{
  "best_metric": 0.6426064968109131,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.03536380514543365,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00017681902572716824,
      "grad_norm": 30.77566909790039,
      "learning_rate": 1e-05,
      "loss": 5.7904,
      "step": 1
    },
    {
      "epoch": 0.00017681902572716824,
      "eval_loss": 1.5057493448257446,
      "eval_runtime": 699.8238,
      "eval_samples_per_second": 13.611,
      "eval_steps_per_second": 3.404,
      "step": 1
    },
    {
      "epoch": 0.0003536380514543365,
      "grad_norm": 33.43643569946289,
      "learning_rate": 2e-05,
      "loss": 6.1229,
      "step": 2
    },
    {
      "epoch": 0.0005304570771815047,
      "grad_norm": 22.859066009521484,
      "learning_rate": 3e-05,
      "loss": 5.832,
      "step": 3
    },
    {
      "epoch": 0.000707276102908673,
      "grad_norm": 16.750789642333984,
      "learning_rate": 4e-05,
      "loss": 4.8626,
      "step": 4
    },
    {
      "epoch": 0.0008840951286358412,
      "grad_norm": 14.933880805969238,
      "learning_rate": 5e-05,
      "loss": 4.4985,
      "step": 5
    },
    {
      "epoch": 0.0010609141543630094,
      "grad_norm": 12.131953239440918,
      "learning_rate": 6e-05,
      "loss": 3.9705,
      "step": 6
    },
    {
      "epoch": 0.0012377331800901778,
      "grad_norm": 10.108991622924805,
      "learning_rate": 7e-05,
      "loss": 3.774,
      "step": 7
    },
    {
      "epoch": 0.001414552205817346,
      "grad_norm": 10.229625701904297,
      "learning_rate": 8e-05,
      "loss": 3.6096,
      "step": 8
    },
    {
      "epoch": 0.001591371231544514,
      "grad_norm": 10.535920143127441,
      "learning_rate": 9e-05,
      "loss": 3.4153,
      "step": 9
    },
    {
      "epoch": 0.0017681902572716825,
      "grad_norm": 9.722245216369629,
      "learning_rate": 0.0001,
      "loss": 3.6757,
      "step": 10
    },
    {
      "epoch": 0.0019450092829988506,
      "grad_norm": 15.117246627807617,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.6501,
      "step": 11
    },
    {
      "epoch": 0.002121828308726019,
      "grad_norm": 8.139172554016113,
      "learning_rate": 9.997266286704631e-05,
      "loss": 3.5474,
      "step": 12
    },
    {
      "epoch": 0.002298647334453187,
      "grad_norm": 9.034537315368652,
      "learning_rate": 9.993849845741524e-05,
      "loss": 3.4964,
      "step": 13
    },
    {
      "epoch": 0.0024754663601803555,
      "grad_norm": 7.310268402099609,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.3639,
      "step": 14
    },
    {
      "epoch": 0.0026522853859075237,
      "grad_norm": 6.865406513214111,
      "learning_rate": 9.98292246503335e-05,
      "loss": 3.2937,
      "step": 15
    },
    {
      "epoch": 0.002829104411634692,
      "grad_norm": 6.944815635681152,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.2835,
      "step": 16
    },
    {
      "epoch": 0.00300592343736186,
      "grad_norm": 6.507901668548584,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.1566,
      "step": 17
    },
    {
      "epoch": 0.003182742463089028,
      "grad_norm": 7.71777868270874,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.2667,
      "step": 18
    },
    {
      "epoch": 0.0033595614888161968,
      "grad_norm": 6.91279935836792,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.1104,
      "step": 19
    },
    {
      "epoch": 0.003536380514543365,
      "grad_norm": 6.425033092498779,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.1124,
      "step": 20
    },
    {
      "epoch": 0.003713199540270533,
      "grad_norm": 6.594813346862793,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.0918,
      "step": 21
    },
    {
      "epoch": 0.0038900185659977013,
      "grad_norm": 6.123743534088135,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.1857,
      "step": 22
    },
    {
      "epoch": 0.00406683759172487,
      "grad_norm": 5.8520989418029785,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.1095,
      "step": 23
    },
    {
      "epoch": 0.004243656617452038,
      "grad_norm": 5.65989351272583,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.1227,
      "step": 24
    },
    {
      "epoch": 0.004420475643179206,
      "grad_norm": 6.06626558303833,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.0941,
      "step": 25
    },
    {
      "epoch": 0.004597294668906374,
      "grad_norm": 6.0456132888793945,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.017,
      "step": 26
    },
    {
      "epoch": 0.0047741136946335425,
      "grad_norm": 5.214839935302734,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.7115,
      "step": 27
    },
    {
      "epoch": 0.004950932720360711,
      "grad_norm": 5.342077732086182,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.1084,
      "step": 28
    },
    {
      "epoch": 0.005127751746087879,
      "grad_norm": 5.682497501373291,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.8929,
      "step": 29
    },
    {
      "epoch": 0.005304570771815047,
      "grad_norm": 5.593482494354248,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.1474,
      "step": 30
    },
    {
      "epoch": 0.005481389797542215,
      "grad_norm": 5.597923278808594,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.746,
      "step": 31
    },
    {
      "epoch": 0.005658208823269384,
      "grad_norm": 6.232513427734375,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.9022,
      "step": 32
    },
    {
      "epoch": 0.005835027848996552,
      "grad_norm": 5.472952365875244,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.9239,
      "step": 33
    },
    {
      "epoch": 0.00601184687472372,
      "grad_norm": 4.986033916473389,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.8302,
      "step": 34
    },
    {
      "epoch": 0.006188665900450889,
      "grad_norm": 4.939537525177002,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.9965,
      "step": 35
    },
    {
      "epoch": 0.006365484926178056,
      "grad_norm": 4.7341628074646,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.8888,
      "step": 36
    },
    {
      "epoch": 0.006542303951905225,
      "grad_norm": 5.597661018371582,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.9968,
      "step": 37
    },
    {
      "epoch": 0.0067191229776323936,
      "grad_norm": 5.1189165115356445,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.8592,
      "step": 38
    },
    {
      "epoch": 0.006895942003359561,
      "grad_norm": 5.132823944091797,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.9616,
      "step": 39
    },
    {
      "epoch": 0.00707276102908673,
      "grad_norm": 5.72930383682251,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.7738,
      "step": 40
    },
    {
      "epoch": 0.007249580054813898,
      "grad_norm": 5.405780792236328,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.9437,
      "step": 41
    },
    {
      "epoch": 0.007426399080541066,
      "grad_norm": 5.408172607421875,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.8486,
      "step": 42
    },
    {
      "epoch": 0.007603218106268235,
      "grad_norm": 4.897181987762451,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.6534,
      "step": 43
    },
    {
      "epoch": 0.0077800371319954025,
      "grad_norm": 4.973101615905762,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.7654,
      "step": 44
    },
    {
      "epoch": 0.007956856157722571,
      "grad_norm": 4.965903282165527,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.7389,
      "step": 45
    },
    {
      "epoch": 0.00813367518344974,
      "grad_norm": 5.6687750816345215,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.6717,
      "step": 46
    },
    {
      "epoch": 0.008310494209176908,
      "grad_norm": 5.180047512054443,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.7604,
      "step": 47
    },
    {
      "epoch": 0.008487313234904075,
      "grad_norm": 4.873654842376709,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.6169,
      "step": 48
    },
    {
      "epoch": 0.008664132260631244,
      "grad_norm": 6.44728946685791,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.6376,
      "step": 49
    },
    {
      "epoch": 0.008840951286358412,
      "grad_norm": 7.412442207336426,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.8407,
      "step": 50
    },
    {
      "epoch": 0.008840951286358412,
      "eval_loss": 0.8082038164138794,
      "eval_runtime": 704.1742,
      "eval_samples_per_second": 13.526,
      "eval_steps_per_second": 3.383,
      "step": 50
    },
    {
      "epoch": 0.009017770312085581,
      "grad_norm": 10.941012382507324,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.9579,
      "step": 51
    },
    {
      "epoch": 0.009194589337812748,
      "grad_norm": 5.875502109527588,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.3415,
      "step": 52
    },
    {
      "epoch": 0.009371408363539916,
      "grad_norm": 5.685242652893066,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.1855,
      "step": 53
    },
    {
      "epoch": 0.009548227389267085,
      "grad_norm": 7.529914855957031,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.3116,
      "step": 54
    },
    {
      "epoch": 0.009725046414994254,
      "grad_norm": 4.6792707443237305,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.022,
      "step": 55
    },
    {
      "epoch": 0.009901865440721422,
      "grad_norm": 4.322147369384766,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.1485,
      "step": 56
    },
    {
      "epoch": 0.010078684466448589,
      "grad_norm": 4.370192050933838,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.0674,
      "step": 57
    },
    {
      "epoch": 0.010255503492175758,
      "grad_norm": 4.796641826629639,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.998,
      "step": 58
    },
    {
      "epoch": 0.010432322517902926,
      "grad_norm": 5.605325698852539,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.1447,
      "step": 59
    },
    {
      "epoch": 0.010609141543630095,
      "grad_norm": 4.807485103607178,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.9791,
      "step": 60
    },
    {
      "epoch": 0.010785960569357263,
      "grad_norm": 4.288360595703125,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.8238,
      "step": 61
    },
    {
      "epoch": 0.01096277959508443,
      "grad_norm": 4.529268264770508,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.0149,
      "step": 62
    },
    {
      "epoch": 0.011139598620811599,
      "grad_norm": 4.85869026184082,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.0277,
      "step": 63
    },
    {
      "epoch": 0.011316417646538767,
      "grad_norm": 4.504361152648926,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.8148,
      "step": 64
    },
    {
      "epoch": 0.011493236672265936,
      "grad_norm": 4.638814449310303,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.145,
      "step": 65
    },
    {
      "epoch": 0.011670055697993105,
      "grad_norm": 4.652398109436035,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.8357,
      "step": 66
    },
    {
      "epoch": 0.011846874723720272,
      "grad_norm": 4.040258884429932,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.7536,
      "step": 67
    },
    {
      "epoch": 0.01202369374944744,
      "grad_norm": 4.549870014190674,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.9259,
      "step": 68
    },
    {
      "epoch": 0.012200512775174609,
      "grad_norm": 3.9912383556365967,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.7272,
      "step": 69
    },
    {
      "epoch": 0.012377331800901777,
      "grad_norm": 4.487094402313232,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.8944,
      "step": 70
    },
    {
      "epoch": 0.012554150826628946,
      "grad_norm": 4.489090919494629,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.9474,
      "step": 71
    },
    {
      "epoch": 0.012730969852356113,
      "grad_norm": 4.538005828857422,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.8556,
      "step": 72
    },
    {
      "epoch": 0.012907788878083281,
      "grad_norm": 4.3861470222473145,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.0211,
      "step": 73
    },
    {
      "epoch": 0.01308460790381045,
      "grad_norm": 4.112513065338135,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.8523,
      "step": 74
    },
    {
      "epoch": 0.013261426929537619,
      "grad_norm": 4.240283489227295,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.7448,
      "step": 75
    },
    {
      "epoch": 0.013438245955264787,
      "grad_norm": 4.358696460723877,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.8285,
      "step": 76
    },
    {
      "epoch": 0.013615064980991954,
      "grad_norm": 4.034807205200195,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.7714,
      "step": 77
    },
    {
      "epoch": 0.013791884006719123,
      "grad_norm": 4.1600518226623535,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.915,
      "step": 78
    },
    {
      "epoch": 0.013968703032446291,
      "grad_norm": 4.095190525054932,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.8567,
      "step": 79
    },
    {
      "epoch": 0.01414552205817346,
      "grad_norm": 4.382471084594727,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.9988,
      "step": 80
    },
    {
      "epoch": 0.014322341083900628,
      "grad_norm": 4.3130269050598145,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.7513,
      "step": 81
    },
    {
      "epoch": 0.014499160109627795,
      "grad_norm": 4.03249454498291,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.8703,
      "step": 82
    },
    {
      "epoch": 0.014675979135354964,
      "grad_norm": 3.758502244949341,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.7361,
      "step": 83
    },
    {
      "epoch": 0.014852798161082132,
      "grad_norm": 4.064537048339844,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.697,
      "step": 84
    },
    {
      "epoch": 0.015029617186809301,
      "grad_norm": 4.2941670417785645,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.9156,
      "step": 85
    },
    {
      "epoch": 0.01520643621253647,
      "grad_norm": 4.217069625854492,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.6571,
      "step": 86
    },
    {
      "epoch": 0.015383255238263636,
      "grad_norm": 4.212507247924805,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.7511,
      "step": 87
    },
    {
      "epoch": 0.015560074263990805,
      "grad_norm": 4.505743980407715,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.6847,
      "step": 88
    },
    {
      "epoch": 0.015736893289717974,
      "grad_norm": 3.9542410373687744,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.4308,
      "step": 89
    },
    {
      "epoch": 0.015913712315445142,
      "grad_norm": 4.286926746368408,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.5392,
      "step": 90
    },
    {
      "epoch": 0.01609053134117231,
      "grad_norm": 4.219004154205322,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.5647,
      "step": 91
    },
    {
      "epoch": 0.01626735036689948,
      "grad_norm": 4.355921268463135,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.7065,
      "step": 92
    },
    {
      "epoch": 0.016444169392626648,
      "grad_norm": 3.816110610961914,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.4166,
      "step": 93
    },
    {
      "epoch": 0.016620988418353817,
      "grad_norm": 3.980553388595581,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.4319,
      "step": 94
    },
    {
      "epoch": 0.01679780744408098,
      "grad_norm": 4.162726879119873,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6132,
      "step": 95
    },
    {
      "epoch": 0.01697462646980815,
      "grad_norm": 3.9977686405181885,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.5249,
      "step": 96
    },
    {
      "epoch": 0.01715144549553532,
      "grad_norm": 4.283101558685303,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.5313,
      "step": 97
    },
    {
      "epoch": 0.017328264521262488,
      "grad_norm": 4.2165656089782715,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.4478,
      "step": 98
    },
    {
      "epoch": 0.017505083546989656,
      "grad_norm": 4.173795223236084,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.427,
      "step": 99
    },
    {
      "epoch": 0.017681902572716825,
      "grad_norm": 4.588566780090332,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4309,
      "step": 100
    },
    {
      "epoch": 0.017681902572716825,
      "eval_loss": 0.7389817237854004,
      "eval_runtime": 704.1343,
      "eval_samples_per_second": 13.527,
      "eval_steps_per_second": 3.383,
      "step": 100
    },
    {
      "epoch": 0.017858721598443993,
      "grad_norm": 8.920741081237793,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.3914,
      "step": 101
    },
    {
      "epoch": 0.018035540624171162,
      "grad_norm": 7.691524982452393,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.4589,
      "step": 102
    },
    {
      "epoch": 0.01821235964989833,
      "grad_norm": 4.382842063903809,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.9892,
      "step": 103
    },
    {
      "epoch": 0.018389178675625496,
      "grad_norm": 4.469934463500977,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.1142,
      "step": 104
    },
    {
      "epoch": 0.018565997701352664,
      "grad_norm": 4.711705207824707,
      "learning_rate": 5e-05,
      "loss": 3.1713,
      "step": 105
    },
    {
      "epoch": 0.018742816727079833,
      "grad_norm": 4.239561080932617,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.0092,
      "step": 106
    },
    {
      "epoch": 0.018919635752807,
      "grad_norm": 3.846534013748169,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.0243,
      "step": 107
    },
    {
      "epoch": 0.01909645477853417,
      "grad_norm": 3.866678476333618,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.8102,
      "step": 108
    },
    {
      "epoch": 0.01927327380426134,
      "grad_norm": 3.4842145442962646,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.7249,
      "step": 109
    },
    {
      "epoch": 0.019450092829988507,
      "grad_norm": 3.4887750148773193,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.8563,
      "step": 110
    },
    {
      "epoch": 0.019626911855715676,
      "grad_norm": 3.680882453918457,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.6713,
      "step": 111
    },
    {
      "epoch": 0.019803730881442844,
      "grad_norm": 4.103943347930908,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.0004,
      "step": 112
    },
    {
      "epoch": 0.019980549907170013,
      "grad_norm": 3.867415428161621,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.8714,
      "step": 113
    },
    {
      "epoch": 0.020157368932897178,
      "grad_norm": 3.6518120765686035,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.4498,
      "step": 114
    },
    {
      "epoch": 0.020334187958624347,
      "grad_norm": 3.914706230163574,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.7675,
      "step": 115
    },
    {
      "epoch": 0.020511006984351515,
      "grad_norm": 3.9167869091033936,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.8318,
      "step": 116
    },
    {
      "epoch": 0.020687826010078684,
      "grad_norm": 3.727733850479126,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.715,
      "step": 117
    },
    {
      "epoch": 0.020864645035805852,
      "grad_norm": 3.9161274433135986,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.7997,
      "step": 118
    },
    {
      "epoch": 0.02104146406153302,
      "grad_norm": 3.6844944953918457,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.752,
      "step": 119
    },
    {
      "epoch": 0.02121828308726019,
      "grad_norm": 4.133869647979736,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.7113,
      "step": 120
    },
    {
      "epoch": 0.021395102112987358,
      "grad_norm": 3.8387954235076904,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.5577,
      "step": 121
    },
    {
      "epoch": 0.021571921138714527,
      "grad_norm": 3.6142640113830566,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.6402,
      "step": 122
    },
    {
      "epoch": 0.021748740164441695,
      "grad_norm": 3.8690361976623535,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.6052,
      "step": 123
    },
    {
      "epoch": 0.02192555919016886,
      "grad_norm": 3.8395142555236816,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.6495,
      "step": 124
    },
    {
      "epoch": 0.02210237821589603,
      "grad_norm": 4.113121032714844,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.6376,
      "step": 125
    },
    {
      "epoch": 0.022279197241623198,
      "grad_norm": 3.76629376411438,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.5573,
      "step": 126
    },
    {
      "epoch": 0.022456016267350366,
      "grad_norm": 3.7748351097106934,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.6338,
      "step": 127
    },
    {
      "epoch": 0.022632835293077535,
      "grad_norm": 3.8269357681274414,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.7441,
      "step": 128
    },
    {
      "epoch": 0.022809654318804704,
      "grad_norm": 3.811927556991577,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.5824,
      "step": 129
    },
    {
      "epoch": 0.022986473344531872,
      "grad_norm": 3.9410016536712646,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.8725,
      "step": 130
    },
    {
      "epoch": 0.02316329237025904,
      "grad_norm": 3.8563146591186523,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.5051,
      "step": 131
    },
    {
      "epoch": 0.02334011139598621,
      "grad_norm": 3.753028392791748,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.6026,
      "step": 132
    },
    {
      "epoch": 0.023516930421713378,
      "grad_norm": 3.807447910308838,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.4886,
      "step": 133
    },
    {
      "epoch": 0.023693749447440543,
      "grad_norm": 3.8522281646728516,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.5633,
      "step": 134
    },
    {
      "epoch": 0.02387056847316771,
      "grad_norm": 3.8721601963043213,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.5071,
      "step": 135
    },
    {
      "epoch": 0.02404738749889488,
      "grad_norm": 4.0354461669921875,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.7198,
      "step": 136
    },
    {
      "epoch": 0.02422420652462205,
      "grad_norm": 3.5654656887054443,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.4604,
      "step": 137
    },
    {
      "epoch": 0.024401025550349217,
      "grad_norm": 3.723776340484619,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.6183,
      "step": 138
    },
    {
      "epoch": 0.024577844576076386,
      "grad_norm": 3.5163791179656982,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.4345,
      "step": 139
    },
    {
      "epoch": 0.024754663601803555,
      "grad_norm": 3.7013096809387207,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.2734,
      "step": 140
    },
    {
      "epoch": 0.024931482627530723,
      "grad_norm": 3.5966262817382812,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.4084,
      "step": 141
    },
    {
      "epoch": 0.025108301653257892,
      "grad_norm": 3.8784823417663574,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.5505,
      "step": 142
    },
    {
      "epoch": 0.02528512067898506,
      "grad_norm": 3.960005760192871,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.3651,
      "step": 143
    },
    {
      "epoch": 0.025461939704712225,
      "grad_norm": 3.905780792236328,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.4459,
      "step": 144
    },
    {
      "epoch": 0.025638758730439394,
      "grad_norm": 3.7308545112609863,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.3803,
      "step": 145
    },
    {
      "epoch": 0.025815577756166563,
      "grad_norm": 4.277618408203125,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.5735,
      "step": 146
    },
    {
      "epoch": 0.02599239678189373,
      "grad_norm": 4.1843180656433105,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.5866,
      "step": 147
    },
    {
      "epoch": 0.0261692158076209,
      "grad_norm": 3.8520991802215576,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.2892,
      "step": 148
    },
    {
      "epoch": 0.02634603483334807,
      "grad_norm": 3.938481330871582,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.2478,
      "step": 149
    },
    {
      "epoch": 0.026522853859075237,
      "grad_norm": 4.657419681549072,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.4547,
      "step": 150
    },
    {
      "epoch": 0.026522853859075237,
      "eval_loss": 0.6781367659568787,
      "eval_runtime": 703.6237,
      "eval_samples_per_second": 13.537,
      "eval_steps_per_second": 3.385,
      "step": 150
    },
    {
      "epoch": 0.026699672884802406,
      "grad_norm": 6.747509479522705,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.3621,
      "step": 151
    },
    {
      "epoch": 0.026876491910529574,
      "grad_norm": 6.714910984039307,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 3.2051,
      "step": 152
    },
    {
      "epoch": 0.027053310936256743,
      "grad_norm": 6.101622104644775,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.1115,
      "step": 153
    },
    {
      "epoch": 0.027230129961983908,
      "grad_norm": 5.840374946594238,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.9658,
      "step": 154
    },
    {
      "epoch": 0.027406948987711077,
      "grad_norm": 4.568004131317139,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.7909,
      "step": 155
    },
    {
      "epoch": 0.027583768013438245,
      "grad_norm": 4.027502059936523,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.7051,
      "step": 156
    },
    {
      "epoch": 0.027760587039165414,
      "grad_norm": 3.8282158374786377,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.8342,
      "step": 157
    },
    {
      "epoch": 0.027937406064892582,
      "grad_norm": 3.634490489959717,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.6897,
      "step": 158
    },
    {
      "epoch": 0.02811422509061975,
      "grad_norm": 3.798839569091797,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.7565,
      "step": 159
    },
    {
      "epoch": 0.02829104411634692,
      "grad_norm": 3.824894666671753,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.9382,
      "step": 160
    },
    {
      "epoch": 0.028467863142074088,
      "grad_norm": 3.9101107120513916,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.8548,
      "step": 161
    },
    {
      "epoch": 0.028644682167801257,
      "grad_norm": 3.989967107772827,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.7325,
      "step": 162
    },
    {
      "epoch": 0.028821501193528425,
      "grad_norm": 4.159213542938232,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.9587,
      "step": 163
    },
    {
      "epoch": 0.02899832021925559,
      "grad_norm": 3.9716978073120117,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.7947,
      "step": 164
    },
    {
      "epoch": 0.02917513924498276,
      "grad_norm": 3.726034641265869,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.6536,
      "step": 165
    },
    {
      "epoch": 0.029351958270709928,
      "grad_norm": 3.7833333015441895,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.6872,
      "step": 166
    },
    {
      "epoch": 0.029528777296437096,
      "grad_norm": 3.8764686584472656,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.7443,
      "step": 167
    },
    {
      "epoch": 0.029705596322164265,
      "grad_norm": 3.609280824661255,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.6044,
      "step": 168
    },
    {
      "epoch": 0.029882415347891433,
      "grad_norm": 4.007650852203369,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.7816,
      "step": 169
    },
    {
      "epoch": 0.030059234373618602,
      "grad_norm": 3.7304792404174805,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.663,
      "step": 170
    },
    {
      "epoch": 0.03023605339934577,
      "grad_norm": 3.7628748416900635,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.6515,
      "step": 171
    },
    {
      "epoch": 0.03041287242507294,
      "grad_norm": 3.6568541526794434,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.651,
      "step": 172
    },
    {
      "epoch": 0.030589691450800108,
      "grad_norm": 3.604257822036743,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.5193,
      "step": 173
    },
    {
      "epoch": 0.030766510476527273,
      "grad_norm": 3.6266162395477295,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.5014,
      "step": 174
    },
    {
      "epoch": 0.03094332950225444,
      "grad_norm": 3.808096170425415,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.7023,
      "step": 175
    },
    {
      "epoch": 0.03112014852798161,
      "grad_norm": 3.9304137229919434,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.666,
      "step": 176
    },
    {
      "epoch": 0.03129696755370878,
      "grad_norm": 3.6244189739227295,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.4847,
      "step": 177
    },
    {
      "epoch": 0.03147378657943595,
      "grad_norm": 3.5086371898651123,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5285,
      "step": 178
    },
    {
      "epoch": 0.03165060560516311,
      "grad_norm": 3.7354354858398438,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.5621,
      "step": 179
    },
    {
      "epoch": 0.031827424630890284,
      "grad_norm": 3.7806384563446045,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.6966,
      "step": 180
    },
    {
      "epoch": 0.03200424365661745,
      "grad_norm": 3.9345269203186035,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5602,
      "step": 181
    },
    {
      "epoch": 0.03218106268234462,
      "grad_norm": 3.4946577548980713,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.5295,
      "step": 182
    },
    {
      "epoch": 0.03235788170807179,
      "grad_norm": 3.708465099334717,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.5623,
      "step": 183
    },
    {
      "epoch": 0.03253470073379896,
      "grad_norm": 3.928293228149414,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.635,
      "step": 184
    },
    {
      "epoch": 0.032711519759526124,
      "grad_norm": 3.5179319381713867,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.3589,
      "step": 185
    },
    {
      "epoch": 0.032888338785253296,
      "grad_norm": 3.815838098526001,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.5558,
      "step": 186
    },
    {
      "epoch": 0.03306515781098046,
      "grad_norm": 3.9550325870513916,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.6194,
      "step": 187
    },
    {
      "epoch": 0.03324197683670763,
      "grad_norm": 3.733525276184082,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.358,
      "step": 188
    },
    {
      "epoch": 0.0334187958624348,
      "grad_norm": 3.733062744140625,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.535,
      "step": 189
    },
    {
      "epoch": 0.03359561488816196,
      "grad_norm": 3.500577926635742,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.3805,
      "step": 190
    },
    {
      "epoch": 0.033772433913889136,
      "grad_norm": 3.7395260334014893,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.5822,
      "step": 191
    },
    {
      "epoch": 0.0339492529396163,
      "grad_norm": 3.801841974258423,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.3383,
      "step": 192
    },
    {
      "epoch": 0.03412607196534347,
      "grad_norm": 3.66439151763916,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.3541,
      "step": 193
    },
    {
      "epoch": 0.03430289099107064,
      "grad_norm": 3.8568992614746094,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.5269,
      "step": 194
    },
    {
      "epoch": 0.03447971001679781,
      "grad_norm": 3.8315274715423584,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.3952,
      "step": 195
    },
    {
      "epoch": 0.034656529042524975,
      "grad_norm": 3.829559326171875,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.4035,
      "step": 196
    },
    {
      "epoch": 0.03483334806825215,
      "grad_norm": 4.215095520019531,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.5518,
      "step": 197
    },
    {
      "epoch": 0.03501016709397931,
      "grad_norm": 4.245261192321777,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.3925,
      "step": 198
    },
    {
      "epoch": 0.03518698611970648,
      "grad_norm": 4.286358833312988,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.5002,
      "step": 199
    },
    {
      "epoch": 0.03536380514543365,
      "grad_norm": 5.465405464172363,
      "learning_rate": 0.0,
      "loss": 2.6677,
      "step": 200
    },
    {
      "epoch": 0.03536380514543365,
      "eval_loss": 0.6426064968109131,
      "eval_runtime": 703.4443,
      "eval_samples_per_second": 13.541,
      "eval_steps_per_second": 3.386,
      "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": 2.861999239200768e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}