{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9976812782939256,
  "global_step": 126500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.9921131914759374e-05,
      "loss": 1.2922,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.984226382951875e-05,
      "loss": 1.3127,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.976339574427812e-05,
      "loss": 1.293,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9684527659037496e-05,
      "loss": 1.2856,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9605659573796868e-05,
      "loss": 1.3064,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9526791488556244e-05,
      "loss": 1.3068,
      "step": 3000
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9447923403315616e-05,
      "loss": 1.2982,
      "step": 3500
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.936905531807499e-05,
      "loss": 1.303,
      "step": 4000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9290187232834363e-05,
      "loss": 1.3007,
      "step": 4500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9211319147593738e-05,
      "loss": 1.3172,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.913245106235311e-05,
      "loss": 1.325,
      "step": 5500
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9053582977112482e-05,
      "loss": 1.2849,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.8974714891871857e-05,
      "loss": 1.3189,
      "step": 6500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.889584680663123e-05,
      "loss": 1.2955,
      "step": 7000
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8816978721390605e-05,
      "loss": 1.3496,
      "step": 7500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8738110636149977e-05,
      "loss": 1.3103,
      "step": 8000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8659242550909352e-05,
      "loss": 1.2921,
      "step": 8500
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8580374465668724e-05,
      "loss": 1.303,
      "step": 9000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.85015063804281e-05,
      "loss": 1.2997,
      "step": 9500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.842263829518747e-05,
      "loss": 1.2692,
      "step": 10000
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8343770209946847e-05,
      "loss": 1.2915,
      "step": 10500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.826490212470622e-05,
      "loss": 1.2907,
      "step": 11000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.818603403946559e-05,
      "loss": 1.3233,
      "step": 11500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8107165954224966e-05,
      "loss": 1.3047,
      "step": 12000
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.8028297868984338e-05,
      "loss": 1.273,
      "step": 12500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.7949429783743713e-05,
      "loss": 1.3118,
      "step": 13000
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7870561698503085e-05,
      "loss": 1.3183,
      "step": 13500
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.779169361326246e-05,
      "loss": 1.304,
      "step": 14000
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7712825528021833e-05,
      "loss": 1.3409,
      "step": 14500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.7633957442781208e-05,
      "loss": 1.3068,
      "step": 15000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.755508935754058e-05,
      "loss": 1.3038,
      "step": 15500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7476221272299955e-05,
      "loss": 1.3331,
      "step": 16000
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7397353187059327e-05,
      "loss": 1.2997,
      "step": 16500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.73184851018187e-05,
      "loss": 1.2833,
      "step": 17000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.723961701657807e-05,
      "loss": 1.3178,
      "step": 17500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7160748931337447e-05,
      "loss": 1.3263,
      "step": 18000
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.708188084609682e-05,
      "loss": 1.2907,
      "step": 18500
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.7003012760856194e-05,
      "loss": 1.3328,
      "step": 19000
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.6924144675615566e-05,
      "loss": 1.3268,
      "step": 19500
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6845276590374938e-05,
      "loss": 1.3161,
      "step": 20000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6766408505134313e-05,
      "loss": 1.3026,
      "step": 20500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.6687540419893685e-05,
      "loss": 1.3163,
      "step": 21000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.660867233465306e-05,
      "loss": 1.3202,
      "step": 21500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.6529804249412432e-05,
      "loss": 1.2872,
      "step": 22000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6450936164171808e-05,
      "loss": 1.3209,
      "step": 22500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.637206807893118e-05,
      "loss": 1.3165,
      "step": 23000
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6293199993690555e-05,
      "loss": 1.3026,
      "step": 23500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6214331908449927e-05,
      "loss": 1.3313,
      "step": 24000
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6135463823209302e-05,
      "loss": 1.3123,
      "step": 24500
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6056595737968674e-05,
      "loss": 1.3124,
      "step": 25000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.5977727652728046e-05,
      "loss": 1.3227,
      "step": 25500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.589885956748742e-05,
      "loss": 1.316,
      "step": 26000
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.5819991482246794e-05,
      "loss": 1.2764,
      "step": 26500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.574112339700617e-05,
      "loss": 1.3096,
      "step": 27000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.566225531176554e-05,
      "loss": 1.3499,
      "step": 27500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5583387226524916e-05,
      "loss": 1.3339,
      "step": 28000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5504519141284288e-05,
      "loss": 1.3423,
      "step": 28500
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.5425651056043664e-05,
      "loss": 1.3295,
      "step": 29000
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.5346782970803036e-05,
      "loss": 1.3226,
      "step": 29500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.526791488556241e-05,
      "loss": 1.2861,
      "step": 30000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5189046800321783e-05,
      "loss": 1.3504,
      "step": 30500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5110178715081157e-05,
      "loss": 1.3619,
      "step": 31000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.503131062984053e-05,
      "loss": 1.329,
      "step": 31500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.4952442544599904e-05,
      "loss": 1.2826,
      "step": 32000
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4873574459359277e-05,
      "loss": 1.3075,
      "step": 32500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4794706374118651e-05,
      "loss": 1.3293,
      "step": 33000
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4715838288878023e-05,
      "loss": 1.2638,
      "step": 33500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.4636970203637397e-05,
      "loss": 1.3282,
      "step": 34000
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.455810211839677e-05,
      "loss": 1.3562,
      "step": 34500
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4479234033156144e-05,
      "loss": 1.3446,
      "step": 35000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4400365947915518e-05,
      "loss": 1.3339,
      "step": 35500
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4321497862674891e-05,
      "loss": 1.31,
      "step": 36000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4242629777434265e-05,
      "loss": 1.3224,
      "step": 36500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4163761692193639e-05,
      "loss": 1.3715,
      "step": 37000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.4084893606953012e-05,
      "loss": 1.3081,
      "step": 37500
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.4006025521712386e-05,
      "loss": 1.2892,
      "step": 38000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.3927157436471758e-05,
      "loss": 1.3046,
      "step": 38500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.3848289351231132e-05,
      "loss": 1.3344,
      "step": 39000
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.3769421265990505e-05,
      "loss": 1.3303,
      "step": 39500
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3690553180749879e-05,
      "loss": 1.3237,
      "step": 40000
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3611685095509253e-05,
      "loss": 1.3162,
      "step": 40500
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3532817010268626e-05,
      "loss": 1.3496,
      "step": 41000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.3453948925028e-05,
      "loss": 1.355,
      "step": 41500
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.3375080839787374e-05,
      "loss": 1.348,
      "step": 42000
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3296212754546747e-05,
      "loss": 1.344,
      "step": 42500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3217344669306121e-05,
      "loss": 1.3256,
      "step": 43000
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3138476584065495e-05,
      "loss": 1.3724,
      "step": 43500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.3059608498824867e-05,
      "loss": 1.2848,
      "step": 44000
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.298074041358424e-05,
      "loss": 1.3318,
      "step": 44500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.2901872328343614e-05,
      "loss": 1.3025,
      "step": 45000
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.2823004243102987e-05,
      "loss": 1.2953,
      "step": 45500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.2744136157862361e-05,
      "loss": 1.3388,
      "step": 46000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2665268072621735e-05,
      "loss": 1.3444,
      "step": 46500
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2586399987381108e-05,
      "loss": 1.3239,
      "step": 47000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2507531902140482e-05,
      "loss": 1.296,
      "step": 47500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.2428663816899856e-05,
      "loss": 1.3024,
      "step": 48000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.234979573165923e-05,
      "loss": 1.33,
      "step": 48500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2270927646418603e-05,
      "loss": 1.3103,
      "step": 49000
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2192059561177975e-05,
      "loss": 1.3585,
      "step": 49500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2113191475937349e-05,
      "loss": 1.3155,
      "step": 50000
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.2034323390696722e-05,
      "loss": 1.3307,
      "step": 50500
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.1955455305456094e-05,
      "loss": 1.3344,
      "step": 51000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1876587220215468e-05,
      "loss": 1.3154,
      "step": 51500
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1797719134974842e-05,
      "loss": 1.3233,
      "step": 52000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1718851049734215e-05,
      "loss": 1.2995,
      "step": 52500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1639982964493587e-05,
      "loss": 1.3273,
      "step": 53000
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1561114879252961e-05,
      "loss": 1.3384,
      "step": 53500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1482246794012335e-05,
      "loss": 1.3127,
      "step": 54000
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1403378708771708e-05,
      "loss": 1.3681,
      "step": 54500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1324510623531082e-05,
      "loss": 1.3499,
      "step": 55000
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1245642538290456e-05,
      "loss": 1.3132,
      "step": 55500
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.116677445304983e-05,
      "loss": 1.3173,
      "step": 56000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1087906367809203e-05,
      "loss": 1.3366,
      "step": 56500
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1009038282568577e-05,
      "loss": 1.3433,
      "step": 57000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.093017019732795e-05,
      "loss": 1.3168,
      "step": 57500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0851302112087324e-05,
      "loss": 1.3489,
      "step": 58000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0772434026846696e-05,
      "loss": 1.3441,
      "step": 58500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.069356594160607e-05,
      "loss": 1.2982,
      "step": 59000
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0614697856365443e-05,
      "loss": 1.3138,
      "step": 59500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0535829771124817e-05,
      "loss": 1.3417,
      "step": 60000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.045696168588419e-05,
      "loss": 1.3107,
      "step": 60500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0378093600643564e-05,
      "loss": 1.3524,
      "step": 61000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0299225515402938e-05,
      "loss": 1.3509,
      "step": 61500
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0220357430162311e-05,
      "loss": 1.3469,
      "step": 62000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0141489344921685e-05,
      "loss": 1.3231,
      "step": 62500
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0062621259681059e-05,
      "loss": 1.3218,
      "step": 63000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.98375317444043e-06,
      "loss": 1.3203,
      "step": 63500
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.904885089199804e-06,
      "loss": 1.286,
      "step": 64000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.826017003959178e-06,
      "loss": 1.2927,
      "step": 64500
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.747148918718552e-06,
      "loss": 1.3631,
      "step": 65000
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.668280833477925e-06,
      "loss": 1.3556,
      "step": 65500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.589412748237299e-06,
      "loss": 1.3153,
      "step": 66000
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.510544662996673e-06,
      "loss": 1.3567,
      "step": 66500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.431676577756046e-06,
      "loss": 1.2935,
      "step": 67000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.35280849251542e-06,
      "loss": 1.3497,
      "step": 67500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.273940407274794e-06,
      "loss": 1.3418,
      "step": 68000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.195072322034167e-06,
      "loss": 1.2812,
      "step": 68500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.11620423679354e-06,
      "loss": 1.3112,
      "step": 69000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.037336151552913e-06,
      "loss": 1.3378,
      "step": 69500
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.958468066312287e-06,
      "loss": 1.3173,
      "step": 70000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.87959998107166e-06,
      "loss": 1.3013,
      "step": 70500
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.800731895831034e-06,
      "loss": 1.3323,
      "step": 71000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.721863810590407e-06,
      "loss": 1.2787,
      "step": 71500
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.642995725349781e-06,
      "loss": 1.3173,
      "step": 72000
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.564127640109155e-06,
      "loss": 1.3196,
      "step": 72500
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.485259554868528e-06,
      "loss": 1.2903,
      "step": 73000
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.406391469627902e-06,
      "loss": 1.3292,
      "step": 73500
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.327523384387276e-06,
      "loss": 1.3402,
      "step": 74000
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.248655299146648e-06,
      "loss": 1.3504,
      "step": 74500
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.169787213906021e-06,
      "loss": 1.318,
      "step": 75000
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.090919128665395e-06,
      "loss": 1.3544,
      "step": 75500
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.012051043424769e-06,
      "loss": 1.2946,
      "step": 76000
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.93318295818414e-06,
      "loss": 1.357,
      "step": 76500
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.854314872943514e-06,
      "loss": 1.3001,
      "step": 77000
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.775446787702888e-06,
      "loss": 1.3008,
      "step": 77500
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.696578702462262e-06,
      "loss": 1.3285,
      "step": 78000
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.617710617221635e-06,
      "loss": 1.3149,
      "step": 78500
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.538842531981009e-06,
      "loss": 1.2988,
      "step": 79000
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.459974446740383e-06,
      "loss": 1.3254,
      "step": 79500
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.381106361499755e-06,
      "loss": 1.3198,
      "step": 80000
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.302238276259129e-06,
      "loss": 1.2625,
      "step": 80500
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.223370191018503e-06,
      "loss": 1.3755,
      "step": 81000
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.144502105777876e-06,
      "loss": 1.3148,
      "step": 81500
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.06563402053725e-06,
      "loss": 1.2799,
      "step": 82000
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.986765935296624e-06,
      "loss": 1.3211,
      "step": 82500
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.9078978500559965e-06,
      "loss": 1.3325,
      "step": 83000
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.82902976481537e-06,
      "loss": 1.2948,
      "step": 83500
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.750161679574744e-06,
      "loss": 1.314,
      "step": 84000
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.6712935943341175e-06,
      "loss": 1.2739,
      "step": 84500
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.592425509093491e-06,
      "loss": 1.3231,
      "step": 85000
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.513557423852864e-06,
      "loss": 1.3044,
      "step": 85500
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.434689338612238e-06,
      "loss": 1.3177,
      "step": 86000
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.355821253371611e-06,
      "loss": 1.3457,
      "step": 86500
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.276953168130985e-06,
      "loss": 1.3338,
      "step": 87000
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.1980850828903586e-06,
      "loss": 1.3145,
      "step": 87500
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.119216997649731e-06,
      "loss": 1.2863,
      "step": 88000
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.040348912409105e-06,
      "loss": 1.3334,
      "step": 88500
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.961480827168479e-06,
      "loss": 1.2809,
      "step": 89000
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.882612741927852e-06,
      "loss": 1.3164,
      "step": 89500
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.803744656687226e-06,
      "loss": 1.2591,
      "step": 90000
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.7248765714466e-06,
      "loss": 1.3127,
      "step": 90500
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.6460084862059725e-06,
      "loss": 1.2838,
      "step": 91000
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.567140400965346e-06,
      "loss": 1.2783,
      "step": 91500
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.48827231572472e-06,
      "loss": 1.2546,
      "step": 92000
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.4094042304840934e-06,
      "loss": 1.3429,
      "step": 92500
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.330536145243467e-06,
      "loss": 1.3223,
      "step": 93000
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.251668060002839e-06,
      "loss": 1.2925,
      "step": 93500
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.172799974762213e-06,
      "loss": 1.2802,
      "step": 94000
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.093931889521586e-06,
      "loss": 1.2491,
      "step": 94500
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.01506380428096e-06,
      "loss": 1.3079,
      "step": 95000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.936195719040334e-06,
      "loss": 1.2788,
      "step": 95500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.857327633799707e-06,
      "loss": 1.2728,
      "step": 96000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.778459548559081e-06,
      "loss": 1.3208,
      "step": 96500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.699591463318454e-06,
      "loss": 1.3279,
      "step": 97000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.6207233780778275e-06,
      "loss": 1.3523,
      "step": 97500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.5418552928372e-06,
      "loss": 1.3168,
      "step": 98000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.462987207596574e-06,
      "loss": 1.2678,
      "step": 98500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.384119122355948e-06,
      "loss": 1.2477,
      "step": 99000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.305251037115321e-06,
      "loss": 1.3298,
      "step": 99500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.226382951874695e-06,
      "loss": 1.2886,
      "step": 100000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.147514866634068e-06,
      "loss": 1.3194,
      "step": 100500
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.068646781393441e-06,
      "loss": 1.3336,
      "step": 101000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.989778696152815e-06,
      "loss": 1.28,
      "step": 101500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.910910610912189e-06,
      "loss": 1.2774,
      "step": 102000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.832042525671562e-06,
      "loss": 1.3041,
      "step": 102500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.7531744404309356e-06,
      "loss": 1.3761,
      "step": 103000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6743063551903092e-06,
      "loss": 1.37,
      "step": 103500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.5954382699496825e-06,
      "loss": 1.3055,
      "step": 104000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.516570184709056e-06,
      "loss": 1.3361,
      "step": 104500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.4377020994684298e-06,
      "loss": 1.3031,
      "step": 105000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.358834014227803e-06,
      "loss": 1.3137,
      "step": 105500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.2799659289871767e-06,
      "loss": 1.2919,
      "step": 106000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.2010978437465495e-06,
      "loss": 1.2852,
      "step": 106500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.122229758505923e-06,
      "loss": 1.2751,
      "step": 107000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.0433616732652964e-06,
      "loss": 1.2906,
      "step": 107500
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.96449358802467e-06,
      "loss": 1.3119,
      "step": 108000
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.8856255027840437e-06,
      "loss": 1.2621,
      "step": 108500
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.806757417543417e-06,
      "loss": 1.2994,
      "step": 109000
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.7278893323027906e-06,
      "loss": 1.2909,
      "step": 109500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.649021247062164e-06,
      "loss": 1.3306,
      "step": 110000
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.5701531618215375e-06,
      "loss": 1.2881,
      "step": 110500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.491285076580911e-06,
      "loss": 1.306,
      "step": 111000
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.4124169913402844e-06,
      "loss": 1.3357,
      "step": 111500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.333548906099658e-06,
      "loss": 1.3152,
      "step": 112000
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.2546808208590317e-06,
      "loss": 1.3011,
      "step": 112500
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.175812735618405e-06,
      "loss": 1.2949,
      "step": 113000
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.096944650377778e-06,
      "loss": 1.2997,
      "step": 113500
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.018076565137152e-06,
      "loss": 1.3148,
      "step": 114000
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.939208479896525e-06,
      "loss": 1.3464,
      "step": 114500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8603403946558987e-06,
      "loss": 1.2722,
      "step": 115000
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7814723094152721e-06,
      "loss": 1.2361,
      "step": 115500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7026042241746458e-06,
      "loss": 1.2932,
      "step": 116000
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6237361389340192e-06,
      "loss": 1.2929,
      "step": 116500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.5448680536933925e-06,
      "loss": 1.2514,
      "step": 117000
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.465999968452766e-06,
      "loss": 1.3133,
      "step": 117500
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.3871318832121394e-06,
      "loss": 1.2766,
      "step": 118000
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.308263797971513e-06,
      "loss": 1.2892,
      "step": 118500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.2293957127308865e-06,
      "loss": 1.3185,
      "step": 119000
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.15052762749026e-06,
      "loss": 1.299,
      "step": 119500
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0716595422496333e-06,
      "loss": 1.2878,
      "step": 120000
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.927914570090068e-07,
      "loss": 1.2827,
      "step": 120500
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.139233717683802e-07,
      "loss": 1.3103,
      "step": 121000
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.350552865277538e-07,
      "loss": 1.3186,
      "step": 121500
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.561872012871272e-07,
      "loss": 1.3046,
      "step": 122000
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.773191160465007e-07,
      "loss": 1.2945,
      "step": 122500
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.984510308058741e-07,
      "loss": 1.3193,
      "step": 123000
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.195829455652477e-07,
      "loss": 1.3048,
      "step": 123500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.4071486032462107e-07,
      "loss": 1.2599,
      "step": 124000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.6184677508399457e-07,
      "loss": 1.2585,
      "step": 124500
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.82978689843368e-07,
      "loss": 1.2506,
      "step": 125000
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.0411060460274149e-07,
      "loss": 1.2928,
      "step": 125500
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.2524251936211493e-07,
      "loss": 1.2959,
      "step": 126000
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.63744341214884e-08,
      "loss": 1.2957,
      "step": 126500
    }
  ],
  "max_steps": 126794,
  "num_train_epochs": 1,
  "total_flos": 4757862742622208.0,
  "trial_name": null,
  "trial_params": null
}