| { |
| "best_global_step": 2000, |
| "best_metric": 0.8704838022863507, |
| "best_model_checkpoint": "./SALAMA_NEWMED9/checkpoint-2000", |
| "epoch": 2.0, |
| "eval_steps": 2000, |
| "global_step": 3338, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.00599250936329588, |
| "grad_norm": 3.4430904388427734, |
| "learning_rate": 1.8e-07, |
| "loss": 0.0305, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.01198501872659176, |
| "grad_norm": 1.3001453876495361, |
| "learning_rate": 3.8e-07, |
| "loss": 0.0217, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.017977528089887642, |
| "grad_norm": 2.323173999786377, |
| "learning_rate": 5.800000000000001e-07, |
| "loss": 0.0259, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.02397003745318352, |
| "grad_norm": 1.480966329574585, |
| "learning_rate": 7.8e-07, |
| "loss": 0.0274, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.0299625468164794, |
| "grad_norm": 0.8444674015045166, |
| "learning_rate": 9.800000000000001e-07, |
| "loss": 0.0204, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.035955056179775284, |
| "grad_norm": 4.492954254150391, |
| "learning_rate": 1.1800000000000001e-06, |
| "loss": 0.0362, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.041947565543071164, |
| "grad_norm": 0.8774360418319702, |
| "learning_rate": 1.3800000000000001e-06, |
| "loss": 0.0283, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.04794007490636704, |
| "grad_norm": 1.3773318529129028, |
| "learning_rate": 1.5800000000000001e-06, |
| "loss": 0.0255, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.05393258426966292, |
| "grad_norm": 1.632542610168457, |
| "learning_rate": 1.7800000000000001e-06, |
| "loss": 0.0186, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.0599250936329588, |
| "grad_norm": 2.4346158504486084, |
| "learning_rate": 1.98e-06, |
| "loss": 0.0311, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.06591760299625468, |
| "grad_norm": 1.4394844770431519, |
| "learning_rate": 2.1800000000000003e-06, |
| "loss": 0.0279, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.07191011235955057, |
| "grad_norm": 2.2031307220458984, |
| "learning_rate": 2.38e-06, |
| "loss": 0.0223, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.07790262172284644, |
| "grad_norm": 1.1016514301300049, |
| "learning_rate": 2.5800000000000003e-06, |
| "loss": 0.0159, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.08389513108614233, |
| "grad_norm": 2.271472454071045, |
| "learning_rate": 2.7800000000000005e-06, |
| "loss": 0.0187, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.0898876404494382, |
| "grad_norm": 1.3717912435531616, |
| "learning_rate": 2.9800000000000003e-06, |
| "loss": 0.0191, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.09588014981273409, |
| "grad_norm": 2.001311779022217, |
| "learning_rate": 3.1800000000000005e-06, |
| "loss": 0.0245, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.10187265917602996, |
| "grad_norm": 2.1671388149261475, |
| "learning_rate": 3.3800000000000007e-06, |
| "loss": 0.0207, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.10786516853932585, |
| "grad_norm": 2.7302298545837402, |
| "learning_rate": 3.58e-06, |
| "loss": 0.0251, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.11385767790262172, |
| "grad_norm": 2.360443115234375, |
| "learning_rate": 3.7800000000000002e-06, |
| "loss": 0.0279, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.1198501872659176, |
| "grad_norm": 2.1473894119262695, |
| "learning_rate": 3.980000000000001e-06, |
| "loss": 0.0166, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.1258426966292135, |
| "grad_norm": 2.1437275409698486, |
| "learning_rate": 4.18e-06, |
| "loss": 0.0296, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.13183520599250936, |
| "grad_norm": 1.155143141746521, |
| "learning_rate": 4.38e-06, |
| "loss": 0.0235, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.13782771535580524, |
| "grad_norm": 1.398034691810608, |
| "learning_rate": 4.58e-06, |
| "loss": 0.0256, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.14382022471910114, |
| "grad_norm": 1.704981803894043, |
| "learning_rate": 4.78e-06, |
| "loss": 0.0314, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.149812734082397, |
| "grad_norm": 2.0833070278167725, |
| "learning_rate": 4.980000000000001e-06, |
| "loss": 0.0292, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.15580524344569288, |
| "grad_norm": 1.6899584531784058, |
| "learning_rate": 5.18e-06, |
| "loss": 0.0226, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.16179775280898875, |
| "grad_norm": 1.5703374147415161, |
| "learning_rate": 5.380000000000001e-06, |
| "loss": 0.0251, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.16779026217228465, |
| "grad_norm": 0.9077961444854736, |
| "learning_rate": 5.580000000000001e-06, |
| "loss": 0.0193, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.17378277153558053, |
| "grad_norm": 2.0035812854766846, |
| "learning_rate": 5.78e-06, |
| "loss": 0.0231, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.1797752808988764, |
| "grad_norm": 2.8311734199523926, |
| "learning_rate": 5.98e-06, |
| "loss": 0.0326, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.18576779026217227, |
| "grad_norm": 1.6007928848266602, |
| "learning_rate": 6.18e-06, |
| "loss": 0.0231, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.19176029962546817, |
| "grad_norm": 1.3683286905288696, |
| "learning_rate": 6.380000000000001e-06, |
| "loss": 0.026, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.19775280898876405, |
| "grad_norm": 2.6587653160095215, |
| "learning_rate": 6.5800000000000005e-06, |
| "loss": 0.0288, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.20374531835205992, |
| "grad_norm": 2.617583751678467, |
| "learning_rate": 6.780000000000001e-06, |
| "loss": 0.0297, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.20973782771535582, |
| "grad_norm": 1.4291918277740479, |
| "learning_rate": 6.98e-06, |
| "loss": 0.0172, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2157303370786517, |
| "grad_norm": 1.9586079120635986, |
| "learning_rate": 7.180000000000001e-06, |
| "loss": 0.0413, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.22172284644194756, |
| "grad_norm": 2.081557273864746, |
| "learning_rate": 7.3800000000000005e-06, |
| "loss": 0.0297, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.22771535580524344, |
| "grad_norm": 1.5032098293304443, |
| "learning_rate": 7.58e-06, |
| "loss": 0.027, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.23370786516853934, |
| "grad_norm": 2.316767454147339, |
| "learning_rate": 7.78e-06, |
| "loss": 0.0291, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.2397003745318352, |
| "grad_norm": 3.565624475479126, |
| "learning_rate": 7.980000000000002e-06, |
| "loss": 0.0335, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.24569288389513108, |
| "grad_norm": 1.4883167743682861, |
| "learning_rate": 8.18e-06, |
| "loss": 0.0254, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.251685393258427, |
| "grad_norm": 2.2969090938568115, |
| "learning_rate": 8.380000000000001e-06, |
| "loss": 0.0401, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.25767790262172285, |
| "grad_norm": 1.9346531629562378, |
| "learning_rate": 8.580000000000001e-06, |
| "loss": 0.0254, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.2636704119850187, |
| "grad_norm": 0.6965954899787903, |
| "learning_rate": 8.78e-06, |
| "loss": 0.0248, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.2696629213483146, |
| "grad_norm": 3.862004041671753, |
| "learning_rate": 8.98e-06, |
| "loss": 0.0392, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.27565543071161047, |
| "grad_norm": 1.418031930923462, |
| "learning_rate": 9.180000000000002e-06, |
| "loss": 0.0364, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.28164794007490634, |
| "grad_norm": 1.2769050598144531, |
| "learning_rate": 9.38e-06, |
| "loss": 0.0239, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.2876404494382023, |
| "grad_norm": 2.03157901763916, |
| "learning_rate": 9.58e-06, |
| "loss": 0.0302, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.29363295880149815, |
| "grad_norm": 2.0723016262054443, |
| "learning_rate": 9.780000000000001e-06, |
| "loss": 0.0329, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.299625468164794, |
| "grad_norm": 1.585113763809204, |
| "learning_rate": 9.980000000000001e-06, |
| "loss": 0.0268, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.3056179775280899, |
| "grad_norm": 2.0149030685424805, |
| "learning_rate": 9.968287526427062e-06, |
| "loss": 0.0321, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.31161048689138576, |
| "grad_norm": 1.5547727346420288, |
| "learning_rate": 9.933051444679352e-06, |
| "loss": 0.0298, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.31760299625468164, |
| "grad_norm": 2.3159093856811523, |
| "learning_rate": 9.897815362931644e-06, |
| "loss": 0.0328, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.3235955056179775, |
| "grad_norm": 2.6754517555236816, |
| "learning_rate": 9.862579281183932e-06, |
| "loss": 0.0437, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.3295880149812734, |
| "grad_norm": 1.470944881439209, |
| "learning_rate": 9.827343199436224e-06, |
| "loss": 0.0324, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.3355805243445693, |
| "grad_norm": 1.461867094039917, |
| "learning_rate": 9.792107117688514e-06, |
| "loss": 0.0391, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.3415730337078652, |
| "grad_norm": 2.258100986480713, |
| "learning_rate": 9.756871035940804e-06, |
| "loss": 0.0382, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.34756554307116105, |
| "grad_norm": 2.2667694091796875, |
| "learning_rate": 9.721634954193094e-06, |
| "loss": 0.0393, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.3535580524344569, |
| "grad_norm": 2.1521716117858887, |
| "learning_rate": 9.686398872445385e-06, |
| "loss": 0.0242, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.3595505617977528, |
| "grad_norm": 2.3627607822418213, |
| "learning_rate": 9.651162790697676e-06, |
| "loss": 0.0303, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.36554307116104867, |
| "grad_norm": 2.5504517555236816, |
| "learning_rate": 9.615926708949965e-06, |
| "loss": 0.0384, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.37153558052434454, |
| "grad_norm": 2.1967527866363525, |
| "learning_rate": 9.580690627202257e-06, |
| "loss": 0.033, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.3775280898876405, |
| "grad_norm": 1.6752692461013794, |
| "learning_rate": 9.545454545454547e-06, |
| "loss": 0.0246, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.38352059925093634, |
| "grad_norm": 1.6692609786987305, |
| "learning_rate": 9.510218463706837e-06, |
| "loss": 0.0339, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.3895131086142322, |
| "grad_norm": 2.7334742546081543, |
| "learning_rate": 9.474982381959127e-06, |
| "loss": 0.0276, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.3955056179775281, |
| "grad_norm": 1.5142534971237183, |
| "learning_rate": 9.439746300211417e-06, |
| "loss": 0.0347, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.40149812734082396, |
| "grad_norm": 2.5328900814056396, |
| "learning_rate": 9.404510218463707e-06, |
| "loss": 0.0295, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.40749063670411984, |
| "grad_norm": 3.101393938064575, |
| "learning_rate": 9.369274136715997e-06, |
| "loss": 0.0409, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.4134831460674157, |
| "grad_norm": 2.7148630619049072, |
| "learning_rate": 9.33403805496829e-06, |
| "loss": 0.0311, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.41947565543071164, |
| "grad_norm": 1.6952073574066162, |
| "learning_rate": 9.298801973220578e-06, |
| "loss": 0.0341, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.4254681647940075, |
| "grad_norm": 2.545931577682495, |
| "learning_rate": 9.26356589147287e-06, |
| "loss": 0.036, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.4314606741573034, |
| "grad_norm": 2.246628999710083, |
| "learning_rate": 9.22832980972516e-06, |
| "loss": 0.0424, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.43745318352059925, |
| "grad_norm": 1.6698744297027588, |
| "learning_rate": 9.19309372797745e-06, |
| "loss": 0.0416, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.4434456928838951, |
| "grad_norm": 1.2881724834442139, |
| "learning_rate": 9.15785764622974e-06, |
| "loss": 0.0366, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.449438202247191, |
| "grad_norm": 1.7771434783935547, |
| "learning_rate": 9.12262156448203e-06, |
| "loss": 0.0309, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.45543071161048687, |
| "grad_norm": 1.563678503036499, |
| "learning_rate": 9.087385482734322e-06, |
| "loss": 0.0376, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.46142322097378274, |
| "grad_norm": 2.24344801902771, |
| "learning_rate": 9.05214940098661e-06, |
| "loss": 0.0332, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.46741573033707867, |
| "grad_norm": 3.014191150665283, |
| "learning_rate": 9.016913319238902e-06, |
| "loss": 0.0288, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.47340823970037454, |
| "grad_norm": 2.138899564743042, |
| "learning_rate": 8.981677237491192e-06, |
| "loss": 0.0332, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.4794007490636704, |
| "grad_norm": 2.4274239540100098, |
| "learning_rate": 8.946441155743482e-06, |
| "loss": 0.0426, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.4853932584269663, |
| "grad_norm": 1.720563292503357, |
| "learning_rate": 8.911205073995772e-06, |
| "loss": 0.0322, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.49138576779026216, |
| "grad_norm": 2.580158233642578, |
| "learning_rate": 8.875968992248062e-06, |
| "loss": 0.0337, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.49737827715355803, |
| "grad_norm": 1.918359398841858, |
| "learning_rate": 8.840732910500353e-06, |
| "loss": 0.0276, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.503370786516854, |
| "grad_norm": 3.0238075256347656, |
| "learning_rate": 8.805496828752643e-06, |
| "loss": 0.0363, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.5093632958801498, |
| "grad_norm": 1.9104974269866943, |
| "learning_rate": 8.770260747004935e-06, |
| "loss": 0.0372, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.5153558052434457, |
| "grad_norm": 4.188656806945801, |
| "learning_rate": 8.735024665257225e-06, |
| "loss": 0.0374, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.5213483146067416, |
| "grad_norm": 2.6487040519714355, |
| "learning_rate": 8.699788583509515e-06, |
| "loss": 0.0305, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.5273408239700375, |
| "grad_norm": 2.9171149730682373, |
| "learning_rate": 8.664552501761805e-06, |
| "loss": 0.0321, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.5333333333333333, |
| "grad_norm": 2.196499824523926, |
| "learning_rate": 8.629316420014095e-06, |
| "loss": 0.0497, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.5393258426966292, |
| "grad_norm": 2.324110507965088, |
| "learning_rate": 8.594080338266385e-06, |
| "loss": 0.0411, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.5453183520599251, |
| "grad_norm": 2.23406720161438, |
| "learning_rate": 8.558844256518675e-06, |
| "loss": 0.0302, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.5513108614232209, |
| "grad_norm": 2.158843994140625, |
| "learning_rate": 8.523608174770967e-06, |
| "loss": 0.0349, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.5573033707865168, |
| "grad_norm": 2.5448203086853027, |
| "learning_rate": 8.488372093023256e-06, |
| "loss": 0.0338, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.5632958801498127, |
| "grad_norm": 2.366697072982788, |
| "learning_rate": 8.453136011275547e-06, |
| "loss": 0.0307, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.5692883895131086, |
| "grad_norm": 1.7669707536697388, |
| "learning_rate": 8.417899929527837e-06, |
| "loss": 0.0374, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.5752808988764045, |
| "grad_norm": 1.8498706817626953, |
| "learning_rate": 8.382663847780128e-06, |
| "loss": 0.0368, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.5812734082397004, |
| "grad_norm": 1.1811097860336304, |
| "learning_rate": 8.347427766032418e-06, |
| "loss": 0.0319, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.5872659176029963, |
| "grad_norm": 1.306626796722412, |
| "learning_rate": 8.312191684284708e-06, |
| "loss": 0.0409, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.5932584269662922, |
| "grad_norm": 1.9113121032714844, |
| "learning_rate": 8.276955602537e-06, |
| "loss": 0.0329, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.599250936329588, |
| "grad_norm": 1.8065577745437622, |
| "learning_rate": 8.241719520789288e-06, |
| "loss": 0.0261, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.6052434456928839, |
| "grad_norm": 1.644216775894165, |
| "learning_rate": 8.20648343904158e-06, |
| "loss": 0.0455, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.6112359550561798, |
| "grad_norm": 2.524791955947876, |
| "learning_rate": 8.17124735729387e-06, |
| "loss": 0.0378, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.6172284644194757, |
| "grad_norm": 4.8924880027771, |
| "learning_rate": 8.13601127554616e-06, |
| "loss": 0.0458, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.6232209737827715, |
| "grad_norm": 2.6112005710601807, |
| "learning_rate": 8.10077519379845e-06, |
| "loss": 0.0388, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.6292134831460674, |
| "grad_norm": 2.2493460178375244, |
| "learning_rate": 8.06553911205074e-06, |
| "loss": 0.0451, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.6352059925093633, |
| "grad_norm": 1.854641079902649, |
| "learning_rate": 8.03030303030303e-06, |
| "loss": 0.0434, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.6411985018726591, |
| "grad_norm": 2.749845027923584, |
| "learning_rate": 7.99506694855532e-06, |
| "loss": 0.0291, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.647191011235955, |
| "grad_norm": 3.08630108833313, |
| "learning_rate": 7.959830866807612e-06, |
| "loss": 0.0256, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.6531835205992509, |
| "grad_norm": 1.4188586473464966, |
| "learning_rate": 7.924594785059903e-06, |
| "loss": 0.0363, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.6591760299625468, |
| "grad_norm": 1.34322988986969, |
| "learning_rate": 7.889358703312193e-06, |
| "loss": 0.0313, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.6651685393258427, |
| "grad_norm": 1.7974929809570312, |
| "learning_rate": 7.854122621564483e-06, |
| "loss": 0.0245, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.6711610486891386, |
| "grad_norm": 2.5316739082336426, |
| "learning_rate": 7.818886539816773e-06, |
| "loss": 0.0355, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.6771535580524345, |
| "grad_norm": 3.0023632049560547, |
| "learning_rate": 7.783650458069063e-06, |
| "loss": 0.0344, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.6831460674157304, |
| "grad_norm": 2.484605550765991, |
| "learning_rate": 7.748414376321353e-06, |
| "loss": 0.0442, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.6891385767790262, |
| "grad_norm": 2.7692389488220215, |
| "learning_rate": 7.713178294573645e-06, |
| "loss": 0.0389, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.6951310861423221, |
| "grad_norm": 1.1416544914245605, |
| "learning_rate": 7.677942212825933e-06, |
| "loss": 0.0357, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.701123595505618, |
| "grad_norm": 2.1747801303863525, |
| "learning_rate": 7.642706131078225e-06, |
| "loss": 0.0329, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.7071161048689139, |
| "grad_norm": 2.4001238346099854, |
| "learning_rate": 7.607470049330515e-06, |
| "loss": 0.0327, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.7131086142322097, |
| "grad_norm": 2.489826202392578, |
| "learning_rate": 7.572233967582805e-06, |
| "loss": 0.038, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.7191011235955056, |
| "grad_norm": 2.055882453918457, |
| "learning_rate": 7.5369978858350956e-06, |
| "loss": 0.0268, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.7250936329588015, |
| "grad_norm": 1.705216646194458, |
| "learning_rate": 7.5017618040873865e-06, |
| "loss": 0.0258, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.7310861423220973, |
| "grad_norm": 1.2729088068008423, |
| "learning_rate": 7.466525722339677e-06, |
| "loss": 0.0352, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.7370786516853932, |
| "grad_norm": 4.454958438873291, |
| "learning_rate": 7.431289640591967e-06, |
| "loss": 0.0343, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.7430711610486891, |
| "grad_norm": 1.8376761674880981, |
| "learning_rate": 7.396053558844257e-06, |
| "loss": 0.0337, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.7490636704119851, |
| "grad_norm": 2.3845252990722656, |
| "learning_rate": 7.360817477096548e-06, |
| "loss": 0.0406, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.755056179775281, |
| "grad_norm": 6.822844982147217, |
| "learning_rate": 7.325581395348837e-06, |
| "loss": 0.0417, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.7610486891385768, |
| "grad_norm": 7.371566295623779, |
| "learning_rate": 7.290345313601128e-06, |
| "loss": 0.0271, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.7670411985018727, |
| "grad_norm": 1.3738433122634888, |
| "learning_rate": 7.255109231853419e-06, |
| "loss": 0.0296, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.7730337078651686, |
| "grad_norm": 1.8998222351074219, |
| "learning_rate": 7.219873150105708e-06, |
| "loss": 0.0317, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.7790262172284644, |
| "grad_norm": 2.569655418395996, |
| "learning_rate": 7.184637068357999e-06, |
| "loss": 0.0305, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.7850187265917603, |
| "grad_norm": 1.7254374027252197, |
| "learning_rate": 7.1494009866102894e-06, |
| "loss": 0.0327, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.7910112359550562, |
| "grad_norm": 2.161518096923828, |
| "learning_rate": 7.1141649048625796e-06, |
| "loss": 0.0329, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.797003745318352, |
| "grad_norm": 1.4740070104599, |
| "learning_rate": 7.07892882311487e-06, |
| "loss": 0.0346, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.8029962546816479, |
| "grad_norm": 2.4593379497528076, |
| "learning_rate": 7.043692741367161e-06, |
| "loss": 0.0267, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.8089887640449438, |
| "grad_norm": 1.8585057258605957, |
| "learning_rate": 7.008456659619452e-06, |
| "loss": 0.0262, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.8149812734082397, |
| "grad_norm": 2.7434608936309814, |
| "learning_rate": 6.973220577871741e-06, |
| "loss": 0.039, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.8209737827715355, |
| "grad_norm": 2.6166908740997314, |
| "learning_rate": 6.937984496124032e-06, |
| "loss": 0.0346, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.8269662921348314, |
| "grad_norm": 1.4991202354431152, |
| "learning_rate": 6.902748414376322e-06, |
| "loss": 0.0355, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.8329588014981273, |
| "grad_norm": 1.4986398220062256, |
| "learning_rate": 6.867512332628612e-06, |
| "loss": 0.0332, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.8389513108614233, |
| "grad_norm": 2.284785509109497, |
| "learning_rate": 6.832276250880902e-06, |
| "loss": 0.0301, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.8449438202247191, |
| "grad_norm": 1.3910906314849854, |
| "learning_rate": 6.797040169133193e-06, |
| "loss": 0.024, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.850936329588015, |
| "grad_norm": 2.0242066383361816, |
| "learning_rate": 6.7618040873854825e-06, |
| "loss": 0.0359, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.8569288389513109, |
| "grad_norm": 2.430185317993164, |
| "learning_rate": 6.7265680056377734e-06, |
| "loss": 0.0226, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.8629213483146068, |
| "grad_norm": 0.8888330459594727, |
| "learning_rate": 6.691331923890064e-06, |
| "loss": 0.0289, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.8689138576779026, |
| "grad_norm": 1.8304884433746338, |
| "learning_rate": 6.656095842142354e-06, |
| "loss": 0.0238, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.8749063670411985, |
| "grad_norm": 2.193533182144165, |
| "learning_rate": 6.620859760394645e-06, |
| "loss": 0.0385, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.8808988764044944, |
| "grad_norm": 2.304433822631836, |
| "learning_rate": 6.585623678646935e-06, |
| "loss": 0.0341, |
| "step": 1470 |
| }, |
| { |
| "epoch": 0.8868913857677903, |
| "grad_norm": 1.412208080291748, |
| "learning_rate": 6.550387596899226e-06, |
| "loss": 0.0262, |
| "step": 1480 |
| }, |
| { |
| "epoch": 0.8928838951310861, |
| "grad_norm": 0.917147696018219, |
| "learning_rate": 6.515151515151516e-06, |
| "loss": 0.0219, |
| "step": 1490 |
| }, |
| { |
| "epoch": 0.898876404494382, |
| "grad_norm": 1.979040265083313, |
| "learning_rate": 6.479915433403806e-06, |
| "loss": 0.0396, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.9048689138576779, |
| "grad_norm": 1.5810662508010864, |
| "learning_rate": 6.444679351656097e-06, |
| "loss": 0.0274, |
| "step": 1510 |
| }, |
| { |
| "epoch": 0.9108614232209737, |
| "grad_norm": 2.405066728591919, |
| "learning_rate": 6.409443269908386e-06, |
| "loss": 0.0388, |
| "step": 1520 |
| }, |
| { |
| "epoch": 0.9168539325842696, |
| "grad_norm": 2.646991014480591, |
| "learning_rate": 6.374207188160677e-06, |
| "loss": 0.0392, |
| "step": 1530 |
| }, |
| { |
| "epoch": 0.9228464419475655, |
| "grad_norm": 2.1389219760894775, |
| "learning_rate": 6.338971106412968e-06, |
| "loss": 0.0345, |
| "step": 1540 |
| }, |
| { |
| "epoch": 0.9288389513108615, |
| "grad_norm": 1.683204174041748, |
| "learning_rate": 6.3037350246652574e-06, |
| "loss": 0.0354, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.9348314606741573, |
| "grad_norm": 2.031403064727783, |
| "learning_rate": 6.268498942917548e-06, |
| "loss": 0.0263, |
| "step": 1560 |
| }, |
| { |
| "epoch": 0.9408239700374532, |
| "grad_norm": 2.195925712585449, |
| "learning_rate": 6.2332628611698385e-06, |
| "loss": 0.028, |
| "step": 1570 |
| }, |
| { |
| "epoch": 0.9468164794007491, |
| "grad_norm": 1.9090993404388428, |
| "learning_rate": 6.198026779422129e-06, |
| "loss": 0.0339, |
| "step": 1580 |
| }, |
| { |
| "epoch": 0.952808988764045, |
| "grad_norm": 2.7698168754577637, |
| "learning_rate": 6.162790697674419e-06, |
| "loss": 0.0412, |
| "step": 1590 |
| }, |
| { |
| "epoch": 0.9588014981273408, |
| "grad_norm": 2.477466106414795, |
| "learning_rate": 6.12755461592671e-06, |
| "loss": 0.0307, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.9647940074906367, |
| "grad_norm": 1.7105556726455688, |
| "learning_rate": 6.092318534179001e-06, |
| "loss": 0.0378, |
| "step": 1610 |
| }, |
| { |
| "epoch": 0.9707865168539326, |
| "grad_norm": 1.9348987340927124, |
| "learning_rate": 6.05708245243129e-06, |
| "loss": 0.0312, |
| "step": 1620 |
| }, |
| { |
| "epoch": 0.9767790262172285, |
| "grad_norm": 2.8063669204711914, |
| "learning_rate": 6.021846370683581e-06, |
| "loss": 0.0321, |
| "step": 1630 |
| }, |
| { |
| "epoch": 0.9827715355805243, |
| "grad_norm": 1.3787035942077637, |
| "learning_rate": 5.986610288935871e-06, |
| "loss": 0.0204, |
| "step": 1640 |
| }, |
| { |
| "epoch": 0.9887640449438202, |
| "grad_norm": 1.0348024368286133, |
| "learning_rate": 5.951374207188161e-06, |
| "loss": 0.0269, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.9947565543071161, |
| "grad_norm": 2.3629918098449707, |
| "learning_rate": 5.916138125440451e-06, |
| "loss": 0.0289, |
| "step": 1660 |
| }, |
| { |
| "epoch": 1.0005992509363295, |
| "grad_norm": 1.635030746459961, |
| "learning_rate": 5.880902043692742e-06, |
| "loss": 0.0271, |
| "step": 1670 |
| }, |
| { |
| "epoch": 1.0065917602996255, |
| "grad_norm": 1.2064359188079834, |
| "learning_rate": 5.8456659619450316e-06, |
| "loss": 0.0068, |
| "step": 1680 |
| }, |
| { |
| "epoch": 1.0125842696629213, |
| "grad_norm": 1.9273971319198608, |
| "learning_rate": 5.8104298801973225e-06, |
| "loss": 0.0074, |
| "step": 1690 |
| }, |
| { |
| "epoch": 1.0185767790262172, |
| "grad_norm": 1.0188418626785278, |
| "learning_rate": 5.7751937984496135e-06, |
| "loss": 0.0124, |
| "step": 1700 |
| }, |
| { |
| "epoch": 1.024569288389513, |
| "grad_norm": 0.795347273349762, |
| "learning_rate": 5.739957716701903e-06, |
| "loss": 0.0102, |
| "step": 1710 |
| }, |
| { |
| "epoch": 1.030561797752809, |
| "grad_norm": 0.6345170736312866, |
| "learning_rate": 5.704721634954194e-06, |
| "loss": 0.009, |
| "step": 1720 |
| }, |
| { |
| "epoch": 1.036554307116105, |
| "grad_norm": 0.6679102778434753, |
| "learning_rate": 5.669485553206484e-06, |
| "loss": 0.0086, |
| "step": 1730 |
| }, |
| { |
| "epoch": 1.0425468164794007, |
| "grad_norm": 2.5230817794799805, |
| "learning_rate": 5.634249471458775e-06, |
| "loss": 0.0076, |
| "step": 1740 |
| }, |
| { |
| "epoch": 1.0485393258426967, |
| "grad_norm": 0.9390186667442322, |
| "learning_rate": 5.599013389711064e-06, |
| "loss": 0.0056, |
| "step": 1750 |
| }, |
| { |
| "epoch": 1.0545318352059925, |
| "grad_norm": 1.2290983200073242, |
| "learning_rate": 5.563777307963355e-06, |
| "loss": 0.0067, |
| "step": 1760 |
| }, |
| { |
| "epoch": 1.0605243445692885, |
| "grad_norm": 1.8171833753585815, |
| "learning_rate": 5.528541226215646e-06, |
| "loss": 0.009, |
| "step": 1770 |
| }, |
| { |
| "epoch": 1.0665168539325842, |
| "grad_norm": 1.3471766710281372, |
| "learning_rate": 5.493305144467935e-06, |
| "loss": 0.0061, |
| "step": 1780 |
| }, |
| { |
| "epoch": 1.0725093632958802, |
| "grad_norm": 0.8515785336494446, |
| "learning_rate": 5.458069062720226e-06, |
| "loss": 0.0059, |
| "step": 1790 |
| }, |
| { |
| "epoch": 1.078501872659176, |
| "grad_norm": 1.4226453304290771, |
| "learning_rate": 5.422832980972516e-06, |
| "loss": 0.0098, |
| "step": 1800 |
| }, |
| { |
| "epoch": 1.084494382022472, |
| "grad_norm": 2.040360927581787, |
| "learning_rate": 5.3875968992248065e-06, |
| "loss": 0.009, |
| "step": 1810 |
| }, |
| { |
| "epoch": 1.0904868913857677, |
| "grad_norm": 1.0943787097930908, |
| "learning_rate": 5.352360817477097e-06, |
| "loss": 0.0107, |
| "step": 1820 |
| }, |
| { |
| "epoch": 1.0964794007490637, |
| "grad_norm": 2.117577314376831, |
| "learning_rate": 5.317124735729388e-06, |
| "loss": 0.0082, |
| "step": 1830 |
| }, |
| { |
| "epoch": 1.1024719101123595, |
| "grad_norm": 1.5584527254104614, |
| "learning_rate": 5.281888653981677e-06, |
| "loss": 0.0058, |
| "step": 1840 |
| }, |
| { |
| "epoch": 1.1084644194756554, |
| "grad_norm": 1.2435089349746704, |
| "learning_rate": 5.246652572233968e-06, |
| "loss": 0.0073, |
| "step": 1850 |
| }, |
| { |
| "epoch": 1.1144569288389512, |
| "grad_norm": 1.123214602470398, |
| "learning_rate": 5.211416490486259e-06, |
| "loss": 0.0081, |
| "step": 1860 |
| }, |
| { |
| "epoch": 1.1204494382022472, |
| "grad_norm": 0.5520896315574646, |
| "learning_rate": 5.176180408738549e-06, |
| "loss": 0.009, |
| "step": 1870 |
| }, |
| { |
| "epoch": 1.1264419475655432, |
| "grad_norm": 1.1572812795639038, |
| "learning_rate": 5.140944326990839e-06, |
| "loss": 0.0062, |
| "step": 1880 |
| }, |
| { |
| "epoch": 1.132434456928839, |
| "grad_norm": 1.0478726625442505, |
| "learning_rate": 5.105708245243129e-06, |
| "loss": 0.0057, |
| "step": 1890 |
| }, |
| { |
| "epoch": 1.138426966292135, |
| "grad_norm": 0.7916937470436096, |
| "learning_rate": 5.07047216349542e-06, |
| "loss": 0.0068, |
| "step": 1900 |
| }, |
| { |
| "epoch": 1.1444194756554307, |
| "grad_norm": 0.804720401763916, |
| "learning_rate": 5.0352360817477094e-06, |
| "loss": 0.0052, |
| "step": 1910 |
| }, |
| { |
| "epoch": 1.1504119850187267, |
| "grad_norm": 0.8383046984672546, |
| "learning_rate": 5e-06, |
| "loss": 0.0103, |
| "step": 1920 |
| }, |
| { |
| "epoch": 1.1564044943820224, |
| "grad_norm": 0.44970327615737915, |
| "learning_rate": 4.9647639182522905e-06, |
| "loss": 0.0086, |
| "step": 1930 |
| }, |
| { |
| "epoch": 1.1623970037453184, |
| "grad_norm": 1.1988434791564941, |
| "learning_rate": 4.9295278365045815e-06, |
| "loss": 0.0091, |
| "step": 1940 |
| }, |
| { |
| "epoch": 1.1683895131086142, |
| "grad_norm": 0.7889288067817688, |
| "learning_rate": 4.894291754756872e-06, |
| "loss": 0.0057, |
| "step": 1950 |
| }, |
| { |
| "epoch": 1.1743820224719101, |
| "grad_norm": 0.5470920205116272, |
| "learning_rate": 4.859055673009162e-06, |
| "loss": 0.0072, |
| "step": 1960 |
| }, |
| { |
| "epoch": 1.180374531835206, |
| "grad_norm": 1.7936517000198364, |
| "learning_rate": 4.823819591261452e-06, |
| "loss": 0.0048, |
| "step": 1970 |
| }, |
| { |
| "epoch": 1.186367041198502, |
| "grad_norm": 1.2022855281829834, |
| "learning_rate": 4.788583509513742e-06, |
| "loss": 0.0066, |
| "step": 1980 |
| }, |
| { |
| "epoch": 1.1923595505617977, |
| "grad_norm": 0.5040098428726196, |
| "learning_rate": 4.753347427766033e-06, |
| "loss": 0.0067, |
| "step": 1990 |
| }, |
| { |
| "epoch": 1.1983520599250936, |
| "grad_norm": 0.6506677269935608, |
| "learning_rate": 4.718111346018323e-06, |
| "loss": 0.0045, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.1983520599250936, |
| "eval_loss": 0.011116987094283104, |
| "eval_runtime": 9693.0647, |
| "eval_samples_per_second": 1.377, |
| "eval_steps_per_second": 0.172, |
| "eval_wer": 0.8704838022863507, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.2043445692883896, |
| "grad_norm": 1.1864808797836304, |
| "learning_rate": 4.682875264270613e-06, |
| "loss": 0.0053, |
| "step": 2010 |
| }, |
| { |
| "epoch": 1.2103370786516854, |
| "grad_norm": 0.9299235939979553, |
| "learning_rate": 4.647639182522904e-06, |
| "loss": 0.0072, |
| "step": 2020 |
| }, |
| { |
| "epoch": 1.2163295880149814, |
| "grad_norm": 1.0193731784820557, |
| "learning_rate": 4.612403100775194e-06, |
| "loss": 0.0088, |
| "step": 2030 |
| }, |
| { |
| "epoch": 1.2223220973782771, |
| "grad_norm": 0.8602034449577332, |
| "learning_rate": 4.577167019027484e-06, |
| "loss": 0.0071, |
| "step": 2040 |
| }, |
| { |
| "epoch": 1.2283146067415731, |
| "grad_norm": 0.3457139730453491, |
| "learning_rate": 4.5419309372797745e-06, |
| "loss": 0.0091, |
| "step": 2050 |
| }, |
| { |
| "epoch": 1.2343071161048689, |
| "grad_norm": 2.478890895843506, |
| "learning_rate": 4.506694855532065e-06, |
| "loss": 0.0118, |
| "step": 2060 |
| }, |
| { |
| "epoch": 1.2402996254681649, |
| "grad_norm": 0.6306349039077759, |
| "learning_rate": 4.471458773784356e-06, |
| "loss": 0.0128, |
| "step": 2070 |
| }, |
| { |
| "epoch": 1.2462921348314606, |
| "grad_norm": 1.3818278312683105, |
| "learning_rate": 4.436222692036646e-06, |
| "loss": 0.0048, |
| "step": 2080 |
| }, |
| { |
| "epoch": 1.2522846441947566, |
| "grad_norm": 1.1697428226470947, |
| "learning_rate": 4.400986610288937e-06, |
| "loss": 0.0084, |
| "step": 2090 |
| }, |
| { |
| "epoch": 1.2582771535580524, |
| "grad_norm": 1.0534144639968872, |
| "learning_rate": 4.365750528541227e-06, |
| "loss": 0.0072, |
| "step": 2100 |
| }, |
| { |
| "epoch": 1.2642696629213483, |
| "grad_norm": 0.4338432550430298, |
| "learning_rate": 4.330514446793517e-06, |
| "loss": 0.008, |
| "step": 2110 |
| }, |
| { |
| "epoch": 1.270262172284644, |
| "grad_norm": 0.39000409841537476, |
| "learning_rate": 4.295278365045807e-06, |
| "loss": 0.0054, |
| "step": 2120 |
| }, |
| { |
| "epoch": 1.27625468164794, |
| "grad_norm": 2.452831745147705, |
| "learning_rate": 4.260042283298097e-06, |
| "loss": 0.0083, |
| "step": 2130 |
| }, |
| { |
| "epoch": 1.282247191011236, |
| "grad_norm": 1.2903640270233154, |
| "learning_rate": 4.224806201550387e-06, |
| "loss": 0.0062, |
| "step": 2140 |
| }, |
| { |
| "epoch": 1.2882397003745318, |
| "grad_norm": 0.5914607644081116, |
| "learning_rate": 4.189570119802678e-06, |
| "loss": 0.0087, |
| "step": 2150 |
| }, |
| { |
| "epoch": 1.2942322097378276, |
| "grad_norm": 0.9648634195327759, |
| "learning_rate": 4.154334038054968e-06, |
| "loss": 0.008, |
| "step": 2160 |
| }, |
| { |
| "epoch": 1.3002247191011236, |
| "grad_norm": 0.7724722623825073, |
| "learning_rate": 4.119097956307259e-06, |
| "loss": 0.0078, |
| "step": 2170 |
| }, |
| { |
| "epoch": 1.3062172284644196, |
| "grad_norm": 1.0104018449783325, |
| "learning_rate": 4.0838618745595495e-06, |
| "loss": 0.0063, |
| "step": 2180 |
| }, |
| { |
| "epoch": 1.3122097378277153, |
| "grad_norm": 1.2675261497497559, |
| "learning_rate": 4.04862579281184e-06, |
| "loss": 0.0085, |
| "step": 2190 |
| }, |
| { |
| "epoch": 1.3182022471910113, |
| "grad_norm": 2.127049446105957, |
| "learning_rate": 4.01338971106413e-06, |
| "loss": 0.0075, |
| "step": 2200 |
| }, |
| { |
| "epoch": 1.324194756554307, |
| "grad_norm": 0.3128371834754944, |
| "learning_rate": 3.97815362931642e-06, |
| "loss": 0.0067, |
| "step": 2210 |
| }, |
| { |
| "epoch": 1.330187265917603, |
| "grad_norm": 0.8424582481384277, |
| "learning_rate": 3.942917547568711e-06, |
| "loss": 0.0087, |
| "step": 2220 |
| }, |
| { |
| "epoch": 1.3361797752808988, |
| "grad_norm": 0.6332743763923645, |
| "learning_rate": 3.907681465821001e-06, |
| "loss": 0.0059, |
| "step": 2230 |
| }, |
| { |
| "epoch": 1.3421722846441948, |
| "grad_norm": 0.41060033440589905, |
| "learning_rate": 3.872445384073291e-06, |
| "loss": 0.0046, |
| "step": 2240 |
| }, |
| { |
| "epoch": 1.3481647940074906, |
| "grad_norm": 1.4038411378860474, |
| "learning_rate": 3.837209302325582e-06, |
| "loss": 0.0073, |
| "step": 2250 |
| }, |
| { |
| "epoch": 1.3541573033707865, |
| "grad_norm": 1.3935422897338867, |
| "learning_rate": 3.801973220577872e-06, |
| "loss": 0.008, |
| "step": 2260 |
| }, |
| { |
| "epoch": 1.3601498127340825, |
| "grad_norm": 0.430426687002182, |
| "learning_rate": 3.7667371388301623e-06, |
| "loss": 0.0069, |
| "step": 2270 |
| }, |
| { |
| "epoch": 1.3661423220973783, |
| "grad_norm": 1.468438982963562, |
| "learning_rate": 3.731501057082453e-06, |
| "loss": 0.0071, |
| "step": 2280 |
| }, |
| { |
| "epoch": 1.372134831460674, |
| "grad_norm": 0.5553675293922424, |
| "learning_rate": 3.696264975334743e-06, |
| "loss": 0.0068, |
| "step": 2290 |
| }, |
| { |
| "epoch": 1.37812734082397, |
| "grad_norm": 0.6449081897735596, |
| "learning_rate": 3.6610288935870335e-06, |
| "loss": 0.0091, |
| "step": 2300 |
| }, |
| { |
| "epoch": 1.384119850187266, |
| "grad_norm": 1.5938137769699097, |
| "learning_rate": 3.6257928118393236e-06, |
| "loss": 0.0089, |
| "step": 2310 |
| }, |
| { |
| "epoch": 1.3901123595505618, |
| "grad_norm": 0.6306646466255188, |
| "learning_rate": 3.5905567300916137e-06, |
| "loss": 0.0042, |
| "step": 2320 |
| }, |
| { |
| "epoch": 1.3961048689138575, |
| "grad_norm": 0.5056957006454468, |
| "learning_rate": 3.5553206483439047e-06, |
| "loss": 0.0046, |
| "step": 2330 |
| }, |
| { |
| "epoch": 1.4020973782771535, |
| "grad_norm": 0.8323089480400085, |
| "learning_rate": 3.520084566596195e-06, |
| "loss": 0.0047, |
| "step": 2340 |
| }, |
| { |
| "epoch": 1.4080898876404495, |
| "grad_norm": 0.9807717204093933, |
| "learning_rate": 3.4848484848484854e-06, |
| "loss": 0.007, |
| "step": 2350 |
| }, |
| { |
| "epoch": 1.4140823970037453, |
| "grad_norm": 0.182850182056427, |
| "learning_rate": 3.4496124031007755e-06, |
| "loss": 0.0079, |
| "step": 2360 |
| }, |
| { |
| "epoch": 1.4200749063670413, |
| "grad_norm": 0.37714436650276184, |
| "learning_rate": 3.4143763213530656e-06, |
| "loss": 0.0052, |
| "step": 2370 |
| }, |
| { |
| "epoch": 1.426067415730337, |
| "grad_norm": 0.9423696398735046, |
| "learning_rate": 3.379140239605356e-06, |
| "loss": 0.0104, |
| "step": 2380 |
| }, |
| { |
| "epoch": 1.432059925093633, |
| "grad_norm": 0.8269687294960022, |
| "learning_rate": 3.3439041578576463e-06, |
| "loss": 0.0068, |
| "step": 2390 |
| }, |
| { |
| "epoch": 1.4380524344569288, |
| "grad_norm": 1.2370884418487549, |
| "learning_rate": 3.3086680761099373e-06, |
| "loss": 0.0064, |
| "step": 2400 |
| }, |
| { |
| "epoch": 1.4440449438202247, |
| "grad_norm": 1.2782443761825562, |
| "learning_rate": 3.2734319943622274e-06, |
| "loss": 0.007, |
| "step": 2410 |
| }, |
| { |
| "epoch": 1.4500374531835205, |
| "grad_norm": 0.65436190366745, |
| "learning_rate": 3.2381959126145175e-06, |
| "loss": 0.0093, |
| "step": 2420 |
| }, |
| { |
| "epoch": 1.4560299625468165, |
| "grad_norm": 0.42517855763435364, |
| "learning_rate": 3.202959830866808e-06, |
| "loss": 0.0071, |
| "step": 2430 |
| }, |
| { |
| "epoch": 1.4620224719101125, |
| "grad_norm": 1.214666724205017, |
| "learning_rate": 3.167723749119098e-06, |
| "loss": 0.0073, |
| "step": 2440 |
| }, |
| { |
| "epoch": 1.4680149812734082, |
| "grad_norm": 1.4744590520858765, |
| "learning_rate": 3.1324876673713883e-06, |
| "loss": 0.0065, |
| "step": 2450 |
| }, |
| { |
| "epoch": 1.474007490636704, |
| "grad_norm": 1.2045056819915771, |
| "learning_rate": 3.097251585623679e-06, |
| "loss": 0.0094, |
| "step": 2460 |
| }, |
| { |
| "epoch": 1.48, |
| "grad_norm": 0.7647907137870789, |
| "learning_rate": 3.062015503875969e-06, |
| "loss": 0.0082, |
| "step": 2470 |
| }, |
| { |
| "epoch": 1.485992509363296, |
| "grad_norm": 0.8723673820495605, |
| "learning_rate": 3.02677942212826e-06, |
| "loss": 0.0053, |
| "step": 2480 |
| }, |
| { |
| "epoch": 1.4919850187265917, |
| "grad_norm": 0.8856358528137207, |
| "learning_rate": 2.99154334038055e-06, |
| "loss": 0.0057, |
| "step": 2490 |
| }, |
| { |
| "epoch": 1.4979775280898877, |
| "grad_norm": 1.0945813655853271, |
| "learning_rate": 2.95630725863284e-06, |
| "loss": 0.0072, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.5039700374531835, |
| "grad_norm": 0.4797390103340149, |
| "learning_rate": 2.9210711768851307e-06, |
| "loss": 0.0069, |
| "step": 2510 |
| }, |
| { |
| "epoch": 1.5099625468164795, |
| "grad_norm": 2.183314323425293, |
| "learning_rate": 2.885835095137421e-06, |
| "loss": 0.0065, |
| "step": 2520 |
| }, |
| { |
| "epoch": 1.5159550561797754, |
| "grad_norm": 0.7189924120903015, |
| "learning_rate": 2.8505990133897114e-06, |
| "loss": 0.0054, |
| "step": 2530 |
| }, |
| { |
| "epoch": 1.5219475655430712, |
| "grad_norm": 0.9481498003005981, |
| "learning_rate": 2.8153629316420015e-06, |
| "loss": 0.0049, |
| "step": 2540 |
| }, |
| { |
| "epoch": 1.527940074906367, |
| "grad_norm": 2.258924961090088, |
| "learning_rate": 2.7801268498942916e-06, |
| "loss": 0.0077, |
| "step": 2550 |
| }, |
| { |
| "epoch": 1.533932584269663, |
| "grad_norm": 0.1977342665195465, |
| "learning_rate": 2.7448907681465826e-06, |
| "loss": 0.0085, |
| "step": 2560 |
| }, |
| { |
| "epoch": 1.539925093632959, |
| "grad_norm": 1.3324122428894043, |
| "learning_rate": 2.7096546863988727e-06, |
| "loss": 0.005, |
| "step": 2570 |
| }, |
| { |
| "epoch": 1.5459176029962547, |
| "grad_norm": 0.5444143414497375, |
| "learning_rate": 2.674418604651163e-06, |
| "loss": 0.0051, |
| "step": 2580 |
| }, |
| { |
| "epoch": 1.5519101123595505, |
| "grad_norm": 0.62740558385849, |
| "learning_rate": 2.6391825229034534e-06, |
| "loss": 0.007, |
| "step": 2590 |
| }, |
| { |
| "epoch": 1.5579026217228464, |
| "grad_norm": 0.43169164657592773, |
| "learning_rate": 2.6039464411557435e-06, |
| "loss": 0.0079, |
| "step": 2600 |
| }, |
| { |
| "epoch": 1.5638951310861424, |
| "grad_norm": 1.0292989015579224, |
| "learning_rate": 2.5687103594080345e-06, |
| "loss": 0.0055, |
| "step": 2610 |
| }, |
| { |
| "epoch": 1.5698876404494382, |
| "grad_norm": 1.001236915588379, |
| "learning_rate": 2.5334742776603246e-06, |
| "loss": 0.0083, |
| "step": 2620 |
| }, |
| { |
| "epoch": 1.575880149812734, |
| "grad_norm": 1.0882340669631958, |
| "learning_rate": 2.4982381959126147e-06, |
| "loss": 0.0046, |
| "step": 2630 |
| }, |
| { |
| "epoch": 1.58187265917603, |
| "grad_norm": 0.7527475953102112, |
| "learning_rate": 2.4630021141649053e-06, |
| "loss": 0.0051, |
| "step": 2640 |
| }, |
| { |
| "epoch": 1.587865168539326, |
| "grad_norm": 1.0273209810256958, |
| "learning_rate": 2.4277660324171954e-06, |
| "loss": 0.0085, |
| "step": 2650 |
| }, |
| { |
| "epoch": 1.5938576779026217, |
| "grad_norm": 1.1439406871795654, |
| "learning_rate": 2.392529950669486e-06, |
| "loss": 0.0054, |
| "step": 2660 |
| }, |
| { |
| "epoch": 1.5998501872659177, |
| "grad_norm": 1.746701955795288, |
| "learning_rate": 2.357293868921776e-06, |
| "loss": 0.0083, |
| "step": 2670 |
| }, |
| { |
| "epoch": 1.6058426966292134, |
| "grad_norm": 0.3131077289581299, |
| "learning_rate": 2.3220577871740666e-06, |
| "loss": 0.0039, |
| "step": 2680 |
| }, |
| { |
| "epoch": 1.6118352059925094, |
| "grad_norm": 0.9758055210113525, |
| "learning_rate": 2.2868217054263567e-06, |
| "loss": 0.0042, |
| "step": 2690 |
| }, |
| { |
| "epoch": 1.6178277153558054, |
| "grad_norm": 0.5772675275802612, |
| "learning_rate": 2.2515856236786473e-06, |
| "loss": 0.0039, |
| "step": 2700 |
| }, |
| { |
| "epoch": 1.6238202247191011, |
| "grad_norm": 0.8181300759315491, |
| "learning_rate": 2.2163495419309374e-06, |
| "loss": 0.0101, |
| "step": 2710 |
| }, |
| { |
| "epoch": 1.629812734082397, |
| "grad_norm": 0.3675064146518707, |
| "learning_rate": 2.181113460183228e-06, |
| "loss": 0.0054, |
| "step": 2720 |
| }, |
| { |
| "epoch": 1.635805243445693, |
| "grad_norm": 1.7257589101791382, |
| "learning_rate": 2.145877378435518e-06, |
| "loss": 0.008, |
| "step": 2730 |
| }, |
| { |
| "epoch": 1.6417977528089889, |
| "grad_norm": 0.25902286171913147, |
| "learning_rate": 2.1106412966878086e-06, |
| "loss": 0.0077, |
| "step": 2740 |
| }, |
| { |
| "epoch": 1.6477902621722846, |
| "grad_norm": 0.8033377528190613, |
| "learning_rate": 2.0754052149400987e-06, |
| "loss": 0.0039, |
| "step": 2750 |
| }, |
| { |
| "epoch": 1.6537827715355804, |
| "grad_norm": 0.5998724699020386, |
| "learning_rate": 2.0401691331923893e-06, |
| "loss": 0.008, |
| "step": 2760 |
| }, |
| { |
| "epoch": 1.6597752808988764, |
| "grad_norm": 1.0093644857406616, |
| "learning_rate": 2.00493305144468e-06, |
| "loss": 0.0055, |
| "step": 2770 |
| }, |
| { |
| "epoch": 1.6657677902621724, |
| "grad_norm": 0.11699071526527405, |
| "learning_rate": 1.96969696969697e-06, |
| "loss": 0.0067, |
| "step": 2780 |
| }, |
| { |
| "epoch": 1.6717602996254681, |
| "grad_norm": 0.24987554550170898, |
| "learning_rate": 1.93446088794926e-06, |
| "loss": 0.0047, |
| "step": 2790 |
| }, |
| { |
| "epoch": 1.6777528089887639, |
| "grad_norm": 0.580162763595581, |
| "learning_rate": 1.8992248062015506e-06, |
| "loss": 0.0097, |
| "step": 2800 |
| }, |
| { |
| "epoch": 1.6837453183520599, |
| "grad_norm": 0.7086832523345947, |
| "learning_rate": 1.863988724453841e-06, |
| "loss": 0.0048, |
| "step": 2810 |
| }, |
| { |
| "epoch": 1.6897378277153559, |
| "grad_norm": 0.9529427886009216, |
| "learning_rate": 1.828752642706131e-06, |
| "loss": 0.0065, |
| "step": 2820 |
| }, |
| { |
| "epoch": 1.6957303370786518, |
| "grad_norm": 0.6407700777053833, |
| "learning_rate": 1.7935165609584216e-06, |
| "loss": 0.0119, |
| "step": 2830 |
| }, |
| { |
| "epoch": 1.7017228464419476, |
| "grad_norm": 1.1374036073684692, |
| "learning_rate": 1.758280479210712e-06, |
| "loss": 0.0047, |
| "step": 2840 |
| }, |
| { |
| "epoch": 1.7077153558052434, |
| "grad_norm": 0.11668060719966888, |
| "learning_rate": 1.7230443974630023e-06, |
| "loss": 0.007, |
| "step": 2850 |
| }, |
| { |
| "epoch": 1.7137078651685393, |
| "grad_norm": 2.140202760696411, |
| "learning_rate": 1.6878083157152924e-06, |
| "loss": 0.0081, |
| "step": 2860 |
| }, |
| { |
| "epoch": 1.7197003745318353, |
| "grad_norm": 0.5463646054267883, |
| "learning_rate": 1.652572233967583e-06, |
| "loss": 0.007, |
| "step": 2870 |
| }, |
| { |
| "epoch": 1.725692883895131, |
| "grad_norm": 0.26337623596191406, |
| "learning_rate": 1.6173361522198733e-06, |
| "loss": 0.0079, |
| "step": 2880 |
| }, |
| { |
| "epoch": 1.7316853932584269, |
| "grad_norm": 1.1710906028747559, |
| "learning_rate": 1.5821000704721636e-06, |
| "loss": 0.0052, |
| "step": 2890 |
| }, |
| { |
| "epoch": 1.7376779026217228, |
| "grad_norm": 0.4584875702857971, |
| "learning_rate": 1.5468639887244541e-06, |
| "loss": 0.0037, |
| "step": 2900 |
| }, |
| { |
| "epoch": 1.7436704119850188, |
| "grad_norm": 1.4522494077682495, |
| "learning_rate": 1.5116279069767443e-06, |
| "loss": 0.0063, |
| "step": 2910 |
| }, |
| { |
| "epoch": 1.7496629213483146, |
| "grad_norm": 0.5847310423851013, |
| "learning_rate": 1.4763918252290346e-06, |
| "loss": 0.0077, |
| "step": 2920 |
| }, |
| { |
| "epoch": 1.7556554307116103, |
| "grad_norm": 1.927170753479004, |
| "learning_rate": 1.4411557434813251e-06, |
| "loss": 0.0068, |
| "step": 2930 |
| }, |
| { |
| "epoch": 1.7616479400749063, |
| "grad_norm": 0.5504691004753113, |
| "learning_rate": 1.4059196617336155e-06, |
| "loss": 0.0044, |
| "step": 2940 |
| }, |
| { |
| "epoch": 1.7676404494382023, |
| "grad_norm": 0.38014161586761475, |
| "learning_rate": 1.3706835799859056e-06, |
| "loss": 0.0044, |
| "step": 2950 |
| }, |
| { |
| "epoch": 1.7736329588014983, |
| "grad_norm": 0.6880072951316833, |
| "learning_rate": 1.335447498238196e-06, |
| "loss": 0.0061, |
| "step": 2960 |
| }, |
| { |
| "epoch": 1.779625468164794, |
| "grad_norm": 0.2847464382648468, |
| "learning_rate": 1.3002114164904865e-06, |
| "loss": 0.0059, |
| "step": 2970 |
| }, |
| { |
| "epoch": 1.7856179775280898, |
| "grad_norm": 0.5304543375968933, |
| "learning_rate": 1.2649753347427768e-06, |
| "loss": 0.005, |
| "step": 2980 |
| }, |
| { |
| "epoch": 1.7916104868913858, |
| "grad_norm": 0.3508507013320923, |
| "learning_rate": 1.2297392529950671e-06, |
| "loss": 0.0062, |
| "step": 2990 |
| }, |
| { |
| "epoch": 1.7976029962546818, |
| "grad_norm": 0.5547974705696106, |
| "learning_rate": 1.1945031712473575e-06, |
| "loss": 0.012, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.8035955056179775, |
| "grad_norm": 0.08585397154092789, |
| "learning_rate": 1.1592670894996478e-06, |
| "loss": 0.004, |
| "step": 3010 |
| }, |
| { |
| "epoch": 1.8095880149812733, |
| "grad_norm": 1.0728428363800049, |
| "learning_rate": 1.1240310077519381e-06, |
| "loss": 0.0053, |
| "step": 3020 |
| }, |
| { |
| "epoch": 1.8155805243445693, |
| "grad_norm": 0.8966228365898132, |
| "learning_rate": 1.0887949260042285e-06, |
| "loss": 0.0079, |
| "step": 3030 |
| }, |
| { |
| "epoch": 1.8215730337078653, |
| "grad_norm": 1.4710865020751953, |
| "learning_rate": 1.0535588442565188e-06, |
| "loss": 0.0064, |
| "step": 3040 |
| }, |
| { |
| "epoch": 1.827565543071161, |
| "grad_norm": 1.2710106372833252, |
| "learning_rate": 1.0183227625088091e-06, |
| "loss": 0.0045, |
| "step": 3050 |
| }, |
| { |
| "epoch": 1.8335580524344568, |
| "grad_norm": 0.39145779609680176, |
| "learning_rate": 9.830866807610995e-07, |
| "loss": 0.0064, |
| "step": 3060 |
| }, |
| { |
| "epoch": 1.8395505617977528, |
| "grad_norm": 0.9915968179702759, |
| "learning_rate": 9.478505990133897e-07, |
| "loss": 0.0037, |
| "step": 3070 |
| }, |
| { |
| "epoch": 1.8455430711610488, |
| "grad_norm": 0.36355727910995483, |
| "learning_rate": 9.126145172656801e-07, |
| "loss": 0.0042, |
| "step": 3080 |
| }, |
| { |
| "epoch": 1.8515355805243445, |
| "grad_norm": 0.560112476348877, |
| "learning_rate": 8.773784355179705e-07, |
| "loss": 0.0038, |
| "step": 3090 |
| }, |
| { |
| "epoch": 1.8575280898876403, |
| "grad_norm": 0.7680849432945251, |
| "learning_rate": 8.421423537702608e-07, |
| "loss": 0.0043, |
| "step": 3100 |
| }, |
| { |
| "epoch": 1.8635205992509363, |
| "grad_norm": 0.31203821301460266, |
| "learning_rate": 8.069062720225511e-07, |
| "loss": 0.0068, |
| "step": 3110 |
| }, |
| { |
| "epoch": 1.8695131086142323, |
| "grad_norm": 0.6210047602653503, |
| "learning_rate": 7.716701902748415e-07, |
| "loss": 0.0084, |
| "step": 3120 |
| }, |
| { |
| "epoch": 1.8755056179775282, |
| "grad_norm": 1.11445951461792, |
| "learning_rate": 7.364341085271319e-07, |
| "loss": 0.0075, |
| "step": 3130 |
| }, |
| { |
| "epoch": 1.881498127340824, |
| "grad_norm": 0.7059141993522644, |
| "learning_rate": 7.011980267794221e-07, |
| "loss": 0.0057, |
| "step": 3140 |
| }, |
| { |
| "epoch": 1.8874906367041198, |
| "grad_norm": 0.5541854500770569, |
| "learning_rate": 6.659619450317126e-07, |
| "loss": 0.0046, |
| "step": 3150 |
| }, |
| { |
| "epoch": 1.8934831460674157, |
| "grad_norm": 1.166515588760376, |
| "learning_rate": 6.307258632840028e-07, |
| "loss": 0.0041, |
| "step": 3160 |
| }, |
| { |
| "epoch": 1.8994756554307117, |
| "grad_norm": 0.43796196579933167, |
| "learning_rate": 5.954897815362931e-07, |
| "loss": 0.0117, |
| "step": 3170 |
| }, |
| { |
| "epoch": 1.9054681647940075, |
| "grad_norm": 0.6695800423622131, |
| "learning_rate": 5.602536997885836e-07, |
| "loss": 0.0056, |
| "step": 3180 |
| }, |
| { |
| "epoch": 1.9114606741573033, |
| "grad_norm": 0.248644158244133, |
| "learning_rate": 5.250176180408739e-07, |
| "loss": 0.0055, |
| "step": 3190 |
| }, |
| { |
| "epoch": 1.9174531835205992, |
| "grad_norm": 0.7921516299247742, |
| "learning_rate": 4.897815362931642e-07, |
| "loss": 0.0039, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1.9234456928838952, |
| "grad_norm": 0.6907849311828613, |
| "learning_rate": 4.5454545454545457e-07, |
| "loss": 0.0071, |
| "step": 3210 |
| }, |
| { |
| "epoch": 1.929438202247191, |
| "grad_norm": 0.38202527165412903, |
| "learning_rate": 4.193093727977449e-07, |
| "loss": 0.0111, |
| "step": 3220 |
| }, |
| { |
| "epoch": 1.9354307116104867, |
| "grad_norm": 0.1203591376543045, |
| "learning_rate": 3.840732910500353e-07, |
| "loss": 0.0071, |
| "step": 3230 |
| }, |
| { |
| "epoch": 1.9414232209737827, |
| "grad_norm": 0.10551086813211441, |
| "learning_rate": 3.488372093023256e-07, |
| "loss": 0.0031, |
| "step": 3240 |
| }, |
| { |
| "epoch": 1.9474157303370787, |
| "grad_norm": 0.2664087116718292, |
| "learning_rate": 3.1360112755461596e-07, |
| "loss": 0.0047, |
| "step": 3250 |
| }, |
| { |
| "epoch": 1.9534082397003747, |
| "grad_norm": 0.3722684383392334, |
| "learning_rate": 2.783650458069063e-07, |
| "loss": 0.0052, |
| "step": 3260 |
| }, |
| { |
| "epoch": 1.9594007490636705, |
| "grad_norm": 1.6388803720474243, |
| "learning_rate": 2.431289640591966e-07, |
| "loss": 0.0048, |
| "step": 3270 |
| }, |
| { |
| "epoch": 1.9653932584269662, |
| "grad_norm": 0.32303690910339355, |
| "learning_rate": 2.0789288231148699e-07, |
| "loss": 0.0043, |
| "step": 3280 |
| }, |
| { |
| "epoch": 1.9713857677902622, |
| "grad_norm": 0.1400229036808014, |
| "learning_rate": 1.7265680056377732e-07, |
| "loss": 0.0057, |
| "step": 3290 |
| }, |
| { |
| "epoch": 1.9773782771535582, |
| "grad_norm": 0.2554524540901184, |
| "learning_rate": 1.3742071881606765e-07, |
| "loss": 0.0059, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1.983370786516854, |
| "grad_norm": 0.8911289572715759, |
| "learning_rate": 1.0218463706835801e-07, |
| "loss": 0.0042, |
| "step": 3310 |
| }, |
| { |
| "epoch": 1.9893632958801497, |
| "grad_norm": 1.1008208990097046, |
| "learning_rate": 6.694855532064835e-08, |
| "loss": 0.0054, |
| "step": 3320 |
| }, |
| { |
| "epoch": 1.9953558052434457, |
| "grad_norm": 1.4023923873901367, |
| "learning_rate": 3.171247357293869e-08, |
| "loss": 0.005, |
| "step": 3330 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 3338, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 2000, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.09000604123136e+20, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
