{
  "best_metric": 10.350133895874023,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.42621204049014383,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002131060202450719,
      "grad_norm": 0.09248433262109756,
      "learning_rate": 1e-05,
      "loss": 10.3708,
      "step": 1
    },
    {
      "epoch": 0.002131060202450719,
      "eval_loss": 10.376428604125977,
      "eval_runtime": 1.9435,
      "eval_samples_per_second": 406.48,
      "eval_steps_per_second": 101.877,
      "step": 1
    },
    {
      "epoch": 0.004262120404901438,
      "grad_norm": 0.04823608323931694,
      "learning_rate": 2e-05,
      "loss": 10.3791,
      "step": 2
    },
    {
      "epoch": 0.006393180607352157,
      "grad_norm": 0.04629861190915108,
      "learning_rate": 3e-05,
      "loss": 10.3807,
      "step": 3
    },
    {
      "epoch": 0.008524240809802876,
      "grad_norm": 0.0670231506228447,
      "learning_rate": 4e-05,
      "loss": 10.3781,
      "step": 4
    },
    {
      "epoch": 0.010655301012253596,
      "grad_norm": 0.0665554478764534,
      "learning_rate": 5e-05,
      "loss": 10.3818,
      "step": 5
    },
    {
      "epoch": 0.012786361214704315,
      "grad_norm": 0.07088156789541245,
      "learning_rate": 6e-05,
      "loss": 10.3809,
      "step": 6
    },
    {
      "epoch": 0.014917421417155035,
      "grad_norm": 0.041599053889513016,
      "learning_rate": 7e-05,
      "loss": 10.3779,
      "step": 7
    },
    {
      "epoch": 0.017048481619605753,
      "grad_norm": 0.040396321564912796,
      "learning_rate": 8e-05,
      "loss": 10.3772,
      "step": 8
    },
    {
      "epoch": 0.019179541822056473,
      "grad_norm": 0.06732139736413956,
      "learning_rate": 9e-05,
      "loss": 10.3721,
      "step": 9
    },
    {
      "epoch": 0.021310602024507193,
      "grad_norm": 0.06403470784425735,
      "learning_rate": 0.0001,
      "loss": 10.3715,
      "step": 10
    },
    {
      "epoch": 0.023441662226957913,
      "grad_norm": 0.06851677596569061,
      "learning_rate": 9.999316524962345e-05,
      "loss": 10.3769,
      "step": 11
    },
    {
      "epoch": 0.02557272242940863,
      "grad_norm": 0.0863216295838356,
      "learning_rate": 9.997266286704631e-05,
      "loss": 10.3787,
      "step": 12
    },
    {
      "epoch": 0.02770378263185935,
      "grad_norm": 0.09745137393474579,
      "learning_rate": 9.993849845741524e-05,
      "loss": 10.3784,
      "step": 13
    },
    {
      "epoch": 0.02983484283431007,
      "grad_norm": 0.0846598818898201,
      "learning_rate": 9.989068136093873e-05,
      "loss": 10.3748,
      "step": 14
    },
    {
      "epoch": 0.031965903036760786,
      "grad_norm": 0.06447857618331909,
      "learning_rate": 9.98292246503335e-05,
      "loss": 10.3745,
      "step": 15
    },
    {
      "epoch": 0.034096963239211506,
      "grad_norm": 0.07761789858341217,
      "learning_rate": 9.975414512725057e-05,
      "loss": 10.3739,
      "step": 16
    },
    {
      "epoch": 0.036228023441662226,
      "grad_norm": 0.07270949333906174,
      "learning_rate": 9.966546331768191e-05,
      "loss": 10.37,
      "step": 17
    },
    {
      "epoch": 0.038359083644112946,
      "grad_norm": 0.07797560095787048,
      "learning_rate": 9.956320346634876e-05,
      "loss": 10.3704,
      "step": 18
    },
    {
      "epoch": 0.040490143846563666,
      "grad_norm": 0.08227936923503876,
      "learning_rate": 9.944739353007344e-05,
      "loss": 10.3694,
      "step": 19
    },
    {
      "epoch": 0.042621204049014386,
      "grad_norm": 0.06472763419151306,
      "learning_rate": 9.931806517013612e-05,
      "loss": 10.3753,
      "step": 20
    },
    {
      "epoch": 0.044752264251465106,
      "grad_norm": 0.05423372983932495,
      "learning_rate": 9.917525374361912e-05,
      "loss": 10.3756,
      "step": 21
    },
    {
      "epoch": 0.046883324453915826,
      "grad_norm": 0.07389745861291885,
      "learning_rate": 9.901899829374047e-05,
      "loss": 10.3735,
      "step": 22
    },
    {
      "epoch": 0.049014384656366546,
      "grad_norm": 0.08643969148397446,
      "learning_rate": 9.884934153917997e-05,
      "loss": 10.3665,
      "step": 23
    },
    {
      "epoch": 0.05114544485881726,
      "grad_norm": 0.09035203605890274,
      "learning_rate": 9.86663298624003e-05,
      "loss": 10.366,
      "step": 24
    },
    {
      "epoch": 0.05327650506126798,
      "grad_norm": 0.11810406297445297,
      "learning_rate": 9.847001329696653e-05,
      "loss": 10.3673,
      "step": 25
    },
    {
      "epoch": 0.0554075652637187,
      "grad_norm": 0.08933798223733902,
      "learning_rate": 9.826044551386744e-05,
      "loss": 10.3702,
      "step": 26
    },
    {
      "epoch": 0.05753862546616942,
      "grad_norm": 0.08049152046442032,
      "learning_rate": 9.803768380684242e-05,
      "loss": 10.3735,
      "step": 27
    },
    {
      "epoch": 0.05966968566862014,
      "grad_norm": 0.08148496598005295,
      "learning_rate": 9.780178907671789e-05,
      "loss": 10.3753,
      "step": 28
    },
    {
      "epoch": 0.06180074587107086,
      "grad_norm": 0.09913858026266098,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.3713,
      "step": 29
    },
    {
      "epoch": 0.06393180607352157,
      "grad_norm": 0.20390036702156067,
      "learning_rate": 9.729086208503174e-05,
      "loss": 10.3685,
      "step": 30
    },
    {
      "epoch": 0.0660628662759723,
      "grad_norm": 0.15267124772071838,
      "learning_rate": 9.701596950580806e-05,
      "loss": 10.3716,
      "step": 31
    },
    {
      "epoch": 0.06819392647842301,
      "grad_norm": 0.1439274102449417,
      "learning_rate": 9.672822322997305e-05,
      "loss": 10.3751,
      "step": 32
    },
    {
      "epoch": 0.07032498668087374,
      "grad_norm": 0.18122100830078125,
      "learning_rate": 9.642770192448536e-05,
      "loss": 10.3711,
      "step": 33
    },
    {
      "epoch": 0.07245604688332445,
      "grad_norm": 0.18494148552417755,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.3709,
      "step": 34
    },
    {
      "epoch": 0.07458710708577518,
      "grad_norm": 0.2327541559934616,
      "learning_rate": 9.578866633275288e-05,
      "loss": 10.371,
      "step": 35
    },
    {
      "epoch": 0.07671816728822589,
      "grad_norm": 0.21441899240016937,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.3588,
      "step": 36
    },
    {
      "epoch": 0.07884922749067662,
      "grad_norm": 0.2696651220321655,
      "learning_rate": 9.509956150664796e-05,
      "loss": 10.351,
      "step": 37
    },
    {
      "epoch": 0.08098028769312733,
      "grad_norm": 0.18598511815071106,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.37,
      "step": 38
    },
    {
      "epoch": 0.08311134789557804,
      "grad_norm": 0.1918763369321823,
      "learning_rate": 9.43611409721806e-05,
      "loss": 10.3599,
      "step": 39
    },
    {
      "epoch": 0.08524240809802877,
      "grad_norm": 0.22185911238193512,
      "learning_rate": 9.397368756032445e-05,
      "loss": 10.3521,
      "step": 40
    },
    {
      "epoch": 0.08737346830047948,
      "grad_norm": 0.207609623670578,
      "learning_rate": 9.357421218136386e-05,
      "loss": 10.3598,
      "step": 41
    },
    {
      "epoch": 0.08950452850293021,
      "grad_norm": 0.22751246392726898,
      "learning_rate": 9.316282404787871e-05,
      "loss": 10.3644,
      "step": 42
    },
    {
      "epoch": 0.09163558870538092,
      "grad_norm": 0.18040315806865692,
      "learning_rate": 9.273963562927695e-05,
      "loss": 10.3668,
      "step": 43
    },
    {
      "epoch": 0.09376664890783165,
      "grad_norm": 0.12984536588191986,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.3669,
      "step": 44
    },
    {
      "epoch": 0.09589770911028236,
      "grad_norm": 0.17721785604953766,
      "learning_rate": 9.185832391312644e-05,
      "loss": 10.3689,
      "step": 45
    },
    {
      "epoch": 0.09802876931273309,
      "grad_norm": 0.1179790049791336,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.3663,
      "step": 46
    },
    {
      "epoch": 0.1001598295151838,
      "grad_norm": 0.1524268388748169,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.3647,
      "step": 47
    },
    {
      "epoch": 0.10229088971763452,
      "grad_norm": 0.15119385719299316,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.3677,
      "step": 48
    },
    {
      "epoch": 0.10442194992008524,
      "grad_norm": 0.15441370010375977,
      "learning_rate": 8.995939984474624e-05,
      "loss": 10.3596,
      "step": 49
    },
    {
      "epoch": 0.10655301012253596,
      "grad_norm": 0.19342565536499023,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.3564,
      "step": 50
    },
    {
      "epoch": 0.10655301012253596,
      "eval_loss": 10.356715202331543,
      "eval_runtime": 1.982,
      "eval_samples_per_second": 398.591,
      "eval_steps_per_second": 99.9,
      "step": 50
    },
    {
      "epoch": 0.10868407032498668,
      "grad_norm": 0.16328439116477966,
      "learning_rate": 8.894386393810563e-05,
      "loss": 10.3508,
      "step": 51
    },
    {
      "epoch": 0.1108151305274374,
      "grad_norm": 0.1387805938720703,
      "learning_rate": 8.842005554284296e-05,
      "loss": 10.3663,
      "step": 52
    },
    {
      "epoch": 0.11294619072988812,
      "grad_norm": 0.19031251966953278,
      "learning_rate": 8.788574348801675e-05,
      "loss": 10.3759,
      "step": 53
    },
    {
      "epoch": 0.11507725093233884,
      "grad_norm": 0.1813199818134308,
      "learning_rate": 8.73410738492077e-05,
      "loss": 10.3571,
      "step": 54
    },
    {
      "epoch": 0.11720831113478956,
      "grad_norm": 0.20543202757835388,
      "learning_rate": 8.678619553365659e-05,
      "loss": 10.3503,
      "step": 55
    },
    {
      "epoch": 0.11933937133724028,
      "grad_norm": 0.18195782601833344,
      "learning_rate": 8.622126023955446e-05,
      "loss": 10.3558,
      "step": 56
    },
    {
      "epoch": 0.12147043153969099,
      "grad_norm": 0.14910073578357697,
      "learning_rate": 8.564642241456986e-05,
      "loss": 10.3546,
      "step": 57
    },
    {
      "epoch": 0.12360149174214172,
      "grad_norm": 0.13736093044281006,
      "learning_rate": 8.506183921362443e-05,
      "loss": 10.3546,
      "step": 58
    },
    {
      "epoch": 0.12573255194459243,
      "grad_norm": 0.10152393579483032,
      "learning_rate": 8.44676704559283e-05,
      "loss": 10.3651,
      "step": 59
    },
    {
      "epoch": 0.12786361214704314,
      "grad_norm": 0.13326126337051392,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.3562,
      "step": 60
    },
    {
      "epoch": 0.12999467234949388,
      "grad_norm": 0.11736905574798584,
      "learning_rate": 8.32512286056924e-05,
      "loss": 10.351,
      "step": 61
    },
    {
      "epoch": 0.1321257325519446,
      "grad_norm": 0.13760432600975037,
      "learning_rate": 8.262928807620843e-05,
      "loss": 10.353,
      "step": 62
    },
    {
      "epoch": 0.1342567927543953,
      "grad_norm": 0.12203924357891083,
      "learning_rate": 8.199842702516583e-05,
      "loss": 10.3496,
      "step": 63
    },
    {
      "epoch": 0.13638785295684602,
      "grad_norm": 0.1889341175556183,
      "learning_rate": 8.135881792367686e-05,
      "loss": 10.348,
      "step": 64
    },
    {
      "epoch": 0.13851891315929676,
      "grad_norm": 0.17057493329048157,
      "learning_rate": 8.07106356344834e-05,
      "loss": 10.3484,
      "step": 65
    },
    {
      "epoch": 0.14064997336174748,
      "grad_norm": 0.14904175698757172,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.3517,
      "step": 66
    },
    {
      "epoch": 0.1427810335641982,
      "grad_norm": 0.1501254439353943,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.3513,
      "step": 67
    },
    {
      "epoch": 0.1449120937666489,
      "grad_norm": 0.18069830536842346,
      "learning_rate": 7.871643313414718e-05,
      "loss": 10.3446,
      "step": 68
    },
    {
      "epoch": 0.14704315396909962,
      "grad_norm": 0.1747799813747406,
      "learning_rate": 7.803575286758364e-05,
      "loss": 10.3439,
      "step": 69
    },
    {
      "epoch": 0.14917421417155036,
      "grad_norm": 0.13478559255599976,
      "learning_rate": 7.734740790612136e-05,
      "loss": 10.3438,
      "step": 70
    },
    {
      "epoch": 0.15130527437400107,
      "grad_norm": 0.16136078536510468,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.3511,
      "step": 71
    },
    {
      "epoch": 0.15343633457645178,
      "grad_norm": 0.0983903557062149,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.3564,
      "step": 72
    },
    {
      "epoch": 0.1555673947789025,
      "grad_norm": 0.16878783702850342,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.3593,
      "step": 73
    },
    {
      "epoch": 0.15769845498135324,
      "grad_norm": 0.13822585344314575,
      "learning_rate": 7.452117519152542e-05,
      "loss": 10.3478,
      "step": 74
    },
    {
      "epoch": 0.15982951518380395,
      "grad_norm": 0.11573084443807602,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.3509,
      "step": 75
    },
    {
      "epoch": 0.16196057538625466,
      "grad_norm": 0.19796794652938843,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.3543,
      "step": 76
    },
    {
      "epoch": 0.16409163558870538,
      "grad_norm": 0.08479451388120651,
      "learning_rate": 7.233044034264034e-05,
      "loss": 10.3574,
      "step": 77
    },
    {
      "epoch": 0.1662226957911561,
      "grad_norm": 0.08592009544372559,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.3558,
      "step": 78
    },
    {
      "epoch": 0.16835375599360683,
      "grad_norm": 0.08360584080219269,
      "learning_rate": 7.083909302476453e-05,
      "loss": 10.3564,
      "step": 79
    },
    {
      "epoch": 0.17048481619605754,
      "grad_norm": 0.12434081733226776,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.3532,
      "step": 80
    },
    {
      "epoch": 0.17261587639850826,
      "grad_norm": 0.18514879047870636,
      "learning_rate": 6.932495846462261e-05,
      "loss": 10.3495,
      "step": 81
    },
    {
      "epoch": 0.17474693660095897,
      "grad_norm": 0.128123477101326,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.3531,
      "step": 82
    },
    {
      "epoch": 0.1768779968034097,
      "grad_norm": 0.09085346758365631,
      "learning_rate": 6.778969234612584e-05,
      "loss": 10.3519,
      "step": 83
    },
    {
      "epoch": 0.17900905700586042,
      "grad_norm": 0.13225452601909637,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.3509,
      "step": 84
    },
    {
      "epoch": 0.18114011720831114,
      "grad_norm": 0.16030804812908173,
      "learning_rate": 6.623497346023418e-05,
      "loss": 10.3457,
      "step": 85
    },
    {
      "epoch": 0.18327117741076185,
      "grad_norm": 0.1662900149822235,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.3447,
      "step": 86
    },
    {
      "epoch": 0.18540223761321256,
      "grad_norm": 0.2558648884296417,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.3411,
      "step": 87
    },
    {
      "epoch": 0.1875332978156633,
      "grad_norm": 0.17125572264194489,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.3442,
      "step": 88
    },
    {
      "epoch": 0.18966435801811402,
      "grad_norm": 0.1528247594833374,
      "learning_rate": 6.307399704769099e-05,
      "loss": 10.3493,
      "step": 89
    },
    {
      "epoch": 0.19179541822056473,
      "grad_norm": 0.10840792953968048,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.3439,
      "step": 90
    },
    {
      "epoch": 0.19392647842301544,
      "grad_norm": 0.11453763395547867,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.3431,
      "step": 91
    },
    {
      "epoch": 0.19605753862546618,
      "grad_norm": 0.19489312171936035,
      "learning_rate": 6.066498153718735e-05,
      "loss": 10.3472,
      "step": 92
    },
    {
      "epoch": 0.1981885988279169,
      "grad_norm": 0.13528898358345032,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.3562,
      "step": 93
    },
    {
      "epoch": 0.2003196590303676,
      "grad_norm": 0.15263991057872772,
      "learning_rate": 5.90440267166055e-05,
      "loss": 10.3625,
      "step": 94
    },
    {
      "epoch": 0.20245071923281832,
      "grad_norm": 0.16219700872898102,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.3658,
      "step": 95
    },
    {
      "epoch": 0.20458177943526903,
      "grad_norm": 0.09309562295675278,
      "learning_rate": 5.74131823855921e-05,
      "loss": 10.3573,
      "step": 96
    },
    {
      "epoch": 0.20671283963771978,
      "grad_norm": 0.15279202163219452,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 10.3554,
      "step": 97
    },
    {
      "epoch": 0.2088438998401705,
      "grad_norm": 0.13374289870262146,
      "learning_rate": 5.577423184847932e-05,
      "loss": 10.3546,
      "step": 98
    },
    {
      "epoch": 0.2109749600426212,
      "grad_norm": 0.1342548131942749,
      "learning_rate": 5.495227651252315e-05,
      "loss": 10.3515,
      "step": 99
    },
    {
      "epoch": 0.21310602024507191,
      "grad_norm": 0.1506897509098053,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.3508,
      "step": 100
    },
    {
      "epoch": 0.21310602024507191,
      "eval_loss": 10.3510160446167,
      "eval_runtime": 2.0514,
      "eval_samples_per_second": 385.097,
      "eval_steps_per_second": 96.518,
      "step": 100
    },
    {
      "epoch": 0.21523708044752266,
      "grad_norm": 0.13816402852535248,
      "learning_rate": 5.330452921628497e-05,
      "loss": 10.35,
      "step": 101
    },
    {
      "epoch": 0.21736814064997337,
      "grad_norm": 0.10866691917181015,
      "learning_rate": 5.247918773366112e-05,
      "loss": 10.3565,
      "step": 102
    },
    {
      "epoch": 0.21949920085242408,
      "grad_norm": 0.1446610540151596,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.358,
      "step": 103
    },
    {
      "epoch": 0.2216302610548748,
      "grad_norm": 0.13828980922698975,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 10.3422,
      "step": 104
    },
    {
      "epoch": 0.2237613212573255,
      "grad_norm": 0.08554953336715698,
      "learning_rate": 5e-05,
      "loss": 10.3471,
      "step": 105
    },
    {
      "epoch": 0.22589238145977625,
      "grad_norm": 0.08408614993095398,
      "learning_rate": 4.917330276168208e-05,
      "loss": 10.3488,
      "step": 106
    },
    {
      "epoch": 0.22802344166222696,
      "grad_norm": 0.11610912531614304,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.3496,
      "step": 107
    },
    {
      "epoch": 0.23015450186467767,
      "grad_norm": 0.08962419629096985,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 10.3593,
      "step": 108
    },
    {
      "epoch": 0.2322855620671284,
      "grad_norm": 0.09933055192232132,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.3542,
      "step": 109
    },
    {
      "epoch": 0.23441662226957913,
      "grad_norm": 0.18968376517295837,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.3474,
      "step": 110
    },
    {
      "epoch": 0.23654768247202984,
      "grad_norm": 0.09397703409194946,
      "learning_rate": 4.504772348747687e-05,
      "loss": 10.3491,
      "step": 111
    },
    {
      "epoch": 0.23867874267448055,
      "grad_norm": 0.11786504089832306,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.3497,
      "step": 112
    },
    {
      "epoch": 0.24080980287693127,
      "grad_norm": 0.1807263344526291,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 10.3461,
      "step": 113
    },
    {
      "epoch": 0.24294086307938198,
      "grad_norm": 0.11689487099647522,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.3457,
      "step": 114
    },
    {
      "epoch": 0.24507192328183272,
      "grad_norm": 0.1555974781513214,
      "learning_rate": 4.17702704859633e-05,
      "loss": 10.3454,
      "step": 115
    },
    {
      "epoch": 0.24720298348428343,
      "grad_norm": 0.16858868300914764,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.3531,
      "step": 116
    },
    {
      "epoch": 0.24933404368673415,
      "grad_norm": 0.13661743700504303,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.3489,
      "step": 117
    },
    {
      "epoch": 0.25146510388918486,
      "grad_norm": 0.1304975152015686,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.3424,
      "step": 118
    },
    {
      "epoch": 0.2535961640916356,
      "grad_norm": 0.15624232590198517,
      "learning_rate": 3.852880399766243e-05,
      "loss": 10.3417,
      "step": 119
    },
    {
      "epoch": 0.2557272242940863,
      "grad_norm": 0.15448378026485443,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.3419,
      "step": 120
    },
    {
      "epoch": 0.257858284496537,
      "grad_norm": 0.14038050174713135,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 10.349,
      "step": 121
    },
    {
      "epoch": 0.25998934469898777,
      "grad_norm": 0.0935085341334343,
      "learning_rate": 3.612985456190778e-05,
      "loss": 10.3517,
      "step": 122
    },
    {
      "epoch": 0.26212040490143845,
      "grad_norm": 0.15683352947235107,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.3576,
      "step": 123
    },
    {
      "epoch": 0.2642514651038892,
      "grad_norm": 0.13154317438602448,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.3461,
      "step": 124
    },
    {
      "epoch": 0.2663825253063399,
      "grad_norm": 0.1440422534942627,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.3518,
      "step": 125
    },
    {
      "epoch": 0.2685135855087906,
      "grad_norm": 0.11527030915021896,
      "learning_rate": 3.298534127791785e-05,
      "loss": 10.3544,
      "step": 126
    },
    {
      "epoch": 0.27064464571124136,
      "grad_norm": 0.09047039598226547,
      "learning_rate": 3.221030765387417e-05,
      "loss": 10.3566,
      "step": 127
    },
    {
      "epoch": 0.27277570591369205,
      "grad_norm": 0.08020135760307312,
      "learning_rate": 3.144013755408895e-05,
      "loss": 10.3561,
      "step": 128
    },
    {
      "epoch": 0.2749067661161428,
      "grad_norm": 0.07745938748121262,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 10.3547,
      "step": 129
    },
    {
      "epoch": 0.2770378263185935,
      "grad_norm": 0.18570277094841003,
      "learning_rate": 2.991522876735154e-05,
      "loss": 10.3503,
      "step": 130
    },
    {
      "epoch": 0.2791688865210442,
      "grad_norm": 0.120386503636837,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.3502,
      "step": 131
    },
    {
      "epoch": 0.28129994672349495,
      "grad_norm": 0.13220550119876862,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 10.3514,
      "step": 132
    },
    {
      "epoch": 0.28343100692594564,
      "grad_norm": 0.08243604749441147,
      "learning_rate": 2.766955965735968e-05,
      "loss": 10.35,
      "step": 133
    },
    {
      "epoch": 0.2855620671283964,
      "grad_norm": 0.12599261105060577,
      "learning_rate": 2.693294185106562e-05,
      "loss": 10.35,
      "step": 134
    },
    {
      "epoch": 0.2876931273308471,
      "grad_norm": 0.15206916630268097,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 10.3444,
      "step": 135
    },
    {
      "epoch": 0.2898241875332978,
      "grad_norm": 0.12433921545743942,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.3427,
      "step": 136
    },
    {
      "epoch": 0.29195524773574855,
      "grad_norm": 0.23569245636463165,
      "learning_rate": 2.476172311325783e-05,
      "loss": 10.3396,
      "step": 137
    },
    {
      "epoch": 0.29408630793819923,
      "grad_norm": 0.13613857328891754,
      "learning_rate": 2.405152131093926e-05,
      "loss": 10.3461,
      "step": 138
    },
    {
      "epoch": 0.29621736814065,
      "grad_norm": 0.09155803918838501,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 10.3461,
      "step": 139
    },
    {
      "epoch": 0.2983484283431007,
      "grad_norm": 0.1303211897611618,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 10.3418,
      "step": 140
    },
    {
      "epoch": 0.3004794885455514,
      "grad_norm": 0.1281050443649292,
      "learning_rate": 2.196424713241637e-05,
      "loss": 10.3454,
      "step": 141
    },
    {
      "epoch": 0.30261054874800214,
      "grad_norm": 0.18241943418979645,
      "learning_rate": 2.128356686585282e-05,
      "loss": 10.3463,
      "step": 142
    },
    {
      "epoch": 0.3047416089504528,
      "grad_norm": 0.1735762655735016,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.3601,
      "step": 143
    },
    {
      "epoch": 0.30687266915290357,
      "grad_norm": 0.0996169000864029,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 10.3629,
      "step": 144
    },
    {
      "epoch": 0.3090037293553543,
      "grad_norm": 0.16868586838245392,
      "learning_rate": 1.928936436551661e-05,
      "loss": 10.3647,
      "step": 145
    },
    {
      "epoch": 0.311134789557805,
      "grad_norm": 0.11873532086610794,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 10.3578,
      "step": 146
    },
    {
      "epoch": 0.31326584976025573,
      "grad_norm": 0.16126777231693268,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.3539,
      "step": 147
    },
    {
      "epoch": 0.3153969099627065,
      "grad_norm": 0.15787655115127563,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 10.3549,
      "step": 148
    },
    {
      "epoch": 0.31752797016515716,
      "grad_norm": 0.10407184064388275,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 10.3518,
      "step": 149
    },
    {
      "epoch": 0.3196590303676079,
      "grad_norm": 0.15440940856933594,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 10.3513,
      "step": 150
    },
    {
      "epoch": 0.3196590303676079,
      "eval_loss": 10.350295066833496,
      "eval_runtime": 1.9787,
      "eval_samples_per_second": 399.251,
      "eval_steps_per_second": 100.066,
      "step": 150
    },
    {
      "epoch": 0.3217900905700586,
      "grad_norm": 0.13332189619541168,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.349,
      "step": 151
    },
    {
      "epoch": 0.3239211507725093,
      "grad_norm": 0.10957372933626175,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.3555,
      "step": 152
    },
    {
      "epoch": 0.32605221097496007,
      "grad_norm": 0.20130348205566406,
      "learning_rate": 1.435357758543015e-05,
      "loss": 10.3681,
      "step": 153
    },
    {
      "epoch": 0.32818327117741075,
      "grad_norm": 0.1332850605249405,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 10.3548,
      "step": 154
    },
    {
      "epoch": 0.3303143313798615,
      "grad_norm": 0.12404490262269974,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 10.3412,
      "step": 155
    },
    {
      "epoch": 0.3324453915823122,
      "grad_norm": 0.09010142087936401,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 10.3478,
      "step": 156
    },
    {
      "epoch": 0.3345764517847629,
      "grad_norm": 0.08030810952186584,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 10.3484,
      "step": 157
    },
    {
      "epoch": 0.33670751198721366,
      "grad_norm": 0.11509815603494644,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.3497,
      "step": 158
    },
    {
      "epoch": 0.33883857218966434,
      "grad_norm": 0.1001550704240799,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 10.3613,
      "step": 159
    },
    {
      "epoch": 0.3409696323921151,
      "grad_norm": 0.0986156240105629,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.3609,
      "step": 160
    },
    {
      "epoch": 0.34310069259456577,
      "grad_norm": 0.1840796172618866,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 10.3461,
      "step": 161
    },
    {
      "epoch": 0.3452317527970165,
      "grad_norm": 0.08613578975200653,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.3486,
      "step": 162
    },
    {
      "epoch": 0.34736281299946725,
      "grad_norm": 0.11978252232074738,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.3489,
      "step": 163
    },
    {
      "epoch": 0.34949387320191794,
      "grad_norm": 0.17610830068588257,
      "learning_rate": 8.599558442598998e-06,
      "loss": 10.3461,
      "step": 164
    },
    {
      "epoch": 0.3516249334043687,
      "grad_norm": 0.11621057242155075,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.3458,
      "step": 165
    },
    {
      "epoch": 0.3537559936068194,
      "grad_norm": 0.12426748871803284,
      "learning_rate": 7.695237378953223e-06,
      "loss": 10.3479,
      "step": 166
    },
    {
      "epoch": 0.3558870538092701,
      "grad_norm": 0.18053308129310608,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.3528,
      "step": 167
    },
    {
      "epoch": 0.35801811401172084,
      "grad_norm": 0.10976419597864151,
      "learning_rate": 6.837175952121306e-06,
      "loss": 10.3476,
      "step": 168
    },
    {
      "epoch": 0.36014917421417153,
      "grad_norm": 0.13435348868370056,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.3413,
      "step": 169
    },
    {
      "epoch": 0.36228023441662227,
      "grad_norm": 0.14827629923820496,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.3406,
      "step": 170
    },
    {
      "epoch": 0.364411294619073,
      "grad_norm": 0.11852351576089859,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.3464,
      "step": 171
    },
    {
      "epoch": 0.3665423548215237,
      "grad_norm": 0.0945136770606041,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.3507,
      "step": 172
    },
    {
      "epoch": 0.36867341502397444,
      "grad_norm": 0.15945258736610413,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.3584,
      "step": 173
    },
    {
      "epoch": 0.3708044752264251,
      "grad_norm": 0.1315777599811554,
      "learning_rate": 4.549673247541875e-06,
      "loss": 10.3468,
      "step": 174
    },
    {
      "epoch": 0.37293553542887586,
      "grad_norm": 0.09532032161951065,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.3478,
      "step": 175
    },
    {
      "epoch": 0.3750665956313266,
      "grad_norm": 0.17851229012012482,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.3524,
      "step": 176
    },
    {
      "epoch": 0.3771976558337773,
      "grad_norm": 0.07164841890335083,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 10.355,
      "step": 177
    },
    {
      "epoch": 0.37932871603622803,
      "grad_norm": 0.07830449938774109,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.3555,
      "step": 178
    },
    {
      "epoch": 0.3814597762386787,
      "grad_norm": 0.07385001331567764,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.355,
      "step": 179
    },
    {
      "epoch": 0.38359083644112946,
      "grad_norm": 0.07797562330961227,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 10.3554,
      "step": 180
    },
    {
      "epoch": 0.3857218966435802,
      "grad_norm": 0.17870692908763885,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.3488,
      "step": 181
    },
    {
      "epoch": 0.3878529568460309,
      "grad_norm": 0.14780284464359283,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 10.3493,
      "step": 182
    },
    {
      "epoch": 0.3899840170484816,
      "grad_norm": 0.13054999709129333,
      "learning_rate": 1.962316193157593e-06,
      "loss": 10.351,
      "step": 183
    },
    {
      "epoch": 0.39211507725093236,
      "grad_norm": 0.07900195568799973,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.3504,
      "step": 184
    },
    {
      "epoch": 0.39424613745338305,
      "grad_norm": 0.11692091822624207,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 10.3495,
      "step": 185
    },
    {
      "epoch": 0.3963771976558338,
      "grad_norm": 0.16850504279136658,
      "learning_rate": 1.333670137599713e-06,
      "loss": 10.3431,
      "step": 186
    },
    {
      "epoch": 0.3985082578582845,
      "grad_norm": 0.11325499415397644,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 10.3417,
      "step": 187
    },
    {
      "epoch": 0.4006393180607352,
      "grad_norm": 0.22079791128635406,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.3391,
      "step": 188
    },
    {
      "epoch": 0.40277037826318596,
      "grad_norm": 0.1437220424413681,
      "learning_rate": 8.247462563808817e-07,
      "loss": 10.3479,
      "step": 189
    },
    {
      "epoch": 0.40490143846563664,
      "grad_norm": 0.08985292166471481,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.3458,
      "step": 190
    },
    {
      "epoch": 0.4070324986680874,
      "grad_norm": 0.1367080807685852,
      "learning_rate": 5.526064699265753e-07,
      "loss": 10.3417,
      "step": 191
    },
    {
      "epoch": 0.40916355887053807,
      "grad_norm": 0.12252617627382278,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.3447,
      "step": 192
    },
    {
      "epoch": 0.4112946190729888,
      "grad_norm": 0.16251246631145477,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.3476,
      "step": 193
    },
    {
      "epoch": 0.41342567927543955,
      "grad_norm": 0.19158634543418884,
      "learning_rate": 2.458548727494292e-07,
      "loss": 10.3609,
      "step": 194
    },
    {
      "epoch": 0.41555673947789024,
      "grad_norm": 0.1534237414598465,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.3649,
      "step": 195
    },
    {
      "epoch": 0.417687799680341,
      "grad_norm": 0.12606020271778107,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 10.3583,
      "step": 196
    },
    {
      "epoch": 0.41981885988279166,
      "grad_norm": 0.12436086684465408,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.352,
      "step": 197
    },
    {
      "epoch": 0.4219499200852424,
      "grad_norm": 0.1406901329755783,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.3524,
      "step": 198
    },
    {
      "epoch": 0.42408098028769314,
      "grad_norm": 0.10922940075397491,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.3512,
      "step": 199
    },
    {
      "epoch": 0.42621204049014383,
      "grad_norm": 0.15437044203281403,
      "learning_rate": 0.0,
      "loss": 10.3498,
      "step": 200
    },
    {
      "epoch": 0.42621204049014383,
      "eval_loss": 10.350133895874023,
      "eval_runtime": 1.9731,
      "eval_samples_per_second": 400.39,
      "eval_steps_per_second": 100.351,
      "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": 22310132121600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}