| { | |
| "best_metric": 1.9903991222381592, | |
| "best_model_checkpoint": "output/eminem/checkpoint-1824", | |
| "epoch": 4.0, | |
| "global_step": 1824, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7203068149785564e-07, | |
| "loss": 2.6483, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.899165231781454e-07, | |
| "loss": 2.495, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.1886129228032032e-06, | |
| "loss": 2.7138, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.6676447463390023e-06, | |
| "loss": 2.5829, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.2264411225088625e-06, | |
| "loss": 2.7096, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.8643361231111207e-06, | |
| "loss": 2.4466, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.580569556618851e-06, | |
| "loss": 2.7, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.3742878741139976e-06, | |
| "loss": 2.718, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.244545186477934e-06, | |
| "loss": 2.6071, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 6.190304391625697e-06, | |
| "loss": 2.5909, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.210438410439488e-06, | |
| "loss": 2.5627, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.303731529930944e-06, | |
| "loss": 2.4577, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.468880852028573e-06, | |
| "loss": 2.5309, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.0704497846266275e-05, | |
| "loss": 2.7378, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.200911000452175e-05, | |
| "loss": 2.6599, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.338116259583136e-05, | |
| "loss": 2.3352, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.481902051919347e-05, | |
| "loss": 2.7494, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.6320970252148078e-05, | |
| "loss": 2.395, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7885221892815092e-05, | |
| "loss": 2.5425, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.950991129295317e-05, | |
| "loss": 2.6284, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.11931022795012e-05, | |
| "loss": 2.4998, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.2932788961951275e-05, | |
| "loss": 2.5979, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.4726898122808088e-05, | |
| "loss": 2.7143, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.6573291688279658e-05, | |
| "loss": 2.4124, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.8469769276260912e-05, | |
| "loss": 2.6573, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.0414070818571297e-05, | |
| "loss": 2.6071, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.24038792543193e-05, | |
| "loss": 2.8709, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4436823291189305e-05, | |
| "loss": 2.6786, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.651048023135391e-05, | |
| "loss": 2.6154, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.8622378858650226e-05, | |
| "loss": 2.6546, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.077000238357357e-05, | |
| "loss": 2.6476, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.295079144258355e-05, | |
| "loss": 2.5445, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.516214714814676e-05, | |
| "loss": 2.4619, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.740143418587853e-05, | |
| "loss": 2.5966, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.9665983955098734e-05, | |
| "loss": 2.4876, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.1953097749051775e-05, | |
| "loss": 2.6759, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.426004997100832e-05, | |
| "loss": 2.4991, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.658409138240877e-05, | |
| "loss": 2.5614, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.892245237918353e-05, | |
| "loss": 2.6733, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.127234629234317e-05, | |
| "loss": 2.7397, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.363097270890295e-05, | |
| "loss": 2.7079, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.599552080918974e-05, | |
| "loss": 2.5227, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 6.836317271654733e-05, | |
| "loss": 2.6247, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.073110685545387e-05, | |
| "loss": 2.5805, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.309650131404767e-05, | |
| "loss": 2.7595, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.545653720705101e-05, | |
| "loss": 2.5378, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 7.780840203509109e-05, | |
| "loss": 2.6695, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.014929303640662e-05, | |
| "loss": 2.6016, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.2476420526954e-05, | |
| "loss": 2.8115, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.478701122492476e-05, | |
| "loss": 2.8291, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.707831155571868e-05, | |
| "loss": 2.4987, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.934759093343136e-05, | |
| "loss": 2.5696, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.15921450149434e-05, | |
| "loss": 2.6025, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.380929892273904e-05, | |
| "loss": 2.6397, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.599641043260666e-05, | |
| "loss": 2.5424, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.815087312242802e-05, | |
| "loss": 2.7226, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00010027011947830137, | |
| "loss": 2.6259, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00010235162395429476, | |
| "loss": 2.6201, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001043929059821886, | |
| "loss": 2.7705, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00010639153292761396, | |
| "loss": 2.7333, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00010834512298907064, | |
| "loss": 2.6623, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00011025134803636347, | |
| "loss": 2.8, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00011210793638507988, | |
| "loss": 2.6572, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00011391267550379957, | |
| "loss": 2.8413, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00011566341465080861, | |
| "loss": 2.6584, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00011735806743718085, | |
| "loss": 2.743, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00011899461431316495, | |
| "loss": 2.6499, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00012057110497492171, | |
| "loss": 2.6371, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00012208566068873709, | |
| "loss": 2.7359, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00012353647652994535, | |
| "loss": 2.5658, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00012492182353389096, | |
| "loss": 2.7984, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0001262400507563695, | |
| "loss": 2.6983, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0001274895872410881, | |
| "loss": 2.7025, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00012866894389180423, | |
| "loss": 2.6162, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00012977671524691021, | |
| "loss": 2.7679, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0001308115811543474, | |
| "loss": 2.408, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00013177230834485735, | |
| "loss": 2.7832, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00013265775190169125, | |
| "loss": 2.7911, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00013346685662502964, | |
| "loss": 2.7389, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00013419865828948326, | |
| "loss": 2.6844, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00013485228479317903, | |
| "loss": 2.8377, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.00013542695719706048, | |
| "loss": 2.4116, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00013592199065316375, | |
| "loss": 2.5933, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.00013633679522076416, | |
| "loss": 2.7723, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0001366708765694196, | |
| "loss": 2.7647, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0001369238365680734, | |
| "loss": 2.6368, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0001370953737595148, | |
| "loss": 2.3958, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00013718528371963105, | |
| "loss": 2.7726, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00013719345930102362, | |
| "loss": 2.6424, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00013711989076069762, | |
| "loss": 2.807, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00013696466577167284, | |
| "loss": 2.5944, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 2.5120232105255127, | |
| "eval_runtime": 30.8025, | |
| "eval_samples_per_second": 20.81, | |
| "eval_steps_per_second": 2.63, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00013672796931850215, | |
| "loss": 2.5617, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00013641008347682184, | |
| "loss": 2.3941, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0001360113870771968, | |
| "loss": 2.609, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.000135532355253661, | |
| "loss": 2.6935, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00013497355887749114, | |
| "loss": 2.4847, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00013433566387688887, | |
| "loss": 2.5804, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00013361943044338116, | |
| "loss": 2.493, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.000132825712125886, | |
| "loss": 2.432, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00013195545481352208, | |
| "loss": 2.3443, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0001310096956083743, | |
| "loss": 2.5815, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0001299895615895605, | |
| "loss": 2.5337, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00012889626847006907, | |
| "loss": 2.3796, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00012773111914797142, | |
| "loss": 2.5178, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00012649550215373373, | |
| "loss": 2.4299, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0001251908899954784, | |
| "loss": 2.4451, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00012381883740416864, | |
| "loss": 2.6521, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00012238097948080652, | |
| "loss": 2.403, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00012087902974785178, | |
| "loss": 2.601, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00011931477810718475, | |
| "loss": 2.4694, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00011769008870704702, | |
| "loss": 2.5935, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0001160068977204988, | |
| "loss": 2.5179, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00011426721103804855, | |
| "loss": 2.7228, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011247310187719211, | |
| "loss": 2.5971, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011062670831172054, | |
| "loss": 2.5625, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0001087302307237391, | |
| "loss": 2.7913, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00010678592918142871, | |
| "loss": 2.4481, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00010479612074568051, | |
| "loss": 2.4637, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00010276317670881092, | |
| "loss": 2.7515, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0001006895197686461, | |
| "loss": 2.685, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.857762114134977e-05, | |
| "loss": 2.6865, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.642999761642645e-05, | |
| "loss": 2.586, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.424920855741624e-05, | |
| "loss": 2.5638, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.203785285185348e-05, | |
| "loss": 2.8532, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 8.979856581412148e-05, | |
| "loss": 2.4697, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.753401604490128e-05, | |
| "loss": 2.4653, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 8.5246902250948e-05, | |
| "loss": 2.4832, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 8.293995002899144e-05, | |
| "loss": 2.5514, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.061590861759147e-05, | |
| "loss": 2.6216, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.827754762081649e-05, | |
| "loss": 2.4314, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.592765370765685e-05, | |
| "loss": 2.702, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.356902729109682e-05, | |
| "loss": 2.3754, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.120447919081049e-05, | |
| "loss": 2.4599, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.883682728345293e-05, | |
| "loss": 2.7681, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.646889314454614e-05, | |
| "loss": 2.4768, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.410349868595235e-05, | |
| "loss": 2.6323, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.174346279294876e-05, | |
| "loss": 2.4092, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.939159796490916e-05, | |
| "loss": 2.6557, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.705070696359339e-05, | |
| "loss": 2.3712, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 5.4723579473046015e-05, | |
| "loss": 2.3837, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 5.241298877507526e-05, | |
| "loss": 2.4259, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 5.01216884442811e-05, | |
| "loss": 2.6717, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.785240906656888e-05, | |
| "loss": 2.361, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.560785498505662e-05, | |
| "loss": 2.4714, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.339070107726098e-05, | |
| "loss": 2.6358, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.120358956739335e-05, | |
| "loss": 2.5129, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.904912687757176e-05, | |
| "loss": 2.5261, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.6929880521698846e-05, | |
| "loss": 2.5826, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.484837604570524e-05, | |
| "loss": 2.3181, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.2807094017811405e-05, | |
| "loss": 2.2812, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.080846707238584e-05, | |
| "loss": 2.479, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.885487701092958e-05, | |
| "loss": 2.6417, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.694865196363672e-05, | |
| "loss": 2.6846, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.509206361492011e-05, | |
| "loss": 2.4952, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.3287324496200448e-05, | |
| "loss": 2.5992, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1536585349191225e-05, | |
| "loss": 2.585, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.9841932562819154e-05, | |
| "loss": 2.597, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.8205385686835072e-05, | |
| "loss": 2.2072, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.6628895025078307e-05, | |
| "loss": 2.5062, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.5114339311262777e-05, | |
| "loss": 2.5392, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.3663523470054795e-05, | |
| "loss": 2.5593, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.2278176466109037e-05, | |
| "loss": 2.5837, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.0959949243630505e-05, | |
| "loss": 2.3897, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.710412758911909e-06, | |
| "loss": 2.4527, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.531056108195637e-06, | |
| "loss": 2.5424, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.423284753089891e-06, | |
| "loss": 2.4882, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.3884188456526086e-06, | |
| "loss": 2.4503, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 5.427691655142655e-06, | |
| "loss": 2.3948, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.5422480983086555e-06, | |
| "loss": 2.6461, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.7331433749702816e-06, | |
| "loss": 2.4818, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.0013417105168102e-06, | |
| "loss": 2.6764, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.347715206820977e-06, | |
| "loss": 2.4561, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.7730428029395095e-06, | |
| "loss": 2.4363, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.2780093468362156e-06, | |
| "loss": 2.5073, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.632047792358677e-07, | |
| "loss": 2.3678, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.291234305804284e-07, | |
| "loss": 2.5126, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.761634319266075e-07, | |
| "loss": 2.5024, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0462624048520178e-07, | |
| "loss": 2.4416, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.4716280368937374e-08, | |
| "loss": 2.3922, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.540698976388004e-09, | |
| "loss": 2.7093, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 8.010923930238477e-08, | |
| "loss": 2.4296, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.3533422832715752e-07, | |
| "loss": 2.6948, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 2.4912426471710205, | |
| "eval_runtime": 30.6235, | |
| "eval_samples_per_second": 20.932, | |
| "eval_steps_per_second": 2.645, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.1246491964533778e-05, | |
| "loss": 2.424, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.2598018517843792e-05, | |
| "loss": 2.5084, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.4018079085430647e-05, | |
| "loss": 2.3701, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5504935827871155e-05, | |
| "loss": 2.4517, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.7056769162026474e-05, | |
| "loss": 2.6206, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.867167998781036e-05, | |
| "loss": 2.5704, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.0347692012266013e-05, | |
| "loss": 2.4849, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.208275416810954e-05, | |
| "loss": 2.5616, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.3874743123783855e-05, | |
| "loss": 2.3109, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.572146588194454e-05, | |
| "loss": 2.4892, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 2.7620662463202956e-05, | |
| "loss": 2.443, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.957000867184013e-05, | |
| "loss": 2.4545, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.156711894010725e-05, | |
| "loss": 2.4657, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.360954924763191e-05, | |
| "loss": 2.1432, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.569480011235739e-05, | |
| "loss": 2.5793, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.782031964935521e-05, | |
| "loss": 2.3062, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.998350669376499e-05, | |
| "loss": 2.3345, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.218171398404682e-05, | |
| "loss": 2.5329, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.4412251401639226e-05, | |
| "loss": 2.201, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.667238926307116e-05, | |
| "loss": 2.5198, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.895936166048787e-05, | |
| "loss": 2.5241, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.1270369846509574e-05, | |
| "loss": 2.4079, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.360258565927803e-05, | |
| "loss": 2.5147, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.595315498349983e-05, | |
| "loss": 2.5063, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.83192012432515e-05, | |
| "loss": 2.3742, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 6.069782892226903e-05, | |
| "loss": 2.625, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 6.308612710742091e-05, | |
| "loss": 2.4382, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 6.548117305101883e-05, | |
| "loss": 2.4238, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 6.788003574761404e-05, | |
| "loss": 2.4157, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.027977952089649e-05, | |
| "loss": 2.5458, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.267746761631506e-05, | |
| "loss": 2.5048, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.507016579501221e-05, | |
| "loss": 2.5779, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.745494592468267e-05, | |
| "loss": 2.5149, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.98288895629588e-05, | |
| "loss": 2.4972, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.218909152893547e-05, | |
| "loss": 2.2248, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.453266345847049e-05, | |
| "loss": 2.3945, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.685673733890025e-05, | |
| "loss": 2.4223, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 8.915846901885212e-05, | |
| "loss": 2.5513, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.143504168885545e-05, | |
| "loss": 2.527, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.368366932849205e-05, | |
| "loss": 2.6526, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.590160011586822e-05, | |
| "loss": 2.6926, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 9.808611979523307e-05, | |
| "loss": 2.4965, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.00010023455499862926, | |
| "loss": 2.4985, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00010234427651749925, | |
| "loss": 2.6502, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.00010441270252025704, | |
| "loss": 2.5715, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.00010643730171187608, | |
| "loss": 2.1817, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.00010841559643163376, | |
| "loss": 2.3754, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.00011034516568521922, | |
| "loss": 2.8706, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.0001122236481074937, | |
| "loss": 2.6822, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.0001140487448522782, | |
| "loss": 2.5523, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.0001158182224056317, | |
| "loss": 2.4356, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.00011752991531917766, | |
| "loss": 2.5516, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.00011918172886013147, | |
| "loss": 2.4273, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.00012077164157479157, | |
| "loss": 2.2673, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.00012229770776234816, | |
| "loss": 2.6163, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.000123758059855992, | |
| "loss": 2.5445, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.00012515091070840103, | |
| "loss": 2.3948, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.00012647455577881304, | |
| "loss": 2.5891, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.00012772737521900558, | |
| "loss": 2.5894, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 0.00012890783585563144, | |
| "loss": 2.5051, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.0001300144930664832, | |
| "loss": 2.4751, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.00013104599254838963, | |
| "loss": 2.4493, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.00013200107197458417, | |
| "loss": 2.4946, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.0001328785625395121, | |
| "loss": 2.4618, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.00013367739038918988, | |
| "loss": 2.5223, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.0001343965779353643, | |
| "loss": 2.4409, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.0001350352450518637, | |
| "loss": 2.3083, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.00013559261015167785, | |
| "loss": 2.4411, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0001360679911434468, | |
| "loss": 2.3657, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.00013646080626618978, | |
| "loss": 2.4467, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0001367705748012514, | |
| "loss": 2.3811, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0001369969176605951, | |
| "loss": 2.5053, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.00013713955785072274, | |
| "loss": 2.7194, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0001371983208116533, | |
| "loss": 2.5204, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 0.0001371731346305456, | |
| "loss": 2.5025, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.00013706403012970347, | |
| "loss": 2.4982, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 0.00013687114082885652, | |
| "loss": 2.7208, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.00013659470278176106, | |
| "loss": 2.5839, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.00013623505428732318, | |
| "loss": 2.5333, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.0001357926354755953, | |
| "loss": 2.5095, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.0001352679877691551, | |
| "loss": 2.2663, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.00013466175322052366, | |
| "loss": 2.5816, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.00013397467372643594, | |
| "loss": 2.4828, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.00013320759011992343, | |
| "loss": 2.5478, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.00013236144114132077, | |
| "loss": 2.5125, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0001314372622894558, | |
| "loss": 2.5845, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.00013043618455442838, | |
| "loss": 2.6289, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 2.2795188426971436, | |
| "eval_runtime": 30.9908, | |
| "eval_samples_per_second": 22.329, | |
| "eval_steps_per_second": 2.807, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.000135403446735711, | |
| "loss": 2.1728, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.00013482119897697306, | |
| "loss": 2.2896, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 2.07898211479187, | |
| "eval_runtime": 29.7025, | |
| "eval_samples_per_second": 22.523, | |
| "eval_steps_per_second": 2.828, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 0.0001341589832217866, | |
| "loss": 2.2492, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 0.00013341759915466326, | |
| "loss": 2.4062, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 0.00013259794206290322, | |
| "loss": 2.4505, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 0.00013170100175545615, | |
| "loss": 2.3168, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 0.00013072786136763862, | |
| "loss": 2.4331, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 0.0001296796960531519, | |
| "loss": 2.3448, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 0.00012855777156497896, | |
| "loss": 2.5707, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.00012736344272687528, | |
| "loss": 2.3018, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 0.00012609815179729852, | |
| "loss": 2.2497, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 0.00012476342672775167, | |
| "loss": 2.2297, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 0.00012336087931764827, | |
| "loss": 2.5308, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 0.00012189220326791867, | |
| "loss": 2.4956, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 0.00012035917213571493, | |
| "loss": 2.259, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 0.0001187636371926859, | |
| "loss": 2.5147, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 0.00011710752518939737, | |
| "loss": 2.5795, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 0.00011539283602861218, | |
| "loss": 2.4607, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00011362164035022977, | |
| "loss": 2.4427, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 0.00011179607703080082, | |
| "loss": 2.3976, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 0.00010991835060064675, | |
| "loss": 2.3127, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.00010799072858169134, | |
| "loss": 2.1402, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 0.0001060155387492263, | |
| "loss": 2.319, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 0.00010399516632091525, | |
| "loss": 2.6207, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 0.000101932051076431, | |
| "loss": 2.5146, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 9.982868441120418e-05, | |
| "loss": 2.513, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 9.768760632784142e-05, | |
| "loss": 2.5331, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 9.551140236884636e-05, | |
| "loss": 2.283, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 9.330270049434448e-05, | |
| "loss": 2.5398, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 9.106416790859171e-05, | |
| "loss": 2.4427, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 8.879850783908373e-05, | |
| "loss": 2.4624, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 8.650845627216773e-05, | |
| "loss": 2.3922, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 8.419677864910008e-05, | |
| "loss": 2.3814, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 8.186626652652247e-05, | |
| "loss": 2.4956, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 7.951973420541575e-05, | |
| "loss": 2.4434, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 7.716001533257615e-05, | |
| "loss": 2.2408, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.478995947873418e-05, | |
| "loss": 2.333, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 7.241242869744915e-05, | |
| "loss": 2.412, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 7.00302940689176e-05, | |
| "loss": 2.1938, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 6.764643223289375e-05, | |
| "loss": 2.4952, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 6.526372191488815e-05, | |
| "loss": 2.346, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 6.288504044985084e-05, | |
| "loss": 2.4523, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.051326030753298e-05, | |
| "loss": 2.3246, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 5.815124562372384e-05, | |
| "loss": 2.3573, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 5.580184874155164e-05, | |
| "loss": 2.5265, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 5.346790676702522e-05, | |
| "loss": 2.4817, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 5.115223814297577e-05, | |
| "loss": 2.3745, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 4.885763924553609e-05, | |
| "loss": 2.3791, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 4.658688100726727e-05, | |
| "loss": 2.438, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 4.434270557101133e-05, | |
| "loss": 2.2344, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 4.212782297850713e-05, | |
| "loss": 2.3451, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.9944907897778607e-05, | |
| "loss": 2.4611, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.7796596393228573e-05, | |
| "loss": 2.4182, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.568548274236061e-05, | |
| "loss": 2.5135, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.361411630295785e-05, | |
| "loss": 2.5817, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.158499843450252e-05, | |
| "loss": 2.3341, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.9600579477565787e-05, | |
| "loss": 2.3179, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 2.7663255794801203e-05, | |
| "loss": 2.29, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 2.57753668771237e-05, | |
| "loss": 2.5001, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 2.3939192518565938e-05, | |
| "loss": 2.5561, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.215695006322396e-05, | |
| "loss": 2.4063, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 2.0430791727616646e-05, | |
| "loss": 2.377, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.8762802001692604e-05, | |
| "loss": 2.6203, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 1.715499513162315e-05, | |
| "loss": 2.434, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 1.5609312687419068e-05, | |
| "loss": 2.2921, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.4127621218315069e-05, | |
| "loss": 2.2474, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 1.2711709998742125e-05, | |
| "loss": 2.4056, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 1.1363288867620807e-05, | |
| "loss": 2.1388, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.0083986163578448e-05, | |
| "loss": 2.4683, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 8.875346758583873e-06, | |
| "loss": 2.2937, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 7.73883019237949e-06, | |
| "loss": 2.4202, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 6.6758089099546965e-06, | |
| "loss": 2.5003, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 5.6875666041964786e-06, | |
| "loss": 2.3068, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 4.775296665718002e-06, | |
| "loss": 2.2147, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.940100741730249e-06, | |
| "loss": 2.4371, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.1829874057071e-06, | |
| "loss": 2.3255, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.504870939442012e-06, | |
| "loss": 2.3511, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 1.9065702289715435e-06, | |
| "loss": 2.587, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.3888077756974934e-06, | |
| "loss": 2.4066, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 9.522088239021213e-07, | |
| "loss": 2.3034, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 5.973006057099719e-07, | |
| "loss": 2.2354, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.2451170440812575e-07, | |
| "loss": 2.4044, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.3417153689369494e-07, | |
| "loss": 2.2793, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.650995587359215e-08, | |
| "loss": 2.3649, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.6569722969137058e-09, | |
| "loss": 2.5027, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 5.964259835509249e-08, | |
| "loss": 2.3313, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 2.0039681123962964e-07, | |
| "loss": 2.3823, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 4.2374963770060374e-07, | |
| "loss": 2.3593, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 7.294313593049221e-07, | |
| "loss": 2.4781, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.1170728381450536e-06, | |
| "loss": 2.4511, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.5862059626056908e-06, | |
| "loss": 2.0358, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 2.1362642126510735e-06, | |
| "loss": 2.4982, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 2.0775227546691895, | |
| "eval_runtime": 29.8853, | |
| "eval_samples_per_second": 22.386, | |
| "eval_steps_per_second": 2.811, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.9689817301781333e-07, | |
| "loss": 2.3835, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 5.860097463115006e-08, | |
| "loss": 2.2235, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 1.628030220611354e-09, | |
| "loss": 2.1558, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.9903991222381592, | |
| "eval_runtime": 27.7097, | |
| "eval_samples_per_second": 22.844, | |
| "eval_steps_per_second": 2.887, | |
| "step": 1824 | |
| } | |
| ], | |
| "max_steps": 2280, | |
| "num_train_epochs": 5, | |
| "total_flos": 1902336638976000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |