| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 5656890, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.991162104972874e-05, | |
| "loss": 5.8101, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.982326861579419e-05, | |
| "loss": 4.147, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.973490734308074e-05, | |
| "loss": 3.446, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.964655490914619e-05, | |
| "loss": 3.0968, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9558193636432744e-05, | |
| "loss": 2.8867, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.94698500412771e-05, | |
| "loss": 2.7374, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.938149760734255e-05, | |
| "loss": 2.6213, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.92931363346291e-05, | |
| "loss": 2.5359, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.920478390069455e-05, | |
| "loss": 2.4665, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.91164226279811e-05, | |
| "loss": 2.4067, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.902807019404656e-05, | |
| "loss": 2.3518, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.893971776011201e-05, | |
| "loss": 2.311, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.885134764861965e-05, | |
| "loss": 2.2728, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.87629952146851e-05, | |
| "loss": 2.2382, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.867464278075056e-05, | |
| "loss": 2.2096, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.858629034681601e-05, | |
| "loss": 2.178, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.849793791288146e-05, | |
| "loss": 2.1548, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.840958547894692e-05, | |
| "loss": 2.1315, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8321233045012365e-05, | |
| "loss": 2.1091, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.823288061107782e-05, | |
| "loss": 2.0882, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.814452817714327e-05, | |
| "loss": 2.0706, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.805618458198763e-05, | |
| "loss": 2.0587, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.7967832148053085e-05, | |
| "loss": 2.0406, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.787947971411853e-05, | |
| "loss": 2.0289, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.779112728018399e-05, | |
| "loss": 2.013, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.770276600747054e-05, | |
| "loss": 1.9989, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.7614413573535995e-05, | |
| "loss": 1.9863, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.752606997838035e-05, | |
| "loss": 1.9783, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.7437717544445804e-05, | |
| "loss": 1.9657, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.734935627173235e-05, | |
| "loss": 1.9566, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.72610038377978e-05, | |
| "loss": 1.9453, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7172660242642156e-05, | |
| "loss": 1.9367, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.708430780870761e-05, | |
| "loss": 1.9306, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.699595537477307e-05, | |
| "loss": 1.9191, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.690761177961742e-05, | |
| "loss": 1.9128, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.6819259345682876e-05, | |
| "loss": 1.9027, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.673090691174833e-05, | |
| "loss": 1.8961, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.664254563903488e-05, | |
| "loss": 1.8879, | |
| "step": 380000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.6554202043879235e-05, | |
| "loss": 1.8775, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.646584960994469e-05, | |
| "loss": 1.8793, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.6377506014789044e-05, | |
| "loss": 1.8719, | |
| "step": 410000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.6289171258412315e-05, | |
| "loss": 1.8633, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.6200818824477763e-05, | |
| "loss": 1.8573, | |
| "step": 430000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.611245755176431e-05, | |
| "loss": 1.8464, | |
| "step": 440000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.602410511782976e-05, | |
| "loss": 1.8467, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.593575268389522e-05, | |
| "loss": 1.8403, | |
| "step": 460000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.584740024996067e-05, | |
| "loss": 1.836, | |
| "step": 470000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.575906549358393e-05, | |
| "loss": 1.8332, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.567071305964939e-05, | |
| "loss": 1.8271, | |
| "step": 490000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.5582360625714835e-05, | |
| "loss": 1.8172, | |
| "step": 500000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.5494017030559195e-05, | |
| "loss": 1.8169, | |
| "step": 510000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.540566459662465e-05, | |
| "loss": 1.8168, | |
| "step": 520000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.53173121626901e-05, | |
| "loss": 1.8088, | |
| "step": 530000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.522896856753446e-05, | |
| "loss": 1.801, | |
| "step": 540000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.514062497237882e-05, | |
| "loss": 1.8026, | |
| "step": 550000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5052272538444275e-05, | |
| "loss": 1.7986, | |
| "step": 560000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 28.7509, | |
| "eval_gen_len": 66.1132, | |
| "eval_loss": 1.8565547466278076, | |
| "eval_runtime": 2948.752, | |
| "eval_samples_per_second": 5.821, | |
| "eval_steps_per_second": 0.364, | |
| "step": 565689 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.496392894328863e-05, | |
| "loss": 1.7781, | |
| "step": 570000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.487557650935408e-05, | |
| "loss": 1.7622, | |
| "step": 580000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.478722407541954e-05, | |
| "loss": 1.7623, | |
| "step": 590000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.469887164148499e-05, | |
| "loss": 1.7613, | |
| "step": 600000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.461052804632935e-05, | |
| "loss": 1.7595, | |
| "step": 610000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.45221756123948e-05, | |
| "loss": 1.7563, | |
| "step": 620000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.443382317846025e-05, | |
| "loss": 1.7571, | |
| "step": 630000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.434547958330461e-05, | |
| "loss": 1.7521, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.4257127149370066e-05, | |
| "loss": 1.7513, | |
| "step": 650000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.4168774715435515e-05, | |
| "loss": 1.7471, | |
| "step": 660000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.4080422281500964e-05, | |
| "loss": 1.7532, | |
| "step": 670000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.399207868634533e-05, | |
| "loss": 1.7469, | |
| "step": 680000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.390372625241078e-05, | |
| "loss": 1.7447, | |
| "step": 690000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.381537381847623e-05, | |
| "loss": 1.7385, | |
| "step": 700000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.372702138454168e-05, | |
| "loss": 1.7426, | |
| "step": 710000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.363867778938604e-05, | |
| "loss": 1.7381, | |
| "step": 720000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.355032535545149e-05, | |
| "loss": 1.7338, | |
| "step": 730000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.346197292151695e-05, | |
| "loss": 1.7307, | |
| "step": 740000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.33736204875824e-05, | |
| "loss": 1.7319, | |
| "step": 750000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.328525921486895e-05, | |
| "loss": 1.7279, | |
| "step": 760000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.3196915619713307e-05, | |
| "loss": 1.729, | |
| "step": 770000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.310856318577876e-05, | |
| "loss": 1.7254, | |
| "step": 780000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.3020201913065306e-05, | |
| "loss": 1.7207, | |
| "step": 790000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.293184947913076e-05, | |
| "loss": 1.7183, | |
| "step": 800000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.284350588397512e-05, | |
| "loss": 1.7184, | |
| "step": 810000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.275515345004057e-05, | |
| "loss": 1.7139, | |
| "step": 820000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.266679217732712e-05, | |
| "loss": 1.7155, | |
| "step": 830000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.257843090461367e-05, | |
| "loss": 1.7122, | |
| "step": 840000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.2490087309458025e-05, | |
| "loss": 1.7111, | |
| "step": 850000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.2401717197965665e-05, | |
| "loss": 1.7106, | |
| "step": 860000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.2313373602810025e-05, | |
| "loss": 1.7071, | |
| "step": 870000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.222502116887548e-05, | |
| "loss": 1.7081, | |
| "step": 880000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.213666873494093e-05, | |
| "loss": 1.7053, | |
| "step": 890000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.2048316301006384e-05, | |
| "loss": 1.7023, | |
| "step": 900000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.195996386707184e-05, | |
| "loss": 1.7017, | |
| "step": 910000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.1871611433137295e-05, | |
| "loss": 1.6979, | |
| "step": 920000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.178325016042384e-05, | |
| "loss": 1.6953, | |
| "step": 930000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.169489772648929e-05, | |
| "loss": 1.693, | |
| "step": 940000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.160653645377584e-05, | |
| "loss": 1.6934, | |
| "step": 950000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.1518184019841294e-05, | |
| "loss": 1.6899, | |
| "step": 960000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.142983158590675e-05, | |
| "loss": 1.6913, | |
| "step": 970000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.1341470313193294e-05, | |
| "loss": 1.691, | |
| "step": 980000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.125311787925875e-05, | |
| "loss": 1.6888, | |
| "step": 990000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.11647742841031e-05, | |
| "loss": 1.6888, | |
| "step": 1000000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.107641301138965e-05, | |
| "loss": 1.6857, | |
| "step": 1010000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.098806057745511e-05, | |
| "loss": 1.6854, | |
| "step": 1020000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.089970814352056e-05, | |
| "loss": 1.679, | |
| "step": 1030000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.08113468708071e-05, | |
| "loss": 1.68, | |
| "step": 1040000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.072299443687256e-05, | |
| "loss": 1.681, | |
| "step": 1050000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.063464200293801e-05, | |
| "loss": 1.6783, | |
| "step": 1060000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.0546298407782365e-05, | |
| "loss": 1.6766, | |
| "step": 1070000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.0457937135068916e-05, | |
| "loss": 1.6762, | |
| "step": 1080000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.036957586235546e-05, | |
| "loss": 1.6753, | |
| "step": 1090000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.0281223428420916e-05, | |
| "loss": 1.671, | |
| "step": 1100000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.019287099448637e-05, | |
| "loss": 1.668, | |
| "step": 1110000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.0104527399330724e-05, | |
| "loss": 1.6676, | |
| "step": 1120000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.0016166126617275e-05, | |
| "loss": 1.6695, | |
| "step": 1130000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 29.525, | |
| "eval_gen_len": 66.2651, | |
| "eval_loss": 1.7653018236160278, | |
| "eval_runtime": 3037.2133, | |
| "eval_samples_per_second": 5.652, | |
| "eval_steps_per_second": 0.353, | |
| "step": 1131378 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.992781369268273e-05, | |
| "loss": 1.6357, | |
| "step": 1140000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.9839461258748186e-05, | |
| "loss": 1.6309, | |
| "step": 1150000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.975109998603473e-05, | |
| "loss": 1.6355, | |
| "step": 1160000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.966275639087909e-05, | |
| "loss": 1.6321, | |
| "step": 1170000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.9574395118165634e-05, | |
| "loss": 1.6344, | |
| "step": 1180000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.948604268423109e-05, | |
| "loss": 1.6312, | |
| "step": 1190000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.939768141151764e-05, | |
| "loss": 1.6324, | |
| "step": 1200000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.9309328977583096e-05, | |
| "loss": 1.6357, | |
| "step": 1210000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.922097654364854e-05, | |
| "loss": 1.6325, | |
| "step": 1220000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.913262410971399e-05, | |
| "loss": 1.6327, | |
| "step": 1230000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.9044262837000544e-05, | |
| "loss": 1.632, | |
| "step": 1240000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.8955910403066e-05, | |
| "loss": 1.6315, | |
| "step": 1250000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.8867557969131455e-05, | |
| "loss": 1.6322, | |
| "step": 1260000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.877921437397581e-05, | |
| "loss": 1.6314, | |
| "step": 1270000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.8690861940041264e-05, | |
| "loss": 1.6317, | |
| "step": 1280000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.860250066732781e-05, | |
| "loss": 1.6275, | |
| "step": 1290000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.851415707217217e-05, | |
| "loss": 1.6296, | |
| "step": 1300000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.842579579945871e-05, | |
| "loss": 1.6305, | |
| "step": 1310000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.833743452674526e-05, | |
| "loss": 1.6266, | |
| "step": 1320000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.824908209281072e-05, | |
| "loss": 1.6273, | |
| "step": 1330000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.8160729658876174e-05, | |
| "loss": 1.6283, | |
| "step": 1340000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 3.807237722494162e-05, | |
| "loss": 1.629, | |
| "step": 1350000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.798403362978598e-05, | |
| "loss": 1.6269, | |
| "step": 1360000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.7895672357072526e-05, | |
| "loss": 1.6252, | |
| "step": 1370000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.7807328761916886e-05, | |
| "loss": 1.6221, | |
| "step": 1380000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.7718976327982335e-05, | |
| "loss": 1.6229, | |
| "step": 1390000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.7630615055268886e-05, | |
| "loss": 1.6226, | |
| "step": 1400000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.754226262133434e-05, | |
| "loss": 1.6224, | |
| "step": 1410000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.74539101873998e-05, | |
| "loss": 1.6207, | |
| "step": 1420000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.7365557753465245e-05, | |
| "loss": 1.6204, | |
| "step": 1430000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 3.72772053195307e-05, | |
| "loss": 1.6183, | |
| "step": 1440000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.7188861724375054e-05, | |
| "loss": 1.6235, | |
| "step": 1450000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.710050929044051e-05, | |
| "loss": 1.6207, | |
| "step": 1460000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.7012156856505965e-05, | |
| "loss": 1.618, | |
| "step": 1470000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.692381326135032e-05, | |
| "loss": 1.6162, | |
| "step": 1480000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.683546082741577e-05, | |
| "loss": 1.6194, | |
| "step": 1490000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.674711723226013e-05, | |
| "loss": 1.6154, | |
| "step": 1500000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.6658755959546684e-05, | |
| "loss": 1.6156, | |
| "step": 1510000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.6570403525612126e-05, | |
| "loss": 1.6144, | |
| "step": 1520000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 3.648205993045649e-05, | |
| "loss": 1.6125, | |
| "step": 1530000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.639369865774304e-05, | |
| "loss": 1.6134, | |
| "step": 1540000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.630534622380849e-05, | |
| "loss": 1.6102, | |
| "step": 1550000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.6217002628652845e-05, | |
| "loss": 1.6065, | |
| "step": 1560000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.6128641355939396e-05, | |
| "loss": 1.6073, | |
| "step": 1570000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.604028008322594e-05, | |
| "loss": 1.6071, | |
| "step": 1580000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.59519364880703e-05, | |
| "loss": 1.6046, | |
| "step": 1590000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.586359289291466e-05, | |
| "loss": 1.607, | |
| "step": 1600000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 3.5775231620201204e-05, | |
| "loss": 1.6069, | |
| "step": 1610000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.5686888025045564e-05, | |
| "loss": 1.6058, | |
| "step": 1620000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.5598526752332115e-05, | |
| "loss": 1.6031, | |
| "step": 1630000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.551018315717647e-05, | |
| "loss": 1.6047, | |
| "step": 1640000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.542182188446302e-05, | |
| "loss": 1.6042, | |
| "step": 1650000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.5333469450528475e-05, | |
| "loss": 1.6025, | |
| "step": 1660000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.524511701659392e-05, | |
| "loss": 1.6036, | |
| "step": 1670000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.515677342143828e-05, | |
| "loss": 1.6011, | |
| "step": 1680000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.506842098750374e-05, | |
| "loss": 1.6038, | |
| "step": 1690000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 29.8841, | |
| "eval_gen_len": 66.1849, | |
| "eval_loss": 1.7081401348114014, | |
| "eval_runtime": 2996.6763, | |
| "eval_samples_per_second": 5.728, | |
| "eval_steps_per_second": 0.358, | |
| "step": 1697067 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.498005971479029e-05, | |
| "loss": 1.5881, | |
| "step": 1700000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.489170728085573e-05, | |
| "loss": 1.5596, | |
| "step": 1710000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.48033636857001e-05, | |
| "loss": 1.5643, | |
| "step": 1720000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.471502009054445e-05, | |
| "loss": 1.563, | |
| "step": 1730000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.462667649538881e-05, | |
| "loss": 1.5633, | |
| "step": 1740000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.4538324061454266e-05, | |
| "loss": 1.5678, | |
| "step": 1750000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.444996278874081e-05, | |
| "loss": 1.5677, | |
| "step": 1760000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.436161919358517e-05, | |
| "loss": 1.5664, | |
| "step": 1770000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.427327559842953e-05, | |
| "loss": 1.57, | |
| "step": 1780000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.418492316449498e-05, | |
| "loss": 1.5695, | |
| "step": 1790000 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.4096570730560434e-05, | |
| "loss": 1.5693, | |
| "step": 1800000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.4008227135404794e-05, | |
| "loss": 1.5669, | |
| "step": 1810000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.391987470147024e-05, | |
| "loss": 1.5677, | |
| "step": 1820000 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.38315222675357e-05, | |
| "loss": 1.5689, | |
| "step": 1830000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.3743169833601154e-05, | |
| "loss": 1.5711, | |
| "step": 1840000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.36548173996666e-05, | |
| "loss": 1.5679, | |
| "step": 1850000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.3566456126953147e-05, | |
| "loss": 1.5705, | |
| "step": 1860000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.3478112531797506e-05, | |
| "loss": 1.5689, | |
| "step": 1870000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.338976009786296e-05, | |
| "loss": 1.5657, | |
| "step": 1880000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.330139882514951e-05, | |
| "loss": 1.5658, | |
| "step": 1890000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.3213055229993866e-05, | |
| "loss": 1.5675, | |
| "step": 1900000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.312469395728042e-05, | |
| "loss": 1.5664, | |
| "step": 1910000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.303635036212478e-05, | |
| "loss": 1.5668, | |
| "step": 1920000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.2947997928190225e-05, | |
| "loss": 1.5653, | |
| "step": 1930000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.285963665547677e-05, | |
| "loss": 1.5677, | |
| "step": 1940000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.2771293060321136e-05, | |
| "loss": 1.5633, | |
| "step": 1950000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.2682940626386585e-05, | |
| "loss": 1.5638, | |
| "step": 1960000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.2594579353673136e-05, | |
| "loss": 1.5635, | |
| "step": 1970000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.2506226919738584e-05, | |
| "loss": 1.566, | |
| "step": 1980000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.241787448580404e-05, | |
| "loss": 1.5641, | |
| "step": 1990000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.232950437431168e-05, | |
| "loss": 1.5634, | |
| "step": 2000000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.224116077915604e-05, | |
| "loss": 1.5658, | |
| "step": 2010000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.215280834522149e-05, | |
| "loss": 1.5644, | |
| "step": 2020000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.2064455911286943e-05, | |
| "loss": 1.5625, | |
| "step": 2030000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.19761034773524e-05, | |
| "loss": 1.562, | |
| "step": 2040000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.1887751043417854e-05, | |
| "loss": 1.5634, | |
| "step": 2050000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.179940744826221e-05, | |
| "loss": 1.5595, | |
| "step": 2060000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.171104617554876e-05, | |
| "loss": 1.5594, | |
| "step": 2070000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.1622693741614214e-05, | |
| "loss": 1.5609, | |
| "step": 2080000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.153434130767966e-05, | |
| "loss": 1.5606, | |
| "step": 2090000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.144598003496621e-05, | |
| "loss": 1.562, | |
| "step": 2100000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.135762760103166e-05, | |
| "loss": 1.5601, | |
| "step": 2110000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.126927516709712e-05, | |
| "loss": 1.5574, | |
| "step": 2120000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.118092273316257e-05, | |
| "loss": 1.557, | |
| "step": 2130000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.109257029922802e-05, | |
| "loss": 1.558, | |
| "step": 2140000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.100420902651457e-05, | |
| "loss": 1.5589, | |
| "step": 2150000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.0915865431358926e-05, | |
| "loss": 1.5563, | |
| "step": 2160000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.082751299742438e-05, | |
| "loss": 1.557, | |
| "step": 2170000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.073915172471093e-05, | |
| "loss": 1.5517, | |
| "step": 2180000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.0650808129555285e-05, | |
| "loss": 1.5569, | |
| "step": 2190000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.0562438018062925e-05, | |
| "loss": 1.5561, | |
| "step": 2200000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.047409442290729e-05, | |
| "loss": 1.5536, | |
| "step": 2210000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.0385733150193836e-05, | |
| "loss": 1.5567, | |
| "step": 2220000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.0297371877480386e-05, | |
| "loss": 1.5514, | |
| "step": 2230000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.0209028282324743e-05, | |
| "loss": 1.5542, | |
| "step": 2240000 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.0120675848390195e-05, | |
| "loss": 1.5515, | |
| "step": 2250000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.0032323414455647e-05, | |
| "loss": 1.5515, | |
| "step": 2260000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 30.588, | |
| "eval_gen_len": 65.9093, | |
| "eval_loss": 1.6601390838623047, | |
| "eval_runtime": 3100.6175, | |
| "eval_samples_per_second": 5.536, | |
| "eval_steps_per_second": 0.346, | |
| "step": 2262756 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.9943970980521102e-05, | |
| "loss": 1.5183, | |
| "step": 2270000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 2.9855609707807647e-05, | |
| "loss": 1.5123, | |
| "step": 2280000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 2.9767257273873102e-05, | |
| "loss": 1.5146, | |
| "step": 2290000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 2.9678904839938554e-05, | |
| "loss": 1.5144, | |
| "step": 2300000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 2.9590552406004006e-05, | |
| "loss": 1.5173, | |
| "step": 2310000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 2.9502191133290557e-05, | |
| "loss": 1.5188, | |
| "step": 2320000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.9413838699356006e-05, | |
| "loss": 1.5195, | |
| "step": 2330000 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.932549510420037e-05, | |
| "loss": 1.5181, | |
| "step": 2340000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 2.923714267026582e-05, | |
| "loss": 1.5199, | |
| "step": 2350000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 2.9148799075110177e-05, | |
| "loss": 1.5206, | |
| "step": 2360000 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 2.9060446641175633e-05, | |
| "loss": 1.5207, | |
| "step": 2370000 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 2.8972094207241085e-05, | |
| "loss": 1.5209, | |
| "step": 2380000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 2.8883741773306534e-05, | |
| "loss": 1.5192, | |
| "step": 2390000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 2.8795380500593084e-05, | |
| "loss": 1.5231, | |
| "step": 2400000 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 2.870703690543744e-05, | |
| "loss": 1.5201, | |
| "step": 2410000 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 2.8618684471502893e-05, | |
| "loss": 1.5204, | |
| "step": 2420000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 2.853033203756835e-05, | |
| "loss": 1.5224, | |
| "step": 2430000 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 2.84419796036338e-05, | |
| "loss": 1.5221, | |
| "step": 2440000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 2.8353627169699252e-05, | |
| "loss": 1.52, | |
| "step": 2450000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.8265274735764708e-05, | |
| "loss": 1.5212, | |
| "step": 2460000 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 2.8176931140609064e-05, | |
| "loss": 1.5241, | |
| "step": 2470000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 2.8088578706674516e-05, | |
| "loss": 1.5224, | |
| "step": 2480000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 2.8000235111518873e-05, | |
| "loss": 1.5185, | |
| "step": 2490000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.791188267758433e-05, | |
| "loss": 1.5184, | |
| "step": 2500000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.7823539082428685e-05, | |
| "loss": 1.5199, | |
| "step": 2510000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 2.7735186648494137e-05, | |
| "loss": 1.5198, | |
| "step": 2520000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 2.7646834214559592e-05, | |
| "loss": 1.5195, | |
| "step": 2530000 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.7558481780625044e-05, | |
| "loss": 1.522, | |
| "step": 2540000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.74701293466905e-05, | |
| "loss": 1.5223, | |
| "step": 2550000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 2.7381776912755952e-05, | |
| "loss": 1.5183, | |
| "step": 2560000 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 2.7293433317600308e-05, | |
| "loss": 1.5225, | |
| "step": 2570000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.7205080883665764e-05, | |
| "loss": 1.5198, | |
| "step": 2580000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.7116728449731216e-05, | |
| "loss": 1.5201, | |
| "step": 2590000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.7028384854575572e-05, | |
| "loss": 1.5195, | |
| "step": 2600000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.6940032420641028e-05, | |
| "loss": 1.52, | |
| "step": 2610000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.6851688825485384e-05, | |
| "loss": 1.5202, | |
| "step": 2620000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.6763336391550836e-05, | |
| "loss": 1.5192, | |
| "step": 2630000 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.6674983957616288e-05, | |
| "loss": 1.5176, | |
| "step": 2640000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.6586631523681744e-05, | |
| "loss": 1.5153, | |
| "step": 2650000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.64982879285261e-05, | |
| "loss": 1.5156, | |
| "step": 2660000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.640992665581265e-05, | |
| "loss": 1.5191, | |
| "step": 2670000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.6321583060657008e-05, | |
| "loss": 1.5147, | |
| "step": 2680000 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.6233239465501364e-05, | |
| "loss": 1.5169, | |
| "step": 2690000 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.614489587034572e-05, | |
| "loss": 1.5174, | |
| "step": 2700000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.605653459763227e-05, | |
| "loss": 1.5164, | |
| "step": 2710000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.5968182163697724e-05, | |
| "loss": 1.5134, | |
| "step": 2720000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.587983856854208e-05, | |
| "loss": 1.5166, | |
| "step": 2730000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 2.5791486134607535e-05, | |
| "loss": 1.5162, | |
| "step": 2740000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 2.5703142539451892e-05, | |
| "loss": 1.5122, | |
| "step": 2750000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 2.5614781266738443e-05, | |
| "loss": 1.5134, | |
| "step": 2760000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 2.55264376715828e-05, | |
| "loss": 1.5135, | |
| "step": 2770000 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 2.543808523764825e-05, | |
| "loss": 1.511, | |
| "step": 2780000 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 2.5349741642492608e-05, | |
| "loss": 1.5089, | |
| "step": 2790000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 2.5261389208558063e-05, | |
| "loss": 1.5149, | |
| "step": 2800000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 2.517304561340242e-05, | |
| "loss": 1.5121, | |
| "step": 2810000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 2.5084693179467872e-05, | |
| "loss": 1.5115, | |
| "step": 2820000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bleu": 30.9726, | |
| "eval_gen_len": 66.2171, | |
| "eval_loss": 1.6359007358551025, | |
| "eval_runtime": 3064.0636, | |
| "eval_samples_per_second": 5.602, | |
| "eval_steps_per_second": 0.35, | |
| "step": 2828445 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.4996340745533324e-05, | |
| "loss": 1.5045, | |
| "step": 2830000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.4907988311598776e-05, | |
| "loss": 1.4705, | |
| "step": 2840000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 2.4819635877664228e-05, | |
| "loss": 1.4722, | |
| "step": 2850000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.4731292282508588e-05, | |
| "loss": 1.4756, | |
| "step": 2860000 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.464293984857404e-05, | |
| "loss": 1.473, | |
| "step": 2870000 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.45545962534184e-05, | |
| "loss": 1.4754, | |
| "step": 2880000 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.4466252658262756e-05, | |
| "loss": 1.4775, | |
| "step": 2890000 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.4377909063107116e-05, | |
| "loss": 1.4779, | |
| "step": 2900000 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 2.4289556629172568e-05, | |
| "loss": 1.4802, | |
| "step": 2910000 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 2.420120419523802e-05, | |
| "loss": 1.479, | |
| "step": 2920000 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 2.4112851761303472e-05, | |
| "loss": 1.4802, | |
| "step": 2930000 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.4024499327368928e-05, | |
| "loss": 1.4796, | |
| "step": 2940000 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 2.3936155732213284e-05, | |
| "loss": 1.4798, | |
| "step": 2950000 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 2.3847812137057644e-05, | |
| "loss": 1.4808, | |
| "step": 2960000 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 2.3759459703123096e-05, | |
| "loss": 1.4811, | |
| "step": 2970000 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 2.3671107269188548e-05, | |
| "loss": 1.4795, | |
| "step": 2980000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.3582763674032908e-05, | |
| "loss": 1.4812, | |
| "step": 2990000 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 2.349441124009836e-05, | |
| "loss": 1.4828, | |
| "step": 3000000 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 2.3406067644942717e-05, | |
| "loss": 1.4809, | |
| "step": 3010000 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 2.3317715211008172e-05, | |
| "loss": 1.4843, | |
| "step": 3020000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 2.3229362777073624e-05, | |
| "loss": 1.4821, | |
| "step": 3030000 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 2.3141010343139076e-05, | |
| "loss": 1.4808, | |
| "step": 3040000 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 2.3052666747983432e-05, | |
| "loss": 1.4815, | |
| "step": 3050000 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 2.2964323152827792e-05, | |
| "loss": 1.4796, | |
| "step": 3060000 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 2.2875970718893244e-05, | |
| "loss": 1.4823, | |
| "step": 3070000 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 2.2787618284958696e-05, | |
| "loss": 1.4802, | |
| "step": 3080000 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 2.2699265851024152e-05, | |
| "loss": 1.4814, | |
| "step": 3090000 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 2.26109134170896e-05, | |
| "loss": 1.482, | |
| "step": 3100000 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 2.2522560983155056e-05, | |
| "loss": 1.4808, | |
| "step": 3110000 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 2.2434208549220508e-05, | |
| "loss": 1.4823, | |
| "step": 3120000 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 2.2345856115285963e-05, | |
| "loss": 1.4807, | |
| "step": 3130000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 2.225751252013032e-05, | |
| "loss": 1.4821, | |
| "step": 3140000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 2.216916892497468e-05, | |
| "loss": 1.4789, | |
| "step": 3150000 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 2.2080825329819036e-05, | |
| "loss": 1.4809, | |
| "step": 3160000 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 2.199245521832668e-05, | |
| "loss": 1.478, | |
| "step": 3170000 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 2.1904111623171036e-05, | |
| "loss": 1.4817, | |
| "step": 3180000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 2.1815768028015396e-05, | |
| "loss": 1.4811, | |
| "step": 3190000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 2.1727415594080848e-05, | |
| "loss": 1.4791, | |
| "step": 3200000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 2.16390631601463e-05, | |
| "loss": 1.4789, | |
| "step": 3210000 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 2.155071956499066e-05, | |
| "loss": 1.4778, | |
| "step": 3220000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 2.1462358292277207e-05, | |
| "loss": 1.4785, | |
| "step": 3230000 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 2.137400585834266e-05, | |
| "loss": 1.4769, | |
| "step": 3240000 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 2.128565342440811e-05, | |
| "loss": 1.4783, | |
| "step": 3250000 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 2.119730982925247e-05, | |
| "loss": 1.4793, | |
| "step": 3260000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 2.1108966234096828e-05, | |
| "loss": 1.48, | |
| "step": 3270000 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 2.102061380016228e-05, | |
| "loss": 1.4794, | |
| "step": 3280000 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.0932261366227735e-05, | |
| "loss": 1.4788, | |
| "step": 3290000 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 2.084391777107209e-05, | |
| "loss": 1.4791, | |
| "step": 3300000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 2.075557417591645e-05, | |
| "loss": 1.4779, | |
| "step": 3310000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 2.06672217419819e-05, | |
| "loss": 1.4751, | |
| "step": 3320000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 2.0578869308047356e-05, | |
| "loss": 1.4763, | |
| "step": 3330000 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 2.0490508035333903e-05, | |
| "loss": 1.4751, | |
| "step": 3340000 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 2.0402164440178263e-05, | |
| "loss": 1.4753, | |
| "step": 3350000 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 2.0313812006243715e-05, | |
| "loss": 1.4737, | |
| "step": 3360000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 2.022546841108807e-05, | |
| "loss": 1.4755, | |
| "step": 3370000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 2.013710713837462e-05, | |
| "loss": 1.4756, | |
| "step": 3380000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 2.004876354321898e-05, | |
| "loss": 1.474, | |
| "step": 3390000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bleu": 31.3244, | |
| "eval_gen_len": 66.1843, | |
| "eval_loss": 1.6097419261932373, | |
| "eval_runtime": 3209.3116, | |
| "eval_samples_per_second": 5.348, | |
| "eval_steps_per_second": 0.334, | |
| "step": 3394134 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 1.9960402270505527e-05, | |
| "loss": 1.4509, | |
| "step": 3400000 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.9872058675349883e-05, | |
| "loss": 1.4364, | |
| "step": 3410000 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.978370624141534e-05, | |
| "loss": 1.4356, | |
| "step": 3420000 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.969535380748079e-05, | |
| "loss": 1.4374, | |
| "step": 3430000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.9606992534767338e-05, | |
| "loss": 1.4387, | |
| "step": 3440000 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.951864010083279e-05, | |
| "loss": 1.4395, | |
| "step": 3450000 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.9430278828119338e-05, | |
| "loss": 1.4417, | |
| "step": 3460000 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 1.9341935232963694e-05, | |
| "loss": 1.4411, | |
| "step": 3470000 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 1.9253573960250245e-05, | |
| "loss": 1.4406, | |
| "step": 3480000 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 1.9165221526315697e-05, | |
| "loss": 1.4435, | |
| "step": 3490000 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 1.907686909238115e-05, | |
| "loss": 1.4451, | |
| "step": 3500000 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 1.8988516658446605e-05, | |
| "loss": 1.4467, | |
| "step": 3510000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 1.8900164224512057e-05, | |
| "loss": 1.443, | |
| "step": 3520000 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 1.8811811790577512e-05, | |
| "loss": 1.4465, | |
| "step": 3530000 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 1.872345051786406e-05, | |
| "loss": 1.4401, | |
| "step": 3540000 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 1.8635098083929508e-05, | |
| "loss": 1.4423, | |
| "step": 3550000 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 1.8546745649994964e-05, | |
| "loss": 1.4443, | |
| "step": 3560000 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 1.8458393216060416e-05, | |
| "loss": 1.4441, | |
| "step": 3570000 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 1.8370040782125868e-05, | |
| "loss": 1.4448, | |
| "step": 3580000 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 1.8281688348191323e-05, | |
| "loss": 1.4438, | |
| "step": 3590000 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 1.819333591425677e-05, | |
| "loss": 1.4479, | |
| "step": 3600000 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 1.8104983480322227e-05, | |
| "loss": 1.4444, | |
| "step": 3610000 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 1.8016622207608775e-05, | |
| "loss": 1.4464, | |
| "step": 3620000 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 1.7928278612453135e-05, | |
| "loss": 1.4453, | |
| "step": 3630000 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 1.7839917339739682e-05, | |
| "loss": 1.4437, | |
| "step": 3640000 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 1.775157374458404e-05, | |
| "loss": 1.4461, | |
| "step": 3650000 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 1.766322131064949e-05, | |
| "loss": 1.4461, | |
| "step": 3660000 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 1.7574868876714946e-05, | |
| "loss": 1.4452, | |
| "step": 3670000 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 1.7486516442780398e-05, | |
| "loss": 1.4465, | |
| "step": 3680000 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 1.7398164008845853e-05, | |
| "loss": 1.4484, | |
| "step": 3690000 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 1.7309793897353493e-05, | |
| "loss": 1.4447, | |
| "step": 3700000 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 1.7221450302197853e-05, | |
| "loss": 1.4449, | |
| "step": 3710000 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 1.71330890294844e-05, | |
| "loss": 1.4437, | |
| "step": 3720000 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 1.7044736595549853e-05, | |
| "loss": 1.4435, | |
| "step": 3730000 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 1.695639300039421e-05, | |
| "loss": 1.4453, | |
| "step": 3740000 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 1.6868031727680757e-05, | |
| "loss": 1.4469, | |
| "step": 3750000 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 1.6779688132525113e-05, | |
| "loss": 1.4446, | |
| "step": 3760000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 1.6691326859811664e-05, | |
| "loss": 1.4432, | |
| "step": 3770000 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 1.6602974425877116e-05, | |
| "loss": 1.4411, | |
| "step": 3780000 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.6514621991942568e-05, | |
| "loss": 1.4424, | |
| "step": 3790000 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 1.6426260719229116e-05, | |
| "loss": 1.4423, | |
| "step": 3800000 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 1.633790828529457e-05, | |
| "loss": 1.4437, | |
| "step": 3810000 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 1.6249555851360023e-05, | |
| "loss": 1.4468, | |
| "step": 3820000 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 1.616119457864657e-05, | |
| "loss": 1.4434, | |
| "step": 3830000 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 1.607285098349093e-05, | |
| "loss": 1.4394, | |
| "step": 3840000 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.5984498549556383e-05, | |
| "loss": 1.4409, | |
| "step": 3850000 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 1.5896146115621835e-05, | |
| "loss": 1.4404, | |
| "step": 3860000 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 1.580779368168729e-05, | |
| "loss": 1.4402, | |
| "step": 3870000 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 1.5719441247752742e-05, | |
| "loss": 1.4409, | |
| "step": 3880000 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 1.56310976525971e-05, | |
| "loss": 1.4401, | |
| "step": 3890000 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 1.5542736379883646e-05, | |
| "loss": 1.4392, | |
| "step": 3900000 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 1.5454392784728006e-05, | |
| "loss": 1.4408, | |
| "step": 3910000 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 1.5366031512014554e-05, | |
| "loss": 1.4408, | |
| "step": 3920000 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 1.5277679078080006e-05, | |
| "loss": 1.4398, | |
| "step": 3930000 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 1.5189326644145458e-05, | |
| "loss": 1.4415, | |
| "step": 3940000 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 1.5100974210210911e-05, | |
| "loss": 1.4425, | |
| "step": 3950000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bleu": 31.557, | |
| "eval_gen_len": 66.1481, | |
| "eval_loss": 1.5914360284805298, | |
| "eval_runtime": 3218.9621, | |
| "eval_samples_per_second": 5.332, | |
| "eval_steps_per_second": 0.333, | |
| "step": 3959823 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.5012621776276365e-05, | |
| "loss": 1.4396, | |
| "step": 3960000 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.4924269342341817e-05, | |
| "loss": 1.4025, | |
| "step": 3970000 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.4835916908407271e-05, | |
| "loss": 1.4013, | |
| "step": 3980000 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.4747564474472723e-05, | |
| "loss": 1.4035, | |
| "step": 3990000 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 1.4659212040538175e-05, | |
| "loss": 1.4054, | |
| "step": 4000000 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 1.4570850767824722e-05, | |
| "loss": 1.4067, | |
| "step": 4010000 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 1.4482498333890176e-05, | |
| "loss": 1.4044, | |
| "step": 4020000 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 1.439414589995563e-05, | |
| "loss": 1.4081, | |
| "step": 4030000 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 1.4305802304799986e-05, | |
| "loss": 1.4049, | |
| "step": 4040000 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 1.4217441032086537e-05, | |
| "loss": 1.4072, | |
| "step": 4050000 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 1.4129079759373085e-05, | |
| "loss": 1.4085, | |
| "step": 4060000 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 1.4040727325438539e-05, | |
| "loss": 1.4078, | |
| "step": 4070000 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 1.3952383730282895e-05, | |
| "loss": 1.4066, | |
| "step": 4080000 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 1.3864031296348349e-05, | |
| "loss": 1.4105, | |
| "step": 4090000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 1.3775678862413801e-05, | |
| "loss": 1.4116, | |
| "step": 4100000 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 1.3687326428479253e-05, | |
| "loss": 1.4095, | |
| "step": 4110000 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 1.3598973994544707e-05, | |
| "loss": 1.4104, | |
| "step": 4120000 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 1.3510621560610159e-05, | |
| "loss": 1.4105, | |
| "step": 4130000 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 1.3422269126675612e-05, | |
| "loss": 1.4104, | |
| "step": 4140000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 1.3333916692741066e-05, | |
| "loss": 1.4106, | |
| "step": 4150000 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 1.3245564258806516e-05, | |
| "loss": 1.4114, | |
| "step": 4160000 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 1.315721182487197e-05, | |
| "loss": 1.4108, | |
| "step": 4170000 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 1.306886822971633e-05, | |
| "loss": 1.409, | |
| "step": 4180000 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 1.298051579578178e-05, | |
| "loss": 1.4106, | |
| "step": 4190000 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 1.2892163361847234e-05, | |
| "loss": 1.4118, | |
| "step": 4200000 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 1.2803819766691594e-05, | |
| "loss": 1.4115, | |
| "step": 4210000 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 1.2715467332757044e-05, | |
| "loss": 1.4135, | |
| "step": 4220000 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 1.2627123737601404e-05, | |
| "loss": 1.4104, | |
| "step": 4230000 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 1.2538771303666858e-05, | |
| "loss": 1.4124, | |
| "step": 4240000 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 1.2450427708511214e-05, | |
| "loss": 1.4094, | |
| "step": 4250000 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 1.2362075274576668e-05, | |
| "loss": 1.4058, | |
| "step": 4260000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 1.227372284064212e-05, | |
| "loss": 1.4108, | |
| "step": 4270000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 1.2185379245486478e-05, | |
| "loss": 1.411, | |
| "step": 4280000 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 1.2097035650330837e-05, | |
| "loss": 1.4102, | |
| "step": 4290000 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 1.200868321639629e-05, | |
| "loss": 1.4126, | |
| "step": 4300000 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 1.1920330782461742e-05, | |
| "loss": 1.4104, | |
| "step": 4310000 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 1.18319871873061e-05, | |
| "loss": 1.4127, | |
| "step": 4320000 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 1.1743634753371553e-05, | |
| "loss": 1.4094, | |
| "step": 4330000 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 1.165529115821591e-05, | |
| "loss": 1.4097, | |
| "step": 4340000 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 1.1566938724281363e-05, | |
| "loss": 1.4095, | |
| "step": 4350000 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 1.1478595129125721e-05, | |
| "loss": 1.4079, | |
| "step": 4360000 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 1.1390242695191175e-05, | |
| "loss": 1.4127, | |
| "step": 4370000 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 1.1301890261256627e-05, | |
| "loss": 1.4079, | |
| "step": 4380000 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 1.1213546666100985e-05, | |
| "loss": 1.4065, | |
| "step": 4390000 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 1.1125203070945343e-05, | |
| "loss": 1.4098, | |
| "step": 4400000 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 1.1036850637010797e-05, | |
| "loss": 1.4123, | |
| "step": 4410000 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.0948507041855153e-05, | |
| "loss": 1.409, | |
| "step": 4420000 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 1.0860163446699512e-05, | |
| "loss": 1.4045, | |
| "step": 4430000 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 1.0771811012764965e-05, | |
| "loss": 1.4102, | |
| "step": 4440000 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 1.0683467417609323e-05, | |
| "loss": 1.4085, | |
| "step": 4450000 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 1.0595114983674775e-05, | |
| "loss": 1.4038, | |
| "step": 4460000 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 1.0506771388519134e-05, | |
| "loss": 1.4052, | |
| "step": 4470000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 1.0418427793363492e-05, | |
| "loss": 1.4094, | |
| "step": 4480000 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 1.033008419820785e-05, | |
| "loss": 1.4071, | |
| "step": 4490000 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 1.0241731764273304e-05, | |
| "loss": 1.4075, | |
| "step": 4500000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 1.0153379330338756e-05, | |
| "loss": 1.4047, | |
| "step": 4510000 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 1.0065035735183114e-05, | |
| "loss": 1.4063, | |
| "step": 4520000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bleu": 32.0886, | |
| "eval_gen_len": 65.8595, | |
| "eval_loss": 1.5665596723556519, | |
| "eval_runtime": 3002.5617, | |
| "eval_samples_per_second": 5.717, | |
| "eval_steps_per_second": 0.357, | |
| "step": 4525512 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 9.976692140027472e-06, | |
| "loss": 1.3896, | |
| "step": 4530000 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 9.88834854487183e-06, | |
| "loss": 1.3736, | |
| "step": 4540000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 9.799996110937282e-06, | |
| "loss": 1.3741, | |
| "step": 4550000 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 9.71165251578164e-06, | |
| "loss": 1.3717, | |
| "step": 4560000 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 9.623308920625997e-06, | |
| "loss": 1.3731, | |
| "step": 4570000 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 9.534965325470355e-06, | |
| "loss": 1.375, | |
| "step": 4580000 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 9.446612891535809e-06, | |
| "loss": 1.3764, | |
| "step": 4590000 | |
| }, | |
| { | |
| "epoch": 8.13, | |
| "learning_rate": 9.358269296380166e-06, | |
| "loss": 1.3752, | |
| "step": 4600000 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 9.26991686244562e-06, | |
| "loss": 1.3768, | |
| "step": 4610000 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 9.181573267289978e-06, | |
| "loss": 1.3767, | |
| "step": 4620000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 9.093229672134336e-06, | |
| "loss": 1.3769, | |
| "step": 4630000 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 9.004886076978694e-06, | |
| "loss": 1.3772, | |
| "step": 4640000 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 8.916533643044146e-06, | |
| "loss": 1.3766, | |
| "step": 4650000 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 8.828190047888504e-06, | |
| "loss": 1.38, | |
| "step": 4660000 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 8.739846452732862e-06, | |
| "loss": 1.3764, | |
| "step": 4670000 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 8.65150285757722e-06, | |
| "loss": 1.3765, | |
| "step": 4680000 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 8.563150423642673e-06, | |
| "loss": 1.3765, | |
| "step": 4690000 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 8.47480682848703e-06, | |
| "loss": 1.3756, | |
| "step": 4700000 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 8.386463233331389e-06, | |
| "loss": 1.3781, | |
| "step": 4710000 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 8.298110799396843e-06, | |
| "loss": 1.3788, | |
| "step": 4720000 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 8.209767204241201e-06, | |
| "loss": 1.3759, | |
| "step": 4730000 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 8.121423609085559e-06, | |
| "loss": 1.3783, | |
| "step": 4740000 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 8.033080013929916e-06, | |
| "loss": 1.3782, | |
| "step": 4750000 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 7.944736418774274e-06, | |
| "loss": 1.3792, | |
| "step": 4760000 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 7.856383984839727e-06, | |
| "loss": 1.3775, | |
| "step": 4770000 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 7.768040389684086e-06, | |
| "loss": 1.3779, | |
| "step": 4780000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 7.679696794528444e-06, | |
| "loss": 1.3797, | |
| "step": 4790000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 7.591353199372801e-06, | |
| "loss": 1.3761, | |
| "step": 4800000 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 7.503009604217158e-06, | |
| "loss": 1.3784, | |
| "step": 4810000 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 7.414666009061516e-06, | |
| "loss": 1.3769, | |
| "step": 4820000 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 7.32631357512697e-06, | |
| "loss": 1.3764, | |
| "step": 4830000 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 7.237969979971328e-06, | |
| "loss": 1.3818, | |
| "step": 4840000 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 7.149626384815686e-06, | |
| "loss": 1.3787, | |
| "step": 4850000 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 7.061282789660044e-06, | |
| "loss": 1.3762, | |
| "step": 4860000 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 6.972939194504401e-06, | |
| "loss": 1.3788, | |
| "step": 4870000 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 6.884595599348759e-06, | |
| "loss": 1.3752, | |
| "step": 4880000 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 6.796243165414212e-06, | |
| "loss": 1.3771, | |
| "step": 4890000 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "learning_rate": 6.70789957025857e-06, | |
| "loss": 1.3785, | |
| "step": 4900000 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 6.619555975102928e-06, | |
| "loss": 1.3746, | |
| "step": 4910000 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 6.531212379947286e-06, | |
| "loss": 1.3769, | |
| "step": 4920000 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 6.4428599460127385e-06, | |
| "loss": 1.3781, | |
| "step": 4930000 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 6.354516350857097e-06, | |
| "loss": 1.3756, | |
| "step": 4940000 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 6.266172755701455e-06, | |
| "loss": 1.3761, | |
| "step": 4950000 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 6.177829160545812e-06, | |
| "loss": 1.3762, | |
| "step": 4960000 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 6.089476726611266e-06, | |
| "loss": 1.3754, | |
| "step": 4970000 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 6.001133131455623e-06, | |
| "loss": 1.3728, | |
| "step": 4980000 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 5.9127895362999815e-06, | |
| "loss": 1.3759, | |
| "step": 4990000 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 5.82444594114434e-06, | |
| "loss": 1.374, | |
| "step": 5000000 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 5.736102345988698e-06, | |
| "loss": 1.3737, | |
| "step": 5010000 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 5.647758750833055e-06, | |
| "loss": 1.3749, | |
| "step": 5020000 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 5.559406316898508e-06, | |
| "loss": 1.3757, | |
| "step": 5030000 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 5.4710627217428655e-06, | |
| "loss": 1.376, | |
| "step": 5040000 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 5.382719126587224e-06, | |
| "loss": 1.3753, | |
| "step": 5050000 | |
| }, | |
| { | |
| "epoch": 8.94, | |
| "learning_rate": 5.294375531431582e-06, | |
| "loss": 1.3763, | |
| "step": 5060000 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.206031936275939e-06, | |
| "loss": 1.3731, | |
| "step": 5070000 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 5.117679502341393e-06, | |
| "loss": 1.3711, | |
| "step": 5080000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5.02933590718575e-06, | |
| "loss": 1.3724, | |
| "step": 5090000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_bleu": 32.3644, | |
| "eval_gen_len": 66.1648, | |
| "eval_loss": 1.5537199974060059, | |
| "eval_runtime": 3034.8877, | |
| "eval_samples_per_second": 5.656, | |
| "eval_steps_per_second": 0.354, | |
| "step": 5091201 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 4.9409923120301085e-06, | |
| "loss": 1.3485, | |
| "step": 5100000 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 4.852648716874467e-06, | |
| "loss": 1.3453, | |
| "step": 5110000 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 4.764305121718825e-06, | |
| "loss": 1.3439, | |
| "step": 5120000 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 4.675961526563182e-06, | |
| "loss": 1.3475, | |
| "step": 5130000 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.5876179314075404e-06, | |
| "loss": 1.3443, | |
| "step": 5140000 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 4.499265497472993e-06, | |
| "loss": 1.3456, | |
| "step": 5150000 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 4.4109219023173515e-06, | |
| "loss": 1.3474, | |
| "step": 5160000 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 4.32257830716171e-06, | |
| "loss": 1.3491, | |
| "step": 5170000 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 4.234234712006067e-06, | |
| "loss": 1.3491, | |
| "step": 5180000 | |
| }, | |
| { | |
| "epoch": 9.17, | |
| "learning_rate": 4.14588227807152e-06, | |
| "loss": 1.3493, | |
| "step": 5190000 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 4.057538682915878e-06, | |
| "loss": 1.3485, | |
| "step": 5200000 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 3.9691950877602355e-06, | |
| "loss": 1.3506, | |
| "step": 5210000 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 3.880842653825688e-06, | |
| "loss": 1.3498, | |
| "step": 5220000 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 3.7924990586700466e-06, | |
| "loss": 1.3453, | |
| "step": 5230000 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 3.7041554635144048e-06, | |
| "loss": 1.3475, | |
| "step": 5240000 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 3.615811868358763e-06, | |
| "loss": 1.3469, | |
| "step": 5250000 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 3.5274682732031203e-06, | |
| "loss": 1.3477, | |
| "step": 5260000 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 3.4391246780474785e-06, | |
| "loss": 1.3443, | |
| "step": 5270000 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 3.3507810828918367e-06, | |
| "loss": 1.3478, | |
| "step": 5280000 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 3.2624374877361945e-06, | |
| "loss": 1.3473, | |
| "step": 5290000 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 3.1740850538016474e-06, | |
| "loss": 1.349, | |
| "step": 5300000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "learning_rate": 3.085741458646005e-06, | |
| "loss": 1.3485, | |
| "step": 5310000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 2.9973978634903633e-06, | |
| "loss": 1.349, | |
| "step": 5320000 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 2.909054268334721e-06, | |
| "loss": 1.3478, | |
| "step": 5330000 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.8207106731790793e-06, | |
| "loss": 1.3489, | |
| "step": 5340000 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 2.732367078023437e-06, | |
| "loss": 1.3446, | |
| "step": 5350000 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 2.644023482867795e-06, | |
| "loss": 1.3478, | |
| "step": 5360000 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 2.5556798877121526e-06, | |
| "loss": 1.3491, | |
| "step": 5370000 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 2.467336292556511e-06, | |
| "loss": 1.3467, | |
| "step": 5380000 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 2.3789926974008686e-06, | |
| "loss": 1.3492, | |
| "step": 5390000 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 2.290640263466322e-06, | |
| "loss": 1.3472, | |
| "step": 5400000 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.2022966683106797e-06, | |
| "loss": 1.3439, | |
| "step": 5410000 | |
| }, | |
| { | |
| "epoch": 9.58, | |
| "learning_rate": 2.113953073155038e-06, | |
| "loss": 1.3474, | |
| "step": 5420000 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 2.0256094779993956e-06, | |
| "loss": 1.3459, | |
| "step": 5430000 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 1.9372658828437534e-06, | |
| "loss": 1.3485, | |
| "step": 5440000 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 1.8489134489092063e-06, | |
| "loss": 1.3438, | |
| "step": 5450000 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 1.7605698537535645e-06, | |
| "loss": 1.3449, | |
| "step": 5460000 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 1.6722262585979223e-06, | |
| "loss": 1.3472, | |
| "step": 5470000 | |
| }, | |
| { | |
| "epoch": 9.69, | |
| "learning_rate": 1.5838826634422803e-06, | |
| "loss": 1.3476, | |
| "step": 5480000 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 1.495539068286638e-06, | |
| "loss": 1.3462, | |
| "step": 5490000 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 1.407195473130996e-06, | |
| "loss": 1.3479, | |
| "step": 5500000 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 1.318851877975354e-06, | |
| "loss": 1.3441, | |
| "step": 5510000 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 1.2304994440408069e-06, | |
| "loss": 1.3439, | |
| "step": 5520000 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 1.1421558488851649e-06, | |
| "loss": 1.3459, | |
| "step": 5530000 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 1.0538122537295229e-06, | |
| "loss": 1.3443, | |
| "step": 5540000 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 9.654686585738808e-07, | |
| "loss": 1.3455, | |
| "step": 5550000 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 8.771162246393337e-07, | |
| "loss": 1.3448, | |
| "step": 5560000 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 7.887726294836917e-07, | |
| "loss": 1.3432, | |
| "step": 5570000 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 7.004290343280496e-07, | |
| "loss": 1.3444, | |
| "step": 5580000 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 6.120854391724075e-07, | |
| "loss": 1.3432, | |
| "step": 5590000 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 5.237418440167655e-07, | |
| "loss": 1.3454, | |
| "step": 5600000 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 4.353894100822184e-07, | |
| "loss": 1.3437, | |
| "step": 5610000 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 3.4704581492657626e-07, | |
| "loss": 1.3447, | |
| "step": 5620000 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 2.587022197709342e-07, | |
| "loss": 1.3441, | |
| "step": 5630000 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 1.7035862461529215e-07, | |
| "loss": 1.3426, | |
| "step": 5640000 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "learning_rate": 8.201502945965009e-08, | |
| "loss": 1.3452, | |
| "step": 5650000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_bleu": 32.4724, | |
| "eval_gen_len": 66.1539, | |
| "eval_loss": 1.5473366975784302, | |
| "eval_runtime": 3064.956, | |
| "eval_samples_per_second": 5.6, | |
| "eval_steps_per_second": 0.35, | |
| "step": 5656890 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 5656890, | |
| "total_flos": 1.8656478019360383e+19, | |
| "train_loss": 1.5652431253911159, | |
| "train_runtime": 1147827.3858, | |
| "train_samples_per_second": 78.854, | |
| "train_steps_per_second": 4.928 | |
| } | |
| ], | |
| "max_steps": 5656890, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.8656478019360383e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |