{
  "best_metric": 0.350641667842865,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 2.452599388379205,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012232415902140673,
      "grad_norm": 0.49203047156333923,
      "learning_rate": 1e-05,
      "loss": 0.618,
      "step": 1
    },
    {
      "epoch": 0.012232415902140673,
      "eval_loss": 0.8547034859657288,
      "eval_runtime": 10.6767,
      "eval_samples_per_second": 12.925,
      "eval_steps_per_second": 3.278,
      "step": 1
    },
    {
      "epoch": 0.024464831804281346,
      "grad_norm": 0.7335397005081177,
      "learning_rate": 2e-05,
      "loss": 0.7154,
      "step": 2
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 0.842539370059967,
      "learning_rate": 3e-05,
      "loss": 0.6231,
      "step": 3
    },
    {
      "epoch": 0.04892966360856269,
      "grad_norm": 0.721678614616394,
      "learning_rate": 4e-05,
      "loss": 0.7223,
      "step": 4
    },
    {
      "epoch": 0.06116207951070336,
      "grad_norm": 0.6813793182373047,
      "learning_rate": 5e-05,
      "loss": 0.6558,
      "step": 5
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 0.6883832216262817,
      "learning_rate": 6e-05,
      "loss": 0.6508,
      "step": 6
    },
    {
      "epoch": 0.0856269113149847,
      "grad_norm": 0.7799242734909058,
      "learning_rate": 7e-05,
      "loss": 0.5509,
      "step": 7
    },
    {
      "epoch": 0.09785932721712538,
      "grad_norm": 0.7779672145843506,
      "learning_rate": 8e-05,
      "loss": 0.6111,
      "step": 8
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.6614462733268738,
      "learning_rate": 9e-05,
      "loss": 0.5749,
      "step": 9
    },
    {
      "epoch": 0.12232415902140673,
      "grad_norm": 0.6100348830223083,
      "learning_rate": 0.0001,
      "loss": 0.5295,
      "step": 10
    },
    {
      "epoch": 0.1345565749235474,
      "grad_norm": 0.5878755450248718,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.4425,
      "step": 11
    },
    {
      "epoch": 0.14678899082568808,
      "grad_norm": 0.5511037111282349,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.4425,
      "step": 12
    },
    {
      "epoch": 0.15902140672782875,
      "grad_norm": 0.6117210984230042,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.3974,
      "step": 13
    },
    {
      "epoch": 0.1712538226299694,
      "grad_norm": 0.5576888918876648,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.3981,
      "step": 14
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 0.5544164180755615,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.4205,
      "step": 15
    },
    {
      "epoch": 0.19571865443425077,
      "grad_norm": 0.48833444714546204,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.4293,
      "step": 16
    },
    {
      "epoch": 0.20795107033639143,
      "grad_norm": 0.5300983786582947,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.4389,
      "step": 17
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.4910358190536499,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.3726,
      "step": 18
    },
    {
      "epoch": 0.2324159021406728,
      "grad_norm": 0.536125123500824,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.3854,
      "step": 19
    },
    {
      "epoch": 0.24464831804281345,
      "grad_norm": 0.6423019170761108,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.3626,
      "step": 20
    },
    {
      "epoch": 0.25688073394495414,
      "grad_norm": 0.3751331865787506,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.5194,
      "step": 21
    },
    {
      "epoch": 0.2691131498470948,
      "grad_norm": 0.3893100619316101,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.5131,
      "step": 22
    },
    {
      "epoch": 0.28134556574923547,
      "grad_norm": 0.38985005021095276,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.4072,
      "step": 23
    },
    {
      "epoch": 0.29357798165137616,
      "grad_norm": 0.36344438791275024,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.4463,
      "step": 24
    },
    {
      "epoch": 0.3058103975535168,
      "grad_norm": 0.34405001997947693,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.5112,
      "step": 25
    },
    {
      "epoch": 0.3180428134556575,
      "grad_norm": 0.37133195996284485,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.4333,
      "step": 26
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.3859463334083557,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.395,
      "step": 27
    },
    {
      "epoch": 0.3425076452599388,
      "grad_norm": 0.39786985516548157,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.4617,
      "step": 28
    },
    {
      "epoch": 0.3547400611620795,
      "grad_norm": 0.35098281502723694,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.3966,
      "step": 29
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 0.3567488491535187,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.4175,
      "step": 30
    },
    {
      "epoch": 0.37920489296636084,
      "grad_norm": 0.354366660118103,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.395,
      "step": 31
    },
    {
      "epoch": 0.39143730886850153,
      "grad_norm": 0.36559244990348816,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.3705,
      "step": 32
    },
    {
      "epoch": 0.4036697247706422,
      "grad_norm": 0.39787328243255615,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.3671,
      "step": 33
    },
    {
      "epoch": 0.41590214067278286,
      "grad_norm": 0.4002894461154938,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.3386,
      "step": 34
    },
    {
      "epoch": 0.42813455657492355,
      "grad_norm": 0.38766971230506897,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.3298,
      "step": 35
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.3920033276081085,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.3321,
      "step": 36
    },
    {
      "epoch": 0.4525993883792049,
      "grad_norm": 0.44283950328826904,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.3497,
      "step": 37
    },
    {
      "epoch": 0.4648318042813456,
      "grad_norm": 0.3867936134338379,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.3333,
      "step": 38
    },
    {
      "epoch": 0.47706422018348627,
      "grad_norm": 0.3656964600086212,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.2946,
      "step": 39
    },
    {
      "epoch": 0.4892966360856269,
      "grad_norm": 0.479472815990448,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.3368,
      "step": 40
    },
    {
      "epoch": 0.5015290519877675,
      "grad_norm": 0.32764238119125366,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3651,
      "step": 41
    },
    {
      "epoch": 0.5137614678899083,
      "grad_norm": 0.3143838346004486,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.3543,
      "step": 42
    },
    {
      "epoch": 0.5259938837920489,
      "grad_norm": 0.32847076654434204,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.3966,
      "step": 43
    },
    {
      "epoch": 0.5382262996941896,
      "grad_norm": 0.3260630667209625,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4229,
      "step": 44
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.29270943999290466,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.3809,
      "step": 45
    },
    {
      "epoch": 0.5626911314984709,
      "grad_norm": 0.31629636883735657,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.3959,
      "step": 46
    },
    {
      "epoch": 0.5749235474006116,
      "grad_norm": 0.31145551800727844,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.3587,
      "step": 47
    },
    {
      "epoch": 0.5871559633027523,
      "grad_norm": 0.3356291651725769,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.3907,
      "step": 48
    },
    {
      "epoch": 0.599388379204893,
      "grad_norm": 0.3308155834674835,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.4293,
      "step": 49
    },
    {
      "epoch": 0.6116207951070336,
      "grad_norm": 0.3207649290561676,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.3904,
      "step": 50
    },
    {
      "epoch": 0.6116207951070336,
      "eval_loss": 0.3694062829017639,
      "eval_runtime": 10.8581,
      "eval_samples_per_second": 12.709,
      "eval_steps_per_second": 3.223,
      "step": 50
    },
    {
      "epoch": 0.6238532110091743,
      "grad_norm": 0.30467653274536133,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.3054,
      "step": 51
    },
    {
      "epoch": 0.636085626911315,
      "grad_norm": 0.35686832666397095,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.3896,
      "step": 52
    },
    {
      "epoch": 0.6483180428134556,
      "grad_norm": 0.41116029024124146,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.3654,
      "step": 53
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.3390863537788391,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.3115,
      "step": 54
    },
    {
      "epoch": 0.672782874617737,
      "grad_norm": 0.3317573368549347,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.3011,
      "step": 55
    },
    {
      "epoch": 0.6850152905198776,
      "grad_norm": 0.35453498363494873,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3041,
      "step": 56
    },
    {
      "epoch": 0.6972477064220184,
      "grad_norm": 0.4054669737815857,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3466,
      "step": 57
    },
    {
      "epoch": 0.709480122324159,
      "grad_norm": 0.345628559589386,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.2649,
      "step": 58
    },
    {
      "epoch": 0.7217125382262997,
      "grad_norm": 0.4063998758792877,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.306,
      "step": 59
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 0.4322091341018677,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.2841,
      "step": 60
    },
    {
      "epoch": 0.746177370030581,
      "grad_norm": 0.2798269987106323,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.4599,
      "step": 61
    },
    {
      "epoch": 0.7584097859327217,
      "grad_norm": 0.28430086374282837,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.3556,
      "step": 62
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.2860104441642761,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.3443,
      "step": 63
    },
    {
      "epoch": 0.7828746177370031,
      "grad_norm": 0.2921513020992279,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.3482,
      "step": 64
    },
    {
      "epoch": 0.7951070336391437,
      "grad_norm": 0.2988837957382202,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.3915,
      "step": 65
    },
    {
      "epoch": 0.8073394495412844,
      "grad_norm": 0.2925814688205719,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.3744,
      "step": 66
    },
    {
      "epoch": 0.8195718654434251,
      "grad_norm": 0.3107154369354248,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.3613,
      "step": 67
    },
    {
      "epoch": 0.8318042813455657,
      "grad_norm": 0.3237391412258148,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3859,
      "step": 68
    },
    {
      "epoch": 0.8440366972477065,
      "grad_norm": 0.29954588413238525,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3692,
      "step": 69
    },
    {
      "epoch": 0.8562691131498471,
      "grad_norm": 0.3180379867553711,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3455,
      "step": 70
    },
    {
      "epoch": 0.8685015290519877,
      "grad_norm": 0.3111006021499634,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3319,
      "step": 71
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.3059518337249756,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3061,
      "step": 72
    },
    {
      "epoch": 0.8929663608562691,
      "grad_norm": 0.3422292172908783,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.3409,
      "step": 73
    },
    {
      "epoch": 0.9051987767584098,
      "grad_norm": 0.31924888491630554,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.2662,
      "step": 74
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 0.3243701159954071,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.2989,
      "step": 75
    },
    {
      "epoch": 0.9296636085626911,
      "grad_norm": 0.32233110070228577,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.2606,
      "step": 76
    },
    {
      "epoch": 0.9418960244648318,
      "grad_norm": 0.3271600604057312,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.2594,
      "step": 77
    },
    {
      "epoch": 0.9541284403669725,
      "grad_norm": 0.3703632652759552,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.2993,
      "step": 78
    },
    {
      "epoch": 0.9663608562691132,
      "grad_norm": 0.3769654929637909,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.2884,
      "step": 79
    },
    {
      "epoch": 0.9785932721712538,
      "grad_norm": 0.47278931736946106,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3175,
      "step": 80
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.3103530704975128,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3944,
      "step": 81
    },
    {
      "epoch": 1.0061162079510704,
      "grad_norm": 0.5222724080085754,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.5874,
      "step": 82
    },
    {
      "epoch": 1.018348623853211,
      "grad_norm": 0.2140042781829834,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.2725,
      "step": 83
    },
    {
      "epoch": 1.0305810397553516,
      "grad_norm": 0.25839629769325256,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.305,
      "step": 84
    },
    {
      "epoch": 1.0428134556574924,
      "grad_norm": 0.2557167708873749,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3269,
      "step": 85
    },
    {
      "epoch": 1.0550458715596331,
      "grad_norm": 0.24326728284358978,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.3201,
      "step": 86
    },
    {
      "epoch": 1.0672782874617737,
      "grad_norm": 0.23968897759914398,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.2505,
      "step": 87
    },
    {
      "epoch": 1.0795107033639144,
      "grad_norm": 0.27097949385643005,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.2604,
      "step": 88
    },
    {
      "epoch": 1.091743119266055,
      "grad_norm": 0.2654765844345093,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.2816,
      "step": 89
    },
    {
      "epoch": 1.1039755351681957,
      "grad_norm": 0.2873367965221405,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.3143,
      "step": 90
    },
    {
      "epoch": 1.1162079510703364,
      "grad_norm": 0.3078736364841461,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3264,
      "step": 91
    },
    {
      "epoch": 1.1284403669724772,
      "grad_norm": 0.3022758662700653,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.2882,
      "step": 92
    },
    {
      "epoch": 1.1406727828746177,
      "grad_norm": 0.3040107786655426,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.2492,
      "step": 93
    },
    {
      "epoch": 1.1529051987767585,
      "grad_norm": 0.3144991397857666,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.258,
      "step": 94
    },
    {
      "epoch": 1.165137614678899,
      "grad_norm": 0.3165133595466614,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.2217,
      "step": 95
    },
    {
      "epoch": 1.1773700305810397,
      "grad_norm": 0.36050423979759216,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.2434,
      "step": 96
    },
    {
      "epoch": 1.1896024464831805,
      "grad_norm": 0.35134977102279663,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2305,
      "step": 97
    },
    {
      "epoch": 1.2018348623853212,
      "grad_norm": 0.319259911775589,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.1912,
      "step": 98
    },
    {
      "epoch": 1.2140672782874617,
      "grad_norm": 0.4008598327636719,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.2208,
      "step": 99
    },
    {
      "epoch": 1.2262996941896025,
      "grad_norm": 0.38668161630630493,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.1979,
      "step": 100
    },
    {
      "epoch": 1.2262996941896025,
      "eval_loss": 0.368563711643219,
      "eval_runtime": 10.8465,
      "eval_samples_per_second": 12.723,
      "eval_steps_per_second": 3.227,
      "step": 100
    },
    {
      "epoch": 1.238532110091743,
      "grad_norm": 0.38040223717689514,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.1722,
      "step": 101
    },
    {
      "epoch": 1.2507645259938838,
      "grad_norm": 0.4215383231639862,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.3742,
      "step": 102
    },
    {
      "epoch": 1.2629969418960245,
      "grad_norm": 0.4104975461959839,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.335,
      "step": 103
    },
    {
      "epoch": 1.2752293577981653,
      "grad_norm": 0.4201478064060211,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3549,
      "step": 104
    },
    {
      "epoch": 1.2874617737003058,
      "grad_norm": 0.3632720708847046,
      "learning_rate": 5e-05,
      "loss": 0.3317,
      "step": 105
    },
    {
      "epoch": 1.2996941896024465,
      "grad_norm": 0.3862139880657196,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3799,
      "step": 106
    },
    {
      "epoch": 1.311926605504587,
      "grad_norm": 0.3020176887512207,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.2425,
      "step": 107
    },
    {
      "epoch": 1.3241590214067278,
      "grad_norm": 0.328985333442688,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.3072,
      "step": 108
    },
    {
      "epoch": 1.3363914373088686,
      "grad_norm": 0.3140556812286377,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.299,
      "step": 109
    },
    {
      "epoch": 1.3486238532110093,
      "grad_norm": 0.3671078383922577,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3414,
      "step": 110
    },
    {
      "epoch": 1.3608562691131498,
      "grad_norm": 0.31250542402267456,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.3184,
      "step": 111
    },
    {
      "epoch": 1.3730886850152906,
      "grad_norm": 0.32229185104370117,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.2692,
      "step": 112
    },
    {
      "epoch": 1.385321100917431,
      "grad_norm": 0.29910027980804443,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.2388,
      "step": 113
    },
    {
      "epoch": 1.3975535168195719,
      "grad_norm": 0.3154756426811218,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.2764,
      "step": 114
    },
    {
      "epoch": 1.4097859327217126,
      "grad_norm": 0.330838143825531,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2738,
      "step": 115
    },
    {
      "epoch": 1.4220183486238533,
      "grad_norm": 0.3221725821495056,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.204,
      "step": 116
    },
    {
      "epoch": 1.4342507645259939,
      "grad_norm": 0.3018253445625305,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.2169,
      "step": 117
    },
    {
      "epoch": 1.4464831804281346,
      "grad_norm": 0.31192296743392944,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.2285,
      "step": 118
    },
    {
      "epoch": 1.4587155963302751,
      "grad_norm": 0.3444978892803192,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.2291,
      "step": 119
    },
    {
      "epoch": 1.470948012232416,
      "grad_norm": 0.3290013074874878,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2035,
      "step": 120
    },
    {
      "epoch": 1.4831804281345566,
      "grad_norm": 0.3425545394420624,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1777,
      "step": 121
    },
    {
      "epoch": 1.4954128440366974,
      "grad_norm": 0.3039463460445404,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3634,
      "step": 122
    },
    {
      "epoch": 1.507645259938838,
      "grad_norm": 0.23458532989025116,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3163,
      "step": 123
    },
    {
      "epoch": 1.5198776758409784,
      "grad_norm": 0.30795058608055115,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3582,
      "step": 124
    },
    {
      "epoch": 1.5321100917431192,
      "grad_norm": 0.3516447842121124,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3484,
      "step": 125
    },
    {
      "epoch": 1.54434250764526,
      "grad_norm": 0.3327048122882843,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3578,
      "step": 126
    },
    {
      "epoch": 1.5565749235474007,
      "grad_norm": 0.33929526805877686,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3533,
      "step": 127
    },
    {
      "epoch": 1.5688073394495414,
      "grad_norm": 0.2988111972808838,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.2624,
      "step": 128
    },
    {
      "epoch": 1.581039755351682,
      "grad_norm": 0.3317740857601166,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.2714,
      "step": 129
    },
    {
      "epoch": 1.5932721712538225,
      "grad_norm": 0.38439252972602844,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3806,
      "step": 130
    },
    {
      "epoch": 1.6055045871559632,
      "grad_norm": 0.3128364086151123,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2617,
      "step": 131
    },
    {
      "epoch": 1.617737003058104,
      "grad_norm": 0.38125112652778625,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.2762,
      "step": 132
    },
    {
      "epoch": 1.6299694189602447,
      "grad_norm": 0.3580372631549835,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.2729,
      "step": 133
    },
    {
      "epoch": 1.6422018348623855,
      "grad_norm": 0.342160165309906,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.2579,
      "step": 134
    },
    {
      "epoch": 1.654434250764526,
      "grad_norm": 0.32111403346061707,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.2332,
      "step": 135
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.3267934322357178,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2047,
      "step": 136
    },
    {
      "epoch": 1.6788990825688073,
      "grad_norm": 0.33800366520881653,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.2091,
      "step": 137
    },
    {
      "epoch": 1.691131498470948,
      "grad_norm": 0.35488492250442505,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.214,
      "step": 138
    },
    {
      "epoch": 1.7033639143730888,
      "grad_norm": 0.33737078309059143,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.1861,
      "step": 139
    },
    {
      "epoch": 1.7155963302752295,
      "grad_norm": 0.33756589889526367,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.1827,
      "step": 140
    },
    {
      "epoch": 1.72782874617737,
      "grad_norm": 0.3880833685398102,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.1774,
      "step": 141
    },
    {
      "epoch": 1.7400611620795106,
      "grad_norm": 0.3036118149757385,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2856,
      "step": 142
    },
    {
      "epoch": 1.7522935779816513,
      "grad_norm": 0.254378080368042,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3135,
      "step": 143
    },
    {
      "epoch": 1.764525993883792,
      "grad_norm": 0.3575253486633301,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.272,
      "step": 144
    },
    {
      "epoch": 1.7767584097859328,
      "grad_norm": 0.31142857670783997,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2814,
      "step": 145
    },
    {
      "epoch": 1.7889908256880735,
      "grad_norm": 0.3320716321468353,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.3465,
      "step": 146
    },
    {
      "epoch": 1.801223241590214,
      "grad_norm": 0.3244940936565399,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3303,
      "step": 147
    },
    {
      "epoch": 1.8134556574923546,
      "grad_norm": 0.32819393277168274,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.3307,
      "step": 148
    },
    {
      "epoch": 1.8256880733944953,
      "grad_norm": 0.35222482681274414,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3208,
      "step": 149
    },
    {
      "epoch": 1.837920489296636,
      "grad_norm": 0.3394937813282013,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.3134,
      "step": 150
    },
    {
      "epoch": 1.837920489296636,
      "eval_loss": 0.350641667842865,
      "eval_runtime": 10.8646,
      "eval_samples_per_second": 12.702,
      "eval_steps_per_second": 3.221,
      "step": 150
    },
    {
      "epoch": 1.8501529051987768,
      "grad_norm": 0.34158000349998474,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2468,
      "step": 151
    },
    {
      "epoch": 1.8623853211009176,
      "grad_norm": 0.3191375732421875,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.232,
      "step": 152
    },
    {
      "epoch": 1.8746177370030581,
      "grad_norm": 0.32770881056785583,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.233,
      "step": 153
    },
    {
      "epoch": 1.8868501529051986,
      "grad_norm": 0.33779266476631165,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.2322,
      "step": 154
    },
    {
      "epoch": 1.8990825688073394,
      "grad_norm": 0.3396666944026947,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.2165,
      "step": 155
    },
    {
      "epoch": 1.9113149847094801,
      "grad_norm": 0.3281339108943939,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2128,
      "step": 156
    },
    {
      "epoch": 1.9235474006116209,
      "grad_norm": 0.3483438193798065,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.2169,
      "step": 157
    },
    {
      "epoch": 1.9357798165137616,
      "grad_norm": 0.3689464032649994,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2426,
      "step": 158
    },
    {
      "epoch": 1.9480122324159022,
      "grad_norm": 0.36408838629722595,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.2115,
      "step": 159
    },
    {
      "epoch": 1.9602446483180427,
      "grad_norm": 0.42640426754951477,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.25,
      "step": 160
    },
    {
      "epoch": 1.9724770642201834,
      "grad_norm": 0.41972142457962036,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.2345,
      "step": 161
    },
    {
      "epoch": 1.9847094801223242,
      "grad_norm": 0.3567162752151489,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2214,
      "step": 162
    },
    {
      "epoch": 1.996941896024465,
      "grad_norm": 0.5055474042892456,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2929,
      "step": 163
    },
    {
      "epoch": 2.0122324159021407,
      "grad_norm": 0.22303302586078644,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2807,
      "step": 164
    },
    {
      "epoch": 2.0244648318042815,
      "grad_norm": 0.26136767864227295,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2843,
      "step": 165
    },
    {
      "epoch": 2.036697247706422,
      "grad_norm": 0.2842232882976532,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.2911,
      "step": 166
    },
    {
      "epoch": 2.0489296636085625,
      "grad_norm": 0.2708796262741089,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.2626,
      "step": 167
    },
    {
      "epoch": 2.0611620795107033,
      "grad_norm": 0.2718449532985687,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2543,
      "step": 168
    },
    {
      "epoch": 2.073394495412844,
      "grad_norm": 0.29201751947402954,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2759,
      "step": 169
    },
    {
      "epoch": 2.085626911314985,
      "grad_norm": 0.29264751076698303,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.2276,
      "step": 170
    },
    {
      "epoch": 2.0978593272171255,
      "grad_norm": 0.3177122175693512,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.2929,
      "step": 171
    },
    {
      "epoch": 2.1100917431192663,
      "grad_norm": 0.2933153212070465,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.2456,
      "step": 172
    },
    {
      "epoch": 2.1223241590214066,
      "grad_norm": 0.2777900695800781,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.1786,
      "step": 173
    },
    {
      "epoch": 2.1345565749235473,
      "grad_norm": 0.2833649814128876,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.1785,
      "step": 174
    },
    {
      "epoch": 2.146788990825688,
      "grad_norm": 0.30753934383392334,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.1887,
      "step": 175
    },
    {
      "epoch": 2.159021406727829,
      "grad_norm": 0.3037162125110626,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1883,
      "step": 176
    },
    {
      "epoch": 2.1712538226299696,
      "grad_norm": 0.2993669807910919,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1701,
      "step": 177
    },
    {
      "epoch": 2.18348623853211,
      "grad_norm": 0.3263915479183197,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.1783,
      "step": 178
    },
    {
      "epoch": 2.1957186544342506,
      "grad_norm": 0.3346138596534729,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.1764,
      "step": 179
    },
    {
      "epoch": 2.2079510703363914,
      "grad_norm": 0.6663686633110046,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.2113,
      "step": 180
    },
    {
      "epoch": 2.220183486238532,
      "grad_norm": 0.3060767948627472,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1483,
      "step": 181
    },
    {
      "epoch": 2.232415902140673,
      "grad_norm": 0.3507230579853058,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.1656,
      "step": 182
    },
    {
      "epoch": 2.2446483180428136,
      "grad_norm": 0.3343973457813263,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.1368,
      "step": 183
    },
    {
      "epoch": 2.2568807339449544,
      "grad_norm": 0.2353735864162445,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.2903,
      "step": 184
    },
    {
      "epoch": 2.2691131498470947,
      "grad_norm": 0.25053614377975464,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.2484,
      "step": 185
    },
    {
      "epoch": 2.2813455657492354,
      "grad_norm": 0.27156978845596313,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2264,
      "step": 186
    },
    {
      "epoch": 2.293577981651376,
      "grad_norm": 0.3115900158882141,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.3016,
      "step": 187
    },
    {
      "epoch": 2.305810397553517,
      "grad_norm": 0.2937740683555603,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2485,
      "step": 188
    },
    {
      "epoch": 2.3180428134556577,
      "grad_norm": 0.3086647689342499,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2503,
      "step": 189
    },
    {
      "epoch": 2.330275229357798,
      "grad_norm": 0.26695045828819275,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.1979,
      "step": 190
    },
    {
      "epoch": 2.3425076452599387,
      "grad_norm": 0.3203602135181427,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2646,
      "step": 191
    },
    {
      "epoch": 2.3547400611620795,
      "grad_norm": 0.3068872094154358,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.2138,
      "step": 192
    },
    {
      "epoch": 2.36697247706422,
      "grad_norm": 0.3405860662460327,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.258,
      "step": 193
    },
    {
      "epoch": 2.379204892966361,
      "grad_norm": 0.3421233296394348,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2252,
      "step": 194
    },
    {
      "epoch": 2.3914373088685017,
      "grad_norm": 0.319794625043869,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1911,
      "step": 195
    },
    {
      "epoch": 2.4036697247706424,
      "grad_norm": 0.3064125180244446,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.1772,
      "step": 196
    },
    {
      "epoch": 2.4159021406727827,
      "grad_norm": 0.3086625337600708,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.1514,
      "step": 197
    },
    {
      "epoch": 2.4281345565749235,
      "grad_norm": 0.3053826093673706,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1626,
      "step": 198
    },
    {
      "epoch": 2.4403669724770642,
      "grad_norm": 0.32307735085487366,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.177,
      "step": 199
    },
    {
      "epoch": 2.452599388379205,
      "grad_norm": 0.31589850783348083,
      "learning_rate": 0.0,
      "loss": 0.1692,
      "step": 200
    },
    {
      "epoch": 2.452599388379205,
      "eval_loss": 0.35414132475852966,
      "eval_runtime": 10.8556,
      "eval_samples_per_second": 12.712,
      "eval_steps_per_second": 3.224,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.0547399969603584e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}