qcpg-parabk2-sbert-lr1e-4 / trainer_state.json
madhavsankar's picture
Upload 12 files
facf72a
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.0,
"global_step": 337500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 0.0001,
"loss": 1.6466,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 0.0001,
"loss": 1.3028,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 0.0001,
"loss": 1.2357,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 0.0001,
"loss": 1.2009,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 0.0001,
"loss": 1.1854,
"step": 2500
},
{
"epoch": 0.05,
"learning_rate": 0.0001,
"loss": 1.1657,
"step": 3000
},
{
"epoch": 0.06,
"learning_rate": 0.0001,
"loss": 1.1535,
"step": 3500
},
{
"epoch": 0.07,
"learning_rate": 0.0001,
"loss": 1.1177,
"step": 4000
},
{
"epoch": 0.08,
"learning_rate": 0.0001,
"loss": 1.1035,
"step": 4500
},
{
"epoch": 0.09,
"learning_rate": 0.0001,
"loss": 1.1204,
"step": 5000
},
{
"epoch": 0.1,
"learning_rate": 0.0001,
"loss": 1.1042,
"step": 5500
},
{
"epoch": 0.11,
"learning_rate": 0.0001,
"loss": 1.0938,
"step": 6000
},
{
"epoch": 0.12,
"learning_rate": 0.0001,
"loss": 1.0782,
"step": 6500
},
{
"epoch": 0.12,
"learning_rate": 0.0001,
"loss": 1.0673,
"step": 7000
},
{
"epoch": 0.13,
"learning_rate": 0.0001,
"loss": 1.0629,
"step": 7500
},
{
"epoch": 0.14,
"learning_rate": 0.0001,
"loss": 1.0525,
"step": 8000
},
{
"epoch": 0.15,
"learning_rate": 0.0001,
"loss": 1.0601,
"step": 8500
},
{
"epoch": 0.16,
"learning_rate": 0.0001,
"loss": 1.0417,
"step": 9000
},
{
"epoch": 0.17,
"learning_rate": 0.0001,
"loss": 1.042,
"step": 9500
},
{
"epoch": 0.18,
"learning_rate": 0.0001,
"loss": 1.0324,
"step": 10000
},
{
"epoch": 0.19,
"learning_rate": 0.0001,
"loss": 1.0258,
"step": 10500
},
{
"epoch": 0.2,
"learning_rate": 0.0001,
"loss": 1.0202,
"step": 11000
},
{
"epoch": 0.2,
"learning_rate": 0.0001,
"loss": 1.0232,
"step": 11500
},
{
"epoch": 0.21,
"learning_rate": 0.0001,
"loss": 1.0247,
"step": 12000
},
{
"epoch": 0.22,
"learning_rate": 0.0001,
"loss": 1.0115,
"step": 12500
},
{
"epoch": 0.23,
"learning_rate": 0.0001,
"loss": 1.0048,
"step": 13000
},
{
"epoch": 0.24,
"learning_rate": 0.0001,
"loss": 1.0002,
"step": 13500
},
{
"epoch": 0.25,
"learning_rate": 0.0001,
"loss": 1.0041,
"step": 14000
},
{
"epoch": 0.26,
"learning_rate": 0.0001,
"loss": 0.9986,
"step": 14500
},
{
"epoch": 0.27,
"learning_rate": 0.0001,
"loss": 0.9897,
"step": 15000
},
{
"epoch": 0.28,
"learning_rate": 0.0001,
"loss": 0.9839,
"step": 15500
},
{
"epoch": 0.28,
"learning_rate": 0.0001,
"loss": 0.9838,
"step": 16000
},
{
"epoch": 0.29,
"learning_rate": 0.0001,
"loss": 0.9673,
"step": 16500
},
{
"epoch": 0.3,
"learning_rate": 0.0001,
"loss": 0.9832,
"step": 17000
},
{
"epoch": 0.31,
"learning_rate": 0.0001,
"loss": 0.9699,
"step": 17500
},
{
"epoch": 0.32,
"learning_rate": 0.0001,
"loss": 0.9598,
"step": 18000
},
{
"epoch": 0.33,
"learning_rate": 0.0001,
"loss": 0.9716,
"step": 18500
},
{
"epoch": 0.34,
"learning_rate": 0.0001,
"loss": 0.9631,
"step": 19000
},
{
"epoch": 0.35,
"learning_rate": 0.0001,
"loss": 0.9453,
"step": 19500
},
{
"epoch": 0.36,
"learning_rate": 0.0001,
"loss": 0.9591,
"step": 20000
},
{
"epoch": 0.36,
"learning_rate": 0.0001,
"loss": 0.9564,
"step": 20500
},
{
"epoch": 0.37,
"learning_rate": 0.0001,
"loss": 0.9462,
"step": 21000
},
{
"epoch": 0.38,
"learning_rate": 0.0001,
"loss": 0.9386,
"step": 21500
},
{
"epoch": 0.39,
"learning_rate": 0.0001,
"loss": 0.9359,
"step": 22000
},
{
"epoch": 0.4,
"learning_rate": 0.0001,
"loss": 0.9252,
"step": 22500
},
{
"epoch": 0.41,
"learning_rate": 0.0001,
"loss": 0.9284,
"step": 23000
},
{
"epoch": 0.42,
"learning_rate": 0.0001,
"loss": 0.9256,
"step": 23500
},
{
"epoch": 0.43,
"learning_rate": 0.0001,
"loss": 0.9236,
"step": 24000
},
{
"epoch": 0.44,
"learning_rate": 0.0001,
"loss": 0.9108,
"step": 24500
},
{
"epoch": 0.44,
"learning_rate": 0.0001,
"loss": 0.9228,
"step": 25000
},
{
"epoch": 0.45,
"learning_rate": 0.0001,
"loss": 0.923,
"step": 25500
},
{
"epoch": 0.46,
"learning_rate": 0.0001,
"loss": 0.916,
"step": 26000
},
{
"epoch": 0.47,
"learning_rate": 0.0001,
"loss": 0.9148,
"step": 26500
},
{
"epoch": 0.48,
"learning_rate": 0.0001,
"loss": 0.9073,
"step": 27000
},
{
"epoch": 0.49,
"learning_rate": 0.0001,
"loss": 0.9127,
"step": 27500
},
{
"epoch": 0.5,
"learning_rate": 0.0001,
"loss": 0.9025,
"step": 28000
},
{
"epoch": 0.51,
"learning_rate": 0.0001,
"loss": 0.9004,
"step": 28500
},
{
"epoch": 0.52,
"learning_rate": 0.0001,
"loss": 0.9,
"step": 29000
},
{
"epoch": 0.52,
"learning_rate": 0.0001,
"loss": 0.8953,
"step": 29500
},
{
"epoch": 0.53,
"learning_rate": 0.0001,
"loss": 0.8934,
"step": 30000
},
{
"epoch": 0.54,
"learning_rate": 0.0001,
"loss": 0.9058,
"step": 30500
},
{
"epoch": 0.55,
"learning_rate": 0.0001,
"loss": 0.8879,
"step": 31000
},
{
"epoch": 0.56,
"learning_rate": 0.0001,
"loss": 0.8854,
"step": 31500
},
{
"epoch": 0.57,
"learning_rate": 0.0001,
"loss": 0.8748,
"step": 32000
},
{
"epoch": 0.58,
"learning_rate": 0.0001,
"loss": 0.8872,
"step": 32500
},
{
"epoch": 0.59,
"learning_rate": 0.0001,
"loss": 0.8815,
"step": 33000
},
{
"epoch": 0.6,
"learning_rate": 0.0001,
"loss": 0.8901,
"step": 33500
},
{
"epoch": 0.6,
"learning_rate": 0.0001,
"loss": 0.8732,
"step": 34000
},
{
"epoch": 0.61,
"learning_rate": 0.0001,
"loss": 0.8822,
"step": 34500
},
{
"epoch": 0.62,
"learning_rate": 0.0001,
"loss": 0.8726,
"step": 35000
},
{
"epoch": 0.63,
"learning_rate": 0.0001,
"loss": 0.8695,
"step": 35500
},
{
"epoch": 0.64,
"learning_rate": 0.0001,
"loss": 0.8571,
"step": 36000
},
{
"epoch": 0.65,
"learning_rate": 0.0001,
"loss": 0.872,
"step": 36500
},
{
"epoch": 0.66,
"learning_rate": 0.0001,
"loss": 0.8629,
"step": 37000
},
{
"epoch": 0.67,
"learning_rate": 0.0001,
"loss": 0.8602,
"step": 37500
},
{
"epoch": 0.68,
"learning_rate": 0.0001,
"loss": 0.8649,
"step": 38000
},
{
"epoch": 0.68,
"learning_rate": 0.0001,
"loss": 0.8608,
"step": 38500
},
{
"epoch": 0.69,
"learning_rate": 0.0001,
"loss": 0.8579,
"step": 39000
},
{
"epoch": 0.7,
"learning_rate": 0.0001,
"loss": 0.8529,
"step": 39500
},
{
"epoch": 0.71,
"learning_rate": 0.0001,
"loss": 0.8529,
"step": 40000
},
{
"epoch": 0.72,
"learning_rate": 0.0001,
"loss": 0.8512,
"step": 40500
},
{
"epoch": 0.73,
"learning_rate": 0.0001,
"loss": 0.8479,
"step": 41000
},
{
"epoch": 0.74,
"learning_rate": 0.0001,
"loss": 0.8479,
"step": 41500
},
{
"epoch": 0.75,
"learning_rate": 0.0001,
"loss": 0.8399,
"step": 42000
},
{
"epoch": 0.76,
"learning_rate": 0.0001,
"loss": 0.8598,
"step": 42500
},
{
"epoch": 0.76,
"learning_rate": 0.0001,
"loss": 0.8481,
"step": 43000
},
{
"epoch": 0.77,
"learning_rate": 0.0001,
"loss": 0.8263,
"step": 43500
},
{
"epoch": 0.78,
"learning_rate": 0.0001,
"loss": 0.8375,
"step": 44000
},
{
"epoch": 0.79,
"learning_rate": 0.0001,
"loss": 0.8458,
"step": 44500
},
{
"epoch": 0.8,
"learning_rate": 0.0001,
"loss": 0.8417,
"step": 45000
},
{
"epoch": 0.81,
"learning_rate": 0.0001,
"loss": 0.8409,
"step": 45500
},
{
"epoch": 0.82,
"learning_rate": 0.0001,
"loss": 0.8422,
"step": 46000
},
{
"epoch": 0.83,
"learning_rate": 0.0001,
"loss": 0.8315,
"step": 46500
},
{
"epoch": 0.84,
"learning_rate": 0.0001,
"loss": 0.8342,
"step": 47000
},
{
"epoch": 0.84,
"learning_rate": 0.0001,
"loss": 0.8132,
"step": 47500
},
{
"epoch": 0.85,
"learning_rate": 0.0001,
"loss": 0.8279,
"step": 48000
},
{
"epoch": 0.86,
"learning_rate": 0.0001,
"loss": 0.8248,
"step": 48500
},
{
"epoch": 0.87,
"learning_rate": 0.0001,
"loss": 0.8186,
"step": 49000
},
{
"epoch": 0.88,
"learning_rate": 0.0001,
"loss": 0.834,
"step": 49500
},
{
"epoch": 0.89,
"learning_rate": 0.0001,
"loss": 0.821,
"step": 50000
},
{
"epoch": 0.9,
"learning_rate": 0.0001,
"loss": 0.8132,
"step": 50500
},
{
"epoch": 0.91,
"learning_rate": 0.0001,
"loss": 0.8173,
"step": 51000
},
{
"epoch": 0.92,
"learning_rate": 0.0001,
"loss": 0.8093,
"step": 51500
},
{
"epoch": 0.92,
"learning_rate": 0.0001,
"loss": 0.8149,
"step": 52000
},
{
"epoch": 0.93,
"learning_rate": 0.0001,
"loss": 0.8037,
"step": 52500
},
{
"epoch": 0.94,
"learning_rate": 0.0001,
"loss": 0.8117,
"step": 53000
},
{
"epoch": 0.95,
"learning_rate": 0.0001,
"loss": 0.7952,
"step": 53500
},
{
"epoch": 0.96,
"learning_rate": 0.0001,
"loss": 0.8053,
"step": 54000
},
{
"epoch": 0.97,
"learning_rate": 0.0001,
"loss": 0.8129,
"step": 54500
},
{
"epoch": 0.98,
"learning_rate": 0.0001,
"loss": 0.8049,
"step": 55000
},
{
"epoch": 0.99,
"learning_rate": 0.0001,
"loss": 0.7965,
"step": 55500
},
{
"epoch": 1.0,
"learning_rate": 0.0001,
"loss": 0.8058,
"step": 56000
},
{
"epoch": 1.0,
"eval_bleu": 27.3819,
"eval_gen_len": 11.0991,
"eval_loss": 0.9401872754096985,
"eval_runtime": 720.1583,
"eval_samples_per_second": 19.44,
"step": 56250
},
{
"epoch": 1.0,
"learning_rate": 0.0001,
"loss": 0.7812,
"step": 56500
},
{
"epoch": 1.01,
"learning_rate": 0.0001,
"loss": 0.7532,
"step": 57000
},
{
"epoch": 1.02,
"learning_rate": 0.0001,
"loss": 0.7441,
"step": 57500
},
{
"epoch": 1.03,
"learning_rate": 0.0001,
"loss": 0.7592,
"step": 58000
},
{
"epoch": 1.04,
"learning_rate": 0.0001,
"loss": 0.7553,
"step": 58500
},
{
"epoch": 1.05,
"learning_rate": 0.0001,
"loss": 0.7495,
"step": 59000
},
{
"epoch": 1.06,
"learning_rate": 0.0001,
"loss": 0.7532,
"step": 59500
},
{
"epoch": 1.07,
"learning_rate": 0.0001,
"loss": 0.743,
"step": 60000
},
{
"epoch": 1.08,
"learning_rate": 0.0001,
"loss": 0.7429,
"step": 60500
},
{
"epoch": 1.08,
"learning_rate": 0.0001,
"loss": 0.7572,
"step": 61000
},
{
"epoch": 1.09,
"learning_rate": 0.0001,
"loss": 0.7431,
"step": 61500
},
{
"epoch": 1.1,
"learning_rate": 0.0001,
"loss": 0.738,
"step": 62000
},
{
"epoch": 1.11,
"learning_rate": 0.0001,
"loss": 0.7448,
"step": 62500
},
{
"epoch": 1.12,
"learning_rate": 0.0001,
"loss": 0.7441,
"step": 63000
},
{
"epoch": 1.13,
"learning_rate": 0.0001,
"loss": 0.7371,
"step": 63500
},
{
"epoch": 1.14,
"learning_rate": 0.0001,
"loss": 0.7486,
"step": 64000
},
{
"epoch": 1.15,
"learning_rate": 0.0001,
"loss": 0.7401,
"step": 64500
},
{
"epoch": 1.16,
"learning_rate": 0.0001,
"loss": 0.7467,
"step": 65000
},
{
"epoch": 1.16,
"learning_rate": 0.0001,
"loss": 0.7277,
"step": 65500
},
{
"epoch": 1.17,
"learning_rate": 0.0001,
"loss": 0.723,
"step": 66000
},
{
"epoch": 1.18,
"learning_rate": 0.0001,
"loss": 0.7261,
"step": 66500
},
{
"epoch": 1.19,
"learning_rate": 0.0001,
"loss": 0.7262,
"step": 67000
},
{
"epoch": 1.2,
"learning_rate": 0.0001,
"loss": 0.7248,
"step": 67500
},
{
"epoch": 1.21,
"learning_rate": 0.0001,
"loss": 0.7137,
"step": 68000
},
{
"epoch": 1.22,
"learning_rate": 0.0001,
"loss": 0.7215,
"step": 68500
},
{
"epoch": 1.23,
"learning_rate": 0.0001,
"loss": 0.726,
"step": 69000
},
{
"epoch": 1.24,
"learning_rate": 0.0001,
"loss": 0.7209,
"step": 69500
},
{
"epoch": 1.24,
"learning_rate": 0.0001,
"loss": 0.7266,
"step": 70000
},
{
"epoch": 1.25,
"learning_rate": 0.0001,
"loss": 0.7253,
"step": 70500
},
{
"epoch": 1.26,
"learning_rate": 0.0001,
"loss": 0.7189,
"step": 71000
},
{
"epoch": 1.27,
"learning_rate": 0.0001,
"loss": 0.722,
"step": 71500
},
{
"epoch": 1.28,
"learning_rate": 0.0001,
"loss": 0.7258,
"step": 72000
},
{
"epoch": 1.29,
"learning_rate": 0.0001,
"loss": 0.7173,
"step": 72500
},
{
"epoch": 1.3,
"learning_rate": 0.0001,
"loss": 0.7257,
"step": 73000
},
{
"epoch": 1.31,
"learning_rate": 0.0001,
"loss": 0.7233,
"step": 73500
},
{
"epoch": 1.32,
"learning_rate": 0.0001,
"loss": 0.7303,
"step": 74000
},
{
"epoch": 1.32,
"learning_rate": 0.0001,
"loss": 0.7097,
"step": 74500
},
{
"epoch": 1.33,
"learning_rate": 0.0001,
"loss": 0.7245,
"step": 75000
},
{
"epoch": 1.34,
"learning_rate": 0.0001,
"loss": 0.7174,
"step": 75500
},
{
"epoch": 1.35,
"learning_rate": 0.0001,
"loss": 0.7163,
"step": 76000
},
{
"epoch": 1.36,
"learning_rate": 0.0001,
"loss": 0.7103,
"step": 76500
},
{
"epoch": 1.37,
"learning_rate": 0.0001,
"loss": 0.7252,
"step": 77000
},
{
"epoch": 1.38,
"learning_rate": 0.0001,
"loss": 0.7096,
"step": 77500
},
{
"epoch": 1.39,
"learning_rate": 0.0001,
"loss": 0.7087,
"step": 78000
},
{
"epoch": 1.4,
"learning_rate": 0.0001,
"loss": 0.7158,
"step": 78500
},
{
"epoch": 1.4,
"learning_rate": 0.0001,
"loss": 0.7081,
"step": 79000
},
{
"epoch": 1.41,
"learning_rate": 0.0001,
"loss": 0.7161,
"step": 79500
},
{
"epoch": 1.42,
"learning_rate": 0.0001,
"loss": 0.7083,
"step": 80000
},
{
"epoch": 1.43,
"learning_rate": 0.0001,
"loss": 0.7089,
"step": 80500
},
{
"epoch": 1.44,
"learning_rate": 0.0001,
"loss": 0.7056,
"step": 81000
},
{
"epoch": 1.45,
"learning_rate": 0.0001,
"loss": 0.7085,
"step": 81500
},
{
"epoch": 1.46,
"learning_rate": 0.0001,
"loss": 0.7089,
"step": 82000
},
{
"epoch": 1.47,
"learning_rate": 0.0001,
"loss": 0.7062,
"step": 82500
},
{
"epoch": 1.48,
"learning_rate": 0.0001,
"loss": 0.6948,
"step": 83000
},
{
"epoch": 1.48,
"learning_rate": 0.0001,
"loss": 0.7064,
"step": 83500
},
{
"epoch": 1.49,
"learning_rate": 0.0001,
"loss": 0.6978,
"step": 84000
},
{
"epoch": 1.5,
"learning_rate": 0.0001,
"loss": 0.7055,
"step": 84500
},
{
"epoch": 1.51,
"learning_rate": 0.0001,
"loss": 0.6973,
"step": 85000
},
{
"epoch": 1.52,
"learning_rate": 0.0001,
"loss": 0.6948,
"step": 85500
},
{
"epoch": 1.53,
"learning_rate": 0.0001,
"loss": 0.6982,
"step": 86000
},
{
"epoch": 1.54,
"learning_rate": 0.0001,
"loss": 0.6973,
"step": 86500
},
{
"epoch": 1.55,
"learning_rate": 0.0001,
"loss": 0.6986,
"step": 87000
},
{
"epoch": 1.56,
"learning_rate": 0.0001,
"loss": 0.6964,
"step": 87500
},
{
"epoch": 1.56,
"learning_rate": 0.0001,
"loss": 0.6968,
"step": 88000
},
{
"epoch": 1.57,
"learning_rate": 0.0001,
"loss": 0.7013,
"step": 88500
},
{
"epoch": 1.58,
"learning_rate": 0.0001,
"loss": 0.6974,
"step": 89000
},
{
"epoch": 1.59,
"learning_rate": 0.0001,
"loss": 0.6921,
"step": 89500
},
{
"epoch": 1.6,
"learning_rate": 0.0001,
"loss": 0.6989,
"step": 90000
},
{
"epoch": 1.61,
"learning_rate": 0.0001,
"loss": 0.7024,
"step": 90500
},
{
"epoch": 1.62,
"learning_rate": 0.0001,
"loss": 0.6968,
"step": 91000
},
{
"epoch": 1.63,
"learning_rate": 0.0001,
"loss": 0.6997,
"step": 91500
},
{
"epoch": 1.64,
"learning_rate": 0.0001,
"loss": 0.6999,
"step": 92000
},
{
"epoch": 1.64,
"learning_rate": 0.0001,
"loss": 0.6892,
"step": 92500
},
{
"epoch": 1.65,
"learning_rate": 0.0001,
"loss": 0.6888,
"step": 93000
},
{
"epoch": 1.66,
"learning_rate": 0.0001,
"loss": 0.6958,
"step": 93500
},
{
"epoch": 1.67,
"learning_rate": 0.0001,
"loss": 0.6947,
"step": 94000
},
{
"epoch": 1.68,
"learning_rate": 0.0001,
"loss": 0.6973,
"step": 94500
},
{
"epoch": 1.69,
"learning_rate": 0.0001,
"loss": 0.6913,
"step": 95000
},
{
"epoch": 1.7,
"learning_rate": 0.0001,
"loss": 0.6765,
"step": 95500
},
{
"epoch": 1.71,
"learning_rate": 0.0001,
"loss": 0.6893,
"step": 96000
},
{
"epoch": 1.72,
"learning_rate": 0.0001,
"loss": 0.6828,
"step": 96500
},
{
"epoch": 1.72,
"learning_rate": 0.0001,
"loss": 0.6892,
"step": 97000
},
{
"epoch": 1.73,
"learning_rate": 0.0001,
"loss": 0.6849,
"step": 97500
},
{
"epoch": 1.74,
"learning_rate": 0.0001,
"loss": 0.6864,
"step": 98000
},
{
"epoch": 1.75,
"learning_rate": 0.0001,
"loss": 0.6776,
"step": 98500
},
{
"epoch": 1.76,
"learning_rate": 0.0001,
"loss": 0.683,
"step": 99000
},
{
"epoch": 1.77,
"learning_rate": 0.0001,
"loss": 0.6771,
"step": 99500
},
{
"epoch": 1.78,
"learning_rate": 0.0001,
"loss": 0.6794,
"step": 100000
},
{
"epoch": 1.79,
"learning_rate": 0.0001,
"loss": 0.684,
"step": 100500
},
{
"epoch": 1.8,
"learning_rate": 0.0001,
"loss": 0.6934,
"step": 101000
},
{
"epoch": 1.8,
"learning_rate": 0.0001,
"loss": 0.6747,
"step": 101500
},
{
"epoch": 1.81,
"learning_rate": 0.0001,
"loss": 0.6915,
"step": 102000
},
{
"epoch": 1.82,
"learning_rate": 0.0001,
"loss": 0.6744,
"step": 102500
},
{
"epoch": 1.83,
"learning_rate": 0.0001,
"loss": 0.6745,
"step": 103000
},
{
"epoch": 1.84,
"learning_rate": 0.0001,
"loss": 0.6762,
"step": 103500
},
{
"epoch": 1.85,
"learning_rate": 0.0001,
"loss": 0.6806,
"step": 104000
},
{
"epoch": 1.86,
"learning_rate": 0.0001,
"loss": 0.6786,
"step": 104500
},
{
"epoch": 1.87,
"learning_rate": 0.0001,
"loss": 0.6662,
"step": 105000
},
{
"epoch": 1.88,
"learning_rate": 0.0001,
"loss": 0.6724,
"step": 105500
},
{
"epoch": 1.88,
"learning_rate": 0.0001,
"loss": 0.6762,
"step": 106000
},
{
"epoch": 1.89,
"learning_rate": 0.0001,
"loss": 0.6781,
"step": 106500
},
{
"epoch": 1.9,
"learning_rate": 0.0001,
"loss": 0.6681,
"step": 107000
},
{
"epoch": 1.91,
"learning_rate": 0.0001,
"loss": 0.6669,
"step": 107500
},
{
"epoch": 1.92,
"learning_rate": 0.0001,
"loss": 0.6687,
"step": 108000
},
{
"epoch": 1.93,
"learning_rate": 0.0001,
"loss": 0.6624,
"step": 108500
},
{
"epoch": 1.94,
"learning_rate": 0.0001,
"loss": 0.6671,
"step": 109000
},
{
"epoch": 1.95,
"learning_rate": 0.0001,
"loss": 0.673,
"step": 109500
},
{
"epoch": 1.96,
"learning_rate": 0.0001,
"loss": 0.6748,
"step": 110000
},
{
"epoch": 1.96,
"learning_rate": 0.0001,
"loss": 0.6623,
"step": 110500
},
{
"epoch": 1.97,
"learning_rate": 0.0001,
"loss": 0.6635,
"step": 111000
},
{
"epoch": 1.98,
"learning_rate": 0.0001,
"loss": 0.6768,
"step": 111500
},
{
"epoch": 1.99,
"learning_rate": 0.0001,
"loss": 0.6586,
"step": 112000
},
{
"epoch": 2.0,
"learning_rate": 0.0001,
"loss": 0.668,
"step": 112500
},
{
"epoch": 2.0,
"eval_bleu": 27.3638,
"eval_gen_len": 11.1549,
"eval_loss": 0.9484143257141113,
"eval_runtime": 711.1842,
"eval_samples_per_second": 19.685,
"step": 112500
},
{
"epoch": 2.01,
"learning_rate": 0.0001,
"loss": 0.6175,
"step": 113000
},
{
"epoch": 2.02,
"learning_rate": 0.0001,
"loss": 0.625,
"step": 113500
},
{
"epoch": 2.03,
"learning_rate": 0.0001,
"loss": 0.6285,
"step": 114000
},
{
"epoch": 2.04,
"learning_rate": 0.0001,
"loss": 0.6191,
"step": 114500
},
{
"epoch": 2.04,
"learning_rate": 0.0001,
"loss": 0.6246,
"step": 115000
},
{
"epoch": 2.05,
"learning_rate": 0.0001,
"loss": 0.6186,
"step": 115500
},
{
"epoch": 2.06,
"learning_rate": 0.0001,
"loss": 0.6224,
"step": 116000
},
{
"epoch": 2.07,
"learning_rate": 0.0001,
"loss": 0.6282,
"step": 116500
},
{
"epoch": 2.08,
"learning_rate": 0.0001,
"loss": 0.6186,
"step": 117000
},
{
"epoch": 2.09,
"learning_rate": 0.0001,
"loss": 0.6159,
"step": 117500
},
{
"epoch": 2.1,
"learning_rate": 0.0001,
"loss": 0.6237,
"step": 118000
},
{
"epoch": 2.11,
"learning_rate": 0.0001,
"loss": 0.6194,
"step": 118500
},
{
"epoch": 2.12,
"learning_rate": 0.0001,
"loss": 0.62,
"step": 119000
},
{
"epoch": 2.12,
"learning_rate": 0.0001,
"loss": 0.6232,
"step": 119500
},
{
"epoch": 2.13,
"learning_rate": 0.0001,
"loss": 0.6258,
"step": 120000
},
{
"epoch": 2.14,
"learning_rate": 0.0001,
"loss": 0.6224,
"step": 120500
},
{
"epoch": 2.15,
"learning_rate": 0.0001,
"loss": 0.619,
"step": 121000
},
{
"epoch": 2.16,
"learning_rate": 0.0001,
"loss": 0.6141,
"step": 121500
},
{
"epoch": 2.17,
"learning_rate": 0.0001,
"loss": 0.6127,
"step": 122000
},
{
"epoch": 2.18,
"learning_rate": 0.0001,
"loss": 0.6202,
"step": 122500
},
{
"epoch": 2.19,
"learning_rate": 0.0001,
"loss": 0.6151,
"step": 123000
},
{
"epoch": 2.2,
"learning_rate": 0.0001,
"loss": 0.6248,
"step": 123500
},
{
"epoch": 2.2,
"learning_rate": 0.0001,
"loss": 0.6247,
"step": 124000
},
{
"epoch": 2.21,
"learning_rate": 0.0001,
"loss": 0.6167,
"step": 124500
},
{
"epoch": 2.22,
"learning_rate": 0.0001,
"loss": 0.6201,
"step": 125000
},
{
"epoch": 2.23,
"learning_rate": 0.0001,
"loss": 0.6142,
"step": 125500
},
{
"epoch": 2.24,
"learning_rate": 0.0001,
"loss": 0.6157,
"step": 126000
},
{
"epoch": 2.25,
"learning_rate": 0.0001,
"loss": 0.6144,
"step": 126500
},
{
"epoch": 2.26,
"learning_rate": 0.0001,
"loss": 0.6178,
"step": 127000
},
{
"epoch": 2.27,
"learning_rate": 0.0001,
"loss": 0.6125,
"step": 127500
},
{
"epoch": 2.28,
"learning_rate": 0.0001,
"loss": 0.6166,
"step": 128000
},
{
"epoch": 2.28,
"learning_rate": 0.0001,
"loss": 0.6166,
"step": 128500
},
{
"epoch": 2.29,
"learning_rate": 0.0001,
"loss": 0.6111,
"step": 129000
},
{
"epoch": 2.3,
"learning_rate": 0.0001,
"loss": 0.6155,
"step": 129500
},
{
"epoch": 2.31,
"learning_rate": 0.0001,
"loss": 0.6103,
"step": 130000
},
{
"epoch": 2.32,
"learning_rate": 0.0001,
"loss": 0.6118,
"step": 130500
},
{
"epoch": 2.33,
"learning_rate": 0.0001,
"loss": 0.6132,
"step": 131000
},
{
"epoch": 2.34,
"learning_rate": 0.0001,
"loss": 0.6139,
"step": 131500
},
{
"epoch": 2.35,
"learning_rate": 0.0001,
"loss": 0.6064,
"step": 132000
},
{
"epoch": 2.36,
"learning_rate": 0.0001,
"loss": 0.6088,
"step": 132500
},
{
"epoch": 2.36,
"learning_rate": 0.0001,
"loss": 0.6163,
"step": 133000
},
{
"epoch": 2.37,
"learning_rate": 0.0001,
"loss": 0.6147,
"step": 133500
},
{
"epoch": 2.38,
"learning_rate": 0.0001,
"loss": 0.6138,
"step": 134000
},
{
"epoch": 2.39,
"learning_rate": 0.0001,
"loss": 0.6126,
"step": 134500
},
{
"epoch": 2.4,
"learning_rate": 0.0001,
"loss": 0.607,
"step": 135000
},
{
"epoch": 2.41,
"learning_rate": 0.0001,
"loss": 0.5889,
"step": 135500
},
{
"epoch": 2.42,
"learning_rate": 0.0001,
"loss": 0.585,
"step": 136000
},
{
"epoch": 2.43,
"learning_rate": 0.0001,
"loss": 0.5882,
"step": 136500
},
{
"epoch": 2.44,
"learning_rate": 0.0001,
"loss": 0.5779,
"step": 137000
},
{
"epoch": 2.44,
"learning_rate": 0.0001,
"loss": 0.5921,
"step": 137500
},
{
"epoch": 2.45,
"learning_rate": 0.0001,
"loss": 0.5974,
"step": 138000
},
{
"epoch": 2.46,
"learning_rate": 0.0001,
"loss": 0.5931,
"step": 138500
},
{
"epoch": 2.47,
"learning_rate": 0.0001,
"loss": 0.5868,
"step": 139000
},
{
"epoch": 2.48,
"learning_rate": 0.0001,
"loss": 0.5879,
"step": 139500
},
{
"epoch": 2.49,
"learning_rate": 0.0001,
"loss": 0.5943,
"step": 140000
},
{
"epoch": 2.5,
"learning_rate": 0.0001,
"loss": 0.5946,
"step": 140500
},
{
"epoch": 2.51,
"learning_rate": 0.0001,
"loss": 0.5945,
"step": 141000
},
{
"epoch": 2.52,
"learning_rate": 0.0001,
"loss": 0.5881,
"step": 141500
},
{
"epoch": 2.52,
"learning_rate": 0.0001,
"loss": 0.5846,
"step": 142000
},
{
"epoch": 2.53,
"learning_rate": 0.0001,
"loss": 0.5827,
"step": 142500
},
{
"epoch": 2.54,
"learning_rate": 0.0001,
"loss": 0.5894,
"step": 143000
},
{
"epoch": 2.55,
"learning_rate": 0.0001,
"loss": 0.5852,
"step": 143500
},
{
"epoch": 2.56,
"learning_rate": 0.0001,
"loss": 0.588,
"step": 144000
},
{
"epoch": 2.57,
"learning_rate": 0.0001,
"loss": 0.5854,
"step": 144500
},
{
"epoch": 2.58,
"learning_rate": 0.0001,
"loss": 0.5919,
"step": 145000
},
{
"epoch": 2.59,
"learning_rate": 0.0001,
"loss": 0.589,
"step": 145500
},
{
"epoch": 2.6,
"learning_rate": 0.0001,
"loss": 0.5859,
"step": 146000
},
{
"epoch": 2.6,
"learning_rate": 0.0001,
"loss": 0.5856,
"step": 146500
},
{
"epoch": 2.61,
"learning_rate": 0.0001,
"loss": 0.5859,
"step": 147000
},
{
"epoch": 2.62,
"learning_rate": 0.0001,
"loss": 0.5966,
"step": 147500
},
{
"epoch": 2.63,
"learning_rate": 0.0001,
"loss": 0.594,
"step": 148000
},
{
"epoch": 2.64,
"learning_rate": 0.0001,
"loss": 0.5934,
"step": 148500
},
{
"epoch": 2.65,
"learning_rate": 0.0001,
"loss": 0.5864,
"step": 149000
},
{
"epoch": 2.66,
"learning_rate": 0.0001,
"loss": 0.5834,
"step": 149500
},
{
"epoch": 2.67,
"learning_rate": 0.0001,
"loss": 0.59,
"step": 150000
},
{
"epoch": 2.68,
"learning_rate": 0.0001,
"loss": 0.5855,
"step": 150500
},
{
"epoch": 2.68,
"learning_rate": 0.0001,
"loss": 0.5832,
"step": 151000
},
{
"epoch": 2.69,
"learning_rate": 0.0001,
"loss": 0.587,
"step": 151500
},
{
"epoch": 2.7,
"learning_rate": 0.0001,
"loss": 0.5853,
"step": 152000
},
{
"epoch": 2.71,
"learning_rate": 0.0001,
"loss": 0.5857,
"step": 152500
},
{
"epoch": 2.72,
"learning_rate": 0.0001,
"loss": 0.5816,
"step": 153000
},
{
"epoch": 2.73,
"learning_rate": 0.0001,
"loss": 0.5796,
"step": 153500
},
{
"epoch": 2.74,
"learning_rate": 0.0001,
"loss": 0.5892,
"step": 154000
},
{
"epoch": 2.75,
"learning_rate": 0.0001,
"loss": 0.5944,
"step": 154500
},
{
"epoch": 2.76,
"learning_rate": 0.0001,
"loss": 0.5821,
"step": 155000
},
{
"epoch": 2.76,
"learning_rate": 0.0001,
"loss": 0.5853,
"step": 155500
},
{
"epoch": 2.77,
"learning_rate": 0.0001,
"loss": 0.5894,
"step": 156000
},
{
"epoch": 2.78,
"learning_rate": 0.0001,
"loss": 0.5848,
"step": 156500
},
{
"epoch": 2.79,
"learning_rate": 0.0001,
"loss": 0.588,
"step": 157000
},
{
"epoch": 2.8,
"learning_rate": 0.0001,
"loss": 0.5799,
"step": 157500
},
{
"epoch": 2.81,
"learning_rate": 0.0001,
"loss": 0.5777,
"step": 158000
},
{
"epoch": 2.82,
"learning_rate": 0.0001,
"loss": 0.5848,
"step": 158500
},
{
"epoch": 2.83,
"learning_rate": 0.0001,
"loss": 0.5735,
"step": 159000
},
{
"epoch": 2.84,
"learning_rate": 0.0001,
"loss": 0.5887,
"step": 159500
},
{
"epoch": 2.84,
"learning_rate": 0.0001,
"loss": 0.5775,
"step": 160000
},
{
"epoch": 2.85,
"learning_rate": 0.0001,
"loss": 0.5897,
"step": 160500
},
{
"epoch": 2.86,
"learning_rate": 0.0001,
"loss": 0.5844,
"step": 161000
},
{
"epoch": 2.87,
"learning_rate": 0.0001,
"loss": 0.5793,
"step": 161500
},
{
"epoch": 2.88,
"learning_rate": 0.0001,
"loss": 0.5863,
"step": 162000
},
{
"epoch": 2.89,
"learning_rate": 0.0001,
"loss": 0.5794,
"step": 162500
},
{
"epoch": 2.9,
"learning_rate": 0.0001,
"loss": 0.5794,
"step": 163000
},
{
"epoch": 2.91,
"learning_rate": 0.0001,
"loss": 0.5743,
"step": 163500
},
{
"epoch": 2.92,
"learning_rate": 0.0001,
"loss": 0.572,
"step": 164000
},
{
"epoch": 2.92,
"learning_rate": 0.0001,
"loss": 0.5855,
"step": 164500
},
{
"epoch": 2.93,
"learning_rate": 0.0001,
"loss": 0.5769,
"step": 165000
},
{
"epoch": 2.94,
"learning_rate": 0.0001,
"loss": 0.5734,
"step": 165500
},
{
"epoch": 2.95,
"learning_rate": 0.0001,
"loss": 0.5767,
"step": 166000
},
{
"epoch": 2.96,
"learning_rate": 0.0001,
"loss": 0.5697,
"step": 166500
},
{
"epoch": 2.97,
"learning_rate": 0.0001,
"loss": 0.5791,
"step": 167000
},
{
"epoch": 2.98,
"learning_rate": 0.0001,
"loss": 0.5728,
"step": 167500
},
{
"epoch": 2.99,
"learning_rate": 0.0001,
"loss": 0.5723,
"step": 168000
},
{
"epoch": 3.0,
"learning_rate": 0.0001,
"loss": 0.5786,
"step": 168500
},
{
"epoch": 3.0,
"eval_bleu": 27.3165,
"eval_gen_len": 11.202,
"eval_loss": 0.9859131574630737,
"eval_runtime": 722.7121,
"eval_samples_per_second": 19.371,
"step": 168750
},
{
"epoch": 3.0,
"learning_rate": 0.0001,
"loss": 0.5554,
"step": 169000
},
{
"epoch": 3.01,
"learning_rate": 0.0001,
"loss": 0.5383,
"step": 169500
},
{
"epoch": 3.02,
"learning_rate": 0.0001,
"loss": 0.5378,
"step": 170000
},
{
"epoch": 3.03,
"learning_rate": 0.0001,
"loss": 0.5397,
"step": 170500
},
{
"epoch": 3.04,
"learning_rate": 0.0001,
"loss": 0.5397,
"step": 171000
},
{
"epoch": 3.05,
"learning_rate": 0.0001,
"loss": 0.5403,
"step": 171500
},
{
"epoch": 3.06,
"learning_rate": 0.0001,
"loss": 0.5356,
"step": 172000
},
{
"epoch": 3.07,
"learning_rate": 0.0001,
"loss": 0.5447,
"step": 172500
},
{
"epoch": 3.08,
"learning_rate": 0.0001,
"loss": 0.5327,
"step": 173000
},
{
"epoch": 3.08,
"learning_rate": 0.0001,
"loss": 0.5462,
"step": 173500
},
{
"epoch": 3.09,
"learning_rate": 0.0001,
"loss": 0.5395,
"step": 174000
},
{
"epoch": 3.1,
"learning_rate": 0.0001,
"loss": 0.5421,
"step": 174500
},
{
"epoch": 3.11,
"learning_rate": 0.0001,
"loss": 0.5402,
"step": 175000
},
{
"epoch": 3.12,
"learning_rate": 0.0001,
"loss": 0.5431,
"step": 175500
},
{
"epoch": 3.13,
"learning_rate": 0.0001,
"loss": 0.5407,
"step": 176000
},
{
"epoch": 3.14,
"learning_rate": 0.0001,
"loss": 0.5483,
"step": 176500
},
{
"epoch": 3.15,
"learning_rate": 0.0001,
"loss": 0.5414,
"step": 177000
},
{
"epoch": 3.16,
"learning_rate": 0.0001,
"loss": 0.5402,
"step": 177500
},
{
"epoch": 3.16,
"learning_rate": 0.0001,
"loss": 0.5377,
"step": 178000
},
{
"epoch": 3.17,
"learning_rate": 0.0001,
"loss": 0.5392,
"step": 178500
},
{
"epoch": 3.18,
"learning_rate": 0.0001,
"loss": 0.5448,
"step": 179000
},
{
"epoch": 3.19,
"learning_rate": 0.0001,
"loss": 0.5463,
"step": 179500
},
{
"epoch": 3.2,
"learning_rate": 0.0001,
"loss": 0.5425,
"step": 180000
},
{
"epoch": 3.21,
"learning_rate": 0.0001,
"loss": 0.5103,
"step": 180500
},
{
"epoch": 3.22,
"learning_rate": 0.0001,
"loss": 0.5081,
"step": 181000
},
{
"epoch": 3.23,
"learning_rate": 0.0001,
"loss": 0.5068,
"step": 181500
},
{
"epoch": 3.24,
"learning_rate": 0.0001,
"loss": 0.5118,
"step": 182000
},
{
"epoch": 3.24,
"learning_rate": 0.0001,
"loss": 0.5117,
"step": 182500
},
{
"epoch": 3.25,
"learning_rate": 0.0001,
"loss": 0.5121,
"step": 183000
},
{
"epoch": 3.26,
"learning_rate": 0.0001,
"loss": 0.5102,
"step": 183500
},
{
"epoch": 3.27,
"learning_rate": 0.0001,
"loss": 0.5098,
"step": 184000
},
{
"epoch": 3.28,
"learning_rate": 0.0001,
"loss": 0.5168,
"step": 184500
},
{
"epoch": 3.29,
"learning_rate": 0.0001,
"loss": 0.5135,
"step": 185000
},
{
"epoch": 3.3,
"learning_rate": 0.0001,
"loss": 0.5062,
"step": 185500
},
{
"epoch": 3.31,
"learning_rate": 0.0001,
"loss": 0.5156,
"step": 186000
},
{
"epoch": 3.32,
"learning_rate": 0.0001,
"loss": 0.5088,
"step": 186500
},
{
"epoch": 3.32,
"learning_rate": 0.0001,
"loss": 0.509,
"step": 187000
},
{
"epoch": 3.33,
"learning_rate": 0.0001,
"loss": 0.5106,
"step": 187500
},
{
"epoch": 3.34,
"learning_rate": 0.0001,
"loss": 0.511,
"step": 188000
},
{
"epoch": 3.35,
"learning_rate": 0.0001,
"loss": 0.5111,
"step": 188500
},
{
"epoch": 3.36,
"learning_rate": 0.0001,
"loss": 0.5125,
"step": 189000
},
{
"epoch": 3.37,
"learning_rate": 0.0001,
"loss": 0.5173,
"step": 189500
},
{
"epoch": 3.38,
"learning_rate": 0.0001,
"loss": 0.5123,
"step": 190000
},
{
"epoch": 3.39,
"learning_rate": 0.0001,
"loss": 0.5168,
"step": 190500
},
{
"epoch": 3.4,
"learning_rate": 0.0001,
"loss": 0.5084,
"step": 191000
},
{
"epoch": 3.4,
"learning_rate": 0.0001,
"loss": 0.5273,
"step": 191500
},
{
"epoch": 3.41,
"learning_rate": 0.0001,
"loss": 0.5402,
"step": 192000
},
{
"epoch": 3.42,
"learning_rate": 0.0001,
"loss": 0.5335,
"step": 192500
},
{
"epoch": 3.43,
"learning_rate": 0.0001,
"loss": 0.5402,
"step": 193000
},
{
"epoch": 3.44,
"learning_rate": 0.0001,
"loss": 0.5325,
"step": 193500
},
{
"epoch": 3.45,
"learning_rate": 0.0001,
"loss": 0.538,
"step": 194000
},
{
"epoch": 3.46,
"learning_rate": 0.0001,
"loss": 0.5389,
"step": 194500
},
{
"epoch": 3.47,
"learning_rate": 0.0001,
"loss": 0.5328,
"step": 195000
},
{
"epoch": 3.48,
"learning_rate": 0.0001,
"loss": 0.533,
"step": 195500
},
{
"epoch": 3.48,
"learning_rate": 0.0001,
"loss": 0.5366,
"step": 196000
},
{
"epoch": 3.49,
"learning_rate": 0.0001,
"loss": 0.5372,
"step": 196500
},
{
"epoch": 3.5,
"learning_rate": 0.0001,
"loss": 0.546,
"step": 197000
},
{
"epoch": 3.51,
"learning_rate": 0.0001,
"loss": 0.5399,
"step": 197500
},
{
"epoch": 3.52,
"learning_rate": 0.0001,
"loss": 0.5306,
"step": 198000
},
{
"epoch": 3.53,
"learning_rate": 0.0001,
"loss": 0.5359,
"step": 198500
},
{
"epoch": 3.54,
"learning_rate": 0.0001,
"loss": 0.5361,
"step": 199000
},
{
"epoch": 3.55,
"learning_rate": 0.0001,
"loss": 0.5358,
"step": 199500
},
{
"epoch": 3.56,
"learning_rate": 0.0001,
"loss": 0.5429,
"step": 200000
},
{
"epoch": 3.56,
"learning_rate": 0.0001,
"loss": 0.5325,
"step": 200500
},
{
"epoch": 3.57,
"learning_rate": 0.0001,
"loss": 0.5409,
"step": 201000
},
{
"epoch": 3.58,
"learning_rate": 0.0001,
"loss": 0.5361,
"step": 201500
},
{
"epoch": 3.59,
"learning_rate": 0.0001,
"loss": 0.537,
"step": 202000
},
{
"epoch": 3.6,
"learning_rate": 0.0001,
"loss": 0.5319,
"step": 202500
},
{
"epoch": 3.61,
"learning_rate": 0.0001,
"loss": 0.5353,
"step": 203000
},
{
"epoch": 3.62,
"learning_rate": 0.0001,
"loss": 0.537,
"step": 203500
},
{
"epoch": 3.63,
"learning_rate": 0.0001,
"loss": 0.529,
"step": 204000
},
{
"epoch": 3.64,
"learning_rate": 0.0001,
"loss": 0.536,
"step": 204500
},
{
"epoch": 3.64,
"learning_rate": 0.0001,
"loss": 0.5322,
"step": 205000
},
{
"epoch": 3.65,
"learning_rate": 0.0001,
"loss": 0.5305,
"step": 205500
},
{
"epoch": 3.66,
"learning_rate": 0.0001,
"loss": 0.537,
"step": 206000
},
{
"epoch": 3.67,
"learning_rate": 0.0001,
"loss": 0.5324,
"step": 206500
},
{
"epoch": 3.68,
"learning_rate": 0.0001,
"loss": 0.5388,
"step": 207000
},
{
"epoch": 3.69,
"learning_rate": 0.0001,
"loss": 0.5323,
"step": 207500
},
{
"epoch": 3.7,
"learning_rate": 0.0001,
"loss": 0.5367,
"step": 208000
},
{
"epoch": 3.71,
"learning_rate": 0.0001,
"loss": 0.5286,
"step": 208500
},
{
"epoch": 3.72,
"learning_rate": 0.0001,
"loss": 0.5373,
"step": 209000
},
{
"epoch": 3.72,
"learning_rate": 0.0001,
"loss": 0.532,
"step": 209500
},
{
"epoch": 3.73,
"learning_rate": 0.0001,
"loss": 0.5238,
"step": 210000
},
{
"epoch": 3.74,
"learning_rate": 0.0001,
"loss": 0.5297,
"step": 210500
},
{
"epoch": 3.75,
"learning_rate": 0.0001,
"loss": 0.5348,
"step": 211000
},
{
"epoch": 3.76,
"learning_rate": 0.0001,
"loss": 0.5324,
"step": 211500
},
{
"epoch": 3.77,
"learning_rate": 0.0001,
"loss": 0.5247,
"step": 212000
},
{
"epoch": 3.78,
"learning_rate": 0.0001,
"loss": 0.5294,
"step": 212500
},
{
"epoch": 3.79,
"learning_rate": 0.0001,
"loss": 0.5322,
"step": 213000
},
{
"epoch": 3.8,
"learning_rate": 0.0001,
"loss": 0.5329,
"step": 213500
},
{
"epoch": 3.8,
"learning_rate": 0.0001,
"loss": 0.5348,
"step": 214000
},
{
"epoch": 3.81,
"learning_rate": 0.0001,
"loss": 0.5256,
"step": 214500
},
{
"epoch": 3.82,
"learning_rate": 0.0001,
"loss": 0.5353,
"step": 215000
},
{
"epoch": 3.83,
"learning_rate": 0.0001,
"loss": 0.5264,
"step": 215500
},
{
"epoch": 3.84,
"learning_rate": 0.0001,
"loss": 0.5245,
"step": 216000
},
{
"epoch": 3.85,
"learning_rate": 0.0001,
"loss": 0.5207,
"step": 216500
},
{
"epoch": 3.86,
"learning_rate": 0.0001,
"loss": 0.5289,
"step": 217000
},
{
"epoch": 3.87,
"learning_rate": 0.0001,
"loss": 0.5267,
"step": 217500
},
{
"epoch": 3.88,
"learning_rate": 0.0001,
"loss": 0.5267,
"step": 218000
},
{
"epoch": 3.88,
"learning_rate": 0.0001,
"loss": 0.5328,
"step": 218500
},
{
"epoch": 3.89,
"learning_rate": 0.0001,
"loss": 0.5271,
"step": 219000
},
{
"epoch": 3.9,
"learning_rate": 0.0001,
"loss": 0.5204,
"step": 219500
},
{
"epoch": 3.91,
"learning_rate": 0.0001,
"loss": 0.533,
"step": 220000
},
{
"epoch": 3.92,
"learning_rate": 0.0001,
"loss": 0.5267,
"step": 220500
},
{
"epoch": 3.93,
"learning_rate": 0.0001,
"loss": 0.5261,
"step": 221000
},
{
"epoch": 3.94,
"learning_rate": 0.0001,
"loss": 0.5276,
"step": 221500
},
{
"epoch": 3.95,
"learning_rate": 0.0001,
"loss": 0.5281,
"step": 222000
},
{
"epoch": 3.96,
"learning_rate": 0.0001,
"loss": 0.5193,
"step": 222500
},
{
"epoch": 3.96,
"learning_rate": 0.0001,
"loss": 0.5229,
"step": 223000
},
{
"epoch": 3.97,
"learning_rate": 0.0001,
"loss": 0.5189,
"step": 223500
},
{
"epoch": 3.98,
"learning_rate": 0.0001,
"loss": 0.5206,
"step": 224000
},
{
"epoch": 3.99,
"learning_rate": 0.0001,
"loss": 0.5229,
"step": 224500
},
{
"epoch": 4.0,
"learning_rate": 0.0001,
"loss": 0.5258,
"step": 225000
},
{
"epoch": 4.0,
"eval_bleu": 27.2229,
"eval_gen_len": 11.222,
"eval_loss": 1.0146929025650024,
"eval_runtime": 724.1714,
"eval_samples_per_second": 19.332,
"step": 225000
},
{
"epoch": 4.01,
"learning_rate": 0.0001,
"loss": 0.4757,
"step": 225500
},
{
"epoch": 4.02,
"learning_rate": 0.0001,
"loss": 0.4793,
"step": 226000
},
{
"epoch": 4.03,
"learning_rate": 0.0001,
"loss": 0.4781,
"step": 226500
},
{
"epoch": 4.04,
"learning_rate": 0.0001,
"loss": 0.48,
"step": 227000
},
{
"epoch": 4.04,
"learning_rate": 0.0001,
"loss": 0.4798,
"step": 227500
},
{
"epoch": 4.05,
"learning_rate": 0.0001,
"loss": 0.476,
"step": 228000
},
{
"epoch": 4.06,
"learning_rate": 0.0001,
"loss": 0.4802,
"step": 228500
},
{
"epoch": 4.07,
"learning_rate": 0.0001,
"loss": 0.4761,
"step": 229000
},
{
"epoch": 4.08,
"learning_rate": 0.0001,
"loss": 0.4783,
"step": 229500
},
{
"epoch": 4.09,
"learning_rate": 0.0001,
"loss": 0.4801,
"step": 230000
},
{
"epoch": 4.1,
"learning_rate": 0.0001,
"loss": 0.477,
"step": 230500
},
{
"epoch": 4.11,
"learning_rate": 0.0001,
"loss": 0.4756,
"step": 231000
},
{
"epoch": 4.12,
"learning_rate": 0.0001,
"loss": 0.4851,
"step": 231500
},
{
"epoch": 4.12,
"learning_rate": 0.0001,
"loss": 0.4884,
"step": 232000
},
{
"epoch": 4.13,
"learning_rate": 0.0001,
"loss": 0.485,
"step": 232500
},
{
"epoch": 4.14,
"learning_rate": 0.0001,
"loss": 0.4844,
"step": 233000
},
{
"epoch": 4.15,
"learning_rate": 0.0001,
"loss": 0.4819,
"step": 233500
},
{
"epoch": 4.16,
"learning_rate": 0.0001,
"loss": 0.4795,
"step": 234000
},
{
"epoch": 4.17,
"learning_rate": 0.0001,
"loss": 0.4779,
"step": 234500
},
{
"epoch": 4.18,
"learning_rate": 0.0001,
"loss": 0.4826,
"step": 235000
},
{
"epoch": 4.19,
"learning_rate": 0.0001,
"loss": 0.4774,
"step": 235500
},
{
"epoch": 4.2,
"learning_rate": 0.0001,
"loss": 0.4812,
"step": 236000
},
{
"epoch": 4.2,
"learning_rate": 0.0001,
"loss": 0.4781,
"step": 236500
},
{
"epoch": 4.21,
"learning_rate": 0.0001,
"loss": 0.4776,
"step": 237000
},
{
"epoch": 4.22,
"learning_rate": 0.0001,
"loss": 0.4874,
"step": 237500
},
{
"epoch": 4.23,
"learning_rate": 0.0001,
"loss": 0.4743,
"step": 238000
},
{
"epoch": 4.24,
"learning_rate": 0.0001,
"loss": 0.4852,
"step": 238500
},
{
"epoch": 4.25,
"learning_rate": 0.0001,
"loss": 0.4791,
"step": 239000
},
{
"epoch": 4.26,
"learning_rate": 0.0001,
"loss": 0.4866,
"step": 239500
},
{
"epoch": 4.27,
"learning_rate": 0.0001,
"loss": 0.487,
"step": 240000
},
{
"epoch": 4.28,
"learning_rate": 0.0001,
"loss": 0.4712,
"step": 240500
},
{
"epoch": 4.28,
"learning_rate": 0.0001,
"loss": 0.4813,
"step": 241000
},
{
"epoch": 4.29,
"learning_rate": 0.0001,
"loss": 0.487,
"step": 241500
},
{
"epoch": 4.3,
"learning_rate": 0.0001,
"loss": 0.4775,
"step": 242000
},
{
"epoch": 4.31,
"learning_rate": 0.0001,
"loss": 0.4807,
"step": 242500
},
{
"epoch": 4.32,
"learning_rate": 0.0001,
"loss": 0.4863,
"step": 243000
},
{
"epoch": 4.33,
"learning_rate": 0.0001,
"loss": 0.4738,
"step": 243500
},
{
"epoch": 4.34,
"learning_rate": 0.0001,
"loss": 0.4798,
"step": 244000
},
{
"epoch": 4.35,
"learning_rate": 0.0001,
"loss": 0.4799,
"step": 244500
},
{
"epoch": 4.36,
"learning_rate": 0.0001,
"loss": 0.4845,
"step": 245000
},
{
"epoch": 4.36,
"learning_rate": 0.0001,
"loss": 0.4669,
"step": 245500
},
{
"epoch": 4.37,
"learning_rate": 0.0001,
"loss": 0.4614,
"step": 246000
},
{
"epoch": 4.38,
"learning_rate": 0.0001,
"loss": 0.4641,
"step": 246500
},
{
"epoch": 4.39,
"learning_rate": 0.0001,
"loss": 0.4666,
"step": 247000
},
{
"epoch": 4.4,
"learning_rate": 0.0001,
"loss": 0.4572,
"step": 247500
},
{
"epoch": 4.41,
"learning_rate": 0.0001,
"loss": 0.4639,
"step": 248000
},
{
"epoch": 4.42,
"learning_rate": 0.0001,
"loss": 0.4663,
"step": 248500
},
{
"epoch": 4.43,
"learning_rate": 0.0001,
"loss": 0.4677,
"step": 249000
},
{
"epoch": 4.44,
"learning_rate": 0.0001,
"loss": 0.4696,
"step": 249500
},
{
"epoch": 4.44,
"learning_rate": 0.0001,
"loss": 0.4725,
"step": 250000
},
{
"epoch": 4.45,
"learning_rate": 0.0001,
"loss": 0.4664,
"step": 250500
},
{
"epoch": 4.46,
"learning_rate": 0.0001,
"loss": 0.4673,
"step": 251000
},
{
"epoch": 4.47,
"learning_rate": 0.0001,
"loss": 0.4651,
"step": 251500
},
{
"epoch": 4.48,
"learning_rate": 0.0001,
"loss": 0.4708,
"step": 252000
},
{
"epoch": 4.49,
"learning_rate": 0.0001,
"loss": 0.4736,
"step": 252500
},
{
"epoch": 4.5,
"learning_rate": 0.0001,
"loss": 0.4668,
"step": 253000
},
{
"epoch": 4.51,
"learning_rate": 0.0001,
"loss": 0.4744,
"step": 253500
},
{
"epoch": 4.52,
"learning_rate": 0.0001,
"loss": 0.4671,
"step": 254000
},
{
"epoch": 4.52,
"learning_rate": 0.0001,
"loss": 0.472,
"step": 254500
},
{
"epoch": 4.53,
"learning_rate": 0.0001,
"loss": 0.4655,
"step": 255000
},
{
"epoch": 4.54,
"learning_rate": 0.0001,
"loss": 0.4667,
"step": 255500
},
{
"epoch": 4.55,
"learning_rate": 0.0001,
"loss": 0.4682,
"step": 256000
},
{
"epoch": 4.56,
"learning_rate": 0.0001,
"loss": 0.4755,
"step": 256500
},
{
"epoch": 4.57,
"learning_rate": 0.0001,
"loss": 0.4672,
"step": 257000
},
{
"epoch": 4.58,
"learning_rate": 0.0001,
"loss": 0.4639,
"step": 257500
},
{
"epoch": 4.59,
"learning_rate": 0.0001,
"loss": 0.4671,
"step": 258000
},
{
"epoch": 4.6,
"learning_rate": 0.0001,
"loss": 0.4688,
"step": 258500
},
{
"epoch": 4.6,
"learning_rate": 0.0001,
"loss": 0.4782,
"step": 259000
},
{
"epoch": 4.61,
"learning_rate": 0.0001,
"loss": 0.4697,
"step": 259500
},
{
"epoch": 4.62,
"learning_rate": 0.0001,
"loss": 0.468,
"step": 260000
},
{
"epoch": 4.63,
"learning_rate": 0.0001,
"loss": 0.4668,
"step": 260500
},
{
"epoch": 4.64,
"learning_rate": 0.0001,
"loss": 0.4745,
"step": 261000
},
{
"epoch": 4.65,
"learning_rate": 0.0001,
"loss": 0.4685,
"step": 261500
},
{
"epoch": 4.66,
"learning_rate": 0.0001,
"loss": 0.4682,
"step": 262000
},
{
"epoch": 4.67,
"learning_rate": 0.0001,
"loss": 0.47,
"step": 262500
},
{
"epoch": 4.68,
"learning_rate": 0.0001,
"loss": 0.4715,
"step": 263000
},
{
"epoch": 4.68,
"learning_rate": 0.0001,
"loss": 0.4677,
"step": 263500
},
{
"epoch": 4.69,
"learning_rate": 0.0001,
"loss": 0.4727,
"step": 264000
},
{
"epoch": 4.7,
"learning_rate": 0.0001,
"loss": 0.4635,
"step": 264500
},
{
"epoch": 4.71,
"learning_rate": 0.0001,
"loss": 0.4661,
"step": 265000
},
{
"epoch": 4.72,
"learning_rate": 0.0001,
"loss": 0.4682,
"step": 265500
},
{
"epoch": 4.73,
"learning_rate": 0.0001,
"loss": 0.4662,
"step": 266000
},
{
"epoch": 4.74,
"learning_rate": 0.0001,
"loss": 0.4673,
"step": 266500
},
{
"epoch": 4.75,
"learning_rate": 0.0001,
"loss": 0.4713,
"step": 267000
},
{
"epoch": 4.76,
"learning_rate": 0.0001,
"loss": 0.4754,
"step": 267500
},
{
"epoch": 4.76,
"learning_rate": 0.0001,
"loss": 0.4654,
"step": 268000
},
{
"epoch": 4.77,
"learning_rate": 0.0001,
"loss": 0.4692,
"step": 268500
},
{
"epoch": 4.78,
"learning_rate": 0.0001,
"loss": 0.4684,
"step": 269000
},
{
"epoch": 4.79,
"learning_rate": 0.0001,
"loss": 0.4652,
"step": 269500
},
{
"epoch": 4.8,
"learning_rate": 0.0001,
"loss": 0.4631,
"step": 270000
},
{
"epoch": 4.81,
"learning_rate": 0.0001,
"loss": 0.4613,
"step": 270500
},
{
"epoch": 4.82,
"learning_rate": 0.0001,
"loss": 0.4607,
"step": 271000
},
{
"epoch": 4.83,
"learning_rate": 0.0001,
"loss": 0.4695,
"step": 271500
},
{
"epoch": 4.84,
"learning_rate": 0.0001,
"loss": 0.4634,
"step": 272000
},
{
"epoch": 4.84,
"learning_rate": 0.0001,
"loss": 0.4645,
"step": 272500
},
{
"epoch": 4.85,
"learning_rate": 0.0001,
"loss": 0.4713,
"step": 273000
},
{
"epoch": 4.86,
"learning_rate": 0.0001,
"loss": 0.4669,
"step": 273500
},
{
"epoch": 4.87,
"learning_rate": 0.0001,
"loss": 0.4665,
"step": 274000
},
{
"epoch": 4.88,
"learning_rate": 0.0001,
"loss": 0.4652,
"step": 274500
},
{
"epoch": 4.89,
"learning_rate": 0.0001,
"loss": 0.4766,
"step": 275000
},
{
"epoch": 4.9,
"learning_rate": 0.0001,
"loss": 0.4719,
"step": 275500
},
{
"epoch": 4.91,
"learning_rate": 0.0001,
"loss": 0.461,
"step": 276000
},
{
"epoch": 4.92,
"learning_rate": 0.0001,
"loss": 0.4636,
"step": 276500
},
{
"epoch": 4.92,
"learning_rate": 0.0001,
"loss": 0.4639,
"step": 277000
},
{
"epoch": 4.93,
"learning_rate": 0.0001,
"loss": 0.4631,
"step": 277500
},
{
"epoch": 4.94,
"learning_rate": 0.0001,
"loss": 0.4644,
"step": 278000
},
{
"epoch": 4.95,
"learning_rate": 0.0001,
"loss": 0.4663,
"step": 278500
},
{
"epoch": 4.96,
"learning_rate": 0.0001,
"loss": 0.4659,
"step": 279000
},
{
"epoch": 4.97,
"learning_rate": 0.0001,
"loss": 0.4642,
"step": 279500
},
{
"epoch": 4.98,
"learning_rate": 0.0001,
"loss": 0.4581,
"step": 280000
},
{
"epoch": 4.99,
"learning_rate": 0.0001,
"loss": 0.468,
"step": 280500
},
{
"epoch": 5.0,
"learning_rate": 0.0001,
"loss": 0.4658,
"step": 281000
},
{
"epoch": 5.0,
"eval_bleu": 27.1785,
"eval_gen_len": 11.2621,
"eval_loss": 1.0581830739974976,
"eval_runtime": 729.5414,
"eval_samples_per_second": 19.19,
"step": 281250
},
{
"epoch": 5.0,
"learning_rate": 0.0001,
"loss": 0.4516,
"step": 281500
},
{
"epoch": 5.01,
"learning_rate": 0.0001,
"loss": 0.4347,
"step": 282000
},
{
"epoch": 5.02,
"learning_rate": 0.0001,
"loss": 0.4341,
"step": 282500
},
{
"epoch": 5.03,
"learning_rate": 0.0001,
"loss": 0.4305,
"step": 283000
},
{
"epoch": 5.04,
"learning_rate": 0.0001,
"loss": 0.4337,
"step": 283500
},
{
"epoch": 5.05,
"learning_rate": 0.0001,
"loss": 0.4271,
"step": 284000
},
{
"epoch": 5.06,
"learning_rate": 0.0001,
"loss": 0.4306,
"step": 284500
},
{
"epoch": 5.07,
"learning_rate": 0.0001,
"loss": 0.4276,
"step": 285000
},
{
"epoch": 5.08,
"learning_rate": 0.0001,
"loss": 0.4048,
"step": 285500
},
{
"epoch": 5.08,
"learning_rate": 0.0001,
"loss": 0.4116,
"step": 286000
},
{
"epoch": 5.09,
"learning_rate": 0.0001,
"loss": 0.4118,
"step": 286500
},
{
"epoch": 5.1,
"learning_rate": 0.0001,
"loss": 0.4096,
"step": 287000
},
{
"epoch": 5.11,
"learning_rate": 0.0001,
"loss": 0.4154,
"step": 287500
},
{
"epoch": 5.12,
"learning_rate": 0.0001,
"loss": 0.4172,
"step": 288000
},
{
"epoch": 5.13,
"learning_rate": 0.0001,
"loss": 0.4194,
"step": 288500
},
{
"epoch": 5.14,
"learning_rate": 0.0001,
"loss": 0.4182,
"step": 289000
},
{
"epoch": 5.15,
"learning_rate": 0.0001,
"loss": 0.4184,
"step": 289500
},
{
"epoch": 5.16,
"learning_rate": 0.0001,
"loss": 0.4103,
"step": 290000
},
{
"epoch": 5.16,
"learning_rate": 0.0001,
"loss": 0.4165,
"step": 290500
},
{
"epoch": 5.17,
"learning_rate": 0.0001,
"loss": 0.4135,
"step": 291000
},
{
"epoch": 5.18,
"learning_rate": 0.0001,
"loss": 0.4132,
"step": 291500
},
{
"epoch": 5.19,
"learning_rate": 0.0001,
"loss": 0.4154,
"step": 292000
},
{
"epoch": 5.2,
"learning_rate": 0.0001,
"loss": 0.4154,
"step": 292500
},
{
"epoch": 5.21,
"learning_rate": 0.0001,
"loss": 0.4202,
"step": 293000
},
{
"epoch": 5.22,
"learning_rate": 0.0001,
"loss": 0.4207,
"step": 293500
},
{
"epoch": 5.23,
"learning_rate": 0.0001,
"loss": 0.417,
"step": 294000
},
{
"epoch": 5.24,
"learning_rate": 0.0001,
"loss": 0.414,
"step": 294500
},
{
"epoch": 5.24,
"learning_rate": 0.0001,
"loss": 0.4177,
"step": 295000
},
{
"epoch": 5.25,
"learning_rate": 0.0001,
"loss": 0.4179,
"step": 295500
},
{
"epoch": 5.26,
"learning_rate": 0.0001,
"loss": 0.4167,
"step": 296000
},
{
"epoch": 5.27,
"learning_rate": 0.0001,
"loss": 0.4156,
"step": 296500
},
{
"epoch": 5.28,
"learning_rate": 0.0001,
"loss": 0.4143,
"step": 297000
},
{
"epoch": 5.29,
"learning_rate": 0.0001,
"loss": 0.4155,
"step": 297500
},
{
"epoch": 5.3,
"learning_rate": 0.0001,
"loss": 0.4173,
"step": 298000
},
{
"epoch": 5.31,
"learning_rate": 0.0001,
"loss": 0.4176,
"step": 298500
},
{
"epoch": 5.32,
"learning_rate": 0.0001,
"loss": 0.423,
"step": 299000
},
{
"epoch": 5.32,
"learning_rate": 0.0001,
"loss": 0.419,
"step": 299500
},
{
"epoch": 5.33,
"learning_rate": 0.0001,
"loss": 0.4142,
"step": 300000
},
{
"epoch": 5.34,
"learning_rate": 0.0001,
"loss": 0.4169,
"step": 300500
},
{
"epoch": 5.35,
"learning_rate": 0.0001,
"loss": 0.4161,
"step": 301000
},
{
"epoch": 5.36,
"learning_rate": 0.0001,
"loss": 0.4304,
"step": 301500
},
{
"epoch": 5.37,
"learning_rate": 0.0001,
"loss": 0.4479,
"step": 302000
},
{
"epoch": 5.38,
"learning_rate": 0.0001,
"loss": 0.4499,
"step": 302500
},
{
"epoch": 5.39,
"learning_rate": 0.0001,
"loss": 0.4455,
"step": 303000
},
{
"epoch": 5.4,
"learning_rate": 0.0001,
"loss": 0.4471,
"step": 303500
},
{
"epoch": 5.4,
"learning_rate": 0.0001,
"loss": 0.4418,
"step": 304000
},
{
"epoch": 5.41,
"learning_rate": 0.0001,
"loss": 0.4493,
"step": 304500
},
{
"epoch": 5.42,
"learning_rate": 0.0001,
"loss": 0.4422,
"step": 305000
},
{
"epoch": 5.43,
"learning_rate": 0.0001,
"loss": 0.4418,
"step": 305500
},
{
"epoch": 5.44,
"learning_rate": 0.0001,
"loss": 0.4463,
"step": 306000
},
{
"epoch": 5.45,
"learning_rate": 0.0001,
"loss": 0.4413,
"step": 306500
},
{
"epoch": 5.46,
"learning_rate": 0.0001,
"loss": 0.4447,
"step": 307000
},
{
"epoch": 5.47,
"learning_rate": 0.0001,
"loss": 0.4442,
"step": 307500
},
{
"epoch": 5.48,
"learning_rate": 0.0001,
"loss": 0.4409,
"step": 308000
},
{
"epoch": 5.48,
"learning_rate": 0.0001,
"loss": 0.4383,
"step": 308500
},
{
"epoch": 5.49,
"learning_rate": 0.0001,
"loss": 0.4397,
"step": 309000
},
{
"epoch": 5.5,
"learning_rate": 0.0001,
"loss": 0.4416,
"step": 309500
},
{
"epoch": 5.51,
"learning_rate": 0.0001,
"loss": 0.4428,
"step": 310000
},
{
"epoch": 5.52,
"learning_rate": 0.0001,
"loss": 0.4431,
"step": 310500
},
{
"epoch": 5.53,
"learning_rate": 0.0001,
"loss": 0.4393,
"step": 311000
},
{
"epoch": 5.54,
"learning_rate": 0.0001,
"loss": 0.4341,
"step": 311500
},
{
"epoch": 5.55,
"learning_rate": 0.0001,
"loss": 0.444,
"step": 312000
},
{
"epoch": 5.56,
"learning_rate": 0.0001,
"loss": 0.4426,
"step": 312500
},
{
"epoch": 5.56,
"learning_rate": 0.0001,
"loss": 0.4454,
"step": 313000
},
{
"epoch": 5.57,
"learning_rate": 0.0001,
"loss": 0.4384,
"step": 313500
},
{
"epoch": 5.58,
"learning_rate": 0.0001,
"loss": 0.4316,
"step": 314000
},
{
"epoch": 5.59,
"learning_rate": 0.0001,
"loss": 0.4427,
"step": 314500
},
{
"epoch": 5.6,
"learning_rate": 0.0001,
"loss": 0.4453,
"step": 315000
},
{
"epoch": 5.61,
"learning_rate": 0.0001,
"loss": 0.436,
"step": 315500
},
{
"epoch": 5.62,
"learning_rate": 0.0001,
"loss": 0.4402,
"step": 316000
},
{
"epoch": 5.63,
"learning_rate": 0.0001,
"loss": 0.4376,
"step": 316500
},
{
"epoch": 5.64,
"learning_rate": 0.0001,
"loss": 0.4399,
"step": 317000
},
{
"epoch": 5.64,
"learning_rate": 0.0001,
"loss": 0.4443,
"step": 317500
},
{
"epoch": 5.65,
"learning_rate": 0.0001,
"loss": 0.4439,
"step": 318000
},
{
"epoch": 5.66,
"learning_rate": 0.0001,
"loss": 0.4402,
"step": 318500
},
{
"epoch": 5.67,
"learning_rate": 0.0001,
"loss": 0.4357,
"step": 319000
},
{
"epoch": 5.68,
"learning_rate": 0.0001,
"loss": 0.4449,
"step": 319500
},
{
"epoch": 5.69,
"learning_rate": 0.0001,
"loss": 0.4394,
"step": 320000
},
{
"epoch": 5.7,
"learning_rate": 0.0001,
"loss": 0.4415,
"step": 320500
},
{
"epoch": 5.71,
"learning_rate": 0.0001,
"loss": 0.4417,
"step": 321000
},
{
"epoch": 5.72,
"learning_rate": 0.0001,
"loss": 0.4418,
"step": 321500
},
{
"epoch": 5.72,
"learning_rate": 0.0001,
"loss": 0.4398,
"step": 322000
},
{
"epoch": 5.73,
"learning_rate": 0.0001,
"loss": 0.4362,
"step": 322500
},
{
"epoch": 5.74,
"learning_rate": 0.0001,
"loss": 0.4383,
"step": 323000
},
{
"epoch": 5.75,
"learning_rate": 0.0001,
"loss": 0.4299,
"step": 323500
},
{
"epoch": 5.76,
"learning_rate": 0.0001,
"loss": 0.4391,
"step": 324000
},
{
"epoch": 5.77,
"learning_rate": 0.0001,
"loss": 0.4328,
"step": 324500
},
{
"epoch": 5.78,
"learning_rate": 0.0001,
"loss": 0.4364,
"step": 325000
},
{
"epoch": 5.79,
"learning_rate": 0.0001,
"loss": 0.4424,
"step": 325500
},
{
"epoch": 5.8,
"learning_rate": 0.0001,
"loss": 0.4362,
"step": 326000
},
{
"epoch": 5.8,
"learning_rate": 0.0001,
"loss": 0.4362,
"step": 326500
},
{
"epoch": 5.81,
"learning_rate": 0.0001,
"loss": 0.4401,
"step": 327000
},
{
"epoch": 5.82,
"learning_rate": 0.0001,
"loss": 0.435,
"step": 327500
},
{
"epoch": 5.83,
"learning_rate": 0.0001,
"loss": 0.4407,
"step": 328000
},
{
"epoch": 5.84,
"learning_rate": 0.0001,
"loss": 0.4341,
"step": 328500
},
{
"epoch": 5.85,
"learning_rate": 0.0001,
"loss": 0.4432,
"step": 329000
},
{
"epoch": 5.86,
"learning_rate": 0.0001,
"loss": 0.4371,
"step": 329500
},
{
"epoch": 5.87,
"learning_rate": 0.0001,
"loss": 0.4366,
"step": 330000
},
{
"epoch": 5.88,
"learning_rate": 0.0001,
"loss": 0.4371,
"step": 330500
},
{
"epoch": 5.88,
"learning_rate": 0.0001,
"loss": 0.4381,
"step": 331000
},
{
"epoch": 5.89,
"learning_rate": 0.0001,
"loss": 0.4376,
"step": 331500
},
{
"epoch": 5.9,
"learning_rate": 0.0001,
"loss": 0.4356,
"step": 332000
},
{
"epoch": 5.91,
"learning_rate": 0.0001,
"loss": 0.4385,
"step": 332500
},
{
"epoch": 5.92,
"learning_rate": 0.0001,
"loss": 0.4406,
"step": 333000
},
{
"epoch": 5.93,
"learning_rate": 0.0001,
"loss": 0.4348,
"step": 333500
},
{
"epoch": 5.94,
"learning_rate": 0.0001,
"loss": 0.433,
"step": 334000
},
{
"epoch": 5.95,
"learning_rate": 0.0001,
"loss": 0.4328,
"step": 334500
},
{
"epoch": 5.96,
"learning_rate": 0.0001,
"loss": 0.4345,
"step": 335000
},
{
"epoch": 5.96,
"learning_rate": 0.0001,
"loss": 0.4367,
"step": 335500
},
{
"epoch": 5.97,
"learning_rate": 0.0001,
"loss": 0.4316,
"step": 336000
},
{
"epoch": 5.98,
"learning_rate": 0.0001,
"loss": 0.426,
"step": 336500
},
{
"epoch": 5.99,
"learning_rate": 0.0001,
"loss": 0.4322,
"step": 337000
},
{
"epoch": 6.0,
"learning_rate": 0.0001,
"loss": 0.4351,
"step": 337500
},
{
"epoch": 6.0,
"eval_bleu": 27.4783,
"eval_gen_len": 11.2419,
"eval_loss": 1.0986744165420532,
"eval_runtime": 809.2585,
"eval_samples_per_second": 17.3,
"step": 337500
},
{
"epoch": 6.0,
"step": 337500,
"total_flos": 4721657962807296.0,
"train_runtime": 4215.8001,
"train_samples_per_second": 80.056
}
],
"max_steps": 337500,
"num_train_epochs": 6,
"total_flos": 4721657962807296.0,
"trial_name": null,
"trial_params": null
}