| { | |
| "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": 20.458864687141354, | |
| "learning_rate": 6.530612244897961e-07, | |
| "loss": 1.0261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 1.017302393913269, | |
| "step": 5, | |
| "valid_targets_mean": 1250.2, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 0.02457002457002457, | |
| "grad_norm": 16.62050892535432, | |
| "learning_rate": 1.469387755102041e-06, | |
| "loss": 0.9884, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.9597591161727905, | |
| "step": 10, | |
| "valid_targets_mean": 1490.1, | |
| "valid_targets_min": 971 | |
| }, | |
| { | |
| "epoch": 0.036855036855036855, | |
| "grad_norm": 16.77513160699125, | |
| "learning_rate": 2.285714285714286e-06, | |
| "loss": 0.9736, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.9614359140396118, | |
| "step": 15, | |
| "valid_targets_mean": 1209.8, | |
| "valid_targets_min": 739 | |
| }, | |
| { | |
| "epoch": 0.04914004914004914, | |
| "grad_norm": 10.374535357780216, | |
| "learning_rate": 3.1020408163265307e-06, | |
| "loss": 0.864, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.8312211036682129, | |
| "step": 20, | |
| "valid_targets_mean": 1436.6, | |
| "valid_targets_min": 853 | |
| }, | |
| { | |
| "epoch": 0.06142506142506143, | |
| "grad_norm": 6.786503884512912, | |
| "learning_rate": 3.9183673469387755e-06, | |
| "loss": 0.7762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.7279175519943237, | |
| "step": 25, | |
| "valid_targets_mean": 1216.4, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 0.07371007371007371, | |
| "grad_norm": 4.583415663665711, | |
| "learning_rate": 4.734693877551021e-06, | |
| "loss": 0.677, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6565301418304443, | |
| "step": 30, | |
| "valid_targets_mean": 1385.0, | |
| "valid_targets_min": 568 | |
| }, | |
| { | |
| "epoch": 0.085995085995086, | |
| "grad_norm": 2.731049507027889, | |
| "learning_rate": 5.551020408163266e-06, | |
| "loss": 0.612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5778182148933411, | |
| "step": 35, | |
| "valid_targets_mean": 1544.8, | |
| "valid_targets_min": 654 | |
| }, | |
| { | |
| "epoch": 0.09828009828009827, | |
| "grad_norm": 1.9598661561285227, | |
| "learning_rate": 6.36734693877551e-06, | |
| "loss": 0.5422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5209269523620605, | |
| "step": 40, | |
| "valid_targets_mean": 1367.7, | |
| "valid_targets_min": 574 | |
| }, | |
| { | |
| "epoch": 0.11056511056511056, | |
| "grad_norm": 1.638653833946308, | |
| "learning_rate": 7.183673469387755e-06, | |
| "loss": 0.4922, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.47820889949798584, | |
| "step": 45, | |
| "valid_targets_mean": 1329.5, | |
| "valid_targets_min": 531 | |
| }, | |
| { | |
| "epoch": 0.12285012285012285, | |
| "grad_norm": 1.6480129713282212, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.4663, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4506987929344177, | |
| "step": 50, | |
| "valid_targets_mean": 1316.3, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 0.13513513513513514, | |
| "grad_norm": 1.3915776930314507, | |
| "learning_rate": 8.816326530612247e-06, | |
| "loss": 0.421, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.41274315118789673, | |
| "step": 55, | |
| "valid_targets_mean": 1350.4, | |
| "valid_targets_min": 794 | |
| }, | |
| { | |
| "epoch": 0.14742014742014742, | |
| "grad_norm": 1.8150172535705966, | |
| "learning_rate": 9.63265306122449e-06, | |
| "loss": 0.4052, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37979835271835327, | |
| "step": 60, | |
| "valid_targets_mean": 1348.1, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 0.1597051597051597, | |
| "grad_norm": 1.2911082102402043, | |
| "learning_rate": 1.0448979591836737e-05, | |
| "loss": 0.3751, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36601579189300537, | |
| "step": 65, | |
| "valid_targets_mean": 1233.3, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 0.171990171990172, | |
| "grad_norm": 1.1066917837756176, | |
| "learning_rate": 1.126530612244898e-05, | |
| "loss": 0.3725, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3778034746646881, | |
| "step": 70, | |
| "valid_targets_mean": 1599.6, | |
| "valid_targets_min": 768 | |
| }, | |
| { | |
| "epoch": 0.18427518427518427, | |
| "grad_norm": 1.2330283871793921, | |
| "learning_rate": 1.2081632653061225e-05, | |
| "loss": 0.3497, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33981752395629883, | |
| "step": 75, | |
| "valid_targets_mean": 1175.1, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 0.19656019656019655, | |
| "grad_norm": 1.1955197787044425, | |
| "learning_rate": 1.2897959183673469e-05, | |
| "loss": 0.3275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3365780711174011, | |
| "step": 80, | |
| "valid_targets_mean": 1151.1, | |
| "valid_targets_min": 712 | |
| }, | |
| { | |
| "epoch": 0.20884520884520885, | |
| "grad_norm": 1.1040514955581373, | |
| "learning_rate": 1.3714285714285716e-05, | |
| "loss": 0.3205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31247782707214355, | |
| "step": 85, | |
| "valid_targets_mean": 1273.6, | |
| "valid_targets_min": 590 | |
| }, | |
| { | |
| "epoch": 0.22113022113022113, | |
| "grad_norm": 1.1067102481311106, | |
| "learning_rate": 1.4530612244897961e-05, | |
| "loss": 0.3052, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3124191164970398, | |
| "step": 90, | |
| "valid_targets_mean": 1315.8, | |
| "valid_targets_min": 649 | |
| }, | |
| { | |
| "epoch": 0.2334152334152334, | |
| "grad_norm": 1.1174617597851082, | |
| "learning_rate": 1.5346938775510204e-05, | |
| "loss": 0.3146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31378817558288574, | |
| "step": 95, | |
| "valid_targets_mean": 1209.7, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.2457002457002457, | |
| "grad_norm": 1.0357427151455243, | |
| "learning_rate": 1.616326530612245e-05, | |
| "loss": 0.3138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29703712463378906, | |
| "step": 100, | |
| "valid_targets_mean": 1322.4, | |
| "valid_targets_min": 940 | |
| }, | |
| { | |
| "epoch": 0.257985257985258, | |
| "grad_norm": 1.1422371590736977, | |
| "learning_rate": 1.6979591836734695e-05, | |
| "loss": 0.3098, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3078041970729828, | |
| "step": 105, | |
| "valid_targets_mean": 1323.4, | |
| "valid_targets_min": 711 | |
| }, | |
| { | |
| "epoch": 0.2702702702702703, | |
| "grad_norm": 1.0038192317155883, | |
| "learning_rate": 1.779591836734694e-05, | |
| "loss": 0.2963, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3173219859600067, | |
| "step": 110, | |
| "valid_targets_mean": 1408.8, | |
| "valid_targets_min": 897 | |
| }, | |
| { | |
| "epoch": 0.28255528255528256, | |
| "grad_norm": 0.9855120304223134, | |
| "learning_rate": 1.8612244897959185e-05, | |
| "loss": 0.2866, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2685781717300415, | |
| "step": 115, | |
| "valid_targets_mean": 1392.1, | |
| "valid_targets_min": 813 | |
| }, | |
| { | |
| "epoch": 0.29484029484029484, | |
| "grad_norm": 1.0672252795112558, | |
| "learning_rate": 1.942857142857143e-05, | |
| "loss": 0.289, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2954533100128174, | |
| "step": 120, | |
| "valid_targets_mean": 1364.4, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 0.3071253071253071, | |
| "grad_norm": 1.0799002257104326, | |
| "learning_rate": 2.0244897959183672e-05, | |
| "loss": 0.2895, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2752310037612915, | |
| "step": 125, | |
| "valid_targets_mean": 1465.1, | |
| "valid_targets_min": 941 | |
| }, | |
| { | |
| "epoch": 0.3194103194103194, | |
| "grad_norm": 1.3135467049033194, | |
| "learning_rate": 2.106122448979592e-05, | |
| "loss": 0.2925, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29970723390579224, | |
| "step": 130, | |
| "valid_targets_mean": 1072.5, | |
| "valid_targets_min": 663 | |
| }, | |
| { | |
| "epoch": 0.3316953316953317, | |
| "grad_norm": 1.1052416113035322, | |
| "learning_rate": 2.1877551020408166e-05, | |
| "loss": 0.2846, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3040231466293335, | |
| "step": 135, | |
| "valid_targets_mean": 1425.1, | |
| "valid_targets_min": 727 | |
| }, | |
| { | |
| "epoch": 0.343980343980344, | |
| "grad_norm": 1.0034696175995714, | |
| "learning_rate": 2.269387755102041e-05, | |
| "loss": 0.2975, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31153547763824463, | |
| "step": 140, | |
| "valid_targets_mean": 1287.7, | |
| "valid_targets_min": 637 | |
| }, | |
| { | |
| "epoch": 0.35626535626535627, | |
| "grad_norm": 1.0317670697456185, | |
| "learning_rate": 2.3510204081632656e-05, | |
| "loss": 0.275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28750118613243103, | |
| "step": 145, | |
| "valid_targets_mean": 1450.8, | |
| "valid_targets_min": 724 | |
| }, | |
| { | |
| "epoch": 0.36855036855036855, | |
| "grad_norm": 1.045825397854041, | |
| "learning_rate": 2.4326530612244898e-05, | |
| "loss": 0.2869, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28550174832344055, | |
| "step": 150, | |
| "valid_targets_mean": 1355.8, | |
| "valid_targets_min": 646 | |
| }, | |
| { | |
| "epoch": 0.3808353808353808, | |
| "grad_norm": 1.0635125133091923, | |
| "learning_rate": 2.5142857142857143e-05, | |
| "loss": 0.2729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2671389877796173, | |
| "step": 155, | |
| "valid_targets_mean": 1190.7, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 0.3931203931203931, | |
| "grad_norm": 1.0793917463505718, | |
| "learning_rate": 2.5959183673469392e-05, | |
| "loss": 0.2829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2713095545768738, | |
| "step": 160, | |
| "valid_targets_mean": 1198.3, | |
| "valid_targets_min": 800 | |
| }, | |
| { | |
| "epoch": 0.40540540540540543, | |
| "grad_norm": 1.2207070496489467, | |
| "learning_rate": 2.6775510204081637e-05, | |
| "loss": 0.2733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2674906253814697, | |
| "step": 165, | |
| "valid_targets_mean": 1239.4, | |
| "valid_targets_min": 713 | |
| }, | |
| { | |
| "epoch": 0.4176904176904177, | |
| "grad_norm": 1.0046068676053646, | |
| "learning_rate": 2.7591836734693882e-05, | |
| "loss": 0.2754, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26432591676712036, | |
| "step": 170, | |
| "valid_targets_mean": 1440.5, | |
| "valid_targets_min": 598 | |
| }, | |
| { | |
| "epoch": 0.42997542997543, | |
| "grad_norm": 0.9375984105225126, | |
| "learning_rate": 2.8408163265306124e-05, | |
| "loss": 0.2772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2688313126564026, | |
| "step": 175, | |
| "valid_targets_mean": 1560.9, | |
| "valid_targets_min": 872 | |
| }, | |
| { | |
| "epoch": 0.44226044226044225, | |
| "grad_norm": 1.1180598027008593, | |
| "learning_rate": 2.922448979591837e-05, | |
| "loss": 0.2782, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2788759171962738, | |
| "step": 180, | |
| "valid_targets_mean": 1178.6, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 0.45454545454545453, | |
| "grad_norm": 1.115636815298234, | |
| "learning_rate": 3.0040816326530614e-05, | |
| "loss": 0.2688, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27034085988998413, | |
| "step": 185, | |
| "valid_targets_mean": 1167.0, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 0.4668304668304668, | |
| "grad_norm": 1.1455483058873124, | |
| "learning_rate": 3.085714285714286e-05, | |
| "loss": 0.264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28708407282829285, | |
| "step": 190, | |
| "valid_targets_mean": 1329.9, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 0.47911547911547914, | |
| "grad_norm": 1.1215868583441444, | |
| "learning_rate": 3.1673469387755105e-05, | |
| "loss": 0.26, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24477200210094452, | |
| "step": 195, | |
| "valid_targets_mean": 1164.9, | |
| "valid_targets_min": 756 | |
| }, | |
| { | |
| "epoch": 0.4914004914004914, | |
| "grad_norm": 1.0774130715647405, | |
| "learning_rate": 3.2489795918367346e-05, | |
| "loss": 0.252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22574767470359802, | |
| "step": 200, | |
| "valid_targets_mean": 1299.4, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 0.5036855036855037, | |
| "grad_norm": 1.4181221032778561, | |
| "learning_rate": 3.3306122448979595e-05, | |
| "loss": 0.2661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26352864503860474, | |
| "step": 205, | |
| "valid_targets_mean": 1260.2, | |
| "valid_targets_min": 874 | |
| }, | |
| { | |
| "epoch": 0.515970515970516, | |
| "grad_norm": 1.147550012481648, | |
| "learning_rate": 3.4122448979591843e-05, | |
| "loss": 0.261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.261576384305954, | |
| "step": 210, | |
| "valid_targets_mean": 1420.3, | |
| "valid_targets_min": 861 | |
| }, | |
| { | |
| "epoch": 0.5282555282555282, | |
| "grad_norm": 0.9387739790313563, | |
| "learning_rate": 3.4938775510204085e-05, | |
| "loss": 0.2496, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2405305653810501, | |
| "step": 215, | |
| "valid_targets_mean": 1357.9, | |
| "valid_targets_min": 1000 | |
| }, | |
| { | |
| "epoch": 0.5405405405405406, | |
| "grad_norm": 1.0028258368925376, | |
| "learning_rate": 3.575510204081633e-05, | |
| "loss": 0.2573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25683438777923584, | |
| "step": 220, | |
| "valid_targets_mean": 1370.2, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 0.5528255528255528, | |
| "grad_norm": 1.0087280455087901, | |
| "learning_rate": 3.6571428571428576e-05, | |
| "loss": 0.2552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2451547086238861, | |
| "step": 225, | |
| "valid_targets_mean": 1205.2, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 0.5651105651105651, | |
| "grad_norm": 1.0332327473283356, | |
| "learning_rate": 3.738775510204082e-05, | |
| "loss": 0.2577, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2729160785675049, | |
| "step": 230, | |
| "valid_targets_mean": 1312.5, | |
| "valid_targets_min": 809 | |
| }, | |
| { | |
| "epoch": 0.5773955773955773, | |
| "grad_norm": 1.0664198583479307, | |
| "learning_rate": 3.8204081632653066e-05, | |
| "loss": 0.251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24643297493457794, | |
| "step": 235, | |
| "valid_targets_mean": 1375.5, | |
| "valid_targets_min": 725 | |
| }, | |
| { | |
| "epoch": 0.5896805896805897, | |
| "grad_norm": 1.3724435186192199, | |
| "learning_rate": 3.902040816326531e-05, | |
| "loss": 0.2615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23138344287872314, | |
| "step": 240, | |
| "valid_targets_mean": 1403.4, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.601965601965602, | |
| "grad_norm": 0.9920775530391814, | |
| "learning_rate": 3.983673469387755e-05, | |
| "loss": 0.262, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2616458237171173, | |
| "step": 245, | |
| "valid_targets_mean": 1307.3, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 0.6142506142506142, | |
| "grad_norm": 1.0722956456758705, | |
| "learning_rate": 3.9999672841332876e-05, | |
| "loss": 0.2551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23813892900943756, | |
| "step": 250, | |
| "valid_targets_mean": 1384.2, | |
| "valid_targets_min": 643 | |
| }, | |
| { | |
| "epoch": 0.6265356265356266, | |
| "grad_norm": 1.187546618081259, | |
| "learning_rate": 3.999834377759164e-05, | |
| "loss": 0.2605, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25846004486083984, | |
| "step": 255, | |
| "valid_targets_mean": 1287.4, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 0.6388206388206388, | |
| "grad_norm": 1.0313710783686258, | |
| "learning_rate": 3.999599242924703e-05, | |
| "loss": 0.2569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25932109355926514, | |
| "step": 260, | |
| "valid_targets_mean": 1212.7, | |
| "valid_targets_min": 577 | |
| }, | |
| { | |
| "epoch": 0.6511056511056511, | |
| "grad_norm": 0.9509239983764558, | |
| "learning_rate": 3.999261891649637e-05, | |
| "loss": 0.2569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23473910987377167, | |
| "step": 265, | |
| "valid_targets_mean": 1461.9, | |
| "valid_targets_min": 979 | |
| }, | |
| { | |
| "epoch": 0.6633906633906634, | |
| "grad_norm": 0.9843909928851137, | |
| "learning_rate": 3.9988223411788436e-05, | |
| "loss": 0.2612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2610856294631958, | |
| "step": 270, | |
| "valid_targets_mean": 1577.8, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 0.6756756756756757, | |
| "grad_norm": 2.3257950263930125, | |
| "learning_rate": 3.998280613981468e-05, | |
| "loss": 0.2481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24677515029907227, | |
| "step": 275, | |
| "valid_targets_mean": 1449.1, | |
| "valid_targets_min": 819 | |
| }, | |
| { | |
| "epoch": 0.687960687960688, | |
| "grad_norm": 1.0120069901357338, | |
| "learning_rate": 3.9976367377497725e-05, | |
| "loss": 0.252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.243765190243721, | |
| "step": 280, | |
| "valid_targets_mean": 1120.2, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.7002457002457002, | |
| "grad_norm": 1.0056916875862807, | |
| "learning_rate": 3.99689074539772e-05, | |
| "loss": 0.2498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24412012100219727, | |
| "step": 285, | |
| "valid_targets_mean": 1168.4, | |
| "valid_targets_min": 599 | |
| }, | |
| { | |
| "epoch": 0.7125307125307125, | |
| "grad_norm": 0.9928969394751843, | |
| "learning_rate": 3.9960426750592936e-05, | |
| "loss": 0.2447, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25321725010871887, | |
| "step": 290, | |
| "valid_targets_mean": 1205.3, | |
| "valid_targets_min": 833 | |
| }, | |
| { | |
| "epoch": 0.7248157248157249, | |
| "grad_norm": 1.0035539521741665, | |
| "learning_rate": 3.995092570086546e-05, | |
| "loss": 0.2502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23981334269046783, | |
| "step": 295, | |
| "valid_targets_mean": 1309.0, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 0.7371007371007371, | |
| "grad_norm": 0.9217446118721824, | |
| "learning_rate": 3.9940404790473825e-05, | |
| "loss": 0.2461, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22987031936645508, | |
| "step": 300, | |
| "valid_targets_mean": 1249.8, | |
| "valid_targets_min": 656 | |
| }, | |
| { | |
| "epoch": 0.7493857493857494, | |
| "grad_norm": 0.9911047619338179, | |
| "learning_rate": 3.992886455723082e-05, | |
| "loss": 0.2502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25888746976852417, | |
| "step": 305, | |
| "valid_targets_mean": 1174.4, | |
| "valid_targets_min": 546 | |
| }, | |
| { | |
| "epoch": 0.7616707616707616, | |
| "grad_norm": 0.8658673852164612, | |
| "learning_rate": 3.991630559105541e-05, | |
| "loss": 0.236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22796308994293213, | |
| "step": 310, | |
| "valid_targets_mean": 1366.1, | |
| "valid_targets_min": 828 | |
| }, | |
| { | |
| "epoch": 0.773955773955774, | |
| "grad_norm": 1.0842055642509907, | |
| "learning_rate": 3.990272853394268e-05, | |
| "loss": 0.2446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23061205446720123, | |
| "step": 315, | |
| "valid_targets_mean": 1168.6, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 0.7862407862407862, | |
| "grad_norm": 0.9014834838441966, | |
| "learning_rate": 3.988813407993089e-05, | |
| "loss": 0.2254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23373663425445557, | |
| "step": 320, | |
| "valid_targets_mean": 1399.3, | |
| "valid_targets_min": 848 | |
| }, | |
| { | |
| "epoch": 0.7985257985257985, | |
| "grad_norm": 0.9338172197869813, | |
| "learning_rate": 3.987252297506613e-05, | |
| "loss": 0.2405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2474566400051117, | |
| "step": 325, | |
| "valid_targets_mean": 1277.9, | |
| "valid_targets_min": 659 | |
| }, | |
| { | |
| "epoch": 0.8108108108108109, | |
| "grad_norm": 0.913549961658726, | |
| "learning_rate": 3.9855896017364075e-05, | |
| "loss": 0.2373, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24477137625217438, | |
| "step": 330, | |
| "valid_targets_mean": 1268.5, | |
| "valid_targets_min": 559 | |
| }, | |
| { | |
| "epoch": 0.8230958230958231, | |
| "grad_norm": 0.9127581901150593, | |
| "learning_rate": 3.983825405676927e-05, | |
| "loss": 0.2442, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2348695546388626, | |
| "step": 335, | |
| "valid_targets_mean": 1384.6, | |
| "valid_targets_min": 898 | |
| }, | |
| { | |
| "epoch": 0.8353808353808354, | |
| "grad_norm": 0.886670563471073, | |
| "learning_rate": 3.981959799511161e-05, | |
| "loss": 0.242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23363958299160004, | |
| "step": 340, | |
| "valid_targets_mean": 1429.6, | |
| "valid_targets_min": 613 | |
| }, | |
| { | |
| "epoch": 0.8476658476658476, | |
| "grad_norm": 1.101260931849648, | |
| "learning_rate": 3.979992878606032e-05, | |
| "loss": 0.2486, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23046278953552246, | |
| "step": 345, | |
| "valid_targets_mean": 1162.9, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 0.85995085995086, | |
| "grad_norm": 0.9340190900844615, | |
| "learning_rate": 3.977924743507513e-05, | |
| "loss": 0.2479, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23310105502605438, | |
| "step": 350, | |
| "valid_targets_mean": 1256.2, | |
| "valid_targets_min": 592 | |
| }, | |
| { | |
| "epoch": 0.8722358722358723, | |
| "grad_norm": 0.941031348253831, | |
| "learning_rate": 3.975755499935492e-05, | |
| "loss": 0.2552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27752816677093506, | |
| "step": 355, | |
| "valid_targets_mean": 1220.6, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 0.8845208845208845, | |
| "grad_norm": 0.7902767470010282, | |
| "learning_rate": 3.973485258778368e-05, | |
| "loss": 0.2316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.230636328458786, | |
| "step": 360, | |
| "valid_targets_mean": 1582.8, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 0.8968058968058968, | |
| "grad_norm": 0.9584038623179534, | |
| "learning_rate": 3.971114136087379e-05, | |
| "loss": 0.2409, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24100783467292786, | |
| "step": 365, | |
| "valid_targets_mean": 1402.8, | |
| "valid_targets_min": 834 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 0.9401862753457677, | |
| "learning_rate": 3.968642253070675e-05, | |
| "loss": 0.2353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23034992814064026, | |
| "step": 370, | |
| "valid_targets_mean": 1309.1, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 0.9213759213759214, | |
| "grad_norm": 0.926409081388446, | |
| "learning_rate": 3.966069736087116e-05, | |
| "loss": 0.2481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24051252007484436, | |
| "step": 375, | |
| "valid_targets_mean": 1465.2, | |
| "valid_targets_min": 569 | |
| }, | |
| { | |
| "epoch": 0.9336609336609336, | |
| "grad_norm": 0.9392772230831478, | |
| "learning_rate": 3.963396716639818e-05, | |
| "loss": 0.2361, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2311394214630127, | |
| "step": 380, | |
| "valid_targets_mean": 1381.5, | |
| "valid_targets_min": 855 | |
| }, | |
| { | |
| "epoch": 0.9459459459459459, | |
| "grad_norm": 1.1295683321200982, | |
| "learning_rate": 3.960623331369427e-05, | |
| "loss": 0.2355, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20992863178253174, | |
| "step": 385, | |
| "valid_targets_mean": 1067.1, | |
| "valid_targets_min": 833 | |
| }, | |
| { | |
| "epoch": 0.9582309582309583, | |
| "grad_norm": 0.8472505054920229, | |
| "learning_rate": 3.957749722047138e-05, | |
| "loss": 0.231, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2383286952972412, | |
| "step": 390, | |
| "valid_targets_mean": 1291.5, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 0.9705159705159705, | |
| "grad_norm": 0.8747395591274454, | |
| "learning_rate": 3.9547760355674405e-05, | |
| "loss": 0.2226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22372183203697205, | |
| "step": 395, | |
| "valid_targets_mean": 1429.6, | |
| "valid_targets_min": 860 | |
| }, | |
| { | |
| "epoch": 0.9828009828009828, | |
| "grad_norm": 0.8723755304470501, | |
| "learning_rate": 3.951702423940621e-05, | |
| "loss": 0.235, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2421160787343979, | |
| "step": 400, | |
| "valid_targets_mean": 1270.6, | |
| "valid_targets_min": 668 | |
| }, | |
| { | |
| "epoch": 0.995085995085995, | |
| "grad_norm": 1.05363286328019, | |
| "learning_rate": 3.948529044284981e-05, | |
| "loss": 0.2302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24994471669197083, | |
| "step": 405, | |
| "valid_targets_mean": 1449.1, | |
| "valid_targets_min": 605 | |
| }, | |
| { | |
| "epoch": 1.0073710073710074, | |
| "grad_norm": 0.8920261500246442, | |
| "learning_rate": 3.9452560588188135e-05, | |
| "loss": 0.2446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2395254522562027, | |
| "step": 410, | |
| "valid_targets_mean": 1442.7, | |
| "valid_targets_min": 625 | |
| }, | |
| { | |
| "epoch": 1.0196560196560196, | |
| "grad_norm": 0.8788201882105066, | |
| "learning_rate": 3.9418836348521045e-05, | |
| "loss": 0.2311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22359757125377655, | |
| "step": 415, | |
| "valid_targets_mean": 1561.7, | |
| "valid_targets_min": 854 | |
| }, | |
| { | |
| "epoch": 1.031941031941032, | |
| "grad_norm": 0.8963576975802414, | |
| "learning_rate": 3.9384119447779854e-05, | |
| "loss": 0.2234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22188884019851685, | |
| "step": 420, | |
| "valid_targets_mean": 1369.3, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 1.0442260442260443, | |
| "grad_norm": 0.9312090201385392, | |
| "learning_rate": 3.934841166063919e-05, | |
| "loss": 0.2195, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21513287723064423, | |
| "step": 425, | |
| "valid_targets_mean": 1252.8, | |
| "valid_targets_min": 555 | |
| }, | |
| { | |
| "epoch": 1.0565110565110565, | |
| "grad_norm": 0.9090544401101063, | |
| "learning_rate": 3.931171481242625e-05, | |
| "loss": 0.2099, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20362290740013123, | |
| "step": 430, | |
| "valid_targets_mean": 1281.4, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 1.0687960687960687, | |
| "grad_norm": 0.8960022842390729, | |
| "learning_rate": 3.927403077902753e-05, | |
| "loss": 0.2134, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22219829261302948, | |
| "step": 435, | |
| "valid_targets_mean": 1247.5, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 1.0810810810810811, | |
| "grad_norm": 1.096520420802368, | |
| "learning_rate": 3.9235361486792905e-05, | |
| "loss": 0.2235, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24039161205291748, | |
| "step": 440, | |
| "valid_targets_mean": 1353.1, | |
| "valid_targets_min": 859 | |
| }, | |
| { | |
| "epoch": 1.0933660933660934, | |
| "grad_norm": 0.8339795162374838, | |
| "learning_rate": 3.9195708912437176e-05, | |
| "loss": 0.2176, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1995536983013153, | |
| "step": 445, | |
| "valid_targets_mean": 1291.4, | |
| "valid_targets_min": 832 | |
| }, | |
| { | |
| "epoch": 1.1056511056511056, | |
| "grad_norm": 0.8344607923251784, | |
| "learning_rate": 3.915507508293901e-05, | |
| "loss": 0.2135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2221432775259018, | |
| "step": 450, | |
| "valid_targets_mean": 1466.1, | |
| "valid_targets_min": 928 | |
| }, | |
| { | |
| "epoch": 1.117936117936118, | |
| "grad_norm": 0.8563952361791388, | |
| "learning_rate": 3.911346207543734e-05, | |
| "loss": 0.2097, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2145756334066391, | |
| "step": 455, | |
| "valid_targets_mean": 1386.1, | |
| "valid_targets_min": 676 | |
| }, | |
| { | |
| "epoch": 1.1302211302211302, | |
| "grad_norm": 1.0053828040779496, | |
| "learning_rate": 3.907087201712515e-05, | |
| "loss": 0.2339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23731562495231628, | |
| "step": 460, | |
| "valid_targets_mean": 1351.2, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 1.1425061425061425, | |
| "grad_norm": 0.9200687870617267, | |
| "learning_rate": 3.902730708514078e-05, | |
| "loss": 0.2079, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19236290454864502, | |
| "step": 465, | |
| "valid_targets_mean": 1237.4, | |
| "valid_targets_min": 707 | |
| }, | |
| { | |
| "epoch": 1.154791154791155, | |
| "grad_norm": 1.956829606116063, | |
| "learning_rate": 3.8982769506456616e-05, | |
| "loss": 0.2138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2312776744365692, | |
| "step": 470, | |
| "valid_targets_mean": 1205.5, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 1.1670761670761671, | |
| "grad_norm": 0.8376975426348914, | |
| "learning_rate": 3.893726155776524e-05, | |
| "loss": 0.2174, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20141255855560303, | |
| "step": 475, | |
| "valid_targets_mean": 1293.6, | |
| "valid_targets_min": 552 | |
| }, | |
| { | |
| "epoch": 1.1793611793611793, | |
| "grad_norm": 0.9106721894053932, | |
| "learning_rate": 3.8890785565363046e-05, | |
| "loss": 0.2105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21375897526741028, | |
| "step": 480, | |
| "valid_targets_mean": 1276.3, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 1.1916461916461916, | |
| "grad_norm": 0.9575923245260296, | |
| "learning_rate": 3.884334390503136e-05, | |
| "loss": 0.219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21761855483055115, | |
| "step": 485, | |
| "valid_targets_mean": 1113.1, | |
| "valid_targets_min": 623 | |
| }, | |
| { | |
| "epoch": 1.203931203931204, | |
| "grad_norm": 0.9627502770339411, | |
| "learning_rate": 3.8794939001914955e-05, | |
| "loss": 0.2135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22376301884651184, | |
| "step": 490, | |
| "valid_targets_mean": 1444.7, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 1.2162162162162162, | |
| "grad_norm": 0.9284874253755052, | |
| "learning_rate": 3.87455733303981e-05, | |
| "loss": 0.2181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22301726043224335, | |
| "step": 495, | |
| "valid_targets_mean": 1192.3, | |
| "valid_targets_min": 559 | |
| }, | |
| { | |
| "epoch": 1.2285012285012284, | |
| "grad_norm": 0.9095646070613896, | |
| "learning_rate": 3.869524941397805e-05, | |
| "loss": 0.2138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2163151502609253, | |
| "step": 500, | |
| "valid_targets_mean": 1267.6, | |
| "valid_targets_min": 702 | |
| }, | |
| { | |
| "epoch": 1.2407862407862407, | |
| "grad_norm": 2.108591173486237, | |
| "learning_rate": 3.8643969825136095e-05, | |
| "loss": 0.2161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22375309467315674, | |
| "step": 505, | |
| "valid_targets_mean": 1189.6, | |
| "valid_targets_min": 599 | |
| }, | |
| { | |
| "epoch": 1.253071253071253, | |
| "grad_norm": 0.8526514308763375, | |
| "learning_rate": 3.8591737185206024e-05, | |
| "loss": 0.2155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20474407076835632, | |
| "step": 510, | |
| "valid_targets_mean": 1360.9, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 1.2653562653562653, | |
| "grad_norm": 0.8781899367047357, | |
| "learning_rate": 3.853855416424011e-05, | |
| "loss": 0.2153, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22370408475399017, | |
| "step": 515, | |
| "valid_targets_mean": 1262.0, | |
| "valid_targets_min": 666 | |
| }, | |
| { | |
| "epoch": 1.2776412776412776, | |
| "grad_norm": 0.9403635720815047, | |
| "learning_rate": 3.848442348087267e-05, | |
| "loss": 0.2134, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.202105313539505, | |
| "step": 520, | |
| "valid_targets_mean": 1425.7, | |
| "valid_targets_min": 764 | |
| }, | |
| { | |
| "epoch": 1.28992628992629, | |
| "grad_norm": 0.8090564776917256, | |
| "learning_rate": 3.842934790218106e-05, | |
| "loss": 0.2104, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21103046834468842, | |
| "step": 525, | |
| "valid_targets_mean": 1308.6, | |
| "valid_targets_min": 856 | |
| }, | |
| { | |
| "epoch": 1.3022113022113022, | |
| "grad_norm": 1.0851226060699104, | |
| "learning_rate": 3.837333024354422e-05, | |
| "loss": 0.215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23397046327590942, | |
| "step": 530, | |
| "valid_targets_mean": 1284.8, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 1.3144963144963144, | |
| "grad_norm": 0.8157174831500646, | |
| "learning_rate": 3.8316373368498794e-05, | |
| "loss": 0.1986, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21336279809474945, | |
| "step": 535, | |
| "valid_targets_mean": 1468.4, | |
| "valid_targets_min": 604 | |
| }, | |
| { | |
| "epoch": 1.3267813267813269, | |
| "grad_norm": 0.8259937885003744, | |
| "learning_rate": 3.82584801885927e-05, | |
| "loss": 0.2177, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21354883909225464, | |
| "step": 540, | |
| "valid_targets_mean": 1441.9, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 1.339066339066339, | |
| "grad_norm": 0.7719178515354956, | |
| "learning_rate": 3.8199653663236336e-05, | |
| "loss": 0.2204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22929129004478455, | |
| "step": 545, | |
| "valid_targets_mean": 1425.4, | |
| "valid_targets_min": 957 | |
| }, | |
| { | |
| "epoch": 1.3513513513513513, | |
| "grad_norm": 0.9713509399154334, | |
| "learning_rate": 3.813989679955128e-05, | |
| "loss": 0.2103, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20775477588176727, | |
| "step": 550, | |
| "valid_targets_mean": 1368.8, | |
| "valid_targets_min": 826 | |
| }, | |
| { | |
| "epoch": 1.3636363636363638, | |
| "grad_norm": 1.5657871968007089, | |
| "learning_rate": 3.8079212652216595e-05, | |
| "loss": 0.2053, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2107270210981369, | |
| "step": 555, | |
| "valid_targets_mean": 1369.9, | |
| "valid_targets_min": 602 | |
| }, | |
| { | |
| "epoch": 1.375921375921376, | |
| "grad_norm": 0.7739864999218562, | |
| "learning_rate": 3.8017604323312616e-05, | |
| "loss": 0.2089, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20910859107971191, | |
| "step": 560, | |
| "valid_targets_mean": 1351.6, | |
| "valid_targets_min": 717 | |
| }, | |
| { | |
| "epoch": 1.3882063882063882, | |
| "grad_norm": 0.8934410106791703, | |
| "learning_rate": 3.795507496216246e-05, | |
| "loss": 0.2218, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22969551384449005, | |
| "step": 565, | |
| "valid_targets_mean": 1307.3, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 1.4004914004914004, | |
| "grad_norm": 0.8511838263014904, | |
| "learning_rate": 3.789162776517098e-05, | |
| "loss": 0.2131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2171616405248642, | |
| "step": 570, | |
| "valid_targets_mean": 1304.6, | |
| "valid_targets_min": 677 | |
| }, | |
| { | |
| "epoch": 1.4127764127764126, | |
| "grad_norm": 0.8141913250178237, | |
| "learning_rate": 3.78272659756614e-05, | |
| "loss": 0.2192, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23532284796237946, | |
| "step": 575, | |
| "valid_targets_mean": 1437.6, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 1.425061425061425, | |
| "grad_norm": 0.8420915956597627, | |
| "learning_rate": 3.776199288370948e-05, | |
| "loss": 0.2192, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23164448142051697, | |
| "step": 580, | |
| "valid_targets_mean": 1419.3, | |
| "valid_targets_min": 851 | |
| }, | |
| { | |
| "epoch": 1.4373464373464373, | |
| "grad_norm": 0.8968223148938935, | |
| "learning_rate": 3.7695811825975386e-05, | |
| "loss": 0.2063, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19434265792369843, | |
| "step": 585, | |
| "valid_targets_mean": 1047.1, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 1.4496314496314495, | |
| "grad_norm": 0.7610632883336738, | |
| "learning_rate": 3.76287261855331e-05, | |
| "loss": 0.208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20634359121322632, | |
| "step": 590, | |
| "valid_targets_mean": 1468.5, | |
| "valid_targets_min": 906 | |
| }, | |
| { | |
| "epoch": 1.461916461916462, | |
| "grad_norm": 0.7589326597800894, | |
| "learning_rate": 3.7560739391697465e-05, | |
| "loss": 0.2154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18889199197292328, | |
| "step": 595, | |
| "valid_targets_mean": 1226.5, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 1.4742014742014742, | |
| "grad_norm": 0.8216098043735577, | |
| "learning_rate": 3.749185491984891e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20000270009040833, | |
| "step": 600, | |
| "valid_targets_mean": 1265.7, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 1.4864864864864864, | |
| "grad_norm": 0.8373041646274438, | |
| "learning_rate": 3.7422076291255785e-05, | |
| "loss": 0.1991, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1893223524093628, | |
| "step": 605, | |
| "valid_targets_mean": 1247.5, | |
| "valid_targets_min": 677 | |
| }, | |
| { | |
| "epoch": 1.4987714987714988, | |
| "grad_norm": 0.8414739647341087, | |
| "learning_rate": 3.7351407072894356e-05, | |
| "loss": 0.2101, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20384536683559418, | |
| "step": 610, | |
| "valid_targets_mean": 1286.6, | |
| "valid_targets_min": 957 | |
| }, | |
| { | |
| "epoch": 1.511056511056511, | |
| "grad_norm": 0.869973506541461, | |
| "learning_rate": 3.7279850877266486e-05, | |
| "loss": 0.2223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22863122820854187, | |
| "step": 615, | |
| "valid_targets_mean": 1334.2, | |
| "valid_targets_min": 727 | |
| }, | |
| { | |
| "epoch": 1.5233415233415233, | |
| "grad_norm": 0.8167658605627112, | |
| "learning_rate": 3.720741136221491e-05, | |
| "loss": 0.2093, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2142532467842102, | |
| "step": 620, | |
| "valid_targets_mean": 1292.1, | |
| "valid_targets_min": 704 | |
| }, | |
| { | |
| "epoch": 1.5356265356265357, | |
| "grad_norm": 0.862827915655825, | |
| "learning_rate": 3.713409223073636e-05, | |
| "loss": 0.2175, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21128416061401367, | |
| "step": 625, | |
| "valid_targets_mean": 1296.1, | |
| "valid_targets_min": 501 | |
| }, | |
| { | |
| "epoch": 1.547911547911548, | |
| "grad_norm": 0.8137378186782543, | |
| "learning_rate": 3.705989723079214e-05, | |
| "loss": 0.2188, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22958746552467346, | |
| "step": 630, | |
| "valid_targets_mean": 1526.4, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 1.5601965601965602, | |
| "grad_norm": 0.8249041342154185, | |
| "learning_rate": 3.698483015511665e-05, | |
| "loss": 0.2115, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21446937322616577, | |
| "step": 635, | |
| "valid_targets_mean": 1382.8, | |
| "valid_targets_min": 608 | |
| }, | |
| { | |
| "epoch": 1.5724815724815726, | |
| "grad_norm": 0.7761727919338764, | |
| "learning_rate": 3.690889484102344e-05, | |
| "loss": 0.2082, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1898980289697647, | |
| "step": 640, | |
| "valid_targets_mean": 1198.1, | |
| "valid_targets_min": 587 | |
| }, | |
| { | |
| "epoch": 1.5847665847665846, | |
| "grad_norm": 0.85132599232256, | |
| "learning_rate": 3.683209517020908e-05, | |
| "loss": 0.2111, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2267095148563385, | |
| "step": 645, | |
| "valid_targets_mean": 1355.8, | |
| "valid_targets_min": 678 | |
| }, | |
| { | |
| "epoch": 1.597051597051597, | |
| "grad_norm": 0.8632075917415805, | |
| "learning_rate": 3.675443506855473e-05, | |
| "loss": 0.2149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2519699037075043, | |
| "step": 650, | |
| "valid_targets_mean": 1355.8, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 1.6093366093366095, | |
| "grad_norm": 0.8029636768151777, | |
| "learning_rate": 3.6675918505925456e-05, | |
| "loss": 0.2128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19434180855751038, | |
| "step": 655, | |
| "valid_targets_mean": 1209.3, | |
| "valid_targets_min": 622 | |
| }, | |
| { | |
| "epoch": 1.6216216216216215, | |
| "grad_norm": 0.8533947419088043, | |
| "learning_rate": 3.6596549495967276e-05, | |
| "loss": 0.2128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22300590574741364, | |
| "step": 660, | |
| "valid_targets_mean": 1284.0, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 1.633906633906634, | |
| "grad_norm": 0.7643303692582221, | |
| "learning_rate": 3.651633209590202e-05, | |
| "loss": 0.213, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18270307779312134, | |
| "step": 665, | |
| "valid_targets_mean": 1283.6, | |
| "valid_targets_min": 532 | |
| }, | |
| { | |
| "epoch": 1.6461916461916462, | |
| "grad_norm": 0.7807796150123283, | |
| "learning_rate": 3.6435270406319914e-05, | |
| "loss": 0.2281, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21742983162403107, | |
| "step": 670, | |
| "valid_targets_mean": 1435.8, | |
| "valid_targets_min": 691 | |
| }, | |
| { | |
| "epoch": 1.6584766584766584, | |
| "grad_norm": 0.8076168248855204, | |
| "learning_rate": 3.635336857096997e-05, | |
| "loss": 0.2102, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20645713806152344, | |
| "step": 675, | |
| "valid_targets_mean": 1312.5, | |
| "valid_targets_min": 863 | |
| }, | |
| { | |
| "epoch": 1.6707616707616708, | |
| "grad_norm": 0.7686324259536815, | |
| "learning_rate": 3.627063077654815e-05, | |
| "loss": 0.2117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2024710476398468, | |
| "step": 680, | |
| "valid_targets_mean": 1363.4, | |
| "valid_targets_min": 538 | |
| }, | |
| { | |
| "epoch": 1.683046683046683, | |
| "grad_norm": 0.8120597616491396, | |
| "learning_rate": 3.618706125248337e-05, | |
| "loss": 0.2057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20566058158874512, | |
| "step": 685, | |
| "valid_targets_mean": 1409.5, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 1.6953316953316953, | |
| "grad_norm": 0.7975177403301514, | |
| "learning_rate": 3.6102664270721275e-05, | |
| "loss": 0.2171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21366316080093384, | |
| "step": 690, | |
| "valid_targets_mean": 1345.3, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 1.7076167076167077, | |
| "grad_norm": 0.693128955960592, | |
| "learning_rate": 3.601744414550589e-05, | |
| "loss": 0.2072, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.203857883810997, | |
| "step": 695, | |
| "valid_targets_mean": 1457.3, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 1.71990171990172, | |
| "grad_norm": 0.8655978074299355, | |
| "learning_rate": 3.593140523315906e-05, | |
| "loss": 0.2153, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21074074506759644, | |
| "step": 700, | |
| "valid_targets_mean": 1263.1, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 1.7321867321867321, | |
| "grad_norm": 0.8174739337432844, | |
| "learning_rate": 3.584455193185778e-05, | |
| "loss": 0.2155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2070631980895996, | |
| "step": 705, | |
| "valid_targets_mean": 1293.8, | |
| "valid_targets_min": 641 | |
| }, | |
| { | |
| "epoch": 1.7444717444717446, | |
| "grad_norm": 0.821190279821601, | |
| "learning_rate": 3.575688868140933e-05, | |
| "loss": 0.22, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20903602242469788, | |
| "step": 710, | |
| "valid_targets_mean": 1464.4, | |
| "valid_targets_min": 796 | |
| }, | |
| { | |
| "epoch": 1.7567567567567568, | |
| "grad_norm": 0.7124088474881533, | |
| "learning_rate": 3.566841996302438e-05, | |
| "loss": 0.2083, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2009987235069275, | |
| "step": 715, | |
| "valid_targets_mean": 1362.8, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 1.769041769041769, | |
| "grad_norm": 0.7766872594933402, | |
| "learning_rate": 3.557915029908787e-05, | |
| "loss": 0.2159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22496528923511505, | |
| "step": 720, | |
| "valid_targets_mean": 1335.9, | |
| "valid_targets_min": 775 | |
| }, | |
| { | |
| "epoch": 1.7813267813267815, | |
| "grad_norm": 0.778311753577005, | |
| "learning_rate": 3.548908425292784e-05, | |
| "loss": 0.2102, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20054762065410614, | |
| "step": 725, | |
| "valid_targets_mean": 1222.9, | |
| "valid_targets_min": 765 | |
| }, | |
| { | |
| "epoch": 1.7936117936117935, | |
| "grad_norm": 0.7697237115798328, | |
| "learning_rate": 3.5398226428582165e-05, | |
| "loss": 0.2051, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18973074853420258, | |
| "step": 730, | |
| "valid_targets_mean": 1225.9, | |
| "valid_targets_min": 576 | |
| }, | |
| { | |
| "epoch": 1.805896805896806, | |
| "grad_norm": 0.7435081743751965, | |
| "learning_rate": 3.530658147056321e-05, | |
| "loss": 0.2074, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21077962219715118, | |
| "step": 735, | |
| "valid_targets_mean": 1544.0, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 1.8181818181818183, | |
| "grad_norm": 0.8072606835461681, | |
| "learning_rate": 3.521415406362041e-05, | |
| "loss": 0.2111, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22379997372627258, | |
| "step": 740, | |
| "valid_targets_mean": 1295.1, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 1.8304668304668303, | |
| "grad_norm": 0.8303839843270181, | |
| "learning_rate": 3.512094893250076e-05, | |
| "loss": 0.2112, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19904112815856934, | |
| "step": 745, | |
| "valid_targets_mean": 1146.8, | |
| "valid_targets_min": 592 | |
| }, | |
| { | |
| "epoch": 1.8427518427518428, | |
| "grad_norm": 0.806403930167987, | |
| "learning_rate": 3.5026970841707366e-05, | |
| "loss": 0.2095, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22263360023498535, | |
| "step": 750, | |
| "valid_targets_mean": 1444.1, | |
| "valid_targets_min": 936 | |
| }, | |
| { | |
| "epoch": 1.855036855036855, | |
| "grad_norm": 0.832539093115592, | |
| "learning_rate": 3.493222459525579e-05, | |
| "loss": 0.1997, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2039949595928192, | |
| "step": 755, | |
| "valid_targets_mean": 1232.6, | |
| "valid_targets_min": 750 | |
| }, | |
| { | |
| "epoch": 1.8673218673218672, | |
| "grad_norm": 0.7634964231003126, | |
| "learning_rate": 3.483671503642858e-05, | |
| "loss": 0.2061, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20757383108139038, | |
| "step": 760, | |
| "valid_targets_mean": 1327.8, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 1.8796068796068797, | |
| "grad_norm": 0.734644093470408, | |
| "learning_rate": 3.474044704752761e-05, | |
| "loss": 0.2108, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20907098054885864, | |
| "step": 765, | |
| "valid_targets_mean": 1293.1, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 1.8918918918918919, | |
| "grad_norm": 0.7599652457636882, | |
| "learning_rate": 3.464342554962454e-05, | |
| "loss": 0.1995, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20856548845767975, | |
| "step": 770, | |
| "valid_targets_mean": 1729.1, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 1.904176904176904, | |
| "grad_norm": 0.7360153071481739, | |
| "learning_rate": 3.4545655502309254e-05, | |
| "loss": 0.2031, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18852345645427704, | |
| "step": 775, | |
| "valid_targets_mean": 1302.1, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 1.9164619164619165, | |
| "grad_norm": 0.7714929717750981, | |
| "learning_rate": 3.444714190343633e-05, | |
| "loss": 0.2085, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2067737728357315, | |
| "step": 780, | |
| "valid_targets_mean": 1459.2, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 1.9287469287469288, | |
| "grad_norm": 0.7966394203579434, | |
| "learning_rate": 3.434788978886957e-05, | |
| "loss": 0.2102, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21493887901306152, | |
| "step": 785, | |
| "valid_targets_mean": 1388.4, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 1.941031941031941, | |
| "grad_norm": 0.7664698984697156, | |
| "learning_rate": 3.424790423222455e-05, | |
| "loss": 0.1962, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2068059891462326, | |
| "step": 790, | |
| "valid_targets_mean": 1329.3, | |
| "valid_targets_min": 578 | |
| }, | |
| { | |
| "epoch": 1.9533169533169534, | |
| "grad_norm": 0.7879676439663363, | |
| "learning_rate": 3.414719034460928e-05, | |
| "loss": 0.2001, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2054266482591629, | |
| "step": 795, | |
| "valid_targets_mean": 1332.4, | |
| "valid_targets_min": 854 | |
| }, | |
| { | |
| "epoch": 1.9656019656019657, | |
| "grad_norm": 0.7910168582778149, | |
| "learning_rate": 3.404575327436294e-05, | |
| "loss": 0.2026, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19492971897125244, | |
| "step": 800, | |
| "valid_targets_mean": 1276.2, | |
| "valid_targets_min": 620 | |
| }, | |
| { | |
| "epoch": 1.9778869778869779, | |
| "grad_norm": 0.8143146289838762, | |
| "learning_rate": 3.3943598206792665e-05, | |
| "loss": 0.1994, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19378644227981567, | |
| "step": 805, | |
| "valid_targets_mean": 1181.9, | |
| "valid_targets_min": 665 | |
| }, | |
| { | |
| "epoch": 1.9901719901719903, | |
| "grad_norm": 0.7759024846717739, | |
| "learning_rate": 3.384073036390857e-05, | |
| "loss": 0.2021, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20919588208198547, | |
| "step": 810, | |
| "valid_targets_mean": 1480.4, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 2.0024570024570023, | |
| "grad_norm": 0.8889852312804324, | |
| "learning_rate": 3.373715500415667e-05, | |
| "loss": 0.203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17950613796710968, | |
| "step": 815, | |
| "valid_targets_mean": 1413.7, | |
| "valid_targets_min": 758 | |
| }, | |
| { | |
| "epoch": 2.0147420147420148, | |
| "grad_norm": 0.8529043749791839, | |
| "learning_rate": 3.363287742215023e-05, | |
| "loss": 0.1742, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1715715229511261, | |
| "step": 820, | |
| "valid_targets_mean": 1223.9, | |
| "valid_targets_min": 806 | |
| }, | |
| { | |
| "epoch": 2.027027027027027, | |
| "grad_norm": 0.861295844522495, | |
| "learning_rate": 3.352790294839898e-05, | |
| "loss": 0.1838, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17608895897865295, | |
| "step": 825, | |
| "valid_targets_mean": 1208.3, | |
| "valid_targets_min": 608 | |
| }, | |
| { | |
| "epoch": 2.039312039312039, | |
| "grad_norm": 0.8658818114720375, | |
| "learning_rate": 3.3422236949036726e-05, | |
| "loss": 0.1778, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18546447157859802, | |
| "step": 830, | |
| "valid_targets_mean": 1115.2, | |
| "valid_targets_min": 527 | |
| }, | |
| { | |
| "epoch": 2.0515970515970516, | |
| "grad_norm": 0.8483551528421704, | |
| "learning_rate": 3.331588482554697e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17613749206066132, | |
| "step": 835, | |
| "valid_targets_mean": 1319.5, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 2.063882063882064, | |
| "grad_norm": 0.7743600766705527, | |
| "learning_rate": 3.320885201448684e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17434000968933105, | |
| "step": 840, | |
| "valid_targets_mean": 1287.9, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 2.076167076167076, | |
| "grad_norm": 0.8493923341230093, | |
| "learning_rate": 3.310114398720917e-05, | |
| "loss": 0.1824, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17663569748401642, | |
| "step": 845, | |
| "valid_targets_mean": 1190.2, | |
| "valid_targets_min": 713 | |
| }, | |
| { | |
| "epoch": 2.0884520884520885, | |
| "grad_norm": 0.8478553827027294, | |
| "learning_rate": 3.299276624958281e-05, | |
| "loss": 0.1849, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19623138010501862, | |
| "step": 850, | |
| "valid_targets_mean": 1453.6, | |
| "valid_targets_min": 672 | |
| }, | |
| { | |
| "epoch": 2.100737100737101, | |
| "grad_norm": 0.8011647896279039, | |
| "learning_rate": 3.288372434171116e-05, | |
| "loss": 0.1776, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17543333768844604, | |
| "step": 855, | |
| "valid_targets_mean": 1394.6, | |
| "valid_targets_min": 742 | |
| }, | |
| { | |
| "epoch": 2.113022113022113, | |
| "grad_norm": 0.8904716311684531, | |
| "learning_rate": 3.2774023837648986e-05, | |
| "loss": 0.1849, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20401468873023987, | |
| "step": 860, | |
| "valid_targets_mean": 1273.4, | |
| "valid_targets_min": 591 | |
| }, | |
| { | |
| "epoch": 2.1253071253071254, | |
| "grad_norm": 0.7994042104976049, | |
| "learning_rate": 3.26636703451175e-05, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16975076496601105, | |
| "step": 865, | |
| "valid_targets_mean": 1109.5, | |
| "valid_targets_min": 536 | |
| }, | |
| { | |
| "epoch": 2.1375921375921374, | |
| "grad_norm": 0.815977947868972, | |
| "learning_rate": 3.2552669505217646e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18750914931297302, | |
| "step": 870, | |
| "valid_targets_mean": 1309.8, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 2.14987714987715, | |
| "grad_norm": 0.8080552902212362, | |
| "learning_rate": 3.24410269921418e-05, | |
| "loss": 0.1792, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16167230904102325, | |
| "step": 875, | |
| "valid_targets_mean": 1132.6, | |
| "valid_targets_min": 807 | |
| }, | |
| { | |
| "epoch": 2.1621621621621623, | |
| "grad_norm": 0.8331786616540037, | |
| "learning_rate": 3.232874851288367e-05, | |
| "loss": 0.1824, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1691512167453766, | |
| "step": 880, | |
| "valid_targets_mean": 1280.0, | |
| "valid_targets_min": 706 | |
| }, | |
| { | |
| "epoch": 2.1744471744471743, | |
| "grad_norm": 0.9322586407992784, | |
| "learning_rate": 3.221583980694659e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1737692803144455, | |
| "step": 885, | |
| "valid_targets_mean": 1235.9, | |
| "valid_targets_min": 742 | |
| }, | |
| { | |
| "epoch": 2.1867321867321867, | |
| "grad_norm": 0.7957232692494968, | |
| "learning_rate": 3.21023066460501e-05, | |
| "loss": 0.1747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18011856079101562, | |
| "step": 890, | |
| "valid_targets_mean": 1525.5, | |
| "valid_targets_min": 826 | |
| }, | |
| { | |
| "epoch": 2.199017199017199, | |
| "grad_norm": 0.836674749989076, | |
| "learning_rate": 3.198815483383492e-05, | |
| "loss": 0.1812, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17540715634822845, | |
| "step": 895, | |
| "valid_targets_mean": 1178.8, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 2.211302211302211, | |
| "grad_norm": 0.8699604703326298, | |
| "learning_rate": 3.1873390205566295e-05, | |
| "loss": 0.1836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.194318026304245, | |
| "step": 900, | |
| "valid_targets_mean": 1225.6, | |
| "valid_targets_min": 647 | |
| }, | |
| { | |
| "epoch": 2.2235872235872236, | |
| "grad_norm": 0.7976579724339753, | |
| "learning_rate": 3.175801862783565e-05, | |
| "loss": 0.1826, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19349230825901031, | |
| "step": 905, | |
| "valid_targets_mean": 1512.1, | |
| "valid_targets_min": 762 | |
| }, | |
| { | |
| "epoch": 2.235872235872236, | |
| "grad_norm": 0.8510536664782925, | |
| "learning_rate": 3.164204599826077e-05, | |
| "loss": 0.1807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1855241358280182, | |
| "step": 910, | |
| "valid_targets_mean": 1290.1, | |
| "valid_targets_min": 707 | |
| }, | |
| { | |
| "epoch": 2.248157248157248, | |
| "grad_norm": 1.2961464486220478, | |
| "learning_rate": 3.1525478245184245e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20424135029315948, | |
| "step": 915, | |
| "valid_targets_mean": 1385.6, | |
| "valid_targets_min": 760 | |
| }, | |
| { | |
| "epoch": 2.2604422604422605, | |
| "grad_norm": 0.723340082244006, | |
| "learning_rate": 3.140832132737051e-05, | |
| "loss": 0.1761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17882639169692993, | |
| "step": 920, | |
| "valid_targets_mean": 1393.4, | |
| "valid_targets_min": 745 | |
| }, | |
| { | |
| "epoch": 2.2727272727272725, | |
| "grad_norm": 0.7235527044863573, | |
| "learning_rate": 3.129058123370116e-05, | |
| "loss": 0.1817, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17998242378234863, | |
| "step": 925, | |
| "valid_targets_mean": 1463.0, | |
| "valid_targets_min": 806 | |
| }, | |
| { | |
| "epoch": 2.285012285012285, | |
| "grad_norm": 0.8449406721479917, | |
| "learning_rate": 3.117226398286887e-05, | |
| "loss": 0.1759, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17853769659996033, | |
| "step": 930, | |
| "valid_targets_mean": 1234.6, | |
| "valid_targets_min": 637 | |
| }, | |
| { | |
| "epoch": 2.2972972972972974, | |
| "grad_norm": 0.7732211800284947, | |
| "learning_rate": 3.105337562306968e-05, | |
| "loss": 0.1872, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.191191628575325, | |
| "step": 935, | |
| "valid_targets_mean": 1597.8, | |
| "valid_targets_min": 939 | |
| }, | |
| { | |
| "epoch": 2.30958230958231, | |
| "grad_norm": 0.8005078939965097, | |
| "learning_rate": 3.0933922231693854e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1898137778043747, | |
| "step": 940, | |
| "valid_targets_mean": 1381.9, | |
| "valid_targets_min": 768 | |
| }, | |
| { | |
| "epoch": 2.321867321867322, | |
| "grad_norm": 0.9221082841518858, | |
| "learning_rate": 3.08139099150152e-05, | |
| "loss": 0.1886, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1998477727174759, | |
| "step": 945, | |
| "valid_targets_mean": 1377.6, | |
| "valid_targets_min": 902 | |
| }, | |
| { | |
| "epoch": 2.3341523341523343, | |
| "grad_norm": 0.7707479063679702, | |
| "learning_rate": 3.069334480787893e-05, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17489567399024963, | |
| "step": 950, | |
| "valid_targets_mean": 1442.1, | |
| "valid_targets_min": 564 | |
| }, | |
| { | |
| "epoch": 2.3464373464373462, | |
| "grad_norm": 0.8256834025961682, | |
| "learning_rate": 3.057223307338806e-05, | |
| "loss": 0.1837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18947067856788635, | |
| "step": 955, | |
| "valid_targets_mean": 1536.8, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 2.3587223587223587, | |
| "grad_norm": 0.8531224678622473, | |
| "learning_rate": 3.0450580902588346e-05, | |
| "loss": 0.1773, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1829451620578766, | |
| "step": 960, | |
| "valid_targets_mean": 1313.1, | |
| "valid_targets_min": 777 | |
| }, | |
| { | |
| "epoch": 2.371007371007371, | |
| "grad_norm": 0.8371259257645897, | |
| "learning_rate": 3.032839451415182e-05, | |
| "loss": 0.1835, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17725329101085663, | |
| "step": 965, | |
| "valid_targets_mean": 1203.5, | |
| "valid_targets_min": 457 | |
| }, | |
| { | |
| "epoch": 2.383292383292383, | |
| "grad_norm": 0.717363860123874, | |
| "learning_rate": 3.0205680154058904e-05, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1754624843597412, | |
| "step": 970, | |
| "valid_targets_mean": 1550.9, | |
| "valid_targets_min": 697 | |
| }, | |
| { | |
| "epoch": 2.3955773955773956, | |
| "grad_norm": 0.8214153230293283, | |
| "learning_rate": 3.0082444095279117e-05, | |
| "loss": 0.1802, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20176392793655396, | |
| "step": 975, | |
| "valid_targets_mean": 1422.4, | |
| "valid_targets_min": 708 | |
| }, | |
| { | |
| "epoch": 2.407862407862408, | |
| "grad_norm": 0.8026368482781924, | |
| "learning_rate": 2.9958692637450406e-05, | |
| "loss": 0.1767, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16252771019935608, | |
| "step": 980, | |
| "valid_targets_mean": 1248.3, | |
| "valid_targets_min": 643 | |
| }, | |
| { | |
| "epoch": 2.42014742014742, | |
| "grad_norm": 0.8815336408634351, | |
| "learning_rate": 2.983443210655714e-05, | |
| "loss": 0.1752, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17958936095237732, | |
| "step": 985, | |
| "valid_targets_mean": 1289.1, | |
| "valid_targets_min": 682 | |
| }, | |
| { | |
| "epoch": 2.4324324324324325, | |
| "grad_norm": 0.7836836708413403, | |
| "learning_rate": 2.9709668854606706e-05, | |
| "loss": 0.1896, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1911127269268036, | |
| "step": 990, | |
| "valid_targets_mean": 1440.7, | |
| "valid_targets_min": 614 | |
| }, | |
| { | |
| "epoch": 2.444717444717445, | |
| "grad_norm": 0.7554743565043565, | |
| "learning_rate": 2.9584409259304828e-05, | |
| "loss": 0.1845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1768302172422409, | |
| "step": 995, | |
| "valid_targets_mean": 1373.0, | |
| "valid_targets_min": 770 | |
| }, | |
| { | |
| "epoch": 2.457002457002457, | |
| "grad_norm": 0.7711572760011336, | |
| "learning_rate": 2.945865972372954e-05, | |
| "loss": 0.1806, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1825973093509674, | |
| "step": 1000, | |
| "valid_targets_mean": 1467.0, | |
| "valid_targets_min": 958 | |
| }, | |
| { | |
| "epoch": 2.4692874692874693, | |
| "grad_norm": 0.7763115308115468, | |
| "learning_rate": 2.9332426676003858e-05, | |
| "loss": 0.1843, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18232324719429016, | |
| "step": 1005, | |
| "valid_targets_mean": 1405.0, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 2.4815724815724813, | |
| "grad_norm": 0.8101073760100002, | |
| "learning_rate": 2.920571656896722e-05, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17708787322044373, | |
| "step": 1010, | |
| "valid_targets_mean": 1309.1, | |
| "valid_targets_min": 782 | |
| }, | |
| { | |
| "epoch": 2.493857493857494, | |
| "grad_norm": 0.7835475920518248, | |
| "learning_rate": 2.907853587984558e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18842971324920654, | |
| "step": 1015, | |
| "valid_targets_mean": 1459.9, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 2.506142506142506, | |
| "grad_norm": 0.7907834667766721, | |
| "learning_rate": 2.8950891109920333e-05, | |
| "loss": 0.1807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18541795015335083, | |
| "step": 1020, | |
| "valid_targets_mean": 1363.8, | |
| "valid_targets_min": 770 | |
| }, | |
| { | |
| "epoch": 2.5184275184275187, | |
| "grad_norm": 0.7806222134145253, | |
| "learning_rate": 2.882278878419597e-05, | |
| "loss": 0.1765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17934384942054749, | |
| "step": 1025, | |
| "valid_targets_mean": 1293.1, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 2.5307125307125307, | |
| "grad_norm": 0.7736823415875187, | |
| "learning_rate": 2.8694235451066538e-05, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18719598650932312, | |
| "step": 1030, | |
| "valid_targets_mean": 1573.0, | |
| "valid_targets_min": 856 | |
| }, | |
| { | |
| "epoch": 2.542997542997543, | |
| "grad_norm": 0.7531581920130256, | |
| "learning_rate": 2.8565237681980876e-05, | |
| "loss": 0.1743, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1793087124824524, | |
| "step": 1035, | |
| "valid_targets_mean": 1425.1, | |
| "valid_targets_min": 836 | |
| }, | |
| { | |
| "epoch": 2.555282555282555, | |
| "grad_norm": 0.7399212256403678, | |
| "learning_rate": 2.843580207110672e-05, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16769655048847198, | |
| "step": 1040, | |
| "valid_targets_mean": 1410.6, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 2.5675675675675675, | |
| "grad_norm": 0.741628811394016, | |
| "learning_rate": 2.830593523499361e-05, | |
| "loss": 0.1815, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18228626251220703, | |
| "step": 1045, | |
| "valid_targets_mean": 1439.1, | |
| "valid_targets_min": 771 | |
| }, | |
| { | |
| "epoch": 2.57985257985258, | |
| "grad_norm": 0.7969152441113768, | |
| "learning_rate": 2.8175643812234627e-05, | |
| "loss": 0.1798, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17840053141117096, | |
| "step": 1050, | |
| "valid_targets_mean": 1492.2, | |
| "valid_targets_min": 826 | |
| }, | |
| { | |
| "epoch": 2.592137592137592, | |
| "grad_norm": 0.8367342853086918, | |
| "learning_rate": 2.8044934463127108e-05, | |
| "loss": 0.1815, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18805266916751862, | |
| "step": 1055, | |
| "valid_targets_mean": 1261.8, | |
| "valid_targets_min": 759 | |
| }, | |
| { | |
| "epoch": 2.6044226044226044, | |
| "grad_norm": 0.7677291227603371, | |
| "learning_rate": 2.7913813869332112e-05, | |
| "loss": 0.1787, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17599162459373474, | |
| "step": 1060, | |
| "valid_targets_mean": 1296.4, | |
| "valid_targets_min": 654 | |
| }, | |
| { | |
| "epoch": 2.616707616707617, | |
| "grad_norm": 0.7992239103679709, | |
| "learning_rate": 2.7782288733532915e-05, | |
| "loss": 0.1795, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17146100103855133, | |
| "step": 1065, | |
| "valid_targets_mean": 1234.5, | |
| "valid_targets_min": 671 | |
| }, | |
| { | |
| "epoch": 2.628992628992629, | |
| "grad_norm": 0.8304316096602254, | |
| "learning_rate": 2.7650365779092346e-05, | |
| "loss": 0.177, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17182031273841858, | |
| "step": 1070, | |
| "valid_targets_mean": 1255.4, | |
| "valid_targets_min": 632 | |
| }, | |
| { | |
| "epoch": 2.6412776412776413, | |
| "grad_norm": 0.805604710666297, | |
| "learning_rate": 2.751805174970912e-05, | |
| "loss": 0.1825, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19578979909420013, | |
| "step": 1075, | |
| "valid_targets_mean": 1437.3, | |
| "valid_targets_min": 724 | |
| }, | |
| { | |
| "epoch": 2.6535626535626538, | |
| "grad_norm": 0.7937637333321657, | |
| "learning_rate": 2.7385353409073093e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18653550744056702, | |
| "step": 1080, | |
| "valid_targets_mean": 1224.4, | |
| "valid_targets_min": 626 | |
| }, | |
| { | |
| "epoch": 2.6658476658476657, | |
| "grad_norm": 0.8392997092676029, | |
| "learning_rate": 2.725227754051953e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18633800745010376, | |
| "step": 1085, | |
| "valid_targets_mean": 1351.3, | |
| "valid_targets_min": 711 | |
| }, | |
| { | |
| "epoch": 2.678132678132678, | |
| "grad_norm": 0.82826418253686, | |
| "learning_rate": 2.711883094668234e-05, | |
| "loss": 0.1772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16905131936073303, | |
| "step": 1090, | |
| "valid_targets_mean": 1228.1, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 2.69041769041769, | |
| "grad_norm": 0.7749583275108434, | |
| "learning_rate": 2.698502044914633e-05, | |
| "loss": 0.1735, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18027430772781372, | |
| "step": 1095, | |
| "valid_targets_mean": 1298.4, | |
| "valid_targets_min": 814 | |
| }, | |
| { | |
| "epoch": 2.7027027027027026, | |
| "grad_norm": 0.8377694942716309, | |
| "learning_rate": 2.685085288809853e-05, | |
| "loss": 0.1817, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18739664554595947, | |
| "step": 1100, | |
| "valid_targets_mean": 1326.9, | |
| "valid_targets_min": 763 | |
| }, | |
| { | |
| "epoch": 2.714987714987715, | |
| "grad_norm": 0.7605797734549805, | |
| "learning_rate": 2.671633512197848e-05, | |
| "loss": 0.1831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1845276653766632, | |
| "step": 1105, | |
| "valid_targets_mean": 1511.4, | |
| "valid_targets_min": 1088 | |
| }, | |
| { | |
| "epoch": 2.7272727272727275, | |
| "grad_norm": 0.734192726408221, | |
| "learning_rate": 2.658147402712768e-05, | |
| "loss": 0.1798, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1821923404932022, | |
| "step": 1110, | |
| "valid_targets_mean": 1476.5, | |
| "valid_targets_min": 968 | |
| }, | |
| { | |
| "epoch": 2.7395577395577395, | |
| "grad_norm": 0.933923702321532, | |
| "learning_rate": 2.6446276497438064e-05, | |
| "loss": 0.1772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1841639280319214, | |
| "step": 1115, | |
| "valid_targets_mean": 1164.2, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 2.751842751842752, | |
| "grad_norm": 0.8577205410075446, | |
| "learning_rate": 2.6310749443999593e-05, | |
| "loss": 0.1853, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19331184029579163, | |
| "step": 1120, | |
| "valid_targets_mean": 1148.2, | |
| "valid_targets_min": 782 | |
| }, | |
| { | |
| "epoch": 2.764127764127764, | |
| "grad_norm": 0.8240964794783217, | |
| "learning_rate": 2.617489979474699e-05, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1863320767879486, | |
| "step": 1125, | |
| "valid_targets_mean": 1221.2, | |
| "valid_targets_min": 620 | |
| }, | |
| { | |
| "epoch": 2.7764127764127764, | |
| "grad_norm": 0.6884072653384222, | |
| "learning_rate": 2.6038734494105562e-05, | |
| "loss": 0.1832, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19115372002124786, | |
| "step": 1130, | |
| "valid_targets_mean": 1645.6, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 2.788697788697789, | |
| "grad_norm": 0.8365132438504224, | |
| "learning_rate": 2.590226050263625e-05, | |
| "loss": 0.1738, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17121455073356628, | |
| "step": 1135, | |
| "valid_targets_mean": 1232.6, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 2.800982800982801, | |
| "grad_norm": 0.8932011031329712, | |
| "learning_rate": 2.5765484796679768e-05, | |
| "loss": 0.1766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1942731738090515, | |
| "step": 1140, | |
| "valid_targets_mean": 1251.2, | |
| "valid_targets_min": 674 | |
| }, | |
| { | |
| "epoch": 2.8132678132678133, | |
| "grad_norm": 0.8005242041150714, | |
| "learning_rate": 2.5628414368000035e-05, | |
| "loss": 0.1836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1775723397731781, | |
| "step": 1145, | |
| "valid_targets_mean": 1480.1, | |
| "valid_targets_min": 784 | |
| }, | |
| { | |
| "epoch": 2.8255528255528253, | |
| "grad_norm": 0.7503467618496086, | |
| "learning_rate": 2.5491056223426746e-05, | |
| "loss": 0.1848, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18719075620174408, | |
| "step": 1150, | |
| "valid_targets_mean": 1456.2, | |
| "valid_targets_min": 873 | |
| }, | |
| { | |
| "epoch": 2.8378378378378377, | |
| "grad_norm": 0.8501440933671442, | |
| "learning_rate": 2.5353417384497166e-05, | |
| "loss": 0.1798, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1862514168024063, | |
| "step": 1155, | |
| "valid_targets_mean": 1129.2, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 2.85012285012285, | |
| "grad_norm": 0.8153346046411876, | |
| "learning_rate": 2.5215504887097243e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20934267342090607, | |
| "step": 1160, | |
| "valid_targets_mean": 1318.6, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 2.8624078624078626, | |
| "grad_norm": 0.7763002556957465, | |
| "learning_rate": 2.5077325781101918e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17470885813236237, | |
| "step": 1165, | |
| "valid_targets_mean": 1339.2, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 2.8746928746928746, | |
| "grad_norm": 0.8016048702566863, | |
| "learning_rate": 2.493888713001476e-05, | |
| "loss": 0.1793, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18066394329071045, | |
| "step": 1170, | |
| "valid_targets_mean": 1284.4, | |
| "valid_targets_min": 691 | |
| }, | |
| { | |
| "epoch": 2.886977886977887, | |
| "grad_norm": 0.7717525435471911, | |
| "learning_rate": 2.480019601060687e-05, | |
| "loss": 0.1789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1963619589805603, | |
| "step": 1175, | |
| "valid_targets_mean": 1382.4, | |
| "valid_targets_min": 830 | |
| }, | |
| { | |
| "epoch": 2.899262899262899, | |
| "grad_norm": 0.7784504210078957, | |
| "learning_rate": 2.4661259512555176e-05, | |
| "loss": 0.1811, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18142808973789215, | |
| "step": 1180, | |
| "valid_targets_mean": 1400.8, | |
| "valid_targets_min": 1080 | |
| }, | |
| { | |
| "epoch": 2.9115479115479115, | |
| "grad_norm": 0.7978518183157296, | |
| "learning_rate": 2.4522084738079933e-05, | |
| "loss": 0.1784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1911832094192505, | |
| "step": 1185, | |
| "valid_targets_mean": 1342.9, | |
| "valid_targets_min": 773 | |
| }, | |
| { | |
| "epoch": 2.923832923832924, | |
| "grad_norm": 0.7773097305802383, | |
| "learning_rate": 2.4382678801581762e-05, | |
| "loss": 0.1757, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16415411233901978, | |
| "step": 1190, | |
| "valid_targets_mean": 1346.6, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 2.9361179361179364, | |
| "grad_norm": 0.7149177870829416, | |
| "learning_rate": 2.4243048829277916e-05, | |
| "loss": 0.1846, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16982388496398926, | |
| "step": 1195, | |
| "valid_targets_mean": 1398.8, | |
| "valid_targets_min": 615 | |
| }, | |
| { | |
| "epoch": 2.9484029484029484, | |
| "grad_norm": 0.8133991765071973, | |
| "learning_rate": 2.410320195883802e-05, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1808069348335266, | |
| "step": 1200, | |
| "valid_targets_mean": 1261.6, | |
| "valid_targets_min": 576 | |
| }, | |
| { | |
| "epoch": 2.960687960687961, | |
| "grad_norm": 0.8035380696411945, | |
| "learning_rate": 2.396314533901918e-05, | |
| "loss": 0.1799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17814520001411438, | |
| "step": 1205, | |
| "valid_targets_mean": 1265.3, | |
| "valid_targets_min": 605 | |
| }, | |
| { | |
| "epoch": 2.972972972972973, | |
| "grad_norm": 0.8103346517230893, | |
| "learning_rate": 2.3822886129300603e-05, | |
| "loss": 0.1814, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18759378790855408, | |
| "step": 1210, | |
| "valid_targets_mean": 1393.0, | |
| "valid_targets_min": 898 | |
| }, | |
| { | |
| "epoch": 2.9852579852579852, | |
| "grad_norm": 0.9427441660740232, | |
| "learning_rate": 2.368243149951755e-05, | |
| "loss": 0.1833, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1856885552406311, | |
| "step": 1215, | |
| "valid_targets_mean": 1184.1, | |
| "valid_targets_min": 623 | |
| }, | |
| { | |
| "epoch": 2.9975429975429977, | |
| "grad_norm": 0.740828622122419, | |
| "learning_rate": 2.3541788629494865e-05, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17210790514945984, | |
| "step": 1220, | |
| "valid_targets_mean": 1348.9, | |
| "valid_targets_min": 635 | |
| }, | |
| { | |
| "epoch": 3.0098280098280097, | |
| "grad_norm": 0.7164578724146792, | |
| "learning_rate": 2.3400964708679944e-05, | |
| "loss": 0.1549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13863927125930786, | |
| "step": 1225, | |
| "valid_targets_mean": 1282.5, | |
| "valid_targets_min": 651 | |
| }, | |
| { | |
| "epoch": 3.022113022113022, | |
| "grad_norm": 0.8312096014116472, | |
| "learning_rate": 2.325996693577522e-05, | |
| "loss": 0.159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15467259287834167, | |
| "step": 1230, | |
| "valid_targets_mean": 1422.1, | |
| "valid_targets_min": 799 | |
| }, | |
| { | |
| "epoch": 3.0343980343980346, | |
| "grad_norm": 0.8375454565582668, | |
| "learning_rate": 2.311880251837019e-05, | |
| "loss": 0.1522, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15046252310276031, | |
| "step": 1235, | |
| "valid_targets_mean": 1260.6, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 3.0466830466830466, | |
| "grad_norm": 0.8180945627432928, | |
| "learning_rate": 2.2977478672572933e-05, | |
| "loss": 0.154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15951278805732727, | |
| "step": 1240, | |
| "valid_targets_mean": 1366.7, | |
| "valid_targets_min": 666 | |
| }, | |
| { | |
| "epoch": 3.058968058968059, | |
| "grad_norm": 1.4817440168716653, | |
| "learning_rate": 2.2836002622641297e-05, | |
| "loss": 0.1502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16168934106826782, | |
| "step": 1245, | |
| "valid_targets_mean": 1357.0, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 3.0712530712530715, | |
| "grad_norm": 0.789782000868707, | |
| "learning_rate": 2.269438160061354e-05, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14563557505607605, | |
| "step": 1250, | |
| "valid_targets_mean": 1295.6, | |
| "valid_targets_min": 810 | |
| }, | |
| { | |
| "epoch": 3.0835380835380835, | |
| "grad_norm": 0.7958342463964752, | |
| "learning_rate": 2.2552622845938698e-05, | |
| "loss": 0.1565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16075897216796875, | |
| "step": 1255, | |
| "valid_targets_mean": 1529.2, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 3.095823095823096, | |
| "grad_norm": 0.8394291652788792, | |
| "learning_rate": 2.2410733605106462e-05, | |
| "loss": 0.1581, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16368389129638672, | |
| "step": 1260, | |
| "valid_targets_mean": 1204.4, | |
| "valid_targets_min": 604 | |
| }, | |
| { | |
| "epoch": 3.108108108108108, | |
| "grad_norm": 0.8183277107345958, | |
| "learning_rate": 2.2268721131276805e-05, | |
| "loss": 0.1512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1558169722557068, | |
| "step": 1265, | |
| "valid_targets_mean": 1444.3, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 3.1203931203931203, | |
| "grad_norm": 0.8691016138218304, | |
| "learning_rate": 2.2126592683909154e-05, | |
| "loss": 0.1501, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15810123085975647, | |
| "step": 1270, | |
| "valid_targets_mean": 1228.3, | |
| "valid_targets_min": 614 | |
| }, | |
| { | |
| "epoch": 3.1326781326781328, | |
| "grad_norm": 0.7988341801875339, | |
| "learning_rate": 2.1984355528391342e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1563870906829834, | |
| "step": 1275, | |
| "valid_targets_mean": 1397.4, | |
| "valid_targets_min": 758 | |
| }, | |
| { | |
| "epoch": 3.1449631449631448, | |
| "grad_norm": 0.8981627797019425, | |
| "learning_rate": 2.1842016935668188e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.172378808259964, | |
| "step": 1280, | |
| "valid_targets_mean": 1452.6, | |
| "valid_targets_min": 746 | |
| }, | |
| { | |
| "epoch": 3.157248157248157, | |
| "grad_norm": 0.8274025893882145, | |
| "learning_rate": 2.169958418186982e-05, | |
| "loss": 0.1624, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1557457149028778, | |
| "step": 1285, | |
| "valid_targets_mean": 1294.4, | |
| "valid_targets_min": 738 | |
| }, | |
| { | |
| "epoch": 3.1695331695331697, | |
| "grad_norm": 0.8220493363877825, | |
| "learning_rate": 2.1557064547939754e-05, | |
| "loss": 0.1576, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14473125338554382, | |
| "step": 1290, | |
| "valid_targets_mean": 1266.2, | |
| "valid_targets_min": 810 | |
| }, | |
| { | |
| "epoch": 3.1818181818181817, | |
| "grad_norm": 0.7360182914990048, | |
| "learning_rate": 2.1414465319262666e-05, | |
| "loss": 0.1564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14730754494667053, | |
| "step": 1295, | |
| "valid_targets_mean": 1345.4, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 3.194103194103194, | |
| "grad_norm": 0.761094766359276, | |
| "learning_rate": 2.1271793785291997e-05, | |
| "loss": 0.151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1422240436077118, | |
| "step": 1300, | |
| "valid_targets_mean": 1342.6, | |
| "valid_targets_min": 602 | |
| }, | |
| { | |
| "epoch": 3.2063882063882065, | |
| "grad_norm": 0.8176967843846148, | |
| "learning_rate": 2.1129057239177337e-05, | |
| "loss": 0.1569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1573348492383957, | |
| "step": 1305, | |
| "valid_targets_mean": 1261.3, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 3.2186732186732185, | |
| "grad_norm": 0.8262216828068331, | |
| "learning_rate": 2.0986262977391577e-05, | |
| "loss": 0.154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15253204107284546, | |
| "step": 1310, | |
| "valid_targets_mean": 1228.0, | |
| "valid_targets_min": 728 | |
| }, | |
| { | |
| "epoch": 3.230958230958231, | |
| "grad_norm": 0.8436083601387745, | |
| "learning_rate": 2.084341829935796e-05, | |
| "loss": 0.1537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16099226474761963, | |
| "step": 1315, | |
| "valid_targets_mean": 1279.6, | |
| "valid_targets_min": 664 | |
| }, | |
| { | |
| "epoch": 3.2432432432432434, | |
| "grad_norm": 0.8107364765343557, | |
| "learning_rate": 2.0700530507076916e-05, | |
| "loss": 0.1469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14447104930877686, | |
| "step": 1320, | |
| "valid_targets_mean": 1242.2, | |
| "valid_targets_min": 807 | |
| }, | |
| { | |
| "epoch": 3.2555282555282554, | |
| "grad_norm": 0.8823201088197631, | |
| "learning_rate": 2.0557606904752833e-05, | |
| "loss": 0.1535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15442629158496857, | |
| "step": 1325, | |
| "valid_targets_mean": 1201.7, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 3.267813267813268, | |
| "grad_norm": 0.8163225673327624, | |
| "learning_rate": 2.0414654798420622e-05, | |
| "loss": 0.145, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15767508745193481, | |
| "step": 1330, | |
| "valid_targets_mean": 1397.9, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 3.2800982800982803, | |
| "grad_norm": 0.811563363023717, | |
| "learning_rate": 2.02716814955723e-05, | |
| "loss": 0.1551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14335688948631287, | |
| "step": 1335, | |
| "valid_targets_mean": 1254.7, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 3.2923832923832923, | |
| "grad_norm": 0.8679959811700956, | |
| "learning_rate": 2.0128694304783406e-05, | |
| "loss": 0.1568, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15623366832733154, | |
| "step": 1340, | |
| "valid_targets_mean": 1235.6, | |
| "valid_targets_min": 730 | |
| }, | |
| { | |
| "epoch": 3.3046683046683047, | |
| "grad_norm": 0.8311214799635258, | |
| "learning_rate": 1.9985700535339406e-05, | |
| "loss": 0.1591, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15222451090812683, | |
| "step": 1345, | |
| "valid_targets_mean": 1168.9, | |
| "valid_targets_min": 845 | |
| }, | |
| { | |
| "epoch": 3.3169533169533167, | |
| "grad_norm": 1.5033255139159805, | |
| "learning_rate": 1.984270749686207e-05, | |
| "loss": 0.155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13309861719608307, | |
| "step": 1350, | |
| "valid_targets_mean": 1288.8, | |
| "valid_targets_min": 536 | |
| }, | |
| { | |
| "epoch": 3.329238329238329, | |
| "grad_norm": 0.845526593227459, | |
| "learning_rate": 1.9699722498935786e-05, | |
| "loss": 0.1545, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1539156436920166, | |
| "step": 1355, | |
| "valid_targets_mean": 1236.1, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 3.3415233415233416, | |
| "grad_norm": 0.7782983419960925, | |
| "learning_rate": 1.9556752850733933e-05, | |
| "loss": 0.1525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14023908972740173, | |
| "step": 1360, | |
| "valid_targets_mean": 1311.5, | |
| "valid_targets_min": 706 | |
| }, | |
| { | |
| "epoch": 3.3538083538083536, | |
| "grad_norm": 0.9209403431550032, | |
| "learning_rate": 1.9413805860645242e-05, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17440494894981384, | |
| "step": 1365, | |
| "valid_targets_mean": 1309.7, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 3.366093366093366, | |
| "grad_norm": 0.9139513931631872, | |
| "learning_rate": 1.9270888835900165e-05, | |
| "loss": 0.1618, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16123470664024353, | |
| "step": 1370, | |
| "valid_targets_mean": 1285.9, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 3.3783783783783785, | |
| "grad_norm": 0.896363024278394, | |
| "learning_rate": 1.9128009082197417e-05, | |
| "loss": 0.1507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14758946001529694, | |
| "step": 1375, | |
| "valid_targets_mean": 1197.5, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 3.3906633906633905, | |
| "grad_norm": 0.7857845304860243, | |
| "learning_rate": 1.8985173903330428e-05, | |
| "loss": 0.1538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15281282365322113, | |
| "step": 1380, | |
| "valid_targets_mean": 1328.5, | |
| "valid_targets_min": 494 | |
| }, | |
| { | |
| "epoch": 3.402948402948403, | |
| "grad_norm": 0.7926903268037055, | |
| "learning_rate": 1.884239060081407e-05, | |
| "loss": 0.1454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14091375470161438, | |
| "step": 1385, | |
| "valid_targets_mean": 1239.3, | |
| "valid_targets_min": 808 | |
| }, | |
| { | |
| "epoch": 3.4152334152334154, | |
| "grad_norm": 0.8981590353433372, | |
| "learning_rate": 1.869966647351135e-05, | |
| "loss": 0.1553, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15038526058197021, | |
| "step": 1390, | |
| "valid_targets_mean": 1271.1, | |
| "valid_targets_min": 667 | |
| }, | |
| { | |
| "epoch": 3.4275184275184274, | |
| "grad_norm": 0.781660273156141, | |
| "learning_rate": 1.8557008817260343e-05, | |
| "loss": 0.156, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14509668946266174, | |
| "step": 1395, | |
| "valid_targets_mean": 1379.3, | |
| "valid_targets_min": 871 | |
| }, | |
| { | |
| "epoch": 3.43980343980344, | |
| "grad_norm": 0.8026197207302569, | |
| "learning_rate": 1.8414424924501222e-05, | |
| "loss": 0.1615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14742882549762726, | |
| "step": 1400, | |
| "valid_targets_mean": 1413.8, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 3.4520884520884523, | |
| "grad_norm": 0.8123747210830401, | |
| "learning_rate": 1.827192208390347e-05, | |
| "loss": 0.1488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14872382581233978, | |
| "step": 1405, | |
| "valid_targets_mean": 1303.1, | |
| "valid_targets_min": 532 | |
| }, | |
| { | |
| "epoch": 3.4643734643734643, | |
| "grad_norm": 0.8341496291476699, | |
| "learning_rate": 1.812950757999334e-05, | |
| "loss": 0.1573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1435258984565735, | |
| "step": 1410, | |
| "valid_targets_mean": 1249.4, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 3.4766584766584767, | |
| "grad_norm": 0.8201503217545428, | |
| "learning_rate": 1.7987188692781417e-05, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15042570233345032, | |
| "step": 1415, | |
| "valid_targets_mean": 1345.8, | |
| "valid_targets_min": 828 | |
| }, | |
| { | |
| "epoch": 3.488943488943489, | |
| "grad_norm": 0.7853127186556507, | |
| "learning_rate": 1.784497269739052e-05, | |
| "loss": 0.1546, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14669057726860046, | |
| "step": 1420, | |
| "valid_targets_mean": 1384.6, | |
| "valid_targets_min": 477 | |
| }, | |
| { | |
| "epoch": 3.501228501228501, | |
| "grad_norm": 0.8842882650050354, | |
| "learning_rate": 1.770286686368381e-05, | |
| "loss": 0.1527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15415064990520477, | |
| "step": 1425, | |
| "valid_targets_mean": 1237.2, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 3.5135135135135136, | |
| "grad_norm": 0.813011039927028, | |
| "learning_rate": 1.756087845589312e-05, | |
| "loss": 0.1573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14477777481079102, | |
| "step": 1430, | |
| "valid_targets_mean": 1236.9, | |
| "valid_targets_min": 599 | |
| }, | |
| { | |
| "epoch": 3.5257985257985256, | |
| "grad_norm": 0.7872126972705287, | |
| "learning_rate": 1.7419014732247683e-05, | |
| "loss": 0.1584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16638851165771484, | |
| "step": 1435, | |
| "valid_targets_mean": 1305.9, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 3.538083538083538, | |
| "grad_norm": 0.858779956828706, | |
| "learning_rate": 1.7277282944603047e-05, | |
| "loss": 0.1621, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16495858132839203, | |
| "step": 1440, | |
| "valid_targets_mean": 1440.2, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 3.5503685503685505, | |
| "grad_norm": 0.8728820790914598, | |
| "learning_rate": 1.713569033807041e-05, | |
| "loss": 0.1538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1443580836057663, | |
| "step": 1445, | |
| "valid_targets_mean": 1183.0, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 3.562653562653563, | |
| "grad_norm": 0.6901953229711102, | |
| "learning_rate": 1.6994244150646244e-05, | |
| "loss": 0.1491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.135453999042511, | |
| "step": 1450, | |
| "valid_targets_mean": 1456.6, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 3.574938574938575, | |
| "grad_norm": 0.7882552220256581, | |
| "learning_rate": 1.6852951612842278e-05, | |
| "loss": 0.1543, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15203431248664856, | |
| "step": 1455, | |
| "valid_targets_mean": 1427.3, | |
| "valid_targets_min": 922 | |
| }, | |
| { | |
| "epoch": 3.5872235872235874, | |
| "grad_norm": 0.802201807828957, | |
| "learning_rate": 1.671181994731595e-05, | |
| "loss": 0.1535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14671474695205688, | |
| "step": 1460, | |
| "valid_targets_mean": 1319.3, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 3.5995085995085994, | |
| "grad_norm": 0.855708476401229, | |
| "learning_rate": 1.6570856368501108e-05, | |
| "loss": 0.16, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1802348643541336, | |
| "step": 1465, | |
| "valid_targets_mean": 1353.6, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 3.611793611793612, | |
| "grad_norm": 0.8013086713913176, | |
| "learning_rate": 1.643006808223931e-05, | |
| "loss": 0.1513, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14372509717941284, | |
| "step": 1470, | |
| "valid_targets_mean": 1331.5, | |
| "valid_targets_min": 576 | |
| }, | |
| { | |
| "epoch": 3.6240786240786242, | |
| "grad_norm": 0.8833598909227577, | |
| "learning_rate": 1.6289462285411387e-05, | |
| "loss": 0.151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1556214988231659, | |
| "step": 1475, | |
| "valid_targets_mean": 1267.2, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 3.6363636363636362, | |
| "grad_norm": 0.8363554061363515, | |
| "learning_rate": 1.614904616556962e-05, | |
| "loss": 0.1453, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.144898921251297, | |
| "step": 1480, | |
| "valid_targets_mean": 1380.3, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 3.6486486486486487, | |
| "grad_norm": 0.8598319240396789, | |
| "learning_rate": 1.6008826900570294e-05, | |
| "loss": 0.1513, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14435847103595734, | |
| "step": 1485, | |
| "valid_targets_mean": 1139.2, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 3.6609336609336607, | |
| "grad_norm": 0.8495838229620162, | |
| "learning_rate": 1.586881165820675e-05, | |
| "loss": 0.1553, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15599597990512848, | |
| "step": 1490, | |
| "valid_targets_mean": 1231.2, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 3.673218673218673, | |
| "grad_norm": 0.7933104198530476, | |
| "learning_rate": 1.5729007595843037e-05, | |
| "loss": 0.1527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1428956240415573, | |
| "step": 1495, | |
| "valid_targets_mean": 1268.3, | |
| "valid_targets_min": 776 | |
| }, | |
| { | |
| "epoch": 3.6855036855036856, | |
| "grad_norm": 0.9253518944844057, | |
| "learning_rate": 1.5589421860047986e-05, | |
| "loss": 0.1556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15546691417694092, | |
| "step": 1500, | |
| "valid_targets_mean": 1399.1, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 3.697788697788698, | |
| "grad_norm": 0.8710193566221253, | |
| "learning_rate": 1.5450061586229903e-05, | |
| "loss": 0.1572, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15553036332130432, | |
| "step": 1505, | |
| "valid_targets_mean": 1175.8, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 3.71007371007371, | |
| "grad_norm": 0.8088094494880697, | |
| "learning_rate": 1.5310933898271864e-05, | |
| "loss": 0.1526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15640608966350555, | |
| "step": 1510, | |
| "valid_targets_mean": 1268.0, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 3.7223587223587224, | |
| "grad_norm": 0.927057394855993, | |
| "learning_rate": 1.5172045908167462e-05, | |
| "loss": 0.1498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15059244632720947, | |
| "step": 1515, | |
| "valid_targets_mean": 1296.2, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 3.7346437346437344, | |
| "grad_norm": 0.8799125344635643, | |
| "learning_rate": 1.5033404715657344e-05, | |
| "loss": 0.1527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14563912153244019, | |
| "step": 1520, | |
| "valid_targets_mean": 1261.3, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 3.746928746928747, | |
| "grad_norm": 0.9103297618631646, | |
| "learning_rate": 1.4895017407866217e-05, | |
| "loss": 0.158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.155216783285141, | |
| "step": 1525, | |
| "valid_targets_mean": 1103.7, | |
| "valid_targets_min": 578 | |
| }, | |
| { | |
| "epoch": 3.7592137592137593, | |
| "grad_norm": 0.7687412095426766, | |
| "learning_rate": 1.4756891058940606e-05, | |
| "loss": 0.1576, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16270983219146729, | |
| "step": 1530, | |
| "valid_targets_mean": 1470.4, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 3.7714987714987718, | |
| "grad_norm": 0.9064979122055239, | |
| "learning_rate": 1.4619032729687223e-05, | |
| "loss": 0.1525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15882454812526703, | |
| "step": 1535, | |
| "valid_targets_mean": 1394.7, | |
| "valid_targets_min": 703 | |
| }, | |
| { | |
| "epoch": 3.7837837837837838, | |
| "grad_norm": 0.8424145043338477, | |
| "learning_rate": 1.4481449467212004e-05, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14801853895187378, | |
| "step": 1540, | |
| "valid_targets_mean": 1197.9, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 3.796068796068796, | |
| "grad_norm": 0.8343369574704005, | |
| "learning_rate": 1.4344148304559926e-05, | |
| "loss": 0.1525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15639081597328186, | |
| "step": 1545, | |
| "valid_targets_mean": 1394.4, | |
| "valid_targets_min": 930 | |
| }, | |
| { | |
| "epoch": 3.808353808353808, | |
| "grad_norm": 0.8129609975131263, | |
| "learning_rate": 1.4207136260355426e-05, | |
| "loss": 0.1512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15776801109313965, | |
| "step": 1550, | |
| "valid_targets_mean": 1388.5, | |
| "valid_targets_min": 844 | |
| }, | |
| { | |
| "epoch": 3.8206388206388207, | |
| "grad_norm": 0.8580808574668879, | |
| "learning_rate": 1.4070420338443667e-05, | |
| "loss": 0.1569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15302756428718567, | |
| "step": 1555, | |
| "valid_targets_mean": 1418.8, | |
| "valid_targets_min": 652 | |
| }, | |
| { | |
| "epoch": 3.832923832923833, | |
| "grad_norm": 0.777430435398723, | |
| "learning_rate": 1.3934007527532494e-05, | |
| "loss": 0.1559, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16227784752845764, | |
| "step": 1560, | |
| "valid_targets_mean": 1578.3, | |
| "valid_targets_min": 1012 | |
| }, | |
| { | |
| "epoch": 3.845208845208845, | |
| "grad_norm": 0.793914409365788, | |
| "learning_rate": 1.3797904800835174e-05, | |
| "loss": 0.1537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14964772760868073, | |
| "step": 1565, | |
| "valid_targets_mean": 1403.3, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 3.8574938574938575, | |
| "grad_norm": 0.8085824166690417, | |
| "learning_rate": 1.3662119115713968e-05, | |
| "loss": 0.1547, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13584861159324646, | |
| "step": 1570, | |
| "valid_targets_mean": 1334.8, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 3.8697788697788695, | |
| "grad_norm": 0.8070987527644918, | |
| "learning_rate": 1.3526657413324427e-05, | |
| "loss": 0.1491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14322111010551453, | |
| "step": 1575, | |
| "valid_targets_mean": 1394.0, | |
| "valid_targets_min": 755 | |
| }, | |
| { | |
| "epoch": 3.882063882063882, | |
| "grad_norm": 0.7998968510753872, | |
| "learning_rate": 1.3391526618260636e-05, | |
| "loss": 0.152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15165430307388306, | |
| "step": 1580, | |
| "valid_targets_mean": 1313.9, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 3.8943488943488944, | |
| "grad_norm": 0.8639486361401842, | |
| "learning_rate": 1.3256733638201172e-05, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16121874749660492, | |
| "step": 1585, | |
| "valid_targets_mean": 1215.0, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 3.906633906633907, | |
| "grad_norm": 0.8360820023117336, | |
| "learning_rate": 1.3122285363556053e-05, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14884233474731445, | |
| "step": 1590, | |
| "valid_targets_mean": 1404.9, | |
| "valid_targets_min": 552 | |
| }, | |
| { | |
| "epoch": 3.918918918918919, | |
| "grad_norm": 1.0324227056373336, | |
| "learning_rate": 1.2988188667114487e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17249566316604614, | |
| "step": 1595, | |
| "valid_targets_mean": 1450.9, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 3.9312039312039313, | |
| "grad_norm": 0.7758969682797754, | |
| "learning_rate": 1.2854450403693526e-05, | |
| "loss": 0.1552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1592610478401184, | |
| "step": 1600, | |
| "valid_targets_mean": 1363.2, | |
| "valid_targets_min": 670 | |
| }, | |
| { | |
| "epoch": 3.9434889434889433, | |
| "grad_norm": 0.8109033037431925, | |
| "learning_rate": 1.272107740978769e-05, | |
| "loss": 0.155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1471540331840515, | |
| "step": 1605, | |
| "valid_targets_mean": 1326.2, | |
| "valid_targets_min": 612 | |
| }, | |
| { | |
| "epoch": 3.9557739557739557, | |
| "grad_norm": 0.8004506548550595, | |
| "learning_rate": 1.2588076503219475e-05, | |
| "loss": 0.1545, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15889285504817963, | |
| "step": 1610, | |
| "valid_targets_mean": 1378.2, | |
| "valid_targets_min": 792 | |
| }, | |
| { | |
| "epoch": 3.968058968058968, | |
| "grad_norm": 0.7415942779166481, | |
| "learning_rate": 1.2455454482790859e-05, | |
| "loss": 0.1607, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15655523538589478, | |
| "step": 1615, | |
| "valid_targets_mean": 1420.8, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 3.98034398034398, | |
| "grad_norm": 0.8039429952990631, | |
| "learning_rate": 1.2323218127935714e-05, | |
| "loss": 0.1508, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1585637331008911, | |
| "step": 1620, | |
| "valid_targets_mean": 1429.4, | |
| "valid_targets_min": 706 | |
| }, | |
| { | |
| "epoch": 3.9926289926289926, | |
| "grad_norm": 0.8692606501071803, | |
| "learning_rate": 1.2191374198373309e-05, | |
| "loss": 0.1531, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1549098938703537, | |
| "step": 1625, | |
| "valid_targets_mean": 1184.1, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 4.004914004914005, | |
| "grad_norm": 0.7418231127350439, | |
| "learning_rate": 1.2059929433762734e-05, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13540256023406982, | |
| "step": 1630, | |
| "valid_targets_mean": 1401.8, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 4.017199017199017, | |
| "grad_norm": 0.7491106138112695, | |
| "learning_rate": 1.1928890553358352e-05, | |
| "loss": 0.136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1356310099363327, | |
| "step": 1635, | |
| "valid_targets_mean": 1565.2, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 4.0294840294840295, | |
| "grad_norm": 0.8758510629920021, | |
| "learning_rate": 1.1798264255666387e-05, | |
| "loss": 0.1324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14381960034370422, | |
| "step": 1640, | |
| "valid_targets_mean": 1300.3, | |
| "valid_targets_min": 602 | |
| }, | |
| { | |
| "epoch": 4.041769041769042, | |
| "grad_norm": 0.7566832249876295, | |
| "learning_rate": 1.1668057218102436e-05, | |
| "loss": 0.1289, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11850036680698395, | |
| "step": 1645, | |
| "valid_targets_mean": 1487.4, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 4.054054054054054, | |
| "grad_norm": 0.8861198265110193, | |
| "learning_rate": 1.1538276096650175e-05, | |
| "loss": 0.1321, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13395559787750244, | |
| "step": 1650, | |
| "valid_targets_mean": 1204.8, | |
| "valid_targets_min": 510 | |
| }, | |
| { | |
| "epoch": 4.066339066339066, | |
| "grad_norm": 0.7931290609747208, | |
| "learning_rate": 1.1408927525521118e-05, | |
| "loss": 0.1356, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12991078197956085, | |
| "step": 1655, | |
| "valid_targets_mean": 1414.1, | |
| "valid_targets_min": 768 | |
| }, | |
| { | |
| "epoch": 4.078624078624078, | |
| "grad_norm": 0.8148588065033624, | |
| "learning_rate": 1.1280018116815438e-05, | |
| "loss": 0.1312, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1274513453245163, | |
| "step": 1660, | |
| "valid_targets_mean": 1330.9, | |
| "valid_targets_min": 675 | |
| }, | |
| { | |
| "epoch": 4.090909090909091, | |
| "grad_norm": 0.8651740089430039, | |
| "learning_rate": 1.115155446018404e-05, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12686586380004883, | |
| "step": 1665, | |
| "valid_targets_mean": 1129.1, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 4.103194103194103, | |
| "grad_norm": 1.0152247871489999, | |
| "learning_rate": 1.1023543122491626e-05, | |
| "loss": 0.1329, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14235243201255798, | |
| "step": 1670, | |
| "valid_targets_mean": 1496.2, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 4.115479115479116, | |
| "grad_norm": 0.8387772219526038, | |
| "learning_rate": 1.089599064748108e-05, | |
| "loss": 0.1343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12564003467559814, | |
| "step": 1675, | |
| "valid_targets_mean": 1218.8, | |
| "valid_targets_min": 829 | |
| }, | |
| { | |
| "epoch": 4.127764127764128, | |
| "grad_norm": 0.8758223941652693, | |
| "learning_rate": 1.0768903555438927e-05, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13461729884147644, | |
| "step": 1680, | |
| "valid_targets_mean": 1221.2, | |
| "valid_targets_min": 866 | |
| }, | |
| { | |
| "epoch": 4.14004914004914, | |
| "grad_norm": 0.8892247132266647, | |
| "learning_rate": 1.0642288342862007e-05, | |
| "loss": 0.1296, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1310172975063324, | |
| "step": 1685, | |
| "valid_targets_mean": 1356.1, | |
| "valid_targets_min": 821 | |
| }, | |
| { | |
| "epoch": 4.152334152334152, | |
| "grad_norm": 0.9050342496671671, | |
| "learning_rate": 1.051615148212544e-05, | |
| "loss": 0.1327, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13743871450424194, | |
| "step": 1690, | |
| "valid_targets_mean": 1221.8, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 4.164619164619165, | |
| "grad_norm": 0.7939266146430992, | |
| "learning_rate": 1.0390499421151706e-05, | |
| "loss": 0.1303, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1256050169467926, | |
| "step": 1695, | |
| "valid_targets_mean": 1561.8, | |
| "valid_targets_min": 838 | |
| }, | |
| { | |
| "epoch": 4.176904176904177, | |
| "grad_norm": 0.8958126329856702, | |
| "learning_rate": 1.0265338583081088e-05, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13179880380630493, | |
| "step": 1700, | |
| "valid_targets_mean": 1211.4, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 4.1891891891891895, | |
| "grad_norm": 0.937480505694556, | |
| "learning_rate": 1.0140675365943284e-05, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14077922701835632, | |
| "step": 1705, | |
| "valid_targets_mean": 1244.7, | |
| "valid_targets_min": 740 | |
| }, | |
| { | |
| "epoch": 4.201474201474202, | |
| "grad_norm": 0.8649258597004559, | |
| "learning_rate": 1.0016516142330404e-05, | |
| "loss": 0.1353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13584747910499573, | |
| "step": 1710, | |
| "valid_targets_mean": 1241.2, | |
| "valid_targets_min": 817 | |
| }, | |
| { | |
| "epoch": 4.2137592137592135, | |
| "grad_norm": 0.8702025042777511, | |
| "learning_rate": 9.89286725907117e-06, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1355808675289154, | |
| "step": 1715, | |
| "valid_targets_mean": 1278.6, | |
| "valid_targets_min": 764 | |
| }, | |
| { | |
| "epoch": 4.226044226044226, | |
| "grad_norm": 1.0405293797225805, | |
| "learning_rate": 9.769735036906475e-06, | |
| "loss": 0.1348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13069278001785278, | |
| "step": 1720, | |
| "valid_targets_mean": 1170.8, | |
| "valid_targets_min": 686 | |
| }, | |
| { | |
| "epoch": 4.238329238329238, | |
| "grad_norm": 0.8353293938513687, | |
| "learning_rate": 9.647125770166321e-06, | |
| "loss": 0.1323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13137099146842957, | |
| "step": 1725, | |
| "valid_targets_mean": 1290.1, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 4.250614250614251, | |
| "grad_norm": 0.8776999596670313, | |
| "learning_rate": 9.525045726448001e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1377831995487213, | |
| "step": 1730, | |
| "valid_targets_mean": 1338.8, | |
| "valid_targets_min": 708 | |
| }, | |
| { | |
| "epoch": 4.262899262899263, | |
| "grad_norm": 0.876277681541209, | |
| "learning_rate": 9.40350114629577e-06, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13550785183906555, | |
| "step": 1735, | |
| "valid_targets_mean": 1207.9, | |
| "valid_targets_min": 702 | |
| }, | |
| { | |
| "epoch": 4.275184275184275, | |
| "grad_norm": 0.8944643000153863, | |
| "learning_rate": 9.282498242881784e-06, | |
| "loss": 0.1339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1361280232667923, | |
| "step": 1740, | |
| "valid_targets_mean": 1364.2, | |
| "valid_targets_min": 687 | |
| }, | |
| { | |
| "epoch": 4.287469287469287, | |
| "grad_norm": 0.8582961426311289, | |
| "learning_rate": 9.162043201688517e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12418873608112335, | |
| "step": 1745, | |
| "valid_targets_mean": 1242.6, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 4.2997542997543, | |
| "grad_norm": 0.8460722988037566, | |
| "learning_rate": 9.042142180192596e-06, | |
| "loss": 0.1384, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13723894953727722, | |
| "step": 1750, | |
| "valid_targets_mean": 1232.2, | |
| "valid_targets_min": 750 | |
| }, | |
| { | |
| "epoch": 4.312039312039312, | |
| "grad_norm": 0.8185360622222694, | |
| "learning_rate": 8.92280130754998e-06, | |
| "loss": 0.1319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13691049814224243, | |
| "step": 1755, | |
| "valid_targets_mean": 1385.4, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 4.324324324324325, | |
| "grad_norm": 0.8089554216095896, | |
| "learning_rate": 8.804026684282694e-06, | |
| "loss": 0.1347, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1337611973285675, | |
| "step": 1760, | |
| "valid_targets_mean": 1322.9, | |
| "valid_targets_min": 723 | |
| }, | |
| { | |
| "epoch": 4.336609336609337, | |
| "grad_norm": 2.6163400061885085, | |
| "learning_rate": 8.685824381966975e-06, | |
| "loss": 0.134, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14328783750534058, | |
| "step": 1765, | |
| "valid_targets_mean": 1257.4, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 4.348894348894349, | |
| "grad_norm": 0.8727646019220853, | |
| "learning_rate": 8.568200442922865e-06, | |
| "loss": 0.1346, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13159531354904175, | |
| "step": 1770, | |
| "valid_targets_mean": 1367.1, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 4.361179361179361, | |
| "grad_norm": 0.8996659837757238, | |
| "learning_rate": 8.451160879905398e-06, | |
| "loss": 0.1337, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1381177306175232, | |
| "step": 1775, | |
| "valid_targets_mean": 1295.1, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 4.3734643734643734, | |
| "grad_norm": 0.8221175548847971, | |
| "learning_rate": 8.33471167579717e-06, | |
| "loss": 0.1368, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14100399613380432, | |
| "step": 1780, | |
| "valid_targets_mean": 1438.7, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 4.385749385749386, | |
| "grad_norm": 0.9160963463737738, | |
| "learning_rate": 8.218858783302566e-06, | |
| "loss": 0.1372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13693374395370483, | |
| "step": 1785, | |
| "valid_targets_mean": 1326.6, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 4.398034398034398, | |
| "grad_norm": 0.8441030171356969, | |
| "learning_rate": 8.103608124643412e-06, | |
| "loss": 0.1298, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12508758902549744, | |
| "step": 1790, | |
| "valid_targets_mean": 1355.0, | |
| "valid_targets_min": 745 | |
| }, | |
| { | |
| "epoch": 4.41031941031941, | |
| "grad_norm": 0.8520294326439847, | |
| "learning_rate": 7.988965591256284e-06, | |
| "loss": 0.1393, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13980035483837128, | |
| "step": 1795, | |
| "valid_targets_mean": 1313.2, | |
| "valid_targets_min": 677 | |
| }, | |
| { | |
| "epoch": 4.422604422604422, | |
| "grad_norm": 0.8246418406935714, | |
| "learning_rate": 7.874937043491331e-06, | |
| "loss": 0.1364, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13462932407855988, | |
| "step": 1800, | |
| "valid_targets_mean": 1353.4, | |
| "valid_targets_min": 654 | |
| }, | |
| { | |
| "epoch": 4.434889434889435, | |
| "grad_norm": 0.8457223705197832, | |
| "learning_rate": 7.761528310312679e-06, | |
| "loss": 0.136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13229894638061523, | |
| "step": 1805, | |
| "valid_targets_mean": 1344.6, | |
| "valid_targets_min": 658 | |
| }, | |
| { | |
| "epoch": 4.447174447174447, | |
| "grad_norm": 0.833607462401412, | |
| "learning_rate": 7.648745189000511e-06, | |
| "loss": 0.129, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12600962817668915, | |
| "step": 1810, | |
| "valid_targets_mean": 1313.6, | |
| "valid_targets_min": 607 | |
| }, | |
| { | |
| "epoch": 4.45945945945946, | |
| "grad_norm": 0.8510312578108182, | |
| "learning_rate": 7.536593444854663e-06, | |
| "loss": 0.1324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13862791657447815, | |
| "step": 1815, | |
| "valid_targets_mean": 1411.5, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 4.471744471744472, | |
| "grad_norm": 0.9145076730321617, | |
| "learning_rate": 7.4250788108999686e-06, | |
| "loss": 0.136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13466450572013855, | |
| "step": 1820, | |
| "valid_targets_mean": 1134.2, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 4.484029484029484, | |
| "grad_norm": 0.8548827601262811, | |
| "learning_rate": 7.314206987593162e-06, | |
| "loss": 0.1404, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13373500108718872, | |
| "step": 1825, | |
| "valid_targets_mean": 1223.4, | |
| "valid_targets_min": 776 | |
| }, | |
| { | |
| "epoch": 4.496314496314496, | |
| "grad_norm": 0.8938428905617729, | |
| "learning_rate": 7.203983642531462e-06, | |
| "loss": 0.1365, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13550472259521484, | |
| "step": 1830, | |
| "valid_targets_mean": 1318.0, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 4.5085995085995085, | |
| "grad_norm": 0.8603104670065109, | |
| "learning_rate": 7.094414410162913e-06, | |
| "loss": 0.1388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13778984546661377, | |
| "step": 1835, | |
| "valid_targets_mean": 1195.8, | |
| "valid_targets_min": 689 | |
| }, | |
| { | |
| "epoch": 4.520884520884521, | |
| "grad_norm": 0.7919032263244239, | |
| "learning_rate": 6.985504891498291e-06, | |
| "loss": 0.1329, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1213904470205307, | |
| "step": 1840, | |
| "valid_targets_mean": 1309.6, | |
| "valid_targets_min": 661 | |
| }, | |
| { | |
| "epoch": 4.533169533169533, | |
| "grad_norm": 0.8592680814494961, | |
| "learning_rate": 6.8772606538248285e-06, | |
| "loss": 0.1337, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13496822118759155, | |
| "step": 1845, | |
| "valid_targets_mean": 1331.1, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 4.545454545454545, | |
| "grad_norm": 0.8466687407616076, | |
| "learning_rate": 6.769687230421638e-06, | |
| "loss": 0.1348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13918258249759674, | |
| "step": 1850, | |
| "valid_targets_mean": 1431.8, | |
| "valid_targets_min": 637 | |
| }, | |
| { | |
| "epoch": 4.557739557739557, | |
| "grad_norm": 0.8216050231021137, | |
| "learning_rate": 6.662790120276803e-06, | |
| "loss": 0.138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13874514400959015, | |
| "step": 1855, | |
| "valid_targets_mean": 1524.4, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 4.57002457002457, | |
| "grad_norm": 0.9669401045446978, | |
| "learning_rate": 6.556574787806344e-06, | |
| "loss": 0.1337, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14022761583328247, | |
| "step": 1860, | |
| "valid_targets_mean": 1152.1, | |
| "valid_targets_min": 578 | |
| }, | |
| { | |
| "epoch": 4.582309582309582, | |
| "grad_norm": 0.9375234740980322, | |
| "learning_rate": 6.451046662574831e-06, | |
| "loss": 0.1377, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13871565461158752, | |
| "step": 1865, | |
| "valid_targets_mean": 1150.8, | |
| "valid_targets_min": 640 | |
| }, | |
| { | |
| "epoch": 4.594594594594595, | |
| "grad_norm": 0.836770791087053, | |
| "learning_rate": 6.346211139017877e-06, | |
| "loss": 0.1373, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13808581233024597, | |
| "step": 1870, | |
| "valid_targets_mean": 1486.2, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 4.606879606879607, | |
| "grad_norm": 0.950259784141947, | |
| "learning_rate": 6.242073576166337e-06, | |
| "loss": 0.1334, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13348250091075897, | |
| "step": 1875, | |
| "valid_targets_mean": 1282.6, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 4.61916461916462, | |
| "grad_norm": 0.9276466101362375, | |
| "learning_rate": 6.138639297372404e-06, | |
| "loss": 0.1287, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13664673268795013, | |
| "step": 1880, | |
| "valid_targets_mean": 1407.1, | |
| "valid_targets_min": 873 | |
| }, | |
| { | |
| "epoch": 4.631449631449631, | |
| "grad_norm": 0.9134688417135232, | |
| "learning_rate": 6.035913590037479e-06, | |
| "loss": 0.1329, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13874787092208862, | |
| "step": 1885, | |
| "valid_targets_mean": 1228.9, | |
| "valid_targets_min": 726 | |
| }, | |
| { | |
| "epoch": 4.643734643734644, | |
| "grad_norm": 0.9364687166887621, | |
| "learning_rate": 5.933901705341851e-06, | |
| "loss": 0.1351, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13511444628238678, | |
| "step": 1890, | |
| "valid_targets_mean": 1213.6, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 4.656019656019656, | |
| "grad_norm": 0.8226051755557579, | |
| "learning_rate": 5.832608857976321e-06, | |
| "loss": 0.1323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14297759532928467, | |
| "step": 1895, | |
| "valid_targets_mean": 1434.4, | |
| "valid_targets_min": 604 | |
| }, | |
| { | |
| "epoch": 4.6683046683046685, | |
| "grad_norm": 0.8928576176089511, | |
| "learning_rate": 5.732040225875584e-06, | |
| "loss": 0.138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14191249012947083, | |
| "step": 1900, | |
| "valid_targets_mean": 1351.0, | |
| "valid_targets_min": 740 | |
| }, | |
| { | |
| "epoch": 4.680589680589681, | |
| "grad_norm": 0.8362193728819766, | |
| "learning_rate": 5.632200949953579e-06, | |
| "loss": 0.1324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13142593204975128, | |
| "step": 1905, | |
| "valid_targets_mean": 1434.6, | |
| "valid_targets_min": 508 | |
| }, | |
| { | |
| "epoch": 4.6928746928746925, | |
| "grad_norm": 0.8558162872587993, | |
| "learning_rate": 5.533096133840677e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1275939792394638, | |
| "step": 1910, | |
| "valid_targets_mean": 1292.1, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 4.705159705159705, | |
| "grad_norm": 0.7983343068498832, | |
| "learning_rate": 5.434730843622778e-06, | |
| "loss": 0.1294, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1256292462348938, | |
| "step": 1915, | |
| "valid_targets_mean": 1475.9, | |
| "valid_targets_min": 941 | |
| }, | |
| { | |
| "epoch": 4.717444717444717, | |
| "grad_norm": 1.3118893758847399, | |
| "learning_rate": 5.337110107582377e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13079731166362762, | |
| "step": 1920, | |
| "valid_targets_mean": 1585.5, | |
| "valid_targets_min": 933 | |
| }, | |
| { | |
| "epoch": 4.72972972972973, | |
| "grad_norm": 0.8080898258134603, | |
| "learning_rate": 5.2402389159414755e-06, | |
| "loss": 0.1343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13039739429950714, | |
| "step": 1925, | |
| "valid_targets_mean": 1359.4, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 4.742014742014742, | |
| "grad_norm": 1.0236112284503025, | |
| "learning_rate": 5.144122220606542e-06, | |
| "loss": 0.1325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12727615237236023, | |
| "step": 1930, | |
| "valid_targets_mean": 1334.2, | |
| "valid_targets_min": 836 | |
| }, | |
| { | |
| "epoch": 4.754299754299755, | |
| "grad_norm": 0.8608999172912609, | |
| "learning_rate": 5.048764934915349e-06, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12973430752754211, | |
| "step": 1935, | |
| "valid_targets_mean": 1372.7, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 4.766584766584766, | |
| "grad_norm": 0.8457132249837475, | |
| "learning_rate": 4.954171933385805e-06, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12466098368167877, | |
| "step": 1940, | |
| "valid_targets_mean": 1208.1, | |
| "valid_targets_min": 664 | |
| }, | |
| { | |
| "epoch": 4.778869778869779, | |
| "grad_norm": 0.8118421199942646, | |
| "learning_rate": 4.8603480514667836e-06, | |
| "loss": 0.1385, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1356389820575714, | |
| "step": 1945, | |
| "valid_targets_mean": 1539.8, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 4.791154791154791, | |
| "grad_norm": 0.9109442080712121, | |
| "learning_rate": 4.767298085290963e-06, | |
| "loss": 0.1407, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13706953823566437, | |
| "step": 1950, | |
| "valid_targets_mean": 1200.8, | |
| "valid_targets_min": 814 | |
| }, | |
| { | |
| "epoch": 4.803439803439804, | |
| "grad_norm": 0.7836538521572521, | |
| "learning_rate": 4.675026791429624e-06, | |
| "loss": 0.132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13943910598754883, | |
| "step": 1955, | |
| "valid_targets_mean": 1431.2, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 4.815724815724816, | |
| "grad_norm": 0.8338736840520049, | |
| "learning_rate": 4.583538886649525e-06, | |
| "loss": 0.1281, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11881544440984726, | |
| "step": 1960, | |
| "valid_targets_mean": 1207.3, | |
| "valid_targets_min": 769 | |
| }, | |
| { | |
| "epoch": 4.828009828009828, | |
| "grad_norm": 0.8631974963169733, | |
| "learning_rate": 4.492839047671764e-06, | |
| "loss": 0.1387, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13759952783584595, | |
| "step": 1965, | |
| "valid_targets_mean": 1240.6, | |
| "valid_targets_min": 716 | |
| }, | |
| { | |
| "epoch": 4.84029484029484, | |
| "grad_norm": 0.8445274635568915, | |
| "learning_rate": 4.4029319109327465e-06, | |
| "loss": 0.1348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1327393352985382, | |
| "step": 1970, | |
| "valid_targets_mean": 1349.8, | |
| "valid_targets_min": 569 | |
| }, | |
| { | |
| "epoch": 4.8525798525798525, | |
| "grad_norm": 0.8379355840832601, | |
| "learning_rate": 4.313822072347136e-06, | |
| "loss": 0.1293, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13074269890785217, | |
| "step": 1975, | |
| "valid_targets_mean": 1367.1, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 4.864864864864865, | |
| "grad_norm": 0.9109766321758263, | |
| "learning_rate": 4.22551408707296e-06, | |
| "loss": 0.127, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12870089709758759, | |
| "step": 1980, | |
| "valid_targets_mean": 1204.9, | |
| "valid_targets_min": 608 | |
| }, | |
| { | |
| "epoch": 4.877149877149877, | |
| "grad_norm": 0.826848683962763, | |
| "learning_rate": 4.138012469278714e-06, | |
| "loss": 0.1302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12679742276668549, | |
| "step": 1985, | |
| "valid_targets_mean": 1459.3, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 4.88943488943489, | |
| "grad_norm": 1.0072797324517473, | |
| "learning_rate": 4.051321691912649e-06, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1277749389410019, | |
| "step": 1990, | |
| "valid_targets_mean": 1285.3, | |
| "valid_targets_min": 837 | |
| }, | |
| { | |
| "epoch": 4.901719901719901, | |
| "grad_norm": 0.8582064043845651, | |
| "learning_rate": 3.9654461864740935e-06, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12809713184833527, | |
| "step": 1995, | |
| "valid_targets_mean": 1380.1, | |
| "valid_targets_min": 808 | |
| }, | |
| { | |
| "epoch": 4.914004914004914, | |
| "grad_norm": 0.8154469160438075, | |
| "learning_rate": 3.880390342786915e-06, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13291479647159576, | |
| "step": 2000, | |
| "valid_targets_mean": 1429.1, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 4.926289926289926, | |
| "grad_norm": 0.8649560778284509, | |
| "learning_rate": 3.7961585087751516e-06, | |
| "loss": 0.1318, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12271484732627869, | |
| "step": 2005, | |
| "valid_targets_mean": 1190.2, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 4.938574938574939, | |
| "grad_norm": 0.8829456549947029, | |
| "learning_rate": 3.71275499024071e-06, | |
| "loss": 0.1323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12908603250980377, | |
| "step": 2010, | |
| "valid_targets_mean": 1133.1, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 4.950859950859951, | |
| "grad_norm": 0.8485559058439717, | |
| "learning_rate": 3.6301840506433083e-06, | |
| "loss": 0.1292, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12414588779211044, | |
| "step": 2015, | |
| "valid_targets_mean": 1268.0, | |
| "valid_targets_min": 672 | |
| }, | |
| { | |
| "epoch": 4.963144963144963, | |
| "grad_norm": 0.9312954958571251, | |
| "learning_rate": 3.5484499108824853e-06, | |
| "loss": 0.1349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13587301969528198, | |
| "step": 2020, | |
| "valid_targets_mean": 1167.4, | |
| "valid_targets_min": 685 | |
| }, | |
| { | |
| "epoch": 4.975429975429975, | |
| "grad_norm": 0.8826698941979991, | |
| "learning_rate": 3.4675567490818727e-06, | |
| "loss": 0.1396, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13824445009231567, | |
| "step": 2025, | |
| "valid_targets_mean": 1295.9, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 4.987714987714988, | |
| "grad_norm": 0.842942106749072, | |
| "learning_rate": 3.3875087003756036e-06, | |
| "loss": 0.133, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1280279904603958, | |
| "step": 2030, | |
| "valid_targets_mean": 1316.2, | |
| "valid_targets_min": 714 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 1.560417863126269, | |
| "learning_rate": 3.30830985669691e-06, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12781304121017456, | |
| "step": 2035, | |
| "valid_targets_mean": 1413.3, | |
| "valid_targets_min": 603 | |
| }, | |
| { | |
| "epoch": 5.012285012285012, | |
| "grad_norm": 0.7931305692109261, | |
| "learning_rate": 3.22996426656899e-06, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11616753041744232, | |
| "step": 2040, | |
| "valid_targets_mean": 1353.3, | |
| "valid_targets_min": 818 | |
| }, | |
| { | |
| "epoch": 5.024570024570025, | |
| "grad_norm": 1.0099583617389296, | |
| "learning_rate": 3.1524759348980096e-06, | |
| "loss": 0.1222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12158218771219254, | |
| "step": 2045, | |
| "valid_targets_mean": 1211.6, | |
| "valid_targets_min": 668 | |
| }, | |
| { | |
| "epoch": 5.036855036855036, | |
| "grad_norm": 0.8500902588731212, | |
| "learning_rate": 3.0758488227684212e-06, | |
| "loss": 0.1223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12790502607822418, | |
| "step": 2050, | |
| "valid_targets_mean": 1328.2, | |
| "valid_targets_min": 546 | |
| }, | |
| { | |
| "epoch": 5.049140049140049, | |
| "grad_norm": 0.928449305764292, | |
| "learning_rate": 3.0000868472404423e-06, | |
| "loss": 0.1266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12052057683467865, | |
| "step": 2055, | |
| "valid_targets_mean": 1168.4, | |
| "valid_targets_min": 666 | |
| }, | |
| { | |
| "epoch": 5.061425061425061, | |
| "grad_norm": 0.88759587464759, | |
| "learning_rate": 2.9251938811498436e-06, | |
| "loss": 0.1226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11871778219938278, | |
| "step": 2060, | |
| "valid_targets_mean": 1182.1, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 5.073710073710074, | |
| "grad_norm": 0.8236200534470612, | |
| "learning_rate": 2.8511737529099704e-06, | |
| "loss": 0.1218, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11925345659255981, | |
| "step": 2065, | |
| "valid_targets_mean": 1338.1, | |
| "valid_targets_min": 741 | |
| }, | |
| { | |
| "epoch": 5.085995085995086, | |
| "grad_norm": 0.9644824242433888, | |
| "learning_rate": 2.7780302463160235e-06, | |
| "loss": 0.1281, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13304205238819122, | |
| "step": 2070, | |
| "valid_targets_mean": 1192.4, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 5.098280098280099, | |
| "grad_norm": 0.8128725043575888, | |
| "learning_rate": 2.705767100351673e-06, | |
| "loss": 0.1279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11420956254005432, | |
| "step": 2075, | |
| "valid_targets_mean": 1345.1, | |
| "valid_targets_min": 564 | |
| }, | |
| { | |
| "epoch": 5.11056511056511, | |
| "grad_norm": 0.9134050119992433, | |
| "learning_rate": 2.634388008997899e-06, | |
| "loss": 0.1274, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12336855381727219, | |
| "step": 2080, | |
| "valid_targets_mean": 1291.5, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 5.122850122850123, | |
| "grad_norm": 0.8346188690601416, | |
| "learning_rate": 2.5638966210441597e-06, | |
| "loss": 0.1185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11434096097946167, | |
| "step": 2085, | |
| "valid_targets_mean": 1294.2, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 5.135135135135135, | |
| "grad_norm": 0.8971067231810141, | |
| "learning_rate": 2.4942965399018926e-06, | |
| "loss": 0.1244, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12700599431991577, | |
| "step": 2090, | |
| "valid_targets_mean": 1269.8, | |
| "valid_targets_min": 477 | |
| }, | |
| { | |
| "epoch": 5.1474201474201475, | |
| "grad_norm": 0.8396329615955295, | |
| "learning_rate": 2.425591323420289e-06, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12476441264152527, | |
| "step": 2095, | |
| "valid_targets_mean": 1452.9, | |
| "valid_targets_min": 725 | |
| }, | |
| { | |
| "epoch": 5.15970515970516, | |
| "grad_norm": 0.9668038964121843, | |
| "learning_rate": 2.357784483704444e-06, | |
| "loss": 0.1218, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12222275882959366, | |
| "step": 2100, | |
| "valid_targets_mean": 1202.8, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 5.171990171990172, | |
| "grad_norm": 0.8530860924157005, | |
| "learning_rate": 2.2908794869358044e-06, | |
| "loss": 0.1216, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11945752799510956, | |
| "step": 2105, | |
| "valid_targets_mean": 1264.2, | |
| "valid_targets_min": 669 | |
| }, | |
| { | |
| "epoch": 5.184275184275184, | |
| "grad_norm": 0.8756759676757957, | |
| "learning_rate": 2.2248797531949952e-06, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12002584338188171, | |
| "step": 2110, | |
| "valid_targets_mean": 1310.5, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 5.196560196560196, | |
| "grad_norm": 0.8828888300496266, | |
| "learning_rate": 2.1597886562869917e-06, | |
| "loss": 0.1242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13054341077804565, | |
| "step": 2115, | |
| "valid_targets_mean": 1316.6, | |
| "valid_targets_min": 609 | |
| }, | |
| { | |
| "epoch": 5.208845208845209, | |
| "grad_norm": 0.8394875589131532, | |
| "learning_rate": 2.095609523568638e-06, | |
| "loss": 0.1186, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1151779368519783, | |
| "step": 2120, | |
| "valid_targets_mean": 1418.0, | |
| "valid_targets_min": 800 | |
| }, | |
| { | |
| "epoch": 5.221130221130221, | |
| "grad_norm": 0.8903601792892257, | |
| "learning_rate": 2.0323456357785855e-06, | |
| "loss": 0.1241, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1157693862915039, | |
| "step": 2125, | |
| "valid_targets_mean": 1233.6, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 5.233415233415234, | |
| "grad_norm": 0.8083599530344519, | |
| "learning_rate": 1.970000226869553e-06, | |
| "loss": 0.1208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11680185794830322, | |
| "step": 2130, | |
| "valid_targets_mean": 1303.5, | |
| "valid_targets_min": 698 | |
| }, | |
| { | |
| "epoch": 5.245700245700245, | |
| "grad_norm": 0.887959202055401, | |
| "learning_rate": 1.90857648384305e-06, | |
| "loss": 0.1198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11497928202152252, | |
| "step": 2135, | |
| "valid_targets_mean": 1166.8, | |
| "valid_targets_min": 849 | |
| }, | |
| { | |
| "epoch": 5.257985257985258, | |
| "grad_norm": 0.8924545156140798, | |
| "learning_rate": 1.848077546586431e-06, | |
| "loss": 0.1194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1226976066827774, | |
| "step": 2140, | |
| "valid_targets_mean": 1300.3, | |
| "valid_targets_min": 585 | |
| }, | |
| { | |
| "epoch": 5.27027027027027, | |
| "grad_norm": 0.887818676899762, | |
| "learning_rate": 1.7885065077123976e-06, | |
| "loss": 0.1246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11834403872489929, | |
| "step": 2145, | |
| "valid_targets_mean": 1208.6, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 5.282555282555283, | |
| "grad_norm": 0.8761041176465963, | |
| "learning_rate": 1.7298664124009245e-06, | |
| "loss": 0.1257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12676090002059937, | |
| "step": 2150, | |
| "valid_targets_mean": 1258.4, | |
| "valid_targets_min": 774 | |
| }, | |
| { | |
| "epoch": 5.294840294840295, | |
| "grad_norm": 0.9376811939994709, | |
| "learning_rate": 1.672160258243567e-06, | |
| "loss": 0.1188, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11709423363208771, | |
| "step": 2155, | |
| "valid_targets_mean": 1109.1, | |
| "valid_targets_min": 577 | |
| }, | |
| { | |
| "epoch": 5.3071253071253075, | |
| "grad_norm": 0.9358803313254709, | |
| "learning_rate": 1.615390995090258e-06, | |
| "loss": 0.1197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1253659427165985, | |
| "step": 2160, | |
| "valid_targets_mean": 1176.1, | |
| "valid_targets_min": 716 | |
| }, | |
| { | |
| "epoch": 5.319410319410319, | |
| "grad_norm": 0.9101281679651285, | |
| "learning_rate": 1.559561524898492e-06, | |
| "loss": 0.1225, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12124703824520111, | |
| "step": 2165, | |
| "valid_targets_mean": 1176.6, | |
| "valid_targets_min": 697 | |
| }, | |
| { | |
| "epoch": 5.3316953316953315, | |
| "grad_norm": 0.9020013037671613, | |
| "learning_rate": 1.5046747015849893e-06, | |
| "loss": 0.1181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12182813882827759, | |
| "step": 2170, | |
| "valid_targets_mean": 1287.6, | |
| "valid_targets_min": 943 | |
| }, | |
| { | |
| "epoch": 5.343980343980344, | |
| "grad_norm": 0.8665564025261498, | |
| "learning_rate": 1.4507333308798255e-06, | |
| "loss": 0.1251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11275840550661087, | |
| "step": 2175, | |
| "valid_targets_mean": 1213.7, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 5.356265356265356, | |
| "grad_norm": 0.7948586556833295, | |
| "learning_rate": 1.3977401701829752e-06, | |
| "loss": 0.1148, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11124136298894882, | |
| "step": 2180, | |
| "valid_targets_mean": 1512.9, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 5.368550368550369, | |
| "grad_norm": 0.8479955134968785, | |
| "learning_rate": 1.345697928423384e-06, | |
| "loss": 0.1222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11807121336460114, | |
| "step": 2185, | |
| "valid_targets_mean": 1355.6, | |
| "valid_targets_min": 841 | |
| }, | |
| { | |
| "epoch": 5.38083538083538, | |
| "grad_norm": 0.913604687178217, | |
| "learning_rate": 1.2946092659204767e-06, | |
| "loss": 0.1226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11881860345602036, | |
| "step": 2190, | |
| "valid_targets_mean": 1087.2, | |
| "valid_targets_min": 626 | |
| }, | |
| { | |
| "epoch": 5.393120393120393, | |
| "grad_norm": 0.9143190692852058, | |
| "learning_rate": 1.244476794248175e-06, | |
| "loss": 0.1225, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12613148987293243, | |
| "step": 2195, | |
| "valid_targets_mean": 1249.9, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 5.405405405405405, | |
| "grad_norm": 0.9554493796932091, | |
| "learning_rate": 1.1953030761014017e-06, | |
| "loss": 0.12, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1262340396642685, | |
| "step": 2200, | |
| "valid_targets_mean": 1098.0, | |
| "valid_targets_min": 587 | |
| }, | |
| { | |
| "epoch": 5.417690417690418, | |
| "grad_norm": 0.8665422814682507, | |
| "learning_rate": 1.147090625165055e-06, | |
| "loss": 0.1206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11446169763803482, | |
| "step": 2205, | |
| "valid_targets_mean": 1287.6, | |
| "valid_targets_min": 650 | |
| }, | |
| { | |
| "epoch": 5.42997542997543, | |
| "grad_norm": 0.8488658718651519, | |
| "learning_rate": 1.0998419059855503e-06, | |
| "loss": 0.1246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11071212589740753, | |
| "step": 2210, | |
| "valid_targets_mean": 1244.9, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 5.442260442260443, | |
| "grad_norm": 0.9677990651511228, | |
| "learning_rate": 1.053559333844798e-06, | |
| "loss": 0.1209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12183261662721634, | |
| "step": 2215, | |
| "valid_targets_mean": 1159.0, | |
| "valid_targets_min": 619 | |
| }, | |
| { | |
| "epoch": 5.454545454545454, | |
| "grad_norm": 1.0224261972543698, | |
| "learning_rate": 1.0082452746367721e-06, | |
| "loss": 0.1186, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1227094978094101, | |
| "step": 2220, | |
| "valid_targets_mean": 1384.5, | |
| "valid_targets_min": 695 | |
| }, | |
| { | |
| "epoch": 5.466830466830467, | |
| "grad_norm": 0.9075016092237645, | |
| "learning_rate": 9.639020447465475e-07, | |
| "loss": 0.1229, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12406787276268005, | |
| "step": 2225, | |
| "valid_targets_mean": 1512.6, | |
| "valid_targets_min": 864 | |
| }, | |
| { | |
| "epoch": 5.479115479115479, | |
| "grad_norm": 0.854872277774117, | |
| "learning_rate": 9.205319109318922e-07, | |
| "loss": 0.118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11251235008239746, | |
| "step": 2230, | |
| "valid_targets_mean": 1181.9, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 5.4914004914004915, | |
| "grad_norm": 0.8276956734834705, | |
| "learning_rate": 8.781370902074049e-07, | |
| "loss": 0.1213, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11970800161361694, | |
| "step": 2235, | |
| "valid_targets_mean": 1536.9, | |
| "valid_targets_min": 789 | |
| }, | |
| { | |
| "epoch": 5.503685503685504, | |
| "grad_norm": 0.8219281728800845, | |
| "learning_rate": 8.367197497311719e-07, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11715812981128693, | |
| "step": 2240, | |
| "valid_targets_mean": 1464.1, | |
| "valid_targets_min": 881 | |
| }, | |
| { | |
| "epoch": 5.515970515970516, | |
| "grad_norm": 0.9336086469402823, | |
| "learning_rate": 7.962820066939958e-07, | |
| "loss": 0.1255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1221204623579979, | |
| "step": 2245, | |
| "valid_targets_mean": 1244.7, | |
| "valid_targets_min": 651 | |
| }, | |
| { | |
| "epoch": 5.528255528255528, | |
| "grad_norm": 0.8878725870062483, | |
| "learning_rate": 7.568259282111645e-07, | |
| "loss": 0.1224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1210436075925827, | |
| "step": 2250, | |
| "valid_targets_mean": 1273.2, | |
| "valid_targets_min": 525 | |
| }, | |
| { | |
| "epoch": 5.54054054054054, | |
| "grad_norm": 0.9305753159284449, | |
| "learning_rate": 7.183535312167755e-07, | |
| "loss": 0.1259, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12594647705554962, | |
| "step": 2255, | |
| "valid_targets_mean": 1302.9, | |
| "valid_targets_min": 711 | |
| }, | |
| { | |
| "epoch": 5.552825552825553, | |
| "grad_norm": 0.8196405271986151, | |
| "learning_rate": 6.808667823606474e-07, | |
| "loss": 0.1191, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11032858490943909, | |
| "step": 2260, | |
| "valid_targets_mean": 1217.7, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 5.565110565110565, | |
| "grad_norm": 0.9629415002181919, | |
| "learning_rate": 6.443675979077779e-07, | |
| "loss": 0.1194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11967958509922028, | |
| "step": 2265, | |
| "valid_targets_mean": 1159.2, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 5.577395577395578, | |
| "grad_norm": 0.8877542687460271, | |
| "learning_rate": 6.088578436403847e-07, | |
| "loss": 0.1197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12060631066560745, | |
| "step": 2270, | |
| "valid_targets_mean": 1184.3, | |
| "valid_targets_min": 773 | |
| }, | |
| { | |
| "epoch": 5.58968058968059, | |
| "grad_norm": 0.9064893261528624, | |
| "learning_rate": 5.743393347625436e-07, | |
| "loss": 0.1203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12473144382238388, | |
| "step": 2275, | |
| "valid_targets_mean": 1232.4, | |
| "valid_targets_min": 682 | |
| }, | |
| { | |
| "epoch": 5.601965601965602, | |
| "grad_norm": 0.891255492857794, | |
| "learning_rate": 5.408138358073833e-07, | |
| "loss": 0.1215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11696426570415497, | |
| "step": 2280, | |
| "valid_targets_mean": 1211.9, | |
| "valid_targets_min": 817 | |
| }, | |
| { | |
| "epoch": 5.614250614250614, | |
| "grad_norm": 0.8804064021646336, | |
| "learning_rate": 5.082830605468969e-07, | |
| "loss": 0.1249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12240994721651077, | |
| "step": 2285, | |
| "valid_targets_mean": 1465.6, | |
| "valid_targets_min": 693 | |
| }, | |
| { | |
| "epoch": 5.6265356265356266, | |
| "grad_norm": 0.884415144400364, | |
| "learning_rate": 4.767486719043235e-07, | |
| "loss": 0.1265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12938669323921204, | |
| "step": 2290, | |
| "valid_targets_mean": 1303.9, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 5.638820638820639, | |
| "grad_norm": 0.853083172107117, | |
| "learning_rate": 4.4621228186915833e-07, | |
| "loss": 0.1199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11663314700126648, | |
| "step": 2295, | |
| "valid_targets_mean": 1375.6, | |
| "valid_targets_min": 663 | |
| }, | |
| { | |
| "epoch": 5.651105651105651, | |
| "grad_norm": 0.8179139441747029, | |
| "learning_rate": 4.166754514147275e-07, | |
| "loss": 0.1205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11481982469558716, | |
| "step": 2300, | |
| "valid_targets_mean": 1349.8, | |
| "valid_targets_min": 694 | |
| }, | |
| { | |
| "epoch": 5.663390663390663, | |
| "grad_norm": 0.8642291461817891, | |
| "learning_rate": 3.881396904184231e-07, | |
| "loss": 0.1249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13626883924007416, | |
| "step": 2305, | |
| "valid_targets_mean": 1455.5, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 5.675675675675675, | |
| "grad_norm": 0.855015151000873, | |
| "learning_rate": 3.6060645758449584e-07, | |
| "loss": 0.118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10772322118282318, | |
| "step": 2310, | |
| "valid_targets_mean": 1306.6, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 5.687960687960688, | |
| "grad_norm": 0.939814937762572, | |
| "learning_rate": 3.34077160369497e-07, | |
| "loss": 0.123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12797702848911285, | |
| "step": 2315, | |
| "valid_targets_mean": 1318.8, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 5.7002457002457, | |
| "grad_norm": 0.8694466054617536, | |
| "learning_rate": 3.08553154910336e-07, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11525967717170715, | |
| "step": 2320, | |
| "valid_targets_mean": 1257.1, | |
| "valid_targets_min": 745 | |
| }, | |
| { | |
| "epoch": 5.712530712530713, | |
| "grad_norm": 0.9213076662745712, | |
| "learning_rate": 2.840357459549492e-07, | |
| "loss": 0.1209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12923762202262878, | |
| "step": 2325, | |
| "valid_targets_mean": 1301.2, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 5.724815724815725, | |
| "grad_norm": 0.8510568275894363, | |
| "learning_rate": 2.6052618679560884e-07, | |
| "loss": 0.124, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12109372019767761, | |
| "step": 2330, | |
| "valid_targets_mean": 1400.9, | |
| "valid_targets_min": 586 | |
| }, | |
| { | |
| "epoch": 5.737100737100737, | |
| "grad_norm": 0.9026468731374053, | |
| "learning_rate": 2.380256792048541e-07, | |
| "loss": 0.117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11053834855556488, | |
| "step": 2335, | |
| "valid_targets_mean": 1192.9, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 5.749385749385749, | |
| "grad_norm": 0.8932267089116724, | |
| "learning_rate": 2.1653537337405383e-07, | |
| "loss": 0.1199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12436593323945999, | |
| "step": 2340, | |
| "valid_targets_mean": 1306.6, | |
| "valid_targets_min": 683 | |
| }, | |
| { | |
| "epoch": 5.761670761670762, | |
| "grad_norm": 0.8144937839460926, | |
| "learning_rate": 1.9605636785462234e-07, | |
| "loss": 0.1179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11719940602779388, | |
| "step": 2345, | |
| "valid_targets_mean": 1488.6, | |
| "valid_targets_min": 647 | |
| }, | |
| { | |
| "epoch": 5.773955773955774, | |
| "grad_norm": 0.8509307547721932, | |
| "learning_rate": 1.7658970950185095e-07, | |
| "loss": 0.1189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1175820529460907, | |
| "step": 2350, | |
| "valid_targets_mean": 1377.8, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 5.7862407862407865, | |
| "grad_norm": 0.8755385347465204, | |
| "learning_rate": 1.5813639342140197e-07, | |
| "loss": 0.1207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.122462198138237, | |
| "step": 2355, | |
| "valid_targets_mean": 1414.1, | |
| "valid_targets_min": 559 | |
| }, | |
| { | |
| "epoch": 5.798525798525798, | |
| "grad_norm": 0.8354198048278062, | |
| "learning_rate": 1.4069736291843605e-07, | |
| "loss": 0.1216, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12454007565975189, | |
| "step": 2360, | |
| "valid_targets_mean": 1355.3, | |
| "valid_targets_min": 769 | |
| }, | |
| { | |
| "epoch": 5.8108108108108105, | |
| "grad_norm": 0.9279865336013997, | |
| "learning_rate": 1.242735094493952e-07, | |
| "loss": 0.1267, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1297607570886612, | |
| "step": 2365, | |
| "valid_targets_mean": 1301.8, | |
| "valid_targets_min": 709 | |
| }, | |
| { | |
| "epoch": 5.823095823095823, | |
| "grad_norm": 0.8356599989544432, | |
| "learning_rate": 1.0886567257643033e-07, | |
| "loss": 0.1246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13703671097755432, | |
| "step": 2370, | |
| "valid_targets_mean": 1503.9, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 5.835380835380835, | |
| "grad_norm": 0.8976360214913625, | |
| "learning_rate": 9.447463992448891e-08, | |
| "loss": 0.1223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1237124353647232, | |
| "step": 2375, | |
| "valid_targets_mean": 1253.7, | |
| "valid_targets_min": 728 | |
| }, | |
| { | |
| "epoch": 5.847665847665848, | |
| "grad_norm": 0.8469797015789384, | |
| "learning_rate": 8.110114714104277e-08, | |
| "loss": 0.1194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12047050893306732, | |
| "step": 2380, | |
| "valid_targets_mean": 1357.1, | |
| "valid_targets_min": 798 | |
| }, | |
| { | |
| "epoch": 5.85995085995086, | |
| "grad_norm": 0.8474816619307004, | |
| "learning_rate": 6.874587785849152e-08, | |
| "loss": 0.1159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11204469203948975, | |
| "step": 2385, | |
| "valid_targets_mean": 1219.0, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 5.872235872235873, | |
| "grad_norm": 0.7930581021314657, | |
| "learning_rate": 5.7409463659219286e-08, | |
| "loss": 0.1248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12501925230026245, | |
| "step": 2390, | |
| "valid_targets_mean": 1584.6, | |
| "valid_targets_min": 639 | |
| }, | |
| { | |
| "epoch": 5.884520884520884, | |
| "grad_norm": 0.8356276415586511, | |
| "learning_rate": 4.709248404329625e-08, | |
| "loss": 0.1185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11507894843816757, | |
| "step": 2395, | |
| "valid_targets_mean": 1260.2, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 5.896805896805897, | |
| "grad_norm": 0.902619975905144, | |
| "learning_rate": 3.7795466398868885e-08, | |
| "loss": 0.1201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11913271248340607, | |
| "step": 2400, | |
| "valid_targets_mean": 1240.2, | |
| "valid_targets_min": 633 | |
| }, | |
| { | |
| "epoch": 5.909090909090909, | |
| "grad_norm": 0.8934023630859167, | |
| "learning_rate": 2.9518885975192702e-08, | |
| "loss": 0.1223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12368829548358917, | |
| "step": 2405, | |
| "valid_targets_mean": 1286.1, | |
| "valid_targets_min": 679 | |
| }, | |
| { | |
| "epoch": 5.921375921375922, | |
| "grad_norm": 0.9345968672915899, | |
| "learning_rate": 2.226316585833832e-08, | |
| "loss": 0.1244, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13608361780643463, | |
| "step": 2410, | |
| "valid_targets_mean": 1285.1, | |
| "valid_targets_min": 538 | |
| }, | |
| { | |
| "epoch": 5.933660933660933, | |
| "grad_norm": 0.9107913050491543, | |
| "learning_rate": 1.6028676949570997e-08, | |
| "loss": 0.1272, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1253403127193451, | |
| "step": 2415, | |
| "valid_targets_mean": 1235.6, | |
| "valid_targets_min": 770 | |
| }, | |
| { | |
| "epoch": 5.945945945945946, | |
| "grad_norm": 0.84381108981527, | |
| "learning_rate": 1.0815737946383575e-08, | |
| "loss": 0.1185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11838357150554657, | |
| "step": 2420, | |
| "valid_targets_mean": 1372.6, | |
| "valid_targets_min": 965 | |
| }, | |
| { | |
| "epoch": 5.958230958230958, | |
| "grad_norm": 0.8810683681797915, | |
| "learning_rate": 6.624615326207284e-09, | |
| "loss": 0.1186, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1146058738231659, | |
| "step": 2425, | |
| "valid_targets_mean": 1213.4, | |
| "valid_targets_min": 575 | |
| }, | |
| { | |
| "epoch": 5.9705159705159705, | |
| "grad_norm": 0.7852839819224546, | |
| "learning_rate": 3.4555233327893124e-09, | |
| "loss": 0.1289, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11223162710666656, | |
| "step": 2430, | |
| "valid_targets_mean": 1430.3, | |
| "valid_targets_min": 724 | |
| }, | |
| { | |
| "epoch": 5.982800982800983, | |
| "grad_norm": 0.8528999444132811, | |
| "learning_rate": 1.3086239652415621e-09, | |
| "loss": 0.1189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11062179505825043, | |
| "step": 2435, | |
| "valid_targets_mean": 1334.7, | |
| "valid_targets_min": 751 | |
| }, | |
| { | |
| "epoch": 5.995085995085995, | |
| "grad_norm": 0.8859257547201402, | |
| "learning_rate": 1.840269697628294e-10, | |
| "loss": 0.1209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12078575789928436, | |
| "step": 2440, | |
| "valid_targets_mean": 1338.7, | |
| "valid_targets_min": 687 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12043584883213043, | |
| "step": 2442, | |
| "total_flos": 254043172110336.0, | |
| "train_loss": 0.18853386797248878, | |
| "train_runtime": 8750.677, | |
| "train_samples_per_second": 4.459, | |
| "train_steps_per_second": 0.279, | |
| "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": 254043172110336.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |