{
  "best_metric": 1.6108735799789429,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.1157601115760112,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005578800557880056,
      "grad_norm": 2.9317691326141357,
      "learning_rate": 1e-05,
      "loss": 2.5854,
      "step": 1
    },
    {
      "epoch": 0.005578800557880056,
      "eval_loss": 2.741358995437622,
      "eval_runtime": 22.5485,
      "eval_samples_per_second": 13.393,
      "eval_steps_per_second": 3.371,
      "step": 1
    },
    {
      "epoch": 0.011157601115760111,
      "grad_norm": 3.3476312160491943,
      "learning_rate": 2e-05,
      "loss": 2.8058,
      "step": 2
    },
    {
      "epoch": 0.016736401673640166,
      "grad_norm": 3.5412299633026123,
      "learning_rate": 3e-05,
      "loss": 2.8509,
      "step": 3
    },
    {
      "epoch": 0.022315202231520222,
      "grad_norm": 3.574714422225952,
      "learning_rate": 4e-05,
      "loss": 2.7585,
      "step": 4
    },
    {
      "epoch": 0.02789400278940028,
      "grad_norm": 2.883089780807495,
      "learning_rate": 5e-05,
      "loss": 2.4545,
      "step": 5
    },
    {
      "epoch": 0.03347280334728033,
      "grad_norm": 2.724461555480957,
      "learning_rate": 6e-05,
      "loss": 2.3378,
      "step": 6
    },
    {
      "epoch": 0.03905160390516039,
      "grad_norm": 2.381211042404175,
      "learning_rate": 7e-05,
      "loss": 2.1702,
      "step": 7
    },
    {
      "epoch": 0.044630404463040445,
      "grad_norm": 2.1633474826812744,
      "learning_rate": 8e-05,
      "loss": 1.9864,
      "step": 8
    },
    {
      "epoch": 0.0502092050209205,
      "grad_norm": 2.269535779953003,
      "learning_rate": 9e-05,
      "loss": 2.3283,
      "step": 9
    },
    {
      "epoch": 0.05578800557880056,
      "grad_norm": 2.380795478820801,
      "learning_rate": 0.0001,
      "loss": 1.9662,
      "step": 10
    },
    {
      "epoch": 0.061366806136680614,
      "grad_norm": 1.9706579446792603,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.9527,
      "step": 11
    },
    {
      "epoch": 0.06694560669456066,
      "grad_norm": 1.7410569190979004,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.744,
      "step": 12
    },
    {
      "epoch": 0.07252440725244072,
      "grad_norm": 1.7006655931472778,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.8935,
      "step": 13
    },
    {
      "epoch": 0.07810320781032078,
      "grad_norm": 1.725269079208374,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.8607,
      "step": 14
    },
    {
      "epoch": 0.08368200836820083,
      "grad_norm": 1.9727450609207153,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.8936,
      "step": 15
    },
    {
      "epoch": 0.08926080892608089,
      "grad_norm": 1.719459056854248,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.8301,
      "step": 16
    },
    {
      "epoch": 0.09483960948396095,
      "grad_norm": 1.3618836402893066,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.7115,
      "step": 17
    },
    {
      "epoch": 0.100418410041841,
      "grad_norm": 1.5107271671295166,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.5628,
      "step": 18
    },
    {
      "epoch": 0.10599721059972106,
      "grad_norm": 1.3186451196670532,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6618,
      "step": 19
    },
    {
      "epoch": 0.11157601115760112,
      "grad_norm": 1.2762210369110107,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.8504,
      "step": 20
    },
    {
      "epoch": 0.11715481171548117,
      "grad_norm": 1.3136497735977173,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.8117,
      "step": 21
    },
    {
      "epoch": 0.12273361227336123,
      "grad_norm": 1.2667301893234253,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.6778,
      "step": 22
    },
    {
      "epoch": 0.12831241283124128,
      "grad_norm": 1.3415364027023315,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.8092,
      "step": 23
    },
    {
      "epoch": 0.13389121338912133,
      "grad_norm": 1.204180359840393,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.5573,
      "step": 24
    },
    {
      "epoch": 0.1394700139470014,
      "grad_norm": 1.2918239831924438,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6006,
      "step": 25
    },
    {
      "epoch": 0.14504881450488144,
      "grad_norm": 1.2955774068832397,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.6195,
      "step": 26
    },
    {
      "epoch": 0.1506276150627615,
      "grad_norm": 1.171953558921814,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5737,
      "step": 27
    },
    {
      "epoch": 0.15620641562064155,
      "grad_norm": 1.2768067121505737,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.6768,
      "step": 28
    },
    {
      "epoch": 0.16178521617852162,
      "grad_norm": 1.1846047639846802,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.6967,
      "step": 29
    },
    {
      "epoch": 0.16736401673640167,
      "grad_norm": 1.2360256910324097,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.6948,
      "step": 30
    },
    {
      "epoch": 0.17294281729428174,
      "grad_norm": 1.273258924484253,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.7421,
      "step": 31
    },
    {
      "epoch": 0.17852161785216178,
      "grad_norm": 1.2629722356796265,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.6709,
      "step": 32
    },
    {
      "epoch": 0.18410041841004185,
      "grad_norm": 1.22515869140625,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.6185,
      "step": 33
    },
    {
      "epoch": 0.1896792189679219,
      "grad_norm": 1.1701767444610596,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5548,
      "step": 34
    },
    {
      "epoch": 0.19525801952580196,
      "grad_norm": 1.0819822549819946,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.6301,
      "step": 35
    },
    {
      "epoch": 0.200836820083682,
      "grad_norm": 1.248599648475647,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3999,
      "step": 36
    },
    {
      "epoch": 0.20641562064156208,
      "grad_norm": 1.120344638824463,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.486,
      "step": 37
    },
    {
      "epoch": 0.21199442119944212,
      "grad_norm": 1.1335872411727905,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5043,
      "step": 38
    },
    {
      "epoch": 0.2175732217573222,
      "grad_norm": 1.2641167640686035,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6944,
      "step": 39
    },
    {
      "epoch": 0.22315202231520223,
      "grad_norm": 1.3135420083999634,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4997,
      "step": 40
    },
    {
      "epoch": 0.22873082287308227,
      "grad_norm": 1.128016710281372,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.6073,
      "step": 41
    },
    {
      "epoch": 0.23430962343096234,
      "grad_norm": 1.0898466110229492,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5979,
      "step": 42
    },
    {
      "epoch": 0.2398884239888424,
      "grad_norm": 1.1444522142410278,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.6797,
      "step": 43
    },
    {
      "epoch": 0.24546722454672246,
      "grad_norm": 1.1293357610702515,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.6039,
      "step": 44
    },
    {
      "epoch": 0.2510460251046025,
      "grad_norm": 1.0889806747436523,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.601,
      "step": 45
    },
    {
      "epoch": 0.25662482566248257,
      "grad_norm": 1.305039882659912,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.7114,
      "step": 46
    },
    {
      "epoch": 0.26220362622036264,
      "grad_norm": 1.0812164545059204,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.6837,
      "step": 47
    },
    {
      "epoch": 0.26778242677824265,
      "grad_norm": 0.9839636087417603,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.6442,
      "step": 48
    },
    {
      "epoch": 0.2733612273361227,
      "grad_norm": 1.173969030380249,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.9296,
      "step": 49
    },
    {
      "epoch": 0.2789400278940028,
      "grad_norm": 1.1880217790603638,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.6267,
      "step": 50
    },
    {
      "epoch": 0.2789400278940028,
      "eval_loss": 1.686349868774414,
      "eval_runtime": 22.9173,
      "eval_samples_per_second": 13.178,
      "eval_steps_per_second": 3.316,
      "step": 50
    },
    {
      "epoch": 0.28451882845188287,
      "grad_norm": 1.253607153892517,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.7952,
      "step": 51
    },
    {
      "epoch": 0.2900976290097629,
      "grad_norm": 1.140913724899292,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.6293,
      "step": 52
    },
    {
      "epoch": 0.29567642956764295,
      "grad_norm": 1.3402795791625977,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.9089,
      "step": 53
    },
    {
      "epoch": 0.301255230125523,
      "grad_norm": 1.0494638681411743,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.7221,
      "step": 54
    },
    {
      "epoch": 0.3068340306834031,
      "grad_norm": 1.1479544639587402,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5917,
      "step": 55
    },
    {
      "epoch": 0.3124128312412831,
      "grad_norm": 0.993493378162384,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5355,
      "step": 56
    },
    {
      "epoch": 0.3179916317991632,
      "grad_norm": 1.088383674621582,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.8137,
      "step": 57
    },
    {
      "epoch": 0.32357043235704325,
      "grad_norm": 1.0391969680786133,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.6123,
      "step": 58
    },
    {
      "epoch": 0.3291492329149233,
      "grad_norm": 1.0096733570098877,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.7531,
      "step": 59
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 0.9810142517089844,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7138,
      "step": 60
    },
    {
      "epoch": 0.3403068340306834,
      "grad_norm": 0.9760948419570923,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.7188,
      "step": 61
    },
    {
      "epoch": 0.3458856345885635,
      "grad_norm": 0.9632744193077087,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.6227,
      "step": 62
    },
    {
      "epoch": 0.3514644351464435,
      "grad_norm": 1.0082738399505615,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.6589,
      "step": 63
    },
    {
      "epoch": 0.35704323570432356,
      "grad_norm": 1.2210134267807007,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5379,
      "step": 64
    },
    {
      "epoch": 0.36262203626220363,
      "grad_norm": 1.0689101219177246,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6047,
      "step": 65
    },
    {
      "epoch": 0.3682008368200837,
      "grad_norm": 1.2459222078323364,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.8272,
      "step": 66
    },
    {
      "epoch": 0.3737796373779637,
      "grad_norm": 1.0078871250152588,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5564,
      "step": 67
    },
    {
      "epoch": 0.3793584379358438,
      "grad_norm": 1.0605857372283936,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.7475,
      "step": 68
    },
    {
      "epoch": 0.38493723849372385,
      "grad_norm": 1.03123140335083,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.7894,
      "step": 69
    },
    {
      "epoch": 0.3905160390516039,
      "grad_norm": 1.1406044960021973,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4748,
      "step": 70
    },
    {
      "epoch": 0.39609483960948394,
      "grad_norm": 0.9987372159957886,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.6649,
      "step": 71
    },
    {
      "epoch": 0.401673640167364,
      "grad_norm": 0.9522144198417664,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.5148,
      "step": 72
    },
    {
      "epoch": 0.4072524407252441,
      "grad_norm": 0.9634456634521484,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4295,
      "step": 73
    },
    {
      "epoch": 0.41283124128312415,
      "grad_norm": 1.0577412843704224,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.7197,
      "step": 74
    },
    {
      "epoch": 0.41841004184100417,
      "grad_norm": 0.991331934928894,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.603,
      "step": 75
    },
    {
      "epoch": 0.42398884239888424,
      "grad_norm": 1.1320452690124512,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6711,
      "step": 76
    },
    {
      "epoch": 0.4295676429567643,
      "grad_norm": 1.0363842248916626,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7343,
      "step": 77
    },
    {
      "epoch": 0.4351464435146444,
      "grad_norm": 1.0597922801971436,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.6004,
      "step": 78
    },
    {
      "epoch": 0.4407252440725244,
      "grad_norm": 1.006777286529541,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.668,
      "step": 79
    },
    {
      "epoch": 0.44630404463040446,
      "grad_norm": 1.0739244222640991,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.6266,
      "step": 80
    },
    {
      "epoch": 0.45188284518828453,
      "grad_norm": 1.0785304307937622,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.6618,
      "step": 81
    },
    {
      "epoch": 0.45746164574616455,
      "grad_norm": 0.9765123724937439,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.4684,
      "step": 82
    },
    {
      "epoch": 0.4630404463040446,
      "grad_norm": 1.0404711961746216,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.6295,
      "step": 83
    },
    {
      "epoch": 0.4686192468619247,
      "grad_norm": 1.0764896869659424,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4742,
      "step": 84
    },
    {
      "epoch": 0.47419804741980476,
      "grad_norm": 1.003179907798767,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.5496,
      "step": 85
    },
    {
      "epoch": 0.4797768479776848,
      "grad_norm": 1.0054682493209839,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.646,
      "step": 86
    },
    {
      "epoch": 0.48535564853556484,
      "grad_norm": 0.9638222455978394,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5559,
      "step": 87
    },
    {
      "epoch": 0.4909344490934449,
      "grad_norm": 1.1578853130340576,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.6149,
      "step": 88
    },
    {
      "epoch": 0.496513249651325,
      "grad_norm": 1.1789309978485107,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.8483,
      "step": 89
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 1.025229573249817,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.7108,
      "step": 90
    },
    {
      "epoch": 0.5076708507670851,
      "grad_norm": 1.013513207435608,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.723,
      "step": 91
    },
    {
      "epoch": 0.5132496513249651,
      "grad_norm": 0.9646743535995483,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.6867,
      "step": 92
    },
    {
      "epoch": 0.5188284518828452,
      "grad_norm": 0.9263551831245422,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.7071,
      "step": 93
    },
    {
      "epoch": 0.5244072524407253,
      "grad_norm": 1.0874505043029785,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.8197,
      "step": 94
    },
    {
      "epoch": 0.5299860529986054,
      "grad_norm": 0.9751068353652954,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.4683,
      "step": 95
    },
    {
      "epoch": 0.5355648535564853,
      "grad_norm": 0.9625138640403748,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.6623,
      "step": 96
    },
    {
      "epoch": 0.5411436541143654,
      "grad_norm": 0.9792675375938416,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.5511,
      "step": 97
    },
    {
      "epoch": 0.5467224546722455,
      "grad_norm": 1.0116002559661865,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.662,
      "step": 98
    },
    {
      "epoch": 0.5523012552301255,
      "grad_norm": 1.0003246068954468,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.6437,
      "step": 99
    },
    {
      "epoch": 0.5578800557880056,
      "grad_norm": 1.0759495496749878,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.7573,
      "step": 100
    },
    {
      "epoch": 0.5578800557880056,
      "eval_loss": 1.6395673751831055,
      "eval_runtime": 22.9159,
      "eval_samples_per_second": 13.179,
      "eval_steps_per_second": 3.316,
      "step": 100
    },
    {
      "epoch": 0.5634588563458857,
      "grad_norm": 0.9676824808120728,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.5524,
      "step": 101
    },
    {
      "epoch": 0.5690376569037657,
      "grad_norm": 1.0113121271133423,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.7142,
      "step": 102
    },
    {
      "epoch": 0.5746164574616457,
      "grad_norm": 1.0064094066619873,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5214,
      "step": 103
    },
    {
      "epoch": 0.5801952580195258,
      "grad_norm": 1.0235830545425415,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5698,
      "step": 104
    },
    {
      "epoch": 0.5857740585774058,
      "grad_norm": 1.1040574312210083,
      "learning_rate": 5e-05,
      "loss": 1.8354,
      "step": 105
    },
    {
      "epoch": 0.5913528591352859,
      "grad_norm": 0.9078890085220337,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.4956,
      "step": 106
    },
    {
      "epoch": 0.596931659693166,
      "grad_norm": 1.0370744466781616,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.6804,
      "step": 107
    },
    {
      "epoch": 0.602510460251046,
      "grad_norm": 0.848264753818512,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4458,
      "step": 108
    },
    {
      "epoch": 0.6080892608089261,
      "grad_norm": 0.9408484697341919,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4371,
      "step": 109
    },
    {
      "epoch": 0.6136680613668062,
      "grad_norm": 0.9886391162872314,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.6642,
      "step": 110
    },
    {
      "epoch": 0.6192468619246861,
      "grad_norm": 1.0459530353546143,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.7468,
      "step": 111
    },
    {
      "epoch": 0.6248256624825662,
      "grad_norm": 0.9325463771820068,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.5876,
      "step": 112
    },
    {
      "epoch": 0.6304044630404463,
      "grad_norm": 0.9819180965423584,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.6073,
      "step": 113
    },
    {
      "epoch": 0.6359832635983264,
      "grad_norm": 1.0222245454788208,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6752,
      "step": 114
    },
    {
      "epoch": 0.6415620641562064,
      "grad_norm": 0.9456518888473511,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.6138,
      "step": 115
    },
    {
      "epoch": 0.6471408647140865,
      "grad_norm": 0.9613638520240784,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5099,
      "step": 116
    },
    {
      "epoch": 0.6527196652719666,
      "grad_norm": 0.9034593105316162,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.4677,
      "step": 117
    },
    {
      "epoch": 0.6582984658298466,
      "grad_norm": 0.9895356297492981,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.596,
      "step": 118
    },
    {
      "epoch": 0.6638772663877266,
      "grad_norm": 1.0307013988494873,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7299,
      "step": 119
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 0.9757693409919739,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.5457,
      "step": 120
    },
    {
      "epoch": 0.6750348675034867,
      "grad_norm": 0.9991773962974548,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.702,
      "step": 121
    },
    {
      "epoch": 0.6806136680613668,
      "grad_norm": 0.9439085125923157,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.5062,
      "step": 122
    },
    {
      "epoch": 0.6861924686192469,
      "grad_norm": 0.9614840745925903,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.4849,
      "step": 123
    },
    {
      "epoch": 0.691771269177127,
      "grad_norm": 0.9206346273422241,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.5488,
      "step": 124
    },
    {
      "epoch": 0.697350069735007,
      "grad_norm": 0.9437282681465149,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.6928,
      "step": 125
    },
    {
      "epoch": 0.702928870292887,
      "grad_norm": 1.0243704319000244,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7343,
      "step": 126
    },
    {
      "epoch": 0.708507670850767,
      "grad_norm": 0.9115479588508606,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.4079,
      "step": 127
    },
    {
      "epoch": 0.7140864714086471,
      "grad_norm": 0.9688447713851929,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.5035,
      "step": 128
    },
    {
      "epoch": 0.7196652719665272,
      "grad_norm": 1.0060139894485474,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4799,
      "step": 129
    },
    {
      "epoch": 0.7252440725244073,
      "grad_norm": 1.0383399724960327,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.671,
      "step": 130
    },
    {
      "epoch": 0.7308228730822873,
      "grad_norm": 0.9058457016944885,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3957,
      "step": 131
    },
    {
      "epoch": 0.7364016736401674,
      "grad_norm": 1.0301634073257446,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.5478,
      "step": 132
    },
    {
      "epoch": 0.7419804741980475,
      "grad_norm": 0.979134202003479,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.7334,
      "step": 133
    },
    {
      "epoch": 0.7475592747559274,
      "grad_norm": 0.9294139742851257,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.6679,
      "step": 134
    },
    {
      "epoch": 0.7531380753138075,
      "grad_norm": 0.9550366997718811,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.6204,
      "step": 135
    },
    {
      "epoch": 0.7587168758716876,
      "grad_norm": 1.000625729560852,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.686,
      "step": 136
    },
    {
      "epoch": 0.7642956764295676,
      "grad_norm": 0.9747148752212524,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.6375,
      "step": 137
    },
    {
      "epoch": 0.7698744769874477,
      "grad_norm": 1.005576491355896,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.7568,
      "step": 138
    },
    {
      "epoch": 0.7754532775453278,
      "grad_norm": 1.0260334014892578,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.6025,
      "step": 139
    },
    {
      "epoch": 0.7810320781032078,
      "grad_norm": 0.8803253769874573,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.5848,
      "step": 140
    },
    {
      "epoch": 0.7866108786610879,
      "grad_norm": 0.9695632457733154,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.6456,
      "step": 141
    },
    {
      "epoch": 0.7921896792189679,
      "grad_norm": 1.0484915971755981,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.7941,
      "step": 142
    },
    {
      "epoch": 0.797768479776848,
      "grad_norm": 0.9246304035186768,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.6247,
      "step": 143
    },
    {
      "epoch": 0.803347280334728,
      "grad_norm": 0.8872515559196472,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.5287,
      "step": 144
    },
    {
      "epoch": 0.8089260808926081,
      "grad_norm": 1.077502965927124,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.5811,
      "step": 145
    },
    {
      "epoch": 0.8145048814504882,
      "grad_norm": 0.9123009443283081,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.6973,
      "step": 146
    },
    {
      "epoch": 0.8200836820083682,
      "grad_norm": 0.9766699075698853,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.714,
      "step": 147
    },
    {
      "epoch": 0.8256624825662483,
      "grad_norm": 0.9173923134803772,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.5935,
      "step": 148
    },
    {
      "epoch": 0.8312412831241283,
      "grad_norm": 1.121888518333435,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.8199,
      "step": 149
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 0.9458000659942627,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.6834,
      "step": 150
    },
    {
      "epoch": 0.8368200836820083,
      "eval_loss": 1.6211026906967163,
      "eval_runtime": 22.9007,
      "eval_samples_per_second": 13.187,
      "eval_steps_per_second": 3.319,
      "step": 150
    },
    {
      "epoch": 0.8423988842398884,
      "grad_norm": 1.0237611532211304,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.6682,
      "step": 151
    },
    {
      "epoch": 0.8479776847977685,
      "grad_norm": 0.9446094632148743,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.6208,
      "step": 152
    },
    {
      "epoch": 0.8535564853556485,
      "grad_norm": 0.9437018632888794,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.6434,
      "step": 153
    },
    {
      "epoch": 0.8591352859135286,
      "grad_norm": 0.8713282942771912,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.5361,
      "step": 154
    },
    {
      "epoch": 0.8647140864714087,
      "grad_norm": 0.9705963134765625,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.877,
      "step": 155
    },
    {
      "epoch": 0.8702928870292888,
      "grad_norm": 0.9434710144996643,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.558,
      "step": 156
    },
    {
      "epoch": 0.8758716875871687,
      "grad_norm": 1.0295372009277344,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.7604,
      "step": 157
    },
    {
      "epoch": 0.8814504881450488,
      "grad_norm": 0.9482672214508057,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.5489,
      "step": 158
    },
    {
      "epoch": 0.8870292887029289,
      "grad_norm": 0.8490006327629089,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.5928,
      "step": 159
    },
    {
      "epoch": 0.8926080892608089,
      "grad_norm": 0.9739740490913391,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.5674,
      "step": 160
    },
    {
      "epoch": 0.898186889818689,
      "grad_norm": 0.9436212778091431,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.6486,
      "step": 161
    },
    {
      "epoch": 0.9037656903765691,
      "grad_norm": 0.8999307155609131,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.6498,
      "step": 162
    },
    {
      "epoch": 0.9093444909344491,
      "grad_norm": 0.9686093926429749,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.4199,
      "step": 163
    },
    {
      "epoch": 0.9149232914923291,
      "grad_norm": 1.017491340637207,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.7666,
      "step": 164
    },
    {
      "epoch": 0.9205020920502092,
      "grad_norm": 0.9168091416358948,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.4488,
      "step": 165
    },
    {
      "epoch": 0.9260808926080892,
      "grad_norm": 0.9157407879829407,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.4939,
      "step": 166
    },
    {
      "epoch": 0.9316596931659693,
      "grad_norm": 0.8908804655075073,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.3794,
      "step": 167
    },
    {
      "epoch": 0.9372384937238494,
      "grad_norm": 0.9589452743530273,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5704,
      "step": 168
    },
    {
      "epoch": 0.9428172942817294,
      "grad_norm": 1.066652774810791,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.4166,
      "step": 169
    },
    {
      "epoch": 0.9483960948396095,
      "grad_norm": 1.095902681350708,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.7217,
      "step": 170
    },
    {
      "epoch": 0.9539748953974896,
      "grad_norm": 1.0956112146377563,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.5452,
      "step": 171
    },
    {
      "epoch": 0.9595536959553695,
      "grad_norm": 1.0670567750930786,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.6629,
      "step": 172
    },
    {
      "epoch": 0.9651324965132496,
      "grad_norm": 0.9397924542427063,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.3279,
      "step": 173
    },
    {
      "epoch": 0.9707112970711297,
      "grad_norm": 0.9698524475097656,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.5297,
      "step": 174
    },
    {
      "epoch": 0.9762900976290098,
      "grad_norm": 0.9127272367477417,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.5907,
      "step": 175
    },
    {
      "epoch": 0.9818688981868898,
      "grad_norm": 1.159216284751892,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.64,
      "step": 176
    },
    {
      "epoch": 0.9874476987447699,
      "grad_norm": 0.8653454184532166,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.5639,
      "step": 177
    },
    {
      "epoch": 0.99302649930265,
      "grad_norm": 0.9482781887054443,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.6305,
      "step": 178
    },
    {
      "epoch": 0.99860529986053,
      "grad_norm": 0.9488359689712524,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.494,
      "step": 179
    },
    {
      "epoch": 1.00418410041841,
      "grad_norm": 2.6767518520355225,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.6374,
      "step": 180
    },
    {
      "epoch": 1.00976290097629,
      "grad_norm": 0.9007354378700256,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.5079,
      "step": 181
    },
    {
      "epoch": 1.0153417015341701,
      "grad_norm": 0.8877829313278198,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.3648,
      "step": 182
    },
    {
      "epoch": 1.0209205020920502,
      "grad_norm": 0.9139792323112488,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.527,
      "step": 183
    },
    {
      "epoch": 1.0264993026499303,
      "grad_norm": 0.9384551048278809,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.4944,
      "step": 184
    },
    {
      "epoch": 1.0320781032078103,
      "grad_norm": 0.891731321811676,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.5088,
      "step": 185
    },
    {
      "epoch": 1.0376569037656904,
      "grad_norm": 0.9330790042877197,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.3797,
      "step": 186
    },
    {
      "epoch": 1.0432357043235705,
      "grad_norm": 0.9671591520309448,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.3486,
      "step": 187
    },
    {
      "epoch": 1.0488145048814506,
      "grad_norm": 0.9701920747756958,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.4411,
      "step": 188
    },
    {
      "epoch": 1.0543933054393306,
      "grad_norm": 0.9812690615653992,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.5248,
      "step": 189
    },
    {
      "epoch": 1.0599721059972107,
      "grad_norm": 1.0303298234939575,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.4578,
      "step": 190
    },
    {
      "epoch": 1.0655509065550905,
      "grad_norm": 0.9400258660316467,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.5059,
      "step": 191
    },
    {
      "epoch": 1.0711297071129706,
      "grad_norm": 0.8996063470840454,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.4761,
      "step": 192
    },
    {
      "epoch": 1.0767085076708507,
      "grad_norm": 0.9756540656089783,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.3862,
      "step": 193
    },
    {
      "epoch": 1.0822873082287308,
      "grad_norm": 0.912513256072998,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.3678,
      "step": 194
    },
    {
      "epoch": 1.0878661087866108,
      "grad_norm": 0.8697913885116577,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.3496,
      "step": 195
    },
    {
      "epoch": 1.093444909344491,
      "grad_norm": 0.9285879731178284,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.4305,
      "step": 196
    },
    {
      "epoch": 1.099023709902371,
      "grad_norm": 0.8984832763671875,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.3197,
      "step": 197
    },
    {
      "epoch": 1.104602510460251,
      "grad_norm": 0.8145501613616943,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.2339,
      "step": 198
    },
    {
      "epoch": 1.1101813110181311,
      "grad_norm": 0.9698240756988525,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.4373,
      "step": 199
    },
    {
      "epoch": 1.1157601115760112,
      "grad_norm": 0.9802901744842529,
      "learning_rate": 0.0,
      "loss": 1.3589,
      "step": 200
    },
    {
      "epoch": 1.1157601115760112,
      "eval_loss": 1.6108735799789429,
      "eval_runtime": 22.9177,
      "eval_samples_per_second": 13.178,
      "eval_steps_per_second": 3.316,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 3.0161358600064205e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}