{
  "best_metric": 0.5490885972976685,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7279344858962693,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003639672429481347,
      "grad_norm": 1.6156786680221558,
      "learning_rate": 1e-05,
      "loss": 4.1589,
      "step": 1
    },
    {
      "epoch": 0.003639672429481347,
      "eval_loss": 1.79654061794281,
      "eval_runtime": 47.1389,
      "eval_samples_per_second": 9.822,
      "eval_steps_per_second": 2.461,
      "step": 1
    },
    {
      "epoch": 0.007279344858962694,
      "grad_norm": 1.8971140384674072,
      "learning_rate": 2e-05,
      "loss": 5.4649,
      "step": 2
    },
    {
      "epoch": 0.01091901728844404,
      "grad_norm": 2.1200432777404785,
      "learning_rate": 3e-05,
      "loss": 6.044,
      "step": 3
    },
    {
      "epoch": 0.014558689717925387,
      "grad_norm": 1.9115948677062988,
      "learning_rate": 4e-05,
      "loss": 5.4719,
      "step": 4
    },
    {
      "epoch": 0.018198362147406732,
      "grad_norm": 2.261807680130005,
      "learning_rate": 5e-05,
      "loss": 7.0767,
      "step": 5
    },
    {
      "epoch": 0.02183803457688808,
      "grad_norm": 2.1007025241851807,
      "learning_rate": 6e-05,
      "loss": 5.9653,
      "step": 6
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 2.007903814315796,
      "learning_rate": 7e-05,
      "loss": 6.3614,
      "step": 7
    },
    {
      "epoch": 0.029117379435850774,
      "grad_norm": 2.3542909622192383,
      "learning_rate": 8e-05,
      "loss": 7.0339,
      "step": 8
    },
    {
      "epoch": 0.03275705186533212,
      "grad_norm": 2.008129119873047,
      "learning_rate": 9e-05,
      "loss": 4.6048,
      "step": 9
    },
    {
      "epoch": 0.036396724294813464,
      "grad_norm": 2.67560076713562,
      "learning_rate": 0.0001,
      "loss": 4.6206,
      "step": 10
    },
    {
      "epoch": 0.040036396724294813,
      "grad_norm": 2.982059955596924,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.9534,
      "step": 11
    },
    {
      "epoch": 0.04367606915377616,
      "grad_norm": 3.115858793258667,
      "learning_rate": 9.997266286704631e-05,
      "loss": 5.3116,
      "step": 12
    },
    {
      "epoch": 0.047315741583257506,
      "grad_norm": 2.690213918685913,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.4884,
      "step": 13
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 2.5859742164611816,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.3171,
      "step": 14
    },
    {
      "epoch": 0.0545950864422202,
      "grad_norm": 2.9327213764190674,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.0863,
      "step": 15
    },
    {
      "epoch": 0.05823475887170155,
      "grad_norm": 3.020179510116577,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.6601,
      "step": 16
    },
    {
      "epoch": 0.06187443130118289,
      "grad_norm": 3.243584394454956,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.1835,
      "step": 17
    },
    {
      "epoch": 0.06551410373066424,
      "grad_norm": 2.7714054584503174,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.821,
      "step": 18
    },
    {
      "epoch": 0.06915377616014559,
      "grad_norm": 2.482257843017578,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.0616,
      "step": 19
    },
    {
      "epoch": 0.07279344858962693,
      "grad_norm": 2.5543320178985596,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.5511,
      "step": 20
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 3.0923211574554443,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.3053,
      "step": 21
    },
    {
      "epoch": 0.08007279344858963,
      "grad_norm": 3.088024377822876,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.7435,
      "step": 22
    },
    {
      "epoch": 0.08371246587807098,
      "grad_norm": 2.2931549549102783,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.3854,
      "step": 23
    },
    {
      "epoch": 0.08735213830755233,
      "grad_norm": 3.452805519104004,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.6226,
      "step": 24
    },
    {
      "epoch": 0.09099181073703366,
      "grad_norm": 4.084782600402832,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.2288,
      "step": 25
    },
    {
      "epoch": 0.09463148316651501,
      "grad_norm": 3.8011066913604736,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.476,
      "step": 26
    },
    {
      "epoch": 0.09827115559599636,
      "grad_norm": 2.913602590560913,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.6165,
      "step": 27
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 3.3534297943115234,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.1292,
      "step": 28
    },
    {
      "epoch": 0.10555050045495905,
      "grad_norm": 3.159245729446411,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.6337,
      "step": 29
    },
    {
      "epoch": 0.1091901728844404,
      "grad_norm": 4.087600231170654,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.2053,
      "step": 30
    },
    {
      "epoch": 0.11282984531392175,
      "grad_norm": 3.511906385421753,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.6733,
      "step": 31
    },
    {
      "epoch": 0.1164695177434031,
      "grad_norm": 3.3889596462249756,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.7836,
      "step": 32
    },
    {
      "epoch": 0.12010919017288443,
      "grad_norm": 3.1994011402130127,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.3278,
      "step": 33
    },
    {
      "epoch": 0.12374886260236578,
      "grad_norm": 3.325439691543579,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.4156,
      "step": 34
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 3.169724464416504,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.1496,
      "step": 35
    },
    {
      "epoch": 0.13102820746132848,
      "grad_norm": 3.8054027557373047,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.4958,
      "step": 36
    },
    {
      "epoch": 0.13466787989080983,
      "grad_norm": 3.272763729095459,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.3798,
      "step": 37
    },
    {
      "epoch": 0.13830755232029118,
      "grad_norm": 4.186960220336914,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.9526,
      "step": 38
    },
    {
      "epoch": 0.14194722474977253,
      "grad_norm": 3.5222666263580322,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.53,
      "step": 39
    },
    {
      "epoch": 0.14558689717925385,
      "grad_norm": 3.084665536880493,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.8511,
      "step": 40
    },
    {
      "epoch": 0.1492265696087352,
      "grad_norm": 3.7815001010894775,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.0269,
      "step": 41
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 3.346134901046753,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.3294,
      "step": 42
    },
    {
      "epoch": 0.1565059144676979,
      "grad_norm": 3.176572799682617,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.0126,
      "step": 43
    },
    {
      "epoch": 0.16014558689717925,
      "grad_norm": 3.521372079849243,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.3762,
      "step": 44
    },
    {
      "epoch": 0.1637852593266606,
      "grad_norm": 3.4461355209350586,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.0099,
      "step": 45
    },
    {
      "epoch": 0.16742493175614195,
      "grad_norm": 3.5269594192504883,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.3313,
      "step": 46
    },
    {
      "epoch": 0.1710646041856233,
      "grad_norm": 4.0336198806762695,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.4732,
      "step": 47
    },
    {
      "epoch": 0.17470427661510465,
      "grad_norm": 4.041784763336182,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.4551,
      "step": 48
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 4.23267936706543,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.1445,
      "step": 49
    },
    {
      "epoch": 0.18198362147406733,
      "grad_norm": 5.671982765197754,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.9622,
      "step": 50
    },
    {
      "epoch": 0.18198362147406733,
      "eval_loss": 0.7934085130691528,
      "eval_runtime": 47.5584,
      "eval_samples_per_second": 9.735,
      "eval_steps_per_second": 2.439,
      "step": 50
    },
    {
      "epoch": 0.18562329390354868,
      "grad_norm": 6.55305814743042,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.676,
      "step": 51
    },
    {
      "epoch": 0.18926296633303002,
      "grad_norm": 5.221794128417969,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.2433,
      "step": 52
    },
    {
      "epoch": 0.19290263876251137,
      "grad_norm": 3.5830085277557373,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.1168,
      "step": 53
    },
    {
      "epoch": 0.19654231119199272,
      "grad_norm": 3.0552215576171875,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.219,
      "step": 54
    },
    {
      "epoch": 0.20018198362147407,
      "grad_norm": 2.9120144844055176,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.1183,
      "step": 55
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 2.764294147491455,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.4571,
      "step": 56
    },
    {
      "epoch": 0.20746132848043677,
      "grad_norm": 2.788884401321411,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.9288,
      "step": 57
    },
    {
      "epoch": 0.2111010009099181,
      "grad_norm": 3.0763115882873535,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.6845,
      "step": 58
    },
    {
      "epoch": 0.21474067333939945,
      "grad_norm": 2.6628854274749756,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.1692,
      "step": 59
    },
    {
      "epoch": 0.2183803457688808,
      "grad_norm": 2.8497138023376465,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.0371,
      "step": 60
    },
    {
      "epoch": 0.22202001819836215,
      "grad_norm": 2.802095651626587,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.1548,
      "step": 61
    },
    {
      "epoch": 0.2256596906278435,
      "grad_norm": 3.0266306400299072,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.0323,
      "step": 62
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 2.807077169418335,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.0861,
      "step": 63
    },
    {
      "epoch": 0.2329390354868062,
      "grad_norm": 2.65042781829834,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.9227,
      "step": 64
    },
    {
      "epoch": 0.23657870791628755,
      "grad_norm": 3.1444993019104004,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.0296,
      "step": 65
    },
    {
      "epoch": 0.24021838034576887,
      "grad_norm": 2.8281092643737793,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.9286,
      "step": 66
    },
    {
      "epoch": 0.24385805277525022,
      "grad_norm": 3.051706552505493,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.1729,
      "step": 67
    },
    {
      "epoch": 0.24749772520473157,
      "grad_norm": 2.6119096279144287,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3656,
      "step": 68
    },
    {
      "epoch": 0.25113739763421294,
      "grad_norm": 2.4973325729370117,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.4181,
      "step": 69
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 3.098081350326538,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.2116,
      "step": 70
    },
    {
      "epoch": 0.2584167424931756,
      "grad_norm": 3.3221497535705566,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.1,
      "step": 71
    },
    {
      "epoch": 0.26205641492265697,
      "grad_norm": 2.184515953063965,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.9502,
      "step": 72
    },
    {
      "epoch": 0.2656960873521383,
      "grad_norm": 2.9642996788024902,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.4756,
      "step": 73
    },
    {
      "epoch": 0.26933575978161967,
      "grad_norm": 2.822319984436035,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.5607,
      "step": 74
    },
    {
      "epoch": 0.272975432211101,
      "grad_norm": 3.025736093521118,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.6181,
      "step": 75
    },
    {
      "epoch": 0.27661510464058237,
      "grad_norm": 2.5108871459960938,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.0465,
      "step": 76
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 3.347212314605713,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.642,
      "step": 77
    },
    {
      "epoch": 0.28389444949954507,
      "grad_norm": 3.042344331741333,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.2472,
      "step": 78
    },
    {
      "epoch": 0.2875341219290264,
      "grad_norm": 3.8693716526031494,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.796,
      "step": 79
    },
    {
      "epoch": 0.2911737943585077,
      "grad_norm": 3.328456163406372,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.3817,
      "step": 80
    },
    {
      "epoch": 0.2948134667879891,
      "grad_norm": 3.721893310546875,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.6781,
      "step": 81
    },
    {
      "epoch": 0.2984531392174704,
      "grad_norm": 3.2879714965820312,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.3935,
      "step": 82
    },
    {
      "epoch": 0.3020928116469518,
      "grad_norm": 3.9299476146698,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.7053,
      "step": 83
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": 3.058401346206665,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.0633,
      "step": 84
    },
    {
      "epoch": 0.3093721565059145,
      "grad_norm": 3.3975331783294678,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.5576,
      "step": 85
    },
    {
      "epoch": 0.3130118289353958,
      "grad_norm": 3.3951075077056885,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.2732,
      "step": 86
    },
    {
      "epoch": 0.3166515013648772,
      "grad_norm": 3.2836809158325195,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.4419,
      "step": 87
    },
    {
      "epoch": 0.3202911737943585,
      "grad_norm": 2.5833373069763184,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.8533,
      "step": 88
    },
    {
      "epoch": 0.32393084622383983,
      "grad_norm": 3.167266368865967,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.1784,
      "step": 89
    },
    {
      "epoch": 0.3275705186533212,
      "grad_norm": 2.5674991607666016,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.5591,
      "step": 90
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": 3.5078673362731934,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.4201,
      "step": 91
    },
    {
      "epoch": 0.3348498635122839,
      "grad_norm": 2.5641889572143555,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.6499,
      "step": 92
    },
    {
      "epoch": 0.33848953594176523,
      "grad_norm": 3.426968812942505,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.1176,
      "step": 93
    },
    {
      "epoch": 0.3421292083712466,
      "grad_norm": 2.7209270000457764,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.7467,
      "step": 94
    },
    {
      "epoch": 0.34576888080072793,
      "grad_norm": 3.288771152496338,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.9233,
      "step": 95
    },
    {
      "epoch": 0.3494085532302093,
      "grad_norm": 2.536733627319336,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.6057,
      "step": 96
    },
    {
      "epoch": 0.35304822565969063,
      "grad_norm": 3.5754098892211914,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.2117,
      "step": 97
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": 3.674102783203125,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.4695,
      "step": 98
    },
    {
      "epoch": 0.36032757051865333,
      "grad_norm": 4.0804338455200195,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.3888,
      "step": 99
    },
    {
      "epoch": 0.36396724294813465,
      "grad_norm": 3.5864241123199463,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.9172,
      "step": 100
    },
    {
      "epoch": 0.36396724294813465,
      "eval_loss": 0.647607684135437,
      "eval_runtime": 47.5852,
      "eval_samples_per_second": 9.73,
      "eval_steps_per_second": 2.438,
      "step": 100
    },
    {
      "epoch": 0.36760691537761603,
      "grad_norm": 3.283327102661133,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.7434,
      "step": 101
    },
    {
      "epoch": 0.37124658780709735,
      "grad_norm": 3.6895530223846436,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.405,
      "step": 102
    },
    {
      "epoch": 0.37488626023657873,
      "grad_norm": 3.438565254211426,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.0602,
      "step": 103
    },
    {
      "epoch": 0.37852593266606005,
      "grad_norm": 2.9382526874542236,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.9614,
      "step": 104
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": 2.9088261127471924,
      "learning_rate": 5e-05,
      "loss": 3.1809,
      "step": 105
    },
    {
      "epoch": 0.38580527752502275,
      "grad_norm": 3.2111284732818604,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.2037,
      "step": 106
    },
    {
      "epoch": 0.38944494995450407,
      "grad_norm": 3.026944875717163,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.5522,
      "step": 107
    },
    {
      "epoch": 0.39308462238398545,
      "grad_norm": 2.5377681255340576,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.0077,
      "step": 108
    },
    {
      "epoch": 0.39672429481346677,
      "grad_norm": 2.589021921157837,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.8889,
      "step": 109
    },
    {
      "epoch": 0.40036396724294815,
      "grad_norm": 2.914215564727783,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.9254,
      "step": 110
    },
    {
      "epoch": 0.40400363967242947,
      "grad_norm": 2.4525372982025146,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.8066,
      "step": 111
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": 2.617382049560547,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.7077,
      "step": 112
    },
    {
      "epoch": 0.41128298453139217,
      "grad_norm": 2.7838759422302246,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.9065,
      "step": 113
    },
    {
      "epoch": 0.41492265696087355,
      "grad_norm": 2.6013290882110596,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.8385,
      "step": 114
    },
    {
      "epoch": 0.41856232939035487,
      "grad_norm": 2.6603236198425293,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.5664,
      "step": 115
    },
    {
      "epoch": 0.4222020018198362,
      "grad_norm": 2.7078967094421387,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.904,
      "step": 116
    },
    {
      "epoch": 0.42584167424931757,
      "grad_norm": 2.647740125656128,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.7269,
      "step": 117
    },
    {
      "epoch": 0.4294813466787989,
      "grad_norm": 2.8882391452789307,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.7514,
      "step": 118
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": 2.4470083713531494,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.3616,
      "step": 119
    },
    {
      "epoch": 0.4367606915377616,
      "grad_norm": 2.3307876586914062,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.1264,
      "step": 120
    },
    {
      "epoch": 0.44040036396724297,
      "grad_norm": 2.5252671241760254,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.5383,
      "step": 121
    },
    {
      "epoch": 0.4440400363967243,
      "grad_norm": 2.5042293071746826,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.9413,
      "step": 122
    },
    {
      "epoch": 0.44767970882620567,
      "grad_norm": 3.003763437271118,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.7041,
      "step": 123
    },
    {
      "epoch": 0.451319381255687,
      "grad_norm": 3.1650874614715576,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.6616,
      "step": 124
    },
    {
      "epoch": 0.4549590536851683,
      "grad_norm": 4.306475639343262,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.1808,
      "step": 125
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": 2.983140230178833,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.3687,
      "step": 126
    },
    {
      "epoch": 0.462238398544131,
      "grad_norm": 2.767782211303711,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.309,
      "step": 127
    },
    {
      "epoch": 0.4658780709736124,
      "grad_norm": 2.729362964630127,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.0803,
      "step": 128
    },
    {
      "epoch": 0.4695177434030937,
      "grad_norm": 2.3514013290405273,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.7072,
      "step": 129
    },
    {
      "epoch": 0.4731574158325751,
      "grad_norm": 2.9426000118255615,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.96,
      "step": 130
    },
    {
      "epoch": 0.4767970882620564,
      "grad_norm": 3.09873104095459,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.103,
      "step": 131
    },
    {
      "epoch": 0.48043676069153773,
      "grad_norm": 2.6302735805511475,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.7701,
      "step": 132
    },
    {
      "epoch": 0.4840764331210191,
      "grad_norm": 3.5831170082092285,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.3424,
      "step": 133
    },
    {
      "epoch": 0.48771610555050043,
      "grad_norm": 2.947216510772705,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.005,
      "step": 134
    },
    {
      "epoch": 0.4913557779799818,
      "grad_norm": 2.44584059715271,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.8008,
      "step": 135
    },
    {
      "epoch": 0.49499545040946313,
      "grad_norm": 2.835815668106079,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.6431,
      "step": 136
    },
    {
      "epoch": 0.4986351228389445,
      "grad_norm": 2.9566168785095215,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.8972,
      "step": 137
    },
    {
      "epoch": 0.5022747952684259,
      "grad_norm": 2.8612327575683594,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.6708,
      "step": 138
    },
    {
      "epoch": 0.5059144676979072,
      "grad_norm": 3.2084693908691406,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.0785,
      "step": 139
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 2.7801413536071777,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.9208,
      "step": 140
    },
    {
      "epoch": 0.5131938125568699,
      "grad_norm": 3.053459644317627,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.7271,
      "step": 141
    },
    {
      "epoch": 0.5168334849863512,
      "grad_norm": 3.345618486404419,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.8456,
      "step": 142
    },
    {
      "epoch": 0.5204731574158326,
      "grad_norm": 2.718886613845825,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.7926,
      "step": 143
    },
    {
      "epoch": 0.5241128298453139,
      "grad_norm": 3.0628650188446045,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.6218,
      "step": 144
    },
    {
      "epoch": 0.5277525022747953,
      "grad_norm": 3.3708438873291016,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.2188,
      "step": 145
    },
    {
      "epoch": 0.5313921747042766,
      "grad_norm": 3.4302098751068115,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.9789,
      "step": 146
    },
    {
      "epoch": 0.535031847133758,
      "grad_norm": 3.524137258529663,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.1485,
      "step": 147
    },
    {
      "epoch": 0.5386715195632393,
      "grad_norm": 3.4503889083862305,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0008,
      "step": 148
    },
    {
      "epoch": 0.5423111919927207,
      "grad_norm": 4.520384311676025,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.6887,
      "step": 149
    },
    {
      "epoch": 0.545950864422202,
      "grad_norm": 4.837410926818848,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.6012,
      "step": 150
    },
    {
      "epoch": 0.545950864422202,
      "eval_loss": 0.5721358060836792,
      "eval_runtime": 47.5707,
      "eval_samples_per_second": 9.733,
      "eval_steps_per_second": 2.438,
      "step": 150
    },
    {
      "epoch": 0.5495905368516834,
      "grad_norm": 2.302870035171509,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.8646,
      "step": 151
    },
    {
      "epoch": 0.5532302092811647,
      "grad_norm": 2.3190178871154785,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9375,
      "step": 152
    },
    {
      "epoch": 0.556869881710646,
      "grad_norm": 3.0300214290618896,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.7096,
      "step": 153
    },
    {
      "epoch": 0.5605095541401274,
      "grad_norm": 3.025632381439209,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.897,
      "step": 154
    },
    {
      "epoch": 0.5641492265696088,
      "grad_norm": 2.839578866958618,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.6637,
      "step": 155
    },
    {
      "epoch": 0.5677888989990901,
      "grad_norm": 3.25244402885437,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.0786,
      "step": 156
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 3.0265636444091797,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.5205,
      "step": 157
    },
    {
      "epoch": 0.5750682438580528,
      "grad_norm": 2.8974204063415527,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.3908,
      "step": 158
    },
    {
      "epoch": 0.5787079162875342,
      "grad_norm": 3.050604820251465,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.93,
      "step": 159
    },
    {
      "epoch": 0.5823475887170154,
      "grad_norm": 3.0116374492645264,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.7721,
      "step": 160
    },
    {
      "epoch": 0.5859872611464968,
      "grad_norm": 3.1212122440338135,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.6728,
      "step": 161
    },
    {
      "epoch": 0.5896269335759782,
      "grad_norm": 3.024791955947876,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.8683,
      "step": 162
    },
    {
      "epoch": 0.5932666060054596,
      "grad_norm": 3.1855437755584717,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.9764,
      "step": 163
    },
    {
      "epoch": 0.5969062784349408,
      "grad_norm": 2.8845460414886475,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.5825,
      "step": 164
    },
    {
      "epoch": 0.6005459508644222,
      "grad_norm": 2.7672841548919678,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.5022,
      "step": 165
    },
    {
      "epoch": 0.6041856232939036,
      "grad_norm": 2.711394786834717,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.158,
      "step": 166
    },
    {
      "epoch": 0.607825295723385,
      "grad_norm": 2.831782579421997,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.2504,
      "step": 167
    },
    {
      "epoch": 0.6114649681528662,
      "grad_norm": 2.912217378616333,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.2321,
      "step": 168
    },
    {
      "epoch": 0.6151046405823476,
      "grad_norm": 2.7623188495635986,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.2487,
      "step": 169
    },
    {
      "epoch": 0.618744313011829,
      "grad_norm": 2.90129017829895,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.3792,
      "step": 170
    },
    {
      "epoch": 0.6223839854413102,
      "grad_norm": 3.3646767139434814,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.8504,
      "step": 171
    },
    {
      "epoch": 0.6260236578707916,
      "grad_norm": 3.3314194679260254,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.4542,
      "step": 172
    },
    {
      "epoch": 0.629663330300273,
      "grad_norm": 3.1008663177490234,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.2329,
      "step": 173
    },
    {
      "epoch": 0.6333030027297544,
      "grad_norm": 2.999218225479126,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.4201,
      "step": 174
    },
    {
      "epoch": 0.6369426751592356,
      "grad_norm": 2.6040823459625244,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.959,
      "step": 175
    },
    {
      "epoch": 0.640582347588717,
      "grad_norm": 2.9458696842193604,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2246,
      "step": 176
    },
    {
      "epoch": 0.6442220200181984,
      "grad_norm": 3.246384620666504,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.6504,
      "step": 177
    },
    {
      "epoch": 0.6478616924476797,
      "grad_norm": 3.316885232925415,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5356,
      "step": 178
    },
    {
      "epoch": 0.651501364877161,
      "grad_norm": 2.5368857383728027,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.9127,
      "step": 179
    },
    {
      "epoch": 0.6551410373066424,
      "grad_norm": 2.7126615047454834,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.0505,
      "step": 180
    },
    {
      "epoch": 0.6587807097361238,
      "grad_norm": 3.049729585647583,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.2564,
      "step": 181
    },
    {
      "epoch": 0.6624203821656051,
      "grad_norm": 3.1560466289520264,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.059,
      "step": 182
    },
    {
      "epoch": 0.6660600545950864,
      "grad_norm": 2.5500495433807373,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.6247,
      "step": 183
    },
    {
      "epoch": 0.6696997270245678,
      "grad_norm": 2.9407999515533447,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.1153,
      "step": 184
    },
    {
      "epoch": 0.6733393994540491,
      "grad_norm": 2.8330116271972656,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.7676,
      "step": 185
    },
    {
      "epoch": 0.6769790718835305,
      "grad_norm": 2.8989205360412598,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.7266,
      "step": 186
    },
    {
      "epoch": 0.6806187443130118,
      "grad_norm": 3.519871473312378,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.0414,
      "step": 187
    },
    {
      "epoch": 0.6842584167424932,
      "grad_norm": 2.6813058853149414,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.684,
      "step": 188
    },
    {
      "epoch": 0.6878980891719745,
      "grad_norm": 2.557551860809326,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.6045,
      "step": 189
    },
    {
      "epoch": 0.6915377616014559,
      "grad_norm": 3.1362481117248535,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.0943,
      "step": 190
    },
    {
      "epoch": 0.6951774340309372,
      "grad_norm": 2.4160473346710205,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.6056,
      "step": 191
    },
    {
      "epoch": 0.6988171064604186,
      "grad_norm": 3.358062744140625,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.0739,
      "step": 192
    },
    {
      "epoch": 0.7024567788898999,
      "grad_norm": 3.103926420211792,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.9202,
      "step": 193
    },
    {
      "epoch": 0.7060964513193813,
      "grad_norm": 3.2157270908355713,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.9288,
      "step": 194
    },
    {
      "epoch": 0.7097361237488626,
      "grad_norm": 2.6933817863464355,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.8028,
      "step": 195
    },
    {
      "epoch": 0.7133757961783439,
      "grad_norm": 3.2947006225585938,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.1877,
      "step": 196
    },
    {
      "epoch": 0.7170154686078253,
      "grad_norm": 3.6697769165039062,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.9777,
      "step": 197
    },
    {
      "epoch": 0.7206551410373067,
      "grad_norm": 3.170090913772583,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.0089,
      "step": 198
    },
    {
      "epoch": 0.724294813466788,
      "grad_norm": 4.24803352355957,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.5854,
      "step": 199
    },
    {
      "epoch": 0.7279344858962693,
      "grad_norm": 4.326064109802246,
      "learning_rate": 0.0,
      "loss": 2.2922,
      "step": 200
    },
    {
      "epoch": 0.7279344858962693,
      "eval_loss": 0.5490885972976685,
      "eval_runtime": 47.5809,
      "eval_samples_per_second": 9.731,
      "eval_steps_per_second": 2.438,
      "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": 5.09187566075904e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}