het5_small_summarization / trainer_state.json
imvladikon's picture
init
9892458
raw
history blame contribute delete
No virus
27.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.518549993730757,
"global_step": 109000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.993034174340685e-05,
"loss": 2.3834,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.986068348681369e-05,
"loss": 2.3564,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.979102523022054e-05,
"loss": 2.3619,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 4.972136697362739e-05,
"loss": 2.3887,
"step": 2000
},
{
"epoch": 0.03,
"learning_rate": 4.965170871703423e-05,
"loss": 2.3675,
"step": 2500
},
{
"epoch": 0.04,
"learning_rate": 4.958205046044108e-05,
"loss": 2.3616,
"step": 3000
},
{
"epoch": 0.05,
"learning_rate": 4.951239220384793e-05,
"loss": 2.3612,
"step": 3500
},
{
"epoch": 0.06,
"learning_rate": 4.944273394725477e-05,
"loss": 2.3387,
"step": 4000
},
{
"epoch": 0.06,
"learning_rate": 4.937307569066162e-05,
"loss": 2.3691,
"step": 4500
},
{
"epoch": 0.07,
"learning_rate": 4.9303417434068464e-05,
"loss": 2.3528,
"step": 5000
},
{
"epoch": 0.08,
"learning_rate": 4.923375917747531e-05,
"loss": 2.3344,
"step": 5500
},
{
"epoch": 0.08,
"learning_rate": 4.916410092088216e-05,
"loss": 2.3454,
"step": 6000
},
{
"epoch": 0.09,
"learning_rate": 4.9094442664289e-05,
"loss": 2.3597,
"step": 6500
},
{
"epoch": 0.1,
"learning_rate": 4.9024784407695844e-05,
"loss": 2.3215,
"step": 7000
},
{
"epoch": 0.1,
"learning_rate": 4.8955126151102695e-05,
"loss": 2.3358,
"step": 7500
},
{
"epoch": 0.11,
"learning_rate": 4.888546789450954e-05,
"loss": 2.3134,
"step": 8000
},
{
"epoch": 0.12,
"learning_rate": 4.881580963791638e-05,
"loss": 2.3409,
"step": 8500
},
{
"epoch": 0.13,
"learning_rate": 4.874615138132323e-05,
"loss": 2.3431,
"step": 9000
},
{
"epoch": 0.13,
"learning_rate": 4.8676493124730075e-05,
"loss": 2.3337,
"step": 9500
},
{
"epoch": 0.14,
"learning_rate": 4.8606834868136925e-05,
"loss": 2.3242,
"step": 10000
},
{
"epoch": 0.15,
"learning_rate": 4.853717661154377e-05,
"loss": 2.3335,
"step": 10500
},
{
"epoch": 0.15,
"learning_rate": 4.846751835495061e-05,
"loss": 2.3331,
"step": 11000
},
{
"epoch": 0.16,
"learning_rate": 4.839786009835746e-05,
"loss": 2.3122,
"step": 11500
},
{
"epoch": 0.17,
"learning_rate": 4.8328201841764306e-05,
"loss": 2.3356,
"step": 12000
},
{
"epoch": 0.17,
"learning_rate": 4.825854358517115e-05,
"loss": 2.3029,
"step": 12500
},
{
"epoch": 0.18,
"learning_rate": 4.8188885328578e-05,
"loss": 2.2891,
"step": 13000
},
{
"epoch": 0.19,
"learning_rate": 4.811922707198484e-05,
"loss": 2.3112,
"step": 13500
},
{
"epoch": 0.2,
"learning_rate": 4.804956881539169e-05,
"loss": 2.3219,
"step": 14000
},
{
"epoch": 0.2,
"learning_rate": 4.797991055879854e-05,
"loss": 2.3077,
"step": 14500
},
{
"epoch": 0.21,
"learning_rate": 4.791025230220538e-05,
"loss": 2.3153,
"step": 15000
},
{
"epoch": 0.22,
"learning_rate": 4.784059404561223e-05,
"loss": 2.2975,
"step": 15500
},
{
"epoch": 0.22,
"learning_rate": 4.7770935789019074e-05,
"loss": 2.3129,
"step": 16000
},
{
"epoch": 0.23,
"learning_rate": 4.770127753242592e-05,
"loss": 2.3149,
"step": 16500
},
{
"epoch": 0.24,
"learning_rate": 4.763161927583277e-05,
"loss": 2.3233,
"step": 17000
},
{
"epoch": 0.24,
"learning_rate": 4.756196101923962e-05,
"loss": 2.2995,
"step": 17500
},
{
"epoch": 0.25,
"learning_rate": 4.7492302762646454e-05,
"loss": 2.3327,
"step": 18000
},
{
"epoch": 0.26,
"learning_rate": 4.7422644506053305e-05,
"loss": 2.2787,
"step": 18500
},
{
"epoch": 0.26,
"learning_rate": 4.7352986249460155e-05,
"loss": 2.2952,
"step": 19000
},
{
"epoch": 0.27,
"learning_rate": 4.7283327992867e-05,
"loss": 2.2961,
"step": 19500
},
{
"epoch": 0.28,
"learning_rate": 4.721366973627384e-05,
"loss": 2.2844,
"step": 20000
},
{
"epoch": 0.29,
"learning_rate": 4.714401147968069e-05,
"loss": 2.3003,
"step": 20500
},
{
"epoch": 0.29,
"learning_rate": 4.7074353223087535e-05,
"loss": 2.2918,
"step": 21000
},
{
"epoch": 0.3,
"learning_rate": 4.700469496649438e-05,
"loss": 2.2911,
"step": 21500
},
{
"epoch": 0.31,
"learning_rate": 4.693503670990122e-05,
"loss": 2.3298,
"step": 22000
},
{
"epoch": 0.31,
"learning_rate": 4.686537845330807e-05,
"loss": 2.288,
"step": 22500
},
{
"epoch": 0.32,
"learning_rate": 4.679572019671492e-05,
"loss": 2.3024,
"step": 23000
},
{
"epoch": 0.33,
"learning_rate": 4.672606194012176e-05,
"loss": 2.3144,
"step": 23500
},
{
"epoch": 0.33,
"learning_rate": 4.665640368352861e-05,
"loss": 2.2914,
"step": 24000
},
{
"epoch": 0.34,
"learning_rate": 4.658674542693546e-05,
"loss": 2.3032,
"step": 24500
},
{
"epoch": 0.35,
"learning_rate": 4.65170871703423e-05,
"loss": 2.285,
"step": 25000
},
{
"epoch": 0.36,
"learning_rate": 4.644742891374915e-05,
"loss": 2.3003,
"step": 25500
},
{
"epoch": 0.36,
"learning_rate": 4.6377770657156e-05,
"loss": 2.3186,
"step": 26000
},
{
"epoch": 0.37,
"learning_rate": 4.630811240056284e-05,
"loss": 2.3173,
"step": 26500
},
{
"epoch": 0.38,
"learning_rate": 4.623845414396969e-05,
"loss": 2.2682,
"step": 27000
},
{
"epoch": 0.38,
"learning_rate": 4.6168795887376534e-05,
"loss": 2.2887,
"step": 27500
},
{
"epoch": 0.39,
"learning_rate": 4.609913763078338e-05,
"loss": 2.3012,
"step": 28000
},
{
"epoch": 0.4,
"learning_rate": 4.602947937419023e-05,
"loss": 2.2775,
"step": 28500
},
{
"epoch": 0.4,
"learning_rate": 4.595982111759707e-05,
"loss": 2.2973,
"step": 29000
},
{
"epoch": 0.41,
"learning_rate": 4.5890162861003915e-05,
"loss": 2.2992,
"step": 29500
},
{
"epoch": 0.42,
"learning_rate": 4.5820504604410765e-05,
"loss": 2.2588,
"step": 30000
},
{
"epoch": 0.42,
"learning_rate": 4.575084634781761e-05,
"loss": 2.2797,
"step": 30500
},
{
"epoch": 0.43,
"learning_rate": 4.568118809122445e-05,
"loss": 2.2706,
"step": 31000
},
{
"epoch": 0.44,
"learning_rate": 4.56115298346313e-05,
"loss": 2.3081,
"step": 31500
},
{
"epoch": 0.45,
"learning_rate": 4.5541871578038145e-05,
"loss": 2.2698,
"step": 32000
},
{
"epoch": 0.45,
"learning_rate": 4.5472213321444996e-05,
"loss": 2.2789,
"step": 32500
},
{
"epoch": 0.46,
"learning_rate": 4.540255506485184e-05,
"loss": 2.3017,
"step": 33000
},
{
"epoch": 0.47,
"learning_rate": 4.533289680825868e-05,
"loss": 2.2938,
"step": 33500
},
{
"epoch": 0.47,
"learning_rate": 4.526323855166553e-05,
"loss": 2.3085,
"step": 34000
},
{
"epoch": 0.48,
"learning_rate": 4.519358029507238e-05,
"loss": 2.3055,
"step": 34500
},
{
"epoch": 0.49,
"learning_rate": 4.512392203847922e-05,
"loss": 2.2903,
"step": 35000
},
{
"epoch": 0.49,
"learning_rate": 4.505426378188607e-05,
"loss": 2.2923,
"step": 35500
},
{
"epoch": 0.5,
"learning_rate": 4.498460552529291e-05,
"loss": 2.2878,
"step": 36000
},
{
"epoch": 0.51,
"learning_rate": 4.491494726869976e-05,
"loss": 2.2875,
"step": 36500
},
{
"epoch": 0.52,
"learning_rate": 4.484528901210661e-05,
"loss": 2.2575,
"step": 37000
},
{
"epoch": 0.52,
"learning_rate": 4.477563075551345e-05,
"loss": 2.3053,
"step": 37500
},
{
"epoch": 0.53,
"learning_rate": 4.47059724989203e-05,
"loss": 2.2795,
"step": 38000
},
{
"epoch": 0.54,
"learning_rate": 4.4636314242327144e-05,
"loss": 2.2766,
"step": 38500
},
{
"epoch": 0.54,
"learning_rate": 4.456665598573399e-05,
"loss": 2.2476,
"step": 39000
},
{
"epoch": 0.55,
"learning_rate": 4.449699772914084e-05,
"loss": 2.3061,
"step": 39500
},
{
"epoch": 0.56,
"learning_rate": 4.442733947254769e-05,
"loss": 2.2821,
"step": 40000
},
{
"epoch": 0.56,
"learning_rate": 4.4357681215954525e-05,
"loss": 2.282,
"step": 40500
},
{
"epoch": 0.57,
"learning_rate": 4.4288022959361375e-05,
"loss": 2.276,
"step": 41000
},
{
"epoch": 0.58,
"learning_rate": 4.4218364702768225e-05,
"loss": 2.2973,
"step": 41500
},
{
"epoch": 0.59,
"learning_rate": 4.414870644617507e-05,
"loss": 2.2998,
"step": 42000
},
{
"epoch": 0.59,
"learning_rate": 4.407904818958191e-05,
"loss": 2.271,
"step": 42500
},
{
"epoch": 0.6,
"learning_rate": 4.400938993298876e-05,
"loss": 2.268,
"step": 43000
},
{
"epoch": 0.61,
"learning_rate": 4.3939731676395606e-05,
"loss": 2.2975,
"step": 43500
},
{
"epoch": 0.61,
"learning_rate": 4.387007341980245e-05,
"loss": 2.2643,
"step": 44000
},
{
"epoch": 0.62,
"learning_rate": 4.38004151632093e-05,
"loss": 2.3335,
"step": 44500
},
{
"epoch": 0.63,
"learning_rate": 4.373075690661614e-05,
"loss": 2.3195,
"step": 45000
},
{
"epoch": 0.63,
"learning_rate": 4.366109865002299e-05,
"loss": 2.3385,
"step": 45500
},
{
"epoch": 0.64,
"learning_rate": 4.3591440393429836e-05,
"loss": 2.3292,
"step": 46000
},
{
"epoch": 0.65,
"learning_rate": 4.352178213683668e-05,
"loss": 2.329,
"step": 46500
},
{
"epoch": 0.65,
"learning_rate": 4.345212388024353e-05,
"loss": 2.331,
"step": 47000
},
{
"epoch": 0.66,
"learning_rate": 4.3382465623650374e-05,
"loss": 2.3483,
"step": 47500
},
{
"epoch": 0.67,
"learning_rate": 4.331280736705722e-05,
"loss": 2.3124,
"step": 48000
},
{
"epoch": 0.68,
"learning_rate": 4.324314911046407e-05,
"loss": 2.3102,
"step": 48500
},
{
"epoch": 0.68,
"learning_rate": 4.317349085387091e-05,
"loss": 2.3151,
"step": 49000
},
{
"epoch": 0.69,
"learning_rate": 4.310383259727776e-05,
"loss": 2.323,
"step": 49500
},
{
"epoch": 0.7,
"learning_rate": 4.3034174340684604e-05,
"loss": 2.3174,
"step": 50000
},
{
"epoch": 0.7,
"learning_rate": 4.296451608409145e-05,
"loss": 2.3188,
"step": 50500
},
{
"epoch": 0.71,
"learning_rate": 4.28948578274983e-05,
"loss": 2.3054,
"step": 51000
},
{
"epoch": 0.72,
"learning_rate": 4.282519957090514e-05,
"loss": 2.3133,
"step": 51500
},
{
"epoch": 0.72,
"learning_rate": 4.2755541314311985e-05,
"loss": 2.3224,
"step": 52000
},
{
"epoch": 0.73,
"learning_rate": 4.2685883057718835e-05,
"loss": 2.3354,
"step": 52500
},
{
"epoch": 0.74,
"learning_rate": 4.261622480112568e-05,
"loss": 2.3386,
"step": 53000
},
{
"epoch": 0.75,
"learning_rate": 4.254656654453252e-05,
"loss": 2.3276,
"step": 53500
},
{
"epoch": 0.75,
"learning_rate": 4.247690828793937e-05,
"loss": 2.3057,
"step": 54000
},
{
"epoch": 0.76,
"learning_rate": 4.2407250031346216e-05,
"loss": 2.2921,
"step": 54500
},
{
"epoch": 0.77,
"learning_rate": 4.2337591774753066e-05,
"loss": 2.3185,
"step": 55000
},
{
"epoch": 0.77,
"learning_rate": 4.226793351815991e-05,
"loss": 2.3046,
"step": 55500
},
{
"epoch": 0.78,
"learning_rate": 4.219827526156675e-05,
"loss": 2.3018,
"step": 56000
},
{
"epoch": 0.79,
"learning_rate": 4.21286170049736e-05,
"loss": 2.2825,
"step": 56500
},
{
"epoch": 0.79,
"learning_rate": 4.2058958748380446e-05,
"loss": 2.3068,
"step": 57000
},
{
"epoch": 0.8,
"learning_rate": 4.198930049178729e-05,
"loss": 2.3115,
"step": 57500
},
{
"epoch": 0.81,
"learning_rate": 4.191964223519414e-05,
"loss": 2.3247,
"step": 58000
},
{
"epoch": 0.82,
"learning_rate": 4.184998397860099e-05,
"loss": 2.2919,
"step": 58500
},
{
"epoch": 0.82,
"learning_rate": 4.178032572200783e-05,
"loss": 2.345,
"step": 59000
},
{
"epoch": 0.83,
"learning_rate": 4.171066746541468e-05,
"loss": 2.298,
"step": 59500
},
{
"epoch": 0.84,
"learning_rate": 4.164100920882153e-05,
"loss": 2.3093,
"step": 60000
},
{
"epoch": 0.84,
"learning_rate": 4.157135095222837e-05,
"loss": 2.3259,
"step": 60500
},
{
"epoch": 0.85,
"learning_rate": 4.1501692695635214e-05,
"loss": 2.3174,
"step": 61000
},
{
"epoch": 0.86,
"learning_rate": 4.1432034439042064e-05,
"loss": 2.3268,
"step": 61500
},
{
"epoch": 0.86,
"learning_rate": 4.136237618244891e-05,
"loss": 2.2997,
"step": 62000
},
{
"epoch": 0.87,
"learning_rate": 4.129271792585576e-05,
"loss": 2.3239,
"step": 62500
},
{
"epoch": 0.88,
"learning_rate": 4.1223059669262595e-05,
"loss": 2.3302,
"step": 63000
},
{
"epoch": 0.88,
"learning_rate": 4.1153401412669445e-05,
"loss": 2.322,
"step": 63500
},
{
"epoch": 0.89,
"learning_rate": 4.1083743156076295e-05,
"loss": 2.3216,
"step": 64000
},
{
"epoch": 0.9,
"learning_rate": 4.101408489948313e-05,
"loss": 2.319,
"step": 64500
},
{
"epoch": 0.91,
"learning_rate": 4.094442664288998e-05,
"loss": 2.304,
"step": 65000
},
{
"epoch": 0.91,
"learning_rate": 4.087476838629683e-05,
"loss": 2.2772,
"step": 65500
},
{
"epoch": 0.92,
"learning_rate": 4.0805110129703676e-05,
"loss": 2.3205,
"step": 66000
},
{
"epoch": 0.93,
"learning_rate": 4.073545187311052e-05,
"loss": 2.3053,
"step": 66500
},
{
"epoch": 0.93,
"learning_rate": 4.066579361651737e-05,
"loss": 2.3196,
"step": 67000
},
{
"epoch": 0.94,
"learning_rate": 4.059613535992421e-05,
"loss": 2.3245,
"step": 67500
},
{
"epoch": 0.95,
"learning_rate": 4.052647710333106e-05,
"loss": 2.2804,
"step": 68000
},
{
"epoch": 0.95,
"learning_rate": 4.0456818846737907e-05,
"loss": 2.2981,
"step": 68500
},
{
"epoch": 0.96,
"learning_rate": 4.038716059014475e-05,
"loss": 2.307,
"step": 69000
},
{
"epoch": 0.97,
"learning_rate": 4.03175023335516e-05,
"loss": 2.3003,
"step": 69500
},
{
"epoch": 0.98,
"learning_rate": 4.0247844076958444e-05,
"loss": 2.2922,
"step": 70000
},
{
"epoch": 0.98,
"learning_rate": 4.017818582036529e-05,
"loss": 2.351,
"step": 70500
},
{
"epoch": 0.99,
"learning_rate": 4.010852756377214e-05,
"loss": 2.3629,
"step": 71000
},
{
"epoch": 1.0,
"learning_rate": 4.003886930717898e-05,
"loss": 2.3658,
"step": 71500
},
{
"epoch": 1.0,
"eval_gen_len": 19.0,
"eval_loss": 2.0120694637298584,
"eval_rouge1": 13.694,
"eval_rouge2": 3.0324,
"eval_rougeL": 13.5592,
"eval_rougeLsum": 13.6473,
"eval_runtime": 1519.7773,
"eval_samples_per_second": 8.796,
"eval_steps_per_second": 2.199,
"step": 71779
},
{
"epoch": 1.0,
"learning_rate": 3.9969211050585824e-05,
"loss": 2.3626,
"step": 72000
},
{
"epoch": 1.01,
"learning_rate": 3.9899552793992674e-05,
"loss": 2.3326,
"step": 72500
},
{
"epoch": 1.02,
"learning_rate": 3.982989453739952e-05,
"loss": 2.33,
"step": 73000
},
{
"epoch": 1.02,
"learning_rate": 3.976023628080637e-05,
"loss": 2.3426,
"step": 73500
},
{
"epoch": 1.03,
"learning_rate": 3.969057802421321e-05,
"loss": 2.3469,
"step": 74000
},
{
"epoch": 1.04,
"learning_rate": 3.9620919767620055e-05,
"loss": 2.3075,
"step": 74500
},
{
"epoch": 1.04,
"learning_rate": 3.9551261511026905e-05,
"loss": 2.3124,
"step": 75000
},
{
"epoch": 1.05,
"learning_rate": 3.9481603254433755e-05,
"loss": 2.344,
"step": 75500
},
{
"epoch": 1.06,
"learning_rate": 3.941194499784059e-05,
"loss": 2.3329,
"step": 76000
},
{
"epoch": 1.07,
"learning_rate": 3.934228674124744e-05,
"loss": 2.3163,
"step": 76500
},
{
"epoch": 1.07,
"learning_rate": 3.9272628484654286e-05,
"loss": 2.3322,
"step": 77000
},
{
"epoch": 1.08,
"learning_rate": 3.9202970228061136e-05,
"loss": 2.3133,
"step": 77500
},
{
"epoch": 1.09,
"learning_rate": 3.913331197146798e-05,
"loss": 2.3388,
"step": 78000
},
{
"epoch": 1.09,
"learning_rate": 3.906365371487482e-05,
"loss": 2.289,
"step": 78500
},
{
"epoch": 1.1,
"learning_rate": 3.899399545828167e-05,
"loss": 2.2948,
"step": 79000
},
{
"epoch": 1.11,
"learning_rate": 3.8924337201688517e-05,
"loss": 2.3597,
"step": 79500
},
{
"epoch": 1.11,
"learning_rate": 3.885467894509536e-05,
"loss": 2.3136,
"step": 80000
},
{
"epoch": 1.12,
"learning_rate": 3.878502068850221e-05,
"loss": 2.3143,
"step": 80500
},
{
"epoch": 1.13,
"learning_rate": 3.871536243190906e-05,
"loss": 2.3085,
"step": 81000
},
{
"epoch": 1.14,
"learning_rate": 3.86457041753159e-05,
"loss": 2.3485,
"step": 81500
},
{
"epoch": 1.14,
"learning_rate": 3.857604591872275e-05,
"loss": 2.2972,
"step": 82000
},
{
"epoch": 1.15,
"learning_rate": 3.85063876621296e-05,
"loss": 2.3323,
"step": 82500
},
{
"epoch": 1.16,
"learning_rate": 3.843672940553644e-05,
"loss": 2.3263,
"step": 83000
},
{
"epoch": 1.16,
"learning_rate": 3.8367071148943284e-05,
"loss": 2.2894,
"step": 83500
},
{
"epoch": 1.17,
"learning_rate": 3.8297412892350135e-05,
"loss": 2.3156,
"step": 84000
},
{
"epoch": 1.18,
"learning_rate": 3.822775463575698e-05,
"loss": 2.3401,
"step": 84500
},
{
"epoch": 1.18,
"learning_rate": 3.815809637916383e-05,
"loss": 2.3034,
"step": 85000
},
{
"epoch": 1.19,
"learning_rate": 3.808843812257067e-05,
"loss": 2.3141,
"step": 85500
},
{
"epoch": 1.2,
"learning_rate": 3.8018779865977515e-05,
"loss": 2.3265,
"step": 86000
},
{
"epoch": 1.21,
"learning_rate": 3.7949121609384365e-05,
"loss": 2.3211,
"step": 86500
},
{
"epoch": 1.21,
"learning_rate": 3.787946335279121e-05,
"loss": 2.3137,
"step": 87000
},
{
"epoch": 1.22,
"learning_rate": 3.780980509619805e-05,
"loss": 2.3216,
"step": 87500
},
{
"epoch": 1.23,
"learning_rate": 3.77401468396049e-05,
"loss": 2.3043,
"step": 88000
},
{
"epoch": 1.23,
"learning_rate": 3.7670488583011746e-05,
"loss": 2.3245,
"step": 88500
},
{
"epoch": 1.24,
"learning_rate": 3.760083032641859e-05,
"loss": 2.3319,
"step": 89000
},
{
"epoch": 1.25,
"learning_rate": 3.753117206982544e-05,
"loss": 2.3025,
"step": 89500
},
{
"epoch": 1.25,
"learning_rate": 3.746151381323228e-05,
"loss": 2.2983,
"step": 90000
},
{
"epoch": 1.26,
"learning_rate": 3.739185555663913e-05,
"loss": 2.3107,
"step": 90500
},
{
"epoch": 1.27,
"learning_rate": 3.732219730004598e-05,
"loss": 2.3137,
"step": 91000
},
{
"epoch": 1.27,
"learning_rate": 3.725253904345282e-05,
"loss": 2.3006,
"step": 91500
},
{
"epoch": 1.28,
"learning_rate": 3.718288078685967e-05,
"loss": 2.3058,
"step": 92000
},
{
"epoch": 1.29,
"learning_rate": 3.7113222530266514e-05,
"loss": 2.3175,
"step": 92500
},
{
"epoch": 1.3,
"learning_rate": 3.704356427367336e-05,
"loss": 2.2991,
"step": 93000
},
{
"epoch": 1.3,
"learning_rate": 3.697390601708021e-05,
"loss": 2.302,
"step": 93500
},
{
"epoch": 1.31,
"learning_rate": 3.690424776048705e-05,
"loss": 2.3089,
"step": 94000
},
{
"epoch": 1.32,
"learning_rate": 3.6834589503893894e-05,
"loss": 2.2957,
"step": 94500
},
{
"epoch": 1.32,
"learning_rate": 3.6764931247300745e-05,
"loss": 2.331,
"step": 95000
},
{
"epoch": 1.33,
"learning_rate": 3.669527299070759e-05,
"loss": 2.3044,
"step": 95500
},
{
"epoch": 1.34,
"learning_rate": 3.662561473411444e-05,
"loss": 2.2887,
"step": 96000
},
{
"epoch": 1.34,
"learning_rate": 3.655595647752128e-05,
"loss": 2.2816,
"step": 96500
},
{
"epoch": 1.35,
"learning_rate": 3.6486298220928125e-05,
"loss": 2.3284,
"step": 97000
},
{
"epoch": 1.36,
"learning_rate": 3.6416639964334975e-05,
"loss": 2.2985,
"step": 97500
},
{
"epoch": 1.37,
"learning_rate": 3.6346981707741826e-05,
"loss": 2.3366,
"step": 98000
},
{
"epoch": 1.37,
"learning_rate": 3.627732345114866e-05,
"loss": 2.3232,
"step": 98500
},
{
"epoch": 1.38,
"learning_rate": 3.620766519455551e-05,
"loss": 2.2867,
"step": 99000
},
{
"epoch": 1.39,
"learning_rate": 3.613800693796236e-05,
"loss": 2.2818,
"step": 99500
},
{
"epoch": 1.39,
"learning_rate": 3.60683486813692e-05,
"loss": 2.3103,
"step": 100000
},
{
"epoch": 1.4,
"learning_rate": 3.599869042477605e-05,
"loss": 2.2815,
"step": 100500
},
{
"epoch": 1.41,
"learning_rate": 3.59290321681829e-05,
"loss": 2.2924,
"step": 101000
},
{
"epoch": 1.41,
"learning_rate": 3.585937391158974e-05,
"loss": 2.3227,
"step": 101500
},
{
"epoch": 1.42,
"learning_rate": 3.578971565499659e-05,
"loss": 2.3177,
"step": 102000
},
{
"epoch": 1.43,
"learning_rate": 3.572005739840344e-05,
"loss": 2.3085,
"step": 102500
},
{
"epoch": 1.43,
"learning_rate": 3.565039914181028e-05,
"loss": 2.2983,
"step": 103000
},
{
"epoch": 1.44,
"learning_rate": 3.558074088521713e-05,
"loss": 2.2718,
"step": 103500
},
{
"epoch": 1.45,
"learning_rate": 3.551108262862397e-05,
"loss": 2.2958,
"step": 104000
},
{
"epoch": 1.46,
"learning_rate": 3.544142437203082e-05,
"loss": 2.3368,
"step": 104500
},
{
"epoch": 1.46,
"learning_rate": 3.537176611543767e-05,
"loss": 2.2821,
"step": 105000
},
{
"epoch": 1.47,
"learning_rate": 3.530210785884451e-05,
"loss": 2.2903,
"step": 105500
},
{
"epoch": 1.48,
"learning_rate": 3.5232449602251355e-05,
"loss": 2.2941,
"step": 106000
},
{
"epoch": 1.48,
"learning_rate": 3.5162791345658205e-05,
"loss": 2.2957,
"step": 106500
},
{
"epoch": 1.49,
"learning_rate": 3.509313308906505e-05,
"loss": 2.2894,
"step": 107000
},
{
"epoch": 1.5,
"learning_rate": 3.502347483247189e-05,
"loss": 2.3138,
"step": 107500
},
{
"epoch": 1.5,
"learning_rate": 3.495381657587874e-05,
"loss": 2.3015,
"step": 108000
},
{
"epoch": 1.51,
"learning_rate": 3.4884158319285585e-05,
"loss": 2.315,
"step": 108500
},
{
"epoch": 1.52,
"learning_rate": 3.4814500062692436e-05,
"loss": 2.2956,
"step": 109000
}
],
"max_steps": 358895,
"num_train_epochs": 5,
"total_flos": 2.305332748437504e+17,
"trial_name": null,
"trial_params": null
}