{
  "best_metric": 2.1752114295959473,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.05201222287237501,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000260061114361875,
      "grad_norm": 11.279027938842773,
      "learning_rate": 1e-05,
      "loss": 3.8545,
      "step": 1
    },
    {
      "epoch": 0.000260061114361875,
      "eval_loss": 4.0225372314453125,
      "eval_runtime": 226.4124,
      "eval_samples_per_second": 28.607,
      "eval_steps_per_second": 7.155,
      "step": 1
    },
    {
      "epoch": 0.00052012222872375,
      "grad_norm": 12.146949768066406,
      "learning_rate": 2e-05,
      "loss": 4.1702,
      "step": 2
    },
    {
      "epoch": 0.0007801833430856252,
      "grad_norm": 12.14774227142334,
      "learning_rate": 3e-05,
      "loss": 3.9964,
      "step": 3
    },
    {
      "epoch": 0.0010402444574475,
      "grad_norm": 9.804224014282227,
      "learning_rate": 4e-05,
      "loss": 3.592,
      "step": 4
    },
    {
      "epoch": 0.001300305571809375,
      "grad_norm": 7.959303379058838,
      "learning_rate": 5e-05,
      "loss": 3.2399,
      "step": 5
    },
    {
      "epoch": 0.0015603666861712503,
      "grad_norm": 5.706275939941406,
      "learning_rate": 6e-05,
      "loss": 3.0075,
      "step": 6
    },
    {
      "epoch": 0.0018204278005331254,
      "grad_norm": 4.102785110473633,
      "learning_rate": 7e-05,
      "loss": 2.9418,
      "step": 7
    },
    {
      "epoch": 0.002080488914895,
      "grad_norm": 4.174167156219482,
      "learning_rate": 8e-05,
      "loss": 2.7187,
      "step": 8
    },
    {
      "epoch": 0.0023405500292568754,
      "grad_norm": 4.352199554443359,
      "learning_rate": 9e-05,
      "loss": 2.4824,
      "step": 9
    },
    {
      "epoch": 0.00260061114361875,
      "grad_norm": 3.7645320892333984,
      "learning_rate": 0.0001,
      "loss": 2.6324,
      "step": 10
    },
    {
      "epoch": 0.0028606722579806254,
      "grad_norm": 3.2619616985321045,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.5213,
      "step": 11
    },
    {
      "epoch": 0.0031207333723425007,
      "grad_norm": 3.8362362384796143,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.546,
      "step": 12
    },
    {
      "epoch": 0.0033807944867043755,
      "grad_norm": 3.2334647178649902,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.3918,
      "step": 13
    },
    {
      "epoch": 0.0036408556010662507,
      "grad_norm": 2.9965999126434326,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.516,
      "step": 14
    },
    {
      "epoch": 0.0039009167154281255,
      "grad_norm": 2.900367021560669,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.5994,
      "step": 15
    },
    {
      "epoch": 0.00416097782979,
      "grad_norm": 3.021207094192505,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.468,
      "step": 16
    },
    {
      "epoch": 0.004421038944151876,
      "grad_norm": 2.7395761013031006,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.4441,
      "step": 17
    },
    {
      "epoch": 0.004681100058513751,
      "grad_norm": 2.5328333377838135,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.438,
      "step": 18
    },
    {
      "epoch": 0.004941161172875626,
      "grad_norm": 2.5700788497924805,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.4234,
      "step": 19
    },
    {
      "epoch": 0.0052012222872375,
      "grad_norm": 2.506720542907715,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.1949,
      "step": 20
    },
    {
      "epoch": 0.005461283401599376,
      "grad_norm": 2.8594417572021484,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.5811,
      "step": 21
    },
    {
      "epoch": 0.005721344515961251,
      "grad_norm": 2.623512029647827,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.4576,
      "step": 22
    },
    {
      "epoch": 0.005981405630323126,
      "grad_norm": 2.5077295303344727,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.4548,
      "step": 23
    },
    {
      "epoch": 0.006241466744685001,
      "grad_norm": 2.7206809520721436,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.3755,
      "step": 24
    },
    {
      "epoch": 0.006501527859046876,
      "grad_norm": 2.7397067546844482,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.2536,
      "step": 25
    },
    {
      "epoch": 0.006761588973408751,
      "grad_norm": 2.648794412612915,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.2745,
      "step": 26
    },
    {
      "epoch": 0.007021650087770626,
      "grad_norm": 2.612983465194702,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.3835,
      "step": 27
    },
    {
      "epoch": 0.0072817112021325014,
      "grad_norm": 2.662501811981201,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.3312,
      "step": 28
    },
    {
      "epoch": 0.007541772316494376,
      "grad_norm": 2.3952796459198,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.2493,
      "step": 29
    },
    {
      "epoch": 0.007801833430856251,
      "grad_norm": 2.3468735218048096,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.179,
      "step": 30
    },
    {
      "epoch": 0.008061894545218126,
      "grad_norm": 2.5587215423583984,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2989,
      "step": 31
    },
    {
      "epoch": 0.00832195565958,
      "grad_norm": 2.3676464557647705,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.2952,
      "step": 32
    },
    {
      "epoch": 0.008582016773941877,
      "grad_norm": 2.4102590084075928,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.233,
      "step": 33
    },
    {
      "epoch": 0.008842077888303752,
      "grad_norm": 2.449181079864502,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.2371,
      "step": 34
    },
    {
      "epoch": 0.009102139002665627,
      "grad_norm": 2.5043232440948486,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.2633,
      "step": 35
    },
    {
      "epoch": 0.009362200117027502,
      "grad_norm": 2.6106226444244385,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.3006,
      "step": 36
    },
    {
      "epoch": 0.009622261231389376,
      "grad_norm": 2.48602294921875,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.3724,
      "step": 37
    },
    {
      "epoch": 0.009882322345751251,
      "grad_norm": 2.424474000930786,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.1156,
      "step": 38
    },
    {
      "epoch": 0.010142383460113126,
      "grad_norm": 2.571842670440674,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.4742,
      "step": 39
    },
    {
      "epoch": 0.010402444574475,
      "grad_norm": 2.3028788566589355,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.3561,
      "step": 40
    },
    {
      "epoch": 0.010662505688836877,
      "grad_norm": 2.419778823852539,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.212,
      "step": 41
    },
    {
      "epoch": 0.010922566803198752,
      "grad_norm": 2.5612616539001465,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.5184,
      "step": 42
    },
    {
      "epoch": 0.011182627917560627,
      "grad_norm": 2.2547926902770996,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.091,
      "step": 43
    },
    {
      "epoch": 0.011442689031922502,
      "grad_norm": 2.205031156539917,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.2702,
      "step": 44
    },
    {
      "epoch": 0.011702750146284377,
      "grad_norm": 2.261950969696045,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.1154,
      "step": 45
    },
    {
      "epoch": 0.011962811260646251,
      "grad_norm": 2.453289747238159,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.4655,
      "step": 46
    },
    {
      "epoch": 0.012222872375008126,
      "grad_norm": 2.325051784515381,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.0918,
      "step": 47
    },
    {
      "epoch": 0.012482933489370003,
      "grad_norm": 2.4815924167633057,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.2778,
      "step": 48
    },
    {
      "epoch": 0.012742994603731878,
      "grad_norm": 2.5599594116210938,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.4497,
      "step": 49
    },
    {
      "epoch": 0.013003055718093752,
      "grad_norm": 2.5239109992980957,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.445,
      "step": 50
    },
    {
      "epoch": 0.013003055718093752,
      "eval_loss": 2.2580556869506836,
      "eval_runtime": 227.6834,
      "eval_samples_per_second": 28.447,
      "eval_steps_per_second": 7.115,
      "step": 50
    },
    {
      "epoch": 0.013263116832455627,
      "grad_norm": 2.1725056171417236,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.0836,
      "step": 51
    },
    {
      "epoch": 0.013523177946817502,
      "grad_norm": 2.151031255722046,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.1933,
      "step": 52
    },
    {
      "epoch": 0.013783239061179377,
      "grad_norm": 2.312286376953125,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.2838,
      "step": 53
    },
    {
      "epoch": 0.014043300175541252,
      "grad_norm": 2.3974902629852295,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.3486,
      "step": 54
    },
    {
      "epoch": 0.014303361289903128,
      "grad_norm": 2.347676992416382,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1798,
      "step": 55
    },
    {
      "epoch": 0.014563422404265003,
      "grad_norm": 2.500965118408203,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.3955,
      "step": 56
    },
    {
      "epoch": 0.014823483518626878,
      "grad_norm": 2.5651566982269287,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.1996,
      "step": 57
    },
    {
      "epoch": 0.015083544632988753,
      "grad_norm": 2.4082300662994385,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.2885,
      "step": 58
    },
    {
      "epoch": 0.015343605747350627,
      "grad_norm": 2.3121886253356934,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.3102,
      "step": 59
    },
    {
      "epoch": 0.015603666861712502,
      "grad_norm": 2.371232032775879,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.4004,
      "step": 60
    },
    {
      "epoch": 0.01586372797607438,
      "grad_norm": 2.2909135818481445,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.1355,
      "step": 61
    },
    {
      "epoch": 0.01612378909043625,
      "grad_norm": 2.3039820194244385,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.092,
      "step": 62
    },
    {
      "epoch": 0.016383850204798128,
      "grad_norm": 2.4646389484405518,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.2632,
      "step": 63
    },
    {
      "epoch": 0.01664391131916,
      "grad_norm": 2.2102277278900146,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.1889,
      "step": 64
    },
    {
      "epoch": 0.016903972433521878,
      "grad_norm": 2.2069945335388184,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.2429,
      "step": 65
    },
    {
      "epoch": 0.017164033547883754,
      "grad_norm": 2.3962416648864746,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.1844,
      "step": 66
    },
    {
      "epoch": 0.017424094662245627,
      "grad_norm": 2.281733751296997,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.2889,
      "step": 67
    },
    {
      "epoch": 0.017684155776607504,
      "grad_norm": 2.2765684127807617,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.1344,
      "step": 68
    },
    {
      "epoch": 0.017944216890969377,
      "grad_norm": 2.3645732402801514,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.3186,
      "step": 69
    },
    {
      "epoch": 0.018204278005331254,
      "grad_norm": 2.2298169136047363,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.2085,
      "step": 70
    },
    {
      "epoch": 0.018464339119693127,
      "grad_norm": 2.210784912109375,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.1992,
      "step": 71
    },
    {
      "epoch": 0.018724400234055003,
      "grad_norm": 2.1122100353240967,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.1373,
      "step": 72
    },
    {
      "epoch": 0.018984461348416876,
      "grad_norm": 2.3805735111236572,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.2175,
      "step": 73
    },
    {
      "epoch": 0.019244522462778753,
      "grad_norm": 2.3300492763519287,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.2024,
      "step": 74
    },
    {
      "epoch": 0.01950458357714063,
      "grad_norm": 2.2528269290924072,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.2647,
      "step": 75
    },
    {
      "epoch": 0.019764644691502502,
      "grad_norm": 2.152442693710327,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.074,
      "step": 76
    },
    {
      "epoch": 0.02002470580586438,
      "grad_norm": 2.3002822399139404,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.1505,
      "step": 77
    },
    {
      "epoch": 0.020284766920226252,
      "grad_norm": 2.270409107208252,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.2814,
      "step": 78
    },
    {
      "epoch": 0.02054482803458813,
      "grad_norm": 2.2066092491149902,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.2664,
      "step": 79
    },
    {
      "epoch": 0.02080488914895,
      "grad_norm": 2.364281177520752,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.283,
      "step": 80
    },
    {
      "epoch": 0.021064950263311878,
      "grad_norm": 2.304248332977295,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.1214,
      "step": 81
    },
    {
      "epoch": 0.021325011377673755,
      "grad_norm": 2.186495065689087,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.0971,
      "step": 82
    },
    {
      "epoch": 0.021585072492035628,
      "grad_norm": 2.2885663509368896,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.1067,
      "step": 83
    },
    {
      "epoch": 0.021845133606397504,
      "grad_norm": 2.321530818939209,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.2285,
      "step": 84
    },
    {
      "epoch": 0.022105194720759377,
      "grad_norm": 2.183454751968384,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.0823,
      "step": 85
    },
    {
      "epoch": 0.022365255835121254,
      "grad_norm": 2.2079780101776123,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.2924,
      "step": 86
    },
    {
      "epoch": 0.022625316949483127,
      "grad_norm": 2.193880319595337,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.024,
      "step": 87
    },
    {
      "epoch": 0.022885378063845004,
      "grad_norm": 2.5025057792663574,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.3183,
      "step": 88
    },
    {
      "epoch": 0.02314543917820688,
      "grad_norm": 2.3934152126312256,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.1079,
      "step": 89
    },
    {
      "epoch": 0.023405500292568753,
      "grad_norm": 2.3551230430603027,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.2039,
      "step": 90
    },
    {
      "epoch": 0.02366556140693063,
      "grad_norm": 2.217134952545166,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.1485,
      "step": 91
    },
    {
      "epoch": 0.023925622521292503,
      "grad_norm": 2.3489155769348145,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.2276,
      "step": 92
    },
    {
      "epoch": 0.02418568363565438,
      "grad_norm": 2.3202967643737793,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.0705,
      "step": 93
    },
    {
      "epoch": 0.024445744750016252,
      "grad_norm": 2.356043815612793,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.2374,
      "step": 94
    },
    {
      "epoch": 0.02470580586437813,
      "grad_norm": 2.2478909492492676,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.9617,
      "step": 95
    },
    {
      "epoch": 0.024965866978740005,
      "grad_norm": 2.2553298473358154,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.0932,
      "step": 96
    },
    {
      "epoch": 0.02522592809310188,
      "grad_norm": 2.3768444061279297,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.2258,
      "step": 97
    },
    {
      "epoch": 0.025485989207463755,
      "grad_norm": 2.224522590637207,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.0418,
      "step": 98
    },
    {
      "epoch": 0.025746050321825628,
      "grad_norm": 2.294828414916992,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.1561,
      "step": 99
    },
    {
      "epoch": 0.026006111436187505,
      "grad_norm": 2.517188787460327,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.3157,
      "step": 100
    },
    {
      "epoch": 0.026006111436187505,
      "eval_loss": 2.2096688747406006,
      "eval_runtime": 227.6104,
      "eval_samples_per_second": 28.457,
      "eval_steps_per_second": 7.117,
      "step": 100
    },
    {
      "epoch": 0.026266172550549378,
      "grad_norm": 2.1049461364746094,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.1924,
      "step": 101
    },
    {
      "epoch": 0.026526233664911254,
      "grad_norm": 2.2002434730529785,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.1416,
      "step": 102
    },
    {
      "epoch": 0.02678629477927313,
      "grad_norm": 2.338472843170166,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.3528,
      "step": 103
    },
    {
      "epoch": 0.027046355893635004,
      "grad_norm": 2.228801727294922,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.2065,
      "step": 104
    },
    {
      "epoch": 0.02730641700799688,
      "grad_norm": 2.0818748474121094,
      "learning_rate": 5e-05,
      "loss": 2.1229,
      "step": 105
    },
    {
      "epoch": 0.027566478122358753,
      "grad_norm": 2.0052101612091064,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.027,
      "step": 106
    },
    {
      "epoch": 0.02782653923672063,
      "grad_norm": 1.970994472503662,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.0914,
      "step": 107
    },
    {
      "epoch": 0.028086600351082503,
      "grad_norm": 2.210747480392456,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.166,
      "step": 108
    },
    {
      "epoch": 0.02834666146544438,
      "grad_norm": 2.1865952014923096,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.1483,
      "step": 109
    },
    {
      "epoch": 0.028606722579806256,
      "grad_norm": 2.214153528213501,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.1773,
      "step": 110
    },
    {
      "epoch": 0.02886678369416813,
      "grad_norm": 2.086395263671875,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.111,
      "step": 111
    },
    {
      "epoch": 0.029126844808530006,
      "grad_norm": 2.1963107585906982,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.1772,
      "step": 112
    },
    {
      "epoch": 0.02938690592289188,
      "grad_norm": 2.156548023223877,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.9825,
      "step": 113
    },
    {
      "epoch": 0.029646967037253755,
      "grad_norm": 2.24280047416687,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.2053,
      "step": 114
    },
    {
      "epoch": 0.02990702815161563,
      "grad_norm": 2.130927562713623,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.0353,
      "step": 115
    },
    {
      "epoch": 0.030167089265977505,
      "grad_norm": 2.3154146671295166,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.4269,
      "step": 116
    },
    {
      "epoch": 0.030427150380339378,
      "grad_norm": 2.136002779006958,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.2096,
      "step": 117
    },
    {
      "epoch": 0.030687211494701255,
      "grad_norm": 2.1550419330596924,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.1687,
      "step": 118
    },
    {
      "epoch": 0.03094727260906313,
      "grad_norm": 2.0902926921844482,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.281,
      "step": 119
    },
    {
      "epoch": 0.031207333723425004,
      "grad_norm": 2.546315908432007,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.0914,
      "step": 120
    },
    {
      "epoch": 0.03146739483778688,
      "grad_norm": 2.1348767280578613,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.2393,
      "step": 121
    },
    {
      "epoch": 0.03172745595214876,
      "grad_norm": 2.077097177505493,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.1815,
      "step": 122
    },
    {
      "epoch": 0.03198751706651063,
      "grad_norm": 2.316852331161499,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.1184,
      "step": 123
    },
    {
      "epoch": 0.0322475781808725,
      "grad_norm": 2.2091751098632812,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.2969,
      "step": 124
    },
    {
      "epoch": 0.03250763929523438,
      "grad_norm": 2.3899569511413574,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.1966,
      "step": 125
    },
    {
      "epoch": 0.032767700409596257,
      "grad_norm": 2.2248878479003906,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.1744,
      "step": 126
    },
    {
      "epoch": 0.03302776152395813,
      "grad_norm": 2.2646782398223877,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.1605,
      "step": 127
    },
    {
      "epoch": 0.03328782263832,
      "grad_norm": 2.08091139793396,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.228,
      "step": 128
    },
    {
      "epoch": 0.03354788375268188,
      "grad_norm": 2.125476121902466,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.9854,
      "step": 129
    },
    {
      "epoch": 0.033807944867043756,
      "grad_norm": 2.0864744186401367,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.983,
      "step": 130
    },
    {
      "epoch": 0.03406800598140563,
      "grad_norm": 2.199686050415039,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.2645,
      "step": 131
    },
    {
      "epoch": 0.03432806709576751,
      "grad_norm": 2.2540361881256104,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.2981,
      "step": 132
    },
    {
      "epoch": 0.03458812821012938,
      "grad_norm": 2.2063488960266113,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.1369,
      "step": 133
    },
    {
      "epoch": 0.034848189324491255,
      "grad_norm": 2.3302245140075684,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.3653,
      "step": 134
    },
    {
      "epoch": 0.03510825043885313,
      "grad_norm": 2.8641624450683594,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.2025,
      "step": 135
    },
    {
      "epoch": 0.03536831155321501,
      "grad_norm": 2.0843074321746826,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.1321,
      "step": 136
    },
    {
      "epoch": 0.03562837266757688,
      "grad_norm": 2.185920000076294,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.1125,
      "step": 137
    },
    {
      "epoch": 0.035888433781938754,
      "grad_norm": 2.1308934688568115,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.1057,
      "step": 138
    },
    {
      "epoch": 0.03614849489630063,
      "grad_norm": 2.190056562423706,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.088,
      "step": 139
    },
    {
      "epoch": 0.03640855601066251,
      "grad_norm": 2.091644048690796,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.1628,
      "step": 140
    },
    {
      "epoch": 0.03666861712502438,
      "grad_norm": 2.146914482116699,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.0045,
      "step": 141
    },
    {
      "epoch": 0.03692867823938625,
      "grad_norm": 2.2308313846588135,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.0814,
      "step": 142
    },
    {
      "epoch": 0.03718873935374813,
      "grad_norm": 2.394566774368286,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.2715,
      "step": 143
    },
    {
      "epoch": 0.037448800468110006,
      "grad_norm": 2.3659446239471436,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.1684,
      "step": 144
    },
    {
      "epoch": 0.03770886158247188,
      "grad_norm": 2.3128445148468018,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.0926,
      "step": 145
    },
    {
      "epoch": 0.03796892269683375,
      "grad_norm": 2.3206710815429688,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.2095,
      "step": 146
    },
    {
      "epoch": 0.03822898381119563,
      "grad_norm": 2.2718405723571777,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.1025,
      "step": 147
    },
    {
      "epoch": 0.038489044925557506,
      "grad_norm": 2.218853712081909,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0113,
      "step": 148
    },
    {
      "epoch": 0.03874910603991938,
      "grad_norm": 2.686237335205078,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.3147,
      "step": 149
    },
    {
      "epoch": 0.03900916715428126,
      "grad_norm": 2.6437342166900635,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.535,
      "step": 150
    },
    {
      "epoch": 0.03900916715428126,
      "eval_loss": 2.181919574737549,
      "eval_runtime": 227.6287,
      "eval_samples_per_second": 28.454,
      "eval_steps_per_second": 7.117,
      "step": 150
    },
    {
      "epoch": 0.03926922826864313,
      "grad_norm": 2.149697780609131,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.0546,
      "step": 151
    },
    {
      "epoch": 0.039529289383005005,
      "grad_norm": 2.1155319213867188,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.0647,
      "step": 152
    },
    {
      "epoch": 0.03978935049736688,
      "grad_norm": 2.1451268196105957,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.1496,
      "step": 153
    },
    {
      "epoch": 0.04004941161172876,
      "grad_norm": 2.4642882347106934,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.0772,
      "step": 154
    },
    {
      "epoch": 0.04030947272609063,
      "grad_norm": 2.1916542053222656,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.2441,
      "step": 155
    },
    {
      "epoch": 0.040569533840452504,
      "grad_norm": 2.1746671199798584,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.1504,
      "step": 156
    },
    {
      "epoch": 0.040829594954814384,
      "grad_norm": 2.157935857772827,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.107,
      "step": 157
    },
    {
      "epoch": 0.04108965606917626,
      "grad_norm": 2.062431812286377,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.081,
      "step": 158
    },
    {
      "epoch": 0.04134971718353813,
      "grad_norm": 2.142822027206421,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.0843,
      "step": 159
    },
    {
      "epoch": 0.0416097782979,
      "grad_norm": 2.1664884090423584,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.0942,
      "step": 160
    },
    {
      "epoch": 0.04186983941226188,
      "grad_norm": 2.2922542095184326,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.205,
      "step": 161
    },
    {
      "epoch": 0.042129900526623756,
      "grad_norm": 2.0822224617004395,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.1184,
      "step": 162
    },
    {
      "epoch": 0.04238996164098563,
      "grad_norm": 2.0957558155059814,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.0645,
      "step": 163
    },
    {
      "epoch": 0.04265002275534751,
      "grad_norm": 2.269651174545288,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.1873,
      "step": 164
    },
    {
      "epoch": 0.04291008386970938,
      "grad_norm": 2.290019989013672,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.2227,
      "step": 165
    },
    {
      "epoch": 0.043170144984071256,
      "grad_norm": 2.2755532264709473,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.2698,
      "step": 166
    },
    {
      "epoch": 0.04343020609843313,
      "grad_norm": 2.457045316696167,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.1578,
      "step": 167
    },
    {
      "epoch": 0.04369026721279501,
      "grad_norm": 2.1705782413482666,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.1767,
      "step": 168
    },
    {
      "epoch": 0.04395032832715688,
      "grad_norm": 2.231419563293457,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.2779,
      "step": 169
    },
    {
      "epoch": 0.044210389441518755,
      "grad_norm": 2.3254246711730957,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.2079,
      "step": 170
    },
    {
      "epoch": 0.044470450555880635,
      "grad_norm": 2.2333338260650635,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.1548,
      "step": 171
    },
    {
      "epoch": 0.04473051167024251,
      "grad_norm": 2.515716791152954,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.1873,
      "step": 172
    },
    {
      "epoch": 0.04499057278460438,
      "grad_norm": 2.3713366985321045,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.1824,
      "step": 173
    },
    {
      "epoch": 0.045250633898966254,
      "grad_norm": 2.3272628784179688,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.212,
      "step": 174
    },
    {
      "epoch": 0.045510695013328134,
      "grad_norm": 2.1713898181915283,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.127,
      "step": 175
    },
    {
      "epoch": 0.04577075612769001,
      "grad_norm": 2.2558517456054688,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2112,
      "step": 176
    },
    {
      "epoch": 0.04603081724205188,
      "grad_norm": 2.3417723178863525,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.0578,
      "step": 177
    },
    {
      "epoch": 0.04629087835641376,
      "grad_norm": 2.1906392574310303,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.2454,
      "step": 178
    },
    {
      "epoch": 0.04655093947077563,
      "grad_norm": 2.1303303241729736,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.9599,
      "step": 179
    },
    {
      "epoch": 0.046811000585137506,
      "grad_norm": 2.1788878440856934,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.022,
      "step": 180
    },
    {
      "epoch": 0.04707106169949938,
      "grad_norm": 2.2112619876861572,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.2099,
      "step": 181
    },
    {
      "epoch": 0.04733112281386126,
      "grad_norm": 2.2090134620666504,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1532,
      "step": 182
    },
    {
      "epoch": 0.04759118392822313,
      "grad_norm": 2.212982416152954,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.1756,
      "step": 183
    },
    {
      "epoch": 0.047851245042585006,
      "grad_norm": 2.3446829319000244,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.2293,
      "step": 184
    },
    {
      "epoch": 0.048111306156946886,
      "grad_norm": 2.1704304218292236,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.9746,
      "step": 185
    },
    {
      "epoch": 0.04837136727130876,
      "grad_norm": 2.264618158340454,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.2944,
      "step": 186
    },
    {
      "epoch": 0.04863142838567063,
      "grad_norm": 2.402517080307007,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.1472,
      "step": 187
    },
    {
      "epoch": 0.048891489500032505,
      "grad_norm": 2.2531824111938477,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.0401,
      "step": 188
    },
    {
      "epoch": 0.049151550614394385,
      "grad_norm": 2.224325656890869,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.0878,
      "step": 189
    },
    {
      "epoch": 0.04941161172875626,
      "grad_norm": 2.3533246517181396,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.0941,
      "step": 190
    },
    {
      "epoch": 0.04967167284311813,
      "grad_norm": 2.397167921066284,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.1501,
      "step": 191
    },
    {
      "epoch": 0.04993173395748001,
      "grad_norm": 2.233166456222534,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.0635,
      "step": 192
    },
    {
      "epoch": 0.050191795071841884,
      "grad_norm": 2.260296583175659,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.0526,
      "step": 193
    },
    {
      "epoch": 0.05045185618620376,
      "grad_norm": 2.3124306201934814,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.136,
      "step": 194
    },
    {
      "epoch": 0.05071191730056563,
      "grad_norm": 2.3403406143188477,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.1909,
      "step": 195
    },
    {
      "epoch": 0.05097197841492751,
      "grad_norm": 2.357811212539673,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.2516,
      "step": 196
    },
    {
      "epoch": 0.05123203952928938,
      "grad_norm": 2.270812511444092,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.1596,
      "step": 197
    },
    {
      "epoch": 0.051492100643651256,
      "grad_norm": 2.2893905639648438,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.1138,
      "step": 198
    },
    {
      "epoch": 0.051752161758013136,
      "grad_norm": 2.342320442199707,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.3086,
      "step": 199
    },
    {
      "epoch": 0.05201222287237501,
      "grad_norm": 2.4791676998138428,
      "learning_rate": 0.0,
      "loss": 2.4156,
      "step": 200
    },
    {
      "epoch": 0.05201222287237501,
      "eval_loss": 2.1752114295959473,
      "eval_runtime": 227.536,
      "eval_samples_per_second": 28.466,
      "eval_steps_per_second": 7.12,
      "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": 8.10155117641728e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}