{
  "best_metric": 2.056821584701538,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9248554913294798,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004624277456647399,
      "grad_norm": 13.66818618774414,
      "learning_rate": 1e-05,
      "loss": 4.4304,
      "step": 1
    },
    {
      "epoch": 0.004624277456647399,
      "eval_loss": 5.265167713165283,
      "eval_runtime": 6.4094,
      "eval_samples_per_second": 56.947,
      "eval_steps_per_second": 14.354,
      "step": 1
    },
    {
      "epoch": 0.009248554913294798,
      "grad_norm": 16.54599380493164,
      "learning_rate": 2e-05,
      "loss": 4.5779,
      "step": 2
    },
    {
      "epoch": 0.013872832369942197,
      "grad_norm": 15.252068519592285,
      "learning_rate": 3e-05,
      "loss": 4.2209,
      "step": 3
    },
    {
      "epoch": 0.018497109826589597,
      "grad_norm": 14.623470306396484,
      "learning_rate": 4e-05,
      "loss": 3.6107,
      "step": 4
    },
    {
      "epoch": 0.023121387283236993,
      "grad_norm": 17.961008071899414,
      "learning_rate": 5e-05,
      "loss": 3.8596,
      "step": 5
    },
    {
      "epoch": 0.027745664739884393,
      "grad_norm": 10.022628784179688,
      "learning_rate": 6e-05,
      "loss": 2.8527,
      "step": 6
    },
    {
      "epoch": 0.03236994219653179,
      "grad_norm": 9.890962600708008,
      "learning_rate": 7e-05,
      "loss": 2.9021,
      "step": 7
    },
    {
      "epoch": 0.03699421965317919,
      "grad_norm": 9.332515716552734,
      "learning_rate": 8e-05,
      "loss": 2.5086,
      "step": 8
    },
    {
      "epoch": 0.04161849710982659,
      "grad_norm": 6.255437850952148,
      "learning_rate": 9e-05,
      "loss": 2.2377,
      "step": 9
    },
    {
      "epoch": 0.046242774566473986,
      "grad_norm": 6.671128273010254,
      "learning_rate": 0.0001,
      "loss": 2.6216,
      "step": 10
    },
    {
      "epoch": 0.05086705202312139,
      "grad_norm": 7.9394025802612305,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.2707,
      "step": 11
    },
    {
      "epoch": 0.055491329479768786,
      "grad_norm": 6.322081089019775,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.3732,
      "step": 12
    },
    {
      "epoch": 0.06011560693641618,
      "grad_norm": 6.342021942138672,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.4865,
      "step": 13
    },
    {
      "epoch": 0.06473988439306358,
      "grad_norm": 6.89796781539917,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.3947,
      "step": 14
    },
    {
      "epoch": 0.06936416184971098,
      "grad_norm": 7.137399673461914,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.5294,
      "step": 15
    },
    {
      "epoch": 0.07398843930635839,
      "grad_norm": 6.32910680770874,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.5101,
      "step": 16
    },
    {
      "epoch": 0.07861271676300578,
      "grad_norm": 5.440783500671387,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.2562,
      "step": 17
    },
    {
      "epoch": 0.08323699421965318,
      "grad_norm": 6.01143741607666,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.5165,
      "step": 18
    },
    {
      "epoch": 0.08786127167630058,
      "grad_norm": 6.952459335327148,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.0098,
      "step": 19
    },
    {
      "epoch": 0.09248554913294797,
      "grad_norm": 5.688777446746826,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.4151,
      "step": 20
    },
    {
      "epoch": 0.09710982658959537,
      "grad_norm": 6.305129051208496,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.2242,
      "step": 21
    },
    {
      "epoch": 0.10173410404624278,
      "grad_norm": 7.090821266174316,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.5527,
      "step": 22
    },
    {
      "epoch": 0.10635838150289018,
      "grad_norm": 6.640192031860352,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2266,
      "step": 23
    },
    {
      "epoch": 0.11098265895953757,
      "grad_norm": 6.502255916595459,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.7065,
      "step": 24
    },
    {
      "epoch": 0.11560693641618497,
      "grad_norm": 5.606374263763428,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.1715,
      "step": 25
    },
    {
      "epoch": 0.12023121387283237,
      "grad_norm": 7.029818534851074,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.6427,
      "step": 26
    },
    {
      "epoch": 0.12485549132947976,
      "grad_norm": 6.205046653747559,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.3088,
      "step": 27
    },
    {
      "epoch": 0.12947976878612716,
      "grad_norm": 7.449288368225098,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.3148,
      "step": 28
    },
    {
      "epoch": 0.13410404624277455,
      "grad_norm": 7.620692729949951,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3834,
      "step": 29
    },
    {
      "epoch": 0.13872832369942195,
      "grad_norm": 7.837033271789551,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.7283,
      "step": 30
    },
    {
      "epoch": 0.14335260115606938,
      "grad_norm": 5.987688064575195,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.9999,
      "step": 31
    },
    {
      "epoch": 0.14797687861271677,
      "grad_norm": 6.761056900024414,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.618,
      "step": 32
    },
    {
      "epoch": 0.15260115606936417,
      "grad_norm": 6.640378952026367,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.195,
      "step": 33
    },
    {
      "epoch": 0.15722543352601157,
      "grad_norm": 6.470536708831787,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.2695,
      "step": 34
    },
    {
      "epoch": 0.16184971098265896,
      "grad_norm": 6.4953718185424805,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.3562,
      "step": 35
    },
    {
      "epoch": 0.16647398843930636,
      "grad_norm": 6.763280391693115,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.2843,
      "step": 36
    },
    {
      "epoch": 0.17109826589595376,
      "grad_norm": 6.885309219360352,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.6539,
      "step": 37
    },
    {
      "epoch": 0.17572254335260115,
      "grad_norm": 8.713134765625,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.2969,
      "step": 38
    },
    {
      "epoch": 0.18034682080924855,
      "grad_norm": 7.597935676574707,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.3745,
      "step": 39
    },
    {
      "epoch": 0.18497109826589594,
      "grad_norm": 6.970432281494141,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.0854,
      "step": 40
    },
    {
      "epoch": 0.18959537572254334,
      "grad_norm": 7.277366638183594,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.4324,
      "step": 41
    },
    {
      "epoch": 0.19421965317919074,
      "grad_norm": 6.135131359100342,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.8944,
      "step": 42
    },
    {
      "epoch": 0.19884393063583816,
      "grad_norm": 7.045614242553711,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.0352,
      "step": 43
    },
    {
      "epoch": 0.20346820809248556,
      "grad_norm": 7.250356674194336,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.2999,
      "step": 44
    },
    {
      "epoch": 0.20809248554913296,
      "grad_norm": 6.6403117179870605,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.0029,
      "step": 45
    },
    {
      "epoch": 0.21271676300578035,
      "grad_norm": 7.807745933532715,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.3155,
      "step": 46
    },
    {
      "epoch": 0.21734104046242775,
      "grad_norm": 10.786064147949219,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.7446,
      "step": 47
    },
    {
      "epoch": 0.22196531791907514,
      "grad_norm": 9.124238014221191,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.8679,
      "step": 48
    },
    {
      "epoch": 0.22658959537572254,
      "grad_norm": 9.356467247009277,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.5388,
      "step": 49
    },
    {
      "epoch": 0.23121387283236994,
      "grad_norm": 8.835561752319336,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.7724,
      "step": 50
    },
    {
      "epoch": 0.23121387283236994,
      "eval_loss": 2.492640256881714,
      "eval_runtime": 6.3929,
      "eval_samples_per_second": 57.094,
      "eval_steps_per_second": 14.391,
      "step": 50
    },
    {
      "epoch": 0.23583815028901733,
      "grad_norm": 11.168318748474121,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.3551,
      "step": 51
    },
    {
      "epoch": 0.24046242774566473,
      "grad_norm": 9.26650333404541,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.3905,
      "step": 52
    },
    {
      "epoch": 0.24508670520231213,
      "grad_norm": 6.986851692199707,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.576,
      "step": 53
    },
    {
      "epoch": 0.24971098265895952,
      "grad_norm": 5.855767250061035,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.9489,
      "step": 54
    },
    {
      "epoch": 0.2543352601156069,
      "grad_norm": 5.371345043182373,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1988,
      "step": 55
    },
    {
      "epoch": 0.2589595375722543,
      "grad_norm": 5.048678874969482,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.9993,
      "step": 56
    },
    {
      "epoch": 0.2635838150289017,
      "grad_norm": 6.399174213409424,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3613,
      "step": 57
    },
    {
      "epoch": 0.2682080924855491,
      "grad_norm": 5.449174880981445,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.2427,
      "step": 58
    },
    {
      "epoch": 0.2728323699421965,
      "grad_norm": 4.608231544494629,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.1345,
      "step": 59
    },
    {
      "epoch": 0.2774566473988439,
      "grad_norm": 6.366066932678223,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.2746,
      "step": 60
    },
    {
      "epoch": 0.28208092485549136,
      "grad_norm": 5.459083557128906,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.9963,
      "step": 61
    },
    {
      "epoch": 0.28670520231213875,
      "grad_norm": 5.977375507354736,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.5899,
      "step": 62
    },
    {
      "epoch": 0.29132947976878615,
      "grad_norm": 5.339417457580566,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.1859,
      "step": 63
    },
    {
      "epoch": 0.29595375722543354,
      "grad_norm": 5.901102542877197,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.5519,
      "step": 64
    },
    {
      "epoch": 0.30057803468208094,
      "grad_norm": 4.8512163162231445,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.1401,
      "step": 65
    },
    {
      "epoch": 0.30520231213872834,
      "grad_norm": 6.103243350982666,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.0858,
      "step": 66
    },
    {
      "epoch": 0.30982658959537573,
      "grad_norm": 4.9318108558654785,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.5098,
      "step": 67
    },
    {
      "epoch": 0.31445086705202313,
      "grad_norm": 5.127270221710205,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.2893,
      "step": 68
    },
    {
      "epoch": 0.3190751445086705,
      "grad_norm": 5.167647361755371,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.1235,
      "step": 69
    },
    {
      "epoch": 0.3236994219653179,
      "grad_norm": 5.155222415924072,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.2119,
      "step": 70
    },
    {
      "epoch": 0.3283236994219653,
      "grad_norm": 5.356748104095459,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.9708,
      "step": 71
    },
    {
      "epoch": 0.3329479768786127,
      "grad_norm": 5.793696403503418,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.7301,
      "step": 72
    },
    {
      "epoch": 0.3375722543352601,
      "grad_norm": 6.94437313079834,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.9874,
      "step": 73
    },
    {
      "epoch": 0.3421965317919075,
      "grad_norm": 5.386171340942383,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.6467,
      "step": 74
    },
    {
      "epoch": 0.3468208092485549,
      "grad_norm": 5.410337924957275,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.0862,
      "step": 75
    },
    {
      "epoch": 0.3514450867052023,
      "grad_norm": 5.958206653594971,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.1172,
      "step": 76
    },
    {
      "epoch": 0.3560693641618497,
      "grad_norm": 6.193649768829346,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.9514,
      "step": 77
    },
    {
      "epoch": 0.3606936416184971,
      "grad_norm": 5.2359418869018555,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.0575,
      "step": 78
    },
    {
      "epoch": 0.3653179190751445,
      "grad_norm": 5.117005825042725,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.2507,
      "step": 79
    },
    {
      "epoch": 0.3699421965317919,
      "grad_norm": 5.684737205505371,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.1911,
      "step": 80
    },
    {
      "epoch": 0.3745664739884393,
      "grad_norm": 5.511053562164307,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.4459,
      "step": 81
    },
    {
      "epoch": 0.3791907514450867,
      "grad_norm": 6.717464923858643,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.4078,
      "step": 82
    },
    {
      "epoch": 0.3838150289017341,
      "grad_norm": 5.559933662414551,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.9923,
      "step": 83
    },
    {
      "epoch": 0.3884393063583815,
      "grad_norm": 5.572854995727539,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.9762,
      "step": 84
    },
    {
      "epoch": 0.3930635838150289,
      "grad_norm": 6.934199333190918,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.5425,
      "step": 85
    },
    {
      "epoch": 0.3976878612716763,
      "grad_norm": 5.862249851226807,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.2705,
      "step": 86
    },
    {
      "epoch": 0.4023121387283237,
      "grad_norm": 5.357310771942139,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.9126,
      "step": 87
    },
    {
      "epoch": 0.4069364161849711,
      "grad_norm": 6.185244560241699,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.0877,
      "step": 88
    },
    {
      "epoch": 0.4115606936416185,
      "grad_norm": 6.051638126373291,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.8228,
      "step": 89
    },
    {
      "epoch": 0.4161849710982659,
      "grad_norm": 5.963949680328369,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.7054,
      "step": 90
    },
    {
      "epoch": 0.4208092485549133,
      "grad_norm": 7.951855659484863,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.5264,
      "step": 91
    },
    {
      "epoch": 0.4254335260115607,
      "grad_norm": 6.878432750701904,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.2679,
      "step": 92
    },
    {
      "epoch": 0.4300578034682081,
      "grad_norm": 6.006994724273682,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.3064,
      "step": 93
    },
    {
      "epoch": 0.4346820809248555,
      "grad_norm": 8.550135612487793,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.3091,
      "step": 94
    },
    {
      "epoch": 0.4393063583815029,
      "grad_norm": 5.893960952758789,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.8043,
      "step": 95
    },
    {
      "epoch": 0.4439306358381503,
      "grad_norm": 6.486201763153076,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.9566,
      "step": 96
    },
    {
      "epoch": 0.4485549132947977,
      "grad_norm": 7.376112937927246,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.6098,
      "step": 97
    },
    {
      "epoch": 0.4531791907514451,
      "grad_norm": 8.397552490234375,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.113,
      "step": 98
    },
    {
      "epoch": 0.4578034682080925,
      "grad_norm": 8.102672576904297,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.5676,
      "step": 99
    },
    {
      "epoch": 0.4624277456647399,
      "grad_norm": 10.339282989501953,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.9121,
      "step": 100
    },
    {
      "epoch": 0.4624277456647399,
      "eval_loss": 2.251077651977539,
      "eval_runtime": 6.3846,
      "eval_samples_per_second": 57.169,
      "eval_steps_per_second": 14.41,
      "step": 100
    },
    {
      "epoch": 0.46705202312138727,
      "grad_norm": 4.6681437492370605,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.291,
      "step": 101
    },
    {
      "epoch": 0.47167630057803467,
      "grad_norm": 4.3864827156066895,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.0247,
      "step": 102
    },
    {
      "epoch": 0.47630057803468207,
      "grad_norm": 5.49507474899292,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.0726,
      "step": 103
    },
    {
      "epoch": 0.48092485549132946,
      "grad_norm": 6.119171619415283,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.174,
      "step": 104
    },
    {
      "epoch": 0.48554913294797686,
      "grad_norm": 5.9026408195495605,
      "learning_rate": 5e-05,
      "loss": 2.1384,
      "step": 105
    },
    {
      "epoch": 0.49017341040462425,
      "grad_norm": 5.539081573486328,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.4496,
      "step": 106
    },
    {
      "epoch": 0.49479768786127165,
      "grad_norm": 5.595170974731445,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.4467,
      "step": 107
    },
    {
      "epoch": 0.49942196531791905,
      "grad_norm": 5.280254364013672,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.1806,
      "step": 108
    },
    {
      "epoch": 0.5040462427745664,
      "grad_norm": 4.653598308563232,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0262,
      "step": 109
    },
    {
      "epoch": 0.5086705202312138,
      "grad_norm": 4.762678623199463,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.0287,
      "step": 110
    },
    {
      "epoch": 0.5132947976878612,
      "grad_norm": 5.388850212097168,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.2743,
      "step": 111
    },
    {
      "epoch": 0.5179190751445086,
      "grad_norm": 4.635728359222412,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.8908,
      "step": 112
    },
    {
      "epoch": 0.522543352601156,
      "grad_norm": 4.689375877380371,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.1411,
      "step": 113
    },
    {
      "epoch": 0.5271676300578034,
      "grad_norm": 6.601043701171875,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.597,
      "step": 114
    },
    {
      "epoch": 0.5317919075144508,
      "grad_norm": 4.445426940917969,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.4877,
      "step": 115
    },
    {
      "epoch": 0.5364161849710982,
      "grad_norm": 5.040781021118164,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.4113,
      "step": 116
    },
    {
      "epoch": 0.5410404624277456,
      "grad_norm": 4.245193958282471,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.0575,
      "step": 117
    },
    {
      "epoch": 0.545664739884393,
      "grad_norm": 4.230079650878906,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.3154,
      "step": 118
    },
    {
      "epoch": 0.5502890173410404,
      "grad_norm": 5.019067764282227,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.9595,
      "step": 119
    },
    {
      "epoch": 0.5549132947976878,
      "grad_norm": 4.667409420013428,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.3683,
      "step": 120
    },
    {
      "epoch": 0.5595375722543353,
      "grad_norm": 4.534331321716309,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.0588,
      "step": 121
    },
    {
      "epoch": 0.5641618497109827,
      "grad_norm": 3.8424460887908936,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.9173,
      "step": 122
    },
    {
      "epoch": 0.5687861271676301,
      "grad_norm": 4.836106777191162,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.833,
      "step": 123
    },
    {
      "epoch": 0.5734104046242775,
      "grad_norm": 4.822645664215088,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.8499,
      "step": 124
    },
    {
      "epoch": 0.5780346820809249,
      "grad_norm": 5.111135005950928,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.9122,
      "step": 125
    },
    {
      "epoch": 0.5826589595375723,
      "grad_norm": 4.807843208312988,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.1549,
      "step": 126
    },
    {
      "epoch": 0.5872832369942197,
      "grad_norm": 6.003751277923584,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.768,
      "step": 127
    },
    {
      "epoch": 0.5919075144508671,
      "grad_norm": 5.239036560058594,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.1503,
      "step": 128
    },
    {
      "epoch": 0.5965317919075145,
      "grad_norm": 4.235095024108887,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.939,
      "step": 129
    },
    {
      "epoch": 0.6011560693641619,
      "grad_norm": 6.588282108306885,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.0709,
      "step": 130
    },
    {
      "epoch": 0.6057803468208093,
      "grad_norm": 6.036434650421143,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.0781,
      "step": 131
    },
    {
      "epoch": 0.6104046242774567,
      "grad_norm": 5.442200660705566,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.1234,
      "step": 132
    },
    {
      "epoch": 0.6150289017341041,
      "grad_norm": 5.534650802612305,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.0043,
      "step": 133
    },
    {
      "epoch": 0.6196531791907515,
      "grad_norm": 4.898402214050293,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.7237,
      "step": 134
    },
    {
      "epoch": 0.6242774566473989,
      "grad_norm": 4.983450889587402,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.6327,
      "step": 135
    },
    {
      "epoch": 0.6289017341040463,
      "grad_norm": 5.243624687194824,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.9394,
      "step": 136
    },
    {
      "epoch": 0.6335260115606937,
      "grad_norm": 5.744765758514404,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.8547,
      "step": 137
    },
    {
      "epoch": 0.638150289017341,
      "grad_norm": 5.003788471221924,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.9668,
      "step": 138
    },
    {
      "epoch": 0.6427745664739885,
      "grad_norm": 6.371557235717773,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.7128,
      "step": 139
    },
    {
      "epoch": 0.6473988439306358,
      "grad_norm": 5.272283554077148,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.9249,
      "step": 140
    },
    {
      "epoch": 0.6520231213872832,
      "grad_norm": 5.948561668395996,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.4251,
      "step": 141
    },
    {
      "epoch": 0.6566473988439306,
      "grad_norm": 5.489185333251953,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.1093,
      "step": 142
    },
    {
      "epoch": 0.661271676300578,
      "grad_norm": 5.949795722961426,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.9437,
      "step": 143
    },
    {
      "epoch": 0.6658959537572254,
      "grad_norm": 6.4457688331604,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.181,
      "step": 144
    },
    {
      "epoch": 0.6705202312138728,
      "grad_norm": 6.602534294128418,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.3516,
      "step": 145
    },
    {
      "epoch": 0.6751445086705202,
      "grad_norm": 6.211402893066406,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.2818,
      "step": 146
    },
    {
      "epoch": 0.6797687861271676,
      "grad_norm": 6.284914016723633,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.2245,
      "step": 147
    },
    {
      "epoch": 0.684393063583815,
      "grad_norm": 6.084216594696045,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0769,
      "step": 148
    },
    {
      "epoch": 0.6890173410404624,
      "grad_norm": 7.297597408294678,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.2302,
      "step": 149
    },
    {
      "epoch": 0.6936416184971098,
      "grad_norm": 7.841131687164307,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 3.0472,
      "step": 150
    },
    {
      "epoch": 0.6936416184971098,
      "eval_loss": 2.1150963306427,
      "eval_runtime": 6.3813,
      "eval_samples_per_second": 57.198,
      "eval_steps_per_second": 14.417,
      "step": 150
    },
    {
      "epoch": 0.6982658959537572,
      "grad_norm": 4.8362016677856445,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.1981,
      "step": 151
    },
    {
      "epoch": 0.7028901734104046,
      "grad_norm": 5.259777069091797,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9558,
      "step": 152
    },
    {
      "epoch": 0.707514450867052,
      "grad_norm": 4.200504302978516,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.063,
      "step": 153
    },
    {
      "epoch": 0.7121387283236994,
      "grad_norm": 4.8924994468688965,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.3358,
      "step": 154
    },
    {
      "epoch": 0.7167630057803468,
      "grad_norm": 4.375818729400635,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.0129,
      "step": 155
    },
    {
      "epoch": 0.7213872832369942,
      "grad_norm": 3.907277822494507,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.8845,
      "step": 156
    },
    {
      "epoch": 0.7260115606936416,
      "grad_norm": 4.529579162597656,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.1907,
      "step": 157
    },
    {
      "epoch": 0.730635838150289,
      "grad_norm": 4.412952899932861,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.143,
      "step": 158
    },
    {
      "epoch": 0.7352601156069364,
      "grad_norm": 4.717729091644287,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.0534,
      "step": 159
    },
    {
      "epoch": 0.7398843930635838,
      "grad_norm": 4.024964809417725,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.2121,
      "step": 160
    },
    {
      "epoch": 0.7445086705202312,
      "grad_norm": 4.3739914894104,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.1085,
      "step": 161
    },
    {
      "epoch": 0.7491329479768786,
      "grad_norm": 4.6576972007751465,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.9944,
      "step": 162
    },
    {
      "epoch": 0.753757225433526,
      "grad_norm": 4.177798271179199,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.4119,
      "step": 163
    },
    {
      "epoch": 0.7583815028901734,
      "grad_norm": 4.690871715545654,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.631,
      "step": 164
    },
    {
      "epoch": 0.7630057803468208,
      "grad_norm": 4.558809757232666,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.8138,
      "step": 165
    },
    {
      "epoch": 0.7676300578034682,
      "grad_norm": 4.814574241638184,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.1725,
      "step": 166
    },
    {
      "epoch": 0.7722543352601156,
      "grad_norm": 5.367768287658691,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.3555,
      "step": 167
    },
    {
      "epoch": 0.776878612716763,
      "grad_norm": 5.063154220581055,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.1287,
      "step": 168
    },
    {
      "epoch": 0.7815028901734105,
      "grad_norm": 4.955220699310303,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.2239,
      "step": 169
    },
    {
      "epoch": 0.7861271676300579,
      "grad_norm": 5.407108783721924,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.3707,
      "step": 170
    },
    {
      "epoch": 0.7907514450867053,
      "grad_norm": 5.4429826736450195,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.8507,
      "step": 171
    },
    {
      "epoch": 0.7953757225433526,
      "grad_norm": 4.828087329864502,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.2035,
      "step": 172
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.1899237632751465,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.0759,
      "step": 173
    },
    {
      "epoch": 0.8046242774566474,
      "grad_norm": 5.426806926727295,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.9159,
      "step": 174
    },
    {
      "epoch": 0.8092485549132948,
      "grad_norm": 6.05827522277832,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.067,
      "step": 175
    },
    {
      "epoch": 0.8138728323699422,
      "grad_norm": 4.956657409667969,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.121,
      "step": 176
    },
    {
      "epoch": 0.8184971098265896,
      "grad_norm": 4.764184474945068,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.7505,
      "step": 177
    },
    {
      "epoch": 0.823121387283237,
      "grad_norm": 5.682034492492676,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.1842,
      "step": 178
    },
    {
      "epoch": 0.8277456647398844,
      "grad_norm": 4.754544734954834,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.2167,
      "step": 179
    },
    {
      "epoch": 0.8323699421965318,
      "grad_norm": 5.2762250900268555,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.0995,
      "step": 180
    },
    {
      "epoch": 0.8369942196531792,
      "grad_norm": 4.626994609832764,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.697,
      "step": 181
    },
    {
      "epoch": 0.8416184971098266,
      "grad_norm": 4.7468180656433105,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.9327,
      "step": 182
    },
    {
      "epoch": 0.846242774566474,
      "grad_norm": 5.475385665893555,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.1992,
      "step": 183
    },
    {
      "epoch": 0.8508670520231214,
      "grad_norm": 5.75460958480835,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.6458,
      "step": 184
    },
    {
      "epoch": 0.8554913294797688,
      "grad_norm": 5.448469638824463,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.9735,
      "step": 185
    },
    {
      "epoch": 0.8601156069364162,
      "grad_norm": 5.028481960296631,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.0669,
      "step": 186
    },
    {
      "epoch": 0.8647398843930636,
      "grad_norm": 5.3170270919799805,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.8385,
      "step": 187
    },
    {
      "epoch": 0.869364161849711,
      "grad_norm": 4.953030586242676,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.911,
      "step": 188
    },
    {
      "epoch": 0.8739884393063584,
      "grad_norm": 6.173885345458984,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.0521,
      "step": 189
    },
    {
      "epoch": 0.8786127167630058,
      "grad_norm": 5.352744102478027,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.1157,
      "step": 190
    },
    {
      "epoch": 0.8832369942196532,
      "grad_norm": 5.833892345428467,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.409,
      "step": 191
    },
    {
      "epoch": 0.8878612716763006,
      "grad_norm": 6.850665092468262,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.0101,
      "step": 192
    },
    {
      "epoch": 0.892485549132948,
      "grad_norm": 4.598543643951416,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.5818,
      "step": 193
    },
    {
      "epoch": 0.8971098265895954,
      "grad_norm": 6.304871082305908,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.9805,
      "step": 194
    },
    {
      "epoch": 0.9017341040462428,
      "grad_norm": 6.471837520599365,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.444,
      "step": 195
    },
    {
      "epoch": 0.9063583815028902,
      "grad_norm": 6.678435325622559,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.2543,
      "step": 196
    },
    {
      "epoch": 0.9109826589595376,
      "grad_norm": 7.933207988739014,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.4777,
      "step": 197
    },
    {
      "epoch": 0.915606936416185,
      "grad_norm": 8.702287673950195,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.6437,
      "step": 198
    },
    {
      "epoch": 0.9202312138728324,
      "grad_norm": 6.499545574188232,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.6723,
      "step": 199
    },
    {
      "epoch": 0.9248554913294798,
      "grad_norm": 8.72105884552002,
      "learning_rate": 0.0,
      "loss": 2.2872,
      "step": 200
    },
    {
      "epoch": 0.9248554913294798,
      "eval_loss": 2.056821584701538,
      "eval_runtime": 6.394,
      "eval_samples_per_second": 57.085,
      "eval_steps_per_second": 14.388,
      "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": 1.54569648635904e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}