diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,9650 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 7.0, + "eval_steps": 500, + "global_step": 4368, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.008012820512820512, + "grad_norm": 19.879163519026537, + "learning_rate": 3.661327231121282e-07, + "loss": 0.8272, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.8405616879463196, + "step": 5, + "valid_targets_mean": 3674.4, + "valid_targets_min": 1689 + }, + { + "epoch": 0.016025641025641024, + "grad_norm": 16.241622640859035, + "learning_rate": 8.237986270022884e-07, + "loss": 0.7792, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.8294421434402466, + "step": 10, + "valid_targets_mean": 3478.4, + "valid_targets_min": 320 + }, + { + "epoch": 0.02403846153846154, + "grad_norm": 13.729710057537002, + "learning_rate": 1.2814645308924487e-06, + "loss": 0.7564, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.7618365287780762, + "step": 15, + "valid_targets_mean": 4128.4, + "valid_targets_min": 1643 + }, + { + "epoch": 0.03205128205128205, + "grad_norm": 13.07548119294594, + "learning_rate": 1.7391304347826088e-06, + "loss": 0.76, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.8027458190917969, + "step": 20, + "valid_targets_mean": 3266.2, + "valid_targets_min": 1715 + }, + { + "epoch": 0.04006410256410257, + "grad_norm": 8.584873330406953, + "learning_rate": 2.196796338672769e-06, + "loss": 0.6846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.6333841681480408, + "step": 25, + "valid_targets_mean": 3774.2, + "valid_targets_min": 1811 + }, + { + "epoch": 0.04807692307692308, + "grad_norm": 4.675883893427763, + "learning_rate": 2.654462242562929e-06, + "loss": 0.6292, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5961663126945496, + "step": 30, + "valid_targets_mean": 3380.9, + "valid_targets_min": 247 + }, + { + "epoch": 0.05608974358974359, + "grad_norm": 2.8044998329599147, + "learning_rate": 3.1121281464530894e-06, + "loss": 0.5932, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5892564058303833, + "step": 35, + "valid_targets_mean": 3549.8, + "valid_targets_min": 590 + }, + { + "epoch": 0.0641025641025641, + "grad_norm": 1.93974754269056, + "learning_rate": 3.56979405034325e-06, + "loss": 0.565, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.534325897693634, + "step": 40, + "valid_targets_mean": 3281.8, + "valid_targets_min": 1729 + }, + { + "epoch": 0.07211538461538461, + "grad_norm": 1.7246213352284163, + "learning_rate": 4.0274599542334094e-06, + "loss": 0.57, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.6538747549057007, + "step": 45, + "valid_targets_mean": 3394.2, + "valid_targets_min": 2063 + }, + { + "epoch": 0.08012820512820513, + "grad_norm": 1.2010022045969184, + "learning_rate": 4.48512585812357e-06, + "loss": 0.4973, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4948710501194, + "step": 50, + "valid_targets_mean": 3626.2, + "valid_targets_min": 1684 + }, + { + "epoch": 0.08814102564102565, + "grad_norm": 1.0021386564368988, + "learning_rate": 4.94279176201373e-06, + "loss": 0.4607, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4545423090457916, + "step": 55, + "valid_targets_mean": 3759.3, + "valid_targets_min": 1884 + }, + { + "epoch": 0.09615384615384616, + "grad_norm": 0.8828334064941602, + "learning_rate": 5.400457665903891e-06, + "loss": 0.5137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5267658233642578, + "step": 60, + "valid_targets_mean": 3870.8, + "valid_targets_min": 2195 + }, + { + "epoch": 0.10416666666666667, + "grad_norm": 0.8508838206539899, + "learning_rate": 5.858123569794051e-06, + "loss": 0.4561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5161157250404358, + "step": 65, + "valid_targets_mean": 3584.2, + "valid_targets_min": 1879 + }, + { + "epoch": 0.11217948717948718, + "grad_norm": 0.8490931245749613, + "learning_rate": 6.31578947368421e-06, + "loss": 0.4702, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.43000292778015137, + "step": 70, + "valid_targets_mean": 2902.6, + "valid_targets_min": 418 + }, + { + "epoch": 0.1201923076923077, + "grad_norm": 0.7905351709375654, + "learning_rate": 6.773455377574372e-06, + "loss": 0.459, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4744901657104492, + "step": 75, + "valid_targets_mean": 2864.9, + "valid_targets_min": 341 + }, + { + "epoch": 0.1282051282051282, + "grad_norm": 0.6208154753019489, + "learning_rate": 7.231121281464531e-06, + "loss": 0.3937, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3898765444755554, + "step": 80, + "valid_targets_mean": 4238.9, + "valid_targets_min": 2064 + }, + { + "epoch": 0.1362179487179487, + "grad_norm": 0.6498817168712755, + "learning_rate": 7.688787185354691e-06, + "loss": 0.4217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4282737374305725, + "step": 85, + "valid_targets_mean": 3655.1, + "valid_targets_min": 2268 + }, + { + "epoch": 0.14423076923076922, + "grad_norm": 0.6460440529509714, + "learning_rate": 8.146453089244852e-06, + "loss": 0.4334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.391543447971344, + "step": 90, + "valid_targets_mean": 3515.1, + "valid_targets_min": 2125 + }, + { + "epoch": 0.15224358974358973, + "grad_norm": 0.6964132469349131, + "learning_rate": 8.604118993135013e-06, + "loss": 0.4022, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.39982515573501587, + "step": 95, + "valid_targets_mean": 3153.3, + "valid_targets_min": 668 + }, + { + "epoch": 0.16025641025641027, + "grad_norm": 0.7900128089325914, + "learning_rate": 9.061784897025172e-06, + "loss": 0.4219, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4130443036556244, + "step": 100, + "valid_targets_mean": 3466.8, + "valid_targets_min": 1967 + }, + { + "epoch": 0.16826923076923078, + "grad_norm": 0.7439846618966071, + "learning_rate": 9.519450800915333e-06, + "loss": 0.3881, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4345982074737549, + "step": 105, + "valid_targets_mean": 3018.2, + "valid_targets_min": 1376 + }, + { + "epoch": 0.1762820512820513, + "grad_norm": 0.7102582754236836, + "learning_rate": 9.977116704805492e-06, + "loss": 0.4092, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.43670815229415894, + "step": 110, + "valid_targets_mean": 3333.6, + "valid_targets_min": 600 + }, + { + "epoch": 0.1842948717948718, + "grad_norm": 0.7593872791968305, + "learning_rate": 1.0434782608695653e-05, + "loss": 0.3711, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3753435015678406, + "step": 115, + "valid_targets_mean": 2670.1, + "valid_targets_min": 316 + }, + { + "epoch": 0.19230769230769232, + "grad_norm": 0.58061569172075, + "learning_rate": 1.0892448512585814e-05, + "loss": 0.3748, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.37627002596855164, + "step": 120, + "valid_targets_mean": 3750.1, + "valid_targets_min": 2147 + }, + { + "epoch": 0.20032051282051283, + "grad_norm": 0.7535485276584365, + "learning_rate": 1.1350114416475973e-05, + "loss": 0.4141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4423384666442871, + "step": 125, + "valid_targets_mean": 2803.3, + "valid_targets_min": 900 + }, + { + "epoch": 0.20833333333333334, + "grad_norm": 0.6813290969582381, + "learning_rate": 1.1807780320366134e-05, + "loss": 0.3891, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4192976951599121, + "step": 130, + "valid_targets_mean": 3172.6, + "valid_targets_min": 2085 + }, + { + "epoch": 0.21634615384615385, + "grad_norm": 0.5032182735262825, + "learning_rate": 1.2265446224256295e-05, + "loss": 0.3456, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3107292056083679, + "step": 135, + "valid_targets_mean": 4446.8, + "valid_targets_min": 1607 + }, + { + "epoch": 0.22435897435897437, + "grad_norm": 0.5595984890274238, + "learning_rate": 1.2723112128146454e-05, + "loss": 0.3679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.37036430835723877, + "step": 140, + "valid_targets_mean": 3787.6, + "valid_targets_min": 1870 + }, + { + "epoch": 0.23237179487179488, + "grad_norm": 0.6093079141585656, + "learning_rate": 1.3180778032036615e-05, + "loss": 0.3286, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3455902338027954, + "step": 145, + "valid_targets_mean": 3455.2, + "valid_targets_min": 2049 + }, + { + "epoch": 0.2403846153846154, + "grad_norm": 0.6053924365486686, + "learning_rate": 1.3638443935926776e-05, + "loss": 0.3574, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31911373138427734, + "step": 150, + "valid_targets_mean": 3641.8, + "valid_targets_min": 379 + }, + { + "epoch": 0.2483974358974359, + "grad_norm": 0.610771646331601, + "learning_rate": 1.4096109839816933e-05, + "loss": 0.3313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3442545235157013, + "step": 155, + "valid_targets_mean": 3966.2, + "valid_targets_min": 1755 + }, + { + "epoch": 0.2564102564102564, + "grad_norm": 0.6293411508598329, + "learning_rate": 1.4553775743707096e-05, + "loss": 0.3544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34595659375190735, + "step": 160, + "valid_targets_mean": 3334.4, + "valid_targets_min": 582 + }, + { + "epoch": 0.2644230769230769, + "grad_norm": 0.6171797139949151, + "learning_rate": 1.5011441647597256e-05, + "loss": 0.3211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2953029274940491, + "step": 165, + "valid_targets_mean": 3222.6, + "valid_targets_min": 440 + }, + { + "epoch": 0.2724358974358974, + "grad_norm": 0.6190047507550094, + "learning_rate": 1.5469107551487414e-05, + "loss": 0.3297, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35699936747550964, + "step": 170, + "valid_targets_mean": 3371.2, + "valid_targets_min": 1543 + }, + { + "epoch": 0.28044871794871795, + "grad_norm": 0.7333222604739893, + "learning_rate": 1.5926773455377575e-05, + "loss": 0.3601, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3310508131980896, + "step": 175, + "valid_targets_mean": 2892.6, + "valid_targets_min": 254 + }, + { + "epoch": 0.28846153846153844, + "grad_norm": 0.5628086160657125, + "learning_rate": 1.6384439359267736e-05, + "loss": 0.3672, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30004677176475525, + "step": 180, + "valid_targets_mean": 3743.0, + "valid_targets_min": 1925 + }, + { + "epoch": 0.296474358974359, + "grad_norm": 0.6768840470894368, + "learning_rate": 1.6842105263157896e-05, + "loss": 0.3621, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3543248176574707, + "step": 185, + "valid_targets_mean": 3018.8, + "valid_targets_min": 273 + }, + { + "epoch": 0.30448717948717946, + "grad_norm": 0.5051527163277465, + "learning_rate": 1.7299771167048057e-05, + "loss": 0.3476, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3271610736846924, + "step": 190, + "valid_targets_mean": 4514.3, + "valid_targets_min": 1734 + }, + { + "epoch": 0.3125, + "grad_norm": 0.5888252395327391, + "learning_rate": 1.7757437070938218e-05, + "loss": 0.3522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3883243203163147, + "step": 195, + "valid_targets_mean": 3935.9, + "valid_targets_min": 351 + }, + { + "epoch": 0.32051282051282054, + "grad_norm": 0.6576488120247566, + "learning_rate": 1.8215102974828376e-05, + "loss": 0.3276, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.318558931350708, + "step": 200, + "valid_targets_mean": 3090.2, + "valid_targets_min": 811 + }, + { + "epoch": 0.328525641025641, + "grad_norm": 0.6235407940452166, + "learning_rate": 1.8672768878718537e-05, + "loss": 0.3255, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3435075879096985, + "step": 205, + "valid_targets_mean": 3503.6, + "valid_targets_min": 2068 + }, + { + "epoch": 0.33653846153846156, + "grad_norm": 0.6927027884314421, + "learning_rate": 1.9130434782608697e-05, + "loss": 0.3368, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35067087411880493, + "step": 210, + "valid_targets_mean": 3377.3, + "valid_targets_min": 2069 + }, + { + "epoch": 0.34455128205128205, + "grad_norm": 0.659190951187864, + "learning_rate": 1.9588100686498858e-05, + "loss": 0.3524, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.373557448387146, + "step": 215, + "valid_targets_mean": 3157.6, + "valid_targets_min": 307 + }, + { + "epoch": 0.3525641025641026, + "grad_norm": 0.7225360744876216, + "learning_rate": 2.004576659038902e-05, + "loss": 0.3242, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35884422063827515, + "step": 220, + "valid_targets_mean": 2915.2, + "valid_targets_min": 1351 + }, + { + "epoch": 0.3605769230769231, + "grad_norm": 0.6968676707115277, + "learning_rate": 2.050343249427918e-05, + "loss": 0.3248, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34566766023635864, + "step": 225, + "valid_targets_mean": 3111.0, + "valid_targets_min": 1926 + }, + { + "epoch": 0.3685897435897436, + "grad_norm": 0.6527690721075753, + "learning_rate": 2.0961098398169337e-05, + "loss": 0.3261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.348101943731308, + "step": 230, + "valid_targets_mean": 3777.4, + "valid_targets_min": 764 + }, + { + "epoch": 0.3766025641025641, + "grad_norm": 0.5509178981095164, + "learning_rate": 2.14187643020595e-05, + "loss": 0.336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3583675026893616, + "step": 235, + "valid_targets_mean": 4796.3, + "valid_targets_min": 1766 + }, + { + "epoch": 0.38461538461538464, + "grad_norm": 0.6953704961476025, + "learning_rate": 2.187643020594966e-05, + "loss": 0.3189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33428576588630676, + "step": 240, + "valid_targets_mean": 3084.4, + "valid_targets_min": 509 + }, + { + "epoch": 0.3926282051282051, + "grad_norm": 0.6779641767614547, + "learning_rate": 2.2334096109839817e-05, + "loss": 0.3149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32355090975761414, + "step": 245, + "valid_targets_mean": 3073.9, + "valid_targets_min": 317 + }, + { + "epoch": 0.40064102564102566, + "grad_norm": 0.5552383550989171, + "learning_rate": 2.279176201372998e-05, + "loss": 0.3455, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28732290863990784, + "step": 250, + "valid_targets_mean": 3637.4, + "valid_targets_min": 1476 + }, + { + "epoch": 0.40865384615384615, + "grad_norm": 0.5245790306333418, + "learning_rate": 2.3249427917620138e-05, + "loss": 0.2893, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2623603641986847, + "step": 255, + "valid_targets_mean": 5096.8, + "valid_targets_min": 2403 + }, + { + "epoch": 0.4166666666666667, + "grad_norm": 0.6195113913098604, + "learning_rate": 2.37070938215103e-05, + "loss": 0.3117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30828219652175903, + "step": 260, + "valid_targets_mean": 3688.0, + "valid_targets_min": 1285 + }, + { + "epoch": 0.42467948717948717, + "grad_norm": 0.6048964562434808, + "learning_rate": 2.4164759725400463e-05, + "loss": 0.3035, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3320136070251465, + "step": 265, + "valid_targets_mean": 4328.0, + "valid_targets_min": 469 + }, + { + "epoch": 0.4326923076923077, + "grad_norm": 0.7734208396405547, + "learning_rate": 2.462242562929062e-05, + "loss": 0.3183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33223971724510193, + "step": 270, + "valid_targets_mean": 3306.9, + "valid_targets_min": 383 + }, + { + "epoch": 0.4407051282051282, + "grad_norm": 0.693343532490161, + "learning_rate": 2.508009153318078e-05, + "loss": 0.3216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.36366701126098633, + "step": 275, + "valid_targets_mean": 3546.5, + "valid_targets_min": 2021 + }, + { + "epoch": 0.44871794871794873, + "grad_norm": 0.5065572861491737, + "learning_rate": 2.5537757437070943e-05, + "loss": 0.3094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26778098940849304, + "step": 280, + "valid_targets_mean": 4116.9, + "valid_targets_min": 1896 + }, + { + "epoch": 0.4567307692307692, + "grad_norm": 0.7350400440914835, + "learning_rate": 2.59954233409611e-05, + "loss": 0.3018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32177820801734924, + "step": 285, + "valid_targets_mean": 3166.9, + "valid_targets_min": 613 + }, + { + "epoch": 0.46474358974358976, + "grad_norm": 0.6174748140430125, + "learning_rate": 2.645308924485126e-05, + "loss": 0.3169, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31665194034576416, + "step": 290, + "valid_targets_mean": 3801.4, + "valid_targets_min": 511 + }, + { + "epoch": 0.47275641025641024, + "grad_norm": 0.6946647941876152, + "learning_rate": 2.6910755148741422e-05, + "loss": 0.301, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32155466079711914, + "step": 295, + "valid_targets_mean": 3313.8, + "valid_targets_min": 396 + }, + { + "epoch": 0.4807692307692308, + "grad_norm": 0.5710354149261667, + "learning_rate": 2.7368421052631583e-05, + "loss": 0.3263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25513502955436707, + "step": 300, + "valid_targets_mean": 3716.9, + "valid_targets_min": 2064 + }, + { + "epoch": 0.48878205128205127, + "grad_norm": 0.7373833498036741, + "learning_rate": 2.782608695652174e-05, + "loss": 0.3068, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.289334237575531, + "step": 305, + "valid_targets_mean": 3066.1, + "valid_targets_min": 432 + }, + { + "epoch": 0.4967948717948718, + "grad_norm": 0.5591989242192875, + "learning_rate": 2.8283752860411904e-05, + "loss": 0.3046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27973124384880066, + "step": 310, + "valid_targets_mean": 3920.6, + "valid_targets_min": 450 + }, + { + "epoch": 0.5048076923076923, + "grad_norm": 0.6782061420012974, + "learning_rate": 2.8741418764302062e-05, + "loss": 0.3162, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34713077545166016, + "step": 315, + "valid_targets_mean": 3371.8, + "valid_targets_min": 1638 + }, + { + "epoch": 0.5128205128205128, + "grad_norm": 0.6699054785113124, + "learning_rate": 2.9199084668192223e-05, + "loss": 0.2894, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23538798093795776, + "step": 320, + "valid_targets_mean": 3664.9, + "valid_targets_min": 2039 + }, + { + "epoch": 0.5208333333333334, + "grad_norm": 0.6116276917307956, + "learning_rate": 2.9656750572082384e-05, + "loss": 0.3189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3189501166343689, + "step": 325, + "valid_targets_mean": 3389.8, + "valid_targets_min": 1667 + }, + { + "epoch": 0.5288461538461539, + "grad_norm": 0.5857592319111676, + "learning_rate": 3.0114416475972544e-05, + "loss": 0.2898, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23993510007858276, + "step": 330, + "valid_targets_mean": 3283.4, + "valid_targets_min": 648 + }, + { + "epoch": 0.5368589743589743, + "grad_norm": 0.6736826581101825, + "learning_rate": 3.05720823798627e-05, + "loss": 0.3304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.36864790320396423, + "step": 335, + "valid_targets_mean": 3316.4, + "valid_targets_min": 1653 + }, + { + "epoch": 0.5448717948717948, + "grad_norm": 0.7664383053440278, + "learning_rate": 3.102974828375286e-05, + "loss": 0.3013, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2832982838153839, + "step": 340, + "valid_targets_mean": 3277.5, + "valid_targets_min": 1646 + }, + { + "epoch": 0.5528846153846154, + "grad_norm": 0.6889272863397136, + "learning_rate": 3.1487414187643024e-05, + "loss": 0.2838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2998211979866028, + "step": 345, + "valid_targets_mean": 3290.5, + "valid_targets_min": 407 + }, + { + "epoch": 0.5608974358974359, + "grad_norm": 0.6193116923226962, + "learning_rate": 3.1945080091533184e-05, + "loss": 0.2742, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30396538972854614, + "step": 350, + "valid_targets_mean": 3548.9, + "valid_targets_min": 1571 + }, + { + "epoch": 0.5689102564102564, + "grad_norm": 0.6347900047328812, + "learning_rate": 3.240274599542334e-05, + "loss": 0.273, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25522372126579285, + "step": 355, + "valid_targets_mean": 3351.9, + "valid_targets_min": 1902 + }, + { + "epoch": 0.5769230769230769, + "grad_norm": 0.573848093038453, + "learning_rate": 3.2860411899313506e-05, + "loss": 0.3059, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26962196826934814, + "step": 360, + "valid_targets_mean": 4155.2, + "valid_targets_min": 1606 + }, + { + "epoch": 0.5849358974358975, + "grad_norm": 0.6857168008513769, + "learning_rate": 3.331807780320366e-05, + "loss": 0.3306, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34512871503829956, + "step": 365, + "valid_targets_mean": 3127.6, + "valid_targets_min": 1484 + }, + { + "epoch": 0.592948717948718, + "grad_norm": 0.6299501708112862, + "learning_rate": 3.377574370709382e-05, + "loss": 0.3045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.281450092792511, + "step": 370, + "valid_targets_mean": 3588.4, + "valid_targets_min": 2335 + }, + { + "epoch": 0.6009615384615384, + "grad_norm": 0.650842919082811, + "learning_rate": 3.423340961098399e-05, + "loss": 0.283, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2898961901664734, + "step": 375, + "valid_targets_mean": 4018.8, + "valid_targets_min": 1989 + }, + { + "epoch": 0.6089743589743589, + "grad_norm": 0.5459269291399794, + "learning_rate": 3.469107551487414e-05, + "loss": 0.2683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25225377082824707, + "step": 380, + "valid_targets_mean": 3866.1, + "valid_targets_min": 2103 + }, + { + "epoch": 0.6169871794871795, + "grad_norm": 0.615096370203391, + "learning_rate": 3.5148741418764304e-05, + "loss": 0.314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3192776143550873, + "step": 385, + "valid_targets_mean": 3499.0, + "valid_targets_min": 1815 + }, + { + "epoch": 0.625, + "grad_norm": 0.5876066893036459, + "learning_rate": 3.5606407322654464e-05, + "loss": 0.3292, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.40550702810287476, + "step": 390, + "valid_targets_mean": 4450.1, + "valid_targets_min": 1736 + }, + { + "epoch": 0.6330128205128205, + "grad_norm": 0.636103916586035, + "learning_rate": 3.6064073226544625e-05, + "loss": 0.2862, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3379051685333252, + "step": 395, + "valid_targets_mean": 3188.5, + "valid_targets_min": 388 + }, + { + "epoch": 0.6410256410256411, + "grad_norm": 0.5330337179042688, + "learning_rate": 3.6521739130434786e-05, + "loss": 0.2867, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22062329947948456, + "step": 400, + "valid_targets_mean": 3464.9, + "valid_targets_min": 445 + }, + { + "epoch": 0.6490384615384616, + "grad_norm": 0.5889308075943929, + "learning_rate": 3.697940503432495e-05, + "loss": 0.2936, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30935603380203247, + "step": 405, + "valid_targets_mean": 3471.9, + "valid_targets_min": 1535 + }, + { + "epoch": 0.657051282051282, + "grad_norm": 0.5895846290146585, + "learning_rate": 3.743707093821511e-05, + "loss": 0.2945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2988286018371582, + "step": 410, + "valid_targets_mean": 3568.8, + "valid_targets_min": 1891 + }, + { + "epoch": 0.6650641025641025, + "grad_norm": 0.6156981066113356, + "learning_rate": 3.789473684210526e-05, + "loss": 0.2943, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2502792775630951, + "step": 415, + "valid_targets_mean": 3166.1, + "valid_targets_min": 264 + }, + { + "epoch": 0.6730769230769231, + "grad_norm": 0.6129692991693171, + "learning_rate": 3.835240274599543e-05, + "loss": 0.2829, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3190169930458069, + "step": 420, + "valid_targets_mean": 3300.9, + "valid_targets_min": 290 + }, + { + "epoch": 0.6810897435897436, + "grad_norm": 0.5946384971781778, + "learning_rate": 3.8810068649885584e-05, + "loss": 0.2926, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28893283009529114, + "step": 425, + "valid_targets_mean": 3541.3, + "valid_targets_min": 2064 + }, + { + "epoch": 0.6891025641025641, + "grad_norm": 0.5928560768818655, + "learning_rate": 3.9267734553775745e-05, + "loss": 0.2969, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30574142932891846, + "step": 430, + "valid_targets_mean": 3345.2, + "valid_targets_min": 1593 + }, + { + "epoch": 0.6971153846153846, + "grad_norm": 0.6686330453947963, + "learning_rate": 3.9725400457665905e-05, + "loss": 0.2876, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2704697847366333, + "step": 435, + "valid_targets_mean": 3345.6, + "valid_targets_min": 317 + }, + { + "epoch": 0.7051282051282052, + "grad_norm": 0.8157234796308768, + "learning_rate": 3.999997445219712e-05, + "loss": 0.2731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2545115351676941, + "step": 440, + "valid_targets_mean": 3026.8, + "valid_targets_min": 1080 + }, + { + "epoch": 0.7131410256410257, + "grad_norm": 0.6376986115267951, + "learning_rate": 3.999968704016428e-05, + "loss": 0.3057, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31646421551704407, + "step": 445, + "valid_targets_mean": 3348.3, + "valid_targets_min": 1511 + }, + { + "epoch": 0.7211538461538461, + "grad_norm": 0.6242518345573395, + "learning_rate": 3.9999080285949514e-05, + "loss": 0.2815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2892857491970062, + "step": 450, + "valid_targets_mean": 4161.2, + "valid_targets_min": 384 + }, + { + "epoch": 0.7291666666666666, + "grad_norm": 0.6432850525140379, + "learning_rate": 3.999815419924108e-05, + "loss": 0.2869, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2984856069087982, + "step": 455, + "valid_targets_mean": 3093.6, + "valid_targets_min": 367 + }, + { + "epoch": 0.7371794871794872, + "grad_norm": 0.6612758056837322, + "learning_rate": 3.999690879482614e-05, + "loss": 0.3013, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33815231919288635, + "step": 460, + "valid_targets_mean": 3653.8, + "valid_targets_min": 1170 + }, + { + "epoch": 0.7451923076923077, + "grad_norm": 0.72681333588269, + "learning_rate": 3.9995344092590506e-05, + "loss": 0.2861, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31242284178733826, + "step": 465, + "valid_targets_mean": 2851.2, + "valid_targets_min": 376 + }, + { + "epoch": 0.7532051282051282, + "grad_norm": 0.6194654033752973, + "learning_rate": 3.999346011751835e-05, + "loss": 0.2837, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2921714782714844, + "step": 470, + "valid_targets_mean": 3197.2, + "valid_targets_min": 831 + }, + { + "epoch": 0.7612179487179487, + "grad_norm": 0.4808513607264086, + "learning_rate": 3.999125689969176e-05, + "loss": 0.2757, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23430654406547546, + "step": 475, + "valid_targets_mean": 4196.6, + "valid_targets_min": 1727 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.5941447275387179, + "learning_rate": 3.9988734474290324e-05, + "loss": 0.276, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2832925021648407, + "step": 480, + "valid_targets_mean": 3244.9, + "valid_targets_min": 321 + }, + { + "epoch": 0.7772435897435898, + "grad_norm": 0.5783663517369084, + "learning_rate": 3.9985892881590513e-05, + "loss": 0.3073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3123394846916199, + "step": 485, + "valid_targets_mean": 4001.1, + "valid_targets_min": 352 + }, + { + "epoch": 0.7852564102564102, + "grad_norm": 0.6290196783075886, + "learning_rate": 3.9982732166965054e-05, + "loss": 0.2838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26579058170318604, + "step": 490, + "valid_targets_mean": 3206.8, + "valid_targets_min": 274 + }, + { + "epoch": 0.7932692307692307, + "grad_norm": 0.5515059267107406, + "learning_rate": 3.997925238088221e-05, + "loss": 0.3046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2969035804271698, + "step": 495, + "valid_targets_mean": 3815.4, + "valid_targets_min": 2430 + }, + { + "epoch": 0.8012820512820513, + "grad_norm": 0.6216409807648683, + "learning_rate": 3.9975453578904975e-05, + "loss": 0.2755, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29659634828567505, + "step": 500, + "valid_targets_mean": 3265.8, + "valid_targets_min": 2020 + }, + { + "epoch": 0.8092948717948718, + "grad_norm": 0.5969522650575887, + "learning_rate": 3.997133582169018e-05, + "loss": 0.299, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3164765238761902, + "step": 505, + "valid_targets_mean": 3195.0, + "valid_targets_min": 1424 + }, + { + "epoch": 0.8173076923076923, + "grad_norm": 0.5979583258428498, + "learning_rate": 3.996689917498754e-05, + "loss": 0.3014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30864858627319336, + "step": 510, + "valid_targets_mean": 3398.8, + "valid_targets_min": 403 + }, + { + "epoch": 0.8253205128205128, + "grad_norm": 0.6316919330012107, + "learning_rate": 3.9962143709638585e-05, + "loss": 0.2728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30174243450164795, + "step": 515, + "valid_targets_mean": 3205.8, + "valid_targets_min": 898 + }, + { + "epoch": 0.8333333333333334, + "grad_norm": 0.5430126120597503, + "learning_rate": 3.995706950157554e-05, + "loss": 0.2914, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23059752583503723, + "step": 520, + "valid_targets_mean": 3918.1, + "valid_targets_min": 290 + }, + { + "epoch": 0.8413461538461539, + "grad_norm": 0.7031490460182316, + "learning_rate": 3.995167663182008e-05, + "loss": 0.3043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33885687589645386, + "step": 525, + "valid_targets_mean": 3232.4, + "valid_targets_min": 1682 + }, + { + "epoch": 0.8493589743589743, + "grad_norm": 0.519767450790238, + "learning_rate": 3.994596518648214e-05, + "loss": 0.2889, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27818453311920166, + "step": 530, + "valid_targets_mean": 3525.6, + "valid_targets_min": 366 + }, + { + "epoch": 0.8573717948717948, + "grad_norm": 0.563048053238403, + "learning_rate": 3.993993525675838e-05, + "loss": 0.2727, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2743377089500427, + "step": 535, + "valid_targets_mean": 3589.6, + "valid_targets_min": 380 + }, + { + "epoch": 0.8653846153846154, + "grad_norm": 0.6322924611591424, + "learning_rate": 3.993358693893086e-05, + "loss": 0.2886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31251460313796997, + "step": 540, + "valid_targets_mean": 2750.7, + "valid_targets_min": 288 + }, + { + "epoch": 0.8733974358974359, + "grad_norm": 0.5801810360218201, + "learning_rate": 3.9926920334365457e-05, + "loss": 0.3023, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3175027668476105, + "step": 545, + "valid_targets_mean": 3746.8, + "valid_targets_min": 1828 + }, + { + "epoch": 0.8814102564102564, + "grad_norm": 0.5092914875070896, + "learning_rate": 3.991993554951023e-05, + "loss": 0.2896, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26988404989242554, + "step": 550, + "valid_targets_mean": 3991.2, + "valid_targets_min": 1851 + }, + { + "epoch": 0.8894230769230769, + "grad_norm": 0.5005063451467932, + "learning_rate": 3.991263269589376e-05, + "loss": 0.2507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2650403380393982, + "step": 555, + "valid_targets_mean": 3498.0, + "valid_targets_min": 1983 + }, + { + "epoch": 0.8974358974358975, + "grad_norm": 0.6082126141999025, + "learning_rate": 3.990501189012332e-05, + "loss": 0.2763, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27788347005844116, + "step": 560, + "valid_targets_mean": 3092.6, + "valid_targets_min": 354 + }, + { + "epoch": 0.905448717948718, + "grad_norm": 0.5009998118620846, + "learning_rate": 3.989707325388305e-05, + "loss": 0.2726, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2369709610939026, + "step": 565, + "valid_targets_mean": 4009.0, + "valid_targets_min": 814 + }, + { + "epoch": 0.9134615384615384, + "grad_norm": 0.6500666574085029, + "learning_rate": 3.9888816913932016e-05, + "loss": 0.2986, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28920280933380127, + "step": 570, + "valid_targets_mean": 3343.9, + "valid_targets_min": 2076 + }, + { + "epoch": 0.9214743589743589, + "grad_norm": 0.512455392251192, + "learning_rate": 3.988024300210215e-05, + "loss": 0.2624, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2702721059322357, + "step": 575, + "valid_targets_mean": 3814.2, + "valid_targets_min": 272 + }, + { + "epoch": 0.9294871794871795, + "grad_norm": 0.4624357477991238, + "learning_rate": 3.987135165529618e-05, + "loss": 0.2864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.226173996925354, + "step": 580, + "valid_targets_mean": 3846.5, + "valid_targets_min": 2247 + }, + { + "epoch": 0.9375, + "grad_norm": 0.4416044819347505, + "learning_rate": 3.9862143015485446e-05, + "loss": 0.28, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23104633390903473, + "step": 585, + "valid_targets_mean": 4618.9, + "valid_targets_min": 1440 + }, + { + "epoch": 0.9455128205128205, + "grad_norm": 0.6903001095230034, + "learning_rate": 3.985261722970759e-05, + "loss": 0.2938, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31117039918899536, + "step": 590, + "valid_targets_mean": 2427.8, + "valid_targets_min": 258 + }, + { + "epoch": 0.9535256410256411, + "grad_norm": 0.5555874802584456, + "learning_rate": 3.984277445006426e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.264557421207428, + "step": 595, + "valid_targets_mean": 3561.6, + "valid_targets_min": 1795 + }, + { + "epoch": 0.9615384615384616, + "grad_norm": 0.5008464052732203, + "learning_rate": 3.9832614833718654e-05, + "loss": 0.2679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2578531503677368, + "step": 600, + "valid_targets_mean": 3733.9, + "valid_targets_min": 2536 + }, + { + "epoch": 0.969551282051282, + "grad_norm": 0.5124597026534315, + "learning_rate": 3.9822138542893005e-05, + "loss": 0.2977, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29258236289024353, + "step": 605, + "valid_targets_mean": 4174.6, + "valid_targets_min": 369 + }, + { + "epoch": 0.9775641025641025, + "grad_norm": 0.5830748091656482, + "learning_rate": 3.9811345744866014e-05, + "loss": 0.279, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.274571031332016, + "step": 610, + "valid_targets_mean": 3175.8, + "valid_targets_min": 2039 + }, + { + "epoch": 0.9855769230769231, + "grad_norm": 0.5812110493543656, + "learning_rate": 3.980023661197016e-05, + "loss": 0.2714, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2767738103866577, + "step": 615, + "valid_targets_mean": 3417.7, + "valid_targets_min": 2031 + }, + { + "epoch": 0.9935897435897436, + "grad_norm": 0.5058825321899566, + "learning_rate": 3.978881132158896e-05, + "loss": 0.2691, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24452310800552368, + "step": 620, + "valid_targets_mean": 4135.4, + "valid_targets_min": 317 + }, + { + "epoch": 1.001602564102564, + "grad_norm": 0.6221013298738255, + "learning_rate": 3.9777070056154124e-05, + "loss": 0.2703, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3130282759666443, + "step": 625, + "valid_targets_mean": 3976.1, + "valid_targets_min": 567 + }, + { + "epoch": 1.0096153846153846, + "grad_norm": 0.5498571953724251, + "learning_rate": 3.976501300314264e-05, + "loss": 0.2692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24536371231079102, + "step": 630, + "valid_targets_mean": 3277.8, + "valid_targets_min": 1944 + }, + { + "epoch": 1.017628205128205, + "grad_norm": 0.5803974168836421, + "learning_rate": 3.9752640355073825e-05, + "loss": 0.2668, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2587840259075165, + "step": 635, + "valid_targets_mean": 3612.0, + "valid_targets_min": 2015 + }, + { + "epoch": 1.0256410256410255, + "grad_norm": 0.4989478569760299, + "learning_rate": 3.9739952309506175e-05, + "loss": 0.2815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2712844908237457, + "step": 640, + "valid_targets_mean": 3861.4, + "valid_targets_min": 2026 + }, + { + "epoch": 1.0336538461538463, + "grad_norm": 0.5731516438747354, + "learning_rate": 3.972694906903427e-05, + "loss": 0.2505, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2606925070285797, + "step": 645, + "valid_targets_mean": 3275.2, + "valid_targets_min": 1505 + }, + { + "epoch": 1.0416666666666667, + "grad_norm": 0.5992339614399704, + "learning_rate": 3.971363084128552e-05, + "loss": 0.2727, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3189937174320221, + "step": 650, + "valid_targets_mean": 3909.1, + "valid_targets_min": 1758 + }, + { + "epoch": 1.0496794871794872, + "grad_norm": 0.6286123050979071, + "learning_rate": 3.969999783891685e-05, + "loss": 0.2517, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24376559257507324, + "step": 655, + "valid_targets_mean": 2944.8, + "valid_targets_min": 1765 + }, + { + "epoch": 1.0576923076923077, + "grad_norm": 0.4966019659995407, + "learning_rate": 3.96860502796113e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25725221633911133, + "step": 660, + "valid_targets_mean": 4015.9, + "valid_targets_min": 1927 + }, + { + "epoch": 1.0657051282051282, + "grad_norm": 0.5176531998992511, + "learning_rate": 3.967178838607456e-05, + "loss": 0.2498, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29367154836654663, + "step": 665, + "valid_targets_mean": 4043.6, + "valid_targets_min": 1759 + }, + { + "epoch": 1.0737179487179487, + "grad_norm": 0.5958021710987137, + "learning_rate": 3.965721238603139e-05, + "loss": 0.2745, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2784119248390198, + "step": 670, + "valid_targets_mean": 3578.6, + "valid_targets_min": 2088 + }, + { + "epoch": 1.0817307692307692, + "grad_norm": 0.5761545845245867, + "learning_rate": 3.964232251222203e-05, + "loss": 0.2832, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27019691467285156, + "step": 675, + "valid_targets_mean": 3691.7, + "valid_targets_min": 1619 + }, + { + "epoch": 1.0897435897435896, + "grad_norm": 0.46464694268546214, + "learning_rate": 3.962711900239844e-05, + "loss": 0.255, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22501473128795624, + "step": 680, + "valid_targets_mean": 4020.6, + "valid_targets_min": 428 + }, + { + "epoch": 1.0977564102564104, + "grad_norm": 0.5923617331224703, + "learning_rate": 3.961160209932051e-05, + "loss": 0.2676, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.300213098526001, + "step": 685, + "valid_targets_mean": 3534.9, + "valid_targets_min": 275 + }, + { + "epoch": 1.1057692307692308, + "grad_norm": 0.49800360693237433, + "learning_rate": 3.95957720507522e-05, + "loss": 0.259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2590849697589874, + "step": 690, + "valid_targets_mean": 4144.4, + "valid_targets_min": 312 + }, + { + "epoch": 1.1137820512820513, + "grad_norm": 0.5249286912076265, + "learning_rate": 3.957962910945759e-05, + "loss": 0.2532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2369239628314972, + "step": 695, + "valid_targets_mean": 3731.9, + "valid_targets_min": 287 + }, + { + "epoch": 1.1217948717948718, + "grad_norm": 0.595022424517922, + "learning_rate": 3.9563173533196805e-05, + "loss": 0.2544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2349884808063507, + "step": 700, + "valid_targets_mean": 3039.2, + "valid_targets_min": 1905 + }, + { + "epoch": 1.1298076923076923, + "grad_norm": 0.46206851258117854, + "learning_rate": 3.954640558472195e-05, + "loss": 0.2535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2336539924144745, + "step": 705, + "valid_targets_mean": 4502.6, + "valid_targets_min": 1925 + }, + { + "epoch": 1.1378205128205128, + "grad_norm": 0.4976259228450712, + "learning_rate": 3.952932553177287e-05, + "loss": 0.2676, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27928584814071655, + "step": 710, + "valid_targets_mean": 4550.8, + "valid_targets_min": 2210 + }, + { + "epoch": 1.1458333333333333, + "grad_norm": 0.5827290190538823, + "learning_rate": 3.95119336470729e-05, + "loss": 0.2562, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30208781361579895, + "step": 715, + "valid_targets_mean": 3313.2, + "valid_targets_min": 322 + }, + { + "epoch": 1.1538461538461537, + "grad_norm": 0.5330675063388832, + "learning_rate": 3.949423020832451e-05, + "loss": 0.2713, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26606252789497375, + "step": 720, + "valid_targets_mean": 3708.1, + "valid_targets_min": 1381 + }, + { + "epoch": 1.1618589743589745, + "grad_norm": 0.7492896337130134, + "learning_rate": 3.947621549820485e-05, + "loss": 0.2862, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3572145700454712, + "step": 725, + "valid_targets_mean": 4327.9, + "valid_targets_min": 1430 + }, + { + "epoch": 1.169871794871795, + "grad_norm": 0.5596584335249928, + "learning_rate": 3.945788980436129e-05, + "loss": 0.2547, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23527881503105164, + "step": 730, + "valid_targets_mean": 3265.1, + "valid_targets_min": 281 + }, + { + "epoch": 1.1778846153846154, + "grad_norm": 0.523547937494331, + "learning_rate": 3.943925341940673e-05, + "loss": 0.2968, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2402142584323883, + "step": 735, + "valid_targets_mean": 3584.3, + "valid_targets_min": 2159 + }, + { + "epoch": 1.185897435897436, + "grad_norm": 0.6556254274252725, + "learning_rate": 3.942030664091503e-05, + "loss": 0.2508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2958112955093384, + "step": 740, + "valid_targets_mean": 3284.9, + "valid_targets_min": 304 + }, + { + "epoch": 1.1939102564102564, + "grad_norm": 0.7172672140383616, + "learning_rate": 3.9401049771416214e-05, + "loss": 0.298, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2700616121292114, + "step": 745, + "valid_targets_mean": 2539.8, + "valid_targets_min": 366 + }, + { + "epoch": 1.2019230769230769, + "grad_norm": 0.49454796376028104, + "learning_rate": 3.938148311839162e-05, + "loss": 0.2489, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2377432882785797, + "step": 750, + "valid_targets_mean": 3485.8, + "valid_targets_min": 1991 + }, + { + "epoch": 1.2099358974358974, + "grad_norm": 0.5811069095247919, + "learning_rate": 3.9361606994269014e-05, + "loss": 0.2564, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2677744925022125, + "step": 755, + "valid_targets_mean": 3164.1, + "valid_targets_min": 322 + }, + { + "epoch": 1.217948717948718, + "grad_norm": 0.5648309056629999, + "learning_rate": 3.934142171641763e-05, + "loss": 0.2659, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2626572847366333, + "step": 760, + "valid_targets_mean": 4323.2, + "valid_targets_min": 1127 + }, + { + "epoch": 1.2259615384615385, + "grad_norm": 0.5991473267432794, + "learning_rate": 3.9320927607143003e-05, + "loss": 0.2553, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28201889991760254, + "step": 765, + "valid_targets_mean": 3157.0, + "valid_targets_min": 1836 + }, + { + "epoch": 1.233974358974359, + "grad_norm": 0.5176752233150077, + "learning_rate": 3.9300124993681976e-05, + "loss": 0.2517, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24023833870887756, + "step": 770, + "valid_targets_mean": 3602.4, + "valid_targets_min": 1934 + }, + { + "epoch": 1.2419871794871795, + "grad_norm": 0.5206801167112293, + "learning_rate": 3.9279014208197317e-05, + "loss": 0.2325, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2505241930484772, + "step": 775, + "valid_targets_mean": 3489.1, + "valid_targets_min": 1891 + }, + { + "epoch": 1.25, + "grad_norm": 0.6675315221781205, + "learning_rate": 3.925759558777252e-05, + "loss": 0.2597, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33495059609413147, + "step": 780, + "valid_targets_mean": 3257.1, + "valid_targets_min": 2044 + }, + { + "epoch": 1.2580128205128205, + "grad_norm": 0.6330199967235978, + "learning_rate": 3.923586947440639e-05, + "loss": 0.2771, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2804538607597351, + "step": 785, + "valid_targets_mean": 3040.4, + "valid_targets_min": 1650 + }, + { + "epoch": 1.266025641025641, + "grad_norm": 0.5983476006441762, + "learning_rate": 3.921383621500758e-05, + "loss": 0.2759, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28068268299102783, + "step": 790, + "valid_targets_mean": 3209.1, + "valid_targets_min": 1899 + }, + { + "epoch": 1.2740384615384617, + "grad_norm": 0.5652783732328699, + "learning_rate": 3.919149616138906e-05, + "loss": 0.28, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3000967502593994, + "step": 795, + "valid_targets_mean": 3471.8, + "valid_targets_min": 1871 + }, + { + "epoch": 1.282051282051282, + "grad_norm": 0.48020791065961055, + "learning_rate": 3.916884967026246e-05, + "loss": 0.2642, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2259122133255005, + "step": 800, + "valid_targets_mean": 3463.9, + "valid_targets_min": 318 + }, + { + "epoch": 1.2900641025641026, + "grad_norm": 0.5952179578546953, + "learning_rate": 3.914589710323245e-05, + "loss": 0.2771, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29001694917678833, + "step": 805, + "valid_targets_mean": 3119.9, + "valid_targets_min": 328 + }, + { + "epoch": 1.2980769230769231, + "grad_norm": 0.4707806754969901, + "learning_rate": 3.912263882679091e-05, + "loss": 0.2584, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26366129517555237, + "step": 810, + "valid_targets_mean": 4289.3, + "valid_targets_min": 2249 + }, + { + "epoch": 1.3060897435897436, + "grad_norm": 0.5716840203449413, + "learning_rate": 3.9099075212311076e-05, + "loss": 0.2559, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2654898166656494, + "step": 815, + "valid_targets_mean": 2866.3, + "valid_targets_min": 2021 + }, + { + "epoch": 1.314102564102564, + "grad_norm": 0.6469688497693485, + "learning_rate": 3.9075206636041646e-05, + "loss": 0.2969, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2785649597644806, + "step": 820, + "valid_targets_mean": 2493.1, + "valid_targets_min": 325 + }, + { + "epoch": 1.3221153846153846, + "grad_norm": 0.5959874599715644, + "learning_rate": 3.905103347910075e-05, + "loss": 0.2577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2832576632499695, + "step": 825, + "valid_targets_mean": 2864.1, + "valid_targets_min": 377 + }, + { + "epoch": 1.330128205128205, + "grad_norm": 0.9264271733644018, + "learning_rate": 3.902655612746985e-05, + "loss": 0.2487, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25454455614089966, + "step": 830, + "valid_targets_mean": 3534.1, + "valid_targets_min": 322 + }, + { + "epoch": 1.3381410256410255, + "grad_norm": 0.480142441096767, + "learning_rate": 3.900177497198761e-05, + "loss": 0.2384, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22106271982192993, + "step": 835, + "valid_targets_mean": 4238.7, + "valid_targets_min": 2572 + }, + { + "epoch": 1.3461538461538463, + "grad_norm": 0.5864396452941366, + "learning_rate": 3.8976690408343635e-05, + "loss": 0.2846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24299442768096924, + "step": 840, + "valid_targets_mean": 2960.3, + "valid_targets_min": 292 + }, + { + "epoch": 1.3541666666666667, + "grad_norm": 0.5825994482188888, + "learning_rate": 3.8951302837072165e-05, + "loss": 0.2484, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2334529161453247, + "step": 845, + "valid_targets_mean": 3193.8, + "valid_targets_min": 325 + }, + { + "epoch": 1.3621794871794872, + "grad_norm": 0.5501476189263841, + "learning_rate": 3.892561266354566e-05, + "loss": 0.2824, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27922576665878296, + "step": 850, + "valid_targets_mean": 3318.4, + "valid_targets_min": 1616 + }, + { + "epoch": 1.3701923076923077, + "grad_norm": 0.547822690666954, + "learning_rate": 3.889962029796833e-05, + "loss": 0.2873, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3011276125907898, + "step": 855, + "valid_targets_mean": 3529.6, + "valid_targets_min": 1579 + }, + { + "epoch": 1.3782051282051282, + "grad_norm": 0.5347820792952802, + "learning_rate": 3.887332615536962e-05, + "loss": 0.2754, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26003679633140564, + "step": 860, + "valid_targets_mean": 3406.0, + "valid_targets_min": 651 + }, + { + "epoch": 1.3862179487179487, + "grad_norm": 0.45056781979483984, + "learning_rate": 3.8846730655597535e-05, + "loss": 0.2827, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22900214791297913, + "step": 865, + "valid_targets_mean": 3888.4, + "valid_targets_min": 600 + }, + { + "epoch": 1.3942307692307692, + "grad_norm": 0.5328070405748072, + "learning_rate": 3.881983422331198e-05, + "loss": 0.2511, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2603125274181366, + "step": 870, + "valid_targets_mean": 3456.5, + "valid_targets_min": 264 + }, + { + "epoch": 1.4022435897435899, + "grad_norm": 0.4393271897940895, + "learning_rate": 3.879263728797792e-05, + "loss": 0.2743, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24921280145645142, + "step": 875, + "valid_targets_mean": 4419.8, + "valid_targets_min": 314 + }, + { + "epoch": 1.4102564102564101, + "grad_norm": 0.5719674452916136, + "learning_rate": 3.876514028385861e-05, + "loss": 0.255, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2597891092300415, + "step": 880, + "valid_targets_mean": 2635.8, + "valid_targets_min": 298 + }, + { + "epoch": 1.4182692307692308, + "grad_norm": 0.5404887543205124, + "learning_rate": 3.873734365000857e-05, + "loss": 0.2687, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2549996078014374, + "step": 885, + "valid_targets_mean": 3332.9, + "valid_targets_min": 419 + }, + { + "epoch": 1.4262820512820513, + "grad_norm": 0.5610352900575674, + "learning_rate": 3.870924783026663e-05, + "loss": 0.2591, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3065013587474823, + "step": 890, + "valid_targets_mean": 4062.5, + "valid_targets_min": 576 + }, + { + "epoch": 1.4342948717948718, + "grad_norm": 0.5589275153379575, + "learning_rate": 3.8680853273248826e-05, + "loss": 0.2588, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23422890901565552, + "step": 895, + "valid_targets_mean": 3403.2, + "valid_targets_min": 2185 + }, + { + "epoch": 1.4423076923076923, + "grad_norm": 0.553592927050287, + "learning_rate": 3.865216043234126e-05, + "loss": 0.2565, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.253162145614624, + "step": 900, + "valid_targets_mean": 2935.9, + "valid_targets_min": 1675 + }, + { + "epoch": 1.4503205128205128, + "grad_norm": 0.4025061423275714, + "learning_rate": 3.862316976569281e-05, + "loss": 0.2328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20285256206989288, + "step": 905, + "valid_targets_mean": 4851.9, + "valid_targets_min": 2013 + }, + { + "epoch": 1.4583333333333333, + "grad_norm": 0.4772579094391561, + "learning_rate": 3.859388173620785e-05, + "loss": 0.2638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1970895528793335, + "step": 910, + "valid_targets_mean": 3699.8, + "valid_targets_min": 2229 + }, + { + "epoch": 1.4663461538461537, + "grad_norm": 0.46501156353751144, + "learning_rate": 3.8564296811538874e-05, + "loss": 0.2464, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.223637655377388, + "step": 915, + "valid_targets_mean": 3805.1, + "valid_targets_min": 1955 + }, + { + "epoch": 1.4743589743589745, + "grad_norm": 0.43790133057938885, + "learning_rate": 3.853441546407898e-05, + "loss": 0.2515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20834773778915405, + "step": 920, + "valid_targets_mean": 4177.2, + "valid_targets_min": 607 + }, + { + "epoch": 1.482371794871795, + "grad_norm": 0.5454748868394167, + "learning_rate": 3.850423817095438e-05, + "loss": 0.286, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34521013498306274, + "step": 925, + "valid_targets_mean": 4365.6, + "valid_targets_min": 1956 + }, + { + "epoch": 1.4903846153846154, + "grad_norm": 0.6094400809944239, + "learning_rate": 3.847376541401674e-05, + "loss": 0.2749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2916841506958008, + "step": 930, + "valid_targets_mean": 3232.4, + "valid_targets_min": 317 + }, + { + "epoch": 1.498397435897436, + "grad_norm": 0.5102372386762555, + "learning_rate": 3.844299767983551e-05, + "loss": 0.2502, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22648373246192932, + "step": 935, + "valid_targets_mean": 3722.2, + "valid_targets_min": 2282 + }, + { + "epoch": 1.5064102564102564, + "grad_norm": 0.5937064215878244, + "learning_rate": 3.841193545969015e-05, + "loss": 0.2426, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2533378005027771, + "step": 940, + "valid_targets_mean": 3083.6, + "valid_targets_min": 400 + }, + { + "epoch": 1.5144230769230769, + "grad_norm": 0.5764452712073888, + "learning_rate": 3.8380579249562265e-05, + "loss": 0.2503, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23182162642478943, + "step": 945, + "valid_targets_mean": 4151.0, + "valid_targets_min": 300 + }, + { + "epoch": 1.5224358974358974, + "grad_norm": 0.49751167575454874, + "learning_rate": 3.8348929550127734e-05, + "loss": 0.2524, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20357897877693176, + "step": 950, + "valid_targets_mean": 3389.6, + "valid_targets_min": 230 + }, + { + "epoch": 1.530448717948718, + "grad_norm": 0.5830545941112953, + "learning_rate": 3.831698686674866e-05, + "loss": 0.2769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28848379850387573, + "step": 955, + "valid_targets_mean": 3175.4, + "valid_targets_min": 317 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.5537102139254313, + "learning_rate": 3.828475170946534e-05, + "loss": 0.2612, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27478456497192383, + "step": 960, + "valid_targets_mean": 3087.1, + "valid_targets_min": 1798 + }, + { + "epoch": 1.546474358974359, + "grad_norm": 0.48142758344751735, + "learning_rate": 3.8252224592988087e-05, + "loss": 0.2909, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2891562581062317, + "step": 965, + "valid_targets_mean": 4012.2, + "valid_targets_min": 1886 + }, + { + "epoch": 1.5544871794871795, + "grad_norm": 0.528773456282276, + "learning_rate": 3.821940603668906e-05, + "loss": 0.2821, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26372164487838745, + "step": 970, + "valid_targets_mean": 3712.9, + "valid_targets_min": 1890 + }, + { + "epoch": 1.5625, + "grad_norm": 0.5233280901371568, + "learning_rate": 3.8186296564593924e-05, + "loss": 0.2583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28313058614730835, + "step": 975, + "valid_targets_mean": 3477.3, + "valid_targets_min": 1834 + }, + { + "epoch": 1.5705128205128205, + "grad_norm": 0.4892747737589277, + "learning_rate": 3.815289670537351e-05, + "loss": 0.2714, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2609057128429413, + "step": 980, + "valid_targets_mean": 3832.3, + "valid_targets_min": 1283 + }, + { + "epoch": 1.578525641025641, + "grad_norm": 0.5618123875185596, + "learning_rate": 3.811920699233535e-05, + "loss": 0.2695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2744769752025604, + "step": 985, + "valid_targets_mean": 3458.1, + "valid_targets_min": 2020 + }, + { + "epoch": 1.5865384615384617, + "grad_norm": 0.5183652853775259, + "learning_rate": 3.8085227963415186e-05, + "loss": 0.2781, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2586187720298767, + "step": 990, + "valid_targets_mean": 3215.7, + "valid_targets_min": 447 + }, + { + "epoch": 1.594551282051282, + "grad_norm": 0.578243895186001, + "learning_rate": 3.805096016116838e-05, + "loss": 0.2548, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26217278838157654, + "step": 995, + "valid_targets_mean": 2837.5, + "valid_targets_min": 303 + }, + { + "epoch": 1.6025641025641026, + "grad_norm": 0.44822872043640527, + "learning_rate": 3.801640413276121e-05, + "loss": 0.2587, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2369627058506012, + "step": 1000, + "valid_targets_mean": 4413.8, + "valid_targets_min": 2681 + }, + { + "epoch": 1.6105769230769231, + "grad_norm": 0.48293479766091263, + "learning_rate": 3.7981560429962204e-05, + "loss": 0.2769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25961318612098694, + "step": 1005, + "valid_targets_mean": 3498.2, + "valid_targets_min": 251 + }, + { + "epoch": 1.6185897435897436, + "grad_norm": 0.5671456917361367, + "learning_rate": 3.7946429609133274e-05, + "loss": 0.2512, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24469730257987976, + "step": 1010, + "valid_targets_mean": 3181.0, + "valid_targets_min": 655 + }, + { + "epoch": 1.626602564102564, + "grad_norm": 0.5238584082964504, + "learning_rate": 3.791101223122084e-05, + "loss": 0.2757, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2893809676170349, + "step": 1015, + "valid_targets_mean": 3797.9, + "valid_targets_min": 2157 + }, + { + "epoch": 1.6346153846153846, + "grad_norm": 0.49326302948662853, + "learning_rate": 3.787530886174688e-05, + "loss": 0.2474, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25847795605659485, + "step": 1020, + "valid_targets_mean": 3775.4, + "valid_targets_min": 1848 + }, + { + "epoch": 1.6426282051282053, + "grad_norm": 0.5099061677777234, + "learning_rate": 3.783932007079992e-05, + "loss": 0.2276, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23870500922203064, + "step": 1025, + "valid_targets_mean": 3440.9, + "valid_targets_min": 279 + }, + { + "epoch": 1.6506410256410255, + "grad_norm": 0.47352637734693354, + "learning_rate": 3.7803046433025905e-05, + "loss": 0.2552, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2686188220977783, + "step": 1030, + "valid_targets_mean": 4180.7, + "valid_targets_min": 2031 + }, + { + "epoch": 1.6586538461538463, + "grad_norm": 0.5314853755634682, + "learning_rate": 3.7766488527619024e-05, + "loss": 0.2838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3336718678474426, + "step": 1035, + "valid_targets_mean": 3948.1, + "valid_targets_min": 335 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 0.4921049254100938, + "learning_rate": 3.772964693831247e-05, + "loss": 0.2593, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2925066351890564, + "step": 1040, + "valid_targets_mean": 4141.9, + "valid_targets_min": 1127 + }, + { + "epoch": 1.6746794871794872, + "grad_norm": 0.4877598437145359, + "learning_rate": 3.7692522253369136e-05, + "loss": 0.2611, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22297003865242004, + "step": 1045, + "valid_targets_mean": 3177.9, + "valid_targets_min": 332 + }, + { + "epoch": 1.6826923076923077, + "grad_norm": 0.4446634577336513, + "learning_rate": 3.7655115065572194e-05, + "loss": 0.236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24668602645397186, + "step": 1050, + "valid_targets_mean": 4311.4, + "valid_targets_min": 2154 + }, + { + "epoch": 1.6907051282051282, + "grad_norm": 0.4967124079594226, + "learning_rate": 3.7617425972215626e-05, + "loss": 0.2499, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2651454210281372, + "step": 1055, + "valid_targets_mean": 3677.1, + "valid_targets_min": 2064 + }, + { + "epoch": 1.6987179487179487, + "grad_norm": 0.4910030484682953, + "learning_rate": 3.757945557509472e-05, + "loss": 0.2641, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22594448924064636, + "step": 1060, + "valid_targets_mean": 3567.4, + "valid_targets_min": 306 + }, + { + "epoch": 1.7067307692307692, + "grad_norm": 0.6228057575059062, + "learning_rate": 3.7541204480496444e-05, + "loss": 0.2692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2984437346458435, + "step": 1065, + "valid_targets_mean": 2969.2, + "valid_targets_min": 330 + }, + { + "epoch": 1.7147435897435899, + "grad_norm": 0.5067614506057814, + "learning_rate": 3.7502673299189745e-05, + "loss": 0.2402, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21630270779132843, + "step": 1070, + "valid_targets_mean": 3610.4, + "valid_targets_min": 2071 + }, + { + "epoch": 1.7227564102564101, + "grad_norm": 0.5417834073613709, + "learning_rate": 3.746386264641583e-05, + "loss": 0.2333, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.261353075504303, + "step": 1075, + "valid_targets_mean": 3262.4, + "valid_targets_min": 1729 + }, + { + "epoch": 1.7307692307692308, + "grad_norm": 0.441919197890755, + "learning_rate": 3.7424773141878324e-05, + "loss": 0.2458, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.267980694770813, + "step": 1080, + "valid_targets_mean": 4533.0, + "valid_targets_min": 1974 + }, + { + "epoch": 1.7387820512820513, + "grad_norm": 0.5705844782040119, + "learning_rate": 3.738540540973338e-05, + "loss": 0.2761, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3395107388496399, + "step": 1085, + "valid_targets_mean": 3350.6, + "valid_targets_min": 290 + }, + { + "epoch": 1.7467948717948718, + "grad_norm": 0.5572125349129156, + "learning_rate": 3.7345760078579695e-05, + "loss": 0.2576, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24319365620613098, + "step": 1090, + "valid_targets_mean": 3060.1, + "valid_targets_min": 1856 + }, + { + "epoch": 1.7548076923076923, + "grad_norm": 0.48020365542220134, + "learning_rate": 3.730583778144852e-05, + "loss": 0.2521, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23472854495048523, + "step": 1095, + "valid_targets_mean": 3657.5, + "valid_targets_min": 1830 + }, + { + "epoch": 1.7628205128205128, + "grad_norm": 0.5070561827519618, + "learning_rate": 3.7265639155793494e-05, + "loss": 0.2576, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25640958547592163, + "step": 1100, + "valid_targets_mean": 3536.3, + "valid_targets_min": 260 + }, + { + "epoch": 1.7708333333333335, + "grad_norm": 0.6071881430446543, + "learning_rate": 3.7225164843480503e-05, + "loss": 0.2731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2952703535556793, + "step": 1105, + "valid_targets_mean": 2960.6, + "valid_targets_min": 1856 + }, + { + "epoch": 1.7788461538461537, + "grad_norm": 0.4365264125375587, + "learning_rate": 3.7184415490777426e-05, + "loss": 0.2316, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23112201690673828, + "step": 1110, + "valid_targets_mean": 4157.2, + "valid_targets_min": 1170 + }, + { + "epoch": 1.7868589743589745, + "grad_norm": 0.5330168750066543, + "learning_rate": 3.714339174834379e-05, + "loss": 0.259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23355525732040405, + "step": 1115, + "valid_targets_mean": 2776.1, + "valid_targets_min": 465 + }, + { + "epoch": 1.7948717948717947, + "grad_norm": 0.49761685732908006, + "learning_rate": 3.710209427122044e-05, + "loss": 0.2194, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2513016164302826, + "step": 1120, + "valid_targets_mean": 3535.9, + "valid_targets_min": 250 + }, + { + "epoch": 1.8028846153846154, + "grad_norm": 0.4539573331647677, + "learning_rate": 3.7060523718819e-05, + "loss": 0.2358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20890021324157715, + "step": 1125, + "valid_targets_mean": 3893.2, + "valid_targets_min": 2223 + }, + { + "epoch": 1.810897435897436, + "grad_norm": 0.4499379546162627, + "learning_rate": 3.701868075491139e-05, + "loss": 0.2285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22595152258872986, + "step": 1130, + "valid_targets_mean": 3673.0, + "valid_targets_min": 1725 + }, + { + "epoch": 1.8189102564102564, + "grad_norm": 0.5006718742013021, + "learning_rate": 3.697656604761926e-05, + "loss": 0.2603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28063613176345825, + "step": 1135, + "valid_targets_mean": 3769.2, + "valid_targets_min": 2063 + }, + { + "epoch": 1.8269230769230769, + "grad_norm": 0.47557270821506, + "learning_rate": 3.693418026940325e-05, + "loss": 0.2421, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2146185040473938, + "step": 1140, + "valid_targets_mean": 3377.9, + "valid_targets_min": 1997 + }, + { + "epoch": 1.8349358974358974, + "grad_norm": 0.5718766251672308, + "learning_rate": 3.689152409705229e-05, + "loss": 0.2629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26902586221694946, + "step": 1145, + "valid_targets_mean": 2666.6, + "valid_targets_min": 273 + }, + { + "epoch": 1.842948717948718, + "grad_norm": 0.49857619295485034, + "learning_rate": 3.6848598211672794e-05, + "loss": 0.2329, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21846452355384827, + "step": 1150, + "valid_targets_mean": 3345.5, + "valid_targets_min": 1624 + }, + { + "epoch": 1.8509615384615383, + "grad_norm": 0.5385567351558466, + "learning_rate": 3.6805403298677797e-05, + "loss": 0.253, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20614317059516907, + "step": 1155, + "valid_targets_mean": 2915.1, + "valid_targets_min": 401 + }, + { + "epoch": 1.858974358974359, + "grad_norm": 0.5178461800642131, + "learning_rate": 3.6761940047775966e-05, + "loss": 0.2605, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2558942437171936, + "step": 1160, + "valid_targets_mean": 3066.9, + "valid_targets_min": 2139 + }, + { + "epoch": 1.8669871794871795, + "grad_norm": 0.5473587738559478, + "learning_rate": 3.671820915296063e-05, + "loss": 0.2569, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2264281064271927, + "step": 1165, + "valid_targets_mean": 2752.2, + "valid_targets_min": 354 + }, + { + "epoch": 1.875, + "grad_norm": 0.5742898492979889, + "learning_rate": 3.667421131249869e-05, + "loss": 0.2599, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2911713123321533, + "step": 1170, + "valid_targets_mean": 2875.4, + "valid_targets_min": 351 + }, + { + "epoch": 1.8830128205128205, + "grad_norm": 0.5463479296932019, + "learning_rate": 3.662994722891946e-05, + "loss": 0.2639, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2465297281742096, + "step": 1175, + "valid_targets_mean": 3061.2, + "valid_targets_min": 361 + }, + { + "epoch": 1.891025641025641, + "grad_norm": 0.40591617828758964, + "learning_rate": 3.658541760900344e-05, + "loss": 0.2529, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2039278745651245, + "step": 1180, + "valid_targets_mean": 4176.2, + "valid_targets_min": 345 + }, + { + "epoch": 1.8990384615384617, + "grad_norm": 0.516920853045726, + "learning_rate": 3.654062316377106e-05, + "loss": 0.2833, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26839733123779297, + "step": 1185, + "valid_targets_mean": 3415.8, + "valid_targets_min": 2208 + }, + { + "epoch": 1.907051282051282, + "grad_norm": 0.49950356462293516, + "learning_rate": 3.649556460847131e-05, + "loss": 0.2479, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20692500472068787, + "step": 1190, + "valid_targets_mean": 3404.6, + "valid_targets_min": 429 + }, + { + "epoch": 1.9150641025641026, + "grad_norm": 0.8809308213127865, + "learning_rate": 3.6450242662570314e-05, + "loss": 0.2411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2121295928955078, + "step": 1195, + "valid_targets_mean": 4477.4, + "valid_targets_min": 1613 + }, + { + "epoch": 1.9230769230769231, + "grad_norm": 0.4779923123123419, + "learning_rate": 3.6404658049739854e-05, + "loss": 0.2313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2340589463710785, + "step": 1200, + "valid_targets_mean": 3211.6, + "valid_targets_min": 250 + }, + { + "epoch": 1.9310897435897436, + "grad_norm": 0.414184625487714, + "learning_rate": 3.63588114978458e-05, + "loss": 0.2468, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2192663848400116, + "step": 1205, + "valid_targets_mean": 4318.8, + "valid_targets_min": 266 + }, + { + "epoch": 1.939102564102564, + "grad_norm": 0.5303697493272574, + "learning_rate": 3.6312703738936504e-05, + "loss": 0.2409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22840480506420135, + "step": 1210, + "valid_targets_mean": 3477.6, + "valid_targets_min": 2264 + }, + { + "epoch": 1.9471153846153846, + "grad_norm": 0.43566224462419423, + "learning_rate": 3.626633550923111e-05, + "loss": 0.2545, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2339707314968109, + "step": 1215, + "valid_targets_mean": 4116.4, + "valid_targets_min": 1590 + }, + { + "epoch": 1.9551282051282053, + "grad_norm": 0.5197010556809256, + "learning_rate": 3.621970754910778e-05, + "loss": 0.2377, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22132663428783417, + "step": 1220, + "valid_targets_mean": 2858.6, + "valid_targets_min": 434 + }, + { + "epoch": 1.9631410256410255, + "grad_norm": 0.5473460491584194, + "learning_rate": 3.6172820603091885e-05, + "loss": 0.2444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2865564823150635, + "step": 1225, + "valid_targets_mean": 3204.2, + "valid_targets_min": 2245 + }, + { + "epoch": 1.9711538461538463, + "grad_norm": 0.4360570251401029, + "learning_rate": 3.612567541984413e-05, + "loss": 0.2297, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23854400217533112, + "step": 1230, + "valid_targets_mean": 4248.2, + "valid_targets_min": 1842 + }, + { + "epoch": 1.9791666666666665, + "grad_norm": 0.4735446183087179, + "learning_rate": 3.6078272752148574e-05, + "loss": 0.2314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1998017132282257, + "step": 1235, + "valid_targets_mean": 3103.8, + "valid_targets_min": 1315 + }, + { + "epoch": 1.9871794871794872, + "grad_norm": 0.4556927270069073, + "learning_rate": 3.6030613356900635e-05, + "loss": 0.2312, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18791323900222778, + "step": 1240, + "valid_targets_mean": 3856.6, + "valid_targets_min": 449 + }, + { + "epoch": 1.9951923076923077, + "grad_norm": 0.45598040579973126, + "learning_rate": 3.598269799509498e-05, + "loss": 0.2629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2848716378211975, + "step": 1245, + "valid_targets_mean": 4495.0, + "valid_targets_min": 2194 + }, + { + "epoch": 2.003205128205128, + "grad_norm": 0.48113601096988307, + "learning_rate": 3.5934527431813385e-05, + "loss": 0.2382, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20847183465957642, + "step": 1250, + "valid_targets_mean": 3337.9, + "valid_targets_min": 2074 + }, + { + "epoch": 2.011217948717949, + "grad_norm": 0.4464604635105121, + "learning_rate": 3.5886102436212536e-05, + "loss": 0.244, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2278617024421692, + "step": 1255, + "valid_targets_mean": 4564.4, + "valid_targets_min": 1381 + }, + { + "epoch": 2.019230769230769, + "grad_norm": 0.46286257830943056, + "learning_rate": 3.583742378151171e-05, + "loss": 0.2247, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21383799612522125, + "step": 1260, + "valid_targets_mean": 4035.4, + "valid_targets_min": 1676 + }, + { + "epoch": 2.02724358974359, + "grad_norm": 0.5034513672207104, + "learning_rate": 3.5788492244980464e-05, + "loss": 0.2469, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22386686503887177, + "step": 1265, + "valid_targets_mean": 3764.6, + "valid_targets_min": 646 + }, + { + "epoch": 2.03525641025641, + "grad_norm": 0.5178445959097301, + "learning_rate": 3.573930860792621e-05, + "loss": 0.2243, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22068476676940918, + "step": 1270, + "valid_targets_mean": 2955.2, + "valid_targets_min": 233 + }, + { + "epoch": 2.043269230769231, + "grad_norm": 0.4828450286557547, + "learning_rate": 3.568987365568173e-05, + "loss": 0.2393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.271858274936676, + "step": 1275, + "valid_targets_mean": 4249.4, + "valid_targets_min": 1675 + }, + { + "epoch": 2.051282051282051, + "grad_norm": 0.5734483721951875, + "learning_rate": 3.564018817759266e-05, + "loss": 0.2535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2882147431373596, + "step": 1280, + "valid_targets_mean": 3304.8, + "valid_targets_min": 1263 + }, + { + "epoch": 2.059294871794872, + "grad_norm": 0.5637741231035333, + "learning_rate": 3.559025296700484e-05, + "loss": 0.2239, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21064125001430511, + "step": 1285, + "valid_targets_mean": 2992.1, + "valid_targets_min": 469 + }, + { + "epoch": 2.0673076923076925, + "grad_norm": 0.5631234037431226, + "learning_rate": 3.554006882125173e-05, + "loss": 0.2459, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24107757210731506, + "step": 1290, + "valid_targets_mean": 3415.8, + "valid_targets_min": 1812 + }, + { + "epoch": 2.0753205128205128, + "grad_norm": 0.47018159183118846, + "learning_rate": 3.5489636541641586e-05, + "loss": 0.2144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20437480509281158, + "step": 1295, + "valid_targets_mean": 3452.8, + "valid_targets_min": 332 + }, + { + "epoch": 2.0833333333333335, + "grad_norm": 0.4880151515161931, + "learning_rate": 3.543895693344472e-05, + "loss": 0.2334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24443921446800232, + "step": 1300, + "valid_targets_mean": 3210.9, + "valid_targets_min": 383 + }, + { + "epoch": 2.0913461538461537, + "grad_norm": 0.5616738633854279, + "learning_rate": 3.538803080588063e-05, + "loss": 0.2552, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2829934358596802, + "step": 1305, + "valid_targets_mean": 3090.2, + "valid_targets_min": 830 + }, + { + "epoch": 2.0993589743589745, + "grad_norm": 0.5145409981741939, + "learning_rate": 3.5336858972105076e-05, + "loss": 0.2285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26258933544158936, + "step": 1310, + "valid_targets_mean": 3607.2, + "valid_targets_min": 1777 + }, + { + "epoch": 2.1073717948717947, + "grad_norm": 0.4770200458940273, + "learning_rate": 3.528544224919708e-05, + "loss": 0.2237, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20944449305534363, + "step": 1315, + "valid_targets_mean": 3943.2, + "valid_targets_min": 1951 + }, + { + "epoch": 2.1153846153846154, + "grad_norm": 0.5351286659289585, + "learning_rate": 3.5233781458145934e-05, + "loss": 0.2468, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.264823317527771, + "step": 1320, + "valid_targets_mean": 3264.5, + "valid_targets_min": 1890 + }, + { + "epoch": 2.123397435897436, + "grad_norm": 0.4364617795621253, + "learning_rate": 3.5181877423838034e-05, + "loss": 0.227, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21104183793067932, + "step": 1325, + "valid_targets_mean": 3822.4, + "valid_targets_min": 544 + }, + { + "epoch": 2.1314102564102564, + "grad_norm": 0.5579212740013267, + "learning_rate": 3.512973097504371e-05, + "loss": 0.2463, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25203123688697815, + "step": 1330, + "valid_targets_mean": 3057.1, + "valid_targets_min": 874 + }, + { + "epoch": 2.139423076923077, + "grad_norm": 0.5560163039281574, + "learning_rate": 3.507734294440403e-05, + "loss": 0.2273, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21084235608577728, + "step": 1335, + "valid_targets_mean": 3401.2, + "valid_targets_min": 1968 + }, + { + "epoch": 2.1474358974358974, + "grad_norm": 0.5550177081038501, + "learning_rate": 3.50247141684175e-05, + "loss": 0.2409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2602787911891937, + "step": 1340, + "valid_targets_mean": 2963.6, + "valid_targets_min": 476 + }, + { + "epoch": 2.155448717948718, + "grad_norm": 0.4025082575367705, + "learning_rate": 3.497184548742667e-05, + "loss": 0.2287, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22489029169082642, + "step": 1345, + "valid_targets_mean": 4617.6, + "valid_targets_min": 1693 + }, + { + "epoch": 2.1634615384615383, + "grad_norm": 0.5459612756781311, + "learning_rate": 3.491873774560473e-05, + "loss": 0.2304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25036120414733887, + "step": 1350, + "valid_targets_mean": 3089.4, + "valid_targets_min": 1401 + }, + { + "epoch": 2.171474358974359, + "grad_norm": 0.5395675971796833, + "learning_rate": 3.486539179094208e-05, + "loss": 0.2489, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29087430238723755, + "step": 1355, + "valid_targets_mean": 3440.6, + "valid_targets_min": 332 + }, + { + "epoch": 2.1794871794871793, + "grad_norm": 0.5092018797334458, + "learning_rate": 3.481180847523272e-05, + "loss": 0.2289, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21696212887763977, + "step": 1360, + "valid_targets_mean": 3313.4, + "valid_targets_min": 2144 + }, + { + "epoch": 2.1875, + "grad_norm": 0.5295673988104233, + "learning_rate": 3.4757988654060684e-05, + "loss": 0.2212, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2391449213027954, + "step": 1365, + "valid_targets_mean": 3206.8, + "valid_targets_min": 1028 + }, + { + "epoch": 2.1955128205128207, + "grad_norm": 1.3613971129842901, + "learning_rate": 3.470393318678637e-05, + "loss": 0.2473, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.222117081284523, + "step": 1370, + "valid_targets_mean": 3327.1, + "valid_targets_min": 2051 + }, + { + "epoch": 2.203525641025641, + "grad_norm": 0.4981966116272187, + "learning_rate": 3.4649642936532836e-05, + "loss": 0.2289, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23258623480796814, + "step": 1375, + "valid_targets_mean": 3291.8, + "valid_targets_min": 314 + }, + { + "epoch": 2.2115384615384617, + "grad_norm": 0.43365813086550076, + "learning_rate": 3.4595118770171984e-05, + "loss": 0.2328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19812996685504913, + "step": 1380, + "valid_targets_mean": 3913.2, + "valid_targets_min": 368 + }, + { + "epoch": 2.219551282051282, + "grad_norm": 0.5247182768095835, + "learning_rate": 3.454036155831077e-05, + "loss": 0.2429, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1894225925207138, + "step": 1385, + "valid_targets_mean": 3751.2, + "valid_targets_min": 288 + }, + { + "epoch": 2.2275641025641026, + "grad_norm": 0.553269199889545, + "learning_rate": 3.4485372175277236e-05, + "loss": 0.2224, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2299507111310959, + "step": 1390, + "valid_targets_mean": 2955.2, + "valid_targets_min": 1472 + }, + { + "epoch": 2.235576923076923, + "grad_norm": 0.5464234829613472, + "learning_rate": 3.44301514991066e-05, + "loss": 0.2146, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23231425881385803, + "step": 1395, + "valid_targets_mean": 3088.8, + "valid_targets_min": 610 + }, + { + "epoch": 2.2435897435897436, + "grad_norm": 0.5180228358093808, + "learning_rate": 3.4374700411527225e-05, + "loss": 0.2386, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22594726085662842, + "step": 1400, + "valid_targets_mean": 3432.4, + "valid_targets_min": 1748 + }, + { + "epoch": 2.251602564102564, + "grad_norm": 0.5488375513914536, + "learning_rate": 3.431901979794653e-05, + "loss": 0.2476, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24900490045547485, + "step": 1405, + "valid_targets_mean": 3217.9, + "valid_targets_min": 1691 + }, + { + "epoch": 2.2596153846153846, + "grad_norm": 0.505291776579264, + "learning_rate": 3.426311054743685e-05, + "loss": 0.2405, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23657619953155518, + "step": 1410, + "valid_targets_mean": 3181.9, + "valid_targets_min": 420 + }, + { + "epoch": 2.2676282051282053, + "grad_norm": 0.5373306561158637, + "learning_rate": 3.420697355272127e-05, + "loss": 0.222, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20928433537483215, + "step": 1415, + "valid_targets_mean": 3262.9, + "valid_targets_min": 408 + }, + { + "epoch": 2.2756410256410255, + "grad_norm": 0.43766174495312943, + "learning_rate": 3.415060971015933e-05, + "loss": 0.2297, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22512772679328918, + "step": 1420, + "valid_targets_mean": 3874.0, + "valid_targets_min": 2642 + }, + { + "epoch": 2.2836538461538463, + "grad_norm": 0.5536725881247396, + "learning_rate": 3.4094019919732736e-05, + "loss": 0.2472, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2986356019973755, + "step": 1425, + "valid_targets_mean": 3210.8, + "valid_targets_min": 1565 + }, + { + "epoch": 2.2916666666666665, + "grad_norm": 0.4934531732234053, + "learning_rate": 3.403720508503098e-05, + "loss": 0.2235, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2186366319656372, + "step": 1430, + "valid_targets_mean": 3384.0, + "valid_targets_min": 406 + }, + { + "epoch": 2.2996794871794872, + "grad_norm": 0.5009294955215486, + "learning_rate": 3.398016611323693e-05, + "loss": 0.2224, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21268531680107117, + "step": 1435, + "valid_targets_mean": 3506.9, + "valid_targets_min": 1143 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.48874671172712036, + "learning_rate": 3.392290391511232e-05, + "loss": 0.2137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21587032079696655, + "step": 1440, + "valid_targets_mean": 3306.5, + "valid_targets_min": 2000 + }, + { + "epoch": 2.315705128205128, + "grad_norm": 0.5103195501063624, + "learning_rate": 3.386541940498322e-05, + "loss": 0.2263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2232041209936142, + "step": 1445, + "valid_targets_mean": 3811.5, + "valid_targets_min": 917 + }, + { + "epoch": 2.323717948717949, + "grad_norm": 0.4970664229701647, + "learning_rate": 3.380771350072543e-05, + "loss": 0.236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23805755376815796, + "step": 1450, + "valid_targets_mean": 3688.0, + "valid_targets_min": 1925 + }, + { + "epoch": 2.331730769230769, + "grad_norm": 0.4510517770993502, + "learning_rate": 3.374978712374986e-05, + "loss": 0.23, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21762406826019287, + "step": 1455, + "valid_targets_mean": 4083.8, + "valid_targets_min": 2197 + }, + { + "epoch": 2.33974358974359, + "grad_norm": 0.4381525573648137, + "learning_rate": 3.369164119898774e-05, + "loss": 0.2375, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1905115395784378, + "step": 1460, + "valid_targets_mean": 3869.7, + "valid_targets_min": 1743 + }, + { + "epoch": 2.34775641025641, + "grad_norm": 0.4764400553773267, + "learning_rate": 3.363327665487593e-05, + "loss": 0.2547, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23225094377994537, + "step": 1465, + "valid_targets_mean": 3675.2, + "valid_targets_min": 1755 + }, + { + "epoch": 2.355769230769231, + "grad_norm": 0.4939814659194403, + "learning_rate": 3.357469442334206e-05, + "loss": 0.241, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26246967911720276, + "step": 1470, + "valid_targets_mean": 3462.3, + "valid_targets_min": 576 + }, + { + "epoch": 2.363782051282051, + "grad_norm": 0.5401786052759295, + "learning_rate": 3.351589543978965e-05, + "loss": 0.2221, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23349204659461975, + "step": 1475, + "valid_targets_mean": 2899.4, + "valid_targets_min": 361 + }, + { + "epoch": 2.371794871794872, + "grad_norm": 0.5472139904225963, + "learning_rate": 3.345688064308317e-05, + "loss": 0.2216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2089230865240097, + "step": 1480, + "valid_targets_mean": 3110.4, + "valid_targets_min": 245 + }, + { + "epoch": 2.3798076923076925, + "grad_norm": 0.458331587340864, + "learning_rate": 3.339765097553307e-05, + "loss": 0.2369, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21575820446014404, + "step": 1485, + "valid_targets_mean": 3449.1, + "valid_targets_min": 580 + }, + { + "epoch": 2.3878205128205128, + "grad_norm": 0.49455068163325516, + "learning_rate": 3.33382073828807e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24723072350025177, + "step": 1490, + "valid_targets_mean": 3645.0, + "valid_targets_min": 446 + }, + { + "epoch": 2.3958333333333335, + "grad_norm": 0.4386780332732727, + "learning_rate": 3.327855081428326e-05, + "loss": 0.2594, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22455403208732605, + "step": 1495, + "valid_targets_mean": 4253.0, + "valid_targets_min": 2122 + }, + { + "epoch": 2.4038461538461537, + "grad_norm": 0.5595129873547966, + "learning_rate": 3.3218682222298584e-05, + "loss": 0.2376, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26848047971725464, + "step": 1500, + "valid_targets_mean": 3699.6, + "valid_targets_min": 582 + }, + { + "epoch": 2.4118589743589745, + "grad_norm": 0.5203649586160911, + "learning_rate": 3.315860256286996e-05, + "loss": 0.2121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2161019742488861, + "step": 1505, + "valid_targets_mean": 2969.0, + "valid_targets_min": 294 + }, + { + "epoch": 2.4198717948717947, + "grad_norm": 0.48238723093848074, + "learning_rate": 3.3098312795310894e-05, + "loss": 0.2217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20345598459243774, + "step": 1510, + "valid_targets_mean": 3430.6, + "valid_targets_min": 314 + }, + { + "epoch": 2.4278846153846154, + "grad_norm": 0.6327412024094834, + "learning_rate": 3.303781388228974e-05, + "loss": 0.222, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22221553325653076, + "step": 1515, + "valid_targets_mean": 2144.2, + "valid_targets_min": 250 + }, + { + "epoch": 2.435897435897436, + "grad_norm": 0.6837869827215011, + "learning_rate": 3.297710678981435e-05, + "loss": 0.2246, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1765064001083374, + "step": 1520, + "valid_targets_mean": 4076.7, + "valid_targets_min": 2191 + }, + { + "epoch": 2.4439102564102564, + "grad_norm": 0.4872367895970891, + "learning_rate": 3.291619248721667e-05, + "loss": 0.2337, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2066897302865982, + "step": 1525, + "valid_targets_mean": 3390.9, + "valid_targets_min": 1469 + }, + { + "epoch": 2.451923076923077, + "grad_norm": 0.4839123688592673, + "learning_rate": 3.285507194713724e-05, + "loss": 0.2202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22590945661067963, + "step": 1530, + "valid_targets_mean": 3387.2, + "valid_targets_min": 2066 + }, + { + "epoch": 2.4599358974358974, + "grad_norm": 0.5310641416737413, + "learning_rate": 3.279374614550966e-05, + "loss": 0.2291, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22865507006645203, + "step": 1535, + "valid_targets_mean": 3314.9, + "valid_targets_min": 1059 + }, + { + "epoch": 2.467948717948718, + "grad_norm": 0.5984703425117173, + "learning_rate": 3.2732216061545e-05, + "loss": 0.2197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21253696084022522, + "step": 1540, + "valid_targets_mean": 2831.1, + "valid_targets_min": 250 + }, + { + "epoch": 2.4759615384615383, + "grad_norm": 0.5213539179148546, + "learning_rate": 3.2670482677716214e-05, + "loss": 0.2369, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26152336597442627, + "step": 1545, + "valid_targets_mean": 3748.8, + "valid_targets_min": 2097 + }, + { + "epoch": 2.483974358974359, + "grad_norm": 0.5596224583501783, + "learning_rate": 3.2608546979742394e-05, + "loss": 0.2292, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24703338742256165, + "step": 1550, + "valid_targets_mean": 3326.6, + "valid_targets_min": 1712 + }, + { + "epoch": 2.4919871794871793, + "grad_norm": 0.5054522907981005, + "learning_rate": 3.254640995657307e-05, + "loss": 0.2312, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18621867895126343, + "step": 1555, + "valid_targets_mean": 2977.2, + "valid_targets_min": 576 + }, + { + "epoch": 2.5, + "grad_norm": 0.5207121321246797, + "learning_rate": 3.248407260037239e-05, + "loss": 0.2126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2505119740962982, + "step": 1560, + "valid_targets_mean": 3519.9, + "valid_targets_min": 2315 + }, + { + "epoch": 2.5080128205128203, + "grad_norm": 0.5397392067361689, + "learning_rate": 3.24215359065033e-05, + "loss": 0.2315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20214936137199402, + "step": 1565, + "valid_targets_mean": 3165.1, + "valid_targets_min": 1968 + }, + { + "epoch": 2.516025641025641, + "grad_norm": 0.5486644133797549, + "learning_rate": 3.235880087351164e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2726089358329773, + "step": 1570, + "valid_targets_mean": 3464.1, + "valid_targets_min": 1756 + }, + { + "epoch": 2.5240384615384617, + "grad_norm": 0.554056515656035, + "learning_rate": 3.2295868503110184e-05, + "loss": 0.2275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18808777630329132, + "step": 1575, + "valid_targets_mean": 4925.1, + "valid_targets_min": 1712 + }, + { + "epoch": 2.532051282051282, + "grad_norm": 0.5753807285501622, + "learning_rate": 3.22327398001627e-05, + "loss": 0.2258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23577101528644562, + "step": 1580, + "valid_targets_mean": 3408.1, + "valid_targets_min": 1686 + }, + { + "epoch": 2.5400641025641026, + "grad_norm": 0.47516390006880044, + "learning_rate": 3.216941577266783e-05, + "loss": 0.2541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19967889785766602, + "step": 1585, + "valid_targets_mean": 3382.3, + "valid_targets_min": 360 + }, + { + "epoch": 2.5480769230769234, + "grad_norm": 0.4871279952298379, + "learning_rate": 3.210589743174308e-05, + "loss": 0.2129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20786455273628235, + "step": 1590, + "valid_targets_mean": 3622.5, + "valid_targets_min": 1935 + }, + { + "epoch": 2.5560897435897436, + "grad_norm": 0.5420223468811577, + "learning_rate": 3.204218579160857e-05, + "loss": 0.2062, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18591973185539246, + "step": 1595, + "valid_targets_mean": 3835.8, + "valid_targets_min": 300 + }, + { + "epoch": 2.564102564102564, + "grad_norm": 0.4941582897081752, + "learning_rate": 3.197828186957094e-05, + "loss": 0.2051, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23235294222831726, + "step": 1600, + "valid_targets_mean": 3895.9, + "valid_targets_min": 1975 + }, + { + "epoch": 2.5721153846153846, + "grad_norm": 0.5759213015140178, + "learning_rate": 3.191418668600705e-05, + "loss": 0.2166, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27510184049606323, + "step": 1605, + "valid_targets_mean": 3370.2, + "valid_targets_min": 1699 + }, + { + "epoch": 2.5801282051282053, + "grad_norm": 0.4916915979617188, + "learning_rate": 3.184990126434771e-05, + "loss": 0.2317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2400546669960022, + "step": 1610, + "valid_targets_mean": 3853.6, + "valid_targets_min": 346 + }, + { + "epoch": 2.5881410256410255, + "grad_norm": 0.5134223213888042, + "learning_rate": 3.178542663106131e-05, + "loss": 0.2449, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25739049911499023, + "step": 1615, + "valid_targets_mean": 3560.9, + "valid_targets_min": 2072 + }, + { + "epoch": 2.5961538461538463, + "grad_norm": 0.4647485387209341, + "learning_rate": 3.172076381563748e-05, + "loss": 0.225, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2345583736896515, + "step": 1620, + "valid_targets_mean": 4102.4, + "valid_targets_min": 1909 + }, + { + "epoch": 2.6041666666666665, + "grad_norm": 0.4613965020941828, + "learning_rate": 3.165591385057058e-05, + "loss": 0.199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18396924436092377, + "step": 1625, + "valid_targets_mean": 3841.2, + "valid_targets_min": 1765 + }, + { + "epoch": 2.6121794871794872, + "grad_norm": 0.5606790940088371, + "learning_rate": 3.1590877771343316e-05, + "loss": 0.2211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2248254120349884, + "step": 1630, + "valid_targets_mean": 3164.4, + "valid_targets_min": 488 + }, + { + "epoch": 2.6201923076923075, + "grad_norm": 0.5628286317328028, + "learning_rate": 3.152565661641008e-05, + "loss": 0.245, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2901548445224762, + "step": 1635, + "valid_targets_mean": 3357.9, + "valid_targets_min": 2079 + }, + { + "epoch": 2.628205128205128, + "grad_norm": 0.4835026603266905, + "learning_rate": 3.1460251427180474e-05, + "loss": 0.2333, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21858114004135132, + "step": 1640, + "valid_targets_mean": 3778.3, + "valid_targets_min": 302 + }, + { + "epoch": 2.636217948717949, + "grad_norm": 0.44764340943402625, + "learning_rate": 3.139466324800263e-05, + "loss": 0.2205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19746120274066925, + "step": 1645, + "valid_targets_mean": 3783.5, + "valid_targets_min": 361 + }, + { + "epoch": 2.644230769230769, + "grad_norm": 0.4756491019431985, + "learning_rate": 3.132889312614655e-05, + "loss": 0.2274, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2135154902935028, + "step": 1650, + "valid_targets_mean": 4373.3, + "valid_targets_min": 2480 + }, + { + "epoch": 2.65224358974359, + "grad_norm": 0.499792970874057, + "learning_rate": 3.126294211178737e-05, + "loss": 0.2076, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21535325050354004, + "step": 1655, + "valid_targets_mean": 3278.2, + "valid_targets_min": 2029 + }, + { + "epoch": 2.66025641025641, + "grad_norm": 0.509858262662174, + "learning_rate": 3.1196811257988634e-05, + "loss": 0.2303, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24099260568618774, + "step": 1660, + "valid_targets_mean": 3947.8, + "valid_targets_min": 2214 + }, + { + "epoch": 2.668269230769231, + "grad_norm": 0.49425844374690214, + "learning_rate": 3.1130501620685394e-05, + "loss": 0.2134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2178635597229004, + "step": 1665, + "valid_targets_mean": 3340.4, + "valid_targets_min": 1623 + }, + { + "epoch": 2.676282051282051, + "grad_norm": 0.5160738309260722, + "learning_rate": 3.106401425866745e-05, + "loss": 0.2197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2119971215724945, + "step": 1670, + "valid_targets_mean": 3309.1, + "valid_targets_min": 388 + }, + { + "epoch": 2.684294871794872, + "grad_norm": 0.4827614168487263, + "learning_rate": 3.099735023356236e-05, + "loss": 0.2228, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2164524495601654, + "step": 1675, + "valid_targets_mean": 3530.8, + "valid_targets_min": 302 + }, + { + "epoch": 2.6923076923076925, + "grad_norm": 0.5031362133138242, + "learning_rate": 3.0930510609818564e-05, + "loss": 0.2302, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2503780126571655, + "step": 1680, + "valid_targets_mean": 3499.4, + "valid_targets_min": 1833 + }, + { + "epoch": 2.7003205128205128, + "grad_norm": 0.4600421225293802, + "learning_rate": 3.086349645468831e-05, + "loss": 0.2189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20820432901382446, + "step": 1685, + "valid_targets_mean": 3730.4, + "valid_targets_min": 1619 + }, + { + "epoch": 2.7083333333333335, + "grad_norm": 0.4879718110605468, + "learning_rate": 3.079630883821067e-05, + "loss": 0.2129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23680952191352844, + "step": 1690, + "valid_targets_mean": 3882.0, + "valid_targets_min": 2360 + }, + { + "epoch": 2.7163461538461537, + "grad_norm": 0.43727736757613994, + "learning_rate": 3.0728948833194436e-05, + "loss": 0.2174, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17322327196598053, + "step": 1695, + "valid_targets_mean": 3951.0, + "valid_targets_min": 2267 + }, + { + "epoch": 2.7243589743589745, + "grad_norm": 0.4879962165083738, + "learning_rate": 3.066141751520099e-05, + "loss": 0.2221, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19267430901527405, + "step": 1700, + "valid_targets_mean": 3747.5, + "valid_targets_min": 1944 + }, + { + "epoch": 2.7323717948717947, + "grad_norm": 0.457795201986618, + "learning_rate": 3.059371596252712e-05, + "loss": 0.225, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18401110172271729, + "step": 1705, + "valid_targets_mean": 3678.1, + "valid_targets_min": 534 + }, + { + "epoch": 2.7403846153846154, + "grad_norm": 0.5151213135732414, + "learning_rate": 3.0525845256187834e-05, + "loss": 0.2267, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2076324224472046, + "step": 1710, + "valid_targets_mean": 3722.7, + "valid_targets_min": 290 + }, + { + "epoch": 2.748397435897436, + "grad_norm": 0.5497157184393276, + "learning_rate": 3.0457806479899044e-05, + "loss": 0.2258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23892144858837128, + "step": 1715, + "valid_targets_mean": 3113.8, + "valid_targets_min": 1779 + }, + { + "epoch": 2.7564102564102564, + "grad_norm": 0.4177740808857472, + "learning_rate": 3.0389600720060318e-05, + "loss": 0.2151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19988121092319489, + "step": 1720, + "valid_targets_mean": 4454.9, + "valid_targets_min": 2529 + }, + { + "epoch": 2.7644230769230766, + "grad_norm": 0.40429971426500544, + "learning_rate": 3.0321229065737522e-05, + "loss": 0.2071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1552007794380188, + "step": 1725, + "valid_targets_mean": 3879.1, + "valid_targets_min": 2008 + }, + { + "epoch": 2.7724358974358974, + "grad_norm": 0.5259528054083097, + "learning_rate": 3.0252692608645384e-05, + "loss": 0.2322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2267281860113144, + "step": 1730, + "valid_targets_mean": 3345.8, + "valid_targets_min": 2071 + }, + { + "epoch": 2.780448717948718, + "grad_norm": 0.5358999746656999, + "learning_rate": 3.0183992443130127e-05, + "loss": 0.2302, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26404300332069397, + "step": 1735, + "valid_targets_mean": 3579.4, + "valid_targets_min": 1938 + }, + { + "epoch": 2.7884615384615383, + "grad_norm": 0.511982571691961, + "learning_rate": 3.011512966615195e-05, + "loss": 0.2274, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23434904217720032, + "step": 1740, + "valid_targets_mean": 3328.4, + "valid_targets_min": 651 + }, + { + "epoch": 2.796474358974359, + "grad_norm": 0.48335543579817647, + "learning_rate": 3.0046105377267523e-05, + "loss": 0.223, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1833260953426361, + "step": 1745, + "valid_targets_mean": 3190.2, + "valid_targets_min": 607 + }, + { + "epoch": 2.8044871794871797, + "grad_norm": 0.43039906371305947, + "learning_rate": 2.9976920678612456e-05, + "loss": 0.2211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1997445970773697, + "step": 1750, + "valid_targets_mean": 4131.9, + "valid_targets_min": 1676 + }, + { + "epoch": 2.8125, + "grad_norm": 0.4933277471312345, + "learning_rate": 2.9907576674883664e-05, + "loss": 0.2437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23295000195503235, + "step": 1755, + "valid_targets_mean": 3472.1, + "valid_targets_min": 1975 + }, + { + "epoch": 2.8205128205128203, + "grad_norm": 0.5681686885351188, + "learning_rate": 2.983807447332174e-05, + "loss": 0.2174, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20448961853981018, + "step": 1760, + "valid_targets_mean": 2844.5, + "valid_targets_min": 314 + }, + { + "epoch": 2.828525641025641, + "grad_norm": 0.5668449658402579, + "learning_rate": 2.9768415183693293e-05, + "loss": 0.2335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24366453289985657, + "step": 1765, + "valid_targets_mean": 2975.3, + "valid_targets_min": 322 + }, + { + "epoch": 2.8365384615384617, + "grad_norm": 0.5226268711251213, + "learning_rate": 2.9698599918273197e-05, + "loss": 0.2155, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19929251074790955, + "step": 1770, + "valid_targets_mean": 3103.5, + "valid_targets_min": 1884 + }, + { + "epoch": 2.844551282051282, + "grad_norm": 0.5302151675543736, + "learning_rate": 2.962862979182686e-05, + "loss": 0.2358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21836702525615692, + "step": 1775, + "valid_targets_mean": 3503.3, + "valid_targets_min": 2282 + }, + { + "epoch": 2.8525641025641026, + "grad_norm": 0.5027916097789561, + "learning_rate": 2.95585059215924e-05, + "loss": 0.218, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17366638779640198, + "step": 1780, + "valid_targets_mean": 3490.8, + "valid_targets_min": 1933 + }, + { + "epoch": 2.8605769230769234, + "grad_norm": 0.4601505846710778, + "learning_rate": 2.948822942726284e-05, + "loss": 0.2163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18435326218605042, + "step": 1785, + "valid_targets_mean": 4212.1, + "valid_targets_min": 1640 + }, + { + "epoch": 2.8685897435897436, + "grad_norm": 0.5293089677478494, + "learning_rate": 2.941780143096817e-05, + "loss": 0.2359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23952870070934296, + "step": 1790, + "valid_targets_mean": 3439.8, + "valid_targets_min": 1702 + }, + { + "epoch": 2.876602564102564, + "grad_norm": 0.5897045687575828, + "learning_rate": 2.9347223057257505e-05, + "loss": 0.2475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2963252067565918, + "step": 1795, + "valid_targets_mean": 3266.6, + "valid_targets_min": 246 + }, + { + "epoch": 2.8846153846153846, + "grad_norm": 0.48554553955268526, + "learning_rate": 2.927649543308106e-05, + "loss": 0.2004, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1865456998348236, + "step": 1800, + "valid_targets_mean": 3357.2, + "valid_targets_min": 501 + }, + { + "epoch": 2.8926282051282053, + "grad_norm": 0.4898939438514679, + "learning_rate": 2.9205619687772212e-05, + "loss": 0.2122, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19000311195850372, + "step": 1805, + "valid_targets_mean": 3306.9, + "valid_targets_min": 659 + }, + { + "epoch": 2.9006410256410255, + "grad_norm": 0.44510432099453506, + "learning_rate": 2.9134596953029413e-05, + "loss": 0.2171, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19600239396095276, + "step": 1810, + "valid_targets_mean": 3557.1, + "valid_targets_min": 1282 + }, + { + "epoch": 2.9086538461538463, + "grad_norm": 0.5066463304119703, + "learning_rate": 2.9063428362898168e-05, + "loss": 0.2289, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23164188861846924, + "step": 1815, + "valid_targets_mean": 3555.6, + "valid_targets_min": 2122 + }, + { + "epoch": 2.9166666666666665, + "grad_norm": 0.5756911724515725, + "learning_rate": 2.8992115053752905e-05, + "loss": 0.2197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2190612107515335, + "step": 1820, + "valid_targets_mean": 2842.9, + "valid_targets_min": 418 + }, + { + "epoch": 2.9246794871794872, + "grad_norm": 0.5255305070731653, + "learning_rate": 2.8920658164278816e-05, + "loss": 0.2115, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22007790207862854, + "step": 1825, + "valid_targets_mean": 3013.8, + "valid_targets_min": 277 + }, + { + "epoch": 2.9326923076923075, + "grad_norm": 0.4959332729505306, + "learning_rate": 2.884905883545373e-05, + "loss": 0.2317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2223539650440216, + "step": 1830, + "valid_targets_mean": 3726.8, + "valid_targets_min": 2423 + }, + { + "epoch": 2.940705128205128, + "grad_norm": 0.4602219302613781, + "learning_rate": 2.877731821052981e-05, + "loss": 0.2108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20421993732452393, + "step": 1835, + "valid_targets_mean": 3851.6, + "valid_targets_min": 777 + }, + { + "epoch": 2.948717948717949, + "grad_norm": 0.5758106780745842, + "learning_rate": 2.8705437435015375e-05, + "loss": 0.2275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22423945367336273, + "step": 1840, + "valid_targets_mean": 2997.2, + "valid_targets_min": 309 + }, + { + "epoch": 2.956730769230769, + "grad_norm": 0.5555841227697801, + "learning_rate": 2.8633417656656566e-05, + "loss": 0.2108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24468722939491272, + "step": 1845, + "valid_targets_mean": 2970.1, + "valid_targets_min": 434 + }, + { + "epoch": 2.96474358974359, + "grad_norm": 0.5519550879359697, + "learning_rate": 2.8561260025419036e-05, + "loss": 0.2056, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2539096474647522, + "step": 1850, + "valid_targets_mean": 3302.1, + "valid_targets_min": 1767 + }, + { + "epoch": 2.97275641025641, + "grad_norm": 0.5521763069467881, + "learning_rate": 2.8488965693469583e-05, + "loss": 0.2346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20984898507595062, + "step": 1855, + "valid_targets_mean": 3273.4, + "valid_targets_min": 1086 + }, + { + "epoch": 2.980769230769231, + "grad_norm": 0.3863036704502182, + "learning_rate": 2.8416535815157763e-05, + "loss": 0.2223, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.183609738945961, + "step": 1860, + "valid_targets_mean": 4792.1, + "valid_targets_min": 582 + }, + { + "epoch": 2.988782051282051, + "grad_norm": 0.5142968904085405, + "learning_rate": 2.8343971546997434e-05, + "loss": 0.2147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2554502487182617, + "step": 1865, + "valid_targets_mean": 4478.8, + "valid_targets_min": 2145 + }, + { + "epoch": 2.996794871794872, + "grad_norm": 0.4895495909278828, + "learning_rate": 2.827127404764831e-05, + "loss": 0.2082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1824975162744522, + "step": 1870, + "valid_targets_mean": 3639.9, + "valid_targets_min": 655 + }, + { + "epoch": 3.0048076923076925, + "grad_norm": 0.5077644475380257, + "learning_rate": 2.8198444477897467e-05, + "loss": 0.2074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20675477385520935, + "step": 1875, + "valid_targets_mean": 3273.4, + "valid_targets_min": 834 + }, + { + "epoch": 3.0128205128205128, + "grad_norm": 0.7231231356441381, + "learning_rate": 2.8125484000640787e-05, + "loss": 0.2175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21694618463516235, + "step": 1880, + "valid_targets_mean": 2460.2, + "valid_targets_min": 290 + }, + { + "epoch": 3.0208333333333335, + "grad_norm": 0.5418323064797319, + "learning_rate": 2.8052393780864394e-05, + "loss": 0.2214, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2313728928565979, + "step": 1885, + "valid_targets_mean": 3570.8, + "valid_targets_min": 2119 + }, + { + "epoch": 3.0288461538461537, + "grad_norm": 0.4734695604472562, + "learning_rate": 2.797917498562607e-05, + "loss": 0.2044, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20120325684547424, + "step": 1890, + "valid_targets_mean": 3707.3, + "valid_targets_min": 1993 + }, + { + "epoch": 3.0368589743589745, + "grad_norm": 0.5208029845200504, + "learning_rate": 2.7905828784036596e-05, + "loss": 0.2184, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20613202452659607, + "step": 1895, + "valid_targets_mean": 3343.9, + "valid_targets_min": 1936 + }, + { + "epoch": 3.0448717948717947, + "grad_norm": 0.5636930047627939, + "learning_rate": 2.78323563472411e-05, + "loss": 0.2206, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2063179761171341, + "step": 1900, + "valid_targets_mean": 3101.1, + "valid_targets_min": 390 + }, + { + "epoch": 3.0528846153846154, + "grad_norm": 0.47028704072176375, + "learning_rate": 2.7758758848400354e-05, + "loss": 0.2141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2136302888393402, + "step": 1905, + "valid_targets_mean": 4080.3, + "valid_targets_min": 1671 + }, + { + "epoch": 3.0608974358974357, + "grad_norm": 0.6406329634245468, + "learning_rate": 2.7685037462672043e-05, + "loss": 0.2067, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20517919957637787, + "step": 1910, + "valid_targets_mean": 2988.4, + "valid_targets_min": 340 + }, + { + "epoch": 3.0689102564102564, + "grad_norm": 0.5208968741487041, + "learning_rate": 2.7611193367191993e-05, + "loss": 0.216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2274111807346344, + "step": 1915, + "valid_targets_mean": 3925.6, + "valid_targets_min": 1472 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.6213844783621272, + "learning_rate": 2.7537227741055378e-05, + "loss": 0.2291, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29447296261787415, + "step": 1920, + "valid_targets_mean": 3099.1, + "valid_targets_min": 1267 + }, + { + "epoch": 3.0849358974358974, + "grad_norm": 0.6223062142576921, + "learning_rate": 2.746314176529791e-05, + "loss": 0.2195, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2739785313606262, + "step": 1925, + "valid_targets_mean": 2990.8, + "valid_targets_min": 1535 + }, + { + "epoch": 3.092948717948718, + "grad_norm": 0.5068820414313022, + "learning_rate": 2.7388936622876957e-05, + "loss": 0.2228, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2140953093767166, + "step": 1930, + "valid_targets_mean": 3561.2, + "valid_targets_min": 1882 + }, + { + "epoch": 3.1009615384615383, + "grad_norm": 0.6073432596277979, + "learning_rate": 2.7314613498652663e-05, + "loss": 0.2248, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23374855518341064, + "step": 1935, + "valid_targets_mean": 3133.8, + "valid_targets_min": 1705 + }, + { + "epoch": 3.108974358974359, + "grad_norm": 0.5401937808022614, + "learning_rate": 2.7240173579369025e-05, + "loss": 0.2127, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21288909018039703, + "step": 1940, + "valid_targets_mean": 3231.6, + "valid_targets_min": 383 + }, + { + "epoch": 3.1169871794871793, + "grad_norm": 0.5574637188150374, + "learning_rate": 2.7165618053634962e-05, + "loss": 0.2236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27656134963035583, + "step": 1945, + "valid_targets_mean": 3314.0, + "valid_targets_min": 2044 + }, + { + "epoch": 3.125, + "grad_norm": 0.49173386570489425, + "learning_rate": 2.7090948111905304e-05, + "loss": 0.2125, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18874695897102356, + "step": 1950, + "valid_targets_mean": 3627.7, + "valid_targets_min": 2175 + }, + { + "epoch": 3.1330128205128207, + "grad_norm": 0.6238382719891059, + "learning_rate": 2.701616494646183e-05, + "loss": 0.2037, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20941844582557678, + "step": 1955, + "valid_targets_mean": 3470.2, + "valid_targets_min": 2062 + }, + { + "epoch": 3.141025641025641, + "grad_norm": 0.487803607393044, + "learning_rate": 2.6941269751394174e-05, + "loss": 0.1959, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16428649425506592, + "step": 1960, + "valid_targets_mean": 3103.4, + "valid_targets_min": 265 + }, + { + "epoch": 3.1490384615384617, + "grad_norm": 0.6360432454399796, + "learning_rate": 2.686626372258081e-05, + "loss": 0.2169, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2702599763870239, + "step": 1965, + "valid_targets_mean": 2839.3, + "valid_targets_min": 1024 + }, + { + "epoch": 3.157051282051282, + "grad_norm": 0.49950293386656874, + "learning_rate": 2.6791148057669913e-05, + "loss": 0.2133, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20391499996185303, + "step": 1970, + "valid_targets_mean": 3623.2, + "valid_targets_min": 1986 + }, + { + "epoch": 3.1650641025641026, + "grad_norm": 0.4641725143928944, + "learning_rate": 2.671592395606027e-05, + "loss": 0.2351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23568058013916016, + "step": 1975, + "valid_targets_mean": 4207.4, + "valid_targets_min": 1762 + }, + { + "epoch": 3.173076923076923, + "grad_norm": 0.5731658630297737, + "learning_rate": 2.6640592618882114e-05, + "loss": 0.206, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18935617804527283, + "step": 1980, + "valid_targets_mean": 2550.4, + "valid_targets_min": 290 + }, + { + "epoch": 3.1810897435897436, + "grad_norm": 0.4182322068592183, + "learning_rate": 2.656515524897795e-05, + "loss": 0.2097, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1807328462600708, + "step": 1985, + "valid_targets_mean": 4199.2, + "valid_targets_min": 308 + }, + { + "epoch": 3.189102564102564, + "grad_norm": 0.56077657394403, + "learning_rate": 2.6489613050883343e-05, + "loss": 0.2156, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19504913687705994, + "step": 1990, + "valid_targets_mean": 3384.6, + "valid_targets_min": 874 + }, + { + "epoch": 3.1971153846153846, + "grad_norm": 0.48875364275581146, + "learning_rate": 2.6413967230807677e-05, + "loss": 0.2289, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24835677444934845, + "step": 1995, + "valid_targets_mean": 4170.1, + "valid_targets_min": 1950 + }, + { + "epoch": 3.2051282051282053, + "grad_norm": 0.465054834927291, + "learning_rate": 2.6338218996614924e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20396825671195984, + "step": 2000, + "valid_targets_mean": 4063.9, + "valid_targets_min": 1911 + }, + { + "epoch": 3.2131410256410255, + "grad_norm": 0.5825104361550973, + "learning_rate": 2.6262369557804325e-05, + "loss": 0.2194, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26649874448776245, + "step": 2005, + "valid_targets_mean": 3483.6, + "valid_targets_min": 1575 + }, + { + "epoch": 3.2211538461538463, + "grad_norm": 0.5518087060055086, + "learning_rate": 2.6186420125491094e-05, + "loss": 0.2045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21316856145858765, + "step": 2010, + "valid_targets_mean": 3461.9, + "valid_targets_min": 262 + }, + { + "epoch": 3.2291666666666665, + "grad_norm": 0.4932050175961384, + "learning_rate": 2.6110371912387083e-05, + "loss": 0.2251, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2211998999118805, + "step": 2015, + "valid_targets_mean": 3540.0, + "valid_targets_min": 2048 + }, + { + "epoch": 3.2371794871794872, + "grad_norm": 0.9755740013917427, + "learning_rate": 2.6034226132781407e-05, + "loss": 0.2155, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17934134602546692, + "step": 2020, + "valid_targets_mean": 3694.6, + "valid_targets_min": 352 + }, + { + "epoch": 3.2451923076923075, + "grad_norm": 0.5365875996452869, + "learning_rate": 2.5957984002521066e-05, + "loss": 0.2073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2610040307044983, + "step": 2025, + "valid_targets_mean": 3428.6, + "valid_targets_min": 1713 + }, + { + "epoch": 3.253205128205128, + "grad_norm": 0.5319413568742882, + "learning_rate": 2.588164673899151e-05, + "loss": 0.1979, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22881156206130981, + "step": 2030, + "valid_targets_mean": 3281.9, + "valid_targets_min": 2165 + }, + { + "epoch": 3.261217948717949, + "grad_norm": 0.4728884688960481, + "learning_rate": 2.580521556109724e-05, + "loss": 0.2176, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23082220554351807, + "step": 2035, + "valid_targets_mean": 4267.6, + "valid_targets_min": 2654 + }, + { + "epoch": 3.269230769230769, + "grad_norm": 0.5954186304706786, + "learning_rate": 2.57286916892423e-05, + "loss": 0.2096, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2065027952194214, + "step": 2040, + "valid_targets_mean": 3015.3, + "valid_targets_min": 363 + }, + { + "epoch": 3.27724358974359, + "grad_norm": 0.4581325903962391, + "learning_rate": 2.5652076345310822e-05, + "loss": 0.2264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1993183046579361, + "step": 2045, + "valid_targets_mean": 3884.2, + "valid_targets_min": 1911 + }, + { + "epoch": 3.28525641025641, + "grad_norm": 0.4978220491849944, + "learning_rate": 2.5575370752647507e-05, + "loss": 0.2124, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21730056405067444, + "step": 2050, + "valid_targets_mean": 3592.5, + "valid_targets_min": 1719 + }, + { + "epoch": 3.293269230769231, + "grad_norm": 0.48150684905539476, + "learning_rate": 2.5498576136038077e-05, + "loss": 0.2143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2327374815940857, + "step": 2055, + "valid_targets_mean": 3775.5, + "valid_targets_min": 315 + }, + { + "epoch": 3.301282051282051, + "grad_norm": 0.6130312357702806, + "learning_rate": 2.542169372168976e-05, + "loss": 0.2031, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21109043061733246, + "step": 2060, + "valid_targets_mean": 3150.4, + "valid_targets_min": 1877 + }, + { + "epoch": 3.309294871794872, + "grad_norm": 0.5526478381871899, + "learning_rate": 2.5344724737211646e-05, + "loss": 0.211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24314221739768982, + "step": 2065, + "valid_targets_mean": 3232.4, + "valid_targets_min": 1884 + }, + { + "epoch": 3.3173076923076925, + "grad_norm": 0.6292693413383127, + "learning_rate": 2.5267670411595152e-05, + "loss": 0.2135, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24798676371574402, + "step": 2070, + "valid_targets_mean": 3371.9, + "valid_targets_min": 522 + }, + { + "epoch": 3.3253205128205128, + "grad_norm": 0.5349155064935684, + "learning_rate": 2.5190531975194345e-05, + "loss": 0.2385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24136458337306976, + "step": 2075, + "valid_targets_mean": 3143.9, + "valid_targets_min": 1876 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.5778043108253703, + "learning_rate": 2.5113310659706322e-05, + "loss": 0.2454, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.254383385181427, + "step": 2080, + "valid_targets_mean": 3132.8, + "valid_targets_min": 318 + }, + { + "epoch": 3.3413461538461537, + "grad_norm": 0.4561195758923805, + "learning_rate": 2.5036007698151553e-05, + "loss": 0.2096, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16544273495674133, + "step": 2085, + "valid_targets_mean": 3583.4, + "valid_targets_min": 2025 + }, + { + "epoch": 3.3493589743589745, + "grad_norm": 0.472371073834948, + "learning_rate": 2.4958624324854185e-05, + "loss": 0.2072, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1999911367893219, + "step": 2090, + "valid_targets_mean": 3940.1, + "valid_targets_min": 251 + }, + { + "epoch": 3.3573717948717947, + "grad_norm": 0.47468781614384653, + "learning_rate": 2.4881161775422303e-05, + "loss": 0.2189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20175859332084656, + "step": 2095, + "valid_targets_mean": 3614.2, + "valid_targets_min": 1835 + }, + { + "epoch": 3.3653846153846154, + "grad_norm": 0.5463743266445147, + "learning_rate": 2.480362128672824e-05, + "loss": 0.2161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2244294434785843, + "step": 2100, + "valid_targets_mean": 3027.9, + "valid_targets_min": 1231 + }, + { + "epoch": 3.373397435897436, + "grad_norm": 0.5091167365835734, + "learning_rate": 2.4726004096888817e-05, + "loss": 0.2244, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22996219992637634, + "step": 2105, + "valid_targets_mean": 3577.7, + "valid_targets_min": 2051 + }, + { + "epoch": 3.3814102564102564, + "grad_norm": 0.5525551859037144, + "learning_rate": 2.4648311445245558e-05, + "loss": 0.2266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22371171414852142, + "step": 2110, + "valid_targets_mean": 3024.2, + "valid_targets_min": 327 + }, + { + "epoch": 3.389423076923077, + "grad_norm": 0.5245484166047455, + "learning_rate": 2.457054457234493e-05, + "loss": 0.2354, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2585659623146057, + "step": 2115, + "valid_targets_mean": 3375.2, + "valid_targets_min": 2277 + }, + { + "epoch": 3.3974358974358974, + "grad_norm": 0.5021432681126327, + "learning_rate": 2.4492704719918497e-05, + "loss": 0.1972, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20290249586105347, + "step": 2120, + "valid_targets_mean": 3324.5, + "valid_targets_min": 1705 + }, + { + "epoch": 3.405448717948718, + "grad_norm": 0.4075352140717916, + "learning_rate": 2.4414793130863134e-05, + "loss": 0.1947, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17111021280288696, + "step": 2125, + "valid_targets_mean": 4749.3, + "valid_targets_min": 1861 + }, + { + "epoch": 3.4134615384615383, + "grad_norm": 0.4912694843388253, + "learning_rate": 2.433681104922114e-05, + "loss": 0.205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19937452673912048, + "step": 2130, + "valid_targets_mean": 3759.5, + "valid_targets_min": 1848 + }, + { + "epoch": 3.421474358974359, + "grad_norm": 0.46633524863143777, + "learning_rate": 2.4258759720160412e-05, + "loss": 0.199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20374462008476257, + "step": 2135, + "valid_targets_mean": 4000.0, + "valid_targets_min": 2040 + }, + { + "epoch": 3.4294871794871793, + "grad_norm": 0.5353379245393521, + "learning_rate": 2.4180640389954534e-05, + "loss": 0.2123, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19159536063671112, + "step": 2140, + "valid_targets_mean": 3524.2, + "valid_targets_min": 1662 + }, + { + "epoch": 3.4375, + "grad_norm": 0.5106404124718604, + "learning_rate": 2.4102454305962892e-05, + "loss": 0.2012, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20838648080825806, + "step": 2145, + "valid_targets_mean": 3315.6, + "valid_targets_min": 298 + }, + { + "epoch": 3.4455128205128207, + "grad_norm": 0.5117939711713273, + "learning_rate": 2.402420271661076e-05, + "loss": 0.2253, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21036981046199799, + "step": 2150, + "valid_targets_mean": 3242.2, + "valid_targets_min": 244 + }, + { + "epoch": 3.453525641025641, + "grad_norm": 0.5489890660628814, + "learning_rate": 2.3945886871369338e-05, + "loss": 0.2311, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2170429229736328, + "step": 2155, + "valid_targets_mean": 2959.5, + "valid_targets_min": 1973 + }, + { + "epoch": 3.4615384615384617, + "grad_norm": 0.5220848220225659, + "learning_rate": 2.3867508020735865e-05, + "loss": 0.2157, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23258650302886963, + "step": 2160, + "valid_targets_mean": 3513.6, + "valid_targets_min": 290 + }, + { + "epoch": 3.469551282051282, + "grad_norm": 0.5177385972382765, + "learning_rate": 2.3789067416213568e-05, + "loss": 0.2247, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24745973944664001, + "step": 2165, + "valid_targets_mean": 3519.9, + "valid_targets_min": 837 + }, + { + "epoch": 3.4775641025641026, + "grad_norm": 0.5008188668113256, + "learning_rate": 2.3710566310291733e-05, + "loss": 0.2163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23902934789657593, + "step": 2170, + "valid_targets_mean": 3710.1, + "valid_targets_min": 1978 + }, + { + "epoch": 3.485576923076923, + "grad_norm": 0.47611524773645536, + "learning_rate": 2.36320059564257e-05, + "loss": 0.2197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20398235321044922, + "step": 2175, + "valid_targets_mean": 3758.6, + "valid_targets_min": 1973 + }, + { + "epoch": 3.4935897435897436, + "grad_norm": 0.5366226972721881, + "learning_rate": 2.3553387609016833e-05, + "loss": 0.2004, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20256412029266357, + "step": 2180, + "valid_targets_mean": 2980.2, + "valid_targets_min": 1285 + }, + { + "epoch": 3.501602564102564, + "grad_norm": 0.49018507672277717, + "learning_rate": 2.347471252339252e-05, + "loss": 0.2211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20523664355278015, + "step": 2185, + "valid_targets_mean": 3595.4, + "valid_targets_min": 1871 + }, + { + "epoch": 3.5096153846153846, + "grad_norm": 0.5651047221687497, + "learning_rate": 2.339598195578608e-05, + "loss": 0.2273, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22433197498321533, + "step": 2190, + "valid_targets_mean": 3616.5, + "valid_targets_min": 343 + }, + { + "epoch": 3.5176282051282053, + "grad_norm": 0.49048026751855617, + "learning_rate": 2.3317197163316757e-05, + "loss": 0.2301, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22643204033374786, + "step": 2195, + "valid_targets_mean": 3851.8, + "valid_targets_min": 686 + }, + { + "epoch": 3.5256410256410255, + "grad_norm": 0.48873357577242255, + "learning_rate": 2.3238359403969608e-05, + "loss": 0.2158, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21824310719966888, + "step": 2200, + "valid_targets_mean": 3577.9, + "valid_targets_min": 1968 + }, + { + "epoch": 3.5336538461538463, + "grad_norm": 0.48623991711741665, + "learning_rate": 2.315946993657543e-05, + "loss": 0.2086, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19976502656936646, + "step": 2205, + "valid_targets_mean": 3920.6, + "valid_targets_min": 1922 + }, + { + "epoch": 3.5416666666666665, + "grad_norm": 0.4839874669390242, + "learning_rate": 2.3080530020790673e-05, + "loss": 0.2162, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20096611976623535, + "step": 2210, + "valid_targets_mean": 3286.7, + "valid_targets_min": 646 + }, + { + "epoch": 3.5496794871794872, + "grad_norm": 0.4831112771937781, + "learning_rate": 2.300154091707731e-05, + "loss": 0.209, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.225946307182312, + "step": 2215, + "valid_targets_mean": 3587.2, + "valid_targets_min": 1718 + }, + { + "epoch": 3.5576923076923075, + "grad_norm": 0.5885045198895541, + "learning_rate": 2.2922503886682706e-05, + "loss": 0.2266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22732587158679962, + "step": 2220, + "valid_targets_mean": 3120.9, + "valid_targets_min": 334 + }, + { + "epoch": 3.565705128205128, + "grad_norm": 0.550148887781423, + "learning_rate": 2.28434201916195e-05, + "loss": 0.2152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22492703795433044, + "step": 2225, + "valid_targets_mean": 3465.1, + "valid_targets_min": 2393 + }, + { + "epoch": 3.573717948717949, + "grad_norm": 0.4938697286515837, + "learning_rate": 2.2764291094645446e-05, + "loss": 0.2135, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1986878365278244, + "step": 2230, + "valid_targets_mean": 3271.9, + "valid_targets_min": 290 + }, + { + "epoch": 3.581730769230769, + "grad_norm": 0.5557254876044349, + "learning_rate": 2.2685117859243223e-05, + "loss": 0.2045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2523908019065857, + "step": 2235, + "valid_targets_mean": 3248.2, + "valid_targets_min": 273 + }, + { + "epoch": 3.58974358974359, + "grad_norm": 0.6008327672276413, + "learning_rate": 2.2605901749600312e-05, + "loss": 0.2042, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25647902488708496, + "step": 2240, + "valid_targets_mean": 3110.2, + "valid_targets_min": 389 + }, + { + "epoch": 3.59775641025641, + "grad_norm": 0.532844954299154, + "learning_rate": 2.2526644030588764e-05, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2435721606016159, + "step": 2245, + "valid_targets_mean": 3189.6, + "valid_targets_min": 277 + }, + { + "epoch": 3.605769230769231, + "grad_norm": 0.545878516940055, + "learning_rate": 2.2447345967745036e-05, + "loss": 0.2107, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22559991478919983, + "step": 2250, + "valid_targets_mean": 3135.8, + "valid_targets_min": 1988 + }, + { + "epoch": 3.613782051282051, + "grad_norm": 0.4562069435188204, + "learning_rate": 2.2368008827249756e-05, + "loss": 0.2161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20217636227607727, + "step": 2255, + "valid_targets_mean": 3706.4, + "valid_targets_min": 1288 + }, + { + "epoch": 3.621794871794872, + "grad_norm": 0.458353746283704, + "learning_rate": 2.228863387590752e-05, + "loss": 0.2083, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18142034113407135, + "step": 2260, + "valid_targets_mean": 3865.8, + "valid_targets_min": 1511 + }, + { + "epoch": 3.6298076923076925, + "grad_norm": 0.5216787285174662, + "learning_rate": 2.2209222381126687e-05, + "loss": 0.2036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20579573512077332, + "step": 2265, + "valid_targets_mean": 3342.1, + "valid_targets_min": 1371 + }, + { + "epoch": 3.6378205128205128, + "grad_norm": 0.5925836434071377, + "learning_rate": 2.212977561089908e-05, + "loss": 0.2428, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2542372941970825, + "step": 2270, + "valid_targets_mean": 2954.4, + "valid_targets_min": 1127 + }, + { + "epoch": 3.6458333333333335, + "grad_norm": 0.5313423229891275, + "learning_rate": 2.20502948337798e-05, + "loss": 0.2107, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20688408613204956, + "step": 2275, + "valid_targets_mean": 3146.2, + "valid_targets_min": 1886 + }, + { + "epoch": 3.6538461538461537, + "grad_norm": 0.519268139967825, + "learning_rate": 2.1970781318866953e-05, + "loss": 0.2116, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26766157150268555, + "step": 2280, + "valid_targets_mean": 3711.6, + "valid_targets_min": 1992 + }, + { + "epoch": 3.6618589743589745, + "grad_norm": 0.5198099851729401, + "learning_rate": 2.1891236335781363e-05, + "loss": 0.2364, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24527433514595032, + "step": 2285, + "valid_targets_mean": 3303.0, + "valid_targets_min": 2143 + }, + { + "epoch": 3.6698717948717947, + "grad_norm": 0.4881045184124244, + "learning_rate": 2.1811661154646332e-05, + "loss": 0.2005, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2145967185497284, + "step": 2290, + "valid_targets_mean": 3696.6, + "valid_targets_min": 1955 + }, + { + "epoch": 3.6778846153846154, + "grad_norm": 0.5149448739418462, + "learning_rate": 2.173205704606735e-05, + "loss": 0.2164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21173402667045593, + "step": 2295, + "valid_targets_mean": 3381.4, + "valid_targets_min": 362 + }, + { + "epoch": 3.685897435897436, + "grad_norm": 0.46762585987689204, + "learning_rate": 2.1652425281111785e-05, + "loss": 0.237, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18129359185695648, + "step": 2300, + "valid_targets_mean": 3209.1, + "valid_targets_min": 499 + }, + { + "epoch": 3.6939102564102564, + "grad_norm": 0.5016573564761132, + "learning_rate": 2.1572767131288607e-05, + "loss": 0.2136, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1932133138179779, + "step": 2305, + "valid_targets_mean": 3831.2, + "valid_targets_min": 2363 + }, + { + "epoch": 3.7019230769230766, + "grad_norm": 0.553031313219947, + "learning_rate": 2.1493083868528095e-05, + "loss": 0.2453, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2638936936855316, + "step": 2310, + "valid_targets_mean": 3288.8, + "valid_targets_min": 697 + }, + { + "epoch": 3.7099358974358974, + "grad_norm": 0.5004078443376743, + "learning_rate": 2.141337676516151e-05, + "loss": 0.2332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2213425636291504, + "step": 2315, + "valid_targets_mean": 3674.1, + "valid_targets_min": 497 + }, + { + "epoch": 3.717948717948718, + "grad_norm": 0.49828281149304193, + "learning_rate": 2.1333647093900772e-05, + "loss": 0.2094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19858595728874207, + "step": 2320, + "valid_targets_mean": 3288.1, + "valid_targets_min": 1943 + }, + { + "epoch": 3.7259615384615383, + "grad_norm": 0.516956470560625, + "learning_rate": 2.1253896127818175e-05, + "loss": 0.2002, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19624871015548706, + "step": 2325, + "valid_targets_mean": 3053.8, + "valid_targets_min": 1660 + }, + { + "epoch": 3.733974358974359, + "grad_norm": 0.6330396283922977, + "learning_rate": 2.1174125140326013e-05, + "loss": 0.2176, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25099366903305054, + "step": 2330, + "valid_targets_mean": 2462.1, + "valid_targets_min": 298 + }, + { + "epoch": 3.7419871794871797, + "grad_norm": 0.4195583425046997, + "learning_rate": 2.1094335405156277e-05, + "loss": 0.2132, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1856677234172821, + "step": 2335, + "valid_targets_mean": 4358.2, + "valid_targets_min": 567 + }, + { + "epoch": 3.75, + "grad_norm": 0.5579122745611423, + "learning_rate": 2.1014528196340316e-05, + "loss": 0.2216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21907992660999298, + "step": 2340, + "valid_targets_mean": 2995.2, + "valid_targets_min": 1670 + }, + { + "epoch": 3.7580128205128203, + "grad_norm": 0.4755276678031606, + "learning_rate": 2.093470478818847e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19798678159713745, + "step": 2345, + "valid_targets_mean": 3438.9, + "valid_targets_min": 1940 + }, + { + "epoch": 3.766025641025641, + "grad_norm": 0.5095919410864391, + "learning_rate": 2.0854866455269756e-05, + "loss": 0.2187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24966293573379517, + "step": 2350, + "valid_targets_mean": 3300.4, + "valid_targets_min": 673 + }, + { + "epoch": 3.7740384615384617, + "grad_norm": 0.43308007874052606, + "learning_rate": 2.0775014472391496e-05, + "loss": 0.2094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1921013593673706, + "step": 2355, + "valid_targets_mean": 4197.9, + "valid_targets_min": 548 + }, + { + "epoch": 3.782051282051282, + "grad_norm": 0.5656612684947171, + "learning_rate": 2.0695150114578958e-05, + "loss": 0.2245, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2663126587867737, + "step": 2360, + "valid_targets_mean": 3496.9, + "valid_targets_min": 427 + }, + { + "epoch": 3.7900641025641026, + "grad_norm": 0.5350045307116155, + "learning_rate": 2.061527465705502e-05, + "loss": 0.2215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25265276432037354, + "step": 2365, + "valid_targets_mean": 3458.9, + "valid_targets_min": 274 + }, + { + "epoch": 3.7980769230769234, + "grad_norm": 0.4972661860936192, + "learning_rate": 2.0535389375219773e-05, + "loss": 0.2134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22089067101478577, + "step": 2370, + "valid_targets_mean": 3173.3, + "valid_targets_min": 366 + }, + { + "epoch": 3.8060897435897436, + "grad_norm": 0.5148532675806966, + "learning_rate": 2.045549554463019e-05, + "loss": 0.2092, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19942928850650787, + "step": 2375, + "valid_targets_mean": 3215.6, + "valid_targets_min": 1924 + }, + { + "epoch": 3.814102564102564, + "grad_norm": 0.5911016991139373, + "learning_rate": 2.0375594440979744e-05, + "loss": 0.216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21095842123031616, + "step": 2380, + "valid_targets_mean": 3474.6, + "valid_targets_min": 1719 + }, + { + "epoch": 3.8221153846153846, + "grad_norm": 0.597719336924374, + "learning_rate": 2.0295687340078037e-05, + "loss": 0.2304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2182496190071106, + "step": 2385, + "valid_targets_mean": 2813.0, + "valid_targets_min": 382 + }, + { + "epoch": 3.8301282051282053, + "grad_norm": 0.530072710758143, + "learning_rate": 2.0215775517830437e-05, + "loss": 0.2067, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22930724918842316, + "step": 2390, + "valid_targets_mean": 3318.5, + "valid_targets_min": 488 + }, + { + "epoch": 3.8381410256410255, + "grad_norm": 0.5154279772668271, + "learning_rate": 2.013586025021769e-05, + "loss": 0.245, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26545244455337524, + "step": 2395, + "valid_targets_mean": 3903.9, + "valid_targets_min": 1511 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.5746132815293595, + "learning_rate": 2.0055942813275564e-05, + "loss": 0.2292, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23045890033245087, + "step": 2400, + "valid_targets_mean": 2854.4, + "valid_targets_min": 465 + }, + { + "epoch": 3.8541666666666665, + "grad_norm": 0.6238290798209558, + "learning_rate": 1.9976024483074456e-05, + "loss": 0.2065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23214343190193176, + "step": 2405, + "valid_targets_mean": 2401.7, + "valid_targets_min": 245 + }, + { + "epoch": 3.8621794871794872, + "grad_norm": 0.5664988549700263, + "learning_rate": 1.9896106535699025e-05, + "loss": 0.2114, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2216949760913849, + "step": 2410, + "valid_targets_mean": 2997.2, + "valid_targets_min": 1476 + }, + { + "epoch": 3.8701923076923075, + "grad_norm": 0.6198688515244216, + "learning_rate": 1.9816190247227834e-05, + "loss": 0.2014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24409687519073486, + "step": 2415, + "valid_targets_mean": 2904.2, + "valid_targets_min": 395 + }, + { + "epoch": 3.878205128205128, + "grad_norm": 0.5277766343405376, + "learning_rate": 1.9736276893712954e-05, + "loss": 0.2264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2225913405418396, + "step": 2420, + "valid_targets_mean": 3026.1, + "valid_targets_min": 601 + }, + { + "epoch": 3.886217948717949, + "grad_norm": 0.49932532353423154, + "learning_rate": 1.9656367751159565e-05, + "loss": 0.2242, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21792566776275635, + "step": 2425, + "valid_targets_mean": 3857.0, + "valid_targets_min": 2043 + }, + { + "epoch": 3.894230769230769, + "grad_norm": 0.5057026335085257, + "learning_rate": 1.957646409550565e-05, + "loss": 0.2307, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26214978098869324, + "step": 2430, + "valid_targets_mean": 3946.2, + "valid_targets_min": 2392 + }, + { + "epoch": 3.90224358974359, + "grad_norm": 0.4619341239060796, + "learning_rate": 1.9496567202601545e-05, + "loss": 0.216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20152977108955383, + "step": 2435, + "valid_targets_mean": 3997.2, + "valid_targets_min": 2271 + }, + { + "epoch": 3.91025641025641, + "grad_norm": 0.48590936854836037, + "learning_rate": 1.9416678348189627e-05, + "loss": 0.2118, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1917511522769928, + "step": 2440, + "valid_targets_mean": 3163.1, + "valid_targets_min": 734 + }, + { + "epoch": 3.918269230769231, + "grad_norm": 0.5804332092212046, + "learning_rate": 1.9336798807883907e-05, + "loss": 0.2137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23004987835884094, + "step": 2445, + "valid_targets_mean": 3031.5, + "valid_targets_min": 302 + }, + { + "epoch": 3.926282051282051, + "grad_norm": 0.5355031811260637, + "learning_rate": 1.9256929857149686e-05, + "loss": 0.2258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2335416078567505, + "step": 2450, + "valid_targets_mean": 3444.2, + "valid_targets_min": 1546 + }, + { + "epoch": 3.934294871794872, + "grad_norm": 0.4376032631537377, + "learning_rate": 1.9177072771283167e-05, + "loss": 0.2126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19955921173095703, + "step": 2455, + "valid_targets_mean": 4088.3, + "valid_targets_min": 1712 + }, + { + "epoch": 3.9423076923076925, + "grad_norm": 0.4972198496907315, + "learning_rate": 1.9097228825391087e-05, + "loss": 0.2098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20892062783241272, + "step": 2460, + "valid_targets_mean": 3496.2, + "valid_targets_min": 1704 + }, + { + "epoch": 3.9503205128205128, + "grad_norm": 0.5119397009023006, + "learning_rate": 1.9017399294370413e-05, + "loss": 0.2117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20475465059280396, + "step": 2465, + "valid_targets_mean": 3078.6, + "valid_targets_min": 1843 + }, + { + "epoch": 3.9583333333333335, + "grad_norm": 0.46958127089075363, + "learning_rate": 1.893758545288791e-05, + "loss": 0.2208, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23267512023448944, + "step": 2470, + "valid_targets_mean": 3967.6, + "valid_targets_min": 346 + }, + { + "epoch": 3.9663461538461537, + "grad_norm": 0.4422338855672534, + "learning_rate": 1.8857788575359847e-05, + "loss": 0.1968, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18764463067054749, + "step": 2475, + "valid_targets_mean": 3658.1, + "valid_targets_min": 282 + }, + { + "epoch": 3.9743589743589745, + "grad_norm": 0.45411203405757655, + "learning_rate": 1.87780099359316e-05, + "loss": 0.1988, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16824612021446228, + "step": 2480, + "valid_targets_mean": 3436.4, + "valid_targets_min": 290 + }, + { + "epoch": 3.9823717948717947, + "grad_norm": 0.5975349252417311, + "learning_rate": 1.869825080845734e-05, + "loss": 0.2534, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27063214778900146, + "step": 2485, + "valid_targets_mean": 2898.2, + "valid_targets_min": 1411 + }, + { + "epoch": 3.9903846153846154, + "grad_norm": 0.42022271538873057, + "learning_rate": 1.8618512466479686e-05, + "loss": 0.1997, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16057726740837097, + "step": 2490, + "valid_targets_mean": 4243.4, + "valid_targets_min": 1836 + }, + { + "epoch": 3.998397435897436, + "grad_norm": 0.5012809468885003, + "learning_rate": 1.8538796183209373e-05, + "loss": 0.2118, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2236367017030716, + "step": 2495, + "valid_targets_mean": 3480.9, + "valid_targets_min": 1944 + }, + { + "epoch": 4.006410256410256, + "grad_norm": 0.4195723414242144, + "learning_rate": 1.845910323150491e-05, + "loss": 0.1996, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16902586817741394, + "step": 2500, + "valid_targets_mean": 4047.4, + "valid_targets_min": 2083 + }, + { + "epoch": 4.014423076923077, + "grad_norm": 0.5145131919278905, + "learning_rate": 1.8379434883852255e-05, + "loss": 0.2, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19503673911094666, + "step": 2505, + "valid_targets_mean": 3389.7, + "valid_targets_min": 427 + }, + { + "epoch": 4.022435897435898, + "grad_norm": 0.638446760469446, + "learning_rate": 1.8299792412344524e-05, + "loss": 0.2076, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2457159012556076, + "step": 2510, + "valid_targets_mean": 3125.1, + "valid_targets_min": 1945 + }, + { + "epoch": 4.030448717948718, + "grad_norm": 0.5320510562195919, + "learning_rate": 1.8220177088661635e-05, + "loss": 0.2172, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28126007318496704, + "step": 2515, + "valid_targets_mean": 4158.2, + "valid_targets_min": 272 + }, + { + "epoch": 4.038461538461538, + "grad_norm": 0.47272452528662334, + "learning_rate": 1.814059018405004e-05, + "loss": 0.2106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2101389467716217, + "step": 2520, + "valid_targets_mean": 4075.9, + "valid_targets_min": 1696 + }, + { + "epoch": 4.046474358974359, + "grad_norm": 0.45352211553110705, + "learning_rate": 1.806103296930243e-05, + "loss": 0.196, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17897215485572815, + "step": 2525, + "valid_targets_mean": 3710.6, + "valid_targets_min": 1507 + }, + { + "epoch": 4.05448717948718, + "grad_norm": 0.5064275499739576, + "learning_rate": 1.7981506714737392e-05, + "loss": 0.191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20361468195915222, + "step": 2530, + "valid_targets_mean": 3783.2, + "valid_targets_min": 314 + }, + { + "epoch": 4.0625, + "grad_norm": 0.49068392176466213, + "learning_rate": 1.7902012690179188e-05, + "loss": 0.2089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23372966051101685, + "step": 2535, + "valid_targets_mean": 4053.6, + "valid_targets_min": 2273 + }, + { + "epoch": 4.07051282051282, + "grad_norm": 0.3999280296624379, + "learning_rate": 1.7822552164937437e-05, + "loss": 0.1898, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15756747126579285, + "step": 2540, + "valid_targets_mean": 4231.2, + "valid_targets_min": 2032 + }, + { + "epoch": 4.078525641025641, + "grad_norm": 0.6411212991277369, + "learning_rate": 1.7743126407786873e-05, + "loss": 0.2018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18768355250358582, + "step": 2545, + "valid_targets_mean": 3995.1, + "valid_targets_min": 362 + }, + { + "epoch": 4.086538461538462, + "grad_norm": 0.632615400224224, + "learning_rate": 1.766373668694707e-05, + "loss": 0.2203, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23910491168498993, + "step": 2550, + "valid_targets_mean": 2928.5, + "valid_targets_min": 686 + }, + { + "epoch": 4.094551282051282, + "grad_norm": 0.5054791393998274, + "learning_rate": 1.7584384270062195e-05, + "loss": 0.2221, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20476405322551727, + "step": 2555, + "valid_targets_mean": 3803.2, + "valid_targets_min": 2087 + }, + { + "epoch": 4.102564102564102, + "grad_norm": 0.6073867684870838, + "learning_rate": 1.7505070424180772e-05, + "loss": 0.2018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23343729972839355, + "step": 2560, + "valid_targets_mean": 3272.6, + "valid_targets_min": 1870 + }, + { + "epoch": 4.110576923076923, + "grad_norm": 0.48654884113244695, + "learning_rate": 1.7425796415735454e-05, + "loss": 0.1941, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1814398467540741, + "step": 2565, + "valid_targets_mean": 3609.6, + "valid_targets_min": 1476 + }, + { + "epoch": 4.118589743589744, + "grad_norm": 0.5334852365572529, + "learning_rate": 1.7346563510522783e-05, + "loss": 0.1964, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18280336260795593, + "step": 2570, + "valid_targets_mean": 3150.2, + "valid_targets_min": 455 + }, + { + "epoch": 4.126602564102564, + "grad_norm": 0.49068187641004996, + "learning_rate": 1.7267372973682998e-05, + "loss": 0.1989, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18673810362815857, + "step": 2575, + "valid_targets_mean": 3891.6, + "valid_targets_min": 1943 + }, + { + "epoch": 4.134615384615385, + "grad_norm": 0.5199937773523386, + "learning_rate": 1.7188226069679834e-05, + "loss": 0.2129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19363847374916077, + "step": 2580, + "valid_targets_mean": 3347.2, + "valid_targets_min": 478 + }, + { + "epoch": 4.142628205128205, + "grad_norm": 0.5175812136446701, + "learning_rate": 1.7109124062280307e-05, + "loss": 0.2145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20378626883029938, + "step": 2585, + "valid_targets_mean": 3837.9, + "valid_targets_min": 2305 + }, + { + "epoch": 4.1506410256410255, + "grad_norm": 0.6269435116455723, + "learning_rate": 1.7030068214534567e-05, + "loss": 0.1875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1834736168384552, + "step": 2590, + "valid_targets_mean": 3715.1, + "valid_targets_min": 325 + }, + { + "epoch": 4.158653846153846, + "grad_norm": 0.5589538239959635, + "learning_rate": 1.695105978875572e-05, + "loss": 0.1915, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21350257098674774, + "step": 2595, + "valid_targets_mean": 2966.5, + "valid_targets_min": 1723 + }, + { + "epoch": 4.166666666666667, + "grad_norm": 0.503812150208315, + "learning_rate": 1.687210004649965e-05, + "loss": 0.2164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17916864156723022, + "step": 2600, + "valid_targets_mean": 3349.9, + "valid_targets_min": 1798 + }, + { + "epoch": 4.174679487179487, + "grad_norm": 0.45804240872640667, + "learning_rate": 1.679319024854491e-05, + "loss": 0.1885, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17588487267494202, + "step": 2605, + "valid_targets_mean": 3999.6, + "valid_targets_min": 448 + }, + { + "epoch": 4.1826923076923075, + "grad_norm": 0.5806614439136704, + "learning_rate": 1.6714331654872564e-05, + "loss": 0.1964, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2032093107700348, + "step": 2610, + "valid_targets_mean": 3521.7, + "valid_targets_min": 1517 + }, + { + "epoch": 4.190705128205128, + "grad_norm": 0.51365300180505, + "learning_rate": 1.663552552464609e-05, + "loss": 0.1892, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22034162282943726, + "step": 2615, + "valid_targets_mean": 3520.1, + "valid_targets_min": 1472 + }, + { + "epoch": 4.198717948717949, + "grad_norm": 0.46595977542803346, + "learning_rate": 1.6556773116191257e-05, + "loss": 0.1897, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16177070140838623, + "step": 2620, + "valid_targets_mean": 3612.4, + "valid_targets_min": 377 + }, + { + "epoch": 4.206730769230769, + "grad_norm": 0.5934619647470252, + "learning_rate": 1.647807568697603e-05, + "loss": 0.1863, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1621251404285431, + "step": 2625, + "valid_targets_mean": 3973.6, + "valid_targets_min": 1866 + }, + { + "epoch": 4.214743589743589, + "grad_norm": 0.5279494545170368, + "learning_rate": 1.6399434493590524e-05, + "loss": 0.193, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20115870237350464, + "step": 2630, + "valid_targets_mean": 3740.2, + "valid_targets_min": 2234 + }, + { + "epoch": 4.222756410256411, + "grad_norm": 0.5402676551960568, + "learning_rate": 1.6320850791726884e-05, + "loss": 0.1956, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18363741040229797, + "step": 2635, + "valid_targets_mean": 3030.2, + "valid_targets_min": 1729 + }, + { + "epoch": 4.230769230769231, + "grad_norm": 0.5309143923757463, + "learning_rate": 1.6242325836159304e-05, + "loss": 0.195, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20168320834636688, + "step": 2640, + "valid_targets_mean": 3316.9, + "valid_targets_min": 299 + }, + { + "epoch": 4.238782051282051, + "grad_norm": 0.5108105368110876, + "learning_rate": 1.6163860880723923e-05, + "loss": 0.1961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1702432930469513, + "step": 2645, + "valid_targets_mean": 3533.9, + "valid_targets_min": 2080 + }, + { + "epoch": 4.246794871794872, + "grad_norm": 0.519843760147622, + "learning_rate": 1.6085457178298866e-05, + "loss": 0.1972, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18240828812122345, + "step": 2650, + "valid_targets_mean": 3165.8, + "valid_targets_min": 1670 + }, + { + "epoch": 4.2548076923076925, + "grad_norm": 0.5077601861968751, + "learning_rate": 1.6007115980784182e-05, + "loss": 0.2098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21437841653823853, + "step": 2655, + "valid_targets_mean": 3410.4, + "valid_targets_min": 1535 + }, + { + "epoch": 4.262820512820513, + "grad_norm": 0.43324095570540344, + "learning_rate": 1.592883853908188e-05, + "loss": 0.1961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16936254501342773, + "step": 2660, + "valid_targets_mean": 4128.1, + "valid_targets_min": 2076 + }, + { + "epoch": 4.270833333333333, + "grad_norm": 0.5434487721223978, + "learning_rate": 1.585062610307599e-05, + "loss": 0.2005, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18581029772758484, + "step": 2665, + "valid_targets_mean": 3207.5, + "valid_targets_min": 318 + }, + { + "epoch": 4.278846153846154, + "grad_norm": 0.5185309476056992, + "learning_rate": 1.5772479921612543e-05, + "loss": 0.226, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19693030416965485, + "step": 2670, + "valid_targets_mean": 3433.8, + "valid_targets_min": 1994 + }, + { + "epoch": 4.2868589743589745, + "grad_norm": 0.49328604974342904, + "learning_rate": 1.5694401242479677e-05, + "loss": 0.1951, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20844605565071106, + "step": 2675, + "valid_targets_mean": 3896.2, + "valid_targets_min": 244 + }, + { + "epoch": 4.294871794871795, + "grad_norm": 0.5315762410373522, + "learning_rate": 1.5616391312387683e-05, + "loss": 0.2035, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19662663340568542, + "step": 2680, + "valid_targets_mean": 3516.6, + "valid_targets_min": 466 + }, + { + "epoch": 4.302884615384615, + "grad_norm": 0.5634714984529963, + "learning_rate": 1.5538451376949106e-05, + "loss": 0.2133, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21461111307144165, + "step": 2685, + "valid_targets_mean": 3134.6, + "valid_targets_min": 543 + }, + { + "epoch": 4.310897435897436, + "grad_norm": 0.5332640250660052, + "learning_rate": 1.5460582680658888e-05, + "loss": 0.194, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2080698311328888, + "step": 2690, + "valid_targets_mean": 3422.9, + "valid_targets_min": 1331 + }, + { + "epoch": 4.318910256410256, + "grad_norm": 0.5187079147122462, + "learning_rate": 1.5382786466874446e-05, + "loss": 0.1822, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17928150296211243, + "step": 2695, + "valid_targets_mean": 2967.9, + "valid_targets_min": 455 + }, + { + "epoch": 4.326923076923077, + "grad_norm": 0.5327115710533992, + "learning_rate": 1.5305063977795856e-05, + "loss": 0.2029, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20295903086662292, + "step": 2700, + "valid_targets_mean": 3376.3, + "valid_targets_min": 1958 + }, + { + "epoch": 4.334935897435898, + "grad_norm": 0.4895315499876015, + "learning_rate": 1.5227416454445995e-05, + "loss": 0.1834, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18504248559474945, + "step": 2705, + "valid_targets_mean": 3932.8, + "valid_targets_min": 2011 + }, + { + "epoch": 4.342948717948718, + "grad_norm": 0.5046175280672212, + "learning_rate": 1.5149845136650748e-05, + "loss": 0.2046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1886163204908371, + "step": 2710, + "valid_targets_mean": 3494.1, + "valid_targets_min": 1993 + }, + { + "epoch": 4.350961538461538, + "grad_norm": 0.4259723146821667, + "learning_rate": 1.5072351263019177e-05, + "loss": 0.1843, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15069478750228882, + "step": 2715, + "valid_targets_mean": 3969.4, + "valid_targets_min": 356 + }, + { + "epoch": 4.358974358974359, + "grad_norm": 0.5165494690418552, + "learning_rate": 1.4994936070923784e-05, + "loss": 0.2042, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21728010475635529, + "step": 2720, + "valid_targets_mean": 3352.1, + "valid_targets_min": 1653 + }, + { + "epoch": 4.36698717948718, + "grad_norm": 0.5150257406953364, + "learning_rate": 1.4917600796480745e-05, + "loss": 0.1942, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16228248178958893, + "step": 2725, + "valid_targets_mean": 2981.2, + "valid_targets_min": 230 + }, + { + "epoch": 4.375, + "grad_norm": 0.6145951677631153, + "learning_rate": 1.4840346674530122e-05, + "loss": 0.2032, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21405372023582458, + "step": 2730, + "valid_targets_mean": 2825.2, + "valid_targets_min": 332 + }, + { + "epoch": 4.38301282051282, + "grad_norm": 0.5918044868667718, + "learning_rate": 1.4763174938616232e-05, + "loss": 0.2194, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22870035469532013, + "step": 2735, + "valid_targets_mean": 2888.4, + "valid_targets_min": 1671 + }, + { + "epoch": 4.391025641025641, + "grad_norm": 0.5192510906056325, + "learning_rate": 1.4686086820967865e-05, + "loss": 0.1986, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20790967345237732, + "step": 2740, + "valid_targets_mean": 3779.0, + "valid_targets_min": 1993 + }, + { + "epoch": 4.399038461538462, + "grad_norm": 0.6197195127482825, + "learning_rate": 1.460908355247868e-05, + "loss": 0.2031, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23618865013122559, + "step": 2745, + "valid_targets_mean": 2836.0, + "valid_targets_min": 248 + }, + { + "epoch": 4.407051282051282, + "grad_norm": 0.4760638930188835, + "learning_rate": 1.4532166362687507e-05, + "loss": 0.186, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1725659966468811, + "step": 2750, + "valid_targets_mean": 3864.0, + "valid_targets_min": 1833 + }, + { + "epoch": 4.415064102564102, + "grad_norm": 0.511002075402515, + "learning_rate": 1.445533647975871e-05, + "loss": 0.203, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19956639409065247, + "step": 2755, + "valid_targets_mean": 3626.1, + "valid_targets_min": 1713 + }, + { + "epoch": 4.423076923076923, + "grad_norm": 0.5438811518715806, + "learning_rate": 1.437859513046263e-05, + "loss": 0.2, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23376289010047913, + "step": 2760, + "valid_targets_mean": 3961.9, + "valid_targets_min": 646 + }, + { + "epoch": 4.431089743589744, + "grad_norm": 0.4898545257870175, + "learning_rate": 1.4301943540155914e-05, + "loss": 0.2097, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17090612649917603, + "step": 2765, + "valid_targets_mean": 3729.2, + "valid_targets_min": 1751 + }, + { + "epoch": 4.439102564102564, + "grad_norm": 0.48058927205208335, + "learning_rate": 1.4225382932762033e-05, + "loss": 0.2201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15635713934898376, + "step": 2770, + "valid_targets_mean": 3476.9, + "valid_targets_min": 315 + }, + { + "epoch": 4.447115384615385, + "grad_norm": 0.5442851597081118, + "learning_rate": 1.4148914530751681e-05, + "loss": 0.2124, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19521600008010864, + "step": 2775, + "valid_targets_mean": 3143.2, + "valid_targets_min": 532 + }, + { + "epoch": 4.455128205128205, + "grad_norm": 0.5770583514083514, + "learning_rate": 1.4072539555123292e-05, + "loss": 0.2016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20591652393341064, + "step": 2780, + "valid_targets_mean": 3650.2, + "valid_targets_min": 258 + }, + { + "epoch": 4.4631410256410255, + "grad_norm": 0.7251203058744868, + "learning_rate": 1.3996259225383514e-05, + "loss": 0.1929, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1771497130393982, + "step": 2785, + "valid_targets_mean": 3405.9, + "valid_targets_min": 1786 + }, + { + "epoch": 4.471153846153846, + "grad_norm": 0.6076887775424848, + "learning_rate": 1.3920074759527737e-05, + "loss": 0.2008, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22005143761634827, + "step": 2790, + "valid_targets_mean": 3131.5, + "valid_targets_min": 401 + }, + { + "epoch": 4.479166666666667, + "grad_norm": 0.524736934106875, + "learning_rate": 1.3843987374020689e-05, + "loss": 0.2026, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2200172245502472, + "step": 2795, + "valid_targets_mean": 3403.7, + "valid_targets_min": 300 + }, + { + "epoch": 4.487179487179487, + "grad_norm": 0.5476458833911162, + "learning_rate": 1.376799828377696e-05, + "loss": 0.1895, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19838351011276245, + "step": 2800, + "valid_targets_mean": 2975.4, + "valid_targets_min": 306 + }, + { + "epoch": 4.4951923076923075, + "grad_norm": 0.49700654071688327, + "learning_rate": 1.3692108702141642e-05, + "loss": 0.1991, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21074606478214264, + "step": 2805, + "valid_targets_mean": 3779.0, + "valid_targets_min": 1566 + }, + { + "epoch": 4.503205128205128, + "grad_norm": 0.5685422355069307, + "learning_rate": 1.361631984087091e-05, + "loss": 0.2147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2041737586259842, + "step": 2810, + "valid_targets_mean": 2877.1, + "valid_targets_min": 245 + }, + { + "epoch": 4.511217948717949, + "grad_norm": 0.5542806088102038, + "learning_rate": 1.354063291011273e-05, + "loss": 0.202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19276300072669983, + "step": 2815, + "valid_targets_mean": 3169.6, + "valid_targets_min": 418 + }, + { + "epoch": 4.519230769230769, + "grad_norm": 0.5355827919010402, + "learning_rate": 1.3465049118387486e-05, + "loss": 0.1894, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19092710316181183, + "step": 2820, + "valid_targets_mean": 3060.0, + "valid_targets_min": 377 + }, + { + "epoch": 4.527243589743589, + "grad_norm": 0.5027999913792417, + "learning_rate": 1.3389569672568707e-05, + "loss": 0.2027, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19048264622688293, + "step": 2825, + "valid_targets_mean": 3426.3, + "valid_targets_min": 2489 + }, + { + "epoch": 4.535256410256411, + "grad_norm": 0.5431332317741009, + "learning_rate": 1.331419577786381e-05, + "loss": 0.2054, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22947734594345093, + "step": 2830, + "valid_targets_mean": 3533.0, + "valid_targets_min": 351 + }, + { + "epoch": 4.543269230769231, + "grad_norm": 0.49143858468037405, + "learning_rate": 1.3238928637794816e-05, + "loss": 0.1912, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1725982129573822, + "step": 2835, + "valid_targets_mean": 3654.7, + "valid_targets_min": 2004 + }, + { + "epoch": 4.551282051282051, + "grad_norm": 0.5466219539220681, + "learning_rate": 1.3163769454179183e-05, + "loss": 0.1993, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22573739290237427, + "step": 2840, + "valid_targets_mean": 3480.4, + "valid_targets_min": 352 + }, + { + "epoch": 4.559294871794872, + "grad_norm": 0.4499086516520062, + "learning_rate": 1.3088719427110552e-05, + "loss": 0.188, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15970298647880554, + "step": 2845, + "valid_targets_mean": 3769.4, + "valid_targets_min": 2111 + }, + { + "epoch": 4.5673076923076925, + "grad_norm": 0.5194647525232524, + "learning_rate": 1.3013779754939666e-05, + "loss": 0.2146, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17854247987270355, + "step": 2850, + "valid_targets_mean": 3513.4, + "valid_targets_min": 2089 + }, + { + "epoch": 4.575320512820513, + "grad_norm": 0.46776306719156063, + "learning_rate": 1.2938951634255164e-05, + "loss": 0.1957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18298465013504028, + "step": 2855, + "valid_targets_mean": 3878.9, + "valid_targets_min": 1974 + }, + { + "epoch": 4.583333333333333, + "grad_norm": 0.5576953176753099, + "learning_rate": 1.2864236259864495e-05, + "loss": 0.1974, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19287380576133728, + "step": 2860, + "valid_targets_mean": 2957.9, + "valid_targets_min": 352 + }, + { + "epoch": 4.591346153846154, + "grad_norm": 0.4951258213397684, + "learning_rate": 1.2789634824774887e-05, + "loss": 0.2147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17422738671302795, + "step": 2865, + "valid_targets_mean": 3260.9, + "valid_targets_min": 2020 + }, + { + "epoch": 4.5993589743589745, + "grad_norm": 0.46464947946698526, + "learning_rate": 1.2715148520174206e-05, + "loss": 0.1847, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15368060767650604, + "step": 2870, + "valid_targets_mean": 3799.4, + "valid_targets_min": 1561 + }, + { + "epoch": 4.607371794871795, + "grad_norm": 0.4061095560086419, + "learning_rate": 1.2640778535412036e-05, + "loss": 0.1838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1623319685459137, + "step": 2875, + "valid_targets_mean": 4657.2, + "valid_targets_min": 1943 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.4838245534820288, + "learning_rate": 1.2566526057980608e-05, + "loss": 0.1865, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18071702122688293, + "step": 2880, + "valid_targets_mean": 3507.5, + "valid_targets_min": 2029 + }, + { + "epoch": 4.623397435897436, + "grad_norm": 0.5243111054303644, + "learning_rate": 1.2492392273495879e-05, + "loss": 0.2015, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18476027250289917, + "step": 2885, + "valid_targets_mean": 3678.5, + "valid_targets_min": 330 + }, + { + "epoch": 4.631410256410256, + "grad_norm": 0.553281734097847, + "learning_rate": 1.2418378365678612e-05, + "loss": 0.1966, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20735764503479004, + "step": 2890, + "valid_targets_mean": 3342.6, + "valid_targets_min": 1528 + }, + { + "epoch": 4.639423076923077, + "grad_norm": 0.41114325272563546, + "learning_rate": 1.234448551633542e-05, + "loss": 0.2043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18882155418395996, + "step": 2895, + "valid_targets_mean": 4904.7, + "valid_targets_min": 1084 + }, + { + "epoch": 4.647435897435898, + "grad_norm": 0.5606770649530616, + "learning_rate": 1.2270714905339969e-05, + "loss": 0.2084, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21375831961631775, + "step": 2900, + "valid_targets_mean": 3524.6, + "valid_targets_min": 1736 + }, + { + "epoch": 4.655448717948718, + "grad_norm": 0.5162069270799858, + "learning_rate": 1.2197067710614075e-05, + "loss": 0.1927, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18576446175575256, + "step": 2905, + "valid_targets_mean": 3381.0, + "valid_targets_min": 1666 + }, + { + "epoch": 4.663461538461538, + "grad_norm": 0.5597459395036315, + "learning_rate": 1.2123545108108943e-05, + "loss": 0.2117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22167593240737915, + "step": 2910, + "valid_targets_mean": 3257.6, + "valid_targets_min": 391 + }, + { + "epoch": 4.671474358974359, + "grad_norm": 0.4781708796025318, + "learning_rate": 1.2050148271786348e-05, + "loss": 0.2101, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17129084467887878, + "step": 2915, + "valid_targets_mean": 3561.2, + "valid_targets_min": 444 + }, + { + "epoch": 4.67948717948718, + "grad_norm": 0.4997511060015055, + "learning_rate": 1.1976878373599928e-05, + "loss": 0.1982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16873495280742645, + "step": 2920, + "valid_targets_mean": 3391.8, + "valid_targets_min": 291 + }, + { + "epoch": 4.6875, + "grad_norm": 0.4902845974414694, + "learning_rate": 1.1903736583476441e-05, + "loss": 0.1979, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2030390202999115, + "step": 2925, + "valid_targets_mean": 3732.4, + "valid_targets_min": 2031 + }, + { + "epoch": 4.69551282051282, + "grad_norm": 0.5193517614088814, + "learning_rate": 1.1830724069297106e-05, + "loss": 0.205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19543379545211792, + "step": 2930, + "valid_targets_mean": 3282.8, + "valid_targets_min": 353 + }, + { + "epoch": 4.703525641025641, + "grad_norm": 0.5704804701796626, + "learning_rate": 1.1757841996878957e-05, + "loss": 0.1936, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19900594651699066, + "step": 2935, + "valid_targets_mean": 3075.9, + "valid_targets_min": 418 + }, + { + "epoch": 4.711538461538462, + "grad_norm": 0.5826653409132639, + "learning_rate": 1.1685091529956187e-05, + "loss": 0.2073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2115277647972107, + "step": 2940, + "valid_targets_mean": 3320.6, + "valid_targets_min": 1780 + }, + { + "epoch": 4.719551282051282, + "grad_norm": 0.5021845306839792, + "learning_rate": 1.161247383016163e-05, + "loss": 0.2036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20887619256973267, + "step": 2945, + "valid_targets_mean": 3432.6, + "valid_targets_min": 688 + }, + { + "epoch": 4.727564102564102, + "grad_norm": 0.6114150346499213, + "learning_rate": 1.1539990057008166e-05, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18386265635490417, + "step": 2950, + "valid_targets_mean": 3268.9, + "valid_targets_min": 590 + }, + { + "epoch": 4.735576923076923, + "grad_norm": 0.5512196387593564, + "learning_rate": 1.1467641367870198e-05, + "loss": 0.1945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22034752368927002, + "step": 2955, + "valid_targets_mean": 3414.5, + "valid_targets_min": 1828 + }, + { + "epoch": 4.743589743589744, + "grad_norm": 0.5647291311418439, + "learning_rate": 1.1395428917965239e-05, + "loss": 0.1825, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19606733322143555, + "step": 2960, + "valid_targets_mean": 3166.5, + "valid_targets_min": 1982 + }, + { + "epoch": 4.751602564102564, + "grad_norm": 0.6519634295021054, + "learning_rate": 1.1323353860335385e-05, + "loss": 0.2091, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2246025651693344, + "step": 2965, + "valid_targets_mean": 2595.8, + "valid_targets_min": 282 + }, + { + "epoch": 4.759615384615385, + "grad_norm": 0.4555433892994592, + "learning_rate": 1.1251417345828962e-05, + "loss": 0.2137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18713593482971191, + "step": 2970, + "valid_targets_mean": 3909.8, + "valid_targets_min": 607 + }, + { + "epoch": 4.767628205128205, + "grad_norm": 0.5112213787584956, + "learning_rate": 1.1179620523082107e-05, + "loss": 0.192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18104520440101624, + "step": 2975, + "valid_targets_mean": 3262.2, + "valid_targets_min": 274 + }, + { + "epoch": 4.7756410256410255, + "grad_norm": 0.5695788763309223, + "learning_rate": 1.110796453850047e-05, + "loss": 0.2126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21264362335205078, + "step": 2980, + "valid_targets_mean": 3146.4, + "valid_targets_min": 1505 + }, + { + "epoch": 4.783653846153846, + "grad_norm": 0.4658028855567424, + "learning_rate": 1.1036450536240877e-05, + "loss": 0.1921, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16871154308319092, + "step": 2985, + "valid_targets_mean": 4409.3, + "valid_targets_min": 2097 + }, + { + "epoch": 4.791666666666667, + "grad_norm": 0.4841046230771778, + "learning_rate": 1.0965079658193068e-05, + "loss": 0.2033, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2128428816795349, + "step": 2990, + "valid_targets_mean": 3797.4, + "valid_targets_min": 1539 + }, + { + "epoch": 4.799679487179487, + "grad_norm": 0.5488706583686694, + "learning_rate": 1.0893853043961475e-05, + "loss": 0.2028, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20432452857494354, + "step": 2995, + "valid_targets_mean": 3163.5, + "valid_targets_min": 1705 + }, + { + "epoch": 4.8076923076923075, + "grad_norm": 0.3993765560711606, + "learning_rate": 1.0822771830847011e-05, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18088233470916748, + "step": 3000, + "valid_targets_mean": 5041.3, + "valid_targets_min": 837 + }, + { + "epoch": 4.815705128205128, + "grad_norm": 0.5478814229443099, + "learning_rate": 1.0751837153828926e-05, + "loss": 0.203, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19549459218978882, + "step": 3005, + "valid_targets_mean": 3135.2, + "valid_targets_min": 287 + }, + { + "epoch": 4.823717948717949, + "grad_norm": 0.5340723906442628, + "learning_rate": 1.0681050145546666e-05, + "loss": 0.2036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22188079357147217, + "step": 3010, + "valid_targets_mean": 3478.6, + "valid_targets_min": 831 + }, + { + "epoch": 4.831730769230769, + "grad_norm": 0.5305728395941428, + "learning_rate": 1.0610411936281801e-05, + "loss": 0.2003, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21828171610832214, + "step": 3015, + "valid_targets_mean": 3555.6, + "valid_targets_min": 368 + }, + { + "epoch": 4.839743589743589, + "grad_norm": 0.5786196807883349, + "learning_rate": 1.0539923653939978e-05, + "loss": 0.2149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20064550638198853, + "step": 3020, + "valid_targets_mean": 2935.4, + "valid_targets_min": 590 + }, + { + "epoch": 4.847756410256411, + "grad_norm": 0.6004313342633323, + "learning_rate": 1.0469586424032903e-05, + "loss": 0.1965, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16980455815792084, + "step": 3025, + "valid_targets_mean": 2810.4, + "valid_targets_min": 309 + }, + { + "epoch": 4.855769230769231, + "grad_norm": 0.6131691098928032, + "learning_rate": 1.0399401369660369e-05, + "loss": 0.1898, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19341379404067993, + "step": 3030, + "valid_targets_mean": 2728.5, + "valid_targets_min": 302 + }, + { + "epoch": 4.863782051282051, + "grad_norm": 0.5544720364395579, + "learning_rate": 1.0329369611492334e-05, + "loss": 0.2055, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2462378740310669, + "step": 3035, + "valid_targets_mean": 3388.8, + "valid_targets_min": 334 + }, + { + "epoch": 4.871794871794872, + "grad_norm": 0.5733544279629245, + "learning_rate": 1.0259492267751022e-05, + "loss": 0.2091, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20790205895900726, + "step": 3040, + "valid_targets_mean": 2837.6, + "valid_targets_min": 279 + }, + { + "epoch": 4.8798076923076925, + "grad_norm": 0.531980939673929, + "learning_rate": 1.0189770454193052e-05, + "loss": 0.1957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2089274674654007, + "step": 3045, + "valid_targets_mean": 3406.5, + "valid_targets_min": 1848 + }, + { + "epoch": 4.887820512820513, + "grad_norm": 0.5854718977302726, + "learning_rate": 1.0120205284091673e-05, + "loss": 0.2063, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23034754395484924, + "step": 3050, + "valid_targets_mean": 2694.4, + "valid_targets_min": 290 + }, + { + "epoch": 4.895833333333333, + "grad_norm": 0.5806071455397431, + "learning_rate": 1.0050797868218907e-05, + "loss": 0.1945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18953976035118103, + "step": 3055, + "valid_targets_mean": 2736.9, + "valid_targets_min": 1804 + }, + { + "epoch": 4.903846153846154, + "grad_norm": 0.4758714435534047, + "learning_rate": 9.981549314827876e-06, + "loss": 0.1938, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17658931016921997, + "step": 3060, + "valid_targets_mean": 3758.4, + "valid_targets_min": 548 + }, + { + "epoch": 4.9118589743589745, + "grad_norm": 0.5151085606255328, + "learning_rate": 9.912460729635097e-06, + "loss": 0.2001, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1856827735900879, + "step": 3065, + "valid_targets_mean": 3886.4, + "valid_targets_min": 2076 + }, + { + "epoch": 4.919871794871795, + "grad_norm": 0.5600078970119667, + "learning_rate": 9.843533215802796e-06, + "loss": 0.199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22082971036434174, + "step": 3070, + "valid_targets_mean": 3223.6, + "valid_targets_min": 1779 + }, + { + "epoch": 4.927884615384615, + "grad_norm": 0.5605700075398913, + "learning_rate": 9.774767873921357e-06, + "loss": 0.2142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2197868525981903, + "step": 3075, + "valid_targets_mean": 3292.1, + "valid_targets_min": 1981 + }, + { + "epoch": 4.935897435897436, + "grad_norm": 0.5388453900657821, + "learning_rate": 9.706165801991651e-06, + "loss": 0.1945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20840944349765778, + "step": 3080, + "valid_targets_mean": 3646.8, + "valid_targets_min": 2025 + }, + { + "epoch": 4.943910256410256, + "grad_norm": 0.5337480293840772, + "learning_rate": 9.637728095407593e-06, + "loss": 0.2032, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22162067890167236, + "step": 3085, + "valid_targets_mean": 3837.4, + "valid_targets_min": 2260 + }, + { + "epoch": 4.951923076923077, + "grad_norm": 0.5315055333333628, + "learning_rate": 9.56945584693861e-06, + "loss": 0.1986, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1910790205001831, + "step": 3090, + "valid_targets_mean": 2969.9, + "valid_targets_min": 1224 + }, + { + "epoch": 4.959935897435898, + "grad_norm": 0.531450262082612, + "learning_rate": 9.501350146712193e-06, + "loss": 0.2192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2249545156955719, + "step": 3095, + "valid_targets_mean": 3650.4, + "valid_targets_min": 582 + }, + { + "epoch": 4.967948717948718, + "grad_norm": 0.514593402922467, + "learning_rate": 9.433412082196527e-06, + "loss": 0.1948, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1740938276052475, + "step": 3100, + "valid_targets_mean": 3389.1, + "valid_targets_min": 567 + }, + { + "epoch": 4.975961538461538, + "grad_norm": 0.479202568267713, + "learning_rate": 9.365642738183044e-06, + "loss": 0.1976, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18655627965927124, + "step": 3105, + "valid_targets_mean": 3681.2, + "valid_targets_min": 1612 + }, + { + "epoch": 4.983974358974359, + "grad_norm": 0.5530192556043516, + "learning_rate": 9.298043196769217e-06, + "loss": 0.2213, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2281743586063385, + "step": 3110, + "valid_targets_mean": 3384.4, + "valid_targets_min": 320 + }, + { + "epoch": 4.99198717948718, + "grad_norm": 0.44807416141894424, + "learning_rate": 9.230614537341167e-06, + "loss": 0.1976, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1691162884235382, + "step": 3115, + "valid_targets_mean": 3748.1, + "valid_targets_min": 1861 + }, + { + "epoch": 5.0, + "grad_norm": 0.5409391376747347, + "learning_rate": 9.163357836556498e-06, + "loss": 0.2145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19238653779029846, + "step": 3120, + "valid_targets_mean": 3047.0, + "valid_targets_min": 288 + }, + { + "epoch": 5.00801282051282, + "grad_norm": 0.5113284257073623, + "learning_rate": 9.096274168327122e-06, + "loss": 0.1968, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1752883791923523, + "step": 3125, + "valid_targets_mean": 3437.4, + "valid_targets_min": 550 + }, + { + "epoch": 5.016025641025641, + "grad_norm": 0.5447988674027889, + "learning_rate": 9.029364603802017e-06, + "loss": 0.1885, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16521689295768738, + "step": 3130, + "valid_targets_mean": 3108.8, + "valid_targets_min": 1906 + }, + { + "epoch": 5.024038461538462, + "grad_norm": 0.648454130718421, + "learning_rate": 8.962630211350248e-06, + "loss": 0.1887, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18997415900230408, + "step": 3135, + "valid_targets_mean": 2555.8, + "valid_targets_min": 377 + }, + { + "epoch": 5.032051282051282, + "grad_norm": 0.5763204739253073, + "learning_rate": 8.89607205654378e-06, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17350471019744873, + "step": 3140, + "valid_targets_mean": 3285.9, + "valid_targets_min": 325 + }, + { + "epoch": 5.040064102564102, + "grad_norm": 0.5162424230093436, + "learning_rate": 8.829691202140591e-06, + "loss": 0.1878, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1592182070016861, + "step": 3145, + "valid_targets_mean": 3308.9, + "valid_targets_min": 391 + }, + { + "epoch": 5.048076923076923, + "grad_norm": 0.6082522256664168, + "learning_rate": 8.763488708067604e-06, + "loss": 0.1871, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2013545036315918, + "step": 3150, + "valid_targets_mean": 3266.9, + "valid_targets_min": 2157 + }, + { + "epoch": 5.056089743589744, + "grad_norm": 0.6021413534717205, + "learning_rate": 8.69746563140379e-06, + "loss": 0.2033, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21238872408866882, + "step": 3155, + "valid_targets_mean": 3137.6, + "valid_targets_min": 1170 + }, + { + "epoch": 5.064102564102564, + "grad_norm": 0.46975098631980283, + "learning_rate": 8.631623026363331e-06, + "loss": 0.1861, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13161732256412506, + "step": 3160, + "valid_targets_mean": 3321.1, + "valid_targets_min": 1695 + }, + { + "epoch": 5.072115384615385, + "grad_norm": 0.5887772251561811, + "learning_rate": 8.56596194427873e-06, + "loss": 0.1761, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20781591534614563, + "step": 3165, + "valid_targets_mean": 3307.9, + "valid_targets_min": 352 + }, + { + "epoch": 5.080128205128205, + "grad_norm": 0.5390332643634987, + "learning_rate": 8.500483433584054e-06, + "loss": 0.1769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16811692714691162, + "step": 3170, + "valid_targets_mean": 3268.9, + "valid_targets_min": 326 + }, + { + "epoch": 5.0881410256410255, + "grad_norm": 0.5460202693884373, + "learning_rate": 8.435188539798187e-06, + "loss": 0.205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23228251934051514, + "step": 3175, + "valid_targets_mean": 3874.6, + "valid_targets_min": 1886 + }, + { + "epoch": 5.096153846153846, + "grad_norm": 0.5859478298591612, + "learning_rate": 8.370078305508136e-06, + "loss": 0.2006, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21061252057552338, + "step": 3180, + "valid_targets_mean": 3122.8, + "valid_targets_min": 2093 + }, + { + "epoch": 5.104166666666667, + "grad_norm": 0.5059924010279726, + "learning_rate": 8.305153770352384e-06, + "loss": 0.2068, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2043697088956833, + "step": 3185, + "valid_targets_mean": 4493.2, + "valid_targets_min": 1944 + }, + { + "epoch": 5.112179487179487, + "grad_norm": 0.5253391559679185, + "learning_rate": 8.240415971004285e-06, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19598671793937683, + "step": 3190, + "valid_targets_mean": 3796.3, + "valid_targets_min": 2472 + }, + { + "epoch": 5.1201923076923075, + "grad_norm": 0.5296821750414983, + "learning_rate": 8.175865941155525e-06, + "loss": 0.2053, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1722840517759323, + "step": 3195, + "valid_targets_mean": 3151.8, + "valid_targets_min": 1415 + }, + { + "epoch": 5.128205128205128, + "grad_norm": 0.453908424287244, + "learning_rate": 8.111504711499598e-06, + "loss": 0.1865, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18074356019496918, + "step": 3200, + "valid_targets_mean": 4311.1, + "valid_targets_min": 835 + }, + { + "epoch": 5.136217948717949, + "grad_norm": 0.48570840686600897, + "learning_rate": 8.04733330971536e-06, + "loss": 0.1886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18450653553009033, + "step": 3205, + "valid_targets_mean": 4052.2, + "valid_targets_min": 2120 + }, + { + "epoch": 5.144230769230769, + "grad_norm": 0.6526366021581173, + "learning_rate": 7.983352760450618e-06, + "loss": 0.2022, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19894084334373474, + "step": 3210, + "valid_targets_mean": 2804.1, + "valid_targets_min": 338 + }, + { + "epoch": 5.152243589743589, + "grad_norm": 0.44063183279798923, + "learning_rate": 7.919564085305768e-06, + "loss": 0.1802, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15730887651443481, + "step": 3215, + "valid_targets_mean": 4473.3, + "valid_targets_min": 2253 + }, + { + "epoch": 5.160256410256411, + "grad_norm": 0.6712742300870267, + "learning_rate": 7.855968302817487e-06, + "loss": 0.1948, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2181408405303955, + "step": 3220, + "valid_targets_mean": 3240.4, + "valid_targets_min": 1565 + }, + { + "epoch": 5.168269230769231, + "grad_norm": 0.5614510327024675, + "learning_rate": 7.792566428442456e-06, + "loss": 0.1782, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17663878202438354, + "step": 3225, + "valid_targets_mean": 3257.2, + "valid_targets_min": 1846 + }, + { + "epoch": 5.176282051282051, + "grad_norm": 0.5163523322212359, + "learning_rate": 7.729359474541168e-06, + "loss": 0.1995, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2008240520954132, + "step": 3230, + "valid_targets_mean": 3838.0, + "valid_targets_min": 2021 + }, + { + "epoch": 5.184294871794872, + "grad_norm": 0.5690798020020454, + "learning_rate": 7.666348450361737e-06, + "loss": 0.192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2337350696325302, + "step": 3235, + "valid_targets_mean": 3405.2, + "valid_targets_min": 246 + }, + { + "epoch": 5.1923076923076925, + "grad_norm": 0.5742136826338571, + "learning_rate": 7.60353436202381e-06, + "loss": 0.1766, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1858733743429184, + "step": 3240, + "valid_targets_mean": 3966.5, + "valid_targets_min": 1872 + }, + { + "epoch": 5.200320512820513, + "grad_norm": 0.5122166596394158, + "learning_rate": 7.540918212502479e-06, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1741316318511963, + "step": 3245, + "valid_targets_mean": 3365.6, + "valid_targets_min": 1880 + }, + { + "epoch": 5.208333333333333, + "grad_norm": 0.49354163626246467, + "learning_rate": 7.478501001612281e-06, + "loss": 0.1886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17873799800872803, + "step": 3250, + "valid_targets_mean": 3825.0, + "valid_targets_min": 1619 + }, + { + "epoch": 5.216346153846154, + "grad_norm": 0.5443684108332746, + "learning_rate": 7.416283725991229e-06, + "loss": 0.1882, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19958025217056274, + "step": 3255, + "valid_targets_mean": 3123.3, + "valid_targets_min": 2049 + }, + { + "epoch": 5.2243589743589745, + "grad_norm": 0.518143554694147, + "learning_rate": 7.354267379084896e-06, + "loss": 0.192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18749485909938812, + "step": 3260, + "valid_targets_mean": 3432.8, + "valid_targets_min": 2052 + }, + { + "epoch": 5.232371794871795, + "grad_norm": 0.5727610082043441, + "learning_rate": 7.292452951130548e-06, + "loss": 0.1913, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22807061672210693, + "step": 3265, + "valid_targets_mean": 3272.8, + "valid_targets_min": 440 + }, + { + "epoch": 5.240384615384615, + "grad_norm": 0.5171761185493424, + "learning_rate": 7.230841429141347e-06, + "loss": 0.1736, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1712077558040619, + "step": 3270, + "valid_targets_mean": 3486.1, + "valid_targets_min": 590 + }, + { + "epoch": 5.248397435897436, + "grad_norm": 0.5753726653347081, + "learning_rate": 7.169433796890595e-06, + "loss": 0.1769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18391674757003784, + "step": 3275, + "valid_targets_mean": 3149.1, + "valid_targets_min": 600 + }, + { + "epoch": 5.256410256410256, + "grad_norm": 0.5827401383446655, + "learning_rate": 7.108231034895976e-06, + "loss": 0.1898, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22402942180633545, + "step": 3280, + "valid_targets_mean": 3546.9, + "valid_targets_min": 2170 + }, + { + "epoch": 5.264423076923077, + "grad_norm": 0.543075705148801, + "learning_rate": 7.047234120403972e-06, + "loss": 0.2025, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19148463010787964, + "step": 3285, + "valid_targets_mean": 3322.8, + "valid_targets_min": 1767 + }, + { + "epoch": 5.272435897435898, + "grad_norm": 0.5626066912048262, + "learning_rate": 6.986444027374211e-06, + "loss": 0.19, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19853419065475464, + "step": 3290, + "valid_targets_mean": 3568.6, + "valid_targets_min": 2061 + }, + { + "epoch": 5.280448717948718, + "grad_norm": 0.572779952262963, + "learning_rate": 6.925861726463919e-06, + "loss": 0.2096, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21987566351890564, + "step": 3295, + "valid_targets_mean": 3446.1, + "valid_targets_min": 1877 + }, + { + "epoch": 5.288461538461538, + "grad_norm": 0.577913018345779, + "learning_rate": 6.865488185012464e-06, + "loss": 0.1899, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18918846547603607, + "step": 3300, + "valid_targets_mean": 3336.0, + "valid_targets_min": 2215 + }, + { + "epoch": 5.296474358974359, + "grad_norm": 0.5075057087612974, + "learning_rate": 6.805324367025825e-06, + "loss": 0.1923, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1787465363740921, + "step": 3305, + "valid_targets_mean": 3670.3, + "valid_targets_min": 1922 + }, + { + "epoch": 5.30448717948718, + "grad_norm": 0.5001681486648626, + "learning_rate": 6.745371233161309e-06, + "loss": 0.1887, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19627542793750763, + "step": 3310, + "valid_targets_mean": 3670.1, + "valid_targets_min": 1944 + }, + { + "epoch": 5.3125, + "grad_norm": 0.48278736188583865, + "learning_rate": 6.685629740712103e-06, + "loss": 0.1851, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16993483901023865, + "step": 3315, + "valid_targets_mean": 3727.1, + "valid_targets_min": 1822 + }, + { + "epoch": 5.32051282051282, + "grad_norm": 0.5575051576217561, + "learning_rate": 6.6261008435920605e-06, + "loss": 0.2214, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20006653666496277, + "step": 3320, + "valid_targets_mean": 3215.5, + "valid_targets_min": 1937 + }, + { + "epoch": 5.328525641025641, + "grad_norm": 0.5388816043955157, + "learning_rate": 6.566785492320471e-06, + "loss": 0.2074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1908368021249771, + "step": 3325, + "valid_targets_mean": 3455.5, + "valid_targets_min": 2023 + }, + { + "epoch": 5.336538461538462, + "grad_norm": 0.5400123521185616, + "learning_rate": 6.507684634006815e-06, + "loss": 0.1925, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16674238443374634, + "step": 3330, + "valid_targets_mean": 3334.6, + "valid_targets_min": 1996 + }, + { + "epoch": 5.344551282051282, + "grad_norm": 0.5270342093008171, + "learning_rate": 6.448799212335734e-06, + "loss": 0.1982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1765822470188141, + "step": 3335, + "valid_targets_mean": 3522.7, + "valid_targets_min": 1690 + }, + { + "epoch": 5.352564102564102, + "grad_norm": 0.4636706571549095, + "learning_rate": 6.390130167551869e-06, + "loss": 0.1709, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14373044669628143, + "step": 3340, + "valid_targets_mean": 3593.2, + "valid_targets_min": 345 + }, + { + "epoch": 5.360576923076923, + "grad_norm": 0.5805185030734262, + "learning_rate": 6.331678436444939e-06, + "loss": 0.1792, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17474153637886047, + "step": 3345, + "valid_targets_mean": 2900.4, + "valid_targets_min": 377 + }, + { + "epoch": 5.368589743589744, + "grad_norm": 0.5396103910816336, + "learning_rate": 6.273444952334713e-06, + "loss": 0.1923, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19289200007915497, + "step": 3350, + "valid_targets_mean": 3375.8, + "valid_targets_min": 1496 + }, + { + "epoch": 5.376602564102564, + "grad_norm": 0.5447076740652468, + "learning_rate": 6.2154306450561175e-06, + "loss": 0.1782, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1780100166797638, + "step": 3355, + "valid_targets_mean": 3710.4, + "valid_targets_min": 646 + }, + { + "epoch": 5.384615384615385, + "grad_norm": 0.5516827352859298, + "learning_rate": 6.157636440944445e-06, + "loss": 0.1874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.191841721534729, + "step": 3360, + "valid_targets_mean": 3302.8, + "valid_targets_min": 328 + }, + { + "epoch": 5.392628205128205, + "grad_norm": 0.8580701767767885, + "learning_rate": 6.100063262820474e-06, + "loss": 0.18, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.187397301197052, + "step": 3365, + "valid_targets_mean": 3643.5, + "valid_targets_min": 1684 + }, + { + "epoch": 5.4006410256410255, + "grad_norm": 0.506561834066886, + "learning_rate": 6.0427120299758236e-06, + "loss": 0.2055, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1966952383518219, + "step": 3370, + "valid_targets_mean": 3982.6, + "valid_targets_min": 2118 + }, + { + "epoch": 5.408653846153846, + "grad_norm": 0.5289407099998846, + "learning_rate": 5.985583658158212e-06, + "loss": 0.1973, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20928660035133362, + "step": 3375, + "valid_targets_mean": 3634.4, + "valid_targets_min": 382 + }, + { + "epoch": 5.416666666666667, + "grad_norm": 0.47593762647730675, + "learning_rate": 5.928679059556852e-06, + "loss": 0.1747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15979638695716858, + "step": 3380, + "valid_targets_mean": 4029.1, + "valid_targets_min": 1971 + }, + { + "epoch": 5.424679487179487, + "grad_norm": 0.54615681581572, + "learning_rate": 5.871999142787908e-06, + "loss": 0.1939, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18546456098556519, + "step": 3385, + "valid_targets_mean": 3214.1, + "valid_targets_min": 1245 + }, + { + "epoch": 5.4326923076923075, + "grad_norm": 0.5240694658371231, + "learning_rate": 5.815544812879936e-06, + "loss": 0.1886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20221960544586182, + "step": 3390, + "valid_targets_mean": 3662.1, + "valid_targets_min": 2329 + }, + { + "epoch": 5.440705128205128, + "grad_norm": 0.6164173562320571, + "learning_rate": 5.759316971259503e-06, + "loss": 0.189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24456317722797394, + "step": 3395, + "valid_targets_mean": 3020.1, + "valid_targets_min": 489 + }, + { + "epoch": 5.448717948717949, + "grad_norm": 0.5726966590420226, + "learning_rate": 5.703316515736734e-06, + "loss": 0.175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1823807656764984, + "step": 3400, + "valid_targets_mean": 2954.6, + "valid_targets_min": 383 + }, + { + "epoch": 5.456730769230769, + "grad_norm": 0.514223543564771, + "learning_rate": 5.647544340491007e-06, + "loss": 0.1919, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1850132793188095, + "step": 3405, + "valid_targets_mean": 3505.2, + "valid_targets_min": 831 + }, + { + "epoch": 5.464743589743589, + "grad_norm": 0.6189202501534701, + "learning_rate": 5.592001336056659e-06, + "loss": 0.2149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1900269091129303, + "step": 3410, + "valid_targets_mean": 3419.9, + "valid_targets_min": 2164 + }, + { + "epoch": 5.472756410256411, + "grad_norm": 0.6094758033614434, + "learning_rate": 5.536688389308782e-06, + "loss": 0.1864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20669332146644592, + "step": 3415, + "valid_targets_mean": 2933.1, + "valid_targets_min": 303 + }, + { + "epoch": 5.480769230769231, + "grad_norm": 0.437352007826735, + "learning_rate": 5.4816063834490496e-06, + "loss": 0.1736, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18805278837680817, + "step": 3420, + "valid_targets_mean": 4861.1, + "valid_targets_min": 239 + }, + { + "epoch": 5.488782051282051, + "grad_norm": 0.5937375893287633, + "learning_rate": 5.426756197991625e-06, + "loss": 0.2026, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19615274667739868, + "step": 3425, + "valid_targets_mean": 3083.9, + "valid_targets_min": 281 + }, + { + "epoch": 5.496794871794872, + "grad_norm": 0.5292374325348774, + "learning_rate": 5.372138708749104e-06, + "loss": 0.1829, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17313215136528015, + "step": 3430, + "valid_targets_mean": 3637.6, + "valid_targets_min": 1848 + }, + { + "epoch": 5.5048076923076925, + "grad_norm": 0.5904116952834751, + "learning_rate": 5.3177547878185436e-06, + "loss": 0.19, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18827790021896362, + "step": 3435, + "valid_targets_mean": 3208.8, + "valid_targets_min": 334 + }, + { + "epoch": 5.512820512820513, + "grad_norm": 0.5009765474233788, + "learning_rate": 5.263605303567532e-06, + "loss": 0.179, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15040534734725952, + "step": 3440, + "valid_targets_mean": 3404.1, + "valid_targets_min": 414 + }, + { + "epoch": 5.520833333333333, + "grad_norm": 0.46199036273708716, + "learning_rate": 5.20969112062032e-06, + "loss": 0.1767, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.155937060713768, + "step": 3445, + "valid_targets_mean": 3935.8, + "valid_targets_min": 464 + }, + { + "epoch": 5.528846153846154, + "grad_norm": 0.5119196894630544, + "learning_rate": 5.156013099844017e-06, + "loss": 0.1809, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1872662901878357, + "step": 3450, + "valid_targets_mean": 3880.6, + "valid_targets_min": 1893 + }, + { + "epoch": 5.5368589743589745, + "grad_norm": 0.566007537463004, + "learning_rate": 5.1025720983348544e-06, + "loss": 0.2005, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19776442646980286, + "step": 3455, + "valid_targets_mean": 3433.2, + "valid_targets_min": 1809 + }, + { + "epoch": 5.544871794871795, + "grad_norm": 0.5514081766003055, + "learning_rate": 5.049368969404484e-06, + "loss": 0.2006, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1966758370399475, + "step": 3460, + "valid_targets_mean": 3726.0, + "valid_targets_min": 2228 + }, + { + "epoch": 5.552884615384615, + "grad_norm": 0.5654081475081372, + "learning_rate": 4.99640456256636e-06, + "loss": 0.1994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18677841126918793, + "step": 3465, + "valid_targets_mean": 3471.7, + "valid_targets_min": 432 + }, + { + "epoch": 5.560897435897436, + "grad_norm": 0.5163881738271517, + "learning_rate": 4.9436797235221814e-06, + "loss": 0.1762, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18354731798171997, + "step": 3470, + "valid_targets_mean": 3611.6, + "valid_targets_min": 250 + }, + { + "epoch": 5.568910256410256, + "grad_norm": 0.5434683494215142, + "learning_rate": 4.891195294148376e-06, + "loss": 0.2059, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1964164674282074, + "step": 3475, + "valid_targets_mean": 3558.6, + "valid_targets_min": 1768 + }, + { + "epoch": 5.576923076923077, + "grad_norm": 0.5336897538504671, + "learning_rate": 4.838952112482671e-06, + "loss": 0.1916, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16893991827964783, + "step": 3480, + "valid_targets_mean": 3513.5, + "valid_targets_min": 1784 + }, + { + "epoch": 5.584935897435898, + "grad_norm": 0.42933770422501677, + "learning_rate": 4.786951012710699e-06, + "loss": 0.1862, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15886548161506653, + "step": 3485, + "valid_targets_mean": 4568.1, + "valid_targets_min": 1430 + }, + { + "epoch": 5.592948717948718, + "grad_norm": 0.4188860911448219, + "learning_rate": 4.735192825152686e-06, + "loss": 0.1701, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16202595829963684, + "step": 3490, + "valid_targets_mean": 4717.6, + "valid_targets_min": 1939 + }, + { + "epoch": 5.600961538461538, + "grad_norm": 0.4593864099417896, + "learning_rate": 4.683678376250189e-06, + "loss": 0.1876, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17314445972442627, + "step": 3495, + "valid_targets_mean": 3925.8, + "valid_targets_min": 377 + }, + { + "epoch": 5.608974358974359, + "grad_norm": 0.48198100944237676, + "learning_rate": 4.6324084885529086e-06, + "loss": 0.1793, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16028594970703125, + "step": 3500, + "valid_targets_mean": 3850.2, + "valid_targets_min": 2018 + }, + { + "epoch": 5.61698717948718, + "grad_norm": 0.6326650398575101, + "learning_rate": 4.581383980705538e-06, + "loss": 0.2071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2744472324848175, + "step": 3505, + "valid_targets_mean": 3211.5, + "valid_targets_min": 1897 + }, + { + "epoch": 5.625, + "grad_norm": 0.492480738098872, + "learning_rate": 4.530605667434727e-06, + "loss": 0.1906, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18184933066368103, + "step": 3510, + "valid_targets_mean": 3921.9, + "valid_targets_min": 1666 + }, + { + "epoch": 5.63301282051282, + "grad_norm": 0.5627002261210652, + "learning_rate": 4.480074359536013e-06, + "loss": 0.1957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19673556089401245, + "step": 3515, + "valid_targets_mean": 3383.2, + "valid_targets_min": 288 + }, + { + "epoch": 5.641025641025641, + "grad_norm": 0.5402686277972082, + "learning_rate": 4.429790863860934e-06, + "loss": 0.1932, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19620636105537415, + "step": 3520, + "valid_targets_mean": 3515.4, + "valid_targets_min": 578 + }, + { + "epoch": 5.649038461538462, + "grad_norm": 0.455833870213205, + "learning_rate": 4.3797559833041146e-06, + "loss": 0.1787, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15487775206565857, + "step": 3525, + "valid_targets_mean": 3998.6, + "valid_targets_min": 1971 + }, + { + "epoch": 5.657051282051282, + "grad_norm": 0.5107407870906246, + "learning_rate": 4.329970516790447e-06, + "loss": 0.1827, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1705532819032669, + "step": 3530, + "valid_targets_mean": 3715.2, + "valid_targets_min": 1940 + }, + { + "epoch": 5.665064102564102, + "grad_norm": 0.46992914278548026, + "learning_rate": 4.280435259262363e-06, + "loss": 0.1979, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14910975098609924, + "step": 3535, + "valid_targets_mean": 4047.2, + "valid_targets_min": 233 + }, + { + "epoch": 5.673076923076923, + "grad_norm": 0.4964648069391279, + "learning_rate": 4.231151001667077e-06, + "loss": 0.2083, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2373436838388443, + "step": 3540, + "valid_targets_mean": 4289.2, + "valid_targets_min": 1797 + }, + { + "epoch": 5.681089743589744, + "grad_norm": 0.5259660873921462, + "learning_rate": 4.182118530944044e-06, + "loss": 0.1942, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18008257448673248, + "step": 3545, + "valid_targets_mean": 3665.4, + "valid_targets_min": 380 + }, + { + "epoch": 5.689102564102564, + "grad_norm": 0.5056551983974236, + "learning_rate": 4.133338630012307e-06, + "loss": 0.1759, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19457197189331055, + "step": 3550, + "valid_targets_mean": 3579.0, + "valid_targets_min": 298 + }, + { + "epoch": 5.697115384615385, + "grad_norm": 0.6123253119930623, + "learning_rate": 4.0848120777580554e-06, + "loss": 0.2028, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18530994653701782, + "step": 3555, + "valid_targets_mean": 2964.7, + "valid_targets_min": 821 + }, + { + "epoch": 5.705128205128205, + "grad_norm": 0.5139721617285482, + "learning_rate": 4.036539649022182e-06, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20200860500335693, + "step": 3560, + "valid_targets_mean": 3852.3, + "valid_targets_min": 351 + }, + { + "epoch": 5.7131410256410255, + "grad_norm": 0.5399140803806365, + "learning_rate": 3.988522114587865e-06, + "loss": 0.1877, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1806945502758026, + "step": 3565, + "valid_targets_mean": 3429.0, + "valid_targets_min": 1667 + }, + { + "epoch": 5.721153846153846, + "grad_norm": 0.5501809465020228, + "learning_rate": 3.940760241168331e-06, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1809982806444168, + "step": 3570, + "valid_targets_mean": 2989.8, + "valid_targets_min": 318 + }, + { + "epoch": 5.729166666666667, + "grad_norm": 0.5844901005312456, + "learning_rate": 3.893254791394541e-06, + "loss": 0.1938, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19920454919338226, + "step": 3575, + "valid_targets_mean": 3151.3, + "valid_targets_min": 917 + }, + { + "epoch": 5.737179487179487, + "grad_norm": 0.501643292559648, + "learning_rate": 3.846006523803074e-06, + "loss": 0.1781, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1691002994775772, + "step": 3580, + "valid_targets_mean": 3593.8, + "valid_targets_min": 1719 + }, + { + "epoch": 5.7451923076923075, + "grad_norm": 0.5671413233547391, + "learning_rate": 3.799016192823981e-06, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17352180182933807, + "step": 3585, + "valid_targets_mean": 3414.9, + "valid_targets_min": 290 + }, + { + "epoch": 5.753205128205128, + "grad_norm": 0.5666972330482487, + "learning_rate": 3.7522845487687276e-06, + "loss": 0.1839, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1914469301700592, + "step": 3590, + "valid_targets_mean": 3221.8, + "valid_targets_min": 1045 + }, + { + "epoch": 5.761217948717949, + "grad_norm": 0.5351797553861585, + "learning_rate": 3.7058123378182664e-06, + "loss": 0.1948, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2058011293411255, + "step": 3595, + "valid_targets_mean": 3487.9, + "valid_targets_min": 304 + }, + { + "epoch": 5.769230769230769, + "grad_norm": 0.4599129294891559, + "learning_rate": 3.6596003020110636e-06, + "loss": 0.1853, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18410655856132507, + "step": 3600, + "valid_targets_mean": 4461.4, + "valid_targets_min": 1898 + }, + { + "epoch": 5.777243589743589, + "grad_norm": 0.4957453124553632, + "learning_rate": 3.613649179231287e-06, + "loss": 0.1816, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17034442722797394, + "step": 3605, + "valid_targets_mean": 3607.4, + "valid_targets_min": 836 + }, + { + "epoch": 5.785256410256411, + "grad_norm": 0.47090246675664355, + "learning_rate": 3.5679597031970017e-06, + "loss": 0.1861, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1710040122270584, + "step": 3610, + "valid_targets_mean": 4074.2, + "valid_targets_min": 2139 + }, + { + "epoch": 5.793269230769231, + "grad_norm": 0.5611368048797609, + "learning_rate": 3.5225326034484764e-06, + "loss": 0.1823, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19601497054100037, + "step": 3615, + "valid_targets_mean": 3592.8, + "valid_targets_min": 418 + }, + { + "epoch": 5.801282051282051, + "grad_norm": 0.5429508267587924, + "learning_rate": 3.4773686053365197e-06, + "loss": 0.1956, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25647762417793274, + "step": 3620, + "valid_targets_mean": 3899.1, + "valid_targets_min": 274 + }, + { + "epoch": 5.809294871794872, + "grad_norm": 0.6267818504699711, + "learning_rate": 3.4324684300109003e-06, + "loss": 0.1873, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22126325964927673, + "step": 3625, + "valid_targets_mean": 3169.4, + "valid_targets_min": 340 + }, + { + "epoch": 5.8173076923076925, + "grad_norm": 0.6003395101225485, + "learning_rate": 3.387832794408832e-06, + "loss": 0.1862, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19179658591747284, + "step": 3630, + "valid_targets_mean": 3175.5, + "valid_targets_min": 1761 + }, + { + "epoch": 5.825320512820513, + "grad_norm": 0.8231448274682673, + "learning_rate": 3.3434624112435342e-06, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15372401475906372, + "step": 3635, + "valid_targets_mean": 3813.9, + "valid_targets_min": 2058 + }, + { + "epoch": 5.833333333333333, + "grad_norm": 0.5695754219453818, + "learning_rate": 3.2993579889928397e-06, + "loss": 0.1975, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23370450735092163, + "step": 3640, + "valid_targets_mean": 3412.4, + "valid_targets_min": 2048 + }, + { + "epoch": 5.841346153846154, + "grad_norm": 0.6502128293688998, + "learning_rate": 3.25552023188789e-06, + "loss": 0.1786, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21419233083724976, + "step": 3645, + "valid_targets_mean": 3037.8, + "valid_targets_min": 316 + }, + { + "epoch": 5.8493589743589745, + "grad_norm": 0.5359902686377935, + "learning_rate": 3.211949839901889e-06, + "loss": 0.1982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2461708039045334, + "step": 3650, + "valid_targets_mean": 3926.8, + "valid_targets_min": 1331 + }, + { + "epoch": 5.857371794871795, + "grad_norm": 0.5659386559768532, + "learning_rate": 3.168647508738927e-06, + "loss": 0.1948, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24257324635982513, + "step": 3655, + "valid_targets_mean": 3724.2, + "valid_targets_min": 306 + }, + { + "epoch": 5.865384615384615, + "grad_norm": 0.37056662744745467, + "learning_rate": 3.125613929822866e-06, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12163887917995453, + "step": 3660, + "valid_targets_mean": 5052.7, + "valid_targets_min": 2161 + }, + { + "epoch": 5.873397435897436, + "grad_norm": 0.5838541068133365, + "learning_rate": 3.0828497902863106e-06, + "loss": 0.2097, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22443059086799622, + "step": 3665, + "valid_targets_mean": 3060.1, + "valid_targets_min": 903 + }, + { + "epoch": 5.881410256410256, + "grad_norm": 0.5682074734014336, + "learning_rate": 3.0403557729596267e-06, + "loss": 0.1946, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19512076675891876, + "step": 3670, + "valid_targets_mean": 3185.4, + "valid_targets_min": 1948 + }, + { + "epoch": 5.889423076923077, + "grad_norm": 0.6022044204063187, + "learning_rate": 2.998132556360038e-06, + "loss": 0.1907, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22192490100860596, + "step": 3675, + "valid_targets_mean": 3249.8, + "valid_targets_min": 1676 + }, + { + "epoch": 5.897435897435898, + "grad_norm": 0.6195066221518493, + "learning_rate": 2.9561808146808068e-06, + "loss": 0.2094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19596882164478302, + "step": 3680, + "valid_targets_mean": 2721.9, + "valid_targets_min": 1038 + }, + { + "epoch": 5.905448717948718, + "grad_norm": 0.574156043742674, + "learning_rate": 2.9145012177804476e-06, + "loss": 0.2033, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21469567716121674, + "step": 3685, + "valid_targets_mean": 3452.7, + "valid_targets_min": 369 + }, + { + "epoch": 5.913461538461538, + "grad_norm": 0.5087952282766907, + "learning_rate": 2.8730944311720454e-06, + "loss": 0.1961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2278899997472763, + "step": 3690, + "valid_targets_mean": 4572.1, + "valid_targets_min": 2118 + }, + { + "epoch": 5.921474358974359, + "grad_norm": 0.47323882313623933, + "learning_rate": 2.8319611160126226e-06, + "loss": 0.1876, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16829034686088562, + "step": 3695, + "valid_targets_mean": 4650.4, + "valid_targets_min": 1963 + }, + { + "epoch": 5.92948717948718, + "grad_norm": 0.4862546032433573, + "learning_rate": 2.791101929092592e-06, + "loss": 0.1908, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17378859221935272, + "step": 3700, + "valid_targets_mean": 3729.2, + "valid_targets_min": 836 + }, + { + "epoch": 5.9375, + "grad_norm": 0.5851260153472438, + "learning_rate": 2.750517522825251e-06, + "loss": 0.1826, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2061779797077179, + "step": 3705, + "valid_targets_mean": 3207.8, + "valid_targets_min": 1876 + }, + { + "epoch": 5.94551282051282, + "grad_norm": 0.5347583134407086, + "learning_rate": 2.710208545236397e-06, + "loss": 0.189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17754028737545013, + "step": 3710, + "valid_targets_mean": 3264.6, + "valid_targets_min": 420 + }, + { + "epoch": 5.953525641025641, + "grad_norm": 0.5584631349323056, + "learning_rate": 2.670175639953929e-06, + "loss": 0.1806, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18711164593696594, + "step": 3715, + "valid_targets_mean": 3365.9, + "valid_targets_min": 1625 + }, + { + "epoch": 5.961538461538462, + "grad_norm": 0.5689525852014578, + "learning_rate": 2.6304194461976207e-06, + "loss": 0.1785, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17805981636047363, + "step": 3720, + "valid_targets_mean": 2968.3, + "valid_targets_min": 1918 + }, + { + "epoch": 5.969551282051282, + "grad_norm": 0.5742462138075193, + "learning_rate": 2.5909405987688896e-06, + "loss": 0.1828, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18538996577262878, + "step": 3725, + "valid_targets_mean": 2999.2, + "valid_targets_min": 450 + }, + { + "epoch": 5.977564102564102, + "grad_norm": 0.5474544354626358, + "learning_rate": 2.5517397280406565e-06, + "loss": 0.1839, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18040937185287476, + "step": 3730, + "valid_targets_mean": 3083.9, + "valid_targets_min": 1855 + }, + { + "epoch": 5.985576923076923, + "grad_norm": 0.5370344222680715, + "learning_rate": 2.512817459947312e-06, + "loss": 0.1937, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1840437948703766, + "step": 3735, + "valid_targets_mean": 3041.1, + "valid_targets_min": 833 + }, + { + "epoch": 5.993589743589744, + "grad_norm": 0.557742421962461, + "learning_rate": 2.4741744159746618e-06, + "loss": 0.1953, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20614948868751526, + "step": 3740, + "valid_targets_mean": 3577.1, + "valid_targets_min": 322 + }, + { + "epoch": 6.001602564102564, + "grad_norm": 0.5492539692183419, + "learning_rate": 2.435811213150079e-06, + "loss": 0.1933, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20480048656463623, + "step": 3745, + "valid_targets_mean": 3380.0, + "valid_targets_min": 2101 + }, + { + "epoch": 6.009615384615385, + "grad_norm": 0.5012174932189678, + "learning_rate": 2.3977284640325805e-06, + "loss": 0.1865, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16375727951526642, + "step": 3750, + "valid_targets_mean": 3257.8, + "valid_targets_min": 279 + }, + { + "epoch": 6.017628205128205, + "grad_norm": 0.5150033985791577, + "learning_rate": 2.359926776703092e-06, + "loss": 0.1959, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17800338566303253, + "step": 3755, + "valid_targets_mean": 3647.5, + "valid_targets_min": 2028 + }, + { + "epoch": 6.0256410256410255, + "grad_norm": 0.6125072628848407, + "learning_rate": 2.3224067547547357e-06, + "loss": 0.1902, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24423670768737793, + "step": 3760, + "valid_targets_mean": 3147.4, + "valid_targets_min": 1749 + }, + { + "epoch": 6.033653846153846, + "grad_norm": 0.5290504030535882, + "learning_rate": 2.2851689972831536e-06, + "loss": 0.1917, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1971917450428009, + "step": 3765, + "valid_targets_mean": 3512.4, + "valid_targets_min": 363 + }, + { + "epoch": 6.041666666666667, + "grad_norm": 0.5074131034125838, + "learning_rate": 2.248214098877002e-06, + "loss": 0.1803, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17116963863372803, + "step": 3770, + "valid_targets_mean": 3892.2, + "valid_targets_min": 326 + }, + { + "epoch": 6.049679487179487, + "grad_norm": 0.6370750582627046, + "learning_rate": 2.2115426496083958e-06, + "loss": 0.1997, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24153323471546173, + "step": 3775, + "valid_targets_mean": 3046.4, + "valid_targets_min": 1685 + }, + { + "epoch": 6.0576923076923075, + "grad_norm": 0.5857968742273449, + "learning_rate": 2.175155235023536e-06, + "loss": 0.1864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22333982586860657, + "step": 3780, + "valid_targets_mean": 3206.4, + "valid_targets_min": 764 + }, + { + "epoch": 6.065705128205129, + "grad_norm": 0.5720754742458641, + "learning_rate": 2.1390524361333355e-06, + "loss": 0.1836, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1934468150138855, + "step": 3785, + "valid_targets_mean": 3093.9, + "valid_targets_min": 469 + }, + { + "epoch": 6.073717948717949, + "grad_norm": 0.6235776979123787, + "learning_rate": 2.1032348294041305e-06, + "loss": 0.1835, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17414945363998413, + "step": 3790, + "valid_targets_mean": 2934.5, + "valid_targets_min": 1846 + }, + { + "epoch": 6.081730769230769, + "grad_norm": 0.5800575852208971, + "learning_rate": 2.067702986748521e-06, + "loss": 0.1876, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21823963522911072, + "step": 3795, + "valid_targets_mean": 3321.6, + "valid_targets_min": 1872 + }, + { + "epoch": 6.089743589743589, + "grad_norm": 0.5380548846853326, + "learning_rate": 2.0324574755161764e-06, + "loss": 0.1909, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1871141493320465, + "step": 3800, + "valid_targets_mean": 3578.1, + "valid_targets_min": 1717 + }, + { + "epoch": 6.097756410256411, + "grad_norm": 0.6216251637611282, + "learning_rate": 1.9974988584848385e-06, + "loss": 0.1981, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21641653776168823, + "step": 3805, + "valid_targets_mean": 3018.4, + "valid_targets_min": 1902 + }, + { + "epoch": 6.105769230769231, + "grad_norm": 0.4499232995661724, + "learning_rate": 1.96282769385129e-06, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13558490574359894, + "step": 3810, + "valid_targets_mean": 3978.8, + "valid_targets_min": 1748 + }, + { + "epoch": 6.113782051282051, + "grad_norm": 0.5440572437785779, + "learning_rate": 1.9284445352224625e-06, + "loss": 0.1821, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1783181130886078, + "step": 3815, + "valid_targets_mean": 3168.1, + "valid_targets_min": 1751 + }, + { + "epoch": 6.121794871794871, + "grad_norm": 0.5104998731463487, + "learning_rate": 1.894349931606596e-06, + "loss": 0.1635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18172745406627655, + "step": 3820, + "valid_targets_mean": 3574.4, + "valid_targets_min": 314 + }, + { + "epoch": 6.1298076923076925, + "grad_norm": 0.4896220667407711, + "learning_rate": 1.8605444274044493e-06, + "loss": 0.1899, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18478181958198547, + "step": 3825, + "valid_targets_mean": 3944.8, + "valid_targets_min": 2045 + }, + { + "epoch": 6.137820512820513, + "grad_norm": 0.48989400822262036, + "learning_rate": 1.827028562400659e-06, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16057899594306946, + "step": 3830, + "valid_targets_mean": 3592.7, + "valid_targets_min": 352 + }, + { + "epoch": 6.145833333333333, + "grad_norm": 0.47494365061846483, + "learning_rate": 1.793802871755066e-06, + "loss": 0.1753, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14633449912071228, + "step": 3835, + "valid_targets_mean": 3542.4, + "valid_targets_min": 2396 + }, + { + "epoch": 6.153846153846154, + "grad_norm": 0.5904921538812401, + "learning_rate": 1.760867885994202e-06, + "loss": 0.17, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2000691294670105, + "step": 3840, + "valid_targets_mean": 3720.2, + "valid_targets_min": 1964 + }, + { + "epoch": 6.1618589743589745, + "grad_norm": 0.5080439984155285, + "learning_rate": 1.7282241310028047e-06, + "loss": 0.1722, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16287338733673096, + "step": 3845, + "valid_targets_mean": 3678.0, + "valid_targets_min": 2342 + }, + { + "epoch": 6.169871794871795, + "grad_norm": 0.6316969144820093, + "learning_rate": 1.6958721280154232e-06, + "loss": 0.184, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20093393325805664, + "step": 3850, + "valid_targets_mean": 2975.8, + "valid_targets_min": 2118 + }, + { + "epoch": 6.177884615384615, + "grad_norm": 0.5629266503674685, + "learning_rate": 1.6638123936081085e-06, + "loss": 0.1945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1980648934841156, + "step": 3855, + "valid_targets_mean": 3402.4, + "valid_targets_min": 361 + }, + { + "epoch": 6.185897435897436, + "grad_norm": 0.4901786276304485, + "learning_rate": 1.6320454396901463e-06, + "loss": 0.1875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1593923568725586, + "step": 3860, + "valid_targets_mean": 3699.1, + "valid_targets_min": 2111 + }, + { + "epoch": 6.193910256410256, + "grad_norm": 0.496727862672396, + "learning_rate": 1.6005717734958914e-06, + "loss": 0.1742, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14378324151039124, + "step": 3865, + "valid_targets_mean": 3391.4, + "valid_targets_min": 322 + }, + { + "epoch": 6.201923076923077, + "grad_norm": 0.5874727303234155, + "learning_rate": 1.569391897576671e-06, + "loss": 0.1857, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17528510093688965, + "step": 3870, + "valid_targets_mean": 3422.4, + "valid_targets_min": 366 + }, + { + "epoch": 6.209935897435898, + "grad_norm": 0.49680224237276693, + "learning_rate": 1.5385063097927533e-06, + "loss": 0.2021, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18529458343982697, + "step": 3875, + "valid_targets_mean": 3745.9, + "valid_targets_min": 1508 + }, + { + "epoch": 6.217948717948718, + "grad_norm": 0.498633204184822, + "learning_rate": 1.5079155033054104e-06, + "loss": 0.1791, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16825734078884125, + "step": 3880, + "valid_targets_mean": 4082.2, + "valid_targets_min": 1906 + }, + { + "epoch": 6.225961538461538, + "grad_norm": 0.4756372354476479, + "learning_rate": 1.4776199665690239e-06, + "loss": 0.197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15564313530921936, + "step": 3885, + "valid_targets_mean": 3845.8, + "valid_targets_min": 1946 + }, + { + "epoch": 6.233974358974359, + "grad_norm": 0.5104917457868668, + "learning_rate": 1.4476201833233084e-06, + "loss": 0.1815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1515727937221527, + "step": 3890, + "valid_targets_mean": 3294.8, + "valid_targets_min": 2030 + }, + { + "epoch": 6.24198717948718, + "grad_norm": 0.5320951676861468, + "learning_rate": 1.4179166325855676e-06, + "loss": 0.1914, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17016607522964478, + "step": 3895, + "valid_targets_mean": 3325.2, + "valid_targets_min": 1080 + }, + { + "epoch": 6.25, + "grad_norm": 0.5226640302399762, + "learning_rate": 1.3885097886430599e-06, + "loss": 0.1822, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17012283205986023, + "step": 3900, + "valid_targets_mean": 3378.5, + "valid_targets_min": 277 + }, + { + "epoch": 6.25801282051282, + "grad_norm": 0.4885361914877402, + "learning_rate": 1.35940012104542e-06, + "loss": 0.1894, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1809297651052475, + "step": 3905, + "valid_targets_mean": 4060.0, + "valid_targets_min": 1970 + }, + { + "epoch": 6.266025641025641, + "grad_norm": 0.5500088608225795, + "learning_rate": 1.3305880945971583e-06, + "loss": 0.2108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20557723939418793, + "step": 3910, + "valid_targets_mean": 3452.0, + "valid_targets_min": 1545 + }, + { + "epoch": 6.274038461538462, + "grad_norm": 0.5343047377776239, + "learning_rate": 1.3020741693502403e-06, + "loss": 0.1887, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1847967505455017, + "step": 3915, + "valid_targets_mean": 3522.4, + "valid_targets_min": 814 + }, + { + "epoch": 6.282051282051282, + "grad_norm": 0.5724110582086094, + "learning_rate": 1.27385880059675e-06, + "loss": 0.187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15530461072921753, + "step": 3920, + "valid_targets_mean": 3021.8, + "valid_targets_min": 233 + }, + { + "epoch": 6.290064102564102, + "grad_norm": 0.5189903179373517, + "learning_rate": 1.245942438861607e-06, + "loss": 0.1852, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16924820840358734, + "step": 3925, + "valid_targets_mean": 3413.4, + "valid_targets_min": 1713 + }, + { + "epoch": 6.298076923076923, + "grad_norm": 0.5487751899432807, + "learning_rate": 1.2183255298953788e-06, + "loss": 0.1835, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17502455413341522, + "step": 3930, + "valid_targets_mean": 3519.8, + "valid_targets_min": 260 + }, + { + "epoch": 6.306089743589744, + "grad_norm": 0.5903657333858096, + "learning_rate": 1.1910085146671645e-06, + "loss": 0.1868, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18399037420749664, + "step": 3935, + "valid_targets_mean": 3115.6, + "valid_targets_min": 224 + }, + { + "epoch": 6.314102564102564, + "grad_norm": 0.5420035653385537, + "learning_rate": 1.1639918293575492e-06, + "loss": 0.1921, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21313560009002686, + "step": 3940, + "valid_targets_mean": 3612.5, + "valid_targets_min": 600 + }, + { + "epoch": 6.322115384615385, + "grad_norm": 0.5477870467404146, + "learning_rate": 1.1372759053516536e-06, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1668134331703186, + "step": 3945, + "valid_targets_mean": 3054.5, + "valid_targets_min": 466 + }, + { + "epoch": 6.330128205128205, + "grad_norm": 0.407118652221098, + "learning_rate": 1.1108611692322157e-06, + "loss": 0.16, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13269789516925812, + "step": 3950, + "valid_targets_mean": 4318.4, + "valid_targets_min": 1846 + }, + { + "epoch": 6.3381410256410255, + "grad_norm": 0.6041896354253443, + "learning_rate": 1.0847480427728142e-06, + "loss": 0.181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19352270662784576, + "step": 3955, + "valid_targets_mean": 2949.6, + "valid_targets_min": 244 + }, + { + "epoch": 6.346153846153846, + "grad_norm": 0.44110066264683917, + "learning_rate": 1.0589369429311125e-06, + "loss": 0.1672, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1455502212047577, + "step": 3960, + "valid_targets_mean": 4418.0, + "valid_targets_min": 290 + }, + { + "epoch": 6.354166666666667, + "grad_norm": 0.5549047133023548, + "learning_rate": 1.0334282818422037e-06, + "loss": 0.1935, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17327052354812622, + "step": 3965, + "valid_targets_mean": 3370.1, + "valid_targets_min": 1507 + }, + { + "epoch": 6.362179487179487, + "grad_norm": 0.5740600633298263, + "learning_rate": 1.008222466812041e-06, + "loss": 0.1991, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2384001463651657, + "step": 3970, + "valid_targets_mean": 3610.7, + "valid_targets_min": 1958 + }, + { + "epoch": 6.3701923076923075, + "grad_norm": 0.5797903414700959, + "learning_rate": 9.83319900310915e-07, + "loss": 0.1933, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2142302691936493, + "step": 3975, + "valid_targets_mean": 3315.9, + "valid_targets_min": 735 + }, + { + "epoch": 6.378205128205128, + "grad_norm": 0.48995819790666606, + "learning_rate": 9.587209799670495e-07, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17958799004554749, + "step": 3980, + "valid_targets_mean": 4395.9, + "valid_targets_min": 2026 + }, + { + "epoch": 6.386217948717949, + "grad_norm": 0.5696379795450798, + "learning_rate": 9.344260985602327e-07, + "loss": 0.1761, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16386675834655762, + "step": 3985, + "valid_targets_mean": 3073.3, + "valid_targets_min": 239 + }, + { + "epoch": 6.394230769230769, + "grad_norm": 0.534668018184061, + "learning_rate": 9.104356440155526e-07, + "loss": 0.1692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17810484766960144, + "step": 3990, + "valid_targets_mean": 3377.9, + "valid_targets_min": 302 + }, + { + "epoch": 6.402243589743589, + "grad_norm": 0.4734615929959692, + "learning_rate": 8.867499993972162e-07, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16918233036994934, + "step": 3995, + "valid_targets_mean": 4161.4, + "valid_targets_min": 2309 + }, + { + "epoch": 6.410256410256411, + "grad_norm": 0.5439903809679225, + "learning_rate": 8.633695429024058e-07, + "loss": 0.1951, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18606598675251007, + "step": 4000, + "valid_targets_mean": 3215.1, + "valid_targets_min": 1557 + }, + { + "epoch": 6.418269230769231, + "grad_norm": 0.5942152209139376, + "learning_rate": 8.402946478552732e-07, + "loss": 0.1803, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18717849254608154, + "step": 4005, + "valid_targets_mean": 2919.1, + "valid_targets_min": 360 + }, + { + "epoch": 6.426282051282051, + "grad_norm": 0.5775374131849361, + "learning_rate": 8.175256827009392e-07, + "loss": 0.1788, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2013743817806244, + "step": 4010, + "valid_targets_mean": 3049.1, + "valid_targets_min": 394 + }, + { + "epoch": 6.434294871794872, + "grad_norm": 0.5969180238661153, + "learning_rate": 7.95063010999646e-07, + "loss": 0.1767, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18375027179718018, + "step": 4015, + "valid_targets_mean": 3105.9, + "valid_targets_min": 376 + }, + { + "epoch": 6.4423076923076925, + "grad_norm": 0.6564343007785355, + "learning_rate": 7.729069914209409e-07, + "loss": 0.1846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2291739583015442, + "step": 4020, + "valid_targets_mean": 3057.9, + "valid_targets_min": 1331 + }, + { + "epoch": 6.450320512820513, + "grad_norm": 0.5128523308822036, + "learning_rate": 7.510579777379345e-07, + "loss": 0.1749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16211318969726562, + "step": 4025, + "valid_targets_mean": 3331.9, + "valid_targets_min": 1713 + }, + { + "epoch": 6.458333333333333, + "grad_norm": 0.6207413303214334, + "learning_rate": 7.295163188216792e-07, + "loss": 0.1829, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21440479159355164, + "step": 4030, + "valid_targets_mean": 3360.0, + "valid_targets_min": 1856 + }, + { + "epoch": 6.466346153846154, + "grad_norm": 0.46974688232532347, + "learning_rate": 7.08282358635568e-07, + "loss": 0.1787, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16598930954933167, + "step": 4035, + "valid_targets_mean": 4054.6, + "valid_targets_min": 1918 + }, + { + "epoch": 6.4743589743589745, + "grad_norm": 0.5760375802259009, + "learning_rate": 6.87356436229869e-07, + "loss": 0.1868, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1884935200214386, + "step": 4040, + "valid_targets_mean": 2843.3, + "valid_targets_min": 272 + }, + { + "epoch": 6.482371794871795, + "grad_norm": 0.4862966080647505, + "learning_rate": 6.667388857362977e-07, + "loss": 0.1864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20953023433685303, + "step": 4045, + "valid_targets_mean": 4378.8, + "valid_targets_min": 2179 + }, + { + "epoch": 6.490384615384615, + "grad_norm": 0.48237548817545783, + "learning_rate": 6.464300363626797e-07, + "loss": 0.192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20548981428146362, + "step": 4050, + "valid_targets_mean": 4769.5, + "valid_targets_min": 1832 + }, + { + "epoch": 6.498397435897436, + "grad_norm": 0.5504436514800234, + "learning_rate": 6.264302123877053e-07, + "loss": 0.1937, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17251989245414734, + "step": 4055, + "valid_targets_mean": 3314.0, + "valid_targets_min": 1224 + }, + { + "epoch": 6.506410256410256, + "grad_norm": 0.6176400060173131, + "learning_rate": 6.067397331557412e-07, + "loss": 0.1832, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21772697567939758, + "step": 4060, + "valid_targets_mean": 3025.1, + "valid_targets_min": 1704 + }, + { + "epoch": 6.514423076923077, + "grad_norm": 0.6128183784325526, + "learning_rate": 5.873589130717405e-07, + "loss": 0.1742, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18539603054523468, + "step": 4065, + "valid_targets_mean": 3641.0, + "valid_targets_min": 1606 + }, + { + "epoch": 6.522435897435898, + "grad_norm": 0.5869844615039573, + "learning_rate": 5.682880615962116e-07, + "loss": 0.1842, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18697796761989594, + "step": 4070, + "valid_targets_mean": 3250.5, + "valid_targets_min": 1619 + }, + { + "epoch": 6.530448717948718, + "grad_norm": 0.45395568910065465, + "learning_rate": 5.495274832402841e-07, + "loss": 0.1878, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18877571821212769, + "step": 4075, + "valid_targets_mean": 4260.9, + "valid_targets_min": 318 + }, + { + "epoch": 6.538461538461538, + "grad_norm": 0.5811382204791522, + "learning_rate": 5.310774775608529e-07, + "loss": 0.1987, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20900984108448029, + "step": 4080, + "valid_targets_mean": 3326.0, + "valid_targets_min": 1659 + }, + { + "epoch": 6.546474358974359, + "grad_norm": 0.5821174706289566, + "learning_rate": 5.129383391557751e-07, + "loss": 0.1826, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21676848828792572, + "step": 4085, + "valid_targets_mean": 3183.1, + "valid_targets_min": 1476 + }, + { + "epoch": 6.55448717948718, + "grad_norm": 0.6460932248991629, + "learning_rate": 4.951103576591876e-07, + "loss": 0.1724, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18827717006206512, + "step": 4090, + "valid_targets_mean": 2421.1, + "valid_targets_min": 410 + }, + { + "epoch": 6.5625, + "grad_norm": 0.49037830926652504, + "learning_rate": 4.7759381773687e-07, + "loss": 0.2052, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21589890122413635, + "step": 4095, + "valid_targets_mean": 4651.1, + "valid_targets_min": 605 + }, + { + "epoch": 6.57051282051282, + "grad_norm": 0.5084810119478969, + "learning_rate": 4.6038899908170234e-07, + "loss": 0.1703, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15555456280708313, + "step": 4100, + "valid_targets_mean": 3502.8, + "valid_targets_min": 702 + }, + { + "epoch": 6.578525641025641, + "grad_norm": 0.5431185847935183, + "learning_rate": 4.4349617640920164e-07, + "loss": 0.2007, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21165643632411957, + "step": 4105, + "valid_targets_mean": 3681.2, + "valid_targets_min": 2204 + }, + { + "epoch": 6.586538461538462, + "grad_norm": 0.5911359764831849, + "learning_rate": 4.2691561945312764e-07, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18048742413520813, + "step": 4110, + "valid_targets_mean": 2909.3, + "valid_targets_min": 1491 + }, + { + "epoch": 6.594551282051282, + "grad_norm": 0.6130091429742942, + "learning_rate": 4.106475929611886e-07, + "loss": 0.183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19334766268730164, + "step": 4115, + "valid_targets_mean": 3284.5, + "valid_targets_min": 266 + }, + { + "epoch": 6.602564102564102, + "grad_norm": 0.6142827483556337, + "learning_rate": 3.9469235669080007e-07, + "loss": 0.1915, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20154422521591187, + "step": 4120, + "valid_targets_mean": 2889.7, + "valid_targets_min": 294 + }, + { + "epoch": 6.610576923076923, + "grad_norm": 0.6990866949863356, + "learning_rate": 3.7905016540495053e-07, + "loss": 0.2014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24647264182567596, + "step": 4125, + "valid_targets_mean": 2540.8, + "valid_targets_min": 406 + }, + { + "epoch": 6.618589743589744, + "grad_norm": 0.46263509014103377, + "learning_rate": 3.63721268868118e-07, + "loss": 0.1769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.158363938331604, + "step": 4130, + "valid_targets_mean": 4078.2, + "valid_targets_min": 1028 + }, + { + "epoch": 6.626602564102564, + "grad_norm": 0.5032941205599963, + "learning_rate": 3.487059118422997e-07, + "loss": 0.1817, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15640424191951752, + "step": 4135, + "valid_targets_mean": 3854.6, + "valid_targets_min": 1561 + }, + { + "epoch": 6.634615384615385, + "grad_norm": 0.4887526857319112, + "learning_rate": 3.3400433408308895e-07, + "loss": 0.1737, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16593563556671143, + "step": 4140, + "valid_targets_mean": 3915.2, + "valid_targets_min": 1696 + }, + { + "epoch": 6.642628205128205, + "grad_norm": 0.5614911252396141, + "learning_rate": 3.196167703358577e-07, + "loss": 0.1709, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18777693808078766, + "step": 4145, + "valid_targets_mean": 3050.1, + "valid_targets_min": 320 + }, + { + "epoch": 6.6506410256410255, + "grad_norm": 0.5358038811905682, + "learning_rate": 3.0554345033199985e-07, + "loss": 0.1877, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18092574179172516, + "step": 4150, + "valid_targets_mean": 3627.1, + "valid_targets_min": 1736 + }, + { + "epoch": 6.658653846153846, + "grad_norm": 0.5716438360594936, + "learning_rate": 2.917845987852652e-07, + "loss": 0.1948, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17830929160118103, + "step": 4155, + "valid_targets_mean": 3280.9, + "valid_targets_min": 1690 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.5430177303430579, + "learning_rate": 2.783404353881758e-07, + "loss": 0.1626, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16124941408634186, + "step": 4160, + "valid_targets_mean": 3723.2, + "valid_targets_min": 2203 + }, + { + "epoch": 6.674679487179487, + "grad_norm": 0.5549550586423444, + "learning_rate": 2.652111748085151e-07, + "loss": 0.1939, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1879281997680664, + "step": 4165, + "valid_targets_mean": 3579.1, + "valid_targets_min": 1680 + }, + { + "epoch": 6.6826923076923075, + "grad_norm": 0.5422698331625211, + "learning_rate": 2.523970266859044e-07, + "loss": 0.1857, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18413355946540833, + "step": 4170, + "valid_targets_mean": 3527.7, + "valid_targets_min": 272 + }, + { + "epoch": 6.690705128205128, + "grad_norm": 0.5692580709391031, + "learning_rate": 2.398981956284363e-07, + "loss": 0.1855, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24329707026481628, + "step": 4175, + "valid_targets_mean": 3685.5, + "valid_targets_min": 1907 + }, + { + "epoch": 6.698717948717949, + "grad_norm": 0.4258839271286024, + "learning_rate": 2.2771488120944207e-07, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15105578303337097, + "step": 4180, + "valid_targets_mean": 4621.4, + "valid_targets_min": 250 + }, + { + "epoch": 6.706730769230769, + "grad_norm": 0.5449932476686346, + "learning_rate": 2.1584727796427174e-07, + "loss": 0.1741, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16315647959709167, + "step": 4185, + "valid_targets_mean": 3350.1, + "valid_targets_min": 2204 + }, + { + "epoch": 6.714743589743589, + "grad_norm": 0.4646657622725054, + "learning_rate": 2.0429557538720556e-07, + "loss": 0.172, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14010201394557953, + "step": 4190, + "valid_targets_mean": 3919.2, + "valid_targets_min": 1435 + }, + { + "epoch": 6.722756410256411, + "grad_norm": 0.5246048209942255, + "learning_rate": 1.930599579284298e-07, + "loss": 0.1805, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1596519947052002, + "step": 4195, + "valid_targets_mean": 3963.7, + "valid_targets_min": 2061 + }, + { + "epoch": 6.730769230769231, + "grad_norm": 0.5620833700746424, + "learning_rate": 1.8214060499107679e-07, + "loss": 0.1702, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17860868573188782, + "step": 4200, + "valid_targets_mean": 3272.2, + "valid_targets_min": 1982 + }, + { + "epoch": 6.738782051282051, + "grad_norm": 0.5754656693227905, + "learning_rate": 1.7153769092837614e-07, + "loss": 0.1844, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19689905643463135, + "step": 4205, + "valid_targets_mean": 3276.7, + "valid_targets_min": 1811 + }, + { + "epoch": 6.746794871794872, + "grad_norm": 0.6155174380330322, + "learning_rate": 1.6125138504086146e-07, + "loss": 0.2014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2084151804447174, + "step": 4210, + "valid_targets_mean": 3323.5, + "valid_targets_min": 1285 + }, + { + "epoch": 6.7548076923076925, + "grad_norm": 0.5014874603446833, + "learning_rate": 1.5128185157367247e-07, + "loss": 0.1813, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17992736399173737, + "step": 4215, + "valid_targets_mean": 3556.7, + "valid_targets_min": 2075 + }, + { + "epoch": 6.762820512820513, + "grad_norm": 0.5670108496497538, + "learning_rate": 1.4162924971393044e-07, + "loss": 0.1861, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17421139776706696, + "step": 4220, + "valid_targets_mean": 3347.7, + "valid_targets_min": 1896 + }, + { + "epoch": 6.770833333333333, + "grad_norm": 0.45413689172272814, + "learning_rate": 1.322937335881891e-07, + "loss": 0.1791, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1506701111793518, + "step": 4225, + "valid_targets_mean": 3958.9, + "valid_targets_min": 389 + }, + { + "epoch": 6.778846153846154, + "grad_norm": 0.5306931473704816, + "learning_rate": 1.2327545225999215e-07, + "loss": 0.1801, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16012093424797058, + "step": 4230, + "valid_targets_mean": 3542.8, + "valid_targets_min": 1869 + }, + { + "epoch": 6.7868589743589745, + "grad_norm": 0.4597430728876461, + "learning_rate": 1.145745497274664e-07, + "loss": 0.1727, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1277555525302887, + "step": 4235, + "valid_targets_mean": 4133.0, + "valid_targets_min": 2238 + }, + { + "epoch": 6.794871794871795, + "grad_norm": 0.5289155251909863, + "learning_rate": 1.061911649210523e-07, + "loss": 0.1972, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1734294891357422, + "step": 4240, + "valid_targets_mean": 3650.6, + "valid_targets_min": 1200 + }, + { + "epoch": 6.802884615384615, + "grad_norm": 0.8185197904986652, + "learning_rate": 9.812543170126365e-08, + "loss": 0.2023, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18272753059864044, + "step": 4245, + "valid_targets_mean": 3363.8, + "valid_targets_min": 1712 + }, + { + "epoch": 6.810897435897436, + "grad_norm": 0.6406656101310069, + "learning_rate": 9.03774788565559e-08, + "loss": 0.1816, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21644482016563416, + "step": 4250, + "valid_targets_mean": 3481.6, + "valid_targets_min": 457 + }, + { + "epoch": 6.818910256410256, + "grad_norm": 0.5737761481772319, + "learning_rate": 8.294743010127448e-08, + "loss": 0.1963, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17666317522525787, + "step": 4255, + "valid_targets_mean": 3081.4, + "valid_targets_min": 1722 + }, + { + "epoch": 6.826923076923077, + "grad_norm": 0.5417175250426073, + "learning_rate": 7.583540407367418e-08, + "loss": 0.1782, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1564117670059204, + "step": 4260, + "valid_targets_mean": 3170.7, + "valid_targets_min": 455 + }, + { + "epoch": 6.834935897435898, + "grad_norm": 0.6257507553189467, + "learning_rate": 6.904151433402728e-08, + "loss": 0.1899, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22169390320777893, + "step": 4265, + "valid_targets_mean": 2890.9, + "valid_targets_min": 288 + }, + { + "epoch": 6.842948717948718, + "grad_norm": 0.5956634707447183, + "learning_rate": 6.256586936281172e-08, + "loss": 0.1814, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17920449376106262, + "step": 4270, + "valid_targets_mean": 2817.1, + "valid_targets_min": 430 + }, + { + "epoch": 6.850961538461538, + "grad_norm": 0.5901129809534669, + "learning_rate": 5.6408572558972475e-08, + "loss": 0.188, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18381944298744202, + "step": 4275, + "valid_targets_mean": 3248.1, + "valid_targets_min": 1796 + }, + { + "epoch": 6.858974358974359, + "grad_norm": 0.5375203922856862, + "learning_rate": 5.0569722238280605e-08, + "loss": 0.1864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16048292815685272, + "step": 4280, + "valid_targets_mean": 3052.4, + "valid_targets_min": 1575 + }, + { + "epoch": 6.86698717948718, + "grad_norm": 0.4429837357031785, + "learning_rate": 4.504941163175236e-08, + "loss": 0.1959, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1303357481956482, + "step": 4285, + "valid_targets_mean": 4013.2, + "valid_targets_min": 288 + }, + { + "epoch": 6.875, + "grad_norm": 0.46755542056895144, + "learning_rate": 3.984772888417032e-08, + "loss": 0.1992, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15034985542297363, + "step": 4290, + "valid_targets_mean": 3739.6, + "valid_targets_min": 2020 + }, + { + "epoch": 6.88301282051282, + "grad_norm": 0.5225117765429849, + "learning_rate": 3.4964757052671216e-08, + "loss": 0.1898, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16861560940742493, + "step": 4295, + "valid_targets_mean": 3642.2, + "valid_targets_min": 1808 + }, + { + "epoch": 6.891025641025641, + "grad_norm": 0.5708858038464354, + "learning_rate": 3.0400574105415856e-08, + "loss": 0.1668, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16104033589363098, + "step": 4300, + "valid_targets_mean": 3452.1, + "valid_targets_min": 273 + }, + { + "epoch": 6.899038461538462, + "grad_norm": 0.5338479864413854, + "learning_rate": 2.615525292035459e-08, + "loss": 0.2133, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19287021458148956, + "step": 4305, + "valid_targets_mean": 3645.0, + "valid_targets_min": 2051 + }, + { + "epoch": 6.907051282051282, + "grad_norm": 0.5011813250662843, + "learning_rate": 2.222886128405266e-08, + "loss": 0.177, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17692403495311737, + "step": 4310, + "valid_targets_mean": 3748.9, + "valid_targets_min": 274 + }, + { + "epoch": 6.915064102564102, + "grad_norm": 0.626908983487979, + "learning_rate": 1.8621461890617752e-08, + "loss": 0.1817, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17852284014225006, + "step": 4315, + "valid_targets_mean": 2626.8, + "valid_targets_min": 230 + }, + { + "epoch": 6.923076923076923, + "grad_norm": 0.5543654144483077, + "learning_rate": 1.5333112340687463e-08, + "loss": 0.1764, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16991391777992249, + "step": 4320, + "valid_targets_mean": 3477.0, + "valid_targets_min": 2240 + }, + { + "epoch": 6.931089743589744, + "grad_norm": 0.518706878581187, + "learning_rate": 1.2363865140518905e-08, + "loss": 0.1825, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17875301837921143, + "step": 4325, + "valid_targets_mean": 3591.9, + "valid_targets_min": 1830 + }, + { + "epoch": 6.939102564102564, + "grad_norm": 0.49639795431937256, + "learning_rate": 9.713767701151621e-09, + "loss": 0.1668, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16456924378871918, + "step": 4330, + "valid_targets_mean": 3846.6, + "valid_targets_min": 351 + }, + { + "epoch": 6.947115384615385, + "grad_norm": 0.561660195109291, + "learning_rate": 7.382862337641516e-09, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20046007633209229, + "step": 4335, + "valid_targets_mean": 3275.1, + "valid_targets_min": 1743 + }, + { + "epoch": 6.955128205128205, + "grad_norm": 0.5421335045005533, + "learning_rate": 5.371186268390283e-09, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1522393673658371, + "step": 4340, + "valid_targets_mean": 3741.4, + "valid_targets_min": 940 + }, + { + "epoch": 6.9631410256410255, + "grad_norm": 0.5325986102299718, + "learning_rate": 3.678771614550325e-09, + "loss": 0.1848, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18922027945518494, + "step": 4345, + "valid_targets_mean": 3629.2, + "valid_targets_min": 282 + }, + { + "epoch": 6.971153846153846, + "grad_norm": 0.4855368562058737, + "learning_rate": 2.3056453995162763e-09, + "loss": 0.1819, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17648158967494965, + "step": 4350, + "valid_targets_mean": 3896.1, + "valid_targets_min": 1649 + }, + { + "epoch": 6.979166666666667, + "grad_norm": 0.5169470715238487, + "learning_rate": 1.2518295484875708e-09, + "loss": 0.1879, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1792326271533966, + "step": 4355, + "valid_targets_mean": 3725.2, + "valid_targets_min": 377 + }, + { + "epoch": 6.987179487179487, + "grad_norm": 0.5051351410792214, + "learning_rate": 5.173408881198328e-10, + "loss": 0.1669, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18018975853919983, + "step": 4360, + "valid_targets_mean": 3765.9, + "valid_targets_min": 419 + }, + { + "epoch": 6.9951923076923075, + "grad_norm": 0.4793212594197615, + "learning_rate": 1.0219114625398263e-10, + "loss": 0.1811, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14409731328487396, + "step": 4365, + "valid_targets_mean": 3652.4, + "valid_targets_min": 1739 + }, + { + "epoch": 7.0, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1759420782327652, + "step": 4368, + "total_flos": 875542416195584.0, + "train_loss": 0.13199154654647405, + "train_runtime": 8877.0231, + "train_samples_per_second": 7.872, + "train_steps_per_second": 0.492, + "valid_targets_mean": 3122.6, + "valid_targets_min": 2000 + } + ], + "logging_steps": 5, + "max_steps": 4368, + "num_input_tokens_seen": 0, + "num_train_epochs": 7, + "save_steps": 1500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 875542416195584.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}