| { | |
| "best_metric": 0.16020862758159637, | |
| "best_model_checkpoint": "output/eminem/checkpoint-910", | |
| "epoch": 2.0, | |
| "global_step": 910, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.638262768466246e-06, | |
| "loss": 0.3685, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.618083216571314e-06, | |
| "loss": 0.32, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.672094408674214e-06, | |
| "loss": 0.3424, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.79903472075949e-06, | |
| "loss": 0.3602, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.997555234556058e-06, | |
| "loss": 0.3777, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.126622135214757e-05, | |
| "loss": 0.374, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.2603514513152213e-05, | |
| "loss": 0.3274, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.4007834012404736e-05, | |
| "loss": 0.3659, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.5477498915945268e-05, | |
| "loss": 0.3498, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.7010750073075836e-05, | |
| "loss": 0.3689, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.8605752221991583e-05, | |
| "loss": 0.3778, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.026059618656492e-05, | |
| "loss": 0.3858, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.1973301161563144e-05, | |
| "loss": 0.3659, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.3741817083632926e-05, | |
| "loss": 0.3721, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.5564027085161653e-05, | |
| "loss": 0.3581, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.7437750028124533e-05, | |
| "loss": 0.3966, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.9360743114838212e-05, | |
| "loss": 0.4075, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.1330704572545265e-05, | |
| "loss": 0.3213, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.334527640855557e-05, | |
| "loss": 0.3886, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.540204723273326e-05, | |
| "loss": 0.3776, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.7498555143833497e-05, | |
| "loss": 0.369, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.963229067635737e-05, | |
| "loss": 0.3304, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.180069980431396e-05, | |
| "loss": 0.3778, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.400118699831243e-05, | |
| "loss": 0.3336, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.623111833235935e-05, | |
| "loss": 0.3594, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.848782463657604e-05, | |
| "loss": 0.3779, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.076860469216043e-05, | |
| "loss": 0.3485, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.3070728464635136e-05, | |
| "loss": 0.3823, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.539144037166318e-05, | |
| "loss": 0.3733, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.7727962581384004e-05, | |
| "loss": 0.3473, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.007749833742314e-05, | |
| "loss": 0.4018, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.243723530652164e-05, | |
| "loss": 0.396, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.480434894484319e-05, | |
| "loss": 0.3803, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.71760058788546e-05, | |
| "loss": 0.3732, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.954936729683709e-05, | |
| "loss": 0.3755, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.192159234682948e-05, | |
| "loss": 0.3447, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.428984153708128e-05, | |
| "loss": 0.3732, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 7.665128013484672e-05, | |
| "loss": 0.3869, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.900308155947452e-05, | |
| "loss": 0.387, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.134243076576889e-05, | |
| "loss": 0.426, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.366652761350052e-05, | |
| "loss": 0.3936, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.597259021913802e-05, | |
| "loss": 0.4021, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.825785828564833e-05, | |
| "loss": 0.3664, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.051959640653656e-05, | |
| "loss": 0.3974, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.275509734003202e-05, | |
| "loss": 0.3667, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.496168524960306e-05, | |
| "loss": 0.4155, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.713671890684909e-05, | |
| "loss": 0.371, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.927759485299701e-05, | |
| "loss": 0.3468, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00010138175051515928, | |
| "loss": 0.4078, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0001034466672736829, | |
| "loss": 0.387, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00010546987347685277, | |
| "loss": 0.4011, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00010744894739941007, | |
| "loss": 0.4018, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00010938152014129237, | |
| "loss": 0.4338, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00011126527846313597, | |
| "loss": 0.4123, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00011309796755517588, | |
| "loss": 0.3855, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00011487739373618224, | |
| "loss": 0.4182, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00011660142707925318, | |
| "loss": 0.3913, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00011826800396126461, | |
| "loss": 0.477, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00011987512953299345, | |
| "loss": 0.3962, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00012142088010688345, | |
| "loss": 0.4238, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001229034054596614, | |
| "loss": 0.4407, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00012432093104699607, | |
| "loss": 0.4689, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00012567176012759143, | |
| "loss": 0.4264, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0001269542757941326, | |
| "loss": 0.3877, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0001281669429086917, | |
| "loss": 0.4585, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0001293083099402366, | |
| "loss": 0.4342, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0001303770107020854, | |
| "loss": 0.4388, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00013137176598719452, | |
| "loss": 0.4933, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00013229138509933145, | |
| "loss": 0.4994, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00013313476727831372, | |
| "loss": 0.4215, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00013390090301758416, | |
| "loss": 0.4488, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00013458887527257018, | |
| "loss": 0.4785, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0001351978605583545, | |
| "loss": 0.444, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00013572712993537543, | |
| "loss": 0.409, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00013617604988193778, | |
| "loss": 0.4202, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00013654408305253035, | |
| "loss": 0.4465, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0001368307889210095, | |
| "loss": 0.487, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001370358243079002, | |
| "loss": 0.4389, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00013715894379117118, | |
| "loss": 0.4761, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0001372, | |
| "loss": 0.4531, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0001371589437911712, | |
| "loss": 0.4367, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001370358243079003, | |
| "loss": 0.4265, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0001368307889210096, | |
| "loss": 0.4721, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00013654408305253054, | |
| "loss": 0.3988, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00013617604988193797, | |
| "loss": 0.4344, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00013572712993537567, | |
| "loss": 0.4899, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00013519786055835476, | |
| "loss": 0.5019, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0001345888752725705, | |
| "loss": 0.4645, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00013390090301758454, | |
| "loss": 0.3885, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0001331347672783141, | |
| "loss": 0.426, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.21941067278385162, | |
| "eval_runtime": 15.1157, | |
| "eval_samples_per_second": 42.803, | |
| "eval_steps_per_second": 5.359, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00013229138509933188, | |
| "loss": 0.4675, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00013137176598719498, | |
| "loss": 0.3498, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00013037701070208588, | |
| "loss": 0.4028, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00012930830994023715, | |
| "loss": 0.3913, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00012816694290869227, | |
| "loss": 0.427, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00012695427579413322, | |
| "loss": 0.3914, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00012567176012759208, | |
| "loss": 0.4128, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00012432093104699674, | |
| "loss": 0.3783, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00012290340545966207, | |
| "loss": 0.4026, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00012142088010688418, | |
| "loss": 0.3649, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00011987512953299421, | |
| "loss": 0.4037, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00011826800396126541, | |
| "loss": 0.3905, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.000116601427079254, | |
| "loss": 0.3741, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00011487739373618308, | |
| "loss": 0.3776, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00011309796755517674, | |
| "loss": 0.3915, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00011126527846313685, | |
| "loss": 0.3885, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00010938152014129329, | |
| "loss": 0.3599, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00010744894739941099, | |
| "loss": 0.3955, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00010546987347685372, | |
| "loss": 0.3638, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00010344666727368387, | |
| "loss": 0.4055, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00010138175051516028, | |
| "loss": 0.3965, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.927759485299804e-05, | |
| "loss": 0.3767, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 9.713671890685015e-05, | |
| "loss": 0.4088, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.49616852496041e-05, | |
| "loss": 0.4329, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.275509734003309e-05, | |
| "loss": 0.3581, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.051959640653763e-05, | |
| "loss": 0.4084, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.825785828564943e-05, | |
| "loss": 0.3755, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.597259021913913e-05, | |
| "loss": 0.4321, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.366652761350163e-05, | |
| "loss": 0.4021, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.134243076577001e-05, | |
| "loss": 0.4115, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 7.900308155947565e-05, | |
| "loss": 0.3358, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.665128013484786e-05, | |
| "loss": 0.3576, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.428984153708243e-05, | |
| "loss": 0.4326, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.192159234683063e-05, | |
| "loss": 0.3627, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.954936729683823e-05, | |
| "loss": 0.3891, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.717600587885573e-05, | |
| "loss": 0.3983, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.480434894484433e-05, | |
| "loss": 0.3504, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.243723530652277e-05, | |
| "loss": 0.3806, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.007749833742428e-05, | |
| "loss": 0.3616, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.772796258138513e-05, | |
| "loss": 0.4417, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.53914403716643e-05, | |
| "loss": 0.4272, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.307072846463625e-05, | |
| "loss": 0.3909, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.076860469216153e-05, | |
| "loss": 0.3833, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.848782463657713e-05, | |
| "loss": 0.3611, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.6231118332360436e-05, | |
| "loss": 0.415, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.40011869983135e-05, | |
| "loss": 0.4037, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.1800699804315014e-05, | |
| "loss": 0.3836, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 3.9632290676358406e-05, | |
| "loss": 0.3564, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 3.749855514383451e-05, | |
| "loss": 0.4255, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 3.5402047232734255e-05, | |
| "loss": 0.3377, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 3.334527640855654e-05, | |
| "loss": 0.3671, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.133070457254623e-05, | |
| "loss": 0.3735, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.9360743114839147e-05, | |
| "loss": 0.3614, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.7437750028125448e-05, | |
| "loss": 0.373, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.5564027085162544e-05, | |
| "loss": 0.3795, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3741817083633794e-05, | |
| "loss": 0.3621, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.197330116156398e-05, | |
| "loss": 0.3882, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.026059618656573e-05, | |
| "loss": 0.3836, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.860575222199237e-05, | |
| "loss": 0.3413, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.701075007307659e-05, | |
| "loss": 0.3664, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.5477498915945983e-05, | |
| "loss": 0.3463, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.4007834012405429e-05, | |
| "loss": 0.3331, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2603514513152868e-05, | |
| "loss": 0.3815, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1266221352148201e-05, | |
| "loss": 0.3769, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 9.997555234556651e-06, | |
| "loss": 0.4108, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.799034720760045e-06, | |
| "loss": 0.3722, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 7.672094408674733e-06, | |
| "loss": 0.4017, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 6.6180832165718085e-06, | |
| "loss": 0.4178, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.638262768466695e-06, | |
| "loss": 0.3899, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.733805883975914e-06, | |
| "loss": 0.3741, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.905795174492382e-06, | |
| "loss": 0.3831, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.15522174732089e-06, | |
| "loss": 0.3682, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.4829840193538357e-06, | |
| "loss": 0.391, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.88988664168465e-06, | |
| "loss": 0.3597, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.3766395364657029e-06, | |
| "loss": 0.379, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 9.43857047145762e-07, | |
| "loss": 0.4247, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.920572031218176e-07, | |
| "loss": 0.3705, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.2166109966507624e-07, | |
| "loss": 0.3148, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.3299239388725518e-07, | |
| "loss": 0.3331, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.6276917326658224e-08, | |
| "loss": 0.3802, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.6424056362936757e-09, | |
| "loss": 0.324, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 5.911834568650871e-08, | |
| "loss": 0.3751, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.9863594027063992e-07, | |
| "loss": 0.3792, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.200281904521944e-07, | |
| "loss": 0.3776, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.230300954609269e-07, | |
| "loss": 0.3469, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.1072789698879036e-06, | |
| "loss": 0.3464, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.572314877814697e-06, | |
| "loss": 0.3726, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.11758118334118e-06, | |
| "loss": 0.3881, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.742425216867453e-06, | |
| "loss": 0.3518, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.4460990563188536e-06, | |
| "loss": 0.3839, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.227760422393333e-06, | |
| "loss": 0.347, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.20774193108081818, | |
| "eval_runtime": 15.111, | |
| "eval_samples_per_second": 42.816, | |
| "eval_steps_per_second": 5.36, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.761634319266075e-07, | |
| "loss": 0.4447, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.16020862758159637, | |
| "eval_runtime": 13.0694, | |
| "eval_samples_per_second": 49.046, | |
| "eval_steps_per_second": 6.198, | |
| "step": 910 | |
| } | |
| ], | |
| "max_steps": 1365, | |
| "num_train_epochs": 3, | |
| "total_flos": 950057828352000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |