| { | |
| "best_metric": 3.1907694339752197, | |
| "best_model_checkpoint": "output/snoop-dogg/checkpoint-1614", | |
| "epoch": 3.0, | |
| "global_step": 1614, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013717076261874923, | |
| "loss": 4.1063, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013708307539699282, | |
| "loss": 4.1679, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00013693701307947525, | |
| "loss": 4.0154, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001367327001699768, | |
| "loss": 3.9048, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013647031082518634, | |
| "loss": 3.9959, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00013615006870624992, | |
| "loss": 3.8299, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00013577224678812164, | |
| "loss": 3.7535, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00013533716712687914, | |
| "loss": 3.7139, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0001348452005852027, | |
| "loss": 3.5272, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00013429676651625084, | |
| "loss": 3.9259, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0001336923324062034, | |
| "loss": 3.646, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00013303241347577507, | |
| "loss": 3.7353, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013231757224104065, | |
| "loss": 3.7269, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001315484180339451, | |
| "loss": 3.6971, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001307256064829081, | |
| "loss": 3.6029, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00012984983895396502, | |
| "loss": 3.6687, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012892186195292147, | |
| "loss": 3.5599, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00012794246648903024, | |
| "loss": 3.4859, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00012691248740073373, | |
| "loss": 3.7556, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00012583280264404602, | |
| "loss": 3.6952, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00012470433254418178, | |
| "loss": 3.7874, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00012352803901106928, | |
| "loss": 3.6861, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00012230492471941668, | |
| "loss": 3.7332, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00012103603225403036, | |
| "loss": 3.6027, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00011972244322111374, | |
| "loss": 3.6728, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00011836527732630418, | |
| "loss": 3.6239, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00011696569142023398, | |
| "loss": 3.6474, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00011552487851242871, | |
| "loss": 3.6182, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00011404406675438398, | |
| "loss": 3.6673, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00011252451839268661, | |
| "loss": 3.6792, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00011096752869307368, | |
| "loss": 3.7147, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001093744248363455, | |
| "loss": 3.7344, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0001077465647870744, | |
| "loss": 3.672, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0001060853361360733, | |
| "loss": 3.5366, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00010439215491761087, | |
| "loss": 3.7628, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00010266846440238127, | |
| "loss": 3.4477, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00010091573386725777, | |
| "loss": 3.641, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.913545734287838e-05, | |
| "loss": 3.5141, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.732915234013161e-05, | |
| "loss": 3.7305, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.549835855662742e-05, | |
| "loss": 3.5619, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.364463656425613e-05, | |
| "loss": 3.4244, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.176956647895434e-05, | |
| "loss": 3.5957, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.987474661381113e-05, | |
| "loss": 3.579, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.79617921166632e-05, | |
| "loss": 3.4737, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.603233359333994e-05, | |
| "loss": 3.5314, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.408801571773217e-05, | |
| "loss": 3.6413, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.213049582986904e-05, | |
| "loss": 3.616, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.016144252319865e-05, | |
| "loss": 3.6094, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.818253422227576e-05, | |
| "loss": 3.4714, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.619545775206993e-05, | |
| "loss": 3.6148, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.420190690011274e-05, | |
| "loss": 3.393, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.220358097271036e-05, | |
| "loss": 3.7299, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.02021833464518e-05, | |
| "loss": 3.5671, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6.819942001624763e-05, | |
| "loss": 3.5912, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.619699814113692e-05, | |
| "loss": 3.4808, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.419662458910188e-05, | |
| "loss": 3.5158, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.220000448213035e-05, | |
| "loss": 3.4962, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 6.020883974276696e-05, | |
| "loss": 3.5426, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.822482764339123e-05, | |
| "loss": 3.4774, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.6249659359459626e-05, | |
| "loss": 3.6775, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.4285018527944484e-05, | |
| "loss": 3.4415, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.233257981219891e-05, | |
| "loss": 3.2558, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.0394007474470735e-05, | |
| "loss": 3.4333, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.8470953957282466e-05, | |
| "loss": 3.4511, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.656505847488621e-05, | |
| "loss": 3.439, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.4677945615994564e-05, | |
| "loss": 3.5292, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.281122395897837e-05, | |
| "loss": 3.5033, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.096648470071148e-05, | |
| "loss": 3.4235, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.914530030023193e-05, | |
| "loss": 3.4522, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.734922313837482e-05, | |
| "loss": 3.4705, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.5579784194520246e-05, | |
| "loss": 3.5511, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.383849174158367e-05, | |
| "loss": 3.3867, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.2126830060361513e-05, | |
| "loss": 3.6066, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.044625817432741e-05, | |
| "loss": 3.3799, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.879820860595807e-05, | |
| "loss": 3.3825, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.718408615564853e-05, | |
| "loss": 3.4277, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.560526670425788e-05, | |
| "loss": 3.6353, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.4063096040305828e-05, | |
| "loss": 3.388, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.2558888712820237e-05, | |
| "loss": 3.4079, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.1093926910813165e-05, | |
| "loss": 3.6123, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.966945937034066e-05, | |
| "loss": 3.4731, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.828670031007799e-05, | |
| "loss": 3.6209, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.694682839631735e-05, | |
| "loss": 3.6534, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5650985738270693e-05, | |
| "loss": 3.5689, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4400276914533763e-05, | |
| "loss": 3.5702, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.319576803154135e-05, | |
| "loss": 3.4445, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.2038485814816265e-05, | |
| "loss": 3.3746, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.092941673378663e-05, | |
| "loss": 3.5187, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.869506160917616e-06, | |
| "loss": 3.4482, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.859657565874292e-06, | |
| "loss": 3.5279, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.900731745402374e-06, | |
| "loss": 3.4626, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.9935460895836276e-06, | |
| "loss": 3.6454, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.138873885091061e-06, | |
| "loss": 3.4554, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.3374436560379995e-06, | |
| "loss": 3.4318, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.589938542982855e-06, | |
| "loss": 3.5353, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.896995720618831e-06, | |
| "loss": 3.3728, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.2592058546451155e-06, | |
| "loss": 3.5394, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.677112598282391e-06, | |
| "loss": 3.4868, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.1512121288618626e-06, | |
| "loss": 3.3146, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.6819527248827737e-06, | |
| "loss": 3.286, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.269734383899055e-06, | |
| "loss": 3.3572, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.149084815606564e-07, | |
| "loss": 3.4358, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.177774721003861e-07, | |
| "loss": 3.404, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.7859463052128633e-07, | |
| "loss": 3.4824, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.9756383670464207e-07, | |
| "loss": 3.3723, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.483940162237093e-08, | |
| "loss": 3.4703, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0525935802102593e-08, | |
| "loss": 3.3436, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 3.417607307434082, | |
| "eval_runtime": 34.9657, | |
| "eval_samples_per_second": 21.535, | |
| "eval_steps_per_second": 2.717, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.678260156942859e-09, | |
| "loss": 3.4076, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.730135925604576e-08, | |
| "loss": 3.3309, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.683503770757803e-07, | |
| "loss": 3.3307, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.377306552350485e-07, | |
| "loss": 3.1417, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.6529781368228e-07, | |
| "loss": 3.2028, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.508578737652368e-07, | |
| "loss": 3.4381, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.1941674235787645e-06, | |
| "loss": 3.2711, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.5949338254495494e-06, | |
| "loss": 3.2736, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.0528154653809706e-06, | |
| "loss": 3.2426, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.5674220442454935e-06, | |
| "loss": 3.3709, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.13831491047635e-06, | |
| "loss": 3.5776, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.7650074339748796e-06, | |
| "loss": 3.3789, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.446965420914962e-06, | |
| "loss": 3.4289, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.1836075690907296e-06, | |
| "loss": 3.3452, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.974305963419616e-06, | |
| "loss": 3.3496, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.818386611178326e-06, | |
| "loss": 3.3166, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.715130016515405e-06, | |
| "loss": 3.3951, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.66377179375079e-06, | |
| "loss": 3.3529, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.66350331893958e-06, | |
| "loss": 3.3422, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.0713472419144369e-05, | |
| "loss": 3.3403, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.1812784098829224e-05, | |
| "loss": 3.3656, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2960501302755281e-05, | |
| "loss": 3.2795, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.41556457147284e-05, | |
| "loss": 3.3825, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.5397198591517605e-05, | |
| "loss": 3.1902, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.668410163123363e-05, | |
| "loss": 3.3563, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.8015257875427347e-05, | |
| "loss": 3.2479, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.9389532644139112e-05, | |
| "loss": 3.3764, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.080575450310203e-05, | |
| "loss": 3.2164, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.226271626227443e-05, | |
| "loss": 3.1518, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.375917600485109e-05, | |
| "loss": 3.2922, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.5293858145875134e-05, | |
| "loss": 3.4571, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.6865454519549044e-05, | |
| "loss": 3.3273, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8472625494317417e-05, | |
| "loss": 3.2389, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.011400111477139e-05, | |
| "loss": 3.2702, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.178818226940102e-05, | |
| "loss": 3.6087, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.349374188320044e-05, | |
| "loss": 3.3782, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.522922613410924e-05, | |
| "loss": 3.428, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.69931556922527e-05, | |
| "loss": 3.2301, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.878402698092561e-05, | |
| "loss": 3.3942, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.0600313458243476e-05, | |
| "loss": 3.3824, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.244046691836963e-05, | |
| "loss": 3.4915, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.430291881120861e-05, | |
| "loss": 3.2738, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.618608157944113e-05, | |
| "loss": 3.3024, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.808835001176069e-05, | |
| "loss": 3.3214, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.000810261115872e-05, | |
| "loss": 3.5582, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.194370297709118e-05, | |
| "loss": 3.2745, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.3893501200349724e-05, | |
| "loss": 3.2672, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.585583526944705e-05, | |
| "loss": 3.3916, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.7829032487318437e-05, | |
| "loss": 3.5783, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.9811410897131736e-05, | |
| "loss": 3.1478, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.180128071599044e-05, | |
| "loss": 3.2909, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.379694577530755e-05, | |
| "loss": 3.4915, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.57967049666228e-05, | |
| "loss": 3.1955, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.77988536916305e-05, | |
| "loss": 3.4232, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.980168531518204e-05, | |
| "loss": 3.3181, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.1803492620025e-05, | |
| "loss": 3.4153, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.380256926203807e-05, | |
| "loss": 3.4825, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.579721122472176e-05, | |
| "loss": 3.2802, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 7.778571827170516e-05, | |
| "loss": 3.4058, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.976639539603049e-05, | |
| "loss": 3.385, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.173755426497999e-05, | |
| "loss": 3.361, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.36975146592138e-05, | |
| "loss": 3.3453, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.564460590499189e-05, | |
| "loss": 3.2614, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.757716829825938e-05, | |
| "loss": 3.2934, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.949355451938121e-05, | |
| "loss": 3.2375, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.139213103732031e-05, | |
| "loss": 3.5485, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.327127950206244e-05, | |
| "loss": 3.6656, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.512939812410057e-05, | |
| "loss": 3.2755, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.69649030398032e-05, | |
| "loss": 3.3555, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.877622966150243e-05, | |
| "loss": 3.3756, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00010056183401115142, | |
| "loss": 3.0988, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00010232019403641382, | |
| "loss": 3.2778, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00010404981090806416, | |
| "loss": 3.4258, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00010574921029759243, | |
| "loss": 3.3742, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00010741694363392428, | |
| "loss": 3.3941, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00010905158933818573, | |
| "loss": 3.3793, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011065175403545939, | |
| "loss": 3.3372, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00011221607374249983, | |
| "loss": 3.3405, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011374321503039532, | |
| "loss": 3.2013, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.000115231876161185, | |
| "loss": 3.3862, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00011668078819746258, | |
| "loss": 3.3706, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011808871608402098, | |
| "loss": 3.5269, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00011945445970061549, | |
| "loss": 3.4064, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00012077685488494853, | |
| "loss": 3.4603, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00012205477442500345, | |
| "loss": 3.3909, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00012328712901988232, | |
| "loss": 3.3253, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0001244728682083278, | |
| "loss": 3.384, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00012561098126413823, | |
| "loss": 3.3282, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0001267004980577125, | |
| "loss": 3.3693, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0001277404898829901, | |
| "loss": 3.3716, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00012873007024908207, | |
| "loss": 3.4564, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.0001296683956359172, | |
| "loss": 3.4711, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00013055466621326016, | |
| "loss": 3.2512, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0001313881265224882, | |
| "loss": 3.3624, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00013216806612054534, | |
| "loss": 3.3469, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00013289382018552548, | |
| "loss": 3.3339, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.00013356477008336762, | |
| "loss": 3.5051, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013418034389518066, | |
| "loss": 3.6009, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013474001690474816, | |
| "loss": 3.4188, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00013524331204579738, | |
| "loss": 3.3583, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00013568980030865128, | |
| "loss": 3.4374, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013607910110591736, | |
| "loss": 3.5138, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.0001364108825969008, | |
| "loss": 3.2506, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.00013668486197046618, | |
| "loss": 3.313, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00013690080568610629, | |
| "loss": 3.4122, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00013705852967301225, | |
| "loss": 3.2444, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00013715789948697598, | |
| "loss": 3.3139, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0001371988304249906, | |
| "loss": 3.4089, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.2667410373687744, | |
| "eval_runtime": 18.6928, | |
| "eval_samples_per_second": 40.283, | |
| "eval_steps_per_second": 5.082, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.00013718128759745147, | |
| "loss": 3.1857, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.00013710528595789614, | |
| "loss": 3.1502, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.00013697089029025781, | |
| "loss": 3.244, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00013677821515364362, | |
| "loss": 2.9782, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00013652742478468407, | |
| "loss": 3.4309, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.00013621873295753753, | |
| "loss": 3.3141, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.0001358524028016687, | |
| "loss": 3.2167, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0001354287465775566, | |
| "loss": 3.2457, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.00013494812541052316, | |
| "loss": 3.2054, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.00013441094898290952, | |
| "loss": 3.1213, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00013381767518486174, | |
| "loss": 3.1849, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.00013316880972402475, | |
| "loss": 3.2974, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0001324649056944761, | |
| "loss": 3.0743, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0001317065631052675, | |
| "loss": 3.3105, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.00013089442836897625, | |
| "loss": 3.0161, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.00013002919375070177, | |
| "loss": 3.1369, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0001291115967779778, | |
| "loss": 3.2544, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.00012814241961210242, | |
| "loss": 3.2921, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.00012712248838142262, | |
| "loss": 3.2056, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00012605267247714066, | |
| "loss": 3.0081, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.00012493388381224383, | |
| "loss": 3.3426, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.00012376707604418792, | |
| "loss": 3.4155, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.00012255324376199802, | |
| "loss": 3.2321, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0001212934216384791, | |
| "loss": 3.3063, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.00011998868354825889, | |
| "loss": 3.3722, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.0001186401416524155, | |
| "loss": 3.0143, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.000117248945450469, | |
| "loss": 3.0218, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00011581628080054626, | |
| "loss": 3.0963, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.00011434336890855335, | |
| "loss": 3.2636, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.00011283146528721704, | |
| "loss": 3.147, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.00011128185868588423, | |
| "loss": 3.1549, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.0001096958699919889, | |
| "loss": 3.2418, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.00010807485110512604, | |
| "loss": 3.3284, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.00010642018378468939, | |
| "loss": 2.9988, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00010473327847205782, | |
| "loss": 3.2944, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.00010301557308833204, | |
| "loss": 3.3507, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.00010126853180864858, | |
| "loss": 3.1687, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.949364381411411e-05, | |
| "loss": 3.0591, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.769242202242448e-05, | |
| "loss": 3.1366, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 9.586640179825159e-05, | |
| "loss": 3.1332, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.401713964449514e-05, | |
| "loss": 3.17, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 9.21462118755175e-05, | |
| "loss": 3.2294, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.025521327348971e-05, | |
| "loss": 2.9817, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 8.834575572899717e-05, | |
| "loss": 3.1762, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.641946686706039e-05, | |
| "loss": 3.3385, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.447798865974526e-05, | |
| "loss": 3.0338, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.252297602654276e-05, | |
| "loss": 3.1138, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.055609542371231e-05, | |
| "loss": 3.3565, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.857902342379239e-05, | |
| "loss": 3.3677, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.659344528648596e-05, | |
| "loss": 3.0411, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.46010535221433e-05, | |
| "loss": 3.1615, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 7.26035464490619e-05, | |
| "loss": 3.3576, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.060262674583818e-05, | |
| "loss": 3.1091, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.860000000000001e-05, | |
| "loss": 3.4431, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.659737325416186e-05, | |
| "loss": 3.2144, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.459645355093813e-05, | |
| "loss": 3.1546, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.259894647785675e-05, | |
| "loss": 2.963, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.060655471351407e-05, | |
| "loss": 3.1648, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.862097657620766e-05, | |
| "loss": 3.1867, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.664390457628773e-05, | |
| "loss": 3.11, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.467702397345729e-05, | |
| "loss": 3.0585, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.272201134025478e-05, | |
| "loss": 3.0417, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.078053313293964e-05, | |
| "loss": 3.078, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.885424427100287e-05, | |
| "loss": 3.2723, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.694478672651034e-05, | |
| "loss": 3.0448, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.505378812448254e-05, | |
| "loss": 3.1823, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.31828603555049e-05, | |
| "loss": 3.1987, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.133359820174844e-05, | |
| "loss": 3.1579, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.950757797757557e-05, | |
| "loss": 3.0757, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.7706356185885915e-05, | |
| "loss": 3.1036, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.593146819135146e-05, | |
| "loss": 3.0992, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.418442691166801e-05, | |
| "loss": 3.1336, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.246672152794222e-05, | |
| "loss": 3.1688, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.077981621531067e-05, | |
| "loss": 2.9663, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.9125148894873993e-05, | |
| "loss": 3.057, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.750413000801114e-05, | |
| "loss": 3.1822, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.5918141314115796e-05, | |
| "loss": 2.9519, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.436853471278299e-05, | |
| "loss": 3.0403, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.2856631091446682e-05, | |
| "loss": 3.0835, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.138371919945377e-05, | |
| "loss": 3.0139, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.9951054549531022e-05, | |
| "loss": 3.0177, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.85598583475845e-05, | |
| "loss": 3.3141, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.7211316451741122e-05, | |
| "loss": 3.1076, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.59065783615209e-05, | |
| "loss": 3.1387, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.4646756238001999e-05, | |
| "loss": 3.1493, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.3432923955812079e-05, | |
| "loss": 3.2037, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.22661161877562e-05, | |
| "loss": 3.033, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.114732752285933e-05, | |
| "loss": 3.1567, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.0077511618577358e-05, | |
| "loss": 2.9584, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 9.057580387897574e-06, | |
| "loss": 3.1304, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 8.088403222022203e-06, | |
| "loss": 3.0946, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.1708062492982514e-06, | |
| "loss": 3.1109, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 6.3055716310237545e-06, | |
| "loss": 3.1375, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 5.493436894732511e-06, | |
| "loss": 3.1381, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.735094305523896e-06, | |
| "loss": 3.1063, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.031190275975246e-06, | |
| "loss": 3.1137, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.3823248151382478e-06, | |
| "loss": 3.1414, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.789051017090463e-06, | |
| "loss": 3.2499, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.251874589476814e-06, | |
| "loss": 3.0676, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.7712534224433918e-06, | |
| "loss": 3.0984, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.3475971983312994e-06, | |
| "loss": 3.2583, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 9.812670424624475e-07, | |
| "loss": 3.1486, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.725752153159165e-07, | |
| "loss": 3.1602, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.217848463563714e-07, | |
| "loss": 3.0449, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.291097097421942e-07, | |
| "loss": 3.1203, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 9.471404210387107e-08, | |
| "loss": 3.1678, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.871240254851856e-08, | |
| "loss": 3.3214, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 3.1907694339752197, | |
| "eval_runtime": 18.9242, | |
| "eval_samples_per_second": 39.79, | |
| "eval_steps_per_second": 5.02, | |
| "step": 1614 | |
| } | |
| ], | |
| "max_steps": 1614, | |
| "num_train_epochs": 3, | |
| "total_flos": 1685464252416000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |