| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "global_step": 3031, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00013697367205542726, |
| "loss": 4.515, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001367473441108545, |
| "loss": 4.4104, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00013652101616628176, |
| "loss": 4.1934, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.000136294688221709, |
| "loss": 4.2499, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00013606836027713626, |
| "loss": 4.1236, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00013584203233256352, |
| "loss": 4.1098, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00013561570438799076, |
| "loss": 4.0061, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.000135389376443418, |
| "loss": 3.8647, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00013516304849884526, |
| "loss": 3.9263, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00013493672055427252, |
| "loss": 3.8562, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00013471039260969976, |
| "loss": 3.9549, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00013448406466512702, |
| "loss": 3.947, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00013425773672055428, |
| "loss": 3.9132, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00013403140877598152, |
| "loss": 3.7056, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00013380508083140878, |
| "loss": 3.9002, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00013357875288683602, |
| "loss": 3.9315, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00013335242494226328, |
| "loss": 3.8103, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00013312609699769055, |
| "loss": 3.7449, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00013289976905311778, |
| "loss": 3.7089, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00013267344110854504, |
| "loss": 3.8787, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00013244711316397228, |
| "loss": 3.797, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00013222078521939954, |
| "loss": 3.5114, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00013199445727482678, |
| "loss": 3.8432, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00013176812933025404, |
| "loss": 3.4919, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00013154180138568128, |
| "loss": 3.683, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00013131547344110854, |
| "loss": 3.9541, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001310891454965358, |
| "loss": 3.6708, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00013086281755196304, |
| "loss": 3.7126, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001306364896073903, |
| "loss": 3.6123, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00013041016166281757, |
| "loss": 3.7501, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001301838337182448, |
| "loss": 3.6422, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00012995750577367206, |
| "loss": 3.5964, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001297311778290993, |
| "loss": 3.7605, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00012950484988452656, |
| "loss": 3.5969, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00012927852193995383, |
| "loss": 3.6422, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00012905219399538106, |
| "loss": 3.3317, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00012882586605080833, |
| "loss": 3.3237, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00012859953810623556, |
| "loss": 3.511, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00012837321016166282, |
| "loss": 3.634, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00012814688221709006, |
| "loss": 3.2707, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00012792055427251732, |
| "loss": 3.7581, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00012769422632794456, |
| "loss": 3.7056, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00012746789838337182, |
| "loss": 3.784, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00012724157043879909, |
| "loss": 2.9932, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00012701524249422632, |
| "loss": 3.6009, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00012678891454965358, |
| "loss": 3.7976, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00012656258660508085, |
| "loss": 3.7625, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00012633625866050808, |
| "loss": 3.264, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00012610993071593535, |
| "loss": 3.7345, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00012588360277136258, |
| "loss": 3.5308, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00012565727482678985, |
| "loss": 3.7175, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001254309468822171, |
| "loss": 3.5114, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00012520461893764434, |
| "loss": 3.7169, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00012497829099307158, |
| "loss": 3.4318, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00012475196304849884, |
| "loss": 3.3658, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001245256351039261, |
| "loss": 3.7082, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00012429930715935334, |
| "loss": 3.5838, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001240729792147806, |
| "loss": 2.8508, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00012384665127020784, |
| "loss": 3.2041, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001236203233256351, |
| "loss": 3.1151, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00012339399538106237, |
| "loss": 3.5128, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001231676674364896, |
| "loss": 3.6807, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00012294133949191687, |
| "loss": 3.6973, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00012271501154734413, |
| "loss": 3.7338, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00012248868360277136, |
| "loss": 3.3445, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00012226235565819863, |
| "loss": 3.6292, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00012203602771362588, |
| "loss": 3.6886, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00012180969976905311, |
| "loss": 2.6932, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00012158337182448036, |
| "loss": 3.4789, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00012135704387990761, |
| "loss": 3.4813, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00012113071593533487, |
| "loss": 3.6185, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00012090438799076212, |
| "loss": 3.5236, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00012067806004618937, |
| "loss": 3.6228, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00012045173210161662, |
| "loss": 3.414, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00012022540415704389, |
| "loss": 3.6309, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00011999907621247114, |
| "loss": 3.5011, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00011977274826789839, |
| "loss": 3.5267, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00011954642032332563, |
| "loss": 3.5156, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001193200923787529, |
| "loss": 3.3563, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00011909376443418015, |
| "loss": 3.6401, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001188674364896074, |
| "loss": 3.4525, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00011864110854503465, |
| "loss": 3.5465, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00011841478060046188, |
| "loss": 2.87, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00011818845265588914, |
| "loss": 2.6848, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001179621247113164, |
| "loss": 3.5327, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00011773579676674364, |
| "loss": 3.1602, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001175094688221709, |
| "loss": 3.1387, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00011728314087759816, |
| "loss": 3.1799, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001170568129330254, |
| "loss": 3.1679, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00011683048498845266, |
| "loss": 3.0457, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001166041570438799, |
| "loss": 3.5954, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00011637782909930717, |
| "loss": 3.008, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00011615150115473442, |
| "loss": 3.6305, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011592517321016167, |
| "loss": 3.056, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011569884526558892, |
| "loss": 3.0358, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011547251732101618, |
| "loss": 3.3535, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011524618937644343, |
| "loss": 3.5352, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011501986143187066, |
| "loss": 3.5372, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011479353348729791, |
| "loss": 2.9131, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00011456720554272516, |
| "loss": 3.4774, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00011434087759815243, |
| "loss": 3.5785, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00011411454965357968, |
| "loss": 3.0545, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00011388822170900693, |
| "loss": 3.6184, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00011366189376443417, |
| "loss": 3.536, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00011343556581986144, |
| "loss": 3.4078, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00011320923787528869, |
| "loss": 3.0253, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00011298290993071594, |
| "loss": 3.5612, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00011275658198614319, |
| "loss": 3.609, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00011253025404157045, |
| "loss": 3.48, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001123039260969977, |
| "loss": 3.0283, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00011207759815242495, |
| "loss": 2.9956, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001118512702078522, |
| "loss": 3.4985, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00011162494226327943, |
| "loss": 3.6792, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001113986143187067, |
| "loss": 3.0146, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00011117228637413395, |
| "loss": 3.3542, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001109459584295612, |
| "loss": 3.4283, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00011071963048498844, |
| "loss": 3.4591, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00011049330254041571, |
| "loss": 2.725, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00011026697459584296, |
| "loss": 2.8868, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001100406466512702, |
| "loss": 2.981, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00010981431870669746, |
| "loss": 3.4859, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00010958799076212472, |
| "loss": 3.2683, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00010936166281755197, |
| "loss": 3.5555, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00010913533487297922, |
| "loss": 2.9897, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00010890900692840647, |
| "loss": 3.227, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00010868267898383373, |
| "loss": 3.2891, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00010845635103926098, |
| "loss": 3.5478, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00010823002309468822, |
| "loss": 2.9625, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00010800369515011547, |
| "loss": 3.2351, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00010777736720554271, |
| "loss": 3.4041, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00010755103926096998, |
| "loss": 3.5005, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00010732471131639723, |
| "loss": 3.4141, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00010709838337182448, |
| "loss": 3.0204, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00010687205542725173, |
| "loss": 2.8737, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00010664572748267899, |
| "loss": 3.44, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00010641939953810624, |
| "loss": 3.4723, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.00010619307159353349, |
| "loss": 2.8187, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.00010596674364896074, |
| "loss": 3.5171, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.000105740415704388, |
| "loss": 3.346, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.00010551408775981525, |
| "loss": 3.3495, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.0001052877598152425, |
| "loss": 3.4582, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.00010506143187066975, |
| "loss": 3.4312, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.00010483510392609698, |
| "loss": 2.9329, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.00010460877598152425, |
| "loss": 3.4131, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.0001043824480369515, |
| "loss": 3.432, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.00010415612009237875, |
| "loss": 3.4073, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.000103929792147806, |
| "loss": 3.0032, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.00010370346420323326, |
| "loss": 3.3989, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00010347713625866051, |
| "loss": 2.5839, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00010325080831408776, |
| "loss": 3.5058, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00010302448036951501, |
| "loss": 2.8574, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00010279815242494227, |
| "loss": 3.2324, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00010257182448036952, |
| "loss": 3.4433, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00010234549653579677, |
| "loss": 3.2637, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00010211916859122402, |
| "loss": 3.4732, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00010189284064665128, |
| "loss": 3.215, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00010166651270207853, |
| "loss": 3.0964, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00010144018475750577, |
| "loss": 3.2448, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00010121385681293302, |
| "loss": 3.3717, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00010098752886836027, |
| "loss": 3.4094, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 0.00010076120092378753, |
| "loss": 2.8995, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 0.00010053487297921478, |
| "loss": 3.4194, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 0.00010030854503464203, |
| "loss": 2.8647, |
| "step": 815 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 0.00010008221709006928, |
| "loss": 2.649, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 9.985588914549654e-05, |
| "loss": 2.593, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 9.962956120092379e-05, |
| "loss": 2.7692, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.940323325635104e-05, |
| "loss": 2.923, |
| "step": 835 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.917690531177829e-05, |
| "loss": 2.7793, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.895057736720555e-05, |
| "loss": 2.3173, |
| "step": 845 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.87242494226328e-05, |
| "loss": 3.3614, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.849792147806005e-05, |
| "loss": 2.8285, |
| "step": 855 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 9.82715935334873e-05, |
| "loss": 3.1578, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.804526558891455e-05, |
| "loss": 3.2701, |
| "step": 865 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.78189376443418e-05, |
| "loss": 3.1701, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.759260969976905e-05, |
| "loss": 3.3629, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.73662817551963e-05, |
| "loss": 3.2969, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.713995381062355e-05, |
| "loss": 2.4782, |
| "step": 885 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 9.691362586605081e-05, |
| "loss": 3.3053, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.668729792147806e-05, |
| "loss": 3.3026, |
| "step": 895 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.646096997690531e-05, |
| "loss": 3.353, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.623464203233256e-05, |
| "loss": 2.7781, |
| "step": 905 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.600831408775982e-05, |
| "loss": 2.8703, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.578198614318707e-05, |
| "loss": 3.1661, |
| "step": 915 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 9.555565819861432e-05, |
| "loss": 2.7715, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 9.532933025404157e-05, |
| "loss": 2.8061, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 9.510300230946883e-05, |
| "loss": 3.2015, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 9.487667436489608e-05, |
| "loss": 2.4464, |
| "step": 935 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 9.465034642032332e-05, |
| "loss": 2.5126, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 9.442401847575057e-05, |
| "loss": 3.3578, |
| "step": 945 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 9.419769053117783e-05, |
| "loss": 3.2997, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.397136258660508e-05, |
| "loss": 3.0528, |
| "step": 955 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.374503464203233e-05, |
| "loss": 3.2615, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.351870669745958e-05, |
| "loss": 2.2974, |
| "step": 965 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.329237875288683e-05, |
| "loss": 2.4089, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.306605080831409e-05, |
| "loss": 3.3305, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.283972286374134e-05, |
| "loss": 2.9349, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 9.261339491916859e-05, |
| "loss": 3.026, |
| "step": 985 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 9.238706697459584e-05, |
| "loss": 3.2153, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 9.21607390300231e-05, |
| "loss": 2.754, |
| "step": 995 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 9.193441108545035e-05, |
| "loss": 3.3203, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 9.17080831408776e-05, |
| "loss": 3.3236, |
| "step": 1005 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 9.148175519630485e-05, |
| "loss": 3.2072, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 9.12554272517321e-05, |
| "loss": 3.271, |
| "step": 1015 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 9.102909930715935e-05, |
| "loss": 3.3537, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 9.08027713625866e-05, |
| "loss": 3.0024, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 9.057644341801385e-05, |
| "loss": 3.4014, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 9.035011547344111e-05, |
| "loss": 2.6294, |
| "step": 1035 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 9.012378752886836e-05, |
| "loss": 2.736, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 8.989745958429561e-05, |
| "loss": 2.1099, |
| "step": 1045 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.967113163972286e-05, |
| "loss": 3.1103, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.944480369515012e-05, |
| "loss": 2.7629, |
| "step": 1055 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.921847575057737e-05, |
| "loss": 2.9411, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.899214780600462e-05, |
| "loss": 2.6607, |
| "step": 1065 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.876581986143187e-05, |
| "loss": 1.6034, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.853949191685912e-05, |
| "loss": 2.7623, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.831316397228638e-05, |
| "loss": 3.3491, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.808683602771363e-05, |
| "loss": 3.3623, |
| "step": 1085 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.786050808314087e-05, |
| "loss": 3.3125, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.763418013856812e-05, |
| "loss": 3.2459, |
| "step": 1095 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.740785219399538e-05, |
| "loss": 2.9697, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.718152424942263e-05, |
| "loss": 2.063, |
| "step": 1105 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.695519630484988e-05, |
| "loss": 2.6997, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.672886836027713e-05, |
| "loss": 3.2616, |
| "step": 1115 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.65025404157044e-05, |
| "loss": 3.327, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.627621247113164e-05, |
| "loss": 2.6713, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.604988452655889e-05, |
| "loss": 2.8985, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 8.582355658198614e-05, |
| "loss": 3.3272, |
| "step": 1135 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.55972286374134e-05, |
| "loss": 3.2654, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.537090069284065e-05, |
| "loss": 3.2365, |
| "step": 1145 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.51445727482679e-05, |
| "loss": 2.3625, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.491824480369515e-05, |
| "loss": 3.3359, |
| "step": 1155 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.46919168591224e-05, |
| "loss": 3.2978, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 8.446558891454965e-05, |
| "loss": 3.2987, |
| "step": 1165 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.42392609699769e-05, |
| "loss": 3.1829, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.401293302540415e-05, |
| "loss": 2.9166, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.37866050808314e-05, |
| "loss": 2.8781, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.356027713625866e-05, |
| "loss": 3.311, |
| "step": 1185 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.333394919168591e-05, |
| "loss": 3.2549, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 8.310762124711316e-05, |
| "loss": 3.3449, |
| "step": 1195 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.288129330254041e-05, |
| "loss": 3.1834, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.265496535796768e-05, |
| "loss": 2.6987, |
| "step": 1205 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.242863741339492e-05, |
| "loss": 3.2679, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.220230946882217e-05, |
| "loss": 2.2549, |
| "step": 1215 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.197598152424942e-05, |
| "loss": 2.9217, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 8.174965357967669e-05, |
| "loss": 3.2378, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.152332563510394e-05, |
| "loss": 3.2458, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.129699769053119e-05, |
| "loss": 2.5979, |
| "step": 1235 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.107066974595842e-05, |
| "loss": 3.141, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.084434180138567e-05, |
| "loss": 3.2136, |
| "step": 1245 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.061801385681293e-05, |
| "loss": 2.5998, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.039168591224018e-05, |
| "loss": 1.9745, |
| "step": 1255 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 8.016535796766743e-05, |
| "loss": 3.2393, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 7.993903002309468e-05, |
| "loss": 1.985, |
| "step": 1265 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 7.971270207852195e-05, |
| "loss": 3.1701, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 7.94863741339492e-05, |
| "loss": 3.2786, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 7.926004618937644e-05, |
| "loss": 3.3514, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 7.90337182448037e-05, |
| "loss": 2.6101, |
| "step": 1285 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.880739030023096e-05, |
| "loss": 3.2465, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.85810623556582e-05, |
| "loss": 3.1456, |
| "step": 1295 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.835473441108546e-05, |
| "loss": 3.1528, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.81284064665127e-05, |
| "loss": 2.3822, |
| "step": 1305 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.790207852193997e-05, |
| "loss": 3.1659, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.76757505773672e-05, |
| "loss": 2.6111, |
| "step": 1315 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.744942263279445e-05, |
| "loss": 2.6443, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.72230946882217e-05, |
| "loss": 2.184, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.699676674364895e-05, |
| "loss": 3.2053, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.677043879907622e-05, |
| "loss": 3.323, |
| "step": 1335 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.654411085450346e-05, |
| "loss": 2.2732, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.631778290993071e-05, |
| "loss": 3.1413, |
| "step": 1345 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.609145496535796e-05, |
| "loss": 3.2421, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.586512702078523e-05, |
| "loss": 3.1688, |
| "step": 1355 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.563879907621248e-05, |
| "loss": 3.1884, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.541247113163973e-05, |
| "loss": 3.1239, |
| "step": 1365 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.518614318706697e-05, |
| "loss": 2.6056, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.495981524249424e-05, |
| "loss": 3.1442, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.473348729792149e-05, |
| "loss": 3.2654, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.450715935334874e-05, |
| "loss": 3.1673, |
| "step": 1385 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.428083140877597e-05, |
| "loss": 2.8148, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.405450346420322e-05, |
| "loss": 2.5837, |
| "step": 1395 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.382817551963049e-05, |
| "loss": 2.2148, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.360184757505773e-05, |
| "loss": 3.1805, |
| "step": 1405 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.337551963048498e-05, |
| "loss": 2.798, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.314919168591223e-05, |
| "loss": 3.2318, |
| "step": 1415 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.29228637413395e-05, |
| "loss": 2.8976, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.269653579676675e-05, |
| "loss": 3.0862, |
| "step": 1425 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.2470207852194e-05, |
| "loss": 2.5277, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.224387990762124e-05, |
| "loss": 2.0882, |
| "step": 1435 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.201755196304851e-05, |
| "loss": 3.0711, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.179122401847576e-05, |
| "loss": 2.6206, |
| "step": 1445 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.156489607390301e-05, |
| "loss": 2.0745, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.133856812933026e-05, |
| "loss": 2.5786, |
| "step": 1455 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.111224018475752e-05, |
| "loss": 2.6242, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.088591224018476e-05, |
| "loss": 2.506, |
| "step": 1465 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.0659584295612e-05, |
| "loss": 2.5967, |
| "step": 1470 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 7.043325635103925e-05, |
| "loss": 3.1206, |
| "step": 1475 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 7.02069284064665e-05, |
| "loss": 3.1769, |
| "step": 1480 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 6.998060046189377e-05, |
| "loss": 2.7589, |
| "step": 1485 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 6.975427251732102e-05, |
| "loss": 2.892, |
| "step": 1490 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 6.952794457274827e-05, |
| "loss": 2.7783, |
| "step": 1495 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 6.930161662817551e-05, |
| "loss": 3.0331, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.907528868360278e-05, |
| "loss": 3.1017, |
| "step": 1505 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.884896073903003e-05, |
| "loss": 2.909, |
| "step": 1510 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.862263279445728e-05, |
| "loss": 2.5939, |
| "step": 1515 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.839630484988453e-05, |
| "loss": 3.1691, |
| "step": 1520 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.816997690531178e-05, |
| "loss": 2.9704, |
| "step": 1525 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.794364896073903e-05, |
| "loss": 2.3905, |
| "step": 1530 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.771732101616629e-05, |
| "loss": 2.1867, |
| "step": 1535 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.749099307159354e-05, |
| "loss": 2.147, |
| "step": 1540 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.726466512702079e-05, |
| "loss": 2.7461, |
| "step": 1545 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.703833718244804e-05, |
| "loss": 3.1617, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.681200923787529e-05, |
| "loss": 3.1386, |
| "step": 1555 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.658568129330254e-05, |
| "loss": 2.7372, |
| "step": 1560 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.635935334872979e-05, |
| "loss": 3.1669, |
| "step": 1565 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.613302540415705e-05, |
| "loss": 2.5412, |
| "step": 1570 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.59066974595843e-05, |
| "loss": 2.7079, |
| "step": 1575 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.568036951501155e-05, |
| "loss": 2.5812, |
| "step": 1580 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.54540415704388e-05, |
| "loss": 2.575, |
| "step": 1585 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.522771362586606e-05, |
| "loss": 3.1151, |
| "step": 1590 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.50013856812933e-05, |
| "loss": 3.0427, |
| "step": 1595 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.477505773672056e-05, |
| "loss": 3.0496, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.454872979214781e-05, |
| "loss": 3.1504, |
| "step": 1605 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.432240184757506e-05, |
| "loss": 2.7546, |
| "step": 1610 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.40960739030023e-05, |
| "loss": 2.5357, |
| "step": 1615 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.386974595842957e-05, |
| "loss": 2.3116, |
| "step": 1620 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.364341801385682e-05, |
| "loss": 2.7059, |
| "step": 1625 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.341709006928407e-05, |
| "loss": 2.5225, |
| "step": 1630 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.319076212471132e-05, |
| "loss": 2.5516, |
| "step": 1635 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.296443418013857e-05, |
| "loss": 2.677, |
| "step": 1640 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.273810623556582e-05, |
| "loss": 1.9059, |
| "step": 1645 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.251177829099307e-05, |
| "loss": 1.3935, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.228545034642033e-05, |
| "loss": 2.6359, |
| "step": 1655 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.205912240184758e-05, |
| "loss": 2.7623, |
| "step": 1660 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.183279445727483e-05, |
| "loss": 3.0703, |
| "step": 1665 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.160646651270208e-05, |
| "loss": 2.7344, |
| "step": 1670 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.138013856812933e-05, |
| "loss": 3.0174, |
| "step": 1675 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.115381062355658e-05, |
| "loss": 2.5429, |
| "step": 1680 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 6.092748267898383e-05, |
| "loss": 3.092, |
| "step": 1685 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 6.070115473441109e-05, |
| "loss": 2.0041, |
| "step": 1690 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 6.047482678983834e-05, |
| "loss": 1.9942, |
| "step": 1695 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 6.0248498845265595e-05, |
| "loss": 3.0518, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 6.0022170900692844e-05, |
| "loss": 3.1302, |
| "step": 1705 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 5.979584295612009e-05, |
| "loss": 2.6788, |
| "step": 1710 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.956951501154734e-05, |
| "loss": 2.6504, |
| "step": 1715 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.934318706697459e-05, |
| "loss": 3.1506, |
| "step": 1720 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.911685912240185e-05, |
| "loss": 2.5277, |
| "step": 1725 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.88905311778291e-05, |
| "loss": 2.7044, |
| "step": 1730 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.8664203233256355e-05, |
| "loss": 3.0895, |
| "step": 1735 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.8437875288683604e-05, |
| "loss": 2.7463, |
| "step": 1740 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 5.8211547344110853e-05, |
| "loss": 2.5867, |
| "step": 1745 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 5.79852193995381e-05, |
| "loss": 2.5577, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 5.775889145496536e-05, |
| "loss": 2.0084, |
| "step": 1755 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 5.753256351039261e-05, |
| "loss": 2.656, |
| "step": 1760 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 5.7306235565819865e-05, |
| "loss": 3.0814, |
| "step": 1765 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 5.7079907621247114e-05, |
| "loss": 3.0512, |
| "step": 1770 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.685357967667437e-05, |
| "loss": 2.4837, |
| "step": 1775 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.662725173210162e-05, |
| "loss": 3.0879, |
| "step": 1780 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.640092378752886e-05, |
| "loss": 3.0925, |
| "step": 1785 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.617459584295612e-05, |
| "loss": 2.9465, |
| "step": 1790 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.594826789838337e-05, |
| "loss": 2.1648, |
| "step": 1795 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.5721939953810625e-05, |
| "loss": 2.4771, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.5495612009237874e-05, |
| "loss": 3.0954, |
| "step": 1805 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.526928406466513e-05, |
| "loss": 2.481, |
| "step": 1810 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.504295612009238e-05, |
| "loss": 3.0379, |
| "step": 1815 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.481662817551963e-05, |
| "loss": 2.5866, |
| "step": 1820 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.459030023094688e-05, |
| "loss": 2.4673, |
| "step": 1825 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.4363972286374135e-05, |
| "loss": 2.0428, |
| "step": 1830 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.4137644341801384e-05, |
| "loss": 2.9469, |
| "step": 1835 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.391131639722864e-05, |
| "loss": 3.065, |
| "step": 1840 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.368498845265589e-05, |
| "loss": 2.3777, |
| "step": 1845 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.3458660508083146e-05, |
| "loss": 2.4645, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.3232332563510396e-05, |
| "loss": 2.9935, |
| "step": 1855 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.300600461893764e-05, |
| "loss": 3.0377, |
| "step": 1860 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.2779676674364895e-05, |
| "loss": 2.2265, |
| "step": 1865 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.2553348729792144e-05, |
| "loss": 2.0621, |
| "step": 1870 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.23270207852194e-05, |
| "loss": 2.285, |
| "step": 1875 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.210069284064665e-05, |
| "loss": 2.605, |
| "step": 1880 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.1874364896073906e-05, |
| "loss": 2.4744, |
| "step": 1885 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.1648036951501155e-05, |
| "loss": 2.3932, |
| "step": 1890 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.1421709006928405e-05, |
| "loss": 3.0259, |
| "step": 1895 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.1195381062355654e-05, |
| "loss": 2.9768, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.096905311778291e-05, |
| "loss": 2.4951, |
| "step": 1905 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.074272517321016e-05, |
| "loss": 2.9348, |
| "step": 1910 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.0516397228637416e-05, |
| "loss": 3.0326, |
| "step": 1915 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.0290069284064666e-05, |
| "loss": 2.544, |
| "step": 1920 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 5.006374133949192e-05, |
| "loss": 2.9794, |
| "step": 1925 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.983741339491917e-05, |
| "loss": 2.2389, |
| "step": 1930 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.961108545034642e-05, |
| "loss": 2.5411, |
| "step": 1935 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.938475750577367e-05, |
| "loss": 2.1246, |
| "step": 1940 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.915842956120092e-05, |
| "loss": 2.549, |
| "step": 1945 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.8932101616628176e-05, |
| "loss": 2.9752, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.8705773672055425e-05, |
| "loss": 2.2107, |
| "step": 1955 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.847944572748268e-05, |
| "loss": 2.9383, |
| "step": 1960 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.825311778290993e-05, |
| "loss": 2.4591, |
| "step": 1965 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.802678983833718e-05, |
| "loss": 2.5564, |
| "step": 1970 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.780046189376443e-05, |
| "loss": 2.6209, |
| "step": 1975 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.7574133949191686e-05, |
| "loss": 2.4038, |
| "step": 1980 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.7347806004618936e-05, |
| "loss": 2.5304, |
| "step": 1985 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.712147806004619e-05, |
| "loss": 2.6632, |
| "step": 1990 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.689515011547344e-05, |
| "loss": 1.937, |
| "step": 1995 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.66688221709007e-05, |
| "loss": 2.4658, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.644249422632795e-05, |
| "loss": 2.5939, |
| "step": 2005 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.6216166281755197e-05, |
| "loss": 2.618, |
| "step": 2010 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.5989838337182446e-05, |
| "loss": 3.0384, |
| "step": 2015 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.57635103926097e-05, |
| "loss": 2.9713, |
| "step": 2020 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.553718244803695e-05, |
| "loss": 3.0171, |
| "step": 2025 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.53108545034642e-05, |
| "loss": 3.0126, |
| "step": 2030 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.508452655889146e-05, |
| "loss": 2.9481, |
| "step": 2035 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.485819861431871e-05, |
| "loss": 2.4466, |
| "step": 2040 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.4631870669745956e-05, |
| "loss": 2.0695, |
| "step": 2045 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.4405542725173206e-05, |
| "loss": 3.0076, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.417921478060046e-05, |
| "loss": 3.0097, |
| "step": 2055 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.395288683602771e-05, |
| "loss": 2.4857, |
| "step": 2060 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.372655889145497e-05, |
| "loss": 3.0361, |
| "step": 2065 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.350023094688222e-05, |
| "loss": 1.9166, |
| "step": 2070 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.327390300230947e-05, |
| "loss": 2.9817, |
| "step": 2075 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.304757505773672e-05, |
| "loss": 2.9798, |
| "step": 2080 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.282124711316397e-05, |
| "loss": 2.4675, |
| "step": 2085 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.259491916859122e-05, |
| "loss": 3.0161, |
| "step": 2090 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.236859122401848e-05, |
| "loss": 1.9453, |
| "step": 2095 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.214226327944573e-05, |
| "loss": 3.0623, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.1915935334872984e-05, |
| "loss": 3.0624, |
| "step": 2105 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.168960739030023e-05, |
| "loss": 2.4797, |
| "step": 2110 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.146327944572748e-05, |
| "loss": 2.0303, |
| "step": 2115 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.123695150115473e-05, |
| "loss": 2.9842, |
| "step": 2120 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.101062355658198e-05, |
| "loss": 2.5711, |
| "step": 2125 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.078429561200924e-05, |
| "loss": 2.409, |
| "step": 2130 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.055796766743649e-05, |
| "loss": 2.473, |
| "step": 2135 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.033163972286374e-05, |
| "loss": 1.9119, |
| "step": 2140 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.010531177829099e-05, |
| "loss": 2.9008, |
| "step": 2145 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.987898383371825e-05, |
| "loss": 3.032, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.96526558891455e-05, |
| "loss": 2.3587, |
| "step": 2155 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.942632794457275e-05, |
| "loss": 2.9925, |
| "step": 2160 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.92e-05, |
| "loss": 2.9025, |
| "step": 2165 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.8973672055427254e-05, |
| "loss": 3.0899, |
| "step": 2170 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.87473441108545e-05, |
| "loss": 2.9319, |
| "step": 2175 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.852101616628176e-05, |
| "loss": 2.453, |
| "step": 2180 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.829468822170901e-05, |
| "loss": 2.0663, |
| "step": 2185 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.8068360277136265e-05, |
| "loss": 2.4464, |
| "step": 2190 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.784203233256351e-05, |
| "loss": 2.5029, |
| "step": 2195 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.761570438799076e-05, |
| "loss": 2.4315, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.738937644341801e-05, |
| "loss": 2.9723, |
| "step": 2205 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.716304849884526e-05, |
| "loss": 2.9545, |
| "step": 2210 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.693672055427252e-05, |
| "loss": 2.4394, |
| "step": 2215 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.671039260969977e-05, |
| "loss": 2.8852, |
| "step": 2220 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.6484064665127025e-05, |
| "loss": 2.5524, |
| "step": 2225 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.6257736720554274e-05, |
| "loss": 2.5424, |
| "step": 2230 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.6031408775981524e-05, |
| "loss": 2.4416, |
| "step": 2235 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.580508083140877e-05, |
| "loss": 2.3582, |
| "step": 2240 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.557875288683603e-05, |
| "loss": 2.9595, |
| "step": 2245 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.535242494226328e-05, |
| "loss": 3.0225, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.5126096997690535e-05, |
| "loss": 2.0518, |
| "step": 2255 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.4899769053117784e-05, |
| "loss": 2.9843, |
| "step": 2260 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.467344110854504e-05, |
| "loss": 2.3971, |
| "step": 2265 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.444711316397228e-05, |
| "loss": 2.9604, |
| "step": 2270 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.422078521939954e-05, |
| "loss": 1.526, |
| "step": 2275 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.399445727482679e-05, |
| "loss": 2.5448, |
| "step": 2280 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.376812933025404e-05, |
| "loss": 2.5325, |
| "step": 2285 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.3541801385681295e-05, |
| "loss": 2.0032, |
| "step": 2290 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.3315473441108544e-05, |
| "loss": 2.447, |
| "step": 2295 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.3089145496535794e-05, |
| "loss": 2.9568, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.286281755196305e-05, |
| "loss": 2.5495, |
| "step": 2305 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.26364896073903e-05, |
| "loss": 2.4766, |
| "step": 2310 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.2410161662817556e-05, |
| "loss": 2.9513, |
| "step": 2315 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.2183833718244805e-05, |
| "loss": 1.9875, |
| "step": 2320 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.1957505773672054e-05, |
| "loss": 2.0134, |
| "step": 2325 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.173117782909931e-05, |
| "loss": 2.5515, |
| "step": 2330 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.150484988452656e-05, |
| "loss": 2.9005, |
| "step": 2335 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.127852193995381e-05, |
| "loss": 2.8825, |
| "step": 2340 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.1052193995381066e-05, |
| "loss": 1.8196, |
| "step": 2345 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.0825866050808315e-05, |
| "loss": 2.8672, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.0599538106235565e-05, |
| "loss": 2.5313, |
| "step": 2355 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.0373210161662818e-05, |
| "loss": 2.4912, |
| "step": 2360 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.014688221709007e-05, |
| "loss": 2.851, |
| "step": 2365 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.9920554272517323e-05, |
| "loss": 2.9382, |
| "step": 2370 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.9694226327944573e-05, |
| "loss": 2.8546, |
| "step": 2375 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.9467898383371826e-05, |
| "loss": 2.9912, |
| "step": 2380 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.924157043879908e-05, |
| "loss": 2.4831, |
| "step": 2385 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.901524249422633e-05, |
| "loss": 2.96, |
| "step": 2390 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.8788914549653577e-05, |
| "loss": 2.916, |
| "step": 2395 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.856258660508083e-05, |
| "loss": 2.9423, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.8336258660508083e-05, |
| "loss": 2.3475, |
| "step": 2405 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.8109930715935336e-05, |
| "loss": 2.4757, |
| "step": 2410 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.7883602771362585e-05, |
| "loss": 2.497, |
| "step": 2415 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.7657274826789838e-05, |
| "loss": 2.8617, |
| "step": 2420 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.743094688221709e-05, |
| "loss": 2.5424, |
| "step": 2425 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.720461893764434e-05, |
| "loss": 1.9343, |
| "step": 2430 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.6978290993071593e-05, |
| "loss": 2.3538, |
| "step": 2435 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.6751963048498846e-05, |
| "loss": 2.457, |
| "step": 2440 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.65256351039261e-05, |
| "loss": 1.8267, |
| "step": 2445 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.629930715935335e-05, |
| "loss": 2.8956, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.60729792147806e-05, |
| "loss": 2.4804, |
| "step": 2455 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.5846651270207854e-05, |
| "loss": 0.9658, |
| "step": 2460 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.5620323325635107e-05, |
| "loss": 2.5179, |
| "step": 2465 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 2.5393995381062353e-05, |
| "loss": 2.0064, |
| "step": 2470 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 2.5167667436489606e-05, |
| "loss": 2.9194, |
| "step": 2475 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 2.494133949191686e-05, |
| "loss": 2.9696, |
| "step": 2480 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 2.471501154734411e-05, |
| "loss": 2.3919, |
| "step": 2485 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 2.448868360277136e-05, |
| "loss": 2.9989, |
| "step": 2490 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 2.4262355658198614e-05, |
| "loss": 1.9785, |
| "step": 2495 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 2.4036027713625867e-05, |
| "loss": 2.4736, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 2.3809699769053116e-05, |
| "loss": 2.9508, |
| "step": 2505 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 2.358337182448037e-05, |
| "loss": 3.0043, |
| "step": 2510 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 2.3357043879907622e-05, |
| "loss": 2.3829, |
| "step": 2515 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 2.3130715935334875e-05, |
| "loss": 2.4771, |
| "step": 2520 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 2.2904387990762124e-05, |
| "loss": 2.4709, |
| "step": 2525 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 2.2678060046189377e-05, |
| "loss": 2.8935, |
| "step": 2530 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 2.245173210161663e-05, |
| "loss": 2.9177, |
| "step": 2535 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 2.2225404157043883e-05, |
| "loss": 2.9019, |
| "step": 2540 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 2.1999076212471132e-05, |
| "loss": 2.944, |
| "step": 2545 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 2.1772748267898385e-05, |
| "loss": 2.4892, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 2.1546420323325638e-05, |
| "loss": 2.8733, |
| "step": 2555 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 2.1320092378752887e-05, |
| "loss": 1.6182, |
| "step": 2560 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 2.1093764434180137e-05, |
| "loss": 2.8951, |
| "step": 2565 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 2.086743648960739e-05, |
| "loss": 2.903, |
| "step": 2570 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 2.0641108545034642e-05, |
| "loss": 2.3775, |
| "step": 2575 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 2.0414780600461892e-05, |
| "loss": 2.5025, |
| "step": 2580 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 2.0188452655889145e-05, |
| "loss": 2.3309, |
| "step": 2585 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.9962124711316398e-05, |
| "loss": 2.5212, |
| "step": 2590 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.973579676674365e-05, |
| "loss": 2.8385, |
| "step": 2595 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.95094688221709e-05, |
| "loss": 2.3192, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.9283140877598153e-05, |
| "loss": 2.8028, |
| "step": 2605 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.9056812933025406e-05, |
| "loss": 2.9434, |
| "step": 2610 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.883048498845266e-05, |
| "loss": 2.9547, |
| "step": 2615 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.8604157043879908e-05, |
| "loss": 1.8949, |
| "step": 2620 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.837782909930716e-05, |
| "loss": 1.8706, |
| "step": 2625 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.8151501154734413e-05, |
| "loss": 2.424, |
| "step": 2630 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.7925173210161666e-05, |
| "loss": 2.3591, |
| "step": 2635 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.7698845265588912e-05, |
| "loss": 2.8878, |
| "step": 2640 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.7472517321016165e-05, |
| "loss": 2.8651, |
| "step": 2645 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.7246189376443418e-05, |
| "loss": 2.8761, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 1.701986143187067e-05, |
| "loss": 2.833, |
| "step": 2655 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 1.679353348729792e-05, |
| "loss": 2.395, |
| "step": 2660 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 1.6567205542725173e-05, |
| "loss": 2.9417, |
| "step": 2665 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 1.6340877598152426e-05, |
| "loss": 2.9002, |
| "step": 2670 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 1.611454965357968e-05, |
| "loss": 2.3858, |
| "step": 2675 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 1.588822170900693e-05, |
| "loss": 2.4055, |
| "step": 2680 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.566189376443418e-05, |
| "loss": 2.8661, |
| "step": 2685 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.543556581986143e-05, |
| "loss": 2.4005, |
| "step": 2690 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.5209237875288685e-05, |
| "loss": 2.8337, |
| "step": 2695 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.4982909930715935e-05, |
| "loss": 1.8415, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.4756581986143188e-05, |
| "loss": 2.8588, |
| "step": 2705 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.4530254041570439e-05, |
| "loss": 2.8498, |
| "step": 2710 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 1.430392609699769e-05, |
| "loss": 2.3665, |
| "step": 2715 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 1.4077598152424943e-05, |
| "loss": 2.9143, |
| "step": 2720 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 1.3851270207852194e-05, |
| "loss": 2.8927, |
| "step": 2725 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 1.3624942263279447e-05, |
| "loss": 2.2419, |
| "step": 2730 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 1.3398614318706696e-05, |
| "loss": 2.8982, |
| "step": 2735 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 1.3172286374133949e-05, |
| "loss": 2.3598, |
| "step": 2740 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 1.29459584295612e-05, |
| "loss": 1.9034, |
| "step": 2745 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 1.2719630484988453e-05, |
| "loss": 2.8746, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 1.2493302540415704e-05, |
| "loss": 2.5453, |
| "step": 2755 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 1.2266974595842957e-05, |
| "loss": 2.3748, |
| "step": 2760 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 1.2040646651270208e-05, |
| "loss": 2.3395, |
| "step": 2765 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 1.1814318706697461e-05, |
| "loss": 1.8988, |
| "step": 2770 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.1587990762124712e-05, |
| "loss": 2.832, |
| "step": 2775 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.1361662817551963e-05, |
| "loss": 2.9164, |
| "step": 2780 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.1135334872979214e-05, |
| "loss": 1.3357, |
| "step": 2785 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.0909006928406465e-05, |
| "loss": 2.3844, |
| "step": 2790 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.0682678983833718e-05, |
| "loss": 2.3568, |
| "step": 2795 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.045635103926097e-05, |
| "loss": 2.9659, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 1.0230023094688222e-05, |
| "loss": 2.3948, |
| "step": 2805 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 1.0003695150115473e-05, |
| "loss": 2.3828, |
| "step": 2810 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 9.777367205542726e-06, |
| "loss": 2.3914, |
| "step": 2815 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 9.551039260969976e-06, |
| "loss": 2.4423, |
| "step": 2820 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 9.324711316397229e-06, |
| "loss": 2.4067, |
| "step": 2825 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 9.09838337182448e-06, |
| "loss": 2.8074, |
| "step": 2830 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 8.872055427251733e-06, |
| "loss": 2.9147, |
| "step": 2835 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 8.645727482678984e-06, |
| "loss": 2.4484, |
| "step": 2840 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 8.419399538106235e-06, |
| "loss": 2.8331, |
| "step": 2845 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 8.193071593533488e-06, |
| "loss": 2.3605, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 7.966743648960739e-06, |
| "loss": 2.0343, |
| "step": 2855 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 7.74041570438799e-06, |
| "loss": 2.8557, |
| "step": 2860 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 7.514087759815243e-06, |
| "loss": 2.4906, |
| "step": 2865 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 7.287759815242494e-06, |
| "loss": 2.2705, |
| "step": 2870 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 7.061431870669746e-06, |
| "loss": 2.1079, |
| "step": 2875 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.835103926096998e-06, |
| "loss": 2.9184, |
| "step": 2880 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.60877598152425e-06, |
| "loss": 2.7964, |
| "step": 2885 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.382448036951501e-06, |
| "loss": 2.2701, |
| "step": 2890 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 6.156120092378753e-06, |
| "loss": 2.4188, |
| "step": 2895 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 5.929792147806005e-06, |
| "loss": 2.4344, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 5.703464203233256e-06, |
| "loss": 2.0036, |
| "step": 2905 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 5.4771362586605075e-06, |
| "loss": 2.3556, |
| "step": 2910 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 5.2508083140877595e-06, |
| "loss": 2.8956, |
| "step": 2915 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 5.0244803695150115e-06, |
| "loss": 2.9212, |
| "step": 2920 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.7981524249422635e-06, |
| "loss": 2.325, |
| "step": 2925 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.571824480369515e-06, |
| "loss": 2.9001, |
| "step": 2930 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.345496535796767e-06, |
| "loss": 2.0078, |
| "step": 2935 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.119168591224019e-06, |
| "loss": 1.8776, |
| "step": 2940 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.892840646651271e-06, |
| "loss": 2.315, |
| "step": 2945 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.666512702078522e-06, |
| "loss": 1.9461, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.4401847575057737e-06, |
| "loss": 2.8444, |
| "step": 2955 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.2138568129330253e-06, |
| "loss": 2.7964, |
| "step": 2960 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.987528868360277e-06, |
| "loss": 2.2962, |
| "step": 2965 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.761200923787529e-06, |
| "loss": 2.9162, |
| "step": 2970 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.5348729792147804e-06, |
| "loss": 2.3761, |
| "step": 2975 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.3085450346420324e-06, |
| "loss": 2.7675, |
| "step": 2980 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.082217090069284e-06, |
| "loss": 2.3303, |
| "step": 2985 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 1.8558891454965358e-06, |
| "loss": 2.3253, |
| "step": 2990 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 1.6295612009237876e-06, |
| "loss": 2.8782, |
| "step": 2995 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 1.4032332563510394e-06, |
| "loss": 2.7999, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 1.1769053117782911e-06, |
| "loss": 2.7688, |
| "step": 3005 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 9.505773672055428e-07, |
| "loss": 2.8891, |
| "step": 3010 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 7.242494226327945e-07, |
| "loss": 2.4505, |
| "step": 3015 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.979214780600462e-07, |
| "loss": 2.8581, |
| "step": 3020 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 2.715935334872979e-07, |
| "loss": 2.3888, |
| "step": 3025 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.5265588914549656e-08, |
| "loss": 2.2638, |
| "step": 3030 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 3031, |
| "total_flos": 5629783610228736.0, |
| "train_loss": 2.9040869920020134, |
| "train_runtime": 3297.2859, |
| "train_samples_per_second": 0.919, |
| "train_steps_per_second": 0.919 |
| } |
| ], |
| "max_steps": 3031, |
| "num_train_epochs": 1, |
| "total_flos": 5629783610228736.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|