{ "best_metric": 0.47342389822006226, "best_model_checkpoint": "seizure_vit/seizure_vit_jlb_231108_iir_adjusted/checkpoint-1000", "epoch": 5.0, "eval_steps": 1000, "global_step": 14685, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 9.986380660537965e-06, "loss": 0.6425, "step": 20 }, { "epoch": 0.01, "learning_rate": 9.972761321075928e-06, "loss": 0.5846, "step": 40 }, { "epoch": 0.02, "learning_rate": 9.959141981613892e-06, "loss": 0.549, "step": 60 }, { "epoch": 0.03, "learning_rate": 9.945522642151857e-06, "loss": 0.5503, "step": 80 }, { "epoch": 0.03, "learning_rate": 9.931903302689821e-06, "loss": 0.5304, "step": 100 }, { "epoch": 0.04, "learning_rate": 9.918283963227783e-06, "loss": 0.5064, "step": 120 }, { "epoch": 0.05, "learning_rate": 9.904664623765749e-06, "loss": 0.4953, "step": 140 }, { "epoch": 0.05, "learning_rate": 9.891045284303712e-06, "loss": 0.5028, "step": 160 }, { "epoch": 0.06, "learning_rate": 9.877425944841676e-06, "loss": 0.4742, "step": 180 }, { "epoch": 0.07, "learning_rate": 9.86380660537964e-06, "loss": 0.4921, "step": 200 }, { "epoch": 0.07, "learning_rate": 9.850187265917604e-06, "loss": 0.4961, "step": 220 }, { "epoch": 0.08, "learning_rate": 9.836567926455567e-06, "loss": 0.438, "step": 240 }, { "epoch": 0.09, "learning_rate": 9.822948586993531e-06, "loss": 0.4273, "step": 260 }, { "epoch": 0.1, "learning_rate": 9.809329247531496e-06, "loss": 0.4606, "step": 280 }, { "epoch": 0.1, "learning_rate": 9.79570990806946e-06, "loss": 0.4513, "step": 300 }, { "epoch": 0.11, "learning_rate": 9.782090568607424e-06, "loss": 0.4375, "step": 320 }, { "epoch": 0.12, "learning_rate": 9.768471229145388e-06, "loss": 0.4441, "step": 340 }, { "epoch": 0.12, "learning_rate": 9.754851889683351e-06, "loss": 0.4187, "step": 360 }, { "epoch": 0.13, "learning_rate": 9.741232550221315e-06, "loss": 0.4336, "step": 380 }, { "epoch": 0.14, "learning_rate": 9.727613210759279e-06, "loss": 0.4258, "step": 400 }, { "epoch": 0.14, "learning_rate": 9.713993871297242e-06, "loss": 0.4225, "step": 420 }, { "epoch": 0.15, "learning_rate": 9.700374531835208e-06, "loss": 0.4119, "step": 440 }, { "epoch": 0.16, "learning_rate": 9.68675519237317e-06, "loss": 0.4016, "step": 460 }, { "epoch": 0.16, "learning_rate": 9.673135852911134e-06, "loss": 0.43, "step": 480 }, { "epoch": 0.17, "learning_rate": 9.659516513449099e-06, "loss": 0.4397, "step": 500 }, { "epoch": 0.18, "learning_rate": 9.645897173987063e-06, "loss": 0.376, "step": 520 }, { "epoch": 0.18, "learning_rate": 9.632277834525027e-06, "loss": 0.4556, "step": 540 }, { "epoch": 0.19, "learning_rate": 9.61865849506299e-06, "loss": 0.4067, "step": 560 }, { "epoch": 0.2, "learning_rate": 9.605039155600954e-06, "loss": 0.4005, "step": 580 }, { "epoch": 0.2, "learning_rate": 9.591419816138918e-06, "loss": 0.3618, "step": 600 }, { "epoch": 0.21, "learning_rate": 9.577800476676881e-06, "loss": 0.4302, "step": 620 }, { "epoch": 0.22, "learning_rate": 9.564181137214847e-06, "loss": 0.4234, "step": 640 }, { "epoch": 0.22, "learning_rate": 9.55056179775281e-06, "loss": 0.4304, "step": 660 }, { "epoch": 0.23, "learning_rate": 9.536942458290773e-06, "loss": 0.3656, "step": 680 }, { "epoch": 0.24, "learning_rate": 9.523323118828738e-06, "loss": 0.3883, "step": 700 }, { "epoch": 0.25, "learning_rate": 9.509703779366702e-06, "loss": 0.3971, "step": 720 }, { "epoch": 0.25, "learning_rate": 9.496084439904665e-06, "loss": 0.4155, "step": 740 }, { "epoch": 0.26, "learning_rate": 9.48246510044263e-06, "loss": 0.3724, "step": 760 }, { "epoch": 0.27, "learning_rate": 9.468845760980593e-06, "loss": 0.4199, "step": 780 }, { "epoch": 0.27, "learning_rate": 9.455226421518557e-06, "loss": 0.3955, "step": 800 }, { "epoch": 0.28, "learning_rate": 9.44160708205652e-06, "loss": 0.3703, "step": 820 }, { "epoch": 0.29, "learning_rate": 9.427987742594486e-06, "loss": 0.3956, "step": 840 }, { "epoch": 0.29, "learning_rate": 9.41436840313245e-06, "loss": 0.362, "step": 860 }, { "epoch": 0.3, "learning_rate": 9.400749063670412e-06, "loss": 0.3908, "step": 880 }, { "epoch": 0.31, "learning_rate": 9.387129724208377e-06, "loss": 0.3674, "step": 900 }, { "epoch": 0.31, "learning_rate": 9.37351038474634e-06, "loss": 0.3814, "step": 920 }, { "epoch": 0.32, "learning_rate": 9.359891045284304e-06, "loss": 0.4162, "step": 940 }, { "epoch": 0.33, "learning_rate": 9.346271705822268e-06, "loss": 0.4027, "step": 960 }, { "epoch": 0.33, "learning_rate": 9.332652366360232e-06, "loss": 0.3573, "step": 980 }, { "epoch": 0.34, "learning_rate": 9.319033026898197e-06, "loss": 0.3803, "step": 1000 }, { "epoch": 0.34, "eval_loss": 0.47342389822006226, "eval_roc_auc": 0.7746478873239437, "eval_runtime": 223.2127, "eval_samples_per_second": 66.797, "eval_steps_per_second": 8.351, "step": 1000 }, { "epoch": 0.35, "learning_rate": 9.30541368743616e-06, "loss": 0.3728, "step": 1020 }, { "epoch": 0.35, "learning_rate": 9.291794347974123e-06, "loss": 0.3511, "step": 1040 }, { "epoch": 0.36, "learning_rate": 9.278175008512089e-06, "loss": 0.3722, "step": 1060 }, { "epoch": 0.37, "learning_rate": 9.264555669050052e-06, "loss": 0.336, "step": 1080 }, { "epoch": 0.37, "learning_rate": 9.250936329588016e-06, "loss": 0.4234, "step": 1100 }, { "epoch": 0.38, "learning_rate": 9.23731699012598e-06, "loss": 0.3634, "step": 1120 }, { "epoch": 0.39, "learning_rate": 9.223697650663943e-06, "loss": 0.3378, "step": 1140 }, { "epoch": 0.39, "learning_rate": 9.210078311201907e-06, "loss": 0.3802, "step": 1160 }, { "epoch": 0.4, "learning_rate": 9.196458971739871e-06, "loss": 0.3998, "step": 1180 }, { "epoch": 0.41, "learning_rate": 9.182839632277836e-06, "loss": 0.3611, "step": 1200 }, { "epoch": 0.42, "learning_rate": 9.169220292815798e-06, "loss": 0.3884, "step": 1220 }, { "epoch": 0.42, "learning_rate": 9.155600953353762e-06, "loss": 0.3412, "step": 1240 }, { "epoch": 0.43, "learning_rate": 9.141981613891727e-06, "loss": 0.4026, "step": 1260 }, { "epoch": 0.44, "learning_rate": 9.128362274429691e-06, "loss": 0.3619, "step": 1280 }, { "epoch": 0.44, "learning_rate": 9.114742934967655e-06, "loss": 0.3291, "step": 1300 }, { "epoch": 0.45, "learning_rate": 9.101123595505619e-06, "loss": 0.3514, "step": 1320 }, { "epoch": 0.46, "learning_rate": 9.087504256043582e-06, "loss": 0.3353, "step": 1340 }, { "epoch": 0.46, "learning_rate": 9.073884916581546e-06, "loss": 0.3713, "step": 1360 }, { "epoch": 0.47, "learning_rate": 9.06026557711951e-06, "loss": 0.3571, "step": 1380 }, { "epoch": 0.48, "learning_rate": 9.046646237657475e-06, "loss": 0.3406, "step": 1400 }, { "epoch": 0.48, "learning_rate": 9.033026898195439e-06, "loss": 0.3785, "step": 1420 }, { "epoch": 0.49, "learning_rate": 9.019407558733401e-06, "loss": 0.3618, "step": 1440 }, { "epoch": 0.5, "learning_rate": 9.005788219271366e-06, "loss": 0.323, "step": 1460 }, { "epoch": 0.5, "learning_rate": 8.99216887980933e-06, "loss": 0.2908, "step": 1480 }, { "epoch": 0.51, "learning_rate": 8.978549540347294e-06, "loss": 0.372, "step": 1500 }, { "epoch": 0.52, "learning_rate": 8.964930200885258e-06, "loss": 0.3317, "step": 1520 }, { "epoch": 0.52, "learning_rate": 8.951310861423221e-06, "loss": 0.3953, "step": 1540 }, { "epoch": 0.53, "learning_rate": 8.937691521961187e-06, "loss": 0.3151, "step": 1560 }, { "epoch": 0.54, "learning_rate": 8.924072182499149e-06, "loss": 0.3391, "step": 1580 }, { "epoch": 0.54, "learning_rate": 8.910452843037113e-06, "loss": 0.3281, "step": 1600 }, { "epoch": 0.55, "learning_rate": 8.896833503575078e-06, "loss": 0.3613, "step": 1620 }, { "epoch": 0.56, "learning_rate": 8.883214164113042e-06, "loss": 0.3444, "step": 1640 }, { "epoch": 0.57, "learning_rate": 8.869594824651005e-06, "loss": 0.3122, "step": 1660 }, { "epoch": 0.57, "learning_rate": 8.855975485188969e-06, "loss": 0.2896, "step": 1680 }, { "epoch": 0.58, "learning_rate": 8.842356145726933e-06, "loss": 0.3603, "step": 1700 }, { "epoch": 0.59, "learning_rate": 8.828736806264897e-06, "loss": 0.3173, "step": 1720 }, { "epoch": 0.59, "learning_rate": 8.81511746680286e-06, "loss": 0.3016, "step": 1740 }, { "epoch": 0.6, "learning_rate": 8.801498127340826e-06, "loss": 0.3522, "step": 1760 }, { "epoch": 0.61, "learning_rate": 8.787878787878788e-06, "loss": 0.3482, "step": 1780 }, { "epoch": 0.61, "learning_rate": 8.774259448416752e-06, "loss": 0.2981, "step": 1800 }, { "epoch": 0.62, "learning_rate": 8.760640108954717e-06, "loss": 0.3467, "step": 1820 }, { "epoch": 0.63, "learning_rate": 8.74702076949268e-06, "loss": 0.373, "step": 1840 }, { "epoch": 0.63, "learning_rate": 8.733401430030644e-06, "loss": 0.3449, "step": 1860 }, { "epoch": 0.64, "learning_rate": 8.719782090568608e-06, "loss": 0.3541, "step": 1880 }, { "epoch": 0.65, "learning_rate": 8.706162751106572e-06, "loss": 0.3123, "step": 1900 }, { "epoch": 0.65, "learning_rate": 8.692543411644536e-06, "loss": 0.3103, "step": 1920 }, { "epoch": 0.66, "learning_rate": 8.6789240721825e-06, "loss": 0.3029, "step": 1940 }, { "epoch": 0.67, "learning_rate": 8.665304732720465e-06, "loss": 0.3077, "step": 1960 }, { "epoch": 0.67, "learning_rate": 8.651685393258428e-06, "loss": 0.3612, "step": 1980 }, { "epoch": 0.68, "learning_rate": 8.63806605379639e-06, "loss": 0.3456, "step": 2000 }, { "epoch": 0.68, "eval_loss": 0.4863223433494568, "eval_roc_auc": 0.7782025486250839, "eval_runtime": 226.9387, "eval_samples_per_second": 65.701, "eval_steps_per_second": 8.214, "step": 2000 }, { "epoch": 0.69, "learning_rate": 8.624446714334356e-06, "loss": 0.308, "step": 2020 }, { "epoch": 0.69, "learning_rate": 8.61082737487232e-06, "loss": 0.314, "step": 2040 }, { "epoch": 0.7, "learning_rate": 8.597208035410283e-06, "loss": 0.3102, "step": 2060 }, { "epoch": 0.71, "learning_rate": 8.583588695948247e-06, "loss": 0.3176, "step": 2080 }, { "epoch": 0.72, "learning_rate": 8.56996935648621e-06, "loss": 0.3541, "step": 2100 }, { "epoch": 0.72, "learning_rate": 8.556350017024175e-06, "loss": 0.3057, "step": 2120 }, { "epoch": 0.73, "learning_rate": 8.542730677562138e-06, "loss": 0.2985, "step": 2140 }, { "epoch": 0.74, "learning_rate": 8.529111338100104e-06, "loss": 0.2799, "step": 2160 }, { "epoch": 0.74, "learning_rate": 8.515491998638067e-06, "loss": 0.3107, "step": 2180 }, { "epoch": 0.75, "learning_rate": 8.501872659176031e-06, "loss": 0.3512, "step": 2200 }, { "epoch": 0.76, "learning_rate": 8.488253319713995e-06, "loss": 0.311, "step": 2220 }, { "epoch": 0.76, "learning_rate": 8.474633980251959e-06, "loss": 0.3146, "step": 2240 }, { "epoch": 0.77, "learning_rate": 8.461014640789922e-06, "loss": 0.3202, "step": 2260 }, { "epoch": 0.78, "learning_rate": 8.447395301327886e-06, "loss": 0.3073, "step": 2280 }, { "epoch": 0.78, "learning_rate": 8.43377596186585e-06, "loss": 0.3144, "step": 2300 }, { "epoch": 0.79, "learning_rate": 8.420156622403815e-06, "loss": 0.3132, "step": 2320 }, { "epoch": 0.8, "learning_rate": 8.406537282941777e-06, "loss": 0.3162, "step": 2340 }, { "epoch": 0.8, "learning_rate": 8.392917943479741e-06, "loss": 0.3047, "step": 2360 }, { "epoch": 0.81, "learning_rate": 8.379298604017706e-06, "loss": 0.3362, "step": 2380 }, { "epoch": 0.82, "learning_rate": 8.36567926455567e-06, "loss": 0.3045, "step": 2400 }, { "epoch": 0.82, "learning_rate": 8.352059925093634e-06, "loss": 0.301, "step": 2420 }, { "epoch": 0.83, "learning_rate": 8.338440585631598e-06, "loss": 0.3196, "step": 2440 }, { "epoch": 0.84, "learning_rate": 8.324821246169561e-06, "loss": 0.3542, "step": 2460 }, { "epoch": 0.84, "learning_rate": 8.311201906707525e-06, "loss": 0.3279, "step": 2480 }, { "epoch": 0.85, "learning_rate": 8.297582567245489e-06, "loss": 0.3157, "step": 2500 }, { "epoch": 0.86, "learning_rate": 8.283963227783454e-06, "loss": 0.2519, "step": 2520 }, { "epoch": 0.86, "learning_rate": 8.270343888321418e-06, "loss": 0.3058, "step": 2540 }, { "epoch": 0.87, "learning_rate": 8.25672454885938e-06, "loss": 0.2764, "step": 2560 }, { "epoch": 0.88, "learning_rate": 8.243105209397345e-06, "loss": 0.2877, "step": 2580 }, { "epoch": 0.89, "learning_rate": 8.229485869935309e-06, "loss": 0.3564, "step": 2600 }, { "epoch": 0.89, "learning_rate": 8.215866530473273e-06, "loss": 0.2914, "step": 2620 }, { "epoch": 0.9, "learning_rate": 8.202247191011237e-06, "loss": 0.2847, "step": 2640 }, { "epoch": 0.91, "learning_rate": 8.1886278515492e-06, "loss": 0.3405, "step": 2660 }, { "epoch": 0.91, "learning_rate": 8.175008512087164e-06, "loss": 0.3281, "step": 2680 }, { "epoch": 0.92, "learning_rate": 8.161389172625128e-06, "loss": 0.263, "step": 2700 }, { "epoch": 0.93, "learning_rate": 8.147769833163093e-06, "loss": 0.2724, "step": 2720 }, { "epoch": 0.93, "learning_rate": 8.134150493701057e-06, "loss": 0.342, "step": 2740 }, { "epoch": 0.94, "learning_rate": 8.120531154239019e-06, "loss": 0.3156, "step": 2760 }, { "epoch": 0.95, "learning_rate": 8.106911814776984e-06, "loss": 0.3101, "step": 2780 }, { "epoch": 0.95, "learning_rate": 8.093292475314948e-06, "loss": 0.3083, "step": 2800 }, { "epoch": 0.96, "learning_rate": 8.079673135852912e-06, "loss": 0.304, "step": 2820 }, { "epoch": 0.97, "learning_rate": 8.066053796390875e-06, "loss": 0.3087, "step": 2840 }, { "epoch": 0.97, "learning_rate": 8.05243445692884e-06, "loss": 0.3681, "step": 2860 }, { "epoch": 0.98, "learning_rate": 8.038815117466805e-06, "loss": 0.2919, "step": 2880 }, { "epoch": 0.99, "learning_rate": 8.025195778004767e-06, "loss": 0.3218, "step": 2900 }, { "epoch": 0.99, "learning_rate": 8.01157643854273e-06, "loss": 0.3012, "step": 2920 }, { "epoch": 1.0, "learning_rate": 7.997957099080696e-06, "loss": 0.2952, "step": 2940 }, { "epoch": 1.01, "learning_rate": 7.98433775961866e-06, "loss": 0.2597, "step": 2960 }, { "epoch": 1.01, "learning_rate": 7.970718420156623e-06, "loss": 0.2836, "step": 2980 }, { "epoch": 1.02, "learning_rate": 7.957099080694587e-06, "loss": 0.2831, "step": 3000 }, { "epoch": 1.02, "eval_loss": 0.48173409700393677, "eval_roc_auc": 0.7896713615023474, "eval_runtime": 226.5392, "eval_samples_per_second": 65.816, "eval_steps_per_second": 8.228, "step": 3000 }, { "epoch": 1.03, "learning_rate": 7.94347974123255e-06, "loss": 0.2553, "step": 3020 }, { "epoch": 1.04, "learning_rate": 7.929860401770514e-06, "loss": 0.2786, "step": 3040 }, { "epoch": 1.04, "learning_rate": 7.916241062308478e-06, "loss": 0.2636, "step": 3060 }, { "epoch": 1.05, "learning_rate": 7.902621722846444e-06, "loss": 0.2512, "step": 3080 }, { "epoch": 1.06, "learning_rate": 7.889002383384407e-06, "loss": 0.2755, "step": 3100 }, { "epoch": 1.06, "learning_rate": 7.87538304392237e-06, "loss": 0.2738, "step": 3120 }, { "epoch": 1.07, "learning_rate": 7.861763704460335e-06, "loss": 0.2653, "step": 3140 }, { "epoch": 1.08, "learning_rate": 7.848144364998298e-06, "loss": 0.2944, "step": 3160 }, { "epoch": 1.08, "learning_rate": 7.834525025536262e-06, "loss": 0.2624, "step": 3180 }, { "epoch": 1.09, "learning_rate": 7.820905686074226e-06, "loss": 0.2323, "step": 3200 }, { "epoch": 1.1, "learning_rate": 7.80728634661219e-06, "loss": 0.2742, "step": 3220 }, { "epoch": 1.1, "learning_rate": 7.793667007150153e-06, "loss": 0.2747, "step": 3240 }, { "epoch": 1.11, "learning_rate": 7.780047667688117e-06, "loss": 0.2377, "step": 3260 }, { "epoch": 1.12, "learning_rate": 7.766428328226083e-06, "loss": 0.2602, "step": 3280 }, { "epoch": 1.12, "learning_rate": 7.752808988764046e-06, "loss": 0.2795, "step": 3300 }, { "epoch": 1.13, "learning_rate": 7.739189649302008e-06, "loss": 0.2344, "step": 3320 }, { "epoch": 1.14, "learning_rate": 7.725570309839974e-06, "loss": 0.2519, "step": 3340 }, { "epoch": 1.14, "learning_rate": 7.711950970377937e-06, "loss": 0.2342, "step": 3360 }, { "epoch": 1.15, "learning_rate": 7.698331630915901e-06, "loss": 0.2442, "step": 3380 }, { "epoch": 1.16, "learning_rate": 7.684712291453865e-06, "loss": 0.2626, "step": 3400 }, { "epoch": 1.16, "learning_rate": 7.671092951991829e-06, "loss": 0.2381, "step": 3420 }, { "epoch": 1.17, "learning_rate": 7.657473612529794e-06, "loss": 0.2341, "step": 3440 }, { "epoch": 1.18, "learning_rate": 7.643854273067756e-06, "loss": 0.2994, "step": 3460 }, { "epoch": 1.18, "learning_rate": 7.63023493360572e-06, "loss": 0.2759, "step": 3480 }, { "epoch": 1.19, "learning_rate": 7.616615594143684e-06, "loss": 0.2927, "step": 3500 }, { "epoch": 1.2, "learning_rate": 7.602996254681648e-06, "loss": 0.2327, "step": 3520 }, { "epoch": 1.21, "learning_rate": 7.589376915219613e-06, "loss": 0.3248, "step": 3540 }, { "epoch": 1.21, "learning_rate": 7.5757575757575764e-06, "loss": 0.2836, "step": 3560 }, { "epoch": 1.22, "learning_rate": 7.56213823629554e-06, "loss": 0.2545, "step": 3580 }, { "epoch": 1.23, "learning_rate": 7.548518896833505e-06, "loss": 0.2596, "step": 3600 }, { "epoch": 1.23, "learning_rate": 7.534899557371468e-06, "loss": 0.2308, "step": 3620 }, { "epoch": 1.24, "learning_rate": 7.521280217909432e-06, "loss": 0.2462, "step": 3640 }, { "epoch": 1.25, "learning_rate": 7.507660878447396e-06, "loss": 0.2763, "step": 3660 }, { "epoch": 1.25, "learning_rate": 7.49404153898536e-06, "loss": 0.271, "step": 3680 }, { "epoch": 1.26, "learning_rate": 7.480422199523324e-06, "loss": 0.2717, "step": 3700 }, { "epoch": 1.27, "learning_rate": 7.466802860061287e-06, "loss": 0.2995, "step": 3720 }, { "epoch": 1.27, "learning_rate": 7.453183520599252e-06, "loss": 0.2812, "step": 3740 }, { "epoch": 1.28, "learning_rate": 7.439564181137215e-06, "loss": 0.2863, "step": 3760 }, { "epoch": 1.29, "learning_rate": 7.425944841675179e-06, "loss": 0.2468, "step": 3780 }, { "epoch": 1.29, "learning_rate": 7.412325502213144e-06, "loss": 0.2542, "step": 3800 }, { "epoch": 1.3, "learning_rate": 7.3987061627511066e-06, "loss": 0.2439, "step": 3820 }, { "epoch": 1.31, "learning_rate": 7.385086823289071e-06, "loss": 0.2923, "step": 3840 }, { "epoch": 1.31, "learning_rate": 7.371467483827035e-06, "loss": 0.2809, "step": 3860 }, { "epoch": 1.32, "learning_rate": 7.357848144364999e-06, "loss": 0.2601, "step": 3880 }, { "epoch": 1.33, "learning_rate": 7.344228804902963e-06, "loss": 0.2481, "step": 3900 }, { "epoch": 1.33, "learning_rate": 7.330609465440927e-06, "loss": 0.2886, "step": 3920 }, { "epoch": 1.34, "learning_rate": 7.3169901259788915e-06, "loss": 0.2546, "step": 3940 }, { "epoch": 1.35, "learning_rate": 7.303370786516854e-06, "loss": 0.2393, "step": 3960 }, { "epoch": 1.36, "learning_rate": 7.289751447054818e-06, "loss": 0.2839, "step": 3980 }, { "epoch": 1.36, "learning_rate": 7.276132107592783e-06, "loss": 0.2781, "step": 4000 }, { "epoch": 1.36, "eval_loss": 0.5417753458023071, "eval_roc_auc": 0.7655935613682091, "eval_runtime": 226.2172, "eval_samples_per_second": 65.91, "eval_steps_per_second": 8.24, "step": 4000 }, { "epoch": 1.37, "learning_rate": 7.262512768130746e-06, "loss": 0.2696, "step": 4020 }, { "epoch": 1.38, "learning_rate": 7.248893428668711e-06, "loss": 0.2458, "step": 4040 }, { "epoch": 1.38, "learning_rate": 7.235274089206674e-06, "loss": 0.269, "step": 4060 }, { "epoch": 1.39, "learning_rate": 7.2216547497446376e-06, "loss": 0.2798, "step": 4080 }, { "epoch": 1.4, "learning_rate": 7.208035410282602e-06, "loss": 0.2501, "step": 4100 }, { "epoch": 1.4, "learning_rate": 7.194416070820566e-06, "loss": 0.2655, "step": 4120 }, { "epoch": 1.41, "learning_rate": 7.180796731358529e-06, "loss": 0.2207, "step": 4140 }, { "epoch": 1.42, "learning_rate": 7.167177391896494e-06, "loss": 0.2381, "step": 4160 }, { "epoch": 1.42, "learning_rate": 7.153558052434457e-06, "loss": 0.2929, "step": 4180 }, { "epoch": 1.43, "learning_rate": 7.139938712972422e-06, "loss": 0.2455, "step": 4200 }, { "epoch": 1.44, "learning_rate": 7.126319373510385e-06, "loss": 0.2443, "step": 4220 }, { "epoch": 1.44, "learning_rate": 7.112700034048349e-06, "loss": 0.2439, "step": 4240 }, { "epoch": 1.45, "learning_rate": 7.099080694586314e-06, "loss": 0.2799, "step": 4260 }, { "epoch": 1.46, "learning_rate": 7.0854613551242765e-06, "loss": 0.2492, "step": 4280 }, { "epoch": 1.46, "learning_rate": 7.071842015662241e-06, "loss": 0.281, "step": 4300 }, { "epoch": 1.47, "learning_rate": 7.058222676200205e-06, "loss": 0.2539, "step": 4320 }, { "epoch": 1.48, "learning_rate": 7.0446033367381686e-06, "loss": 0.2464, "step": 4340 }, { "epoch": 1.48, "learning_rate": 7.030983997276133e-06, "loss": 0.2929, "step": 4360 }, { "epoch": 1.49, "learning_rate": 7.017364657814096e-06, "loss": 0.2679, "step": 4380 }, { "epoch": 1.5, "learning_rate": 7.003745318352061e-06, "loss": 0.2155, "step": 4400 }, { "epoch": 1.5, "learning_rate": 6.990125978890024e-06, "loss": 0.247, "step": 4420 }, { "epoch": 1.51, "learning_rate": 6.976506639427988e-06, "loss": 0.2069, "step": 4440 }, { "epoch": 1.52, "learning_rate": 6.962887299965953e-06, "loss": 0.2544, "step": 4460 }, { "epoch": 1.53, "learning_rate": 6.949267960503916e-06, "loss": 0.2162, "step": 4480 }, { "epoch": 1.53, "learning_rate": 6.935648621041881e-06, "loss": 0.252, "step": 4500 }, { "epoch": 1.54, "learning_rate": 6.922029281579844e-06, "loss": 0.2548, "step": 4520 }, { "epoch": 1.55, "learning_rate": 6.9084099421178075e-06, "loss": 0.27, "step": 4540 }, { "epoch": 1.55, "learning_rate": 6.894790602655772e-06, "loss": 0.2314, "step": 4560 }, { "epoch": 1.56, "learning_rate": 6.881171263193736e-06, "loss": 0.2381, "step": 4580 }, { "epoch": 1.57, "learning_rate": 6.8675519237317e-06, "loss": 0.2739, "step": 4600 }, { "epoch": 1.57, "learning_rate": 6.853932584269663e-06, "loss": 0.2185, "step": 4620 }, { "epoch": 1.58, "learning_rate": 6.840313244807627e-06, "loss": 0.2458, "step": 4640 }, { "epoch": 1.59, "learning_rate": 6.826693905345592e-06, "loss": 0.3119, "step": 4660 }, { "epoch": 1.59, "learning_rate": 6.813074565883555e-06, "loss": 0.2082, "step": 4680 }, { "epoch": 1.6, "learning_rate": 6.799455226421518e-06, "loss": 0.2772, "step": 4700 }, { "epoch": 1.61, "learning_rate": 6.785835886959483e-06, "loss": 0.238, "step": 4720 }, { "epoch": 1.61, "learning_rate": 6.7722165474974465e-06, "loss": 0.2403, "step": 4740 }, { "epoch": 1.62, "learning_rate": 6.758597208035411e-06, "loss": 0.2496, "step": 4760 }, { "epoch": 1.63, "learning_rate": 6.744977868573375e-06, "loss": 0.258, "step": 4780 }, { "epoch": 1.63, "learning_rate": 6.7313585291113385e-06, "loss": 0.2462, "step": 4800 }, { "epoch": 1.64, "learning_rate": 6.717739189649303e-06, "loss": 0.3012, "step": 4820 }, { "epoch": 1.65, "learning_rate": 6.704119850187266e-06, "loss": 0.2331, "step": 4840 }, { "epoch": 1.65, "learning_rate": 6.6905005107252305e-06, "loss": 0.2477, "step": 4860 }, { "epoch": 1.66, "learning_rate": 6.676881171263194e-06, "loss": 0.234, "step": 4880 }, { "epoch": 1.67, "learning_rate": 6.663261831801158e-06, "loss": 0.2434, "step": 4900 }, { "epoch": 1.68, "learning_rate": 6.649642492339123e-06, "loss": 0.3011, "step": 4920 }, { "epoch": 1.68, "learning_rate": 6.6360231528770855e-06, "loss": 0.2587, "step": 4940 }, { "epoch": 1.69, "learning_rate": 6.62240381341505e-06, "loss": 0.2252, "step": 4960 }, { "epoch": 1.7, "learning_rate": 6.608784473953014e-06, "loss": 0.2443, "step": 4980 }, { "epoch": 1.7, "learning_rate": 6.5951651344909775e-06, "loss": 0.2355, "step": 5000 }, { "epoch": 1.7, "eval_loss": 0.5398261547088623, "eval_roc_auc": 0.7786049631120053, "eval_runtime": 224.1313, "eval_samples_per_second": 66.524, "eval_steps_per_second": 8.317, "step": 5000 }, { "epoch": 1.71, "learning_rate": 6.581545795028942e-06, "loss": 0.2645, "step": 5020 }, { "epoch": 1.72, "learning_rate": 6.567926455566905e-06, "loss": 0.2447, "step": 5040 }, { "epoch": 1.72, "learning_rate": 6.5543071161048695e-06, "loss": 0.234, "step": 5060 }, { "epoch": 1.73, "learning_rate": 6.540687776642833e-06, "loss": 0.2343, "step": 5080 }, { "epoch": 1.74, "learning_rate": 6.527068437180797e-06, "loss": 0.2592, "step": 5100 }, { "epoch": 1.74, "learning_rate": 6.5134490977187615e-06, "loss": 0.2689, "step": 5120 }, { "epoch": 1.75, "learning_rate": 6.499829758256725e-06, "loss": 0.2535, "step": 5140 }, { "epoch": 1.76, "learning_rate": 6.48621041879469e-06, "loss": 0.2059, "step": 5160 }, { "epoch": 1.76, "learning_rate": 6.472591079332653e-06, "loss": 0.2646, "step": 5180 }, { "epoch": 1.77, "learning_rate": 6.4589717398706165e-06, "loss": 0.2203, "step": 5200 }, { "epoch": 1.78, "learning_rate": 6.445352400408581e-06, "loss": 0.2413, "step": 5220 }, { "epoch": 1.78, "learning_rate": 6.431733060946545e-06, "loss": 0.2399, "step": 5240 }, { "epoch": 1.79, "learning_rate": 6.418113721484509e-06, "loss": 0.2313, "step": 5260 }, { "epoch": 1.8, "learning_rate": 6.404494382022472e-06, "loss": 0.2082, "step": 5280 }, { "epoch": 1.8, "learning_rate": 6.390875042560436e-06, "loss": 0.2555, "step": 5300 }, { "epoch": 1.81, "learning_rate": 6.3772557030984005e-06, "loss": 0.2664, "step": 5320 }, { "epoch": 1.82, "learning_rate": 6.363636363636364e-06, "loss": 0.2367, "step": 5340 }, { "epoch": 1.82, "learning_rate": 6.350017024174327e-06, "loss": 0.2643, "step": 5360 }, { "epoch": 1.83, "learning_rate": 6.336397684712292e-06, "loss": 0.1894, "step": 5380 }, { "epoch": 1.84, "learning_rate": 6.322778345250255e-06, "loss": 0.2024, "step": 5400 }, { "epoch": 1.85, "learning_rate": 6.30915900578822e-06, "loss": 0.2293, "step": 5420 }, { "epoch": 1.85, "learning_rate": 6.295539666326184e-06, "loss": 0.2299, "step": 5440 }, { "epoch": 1.86, "learning_rate": 6.2819203268641474e-06, "loss": 0.2658, "step": 5460 }, { "epoch": 1.87, "learning_rate": 6.268300987402112e-06, "loss": 0.2577, "step": 5480 }, { "epoch": 1.87, "learning_rate": 6.254681647940075e-06, "loss": 0.2487, "step": 5500 }, { "epoch": 1.88, "learning_rate": 6.2410623084780395e-06, "loss": 0.2413, "step": 5520 }, { "epoch": 1.89, "learning_rate": 6.227442969016003e-06, "loss": 0.2544, "step": 5540 }, { "epoch": 1.89, "learning_rate": 6.213823629553967e-06, "loss": 0.2531, "step": 5560 }, { "epoch": 1.9, "learning_rate": 6.2002042900919315e-06, "loss": 0.293, "step": 5580 }, { "epoch": 1.91, "learning_rate": 6.186584950629894e-06, "loss": 0.2563, "step": 5600 }, { "epoch": 1.91, "learning_rate": 6.172965611167859e-06, "loss": 0.2549, "step": 5620 }, { "epoch": 1.92, "learning_rate": 6.159346271705823e-06, "loss": 0.2166, "step": 5640 }, { "epoch": 1.93, "learning_rate": 6.145726932243786e-06, "loss": 0.2084, "step": 5660 }, { "epoch": 1.93, "learning_rate": 6.132107592781751e-06, "loss": 0.2599, "step": 5680 }, { "epoch": 1.94, "learning_rate": 6.118488253319715e-06, "loss": 0.2266, "step": 5700 }, { "epoch": 1.95, "learning_rate": 6.104868913857679e-06, "loss": 0.2363, "step": 5720 }, { "epoch": 1.95, "learning_rate": 6.091249574395642e-06, "loss": 0.2586, "step": 5740 }, { "epoch": 1.96, "learning_rate": 6.077630234933606e-06, "loss": 0.2428, "step": 5760 }, { "epoch": 1.97, "learning_rate": 6.0640108954715705e-06, "loss": 0.269, "step": 5780 }, { "epoch": 1.97, "learning_rate": 6.050391556009534e-06, "loss": 0.2265, "step": 5800 }, { "epoch": 1.98, "learning_rate": 6.036772216547499e-06, "loss": 0.2594, "step": 5820 }, { "epoch": 1.99, "learning_rate": 6.023152877085462e-06, "loss": 0.2552, "step": 5840 }, { "epoch": 2.0, "learning_rate": 6.009533537623425e-06, "loss": 0.2279, "step": 5860 }, { "epoch": 2.0, "learning_rate": 5.99591419816139e-06, "loss": 0.2381, "step": 5880 }, { "epoch": 2.01, "learning_rate": 5.982294858699354e-06, "loss": 0.2271, "step": 5900 }, { "epoch": 2.02, "learning_rate": 5.968675519237318e-06, "loss": 0.1785, "step": 5920 }, { "epoch": 2.02, "learning_rate": 5.955056179775281e-06, "loss": 0.2066, "step": 5940 }, { "epoch": 2.03, "learning_rate": 5.941436840313245e-06, "loss": 0.219, "step": 5960 }, { "epoch": 2.04, "learning_rate": 5.9278175008512094e-06, "loss": 0.1896, "step": 5980 }, { "epoch": 2.04, "learning_rate": 5.914198161389173e-06, "loss": 0.1978, "step": 6000 }, { "epoch": 2.04, "eval_loss": 0.6120939254760742, "eval_roc_auc": 0.7648558014755198, "eval_runtime": 225.5238, "eval_samples_per_second": 66.113, "eval_steps_per_second": 8.265, "step": 6000 }, { "epoch": 2.05, "learning_rate": 5.900578821927137e-06, "loss": 0.1811, "step": 6020 }, { "epoch": 2.06, "learning_rate": 5.8869594824651015e-06, "loss": 0.2201, "step": 6040 }, { "epoch": 2.06, "learning_rate": 5.873340143003064e-06, "loss": 0.2118, "step": 6060 }, { "epoch": 2.07, "learning_rate": 5.859720803541029e-06, "loss": 0.1765, "step": 6080 }, { "epoch": 2.08, "learning_rate": 5.846101464078993e-06, "loss": 0.2063, "step": 6100 }, { "epoch": 2.08, "learning_rate": 5.832482124616956e-06, "loss": 0.2055, "step": 6120 }, { "epoch": 2.09, "learning_rate": 5.818862785154921e-06, "loss": 0.1919, "step": 6140 }, { "epoch": 2.1, "learning_rate": 5.805243445692884e-06, "loss": 0.1999, "step": 6160 }, { "epoch": 2.1, "learning_rate": 5.791624106230848e-06, "loss": 0.1761, "step": 6180 }, { "epoch": 2.11, "learning_rate": 5.778004766768812e-06, "loss": 0.1917, "step": 6200 }, { "epoch": 2.12, "learning_rate": 5.764385427306776e-06, "loss": 0.2086, "step": 6220 }, { "epoch": 2.12, "learning_rate": 5.7507660878447404e-06, "loss": 0.195, "step": 6240 }, { "epoch": 2.13, "learning_rate": 5.737146748382703e-06, "loss": 0.2244, "step": 6260 }, { "epoch": 2.14, "learning_rate": 5.723527408920668e-06, "loss": 0.2411, "step": 6280 }, { "epoch": 2.15, "learning_rate": 5.709908069458632e-06, "loss": 0.1984, "step": 6300 }, { "epoch": 2.15, "learning_rate": 5.696288729996595e-06, "loss": 0.1899, "step": 6320 }, { "epoch": 2.16, "learning_rate": 5.68266939053456e-06, "loss": 0.2241, "step": 6340 }, { "epoch": 2.17, "learning_rate": 5.669050051072524e-06, "loss": 0.2296, "step": 6360 }, { "epoch": 2.17, "learning_rate": 5.655430711610488e-06, "loss": 0.2082, "step": 6380 }, { "epoch": 2.18, "learning_rate": 5.641811372148451e-06, "loss": 0.1981, "step": 6400 }, { "epoch": 2.19, "learning_rate": 5.628192032686415e-06, "loss": 0.2306, "step": 6420 }, { "epoch": 2.19, "learning_rate": 5.614572693224379e-06, "loss": 0.1962, "step": 6440 }, { "epoch": 2.2, "learning_rate": 5.600953353762343e-06, "loss": 0.1773, "step": 6460 }, { "epoch": 2.21, "learning_rate": 5.587334014300308e-06, "loss": 0.2058, "step": 6480 }, { "epoch": 2.21, "learning_rate": 5.5737146748382706e-06, "loss": 0.1875, "step": 6500 }, { "epoch": 2.22, "learning_rate": 5.560095335376234e-06, "loss": 0.208, "step": 6520 }, { "epoch": 2.23, "learning_rate": 5.546475995914199e-06, "loss": 0.2189, "step": 6540 }, { "epoch": 2.23, "learning_rate": 5.532856656452163e-06, "loss": 0.1869, "step": 6560 }, { "epoch": 2.24, "learning_rate": 5.5192373169901255e-06, "loss": 0.1737, "step": 6580 }, { "epoch": 2.25, "learning_rate": 5.50561797752809e-06, "loss": 0.1807, "step": 6600 }, { "epoch": 2.25, "learning_rate": 5.491998638066054e-06, "loss": 0.2236, "step": 6620 }, { "epoch": 2.26, "learning_rate": 5.478379298604018e-06, "loss": 0.2033, "step": 6640 }, { "epoch": 2.27, "learning_rate": 5.464759959141982e-06, "loss": 0.177, "step": 6660 }, { "epoch": 2.27, "learning_rate": 5.451140619679946e-06, "loss": 0.1955, "step": 6680 }, { "epoch": 2.28, "learning_rate": 5.43752128021791e-06, "loss": 0.1657, "step": 6700 }, { "epoch": 2.29, "learning_rate": 5.423901940755873e-06, "loss": 0.2116, "step": 6720 }, { "epoch": 2.29, "learning_rate": 5.410282601293838e-06, "loss": 0.1991, "step": 6740 }, { "epoch": 2.3, "learning_rate": 5.3966632618318016e-06, "loss": 0.2085, "step": 6760 }, { "epoch": 2.31, "learning_rate": 5.383043922369765e-06, "loss": 0.2189, "step": 6780 }, { "epoch": 2.32, "learning_rate": 5.36942458290773e-06, "loss": 0.1781, "step": 6800 }, { "epoch": 2.32, "learning_rate": 5.355805243445693e-06, "loss": 0.192, "step": 6820 }, { "epoch": 2.33, "learning_rate": 5.342185903983657e-06, "loss": 0.1436, "step": 6840 }, { "epoch": 2.34, "learning_rate": 5.328566564521621e-06, "loss": 0.2369, "step": 6860 }, { "epoch": 2.34, "learning_rate": 5.314947225059585e-06, "loss": 0.1851, "step": 6880 }, { "epoch": 2.35, "learning_rate": 5.301327885597549e-06, "loss": 0.1813, "step": 6900 }, { "epoch": 2.36, "learning_rate": 5.287708546135513e-06, "loss": 0.1995, "step": 6920 }, { "epoch": 2.36, "learning_rate": 5.274089206673478e-06, "loss": 0.1771, "step": 6940 }, { "epoch": 2.37, "learning_rate": 5.2604698672114405e-06, "loss": 0.2012, "step": 6960 }, { "epoch": 2.38, "learning_rate": 5.246850527749404e-06, "loss": 0.2236, "step": 6980 }, { "epoch": 2.38, "learning_rate": 5.233231188287369e-06, "loss": 0.149, "step": 7000 }, { "epoch": 2.38, "eval_loss": 0.6401586532592773, "eval_roc_auc": 0.7706237424547283, "eval_runtime": 224.5054, "eval_samples_per_second": 66.413, "eval_steps_per_second": 8.303, "step": 7000 }, { "epoch": 2.39, "learning_rate": 5.2196118488253326e-06, "loss": 0.2141, "step": 7020 }, { "epoch": 2.4, "learning_rate": 5.205992509363297e-06, "loss": 0.1899, "step": 7040 }, { "epoch": 2.4, "learning_rate": 5.19237316990126e-06, "loss": 0.1737, "step": 7060 }, { "epoch": 2.41, "learning_rate": 5.178753830439224e-06, "loss": 0.182, "step": 7080 }, { "epoch": 2.42, "learning_rate": 5.165134490977188e-06, "loss": 0.1781, "step": 7100 }, { "epoch": 2.42, "learning_rate": 5.151515151515152e-06, "loss": 0.1758, "step": 7120 }, { "epoch": 2.43, "learning_rate": 5.137895812053117e-06, "loss": 0.2001, "step": 7140 }, { "epoch": 2.44, "learning_rate": 5.1242764725910795e-06, "loss": 0.1888, "step": 7160 }, { "epoch": 2.44, "learning_rate": 5.110657133129043e-06, "loss": 0.1871, "step": 7180 }, { "epoch": 2.45, "learning_rate": 5.097037793667008e-06, "loss": 0.2059, "step": 7200 }, { "epoch": 2.46, "learning_rate": 5.0834184542049715e-06, "loss": 0.2047, "step": 7220 }, { "epoch": 2.47, "learning_rate": 5.069799114742935e-06, "loss": 0.1985, "step": 7240 }, { "epoch": 2.47, "learning_rate": 5.0561797752809e-06, "loss": 0.2316, "step": 7260 }, { "epoch": 2.48, "learning_rate": 5.042560435818863e-06, "loss": 0.2107, "step": 7280 }, { "epoch": 2.49, "learning_rate": 5.028941096356827e-06, "loss": 0.1963, "step": 7300 }, { "epoch": 2.49, "learning_rate": 5.015321756894791e-06, "loss": 0.209, "step": 7320 }, { "epoch": 2.5, "learning_rate": 5.001702417432755e-06, "loss": 0.2221, "step": 7340 }, { "epoch": 2.51, "learning_rate": 4.988083077970719e-06, "loss": 0.1636, "step": 7360 }, { "epoch": 2.51, "learning_rate": 4.974463738508682e-06, "loss": 0.2369, "step": 7380 }, { "epoch": 2.52, "learning_rate": 4.960844399046647e-06, "loss": 0.2145, "step": 7400 }, { "epoch": 2.53, "learning_rate": 4.9472250595846105e-06, "loss": 0.1966, "step": 7420 }, { "epoch": 2.53, "learning_rate": 4.933605720122574e-06, "loss": 0.2195, "step": 7440 }, { "epoch": 2.54, "learning_rate": 4.919986380660539e-06, "loss": 0.1953, "step": 7460 }, { "epoch": 2.55, "learning_rate": 4.906367041198502e-06, "loss": 0.1959, "step": 7480 }, { "epoch": 2.55, "learning_rate": 4.892747701736466e-06, "loss": 0.1888, "step": 7500 }, { "epoch": 2.56, "learning_rate": 4.87912836227443e-06, "loss": 0.1777, "step": 7520 }, { "epoch": 2.57, "learning_rate": 4.8655090228123945e-06, "loss": 0.2161, "step": 7540 }, { "epoch": 2.57, "learning_rate": 4.851889683350358e-06, "loss": 0.2002, "step": 7560 }, { "epoch": 2.58, "learning_rate": 4.838270343888322e-06, "loss": 0.212, "step": 7580 }, { "epoch": 2.59, "learning_rate": 4.824651004426286e-06, "loss": 0.2396, "step": 7600 }, { "epoch": 2.59, "learning_rate": 4.8110316649642495e-06, "loss": 0.161, "step": 7620 }, { "epoch": 2.6, "learning_rate": 4.797412325502214e-06, "loss": 0.2192, "step": 7640 }, { "epoch": 2.61, "learning_rate": 4.783792986040177e-06, "loss": 0.1878, "step": 7660 }, { "epoch": 2.61, "learning_rate": 4.7701736465781415e-06, "loss": 0.2237, "step": 7680 }, { "epoch": 2.62, "learning_rate": 4.756554307116105e-06, "loss": 0.1654, "step": 7700 }, { "epoch": 2.63, "learning_rate": 4.742934967654069e-06, "loss": 0.1884, "step": 7720 }, { "epoch": 2.64, "learning_rate": 4.7293156281920335e-06, "loss": 0.2605, "step": 7740 }, { "epoch": 2.64, "learning_rate": 4.715696288729996e-06, "loss": 0.1947, "step": 7760 }, { "epoch": 2.65, "learning_rate": 4.702076949267961e-06, "loss": 0.2241, "step": 7780 }, { "epoch": 2.66, "learning_rate": 4.688457609805925e-06, "loss": 0.2135, "step": 7800 }, { "epoch": 2.66, "learning_rate": 4.674838270343888e-06, "loss": 0.1556, "step": 7820 }, { "epoch": 2.67, "learning_rate": 4.661218930881853e-06, "loss": 0.2006, "step": 7840 }, { "epoch": 2.68, "learning_rate": 4.647599591419817e-06, "loss": 0.168, "step": 7860 }, { "epoch": 2.68, "learning_rate": 4.6339802519577804e-06, "loss": 0.1939, "step": 7880 }, { "epoch": 2.69, "learning_rate": 4.620360912495744e-06, "loss": 0.1877, "step": 7900 }, { "epoch": 2.7, "learning_rate": 4.606741573033709e-06, "loss": 0.2254, "step": 7920 }, { "epoch": 2.7, "learning_rate": 4.5931222335716725e-06, "loss": 0.1725, "step": 7940 }, { "epoch": 2.71, "learning_rate": 4.579502894109636e-06, "loss": 0.2169, "step": 7960 }, { "epoch": 2.72, "learning_rate": 4.5658835546476e-06, "loss": 0.1552, "step": 7980 }, { "epoch": 2.72, "learning_rate": 4.552264215185564e-06, "loss": 0.1766, "step": 8000 }, { "epoch": 2.72, "eval_loss": 0.6767948865890503, "eval_roc_auc": 0.7610328638497652, "eval_runtime": 224.0971, "eval_samples_per_second": 66.534, "eval_steps_per_second": 8.318, "step": 8000 }, { "epoch": 2.73, "learning_rate": 4.538644875723528e-06, "loss": 0.1792, "step": 8020 }, { "epoch": 2.74, "learning_rate": 4.525025536261491e-06, "loss": 0.2031, "step": 8040 }, { "epoch": 2.74, "learning_rate": 4.511406196799456e-06, "loss": 0.1644, "step": 8060 }, { "epoch": 2.75, "learning_rate": 4.497786857337419e-06, "loss": 0.1895, "step": 8080 }, { "epoch": 2.76, "learning_rate": 4.484167517875383e-06, "loss": 0.1915, "step": 8100 }, { "epoch": 2.76, "learning_rate": 4.470548178413348e-06, "loss": 0.1828, "step": 8120 }, { "epoch": 2.77, "learning_rate": 4.456928838951311e-06, "loss": 0.2172, "step": 8140 }, { "epoch": 2.78, "learning_rate": 4.443309499489275e-06, "loss": 0.1817, "step": 8160 }, { "epoch": 2.79, "learning_rate": 4.429690160027239e-06, "loss": 0.1833, "step": 8180 }, { "epoch": 2.79, "learning_rate": 4.4160708205652035e-06, "loss": 0.1648, "step": 8200 }, { "epoch": 2.8, "learning_rate": 4.402451481103167e-06, "loss": 0.1905, "step": 8220 }, { "epoch": 2.81, "learning_rate": 4.388832141641131e-06, "loss": 0.1964, "step": 8240 }, { "epoch": 2.81, "learning_rate": 4.375212802179095e-06, "loss": 0.2237, "step": 8260 }, { "epoch": 2.82, "learning_rate": 4.361593462717058e-06, "loss": 0.1929, "step": 8280 }, { "epoch": 2.83, "learning_rate": 4.347974123255023e-06, "loss": 0.2196, "step": 8300 }, { "epoch": 2.83, "learning_rate": 4.334354783792986e-06, "loss": 0.1876, "step": 8320 }, { "epoch": 2.84, "learning_rate": 4.32073544433095e-06, "loss": 0.2213, "step": 8340 }, { "epoch": 2.85, "learning_rate": 4.307116104868914e-06, "loss": 0.2043, "step": 8360 }, { "epoch": 2.85, "learning_rate": 4.293496765406878e-06, "loss": 0.1865, "step": 8380 }, { "epoch": 2.86, "learning_rate": 4.2798774259448424e-06, "loss": 0.2067, "step": 8400 }, { "epoch": 2.87, "learning_rate": 4.266258086482805e-06, "loss": 0.1983, "step": 8420 }, { "epoch": 2.87, "learning_rate": 4.25263874702077e-06, "loss": 0.1724, "step": 8440 }, { "epoch": 2.88, "learning_rate": 4.239019407558734e-06, "loss": 0.2027, "step": 8460 }, { "epoch": 2.89, "learning_rate": 4.225400068096698e-06, "loss": 0.1854, "step": 8480 }, { "epoch": 2.89, "learning_rate": 4.211780728634662e-06, "loss": 0.1886, "step": 8500 }, { "epoch": 2.9, "learning_rate": 4.198161389172626e-06, "loss": 0.1882, "step": 8520 }, { "epoch": 2.91, "learning_rate": 4.184542049710589e-06, "loss": 0.221, "step": 8540 }, { "epoch": 2.91, "learning_rate": 4.170922710248553e-06, "loss": 0.1875, "step": 8560 }, { "epoch": 2.92, "learning_rate": 4.157303370786518e-06, "loss": 0.162, "step": 8580 }, { "epoch": 2.93, "learning_rate": 4.1436840313244805e-06, "loss": 0.1633, "step": 8600 }, { "epoch": 2.93, "learning_rate": 4.130064691862445e-06, "loss": 0.2134, "step": 8620 }, { "epoch": 2.94, "learning_rate": 4.116445352400409e-06, "loss": 0.2259, "step": 8640 }, { "epoch": 2.95, "learning_rate": 4.102826012938373e-06, "loss": 0.1672, "step": 8660 }, { "epoch": 2.96, "learning_rate": 4.089206673476337e-06, "loss": 0.1891, "step": 8680 }, { "epoch": 2.96, "learning_rate": 4.0755873340143e-06, "loss": 0.1633, "step": 8700 }, { "epoch": 2.97, "learning_rate": 4.061967994552265e-06, "loss": 0.192, "step": 8720 }, { "epoch": 2.98, "learning_rate": 4.048348655090228e-06, "loss": 0.1875, "step": 8740 }, { "epoch": 2.98, "learning_rate": 4.034729315628192e-06, "loss": 0.1698, "step": 8760 }, { "epoch": 2.99, "learning_rate": 4.021109976166157e-06, "loss": 0.1905, "step": 8780 }, { "epoch": 3.0, "learning_rate": 4.00749063670412e-06, "loss": 0.1735, "step": 8800 }, { "epoch": 3.0, "learning_rate": 3.993871297242084e-06, "loss": 0.1437, "step": 8820 }, { "epoch": 3.01, "learning_rate": 3.980251957780048e-06, "loss": 0.1843, "step": 8840 }, { "epoch": 3.02, "learning_rate": 3.966632618318012e-06, "loss": 0.1405, "step": 8860 }, { "epoch": 3.02, "learning_rate": 3.953013278855976e-06, "loss": 0.1279, "step": 8880 }, { "epoch": 3.03, "learning_rate": 3.93939393939394e-06, "loss": 0.1407, "step": 8900 }, { "epoch": 3.04, "learning_rate": 3.9257745999319036e-06, "loss": 0.1751, "step": 8920 }, { "epoch": 3.04, "learning_rate": 3.912155260469867e-06, "loss": 0.1507, "step": 8940 }, { "epoch": 3.05, "learning_rate": 3.898535921007832e-06, "loss": 0.1636, "step": 8960 }, { "epoch": 3.06, "learning_rate": 3.884916581545795e-06, "loss": 0.1631, "step": 8980 }, { "epoch": 3.06, "learning_rate": 3.871297242083759e-06, "loss": 0.1496, "step": 9000 }, { "epoch": 3.06, "eval_loss": 0.6239368915557861, "eval_roc_auc": 0.7733065057008719, "eval_runtime": 227.7765, "eval_samples_per_second": 65.459, "eval_steps_per_second": 8.183, "step": 9000 }, { "epoch": 3.07, "learning_rate": 3.857677902621723e-06, "loss": 0.1598, "step": 9020 }, { "epoch": 3.08, "learning_rate": 3.844058563159687e-06, "loss": 0.1247, "step": 9040 }, { "epoch": 3.08, "learning_rate": 3.830439223697651e-06, "loss": 0.1712, "step": 9060 }, { "epoch": 3.09, "learning_rate": 3.816819884235615e-06, "loss": 0.1623, "step": 9080 }, { "epoch": 3.1, "learning_rate": 3.803200544773579e-06, "loss": 0.1359, "step": 9100 }, { "epoch": 3.11, "learning_rate": 3.7895812053115425e-06, "loss": 0.1476, "step": 9120 }, { "epoch": 3.11, "learning_rate": 3.7759618658495067e-06, "loss": 0.1505, "step": 9140 }, { "epoch": 3.12, "learning_rate": 3.762342526387471e-06, "loss": 0.1779, "step": 9160 }, { "epoch": 3.13, "learning_rate": 3.748723186925434e-06, "loss": 0.1288, "step": 9180 }, { "epoch": 3.13, "learning_rate": 3.7351038474633983e-06, "loss": 0.1545, "step": 9200 }, { "epoch": 3.14, "learning_rate": 3.7214845080013624e-06, "loss": 0.1543, "step": 9220 }, { "epoch": 3.15, "learning_rate": 3.707865168539326e-06, "loss": 0.1486, "step": 9240 }, { "epoch": 3.15, "learning_rate": 3.69424582907729e-06, "loss": 0.1456, "step": 9260 }, { "epoch": 3.16, "learning_rate": 3.6806264896152536e-06, "loss": 0.1582, "step": 9280 }, { "epoch": 3.17, "learning_rate": 3.6670071501532178e-06, "loss": 0.1314, "step": 9300 }, { "epoch": 3.17, "learning_rate": 3.653387810691182e-06, "loss": 0.1784, "step": 9320 }, { "epoch": 3.18, "learning_rate": 3.639768471229146e-06, "loss": 0.1799, "step": 9340 }, { "epoch": 3.19, "learning_rate": 3.6261491317671094e-06, "loss": 0.1634, "step": 9360 }, { "epoch": 3.19, "learning_rate": 3.6125297923050735e-06, "loss": 0.1668, "step": 9380 }, { "epoch": 3.2, "learning_rate": 3.5989104528430373e-06, "loss": 0.1717, "step": 9400 }, { "epoch": 3.21, "learning_rate": 3.5852911133810014e-06, "loss": 0.1076, "step": 9420 }, { "epoch": 3.21, "learning_rate": 3.5716717739189656e-06, "loss": 0.1087, "step": 9440 }, { "epoch": 3.22, "learning_rate": 3.558052434456929e-06, "loss": 0.158, "step": 9460 }, { "epoch": 3.23, "learning_rate": 3.544433094994893e-06, "loss": 0.1229, "step": 9480 }, { "epoch": 3.23, "learning_rate": 3.530813755532857e-06, "loss": 0.1428, "step": 9500 }, { "epoch": 3.24, "learning_rate": 3.517194416070821e-06, "loss": 0.1513, "step": 9520 }, { "epoch": 3.25, "learning_rate": 3.5035750766087846e-06, "loss": 0.1302, "step": 9540 }, { "epoch": 3.26, "learning_rate": 3.4899557371467483e-06, "loss": 0.1416, "step": 9560 }, { "epoch": 3.26, "learning_rate": 3.4763363976847125e-06, "loss": 0.1926, "step": 9580 }, { "epoch": 3.27, "learning_rate": 3.4627170582226766e-06, "loss": 0.1431, "step": 9600 }, { "epoch": 3.28, "learning_rate": 3.449097718760641e-06, "loss": 0.1235, "step": 9620 }, { "epoch": 3.28, "learning_rate": 3.435478379298604e-06, "loss": 0.1897, "step": 9640 }, { "epoch": 3.29, "learning_rate": 3.4218590398365683e-06, "loss": 0.1464, "step": 9660 }, { "epoch": 3.3, "learning_rate": 3.408239700374532e-06, "loss": 0.1889, "step": 9680 }, { "epoch": 3.3, "learning_rate": 3.394620360912496e-06, "loss": 0.1379, "step": 9700 }, { "epoch": 3.31, "learning_rate": 3.3810010214504603e-06, "loss": 0.1677, "step": 9720 }, { "epoch": 3.32, "learning_rate": 3.3673816819884236e-06, "loss": 0.1461, "step": 9740 }, { "epoch": 3.32, "learning_rate": 3.3537623425263877e-06, "loss": 0.1661, "step": 9760 }, { "epoch": 3.33, "learning_rate": 3.340143003064352e-06, "loss": 0.1571, "step": 9780 }, { "epoch": 3.34, "learning_rate": 3.3265236636023156e-06, "loss": 0.157, "step": 9800 }, { "epoch": 3.34, "learning_rate": 3.3129043241402798e-06, "loss": 0.1774, "step": 9820 }, { "epoch": 3.35, "learning_rate": 3.299284984678243e-06, "loss": 0.1463, "step": 9840 }, { "epoch": 3.36, "learning_rate": 3.2856656452162072e-06, "loss": 0.1568, "step": 9860 }, { "epoch": 3.36, "learning_rate": 3.2720463057541714e-06, "loss": 0.1165, "step": 9880 }, { "epoch": 3.37, "learning_rate": 3.258426966292135e-06, "loss": 0.1414, "step": 9900 }, { "epoch": 3.38, "learning_rate": 3.244807626830099e-06, "loss": 0.133, "step": 9920 }, { "epoch": 3.38, "learning_rate": 3.231188287368063e-06, "loss": 0.1348, "step": 9940 }, { "epoch": 3.39, "learning_rate": 3.2175689479060267e-06, "loss": 0.1444, "step": 9960 }, { "epoch": 3.4, "learning_rate": 3.203949608443991e-06, "loss": 0.1566, "step": 9980 }, { "epoch": 3.4, "learning_rate": 3.190330268981955e-06, "loss": 0.155, "step": 10000 }, { "epoch": 3.4, "eval_loss": 0.7332798838615417, "eval_roc_auc": 0.7601609657947686, "eval_runtime": 222.7036, "eval_samples_per_second": 66.95, "eval_steps_per_second": 8.37, "step": 10000 }, { "epoch": 3.41, "learning_rate": 3.1767109295199183e-06, "loss": 0.1454, "step": 10020 }, { "epoch": 3.42, "learning_rate": 3.1630915900578825e-06, "loss": 0.1361, "step": 10040 }, { "epoch": 3.43, "learning_rate": 3.149472250595846e-06, "loss": 0.153, "step": 10060 }, { "epoch": 3.43, "learning_rate": 3.1358529111338103e-06, "loss": 0.1855, "step": 10080 }, { "epoch": 3.44, "learning_rate": 3.1222335716717745e-06, "loss": 0.1164, "step": 10100 }, { "epoch": 3.45, "learning_rate": 3.1086142322097378e-06, "loss": 0.1846, "step": 10120 }, { "epoch": 3.45, "learning_rate": 3.094994892747702e-06, "loss": 0.1596, "step": 10140 }, { "epoch": 3.46, "learning_rate": 3.081375553285666e-06, "loss": 0.1693, "step": 10160 }, { "epoch": 3.47, "learning_rate": 3.06775621382363e-06, "loss": 0.1359, "step": 10180 }, { "epoch": 3.47, "learning_rate": 3.0541368743615935e-06, "loss": 0.17, "step": 10200 }, { "epoch": 3.48, "learning_rate": 3.0405175348995573e-06, "loss": 0.1185, "step": 10220 }, { "epoch": 3.49, "learning_rate": 3.0268981954375214e-06, "loss": 0.111, "step": 10240 }, { "epoch": 3.49, "learning_rate": 3.0132788559754856e-06, "loss": 0.1632, "step": 10260 }, { "epoch": 3.5, "learning_rate": 2.9996595165134497e-06, "loss": 0.1217, "step": 10280 }, { "epoch": 3.51, "learning_rate": 2.986040177051413e-06, "loss": 0.1515, "step": 10300 }, { "epoch": 3.51, "learning_rate": 2.972420837589377e-06, "loss": 0.1249, "step": 10320 }, { "epoch": 3.52, "learning_rate": 2.958801498127341e-06, "loss": 0.1566, "step": 10340 }, { "epoch": 3.53, "learning_rate": 2.945182158665305e-06, "loss": 0.1707, "step": 10360 }, { "epoch": 3.53, "learning_rate": 2.931562819203269e-06, "loss": 0.1395, "step": 10380 }, { "epoch": 3.54, "learning_rate": 2.9179434797412325e-06, "loss": 0.1631, "step": 10400 }, { "epoch": 3.55, "learning_rate": 2.9043241402791967e-06, "loss": 0.1359, "step": 10420 }, { "epoch": 3.55, "learning_rate": 2.890704800817161e-06, "loss": 0.1413, "step": 10440 }, { "epoch": 3.56, "learning_rate": 2.8770854613551245e-06, "loss": 0.1415, "step": 10460 }, { "epoch": 3.57, "learning_rate": 2.8634661218930883e-06, "loss": 0.1439, "step": 10480 }, { "epoch": 3.58, "learning_rate": 2.849846782431052e-06, "loss": 0.1997, "step": 10500 }, { "epoch": 3.58, "learning_rate": 2.836227442969016e-06, "loss": 0.1727, "step": 10520 }, { "epoch": 3.59, "learning_rate": 2.8226081035069803e-06, "loss": 0.1468, "step": 10540 }, { "epoch": 3.6, "learning_rate": 2.8089887640449444e-06, "loss": 0.1451, "step": 10560 }, { "epoch": 3.6, "learning_rate": 2.7953694245829077e-06, "loss": 0.1466, "step": 10580 }, { "epoch": 3.61, "learning_rate": 2.781750085120872e-06, "loss": 0.1434, "step": 10600 }, { "epoch": 3.62, "learning_rate": 2.7681307456588356e-06, "loss": 0.1302, "step": 10620 }, { "epoch": 3.62, "learning_rate": 2.7545114061967998e-06, "loss": 0.1318, "step": 10640 }, { "epoch": 3.63, "learning_rate": 2.740892066734764e-06, "loss": 0.1516, "step": 10660 }, { "epoch": 3.64, "learning_rate": 2.7272727272727272e-06, "loss": 0.1508, "step": 10680 }, { "epoch": 3.64, "learning_rate": 2.7136533878106914e-06, "loss": 0.1838, "step": 10700 }, { "epoch": 3.65, "learning_rate": 2.7000340483486555e-06, "loss": 0.1497, "step": 10720 }, { "epoch": 3.66, "learning_rate": 2.6864147088866193e-06, "loss": 0.1694, "step": 10740 }, { "epoch": 3.66, "learning_rate": 2.6727953694245834e-06, "loss": 0.1789, "step": 10760 }, { "epoch": 3.67, "learning_rate": 2.6591760299625467e-06, "loss": 0.1369, "step": 10780 }, { "epoch": 3.68, "learning_rate": 2.645556690500511e-06, "loss": 0.1195, "step": 10800 }, { "epoch": 3.68, "learning_rate": 2.631937351038475e-06, "loss": 0.1486, "step": 10820 }, { "epoch": 3.69, "learning_rate": 2.6183180115764387e-06, "loss": 0.1374, "step": 10840 }, { "epoch": 3.7, "learning_rate": 2.6046986721144025e-06, "loss": 0.1454, "step": 10860 }, { "epoch": 3.7, "learning_rate": 2.5910793326523666e-06, "loss": 0.1969, "step": 10880 }, { "epoch": 3.71, "learning_rate": 2.5774599931903303e-06, "loss": 0.1388, "step": 10900 }, { "epoch": 3.72, "learning_rate": 2.5638406537282945e-06, "loss": 0.1714, "step": 10920 }, { "epoch": 3.72, "learning_rate": 2.5502213142662586e-06, "loss": 0.1339, "step": 10940 }, { "epoch": 3.73, "learning_rate": 2.536601974804222e-06, "loss": 0.1532, "step": 10960 }, { "epoch": 3.74, "learning_rate": 2.522982635342186e-06, "loss": 0.1918, "step": 10980 }, { "epoch": 3.75, "learning_rate": 2.5093632958801502e-06, "loss": 0.1238, "step": 11000 }, { "epoch": 3.75, "eval_loss": 0.6513485312461853, "eval_roc_auc": 0.7726358148893361, "eval_runtime": 227.7851, "eval_samples_per_second": 65.456, "eval_steps_per_second": 8.183, "step": 11000 }, { "epoch": 3.75, "learning_rate": 2.495743956418114e-06, "loss": 0.169, "step": 11020 }, { "epoch": 3.76, "learning_rate": 2.4821246169560777e-06, "loss": 0.1273, "step": 11040 }, { "epoch": 3.77, "learning_rate": 2.468505277494042e-06, "loss": 0.1087, "step": 11060 }, { "epoch": 3.77, "learning_rate": 2.4548859380320056e-06, "loss": 0.1648, "step": 11080 }, { "epoch": 3.78, "learning_rate": 2.4412665985699697e-06, "loss": 0.1748, "step": 11100 }, { "epoch": 3.79, "learning_rate": 2.4276472591079335e-06, "loss": 0.1964, "step": 11120 }, { "epoch": 3.79, "learning_rate": 2.414027919645897e-06, "loss": 0.1551, "step": 11140 }, { "epoch": 3.8, "learning_rate": 2.4004085801838613e-06, "loss": 0.175, "step": 11160 }, { "epoch": 3.81, "learning_rate": 2.386789240721825e-06, "loss": 0.1273, "step": 11180 }, { "epoch": 3.81, "learning_rate": 2.3731699012597892e-06, "loss": 0.1746, "step": 11200 }, { "epoch": 3.82, "learning_rate": 2.359550561797753e-06, "loss": 0.1182, "step": 11220 }, { "epoch": 3.83, "learning_rate": 2.345931222335717e-06, "loss": 0.1066, "step": 11240 }, { "epoch": 3.83, "learning_rate": 2.332311882873681e-06, "loss": 0.1269, "step": 11260 }, { "epoch": 3.84, "learning_rate": 2.3186925434116445e-06, "loss": 0.0901, "step": 11280 }, { "epoch": 3.85, "learning_rate": 2.3050732039496087e-06, "loss": 0.206, "step": 11300 }, { "epoch": 3.85, "learning_rate": 2.2914538644875724e-06, "loss": 0.1164, "step": 11320 }, { "epoch": 3.86, "learning_rate": 2.2778345250255366e-06, "loss": 0.1698, "step": 11340 }, { "epoch": 3.87, "learning_rate": 2.2642151855635003e-06, "loss": 0.1279, "step": 11360 }, { "epoch": 3.87, "learning_rate": 2.2505958461014645e-06, "loss": 0.1477, "step": 11380 }, { "epoch": 3.88, "learning_rate": 2.236976506639428e-06, "loss": 0.136, "step": 11400 }, { "epoch": 3.89, "learning_rate": 2.223357167177392e-06, "loss": 0.122, "step": 11420 }, { "epoch": 3.9, "learning_rate": 2.209737827715356e-06, "loss": 0.138, "step": 11440 }, { "epoch": 3.9, "learning_rate": 2.1961184882533198e-06, "loss": 0.161, "step": 11460 }, { "epoch": 3.91, "learning_rate": 2.182499148791284e-06, "loss": 0.1685, "step": 11480 }, { "epoch": 3.92, "learning_rate": 2.1688798093292477e-06, "loss": 0.1503, "step": 11500 }, { "epoch": 3.92, "learning_rate": 2.155260469867212e-06, "loss": 0.1933, "step": 11520 }, { "epoch": 3.93, "learning_rate": 2.1416411304051755e-06, "loss": 0.199, "step": 11540 }, { "epoch": 3.94, "learning_rate": 2.1280217909431393e-06, "loss": 0.1233, "step": 11560 }, { "epoch": 3.94, "learning_rate": 2.1144024514811034e-06, "loss": 0.1485, "step": 11580 }, { "epoch": 3.95, "learning_rate": 2.100783112019067e-06, "loss": 0.0955, "step": 11600 }, { "epoch": 3.96, "learning_rate": 2.0871637725570313e-06, "loss": 0.1369, "step": 11620 }, { "epoch": 3.96, "learning_rate": 2.073544433094995e-06, "loss": 0.1463, "step": 11640 }, { "epoch": 3.97, "learning_rate": 2.059925093632959e-06, "loss": 0.1499, "step": 11660 }, { "epoch": 3.98, "learning_rate": 2.046305754170923e-06, "loss": 0.1376, "step": 11680 }, { "epoch": 3.98, "learning_rate": 2.0326864147088866e-06, "loss": 0.1385, "step": 11700 }, { "epoch": 3.99, "learning_rate": 2.0190670752468508e-06, "loss": 0.1194, "step": 11720 }, { "epoch": 4.0, "learning_rate": 2.0054477357848145e-06, "loss": 0.1677, "step": 11740 }, { "epoch": 4.0, "learning_rate": 1.9918283963227787e-06, "loss": 0.1102, "step": 11760 }, { "epoch": 4.01, "learning_rate": 1.9782090568607424e-06, "loss": 0.114, "step": 11780 }, { "epoch": 4.02, "learning_rate": 1.9645897173987065e-06, "loss": 0.1493, "step": 11800 }, { "epoch": 4.02, "learning_rate": 1.9509703779366703e-06, "loss": 0.1083, "step": 11820 }, { "epoch": 4.03, "learning_rate": 1.937351038474634e-06, "loss": 0.0971, "step": 11840 }, { "epoch": 4.04, "learning_rate": 1.923731699012598e-06, "loss": 0.1186, "step": 11860 }, { "epoch": 4.04, "learning_rate": 1.910112359550562e-06, "loss": 0.0883, "step": 11880 }, { "epoch": 4.05, "learning_rate": 1.896493020088526e-06, "loss": 0.1268, "step": 11900 }, { "epoch": 4.06, "learning_rate": 1.8828736806264897e-06, "loss": 0.1449, "step": 11920 }, { "epoch": 4.07, "learning_rate": 1.8692543411644537e-06, "loss": 0.1092, "step": 11940 }, { "epoch": 4.07, "learning_rate": 1.8556350017024174e-06, "loss": 0.1058, "step": 11960 }, { "epoch": 4.08, "learning_rate": 1.8420156622403816e-06, "loss": 0.0781, "step": 11980 }, { "epoch": 4.09, "learning_rate": 1.8283963227783455e-06, "loss": 0.1054, "step": 12000 }, { "epoch": 4.09, "eval_loss": 0.7551047801971436, "eval_roc_auc": 0.7666666666666667, "eval_runtime": 225.8268, "eval_samples_per_second": 66.024, "eval_steps_per_second": 8.254, "step": 12000 }, { "epoch": 4.09, "learning_rate": 1.8147769833163092e-06, "loss": 0.1366, "step": 12020 }, { "epoch": 4.1, "learning_rate": 1.8011576438542734e-06, "loss": 0.1395, "step": 12040 }, { "epoch": 4.11, "learning_rate": 1.787538304392237e-06, "loss": 0.094, "step": 12060 }, { "epoch": 4.11, "learning_rate": 1.773918964930201e-06, "loss": 0.1077, "step": 12080 }, { "epoch": 4.12, "learning_rate": 1.760299625468165e-06, "loss": 0.1134, "step": 12100 }, { "epoch": 4.13, "learning_rate": 1.746680286006129e-06, "loss": 0.1002, "step": 12120 }, { "epoch": 4.13, "learning_rate": 1.7330609465440929e-06, "loss": 0.0906, "step": 12140 }, { "epoch": 4.14, "learning_rate": 1.7194416070820566e-06, "loss": 0.1249, "step": 12160 }, { "epoch": 4.15, "learning_rate": 1.7058222676200205e-06, "loss": 0.1197, "step": 12180 }, { "epoch": 4.15, "learning_rate": 1.6922029281579845e-06, "loss": 0.1198, "step": 12200 }, { "epoch": 4.16, "learning_rate": 1.6785835886959484e-06, "loss": 0.1104, "step": 12220 }, { "epoch": 4.17, "learning_rate": 1.6649642492339123e-06, "loss": 0.1034, "step": 12240 }, { "epoch": 4.17, "learning_rate": 1.651344909771876e-06, "loss": 0.0803, "step": 12260 }, { "epoch": 4.18, "learning_rate": 1.6377255703098402e-06, "loss": 0.1039, "step": 12280 }, { "epoch": 4.19, "learning_rate": 1.624106230847804e-06, "loss": 0.1119, "step": 12300 }, { "epoch": 4.19, "learning_rate": 1.6104868913857679e-06, "loss": 0.1463, "step": 12320 }, { "epoch": 4.2, "learning_rate": 1.5968675519237318e-06, "loss": 0.1201, "step": 12340 }, { "epoch": 4.21, "learning_rate": 1.5832482124616958e-06, "loss": 0.1104, "step": 12360 }, { "epoch": 4.22, "learning_rate": 1.5696288729996597e-06, "loss": 0.1164, "step": 12380 }, { "epoch": 4.22, "learning_rate": 1.5560095335376234e-06, "loss": 0.0739, "step": 12400 }, { "epoch": 4.23, "learning_rate": 1.5423901940755876e-06, "loss": 0.1345, "step": 12420 }, { "epoch": 4.24, "learning_rate": 1.5287708546135513e-06, "loss": 0.11, "step": 12440 }, { "epoch": 4.24, "learning_rate": 1.5151515151515152e-06, "loss": 0.0976, "step": 12460 }, { "epoch": 4.25, "learning_rate": 1.501532175689479e-06, "loss": 0.1323, "step": 12480 }, { "epoch": 4.26, "learning_rate": 1.4879128362274431e-06, "loss": 0.1222, "step": 12500 }, { "epoch": 4.26, "learning_rate": 1.474293496765407e-06, "loss": 0.1165, "step": 12520 }, { "epoch": 4.27, "learning_rate": 1.4606741573033708e-06, "loss": 0.1214, "step": 12540 }, { "epoch": 4.28, "learning_rate": 1.447054817841335e-06, "loss": 0.0871, "step": 12560 }, { "epoch": 4.28, "learning_rate": 1.4334354783792987e-06, "loss": 0.1088, "step": 12580 }, { "epoch": 4.29, "learning_rate": 1.4198161389172626e-06, "loss": 0.0892, "step": 12600 }, { "epoch": 4.3, "learning_rate": 1.4061967994552263e-06, "loss": 0.1352, "step": 12620 }, { "epoch": 4.3, "learning_rate": 1.3925774599931905e-06, "loss": 0.0977, "step": 12640 }, { "epoch": 4.31, "learning_rate": 1.3789581205311544e-06, "loss": 0.1374, "step": 12660 }, { "epoch": 4.32, "learning_rate": 1.3653387810691181e-06, "loss": 0.0864, "step": 12680 }, { "epoch": 4.32, "learning_rate": 1.3517194416070823e-06, "loss": 0.089, "step": 12700 }, { "epoch": 4.33, "learning_rate": 1.338100102145046e-06, "loss": 0.0961, "step": 12720 }, { "epoch": 4.34, "learning_rate": 1.32448076268301e-06, "loss": 0.1148, "step": 12740 }, { "epoch": 4.34, "learning_rate": 1.3108614232209737e-06, "loss": 0.1113, "step": 12760 }, { "epoch": 4.35, "learning_rate": 1.2972420837589378e-06, "loss": 0.0984, "step": 12780 }, { "epoch": 4.36, "learning_rate": 1.2836227442969018e-06, "loss": 0.1134, "step": 12800 }, { "epoch": 4.36, "learning_rate": 1.2700034048348655e-06, "loss": 0.1126, "step": 12820 }, { "epoch": 4.37, "learning_rate": 1.2563840653728297e-06, "loss": 0.1318, "step": 12840 }, { "epoch": 4.38, "learning_rate": 1.2427647259107934e-06, "loss": 0.1326, "step": 12860 }, { "epoch": 4.39, "learning_rate": 1.2291453864487573e-06, "loss": 0.0954, "step": 12880 }, { "epoch": 4.39, "learning_rate": 1.2155260469867213e-06, "loss": 0.1549, "step": 12900 }, { "epoch": 4.4, "learning_rate": 1.2019067075246852e-06, "loss": 0.1077, "step": 12920 }, { "epoch": 4.41, "learning_rate": 1.188287368062649e-06, "loss": 0.0598, "step": 12940 }, { "epoch": 4.41, "learning_rate": 1.1746680286006129e-06, "loss": 0.1317, "step": 12960 }, { "epoch": 4.42, "learning_rate": 1.161048689138577e-06, "loss": 0.0877, "step": 12980 }, { "epoch": 4.43, "learning_rate": 1.1474293496765407e-06, "loss": 0.1076, "step": 13000 }, { "epoch": 4.43, "eval_loss": 0.8132159113883972, "eval_roc_auc": 0.7627095908786049, "eval_runtime": 227.818, "eval_samples_per_second": 65.447, "eval_steps_per_second": 8.182, "step": 13000 }, { "epoch": 4.43, "learning_rate": 1.1338100102145047e-06, "loss": 0.1132, "step": 13020 }, { "epoch": 4.44, "learning_rate": 1.1201906707524686e-06, "loss": 0.0821, "step": 13040 }, { "epoch": 4.45, "learning_rate": 1.1065713312904326e-06, "loss": 0.0979, "step": 13060 }, { "epoch": 4.45, "learning_rate": 1.0929519918283963e-06, "loss": 0.1109, "step": 13080 }, { "epoch": 4.46, "learning_rate": 1.0793326523663602e-06, "loss": 0.1429, "step": 13100 }, { "epoch": 4.47, "learning_rate": 1.0657133129043244e-06, "loss": 0.1369, "step": 13120 }, { "epoch": 4.47, "learning_rate": 1.052093973442288e-06, "loss": 0.103, "step": 13140 }, { "epoch": 4.48, "learning_rate": 1.038474633980252e-06, "loss": 0.1071, "step": 13160 }, { "epoch": 4.49, "learning_rate": 1.024855294518216e-06, "loss": 0.0864, "step": 13180 }, { "epoch": 4.49, "learning_rate": 1.01123595505618e-06, "loss": 0.1063, "step": 13200 }, { "epoch": 4.5, "learning_rate": 9.976166155941436e-07, "loss": 0.1019, "step": 13220 }, { "epoch": 4.51, "learning_rate": 9.839972761321076e-07, "loss": 0.1222, "step": 13240 }, { "epoch": 4.51, "learning_rate": 9.703779366700715e-07, "loss": 0.0911, "step": 13260 }, { "epoch": 4.52, "learning_rate": 9.567585972080355e-07, "loss": 0.1774, "step": 13280 }, { "epoch": 4.53, "learning_rate": 9.431392577459994e-07, "loss": 0.1098, "step": 13300 }, { "epoch": 4.54, "learning_rate": 9.295199182839632e-07, "loss": 0.1371, "step": 13320 }, { "epoch": 4.54, "learning_rate": 9.159005788219272e-07, "loss": 0.1277, "step": 13340 }, { "epoch": 4.55, "learning_rate": 9.022812393598911e-07, "loss": 0.0962, "step": 13360 }, { "epoch": 4.56, "learning_rate": 8.886618998978551e-07, "loss": 0.1411, "step": 13380 }, { "epoch": 4.56, "learning_rate": 8.75042560435819e-07, "loss": 0.1044, "step": 13400 }, { "epoch": 4.57, "learning_rate": 8.614232209737828e-07, "loss": 0.0907, "step": 13420 }, { "epoch": 4.58, "learning_rate": 8.478038815117468e-07, "loss": 0.1084, "step": 13440 }, { "epoch": 4.58, "learning_rate": 8.341845420497106e-07, "loss": 0.1198, "step": 13460 }, { "epoch": 4.59, "learning_rate": 8.205652025876745e-07, "loss": 0.0946, "step": 13480 }, { "epoch": 4.6, "learning_rate": 8.069458631256384e-07, "loss": 0.1361, "step": 13500 }, { "epoch": 4.6, "learning_rate": 7.933265236636024e-07, "loss": 0.0859, "step": 13520 }, { "epoch": 4.61, "learning_rate": 7.797071842015664e-07, "loss": 0.1314, "step": 13540 }, { "epoch": 4.62, "learning_rate": 7.660878447395302e-07, "loss": 0.121, "step": 13560 }, { "epoch": 4.62, "learning_rate": 7.524685052774941e-07, "loss": 0.109, "step": 13580 }, { "epoch": 4.63, "learning_rate": 7.38849165815458e-07, "loss": 0.0947, "step": 13600 }, { "epoch": 4.64, "learning_rate": 7.252298263534219e-07, "loss": 0.105, "step": 13620 }, { "epoch": 4.64, "learning_rate": 7.116104868913857e-07, "loss": 0.1235, "step": 13640 }, { "epoch": 4.65, "learning_rate": 6.979911474293498e-07, "loss": 0.1332, "step": 13660 }, { "epoch": 4.66, "learning_rate": 6.843718079673137e-07, "loss": 0.0717, "step": 13680 }, { "epoch": 4.66, "learning_rate": 6.707524685052775e-07, "loss": 0.1095, "step": 13700 }, { "epoch": 4.67, "learning_rate": 6.571331290432415e-07, "loss": 0.1468, "step": 13720 }, { "epoch": 4.68, "learning_rate": 6.435137895812053e-07, "loss": 0.1202, "step": 13740 }, { "epoch": 4.69, "learning_rate": 6.298944501191693e-07, "loss": 0.0621, "step": 13760 }, { "epoch": 4.69, "learning_rate": 6.162751106571332e-07, "loss": 0.0852, "step": 13780 }, { "epoch": 4.7, "learning_rate": 6.02655771195097e-07, "loss": 0.1006, "step": 13800 }, { "epoch": 4.71, "learning_rate": 5.89036431733061e-07, "loss": 0.1046, "step": 13820 }, { "epoch": 4.71, "learning_rate": 5.754170922710249e-07, "loss": 0.1105, "step": 13840 }, { "epoch": 4.72, "learning_rate": 5.617977528089888e-07, "loss": 0.0842, "step": 13860 }, { "epoch": 4.73, "learning_rate": 5.481784133469528e-07, "loss": 0.0779, "step": 13880 }, { "epoch": 4.73, "learning_rate": 5.345590738849166e-07, "loss": 0.1007, "step": 13900 }, { "epoch": 4.74, "learning_rate": 5.209397344228806e-07, "loss": 0.0784, "step": 13920 }, { "epoch": 4.75, "learning_rate": 5.073203949608444e-07, "loss": 0.121, "step": 13940 }, { "epoch": 4.75, "learning_rate": 4.937010554988083e-07, "loss": 0.1562, "step": 13960 }, { "epoch": 4.76, "learning_rate": 4.800817160367723e-07, "loss": 0.1098, "step": 13980 }, { "epoch": 4.77, "learning_rate": 4.6646237657473615e-07, "loss": 0.1321, "step": 14000 }, { "epoch": 4.77, "eval_loss": 0.815223753452301, "eval_roc_auc": 0.7586854460093896, "eval_runtime": 225.105, "eval_samples_per_second": 66.236, "eval_steps_per_second": 8.281, "step": 14000 }, { "epoch": 4.77, "learning_rate": 4.528430371127001e-07, "loss": 0.1244, "step": 14020 }, { "epoch": 4.78, "learning_rate": 4.39223697650664e-07, "loss": 0.1415, "step": 14040 }, { "epoch": 4.79, "learning_rate": 4.2560435818862786e-07, "loss": 0.0982, "step": 14060 }, { "epoch": 4.79, "learning_rate": 4.1198501872659175e-07, "loss": 0.0874, "step": 14080 }, { "epoch": 4.8, "learning_rate": 3.9836567926455574e-07, "loss": 0.1736, "step": 14100 }, { "epoch": 4.81, "learning_rate": 3.8474633980251963e-07, "loss": 0.1213, "step": 14120 }, { "epoch": 4.81, "learning_rate": 3.711270003404835e-07, "loss": 0.1008, "step": 14140 }, { "epoch": 4.82, "learning_rate": 3.5750766087844745e-07, "loss": 0.1624, "step": 14160 }, { "epoch": 4.83, "learning_rate": 3.4388832141641134e-07, "loss": 0.1073, "step": 14180 }, { "epoch": 4.83, "learning_rate": 3.302689819543752e-07, "loss": 0.0801, "step": 14200 }, { "epoch": 4.84, "learning_rate": 3.1664964249233916e-07, "loss": 0.1252, "step": 14220 }, { "epoch": 4.85, "learning_rate": 3.0303030303030305e-07, "loss": 0.0988, "step": 14240 }, { "epoch": 4.86, "learning_rate": 2.8941096356826693e-07, "loss": 0.0818, "step": 14260 }, { "epoch": 4.86, "learning_rate": 2.7579162410623087e-07, "loss": 0.1005, "step": 14280 }, { "epoch": 4.87, "learning_rate": 2.621722846441948e-07, "loss": 0.1191, "step": 14300 }, { "epoch": 4.88, "learning_rate": 2.485529451821587e-07, "loss": 0.1062, "step": 14320 }, { "epoch": 4.88, "learning_rate": 2.349336057201226e-07, "loss": 0.115, "step": 14340 }, { "epoch": 4.89, "learning_rate": 2.213142662580865e-07, "loss": 0.1172, "step": 14360 }, { "epoch": 4.9, "learning_rate": 2.076949267960504e-07, "loss": 0.099, "step": 14380 }, { "epoch": 4.9, "learning_rate": 1.9407558733401432e-07, "loss": 0.127, "step": 14400 }, { "epoch": 4.91, "learning_rate": 1.804562478719782e-07, "loss": 0.0879, "step": 14420 }, { "epoch": 4.92, "learning_rate": 1.6683690840994215e-07, "loss": 0.1284, "step": 14440 }, { "epoch": 4.92, "learning_rate": 1.5321756894790606e-07, "loss": 0.1249, "step": 14460 }, { "epoch": 4.93, "learning_rate": 1.3959822948586994e-07, "loss": 0.1314, "step": 14480 }, { "epoch": 4.94, "learning_rate": 1.2597889002383386e-07, "loss": 0.1307, "step": 14500 }, { "epoch": 4.94, "learning_rate": 1.1235955056179776e-07, "loss": 0.092, "step": 14520 }, { "epoch": 4.95, "learning_rate": 9.874021109976167e-08, "loss": 0.084, "step": 14540 }, { "epoch": 4.96, "learning_rate": 8.512087163772558e-08, "loss": 0.1793, "step": 14560 }, { "epoch": 4.96, "learning_rate": 7.150153217568949e-08, "loss": 0.1071, "step": 14580 }, { "epoch": 4.97, "learning_rate": 5.788219271365339e-08, "loss": 0.103, "step": 14600 }, { "epoch": 4.98, "learning_rate": 4.42628532516173e-08, "loss": 0.1024, "step": 14620 }, { "epoch": 4.98, "learning_rate": 3.0643513789581204e-08, "loss": 0.0751, "step": 14640 }, { "epoch": 4.99, "learning_rate": 1.7024174327545116e-08, "loss": 0.0811, "step": 14660 }, { "epoch": 5.0, "learning_rate": 3.4048348655090227e-09, "loss": 0.0877, "step": 14680 }, { "epoch": 5.0, "step": 14685, "total_flos": 3.6406511640831468e+19, "train_loss": 0.21540352483164263, "train_runtime": 8294.2837, "train_samples_per_second": 56.643, "train_steps_per_second": 1.77 } ], "logging_steps": 20, "max_steps": 14685, "num_train_epochs": 5, "save_steps": 1000, "total_flos": 3.6406511640831468e+19, "trial_name": null, "trial_params": null }