AhmedSSoliman's picture
Upload 12 files
3841fa3
raw
history blame
No virus
46.7 kB
{
"best_metric": 0.2726329565048218,
"best_model_checkpoint": "./distilroberta-marian-training1/checkpoint-144000",
"epoch": 10.0,
"global_step": 160000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 6.225000000000001e-07,
"loss": 0.1837,
"step": 500
},
{
"epoch": 0.06,
"learning_rate": 1.2462500000000001e-06,
"loss": 0.2157,
"step": 1000
},
{
"epoch": 0.09,
"learning_rate": 1.8712500000000003e-06,
"loss": 0.22,
"step": 1500
},
{
"epoch": 0.12,
"learning_rate": 2.4962500000000005e-06,
"loss": 0.1969,
"step": 2000
},
{
"epoch": 0.16,
"learning_rate": 3.12125e-06,
"loss": 0.2,
"step": 2500
},
{
"epoch": 0.19,
"learning_rate": 3.745e-06,
"loss": 0.2612,
"step": 3000
},
{
"epoch": 0.22,
"learning_rate": 4.3700000000000005e-06,
"loss": 0.2172,
"step": 3500
},
{
"epoch": 0.25,
"learning_rate": 4.9950000000000005e-06,
"loss": 0.1853,
"step": 4000
},
{
"epoch": 0.28,
"learning_rate": 5.620000000000001e-06,
"loss": 0.1987,
"step": 4500
},
{
"epoch": 0.31,
"learning_rate": 6.243750000000001e-06,
"loss": 0.232,
"step": 5000
},
{
"epoch": 0.34,
"learning_rate": 6.868750000000001e-06,
"loss": 0.2513,
"step": 5500
},
{
"epoch": 0.38,
"learning_rate": 7.493750000000001e-06,
"loss": 0.2117,
"step": 6000
},
{
"epoch": 0.41,
"learning_rate": 8.11875e-06,
"loss": 0.2349,
"step": 6500
},
{
"epoch": 0.44,
"learning_rate": 8.74375e-06,
"loss": 0.2328,
"step": 7000
},
{
"epoch": 0.47,
"learning_rate": 9.3675e-06,
"loss": 0.2738,
"step": 7500
},
{
"epoch": 0.5,
"learning_rate": 9.992500000000001e-06,
"loss": 0.2504,
"step": 8000
},
{
"epoch": 0.53,
"learning_rate": 9.9675e-06,
"loss": 0.246,
"step": 8500
},
{
"epoch": 0.56,
"learning_rate": 9.934605263157895e-06,
"loss": 0.2218,
"step": 9000
},
{
"epoch": 0.59,
"learning_rate": 9.901776315789475e-06,
"loss": 0.2868,
"step": 9500
},
{
"epoch": 0.62,
"learning_rate": 9.86888157894737e-06,
"loss": 0.2406,
"step": 10000
},
{
"epoch": 0.66,
"learning_rate": 9.835986842105264e-06,
"loss": 0.2783,
"step": 10500
},
{
"epoch": 0.69,
"learning_rate": 9.803092105263159e-06,
"loss": 0.2784,
"step": 11000
},
{
"epoch": 0.72,
"learning_rate": 9.770197368421053e-06,
"loss": 0.2379,
"step": 11500
},
{
"epoch": 0.75,
"learning_rate": 9.737302631578948e-06,
"loss": 0.2887,
"step": 12000
},
{
"epoch": 0.78,
"learning_rate": 9.704473684210527e-06,
"loss": 0.2569,
"step": 12500
},
{
"epoch": 0.81,
"learning_rate": 9.671578947368422e-06,
"loss": 0.2794,
"step": 13000
},
{
"epoch": 0.84,
"learning_rate": 9.638684210526316e-06,
"loss": 0.2573,
"step": 13500
},
{
"epoch": 0.88,
"learning_rate": 9.605789473684211e-06,
"loss": 0.251,
"step": 14000
},
{
"epoch": 0.91,
"learning_rate": 9.572894736842107e-06,
"loss": 0.268,
"step": 14500
},
{
"epoch": 0.94,
"learning_rate": 9.54e-06,
"loss": 0.2304,
"step": 15000
},
{
"epoch": 0.97,
"learning_rate": 9.507105263157895e-06,
"loss": 0.2631,
"step": 15500
},
{
"epoch": 1.0,
"learning_rate": 9.474210526315791e-06,
"loss": 0.2674,
"step": 16000
},
{
"epoch": 1.0,
"eval_BLEU": 73.83414614372862,
"eval_BLEU-Bigram-Precision": 79.1290558672847,
"eval_BLEU-Trigram-Precision": 71.86590940573487,
"eval_BLEU-Unigram-Precision": 87.71472059143292,
"eval_ROUGE-2": 72.64315140484014,
"eval_ROUGE-L": 88.44220304691733,
"eval_Sacre-Bigram-Precision": 79.870557261957,
"eval_Sacre-Trigram-Precision": 73.81098685019049,
"eval_Sacre-Unigram-Precision": 87.78173190984579,
"eval_SacreBLEU": 75.40013818024555,
"eval_loss": 0.32742100954055786,
"eval_runtime": 88.0501,
"eval_samples_per_second": 11.357,
"eval_steps_per_second": 11.357,
"step": 16000
},
{
"epoch": 1.03,
"learning_rate": 9.441315789473686e-06,
"loss": 0.1756,
"step": 16500
},
{
"epoch": 1.06,
"learning_rate": 9.408486842105265e-06,
"loss": 0.277,
"step": 17000
},
{
"epoch": 1.09,
"learning_rate": 9.375592105263158e-06,
"loss": 0.2313,
"step": 17500
},
{
"epoch": 1.12,
"learning_rate": 9.342697368421053e-06,
"loss": 0.2596,
"step": 18000
},
{
"epoch": 1.16,
"learning_rate": 9.309802631578949e-06,
"loss": 0.2473,
"step": 18500
},
{
"epoch": 1.19,
"learning_rate": 9.276907894736843e-06,
"loss": 0.2274,
"step": 19000
},
{
"epoch": 1.22,
"learning_rate": 9.244144736842106e-06,
"loss": 0.2314,
"step": 19500
},
{
"epoch": 1.25,
"learning_rate": 9.211250000000001e-06,
"loss": 0.1981,
"step": 20000
},
{
"epoch": 1.28,
"learning_rate": 9.178355263157896e-06,
"loss": 0.2544,
"step": 20500
},
{
"epoch": 1.31,
"learning_rate": 9.14546052631579e-06,
"loss": 0.2123,
"step": 21000
},
{
"epoch": 1.34,
"learning_rate": 9.112565789473685e-06,
"loss": 0.2215,
"step": 21500
},
{
"epoch": 1.38,
"learning_rate": 9.079671052631581e-06,
"loss": 0.2134,
"step": 22000
},
{
"epoch": 1.41,
"learning_rate": 9.046776315789474e-06,
"loss": 0.2684,
"step": 22500
},
{
"epoch": 1.44,
"learning_rate": 9.013881578947368e-06,
"loss": 0.2485,
"step": 23000
},
{
"epoch": 1.47,
"learning_rate": 8.981052631578948e-06,
"loss": 0.1944,
"step": 23500
},
{
"epoch": 1.5,
"learning_rate": 8.948157894736843e-06,
"loss": 0.2126,
"step": 24000
},
{
"epoch": 1.53,
"learning_rate": 8.915263157894739e-06,
"loss": 0.2315,
"step": 24500
},
{
"epoch": 1.56,
"learning_rate": 8.882368421052632e-06,
"loss": 0.224,
"step": 25000
},
{
"epoch": 1.59,
"learning_rate": 8.849473684210526e-06,
"loss": 0.2295,
"step": 25500
},
{
"epoch": 1.62,
"learning_rate": 8.81657894736842e-06,
"loss": 0.1894,
"step": 26000
},
{
"epoch": 1.66,
"learning_rate": 8.783684210526317e-06,
"loss": 0.2191,
"step": 26500
},
{
"epoch": 1.69,
"learning_rate": 8.750855263157895e-06,
"loss": 0.1989,
"step": 27000
},
{
"epoch": 1.72,
"learning_rate": 8.717960526315791e-06,
"loss": 0.1923,
"step": 27500
},
{
"epoch": 1.75,
"learning_rate": 8.685065789473684e-06,
"loss": 0.1979,
"step": 28000
},
{
"epoch": 1.78,
"learning_rate": 8.652171052631579e-06,
"loss": 0.2166,
"step": 28500
},
{
"epoch": 1.81,
"learning_rate": 8.619342105263158e-06,
"loss": 0.234,
"step": 29000
},
{
"epoch": 1.84,
"learning_rate": 8.586447368421053e-06,
"loss": 0.2352,
"step": 29500
},
{
"epoch": 1.88,
"learning_rate": 8.553552631578949e-06,
"loss": 0.1886,
"step": 30000
},
{
"epoch": 1.91,
"learning_rate": 8.520657894736842e-06,
"loss": 0.2233,
"step": 30500
},
{
"epoch": 1.94,
"learning_rate": 8.487763157894737e-06,
"loss": 0.2737,
"step": 31000
},
{
"epoch": 1.97,
"learning_rate": 8.454868421052633e-06,
"loss": 0.1909,
"step": 31500
},
{
"epoch": 2.0,
"learning_rate": 8.421973684210527e-06,
"loss": 0.1936,
"step": 32000
},
{
"epoch": 2.0,
"eval_BLEU": 75.36161100581221,
"eval_BLEU-Bigram-Precision": 80.48572125945815,
"eval_BLEU-Trigram-Precision": 73.76299376299377,
"eval_BLEU-Unigram-Precision": 88.22057863824233,
"eval_ROUGE-2": 74.09506184806267,
"eval_ROUGE-L": 89.05035697782122,
"eval_Sacre-Bigram-Precision": 81.32291437376183,
"eval_Sacre-Trigram-Precision": 75.74935241149623,
"eval_Sacre-Unigram-Precision": 88.4195915129883,
"eval_SacreBLEU": 76.75621822222277,
"eval_loss": 0.3107452392578125,
"eval_runtime": 88.8932,
"eval_samples_per_second": 11.249,
"eval_steps_per_second": 11.249,
"step": 32000
},
{
"epoch": 2.03,
"learning_rate": 8.389078947368422e-06,
"loss": 0.1905,
"step": 32500
},
{
"epoch": 2.06,
"learning_rate": 8.356184210526316e-06,
"loss": 0.1765,
"step": 33000
},
{
"epoch": 2.09,
"learning_rate": 8.323355263157894e-06,
"loss": 0.1643,
"step": 33500
},
{
"epoch": 2.12,
"learning_rate": 8.29046052631579e-06,
"loss": 0.1843,
"step": 34000
},
{
"epoch": 2.16,
"learning_rate": 8.257565789473685e-06,
"loss": 0.1825,
"step": 34500
},
{
"epoch": 2.19,
"learning_rate": 8.224736842105265e-06,
"loss": 0.1374,
"step": 35000
},
{
"epoch": 2.22,
"learning_rate": 8.191842105263158e-06,
"loss": 0.2312,
"step": 35500
},
{
"epoch": 2.25,
"learning_rate": 8.158947368421052e-06,
"loss": 0.217,
"step": 36000
},
{
"epoch": 2.28,
"learning_rate": 8.126052631578949e-06,
"loss": 0.174,
"step": 36500
},
{
"epoch": 2.31,
"learning_rate": 8.093157894736843e-06,
"loss": 0.1485,
"step": 37000
},
{
"epoch": 2.34,
"learning_rate": 8.060263157894738e-06,
"loss": 0.1364,
"step": 37500
},
{
"epoch": 2.38,
"learning_rate": 8.027434210526317e-06,
"loss": 0.1894,
"step": 38000
},
{
"epoch": 2.41,
"learning_rate": 7.99453947368421e-06,
"loss": 0.1658,
"step": 38500
},
{
"epoch": 2.44,
"learning_rate": 7.961644736842106e-06,
"loss": 0.2059,
"step": 39000
},
{
"epoch": 2.47,
"learning_rate": 7.928750000000001e-06,
"loss": 0.2102,
"step": 39500
},
{
"epoch": 2.5,
"learning_rate": 7.89592105263158e-06,
"loss": 0.2032,
"step": 40000
},
{
"epoch": 2.53,
"learning_rate": 7.863026315789475e-06,
"loss": 0.2062,
"step": 40500
},
{
"epoch": 2.56,
"learning_rate": 7.830131578947368e-06,
"loss": 0.1848,
"step": 41000
},
{
"epoch": 2.59,
"learning_rate": 7.797236842105264e-06,
"loss": 0.2161,
"step": 41500
},
{
"epoch": 2.62,
"learning_rate": 7.764342105263159e-06,
"loss": 0.1746,
"step": 42000
},
{
"epoch": 2.66,
"learning_rate": 7.731447368421053e-06,
"loss": 0.1979,
"step": 42500
},
{
"epoch": 2.69,
"learning_rate": 7.698618421052633e-06,
"loss": 0.1569,
"step": 43000
},
{
"epoch": 2.72,
"learning_rate": 7.665723684210526e-06,
"loss": 0.2154,
"step": 43500
},
{
"epoch": 2.75,
"learning_rate": 7.632828947368422e-06,
"loss": 0.1723,
"step": 44000
},
{
"epoch": 2.78,
"learning_rate": 7.599934210526317e-06,
"loss": 0.1751,
"step": 44500
},
{
"epoch": 2.81,
"learning_rate": 7.5671052631578955e-06,
"loss": 0.1828,
"step": 45000
},
{
"epoch": 2.84,
"learning_rate": 7.534210526315791e-06,
"loss": 0.2034,
"step": 45500
},
{
"epoch": 2.88,
"learning_rate": 7.501315789473685e-06,
"loss": 0.1736,
"step": 46000
},
{
"epoch": 2.91,
"learning_rate": 7.468421052631579e-06,
"loss": 0.1858,
"step": 46500
},
{
"epoch": 2.94,
"learning_rate": 7.435526315789474e-06,
"loss": 0.19,
"step": 47000
},
{
"epoch": 2.97,
"learning_rate": 7.402697368421053e-06,
"loss": 0.176,
"step": 47500
},
{
"epoch": 3.0,
"learning_rate": 7.369802631578949e-06,
"loss": 0.17,
"step": 48000
},
{
"epoch": 3.0,
"eval_BLEU": 76.96135367983078,
"eval_BLEU-Bigram-Precision": 80.3268909568122,
"eval_BLEU-Trigram-Precision": 73.89894623074844,
"eval_BLEU-Unigram-Precision": 88.08356427201022,
"eval_ROUGE-2": 74.80911471749707,
"eval_ROUGE-L": 89.60895451818533,
"eval_Sacre-Bigram-Precision": 81.12129380053908,
"eval_Sacre-Trigram-Precision": 75.68414707655214,
"eval_Sacre-Unigram-Precision": 88.19464720194647,
"eval_SacreBLEU": 78.11560799454091,
"eval_loss": 0.3022772967815399,
"eval_runtime": 90.1953,
"eval_samples_per_second": 11.087,
"eval_steps_per_second": 11.087,
"step": 48000
},
{
"epoch": 3.03,
"learning_rate": 7.3369078947368425e-06,
"loss": 0.1881,
"step": 48500
},
{
"epoch": 3.06,
"learning_rate": 7.304013157894737e-06,
"loss": 0.148,
"step": 49000
},
{
"epoch": 3.09,
"learning_rate": 7.271118421052632e-06,
"loss": 0.1596,
"step": 49500
},
{
"epoch": 3.12,
"learning_rate": 7.238223684210527e-06,
"loss": 0.1553,
"step": 50000
},
{
"epoch": 3.16,
"learning_rate": 7.205328947368422e-06,
"loss": 0.1953,
"step": 50500
},
{
"epoch": 3.19,
"learning_rate": 7.172434210526317e-06,
"loss": 0.1691,
"step": 51000
},
{
"epoch": 3.22,
"learning_rate": 7.1395394736842115e-06,
"loss": 0.1448,
"step": 51500
},
{
"epoch": 3.25,
"learning_rate": 7.1067105263157895e-06,
"loss": 0.1379,
"step": 52000
},
{
"epoch": 3.28,
"learning_rate": 7.073881578947369e-06,
"loss": 0.1322,
"step": 52500
},
{
"epoch": 3.31,
"learning_rate": 7.040986842105264e-06,
"loss": 0.1439,
"step": 53000
},
{
"epoch": 3.34,
"learning_rate": 7.008092105263159e-06,
"loss": 0.1838,
"step": 53500
},
{
"epoch": 3.38,
"learning_rate": 6.975197368421053e-06,
"loss": 0.1548,
"step": 54000
},
{
"epoch": 3.41,
"learning_rate": 6.9423684210526324e-06,
"loss": 0.1375,
"step": 54500
},
{
"epoch": 3.44,
"learning_rate": 6.909473684210527e-06,
"loss": 0.1658,
"step": 55000
},
{
"epoch": 3.47,
"learning_rate": 6.876644736842107e-06,
"loss": 0.1514,
"step": 55500
},
{
"epoch": 3.5,
"learning_rate": 6.84375e-06,
"loss": 0.1835,
"step": 56000
},
{
"epoch": 3.53,
"learning_rate": 6.810855263157895e-06,
"loss": 0.1397,
"step": 56500
},
{
"epoch": 3.56,
"learning_rate": 6.77796052631579e-06,
"loss": 0.1324,
"step": 57000
},
{
"epoch": 3.59,
"learning_rate": 6.745065789473685e-06,
"loss": 0.1426,
"step": 57500
},
{
"epoch": 3.62,
"learning_rate": 6.7121710526315794e-06,
"loss": 0.1932,
"step": 58000
},
{
"epoch": 3.66,
"learning_rate": 6.679276315789475e-06,
"loss": 0.1643,
"step": 58500
},
{
"epoch": 3.69,
"learning_rate": 6.6463815789473685e-06,
"loss": 0.1606,
"step": 59000
},
{
"epoch": 3.72,
"learning_rate": 6.613486842105263e-06,
"loss": 0.1627,
"step": 59500
},
{
"epoch": 3.75,
"learning_rate": 6.5805921052631585e-06,
"loss": 0.1643,
"step": 60000
},
{
"epoch": 3.78,
"learning_rate": 6.547697368421053e-06,
"loss": 0.1321,
"step": 60500
},
{
"epoch": 3.81,
"learning_rate": 6.5148026315789485e-06,
"loss": 0.1282,
"step": 61000
},
{
"epoch": 3.84,
"learning_rate": 6.481907894736843e-06,
"loss": 0.1495,
"step": 61500
},
{
"epoch": 3.88,
"learning_rate": 6.449078947368421e-06,
"loss": 0.1518,
"step": 62000
},
{
"epoch": 3.91,
"learning_rate": 6.416184210526316e-06,
"loss": 0.1508,
"step": 62500
},
{
"epoch": 3.94,
"learning_rate": 6.383289473684211e-06,
"loss": 0.1487,
"step": 63000
},
{
"epoch": 3.97,
"learning_rate": 6.350394736842106e-06,
"loss": 0.1368,
"step": 63500
},
{
"epoch": 4.0,
"learning_rate": 6.317500000000001e-06,
"loss": 0.1371,
"step": 64000
},
{
"epoch": 4.0,
"eval_BLEU": 76.97562999548838,
"eval_BLEU-Bigram-Precision": 80.0071106897369,
"eval_BLEU-Trigram-Precision": 73.54518637704479,
"eval_BLEU-Unigram-Precision": 87.63509218054672,
"eval_ROUGE-2": 75.31771486560173,
"eval_ROUGE-L": 89.81535115722117,
"eval_Sacre-Bigram-Precision": 80.88565621991657,
"eval_Sacre-Trigram-Precision": 75.51678814673198,
"eval_Sacre-Unigram-Precision": 87.88288723548169,
"eval_SacreBLEU": 78.47888708880886,
"eval_loss": 0.2903847098350525,
"eval_runtime": 90.0073,
"eval_samples_per_second": 11.11,
"eval_steps_per_second": 11.11,
"step": 64000
},
{
"epoch": 4.03,
"learning_rate": 6.2846052631578955e-06,
"loss": 0.1406,
"step": 64500
},
{
"epoch": 4.06,
"learning_rate": 6.251710526315789e-06,
"loss": 0.1134,
"step": 65000
},
{
"epoch": 4.09,
"learning_rate": 6.218815789473685e-06,
"loss": 0.126,
"step": 65500
},
{
"epoch": 4.12,
"learning_rate": 6.1860526315789485e-06,
"loss": 0.1274,
"step": 66000
},
{
"epoch": 4.16,
"learning_rate": 6.153157894736843e-06,
"loss": 0.134,
"step": 66500
},
{
"epoch": 4.19,
"learning_rate": 6.120263157894737e-06,
"loss": 0.1462,
"step": 67000
},
{
"epoch": 4.22,
"learning_rate": 6.087368421052632e-06,
"loss": 0.1203,
"step": 67500
},
{
"epoch": 4.25,
"learning_rate": 6.054539473684211e-06,
"loss": 0.1602,
"step": 68000
},
{
"epoch": 4.28,
"learning_rate": 6.021644736842106e-06,
"loss": 0.1367,
"step": 68500
},
{
"epoch": 4.31,
"learning_rate": 5.988750000000001e-06,
"loss": 0.1087,
"step": 69000
},
{
"epoch": 4.34,
"learning_rate": 5.955921052631579e-06,
"loss": 0.1203,
"step": 69500
},
{
"epoch": 4.38,
"learning_rate": 5.923026315789474e-06,
"loss": 0.1199,
"step": 70000
},
{
"epoch": 4.41,
"learning_rate": 5.890131578947369e-06,
"loss": 0.1183,
"step": 70500
},
{
"epoch": 4.44,
"learning_rate": 5.857236842105264e-06,
"loss": 0.1366,
"step": 71000
},
{
"epoch": 4.47,
"learning_rate": 5.824407894736842e-06,
"loss": 0.1205,
"step": 71500
},
{
"epoch": 4.5,
"learning_rate": 5.791513157894737e-06,
"loss": 0.1537,
"step": 72000
},
{
"epoch": 4.53,
"learning_rate": 5.758618421052632e-06,
"loss": 0.1386,
"step": 72500
},
{
"epoch": 4.56,
"learning_rate": 5.725723684210527e-06,
"loss": 0.1634,
"step": 73000
},
{
"epoch": 4.59,
"learning_rate": 5.692828947368422e-06,
"loss": 0.1005,
"step": 73500
},
{
"epoch": 4.62,
"learning_rate": 5.659934210526317e-06,
"loss": 0.1546,
"step": 74000
},
{
"epoch": 4.66,
"learning_rate": 5.62703947368421e-06,
"loss": 0.1212,
"step": 74500
},
{
"epoch": 4.69,
"learning_rate": 5.594144736842106e-06,
"loss": 0.1446,
"step": 75000
},
{
"epoch": 4.72,
"learning_rate": 5.5613157894736846e-06,
"loss": 0.1339,
"step": 75500
},
{
"epoch": 4.75,
"learning_rate": 5.52842105263158e-06,
"loss": 0.1353,
"step": 76000
},
{
"epoch": 4.78,
"learning_rate": 5.4955263157894745e-06,
"loss": 0.1512,
"step": 76500
},
{
"epoch": 4.81,
"learning_rate": 5.462631578947369e-06,
"loss": 0.1217,
"step": 77000
},
{
"epoch": 4.84,
"learning_rate": 5.429736842105263e-06,
"loss": 0.1288,
"step": 77500
},
{
"epoch": 4.88,
"learning_rate": 5.3969078947368424e-06,
"loss": 0.155,
"step": 78000
},
{
"epoch": 4.91,
"learning_rate": 5.364013157894738e-06,
"loss": 0.128,
"step": 78500
},
{
"epoch": 4.94,
"learning_rate": 5.331118421052632e-06,
"loss": 0.1224,
"step": 79000
},
{
"epoch": 4.97,
"learning_rate": 5.298223684210527e-06,
"loss": 0.1245,
"step": 79500
},
{
"epoch": 5.0,
"learning_rate": 5.265394736842106e-06,
"loss": 0.1314,
"step": 80000
},
{
"epoch": 5.0,
"eval_BLEU": 77.77478989310802,
"eval_BLEU-Bigram-Precision": 82.58159149248259,
"eval_BLEU-Trigram-Precision": 76.65601999722261,
"eval_BLEU-Unigram-Precision": 89.58719093780634,
"eval_ROUGE-2": 76.21694739487536,
"eval_ROUGE-L": 90.30321591632153,
"eval_Sacre-Bigram-Precision": 83.30402286216751,
"eval_Sacre-Trigram-Precision": 78.39369302783938,
"eval_Sacre-Unigram-Precision": 89.64151317092494,
"eval_SacreBLEU": 79.14676248093514,
"eval_loss": 0.2857072055339813,
"eval_runtime": 88.6134,
"eval_samples_per_second": 11.285,
"eval_steps_per_second": 11.285,
"step": 80000
},
{
"epoch": 5.03,
"learning_rate": 5.2325e-06,
"loss": 0.0962,
"step": 80500
},
{
"epoch": 5.06,
"learning_rate": 5.199605263157895e-06,
"loss": 0.1223,
"step": 81000
},
{
"epoch": 5.09,
"learning_rate": 5.1667763157894745e-06,
"loss": 0.1234,
"step": 81500
},
{
"epoch": 5.12,
"learning_rate": 5.133881578947368e-06,
"loss": 0.0942,
"step": 82000
},
{
"epoch": 5.16,
"learning_rate": 5.100986842105264e-06,
"loss": 0.088,
"step": 82500
},
{
"epoch": 5.19,
"learning_rate": 5.068092105263158e-06,
"loss": 0.1013,
"step": 83000
},
{
"epoch": 5.22,
"learning_rate": 5.035197368421053e-06,
"loss": 0.1243,
"step": 83500
},
{
"epoch": 5.25,
"learning_rate": 5.002302631578948e-06,
"loss": 0.1097,
"step": 84000
},
{
"epoch": 5.28,
"learning_rate": 4.969407894736843e-06,
"loss": 0.1122,
"step": 84500
},
{
"epoch": 5.31,
"learning_rate": 4.936513157894737e-06,
"loss": 0.1474,
"step": 85000
},
{
"epoch": 5.34,
"learning_rate": 4.903618421052632e-06,
"loss": 0.1107,
"step": 85500
},
{
"epoch": 5.38,
"learning_rate": 4.870723684210526e-06,
"loss": 0.119,
"step": 86000
},
{
"epoch": 5.41,
"learning_rate": 4.837894736842106e-06,
"loss": 0.0859,
"step": 86500
},
{
"epoch": 5.44,
"learning_rate": 4.805000000000001e-06,
"loss": 0.1172,
"step": 87000
},
{
"epoch": 5.47,
"learning_rate": 4.772171052631579e-06,
"loss": 0.1017,
"step": 87500
},
{
"epoch": 5.5,
"learning_rate": 4.739276315789474e-06,
"loss": 0.0847,
"step": 88000
},
{
"epoch": 5.53,
"learning_rate": 4.7063815789473685e-06,
"loss": 0.1437,
"step": 88500
},
{
"epoch": 5.56,
"learning_rate": 4.673486842105264e-06,
"loss": 0.1294,
"step": 89000
},
{
"epoch": 5.59,
"learning_rate": 4.6405921052631585e-06,
"loss": 0.1037,
"step": 89500
},
{
"epoch": 5.62,
"learning_rate": 4.607697368421053e-06,
"loss": 0.1108,
"step": 90000
},
{
"epoch": 5.66,
"learning_rate": 4.574868421052632e-06,
"loss": 0.1186,
"step": 90500
},
{
"epoch": 5.69,
"learning_rate": 4.541973684210526e-06,
"loss": 0.1252,
"step": 91000
},
{
"epoch": 5.72,
"learning_rate": 4.509078947368422e-06,
"loss": 0.1202,
"step": 91500
},
{
"epoch": 5.75,
"learning_rate": 4.476184210526316e-06,
"loss": 0.122,
"step": 92000
},
{
"epoch": 5.78,
"learning_rate": 4.443289473684211e-06,
"loss": 0.136,
"step": 92500
},
{
"epoch": 5.81,
"learning_rate": 4.4103947368421055e-06,
"loss": 0.1213,
"step": 93000
},
{
"epoch": 5.84,
"learning_rate": 4.3775e-06,
"loss": 0.1041,
"step": 93500
},
{
"epoch": 5.88,
"learning_rate": 4.344605263157895e-06,
"loss": 0.1339,
"step": 94000
},
{
"epoch": 5.91,
"learning_rate": 4.31171052631579e-06,
"loss": 0.0929,
"step": 94500
},
{
"epoch": 5.94,
"learning_rate": 4.2788157894736845e-06,
"loss": 0.1185,
"step": 95000
},
{
"epoch": 5.97,
"learning_rate": 4.245986842105263e-06,
"loss": 0.1242,
"step": 95500
},
{
"epoch": 6.0,
"learning_rate": 4.213092105263159e-06,
"loss": 0.1,
"step": 96000
},
{
"epoch": 6.0,
"eval_BLEU": 79.11019971744633,
"eval_BLEU-Bigram-Precision": 85.00372116100223,
"eval_BLEU-Trigram-Precision": 79.31375317706862,
"eval_BLEU-Unigram-Precision": 91.3153829176782,
"eval_ROUGE-2": 76.72218981179273,
"eval_ROUGE-L": 90.74286437693775,
"eval_Sacre-Bigram-Precision": 85.59587813620071,
"eval_Sacre-Trigram-Precision": 81.08958228485153,
"eval_Sacre-Unigram-Precision": 91.41821112006447,
"eval_SacreBLEU": 80.09298309615393,
"eval_loss": 0.2855425179004669,
"eval_runtime": 87.7278,
"eval_samples_per_second": 11.399,
"eval_steps_per_second": 11.399,
"step": 96000
},
{
"epoch": 6.03,
"learning_rate": 4.1801973684210525e-06,
"loss": 0.0992,
"step": 96500
},
{
"epoch": 6.06,
"learning_rate": 4.147302631578948e-06,
"loss": 0.0906,
"step": 97000
},
{
"epoch": 6.09,
"learning_rate": 4.114407894736842e-06,
"loss": 0.114,
"step": 97500
},
{
"epoch": 6.12,
"learning_rate": 4.081578947368421e-06,
"loss": 0.0868,
"step": 98000
},
{
"epoch": 6.16,
"learning_rate": 4.048684210526317e-06,
"loss": 0.0837,
"step": 98500
},
{
"epoch": 6.19,
"learning_rate": 4.01578947368421e-06,
"loss": 0.1157,
"step": 99000
},
{
"epoch": 6.22,
"learning_rate": 3.982894736842106e-06,
"loss": 0.0993,
"step": 99500
},
{
"epoch": 6.25,
"learning_rate": 3.95e-06,
"loss": 0.1214,
"step": 100000
},
{
"epoch": 6.28,
"learning_rate": 3.917171052631579e-06,
"loss": 0.1167,
"step": 100500
},
{
"epoch": 6.31,
"learning_rate": 3.884342105263158e-06,
"loss": 0.1021,
"step": 101000
},
{
"epoch": 6.34,
"learning_rate": 3.851447368421053e-06,
"loss": 0.0963,
"step": 101500
},
{
"epoch": 6.38,
"learning_rate": 3.818552631578948e-06,
"loss": 0.1014,
"step": 102000
},
{
"epoch": 6.41,
"learning_rate": 3.7856578947368424e-06,
"loss": 0.096,
"step": 102500
},
{
"epoch": 6.44,
"learning_rate": 3.752763157894737e-06,
"loss": 0.105,
"step": 103000
},
{
"epoch": 6.47,
"learning_rate": 3.719868421052632e-06,
"loss": 0.095,
"step": 103500
},
{
"epoch": 6.5,
"learning_rate": 3.6870394736842107e-06,
"loss": 0.1049,
"step": 104000
},
{
"epoch": 6.53,
"learning_rate": 3.6541447368421057e-06,
"loss": 0.1037,
"step": 104500
},
{
"epoch": 6.56,
"learning_rate": 3.62125e-06,
"loss": 0.0754,
"step": 105000
},
{
"epoch": 6.59,
"learning_rate": 3.588355263157895e-06,
"loss": 0.0996,
"step": 105500
},
{
"epoch": 6.62,
"learning_rate": 3.55546052631579e-06,
"loss": 0.1024,
"step": 106000
},
{
"epoch": 6.66,
"learning_rate": 3.522565789473685e-06,
"loss": 0.0926,
"step": 106500
},
{
"epoch": 6.69,
"learning_rate": 3.489671052631579e-06,
"loss": 0.0782,
"step": 107000
},
{
"epoch": 6.72,
"learning_rate": 3.456776315789474e-06,
"loss": 0.1045,
"step": 107500
},
{
"epoch": 6.75,
"learning_rate": 3.423881578947369e-06,
"loss": 0.1079,
"step": 108000
},
{
"epoch": 6.78,
"learning_rate": 3.3910526315789477e-06,
"loss": 0.0924,
"step": 108500
},
{
"epoch": 6.81,
"learning_rate": 3.3582236842105265e-06,
"loss": 0.1352,
"step": 109000
},
{
"epoch": 6.84,
"learning_rate": 3.3253289473684215e-06,
"loss": 0.0865,
"step": 109500
},
{
"epoch": 6.88,
"learning_rate": 3.2924342105263164e-06,
"loss": 0.0924,
"step": 110000
},
{
"epoch": 6.91,
"learning_rate": 3.2595394736842106e-06,
"loss": 0.0812,
"step": 110500
},
{
"epoch": 6.94,
"learning_rate": 3.2266447368421056e-06,
"loss": 0.1081,
"step": 111000
},
{
"epoch": 6.97,
"learning_rate": 3.1937500000000006e-06,
"loss": 0.0872,
"step": 111500
},
{
"epoch": 7.0,
"learning_rate": 3.1608552631578947e-06,
"loss": 0.0855,
"step": 112000
},
{
"epoch": 7.0,
"eval_BLEU": 79.18243013411383,
"eval_BLEU-Bigram-Precision": 83.65829880575188,
"eval_BLEU-Trigram-Precision": 78.10683642402437,
"eval_BLEU-Unigram-Precision": 90.08255485552901,
"eval_ROUGE-2": 76.92605201616105,
"eval_ROUGE-L": 90.80400601001202,
"eval_Sacre-Bigram-Precision": 84.62469733656174,
"eval_Sacre-Trigram-Precision": 80.1505057981742,
"eval_Sacre-Unigram-Precision": 90.4322823716042,
"eval_SacreBLEU": 80.51163956701332,
"eval_loss": 0.2782886326313019,
"eval_runtime": 88.5286,
"eval_samples_per_second": 11.296,
"eval_steps_per_second": 11.296,
"step": 112000
},
{
"epoch": 7.03,
"learning_rate": 3.1279605263157897e-06,
"loss": 0.0686,
"step": 112500
},
{
"epoch": 7.06,
"learning_rate": 3.0951315789473685e-06,
"loss": 0.1035,
"step": 113000
},
{
"epoch": 7.09,
"learning_rate": 3.0623026315789477e-06,
"loss": 0.0822,
"step": 113500
},
{
"epoch": 7.12,
"learning_rate": 3.0294078947368422e-06,
"loss": 0.074,
"step": 114000
},
{
"epoch": 7.16,
"learning_rate": 2.996513157894737e-06,
"loss": 0.0931,
"step": 114500
},
{
"epoch": 7.19,
"learning_rate": 2.9636184210526318e-06,
"loss": 0.0849,
"step": 115000
},
{
"epoch": 7.22,
"learning_rate": 2.9307236842105263e-06,
"loss": 0.1084,
"step": 115500
},
{
"epoch": 7.25,
"learning_rate": 2.8978289473684213e-06,
"loss": 0.0754,
"step": 116000
},
{
"epoch": 7.28,
"learning_rate": 2.8649342105263163e-06,
"loss": 0.0711,
"step": 116500
},
{
"epoch": 7.31,
"learning_rate": 2.8320394736842104e-06,
"loss": 0.0965,
"step": 117000
},
{
"epoch": 7.34,
"learning_rate": 2.7992105263157896e-06,
"loss": 0.097,
"step": 117500
},
{
"epoch": 7.38,
"learning_rate": 2.766315789473684e-06,
"loss": 0.0852,
"step": 118000
},
{
"epoch": 7.41,
"learning_rate": 2.733421052631579e-06,
"loss": 0.0847,
"step": 118500
},
{
"epoch": 7.44,
"learning_rate": 2.700526315789474e-06,
"loss": 0.0935,
"step": 119000
},
{
"epoch": 7.47,
"learning_rate": 2.6676315789473687e-06,
"loss": 0.0728,
"step": 119500
},
{
"epoch": 7.5,
"learning_rate": 2.6347368421052633e-06,
"loss": 0.0912,
"step": 120000
},
{
"epoch": 7.53,
"learning_rate": 2.6018421052631583e-06,
"loss": 0.0769,
"step": 120500
},
{
"epoch": 7.56,
"learning_rate": 2.5689473684210533e-06,
"loss": 0.0874,
"step": 121000
},
{
"epoch": 7.59,
"learning_rate": 2.5360526315789474e-06,
"loss": 0.0841,
"step": 121500
},
{
"epoch": 7.62,
"learning_rate": 2.5031578947368424e-06,
"loss": 0.0785,
"step": 122000
},
{
"epoch": 7.66,
"learning_rate": 2.470328947368421e-06,
"loss": 0.0729,
"step": 122500
},
{
"epoch": 7.69,
"learning_rate": 2.437434210526316e-06,
"loss": 0.0871,
"step": 123000
},
{
"epoch": 7.72,
"learning_rate": 2.4045394736842107e-06,
"loss": 0.1076,
"step": 123500
},
{
"epoch": 7.75,
"learning_rate": 2.3716447368421057e-06,
"loss": 0.1002,
"step": 124000
},
{
"epoch": 7.78,
"learning_rate": 2.3388157894736845e-06,
"loss": 0.0812,
"step": 124500
},
{
"epoch": 7.81,
"learning_rate": 2.305921052631579e-06,
"loss": 0.087,
"step": 125000
},
{
"epoch": 7.84,
"learning_rate": 2.273026315789474e-06,
"loss": 0.0767,
"step": 125500
},
{
"epoch": 7.88,
"learning_rate": 2.2401315789473686e-06,
"loss": 0.1071,
"step": 126000
},
{
"epoch": 7.91,
"learning_rate": 2.2072368421052636e-06,
"loss": 0.1095,
"step": 126500
},
{
"epoch": 7.94,
"learning_rate": 2.174342105263158e-06,
"loss": 0.0983,
"step": 127000
},
{
"epoch": 7.97,
"learning_rate": 2.141447368421053e-06,
"loss": 0.0786,
"step": 127500
},
{
"epoch": 8.0,
"learning_rate": 2.1085526315789477e-06,
"loss": 0.0941,
"step": 128000
},
{
"epoch": 8.0,
"eval_BLEU": 79.82458025693062,
"eval_BLEU-Bigram-Precision": 83.85884065001213,
"eval_BLEU-Trigram-Precision": 78.5439031103771,
"eval_BLEU-Unigram-Precision": 90.06056673155959,
"eval_ROUGE-2": 77.16039909105272,
"eval_ROUGE-L": 90.85908984738971,
"eval_Sacre-Bigram-Precision": 84.34089917066783,
"eval_Sacre-Trigram-Precision": 80.05865102639297,
"eval_Sacre-Unigram-Precision": 89.97441951987406,
"eval_SacreBLEU": 80.93271666051959,
"eval_loss": 0.2738938629627228,
"eval_runtime": 88.7035,
"eval_samples_per_second": 11.274,
"eval_steps_per_second": 11.274,
"step": 128000
},
{
"epoch": 8.03,
"learning_rate": 2.0757236842105265e-06,
"loss": 0.085,
"step": 128500
},
{
"epoch": 8.06,
"learning_rate": 2.0428947368421053e-06,
"loss": 0.099,
"step": 129000
},
{
"epoch": 8.09,
"learning_rate": 2.0100000000000002e-06,
"loss": 0.0742,
"step": 129500
},
{
"epoch": 8.12,
"learning_rate": 1.977105263157895e-06,
"loss": 0.0918,
"step": 130000
},
{
"epoch": 8.16,
"learning_rate": 1.9442105263157894e-06,
"loss": 0.0821,
"step": 130500
},
{
"epoch": 8.19,
"learning_rate": 1.9113157894736843e-06,
"loss": 0.0724,
"step": 131000
},
{
"epoch": 8.22,
"learning_rate": 1.8784210526315791e-06,
"loss": 0.0654,
"step": 131500
},
{
"epoch": 8.25,
"learning_rate": 1.8455263157894737e-06,
"loss": 0.0906,
"step": 132000
},
{
"epoch": 8.28,
"learning_rate": 1.8126315789473687e-06,
"loss": 0.1159,
"step": 132500
},
{
"epoch": 8.31,
"learning_rate": 1.7797368421052632e-06,
"loss": 0.0841,
"step": 133000
},
{
"epoch": 8.34,
"learning_rate": 1.7468421052631582e-06,
"loss": 0.0626,
"step": 133500
},
{
"epoch": 8.38,
"learning_rate": 1.7139473684210528e-06,
"loss": 0.0721,
"step": 134000
},
{
"epoch": 8.41,
"learning_rate": 1.6810526315789473e-06,
"loss": 0.0775,
"step": 134500
},
{
"epoch": 8.44,
"learning_rate": 1.6481578947368423e-06,
"loss": 0.0756,
"step": 135000
},
{
"epoch": 8.47,
"learning_rate": 1.615328947368421e-06,
"loss": 0.0736,
"step": 135500
},
{
"epoch": 8.5,
"learning_rate": 1.5825e-06,
"loss": 0.086,
"step": 136000
},
{
"epoch": 8.53,
"learning_rate": 1.5496052631578949e-06,
"loss": 0.0776,
"step": 136500
},
{
"epoch": 8.56,
"learning_rate": 1.5167105263157896e-06,
"loss": 0.0785,
"step": 137000
},
{
"epoch": 8.59,
"learning_rate": 1.4838157894736844e-06,
"loss": 0.0761,
"step": 137500
},
{
"epoch": 8.62,
"learning_rate": 1.450921052631579e-06,
"loss": 0.0919,
"step": 138000
},
{
"epoch": 8.66,
"learning_rate": 1.418092105263158e-06,
"loss": 0.0963,
"step": 138500
},
{
"epoch": 8.69,
"learning_rate": 1.3851973684210525e-06,
"loss": 0.0682,
"step": 139000
},
{
"epoch": 8.72,
"learning_rate": 1.3523026315789475e-06,
"loss": 0.0701,
"step": 139500
},
{
"epoch": 8.75,
"learning_rate": 1.319407894736842e-06,
"loss": 0.0873,
"step": 140000
},
{
"epoch": 8.78,
"learning_rate": 1.286513157894737e-06,
"loss": 0.0778,
"step": 140500
},
{
"epoch": 8.81,
"learning_rate": 1.2536184210526316e-06,
"loss": 0.0854,
"step": 141000
},
{
"epoch": 8.84,
"learning_rate": 1.2207236842105264e-06,
"loss": 0.0727,
"step": 141500
},
{
"epoch": 8.88,
"learning_rate": 1.1878289473684212e-06,
"loss": 0.0798,
"step": 142000
},
{
"epoch": 8.91,
"learning_rate": 1.1550657894736842e-06,
"loss": 0.0629,
"step": 142500
},
{
"epoch": 8.94,
"learning_rate": 1.122171052631579e-06,
"loss": 0.0746,
"step": 143000
},
{
"epoch": 8.97,
"learning_rate": 1.0892763157894737e-06,
"loss": 0.0775,
"step": 143500
},
{
"epoch": 9.0,
"learning_rate": 1.0563815789473685e-06,
"loss": 0.0825,
"step": 144000
},
{
"epoch": 9.0,
"eval_BLEU": 79.76894087868382,
"eval_BLEU-Bigram-Precision": 82.89615522817104,
"eval_BLEU-Trigram-Precision": 77.52747998371557,
"eval_BLEU-Unigram-Precision": 89.2715798481121,
"eval_ROUGE-2": 77.13688197598398,
"eval_ROUGE-L": 90.87326567997422,
"eval_Sacre-Bigram-Precision": 83.55085751267393,
"eval_Sacre-Trigram-Precision": 79.18224580870823,
"eval_Sacre-Unigram-Precision": 89.34865154317983,
"eval_SacreBLEU": 80.98475876485695,
"eval_loss": 0.2726329565048218,
"eval_runtime": 89.7408,
"eval_samples_per_second": 11.143,
"eval_steps_per_second": 11.143,
"step": 144000
},
{
"epoch": 9.03,
"learning_rate": 1.0234868421052633e-06,
"loss": 0.0772,
"step": 144500
},
{
"epoch": 9.06,
"learning_rate": 9.90592105263158e-07,
"loss": 0.075,
"step": 145000
},
{
"epoch": 9.09,
"learning_rate": 9.577631578947368e-07,
"loss": 0.0837,
"step": 145500
},
{
"epoch": 9.12,
"learning_rate": 9.248684210526317e-07,
"loss": 0.0809,
"step": 146000
},
{
"epoch": 9.16,
"learning_rate": 8.919736842105265e-07,
"loss": 0.0698,
"step": 146500
},
{
"epoch": 9.19,
"learning_rate": 8.591447368421054e-07,
"loss": 0.0851,
"step": 147000
},
{
"epoch": 9.22,
"learning_rate": 8.262500000000001e-07,
"loss": 0.0698,
"step": 147500
},
{
"epoch": 9.25,
"learning_rate": 7.933552631578947e-07,
"loss": 0.0822,
"step": 148000
},
{
"epoch": 9.28,
"learning_rate": 7.604605263157895e-07,
"loss": 0.0786,
"step": 148500
},
{
"epoch": 9.31,
"learning_rate": 7.275657894736842e-07,
"loss": 0.078,
"step": 149000
},
{
"epoch": 9.34,
"learning_rate": 6.94671052631579e-07,
"loss": 0.0762,
"step": 149500
},
{
"epoch": 9.38,
"learning_rate": 6.617763157894737e-07,
"loss": 0.071,
"step": 150000
},
{
"epoch": 9.41,
"learning_rate": 6.288815789473685e-07,
"loss": 0.0742,
"step": 150500
},
{
"epoch": 9.44,
"learning_rate": 5.959868421052632e-07,
"loss": 0.0716,
"step": 151000
},
{
"epoch": 9.47,
"learning_rate": 5.630921052631579e-07,
"loss": 0.0602,
"step": 151500
},
{
"epoch": 9.5,
"learning_rate": 5.302631578947369e-07,
"loss": 0.0978,
"step": 152000
},
{
"epoch": 9.53,
"learning_rate": 4.973684210526316e-07,
"loss": 0.0731,
"step": 152500
},
{
"epoch": 9.56,
"learning_rate": 4.6453947368421055e-07,
"loss": 0.0887,
"step": 153000
},
{
"epoch": 9.59,
"learning_rate": 4.316447368421053e-07,
"loss": 0.0895,
"step": 153500
},
{
"epoch": 9.62,
"learning_rate": 3.9875000000000005e-07,
"loss": 0.0756,
"step": 154000
},
{
"epoch": 9.66,
"learning_rate": 3.6585526315789477e-07,
"loss": 0.0795,
"step": 154500
},
{
"epoch": 9.69,
"learning_rate": 3.329605263157895e-07,
"loss": 0.0836,
"step": 155000
},
{
"epoch": 9.72,
"learning_rate": 3.0006578947368426e-07,
"loss": 0.08,
"step": 155500
},
{
"epoch": 9.75,
"learning_rate": 2.67171052631579e-07,
"loss": 0.0676,
"step": 156000
},
{
"epoch": 9.78,
"learning_rate": 2.3434210526315792e-07,
"loss": 0.0931,
"step": 156500
},
{
"epoch": 9.81,
"learning_rate": 2.0144736842105264e-07,
"loss": 0.0717,
"step": 157000
},
{
"epoch": 9.84,
"learning_rate": 1.685526315789474e-07,
"loss": 0.078,
"step": 157500
},
{
"epoch": 9.88,
"learning_rate": 1.356578947368421e-07,
"loss": 0.0618,
"step": 158000
},
{
"epoch": 9.91,
"learning_rate": 1.0276315789473684e-07,
"loss": 0.0783,
"step": 158500
},
{
"epoch": 9.94,
"learning_rate": 6.986842105263158e-08,
"loss": 0.0823,
"step": 159000
},
{
"epoch": 9.97,
"learning_rate": 3.6973684210526316e-08,
"loss": 0.0558,
"step": 159500
},
{
"epoch": 10.0,
"learning_rate": 4.078947368421053e-09,
"loss": 0.0776,
"step": 160000
},
{
"epoch": 10.0,
"eval_BLEU": 80.25661524924499,
"eval_BLEU-Bigram-Precision": 83.93569442765623,
"eval_BLEU-Trigram-Precision": 78.80159056629645,
"eval_BLEU-Unigram-Precision": 90.10029116790682,
"eval_ROUGE-2": 77.49770887608805,
"eval_ROUGE-L": 91.11406002685167,
"eval_Sacre-Bigram-Precision": 84.69510199629104,
"eval_Sacre-Trigram-Precision": 80.59118114083303,
"eval_Sacre-Unigram-Precision": 90.23310711124225,
"eval_SacreBLEU": 81.40117848493372,
"eval_loss": 0.27461713552474976,
"eval_runtime": 88.6539,
"eval_samples_per_second": 11.28,
"eval_steps_per_second": 11.28,
"step": 160000
},
{
"epoch": 10.0,
"step": 160000,
"total_flos": 2111609241600000.0,
"train_loss": 0.13924457960128783,
"train_runtime": 13035.4538,
"train_samples_per_second": 12.274,
"train_steps_per_second": 12.274
}
],
"max_steps": 160000,
"num_train_epochs": 10,
"total_flos": 2111609241600000.0,
"trial_name": null,
"trial_params": null
}