| { | |
| "best_metric": 0.21771365404129028, | |
| "best_model_checkpoint": "output/eminem/checkpoint-1812", | |
| "epoch": 4.0, | |
| "global_step": 1812, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.438570471456781e-07, | |
| "loss": 0.5631, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.3766395364652078e-06, | |
| "loss": 0.4984, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.8898866416840713e-06, | |
| "loss": 0.5528, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.4829840193536984e-06, | |
| "loss": 0.6369, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.15522174732073e-06, | |
| "loss": 0.5848, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.9057951744922e-06, | |
| "loss": 0.5654, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.733805883976431e-06, | |
| "loss": 0.5758, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5.638262768466482e-06, | |
| "loss": 0.5143, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.618083216572411e-06, | |
| "loss": 0.5341, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 7.67209440867359e-06, | |
| "loss": 0.4776, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.799034720759786e-06, | |
| "loss": 0.5263, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.99755523455637e-06, | |
| "loss": 0.5323, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.1266221352146829e-05, | |
| "loss": 0.5985, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.2603514513152563e-05, | |
| "loss": 0.5842, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.4007834012403921e-05, | |
| "loss": 0.652, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.5477498915945647e-05, | |
| "loss": 0.5753, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.7010750073074952e-05, | |
| "loss": 0.5635, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8605752221990662e-05, | |
| "loss": 0.4981, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.0260596186565352e-05, | |
| "loss": 0.5425, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.197330116156359e-05, | |
| "loss": 0.4698, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.374181708363339e-05, | |
| "loss": 0.5, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.5564027085163642e-05, | |
| "loss": 0.5386, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.7437750028125014e-05, | |
| "loss": 0.5554, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.9360743114840313e-05, | |
| "loss": 0.5243, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.133070457254414e-05, | |
| "loss": 0.5746, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.334527640855775e-05, | |
| "loss": 0.5865, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.5402047232733794e-05, | |
| "loss": 0.5271, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.74985551438323e-05, | |
| "loss": 0.5387, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.963229067635792e-05, | |
| "loss": 0.5834, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.180069980431271e-05, | |
| "loss": 0.5346, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.4001186998313e-05, | |
| "loss": 0.5753, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.623111833235809e-05, | |
| "loss": 0.5956, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.848782463657661e-05, | |
| "loss": 0.556, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.076860469215913e-05, | |
| "loss": 0.6621, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.307072846463573e-05, | |
| "loss": 0.5567, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.5391440371663775e-05, | |
| "loss": 0.5614, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.7727962581384594e-05, | |
| "loss": 0.541, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.007749833742374e-05, | |
| "loss": 0.6, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.243723530652418e-05, | |
| "loss": 0.5782, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.480434894484185e-05, | |
| "loss": 0.6341, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.717600587885714e-05, | |
| "loss": 0.6118, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.95493672968377e-05, | |
| "loss": 0.5678, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.192159234682814e-05, | |
| "loss": 0.576, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.428984153708189e-05, | |
| "loss": 0.4995, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 7.66512801348454e-05, | |
| "loss": 0.5505, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.90030815594751e-05, | |
| "loss": 0.5138, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.134243076576756e-05, | |
| "loss": 0.642, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.36665276135011e-05, | |
| "loss": 0.5566, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.597259021913672e-05, | |
| "loss": 0.6086, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.82578582856489e-05, | |
| "loss": 0.5626, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.051959640653713e-05, | |
| "loss": 0.5904, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.275509734003259e-05, | |
| "loss": 0.5694, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.49616852496036e-05, | |
| "loss": 0.5298, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.713671890685142e-05, | |
| "loss": 0.5779, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.927759485299755e-05, | |
| "loss": 0.5908, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00010138175051516152, | |
| "loss": 0.5923, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00010344666727368174, | |
| "loss": 0.5174, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00010546987347685163, | |
| "loss": 0.5551, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00010744894739941056, | |
| "loss": 0.5551, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00010938152014129127, | |
| "loss": 0.5935, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00011126527846313643, | |
| "loss": 0.6027, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00011309796755517486, | |
| "loss": 0.6107, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00011487739373618269, | |
| "loss": 0.6027, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00011660142707925222, | |
| "loss": 0.6548, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00011826800396126505, | |
| "loss": 0.576, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00011987512953299384, | |
| "loss": 0.6334, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00012142088010688381, | |
| "loss": 0.5323, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00012290340545966178, | |
| "loss": 0.6278, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00012432093104699756, | |
| "loss": 0.7043, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00012567176012759175, | |
| "loss": 0.5754, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00012695427579413395, | |
| "loss": 0.5657, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00012816694290869102, | |
| "loss": 0.654, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00012930830994023598, | |
| "loss": 0.687, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00013037701070208566, | |
| "loss": 0.7228, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00013137176598719397, | |
| "loss": 0.583, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00013229138509933167, | |
| "loss": 0.6198, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00013313476727831326, | |
| "loss": 0.6863, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00013390090301758438, | |
| "loss": 0.6002, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00013458887527256983, | |
| "loss": 0.5305, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0001351978605583551, | |
| "loss": 0.647, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00013572712993537557, | |
| "loss": 0.6578, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001361760498819379, | |
| "loss": 0.694, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00013654408305253043, | |
| "loss": 0.6405, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00013683078892100958, | |
| "loss": 0.598, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00013703582430790024, | |
| "loss": 0.6474, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00013715894379117126, | |
| "loss": 0.6811, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0001372, | |
| "loss": 0.6149, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00013715894379117126, | |
| "loss": 0.6465, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00013703582430790037, | |
| "loss": 0.6784, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00013683078892100977, | |
| "loss": 0.6619, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.36211052536964417, | |
| "eval_runtime": 7.9758, | |
| "eval_samples_per_second": 81.371, | |
| "eval_steps_per_second": 10.281, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00013654408305253043, | |
| "loss": 0.7554, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0001361760498819382, | |
| "loss": 0.5685, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00013572712993537557, | |
| "loss": 0.4916, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0001351978605583551, | |
| "loss": 0.6418, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0001345888752725698, | |
| "loss": 0.5455, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00013390090301758435, | |
| "loss": 0.5476, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0001331347672783139, | |
| "loss": 0.5503, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00013229138509933164, | |
| "loss": 0.5999, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00013137176598719473, | |
| "loss": 0.5831, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00013037701070208564, | |
| "loss": 0.5574, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00012930830994023595, | |
| "loss": 0.6027, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00012816694290869194, | |
| "loss": 0.6368, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00012695427579413393, | |
| "loss": 0.6659, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0001256717601275928, | |
| "loss": 0.5791, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00012432093104699753, | |
| "loss": 0.6579, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00012290340545966172, | |
| "loss": 0.5917, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00012142088010688503, | |
| "loss": 0.5659, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0001198751295329938, | |
| "loss": 0.5619, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00011826800396126635, | |
| "loss": 0.5993, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00011660142707925218, | |
| "loss": 0.5983, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00011487739373618408, | |
| "loss": 0.5408, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00011309796755517628, | |
| "loss": 0.5652, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00011126527846313638, | |
| "loss": 0.6357, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0001093815201412928, | |
| "loss": 0.6156, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010744894739941049, | |
| "loss": 0.5459, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00010546987347685321, | |
| "loss": 0.5747, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00010344666727368336, | |
| "loss": 0.6251, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00010138175051516146, | |
| "loss": 0.6133, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.927759485299923e-05, | |
| "loss": 0.6489, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 9.713671890685137e-05, | |
| "loss": 0.6098, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.496168524960355e-05, | |
| "loss": 0.6389, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.275509734003435e-05, | |
| "loss": 0.5117, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.051959640653706e-05, | |
| "loss": 0.5879, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 8.825785828565071e-05, | |
| "loss": 0.5736, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.597259021913666e-05, | |
| "loss": 0.5553, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 8.366652761350294e-05, | |
| "loss": 0.6493, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 8.13424307657675e-05, | |
| "loss": 0.5273, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.900308155947505e-05, | |
| "loss": 0.5613, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 7.665128013484727e-05, | |
| "loss": 0.55, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.428984153708183e-05, | |
| "loss": 0.5884, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.192159234683002e-05, | |
| "loss": 0.5982, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 6.954936729683765e-05, | |
| "loss": 0.5926, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.717600587885708e-05, | |
| "loss": 0.5806, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.480434894484373e-05, | |
| "loss": 0.5951, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.243723530652412e-05, | |
| "loss": 0.5371, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.007749833742561e-05, | |
| "loss": 0.607, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.772796258138646e-05, | |
| "loss": 0.5796, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.5391440371663714e-05, | |
| "loss": 0.6185, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.307072846463756e-05, | |
| "loss": 0.5427, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 5.076860469215906e-05, | |
| "loss": 0.5822, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.8487824636578415e-05, | |
| "loss": 0.6119, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.623111833235802e-05, | |
| "loss": 0.4964, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.400118699831475e-05, | |
| "loss": 0.5858, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.180069980431445e-05, | |
| "loss": 0.5173, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.963229067635786e-05, | |
| "loss": 0.4972, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.749855514383397e-05, | |
| "loss": 0.5399, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.540204723273373e-05, | |
| "loss": 0.5664, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.33452764085577e-05, | |
| "loss": 0.5974, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.133070457254572e-05, | |
| "loss": 0.5232, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.9360743114840252e-05, | |
| "loss": 0.6258, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.7437750028126522e-05, | |
| "loss": 0.5125, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.5564027085163588e-05, | |
| "loss": 0.6473, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.374181708363334e-05, | |
| "loss": 0.5284, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.197330116156497e-05, | |
| "loss": 0.5848, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.0260596186565298e-05, | |
| "loss": 0.535, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.860575222199195e-05, | |
| "loss": 0.5386, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.7010750073074908e-05, | |
| "loss": 0.5661, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.5477498915946836e-05, | |
| "loss": 0.5236, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.4007834012403882e-05, | |
| "loss": 0.5898, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2603514513152518e-05, | |
| "loss": 0.6072, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.1266221352147866e-05, | |
| "loss": 0.4956, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.99755523455634e-06, | |
| "loss": 0.5137, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.799034720760707e-06, | |
| "loss": 0.4718, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.672094408674458e-06, | |
| "loss": 0.5598, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.61808321657238e-06, | |
| "loss": 0.5329, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.638262768467229e-06, | |
| "loss": 0.5635, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.733805883976409e-06, | |
| "loss": 0.6177, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.905795174492832e-06, | |
| "loss": 0.5325, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.1552217473212937e-06, | |
| "loss": 0.6092, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.4829840193536756e-06, | |
| "loss": 0.5838, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8898866416845054e-06, | |
| "loss": 0.5179, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.3766395364651925e-06, | |
| "loss": 0.587, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.438570471459905e-07, | |
| "loss": 0.5222, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.920572031214825e-07, | |
| "loss": 0.6116, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.2166109966502293e-07, | |
| "loss": 0.4959, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.329923938872171e-07, | |
| "loss": 0.5544, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.627691732664299e-08, | |
| "loss": 0.571, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.6424056362860595e-09, | |
| "loss": 0.5854, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.911834568653918e-08, | |
| "loss": 0.5492, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.986359402705409e-07, | |
| "loss": 0.5403, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.2002819045225533e-07, | |
| "loss": 0.5372, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.35109472274780273, | |
| "eval_runtime": 7.9799, | |
| "eval_samples_per_second": 81.33, | |
| "eval_steps_per_second": 10.276, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.30995090154817e-06, | |
| "loss": 0.4526, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.0541855162378925e-05, | |
| "loss": 0.5261, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.1843560639183314e-05, | |
| "loss": 0.5046, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.321350233836306e-05, | |
| "loss": 0.6349, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.4650033228364816e-05, | |
| "loss": 0.5519, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.615142621984976e-05, | |
| "loss": 0.6092, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.7715876242113545e-05, | |
| "loss": 0.5618, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.9341502413255598e-05, | |
| "loss": 0.5784, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.10263503014928e-05, | |
| "loss": 0.6022, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.2768394274904872e-05, | |
| "loss": 0.5783, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.4565539936783126e-05, | |
| "loss": 0.551, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.6415626643638117e-05, | |
| "loss": 0.524, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8316430102875185e-05, | |
| "loss": 0.5542, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.026566504696925e-05, | |
| "loss": 0.5175, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.226098798097204e-05, | |
| "loss": 0.5524, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.4300000000000576e-05, | |
| "loss": 0.6188, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.638024967335875e-05, | |
| "loss": 0.5017, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.8499235991807596e-05, | |
| "loss": 0.5877, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.065441137443541e-05, | |
| "loss": 0.4836, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.2843184731518815e-05, | |
| "loss": 0.4962, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.506292457970005e-05, | |
| "loss": 0.5619, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.731096220573011e-05, | |
| "loss": 0.5061, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.9584594874953847e-05, | |
| "loss": 0.5603, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5.188108908072324e-05, | |
| "loss": 0.6256, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.419768383078854e-05, | |
| "loss": 0.5088, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.653159396673597e-05, | |
| "loss": 0.5927, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.888001351248702e-05, | |
| "loss": 0.5919, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 6.124011904782824e-05, | |
| "loss": 0.5993, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 6.36090731028942e-05, | |
| "loss": 0.5514, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 6.598402756957074e-05, | |
| "loss": 0.6111, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 6.836212712565718e-05, | |
| "loss": 0.5716, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.074051266773294e-05, | |
| "loss": 0.5673, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.311632474854093e-05, | |
| "loss": 0.6117, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.54867070148039e-05, | |
| "loss": 0.5824, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.784880964131884e-05, | |
| "loss": 0.5873, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 8.019979275719533e-05, | |
| "loss": 0.6133, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 8.253682986012616e-05, | |
| "loss": 0.6222, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.485711121459308e-05, | |
| "loss": 0.5468, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.715784722991633e-05, | |
| "loss": 0.5227, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.943627181406567e-05, | |
| "loss": 0.5709, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.168964569924647e-05, | |
| "loss": 0.6087, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.391525973521692e-05, | |
| "loss": 0.5605, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.611043814639816e-05, | |
| "loss": 0.6505, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.827254174886847e-05, | |
| "loss": 0.5743, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 0.00010039897112335537, | |
| "loss": 0.5898, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.000102487169740429, | |
| "loss": 0.5651, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00010453462703412173, | |
| "loss": 0.5988, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.00010653888142028633, | |
| "loss": 0.6178, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.00010849752325608153, | |
| "loss": 0.5959, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.00011040819773698719, | |
| "loss": 0.6657, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.00011226860772790607, | |
| "loss": 0.6, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.0001140765165249296, | |
| "loss": 0.554, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.00011582975054444297, | |
| "loss": 0.5723, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.00011752620193634387, | |
| "loss": 0.6071, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.00011916383111823773, | |
| "loss": 0.5496, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.00012074066922755797, | |
| "loss": 0.6531, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.00012225482048864924, | |
| "loss": 0.6045, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.00012370446449200064, | |
| "loss": 0.6114, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.00012508785838284846, | |
| "loss": 0.6003, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0001264033389565572, | |
| "loss": 0.6637, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.00012764932465822253, | |
| "loss": 0.6116, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.00012882431748412055, | |
| "loss": 0.6878, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 0.00012992690478271, | |
| "loss": 0.5817, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.0001309557609530009, | |
| "loss": 0.6061, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.0001319096490382838, | |
| "loss": 0.6419, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.00013278742221327952, | |
| "loss": 0.5573, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.0001335880251629203, | |
| "loss": 0.6822, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0001343104953511231, | |
| "loss": 0.6704, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.00013495396417800677, | |
| "loss": 0.6309, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.00013551765802418287, | |
| "loss": 0.5814, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.0001360008991808469, | |
| "loss": 0.6252, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.00013640310666456482, | |
| "loss": 0.6373, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00013672379691576902, | |
| "loss": 0.6937, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.0001369625843801236, | |
| "loss": 0.6196, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0001371191819720623, | |
| "loss": 0.6718, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.0001371934014199418, | |
| "loss": 0.6286, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.00013718515349239374, | |
| "loss": 0.6942, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 0.00013709444810560437, | |
| "loss": 0.7029, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.00013692139431139254, | |
| "loss": 0.6829, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 0.00013666620016610043, | |
| "loss": 0.6676, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.0001363291724804556, | |
| "loss": 0.6491, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.00013591071645070298, | |
| "loss": 0.6058, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.00013541133517145257, | |
| "loss": 0.6635, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.00013483162903082574, | |
| "loss": 0.6435, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.00013417229498863067, | |
| "loss": 0.6009, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.00013343412573843465, | |
| "loss": 0.6296, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.0001326180087545332, | |
| "loss": 0.585, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.00013172492522497574, | |
| "loss": 0.6359, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.00013075594887191708, | |
| "loss": 0.5417, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.0001297122446607181, | |
| "loss": 0.7232, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.3138461709022522, | |
| "eval_runtime": 8.6365, | |
| "eval_samples_per_second": 76.072, | |
| "eval_steps_per_second": 9.61, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.4667747726642993e-05, | |
| "loss": 0.6323, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.3214755505568174e-05, | |
| "loss": 0.5288, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.1829845442054389e-05, | |
| "loss": 0.4822, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.0514719932940577e-05, | |
| "loss": 0.5699, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.270995592595549e-06, | |
| "loss": 0.5351, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 8.100201265703248e-06, | |
| "loss": 0.6382, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.0037761479277354e-06, | |
| "loss": 0.5365, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.983068016793553e-06, | |
| "loss": 0.5665, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.039331574927566e-06, | |
| "loss": 0.5169, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.173726907722877e-06, | |
| "loss": 0.5302, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.387318057306161e-06, | |
| "loss": 0.5061, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.6810717145662967e-06, | |
| "loss": 0.5636, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.0558560308545347e-06, | |
| "loss": 0.5727, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.5124395508079945e-06, | |
| "loss": 0.5178, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.0514902676223786e-06, | |
| "loss": 0.5668, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 6.735748019196765e-07, | |
| "loss": 0.5409, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.791577052345188e-07, | |
| "loss": 0.5488, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.6860088896217012e-07, | |
| "loss": 0.5891, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 4.216317948330257e-08, | |
| "loss": 0.6293, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 0.0, | |
| "loss": 0.5776, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 4.216317948328734e-08, | |
| "loss": 0.6373, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.6860088896213203e-07, | |
| "loss": 0.6081, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.791577052344655e-07, | |
| "loss": 0.5801, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.73574801919608e-07, | |
| "loss": 0.5491, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.051490267622295e-06, | |
| "loss": 0.5859, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.5124395508078954e-06, | |
| "loss": 0.5745, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 2.055856030853948e-06, | |
| "loss": 0.4955, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 2.6810717145661676e-06, | |
| "loss": 0.5646, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.387318057305407e-06, | |
| "loss": 0.6039, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 4.173726907722717e-06, | |
| "loss": 0.5249, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 5.039331574926652e-06, | |
| "loss": 0.5415, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 5.983068016793356e-06, | |
| "loss": 0.5852, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 7.003776147927523e-06, | |
| "loss": 0.5236, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 8.100201265703026e-06, | |
| "loss": 0.5849, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 9.270995592595313e-06, | |
| "loss": 0.5388, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.051471993293929e-05, | |
| "loss": 0.5372, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.1829845442054121e-05, | |
| "loss": 0.5343, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.3214755505566749e-05, | |
| "loss": 0.5689, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 1.4667747726642696e-05, | |
| "loss": 0.5281, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.6187036018636972e-05, | |
| "loss": 0.5893, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 1.7770752800649943e-05, | |
| "loss": 0.538, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.941695129323498e-05, | |
| "loss": 0.5953, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 2.1123607911480585e-05, | |
| "loss": 0.5626, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 2.2888624752492827e-05, | |
| "loss": 0.5227, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 2.4709832174232166e-05, | |
| "loss": 0.4595, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 2.658499146254842e-05, | |
| "loss": 0.4945, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 2.8511797583108267e-05, | |
| "loss": 0.4842, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.0487882014855305e-05, | |
| "loss": 0.5576, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.251081566150148e-05, | |
| "loss": 0.5323, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.45781118374983e-05, | |
| "loss": 0.551, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.668722932477494e-05, | |
| "loss": 0.6077, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.883557549653519e-05, | |
| "loss": 0.5575, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.102050950423e-05, | |
| "loss": 0.5375, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.3239345523814906e-05, | |
| "loss": 0.5252, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.548935605729356e-05, | |
| "loss": 0.5687, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.776777528547973e-05, | |
| "loss": 0.5572, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 5.007180246788566e-05, | |
| "loss": 0.4855, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 5.239860538550981e-05, | |
| "loss": 0.5661, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 5.4745323822347075e-05, | |
| "loss": 0.4792, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 5.710907308129398e-05, | |
| "loss": 0.568, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 5.948694753015834e-05, | |
| "loss": 0.5744, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 6.18760241733918e-05, | |
| "loss": 0.5533, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 6.427336624518557e-05, | |
| "loss": 0.5154, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 6.667602681947584e-05, | |
| "loss": 0.5253, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 6.908105243243853e-05, | |
| "loss": 0.511, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 7.148548671303826e-05, | |
| "loss": 0.5867, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 7.388637401713134e-05, | |
| "loss": 0.5813, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 7.628076306068698e-05, | |
| "loss": 0.5911, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 7.866571054763731e-05, | |
| "loss": 0.5641, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 8.103828478792738e-05, | |
| "loss": 0.5811, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 8.339556930128123e-05, | |
| "loss": 0.4312, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 8.573466640227146e-05, | |
| "loss": 0.5149, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 8.805270076230236e-05, | |
| "loss": 0.5371, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 9.0346822944092e-05, | |
| "loss": 0.5879, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 9.261421290433879e-05, | |
| "loss": 0.6111, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 9.485208346024451e-05, | |
| "loss": 0.5744, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 9.705768371566385e-05, | |
| "loss": 0.5572, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 9.922830244262129e-05, | |
| "loss": 0.5072, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.00010136127141409314, | |
| "loss": 0.6262, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 0.00010345396868391098, | |
| "loss": 0.5653, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00010550382180978354, | |
| "loss": 0.629, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 0.00010750831101546573, | |
| "loss": 0.501, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.00010946497228818022, | |
| "loss": 0.6152, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.0001113714004075095, | |
| "loss": 0.5848, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 0.00011322525190199372, | |
| "loss": 0.6411, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 0.00011502424792984731, | |
| "loss": 0.5627, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.00011676617708021282, | |
| "loss": 0.62, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.289399653673172, | |
| "eval_runtime": 9.1611, | |
| "eval_samples_per_second": 75.973, | |
| "eval_steps_per_second": 9.497, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 4.9300467408522294e-06, | |
| "loss": 0.5907, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 5.853621386191686e-06, | |
| "loss": 0.551, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 6.85263374207898e-06, | |
| "loss": 0.6016, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 7.92588273195002e-06, | |
| "loss": 0.5671, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.21771365404129028, | |
| "eval_runtime": 29.2775, | |
| "eval_samples_per_second": 22.44, | |
| "eval_steps_per_second": 2.835, | |
| "step": 1812 | |
| } | |
| ], | |
| "max_steps": 2265, | |
| "num_train_epochs": 5, | |
| "total_flos": 1891623665664000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |