{
  "best_metric": 10.234081268310547,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.11245431543435479,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000562271577171774,
      "grad_norm": 0.12471116334199905,
      "learning_rate": 1e-05,
      "loss": 10.3801,
      "step": 1
    },
    {
      "epoch": 0.000562271577171774,
      "eval_loss": 10.378522872924805,
      "eval_runtime": 6.7003,
      "eval_samples_per_second": 446.994,
      "eval_steps_per_second": 111.786,
      "step": 1
    },
    {
      "epoch": 0.001124543154343548,
      "grad_norm": 0.12479763478040695,
      "learning_rate": 2e-05,
      "loss": 10.3789,
      "step": 2
    },
    {
      "epoch": 0.0016868147315153219,
      "grad_norm": 0.11667029559612274,
      "learning_rate": 3e-05,
      "loss": 10.3794,
      "step": 3
    },
    {
      "epoch": 0.002249086308687096,
      "grad_norm": 0.12813054025173187,
      "learning_rate": 4e-05,
      "loss": 10.3798,
      "step": 4
    },
    {
      "epoch": 0.0028113578858588698,
      "grad_norm": 0.12432009726762772,
      "learning_rate": 5e-05,
      "loss": 10.3807,
      "step": 5
    },
    {
      "epoch": 0.0033736294630306437,
      "grad_norm": 0.11905932426452637,
      "learning_rate": 6e-05,
      "loss": 10.3796,
      "step": 6
    },
    {
      "epoch": 0.003935901040202418,
      "grad_norm": 0.12134478241205215,
      "learning_rate": 7e-05,
      "loss": 10.3766,
      "step": 7
    },
    {
      "epoch": 0.004498172617374192,
      "grad_norm": 0.12212009727954865,
      "learning_rate": 8e-05,
      "loss": 10.3773,
      "step": 8
    },
    {
      "epoch": 0.005060444194545966,
      "grad_norm": 0.12545165419578552,
      "learning_rate": 9e-05,
      "loss": 10.377,
      "step": 9
    },
    {
      "epoch": 0.0056227157717177395,
      "grad_norm": 0.12821167707443237,
      "learning_rate": 0.0001,
      "loss": 10.3756,
      "step": 10
    },
    {
      "epoch": 0.006184987348889514,
      "grad_norm": 0.12199673801660538,
      "learning_rate": 9.999316524962345e-05,
      "loss": 10.3744,
      "step": 11
    },
    {
      "epoch": 0.0067472589260612875,
      "grad_norm": 0.14094240963459015,
      "learning_rate": 9.997266286704631e-05,
      "loss": 10.3723,
      "step": 12
    },
    {
      "epoch": 0.007309530503233062,
      "grad_norm": 0.14446423947811127,
      "learning_rate": 9.993849845741524e-05,
      "loss": 10.3725,
      "step": 13
    },
    {
      "epoch": 0.007871802080404836,
      "grad_norm": 0.146043062210083,
      "learning_rate": 9.989068136093873e-05,
      "loss": 10.3703,
      "step": 14
    },
    {
      "epoch": 0.00843407365757661,
      "grad_norm": 0.14884252846240997,
      "learning_rate": 9.98292246503335e-05,
      "loss": 10.37,
      "step": 15
    },
    {
      "epoch": 0.008996345234748383,
      "grad_norm": 0.16853074729442596,
      "learning_rate": 9.975414512725057e-05,
      "loss": 10.3688,
      "step": 16
    },
    {
      "epoch": 0.009558616811920157,
      "grad_norm": 0.16209086775779724,
      "learning_rate": 9.966546331768191e-05,
      "loss": 10.367,
      "step": 17
    },
    {
      "epoch": 0.010120888389091932,
      "grad_norm": 0.179716095328331,
      "learning_rate": 9.956320346634876e-05,
      "loss": 10.3655,
      "step": 18
    },
    {
      "epoch": 0.010683159966263706,
      "grad_norm": 0.17729488015174866,
      "learning_rate": 9.944739353007344e-05,
      "loss": 10.3632,
      "step": 19
    },
    {
      "epoch": 0.011245431543435479,
      "grad_norm": 0.17712004482746124,
      "learning_rate": 9.931806517013612e-05,
      "loss": 10.3628,
      "step": 20
    },
    {
      "epoch": 0.011807703120607253,
      "grad_norm": 0.19017107784748077,
      "learning_rate": 9.917525374361912e-05,
      "loss": 10.361,
      "step": 21
    },
    {
      "epoch": 0.012369974697779028,
      "grad_norm": 0.19334553182125092,
      "learning_rate": 9.901899829374047e-05,
      "loss": 10.3592,
      "step": 22
    },
    {
      "epoch": 0.012932246274950801,
      "grad_norm": 0.20680536329746246,
      "learning_rate": 9.884934153917997e-05,
      "loss": 10.3556,
      "step": 23
    },
    {
      "epoch": 0.013494517852122575,
      "grad_norm": 0.22241535782814026,
      "learning_rate": 9.86663298624003e-05,
      "loss": 10.3532,
      "step": 24
    },
    {
      "epoch": 0.014056789429294348,
      "grad_norm": 0.2301526516675949,
      "learning_rate": 9.847001329696653e-05,
      "loss": 10.3517,
      "step": 25
    },
    {
      "epoch": 0.014619061006466124,
      "grad_norm": 0.21980610489845276,
      "learning_rate": 9.826044551386744e-05,
      "loss": 10.3528,
      "step": 26
    },
    {
      "epoch": 0.015181332583637897,
      "grad_norm": 0.22108517587184906,
      "learning_rate": 9.803768380684242e-05,
      "loss": 10.3497,
      "step": 27
    },
    {
      "epoch": 0.015743604160809672,
      "grad_norm": 0.21716654300689697,
      "learning_rate": 9.780178907671789e-05,
      "loss": 10.3461,
      "step": 28
    },
    {
      "epoch": 0.016305875737981444,
      "grad_norm": 0.21913281083106995,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.344,
      "step": 29
    },
    {
      "epoch": 0.01686814731515322,
      "grad_norm": 0.22199086844921112,
      "learning_rate": 9.729086208503174e-05,
      "loss": 10.34,
      "step": 30
    },
    {
      "epoch": 0.01743041889232499,
      "grad_norm": 0.19991736114025116,
      "learning_rate": 9.701596950580806e-05,
      "loss": 10.3418,
      "step": 31
    },
    {
      "epoch": 0.017992690469496767,
      "grad_norm": 0.20516079664230347,
      "learning_rate": 9.672822322997305e-05,
      "loss": 10.3384,
      "step": 32
    },
    {
      "epoch": 0.018554962046668542,
      "grad_norm": 0.20187082886695862,
      "learning_rate": 9.642770192448536e-05,
      "loss": 10.3351,
      "step": 33
    },
    {
      "epoch": 0.019117233623840314,
      "grad_norm": 0.20600520074367523,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.3357,
      "step": 34
    },
    {
      "epoch": 0.01967950520101209,
      "grad_norm": 0.19619739055633545,
      "learning_rate": 9.578866633275288e-05,
      "loss": 10.3341,
      "step": 35
    },
    {
      "epoch": 0.020241776778183864,
      "grad_norm": 0.19581444561481476,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.3291,
      "step": 36
    },
    {
      "epoch": 0.020804048355355636,
      "grad_norm": 0.18523846566677094,
      "learning_rate": 9.509956150664796e-05,
      "loss": 10.3285,
      "step": 37
    },
    {
      "epoch": 0.02136631993252741,
      "grad_norm": 0.18195948004722595,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.3249,
      "step": 38
    },
    {
      "epoch": 0.021928591509699186,
      "grad_norm": 0.1719619631767273,
      "learning_rate": 9.43611409721806e-05,
      "loss": 10.3242,
      "step": 39
    },
    {
      "epoch": 0.022490863086870958,
      "grad_norm": 0.17031341791152954,
      "learning_rate": 9.397368756032445e-05,
      "loss": 10.3224,
      "step": 40
    },
    {
      "epoch": 0.023053134664042733,
      "grad_norm": 0.17672231793403625,
      "learning_rate": 9.357421218136386e-05,
      "loss": 10.3176,
      "step": 41
    },
    {
      "epoch": 0.023615406241214505,
      "grad_norm": 0.17666015028953552,
      "learning_rate": 9.316282404787871e-05,
      "loss": 10.3184,
      "step": 42
    },
    {
      "epoch": 0.02417767781838628,
      "grad_norm": 0.18161064386367798,
      "learning_rate": 9.273963562927695e-05,
      "loss": 10.3145,
      "step": 43
    },
    {
      "epoch": 0.024739949395558056,
      "grad_norm": 0.17894713580608368,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.3147,
      "step": 44
    },
    {
      "epoch": 0.025302220972729828,
      "grad_norm": 0.18641634285449982,
      "learning_rate": 9.185832391312644e-05,
      "loss": 10.3115,
      "step": 45
    },
    {
      "epoch": 0.025864492549901603,
      "grad_norm": 0.177683487534523,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.3072,
      "step": 46
    },
    {
      "epoch": 0.026426764127073378,
      "grad_norm": 0.19490234553813934,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.3136,
      "step": 47
    },
    {
      "epoch": 0.02698903570424515,
      "grad_norm": 0.17797794938087463,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.3121,
      "step": 48
    },
    {
      "epoch": 0.027551307281416925,
      "grad_norm": 0.18652331829071045,
      "learning_rate": 8.995939984474624e-05,
      "loss": 10.3063,
      "step": 49
    },
    {
      "epoch": 0.028113578858588697,
      "grad_norm": 0.2340773046016693,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.311,
      "step": 50
    },
    {
      "epoch": 0.028113578858588697,
      "eval_loss": 10.307427406311035,
      "eval_runtime": 6.673,
      "eval_samples_per_second": 448.824,
      "eval_steps_per_second": 112.244,
      "step": 50
    },
    {
      "epoch": 0.028675850435760472,
      "grad_norm": 0.1771995574235916,
      "learning_rate": 8.894386393810563e-05,
      "loss": 10.3135,
      "step": 51
    },
    {
      "epoch": 0.029238122012932247,
      "grad_norm": 0.181050106883049,
      "learning_rate": 8.842005554284296e-05,
      "loss": 10.3104,
      "step": 52
    },
    {
      "epoch": 0.02980039359010402,
      "grad_norm": 0.19221681356430054,
      "learning_rate": 8.788574348801675e-05,
      "loss": 10.3078,
      "step": 53
    },
    {
      "epoch": 0.030362665167275794,
      "grad_norm": 0.2057328224182129,
      "learning_rate": 8.73410738492077e-05,
      "loss": 10.3085,
      "step": 54
    },
    {
      "epoch": 0.03092493674444757,
      "grad_norm": 0.21575243771076202,
      "learning_rate": 8.678619553365659e-05,
      "loss": 10.3072,
      "step": 55
    },
    {
      "epoch": 0.031487208321619345,
      "grad_norm": 0.22473597526550293,
      "learning_rate": 8.622126023955446e-05,
      "loss": 10.304,
      "step": 56
    },
    {
      "epoch": 0.03204947989879112,
      "grad_norm": 0.2214728593826294,
      "learning_rate": 8.564642241456986e-05,
      "loss": 10.302,
      "step": 57
    },
    {
      "epoch": 0.03261175147596289,
      "grad_norm": 0.20758339762687683,
      "learning_rate": 8.506183921362443e-05,
      "loss": 10.2986,
      "step": 58
    },
    {
      "epoch": 0.03317402305313467,
      "grad_norm": 0.2128066122531891,
      "learning_rate": 8.44676704559283e-05,
      "loss": 10.2983,
      "step": 59
    },
    {
      "epoch": 0.03373629463030644,
      "grad_norm": 0.23017951846122742,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.2956,
      "step": 60
    },
    {
      "epoch": 0.03429856620747821,
      "grad_norm": 0.18454602360725403,
      "learning_rate": 8.32512286056924e-05,
      "loss": 10.2912,
      "step": 61
    },
    {
      "epoch": 0.03486083778464998,
      "grad_norm": 0.17630921304225922,
      "learning_rate": 8.262928807620843e-05,
      "loss": 10.2923,
      "step": 62
    },
    {
      "epoch": 0.03542310936182176,
      "grad_norm": 0.1873309314250946,
      "learning_rate": 8.199842702516583e-05,
      "loss": 10.2882,
      "step": 63
    },
    {
      "epoch": 0.03598538093899353,
      "grad_norm": 0.16678327322006226,
      "learning_rate": 8.135881792367686e-05,
      "loss": 10.2902,
      "step": 64
    },
    {
      "epoch": 0.036547652516165305,
      "grad_norm": 0.18229888379573822,
      "learning_rate": 8.07106356344834e-05,
      "loss": 10.2889,
      "step": 65
    },
    {
      "epoch": 0.037109924093337084,
      "grad_norm": 0.16722778975963593,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.2866,
      "step": 66
    },
    {
      "epoch": 0.037672195670508855,
      "grad_norm": 0.15735368430614471,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.2793,
      "step": 67
    },
    {
      "epoch": 0.03823446724768063,
      "grad_norm": 0.1543576866388321,
      "learning_rate": 7.871643313414718e-05,
      "loss": 10.2785,
      "step": 68
    },
    {
      "epoch": 0.038796738824852406,
      "grad_norm": 0.1513262838125229,
      "learning_rate": 7.803575286758364e-05,
      "loss": 10.2766,
      "step": 69
    },
    {
      "epoch": 0.03935901040202418,
      "grad_norm": 0.14744438230991364,
      "learning_rate": 7.734740790612136e-05,
      "loss": 10.2804,
      "step": 70
    },
    {
      "epoch": 0.03992128197919595,
      "grad_norm": 0.14749851822853088,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.2747,
      "step": 71
    },
    {
      "epoch": 0.04048355355636773,
      "grad_norm": 0.1528133749961853,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.2736,
      "step": 72
    },
    {
      "epoch": 0.0410458251335395,
      "grad_norm": 0.15457065403461456,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.2754,
      "step": 73
    },
    {
      "epoch": 0.04160809671071127,
      "grad_norm": 0.13954219222068787,
      "learning_rate": 7.452117519152542e-05,
      "loss": 10.2713,
      "step": 74
    },
    {
      "epoch": 0.04217036828788305,
      "grad_norm": 0.13873203098773956,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.2743,
      "step": 75
    },
    {
      "epoch": 0.04273263986505482,
      "grad_norm": 0.15091167390346527,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.2734,
      "step": 76
    },
    {
      "epoch": 0.043294911442226594,
      "grad_norm": 0.1346290558576584,
      "learning_rate": 7.233044034264034e-05,
      "loss": 10.2676,
      "step": 77
    },
    {
      "epoch": 0.04385718301939837,
      "grad_norm": 0.14285773038864136,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.2691,
      "step": 78
    },
    {
      "epoch": 0.044419454596570145,
      "grad_norm": 0.14212925732135773,
      "learning_rate": 7.083909302476453e-05,
      "loss": 10.2649,
      "step": 79
    },
    {
      "epoch": 0.044981726173741916,
      "grad_norm": 0.13048137724399567,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.2644,
      "step": 80
    },
    {
      "epoch": 0.04554399775091369,
      "grad_norm": 0.13570630550384521,
      "learning_rate": 6.932495846462261e-05,
      "loss": 10.2618,
      "step": 81
    },
    {
      "epoch": 0.04610626932808547,
      "grad_norm": 0.13826949894428253,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.2606,
      "step": 82
    },
    {
      "epoch": 0.04666854090525724,
      "grad_norm": 0.12882477045059204,
      "learning_rate": 6.778969234612584e-05,
      "loss": 10.2601,
      "step": 83
    },
    {
      "epoch": 0.04723081248242901,
      "grad_norm": 0.12722669541835785,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.2603,
      "step": 84
    },
    {
      "epoch": 0.04779308405960079,
      "grad_norm": 0.12323813140392303,
      "learning_rate": 6.623497346023418e-05,
      "loss": 10.2559,
      "step": 85
    },
    {
      "epoch": 0.04835535563677256,
      "grad_norm": 0.12103649228811264,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.2588,
      "step": 86
    },
    {
      "epoch": 0.04891762721394433,
      "grad_norm": 0.11738058924674988,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.2602,
      "step": 87
    },
    {
      "epoch": 0.04947989879111611,
      "grad_norm": 0.12051254510879517,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.2531,
      "step": 88
    },
    {
      "epoch": 0.05004217036828788,
      "grad_norm": 0.1199101135134697,
      "learning_rate": 6.307399704769099e-05,
      "loss": 10.2523,
      "step": 89
    },
    {
      "epoch": 0.050604441945459655,
      "grad_norm": 0.124351866543293,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.2505,
      "step": 90
    },
    {
      "epoch": 0.051166713522631434,
      "grad_norm": 0.12274583429098129,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.2475,
      "step": 91
    },
    {
      "epoch": 0.051728985099803206,
      "grad_norm": 0.12369772791862488,
      "learning_rate": 6.066498153718735e-05,
      "loss": 10.2444,
      "step": 92
    },
    {
      "epoch": 0.05229125667697498,
      "grad_norm": 0.11984284222126007,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.2437,
      "step": 93
    },
    {
      "epoch": 0.052853528254146756,
      "grad_norm": 0.1206233873963356,
      "learning_rate": 5.90440267166055e-05,
      "loss": 10.248,
      "step": 94
    },
    {
      "epoch": 0.05341579983131853,
      "grad_norm": 0.11975141614675522,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.2505,
      "step": 95
    },
    {
      "epoch": 0.0539780714084903,
      "grad_norm": 0.11992384493350983,
      "learning_rate": 5.74131823855921e-05,
      "loss": 10.2515,
      "step": 96
    },
    {
      "epoch": 0.05454034298566207,
      "grad_norm": 0.1408284604549408,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 10.2617,
      "step": 97
    },
    {
      "epoch": 0.05510261456283385,
      "grad_norm": 0.14512471854686737,
      "learning_rate": 5.577423184847932e-05,
      "loss": 10.2552,
      "step": 98
    },
    {
      "epoch": 0.05566488614000562,
      "grad_norm": 0.17660699784755707,
      "learning_rate": 5.495227651252315e-05,
      "loss": 10.2554,
      "step": 99
    },
    {
      "epoch": 0.056227157717177394,
      "grad_norm": 0.248210608959198,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.2632,
      "step": 100
    },
    {
      "epoch": 0.056227157717177394,
      "eval_loss": 10.24990463256836,
      "eval_runtime": 6.6864,
      "eval_samples_per_second": 447.925,
      "eval_steps_per_second": 112.019,
      "step": 100
    },
    {
      "epoch": 0.05678942929434917,
      "grad_norm": 0.11586419492959976,
      "learning_rate": 5.330452921628497e-05,
      "loss": 10.26,
      "step": 101
    },
    {
      "epoch": 0.057351700871520944,
      "grad_norm": 0.10389582812786102,
      "learning_rate": 5.247918773366112e-05,
      "loss": 10.2576,
      "step": 102
    },
    {
      "epoch": 0.057913972448692716,
      "grad_norm": 0.1078716516494751,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.2547,
      "step": 103
    },
    {
      "epoch": 0.058476244025864495,
      "grad_norm": 0.11110513657331467,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 10.2563,
      "step": 104
    },
    {
      "epoch": 0.059038515603036266,
      "grad_norm": 0.09864146262407303,
      "learning_rate": 5e-05,
      "loss": 10.2552,
      "step": 105
    },
    {
      "epoch": 0.05960078718020804,
      "grad_norm": 0.10099820792675018,
      "learning_rate": 4.917330276168208e-05,
      "loss": 10.2526,
      "step": 106
    },
    {
      "epoch": 0.06016305875737982,
      "grad_norm": 0.1103537380695343,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.2541,
      "step": 107
    },
    {
      "epoch": 0.06072533033455159,
      "grad_norm": 0.10246473550796509,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 10.2512,
      "step": 108
    },
    {
      "epoch": 0.06128760191172336,
      "grad_norm": 0.1036551296710968,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.2507,
      "step": 109
    },
    {
      "epoch": 0.06184987348889514,
      "grad_norm": 0.10316862910985947,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.2537,
      "step": 110
    },
    {
      "epoch": 0.06241214506606691,
      "grad_norm": 0.10438147932291031,
      "learning_rate": 4.504772348747687e-05,
      "loss": 10.2475,
      "step": 111
    },
    {
      "epoch": 0.06297441664323869,
      "grad_norm": 0.10019267350435257,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.2506,
      "step": 112
    },
    {
      "epoch": 0.06353668822041046,
      "grad_norm": 0.09953270852565765,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 10.2489,
      "step": 113
    },
    {
      "epoch": 0.06409895979758223,
      "grad_norm": 0.09329447895288467,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.2445,
      "step": 114
    },
    {
      "epoch": 0.064661231374754,
      "grad_norm": 0.09427279233932495,
      "learning_rate": 4.17702704859633e-05,
      "loss": 10.2505,
      "step": 115
    },
    {
      "epoch": 0.06522350295192578,
      "grad_norm": 0.09145309776067734,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.2442,
      "step": 116
    },
    {
      "epoch": 0.06578577452909755,
      "grad_norm": 0.09164765477180481,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.2423,
      "step": 117
    },
    {
      "epoch": 0.06634804610626933,
      "grad_norm": 0.08979421108961105,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.2463,
      "step": 118
    },
    {
      "epoch": 0.0669103176834411,
      "grad_norm": 0.09118715673685074,
      "learning_rate": 3.852880399766243e-05,
      "loss": 10.2435,
      "step": 119
    },
    {
      "epoch": 0.06747258926061288,
      "grad_norm": 0.09121278673410416,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.2455,
      "step": 120
    },
    {
      "epoch": 0.06803486083778465,
      "grad_norm": 0.09374017268419266,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 10.2431,
      "step": 121
    },
    {
      "epoch": 0.06859713241495642,
      "grad_norm": 0.09250976145267487,
      "learning_rate": 3.612985456190778e-05,
      "loss": 10.2426,
      "step": 122
    },
    {
      "epoch": 0.0691594039921282,
      "grad_norm": 0.08664480596780777,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.2412,
      "step": 123
    },
    {
      "epoch": 0.06972167556929997,
      "grad_norm": 0.0919286236166954,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.2453,
      "step": 124
    },
    {
      "epoch": 0.07028394714647175,
      "grad_norm": 0.0888880118727684,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.2446,
      "step": 125
    },
    {
      "epoch": 0.07084621872364352,
      "grad_norm": 0.09760021418333054,
      "learning_rate": 3.298534127791785e-05,
      "loss": 10.2408,
      "step": 126
    },
    {
      "epoch": 0.0714084903008153,
      "grad_norm": 0.08498180657625198,
      "learning_rate": 3.221030765387417e-05,
      "loss": 10.2426,
      "step": 127
    },
    {
      "epoch": 0.07197076187798707,
      "grad_norm": 0.09053520858287811,
      "learning_rate": 3.144013755408895e-05,
      "loss": 10.2429,
      "step": 128
    },
    {
      "epoch": 0.07253303345515884,
      "grad_norm": 0.09129844605922699,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 10.2369,
      "step": 129
    },
    {
      "epoch": 0.07309530503233061,
      "grad_norm": 0.09073355793952942,
      "learning_rate": 2.991522876735154e-05,
      "loss": 10.2375,
      "step": 130
    },
    {
      "epoch": 0.0736575766095024,
      "grad_norm": 0.08949051052331924,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.2397,
      "step": 131
    },
    {
      "epoch": 0.07421984818667417,
      "grad_norm": 0.08686964958906174,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 10.2408,
      "step": 132
    },
    {
      "epoch": 0.07478211976384594,
      "grad_norm": 0.08766482025384903,
      "learning_rate": 2.766955965735968e-05,
      "loss": 10.2383,
      "step": 133
    },
    {
      "epoch": 0.07534439134101771,
      "grad_norm": 0.09003995358943939,
      "learning_rate": 2.693294185106562e-05,
      "loss": 10.2365,
      "step": 134
    },
    {
      "epoch": 0.07590666291818948,
      "grad_norm": 0.0899474248290062,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 10.2348,
      "step": 135
    },
    {
      "epoch": 0.07646893449536125,
      "grad_norm": 0.08693861961364746,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.2351,
      "step": 136
    },
    {
      "epoch": 0.07703120607253304,
      "grad_norm": 0.09370850771665573,
      "learning_rate": 2.476172311325783e-05,
      "loss": 10.2345,
      "step": 137
    },
    {
      "epoch": 0.07759347764970481,
      "grad_norm": 0.08882737904787064,
      "learning_rate": 2.405152131093926e-05,
      "loss": 10.2328,
      "step": 138
    },
    {
      "epoch": 0.07815574922687658,
      "grad_norm": 0.09736776351928711,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 10.2314,
      "step": 139
    },
    {
      "epoch": 0.07871802080404836,
      "grad_norm": 0.08899082988500595,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 10.2293,
      "step": 140
    },
    {
      "epoch": 0.07928029238122013,
      "grad_norm": 0.09270042181015015,
      "learning_rate": 2.196424713241637e-05,
      "loss": 10.2341,
      "step": 141
    },
    {
      "epoch": 0.0798425639583919,
      "grad_norm": 0.0994020327925682,
      "learning_rate": 2.128356686585282e-05,
      "loss": 10.2273,
      "step": 142
    },
    {
      "epoch": 0.08040483553556367,
      "grad_norm": 0.0926503911614418,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.2285,
      "step": 143
    },
    {
      "epoch": 0.08096710711273546,
      "grad_norm": 0.1080385372042656,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 10.2239,
      "step": 144
    },
    {
      "epoch": 0.08152937868990723,
      "grad_norm": 0.09848050773143768,
      "learning_rate": 1.928936436551661e-05,
      "loss": 10.2268,
      "step": 145
    },
    {
      "epoch": 0.082091650267079,
      "grad_norm": 0.0971171110868454,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 10.2317,
      "step": 146
    },
    {
      "epoch": 0.08265392184425077,
      "grad_norm": 0.09865108132362366,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.2354,
      "step": 147
    },
    {
      "epoch": 0.08321619342142254,
      "grad_norm": 0.11976338177919388,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 10.2353,
      "step": 148
    },
    {
      "epoch": 0.08377846499859432,
      "grad_norm": 0.15529459714889526,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 10.2442,
      "step": 149
    },
    {
      "epoch": 0.0843407365757661,
      "grad_norm": 0.20458906888961792,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 10.2474,
      "step": 150
    },
    {
      "epoch": 0.0843407365757661,
      "eval_loss": 10.23597526550293,
      "eval_runtime": 6.6724,
      "eval_samples_per_second": 448.866,
      "eval_steps_per_second": 112.254,
      "step": 150
    },
    {
      "epoch": 0.08490300815293787,
      "grad_norm": 0.1427125632762909,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.2488,
      "step": 151
    },
    {
      "epoch": 0.08546527973010964,
      "grad_norm": 0.12390498071908951,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.2464,
      "step": 152
    },
    {
      "epoch": 0.08602755130728142,
      "grad_norm": 0.10369152575731277,
      "learning_rate": 1.435357758543015e-05,
      "loss": 10.2409,
      "step": 153
    },
    {
      "epoch": 0.08658982288445319,
      "grad_norm": 0.11705708503723145,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 10.2447,
      "step": 154
    },
    {
      "epoch": 0.08715209446162496,
      "grad_norm": 0.11495331674814224,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 10.244,
      "step": 155
    },
    {
      "epoch": 0.08771436603879675,
      "grad_norm": 0.09831446409225464,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 10.242,
      "step": 156
    },
    {
      "epoch": 0.08827663761596852,
      "grad_norm": 0.08565042912960052,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 10.2435,
      "step": 157
    },
    {
      "epoch": 0.08883890919314029,
      "grad_norm": 0.1082313284277916,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.242,
      "step": 158
    },
    {
      "epoch": 0.08940118077031206,
      "grad_norm": 0.09770632535219193,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 10.2413,
      "step": 159
    },
    {
      "epoch": 0.08996345234748383,
      "grad_norm": 0.08227454870939255,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.2386,
      "step": 160
    },
    {
      "epoch": 0.0905257239246556,
      "grad_norm": 0.08055034279823303,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 10.2352,
      "step": 161
    },
    {
      "epoch": 0.09108799550182738,
      "grad_norm": 0.08792147040367126,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.2372,
      "step": 162
    },
    {
      "epoch": 0.09165026707899916,
      "grad_norm": 0.08439870178699493,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.2407,
      "step": 163
    },
    {
      "epoch": 0.09221253865617093,
      "grad_norm": 0.08185575902462006,
      "learning_rate": 8.599558442598998e-06,
      "loss": 10.2448,
      "step": 164
    },
    {
      "epoch": 0.0927748102333427,
      "grad_norm": 0.07867370545864105,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.2394,
      "step": 165
    },
    {
      "epoch": 0.09333708181051448,
      "grad_norm": 0.0815424993634224,
      "learning_rate": 7.695237378953223e-06,
      "loss": 10.2385,
      "step": 166
    },
    {
      "epoch": 0.09389935338768625,
      "grad_norm": 0.081153504550457,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.2368,
      "step": 167
    },
    {
      "epoch": 0.09446162496485802,
      "grad_norm": 0.08403248339891434,
      "learning_rate": 6.837175952121306e-06,
      "loss": 10.2421,
      "step": 168
    },
    {
      "epoch": 0.0950238965420298,
      "grad_norm": 0.08262819796800613,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.243,
      "step": 169
    },
    {
      "epoch": 0.09558616811920158,
      "grad_norm": 0.07913655787706375,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.2356,
      "step": 170
    },
    {
      "epoch": 0.09614843969637335,
      "grad_norm": 0.08381342887878418,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.2365,
      "step": 171
    },
    {
      "epoch": 0.09671071127354512,
      "grad_norm": 0.07874435931444168,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.2393,
      "step": 172
    },
    {
      "epoch": 0.0972729828507169,
      "grad_norm": 0.08159884810447693,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.2368,
      "step": 173
    },
    {
      "epoch": 0.09783525442788867,
      "grad_norm": 0.10534994304180145,
      "learning_rate": 4.549673247541875e-06,
      "loss": 10.24,
      "step": 174
    },
    {
      "epoch": 0.09839752600506045,
      "grad_norm": 0.07906033843755722,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.2379,
      "step": 175
    },
    {
      "epoch": 0.09895979758223222,
      "grad_norm": 0.07596909999847412,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.2347,
      "step": 176
    },
    {
      "epoch": 0.099522069159404,
      "grad_norm": 0.09577859938144684,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 10.2329,
      "step": 177
    },
    {
      "epoch": 0.10008434073657577,
      "grad_norm": 0.08976684510707855,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.2325,
      "step": 178
    },
    {
      "epoch": 0.10064661231374754,
      "grad_norm": 0.08572094142436981,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.2393,
      "step": 179
    },
    {
      "epoch": 0.10120888389091931,
      "grad_norm": 0.080301433801651,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 10.2385,
      "step": 180
    },
    {
      "epoch": 0.10177115546809108,
      "grad_norm": 0.08220746368169785,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.2332,
      "step": 181
    },
    {
      "epoch": 0.10233342704526287,
      "grad_norm": 0.10692481696605682,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 10.2315,
      "step": 182
    },
    {
      "epoch": 0.10289569862243464,
      "grad_norm": 0.10594671219587326,
      "learning_rate": 1.962316193157593e-06,
      "loss": 10.2359,
      "step": 183
    },
    {
      "epoch": 0.10345797019960641,
      "grad_norm": 0.08238958567380905,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.2352,
      "step": 184
    },
    {
      "epoch": 0.10402024177677818,
      "grad_norm": 0.09819667786359787,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 10.2286,
      "step": 185
    },
    {
      "epoch": 0.10458251335394995,
      "grad_norm": 0.0907939150929451,
      "learning_rate": 1.333670137599713e-06,
      "loss": 10.2316,
      "step": 186
    },
    {
      "epoch": 0.10514478493112173,
      "grad_norm": 0.09807981550693512,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 10.2346,
      "step": 187
    },
    {
      "epoch": 0.10570705650829351,
      "grad_norm": 0.08334434032440186,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.231,
      "step": 188
    },
    {
      "epoch": 0.10626932808546528,
      "grad_norm": 0.08330542594194412,
      "learning_rate": 8.247462563808817e-07,
      "loss": 10.2293,
      "step": 189
    },
    {
      "epoch": 0.10683159966263706,
      "grad_norm": 0.08413606137037277,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.2298,
      "step": 190
    },
    {
      "epoch": 0.10739387123980883,
      "grad_norm": 0.09357918053865433,
      "learning_rate": 5.526064699265753e-07,
      "loss": 10.2258,
      "step": 191
    },
    {
      "epoch": 0.1079561428169806,
      "grad_norm": 0.0894232839345932,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.2238,
      "step": 192
    },
    {
      "epoch": 0.10851841439415237,
      "grad_norm": 0.09096109122037888,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.2255,
      "step": 193
    },
    {
      "epoch": 0.10908068597132414,
      "grad_norm": 0.10491399466991425,
      "learning_rate": 2.458548727494292e-07,
      "loss": 10.2209,
      "step": 194
    },
    {
      "epoch": 0.10964295754849593,
      "grad_norm": 0.10258984565734863,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.2231,
      "step": 195
    },
    {
      "epoch": 0.1102052291256677,
      "grad_norm": 0.12160323560237885,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 10.2243,
      "step": 196
    },
    {
      "epoch": 0.11076750070283947,
      "grad_norm": 0.0939982607960701,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.2372,
      "step": 197
    },
    {
      "epoch": 0.11132977228001124,
      "grad_norm": 0.10831650346517563,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.2312,
      "step": 198
    },
    {
      "epoch": 0.11189204385718302,
      "grad_norm": 0.13033053278923035,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.2401,
      "step": 199
    },
    {
      "epoch": 0.11245431543435479,
      "grad_norm": 0.17492900788784027,
      "learning_rate": 0.0,
      "loss": 10.242,
      "step": 200
    },
    {
      "epoch": 0.11245431543435479,
      "eval_loss": 10.234081268310547,
      "eval_runtime": 6.6909,
      "eval_samples_per_second": 447.621,
      "eval_steps_per_second": 111.943,
      "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": 22310132121600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}