| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9998528329654155, | |
| "eval_steps": 500, | |
| "global_step": 7644, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003924454255580083, | |
| "grad_norm": 1.59566508589272, | |
| "learning_rate": 8.695652173913044e-06, | |
| "loss": 1.7431, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007848908511160166, | |
| "grad_norm": 1.9218722741713965, | |
| "learning_rate": 1.739130434782609e-05, | |
| "loss": 1.8901, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01177336276674025, | |
| "grad_norm": 1.017626221414331, | |
| "learning_rate": 2.608695652173913e-05, | |
| "loss": 1.5468, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.015697817022320332, | |
| "grad_norm": 1.215416694846421, | |
| "learning_rate": 3.478260869565218e-05, | |
| "loss": 1.6517, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.019622271277900418, | |
| "grad_norm": 4.053774614027623, | |
| "learning_rate": 4.347826086956522e-05, | |
| "loss": 1.5986, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0235467255334805, | |
| "grad_norm": 0.9611699733801534, | |
| "learning_rate": 5.217391304347826e-05, | |
| "loss": 1.4807, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.027471179789060583, | |
| "grad_norm": 1.508151250079747, | |
| "learning_rate": 6.086956521739131e-05, | |
| "loss": 1.6647, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.031395634044640665, | |
| "grad_norm": 0.8236871461279531, | |
| "learning_rate": 6.956521739130436e-05, | |
| "loss": 1.3701, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03532008830022075, | |
| "grad_norm": 1.0747931175792689, | |
| "learning_rate": 7.82608695652174e-05, | |
| "loss": 1.6309, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.039244542555800836, | |
| "grad_norm": 2.558414240808255, | |
| "learning_rate": 8.695652173913044e-05, | |
| "loss": 1.5057, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.043168996811380915, | |
| "grad_norm": 0.8766943809051262, | |
| "learning_rate": 9.565217391304348e-05, | |
| "loss": 1.571, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.047093451066961, | |
| "grad_norm": 1.3065970989255622, | |
| "learning_rate": 0.00010434782608695653, | |
| "loss": 1.629, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.051017905322541086, | |
| "grad_norm": 0.7682147702990969, | |
| "learning_rate": 0.00011304347826086956, | |
| "loss": 1.4162, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.054942359578121165, | |
| "grad_norm": 1.0400539633177819, | |
| "learning_rate": 0.00012173913043478263, | |
| "loss": 1.6675, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05886681383370125, | |
| "grad_norm": 2.184120628414497, | |
| "learning_rate": 0.00013043478260869567, | |
| "loss": 1.4811, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06279126808928133, | |
| "grad_norm": 0.8493857009549972, | |
| "learning_rate": 0.0001391304347826087, | |
| "loss": 1.6556, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06671572234486142, | |
| "grad_norm": 1.306334292232353, | |
| "learning_rate": 0.00014782608695652173, | |
| "loss": 1.6917, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0706401766004415, | |
| "grad_norm": 0.7818602938197371, | |
| "learning_rate": 0.0001565217391304348, | |
| "loss": 1.5013, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.07456463085602158, | |
| "grad_norm": 0.9014048393310743, | |
| "learning_rate": 0.00016521739130434784, | |
| "loss": 1.7288, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07848908511160167, | |
| "grad_norm": 2.145130580339822, | |
| "learning_rate": 0.00017391304347826088, | |
| "loss": 1.5501, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08241353936718175, | |
| "grad_norm": 0.80854236230725, | |
| "learning_rate": 0.00018260869565217392, | |
| "loss": 1.7233, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08633799362276183, | |
| "grad_norm": 1.2005909273061688, | |
| "learning_rate": 0.00019130434782608697, | |
| "loss": 1.7155, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.09026244787834192, | |
| "grad_norm": 0.779032139448437, | |
| "learning_rate": 0.0002, | |
| "loss": 1.5328, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.094186902133922, | |
| "grad_norm": 0.895984457831143, | |
| "learning_rate": 0.00019999910223238217, | |
| "loss": 1.8188, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.09811135638950208, | |
| "grad_norm": 1.8450636775788756, | |
| "learning_rate": 0.00019999640894564831, | |
| "loss": 1.5566, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.10203581064508217, | |
| "grad_norm": 0.7128469148594528, | |
| "learning_rate": 0.00019999192018815736, | |
| "loss": 1.7639, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.10596026490066225, | |
| "grad_norm": 1.1219978957787244, | |
| "learning_rate": 0.0001999856360405066, | |
| "loss": 1.7703, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.10988471915624233, | |
| "grad_norm": 0.6952941396730594, | |
| "learning_rate": 0.00019997755661553006, | |
| "loss": 1.5359, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.11380917341182242, | |
| "grad_norm": 0.8480327494911744, | |
| "learning_rate": 0.00019996768205829665, | |
| "loss": 1.7883, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.1177336276674025, | |
| "grad_norm": 1.6877738726930545, | |
| "learning_rate": 0.00019995601254610757, | |
| "loss": 1.5717, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.12165808192298258, | |
| "grad_norm": 0.6957803276899013, | |
| "learning_rate": 0.000199942548288493, | |
| "loss": 1.7741, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.12558253617856266, | |
| "grad_norm": 1.0703624063881845, | |
| "learning_rate": 0.0001999272895272084, | |
| "loss": 1.7479, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.12950699043414277, | |
| "grad_norm": 0.6703120010200306, | |
| "learning_rate": 0.0001999102365362303, | |
| "loss": 1.531, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.13343144468972284, | |
| "grad_norm": 0.7751374462026334, | |
| "learning_rate": 0.00019989138962175104, | |
| "loss": 1.8117, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.13735589894530292, | |
| "grad_norm": 1.389578665909138, | |
| "learning_rate": 0.00019987074912217369, | |
| "loss": 1.5659, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.141280353200883, | |
| "grad_norm": 0.6627080244627721, | |
| "learning_rate": 0.00019984831540810567, | |
| "loss": 1.7674, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.14520480745646308, | |
| "grad_norm": 1.0379002029446147, | |
| "learning_rate": 0.00019982408888235225, | |
| "loss": 1.7571, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.14912926171204316, | |
| "grad_norm": 0.5757907603895639, | |
| "learning_rate": 0.00019979806997990918, | |
| "loss": 1.501, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.15305371596762327, | |
| "grad_norm": 0.7320901283579814, | |
| "learning_rate": 0.00019977025916795502, | |
| "loss": 1.8069, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.15697817022320334, | |
| "grad_norm": 1.505755570211187, | |
| "learning_rate": 0.00019974065694584279, | |
| "loss": 1.5442, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.16090262447878342, | |
| "grad_norm": 0.6462386467537791, | |
| "learning_rate": 0.00019970926384509072, | |
| "loss": 1.7406, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1648270787343635, | |
| "grad_norm": 0.9881025653862135, | |
| "learning_rate": 0.00019967608042937302, | |
| "loss": 1.74, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.16875153298994358, | |
| "grad_norm": 0.5198351999565215, | |
| "learning_rate": 0.00019964110729450964, | |
| "loss": 1.4838, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.17267598724552366, | |
| "grad_norm": 0.6976776100585178, | |
| "learning_rate": 0.00019960434506845553, | |
| "loss": 1.8036, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.17660044150110377, | |
| "grad_norm": 1.3406000724813996, | |
| "learning_rate": 0.00019956579441128943, | |
| "loss": 1.5183, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.18052489575668385, | |
| "grad_norm": 0.6897390365275573, | |
| "learning_rate": 0.0001995254560152019, | |
| "loss": 1.7454, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.18444935001226392, | |
| "grad_norm": 0.9102386307226423, | |
| "learning_rate": 0.0001994833306044831, | |
| "loss": 1.7144, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.188373804267844, | |
| "grad_norm": 0.5439902552034376, | |
| "learning_rate": 0.00019943941893550967, | |
| "loss": 1.463, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.19229825852342408, | |
| "grad_norm": 0.6785386985985739, | |
| "learning_rate": 0.00019939372179673105, | |
| "loss": 1.825, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.19622271277900416, | |
| "grad_norm": 1.3124987116714535, | |
| "learning_rate": 0.00019934624000865543, | |
| "loss": 1.5003, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.20014716703458427, | |
| "grad_norm": 0.601716437865292, | |
| "learning_rate": 0.00019929697442383512, | |
| "loss": 1.7115, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.20407162129016435, | |
| "grad_norm": 0.8604098174693385, | |
| "learning_rate": 0.00019924592592685103, | |
| "loss": 1.7183, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.20799607554574442, | |
| "grad_norm": 0.5714207763143907, | |
| "learning_rate": 0.0001991930954342969, | |
| "loss": 1.448, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.2119205298013245, | |
| "grad_norm": 0.6843443496377163, | |
| "learning_rate": 0.00019913848389476285, | |
| "loss": 1.7846, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.21584498405690458, | |
| "grad_norm": 1.2172413322873215, | |
| "learning_rate": 0.00019908209228881827, | |
| "loss": 1.4919, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.21976943831248466, | |
| "grad_norm": 0.5687111213051049, | |
| "learning_rate": 0.0001990239216289944, | |
| "loss": 1.6978, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.22369389256806477, | |
| "grad_norm": 0.8672964829806664, | |
| "learning_rate": 0.00019896397295976584, | |
| "loss": 1.7201, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.22761834682364485, | |
| "grad_norm": 0.5628246824188602, | |
| "learning_rate": 0.00019890224735753215, | |
| "loss": 1.4317, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.23154280107922492, | |
| "grad_norm": 0.6390491703870059, | |
| "learning_rate": 0.00019883874593059825, | |
| "loss": 1.7333, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.235467255334805, | |
| "grad_norm": 1.314195280790082, | |
| "learning_rate": 0.0001987734698191546, | |
| "loss": 1.507, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.23939170959038508, | |
| "grad_norm": 0.6067163458181883, | |
| "learning_rate": 0.00019870642019525678, | |
| "loss": 1.6951, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.24331616384596516, | |
| "grad_norm": 0.8485342996987779, | |
| "learning_rate": 0.00019863759826280447, | |
| "loss": 1.7011, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.24724061810154527, | |
| "grad_norm": 0.49167624613659977, | |
| "learning_rate": 0.00019856700525751965, | |
| "loss": 1.4403, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.2511650723571253, | |
| "grad_norm": 0.6096757587397276, | |
| "learning_rate": 0.00019849464244692467, | |
| "loss": 1.7492, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2550895266127054, | |
| "grad_norm": 1.312860374415334, | |
| "learning_rate": 0.00019842051113031924, | |
| "loss": 1.4578, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.25901398086828553, | |
| "grad_norm": 0.5768345844397214, | |
| "learning_rate": 0.00019834461263875728, | |
| "loss": 1.7214, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2629384351238656, | |
| "grad_norm": 0.8721739477657561, | |
| "learning_rate": 0.00019826694833502295, | |
| "loss": 1.693, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.2668628893794457, | |
| "grad_norm": 0.46312917981419044, | |
| "learning_rate": 0.00019818751961360622, | |
| "loss": 1.4422, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.27078734363502577, | |
| "grad_norm": 0.6148502880249199, | |
| "learning_rate": 0.00019810632790067772, | |
| "loss": 1.7703, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.27471179789060585, | |
| "grad_norm": 1.2290842246551827, | |
| "learning_rate": 0.0001980233746540633, | |
| "loss": 1.5103, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.2786362521461859, | |
| "grad_norm": 0.5576071915309164, | |
| "learning_rate": 0.00019793866136321773, | |
| "loss": 1.694, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.282560706401766, | |
| "grad_norm": 0.8445183650392717, | |
| "learning_rate": 0.00019785218954919797, | |
| "loss": 1.6649, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.2864851606573461, | |
| "grad_norm": 0.48393246990590494, | |
| "learning_rate": 0.00019776396076463595, | |
| "loss": 1.4157, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.29040961491292616, | |
| "grad_norm": 0.6051599690454587, | |
| "learning_rate": 0.00019767397659371057, | |
| "loss": 1.7666, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.29433406916850624, | |
| "grad_norm": 1.1468440741020822, | |
| "learning_rate": 0.0001975822386521193, | |
| "loss": 1.5071, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.2982585234240863, | |
| "grad_norm": 0.5429139129648298, | |
| "learning_rate": 0.00019748874858704923, | |
| "loss": 1.6465, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.3021829776796664, | |
| "grad_norm": 0.8287655903162845, | |
| "learning_rate": 0.0001973935080771474, | |
| "loss": 1.6809, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.30610743193524653, | |
| "grad_norm": 0.4518816067543287, | |
| "learning_rate": 0.00019729651883249074, | |
| "loss": 1.4447, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.3100318861908266, | |
| "grad_norm": 0.6201255078052008, | |
| "learning_rate": 0.00019719778259455532, | |
| "loss": 1.736, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.3139563404464067, | |
| "grad_norm": 1.1784840531264669, | |
| "learning_rate": 0.00019709730113618508, | |
| "loss": 1.4831, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.31788079470198677, | |
| "grad_norm": 0.5371643182619428, | |
| "learning_rate": 0.00019699507626156, | |
| "loss": 1.6586, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.32180524895756685, | |
| "grad_norm": 0.7893259051605654, | |
| "learning_rate": 0.0001968911098061637, | |
| "loss": 1.6784, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3257297032131469, | |
| "grad_norm": 0.5247266380599999, | |
| "learning_rate": 0.0001967854036367506, | |
| "loss": 1.4233, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.329654157468727, | |
| "grad_norm": 0.6788903735157552, | |
| "learning_rate": 0.00019667795965131214, | |
| "loss": 1.7541, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3335786117243071, | |
| "grad_norm": 1.2469614542936243, | |
| "learning_rate": 0.00019656877977904298, | |
| "loss": 1.4887, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.33750306597988716, | |
| "grad_norm": 0.563418856759311, | |
| "learning_rate": 0.00019645786598030616, | |
| "loss": 1.6679, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.34142752023546724, | |
| "grad_norm": 0.7810287126487964, | |
| "learning_rate": 0.00019634522024659804, | |
| "loss": 1.6544, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.3453519744910473, | |
| "grad_norm": 0.4299715447918262, | |
| "learning_rate": 0.00019623084460051248, | |
| "loss": 1.3657, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.3492764287466274, | |
| "grad_norm": 0.6056642192549682, | |
| "learning_rate": 0.00019611474109570444, | |
| "loss": 1.7141, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.35320088300220753, | |
| "grad_norm": 1.2290595368609403, | |
| "learning_rate": 0.00019599691181685334, | |
| "loss": 1.4617, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3571253372577876, | |
| "grad_norm": 0.5881723670713188, | |
| "learning_rate": 0.00019587735887962533, | |
| "loss": 1.6506, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.3610497915133677, | |
| "grad_norm": 0.7896617082024844, | |
| "learning_rate": 0.00019575608443063558, | |
| "loss": 1.6519, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.36497424576894777, | |
| "grad_norm": 0.4217609244092629, | |
| "learning_rate": 0.00019563309064740954, | |
| "loss": 1.4036, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.36889870002452785, | |
| "grad_norm": 0.5961043232650607, | |
| "learning_rate": 0.0001955083797383439, | |
| "loss": 1.7273, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.3728231542801079, | |
| "grad_norm": 1.2028843643313036, | |
| "learning_rate": 0.000195381953942667, | |
| "loss": 1.4575, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.376747608535688, | |
| "grad_norm": 0.525632267831757, | |
| "learning_rate": 0.00019525381553039852, | |
| "loss": 1.6453, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.3806720627912681, | |
| "grad_norm": 0.7932640956213916, | |
| "learning_rate": 0.00019512396680230882, | |
| "loss": 1.6651, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.38459651704684816, | |
| "grad_norm": 0.5172893758056729, | |
| "learning_rate": 0.0001949924100898776, | |
| "loss": 1.4058, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.38852097130242824, | |
| "grad_norm": 0.6029288073427342, | |
| "learning_rate": 0.00019485914775525193, | |
| "loss": 1.715, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.3924454255580083, | |
| "grad_norm": 1.2117347595679093, | |
| "learning_rate": 0.000194724182191204, | |
| "loss": 1.4449, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3963698798135884, | |
| "grad_norm": 0.5304032050972481, | |
| "learning_rate": 0.0001945875158210881, | |
| "loss": 1.6277, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.40029433406916853, | |
| "grad_norm": 0.7927902226112815, | |
| "learning_rate": 0.00019444915109879704, | |
| "loss": 1.6564, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.4042187883247486, | |
| "grad_norm": 0.47363394556772503, | |
| "learning_rate": 0.00019430909050871815, | |
| "loss": 1.3962, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.4081432425803287, | |
| "grad_norm": 0.5696160467482098, | |
| "learning_rate": 0.00019416733656568868, | |
| "loss": 1.7328, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.41206769683590877, | |
| "grad_norm": 1.0648480652851804, | |
| "learning_rate": 0.00019402389181495062, | |
| "loss": 1.4136, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.41599215109148885, | |
| "grad_norm": 0.5494481753899966, | |
| "learning_rate": 0.00019387875883210507, | |
| "loss": 1.6364, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.4199166053470689, | |
| "grad_norm": 0.7954824302893352, | |
| "learning_rate": 0.00019373194022306576, | |
| "loss": 1.6001, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.423841059602649, | |
| "grad_norm": 0.46336376006101543, | |
| "learning_rate": 0.0001935834386240127, | |
| "loss": 1.4158, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.4277655138582291, | |
| "grad_norm": 0.6023768618137292, | |
| "learning_rate": 0.00019343325670134428, | |
| "loss": 1.7213, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.43168996811380916, | |
| "grad_norm": 1.0597541991933253, | |
| "learning_rate": 0.00019328139715162995, | |
| "loss": 1.4421, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.43561442236938924, | |
| "grad_norm": 0.5271067584885061, | |
| "learning_rate": 0.00019312786270156136, | |
| "loss": 1.6042, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.4395388766249693, | |
| "grad_norm": 0.7539995684837235, | |
| "learning_rate": 0.0001929726561079037, | |
| "loss": 1.6369, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.4434633308805494, | |
| "grad_norm": 0.4321890326688252, | |
| "learning_rate": 0.00019281578015744604, | |
| "loss": 1.3712, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.44738778513612953, | |
| "grad_norm": 0.5971842639069664, | |
| "learning_rate": 0.00019265723766695133, | |
| "loss": 1.6896, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.4513122393917096, | |
| "grad_norm": 1.0452914937951234, | |
| "learning_rate": 0.00019249703148310586, | |
| "loss": 1.4343, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.4552366936472897, | |
| "grad_norm": 0.5404851919835004, | |
| "learning_rate": 0.00019233516448246814, | |
| "loss": 1.652, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.45916114790286977, | |
| "grad_norm": 0.7651102413452125, | |
| "learning_rate": 0.00019217163957141716, | |
| "loss": 1.6139, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.46308560215844985, | |
| "grad_norm": 0.38127929097760105, | |
| "learning_rate": 0.00019200645968610035, | |
| "loss": 1.3615, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.46701005641402993, | |
| "grad_norm": 0.5692440260577405, | |
| "learning_rate": 0.0001918396277923807, | |
| "loss": 1.6775, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.47093451066961, | |
| "grad_norm": 1.0414629176757495, | |
| "learning_rate": 0.0001916711468857837, | |
| "loss": 1.4048, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.4748589649251901, | |
| "grad_norm": 0.5043453926125885, | |
| "learning_rate": 0.00019150101999144338, | |
| "loss": 1.6243, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.47878341918077016, | |
| "grad_norm": 0.7872249501646571, | |
| "learning_rate": 0.00019132925016404804, | |
| "loss": 1.6145, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.48270787343635024, | |
| "grad_norm": 0.39550838826464113, | |
| "learning_rate": 0.0001911558404877855, | |
| "loss": 1.3645, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.4866323276919303, | |
| "grad_norm": 0.5650717820779941, | |
| "learning_rate": 0.0001909807940762876, | |
| "loss": 1.6753, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.4905567819475104, | |
| "grad_norm": 1.1096228154114907, | |
| "learning_rate": 0.00019080411407257431, | |
| "loss": 1.3791, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.49448123620309054, | |
| "grad_norm": 0.5219136400722999, | |
| "learning_rate": 0.00019062580364899732, | |
| "loss": 1.6407, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.4984056904586706, | |
| "grad_norm": 0.7223624826857745, | |
| "learning_rate": 0.00019044586600718322, | |
| "loss": 1.6447, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.5023301447142506, | |
| "grad_norm": 0.42218040332652595, | |
| "learning_rate": 0.00019026430437797567, | |
| "loss": 1.3344, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.5062545989698307, | |
| "grad_norm": 0.564596747533548, | |
| "learning_rate": 0.00019008112202137777, | |
| "loss": 1.6916, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.5101790532254108, | |
| "grad_norm": 1.0693549554460788, | |
| "learning_rate": 0.0001898963222264932, | |
| "loss": 1.4193, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5141035074809909, | |
| "grad_norm": 0.4841799865635874, | |
| "learning_rate": 0.00018970990831146744, | |
| "loss": 1.5982, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.5180279617365711, | |
| "grad_norm": 0.7601244467039746, | |
| "learning_rate": 0.00018952188362342804, | |
| "loss": 1.6212, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.5219524159921511, | |
| "grad_norm": 0.4350972445998026, | |
| "learning_rate": 0.00018933225153842446, | |
| "loss": 1.3561, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.5258768702477312, | |
| "grad_norm": 0.5405853945062638, | |
| "learning_rate": 0.00018914101546136766, | |
| "loss": 1.6763, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.5298013245033113, | |
| "grad_norm": 1.0688201126520391, | |
| "learning_rate": 0.00018894817882596877, | |
| "loss": 1.4251, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.5337257787588914, | |
| "grad_norm": 0.4951679408076343, | |
| "learning_rate": 0.00018875374509467755, | |
| "loss": 1.628, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.5376502330144715, | |
| "grad_norm": 0.7557600408769889, | |
| "learning_rate": 0.00018855771775862012, | |
| "loss": 1.6191, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.5415746872700515, | |
| "grad_norm": 0.377073658784635, | |
| "learning_rate": 0.00018836010033753637, | |
| "loss": 1.3429, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.5454991415256316, | |
| "grad_norm": 0.5485259271202717, | |
| "learning_rate": 0.00018816089637971675, | |
| "loss": 1.6978, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.5494235957812117, | |
| "grad_norm": 1.0119983873800584, | |
| "learning_rate": 0.0001879601094619385, | |
| "loss": 1.3818, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.5533480500367918, | |
| "grad_norm": 0.46457826851278217, | |
| "learning_rate": 0.0001877577431894015, | |
| "loss": 1.5829, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.5572725042923719, | |
| "grad_norm": 0.6955742080101969, | |
| "learning_rate": 0.00018755380119566341, | |
| "loss": 1.6086, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.5611969585479519, | |
| "grad_norm": 0.4318272880121663, | |
| "learning_rate": 0.0001873482871425747, | |
| "loss": 1.3767, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.565121412803532, | |
| "grad_norm": 0.544326673848161, | |
| "learning_rate": 0.0001871412047202125, | |
| "loss": 1.655, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.5690458670591121, | |
| "grad_norm": 1.049074233448828, | |
| "learning_rate": 0.00018693255764681474, | |
| "loss": 1.4151, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.5729703213146922, | |
| "grad_norm": 0.5117822783483642, | |
| "learning_rate": 0.0001867223496687131, | |
| "loss": 1.6108, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.5768947755702722, | |
| "grad_norm": 0.7370386974091027, | |
| "learning_rate": 0.00018651058456026592, | |
| "loss": 1.5762, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.5808192298258523, | |
| "grad_norm": 0.39768741853151723, | |
| "learning_rate": 0.00018629726612379033, | |
| "loss": 1.3514, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.5847436840814324, | |
| "grad_norm": 0.5588937021023881, | |
| "learning_rate": 0.000186082398189494, | |
| "loss": 1.7102, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.5886681383370125, | |
| "grad_norm": 0.9862206946512853, | |
| "learning_rate": 0.00018586598461540644, | |
| "loss": 1.3736, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5925925925925926, | |
| "grad_norm": 0.47368208307233367, | |
| "learning_rate": 0.00018564802928730965, | |
| "loss": 1.6015, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.5965170468481726, | |
| "grad_norm": 0.7567658413720824, | |
| "learning_rate": 0.00018542853611866825, | |
| "loss": 1.6137, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.6004415011037527, | |
| "grad_norm": 0.3942603134454529, | |
| "learning_rate": 0.00018520750905055947, | |
| "loss": 1.3276, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.6043659553593328, | |
| "grad_norm": 0.5559856568089228, | |
| "learning_rate": 0.00018498495205160226, | |
| "loss": 1.6358, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.6082904096149129, | |
| "grad_norm": 0.9896614064086859, | |
| "learning_rate": 0.00018476086911788585, | |
| "loss": 1.3846, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.6122148638704931, | |
| "grad_norm": 0.47872164084060653, | |
| "learning_rate": 0.00018453526427289835, | |
| "loss": 1.5684, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.6161393181260731, | |
| "grad_norm": 0.6893394791051162, | |
| "learning_rate": 0.0001843081415674542, | |
| "loss": 1.5812, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.6200637723816532, | |
| "grad_norm": 0.3705155025830917, | |
| "learning_rate": 0.00018407950507962165, | |
| "loss": 1.3208, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.6239882266372333, | |
| "grad_norm": 0.5616643490216171, | |
| "learning_rate": 0.0001838493589146494, | |
| "loss": 1.6824, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.6279126808928134, | |
| "grad_norm": 1.031792835439678, | |
| "learning_rate": 0.00018361770720489286, | |
| "loss": 1.345, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.6318371351483935, | |
| "grad_norm": 0.471794875435998, | |
| "learning_rate": 0.0001833845541097402, | |
| "loss": 1.5665, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.6357615894039735, | |
| "grad_norm": 0.7296970931562926, | |
| "learning_rate": 0.00018314990381553727, | |
| "loss": 1.6028, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.6396860436595536, | |
| "grad_norm": 0.4020800963657848, | |
| "learning_rate": 0.00018291376053551292, | |
| "loss": 1.3523, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.6436104979151337, | |
| "grad_norm": 0.5566755820686294, | |
| "learning_rate": 0.0001826761285097029, | |
| "loss": 1.6999, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.6475349521707138, | |
| "grad_norm": 1.011677825927515, | |
| "learning_rate": 0.0001824370120048739, | |
| "loss": 1.3937, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.6514594064262939, | |
| "grad_norm": 0.48683829889667224, | |
| "learning_rate": 0.00018219641531444714, | |
| "loss": 1.5761, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.6553838606818739, | |
| "grad_norm": 0.6826355942276628, | |
| "learning_rate": 0.0001819543427584209, | |
| "loss": 1.5739, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.659308314937454, | |
| "grad_norm": 0.41583770045895846, | |
| "learning_rate": 0.0001817107986832932, | |
| "loss": 1.3445, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.6632327691930341, | |
| "grad_norm": 0.5164793678992897, | |
| "learning_rate": 0.00018146578746198376, | |
| "loss": 1.6682, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.6671572234486142, | |
| "grad_norm": 1.0096758161757577, | |
| "learning_rate": 0.00018121931349375538, | |
| "loss": 1.3767, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.6710816777041942, | |
| "grad_norm": 0.47287092245663637, | |
| "learning_rate": 0.000180971381204135, | |
| "loss": 1.577, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.6750061319597743, | |
| "grad_norm": 0.7228044254667811, | |
| "learning_rate": 0.00018072199504483427, | |
| "loss": 1.5365, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.6789305862153544, | |
| "grad_norm": 0.38003400406471977, | |
| "learning_rate": 0.00018047115949366952, | |
| "loss": 1.3279, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.6828550404709345, | |
| "grad_norm": 0.5254748001755114, | |
| "learning_rate": 0.00018021887905448147, | |
| "loss": 1.6652, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.6867794947265146, | |
| "grad_norm": 0.9557518673713279, | |
| "learning_rate": 0.0001799651582570543, | |
| "loss": 1.351, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.6907039489820946, | |
| "grad_norm": 0.470478175765928, | |
| "learning_rate": 0.00017971000165703434, | |
| "loss": 1.5682, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.6946284032376747, | |
| "grad_norm": 0.6819419573938305, | |
| "learning_rate": 0.00017945341383584818, | |
| "loss": 1.5558, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.6985528574932548, | |
| "grad_norm": 0.40304828553681576, | |
| "learning_rate": 0.00017919539940062067, | |
| "loss": 1.3134, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.7024773117488349, | |
| "grad_norm": 0.5161923622518703, | |
| "learning_rate": 0.00017893596298409182, | |
| "loss": 1.6667, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.7064017660044151, | |
| "grad_norm": 0.9882742955939973, | |
| "learning_rate": 0.00017867510924453393, | |
| "loss": 1.3512, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7103262202599951, | |
| "grad_norm": 0.44332053639010927, | |
| "learning_rate": 0.00017841284286566783, | |
| "loss": 1.5857, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.7142506745155752, | |
| "grad_norm": 0.6700436745473688, | |
| "learning_rate": 0.00017814916855657872, | |
| "loss": 1.571, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.7181751287711553, | |
| "grad_norm": 0.3683751268099129, | |
| "learning_rate": 0.00017788409105163176, | |
| "loss": 1.3221, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.7220995830267354, | |
| "grad_norm": 0.5146579028792997, | |
| "learning_rate": 0.00017761761511038695, | |
| "loss": 1.6366, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.7260240372823155, | |
| "grad_norm": 0.9944553655446299, | |
| "learning_rate": 0.0001773497455175137, | |
| "loss": 1.3721, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.7299484915378955, | |
| "grad_norm": 0.4734411042764125, | |
| "learning_rate": 0.00017708048708270496, | |
| "loss": 1.5784, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.7338729457934756, | |
| "grad_norm": 0.6783920948204912, | |
| "learning_rate": 0.00017680984464059078, | |
| "loss": 1.5776, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.7377974000490557, | |
| "grad_norm": 0.364477492783294, | |
| "learning_rate": 0.00017653782305065158, | |
| "loss": 1.3224, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.7417218543046358, | |
| "grad_norm": 0.5348683245844875, | |
| "learning_rate": 0.00017626442719713084, | |
| "loss": 1.6972, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.7456463085602159, | |
| "grad_norm": 0.9521510487113936, | |
| "learning_rate": 0.00017598966198894745, | |
| "loss": 1.3557, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.7495707628157959, | |
| "grad_norm": 0.4632996337357782, | |
| "learning_rate": 0.00017571353235960752, | |
| "loss": 1.5654, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.753495217071376, | |
| "grad_norm": 0.6787226249761125, | |
| "learning_rate": 0.00017543604326711592, | |
| "loss": 1.5723, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.7574196713269561, | |
| "grad_norm": 0.3619073984099986, | |
| "learning_rate": 0.00017515719969388697, | |
| "loss": 1.3466, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.7613441255825362, | |
| "grad_norm": 0.5218296341030427, | |
| "learning_rate": 0.00017487700664665535, | |
| "loss": 1.6507, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.7652685798381162, | |
| "grad_norm": 0.9413985897827517, | |
| "learning_rate": 0.00017459546915638593, | |
| "loss": 1.362, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.7691930340936963, | |
| "grad_norm": 0.4876406130122629, | |
| "learning_rate": 0.0001743125922781836, | |
| "loss": 1.5667, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.7731174883492764, | |
| "grad_norm": 0.6727712432072103, | |
| "learning_rate": 0.00017402838109120228, | |
| "loss": 1.5532, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.7770419426048565, | |
| "grad_norm": 0.3876035744275477, | |
| "learning_rate": 0.0001737428406985541, | |
| "loss": 1.2998, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.7809663968604366, | |
| "grad_norm": 0.5398370250844239, | |
| "learning_rate": 0.00017345597622721727, | |
| "loss": 1.6241, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.7848908511160166, | |
| "grad_norm": 0.9816739528987963, | |
| "learning_rate": 0.00017316779282794458, | |
| "loss": 1.3535, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.7888153053715967, | |
| "grad_norm": 0.46130535908354453, | |
| "learning_rate": 0.00017287829567517047, | |
| "loss": 1.5538, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.7927397596271768, | |
| "grad_norm": 0.6956873616607866, | |
| "learning_rate": 0.00017258748996691828, | |
| "loss": 1.5198, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.7966642138827569, | |
| "grad_norm": 0.3358494642080284, | |
| "learning_rate": 0.00017229538092470708, | |
| "loss": 1.3128, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.8005886681383371, | |
| "grad_norm": 0.4858973694156994, | |
| "learning_rate": 0.00017200197379345751, | |
| "loss": 1.6416, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.8045131223939171, | |
| "grad_norm": 0.9466638224977212, | |
| "learning_rate": 0.00017170727384139807, | |
| "loss": 1.3536, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.8084375766494972, | |
| "grad_norm": 0.44501519691610336, | |
| "learning_rate": 0.00017141128635997027, | |
| "loss": 1.5411, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.8123620309050773, | |
| "grad_norm": 0.6689890751102379, | |
| "learning_rate": 0.00017111401666373356, | |
| "loss": 1.5584, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.8162864851606574, | |
| "grad_norm": 0.39182526504709664, | |
| "learning_rate": 0.00017081547009027012, | |
| "loss": 1.3139, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.8202109394162375, | |
| "grad_norm": 0.49953839719232435, | |
| "learning_rate": 0.00017051565200008886, | |
| "loss": 1.6124, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.8241353936718175, | |
| "grad_norm": 0.9166157830601325, | |
| "learning_rate": 0.00017021456777652926, | |
| "loss": 1.3402, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8280598479273976, | |
| "grad_norm": 0.46688438137028465, | |
| "learning_rate": 0.00016991222282566464, | |
| "loss": 1.5058, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.8319843021829777, | |
| "grad_norm": 0.6852347408779391, | |
| "learning_rate": 0.00016960862257620507, | |
| "loss": 1.5324, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.8359087564385578, | |
| "grad_norm": 0.365213783518844, | |
| "learning_rate": 0.00016930377247940004, | |
| "loss": 1.2901, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.8398332106941379, | |
| "grad_norm": 0.5121205291421818, | |
| "learning_rate": 0.0001689976780089405, | |
| "loss": 1.5971, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.8437576649497179, | |
| "grad_norm": 1.0349784862519806, | |
| "learning_rate": 0.00016869034466086044, | |
| "loss": 1.3376, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.847682119205298, | |
| "grad_norm": 0.43878627501827516, | |
| "learning_rate": 0.00016838177795343846, | |
| "loss": 1.5425, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.8516065734608781, | |
| "grad_norm": 0.6891131722833327, | |
| "learning_rate": 0.00016807198342709857, | |
| "loss": 1.5461, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.8555310277164582, | |
| "grad_norm": 0.3652326816933217, | |
| "learning_rate": 0.0001677609666443105, | |
| "loss": 1.29, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.8594554819720382, | |
| "grad_norm": 0.5414689329724711, | |
| "learning_rate": 0.00016744873318949032, | |
| "loss": 1.6727, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.8633799362276183, | |
| "grad_norm": 1.0401307493342142, | |
| "learning_rate": 0.00016713528866889966, | |
| "loss": 1.3649, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.8673043904831984, | |
| "grad_norm": 0.4541909089910171, | |
| "learning_rate": 0.00016682063871054534, | |
| "loss": 1.5502, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.8712288447387785, | |
| "grad_norm": 0.7312632126983858, | |
| "learning_rate": 0.00016650478896407825, | |
| "loss": 1.5389, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.8751532989943586, | |
| "grad_norm": 0.3481202115883989, | |
| "learning_rate": 0.00016618774510069185, | |
| "loss": 1.298, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.8790777532499386, | |
| "grad_norm": 0.5175233572308282, | |
| "learning_rate": 0.00016586951281302045, | |
| "loss": 1.6264, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.8830022075055187, | |
| "grad_norm": 0.9773721796179596, | |
| "learning_rate": 0.0001655500978150369, | |
| "loss": 1.3322, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.8869266617610988, | |
| "grad_norm": 0.4696924761364589, | |
| "learning_rate": 0.00016522950584195, | |
| "loss": 1.5403, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.8908511160166789, | |
| "grad_norm": 0.6732142850422981, | |
| "learning_rate": 0.00016490774265010168, | |
| "loss": 1.5263, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.8947755702722591, | |
| "grad_norm": 0.36904117243032897, | |
| "learning_rate": 0.0001645848140168633, | |
| "loss": 1.2908, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.8987000245278391, | |
| "grad_norm": 0.5013283249137754, | |
| "learning_rate": 0.00016426072574053237, | |
| "loss": 1.5971, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.9026244787834192, | |
| "grad_norm": 0.963803419830796, | |
| "learning_rate": 0.00016393548364022803, | |
| "loss": 1.3281, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.9065489330389993, | |
| "grad_norm": 0.4616755659845405, | |
| "learning_rate": 0.0001636090935557868, | |
| "loss": 1.546, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.9104733872945794, | |
| "grad_norm": 0.6571907439128025, | |
| "learning_rate": 0.00016328156134765763, | |
| "loss": 1.5314, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.9143978415501595, | |
| "grad_norm": 0.3621297015795009, | |
| "learning_rate": 0.00016295289289679674, | |
| "loss": 1.3081, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.9183222958057395, | |
| "grad_norm": 0.5240312614638241, | |
| "learning_rate": 0.000162623094104562, | |
| "loss": 1.6003, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.9222467500613196, | |
| "grad_norm": 0.9499517711127529, | |
| "learning_rate": 0.00016229217089260692, | |
| "loss": 1.3211, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.9261712043168997, | |
| "grad_norm": 0.46177124841897793, | |
| "learning_rate": 0.00016196012920277438, | |
| "loss": 1.5432, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.9300956585724798, | |
| "grad_norm": 0.6638253799984756, | |
| "learning_rate": 0.00016162697499698986, | |
| "loss": 1.5379, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.9340201128280599, | |
| "grad_norm": 0.36010638182220983, | |
| "learning_rate": 0.0001612927142571546, | |
| "loss": 1.291, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.9379445670836399, | |
| "grad_norm": 0.5181849495627148, | |
| "learning_rate": 0.00016095735298503788, | |
| "loss": 1.6005, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.94186902133922, | |
| "grad_norm": 0.9287232179906421, | |
| "learning_rate": 0.00016062089720216956, | |
| "loss": 1.3124, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.9457934755948001, | |
| "grad_norm": 0.47523102829495245, | |
| "learning_rate": 0.0001602833529497318, | |
| "loss": 1.5387, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.9497179298503802, | |
| "grad_norm": 0.6563799414225217, | |
| "learning_rate": 0.00015994472628845055, | |
| "loss": 1.5288, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.9536423841059603, | |
| "grad_norm": 0.3602013042677379, | |
| "learning_rate": 0.0001596050232984868, | |
| "loss": 1.263, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.9575668383615403, | |
| "grad_norm": 0.5017795257825807, | |
| "learning_rate": 0.00015926425007932747, | |
| "loss": 1.5703, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.9614912926171204, | |
| "grad_norm": 0.951556449344021, | |
| "learning_rate": 0.00015892241274967578, | |
| "loss": 1.3134, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.9654157468727005, | |
| "grad_norm": 0.45254191277547945, | |
| "learning_rate": 0.00015857951744734146, | |
| "loss": 1.5605, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.9693402011282806, | |
| "grad_norm": 0.6738702773959866, | |
| "learning_rate": 0.00015823557032913044, | |
| "loss": 1.5254, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.9732646553838606, | |
| "grad_norm": 0.34800595098299636, | |
| "learning_rate": 0.00015789057757073445, | |
| "loss": 1.2737, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.9771891096394407, | |
| "grad_norm": 0.5020371295095916, | |
| "learning_rate": 0.00015754454536662, | |
| "loss": 1.5852, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.9811135638950208, | |
| "grad_norm": 0.9204184394254742, | |
| "learning_rate": 0.0001571974799299172, | |
| "loss": 1.3222, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.9850380181506009, | |
| "grad_norm": 0.45295537620236953, | |
| "learning_rate": 0.0001568493874923084, | |
| "loss": 1.5256, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.9889624724061811, | |
| "grad_norm": 0.6706362891939983, | |
| "learning_rate": 0.00015650027430391585, | |
| "loss": 1.4799, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.9928869266617611, | |
| "grad_norm": 0.38310437990552804, | |
| "learning_rate": 0.0001561501466331899, | |
| "loss": 1.2875, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.9968113809173412, | |
| "grad_norm": 0.5382699336933046, | |
| "learning_rate": 0.00015579901076679623, | |
| "loss": 1.5803, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.9999509443218052, | |
| "eval_loss": 1.3525456190109253, | |
| "eval_runtime": 556.6297, | |
| "eval_samples_per_second": 44.913, | |
| "eval_steps_per_second": 5.614, | |
| "step": 2548 | |
| }, | |
| { | |
| "epoch": 1.0007358351729212, | |
| "grad_norm": 0.39067263443401984, | |
| "learning_rate": 0.00015544687300950307, | |
| "loss": 1.2946, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.0046602894285013, | |
| "grad_norm": 0.4861682022299641, | |
| "learning_rate": 0.0001550937396840679, | |
| "loss": 1.4272, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.0085847436840814, | |
| "grad_norm": 0.6755707562217755, | |
| "learning_rate": 0.00015473961713112405, | |
| "loss": 1.1273, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 1.0125091979396614, | |
| "grad_norm": 0.44579419792853786, | |
| "learning_rate": 0.00015438451170906673, | |
| "loss": 1.227, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.0164336521952415, | |
| "grad_norm": 0.588954494604844, | |
| "learning_rate": 0.00015402842979393884, | |
| "loss": 1.2956, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 1.0203581064508216, | |
| "grad_norm": 0.48645722242198675, | |
| "learning_rate": 0.00015367137777931673, | |
| "loss": 0.9958, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.0242825607064017, | |
| "grad_norm": 0.460191409355804, | |
| "learning_rate": 0.00015331336207619506, | |
| "loss": 1.4128, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 1.0282070149619817, | |
| "grad_norm": 0.6615931521697652, | |
| "learning_rate": 0.000152954389112872, | |
| "loss": 1.1212, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.0321314692175618, | |
| "grad_norm": 0.4330947038889887, | |
| "learning_rate": 0.00015259446533483356, | |
| "loss": 1.2043, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 1.0360559234731421, | |
| "grad_norm": 0.5806168970842769, | |
| "learning_rate": 0.00015223359720463795, | |
| "loss": 1.3012, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.0399803777287222, | |
| "grad_norm": 0.5569591923613705, | |
| "learning_rate": 0.0001518717912017997, | |
| "loss": 0.981, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.0439048319843023, | |
| "grad_norm": 0.4525215018725937, | |
| "learning_rate": 0.00015150905382267296, | |
| "loss": 1.4244, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.0478292862398824, | |
| "grad_norm": 0.7111164858512495, | |
| "learning_rate": 0.00015114539158033529, | |
| "loss": 1.1507, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 1.0517537404954624, | |
| "grad_norm": 0.4079570460236807, | |
| "learning_rate": 0.00015078081100447036, | |
| "loss": 1.2094, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.0556781947510425, | |
| "grad_norm": 0.5765034465097496, | |
| "learning_rate": 0.00015041531864125082, | |
| "loss": 1.3134, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 1.0596026490066226, | |
| "grad_norm": 0.5530302320465103, | |
| "learning_rate": 0.0001500489210532209, | |
| "loss": 0.9636, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.0635271032622027, | |
| "grad_norm": 0.4672841495079169, | |
| "learning_rate": 0.00014968162481917836, | |
| "loss": 1.4228, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 1.0674515575177828, | |
| "grad_norm": 0.6518314535364782, | |
| "learning_rate": 0.00014931343653405653, | |
| "loss": 1.14, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.0713760117733628, | |
| "grad_norm": 0.41502979877468016, | |
| "learning_rate": 0.00014894436280880576, | |
| "loss": 1.209, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 1.075300466028943, | |
| "grad_norm": 0.7413512660232928, | |
| "learning_rate": 0.00014857441027027484, | |
| "loss": 1.3134, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.079224920284523, | |
| "grad_norm": 0.5164424667692381, | |
| "learning_rate": 0.000148203585561092, | |
| "loss": 0.9729, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.083149374540103, | |
| "grad_norm": 0.47226263632697146, | |
| "learning_rate": 0.00014783189533954555, | |
| "loss": 1.4034, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.0870738287956832, | |
| "grad_norm": 0.6904793669396148, | |
| "learning_rate": 0.00014745934627946432, | |
| "loss": 1.1214, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 1.0909982830512632, | |
| "grad_norm": 0.3787836150253688, | |
| "learning_rate": 0.00014708594507009806, | |
| "loss": 1.1916, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.0949227373068433, | |
| "grad_norm": 0.5867217911672264, | |
| "learning_rate": 0.00014671169841599694, | |
| "loss": 1.3326, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 1.0988471915624234, | |
| "grad_norm": 0.5673688454510865, | |
| "learning_rate": 0.00014633661303689157, | |
| "loss": 0.9751, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.1027716458180035, | |
| "grad_norm": 0.49813819186380237, | |
| "learning_rate": 0.00014596069566757207, | |
| "loss": 1.4027, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 1.1066961000735835, | |
| "grad_norm": 0.6563155192278646, | |
| "learning_rate": 0.0001455839530577673, | |
| "loss": 1.1347, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.1106205543291636, | |
| "grad_norm": 0.4232226319878965, | |
| "learning_rate": 0.00014520639197202354, | |
| "loss": 1.1997, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 1.1145450085847437, | |
| "grad_norm": 0.5900808405297482, | |
| "learning_rate": 0.0001448280191895831, | |
| "loss": 1.2989, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.1184694628403238, | |
| "grad_norm": 0.5830443313862127, | |
| "learning_rate": 0.00014444884150426269, | |
| "loss": 0.9978, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.1223939170959039, | |
| "grad_norm": 0.4948578637444819, | |
| "learning_rate": 0.00014406886572433113, | |
| "loss": 1.4179, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.126318371351484, | |
| "grad_norm": 0.6825354261561027, | |
| "learning_rate": 0.00014368809867238753, | |
| "loss": 1.1244, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 1.130242825607064, | |
| "grad_norm": 0.3974750472929623, | |
| "learning_rate": 0.00014330654718523847, | |
| "loss": 1.19, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.134167279862644, | |
| "grad_norm": 0.5892705512188874, | |
| "learning_rate": 0.00014292421811377533, | |
| "loss": 1.337, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 1.1380917341182242, | |
| "grad_norm": 0.5325822853964849, | |
| "learning_rate": 0.00014254111832285126, | |
| "loss": 0.9894, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.1420161883738043, | |
| "grad_norm": 0.47538854248715384, | |
| "learning_rate": 0.00014215725469115807, | |
| "loss": 1.4102, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 1.1459406426293843, | |
| "grad_norm": 0.6703710955585621, | |
| "learning_rate": 0.0001417726341111025, | |
| "loss": 1.124, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.1498650968849644, | |
| "grad_norm": 0.38415902579921885, | |
| "learning_rate": 0.00014138726348868249, | |
| "loss": 1.1943, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 1.1537895511405445, | |
| "grad_norm": 0.6046621765615349, | |
| "learning_rate": 0.00014100114974336352, | |
| "loss": 1.3096, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.1577140053961246, | |
| "grad_norm": 0.5442940244286455, | |
| "learning_rate": 0.00014061429980795383, | |
| "loss": 0.9984, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.1616384596517046, | |
| "grad_norm": 0.472199809356922, | |
| "learning_rate": 0.00014022672062848036, | |
| "loss": 1.3941, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.1655629139072847, | |
| "grad_norm": 0.6815697086629298, | |
| "learning_rate": 0.00013983841916406382, | |
| "loss": 1.1274, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 1.1694873681628648, | |
| "grad_norm": 0.39842919627411405, | |
| "learning_rate": 0.00013944940238679382, | |
| "loss": 1.1936, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.1734118224184449, | |
| "grad_norm": 0.6125359956656986, | |
| "learning_rate": 0.00013905967728160368, | |
| "loss": 1.2979, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 1.177336276674025, | |
| "grad_norm": 0.4887759184695441, | |
| "learning_rate": 0.00013866925084614503, | |
| "loss": 1.0004, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.181260730929605, | |
| "grad_norm": 0.47287421057113876, | |
| "learning_rate": 0.00013827813009066202, | |
| "loss": 1.3823, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 1.1851851851851851, | |
| "grad_norm": 0.637392600509122, | |
| "learning_rate": 0.00013788632203786565, | |
| "loss": 1.1273, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.1891096394407652, | |
| "grad_norm": 0.3855982958013649, | |
| "learning_rate": 0.0001374938337228076, | |
| "loss": 1.1998, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 1.1930340936963453, | |
| "grad_norm": 0.5974552616353569, | |
| "learning_rate": 0.00013710067219275382, | |
| "loss": 1.305, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.1969585479519254, | |
| "grad_norm": 0.5362912574676854, | |
| "learning_rate": 0.00013670684450705812, | |
| "loss": 0.9734, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.2008830022075054, | |
| "grad_norm": 0.4760113440821045, | |
| "learning_rate": 0.00013631235773703534, | |
| "loss": 1.3858, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.2048074564630855, | |
| "grad_norm": 0.6603827598013589, | |
| "learning_rate": 0.00013591721896583453, | |
| "loss": 1.1477, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 1.2087319107186656, | |
| "grad_norm": 0.3913108399865662, | |
| "learning_rate": 0.0001355214352883115, | |
| "loss": 1.1867, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.212656364974246, | |
| "grad_norm": 0.5556996510000729, | |
| "learning_rate": 0.00013512501381090158, | |
| "loss": 1.296, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 1.2165808192298257, | |
| "grad_norm": 0.482843274893376, | |
| "learning_rate": 0.00013472796165149217, | |
| "loss": 0.9715, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.220505273485406, | |
| "grad_norm": 0.459640441972295, | |
| "learning_rate": 0.00013433028593929467, | |
| "loss": 1.3808, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 1.224429727740986, | |
| "grad_norm": 0.6638173151369651, | |
| "learning_rate": 0.00013393199381471656, | |
| "loss": 1.101, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.2283541819965662, | |
| "grad_norm": 0.39719527896686296, | |
| "learning_rate": 0.00013353309242923334, | |
| "loss": 1.1904, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 1.232278636252146, | |
| "grad_norm": 0.5703528338894059, | |
| "learning_rate": 0.00013313358894525995, | |
| "loss": 1.3037, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.2362030905077264, | |
| "grad_norm": 0.5092757726206938, | |
| "learning_rate": 0.0001327334905360222, | |
| "loss": 0.9544, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.2401275447633064, | |
| "grad_norm": 0.46737896719226835, | |
| "learning_rate": 0.00013233280438542796, | |
| "loss": 1.4111, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.2440519990188865, | |
| "grad_norm": 0.667186194123786, | |
| "learning_rate": 0.0001319315376879383, | |
| "loss": 1.1044, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 1.2479764532744666, | |
| "grad_norm": 0.387627465331872, | |
| "learning_rate": 0.00013152969764843812, | |
| "loss": 1.1933, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.2519009075300467, | |
| "grad_norm": 0.5845148114080096, | |
| "learning_rate": 0.00013112729148210693, | |
| "loss": 1.3106, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 1.2558253617856268, | |
| "grad_norm": 0.4797852499891708, | |
| "learning_rate": 0.00013072432641428932, | |
| "loss": 0.9721, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.2597498160412068, | |
| "grad_norm": 0.4570115948864019, | |
| "learning_rate": 0.00013032080968036498, | |
| "loss": 1.3856, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 1.263674270296787, | |
| "grad_norm": 0.670985296905577, | |
| "learning_rate": 0.00012991674852561903, | |
| "loss": 1.1332, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.267598724552367, | |
| "grad_norm": 0.3786609746253102, | |
| "learning_rate": 0.00012951215020511197, | |
| "loss": 1.1818, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 1.271523178807947, | |
| "grad_norm": 0.5671179770848731, | |
| "learning_rate": 0.00012910702198354915, | |
| "loss": 1.2913, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.2754476330635272, | |
| "grad_norm": 0.48050089795209666, | |
| "learning_rate": 0.00012870137113515053, | |
| "loss": 0.9789, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.2793720873191072, | |
| "grad_norm": 0.4663825401149495, | |
| "learning_rate": 0.00012829520494352003, | |
| "loss": 1.3647, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.2832965415746873, | |
| "grad_norm": 0.6406781599120368, | |
| "learning_rate": 0.00012788853070151476, | |
| "loss": 1.0934, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 1.2872209958302674, | |
| "grad_norm": 0.36869696729284446, | |
| "learning_rate": 0.00012748135571111403, | |
| "loss": 1.1903, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.2911454500858475, | |
| "grad_norm": 0.5755351144103023, | |
| "learning_rate": 0.00012707368728328826, | |
| "loss": 1.2738, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 1.2950699043414275, | |
| "grad_norm": 0.5277517390621715, | |
| "learning_rate": 0.0001266655327378677, | |
| "loss": 0.9647, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.2989943585970076, | |
| "grad_norm": 0.4751776343259933, | |
| "learning_rate": 0.000126256899403411, | |
| "loss": 1.3603, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 1.3029188128525877, | |
| "grad_norm": 0.7045824297531149, | |
| "learning_rate": 0.00012584779461707375, | |
| "loss": 1.1245, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.3068432671081678, | |
| "grad_norm": 0.3861372204901668, | |
| "learning_rate": 0.0001254382257244765, | |
| "loss": 1.1699, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 1.3107677213637479, | |
| "grad_norm": 0.5585402169076681, | |
| "learning_rate": 0.000125028200079573, | |
| "loss": 1.2789, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.314692175619328, | |
| "grad_norm": 0.45453908046812247, | |
| "learning_rate": 0.0001246177250445182, | |
| "loss": 0.9823, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.318616629874908, | |
| "grad_norm": 0.463513047852769, | |
| "learning_rate": 0.00012420680798953603, | |
| "loss": 1.3859, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.322541084130488, | |
| "grad_norm": 0.6371928863025532, | |
| "learning_rate": 0.00012379545629278692, | |
| "loss": 1.1185, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 1.3264655383860682, | |
| "grad_norm": 0.389389707867679, | |
| "learning_rate": 0.0001233836773402356, | |
| "loss": 1.1818, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.3303899926416483, | |
| "grad_norm": 0.5732460521673318, | |
| "learning_rate": 0.00012297147852551821, | |
| "loss": 1.295, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 1.3343144468972283, | |
| "grad_norm": 0.48896660253260416, | |
| "learning_rate": 0.00012255886724980974, | |
| "loss": 0.9618, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.3382389011528084, | |
| "grad_norm": 0.463865332230611, | |
| "learning_rate": 0.000122145850921691, | |
| "loss": 1.332, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 1.3421633554083885, | |
| "grad_norm": 0.6790426294736074, | |
| "learning_rate": 0.00012173243695701576, | |
| "loss": 1.1214, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.3460878096639686, | |
| "grad_norm": 0.40445411375649887, | |
| "learning_rate": 0.00012131863277877731, | |
| "loss": 1.1863, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 1.3500122639195486, | |
| "grad_norm": 0.5744871135490961, | |
| "learning_rate": 0.0001209044458169756, | |
| "loss": 1.2974, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.3539367181751287, | |
| "grad_norm": 0.4251117118348158, | |
| "learning_rate": 0.00012048988350848338, | |
| "loss": 0.9984, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.3578611724307088, | |
| "grad_norm": 0.43636304441240586, | |
| "learning_rate": 0.00012007495329691301, | |
| "loss": 1.3656, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.3617856266862889, | |
| "grad_norm": 0.6729487071230792, | |
| "learning_rate": 0.00011965966263248267, | |
| "loss": 1.176, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 1.365710080941869, | |
| "grad_norm": 0.3821790113971227, | |
| "learning_rate": 0.00011924401897188248, | |
| "loss": 1.1968, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.369634535197449, | |
| "grad_norm": 0.5960303659851164, | |
| "learning_rate": 0.00011882802977814091, | |
| "loss": 1.268, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 1.3735589894530291, | |
| "grad_norm": 0.5115745025848752, | |
| "learning_rate": 0.0001184117025204905, | |
| "loss": 0.9385, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.3774834437086092, | |
| "grad_norm": 0.46803192538292715, | |
| "learning_rate": 0.0001179950446742338, | |
| "loss": 1.3878, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 1.3814078979641893, | |
| "grad_norm": 0.6674758662705085, | |
| "learning_rate": 0.00011757806372060933, | |
| "loss": 1.1107, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 1.3853323522197694, | |
| "grad_norm": 0.37645044850730486, | |
| "learning_rate": 0.000117160767146657, | |
| "loss": 1.1455, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 1.3892568064753497, | |
| "grad_norm": 0.5793847181153801, | |
| "learning_rate": 0.00011674316244508381, | |
| "loss": 1.2769, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 1.3931812607309295, | |
| "grad_norm": 0.4541582126730007, | |
| "learning_rate": 0.00011632525711412937, | |
| "loss": 0.9922, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.3971057149865098, | |
| "grad_norm": 0.45483223351146684, | |
| "learning_rate": 0.00011590705865743107, | |
| "loss": 1.3513, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 1.4010301692420897, | |
| "grad_norm": 0.6202671853411204, | |
| "learning_rate": 0.00011548857458388967, | |
| "loss": 1.1041, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 1.40495462349767, | |
| "grad_norm": 0.38763825470358887, | |
| "learning_rate": 0.00011506981240753405, | |
| "loss": 1.1803, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 1.4088790777532498, | |
| "grad_norm": 0.5604091368346654, | |
| "learning_rate": 0.00011465077964738673, | |
| "loss": 1.2642, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 1.4128035320088301, | |
| "grad_norm": 0.4991372528453274, | |
| "learning_rate": 0.00011423148382732853, | |
| "loss": 0.9732, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.41672798626441, | |
| "grad_norm": 0.46222388023464656, | |
| "learning_rate": 0.00011381193247596364, | |
| "loss": 1.351, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 1.4206524405199903, | |
| "grad_norm": 0.6374182986958278, | |
| "learning_rate": 0.0001133921331264844, | |
| "loss": 1.1078, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 1.4245768947755701, | |
| "grad_norm": 0.3879783326567086, | |
| "learning_rate": 0.00011297209331653606, | |
| "loss": 1.1537, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 1.4285013490311504, | |
| "grad_norm": 0.5480862792694288, | |
| "learning_rate": 0.00011255182058808143, | |
| "loss": 1.2822, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 1.4324258032867303, | |
| "grad_norm": 0.4402654620390705, | |
| "learning_rate": 0.00011213132248726541, | |
| "loss": 1.0041, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.4363502575423106, | |
| "grad_norm": 0.4371447741099018, | |
| "learning_rate": 0.00011171060656427956, | |
| "loss": 1.3509, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 1.4402747117978907, | |
| "grad_norm": 0.6861160035913758, | |
| "learning_rate": 0.00011128968037322653, | |
| "loss": 1.1255, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 1.4441991660534708, | |
| "grad_norm": 0.3898032640903237, | |
| "learning_rate": 0.00011086855147198441, | |
| "loss": 1.1772, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 1.4481236203090508, | |
| "grad_norm": 0.5513058135122902, | |
| "learning_rate": 0.00011044722742207101, | |
| "loss": 1.2707, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 1.452048074564631, | |
| "grad_norm": 0.45230795161803045, | |
| "learning_rate": 0.00011002571578850806, | |
| "loss": 0.9427, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.455972528820211, | |
| "grad_norm": 0.45392778443777326, | |
| "learning_rate": 0.0001096040241396855, | |
| "loss": 1.3617, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 1.459896983075791, | |
| "grad_norm": 0.6712668888072351, | |
| "learning_rate": 0.0001091821600472255, | |
| "loss": 1.0938, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 1.4638214373313712, | |
| "grad_norm": 0.40198386801340213, | |
| "learning_rate": 0.00010876013108584644, | |
| "loss": 1.1537, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 1.4677458915869512, | |
| "grad_norm": 0.6009065999255034, | |
| "learning_rate": 0.00010833794483322699, | |
| "loss": 1.2851, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 1.4716703458425313, | |
| "grad_norm": 0.45082709035381513, | |
| "learning_rate": 0.00010791560886987016, | |
| "loss": 0.9573, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.4755948000981114, | |
| "grad_norm": 0.47747709627590357, | |
| "learning_rate": 0.00010749313077896697, | |
| "loss": 1.3526, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 1.4795192543536915, | |
| "grad_norm": 0.6360956513010573, | |
| "learning_rate": 0.00010707051814626034, | |
| "loss": 1.1225, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 1.4834437086092715, | |
| "grad_norm": 0.37495759794090583, | |
| "learning_rate": 0.00010664777855990909, | |
| "loss": 1.1305, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 1.4873681628648516, | |
| "grad_norm": 0.5660694467544367, | |
| "learning_rate": 0.00010622491961035139, | |
| "loss": 1.2877, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 1.4912926171204317, | |
| "grad_norm": 0.4334919194599616, | |
| "learning_rate": 0.00010580194889016865, | |
| "loss": 0.9653, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.4952170713760118, | |
| "grad_norm": 0.4617290594719095, | |
| "learning_rate": 0.00010537887399394925, | |
| "loss": 1.3367, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 1.4991415256315919, | |
| "grad_norm": 0.6411737781715829, | |
| "learning_rate": 0.00010495570251815204, | |
| "loss": 1.0823, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 1.503065979887172, | |
| "grad_norm": 0.4013222729452922, | |
| "learning_rate": 0.00010453244206096992, | |
| "loss": 1.1591, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 1.506990434142752, | |
| "grad_norm": 0.5553012628330309, | |
| "learning_rate": 0.00010410910022219356, | |
| "loss": 1.2573, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 1.510914888398332, | |
| "grad_norm": 0.37658055778661353, | |
| "learning_rate": 0.00010368568460307481, | |
| "loss": 0.9725, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.5148393426539122, | |
| "grad_norm": 0.46008802075280786, | |
| "learning_rate": 0.00010326220280619037, | |
| "loss": 1.3322, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 1.5187637969094923, | |
| "grad_norm": 0.6645349117632661, | |
| "learning_rate": 0.00010283866243530505, | |
| "loss": 1.0743, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 1.5226882511650723, | |
| "grad_norm": 0.368969064666281, | |
| "learning_rate": 0.00010241507109523552, | |
| "loss": 1.1543, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 1.5266127054206524, | |
| "grad_norm": 0.5629083690937815, | |
| "learning_rate": 0.00010199143639171349, | |
| "loss": 1.2844, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 1.5305371596762325, | |
| "grad_norm": 0.4276111752221813, | |
| "learning_rate": 0.00010156776593124933, | |
| "loss": 0.9595, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.5344616139318126, | |
| "grad_norm": 0.4489530717278432, | |
| "learning_rate": 0.00010114406732099548, | |
| "loss": 1.3504, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 1.5383860681873927, | |
| "grad_norm": 0.6894932677390223, | |
| "learning_rate": 0.00010072034816860979, | |
| "loss": 1.1052, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 1.5423105224429727, | |
| "grad_norm": 0.3563959414054078, | |
| "learning_rate": 0.00010029661608211885, | |
| "loss": 1.1373, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 1.5462349766985528, | |
| "grad_norm": 0.5697809101980462, | |
| "learning_rate": 9.987287866978169e-05, | |
| "loss": 1.2627, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 1.5501594309541329, | |
| "grad_norm": 0.374519081050482, | |
| "learning_rate": 9.944914353995277e-05, | |
| "loss": 0.9751, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.554083885209713, | |
| "grad_norm": 0.4536155891712509, | |
| "learning_rate": 9.902541830094569e-05, | |
| "loss": 1.3348, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 1.5580083394652933, | |
| "grad_norm": 0.6568988129381994, | |
| "learning_rate": 9.860171056089646e-05, | |
| "loss": 1.0737, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 1.5619327937208731, | |
| "grad_norm": 0.37686268428730374, | |
| "learning_rate": 9.817802792762676e-05, | |
| "loss": 1.1252, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 1.5658572479764534, | |
| "grad_norm": 0.5405309448599105, | |
| "learning_rate": 9.775437800850763e-05, | |
| "loss": 1.2633, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 1.5697817022320333, | |
| "grad_norm": 0.39176031827085467, | |
| "learning_rate": 9.73307684103226e-05, | |
| "loss": 0.9769, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.5737061564876136, | |
| "grad_norm": 0.46120364148936954, | |
| "learning_rate": 9.690720673913133e-05, | |
| "loss": 1.2948, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 1.5776306107431934, | |
| "grad_norm": 0.6458249189660897, | |
| "learning_rate": 9.648370060013279e-05, | |
| "loss": 1.0807, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 1.5815550649987737, | |
| "grad_norm": 0.3634157711798367, | |
| "learning_rate": 9.606025759752895e-05, | |
| "loss": 1.1289, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 1.5854795192543536, | |
| "grad_norm": 0.5646831443545867, | |
| "learning_rate": 9.563688533438819e-05, | |
| "loss": 1.2736, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 1.589403973509934, | |
| "grad_norm": 0.38801379277814285, | |
| "learning_rate": 9.521359141250859e-05, | |
| "loss": 0.9478, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.5933284277655138, | |
| "grad_norm": 0.42890070034919986, | |
| "learning_rate": 9.479038343228173e-05, | |
| "loss": 1.3486, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 1.597252882021094, | |
| "grad_norm": 0.6170653796154466, | |
| "learning_rate": 9.436726899255597e-05, | |
| "loss": 1.1172, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 1.601177336276674, | |
| "grad_norm": 0.3732398908282777, | |
| "learning_rate": 9.394425569050017e-05, | |
| "loss": 1.1479, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 1.6051017905322542, | |
| "grad_norm": 0.5859785916357649, | |
| "learning_rate": 9.352135112146726e-05, | |
| "loss": 1.2461, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 1.609026244787834, | |
| "grad_norm": 0.3767540739368317, | |
| "learning_rate": 9.309856287885774e-05, | |
| "loss": 0.9547, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.6129506990434144, | |
| "grad_norm": 0.44726572676962856, | |
| "learning_rate": 9.267589855398356e-05, | |
| "loss": 1.3181, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 1.6168751532989942, | |
| "grad_norm": 0.6836553009627602, | |
| "learning_rate": 9.225336573593149e-05, | |
| "loss": 1.1064, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 1.6207996075545745, | |
| "grad_norm": 0.3691285475745124, | |
| "learning_rate": 9.183097201142722e-05, | |
| "loss": 1.1259, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 1.6247240618101544, | |
| "grad_norm": 0.5593165674967299, | |
| "learning_rate": 9.14087249646989e-05, | |
| "loss": 1.2457, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 1.6286485160657347, | |
| "grad_norm": 0.3787033632744885, | |
| "learning_rate": 9.098663217734102e-05, | |
| "loss": 0.9439, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.6325729703213145, | |
| "grad_norm": 0.4517409857598967, | |
| "learning_rate": 9.056470122817836e-05, | |
| "loss": 1.3187, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 1.6364974245768948, | |
| "grad_norm": 0.6534420386340076, | |
| "learning_rate": 9.01429396931297e-05, | |
| "loss": 1.0972, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 1.6404218788324747, | |
| "grad_norm": 0.3831415194008848, | |
| "learning_rate": 8.972135514507211e-05, | |
| "loss": 1.131, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 1.644346333088055, | |
| "grad_norm": 0.5448930741769371, | |
| "learning_rate": 8.92999551537046e-05, | |
| "loss": 1.2586, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 1.648270787343635, | |
| "grad_norm": 0.4438694337455833, | |
| "learning_rate": 8.887874728541258e-05, | |
| "loss": 0.9421, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.6521952415992152, | |
| "grad_norm": 0.4403052417064024, | |
| "learning_rate": 8.845773910313168e-05, | |
| "loss": 1.2844, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 1.6561196958547952, | |
| "grad_norm": 0.6772810954011467, | |
| "learning_rate": 8.803693816621217e-05, | |
| "loss": 1.0677, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 1.6600441501103753, | |
| "grad_norm": 0.3838457140012206, | |
| "learning_rate": 8.761635203028318e-05, | |
| "loss": 1.1111, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 1.6639686043659554, | |
| "grad_norm": 0.5716120880534437, | |
| "learning_rate": 8.719598824711693e-05, | |
| "loss": 1.2808, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 1.6678930586215355, | |
| "grad_norm": 0.4108929060218698, | |
| "learning_rate": 8.677585436449333e-05, | |
| "loss": 0.9304, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.6718175128771156, | |
| "grad_norm": 0.4553249562064126, | |
| "learning_rate": 8.635595792606419e-05, | |
| "loss": 1.3062, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 1.6757419671326956, | |
| "grad_norm": 0.6783796378292891, | |
| "learning_rate": 8.593630647121809e-05, | |
| "loss": 1.0553, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 1.6796664213882757, | |
| "grad_norm": 0.38508297414806236, | |
| "learning_rate": 8.551690753494476e-05, | |
| "loss": 1.1138, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 1.6835908756438558, | |
| "grad_norm": 0.567358998045147, | |
| "learning_rate": 8.509776864769981e-05, | |
| "loss": 1.2443, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 1.6875153298994359, | |
| "grad_norm": 0.4091128392537134, | |
| "learning_rate": 8.467889733526977e-05, | |
| "loss": 0.9329, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.691439784155016, | |
| "grad_norm": 0.4508992186968077, | |
| "learning_rate": 8.426030111863653e-05, | |
| "loss": 1.3184, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 1.695364238410596, | |
| "grad_norm": 0.6778472952872802, | |
| "learning_rate": 8.384198751384272e-05, | |
| "loss": 1.0666, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 1.699288692666176, | |
| "grad_norm": 0.3743718225272674, | |
| "learning_rate": 8.342396403185648e-05, | |
| "loss": 1.1191, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 1.7032131469217562, | |
| "grad_norm": 0.5599884756480523, | |
| "learning_rate": 8.300623817843673e-05, | |
| "loss": 1.206, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 1.7071376011773363, | |
| "grad_norm": 0.3621514054356959, | |
| "learning_rate": 8.258881745399837e-05, | |
| "loss": 0.9234, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.7110620554329163, | |
| "grad_norm": 0.4536694351530376, | |
| "learning_rate": 8.217170935347756e-05, | |
| "loss": 1.2667, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 1.7149865096884964, | |
| "grad_norm": 0.6347947487978, | |
| "learning_rate": 8.175492136619729e-05, | |
| "loss": 1.0458, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 1.7189109639440765, | |
| "grad_norm": 0.396820582894786, | |
| "learning_rate": 8.133846097573263e-05, | |
| "loss": 1.1286, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 1.7228354181996566, | |
| "grad_norm": 0.5542555277511924, | |
| "learning_rate": 8.092233565977669e-05, | |
| "loss": 1.2273, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 1.7267598724552367, | |
| "grad_norm": 0.39906196398972954, | |
| "learning_rate": 8.050655289000612e-05, | |
| "loss": 0.9327, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.7306843267108167, | |
| "grad_norm": 0.47332196368084956, | |
| "learning_rate": 8.009112013194706e-05, | |
| "loss": 1.3068, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 1.7346087809663968, | |
| "grad_norm": 0.6912863685561222, | |
| "learning_rate": 7.96760448448411e-05, | |
| "loss": 1.0293, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 1.738533235221977, | |
| "grad_norm": 0.3927092584563822, | |
| "learning_rate": 7.926133448151121e-05, | |
| "loss": 1.12, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 1.7424576894775572, | |
| "grad_norm": 0.5391368144897142, | |
| "learning_rate": 7.884699648822815e-05, | |
| "loss": 1.2309, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 1.746382143733137, | |
| "grad_norm": 0.3528025154533931, | |
| "learning_rate": 7.843303830457655e-05, | |
| "loss": 0.9038, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.7503065979887173, | |
| "grad_norm": 0.45596045361647103, | |
| "learning_rate": 7.801946736332144e-05, | |
| "loss": 1.2885, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 1.7542310522442972, | |
| "grad_norm": 0.6765276895876672, | |
| "learning_rate": 7.760629109027488e-05, | |
| "loss": 1.0474, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 1.7581555064998775, | |
| "grad_norm": 0.3943158006856729, | |
| "learning_rate": 7.719351690416233e-05, | |
| "loss": 1.1118, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 1.7620799607554574, | |
| "grad_norm": 0.5383284131525542, | |
| "learning_rate": 7.678115221648983e-05, | |
| "loss": 1.1949, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 1.7660044150110377, | |
| "grad_norm": 0.3896181496809045, | |
| "learning_rate": 7.636920443141057e-05, | |
| "loss": 0.9383, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.7699288692666175, | |
| "grad_norm": 0.4526612237308675, | |
| "learning_rate": 7.595768094559226e-05, | |
| "loss": 1.2734, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 1.7738533235221978, | |
| "grad_norm": 0.6384117741345257, | |
| "learning_rate": 7.554658914808404e-05, | |
| "loss": 1.0667, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.4330218997806237, | |
| "learning_rate": 7.513593642018398e-05, | |
| "loss": 1.1204, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 1.781702232033358, | |
| "grad_norm": 0.5773351977112755, | |
| "learning_rate": 7.472573013530657e-05, | |
| "loss": 1.2134, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 1.7856266862889378, | |
| "grad_norm": 0.3817639509246172, | |
| "learning_rate": 7.431597765885013e-05, | |
| "loss": 0.9482, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.7895511405445181, | |
| "grad_norm": 0.4695232084681052, | |
| "learning_rate": 7.39066863480648e-05, | |
| "loss": 1.285, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 1.793475594800098, | |
| "grad_norm": 0.6619232019857513, | |
| "learning_rate": 7.349786355192023e-05, | |
| "loss": 1.0449, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 1.7974000490556783, | |
| "grad_norm": 0.3980451698506084, | |
| "learning_rate": 7.308951661097379e-05, | |
| "loss": 1.072, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 1.8013245033112582, | |
| "grad_norm": 0.557270138634308, | |
| "learning_rate": 7.268165285723875e-05, | |
| "loss": 1.2276, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 1.8052489575668385, | |
| "grad_norm": 0.34333165709332275, | |
| "learning_rate": 7.227427961405245e-05, | |
| "loss": 0.9287, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.8091734118224183, | |
| "grad_norm": 0.4594606657362446, | |
| "learning_rate": 7.186740419594505e-05, | |
| "loss": 1.2805, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 1.8130978660779986, | |
| "grad_norm": 0.6609436089789945, | |
| "learning_rate": 7.1461033908508e-05, | |
| "loss": 1.0481, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 1.8170223203335785, | |
| "grad_norm": 0.37720527909192925, | |
| "learning_rate": 7.105517604826309e-05, | |
| "loss": 1.1124, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 1.8209467745891588, | |
| "grad_norm": 0.5459238823894826, | |
| "learning_rate": 7.064983790253108e-05, | |
| "loss": 1.1964, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 1.8248712288447386, | |
| "grad_norm": 0.3703455624784529, | |
| "learning_rate": 7.024502674930132e-05, | |
| "loss": 0.9235, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.828795683100319, | |
| "grad_norm": 0.4416963396647068, | |
| "learning_rate": 6.984074985710067e-05, | |
| "loss": 1.2576, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 1.8327201373558988, | |
| "grad_norm": 0.6659178146037718, | |
| "learning_rate": 6.943701448486313e-05, | |
| "loss": 1.0805, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 1.836644591611479, | |
| "grad_norm": 0.38151503795964853, | |
| "learning_rate": 6.903382788179961e-05, | |
| "loss": 1.119, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 1.8405690458670592, | |
| "grad_norm": 0.5485854183432936, | |
| "learning_rate": 6.863119728726763e-05, | |
| "loss": 1.2125, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 1.8444935001226392, | |
| "grad_norm": 0.34844367285040573, | |
| "learning_rate": 6.82291299306414e-05, | |
| "loss": 0.9221, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.8484179543782193, | |
| "grad_norm": 0.46514385554661264, | |
| "learning_rate": 6.782763303118193e-05, | |
| "loss": 1.2969, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 1.8523424086337994, | |
| "grad_norm": 0.6668885973032611, | |
| "learning_rate": 6.742671379790756e-05, | |
| "loss": 1.0245, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 1.8562668628893795, | |
| "grad_norm": 0.39143531162064665, | |
| "learning_rate": 6.702637942946442e-05, | |
| "loss": 1.0879, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 1.8601913171449596, | |
| "grad_norm": 0.5415666746659333, | |
| "learning_rate": 6.662663711399705e-05, | |
| "loss": 1.1786, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 1.8641157714005396, | |
| "grad_norm": 0.31170511428745284, | |
| "learning_rate": 6.622749402901971e-05, | |
| "loss": 0.9156, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.8680402256561197, | |
| "grad_norm": 0.46313840142270085, | |
| "learning_rate": 6.582895734128702e-05, | |
| "loss": 1.2656, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 1.8719646799116998, | |
| "grad_norm": 0.6733966002909243, | |
| "learning_rate": 6.543103420666569e-05, | |
| "loss": 1.0653, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 1.8758891341672799, | |
| "grad_norm": 0.3605794297518604, | |
| "learning_rate": 6.503373177000581e-05, | |
| "loss": 1.0922, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 1.87981358842286, | |
| "grad_norm": 0.5397865222128698, | |
| "learning_rate": 6.46370571650126e-05, | |
| "loss": 1.2121, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 1.88373804267844, | |
| "grad_norm": 0.35072810119027953, | |
| "learning_rate": 6.424101751411842e-05, | |
| "loss": 0.9015, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.88766249693402, | |
| "grad_norm": 0.4779687330438575, | |
| "learning_rate": 6.384561992835467e-05, | |
| "loss": 1.2459, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 1.8915869511896002, | |
| "grad_norm": 0.6400468572026754, | |
| "learning_rate": 6.34508715072244e-05, | |
| "loss": 1.0246, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 1.8955114054451803, | |
| "grad_norm": 0.3988078020560303, | |
| "learning_rate": 6.305677933857455e-05, | |
| "loss": 1.0855, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 1.8994358597007603, | |
| "grad_norm": 0.5370700832372044, | |
| "learning_rate": 6.266335049846885e-05, | |
| "loss": 1.1744, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 1.9033603139563404, | |
| "grad_norm": 0.3076821202013721, | |
| "learning_rate": 6.227059205106084e-05, | |
| "loss": 0.9041, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.9072847682119205, | |
| "grad_norm": 0.49427056716113366, | |
| "learning_rate": 6.187851104846676e-05, | |
| "loss": 1.2532, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 1.9112092224675006, | |
| "grad_norm": 0.6423321801065038, | |
| "learning_rate": 6.148711453063921e-05, | |
| "loss": 1.0381, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 1.9151336767230807, | |
| "grad_norm": 0.3642625657326075, | |
| "learning_rate": 6.109640952524051e-05, | |
| "loss": 1.0789, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 1.9190581309786607, | |
| "grad_norm": 0.526039647683758, | |
| "learning_rate": 6.0706403047516766e-05, | |
| "loss": 1.2027, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 1.9229825852342408, | |
| "grad_norm": 0.3350525820599839, | |
| "learning_rate": 6.0317102100171706e-05, | |
| "loss": 0.908, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.926907039489821, | |
| "grad_norm": 0.46000799410611715, | |
| "learning_rate": 5.992851367324097e-05, | |
| "loss": 1.232, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 1.930831493745401, | |
| "grad_norm": 0.6683329965204113, | |
| "learning_rate": 5.954064474396674e-05, | |
| "loss": 1.0662, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 1.9347559480009813, | |
| "grad_norm": 0.4236367656377322, | |
| "learning_rate": 5.915350227667225e-05, | |
| "loss": 1.0722, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 1.9386804022565611, | |
| "grad_norm": 0.5478666612010853, | |
| "learning_rate": 5.876709322263696e-05, | |
| "loss": 1.2067, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 1.9426048565121414, | |
| "grad_norm": 0.3360912061778941, | |
| "learning_rate": 5.838142451997155e-05, | |
| "loss": 0.8946, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.9465293107677213, | |
| "grad_norm": 0.4380839320552767, | |
| "learning_rate": 5.799650309349348e-05, | |
| "loss": 1.2648, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 1.9504537650233016, | |
| "grad_norm": 0.6369854781682616, | |
| "learning_rate": 5.761233585460265e-05, | |
| "loss": 1.0353, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 1.9543782192788814, | |
| "grad_norm": 0.37085998767903156, | |
| "learning_rate": 5.722892970115712e-05, | |
| "loss": 1.0542, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 1.9583026735344617, | |
| "grad_norm": 0.5391893100378502, | |
| "learning_rate": 5.684629151734948e-05, | |
| "loss": 1.1871, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 1.9622271277900416, | |
| "grad_norm": 0.34289951284996467, | |
| "learning_rate": 5.646442817358317e-05, | |
| "loss": 0.8832, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.966151582045622, | |
| "grad_norm": 0.4947184457404738, | |
| "learning_rate": 5.608334652634913e-05, | |
| "loss": 1.2155, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 1.9700760363012018, | |
| "grad_norm": 0.7080338514664215, | |
| "learning_rate": 5.5703053418102514e-05, | |
| "loss": 1.0497, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 1.974000490556782, | |
| "grad_norm": 0.39503401815625133, | |
| "learning_rate": 5.5323555677140124e-05, | |
| "loss": 1.0627, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 1.977924944812362, | |
| "grad_norm": 0.5312022089780903, | |
| "learning_rate": 5.4944860117477615e-05, | |
| "loss": 1.1627, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 1.9818493990679422, | |
| "grad_norm": 0.3237449233694942, | |
| "learning_rate": 5.456697353872722e-05, | |
| "loss": 0.8909, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.985773853323522, | |
| "grad_norm": 0.4362272741116939, | |
| "learning_rate": 5.418990272597561e-05, | |
| "loss": 1.2569, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 1.9896983075791024, | |
| "grad_norm": 0.6777401015153395, | |
| "learning_rate": 5.381365444966204e-05, | |
| "loss": 1.0136, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 1.9936227618346822, | |
| "grad_norm": 0.4454819883782086, | |
| "learning_rate": 5.343823546545692e-05, | |
| "loss": 1.0453, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 1.9975472160902625, | |
| "grad_norm": 0.5972196280294546, | |
| "learning_rate": 5.306365251414043e-05, | |
| "loss": 1.1407, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 1.9999018886436106, | |
| "eval_loss": 1.0006893873214722, | |
| "eval_runtime": 556.5718, | |
| "eval_samples_per_second": 44.918, | |
| "eval_steps_per_second": 5.615, | |
| "step": 5096 | |
| }, | |
| { | |
| "epoch": 2.0014716703458424, | |
| "grad_norm": 0.2730868912445558, | |
| "learning_rate": 5.268991232148137e-05, | |
| "loss": 0.8772, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.0053961246014227, | |
| "grad_norm": 0.550752235572184, | |
| "learning_rate": 5.2317021598116634e-05, | |
| "loss": 1.0705, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 2.0093205788570025, | |
| "grad_norm": 0.5664849757382749, | |
| "learning_rate": 5.194498703943054e-05, | |
| "loss": 0.5327, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 2.013245033112583, | |
| "grad_norm": 0.4646661004269919, | |
| "learning_rate": 5.1573815325434726e-05, | |
| "loss": 1.015, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 2.0171694873681627, | |
| "grad_norm": 0.6201824175407804, | |
| "learning_rate": 5.120351312064802e-05, | |
| "loss": 0.7586, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 2.021093941623743, | |
| "grad_norm": 0.43852988858665654, | |
| "learning_rate": 5.083408707397703e-05, | |
| "loss": 0.6567, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.025018395879323, | |
| "grad_norm": 0.5423770946007749, | |
| "learning_rate": 5.0465543818596626e-05, | |
| "loss": 1.0752, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 2.028942850134903, | |
| "grad_norm": 0.5396758572985769, | |
| "learning_rate": 5.009788997183073e-05, | |
| "loss": 0.5365, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 2.032867304390483, | |
| "grad_norm": 0.45976628107394585, | |
| "learning_rate": 4.9731132135033786e-05, | |
| "loss": 1.037, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 2.0367917586460633, | |
| "grad_norm": 0.604403468186359, | |
| "learning_rate": 4.936527689347194e-05, | |
| "loss": 0.7895, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 2.040716212901643, | |
| "grad_norm": 0.3940840852427293, | |
| "learning_rate": 4.9000330816204995e-05, | |
| "loss": 0.6192, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.0446406671572235, | |
| "grad_norm": 0.5412239502974017, | |
| "learning_rate": 4.863630045596837e-05, | |
| "loss": 1.0303, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 2.0485651214128033, | |
| "grad_norm": 0.5819941719993249, | |
| "learning_rate": 4.827319234905541e-05, | |
| "loss": 0.5216, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 2.0524895756683836, | |
| "grad_norm": 0.48772737176600944, | |
| "learning_rate": 4.791101301520016e-05, | |
| "loss": 1.0116, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 2.0564140299239635, | |
| "grad_norm": 0.6082596900422584, | |
| "learning_rate": 4.754976895746007e-05, | |
| "loss": 0.7864, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 2.060338484179544, | |
| "grad_norm": 0.3651452949364422, | |
| "learning_rate": 4.7189466662099654e-05, | |
| "loss": 0.6605, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.0642629384351237, | |
| "grad_norm": 0.5417504241420229, | |
| "learning_rate": 4.683011259847345e-05, | |
| "loss": 1.0325, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 2.068187392690704, | |
| "grad_norm": 0.5561057997044287, | |
| "learning_rate": 4.647171321891034e-05, | |
| "loss": 0.542, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 2.0721118469462843, | |
| "grad_norm": 0.45769193991204427, | |
| "learning_rate": 4.61142749585975e-05, | |
| "loss": 1.008, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 2.076036301201864, | |
| "grad_norm": 0.6003096603859527, | |
| "learning_rate": 4.575780423546476e-05, | |
| "loss": 0.7667, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 2.0799607554574444, | |
| "grad_norm": 0.3258995704505201, | |
| "learning_rate": 4.540230745006961e-05, | |
| "loss": 0.6279, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.0838852097130243, | |
| "grad_norm": 0.5530991772028139, | |
| "learning_rate": 4.504779098548209e-05, | |
| "loss": 1.0134, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 2.0878096639686046, | |
| "grad_norm": 0.5392064594322553, | |
| "learning_rate": 4.469426120717025e-05, | |
| "loss": 0.5139, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 2.0917341182241844, | |
| "grad_norm": 0.49992862394626564, | |
| "learning_rate": 4.4341724462885794e-05, | |
| "loss": 0.9829, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 2.0956585724797647, | |
| "grad_norm": 0.6107831559205753, | |
| "learning_rate": 4.399018708255018e-05, | |
| "loss": 0.7776, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 2.0995830267353446, | |
| "grad_norm": 0.32116526651884425, | |
| "learning_rate": 4.363965537814102e-05, | |
| "loss": 0.6386, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.103507480990925, | |
| "grad_norm": 0.53318983972012, | |
| "learning_rate": 4.3290135643578476e-05, | |
| "loss": 1.0373, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 2.1074319352465047, | |
| "grad_norm": 0.5218824195477701, | |
| "learning_rate": 4.294163415461258e-05, | |
| "loss": 0.5416, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 2.111356389502085, | |
| "grad_norm": 0.46647785162665695, | |
| "learning_rate": 4.259415716871037e-05, | |
| "loss": 0.9936, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 2.115280843757665, | |
| "grad_norm": 0.6186972870715647, | |
| "learning_rate": 4.224771092494355e-05, | |
| "loss": 0.7696, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 2.119205298013245, | |
| "grad_norm": 0.32939000520905415, | |
| "learning_rate": 4.190230164387655e-05, | |
| "loss": 0.6354, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.123129752268825, | |
| "grad_norm": 0.5673496606713442, | |
| "learning_rate": 4.155793552745465e-05, | |
| "loss": 1.0143, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 2.1270542065244054, | |
| "grad_norm": 0.5621390987987777, | |
| "learning_rate": 4.1214618758892866e-05, | |
| "loss": 0.5293, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 2.130978660779985, | |
| "grad_norm": 0.48313945689223514, | |
| "learning_rate": 4.0872357502564686e-05, | |
| "loss": 1.0082, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 2.1349031150355655, | |
| "grad_norm": 0.606093414487324, | |
| "learning_rate": 4.053115790389159e-05, | |
| "loss": 0.7769, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 2.1388275692911454, | |
| "grad_norm": 0.32141076361893256, | |
| "learning_rate": 4.019102608923262e-05, | |
| "loss": 0.64, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.1427520235467257, | |
| "grad_norm": 0.5713343169438715, | |
| "learning_rate": 3.985196816577433e-05, | |
| "loss": 1.0242, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 2.1466764778023055, | |
| "grad_norm": 0.4976558859683901, | |
| "learning_rate": 3.951399022142127e-05, | |
| "loss": 0.5131, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 2.150600932057886, | |
| "grad_norm": 0.4610312261691466, | |
| "learning_rate": 3.9177098324686413e-05, | |
| "loss": 1.0108, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 2.1545253863134657, | |
| "grad_norm": 0.6114777422135642, | |
| "learning_rate": 3.884129852458254e-05, | |
| "loss": 0.7802, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 2.158449840569046, | |
| "grad_norm": 0.32303096358821215, | |
| "learning_rate": 3.8506596850513355e-05, | |
| "loss": 0.64, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.162374294824626, | |
| "grad_norm": 0.5462414062190428, | |
| "learning_rate": 3.817299931216537e-05, | |
| "loss": 0.9964, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 2.166298749080206, | |
| "grad_norm": 0.5869028357971195, | |
| "learning_rate": 3.7840511899399954e-05, | |
| "loss": 0.5325, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 2.170223203335786, | |
| "grad_norm": 0.4746903696326467, | |
| "learning_rate": 3.7509140582145706e-05, | |
| "loss": 0.9982, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 2.1741476575913663, | |
| "grad_norm": 0.619021723614995, | |
| "learning_rate": 3.7178891310291444e-05, | |
| "loss": 0.7559, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 2.178072111846946, | |
| "grad_norm": 0.30741535706965134, | |
| "learning_rate": 3.684977001357913e-05, | |
| "loss": 0.6586, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.1819965661025265, | |
| "grad_norm": 0.5466186362290557, | |
| "learning_rate": 3.6521782601497677e-05, | |
| "loss": 1.0082, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 2.1859210203581063, | |
| "grad_norm": 0.5620478806855952, | |
| "learning_rate": 3.619493496317662e-05, | |
| "loss": 0.5023, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 2.1898454746136866, | |
| "grad_norm": 0.44542545734748706, | |
| "learning_rate": 3.586923296728046e-05, | |
| "loss": 0.9884, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 2.1937699288692665, | |
| "grad_norm": 0.6090541537139627, | |
| "learning_rate": 3.5544682461903366e-05, | |
| "loss": 0.7936, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 2.1976943831248468, | |
| "grad_norm": 0.3081190678753379, | |
| "learning_rate": 3.522128927446392e-05, | |
| "loss": 0.6317, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.2016188373804266, | |
| "grad_norm": 0.548615459706809, | |
| "learning_rate": 3.4899059211600825e-05, | |
| "loss": 1.0187, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 2.205543291636007, | |
| "grad_norm": 0.5641989254671493, | |
| "learning_rate": 3.457799805906835e-05, | |
| "loss": 0.5163, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 2.209467745891587, | |
| "grad_norm": 0.44170413303918793, | |
| "learning_rate": 3.4258111581632636e-05, | |
| "loss": 1.0185, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 2.213392200147167, | |
| "grad_norm": 0.6105984826873107, | |
| "learning_rate": 3.3939405522968105e-05, | |
| "loss": 0.761, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 2.217316654402747, | |
| "grad_norm": 0.28933348260714464, | |
| "learning_rate": 3.362188560555434e-05, | |
| "loss": 0.6266, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.2212411086583272, | |
| "grad_norm": 0.5527907268845692, | |
| "learning_rate": 3.330555753057336e-05, | |
| "loss": 1.0289, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 2.225165562913907, | |
| "grad_norm": 0.5573465125820448, | |
| "learning_rate": 3.2990426977807156e-05, | |
| "loss": 0.526, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 2.2290900171694874, | |
| "grad_norm": 0.44137772240686285, | |
| "learning_rate": 3.2676499605535914e-05, | |
| "loss": 0.9743, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 2.2330144714250673, | |
| "grad_norm": 0.5697028995274115, | |
| "learning_rate": 3.23637810504361e-05, | |
| "loss": 0.7512, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 2.2369389256806476, | |
| "grad_norm": 0.30127325007968614, | |
| "learning_rate": 3.205227692747968e-05, | |
| "loss": 0.617, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.2408633799362274, | |
| "grad_norm": 0.5573021126219747, | |
| "learning_rate": 3.174199282983292e-05, | |
| "loss": 1.0198, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 2.2447878341918077, | |
| "grad_norm": 0.5535001985257877, | |
| "learning_rate": 3.143293432875607e-05, | |
| "loss": 0.5182, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 2.248712288447388, | |
| "grad_norm": 0.4393397700617783, | |
| "learning_rate": 3.1125106973503485e-05, | |
| "loss": 1.0059, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 2.252636742702968, | |
| "grad_norm": 0.6039985548905238, | |
| "learning_rate": 3.081851629122372e-05, | |
| "loss": 0.7838, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 2.2565611969585477, | |
| "grad_norm": 0.2927590960208694, | |
| "learning_rate": 3.051316778686055e-05, | |
| "loss": 0.6244, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.260485651214128, | |
| "grad_norm": 0.5554823525982631, | |
| "learning_rate": 3.0209066943053943e-05, | |
| "loss": 1.0175, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 2.2644101054697083, | |
| "grad_norm": 0.5839985505312678, | |
| "learning_rate": 2.990621922004172e-05, | |
| "loss": 0.4904, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 2.268334559725288, | |
| "grad_norm": 0.4511527552060238, | |
| "learning_rate": 2.960463005556149e-05, | |
| "loss": 1.0089, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 2.272259013980868, | |
| "grad_norm": 0.6091666486579106, | |
| "learning_rate": 2.9304304864752884e-05, | |
| "loss": 0.751, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 2.2761834682364483, | |
| "grad_norm": 0.3020730129642747, | |
| "learning_rate": 2.9005249040060612e-05, | |
| "loss": 0.6115, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.2801079224920286, | |
| "grad_norm": 0.5524614321832616, | |
| "learning_rate": 2.87074679511373e-05, | |
| "loss": 0.9774, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 2.2840323767476085, | |
| "grad_norm": 0.547095199501154, | |
| "learning_rate": 2.8410966944747376e-05, | |
| "loss": 0.4999, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 2.287956831003189, | |
| "grad_norm": 0.4392171366832068, | |
| "learning_rate": 2.811575134467086e-05, | |
| "loss": 0.9939, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 2.2918812852587687, | |
| "grad_norm": 0.6316197412044359, | |
| "learning_rate": 2.782182645160789e-05, | |
| "loss": 0.7447, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 2.295805739514349, | |
| "grad_norm": 0.3023577492123374, | |
| "learning_rate": 2.7529197543083508e-05, | |
| "loss": 0.6091, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.299730193769929, | |
| "grad_norm": 0.5742017678061654, | |
| "learning_rate": 2.7237869873352827e-05, | |
| "loss": 1.0067, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 2.303654648025509, | |
| "grad_norm": 0.5205864945203419, | |
| "learning_rate": 2.6947848673306853e-05, | |
| "loss": 0.5235, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 2.307579102281089, | |
| "grad_norm": 0.4459801767926128, | |
| "learning_rate": 2.6659139150378377e-05, | |
| "loss": 0.9867, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 2.3115035565366693, | |
| "grad_norm": 0.629580396153204, | |
| "learning_rate": 2.6371746488448613e-05, | |
| "loss": 0.7454, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 2.315428010792249, | |
| "grad_norm": 0.2686612734189131, | |
| "learning_rate": 2.6085675847754153e-05, | |
| "loss": 0.5835, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.3193524650478294, | |
| "grad_norm": 0.5849756819258916, | |
| "learning_rate": 2.580093236479406e-05, | |
| "loss": 0.9788, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 2.3232769193034093, | |
| "grad_norm": 0.5476494028140071, | |
| "learning_rate": 2.5517521152237965e-05, | |
| "loss": 0.5043, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 2.3272013735589896, | |
| "grad_norm": 0.4768610468530905, | |
| "learning_rate": 2.5235447298834004e-05, | |
| "loss": 0.996, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 2.3311258278145695, | |
| "grad_norm": 0.6227519349566278, | |
| "learning_rate": 2.4954715869317603e-05, | |
| "loss": 0.7466, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 2.3350502820701498, | |
| "grad_norm": 0.27687508559092505, | |
| "learning_rate": 2.4675331904320532e-05, | |
| "loss": 0.6062, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.3389747363257296, | |
| "grad_norm": 0.5789525469249787, | |
| "learning_rate": 2.43973004202803e-05, | |
| "loss": 1.0161, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 2.34289919058131, | |
| "grad_norm": 0.578017710538773, | |
| "learning_rate": 2.4120626409350212e-05, | |
| "loss": 0.4955, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 2.3468236448368898, | |
| "grad_norm": 0.5068533593849457, | |
| "learning_rate": 2.3845314839309564e-05, | |
| "loss": 0.9908, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 2.35074809909247, | |
| "grad_norm": 0.6278461330599057, | |
| "learning_rate": 2.3571370653474657e-05, | |
| "loss": 0.7461, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 2.35467255334805, | |
| "grad_norm": 0.29069555207045233, | |
| "learning_rate": 2.329879877060981e-05, | |
| "loss": 0.6023, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.3585970076036302, | |
| "grad_norm": 0.5563305537677198, | |
| "learning_rate": 2.302760408483926e-05, | |
| "loss": 1.0006, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 2.36252146185921, | |
| "grad_norm": 0.5853263403065023, | |
| "learning_rate": 2.2757791465559153e-05, | |
| "loss": 0.4994, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 2.3664459161147904, | |
| "grad_norm": 0.4865967564016656, | |
| "learning_rate": 2.2489365757350135e-05, | |
| "loss": 0.96, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 2.3703703703703702, | |
| "grad_norm": 0.6316383739440334, | |
| "learning_rate": 2.2222331779890394e-05, | |
| "loss": 0.7733, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 2.3742948246259505, | |
| "grad_norm": 0.2663124968094402, | |
| "learning_rate": 2.1956694327869042e-05, | |
| "loss": 0.6184, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.3782192788815304, | |
| "grad_norm": 0.5446521838696877, | |
| "learning_rate": 2.1692458170900198e-05, | |
| "loss": 0.9623, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 2.3821437331371107, | |
| "grad_norm": 0.5531148060056698, | |
| "learning_rate": 2.1429628053437078e-05, | |
| "loss": 0.5051, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 2.3860681873926906, | |
| "grad_norm": 0.429857502647415, | |
| "learning_rate": 2.116820869468711e-05, | |
| "loss": 0.999, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 2.389992641648271, | |
| "grad_norm": 0.6463967530114474, | |
| "learning_rate": 2.0908204788526964e-05, | |
| "loss": 0.7962, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 2.3939170959038507, | |
| "grad_norm": 0.274247420526792, | |
| "learning_rate": 2.064962100341842e-05, | |
| "loss": 0.6123, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.397841550159431, | |
| "grad_norm": 0.5498162429505886, | |
| "learning_rate": 2.0392461982324463e-05, | |
| "loss": 0.9656, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 2.401766004415011, | |
| "grad_norm": 0.5956462304875728, | |
| "learning_rate": 2.0136732342625875e-05, | |
| "loss": 0.5042, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 2.405690458670591, | |
| "grad_norm": 0.4134478838561606, | |
| "learning_rate": 1.9882436676038472e-05, | |
| "loss": 0.9974, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 2.409614912926171, | |
| "grad_norm": 0.6213757004464778, | |
| "learning_rate": 1.9629579548530553e-05, | |
| "loss": 0.7679, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 2.4135393671817513, | |
| "grad_norm": 0.30291872347068327, | |
| "learning_rate": 1.9378165500240943e-05, | |
| "loss": 0.6272, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.417463821437331, | |
| "grad_norm": 0.5516229929904102, | |
| "learning_rate": 1.9128199045397488e-05, | |
| "loss": 0.9896, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 2.4213882756929115, | |
| "grad_norm": 0.5694018196104392, | |
| "learning_rate": 1.887968467223591e-05, | |
| "loss": 0.513, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 2.425312729948492, | |
| "grad_norm": 0.4435609506243493, | |
| "learning_rate": 1.86326268429194e-05, | |
| "loss": 0.9612, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 2.4292371842040716, | |
| "grad_norm": 0.6430522391241489, | |
| "learning_rate": 1.838702999345827e-05, | |
| "loss": 0.7553, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 2.4331616384596515, | |
| "grad_norm": 0.2574037462816669, | |
| "learning_rate": 1.8142898533630536e-05, | |
| "loss": 0.6249, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.437086092715232, | |
| "grad_norm": 0.5632674815841419, | |
| "learning_rate": 1.7900236846902572e-05, | |
| "loss": 0.9955, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 2.441010546970812, | |
| "grad_norm": 0.5411993774487638, | |
| "learning_rate": 1.765904929035046e-05, | |
| "loss": 0.5061, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 2.444935001226392, | |
| "grad_norm": 0.46101386661846205, | |
| "learning_rate": 1.74193401945818e-05, | |
| "loss": 0.9863, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 2.448859455481972, | |
| "grad_norm": 0.6185384884406199, | |
| "learning_rate": 1.7181113863657805e-05, | |
| "loss": 0.7369, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 2.452783909737552, | |
| "grad_norm": 0.27477608134408393, | |
| "learning_rate": 1.6944374575016253e-05, | |
| "loss": 0.6312, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.4567083639931324, | |
| "grad_norm": 0.5585551171494181, | |
| "learning_rate": 1.670912657939443e-05, | |
| "loss": 0.9948, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 2.4606328182487123, | |
| "grad_norm": 0.5545517749209848, | |
| "learning_rate": 1.6475374100753018e-05, | |
| "loss": 0.4869, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 2.464557272504292, | |
| "grad_norm": 0.43490286688968666, | |
| "learning_rate": 1.6243121336200127e-05, | |
| "loss": 0.9652, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 2.4684817267598724, | |
| "grad_norm": 0.6515308250593556, | |
| "learning_rate": 1.601237245591599e-05, | |
| "loss": 0.7634, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 2.4724061810154527, | |
| "grad_norm": 0.2672185897279334, | |
| "learning_rate": 1.5783131603078083e-05, | |
| "loss": 0.6032, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.4763306352710326, | |
| "grad_norm": 0.5364076090373087, | |
| "learning_rate": 1.555540289378663e-05, | |
| "loss": 0.9989, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 2.480255089526613, | |
| "grad_norm": 0.5573883942902318, | |
| "learning_rate": 1.532919041699089e-05, | |
| "loss": 0.5115, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 2.4841795437821927, | |
| "grad_norm": 0.42286186941787296, | |
| "learning_rate": 1.510449823441561e-05, | |
| "loss": 0.9895, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 2.488103998037773, | |
| "grad_norm": 0.6038158280002404, | |
| "learning_rate": 1.4881330380488013e-05, | |
| "loss": 0.7297, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 2.492028452293353, | |
| "grad_norm": 0.25370650541345074, | |
| "learning_rate": 1.4659690862265673e-05, | |
| "loss": 0.6005, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.495952906548933, | |
| "grad_norm": 0.5422830707931421, | |
| "learning_rate": 1.4439583659364153e-05, | |
| "loss": 0.9838, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 2.499877360804513, | |
| "grad_norm": 0.5206331922620188, | |
| "learning_rate": 1.4221012723885874e-05, | |
| "loss": 0.5133, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 2.5038018150600934, | |
| "grad_norm": 0.4509464187391961, | |
| "learning_rate": 1.4003981980348969e-05, | |
| "loss": 0.9848, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 2.507726269315673, | |
| "grad_norm": 0.624805502305627, | |
| "learning_rate": 1.3788495325616912e-05, | |
| "loss": 0.7538, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 2.5116507235712535, | |
| "grad_norm": 0.26003721025783916, | |
| "learning_rate": 1.357455662882855e-05, | |
| "loss": 0.6082, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.5155751778268334, | |
| "grad_norm": 0.5412625437830425, | |
| "learning_rate": 1.3362169731328534e-05, | |
| "loss": 0.9675, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 2.5194996320824137, | |
| "grad_norm": 0.5845979737642246, | |
| "learning_rate": 1.3151338446598482e-05, | |
| "loss": 0.494, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 2.5234240863379935, | |
| "grad_norm": 0.4405194367976622, | |
| "learning_rate": 1.2942066560188348e-05, | |
| "loss": 0.9648, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 2.527348540593574, | |
| "grad_norm": 0.6186299546598477, | |
| "learning_rate": 1.2734357829648625e-05, | |
| "loss": 0.7183, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 2.5312729948491537, | |
| "grad_norm": 0.26408998869960737, | |
| "learning_rate": 1.2528215984462765e-05, | |
| "loss": 0.5982, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.535197449104734, | |
| "grad_norm": 0.5399304191165905, | |
| "learning_rate": 1.23236447259802e-05, | |
| "loss": 0.9958, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 2.539121903360314, | |
| "grad_norm": 0.5170618190552897, | |
| "learning_rate": 1.2120647727349976e-05, | |
| "loss": 0.4922, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 2.543046357615894, | |
| "grad_norm": 0.42228667586533813, | |
| "learning_rate": 1.1919228633454737e-05, | |
| "loss": 0.995, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 2.546970811871474, | |
| "grad_norm": 0.598622764873492, | |
| "learning_rate": 1.1719391060845298e-05, | |
| "loss": 0.7607, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 2.5508952661270543, | |
| "grad_norm": 0.257178241487348, | |
| "learning_rate": 1.1521138597675651e-05, | |
| "loss": 0.6066, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.554819720382634, | |
| "grad_norm": 0.5560367530670483, | |
| "learning_rate": 1.1324474803638652e-05, | |
| "loss": 1.0327, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 2.5587441746382145, | |
| "grad_norm": 0.5705845825042977, | |
| "learning_rate": 1.1129403209902034e-05, | |
| "loss": 0.507, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 2.5626686288937943, | |
| "grad_norm": 0.4232150358846287, | |
| "learning_rate": 1.0935927319044958e-05, | |
| "loss": 0.9455, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 2.5665930831493746, | |
| "grad_norm": 0.6103636845988778, | |
| "learning_rate": 1.0744050604995237e-05, | |
| "loss": 0.7277, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 2.5705175374049545, | |
| "grad_norm": 0.2608709410626365, | |
| "learning_rate": 1.0553776512966885e-05, | |
| "loss": 0.5893, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.574441991660535, | |
| "grad_norm": 0.5350133340613187, | |
| "learning_rate": 1.0365108459398277e-05, | |
| "loss": 0.9777, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 2.5783664459161146, | |
| "grad_norm": 0.5809985433605673, | |
| "learning_rate": 1.0178049831890768e-05, | |
| "loss": 0.5055, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 2.582290900171695, | |
| "grad_norm": 0.44155199585277793, | |
| "learning_rate": 9.992603989147942e-06, | |
| "loss": 1.0014, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 2.586215354427275, | |
| "grad_norm": 0.6033529635611596, | |
| "learning_rate": 9.808774260915244e-06, | |
| "loss": 0.7684, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 2.590139808682855, | |
| "grad_norm": 0.2726321272697929, | |
| "learning_rate": 9.62656394792023e-06, | |
| "loss": 0.5806, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.594064262938435, | |
| "grad_norm": 0.5642485862211182, | |
| "learning_rate": 9.445976321813277e-06, | |
| "loss": 0.9375, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 2.5979887171940153, | |
| "grad_norm": 0.5793712239036046, | |
| "learning_rate": 9.267014625108805e-06, | |
| "loss": 0.4923, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 2.6019131714495956, | |
| "grad_norm": 0.4428684429759492, | |
| "learning_rate": 9.08968207112717e-06, | |
| "loss": 0.9991, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 2.6058376257051754, | |
| "grad_norm": 0.6388416115247564, | |
| "learning_rate": 8.91398184393687e-06, | |
| "loss": 0.7431, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 2.6097620799607553, | |
| "grad_norm": 0.25614351044864425, | |
| "learning_rate": 8.739917098297356e-06, | |
| "loss": 0.6191, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.6136865342163356, | |
| "grad_norm": 0.568433650366567, | |
| "learning_rate": 8.56749095960251e-06, | |
| "loss": 0.9847, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 2.617610988471916, | |
| "grad_norm": 0.5661010875652216, | |
| "learning_rate": 8.396706523824372e-06, | |
| "loss": 0.4906, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 2.6215354427274957, | |
| "grad_norm": 0.4180552358189544, | |
| "learning_rate": 8.227566857457703e-06, | |
| "loss": 0.9661, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 2.6254598969830756, | |
| "grad_norm": 0.5897297010878825, | |
| "learning_rate": 8.060074997464772e-06, | |
| "loss": 0.7585, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 2.629384351238656, | |
| "grad_norm": 0.2622309480245679, | |
| "learning_rate": 7.894233951220953e-06, | |
| "loss": 0.6032, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.633308805494236, | |
| "grad_norm": 0.5808641293543836, | |
| "learning_rate": 7.73004669646069e-06, | |
| "loss": 0.9593, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 2.637233259749816, | |
| "grad_norm": 0.5368098827956075, | |
| "learning_rate": 7.567516181223966e-06, | |
| "loss": 0.4868, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 2.641157714005396, | |
| "grad_norm": 0.4175313367871406, | |
| "learning_rate": 7.406645323803463e-06, | |
| "loss": 0.9575, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 2.645082168260976, | |
| "grad_norm": 0.6219532946650866, | |
| "learning_rate": 7.247437012692104e-06, | |
| "loss": 0.7476, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 2.6490066225165565, | |
| "grad_norm": 0.26617953027654534, | |
| "learning_rate": 7.089894106531214e-06, | |
| "loss": 0.6107, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.6529310767721364, | |
| "grad_norm": 0.5876780487818677, | |
| "learning_rate": 6.9340194340592135e-06, | |
| "loss": 0.9798, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 2.656855531027716, | |
| "grad_norm": 0.5567127538241098, | |
| "learning_rate": 6.779815794060718e-06, | |
| "loss": 0.4725, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 2.6607799852832965, | |
| "grad_norm": 0.42784281442184585, | |
| "learning_rate": 6.627285955316475e-06, | |
| "loss": 0.9679, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 2.664704439538877, | |
| "grad_norm": 0.631682752559883, | |
| "learning_rate": 6.476432656553411e-06, | |
| "loss": 0.7427, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 2.6686288937944567, | |
| "grad_norm": 0.26704189636870707, | |
| "learning_rate": 6.327258606395736e-06, | |
| "loss": 0.596, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.6725533480500365, | |
| "grad_norm": 0.5460882249722191, | |
| "learning_rate": 6.179766483316041e-06, | |
| "loss": 0.9487, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 2.676477802305617, | |
| "grad_norm": 0.535085364580487, | |
| "learning_rate": 6.033958935587369e-06, | |
| "loss": 0.4902, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 2.680402256561197, | |
| "grad_norm": 0.4152111016469633, | |
| "learning_rate": 5.88983858123564e-06, | |
| "loss": 0.9771, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 2.684326710816777, | |
| "grad_norm": 0.6376790660400137, | |
| "learning_rate": 5.747408007992572e-06, | |
| "loss": 0.736, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 2.6882511650723573, | |
| "grad_norm": 0.2605931598167947, | |
| "learning_rate": 5.606669773249296e-06, | |
| "loss": 0.623, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.692175619327937, | |
| "grad_norm": 0.5438462579254439, | |
| "learning_rate": 5.467626404010406e-06, | |
| "loss": 0.9662, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 2.6961000735835174, | |
| "grad_norm": 0.6017742053140295, | |
| "learning_rate": 5.3302803968486195e-06, | |
| "loss": 0.4905, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 2.7000245278390973, | |
| "grad_norm": 0.4305016597576778, | |
| "learning_rate": 5.19463421785985e-06, | |
| "loss": 0.9687, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 2.7039489820946776, | |
| "grad_norm": 0.6296732720195917, | |
| "learning_rate": 5.0606903026190535e-06, | |
| "loss": 0.741, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 2.7078734363502575, | |
| "grad_norm": 0.25604806566569377, | |
| "learning_rate": 4.928451056136441e-06, | |
| "loss": 0.5991, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.7117978906058378, | |
| "grad_norm": 0.5504634310493172, | |
| "learning_rate": 4.797918852814254e-06, | |
| "loss": 0.9717, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 2.7157223448614176, | |
| "grad_norm": 0.5466407511340324, | |
| "learning_rate": 4.669096036404197e-06, | |
| "loss": 0.4769, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 2.719646799116998, | |
| "grad_norm": 0.4234869348890821, | |
| "learning_rate": 4.541984919965336e-06, | |
| "loss": 0.9565, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 2.7235712533725778, | |
| "grad_norm": 0.5908229563139684, | |
| "learning_rate": 4.416587785822568e-06, | |
| "loss": 0.7359, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 2.727495707628158, | |
| "grad_norm": 0.2551802008593362, | |
| "learning_rate": 4.2929068855256275e-06, | |
| "loss": 0.6036, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.731420161883738, | |
| "grad_norm": 0.5495819804709496, | |
| "learning_rate": 4.170944439808622e-06, | |
| "loss": 0.979, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 2.7353446161393182, | |
| "grad_norm": 0.5635033647660206, | |
| "learning_rate": 4.050702638550275e-06, | |
| "loss": 0.4929, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 2.739269070394898, | |
| "grad_norm": 0.42986375502541047, | |
| "learning_rate": 3.932183640734466e-06, | |
| "loss": 0.9688, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 2.7431935246504784, | |
| "grad_norm": 0.6373184809495214, | |
| "learning_rate": 3.815389574411576e-06, | |
| "loss": 0.7346, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 2.7471179789060582, | |
| "grad_norm": 0.2528218296208618, | |
| "learning_rate": 3.700322536660228e-06, | |
| "loss": 0.5915, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.7510424331616385, | |
| "grad_norm": 0.543251812613198, | |
| "learning_rate": 3.586984593549614e-06, | |
| "loss": 0.9684, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 2.7549668874172184, | |
| "grad_norm": 0.5581086761126619, | |
| "learning_rate": 3.4753777801024513e-06, | |
| "loss": 0.4773, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 2.7588913416727987, | |
| "grad_norm": 0.4120717754972658, | |
| "learning_rate": 3.3655041002583987e-06, | |
| "loss": 0.9498, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 2.7628157959283786, | |
| "grad_norm": 0.6123189485886581, | |
| "learning_rate": 3.2573655268380743e-06, | |
| "loss": 0.7057, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 2.766740250183959, | |
| "grad_norm": 0.26686987477259777, | |
| "learning_rate": 3.1509640015076948e-06, | |
| "loss": 0.6036, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.7706647044395387, | |
| "grad_norm": 0.5324373234404657, | |
| "learning_rate": 3.0463014347441255e-06, | |
| "loss": 0.9754, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 2.774589158695119, | |
| "grad_norm": 0.5674641764211202, | |
| "learning_rate": 2.94337970580062e-06, | |
| "loss": 0.4935, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 2.7785136129506993, | |
| "grad_norm": 0.4109503524310012, | |
| "learning_rate": 2.8422006626731114e-06, | |
| "loss": 0.9642, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 2.782438067206279, | |
| "grad_norm": 0.5787231723843825, | |
| "learning_rate": 2.7427661220669532e-06, | |
| "loss": 0.742, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 2.786362521461859, | |
| "grad_norm": 0.24763534771816254, | |
| "learning_rate": 2.645077869364354e-06, | |
| "loss": 0.5802, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.7902869757174393, | |
| "grad_norm": 0.5558690705526742, | |
| "learning_rate": 2.5491376585923266e-06, | |
| "loss": 0.9613, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 2.7942114299730196, | |
| "grad_norm": 0.5254522273936683, | |
| "learning_rate": 2.4549472123911566e-06, | |
| "loss": 0.4894, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 2.7981358842285995, | |
| "grad_norm": 0.4400451438446854, | |
| "learning_rate": 2.362508221983484e-06, | |
| "loss": 0.9432, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 2.8020603384841793, | |
| "grad_norm": 0.6373034570395104, | |
| "learning_rate": 2.2718223471439815e-06, | |
| "loss": 0.7307, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 2.8059847927397596, | |
| "grad_norm": 0.2501503760455942, | |
| "learning_rate": 2.182891216169447e-06, | |
| "loss": 0.5957, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.80990924699534, | |
| "grad_norm": 0.5787741585612971, | |
| "learning_rate": 2.095716425849703e-06, | |
| "loss": 0.9577, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 2.81383370125092, | |
| "grad_norm": 0.5903261690823448, | |
| "learning_rate": 2.010299541438798e-06, | |
| "loss": 0.4833, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 2.8177581555064997, | |
| "grad_norm": 0.43067331669721515, | |
| "learning_rate": 1.9266420966270184e-06, | |
| "loss": 0.9347, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 2.82168260976208, | |
| "grad_norm": 0.630628866714476, | |
| "learning_rate": 1.8447455935132418e-06, | |
| "loss": 0.7114, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 2.8256070640176603, | |
| "grad_norm": 0.26140745824215283, | |
| "learning_rate": 1.7646115025780507e-06, | |
| "loss": 0.5977, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.82953151827324, | |
| "grad_norm": 0.5431373816957603, | |
| "learning_rate": 1.6862412626572842e-06, | |
| "loss": 0.981, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 2.83345597252882, | |
| "grad_norm": 0.5324832948689947, | |
| "learning_rate": 1.6096362809162047e-06, | |
| "loss": 0.515, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 2.8373804267844003, | |
| "grad_norm": 0.4253349884960846, | |
| "learning_rate": 1.5347979328242611e-06, | |
| "loss": 0.9573, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 2.8413048810399806, | |
| "grad_norm": 0.6199214512195229, | |
| "learning_rate": 1.461727562130344e-06, | |
| "loss": 0.7262, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 2.8452293352955604, | |
| "grad_norm": 0.2520361172356776, | |
| "learning_rate": 1.3904264808387246e-06, | |
| "loss": 0.5852, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.8491537895511403, | |
| "grad_norm": 0.5433229038316792, | |
| "learning_rate": 1.3208959691854539e-06, | |
| "loss": 0.9743, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 2.8530782438067206, | |
| "grad_norm": 0.5283893049011732, | |
| "learning_rate": 1.2531372756153458e-06, | |
| "loss": 0.466, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 2.857002698062301, | |
| "grad_norm": 0.4003559202162091, | |
| "learning_rate": 1.1871516167596186e-06, | |
| "loss": 0.9691, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 2.8609271523178808, | |
| "grad_norm": 0.631799345021962, | |
| "learning_rate": 1.1229401774140448e-06, | |
| "loss": 0.7384, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 2.8648516065734606, | |
| "grad_norm": 0.2506998286496455, | |
| "learning_rate": 1.0605041105176129e-06, | |
| "loss": 0.5921, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.868776060829041, | |
| "grad_norm": 0.5447884513724497, | |
| "learning_rate": 9.998445371319332e-07, | |
| "loss": 0.9453, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 2.872700515084621, | |
| "grad_norm": 0.5921461480701601, | |
| "learning_rate": 9.409625464210092e-07, | |
| "loss": 0.4724, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 2.876624969340201, | |
| "grad_norm": 0.42136042850586963, | |
| "learning_rate": 8.838591956317311e-07, | |
| "loss": 0.9567, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 2.8805494235957814, | |
| "grad_norm": 0.6064332885879969, | |
| "learning_rate": 8.285355100748904e-07, | |
| "loss": 0.7203, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 2.8844738778513612, | |
| "grad_norm": 0.2576285203959204, | |
| "learning_rate": 7.749924831067401e-07, | |
| "loss": 0.6128, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.8883983321069415, | |
| "grad_norm": 0.5613002380897611, | |
| "learning_rate": 7.232310761112082e-07, | |
| "loss": 0.9456, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 2.8923227863625214, | |
| "grad_norm": 0.5378546190378084, | |
| "learning_rate": 6.732522184825896e-07, | |
| "loss": 0.4821, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 2.8962472406181017, | |
| "grad_norm": 0.4290000807899821, | |
| "learning_rate": 6.250568076088815e-07, | |
| "loss": 0.9524, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 2.9001716948736815, | |
| "grad_norm": 0.6642987661044457, | |
| "learning_rate": 5.78645708855674e-07, | |
| "loss": 0.7439, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 2.904096149129262, | |
| "grad_norm": 0.25434202409596207, | |
| "learning_rate": 5.340197555505966e-07, | |
| "loss": 0.5924, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.9080206033848417, | |
| "grad_norm": 0.5522817259850744, | |
| "learning_rate": 4.911797489683733e-07, | |
| "loss": 0.9662, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 2.911945057640422, | |
| "grad_norm": 0.5664174452510993, | |
| "learning_rate": 4.501264583164022e-07, | |
| "loss": 0.4759, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 2.915869511896002, | |
| "grad_norm": 0.39883060993706676, | |
| "learning_rate": 4.108606207209875e-07, | |
| "loss": 0.9605, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 2.919793966151582, | |
| "grad_norm": 0.6162016475029085, | |
| "learning_rate": 3.7338294121407323e-07, | |
| "loss": 0.7546, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 2.923718420407162, | |
| "grad_norm": 0.2551489877481498, | |
| "learning_rate": 3.376940927206196e-07, | |
| "loss": 0.585, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.9276428746627423, | |
| "grad_norm": 0.5596928859876541, | |
| "learning_rate": 3.0379471604645717e-07, | |
| "loss": 0.9388, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 2.931567328918322, | |
| "grad_norm": 0.562278783369554, | |
| "learning_rate": 2.716854198668517e-07, | |
| "loss": 0.481, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 2.9354917831739025, | |
| "grad_norm": 0.46071124182787754, | |
| "learning_rate": 2.41366780715524e-07, | |
| "loss": 0.9545, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 2.9394162374294823, | |
| "grad_norm": 0.6658807006109366, | |
| "learning_rate": 2.1283934297432473e-07, | |
| "loss": 0.7041, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 2.9433406916850626, | |
| "grad_norm": 0.2583340479643208, | |
| "learning_rate": 1.861036188634424e-07, | |
| "loss": 0.5884, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.9472651459406425, | |
| "grad_norm": 0.5310200026480655, | |
| "learning_rate": 1.6116008843224395e-07, | |
| "loss": 0.9831, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 2.951189600196223, | |
| "grad_norm": 0.5446447290659789, | |
| "learning_rate": 1.3800919955058167e-07, | |
| "loss": 0.4769, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 2.9551140544518026, | |
| "grad_norm": 0.44124202787318856, | |
| "learning_rate": 1.1665136790084407e-07, | |
| "loss": 0.9739, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 2.959038508707383, | |
| "grad_norm": 0.6227767656219008, | |
| "learning_rate": 9.708697697040636e-08, | |
| "loss": 0.7241, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 2.962962962962963, | |
| "grad_norm": 0.24729777368740563, | |
| "learning_rate": 7.931637804481362e-08, | |
| "loss": 0.595, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.966887417218543, | |
| "grad_norm": 0.5485826493056302, | |
| "learning_rate": 6.333989020143039e-08, | |
| "loss": 0.9474, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 2.9708118714741234, | |
| "grad_norm": 0.5661841602092826, | |
| "learning_rate": 4.9157800303722965e-08, | |
| "loss": 0.4827, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 2.9747363257297033, | |
| "grad_norm": 0.4278977573161211, | |
| "learning_rate": 3.6770362996108033e-08, | |
| "loss": 0.9621, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 2.978660779985283, | |
| "grad_norm": 0.6247650213188113, | |
| "learning_rate": 2.617780069940068e-08, | |
| "loss": 0.7432, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 2.9825852342408634, | |
| "grad_norm": 0.25222386274315406, | |
| "learning_rate": 1.738030360677323e-08, | |
| "loss": 0.5799, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.9865096884964437, | |
| "grad_norm": 0.5459466087116266, | |
| "learning_rate": 1.0378029680391255e-08, | |
| "loss": 0.9712, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 2.9904341427520236, | |
| "grad_norm": 0.5143187271731847, | |
| "learning_rate": 5.171104648549196e-09, | |
| "loss": 0.5131, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 2.9943585970076034, | |
| "grad_norm": 0.4936150427220661, | |
| "learning_rate": 1.7596220034277188e-09, | |
| "loss": 0.9508, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 2.9982830512631837, | |
| "grad_norm": 0.6054196084941783, | |
| "learning_rate": 1.436429993950661e-10, | |
| "loss": 0.6506, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 2.9998528329654155, | |
| "eval_loss": 0.824774444103241, | |
| "eval_runtime": 568.9573, | |
| "eval_samples_per_second": 43.94, | |
| "eval_steps_per_second": 5.493, | |
| "step": 7644 | |
| }, | |
| { | |
| "epoch": 2.9998528329654155, | |
| "step": 7644, | |
| "total_flos": 2930788677648384.0, | |
| "train_loss": 1.1571133552203934, | |
| "train_runtime": 44503.5033, | |
| "train_samples_per_second": 21.986, | |
| "train_steps_per_second": 0.172 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 7644, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2930788677648384.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |