| { | |
| "best_metric": 0.978066110596231, | |
| "best_model_checkpoint": "spa_images_classifier_jd_v1_convnext/checkpoint-1594", | |
| "epoch": 9.967069154774972, | |
| "eval_steps": 500, | |
| "global_step": 2270, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.1013215859030839e-06, | |
| "loss": 1.4923, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.2026431718061677e-06, | |
| "loss": 1.4857, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.303964757709251e-06, | |
| "loss": 1.4597, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.4052863436123355e-06, | |
| "loss": 1.3989, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.506607929515419e-06, | |
| "loss": 1.3587, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.607929515418502e-06, | |
| "loss": 1.2912, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.709251101321587e-06, | |
| "loss": 1.2097, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.810572687224671e-06, | |
| "loss": 1.1577, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.911894273127754e-06, | |
| "loss": 1.0664, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.1013215859030838e-05, | |
| "loss": 0.9747, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.211453744493392e-05, | |
| "loss": 0.8517, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.3215859030837005e-05, | |
| "loss": 0.7492, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.4317180616740089e-05, | |
| "loss": 0.6764, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5418502202643173e-05, | |
| "loss": 0.6508, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.6519823788546258e-05, | |
| "loss": 0.5314, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.7621145374449342e-05, | |
| "loss": 0.4931, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8722466960352426e-05, | |
| "loss": 0.4168, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.9823788546255507e-05, | |
| "loss": 0.4433, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.092511013215859e-05, | |
| "loss": 0.3664, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.2026431718061676e-05, | |
| "loss": 0.3411, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.3127753303964757e-05, | |
| "loss": 0.356, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.422907488986784e-05, | |
| "loss": 0.3499, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.533039647577093e-05, | |
| "loss": 0.2877, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.643171806167401e-05, | |
| "loss": 0.3416, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.753303964757709e-05, | |
| "loss": 0.3564, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.8634361233480178e-05, | |
| "loss": 0.363, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.973568281938326e-05, | |
| "loss": 0.3397, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.0837004405286347e-05, | |
| "loss": 0.2825, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.1938325991189424e-05, | |
| "loss": 0.2888, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.3039647577092515e-05, | |
| "loss": 0.2989, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.414096916299559e-05, | |
| "loss": 0.2703, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.5242290748898684e-05, | |
| "loss": 0.3281, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.634361233480176e-05, | |
| "loss": 0.2862, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.744493392070485e-05, | |
| "loss": 0.2819, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.854625550660793e-05, | |
| "loss": 0.2488, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.9647577092511014e-05, | |
| "loss": 0.2569, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.07488986784141e-05, | |
| "loss": 0.2378, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.185022026431718e-05, | |
| "loss": 0.3059, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.295154185022027e-05, | |
| "loss": 0.2922, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.405286343612335e-05, | |
| "loss": 0.2422, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.5154185022026436e-05, | |
| "loss": 0.2781, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.625550660792951e-05, | |
| "loss": 0.2178, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.7356828193832604e-05, | |
| "loss": 0.2647, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.845814977973568e-05, | |
| "loss": 0.2728, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.955947136563877e-05, | |
| "loss": 0.2494, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9555143651529193, | |
| "eval_loss": 0.11944440752267838, | |
| "eval_runtime": 6.3116, | |
| "eval_samples_per_second": 512.864, | |
| "eval_steps_per_second": 16.161, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.9926578560939794e-05, | |
| "loss": 0.2155, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.980420949583945e-05, | |
| "loss": 0.2348, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.9681840430739115e-05, | |
| "loss": 0.2388, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.955947136563877e-05, | |
| "loss": 0.2387, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.943710230053842e-05, | |
| "loss": 0.1989, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.931473323543808e-05, | |
| "loss": 0.2726, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.919236417033774e-05, | |
| "loss": 0.2197, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.9069995105237395e-05, | |
| "loss": 0.1965, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.894762604013706e-05, | |
| "loss": 0.2256, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.8825256975036717e-05, | |
| "loss": 0.2344, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.870288790993637e-05, | |
| "loss": 0.208, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.8580518844836024e-05, | |
| "loss": 0.2337, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.845814977973568e-05, | |
| "loss": 0.2424, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.8335780714635346e-05, | |
| "loss": 0.2248, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.8213411649535e-05, | |
| "loss": 0.1925, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.809104258443466e-05, | |
| "loss": 0.2311, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.796867351933432e-05, | |
| "loss": 0.2306, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.784630445423397e-05, | |
| "loss": 0.2249, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.7723935389133625e-05, | |
| "loss": 0.2014, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.760156632403329e-05, | |
| "loss": 0.213, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.747919725893295e-05, | |
| "loss": 0.2185, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.7356828193832604e-05, | |
| "loss": 0.2365, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.723445912873226e-05, | |
| "loss": 0.2222, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.711209006363191e-05, | |
| "loss": 0.2154, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.698972099853157e-05, | |
| "loss": 0.2211, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.686735193343123e-05, | |
| "loss": 0.1725, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.674498286833089e-05, | |
| "loss": 0.2174, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.662261380323055e-05, | |
| "loss": 0.2446, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.6500244738130205e-05, | |
| "loss": 0.1818, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.6377875673029856e-05, | |
| "loss": 0.214, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.625550660792951e-05, | |
| "loss": 0.2328, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.613313754282918e-05, | |
| "loss": 0.2407, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.6010768477728834e-05, | |
| "loss": 0.1649, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.588839941262849e-05, | |
| "loss": 0.2018, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.576603034752815e-05, | |
| "loss": 0.2064, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.56436612824278e-05, | |
| "loss": 0.1923, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.5521292217327464e-05, | |
| "loss": 0.2148, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.539892315222712e-05, | |
| "loss": 0.1885, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.527655408712678e-05, | |
| "loss": 0.2078, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.5154185022026436e-05, | |
| "loss": 0.2378, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.503181595692609e-05, | |
| "loss": 0.1933, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.4909446891825743e-05, | |
| "loss": 0.1607, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.478707782672541e-05, | |
| "loss": 0.1923, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.4664708761625065e-05, | |
| "loss": 0.1835, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.454233969652472e-05, | |
| "loss": 0.1758, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.441997063142438e-05, | |
| "loss": 0.2333, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9635464936669756, | |
| "eval_loss": 0.10080890357494354, | |
| "eval_runtime": 6.5201, | |
| "eval_samples_per_second": 496.466, | |
| "eval_steps_per_second": 15.644, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.429760156632404e-05, | |
| "loss": 0.1713, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.417523250122369e-05, | |
| "loss": 0.1998, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.405286343612335e-05, | |
| "loss": 0.1998, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.393049437102301e-05, | |
| "loss": 0.1966, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.3808125305922666e-05, | |
| "loss": 0.2189, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.368575624082232e-05, | |
| "loss": 0.1618, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.356338717572198e-05, | |
| "loss": 0.1521, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.344101811062164e-05, | |
| "loss": 0.1753, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.3318649045521295e-05, | |
| "loss": 0.1915, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.319627998042095e-05, | |
| "loss": 0.2248, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.307391091532061e-05, | |
| "loss": 0.1692, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.295154185022027e-05, | |
| "loss": 0.1832, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.2829172785119924e-05, | |
| "loss": 0.2099, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.270680372001958e-05, | |
| "loss": 0.2225, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.258443465491924e-05, | |
| "loss": 0.1982, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.2462065589818896e-05, | |
| "loss": 0.1538, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.2339696524718554e-05, | |
| "loss": 0.2139, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.221732745961821e-05, | |
| "loss": 0.1922, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.209495839451787e-05, | |
| "loss": 0.182, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.1972589329417525e-05, | |
| "loss": 0.197, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.185022026431718e-05, | |
| "loss": 0.2057, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.172785119921684e-05, | |
| "loss": 0.1773, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.16054821341165e-05, | |
| "loss": 0.171, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.1483113069016155e-05, | |
| "loss": 0.1484, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.136074400391581e-05, | |
| "loss": 0.1552, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.123837493881547e-05, | |
| "loss": 0.1449, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.1116005873715127e-05, | |
| "loss": 0.2022, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.0993636808614784e-05, | |
| "loss": 0.1795, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.087126774351444e-05, | |
| "loss": 0.1998, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.07488986784141e-05, | |
| "loss": 0.1796, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.0626529613313756e-05, | |
| "loss": 0.1975, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.050416054821341e-05, | |
| "loss": 0.1638, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.038179148311307e-05, | |
| "loss": 0.2266, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.025942241801273e-05, | |
| "loss": 0.1445, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.0137053352912385e-05, | |
| "loss": 0.1611, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.001468428781204e-05, | |
| "loss": 0.132, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.98923152227117e-05, | |
| "loss": 0.1867, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.976994615761136e-05, | |
| "loss": 0.1607, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 3.9647577092511014e-05, | |
| "loss": 0.1225, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.952520802741067e-05, | |
| "loss": 0.1748, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.940283896231033e-05, | |
| "loss": 0.1689, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.9280469897209986e-05, | |
| "loss": 0.1475, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.915810083210964e-05, | |
| "loss": 0.1205, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.90357317670093e-05, | |
| "loss": 0.1689, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.8913362701908965e-05, | |
| "loss": 0.1977, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9703429101019463, | |
| "eval_loss": 0.08545960485935211, | |
| "eval_runtime": 6.4859, | |
| "eval_samples_per_second": 499.084, | |
| "eval_steps_per_second": 15.726, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.8790993636808615e-05, | |
| "loss": 0.1616, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.866862457170827e-05, | |
| "loss": 0.1489, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.854625550660793e-05, | |
| "loss": 0.1394, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.842388644150759e-05, | |
| "loss": 0.1451, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.8301517376407244e-05, | |
| "loss": 0.1603, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.817914831130691e-05, | |
| "loss": 0.1511, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.805677924620656e-05, | |
| "loss": 0.1835, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.7934410181106216e-05, | |
| "loss": 0.1328, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.7812041116005874e-05, | |
| "loss": 0.1751, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.768967205090553e-05, | |
| "loss": 0.1914, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.7567302985805195e-05, | |
| "loss": 0.1538, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.744493392070485e-05, | |
| "loss": 0.149, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.73225648556045e-05, | |
| "loss": 0.1764, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.720019579050416e-05, | |
| "loss": 0.2057, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.707782672540382e-05, | |
| "loss": 0.155, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.6955457660303475e-05, | |
| "loss": 0.1523, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.683308859520314e-05, | |
| "loss": 0.1561, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.6710719530102796e-05, | |
| "loss": 0.1758, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.658835046500245e-05, | |
| "loss": 0.1538, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.6465981399902104e-05, | |
| "loss": 0.1882, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.634361233480176e-05, | |
| "loss": 0.1073, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.622124326970142e-05, | |
| "loss": 0.1692, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.609887420460108e-05, | |
| "loss": 0.1762, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.597650513950074e-05, | |
| "loss": 0.1598, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.585413607440039e-05, | |
| "loss": 0.1442, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.573176700930005e-05, | |
| "loss": 0.1359, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.5609397944199705e-05, | |
| "loss": 0.1585, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.548702887909936e-05, | |
| "loss": 0.2017, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.5364659813999026e-05, | |
| "loss": 0.1459, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.5242290748898684e-05, | |
| "loss": 0.1096, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.5119921683798334e-05, | |
| "loss": 0.1507, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.499755261869799e-05, | |
| "loss": 0.1366, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.487518355359765e-05, | |
| "loss": 0.1524, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.4752814488497306e-05, | |
| "loss": 0.1505, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.463044542339697e-05, | |
| "loss": 0.1553, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.450807635829663e-05, | |
| "loss": 0.1373, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.4385707293196285e-05, | |
| "loss": 0.139, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.4263338228095935e-05, | |
| "loss": 0.1482, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.414096916299559e-05, | |
| "loss": 0.1684, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.401860009789526e-05, | |
| "loss": 0.2081, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.3896231032794914e-05, | |
| "loss": 0.1652, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.377386196769457e-05, | |
| "loss": 0.1833, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.365149290259423e-05, | |
| "loss": 0.1532, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.352912383749388e-05, | |
| "loss": 0.1541, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.3406754772393536e-05, | |
| "loss": 0.1298, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.32843857072932e-05, | |
| "loss": 0.1405, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9743589743589743, | |
| "eval_loss": 0.07923375070095062, | |
| "eval_runtime": 6.6056, | |
| "eval_samples_per_second": 490.041, | |
| "eval_steps_per_second": 15.442, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.316201664219286e-05, | |
| "loss": 0.1584, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.3039647577092515e-05, | |
| "loss": 0.1504, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.291727851199217e-05, | |
| "loss": 0.1893, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.279490944689182e-05, | |
| "loss": 0.1496, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.267254038179148e-05, | |
| "loss": 0.1576, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.2550171316691144e-05, | |
| "loss": 0.1291, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.24278022515908e-05, | |
| "loss": 0.146, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.230543318649046e-05, | |
| "loss": 0.134, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.2183064121390116e-05, | |
| "loss": 0.1643, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.206069505628977e-05, | |
| "loss": 0.1453, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.1938325991189424e-05, | |
| "loss": 0.1635, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.181595692608909e-05, | |
| "loss": 0.1194, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.1693587860988745e-05, | |
| "loss": 0.1532, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.15712187958884e-05, | |
| "loss": 0.1838, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.144884973078806e-05, | |
| "loss": 0.1468, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.132648066568771e-05, | |
| "loss": 0.1203, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.120411160058737e-05, | |
| "loss": 0.1789, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.108174253548703e-05, | |
| "loss": 0.1781, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.095937347038669e-05, | |
| "loss": 0.1299, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.0837004405286347e-05, | |
| "loss": 0.1647, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 3.0714635340186004e-05, | |
| "loss": 0.1811, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 3.059226627508566e-05, | |
| "loss": 0.1386, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.0469897209985315e-05, | |
| "loss": 0.1465, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.0347528144884972e-05, | |
| "loss": 0.1469, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 3.0225159079784633e-05, | |
| "loss": 0.1381, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 3.010279001468429e-05, | |
| "loss": 0.1184, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.9980420949583948e-05, | |
| "loss": 0.1596, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.9858051884483605e-05, | |
| "loss": 0.1305, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.973568281938326e-05, | |
| "loss": 0.1487, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.9613313754282916e-05, | |
| "loss": 0.1359, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.9490944689182577e-05, | |
| "loss": 0.1238, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.9368575624082234e-05, | |
| "loss": 0.1322, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.924620655898189e-05, | |
| "loss": 0.1385, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.9123837493881552e-05, | |
| "loss": 0.1432, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.9001468428781203e-05, | |
| "loss": 0.1494, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.887909936368086e-05, | |
| "loss": 0.1475, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.875673029858052e-05, | |
| "loss": 0.1282, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.8634361233480178e-05, | |
| "loss": 0.1606, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.8511992168379835e-05, | |
| "loss": 0.1333, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 2.8389623103279496e-05, | |
| "loss": 0.1556, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 2.8267254038179147e-05, | |
| "loss": 0.1432, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 2.8144884973078807e-05, | |
| "loss": 0.1776, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 2.8022515907978465e-05, | |
| "loss": 0.1029, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 2.7900146842878122e-05, | |
| "loss": 0.1382, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 0.1575, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9731232622798888, | |
| "eval_loss": 0.07335282117128372, | |
| "eval_runtime": 6.4806, | |
| "eval_samples_per_second": 499.488, | |
| "eval_steps_per_second": 15.739, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.765540871267744e-05, | |
| "loss": 0.1558, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.753303964757709e-05, | |
| "loss": 0.144, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 2.741067058247675e-05, | |
| "loss": 0.1474, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.728830151737641e-05, | |
| "loss": 0.1373, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.7165932452276066e-05, | |
| "loss": 0.0985, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.7043563387175723e-05, | |
| "loss": 0.1163, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 2.6921194322075384e-05, | |
| "loss": 0.157, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 2.6798825256975034e-05, | |
| "loss": 0.1345, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 2.6676456191874695e-05, | |
| "loss": 0.1588, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.6554087126774352e-05, | |
| "loss": 0.1693, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 2.643171806167401e-05, | |
| "loss": 0.1348, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 2.6309348996573667e-05, | |
| "loss": 0.1255, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 2.6186979931473327e-05, | |
| "loss": 0.1401, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.6064610866372985e-05, | |
| "loss": 0.1517, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 2.594224180127264e-05, | |
| "loss": 0.1031, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.5819872736172296e-05, | |
| "loss": 0.1127, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 2.5697503671071953e-05, | |
| "loss": 0.1389, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 2.5575134605971614e-05, | |
| "loss": 0.1207, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 2.545276554087127e-05, | |
| "loss": 0.1462, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 2.533039647577093e-05, | |
| "loss": 0.13, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 2.5208027410670582e-05, | |
| "loss": 0.1039, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 2.508565834557024e-05, | |
| "loss": 0.1208, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 2.4963289280469897e-05, | |
| "loss": 0.153, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 2.4840920215369558e-05, | |
| "loss": 0.17, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 2.471855115026921e-05, | |
| "loss": 0.1431, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 2.459618208516887e-05, | |
| "loss": 0.1625, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 2.447381302006853e-05, | |
| "loss": 0.1591, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 2.4351443954968184e-05, | |
| "loss": 0.1123, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 2.422907488986784e-05, | |
| "loss": 0.1296, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 2.41067058247675e-05, | |
| "loss": 0.0822, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 2.398433675966716e-05, | |
| "loss": 0.1461, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 2.3861967694566813e-05, | |
| "loss": 0.1157, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 2.3739598629466473e-05, | |
| "loss": 0.1121, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 2.361722956436613e-05, | |
| "loss": 0.1568, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 2.3494860499265785e-05, | |
| "loss": 0.1053, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 2.3372491434165445e-05, | |
| "loss": 0.1678, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 2.3250122369065103e-05, | |
| "loss": 0.1168, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.3127753303964757e-05, | |
| "loss": 0.1282, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 2.3005384238864417e-05, | |
| "loss": 0.1203, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.2883015173764075e-05, | |
| "loss": 0.1521, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 2.2760646108663732e-05, | |
| "loss": 0.138, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.263827704356339e-05, | |
| "loss": 0.1253, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 2.2515907978463046e-05, | |
| "loss": 0.1716, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.2393538913362704e-05, | |
| "loss": 0.1308, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 2.227116984826236e-05, | |
| "loss": 0.0879, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 2.214880078316202e-05, | |
| "loss": 0.0948, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9777571825764597, | |
| "eval_loss": 0.0665615126490593, | |
| "eval_runtime": 6.5547, | |
| "eval_samples_per_second": 493.844, | |
| "eval_steps_per_second": 15.561, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.2026431718061676e-05, | |
| "loss": 0.1124, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.1904062652961333e-05, | |
| "loss": 0.1582, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.178169358786099e-05, | |
| "loss": 0.1109, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 2.1659324522760648e-05, | |
| "loss": 0.1194, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.1536955457660305e-05, | |
| "loss": 0.1148, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 2.1414586392559962e-05, | |
| "loss": 0.1465, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 2.129221732745962e-05, | |
| "loss": 0.0916, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 2.1169848262359277e-05, | |
| "loss": 0.1426, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 2.1047479197258934e-05, | |
| "loss": 0.0786, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 6.21, | |
| "learning_rate": 2.092511013215859e-05, | |
| "loss": 0.1473, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 2.080274106705825e-05, | |
| "loss": 0.1174, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 2.0680372001957906e-05, | |
| "loss": 0.0798, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 2.0558002936857563e-05, | |
| "loss": 0.1626, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 2.043563387175722e-05, | |
| "loss": 0.1045, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 2.0313264806656878e-05, | |
| "loss": 0.1391, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 2.0190895741556535e-05, | |
| "loss": 0.1505, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 2.0068526676456192e-05, | |
| "loss": 0.1366, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 1.994615761135585e-05, | |
| "loss": 0.1317, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 1.9823788546255507e-05, | |
| "loss": 0.141, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 1.9701419481155164e-05, | |
| "loss": 0.1258, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 1.957905041605482e-05, | |
| "loss": 0.0975, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 1.9456681350954482e-05, | |
| "loss": 0.1543, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 1.9334312285854136e-05, | |
| "loss": 0.1097, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 1.9211943220753794e-05, | |
| "loss": 0.1185, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 1.9089574155653454e-05, | |
| "loss": 0.129, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 1.8967205090553108e-05, | |
| "loss": 0.1135, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 1.8844836025452765e-05, | |
| "loss": 0.1159, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 1.8722466960352426e-05, | |
| "loss": 0.1321, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 1.860009789525208e-05, | |
| "loss": 0.138, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 1.8477728830151737e-05, | |
| "loss": 0.1002, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 1.8355359765051398e-05, | |
| "loss": 0.1541, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.8232990699951052e-05, | |
| "loss": 0.1175, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 1.811062163485071e-05, | |
| "loss": 0.1713, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 1.798825256975037e-05, | |
| "loss": 0.1168, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 1.7865883504650024e-05, | |
| "loss": 0.1172, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 1.774351443954968e-05, | |
| "loss": 0.1219, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.7621145374449342e-05, | |
| "loss": 0.0906, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 1.7498776309348996e-05, | |
| "loss": 0.0864, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 1.7376407244248653e-05, | |
| "loss": 0.1334, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 1.7254038179148314e-05, | |
| "loss": 0.0803, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 1.7131669114047968e-05, | |
| "loss": 0.116, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 1.700930004894763e-05, | |
| "loss": 0.1122, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 1.6886930983847286e-05, | |
| "loss": 0.1357, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 1.676456191874694e-05, | |
| "loss": 0.1193, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 1.66421928536466e-05, | |
| "loss": 0.1049, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.978066110596231, | |
| "eval_loss": 0.0661974847316742, | |
| "eval_runtime": 6.489, | |
| "eval_samples_per_second": 498.843, | |
| "eval_steps_per_second": 15.719, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.6519823788546258e-05, | |
| "loss": 0.128, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.639745472344591e-05, | |
| "loss": 0.0983, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.6275085658345572e-05, | |
| "loss": 0.1544, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 1.615271659324523e-05, | |
| "loss": 0.1264, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 1.6030347528144883e-05, | |
| "loss": 0.1272, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 1.5907978463044544e-05, | |
| "loss": 0.1168, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 1.57856093979442e-05, | |
| "loss": 0.1087, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 1.5663240332843855e-05, | |
| "loss": 0.1234, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 1.5540871267743516e-05, | |
| "loss": 0.1147, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 1.5418502202643173e-05, | |
| "loss": 0.1111, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 1.529613313754283e-05, | |
| "loss": 0.1356, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 1.5173764072442486e-05, | |
| "loss": 0.121, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 1.5051395007342145e-05, | |
| "loss": 0.0809, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 1.4929025942241802e-05, | |
| "loss": 0.1398, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 1.4806656877141458e-05, | |
| "loss": 0.111, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 1.4684287812041117e-05, | |
| "loss": 0.126, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 1.4561918746940776e-05, | |
| "loss": 0.1364, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 1.443954968184043e-05, | |
| "loss": 0.1225, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 1.4317180616740089e-05, | |
| "loss": 0.126, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 1.4194811551639748e-05, | |
| "loss": 0.1036, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 1.4072442486539404e-05, | |
| "loss": 0.0925, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 1.3950073421439061e-05, | |
| "loss": 0.0956, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 1.382770435633872e-05, | |
| "loss": 0.1141, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 1.3705335291238376e-05, | |
| "loss": 0.0709, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 1.3582966226138033e-05, | |
| "loss": 0.1172, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 1.3460597161037692e-05, | |
| "loss": 0.125, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 1.3338228095937347e-05, | |
| "loss": 0.1254, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 1.3215859030837005e-05, | |
| "loss": 0.0998, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 1.3093489965736664e-05, | |
| "loss": 0.104, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 1.297112090063632e-05, | |
| "loss": 0.1127, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 1.2848751835535977e-05, | |
| "loss": 0.1194, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 1.2726382770435636e-05, | |
| "loss": 0.1094, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 1.2604013705335291e-05, | |
| "loss": 0.0983, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 1.2481644640234949e-05, | |
| "loss": 0.1166, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 1.2359275575134606e-05, | |
| "loss": 0.0939, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 1.2236906510034265e-05, | |
| "loss": 0.1319, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 1.211453744493392e-05, | |
| "loss": 0.0841, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 1.199216837983358e-05, | |
| "loss": 0.1166, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 1.1869799314733237e-05, | |
| "loss": 0.0854, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 1.1747430249632892e-05, | |
| "loss": 0.1122, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 1.1625061184532551e-05, | |
| "loss": 0.1287, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 1.1502692119432209e-05, | |
| "loss": 0.0982, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 1.1380323054331866e-05, | |
| "loss": 0.1172, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 1.1257953989231523e-05, | |
| "loss": 0.1236, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 1.113558492413118e-05, | |
| "loss": 0.12, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 1.1013215859030838e-05, | |
| "loss": 0.0928, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9774482545566883, | |
| "eval_loss": 0.06933724135160446, | |
| "eval_runtime": 6.53, | |
| "eval_samples_per_second": 495.714, | |
| "eval_steps_per_second": 15.62, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 1.0890846793930495e-05, | |
| "loss": 0.0916, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 1.0768477728830152e-05, | |
| "loss": 0.0792, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 1.064610866372981e-05, | |
| "loss": 0.1122, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.0523739598629467e-05, | |
| "loss": 0.1373, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 1.0401370533529124e-05, | |
| "loss": 0.1101, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 1.0279001468428782e-05, | |
| "loss": 0.1064, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 1.0156632403328439e-05, | |
| "loss": 0.1187, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 1.0034263338228096e-05, | |
| "loss": 0.1155, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 9.911894273127754e-06, | |
| "loss": 0.1236, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 9.78952520802741e-06, | |
| "loss": 0.1007, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 8.23, | |
| "learning_rate": 9.667156142927068e-06, | |
| "loss": 0.1158, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 9.544787077826727e-06, | |
| "loss": 0.0951, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 9.422418012726383e-06, | |
| "loss": 0.1196, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 9.30004894762604e-06, | |
| "loss": 0.0829, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 9.177679882525699e-06, | |
| "loss": 0.0879, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 9.055310817425355e-06, | |
| "loss": 0.1147, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 8.932941752325012e-06, | |
| "loss": 0.1241, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "learning_rate": 8.810572687224671e-06, | |
| "loss": 0.089, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 8.688203622124327e-06, | |
| "loss": 0.1278, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 8.565834557023984e-06, | |
| "loss": 0.1134, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 8.443465491923643e-06, | |
| "loss": 0.1079, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 8.3210964268233e-06, | |
| "loss": 0.0967, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 8.198727361722956e-06, | |
| "loss": 0.1004, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 8.076358296622615e-06, | |
| "loss": 0.1147, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 7.953989231522272e-06, | |
| "loss": 0.1276, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 7.831620166421928e-06, | |
| "loss": 0.1199, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 7.709251101321587e-06, | |
| "loss": 0.1014, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 7.586882036221243e-06, | |
| "loss": 0.1082, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 7.464512971120901e-06, | |
| "loss": 0.0872, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 7.3421439060205585e-06, | |
| "loss": 0.0865, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 7.219774840920215e-06, | |
| "loss": 0.1083, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 7.097405775819874e-06, | |
| "loss": 0.1053, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 6.9750367107195305e-06, | |
| "loss": 0.1408, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 6.852667645619188e-06, | |
| "loss": 0.1455, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 6.730298580518846e-06, | |
| "loss": 0.0891, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 6.607929515418502e-06, | |
| "loss": 0.1112, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 6.48556045031816e-06, | |
| "loss": 0.0981, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 6.363191385217818e-06, | |
| "loss": 0.1116, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 6.240822320117474e-06, | |
| "loss": 0.095, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 6.118453255017132e-06, | |
| "loss": 0.1103, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 5.99608418991679e-06, | |
| "loss": 0.1162, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 5.873715124816446e-06, | |
| "loss": 0.0998, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 8.94, | |
| "learning_rate": 5.751346059716104e-06, | |
| "loss": 0.094, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.628976994615762e-06, | |
| "loss": 0.1082, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 5.506607929515419e-06, | |
| "loss": 0.0903, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.9771393265369169, | |
| "eval_loss": 0.07044384628534317, | |
| "eval_runtime": 6.5312, | |
| "eval_samples_per_second": 495.624, | |
| "eval_steps_per_second": 15.617, | |
| "step": 2049 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5.384238864415076e-06, | |
| "loss": 0.1212, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.2618697993147335e-06, | |
| "loss": 0.1009, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 5.139500734214391e-06, | |
| "loss": 0.1079, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 5.017131669114048e-06, | |
| "loss": 0.1134, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.894762604013705e-06, | |
| "loss": 0.092, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 4.7723935389133636e-06, | |
| "loss": 0.1126, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 4.65002447381302e-06, | |
| "loss": 0.0749, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 4.527655408712677e-06, | |
| "loss": 0.1292, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 4.4052863436123355e-06, | |
| "loss": 0.1171, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 4.282917278511992e-06, | |
| "loss": 0.082, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 4.16054821341165e-06, | |
| "loss": 0.1129, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 4.038179148311307e-06, | |
| "loss": 0.0779, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 3.915810083210964e-06, | |
| "loss": 0.1117, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 3.7934410181106215e-06, | |
| "loss": 0.0878, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 3.6710719530102793e-06, | |
| "loss": 0.0925, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 3.548702887909937e-06, | |
| "loss": 0.1107, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 3.426333822809594e-06, | |
| "loss": 0.1063, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 3.303964757709251e-06, | |
| "loss": 0.1042, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 3.181595692608909e-06, | |
| "loss": 0.1076, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 3.059226627508566e-06, | |
| "loss": 0.1014, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.936857562408223e-06, | |
| "loss": 0.1035, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 2.814488497307881e-06, | |
| "loss": 0.1216, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 2.692119432207538e-06, | |
| "loss": 0.0988, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 2.5697503671071954e-06, | |
| "loss": 0.1094, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 2.4473813020068527e-06, | |
| "loss": 0.0732, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 2.32501223690651e-06, | |
| "loss": 0.1133, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 2.2026431718061677e-06, | |
| "loss": 0.1065, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 2.080274106705825e-06, | |
| "loss": 0.0633, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 1.957905041605482e-06, | |
| "loss": 0.0959, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 1.8355359765051396e-06, | |
| "loss": 0.1422, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 1.713166911404797e-06, | |
| "loss": 0.1152, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 1.5907978463044545e-06, | |
| "loss": 0.1046, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 1.4684287812041115e-06, | |
| "loss": 0.0728, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 1.346059716103769e-06, | |
| "loss": 0.1027, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 1.2236906510034264e-06, | |
| "loss": 0.0835, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 1.1013215859030839e-06, | |
| "loss": 0.1145, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 9.78952520802741e-07, | |
| "loss": 0.1031, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 8.565834557023985e-07, | |
| "loss": 0.0955, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 7.342143906020558e-07, | |
| "loss": 0.0845, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 6.118453255017132e-07, | |
| "loss": 0.1065, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 4.894762604013705e-07, | |
| "loss": 0.1896, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 3.671071953010279e-07, | |
| "loss": 0.1049, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 2.4473813020068524e-07, | |
| "loss": 0.1212, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 1.2236906510034262e-07, | |
| "loss": 0.1173, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 0.0, | |
| "loss": 0.0759, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "eval_accuracy": 0.9777571825764597, | |
| "eval_loss": 0.06519713997840881, | |
| "eval_runtime": 6.5757, | |
| "eval_samples_per_second": 492.27, | |
| "eval_steps_per_second": 15.512, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "step": 2270, | |
| "total_flos": 7.217190574941635e+18, | |
| "train_loss": 0.18469989935206946, | |
| "train_runtime": 1472.4515, | |
| "train_samples_per_second": 197.82, | |
| "train_steps_per_second": 1.542 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 2270, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 7.217190574941635e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |