| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.3233760459193985, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9457994579945803e-05, | |
| "loss": 4.3763, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.878048780487805e-05, | |
| "loss": 3.8394, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.81029810298103e-05, | |
| "loss": 3.639, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7425474254742554e-05, | |
| "loss": 3.7962, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6747967479674795e-05, | |
| "loss": 3.5152, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.607046070460705e-05, | |
| "loss": 3.5752, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.53929539295393e-05, | |
| "loss": 3.4463, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.4715447154471546e-05, | |
| "loss": 3.5218, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_loss": 3.2419066429138184, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 180.4791, | |
| "eval_samples_per_second": 0.15, | |
| "eval_steps_per_second": 0.078, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.4037940379403794e-05, | |
| "loss": 3.5264, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.336043360433605e-05, | |
| "loss": 3.6286, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.26829268292683e-05, | |
| "loss": 3.4611, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2005420054200545e-05, | |
| "loss": 3.9319, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.132791327913279e-05, | |
| "loss": 3.7751, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.065040650406504e-05, | |
| "loss": 3.6241, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9972899728997295e-05, | |
| "loss": 3.3906, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9295392953929537e-05, | |
| "loss": 3.1815, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "eval_loss": 3.189028739929199, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 265.9332, | |
| "eval_samples_per_second": 0.102, | |
| "eval_steps_per_second": 0.053, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.861788617886179e-05, | |
| "loss": 3.3673, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.794037940379404e-05, | |
| "loss": 3.3792, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.726287262872629e-05, | |
| "loss": 3.5732, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6585365853658535e-05, | |
| "loss": 3.2772, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.6043360433604336e-05, | |
| "loss": 3.3803, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5365853658536584e-05, | |
| "loss": 3.2597, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.468834688346884e-05, | |
| "loss": 3.4085, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.401084010840109e-05, | |
| "loss": 3.2319, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 3.166618824005127, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 433.3435, | |
| "eval_samples_per_second": 0.062, | |
| "eval_steps_per_second": 0.032, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 3.6139, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.265582655826558e-05, | |
| "loss": 3.5873, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.197831978319784e-05, | |
| "loss": 3.3622, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.130081300813008e-05, | |
| "loss": 3.4854, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0623306233062334e-05, | |
| "loss": 3.4429, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9945799457994585e-05, | |
| "loss": 3.1618, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.926829268292683e-05, | |
| "loss": 3.428, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.859078590785908e-05, | |
| "loss": 3.2305, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_loss": 3.152249336242676, | |
| "eval_rouge2_fmeasure": 0.0037, | |
| "eval_rouge2_precision": 0.0062, | |
| "eval_rouge2_recall": 0.0026, | |
| "eval_runtime": 372.2318, | |
| "eval_samples_per_second": 0.073, | |
| "eval_steps_per_second": 0.038, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7913279132791332e-05, | |
| "loss": 3.0798, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7235772357723577e-05, | |
| "loss": 3.2169, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6558265582655828e-05, | |
| "loss": 3.4364, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.588075880758808e-05, | |
| "loss": 3.2353, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5203252032520324e-05, | |
| "loss": 3.4389, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4525745257452575e-05, | |
| "loss": 3.2847, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3848238482384823e-05, | |
| "loss": 3.0442, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.3170731707317075e-05, | |
| "loss": 3.2043, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "eval_loss": 3.1420483589172363, | |
| "eval_rouge2_fmeasure": 0.0032, | |
| "eval_rouge2_precision": 0.0041, | |
| "eval_rouge2_recall": 0.0026, | |
| "eval_runtime": 417.3311, | |
| "eval_samples_per_second": 0.065, | |
| "eval_steps_per_second": 0.034, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2493224932249323e-05, | |
| "loss": 3.4208, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.181571815718157e-05, | |
| "loss": 3.7157, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.1138211382113822e-05, | |
| "loss": 3.0534, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.046070460704607e-05, | |
| "loss": 3.2667, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.978319783197832e-05, | |
| "loss": 3.3489, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9105691056910573e-05, | |
| "loss": 3.3139, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.842818428184282e-05, | |
| "loss": 3.3155, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.775067750677507e-05, | |
| "loss": 3.603, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 3.135263681411743, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 517.2838, | |
| "eval_samples_per_second": 0.052, | |
| "eval_steps_per_second": 0.027, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.707317073170732e-05, | |
| "loss": 3.4609, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6395663956639568e-05, | |
| "loss": 3.2715, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5718157181571816e-05, | |
| "loss": 3.91, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.5040650406504067e-05, | |
| "loss": 3.0947, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4498644986449867e-05, | |
| "loss": 3.2667, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3821138211382115e-05, | |
| "loss": 3.5289, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.3143631436314363e-05, | |
| "loss": 3.3016, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2466124661246612e-05, | |
| "loss": 3.2669, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "eval_loss": 3.130876064300537, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 441.1701, | |
| "eval_samples_per_second": 0.061, | |
| "eval_steps_per_second": 0.032, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1788617886178862e-05, | |
| "loss": 3.0182, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.1111111111111112e-05, | |
| "loss": 3.4221, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0433604336043361e-05, | |
| "loss": 3.1717, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.756097560975611e-06, | |
| "loss": 3.309, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.078590785907859e-06, | |
| "loss": 3.215, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.401084010840109e-06, | |
| "loss": 3.2693, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.723577235772358e-06, | |
| "loss": 3.3973, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.046070460704607e-06, | |
| "loss": 3.3138, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "eval_loss": 3.127232074737549, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 495.8047, | |
| "eval_samples_per_second": 0.054, | |
| "eval_steps_per_second": 0.028, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.368563685636857e-06, | |
| "loss": 3.5124, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.6910569105691056e-06, | |
| "loss": 3.128, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 5.013550135501355e-06, | |
| "loss": 3.6112, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.336043360433605e-06, | |
| "loss": 3.0695, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.6585365853658537e-06, | |
| "loss": 3.4197, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.116531165311653e-06, | |
| "loss": 3.2216, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.4390243902439027e-06, | |
| "loss": 3.1863, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.7615176151761518e-06, | |
| "loss": 3.3628, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "eval_loss": 3.125962972640991, | |
| "eval_rouge2_fmeasure": 0.0, | |
| "eval_rouge2_precision": 0.0, | |
| "eval_rouge2_recall": 0.0, | |
| "eval_runtime": 439.7386, | |
| "eval_samples_per_second": 0.061, | |
| "eval_steps_per_second": 0.032, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.951491632306573e-05, | |
| "loss": 3.6245, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.950817904977498e-05, | |
| "loss": 3.3963, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.950144177648422e-05, | |
| "loss": 3.1854, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9494704503193465e-05, | |
| "loss": 3.5305, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9487967229902713e-05, | |
| "loss": 3.3036, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.948122995661196e-05, | |
| "loss": 3.8064, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.947449268332121e-05, | |
| "loss": 3.3977, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.946775541003045e-05, | |
| "loss": 3.453, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.94610181367397e-05, | |
| "loss": 3.1789, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.945428086344895e-05, | |
| "loss": 3.3437, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.944754359015819e-05, | |
| "loss": 3.1314, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9440806316867435e-05, | |
| "loss": 3.5475, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9434069043576684e-05, | |
| "loss": 3.145, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.942733177028593e-05, | |
| "loss": 3.3634, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9420594496995175e-05, | |
| "loss": 3.2653, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9413857223704424e-05, | |
| "loss": 3.4231, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.940711995041367e-05, | |
| "loss": 3.4497, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.940173013178107e-05, | |
| "loss": 3.6448, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.939499285849031e-05, | |
| "loss": 3.1871, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.938825558519956e-05, | |
| "loss": 3.0073, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.938151831190881e-05, | |
| "loss": 3.2752, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.937478103861805e-05, | |
| "loss": 3.3738, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.93680437653273e-05, | |
| "loss": 3.1156, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.936130649203655e-05, | |
| "loss": 3.089, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9354569218745796e-05, | |
| "loss": 3.2361, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.934783194545504e-05, | |
| "loss": 3.0767, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.934109467216429e-05, | |
| "loss": 3.4057, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.933435739887353e-05, | |
| "loss": 2.9778, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.932762012558277e-05, | |
| "loss": 3.1364, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.932088285229202e-05, | |
| "loss": 3.5782, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.931414557900127e-05, | |
| "loss": 3.056, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.930740830571052e-05, | |
| "loss": 3.4149, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.930067103241976e-05, | |
| "loss": 3.2642, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.929393375912901e-05, | |
| "loss": 3.0618, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.928719648583826e-05, | |
| "loss": 3.2814, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.92804592125475e-05, | |
| "loss": 3.0362, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.927372193925674e-05, | |
| "loss": 3.4842, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.926698466596599e-05, | |
| "loss": 3.2595, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.926024739267524e-05, | |
| "loss": 3.2634, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.925351011938448e-05, | |
| "loss": 3.492, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.924677284609373e-05, | |
| "loss": 3.5355, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.924003557280298e-05, | |
| "loss": 3.3369, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.923329829951223e-05, | |
| "loss": 3.3085, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.922656102622147e-05, | |
| "loss": 3.0434, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.921982375293071e-05, | |
| "loss": 3.1722, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.921308647963996e-05, | |
| "loss": 3.2637, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9206349206349204e-05, | |
| "loss": 3.5246, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.919961193305845e-05, | |
| "loss": 3.5051, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.91928746597677e-05, | |
| "loss": 3.2346, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.918613738647695e-05, | |
| "loss": 3.2653, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.917940011318619e-05, | |
| "loss": 3.1633, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.917266283989544e-05, | |
| "loss": 3.0785, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.916592556660469e-05, | |
| "loss": 3.1973, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.915918829331393e-05, | |
| "loss": 3.2443, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9152451020023175e-05, | |
| "loss": 3.3169, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9145713746732423e-05, | |
| "loss": 3.4131, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.913897647344167e-05, | |
| "loss": 3.2112, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9132239200150914e-05, | |
| "loss": 2.9755, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.912550192686016e-05, | |
| "loss": 3.5015, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.911876465356941e-05, | |
| "loss": 3.2172, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.911202738027866e-05, | |
| "loss": 3.0727, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.91052901069879e-05, | |
| "loss": 3.1883, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9098552833697145e-05, | |
| "loss": 3.6434, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9091815560406394e-05, | |
| "loss": 3.582, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.908507828711564e-05, | |
| "loss": 3.6279, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9078341013824885e-05, | |
| "loss": 3.4427, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9071603740534134e-05, | |
| "loss": 3.2696, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.906486646724338e-05, | |
| "loss": 3.2136, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9058129193952625e-05, | |
| "loss": 3.4226, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9051391920661874e-05, | |
| "loss": 3.0063, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9044654647371116e-05, | |
| "loss": 3.2971, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9037917374080365e-05, | |
| "loss": 3.4248, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.903118010078961e-05, | |
| "loss": 3.3834, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9024442827498856e-05, | |
| "loss": 3.3403, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9017705554208105e-05, | |
| "loss": 3.1636, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9010968280917353e-05, | |
| "loss": 3.2906, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9004231007626596e-05, | |
| "loss": 3.2874, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8997493734335844e-05, | |
| "loss": 3.1831, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8990756461045086e-05, | |
| "loss": 3.2486, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.898401918775433e-05, | |
| "loss": 3.2148, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.897728191446358e-05, | |
| "loss": 3.2995, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8970544641172826e-05, | |
| "loss": 3.3068, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8963807367882075e-05, | |
| "loss": 3.3789, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.895707009459132e-05, | |
| "loss": 3.2501, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8950332821300566e-05, | |
| "loss": 3.3165, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8943595548009815e-05, | |
| "loss": 3.2548, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8936858274719064e-05, | |
| "loss": 3.0099, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8930121001428306e-05, | |
| "loss": 3.466, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.892338372813755e-05, | |
| "loss": 3.106, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.89166464548468e-05, | |
| "loss": 3.4483, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.890990918155604e-05, | |
| "loss": 3.0073, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.890317190826529e-05, | |
| "loss": 3.1426, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.889643463497454e-05, | |
| "loss": 3.2155, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.8889697361683786e-05, | |
| "loss": 3.2883, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.888296008839303e-05, | |
| "loss": 3.3822, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.887622281510228e-05, | |
| "loss": 3.1398, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.886948554181152e-05, | |
| "loss": 3.252, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.886274826852077e-05, | |
| "loss": 3.0845, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.885601099523001e-05, | |
| "loss": 3.4627, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.884927372193926e-05, | |
| "loss": 3.1775, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.884253644864851e-05, | |
| "loss": 3.5839, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.883579917535775e-05, | |
| "loss": 3.1501, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.8829061902067e-05, | |
| "loss": 3.3487, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.882232462877625e-05, | |
| "loss": 3.1447, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.881558735548549e-05, | |
| "loss": 3.0964, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.880885008219473e-05, | |
| "loss": 3.3593, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.880211280890398e-05, | |
| "loss": 3.0965, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.879537553561323e-05, | |
| "loss": 3.0485, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.878863826232247e-05, | |
| "loss": 3.497, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.878190098903172e-05, | |
| "loss": 3.2529, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.877516371574097e-05, | |
| "loss": 3.2689, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.876842644245022e-05, | |
| "loss": 3.2244, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.876168916915946e-05, | |
| "loss": 3.3559, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.875495189586871e-05, | |
| "loss": 3.1012, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8749562077236104e-05, | |
| "loss": 2.9742, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8742824803945346e-05, | |
| "loss": 3.3356, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8736087530654595e-05, | |
| "loss": 3.0625, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8729350257363844e-05, | |
| "loss": 3.8974, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.872261298407309e-05, | |
| "loss": 3.2096, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8715875710782335e-05, | |
| "loss": 3.32, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8709138437491584e-05, | |
| "loss": 3.3169, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8702401164200826e-05, | |
| "loss": 3.2782, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8695663890910075e-05, | |
| "loss": 3.7628, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.868892661761932e-05, | |
| "loss": 3.3816, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8682189344328566e-05, | |
| "loss": 3.3809, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8675452071037814e-05, | |
| "loss": 3.5458, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8668714797747057e-05, | |
| "loss": 3.3134, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8661977524456305e-05, | |
| "loss": 3.062, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8655240251165554e-05, | |
| "loss": 3.0732, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8648502977874796e-05, | |
| "loss": 3.325, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.864176570458404e-05, | |
| "loss": 3.1344, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.863502843129329e-05, | |
| "loss": 3.4551, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8628291158002536e-05, | |
| "loss": 3.0637, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.862155388471178e-05, | |
| "loss": 3.3949, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.861481661142103e-05, | |
| "loss": 3.172, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8608079338130276e-05, | |
| "loss": 3.1154, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8601342064839525e-05, | |
| "loss": 3.0203, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.859460479154877e-05, | |
| "loss": 3.3186, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.858786751825801e-05, | |
| "loss": 3.2057, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.858113024496726e-05, | |
| "loss": 3.2119, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.857439297167651e-05, | |
| "loss": 3.0146, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.856765569838575e-05, | |
| "loss": 3.0737, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8560918425095e-05, | |
| "loss": 3.118, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.855418115180425e-05, | |
| "loss": 3.2081, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.854744387851349e-05, | |
| "loss": 3.2234, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.854070660522274e-05, | |
| "loss": 3.217, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8533969331931987e-05, | |
| "loss": 3.4233, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.852723205864123e-05, | |
| "loss": 3.3069, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.852049478535047e-05, | |
| "loss": 3.2344, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.851375751205972e-05, | |
| "loss": 3.0722, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.850702023876897e-05, | |
| "loss": 2.9557, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.850028296547822e-05, | |
| "loss": 3.132, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.849354569218746e-05, | |
| "loss": 3.2406, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.848680841889671e-05, | |
| "loss": 3.1061, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.848007114560596e-05, | |
| "loss": 3.2933, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.84733338723152e-05, | |
| "loss": 2.9373, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.846659659902444e-05, | |
| "loss": 3.5017, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.845985932573369e-05, | |
| "loss": 3.1143, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.845312205244294e-05, | |
| "loss": 3.3182, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.844638477915218e-05, | |
| "loss": 3.3364, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.843964750586143e-05, | |
| "loss": 2.8652, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.843291023257068e-05, | |
| "loss": 3.3397, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.842617295927993e-05, | |
| "loss": 3.2569, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.841943568598917e-05, | |
| "loss": 3.1305, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.841269841269841e-05, | |
| "loss": 3.3159, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.840596113940766e-05, | |
| "loss": 3.2887, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.83992238661169e-05, | |
| "loss": 3.4384, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.839248659282615e-05, | |
| "loss": 2.9754, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.83857493195354e-05, | |
| "loss": 3.5198, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.837901204624465e-05, | |
| "loss": 3.1371, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.837227477295389e-05, | |
| "loss": 3.2225, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.836553749966314e-05, | |
| "loss": 3.0097, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.835880022637238e-05, | |
| "loss": 3.2081, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.835206295308163e-05, | |
| "loss": 3.3162, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8345325679790874e-05, | |
| "loss": 3.1359, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.833858840650012e-05, | |
| "loss": 3.0363, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.833185113320937e-05, | |
| "loss": 3.2238, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8325113859918614e-05, | |
| "loss": 3.3966, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.831837658662786e-05, | |
| "loss": 3.4219, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.831163931333711e-05, | |
| "loss": 3.2468, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.830490204004636e-05, | |
| "loss": 3.2864, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.82981647667556e-05, | |
| "loss": 3.2543, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8291427493464844e-05, | |
| "loss": 3.3271, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.828469022017409e-05, | |
| "loss": 3.3202, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8277952946883335e-05, | |
| "loss": 3.3111, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8271215673592584e-05, | |
| "loss": 3.2664, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.826447840030183e-05, | |
| "loss": 3.2784, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.825774112701108e-05, | |
| "loss": 3.2833, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8251003853720324e-05, | |
| "loss": 3.1219, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.824426658042957e-05, | |
| "loss": 3.1704, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8237529307138815e-05, | |
| "loss": 3.1486, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8230792033848064e-05, | |
| "loss": 3.2648, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8224054760557306e-05, | |
| "loss": 3.1059, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8217317487266555e-05, | |
| "loss": 3.3623, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8210580213975804e-05, | |
| "loss": 3.5866, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8203842940685046e-05, | |
| "loss": 3.3165, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8197105667394295e-05, | |
| "loss": 3.3361, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8190368394103544e-05, | |
| "loss": 3.3784, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8183631120812786e-05, | |
| "loss": 2.9877, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.817689384752203e-05, | |
| "loss": 3.5171, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.817015657423128e-05, | |
| "loss": 3.1608, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8163419300940526e-05, | |
| "loss": 3.1863, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8156682027649774e-05, | |
| "loss": 3.1035, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8149944754359017e-05, | |
| "loss": 3.4208, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8143207481068265e-05, | |
| "loss": 3.0785, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8136470207777514e-05, | |
| "loss": 3.1869, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8129732934486756e-05, | |
| "loss": 3.0122, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8122995661196005e-05, | |
| "loss": 3.5176, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.811625838790525e-05, | |
| "loss": 3.3739, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8109521114614496e-05, | |
| "loss": 3.1952, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.810278384132374e-05, | |
| "loss": 3.2811, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.809604656803299e-05, | |
| "loss": 3.4939, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8089309294742236e-05, | |
| "loss": 3.3097, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8082572021451485e-05, | |
| "loss": 3.1636, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.807583474816073e-05, | |
| "loss": 3.1452, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8069097474869976e-05, | |
| "loss": 3.3203, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.806236020157922e-05, | |
| "loss": 3.1017, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.805562292828846e-05, | |
| "loss": 3.1715, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.804888565499771e-05, | |
| "loss": 2.8621, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.804214838170696e-05, | |
| "loss": 3.1176, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.803541110841621e-05, | |
| "loss": 3.5175, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.802867383512545e-05, | |
| "loss": 3.245, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.80219365618347e-05, | |
| "loss": 3.0326, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8015199288543947e-05, | |
| "loss": 3.2505, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.800846201525319e-05, | |
| "loss": 3.36, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.800172474196243e-05, | |
| "loss": 3.3422, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.799498746867168e-05, | |
| "loss": 3.2268, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.798825019538093e-05, | |
| "loss": 3.1994, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7982860376748324e-05, | |
| "loss": 3.1365, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.797612310345757e-05, | |
| "loss": 3.2696, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.796938583016682e-05, | |
| "loss": 2.9565, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.796264855687606e-05, | |
| "loss": 2.9312, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7955911283585305e-05, | |
| "loss": 2.901, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7949174010294554e-05, | |
| "loss": 3.3423, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.79424367370038e-05, | |
| "loss": 3.4224, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7935699463713045e-05, | |
| "loss": 3.1498, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7928962190422294e-05, | |
| "loss": 3.31, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.792222491713154e-05, | |
| "loss": 3.3928, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.791548764384079e-05, | |
| "loss": 3.3894, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7908750370550034e-05, | |
| "loss": 3.056, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.790201309725928e-05, | |
| "loss": 3.1893, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7895275823968525e-05, | |
| "loss": 3.1102, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.788853855067777e-05, | |
| "loss": 3.2477, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7881801277387016e-05, | |
| "loss": 3.1685, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7875064004096265e-05, | |
| "loss": 3.285, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7868326730805514e-05, | |
| "loss": 3.3288, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7861589457514756e-05, | |
| "loss": 2.9146, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7854852184224005e-05, | |
| "loss": 3.3004, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7848114910933254e-05, | |
| "loss": 3.3265, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7841377637642496e-05, | |
| "loss": 3.2319, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.783464036435174e-05, | |
| "loss": 3.2694, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.782790309106099e-05, | |
| "loss": 3.3163, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7821165817770235e-05, | |
| "loss": 2.9109, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.781442854447948e-05, | |
| "loss": 3.1788, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7807691271188726e-05, | |
| "loss": 3.2638, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7800953997897975e-05, | |
| "loss": 3.1715, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7794216724607224e-05, | |
| "loss": 3.0204, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7787479451316466e-05, | |
| "loss": 3.1442, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.778074217802571e-05, | |
| "loss": 3.1554, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.777400490473496e-05, | |
| "loss": 3.0041, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7767267631444206e-05, | |
| "loss": 3.4228, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.776053035815345e-05, | |
| "loss": 3.1159, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.77537930848627e-05, | |
| "loss": 3.3559, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7747055811571946e-05, | |
| "loss": 2.8674, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.774031853828119e-05, | |
| "loss": 3.1701, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.773358126499044e-05, | |
| "loss": 3.3984, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.772684399169968e-05, | |
| "loss": 3.2335, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.772010671840893e-05, | |
| "loss": 3.0153, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.771336944511817e-05, | |
| "loss": 3.2157, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.770663217182742e-05, | |
| "loss": 3.2443, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.769989489853667e-05, | |
| "loss": 3.2683, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.769315762524591e-05, | |
| "loss": 3.3998, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.768642035195516e-05, | |
| "loss": 3.2701, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.767968307866441e-05, | |
| "loss": 3.2304, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7672945805373656e-05, | |
| "loss": 3.4887, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.76662085320829e-05, | |
| "loss": 3.1407, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.765947125879214e-05, | |
| "loss": 3.3806, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.765273398550139e-05, | |
| "loss": 2.9394, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.764599671221064e-05, | |
| "loss": 3.0697, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.763925943891988e-05, | |
| "loss": 3.1154, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.763252216562913e-05, | |
| "loss": 3.308, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.762578489233838e-05, | |
| "loss": 3.1236, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.761904761904762e-05, | |
| "loss": 3.2146, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.761231034575687e-05, | |
| "loss": 3.0922, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.760557307246611e-05, | |
| "loss": 3.2447, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.759883579917536e-05, | |
| "loss": 3.2431, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.75920985258846e-05, | |
| "loss": 3.2935, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.758536125259385e-05, | |
| "loss": 3.0225, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.75786239793031e-05, | |
| "loss": 3.1884, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.757188670601235e-05, | |
| "loss": 3.1911, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.756514943272159e-05, | |
| "loss": 3.1588, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.755841215943084e-05, | |
| "loss": 3.0782, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.755167488614008e-05, | |
| "loss": 3.0288, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7544937612849324e-05, | |
| "loss": 3.0151, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.753820033955857e-05, | |
| "loss": 3.058, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.753146306626782e-05, | |
| "loss": 3.1809, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.752472579297707e-05, | |
| "loss": 3.1528, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.751798851968631e-05, | |
| "loss": 3.1677, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.751125124639556e-05, | |
| "loss": 2.9781, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.750451397310481e-05, | |
| "loss": 3.4727, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.749777669981406e-05, | |
| "loss": 3.0776, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.74910394265233e-05, | |
| "loss": 3.4074, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7484302153232544e-05, | |
| "loss": 3.0213, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.747756487994179e-05, | |
| "loss": 3.2027, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7470827606651035e-05, | |
| "loss": 3.1548, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7464090333360283e-05, | |
| "loss": 3.3882, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.745735306006953e-05, | |
| "loss": 3.0727, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.745061578677878e-05, | |
| "loss": 2.8628, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.744387851348802e-05, | |
| "loss": 3.4395, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.743714124019727e-05, | |
| "loss": 3.1199, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7430403966906514e-05, | |
| "loss": 3.2914, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.742366669361576e-05, | |
| "loss": 3.3878, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7416929420325005e-05, | |
| "loss": 3.0218, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7410192147034254e-05, | |
| "loss": 3.2209, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.74034548737435e-05, | |
| "loss": 3.049, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7396717600452745e-05, | |
| "loss": 3.1507, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7389980327161994e-05, | |
| "loss": 3.2019, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.738324305387124e-05, | |
| "loss": 3.1197, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7376505780580485e-05, | |
| "loss": 3.4023, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.736976850728973e-05, | |
| "loss": 3.0877, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7363031233998976e-05, | |
| "loss": 3.2648, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7356293960708225e-05, | |
| "loss": 2.9336, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7349556687417474e-05, | |
| "loss": 3.0737, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7342819414126716e-05, | |
| "loss": 2.8234, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7336082140835965e-05, | |
| "loss": 3.2141, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7329344867545213e-05, | |
| "loss": 3.1565, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7322607594254456e-05, | |
| "loss": 2.9412, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7315870320963704e-05, | |
| "loss": 3.2878, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7309133047672947e-05, | |
| "loss": 3.0794, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7302395774382195e-05, | |
| "loss": 3.397, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.729565850109144e-05, | |
| "loss": 3.0795, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7288921227800686e-05, | |
| "loss": 3.5331, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7282183954509935e-05, | |
| "loss": 3.1282, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.727544668121918e-05, | |
| "loss": 3.1138, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7268709407928426e-05, | |
| "loss": 3.1175, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7261972134637675e-05, | |
| "loss": 3.0259, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.725523486134692e-05, | |
| "loss": 3.4783, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.724849758805616e-05, | |
| "loss": 3.1384, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.724176031476541e-05, | |
| "loss": 3.2739, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.723502304147466e-05, | |
| "loss": 3.0155, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7228285768183906e-05, | |
| "loss": 3.1998, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.722154849489315e-05, | |
| "loss": 3.0975, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.72148112216024e-05, | |
| "loss": 3.2449, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7208073948311646e-05, | |
| "loss": 2.9318, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.720133667502089e-05, | |
| "loss": 3.0997, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.719459940173013e-05, | |
| "loss": 2.9982, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.718786212843938e-05, | |
| "loss": 2.9443, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.718112485514863e-05, | |
| "loss": 2.889, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.717438758185787e-05, | |
| "loss": 3.4088, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.716765030856712e-05, | |
| "loss": 3.184, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.716091303527637e-05, | |
| "loss": 3.0915, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7154175761985616e-05, | |
| "loss": 3.2379, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.714743848869486e-05, | |
| "loss": 3.1446, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.71407012154041e-05, | |
| "loss": 3.1694, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.713396394211335e-05, | |
| "loss": 3.8897, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.712722666882259e-05, | |
| "loss": 3.0934, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.712048939553184e-05, | |
| "loss": 3.1613, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.711375212224109e-05, | |
| "loss": 3.3189, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.710701484895034e-05, | |
| "loss": 3.218, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.710027757565958e-05, | |
| "loss": 3.1543, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.709354030236883e-05, | |
| "loss": 3.1454, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.708680302907808e-05, | |
| "loss": 2.9892, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.708006575578732e-05, | |
| "loss": 3.2591, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.707332848249656e-05, | |
| "loss": 2.9655, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.706659120920581e-05, | |
| "loss": 3.1065, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.705985393591506e-05, | |
| "loss": 2.8932, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.70531166626243e-05, | |
| "loss": 3.2734, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.704637938933355e-05, | |
| "loss": 3.4112, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.70396421160428e-05, | |
| "loss": 3.039, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.703290484275205e-05, | |
| "loss": 3.3834, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.702616756946129e-05, | |
| "loss": 3.2542, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.701943029617053e-05, | |
| "loss": 3.2943, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.701269302287978e-05, | |
| "loss": 3.1251, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.700595574958903e-05, | |
| "loss": 3.4983, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.699921847629827e-05, | |
| "loss": 3.0634, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.699248120300752e-05, | |
| "loss": 3.2068, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.698574392971677e-05, | |
| "loss": 3.0052, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.697900665642601e-05, | |
| "loss": 2.9226, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.697226938313526e-05, | |
| "loss": 3.0226, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.6965532109844504e-05, | |
| "loss": 3.5132, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.695879483655375e-05, | |
| "loss": 3.0079, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.6952057563262995e-05, | |
| "loss": 3.1758, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.6945320289972243e-05, | |
| "loss": 3.2862, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.693858301668149e-05, | |
| "loss": 3.2161, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.693184574339074e-05, | |
| "loss": 2.9678, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.692510847009998e-05, | |
| "loss": 3.1297, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.691837119680923e-05, | |
| "loss": 3.2068, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.691163392351848e-05, | |
| "loss": 2.9587, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.690489665022772e-05, | |
| "loss": 3.2795, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6898159376936965e-05, | |
| "loss": 3.5198, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6891422103646214e-05, | |
| "loss": 3.1278, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.688468483035546e-05, | |
| "loss": 3.1983, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6877947557064705e-05, | |
| "loss": 2.9866, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6871210283773954e-05, | |
| "loss": 3.1094, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.68644730104832e-05, | |
| "loss": 2.8821, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6857735737192445e-05, | |
| "loss": 3.0871, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6850998463901694e-05, | |
| "loss": 3.3086, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6844261190610936e-05, | |
| "loss": 3.1617, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6837523917320185e-05, | |
| "loss": 3.0705, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.683078664402943e-05, | |
| "loss": 3.3558, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6824049370738676e-05, | |
| "loss": 3.2012, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6817312097447925e-05, | |
| "loss": 3.1229, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6810574824157173e-05, | |
| "loss": 3.1233, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6803837550866416e-05, | |
| "loss": 3.4662, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6797100277575664e-05, | |
| "loss": 3.5225, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6790363004284906e-05, | |
| "loss": 3.1466, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.678362573099415e-05, | |
| "loss": 3.3767, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.67768884577034e-05, | |
| "loss": 3.1327, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6770151184412646e-05, | |
| "loss": 3.0657, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6763413911121895e-05, | |
| "loss": 3.238, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.675667663783114e-05, | |
| "loss": 3.0663, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6749939364540386e-05, | |
| "loss": 3.2997, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6743202091249635e-05, | |
| "loss": 3.1138, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.673646481795888e-05, | |
| "loss": 3.1481, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6729727544668126e-05, | |
| "loss": 3.1374, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.672299027137737e-05, | |
| "loss": 3.137, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.671625299808662e-05, | |
| "loss": 3.0878, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.670951572479586e-05, | |
| "loss": 3.0358, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.670277845150511e-05, | |
| "loss": 3.0953, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.669604117821436e-05, | |
| "loss": 3.0776, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6689303904923606e-05, | |
| "loss": 3.4543, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.668256663163285e-05, | |
| "loss": 3.1356, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.66758293583421e-05, | |
| "loss": 3.3752, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.666909208505134e-05, | |
| "loss": 3.0158, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.666235481176059e-05, | |
| "loss": 3.1832, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.665561753846983e-05, | |
| "loss": 3.2367, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.664888026517908e-05, | |
| "loss": 3.2553, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.664214299188833e-05, | |
| "loss": 3.3684, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.663540571859757e-05, | |
| "loss": 3.157, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.662866844530682e-05, | |
| "loss": 3.3339, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.662193117201607e-05, | |
| "loss": 3.142, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.661519389872531e-05, | |
| "loss": 3.0608, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.660845662543455e-05, | |
| "loss": 3.0232, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.66017193521438e-05, | |
| "loss": 3.0909, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.659498207885305e-05, | |
| "loss": 3.0394, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.65882448055623e-05, | |
| "loss": 3.3022, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.658150753227154e-05, | |
| "loss": 3.0824, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.657477025898079e-05, | |
| "loss": 3.0483, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.656803298569004e-05, | |
| "loss": 3.0755, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.656129571239928e-05, | |
| "loss": 3.4422, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.655455843910852e-05, | |
| "loss": 3.1593, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.654782116581777e-05, | |
| "loss": 3.2496, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.654108389252702e-05, | |
| "loss": 3.3281, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.653434661923626e-05, | |
| "loss": 2.9645, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.652760934594551e-05, | |
| "loss": 3.4181, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.652087207265476e-05, | |
| "loss": 3.1505, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6514134799364e-05, | |
| "loss": 3.2713, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.650739752607325e-05, | |
| "loss": 3.0737, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.65006602527825e-05, | |
| "loss": 3.1046, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.649392297949174e-05, | |
| "loss": 3.0764, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6487185706200984e-05, | |
| "loss": 3.1572, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.648044843291023e-05, | |
| "loss": 3.1946, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.647371115961948e-05, | |
| "loss": 3.0024, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.646697388632873e-05, | |
| "loss": 3.1511, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.646023661303797e-05, | |
| "loss": 3.333, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.645349933974722e-05, | |
| "loss": 2.9438, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.644676206645647e-05, | |
| "loss": 3.327, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.644002479316571e-05, | |
| "loss": 3.0695, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.6433287519874954e-05, | |
| "loss": 3.1144, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.64265502465842e-05, | |
| "loss": 3.4706, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.641981297329345e-05, | |
| "loss": 3.1941, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6413075700002694e-05, | |
| "loss": 3.1067, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.640633842671194e-05, | |
| "loss": 3.1044, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.639960115342119e-05, | |
| "loss": 3.1814, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.639286388013044e-05, | |
| "loss": 3.2541, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.638612660683968e-05, | |
| "loss": 3.1815, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6379389333548925e-05, | |
| "loss": 3.1444, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6372652060258174e-05, | |
| "loss": 3.0728, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6365914786967416e-05, | |
| "loss": 3.2032, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6359177513676665e-05, | |
| "loss": 2.8255, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6352440240385914e-05, | |
| "loss": 3.1325, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.634570296709516e-05, | |
| "loss": 3.2602, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6338965693804405e-05, | |
| "loss": 2.9978, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6332228420513654e-05, | |
| "loss": 3.3716, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.63254911472229e-05, | |
| "loss": 3.2488, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6318753873932145e-05, | |
| "loss": 2.9851, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.631201660064139e-05, | |
| "loss": 2.9942, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6305279327350636e-05, | |
| "loss": 3.2165, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6298542054059884e-05, | |
| "loss": 3.0885, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6291804780769127e-05, | |
| "loss": 3.0966, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6285067507478375e-05, | |
| "loss": 3.2696, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6278330234187624e-05, | |
| "loss": 3.0802, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.627159296089687e-05, | |
| "loss": 3.2361, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6264855687606115e-05, | |
| "loss": 3.0949, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.625811841431536e-05, | |
| "loss": 3.3435, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6251381141024606e-05, | |
| "loss": 3.3882, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6244643867733855e-05, | |
| "loss": 2.8436, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.62379065944431e-05, | |
| "loss": 3.3431, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6231169321152346e-05, | |
| "loss": 3.4118, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6224432047861595e-05, | |
| "loss": 3.1914, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.621769477457084e-05, | |
| "loss": 3.061, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6210957501280086e-05, | |
| "loss": 3.1028, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.620422022798933e-05, | |
| "loss": 3.0601, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.619748295469858e-05, | |
| "loss": 3.3313, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.619074568140782e-05, | |
| "loss": 3.1585, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.618400840811707e-05, | |
| "loss": 2.9804, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.617727113482632e-05, | |
| "loss": 3.197, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6170533861535566e-05, | |
| "loss": 2.823, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.616379658824481e-05, | |
| "loss": 3.0995, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6157059314954057e-05, | |
| "loss": 3.2473, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.61503220416633e-05, | |
| "loss": 3.3361, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.614358476837254e-05, | |
| "loss": 2.9229, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.613684749508179e-05, | |
| "loss": 3.2953, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.613011022179104e-05, | |
| "loss": 3.1187, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.612337294850029e-05, | |
| "loss": 3.009, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.611663567520953e-05, | |
| "loss": 3.5421, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.610989840191878e-05, | |
| "loss": 3.0521, | |
| "step": 2895 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.610316112862803e-05, | |
| "loss": 2.9707, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.609642385533727e-05, | |
| "loss": 3.0522, | |
| "step": 2905 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.608968658204652e-05, | |
| "loss": 3.1429, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.608294930875576e-05, | |
| "loss": 3.1433, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.607621203546501e-05, | |
| "loss": 3.0362, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.606947476217425e-05, | |
| "loss": 3.0698, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.60627374888835e-05, | |
| "loss": 2.9708, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.605600021559275e-05, | |
| "loss": 2.929, | |
| "step": 2935 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.6049262942302e-05, | |
| "loss": 3.0652, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.604252566901124e-05, | |
| "loss": 3.0888, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.603578839572049e-05, | |
| "loss": 3.0333, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.602905112242973e-05, | |
| "loss": 3.1151, | |
| "step": 2955 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.602231384913897e-05, | |
| "loss": 3.1866, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.601557657584822e-05, | |
| "loss": 3.0419, | |
| "step": 2965 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.600883930255747e-05, | |
| "loss": 2.9358, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.600210202926672e-05, | |
| "loss": 3.3567, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.599536475597596e-05, | |
| "loss": 2.9169, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.598862748268521e-05, | |
| "loss": 3.0725, | |
| "step": 2985 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.598189020939446e-05, | |
| "loss": 3.4152, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.59751529361037e-05, | |
| "loss": 3.0889, | |
| "step": 2995 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.5968415662812944e-05, | |
| "loss": 3.3419, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.596167838952219e-05, | |
| "loss": 3.3641, | |
| "step": 3005 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.595494111623144e-05, | |
| "loss": 3.0926, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.5948203842940684e-05, | |
| "loss": 3.0744, | |
| "step": 3015 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.594146656964993e-05, | |
| "loss": 3.0896, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.593472929635918e-05, | |
| "loss": 3.2848, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.592799202306843e-05, | |
| "loss": 3.1748, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.592125474977767e-05, | |
| "loss": 3.3514, | |
| "step": 3035 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.591451747648692e-05, | |
| "loss": 3.2087, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.590778020319616e-05, | |
| "loss": 3.1578, | |
| "step": 3045 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.590104292990541e-05, | |
| "loss": 3.0495, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5894305656614654e-05, | |
| "loss": 3.0536, | |
| "step": 3055 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.58875683833239e-05, | |
| "loss": 3.1117, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.588083111003315e-05, | |
| "loss": 3.3505, | |
| "step": 3065 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5874093836742394e-05, | |
| "loss": 3.3008, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.586735656345164e-05, | |
| "loss": 2.9855, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.586061929016089e-05, | |
| "loss": 3.1741, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5853882016870134e-05, | |
| "loss": 3.1377, | |
| "step": 3085 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5847144743579376e-05, | |
| "loss": 3.2383, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5840407470288625e-05, | |
| "loss": 3.0129, | |
| "step": 3095 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5833670196997874e-05, | |
| "loss": 3.1036, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.582693292370712e-05, | |
| "loss": 3.1, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5820195650416365e-05, | |
| "loss": 3.21, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5813458377125614e-05, | |
| "loss": 3.1048, | |
| "step": 3115 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.580672110383486e-05, | |
| "loss": 3.2457, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5799983830544105e-05, | |
| "loss": 3.1279, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.579324655725335e-05, | |
| "loss": 3.247, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5786509283962596e-05, | |
| "loss": 3.2605, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5779772010671844e-05, | |
| "loss": 3.3872, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5773034737381087e-05, | |
| "loss": 2.9347, | |
| "step": 3145 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5766297464090335e-05, | |
| "loss": 3.0918, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5759560190799584e-05, | |
| "loss": 3.1318, | |
| "step": 3155 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.575282291750883e-05, | |
| "loss": 2.9989, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5746085644218075e-05, | |
| "loss": 3.2661, | |
| "step": 3165 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5739348370927324e-05, | |
| "loss": 3.2538, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5732611097636566e-05, | |
| "loss": 3.2111, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.572587382434581e-05, | |
| "loss": 3.0219, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.571913655105506e-05, | |
| "loss": 3.2369, | |
| "step": 3185 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5712399277764306e-05, | |
| "loss": 3.1409, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5705662004473555e-05, | |
| "loss": 3.1233, | |
| "step": 3195 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.56989247311828e-05, | |
| "loss": 3.2719, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5692187457892046e-05, | |
| "loss": 2.9358, | |
| "step": 3205 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5685450184601295e-05, | |
| "loss": 3.1265, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.567871291131054e-05, | |
| "loss": 3.0724, | |
| "step": 3215 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.567197563801978e-05, | |
| "loss": 3.0934, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.566523836472903e-05, | |
| "loss": 3.3711, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.565850109143828e-05, | |
| "loss": 2.8675, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.565176381814752e-05, | |
| "loss": 3.0045, | |
| "step": 3235 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.564502654485677e-05, | |
| "loss": 3.0661, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5638289271566017e-05, | |
| "loss": 3.1129, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5631551998275265e-05, | |
| "loss": 3.0902, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.562481472498451e-05, | |
| "loss": 3.1345, | |
| "step": 3255 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.561807745169375e-05, | |
| "loss": 2.9902, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5611340178403e-05, | |
| "loss": 3.191, | |
| "step": 3265 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.560460290511224e-05, | |
| "loss": 3.1233, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.559786563182149e-05, | |
| "loss": 3.203, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.559112835853074e-05, | |
| "loss": 3.112, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.558439108523999e-05, | |
| "loss": 3.0281, | |
| "step": 3285 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.557765381194923e-05, | |
| "loss": 3.3552, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.557091653865848e-05, | |
| "loss": 3.0418, | |
| "step": 3295 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.556417926536772e-05, | |
| "loss": 3.0119, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.555744199207697e-05, | |
| "loss": 3.003, | |
| "step": 3305 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.555070471878621e-05, | |
| "loss": 3.1365, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.554396744549546e-05, | |
| "loss": 3.0694, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.553723017220471e-05, | |
| "loss": 2.8513, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.553049289891395e-05, | |
| "loss": 3.1681, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.55237556256232e-05, | |
| "loss": 3.1393, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.551701835233245e-05, | |
| "loss": 3.0198, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.55102810790417e-05, | |
| "loss": 3.0511, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.550354380575094e-05, | |
| "loss": 3.1593, | |
| "step": 3345 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.549680653246018e-05, | |
| "loss": 2.9654, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.549006925916943e-05, | |
| "loss": 3.1572, | |
| "step": 3355 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.548333198587868e-05, | |
| "loss": 3.283, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.547659471258792e-05, | |
| "loss": 3.1824, | |
| "step": 3365 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.546985743929717e-05, | |
| "loss": 3.086, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.546312016600642e-05, | |
| "loss": 3.1806, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.545638289271566e-05, | |
| "loss": 3.2051, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.544964561942491e-05, | |
| "loss": 3.0539, | |
| "step": 3385 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.544290834613415e-05, | |
| "loss": 3.0205, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.54361710728434e-05, | |
| "loss": 3.2201, | |
| "step": 3395 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5429433799552644e-05, | |
| "loss": 3.2691, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.542269652626189e-05, | |
| "loss": 3.1167, | |
| "step": 3405 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.541595925297114e-05, | |
| "loss": 2.9539, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.540922197968039e-05, | |
| "loss": 3.1993, | |
| "step": 3415 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.540248470638963e-05, | |
| "loss": 3.1305, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.539574743309888e-05, | |
| "loss": 3.1626, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.538901015980812e-05, | |
| "loss": 3.2176, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5382272886517365e-05, | |
| "loss": 3.0691, | |
| "step": 3435 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5375535613226614e-05, | |
| "loss": 3.0611, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536879833993586e-05, | |
| "loss": 3.1824, | |
| "step": 3445 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536206106664511e-05, | |
| "loss": 3.1394, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5355323793354354e-05, | |
| "loss": 3.1272, | |
| "step": 3455 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.53485865200636e-05, | |
| "loss": 2.9905, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.534184924677285e-05, | |
| "loss": 2.8781, | |
| "step": 3465 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.53351119734821e-05, | |
| "loss": 3.1255, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.532837470019134e-05, | |
| "loss": 3.0489, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5321637426900585e-05, | |
| "loss": 3.1483, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5314900153609834e-05, | |
| "loss": 2.923, | |
| "step": 3485 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5308162880319076e-05, | |
| "loss": 3.1935, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5301425607028325e-05, | |
| "loss": 2.8445, | |
| "step": 3495 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5294688333737574e-05, | |
| "loss": 3.1171, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.528795106044682e-05, | |
| "loss": 3.2084, | |
| "step": 3505 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5281213787156065e-05, | |
| "loss": 3.1549, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.527447651386531e-05, | |
| "loss": 3.0279, | |
| "step": 3515 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5267739240574555e-05, | |
| "loss": 2.7719, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.52610019672838e-05, | |
| "loss": 3.1109, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5254264693993046e-05, | |
| "loss": 2.9881, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5247527420702295e-05, | |
| "loss": 3.2017, | |
| "step": 3535 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5240790147411544e-05, | |
| "loss": 2.9567, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5234052874120786e-05, | |
| "loss": 3.2274, | |
| "step": 3545 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5227315600830035e-05, | |
| "loss": 3.1517, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5220578327539284e-05, | |
| "loss": 3.0133, | |
| "step": 3555 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5213841054248526e-05, | |
| "loss": 3.2494, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.520710378095777e-05, | |
| "loss": 3.1607, | |
| "step": 3565 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.520036650766702e-05, | |
| "loss": 3.1622, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5193629234376266e-05, | |
| "loss": 3.0597, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.518689196108551e-05, | |
| "loss": 3.1157, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.518015468779476e-05, | |
| "loss": 3.1916, | |
| "step": 3585 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5173417414504006e-05, | |
| "loss": 3.3892, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5166680141213255e-05, | |
| "loss": 3.2352, | |
| "step": 3595 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.51599428679225e-05, | |
| "loss": 2.9455, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.515455304928989e-05, | |
| "loss": 3.2882, | |
| "step": 3605 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.514781577599914e-05, | |
| "loss": 3.1644, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.514107850270838e-05, | |
| "loss": 3.1653, | |
| "step": 3615 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.513434122941763e-05, | |
| "loss": 3.286, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.512760395612688e-05, | |
| "loss": 3.2771, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.512086668283613e-05, | |
| "loss": 3.4175, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.511412940954537e-05, | |
| "loss": 3.0911, | |
| "step": 3635 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.510739213625462e-05, | |
| "loss": 3.0502, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.510065486296386e-05, | |
| "loss": 3.0543, | |
| "step": 3645 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5093917589673105e-05, | |
| "loss": 3.125, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5087180316382353e-05, | |
| "loss": 3.1286, | |
| "step": 3655 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.50804430430916e-05, | |
| "loss": 3.1895, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.507370576980085e-05, | |
| "loss": 3.0078, | |
| "step": 3665 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.506696849651009e-05, | |
| "loss": 3.0766, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.506023122321934e-05, | |
| "loss": 3.4027, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.505349394992859e-05, | |
| "loss": 2.8891, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.504675667663783e-05, | |
| "loss": 3.0093, | |
| "step": 3685 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5040019403347075e-05, | |
| "loss": 3.1543, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5033282130056324e-05, | |
| "loss": 3.0005, | |
| "step": 3695 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.502654485676557e-05, | |
| "loss": 3.2393, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5019807583474815e-05, | |
| "loss": 3.2052, | |
| "step": 3705 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5013070310184064e-05, | |
| "loss": 3.1979, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.500633303689331e-05, | |
| "loss": 3.1785, | |
| "step": 3715 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.499959576360256e-05, | |
| "loss": 3.1174, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4992858490311804e-05, | |
| "loss": 3.0281, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4986121217021046e-05, | |
| "loss": 3.1682, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4979383943730295e-05, | |
| "loss": 2.9703, | |
| "step": 3735 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4972646670439544e-05, | |
| "loss": 3.0364, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4965909397148786e-05, | |
| "loss": 2.8145, | |
| "step": 3745 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4959172123858035e-05, | |
| "loss": 3.0781, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4952434850567283e-05, | |
| "loss": 3.0378, | |
| "step": 3755 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4945697577276526e-05, | |
| "loss": 3.1284, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4938960303985774e-05, | |
| "loss": 3.2207, | |
| "step": 3765 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4932223030695017e-05, | |
| "loss": 3.1658, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4925485757404265e-05, | |
| "loss": 2.9948, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.491874848411351e-05, | |
| "loss": 3.1124, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4912011210822756e-05, | |
| "loss": 2.9172, | |
| "step": 3785 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4905273937532005e-05, | |
| "loss": 3.22, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4898536664241254e-05, | |
| "loss": 2.9467, | |
| "step": 3795 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4891799390950496e-05, | |
| "loss": 3.1856, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4885062117659745e-05, | |
| "loss": 3.3563, | |
| "step": 3805 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4878324844368994e-05, | |
| "loss": 3.1518, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4871587571078236e-05, | |
| "loss": 3.1046, | |
| "step": 3815 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.486485029778748e-05, | |
| "loss": 3.1422, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.485811302449673e-05, | |
| "loss": 2.9912, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4851375751205976e-05, | |
| "loss": 3.3937, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.484463847791522e-05, | |
| "loss": 2.772, | |
| "step": 3835 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.483790120462447e-05, | |
| "loss": 3.1256, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4831163931333716e-05, | |
| "loss": 3.2421, | |
| "step": 3845 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4824426658042965e-05, | |
| "loss": 2.8323, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.481768938475221e-05, | |
| "loss": 3.0066, | |
| "step": 3855 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.481095211146145e-05, | |
| "loss": 3.2898, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.48042148381707e-05, | |
| "loss": 3.0331, | |
| "step": 3865 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.479747756487994e-05, | |
| "loss": 3.0689, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.479074029158919e-05, | |
| "loss": 2.9034, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.478400301829844e-05, | |
| "loss": 2.8014, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4777265745007686e-05, | |
| "loss": 2.984, | |
| "step": 3885 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.477052847171693e-05, | |
| "loss": 3.0642, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.476379119842618e-05, | |
| "loss": 2.9044, | |
| "step": 3895 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.475705392513542e-05, | |
| "loss": 3.2954, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.475031665184466e-05, | |
| "loss": 3.1445, | |
| "step": 3905 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.474357937855391e-05, | |
| "loss": 3.3147, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.473684210526316e-05, | |
| "loss": 3.2616, | |
| "step": 3915 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.473010483197241e-05, | |
| "loss": 3.3639, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.472336755868165e-05, | |
| "loss": 2.9422, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.47166302853909e-05, | |
| "loss": 3.193, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.470989301210015e-05, | |
| "loss": 3.0869, | |
| "step": 3935 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.47031557388094e-05, | |
| "loss": 3.097, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.469641846551864e-05, | |
| "loss": 3.2279, | |
| "step": 3945 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.468968119222788e-05, | |
| "loss": 3.4025, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.468294391893713e-05, | |
| "loss": 3.0516, | |
| "step": 3955 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.467620664564637e-05, | |
| "loss": 3.0323, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.466946937235562e-05, | |
| "loss": 2.7716, | |
| "step": 3965 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.466273209906487e-05, | |
| "loss": 3.1923, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.465599482577412e-05, | |
| "loss": 3.1534, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.464925755248336e-05, | |
| "loss": 2.8173, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.464252027919261e-05, | |
| "loss": 3.0176, | |
| "step": 3985 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.463578300590185e-05, | |
| "loss": 3.2045, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.46290457326111e-05, | |
| "loss": 3.0714, | |
| "step": 3995 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.462230845932034e-05, | |
| "loss": 3.0958, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_loss": 3.0753977298736572, | |
| "eval_rouge2_fmeasure": 0.0055, | |
| "eval_rouge2_precision": 0.0064, | |
| "eval_rouge2_recall": 0.0063, | |
| "eval_runtime": 2992.0589, | |
| "eval_samples_per_second": 0.092, | |
| "eval_steps_per_second": 0.046, | |
| "step": 4000 | |
| } | |
| ], | |
| "max_steps": 37107, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.56090117390336e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |