| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 6.0, |
| "eval_steps": 500, |
| "global_step": 1188, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.025252525252525252, |
| "grad_norm": 20.169370087659352, |
| "learning_rate": 1.3445378151260504e-06, |
| "loss": 1.0237, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 1.0157525539398193, |
| "step": 5, |
| "valid_targets_mean": 1187.8, |
| "valid_targets_min": 751 |
| }, |
| { |
| "epoch": 0.050505050505050504, |
| "grad_norm": 16.675857749316165, |
| "learning_rate": 3.0252100840336137e-06, |
| "loss": 0.9775, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.9797632694244385, |
| "step": 10, |
| "valid_targets_mean": 1131.0, |
| "valid_targets_min": 704 |
| }, |
| { |
| "epoch": 0.07575757575757576, |
| "grad_norm": 8.704541263565613, |
| "learning_rate": 4.705882352941177e-06, |
| "loss": 0.8599, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.7952851057052612, |
| "step": 15, |
| "valid_targets_mean": 1290.6, |
| "valid_targets_min": 687 |
| }, |
| { |
| "epoch": 0.10101010101010101, |
| "grad_norm": 5.301964048196616, |
| "learning_rate": 6.386554621848739e-06, |
| "loss": 0.7379, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.6894418001174927, |
| "step": 20, |
| "valid_targets_mean": 1500.8, |
| "valid_targets_min": 659 |
| }, |
| { |
| "epoch": 0.12626262626262627, |
| "grad_norm": 2.8651616903258805, |
| "learning_rate": 8.067226890756303e-06, |
| "loss": 0.6167, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.5881441831588745, |
| "step": 25, |
| "valid_targets_mean": 1226.6, |
| "valid_targets_min": 693 |
| }, |
| { |
| "epoch": 0.15151515151515152, |
| "grad_norm": 1.9308802697891598, |
| "learning_rate": 9.747899159663867e-06, |
| "loss": 0.5299, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.5140570998191833, |
| "step": 30, |
| "valid_targets_mean": 1314.4, |
| "valid_targets_min": 532 |
| }, |
| { |
| "epoch": 0.17676767676767677, |
| "grad_norm": 1.6911984939158233, |
| "learning_rate": 1.1428571428571429e-05, |
| "loss": 0.4981, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.49357980489730835, |
| "step": 35, |
| "valid_targets_mean": 1118.2, |
| "valid_targets_min": 680 |
| }, |
| { |
| "epoch": 0.20202020202020202, |
| "grad_norm": 1.492452782606785, |
| "learning_rate": 1.3109243697478993e-05, |
| "loss": 0.4407, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.4345753788948059, |
| "step": 40, |
| "valid_targets_mean": 1273.3, |
| "valid_targets_min": 629 |
| }, |
| { |
| "epoch": 0.22727272727272727, |
| "grad_norm": 1.500112384532108, |
| "learning_rate": 1.4789915966386557e-05, |
| "loss": 0.414, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.3712472915649414, |
| "step": 45, |
| "valid_targets_mean": 1263.1, |
| "valid_targets_min": 620 |
| }, |
| { |
| "epoch": 0.25252525252525254, |
| "grad_norm": 1.0754446138300695, |
| "learning_rate": 1.647058823529412e-05, |
| "loss": 0.3676, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.35347896814346313, |
| "step": 50, |
| "valid_targets_mean": 1464.4, |
| "valid_targets_min": 659 |
| }, |
| { |
| "epoch": 0.2777777777777778, |
| "grad_norm": 1.1618002004795733, |
| "learning_rate": 1.8151260504201682e-05, |
| "loss": 0.3464, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.3541208803653717, |
| "step": 55, |
| "valid_targets_mean": 1315.9, |
| "valid_targets_min": 589 |
| }, |
| { |
| "epoch": 0.30303030303030304, |
| "grad_norm": 1.0883564166727384, |
| "learning_rate": 1.9831932773109244e-05, |
| "loss": 0.338, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.3574886620044708, |
| "step": 60, |
| "valid_targets_mean": 1405.6, |
| "valid_targets_min": 758 |
| }, |
| { |
| "epoch": 0.3282828282828283, |
| "grad_norm": 1.0137280896162002, |
| "learning_rate": 2.1512605042016807e-05, |
| "loss": 0.3275, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.3133365511894226, |
| "step": 65, |
| "valid_targets_mean": 1416.2, |
| "valid_targets_min": 753 |
| }, |
| { |
| "epoch": 0.35353535353535354, |
| "grad_norm": 1.0379235855718125, |
| "learning_rate": 2.3193277310924373e-05, |
| "loss": 0.2979, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.3172857463359833, |
| "step": 70, |
| "valid_targets_mean": 1387.2, |
| "valid_targets_min": 615 |
| }, |
| { |
| "epoch": 0.3787878787878788, |
| "grad_norm": 1.0676260962261874, |
| "learning_rate": 2.4873949579831935e-05, |
| "loss": 0.2957, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2801944613456726, |
| "step": 75, |
| "valid_targets_mean": 1275.1, |
| "valid_targets_min": 831 |
| }, |
| { |
| "epoch": 0.40404040404040403, |
| "grad_norm": 1.0045547818037883, |
| "learning_rate": 2.6554621848739497e-05, |
| "loss": 0.3063, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.3241528272628784, |
| "step": 80, |
| "valid_targets_mean": 1368.9, |
| "valid_targets_min": 846 |
| }, |
| { |
| "epoch": 0.4292929292929293, |
| "grad_norm": 0.9923280106185243, |
| "learning_rate": 2.8235294117647063e-05, |
| "loss": 0.2973, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2852829396724701, |
| "step": 85, |
| "valid_targets_mean": 1345.0, |
| "valid_targets_min": 735 |
| }, |
| { |
| "epoch": 0.45454545454545453, |
| "grad_norm": 0.9646459041819732, |
| "learning_rate": 2.9915966386554626e-05, |
| "loss": 0.2854, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2792156934738159, |
| "step": 90, |
| "valid_targets_mean": 1343.4, |
| "valid_targets_min": 738 |
| }, |
| { |
| "epoch": 0.4797979797979798, |
| "grad_norm": 1.052289746435847, |
| "learning_rate": 3.159663865546219e-05, |
| "loss": 0.2698, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.26299500465393066, |
| "step": 95, |
| "valid_targets_mean": 1293.9, |
| "valid_targets_min": 773 |
| }, |
| { |
| "epoch": 0.5050505050505051, |
| "grad_norm": 1.1622301238597719, |
| "learning_rate": 3.3277310924369754e-05, |
| "loss": 0.2831, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.311882883310318, |
| "step": 100, |
| "valid_targets_mean": 1408.7, |
| "valid_targets_min": 738 |
| }, |
| { |
| "epoch": 0.5303030303030303, |
| "grad_norm": 1.1114150871374988, |
| "learning_rate": 3.495798319327731e-05, |
| "loss": 0.2821, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.275522381067276, |
| "step": 105, |
| "valid_targets_mean": 1212.6, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 0.5555555555555556, |
| "grad_norm": 0.9706083743298367, |
| "learning_rate": 3.663865546218488e-05, |
| "loss": 0.2799, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2872806489467621, |
| "step": 110, |
| "valid_targets_mean": 1481.1, |
| "valid_targets_min": 918 |
| }, |
| { |
| "epoch": 0.5808080808080808, |
| "grad_norm": 1.1271397120654052, |
| "learning_rate": 3.8319327731092444e-05, |
| "loss": 0.279, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2863199710845947, |
| "step": 115, |
| "valid_targets_mean": 1283.5, |
| "valid_targets_min": 574 |
| }, |
| { |
| "epoch": 0.6060606060606061, |
| "grad_norm": 1.0471857507926354, |
| "learning_rate": 4e-05, |
| "loss": 0.2694, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.29653966426849365, |
| "step": 120, |
| "valid_targets_mean": 1340.8, |
| "valid_targets_min": 756 |
| }, |
| { |
| "epoch": 0.6313131313131313, |
| "grad_norm": 1.1072262420802004, |
| "learning_rate": 3.999784088124753e-05, |
| "loss": 0.2793, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.28706759214401245, |
| "step": 125, |
| "valid_targets_mean": 1297.0, |
| "valid_targets_min": 784 |
| }, |
| { |
| "epoch": 0.6565656565656566, |
| "grad_norm": 0.9900053971903253, |
| "learning_rate": 3.99913639911695e-05, |
| "loss": 0.2597, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2584039866924286, |
| "step": 130, |
| "valid_targets_mean": 1323.2, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 0.6818181818181818, |
| "grad_norm": 0.9452591379939352, |
| "learning_rate": 3.998057072820338e-05, |
| "loss": 0.2705, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2791949212551117, |
| "step": 135, |
| "valid_targets_mean": 1479.4, |
| "valid_targets_min": 582 |
| }, |
| { |
| "epoch": 0.7070707070707071, |
| "grad_norm": 1.1345903226168657, |
| "learning_rate": 3.996546342274282e-05, |
| "loss": 0.2753, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.270517498254776, |
| "step": 140, |
| "valid_targets_mean": 1323.2, |
| "valid_targets_min": 797 |
| }, |
| { |
| "epoch": 0.7323232323232324, |
| "grad_norm": 1.0964303784318785, |
| "learning_rate": 3.9946045336634485e-05, |
| "loss": 0.2598, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.25797197222709656, |
| "step": 145, |
| "valid_targets_mean": 1143.2, |
| "valid_targets_min": 632 |
| }, |
| { |
| "epoch": 0.7575757575757576, |
| "grad_norm": 1.0480327343570168, |
| "learning_rate": 3.9922320662473755e-05, |
| "loss": 0.2704, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2590729296207428, |
| "step": 150, |
| "valid_targets_mean": 1109.0, |
| "valid_targets_min": 555 |
| }, |
| { |
| "epoch": 0.7828282828282829, |
| "grad_norm": 1.2063540633588965, |
| "learning_rate": 3.989429452269951e-05, |
| "loss": 0.2545, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.24084429442882538, |
| "step": 155, |
| "valid_targets_mean": 1144.1, |
| "valid_targets_min": 685 |
| }, |
| { |
| "epoch": 0.8080808080808081, |
| "grad_norm": 1.1381600500975266, |
| "learning_rate": 3.986197296848816e-05, |
| "loss": 0.2619, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.26328036189079285, |
| "step": 160, |
| "valid_targets_mean": 1215.6, |
| "valid_targets_min": 717 |
| }, |
| { |
| "epoch": 0.8333333333333334, |
| "grad_norm": 0.9369305409921032, |
| "learning_rate": 3.982536297844706e-05, |
| "loss": 0.2628, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.25279849767684937, |
| "step": 165, |
| "valid_targets_mean": 1300.6, |
| "valid_targets_min": 679 |
| }, |
| { |
| "epoch": 0.8585858585858586, |
| "grad_norm": 0.8800799927020014, |
| "learning_rate": 3.978447245710784e-05, |
| "loss": 0.2588, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.25856029987335205, |
| "step": 170, |
| "valid_targets_mean": 1579.3, |
| "valid_targets_min": 821 |
| }, |
| { |
| "epoch": 0.8838383838383839, |
| "grad_norm": 0.9268097877772618, |
| "learning_rate": 3.973931023321962e-05, |
| "loss": 0.2538, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.24270369112491608, |
| "step": 175, |
| "valid_targets_mean": 1203.4, |
| "valid_targets_min": 684 |
| }, |
| { |
| "epoch": 0.9090909090909091, |
| "grad_norm": 0.9474619522436488, |
| "learning_rate": 3.9689886057842866e-05, |
| "loss": 0.2409, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.25069937109947205, |
| "step": 180, |
| "valid_targets_mean": 1398.1, |
| "valid_targets_min": 711 |
| }, |
| { |
| "epoch": 0.9343434343434344, |
| "grad_norm": 1.034637445880305, |
| "learning_rate": 3.963621060224396e-05, |
| "loss": 0.2545, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.24697241187095642, |
| "step": 185, |
| "valid_targets_mean": 1112.4, |
| "valid_targets_min": 582 |
| }, |
| { |
| "epoch": 0.9595959595959596, |
| "grad_norm": 0.9298977899777117, |
| "learning_rate": 3.957829545559118e-05, |
| "loss": 0.2553, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2632555365562439, |
| "step": 190, |
| "valid_targets_mean": 1315.6, |
| "valid_targets_min": 742 |
| }, |
| { |
| "epoch": 0.9848484848484849, |
| "grad_norm": 0.8977159403395447, |
| "learning_rate": 3.951615312245243e-05, |
| "loss": 0.2546, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2525337338447571, |
| "step": 195, |
| "valid_targets_mean": 1346.2, |
| "valid_targets_min": 774 |
| }, |
| { |
| "epoch": 1.0101010101010102, |
| "grad_norm": 1.021475627213122, |
| "learning_rate": 3.9449797020095404e-05, |
| "loss": 0.2535, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2486366480588913, |
| "step": 200, |
| "valid_targets_mean": 1331.2, |
| "valid_targets_min": 680 |
| }, |
| { |
| "epoch": 1.0353535353535352, |
| "grad_norm": 1.0071426224742983, |
| "learning_rate": 3.937924147559059e-05, |
| "loss": 0.2277, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2366667240858078, |
| "step": 205, |
| "valid_targets_mean": 1303.9, |
| "valid_targets_min": 648 |
| }, |
| { |
| "epoch": 1.0606060606060606, |
| "grad_norm": 0.863916533755164, |
| "learning_rate": 3.9304501722717925e-05, |
| "loss": 0.2358, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22564607858657837, |
| "step": 210, |
| "valid_targets_mean": 1418.2, |
| "valid_targets_min": 1067 |
| }, |
| { |
| "epoch": 1.0858585858585859, |
| "grad_norm": 0.932084360690738, |
| "learning_rate": 3.922559389867758e-05, |
| "loss": 0.2333, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.24001938104629517, |
| "step": 215, |
| "valid_targets_mean": 1553.2, |
| "valid_targets_min": 1094 |
| }, |
| { |
| "epoch": 1.1111111111111112, |
| "grad_norm": 0.9418003911519435, |
| "learning_rate": 3.914253504060585e-05, |
| "loss": 0.2289, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2039659321308136, |
| "step": 220, |
| "valid_targets_mean": 1241.6, |
| "valid_targets_min": 594 |
| }, |
| { |
| "epoch": 1.1363636363636362, |
| "grad_norm": 0.9424926403942966, |
| "learning_rate": 3.90553430818965e-05, |
| "loss": 0.225, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22819814085960388, |
| "step": 225, |
| "valid_targets_mean": 1457.8, |
| "valid_targets_min": 837 |
| }, |
| { |
| "epoch": 1.1616161616161615, |
| "grad_norm": 0.8372692329391297, |
| "learning_rate": 3.896403684832887e-05, |
| "loss": 0.23, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22280089557170868, |
| "step": 230, |
| "valid_targets_mean": 1298.8, |
| "valid_targets_min": 791 |
| }, |
| { |
| "epoch": 1.1868686868686869, |
| "grad_norm": 0.9634864270568145, |
| "learning_rate": 3.886863605400306e-05, |
| "loss": 0.2314, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.24971917271614075, |
| "step": 235, |
| "valid_targets_mean": 1264.4, |
| "valid_targets_min": 700 |
| }, |
| { |
| "epoch": 1.2121212121212122, |
| "grad_norm": 0.9062647387121051, |
| "learning_rate": 3.876916129708347e-05, |
| "loss": 0.2231, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22650766372680664, |
| "step": 240, |
| "valid_targets_mean": 1429.9, |
| "valid_targets_min": 854 |
| }, |
| { |
| "epoch": 1.2373737373737375, |
| "grad_norm": 0.9679426116681048, |
| "learning_rate": 3.866563405535142e-05, |
| "loss": 0.2363, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22411300241947174, |
| "step": 245, |
| "valid_targets_mean": 1249.2, |
| "valid_targets_min": 550 |
| }, |
| { |
| "epoch": 1.2626262626262625, |
| "grad_norm": 1.5210230333522776, |
| "learning_rate": 3.855807668156779e-05, |
| "loss": 0.2187, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.23495987057685852, |
| "step": 250, |
| "valid_targets_mean": 1417.0, |
| "valid_targets_min": 720 |
| }, |
| { |
| "epoch": 1.2878787878787878, |
| "grad_norm": 1.0154197633890338, |
| "learning_rate": 3.844651239864687e-05, |
| "loss": 0.2273, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22711053490638733, |
| "step": 255, |
| "valid_targets_mean": 1173.8, |
| "valid_targets_min": 776 |
| }, |
| { |
| "epoch": 1.3131313131313131, |
| "grad_norm": 0.9259864099946552, |
| "learning_rate": 3.8330965294642186e-05, |
| "loss": 0.227, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2507520318031311, |
| "step": 260, |
| "valid_targets_mean": 1453.4, |
| "valid_targets_min": 711 |
| }, |
| { |
| "epoch": 1.3383838383838385, |
| "grad_norm": 0.8413676808910939, |
| "learning_rate": 3.821146031754565e-05, |
| "loss": 0.2331, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.235780268907547, |
| "step": 265, |
| "valid_targets_mean": 1401.7, |
| "valid_targets_min": 681 |
| }, |
| { |
| "epoch": 1.3636363636363638, |
| "grad_norm": 0.9876932162935549, |
| "learning_rate": 3.808802326990096e-05, |
| "loss": 0.2198, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21561799943447113, |
| "step": 270, |
| "valid_targets_mean": 1162.8, |
| "valid_targets_min": 605 |
| }, |
| { |
| "epoch": 1.3888888888888888, |
| "grad_norm": 1.1186277755619236, |
| "learning_rate": 3.7960680803232544e-05, |
| "loss": 0.2212, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2476837933063507, |
| "step": 275, |
| "valid_targets_mean": 1249.6, |
| "valid_targets_min": 644 |
| }, |
| { |
| "epoch": 1.4141414141414141, |
| "grad_norm": 0.9272391389819834, |
| "learning_rate": 3.782946041229119e-05, |
| "loss": 0.2239, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22536993026733398, |
| "step": 280, |
| "valid_targets_mean": 1286.6, |
| "valid_targets_min": 671 |
| }, |
| { |
| "epoch": 1.4393939393939394, |
| "grad_norm": 0.9507970993940866, |
| "learning_rate": 3.769439042911758e-05, |
| "loss": 0.234, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.24206951260566711, |
| "step": 285, |
| "valid_targets_mean": 1180.0, |
| "valid_targets_min": 711 |
| }, |
| { |
| "epoch": 1.4646464646464645, |
| "grad_norm": 0.9667877163253019, |
| "learning_rate": 3.755550001692506e-05, |
| "loss": 0.2173, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21246126294136047, |
| "step": 290, |
| "valid_targets_mean": 1220.2, |
| "valid_targets_min": 770 |
| }, |
| { |
| "epoch": 1.4898989898989898, |
| "grad_norm": 0.9116365111809664, |
| "learning_rate": 3.7412819163802986e-05, |
| "loss": 0.2105, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21166947484016418, |
| "step": 295, |
| "valid_targets_mean": 1263.8, |
| "valid_targets_min": 762 |
| }, |
| { |
| "epoch": 1.5151515151515151, |
| "grad_norm": 0.8811979745066287, |
| "learning_rate": 3.726637867624191e-05, |
| "loss": 0.2359, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20912611484527588, |
| "step": 300, |
| "valid_targets_mean": 1320.4, |
| "valid_targets_min": 683 |
| }, |
| { |
| "epoch": 1.5404040404040404, |
| "grad_norm": 0.9596648525111655, |
| "learning_rate": 3.711621017248213e-05, |
| "loss": 0.2152, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20625783503055573, |
| "step": 305, |
| "valid_targets_mean": 1253.4, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 1.5656565656565657, |
| "grad_norm": 0.9622701229690029, |
| "learning_rate": 3.6962346075686876e-05, |
| "loss": 0.2259, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2124948799610138, |
| "step": 310, |
| "valid_targets_mean": 1200.3, |
| "valid_targets_min": 655 |
| }, |
| { |
| "epoch": 1.5909090909090908, |
| "grad_norm": 1.016874652742493, |
| "learning_rate": 3.680481960694183e-05, |
| "loss": 0.2276, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21283963322639465, |
| "step": 315, |
| "valid_targets_mean": 1326.4, |
| "valid_targets_min": 721 |
| }, |
| { |
| "epoch": 1.6161616161616161, |
| "grad_norm": 0.8226357783546479, |
| "learning_rate": 3.6643664778082254e-05, |
| "loss": 0.2275, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20702260732650757, |
| "step": 320, |
| "valid_targets_mean": 1401.2, |
| "valid_targets_min": 868 |
| }, |
| { |
| "epoch": 1.6414141414141414, |
| "grad_norm": 0.8672807532299297, |
| "learning_rate": 3.6478916384349465e-05, |
| "loss": 0.2193, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.23550966382026672, |
| "step": 325, |
| "valid_targets_mean": 1560.4, |
| "valid_targets_min": 825 |
| }, |
| { |
| "epoch": 1.6666666666666665, |
| "grad_norm": 0.9712639328964777, |
| "learning_rate": 3.631060999687809e-05, |
| "loss": 0.2243, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22655721008777618, |
| "step": 330, |
| "valid_targets_mean": 1239.7, |
| "valid_targets_min": 704 |
| }, |
| { |
| "epoch": 1.691919191919192, |
| "grad_norm": 0.9069646079111703, |
| "learning_rate": 3.613878195501586e-05, |
| "loss": 0.2132, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20803983509540558, |
| "step": 335, |
| "valid_targets_mean": 1168.1, |
| "valid_targets_min": 670 |
| }, |
| { |
| "epoch": 1.7171717171717171, |
| "grad_norm": 0.9455557791381117, |
| "learning_rate": 3.596346935847752e-05, |
| "loss": 0.2148, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21128031611442566, |
| "step": 340, |
| "valid_targets_mean": 1368.7, |
| "valid_targets_min": 753 |
| }, |
| { |
| "epoch": 1.7424242424242424, |
| "grad_norm": 1.1078047614048518, |
| "learning_rate": 3.578471005933454e-05, |
| "loss": 0.225, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2151433229446411, |
| "step": 345, |
| "valid_targets_mean": 1172.9, |
| "valid_targets_min": 764 |
| }, |
| { |
| "epoch": 1.7676767676767677, |
| "grad_norm": 1.025649374731909, |
| "learning_rate": 3.5602542653842416e-05, |
| "loss": 0.2148, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21121200919151306, |
| "step": 350, |
| "valid_targets_mean": 1404.9, |
| "valid_targets_min": 802 |
| }, |
| { |
| "epoch": 1.7929292929292928, |
| "grad_norm": 0.9227312319452774, |
| "learning_rate": 3.541700647410728e-05, |
| "loss": 0.2203, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20654287934303284, |
| "step": 355, |
| "valid_targets_mean": 1270.9, |
| "valid_targets_min": 694 |
| }, |
| { |
| "epoch": 1.8181818181818183, |
| "grad_norm": 0.8839026808611836, |
| "learning_rate": 3.522814157959362e-05, |
| "loss": 0.2307, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21429893374443054, |
| "step": 360, |
| "valid_targets_mean": 1328.3, |
| "valid_targets_min": 774 |
| }, |
| { |
| "epoch": 1.8434343434343434, |
| "grad_norm": 0.8418961695574284, |
| "learning_rate": 3.5035988748474974e-05, |
| "loss": 0.23, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2303229570388794, |
| "step": 365, |
| "valid_targets_mean": 1438.8, |
| "valid_targets_min": 801 |
| }, |
| { |
| "epoch": 1.8686868686868687, |
| "grad_norm": 0.8660276681232518, |
| "learning_rate": 3.484058946882946e-05, |
| "loss": 0.2128, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20196302235126495, |
| "step": 370, |
| "valid_targets_mean": 1269.9, |
| "valid_targets_min": 765 |
| }, |
| { |
| "epoch": 1.893939393939394, |
| "grad_norm": 0.8556571315590435, |
| "learning_rate": 3.4641985929681954e-05, |
| "loss": 0.221, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2154424786567688, |
| "step": 375, |
| "valid_targets_mean": 1228.6, |
| "valid_targets_min": 738 |
| }, |
| { |
| "epoch": 1.9191919191919191, |
| "grad_norm": 0.9057324926130037, |
| "learning_rate": 3.4440221011895026e-05, |
| "loss": 0.2149, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22812026739120483, |
| "step": 380, |
| "valid_targets_mean": 1510.7, |
| "valid_targets_min": 648 |
| }, |
| { |
| "epoch": 1.9444444444444444, |
| "grad_norm": 1.22905062391175, |
| "learning_rate": 3.423533827891044e-05, |
| "loss": 0.2226, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.22331665456295013, |
| "step": 385, |
| "valid_targets_mean": 1256.8, |
| "valid_targets_min": 796 |
| }, |
| { |
| "epoch": 1.9696969696969697, |
| "grad_norm": 0.8864960651317679, |
| "learning_rate": 3.402738196734327e-05, |
| "loss": 0.2269, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2474365234375, |
| "step": 390, |
| "valid_targets_mean": 1468.1, |
| "valid_targets_min": 808 |
| }, |
| { |
| "epoch": 1.9949494949494948, |
| "grad_norm": 0.8116444633238107, |
| "learning_rate": 3.381639697743073e-05, |
| "loss": 0.2196, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21921899914741516, |
| "step": 395, |
| "valid_targets_mean": 1372.4, |
| "valid_targets_min": 696 |
| }, |
| { |
| "epoch": 2.0202020202020203, |
| "grad_norm": 0.8089562588991992, |
| "learning_rate": 3.3602428863337625e-05, |
| "loss": 0.1951, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18117868900299072, |
| "step": 400, |
| "valid_targets_mean": 1378.9, |
| "valid_targets_min": 830 |
| }, |
| { |
| "epoch": 2.0454545454545454, |
| "grad_norm": 0.9722779083533737, |
| "learning_rate": 3.338552382332073e-05, |
| "loss": 0.1904, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20024555921554565, |
| "step": 405, |
| "valid_targets_mean": 1251.8, |
| "valid_targets_min": 644 |
| }, |
| { |
| "epoch": 2.0707070707070705, |
| "grad_norm": 0.9539651111813244, |
| "learning_rate": 3.3165728689753976e-05, |
| "loss": 0.1952, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1934088170528412, |
| "step": 410, |
| "valid_targets_mean": 1190.8, |
| "valid_targets_min": 655 |
| }, |
| { |
| "epoch": 2.095959595959596, |
| "grad_norm": 0.8845502621948085, |
| "learning_rate": 3.2943090919016815e-05, |
| "loss": 0.1938, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21735027432441711, |
| "step": 415, |
| "valid_targets_mean": 1383.0, |
| "valid_targets_min": 651 |
| }, |
| { |
| "epoch": 2.121212121212121, |
| "grad_norm": 0.8702714930026663, |
| "learning_rate": 3.2717658581247844e-05, |
| "loss": 0.1834, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17540085315704346, |
| "step": 420, |
| "valid_targets_mean": 1534.4, |
| "valid_targets_min": 686 |
| }, |
| { |
| "epoch": 2.1464646464646466, |
| "grad_norm": 0.8478046646870934, |
| "learning_rate": 3.248948034996583e-05, |
| "loss": 0.1995, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19789089262485504, |
| "step": 425, |
| "valid_targets_mean": 1489.8, |
| "valid_targets_min": 578 |
| }, |
| { |
| "epoch": 2.1717171717171717, |
| "grad_norm": 0.8469812735159056, |
| "learning_rate": 3.2258605491560606e-05, |
| "loss": 0.184, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17873477935791016, |
| "step": 430, |
| "valid_targets_mean": 1449.3, |
| "valid_targets_min": 964 |
| }, |
| { |
| "epoch": 2.196969696969697, |
| "grad_norm": 1.0152996119033395, |
| "learning_rate": 3.2025083854655776e-05, |
| "loss": 0.1889, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19385069608688354, |
| "step": 435, |
| "valid_targets_mean": 1232.6, |
| "valid_targets_min": 532 |
| }, |
| { |
| "epoch": 2.2222222222222223, |
| "grad_norm": 0.9854965513455352, |
| "learning_rate": 3.178896585934588e-05, |
| "loss": 0.1922, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20682978630065918, |
| "step": 440, |
| "valid_targets_mean": 1247.1, |
| "valid_targets_min": 708 |
| }, |
| { |
| "epoch": 2.2474747474747474, |
| "grad_norm": 0.8783905199567699, |
| "learning_rate": 3.1550302486310076e-05, |
| "loss": 0.181, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17547810077667236, |
| "step": 445, |
| "valid_targets_mean": 1395.0, |
| "valid_targets_min": 986 |
| }, |
| { |
| "epoch": 2.2727272727272725, |
| "grad_norm": 0.7915351995988669, |
| "learning_rate": 3.130914526580478e-05, |
| "loss": 0.1932, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19163087010383606, |
| "step": 450, |
| "valid_targets_mean": 1419.6, |
| "valid_targets_min": 550 |
| }, |
| { |
| "epoch": 2.297979797979798, |
| "grad_norm": 0.8603558279215641, |
| "learning_rate": 3.10655462665377e-05, |
| "loss": 0.1943, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18317505717277527, |
| "step": 455, |
| "valid_targets_mean": 1478.6, |
| "valid_targets_min": 796 |
| }, |
| { |
| "epoch": 2.323232323232323, |
| "grad_norm": 0.9291992362997976, |
| "learning_rate": 3.0819558084425574e-05, |
| "loss": 0.1858, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1851797103881836, |
| "step": 460, |
| "valid_targets_mean": 1252.2, |
| "valid_targets_min": 862 |
| }, |
| { |
| "epoch": 2.3484848484848486, |
| "grad_norm": 0.9994111902225407, |
| "learning_rate": 3.0571233831238093e-05, |
| "loss": 0.1912, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1833767294883728, |
| "step": 465, |
| "valid_targets_mean": 1134.1, |
| "valid_targets_min": 619 |
| }, |
| { |
| "epoch": 2.3737373737373737, |
| "grad_norm": 0.9400802045489894, |
| "learning_rate": 3.032062712313044e-05, |
| "loss": 0.1907, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19591230154037476, |
| "step": 470, |
| "valid_targets_mean": 1326.4, |
| "valid_targets_min": 801 |
| }, |
| { |
| "epoch": 2.398989898989899, |
| "grad_norm": 0.8710505639917704, |
| "learning_rate": 3.0067792069066902e-05, |
| "loss": 0.1963, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.21165084838867188, |
| "step": 475, |
| "valid_targets_mean": 1496.6, |
| "valid_targets_min": 774 |
| }, |
| { |
| "epoch": 2.4242424242424243, |
| "grad_norm": 0.8977896309267136, |
| "learning_rate": 2.9812783259138133e-05, |
| "loss": 0.2008, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20881733298301697, |
| "step": 480, |
| "valid_targets_mean": 1376.1, |
| "valid_targets_min": 600 |
| }, |
| { |
| "epoch": 2.4494949494949494, |
| "grad_norm": 0.8422125060200054, |
| "learning_rate": 2.955565575277449e-05, |
| "loss": 0.1868, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18488860130310059, |
| "step": 485, |
| "valid_targets_mean": 1432.2, |
| "valid_targets_min": 585 |
| }, |
| { |
| "epoch": 2.474747474747475, |
| "grad_norm": 0.841379053094665, |
| "learning_rate": 2.929646506685805e-05, |
| "loss": 0.1896, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19385570287704468, |
| "step": 490, |
| "valid_targets_mean": 1452.9, |
| "valid_targets_min": 811 |
| }, |
| { |
| "epoch": 2.5, |
| "grad_norm": 0.8681195861479708, |
| "learning_rate": 2.9035267163735856e-05, |
| "loss": 0.1853, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18989858031272888, |
| "step": 495, |
| "valid_targets_mean": 1264.6, |
| "valid_targets_min": 674 |
| }, |
| { |
| "epoch": 2.525252525252525, |
| "grad_norm": 0.9486462794838508, |
| "learning_rate": 2.8772118439136972e-05, |
| "loss": 0.1917, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20061013102531433, |
| "step": 500, |
| "valid_targets_mean": 1308.9, |
| "valid_targets_min": 854 |
| }, |
| { |
| "epoch": 2.5505050505050506, |
| "grad_norm": 1.0120219125127008, |
| "learning_rate": 2.8507075709996015e-05, |
| "loss": 0.2, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19788292050361633, |
| "step": 505, |
| "valid_targets_mean": 1336.9, |
| "valid_targets_min": 615 |
| }, |
| { |
| "epoch": 2.5757575757575757, |
| "grad_norm": 0.8552983942324871, |
| "learning_rate": 2.8240196202185636e-05, |
| "loss": 0.1841, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18201503157615662, |
| "step": 510, |
| "valid_targets_mean": 1183.4, |
| "valid_targets_min": 648 |
| }, |
| { |
| "epoch": 2.601010101010101, |
| "grad_norm": 1.0101472372139235, |
| "learning_rate": 2.797153753816084e-05, |
| "loss": 0.2015, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18643571436405182, |
| "step": 515, |
| "valid_targets_mean": 1155.6, |
| "valid_targets_min": 538 |
| }, |
| { |
| "epoch": 2.6262626262626263, |
| "grad_norm": 0.8259754112301633, |
| "learning_rate": 2.770115772451758e-05, |
| "loss": 0.1847, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20032909512519836, |
| "step": 520, |
| "valid_targets_mean": 1452.1, |
| "valid_targets_min": 640 |
| }, |
| { |
| "epoch": 2.6515151515151514, |
| "grad_norm": 1.0080765247154564, |
| "learning_rate": 2.7429115139468443e-05, |
| "loss": 0.1868, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2050439417362213, |
| "step": 525, |
| "valid_targets_mean": 1231.2, |
| "valid_targets_min": 678 |
| }, |
| { |
| "epoch": 2.676767676767677, |
| "grad_norm": 0.9228666377959548, |
| "learning_rate": 2.7155468520238116e-05, |
| "loss": 0.1955, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20735567808151245, |
| "step": 530, |
| "valid_targets_mean": 1336.9, |
| "valid_targets_min": 587 |
| }, |
| { |
| "epoch": 2.702020202020202, |
| "grad_norm": 0.997304280142903, |
| "learning_rate": 2.6880276950381316e-05, |
| "loss": 0.1881, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1983107328414917, |
| "step": 535, |
| "valid_targets_mean": 1249.5, |
| "valid_targets_min": 521 |
| }, |
| { |
| "epoch": 2.7272727272727275, |
| "grad_norm": 0.9193727742835162, |
| "learning_rate": 2.6603599847025935e-05, |
| "loss": 0.1805, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18839475512504578, |
| "step": 540, |
| "valid_targets_mean": 1276.8, |
| "valid_targets_min": 776 |
| }, |
| { |
| "epoch": 2.7525252525252526, |
| "grad_norm": 0.9459373491774038, |
| "learning_rate": 2.63254969480442e-05, |
| "loss": 0.191, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19836702942848206, |
| "step": 545, |
| "valid_targets_mean": 1273.7, |
| "valid_targets_min": 702 |
| }, |
| { |
| "epoch": 2.7777777777777777, |
| "grad_norm": 0.8854618406463927, |
| "learning_rate": 2.6046028299154545e-05, |
| "loss": 0.1875, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18734903633594513, |
| "step": 550, |
| "valid_targets_mean": 1388.6, |
| "valid_targets_min": 546 |
| }, |
| { |
| "epoch": 2.8030303030303028, |
| "grad_norm": 0.7860179708066867, |
| "learning_rate": 2.5765254240957024e-05, |
| "loss": 0.1886, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18938644230365753, |
| "step": 555, |
| "valid_targets_mean": 1631.9, |
| "valid_targets_min": 1077 |
| }, |
| { |
| "epoch": 2.8282828282828283, |
| "grad_norm": 0.8367854200810775, |
| "learning_rate": 2.5483235395905056e-05, |
| "loss": 0.1867, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.20430481433868408, |
| "step": 560, |
| "valid_targets_mean": 1437.8, |
| "valid_targets_min": 690 |
| }, |
| { |
| "epoch": 2.8535353535353534, |
| "grad_norm": 0.9180703166963079, |
| "learning_rate": 2.5200032655216343e-05, |
| "loss": 0.1894, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18853136897087097, |
| "step": 565, |
| "valid_targets_mean": 1259.7, |
| "valid_targets_min": 684 |
| }, |
| { |
| "epoch": 2.878787878787879, |
| "grad_norm": 0.8164719253098022, |
| "learning_rate": 2.4915707165725694e-05, |
| "loss": 0.1858, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18391425907611847, |
| "step": 570, |
| "valid_targets_mean": 1376.6, |
| "valid_targets_min": 802 |
| }, |
| { |
| "epoch": 2.904040404040404, |
| "grad_norm": 0.9943903226603121, |
| "learning_rate": 2.4630320316682724e-05, |
| "loss": 0.1847, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.19583845138549805, |
| "step": 575, |
| "valid_targets_mean": 1316.8, |
| "valid_targets_min": 685 |
| }, |
| { |
| "epoch": 2.929292929292929, |
| "grad_norm": 0.9662414433511091, |
| "learning_rate": 2.4343933726497183e-05, |
| "loss": 0.1856, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17909899353981018, |
| "step": 580, |
| "valid_targets_mean": 1219.6, |
| "valid_targets_min": 603 |
| }, |
| { |
| "epoch": 2.9545454545454546, |
| "grad_norm": 0.8264667060320464, |
| "learning_rate": 2.4056609229434812e-05, |
| "loss": 0.1887, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1930309534072876, |
| "step": 585, |
| "valid_targets_mean": 1353.8, |
| "valid_targets_min": 751 |
| }, |
| { |
| "epoch": 2.9797979797979797, |
| "grad_norm": 0.8672579031747749, |
| "learning_rate": 2.376840886226656e-05, |
| "loss": 0.1867, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1810949146747589, |
| "step": 590, |
| "valid_targets_mean": 1360.6, |
| "valid_targets_min": 681 |
| }, |
| { |
| "epoch": 3.005050505050505, |
| "grad_norm": 0.8370979896650649, |
| "learning_rate": 2.347939485087416e-05, |
| "loss": 0.1831, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16158387064933777, |
| "step": 595, |
| "valid_targets_mean": 1132.1, |
| "valid_targets_min": 475 |
| }, |
| { |
| "epoch": 3.0303030303030303, |
| "grad_norm": 0.8281295886895214, |
| "learning_rate": 2.3189629596814788e-05, |
| "loss": 0.157, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15015463531017303, |
| "step": 600, |
| "valid_targets_mean": 1357.4, |
| "valid_targets_min": 629 |
| }, |
| { |
| "epoch": 3.0555555555555554, |
| "grad_norm": 1.0077015146202375, |
| "learning_rate": 2.2899175663847823e-05, |
| "loss": 0.1605, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17284418642520905, |
| "step": 605, |
| "valid_targets_mean": 1323.3, |
| "valid_targets_min": 681 |
| }, |
| { |
| "epoch": 3.080808080808081, |
| "grad_norm": 0.9055532065630415, |
| "learning_rate": 2.2608095764426602e-05, |
| "loss": 0.1609, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.172052800655365, |
| "step": 610, |
| "valid_targets_mean": 1438.2, |
| "valid_targets_min": 785 |
| }, |
| { |
| "epoch": 3.106060606060606, |
| "grad_norm": 0.9289877404583321, |
| "learning_rate": 2.2316452746158063e-05, |
| "loss": 0.1649, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.18071796000003815, |
| "step": 615, |
| "valid_targets_mean": 1361.3, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 3.1313131313131315, |
| "grad_norm": 0.9997329115169467, |
| "learning_rate": 2.2024309578233174e-05, |
| "loss": 0.161, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16634033620357513, |
| "step": 620, |
| "valid_targets_mean": 1173.6, |
| "valid_targets_min": 690 |
| }, |
| { |
| "epoch": 3.1565656565656566, |
| "grad_norm": 1.009976573995706, |
| "learning_rate": 2.1731729337831173e-05, |
| "loss": 0.1574, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16698682308197021, |
| "step": 625, |
| "valid_targets_mean": 1469.7, |
| "valid_targets_min": 677 |
| }, |
| { |
| "epoch": 3.1818181818181817, |
| "grad_norm": 0.9404190990393296, |
| "learning_rate": 2.143877519650042e-05, |
| "loss": 0.165, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16389097273349762, |
| "step": 630, |
| "valid_targets_mean": 1243.7, |
| "valid_targets_min": 655 |
| }, |
| { |
| "epoch": 3.207070707070707, |
| "grad_norm": 0.9313377151581581, |
| "learning_rate": 2.1145510406518928e-05, |
| "loss": 0.1654, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17033018171787262, |
| "step": 635, |
| "valid_targets_mean": 1205.6, |
| "valid_targets_min": 742 |
| }, |
| { |
| "epoch": 3.2323232323232323, |
| "grad_norm": 0.8337581545510251, |
| "learning_rate": 2.0851998287237452e-05, |
| "loss": 0.1603, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14411024749279022, |
| "step": 640, |
| "valid_targets_mean": 1329.4, |
| "valid_targets_min": 746 |
| }, |
| { |
| "epoch": 3.257575757575758, |
| "grad_norm": 0.8441677147621939, |
| "learning_rate": 2.0558302211408075e-05, |
| "loss": 0.1568, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15460258722305298, |
| "step": 645, |
| "valid_targets_mean": 1355.2, |
| "valid_targets_min": 684 |
| }, |
| { |
| "epoch": 3.282828282828283, |
| "grad_norm": 0.894454117393474, |
| "learning_rate": 2.0264485591501272e-05, |
| "loss": 0.1657, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1565285176038742, |
| "step": 650, |
| "valid_targets_mean": 1227.0, |
| "valid_targets_min": 686 |
| }, |
| { |
| "epoch": 3.308080808080808, |
| "grad_norm": 1.0769409575744318, |
| "learning_rate": 1.9970611866014432e-05, |
| "loss": 0.1626, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1580488085746765, |
| "step": 655, |
| "valid_targets_mean": 1152.6, |
| "valid_targets_min": 576 |
| }, |
| { |
| "epoch": 3.3333333333333335, |
| "grad_norm": 0.9561591809140133, |
| "learning_rate": 1.967674448577471e-05, |
| "loss": 0.1581, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15929345786571503, |
| "step": 660, |
| "valid_targets_mean": 1284.5, |
| "valid_targets_min": 629 |
| }, |
| { |
| "epoch": 3.3585858585858586, |
| "grad_norm": 0.9519946019411174, |
| "learning_rate": 1.9382946900239247e-05, |
| "loss": 0.1586, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1578896939754486, |
| "step": 665, |
| "valid_targets_mean": 1273.5, |
| "valid_targets_min": 841 |
| }, |
| { |
| "epoch": 3.3838383838383836, |
| "grad_norm": 0.9445690376209704, |
| "learning_rate": 1.9089282543795692e-05, |
| "loss": 0.171, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16323482990264893, |
| "step": 670, |
| "valid_targets_mean": 1237.7, |
| "valid_targets_min": 704 |
| }, |
| { |
| "epoch": 3.409090909090909, |
| "grad_norm": 0.9125004973109183, |
| "learning_rate": 1.879581482206592e-05, |
| "loss": 0.1549, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1543537974357605, |
| "step": 675, |
| "valid_targets_mean": 1212.9, |
| "valid_targets_min": 608 |
| }, |
| { |
| "epoch": 3.4343434343434343, |
| "grad_norm": 0.862065205565415, |
| "learning_rate": 1.8502607098216056e-05, |
| "loss": 0.1645, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17540498077869415, |
| "step": 680, |
| "valid_targets_mean": 1439.5, |
| "valid_targets_min": 626 |
| }, |
| { |
| "epoch": 3.45959595959596, |
| "grad_norm": 0.9640325333317653, |
| "learning_rate": 1.8209722679275602e-05, |
| "loss": 0.1636, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16301819682121277, |
| "step": 685, |
| "valid_targets_mean": 1403.1, |
| "valid_targets_min": 758 |
| }, |
| { |
| "epoch": 3.484848484848485, |
| "grad_norm": 0.9118905245785244, |
| "learning_rate": 1.791722480246868e-05, |
| "loss": 0.1615, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1672666370868683, |
| "step": 690, |
| "valid_targets_mean": 1300.1, |
| "valid_targets_min": 859 |
| }, |
| { |
| "epoch": 3.51010101010101, |
| "grad_norm": 0.879350730391716, |
| "learning_rate": 1.762517662156037e-05, |
| "loss": 0.1592, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1608966588973999, |
| "step": 695, |
| "valid_targets_mean": 1377.4, |
| "valid_targets_min": 927 |
| }, |
| { |
| "epoch": 3.5353535353535355, |
| "grad_norm": 0.9462737198336908, |
| "learning_rate": 1.733364119322109e-05, |
| "loss": 0.1648, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17058268189430237, |
| "step": 700, |
| "valid_targets_mean": 1403.2, |
| "valid_targets_min": 609 |
| }, |
| { |
| "epoch": 3.5606060606060606, |
| "grad_norm": 0.9043276008484828, |
| "learning_rate": 1.704268146341185e-05, |
| "loss": 0.1683, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17930132150650024, |
| "step": 705, |
| "valid_targets_mean": 1315.9, |
| "valid_targets_min": 678 |
| }, |
| { |
| "epoch": 3.5858585858585856, |
| "grad_norm": 0.8976270383472466, |
| "learning_rate": 1.675236025379355e-05, |
| "loss": 0.1626, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15831458568572998, |
| "step": 710, |
| "valid_targets_mean": 1351.8, |
| "valid_targets_min": 626 |
| }, |
| { |
| "epoch": 3.611111111111111, |
| "grad_norm": 0.9372138654679043, |
| "learning_rate": 1.6462740248162988e-05, |
| "loss": 0.1628, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16674670577049255, |
| "step": 715, |
| "valid_targets_mean": 1407.9, |
| "valid_targets_min": 869 |
| }, |
| { |
| "epoch": 3.6363636363636362, |
| "grad_norm": 0.8889620690389666, |
| "learning_rate": 1.6173883978918682e-05, |
| "loss": 0.1579, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16323408484458923, |
| "step": 720, |
| "valid_targets_mean": 1347.6, |
| "valid_targets_min": 714 |
| }, |
| { |
| "epoch": 3.6616161616161618, |
| "grad_norm": 0.9461049287423077, |
| "learning_rate": 1.5885853813559392e-05, |
| "loss": 0.1606, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16170771420001984, |
| "step": 725, |
| "valid_targets_mean": 1151.1, |
| "valid_targets_min": 474 |
| }, |
| { |
| "epoch": 3.686868686868687, |
| "grad_norm": 0.9883837352207571, |
| "learning_rate": 1.5598711941218265e-05, |
| "loss": 0.1639, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1605217158794403, |
| "step": 730, |
| "valid_targets_mean": 1241.0, |
| "valid_targets_min": 578 |
| }, |
| { |
| "epoch": 3.712121212121212, |
| "grad_norm": 2.3494111496602836, |
| "learning_rate": 1.531252035923541e-05, |
| "loss": 0.1686, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.17595025897026062, |
| "step": 735, |
| "valid_targets_mean": 1373.8, |
| "valid_targets_min": 591 |
| }, |
| { |
| "epoch": 3.7373737373737375, |
| "grad_norm": 0.9593181868358482, |
| "learning_rate": 1.5027340859771972e-05, |
| "loss": 0.1652, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15891680121421814, |
| "step": 740, |
| "valid_targets_mean": 1432.9, |
| "valid_targets_min": 785 |
| }, |
| { |
| "epoch": 3.7626262626262625, |
| "grad_norm": 1.2303147950630564, |
| "learning_rate": 1.4743235016468474e-05, |
| "loss": 0.1628, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15962883830070496, |
| "step": 745, |
| "valid_targets_mean": 1254.1, |
| "valid_targets_min": 640 |
| }, |
| { |
| "epoch": 3.787878787878788, |
| "grad_norm": 0.868720721287915, |
| "learning_rate": 1.4460264171150296e-05, |
| "loss": 0.1614, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16074195504188538, |
| "step": 750, |
| "valid_targets_mean": 1357.2, |
| "valid_targets_min": 915 |
| }, |
| { |
| "epoch": 3.813131313131313, |
| "grad_norm": 0.8933503401903754, |
| "learning_rate": 1.4178489420583297e-05, |
| "loss": 0.1659, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16842246055603027, |
| "step": 755, |
| "valid_targets_mean": 1366.2, |
| "valid_targets_min": 494 |
| }, |
| { |
| "epoch": 3.8383838383838382, |
| "grad_norm": 0.9117330229229211, |
| "learning_rate": 1.3897971603282278e-05, |
| "loss": 0.1595, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15995121002197266, |
| "step": 760, |
| "valid_targets_mean": 1271.9, |
| "valid_targets_min": 709 |
| }, |
| { |
| "epoch": 3.8636363636363638, |
| "grad_norm": 0.9015209816689598, |
| "learning_rate": 1.36187712863752e-05, |
| "loss": 0.166, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15097013115882874, |
| "step": 765, |
| "valid_targets_mean": 1283.9, |
| "valid_targets_min": 650 |
| }, |
| { |
| "epoch": 3.888888888888889, |
| "grad_norm": 1.0021456080906517, |
| "learning_rate": 1.3340948752526069e-05, |
| "loss": 0.1599, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16016200184822083, |
| "step": 770, |
| "valid_targets_mean": 1304.5, |
| "valid_targets_min": 508 |
| }, |
| { |
| "epoch": 3.9141414141414144, |
| "grad_norm": 0.9443204411485612, |
| "learning_rate": 1.3064563986919142e-05, |
| "loss": 0.1596, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16480416059494019, |
| "step": 775, |
| "valid_targets_mean": 1187.4, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 3.9393939393939394, |
| "grad_norm": 1.0288285211308523, |
| "learning_rate": 1.278967666430745e-05, |
| "loss": 0.1628, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16644668579101562, |
| "step": 780, |
| "valid_targets_mean": 1176.1, |
| "valid_targets_min": 711 |
| }, |
| { |
| "epoch": 3.9646464646464645, |
| "grad_norm": 0.9366989096743141, |
| "learning_rate": 1.2516346136128318e-05, |
| "loss": 0.1641, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1586305946111679, |
| "step": 785, |
| "valid_targets_mean": 1275.4, |
| "valid_targets_min": 615 |
| }, |
| { |
| "epoch": 3.98989898989899, |
| "grad_norm": 1.0032114563953773, |
| "learning_rate": 1.2244631417688632e-05, |
| "loss": 0.1634, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16418969631195068, |
| "step": 790, |
| "valid_targets_mean": 1259.9, |
| "valid_targets_min": 784 |
| }, |
| { |
| "epoch": 4.015151515151516, |
| "grad_norm": 0.8527845312952415, |
| "learning_rate": 1.197459117542278e-05, |
| "loss": 0.1509, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14444586634635925, |
| "step": 795, |
| "valid_targets_mean": 1407.8, |
| "valid_targets_min": 683 |
| }, |
| { |
| "epoch": 4.040404040404041, |
| "grad_norm": 0.9834828014413456, |
| "learning_rate": 1.170628371422587e-05, |
| "loss": 0.1365, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1405264139175415, |
| "step": 800, |
| "valid_targets_mean": 1329.9, |
| "valid_targets_min": 688 |
| }, |
| { |
| "epoch": 4.065656565656566, |
| "grad_norm": 1.0122303972675173, |
| "learning_rate": 1.1439766964864995e-05, |
| "loss": 0.1454, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15851764380931854, |
| "step": 805, |
| "valid_targets_mean": 1417.4, |
| "valid_targets_min": 800 |
| }, |
| { |
| "epoch": 4.090909090909091, |
| "grad_norm": 1.0252397884032003, |
| "learning_rate": 1.117509847147128e-05, |
| "loss": 0.1405, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13895373046398163, |
| "step": 810, |
| "valid_targets_mean": 1156.9, |
| "valid_targets_min": 574 |
| }, |
| { |
| "epoch": 4.116161616161616, |
| "grad_norm": 0.8421871414828316, |
| "learning_rate": 1.0912335379115469e-05, |
| "loss": 0.138, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1370278298854828, |
| "step": 815, |
| "valid_targets_mean": 1569.4, |
| "valid_targets_min": 668 |
| }, |
| { |
| "epoch": 4.141414141414141, |
| "grad_norm": 0.9960492717705738, |
| "learning_rate": 1.0651534421469569e-05, |
| "loss": 0.1441, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1409623920917511, |
| "step": 820, |
| "valid_targets_mean": 1150.9, |
| "valid_targets_min": 604 |
| }, |
| { |
| "epoch": 4.166666666666667, |
| "grad_norm": 0.9475930503556128, |
| "learning_rate": 1.0392751908557406e-05, |
| "loss": 0.1423, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14038413763046265, |
| "step": 825, |
| "valid_targets_mean": 1251.2, |
| "valid_targets_min": 741 |
| }, |
| { |
| "epoch": 4.191919191919192, |
| "grad_norm": 0.9077538390467569, |
| "learning_rate": 1.013604371459663e-05, |
| "loss": 0.1385, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1355401575565338, |
| "step": 830, |
| "valid_targets_mean": 1393.7, |
| "valid_targets_min": 680 |
| }, |
| { |
| "epoch": 4.217171717171717, |
| "grad_norm": 0.9313116277410778, |
| "learning_rate": 9.881465265934802e-06, |
| "loss": 0.1434, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13590630888938904, |
| "step": 835, |
| "valid_targets_mean": 1176.8, |
| "valid_targets_min": 785 |
| }, |
| { |
| "epoch": 4.242424242424242, |
| "grad_norm": 1.0954460592720818, |
| "learning_rate": 9.62907152908215e-06, |
| "loss": 0.1395, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14245356619358063, |
| "step": 840, |
| "valid_targets_mean": 1191.3, |
| "valid_targets_min": 619 |
| }, |
| { |
| "epoch": 4.267676767676767, |
| "grad_norm": 0.9434679350463996, |
| "learning_rate": 9.378916998843716e-06, |
| "loss": 0.145, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1527371108531952, |
| "step": 845, |
| "valid_targets_mean": 1231.9, |
| "valid_targets_min": 714 |
| }, |
| { |
| "epoch": 4.292929292929293, |
| "grad_norm": 0.9392165090395129, |
| "learning_rate": 9.13105568655322e-06, |
| "loss": 0.1407, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14035402238368988, |
| "step": 850, |
| "valid_targets_mean": 1247.8, |
| "valid_targets_min": 494 |
| }, |
| { |
| "epoch": 4.318181818181818, |
| "grad_norm": 0.9610037453369663, |
| "learning_rate": 8.885541108411386e-06, |
| "loss": 0.1433, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13992322981357574, |
| "step": 855, |
| "valid_targets_mean": 1307.4, |
| "valid_targets_min": 681 |
| }, |
| { |
| "epoch": 4.343434343434343, |
| "grad_norm": 0.9683597353689614, |
| "learning_rate": 8.642426273931202e-06, |
| "loss": 0.146, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1510647088289261, |
| "step": 860, |
| "valid_targets_mean": 1492.4, |
| "valid_targets_min": 714 |
| }, |
| { |
| "epoch": 4.3686868686868685, |
| "grad_norm": 0.9533225821806605, |
| "learning_rate": 8.40176367449247e-06, |
| "loss": 0.1386, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1331542581319809, |
| "step": 865, |
| "valid_targets_mean": 1215.2, |
| "valid_targets_min": 576 |
| }, |
| { |
| "epoch": 4.393939393939394, |
| "grad_norm": 1.0553941956845185, |
| "learning_rate": 8.16360527200833e-06, |
| "loss": 0.1461, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13314183056354523, |
| "step": 870, |
| "valid_targets_mean": 1251.7, |
| "valid_targets_min": 696 |
| }, |
| { |
| "epoch": 4.41919191919192, |
| "grad_norm": 0.9400873253173821, |
| "learning_rate": 7.928002487706077e-06, |
| "loss": 0.1456, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14997591078281403, |
| "step": 875, |
| "valid_targets_mean": 1381.4, |
| "valid_targets_min": 954 |
| }, |
| { |
| "epoch": 4.444444444444445, |
| "grad_norm": 0.9551571559692043, |
| "learning_rate": 7.69500619102469e-06, |
| "loss": 0.1423, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1366039514541626, |
| "step": 880, |
| "valid_targets_mean": 1127.9, |
| "valid_targets_min": 702 |
| }, |
| { |
| "epoch": 4.46969696969697, |
| "grad_norm": 3.643078356257917, |
| "learning_rate": 7.464666688631497e-06, |
| "loss": 0.1453, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1459706425666809, |
| "step": 885, |
| "valid_targets_mean": 1290.3, |
| "valid_targets_min": 671 |
| }, |
| { |
| "epoch": 4.494949494949495, |
| "grad_norm": 1.0361986594800512, |
| "learning_rate": 7.237033713560415e-06, |
| "loss": 0.1366, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13341154158115387, |
| "step": 890, |
| "valid_targets_mean": 1187.2, |
| "valid_targets_min": 825 |
| }, |
| { |
| "epoch": 4.52020202020202, |
| "grad_norm": 0.9978450478422115, |
| "learning_rate": 7.01215641447395e-06, |
| "loss": 0.1416, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1472000777721405, |
| "step": 895, |
| "valid_targets_mean": 1245.2, |
| "valid_targets_min": 797 |
| }, |
| { |
| "epoch": 4.545454545454545, |
| "grad_norm": 1.0033638283511224, |
| "learning_rate": 6.790083345051457e-06, |
| "loss": 0.144, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.16042381525039673, |
| "step": 900, |
| "valid_targets_mean": 1246.5, |
| "valid_targets_min": 655 |
| }, |
| { |
| "epoch": 4.570707070707071, |
| "grad_norm": 0.9370414548757213, |
| "learning_rate": 6.570862453505793e-06, |
| "loss": 0.142, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1394890397787094, |
| "step": 905, |
| "valid_targets_mean": 1438.2, |
| "valid_targets_min": 677 |
| }, |
| { |
| "epoch": 4.595959595959596, |
| "grad_norm": 0.8845963118772141, |
| "learning_rate": 6.35454107223074e-06, |
| "loss": 0.1398, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1384766846895218, |
| "step": 910, |
| "valid_targets_mean": 1392.9, |
| "valid_targets_min": 592 |
| }, |
| { |
| "epoch": 4.621212121212121, |
| "grad_norm": 0.9355477324402229, |
| "learning_rate": 6.141165907581395e-06, |
| "loss": 0.1419, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14917227625846863, |
| "step": 915, |
| "valid_targets_mean": 1327.9, |
| "valid_targets_min": 753 |
| }, |
| { |
| "epoch": 4.646464646464646, |
| "grad_norm": 0.9640466514670939, |
| "learning_rate": 5.9307830297896755e-06, |
| "loss": 0.138, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13884782791137695, |
| "step": 920, |
| "valid_targets_mean": 1263.7, |
| "valid_targets_min": 632 |
| }, |
| { |
| "epoch": 4.671717171717171, |
| "grad_norm": 0.9141945168421931, |
| "learning_rate": 5.723437863017256e-06, |
| "loss": 0.1378, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14294667541980743, |
| "step": 925, |
| "valid_targets_mean": 1302.5, |
| "valid_targets_min": 777 |
| }, |
| { |
| "epoch": 4.696969696969697, |
| "grad_norm": 0.9123065154375142, |
| "learning_rate": 5.519175175547919e-06, |
| "loss": 0.138, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1268976479768753, |
| "step": 930, |
| "valid_targets_mean": 1319.0, |
| "valid_targets_min": 661 |
| }, |
| { |
| "epoch": 4.722222222222222, |
| "grad_norm": 0.8964668435132555, |
| "learning_rate": 5.318039070121557e-06, |
| "loss": 0.1466, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14182919263839722, |
| "step": 935, |
| "valid_targets_mean": 1347.6, |
| "valid_targets_min": 804 |
| }, |
| { |
| "epoch": 4.747474747474747, |
| "grad_norm": 0.9249960271962788, |
| "learning_rate": 5.120072974411863e-06, |
| "loss": 0.1371, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13355281949043274, |
| "step": 940, |
| "valid_targets_mean": 1267.5, |
| "valid_targets_min": 631 |
| }, |
| { |
| "epoch": 4.7727272727272725, |
| "grad_norm": 0.9555005569091014, |
| "learning_rate": 4.92531963164981e-06, |
| "loss": 0.1408, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13473376631736755, |
| "step": 945, |
| "valid_targets_mean": 1280.8, |
| "valid_targets_min": 741 |
| }, |
| { |
| "epoch": 4.797979797979798, |
| "grad_norm": 0.9026207825389647, |
| "learning_rate": 4.733821091394841e-06, |
| "loss": 0.1377, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12557560205459595, |
| "step": 950, |
| "valid_targets_mean": 1251.2, |
| "valid_targets_min": 639 |
| }, |
| { |
| "epoch": 4.8232323232323235, |
| "grad_norm": 0.8648311888440021, |
| "learning_rate": 4.5456187004558806e-06, |
| "loss": 0.1341, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13172218203544617, |
| "step": 955, |
| "valid_targets_mean": 1288.4, |
| "valid_targets_min": 615 |
| }, |
| { |
| "epoch": 4.848484848484849, |
| "grad_norm": 0.935270097185908, |
| "learning_rate": 4.360753093964094e-06, |
| "loss": 0.1415, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15985289216041565, |
| "step": 960, |
| "valid_targets_mean": 1484.0, |
| "valid_targets_min": 806 |
| }, |
| { |
| "epoch": 4.873737373737374, |
| "grad_norm": 0.9346939564460294, |
| "learning_rate": 4.179264186599239e-06, |
| "loss": 0.1474, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14724518358707428, |
| "step": 965, |
| "valid_targets_mean": 1514.4, |
| "valid_targets_min": 457 |
| }, |
| { |
| "epoch": 4.898989898989899, |
| "grad_norm": 0.9281283556463727, |
| "learning_rate": 4.001191163971645e-06, |
| "loss": 0.1362, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1386679857969284, |
| "step": 970, |
| "valid_targets_mean": 1296.4, |
| "valid_targets_min": 817 |
| }, |
| { |
| "epoch": 4.924242424242424, |
| "grad_norm": 0.9213870761453768, |
| "learning_rate": 3.826572474161565e-06, |
| "loss": 0.1436, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13845255970954895, |
| "step": 975, |
| "valid_targets_mean": 1287.9, |
| "valid_targets_min": 702 |
| }, |
| { |
| "epoch": 4.94949494949495, |
| "grad_norm": 1.0036592428841529, |
| "learning_rate": 3.65544581941776e-06, |
| "loss": 0.1378, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12950043380260468, |
| "step": 980, |
| "valid_targets_mean": 1237.9, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 4.974747474747475, |
| "grad_norm": 0.9657789228628604, |
| "learning_rate": 3.487848148017161e-06, |
| "loss": 0.1428, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13668538630008698, |
| "step": 985, |
| "valid_targets_mean": 1134.9, |
| "valid_targets_min": 692 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 0.8611796892921691, |
| "learning_rate": 3.3238156462872937e-06, |
| "loss": 0.1371, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12768948078155518, |
| "step": 990, |
| "valid_targets_mean": 1368.5, |
| "valid_targets_min": 651 |
| }, |
| { |
| "epoch": 5.025252525252525, |
| "grad_norm": 0.9261974676714096, |
| "learning_rate": 3.1633837307932037e-06, |
| "loss": 0.1309, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13367873430252075, |
| "step": 995, |
| "valid_targets_mean": 1154.2, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 5.05050505050505, |
| "grad_norm": 1.0036400719596783, |
| "learning_rate": 3.0065870406906094e-06, |
| "loss": 0.1292, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13053901493549347, |
| "step": 1000, |
| "valid_targets_mean": 1190.0, |
| "valid_targets_min": 633 |
| }, |
| { |
| "epoch": 5.075757575757576, |
| "grad_norm": 0.9363551041241059, |
| "learning_rate": 2.8534594302469142e-06, |
| "loss": 0.1253, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12562397122383118, |
| "step": 1005, |
| "valid_targets_mean": 1375.2, |
| "valid_targets_min": 772 |
| }, |
| { |
| "epoch": 5.101010101010101, |
| "grad_norm": 1.2700394350667998, |
| "learning_rate": 2.7040339615316315e-06, |
| "loss": 0.1276, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1437637358903885, |
| "step": 1010, |
| "valid_targets_mean": 1253.7, |
| "valid_targets_min": 837 |
| }, |
| { |
| "epoch": 5.126262626262626, |
| "grad_norm": 0.9276952092042188, |
| "learning_rate": 2.5583428972779236e-06, |
| "loss": 0.1323, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13029523193836212, |
| "step": 1015, |
| "valid_targets_mean": 1374.9, |
| "valid_targets_min": 771 |
| }, |
| { |
| "epoch": 5.151515151515151, |
| "grad_norm": 0.962578989323224, |
| "learning_rate": 2.4164176939166883e-06, |
| "loss": 0.1323, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12893831729888916, |
| "step": 1020, |
| "valid_targets_mean": 1278.4, |
| "valid_targets_min": 623 |
| }, |
| { |
| "epoch": 5.1767676767676765, |
| "grad_norm": 0.8740471250815197, |
| "learning_rate": 2.278288994784723e-06, |
| "loss": 0.1353, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13151079416275024, |
| "step": 1025, |
| "valid_targets_mean": 1495.8, |
| "valid_targets_min": 695 |
| }, |
| { |
| "epoch": 5.202020202020202, |
| "grad_norm": 0.8967346790633705, |
| "learning_rate": 2.143986623508478e-06, |
| "loss": 0.1335, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15209463238716125, |
| "step": 1030, |
| "valid_targets_mean": 1422.6, |
| "valid_targets_min": 587 |
| }, |
| { |
| "epoch": 5.2272727272727275, |
| "grad_norm": 0.994338705656964, |
| "learning_rate": 2.0135395775647916e-06, |
| "loss": 0.1302, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1367867887020111, |
| "step": 1035, |
| "valid_targets_mean": 1365.3, |
| "valid_targets_min": 648 |
| }, |
| { |
| "epoch": 5.252525252525253, |
| "grad_norm": 0.9608409928013174, |
| "learning_rate": 1.8869760220199707e-06, |
| "loss": 0.1292, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12372744828462601, |
| "step": 1040, |
| "valid_targets_mean": 1244.5, |
| "valid_targets_min": 788 |
| }, |
| { |
| "epoch": 5.277777777777778, |
| "grad_norm": 0.8629245570874375, |
| "learning_rate": 1.7643232834486347e-06, |
| "loss": 0.1331, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12655162811279297, |
| "step": 1045, |
| "valid_targets_mean": 1365.6, |
| "valid_targets_min": 813 |
| }, |
| { |
| "epoch": 5.303030303030303, |
| "grad_norm": 0.8668280603831705, |
| "learning_rate": 1.6456078440335699e-06, |
| "loss": 0.1284, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11772658675909042, |
| "step": 1050, |
| "valid_targets_mean": 1353.5, |
| "valid_targets_min": 891 |
| }, |
| { |
| "epoch": 5.328282828282829, |
| "grad_norm": 0.9006577141118385, |
| "learning_rate": 1.530855335847916e-06, |
| "loss": 0.1252, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12341552972793579, |
| "step": 1055, |
| "valid_targets_mean": 1405.5, |
| "valid_targets_min": 582 |
| }, |
| { |
| "epoch": 5.353535353535354, |
| "grad_norm": 0.891591633370806, |
| "learning_rate": 1.4200905353209127e-06, |
| "loss": 0.1281, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11440576612949371, |
| "step": 1060, |
| "valid_targets_mean": 1307.4, |
| "valid_targets_min": 948 |
| }, |
| { |
| "epoch": 5.378787878787879, |
| "grad_norm": 0.9342456964635858, |
| "learning_rate": 1.3133373578883557e-06, |
| "loss": 0.1294, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.130423903465271, |
| "step": 1065, |
| "valid_targets_mean": 1313.1, |
| "valid_targets_min": 618 |
| }, |
| { |
| "epoch": 5.404040404040404, |
| "grad_norm": 1.0164525546198238, |
| "learning_rate": 1.210618852828962e-06, |
| "loss": 0.1299, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11982132494449615, |
| "step": 1070, |
| "valid_targets_mean": 1178.9, |
| "valid_targets_min": 576 |
| }, |
| { |
| "epoch": 5.429292929292929, |
| "grad_norm": 0.9239002515903104, |
| "learning_rate": 1.111957198287792e-06, |
| "loss": 0.1277, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12996883690357208, |
| "step": 1075, |
| "valid_targets_mean": 1293.5, |
| "valid_targets_min": 640 |
| }, |
| { |
| "epoch": 5.454545454545454, |
| "grad_norm": 0.9511383148715302, |
| "learning_rate": 1.0173736964876867e-06, |
| "loss": 0.1263, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12540099024772644, |
| "step": 1080, |
| "valid_targets_mean": 1283.8, |
| "valid_targets_min": 474 |
| }, |
| { |
| "epoch": 5.47979797979798, |
| "grad_norm": 1.0846233711368571, |
| "learning_rate": 9.268887691298878e-07, |
| "loss": 0.1284, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12933316826820374, |
| "step": 1085, |
| "valid_targets_mean": 1257.1, |
| "valid_targets_min": 605 |
| }, |
| { |
| "epoch": 5.505050505050505, |
| "grad_norm": 0.9384952309399883, |
| "learning_rate": 8.405219529847453e-07, |
| "loss": 0.1277, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1350557953119278, |
| "step": 1090, |
| "valid_targets_mean": 1315.2, |
| "valid_targets_min": 629 |
| }, |
| { |
| "epoch": 5.53030303030303, |
| "grad_norm": 0.9014524057924357, |
| "learning_rate": 7.582918956734909e-07, |
| "loss": 0.1345, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12424828857183456, |
| "step": 1095, |
| "valid_targets_mean": 1340.4, |
| "valid_targets_min": 781 |
| }, |
| { |
| "epoch": 5.555555555555555, |
| "grad_norm": 0.9292653851483377, |
| "learning_rate": 6.802163516419979e-07, |
| "loss": 0.1265, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12965571880340576, |
| "step": 1100, |
| "valid_targets_mean": 1274.8, |
| "valid_targets_min": 486 |
| }, |
| { |
| "epoch": 5.58080808080808, |
| "grad_norm": 0.943887372743222, |
| "learning_rate": 6.063121783273907e-07, |
| "loss": 0.1307, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1259051263332367, |
| "step": 1105, |
| "valid_targets_mean": 1327.0, |
| "valid_targets_min": 836 |
| }, |
| { |
| "epoch": 5.606060606060606, |
| "grad_norm": 0.9737257175600421, |
| "learning_rate": 5.365953325183216e-07, |
| "loss": 0.1277, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12629470229148865, |
| "step": 1110, |
| "valid_targets_mean": 1253.1, |
| "valid_targets_min": 762 |
| }, |
| { |
| "epoch": 5.6313131313131315, |
| "grad_norm": 1.0509566598389861, |
| "learning_rate": 4.7108086690970113e-07, |
| "loss": 0.1323, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13950996100902557, |
| "step": 1115, |
| "valid_targets_mean": 1205.4, |
| "valid_targets_min": 655 |
| }, |
| { |
| "epoch": 5.656565656565657, |
| "grad_norm": 0.8734622130361067, |
| "learning_rate": 4.0978292685265765e-07, |
| "loss": 0.1235, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11927177011966705, |
| "step": 1120, |
| "valid_targets_mean": 1288.9, |
| "valid_targets_min": 656 |
| }, |
| { |
| "epoch": 5.681818181818182, |
| "grad_norm": 1.041895306850947, |
| "learning_rate": 3.5271474730037557e-07, |
| "loss": 0.1302, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12915349006652832, |
| "step": 1125, |
| "valid_targets_mean": 1260.8, |
| "valid_targets_min": 804 |
| }, |
| { |
| "epoch": 5.707070707070707, |
| "grad_norm": 0.9809767376936727, |
| "learning_rate": 2.998886499505171e-07, |
| "loss": 0.127, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1269259750843048, |
| "step": 1130, |
| "valid_targets_mean": 1323.3, |
| "valid_targets_min": 741 |
| }, |
| { |
| "epoch": 5.732323232323233, |
| "grad_norm": 1.2138384974786145, |
| "learning_rate": 2.513160405848303e-07, |
| "loss": 0.1345, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.14036789536476135, |
| "step": 1135, |
| "valid_targets_mean": 1340.8, |
| "valid_targets_min": 615 |
| }, |
| { |
| "epoch": 5.757575757575758, |
| "grad_norm": 0.8909820802700256, |
| "learning_rate": 2.0700740660648312e-07, |
| "loss": 0.1328, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11737575381994247, |
| "step": 1140, |
| "valid_targets_mean": 1354.6, |
| "valid_targets_min": 864 |
| }, |
| { |
| "epoch": 5.782828282828283, |
| "grad_norm": 1.0571471194993831, |
| "learning_rate": 1.669723147757263e-07, |
| "loss": 0.1284, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12995797395706177, |
| "step": 1145, |
| "valid_targets_mean": 1081.0, |
| "valid_targets_min": 582 |
| }, |
| { |
| "epoch": 5.808080808080808, |
| "grad_norm": 1.4245132821438702, |
| "learning_rate": 1.312194091443142e-07, |
| "loss": 0.1331, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13467878103256226, |
| "step": 1150, |
| "valid_targets_mean": 1349.2, |
| "valid_targets_min": 656 |
| }, |
| { |
| "epoch": 5.833333333333333, |
| "grad_norm": 0.9507679786850238, |
| "learning_rate": 9.975640918915119e-08, |
| "loss": 0.1306, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12804457545280457, |
| "step": 1155, |
| "valid_targets_mean": 1274.2, |
| "valid_targets_min": 574 |
| }, |
| { |
| "epoch": 5.858585858585858, |
| "grad_norm": 0.9639470109858415, |
| "learning_rate": 7.259010814555378e-08, |
| "loss": 0.1304, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12429563701152802, |
| "step": 1160, |
| "valid_targets_mean": 1331.4, |
| "valid_targets_min": 740 |
| }, |
| { |
| "epoch": 5.883838383838384, |
| "grad_norm": 0.9048090193387625, |
| "learning_rate": 4.972637154052606e-08, |
| "loss": 0.1292, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1266717165708542, |
| "step": 1165, |
| "valid_targets_mean": 1540.2, |
| "valid_targets_min": 1033 |
| }, |
| { |
| "epoch": 5.909090909090909, |
| "grad_norm": 0.9137387217315224, |
| "learning_rate": 3.117013592631501e-08, |
| "loss": 0.1341, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13005393743515015, |
| "step": 1170, |
| "valid_targets_mean": 1407.2, |
| "valid_targets_min": 683 |
| }, |
| { |
| "epoch": 5.934343434343434, |
| "grad_norm": 0.9221913785952132, |
| "learning_rate": 1.6925407814545325e-08, |
| "loss": 0.1267, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1224047988653183, |
| "step": 1175, |
| "valid_targets_mean": 1231.8, |
| "valid_targets_min": 782 |
| }, |
| { |
| "epoch": 5.959595959595959, |
| "grad_norm": 0.89954321451429, |
| "learning_rate": 6.995262811178016e-09, |
| "loss": 0.1297, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12393134087324142, |
| "step": 1180, |
| "valid_targets_mean": 1373.0, |
| "valid_targets_min": 746 |
| }, |
| { |
| "epoch": 5.984848484848484, |
| "grad_norm": 0.9614911698663842, |
| "learning_rate": 1.3818449524416467e-09, |
| "loss": 0.1304, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12797527015209198, |
| "step": 1185, |
| "valid_targets_mean": 1194.4, |
| "valid_targets_min": 743 |
| }, |
| { |
| "epoch": 6.0, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.133074551820755, |
| "step": 1188, |
| "total_flos": 123574112944128.0, |
| "train_loss": 0.2030692549064906, |
| "train_runtime": 3957.8692, |
| "train_samples_per_second": 4.79, |
| "train_steps_per_second": 0.3, |
| "valid_targets_mean": 1319.8, |
| "valid_targets_min": 875 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 1188, |
| "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": 123574112944128.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
