| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 2442, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.012285012285012284, | |
| "grad_norm": 20.35053560709185, | |
| "learning_rate": 6.530612244897961e-07, | |
| "loss": 1.0266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 1.016836166381836, | |
| "step": 5, | |
| "valid_targets_mean": 1221.5, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 0.02457002457002457, | |
| "grad_norm": 16.603510355056123, | |
| "learning_rate": 1.469387755102041e-06, | |
| "loss": 0.9886, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.9607698917388916, | |
| "step": 10, | |
| "valid_targets_mean": 1451.9, | |
| "valid_targets_min": 971 | |
| }, | |
| { | |
| "epoch": 0.036855036855036855, | |
| "grad_norm": 16.816054244261874, | |
| "learning_rate": 2.285714285714286e-06, | |
| "loss": 0.972, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.965086042881012, | |
| "step": 15, | |
| "valid_targets_mean": 1169.6, | |
| "valid_targets_min": 739 | |
| }, | |
| { | |
| "epoch": 0.04914004914004914, | |
| "grad_norm": 10.473733849996067, | |
| "learning_rate": 3.1020408163265307e-06, | |
| "loss": 0.8641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.829804003238678, | |
| "step": 20, | |
| "valid_targets_mean": 1340.2, | |
| "valid_targets_min": 853 | |
| }, | |
| { | |
| "epoch": 0.06142506142506143, | |
| "grad_norm": 6.719886648102692, | |
| "learning_rate": 3.9183673469387755e-06, | |
| "loss": 0.779, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.7282828092575073, | |
| "step": 25, | |
| "valid_targets_mean": 1187.4, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 0.07371007371007371, | |
| "grad_norm": 4.642242575672464, | |
| "learning_rate": 4.734693877551021e-06, | |
| "loss": 0.6757, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6520389914512634, | |
| "step": 30, | |
| "valid_targets_mean": 1321.6, | |
| "valid_targets_min": 568 | |
| }, | |
| { | |
| "epoch": 0.085995085995086, | |
| "grad_norm": 2.7965705811975607, | |
| "learning_rate": 5.551020408163266e-06, | |
| "loss": 0.6122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5742930173873901, | |
| "step": 35, | |
| "valid_targets_mean": 1361.1, | |
| "valid_targets_min": 654 | |
| }, | |
| { | |
| "epoch": 0.09828009828009827, | |
| "grad_norm": 1.9647310620540863, | |
| "learning_rate": 6.36734693877551e-06, | |
| "loss": 0.5417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5204732418060303, | |
| "step": 40, | |
| "valid_targets_mean": 1321.1, | |
| "valid_targets_min": 574 | |
| }, | |
| { | |
| "epoch": 0.11056511056511056, | |
| "grad_norm": 1.6336158421098663, | |
| "learning_rate": 7.183673469387755e-06, | |
| "loss": 0.4913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.48145467042922974, | |
| "step": 45, | |
| "valid_targets_mean": 1290.8, | |
| "valid_targets_min": 531 | |
| }, | |
| { | |
| "epoch": 0.12285012285012285, | |
| "grad_norm": 1.6098593343945324, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.4652, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.45056235790252686, | |
| "step": 50, | |
| "valid_targets_mean": 1294.9, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 0.13513513513513514, | |
| "grad_norm": 1.487675000715524, | |
| "learning_rate": 8.816326530612247e-06, | |
| "loss": 0.4215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.41585201025009155, | |
| "step": 55, | |
| "valid_targets_mean": 1332.2, | |
| "valid_targets_min": 794 | |
| }, | |
| { | |
| "epoch": 0.14742014742014742, | |
| "grad_norm": 1.5044656855552325, | |
| "learning_rate": 9.63265306122449e-06, | |
| "loss": 0.4064, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3813883662223816, | |
| "step": 60, | |
| "valid_targets_mean": 1311.1, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 0.1597051597051597, | |
| "grad_norm": 1.3262454599596, | |
| "learning_rate": 1.0448979591836737e-05, | |
| "loss": 0.3753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36576443910598755, | |
| "step": 65, | |
| "valid_targets_mean": 1226.4, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 0.171990171990172, | |
| "grad_norm": 1.1566857700781938, | |
| "learning_rate": 1.126530612244898e-05, | |
| "loss": 0.372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37514474987983704, | |
| "step": 70, | |
| "valid_targets_mean": 1539.2, | |
| "valid_targets_min": 768 | |
| }, | |
| { | |
| "epoch": 0.18427518427518427, | |
| "grad_norm": 1.2676859471136912, | |
| "learning_rate": 1.2081632653061225e-05, | |
| "loss": 0.3477, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33851388096809387, | |
| "step": 75, | |
| "valid_targets_mean": 1114.2, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 0.19656019656019655, | |
| "grad_norm": 1.3741420718567798, | |
| "learning_rate": 1.2897959183673469e-05, | |
| "loss": 0.3266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3353644013404846, | |
| "step": 80, | |
| "valid_targets_mean": 1143.4, | |
| "valid_targets_min": 712 | |
| }, | |
| { | |
| "epoch": 0.20884520884520885, | |
| "grad_norm": 1.107327074899049, | |
| "learning_rate": 1.3714285714285716e-05, | |
| "loss": 0.3201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31138500571250916, | |
| "step": 85, | |
| "valid_targets_mean": 1249.1, | |
| "valid_targets_min": 590 | |
| }, | |
| { | |
| "epoch": 0.22113022113022113, | |
| "grad_norm": 1.2916481544068563, | |
| "learning_rate": 1.4530612244897961e-05, | |
| "loss": 0.3025, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30402475595474243, | |
| "step": 90, | |
| "valid_targets_mean": 1252.0, | |
| "valid_targets_min": 649 | |
| }, | |
| { | |
| "epoch": 0.2334152334152334, | |
| "grad_norm": 1.2219729652086049, | |
| "learning_rate": 1.5346938775510204e-05, | |
| "loss": 0.3136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31273770332336426, | |
| "step": 95, | |
| "valid_targets_mean": 1185.1, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.2457002457002457, | |
| "grad_norm": 1.0254894202707463, | |
| "learning_rate": 1.616326530612245e-05, | |
| "loss": 0.3132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2911428213119507, | |
| "step": 100, | |
| "valid_targets_mean": 1281.2, | |
| "valid_targets_min": 940 | |
| }, | |
| { | |
| "epoch": 0.257985257985258, | |
| "grad_norm": 1.19134839145044, | |
| "learning_rate": 1.6979591836734695e-05, | |
| "loss": 0.3108, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3086431920528412, | |
| "step": 105, | |
| "valid_targets_mean": 1279.0, | |
| "valid_targets_min": 711 | |
| }, | |
| { | |
| "epoch": 0.2702702702702703, | |
| "grad_norm": 1.051295827944603, | |
| "learning_rate": 1.779591836734694e-05, | |
| "loss": 0.2949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3161376118659973, | |
| "step": 110, | |
| "valid_targets_mean": 1383.6, | |
| "valid_targets_min": 897 | |
| }, | |
| { | |
| "epoch": 0.28255528255528256, | |
| "grad_norm": 1.001465215723732, | |
| "learning_rate": 1.8612244897959185e-05, | |
| "loss": 0.2868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27010855078697205, | |
| "step": 115, | |
| "valid_targets_mean": 1363.2, | |
| "valid_targets_min": 813 | |
| }, | |
| { | |
| "epoch": 0.29484029484029484, | |
| "grad_norm": 1.0939416312953611, | |
| "learning_rate": 1.942857142857143e-05, | |
| "loss": 0.2882, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29551124572753906, | |
| "step": 120, | |
| "valid_targets_mean": 1317.6, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 0.3071253071253071, | |
| "grad_norm": 1.017797618825815, | |
| "learning_rate": 2.0244897959183672e-05, | |
| "loss": 0.2896, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27586787939071655, | |
| "step": 125, | |
| "valid_targets_mean": 1418.5, | |
| "valid_targets_min": 941 | |
| }, | |
| { | |
| "epoch": 0.3194103194103194, | |
| "grad_norm": 1.2127705134680413, | |
| "learning_rate": 2.106122448979592e-05, | |
| "loss": 0.2911, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2948208451271057, | |
| "step": 130, | |
| "valid_targets_mean": 1013.2, | |
| "valid_targets_min": 663 | |
| }, | |
| { | |
| "epoch": 0.3316953316953317, | |
| "grad_norm": 1.1881461995379652, | |
| "learning_rate": 2.1877551020408166e-05, | |
| "loss": 0.2837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30349838733673096, | |
| "step": 135, | |
| "valid_targets_mean": 1376.4, | |
| "valid_targets_min": 727 | |
| }, | |
| { | |
| "epoch": 0.343980343980344, | |
| "grad_norm": 1.0220364541266478, | |
| "learning_rate": 2.269387755102041e-05, | |
| "loss": 0.2945, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31276723742485046, | |
| "step": 140, | |
| "valid_targets_mean": 1266.6, | |
| "valid_targets_min": 637 | |
| }, | |
| { | |
| "epoch": 0.35626535626535627, | |
| "grad_norm": 1.0259584649302287, | |
| "learning_rate": 2.3510204081632656e-05, | |
| "loss": 0.2735, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28521037101745605, | |
| "step": 145, | |
| "valid_targets_mean": 1381.8, | |
| "valid_targets_min": 724 | |
| }, | |
| { | |
| "epoch": 0.36855036855036855, | |
| "grad_norm": 1.387643384965655, | |
| "learning_rate": 2.4326530612244898e-05, | |
| "loss": 0.2848, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28743278980255127, | |
| "step": 150, | |
| "valid_targets_mean": 1333.6, | |
| "valid_targets_min": 646 | |
| }, | |
| { | |
| "epoch": 0.3808353808353808, | |
| "grad_norm": 1.104844187565239, | |
| "learning_rate": 2.5142857142857143e-05, | |
| "loss": 0.2726, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.267228364944458, | |
| "step": 155, | |
| "valid_targets_mean": 1163.2, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 0.3931203931203931, | |
| "grad_norm": 1.0781870060932552, | |
| "learning_rate": 2.5959183673469392e-05, | |
| "loss": 0.2835, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2731441855430603, | |
| "step": 160, | |
| "valid_targets_mean": 1182.2, | |
| "valid_targets_min": 800 | |
| }, | |
| { | |
| "epoch": 0.40540540540540543, | |
| "grad_norm": 1.2518533850963875, | |
| "learning_rate": 2.6775510204081637e-05, | |
| "loss": 0.273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2699888348579407, | |
| "step": 165, | |
| "valid_targets_mean": 1239.4, | |
| "valid_targets_min": 713 | |
| }, | |
| { | |
| "epoch": 0.4176904176904177, | |
| "grad_norm": 1.168759326203342, | |
| "learning_rate": 2.7591836734693882e-05, | |
| "loss": 0.2688, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26571792364120483, | |
| "step": 170, | |
| "valid_targets_mean": 1398.9, | |
| "valid_targets_min": 598 | |
| }, | |
| { | |
| "epoch": 0.42997542997543, | |
| "grad_norm": 0.9396552751253561, | |
| "learning_rate": 2.8408163265306124e-05, | |
| "loss": 0.2762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2691813111305237, | |
| "step": 175, | |
| "valid_targets_mean": 1485.3, | |
| "valid_targets_min": 872 | |
| }, | |
| { | |
| "epoch": 0.44226044226044225, | |
| "grad_norm": 1.2096875417684458, | |
| "learning_rate": 2.922448979591837e-05, | |
| "loss": 0.2787, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2808350920677185, | |
| "step": 180, | |
| "valid_targets_mean": 1162.1, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 0.45454545454545453, | |
| "grad_norm": 1.1999093025296843, | |
| "learning_rate": 3.0040816326530614e-05, | |
| "loss": 0.2667, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2688295841217041, | |
| "step": 185, | |
| "valid_targets_mean": 1101.2, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 0.4668304668304668, | |
| "grad_norm": 1.1647181597312102, | |
| "learning_rate": 3.085714285714286e-05, | |
| "loss": 0.2624, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2831493616104126, | |
| "step": 190, | |
| "valid_targets_mean": 1266.9, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 0.47911547911547914, | |
| "grad_norm": 1.0284185612708663, | |
| "learning_rate": 3.1673469387755105e-05, | |
| "loss": 0.2602, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24393121898174286, | |
| "step": 195, | |
| "valid_targets_mean": 1164.9, | |
| "valid_targets_min": 756 | |
| }, | |
| { | |
| "epoch": 0.4914004914004914, | |
| "grad_norm": 1.0003942487517488, | |
| "learning_rate": 3.2489795918367346e-05, | |
| "loss": 0.2523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22005441784858704, | |
| "step": 200, | |
| "valid_targets_mean": 1244.3, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 0.5036855036855037, | |
| "grad_norm": 1.0107391467706224, | |
| "learning_rate": 3.3306122448979595e-05, | |
| "loss": 0.2636, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2617947459220886, | |
| "step": 205, | |
| "valid_targets_mean": 1234.1, | |
| "valid_targets_min": 874 | |
| }, | |
| { | |
| "epoch": 0.515970515970516, | |
| "grad_norm": 1.2118728098280296, | |
| "learning_rate": 3.4122448979591843e-05, | |
| "loss": 0.2615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2575775384902954, | |
| "step": 210, | |
| "valid_targets_mean": 1321.1, | |
| "valid_targets_min": 861 | |
| }, | |
| { | |
| "epoch": 0.5282555282555282, | |
| "grad_norm": 0.9892491299567787, | |
| "learning_rate": 3.4938775510204085e-05, | |
| "loss": 0.249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2387700229883194, | |
| "step": 215, | |
| "valid_targets_mean": 1299.9, | |
| "valid_targets_min": 1000 | |
| }, | |
| { | |
| "epoch": 0.5405405405405406, | |
| "grad_norm": 1.0482471985131325, | |
| "learning_rate": 3.575510204081633e-05, | |
| "loss": 0.2583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2581407427787781, | |
| "step": 220, | |
| "valid_targets_mean": 1370.2, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 0.5528255528255528, | |
| "grad_norm": 1.0295148997419448, | |
| "learning_rate": 3.6571428571428576e-05, | |
| "loss": 0.2548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24649953842163086, | |
| "step": 225, | |
| "valid_targets_mean": 1169.7, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 0.5651105651105651, | |
| "grad_norm": 0.9681847418509463, | |
| "learning_rate": 3.738775510204082e-05, | |
| "loss": 0.2573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27282947301864624, | |
| "step": 230, | |
| "valid_targets_mean": 1279.6, | |
| "valid_targets_min": 809 | |
| }, | |
| { | |
| "epoch": 0.5773955773955773, | |
| "grad_norm": 0.9238070431221829, | |
| "learning_rate": 3.8204081632653066e-05, | |
| "loss": 0.252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24655944108963013, | |
| "step": 235, | |
| "valid_targets_mean": 1360.4, | |
| "valid_targets_min": 725 | |
| }, | |
| { | |
| "epoch": 0.5896805896805897, | |
| "grad_norm": 0.9098797568997322, | |
| "learning_rate": 3.902040816326531e-05, | |
| "loss": 0.2625, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23307578265666962, | |
| "step": 240, | |
| "valid_targets_mean": 1383.3, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.601965601965602, | |
| "grad_norm": 1.006785214329797, | |
| "learning_rate": 3.983673469387755e-05, | |
| "loss": 0.2594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26228320598602295, | |
| "step": 245, | |
| "valid_targets_mean": 1284.8, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 0.6142506142506142, | |
| "grad_norm": 1.0525191631080537, | |
| "learning_rate": 3.9999672841332876e-05, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23748120665550232, | |
| "step": 250, | |
| "valid_targets_mean": 1343.1, | |
| "valid_targets_min": 643 | |
| }, | |
| { | |
| "epoch": 0.6265356265356266, | |
| "grad_norm": 1.0681183261616904, | |
| "learning_rate": 3.999834377759164e-05, | |
| "loss": 0.2604, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25798824429512024, | |
| "step": 255, | |
| "valid_targets_mean": 1271.4, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 0.6388206388206388, | |
| "grad_norm": 1.039669566160986, | |
| "learning_rate": 3.999599242924703e-05, | |
| "loss": 0.2573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26076722145080566, | |
| "step": 260, | |
| "valid_targets_mean": 1133.4, | |
| "valid_targets_min": 577 | |
| }, | |
| { | |
| "epoch": 0.6511056511056511, | |
| "grad_norm": 1.0724973460550207, | |
| "learning_rate": 3.999261891649637e-05, | |
| "loss": 0.2576, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23584957420825958, | |
| "step": 265, | |
| "valid_targets_mean": 1405.9, | |
| "valid_targets_min": 979 | |
| }, | |
| { | |
| "epoch": 0.6633906633906634, | |
| "grad_norm": 0.9104706365824741, | |
| "learning_rate": 3.9988223411788436e-05, | |
| "loss": 0.2593, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2530933618545532, | |
| "step": 270, | |
| "valid_targets_mean": 1466.0, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 0.6756756756756757, | |
| "grad_norm": 1.0963583541448956, | |
| "learning_rate": 3.998280613981468e-05, | |
| "loss": 0.247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2414504736661911, | |
| "step": 275, | |
| "valid_targets_mean": 1352.9, | |
| "valid_targets_min": 819 | |
| }, | |
| { | |
| "epoch": 0.687960687960688, | |
| "grad_norm": 0.9882989853095971, | |
| "learning_rate": 3.9976367377497725e-05, | |
| "loss": 0.2519, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24604278802871704, | |
| "step": 280, | |
| "valid_targets_mean": 1113.4, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.7002457002457002, | |
| "grad_norm": 0.995524650089483, | |
| "learning_rate": 3.99689074539772e-05, | |
| "loss": 0.2503, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24476918578147888, | |
| "step": 285, | |
| "valid_targets_mean": 1168.4, | |
| "valid_targets_min": 599 | |
| }, | |
| { | |
| "epoch": 0.7125307125307125, | |
| "grad_norm": 0.9891596679542455, | |
| "learning_rate": 3.9960426750592936e-05, | |
| "loss": 0.2454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25576359033584595, | |
| "step": 290, | |
| "valid_targets_mean": 1179.9, | |
| "valid_targets_min": 833 | |
| }, | |
| { | |
| "epoch": 0.7248157248157249, | |
| "grad_norm": 0.9551067978223167, | |
| "learning_rate": 3.995092570086546e-05, | |
| "loss": 0.2498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2365557849407196, | |
| "step": 295, | |
| "valid_targets_mean": 1248.8, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 0.7371007371007371, | |
| "grad_norm": 0.8993619363801122, | |
| "learning_rate": 3.9940404790473825e-05, | |
| "loss": 0.2459, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.230590358376503, | |
| "step": 300, | |
| "valid_targets_mean": 1189.8, | |
| "valid_targets_min": 656 | |
| }, | |
| { | |
| "epoch": 0.7493857493857494, | |
| "grad_norm": 0.9950400883383362, | |
| "learning_rate": 3.992886455723082e-05, | |
| "loss": 0.2491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25279349088668823, | |
| "step": 305, | |
| "valid_targets_mean": 1129.0, | |
| "valid_targets_min": 546 | |
| }, | |
| { | |
| "epoch": 0.7616707616707616, | |
| "grad_norm": 0.8527006472613714, | |
| "learning_rate": 3.991630559105541e-05, | |
| "loss": 0.2356, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22874727845191956, | |
| "step": 310, | |
| "valid_targets_mean": 1332.6, | |
| "valid_targets_min": 828 | |
| }, | |
| { | |
| "epoch": 0.773955773955774, | |
| "grad_norm": 1.0419255233532783, | |
| "learning_rate": 3.990272853394268e-05, | |
| "loss": 0.2446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22870038449764252, | |
| "step": 315, | |
| "valid_targets_mean": 1102.8, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 0.7862407862407862, | |
| "grad_norm": 0.8716345618424153, | |
| "learning_rate": 3.988813407993089e-05, | |
| "loss": 0.2261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23537375032901764, | |
| "step": 320, | |
| "valid_targets_mean": 1293.5, | |
| "valid_targets_min": 848 | |
| }, | |
| { | |
| "epoch": 0.7985257985257985, | |
| "grad_norm": 0.8810969006097469, | |
| "learning_rate": 3.987252297506613e-05, | |
| "loss": 0.24, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24338498711585999, | |
| "step": 325, | |
| "valid_targets_mean": 1233.5, | |
| "valid_targets_min": 659 | |
| }, | |
| { | |
| "epoch": 0.8108108108108109, | |
| "grad_norm": 0.9626770961212678, | |
| "learning_rate": 3.9855896017364075e-05, | |
| "loss": 0.2354, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2470875233411789, | |
| "step": 330, | |
| "valid_targets_mean": 1239.8, | |
| "valid_targets_min": 559 | |
| }, | |
| { | |
| "epoch": 0.8230958230958231, | |
| "grad_norm": 0.9176247579572656, | |
| "learning_rate": 3.983825405676927e-05, | |
| "loss": 0.2432, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23234277963638306, | |
| "step": 335, | |
| "valid_targets_mean": 1354.6, | |
| "valid_targets_min": 898 | |
| }, | |
| { | |
| "epoch": 0.8353808353808354, | |
| "grad_norm": 0.8965361140622664, | |
| "learning_rate": 3.981959799511161e-05, | |
| "loss": 0.24, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23358391225337982, | |
| "step": 340, | |
| "valid_targets_mean": 1369.1, | |
| "valid_targets_min": 613 | |
| }, | |
| { | |
| "epoch": 0.8476658476658476, | |
| "grad_norm": 1.073981004545888, | |
| "learning_rate": 3.979992878606032e-05, | |
| "loss": 0.2499, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2327095866203308, | |
| "step": 345, | |
| "valid_targets_mean": 1122.8, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 0.85995085995086, | |
| "grad_norm": 1.0151351479751622, | |
| "learning_rate": 3.977924743507513e-05, | |
| "loss": 0.2483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23525191843509674, | |
| "step": 350, | |
| "valid_targets_mean": 1256.2, | |
| "valid_targets_min": 592 | |
| }, | |
| { | |
| "epoch": 0.8722358722358723, | |
| "grad_norm": 0.9282381133656055, | |
| "learning_rate": 3.975755499935492e-05, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27992555499076843, | |
| "step": 355, | |
| "valid_targets_mean": 1211.7, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 0.8845208845208845, | |
| "grad_norm": 0.7744300003477048, | |
| "learning_rate": 3.973485258778368e-05, | |
| "loss": 0.2308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2320326268672943, | |
| "step": 360, | |
| "valid_targets_mean": 1523.9, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 0.8968058968058968, | |
| "grad_norm": 0.8915272025257586, | |
| "learning_rate": 3.971114136087379e-05, | |
| "loss": 0.2398, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23646889626979828, | |
| "step": 365, | |
| "valid_targets_mean": 1359.6, | |
| "valid_targets_min": 834 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 0.9986391400817158, | |
| "learning_rate": 3.968642253070675e-05, | |
| "loss": 0.2347, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23117730021476746, | |
| "step": 370, | |
| "valid_targets_mean": 1226.4, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 0.9213759213759214, | |
| "grad_norm": 1.0663907580975795, | |
| "learning_rate": 3.966069736087116e-05, | |
| "loss": 0.2464, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2413148581981659, | |
| "step": 375, | |
| "valid_targets_mean": 1369.7, | |
| "valid_targets_min": 569 | |
| }, | |
| { | |
| "epoch": 0.9336609336609336, | |
| "grad_norm": 0.9688455896189119, | |
| "learning_rate": 3.963396716639818e-05, | |
| "loss": 0.2346, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2318524718284607, | |
| "step": 380, | |
| "valid_targets_mean": 1273.9, | |
| "valid_targets_min": 855 | |
| }, | |
| { | |
| "epoch": 0.9459459459459459, | |
| "grad_norm": 0.9406165224783458, | |
| "learning_rate": 3.960623331369427e-05, | |
| "loss": 0.2352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20918339490890503, | |
| "step": 385, | |
| "valid_targets_mean": 1067.1, | |
| "valid_targets_min": 833 | |
| }, | |
| { | |
| "epoch": 0.9582309582309583, | |
| "grad_norm": 0.8493184795160157, | |
| "learning_rate": 3.957749722047138e-05, | |
| "loss": 0.2309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23976099491119385, | |
| "step": 390, | |
| "valid_targets_mean": 1263.5, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 0.9705159705159705, | |
| "grad_norm": 0.8446197614217856, | |
| "learning_rate": 3.9547760355674405e-05, | |
| "loss": 0.2222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2245807945728302, | |
| "step": 395, | |
| "valid_targets_mean": 1357.5, | |
| "valid_targets_min": 860 | |
| }, | |
| { | |
| "epoch": 0.9828009828009828, | |
| "grad_norm": 0.9002227045931153, | |
| "learning_rate": 3.951702423940621e-05, | |
| "loss": 0.2343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2399313747882843, | |
| "step": 400, | |
| "valid_targets_mean": 1242.1, | |
| "valid_targets_min": 668 | |
| }, | |
| { | |
| "epoch": 0.995085995085995, | |
| "grad_norm": 0.9146265445163199, | |
| "learning_rate": 3.948529044284981e-05, | |
| "loss": 0.2268, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24605955183506012, | |
| "step": 405, | |
| "valid_targets_mean": 1379.3, | |
| "valid_targets_min": 605 | |
| }, | |
| { | |
| "epoch": 1.0073710073710074, | |
| "grad_norm": 0.8846119726465934, | |
| "learning_rate": 3.9452560588188135e-05, | |
| "loss": 0.2383, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21427395939826965, | |
| "step": 410, | |
| "valid_targets_mean": 1292.4, | |
| "valid_targets_min": 625 | |
| }, | |
| { | |
| "epoch": 1.0196560196560196, | |
| "grad_norm": 0.8429800757595586, | |
| "learning_rate": 3.9418836348521045e-05, | |
| "loss": 0.2308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22194769978523254, | |
| "step": 415, | |
| "valid_targets_mean": 1537.8, | |
| "valid_targets_min": 854 | |
| }, | |
| { | |
| "epoch": 1.031941031941032, | |
| "grad_norm": 1.0879669131500758, | |
| "learning_rate": 3.9384119447779854e-05, | |
| "loss": 0.2216, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21954546868801117, | |
| "step": 420, | |
| "valid_targets_mean": 1291.0, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 1.0442260442260443, | |
| "grad_norm": 0.91150482238475, | |
| "learning_rate": 3.934841166063919e-05, | |
| "loss": 0.218, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21371571719646454, | |
| "step": 425, | |
| "valid_targets_mean": 1232.7, | |
| "valid_targets_min": 555 | |
| }, | |
| { | |
| "epoch": 1.0565110565110565, | |
| "grad_norm": 0.9859894476975108, | |
| "learning_rate": 3.931171481242625e-05, | |
| "loss": 0.2078, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20356549322605133, | |
| "step": 430, | |
| "valid_targets_mean": 1253.1, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 1.0687960687960687, | |
| "grad_norm": 0.929488975303167, | |
| "learning_rate": 3.927403077902753e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22038447856903076, | |
| "step": 435, | |
| "valid_targets_mean": 1190.2, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 1.0810810810810811, | |
| "grad_norm": 0.9198148139807539, | |
| "learning_rate": 3.9235361486792905e-05, | |
| "loss": 0.2213, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23593559861183167, | |
| "step": 440, | |
| "valid_targets_mean": 1318.4, | |
| "valid_targets_min": 859 | |
| }, | |
| { | |
| "epoch": 1.0933660933660934, | |
| "grad_norm": 0.865539205726412, | |
| "learning_rate": 3.9195708912437176e-05, | |
| "loss": 0.2166, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.199513241648674, | |
| "step": 445, | |
| "valid_targets_mean": 1262.8, | |
| "valid_targets_min": 832 | |
| }, | |
| { | |
| "epoch": 1.1056511056511056, | |
| "grad_norm": 0.8862960683914471, | |
| "learning_rate": 3.915507508293901e-05, | |
| "loss": 0.2114, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2223977893590927, | |
| "step": 450, | |
| "valid_targets_mean": 1404.8, | |
| "valid_targets_min": 928 | |
| }, | |
| { | |
| "epoch": 1.117936117936118, | |
| "grad_norm": 0.8859574165063384, | |
| "learning_rate": 3.911346207543734e-05, | |
| "loss": 0.2101, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21623465418815613, | |
| "step": 455, | |
| "valid_targets_mean": 1318.5, | |
| "valid_targets_min": 676 | |
| }, | |
| { | |
| "epoch": 1.1302211302211302, | |
| "grad_norm": 1.0054461891918525, | |
| "learning_rate": 3.907087201712515e-05, | |
| "loss": 0.2314, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23490525782108307, | |
| "step": 460, | |
| "valid_targets_mean": 1302.8, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 1.1425061425061425, | |
| "grad_norm": 0.8741596133273296, | |
| "learning_rate": 3.902730708514078e-05, | |
| "loss": 0.2061, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18964213132858276, | |
| "step": 465, | |
| "valid_targets_mean": 1216.9, | |
| "valid_targets_min": 707 | |
| }, | |
| { | |
| "epoch": 1.154791154791155, | |
| "grad_norm": 0.9441571131207358, | |
| "learning_rate": 3.8982769506456616e-05, | |
| "loss": 0.2135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23256821930408478, | |
| "step": 470, | |
| "valid_targets_mean": 1205.5, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 1.1670761670761671, | |
| "grad_norm": 0.8475920842038459, | |
| "learning_rate": 3.893726155776524e-05, | |
| "loss": 0.2165, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19832265377044678, | |
| "step": 475, | |
| "valid_targets_mean": 1206.9, | |
| "valid_targets_min": 552 | |
| }, | |
| { | |
| "epoch": 1.1793611793611793, | |
| "grad_norm": 0.9251072573203696, | |
| "learning_rate": 3.8890785565363046e-05, | |
| "loss": 0.2103, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2146698236465454, | |
| "step": 480, | |
| "valid_targets_mean": 1227.5, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 1.1916461916461916, | |
| "grad_norm": 0.9740444472309585, | |
| "learning_rate": 3.884334390503136e-05, | |
| "loss": 0.2194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21822363138198853, | |
| "step": 485, | |
| "valid_targets_mean": 1097.0, | |
| "valid_targets_min": 623 | |
| }, | |
| { | |
| "epoch": 1.203931203931204, | |
| "grad_norm": 0.8370410301775648, | |
| "learning_rate": 3.8794939001914955e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22085630893707275, | |
| "step": 490, | |
| "valid_targets_mean": 1390.9, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 1.2162162162162162, | |
| "grad_norm": 0.9186018225806699, | |
| "learning_rate": 3.87455733303981e-05, | |
| "loss": 0.2189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2234152853488922, | |
| "step": 495, | |
| "valid_targets_mean": 1152.8, | |
| "valid_targets_min": 559 | |
| }, | |
| { | |
| "epoch": 1.2285012285012284, | |
| "grad_norm": 0.8405706051298435, | |
| "learning_rate": 3.869524941397805e-05, | |
| "loss": 0.2137, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21818791329860687, | |
| "step": 500, | |
| "valid_targets_mean": 1246.1, | |
| "valid_targets_min": 702 | |
| }, | |
| { | |
| "epoch": 1.2407862407862407, | |
| "grad_norm": 0.9209574543994296, | |
| "learning_rate": 3.8643969825136095e-05, | |
| "loss": 0.2152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21757091581821442, | |
| "step": 505, | |
| "valid_targets_mean": 1152.7, | |
| "valid_targets_min": 599 | |
| }, | |
| { | |
| "epoch": 1.253071253071253, | |
| "grad_norm": 0.8067547007661393, | |
| "learning_rate": 3.8591737185206024e-05, | |
| "loss": 0.2153, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.203495055437088, | |
| "step": 510, | |
| "valid_targets_mean": 1283.2, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 1.2653562653562653, | |
| "grad_norm": 0.8723905147518646, | |
| "learning_rate": 3.853855416424011e-05, | |
| "loss": 0.2124, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22284701466560364, | |
| "step": 515, | |
| "valid_targets_mean": 1208.0, | |
| "valid_targets_min": 666 | |
| }, | |
| { | |
| "epoch": 1.2776412776412776, | |
| "grad_norm": 0.8107646956453654, | |
| "learning_rate": 3.848442348087267e-05, | |
| "loss": 0.2131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20091310143470764, | |
| "step": 520, | |
| "valid_targets_mean": 1367.6, | |
| "valid_targets_min": 764 | |
| }, | |
| { | |
| "epoch": 1.28992628992629, | |
| "grad_norm": 0.8840227552984795, | |
| "learning_rate": 3.842934790218106e-05, | |
| "loss": 0.2109, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21159666776657104, | |
| "step": 525, | |
| "valid_targets_mean": 1263.2, | |
| "valid_targets_min": 856 | |
| }, | |
| { | |
| "epoch": 1.3022113022113022, | |
| "grad_norm": 0.9823970363721356, | |
| "learning_rate": 3.837333024354422e-05, | |
| "loss": 0.2136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22919949889183044, | |
| "step": 530, | |
| "valid_targets_mean": 1240.1, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 1.3144963144963144, | |
| "grad_norm": 1.735042209364941, | |
| "learning_rate": 3.8316373368498794e-05, | |
| "loss": 0.1981, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21260014176368713, | |
| "step": 535, | |
| "valid_targets_mean": 1401.2, | |
| "valid_targets_min": 604 | |
| }, | |
| { | |
| "epoch": 1.3267813267813269, | |
| "grad_norm": 0.8022885779180352, | |
| "learning_rate": 3.82584801885927e-05, | |
| "loss": 0.2155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21027901768684387, | |
| "step": 540, | |
| "valid_targets_mean": 1354.0, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 1.339066339066339, | |
| "grad_norm": 0.7971643641976438, | |
| "learning_rate": 3.8199653663236336e-05, | |
| "loss": 0.2175, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22587084770202637, | |
| "step": 545, | |
| "valid_targets_mean": 1381.9, | |
| "valid_targets_min": 957 | |
| }, | |
| { | |
| "epoch": 1.3513513513513513, | |
| "grad_norm": 0.8515238470974948, | |
| "learning_rate": 3.813989679955128e-05, | |
| "loss": 0.2102, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20897898077964783, | |
| "step": 550, | |
| "valid_targets_mean": 1361.9, | |
| "valid_targets_min": 826 | |
| }, | |
| { | |
| "epoch": 1.3636363636363638, | |
| "grad_norm": 0.8555526583280557, | |
| "learning_rate": 3.8079212652216595e-05, | |
| "loss": 0.2045, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20616090297698975, | |
| "step": 555, | |
| "valid_targets_mean": 1297.0, | |
| "valid_targets_min": 602 | |
| }, | |
| { | |
| "epoch": 1.375921375921376, | |
| "grad_norm": 0.8964923469359567, | |
| "learning_rate": 3.8017604323312616e-05, | |
| "loss": 0.2068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20442214608192444, | |
| "step": 560, | |
| "valid_targets_mean": 1286.3, | |
| "valid_targets_min": 717 | |
| }, | |
| { | |
| "epoch": 1.3882063882063882, | |
| "grad_norm": 1.0497145720091325, | |
| "learning_rate": 3.795507496216246e-05, | |
| "loss": 0.2201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22913996875286102, | |
| "step": 565, | |
| "valid_targets_mean": 1299.9, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 1.4004914004914004, | |
| "grad_norm": 0.8618095265826493, | |
| "learning_rate": 3.789162776517098e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2160976678133011, | |
| "step": 570, | |
| "valid_targets_mean": 1281.5, | |
| "valid_targets_min": 677 | |
| }, | |
| { | |
| "epoch": 1.4127764127764126, | |
| "grad_norm": 0.8211183730608299, | |
| "learning_rate": 3.78272659756614e-05, | |
| "loss": 0.2173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23328492045402527, | |
| "step": 575, | |
| "valid_targets_mean": 1361.1, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 1.425061425061425, | |
| "grad_norm": 0.8273749298069341, | |
| "learning_rate": 3.776199288370948e-05, | |
| "loss": 0.2199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23211979866027832, | |
| "step": 580, | |
| "valid_targets_mean": 1328.4, | |
| "valid_targets_min": 851 | |
| }, | |
| { | |
| "epoch": 1.4373464373464373, | |
| "grad_norm": 0.8559553834741782, | |
| "learning_rate": 3.7695811825975386e-05, | |
| "loss": 0.2052, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19386297464370728, | |
| "step": 585, | |
| "valid_targets_mean": 1047.1, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 1.4496314496314495, | |
| "grad_norm": 0.7737724552422564, | |
| "learning_rate": 3.76287261855331e-05, | |
| "loss": 0.2065, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20705577731132507, | |
| "step": 590, | |
| "valid_targets_mean": 1450.2, | |
| "valid_targets_min": 906 | |
| }, | |
| { | |
| "epoch": 1.461916461916462, | |
| "grad_norm": 0.7544548232380602, | |
| "learning_rate": 3.7560739391697465e-05, | |
| "loss": 0.2157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18719275295734406, | |
| "step": 595, | |
| "valid_targets_mean": 1226.5, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 1.4742014742014742, | |
| "grad_norm": 0.8636863005403776, | |
| "learning_rate": 3.749185491984891e-05, | |
| "loss": 0.2099, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20051462948322296, | |
| "step": 600, | |
| "valid_targets_mean": 1257.8, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 1.4864864864864864, | |
| "grad_norm": 0.7920583706263344, | |
| "learning_rate": 3.7422076291255785e-05, | |
| "loss": 0.1975, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19033858180046082, | |
| "step": 605, | |
| "valid_targets_mean": 1164.7, | |
| "valid_targets_min": 677 | |
| }, | |
| { | |
| "epoch": 1.4987714987714988, | |
| "grad_norm": 0.8334219155321024, | |
| "learning_rate": 3.7351407072894356e-05, | |
| "loss": 0.2082, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20213520526885986, | |
| "step": 610, | |
| "valid_targets_mean": 1286.6, | |
| "valid_targets_min": 957 | |
| }, | |
| { | |
| "epoch": 1.511056511056511, | |
| "grad_norm": 0.9313275078782892, | |
| "learning_rate": 3.7279850877266486e-05, | |
| "loss": 0.2194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22305332124233246, | |
| "step": 615, | |
| "valid_targets_mean": 1265.1, | |
| "valid_targets_min": 727 | |
| }, | |
| { | |
| "epoch": 1.5233415233415233, | |
| "grad_norm": 0.8208145495971357, | |
| "learning_rate": 3.720741136221491e-05, | |
| "loss": 0.2074, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21243973076343536, | |
| "step": 620, | |
| "valid_targets_mean": 1292.1, | |
| "valid_targets_min": 704 | |
| }, | |
| { | |
| "epoch": 1.5356265356265357, | |
| "grad_norm": 0.8307696500817636, | |
| "learning_rate": 3.713409223073636e-05, | |
| "loss": 0.214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21302366256713867, | |
| "step": 625, | |
| "valid_targets_mean": 1229.5, | |
| "valid_targets_min": 501 | |
| }, | |
| { | |
| "epoch": 1.547911547911548, | |
| "grad_norm": 0.8467719298005856, | |
| "learning_rate": 3.705989723079214e-05, | |
| "loss": 0.2172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22414334118366241, | |
| "step": 630, | |
| "valid_targets_mean": 1386.8, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 1.5601965601965602, | |
| "grad_norm": 0.8338920310244855, | |
| "learning_rate": 3.698483015511665e-05, | |
| "loss": 0.2093, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21425983309745789, | |
| "step": 635, | |
| "valid_targets_mean": 1345.6, | |
| "valid_targets_min": 608 | |
| }, | |
| { | |
| "epoch": 1.5724815724815726, | |
| "grad_norm": 0.847033756979806, | |
| "learning_rate": 3.690889484102344e-05, | |
| "loss": 0.2073, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19102904200553894, | |
| "step": 640, | |
| "valid_targets_mean": 1151.4, | |
| "valid_targets_min": 587 | |
| }, | |
| { | |
| "epoch": 1.5847665847665846, | |
| "grad_norm": 0.8357252424027696, | |
| "learning_rate": 3.683209517020908e-05, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22609484195709229, | |
| "step": 645, | |
| "valid_targets_mean": 1317.7, | |
| "valid_targets_min": 678 | |
| }, | |
| { | |
| "epoch": 1.597051597051597, | |
| "grad_norm": 0.8978628194123708, | |
| "learning_rate": 3.675443506855473e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25122907757759094, | |
| "step": 650, | |
| "valid_targets_mean": 1355.8, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 1.6093366093366095, | |
| "grad_norm": 0.8144918816074354, | |
| "learning_rate": 3.6675918505925456e-05, | |
| "loss": 0.2125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19278162717819214, | |
| "step": 655, | |
| "valid_targets_mean": 1167.2, | |
| "valid_targets_min": 622 | |
| }, | |
| { | |
| "epoch": 1.6216216216216215, | |
| "grad_norm": 0.8678549604409258, | |
| "learning_rate": 3.6596549495967276e-05, | |
| "loss": 0.212, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2218804657459259, | |
| "step": 660, | |
| "valid_targets_mean": 1284.0, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 1.633906633906634, | |
| "grad_norm": 0.8353937245559396, | |
| "learning_rate": 3.651633209590202e-05, | |
| "loss": 0.2123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18236777186393738, | |
| "step": 665, | |
| "valid_targets_mean": 1267.4, | |
| "valid_targets_min": 532 | |
| }, | |
| { | |
| "epoch": 1.6461916461916462, | |
| "grad_norm": 0.7951973802311657, | |
| "learning_rate": 3.6435270406319914e-05, | |
| "loss": 0.2258, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2141144722700119, | |
| "step": 670, | |
| "valid_targets_mean": 1371.6, | |
| "valid_targets_min": 691 | |
| }, | |
| { | |
| "epoch": 1.6584766584766584, | |
| "grad_norm": 0.8471980022480241, | |
| "learning_rate": 3.635336857096997e-05, | |
| "loss": 0.2086, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2074558138847351, | |
| "step": 675, | |
| "valid_targets_mean": 1272.0, | |
| "valid_targets_min": 863 | |
| }, | |
| { | |
| "epoch": 1.6707616707616708, | |
| "grad_norm": 0.7615407401541552, | |
| "learning_rate": 3.627063077654815e-05, | |
| "loss": 0.2102, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20035895705223083, | |
| "step": 680, | |
| "valid_targets_mean": 1336.2, | |
| "valid_targets_min": 538 | |
| }, | |
| { | |
| "epoch": 1.683046683046683, | |
| "grad_norm": 0.8588070921853436, | |
| "learning_rate": 3.618706125248337e-05, | |
| "loss": 0.2054, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20719598233699799, | |
| "step": 685, | |
| "valid_targets_mean": 1390.2, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 1.6953316953316953, | |
| "grad_norm": 0.8305147911432151, | |
| "learning_rate": 3.6102664270721275e-05, | |
| "loss": 0.2149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21375998854637146, | |
| "step": 690, | |
| "valid_targets_mean": 1308.5, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 1.7076167076167077, | |
| "grad_norm": 0.7458152286575466, | |
| "learning_rate": 3.601744414550589e-05, | |
| "loss": 0.206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20337368547916412, | |
| "step": 695, | |
| "valid_targets_mean": 1402.0, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 1.71990171990172, | |
| "grad_norm": 0.8430079820292512, | |
| "learning_rate": 3.593140523315906e-05, | |
| "loss": 0.2147, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2111029326915741, | |
| "step": 700, | |
| "valid_targets_mean": 1263.1, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 1.7321867321867321, | |
| "grad_norm": 0.8587637085294313, | |
| "learning_rate": 3.584455193185778e-05, | |
| "loss": 0.2138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20514604449272156, | |
| "step": 705, | |
| "valid_targets_mean": 1260.1, | |
| "valid_targets_min": 641 | |
| }, | |
| { | |
| "epoch": 1.7444717444717446, | |
| "grad_norm": 0.8419086609463237, | |
| "learning_rate": 3.575688868140933e-05, | |
| "loss": 0.2184, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20621389150619507, | |
| "step": 710, | |
| "valid_targets_mean": 1387.4, | |
| "valid_targets_min": 796 | |
| }, | |
| { | |
| "epoch": 1.7567567567567568, | |
| "grad_norm": 0.7702914519938786, | |
| "learning_rate": 3.566841996302438e-05, | |
| "loss": 0.209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20190760493278503, | |
| "step": 715, | |
| "valid_targets_mean": 1362.8, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 1.769041769041769, | |
| "grad_norm": 0.8055966192320041, | |
| "learning_rate": 3.557915029908787e-05, | |
| "loss": 0.2151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2246687412261963, | |
| "step": 720, | |
| "valid_targets_mean": 1284.2, | |
| "valid_targets_min": 775 | |
| }, | |
| { | |
| "epoch": 1.7813267813267815, | |
| "grad_norm": 0.7810151508322989, | |
| "learning_rate": 3.548908425292784e-05, | |
| "loss": 0.209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19987574219703674, | |
| "step": 725, | |
| "valid_targets_mean": 1195.1, | |
| "valid_targets_min": 765 | |
| }, | |
| { | |
| "epoch": 1.7936117936117935, | |
| "grad_norm": 0.7745765687362091, | |
| "learning_rate": 3.5398226428582165e-05, | |
| "loss": 0.2051, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1902502477169037, | |
| "step": 730, | |
| "valid_targets_mean": 1208.6, | |
| "valid_targets_min": 576 | |
| }, | |
| { | |
| "epoch": 1.805896805896806, | |
| "grad_norm": 0.7740403978164567, | |
| "learning_rate": 3.530658147056321e-05, | |
| "loss": 0.2059, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21077097952365875, | |
| "step": 735, | |
| "valid_targets_mean": 1501.9, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 1.8181818181818183, | |
| "grad_norm": 0.8831871306665532, | |
| "learning_rate": 3.521415406362041e-05, | |
| "loss": 0.2102, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22450268268585205, | |
| "step": 740, | |
| "valid_targets_mean": 1230.2, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 1.8304668304668303, | |
| "grad_norm": 0.8075930439615072, | |
| "learning_rate": 3.512094893250076e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2014818787574768, | |
| "step": 745, | |
| "valid_targets_mean": 1086.2, | |
| "valid_targets_min": 592 | |
| }, | |
| { | |
| "epoch": 1.8427518427518428, | |
| "grad_norm": 0.9351262133752251, | |
| "learning_rate": 3.5026970841707366e-05, | |
| "loss": 0.2097, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22045360505580902, | |
| "step": 750, | |
| "valid_targets_mean": 1351.4, | |
| "valid_targets_min": 936 | |
| }, | |
| { | |
| "epoch": 1.855036855036855, | |
| "grad_norm": 0.8524642243083412, | |
| "learning_rate": 3.493222459525579e-05, | |
| "loss": 0.198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2044043391942978, | |
| "step": 755, | |
| "valid_targets_mean": 1232.6, | |
| "valid_targets_min": 750 | |
| }, | |
| { | |
| "epoch": 1.8673218673218672, | |
| "grad_norm": 0.8233788904636303, | |
| "learning_rate": 3.483671503642858e-05, | |
| "loss": 0.2055, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20760342478752136, | |
| "step": 760, | |
| "valid_targets_mean": 1309.0, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 1.8796068796068797, | |
| "grad_norm": 0.7752715357533787, | |
| "learning_rate": 3.474044704752761e-05, | |
| "loss": 0.209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20890532433986664, | |
| "step": 765, | |
| "valid_targets_mean": 1293.1, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 1.8918918918918919, | |
| "grad_norm": 0.697679262289484, | |
| "learning_rate": 3.464342554962454e-05, | |
| "loss": 0.1996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20702123641967773, | |
| "step": 770, | |
| "valid_targets_mean": 1639.2, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 1.904176904176904, | |
| "grad_norm": 0.7414242700794923, | |
| "learning_rate": 3.4545655502309254e-05, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18738913536071777, | |
| "step": 775, | |
| "valid_targets_mean": 1272.2, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 1.9164619164619165, | |
| "grad_norm": 0.9334643511915244, | |
| "learning_rate": 3.444714190343633e-05, | |
| "loss": 0.2068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.201995387673378, | |
| "step": 780, | |
| "valid_targets_mean": 1324.6, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 1.9287469287469288, | |
| "grad_norm": 0.8252648993168636, | |
| "learning_rate": 3.434788978886957e-05, | |
| "loss": 0.2086, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21050789952278137, | |
| "step": 785, | |
| "valid_targets_mean": 1299.9, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 1.941031941031941, | |
| "grad_norm": 0.9170958801978001, | |
| "learning_rate": 3.424790423222455e-05, | |
| "loss": 0.1949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20576564967632294, | |
| "step": 790, | |
| "valid_targets_mean": 1248.1, | |
| "valid_targets_min": 578 | |
| }, | |
| { | |
| "epoch": 1.9533169533169534, | |
| "grad_norm": 0.8198947258723515, | |
| "learning_rate": 3.414719034460928e-05, | |
| "loss": 0.199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20504584908485413, | |
| "step": 795, | |
| "valid_targets_mean": 1324.3, | |
| "valid_targets_min": 854 | |
| }, | |
| { | |
| "epoch": 1.9656019656019657, | |
| "grad_norm": 0.8253301027208539, | |
| "learning_rate": 3.404575327436294e-05, | |
| "loss": 0.2024, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1956181824207306, | |
| "step": 800, | |
| "valid_targets_mean": 1276.2, | |
| "valid_targets_min": 620 | |
| }, | |
| { | |
| "epoch": 1.9778869778869779, | |
| "grad_norm": 0.8661318138286722, | |
| "learning_rate": 3.3943598206792665e-05, | |
| "loss": 0.1978, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1938505470752716, | |
| "step": 805, | |
| "valid_targets_mean": 1168.2, | |
| "valid_targets_min": 665 | |
| }, | |
| { | |
| "epoch": 1.9901719901719903, | |
| "grad_norm": 0.8202639838791619, | |
| "learning_rate": 3.384073036390857e-05, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2048560529947281, | |
| "step": 810, | |
| "valid_targets_mean": 1392.2, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 2.0024570024570023, | |
| "grad_norm": 0.7431550083256849, | |
| "learning_rate": 3.373715500415667e-05, | |
| "loss": 0.2027, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17889058589935303, | |
| "step": 815, | |
| "valid_targets_mean": 1384.7, | |
| "valid_targets_min": 758 | |
| }, | |
| { | |
| "epoch": 2.0147420147420148, | |
| "grad_norm": 0.8960360020207319, | |
| "learning_rate": 3.363287742215023e-05, | |
| "loss": 0.1729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1715434193611145, | |
| "step": 820, | |
| "valid_targets_mean": 1223.9, | |
| "valid_targets_min": 806 | |
| }, | |
| { | |
| "epoch": 2.027027027027027, | |
| "grad_norm": 0.8448839647298884, | |
| "learning_rate": 3.352790294839898e-05, | |
| "loss": 0.1839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17698851227760315, | |
| "step": 825, | |
| "valid_targets_mean": 1208.3, | |
| "valid_targets_min": 608 | |
| }, | |
| { | |
| "epoch": 2.039312039312039, | |
| "grad_norm": 0.8771397300799911, | |
| "learning_rate": 3.3422236949036726e-05, | |
| "loss": 0.1775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1854882389307022, | |
| "step": 830, | |
| "valid_targets_mean": 1115.2, | |
| "valid_targets_min": 527 | |
| }, | |
| { | |
| "epoch": 2.0515970515970516, | |
| "grad_norm": 0.9375067706527694, | |
| "learning_rate": 3.331588482554697e-05, | |
| "loss": 0.1765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1746768355369568, | |
| "step": 835, | |
| "valid_targets_mean": 1309.6, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 2.063882063882064, | |
| "grad_norm": 0.9320660058665986, | |
| "learning_rate": 3.320885201448684e-05, | |
| "loss": 0.1804, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1711595505475998, | |
| "step": 840, | |
| "valid_targets_mean": 1270.1, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 2.076167076167076, | |
| "grad_norm": 0.9625843914370038, | |
| "learning_rate": 3.310114398720917e-05, | |
| "loss": 0.1812, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17891713976860046, | |
| "step": 845, | |
| "valid_targets_mean": 1182.7, | |
| "valid_targets_min": 713 | |
| }, | |
| { | |
| "epoch": 2.0884520884520885, | |
| "grad_norm": 0.8667658013763913, | |
| "learning_rate": 3.299276624958281e-05, | |
| "loss": 0.1858, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19869863986968994, | |
| "step": 850, | |
| "valid_targets_mean": 1383.4, | |
| "valid_targets_min": 672 | |
| }, | |
| { | |
| "epoch": 2.100737100737101, | |
| "grad_norm": 0.7829119159941206, | |
| "learning_rate": 3.288372434171116e-05, | |
| "loss": 0.1767, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1694278120994568, | |
| "step": 855, | |
| "valid_targets_mean": 1320.6, | |
| "valid_targets_min": 742 | |
| }, | |
| { | |
| "epoch": 2.113022113022113, | |
| "grad_norm": 1.0847289260076243, | |
| "learning_rate": 3.2774023837648986e-05, | |
| "loss": 0.1836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20229297876358032, | |
| "step": 860, | |
| "valid_targets_mean": 1242.6, | |
| "valid_targets_min": 591 | |
| }, | |
| { | |
| "epoch": 2.1253071253071254, | |
| "grad_norm": 0.7918900295560403, | |
| "learning_rate": 3.26636703451175e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16940709948539734, | |
| "step": 865, | |
| "valid_targets_mean": 1091.4, | |
| "valid_targets_min": 536 | |
| }, | |
| { | |
| "epoch": 2.1375921375921374, | |
| "grad_norm": 0.8050057115605104, | |
| "learning_rate": 3.2552669505217646e-05, | |
| "loss": 0.1804, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18291515111923218, | |
| "step": 870, | |
| "valid_targets_mean": 1270.8, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 2.14987714987715, | |
| "grad_norm": 0.8294882963284373, | |
| "learning_rate": 3.24410269921418e-05, | |
| "loss": 0.1789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16050593554973602, | |
| "step": 875, | |
| "valid_targets_mean": 1112.7, | |
| "valid_targets_min": 807 | |
| }, | |
| { | |
| "epoch": 2.1621621621621623, | |
| "grad_norm": 0.8227131541030049, | |
| "learning_rate": 3.232874851288367e-05, | |
| "loss": 0.1809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1674197018146515, | |
| "step": 880, | |
| "valid_targets_mean": 1254.1, | |
| "valid_targets_min": 706 | |
| }, | |
| { | |
| "epoch": 2.1744471744471743, | |
| "grad_norm": 0.8557010633720166, | |
| "learning_rate": 3.221583980694659e-05, | |
| "loss": 0.1827, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17029841244220734, | |
| "step": 885, | |
| "valid_targets_mean": 1191.7, | |
| "valid_targets_min": 742 | |
| }, | |
| { | |
| "epoch": 2.1867321867321867, | |
| "grad_norm": 0.7837923417661988, | |
| "learning_rate": 3.21023066460501e-05, | |
| "loss": 0.175, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18245768547058105, | |
| "step": 890, | |
| "valid_targets_mean": 1483.2, | |
| "valid_targets_min": 826 | |
| }, | |
| { | |
| "epoch": 2.199017199017199, | |
| "grad_norm": 0.8390698587520268, | |
| "learning_rate": 3.198815483383492e-05, | |
| "loss": 0.1809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17464491724967957, | |
| "step": 895, | |
| "valid_targets_mean": 1178.8, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 2.211302211302211, | |
| "grad_norm": 0.8929645135048639, | |
| "learning_rate": 3.1873390205566295e-05, | |
| "loss": 0.1809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19350141286849976, | |
| "step": 900, | |
| "valid_targets_mean": 1195.6, | |
| "valid_targets_min": 647 | |
| }, | |
| { | |
| "epoch": 2.2235872235872236, | |
| "grad_norm": 0.800799760583441, | |
| "learning_rate": 3.175801862783565e-05, | |
| "loss": 0.1802, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18399956822395325, | |
| "step": 905, | |
| "valid_targets_mean": 1394.0, | |
| "valid_targets_min": 762 | |
| }, | |
| { | |
| "epoch": 2.235872235872236, | |
| "grad_norm": 0.861812106568346, | |
| "learning_rate": 3.164204599826077e-05, | |
| "loss": 0.1794, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18631041049957275, | |
| "step": 910, | |
| "valid_targets_mean": 1266.8, | |
| "valid_targets_min": 707 | |
| }, | |
| { | |
| "epoch": 2.248157248157248, | |
| "grad_norm": 0.8136719833422565, | |
| "learning_rate": 3.1525478245184245e-05, | |
| "loss": 0.1776, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2019607126712799, | |
| "step": 915, | |
| "valid_targets_mean": 1316.0, | |
| "valid_targets_min": 760 | |
| }, | |
| { | |
| "epoch": 2.2604422604422605, | |
| "grad_norm": 0.7104387985554499, | |
| "learning_rate": 3.140832132737051e-05, | |
| "loss": 0.1754, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17880454659461975, | |
| "step": 920, | |
| "valid_targets_mean": 1386.5, | |
| "valid_targets_min": 745 | |
| }, | |
| { | |
| "epoch": 2.2727272727272725, | |
| "grad_norm": 0.723347779565406, | |
| "learning_rate": 3.129058123370116e-05, | |
| "loss": 0.1823, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1798139065504074, | |
| "step": 925, | |
| "valid_targets_mean": 1418.9, | |
| "valid_targets_min": 806 | |
| }, | |
| { | |
| "epoch": 2.285012285012285, | |
| "grad_norm": 0.8227450691436261, | |
| "learning_rate": 3.117226398286887e-05, | |
| "loss": 0.1764, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17978253960609436, | |
| "step": 930, | |
| "valid_targets_mean": 1198.0, | |
| "valid_targets_min": 637 | |
| }, | |
| { | |
| "epoch": 2.2972972972972974, | |
| "grad_norm": 0.8181621906002657, | |
| "learning_rate": 3.105337562306968e-05, | |
| "loss": 0.185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18528032302856445, | |
| "step": 935, | |
| "valid_targets_mean": 1467.1, | |
| "valid_targets_min": 939 | |
| }, | |
| { | |
| "epoch": 2.30958230958231, | |
| "grad_norm": 0.8628823590456987, | |
| "learning_rate": 3.0933922231693854e-05, | |
| "loss": 0.1788, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18646690249443054, | |
| "step": 940, | |
| "valid_targets_mean": 1327.8, | |
| "valid_targets_min": 768 | |
| }, | |
| { | |
| "epoch": 2.321867321867322, | |
| "grad_norm": 0.8312067805411888, | |
| "learning_rate": 3.08139099150152e-05, | |
| "loss": 0.189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19820252060890198, | |
| "step": 945, | |
| "valid_targets_mean": 1358.0, | |
| "valid_targets_min": 902 | |
| }, | |
| { | |
| "epoch": 2.3341523341523343, | |
| "grad_norm": 0.7391010728907489, | |
| "learning_rate": 3.069334480787893e-05, | |
| "loss": 0.1789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17447155714035034, | |
| "step": 950, | |
| "valid_targets_mean": 1403.9, | |
| "valid_targets_min": 564 | |
| }, | |
| { | |
| "epoch": 2.3464373464373462, | |
| "grad_norm": 0.8098616372815506, | |
| "learning_rate": 3.057223307338806e-05, | |
| "loss": 0.1828, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18852627277374268, | |
| "step": 955, | |
| "valid_targets_mean": 1456.1, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 2.3587223587223587, | |
| "grad_norm": 0.8698102007694762, | |
| "learning_rate": 3.0450580902588346e-05, | |
| "loss": 0.1778, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18752536177635193, | |
| "step": 960, | |
| "valid_targets_mean": 1266.6, | |
| "valid_targets_min": 777 | |
| }, | |
| { | |
| "epoch": 2.371007371007371, | |
| "grad_norm": 0.7876664418489986, | |
| "learning_rate": 3.032839451415182e-05, | |
| "loss": 0.1836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1748538315296173, | |
| "step": 965, | |
| "valid_targets_mean": 1193.6, | |
| "valid_targets_min": 457 | |
| }, | |
| { | |
| "epoch": 2.383292383292383, | |
| "grad_norm": 0.7113855892489315, | |
| "learning_rate": 3.0205680154058904e-05, | |
| "loss": 0.1775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1739535927772522, | |
| "step": 970, | |
| "valid_targets_mean": 1514.7, | |
| "valid_targets_min": 697 | |
| }, | |
| { | |
| "epoch": 2.3955773955773956, | |
| "grad_norm": 0.8142211234131692, | |
| "learning_rate": 3.0082444095279117e-05, | |
| "loss": 0.1758, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1813521385192871, | |
| "step": 975, | |
| "valid_targets_mean": 1334.5, | |
| "valid_targets_min": 708 | |
| }, | |
| { | |
| "epoch": 2.407862407862408, | |
| "grad_norm": 0.8197074469826403, | |
| "learning_rate": 2.9958692637450406e-05, | |
| "loss": 0.1753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16367939114570618, | |
| "step": 980, | |
| "valid_targets_mean": 1228.6, | |
| "valid_targets_min": 643 | |
| }, | |
| { | |
| "epoch": 2.42014742014742, | |
| "grad_norm": 0.8541938784954854, | |
| "learning_rate": 2.983443210655714e-05, | |
| "loss": 0.1735, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17588171362876892, | |
| "step": 985, | |
| "valid_targets_mean": 1190.8, | |
| "valid_targets_min": 682 | |
| }, | |
| { | |
| "epoch": 2.4324324324324325, | |
| "grad_norm": 0.7846425314619295, | |
| "learning_rate": 2.9709668854606706e-05, | |
| "loss": 0.1882, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18857240676879883, | |
| "step": 990, | |
| "valid_targets_mean": 1355.1, | |
| "valid_targets_min": 614 | |
| }, | |
| { | |
| "epoch": 2.444717444717445, | |
| "grad_norm": 0.7576298642207422, | |
| "learning_rate": 2.9584409259304828e-05, | |
| "loss": 0.1844, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17525619268417358, | |
| "step": 995, | |
| "valid_targets_mean": 1319.9, | |
| "valid_targets_min": 770 | |
| }, | |
| { | |
| "epoch": 2.457002457002457, | |
| "grad_norm": 0.8880135398527781, | |
| "learning_rate": 2.945865972372954e-05, | |
| "loss": 0.1805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1825505644083023, | |
| "step": 1000, | |
| "valid_targets_mean": 1444.0, | |
| "valid_targets_min": 958 | |
| }, | |
| { | |
| "epoch": 2.4692874692874693, | |
| "grad_norm": 0.744044465614355, | |
| "learning_rate": 2.9332426676003858e-05, | |
| "loss": 0.1837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17693683505058289, | |
| "step": 1005, | |
| "valid_targets_mean": 1339.9, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 2.4815724815724813, | |
| "grad_norm": 0.7799185661658911, | |
| "learning_rate": 2.920571656896722e-05, | |
| "loss": 0.178, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17789432406425476, | |
| "step": 1010, | |
| "valid_targets_mean": 1302.0, | |
| "valid_targets_min": 782 | |
| }, | |
| { | |
| "epoch": 2.493857493857494, | |
| "grad_norm": 0.8361418168804058, | |
| "learning_rate": 2.907853587984558e-05, | |
| "loss": 0.1792, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18084871768951416, | |
| "step": 1015, | |
| "valid_targets_mean": 1297.5, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 2.506142506142506, | |
| "grad_norm": 0.8486435962691962, | |
| "learning_rate": 2.8950891109920333e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18357740342617035, | |
| "step": 1020, | |
| "valid_targets_mean": 1246.6, | |
| "valid_targets_min": 770 | |
| }, | |
| { | |
| "epoch": 2.5184275184275187, | |
| "grad_norm": 0.807405719311044, | |
| "learning_rate": 2.882278878419597e-05, | |
| "loss": 0.1764, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1806846410036087, | |
| "step": 1025, | |
| "valid_targets_mean": 1253.2, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 2.5307125307125307, | |
| "grad_norm": 0.8520779946572644, | |
| "learning_rate": 2.8694235451066538e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1823638379573822, | |
| "step": 1030, | |
| "valid_targets_mean": 1461.3, | |
| "valid_targets_min": 856 | |
| }, | |
| { | |
| "epoch": 2.542997542997543, | |
| "grad_norm": 1.0429862912852512, | |
| "learning_rate": 2.8565237681980876e-05, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17990875244140625, | |
| "step": 1035, | |
| "valid_targets_mean": 1360.1, | |
| "valid_targets_min": 836 | |
| }, | |
| { | |
| "epoch": 2.555282555282555, | |
| "grad_norm": 0.7262570397634495, | |
| "learning_rate": 2.843580207110672e-05, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1686376929283142, | |
| "step": 1040, | |
| "valid_targets_mean": 1346.9, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 2.5675675675675675, | |
| "grad_norm": 0.7531761897578519, | |
| "learning_rate": 2.830593523499361e-05, | |
| "loss": 0.1803, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18114441633224487, | |
| "step": 1045, | |
| "valid_targets_mean": 1371.8, | |
| "valid_targets_min": 771 | |
| }, | |
| { | |
| "epoch": 2.57985257985258, | |
| "grad_norm": 0.793803225751772, | |
| "learning_rate": 2.8175643812234627e-05, | |
| "loss": 0.1778, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17945504188537598, | |
| "step": 1050, | |
| "valid_targets_mean": 1413.5, | |
| "valid_targets_min": 826 | |
| }, | |
| { | |
| "epoch": 2.592137592137592, | |
| "grad_norm": 0.9123261601174015, | |
| "learning_rate": 2.8044934463127108e-05, | |
| "loss": 0.1805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1872619390487671, | |
| "step": 1055, | |
| "valid_targets_mean": 1261.8, | |
| "valid_targets_min": 759 | |
| }, | |
| { | |
| "epoch": 2.6044226044226044, | |
| "grad_norm": 0.7979823264894937, | |
| "learning_rate": 2.7913813869332112e-05, | |
| "loss": 0.1753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1702994406223297, | |
| "step": 1060, | |
| "valid_targets_mean": 1177.8, | |
| "valid_targets_min": 654 | |
| }, | |
| { | |
| "epoch": 2.616707616707617, | |
| "grad_norm": 0.7996971813748655, | |
| "learning_rate": 2.7782288733532915e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1733815222978592, | |
| "step": 1065, | |
| "valid_targets_mean": 1196.7, | |
| "valid_targets_min": 671 | |
| }, | |
| { | |
| "epoch": 2.628992628992629, | |
| "grad_norm": 0.8323833609213411, | |
| "learning_rate": 2.7650365779092346e-05, | |
| "loss": 0.1765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17033511400222778, | |
| "step": 1070, | |
| "valid_targets_mean": 1238.1, | |
| "valid_targets_min": 632 | |
| }, | |
| { | |
| "epoch": 2.6412776412776413, | |
| "grad_norm": 0.8537456313959788, | |
| "learning_rate": 2.751805174970912e-05, | |
| "loss": 0.1815, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1890188306570053, | |
| "step": 1075, | |
| "valid_targets_mean": 1366.6, | |
| "valid_targets_min": 724 | |
| }, | |
| { | |
| "epoch": 2.6535626535626538, | |
| "grad_norm": 0.8217031611080989, | |
| "learning_rate": 2.7385353409073093e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1882573664188385, | |
| "step": 1080, | |
| "valid_targets_mean": 1193.4, | |
| "valid_targets_min": 626 | |
| }, | |
| { | |
| "epoch": 2.6658476658476657, | |
| "grad_norm": 0.8348829168323537, | |
| "learning_rate": 2.725227754051953e-05, | |
| "loss": 0.1816, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18410508334636688, | |
| "step": 1085, | |
| "valid_targets_mean": 1281.5, | |
| "valid_targets_min": 711 | |
| }, | |
| { | |
| "epoch": 2.678132678132678, | |
| "grad_norm": 0.8190757683782894, | |
| "learning_rate": 2.711883094668234e-05, | |
| "loss": 0.1741, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16918015480041504, | |
| "step": 1090, | |
| "valid_targets_mean": 1204.8, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 2.69041769041769, | |
| "grad_norm": 0.7763515026478695, | |
| "learning_rate": 2.698502044914633e-05, | |
| "loss": 0.173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18051014840602875, | |
| "step": 1095, | |
| "valid_targets_mean": 1259.6, | |
| "valid_targets_min": 814 | |
| }, | |
| { | |
| "epoch": 2.7027027027027026, | |
| "grad_norm": 0.8114087614186496, | |
| "learning_rate": 2.685085288809853e-05, | |
| "loss": 0.1812, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18935176730155945, | |
| "step": 1100, | |
| "valid_targets_mean": 1320.4, | |
| "valid_targets_min": 763 | |
| }, | |
| { | |
| "epoch": 2.714987714987715, | |
| "grad_norm": 0.7857431478660847, | |
| "learning_rate": 2.671633512197848e-05, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18005716800689697, | |
| "step": 1105, | |
| "valid_targets_mean": 1404.3, | |
| "valid_targets_min": 1088 | |
| }, | |
| { | |
| "epoch": 2.7272727272727275, | |
| "grad_norm": 0.7736276544807723, | |
| "learning_rate": 2.658147402712768e-05, | |
| "loss": 0.1784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1820794641971588, | |
| "step": 1110, | |
| "valid_targets_mean": 1390.3, | |
| "valid_targets_min": 968 | |
| }, | |
| { | |
| "epoch": 2.7395577395577395, | |
| "grad_norm": 0.9867019712115244, | |
| "learning_rate": 2.6446276497438064e-05, | |
| "loss": 0.1768, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18605422973632812, | |
| "step": 1115, | |
| "valid_targets_mean": 1098.2, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 2.751842751842752, | |
| "grad_norm": 0.84967200704623, | |
| "learning_rate": 2.6310749443999593e-05, | |
| "loss": 0.1841, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1918058693408966, | |
| "step": 1120, | |
| "valid_targets_mean": 1148.2, | |
| "valid_targets_min": 782 | |
| }, | |
| { | |
| "epoch": 2.764127764127764, | |
| "grad_norm": 0.8505940380680246, | |
| "learning_rate": 2.617489979474699e-05, | |
| "loss": 0.1784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18517403304576874, | |
| "step": 1125, | |
| "valid_targets_mean": 1168.7, | |
| "valid_targets_min": 620 | |
| }, | |
| { | |
| "epoch": 2.7764127764127764, | |
| "grad_norm": 0.7530757761401906, | |
| "learning_rate": 2.6038734494105562e-05, | |
| "loss": 0.1833, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1890183687210083, | |
| "step": 1130, | |
| "valid_targets_mean": 1517.5, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 2.788697788697789, | |
| "grad_norm": 0.8401579570867019, | |
| "learning_rate": 2.590226050263625e-05, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16918352246284485, | |
| "step": 1135, | |
| "valid_targets_mean": 1212.0, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 2.800982800982801, | |
| "grad_norm": 0.8601827378880036, | |
| "learning_rate": 2.5765484796679768e-05, | |
| "loss": 0.1774, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1912257969379425, | |
| "step": 1140, | |
| "valid_targets_mean": 1214.9, | |
| "valid_targets_min": 674 | |
| }, | |
| { | |
| "epoch": 2.8132678132678133, | |
| "grad_norm": 0.8033832037490392, | |
| "learning_rate": 2.5628414368000035e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17666852474212646, | |
| "step": 1145, | |
| "valid_targets_mean": 1377.9, | |
| "valid_targets_min": 784 | |
| }, | |
| { | |
| "epoch": 2.8255528255528253, | |
| "grad_norm": 0.7512530664621442, | |
| "learning_rate": 2.5491056223426746e-05, | |
| "loss": 0.184, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18542739748954773, | |
| "step": 1150, | |
| "valid_targets_mean": 1384.6, | |
| "valid_targets_min": 873 | |
| }, | |
| { | |
| "epoch": 2.8378378378378377, | |
| "grad_norm": 0.8782223426492971, | |
| "learning_rate": 2.5353417384497166e-05, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1851997822523117, | |
| "step": 1155, | |
| "valid_targets_mean": 1129.2, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 2.85012285012285, | |
| "grad_norm": 0.8625211224696377, | |
| "learning_rate": 2.5215504887097243e-05, | |
| "loss": 0.1826, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20872226357460022, | |
| "step": 1160, | |
| "valid_targets_mean": 1269.6, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 2.8624078624078626, | |
| "grad_norm": 0.7562673187162171, | |
| "learning_rate": 2.5077325781101918e-05, | |
| "loss": 0.1833, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17310750484466553, | |
| "step": 1165, | |
| "valid_targets_mean": 1283.8, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 2.8746928746928746, | |
| "grad_norm": 0.8744562203843028, | |
| "learning_rate": 2.493888713001476e-05, | |
| "loss": 0.1796, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18026913702487946, | |
| "step": 1170, | |
| "valid_targets_mean": 1212.1, | |
| "valid_targets_min": 691 | |
| }, | |
| { | |
| "epoch": 2.886977886977887, | |
| "grad_norm": 0.7979652681313569, | |
| "learning_rate": 2.480019601060687e-05, | |
| "loss": 0.1787, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19717931747436523, | |
| "step": 1175, | |
| "valid_targets_mean": 1335.1, | |
| "valid_targets_min": 830 | |
| }, | |
| { | |
| "epoch": 2.899262899262899, | |
| "grad_norm": 0.7494625305278567, | |
| "learning_rate": 2.4661259512555176e-05, | |
| "loss": 0.1813, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1821805238723755, | |
| "step": 1180, | |
| "valid_targets_mean": 1392.2, | |
| "valid_targets_min": 1080 | |
| }, | |
| { | |
| "epoch": 2.9115479115479115, | |
| "grad_norm": 0.8592457680594853, | |
| "learning_rate": 2.4522084738079933e-05, | |
| "loss": 0.1782, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19203343987464905, | |
| "step": 1185, | |
| "valid_targets_mean": 1296.5, | |
| "valid_targets_min": 773 | |
| }, | |
| { | |
| "epoch": 2.923832923832924, | |
| "grad_norm": 0.787142547780488, | |
| "learning_rate": 2.4382678801581762e-05, | |
| "loss": 0.1747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16595681011676788, | |
| "step": 1190, | |
| "valid_targets_mean": 1291.5, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 2.9361179361179364, | |
| "grad_norm": 0.7237730028951334, | |
| "learning_rate": 2.4243048829277916e-05, | |
| "loss": 0.1829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16514533758163452, | |
| "step": 1195, | |
| "valid_targets_mean": 1309.0, | |
| "valid_targets_min": 615 | |
| }, | |
| { | |
| "epoch": 2.9484029484029484, | |
| "grad_norm": 0.8443033469283165, | |
| "learning_rate": 2.410320195883802e-05, | |
| "loss": 0.172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18028348684310913, | |
| "step": 1200, | |
| "valid_targets_mean": 1211.9, | |
| "valid_targets_min": 576 | |
| }, | |
| { | |
| "epoch": 2.960687960687961, | |
| "grad_norm": 0.8259230182106898, | |
| "learning_rate": 2.396314533901918e-05, | |
| "loss": 0.1762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17320460081100464, | |
| "step": 1205, | |
| "valid_targets_mean": 1226.4, | |
| "valid_targets_min": 605 | |
| }, | |
| { | |
| "epoch": 2.972972972972973, | |
| "grad_norm": 0.7764557250016658, | |
| "learning_rate": 2.3822886129300603e-05, | |
| "loss": 0.18, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18336926400661469, | |
| "step": 1210, | |
| "valid_targets_mean": 1345.6, | |
| "valid_targets_min": 898 | |
| }, | |
| { | |
| "epoch": 2.9852579852579852, | |
| "grad_norm": 0.8960330072091445, | |
| "learning_rate": 2.368243149951755e-05, | |
| "loss": 0.1823, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18257874250411987, | |
| "step": 1215, | |
| "valid_targets_mean": 1147.5, | |
| "valid_targets_min": 623 | |
| }, | |
| { | |
| "epoch": 2.9975429975429977, | |
| "grad_norm": 0.7570835868850769, | |
| "learning_rate": 2.3541788629494865e-05, | |
| "loss": 0.1805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17076748609542847, | |
| "step": 1220, | |
| "valid_targets_mean": 1296.3, | |
| "valid_targets_min": 635 | |
| }, | |
| { | |
| "epoch": 3.0098280098280097, | |
| "grad_norm": 0.72326162291509, | |
| "learning_rate": 2.3400964708679944e-05, | |
| "loss": 0.1534, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13778720796108246, | |
| "step": 1225, | |
| "valid_targets_mean": 1254.6, | |
| "valid_targets_min": 651 | |
| }, | |
| { | |
| "epoch": 3.022113022113022, | |
| "grad_norm": 0.8572240220731201, | |
| "learning_rate": 2.325996693577522e-05, | |
| "loss": 0.1583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15546341240406036, | |
| "step": 1230, | |
| "valid_targets_mean": 1392.1, | |
| "valid_targets_min": 799 | |
| }, | |
| { | |
| "epoch": 3.0343980343980346, | |
| "grad_norm": 0.8406897333538058, | |
| "learning_rate": 2.311880251837019e-05, | |
| "loss": 0.1517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1490376889705658, | |
| "step": 1235, | |
| "valid_targets_mean": 1217.6, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 3.0466830466830466, | |
| "grad_norm": 0.8128667720747944, | |
| "learning_rate": 2.2977478672572933e-05, | |
| "loss": 0.1523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15579545497894287, | |
| "step": 1240, | |
| "valid_targets_mean": 1323.8, | |
| "valid_targets_min": 666 | |
| }, | |
| { | |
| "epoch": 3.058968058968059, | |
| "grad_norm": 0.8117066580399452, | |
| "learning_rate": 2.2836002622641297e-05, | |
| "loss": 0.1497, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16121074557304382, | |
| "step": 1245, | |
| "valid_targets_mean": 1316.7, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 3.0712530712530715, | |
| "grad_norm": 0.8662224455967267, | |
| "learning_rate": 2.269438160061354e-05, | |
| "loss": 0.1533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1444993019104004, | |
| "step": 1250, | |
| "valid_targets_mean": 1238.2, | |
| "valid_targets_min": 810 | |
| }, | |
| { | |
| "epoch": 3.0835380835380835, | |
| "grad_norm": 0.8817439219548429, | |
| "learning_rate": 2.2552622845938698e-05, | |
| "loss": 0.1533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15430381894111633, | |
| "step": 1255, | |
| "valid_targets_mean": 1404.1, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 3.095823095823096, | |
| "grad_norm": 0.8477302153623169, | |
| "learning_rate": 2.2410733605106462e-05, | |
| "loss": 0.1564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1634480059146881, | |
| "step": 1260, | |
| "valid_targets_mean": 1147.9, | |
| "valid_targets_min": 604 | |
| }, | |
| { | |
| "epoch": 3.108108108108108, | |
| "grad_norm": 0.8243597226463298, | |
| "learning_rate": 2.2268721131276805e-05, | |
| "loss": 0.1516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15756012499332428, | |
| "step": 1265, | |
| "valid_targets_mean": 1393.2, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 3.1203931203931203, | |
| "grad_norm": 0.8937269373033379, | |
| "learning_rate": 2.2126592683909154e-05, | |
| "loss": 0.15, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15642580389976501, | |
| "step": 1270, | |
| "valid_targets_mean": 1184.4, | |
| "valid_targets_min": 614 | |
| }, | |
| { | |
| "epoch": 3.1326781326781328, | |
| "grad_norm": 0.8160928913633115, | |
| "learning_rate": 2.1984355528391342e-05, | |
| "loss": 0.1519, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15530337393283844, | |
| "step": 1275, | |
| "valid_targets_mean": 1350.1, | |
| "valid_targets_min": 758 | |
| }, | |
| { | |
| "epoch": 3.1449631449631448, | |
| "grad_norm": 0.8916891912210628, | |
| "learning_rate": 2.1842016935668188e-05, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17001807689666748, | |
| "step": 1280, | |
| "valid_targets_mean": 1380.8, | |
| "valid_targets_min": 746 | |
| }, | |
| { | |
| "epoch": 3.157248157248157, | |
| "grad_norm": 0.8639371441585546, | |
| "learning_rate": 2.169958418186982e-05, | |
| "loss": 0.1606, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15526992082595825, | |
| "step": 1285, | |
| "valid_targets_mean": 1277.1, | |
| "valid_targets_min": 738 | |
| }, | |
| { | |
| "epoch": 3.1695331695331697, | |
| "grad_norm": 0.8130627821075446, | |
| "learning_rate": 2.1557064547939754e-05, | |
| "loss": 0.1556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14274147152900696, | |
| "step": 1290, | |
| "valid_targets_mean": 1234.9, | |
| "valid_targets_min": 810 | |
| }, | |
| { | |
| "epoch": 3.1818181818181817, | |
| "grad_norm": 0.7746615359719721, | |
| "learning_rate": 2.1414465319262666e-05, | |
| "loss": 0.1547, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14222201704978943, | |
| "step": 1295, | |
| "valid_targets_mean": 1286.1, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 3.194103194103194, | |
| "grad_norm": 0.7520776877457452, | |
| "learning_rate": 2.1271793785291997e-05, | |
| "loss": 0.1498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1390017867088318, | |
| "step": 1300, | |
| "valid_targets_mean": 1272.8, | |
| "valid_targets_min": 602 | |
| }, | |
| { | |
| "epoch": 3.2063882063882065, | |
| "grad_norm": 0.8462893075712905, | |
| "learning_rate": 2.1129057239177337e-05, | |
| "loss": 0.1551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15616267919540405, | |
| "step": 1305, | |
| "valid_targets_mean": 1242.9, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 3.2186732186732185, | |
| "grad_norm": 0.8539575906844962, | |
| "learning_rate": 2.0986262977391577e-05, | |
| "loss": 0.1541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1531214714050293, | |
| "step": 1310, | |
| "valid_targets_mean": 1199.8, | |
| "valid_targets_min": 728 | |
| }, | |
| { | |
| "epoch": 3.230958230958231, | |
| "grad_norm": 0.8428538051981783, | |
| "learning_rate": 2.084341829935796e-05, | |
| "loss": 0.1521, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1577906310558319, | |
| "step": 1315, | |
| "valid_targets_mean": 1258.8, | |
| "valid_targets_min": 664 | |
| }, | |
| { | |
| "epoch": 3.2432432432432434, | |
| "grad_norm": 0.84603528658779, | |
| "learning_rate": 2.0700530507076916e-05, | |
| "loss": 0.1472, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14432299137115479, | |
| "step": 1320, | |
| "valid_targets_mean": 1210.4, | |
| "valid_targets_min": 807 | |
| }, | |
| { | |
| "epoch": 3.2555282555282554, | |
| "grad_norm": 0.9102137746252226, | |
| "learning_rate": 2.0557606904752833e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15487079322338104, | |
| "step": 1325, | |
| "valid_targets_mean": 1167.1, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 3.267813267813268, | |
| "grad_norm": 0.8050078740673396, | |
| "learning_rate": 2.0414654798420622e-05, | |
| "loss": 0.1443, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15707707405090332, | |
| "step": 1330, | |
| "valid_targets_mean": 1354.5, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 3.2800982800982803, | |
| "grad_norm": 0.8353880109503813, | |
| "learning_rate": 2.02716814955723e-05, | |
| "loss": 0.1543, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14280740916728973, | |
| "step": 1335, | |
| "valid_targets_mean": 1208.4, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 3.2923832923832923, | |
| "grad_norm": 0.8659682584614259, | |
| "learning_rate": 2.0128694304783406e-05, | |
| "loss": 0.1551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15830692648887634, | |
| "step": 1340, | |
| "valid_targets_mean": 1201.9, | |
| "valid_targets_min": 730 | |
| }, | |
| { | |
| "epoch": 3.3046683046683047, | |
| "grad_norm": 0.8379988002065017, | |
| "learning_rate": 1.9985700535339406e-05, | |
| "loss": 0.1569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1467028260231018, | |
| "step": 1345, | |
| "valid_targets_mean": 1127.9, | |
| "valid_targets_min": 845 | |
| }, | |
| { | |
| "epoch": 3.3169533169533167, | |
| "grad_norm": 1.1917573647951119, | |
| "learning_rate": 1.984270749686207e-05, | |
| "loss": 0.1543, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13254572451114655, | |
| "step": 1350, | |
| "valid_targets_mean": 1263.1, | |
| "valid_targets_min": 536 | |
| }, | |
| { | |
| "epoch": 3.329238329238329, | |
| "grad_norm": 0.8655024458620978, | |
| "learning_rate": 1.9699722498935786e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15300899744033813, | |
| "step": 1355, | |
| "valid_targets_mean": 1207.4, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 3.3415233415233416, | |
| "grad_norm": 0.7847387083906133, | |
| "learning_rate": 1.9556752850733933e-05, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13689152896404266, | |
| "step": 1360, | |
| "valid_targets_mean": 1255.2, | |
| "valid_targets_min": 706 | |
| }, | |
| { | |
| "epoch": 3.3538083538083536, | |
| "grad_norm": 0.8693425075316302, | |
| "learning_rate": 1.9413805860645242e-05, | |
| "loss": 0.1609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17255070805549622, | |
| "step": 1365, | |
| "valid_targets_mean": 1268.3, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 3.366093366093366, | |
| "grad_norm": 0.8576356082893154, | |
| "learning_rate": 1.9270888835900165e-05, | |
| "loss": 0.1616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1623215675354004, | |
| "step": 1370, | |
| "valid_targets_mean": 1256.3, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 3.3783783783783785, | |
| "grad_norm": 0.8842690706850099, | |
| "learning_rate": 1.9128009082197417e-05, | |
| "loss": 0.1504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14834919571876526, | |
| "step": 1375, | |
| "valid_targets_mean": 1141.1, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 3.3906633906633905, | |
| "grad_norm": 0.803567248692513, | |
| "learning_rate": 1.8985173903330428e-05, | |
| "loss": 0.1521, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15296560525894165, | |
| "step": 1380, | |
| "valid_targets_mean": 1291.1, | |
| "valid_targets_min": 494 | |
| }, | |
| { | |
| "epoch": 3.402948402948403, | |
| "grad_norm": 0.8228636441386455, | |
| "learning_rate": 1.884239060081407e-05, | |
| "loss": 0.1442, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14132916927337646, | |
| "step": 1385, | |
| "valid_targets_mean": 1182.8, | |
| "valid_targets_min": 808 | |
| }, | |
| { | |
| "epoch": 3.4152334152334154, | |
| "grad_norm": 0.8389934209523321, | |
| "learning_rate": 1.869966647351135e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1495848149061203, | |
| "step": 1390, | |
| "valid_targets_mean": 1253.1, | |
| "valid_targets_min": 667 | |
| }, | |
| { | |
| "epoch": 3.4275184275184274, | |
| "grad_norm": 0.7601245259644186, | |
| "learning_rate": 1.8557008817260343e-05, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14485347270965576, | |
| "step": 1395, | |
| "valid_targets_mean": 1379.3, | |
| "valid_targets_min": 871 | |
| }, | |
| { | |
| "epoch": 3.43980343980344, | |
| "grad_norm": 0.8713323917939153, | |
| "learning_rate": 1.8414424924501222e-05, | |
| "loss": 0.1596, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1464325487613678, | |
| "step": 1400, | |
| "valid_targets_mean": 1344.7, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 3.4520884520884523, | |
| "grad_norm": 0.8116825428322517, | |
| "learning_rate": 1.827192208390347e-05, | |
| "loss": 0.1482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14812307059764862, | |
| "step": 1405, | |
| "valid_targets_mean": 1292.6, | |
| "valid_targets_min": 532 | |
| }, | |
| { | |
| "epoch": 3.4643734643734643, | |
| "grad_norm": 0.830495465611543, | |
| "learning_rate": 1.812950757999334e-05, | |
| "loss": 0.1553, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14193110167980194, | |
| "step": 1410, | |
| "valid_targets_mean": 1182.0, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 3.4766584766584767, | |
| "grad_norm": 0.8420966729919014, | |
| "learning_rate": 1.7987188692781417e-05, | |
| "loss": 0.1494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15079256892204285, | |
| "step": 1415, | |
| "valid_targets_mean": 1302.2, | |
| "valid_targets_min": 828 | |
| }, | |
| { | |
| "epoch": 3.488943488943489, | |
| "grad_norm": 0.805934689519132, | |
| "learning_rate": 1.784497269739052e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14082270860671997, | |
| "step": 1420, | |
| "valid_targets_mean": 1309.8, | |
| "valid_targets_min": 477 | |
| }, | |
| { | |
| "epoch": 3.501228501228501, | |
| "grad_norm": 0.9110728170126068, | |
| "learning_rate": 1.770286686368381e-05, | |
| "loss": 0.1516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15163275599479675, | |
| "step": 1425, | |
| "valid_targets_mean": 1216.3, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 3.5135135135135136, | |
| "grad_norm": 1.010011400261342, | |
| "learning_rate": 1.756087845589312e-05, | |
| "loss": 0.1557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1431485116481781, | |
| "step": 1430, | |
| "valid_targets_mean": 1191.9, | |
| "valid_targets_min": 599 | |
| }, | |
| { | |
| "epoch": 3.5257985257985256, | |
| "grad_norm": 0.8232817829751108, | |
| "learning_rate": 1.7419014732247683e-05, | |
| "loss": 0.1573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16463416814804077, | |
| "step": 1435, | |
| "valid_targets_mean": 1286.1, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 3.538083538083538, | |
| "grad_norm": 0.9408275175657097, | |
| "learning_rate": 1.7277282944603047e-05, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1658443808555603, | |
| "step": 1440, | |
| "valid_targets_mean": 1391.1, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 3.5503685503685505, | |
| "grad_norm": 0.8964394826069799, | |
| "learning_rate": 1.713569033807041e-05, | |
| "loss": 0.1535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14551150798797607, | |
| "step": 1445, | |
| "valid_targets_mean": 1158.6, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 3.562653562653563, | |
| "grad_norm": 0.6983703036178323, | |
| "learning_rate": 1.6994244150646244e-05, | |
| "loss": 0.1494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13444580137729645, | |
| "step": 1450, | |
| "valid_targets_mean": 1367.5, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 3.574938574938575, | |
| "grad_norm": 0.8407855929126276, | |
| "learning_rate": 1.6852951612842278e-05, | |
| "loss": 0.1542, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15283367037773132, | |
| "step": 1455, | |
| "valid_targets_mean": 1407.4, | |
| "valid_targets_min": 922 | |
| }, | |
| { | |
| "epoch": 3.5872235872235874, | |
| "grad_norm": 0.8242889096179601, | |
| "learning_rate": 1.671181994731595e-05, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.146068274974823, | |
| "step": 1460, | |
| "valid_targets_mean": 1294.8, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 3.5995085995085994, | |
| "grad_norm": 0.8648972544384124, | |
| "learning_rate": 1.6570856368501108e-05, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17492206394672394, | |
| "step": 1465, | |
| "valid_targets_mean": 1290.1, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 3.611793611793612, | |
| "grad_norm": 0.8457602915141741, | |
| "learning_rate": 1.643006808223931e-05, | |
| "loss": 0.1505, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14468783140182495, | |
| "step": 1470, | |
| "valid_targets_mean": 1287.5, | |
| "valid_targets_min": 576 | |
| }, | |
| { | |
| "epoch": 3.6240786240786242, | |
| "grad_norm": 0.8532991002467062, | |
| "learning_rate": 1.6289462285411387e-05, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1533271074295044, | |
| "step": 1475, | |
| "valid_targets_mean": 1243.2, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 3.6363636363636362, | |
| "grad_norm": 0.8803826617627987, | |
| "learning_rate": 1.614904616556962e-05, | |
| "loss": 0.1445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14523166418075562, | |
| "step": 1480, | |
| "valid_targets_mean": 1301.2, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 3.6486486486486487, | |
| "grad_norm": 0.8542006788844356, | |
| "learning_rate": 1.6008826900570294e-05, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14371222257614136, | |
| "step": 1485, | |
| "valid_targets_mean": 1139.2, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 3.6609336609336607, | |
| "grad_norm": 0.8745016942367656, | |
| "learning_rate": 1.586881165820675e-05, | |
| "loss": 0.1548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.154458150267601, | |
| "step": 1490, | |
| "valid_targets_mean": 1155.7, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 3.673218673218673, | |
| "grad_norm": 0.8703992302998022, | |
| "learning_rate": 1.5729007595843037e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14299608767032623, | |
| "step": 1495, | |
| "valid_targets_mean": 1268.3, | |
| "valid_targets_min": 776 | |
| }, | |
| { | |
| "epoch": 3.6855036855036856, | |
| "grad_norm": 0.8095249679573774, | |
| "learning_rate": 1.5589421860047986e-05, | |
| "loss": 0.1544, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15553806722164154, | |
| "step": 1500, | |
| "valid_targets_mean": 1325.9, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 3.697788697788698, | |
| "grad_norm": 0.8758318030682021, | |
| "learning_rate": 1.5450061586229903e-05, | |
| "loss": 0.1552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15329515933990479, | |
| "step": 1505, | |
| "valid_targets_mean": 1132.9, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 3.71007371007371, | |
| "grad_norm": 0.8363686738335151, | |
| "learning_rate": 1.5310933898271864e-05, | |
| "loss": 0.1515, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15566831827163696, | |
| "step": 1510, | |
| "valid_targets_mean": 1268.0, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 3.7223587223587224, | |
| "grad_norm": 0.9279177914708184, | |
| "learning_rate": 1.5172045908167462e-05, | |
| "loss": 0.149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14874522387981415, | |
| "step": 1515, | |
| "valid_targets_mean": 1278.6, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 3.7346437346437344, | |
| "grad_norm": 0.8889669951596095, | |
| "learning_rate": 1.5033404715657344e-05, | |
| "loss": 0.152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1446603238582611, | |
| "step": 1520, | |
| "valid_targets_mean": 1261.3, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 3.746928746928747, | |
| "grad_norm": 0.9294511912472954, | |
| "learning_rate": 1.4895017407866217e-05, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15708409249782562, | |
| "step": 1525, | |
| "valid_targets_mean": 1087.5, | |
| "valid_targets_min": 578 | |
| }, | |
| { | |
| "epoch": 3.7592137592137593, | |
| "grad_norm": 0.7974453732071705, | |
| "learning_rate": 1.4756891058940606e-05, | |
| "loss": 0.1559, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16070222854614258, | |
| "step": 1530, | |
| "valid_targets_mean": 1359.7, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 3.7714987714987718, | |
| "grad_norm": 0.7840160517261555, | |
| "learning_rate": 1.4619032729687223e-05, | |
| "loss": 0.1519, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15777051448822021, | |
| "step": 1535, | |
| "valid_targets_mean": 1339.7, | |
| "valid_targets_min": 703 | |
| }, | |
| { | |
| "epoch": 3.7837837837837838, | |
| "grad_norm": 0.8550911396739045, | |
| "learning_rate": 1.4481449467212004e-05, | |
| "loss": 0.1497, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14654958248138428, | |
| "step": 1540, | |
| "valid_targets_mean": 1197.9, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 3.796068796068796, | |
| "grad_norm": 1.0416276131378706, | |
| "learning_rate": 1.4344148304559926e-05, | |
| "loss": 0.1508, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1543717384338379, | |
| "step": 1545, | |
| "valid_targets_mean": 1387.0, | |
| "valid_targets_min": 930 | |
| }, | |
| { | |
| "epoch": 3.808353808353808, | |
| "grad_norm": 0.7971903127956029, | |
| "learning_rate": 1.4207136260355426e-05, | |
| "loss": 0.1501, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15580767393112183, | |
| "step": 1550, | |
| "valid_targets_mean": 1369.1, | |
| "valid_targets_min": 844 | |
| }, | |
| { | |
| "epoch": 3.8206388206388207, | |
| "grad_norm": 0.8302888094156373, | |
| "learning_rate": 1.4070420338443667e-05, | |
| "loss": 0.1558, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15264888107776642, | |
| "step": 1555, | |
| "valid_targets_mean": 1372.4, | |
| "valid_targets_min": 652 | |
| }, | |
| { | |
| "epoch": 3.832923832923833, | |
| "grad_norm": 0.8113952097335243, | |
| "learning_rate": 1.3934007527532494e-05, | |
| "loss": 0.1541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15965667366981506, | |
| "step": 1560, | |
| "valid_targets_mean": 1547.5, | |
| "valid_targets_min": 1012 | |
| }, | |
| { | |
| "epoch": 3.845208845208845, | |
| "grad_norm": 0.8016515224809283, | |
| "learning_rate": 1.3797904800835174e-05, | |
| "loss": 0.1517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14814619719982147, | |
| "step": 1565, | |
| "valid_targets_mean": 1327.5, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 3.8574938574938575, | |
| "grad_norm": 0.8101546175320481, | |
| "learning_rate": 1.3662119115713968e-05, | |
| "loss": 0.1543, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13531778752803802, | |
| "step": 1570, | |
| "valid_targets_mean": 1317.8, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 3.8697788697788695, | |
| "grad_norm": 0.8752081945344778, | |
| "learning_rate": 1.3526657413324427e-05, | |
| "loss": 0.1483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1407470703125, | |
| "step": 1575, | |
| "valid_targets_mean": 1331.0, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 3.882063882063882, | |
| "grad_norm": 0.8025227410206387, | |
| "learning_rate": 1.3391526618260636e-05, | |
| "loss": 0.1509, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15067920088768005, | |
| "step": 1580, | |
| "valid_targets_mean": 1262.4, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 3.8943488943488944, | |
| "grad_norm": 0.8935145822922234, | |
| "learning_rate": 1.3256733638201172e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15884637832641602, | |
| "step": 1585, | |
| "valid_targets_mean": 1158.1, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 3.906633906633907, | |
| "grad_norm": 0.7708791703581477, | |
| "learning_rate": 1.3122285363556053e-05, | |
| "loss": 0.15, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14581814408302307, | |
| "step": 1590, | |
| "valid_targets_mean": 1297.9, | |
| "valid_targets_min": 552 | |
| }, | |
| { | |
| "epoch": 3.918918918918919, | |
| "grad_norm": 0.8888029980587986, | |
| "learning_rate": 1.2988188667114487e-05, | |
| "loss": 0.1509, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1706969141960144, | |
| "step": 1595, | |
| "valid_targets_mean": 1370.7, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 3.9312039312039313, | |
| "grad_norm": 0.8292708834052593, | |
| "learning_rate": 1.2854450403693526e-05, | |
| "loss": 0.154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15968310832977295, | |
| "step": 1600, | |
| "valid_targets_mean": 1313.7, | |
| "valid_targets_min": 670 | |
| }, | |
| { | |
| "epoch": 3.9434889434889433, | |
| "grad_norm": 0.788269813578671, | |
| "learning_rate": 1.272107740978769e-05, | |
| "loss": 0.1535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1440616101026535, | |
| "step": 1605, | |
| "valid_targets_mean": 1278.4, | |
| "valid_targets_min": 612 | |
| }, | |
| { | |
| "epoch": 3.9557739557739557, | |
| "grad_norm": 0.8037047070644282, | |
| "learning_rate": 1.2588076503219475e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1564336121082306, | |
| "step": 1610, | |
| "valid_targets_mean": 1324.2, | |
| "valid_targets_min": 792 | |
| }, | |
| { | |
| "epoch": 3.968058968058968, | |
| "grad_norm": 0.7845890150943812, | |
| "learning_rate": 1.2455454482790859e-05, | |
| "loss": 0.1565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1552063524723053, | |
| "step": 1615, | |
| "valid_targets_mean": 1362.6, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 3.98034398034398, | |
| "grad_norm": 0.7747181055399516, | |
| "learning_rate": 1.2323218127935714e-05, | |
| "loss": 0.1481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1518639624118805, | |
| "step": 1620, | |
| "valid_targets_mean": 1321.4, | |
| "valid_targets_min": 706 | |
| }, | |
| { | |
| "epoch": 3.9926289926289926, | |
| "grad_norm": 0.8558446375986389, | |
| "learning_rate": 1.2191374198373309e-05, | |
| "loss": 0.1523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1543593853712082, | |
| "step": 1625, | |
| "valid_targets_mean": 1155.8, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 4.004914004914005, | |
| "grad_norm": 0.781627123143895, | |
| "learning_rate": 1.2059929433762734e-05, | |
| "loss": 0.1497, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13371559977531433, | |
| "step": 1630, | |
| "valid_targets_mean": 1401.8, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 4.017199017199017, | |
| "grad_norm": 0.7723114766334374, | |
| "learning_rate": 1.1928890553358352e-05, | |
| "loss": 0.1353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.135203018784523, | |
| "step": 1635, | |
| "valid_targets_mean": 1515.5, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 4.0294840294840295, | |
| "grad_norm": 0.8735823383942167, | |
| "learning_rate": 1.1798264255666387e-05, | |
| "loss": 0.1315, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14232441782951355, | |
| "step": 1640, | |
| "valid_targets_mean": 1276.2, | |
| "valid_targets_min": 602 | |
| }, | |
| { | |
| "epoch": 4.041769041769042, | |
| "grad_norm": 0.8192596996969622, | |
| "learning_rate": 1.1668057218102436e-05, | |
| "loss": 0.1286, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11941641569137573, | |
| "step": 1645, | |
| "valid_targets_mean": 1408.4, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 4.054054054054054, | |
| "grad_norm": 0.8924526222951878, | |
| "learning_rate": 1.1538276096650175e-05, | |
| "loss": 0.1298, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1319366693496704, | |
| "step": 1650, | |
| "valid_targets_mean": 1181.8, | |
| "valid_targets_min": 510 | |
| }, | |
| { | |
| "epoch": 4.066339066339066, | |
| "grad_norm": 0.7699089718854191, | |
| "learning_rate": 1.1408927525521118e-05, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1287621706724167, | |
| "step": 1655, | |
| "valid_targets_mean": 1414.1, | |
| "valid_targets_min": 768 | |
| }, | |
| { | |
| "epoch": 4.078624078624078, | |
| "grad_norm": 0.8099063799012394, | |
| "learning_rate": 1.1280018116815438e-05, | |
| "loss": 0.1308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1263992041349411, | |
| "step": 1660, | |
| "valid_targets_mean": 1322.7, | |
| "valid_targets_min": 675 | |
| }, | |
| { | |
| "epoch": 4.090909090909091, | |
| "grad_norm": 0.8792757959344022, | |
| "learning_rate": 1.115155446018404e-05, | |
| "loss": 0.1297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12545448541641235, | |
| "step": 1665, | |
| "valid_targets_mean": 1129.1, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 4.103194103194103, | |
| "grad_norm": 0.8012256945891334, | |
| "learning_rate": 1.1023543122491626e-05, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13978534936904907, | |
| "step": 1670, | |
| "valid_targets_mean": 1432.0, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 4.115479115479116, | |
| "grad_norm": 0.8588845221042121, | |
| "learning_rate": 1.089599064748108e-05, | |
| "loss": 0.1319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12195970118045807, | |
| "step": 1675, | |
| "valid_targets_mean": 1170.1, | |
| "valid_targets_min": 829 | |
| }, | |
| { | |
| "epoch": 4.127764127764128, | |
| "grad_norm": 0.8611293324266706, | |
| "learning_rate": 1.0768903555438927e-05, | |
| "loss": 0.1316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1330062747001648, | |
| "step": 1680, | |
| "valid_targets_mean": 1203.4, | |
| "valid_targets_min": 866 | |
| }, | |
| { | |
| "epoch": 4.14004914004914, | |
| "grad_norm": 0.8547247627858316, | |
| "learning_rate": 1.0642288342862007e-05, | |
| "loss": 0.1279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1295037716627121, | |
| "step": 1685, | |
| "valid_targets_mean": 1327.2, | |
| "valid_targets_min": 821 | |
| }, | |
| { | |
| "epoch": 4.152334152334152, | |
| "grad_norm": 0.9284978728629667, | |
| "learning_rate": 1.051615148212544e-05, | |
| "loss": 0.1302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13765308260917664, | |
| "step": 1690, | |
| "valid_targets_mean": 1221.8, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 4.164619164619165, | |
| "grad_norm": 0.8420408633139839, | |
| "learning_rate": 1.0390499421151706e-05, | |
| "loss": 0.1295, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12419085204601288, | |
| "step": 1695, | |
| "valid_targets_mean": 1425.6, | |
| "valid_targets_min": 838 | |
| }, | |
| { | |
| "epoch": 4.176904176904177, | |
| "grad_norm": 0.9003128577554333, | |
| "learning_rate": 1.0265338583081088e-05, | |
| "loss": 0.1287, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1306770145893097, | |
| "step": 1700, | |
| "valid_targets_mean": 1178.5, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 4.1891891891891895, | |
| "grad_norm": 0.9195424473786398, | |
| "learning_rate": 1.0140675365943284e-05, | |
| "loss": 0.1316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13825950026512146, | |
| "step": 1705, | |
| "valid_targets_mean": 1205.4, | |
| "valid_targets_min": 740 | |
| }, | |
| { | |
| "epoch": 4.201474201474202, | |
| "grad_norm": 0.8994495082648482, | |
| "learning_rate": 1.0016516142330404e-05, | |
| "loss": 0.1336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13527360558509827, | |
| "step": 1710, | |
| "valid_targets_mean": 1221.3, | |
| "valid_targets_min": 817 | |
| }, | |
| { | |
| "epoch": 4.2137592137592135, | |
| "grad_norm": 0.8912252331597237, | |
| "learning_rate": 9.89286725907117e-06, | |
| "loss": 0.1296, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13443750143051147, | |
| "step": 1715, | |
| "valid_targets_mean": 1249.4, | |
| "valid_targets_min": 764 | |
| }, | |
| { | |
| "epoch": 4.226044226044226, | |
| "grad_norm": 0.8920482099817841, | |
| "learning_rate": 9.769735036906475e-06, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1290460228919983, | |
| "step": 1720, | |
| "valid_targets_mean": 1142.9, | |
| "valid_targets_min": 686 | |
| }, | |
| { | |
| "epoch": 4.238329238329238, | |
| "grad_norm": 0.8527908969918075, | |
| "learning_rate": 9.647125770166321e-06, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13044829666614532, | |
| "step": 1725, | |
| "valid_targets_mean": 1256.8, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 4.250614250614251, | |
| "grad_norm": 0.8841584686314972, | |
| "learning_rate": 9.525045726448001e-06, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13348722457885742, | |
| "step": 1730, | |
| "valid_targets_mean": 1273.6, | |
| "valid_targets_min": 708 | |
| }, | |
| { | |
| "epoch": 4.262899262899263, | |
| "grad_norm": 0.8702023826857563, | |
| "learning_rate": 9.40350114629577e-06, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13443559408187866, | |
| "step": 1735, | |
| "valid_targets_mean": 1183.1, | |
| "valid_targets_min": 702 | |
| }, | |
| { | |
| "epoch": 4.275184275184275, | |
| "grad_norm": 0.8992471857795535, | |
| "learning_rate": 9.282498242881784e-06, | |
| "loss": 0.1322, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13517756760120392, | |
| "step": 1740, | |
| "valid_targets_mean": 1307.1, | |
| "valid_targets_min": 687 | |
| }, | |
| { | |
| "epoch": 4.287469287469287, | |
| "grad_norm": 0.796244329414364, | |
| "learning_rate": 9.162043201688517e-06, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12337645888328552, | |
| "step": 1745, | |
| "valid_targets_mean": 1191.1, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 4.2997542997543, | |
| "grad_norm": 0.8370509126157433, | |
| "learning_rate": 9.042142180192596e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13218200206756592, | |
| "step": 1750, | |
| "valid_targets_mean": 1195.6, | |
| "valid_targets_min": 750 | |
| }, | |
| { | |
| "epoch": 4.312039312039312, | |
| "grad_norm": 0.8895263747387979, | |
| "learning_rate": 8.92280130754998e-06, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1353050172328949, | |
| "step": 1755, | |
| "valid_targets_mean": 1278.0, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 4.324324324324325, | |
| "grad_norm": 0.8271582266915379, | |
| "learning_rate": 8.804026684282694e-06, | |
| "loss": 0.1339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13450637459754944, | |
| "step": 1760, | |
| "valid_targets_mean": 1315.7, | |
| "valid_targets_min": 723 | |
| }, | |
| { | |
| "epoch": 4.336609336609337, | |
| "grad_norm": 0.9520857741575197, | |
| "learning_rate": 8.685824381966975e-06, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13987067341804504, | |
| "step": 1765, | |
| "valid_targets_mean": 1201.6, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 4.348894348894349, | |
| "grad_norm": 0.8757634606289378, | |
| "learning_rate": 8.568200442922865e-06, | |
| "loss": 0.1325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13092288374900818, | |
| "step": 1770, | |
| "valid_targets_mean": 1367.1, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 4.361179361179361, | |
| "grad_norm": 0.8821249974469216, | |
| "learning_rate": 8.451160879905398e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1396908462047577, | |
| "step": 1775, | |
| "valid_targets_mean": 1268.4, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 4.3734643734643734, | |
| "grad_norm": 0.8236753936558183, | |
| "learning_rate": 8.33471167579717e-06, | |
| "loss": 0.1344, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13866353034973145, | |
| "step": 1780, | |
| "valid_targets_mean": 1363.2, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 4.385749385749386, | |
| "grad_norm": 0.8390859558509741, | |
| "learning_rate": 8.218858783302566e-06, | |
| "loss": 0.1374, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13545829057693481, | |
| "step": 1785, | |
| "valid_targets_mean": 1287.6, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 4.398034398034398, | |
| "grad_norm": 0.8518188407674924, | |
| "learning_rate": 8.103608124643412e-06, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1234915554523468, | |
| "step": 1790, | |
| "valid_targets_mean": 1293.2, | |
| "valid_targets_min": 745 | |
| }, | |
| { | |
| "epoch": 4.41031941031941, | |
| "grad_norm": 0.890765851149973, | |
| "learning_rate": 7.988965591256284e-06, | |
| "loss": 0.1386, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1416628658771515, | |
| "step": 1795, | |
| "valid_targets_mean": 1274.9, | |
| "valid_targets_min": 677 | |
| }, | |
| { | |
| "epoch": 4.422604422604422, | |
| "grad_norm": 0.8536233960569621, | |
| "learning_rate": 7.874937043491331e-06, | |
| "loss": 0.1358, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13207730650901794, | |
| "step": 1800, | |
| "valid_targets_mean": 1314.8, | |
| "valid_targets_min": 654 | |
| }, | |
| { | |
| "epoch": 4.434889434889435, | |
| "grad_norm": 0.8547910171850943, | |
| "learning_rate": 7.761528310312679e-06, | |
| "loss": 0.1339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1310330182313919, | |
| "step": 1805, | |
| "valid_targets_mean": 1319.5, | |
| "valid_targets_min": 658 | |
| }, | |
| { | |
| "epoch": 4.447174447174447, | |
| "grad_norm": 0.8440973307304958, | |
| "learning_rate": 7.648745189000511e-06, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12474372982978821, | |
| "step": 1810, | |
| "valid_targets_mean": 1266.9, | |
| "valid_targets_min": 607 | |
| }, | |
| { | |
| "epoch": 4.45945945945946, | |
| "grad_norm": 0.8526523335996494, | |
| "learning_rate": 7.536593444854663e-06, | |
| "loss": 0.132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13603094220161438, | |
| "step": 1815, | |
| "valid_targets_mean": 1392.1, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 4.471744471744472, | |
| "grad_norm": 0.9451445141228779, | |
| "learning_rate": 7.4250788108999686e-06, | |
| "loss": 0.1332, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13309922814369202, | |
| "step": 1820, | |
| "valid_targets_mean": 1134.2, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 4.484029484029484, | |
| "grad_norm": 0.8373347844464977, | |
| "learning_rate": 7.314206987593162e-06, | |
| "loss": 0.1388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13399043679237366, | |
| "step": 1825, | |
| "valid_targets_mean": 1223.4, | |
| "valid_targets_min": 776 | |
| }, | |
| { | |
| "epoch": 4.496314496314496, | |
| "grad_norm": 0.8625925922345389, | |
| "learning_rate": 7.203983642531462e-06, | |
| "loss": 0.1352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13259084522724152, | |
| "step": 1830, | |
| "valid_targets_mean": 1311.8, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 4.5085995085995085, | |
| "grad_norm": 0.8870716782947434, | |
| "learning_rate": 7.094414410162913e-06, | |
| "loss": 0.1371, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1382933109998703, | |
| "step": 1835, | |
| "valid_targets_mean": 1195.8, | |
| "valid_targets_min": 689 | |
| }, | |
| { | |
| "epoch": 4.520884520884521, | |
| "grad_norm": 0.7998309016979608, | |
| "learning_rate": 6.985504891498291e-06, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11957763135433197, | |
| "step": 1840, | |
| "valid_targets_mean": 1288.3, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 4.533169533169533, | |
| "grad_norm": 0.874780743207264, | |
| "learning_rate": 6.8772606538248285e-06, | |
| "loss": 0.1335, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1352095603942871, | |
| "step": 1845, | |
| "valid_targets_mean": 1310.2, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 4.545454545454545, | |
| "grad_norm": 0.862924674798308, | |
| "learning_rate": 6.769687230421638e-06, | |
| "loss": 0.1332, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1378445029258728, | |
| "step": 1850, | |
| "valid_targets_mean": 1357.0, | |
| "valid_targets_min": 637 | |
| }, | |
| { | |
| "epoch": 4.557739557739557, | |
| "grad_norm": 0.8663590326432032, | |
| "learning_rate": 6.662790120276803e-06, | |
| "loss": 0.1354, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13520276546478271, | |
| "step": 1855, | |
| "valid_targets_mean": 1397.5, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 4.57002457002457, | |
| "grad_norm": 0.9327490817540963, | |
| "learning_rate": 6.556574787806344e-06, | |
| "loss": 0.1325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13693825900554657, | |
| "step": 1860, | |
| "valid_targets_mean": 1106.2, | |
| "valid_targets_min": 578 | |
| }, | |
| { | |
| "epoch": 4.582309582309582, | |
| "grad_norm": 0.9638863910492748, | |
| "learning_rate": 6.451046662574831e-06, | |
| "loss": 0.1362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13662275671958923, | |
| "step": 1865, | |
| "valid_targets_mean": 1150.8, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 4.594594594594595, | |
| "grad_norm": 0.8426325810722405, | |
| "learning_rate": 6.346211139017877e-06, | |
| "loss": 0.135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1304231733083725, | |
| "step": 1870, | |
| "valid_targets_mean": 1320.3, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 4.606879606879607, | |
| "grad_norm": 0.9787706797182959, | |
| "learning_rate": 6.242073576166337e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13047711551189423, | |
| "step": 1875, | |
| "valid_targets_mean": 1202.9, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 4.61916461916462, | |
| "grad_norm": 0.9005324833030341, | |
| "learning_rate": 6.138639297372404e-06, | |
| "loss": 0.1273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13522884249687195, | |
| "step": 1880, | |
| "valid_targets_mean": 1393.9, | |
| "valid_targets_min": 873 | |
| }, | |
| { | |
| "epoch": 4.631449631449631, | |
| "grad_norm": 0.9236230373157407, | |
| "learning_rate": 6.035913590037479e-06, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13806456327438354, | |
| "step": 1885, | |
| "valid_targets_mean": 1168.8, | |
| "valid_targets_min": 726 | |
| }, | |
| { | |
| "epoch": 4.643734643734644, | |
| "grad_norm": 1.0647383277495284, | |
| "learning_rate": 5.933901705341851e-06, | |
| "loss": 0.134, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13539758324623108, | |
| "step": 1890, | |
| "valid_targets_mean": 1166.5, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 4.656019656019656, | |
| "grad_norm": 0.855176271796815, | |
| "learning_rate": 5.832608857976321e-06, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14015159010887146, | |
| "step": 1895, | |
| "valid_targets_mean": 1385.6, | |
| "valid_targets_min": 604 | |
| }, | |
| { | |
| "epoch": 4.6683046683046685, | |
| "grad_norm": 0.9385615083346516, | |
| "learning_rate": 5.732040225875584e-06, | |
| "loss": 0.1369, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1414608359336853, | |
| "step": 1900, | |
| "valid_targets_mean": 1284.1, | |
| "valid_targets_min": 740 | |
| }, | |
| { | |
| "epoch": 4.680589680589681, | |
| "grad_norm": 0.8787370830329286, | |
| "learning_rate": 5.632200949953579e-06, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12882032990455627, | |
| "step": 1905, | |
| "valid_targets_mean": 1343.2, | |
| "valid_targets_min": 508 | |
| }, | |
| { | |
| "epoch": 4.6928746928746925, | |
| "grad_norm": 0.8641680753533743, | |
| "learning_rate": 5.533096133840677e-06, | |
| "loss": 0.1295, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12614941596984863, | |
| "step": 1910, | |
| "valid_targets_mean": 1258.2, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 4.705159705159705, | |
| "grad_norm": 0.802865831972532, | |
| "learning_rate": 5.434730843622778e-06, | |
| "loss": 0.1269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1227787509560585, | |
| "step": 1915, | |
| "valid_targets_mean": 1437.7, | |
| "valid_targets_min": 941 | |
| }, | |
| { | |
| "epoch": 4.717444717444717, | |
| "grad_norm": 0.8310677761380986, | |
| "learning_rate": 5.337110107582377e-06, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12943677604198456, | |
| "step": 1920, | |
| "valid_targets_mean": 1451.6, | |
| "valid_targets_min": 933 | |
| }, | |
| { | |
| "epoch": 4.72972972972973, | |
| "grad_norm": 0.8358326381216636, | |
| "learning_rate": 5.2402389159414755e-06, | |
| "loss": 0.1334, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12806978821754456, | |
| "step": 1925, | |
| "valid_targets_mean": 1312.1, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 4.742014742014742, | |
| "grad_norm": 0.8665797867751196, | |
| "learning_rate": 5.144122220606542e-06, | |
| "loss": 0.1314, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1268056482076645, | |
| "step": 1930, | |
| "valid_targets_mean": 1296.8, | |
| "valid_targets_min": 836 | |
| }, | |
| { | |
| "epoch": 4.754299754299755, | |
| "grad_norm": 0.8718443726733912, | |
| "learning_rate": 5.048764934915349e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12933708727359772, | |
| "step": 1935, | |
| "valid_targets_mean": 1270.9, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 4.766584766584766, | |
| "grad_norm": 0.8487591221702341, | |
| "learning_rate": 4.954171933385805e-06, | |
| "loss": 0.1296, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12197130918502808, | |
| "step": 1940, | |
| "valid_targets_mean": 1190.5, | |
| "valid_targets_min": 664 | |
| }, | |
| { | |
| "epoch": 4.778869778869779, | |
| "grad_norm": 0.84375197913285, | |
| "learning_rate": 4.8603480514667836e-06, | |
| "loss": 0.1369, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1338123381137848, | |
| "step": 1945, | |
| "valid_targets_mean": 1458.1, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 4.791154791154791, | |
| "grad_norm": 0.9416639364152577, | |
| "learning_rate": 4.767298085290963e-06, | |
| "loss": 0.1397, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13794106245040894, | |
| "step": 1950, | |
| "valid_targets_mean": 1145.6, | |
| "valid_targets_min": 814 | |
| }, | |
| { | |
| "epoch": 4.803439803439804, | |
| "grad_norm": 1.04032915765231, | |
| "learning_rate": 4.675026791429624e-06, | |
| "loss": 0.131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1345062255859375, | |
| "step": 1955, | |
| "valid_targets_mean": 1348.6, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 4.815724815724816, | |
| "grad_norm": 0.8449040208170585, | |
| "learning_rate": 4.583538886649525e-06, | |
| "loss": 0.1273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1189974918961525, | |
| "step": 1960, | |
| "valid_targets_mean": 1187.4, | |
| "valid_targets_min": 769 | |
| }, | |
| { | |
| "epoch": 4.828009828009828, | |
| "grad_norm": 0.8678888365592436, | |
| "learning_rate": 4.492839047671764e-06, | |
| "loss": 0.1371, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13721004128456116, | |
| "step": 1965, | |
| "valid_targets_mean": 1184.3, | |
| "valid_targets_min": 716 | |
| }, | |
| { | |
| "epoch": 4.84029484029484, | |
| "grad_norm": 0.8554646724635764, | |
| "learning_rate": 4.4029319109327465e-06, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12759876251220703, | |
| "step": 1970, | |
| "valid_targets_mean": 1261.1, | |
| "valid_targets_min": 569 | |
| }, | |
| { | |
| "epoch": 4.8525798525798525, | |
| "grad_norm": 0.8085855019791796, | |
| "learning_rate": 4.313822072347136e-06, | |
| "loss": 0.1279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12948501110076904, | |
| "step": 1975, | |
| "valid_targets_mean": 1355.0, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 4.864864864864865, | |
| "grad_norm": 0.8874100205326839, | |
| "learning_rate": 4.22551408707296e-06, | |
| "loss": 0.1252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12678705155849457, | |
| "step": 1980, | |
| "valid_targets_mean": 1195.2, | |
| "valid_targets_min": 608 | |
| }, | |
| { | |
| "epoch": 4.877149877149877, | |
| "grad_norm": 0.8319449998626455, | |
| "learning_rate": 4.138012469278714e-06, | |
| "loss": 0.1285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12549124658107758, | |
| "step": 1985, | |
| "valid_targets_mean": 1413.9, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 4.88943488943489, | |
| "grad_norm": 0.8716998013685486, | |
| "learning_rate": 4.051321691912649e-06, | |
| "loss": 0.1314, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12598586082458496, | |
| "step": 1990, | |
| "valid_targets_mean": 1264.1, | |
| "valid_targets_min": 837 | |
| }, | |
| { | |
| "epoch": 4.901719901719901, | |
| "grad_norm": 0.8898202785026332, | |
| "learning_rate": 3.9654461864740935e-06, | |
| "loss": 0.1289, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12699808180332184, | |
| "step": 1995, | |
| "valid_targets_mean": 1339.1, | |
| "valid_targets_min": 808 | |
| }, | |
| { | |
| "epoch": 4.914004914004914, | |
| "grad_norm": 0.8810415756398233, | |
| "learning_rate": 3.880390342786915e-06, | |
| "loss": 0.1288, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1320217102766037, | |
| "step": 2000, | |
| "valid_targets_mean": 1336.9, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 4.926289926289926, | |
| "grad_norm": 0.8849692575005523, | |
| "learning_rate": 3.7961585087751516e-06, | |
| "loss": 0.1306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1198081374168396, | |
| "step": 2005, | |
| "valid_targets_mean": 1173.9, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 4.938574938574939, | |
| "grad_norm": 0.8999635972244436, | |
| "learning_rate": 3.71275499024071e-06, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12783324718475342, | |
| "step": 2010, | |
| "valid_targets_mean": 1133.1, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 4.950859950859951, | |
| "grad_norm": 0.867591599040516, | |
| "learning_rate": 3.6301840506433083e-06, | |
| "loss": 0.1287, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12289173156023026, | |
| "step": 2015, | |
| "valid_targets_mean": 1201.3, | |
| "valid_targets_min": 672 | |
| }, | |
| { | |
| "epoch": 4.963144963144963, | |
| "grad_norm": 1.1069039602237263, | |
| "learning_rate": 3.5484499108824853e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13468243181705475, | |
| "step": 2020, | |
| "valid_targets_mean": 1119.6, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 4.975429975429975, | |
| "grad_norm": 0.8588147361301989, | |
| "learning_rate": 3.4675567490818727e-06, | |
| "loss": 0.139, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13931423425674438, | |
| "step": 2025, | |
| "valid_targets_mean": 1295.9, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 4.987714987714988, | |
| "grad_norm": 0.9052294954681119, | |
| "learning_rate": 3.3875087003756036e-06, | |
| "loss": 0.1306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12855863571166992, | |
| "step": 2030, | |
| "valid_targets_mean": 1279.2, | |
| "valid_targets_min": 714 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.8268644277482449, | |
| "learning_rate": 3.30830985669691e-06, | |
| "loss": 0.1275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12387026846408844, | |
| "step": 2035, | |
| "valid_targets_mean": 1328.5, | |
| "valid_targets_min": 603 | |
| }, | |
| { | |
| "epoch": 5.012285012285012, | |
| "grad_norm": 0.7999113943897552, | |
| "learning_rate": 3.22996426656899e-06, | |
| "loss": 0.1227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11524771898984909, | |
| "step": 2040, | |
| "valid_targets_mean": 1326.2, | |
| "valid_targets_min": 818 | |
| }, | |
| { | |
| "epoch": 5.024570024570025, | |
| "grad_norm": 0.8985607563199448, | |
| "learning_rate": 3.1524759348980096e-06, | |
| "loss": 0.1215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12038896977901459, | |
| "step": 2045, | |
| "valid_targets_mean": 1191.4, | |
| "valid_targets_min": 668 | |
| }, | |
| { | |
| "epoch": 5.036855036855036, | |
| "grad_norm": 0.8643167001377622, | |
| "learning_rate": 3.0758488227684212e-06, | |
| "loss": 0.1203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12632206082344055, | |
| "step": 2050, | |
| "valid_targets_mean": 1276.1, | |
| "valid_targets_min": 546 | |
| }, | |
| { | |
| "epoch": 5.049140049140049, | |
| "grad_norm": 0.9365236976935931, | |
| "learning_rate": 3.0000868472404423e-06, | |
| "loss": 0.1242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1172647699713707, | |
| "step": 2055, | |
| "valid_targets_mean": 1160.2, | |
| "valid_targets_min": 666 | |
| }, | |
| { | |
| "epoch": 5.061425061425061, | |
| "grad_norm": 0.9196384353165374, | |
| "learning_rate": 2.9251938811498436e-06, | |
| "loss": 0.1203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11617661267518997, | |
| "step": 2060, | |
| "valid_targets_mean": 1141.0, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 5.073710073710074, | |
| "grad_norm": 0.8727744261465432, | |
| "learning_rate": 2.8511737529099704e-06, | |
| "loss": 0.1197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11832496523857117, | |
| "step": 2065, | |
| "valid_targets_mean": 1291.8, | |
| "valid_targets_min": 741 | |
| }, | |
| { | |
| "epoch": 5.085995085995086, | |
| "grad_norm": 0.9478190878710969, | |
| "learning_rate": 2.7780302463160235e-06, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12878839671611786, | |
| "step": 2070, | |
| "valid_targets_mean": 1140.1, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 5.098280098280099, | |
| "grad_norm": 0.8266940619604733, | |
| "learning_rate": 2.705767100351673e-06, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11219193786382675, | |
| "step": 2075, | |
| "valid_targets_mean": 1320.9, | |
| "valid_targets_min": 564 | |
| }, | |
| { | |
| "epoch": 5.11056511056511, | |
| "grad_norm": 0.9024231945394202, | |
| "learning_rate": 2.634388008997899e-06, | |
| "loss": 0.127, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12351719290018082, | |
| "step": 2080, | |
| "valid_targets_mean": 1220.6, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 5.122850122850123, | |
| "grad_norm": 0.8549275302001442, | |
| "learning_rate": 2.5638966210441597e-06, | |
| "loss": 0.1172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11528341472148895, | |
| "step": 2085, | |
| "valid_targets_mean": 1270.9, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 5.135135135135135, | |
| "grad_norm": 1.0114375126497932, | |
| "learning_rate": 2.4942965399018926e-06, | |
| "loss": 0.122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12501320242881775, | |
| "step": 2090, | |
| "valid_targets_mean": 1199.1, | |
| "valid_targets_min": 477 | |
| }, | |
| { | |
| "epoch": 5.1474201474201475, | |
| "grad_norm": 0.8425339847473776, | |
| "learning_rate": 2.425591323420289e-06, | |
| "loss": 0.1214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12430179864168167, | |
| "step": 2095, | |
| "valid_targets_mean": 1389.4, | |
| "valid_targets_min": 725 | |
| }, | |
| { | |
| "epoch": 5.15970515970516, | |
| "grad_norm": 0.9764933250131361, | |
| "learning_rate": 2.357784483704444e-06, | |
| "loss": 0.1205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1201491504907608, | |
| "step": 2100, | |
| "valid_targets_mean": 1180.0, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 5.171990171990172, | |
| "grad_norm": 0.9012198614613685, | |
| "learning_rate": 2.2908794869358044e-06, | |
| "loss": 0.1199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11872012913227081, | |
| "step": 2105, | |
| "valid_targets_mean": 1228.8, | |
| "valid_targets_min": 669 | |
| }, | |
| { | |
| "epoch": 5.184275184275184, | |
| "grad_norm": 0.8761086525856171, | |
| "learning_rate": 2.2248797531949952e-06, | |
| "loss": 0.1219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11814196407794952, | |
| "step": 2110, | |
| "valid_targets_mean": 1259.1, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 5.196560196560196, | |
| "grad_norm": 0.8625032263605917, | |
| "learning_rate": 2.1597886562869917e-06, | |
| "loss": 0.1228, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12705105543136597, | |
| "step": 2115, | |
| "valid_targets_mean": 1289.2, | |
| "valid_targets_min": 609 | |
| }, | |
| { | |
| "epoch": 5.208845208845209, | |
| "grad_norm": 0.8323607979530334, | |
| "learning_rate": 2.095609523568638e-06, | |
| "loss": 0.118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11345008760690689, | |
| "step": 2120, | |
| "valid_targets_mean": 1408.8, | |
| "valid_targets_min": 800 | |
| }, | |
| { | |
| "epoch": 5.221130221130221, | |
| "grad_norm": 0.9066171525552188, | |
| "learning_rate": 2.0323456357785855e-06, | |
| "loss": 0.1204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11503900587558746, | |
| "step": 2125, | |
| "valid_targets_mean": 1233.6, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 5.233415233415234, | |
| "grad_norm": 0.8470818657488131, | |
| "learning_rate": 1.970000226869553e-06, | |
| "loss": 0.1193, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11597542464733124, | |
| "step": 2130, | |
| "valid_targets_mean": 1239.8, | |
| "valid_targets_min": 698 | |
| }, | |
| { | |
| "epoch": 5.245700245700245, | |
| "grad_norm": 0.960572215465082, | |
| "learning_rate": 1.90857648384305e-06, | |
| "loss": 0.118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11610577255487442, | |
| "step": 2135, | |
| "valid_targets_mean": 1139.8, | |
| "valid_targets_min": 849 | |
| }, | |
| { | |
| "epoch": 5.257985257985258, | |
| "grad_norm": 0.8836353809793225, | |
| "learning_rate": 1.848077546586431e-06, | |
| "loss": 0.1163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.119760662317276, | |
| "step": 2140, | |
| "valid_targets_mean": 1274.1, | |
| "valid_targets_min": 585 | |
| }, | |
| { | |
| "epoch": 5.27027027027027, | |
| "grad_norm": 0.934917710885557, | |
| "learning_rate": 1.7885065077123976e-06, | |
| "loss": 0.1224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1165575161576271, | |
| "step": 2145, | |
| "valid_targets_mean": 1192.0, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 5.282555282555283, | |
| "grad_norm": 0.9435945221488572, | |
| "learning_rate": 1.7298664124009245e-06, | |
| "loss": 0.1238, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12441343069076538, | |
| "step": 2150, | |
| "valid_targets_mean": 1200.3, | |
| "valid_targets_min": 774 | |
| }, | |
| { | |
| "epoch": 5.294840294840295, | |
| "grad_norm": 0.9104365691078795, | |
| "learning_rate": 1.672160258243567e-06, | |
| "loss": 0.1176, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11665159463882446, | |
| "step": 2155, | |
| "valid_targets_mean": 1109.1, | |
| "valid_targets_min": 577 | |
| }, | |
| { | |
| "epoch": 5.3071253071253075, | |
| "grad_norm": 0.9096445128425704, | |
| "learning_rate": 1.615390995090258e-06, | |
| "loss": 0.1181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1253010630607605, | |
| "step": 2160, | |
| "valid_targets_mean": 1146.1, | |
| "valid_targets_min": 716 | |
| }, | |
| { | |
| "epoch": 5.319410319410319, | |
| "grad_norm": 0.9157686991554701, | |
| "learning_rate": 1.559561524898492e-06, | |
| "loss": 0.1214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12084513902664185, | |
| "step": 2165, | |
| "valid_targets_mean": 1176.6, | |
| "valid_targets_min": 697 | |
| }, | |
| { | |
| "epoch": 5.3316953316953315, | |
| "grad_norm": 0.8993452301524155, | |
| "learning_rate": 1.5046747015849893e-06, | |
| "loss": 0.1168, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12099088728427887, | |
| "step": 2170, | |
| "valid_targets_mean": 1287.6, | |
| "valid_targets_min": 943 | |
| }, | |
| { | |
| "epoch": 5.343980343980344, | |
| "grad_norm": 0.9188461360035166, | |
| "learning_rate": 1.4507333308798255e-06, | |
| "loss": 0.1233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11140099167823792, | |
| "step": 2175, | |
| "valid_targets_mean": 1155.9, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 5.356265356265356, | |
| "grad_norm": 0.8332971885732899, | |
| "learning_rate": 1.3977401701829752e-06, | |
| "loss": 0.1129, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10656285285949707, | |
| "step": 2180, | |
| "valid_targets_mean": 1380.5, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 5.368550368550369, | |
| "grad_norm": 0.8921983319069042, | |
| "learning_rate": 1.345697928423384e-06, | |
| "loss": 0.119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1141204684972763, | |
| "step": 2185, | |
| "valid_targets_mean": 1309.9, | |
| "valid_targets_min": 841 | |
| }, | |
| { | |
| "epoch": 5.38083538083538, | |
| "grad_norm": 0.934941810302748, | |
| "learning_rate": 1.2946092659204767e-06, | |
| "loss": 0.1217, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11862413585186005, | |
| "step": 2190, | |
| "valid_targets_mean": 1051.9, | |
| "valid_targets_min": 626 | |
| }, | |
| { | |
| "epoch": 5.393120393120393, | |
| "grad_norm": 0.9380461194253754, | |
| "learning_rate": 1.244476794248175e-06, | |
| "loss": 0.121, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12333647906780243, | |
| "step": 2195, | |
| "valid_targets_mean": 1232.8, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 5.405405405405405, | |
| "grad_norm": 0.9552431951880422, | |
| "learning_rate": 1.1953030761014017e-06, | |
| "loss": 0.1172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12513390183448792, | |
| "step": 2200, | |
| "valid_targets_mean": 1074.1, | |
| "valid_targets_min": 587 | |
| }, | |
| { | |
| "epoch": 5.417690417690418, | |
| "grad_norm": 0.9014567009232736, | |
| "learning_rate": 1.147090625165055e-06, | |
| "loss": 0.1176, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11377055943012238, | |
| "step": 2205, | |
| "valid_targets_mean": 1243.6, | |
| "valid_targets_min": 650 | |
| }, | |
| { | |
| "epoch": 5.42997542997543, | |
| "grad_norm": 0.8599695105481259, | |
| "learning_rate": 1.0998419059855503e-06, | |
| "loss": 0.1223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10883887857198715, | |
| "step": 2210, | |
| "valid_targets_mean": 1235.0, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 5.442260442260443, | |
| "grad_norm": 0.9571634722287222, | |
| "learning_rate": 1.053559333844798e-06, | |
| "loss": 0.1194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12232570350170135, | |
| "step": 2215, | |
| "valid_targets_mean": 1088.8, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 5.454545454545454, | |
| "grad_norm": 0.8940592339142096, | |
| "learning_rate": 1.0082452746367721e-06, | |
| "loss": 0.1164, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12069477140903473, | |
| "step": 2220, | |
| "valid_targets_mean": 1340.8, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 5.466830466830467, | |
| "grad_norm": 0.8180480402297732, | |
| "learning_rate": 9.639020447465475e-07, | |
| "loss": 0.1214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12307047843933105, | |
| "step": 2225, | |
| "valid_targets_mean": 1483.2, | |
| "valid_targets_min": 864 | |
| }, | |
| { | |
| "epoch": 5.479115479115479, | |
| "grad_norm": 0.8733335204397944, | |
| "learning_rate": 9.205319109318922e-07, | |
| "loss": 0.1163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11095479130744934, | |
| "step": 2230, | |
| "valid_targets_mean": 1155.1, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 5.4914004914004915, | |
| "grad_norm": 0.9078213183092334, | |
| "learning_rate": 8.781370902074049e-07, | |
| "loss": 0.1185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11644881963729858, | |
| "step": 2235, | |
| "valid_targets_mean": 1464.8, | |
| "valid_targets_min": 789 | |
| }, | |
| { | |
| "epoch": 5.503685503685504, | |
| "grad_norm": 0.8376491911217916, | |
| "learning_rate": 8.367197497311719e-07, | |
| "loss": 0.1221, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11460597068071365, | |
| "step": 2240, | |
| "valid_targets_mean": 1391.9, | |
| "valid_targets_min": 881 | |
| }, | |
| { | |
| "epoch": 5.515970515970516, | |
| "grad_norm": 0.9312884060452975, | |
| "learning_rate": 7.962820066939958e-07, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12043467909097672, | |
| "step": 2245, | |
| "valid_targets_mean": 1218.1, | |
| "valid_targets_min": 651 | |
| }, | |
| { | |
| "epoch": 5.528255528255528, | |
| "grad_norm": 0.9108572118360622, | |
| "learning_rate": 7.568259282111645e-07, | |
| "loss": 0.1208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12171508371829987, | |
| "step": 2250, | |
| "valid_targets_mean": 1225.9, | |
| "valid_targets_min": 525 | |
| }, | |
| { | |
| "epoch": 5.54054054054054, | |
| "grad_norm": 1.043375173851359, | |
| "learning_rate": 7.183535312167755e-07, | |
| "loss": 0.1248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12365029007196426, | |
| "step": 2255, | |
| "valid_targets_mean": 1268.5, | |
| "valid_targets_min": 711 | |
| }, | |
| { | |
| "epoch": 5.552825552825553, | |
| "grad_norm": 0.8221303890361632, | |
| "learning_rate": 6.808667823606474e-07, | |
| "loss": 0.1177, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10972858965396881, | |
| "step": 2260, | |
| "valid_targets_mean": 1217.7, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 5.565110565110565, | |
| "grad_norm": 0.9943052251443346, | |
| "learning_rate": 6.443675979077779e-07, | |
| "loss": 0.1187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11821258068084717, | |
| "step": 2265, | |
| "valid_targets_mean": 1140.2, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 5.577395577395578, | |
| "grad_norm": 0.8837595753139642, | |
| "learning_rate": 6.088578436403847e-07, | |
| "loss": 0.118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11932357400655746, | |
| "step": 2270, | |
| "valid_targets_mean": 1177.3, | |
| "valid_targets_min": 773 | |
| }, | |
| { | |
| "epoch": 5.58968058968059, | |
| "grad_norm": 0.9037106518704484, | |
| "learning_rate": 5.743393347625436e-07, | |
| "loss": 0.1183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12482412159442902, | |
| "step": 2275, | |
| "valid_targets_mean": 1232.4, | |
| "valid_targets_min": 682 | |
| }, | |
| { | |
| "epoch": 5.601965601965602, | |
| "grad_norm": 0.909686971601023, | |
| "learning_rate": 5.408138358073833e-07, | |
| "loss": 0.1194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11696803569793701, | |
| "step": 2280, | |
| "valid_targets_mean": 1211.9, | |
| "valid_targets_min": 817 | |
| }, | |
| { | |
| "epoch": 5.614250614250614, | |
| "grad_norm": 0.8679916324190773, | |
| "learning_rate": 5.082830605468969e-07, | |
| "loss": 0.1224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11813192814588547, | |
| "step": 2285, | |
| "valid_targets_mean": 1354.4, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 5.6265356265356266, | |
| "grad_norm": 0.8884471818216881, | |
| "learning_rate": 4.767486719043235e-07, | |
| "loss": 0.1237, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12548132240772247, | |
| "step": 2290, | |
| "valid_targets_mean": 1283.0, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 5.638820638820639, | |
| "grad_norm": 0.8715795708444471, | |
| "learning_rate": 4.4621228186915833e-07, | |
| "loss": 0.1189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11673951148986816, | |
| "step": 2295, | |
| "valid_targets_mean": 1328.6, | |
| "valid_targets_min": 663 | |
| }, | |
| { | |
| "epoch": 5.651105651105651, | |
| "grad_norm": 0.865429287358642, | |
| "learning_rate": 4.166754514147275e-07, | |
| "loss": 0.1192, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1132473275065422, | |
| "step": 2300, | |
| "valid_targets_mean": 1261.5, | |
| "valid_targets_min": 694 | |
| }, | |
| { | |
| "epoch": 5.663390663390663, | |
| "grad_norm": 0.8741499140536886, | |
| "learning_rate": 3.881396904184231e-07, | |
| "loss": 0.1231, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13480767607688904, | |
| "step": 2305, | |
| "valid_targets_mean": 1423.6, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 5.675675675675675, | |
| "grad_norm": 0.8345784138645265, | |
| "learning_rate": 3.6060645758449584e-07, | |
| "loss": 0.1159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10352207720279694, | |
| "step": 2310, | |
| "valid_targets_mean": 1274.6, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 5.687960687960688, | |
| "grad_norm": 1.0316462146052467, | |
| "learning_rate": 3.34077160369497e-07, | |
| "loss": 0.1215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12712368369102478, | |
| "step": 2315, | |
| "valid_targets_mean": 1247.8, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 5.7002457002457, | |
| "grad_norm": 0.8815652322621401, | |
| "learning_rate": 3.08553154910336e-07, | |
| "loss": 0.1217, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11496353149414062, | |
| "step": 2320, | |
| "valid_targets_mean": 1219.0, | |
| "valid_targets_min": 745 | |
| }, | |
| { | |
| "epoch": 5.712530712530713, | |
| "grad_norm": 0.9889734864928303, | |
| "learning_rate": 2.840357459549492e-07, | |
| "loss": 0.1198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12958207726478577, | |
| "step": 2325, | |
| "valid_targets_mean": 1281.6, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 5.724815724815725, | |
| "grad_norm": 0.8735025249752323, | |
| "learning_rate": 2.6052618679560884e-07, | |
| "loss": 0.1207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1203438937664032, | |
| "step": 2330, | |
| "valid_targets_mean": 1358.6, | |
| "valid_targets_min": 586 | |
| }, | |
| { | |
| "epoch": 5.737100737100737, | |
| "grad_norm": 0.9487732323352747, | |
| "learning_rate": 2.380256792048541e-07, | |
| "loss": 0.1154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10913722962141037, | |
| "step": 2335, | |
| "valid_targets_mean": 1167.6, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 5.749385749385749, | |
| "grad_norm": 0.9354429013108673, | |
| "learning_rate": 2.1653537337405383e-07, | |
| "loss": 0.1175, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12233065068721771, | |
| "step": 2340, | |
| "valid_targets_mean": 1233.7, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 5.761670761670762, | |
| "grad_norm": 0.8361440623034612, | |
| "learning_rate": 1.9605636785462234e-07, | |
| "loss": 0.1167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11600156873464584, | |
| "step": 2345, | |
| "valid_targets_mean": 1423.9, | |
| "valid_targets_min": 647 | |
| }, | |
| { | |
| "epoch": 5.773955773955774, | |
| "grad_norm": 0.88426141655337, | |
| "learning_rate": 1.7658970950185095e-07, | |
| "loss": 0.117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11672552675008774, | |
| "step": 2350, | |
| "valid_targets_mean": 1327.4, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 5.7862407862407865, | |
| "grad_norm": 0.8958608074003424, | |
| "learning_rate": 1.5813639342140197e-07, | |
| "loss": 0.1179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.121670201420784, | |
| "step": 2355, | |
| "valid_targets_mean": 1360.6, | |
| "valid_targets_min": 559 | |
| }, | |
| { | |
| "epoch": 5.798525798525798, | |
| "grad_norm": 0.8690995889535993, | |
| "learning_rate": 1.4069736291843605e-07, | |
| "loss": 0.1201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12032966315746307, | |
| "step": 2360, | |
| "valid_targets_mean": 1345.8, | |
| "valid_targets_min": 769 | |
| }, | |
| { | |
| "epoch": 5.8108108108108105, | |
| "grad_norm": 0.9609941874663688, | |
| "learning_rate": 1.242735094493952e-07, | |
| "loss": 0.1249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1279018074274063, | |
| "step": 2365, | |
| "valid_targets_mean": 1229.7, | |
| "valid_targets_min": 709 | |
| }, | |
| { | |
| "epoch": 5.823095823095823, | |
| "grad_norm": 0.8488632044982599, | |
| "learning_rate": 1.0886567257643033e-07, | |
| "loss": 0.1229, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13311204314231873, | |
| "step": 2370, | |
| "valid_targets_mean": 1436.2, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 5.835380835380835, | |
| "grad_norm": 0.8912292020833572, | |
| "learning_rate": 9.447463992448891e-08, | |
| "loss": 0.1203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12233321368694305, | |
| "step": 2375, | |
| "valid_targets_mean": 1253.7, | |
| "valid_targets_min": 728 | |
| }, | |
| { | |
| "epoch": 5.847665847665848, | |
| "grad_norm": 0.866161605046653, | |
| "learning_rate": 8.110114714104277e-08, | |
| "loss": 0.1181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11839167773723602, | |
| "step": 2380, | |
| "valid_targets_mean": 1312.7, | |
| "valid_targets_min": 798 | |
| }, | |
| { | |
| "epoch": 5.85995085995086, | |
| "grad_norm": 0.8639388177292349, | |
| "learning_rate": 6.874587785849152e-08, | |
| "loss": 0.1139, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10952752083539963, | |
| "step": 2385, | |
| "valid_targets_mean": 1177.9, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 5.872235872235873, | |
| "grad_norm": 0.8346070029983881, | |
| "learning_rate": 5.7409463659219286e-08, | |
| "loss": 0.1241, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1254696249961853, | |
| "step": 2390, | |
| "valid_targets_mean": 1511.4, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 5.884520884520884, | |
| "grad_norm": 0.8504954108670092, | |
| "learning_rate": 4.709248404329625e-08, | |
| "loss": 0.1163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1132640540599823, | |
| "step": 2395, | |
| "valid_targets_mean": 1198.4, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 5.896805896805897, | |
| "grad_norm": 0.9088038628349666, | |
| "learning_rate": 3.7795466398868885e-08, | |
| "loss": 0.1191, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11657664179801941, | |
| "step": 2400, | |
| "valid_targets_mean": 1197.9, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 5.909090909090909, | |
| "grad_norm": 0.8865491567876178, | |
| "learning_rate": 2.9518885975192702e-08, | |
| "loss": 0.121, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12183408439159393, | |
| "step": 2405, | |
| "valid_targets_mean": 1268.6, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 5.921375921375922, | |
| "grad_norm": 0.9382041318188953, | |
| "learning_rate": 2.226316585833832e-08, | |
| "loss": 0.1221, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13483551144599915, | |
| "step": 2410, | |
| "valid_targets_mean": 1244.8, | |
| "valid_targets_min": 538 | |
| }, | |
| { | |
| "epoch": 5.933660933660933, | |
| "grad_norm": 0.9374744604838307, | |
| "learning_rate": 1.6028676949570997e-08, | |
| "loss": 0.1261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12365995347499847, | |
| "step": 2415, | |
| "valid_targets_mean": 1196.9, | |
| "valid_targets_min": 770 | |
| }, | |
| { | |
| "epoch": 5.945945945945946, | |
| "grad_norm": 0.894291909380334, | |
| "learning_rate": 1.0815737946383575e-08, | |
| "loss": 0.1169, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11816893517971039, | |
| "step": 2420, | |
| "valid_targets_mean": 1345.9, | |
| "valid_targets_min": 965 | |
| }, | |
| { | |
| "epoch": 5.958230958230958, | |
| "grad_norm": 0.8845005806783053, | |
| "learning_rate": 6.624615326207284e-09, | |
| "loss": 0.117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1128786951303482, | |
| "step": 2425, | |
| "valid_targets_mean": 1206.6, | |
| "valid_targets_min": 575 | |
| }, | |
| { | |
| "epoch": 5.9705159705159705, | |
| "grad_norm": 0.8107444415043257, | |
| "learning_rate": 3.4555233327893124e-09, | |
| "loss": 0.1275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11094975471496582, | |
| "step": 2430, | |
| "valid_targets_mean": 1371.6, | |
| "valid_targets_min": 724 | |
| }, | |
| { | |
| "epoch": 5.982800982800983, | |
| "grad_norm": 0.8488528893584344, | |
| "learning_rate": 1.3086239652415621e-09, | |
| "loss": 0.1178, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10983407497406006, | |
| "step": 2435, | |
| "valid_targets_mean": 1295.4, | |
| "valid_targets_min": 751 | |
| }, | |
| { | |
| "epoch": 5.995085995085995, | |
| "grad_norm": 0.9189796749562797, | |
| "learning_rate": 1.840269697628294e-10, | |
| "loss": 0.1182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11939200758934021, | |
| "step": 2440, | |
| "valid_targets_mean": 1261.1, | |
| "valid_targets_min": 687 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11800891160964966, | |
| "step": 2442, | |
| "total_flos": 247694803206144.0, | |
| "train_loss": 0.1872950934367918, | |
| "train_runtime": 7715.4233, | |
| "train_samples_per_second": 5.057, | |
| "train_steps_per_second": 0.317, | |
| "valid_targets_mean": 1261.1, | |
| "valid_targets_min": 857 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 2442, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 6, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 247694803206144.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |