|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.26666666666666666, |
|
"eval_steps": 500, |
|
"global_step": 2500, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 0.00019920000000000002, |
|
"loss": 0.672, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 0.0001984, |
|
"loss": 0.6133, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 0.0001976, |
|
"loss": 0.585, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 0.0001968, |
|
"loss": 0.5738, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.000196, |
|
"loss": 0.6044, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001952, |
|
"loss": 0.5787, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001944, |
|
"loss": 0.6115, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.00019360000000000002, |
|
"loss": 0.5353, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001928, |
|
"loss": 0.5436, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.000192, |
|
"loss": 0.5947, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001912, |
|
"loss": 0.5773, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001904, |
|
"loss": 0.544, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001896, |
|
"loss": 0.568, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0001888, |
|
"loss": 0.5503, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.000188, |
|
"loss": 0.5963, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.00018720000000000002, |
|
"loss": 0.4925, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.00018640000000000003, |
|
"loss": 0.5489, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0001856, |
|
"loss": 0.5321, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.00018480000000000002, |
|
"loss": 0.6061, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.00018400000000000003, |
|
"loss": 0.5699, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0001832, |
|
"loss": 0.5725, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.00018240000000000002, |
|
"loss": 0.5132, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.00018160000000000002, |
|
"loss": 0.5558, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0001808, |
|
"loss": 0.5192, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.00018, |
|
"loss": 0.5916, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.00017920000000000002, |
|
"loss": 0.5719, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0001784, |
|
"loss": 0.6012, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0001776, |
|
"loss": 0.6292, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.00017680000000000001, |
|
"loss": 0.5424, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.00017600000000000002, |
|
"loss": 0.5623, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0001752, |
|
"loss": 0.5139, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0001744, |
|
"loss": 0.5409, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.00017360000000000002, |
|
"loss": 0.5562, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.0001728, |
|
"loss": 0.5624, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.000172, |
|
"loss": 0.5449, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.00017120000000000001, |
|
"loss": 0.5555, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.0001704, |
|
"loss": 0.5622, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.0001696, |
|
"loss": 0.5714, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.0001688, |
|
"loss": 0.5434, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.000168, |
|
"loss": 0.5492, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.0001672, |
|
"loss": 0.536, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.0001664, |
|
"loss": 0.5307, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.0001656, |
|
"loss": 0.5957, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.0001648, |
|
"loss": 0.5935, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.000164, |
|
"loss": 0.5643, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.0001632, |
|
"loss": 0.5483, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.00016240000000000002, |
|
"loss": 0.5415, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.00016160000000000002, |
|
"loss": 0.5383, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.0001608, |
|
"loss": 0.5444, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.00016, |
|
"loss": 0.592, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.00015920000000000002, |
|
"loss": 0.5295, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.00015840000000000003, |
|
"loss": 0.5505, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.0001576, |
|
"loss": 0.5954, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.00015680000000000002, |
|
"loss": 0.5609, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.00015600000000000002, |
|
"loss": 0.5614, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.0001552, |
|
"loss": 0.5902, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.0001544, |
|
"loss": 0.539, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.00015360000000000002, |
|
"loss": 0.5821, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.0001528, |
|
"loss": 0.5347, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 0.000152, |
|
"loss": 0.5322, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.00015120000000000002, |
|
"loss": 0.6044, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.0001504, |
|
"loss": 0.5941, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.0001496, |
|
"loss": 0.5473, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.0001488, |
|
"loss": 0.5833, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.000148, |
|
"loss": 0.5338, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.0001472, |
|
"loss": 0.5687, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.0001464, |
|
"loss": 0.5591, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.00014560000000000002, |
|
"loss": 0.5691, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.0001448, |
|
"loss": 0.5452, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.000144, |
|
"loss": 0.5752, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0001432, |
|
"loss": 0.6147, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0001424, |
|
"loss": 0.5441, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0001416, |
|
"loss": 0.4872, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0001408, |
|
"loss": 0.5158, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.00014, |
|
"loss": 0.5917, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0001392, |
|
"loss": 0.5357, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0001384, |
|
"loss": 0.5594, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.00013759999999999998, |
|
"loss": 0.5654, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.00013680000000000002, |
|
"loss": 0.5621, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.00013600000000000003, |
|
"loss": 0.5776, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.0001352, |
|
"loss": 0.5439, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.00013440000000000001, |
|
"loss": 0.5433, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.00013360000000000002, |
|
"loss": 0.5765, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.0001328, |
|
"loss": 0.5169, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.000132, |
|
"loss": 0.5308, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.00013120000000000002, |
|
"loss": 0.4969, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.0001304, |
|
"loss": 0.5865, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.0001296, |
|
"loss": 0.5639, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 0.00012880000000000001, |
|
"loss": 0.5635, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.00012800000000000002, |
|
"loss": 0.5438, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.0001272, |
|
"loss": 0.5678, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.0001264, |
|
"loss": 0.538, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.00012560000000000002, |
|
"loss": 0.5491, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.0001248, |
|
"loss": 0.5836, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.000124, |
|
"loss": 0.4828, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.0001232, |
|
"loss": 0.5376, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.0001224, |
|
"loss": 0.5001, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.0001216, |
|
"loss": 0.5483, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001208, |
|
"loss": 0.5528, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.00012, |
|
"loss": 0.5651, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001192, |
|
"loss": 0.5435, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001184, |
|
"loss": 0.588, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001176, |
|
"loss": 0.5459, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.00011679999999999999, |
|
"loss": 0.5923, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.000116, |
|
"loss": 0.5071, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001152, |
|
"loss": 0.5964, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001144, |
|
"loss": 0.5785, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.0001136, |
|
"loss": 0.5776, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00011279999999999999, |
|
"loss": 0.5412, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00011200000000000001, |
|
"loss": 0.5943, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00011120000000000002, |
|
"loss": 0.4882, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00011040000000000001, |
|
"loss": 0.523, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00010960000000000001, |
|
"loss": 0.5669, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00010880000000000002, |
|
"loss": 0.5745, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00010800000000000001, |
|
"loss": 0.5419, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00010720000000000002, |
|
"loss": 0.5433, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00010640000000000001, |
|
"loss": 0.6327, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.0001056, |
|
"loss": 0.5044, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.00010480000000000001, |
|
"loss": 0.5386, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.00010400000000000001, |
|
"loss": 0.5734, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.0001032, |
|
"loss": 0.5193, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.00010240000000000001, |
|
"loss": 0.5335, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.0001016, |
|
"loss": 0.5776, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.00010080000000000001, |
|
"loss": 0.5472, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 0.0001, |
|
"loss": 0.5383, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.92e-05, |
|
"loss": 0.5626, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.84e-05, |
|
"loss": 0.5664, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.76e-05, |
|
"loss": 0.5044, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.680000000000001e-05, |
|
"loss": 0.5203, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.6e-05, |
|
"loss": 0.549, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.52e-05, |
|
"loss": 0.4875, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.44e-05, |
|
"loss": 0.552, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.360000000000001e-05, |
|
"loss": 0.5711, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.28e-05, |
|
"loss": 0.5239, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.200000000000001e-05, |
|
"loss": 0.5446, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.120000000000001e-05, |
|
"loss": 0.5091, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.04e-05, |
|
"loss": 0.5244, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.960000000000001e-05, |
|
"loss": 0.5609, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.88e-05, |
|
"loss": 0.5794, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.800000000000001e-05, |
|
"loss": 0.5486, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.72e-05, |
|
"loss": 0.5047, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.64e-05, |
|
"loss": 0.541, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.560000000000001e-05, |
|
"loss": 0.509, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.48e-05, |
|
"loss": 0.4901, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.4e-05, |
|
"loss": 0.5326, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.32e-05, |
|
"loss": 0.5754, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.24e-05, |
|
"loss": 0.5574, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.16e-05, |
|
"loss": 0.5752, |
|
"step": 1480 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.080000000000001e-05, |
|
"loss": 0.5834, |
|
"step": 1490 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8e-05, |
|
"loss": 0.5168, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 7.920000000000001e-05, |
|
"loss": 0.5751, |
|
"step": 1510 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 7.840000000000001e-05, |
|
"loss": 0.6164, |
|
"step": 1520 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 7.76e-05, |
|
"loss": 0.542, |
|
"step": 1530 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 7.680000000000001e-05, |
|
"loss": 0.5558, |
|
"step": 1540 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.6e-05, |
|
"loss": 0.524, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.52e-05, |
|
"loss": 0.6232, |
|
"step": 1560 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.44e-05, |
|
"loss": 0.4692, |
|
"step": 1570 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.36e-05, |
|
"loss": 0.5604, |
|
"step": 1580 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.280000000000001e-05, |
|
"loss": 0.5242, |
|
"step": 1590 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.2e-05, |
|
"loss": 0.5455, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.12e-05, |
|
"loss": 0.5362, |
|
"step": 1610 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.04e-05, |
|
"loss": 0.5257, |
|
"step": 1620 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 6.96e-05, |
|
"loss": 0.5601, |
|
"step": 1630 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 6.879999999999999e-05, |
|
"loss": 0.5484, |
|
"step": 1640 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.800000000000001e-05, |
|
"loss": 0.5787, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.720000000000001e-05, |
|
"loss": 0.6185, |
|
"step": 1660 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.64e-05, |
|
"loss": 0.5453, |
|
"step": 1670 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.560000000000001e-05, |
|
"loss": 0.5297, |
|
"step": 1680 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.48e-05, |
|
"loss": 0.5538, |
|
"step": 1690 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.400000000000001e-05, |
|
"loss": 0.5651, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.32e-05, |
|
"loss": 0.5764, |
|
"step": 1710 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.24e-05, |
|
"loss": 0.5414, |
|
"step": 1720 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 6.16e-05, |
|
"loss": 0.5264, |
|
"step": 1730 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 6.08e-05, |
|
"loss": 0.5472, |
|
"step": 1740 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 6e-05, |
|
"loss": 0.5876, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.92e-05, |
|
"loss": 0.5346, |
|
"step": 1760 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.8399999999999997e-05, |
|
"loss": 0.5856, |
|
"step": 1770 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.76e-05, |
|
"loss": 0.592, |
|
"step": 1780 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.68e-05, |
|
"loss": 0.535, |
|
"step": 1790 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.6000000000000006e-05, |
|
"loss": 0.5427, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.520000000000001e-05, |
|
"loss": 0.5296, |
|
"step": 1810 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 5.440000000000001e-05, |
|
"loss": 0.5307, |
|
"step": 1820 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 5.360000000000001e-05, |
|
"loss": 0.5943, |
|
"step": 1830 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 5.28e-05, |
|
"loss": 0.5458, |
|
"step": 1840 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 5.2000000000000004e-05, |
|
"loss": 0.5331, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 5.1200000000000004e-05, |
|
"loss": 0.5342, |
|
"step": 1860 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 5.0400000000000005e-05, |
|
"loss": 0.6056, |
|
"step": 1870 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.96e-05, |
|
"loss": 0.5618, |
|
"step": 1880 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.88e-05, |
|
"loss": 0.5471, |
|
"step": 1890 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.8e-05, |
|
"loss": 0.5799, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.72e-05, |
|
"loss": 0.5605, |
|
"step": 1910 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.64e-05, |
|
"loss": 0.5294, |
|
"step": 1920 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.5600000000000004e-05, |
|
"loss": 0.469, |
|
"step": 1930 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.4800000000000005e-05, |
|
"loss": 0.5056, |
|
"step": 1940 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.4000000000000006e-05, |
|
"loss": 0.5084, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.32e-05, |
|
"loss": 0.4769, |
|
"step": 1960 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.24e-05, |
|
"loss": 0.5651, |
|
"step": 1970 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.16e-05, |
|
"loss": 0.5117, |
|
"step": 1980 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.08e-05, |
|
"loss": 0.5165, |
|
"step": 1990 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4e-05, |
|
"loss": 0.5167, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 3.9200000000000004e-05, |
|
"loss": 0.5029, |
|
"step": 2010 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.8400000000000005e-05, |
|
"loss": 0.5706, |
|
"step": 2020 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.76e-05, |
|
"loss": 0.5751, |
|
"step": 2030 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.68e-05, |
|
"loss": 0.6197, |
|
"step": 2040 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.6e-05, |
|
"loss": 0.5016, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.52e-05, |
|
"loss": 0.5219, |
|
"step": 2060 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.4399999999999996e-05, |
|
"loss": 0.5708, |
|
"step": 2070 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.3600000000000004e-05, |
|
"loss": 0.5139, |
|
"step": 2080 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.2800000000000004e-05, |
|
"loss": 0.5225, |
|
"step": 2090 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.2000000000000005e-05, |
|
"loss": 0.5606, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 3.12e-05, |
|
"loss": 0.5264, |
|
"step": 2110 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 3.04e-05, |
|
"loss": 0.5016, |
|
"step": 2120 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.96e-05, |
|
"loss": 0.5609, |
|
"step": 2130 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.88e-05, |
|
"loss": 0.5286, |
|
"step": 2140 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.8000000000000003e-05, |
|
"loss": 0.5327, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.7200000000000004e-05, |
|
"loss": 0.5381, |
|
"step": 2160 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.64e-05, |
|
"loss": 0.5476, |
|
"step": 2170 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.5600000000000002e-05, |
|
"loss": 0.5325, |
|
"step": 2180 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.48e-05, |
|
"loss": 0.571, |
|
"step": 2190 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.4e-05, |
|
"loss": 0.5726, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.32e-05, |
|
"loss": 0.5485, |
|
"step": 2210 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.2400000000000002e-05, |
|
"loss": 0.5847, |
|
"step": 2220 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.16e-05, |
|
"loss": 0.5154, |
|
"step": 2230 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.08e-05, |
|
"loss": 0.6066, |
|
"step": 2240 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2e-05, |
|
"loss": 0.5618, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.9200000000000003e-05, |
|
"loss": 0.5221, |
|
"step": 2260 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.84e-05, |
|
"loss": 0.5052, |
|
"step": 2270 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.76e-05, |
|
"loss": 0.5675, |
|
"step": 2280 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.6800000000000002e-05, |
|
"loss": 0.5574, |
|
"step": 2290 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.6000000000000003e-05, |
|
"loss": 0.5942, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.52e-05, |
|
"loss": 0.6035, |
|
"step": 2310 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.44e-05, |
|
"loss": 0.5454, |
|
"step": 2320 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.3600000000000002e-05, |
|
"loss": 0.5231, |
|
"step": 2330 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.2800000000000001e-05, |
|
"loss": 0.5558, |
|
"step": 2340 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.2e-05, |
|
"loss": 0.5233, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.1200000000000001e-05, |
|
"loss": 0.4485, |
|
"step": 2360 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.04e-05, |
|
"loss": 0.5184, |
|
"step": 2370 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.600000000000001e-06, |
|
"loss": 0.5298, |
|
"step": 2380 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.8e-06, |
|
"loss": 0.5317, |
|
"step": 2390 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 0.5399, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 7.2e-06, |
|
"loss": 0.5639, |
|
"step": 2410 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 6.4000000000000006e-06, |
|
"loss": 0.5616, |
|
"step": 2420 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 5.600000000000001e-06, |
|
"loss": 0.5761, |
|
"step": 2430 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.800000000000001e-06, |
|
"loss": 0.5912, |
|
"step": 2440 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 0.601, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 3.2000000000000003e-06, |
|
"loss": 0.5664, |
|
"step": 2460 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 0.6286, |
|
"step": 2470 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.6000000000000001e-06, |
|
"loss": 0.4989, |
|
"step": 2480 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 8.000000000000001e-07, |
|
"loss": 0.5339, |
|
"step": 2490 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 0.0, |
|
"loss": 0.525, |
|
"step": 2500 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 2500, |
|
"num_train_epochs": 1, |
|
"save_steps": 500, |
|
"total_flos": 4.243363257237504e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|