| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.035964035964036, | |
| "global_step": 2038, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0013736056394868903, | |
| "loss": 1.4723, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0019425717247145285, | |
| "loss": 1.568, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0023791547571544325, | |
| "loss": 1.2706, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0027472112789737805, | |
| "loss": 1.3258, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.003071475584169756, | |
| "loss": 1.2558, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0033646329245522657, | |
| "loss": 0.8074, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0036342189215581556, | |
| "loss": 1.4713, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.003885143449429057, | |
| "loss": 1.2681, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.004120816918460671, | |
| "loss": 1.2291, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.004343722427630694, | |
| "loss": 1.306, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.004555734516094203, | |
| "loss": 1.1953, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.004758309514308865, | |
| "loss": 1.2176, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0049526055654364855, | |
| "loss": 1.2338, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.005139561687500467, | |
| "loss": 1.381, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.005319951765989316, | |
| "loss": 1.5546, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.005494422557947561, | |
| "loss": 1.5126, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.005663521139548541, | |
| "loss": 1.5132, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.005827715174143585, | |
| "loss": 1.2736, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.005987408170800916, | |
| "loss": 1.1807, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.006142951168339512, | |
| "loss": 1.392, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0062946518179668965, | |
| "loss": 1.4068, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0064427815392316505, | |
| "loss": 1.2436, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0065875812264513545, | |
| "loss": 1.6382, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.006729265849104531, | |
| "loss": 1.5855, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.006868028197434452, | |
| "loss": 1.8619, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.007004041959724749, | |
| "loss": 1.4162, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.007137464271463298, | |
| "loss": 1.397, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.007268437843116311, | |
| "loss": 1.376, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0073970927486462865, | |
| "loss": 1.5437, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.007523547938632788, | |
| "loss": 1.8505, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0076479125281174514, | |
| "loss": 1.3138, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.007770286898858114, | |
| "loss": 1.6157, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00789076364767037, | |
| "loss": 1.4061, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.008009428406336274, | |
| "loss": 1.4829, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.008126360553720012, | |
| "loss": 1.3897, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.008241633836921341, | |
| "loss": 1.3801, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.008355316915277182, | |
| "loss": 1.6133, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.008467473838610143, | |
| "loss": 1.6826, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.008578164469184382, | |
| "loss": 1.6101, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.008687444855261389, | |
| "loss": 1.7164, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.008795367562872955, | |
| "loss": 1.6603, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.008901981971385245, | |
| "loss": 1.3604, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.009007334537569819, | |
| "loss": 1.5517, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.009111469032188405, | |
| "loss": 1.6048, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.009214426752509268, | |
| "loss": 1.6281, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.009316246713681893, | |
| "loss": 2.1204, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.009416965821485117, | |
| "loss": 1.6439, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00951661902861773, | |
| "loss": 1.6982, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.009615239476408233, | |
| "loss": 1.857, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.009712858623572643, | |
| "loss": 1.5612, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.009809506363438459, | |
| "loss": 1.4843, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.009905211130872971, | |
| "loss": 1.81, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.01, | |
| "loss": 1.5551, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.009971799492385787, | |
| "loss": 1.8817, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.009943598984771575, | |
| "loss": 1.7807, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.009915398477157361, | |
| "loss": 1.7788, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.009887197969543148, | |
| "loss": 1.9268, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.009858997461928934, | |
| "loss": 1.767, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00983079695431472, | |
| "loss": 2.0681, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.009802596446700508, | |
| "loss": 1.8078, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.009774395939086295, | |
| "loss": 1.5819, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.009746195431472081, | |
| "loss": 1.3608, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00971799492385787, | |
| "loss": 1.4035, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.009689794416243654, | |
| "loss": 1.7838, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.009661593908629442, | |
| "loss": 1.462, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.009633393401015228, | |
| "loss": 1.8537, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.009605192893401015, | |
| "loss": 1.6035, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.009576992385786803, | |
| "loss": 1.7534, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00954879187817259, | |
| "loss": 1.5117, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.009520591370558376, | |
| "loss": 1.6606, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.009492390862944162, | |
| "loss": 2.2899, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.009464190355329949, | |
| "loss": 1.6254, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.009435989847715737, | |
| "loss": 1.7699, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.009407789340101523, | |
| "loss": 1.5836, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00937958883248731, | |
| "loss": 1.9447, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.009351388324873098, | |
| "loss": 1.7334, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.009323187817258882, | |
| "loss": 1.5246, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00929498730964467, | |
| "loss": 1.7258, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.009266786802030457, | |
| "loss": 1.6757, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.009238586294416243, | |
| "loss": 1.7912, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.009210385786802031, | |
| "loss": 1.8435, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.009182185279187818, | |
| "loss": 2.0267, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.009153984771573604, | |
| "loss": 1.6059, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00912578426395939, | |
| "loss": 1.7753, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.009097583756345177, | |
| "loss": 1.988, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.009069383248730965, | |
| "loss": 1.5201, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.009041182741116751, | |
| "loss": 1.397, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.009012982233502538, | |
| "loss": 1.7266, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.008984781725888326, | |
| "loss": 1.6747, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.008956581218274112, | |
| "loss": 1.5811, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.008928380710659899, | |
| "loss": 1.4325, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.008900180203045685, | |
| "loss": 1.9487, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.008871979695431471, | |
| "loss": 1.4969, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00884377918781726, | |
| "loss": 1.6207, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.008815578680203046, | |
| "loss": 1.9026, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.008787378172588832, | |
| "loss": 1.6797, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00875917766497462, | |
| "loss": 1.5079, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.008730977157360405, | |
| "loss": 1.5912, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.008702776649746193, | |
| "loss": 1.706, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00867457614213198, | |
| "loss": 1.6559, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.008646375634517766, | |
| "loss": 1.6109, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.008618175126903554, | |
| "loss": 1.9024, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00858997461928934, | |
| "loss": 1.4508, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.008561774111675127, | |
| "loss": 1.7011, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.008533573604060913, | |
| "loss": 1.4233, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0085053730964467, | |
| "loss": 1.5886, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.008477172588832488, | |
| "loss": 1.6088, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.008448972081218274, | |
| "loss": 2.3507, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00842077157360406, | |
| "loss": 1.457, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.008392571065989849, | |
| "loss": 1.7145, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.008364370558375635, | |
| "loss": 1.6431, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.008336170050761421, | |
| "loss": 1.6156, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.008307969543147208, | |
| "loss": 1.4988, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.008279769035532994, | |
| "loss": 1.6655, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.008251568527918782, | |
| "loss": 1.3964, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.008223368020304569, | |
| "loss": 1.5442, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.008195167512690355, | |
| "loss": 1.576, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.008166967005076143, | |
| "loss": 1.6838, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.008138766497461928, | |
| "loss": 1.5472, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.008110565989847716, | |
| "loss": 1.6603, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.008082365482233502, | |
| "loss": 1.8896, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.008054164974619289, | |
| "loss": 1.3879, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.008025964467005077, | |
| "loss": 1.7506, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.007997763959390863, | |
| "loss": 1.6107, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00796956345177665, | |
| "loss": 1.6015, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.007941362944162436, | |
| "loss": 1.4905, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.007913162436548223, | |
| "loss": 1.7908, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00788496192893401, | |
| "loss": 1.5734, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.007856761421319797, | |
| "loss": 1.4382, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.007828560913705583, | |
| "loss": 1.5483, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00780036040609137, | |
| "loss": 1.4697, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.007772159898477159, | |
| "loss": 1.5878, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.007743959390862944, | |
| "loss": 1.9756, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.007715758883248732, | |
| "loss": 1.6656, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.007687558375634519, | |
| "loss": 1.6744, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.007659357868020306, | |
| "loss": 1.762, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0076311573604060925, | |
| "loss": 1.6982, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.007602956852791878, | |
| "loss": 1.4908, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.007574756345177666, | |
| "loss": 1.6065, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0075465558375634526, | |
| "loss": 1.2293, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.007518355329949239, | |
| "loss": 1.6052, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.007490154822335025, | |
| "loss": 1.3118, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0074619543147208135, | |
| "loss": 1.1742, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0074337538071066, | |
| "loss": 1.632, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.007405553299492385, | |
| "loss": 1.6181, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.007377352791878174, | |
| "loss": 1.6726, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00734915228426396, | |
| "loss": 1.5759, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.007320951776649747, | |
| "loss": 1.3197, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0072927512690355335, | |
| "loss": 1.8665, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.007264550761421322, | |
| "loss": 1.1938, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.007236350253807108, | |
| "loss": 1.2782, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0072081497461928936, | |
| "loss": 1.6276, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.007179949238578682, | |
| "loss": 1.5382, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.007151748730964467, | |
| "loss": 1.5528, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0071235482233502545, | |
| "loss": 1.663, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.007095347715736042, | |
| "loss": 1.5098, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.007067147208121828, | |
| "loss": 1.781, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.007038946700507615, | |
| "loss": 1.3787, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.007010746192893401, | |
| "loss": 2.1853, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.006982545685279189, | |
| "loss": 1.6639, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.006954345177664975, | |
| "loss": 1.7152, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.006926144670050761, | |
| "loss": 1.8148, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.006897944162436549, | |
| "loss": 1.6564, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0068697436548223354, | |
| "loss": 1.3046, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.006841543147208123, | |
| "loss": 1.3462, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.006813342639593909, | |
| "loss": 1.4969, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.006785142131979697, | |
| "loss": 1.7367, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.006756941624365483, | |
| "loss": 1.5385, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00672874111675127, | |
| "loss": 1.5918, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.006700540609137057, | |
| "loss": 1.4092, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.006672340101522844, | |
| "loss": 1.6609, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.006644139593908631, | |
| "loss": 1.6529, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.006615939086294417, | |
| "loss": 2.0543, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0065877385786802046, | |
| "loss": 1.7507, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00655953807106599, | |
| "loss": 1.3364, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0065313375634517765, | |
| "loss": 1.5436, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.006503137055837565, | |
| "loss": 1.5808, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00647493654822335, | |
| "loss": 1.1435, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.006446736040609138, | |
| "loss": 1.7857, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.006418535532994924, | |
| "loss": 1.506, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.006390335025380713, | |
| "loss": 1.5276, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.006362134517766498, | |
| "loss": 1.7656, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.006333934010152285, | |
| "loss": 1.2465, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.006305733502538073, | |
| "loss": 1.1414, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.006277532994923858, | |
| "loss": 1.7263, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0062493324873096456, | |
| "loss": 1.6158, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.006221131979695433, | |
| "loss": 1.3896, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00619293147208122, | |
| "loss": 1.8046, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.006164730964467006, | |
| "loss": 1.6806, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.006136530456852793, | |
| "loss": 1.7003, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.006108329949238579, | |
| "loss": 1.4235, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.006080129441624366, | |
| "loss": 1.5528, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.006051928934010154, | |
| "loss": 1.6251, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.006023728426395939, | |
| "loss": 1.7944, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0059955279187817265, | |
| "loss": 1.6361, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.005967327411167512, | |
| "loss": 1.6142, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0059391269035533, | |
| "loss": 1.6014, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.005910926395939088, | |
| "loss": 1.5756, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.005882725888324872, | |
| "loss": 1.2877, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.005854525380710661, | |
| "loss": 1.6494, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.005826324873096447, | |
| "loss": 1.4704, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.005798124365482234, | |
| "loss": 1.3735, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.005769923857868021, | |
| "loss": 1.3836, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.005741723350253808, | |
| "loss": 1.4496, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.005713522842639595, | |
| "loss": 1.4613, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00568532233502538, | |
| "loss": 1.4161, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0056571218274111676, | |
| "loss": 1.592, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.005628921319796955, | |
| "loss": 1.585, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.005600720812182741, | |
| "loss": 1.2444, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.005572520304568528, | |