| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 780, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03210272873194221, | |
| "grad_norm": 7.443076512121371, | |
| "learning_rate": 2.0512820512820513e-06, | |
| "loss": 0.8032, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21804624795913696, | |
| "step": 5, | |
| "valid_targets_mean": 4336.2, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 0.06420545746388442, | |
| "grad_norm": 5.001718673683824, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 0.7807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21758630871772766, | |
| "step": 10, | |
| "valid_targets_mean": 3736.1, | |
| "valid_targets_min": 1113 | |
| }, | |
| { | |
| "epoch": 0.09630818619582665, | |
| "grad_norm": 2.9021650531373258, | |
| "learning_rate": 7.17948717948718e-06, | |
| "loss": 0.7119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18503277003765106, | |
| "step": 15, | |
| "valid_targets_mean": 3926.9, | |
| "valid_targets_min": 837 | |
| }, | |
| { | |
| "epoch": 0.12841091492776885, | |
| "grad_norm": 1.0697186254007953, | |
| "learning_rate": 9.743589743589744e-06, | |
| "loss": 0.6496, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15346118807792664, | |
| "step": 20, | |
| "valid_targets_mean": 3225.6, | |
| "valid_targets_min": 613 | |
| }, | |
| { | |
| "epoch": 0.16051364365971107, | |
| "grad_norm": 0.9133287345600813, | |
| "learning_rate": 1.230769230769231e-05, | |
| "loss": 0.6397, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16350162029266357, | |
| "step": 25, | |
| "valid_targets_mean": 3446.9, | |
| "valid_targets_min": 1106 | |
| }, | |
| { | |
| "epoch": 0.1926163723916533, | |
| "grad_norm": 0.5525491725186413, | |
| "learning_rate": 1.4871794871794874e-05, | |
| "loss": 0.6154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16575123369693756, | |
| "step": 30, | |
| "valid_targets_mean": 4915.8, | |
| "valid_targets_min": 1045 | |
| }, | |
| { | |
| "epoch": 0.2247191011235955, | |
| "grad_norm": 0.5368849244599112, | |
| "learning_rate": 1.7435897435897438e-05, | |
| "loss": 0.5911, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14013072848320007, | |
| "step": 35, | |
| "valid_targets_mean": 3104.2, | |
| "valid_targets_min": 859 | |
| }, | |
| { | |
| "epoch": 0.2568218298555377, | |
| "grad_norm": 0.4615281982796634, | |
| "learning_rate": 2e-05, | |
| "loss": 0.5892, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.161387100815773, | |
| "step": 40, | |
| "valid_targets_mean": 3632.9, | |
| "valid_targets_min": 659 | |
| }, | |
| { | |
| "epoch": 0.28892455858747995, | |
| "grad_norm": 0.4061270737531561, | |
| "learning_rate": 2.2564102564102566e-05, | |
| "loss": 0.5387, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12181757390499115, | |
| "step": 45, | |
| "valid_targets_mean": 3131.7, | |
| "valid_targets_min": 505 | |
| }, | |
| { | |
| "epoch": 0.32102728731942215, | |
| "grad_norm": 0.33878739727162444, | |
| "learning_rate": 2.512820512820513e-05, | |
| "loss": 0.5399, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12523899972438812, | |
| "step": 50, | |
| "valid_targets_mean": 3602.1, | |
| "valid_targets_min": 609 | |
| }, | |
| { | |
| "epoch": 0.35313001605136435, | |
| "grad_norm": 0.35224338467065713, | |
| "learning_rate": 2.7692307692307694e-05, | |
| "loss": 0.5154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15735869109630585, | |
| "step": 55, | |
| "valid_targets_mean": 4276.2, | |
| "valid_targets_min": 1111 | |
| }, | |
| { | |
| "epoch": 0.3852327447833066, | |
| "grad_norm": 0.29204719939595825, | |
| "learning_rate": 3.0256410256410257e-05, | |
| "loss": 0.505, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.142252117395401, | |
| "step": 60, | |
| "valid_targets_mean": 4205.7, | |
| "valid_targets_min": 1221 | |
| }, | |
| { | |
| "epoch": 0.4173354735152488, | |
| "grad_norm": 0.3108436023652963, | |
| "learning_rate": 3.282051282051282e-05, | |
| "loss": 0.5088, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10603222995996475, | |
| "step": 65, | |
| "valid_targets_mean": 3141.0, | |
| "valid_targets_min": 1008 | |
| }, | |
| { | |
| "epoch": 0.449438202247191, | |
| "grad_norm": 0.2986968060504965, | |
| "learning_rate": 3.538461538461539e-05, | |
| "loss": 0.495, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1236981451511383, | |
| "step": 70, | |
| "valid_targets_mean": 3417.8, | |
| "valid_targets_min": 816 | |
| }, | |
| { | |
| "epoch": 0.48154093097913325, | |
| "grad_norm": 0.6323281716016163, | |
| "learning_rate": 3.794871794871795e-05, | |
| "loss": 0.5013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11899252235889435, | |
| "step": 75, | |
| "valid_targets_mean": 3896.2, | |
| "valid_targets_min": 749 | |
| }, | |
| { | |
| "epoch": 0.5136436597110754, | |
| "grad_norm": 0.3574090578830505, | |
| "learning_rate": 3.999979972589652e-05, | |
| "loss": 0.4691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12434444576501846, | |
| "step": 80, | |
| "valid_targets_mean": 3266.9, | |
| "valid_targets_min": 726 | |
| }, | |
| { | |
| "epoch": 0.5457463884430177, | |
| "grad_norm": 0.3233297771547147, | |
| "learning_rate": 3.999279055341771e-05, | |
| "loss": 0.4792, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13975250720977783, | |
| "step": 85, | |
| "valid_targets_mean": 4060.8, | |
| "valid_targets_min": 1022 | |
| }, | |
| { | |
| "epoch": 0.5778491171749599, | |
| "grad_norm": 0.3481282366313697, | |
| "learning_rate": 3.9975771686375434e-05, | |
| "loss": 0.4805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07897774875164032, | |
| "step": 90, | |
| "valid_targets_mean": 2103.4, | |
| "valid_targets_min": 631 | |
| }, | |
| { | |
| "epoch": 0.6099518459069021, | |
| "grad_norm": 0.342138756435365, | |
| "learning_rate": 3.9948751645524235e-05, | |
| "loss": 0.4749, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11953769624233246, | |
| "step": 95, | |
| "valid_targets_mean": 3517.6, | |
| "valid_targets_min": 1496 | |
| }, | |
| { | |
| "epoch": 0.6420545746388443, | |
| "grad_norm": 0.33501528178098466, | |
| "learning_rate": 3.9911743958858376e-05, | |
| "loss": 0.4818, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11692598462104797, | |
| "step": 100, | |
| "valid_targets_mean": 3028.9, | |
| "valid_targets_min": 1292 | |
| }, | |
| { | |
| "epoch": 0.6741573033707865, | |
| "grad_norm": 0.3030959299753308, | |
| "learning_rate": 3.9864767154838864e-05, | |
| "loss": 0.4759, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08811567723751068, | |
| "step": 105, | |
| "valid_targets_mean": 2688.0, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 0.7062600321027287, | |
| "grad_norm": 0.2891268278078363, | |
| "learning_rate": 3.9807844753116864e-05, | |
| "loss": 0.4494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10534675419330597, | |
| "step": 110, | |
| "valid_targets_mean": 3910.9, | |
| "valid_targets_min": 1069 | |
| }, | |
| { | |
| "epoch": 0.7383627608346709, | |
| "grad_norm": 0.27088323808017967, | |
| "learning_rate": 3.9741005252758255e-05, | |
| "loss": 0.4684, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09811793267726898, | |
| "step": 115, | |
| "valid_targets_mean": 3091.4, | |
| "valid_targets_min": 1057 | |
| }, | |
| { | |
| "epoch": 0.7704654895666132, | |
| "grad_norm": 0.3246822966924674, | |
| "learning_rate": 3.966428211797514e-05, | |
| "loss": 0.4527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11549907922744751, | |
| "step": 120, | |
| "valid_targets_mean": 3803.6, | |
| "valid_targets_min": 992 | |
| }, | |
| { | |
| "epoch": 0.8025682182985554, | |
| "grad_norm": 0.33072065469680983, | |
| "learning_rate": 3.957771376137144e-05, | |
| "loss": 0.4683, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1350668966770172, | |
| "step": 125, | |
| "valid_targets_mean": 3599.0, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 0.8346709470304976, | |
| "grad_norm": 0.35035972572140217, | |
| "learning_rate": 3.9481343524711095e-05, | |
| "loss": 0.4594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12366610765457153, | |
| "step": 130, | |
| "valid_targets_mean": 3637.1, | |
| "valid_targets_min": 780 | |
| }, | |
| { | |
| "epoch": 0.8667736757624398, | |
| "grad_norm": 0.33737136455082983, | |
| "learning_rate": 3.937521965721831e-05, | |
| "loss": 0.4599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11112713068723679, | |
| "step": 135, | |
| "valid_targets_mean": 4047.2, | |
| "valid_targets_min": 1496 | |
| }, | |
| { | |
| "epoch": 0.898876404494382, | |
| "grad_norm": 0.31265257576082217, | |
| "learning_rate": 3.9259395291420864e-05, | |
| "loss": 0.4346, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1048852801322937, | |
| "step": 140, | |
| "valid_targets_mean": 3647.4, | |
| "valid_targets_min": 1385 | |
| }, | |
| { | |
| "epoch": 0.9309791332263242, | |
| "grad_norm": 0.3107204486324316, | |
| "learning_rate": 3.913392841654851e-05, | |
| "loss": 0.452, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13349255919456482, | |
| "step": 145, | |
| "valid_targets_mean": 4499.4, | |
| "valid_targets_min": 791 | |
| }, | |
| { | |
| "epoch": 0.9630818619582665, | |
| "grad_norm": 0.31387648518622074, | |
| "learning_rate": 3.8998881849499756e-05, | |
| "loss": 0.4438, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09112104773521423, | |
| "step": 150, | |
| "valid_targets_mean": 3127.7, | |
| "valid_targets_min": 827 | |
| }, | |
| { | |
| "epoch": 0.9951845906902087, | |
| "grad_norm": 0.31502159646866623, | |
| "learning_rate": 3.885432320339167e-05, | |
| "loss": 0.4491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10973098129034042, | |
| "step": 155, | |
| "valid_targets_mean": 3247.8, | |
| "valid_targets_min": 860 | |
| }, | |
| { | |
| "epoch": 1.0256821829855538, | |
| "grad_norm": 0.2995227413933546, | |
| "learning_rate": 3.8700324853708304e-05, | |
| "loss": 0.4309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12735167145729065, | |
| "step": 160, | |
| "valid_targets_mean": 4369.8, | |
| "valid_targets_min": 1151 | |
| }, | |
| { | |
| "epoch": 1.057784911717496, | |
| "grad_norm": 0.3265629724457608, | |
| "learning_rate": 3.853696390206484e-05, | |
| "loss": 0.4441, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10513167083263397, | |
| "step": 165, | |
| "valid_targets_mean": 3461.3, | |
| "valid_targets_min": 1352 | |
| }, | |
| { | |
| "epoch": 1.0898876404494382, | |
| "grad_norm": 0.32739344706824414, | |
| "learning_rate": 3.8364322137605484e-05, | |
| "loss": 0.4472, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09724493324756622, | |
| "step": 170, | |
| "valid_targets_mean": 3163.2, | |
| "valid_targets_min": 806 | |
| }, | |
| { | |
| "epoch": 1.1219903691813804, | |
| "grad_norm": 0.3171489222363104, | |
| "learning_rate": 3.818248599605448e-05, | |
| "loss": 0.4197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1121516227722168, | |
| "step": 175, | |
| "valid_targets_mean": 4971.7, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 1.1540930979133226, | |
| "grad_norm": 0.3273324285042093, | |
| "learning_rate": 3.799154651644075e-05, | |
| "loss": 0.4474, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11072337627410889, | |
| "step": 180, | |
| "valid_targets_mean": 3005.9, | |
| "valid_targets_min": 1236 | |
| }, | |
| { | |
| "epoch": 1.1861958266452648, | |
| "grad_norm": 0.28545216130602724, | |
| "learning_rate": 3.7791599295517825e-05, | |
| "loss": 0.4382, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09122270345687866, | |
| "step": 185, | |
| "valid_targets_mean": 4056.4, | |
| "valid_targets_min": 941 | |
| }, | |
| { | |
| "epoch": 1.218298555377207, | |
| "grad_norm": 0.35115623987576444, | |
| "learning_rate": 3.75827444399019e-05, | |
| "loss": 0.4368, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12966161966323853, | |
| "step": 190, | |
| "valid_targets_mean": 4549.7, | |
| "valid_targets_min": 1229 | |
| }, | |
| { | |
| "epoch": 1.2504012841091492, | |
| "grad_norm": 0.3024907929000345, | |
| "learning_rate": 3.736508651595188e-05, | |
| "loss": 0.4281, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09971387684345245, | |
| "step": 195, | |
| "valid_targets_mean": 3831.5, | |
| "valid_targets_min": 1330 | |
| }, | |
| { | |
| "epoch": 1.2825040128410916, | |
| "grad_norm": 0.3649481469747776, | |
| "learning_rate": 3.713873449741669e-05, | |
| "loss": 0.4343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12230262160301208, | |
| "step": 200, | |
| "valid_targets_mean": 3829.6, | |
| "valid_targets_min": 517 | |
| }, | |
| { | |
| "epoch": 1.3146067415730336, | |
| "grad_norm": 0.3081241454749465, | |
| "learning_rate": 3.69038017108759e-05, | |
| "loss": 0.4231, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09591006487607956, | |
| "step": 205, | |
| "valid_targets_mean": 3110.9, | |
| "valid_targets_min": 1204 | |
| }, | |
| { | |
| "epoch": 1.346709470304976, | |
| "grad_norm": 0.29545465896778206, | |
| "learning_rate": 3.6660405779001036e-05, | |
| "loss": 0.4419, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08623912930488586, | |
| "step": 210, | |
| "valid_targets_mean": 3680.1, | |
| "valid_targets_min": 796 | |
| }, | |
| { | |
| "epoch": 1.3788121990369182, | |
| "grad_norm": 0.37785862300967726, | |
| "learning_rate": 3.640866856166601e-05, | |
| "loss": 0.4384, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10427422821521759, | |
| "step": 215, | |
| "valid_targets_mean": 3281.7, | |
| "valid_targets_min": 837 | |
| }, | |
| { | |
| "epoch": 1.4109149277688604, | |
| "grad_norm": 0.30674701618911177, | |
| "learning_rate": 3.614871609493614e-05, | |
| "loss": 0.4471, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10923303663730621, | |
| "step": 220, | |
| "valid_targets_mean": 3556.0, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 1.4430176565008026, | |
| "grad_norm": 0.367939390471383, | |
| "learning_rate": 3.5880678527966224e-05, | |
| "loss": 0.4404, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12570562958717346, | |
| "step": 225, | |
| "valid_targets_mean": 3408.8, | |
| "valid_targets_min": 960 | |
| }, | |
| { | |
| "epoch": 1.4751203852327448, | |
| "grad_norm": 0.294671299662254, | |
| "learning_rate": 3.5604690057839456e-05, | |
| "loss": 0.4293, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10687486082315445, | |
| "step": 230, | |
| "valid_targets_mean": 4010.9, | |
| "valid_targets_min": 853 | |
| }, | |
| { | |
| "epoch": 1.507223113964687, | |
| "grad_norm": 0.2958977994024142, | |
| "learning_rate": 3.532088886237956e-05, | |
| "loss": 0.4374, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1045038253068924, | |
| "step": 235, | |
| "valid_targets_mean": 3799.8, | |
| "valid_targets_min": 819 | |
| }, | |
| { | |
| "epoch": 1.5393258426966292, | |
| "grad_norm": 0.29728785349279935, | |
| "learning_rate": 3.5029417030970005e-05, | |
| "loss": 0.4361, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10812290012836456, | |
| "step": 240, | |
| "valid_targets_mean": 3272.2, | |
| "valid_targets_min": 945 | |
| }, | |
| { | |
| "epoch": 1.5714285714285714, | |
| "grad_norm": 0.31688903277050234, | |
| "learning_rate": 3.473042049341474e-05, | |
| "loss": 0.4323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10774408280849457, | |
| "step": 245, | |
| "valid_targets_mean": 3913.6, | |
| "valid_targets_min": 1259 | |
| }, | |
| { | |
| "epoch": 1.6035313001605136, | |
| "grad_norm": 0.3692867105165629, | |
| "learning_rate": 3.442404894687629e-05, | |
| "loss": 0.438, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1150355115532875, | |
| "step": 250, | |
| "valid_targets_mean": 3968.7, | |
| "valid_targets_min": 1449 | |
| }, | |
| { | |
| "epoch": 1.635634028892456, | |
| "grad_norm": 0.29717175235926496, | |
| "learning_rate": 3.411045578092754e-05, | |
| "loss": 0.4232, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11245591938495636, | |
| "step": 255, | |
| "valid_targets_mean": 3593.1, | |
| "valid_targets_min": 836 | |
| }, | |
| { | |
| "epoch": 1.667736757624398, | |
| "grad_norm": 0.3221818053403484, | |
| "learning_rate": 3.3789798000754924e-05, | |
| "loss": 0.4362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11561841517686844, | |
| "step": 260, | |
| "valid_targets_mean": 3729.6, | |
| "valid_targets_min": 1041 | |
| }, | |
| { | |
| "epoch": 1.6998394863563404, | |
| "grad_norm": 0.32510941540680377, | |
| "learning_rate": 3.34622361485514e-05, | |
| "loss": 0.434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1104997992515564, | |
| "step": 265, | |
| "valid_targets_mean": 3289.7, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 1.7319422150882824, | |
| "grad_norm": 0.2850424295372404, | |
| "learning_rate": 3.31279342231386e-05, | |
| "loss": 0.4275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09886403381824493, | |
| "step": 270, | |
| "valid_targets_mean": 3479.6, | |
| "valid_targets_min": 1010 | |
| }, | |
| { | |
| "epoch": 1.7640449438202248, | |
| "grad_norm": 0.29747010204623664, | |
| "learning_rate": 3.278705959785821e-05, | |
| "loss": 0.4151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09024910628795624, | |
| "step": 275, | |
| "valid_targets_mean": 2982.7, | |
| "valid_targets_min": 1051 | |
| }, | |
| { | |
| "epoch": 1.7961476725521668, | |
| "grad_norm": 0.2594879128158275, | |
| "learning_rate": 3.243978293677408e-05, | |
| "loss": 0.4082, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10707713663578033, | |
| "step": 280, | |
| "valid_targets_mean": 4601.4, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 1.8282504012841092, | |
| "grad_norm": 0.3092407046323038, | |
| "learning_rate": 3.208627810922665e-05, | |
| "loss": 0.436, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1255902498960495, | |
| "step": 285, | |
| "valid_targets_mean": 3863.8, | |
| "valid_targets_min": 1065 | |
| }, | |
| { | |
| "epoch": 1.8603531300160514, | |
| "grad_norm": 0.33530232091926576, | |
| "learning_rate": 3.172672210278254e-05, | |
| "loss": 0.4209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08784206211566925, | |
| "step": 290, | |
| "valid_targets_mean": 3285.4, | |
| "valid_targets_min": 691 | |
| }, | |
| { | |
| "epoch": 1.8924558587479936, | |
| "grad_norm": 0.2968668400465552, | |
| "learning_rate": 3.136129493462312e-05, | |
| "loss": 0.4319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12176284193992615, | |
| "step": 295, | |
| "valid_targets_mean": 4110.1, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 1.9245585874799358, | |
| "grad_norm": 0.3002839956959657, | |
| "learning_rate": 3.0990179561416124e-05, | |
| "loss": 0.4201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10375160723924637, | |
| "step": 300, | |
| "valid_targets_mean": 3537.5, | |
| "valid_targets_min": 776 | |
| }, | |
| { | |
| "epoch": 1.956661316211878, | |
| "grad_norm": 0.29150855964222466, | |
| "learning_rate": 3.061356178771564e-05, | |
| "loss": 0.4251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08603081107139587, | |
| "step": 305, | |
| "valid_targets_mean": 2934.9, | |
| "valid_targets_min": 517 | |
| }, | |
| { | |
| "epoch": 1.9887640449438202, | |
| "grad_norm": 0.31045167265910495, | |
| "learning_rate": 3.0231630172936205e-05, | |
| "loss": 0.4147, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09072297811508179, | |
| "step": 310, | |
| "valid_targets_mean": 3133.2, | |
| "valid_targets_min": 983 | |
| }, | |
| { | |
| "epoch": 2.019261637239165, | |
| "grad_norm": 0.28708543588073493, | |
| "learning_rate": 2.98445759369477e-05, | |
| "loss": 0.4074, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09558726847171783, | |
| "step": 315, | |
| "valid_targets_mean": 4008.1, | |
| "valid_targets_min": 515 | |
| }, | |
| { | |
| "epoch": 2.0513643659711076, | |
| "grad_norm": 0.3416876858005374, | |
| "learning_rate": 2.945259286433819e-05, | |
| "loss": 0.4149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11786194890737534, | |
| "step": 320, | |
| "valid_targets_mean": 3077.6, | |
| "valid_targets_min": 520 | |
| }, | |
| { | |
| "epoch": 2.0834670947030496, | |
| "grad_norm": 0.3303316995435761, | |
| "learning_rate": 2.9055877207392752e-05, | |
| "loss": 0.4184, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09221881628036499, | |
| "step": 325, | |
| "valid_targets_mean": 2562.9, | |
| "valid_targets_min": 1311 | |
| }, | |
| { | |
| "epoch": 2.115569823434992, | |
| "grad_norm": 0.3074809241219434, | |
| "learning_rate": 2.8654627587836724e-05, | |
| "loss": 0.4253, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12592661380767822, | |
| "step": 330, | |
| "valid_targets_mean": 4764.7, | |
| "valid_targets_min": 1340 | |
| }, | |
| { | |
| "epoch": 2.1476725521669344, | |
| "grad_norm": 0.3043947470357316, | |
| "learning_rate": 2.8249044897392814e-05, | |
| "loss": 0.4071, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11719249188899994, | |
| "step": 335, | |
| "valid_targets_mean": 4137.8, | |
| "valid_targets_min": 1536 | |
| }, | |
| { | |
| "epoch": 2.1797752808988764, | |
| "grad_norm": 0.2890043796848122, | |
| "learning_rate": 2.7839332197201505e-05, | |
| "loss": 0.4119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10229069739580154, | |
| "step": 340, | |
| "valid_targets_mean": 4544.4, | |
| "valid_targets_min": 1321 | |
| }, | |
| { | |
| "epoch": 2.211878009630819, | |
| "grad_norm": 0.28710289619081225, | |
| "learning_rate": 2.7425694616155474e-05, | |
| "loss": 0.4147, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09542287886142731, | |
| "step": 345, | |
| "valid_targets_mean": 3431.3, | |
| "valid_targets_min": 1012 | |
| }, | |
| { | |
| "epoch": 2.243980738362761, | |
| "grad_norm": 0.33593483468173724, | |
| "learning_rate": 2.700833924819868e-05, | |
| "loss": 0.4137, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10045792162418365, | |
| "step": 350, | |
| "valid_targets_mean": 3555.9, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 2.276083467094703, | |
| "grad_norm": 0.3050460009929215, | |
| "learning_rate": 2.6587475048641596e-05, | |
| "loss": 0.421, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1351967751979828, | |
| "step": 355, | |
| "valid_targets_mean": 4165.8, | |
| "valid_targets_min": 864 | |
| }, | |
| { | |
| "epoch": 2.308186195826645, | |
| "grad_norm": 0.3007356034553604, | |
| "learning_rate": 2.6163312729544616e-05, | |
| "loss": 0.4124, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10512376576662064, | |
| "step": 360, | |
| "valid_targets_mean": 3216.5, | |
| "valid_targets_min": 524 | |
| }, | |
| { | |
| "epoch": 2.3402889245585876, | |
| "grad_norm": 0.27621857982103765, | |
| "learning_rate": 2.5736064654221808e-05, | |
| "loss": 0.3984, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12070536613464355, | |
| "step": 365, | |
| "valid_targets_mean": 4767.9, | |
| "valid_targets_min": 978 | |
| }, | |
| { | |
| "epoch": 2.3723916532905296, | |
| "grad_norm": 0.28304064443701404, | |
| "learning_rate": 2.5305944730918017e-05, | |
| "loss": 0.4133, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09823143482208252, | |
| "step": 370, | |
| "valid_targets_mean": 4099.4, | |
| "valid_targets_min": 1167 | |
| }, | |
| { | |
| "epoch": 2.404494382022472, | |
| "grad_norm": 0.3317797648395537, | |
| "learning_rate": 2.487316830571244e-05, | |
| "loss": 0.4177, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08248773217201233, | |
| "step": 375, | |
| "valid_targets_mean": 3434.5, | |
| "valid_targets_min": 945 | |
| }, | |
| { | |
| "epoch": 2.436597110754414, | |
| "grad_norm": 0.34105769887242393, | |
| "learning_rate": 2.4437952054702354e-05, | |
| "loss": 0.4131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08816108107566833, | |
| "step": 380, | |
| "valid_targets_mean": 3624.6, | |
| "valid_targets_min": 551 | |
| }, | |
| { | |
| "epoch": 2.4686998394863564, | |
| "grad_norm": 0.29849113343556627, | |
| "learning_rate": 2.4000513875520892e-05, | |
| "loss": 0.4183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10371711850166321, | |
| "step": 385, | |
| "valid_targets_mean": 4422.7, | |
| "valid_targets_min": 1146 | |
| }, | |
| { | |
| "epoch": 2.5008025682182984, | |
| "grad_norm": 0.29651959052335053, | |
| "learning_rate": 2.3561072778243335e-05, | |
| "loss": 0.4037, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10162942856550217, | |
| "step": 390, | |
| "valid_targets_mean": 3326.1, | |
| "valid_targets_min": 1415 | |
| }, | |
| { | |
| "epoch": 2.532905296950241, | |
| "grad_norm": 0.30954429361691455, | |
| "learning_rate": 2.311984877573636e-05, | |
| "loss": 0.4075, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09746764600276947, | |
| "step": 395, | |
| "valid_targets_mean": 3306.4, | |
| "valid_targets_min": 1062 | |
| }, | |
| { | |
| "epoch": 2.5650080256821832, | |
| "grad_norm": 0.2904497497376163, | |
| "learning_rate": 2.2677062773505235e-05, | |
| "loss": 0.4158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1188712939620018, | |
| "step": 400, | |
| "valid_targets_mean": 4573.7, | |
| "valid_targets_min": 972 | |
| }, | |
| { | |
| "epoch": 2.597110754414125, | |
| "grad_norm": 0.3164646902586644, | |
| "learning_rate": 2.2232936459094158e-05, | |
| "loss": 0.4183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1208307221531868, | |
| "step": 405, | |
| "valid_targets_mean": 3505.8, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 2.629213483146067, | |
| "grad_norm": 0.36094852940534805, | |
| "learning_rate": 2.1787692191094986e-05, | |
| "loss": 0.3936, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09573773294687271, | |
| "step": 410, | |
| "valid_targets_mean": 3226.8, | |
| "valid_targets_min": 808 | |
| }, | |
| { | |
| "epoch": 2.6613162118780096, | |
| "grad_norm": 0.3494215415068745, | |
| "learning_rate": 2.1341552887820048e-05, | |
| "loss": 0.415, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10980828106403351, | |
| "step": 415, | |
| "valid_targets_mean": 3792.3, | |
| "valid_targets_min": 1168 | |
| }, | |
| { | |
| "epoch": 2.693418940609952, | |
| "grad_norm": 0.3890932782850568, | |
| "learning_rate": 2.0894741915694685e-05, | |
| "loss": 0.4111, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09624197334051132, | |
| "step": 420, | |
| "valid_targets_mean": 3245.2, | |
| "valid_targets_min": 1020 | |
| }, | |
| { | |
| "epoch": 2.725521669341894, | |
| "grad_norm": 0.2721958461218534, | |
| "learning_rate": 2.0447482977425465e-05, | |
| "loss": 0.4047, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0908433347940445, | |
| "step": 425, | |
| "valid_targets_mean": 3821.1, | |
| "valid_targets_min": 1278 | |
| }, | |
| { | |
| "epoch": 2.7576243980738364, | |
| "grad_norm": 0.2925048879553211, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4148, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12087635695934296, | |
| "step": 430, | |
| "valid_targets_mean": 3961.0, | |
| "valid_targets_min": 848 | |
| }, | |
| { | |
| "epoch": 2.7897271268057784, | |
| "grad_norm": 0.2931211576473339, | |
| "learning_rate": 1.9552517022574542e-05, | |
| "loss": 0.4073, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10215561091899872, | |
| "step": 435, | |
| "valid_targets_mean": 4705.2, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 2.821829855537721, | |
| "grad_norm": 0.31375433571050343, | |
| "learning_rate": 1.910525808430532e-05, | |
| "loss": 0.4198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0890008732676506, | |
| "step": 440, | |
| "valid_targets_mean": 3066.1, | |
| "valid_targets_min": 1310 | |
| }, | |
| { | |
| "epoch": 2.853932584269663, | |
| "grad_norm": 0.32732471726116175, | |
| "learning_rate": 1.8658447112179952e-05, | |
| "loss": 0.4064, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1120062842965126, | |
| "step": 445, | |
| "valid_targets_mean": 3805.1, | |
| "valid_targets_min": 816 | |
| }, | |
| { | |
| "epoch": 2.886035313001605, | |
| "grad_norm": 0.32531970253612086, | |
| "learning_rate": 1.821230780890502e-05, | |
| "loss": 0.4039, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08783616870641708, | |
| "step": 450, | |
| "valid_targets_mean": 3062.5, | |
| "valid_targets_min": 930 | |
| }, | |
| { | |
| "epoch": 2.918138041733547, | |
| "grad_norm": 0.30368172861725967, | |
| "learning_rate": 1.776706354090585e-05, | |
| "loss": 0.3946, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10530835390090942, | |
| "step": 455, | |
| "valid_targets_mean": 3589.1, | |
| "valid_targets_min": 1070 | |
| }, | |
| { | |
| "epoch": 2.9502407704654896, | |
| "grad_norm": 0.2990061639291643, | |
| "learning_rate": 1.7322937226494775e-05, | |
| "loss": 0.4167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10799364745616913, | |
| "step": 460, | |
| "valid_targets_mean": 3779.5, | |
| "valid_targets_min": 1047 | |
| }, | |
| { | |
| "epoch": 2.982343499197432, | |
| "grad_norm": 0.3067377142223239, | |
| "learning_rate": 1.6880151224263646e-05, | |
| "loss": 0.4143, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09094572067260742, | |
| "step": 465, | |
| "valid_targets_mean": 3441.7, | |
| "valid_targets_min": 730 | |
| }, | |
| { | |
| "epoch": 3.012841091492777, | |
| "grad_norm": 0.30726836931947277, | |
| "learning_rate": 1.643892722175667e-05, | |
| "loss": 0.3996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09883762896060944, | |
| "step": 470, | |
| "valid_targets_mean": 3639.9, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 3.044943820224719, | |
| "grad_norm": 0.2852457832390904, | |
| "learning_rate": 1.5999486124479115e-05, | |
| "loss": 0.3963, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12268540263175964, | |
| "step": 475, | |
| "valid_targets_mean": 5229.9, | |
| "valid_targets_min": 1711 | |
| }, | |
| { | |
| "epoch": 3.077046548956661, | |
| "grad_norm": 0.34155419020286215, | |
| "learning_rate": 1.5562047945297652e-05, | |
| "loss": 0.3921, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09211775660514832, | |
| "step": 480, | |
| "valid_targets_mean": 2812.4, | |
| "valid_targets_min": 1036 | |
| }, | |
| { | |
| "epoch": 3.1091492776886036, | |
| "grad_norm": 0.2937900120598192, | |
| "learning_rate": 1.5126831694287564e-05, | |
| "loss": 0.3875, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11049457639455795, | |
| "step": 485, | |
| "valid_targets_mean": 3831.8, | |
| "valid_targets_min": 1421 | |
| }, | |
| { | |
| "epoch": 3.1412520064205456, | |
| "grad_norm": 0.3004782410044986, | |
| "learning_rate": 1.4694055269081987e-05, | |
| "loss": 0.3897, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09211912006139755, | |
| "step": 490, | |
| "valid_targets_mean": 3683.4, | |
| "valid_targets_min": 896 | |
| }, | |
| { | |
| "epoch": 3.173354735152488, | |
| "grad_norm": 0.28620629746496773, | |
| "learning_rate": 1.4263935345778202e-05, | |
| "loss": 0.4068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1081249862909317, | |
| "step": 495, | |
| "valid_targets_mean": 4564.4, | |
| "valid_targets_min": 1295 | |
| }, | |
| { | |
| "epoch": 3.20545746388443, | |
| "grad_norm": 0.3065292114147679, | |
| "learning_rate": 1.3836687270455388e-05, | |
| "loss": 0.4003, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1026119515299797, | |
| "step": 500, | |
| "valid_targets_mean": 3764.5, | |
| "valid_targets_min": 1524 | |
| }, | |
| { | |
| "epoch": 3.2375601926163724, | |
| "grad_norm": 0.2862455916184854, | |
| "learning_rate": 1.341252495135841e-05, | |
| "loss": 0.404, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09894335269927979, | |
| "step": 505, | |
| "valid_targets_mean": 4383.1, | |
| "valid_targets_min": 1030 | |
| }, | |
| { | |
| "epoch": 3.2696629213483144, | |
| "grad_norm": 0.3238338379491629, | |
| "learning_rate": 1.299166075180133e-05, | |
| "loss": 0.4105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11098513007164001, | |
| "step": 510, | |
| "valid_targets_mean": 3665.1, | |
| "valid_targets_min": 1206 | |
| }, | |
| { | |
| "epoch": 3.301765650080257, | |
| "grad_norm": 0.31425131498588266, | |
| "learning_rate": 1.2574305383844528e-05, | |
| "loss": 0.4021, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08589449524879456, | |
| "step": 515, | |
| "valid_targets_mean": 3058.8, | |
| "valid_targets_min": 773 | |
| }, | |
| { | |
| "epoch": 3.333868378812199, | |
| "grad_norm": 0.2970738358435746, | |
| "learning_rate": 1.2160667802798503e-05, | |
| "loss": 0.4041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09149523079395294, | |
| "step": 520, | |
| "valid_targets_mean": 3748.1, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 3.365971107544141, | |
| "grad_norm": 0.2780590370917071, | |
| "learning_rate": 1.1750955102607193e-05, | |
| "loss": 0.4017, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09089787304401398, | |
| "step": 525, | |
| "valid_targets_mean": 3762.3, | |
| "valid_targets_min": 888 | |
| }, | |
| { | |
| "epoch": 3.3980738362760836, | |
| "grad_norm": 0.30125085043484123, | |
| "learning_rate": 1.1345372412163276e-05, | |
| "loss": 0.4022, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09668652713298798, | |
| "step": 530, | |
| "valid_targets_mean": 3652.8, | |
| "valid_targets_min": 1198 | |
| }, | |
| { | |
| "epoch": 3.4301765650080256, | |
| "grad_norm": 0.3393832899109687, | |
| "learning_rate": 1.094412279260726e-05, | |
| "loss": 0.4071, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09544992446899414, | |
| "step": 535, | |
| "valid_targets_mean": 3167.6, | |
| "valid_targets_min": 1258 | |
| }, | |
| { | |
| "epoch": 3.462279293739968, | |
| "grad_norm": 0.2950846626090308, | |
| "learning_rate": 1.0547407135661816e-05, | |
| "loss": 0.4, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1026640385389328, | |
| "step": 540, | |
| "valid_targets_mean": 4445.8, | |
| "valid_targets_min": 1229 | |
| }, | |
| { | |
| "epoch": 3.49438202247191, | |
| "grad_norm": 0.2841748611048485, | |
| "learning_rate": 1.0155424063052306e-05, | |
| "loss": 0.3857, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09762070327997208, | |
| "step": 545, | |
| "valid_targets_mean": 4222.4, | |
| "valid_targets_min": 1269 | |
| }, | |
| { | |
| "epoch": 3.5264847512038524, | |
| "grad_norm": 0.3352654173394246, | |
| "learning_rate": 9.768369827063803e-06, | |
| "loss": 0.4029, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11695688217878342, | |
| "step": 550, | |
| "valid_targets_mean": 3240.5, | |
| "valid_targets_min": 932 | |
| }, | |
| { | |
| "epoch": 3.5585874799357944, | |
| "grad_norm": 0.2806686898678227, | |
| "learning_rate": 9.386438212284372e-06, | |
| "loss": 0.4033, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1005910113453865, | |
| "step": 555, | |
| "valid_targets_mean": 3759.2, | |
| "valid_targets_min": 668 | |
| }, | |
| { | |
| "epoch": 3.590690208667737, | |
| "grad_norm": 0.33872852815646304, | |
| "learning_rate": 9.009820438583881e-06, | |
| "loss": 0.4145, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12395333498716354, | |
| "step": 560, | |
| "valid_targets_mean": 3597.6, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 3.6227929373996792, | |
| "grad_norm": 0.2654332382202926, | |
| "learning_rate": 8.638705065376887e-06, | |
| "loss": 0.3852, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08896806091070175, | |
| "step": 565, | |
| "valid_targets_mean": 3946.9, | |
| "valid_targets_min": 1110 | |
| }, | |
| { | |
| "epoch": 3.654895666131621, | |
| "grad_norm": 0.27981716529155093, | |
| "learning_rate": 8.27327789721746e-06, | |
| "loss": 0.376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08400297164916992, | |
| "step": 570, | |
| "valid_targets_mean": 4350.8, | |
| "valid_targets_min": 931 | |
| }, | |
| { | |
| "epoch": 3.686998394863563, | |
| "grad_norm": 0.3385370692244635, | |
| "learning_rate": 7.913721890773354e-06, | |
| "loss": 0.4033, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12360556423664093, | |
| "step": 575, | |
| "valid_targets_mean": 3591.9, | |
| "valid_targets_min": 1549 | |
| }, | |
| { | |
| "epoch": 3.7191011235955056, | |
| "grad_norm": 0.2655051672149985, | |
| "learning_rate": 7.560217063225923e-06, | |
| "loss": 0.398, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08802102506160736, | |
| "step": 580, | |
| "valid_targets_mean": 4639.4, | |
| "valid_targets_min": 1643 | |
| }, | |
| { | |
| "epoch": 3.751203852327448, | |
| "grad_norm": 0.40923046291632936, | |
| "learning_rate": 7.212940402141808e-06, | |
| "loss": 0.4014, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09749025106430054, | |
| "step": 585, | |
| "valid_targets_mean": 3517.1, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 3.78330658105939, | |
| "grad_norm": 0.3117388859264426, | |
| "learning_rate": 6.8720657768614115e-06, | |
| "loss": 0.4097, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11236117780208588, | |
| "step": 590, | |
| "valid_targets_mean": 3592.8, | |
| "valid_targets_min": 529 | |
| }, | |
| { | |
| "epoch": 3.8154093097913324, | |
| "grad_norm": 0.31179165172571244, | |
| "learning_rate": 6.537763851448593e-06, | |
| "loss": 0.4046, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10635977983474731, | |
| "step": 595, | |
| "valid_targets_mean": 3038.1, | |
| "valid_targets_min": 1038 | |
| }, | |
| { | |
| "epoch": 3.8475120385232744, | |
| "grad_norm": 0.30097213091658315, | |
| "learning_rate": 6.210201999245087e-06, | |
| "loss": 0.4153, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09912356734275818, | |
| "step": 600, | |
| "valid_targets_mean": 3369.2, | |
| "valid_targets_min": 301 | |
| }, | |
| { | |
| "epoch": 3.879614767255217, | |
| "grad_norm": 0.3092144816503789, | |
| "learning_rate": 5.889544219072465e-06, | |
| "loss": 0.4028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10438457131385803, | |
| "step": 605, | |
| "valid_targets_mean": 3691.4, | |
| "valid_targets_min": 1000 | |
| }, | |
| { | |
| "epoch": 3.911717495987159, | |
| "grad_norm": 0.28578697194711516, | |
| "learning_rate": 5.5759510531237116e-06, | |
| "loss": 0.3933, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09700490534305573, | |
| "step": 610, | |
| "valid_targets_mean": 3770.8, | |
| "valid_targets_min": 1355 | |
| }, | |
| { | |
| "epoch": 3.943820224719101, | |
| "grad_norm": 0.3272428302409307, | |
| "learning_rate": 5.269579506585259e-06, | |
| "loss": 0.3911, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08706691116094589, | |
| "step": 615, | |
| "valid_targets_mean": 3215.6, | |
| "valid_targets_min": 837 | |
| }, | |
| { | |
| "epoch": 3.975922953451043, | |
| "grad_norm": 0.2811671525014539, | |
| "learning_rate": 4.97058296903e-06, | |
| "loss": 0.401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10784287750720978, | |
| "step": 620, | |
| "valid_targets_mean": 3762.6, | |
| "valid_targets_min": 845 | |
| }, | |
| { | |
| "epoch": 4.006420545746389, | |
| "grad_norm": 0.2917034910670569, | |
| "learning_rate": 4.679111137620442e-06, | |
| "loss": 0.3997, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09262470155954361, | |
| "step": 625, | |
| "valid_targets_mean": 3377.3, | |
| "valid_targets_min": 896 | |
| }, | |
| { | |
| "epoch": 4.03852327447833, | |
| "grad_norm": 0.31513354693746726, | |
| "learning_rate": 4.395309942160551e-06, | |
| "loss": 0.3913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10473042726516724, | |
| "step": 630, | |
| "valid_targets_mean": 3944.6, | |
| "valid_targets_min": 551 | |
| }, | |
| { | |
| "epoch": 4.070626003210273, | |
| "grad_norm": 0.27934962375420436, | |
| "learning_rate": 4.119321472033779e-06, | |
| "loss": 0.3949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0756620317697525, | |
| "step": 635, | |
| "valid_targets_mean": 2551.7, | |
| "valid_targets_min": 536 | |
| }, | |
| { | |
| "epoch": 4.102728731942215, | |
| "grad_norm": 0.37095369408284645, | |
| "learning_rate": 3.851283905063863e-06, | |
| "loss": 0.3946, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12161412835121155, | |
| "step": 640, | |
| "valid_targets_mean": 3931.2, | |
| "valid_targets_min": 1237 | |
| }, | |
| { | |
| "epoch": 4.134831460674158, | |
| "grad_norm": 0.27970269585114166, | |
| "learning_rate": 3.5913314383339937e-06, | |
| "loss": 0.3872, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09911416471004486, | |
| "step": 645, | |
| "valid_targets_mean": 4231.0, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 4.166934189406099, | |
| "grad_norm": 0.3456613071245264, | |
| "learning_rate": 3.3395942209989694e-06, | |
| "loss": 0.4023, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08239921182394028, | |
| "step": 650, | |
| "valid_targets_mean": 2619.8, | |
| "valid_targets_min": 886 | |
| }, | |
| { | |
| "epoch": 4.199036918138042, | |
| "grad_norm": 0.2768716249831359, | |
| "learning_rate": 3.0961982891241083e-06, | |
| "loss": 0.3731, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09643545746803284, | |
| "step": 655, | |
| "valid_targets_mean": 3902.5, | |
| "valid_targets_min": 1105 | |
| }, | |
| { | |
| "epoch": 4.231139646869984, | |
| "grad_norm": 0.3014051499821219, | |
| "learning_rate": 2.8612655025833148e-06, | |
| "loss": 0.395, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11026032269001007, | |
| "step": 660, | |
| "valid_targets_mean": 4226.5, | |
| "valid_targets_min": 982 | |
| }, | |
| { | |
| "epoch": 4.263242375601926, | |
| "grad_norm": 0.28379604796854446, | |
| "learning_rate": 2.6349134840481294e-06, | |
| "loss": 0.3809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10869620740413666, | |
| "step": 665, | |
| "valid_targets_mean": 4199.8, | |
| "valid_targets_min": 1291 | |
| }, | |
| { | |
| "epoch": 4.295345104333869, | |
| "grad_norm": 0.32690267770951353, | |
| "learning_rate": 2.417255560098111e-06, | |
| "loss": 0.3984, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08836868405342102, | |
| "step": 670, | |
| "valid_targets_mean": 3305.6, | |
| "valid_targets_min": 561 | |
| }, | |
| { | |
| "epoch": 4.32744783306581, | |
| "grad_norm": 0.25345525186658413, | |
| "learning_rate": 2.2084007044821764e-06, | |
| "loss": 0.389, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09824281930923462, | |
| "step": 675, | |
| "valid_targets_mean": 4750.1, | |
| "valid_targets_min": 1209 | |
| }, | |
| { | |
| "epoch": 4.359550561797753, | |
| "grad_norm": 0.2689895453757197, | |
| "learning_rate": 2.0084534835592584e-06, | |
| "loss": 0.3961, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10501833260059357, | |
| "step": 680, | |
| "valid_targets_mean": 4509.6, | |
| "valid_targets_min": 1157 | |
| }, | |
| { | |
| "epoch": 4.391653290529695, | |
| "grad_norm": 0.27247257469639, | |
| "learning_rate": 1.817514003945524e-06, | |
| "loss": 0.4008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09853364527225494, | |
| "step": 685, | |
| "valid_targets_mean": 4158.9, | |
| "valid_targets_min": 726 | |
| }, | |
| { | |
| "epoch": 4.423756019261638, | |
| "grad_norm": 0.31501055011520845, | |
| "learning_rate": 1.6356778623945223e-06, | |
| "loss": 0.4052, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1491430252790451, | |
| "step": 690, | |
| "valid_targets_mean": 4383.5, | |
| "valid_targets_min": 1456 | |
| }, | |
| { | |
| "epoch": 4.455858747993579, | |
| "grad_norm": 0.2715083485471513, | |
| "learning_rate": 1.4630360979351644e-06, | |
| "loss": 0.4089, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10639771819114685, | |
| "step": 695, | |
| "valid_targets_mean": 4281.1, | |
| "valid_targets_min": 1882 | |
| }, | |
| { | |
| "epoch": 4.487961476725522, | |
| "grad_norm": 0.269592691484671, | |
| "learning_rate": 1.2996751462917057e-06, | |
| "loss": 0.4004, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.095083087682724, | |
| "step": 700, | |
| "valid_targets_mean": 3574.7, | |
| "valid_targets_min": 934 | |
| }, | |
| { | |
| "epoch": 4.520064205457464, | |
| "grad_norm": 0.28970840272710097, | |
| "learning_rate": 1.1456767966083393e-06, | |
| "loss": 0.3842, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10370156168937683, | |
| "step": 705, | |
| "valid_targets_mean": 3411.7, | |
| "valid_targets_min": 529 | |
| }, | |
| { | |
| "epoch": 4.552166934189406, | |
| "grad_norm": 0.2870909780240681, | |
| "learning_rate": 1.0011181505002488e-06, | |
| "loss": 0.3908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10310973972082138, | |
| "step": 710, | |
| "valid_targets_mean": 4500.2, | |
| "valid_targets_min": 1171 | |
| }, | |
| { | |
| "epoch": 4.584269662921348, | |
| "grad_norm": 0.2688461264330106, | |
| "learning_rate": 8.660715834514977e-07, | |
| "loss": 0.3843, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08639390021562576, | |
| "step": 715, | |
| "valid_targets_mean": 3641.4, | |
| "valid_targets_min": 922 | |
| }, | |
| { | |
| "epoch": 4.61637239165329, | |
| "grad_norm": 0.25765194882082687, | |
| "learning_rate": 7.406047085791379e-07, | |
| "loss": 0.3861, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0939197838306427, | |
| "step": 720, | |
| "valid_targets_mean": 4168.2, | |
| "valid_targets_min": 1460 | |
| }, | |
| { | |
| "epoch": 4.648475120385233, | |
| "grad_norm": 0.30160125278292577, | |
| "learning_rate": 6.247803427816945e-07, | |
| "loss": 0.4133, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11861149966716766, | |
| "step": 725, | |
| "valid_targets_mean": 3679.9, | |
| "valid_targets_min": 1359 | |
| }, | |
| { | |
| "epoch": 4.680577849117175, | |
| "grad_norm": 0.2793314638141966, | |
| "learning_rate": 5.186564752889078e-07, | |
| "loss": 0.3965, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09253215789794922, | |
| "step": 730, | |
| "valid_targets_mean": 3437.8, | |
| "valid_targets_min": 1053 | |
| }, | |
| { | |
| "epoch": 4.712680577849117, | |
| "grad_norm": 0.27123116724226537, | |
| "learning_rate": 4.22286238628562e-07, | |
| "loss": 0.3784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07941226661205292, | |
| "step": 735, | |
| "valid_targets_mean": 3344.3, | |
| "valid_targets_min": 342 | |
| }, | |
| { | |
| "epoch": 4.744783306581059, | |
| "grad_norm": 0.2785272531826042, | |
| "learning_rate": 3.357178820248641e-07, | |
| "loss": 0.3961, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08652195334434509, | |
| "step": 740, | |
| "valid_targets_mean": 3726.1, | |
| "valid_targets_min": 1384 | |
| }, | |
| { | |
| "epoch": 4.776886035313002, | |
| "grad_norm": 0.31544979123824807, | |
| "learning_rate": 2.5899474724174313e-07, | |
| "loss": 0.3988, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08774863183498383, | |
| "step": 745, | |
| "valid_targets_mean": 2830.6, | |
| "valid_targets_min": 467 | |
| }, | |
| { | |
| "epoch": 4.808988764044944, | |
| "grad_norm": 0.296446741850878, | |
| "learning_rate": 1.9215524688313893e-07, | |
| "loss": 0.3927, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10991404950618744, | |
| "step": 750, | |
| "valid_targets_mean": 3905.6, | |
| "valid_targets_min": 516 | |
| }, | |
| { | |
| "epoch": 4.841091492776886, | |
| "grad_norm": 0.27186316337463917, | |
| "learning_rate": 1.3523284516113955e-07, | |
| "loss": 0.387, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09752750396728516, | |
| "step": 755, | |
| "valid_targets_mean": 4108.7, | |
| "valid_targets_min": 962 | |
| }, | |
| { | |
| "epoch": 4.873194221508828, | |
| "grad_norm": 0.3001939708765642, | |
| "learning_rate": 8.825604114162556e-08, | |
| "loss": 0.3918, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08993087708950043, | |
| "step": 760, | |
| "valid_targets_mean": 3836.1, | |
| "valid_targets_min": 1251 | |
| }, | |
| { | |
| "epoch": 4.90529695024077, | |
| "grad_norm": 0.2867156389445644, | |
| "learning_rate": 5.1248354475768034e-08, | |
| "loss": 0.4068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09052161872386932, | |
| "step": 765, | |
| "valid_targets_mean": 3208.5, | |
| "valid_targets_min": 1013 | |
| }, | |
| { | |
| "epoch": 4.937399678972713, | |
| "grad_norm": 0.2766893813163677, | |
| "learning_rate": 2.4228313624568013e-08, | |
| "loss": 0.3876, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1123722642660141, | |
| "step": 770, | |
| "valid_targets_mean": 5074.8, | |
| "valid_targets_min": 1145 | |
| }, | |
| { | |
| "epoch": 4.969502407704655, | |
| "grad_norm": 0.339877298742656, | |
| "learning_rate": 7.209446582292501e-09, | |
| "loss": 0.4201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11707846820354462, | |
| "step": 775, | |
| "valid_targets_mean": 3143.6, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.34958645297704605, | |
| "learning_rate": 2.002741034812239e-10, | |
| "loss": 0.4005, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11478559672832489, | |
| "step": 780, | |
| "valid_targets_mean": 3291.8, | |
| "valid_targets_min": 500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11478559672832489, | |
| "step": 780, | |
| "total_flos": 1.0263973305862062e+18, | |
| "train_loss": 0.4329962073228298, | |
| "train_runtime": 13926.0897, | |
| "train_samples_per_second": 3.575, | |
| "train_steps_per_second": 0.056, | |
| "valid_targets_mean": 3291.8, | |
| "valid_targets_min": 500 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 780, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.0263973305862062e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |