.0669,
      "step": 193
    },
    {
      "epoch": 0.05196196598366144,
      "grad_norm": 1.3257977962493896,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.1898,
      "step": 194
    },
    {
      "epoch": 0.052229811169144234,
      "grad_norm": 1.2066391706466675,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.2373,
      "step": 195
    },
    {
      "epoch": 0.052497656354627026,
      "grad_norm": 1.2316598892211914,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.3087,
      "step": 196
    },
    {
      "epoch": 0.05276550154010982,
      "grad_norm": 1.4028412103652954,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3799,
      "step": 197
    },
    {
      "epoch": 0.05303334672559261,
      "grad_norm": 1.4312825202941895,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.4508,
      "step": 198
    },
    {
      "epoch": 0.0533011919110754,
      "grad_norm": 1.8116132020950317,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.3833,
      "step": 199
    },
    {
      "epoch": 0.05356903709655819,
      "grad_norm": 2.49422025680542,
      "learning_rate": 0.0,
      "loss": 2.6864,
      "step": 200
    },
    {
      "epoch": 0.05356903709655819,
      "eval_loss": 1.8008776903152466,
      "eval_runtime": 379.2758,
      "eval_samples_per_second": 16.579,
      "eval_steps_per_second": 4.145,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1332965512930918e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}