{
  "best_metric": 0.1013597771525383,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.02297266253158741,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00011486331265793706,
      "grad_norm": 19.581317901611328,
      "learning_rate": 5e-06,
      "loss": 1.994,
      "step": 1
    },
    {
      "epoch": 0.00011486331265793706,
      "eval_loss": 0.5246481895446777,
      "eval_runtime": 1466.3245,
      "eval_samples_per_second": 10.0,
      "eval_steps_per_second": 5.0,
      "step": 1
    },
    {
      "epoch": 0.00022972662531587412,
      "grad_norm": 17.236621856689453,
      "learning_rate": 1e-05,
      "loss": 2.0861,
      "step": 2
    },
    {
      "epoch": 0.00034458993797381116,
      "grad_norm": 11.309477806091309,
      "learning_rate": 1.5e-05,
      "loss": 2.1709,
      "step": 3
    },
    {
      "epoch": 0.00045945325063174823,
      "grad_norm": 5.152942180633545,
      "learning_rate": 2e-05,
      "loss": 1.712,
      "step": 4
    },
    {
      "epoch": 0.0005743165632896852,
      "grad_norm": 4.796749114990234,
      "learning_rate": 2.5e-05,
      "loss": 1.881,
      "step": 5
    },
    {
      "epoch": 0.0006891798759476223,
      "grad_norm": 3.7207891941070557,
      "learning_rate": 3e-05,
      "loss": 1.899,
      "step": 6
    },
    {
      "epoch": 0.0008040431886055594,
      "grad_norm": 2.4096086025238037,
      "learning_rate": 3.5e-05,
      "loss": 1.8971,
      "step": 7
    },
    {
      "epoch": 0.0009189065012634965,
      "grad_norm": 4.16554594039917,
      "learning_rate": 4e-05,
      "loss": 1.8821,
      "step": 8
    },
    {
      "epoch": 0.0010337698139214334,
      "grad_norm": 2.0596752166748047,
      "learning_rate": 4.5e-05,
      "loss": 2.1518,
      "step": 9
    },
    {
      "epoch": 0.0011486331265793705,
      "grad_norm": 3.1254591941833496,
      "learning_rate": 5e-05,
      "loss": 1.8471,
      "step": 10
    },
    {
      "epoch": 0.0012634964392373076,
      "grad_norm": 1.5779181718826294,
      "learning_rate": 5.500000000000001e-05,
      "loss": 1.5703,
      "step": 11
    },
    {
      "epoch": 0.0013783597518952446,
      "grad_norm": 4.009050369262695,
      "learning_rate": 6e-05,
      "loss": 1.8472,
      "step": 12
    },
    {
      "epoch": 0.0014932230645531817,
      "grad_norm": 3.238170623779297,
      "learning_rate": 6.500000000000001e-05,
      "loss": 1.6825,
      "step": 13
    },
    {
      "epoch": 0.0016080863772111188,
      "grad_norm": 7.195780277252197,
      "learning_rate": 7e-05,
      "loss": 1.6415,
      "step": 14
    },
    {
      "epoch": 0.0017229496898690559,
      "grad_norm": 1.9010658264160156,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.5549,
      "step": 15
    },
    {
      "epoch": 0.001837813002526993,
      "grad_norm": 2.798220634460449,
      "learning_rate": 8e-05,
      "loss": 2.0429,
      "step": 16
    },
    {
      "epoch": 0.00195267631518493,
      "grad_norm": 2.5278286933898926,
      "learning_rate": 8.5e-05,
      "loss": 1.7963,
      "step": 17
    },
    {
      "epoch": 0.002067539627842867,
      "grad_norm": 1.599841833114624,
      "learning_rate": 9e-05,
      "loss": 1.5025,
      "step": 18
    },
    {
      "epoch": 0.002182402940500804,
      "grad_norm": 2.7871410846710205,
      "learning_rate": 9.5e-05,
      "loss": 1.5943,
      "step": 19
    },
    {
      "epoch": 0.002297266253158741,
      "grad_norm": 1.6238923072814941,
      "learning_rate": 0.0001,
      "loss": 1.5476,
      "step": 20
    },
    {
      "epoch": 0.002412129565816678,
      "grad_norm": 1.6949362754821777,
      "learning_rate": 9.999238475781957e-05,
      "loss": 1.789,
      "step": 21
    },
    {
      "epoch": 0.002526992878474615,
      "grad_norm": 1.7465438842773438,
      "learning_rate": 9.99695413509548e-05,
      "loss": 1.6521,
      "step": 22
    },
    {
      "epoch": 0.002641856191132552,
      "grad_norm": 2.695739507675171,
      "learning_rate": 9.99314767377287e-05,
      "loss": 1.6917,
      "step": 23
    },
    {
      "epoch": 0.0027567195037904893,
      "grad_norm": 2.790419101715088,
      "learning_rate": 9.987820251299122e-05,
      "loss": 1.9761,
      "step": 24
    },
    {
      "epoch": 0.0028715828164484264,
      "grad_norm": 4.338637828826904,
      "learning_rate": 9.980973490458728e-05,
      "loss": 1.7246,
      "step": 25
    },
    {
      "epoch": 0.0029864461291063634,
      "grad_norm": 2.29007625579834,
      "learning_rate": 9.972609476841367e-05,
      "loss": 1.4761,
      "step": 26
    },
    {
      "epoch": 0.0031013094417643005,
      "grad_norm": 1.7599883079528809,
      "learning_rate": 9.962730758206611e-05,
      "loss": 1.6964,
      "step": 27
    },
    {
      "epoch": 0.0032161727544222376,
      "grad_norm": 1.6065396070480347,
      "learning_rate": 9.951340343707852e-05,
      "loss": 1.5706,
      "step": 28
    },
    {
      "epoch": 0.0033310360670801746,
      "grad_norm": 2.0913872718811035,
      "learning_rate": 9.938441702975689e-05,
      "loss": 1.6397,
      "step": 29
    },
    {
      "epoch": 0.0034458993797381117,
      "grad_norm": 1.4033796787261963,
      "learning_rate": 9.924038765061042e-05,
      "loss": 1.493,
      "step": 30
    },
    {
      "epoch": 0.003560762692396049,
      "grad_norm": 1.7742940187454224,
      "learning_rate": 9.908135917238321e-05,
      "loss": 1.5057,
      "step": 31
    },
    {
      "epoch": 0.003675626005053986,
      "grad_norm": 2.7684073448181152,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.6441,
      "step": 32
    },
    {
      "epoch": 0.003790489317711923,
      "grad_norm": 2.2858612537384033,
      "learning_rate": 9.871850323926177e-05,
      "loss": 1.6578,
      "step": 33
    },
    {
      "epoch": 0.00390535263036986,
      "grad_norm": 1.846916913986206,
      "learning_rate": 9.851478631379982e-05,
      "loss": 1.3874,
      "step": 34
    },
    {
      "epoch": 0.004020215943027797,
      "grad_norm": 1.7534124851226807,
      "learning_rate": 9.829629131445342e-05,
      "loss": 1.6324,
      "step": 35
    },
    {
      "epoch": 0.004135079255685734,
      "grad_norm": 2.536592960357666,
      "learning_rate": 9.806308479691595e-05,
      "loss": 1.5962,
      "step": 36
    },
    {
      "epoch": 0.004249942568343671,
      "grad_norm": 2.0697085857391357,
      "learning_rate": 9.781523779815179e-05,
      "loss": 1.3835,
      "step": 37
    },
    {
      "epoch": 0.004364805881001608,
      "grad_norm": 2.500739336013794,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4074,
      "step": 38
    },
    {
      "epoch": 0.004479669193659545,
      "grad_norm": 1.6976559162139893,
      "learning_rate": 9.727592877996585e-05,
      "loss": 1.1213,
      "step": 39
    },
    {
      "epoch": 0.004594532506317482,
      "grad_norm": 1.8822135925292969,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.1204,
      "step": 40
    },
    {
      "epoch": 0.0047093958189754195,
      "grad_norm": 2.3586180210113525,
      "learning_rate": 9.667902132486009e-05,
      "loss": 1.3198,
      "step": 41
    },
    {
      "epoch": 0.004824259131633356,
      "grad_norm": 2.3203370571136475,
      "learning_rate": 9.635919272833938e-05,
      "loss": 1.5349,
      "step": 42
    },
    {
      "epoch": 0.004939122444291294,
      "grad_norm": 2.3646602630615234,
      "learning_rate": 9.602524267262203e-05,
      "loss": 1.4018,
      "step": 43
    },
    {
      "epoch": 0.00505398575694923,
      "grad_norm": 2.3897457122802734,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.3423,
      "step": 44
    },
    {
      "epoch": 0.005168849069607168,
      "grad_norm": 3.052333354949951,
      "learning_rate": 9.53153893518325e-05,
      "loss": 1.4598,
      "step": 45
    },
    {
      "epoch": 0.005283712382265104,
      "grad_norm": 2.3878042697906494,
      "learning_rate": 9.493970231495835e-05,
      "loss": 1.3684,
      "step": 46
    },
    {
      "epoch": 0.005398575694923042,
      "grad_norm": 3.579554557800293,
      "learning_rate": 9.45503262094184e-05,
      "loss": 1.3765,
      "step": 47
    },
    {
      "epoch": 0.005513439007580979,
      "grad_norm": 2.688589096069336,
      "learning_rate": 9.414737964294636e-05,
      "loss": 1.4563,
      "step": 48
    },
    {
      "epoch": 0.005628302320238916,
      "grad_norm": 2.9909780025482178,
      "learning_rate": 9.373098535696979e-05,
      "loss": 1.1294,
      "step": 49
    },
    {
      "epoch": 0.005743165632896853,
      "grad_norm": 1.66656494140625,
      "learning_rate": 9.330127018922194e-05,
      "loss": 1.2631,
      "step": 50
    },
    {
      "epoch": 0.005743165632896853,
      "eval_loss": 0.30388450622558594,
      "eval_runtime": 1471.9652,
      "eval_samples_per_second": 9.962,
      "eval_steps_per_second": 4.981,
      "step": 50
    },
    {
      "epoch": 0.00585802894555479,
      "grad_norm": 2.137791156768799,
      "learning_rate": 9.285836503510562e-05,
      "loss": 1.2646,
      "step": 51
    },
    {
      "epoch": 0.005972892258212727,
      "grad_norm": 2.5329372882843018,
      "learning_rate": 9.24024048078213e-05,
      "loss": 1.0451,
      "step": 52
    },
    {
      "epoch": 0.0060877555708706635,
      "grad_norm": 2.620466470718384,
      "learning_rate": 9.193352839727121e-05,
      "loss": 1.1558,
      "step": 53
    },
    {
      "epoch": 0.006202618883528601,
      "grad_norm": 2.2557244300842285,
      "learning_rate": 9.145187862775209e-05,
      "loss": 1.3604,
      "step": 54
    },
    {
      "epoch": 0.006317482196186538,
      "grad_norm": 1.8479572534561157,
      "learning_rate": 9.09576022144496e-05,
      "loss": 1.2608,
      "step": 55
    },
    {
      "epoch": 0.006432345508844475,
      "grad_norm": 2.1417593955993652,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3318,
      "step": 56
    },
    {
      "epoch": 0.006547208821502412,
      "grad_norm": 2.672443151473999,
      "learning_rate": 8.993177550236464e-05,
      "loss": 1.0541,
      "step": 57
    },
    {
      "epoch": 0.006662072134160349,
      "grad_norm": 2.344717025756836,
      "learning_rate": 8.940053768033609e-05,
      "loss": 1.2237,
      "step": 58
    },
    {
      "epoch": 0.006776935446818286,
      "grad_norm": 2.390180826187134,
      "learning_rate": 8.885729807284856e-05,
      "loss": 0.9352,
      "step": 59
    },
    {
      "epoch": 0.006891798759476223,
      "grad_norm": 3.0167908668518066,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.158,
      "step": 60
    },
    {
      "epoch": 0.00700666207213416,
      "grad_norm": 3.0811192989349365,
      "learning_rate": 8.773547901113862e-05,
      "loss": 1.1718,
      "step": 61
    },
    {
      "epoch": 0.007121525384792098,
      "grad_norm": 3.3268229961395264,
      "learning_rate": 8.715724127386972e-05,
      "loss": 1.2555,
      "step": 62
    },
    {
      "epoch": 0.007236388697450034,
      "grad_norm": 2.296539306640625,
      "learning_rate": 8.656768508095853e-05,
      "loss": 1.0316,
      "step": 63
    },
    {
      "epoch": 0.007351252010107972,
      "grad_norm": 2.4139859676361084,
      "learning_rate": 8.596699001693255e-05,
      "loss": 0.8659,
      "step": 64
    },
    {
      "epoch": 0.007466115322765908,
      "grad_norm": 2.5878524780273438,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.16,
      "step": 65
    },
    {
      "epoch": 0.007580978635423846,
      "grad_norm": 2.180026054382324,
      "learning_rate": 8.473291852294987e-05,
      "loss": 1.1024,
      "step": 66
    },
    {
      "epoch": 0.0076958419480817825,
      "grad_norm": 1.9273844957351685,
      "learning_rate": 8.409991800312493e-05,
      "loss": 0.8108,
      "step": 67
    },
    {
      "epoch": 0.00781070526073972,
      "grad_norm": 3.0962324142456055,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.0571,
      "step": 68
    },
    {
      "epoch": 0.007925568573397657,
      "grad_norm": 2.6275441646575928,
      "learning_rate": 8.280295144952536e-05,
      "loss": 0.8768,
      "step": 69
    },
    {
      "epoch": 0.008040431886055594,
      "grad_norm": 3.156315803527832,
      "learning_rate": 8.213938048432697e-05,
      "loss": 1.0539,
      "step": 70
    },
    {
      "epoch": 0.008155295198713532,
      "grad_norm": 2.818579912185669,
      "learning_rate": 8.146601955249188e-05,
      "loss": 0.9232,
      "step": 71
    },
    {
      "epoch": 0.008270158511371467,
      "grad_norm": 2.329266309738159,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.0234,
      "step": 72
    },
    {
      "epoch": 0.008385021824029405,
      "grad_norm": 3.1287198066711426,
      "learning_rate": 8.009075115760243e-05,
      "loss": 1.2203,
      "step": 73
    },
    {
      "epoch": 0.008499885136687342,
      "grad_norm": 2.1660828590393066,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.8592,
      "step": 74
    },
    {
      "epoch": 0.00861474844934528,
      "grad_norm": 2.362858533859253,
      "learning_rate": 7.86788218175523e-05,
      "loss": 1.0678,
      "step": 75
    },
    {
      "epoch": 0.008729611762003216,
      "grad_norm": 2.5020713806152344,
      "learning_rate": 7.795964517353735e-05,
      "loss": 0.8854,
      "step": 76
    },
    {
      "epoch": 0.008844475074661153,
      "grad_norm": 2.5885136127471924,
      "learning_rate": 7.723195175075136e-05,
      "loss": 0.8222,
      "step": 77
    },
    {
      "epoch": 0.00895933838731909,
      "grad_norm": 3.7722344398498535,
      "learning_rate": 7.649596321166024e-05,
      "loss": 1.1339,
      "step": 78
    },
    {
      "epoch": 0.009074201699977026,
      "grad_norm": 2.223078489303589,
      "learning_rate": 7.575190374550272e-05,
      "loss": 0.7033,
      "step": 79
    },
    {
      "epoch": 0.009189065012634964,
      "grad_norm": 2.346508026123047,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.7375,
      "step": 80
    },
    {
      "epoch": 0.009303928325292901,
      "grad_norm": 2.3102056980133057,
      "learning_rate": 7.424048101231686e-05,
      "loss": 0.8455,
      "step": 81
    },
    {
      "epoch": 0.009418791637950839,
      "grad_norm": 2.787309169769287,
      "learning_rate": 7.347357813929454e-05,
      "loss": 1.0711,
      "step": 82
    },
    {
      "epoch": 0.009533654950608775,
      "grad_norm": 2.9215545654296875,
      "learning_rate": 7.269952498697734e-05,
      "loss": 0.8927,
      "step": 83
    },
    {
      "epoch": 0.009648518263266712,
      "grad_norm": 2.4672605991363525,
      "learning_rate": 7.191855733945387e-05,
      "loss": 0.713,
      "step": 84
    },
    {
      "epoch": 0.00976338157592465,
      "grad_norm": 3.549206018447876,
      "learning_rate": 7.113091308703498e-05,
      "loss": 1.0348,
      "step": 85
    },
    {
      "epoch": 0.009878244888582587,
      "grad_norm": 2.957688093185425,
      "learning_rate": 7.033683215379002e-05,
      "loss": 0.9689,
      "step": 86
    },
    {
      "epoch": 0.009993108201240523,
      "grad_norm": 2.4251294136047363,
      "learning_rate": 6.953655642446368e-05,
      "loss": 0.6512,
      "step": 87
    },
    {
      "epoch": 0.01010797151389846,
      "grad_norm": 2.5786890983581543,
      "learning_rate": 6.873032967079561e-05,
      "loss": 0.7163,
      "step": 88
    },
    {
      "epoch": 0.010222834826556398,
      "grad_norm": 3.341750144958496,
      "learning_rate": 6.7918397477265e-05,
      "loss": 0.6509,
      "step": 89
    },
    {
      "epoch": 0.010337698139214336,
      "grad_norm": 3.332127094268799,
      "learning_rate": 6.710100716628344e-05,
      "loss": 1.0058,
      "step": 90
    },
    {
      "epoch": 0.010452561451872271,
      "grad_norm": 3.31768536567688,
      "learning_rate": 6.627840772285784e-05,
      "loss": 0.7874,
      "step": 91
    },
    {
      "epoch": 0.010567424764530209,
      "grad_norm": 2.435286521911621,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.6309,
      "step": 92
    },
    {
      "epoch": 0.010682288077188146,
      "grad_norm": 2.9863882064819336,
      "learning_rate": 6.461858523613684e-05,
      "loss": 0.8979,
      "step": 93
    },
    {
      "epoch": 0.010797151389846084,
      "grad_norm": 3.188188314437866,
      "learning_rate": 6.378186779084995e-05,
      "loss": 0.8585,
      "step": 94
    },
    {
      "epoch": 0.01091201470250402,
      "grad_norm": 3.706717014312744,
      "learning_rate": 6.294095225512603e-05,
      "loss": 0.8157,
      "step": 95
    },
    {
      "epoch": 0.011026878015161957,
      "grad_norm": 3.348717212677002,
      "learning_rate": 6.209609477998338e-05,
      "loss": 0.6131,
      "step": 96
    },
    {
      "epoch": 0.011141741327819895,
      "grad_norm": 3.345127820968628,
      "learning_rate": 6.124755271719325e-05,
      "loss": 0.7124,
      "step": 97
    },
    {
      "epoch": 0.011256604640477832,
      "grad_norm": 3.588732957839966,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 0.6565,
      "step": 98
    },
    {
      "epoch": 0.011371467953135768,
      "grad_norm": 3.6394217014312744,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 0.6834,
      "step": 99
    },
    {
      "epoch": 0.011486331265793705,
      "grad_norm": 3.545775890350342,
      "learning_rate": 5.868240888334653e-05,
      "loss": 0.9438,
      "step": 100
    },
    {
      "epoch": 0.011486331265793705,
      "eval_loss": 0.17487511038780212,
      "eval_runtime": 1471.3897,
      "eval_samples_per_second": 9.965,
      "eval_steps_per_second": 4.983,
      "step": 100
    },
    {
      "epoch": 0.011601194578451643,
      "grad_norm": 2.6634979248046875,
      "learning_rate": 5.782172325201155e-05,
      "loss": 0.6536,
      "step": 101
    },
    {
      "epoch": 0.01171605789110958,
      "grad_norm": 2.7935872077941895,
      "learning_rate": 5.695865504800327e-05,
      "loss": 0.8657,
      "step": 102
    },
    {
      "epoch": 0.011830921203767516,
      "grad_norm": 3.1238162517547607,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 0.6678,
      "step": 103
    },
    {
      "epoch": 0.011945784516425454,
      "grad_norm": 3.002263069152832,
      "learning_rate": 5.522642316338268e-05,
      "loss": 0.6514,
      "step": 104
    },
    {
      "epoch": 0.012060647829083391,
      "grad_norm": 2.975006580352783,
      "learning_rate": 5.435778713738292e-05,
      "loss": 0.4484,
      "step": 105
    },
    {
      "epoch": 0.012175511141741327,
      "grad_norm": 2.8138067722320557,
      "learning_rate": 5.348782368720626e-05,
      "loss": 0.5668,
      "step": 106
    },
    {
      "epoch": 0.012290374454399264,
      "grad_norm": 2.3226609230041504,
      "learning_rate": 5.26167978121472e-05,
      "loss": 0.4671,
      "step": 107
    },
    {
      "epoch": 0.012405237767057202,
      "grad_norm": 3.791348934173584,
      "learning_rate": 5.174497483512506e-05,
      "loss": 0.6958,
      "step": 108
    },
    {
      "epoch": 0.01252010107971514,
      "grad_norm": 2.7212183475494385,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 0.6117,
      "step": 109
    },
    {
      "epoch": 0.012634964392373075,
      "grad_norm": 3.5150516033172607,
      "learning_rate": 5e-05,
      "loss": 0.822,
      "step": 110
    },
    {
      "epoch": 0.012749827705031013,
      "grad_norm": 2.905332088470459,
      "learning_rate": 4.912737967813583e-05,
      "loss": 0.6823,
      "step": 111
    },
    {
      "epoch": 0.01286469101768895,
      "grad_norm": 3.329963207244873,
      "learning_rate": 4.825502516487497e-05,
      "loss": 0.559,
      "step": 112
    },
    {
      "epoch": 0.012979554330346888,
      "grad_norm": 3.3605265617370605,
      "learning_rate": 4.738320218785281e-05,
      "loss": 0.5544,
      "step": 113
    },
    {
      "epoch": 0.013094417643004824,
      "grad_norm": 3.592366933822632,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 0.5781,
      "step": 114
    },
    {
      "epoch": 0.013209280955662761,
      "grad_norm": 3.577732563018799,
      "learning_rate": 4.564221286261709e-05,
      "loss": 0.6123,
      "step": 115
    },
    {
      "epoch": 0.013324144268320699,
      "grad_norm": 4.342945575714111,
      "learning_rate": 4.477357683661734e-05,
      "loss": 0.7269,
      "step": 116
    },
    {
      "epoch": 0.013439007580978636,
      "grad_norm": 2.921304941177368,
      "learning_rate": 4.390653282974264e-05,
      "loss": 0.5198,
      "step": 117
    },
    {
      "epoch": 0.013553870893636572,
      "grad_norm": 3.515984058380127,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 0.9182,
      "step": 118
    },
    {
      "epoch": 0.01366873420629451,
      "grad_norm": 2.688448190689087,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 0.6087,
      "step": 119
    },
    {
      "epoch": 0.013783597518952447,
      "grad_norm": 2.6708593368530273,
      "learning_rate": 4.131759111665349e-05,
      "loss": 0.5885,
      "step": 120
    },
    {
      "epoch": 0.013898460831610384,
      "grad_norm": 2.5848822593688965,
      "learning_rate": 4.045955023117276e-05,
      "loss": 0.5533,
      "step": 121
    },
    {
      "epoch": 0.01401332414426832,
      "grad_norm": 3.2022759914398193,
      "learning_rate": 3.960441545911204e-05,
      "loss": 0.5861,
      "step": 122
    },
    {
      "epoch": 0.014128187456926258,
      "grad_norm": 2.4787240028381348,
      "learning_rate": 3.875244728280676e-05,
      "loss": 0.4025,
      "step": 123
    },
    {
      "epoch": 0.014243050769584195,
      "grad_norm": 3.095510959625244,
      "learning_rate": 3.790390522001662e-05,
      "loss": 0.5104,
      "step": 124
    },
    {
      "epoch": 0.014357914082242133,
      "grad_norm": 3.846745491027832,
      "learning_rate": 3.705904774487396e-05,
      "loss": 0.5936,
      "step": 125
    },
    {
      "epoch": 0.014472777394900068,
      "grad_norm": 2.9915177822113037,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 0.5739,
      "step": 126
    },
    {
      "epoch": 0.014587640707558006,
      "grad_norm": 2.5920958518981934,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 0.4191,
      "step": 127
    },
    {
      "epoch": 0.014702504020215943,
      "grad_norm": 2.4326400756835938,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.437,
      "step": 128
    },
    {
      "epoch": 0.014817367332873881,
      "grad_norm": 2.664111614227295,
      "learning_rate": 3.372159227714218e-05,
      "loss": 0.4107,
      "step": 129
    },
    {
      "epoch": 0.014932230645531817,
      "grad_norm": 3.015989303588867,
      "learning_rate": 3.289899283371657e-05,
      "loss": 0.6345,
      "step": 130
    },
    {
      "epoch": 0.015047093958189754,
      "grad_norm": 2.8984575271606445,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 0.5547,
      "step": 131
    },
    {
      "epoch": 0.015161957270847692,
      "grad_norm": 3.772346019744873,
      "learning_rate": 3.12696703292044e-05,
      "loss": 0.6239,
      "step": 132
    },
    {
      "epoch": 0.015276820583505627,
      "grad_norm": 3.629082441329956,
      "learning_rate": 3.046344357553632e-05,
      "loss": 0.5739,
      "step": 133
    },
    {
      "epoch": 0.015391683896163565,
      "grad_norm": 3.2363765239715576,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 0.661,
      "step": 134
    },
    {
      "epoch": 0.015506547208821502,
      "grad_norm": 2.6455376148223877,
      "learning_rate": 2.886908691296504e-05,
      "loss": 0.4586,
      "step": 135
    },
    {
      "epoch": 0.01562141052147944,
      "grad_norm": 2.9570345878601074,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 0.6333,
      "step": 136
    },
    {
      "epoch": 0.015736273834137376,
      "grad_norm": 2.019033670425415,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 0.3232,
      "step": 137
    },
    {
      "epoch": 0.015851137146795313,
      "grad_norm": 2.6063356399536133,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 0.4487,
      "step": 138
    },
    {
      "epoch": 0.01596600045945325,
      "grad_norm": 3.5841784477233887,
      "learning_rate": 2.575951898768315e-05,
      "loss": 0.6579,
      "step": 139
    },
    {
      "epoch": 0.01608086377211119,
      "grad_norm": 3.1793031692504883,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.6123,
      "step": 140
    },
    {
      "epoch": 0.016195727084769126,
      "grad_norm": 1.9112600088119507,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 0.2844,
      "step": 141
    },
    {
      "epoch": 0.016310590397427063,
      "grad_norm": 2.6747708320617676,
      "learning_rate": 2.350403678833976e-05,
      "loss": 0.4366,
      "step": 142
    },
    {
      "epoch": 0.016425453710084997,
      "grad_norm": 2.7817370891571045,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 0.4206,
      "step": 143
    },
    {
      "epoch": 0.016540317022742935,
      "grad_norm": 3.0754709243774414,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 0.4934,
      "step": 144
    },
    {
      "epoch": 0.016655180335400872,
      "grad_norm": 2.9075064659118652,
      "learning_rate": 2.132117818244771e-05,
      "loss": 0.3818,
      "step": 145
    },
    {
      "epoch": 0.01677004364805881,
      "grad_norm": 3.9040424823760986,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.6472,
      "step": 146
    },
    {
      "epoch": 0.016884906960716747,
      "grad_norm": 2.7987887859344482,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 0.3802,
      "step": 147
    },
    {
      "epoch": 0.016999770273374685,
      "grad_norm": 3.3494410514831543,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 0.5155,
      "step": 148
    },
    {
      "epoch": 0.017114633586032622,
      "grad_norm": 2.892005681991577,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 0.385,
      "step": 149
    },
    {
      "epoch": 0.01722949689869056,
      "grad_norm": 2.724477529525757,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 0.5024,
      "step": 150
    },
    {
      "epoch": 0.01722949689869056,
      "eval_loss": 0.11594025045633316,
      "eval_runtime": 1471.5665,
      "eval_samples_per_second": 9.964,
      "eval_steps_per_second": 4.982,
      "step": 150
    },
    {
      "epoch": 0.017344360211348494,
      "grad_norm": 3.007096767425537,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 0.5254,
      "step": 151
    },
    {
      "epoch": 0.01745922352400643,
      "grad_norm": 2.6034908294677734,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 0.4203,
      "step": 152
    },
    {
      "epoch": 0.01757408683666437,
      "grad_norm": 3.8404593467712402,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 0.7015,
      "step": 153
    },
    {
      "epoch": 0.017688950149322306,
      "grad_norm": 2.672938346862793,
      "learning_rate": 1.526708147705013e-05,
      "loss": 0.418,
      "step": 154
    },
    {
      "epoch": 0.017803813461980244,
      "grad_norm": 2.8745882511138916,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 0.4368,
      "step": 155
    },
    {
      "epoch": 0.01791867677463818,
      "grad_norm": 1.8944249153137207,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 0.2554,
      "step": 156
    },
    {
      "epoch": 0.01803354008729612,
      "grad_norm": 2.5466866493225098,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 0.4435,
      "step": 157
    },
    {
      "epoch": 0.018148403399954053,
      "grad_norm": 3.0349109172821045,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 0.4948,
      "step": 158
    },
    {
      "epoch": 0.01826326671261199,
      "grad_norm": 2.655149459838867,
      "learning_rate": 1.22645209888614e-05,
      "loss": 0.4062,
      "step": 159
    },
    {
      "epoch": 0.018378130025269928,
      "grad_norm": 2.7073023319244385,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 0.3941,
      "step": 160
    },
    {
      "epoch": 0.018492993337927865,
      "grad_norm": 2.6231350898742676,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 0.4092,
      "step": 161
    },
    {
      "epoch": 0.018607856650585803,
      "grad_norm": 2.469143867492676,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 0.3605,
      "step": 162
    },
    {
      "epoch": 0.01872271996324374,
      "grad_norm": 2.735646963119507,
      "learning_rate": 1.006822449763537e-05,
      "loss": 0.4245,
      "step": 163
    },
    {
      "epoch": 0.018837583275901678,
      "grad_norm": 4.04184627532959,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.5681,
      "step": 164
    },
    {
      "epoch": 0.018952446588559616,
      "grad_norm": 2.9459590911865234,
      "learning_rate": 9.042397785550405e-06,
      "loss": 0.3627,
      "step": 165
    },
    {
      "epoch": 0.01906730990121755,
      "grad_norm": 3.2683775424957275,
      "learning_rate": 8.548121372247918e-06,
      "loss": 0.4684,
      "step": 166
    },
    {
      "epoch": 0.019182173213875487,
      "grad_norm": 2.3902087211608887,
      "learning_rate": 8.066471602728803e-06,
      "loss": 0.3512,
      "step": 167
    },
    {
      "epoch": 0.019297036526533425,
      "grad_norm": 2.420001745223999,
      "learning_rate": 7.597595192178702e-06,
      "loss": 0.3433,
      "step": 168
    },
    {
      "epoch": 0.019411899839191362,
      "grad_norm": 2.396726131439209,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 0.3181,
      "step": 169
    },
    {
      "epoch": 0.0195267631518493,
      "grad_norm": 3.2419331073760986,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.5015,
      "step": 170
    },
    {
      "epoch": 0.019641626464507237,
      "grad_norm": 1.9583468437194824,
      "learning_rate": 6.269014643030213e-06,
      "loss": 0.2504,
      "step": 171
    },
    {
      "epoch": 0.019756489777165175,
      "grad_norm": 3.0308303833007812,
      "learning_rate": 5.852620357053651e-06,
      "loss": 0.5331,
      "step": 172
    },
    {
      "epoch": 0.019871353089823112,
      "grad_norm": 2.8468117713928223,
      "learning_rate": 5.449673790581611e-06,
      "loss": 0.3874,
      "step": 173
    },
    {
      "epoch": 0.019986216402481046,
      "grad_norm": 2.1451475620269775,
      "learning_rate": 5.060297685041659e-06,
      "loss": 0.274,
      "step": 174
    },
    {
      "epoch": 0.020101079715138984,
      "grad_norm": 2.728893756866455,
      "learning_rate": 4.684610648167503e-06,
      "loss": 0.401,
      "step": 175
    },
    {
      "epoch": 0.02021594302779692,
      "grad_norm": 3.027879238128662,
      "learning_rate": 4.322727117869951e-06,
      "loss": 0.4416,
      "step": 176
    },
    {
      "epoch": 0.02033080634045486,
      "grad_norm": 2.050403118133545,
      "learning_rate": 3.974757327377981e-06,
      "loss": 0.2707,
      "step": 177
    },
    {
      "epoch": 0.020445669653112796,
      "grad_norm": 3.0397396087646484,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 0.487,
      "step": 178
    },
    {
      "epoch": 0.020560532965770734,
      "grad_norm": 3.4382946491241455,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 0.6042,
      "step": 179
    },
    {
      "epoch": 0.02067539627842867,
      "grad_norm": 2.5902884006500244,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 0.3982,
      "step": 180
    },
    {
      "epoch": 0.02079025959108661,
      "grad_norm": 2.472612142562866,
      "learning_rate": 2.724071220034158e-06,
      "loss": 0.3634,
      "step": 181
    },
    {
      "epoch": 0.020905122903744543,
      "grad_norm": 2.7177581787109375,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.4134,
      "step": 182
    },
    {
      "epoch": 0.02101998621640248,
      "grad_norm": 2.9707839488983154,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 0.5281,
      "step": 183
    },
    {
      "epoch": 0.021134849529060418,
      "grad_norm": 2.7435781955718994,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 0.4096,
      "step": 184
    },
    {
      "epoch": 0.021249712841718355,
      "grad_norm": 2.72346568107605,
      "learning_rate": 1.70370868554659e-06,
      "loss": 0.4428,
      "step": 185
    },
    {
      "epoch": 0.021364576154376293,
      "grad_norm": 2.1403918266296387,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 0.2873,
      "step": 186
    },
    {
      "epoch": 0.02147943946703423,
      "grad_norm": 3.263788938522339,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 0.5087,
      "step": 187
    },
    {
      "epoch": 0.021594302779692168,
      "grad_norm": 2.661633253097534,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 0.402,
      "step": 188
    },
    {
      "epoch": 0.021709166092350102,
      "grad_norm": 2.249889373779297,
      "learning_rate": 9.186408276168013e-07,
      "loss": 0.3121,
      "step": 189
    },
    {
      "epoch": 0.02182402940500804,
      "grad_norm": 2.8329484462738037,
      "learning_rate": 7.596123493895991e-07,
      "loss": 0.4427,
      "step": 190
    },
    {
      "epoch": 0.021938892717665977,
      "grad_norm": 2.6768405437469482,
      "learning_rate": 6.15582970243117e-07,
      "loss": 0.3991,
      "step": 191
    },
    {
      "epoch": 0.022053756030323914,
      "grad_norm": 2.998131275177002,
      "learning_rate": 4.865965629214819e-07,
      "loss": 0.4568,
      "step": 192
    },
    {
      "epoch": 0.022168619342981852,
      "grad_norm": 3.1869330406188965,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 0.4539,
      "step": 193
    },
    {
      "epoch": 0.02228348265563979,
      "grad_norm": 2.072638511657715,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 0.2798,
      "step": 194
    },
    {
      "epoch": 0.022398345968297727,
      "grad_norm": 2.542144298553467,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 0.3783,
      "step": 195
    },
    {
      "epoch": 0.022513209280955664,
      "grad_norm": 2.675973892211914,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 0.4263,
      "step": 196
    },
    {
      "epoch": 0.0226280725936136,
      "grad_norm": 2.341843366622925,
      "learning_rate": 6.852326227130834e-08,
      "loss": 0.3602,
      "step": 197
    },
    {
      "epoch": 0.022742935906271536,
      "grad_norm": 2.840686082839966,
      "learning_rate": 3.04586490452119e-08,
      "loss": 0.4262,
      "step": 198
    },
    {
      "epoch": 0.022857799218929473,
      "grad_norm": 1.991869568824768,
      "learning_rate": 7.615242180436522e-09,
      "loss": 0.272,
      "step": 199
    },
    {
      "epoch": 0.02297266253158741,
      "grad_norm": 3.1009721755981445,
      "learning_rate": 0.0,
      "loss": 0.5344,
      "step": 200
    },
    {
      "epoch": 0.02297266253158741,
      "eval_loss": 0.1013597771525383,
      "eval_runtime": 1471.8234,
      "eval_samples_per_second": 9.962,
      "eval_steps_per_second": 4.982,
      "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": 2.861996017975296e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}