diff --git "a/multi_student_without_cot/trainer_state.json" "b/multi_student_without_cot/trainer_state.json" new file mode 100644--- /dev/null +++ "b/multi_student_without_cot/trainer_state.json" @@ -0,0 +1,7414 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 5.0, + "eval_steps": 500, + "global_step": 10530, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.004748338081671415, + "grad_norm": 20.21827706063475, + "learning_rate": 8.547008547008548e-07, + "loss": 8.145, + "step": 10 + }, + { + "epoch": 0.00949667616334283, + "grad_norm": 18.24167682921799, + "learning_rate": 1.8043684710351376e-06, + "loss": 8.0172, + "step": 20 + }, + { + "epoch": 0.014245014245014245, + "grad_norm": 11.789191102542203, + "learning_rate": 2.7540360873694207e-06, + "loss": 7.3368, + "step": 30 + }, + { + "epoch": 0.01899335232668566, + "grad_norm": 7.392323142404382, + "learning_rate": 3.7037037037037037e-06, + "loss": 6.3843, + "step": 40 + }, + { + "epoch": 0.023741690408357077, + "grad_norm": 5.964800407301951, + "learning_rate": 4.653371320037987e-06, + "loss": 5.5837, + "step": 50 + }, + { + "epoch": 0.02849002849002849, + "grad_norm": 7.120458677127898, + "learning_rate": 5.60303893637227e-06, + "loss": 4.8579, + "step": 60 + }, + { + "epoch": 0.03323836657169991, + "grad_norm": 8.344343007491538, + "learning_rate": 6.5527065527065525e-06, + "loss": 3.7314, + "step": 70 + }, + { + "epoch": 0.03798670465337132, + "grad_norm": 3.1169502061629655, + "learning_rate": 7.5023741690408365e-06, + "loss": 2.5812, + "step": 80 + }, + { + "epoch": 0.042735042735042736, + "grad_norm": 2.157035593083223, + "learning_rate": 8.452041785375119e-06, + "loss": 2.0913, + "step": 90 + }, + { + "epoch": 0.04748338081671415, + "grad_norm": 1.877682319517242, + "learning_rate": 9.401709401709402e-06, + "loss": 1.9562, + "step": 100 + }, + { + "epoch": 0.052231718898385564, + "grad_norm": 2.6870548427576004, + "learning_rate": 1.0351377018043685e-05, + "loss": 1.8719, + "step": 110 + }, + { + "epoch": 0.05698005698005698, + "grad_norm": 4.980475392094001, + "learning_rate": 1.1301044634377968e-05, + "loss": 1.7976, + "step": 120 + }, + { + "epoch": 0.06172839506172839, + "grad_norm": 8.145965181044149, + "learning_rate": 1.2250712250712251e-05, + "loss": 1.7596, + "step": 130 + }, + { + "epoch": 0.06647673314339982, + "grad_norm": 10.041185447423977, + "learning_rate": 1.3200379867046534e-05, + "loss": 1.7159, + "step": 140 + }, + { + "epoch": 0.07122507122507123, + "grad_norm": 10.11622125534401, + "learning_rate": 1.4150047483380819e-05, + "loss": 1.6898, + "step": 150 + }, + { + "epoch": 0.07597340930674264, + "grad_norm": 8.736601880824086, + "learning_rate": 1.50997150997151e-05, + "loss": 1.6485, + "step": 160 + }, + { + "epoch": 0.08072174738841406, + "grad_norm": 13.03281077165562, + "learning_rate": 1.604938271604938e-05, + "loss": 1.6175, + "step": 170 + }, + { + "epoch": 0.08547008547008547, + "grad_norm": 9.908577710414223, + "learning_rate": 1.6999050332383668e-05, + "loss": 1.5976, + "step": 180 + }, + { + "epoch": 0.09021842355175688, + "grad_norm": 9.237568820174022, + "learning_rate": 1.794871794871795e-05, + "loss": 1.5488, + "step": 190 + }, + { + "epoch": 0.0949667616334283, + "grad_norm": 4.68415488864487, + "learning_rate": 1.889838556505223e-05, + "loss": 1.5014, + "step": 200 + }, + { + "epoch": 0.09971509971509972, + "grad_norm": 4.844935709898889, + "learning_rate": 1.9848053181386517e-05, + "loss": 1.4854, + "step": 210 + }, + { + "epoch": 0.10446343779677113, + "grad_norm": 4.323562398013845, + "learning_rate": 2.07977207977208e-05, + "loss": 1.4586, + "step": 220 + }, + { + "epoch": 0.10921177587844255, + "grad_norm": 3.4385277344372978, + "learning_rate": 2.174738841405508e-05, + "loss": 1.4301, + "step": 230 + }, + { + "epoch": 0.11396011396011396, + "grad_norm": 3.077116780247069, + "learning_rate": 2.2697056030389367e-05, + "loss": 1.3928, + "step": 240 + }, + { + "epoch": 0.11870845204178537, + "grad_norm": 4.302072263978756, + "learning_rate": 2.364672364672365e-05, + "loss": 1.3771, + "step": 250 + }, + { + "epoch": 0.12345679012345678, + "grad_norm": 3.5795185303541572, + "learning_rate": 2.459639126305793e-05, + "loss": 1.3484, + "step": 260 + }, + { + "epoch": 0.1282051282051282, + "grad_norm": 2.071897753499616, + "learning_rate": 2.5546058879392216e-05, + "loss": 1.3107, + "step": 270 + }, + { + "epoch": 0.13295346628679963, + "grad_norm": 2.421429428155855, + "learning_rate": 2.64957264957265e-05, + "loss": 1.2881, + "step": 280 + }, + { + "epoch": 0.13770180436847104, + "grad_norm": 3.700036964420459, + "learning_rate": 2.744539411206078e-05, + "loss": 1.2667, + "step": 290 + }, + { + "epoch": 0.14245014245014245, + "grad_norm": 2.8958286581854944, + "learning_rate": 2.839506172839506e-05, + "loss": 1.247, + "step": 300 + }, + { + "epoch": 0.14719848053181386, + "grad_norm": 2.054414468223543, + "learning_rate": 2.9344729344729345e-05, + "loss": 1.2217, + "step": 310 + }, + { + "epoch": 0.15194681861348527, + "grad_norm": 1.96235600963249, + "learning_rate": 3.029439696106363e-05, + "loss": 1.2, + "step": 320 + }, + { + "epoch": 0.15669515669515668, + "grad_norm": 2.731627929284309, + "learning_rate": 3.1244064577397914e-05, + "loss": 1.1817, + "step": 330 + }, + { + "epoch": 0.16144349477682812, + "grad_norm": 2.1653241264851637, + "learning_rate": 3.2193732193732194e-05, + "loss": 1.1623, + "step": 340 + }, + { + "epoch": 0.16619183285849953, + "grad_norm": 2.519530114010952, + "learning_rate": 3.314339981006648e-05, + "loss": 1.1288, + "step": 350 + }, + { + "epoch": 0.17094017094017094, + "grad_norm": 2.653996907642351, + "learning_rate": 3.409306742640076e-05, + "loss": 1.1081, + "step": 360 + }, + { + "epoch": 0.17568850902184235, + "grad_norm": 2.4977140265839126, + "learning_rate": 3.504273504273504e-05, + "loss": 1.0854, + "step": 370 + }, + { + "epoch": 0.18043684710351376, + "grad_norm": 2.856768490699816, + "learning_rate": 3.599240265906933e-05, + "loss": 1.0636, + "step": 380 + }, + { + "epoch": 0.18518518518518517, + "grad_norm": 2.17649907858301, + "learning_rate": 3.694207027540361e-05, + "loss": 1.0478, + "step": 390 + }, + { + "epoch": 0.1899335232668566, + "grad_norm": 3.9461445509883593, + "learning_rate": 3.789173789173789e-05, + "loss": 1.0281, + "step": 400 + }, + { + "epoch": 0.19468186134852802, + "grad_norm": 2.871728182164183, + "learning_rate": 3.884140550807218e-05, + "loss": 0.9991, + "step": 410 + }, + { + "epoch": 0.19943019943019943, + "grad_norm": 3.013330420912045, + "learning_rate": 3.979107312440646e-05, + "loss": 0.9773, + "step": 420 + }, + { + "epoch": 0.20417853751187084, + "grad_norm": 2.420232752543592, + "learning_rate": 4.074074074074074e-05, + "loss": 0.9586, + "step": 430 + }, + { + "epoch": 0.20892687559354226, + "grad_norm": 2.5768091814664564, + "learning_rate": 4.169040835707503e-05, + "loss": 0.938, + "step": 440 + }, + { + "epoch": 0.21367521367521367, + "grad_norm": 3.113768661249082, + "learning_rate": 4.264007597340931e-05, + "loss": 0.9293, + "step": 450 + }, + { + "epoch": 0.2184235517568851, + "grad_norm": 2.294957751769951, + "learning_rate": 4.358974358974359e-05, + "loss": 0.9049, + "step": 460 + }, + { + "epoch": 0.22317188983855651, + "grad_norm": 2.4492691119209855, + "learning_rate": 4.453941120607788e-05, + "loss": 0.8865, + "step": 470 + }, + { + "epoch": 0.22792022792022792, + "grad_norm": 2.286177237412566, + "learning_rate": 4.548907882241216e-05, + "loss": 0.866, + "step": 480 + }, + { + "epoch": 0.23266856600189934, + "grad_norm": 2.589513226874877, + "learning_rate": 4.643874643874644e-05, + "loss": 0.8457, + "step": 490 + }, + { + "epoch": 0.23741690408357075, + "grad_norm": 3.0229688171419875, + "learning_rate": 4.738841405508073e-05, + "loss": 0.8279, + "step": 500 + }, + { + "epoch": 0.24216524216524216, + "grad_norm": 2.686293470195147, + "learning_rate": 4.833808167141501e-05, + "loss": 0.8147, + "step": 510 + }, + { + "epoch": 0.24691358024691357, + "grad_norm": 2.17482173902714, + "learning_rate": 4.928774928774929e-05, + "loss": 0.7885, + "step": 520 + }, + { + "epoch": 0.251661918328585, + "grad_norm": 2.8541347160003783, + "learning_rate": 5.0237416904083576e-05, + "loss": 0.7784, + "step": 530 + }, + { + "epoch": 0.2564102564102564, + "grad_norm": 2.59370867139089, + "learning_rate": 5.118708452041786e-05, + "loss": 0.7673, + "step": 540 + }, + { + "epoch": 0.2611585944919278, + "grad_norm": 3.1513543951371905, + "learning_rate": 5.213675213675214e-05, + "loss": 0.7479, + "step": 550 + }, + { + "epoch": 0.26590693257359926, + "grad_norm": 2.536640604487201, + "learning_rate": 5.308641975308642e-05, + "loss": 0.7253, + "step": 560 + }, + { + "epoch": 0.2706552706552707, + "grad_norm": 2.8005849730532213, + "learning_rate": 5.403608736942071e-05, + "loss": 0.7154, + "step": 570 + }, + { + "epoch": 0.2754036087369421, + "grad_norm": 2.551754718587974, + "learning_rate": 5.498575498575499e-05, + "loss": 0.6901, + "step": 580 + }, + { + "epoch": 0.2801519468186135, + "grad_norm": 2.6731387859043365, + "learning_rate": 5.5935422602089274e-05, + "loss": 0.6784, + "step": 590 + }, + { + "epoch": 0.2849002849002849, + "grad_norm": 2.558888468671125, + "learning_rate": 5.6885090218423554e-05, + "loss": 0.6634, + "step": 600 + }, + { + "epoch": 0.2896486229819563, + "grad_norm": 2.7304234847433477, + "learning_rate": 5.7834757834757834e-05, + "loss": 0.6528, + "step": 610 + }, + { + "epoch": 0.2943969610636277, + "grad_norm": 2.469120865455278, + "learning_rate": 5.878442545109212e-05, + "loss": 0.6357, + "step": 620 + }, + { + "epoch": 0.29914529914529914, + "grad_norm": 2.361617168312385, + "learning_rate": 5.97340930674264e-05, + "loss": 0.6184, + "step": 630 + }, + { + "epoch": 0.30389363722697055, + "grad_norm": 2.5062066018611544, + "learning_rate": 6.068376068376068e-05, + "loss": 0.6016, + "step": 640 + }, + { + "epoch": 0.30864197530864196, + "grad_norm": 3.35086067384153, + "learning_rate": 6.163342830009497e-05, + "loss": 0.5978, + "step": 650 + }, + { + "epoch": 0.31339031339031337, + "grad_norm": 2.177019547847069, + "learning_rate": 6.258309591642925e-05, + "loss": 0.5774, + "step": 660 + }, + { + "epoch": 0.3181386514719848, + "grad_norm": 2.6347133971158465, + "learning_rate": 6.353276353276355e-05, + "loss": 0.5683, + "step": 670 + }, + { + "epoch": 0.32288698955365625, + "grad_norm": 1.9226802841170278, + "learning_rate": 6.448243114909783e-05, + "loss": 0.5504, + "step": 680 + }, + { + "epoch": 0.32763532763532766, + "grad_norm": 1.9462440337757687, + "learning_rate": 6.54320987654321e-05, + "loss": 0.5393, + "step": 690 + }, + { + "epoch": 0.33238366571699907, + "grad_norm": 2.64699999142231, + "learning_rate": 6.638176638176638e-05, + "loss": 0.5295, + "step": 700 + }, + { + "epoch": 0.3371320037986705, + "grad_norm": 2.428317113805517, + "learning_rate": 6.733143399810066e-05, + "loss": 0.5172, + "step": 710 + }, + { + "epoch": 0.3418803418803419, + "grad_norm": 2.2288280010027854, + "learning_rate": 6.828110161443494e-05, + "loss": 0.5013, + "step": 720 + }, + { + "epoch": 0.3466286799620133, + "grad_norm": 2.417506087683697, + "learning_rate": 6.923076923076924e-05, + "loss": 0.494, + "step": 730 + }, + { + "epoch": 0.3513770180436847, + "grad_norm": 1.9466452806944854, + "learning_rate": 7.018043684710352e-05, + "loss": 0.4766, + "step": 740 + }, + { + "epoch": 0.3561253561253561, + "grad_norm": 2.548627924889759, + "learning_rate": 7.11301044634378e-05, + "loss": 0.4684, + "step": 750 + }, + { + "epoch": 0.36087369420702753, + "grad_norm": 2.157265574032032, + "learning_rate": 7.207977207977208e-05, + "loss": 0.4606, + "step": 760 + }, + { + "epoch": 0.36562203228869894, + "grad_norm": 2.037117411340044, + "learning_rate": 7.302943969610636e-05, + "loss": 0.4485, + "step": 770 + }, + { + "epoch": 0.37037037037037035, + "grad_norm": 1.8379017170113745, + "learning_rate": 7.397910731244065e-05, + "loss": 0.4338, + "step": 780 + }, + { + "epoch": 0.37511870845204176, + "grad_norm": 2.8763875380181734, + "learning_rate": 7.492877492877494e-05, + "loss": 0.4289, + "step": 790 + }, + { + "epoch": 0.3798670465337132, + "grad_norm": 1.8515635424892907, + "learning_rate": 7.587844254510922e-05, + "loss": 0.4179, + "step": 800 + }, + { + "epoch": 0.38461538461538464, + "grad_norm": 1.6717226940996468, + "learning_rate": 7.68281101614435e-05, + "loss": 0.4092, + "step": 810 + }, + { + "epoch": 0.38936372269705605, + "grad_norm": 2.081127977320627, + "learning_rate": 7.777777777777778e-05, + "loss": 0.3973, + "step": 820 + }, + { + "epoch": 0.39411206077872746, + "grad_norm": 1.8193841823557135, + "learning_rate": 7.872744539411206e-05, + "loss": 0.3905, + "step": 830 + }, + { + "epoch": 0.39886039886039887, + "grad_norm": 2.029853671008697, + "learning_rate": 7.967711301044634e-05, + "loss": 0.3829, + "step": 840 + }, + { + "epoch": 0.4036087369420703, + "grad_norm": 1.7467553830120104, + "learning_rate": 8.062678062678063e-05, + "loss": 0.3705, + "step": 850 + }, + { + "epoch": 0.4083570750237417, + "grad_norm": 1.8021994967000676, + "learning_rate": 8.157644824311491e-05, + "loss": 0.3639, + "step": 860 + }, + { + "epoch": 0.4131054131054131, + "grad_norm": 1.641013312643542, + "learning_rate": 8.25261158594492e-05, + "loss": 0.3608, + "step": 870 + }, + { + "epoch": 0.4178537511870845, + "grad_norm": 1.5079068085978562, + "learning_rate": 8.347578347578347e-05, + "loss": 0.3522, + "step": 880 + }, + { + "epoch": 0.4226020892687559, + "grad_norm": 1.4655692445142585, + "learning_rate": 8.442545109211775e-05, + "loss": 0.3443, + "step": 890 + }, + { + "epoch": 0.42735042735042733, + "grad_norm": 1.4792862368111255, + "learning_rate": 8.537511870845205e-05, + "loss": 0.3373, + "step": 900 + }, + { + "epoch": 0.43209876543209874, + "grad_norm": 1.7589212642036423, + "learning_rate": 8.632478632478634e-05, + "loss": 0.3349, + "step": 910 + }, + { + "epoch": 0.4368471035137702, + "grad_norm": 1.3662339866574285, + "learning_rate": 8.727445394112062e-05, + "loss": 0.3264, + "step": 920 + }, + { + "epoch": 0.4415954415954416, + "grad_norm": 1.6091099458218114, + "learning_rate": 8.82241215574549e-05, + "loss": 0.3181, + "step": 930 + }, + { + "epoch": 0.44634377967711303, + "grad_norm": 1.347407542639701, + "learning_rate": 8.917378917378918e-05, + "loss": 0.3095, + "step": 940 + }, + { + "epoch": 0.45109211775878444, + "grad_norm": 1.9418517203504135, + "learning_rate": 9.012345679012346e-05, + "loss": 0.305, + "step": 950 + }, + { + "epoch": 0.45584045584045585, + "grad_norm": 1.3400978709493576, + "learning_rate": 9.107312440645774e-05, + "loss": 0.3008, + "step": 960 + }, + { + "epoch": 0.46058879392212726, + "grad_norm": 1.2579527847439156, + "learning_rate": 9.202279202279203e-05, + "loss": 0.2953, + "step": 970 + }, + { + "epoch": 0.46533713200379867, + "grad_norm": 1.2589304750507126, + "learning_rate": 9.297245963912631e-05, + "loss": 0.2898, + "step": 980 + }, + { + "epoch": 0.4700854700854701, + "grad_norm": 1.3117037531826747, + "learning_rate": 9.392212725546059e-05, + "loss": 0.2843, + "step": 990 + }, + { + "epoch": 0.4748338081671415, + "grad_norm": 1.3291209884932123, + "learning_rate": 9.487179487179487e-05, + "loss": 0.2815, + "step": 1000 + }, + { + "epoch": 0.4795821462488129, + "grad_norm": 1.209968301390431, + "learning_rate": 9.582146248812915e-05, + "loss": 0.2799, + "step": 1010 + }, + { + "epoch": 0.4843304843304843, + "grad_norm": 1.0056322036392946, + "learning_rate": 9.677113010446344e-05, + "loss": 0.275, + "step": 1020 + }, + { + "epoch": 0.4890788224121557, + "grad_norm": 1.2702143901456204, + "learning_rate": 9.772079772079774e-05, + "loss": 0.2691, + "step": 1030 + }, + { + "epoch": 0.49382716049382713, + "grad_norm": 1.0226410737222107, + "learning_rate": 9.867046533713202e-05, + "loss": 0.2616, + "step": 1040 + }, + { + "epoch": 0.4985754985754986, + "grad_norm": 0.9723093234963888, + "learning_rate": 9.96201329534663e-05, + "loss": 0.2579, + "step": 1050 + }, + { + "epoch": 0.50332383665717, + "grad_norm": 1.3912534326159531, + "learning_rate": 9.999990109907572e-05, + "loss": 0.2551, + "step": 1060 + }, + { + "epoch": 0.5080721747388414, + "grad_norm": 1.1085969512788294, + "learning_rate": 9.999929670595522e-05, + "loss": 0.2531, + "step": 1070 + }, + { + "epoch": 0.5128205128205128, + "grad_norm": 1.0552909890539075, + "learning_rate": 9.99981428713058e-05, + "loss": 0.2483, + "step": 1080 + }, + { + "epoch": 0.5175688509021842, + "grad_norm": 1.1040506601466529, + "learning_rate": 9.999643960780693e-05, + "loss": 0.2423, + "step": 1090 + }, + { + "epoch": 0.5223171889838556, + "grad_norm": 0.9501946984359841, + "learning_rate": 9.999418693417573e-05, + "loss": 0.2376, + "step": 1100 + }, + { + "epoch": 0.5270655270655271, + "grad_norm": 1.022708548136873, + "learning_rate": 9.999138487516681e-05, + "loss": 0.2343, + "step": 1110 + }, + { + "epoch": 0.5318138651471985, + "grad_norm": 1.1751853468010314, + "learning_rate": 9.998803346157195e-05, + "loss": 0.2362, + "step": 1120 + }, + { + "epoch": 0.5365622032288699, + "grad_norm": 0.8927081511146825, + "learning_rate": 9.99841327302198e-05, + "loss": 0.2251, + "step": 1130 + }, + { + "epoch": 0.5413105413105413, + "grad_norm": 0.9732415935122696, + "learning_rate": 9.997968272397542e-05, + "loss": 0.2222, + "step": 1140 + }, + { + "epoch": 0.5460588793922128, + "grad_norm": 0.9560857920847607, + "learning_rate": 9.997468349173985e-05, + "loss": 0.2189, + "step": 1150 + }, + { + "epoch": 0.5508072174738842, + "grad_norm": 0.924098382546985, + "learning_rate": 9.99691350884496e-05, + "loss": 0.2179, + "step": 1160 + }, + { + "epoch": 0.5555555555555556, + "grad_norm": 0.828396782450117, + "learning_rate": 9.996303757507602e-05, + "loss": 0.211, + "step": 1170 + }, + { + "epoch": 0.560303893637227, + "grad_norm": 0.9333663244449117, + "learning_rate": 9.995639101862453e-05, + "loss": 0.2102, + "step": 1180 + }, + { + "epoch": 0.5650522317188984, + "grad_norm": 0.9298060626900368, + "learning_rate": 9.994919549213408e-05, + "loss": 0.2074, + "step": 1190 + }, + { + "epoch": 0.5698005698005698, + "grad_norm": 0.7962626517135125, + "learning_rate": 9.99414510746762e-05, + "loss": 0.203, + "step": 1200 + }, + { + "epoch": 0.5745489078822412, + "grad_norm": 0.8580673639697189, + "learning_rate": 9.993315785135416e-05, + "loss": 0.2013, + "step": 1210 + }, + { + "epoch": 0.5792972459639126, + "grad_norm": 0.8048698470169939, + "learning_rate": 9.99243159133021e-05, + "loss": 0.1975, + "step": 1220 + }, + { + "epoch": 0.584045584045584, + "grad_norm": 0.9257889896229621, + "learning_rate": 9.991492535768392e-05, + "loss": 0.1946, + "step": 1230 + }, + { + "epoch": 0.5887939221272555, + "grad_norm": 0.7558640985069887, + "learning_rate": 9.99049862876923e-05, + "loss": 0.1925, + "step": 1240 + }, + { + "epoch": 0.5935422602089269, + "grad_norm": 0.8056728442959841, + "learning_rate": 9.989449881254754e-05, + "loss": 0.1891, + "step": 1250 + }, + { + "epoch": 0.5982905982905983, + "grad_norm": 0.8438539837910541, + "learning_rate": 9.988346304749635e-05, + "loss": 0.1862, + "step": 1260 + }, + { + "epoch": 0.6030389363722697, + "grad_norm": 0.7510354369473087, + "learning_rate": 9.987187911381059e-05, + "loss": 0.1841, + "step": 1270 + }, + { + "epoch": 0.6077872744539411, + "grad_norm": 0.7636415261243265, + "learning_rate": 9.985974713878592e-05, + "loss": 0.1808, + "step": 1280 + }, + { + "epoch": 0.6125356125356125, + "grad_norm": 0.7974385042371772, + "learning_rate": 9.984706725574044e-05, + "loss": 0.1792, + "step": 1290 + }, + { + "epoch": 0.6172839506172839, + "grad_norm": 0.7789431577859948, + "learning_rate": 9.983383960401319e-05, + "loss": 0.1756, + "step": 1300 + }, + { + "epoch": 0.6220322886989553, + "grad_norm": 0.7722426337147819, + "learning_rate": 9.982006432896266e-05, + "loss": 0.176, + "step": 1310 + }, + { + "epoch": 0.6267806267806267, + "grad_norm": 0.7383737481941087, + "learning_rate": 9.980574158196512e-05, + "loss": 0.1723, + "step": 1320 + }, + { + "epoch": 0.6315289648622981, + "grad_norm": 0.7442324964032874, + "learning_rate": 9.979087152041304e-05, + "loss": 0.1698, + "step": 1330 + }, + { + "epoch": 0.6362773029439696, + "grad_norm": 0.7497638974160803, + "learning_rate": 9.977545430771332e-05, + "loss": 0.1673, + "step": 1340 + }, + { + "epoch": 0.6410256410256411, + "grad_norm": 0.8548555257280125, + "learning_rate": 9.975949011328542e-05, + "loss": 0.1636, + "step": 1350 + }, + { + "epoch": 0.6457739791073125, + "grad_norm": 0.8279981954343867, + "learning_rate": 9.97429791125597e-05, + "loss": 0.1655, + "step": 1360 + }, + { + "epoch": 0.6505223171889839, + "grad_norm": 0.6959990480791529, + "learning_rate": 9.972592148697528e-05, + "loss": 0.1612, + "step": 1370 + }, + { + "epoch": 0.6552706552706553, + "grad_norm": 0.828078740940128, + "learning_rate": 9.970831742397815e-05, + "loss": 0.1612, + "step": 1380 + }, + { + "epoch": 0.6600189933523267, + "grad_norm": 0.77793285129046, + "learning_rate": 9.969016711701908e-05, + "loss": 0.1611, + "step": 1390 + }, + { + "epoch": 0.6647673314339981, + "grad_norm": 0.7194789618958793, + "learning_rate": 9.967147076555157e-05, + "loss": 0.1579, + "step": 1400 + }, + { + "epoch": 0.6695156695156695, + "grad_norm": 0.7100892012167056, + "learning_rate": 9.965222857502951e-05, + "loss": 0.1564, + "step": 1410 + }, + { + "epoch": 0.674264007597341, + "grad_norm": 0.7535295766221981, + "learning_rate": 9.963244075690505e-05, + "loss": 0.1519, + "step": 1420 + }, + { + "epoch": 0.6790123456790124, + "grad_norm": 0.706459243665132, + "learning_rate": 9.961210752862626e-05, + "loss": 0.1548, + "step": 1430 + }, + { + "epoch": 0.6837606837606838, + "grad_norm": 0.6799536660584122, + "learning_rate": 9.959122911363463e-05, + "loss": 0.1489, + "step": 1440 + }, + { + "epoch": 0.6885090218423552, + "grad_norm": 0.750734170664292, + "learning_rate": 9.956980574136281e-05, + "loss": 0.1492, + "step": 1450 + }, + { + "epoch": 0.6932573599240266, + "grad_norm": 0.7631275022947855, + "learning_rate": 9.954783764723188e-05, + "loss": 0.1469, + "step": 1460 + }, + { + "epoch": 0.698005698005698, + "grad_norm": 0.7140096944520468, + "learning_rate": 9.952532507264891e-05, + "loss": 0.1456, + "step": 1470 + }, + { + "epoch": 0.7027540360873694, + "grad_norm": 0.6820968891017294, + "learning_rate": 9.950226826500426e-05, + "loss": 0.1459, + "step": 1480 + }, + { + "epoch": 0.7075023741690408, + "grad_norm": 0.6573844442119157, + "learning_rate": 9.947866747766882e-05, + "loss": 0.1432, + "step": 1490 + }, + { + "epoch": 0.7122507122507122, + "grad_norm": 0.6477315679395246, + "learning_rate": 9.945452296999131e-05, + "loss": 0.1418, + "step": 1500 + }, + { + "epoch": 0.7169990503323836, + "grad_norm": 0.7854073646179602, + "learning_rate": 9.942983500729536e-05, + "loss": 0.1415, + "step": 1510 + }, + { + "epoch": 0.7217473884140551, + "grad_norm": 0.6563280291041728, + "learning_rate": 9.940460386087667e-05, + "loss": 0.1396, + "step": 1520 + }, + { + "epoch": 0.7264957264957265, + "grad_norm": 0.6777516445796997, + "learning_rate": 9.937882980799986e-05, + "loss": 0.1374, + "step": 1530 + }, + { + "epoch": 0.7312440645773979, + "grad_norm": 0.6482850562773214, + "learning_rate": 9.935251313189564e-05, + "loss": 0.1341, + "step": 1540 + }, + { + "epoch": 0.7359924026590693, + "grad_norm": 0.6807447861160003, + "learning_rate": 9.93256541217576e-05, + "loss": 0.1328, + "step": 1550 + }, + { + "epoch": 0.7407407407407407, + "grad_norm": 0.6096762595093985, + "learning_rate": 9.929825307273898e-05, + "loss": 0.1323, + "step": 1560 + }, + { + "epoch": 0.7454890788224121, + "grad_norm": 0.6758283907647111, + "learning_rate": 9.927031028594951e-05, + "loss": 0.1323, + "step": 1570 + }, + { + "epoch": 0.7502374169040835, + "grad_norm": 0.6661507717312829, + "learning_rate": 9.924182606845204e-05, + "loss": 0.1292, + "step": 1580 + }, + { + "epoch": 0.7549857549857549, + "grad_norm": 0.7190496153063343, + "learning_rate": 9.921280073325922e-05, + "loss": 0.1288, + "step": 1590 + }, + { + "epoch": 0.7597340930674265, + "grad_norm": 0.7792128467633825, + "learning_rate": 9.918323459933005e-05, + "loss": 0.1315, + "step": 1600 + }, + { + "epoch": 0.7644824311490979, + "grad_norm": 0.7476110366866813, + "learning_rate": 9.915312799156632e-05, + "loss": 0.127, + "step": 1610 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.723682326262704, + "learning_rate": 9.91224812408091e-05, + "loss": 0.1252, + "step": 1620 + }, + { + "epoch": 0.7739791073124407, + "grad_norm": 0.6763342918277632, + "learning_rate": 9.909129468383506e-05, + "loss": 0.1229, + "step": 1630 + }, + { + "epoch": 0.7787274453941121, + "grad_norm": 0.6756385241786282, + "learning_rate": 9.905956866335286e-05, + "loss": 0.1223, + "step": 1640 + }, + { + "epoch": 0.7834757834757835, + "grad_norm": 0.7031791508138145, + "learning_rate": 9.90273035279992e-05, + "loss": 0.121, + "step": 1650 + }, + { + "epoch": 0.7882241215574549, + "grad_norm": 0.6901786661268531, + "learning_rate": 9.899449963233523e-05, + "loss": 0.1218, + "step": 1660 + }, + { + "epoch": 0.7929724596391263, + "grad_norm": 0.655344108091334, + "learning_rate": 9.896115733684244e-05, + "loss": 0.1186, + "step": 1670 + }, + { + "epoch": 0.7977207977207977, + "grad_norm": 0.6429227026044089, + "learning_rate": 9.892727700791881e-05, + "loss": 0.119, + "step": 1680 + }, + { + "epoch": 0.8024691358024691, + "grad_norm": 0.6445284291633588, + "learning_rate": 9.889285901787482e-05, + "loss": 0.1171, + "step": 1690 + }, + { + "epoch": 0.8072174738841406, + "grad_norm": 0.6599500074332483, + "learning_rate": 9.885790374492921e-05, + "loss": 0.114, + "step": 1700 + }, + { + "epoch": 0.811965811965812, + "grad_norm": 0.6458817345788224, + "learning_rate": 9.882241157320497e-05, + "loss": 0.1161, + "step": 1710 + }, + { + "epoch": 0.8167141500474834, + "grad_norm": 0.6510623099667147, + "learning_rate": 9.878638289272507e-05, + "loss": 0.1134, + "step": 1720 + }, + { + "epoch": 0.8214624881291548, + "grad_norm": 0.711735747979104, + "learning_rate": 9.874981809940808e-05, + "loss": 0.1133, + "step": 1730 + }, + { + "epoch": 0.8262108262108262, + "grad_norm": 0.6067344669903363, + "learning_rate": 9.871271759506404e-05, + "loss": 0.111, + "step": 1740 + }, + { + "epoch": 0.8309591642924976, + "grad_norm": 0.6435729954519224, + "learning_rate": 9.86750817873898e-05, + "loss": 0.1101, + "step": 1750 + }, + { + "epoch": 0.835707502374169, + "grad_norm": 0.6594866421615497, + "learning_rate": 9.863691108996471e-05, + "loss": 0.1102, + "step": 1760 + }, + { + "epoch": 0.8404558404558404, + "grad_norm": 0.6463645658702524, + "learning_rate": 9.859820592224599e-05, + "loss": 0.1086, + "step": 1770 + }, + { + "epoch": 0.8452041785375118, + "grad_norm": 0.6384759072033943, + "learning_rate": 9.855896670956415e-05, + "loss": 0.107, + "step": 1780 + }, + { + "epoch": 0.8499525166191833, + "grad_norm": 0.5925061730095992, + "learning_rate": 9.851919388311832e-05, + "loss": 0.1057, + "step": 1790 + }, + { + "epoch": 0.8547008547008547, + "grad_norm": 0.5958838267885576, + "learning_rate": 9.84788878799715e-05, + "loss": 0.1063, + "step": 1800 + }, + { + "epoch": 0.8594491927825261, + "grad_norm": 0.583253660596266, + "learning_rate": 9.843804914304577e-05, + "loss": 0.1033, + "step": 1810 + }, + { + "epoch": 0.8641975308641975, + "grad_norm": 0.6509685918136744, + "learning_rate": 9.839667812111739e-05, + "loss": 0.1039, + "step": 1820 + }, + { + "epoch": 0.8689458689458689, + "grad_norm": 0.6654859819781469, + "learning_rate": 9.835477526881189e-05, + "loss": 0.1026, + "step": 1830 + }, + { + "epoch": 0.8736942070275404, + "grad_norm": 0.6987159643473111, + "learning_rate": 9.831234104659913e-05, + "loss": 0.1015, + "step": 1840 + }, + { + "epoch": 0.8784425451092118, + "grad_norm": 0.5650671462210827, + "learning_rate": 9.826937592078811e-05, + "loss": 0.1011, + "step": 1850 + }, + { + "epoch": 0.8831908831908832, + "grad_norm": 0.5727141961302766, + "learning_rate": 9.822588036352201e-05, + "loss": 0.1012, + "step": 1860 + }, + { + "epoch": 0.8879392212725546, + "grad_norm": 0.6014103938509006, + "learning_rate": 9.81818548527728e-05, + "loss": 0.099, + "step": 1870 + }, + { + "epoch": 0.8926875593542261, + "grad_norm": 0.5734563478604482, + "learning_rate": 9.81372998723362e-05, + "loss": 0.0975, + "step": 1880 + }, + { + "epoch": 0.8974358974358975, + "grad_norm": 0.5684437429333495, + "learning_rate": 9.809221591182623e-05, + "loss": 0.0954, + "step": 1890 + }, + { + "epoch": 0.9021842355175689, + "grad_norm": 0.5550460249948803, + "learning_rate": 9.804660346666984e-05, + "loss": 0.0948, + "step": 1900 + }, + { + "epoch": 0.9069325735992403, + "grad_norm": 0.6030717817112082, + "learning_rate": 9.800046303810154e-05, + "loss": 0.095, + "step": 1910 + }, + { + "epoch": 0.9116809116809117, + "grad_norm": 0.5786096312033695, + "learning_rate": 9.795379513315779e-05, + "loss": 0.0964, + "step": 1920 + }, + { + "epoch": 0.9164292497625831, + "grad_norm": 0.5342317355454276, + "learning_rate": 9.790660026467153e-05, + "loss": 0.0917, + "step": 1930 + }, + { + "epoch": 0.9211775878442545, + "grad_norm": 0.5729999943038201, + "learning_rate": 9.785887895126644e-05, + "loss": 0.0934, + "step": 1940 + }, + { + "epoch": 0.9259259259259259, + "grad_norm": 0.5148337352548606, + "learning_rate": 9.781063171735134e-05, + "loss": 0.0912, + "step": 1950 + }, + { + "epoch": 0.9306742640075973, + "grad_norm": 0.6096023114073594, + "learning_rate": 9.776185909311437e-05, + "loss": 0.0934, + "step": 1960 + }, + { + "epoch": 0.9354226020892688, + "grad_norm": 0.5953694391053918, + "learning_rate": 9.771256161451716e-05, + "loss": 0.0915, + "step": 1970 + }, + { + "epoch": 0.9401709401709402, + "grad_norm": 0.5553421636649856, + "learning_rate": 9.766273982328899e-05, + "loss": 0.0892, + "step": 1980 + }, + { + "epoch": 0.9449192782526116, + "grad_norm": 0.6314566035516022, + "learning_rate": 9.761239426692077e-05, + "loss": 0.0892, + "step": 1990 + }, + { + "epoch": 0.949667616334283, + "grad_norm": 0.553477341094825, + "learning_rate": 9.756152549865907e-05, + "loss": 0.0868, + "step": 2000 + }, + { + "epoch": 0.9544159544159544, + "grad_norm": 0.5855428558450079, + "learning_rate": 9.751013407750004e-05, + "loss": 0.0879, + "step": 2010 + }, + { + "epoch": 0.9591642924976258, + "grad_norm": 0.5564124650704858, + "learning_rate": 9.745822056818324e-05, + "loss": 0.0878, + "step": 2020 + }, + { + "epoch": 0.9639126305792972, + "grad_norm": 0.5648708660408431, + "learning_rate": 9.740578554118545e-05, + "loss": 0.0851, + "step": 2030 + }, + { + "epoch": 0.9686609686609686, + "grad_norm": 0.6100865208168226, + "learning_rate": 9.735282957271438e-05, + "loss": 0.0857, + "step": 2040 + }, + { + "epoch": 0.97340930674264, + "grad_norm": 0.5342516016564061, + "learning_rate": 9.729935324470243e-05, + "loss": 0.0833, + "step": 2050 + }, + { + "epoch": 0.9781576448243114, + "grad_norm": 0.5617810297891285, + "learning_rate": 9.724535714480012e-05, + "loss": 0.0823, + "step": 2060 + }, + { + "epoch": 0.9829059829059829, + "grad_norm": 0.4874563065093935, + "learning_rate": 9.719084186636984e-05, + "loss": 0.0829, + "step": 2070 + }, + { + "epoch": 0.9876543209876543, + "grad_norm": 0.5925800844469832, + "learning_rate": 9.713580800847916e-05, + "loss": 0.0816, + "step": 2080 + }, + { + "epoch": 0.9924026590693258, + "grad_norm": 0.5192721370514558, + "learning_rate": 9.708025617589431e-05, + "loss": 0.0801, + "step": 2090 + }, + { + "epoch": 0.9971509971509972, + "grad_norm": 0.5317846561760016, + "learning_rate": 9.702418697907357e-05, + "loss": 0.08, + "step": 2100 + }, + { + "epoch": 1.0018993352326686, + "grad_norm": 0.5375175588207268, + "learning_rate": 9.696760103416049e-05, + "loss": 0.0765, + "step": 2110 + }, + { + "epoch": 1.00664767331434, + "grad_norm": 0.4995120344713725, + "learning_rate": 9.69104989629772e-05, + "loss": 0.0686, + "step": 2120 + }, + { + "epoch": 1.0113960113960114, + "grad_norm": 0.492492367724788, + "learning_rate": 9.685288139301749e-05, + "loss": 0.0676, + "step": 2130 + }, + { + "epoch": 1.0161443494776827, + "grad_norm": 0.489910902697706, + "learning_rate": 9.679474895744002e-05, + "loss": 0.0673, + "step": 2140 + }, + { + "epoch": 1.0208926875593543, + "grad_norm": 0.4842805102412133, + "learning_rate": 9.673610229506123e-05, + "loss": 0.0657, + "step": 2150 + }, + { + "epoch": 1.0256410256410255, + "grad_norm": 0.500590705014481, + "learning_rate": 9.667694205034842e-05, + "loss": 0.065, + "step": 2160 + }, + { + "epoch": 1.030389363722697, + "grad_norm": 0.5412799666786388, + "learning_rate": 9.661726887341264e-05, + "loss": 0.0661, + "step": 2170 + }, + { + "epoch": 1.0351377018043684, + "grad_norm": 0.49596368987818673, + "learning_rate": 9.655708342000158e-05, + "loss": 0.066, + "step": 2180 + }, + { + "epoch": 1.03988603988604, + "grad_norm": 0.5133006065434564, + "learning_rate": 9.649638635149225e-05, + "loss": 0.0633, + "step": 2190 + }, + { + "epoch": 1.0446343779677112, + "grad_norm": 0.5079148873200521, + "learning_rate": 9.643517833488387e-05, + "loss": 0.0649, + "step": 2200 + }, + { + "epoch": 1.0493827160493827, + "grad_norm": 0.5222502694032142, + "learning_rate": 9.637346004279038e-05, + "loss": 0.0635, + "step": 2210 + }, + { + "epoch": 1.0541310541310542, + "grad_norm": 0.5104557333995293, + "learning_rate": 9.631123215343323e-05, + "loss": 0.0618, + "step": 2220 + }, + { + "epoch": 1.0588793922127255, + "grad_norm": 0.5024303716694327, + "learning_rate": 9.624849535063373e-05, + "loss": 0.0635, + "step": 2230 + }, + { + "epoch": 1.063627730294397, + "grad_norm": 0.5353909988509568, + "learning_rate": 9.61852503238057e-05, + "loss": 0.0623, + "step": 2240 + }, + { + "epoch": 1.0683760683760684, + "grad_norm": 0.548569047624678, + "learning_rate": 9.612149776794776e-05, + "loss": 0.0627, + "step": 2250 + }, + { + "epoch": 1.0731244064577399, + "grad_norm": 0.49061114729721145, + "learning_rate": 9.605723838363586e-05, + "loss": 0.0617, + "step": 2260 + }, + { + "epoch": 1.0778727445394112, + "grad_norm": 0.5118467274517943, + "learning_rate": 9.599247287701539e-05, + "loss": 0.0613, + "step": 2270 + }, + { + "epoch": 1.0826210826210827, + "grad_norm": 0.5106993610815542, + "learning_rate": 9.592720195979353e-05, + "loss": 0.0616, + "step": 2280 + }, + { + "epoch": 1.087369420702754, + "grad_norm": 0.533554957708412, + "learning_rate": 9.586142634923141e-05, + "loss": 0.0607, + "step": 2290 + }, + { + "epoch": 1.0921177587844255, + "grad_norm": 0.4961988723784374, + "learning_rate": 9.579514676813627e-05, + "loss": 0.0603, + "step": 2300 + }, + { + "epoch": 1.0968660968660968, + "grad_norm": 0.5291731061070475, + "learning_rate": 9.572836394485345e-05, + "loss": 0.0578, + "step": 2310 + }, + { + "epoch": 1.1016144349477683, + "grad_norm": 0.4560676421604984, + "learning_rate": 9.566107861325836e-05, + "loss": 0.0585, + "step": 2320 + }, + { + "epoch": 1.1063627730294396, + "grad_norm": 0.5078075639209664, + "learning_rate": 9.559329151274854e-05, + "loss": 0.0578, + "step": 2330 + }, + { + "epoch": 1.1111111111111112, + "grad_norm": 0.4683086272270788, + "learning_rate": 9.552500338823543e-05, + "loss": 0.0562, + "step": 2340 + }, + { + "epoch": 1.1158594491927825, + "grad_norm": 0.49903313629075563, + "learning_rate": 9.54562149901362e-05, + "loss": 0.0574, + "step": 2350 + }, + { + "epoch": 1.120607787274454, + "grad_norm": 0.4952267832995423, + "learning_rate": 9.538692707436554e-05, + "loss": 0.0566, + "step": 2360 + }, + { + "epoch": 1.1253561253561253, + "grad_norm": 0.5143759315257765, + "learning_rate": 9.531714040232733e-05, + "loss": 0.0568, + "step": 2370 + }, + { + "epoch": 1.1301044634377968, + "grad_norm": 0.5333912877931387, + "learning_rate": 9.524685574090627e-05, + "loss": 0.0554, + "step": 2380 + }, + { + "epoch": 1.134852801519468, + "grad_norm": 0.506610426425779, + "learning_rate": 9.517607386245948e-05, + "loss": 0.0557, + "step": 2390 + }, + { + "epoch": 1.1396011396011396, + "grad_norm": 0.4807568687194227, + "learning_rate": 9.510479554480793e-05, + "loss": 0.057, + "step": 2400 + }, + { + "epoch": 1.144349477682811, + "grad_norm": 0.5059220071837027, + "learning_rate": 9.503302157122804e-05, + "loss": 0.0564, + "step": 2410 + }, + { + "epoch": 1.1490978157644824, + "grad_norm": 0.5321634332501042, + "learning_rate": 9.496075273044293e-05, + "loss": 0.0556, + "step": 2420 + }, + { + "epoch": 1.1538461538461537, + "grad_norm": 0.4860319263192551, + "learning_rate": 9.488798981661379e-05, + "loss": 0.0556, + "step": 2430 + }, + { + "epoch": 1.1585944919278253, + "grad_norm": 0.5450716546784171, + "learning_rate": 9.481473362933122e-05, + "loss": 0.0549, + "step": 2440 + }, + { + "epoch": 1.1633428300094968, + "grad_norm": 0.48569228023641087, + "learning_rate": 9.474098497360637e-05, + "loss": 0.0557, + "step": 2450 + }, + { + "epoch": 1.168091168091168, + "grad_norm": 0.5215152833823479, + "learning_rate": 9.466674465986211e-05, + "loss": 0.0552, + "step": 2460 + }, + { + "epoch": 1.1728395061728394, + "grad_norm": 0.4777704133253495, + "learning_rate": 9.459201350392418e-05, + "loss": 0.054, + "step": 2470 + }, + { + "epoch": 1.177587844254511, + "grad_norm": 0.4825911992094268, + "learning_rate": 9.451679232701212e-05, + "loss": 0.0531, + "step": 2480 + }, + { + "epoch": 1.1823361823361824, + "grad_norm": 0.5256696379086168, + "learning_rate": 9.444108195573031e-05, + "loss": 0.0546, + "step": 2490 + }, + { + "epoch": 1.1870845204178537, + "grad_norm": 0.48041145273942765, + "learning_rate": 9.436488322205892e-05, + "loss": 0.0547, + "step": 2500 + }, + { + "epoch": 1.1918328584995252, + "grad_norm": 0.49889044556441, + "learning_rate": 9.42881969633447e-05, + "loss": 0.052, + "step": 2510 + }, + { + "epoch": 1.1965811965811965, + "grad_norm": 0.4691225975029018, + "learning_rate": 9.42110240222918e-05, + "loss": 0.0526, + "step": 2520 + }, + { + "epoch": 1.201329534662868, + "grad_norm": 0.4699543822621789, + "learning_rate": 9.413336524695253e-05, + "loss": 0.0529, + "step": 2530 + }, + { + "epoch": 1.2060778727445394, + "grad_norm": 0.5092846176526173, + "learning_rate": 9.405522149071801e-05, + "loss": 0.0515, + "step": 2540 + }, + { + "epoch": 1.210826210826211, + "grad_norm": 0.48822278879655134, + "learning_rate": 9.397659361230883e-05, + "loss": 0.0518, + "step": 2550 + }, + { + "epoch": 1.2155745489078822, + "grad_norm": 0.4709051625946915, + "learning_rate": 9.389748247576557e-05, + "loss": 0.0513, + "step": 2560 + }, + { + "epoch": 1.2203228869895537, + "grad_norm": 0.43081313275948346, + "learning_rate": 9.381788895043933e-05, + "loss": 0.052, + "step": 2570 + }, + { + "epoch": 1.225071225071225, + "grad_norm": 0.4618949718938529, + "learning_rate": 9.373781391098218e-05, + "loss": 0.0509, + "step": 2580 + }, + { + "epoch": 1.2298195631528965, + "grad_norm": 0.4414468310558702, + "learning_rate": 9.365725823733753e-05, + "loss": 0.0505, + "step": 2590 + }, + { + "epoch": 1.2345679012345678, + "grad_norm": 0.4437487857178328, + "learning_rate": 9.35762228147305e-05, + "loss": 0.0496, + "step": 2600 + }, + { + "epoch": 1.2393162393162394, + "grad_norm": 0.4644815359436171, + "learning_rate": 9.349470853365815e-05, + "loss": 0.0491, + "step": 2610 + }, + { + "epoch": 1.2440645773979107, + "grad_norm": 0.4604315887784035, + "learning_rate": 9.341271628987968e-05, + "loss": 0.0501, + "step": 2620 + }, + { + "epoch": 1.2488129154795822, + "grad_norm": 0.45836637793826146, + "learning_rate": 9.333024698440667e-05, + "loss": 0.0491, + "step": 2630 + }, + { + "epoch": 1.2535612535612537, + "grad_norm": 0.4004595257683082, + "learning_rate": 9.324730152349306e-05, + "loss": 0.0493, + "step": 2640 + }, + { + "epoch": 1.258309591642925, + "grad_norm": 0.4588694709958647, + "learning_rate": 9.31638808186253e-05, + "loss": 0.0485, + "step": 2650 + }, + { + "epoch": 1.2630579297245963, + "grad_norm": 0.4289599019381277, + "learning_rate": 9.307998578651228e-05, + "loss": 0.049, + "step": 2660 + }, + { + "epoch": 1.2678062678062678, + "grad_norm": 0.4727026865591144, + "learning_rate": 9.299561734907528e-05, + "loss": 0.0459, + "step": 2670 + }, + { + "epoch": 1.2725546058879393, + "grad_norm": 0.4229759230480915, + "learning_rate": 9.291077643343776e-05, + "loss": 0.0477, + "step": 2680 + }, + { + "epoch": 1.2773029439696106, + "grad_norm": 0.43707322690226125, + "learning_rate": 9.282546397191531e-05, + "loss": 0.0474, + "step": 2690 + }, + { + "epoch": 1.282051282051282, + "grad_norm": 0.47184366398017824, + "learning_rate": 9.273968090200533e-05, + "loss": 0.0472, + "step": 2700 + }, + { + "epoch": 1.2867996201329535, + "grad_norm": 0.5027834026428316, + "learning_rate": 9.265342816637666e-05, + "loss": 0.0465, + "step": 2710 + }, + { + "epoch": 1.291547958214625, + "grad_norm": 0.4454440035871166, + "learning_rate": 9.256670671285934e-05, + "loss": 0.0462, + "step": 2720 + }, + { + "epoch": 1.2962962962962963, + "grad_norm": 0.43536614903814524, + "learning_rate": 9.247951749443413e-05, + "loss": 0.0459, + "step": 2730 + }, + { + "epoch": 1.3010446343779676, + "grad_norm": 0.49496070699402434, + "learning_rate": 9.239186146922208e-05, + "loss": 0.0451, + "step": 2740 + }, + { + "epoch": 1.305792972459639, + "grad_norm": 0.43375607265420135, + "learning_rate": 9.230373960047394e-05, + "loss": 0.0454, + "step": 2750 + }, + { + "epoch": 1.3105413105413106, + "grad_norm": 0.42267249461132017, + "learning_rate": 9.22151528565596e-05, + "loss": 0.045, + "step": 2760 + }, + { + "epoch": 1.315289648622982, + "grad_norm": 0.47137884223204835, + "learning_rate": 9.212610221095748e-05, + "loss": 0.0453, + "step": 2770 + }, + { + "epoch": 1.3200379867046534, + "grad_norm": 0.5199454437668323, + "learning_rate": 9.203658864224382e-05, + "loss": 0.0439, + "step": 2780 + }, + { + "epoch": 1.3247863247863247, + "grad_norm": 0.4570822874899115, + "learning_rate": 9.194661313408187e-05, + "loss": 0.044, + "step": 2790 + }, + { + "epoch": 1.3295346628679963, + "grad_norm": 0.43113616092372226, + "learning_rate": 9.185617667521117e-05, + "loss": 0.0444, + "step": 2800 + }, + { + "epoch": 1.3342830009496676, + "grad_norm": 0.4222760906355151, + "learning_rate": 9.176528025943663e-05, + "loss": 0.0444, + "step": 2810 + }, + { + "epoch": 1.339031339031339, + "grad_norm": 0.4952066697298851, + "learning_rate": 9.167392488561758e-05, + "loss": 0.0433, + "step": 2820 + }, + { + "epoch": 1.3437796771130104, + "grad_norm": 0.46359163249702656, + "learning_rate": 9.158211155765692e-05, + "loss": 0.0432, + "step": 2830 + }, + { + "epoch": 1.348528015194682, + "grad_norm": 0.47893564069863104, + "learning_rate": 9.14898412844899e-05, + "loss": 0.0437, + "step": 2840 + }, + { + "epoch": 1.3532763532763532, + "grad_norm": 0.4011060385937812, + "learning_rate": 9.13971150800732e-05, + "loss": 0.0414, + "step": 2850 + }, + { + "epoch": 1.3580246913580247, + "grad_norm": 0.46085185858271926, + "learning_rate": 9.130393396337368e-05, + "loss": 0.0418, + "step": 2860 + }, + { + "epoch": 1.362773029439696, + "grad_norm": 0.46535476597347625, + "learning_rate": 9.121029895835731e-05, + "loss": 0.042, + "step": 2870 + }, + { + "epoch": 1.3675213675213675, + "grad_norm": 0.39410170446969295, + "learning_rate": 9.111621109397769e-05, + "loss": 0.0418, + "step": 2880 + }, + { + "epoch": 1.3722697056030388, + "grad_norm": 0.4243876470935551, + "learning_rate": 9.102167140416503e-05, + "loss": 0.041, + "step": 2890 + }, + { + "epoch": 1.3770180436847104, + "grad_norm": 0.4585829009513368, + "learning_rate": 9.092668092781454e-05, + "loss": 0.0414, + "step": 2900 + }, + { + "epoch": 1.381766381766382, + "grad_norm": 0.4262377269345021, + "learning_rate": 9.083124070877516e-05, + "loss": 0.042, + "step": 2910 + }, + { + "epoch": 1.3865147198480532, + "grad_norm": 0.40844548158991334, + "learning_rate": 9.073535179583806e-05, + "loss": 0.0419, + "step": 2920 + }, + { + "epoch": 1.3912630579297245, + "grad_norm": 0.3935442259634633, + "learning_rate": 9.063901524272505e-05, + "loss": 0.0412, + "step": 2930 + }, + { + "epoch": 1.396011396011396, + "grad_norm": 0.3693483555639858, + "learning_rate": 9.054223210807709e-05, + "loss": 0.0402, + "step": 2940 + }, + { + "epoch": 1.4007597340930675, + "grad_norm": 0.40078308269859275, + "learning_rate": 9.044500345544262e-05, + "loss": 0.04, + "step": 2950 + }, + { + "epoch": 1.4055080721747388, + "grad_norm": 0.414448129205789, + "learning_rate": 9.034733035326584e-05, + "loss": 0.0419, + "step": 2960 + }, + { + "epoch": 1.4102564102564101, + "grad_norm": 0.4314293044106671, + "learning_rate": 9.024921387487504e-05, + "loss": 0.0395, + "step": 2970 + }, + { + "epoch": 1.4150047483380817, + "grad_norm": 0.3830262611826239, + "learning_rate": 9.015065509847075e-05, + "loss": 0.0385, + "step": 2980 + }, + { + "epoch": 1.4197530864197532, + "grad_norm": 0.40774106075840866, + "learning_rate": 9.005165510711387e-05, + "loss": 0.0393, + "step": 2990 + }, + { + "epoch": 1.4245014245014245, + "grad_norm": 0.4072063711912796, + "learning_rate": 8.995221498871387e-05, + "loss": 0.0376, + "step": 3000 + }, + { + "epoch": 1.429249762583096, + "grad_norm": 0.418318685699483, + "learning_rate": 8.98523358360167e-05, + "loss": 0.0388, + "step": 3010 + }, + { + "epoch": 1.4339981006647673, + "grad_norm": 0.38188198729370343, + "learning_rate": 8.975201874659292e-05, + "loss": 0.0377, + "step": 3020 + }, + { + "epoch": 1.4387464387464388, + "grad_norm": 0.4240963163227532, + "learning_rate": 8.965126482282549e-05, + "loss": 0.0394, + "step": 3030 + }, + { + "epoch": 1.4434947768281101, + "grad_norm": 0.40891409921975486, + "learning_rate": 8.955007517189784e-05, + "loss": 0.0389, + "step": 3040 + }, + { + "epoch": 1.4482431149097816, + "grad_norm": 0.4343466782916782, + "learning_rate": 8.944845090578149e-05, + "loss": 0.038, + "step": 3050 + }, + { + "epoch": 1.452991452991453, + "grad_norm": 0.47914926317681983, + "learning_rate": 8.934639314122401e-05, + "loss": 0.0383, + "step": 3060 + }, + { + "epoch": 1.4577397910731245, + "grad_norm": 0.4726822552336102, + "learning_rate": 8.924390299973666e-05, + "loss": 0.037, + "step": 3070 + }, + { + "epoch": 1.4624881291547958, + "grad_norm": 0.40639896910669704, + "learning_rate": 8.91409816075821e-05, + "loss": 0.0379, + "step": 3080 + }, + { + "epoch": 1.4672364672364673, + "grad_norm": 0.4297631622330081, + "learning_rate": 8.903763009576195e-05, + "loss": 0.0378, + "step": 3090 + }, + { + "epoch": 1.4719848053181386, + "grad_norm": 0.397553702278778, + "learning_rate": 8.893384960000445e-05, + "loss": 0.0366, + "step": 3100 + }, + { + "epoch": 1.47673314339981, + "grad_norm": 0.3842838408196586, + "learning_rate": 8.882964126075192e-05, + "loss": 0.0355, + "step": 3110 + }, + { + "epoch": 1.4814814814814814, + "grad_norm": 0.4087672118416129, + "learning_rate": 8.872500622314829e-05, + "loss": 0.0363, + "step": 3120 + }, + { + "epoch": 1.486229819563153, + "grad_norm": 0.4098031203055733, + "learning_rate": 8.861994563702638e-05, + "loss": 0.035, + "step": 3130 + }, + { + "epoch": 1.4909781576448244, + "grad_norm": 0.3876997066775908, + "learning_rate": 8.851446065689544e-05, + "loss": 0.0364, + "step": 3140 + }, + { + "epoch": 1.4957264957264957, + "grad_norm": 0.4134151182799379, + "learning_rate": 8.840855244192836e-05, + "loss": 0.0349, + "step": 3150 + }, + { + "epoch": 1.500474833808167, + "grad_norm": 0.3927794557722139, + "learning_rate": 8.83022221559489e-05, + "loss": 0.0357, + "step": 3160 + }, + { + "epoch": 1.5052231718898386, + "grad_norm": 0.41635322736234687, + "learning_rate": 8.819547096741904e-05, + "loss": 0.0346, + "step": 3170 + }, + { + "epoch": 1.50997150997151, + "grad_norm": 0.3773849224872641, + "learning_rate": 8.8088300049426e-05, + "loss": 0.0342, + "step": 3180 + }, + { + "epoch": 1.5147198480531814, + "grad_norm": 0.34735786246540556, + "learning_rate": 8.798071057966935e-05, + "loss": 0.0335, + "step": 3190 + }, + { + "epoch": 1.5194681861348527, + "grad_norm": 0.39208660749913393, + "learning_rate": 8.787270374044824e-05, + "loss": 0.0351, + "step": 3200 + }, + { + "epoch": 1.5242165242165242, + "grad_norm": 0.4048707970102346, + "learning_rate": 8.776428071864819e-05, + "loss": 0.0336, + "step": 3210 + }, + { + "epoch": 1.5289648622981957, + "grad_norm": 0.3843214639707809, + "learning_rate": 8.76554427057282e-05, + "loss": 0.0338, + "step": 3220 + }, + { + "epoch": 1.533713200379867, + "grad_norm": 0.4132367494817212, + "learning_rate": 8.754619089770755e-05, + "loss": 0.0336, + "step": 3230 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.38075123816436185, + "learning_rate": 8.743652649515275e-05, + "loss": 0.0329, + "step": 3240 + }, + { + "epoch": 1.5432098765432098, + "grad_norm": 0.39999216549632843, + "learning_rate": 8.73264507031643e-05, + "loss": 0.0345, + "step": 3250 + }, + { + "epoch": 1.5479582146248814, + "grad_norm": 0.36778246346411, + "learning_rate": 8.721596473136342e-05, + "loss": 0.0323, + "step": 3260 + }, + { + "epoch": 1.5527065527065527, + "grad_norm": 0.41572199384610614, + "learning_rate": 8.710506979387882e-05, + "loss": 0.0321, + "step": 3270 + }, + { + "epoch": 1.557454890788224, + "grad_norm": 0.3608557161549764, + "learning_rate": 8.699376710933336e-05, + "loss": 0.0325, + "step": 3280 + }, + { + "epoch": 1.5622032288698955, + "grad_norm": 0.33558725315518206, + "learning_rate": 8.688205790083053e-05, + "loss": 0.0313, + "step": 3290 + }, + { + "epoch": 1.566951566951567, + "grad_norm": 0.40918172973994715, + "learning_rate": 8.676994339594118e-05, + "loss": 0.0319, + "step": 3300 + }, + { + "epoch": 1.5716999050332383, + "grad_norm": 0.43337497124362845, + "learning_rate": 8.665742482668996e-05, + "loss": 0.0317, + "step": 3310 + }, + { + "epoch": 1.5764482431149098, + "grad_norm": 0.3812998243057314, + "learning_rate": 8.654450342954175e-05, + "loss": 0.0323, + "step": 3320 + }, + { + "epoch": 1.5811965811965814, + "grad_norm": 0.35509189264018565, + "learning_rate": 8.643118044538808e-05, + "loss": 0.0326, + "step": 3330 + }, + { + "epoch": 1.5859449192782527, + "grad_norm": 0.38772936620972853, + "learning_rate": 8.631745711953357e-05, + "loss": 0.0317, + "step": 3340 + }, + { + "epoch": 1.590693257359924, + "grad_norm": 0.4020905021794808, + "learning_rate": 8.620333470168214e-05, + "loss": 0.0322, + "step": 3350 + }, + { + "epoch": 1.5954415954415955, + "grad_norm": 0.42014276075969054, + "learning_rate": 8.608881444592337e-05, + "loss": 0.0319, + "step": 3360 + }, + { + "epoch": 1.600189933523267, + "grad_norm": 0.3486263773747954, + "learning_rate": 8.597389761071862e-05, + "loss": 0.0309, + "step": 3370 + }, + { + "epoch": 1.6049382716049383, + "grad_norm": 0.3635863122166198, + "learning_rate": 8.58585854588873e-05, + "loss": 0.0292, + "step": 3380 + }, + { + "epoch": 1.6096866096866096, + "grad_norm": 0.35660749991777047, + "learning_rate": 8.574287925759294e-05, + "loss": 0.0305, + "step": 3390 + }, + { + "epoch": 1.6144349477682811, + "grad_norm": 0.36832555363654856, + "learning_rate": 8.562678027832925e-05, + "loss": 0.0302, + "step": 3400 + }, + { + "epoch": 1.6191832858499526, + "grad_norm": 0.41904584869446504, + "learning_rate": 8.551028979690621e-05, + "loss": 0.0286, + "step": 3410 + }, + { + "epoch": 1.623931623931624, + "grad_norm": 0.389795337040796, + "learning_rate": 8.539340909343597e-05, + "loss": 0.0298, + "step": 3420 + }, + { + "epoch": 1.6286799620132952, + "grad_norm": 0.36335077372557817, + "learning_rate": 8.527613945231885e-05, + "loss": 0.0298, + "step": 3430 + }, + { + "epoch": 1.6334283000949668, + "grad_norm": 0.34642329798347377, + "learning_rate": 8.51584821622292e-05, + "loss": 0.0302, + "step": 3440 + }, + { + "epoch": 1.6381766381766383, + "grad_norm": 0.35397287574395203, + "learning_rate": 8.504043851610122e-05, + "loss": 0.0285, + "step": 3450 + }, + { + "epoch": 1.6429249762583096, + "grad_norm": 0.35181123197530567, + "learning_rate": 8.492200981111482e-05, + "loss": 0.0276, + "step": 3460 + }, + { + "epoch": 1.6476733143399809, + "grad_norm": 0.39700888323674166, + "learning_rate": 8.480319734868124e-05, + "loss": 0.029, + "step": 3470 + }, + { + "epoch": 1.6524216524216524, + "grad_norm": 0.36508298744053175, + "learning_rate": 8.468400243442888e-05, + "loss": 0.0293, + "step": 3480 + }, + { + "epoch": 1.657169990503324, + "grad_norm": 0.36796649088087074, + "learning_rate": 8.456442637818889e-05, + "loss": 0.0279, + "step": 3490 + }, + { + "epoch": 1.6619183285849952, + "grad_norm": 0.40423201885327376, + "learning_rate": 8.444447049398076e-05, + "loss": 0.0282, + "step": 3500 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 0.4012837410158799, + "learning_rate": 8.432413609999794e-05, + "loss": 0.0282, + "step": 3510 + }, + { + "epoch": 1.671415004748338, + "grad_norm": 0.3591977885417729, + "learning_rate": 8.420342451859329e-05, + "loss": 0.0289, + "step": 3520 + }, + { + "epoch": 1.6761633428300096, + "grad_norm": 0.38049716791304095, + "learning_rate": 8.408233707626458e-05, + "loss": 0.0273, + "step": 3530 + }, + { + "epoch": 1.6809116809116809, + "grad_norm": 0.38381110184655687, + "learning_rate": 8.39608751036399e-05, + "loss": 0.0279, + "step": 3540 + }, + { + "epoch": 1.6856600189933522, + "grad_norm": 0.3758606793118219, + "learning_rate": 8.383903993546311e-05, + "loss": 0.0279, + "step": 3550 + }, + { + "epoch": 1.690408357075024, + "grad_norm": 0.3652014999666907, + "learning_rate": 8.371683291057902e-05, + "loss": 0.0276, + "step": 3560 + }, + { + "epoch": 1.6951566951566952, + "grad_norm": 0.35062250770884223, + "learning_rate": 8.359425537191883e-05, + "loss": 0.0279, + "step": 3570 + }, + { + "epoch": 1.6999050332383665, + "grad_norm": 0.35429427902173394, + "learning_rate": 8.347130866648528e-05, + "loss": 0.027, + "step": 3580 + }, + { + "epoch": 1.704653371320038, + "grad_norm": 0.39021507202575545, + "learning_rate": 8.334799414533791e-05, + "loss": 0.0275, + "step": 3590 + }, + { + "epoch": 1.7094017094017095, + "grad_norm": 0.3337642290497236, + "learning_rate": 8.322431316357815e-05, + "loss": 0.0274, + "step": 3600 + }, + { + "epoch": 1.7141500474833808, + "grad_norm": 0.3369344296932865, + "learning_rate": 8.310026708033447e-05, + "loss": 0.0262, + "step": 3610 + }, + { + "epoch": 1.7188983855650521, + "grad_norm": 0.3152823365250226, + "learning_rate": 8.297585725874743e-05, + "loss": 0.0248, + "step": 3620 + }, + { + "epoch": 1.7236467236467237, + "grad_norm": 0.30821331940982927, + "learning_rate": 8.285108506595471e-05, + "loss": 0.0261, + "step": 3630 + }, + { + "epoch": 1.7283950617283952, + "grad_norm": 0.3477084371470047, + "learning_rate": 8.272595187307612e-05, + "loss": 0.0262, + "step": 3640 + }, + { + "epoch": 1.7331433998100665, + "grad_norm": 0.34605218973502, + "learning_rate": 8.260045905519842e-05, + "loss": 0.0271, + "step": 3650 + }, + { + "epoch": 1.7378917378917378, + "grad_norm": 0.3830251062336052, + "learning_rate": 8.247460799136032e-05, + "loss": 0.0271, + "step": 3660 + }, + { + "epoch": 1.7426400759734093, + "grad_norm": 0.3428683717008426, + "learning_rate": 8.234840006453734e-05, + "loss": 0.0252, + "step": 3670 + }, + { + "epoch": 1.7473884140550808, + "grad_norm": 0.3411101986331494, + "learning_rate": 8.222183666162647e-05, + "loss": 0.025, + "step": 3680 + }, + { + "epoch": 1.7521367521367521, + "grad_norm": 0.3410315014616554, + "learning_rate": 8.209491917343104e-05, + "loss": 0.0251, + "step": 3690 + }, + { + "epoch": 1.7568850902184234, + "grad_norm": 0.3215079510916382, + "learning_rate": 8.196764899464551e-05, + "loss": 0.0245, + "step": 3700 + }, + { + "epoch": 1.761633428300095, + "grad_norm": 0.3315386556077621, + "learning_rate": 8.184002752383993e-05, + "loss": 0.0245, + "step": 3710 + }, + { + "epoch": 1.7663817663817665, + "grad_norm": 0.33403683945023016, + "learning_rate": 8.171205616344478e-05, + "loss": 0.0247, + "step": 3720 + }, + { + "epoch": 1.7711301044634378, + "grad_norm": 0.32301435748615215, + "learning_rate": 8.158373631973542e-05, + "loss": 0.0252, + "step": 3730 + }, + { + "epoch": 1.775878442545109, + "grad_norm": 0.3499811532255015, + "learning_rate": 8.14550694028167e-05, + "loss": 0.0244, + "step": 3740 + }, + { + "epoch": 1.7806267806267806, + "grad_norm": 0.33937435306968816, + "learning_rate": 8.13260568266075e-05, + "loss": 0.0251, + "step": 3750 + }, + { + "epoch": 1.7853751187084521, + "grad_norm": 0.31740106145356095, + "learning_rate": 8.119670000882506e-05, + "loss": 0.0244, + "step": 3760 + }, + { + "epoch": 1.7901234567901234, + "grad_norm": 0.3111005191293526, + "learning_rate": 8.106700037096955e-05, + "loss": 0.024, + "step": 3770 + }, + { + "epoch": 1.7948717948717947, + "grad_norm": 0.32308909823503645, + "learning_rate": 8.093695933830838e-05, + "loss": 0.0237, + "step": 3780 + }, + { + "epoch": 1.7996201329534662, + "grad_norm": 0.3379338471191445, + "learning_rate": 8.080657833986047e-05, + "loss": 0.0236, + "step": 3790 + }, + { + "epoch": 1.8043684710351378, + "grad_norm": 0.2844692645007023, + "learning_rate": 8.067585880838073e-05, + "loss": 0.023, + "step": 3800 + }, + { + "epoch": 1.809116809116809, + "grad_norm": 0.3334579193031102, + "learning_rate": 8.054480218034413e-05, + "loss": 0.0248, + "step": 3810 + }, + { + "epoch": 1.8138651471984806, + "grad_norm": 0.3125645635435408, + "learning_rate": 8.041340989593003e-05, + "loss": 0.0231, + "step": 3820 + }, + { + "epoch": 1.818613485280152, + "grad_norm": 0.31048872061723504, + "learning_rate": 8.028168339900625e-05, + "loss": 0.0225, + "step": 3830 + }, + { + "epoch": 1.8233618233618234, + "grad_norm": 0.3061792228880029, + "learning_rate": 8.014962413711334e-05, + "loss": 0.0236, + "step": 3840 + }, + { + "epoch": 1.8281101614434947, + "grad_norm": 0.34227797990174896, + "learning_rate": 8.001723356144855e-05, + "loss": 0.0233, + "step": 3850 + }, + { + "epoch": 1.8328584995251662, + "grad_norm": 0.3597288464605056, + "learning_rate": 7.988451312684997e-05, + "loss": 0.0226, + "step": 3860 + }, + { + "epoch": 1.8376068376068377, + "grad_norm": 0.3183410163678241, + "learning_rate": 7.975146429178047e-05, + "loss": 0.0228, + "step": 3870 + }, + { + "epoch": 1.842355175688509, + "grad_norm": 0.36742659414237855, + "learning_rate": 7.961808851831172e-05, + "loss": 0.0231, + "step": 3880 + }, + { + "epoch": 1.8471035137701803, + "grad_norm": 0.33353929550439043, + "learning_rate": 7.948438727210814e-05, + "loss": 0.0222, + "step": 3890 + }, + { + "epoch": 1.8518518518518519, + "grad_norm": 0.3051648584398829, + "learning_rate": 7.935036202241071e-05, + "loss": 0.0224, + "step": 3900 + }, + { + "epoch": 1.8566001899335234, + "grad_norm": 0.27472769583326284, + "learning_rate": 7.921601424202091e-05, + "loss": 0.0221, + "step": 3910 + }, + { + "epoch": 1.8613485280151947, + "grad_norm": 0.3141084100485619, + "learning_rate": 7.908134540728452e-05, + "loss": 0.0228, + "step": 3920 + }, + { + "epoch": 1.866096866096866, + "grad_norm": 0.3382576169175025, + "learning_rate": 7.894635699807533e-05, + "loss": 0.0228, + "step": 3930 + }, + { + "epoch": 1.8708452041785375, + "grad_norm": 0.32942766695615683, + "learning_rate": 7.881105049777901e-05, + "loss": 0.0222, + "step": 3940 + }, + { + "epoch": 1.875593542260209, + "grad_norm": 0.3178356174829631, + "learning_rate": 7.867542739327667e-05, + "loss": 0.0218, + "step": 3950 + }, + { + "epoch": 1.8803418803418803, + "grad_norm": 0.29509845464416246, + "learning_rate": 7.853948917492857e-05, + "loss": 0.0218, + "step": 3960 + }, + { + "epoch": 1.8850902184235516, + "grad_norm": 0.31955775449314067, + "learning_rate": 7.840323733655778e-05, + "loss": 0.021, + "step": 3970 + }, + { + "epoch": 1.8898385565052231, + "grad_norm": 0.3272468240136934, + "learning_rate": 7.826667337543377e-05, + "loss": 0.0219, + "step": 3980 + }, + { + "epoch": 1.8945868945868947, + "grad_norm": 0.3240768685105102, + "learning_rate": 7.812979879225581e-05, + "loss": 0.0217, + "step": 3990 + }, + { + "epoch": 1.899335232668566, + "grad_norm": 0.3391729362368217, + "learning_rate": 7.799261509113671e-05, + "loss": 0.0211, + "step": 4000 + }, + { + "epoch": 1.9040835707502373, + "grad_norm": 0.30212662738280743, + "learning_rate": 7.78551237795861e-05, + "loss": 0.0209, + "step": 4010 + }, + { + "epoch": 1.9088319088319088, + "grad_norm": 0.2981138934006492, + "learning_rate": 7.771732636849399e-05, + "loss": 0.0214, + "step": 4020 + }, + { + "epoch": 1.9135802469135803, + "grad_norm": 0.29358086770939895, + "learning_rate": 7.757922437211405e-05, + "loss": 0.0208, + "step": 4030 + }, + { + "epoch": 1.9183285849952516, + "grad_norm": 0.3500792587082307, + "learning_rate": 7.744081930804707e-05, + "loss": 0.0215, + "step": 4040 + }, + { + "epoch": 1.9230769230769231, + "grad_norm": 0.26759064390041853, + "learning_rate": 7.730211269722427e-05, + "loss": 0.0207, + "step": 4050 + }, + { + "epoch": 1.9278252611585946, + "grad_norm": 0.30497949136164654, + "learning_rate": 7.716310606389052e-05, + "loss": 0.0215, + "step": 4060 + }, + { + "epoch": 1.932573599240266, + "grad_norm": 0.2663159270855903, + "learning_rate": 7.702380093558765e-05, + "loss": 0.0202, + "step": 4070 + }, + { + "epoch": 1.9373219373219372, + "grad_norm": 0.31754900048378254, + "learning_rate": 7.688419884313764e-05, + "loss": 0.0198, + "step": 4080 + }, + { + "epoch": 1.9420702754036088, + "grad_norm": 0.32803871842514604, + "learning_rate": 7.674430132062585e-05, + "loss": 0.0213, + "step": 4090 + }, + { + "epoch": 1.9468186134852803, + "grad_norm": 0.29925865560231096, + "learning_rate": 7.660410990538403e-05, + "loss": 0.0205, + "step": 4100 + }, + { + "epoch": 1.9515669515669516, + "grad_norm": 0.3055228244819722, + "learning_rate": 7.646362613797356e-05, + "loss": 0.0206, + "step": 4110 + }, + { + "epoch": 1.956315289648623, + "grad_norm": 0.29029455455854586, + "learning_rate": 7.63228515621685e-05, + "loss": 0.0208, + "step": 4120 + }, + { + "epoch": 1.9610636277302944, + "grad_norm": 0.3137914900937615, + "learning_rate": 7.618178772493856e-05, + "loss": 0.0195, + "step": 4130 + }, + { + "epoch": 1.965811965811966, + "grad_norm": 0.2909821702970031, + "learning_rate": 7.604043617643215e-05, + "loss": 0.0194, + "step": 4140 + }, + { + "epoch": 1.9705603038936372, + "grad_norm": 0.3272183567174432, + "learning_rate": 7.589879846995934e-05, + "loss": 0.0198, + "step": 4150 + }, + { + "epoch": 1.9753086419753085, + "grad_norm": 0.3292146398230906, + "learning_rate": 7.575687616197477e-05, + "loss": 0.02, + "step": 4160 + }, + { + "epoch": 1.98005698005698, + "grad_norm": 0.3201501066613927, + "learning_rate": 7.561467081206061e-05, + "loss": 0.0195, + "step": 4170 + }, + { + "epoch": 1.9848053181386516, + "grad_norm": 0.30345657972087686, + "learning_rate": 7.547218398290932e-05, + "loss": 0.0197, + "step": 4180 + }, + { + "epoch": 1.9895536562203229, + "grad_norm": 0.29368193444912793, + "learning_rate": 7.532941724030654e-05, + "loss": 0.0188, + "step": 4190 + }, + { + "epoch": 1.9943019943019942, + "grad_norm": 0.3031029122165021, + "learning_rate": 7.518637215311388e-05, + "loss": 0.0196, + "step": 4200 + }, + { + "epoch": 1.9990503323836657, + "grad_norm": 0.3162367817687966, + "learning_rate": 7.504305029325172e-05, + "loss": 0.0195, + "step": 4210 + }, + { + "epoch": 2.003798670465337, + "grad_norm": 0.22658229574624864, + "learning_rate": 7.489945323568178e-05, + "loss": 0.0138, + "step": 4220 + }, + { + "epoch": 2.0085470085470085, + "grad_norm": 0.21891975672137579, + "learning_rate": 7.475558255839006e-05, + "loss": 0.011, + "step": 4230 + }, + { + "epoch": 2.01329534662868, + "grad_norm": 0.2530526723841709, + "learning_rate": 7.461143984236924e-05, + "loss": 0.0113, + "step": 4240 + }, + { + "epoch": 2.0180436847103516, + "grad_norm": 0.20305024571246374, + "learning_rate": 7.446702667160154e-05, + "loss": 0.0098, + "step": 4250 + }, + { + "epoch": 2.022792022792023, + "grad_norm": 0.1905882270400326, + "learning_rate": 7.432234463304111e-05, + "loss": 0.0102, + "step": 4260 + }, + { + "epoch": 2.027540360873694, + "grad_norm": 0.21995199109178779, + "learning_rate": 7.417739531659674e-05, + "loss": 0.0097, + "step": 4270 + }, + { + "epoch": 2.0322886989553655, + "grad_norm": 0.255886489253475, + "learning_rate": 7.403218031511428e-05, + "loss": 0.0094, + "step": 4280 + }, + { + "epoch": 2.037037037037037, + "grad_norm": 0.21846327167925236, + "learning_rate": 7.388670122435928e-05, + "loss": 0.0099, + "step": 4290 + }, + { + "epoch": 2.0417853751187085, + "grad_norm": 0.21713197666752804, + "learning_rate": 7.374095964299928e-05, + "loss": 0.0097, + "step": 4300 + }, + { + "epoch": 2.04653371320038, + "grad_norm": 0.22636532369860138, + "learning_rate": 7.359495717258636e-05, + "loss": 0.0101, + "step": 4310 + }, + { + "epoch": 2.051282051282051, + "grad_norm": 0.23234681842362245, + "learning_rate": 7.34486954175395e-05, + "loss": 0.0095, + "step": 4320 + }, + { + "epoch": 2.056030389363723, + "grad_norm": 0.1936836721600177, + "learning_rate": 7.330217598512695e-05, + "loss": 0.0099, + "step": 4330 + }, + { + "epoch": 2.060778727445394, + "grad_norm": 0.2042655714888282, + "learning_rate": 7.315540048544857e-05, + "loss": 0.0092, + "step": 4340 + }, + { + "epoch": 2.0655270655270654, + "grad_norm": 0.20609588799769912, + "learning_rate": 7.300837053141815e-05, + "loss": 0.0096, + "step": 4350 + }, + { + "epoch": 2.0702754036087367, + "grad_norm": 0.21567675178289214, + "learning_rate": 7.286108773874566e-05, + "loss": 0.0091, + "step": 4360 + }, + { + "epoch": 2.0750237416904085, + "grad_norm": 0.25049907200353494, + "learning_rate": 7.271355372591956e-05, + "loss": 0.009, + "step": 4370 + }, + { + "epoch": 2.07977207977208, + "grad_norm": 0.2162437371325694, + "learning_rate": 7.256577011418888e-05, + "loss": 0.0091, + "step": 4380 + }, + { + "epoch": 2.084520417853751, + "grad_norm": 0.21465456792741833, + "learning_rate": 7.241773852754556e-05, + "loss": 0.0087, + "step": 4390 + }, + { + "epoch": 2.0892687559354224, + "grad_norm": 0.21540821815439232, + "learning_rate": 7.226946059270651e-05, + "loss": 0.0091, + "step": 4400 + }, + { + "epoch": 2.094017094017094, + "grad_norm": 0.1976535788276764, + "learning_rate": 7.212093793909573e-05, + "loss": 0.0089, + "step": 4410 + }, + { + "epoch": 2.0987654320987654, + "grad_norm": 0.1870743241825421, + "learning_rate": 7.197217219882647e-05, + "loss": 0.0092, + "step": 4420 + }, + { + "epoch": 2.1035137701804367, + "grad_norm": 0.20277485705414525, + "learning_rate": 7.182316500668326e-05, + "loss": 0.0089, + "step": 4430 + }, + { + "epoch": 2.1082621082621085, + "grad_norm": 0.23541868814783684, + "learning_rate": 7.167391800010389e-05, + "loss": 0.0092, + "step": 4440 + }, + { + "epoch": 2.1130104463437798, + "grad_norm": 0.23445568559985083, + "learning_rate": 7.152443281916151e-05, + "loss": 0.0093, + "step": 4450 + }, + { + "epoch": 2.117758784425451, + "grad_norm": 0.23458169443107807, + "learning_rate": 7.137471110654656e-05, + "loss": 0.0095, + "step": 4460 + }, + { + "epoch": 2.1225071225071224, + "grad_norm": 0.20988484973777136, + "learning_rate": 7.122475450754869e-05, + "loss": 0.0087, + "step": 4470 + }, + { + "epoch": 2.127255460588794, + "grad_norm": 0.22846070547892422, + "learning_rate": 7.107456467003877e-05, + "loss": 0.0093, + "step": 4480 + }, + { + "epoch": 2.1320037986704654, + "grad_norm": 0.2513130275747507, + "learning_rate": 7.092414324445068e-05, + "loss": 0.0095, + "step": 4490 + }, + { + "epoch": 2.1367521367521367, + "grad_norm": 0.2170117639837281, + "learning_rate": 7.077349188376325e-05, + "loss": 0.0096, + "step": 4500 + }, + { + "epoch": 2.141500474833808, + "grad_norm": 0.24496701039946453, + "learning_rate": 7.062261224348203e-05, + "loss": 0.0091, + "step": 4510 + }, + { + "epoch": 2.1462488129154798, + "grad_norm": 0.21621069180143684, + "learning_rate": 7.047150598162114e-05, + "loss": 0.0097, + "step": 4520 + }, + { + "epoch": 2.150997150997151, + "grad_norm": 0.2170119503677641, + "learning_rate": 7.032017475868505e-05, + "loss": 0.0089, + "step": 4530 + }, + { + "epoch": 2.1557454890788224, + "grad_norm": 0.24324610318962608, + "learning_rate": 7.016862023765033e-05, + "loss": 0.009, + "step": 4540 + }, + { + "epoch": 2.1604938271604937, + "grad_norm": 0.2306596862350849, + "learning_rate": 7.001684408394735e-05, + "loss": 0.0089, + "step": 4550 + }, + { + "epoch": 2.1652421652421654, + "grad_norm": 0.22393947017592192, + "learning_rate": 6.986484796544203e-05, + "loss": 0.009, + "step": 4560 + }, + { + "epoch": 2.1699905033238367, + "grad_norm": 0.2270861623853049, + "learning_rate": 6.971263355241742e-05, + "loss": 0.0089, + "step": 4570 + }, + { + "epoch": 2.174738841405508, + "grad_norm": 0.22383137527358174, + "learning_rate": 6.956020251755548e-05, + "loss": 0.0092, + "step": 4580 + }, + { + "epoch": 2.1794871794871793, + "grad_norm": 0.22839704391335555, + "learning_rate": 6.940755653591858e-05, + "loss": 0.0089, + "step": 4590 + }, + { + "epoch": 2.184235517568851, + "grad_norm": 0.2345922267437847, + "learning_rate": 6.925469728493114e-05, + "loss": 0.009, + "step": 4600 + }, + { + "epoch": 2.1889838556505223, + "grad_norm": 0.2261315462401965, + "learning_rate": 6.91016264443612e-05, + "loss": 0.0084, + "step": 4610 + }, + { + "epoch": 2.1937321937321936, + "grad_norm": 0.2743979635129678, + "learning_rate": 6.894834569630195e-05, + "loss": 0.0093, + "step": 4620 + }, + { + "epoch": 2.198480531813865, + "grad_norm": 0.21664998223400408, + "learning_rate": 6.879485672515327e-05, + "loss": 0.0087, + "step": 4630 + }, + { + "epoch": 2.2032288698955367, + "grad_norm": 0.2580920861147472, + "learning_rate": 6.864116121760316e-05, + "loss": 0.0094, + "step": 4640 + }, + { + "epoch": 2.207977207977208, + "grad_norm": 0.242966269636067, + "learning_rate": 6.848726086260929e-05, + "loss": 0.0091, + "step": 4650 + }, + { + "epoch": 2.2127255460588793, + "grad_norm": 0.20989216465974841, + "learning_rate": 6.833315735138035e-05, + "loss": 0.0091, + "step": 4660 + }, + { + "epoch": 2.217473884140551, + "grad_norm": 0.23677198810419076, + "learning_rate": 6.817885237735758e-05, + "loss": 0.0089, + "step": 4670 + }, + { + "epoch": 2.2222222222222223, + "grad_norm": 0.22236072977848959, + "learning_rate": 6.802434763619599e-05, + "loss": 0.0091, + "step": 4680 + }, + { + "epoch": 2.2269705603038936, + "grad_norm": 0.24315146508890714, + "learning_rate": 6.786964482574594e-05, + "loss": 0.0094, + "step": 4690 + }, + { + "epoch": 2.231718898385565, + "grad_norm": 0.22824988577881591, + "learning_rate": 6.771474564603428e-05, + "loss": 0.0094, + "step": 4700 + }, + { + "epoch": 2.2364672364672367, + "grad_norm": 0.2518746282722726, + "learning_rate": 6.755965179924582e-05, + "loss": 0.0089, + "step": 4710 + }, + { + "epoch": 2.241215574548908, + "grad_norm": 0.2228096249646919, + "learning_rate": 6.740436498970452e-05, + "loss": 0.0088, + "step": 4720 + }, + { + "epoch": 2.2459639126305793, + "grad_norm": 0.20412531890833327, + "learning_rate": 6.724888692385485e-05, + "loss": 0.0084, + "step": 4730 + }, + { + "epoch": 2.2507122507122506, + "grad_norm": 0.20433865015043312, + "learning_rate": 6.709321931024293e-05, + "loss": 0.0084, + "step": 4740 + }, + { + "epoch": 2.2554605887939223, + "grad_norm": 0.2453273456503076, + "learning_rate": 6.693736385949788e-05, + "loss": 0.0093, + "step": 4750 + }, + { + "epoch": 2.2602089268755936, + "grad_norm": 0.21355399157391225, + "learning_rate": 6.678132228431294e-05, + "loss": 0.0087, + "step": 4760 + }, + { + "epoch": 2.264957264957265, + "grad_norm": 0.23699124042263617, + "learning_rate": 6.662509629942661e-05, + "loss": 0.0086, + "step": 4770 + }, + { + "epoch": 2.269705603038936, + "grad_norm": 0.22660308516863123, + "learning_rate": 6.646868762160399e-05, + "loss": 0.0084, + "step": 4780 + }, + { + "epoch": 2.274453941120608, + "grad_norm": 0.22840803324110504, + "learning_rate": 6.631209796961767e-05, + "loss": 0.009, + "step": 4790 + }, + { + "epoch": 2.2792022792022792, + "grad_norm": 0.23029050180743044, + "learning_rate": 6.615532906422902e-05, + "loss": 0.0082, + "step": 4800 + }, + { + "epoch": 2.2839506172839505, + "grad_norm": 0.2847858938160236, + "learning_rate": 6.599838262816922e-05, + "loss": 0.0091, + "step": 4810 + }, + { + "epoch": 2.288698955365622, + "grad_norm": 0.24787375223693867, + "learning_rate": 6.584126038612033e-05, + "loss": 0.0088, + "step": 4820 + }, + { + "epoch": 2.2934472934472936, + "grad_norm": 0.2491113404885631, + "learning_rate": 6.568396406469632e-05, + "loss": 0.0084, + "step": 4830 + }, + { + "epoch": 2.298195631528965, + "grad_norm": 0.2542352588549579, + "learning_rate": 6.552649539242414e-05, + "loss": 0.0087, + "step": 4840 + }, + { + "epoch": 2.302943969610636, + "grad_norm": 0.20676553593523492, + "learning_rate": 6.536885609972467e-05, + "loss": 0.0086, + "step": 4850 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.21433361754989655, + "learning_rate": 6.521104791889381e-05, + "loss": 0.0086, + "step": 4860 + }, + { + "epoch": 2.3124406457739792, + "grad_norm": 0.23024653424740793, + "learning_rate": 6.505307258408325e-05, + "loss": 0.0091, + "step": 4870 + }, + { + "epoch": 2.3171889838556505, + "grad_norm": 0.22275204824005482, + "learning_rate": 6.489493183128163e-05, + "loss": 0.0087, + "step": 4880 + }, + { + "epoch": 2.321937321937322, + "grad_norm": 0.330205138024354, + "learning_rate": 6.473662739829533e-05, + "loss": 0.009, + "step": 4890 + }, + { + "epoch": 2.3266856600189936, + "grad_norm": 0.1990880646771009, + "learning_rate": 6.457816102472942e-05, + "loss": 0.0088, + "step": 4900 + }, + { + "epoch": 2.331433998100665, + "grad_norm": 0.2174787322434033, + "learning_rate": 6.44195344519685e-05, + "loss": 0.0086, + "step": 4910 + }, + { + "epoch": 2.336182336182336, + "grad_norm": 0.23786724534040635, + "learning_rate": 6.426074942315766e-05, + "loss": 0.0093, + "step": 4920 + }, + { + "epoch": 2.3409306742640075, + "grad_norm": 0.21666259946655012, + "learning_rate": 6.410180768318321e-05, + "loss": 0.0089, + "step": 4930 + }, + { + "epoch": 2.3456790123456788, + "grad_norm": 0.2525347835917826, + "learning_rate": 6.394271097865355e-05, + "loss": 0.0083, + "step": 4940 + }, + { + "epoch": 2.3504273504273505, + "grad_norm": 0.1818674264346914, + "learning_rate": 6.378346105788002e-05, + "loss": 0.0089, + "step": 4950 + }, + { + "epoch": 2.355175688509022, + "grad_norm": 0.2303694692872279, + "learning_rate": 6.362405967085765e-05, + "loss": 0.0091, + "step": 4960 + }, + { + "epoch": 2.359924026590693, + "grad_norm": 0.2704392175352175, + "learning_rate": 6.346450856924588e-05, + "loss": 0.009, + "step": 4970 + }, + { + "epoch": 2.364672364672365, + "grad_norm": 0.21558301520885845, + "learning_rate": 6.330480950634942e-05, + "loss": 0.0089, + "step": 4980 + }, + { + "epoch": 2.369420702754036, + "grad_norm": 0.25900570019240504, + "learning_rate": 6.314496423709887e-05, + "loss": 0.0086, + "step": 4990 + }, + { + "epoch": 2.3741690408357075, + "grad_norm": 0.21826225969232374, + "learning_rate": 6.298497451803157e-05, + "loss": 0.0083, + "step": 5000 + }, + { + "epoch": 2.3789173789173788, + "grad_norm": 0.2213155712830059, + "learning_rate": 6.282484210727213e-05, + "loss": 0.0088, + "step": 5010 + }, + { + "epoch": 2.3836657169990505, + "grad_norm": 0.232839588184917, + "learning_rate": 6.266456876451324e-05, + "loss": 0.008, + "step": 5020 + }, + { + "epoch": 2.388414055080722, + "grad_norm": 0.21684403014547518, + "learning_rate": 6.25041562509963e-05, + "loss": 0.0083, + "step": 5030 + }, + { + "epoch": 2.393162393162393, + "grad_norm": 0.2185819173868181, + "learning_rate": 6.234360632949201e-05, + "loss": 0.0083, + "step": 5040 + }, + { + "epoch": 2.3979107312440644, + "grad_norm": 0.20709805652078042, + "learning_rate": 6.21829207642811e-05, + "loss": 0.0084, + "step": 5050 + }, + { + "epoch": 2.402659069325736, + "grad_norm": 0.24842481671132302, + "learning_rate": 6.202210132113484e-05, + "loss": 0.0082, + "step": 5060 + }, + { + "epoch": 2.4074074074074074, + "grad_norm": 0.21276226654312166, + "learning_rate": 6.186114976729573e-05, + "loss": 0.0081, + "step": 5070 + }, + { + "epoch": 2.4121557454890787, + "grad_norm": 0.2046114928127734, + "learning_rate": 6.170006787145796e-05, + "loss": 0.0081, + "step": 5080 + }, + { + "epoch": 2.41690408357075, + "grad_norm": 0.23870540105767515, + "learning_rate": 6.153885740374813e-05, + "loss": 0.0082, + "step": 5090 + }, + { + "epoch": 2.421652421652422, + "grad_norm": 0.17136740277225707, + "learning_rate": 6.137752013570566e-05, + "loss": 0.0078, + "step": 5100 + }, + { + "epoch": 2.426400759734093, + "grad_norm": 0.21004603052301513, + "learning_rate": 6.121605784026339e-05, + "loss": 0.008, + "step": 5110 + }, + { + "epoch": 2.4311490978157644, + "grad_norm": 0.20331259316767894, + "learning_rate": 6.10544722917281e-05, + "loss": 0.008, + "step": 5120 + }, + { + "epoch": 2.435897435897436, + "grad_norm": 0.20908634643100213, + "learning_rate": 6.089276526576094e-05, + "loss": 0.0085, + "step": 5130 + }, + { + "epoch": 2.4406457739791074, + "grad_norm": 0.24017891290372667, + "learning_rate": 6.0730938539358084e-05, + "loss": 0.008, + "step": 5140 + }, + { + "epoch": 2.4453941120607787, + "grad_norm": 0.2405191279291125, + "learning_rate": 6.056899389083097e-05, + "loss": 0.0082, + "step": 5150 + }, + { + "epoch": 2.45014245014245, + "grad_norm": 0.21235075163572945, + "learning_rate": 6.0406933099786976e-05, + "loss": 0.0079, + "step": 5160 + }, + { + "epoch": 2.4548907882241213, + "grad_norm": 0.21433156836019124, + "learning_rate": 6.02447579471097e-05, + "loss": 0.0082, + "step": 5170 + }, + { + "epoch": 2.459639126305793, + "grad_norm": 0.2659880698246093, + "learning_rate": 6.0082470214939514e-05, + "loss": 0.0087, + "step": 5180 + }, + { + "epoch": 2.4643874643874644, + "grad_norm": 0.25068926482893866, + "learning_rate": 5.9920071686653855e-05, + "loss": 0.0083, + "step": 5190 + }, + { + "epoch": 2.4691358024691357, + "grad_norm": 0.23675457788589924, + "learning_rate": 5.9757564146847764e-05, + "loss": 0.0078, + "step": 5200 + }, + { + "epoch": 2.4738841405508074, + "grad_norm": 0.2287304600995394, + "learning_rate": 5.959494938131418e-05, + "loss": 0.0079, + "step": 5210 + }, + { + "epoch": 2.4786324786324787, + "grad_norm": 0.2106985950234504, + "learning_rate": 5.9432229177024346e-05, + "loss": 0.0077, + "step": 5220 + }, + { + "epoch": 2.48338081671415, + "grad_norm": 0.24106103368942983, + "learning_rate": 5.926940532210817e-05, + "loss": 0.0079, + "step": 5230 + }, + { + "epoch": 2.4881291547958213, + "grad_norm": 0.20839133989389724, + "learning_rate": 5.910647960583457e-05, + "loss": 0.0085, + "step": 5240 + }, + { + "epoch": 2.492877492877493, + "grad_norm": 0.23884971815449735, + "learning_rate": 5.8943453818591845e-05, + "loss": 0.0076, + "step": 5250 + }, + { + "epoch": 2.4976258309591644, + "grad_norm": 0.2334702836423085, + "learning_rate": 5.878032975186791e-05, + "loss": 0.0081, + "step": 5260 + }, + { + "epoch": 2.5023741690408356, + "grad_norm": 0.2718850049039682, + "learning_rate": 5.8617109198230744e-05, + "loss": 0.0078, + "step": 5270 + }, + { + "epoch": 2.5071225071225074, + "grad_norm": 0.20207803949357483, + "learning_rate": 5.845379395130859e-05, + "loss": 0.0075, + "step": 5280 + }, + { + "epoch": 2.5118708452041787, + "grad_norm": 0.2174070261619274, + "learning_rate": 5.829038580577025e-05, + "loss": 0.008, + "step": 5290 + }, + { + "epoch": 2.51661918328585, + "grad_norm": 0.267796141047918, + "learning_rate": 5.812688655730542e-05, + "loss": 0.008, + "step": 5300 + }, + { + "epoch": 2.5213675213675213, + "grad_norm": 0.23389246236044087, + "learning_rate": 5.796329800260489e-05, + "loss": 0.0086, + "step": 5310 + }, + { + "epoch": 2.5261158594491926, + "grad_norm": 0.23392987968135592, + "learning_rate": 5.7799621939340896e-05, + "loss": 0.0079, + "step": 5320 + }, + { + "epoch": 2.5308641975308643, + "grad_norm": 0.207853162909817, + "learning_rate": 5.7635860166147246e-05, + "loss": 0.0075, + "step": 5330 + }, + { + "epoch": 2.5356125356125356, + "grad_norm": 0.24582483177176384, + "learning_rate": 5.747201448259964e-05, + "loss": 0.0082, + "step": 5340 + }, + { + "epoch": 2.540360873694207, + "grad_norm": 0.25279680949795835, + "learning_rate": 5.730808668919584e-05, + "loss": 0.0075, + "step": 5350 + }, + { + "epoch": 2.5451092117758787, + "grad_norm": 0.20941142050614048, + "learning_rate": 5.714407858733596e-05, + "loss": 0.0082, + "step": 5360 + }, + { + "epoch": 2.54985754985755, + "grad_norm": 0.24807411712432773, + "learning_rate": 5.6979991979302585e-05, + "loss": 0.0079, + "step": 5370 + }, + { + "epoch": 2.5546058879392213, + "grad_norm": 0.23556758125210095, + "learning_rate": 5.6815828668241e-05, + "loss": 0.0079, + "step": 5380 + }, + { + "epoch": 2.5593542260208926, + "grad_norm": 0.22367598584163997, + "learning_rate": 5.6651590458139424e-05, + "loss": 0.0081, + "step": 5390 + }, + { + "epoch": 2.564102564102564, + "grad_norm": 0.23924451571222358, + "learning_rate": 5.6487279153809094e-05, + "loss": 0.0079, + "step": 5400 + }, + { + "epoch": 2.5688509021842356, + "grad_norm": 0.21113415460835394, + "learning_rate": 5.632289656086448e-05, + "loss": 0.0076, + "step": 5410 + }, + { + "epoch": 2.573599240265907, + "grad_norm": 0.22981764125958118, + "learning_rate": 5.6158444485703465e-05, + "loss": 0.0076, + "step": 5420 + }, + { + "epoch": 2.578347578347578, + "grad_norm": 0.20787088653823935, + "learning_rate": 5.5993924735487455e-05, + "loss": 0.0072, + "step": 5430 + }, + { + "epoch": 2.58309591642925, + "grad_norm": 0.22867110723303855, + "learning_rate": 5.582933911812155e-05, + "loss": 0.0072, + "step": 5440 + }, + { + "epoch": 2.5878442545109213, + "grad_norm": 0.20187465437566843, + "learning_rate": 5.5664689442234654e-05, + "loss": 0.007, + "step": 5450 + }, + { + "epoch": 2.5925925925925926, + "grad_norm": 0.2122058246788947, + "learning_rate": 5.549997751715961e-05, + "loss": 0.0076, + "step": 5460 + }, + { + "epoch": 2.597340930674264, + "grad_norm": 0.22055737424473512, + "learning_rate": 5.5335205152913305e-05, + "loss": 0.0076, + "step": 5470 + }, + { + "epoch": 2.602089268755935, + "grad_norm": 0.22173488608270828, + "learning_rate": 5.5170374160176815e-05, + "loss": 0.0072, + "step": 5480 + }, + { + "epoch": 2.606837606837607, + "grad_norm": 0.22739069121971175, + "learning_rate": 5.500548635027547e-05, + "loss": 0.0071, + "step": 5490 + }, + { + "epoch": 2.611585944919278, + "grad_norm": 0.223883027288244, + "learning_rate": 5.484054353515896e-05, + "loss": 0.0082, + "step": 5500 + }, + { + "epoch": 2.61633428300095, + "grad_norm": 0.20691132974179816, + "learning_rate": 5.467554752738143e-05, + "loss": 0.0072, + "step": 5510 + }, + { + "epoch": 2.6210826210826212, + "grad_norm": 0.22464148128229552, + "learning_rate": 5.451050014008157e-05, + "loss": 0.0075, + "step": 5520 + }, + { + "epoch": 2.6258309591642925, + "grad_norm": 0.20116802493379207, + "learning_rate": 5.434540318696265e-05, + "loss": 0.0073, + "step": 5530 + }, + { + "epoch": 2.630579297245964, + "grad_norm": 0.23785190549568347, + "learning_rate": 5.418025848227265e-05, + "loss": 0.0078, + "step": 5540 + }, + { + "epoch": 2.635327635327635, + "grad_norm": 0.19145910011635187, + "learning_rate": 5.401506784078427e-05, + "loss": 0.0067, + "step": 5550 + }, + { + "epoch": 2.640075973409307, + "grad_norm": 0.22079557784470044, + "learning_rate": 5.384983307777502e-05, + "loss": 0.0072, + "step": 5560 + }, + { + "epoch": 2.644824311490978, + "grad_norm": 0.21845821042588354, + "learning_rate": 5.368455600900725e-05, + "loss": 0.0067, + "step": 5570 + }, + { + "epoch": 2.6495726495726495, + "grad_norm": 0.22808805599203474, + "learning_rate": 5.351923845070822e-05, + "loss": 0.007, + "step": 5580 + }, + { + "epoch": 2.6543209876543212, + "grad_norm": 0.19082496383099576, + "learning_rate": 5.335388221955012e-05, + "loss": 0.0067, + "step": 5590 + }, + { + "epoch": 2.6590693257359925, + "grad_norm": 0.18748454748464538, + "learning_rate": 5.3188489132630124e-05, + "loss": 0.0074, + "step": 5600 + }, + { + "epoch": 2.663817663817664, + "grad_norm": 0.18757801450402883, + "learning_rate": 5.3023061007450404e-05, + "loss": 0.0073, + "step": 5610 + }, + { + "epoch": 2.668566001899335, + "grad_norm": 0.19677210569732215, + "learning_rate": 5.285759966189817e-05, + "loss": 0.007, + "step": 5620 + }, + { + "epoch": 2.6733143399810064, + "grad_norm": 0.17201041710410034, + "learning_rate": 5.269210691422569e-05, + "loss": 0.0069, + "step": 5630 + }, + { + "epoch": 2.678062678062678, + "grad_norm": 0.22607120027108688, + "learning_rate": 5.252658458303032e-05, + "loss": 0.0068, + "step": 5640 + }, + { + "epoch": 2.6828110161443495, + "grad_norm": 0.23635307578223666, + "learning_rate": 5.236103448723448e-05, + "loss": 0.0074, + "step": 5650 + }, + { + "epoch": 2.6875593542260208, + "grad_norm": 0.2178800049084607, + "learning_rate": 5.219545844606573e-05, + "loss": 0.007, + "step": 5660 + }, + { + "epoch": 2.6923076923076925, + "grad_norm": 0.17633183749409215, + "learning_rate": 5.202985827903672e-05, + "loss": 0.0069, + "step": 5670 + }, + { + "epoch": 2.697056030389364, + "grad_norm": 0.199247536996637, + "learning_rate": 5.186423580592523e-05, + "loss": 0.0071, + "step": 5680 + }, + { + "epoch": 2.701804368471035, + "grad_norm": 0.18427403488003175, + "learning_rate": 5.1698592846754156e-05, + "loss": 0.0062, + "step": 5690 + }, + { + "epoch": 2.7065527065527064, + "grad_norm": 0.20831941610118215, + "learning_rate": 5.1532931221771515e-05, + "loss": 0.0072, + "step": 5700 + }, + { + "epoch": 2.7113010446343777, + "grad_norm": 0.1903105603799371, + "learning_rate": 5.136725275143045e-05, + "loss": 0.0069, + "step": 5710 + }, + { + "epoch": 2.7160493827160495, + "grad_norm": 0.21351260033832267, + "learning_rate": 5.1201559256369215e-05, + "loss": 0.0067, + "step": 5720 + }, + { + "epoch": 2.7207977207977208, + "grad_norm": 0.20926188125652498, + "learning_rate": 5.103585255739113e-05, + "loss": 0.0063, + "step": 5730 + }, + { + "epoch": 2.725546058879392, + "grad_norm": 0.18127504909031003, + "learning_rate": 5.087013447544469e-05, + "loss": 0.0067, + "step": 5740 + }, + { + "epoch": 2.730294396961064, + "grad_norm": 0.21090770999066813, + "learning_rate": 5.0704406831603414e-05, + "loss": 0.007, + "step": 5750 + }, + { + "epoch": 2.735042735042735, + "grad_norm": 0.22839187262173638, + "learning_rate": 5.0538671447045936e-05, + "loss": 0.0064, + "step": 5760 + }, + { + "epoch": 2.7397910731244064, + "grad_norm": 0.2186527982338498, + "learning_rate": 5.037293014303592e-05, + "loss": 0.0064, + "step": 5770 + }, + { + "epoch": 2.7445394112060777, + "grad_norm": 0.24728984386861405, + "learning_rate": 5.020718474090212e-05, + "loss": 0.0069, + "step": 5780 + }, + { + "epoch": 2.7492877492877494, + "grad_norm": 0.18123448575020748, + "learning_rate": 5.004143706201826e-05, + "loss": 0.0066, + "step": 5790 + }, + { + "epoch": 2.7540360873694207, + "grad_norm": 0.22019810673376586, + "learning_rate": 4.987568892778314e-05, + "loss": 0.0067, + "step": 5800 + }, + { + "epoch": 2.758784425451092, + "grad_norm": 0.2510509078630789, + "learning_rate": 4.970994215960056e-05, + "loss": 0.0062, + "step": 5810 + }, + { + "epoch": 2.763532763532764, + "grad_norm": 0.22874404729215814, + "learning_rate": 4.954419857885926e-05, + "loss": 0.0072, + "step": 5820 + }, + { + "epoch": 2.768281101614435, + "grad_norm": 0.20875914680955318, + "learning_rate": 4.9378460006913035e-05, + "loss": 0.0066, + "step": 5830 + }, + { + "epoch": 2.7730294396961064, + "grad_norm": 0.20225405841783692, + "learning_rate": 4.921272826506054e-05, + "loss": 0.0065, + "step": 5840 + }, + { + "epoch": 2.7777777777777777, + "grad_norm": 0.2510118406207145, + "learning_rate": 4.9047005174525454e-05, + "loss": 0.0071, + "step": 5850 + }, + { + "epoch": 2.782526115859449, + "grad_norm": 0.2039039340856703, + "learning_rate": 4.8881292556436354e-05, + "loss": 0.0065, + "step": 5860 + }, + { + "epoch": 2.7872744539411207, + "grad_norm": 0.20826678612052482, + "learning_rate": 4.871559223180673e-05, + "loss": 0.0064, + "step": 5870 + }, + { + "epoch": 2.792022792022792, + "grad_norm": 0.23632518030480928, + "learning_rate": 4.854990602151502e-05, + "loss": 0.0064, + "step": 5880 + }, + { + "epoch": 2.7967711301044633, + "grad_norm": 0.2546101613543101, + "learning_rate": 4.8384235746284476e-05, + "loss": 0.0067, + "step": 5890 + }, + { + "epoch": 2.801519468186135, + "grad_norm": 0.21264161336971096, + "learning_rate": 4.821858322666333e-05, + "loss": 0.006, + "step": 5900 + }, + { + "epoch": 2.8062678062678064, + "grad_norm": 0.20556278528456787, + "learning_rate": 4.8052950283004637e-05, + "loss": 0.0065, + "step": 5910 + }, + { + "epoch": 2.8110161443494777, + "grad_norm": 0.19532500330000757, + "learning_rate": 4.7887338735446394e-05, + "loss": 0.0065, + "step": 5920 + }, + { + "epoch": 2.815764482431149, + "grad_norm": 0.19037784187185305, + "learning_rate": 4.772175040389138e-05, + "loss": 0.0067, + "step": 5930 + }, + { + "epoch": 2.8205128205128203, + "grad_norm": 0.24442757330922604, + "learning_rate": 4.7556187107987365e-05, + "loss": 0.0063, + "step": 5940 + }, + { + "epoch": 2.825261158594492, + "grad_norm": 0.22079130568953334, + "learning_rate": 4.73906506671069e-05, + "loss": 0.0063, + "step": 5950 + }, + { + "epoch": 2.8300094966761633, + "grad_norm": 0.2188062835892268, + "learning_rate": 4.722514290032753e-05, + "loss": 0.0061, + "step": 5960 + }, + { + "epoch": 2.8347578347578346, + "grad_norm": 0.2372979855276816, + "learning_rate": 4.70596656264116e-05, + "loss": 0.0063, + "step": 5970 + }, + { + "epoch": 2.8395061728395063, + "grad_norm": 0.21555712410903943, + "learning_rate": 4.689422066378644e-05, + "loss": 0.0064, + "step": 5980 + }, + { + "epoch": 2.8442545109211776, + "grad_norm": 0.19063940226848383, + "learning_rate": 4.6728809830524285e-05, + "loss": 0.0065, + "step": 5990 + }, + { + "epoch": 2.849002849002849, + "grad_norm": 0.20431221551473935, + "learning_rate": 4.65634349443223e-05, + "loss": 0.0055, + "step": 6000 + }, + { + "epoch": 2.8537511870845202, + "grad_norm": 0.21175044646763816, + "learning_rate": 4.63980978224827e-05, + "loss": 0.0062, + "step": 6010 + }, + { + "epoch": 2.858499525166192, + "grad_norm": 0.20887196186171553, + "learning_rate": 4.62328002818926e-05, + "loss": 0.0065, + "step": 6020 + }, + { + "epoch": 2.8632478632478633, + "grad_norm": 0.19987422691421336, + "learning_rate": 4.606754413900428e-05, + "loss": 0.0064, + "step": 6030 + }, + { + "epoch": 2.8679962013295346, + "grad_norm": 0.16845929491776215, + "learning_rate": 4.5902331209814995e-05, + "loss": 0.0059, + "step": 6040 + }, + { + "epoch": 2.8727445394112063, + "grad_norm": 0.2076077192589226, + "learning_rate": 4.5737163309847205e-05, + "loss": 0.006, + "step": 6050 + }, + { + "epoch": 2.8774928774928776, + "grad_norm": 0.216301350995299, + "learning_rate": 4.557204225412848e-05, + "loss": 0.0059, + "step": 6060 + }, + { + "epoch": 2.882241215574549, + "grad_norm": 0.19430100060898772, + "learning_rate": 4.540696985717168e-05, + "loss": 0.0061, + "step": 6070 + }, + { + "epoch": 2.8869895536562202, + "grad_norm": 0.20048428461012352, + "learning_rate": 4.524194793295488e-05, + "loss": 0.0063, + "step": 6080 + }, + { + "epoch": 2.8917378917378915, + "grad_norm": 0.20556059740727328, + "learning_rate": 4.5076978294901604e-05, + "loss": 0.0061, + "step": 6090 + }, + { + "epoch": 2.8964862298195633, + "grad_norm": 0.227770251504533, + "learning_rate": 4.491206275586072e-05, + "loss": 0.0062, + "step": 6100 + }, + { + "epoch": 2.9012345679012346, + "grad_norm": 0.1861380364772549, + "learning_rate": 4.4747203128086634e-05, + "loss": 0.0058, + "step": 6110 + }, + { + "epoch": 2.905982905982906, + "grad_norm": 0.20396126797537462, + "learning_rate": 4.458240122321939e-05, + "loss": 0.0065, + "step": 6120 + }, + { + "epoch": 2.9107312440645776, + "grad_norm": 0.18662882523991647, + "learning_rate": 4.4417658852264614e-05, + "loss": 0.0059, + "step": 6130 + }, + { + "epoch": 2.915479582146249, + "grad_norm": 0.23864454253511638, + "learning_rate": 4.425297782557383e-05, + "loss": 0.0054, + "step": 6140 + }, + { + "epoch": 2.92022792022792, + "grad_norm": 0.2148931715518326, + "learning_rate": 4.4088359952824344e-05, + "loss": 0.006, + "step": 6150 + }, + { + "epoch": 2.9249762583095915, + "grad_norm": 0.1788365498147363, + "learning_rate": 4.392380704299955e-05, + "loss": 0.0061, + "step": 6160 + }, + { + "epoch": 2.929724596391263, + "grad_norm": 0.2021166974655368, + "learning_rate": 4.3759320904368913e-05, + "loss": 0.0056, + "step": 6170 + }, + { + "epoch": 2.9344729344729346, + "grad_norm": 0.1922520205391225, + "learning_rate": 4.359490334446819e-05, + "loss": 0.0061, + "step": 6180 + }, + { + "epoch": 2.939221272554606, + "grad_norm": 0.20749029229281218, + "learning_rate": 4.343055617007947e-05, + "loss": 0.0055, + "step": 6190 + }, + { + "epoch": 2.943969610636277, + "grad_norm": 0.1899433281058293, + "learning_rate": 4.326628118721144e-05, + "loss": 0.0059, + "step": 6200 + }, + { + "epoch": 2.948717948717949, + "grad_norm": 0.18302473567640612, + "learning_rate": 4.3102080201079435e-05, + "loss": 0.0056, + "step": 6210 + }, + { + "epoch": 2.95346628679962, + "grad_norm": 0.20792554643012584, + "learning_rate": 4.293795501608565e-05, + "loss": 0.0056, + "step": 6220 + }, + { + "epoch": 2.9582146248812915, + "grad_norm": 0.18515367736829666, + "learning_rate": 4.277390743579932e-05, + "loss": 0.006, + "step": 6230 + }, + { + "epoch": 2.962962962962963, + "grad_norm": 0.1955002202316591, + "learning_rate": 4.2609939262936835e-05, + "loss": 0.0057, + "step": 6240 + }, + { + "epoch": 2.9677113010446345, + "grad_norm": 0.23149844066749445, + "learning_rate": 4.244605229934206e-05, + "loss": 0.0059, + "step": 6250 + }, + { + "epoch": 2.972459639126306, + "grad_norm": 0.1490864691401419, + "learning_rate": 4.228224834596636e-05, + "loss": 0.0054, + "step": 6260 + }, + { + "epoch": 2.977207977207977, + "grad_norm": 0.19992886100490762, + "learning_rate": 4.211852920284899e-05, + "loss": 0.0058, + "step": 6270 + }, + { + "epoch": 2.981956315289649, + "grad_norm": 0.1438328382191847, + "learning_rate": 4.195489666909713e-05, + "loss": 0.0056, + "step": 6280 + }, + { + "epoch": 2.98670465337132, + "grad_norm": 0.2278490693729302, + "learning_rate": 4.179135254286631e-05, + "loss": 0.0057, + "step": 6290 + }, + { + "epoch": 2.9914529914529915, + "grad_norm": 0.20345560589023692, + "learning_rate": 4.162789862134047e-05, + "loss": 0.0063, + "step": 6300 + }, + { + "epoch": 2.996201329534663, + "grad_norm": 0.178431943998008, + "learning_rate": 4.146453670071236e-05, + "loss": 0.0055, + "step": 6310 + }, + { + "epoch": 3.000949667616334, + "grad_norm": 0.10451157629269937, + "learning_rate": 4.1301268576163665e-05, + "loss": 0.005, + "step": 6320 + }, + { + "epoch": 3.005698005698006, + "grad_norm": 0.10757973481364168, + "learning_rate": 4.11380960418454e-05, + "loss": 0.0029, + "step": 6330 + }, + { + "epoch": 3.010446343779677, + "grad_norm": 0.09720672789452832, + "learning_rate": 4.0975020890858115e-05, + "loss": 0.0026, + "step": 6340 + }, + { + "epoch": 3.0151946818613484, + "grad_norm": 0.16741740678593628, + "learning_rate": 4.081204491523222e-05, + "loss": 0.0024, + "step": 6350 + }, + { + "epoch": 3.0199430199430197, + "grad_norm": 0.13094917805203038, + "learning_rate": 4.06491699059083e-05, + "loss": 0.0024, + "step": 6360 + }, + { + "epoch": 3.0246913580246915, + "grad_norm": 0.11083281367168207, + "learning_rate": 4.04863976527174e-05, + "loss": 0.0021, + "step": 6370 + }, + { + "epoch": 3.0294396961063628, + "grad_norm": 0.11667243151229773, + "learning_rate": 4.032372994436142e-05, + "loss": 0.002, + "step": 6380 + }, + { + "epoch": 3.034188034188034, + "grad_norm": 0.09500703305808268, + "learning_rate": 4.016116856839335e-05, + "loss": 0.0019, + "step": 6390 + }, + { + "epoch": 3.038936372269706, + "grad_norm": 0.13655661743345057, + "learning_rate": 3.9998715311197785e-05, + "loss": 0.002, + "step": 6400 + }, + { + "epoch": 3.043684710351377, + "grad_norm": 0.10143166240629597, + "learning_rate": 3.9836371957971106e-05, + "loss": 0.002, + "step": 6410 + }, + { + "epoch": 3.0484330484330484, + "grad_norm": 0.11889668679842612, + "learning_rate": 3.967414029270206e-05, + "loss": 0.0019, + "step": 6420 + }, + { + "epoch": 3.0531813865147197, + "grad_norm": 0.06871550099538261, + "learning_rate": 3.9512022098151966e-05, + "loss": 0.002, + "step": 6430 + }, + { + "epoch": 3.0579297245963915, + "grad_norm": 0.13083863460075268, + "learning_rate": 3.935001915583526e-05, + "loss": 0.002, + "step": 6440 + }, + { + "epoch": 3.0626780626780628, + "grad_norm": 0.10800853794097405, + "learning_rate": 3.918813324599989e-05, + "loss": 0.002, + "step": 6450 + }, + { + "epoch": 3.067426400759734, + "grad_norm": 0.09578044823611381, + "learning_rate": 3.902636614760771e-05, + "loss": 0.0019, + "step": 6460 + }, + { + "epoch": 3.0721747388414054, + "grad_norm": 0.09890540611894306, + "learning_rate": 3.8864719638314955e-05, + "loss": 0.0019, + "step": 6470 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.13488696348247367, + "learning_rate": 3.87031954944527e-05, + "loss": 0.002, + "step": 6480 + }, + { + "epoch": 3.0816714150047484, + "grad_norm": 0.07595606529156355, + "learning_rate": 3.854179549100741e-05, + "loss": 0.0016, + "step": 6490 + }, + { + "epoch": 3.0864197530864197, + "grad_norm": 0.08607444755482696, + "learning_rate": 3.838052140160128e-05, + "loss": 0.0018, + "step": 6500 + }, + { + "epoch": 3.091168091168091, + "grad_norm": 0.09015558255472064, + "learning_rate": 3.821937499847291e-05, + "loss": 0.0019, + "step": 6510 + }, + { + "epoch": 3.0959164292497627, + "grad_norm": 0.09542015272878117, + "learning_rate": 3.805835805245772e-05, + "loss": 0.0017, + "step": 6520 + }, + { + "epoch": 3.100664767331434, + "grad_norm": 0.10577976413046235, + "learning_rate": 3.789747233296852e-05, + "loss": 0.0016, + "step": 6530 + }, + { + "epoch": 3.1054131054131053, + "grad_norm": 0.07869791265478253, + "learning_rate": 3.7736719607976125e-05, + "loss": 0.0016, + "step": 6540 + }, + { + "epoch": 3.1101614434947766, + "grad_norm": 0.09915705114614641, + "learning_rate": 3.7576101643989816e-05, + "loss": 0.0015, + "step": 6550 + }, + { + "epoch": 3.1149097815764484, + "grad_norm": 0.1429685266650327, + "learning_rate": 3.7415620206038036e-05, + "loss": 0.0017, + "step": 6560 + }, + { + "epoch": 3.1196581196581197, + "grad_norm": 0.1605759516781887, + "learning_rate": 3.7255277057648906e-05, + "loss": 0.0018, + "step": 6570 + }, + { + "epoch": 3.124406457739791, + "grad_norm": 0.0989130092327099, + "learning_rate": 3.7095073960830905e-05, + "loss": 0.0018, + "step": 6580 + }, + { + "epoch": 3.1291547958214627, + "grad_norm": 0.14717225232153788, + "learning_rate": 3.69350126760535e-05, + "loss": 0.0016, + "step": 6590 + }, + { + "epoch": 3.133903133903134, + "grad_norm": 0.19138993835268137, + "learning_rate": 3.677509496222776e-05, + "loss": 0.0017, + "step": 6600 + }, + { + "epoch": 3.1386514719848053, + "grad_norm": 0.07020464850249546, + "learning_rate": 3.661532257668705e-05, + "loss": 0.0018, + "step": 6610 + }, + { + "epoch": 3.1433998100664766, + "grad_norm": 0.10938326195616555, + "learning_rate": 3.645569727516779e-05, + "loss": 0.0019, + "step": 6620 + }, + { + "epoch": 3.148148148148148, + "grad_norm": 0.10298496067998036, + "learning_rate": 3.6296220811790015e-05, + "loss": 0.0015, + "step": 6630 + }, + { + "epoch": 3.1528964862298197, + "grad_norm": 0.10372878201912246, + "learning_rate": 3.613689493903819e-05, + "loss": 0.0018, + "step": 6640 + }, + { + "epoch": 3.157644824311491, + "grad_norm": 0.14028379381827602, + "learning_rate": 3.597772140774201e-05, + "loss": 0.0017, + "step": 6650 + }, + { + "epoch": 3.1623931623931623, + "grad_norm": 0.12058136386846804, + "learning_rate": 3.5818701967057014e-05, + "loss": 0.0017, + "step": 6660 + }, + { + "epoch": 3.167141500474834, + "grad_norm": 0.07946275548820565, + "learning_rate": 3.5659838364445505e-05, + "loss": 0.0016, + "step": 6670 + }, + { + "epoch": 3.1718898385565053, + "grad_norm": 0.08606483258687761, + "learning_rate": 3.5501132345657204e-05, + "loss": 0.0018, + "step": 6680 + }, + { + "epoch": 3.1766381766381766, + "grad_norm": 0.11183118138674843, + "learning_rate": 3.5342585654710256e-05, + "loss": 0.0017, + "step": 6690 + }, + { + "epoch": 3.181386514719848, + "grad_norm": 0.08081064536655497, + "learning_rate": 3.518420003387183e-05, + "loss": 0.0016, + "step": 6700 + }, + { + "epoch": 3.1861348528015196, + "grad_norm": 0.14163604956558828, + "learning_rate": 3.502597722363921e-05, + "loss": 0.0017, + "step": 6710 + }, + { + "epoch": 3.190883190883191, + "grad_norm": 0.119214547378923, + "learning_rate": 3.486791896272049e-05, + "loss": 0.0019, + "step": 6720 + }, + { + "epoch": 3.1956315289648622, + "grad_norm": 0.13101845859457906, + "learning_rate": 3.471002698801555e-05, + "loss": 0.0018, + "step": 6730 + }, + { + "epoch": 3.2003798670465335, + "grad_norm": 0.10484762514613516, + "learning_rate": 3.4552303034596965e-05, + "loss": 0.0016, + "step": 6740 + }, + { + "epoch": 3.2051282051282053, + "grad_norm": 0.09090211959992706, + "learning_rate": 3.4394748835690894e-05, + "loss": 0.0017, + "step": 6750 + }, + { + "epoch": 3.2098765432098766, + "grad_norm": 0.13778070921506094, + "learning_rate": 3.423736612265812e-05, + "loss": 0.0017, + "step": 6760 + }, + { + "epoch": 3.214624881291548, + "grad_norm": 0.1164953733737809, + "learning_rate": 3.4080156624974904e-05, + "loss": 0.0017, + "step": 6770 + }, + { + "epoch": 3.219373219373219, + "grad_norm": 0.11550400645927915, + "learning_rate": 3.3923122070214126e-05, + "loss": 0.0018, + "step": 6780 + }, + { + "epoch": 3.224121557454891, + "grad_norm": 0.12204858624482236, + "learning_rate": 3.376626418402612e-05, + "loss": 0.0015, + "step": 6790 + }, + { + "epoch": 3.2288698955365622, + "grad_norm": 0.07906689808808706, + "learning_rate": 3.3609584690119925e-05, + "loss": 0.0016, + "step": 6800 + }, + { + "epoch": 3.2336182336182335, + "grad_norm": 0.11285961916166666, + "learning_rate": 3.345308531024412e-05, + "loss": 0.0019, + "step": 6810 + }, + { + "epoch": 3.238366571699905, + "grad_norm": 0.09271896155249686, + "learning_rate": 3.329676776416811e-05, + "loss": 0.0017, + "step": 6820 + }, + { + "epoch": 3.2431149097815766, + "grad_norm": 0.10200613812442547, + "learning_rate": 3.3140633769663057e-05, + "loss": 0.0018, + "step": 6830 + }, + { + "epoch": 3.247863247863248, + "grad_norm": 0.09734106130901037, + "learning_rate": 3.298468504248313e-05, + "loss": 0.0016, + "step": 6840 + }, + { + "epoch": 3.252611585944919, + "grad_norm": 0.1414843916358992, + "learning_rate": 3.282892329634657e-05, + "loss": 0.0016, + "step": 6850 + }, + { + "epoch": 3.2573599240265905, + "grad_norm": 0.15244507865777615, + "learning_rate": 3.267335024291687e-05, + "loss": 0.0016, + "step": 6860 + }, + { + "epoch": 3.262108262108262, + "grad_norm": 0.0984653030055153, + "learning_rate": 3.251796759178403e-05, + "loss": 0.0015, + "step": 6870 + }, + { + "epoch": 3.2668566001899335, + "grad_norm": 0.11523169075964597, + "learning_rate": 3.236277705044568e-05, + "loss": 0.0016, + "step": 6880 + }, + { + "epoch": 3.271604938271605, + "grad_norm": 0.11635519618824158, + "learning_rate": 3.220778032428839e-05, + "loss": 0.0016, + "step": 6890 + }, + { + "epoch": 3.2763532763532766, + "grad_norm": 0.1190378108744281, + "learning_rate": 3.205297911656886e-05, + "loss": 0.0015, + "step": 6900 + }, + { + "epoch": 3.281101614434948, + "grad_norm": 0.10239012705908038, + "learning_rate": 3.18983751283953e-05, + "loss": 0.0015, + "step": 6910 + }, + { + "epoch": 3.285849952516619, + "grad_norm": 0.12668720003074493, + "learning_rate": 3.174397005870858e-05, + "loss": 0.0016, + "step": 6920 + }, + { + "epoch": 3.2905982905982905, + "grad_norm": 0.08009070068492502, + "learning_rate": 3.158976560426379e-05, + "loss": 0.0016, + "step": 6930 + }, + { + "epoch": 3.295346628679962, + "grad_norm": 0.10947814391707279, + "learning_rate": 3.143576345961132e-05, + "loss": 0.0015, + "step": 6940 + }, + { + "epoch": 3.3000949667616335, + "grad_norm": 0.08674767864631551, + "learning_rate": 3.128196531707849e-05, + "loss": 0.0016, + "step": 6950 + }, + { + "epoch": 3.304843304843305, + "grad_norm": 0.07917117777762983, + "learning_rate": 3.112837286675082e-05, + "loss": 0.0014, + "step": 6960 + }, + { + "epoch": 3.309591642924976, + "grad_norm": 0.1587577555431157, + "learning_rate": 3.097498779645341e-05, + "loss": 0.0014, + "step": 6970 + }, + { + "epoch": 3.314339981006648, + "grad_norm": 0.07978648150918838, + "learning_rate": 3.082181179173257e-05, + "loss": 0.0014, + "step": 6980 + }, + { + "epoch": 3.319088319088319, + "grad_norm": 0.09429891882064968, + "learning_rate": 3.0668846535837114e-05, + "loss": 0.0015, + "step": 6990 + }, + { + "epoch": 3.3238366571699904, + "grad_norm": 0.08026201146426383, + "learning_rate": 3.051609370969999e-05, + "loss": 0.0018, + "step": 7000 + }, + { + "epoch": 3.3285849952516617, + "grad_norm": 0.07935942002069693, + "learning_rate": 3.03635549919197e-05, + "loss": 0.0014, + "step": 7010 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.10785626130671347, + "learning_rate": 3.021123205874199e-05, + "loss": 0.0014, + "step": 7020 + }, + { + "epoch": 3.338081671415005, + "grad_norm": 0.10657228714290067, + "learning_rate": 3.0059126584041275e-05, + "loss": 0.0014, + "step": 7030 + }, + { + "epoch": 3.342830009496676, + "grad_norm": 0.08583088409406038, + "learning_rate": 2.9907240239302393e-05, + "loss": 0.0014, + "step": 7040 + }, + { + "epoch": 3.347578347578348, + "grad_norm": 0.07900720565157145, + "learning_rate": 2.9755574693602083e-05, + "loss": 0.0016, + "step": 7050 + }, + { + "epoch": 3.352326685660019, + "grad_norm": 0.13650510834864663, + "learning_rate": 2.9604131613590824e-05, + "loss": 0.0014, + "step": 7060 + }, + { + "epoch": 3.3570750237416904, + "grad_norm": 0.08823468869951828, + "learning_rate": 2.9452912663474348e-05, + "loss": 0.0016, + "step": 7070 + }, + { + "epoch": 3.3618233618233617, + "grad_norm": 0.08895021658567595, + "learning_rate": 2.9301919504995436e-05, + "loss": 0.0014, + "step": 7080 + }, + { + "epoch": 3.366571699905033, + "grad_norm": 0.10088282705160573, + "learning_rate": 2.915115379741571e-05, + "loss": 0.0014, + "step": 7090 + }, + { + "epoch": 3.3713200379867048, + "grad_norm": 0.06870611685801774, + "learning_rate": 2.900061719749724e-05, + "loss": 0.0014, + "step": 7100 + }, + { + "epoch": 3.376068376068376, + "grad_norm": 0.11777634119927832, + "learning_rate": 2.8850311359484527e-05, + "loss": 0.0016, + "step": 7110 + }, + { + "epoch": 3.3808167141500474, + "grad_norm": 0.15947828434910663, + "learning_rate": 2.870023793508616e-05, + "loss": 0.0015, + "step": 7120 + }, + { + "epoch": 3.385565052231719, + "grad_norm": 0.11428246948087106, + "learning_rate": 2.855039857345681e-05, + "loss": 0.0015, + "step": 7130 + }, + { + "epoch": 3.3903133903133904, + "grad_norm": 0.08951049314265355, + "learning_rate": 2.8400794921178953e-05, + "loss": 0.0015, + "step": 7140 + }, + { + "epoch": 3.3950617283950617, + "grad_norm": 0.09381614633634876, + "learning_rate": 2.825142862224494e-05, + "loss": 0.0015, + "step": 7150 + }, + { + "epoch": 3.399810066476733, + "grad_norm": 0.09533863022309419, + "learning_rate": 2.810230131803878e-05, + "loss": 0.0014, + "step": 7160 + }, + { + "epoch": 3.4045584045584047, + "grad_norm": 0.07223014496978093, + "learning_rate": 2.7953414647318232e-05, + "loss": 0.0014, + "step": 7170 + }, + { + "epoch": 3.409306742640076, + "grad_norm": 0.08396228776894232, + "learning_rate": 2.78047702461967e-05, + "loss": 0.0015, + "step": 7180 + }, + { + "epoch": 3.4140550807217473, + "grad_norm": 0.09888926155032769, + "learning_rate": 2.7656369748125276e-05, + "loss": 0.0014, + "step": 7190 + }, + { + "epoch": 3.4188034188034186, + "grad_norm": 0.13185891357720556, + "learning_rate": 2.7508214783874874e-05, + "loss": 0.0015, + "step": 7200 + }, + { + "epoch": 3.4235517568850904, + "grad_norm": 0.09552101765656103, + "learning_rate": 2.7360306981518146e-05, + "loss": 0.0015, + "step": 7210 + }, + { + "epoch": 3.4283000949667617, + "grad_norm": 0.06984102271804707, + "learning_rate": 2.721264796641178e-05, + "loss": 0.0014, + "step": 7220 + }, + { + "epoch": 3.433048433048433, + "grad_norm": 0.1043893525346158, + "learning_rate": 2.7065239361178464e-05, + "loss": 0.0013, + "step": 7230 + }, + { + "epoch": 3.4377967711301043, + "grad_norm": 0.06890544899146898, + "learning_rate": 2.6918082785689208e-05, + "loss": 0.0011, + "step": 7240 + }, + { + "epoch": 3.442545109211776, + "grad_norm": 0.14594935065839373, + "learning_rate": 2.6771179857045396e-05, + "loss": 0.0013, + "step": 7250 + }, + { + "epoch": 3.4472934472934473, + "grad_norm": 0.12200478574387091, + "learning_rate": 2.6624532189561145e-05, + "loss": 0.0014, + "step": 7260 + }, + { + "epoch": 3.4520417853751186, + "grad_norm": 0.10323538637105818, + "learning_rate": 2.6478141394745524e-05, + "loss": 0.0015, + "step": 7270 + }, + { + "epoch": 3.45679012345679, + "grad_norm": 0.1033828315243673, + "learning_rate": 2.6332009081284754e-05, + "loss": 0.0014, + "step": 7280 + }, + { + "epoch": 3.4615384615384617, + "grad_norm": 0.10378520172824637, + "learning_rate": 2.61861368550247e-05, + "loss": 0.0012, + "step": 7290 + }, + { + "epoch": 3.466286799620133, + "grad_norm": 0.10250890091459743, + "learning_rate": 2.6040526318953052e-05, + "loss": 0.0013, + "step": 7300 + }, + { + "epoch": 3.4710351377018043, + "grad_norm": 0.06809913995404888, + "learning_rate": 2.5895179073181863e-05, + "loss": 0.0014, + "step": 7310 + }, + { + "epoch": 3.4757834757834756, + "grad_norm": 0.10879291913291235, + "learning_rate": 2.575009671492983e-05, + "loss": 0.0015, + "step": 7320 + }, + { + "epoch": 3.4805318138651473, + "grad_norm": 0.10150780212940495, + "learning_rate": 2.5605280838504863e-05, + "loss": 0.0013, + "step": 7330 + }, + { + "epoch": 3.4852801519468186, + "grad_norm": 0.13090716714139125, + "learning_rate": 2.5460733035286432e-05, + "loss": 0.0013, + "step": 7340 + }, + { + "epoch": 3.49002849002849, + "grad_norm": 0.10032583395130536, + "learning_rate": 2.5316454893708252e-05, + "loss": 0.0014, + "step": 7350 + }, + { + "epoch": 3.4947768281101617, + "grad_norm": 0.11088832473195062, + "learning_rate": 2.517244799924064e-05, + "loss": 0.0013, + "step": 7360 + }, + { + "epoch": 3.499525166191833, + "grad_norm": 0.10080319726203933, + "learning_rate": 2.502871393437325e-05, + "loss": 0.0013, + "step": 7370 + }, + { + "epoch": 3.5042735042735043, + "grad_norm": 0.08134863562162185, + "learning_rate": 2.4885254278597564e-05, + "loss": 0.0013, + "step": 7380 + }, + { + "epoch": 3.5090218423551756, + "grad_norm": 0.07771526519709729, + "learning_rate": 2.47420706083896e-05, + "loss": 0.0014, + "step": 7390 + }, + { + "epoch": 3.513770180436847, + "grad_norm": 0.08220937848089546, + "learning_rate": 2.4599164497192617e-05, + "loss": 0.0011, + "step": 7400 + }, + { + "epoch": 3.5185185185185186, + "grad_norm": 0.1169343120029651, + "learning_rate": 2.4456537515399707e-05, + "loss": 0.0012, + "step": 7410 + }, + { + "epoch": 3.52326685660019, + "grad_norm": 0.11323896486199661, + "learning_rate": 2.4314191230336702e-05, + "loss": 0.0015, + "step": 7420 + }, + { + "epoch": 3.528015194681861, + "grad_norm": 0.10448704659072139, + "learning_rate": 2.4172127206244766e-05, + "loss": 0.0012, + "step": 7430 + }, + { + "epoch": 3.532763532763533, + "grad_norm": 0.13276131005267122, + "learning_rate": 2.4030347004263392e-05, + "loss": 0.0016, + "step": 7440 + }, + { + "epoch": 3.5375118708452042, + "grad_norm": 0.11356049037822034, + "learning_rate": 2.3888852182413085e-05, + "loss": 0.0013, + "step": 7450 + }, + { + "epoch": 3.5422602089268755, + "grad_norm": 0.1267906600648582, + "learning_rate": 2.3747644295578364e-05, + "loss": 0.0013, + "step": 7460 + }, + { + "epoch": 3.547008547008547, + "grad_norm": 0.08961176186073622, + "learning_rate": 2.3606724895490568e-05, + "loss": 0.0013, + "step": 7470 + }, + { + "epoch": 3.5517568850902186, + "grad_norm": 0.07378427948986291, + "learning_rate": 2.346609553071093e-05, + "loss": 0.0012, + "step": 7480 + }, + { + "epoch": 3.55650522317189, + "grad_norm": 0.07235876134536022, + "learning_rate": 2.332575774661342e-05, + "loss": 0.0012, + "step": 7490 + }, + { + "epoch": 3.561253561253561, + "grad_norm": 0.07718465016907564, + "learning_rate": 2.3185713085367866e-05, + "loss": 0.0012, + "step": 7500 + }, + { + "epoch": 3.566001899335233, + "grad_norm": 0.11456655023906966, + "learning_rate": 2.3045963085923007e-05, + "loss": 0.0012, + "step": 7510 + }, + { + "epoch": 3.5707502374169042, + "grad_norm": 0.07325271872731128, + "learning_rate": 2.2906509283989462e-05, + "loss": 0.0013, + "step": 7520 + }, + { + "epoch": 3.5754985754985755, + "grad_norm": 0.08677972709172467, + "learning_rate": 2.276735321202303e-05, + "loss": 0.0014, + "step": 7530 + }, + { + "epoch": 3.580246913580247, + "grad_norm": 0.1012841744176148, + "learning_rate": 2.2628496399207683e-05, + "loss": 0.0012, + "step": 7540 + }, + { + "epoch": 3.584995251661918, + "grad_norm": 0.10939119010384225, + "learning_rate": 2.2489940371438912e-05, + "loss": 0.0013, + "step": 7550 + }, + { + "epoch": 3.58974358974359, + "grad_norm": 0.09339784648237463, + "learning_rate": 2.2351686651306815e-05, + "loss": 0.001, + "step": 7560 + }, + { + "epoch": 3.594491927825261, + "grad_norm": 0.07731287987710464, + "learning_rate": 2.2213736758079494e-05, + "loss": 0.0012, + "step": 7570 + }, + { + "epoch": 3.5992402659069325, + "grad_norm": 0.09595986097061222, + "learning_rate": 2.207609220768625e-05, + "loss": 0.0011, + "step": 7580 + }, + { + "epoch": 3.603988603988604, + "grad_norm": 0.1063634983845274, + "learning_rate": 2.1938754512701042e-05, + "loss": 0.0012, + "step": 7590 + }, + { + "epoch": 3.6087369420702755, + "grad_norm": 0.054655067711459726, + "learning_rate": 2.1801725182325733e-05, + "loss": 0.0012, + "step": 7600 + }, + { + "epoch": 3.613485280151947, + "grad_norm": 0.07443683864927408, + "learning_rate": 2.1665005722373588e-05, + "loss": 0.0013, + "step": 7610 + }, + { + "epoch": 3.618233618233618, + "grad_norm": 0.12349933555524534, + "learning_rate": 2.1528597635252756e-05, + "loss": 0.0011, + "step": 7620 + }, + { + "epoch": 3.6229819563152894, + "grad_norm": 0.14306715334968834, + "learning_rate": 2.139250241994964e-05, + "loss": 0.0013, + "step": 7630 + }, + { + "epoch": 3.627730294396961, + "grad_norm": 0.11546620703383373, + "learning_rate": 2.1256721572012584e-05, + "loss": 0.0013, + "step": 7640 + }, + { + "epoch": 3.6324786324786325, + "grad_norm": 0.10352377074724403, + "learning_rate": 2.1121256583535272e-05, + "loss": 0.0013, + "step": 7650 + }, + { + "epoch": 3.6372269705603038, + "grad_norm": 0.10247187325991285, + "learning_rate": 2.098610894314048e-05, + "loss": 0.0013, + "step": 7660 + }, + { + "epoch": 3.6419753086419755, + "grad_norm": 0.053718535329379105, + "learning_rate": 2.085128013596359e-05, + "loss": 0.0011, + "step": 7670 + }, + { + "epoch": 3.646723646723647, + "grad_norm": 0.12494537495294165, + "learning_rate": 2.07167716436364e-05, + "loss": 0.0012, + "step": 7680 + }, + { + "epoch": 3.651471984805318, + "grad_norm": 0.13671566290423642, + "learning_rate": 2.0582584944270694e-05, + "loss": 0.001, + "step": 7690 + }, + { + "epoch": 3.6562203228869894, + "grad_norm": 0.08147317408779682, + "learning_rate": 2.0448721512442153e-05, + "loss": 0.0013, + "step": 7700 + }, + { + "epoch": 3.6609686609686607, + "grad_norm": 0.08048416657443642, + "learning_rate": 2.031518281917403e-05, + "loss": 0.001, + "step": 7710 + }, + { + "epoch": 3.6657169990503324, + "grad_norm": 0.10722102262479334, + "learning_rate": 2.018197033192102e-05, + "loss": 0.0012, + "step": 7720 + }, + { + "epoch": 3.6704653371320037, + "grad_norm": 0.09006599852646181, + "learning_rate": 2.0049085514553183e-05, + "loss": 0.001, + "step": 7730 + }, + { + "epoch": 3.6752136752136755, + "grad_norm": 0.07227302662023415, + "learning_rate": 1.9916529827339825e-05, + "loss": 0.0013, + "step": 7740 + }, + { + "epoch": 3.679962013295347, + "grad_norm": 0.08853426618084388, + "learning_rate": 1.9784304726933383e-05, + "loss": 0.001, + "step": 7750 + }, + { + "epoch": 3.684710351377018, + "grad_norm": 0.09156249686525142, + "learning_rate": 1.9652411666353547e-05, + "loss": 0.0011, + "step": 7760 + }, + { + "epoch": 3.6894586894586894, + "grad_norm": 0.1421246582807511, + "learning_rate": 1.9520852094971215e-05, + "loss": 0.0011, + "step": 7770 + }, + { + "epoch": 3.6942070275403607, + "grad_norm": 0.054123516064340876, + "learning_rate": 1.9389627458492526e-05, + "loss": 0.0011, + "step": 7780 + }, + { + "epoch": 3.6989553656220324, + "grad_norm": 0.14105509148692613, + "learning_rate": 1.9258739198943105e-05, + "loss": 0.0011, + "step": 7790 + }, + { + "epoch": 3.7037037037037037, + "grad_norm": 0.1065073736928368, + "learning_rate": 1.9128188754652042e-05, + "loss": 0.0012, + "step": 7800 + }, + { + "epoch": 3.708452041785375, + "grad_norm": 0.09378520852841594, + "learning_rate": 1.8997977560236276e-05, + "loss": 0.0011, + "step": 7810 + }, + { + "epoch": 3.7132003798670468, + "grad_norm": 0.12495330561669392, + "learning_rate": 1.886810704658467e-05, + "loss": 0.0011, + "step": 7820 + }, + { + "epoch": 3.717948717948718, + "grad_norm": 0.12389934302275724, + "learning_rate": 1.8738578640842353e-05, + "loss": 0.0013, + "step": 7830 + }, + { + "epoch": 3.7226970560303894, + "grad_norm": 0.09327182662323678, + "learning_rate": 1.8609393766395085e-05, + "loss": 0.0011, + "step": 7840 + }, + { + "epoch": 3.7274453941120607, + "grad_norm": 0.10935750897374744, + "learning_rate": 1.8480553842853492e-05, + "loss": 0.0012, + "step": 7850 + }, + { + "epoch": 3.732193732193732, + "grad_norm": 0.14111752628471985, + "learning_rate": 1.8352060286037626e-05, + "loss": 0.0011, + "step": 7860 + }, + { + "epoch": 3.7369420702754037, + "grad_norm": 0.10248488118877974, + "learning_rate": 1.8223914507961232e-05, + "loss": 0.0012, + "step": 7870 + }, + { + "epoch": 3.741690408357075, + "grad_norm": 0.12071236456569236, + "learning_rate": 1.80961179168164e-05, + "loss": 0.0013, + "step": 7880 + }, + { + "epoch": 3.7464387464387463, + "grad_norm": 0.15740223594534422, + "learning_rate": 1.796867191695794e-05, + "loss": 0.0013, + "step": 7890 + }, + { + "epoch": 3.751187084520418, + "grad_norm": 0.11181340063731016, + "learning_rate": 1.784157790888808e-05, + "loss": 0.0011, + "step": 7900 + }, + { + "epoch": 3.7559354226020893, + "grad_norm": 0.10318873973882561, + "learning_rate": 1.7714837289240958e-05, + "loss": 0.0012, + "step": 7910 + }, + { + "epoch": 3.7606837606837606, + "grad_norm": 0.06252018033723301, + "learning_rate": 1.7588451450767396e-05, + "loss": 0.001, + "step": 7920 + }, + { + "epoch": 3.765432098765432, + "grad_norm": 0.10132463915812508, + "learning_rate": 1.7462421782319472e-05, + "loss": 0.0012, + "step": 7930 + }, + { + "epoch": 3.7701804368471032, + "grad_norm": 0.0903457736382397, + "learning_rate": 1.733674966883533e-05, + "loss": 0.0011, + "step": 7940 + }, + { + "epoch": 3.774928774928775, + "grad_norm": 0.08557083052751926, + "learning_rate": 1.721143649132399e-05, + "loss": 0.0011, + "step": 7950 + }, + { + "epoch": 3.7796771130104463, + "grad_norm": 0.09670226114398535, + "learning_rate": 1.7086483626850063e-05, + "loss": 0.001, + "step": 7960 + }, + { + "epoch": 3.784425451092118, + "grad_norm": 0.06416483076913425, + "learning_rate": 1.696189244851874e-05, + "loss": 0.001, + "step": 7970 + }, + { + "epoch": 3.7891737891737893, + "grad_norm": 0.06786885651816917, + "learning_rate": 1.683766432546065e-05, + "loss": 0.001, + "step": 7980 + }, + { + "epoch": 3.7939221272554606, + "grad_norm": 0.0908205317957609, + "learning_rate": 1.671380062281675e-05, + "loss": 0.0009, + "step": 7990 + }, + { + "epoch": 3.798670465337132, + "grad_norm": 0.0995070478791982, + "learning_rate": 1.6590302701723454e-05, + "loss": 0.0012, + "step": 8000 + }, + { + "epoch": 3.8034188034188032, + "grad_norm": 0.06989695024427044, + "learning_rate": 1.6467171919297598e-05, + "loss": 0.001, + "step": 8010 + }, + { + "epoch": 3.808167141500475, + "grad_norm": 0.09779953954059863, + "learning_rate": 1.6344409628621484e-05, + "loss": 0.0012, + "step": 8020 + }, + { + "epoch": 3.8129154795821463, + "grad_norm": 0.059790628661728164, + "learning_rate": 1.6222017178728134e-05, + "loss": 0.0012, + "step": 8030 + }, + { + "epoch": 3.8176638176638176, + "grad_norm": 0.0537379710426186, + "learning_rate": 1.6099995914586347e-05, + "loss": 0.0011, + "step": 8040 + }, + { + "epoch": 3.8224121557454893, + "grad_norm": 0.06901268400903983, + "learning_rate": 1.5978347177085967e-05, + "loss": 0.0011, + "step": 8050 + }, + { + "epoch": 3.8271604938271606, + "grad_norm": 0.10770300221818542, + "learning_rate": 1.5857072303023195e-05, + "loss": 0.0009, + "step": 8060 + }, + { + "epoch": 3.831908831908832, + "grad_norm": 0.08713835045154426, + "learning_rate": 1.5736172625085792e-05, + "loss": 0.0011, + "step": 8070 + }, + { + "epoch": 3.836657169990503, + "grad_norm": 0.09407173901082683, + "learning_rate": 1.5615649471838555e-05, + "loss": 0.0011, + "step": 8080 + }, + { + "epoch": 3.8414055080721745, + "grad_norm": 0.12615702825207997, + "learning_rate": 1.5495504167708592e-05, + "loss": 0.0012, + "step": 8090 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.0747717307697842, + "learning_rate": 1.5375738032970893e-05, + "loss": 0.0009, + "step": 8100 + }, + { + "epoch": 3.8509021842355176, + "grad_norm": 0.08579800008481188, + "learning_rate": 1.5256352383733707e-05, + "loss": 0.001, + "step": 8110 + }, + { + "epoch": 3.855650522317189, + "grad_norm": 0.10021833281793126, + "learning_rate": 1.513734853192419e-05, + "loss": 0.0009, + "step": 8120 + }, + { + "epoch": 3.8603988603988606, + "grad_norm": 0.13237848111958933, + "learning_rate": 1.5018727785273872e-05, + "loss": 0.0009, + "step": 8130 + }, + { + "epoch": 3.865147198480532, + "grad_norm": 0.08008885582859418, + "learning_rate": 1.4900491447304409e-05, + "loss": 0.001, + "step": 8140 + }, + { + "epoch": 3.869895536562203, + "grad_norm": 0.061884886976256485, + "learning_rate": 1.4782640817313149e-05, + "loss": 0.0009, + "step": 8150 + }, + { + "epoch": 3.8746438746438745, + "grad_norm": 0.05911139957237997, + "learning_rate": 1.4665177190358886e-05, + "loss": 0.001, + "step": 8160 + }, + { + "epoch": 3.879392212725546, + "grad_norm": 0.0736445138024089, + "learning_rate": 1.454810185724772e-05, + "loss": 0.0011, + "step": 8170 + }, + { + "epoch": 3.8841405508072175, + "grad_norm": 0.05575627581403319, + "learning_rate": 1.4431416104518708e-05, + "loss": 0.0009, + "step": 8180 + }, + { + "epoch": 3.888888888888889, + "grad_norm": 0.06375136704719277, + "learning_rate": 1.4315121214429882e-05, + "loss": 0.0011, + "step": 8190 + }, + { + "epoch": 3.89363722697056, + "grad_norm": 0.11349248043385199, + "learning_rate": 1.4199218464944042e-05, + "loss": 0.0012, + "step": 8200 + }, + { + "epoch": 3.898385565052232, + "grad_norm": 0.09161932779191004, + "learning_rate": 1.40837091297148e-05, + "loss": 0.001, + "step": 8210 + }, + { + "epoch": 3.903133903133903, + "grad_norm": 0.08694198456623081, + "learning_rate": 1.39685944780725e-05, + "loss": 0.001, + "step": 8220 + }, + { + "epoch": 3.9078822412155745, + "grad_norm": 0.036404382868630576, + "learning_rate": 1.3853875775010355e-05, + "loss": 0.0009, + "step": 8230 + }, + { + "epoch": 3.912630579297246, + "grad_norm": 0.09217147132867635, + "learning_rate": 1.3739554281170509e-05, + "loss": 0.0009, + "step": 8240 + }, + { + "epoch": 3.9173789173789175, + "grad_norm": 0.12331637559433002, + "learning_rate": 1.3625631252830118e-05, + "loss": 0.0011, + "step": 8250 + }, + { + "epoch": 3.922127255460589, + "grad_norm": 0.07638147399027147, + "learning_rate": 1.3512107941887686e-05, + "loss": 0.001, + "step": 8260 + }, + { + "epoch": 3.92687559354226, + "grad_norm": 0.04178827000575975, + "learning_rate": 1.3398985595849156e-05, + "loss": 0.001, + "step": 8270 + }, + { + "epoch": 3.931623931623932, + "grad_norm": 0.07534611132802999, + "learning_rate": 1.3286265457814345e-05, + "loss": 0.0009, + "step": 8280 + }, + { + "epoch": 3.936372269705603, + "grad_norm": 0.07314285016464729, + "learning_rate": 1.3173948766463145e-05, + "loss": 0.0008, + "step": 8290 + }, + { + "epoch": 3.9411206077872745, + "grad_norm": 0.08359320227097702, + "learning_rate": 1.3062036756042034e-05, + "loss": 0.0009, + "step": 8300 + }, + { + "epoch": 3.9458689458689458, + "grad_norm": 0.07090270290005515, + "learning_rate": 1.2950530656350413e-05, + "loss": 0.0009, + "step": 8310 + }, + { + "epoch": 3.950617283950617, + "grad_norm": 0.09975041548756264, + "learning_rate": 1.2839431692727172e-05, + "loss": 0.0009, + "step": 8320 + }, + { + "epoch": 3.955365622032289, + "grad_norm": 0.10287272313793343, + "learning_rate": 1.2728741086037149e-05, + "loss": 0.0008, + "step": 8330 + }, + { + "epoch": 3.96011396011396, + "grad_norm": 0.09058702037745629, + "learning_rate": 1.2618460052657804e-05, + "loss": 0.0009, + "step": 8340 + }, + { + "epoch": 3.9648622981956314, + "grad_norm": 0.07635875135457879, + "learning_rate": 1.2508589804465765e-05, + "loss": 0.0009, + "step": 8350 + }, + { + "epoch": 3.969610636277303, + "grad_norm": 0.09800967965707635, + "learning_rate": 1.2399131548823539e-05, + "loss": 0.0009, + "step": 8360 + }, + { + "epoch": 3.9743589743589745, + "grad_norm": 0.07513784840284557, + "learning_rate": 1.2290086488566305e-05, + "loss": 0.0009, + "step": 8370 + }, + { + "epoch": 3.9791073124406457, + "grad_norm": 0.08710096640970359, + "learning_rate": 1.2181455821988574e-05, + "loss": 0.0009, + "step": 8380 + }, + { + "epoch": 3.983855650522317, + "grad_norm": 0.06031671702498948, + "learning_rate": 1.2073240742831165e-05, + "loss": 0.0008, + "step": 8390 + }, + { + "epoch": 3.9886039886039883, + "grad_norm": 0.10258430057872535, + "learning_rate": 1.1965442440267943e-05, + "loss": 0.0008, + "step": 8400 + }, + { + "epoch": 3.99335232668566, + "grad_norm": 0.09157446729360551, + "learning_rate": 1.185806209889288e-05, + "loss": 0.0009, + "step": 8410 + }, + { + "epoch": 3.9981006647673314, + "grad_norm": 0.06547432995021957, + "learning_rate": 1.1751100898706929e-05, + "loss": 0.0008, + "step": 8420 + }, + { + "epoch": 4.002849002849003, + "grad_norm": 0.02599315912982386, + "learning_rate": 1.1644560015105155e-05, + "loss": 0.0006, + "step": 8430 + }, + { + "epoch": 4.007597340930674, + "grad_norm": 0.03893363494604501, + "learning_rate": 1.1538440618863722e-05, + "loss": 0.0004, + "step": 8440 + }, + { + "epoch": 4.012345679012346, + "grad_norm": 0.031639329682037004, + "learning_rate": 1.1432743876127122e-05, + "loss": 0.0003, + "step": 8450 + }, + { + "epoch": 4.017094017094017, + "grad_norm": 0.0228051884482049, + "learning_rate": 1.1327470948395263e-05, + "loss": 0.0004, + "step": 8460 + }, + { + "epoch": 4.021842355175688, + "grad_norm": 0.016533944524247213, + "learning_rate": 1.1222622992510807e-05, + "loss": 0.0003, + "step": 8470 + }, + { + "epoch": 4.02659069325736, + "grad_norm": 0.016802537176278144, + "learning_rate": 1.1118201160646386e-05, + "loss": 0.0003, + "step": 8480 + }, + { + "epoch": 4.031339031339031, + "grad_norm": 0.01446917946731896, + "learning_rate": 1.1014206600291938e-05, + "loss": 0.0003, + "step": 8490 + }, + { + "epoch": 4.036087369420703, + "grad_norm": 0.03205293659400886, + "learning_rate": 1.0910640454242172e-05, + "loss": 0.0003, + "step": 8500 + }, + { + "epoch": 4.040835707502374, + "grad_norm": 0.026548737463652775, + "learning_rate": 1.0807503860583895e-05, + "loss": 0.0003, + "step": 8510 + }, + { + "epoch": 4.045584045584046, + "grad_norm": 0.02056511004780097, + "learning_rate": 1.0704797952683638e-05, + "loss": 0.0003, + "step": 8520 + }, + { + "epoch": 4.050332383665717, + "grad_norm": 0.027569824235526886, + "learning_rate": 1.0602523859175061e-05, + "loss": 0.0003, + "step": 8530 + }, + { + "epoch": 4.055080721747388, + "grad_norm": 0.015347291083060554, + "learning_rate": 1.0500682703946695e-05, + "loss": 0.0004, + "step": 8540 + }, + { + "epoch": 4.05982905982906, + "grad_norm": 0.026571298937273724, + "learning_rate": 1.0399275606129449e-05, + "loss": 0.0003, + "step": 8550 + }, + { + "epoch": 4.064577397910731, + "grad_norm": 0.030370053410486604, + "learning_rate": 1.0298303680084448e-05, + "loss": 0.0003, + "step": 8560 + }, + { + "epoch": 4.069325735992402, + "grad_norm": 0.06471925852444672, + "learning_rate": 1.0197768035390665e-05, + "loss": 0.0003, + "step": 8570 + }, + { + "epoch": 4.074074074074074, + "grad_norm": 0.020730991411452303, + "learning_rate": 1.0097669776832801e-05, + "loss": 0.0003, + "step": 8580 + }, + { + "epoch": 4.078822412155746, + "grad_norm": 0.026057807920312384, + "learning_rate": 9.998010004389158e-06, + "loss": 0.0003, + "step": 8590 + }, + { + "epoch": 4.083570750237417, + "grad_norm": 0.015758051574172868, + "learning_rate": 9.898789813219467e-06, + "loss": 0.0003, + "step": 8600 + }, + { + "epoch": 4.088319088319088, + "grad_norm": 0.020709335217904246, + "learning_rate": 9.800010293652962e-06, + "loss": 0.0003, + "step": 8610 + }, + { + "epoch": 4.09306742640076, + "grad_norm": 0.02466714317527755, + "learning_rate": 9.701672531176286e-06, + "loss": 0.0003, + "step": 8620 + }, + { + "epoch": 4.097815764482431, + "grad_norm": 0.023642190630476463, + "learning_rate": 9.603777606421682e-06, + "loss": 0.0003, + "step": 8630 + }, + { + "epoch": 4.102564102564102, + "grad_norm": 0.01791312924905593, + "learning_rate": 9.506326595155002e-06, + "loss": 0.0003, + "step": 8640 + }, + { + "epoch": 4.107312440645774, + "grad_norm": 0.052309383641723294, + "learning_rate": 9.40932056826398e-06, + "loss": 0.0003, + "step": 8650 + }, + { + "epoch": 4.112060778727446, + "grad_norm": 0.025346423059104534, + "learning_rate": 9.312760591746377e-06, + "loss": 0.0003, + "step": 8660 + }, + { + "epoch": 4.116809116809117, + "grad_norm": 0.036039699043899365, + "learning_rate": 9.216647726698374e-06, + "loss": 0.0002, + "step": 8670 + }, + { + "epoch": 4.121557454890788, + "grad_norm": 0.030879140218773932, + "learning_rate": 9.120983029302805e-06, + "loss": 0.0003, + "step": 8680 + }, + { + "epoch": 4.12630579297246, + "grad_norm": 0.017524802882518026, + "learning_rate": 9.025767550817588e-06, + "loss": 0.0003, + "step": 8690 + }, + { + "epoch": 4.131054131054131, + "grad_norm": 0.03805745792424522, + "learning_rate": 8.93100233756422e-06, + "loss": 0.0003, + "step": 8700 + }, + { + "epoch": 4.135802469135802, + "grad_norm": 0.04599204118670921, + "learning_rate": 8.83668843091624e-06, + "loss": 0.0003, + "step": 8710 + }, + { + "epoch": 4.1405508072174735, + "grad_norm": 0.02404014654544189, + "learning_rate": 8.74282686728773e-06, + "loss": 0.0003, + "step": 8720 + }, + { + "epoch": 4.145299145299146, + "grad_norm": 0.07647279042120554, + "learning_rate": 8.64941867812204e-06, + "loss": 0.0003, + "step": 8730 + }, + { + "epoch": 4.150047483380817, + "grad_norm": 0.019210915321365034, + "learning_rate": 8.556464889880383e-06, + "loss": 0.0003, + "step": 8740 + }, + { + "epoch": 4.154795821462488, + "grad_norm": 0.014644114836617051, + "learning_rate": 8.46396652403053e-06, + "loss": 0.0003, + "step": 8750 + }, + { + "epoch": 4.15954415954416, + "grad_norm": 0.014638340751767762, + "learning_rate": 8.371924597035685e-06, + "loss": 0.0003, + "step": 8760 + }, + { + "epoch": 4.164292497625831, + "grad_norm": 0.020265349263430748, + "learning_rate": 8.280340120343182e-06, + "loss": 0.0003, + "step": 8770 + }, + { + "epoch": 4.169040835707502, + "grad_norm": 0.0627046829145093, + "learning_rate": 8.189214100373499e-06, + "loss": 0.0003, + "step": 8780 + }, + { + "epoch": 4.1737891737891735, + "grad_norm": 0.11256935663616527, + "learning_rate": 8.09854753850911e-06, + "loss": 0.0003, + "step": 8790 + }, + { + "epoch": 4.178537511870845, + "grad_norm": 0.011100918681496498, + "learning_rate": 8.0083414310835e-06, + "loss": 0.0003, + "step": 8800 + }, + { + "epoch": 4.183285849952517, + "grad_norm": 0.0578729308546147, + "learning_rate": 7.918596769370273e-06, + "loss": 0.0003, + "step": 8810 + }, + { + "epoch": 4.188034188034188, + "grad_norm": 0.04057057465519802, + "learning_rate": 7.829314539572152e-06, + "loss": 0.0003, + "step": 8820 + }, + { + "epoch": 4.1927825261158596, + "grad_norm": 0.0205520993550686, + "learning_rate": 7.740495722810271e-06, + "loss": 0.0003, + "step": 8830 + }, + { + "epoch": 4.197530864197531, + "grad_norm": 0.027015583755476014, + "learning_rate": 7.652141295113246e-06, + "loss": 0.0003, + "step": 8840 + }, + { + "epoch": 4.202279202279202, + "grad_norm": 0.03093406021400823, + "learning_rate": 7.56425222740661e-06, + "loss": 0.0003, + "step": 8850 + }, + { + "epoch": 4.2070275403608735, + "grad_norm": 0.009298335484269265, + "learning_rate": 7.476829485501996e-06, + "loss": 0.0003, + "step": 8860 + }, + { + "epoch": 4.211775878442545, + "grad_norm": 0.017740233638943274, + "learning_rate": 7.389874030086636e-06, + "loss": 0.0003, + "step": 8870 + }, + { + "epoch": 4.216524216524217, + "grad_norm": 0.050244909630937036, + "learning_rate": 7.303386816712715e-06, + "loss": 0.0003, + "step": 8880 + }, + { + "epoch": 4.221272554605888, + "grad_norm": 0.022098390429931083, + "learning_rate": 7.217368795786961e-06, + "loss": 0.0003, + "step": 8890 + }, + { + "epoch": 4.2260208926875595, + "grad_norm": 0.019174173223264264, + "learning_rate": 7.131820912560116e-06, + "loss": 0.0003, + "step": 8900 + }, + { + "epoch": 4.230769230769231, + "grad_norm": 0.01543476816871513, + "learning_rate": 7.046744107116587e-06, + "loss": 0.0003, + "step": 8910 + }, + { + "epoch": 4.235517568850902, + "grad_norm": 0.025821657028533302, + "learning_rate": 6.962139314364152e-06, + "loss": 0.0003, + "step": 8920 + }, + { + "epoch": 4.240265906932573, + "grad_norm": 0.03137567466882486, + "learning_rate": 6.878007464023595e-06, + "loss": 0.0003, + "step": 8930 + }, + { + "epoch": 4.245014245014245, + "grad_norm": 0.011123210289366765, + "learning_rate": 6.794349480618583e-06, + "loss": 0.0002, + "step": 8940 + }, + { + "epoch": 4.249762583095916, + "grad_norm": 0.05925740818922852, + "learning_rate": 6.711166283465464e-06, + "loss": 0.0002, + "step": 8950 + }, + { + "epoch": 4.254510921177588, + "grad_norm": 0.015747603012631683, + "learning_rate": 6.628458786663133e-06, + "loss": 0.0002, + "step": 8960 + }, + { + "epoch": 4.2592592592592595, + "grad_norm": 0.03685637143330534, + "learning_rate": 6.546227899083062e-06, + "loss": 0.0003, + "step": 8970 + }, + { + "epoch": 4.264007597340931, + "grad_norm": 0.03680152447883992, + "learning_rate": 6.464474524359243e-06, + "loss": 0.0003, + "step": 8980 + }, + { + "epoch": 4.268755935422602, + "grad_norm": 0.01586860794779499, + "learning_rate": 6.3831995608783006e-06, + "loss": 0.0003, + "step": 8990 + }, + { + "epoch": 4.273504273504273, + "grad_norm": 0.060457370074189505, + "learning_rate": 6.30240390176961e-06, + "loss": 0.0003, + "step": 9000 + }, + { + "epoch": 4.278252611585945, + "grad_norm": 0.011848087514089329, + "learning_rate": 6.222088434895462e-06, + "loss": 0.0002, + "step": 9010 + }, + { + "epoch": 4.283000949667616, + "grad_norm": 0.007588968172980183, + "learning_rate": 6.142254042841317e-06, + "loss": 0.0002, + "step": 9020 + }, + { + "epoch": 4.287749287749287, + "grad_norm": 0.018813020734882144, + "learning_rate": 6.062901602906146e-06, + "loss": 0.0002, + "step": 9030 + }, + { + "epoch": 4.2924976258309595, + "grad_norm": 0.035822826381899694, + "learning_rate": 5.9840319870927085e-06, + "loss": 0.0003, + "step": 9040 + }, + { + "epoch": 4.297245963912631, + "grad_norm": 0.01542912309457929, + "learning_rate": 5.905646062098069e-06, + "loss": 0.0002, + "step": 9050 + }, + { + "epoch": 4.301994301994302, + "grad_norm": 0.033103480333157506, + "learning_rate": 5.827744689303965e-06, + "loss": 0.0003, + "step": 9060 + }, + { + "epoch": 4.306742640075973, + "grad_norm": 0.009750435837995845, + "learning_rate": 5.750328724767451e-06, + "loss": 0.0002, + "step": 9070 + }, + { + "epoch": 4.311490978157645, + "grad_norm": 0.02598801767575912, + "learning_rate": 5.6733990192113895e-06, + "loss": 0.0002, + "step": 9080 + }, + { + "epoch": 4.316239316239316, + "grad_norm": 0.010245573687772128, + "learning_rate": 5.5969564180151865e-06, + "loss": 0.0003, + "step": 9090 + }, + { + "epoch": 4.320987654320987, + "grad_norm": 0.010359365612193123, + "learning_rate": 5.521001761205441e-06, + "loss": 0.0002, + "step": 9100 + }, + { + "epoch": 4.3257359924026595, + "grad_norm": 0.019016290455767977, + "learning_rate": 5.445535883446767e-06, + "loss": 0.0003, + "step": 9110 + }, + { + "epoch": 4.330484330484331, + "grad_norm": 0.030832535746035557, + "learning_rate": 5.370559614032561e-06, + "loss": 0.0002, + "step": 9120 + }, + { + "epoch": 4.335232668566002, + "grad_norm": 0.024069560114827533, + "learning_rate": 5.296073776875921e-06, + "loss": 0.0002, + "step": 9130 + }, + { + "epoch": 4.339981006647673, + "grad_norm": 0.008753911885530482, + "learning_rate": 5.222079190500634e-06, + "loss": 0.0002, + "step": 9140 + }, + { + "epoch": 4.344729344729345, + "grad_norm": 0.014515827604061322, + "learning_rate": 5.1485766680320845e-06, + "loss": 0.0003, + "step": 9150 + }, + { + "epoch": 4.349477682811016, + "grad_norm": 0.025062559122919725, + "learning_rate": 5.075567017188421e-06, + "loss": 0.0003, + "step": 9160 + }, + { + "epoch": 4.354226020892687, + "grad_norm": 0.059685867782423484, + "learning_rate": 5.003051040271589e-06, + "loss": 0.0003, + "step": 9170 + }, + { + "epoch": 4.358974358974359, + "grad_norm": 0.020195856979814412, + "learning_rate": 4.93102953415861e-06, + "loss": 0.0002, + "step": 9180 + }, + { + "epoch": 4.363722697056031, + "grad_norm": 0.031553444686474724, + "learning_rate": 4.8595032902927216e-06, + "loss": 0.0002, + "step": 9190 + }, + { + "epoch": 4.368471035137702, + "grad_norm": 0.026737432366209037, + "learning_rate": 4.788473094674762e-06, + "loss": 0.0002, + "step": 9200 + }, + { + "epoch": 4.373219373219373, + "grad_norm": 0.023495736950044713, + "learning_rate": 4.717939727854515e-06, + "loss": 0.0002, + "step": 9210 + }, + { + "epoch": 4.377967711301045, + "grad_norm": 0.014474976780606165, + "learning_rate": 4.647903964922068e-06, + "loss": 0.0003, + "step": 9220 + }, + { + "epoch": 4.382716049382716, + "grad_norm": 0.0143353517875932, + "learning_rate": 4.57836657549941e-06, + "loss": 0.0002, + "step": 9230 + }, + { + "epoch": 4.387464387464387, + "grad_norm": 0.017107363736079413, + "learning_rate": 4.509328323731854e-06, + "loss": 0.0002, + "step": 9240 + }, + { + "epoch": 4.392212725546059, + "grad_norm": 0.017791461641481883, + "learning_rate": 4.440789968279746e-06, + "loss": 0.0002, + "step": 9250 + }, + { + "epoch": 4.39696106362773, + "grad_norm": 0.013041598381462837, + "learning_rate": 4.372752262310037e-06, + "loss": 0.0003, + "step": 9260 + }, + { + "epoch": 4.401709401709402, + "grad_norm": 0.02722415847361217, + "learning_rate": 4.30521595348809e-06, + "loss": 0.0002, + "step": 9270 + }, + { + "epoch": 4.406457739791073, + "grad_norm": 0.008822425625742323, + "learning_rate": 4.238181783969375e-06, + "loss": 0.0002, + "step": 9280 + }, + { + "epoch": 4.411206077872745, + "grad_norm": 0.016963293535349793, + "learning_rate": 4.171650490391405e-06, + "loss": 0.0002, + "step": 9290 + }, + { + "epoch": 4.415954415954416, + "grad_norm": 0.02880574520644374, + "learning_rate": 4.105622803865561e-06, + "loss": 0.0002, + "step": 9300 + }, + { + "epoch": 4.420702754036087, + "grad_norm": 0.027639477335558993, + "learning_rate": 4.040099449969126e-06, + "loss": 0.0002, + "step": 9310 + }, + { + "epoch": 4.4254510921177586, + "grad_norm": 0.016093107314697267, + "learning_rate": 3.975081148737253e-06, + "loss": 0.0002, + "step": 9320 + }, + { + "epoch": 4.43019943019943, + "grad_norm": 0.016704570318716194, + "learning_rate": 3.910568614655086e-06, + "loss": 0.0002, + "step": 9330 + }, + { + "epoch": 4.434947768281102, + "grad_norm": 0.012897201801140361, + "learning_rate": 3.846562556649913e-06, + "loss": 0.0002, + "step": 9340 + }, + { + "epoch": 4.439696106362773, + "grad_norm": 0.08980720694714919, + "learning_rate": 3.7830636780833375e-06, + "loss": 0.0002, + "step": 9350 + }, + { + "epoch": 4.444444444444445, + "grad_norm": 0.020088206692188224, + "learning_rate": 3.720072676743619e-06, + "loss": 0.0002, + "step": 9360 + }, + { + "epoch": 4.449192782526116, + "grad_norm": 0.013251103220357021, + "learning_rate": 3.6575902448379105e-06, + "loss": 0.0003, + "step": 9370 + }, + { + "epoch": 4.453941120607787, + "grad_norm": 0.022298951983921615, + "learning_rate": 3.5956170689847525e-06, + "loss": 0.0002, + "step": 9380 + }, + { + "epoch": 4.4586894586894585, + "grad_norm": 0.02629286796195216, + "learning_rate": 3.534153830206427e-06, + "loss": 0.0002, + "step": 9390 + }, + { + "epoch": 4.46343779677113, + "grad_norm": 0.027996970494385005, + "learning_rate": 3.4732012039215776e-06, + "loss": 0.0002, + "step": 9400 + }, + { + "epoch": 4.468186134852801, + "grad_norm": 0.014264164084167388, + "learning_rate": 3.4127598599376866e-06, + "loss": 0.0002, + "step": 9410 + }, + { + "epoch": 4.472934472934473, + "grad_norm": 0.021370039646477674, + "learning_rate": 3.352830462443807e-06, + "loss": 0.0002, + "step": 9420 + }, + { + "epoch": 4.477682811016145, + "grad_norm": 0.02060972457503309, + "learning_rate": 3.2934136700031814e-06, + "loss": 0.0002, + "step": 9430 + }, + { + "epoch": 4.482431149097816, + "grad_norm": 0.010267304435386022, + "learning_rate": 3.2345101355460704e-06, + "loss": 0.0003, + "step": 9440 + }, + { + "epoch": 4.487179487179487, + "grad_norm": 0.011859580830432183, + "learning_rate": 3.176120506362529e-06, + "loss": 0.0002, + "step": 9450 + }, + { + "epoch": 4.4919278252611585, + "grad_norm": 0.010980090286683744, + "learning_rate": 3.1182454240953286e-06, + "loss": 0.0002, + "step": 9460 + }, + { + "epoch": 4.49667616334283, + "grad_norm": 0.01569798750088808, + "learning_rate": 3.060885524732904e-06, + "loss": 0.0002, + "step": 9470 + }, + { + "epoch": 4.501424501424501, + "grad_norm": 0.013675067719950541, + "learning_rate": 3.0040414386023117e-06, + "loss": 0.0002, + "step": 9480 + }, + { + "epoch": 4.506172839506172, + "grad_norm": 0.010971237007864829, + "learning_rate": 2.9477137903623887e-06, + "loss": 0.0003, + "step": 9490 + }, + { + "epoch": 4.510921177587845, + "grad_norm": 0.01367105586360729, + "learning_rate": 2.891903198996815e-06, + "loss": 0.0002, + "step": 9500 + }, + { + "epoch": 4.515669515669516, + "grad_norm": 0.02249242502261443, + "learning_rate": 2.8366102778073554e-06, + "loss": 0.0002, + "step": 9510 + }, + { + "epoch": 4.520417853751187, + "grad_norm": 0.027430127096734522, + "learning_rate": 2.7818356344070985e-06, + "loss": 0.0002, + "step": 9520 + }, + { + "epoch": 4.5251661918328585, + "grad_norm": 0.03501020191825407, + "learning_rate": 2.7275798707137866e-06, + "loss": 0.0003, + "step": 9530 + }, + { + "epoch": 4.52991452991453, + "grad_norm": 0.00797541892421168, + "learning_rate": 2.6738435829432085e-06, + "loss": 0.0002, + "step": 9540 + }, + { + "epoch": 4.534662867996201, + "grad_norm": 0.020727695684549737, + "learning_rate": 2.6206273616026123e-06, + "loss": 0.0002, + "step": 9550 + }, + { + "epoch": 4.539411206077872, + "grad_norm": 0.013624996723948667, + "learning_rate": 2.567931791484285e-06, + "loss": 0.0002, + "step": 9560 + }, + { + "epoch": 4.544159544159545, + "grad_norm": 0.04444446785040936, + "learning_rate": 2.515757451659051e-06, + "loss": 0.0003, + "step": 9570 + }, + { + "epoch": 4.548907882241216, + "grad_norm": 0.0275204657563321, + "learning_rate": 2.464104915469967e-06, + "loss": 0.0002, + "step": 9580 + }, + { + "epoch": 4.553656220322887, + "grad_norm": 0.027438021904890947, + "learning_rate": 2.4129747505259703e-06, + "loss": 0.0002, + "step": 9590 + }, + { + "epoch": 4.5584045584045585, + "grad_norm": 0.018070469572468727, + "learning_rate": 2.3623675186957073e-06, + "loss": 0.0002, + "step": 9600 + }, + { + "epoch": 4.56315289648623, + "grad_norm": 0.02981502249381937, + "learning_rate": 2.3122837761012707e-06, + "loss": 0.0002, + "step": 9610 + }, + { + "epoch": 4.567901234567901, + "grad_norm": 0.011427482603589535, + "learning_rate": 2.2627240731121822e-06, + "loss": 0.0002, + "step": 9620 + }, + { + "epoch": 4.572649572649572, + "grad_norm": 0.02753939617267902, + "learning_rate": 2.2136889543392715e-06, + "loss": 0.0002, + "step": 9630 + }, + { + "epoch": 4.577397910731244, + "grad_norm": 0.03557992034872714, + "learning_rate": 2.1651789586287442e-06, + "loss": 0.0002, + "step": 9640 + }, + { + "epoch": 4.582146248812915, + "grad_norm": 0.01593025040715047, + "learning_rate": 2.117194619056212e-06, + "loss": 0.0003, + "step": 9650 + }, + { + "epoch": 4.586894586894587, + "grad_norm": 0.01883961843434692, + "learning_rate": 2.0697364629208626e-06, + "loss": 0.0002, + "step": 9660 + }, + { + "epoch": 4.5916429249762585, + "grad_norm": 0.03541763353826235, + "learning_rate": 2.0228050117396813e-06, + "loss": 0.0002, + "step": 9670 + }, + { + "epoch": 4.59639126305793, + "grad_norm": 0.0132500542000266, + "learning_rate": 1.976400781241694e-06, + "loss": 0.0002, + "step": 9680 + }, + { + "epoch": 4.601139601139601, + "grad_norm": 0.04348183725202648, + "learning_rate": 1.930524281362278e-06, + "loss": 0.0002, + "step": 9690 + }, + { + "epoch": 4.605887939221272, + "grad_norm": 0.00885631997359244, + "learning_rate": 1.8851760162376275e-06, + "loss": 0.0002, + "step": 9700 + }, + { + "epoch": 4.610636277302944, + "grad_norm": 0.022222594213102916, + "learning_rate": 1.8403564841991515e-06, + "loss": 0.0002, + "step": 9710 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.018684651562609175, + "learning_rate": 1.7960661777680188e-06, + "loss": 0.0002, + "step": 9720 + }, + { + "epoch": 4.620132953466287, + "grad_norm": 0.008378334028475045, + "learning_rate": 1.752305583649755e-06, + "loss": 0.0002, + "step": 9730 + }, + { + "epoch": 4.6248812915479585, + "grad_norm": 0.014433357271074476, + "learning_rate": 1.7090751827288643e-06, + "loss": 0.0002, + "step": 9740 + }, + { + "epoch": 4.62962962962963, + "grad_norm": 0.015032650170829363, + "learning_rate": 1.6663754500635954e-06, + "loss": 0.0002, + "step": 9750 + }, + { + "epoch": 4.634377967711301, + "grad_norm": 0.010642472856450867, + "learning_rate": 1.6242068548806777e-06, + "loss": 0.0002, + "step": 9760 + }, + { + "epoch": 4.639126305792972, + "grad_norm": 0.01071299691958736, + "learning_rate": 1.58256986057016e-06, + "loss": 0.0002, + "step": 9770 + }, + { + "epoch": 4.643874643874644, + "grad_norm": 0.010687582423069143, + "learning_rate": 1.5414649246803693e-06, + "loss": 0.0002, + "step": 9780 + }, + { + "epoch": 4.648622981956315, + "grad_norm": 0.011277484569791184, + "learning_rate": 1.5008924989128258e-06, + "loss": 0.0002, + "step": 9790 + }, + { + "epoch": 4.653371320037987, + "grad_norm": 0.009980400030097775, + "learning_rate": 1.4608530291173317e-06, + "loss": 0.0002, + "step": 9800 + }, + { + "epoch": 4.6581196581196584, + "grad_norm": 0.019618210962098266, + "learning_rate": 1.4213469552870006e-06, + "loss": 0.0002, + "step": 9810 + }, + { + "epoch": 4.66286799620133, + "grad_norm": 0.017812284703304353, + "learning_rate": 1.3823747115535135e-06, + "loss": 0.0002, + "step": 9820 + }, + { + "epoch": 4.667616334283001, + "grad_norm": 0.015089308138744192, + "learning_rate": 1.3439367261822599e-06, + "loss": 0.0002, + "step": 9830 + }, + { + "epoch": 4.672364672364672, + "grad_norm": 0.014709018447468676, + "learning_rate": 1.306033421567704e-06, + "loss": 0.0002, + "step": 9840 + }, + { + "epoch": 4.677113010446344, + "grad_norm": 0.009333902325461128, + "learning_rate": 1.2686652142286926e-06, + "loss": 0.0002, + "step": 9850 + }, + { + "epoch": 4.681861348528015, + "grad_norm": 0.008080473900328485, + "learning_rate": 1.2318325148038989e-06, + "loss": 0.0002, + "step": 9860 + }, + { + "epoch": 4.686609686609686, + "grad_norm": 0.011709176849387054, + "learning_rate": 1.1955357280473145e-06, + "loss": 0.0002, + "step": 9870 + }, + { + "epoch": 4.6913580246913575, + "grad_norm": 0.021540201796061084, + "learning_rate": 1.159775252823786e-06, + "loss": 0.0002, + "step": 9880 + }, + { + "epoch": 4.69610636277303, + "grad_norm": 0.009784332049459889, + "learning_rate": 1.1245514821046467e-06, + "loss": 0.0002, + "step": 9890 + }, + { + "epoch": 4.700854700854701, + "grad_norm": 0.020039584527672457, + "learning_rate": 1.0898648029633807e-06, + "loss": 0.0002, + "step": 9900 + }, + { + "epoch": 4.705603038936372, + "grad_norm": 0.01910268669882035, + "learning_rate": 1.055715596571405e-06, + "loss": 0.0002, + "step": 9910 + }, + { + "epoch": 4.710351377018044, + "grad_norm": 0.01418437471156699, + "learning_rate": 1.0221042381938327e-06, + "loss": 0.0002, + "step": 9920 + }, + { + "epoch": 4.715099715099715, + "grad_norm": 0.013536636487220366, + "learning_rate": 9.890310971853833e-07, + "loss": 0.0002, + "step": 9930 + }, + { + "epoch": 4.719848053181386, + "grad_norm": 0.01235875639613853, + "learning_rate": 9.56496536986312e-07, + "loss": 0.0002, + "step": 9940 + }, + { + "epoch": 4.7245963912630575, + "grad_norm": 0.00986734177801895, + "learning_rate": 9.245009151184203e-07, + "loss": 0.0002, + "step": 9950 + }, + { + "epoch": 4.72934472934473, + "grad_norm": 0.006747972213583124, + "learning_rate": 8.930445831811129e-07, + "loss": 0.0002, + "step": 9960 + }, + { + "epoch": 4.734093067426401, + "grad_norm": 0.029047233370321054, + "learning_rate": 8.62127886847569e-07, + "loss": 0.0002, + "step": 9970 + }, + { + "epoch": 4.738841405508072, + "grad_norm": 0.01953202648972791, + "learning_rate": 8.317511658609e-07, + "loss": 0.0002, + "step": 9980 + }, + { + "epoch": 4.743589743589744, + "grad_norm": 0.009998466801559164, + "learning_rate": 8.019147540304306e-07, + "loss": 0.0002, + "step": 9990 + }, + { + "epoch": 4.748338081671415, + "grad_norm": 0.014731999690461248, + "learning_rate": 7.726189792280625e-07, + "loss": 0.0002, + "step": 10000 + }, + { + "epoch": 4.753086419753086, + "grad_norm": 0.008672989147723817, + "learning_rate": 7.438641633846166e-07, + "loss": 0.0002, + "step": 10010 + }, + { + "epoch": 4.7578347578347575, + "grad_norm": 0.0194420972694553, + "learning_rate": 7.156506224863468e-07, + "loss": 0.0002, + "step": 10020 + }, + { + "epoch": 4.76258309591643, + "grad_norm": 0.010910630811079128, + "learning_rate": 6.879786665714316e-07, + "loss": 0.0002, + "step": 10030 + }, + { + "epoch": 4.767331433998101, + "grad_norm": 0.013212355777722541, + "learning_rate": 6.608485997265878e-07, + "loss": 0.0002, + "step": 10040 + }, + { + "epoch": 4.772079772079772, + "grad_norm": 0.03035738750880529, + "learning_rate": 6.34260720083707e-07, + "loss": 0.0002, + "step": 10050 + }, + { + "epoch": 4.776828110161444, + "grad_norm": 0.02182435274739629, + "learning_rate": 6.08215319816613e-07, + "loss": 0.0002, + "step": 10060 + }, + { + "epoch": 4.781576448243115, + "grad_norm": 0.0074833918644936215, + "learning_rate": 5.827126851378206e-07, + "loss": 0.0002, + "step": 10070 + }, + { + "epoch": 4.786324786324786, + "grad_norm": 0.010652029342376508, + "learning_rate": 5.577530962953992e-07, + "loss": 0.0002, + "step": 10080 + }, + { + "epoch": 4.7910731244064575, + "grad_norm": 0.014074178304652957, + "learning_rate": 5.33336827569908e-07, + "loss": 0.0002, + "step": 10090 + }, + { + "epoch": 4.795821462488129, + "grad_norm": 0.01752597070506098, + "learning_rate": 5.09464147271349e-07, + "loss": 0.0002, + "step": 10100 + }, + { + "epoch": 4.8005698005698, + "grad_norm": 0.04022585086609227, + "learning_rate": 4.861353177362638e-07, + "loss": 0.0002, + "step": 10110 + }, + { + "epoch": 4.805318138651472, + "grad_norm": 0.028673711234783343, + "learning_rate": 4.633505953248074e-07, + "loss": 0.0002, + "step": 10120 + }, + { + "epoch": 4.810066476733144, + "grad_norm": 0.0650520958985828, + "learning_rate": 4.411102304179571e-07, + "loss": 0.0002, + "step": 10130 + }, + { + "epoch": 4.814814814814815, + "grad_norm": 0.02142598250639681, + "learning_rate": 4.1941446741475267e-07, + "loss": 0.0002, + "step": 10140 + }, + { + "epoch": 4.819563152896486, + "grad_norm": 0.016171956740779283, + "learning_rate": 3.9826354472961016e-07, + "loss": 0.0002, + "step": 10150 + }, + { + "epoch": 4.8243114909781575, + "grad_norm": 0.006940652998893257, + "learning_rate": 3.7765769478970704e-07, + "loss": 0.0002, + "step": 10160 + }, + { + "epoch": 4.829059829059829, + "grad_norm": 0.01613379663810309, + "learning_rate": 3.575971440324233e-07, + "loss": 0.0002, + "step": 10170 + }, + { + "epoch": 4.8338081671415, + "grad_norm": 0.01195875519171062, + "learning_rate": 3.380821129028489e-07, + "loss": 0.0002, + "step": 10180 + }, + { + "epoch": 4.838556505223172, + "grad_norm": 0.020924587775578173, + "learning_rate": 3.191128158513745e-07, + "loss": 0.0002, + "step": 10190 + }, + { + "epoch": 4.843304843304844, + "grad_norm": 0.016413886410192223, + "learning_rate": 3.0068946133132713e-07, + "loss": 0.0002, + "step": 10200 + }, + { + "epoch": 4.848053181386515, + "grad_norm": 0.013949870569046934, + "learning_rate": 2.828122517966658e-07, + "loss": 0.0002, + "step": 10210 + }, + { + "epoch": 4.852801519468186, + "grad_norm": 0.015525538086731747, + "learning_rate": 2.6548138369978935e-07, + "loss": 0.0002, + "step": 10220 + }, + { + "epoch": 4.8575498575498575, + "grad_norm": 0.010709827684785575, + "learning_rate": 2.4869704748934907e-07, + "loss": 0.0003, + "step": 10230 + }, + { + "epoch": 4.862298195631529, + "grad_norm": 0.011124039185672178, + "learning_rate": 2.3245942760816152e-07, + "loss": 0.0002, + "step": 10240 + }, + { + "epoch": 4.8670465337132, + "grad_norm": 0.06196431047249913, + "learning_rate": 2.167687024911935e-07, + "loss": 0.0002, + "step": 10250 + }, + { + "epoch": 4.871794871794872, + "grad_norm": 0.03348946247489169, + "learning_rate": 2.0162504456359143e-07, + "loss": 0.0002, + "step": 10260 + }, + { + "epoch": 4.8765432098765435, + "grad_norm": 0.033066730239417634, + "learning_rate": 1.8702862023878277e-07, + "loss": 0.0002, + "step": 10270 + }, + { + "epoch": 4.881291547958215, + "grad_norm": 0.011699311189894486, + "learning_rate": 1.7297958991666085e-07, + "loss": 0.0002, + "step": 10280 + }, + { + "epoch": 4.886039886039886, + "grad_norm": 0.009826311343891937, + "learning_rate": 1.5947810798181416e-07, + "loss": 0.0002, + "step": 10290 + }, + { + "epoch": 4.8907882241215574, + "grad_norm": 0.008330339709963382, + "learning_rate": 1.4652432280182203e-07, + "loss": 0.0002, + "step": 10300 + }, + { + "epoch": 4.895536562203229, + "grad_norm": 0.023800076929515735, + "learning_rate": 1.3411837672564487e-07, + "loss": 0.0002, + "step": 10310 + }, + { + "epoch": 4.9002849002849, + "grad_norm": 0.02878005848707081, + "learning_rate": 1.2226040608202537e-07, + "loss": 0.0002, + "step": 10320 + }, + { + "epoch": 4.905033238366571, + "grad_norm": 0.01653213211869477, + "learning_rate": 1.1095054117804537e-07, + "loss": 0.0002, + "step": 10330 + }, + { + "epoch": 4.909781576448243, + "grad_norm": 0.009200145640811685, + "learning_rate": 1.0018890629762689e-07, + "loss": 0.0002, + "step": 10340 + }, + { + "epoch": 4.914529914529915, + "grad_norm": 0.02119686414836386, + "learning_rate": 8.997561970022217e-08, + "loss": 0.0002, + "step": 10350 + }, + { + "epoch": 4.919278252611586, + "grad_norm": 0.01990993539672601, + "learning_rate": 8.03107936194869e-08, + "loss": 0.0002, + "step": 10360 + }, + { + "epoch": 4.924026590693257, + "grad_norm": 0.015415964718939136, + "learning_rate": 7.119453426205347e-08, + "loss": 0.0002, + "step": 10370 + }, + { + "epoch": 4.928774928774929, + "grad_norm": 0.0251551283710964, + "learning_rate": 6.262694180634853e-08, + "loss": 0.0002, + "step": 10380 + }, + { + "epoch": 4.9335232668566, + "grad_norm": 0.014734088435722246, + "learning_rate": 5.4608110401532796e-08, + "loss": 0.0002, + "step": 10390 + }, + { + "epoch": 4.938271604938271, + "grad_norm": 0.015254406988241554, + "learning_rate": 4.713812816641849e-08, + "loss": 0.0002, + "step": 10400 + }, + { + "epoch": 4.943019943019943, + "grad_norm": 0.009692878353320089, + "learning_rate": 4.0217077188536845e-08, + "loss": 0.0002, + "step": 10410 + }, + { + "epoch": 4.947768281101615, + "grad_norm": 0.027159316174024378, + "learning_rate": 3.38450335231999e-08, + "loss": 0.0002, + "step": 10420 + }, + { + "epoch": 4.952516619183286, + "grad_norm": 0.010723828683855324, + "learning_rate": 2.8022067192712277e-08, + "loss": 0.0002, + "step": 10430 + }, + { + "epoch": 4.957264957264957, + "grad_norm": 0.01243713834746454, + "learning_rate": 2.274824218556071e-08, + "loss": 0.0002, + "step": 10440 + }, + { + "epoch": 4.962013295346629, + "grad_norm": 0.022527727503970245, + "learning_rate": 1.802361645573125e-08, + "loss": 0.0002, + "step": 10450 + }, + { + "epoch": 4.9667616334283, + "grad_norm": 0.017443653758961303, + "learning_rate": 1.3848241922059802e-08, + "loss": 0.0002, + "step": 10460 + }, + { + "epoch": 4.971509971509971, + "grad_norm": 0.01937976501458018, + "learning_rate": 1.0222164467688089e-08, + "loss": 0.0002, + "step": 10470 + }, + { + "epoch": 4.976258309591643, + "grad_norm": 0.02202509619611748, + "learning_rate": 7.1454239395141175e-09, + "loss": 0.0002, + "step": 10480 + }, + { + "epoch": 4.981006647673315, + "grad_norm": 0.02238063953227196, + "learning_rate": 4.618054147803585e-09, + "loss": 0.0002, + "step": 10490 + }, + { + "epoch": 4.985754985754986, + "grad_norm": 0.026393716136063707, + "learning_rate": 2.6400828657791033e-09, + "loss": 0.0002, + "step": 10500 + }, + { + "epoch": 4.990503323836657, + "grad_norm": 0.011804554528937828, + "learning_rate": 1.211531829337087e-09, + "loss": 0.0002, + "step": 10510 + }, + { + "epoch": 4.995251661918329, + "grad_norm": 0.012682770872747743, + "learning_rate": 3.324167368090603e-10, + "loss": 0.0002, + "step": 10520 + }, + { + "epoch": 5.0, + "grad_norm": 0.012742190367527018, + "learning_rate": 2.747248778467437e-12, + "loss": 0.0002, + "step": 10530 + }, + { + "epoch": 5.0, + "step": 10530, + "total_flos": 663559922515968.0, + "train_loss": 0.14455950133211057, + "train_runtime": 42932.9049, + "train_samples_per_second": 125.568, + "train_steps_per_second": 0.245 + } + ], + "logging_steps": 10, + "max_steps": 10530, + "num_input_tokens_seen": 0, + "num_train_epochs": 5, + "save_steps": 1800, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 663559922515968.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +}