| { | |
| "best_metric": 2.6586830615997314, | |
| "best_model_checkpoint": "output/drake/checkpoint-1755", | |
| "epoch": 5.0, | |
| "global_step": 1755, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001371432064088578, | |
| "loss": 4.0588, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00013697291967368176, | |
| "loss": 3.874, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000136689421753516, | |
| "loss": 3.7354, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001362931820613334, | |
| "loss": 3.678, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00013578485668678617, | |
| "loss": 3.4762, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00013516528730985944, | |
| "loss": 3.9075, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000134435499807226, | |
| "loss": 3.8826, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00013359670255361022, | |
| "loss": 3.6832, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013265028442097337, | |
| "loss": 3.5663, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00013159781247883373, | |
| "loss": 3.875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00013044102939952857, | |
| "loss": 3.6027, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012918185057271526, | |
| "loss": 3.7132, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001278223609338884, | |
| "loss": 3.5338, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00012636481151216485, | |
| "loss": 3.8949, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0001248116157030527, | |
| "loss": 3.5964, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0001231653452723755, | |
| "loss": 3.7202, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00012142872609796847, | |
| "loss": 3.6172, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00011960463365619783, | |
| "loss": 3.4145, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00011769608826077622, | |
| "loss": 3.555, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000115706250061758, | |
| "loss": 3.4485, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00011363841381299484, | |
| "loss": 3.7157, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000111496003416716, | |
| "loss": 3.6139, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001092825662542656, | |
| "loss": 3.6335, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00010700176731238488, | |
| "loss": 3.8618, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00010465738311476428, | |
| "loss": 3.5889, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00010225329546891398, | |
| "loss": 3.5039, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.979348503870657e-05, | |
| "loss": 3.456, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.728202475323418e-05, | |
| "loss": 3.5593, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.472307306289397e-05, | |
| "loss": 3.6619, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.21208670538683e-05, | |
| "loss": 3.2545, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.94797154324004e-05, | |
| "loss": 3.5664, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.680399139048255e-05, | |
| "loss": 3.6659, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.409812536476914e-05, | |
| "loss": 3.4141, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.136659770070481e-05, | |
| "loss": 3.3853, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.86139312340139e-05, | |
| "loss": 3.5356, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.584468380183525e-05, | |
| "loss": 3.6393, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.306344069590158e-05, | |
| "loss": 3.7593, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.027480707026031e-05, | |
| "loss": 3.6784, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.748340031610621e-05, | |
| "loss": 3.4365, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.469384241635202e-05, | |
| "loss": 3.4692, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.191075229259615e-05, | |
| "loss": 3.5463, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.9138738157159314e-05, | |
| "loss": 3.4183, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.638238988285346e-05, | |
| "loss": 3.6126, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.364627140311692e-05, | |
| "loss": 3.5828, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.093491315510001e-05, | |
| "loss": 3.4454, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.825280457821314e-05, | |
| "loss": 3.5058, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.560438668055879e-05, | |
| "loss": 3.4436, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.299404468555545e-05, | |
| "loss": 3.3312, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.042610077092918e-05, | |
| "loss": 3.4519, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7904806912095865e-05, | |
| "loss": 3.2724, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.543433784178337e-05, | |
| "loss": 3.708, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.30187841375518e-05, | |
| "loss": 3.3732, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.06621454486565e-05, | |
| "loss": 3.6682, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.8368323873469696e-05, | |
| "loss": 3.3303, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.614111749842566e-05, | |
| "loss": 3.4287, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.39842141091877e-05, | |
| "loss": 3.4274, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.1901185084450396e-05, | |
| "loss": 3.3246, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.9895479482486794e-05, | |
| "loss": 3.4282, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.7970418330233093e-05, | |
| "loss": 3.2375, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.61291891243658e-05, | |
| "loss": 3.581, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4374840553477238e-05, | |
| "loss": 3.42, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2710277450088041e-05, | |
| "loss": 3.4236, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.1138255980854819e-05, | |
| "loss": 3.3657, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.661379082937688e-06, | |
| "loss": 3.1107, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.282092154083231e-06, | |
| "loss": 3.3076, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.002679003559785e-06, | |
| "loss": 3.3086, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.825258070649407e-06, | |
| "loss": 3.3643, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.7517789169574735e-06, | |
| "loss": 3.348, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.7840189983485763e-06, | |
| "loss": 3.391, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.9235807218530457e-06, | |
| "loss": 3.2735, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.171888792417796e-06, | |
| "loss": 3.576, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.5301878538942795e-06, | |
| "loss": 3.2583, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.995404281699138e-07, | |
| "loss": 3.5956, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.808251558551173e-07, | |
| "loss": 3.7098, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.7473534143910847e-07, | |
| "loss": 3.3973, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.177780532344056e-08, | |
| "loss": 3.3842, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.2720446338799106e-09, | |
| "loss": 3.2786, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.0890007647780969e-05, | |
| "loss": 3.2655, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2720089689346961e-05, | |
| "loss": 3.1689, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.4676757700644785e-05, | |
| "loss": 3.15, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.6755579199297876e-05, | |
| "loss": 3.3109, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.8951844985992176e-05, | |
| "loss": 3.3557, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.1260579812327114e-05, | |
| "loss": 3.2466, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.3676553651353433e-05, | |
| "loss": 3.4927, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.6194293545266464e-05, | |
| "loss": 3.5059, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8808096003415798e-05, | |
| "loss": 3.3394, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.151203992254596e-05, | |
| "loss": 2.9776, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.429999999999997e-05, | |
| "loss": 3.1045, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.716566060949963e-05, | |
| "loss": 3.3412, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.0102530108070535e-05, | |
| "loss": 3.3821, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.3103955541701554e-05, | |
| "loss": 3.2771, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.6163137716424864e-05, | |
| "loss": 3.2889, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.927314660067792e-05, | |
| "loss": 3.1803, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.242693702405331e-05, | |
| "loss": 3.3752, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.561736463687583e-05, | |
| "loss": 3.4172, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.883720209445263e-05, | |
| "loss": 3.1247, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.207915542933309e-05, | |
| "loss": 3.1979, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.533588057449125e-05, | |
| "loss": 3.2403, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.859999999999999e-05, | |
| "loss": 3.3824, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.186411942550872e-05, | |
| "loss": 3.2565, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.512084457066689e-05, | |
| "loss": 3.1016, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.836279790554734e-05, | |
| "loss": 3.4194, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.158263536312414e-05, | |
| "loss": 3.1877, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.477306297594667e-05, | |
| "loss": 3.4791, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.792685339932205e-05, | |
| "loss": 3.2052, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.103686228357512e-05, | |
| "loss": 3.4702, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.409604445829843e-05, | |
| "loss": 3.152, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.709746989192944e-05, | |
| "loss": 3.2925, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00010003433939050033, | |
| "loss": 3.2007, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00010290000000000001, | |
| "loss": 3.444, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00010568796007745401, | |
| "loss": 3.227, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00010839190399658417, | |
| "loss": 3.133, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011100570645473351, | |
| "loss": 3.1161, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011352344634864656, | |
| "loss": 3.2043, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00011593942018767285, | |
| "loss": 3.265, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011824815501400781, | |
| "loss": 3.3016, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00012044442080070208, | |
| "loss": 3.3944, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0001225232422993552, | |
| "loss": 3.4325, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00012447991031065301, | |
| "loss": 3.3161, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.000126309992352219, | |
| "loss": 3.2818, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00012800934269961248, | |
| "loss": 3.2362, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00012957411177772773, | |
| "loss": 3.1107, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00013100075488131993, | |
| "loss": 3.2821, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00013228604020490257, | |
| "loss": 3.3709, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00013342705616382626, | |
| "loss": 3.3293, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013442121798995453, | |
| "loss": 3.5064, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00013526627358699495, | |
| "loss": 3.2024, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013596030863222166, | |
| "loss": 3.378, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.000136501750913032, | |
| "loss": 3.215, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0001368893738885136, | |
| "loss": 3.2645, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00013712229946795436, | |
| "loss": 3.2699, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0001372, | |
| "loss": 2.9558, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.1339056491851807, | |
| "eval_runtime": 9.8661, | |
| "eval_samples_per_second": 45.611, | |
| "eval_steps_per_second": 5.777, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.0001369452804670866, | |
| "loss": 2.8801, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00013658418337916058, | |
| "loss": 3.0411, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.00013606719676147636, | |
| "loss": 3.1263, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.00013539550607801564, | |
| "loss": 3.2868, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.00013457065153327282, | |
| "loss": 3.0776, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00013359452454052534, | |
| "loss": 3.1467, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.00013246936338477295, | |
| "loss": 2.9318, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.00013119774809029134, | |
| "loss": 3.2317, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.00012978259450456783, | |
| "loss": 2.9684, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0001282271476121864, | |
| "loss": 3.2684, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.0001265349740939919, | |
| "loss": 3.1091, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.00012470995414859683, | |
| "loss": 3.0846, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.00012275627259498339, | |
| "loss": 3.2676, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.00012067840927660262, | |
| "loss": 3.1157, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.00011848112878897519, | |
| "loss": 3.2578, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00011616946955434783, | |
| "loss": 3.1545, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.00011374873226845785, | |
| "loss": 3.0167, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.00011122446774589795, | |
| "loss": 3.3236, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.00010860246419195128, | |
| "loss": 3.0855, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00010588873393008382, | |
| "loss": 3.1622, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.00010308949961552774, | |
| "loss": 3.1558, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.0001002111799665686, | |
| "loss": 3.1256, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.726037504625446e-05, | |
| "loss": 3.0624, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.424385112827677e-05, | |
| "loss": 3.127, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 9.1168525181726e-05, | |
| "loss": 3.2459, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 8.80414490102987e-05, | |
| "loss": 3.2356, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.486979308232529e-05, | |
| "loss": 3.1039, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.16608300886963e-05, | |
| "loss": 3.1498, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.842191826639015e-05, | |
| "loss": 3.2395, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.516048452584106e-05, | |
| "loss": 3.0791, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 7.18840074208375e-05, | |
| "loss": 2.9722, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.860000000000001e-05, | |
| "loss": 3.0485, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.531599257916255e-05, | |
| "loss": 3.26, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.203951547415898e-05, | |
| "loss": 3.0569, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.877808173360989e-05, | |
| "loss": 3.0239, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.553916991130374e-05, | |
| "loss": 3.1443, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.2330206917674755e-05, | |
| "loss": 3.1212, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.915855098970134e-05, | |
| "loss": 2.8549, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.603147481827398e-05, | |
| "loss": 2.9992, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.2956148871723224e-05, | |
| "loss": 2.8552, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.9939624953745524e-05, | |
| "loss": 3.0432, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.698882003343138e-05, | |
| "loss": 3.028, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.411050038447229e-05, | |
| "loss": 3.1615, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.131126606991618e-05, | |
| "loss": 3.1442, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.8597535808048707e-05, | |
| "loss": 3.0388, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.5975532254102027e-05, | |
| "loss": 2.9324, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.3451267731542143e-05, | |
| "loss": 2.9219, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.1030530445652233e-05, | |
| "loss": 2.8783, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.871887121102488e-05, | |
| "loss": 3.2507, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.6521590723397454e-05, | |
| "loss": 2.969, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.4443727405016684e-05, | |
| "loss": 2.9703, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.249004585140324e-05, | |
| "loss": 3.105, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.0665025906008154e-05, | |
| "loss": 3.1641, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 8.97285238781366e-06, | |
| "loss": 2.9782, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.417405495432199e-06, | |
| "loss": 3.0392, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 6.002251909708709e-06, | |
| "loss": 2.962, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.730636615227076e-06, | |
| "loss": 3.3658, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.605475459474696e-06, | |
| "loss": 2.9995, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.6293484667271903e-06, | |
| "loss": 3.1411, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.8044939219843706e-06, | |
| "loss": 3.2262, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.132803238523659e-06, | |
| "loss": 2.995, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.158166208394294e-07, | |
| "loss": 3.1868, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.547195329134365e-07, | |
| "loss": 3.0908, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 5.033997991443411e-08, | |
| "loss": 3.0958, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 3.012192964553833, | |
| "eval_runtime": 3.2423, | |
| "eval_samples_per_second": 143.415, | |
| "eval_steps_per_second": 18.197, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.1466095605587442e-09, | |
| "loss": 3.2577, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.1324763749765112e-07, | |
| "loss": 2.9737, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.803905991579816e-07, | |
| "loss": 2.7524, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 8.03962928668091e-07, | |
| "loss": 2.9894, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.382993363478614e-06, | |
| "loss": 2.8537, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.116154171494932e-06, | |
| "loss": 2.9448, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.001764195602037e-06, | |
| "loss": 2.8933, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.037792708602087e-06, | |
| "loss": 2.7567, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.22186406972573e-06, | |
| "loss": 2.8581, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 6.55126317203911e-06, | |
| "loss": 3.0563, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 8.022941668255849e-06, | |
| "loss": 3.1461, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 9.633524960678029e-06, | |
| "loss": 2.7869, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.1379319939237678e-05, | |
| "loss": 2.9909, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.3256323449896148e-05, | |
| "loss": 2.7969, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.526023147398194e-05, | |
| "loss": 3.0613, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.7386448997419335e-05, | |
| "loss": 2.9952, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.9630100547217162e-05, | |
| "loss": 2.8488, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 2.198604137105723e-05, | |
| "loss": 2.9286, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 2.444886923434745e-05, | |
| "loss": 2.9456, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 2.7012936807688628e-05, | |
| "loss": 2.9093, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 2.9672364616349706e-05, | |
| "loss": 2.6638, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.242105452205932e-05, | |
| "loss": 2.8781, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.5252703706197704e-05, | |
| "loss": 2.8149, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.816081912232647e-05, | |
| "loss": 2.9128, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 4.113873238491593e-05, | |
| "loss": 2.9003, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 4.4179615060128864e-05, | |
| "loss": 3.0516, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 4.727649432360055e-05, | |
| "loss": 2.9956, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 5.042226894930894e-05, | |
| "loss": 2.9518, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 5.3609725592873736e-05, | |
| "loss": 3.0925, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 5.68315553319459e-05, | |
| "loss": 2.9208, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 6.00803704257593e-05, | |
| "loss": 2.9595, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.334872125541636e-05, | |
| "loss": 2.7912, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 6.662911340606064e-05, | |
| "loss": 3.1123, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 6.991402485176834e-05, | |
| "loss": 3.0401, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.319592320375272e-05, | |
| "loss": 3.007, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 7.646728298233026e-05, | |
| "loss": 3.0709, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 7.97206028730452e-05, | |
| "loss": 2.8136, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.294842292738145e-05, | |
| "loss": 3.0615, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 8.6143341668622e-05, | |
| "loss": 2.8914, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 8.92980330636302e-05, | |
| "loss": 3.0546, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 9.240526332163816e-05, | |
| "loss": 2.7764, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 9.545790748151966e-05, | |
| "loss": 2.9967, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 9.844896574951464e-05, | |
| "loss": 2.8943, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 0.00010137157954994128, | |
| "loss": 2.7756, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.00010421904725209109, | |
| "loss": 2.6902, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 0.0001069848395372458, | |
| "loss": 3.0339, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 0.00010966261437057745, | |
| "loss": 2.9684, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 0.00011224623154360197, | |
| "loss": 2.993, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 0.0001147297667538396, | |
| "loss": 2.8925, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 0.0001171075251893971, | |
| "loss": 3.0326, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00011937405458732195, | |
| "loss": 3.1186, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 0.00012152415773578527, | |
| "loss": 2.9294, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 0.00012355290439142633, | |
| "loss": 3.0367, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.0001254556425845306, | |
| "loss": 2.6842, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 0.00012722800928611938, | |
| "loss": 2.807, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 0.0001288659404124906, | |
| "loss": 2.874, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 0.00013036568014427024, | |
| "loss": 2.9999, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 0.00013172378953860524, | |
| "loss": 3.0116, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 0.00013293715441475062, | |
| "loss": 3.1574, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 0.00013400299249496822, | |
| "loss": 2.6983, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 0.0001349188597843629, | |
| "loss": 2.8115, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00013568265617502779, | |
| "loss": 2.7446, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.0001362926302616468, | |
| "loss": 2.8728, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.0001367473833575137, | |
| "loss": 3.134, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 0.0001370458727017576, | |
| "loss": 2.8436, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 0.0001371874138504211, | |
| "loss": 2.901, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 3.0180249214172363, | |
| "eval_runtime": 3.2449, | |
| "eval_samples_per_second": 143.301, | |
| "eval_steps_per_second": 18.182, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.0001371716822459084, | |
| "loss": 2.7412, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 0.00013699871396120457, | |
| "loss": 2.5891, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 0.00013666890561715917, | |
| "loss": 3.1228, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 0.00013618301347302405, | |
| "loss": 2.9228, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 0.000135542151692331, | |
| "loss": 2.6346, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.0001347477897880855, | |
| "loss": 2.7065, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 0.00013380174925313403, | |
| "loss": 2.7945, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 0.00013270619938343336, | |
| "loss": 2.5638, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 0.00013146365230379744, | |
| "loss": 2.8108, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.0001300769572075284, | |
| "loss": 2.8114, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 0.00012854929382314167, | |
| "loss": 2.7208, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 0.0001268841651231636, | |
| "loss": 2.8124, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 0.00012508538929172376, | |
| "loss": 2.9135, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 0.0001231570909693571, | |
| "loss": 2.5881, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 0.00012110369179509431, | |
| "loss": 2.6689, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 0.00011892990026752702, | |
| "loss": 2.8313, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.0001166407009480955, | |
| "loss": 2.9475, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 0.0001142413430313578, | |
| "loss": 3.0502, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 0.00011173732830844787, | |
| "loss": 2.6531, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.00010913439855132291, | |
| "loss": 2.9528, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 0.00010643852234672852, | |
| "loss": 2.7701, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 0.00010365588141007188, | |
| "loss": 2.6251, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 0.00010079285641058387, | |
| "loss": 2.9979, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 9.785601234027734e-05, | |
| "loss": 2.807, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 9.485208346024522e-05, | |
| "loss": 2.8977, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 9.178795785882326e-05, | |
| "loss": 2.8219, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 8.867066165702077e-05, | |
| "loss": 2.6527, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 8.550734289743871e-05, | |
| "loss": 2.8599, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 8.230525515362062e-05, | |
| "loss": 2.9224, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 7.907174089741485e-05, | |
| "loss": 2.6402, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 7.581421466249422e-05, | |
| "loss": 2.658, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 7.254014604263352e-05, | |
| "loss": 2.6781, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 6.92570425637328e-05, | |
| "loss": 2.8161, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 6.597243246886372e-05, | |
| "loss": 2.7003, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 6.269384745580722e-05, | |
| "loss": 2.6335, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 5.942880540667308e-05, | |
| "loss": 2.9172, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 5.618479314919675e-05, | |
| "loss": 2.6901, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 5.2969249289244765e-05, | |
| "loss": 3.0955, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 4.978954715389626e-05, | |
| "loss": 2.8052, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.6652977884208274e-05, | |
| "loss": 2.5181, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 4.356673371643594e-05, | |
| "loss": 2.8346, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 4.0537891490046174e-05, | |
| "loss": 2.9267, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.757339642033588e-05, | |
| "loss": 2.6995, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.468004617287139e-05, | |
| "loss": 2.8038, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.1864475276260315e-05, | |
| "loss": 2.7572, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.9133139909002586e-05, | |
| "loss": 2.8888, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.6492303095302796e-05, | |
| "loss": 2.5902, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.3948020343789084e-05, | |
| "loss": 2.6841, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.1506125762073123e-05, | |
| "loss": 2.6346, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 1.917221867898604e-05, | |
| "loss": 2.574, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 1.695165080516989e-05, | |
| "loss": 2.742, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 1.4849513961463578e-05, | |
| "loss": 2.8388, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 1.2870628403223036e-05, | |
| "loss": 2.6697, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.101953176734824e-05, | |
| "loss": 2.7003, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 9.30046866736211e-06, | |
| "loss": 3.0252, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 7.717380960398831e-06, | |
| "loss": 2.8193, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 6.273898708422032e-06, | |
| "loss": 2.895, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 4.9733318543963394e-06, | |
| "loss": 2.7687, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 3.818662632500395e-06, | |
| "loss": 2.7358, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 2.812538729785672e-06, | |
| "loss": 2.7698, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 1.9572672149594433e-06, | |
| "loss": 2.7906, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 1.2548092482157443e-06, | |
| "loss": 2.6523, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 7.067755842429393e-07, | |
| "loss": 2.7697, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.144228787205802e-07, | |
| "loss": 2.7642, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 7.865080677495002e-08, | |
| "loss": 2.6786, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 2.8629, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 2.985200881958008, | |
| "eval_runtime": 3.2565, | |
| "eval_samples_per_second": 142.792, | |
| "eval_steps_per_second": 18.118, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.064756856761554e-05, | |
| "loss": 2.7681, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.8129056814413798e-05, | |
| "loss": 2.7734, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.5691584488661662e-05, | |
| "loss": 2.8042, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.3340032409686368e-05, | |
| "loss": 2.8781, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.1079109349241507e-05, | |
| "loss": 2.8162, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 1.891334260261341e-05, | |
| "loss": 2.523, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 1.6847068923119758e-05, | |
| "loss": 2.8495, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 1.4884425838151134e-05, | |
| "loss": 2.791, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 1.3029343364143164e-05, | |
| "loss": 2.6161, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.1285536137072603e-05, | |
| "loss": 2.8099, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 9.65649597423201e-06, | |
| "loss": 2.9416, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 8.145484882179486e-06, | |
| "loss": 2.881, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 6.755528524864052e-06, | |
| "loss": 2.4485, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 5.4894101650047195e-06, | |
| "loss": 2.7043, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 4.349665090857373e-06, | |
| "loss": 2.9465, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.3385755395280496e-06, | |
| "loss": 2.8094, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 2.4581661269975954e-06, | |
| "loss": 2.8454, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 1.7101997940107123e-06, | |
| "loss": 2.717, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 1.0961742759449984e-06, | |
| "loss": 2.6812, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 6.173191037306866e-07, | |
| "loss": 2.7196, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 2.7459314182545106e-07, | |
| "loss": 2.8512, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 6.868266817434572e-08, | |
| "loss": 2.8381, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 2.7245, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 2.6586830615997314, | |
| "eval_runtime": 12.7857, | |
| "eval_samples_per_second": 43.33, | |
| "eval_steps_per_second": 5.475, | |
| "step": 1755 | |
| } | |
| ], | |
| "max_steps": 2106, | |
| "num_train_epochs": 6, | |
| "total_flos": 1831003914240000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |