{
  "best_metric": 0.32964658737182617,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.572655690765927,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002863278453829635,
      "grad_norm": 0.28293463587760925,
      "learning_rate": 1e-05,
      "loss": 0.3532,
      "step": 1
    },
    {
      "epoch": 0.002863278453829635,
      "eval_loss": 0.48722097277641296,
      "eval_runtime": 21.419,
      "eval_samples_per_second": 27.499,
      "eval_steps_per_second": 6.91,
      "step": 1
    },
    {
      "epoch": 0.00572655690765927,
      "grad_norm": 0.296455442905426,
      "learning_rate": 2e-05,
      "loss": 0.3649,
      "step": 2
    },
    {
      "epoch": 0.008589835361488905,
      "grad_norm": 0.2837323248386383,
      "learning_rate": 3e-05,
      "loss": 0.3883,
      "step": 3
    },
    {
      "epoch": 0.01145311381531854,
      "grad_norm": 0.2765909731388092,
      "learning_rate": 4e-05,
      "loss": 0.3925,
      "step": 4
    },
    {
      "epoch": 0.014316392269148175,
      "grad_norm": 0.23322518169879913,
      "learning_rate": 5e-05,
      "loss": 0.3394,
      "step": 5
    },
    {
      "epoch": 0.01717967072297781,
      "grad_norm": 0.1991652399301529,
      "learning_rate": 6e-05,
      "loss": 0.3096,
      "step": 6
    },
    {
      "epoch": 0.020042949176807445,
      "grad_norm": 0.23408232629299164,
      "learning_rate": 7e-05,
      "loss": 0.3743,
      "step": 7
    },
    {
      "epoch": 0.02290622763063708,
      "grad_norm": 0.2550717890262604,
      "learning_rate": 8e-05,
      "loss": 0.3789,
      "step": 8
    },
    {
      "epoch": 0.025769506084466716,
      "grad_norm": 0.2513333261013031,
      "learning_rate": 9e-05,
      "loss": 0.3662,
      "step": 9
    },
    {
      "epoch": 0.02863278453829635,
      "grad_norm": 0.23235957324504852,
      "learning_rate": 0.0001,
      "loss": 0.3443,
      "step": 10
    },
    {
      "epoch": 0.031496062992125984,
      "grad_norm": 0.21944913268089294,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.4101,
      "step": 11
    },
    {
      "epoch": 0.03435934144595562,
      "grad_norm": 0.21626396477222443,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.3642,
      "step": 12
    },
    {
      "epoch": 0.03722261989978525,
      "grad_norm": 0.19759628176689148,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.3668,
      "step": 13
    },
    {
      "epoch": 0.04008589835361489,
      "grad_norm": 0.19423308968544006,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.3531,
      "step": 14
    },
    {
      "epoch": 0.04294917680744453,
      "grad_norm": 0.17444612085819244,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.3546,
      "step": 15
    },
    {
      "epoch": 0.04581245526127416,
      "grad_norm": 0.18707162141799927,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.3897,
      "step": 16
    },
    {
      "epoch": 0.048675733715103794,
      "grad_norm": 0.18808357417583466,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.3754,
      "step": 17
    },
    {
      "epoch": 0.05153901216893343,
      "grad_norm": 0.1850220263004303,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.3599,
      "step": 18
    },
    {
      "epoch": 0.05440229062276306,
      "grad_norm": 0.19139081239700317,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.326,
      "step": 19
    },
    {
      "epoch": 0.0572655690765927,
      "grad_norm": 0.19979405403137207,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.4198,
      "step": 20
    },
    {
      "epoch": 0.06012884753042233,
      "grad_norm": 0.18445749580860138,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.3326,
      "step": 21
    },
    {
      "epoch": 0.06299212598425197,
      "grad_norm": 0.17954690754413605,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.3613,
      "step": 22
    },
    {
      "epoch": 0.0658554044380816,
      "grad_norm": 0.19147607684135437,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.4072,
      "step": 23
    },
    {
      "epoch": 0.06871868289191124,
      "grad_norm": 0.18479324877262115,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.3347,
      "step": 24
    },
    {
      "epoch": 0.07158196134574088,
      "grad_norm": 0.18923068046569824,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.3606,
      "step": 25
    },
    {
      "epoch": 0.0744452397995705,
      "grad_norm": 0.200177863240242,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.3776,
      "step": 26
    },
    {
      "epoch": 0.07730851825340014,
      "grad_norm": 0.20110751688480377,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.3574,
      "step": 27
    },
    {
      "epoch": 0.08017179670722978,
      "grad_norm": 0.19002267718315125,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.3549,
      "step": 28
    },
    {
      "epoch": 0.08303507516105942,
      "grad_norm": 0.1950635313987732,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.3394,
      "step": 29
    },
    {
      "epoch": 0.08589835361488905,
      "grad_norm": 0.21518242359161377,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.4419,
      "step": 30
    },
    {
      "epoch": 0.08876163206871868,
      "grad_norm": 0.20167042315006256,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.3507,
      "step": 31
    },
    {
      "epoch": 0.09162491052254831,
      "grad_norm": 0.20110003650188446,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.3201,
      "step": 32
    },
    {
      "epoch": 0.09448818897637795,
      "grad_norm": 0.20710685849189758,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.3461,
      "step": 33
    },
    {
      "epoch": 0.09735146743020759,
      "grad_norm": 0.20276093482971191,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.3237,
      "step": 34
    },
    {
      "epoch": 0.10021474588403723,
      "grad_norm": 0.2124381959438324,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.3639,
      "step": 35
    },
    {
      "epoch": 0.10307802433786686,
      "grad_norm": 0.21842937171459198,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.3698,
      "step": 36
    },
    {
      "epoch": 0.10594130279169649,
      "grad_norm": 0.20515160262584686,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.3207,
      "step": 37
    },
    {
      "epoch": 0.10880458124552612,
      "grad_norm": 0.22341179847717285,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.398,
      "step": 38
    },
    {
      "epoch": 0.11166785969935576,
      "grad_norm": 0.21295617520809174,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3312,
      "step": 39
    },
    {
      "epoch": 0.1145311381531854,
      "grad_norm": 0.20717103779315948,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.3044,
      "step": 40
    },
    {
      "epoch": 0.11739441660701504,
      "grad_norm": 0.22255851328372955,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3736,
      "step": 41
    },
    {
      "epoch": 0.12025769506084466,
      "grad_norm": 0.23172444105148315,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.3669,
      "step": 42
    },
    {
      "epoch": 0.1231209735146743,
      "grad_norm": 0.2181534469127655,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.2944,
      "step": 43
    },
    {
      "epoch": 0.12598425196850394,
      "grad_norm": 0.24176418781280518,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.3214,
      "step": 44
    },
    {
      "epoch": 0.12884753042233357,
      "grad_norm": 0.257718563079834,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.3842,
      "step": 45
    },
    {
      "epoch": 0.1317108088761632,
      "grad_norm": 0.2542491853237152,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.3028,
      "step": 46
    },
    {
      "epoch": 0.13457408732999285,
      "grad_norm": 0.24146394431591034,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.3178,
      "step": 47
    },
    {
      "epoch": 0.13743736578382248,
      "grad_norm": 0.26331910490989685,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.3187,
      "step": 48
    },
    {
      "epoch": 0.14030064423765212,
      "grad_norm": 0.2696646749973297,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.3099,
      "step": 49
    },
    {
      "epoch": 0.14316392269148176,
      "grad_norm": 0.29912179708480835,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.338,
      "step": 50
    },
    {
      "epoch": 0.14316392269148176,
      "eval_loss": 0.3451482653617859,
      "eval_runtime": 21.5561,
      "eval_samples_per_second": 27.324,
      "eval_steps_per_second": 6.866,
      "step": 50
    },
    {
      "epoch": 0.14602720114531137,
      "grad_norm": 0.1618831604719162,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.2983,
      "step": 51
    },
    {
      "epoch": 0.148890479599141,
      "grad_norm": 0.15669173002243042,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.2869,
      "step": 52
    },
    {
      "epoch": 0.15175375805297064,
      "grad_norm": 0.15685734152793884,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.3182,
      "step": 53
    },
    {
      "epoch": 0.15461703650680028,
      "grad_norm": 0.14314024150371552,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.335,
      "step": 54
    },
    {
      "epoch": 0.15748031496062992,
      "grad_norm": 0.1372736245393753,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.4383,
      "step": 55
    },
    {
      "epoch": 0.16034359341445956,
      "grad_norm": 0.12798769772052765,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3147,
      "step": 56
    },
    {
      "epoch": 0.1632068718682892,
      "grad_norm": 0.13556930422782898,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3493,
      "step": 57
    },
    {
      "epoch": 0.16607015032211883,
      "grad_norm": 0.13310757279396057,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3042,
      "step": 58
    },
    {
      "epoch": 0.16893342877594847,
      "grad_norm": 0.1345643401145935,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3104,
      "step": 59
    },
    {
      "epoch": 0.1717967072297781,
      "grad_norm": 0.14942008256912231,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.3767,
      "step": 60
    },
    {
      "epoch": 0.17465998568360774,
      "grad_norm": 0.1519930511713028,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.4371,
      "step": 61
    },
    {
      "epoch": 0.17752326413743735,
      "grad_norm": 0.15125346183776855,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.3523,
      "step": 62
    },
    {
      "epoch": 0.180386542591267,
      "grad_norm": 0.1467069387435913,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.3193,
      "step": 63
    },
    {
      "epoch": 0.18324982104509663,
      "grad_norm": 0.1478845179080963,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.3235,
      "step": 64
    },
    {
      "epoch": 0.18611309949892627,
      "grad_norm": 0.15619802474975586,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.3421,
      "step": 65
    },
    {
      "epoch": 0.1889763779527559,
      "grad_norm": 0.15994876623153687,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.3767,
      "step": 66
    },
    {
      "epoch": 0.19183965640658554,
      "grad_norm": 0.1619894802570343,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.3652,
      "step": 67
    },
    {
      "epoch": 0.19470293486041518,
      "grad_norm": 0.16099055111408234,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3713,
      "step": 68
    },
    {
      "epoch": 0.19756621331424482,
      "grad_norm": 0.1587141901254654,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3146,
      "step": 69
    },
    {
      "epoch": 0.20042949176807445,
      "grad_norm": 0.15905050933361053,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3327,
      "step": 70
    },
    {
      "epoch": 0.2032927702219041,
      "grad_norm": 0.17514987289905548,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.4074,
      "step": 71
    },
    {
      "epoch": 0.20615604867573373,
      "grad_norm": 0.1616692990064621,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3152,
      "step": 72
    },
    {
      "epoch": 0.20901932712956334,
      "grad_norm": 0.15333983302116394,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.2919,
      "step": 73
    },
    {
      "epoch": 0.21188260558339297,
      "grad_norm": 0.16497908532619476,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.33,
      "step": 74
    },
    {
      "epoch": 0.2147458840372226,
      "grad_norm": 0.16649502515792847,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.3057,
      "step": 75
    },
    {
      "epoch": 0.21760916249105225,
      "grad_norm": 0.16902677714824677,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.3597,
      "step": 76
    },
    {
      "epoch": 0.2204724409448819,
      "grad_norm": 0.163208469748497,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3143,
      "step": 77
    },
    {
      "epoch": 0.22333571939871152,
      "grad_norm": 0.17100782692432404,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.3517,
      "step": 78
    },
    {
      "epoch": 0.22619899785254116,
      "grad_norm": 0.1782175600528717,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.3494,
      "step": 79
    },
    {
      "epoch": 0.2290622763063708,
      "grad_norm": 0.17334997653961182,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3357,
      "step": 80
    },
    {
      "epoch": 0.23192555476020044,
      "grad_norm": 0.1830623298883438,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3334,
      "step": 81
    },
    {
      "epoch": 0.23478883321403007,
      "grad_norm": 0.17625942826271057,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.3197,
      "step": 82
    },
    {
      "epoch": 0.2376521116678597,
      "grad_norm": 0.1825447380542755,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.3181,
      "step": 83
    },
    {
      "epoch": 0.24051539012168932,
      "grad_norm": 0.1905507594347,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.3417,
      "step": 84
    },
    {
      "epoch": 0.24337866857551896,
      "grad_norm": 0.1998002827167511,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3448,
      "step": 85
    },
    {
      "epoch": 0.2462419470293486,
      "grad_norm": 0.21149498224258423,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.4129,
      "step": 86
    },
    {
      "epoch": 0.24910522548317823,
      "grad_norm": 0.20471715927124023,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.3584,
      "step": 87
    },
    {
      "epoch": 0.25196850393700787,
      "grad_norm": 0.20461665093898773,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3095,
      "step": 88
    },
    {
      "epoch": 0.25483178239083754,
      "grad_norm": 0.20610246062278748,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.3549,
      "step": 89
    },
    {
      "epoch": 0.25769506084466715,
      "grad_norm": 0.20702482759952545,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.3287,
      "step": 90
    },
    {
      "epoch": 0.26055833929849676,
      "grad_norm": 0.2018522173166275,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3001,
      "step": 91
    },
    {
      "epoch": 0.2634216177523264,
      "grad_norm": 0.2131468951702118,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.3226,
      "step": 92
    },
    {
      "epoch": 0.26628489620615603,
      "grad_norm": 0.2113342136144638,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.3214,
      "step": 93
    },
    {
      "epoch": 0.2691481746599857,
      "grad_norm": 0.22217495739459991,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3141,
      "step": 94
    },
    {
      "epoch": 0.2720114531138153,
      "grad_norm": 0.22695213556289673,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.2984,
      "step": 95
    },
    {
      "epoch": 0.27487473156764497,
      "grad_norm": 0.2394181191921234,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3414,
      "step": 96
    },
    {
      "epoch": 0.2777380100214746,
      "grad_norm": 0.23146599531173706,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2676,
      "step": 97
    },
    {
      "epoch": 0.28060128847530424,
      "grad_norm": 0.2493172585964203,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3123,
      "step": 98
    },
    {
      "epoch": 0.28346456692913385,
      "grad_norm": 0.24938414990901947,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.2926,
      "step": 99
    },
    {
      "epoch": 0.2863278453829635,
      "grad_norm": 0.2769395709037781,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3201,
      "step": 100
    },
    {
      "epoch": 0.2863278453829635,
      "eval_loss": 0.33835068345069885,
      "eval_runtime": 21.4648,
      "eval_samples_per_second": 27.44,
      "eval_steps_per_second": 6.895,
      "step": 100
    },
    {
      "epoch": 0.28919112383679313,
      "grad_norm": 0.15336519479751587,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.4879,
      "step": 101
    },
    {
      "epoch": 0.29205440229062274,
      "grad_norm": 0.15915046632289886,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.3595,
      "step": 102
    },
    {
      "epoch": 0.2949176807444524,
      "grad_norm": 0.1633385866880417,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3382,
      "step": 103
    },
    {
      "epoch": 0.297780959198282,
      "grad_norm": 0.13733084499835968,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.2963,
      "step": 104
    },
    {
      "epoch": 0.3006442376521117,
      "grad_norm": 0.13345304131507874,
      "learning_rate": 5e-05,
      "loss": 0.3196,
      "step": 105
    },
    {
      "epoch": 0.3035075161059413,
      "grad_norm": 0.12916503846645355,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3063,
      "step": 106
    },
    {
      "epoch": 0.30637079455977095,
      "grad_norm": 0.13057591021060944,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3468,
      "step": 107
    },
    {
      "epoch": 0.30923407301360056,
      "grad_norm": 0.12561380863189697,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.2917,
      "step": 108
    },
    {
      "epoch": 0.31209735146743023,
      "grad_norm": 0.1320079118013382,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.3384,
      "step": 109
    },
    {
      "epoch": 0.31496062992125984,
      "grad_norm": 0.13980033993721008,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3761,
      "step": 110
    },
    {
      "epoch": 0.3178239083750895,
      "grad_norm": 0.1449955552816391,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.3598,
      "step": 111
    },
    {
      "epoch": 0.3206871868289191,
      "grad_norm": 0.1465611457824707,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.3803,
      "step": 112
    },
    {
      "epoch": 0.3235504652827487,
      "grad_norm": 0.14949806034564972,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3847,
      "step": 113
    },
    {
      "epoch": 0.3264137437365784,
      "grad_norm": 0.15518344938755035,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3539,
      "step": 114
    },
    {
      "epoch": 0.329277022190408,
      "grad_norm": 0.15582066774368286,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3507,
      "step": 115
    },
    {
      "epoch": 0.33214030064423766,
      "grad_norm": 0.14963629841804504,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3177,
      "step": 116
    },
    {
      "epoch": 0.3350035790980673,
      "grad_norm": 0.14899323880672455,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3382,
      "step": 117
    },
    {
      "epoch": 0.33786685755189694,
      "grad_norm": 0.1565190851688385,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.35,
      "step": 118
    },
    {
      "epoch": 0.34073013600572655,
      "grad_norm": 0.14976900815963745,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3317,
      "step": 119
    },
    {
      "epoch": 0.3435934144595562,
      "grad_norm": 0.15968285501003265,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3509,
      "step": 120
    },
    {
      "epoch": 0.3464566929133858,
      "grad_norm": 0.15129564702510834,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.292,
      "step": 121
    },
    {
      "epoch": 0.3493199713672155,
      "grad_norm": 0.15293705463409424,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3051,
      "step": 122
    },
    {
      "epoch": 0.3521832498210451,
      "grad_norm": 0.1715923696756363,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3925,
      "step": 123
    },
    {
      "epoch": 0.3550465282748747,
      "grad_norm": 0.17057475447654724,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3442,
      "step": 124
    },
    {
      "epoch": 0.35790980672870437,
      "grad_norm": 0.16853640973567963,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3445,
      "step": 125
    },
    {
      "epoch": 0.360773085182534,
      "grad_norm": 0.18137936294078827,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.4118,
      "step": 126
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.16989320516586304,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3282,
      "step": 127
    },
    {
      "epoch": 0.36649964209019326,
      "grad_norm": 0.17703837156295776,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3445,
      "step": 128
    },
    {
      "epoch": 0.3693629205440229,
      "grad_norm": 0.17757730185985565,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.3631,
      "step": 129
    },
    {
      "epoch": 0.37222619899785253,
      "grad_norm": 0.17194613814353943,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3341,
      "step": 130
    },
    {
      "epoch": 0.3750894774516822,
      "grad_norm": 0.18289253115653992,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.3455,
      "step": 131
    },
    {
      "epoch": 0.3779527559055118,
      "grad_norm": 0.18134768307209015,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3361,
      "step": 132
    },
    {
      "epoch": 0.38081603435934147,
      "grad_norm": 0.18206030130386353,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.3431,
      "step": 133
    },
    {
      "epoch": 0.3836793128131711,
      "grad_norm": 0.18371394276618958,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3462,
      "step": 134
    },
    {
      "epoch": 0.3865425912670007,
      "grad_norm": 0.18305209279060364,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3131,
      "step": 135
    },
    {
      "epoch": 0.38940586972083036,
      "grad_norm": 0.18985417485237122,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2999,
      "step": 136
    },
    {
      "epoch": 0.39226914817465997,
      "grad_norm": 0.1840282678604126,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.304,
      "step": 137
    },
    {
      "epoch": 0.39513242662848963,
      "grad_norm": 0.1849391907453537,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3248,
      "step": 138
    },
    {
      "epoch": 0.39799570508231924,
      "grad_norm": 0.19603252410888672,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3481,
      "step": 139
    },
    {
      "epoch": 0.4008589835361489,
      "grad_norm": 0.19191032648086548,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2969,
      "step": 140
    },
    {
      "epoch": 0.4037222619899785,
      "grad_norm": 0.20839765667915344,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.3766,
      "step": 141
    },
    {
      "epoch": 0.4065855404438082,
      "grad_norm": 0.19553877413272858,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2959,
      "step": 142
    },
    {
      "epoch": 0.4094488188976378,
      "grad_norm": 0.19988681375980377,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.28,
      "step": 143
    },
    {
      "epoch": 0.41231209735146745,
      "grad_norm": 0.2023603767156601,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3327,
      "step": 144
    },
    {
      "epoch": 0.41517537580529706,
      "grad_norm": 0.21418696641921997,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3058,
      "step": 145
    },
    {
      "epoch": 0.4180386542591267,
      "grad_norm": 0.21737684309482574,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.2928,
      "step": 146
    },
    {
      "epoch": 0.42090193271295634,
      "grad_norm": 0.2245236486196518,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2979,
      "step": 147
    },
    {
      "epoch": 0.42376521116678595,
      "grad_norm": 0.2391463667154312,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.3299,
      "step": 148
    },
    {
      "epoch": 0.4266284896206156,
      "grad_norm": 0.2674826979637146,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3223,
      "step": 149
    },
    {
      "epoch": 0.4294917680744452,
      "grad_norm": 0.2706792950630188,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.2979,
      "step": 150
    },
    {
      "epoch": 0.4294917680744452,
      "eval_loss": 0.330860435962677,
      "eval_runtime": 21.4441,
      "eval_samples_per_second": 27.467,
      "eval_steps_per_second": 6.902,
      "step": 150
    },
    {
      "epoch": 0.4323550465282749,
      "grad_norm": 0.12367428094148636,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2857,
      "step": 151
    },
    {
      "epoch": 0.4352183249821045,
      "grad_norm": 0.12729761004447937,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.3397,
      "step": 152
    },
    {
      "epoch": 0.43808160343593416,
      "grad_norm": 0.1257731318473816,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.337,
      "step": 153
    },
    {
      "epoch": 0.4409448818897638,
      "grad_norm": 0.1275026798248291,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3017,
      "step": 154
    },
    {
      "epoch": 0.44380816034359344,
      "grad_norm": 0.12171600759029388,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.2776,
      "step": 155
    },
    {
      "epoch": 0.44667143879742305,
      "grad_norm": 0.13033492863178253,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.3003,
      "step": 156
    },
    {
      "epoch": 0.44953471725125266,
      "grad_norm": 0.1285054087638855,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.2827,
      "step": 157
    },
    {
      "epoch": 0.4523979957050823,
      "grad_norm": 0.12316196411848068,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2823,
      "step": 158
    },
    {
      "epoch": 0.45526127415891193,
      "grad_norm": 0.1336871236562729,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.323,
      "step": 159
    },
    {
      "epoch": 0.4581245526127416,
      "grad_norm": 0.1375422179698944,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.3663,
      "step": 160
    },
    {
      "epoch": 0.4609878310665712,
      "grad_norm": 0.13683415949344635,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3305,
      "step": 161
    },
    {
      "epoch": 0.4638511095204009,
      "grad_norm": 0.13866949081420898,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.3195,
      "step": 162
    },
    {
      "epoch": 0.4667143879742305,
      "grad_norm": 0.1441594958305359,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.3154,
      "step": 163
    },
    {
      "epoch": 0.46957766642806015,
      "grad_norm": 0.14356538653373718,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.3202,
      "step": 164
    },
    {
      "epoch": 0.47244094488188976,
      "grad_norm": 0.13742604851722717,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.3129,
      "step": 165
    },
    {
      "epoch": 0.4753042233357194,
      "grad_norm": 0.14718621969223022,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.3551,
      "step": 166
    },
    {
      "epoch": 0.47816750178954903,
      "grad_norm": 0.1457807570695877,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3406,
      "step": 167
    },
    {
      "epoch": 0.48103078024337864,
      "grad_norm": 0.14546234905719757,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.3002,
      "step": 168
    },
    {
      "epoch": 0.4838940586972083,
      "grad_norm": 0.1526423543691635,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.3228,
      "step": 169
    },
    {
      "epoch": 0.4867573371510379,
      "grad_norm": 0.15682844817638397,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3267,
      "step": 170
    },
    {
      "epoch": 0.4896206156048676,
      "grad_norm": 0.15266835689544678,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3594,
      "step": 171
    },
    {
      "epoch": 0.4924838940586972,
      "grad_norm": 0.1588318943977356,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3587,
      "step": 172
    },
    {
      "epoch": 0.49534717251252686,
      "grad_norm": 0.15571781992912292,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.3407,
      "step": 173
    },
    {
      "epoch": 0.49821045096635647,
      "grad_norm": 0.16197878122329712,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.3177,
      "step": 174
    },
    {
      "epoch": 0.5010737294201861,
      "grad_norm": 0.1555037647485733,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3258,
      "step": 175
    },
    {
      "epoch": 0.5039370078740157,
      "grad_norm": 0.17083266377449036,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3681,
      "step": 176
    },
    {
      "epoch": 0.5068002863278454,
      "grad_norm": 0.17624787986278534,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.4137,
      "step": 177
    },
    {
      "epoch": 0.5096635647816751,
      "grad_norm": 0.17016418278217316,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.3502,
      "step": 178
    },
    {
      "epoch": 0.5125268432355047,
      "grad_norm": 0.16492441296577454,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3317,
      "step": 179
    },
    {
      "epoch": 0.5153901216893343,
      "grad_norm": 0.16688668727874756,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.325,
      "step": 180
    },
    {
      "epoch": 0.5182534001431639,
      "grad_norm": 0.17033790051937103,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3064,
      "step": 181
    },
    {
      "epoch": 0.5211166785969935,
      "grad_norm": 0.17435817420482635,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3438,
      "step": 182
    },
    {
      "epoch": 0.5239799570508232,
      "grad_norm": 0.16841447353363037,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2822,
      "step": 183
    },
    {
      "epoch": 0.5268432355046528,
      "grad_norm": 0.19512639939785004,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3788,
      "step": 184
    },
    {
      "epoch": 0.5297065139584825,
      "grad_norm": 0.18789996206760406,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.3682,
      "step": 185
    },
    {
      "epoch": 0.5325697924123121,
      "grad_norm": 0.18214291334152222,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.3343,
      "step": 186
    },
    {
      "epoch": 0.5354330708661418,
      "grad_norm": 0.17767339944839478,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2733,
      "step": 187
    },
    {
      "epoch": 0.5382963493199714,
      "grad_norm": 0.2006862610578537,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.3713,
      "step": 188
    },
    {
      "epoch": 0.541159627773801,
      "grad_norm": 0.1960771530866623,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.3477,
      "step": 189
    },
    {
      "epoch": 0.5440229062276306,
      "grad_norm": 0.20107898116111755,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.3653,
      "step": 190
    },
    {
      "epoch": 0.5468861846814602,
      "grad_norm": 0.20176103711128235,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.3329,
      "step": 191
    },
    {
      "epoch": 0.5497494631352899,
      "grad_norm": 0.20580102503299713,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.3109,
      "step": 192
    },
    {
      "epoch": 0.5526127415891195,
      "grad_norm": 0.20618514716625214,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2971,
      "step": 193
    },
    {
      "epoch": 0.5554760200429492,
      "grad_norm": 0.21497078239917755,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2956,
      "step": 194
    },
    {
      "epoch": 0.5583392984967788,
      "grad_norm": 0.23176227509975433,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.3758,
      "step": 195
    },
    {
      "epoch": 0.5612025769506085,
      "grad_norm": 0.22729800641536713,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.3288,
      "step": 196
    },
    {
      "epoch": 0.5640658554044381,
      "grad_norm": 0.22445441782474518,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.2772,
      "step": 197
    },
    {
      "epoch": 0.5669291338582677,
      "grad_norm": 0.22773410379886627,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.2505,
      "step": 198
    },
    {
      "epoch": 0.5697924123120973,
      "grad_norm": 0.24435363709926605,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2642,
      "step": 199
    },
    {
      "epoch": 0.572655690765927,
      "grad_norm": 0.28229326009750366,
      "learning_rate": 0.0,
      "loss": 0.3031,
      "step": 200
    },
    {
      "epoch": 0.572655690765927,
      "eval_loss": 0.32964658737182617,
      "eval_runtime": 21.5704,
      "eval_samples_per_second": 27.306,
      "eval_steps_per_second": 6.861,
      "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": 6.762796557533184e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}