mT5-large-iwslt2017-de-en / trainer_state.json
Sehoon's picture
store
2e29a7c
{
"best_metric": 19.9878,
"best_model_checkpoint": "ckpt_mt5/google/mt5-large/iwslt2017_de_en/lr5e-05_e49/checkpoint-64410",
"epoch": 7.0,
"global_step": 90174,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 2.9083,
"step": 500
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 1.7423,
"step": 1000
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 1.6399,
"step": 1500
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 1.5877,
"step": 2000
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 1.5372,
"step": 2500
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 1.5221,
"step": 3000
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 1.5048,
"step": 3500
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 1.4684,
"step": 4000
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 1.461,
"step": 4500
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 1.4498,
"step": 5000
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 1.4377,
"step": 5500
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 1.4265,
"step": 6000
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 1.4181,
"step": 6500
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 1.4023,
"step": 7000
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 1.4015,
"step": 7500
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 1.3927,
"step": 8000
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 1.3676,
"step": 8500
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 1.3643,
"step": 9000
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 1.3544,
"step": 9500
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 1.3459,
"step": 10000
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 1.3652,
"step": 10500
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 1.3385,
"step": 11000
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 1.3395,
"step": 11500
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 1.3303,
"step": 12000
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 1.3309,
"step": 12500
},
{
"epoch": 1.0,
"eval_bleu": 19.0434,
"eval_gen_len": 16.8277,
"eval_loss": 1.0267103910446167,
"eval_runtime": 54.7318,
"eval_samples_per_second": 16.225,
"eval_steps_per_second": 1.023,
"step": 12882
},
{
"epoch": 1.01,
"learning_rate": 5e-05,
"loss": 1.3097,
"step": 13000
},
{
"epoch": 1.05,
"learning_rate": 5e-05,
"loss": 1.2515,
"step": 13500
},
{
"epoch": 1.09,
"learning_rate": 5e-05,
"loss": 1.2635,
"step": 14000
},
{
"epoch": 1.13,
"learning_rate": 5e-05,
"loss": 1.2475,
"step": 14500
},
{
"epoch": 1.16,
"learning_rate": 5e-05,
"loss": 1.2404,
"step": 15000
},
{
"epoch": 1.2,
"learning_rate": 5e-05,
"loss": 1.2447,
"step": 15500
},
{
"epoch": 1.24,
"learning_rate": 5e-05,
"loss": 1.2333,
"step": 16000
},
{
"epoch": 1.28,
"learning_rate": 5e-05,
"loss": 1.2481,
"step": 16500
},
{
"epoch": 1.32,
"learning_rate": 5e-05,
"loss": 1.2257,
"step": 17000
},
{
"epoch": 1.36,
"learning_rate": 5e-05,
"loss": 1.2382,
"step": 17500
},
{
"epoch": 1.4,
"learning_rate": 5e-05,
"loss": 1.228,
"step": 18000
},
{
"epoch": 1.44,
"learning_rate": 5e-05,
"loss": 1.2217,
"step": 18500
},
{
"epoch": 1.47,
"learning_rate": 5e-05,
"loss": 1.221,
"step": 19000
},
{
"epoch": 1.51,
"learning_rate": 5e-05,
"loss": 1.2139,
"step": 19500
},
{
"epoch": 1.55,
"learning_rate": 5e-05,
"loss": 1.2143,
"step": 20000
},
{
"epoch": 1.59,
"learning_rate": 5e-05,
"loss": 1.2305,
"step": 20500
},
{
"epoch": 1.63,
"learning_rate": 5e-05,
"loss": 1.2265,
"step": 21000
},
{
"epoch": 1.67,
"learning_rate": 5e-05,
"loss": 1.2177,
"step": 21500
},
{
"epoch": 1.71,
"learning_rate": 5e-05,
"loss": 1.2248,
"step": 22000
},
{
"epoch": 1.75,
"learning_rate": 5e-05,
"loss": 1.2105,
"step": 22500
},
{
"epoch": 1.79,
"learning_rate": 5e-05,
"loss": 1.198,
"step": 23000
},
{
"epoch": 1.82,
"learning_rate": 5e-05,
"loss": 1.2073,
"step": 23500
},
{
"epoch": 1.86,
"learning_rate": 5e-05,
"loss": 1.2087,
"step": 24000
},
{
"epoch": 1.9,
"learning_rate": 5e-05,
"loss": 1.1997,
"step": 24500
},
{
"epoch": 1.94,
"learning_rate": 5e-05,
"loss": 1.1992,
"step": 25000
},
{
"epoch": 1.98,
"learning_rate": 5e-05,
"loss": 1.1944,
"step": 25500
},
{
"epoch": 2.0,
"eval_bleu": 19.4468,
"eval_gen_len": 16.8266,
"eval_loss": 0.9961099624633789,
"eval_runtime": 55.2432,
"eval_samples_per_second": 16.074,
"eval_steps_per_second": 1.014,
"step": 25764
},
{
"epoch": 2.02,
"learning_rate": 5e-05,
"loss": 1.1576,
"step": 26000
},
{
"epoch": 2.06,
"learning_rate": 5e-05,
"loss": 1.1239,
"step": 26500
},
{
"epoch": 2.1,
"learning_rate": 5e-05,
"loss": 1.1292,
"step": 27000
},
{
"epoch": 2.13,
"learning_rate": 5e-05,
"loss": 1.1238,
"step": 27500
},
{
"epoch": 2.17,
"learning_rate": 5e-05,
"loss": 1.1194,
"step": 28000
},
{
"epoch": 2.21,
"learning_rate": 5e-05,
"loss": 1.1326,
"step": 28500
},
{
"epoch": 2.25,
"learning_rate": 5e-05,
"loss": 1.1292,
"step": 29000
},
{
"epoch": 2.29,
"learning_rate": 5e-05,
"loss": 1.1089,
"step": 29500
},
{
"epoch": 2.33,
"learning_rate": 5e-05,
"loss": 1.1271,
"step": 30000
},
{
"epoch": 2.37,
"learning_rate": 5e-05,
"loss": 1.1199,
"step": 30500
},
{
"epoch": 2.41,
"learning_rate": 5e-05,
"loss": 1.1279,
"step": 31000
},
{
"epoch": 2.45,
"learning_rate": 5e-05,
"loss": 1.124,
"step": 31500
},
{
"epoch": 2.48,
"learning_rate": 5e-05,
"loss": 1.1188,
"step": 32000
},
{
"epoch": 2.52,
"learning_rate": 5e-05,
"loss": 1.1251,
"step": 32500
},
{
"epoch": 2.56,
"learning_rate": 5e-05,
"loss": 1.1231,
"step": 33000
},
{
"epoch": 2.6,
"learning_rate": 5e-05,
"loss": 1.1223,
"step": 33500
},
{
"epoch": 2.64,
"learning_rate": 5e-05,
"loss": 1.1198,
"step": 34000
},
{
"epoch": 2.68,
"learning_rate": 5e-05,
"loss": 1.1248,
"step": 34500
},
{
"epoch": 2.72,
"learning_rate": 5e-05,
"loss": 1.1106,
"step": 35000
},
{
"epoch": 2.76,
"learning_rate": 5e-05,
"loss": 1.1171,
"step": 35500
},
{
"epoch": 2.79,
"learning_rate": 5e-05,
"loss": 1.1133,
"step": 36000
},
{
"epoch": 2.83,
"learning_rate": 5e-05,
"loss": 1.1254,
"step": 36500
},
{
"epoch": 2.87,
"learning_rate": 5e-05,
"loss": 1.1141,
"step": 37000
},
{
"epoch": 2.91,
"learning_rate": 5e-05,
"loss": 1.1238,
"step": 37500
},
{
"epoch": 2.95,
"learning_rate": 5e-05,
"loss": 1.1058,
"step": 38000
},
{
"epoch": 2.99,
"learning_rate": 5e-05,
"loss": 1.1119,
"step": 38500
},
{
"epoch": 3.0,
"eval_bleu": 19.9654,
"eval_gen_len": 16.8525,
"eval_loss": 0.9805649518966675,
"eval_runtime": 55.4985,
"eval_samples_per_second": 16.0,
"eval_steps_per_second": 1.009,
"step": 38646
},
{
"epoch": 3.03,
"learning_rate": 5e-05,
"loss": 1.0585,
"step": 39000
},
{
"epoch": 3.07,
"learning_rate": 5e-05,
"loss": 1.0476,
"step": 39500
},
{
"epoch": 3.11,
"learning_rate": 5e-05,
"loss": 1.0405,
"step": 40000
},
{
"epoch": 3.14,
"learning_rate": 5e-05,
"loss": 1.0549,
"step": 40500
},
{
"epoch": 3.18,
"learning_rate": 5e-05,
"loss": 1.0379,
"step": 41000
},
{
"epoch": 3.22,
"learning_rate": 5e-05,
"loss": 1.049,
"step": 41500
},
{
"epoch": 3.26,
"learning_rate": 5e-05,
"loss": 1.0367,
"step": 42000
},
{
"epoch": 3.3,
"learning_rate": 5e-05,
"loss": 1.0553,
"step": 42500
},
{
"epoch": 3.34,
"learning_rate": 5e-05,
"loss": 1.0408,
"step": 43000
},
{
"epoch": 3.38,
"learning_rate": 5e-05,
"loss": 1.041,
"step": 43500
},
{
"epoch": 3.42,
"learning_rate": 5e-05,
"loss": 1.0516,
"step": 44000
},
{
"epoch": 3.45,
"learning_rate": 5e-05,
"loss": 1.0448,
"step": 44500
},
{
"epoch": 3.49,
"learning_rate": 5e-05,
"loss": 1.0512,
"step": 45000
},
{
"epoch": 3.53,
"learning_rate": 5e-05,
"loss": 1.0487,
"step": 45500
},
{
"epoch": 3.57,
"learning_rate": 5e-05,
"loss": 1.0504,
"step": 46000
},
{
"epoch": 3.61,
"learning_rate": 5e-05,
"loss": 1.0436,
"step": 46500
},
{
"epoch": 3.65,
"learning_rate": 5e-05,
"loss": 1.0439,
"step": 47000
},
{
"epoch": 3.69,
"learning_rate": 5e-05,
"loss": 1.0386,
"step": 47500
},
{
"epoch": 3.73,
"learning_rate": 5e-05,
"loss": 1.0413,
"step": 48000
},
{
"epoch": 3.76,
"learning_rate": 5e-05,
"loss": 1.0512,
"step": 48500
},
{
"epoch": 3.8,
"learning_rate": 5e-05,
"loss": 1.0544,
"step": 49000
},
{
"epoch": 3.84,
"learning_rate": 5e-05,
"loss": 1.042,
"step": 49500
},
{
"epoch": 3.88,
"learning_rate": 5e-05,
"loss": 1.0411,
"step": 50000
},
{
"epoch": 3.92,
"learning_rate": 5e-05,
"loss": 1.0501,
"step": 50500
},
{
"epoch": 3.96,
"learning_rate": 5e-05,
"loss": 1.0397,
"step": 51000
},
{
"epoch": 4.0,
"learning_rate": 5e-05,
"loss": 1.0495,
"step": 51500
},
{
"epoch": 4.0,
"eval_bleu": 19.9788,
"eval_gen_len": 16.8604,
"eval_loss": 0.9743499159812927,
"eval_runtime": 56.6985,
"eval_samples_per_second": 15.662,
"eval_steps_per_second": 0.988,
"step": 51528
},
{
"epoch": 4.04,
"learning_rate": 5e-05,
"loss": 0.9885,
"step": 52000
},
{
"epoch": 4.08,
"learning_rate": 5e-05,
"loss": 0.9636,
"step": 52500
},
{
"epoch": 4.11,
"learning_rate": 5e-05,
"loss": 0.9842,
"step": 53000
},
{
"epoch": 4.15,
"learning_rate": 5e-05,
"loss": 0.9882,
"step": 53500
},
{
"epoch": 4.19,
"learning_rate": 5e-05,
"loss": 0.9787,
"step": 54000
},
{
"epoch": 4.23,
"learning_rate": 5e-05,
"loss": 0.9809,
"step": 54500
},
{
"epoch": 4.27,
"learning_rate": 5e-05,
"loss": 0.9807,
"step": 55000
},
{
"epoch": 4.31,
"learning_rate": 5e-05,
"loss": 0.9824,
"step": 55500
},
{
"epoch": 4.35,
"learning_rate": 5e-05,
"loss": 0.9815,
"step": 56000
},
{
"epoch": 4.39,
"learning_rate": 5e-05,
"loss": 0.9763,
"step": 56500
},
{
"epoch": 4.42,
"learning_rate": 5e-05,
"loss": 0.9809,
"step": 57000
},
{
"epoch": 4.46,
"learning_rate": 5e-05,
"loss": 0.9885,
"step": 57500
},
{
"epoch": 4.5,
"learning_rate": 5e-05,
"loss": 0.9825,
"step": 58000
},
{
"epoch": 4.54,
"learning_rate": 5e-05,
"loss": 0.9856,
"step": 58500
},
{
"epoch": 4.58,
"learning_rate": 5e-05,
"loss": 0.9848,
"step": 59000
},
{
"epoch": 4.62,
"learning_rate": 5e-05,
"loss": 0.9876,
"step": 59500
},
{
"epoch": 4.66,
"learning_rate": 5e-05,
"loss": 0.9834,
"step": 60000
},
{
"epoch": 4.7,
"learning_rate": 5e-05,
"loss": 0.9939,
"step": 60500
},
{
"epoch": 4.74,
"learning_rate": 5e-05,
"loss": 0.9845,
"step": 61000
},
{
"epoch": 4.77,
"learning_rate": 5e-05,
"loss": 0.9858,
"step": 61500
},
{
"epoch": 4.81,
"learning_rate": 5e-05,
"loss": 0.9885,
"step": 62000
},
{
"epoch": 4.85,
"learning_rate": 5e-05,
"loss": 0.9971,
"step": 62500
},
{
"epoch": 4.89,
"learning_rate": 5e-05,
"loss": 0.9867,
"step": 63000
},
{
"epoch": 4.93,
"learning_rate": 5e-05,
"loss": 0.9941,
"step": 63500
},
{
"epoch": 4.97,
"learning_rate": 5e-05,
"loss": 0.9849,
"step": 64000
},
{
"epoch": 5.0,
"eval_bleu": 19.9878,
"eval_gen_len": 16.8468,
"eval_loss": 0.9771544337272644,
"eval_runtime": 56.8362,
"eval_samples_per_second": 15.624,
"eval_steps_per_second": 0.985,
"step": 64410
},
{
"epoch": 5.01,
"learning_rate": 5e-05,
"loss": 0.9758,
"step": 64500
},
{
"epoch": 5.05,
"learning_rate": 5e-05,
"loss": 0.9183,
"step": 65000
},
{
"epoch": 5.08,
"learning_rate": 5e-05,
"loss": 0.9121,
"step": 65500
},
{
"epoch": 5.12,
"learning_rate": 5e-05,
"loss": 0.9197,
"step": 66000
},
{
"epoch": 5.16,
"learning_rate": 5e-05,
"loss": 0.926,
"step": 66500
},
{
"epoch": 5.2,
"learning_rate": 5e-05,
"loss": 0.9233,
"step": 67000
},
{
"epoch": 5.24,
"learning_rate": 5e-05,
"loss": 0.9263,
"step": 67500
},
{
"epoch": 5.28,
"learning_rate": 5e-05,
"loss": 0.9316,
"step": 68000
},
{
"epoch": 5.32,
"learning_rate": 5e-05,
"loss": 0.9251,
"step": 68500
},
{
"epoch": 5.36,
"learning_rate": 5e-05,
"loss": 0.9309,
"step": 69000
},
{
"epoch": 5.4,
"learning_rate": 5e-05,
"loss": 0.9261,
"step": 69500
},
{
"epoch": 5.43,
"learning_rate": 5e-05,
"loss": 0.931,
"step": 70000
},
{
"epoch": 5.47,
"learning_rate": 5e-05,
"loss": 0.9367,
"step": 70500
},
{
"epoch": 5.51,
"learning_rate": 5e-05,
"loss": 0.9347,
"step": 71000
},
{
"epoch": 5.55,
"learning_rate": 5e-05,
"loss": 0.9384,
"step": 71500
},
{
"epoch": 5.59,
"learning_rate": 5e-05,
"loss": 0.9334,
"step": 72000
},
{
"epoch": 5.63,
"learning_rate": 5e-05,
"loss": 0.9353,
"step": 72500
},
{
"epoch": 5.67,
"learning_rate": 5e-05,
"loss": 0.929,
"step": 73000
},
{
"epoch": 5.71,
"learning_rate": 5e-05,
"loss": 0.9419,
"step": 73500
},
{
"epoch": 5.74,
"learning_rate": 5e-05,
"loss": 0.9353,
"step": 74000
},
{
"epoch": 5.78,
"learning_rate": 5e-05,
"loss": 0.9302,
"step": 74500
},
{
"epoch": 5.82,
"learning_rate": 5e-05,
"loss": 0.9448,
"step": 75000
},
{
"epoch": 5.86,
"learning_rate": 5e-05,
"loss": 0.9277,
"step": 75500
},
{
"epoch": 5.9,
"learning_rate": 5e-05,
"loss": 0.9325,
"step": 76000
},
{
"epoch": 5.94,
"learning_rate": 5e-05,
"loss": 0.9451,
"step": 76500
},
{
"epoch": 5.98,
"learning_rate": 5e-05,
"loss": 0.9336,
"step": 77000
},
{
"epoch": 6.0,
"eval_bleu": 19.7392,
"eval_gen_len": 16.7646,
"eval_loss": 0.9833778738975525,
"eval_runtime": 58.8288,
"eval_samples_per_second": 15.095,
"eval_steps_per_second": 0.952,
"step": 77292
},
{
"epoch": 6.02,
"learning_rate": 5e-05,
"loss": 0.9089,
"step": 77500
},
{
"epoch": 6.05,
"learning_rate": 5e-05,
"loss": 0.8559,
"step": 78000
},
{
"epoch": 6.09,
"learning_rate": 5e-05,
"loss": 0.8688,
"step": 78500
},
{
"epoch": 6.13,
"learning_rate": 5e-05,
"loss": 0.872,
"step": 79000
},
{
"epoch": 6.17,
"learning_rate": 5e-05,
"loss": 0.8789,
"step": 79500
},
{
"epoch": 6.21,
"learning_rate": 5e-05,
"loss": 0.8814,
"step": 80000
},
{
"epoch": 6.25,
"learning_rate": 5e-05,
"loss": 0.8721,
"step": 80500
},
{
"epoch": 6.29,
"learning_rate": 5e-05,
"loss": 0.8864,
"step": 81000
},
{
"epoch": 6.33,
"learning_rate": 5e-05,
"loss": 0.8875,
"step": 81500
},
{
"epoch": 6.37,
"learning_rate": 5e-05,
"loss": 0.8751,
"step": 82000
},
{
"epoch": 6.4,
"learning_rate": 5e-05,
"loss": 0.8772,
"step": 82500
},
{
"epoch": 6.44,
"learning_rate": 5e-05,
"loss": 0.8742,
"step": 83000
},
{
"epoch": 6.48,
"learning_rate": 5e-05,
"loss": 0.8773,
"step": 83500
},
{
"epoch": 6.52,
"learning_rate": 5e-05,
"loss": 0.8783,
"step": 84000
},
{
"epoch": 6.56,
"learning_rate": 5e-05,
"loss": 0.8845,
"step": 84500
},
{
"epoch": 6.6,
"learning_rate": 5e-05,
"loss": 0.8718,
"step": 85000
},
{
"epoch": 6.64,
"learning_rate": 5e-05,
"loss": 0.8871,
"step": 85500
},
{
"epoch": 6.68,
"learning_rate": 5e-05,
"loss": 0.8834,
"step": 86000
},
{
"epoch": 6.71,
"learning_rate": 5e-05,
"loss": 0.8979,
"step": 86500
},
{
"epoch": 6.75,
"learning_rate": 5e-05,
"loss": 0.8875,
"step": 87000
},
{
"epoch": 6.79,
"learning_rate": 5e-05,
"loss": 0.8959,
"step": 87500
},
{
"epoch": 6.83,
"learning_rate": 5e-05,
"loss": 0.8874,
"step": 88000
},
{
"epoch": 6.87,
"learning_rate": 5e-05,
"loss": 0.8817,
"step": 88500
},
{
"epoch": 6.91,
"learning_rate": 5e-05,
"loss": 0.8894,
"step": 89000
},
{
"epoch": 6.95,
"learning_rate": 5e-05,
"loss": 0.8925,
"step": 89500
},
{
"epoch": 6.99,
"learning_rate": 5e-05,
"loss": 0.8903,
"step": 90000
},
{
"epoch": 7.0,
"eval_bleu": 19.9464,
"eval_gen_len": 16.8491,
"eval_loss": 0.9870715737342834,
"eval_runtime": 56.1043,
"eval_samples_per_second": 15.828,
"eval_steps_per_second": 0.998,
"step": 90174
}
],
"max_steps": 631218,
"num_train_epochs": 49,
"total_flos": 6.276153481836626e+17,
"trial_name": null,
"trial_params": null
}