| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7025997267496681, | |
| "eval_steps": 500, | |
| "global_step": 2282, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999995319361092e-05, | |
| "loss": 1.7428, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.999988298424633e-05, | |
| "loss": 1.5853, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999531939723856e-05, | |
| "loss": 1.4404, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9998946874648124e-05, | |
| "loss": 1.4037, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9998127802711502e-05, | |
| "loss": 1.3591, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9997074743082852e-05, | |
| "loss": 1.3367, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9995787720407095e-05, | |
| "loss": 1.3399, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9994266764804605e-05, | |
| "loss": 1.3106, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.999251191187054e-05, | |
| "loss": 1.3077, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9990523202673992e-05, | |
| "loss": 1.2856, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9988300683757016e-05, | |
| "loss": 1.3115, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.998584440713356e-05, | |
| "loss": 1.3046, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9983154430288237e-05, | |
| "loss": 1.3149, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9980230816174982e-05, | |
| "loss": 1.3038, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.997707363321557e-05, | |
| "loss": 1.266, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9973682955298038e-05, | |
| "loss": 1.2732, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9970058861774924e-05, | |
| "loss": 1.2475, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.996620143746144e-05, | |
| "loss": 1.2846, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.996211077263347e-05, | |
| "loss": 1.2978, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.995778696302546e-05, | |
| "loss": 1.2706, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.995323010982817e-05, | |
| "loss": 1.2472, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9948440319686334e-05, | |
| "loss": 1.2557, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.994341770469612e-05, | |
| "loss": 1.2501, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9938162382402556e-05, | |
| "loss": 1.2599, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.993267447579673e-05, | |
| "loss": 1.2891, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.992695411331296e-05, | |
| "loss": 1.2394, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.992100142882574e-05, | |
| "loss": 1.2639, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9914816561646648e-05, | |
| "loss": 1.2778, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9908399656521066e-05, | |
| "loss": 1.2688, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9901750863624784e-05, | |
| "loss": 1.2677, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9894870338560512e-05, | |
| "loss": 1.2821, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.988775824235421e-05, | |
| "loss": 1.2956, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9880414741451336e-05, | |
| "loss": 1.2875, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9872840007712953e-05, | |
| "loss": 1.2634, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9865034218411698e-05, | |
| "loss": 1.2588, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9856997556227632e-05, | |
| "loss": 1.2692, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.984873020924397e-05, | |
| "loss": 1.2452, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.984023237094268e-05, | |
| "loss": 1.2333, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9831504240199956e-05, | |
| "loss": 1.251, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.982254602128155e-05, | |
| "loss": 1.2642, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.981335792383802e-05, | |
| "loss": 1.2163, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9803940162899785e-05, | |
| "loss": 1.2599, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.979429295887213e-05, | |
| "loss": 1.2397, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.978441653753003e-05, | |
| "loss": 1.2563, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.977431113001285e-05, | |
| "loss": 1.2619, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9763976972818973e-05, | |
| "loss": 1.2359, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9753414307800238e-05, | |
| "loss": 1.2392, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9742623382156284e-05, | |
| "loss": 1.2283, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9731604448428773e-05, | |
| "loss": 1.225, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9720357764495466e-05, | |
| "loss": 1.2274, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9708883593564207e-05, | |
| "loss": 1.2501, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.969718220416675e-05, | |
| "loss": 1.234, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9685253870152472e-05, | |
| "loss": 1.2111, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.967309887068197e-05, | |
| "loss": 1.2351, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9660717490220535e-05, | |
| "loss": 1.2288, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.964811001853147e-05, | |
| "loss": 1.2718, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9635276750669345e-05, | |
| "loss": 1.2506, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.962221798697306e-05, | |
| "loss": 1.2577, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9608934033058826e-05, | |
| "loss": 1.2609, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9595425199813026e-05, | |
| "loss": 1.2402, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9581691803384914e-05, | |
| "loss": 1.21, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9567734165179236e-05, | |
| "loss": 1.2323, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.955355261184871e-05, | |
| "loss": 1.2391, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.953914747528636e-05, | |
| "loss": 1.236, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9524519092617772e-05, | |
| "loss": 1.2183, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9509667806193186e-05, | |
| "loss": 1.194, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9494593963579503e-05, | |
| "loss": 1.2314, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.947929791755213e-05, | |
| "loss": 1.2567, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9463780026086735e-05, | |
| "loss": 1.2213, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9448040652350878e-05, | |
| "loss": 1.2281, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9432080164695492e-05, | |
| "loss": 1.2395, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9415898936646275e-05, | |
| "loss": 1.2553, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9399497346894942e-05, | |
| "loss": 1.2489, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9382875779290378e-05, | |
| "loss": 1.1973, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.936603462282963e-05, | |
| "loss": 1.1874, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9348974271648822e-05, | |
| "loss": 1.2221, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9331695125013928e-05, | |
| "loss": 1.2517, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9314197587311413e-05, | |
| "loss": 1.2187, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9296482068038797e-05, | |
| "loss": 1.2315, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9278548981795045e-05, | |
| "loss": 1.2588, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9260398748270878e-05, | |
| "loss": 1.2061, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.924203179223895e-05, | |
| "loss": 1.2329, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9223448543543893e-05, | |
| "loss": 1.2245, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9204649437092284e-05, | |
| "loss": 1.2233, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9185634912842445e-05, | |
| "loss": 1.2197, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.916640541579415e-05, | |
| "loss": 1.2045, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9146961395978218e-05, | |
| "loss": 1.2194, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9127303308445978e-05, | |
| "loss": 1.234, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9107431613258614e-05, | |
| "loss": 1.2175, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9087346775476402e-05, | |
| "loss": 1.2364, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.906704926514783e-05, | |
| "loss": 1.238, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9046539557298597e-05, | |
| "loss": 1.223, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9025818131920478e-05, | |
| "loss": 1.2248, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.900488547396012e-05, | |
| "loss": 1.2044, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.8983742073307667e-05, | |
| "loss": 1.216, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.896238842478532e-05, | |
| "loss": 1.2172, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8940825028135726e-05, | |
| "loss": 1.2224, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.891905238801031e-05, | |
| "loss": 1.1959, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8897071013957456e-05, | |
| "loss": 1.2134, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8874881420410566e-05, | |
| "loss": 1.2168, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.885248412667605e-05, | |
| "loss": 1.2193, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8829879656921146e-05, | |
| "loss": 1.1973, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.880706854016166e-05, | |
| "loss": 1.2266, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8784051310249606e-05, | |
| "loss": 1.2047, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.876082850586068e-05, | |
| "loss": 1.2237, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.873740067048168e-05, | |
| "loss": 1.2192, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.871376835239776e-05, | |
| "loss": 1.206, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8689932104679628e-05, | |
| "loss": 1.2098, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.866589248517058e-05, | |
| "loss": 1.2125, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8641650056473457e-05, | |
| "loss": 1.1842, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8617205385937466e-05, | |
| "loss": 1.2151, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8592559045644925e-05, | |
| "loss": 1.2003, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8567711612397847e-05, | |
| "loss": 1.201, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.854266366770445e-05, | |
| "loss": 1.2158, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8517415797765575e-05, | |
| "loss": 1.1945, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8491968593460913e-05, | |
| "loss": 1.1924, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8466322650335234e-05, | |
| "loss": 1.1939, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8440478568584402e-05, | |
| "loss": 1.1886, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.841443695304136e-05, | |
| "loss": 1.2121, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8388198413161962e-05, | |
| "loss": 1.2061, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.836176356301072e-05, | |
| "loss": 1.2154, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8335133021246402e-05, | |
| "loss": 1.1947, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.83083074111076e-05, | |
| "loss": 1.1973, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.828128736039811e-05, | |
| "loss": 1.2221, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8254073501472255e-05, | |
| "loss": 1.1999, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.822666647122007e-05, | |
| "loss": 1.1843, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.819906691105242e-05, | |
| "loss": 1.2095, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8171275466885975e-05, | |
| "loss": 1.2207, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8143292789128085e-05, | |
| "loss": 1.2243, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8115119532661576e-05, | |
| "loss": 1.2123, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.808675635682942e-05, | |
| "loss": 1.2292, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8058203925419298e-05, | |
| "loss": 1.2219, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.802946290664806e-05, | |
| "loss": 1.1849, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8000533973146102e-05, | |
| "loss": 1.2105, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.797141780194162e-05, | |
| "loss": 1.1909, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7942115074444753e-05, | |
| "loss": 1.1912, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7912626476431648e-05, | |
| "loss": 1.1956, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7882952698028414e-05, | |
| "loss": 1.2026, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.785309443369496e-05, | |
| "loss": 1.2417, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7823052382208744e-05, | |
| "loss": 1.1953, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7792827246648422e-05, | |
| "loss": 1.2126, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7762419734377406e-05, | |
| "loss": 1.2442, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7731830557027278e-05, | |
| "loss": 1.1959, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7701060430481178e-05, | |
| "loss": 1.1956, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.767011007485701e-05, | |
| "loss": 1.2106, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.763898021449061e-05, | |
| "loss": 1.191, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7607671577918798e-05, | |
| "loss": 1.1803, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7576184897862313e-05, | |
| "loss": 1.1951, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7544520911208672e-05, | |
| "loss": 1.2031, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7512680358994926e-05, | |
| "loss": 1.1791, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7480663986390324e-05, | |
| "loss": 1.2023, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7448472542678855e-05, | |
| "loss": 1.2087, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7416106781241734e-05, | |
| "loss": 1.2153, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.738356745953975e-05, | |
| "loss": 1.1961, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.735085533909555e-05, | |
| "loss": 1.1993, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7317971185475828e-05, | |
| "loss": 1.193, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.728491576827338e-05, | |
| "loss": 1.1551, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7251689861089126e-05, | |
| "loss": 1.1809, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7218294241513973e-05, | |
| "loss": 1.1911, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7184729691110642e-05, | |
| "loss": 1.1837, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7150996995395367e-05, | |
| "loss": 1.1815, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.711709694381951e-05, | |
| "loss": 1.1964, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7083030329751086e-05, | |
| "loss": 1.1985, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.70487979504562e-05, | |
| "loss": 1.2122, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7014400607080386e-05, | |
| "loss": 1.2333, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.6979839104629853e-05, | |
| "loss": 1.1995, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6945114251952654e-05, | |
| "loss": 1.1991, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.691022686171975e-05, | |
| "loss": 1.1733, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6875177750405998e-05, | |
| "loss": 1.1853, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6839967738271024e-05, | |
| "loss": 1.1959, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.680459764934006e-05, | |
| "loss": 1.2479, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.676906831138462e-05, | |
| "loss": 1.2328, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6733380555903156e-05, | |
| "loss": 1.2156, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6697535218101595e-05, | |
| "loss": 1.2296, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.666153313687378e-05, | |
| "loss": 1.1847, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6625375154781834e-05, | |
| "loss": 1.2071, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6589062118036477e-05, | |
| "loss": 1.2149, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6552594876477173e-05, | |
| "loss": 1.2063, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6515974283552273e-05, | |
| "loss": 1.191, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.647920119629904e-05, | |
| "loss": 1.1844, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6442276475323568e-05, | |
| "loss": 1.2153, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6405200984780677e-05, | |
| "loss": 1.185, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6367975592353652e-05, | |
| "loss": 1.1784, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.633060116923397e-05, | |
| "loss": 1.2095, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6293078590100886e-05, | |
| "loss": 1.2006, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6255408733100977e-05, | |
| "loss": 1.1608, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6217592479827583e-05, | |
| "loss": 1.1822, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.617963071530018e-05, | |
| "loss": 1.1893, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6141524327943666e-05, | |
| "loss": 1.1719, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6103274209567567e-05, | |
| "loss": 1.1842, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6064881255345167e-05, | |
| "loss": 1.1935, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6026346363792565e-05, | |
| "loss": 1.1797, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5987670436747634e-05, | |
| "loss": 1.2015, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.594885437934893e-05, | |
| "loss": 1.1996, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5909899100014485e-05, | |
| "loss": 1.1926, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.587080551042058e-05, | |
| "loss": 1.1997, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5831574525480387e-05, | |
| "loss": 1.1826, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5792207063322554e-05, | |
| "loss": 1.2177, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5752704045269737e-05, | |
| "loss": 1.1831, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5713066395817013e-05, | |
| "loss": 1.1852, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5673295042610278e-05, | |
| "loss": 1.1694, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5633390916424503e-05, | |
| "loss": 1.1886, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5593354951141963e-05, | |
| "loss": 1.1711, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5553188083730392e-05, | |
| "loss": 1.2088, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5512891254221046e-05, | |
| "loss": 1.1905, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5472465405686693e-05, | |
| "loss": 1.1856, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5431911484219565e-05, | |
| "loss": 1.1962, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5391230438909195e-05, | |
| "loss": 1.1836, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5350423221820223e-05, | |
| "loss": 1.1906, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5309490787970094e-05, | |
| "loss": 1.2194, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.526843409530674e-05, | |
| "loss": 1.1873, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5227254104686115e-05, | |
| "loss": 1.1763, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5185951779849754e-05, | |
| "loss": 1.198, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5144528087402191e-05, | |
| "loss": 1.1885, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5102983996788341e-05, | |
| "loss": 1.2233, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5061320480270817e-05, | |
| "loss": 1.173, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5019538512907184e-05, | |
| "loss": 1.1813, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.4977639072527115e-05, | |
| "loss": 1.1783, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4935623139709532e-05, | |
| "loss": 1.2148, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4893491697759642e-05, | |
| "loss": 1.2125, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4851245732685935e-05, | |
| "loss": 1.2219, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4808886233177096e-05, | |
| "loss": 1.1876, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4766414190578879e-05, | |
| "loss": 1.2137, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4723830598870898e-05, | |
| "loss": 1.2021, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.468113645464337e-05, | |
| "loss": 1.1969, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4638332757073785e-05, | |
| "loss": 1.1855, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4595420507903534e-05, | |
| "loss": 1.1714, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.455240071141445e-05, | |
| "loss": 1.1907, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.450927437440531e-05, | |
| "loss": 1.2007, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4466042506168285e-05, | |
| "loss": 1.1712, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4422706118465306e-05, | |
| "loss": 1.1598, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.437926622550438e-05, | |
| "loss": 1.1676, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4335723843915876e-05, | |
| "loss": 1.1773, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4292079992728712e-05, | |
| "loss": 1.1894, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4248335693346513e-05, | |
| "loss": 1.1996, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4204491969523715e-05, | |
| "loss": 1.2128, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4160549847341591e-05, | |
| "loss": 1.1819, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4116510355184256e-05, | |
| "loss": 1.1785, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4072374523714577e-05, | |
| "loss": 1.186, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4028143385850077e-05, | |
| "loss": 1.1752, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.3983817976738736e-05, | |
| "loss": 1.1863, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.3939399333734789e-05, | |
| "loss": 1.1947, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.3894888496374433e-05, | |
| "loss": 1.1798, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.38502865063515e-05, | |
| "loss": 1.1756, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3805594407493093e-05, | |
| "loss": 1.1631, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3760813245735133e-05, | |
| "loss": 1.2141, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3715944069097899e-05, | |
| "loss": 1.1619, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3670987927661498e-05, | |
| "loss": 1.1761, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3625945873541285e-05, | |
| "loss": 1.1849, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3580818960863233e-05, | |
| "loss": 1.1654, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3535608245739293e-05, | |
| "loss": 1.1971, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.349031478624264e-05, | |
| "loss": 1.1807, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3444939642382932e-05, | |
| "loss": 1.199, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3399483876081506e-05, | |
| "loss": 1.1758, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3353948551146511e-05, | |
| "loss": 1.1622, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3308334733248019e-05, | |
| "loss": 1.1608, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3262643489893084e-05, | |
| "loss": 1.2018, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3216875890400762e-05, | |
| "loss": 1.1874, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3171033005877086e-05, | |
| "loss": 1.1819, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3125115909189989e-05, | |
| "loss": 1.1533, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.307912567494421e-05, | |
| "loss": 1.1781, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3033063379456128e-05, | |
| "loss": 1.1568, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.298693010072859e-05, | |
| "loss": 1.1713, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.2940726918425674e-05, | |
| "loss": 1.164, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2894454913847419e-05, | |
| "loss": 1.1971, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2848115169904519e-05, | |
| "loss": 1.1826, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2801708771092991e-05, | |
| "loss": 1.1468, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.275523680346878e-05, | |
| "loss": 1.1665, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.270870035462235e-05, | |
| "loss": 1.1604, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2662100513653226e-05, | |
| "loss": 1.1805, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2615438371144511e-05, | |
| "loss": 1.1966, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.256871501913736e-05, | |
| "loss": 1.1789, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2521931551105427e-05, | |
| "loss": 1.1968, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2475089061929257e-05, | |
| "loss": 1.2045, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2428188647870691e-05, | |
| "loss": 1.2092, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2381231406547189e-05, | |
| "loss": 1.1558, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2334218436906149e-05, | |
| "loss": 1.1617, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.228715083919918e-05, | |
| "loss": 1.1606, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.224002971495637e-05, | |
| "loss": 1.1966, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2192856166960493e-05, | |
| "loss": 1.1851, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2145631299221201e-05, | |
| "loss": 1.1774, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2098356216949196e-05, | |
| "loss": 1.1549, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2051032026530349e-05, | |
| "loss": 1.1633, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2003659835499826e-05, | |
| "loss": 1.2149, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1956240752516153e-05, | |
| "loss": 1.1954, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1908775887335275e-05, | |
| "loss": 1.1987, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1861266350784588e-05, | |
| "loss": 1.1619, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1813713254736938e-05, | |
| "loss": 1.151, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1766117712084599e-05, | |
| "loss": 1.1601, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1718480836713228e-05, | |
| "loss": 1.1541, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1670803743475805e-05, | |
| "loss": 1.1893, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1623087548166525e-05, | |
| "loss": 1.18, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1575333367494698e-05, | |
| "loss": 1.1749, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1527542319058618e-05, | |
| "loss": 1.1631, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1479715521319393e-05, | |
| "loss": 1.1945, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1431854093574776e-05, | |
| "loss": 1.1445, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1383959155932978e-05, | |
| "loss": 1.1709, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1336031829286444e-05, | |
| "loss": 1.1953, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1288073235285626e-05, | |
| "loss": 1.1635, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1240084496312724e-05, | |
| "loss": 1.1638, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1192066735455431e-05, | |
| "loss": 1.2029, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1144021076480644e-05, | |
| "loss": 1.2041, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.109594864380816e-05, | |
| "loss": 1.1587, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.104785056248437e-05, | |
| "loss": 1.2077, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.099972795815592e-05, | |
| "loss": 1.171, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.0951581957043369e-05, | |
| "loss": 1.168, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.0903413685914843e-05, | |
| "loss": 1.1637, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.085522427205965e-05, | |
| "loss": 1.1865, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.0807014843261904e-05, | |
| "loss": 1.1895, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0758786527774142e-05, | |
| "loss": 1.1966, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0710540454290891e-05, | |
| "loss": 1.1916, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0662277751922286e-05, | |
| "loss": 1.1827, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0613999550167626e-05, | |
| "loss": 1.2037, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0565706978888942e-05, | |
| "loss": 1.1763, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.051740116828456e-05, | |
| "loss": 1.1784, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0469083248862645e-05, | |
| "loss": 1.1698, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0420754351414755e-05, | |
| "loss": 1.1708, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0372415606989351e-05, | |
| "loss": 1.16, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0324068146865365e-05, | |
| "loss": 1.2111, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0275713102525689e-05, | |
| "loss": 1.1751, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0227351605630711e-05, | |
| "loss": 1.1805, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0178984787991842e-05, | |
| "loss": 1.2023, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0130613781544998e-05, | |
| "loss": 1.1633, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0082239718324136e-05, | |
| "loss": 1.1667, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0033863730434752e-05, | |
| "loss": 1.1676, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.98548695002738e-06, | |
| "loss": 1.1934, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.937110509271101e-06, | |
| "loss": 1.1881, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.888735540327059e-06, | |
| "loss": 1.1607, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.840363175321932e-06, | |
| "loss": 1.159, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.791994546321478e-06, | |
| "loss": 1.1442, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.743630785304012e-06, | |
| "loss": 1.1564, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.695273024133916e-06, | |
| "loss": 1.1934, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.646922394535173e-06, | |
| "loss": 1.168, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.598580028064851e-06, | |
| "loss": 1.1991, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.550247056086641e-06, | |
| "loss": 1.1808, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.501924609744367e-06, | |
| "loss": 1.1679, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.45361381993553e-06, | |
| "loss": 1.1802, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.405315817284825e-06, | |
| "loss": 1.1756, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.357031732117687e-06, | |
| "loss": 1.1793, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.308762694433842e-06, | |
| "loss": 1.1719, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.260509833880848e-06, | |
| "loss": 1.1508, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.212274279727676e-06, | |
| "loss": 1.146, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.16405716083828e-06, | |
| "loss": 1.2294, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.115859605645149e-06, | |
| "loss": 1.1678, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.067682742122937e-06, | |
| "loss": 1.1477, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.019527697762047e-06, | |
| "loss": 1.1584, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.971395599542232e-06, | |
| "loss": 1.2014, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.92328757390625e-06, | |
| "loss": 1.1781, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.875204746733474e-06, | |
| "loss": 1.1736, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.827148243313553e-06, | |
| "loss": 1.1895, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.779119188320082e-06, | |
| "loss": 1.151, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.73111870578428e-06, | |
| "loss": 1.1404, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.683147919068665e-06, | |
| "loss": 1.1952, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.635207950840795e-06, | |
| "loss": 1.1963, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.587299923046977e-06, | |
| "loss": 1.1789, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.539424956885998e-06, | |
| "loss": 1.1806, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.49158417278291e-06, | |
| "loss": 1.1747, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.443778690362801e-06, | |
| "loss": 1.1907, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.396009628424574e-06, | |
| "loss": 1.1948, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.348278104914791e-06, | |
| "loss": 1.1802, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.300585236901489e-06, | |
| "loss": 1.1969, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.252932140548046e-06, | |
| "loss": 1.1954, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.205319931087062e-06, | |
| "loss": 1.1884, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.15774972279425e-06, | |
| "loss": 1.1829, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.110222628962366e-06, | |
| "loss": 1.1676, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.06273976187515e-06, | |
| "loss": 1.1486, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.0153022327813e-06, | |
| "loss": 1.1814, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 7.967911151868454e-06, | |
| "loss": 1.1653, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.920567628237228e-06, | |
| "loss": 1.1542, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.873272769875244e-06, | |
| "loss": 1.182, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.82602768363119e-06, | |
| "loss": 1.1686, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.778833475188949e-06, | |
| "loss": 1.1805, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.7316912490417e-06, | |
| "loss": 1.1775, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.684602108466055e-06, | |
| "loss": 1.2037, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.637567155496277e-06, | |
| "loss": 1.155, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.59058749089846e-06, | |
| "loss": 1.1766, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.543664214144773e-06, | |
| "loss": 1.1813, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.496798423387733e-06, | |
| "loss": 1.1761, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.449991215434514e-06, | |
| "loss": 1.176, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.403243685721251e-06, | |
| "loss": 1.1658, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.356556928287433e-06, | |
| "loss": 1.1525, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.309932035750276e-06, | |
| "loss": 1.1843, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.263370099279173e-06, | |
| "loss": 1.1704, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.2168722085701416e-06, | |
| "loss": 1.1924, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.170439451820323e-06, | |
| "loss": 1.1638, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.124072915702524e-06, | |
| "loss": 1.1604, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.077773685339783e-06, | |
| "loss": 1.1719, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.031542844279962e-06, | |
| "loss": 1.1768, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.985381474470402e-06, | |
| "loss": 1.1596, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.939290656232604e-06, | |
| "loss": 1.1894, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.89327146823693e-06, | |
| "loss": 1.1764, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.847324987477375e-06, | |
| "loss": 1.1855, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.801452289246356e-06, | |
| "loss": 1.187, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.755654447109538e-06, | |
| "loss": 1.1709, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.709932532880732e-06, | |
| "loss": 1.1836, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.664287616596785e-06, | |
| "loss": 1.1602, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.6187207664925505e-06, | |
| "loss": 1.1752, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.573233048975891e-06, | |
| "loss": 1.16, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.527825528602719e-06, | |
| "loss": 1.1683, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.482499268052077e-06, | |
| "loss": 1.1981, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.4372553281012695e-06, | |
| "loss": 1.1877, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.392094767601051e-06, | |
| "loss": 1.1774, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.347018643450815e-06, | |
| "loss": 1.1758, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.302028010573898e-06, | |
| "loss": 1.1841, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.257123921892862e-06, | |
| "loss": 1.1538, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.212307428304858e-06, | |
| "loss": 1.1824, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.167579578657044e-06, | |
| "loss": 1.1608, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.122941419722032e-06, | |
| "loss": 1.1857, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.078393996173375e-06, | |
| "loss": 1.1987, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.0339383505611444e-06, | |
| "loss": 1.1816, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.9895755232875216e-06, | |
| "loss": 1.1852, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.945306552582431e-06, | |
| "loss": 1.1918, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.9011324744792716e-06, | |
| "loss": 1.1628, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.8570543227906565e-06, | |
| "loss": 1.1497, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.813073129084206e-06, | |
| "loss": 1.1987, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.769189922658423e-06, | |
| "loss": 1.183, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.725405730518616e-06, | |
| "loss": 1.1487, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.681721577352818e-06, | |
| "loss": 1.1558, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.638138485507861e-06, | |
| "loss": 1.186, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.594657474965408e-06, | |
| "loss": 1.1615, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.551279563318106e-06, | |
| "loss": 1.1436, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.508005765745753e-06, | |
| "loss": 1.162, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.464837094991569e-06, | |
| "loss": 1.1526, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.421774561338446e-06, | |
| "loss": 1.1535, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.3788191725853656e-06, | |
| "loss": 1.1648, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.335971934023757e-06, | |
| "loss": 1.1607, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.2932338484140055e-06, | |
| "loss": 1.1613, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.250605915961966e-06, | |
| "loss": 1.1767, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.20808913429558e-06, | |
| "loss": 1.1886, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.1656844984414785e-06, | |
| "loss": 1.1761, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.1233930008017615e-06, | |
| "loss": 1.1817, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.081215631130719e-06, | |
| "loss": 1.1521, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.0391533765116895e-06, | |
| "loss": 1.1811, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.9972072213339555e-06, | |
| "loss": 1.1719, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.955378147269719e-06, | |
| "loss": 1.1766, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.913667133251091e-06, | |
| "loss": 1.1764, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.872075155447231e-06, | |
| "loss": 1.1636, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.830603187241463e-06, | |
| "loss": 1.178, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.789252199208508e-06, | |
| "loss": 1.1748, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.748023159091773e-06, | |
| "loss": 1.1494, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.706917031780698e-06, | |
| "loss": 1.1737, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.6659347792881756e-06, | |
| "loss": 1.1757, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.6250773607280375e-06, | |
| "loss": 1.1464, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.584345732292613e-06, | |
| "loss": 1.2013, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.543740847230341e-06, | |
| "loss": 1.1716, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.503263655823466e-06, | |
| "loss": 1.2016, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.462915105365798e-06, | |
| "loss": 1.1922, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.422696140140547e-06, | |
| "loss": 1.1874, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.382607701398217e-06, | |
| "loss": 1.1498, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.342650727334591e-06, | |
| "loss": 1.1924, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.302826153068741e-06, | |
| "loss": 1.168, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.263134910621196e-06, | |
| "loss": 1.1673, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.223577928892073e-06, | |
| "loss": 1.163, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.184156133639388e-06, | |
| "loss": 1.146, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.1448704474573546e-06, | |
| "loss": 1.1513, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.105721789754805e-06, | |
| "loss": 1.1913, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.066711076733677e-06, | |
| "loss": 1.1826, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 1.1736973524093628, | |
| "eval_runtime": 1794.5729, | |
| "eval_samples_per_second": 12.878, | |
| "eval_steps_per_second": 3.22, | |
| "step": 2282 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "step": 2282, | |
| "total_flos": 165398351708160.0, | |
| "train_loss": 1.200063390965842, | |
| "train_runtime": 41297.3635, | |
| "train_samples_per_second": 5.033, | |
| "train_steps_per_second": 0.079 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 3247, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 165398351708160.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |