| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.14381040036815462, | |
| "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": 0.8433, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001984, | |
| "loss": 0.8318, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001976, | |
| "loss": 0.8495, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001968, | |
| "loss": 0.8214, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000196, | |
| "loss": 0.927, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001952, | |
| "loss": 0.8726, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001944, | |
| "loss": 0.8191, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019360000000000002, | |
| "loss": 0.7768, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001928, | |
| "loss": 0.8353, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000192, | |
| "loss": 0.8387, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001912, | |
| "loss": 0.7577, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001904, | |
| "loss": 0.8497, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001896, | |
| "loss": 0.8589, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001888, | |
| "loss": 0.838, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000188, | |
| "loss": 0.8109, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018720000000000002, | |
| "loss": 0.7742, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018640000000000003, | |
| "loss": 0.8455, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001856, | |
| "loss": 0.8724, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018480000000000002, | |
| "loss": 0.7828, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 0.7898, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001832, | |
| "loss": 0.8237, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018240000000000002, | |
| "loss": 0.8804, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018160000000000002, | |
| "loss": 0.8122, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001808, | |
| "loss": 0.8581, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018, | |
| "loss": 0.8293, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017920000000000002, | |
| "loss": 0.8052, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001784, | |
| "loss": 0.8382, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001776, | |
| "loss": 0.7908, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017680000000000001, | |
| "loss": 0.783, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 0.7574, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001752, | |
| "loss": 0.8422, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001744, | |
| "loss": 0.8644, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017360000000000002, | |
| "loss": 0.8133, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001728, | |
| "loss": 0.8363, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000172, | |
| "loss": 0.7527, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017120000000000001, | |
| "loss": 0.752, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001704, | |
| "loss": 0.8825, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001696, | |
| "loss": 0.754, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001688, | |
| "loss": 0.8108, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000168, | |
| "loss": 0.8138, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001672, | |
| "loss": 0.7985, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001664, | |
| "loss": 0.8676, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001656, | |
| "loss": 0.7805, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001648, | |
| "loss": 0.8159, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000164, | |
| "loss": 0.8497, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001632, | |
| "loss": 0.7753, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016240000000000002, | |
| "loss": 0.7725, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016160000000000002, | |
| "loss": 0.7222, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001608, | |
| "loss": 0.779, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016, | |
| "loss": 0.82, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015920000000000002, | |
| "loss": 0.7797, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015840000000000003, | |
| "loss": 0.764, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001576, | |
| "loss": 0.8109, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015680000000000002, | |
| "loss": 0.7836, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015600000000000002, | |
| "loss": 0.7792, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001552, | |
| "loss": 0.8351, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001544, | |
| "loss": 0.7221, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015360000000000002, | |
| "loss": 0.8869, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001528, | |
| "loss": 0.7721, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000152, | |
| "loss": 0.8214, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015120000000000002, | |
| "loss": 0.8489, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001504, | |
| "loss": 0.8178, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001496, | |
| "loss": 0.8017, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001488, | |
| "loss": 0.8061, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000148, | |
| "loss": 0.7493, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001472, | |
| "loss": 0.7679, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001464, | |
| "loss": 0.783, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00014560000000000002, | |
| "loss": 0.7395, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001448, | |
| "loss": 0.7906, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000144, | |
| "loss": 0.7583, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001432, | |
| "loss": 0.8346, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001424, | |
| "loss": 0.8341, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001416, | |
| "loss": 0.808, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001408, | |
| "loss": 0.8754, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00014, | |
| "loss": 0.8299, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001392, | |
| "loss": 0.868, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001384, | |
| "loss": 0.8581, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00013759999999999998, | |
| "loss": 0.8532, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013680000000000002, | |
| "loss": 0.7706, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 0.8238, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001352, | |
| "loss": 0.7712, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013440000000000001, | |
| "loss": 0.8446, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013360000000000002, | |
| "loss": 0.7679, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001328, | |
| "loss": 0.8012, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000132, | |
| "loss": 0.832, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013120000000000002, | |
| "loss": 0.7592, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001304, | |
| "loss": 0.8324, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001296, | |
| "loss": 0.7662, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012880000000000001, | |
| "loss": 0.8185, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 0.8729, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001272, | |
| "loss": 0.7608, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001264, | |
| "loss": 0.714, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012560000000000002, | |
| "loss": 0.7721, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001248, | |
| "loss": 0.8258, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000124, | |
| "loss": 0.7798, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001232, | |
| "loss": 0.7993, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001224, | |
| "loss": 0.8531, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001216, | |
| "loss": 0.8215, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001208, | |
| "loss": 0.7861, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00012, | |
| "loss": 0.8256, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001192, | |
| "loss": 0.7763, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001184, | |
| "loss": 0.8093, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001176, | |
| "loss": 0.8607, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00011679999999999999, | |
| "loss": 0.8037, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000116, | |
| "loss": 0.8355, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001152, | |
| "loss": 0.7644, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001144, | |
| "loss": 0.8011, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001136, | |
| "loss": 0.7949, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00011279999999999999, | |
| "loss": 0.7978, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 0.8408, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00011120000000000002, | |
| "loss": 0.7807, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00011040000000000001, | |
| "loss": 0.8224, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010960000000000001, | |
| "loss": 0.825, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010880000000000002, | |
| "loss": 0.7957, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010800000000000001, | |
| "loss": 0.8792, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010720000000000002, | |
| "loss": 0.741, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010640000000000001, | |
| "loss": 0.8166, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0001056, | |
| "loss": 0.7794, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010480000000000001, | |
| "loss": 0.8217, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 0.8358, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0001032, | |
| "loss": 0.7606, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010240000000000001, | |
| "loss": 0.7872, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0001016, | |
| "loss": 0.7629, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00010080000000000001, | |
| "loss": 0.7993, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8389, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.92e-05, | |
| "loss": 0.7829, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.84e-05, | |
| "loss": 0.7919, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.76e-05, | |
| "loss": 0.729, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.680000000000001e-05, | |
| "loss": 0.7501, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.6e-05, | |
| "loss": 0.8586, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.52e-05, | |
| "loss": 0.8072, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.44e-05, | |
| "loss": 0.765, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.360000000000001e-05, | |
| "loss": 0.7719, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.28e-05, | |
| "loss": 0.8161, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 0.8253, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.120000000000001e-05, | |
| "loss": 0.7934, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.04e-05, | |
| "loss": 0.8491, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.960000000000001e-05, | |
| "loss": 0.7613, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.88e-05, | |
| "loss": 0.8765, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 0.7236, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.72e-05, | |
| "loss": 0.8744, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.64e-05, | |
| "loss": 0.8126, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.560000000000001e-05, | |
| "loss": 0.7705, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.48e-05, | |
| "loss": 0.7312, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.4e-05, | |
| "loss": 0.7966, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.32e-05, | |
| "loss": 0.838, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.24e-05, | |
| "loss": 0.8533, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.16e-05, | |
| "loss": 0.7606, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 0.8603, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8e-05, | |
| "loss": 0.8102, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.920000000000001e-05, | |
| "loss": 0.84, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.840000000000001e-05, | |
| "loss": 0.8244, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.76e-05, | |
| "loss": 0.7735, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.680000000000001e-05, | |
| "loss": 0.8483, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.6e-05, | |
| "loss": 0.7737, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.52e-05, | |
| "loss": 0.7836, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.44e-05, | |
| "loss": 0.7995, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.36e-05, | |
| "loss": 0.7608, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.280000000000001e-05, | |
| "loss": 0.8071, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.2e-05, | |
| "loss": 0.7472, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.12e-05, | |
| "loss": 0.8213, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.04e-05, | |
| "loss": 0.7683, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.96e-05, | |
| "loss": 0.8117, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.879999999999999e-05, | |
| "loss": 0.813, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 0.8026, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.720000000000001e-05, | |
| "loss": 0.7926, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.64e-05, | |
| "loss": 0.8832, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.560000000000001e-05, | |
| "loss": 0.8102, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.48e-05, | |
| "loss": 0.7791, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 0.8272, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.32e-05, | |
| "loss": 0.8416, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.24e-05, | |
| "loss": 0.8087, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.16e-05, | |
| "loss": 0.8529, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.08e-05, | |
| "loss": 0.6999, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6e-05, | |
| "loss": 0.7759, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.92e-05, | |
| "loss": 0.768, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.8399999999999997e-05, | |
| "loss": 0.7776, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.76e-05, | |
| "loss": 0.8465, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.68e-05, | |
| "loss": 0.8507, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 0.7806, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.520000000000001e-05, | |
| "loss": 0.7733, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.440000000000001e-05, | |
| "loss": 0.8462, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.360000000000001e-05, | |
| "loss": 0.7765, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.28e-05, | |
| "loss": 0.8014, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 0.8057, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.1200000000000004e-05, | |
| "loss": 0.7779, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.0400000000000005e-05, | |
| "loss": 0.8242, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.96e-05, | |
| "loss": 0.7589, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.88e-05, | |
| "loss": 0.8135, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.76, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.72e-05, | |
| "loss": 0.7453, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.64e-05, | |
| "loss": 0.7412, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.5600000000000004e-05, | |
| "loss": 0.7889, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.4800000000000005e-05, | |
| "loss": 0.8541, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.712, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.32e-05, | |
| "loss": 0.7794, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.24e-05, | |
| "loss": 0.7653, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.16e-05, | |
| "loss": 0.8125, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.08e-05, | |
| "loss": 0.6878, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4e-05, | |
| "loss": 0.7914, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.9200000000000004e-05, | |
| "loss": 0.7461, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.8400000000000005e-05, | |
| "loss": 0.6968, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.76e-05, | |
| "loss": 0.8051, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.68e-05, | |
| "loss": 0.8327, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.8272, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.52e-05, | |
| "loss": 0.8849, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.4399999999999996e-05, | |
| "loss": 0.8411, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.3600000000000004e-05, | |
| "loss": 0.7671, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.2800000000000004e-05, | |
| "loss": 0.7921, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.7841, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.12e-05, | |
| "loss": 0.7425, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.04e-05, | |
| "loss": 0.7849, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.96e-05, | |
| "loss": 0.7881, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.88e-05, | |
| "loss": 0.8554, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.8342, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.7200000000000004e-05, | |
| "loss": 0.7181, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.64e-05, | |
| "loss": 0.8053, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.5600000000000002e-05, | |
| "loss": 0.7378, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.48e-05, | |
| "loss": 0.7249, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.8441, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.32e-05, | |
| "loss": 0.7808, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.2400000000000002e-05, | |
| "loss": 0.8122, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.16e-05, | |
| "loss": 0.783, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.08e-05, | |
| "loss": 0.7651, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7669, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 0.8383, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.84e-05, | |
| "loss": 0.8314, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.76e-05, | |
| "loss": 0.7447, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 0.7987, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.6986, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.52e-05, | |
| "loss": 0.8227, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.44e-05, | |
| "loss": 0.7486, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 0.8058, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.2800000000000001e-05, | |
| "loss": 0.8046, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.7885, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.1200000000000001e-05, | |
| "loss": 0.8313, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.04e-05, | |
| "loss": 0.7741, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 0.8277, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.8e-06, | |
| "loss": 0.7546, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.7758, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.2e-06, | |
| "loss": 0.8794, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.7627, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 0.8617, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.7735, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.738, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 0.8132, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.8427, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.8225, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.7452, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0, | |
| "loss": 0.7845, | |
| "step": 2500 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2500, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 1.6856245075968e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |