diff --git "a/ensemble_student_7/trainer_state.json" "b/ensemble_student_7/trainer_state.json" new file mode 100644--- /dev/null +++ "b/ensemble_student_7/trainer_state.json" @@ -0,0 +1,7414 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 5.0, + "eval_steps": 500, + "global_step": 10530, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.004748338081671415, + "grad_norm": 17.432864424122187, + "learning_rate": 8.547008547008548e-07, + "loss": 8.2297, + "step": 10 + }, + { + "epoch": 0.00949667616334283, + "grad_norm": 16.63362455711296, + "learning_rate": 1.8043684710351376e-06, + "loss": 8.1208, + "step": 20 + }, + { + "epoch": 0.014245014245014245, + "grad_norm": 8.67007680301142, + "learning_rate": 2.7540360873694207e-06, + "loss": 7.5424, + "step": 30 + }, + { + "epoch": 0.01899335232668566, + "grad_norm": 4.516306847334219, + "learning_rate": 3.7037037037037037e-06, + "loss": 6.8654, + "step": 40 + }, + { + "epoch": 0.023741690408357077, + "grad_norm": 2.850133400972149, + "learning_rate": 4.653371320037987e-06, + "loss": 6.4317, + "step": 50 + }, + { + "epoch": 0.02849002849002849, + "grad_norm": 2.643737139794088, + "learning_rate": 5.60303893637227e-06, + "loss": 6.1171, + "step": 60 + }, + { + "epoch": 0.03323836657169991, + "grad_norm": 3.5445697531141858, + "learning_rate": 6.5527065527065525e-06, + "loss": 5.7929, + "step": 70 + }, + { + "epoch": 0.03798670465337132, + "grad_norm": 4.058133683478427, + "learning_rate": 7.5023741690408365e-06, + "loss": 5.2787, + "step": 80 + }, + { + "epoch": 0.042735042735042736, + "grad_norm": 5.307334866309611, + "learning_rate": 8.452041785375119e-06, + "loss": 4.7183, + "step": 90 + }, + { + "epoch": 0.04748338081671415, + "grad_norm": 6.384004902162933, + "learning_rate": 9.401709401709402e-06, + "loss": 4.2377, + "step": 100 + }, + { + "epoch": 0.052231718898385564, + "grad_norm": 5.392096618327688, + "learning_rate": 1.0351377018043685e-05, + "loss": 3.8902, + "step": 110 + }, + { + "epoch": 0.05698005698005698, + "grad_norm": 6.261362932318543, + "learning_rate": 1.1301044634377968e-05, + "loss": 3.636, + "step": 120 + }, + { + "epoch": 0.06172839506172839, + "grad_norm": 14.72936446343664, + "learning_rate": 1.2250712250712251e-05, + "loss": 3.4364, + "step": 130 + }, + { + "epoch": 0.06647673314339982, + "grad_norm": 11.787992877993863, + "learning_rate": 1.3200379867046534e-05, + "loss": 3.2742, + "step": 140 + }, + { + "epoch": 0.07122507122507123, + "grad_norm": 7.240669722892527, + "learning_rate": 1.4150047483380819e-05, + "loss": 3.1572, + "step": 150 + }, + { + "epoch": 0.07597340930674264, + "grad_norm": 7.30829025419389, + "learning_rate": 1.50997150997151e-05, + "loss": 3.0368, + "step": 160 + }, + { + "epoch": 0.08072174738841406, + "grad_norm": 6.438283955409255, + "learning_rate": 1.604938271604938e-05, + "loss": 2.9421, + "step": 170 + }, + { + "epoch": 0.08547008547008547, + "grad_norm": 7.488104378967692, + "learning_rate": 1.6999050332383668e-05, + "loss": 2.8323, + "step": 180 + }, + { + "epoch": 0.09021842355175688, + "grad_norm": 7.088085160192624, + "learning_rate": 1.794871794871795e-05, + "loss": 2.7356, + "step": 190 + }, + { + "epoch": 0.0949667616334283, + "grad_norm": 10.337969443899013, + "learning_rate": 1.889838556505223e-05, + "loss": 2.6452, + "step": 200 + }, + { + "epoch": 0.09971509971509972, + "grad_norm": 10.218633436004566, + "learning_rate": 1.9848053181386517e-05, + "loss": 2.5722, + "step": 210 + }, + { + "epoch": 0.10446343779677113, + "grad_norm": 10.736175461807424, + "learning_rate": 2.07977207977208e-05, + "loss": 2.4894, + "step": 220 + }, + { + "epoch": 0.10921177587844255, + "grad_norm": 7.357877887367853, + "learning_rate": 2.174738841405508e-05, + "loss": 2.4288, + "step": 230 + }, + { + "epoch": 0.11396011396011396, + "grad_norm": 12.866419035556016, + "learning_rate": 2.2697056030389367e-05, + "loss": 2.3692, + "step": 240 + }, + { + "epoch": 0.11870845204178537, + "grad_norm": 21.049917736585176, + "learning_rate": 2.364672364672365e-05, + "loss": 2.3364, + "step": 250 + }, + { + "epoch": 0.12345679012345678, + "grad_norm": 11.620896237830344, + "learning_rate": 2.459639126305793e-05, + "loss": 2.2798, + "step": 260 + }, + { + "epoch": 0.1282051282051282, + "grad_norm": 21.721184621887932, + "learning_rate": 2.5546058879392216e-05, + "loss": 2.2205, + "step": 270 + }, + { + "epoch": 0.13295346628679963, + "grad_norm": 12.495211584384712, + "learning_rate": 2.64957264957265e-05, + "loss": 2.1644, + "step": 280 + }, + { + "epoch": 0.13770180436847104, + "grad_norm": 14.812686036110316, + "learning_rate": 2.744539411206078e-05, + "loss": 2.1224, + "step": 290 + }, + { + "epoch": 0.14245014245014245, + "grad_norm": 10.895747006815013, + "learning_rate": 2.839506172839506e-05, + "loss": 2.0066, + "step": 300 + }, + { + "epoch": 0.14719848053181386, + "grad_norm": 8.833936208899251, + "learning_rate": 2.9344729344729345e-05, + "loss": 1.8866, + "step": 310 + }, + { + "epoch": 0.15194681861348527, + "grad_norm": 11.163269694033598, + "learning_rate": 3.029439696106363e-05, + "loss": 1.7675, + "step": 320 + }, + { + "epoch": 0.15669515669515668, + "grad_norm": 8.171052426461534, + "learning_rate": 3.1244064577397914e-05, + "loss": 1.6538, + "step": 330 + }, + { + "epoch": 0.16144349477682812, + "grad_norm": 9.792423767035137, + "learning_rate": 3.2193732193732194e-05, + "loss": 1.5653, + "step": 340 + }, + { + "epoch": 0.16619183285849953, + "grad_norm": 10.853553337054564, + "learning_rate": 3.314339981006648e-05, + "loss": 1.4682, + "step": 350 + }, + { + "epoch": 0.17094017094017094, + "grad_norm": 9.280208505051991, + "learning_rate": 3.409306742640076e-05, + "loss": 1.4044, + "step": 360 + }, + { + "epoch": 0.17568850902184235, + "grad_norm": 6.352523335219748, + "learning_rate": 3.504273504273504e-05, + "loss": 1.3535, + "step": 370 + }, + { + "epoch": 0.18043684710351376, + "grad_norm": 8.054285991410518, + "learning_rate": 3.599240265906933e-05, + "loss": 1.277, + "step": 380 + }, + { + "epoch": 0.18518518518518517, + "grad_norm": 6.082452836212801, + "learning_rate": 3.694207027540361e-05, + "loss": 1.23, + "step": 390 + }, + { + "epoch": 0.1899335232668566, + "grad_norm": 4.780897434524176, + "learning_rate": 3.789173789173789e-05, + "loss": 1.1896, + "step": 400 + }, + { + "epoch": 0.19468186134852802, + "grad_norm": 6.715836306755239, + "learning_rate": 3.884140550807218e-05, + "loss": 1.1603, + "step": 410 + }, + { + "epoch": 0.19943019943019943, + "grad_norm": 3.6229976657401615, + "learning_rate": 3.979107312440646e-05, + "loss": 1.127, + "step": 420 + }, + { + "epoch": 0.20417853751187084, + "grad_norm": 6.2850066380734635, + "learning_rate": 4.074074074074074e-05, + "loss": 1.08, + "step": 430 + }, + { + "epoch": 0.20892687559354226, + "grad_norm": 6.0654255702238355, + "learning_rate": 4.169040835707503e-05, + "loss": 1.0454, + "step": 440 + }, + { + "epoch": 0.21367521367521367, + "grad_norm": 5.000674723406981, + "learning_rate": 4.264007597340931e-05, + "loss": 1.0116, + "step": 450 + }, + { + "epoch": 0.2184235517568851, + "grad_norm": 4.464711378125037, + "learning_rate": 4.358974358974359e-05, + "loss": 0.9829, + "step": 460 + }, + { + "epoch": 0.22317188983855651, + "grad_norm": 5.763312794307584, + "learning_rate": 4.453941120607788e-05, + "loss": 0.9623, + "step": 470 + }, + { + "epoch": 0.22792022792022792, + "grad_norm": 5.940391132882479, + "learning_rate": 4.548907882241216e-05, + "loss": 0.9415, + "step": 480 + }, + { + "epoch": 0.23266856600189934, + "grad_norm": 5.838522317480493, + "learning_rate": 4.643874643874644e-05, + "loss": 0.9178, + "step": 490 + }, + { + "epoch": 0.23741690408357075, + "grad_norm": 5.54688270440205, + "learning_rate": 4.738841405508073e-05, + "loss": 0.8913, + "step": 500 + }, + { + "epoch": 0.24216524216524216, + "grad_norm": 4.001882422919119, + "learning_rate": 4.833808167141501e-05, + "loss": 0.8701, + "step": 510 + }, + { + "epoch": 0.24691358024691357, + "grad_norm": 4.500580556980079, + "learning_rate": 4.928774928774929e-05, + "loss": 0.8424, + "step": 520 + }, + { + "epoch": 0.251661918328585, + "grad_norm": 4.059034891756927, + "learning_rate": 5.0237416904083576e-05, + "loss": 0.8173, + "step": 530 + }, + { + "epoch": 0.2564102564102564, + "grad_norm": 4.26814628059991, + "learning_rate": 5.118708452041786e-05, + "loss": 0.7979, + "step": 540 + }, + { + "epoch": 0.2611585944919278, + "grad_norm": 3.86579517330865, + "learning_rate": 5.213675213675214e-05, + "loss": 0.78, + "step": 550 + }, + { + "epoch": 0.26590693257359926, + "grad_norm": 5.3503784870764175, + "learning_rate": 5.308641975308642e-05, + "loss": 0.7612, + "step": 560 + }, + { + "epoch": 0.2706552706552707, + "grad_norm": 4.018678502930456, + "learning_rate": 5.403608736942071e-05, + "loss": 0.7328, + "step": 570 + }, + { + "epoch": 0.2754036087369421, + "grad_norm": 3.1509535411443617, + "learning_rate": 5.498575498575499e-05, + "loss": 0.7222, + "step": 580 + }, + { + "epoch": 0.2801519468186135, + "grad_norm": 4.711207538954258, + "learning_rate": 5.5935422602089274e-05, + "loss": 0.7028, + "step": 590 + }, + { + "epoch": 0.2849002849002849, + "grad_norm": 3.66201060324775, + "learning_rate": 5.6885090218423554e-05, + "loss": 0.6877, + "step": 600 + }, + { + "epoch": 0.2896486229819563, + "grad_norm": 3.5761217614657603, + "learning_rate": 5.7834757834757834e-05, + "loss": 0.6648, + "step": 610 + }, + { + "epoch": 0.2943969610636277, + "grad_norm": 4.222669360531218, + "learning_rate": 5.878442545109212e-05, + "loss": 0.6335, + "step": 620 + }, + { + "epoch": 0.29914529914529914, + "grad_norm": 3.134889865565898, + "learning_rate": 5.97340930674264e-05, + "loss": 0.6242, + "step": 630 + }, + { + "epoch": 0.30389363722697055, + "grad_norm": 2.8999843858462127, + "learning_rate": 6.068376068376068e-05, + "loss": 0.6016, + "step": 640 + }, + { + "epoch": 0.30864197530864196, + "grad_norm": 4.059745718855991, + "learning_rate": 6.163342830009497e-05, + "loss": 0.5851, + "step": 650 + }, + { + "epoch": 0.31339031339031337, + "grad_norm": 3.5067777927843453, + "learning_rate": 6.258309591642925e-05, + "loss": 0.5708, + "step": 660 + }, + { + "epoch": 0.3181386514719848, + "grad_norm": 3.4958043671141477, + "learning_rate": 6.353276353276355e-05, + "loss": 0.5531, + "step": 670 + }, + { + "epoch": 0.32288698955365625, + "grad_norm": 2.959989452923648, + "learning_rate": 6.448243114909783e-05, + "loss": 0.5254, + "step": 680 + }, + { + "epoch": 0.32763532763532766, + "grad_norm": 2.462523251841697, + "learning_rate": 6.54320987654321e-05, + "loss": 0.5185, + "step": 690 + }, + { + "epoch": 0.33238366571699907, + "grad_norm": 2.5875325150237907, + "learning_rate": 6.638176638176638e-05, + "loss": 0.5091, + "step": 700 + }, + { + "epoch": 0.3371320037986705, + "grad_norm": 3.0856632997431603, + "learning_rate": 6.733143399810066e-05, + "loss": 0.4957, + "step": 710 + }, + { + "epoch": 0.3418803418803419, + "grad_norm": 2.971370030385959, + "learning_rate": 6.828110161443494e-05, + "loss": 0.4694, + "step": 720 + }, + { + "epoch": 0.3466286799620133, + "grad_norm": 1.9696173020757242, + "learning_rate": 6.923076923076924e-05, + "loss": 0.4551, + "step": 730 + }, + { + "epoch": 0.3513770180436847, + "grad_norm": 1.991606368771852, + "learning_rate": 7.018043684710352e-05, + "loss": 0.4401, + "step": 740 + }, + { + "epoch": 0.3561253561253561, + "grad_norm": 2.5992933005421928, + "learning_rate": 7.11301044634378e-05, + "loss": 0.4336, + "step": 750 + }, + { + "epoch": 0.36087369420702753, + "grad_norm": 2.1946162645208873, + "learning_rate": 7.207977207977208e-05, + "loss": 0.4152, + "step": 760 + }, + { + "epoch": 0.36562203228869894, + "grad_norm": 1.8446538182948966, + "learning_rate": 7.302943969610636e-05, + "loss": 0.4028, + "step": 770 + }, + { + "epoch": 0.37037037037037035, + "grad_norm": 2.408349892653379, + "learning_rate": 7.397910731244065e-05, + "loss": 0.3984, + "step": 780 + }, + { + "epoch": 0.37511870845204176, + "grad_norm": 2.764326789904103, + "learning_rate": 7.492877492877494e-05, + "loss": 0.3812, + "step": 790 + }, + { + "epoch": 0.3798670465337132, + "grad_norm": 2.2788441802312476, + "learning_rate": 7.587844254510922e-05, + "loss": 0.3701, + "step": 800 + }, + { + "epoch": 0.38461538461538464, + "grad_norm": 1.446006882668796, + "learning_rate": 7.68281101614435e-05, + "loss": 0.3512, + "step": 810 + }, + { + "epoch": 0.38936372269705605, + "grad_norm": 2.2605925527415116, + "learning_rate": 7.777777777777778e-05, + "loss": 0.3434, + "step": 820 + }, + { + "epoch": 0.39411206077872746, + "grad_norm": 1.9157168896042616, + "learning_rate": 7.872744539411206e-05, + "loss": 0.3402, + "step": 830 + }, + { + "epoch": 0.39886039886039887, + "grad_norm": 1.7875829509487084, + "learning_rate": 7.967711301044634e-05, + "loss": 0.3174, + "step": 840 + }, + { + "epoch": 0.4036087369420703, + "grad_norm": 1.7289430750540042, + "learning_rate": 8.062678062678063e-05, + "loss": 0.3096, + "step": 850 + }, + { + "epoch": 0.4083570750237417, + "grad_norm": 1.5185624571081302, + "learning_rate": 8.157644824311491e-05, + "loss": 0.3036, + "step": 860 + }, + { + "epoch": 0.4131054131054131, + "grad_norm": 1.8866414314506403, + "learning_rate": 8.25261158594492e-05, + "loss": 0.2997, + "step": 870 + }, + { + "epoch": 0.4178537511870845, + "grad_norm": 2.1040398909529285, + "learning_rate": 8.347578347578347e-05, + "loss": 0.2833, + "step": 880 + }, + { + "epoch": 0.4226020892687559, + "grad_norm": 1.3583802407350927, + "learning_rate": 8.442545109211775e-05, + "loss": 0.2778, + "step": 890 + }, + { + "epoch": 0.42735042735042733, + "grad_norm": 1.2610490630270248, + "learning_rate": 8.537511870845205e-05, + "loss": 0.275, + "step": 900 + }, + { + "epoch": 0.43209876543209874, + "grad_norm": 1.4487084630921243, + "learning_rate": 8.632478632478634e-05, + "loss": 0.2608, + "step": 910 + }, + { + "epoch": 0.4368471035137702, + "grad_norm": 1.415918445245649, + "learning_rate": 8.727445394112062e-05, + "loss": 0.2525, + "step": 920 + }, + { + "epoch": 0.4415954415954416, + "grad_norm": 1.273651472070069, + "learning_rate": 8.82241215574549e-05, + "loss": 0.2444, + "step": 930 + }, + { + "epoch": 0.44634377967711303, + "grad_norm": 1.7323109578352192, + "learning_rate": 8.917378917378918e-05, + "loss": 0.24, + "step": 940 + }, + { + "epoch": 0.45109211775878444, + "grad_norm": 1.1641894315690406, + "learning_rate": 9.012345679012346e-05, + "loss": 0.2399, + "step": 950 + }, + { + "epoch": 0.45584045584045585, + "grad_norm": 1.0462887029094885, + "learning_rate": 9.107312440645774e-05, + "loss": 0.2266, + "step": 960 + }, + { + "epoch": 0.46058879392212726, + "grad_norm": 1.0726705256851425, + "learning_rate": 9.202279202279203e-05, + "loss": 0.223, + "step": 970 + }, + { + "epoch": 0.46533713200379867, + "grad_norm": 2.4071360760023297, + "learning_rate": 9.297245963912631e-05, + "loss": 0.2195, + "step": 980 + }, + { + "epoch": 0.4700854700854701, + "grad_norm": 1.1125544913387833, + "learning_rate": 9.392212725546059e-05, + "loss": 0.2174, + "step": 990 + }, + { + "epoch": 0.4748338081671415, + "grad_norm": 1.5052675274629466, + "learning_rate": 9.487179487179487e-05, + "loss": 0.2114, + "step": 1000 + }, + { + "epoch": 0.4795821462488129, + "grad_norm": 0.9418262909872128, + "learning_rate": 9.582146248812915e-05, + "loss": 0.2004, + "step": 1010 + }, + { + "epoch": 0.4843304843304843, + "grad_norm": 0.8619027076153492, + "learning_rate": 9.677113010446344e-05, + "loss": 0.188, + "step": 1020 + }, + { + "epoch": 0.4890788224121557, + "grad_norm": 0.9419391393955927, + "learning_rate": 9.772079772079774e-05, + "loss": 0.189, + "step": 1030 + }, + { + "epoch": 0.49382716049382713, + "grad_norm": 1.1575714488740911, + "learning_rate": 9.867046533713202e-05, + "loss": 0.1829, + "step": 1040 + }, + { + "epoch": 0.4985754985754986, + "grad_norm": 0.8008197249491472, + "learning_rate": 9.96201329534663e-05, + "loss": 0.1835, + "step": 1050 + }, + { + "epoch": 0.50332383665717, + "grad_norm": 0.7460505811746059, + "learning_rate": 9.999990109907572e-05, + "loss": 0.1734, + "step": 1060 + }, + { + "epoch": 0.5080721747388414, + "grad_norm": 1.0748681291087179, + "learning_rate": 9.999929670595522e-05, + "loss": 0.1715, + "step": 1070 + }, + { + "epoch": 0.5128205128205128, + "grad_norm": 0.7150882638695625, + "learning_rate": 9.99981428713058e-05, + "loss": 0.1655, + "step": 1080 + }, + { + "epoch": 0.5175688509021842, + "grad_norm": 0.60451168171035, + "learning_rate": 9.999643960780693e-05, + "loss": 0.1579, + "step": 1090 + }, + { + "epoch": 0.5223171889838556, + "grad_norm": 0.6276787690628904, + "learning_rate": 9.999418693417573e-05, + "loss": 0.1532, + "step": 1100 + }, + { + "epoch": 0.5270655270655271, + "grad_norm": 0.611927883725597, + "learning_rate": 9.999138487516681e-05, + "loss": 0.1475, + "step": 1110 + }, + { + "epoch": 0.5318138651471985, + "grad_norm": 0.8908043781538245, + "learning_rate": 9.998803346157195e-05, + "loss": 0.1491, + "step": 1120 + }, + { + "epoch": 0.5365622032288699, + "grad_norm": 0.6898979212565218, + "learning_rate": 9.99841327302198e-05, + "loss": 0.1465, + "step": 1130 + }, + { + "epoch": 0.5413105413105413, + "grad_norm": 0.6050471714489271, + "learning_rate": 9.997968272397542e-05, + "loss": 0.1396, + "step": 1140 + }, + { + "epoch": 0.5460588793922128, + "grad_norm": 0.6161254824338225, + "learning_rate": 9.997468349173985e-05, + "loss": 0.1376, + "step": 1150 + }, + { + "epoch": 0.5508072174738842, + "grad_norm": 0.6340270927710323, + "learning_rate": 9.99691350884496e-05, + "loss": 0.1353, + "step": 1160 + }, + { + "epoch": 0.5555555555555556, + "grad_norm": 0.507827549497256, + "learning_rate": 9.996303757507602e-05, + "loss": 0.1327, + "step": 1170 + }, + { + "epoch": 0.560303893637227, + "grad_norm": 0.6372025516320821, + "learning_rate": 9.995639101862453e-05, + "loss": 0.1313, + "step": 1180 + }, + { + "epoch": 0.5650522317188984, + "grad_norm": 0.5129726960330293, + "learning_rate": 9.994919549213408e-05, + "loss": 0.1269, + "step": 1190 + }, + { + "epoch": 0.5698005698005698, + "grad_norm": 0.4379521654118294, + "learning_rate": 9.99414510746762e-05, + "loss": 0.1265, + "step": 1200 + }, + { + "epoch": 0.5745489078822412, + "grad_norm": 0.4931697479620591, + "learning_rate": 9.993315785135416e-05, + "loss": 0.1246, + "step": 1210 + }, + { + "epoch": 0.5792972459639126, + "grad_norm": 0.4946812340704433, + "learning_rate": 9.99243159133021e-05, + "loss": 0.1244, + "step": 1220 + }, + { + "epoch": 0.584045584045584, + "grad_norm": 1.0428205759736326, + "learning_rate": 9.991492535768392e-05, + "loss": 0.1211, + "step": 1230 + }, + { + "epoch": 0.5887939221272555, + "grad_norm": 0.481878288753429, + "learning_rate": 9.99049862876923e-05, + "loss": 0.1198, + "step": 1240 + }, + { + "epoch": 0.5935422602089269, + "grad_norm": 0.4352406429768311, + "learning_rate": 9.989449881254754e-05, + "loss": 0.1165, + "step": 1250 + }, + { + "epoch": 0.5982905982905983, + "grad_norm": 0.4390248302706892, + "learning_rate": 9.988346304749635e-05, + "loss": 0.1142, + "step": 1260 + }, + { + "epoch": 0.6030389363722697, + "grad_norm": 0.6910489420787164, + "learning_rate": 9.987187911381059e-05, + "loss": 0.1132, + "step": 1270 + }, + { + "epoch": 0.6077872744539411, + "grad_norm": 0.4153131258757278, + "learning_rate": 9.985974713878592e-05, + "loss": 0.1123, + "step": 1280 + }, + { + "epoch": 0.6125356125356125, + "grad_norm": 0.4069378444824487, + "learning_rate": 9.984706725574044e-05, + "loss": 0.1099, + "step": 1290 + }, + { + "epoch": 0.6172839506172839, + "grad_norm": 0.44642699318779955, + "learning_rate": 9.983383960401319e-05, + "loss": 0.1117, + "step": 1300 + }, + { + "epoch": 0.6220322886989553, + "grad_norm": 0.4787515831332926, + "learning_rate": 9.982006432896266e-05, + "loss": 0.1091, + "step": 1310 + }, + { + "epoch": 0.6267806267806267, + "grad_norm": 0.5289586712008078, + "learning_rate": 9.980574158196512e-05, + "loss": 0.1073, + "step": 1320 + }, + { + "epoch": 0.6315289648622981, + "grad_norm": 0.3759208221644045, + "learning_rate": 9.979087152041304e-05, + "loss": 0.1087, + "step": 1330 + }, + { + "epoch": 0.6362773029439696, + "grad_norm": 0.4090900796347572, + "learning_rate": 9.977545430771332e-05, + "loss": 0.1066, + "step": 1340 + }, + { + "epoch": 0.6410256410256411, + "grad_norm": 0.4097058594047625, + "learning_rate": 9.975949011328542e-05, + "loss": 0.1062, + "step": 1350 + }, + { + "epoch": 0.6457739791073125, + "grad_norm": 0.433259873811403, + "learning_rate": 9.97429791125597e-05, + "loss": 0.1024, + "step": 1360 + }, + { + "epoch": 0.6505223171889839, + "grad_norm": 0.39999005002861243, + "learning_rate": 9.972592148697528e-05, + "loss": 0.103, + "step": 1370 + }, + { + "epoch": 0.6552706552706553, + "grad_norm": 0.34448435460861243, + "learning_rate": 9.970831742397815e-05, + "loss": 0.1029, + "step": 1380 + }, + { + "epoch": 0.6600189933523267, + "grad_norm": 0.6341786601445849, + "learning_rate": 9.969016711701908e-05, + "loss": 0.102, + "step": 1390 + }, + { + "epoch": 0.6647673314339981, + "grad_norm": 0.3450749139048759, + "learning_rate": 9.967147076555157e-05, + "loss": 0.1007, + "step": 1400 + }, + { + "epoch": 0.6695156695156695, + "grad_norm": 0.3645569137708219, + "learning_rate": 9.965222857502951e-05, + "loss": 0.0995, + "step": 1410 + }, + { + "epoch": 0.674264007597341, + "grad_norm": 0.4044502343609566, + "learning_rate": 9.963244075690505e-05, + "loss": 0.0981, + "step": 1420 + }, + { + "epoch": 0.6790123456790124, + "grad_norm": 0.3942169334056769, + "learning_rate": 9.961210752862626e-05, + "loss": 0.0965, + "step": 1430 + }, + { + "epoch": 0.6837606837606838, + "grad_norm": 0.33115911237104434, + "learning_rate": 9.959122911363463e-05, + "loss": 0.0948, + "step": 1440 + }, + { + "epoch": 0.6885090218423552, + "grad_norm": 0.3087435826871738, + "learning_rate": 9.956980574136281e-05, + "loss": 0.0943, + "step": 1450 + }, + { + "epoch": 0.6932573599240266, + "grad_norm": 1.1610723753210634, + "learning_rate": 9.954783764723188e-05, + "loss": 0.0928, + "step": 1460 + }, + { + "epoch": 0.698005698005698, + "grad_norm": 0.5402876781043929, + "learning_rate": 9.952532507264891e-05, + "loss": 0.0955, + "step": 1470 + }, + { + "epoch": 0.7027540360873694, + "grad_norm": 0.32370379649779396, + "learning_rate": 9.950226826500426e-05, + "loss": 0.0955, + "step": 1480 + }, + { + "epoch": 0.7075023741690408, + "grad_norm": 0.3338862076724047, + "learning_rate": 9.947866747766882e-05, + "loss": 0.0943, + "step": 1490 + }, + { + "epoch": 0.7122507122507122, + "grad_norm": 0.3653076901761826, + "learning_rate": 9.945452296999131e-05, + "loss": 0.0926, + "step": 1500 + }, + { + "epoch": 0.7169990503323836, + "grad_norm": 0.3555191161769757, + "learning_rate": 9.942983500729536e-05, + "loss": 0.0914, + "step": 1510 + }, + { + "epoch": 0.7217473884140551, + "grad_norm": 0.3586469381843, + "learning_rate": 9.940460386087667e-05, + "loss": 0.091, + "step": 1520 + }, + { + "epoch": 0.7264957264957265, + "grad_norm": 0.38894924669213865, + "learning_rate": 9.937882980799986e-05, + "loss": 0.0896, + "step": 1530 + }, + { + "epoch": 0.7312440645773979, + "grad_norm": 0.37423188041376426, + "learning_rate": 9.935251313189564e-05, + "loss": 0.0921, + "step": 1540 + }, + { + "epoch": 0.7359924026590693, + "grad_norm": 0.31582582710948615, + "learning_rate": 9.93256541217576e-05, + "loss": 0.0913, + "step": 1550 + }, + { + "epoch": 0.7407407407407407, + "grad_norm": 0.3544845605748088, + "learning_rate": 9.929825307273898e-05, + "loss": 0.088, + "step": 1560 + }, + { + "epoch": 0.7454890788224121, + "grad_norm": 0.36091451801992247, + "learning_rate": 9.927031028594951e-05, + "loss": 0.0888, + "step": 1570 + }, + { + "epoch": 0.7502374169040835, + "grad_norm": 0.4621808633034976, + "learning_rate": 9.924182606845204e-05, + "loss": 0.0869, + "step": 1580 + }, + { + "epoch": 0.7549857549857549, + "grad_norm": 0.34036017277106173, + "learning_rate": 9.921280073325922e-05, + "loss": 0.0866, + "step": 1590 + }, + { + "epoch": 0.7597340930674265, + "grad_norm": 0.34605113849188424, + "learning_rate": 9.918323459933005e-05, + "loss": 0.086, + "step": 1600 + }, + { + "epoch": 0.7644824311490979, + "grad_norm": 0.3620145433165347, + "learning_rate": 9.915312799156632e-05, + "loss": 0.084, + "step": 1610 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.31298412649129087, + "learning_rate": 9.91224812408091e-05, + "loss": 0.0853, + "step": 1620 + }, + { + "epoch": 0.7739791073124407, + "grad_norm": 0.32798290916041184, + "learning_rate": 9.909129468383506e-05, + "loss": 0.0877, + "step": 1630 + }, + { + "epoch": 0.7787274453941121, + "grad_norm": 0.31540269002849874, + "learning_rate": 9.905956866335286e-05, + "loss": 0.0844, + "step": 1640 + }, + { + "epoch": 0.7834757834757835, + "grad_norm": 0.2934555747703243, + "learning_rate": 9.90273035279992e-05, + "loss": 0.0837, + "step": 1650 + }, + { + "epoch": 0.7882241215574549, + "grad_norm": 0.2700536772810812, + "learning_rate": 9.899449963233523e-05, + "loss": 0.0829, + "step": 1660 + }, + { + "epoch": 0.7929724596391263, + "grad_norm": 0.3015082999767182, + "learning_rate": 9.896115733684244e-05, + "loss": 0.0826, + "step": 1670 + }, + { + "epoch": 0.7977207977207977, + "grad_norm": 0.2589046023410531, + "learning_rate": 9.892727700791881e-05, + "loss": 0.081, + "step": 1680 + }, + { + "epoch": 0.8024691358024691, + "grad_norm": 0.30977940432674694, + "learning_rate": 9.889285901787482e-05, + "loss": 0.0834, + "step": 1690 + }, + { + "epoch": 0.8072174738841406, + "grad_norm": 0.28312205588589345, + "learning_rate": 9.885790374492921e-05, + "loss": 0.0837, + "step": 1700 + }, + { + "epoch": 0.811965811965812, + "grad_norm": 0.34503936605051383, + "learning_rate": 9.882241157320497e-05, + "loss": 0.0811, + "step": 1710 + }, + { + "epoch": 0.8167141500474834, + "grad_norm": 0.3226614418501471, + "learning_rate": 9.878638289272507e-05, + "loss": 0.0799, + "step": 1720 + }, + { + "epoch": 0.8214624881291548, + "grad_norm": 0.3112267378858567, + "learning_rate": 9.874981809940808e-05, + "loss": 0.0783, + "step": 1730 + }, + { + "epoch": 0.8262108262108262, + "grad_norm": 0.2720684837491654, + "learning_rate": 9.871271759506404e-05, + "loss": 0.0819, + "step": 1740 + }, + { + "epoch": 0.8309591642924976, + "grad_norm": 0.32534067181747117, + "learning_rate": 9.86750817873898e-05, + "loss": 0.0822, + "step": 1750 + }, + { + "epoch": 0.835707502374169, + "grad_norm": 0.3087109652443099, + "learning_rate": 9.863691108996471e-05, + "loss": 0.0777, + "step": 1760 + }, + { + "epoch": 0.8404558404558404, + "grad_norm": 0.29800225266471947, + "learning_rate": 9.859820592224599e-05, + "loss": 0.0794, + "step": 1770 + }, + { + "epoch": 0.8452041785375118, + "grad_norm": 0.3306248353381974, + "learning_rate": 9.855896670956415e-05, + "loss": 0.08, + "step": 1780 + }, + { + "epoch": 0.8499525166191833, + "grad_norm": 0.286094740379248, + "learning_rate": 9.851919388311832e-05, + "loss": 0.0781, + "step": 1790 + }, + { + "epoch": 0.8547008547008547, + "grad_norm": 0.29705664344681104, + "learning_rate": 9.84788878799715e-05, + "loss": 0.0764, + "step": 1800 + }, + { + "epoch": 0.8594491927825261, + "grad_norm": 0.3371289922525636, + "learning_rate": 9.843804914304577e-05, + "loss": 0.0778, + "step": 1810 + }, + { + "epoch": 0.8641975308641975, + "grad_norm": 0.30757552106706026, + "learning_rate": 9.839667812111739e-05, + "loss": 0.0757, + "step": 1820 + }, + { + "epoch": 0.8689458689458689, + "grad_norm": 0.28013529619640526, + "learning_rate": 9.835477526881189e-05, + "loss": 0.0767, + "step": 1830 + }, + { + "epoch": 0.8736942070275404, + "grad_norm": 0.30272477476810145, + "learning_rate": 9.831234104659913e-05, + "loss": 0.0759, + "step": 1840 + }, + { + "epoch": 0.8784425451092118, + "grad_norm": 0.29453790391537354, + "learning_rate": 9.826937592078811e-05, + "loss": 0.0746, + "step": 1850 + }, + { + "epoch": 0.8831908831908832, + "grad_norm": 0.2571991705787471, + "learning_rate": 9.822588036352201e-05, + "loss": 0.0755, + "step": 1860 + }, + { + "epoch": 0.8879392212725546, + "grad_norm": 0.260727354183401, + "learning_rate": 9.81818548527728e-05, + "loss": 0.0747, + "step": 1870 + }, + { + "epoch": 0.8926875593542261, + "grad_norm": 0.2507230639495949, + "learning_rate": 9.81372998723362e-05, + "loss": 0.0747, + "step": 1880 + }, + { + "epoch": 0.8974358974358975, + "grad_norm": 0.3161145189150504, + "learning_rate": 9.809221591182623e-05, + "loss": 0.0753, + "step": 1890 + }, + { + "epoch": 0.9021842355175689, + "grad_norm": 0.2653844897851644, + "learning_rate": 9.804660346666984e-05, + "loss": 0.073, + "step": 1900 + }, + { + "epoch": 0.9069325735992403, + "grad_norm": 0.2636332309425881, + "learning_rate": 9.800046303810154e-05, + "loss": 0.0729, + "step": 1910 + }, + { + "epoch": 0.9116809116809117, + "grad_norm": 0.2666779021286328, + "learning_rate": 9.795379513315779e-05, + "loss": 0.0737, + "step": 1920 + }, + { + "epoch": 0.9164292497625831, + "grad_norm": 0.26336289703155796, + "learning_rate": 9.790660026467153e-05, + "loss": 0.0716, + "step": 1930 + }, + { + "epoch": 0.9211775878442545, + "grad_norm": 0.25577431456937405, + "learning_rate": 9.785887895126644e-05, + "loss": 0.0707, + "step": 1940 + }, + { + "epoch": 0.9259259259259259, + "grad_norm": 0.2598450781476645, + "learning_rate": 9.781063171735134e-05, + "loss": 0.0691, + "step": 1950 + }, + { + "epoch": 0.9306742640075973, + "grad_norm": 0.30025187884565285, + "learning_rate": 9.776185909311437e-05, + "loss": 0.0697, + "step": 1960 + }, + { + "epoch": 0.9354226020892688, + "grad_norm": 0.2659520157120913, + "learning_rate": 9.771256161451716e-05, + "loss": 0.0713, + "step": 1970 + }, + { + "epoch": 0.9401709401709402, + "grad_norm": 0.2462468896036801, + "learning_rate": 9.766273982328899e-05, + "loss": 0.0689, + "step": 1980 + }, + { + "epoch": 0.9449192782526116, + "grad_norm": 0.2538687273343757, + "learning_rate": 9.761239426692077e-05, + "loss": 0.0687, + "step": 1990 + }, + { + "epoch": 0.949667616334283, + "grad_norm": 0.27548423564151736, + "learning_rate": 9.756152549865907e-05, + "loss": 0.0682, + "step": 2000 + }, + { + "epoch": 0.9544159544159544, + "grad_norm": 0.25948333883037444, + "learning_rate": 9.751013407750004e-05, + "loss": 0.0687, + "step": 2010 + }, + { + "epoch": 0.9591642924976258, + "grad_norm": 0.23983580384123895, + "learning_rate": 9.745822056818324e-05, + "loss": 0.0692, + "step": 2020 + }, + { + "epoch": 0.9639126305792972, + "grad_norm": 0.24620899100875848, + "learning_rate": 9.740578554118545e-05, + "loss": 0.067, + "step": 2030 + }, + { + "epoch": 0.9686609686609686, + "grad_norm": 0.27572340029809134, + "learning_rate": 9.735282957271438e-05, + "loss": 0.0651, + "step": 2040 + }, + { + "epoch": 0.97340930674264, + "grad_norm": 0.23420795574674338, + "learning_rate": 9.729935324470243e-05, + "loss": 0.0674, + "step": 2050 + }, + { + "epoch": 0.9781576448243114, + "grad_norm": 0.2502485281414291, + "learning_rate": 9.724535714480012e-05, + "loss": 0.0654, + "step": 2060 + }, + { + "epoch": 0.9829059829059829, + "grad_norm": 0.32860381858618326, + "learning_rate": 9.719084186636984e-05, + "loss": 0.0683, + "step": 2070 + }, + { + "epoch": 0.9876543209876543, + "grad_norm": 0.8547521950574297, + "learning_rate": 9.713580800847916e-05, + "loss": 0.0675, + "step": 2080 + }, + { + "epoch": 0.9924026590693258, + "grad_norm": 0.24350093538889778, + "learning_rate": 9.708025617589431e-05, + "loss": 0.0681, + "step": 2090 + }, + { + "epoch": 0.9971509971509972, + "grad_norm": 0.25332672615494994, + "learning_rate": 9.702418697907357e-05, + "loss": 0.065, + "step": 2100 + }, + { + "epoch": 1.0018993352326686, + "grad_norm": 0.21912322383267868, + "learning_rate": 9.696760103416049e-05, + "loss": 0.0635, + "step": 2110 + }, + { + "epoch": 1.00664767331434, + "grad_norm": 0.22346655594108236, + "learning_rate": 9.69104989629772e-05, + "loss": 0.0645, + "step": 2120 + }, + { + "epoch": 1.0113960113960114, + "grad_norm": 0.2207833292793415, + "learning_rate": 9.685288139301749e-05, + "loss": 0.0608, + "step": 2130 + }, + { + "epoch": 1.0161443494776827, + "grad_norm": 0.2288327495340157, + "learning_rate": 9.679474895744002e-05, + "loss": 0.0599, + "step": 2140 + }, + { + "epoch": 1.0208926875593543, + "grad_norm": 0.21937739737652542, + "learning_rate": 9.673610229506123e-05, + "loss": 0.0607, + "step": 2150 + }, + { + "epoch": 1.0256410256410255, + "grad_norm": 0.21251886282065818, + "learning_rate": 9.667694205034842e-05, + "loss": 0.0602, + "step": 2160 + }, + { + "epoch": 1.030389363722697, + "grad_norm": 0.23979277147859762, + "learning_rate": 9.661726887341264e-05, + "loss": 0.0607, + "step": 2170 + }, + { + "epoch": 1.0351377018043684, + "grad_norm": 0.22188448542790795, + "learning_rate": 9.655708342000158e-05, + "loss": 0.0582, + "step": 2180 + }, + { + "epoch": 1.03988603988604, + "grad_norm": 0.20772221664306262, + "learning_rate": 9.649638635149225e-05, + "loss": 0.0599, + "step": 2190 + }, + { + "epoch": 1.0446343779677112, + "grad_norm": 0.1897981441126486, + "learning_rate": 9.643517833488387e-05, + "loss": 0.058, + "step": 2200 + }, + { + "epoch": 1.0493827160493827, + "grad_norm": 0.22353239515400097, + "learning_rate": 9.637346004279038e-05, + "loss": 0.0584, + "step": 2210 + }, + { + "epoch": 1.0541310541310542, + "grad_norm": 0.20164721541316527, + "learning_rate": 9.631123215343323e-05, + "loss": 0.0593, + "step": 2220 + }, + { + "epoch": 1.0588793922127255, + "grad_norm": 0.23501613582940448, + "learning_rate": 9.624849535063373e-05, + "loss": 0.0593, + "step": 2230 + }, + { + "epoch": 1.063627730294397, + "grad_norm": 0.21948324888669157, + "learning_rate": 9.61852503238057e-05, + "loss": 0.0595, + "step": 2240 + }, + { + "epoch": 1.0683760683760684, + "grad_norm": 0.2227496002574335, + "learning_rate": 9.612149776794776e-05, + "loss": 0.0587, + "step": 2250 + }, + { + "epoch": 1.0731244064577399, + "grad_norm": 0.19077206116702786, + "learning_rate": 9.605723838363586e-05, + "loss": 0.0574, + "step": 2260 + }, + { + "epoch": 1.0778727445394112, + "grad_norm": 0.21270139796070567, + "learning_rate": 9.599247287701539e-05, + "loss": 0.0581, + "step": 2270 + }, + { + "epoch": 1.0826210826210827, + "grad_norm": 0.24285211637141993, + "learning_rate": 9.592720195979353e-05, + "loss": 0.0586, + "step": 2280 + }, + { + "epoch": 1.087369420702754, + "grad_norm": 0.18310646741381661, + "learning_rate": 9.586142634923141e-05, + "loss": 0.0562, + "step": 2290 + }, + { + "epoch": 1.0921177587844255, + "grad_norm": 0.22859129664217928, + "learning_rate": 9.579514676813627e-05, + "loss": 0.0558, + "step": 2300 + }, + { + "epoch": 1.0968660968660968, + "grad_norm": 0.21334402226775284, + "learning_rate": 9.572836394485345e-05, + "loss": 0.0561, + "step": 2310 + }, + { + "epoch": 1.1016144349477683, + "grad_norm": 0.20719716413244724, + "learning_rate": 9.566107861325836e-05, + "loss": 0.0559, + "step": 2320 + }, + { + "epoch": 1.1063627730294396, + "grad_norm": 0.1842267287823958, + "learning_rate": 9.559329151274854e-05, + "loss": 0.0542, + "step": 2330 + }, + { + "epoch": 1.1111111111111112, + "grad_norm": 0.22227221739789246, + "learning_rate": 9.552500338823543e-05, + "loss": 0.0553, + "step": 2340 + }, + { + "epoch": 1.1158594491927825, + "grad_norm": 0.20383457023544208, + "learning_rate": 9.54562149901362e-05, + "loss": 0.0539, + "step": 2350 + }, + { + "epoch": 1.120607787274454, + "grad_norm": 0.18934441614964076, + "learning_rate": 9.538692707436554e-05, + "loss": 0.0542, + "step": 2360 + }, + { + "epoch": 1.1253561253561253, + "grad_norm": 0.20642667701754755, + "learning_rate": 9.531714040232733e-05, + "loss": 0.0529, + "step": 2370 + }, + { + "epoch": 1.1301044634377968, + "grad_norm": 0.21940421686481923, + "learning_rate": 9.524685574090627e-05, + "loss": 0.0541, + "step": 2380 + }, + { + "epoch": 1.134852801519468, + "grad_norm": 0.1905843945715407, + "learning_rate": 9.517607386245948e-05, + "loss": 0.0518, + "step": 2390 + }, + { + "epoch": 1.1396011396011396, + "grad_norm": 0.20467849635937235, + "learning_rate": 9.510479554480793e-05, + "loss": 0.0521, + "step": 2400 + }, + { + "epoch": 1.144349477682811, + "grad_norm": 0.19687471427249065, + "learning_rate": 9.503302157122804e-05, + "loss": 0.0519, + "step": 2410 + }, + { + "epoch": 1.1490978157644824, + "grad_norm": 0.1925761023881491, + "learning_rate": 9.496075273044293e-05, + "loss": 0.0522, + "step": 2420 + }, + { + "epoch": 1.1538461538461537, + "grad_norm": 0.1957384714679789, + "learning_rate": 9.488798981661379e-05, + "loss": 0.0521, + "step": 2430 + }, + { + "epoch": 1.1585944919278253, + "grad_norm": 0.19537752549995221, + "learning_rate": 9.481473362933122e-05, + "loss": 0.0506, + "step": 2440 + }, + { + "epoch": 1.1633428300094968, + "grad_norm": 0.2023634276429789, + "learning_rate": 9.474098497360637e-05, + "loss": 0.0525, + "step": 2450 + }, + { + "epoch": 1.168091168091168, + "grad_norm": 0.20945425607384524, + "learning_rate": 9.466674465986211e-05, + "loss": 0.0504, + "step": 2460 + }, + { + "epoch": 1.1728395061728394, + "grad_norm": 0.18187730126082718, + "learning_rate": 9.459201350392418e-05, + "loss": 0.0511, + "step": 2470 + }, + { + "epoch": 1.177587844254511, + "grad_norm": 0.18423996040419854, + "learning_rate": 9.451679232701212e-05, + "loss": 0.0511, + "step": 2480 + }, + { + "epoch": 1.1823361823361824, + "grad_norm": 0.1792762400386612, + "learning_rate": 9.444108195573031e-05, + "loss": 0.0512, + "step": 2490 + }, + { + "epoch": 1.1870845204178537, + "grad_norm": 0.17851295562470668, + "learning_rate": 9.436488322205892e-05, + "loss": 0.0513, + "step": 2500 + }, + { + "epoch": 1.1918328584995252, + "grad_norm": 0.17321537102696463, + "learning_rate": 9.42881969633447e-05, + "loss": 0.0498, + "step": 2510 + }, + { + "epoch": 1.1965811965811965, + "grad_norm": 0.15606875969658696, + "learning_rate": 9.42110240222918e-05, + "loss": 0.0495, + "step": 2520 + }, + { + "epoch": 1.201329534662868, + "grad_norm": 0.16935041065793033, + "learning_rate": 9.413336524695253e-05, + "loss": 0.0498, + "step": 2530 + }, + { + "epoch": 1.2060778727445394, + "grad_norm": 0.18173747335605317, + "learning_rate": 9.405522149071801e-05, + "loss": 0.0493, + "step": 2540 + }, + { + "epoch": 1.210826210826211, + "grad_norm": 0.23952437886099476, + "learning_rate": 9.397659361230883e-05, + "loss": 0.049, + "step": 2550 + }, + { + "epoch": 1.2155745489078822, + "grad_norm": 0.16087103473093178, + "learning_rate": 9.389748247576557e-05, + "loss": 0.047, + "step": 2560 + }, + { + "epoch": 1.2203228869895537, + "grad_norm": 0.16230166526290482, + "learning_rate": 9.381788895043933e-05, + "loss": 0.0506, + "step": 2570 + }, + { + "epoch": 1.225071225071225, + "grad_norm": 0.1550053040731914, + "learning_rate": 9.373781391098218e-05, + "loss": 0.0469, + "step": 2580 + }, + { + "epoch": 1.2298195631528965, + "grad_norm": 0.1473224321036201, + "learning_rate": 9.365725823733753e-05, + "loss": 0.0479, + "step": 2590 + }, + { + "epoch": 1.2345679012345678, + "grad_norm": 0.5220140469649693, + "learning_rate": 9.35762228147305e-05, + "loss": 0.0612, + "step": 2600 + }, + { + "epoch": 1.2393162393162394, + "grad_norm": 0.2611768662775588, + "learning_rate": 9.349470853365815e-05, + "loss": 0.0517, + "step": 2610 + }, + { + "epoch": 1.2440645773979107, + "grad_norm": 0.2115029550683315, + "learning_rate": 9.341271628987968e-05, + "loss": 0.0508, + "step": 2620 + }, + { + "epoch": 1.2488129154795822, + "grad_norm": 0.17858362151678428, + "learning_rate": 9.333024698440667e-05, + "loss": 0.0496, + "step": 2630 + }, + { + "epoch": 1.2535612535612537, + "grad_norm": 0.19962221023322999, + "learning_rate": 9.324730152349306e-05, + "loss": 0.0493, + "step": 2640 + }, + { + "epoch": 1.258309591642925, + "grad_norm": 0.15266270618964042, + "learning_rate": 9.31638808186253e-05, + "loss": 0.0484, + "step": 2650 + }, + { + "epoch": 1.2630579297245963, + "grad_norm": 0.18092136509446036, + "learning_rate": 9.307998578651228e-05, + "loss": 0.0487, + "step": 2660 + }, + { + "epoch": 1.2678062678062678, + "grad_norm": 0.16977905170290133, + "learning_rate": 9.299561734907528e-05, + "loss": 0.049, + "step": 2670 + }, + { + "epoch": 1.2725546058879393, + "grad_norm": 0.15848717808477727, + "learning_rate": 9.291077643343776e-05, + "loss": 0.0494, + "step": 2680 + }, + { + "epoch": 1.2773029439696106, + "grad_norm": 0.15394052531679717, + "learning_rate": 9.282546397191531e-05, + "loss": 0.0478, + "step": 2690 + }, + { + "epoch": 1.282051282051282, + "grad_norm": 0.1498167787839862, + "learning_rate": 9.273968090200533e-05, + "loss": 0.0464, + "step": 2700 + }, + { + "epoch": 1.2867996201329535, + "grad_norm": 0.147478666496726, + "learning_rate": 9.265342816637666e-05, + "loss": 0.0461, + "step": 2710 + }, + { + "epoch": 1.291547958214625, + "grad_norm": 0.1444085898152993, + "learning_rate": 9.256670671285934e-05, + "loss": 0.0459, + "step": 2720 + }, + { + "epoch": 1.2962962962962963, + "grad_norm": 0.14939606261681945, + "learning_rate": 9.247951749443413e-05, + "loss": 0.0448, + "step": 2730 + }, + { + "epoch": 1.3010446343779676, + "grad_norm": 0.15029207453897359, + "learning_rate": 9.239186146922208e-05, + "loss": 0.0452, + "step": 2740 + }, + { + "epoch": 1.305792972459639, + "grad_norm": 0.14164642597163712, + "learning_rate": 9.230373960047394e-05, + "loss": 0.0463, + "step": 2750 + }, + { + "epoch": 1.3105413105413106, + "grad_norm": 0.15220212063707358, + "learning_rate": 9.22151528565596e-05, + "loss": 0.0453, + "step": 2760 + }, + { + "epoch": 1.315289648622982, + "grad_norm": 0.16119691543001152, + "learning_rate": 9.212610221095748e-05, + "loss": 0.0468, + "step": 2770 + }, + { + "epoch": 1.3200379867046534, + "grad_norm": 0.7261612451264676, + "learning_rate": 9.203658864224382e-05, + "loss": 0.0543, + "step": 2780 + }, + { + "epoch": 1.3247863247863247, + "grad_norm": 0.27857509760784527, + "learning_rate": 9.194661313408187e-05, + "loss": 0.0546, + "step": 2790 + }, + { + "epoch": 1.3295346628679963, + "grad_norm": 0.1675307975986236, + "learning_rate": 9.185617667521117e-05, + "loss": 0.0482, + "step": 2800 + }, + { + "epoch": 1.3342830009496676, + "grad_norm": 0.165408335565686, + "learning_rate": 9.176528025943663e-05, + "loss": 0.0474, + "step": 2810 + }, + { + "epoch": 1.339031339031339, + "grad_norm": 0.16225375283847396, + "learning_rate": 9.167392488561758e-05, + "loss": 0.0464, + "step": 2820 + }, + { + "epoch": 1.3437796771130104, + "grad_norm": 0.15512358699559678, + "learning_rate": 9.158211155765692e-05, + "loss": 0.0456, + "step": 2830 + }, + { + "epoch": 1.348528015194682, + "grad_norm": 0.13702079862007333, + "learning_rate": 9.14898412844899e-05, + "loss": 0.0453, + "step": 2840 + }, + { + "epoch": 1.3532763532763532, + "grad_norm": 0.12896072018717666, + "learning_rate": 9.13971150800732e-05, + "loss": 0.0448, + "step": 2850 + }, + { + "epoch": 1.3580246913580247, + "grad_norm": 0.15652505223773908, + "learning_rate": 9.130393396337368e-05, + "loss": 0.0458, + "step": 2860 + }, + { + "epoch": 1.362773029439696, + "grad_norm": 0.1522508651001737, + "learning_rate": 9.121029895835731e-05, + "loss": 0.0446, + "step": 2870 + }, + { + "epoch": 1.3675213675213675, + "grad_norm": 0.1276028163943599, + "learning_rate": 9.111621109397769e-05, + "loss": 0.0456, + "step": 2880 + }, + { + "epoch": 1.3722697056030388, + "grad_norm": 0.13869727344939897, + "learning_rate": 9.102167140416503e-05, + "loss": 0.0428, + "step": 2890 + }, + { + "epoch": 1.3770180436847104, + "grad_norm": 0.16312935755283214, + "learning_rate": 9.092668092781454e-05, + "loss": 0.0451, + "step": 2900 + }, + { + "epoch": 1.381766381766382, + "grad_norm": 0.14681004181182833, + "learning_rate": 9.083124070877516e-05, + "loss": 0.0443, + "step": 2910 + }, + { + "epoch": 1.3865147198480532, + "grad_norm": 0.1414240629596368, + "learning_rate": 9.073535179583806e-05, + "loss": 0.0449, + "step": 2920 + }, + { + "epoch": 1.3912630579297245, + "grad_norm": 0.13609146789630597, + "learning_rate": 9.063901524272505e-05, + "loss": 0.044, + "step": 2930 + }, + { + "epoch": 1.396011396011396, + "grad_norm": 0.1396980536936488, + "learning_rate": 9.054223210807709e-05, + "loss": 0.044, + "step": 2940 + }, + { + "epoch": 1.4007597340930675, + "grad_norm": 0.13452472736927784, + "learning_rate": 9.044500345544262e-05, + "loss": 0.0437, + "step": 2950 + }, + { + "epoch": 1.4055080721747388, + "grad_norm": 0.1347515273813185, + "learning_rate": 9.034733035326584e-05, + "loss": 0.0437, + "step": 2960 + }, + { + "epoch": 1.4102564102564101, + "grad_norm": 0.13601750787874367, + "learning_rate": 9.024921387487504e-05, + "loss": 0.0443, + "step": 2970 + }, + { + "epoch": 1.4150047483380817, + "grad_norm": 0.15092324146234815, + "learning_rate": 9.015065509847075e-05, + "loss": 0.0428, + "step": 2980 + }, + { + "epoch": 1.4197530864197532, + "grad_norm": 0.131087509040608, + "learning_rate": 9.005165510711387e-05, + "loss": 0.0444, + "step": 2990 + }, + { + "epoch": 1.4245014245014245, + "grad_norm": 0.11727372514241877, + "learning_rate": 8.995221498871387e-05, + "loss": 0.0436, + "step": 3000 + }, + { + "epoch": 1.429249762583096, + "grad_norm": 0.1182153890465175, + "learning_rate": 8.98523358360167e-05, + "loss": 0.0439, + "step": 3010 + }, + { + "epoch": 1.4339981006647673, + "grad_norm": 0.12249931547934703, + "learning_rate": 8.975201874659292e-05, + "loss": 0.0433, + "step": 3020 + }, + { + "epoch": 1.4387464387464388, + "grad_norm": 0.1425786226434079, + "learning_rate": 8.965126482282549e-05, + "loss": 0.042, + "step": 3030 + }, + { + "epoch": 1.4434947768281101, + "grad_norm": 0.14378943137063568, + "learning_rate": 8.955007517189784e-05, + "loss": 0.0425, + "step": 3040 + }, + { + "epoch": 1.4482431149097816, + "grad_norm": 0.13856051879893824, + "learning_rate": 8.944845090578149e-05, + "loss": 0.0443, + "step": 3050 + }, + { + "epoch": 1.452991452991453, + "grad_norm": 0.15327736567967157, + "learning_rate": 8.934639314122401e-05, + "loss": 0.0445, + "step": 3060 + }, + { + "epoch": 1.4577397910731245, + "grad_norm": 0.1543878554740903, + "learning_rate": 8.924390299973666e-05, + "loss": 0.0426, + "step": 3070 + }, + { + "epoch": 1.4624881291547958, + "grad_norm": 0.1270382086630365, + "learning_rate": 8.91409816075821e-05, + "loss": 0.0441, + "step": 3080 + }, + { + "epoch": 1.4672364672364673, + "grad_norm": 0.12237129208610195, + "learning_rate": 8.903763009576195e-05, + "loss": 0.0429, + "step": 3090 + }, + { + "epoch": 1.4719848053181386, + "grad_norm": 0.12711468071481644, + "learning_rate": 8.893384960000445e-05, + "loss": 0.0423, + "step": 3100 + }, + { + "epoch": 1.47673314339981, + "grad_norm": 0.14274934943665177, + "learning_rate": 8.882964126075192e-05, + "loss": 0.0426, + "step": 3110 + }, + { + "epoch": 1.4814814814814814, + "grad_norm": 0.13599778051924102, + "learning_rate": 8.872500622314829e-05, + "loss": 0.044, + "step": 3120 + }, + { + "epoch": 1.486229819563153, + "grad_norm": 0.1332015476195608, + "learning_rate": 8.861994563702638e-05, + "loss": 0.043, + "step": 3130 + }, + { + "epoch": 1.4909781576448244, + "grad_norm": 0.13636237748902338, + "learning_rate": 8.851446065689544e-05, + "loss": 0.0418, + "step": 3140 + }, + { + "epoch": 1.4957264957264957, + "grad_norm": 0.12508153131116828, + "learning_rate": 8.840855244192836e-05, + "loss": 0.0413, + "step": 3150 + }, + { + "epoch": 1.500474833808167, + "grad_norm": 0.13383901174850002, + "learning_rate": 8.83022221559489e-05, + "loss": 0.0431, + "step": 3160 + }, + { + "epoch": 1.5052231718898386, + "grad_norm": 0.12312804069841883, + "learning_rate": 8.819547096741904e-05, + "loss": 0.0425, + "step": 3170 + }, + { + "epoch": 1.50997150997151, + "grad_norm": 0.11718125254101988, + "learning_rate": 8.8088300049426e-05, + "loss": 0.0427, + "step": 3180 + }, + { + "epoch": 1.5147198480531814, + "grad_norm": 0.11995986857849927, + "learning_rate": 8.798071057966935e-05, + "loss": 0.0404, + "step": 3190 + }, + { + "epoch": 1.5194681861348527, + "grad_norm": 0.140468845705722, + "learning_rate": 8.787270374044824e-05, + "loss": 0.0419, + "step": 3200 + }, + { + "epoch": 1.5242165242165242, + "grad_norm": 0.1329184395064802, + "learning_rate": 8.776428071864819e-05, + "loss": 0.0422, + "step": 3210 + }, + { + "epoch": 1.5289648622981957, + "grad_norm": 0.1281151562706742, + "learning_rate": 8.76554427057282e-05, + "loss": 0.0417, + "step": 3220 + }, + { + "epoch": 1.533713200379867, + "grad_norm": 0.157973166964191, + "learning_rate": 8.754619089770755e-05, + "loss": 0.0411, + "step": 3230 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.1229888410446223, + "learning_rate": 8.743652649515275e-05, + "loss": 0.0418, + "step": 3240 + }, + { + "epoch": 1.5432098765432098, + "grad_norm": 0.12403225068936573, + "learning_rate": 8.73264507031643e-05, + "loss": 0.0416, + "step": 3250 + }, + { + "epoch": 1.5479582146248814, + "grad_norm": 0.1256603260346007, + "learning_rate": 8.721596473136342e-05, + "loss": 0.0416, + "step": 3260 + }, + { + "epoch": 1.5527065527065527, + "grad_norm": 0.1299088240864389, + "learning_rate": 8.710506979387882e-05, + "loss": 0.0419, + "step": 3270 + }, + { + "epoch": 1.557454890788224, + "grad_norm": 0.13751270479938368, + "learning_rate": 8.699376710933336e-05, + "loss": 0.0425, + "step": 3280 + }, + { + "epoch": 1.5622032288698955, + "grad_norm": 0.12892726080049477, + "learning_rate": 8.688205790083053e-05, + "loss": 0.042, + "step": 3290 + }, + { + "epoch": 1.566951566951567, + "grad_norm": 0.12080615004282609, + "learning_rate": 8.676994339594118e-05, + "loss": 0.0416, + "step": 3300 + }, + { + "epoch": 1.5716999050332383, + "grad_norm": 0.12659703415370233, + "learning_rate": 8.665742482668996e-05, + "loss": 0.0417, + "step": 3310 + }, + { + "epoch": 1.5764482431149098, + "grad_norm": 0.11919385279623763, + "learning_rate": 8.654450342954175e-05, + "loss": 0.0413, + "step": 3320 + }, + { + "epoch": 1.5811965811965814, + "grad_norm": 0.11606100188090912, + "learning_rate": 8.643118044538808e-05, + "loss": 0.0418, + "step": 3330 + }, + { + "epoch": 1.5859449192782527, + "grad_norm": 0.12873235710598074, + "learning_rate": 8.631745711953357e-05, + "loss": 0.0414, + "step": 3340 + }, + { + "epoch": 1.590693257359924, + "grad_norm": 0.11589313123593642, + "learning_rate": 8.620333470168214e-05, + "loss": 0.0417, + "step": 3350 + }, + { + "epoch": 1.5954415954415955, + "grad_norm": 0.1219243811158153, + "learning_rate": 8.608881444592337e-05, + "loss": 0.041, + "step": 3360 + }, + { + "epoch": 1.600189933523267, + "grad_norm": 0.10972509536995534, + "learning_rate": 8.597389761071862e-05, + "loss": 0.0402, + "step": 3370 + }, + { + "epoch": 1.6049382716049383, + "grad_norm": 0.1296015281557993, + "learning_rate": 8.58585854588873e-05, + "loss": 0.0412, + "step": 3380 + }, + { + "epoch": 1.6096866096866096, + "grad_norm": 0.10767684917076147, + "learning_rate": 8.574287925759294e-05, + "loss": 0.0399, + "step": 3390 + }, + { + "epoch": 1.6144349477682811, + "grad_norm": 0.13244074208635626, + "learning_rate": 8.562678027832925e-05, + "loss": 0.0405, + "step": 3400 + }, + { + "epoch": 1.6191832858499526, + "grad_norm": 0.11637139867450855, + "learning_rate": 8.551028979690621e-05, + "loss": 0.0401, + "step": 3410 + }, + { + "epoch": 1.623931623931624, + "grad_norm": 0.12222054477045727, + "learning_rate": 8.539340909343597e-05, + "loss": 0.0403, + "step": 3420 + }, + { + "epoch": 1.6286799620132952, + "grad_norm": 0.11514800508327834, + "learning_rate": 8.527613945231885e-05, + "loss": 0.0407, + "step": 3430 + }, + { + "epoch": 1.6334283000949668, + "grad_norm": 0.10793141811121093, + "learning_rate": 8.51584821622292e-05, + "loss": 0.0408, + "step": 3440 + }, + { + "epoch": 1.6381766381766383, + "grad_norm": 0.11557071555291666, + "learning_rate": 8.504043851610122e-05, + "loss": 0.0409, + "step": 3450 + }, + { + "epoch": 1.6429249762583096, + "grad_norm": 0.11808765551904757, + "learning_rate": 8.492200981111482e-05, + "loss": 0.0397, + "step": 3460 + }, + { + "epoch": 1.6476733143399809, + "grad_norm": 0.12075331508547124, + "learning_rate": 8.480319734868124e-05, + "loss": 0.0408, + "step": 3470 + }, + { + "epoch": 1.6524216524216524, + "grad_norm": 0.11520321154123168, + "learning_rate": 8.468400243442888e-05, + "loss": 0.0396, + "step": 3480 + }, + { + "epoch": 1.657169990503324, + "grad_norm": 0.12071641509648842, + "learning_rate": 8.456442637818889e-05, + "loss": 0.039, + "step": 3490 + }, + { + "epoch": 1.6619183285849952, + "grad_norm": 0.1215549065836691, + "learning_rate": 8.444447049398076e-05, + "loss": 0.0403, + "step": 3500 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 0.11633801575196545, + "learning_rate": 8.432413609999794e-05, + "loss": 0.0399, + "step": 3510 + }, + { + "epoch": 1.671415004748338, + "grad_norm": 0.12983909835663454, + "learning_rate": 8.420342451859329e-05, + "loss": 0.0397, + "step": 3520 + }, + { + "epoch": 1.6761633428300096, + "grad_norm": 0.11283828950587488, + "learning_rate": 8.408233707626458e-05, + "loss": 0.0404, + "step": 3530 + }, + { + "epoch": 1.6809116809116809, + "grad_norm": 0.1235547958566696, + "learning_rate": 8.39608751036399e-05, + "loss": 0.0405, + "step": 3540 + }, + { + "epoch": 1.6856600189933522, + "grad_norm": 0.13053400332191945, + "learning_rate": 8.383903993546311e-05, + "loss": 0.0409, + "step": 3550 + }, + { + "epoch": 1.690408357075024, + "grad_norm": 0.09881875731919204, + "learning_rate": 8.371683291057902e-05, + "loss": 0.0401, + "step": 3560 + }, + { + "epoch": 1.6951566951566952, + "grad_norm": 0.11900588175847691, + "learning_rate": 8.359425537191883e-05, + "loss": 0.0405, + "step": 3570 + }, + { + "epoch": 1.6999050332383665, + "grad_norm": 0.1182612167989668, + "learning_rate": 8.347130866648528e-05, + "loss": 0.0403, + "step": 3580 + }, + { + "epoch": 1.704653371320038, + "grad_norm": 0.1332315928606217, + "learning_rate": 8.334799414533791e-05, + "loss": 0.0403, + "step": 3590 + }, + { + "epoch": 1.7094017094017095, + "grad_norm": 0.1123280856865077, + "learning_rate": 8.322431316357815e-05, + "loss": 0.0383, + "step": 3600 + }, + { + "epoch": 1.7141500474833808, + "grad_norm": 0.1206524268810823, + "learning_rate": 8.310026708033447e-05, + "loss": 0.0387, + "step": 3610 + }, + { + "epoch": 1.7188983855650521, + "grad_norm": 0.12464543627184521, + "learning_rate": 8.297585725874743e-05, + "loss": 0.0399, + "step": 3620 + }, + { + "epoch": 1.7236467236467237, + "grad_norm": 0.12020133337636746, + "learning_rate": 8.285108506595471e-05, + "loss": 0.0402, + "step": 3630 + }, + { + "epoch": 1.7283950617283952, + "grad_norm": 0.13318837076393672, + "learning_rate": 8.272595187307612e-05, + "loss": 0.0392, + "step": 3640 + }, + { + "epoch": 1.7331433998100665, + "grad_norm": 0.12218215896754882, + "learning_rate": 8.260045905519842e-05, + "loss": 0.041, + "step": 3650 + }, + { + "epoch": 1.7378917378917378, + "grad_norm": 0.11562501042945135, + "learning_rate": 8.247460799136032e-05, + "loss": 0.0394, + "step": 3660 + }, + { + "epoch": 1.7426400759734093, + "grad_norm": 0.10978918617006504, + "learning_rate": 8.234840006453734e-05, + "loss": 0.0381, + "step": 3670 + }, + { + "epoch": 1.7473884140550808, + "grad_norm": 0.10522213302683346, + "learning_rate": 8.222183666162647e-05, + "loss": 0.0395, + "step": 3680 + }, + { + "epoch": 1.7521367521367521, + "grad_norm": 0.10381036059133479, + "learning_rate": 8.209491917343104e-05, + "loss": 0.0388, + "step": 3690 + }, + { + "epoch": 1.7568850902184234, + "grad_norm": 0.09685438123981976, + "learning_rate": 8.196764899464551e-05, + "loss": 0.0392, + "step": 3700 + }, + { + "epoch": 1.761633428300095, + "grad_norm": 0.11506480811083894, + "learning_rate": 8.184002752383993e-05, + "loss": 0.0396, + "step": 3710 + }, + { + "epoch": 1.7663817663817665, + "grad_norm": 0.10258425650028577, + "learning_rate": 8.171205616344478e-05, + "loss": 0.04, + "step": 3720 + }, + { + "epoch": 1.7711301044634378, + "grad_norm": 0.13351702062046716, + "learning_rate": 8.158373631973542e-05, + "loss": 0.0391, + "step": 3730 + }, + { + "epoch": 1.775878442545109, + "grad_norm": 0.1161969449307836, + "learning_rate": 8.14550694028167e-05, + "loss": 0.0391, + "step": 3740 + }, + { + "epoch": 1.7806267806267806, + "grad_norm": 0.09628463518063829, + "learning_rate": 8.13260568266075e-05, + "loss": 0.0392, + "step": 3750 + }, + { + "epoch": 1.7853751187084521, + "grad_norm": 0.14175985574433034, + "learning_rate": 8.119670000882506e-05, + "loss": 0.0402, + "step": 3760 + }, + { + "epoch": 1.7901234567901234, + "grad_norm": 0.11097406245410807, + "learning_rate": 8.106700037096955e-05, + "loss": 0.0401, + "step": 3770 + }, + { + "epoch": 1.7948717948717947, + "grad_norm": 0.10934435755609073, + "learning_rate": 8.093695933830838e-05, + "loss": 0.0389, + "step": 3780 + }, + { + "epoch": 1.7996201329534662, + "grad_norm": 0.11154977313023136, + "learning_rate": 8.080657833986047e-05, + "loss": 0.0384, + "step": 3790 + }, + { + "epoch": 1.8043684710351378, + "grad_norm": 0.11034118922358843, + "learning_rate": 8.067585880838073e-05, + "loss": 0.0397, + "step": 3800 + }, + { + "epoch": 1.809116809116809, + "grad_norm": 0.10154319489690186, + "learning_rate": 8.054480218034413e-05, + "loss": 0.0392, + "step": 3810 + }, + { + "epoch": 1.8138651471984806, + "grad_norm": 0.11324160917024435, + "learning_rate": 8.041340989593003e-05, + "loss": 0.0387, + "step": 3820 + }, + { + "epoch": 1.818613485280152, + "grad_norm": 0.10812735813775641, + "learning_rate": 8.028168339900625e-05, + "loss": 0.0382, + "step": 3830 + }, + { + "epoch": 1.8233618233618234, + "grad_norm": 0.12202215682123403, + "learning_rate": 8.014962413711334e-05, + "loss": 0.0386, + "step": 3840 + }, + { + "epoch": 1.8281101614434947, + "grad_norm": 0.11494240850641056, + "learning_rate": 8.001723356144855e-05, + "loss": 0.0385, + "step": 3850 + }, + { + "epoch": 1.8328584995251662, + "grad_norm": 0.10028366485676024, + "learning_rate": 7.988451312684997e-05, + "loss": 0.0379, + "step": 3860 + }, + { + "epoch": 1.8376068376068377, + "grad_norm": 0.11918434242549941, + "learning_rate": 7.975146429178047e-05, + "loss": 0.0402, + "step": 3870 + }, + { + "epoch": 1.842355175688509, + "grad_norm": 0.12477382726856209, + "learning_rate": 7.961808851831172e-05, + "loss": 0.0385, + "step": 3880 + }, + { + "epoch": 1.8471035137701803, + "grad_norm": 0.09542100962004467, + "learning_rate": 7.948438727210814e-05, + "loss": 0.038, + "step": 3890 + }, + { + "epoch": 1.8518518518518519, + "grad_norm": 0.12188152766033451, + "learning_rate": 7.935036202241071e-05, + "loss": 0.0384, + "step": 3900 + }, + { + "epoch": 1.8566001899335234, + "grad_norm": 0.1090603641635716, + "learning_rate": 7.921601424202091e-05, + "loss": 0.0382, + "step": 3910 + }, + { + "epoch": 1.8613485280151947, + "grad_norm": 0.11596999332746909, + "learning_rate": 7.908134540728452e-05, + "loss": 0.0395, + "step": 3920 + }, + { + "epoch": 1.866096866096866, + "grad_norm": 0.10987225517374916, + "learning_rate": 7.894635699807533e-05, + "loss": 0.0398, + "step": 3930 + }, + { + "epoch": 1.8708452041785375, + "grad_norm": 0.1095005362591336, + "learning_rate": 7.881105049777901e-05, + "loss": 0.0369, + "step": 3940 + }, + { + "epoch": 1.875593542260209, + "grad_norm": 0.09175314541261252, + "learning_rate": 7.867542739327667e-05, + "loss": 0.0375, + "step": 3950 + }, + { + "epoch": 1.8803418803418803, + "grad_norm": 0.10251392587830827, + "learning_rate": 7.853948917492857e-05, + "loss": 0.0382, + "step": 3960 + }, + { + "epoch": 1.8850902184235516, + "grad_norm": 0.11367932391845673, + "learning_rate": 7.840323733655778e-05, + "loss": 0.0385, + "step": 3970 + }, + { + "epoch": 1.8898385565052231, + "grad_norm": 0.11210942475533581, + "learning_rate": 7.826667337543377e-05, + "loss": 0.0393, + "step": 3980 + }, + { + "epoch": 1.8945868945868947, + "grad_norm": 0.09953666941814746, + "learning_rate": 7.812979879225581e-05, + "loss": 0.0399, + "step": 3990 + }, + { + "epoch": 1.899335232668566, + "grad_norm": 0.10578619892342929, + "learning_rate": 7.799261509113671e-05, + "loss": 0.0371, + "step": 4000 + }, + { + "epoch": 1.9040835707502373, + "grad_norm": 0.1035550862879546, + "learning_rate": 7.78551237795861e-05, + "loss": 0.0375, + "step": 4010 + }, + { + "epoch": 1.9088319088319088, + "grad_norm": 0.09719119646668428, + "learning_rate": 7.771732636849399e-05, + "loss": 0.0387, + "step": 4020 + }, + { + "epoch": 1.9135802469135803, + "grad_norm": 0.09848468793751304, + "learning_rate": 7.757922437211405e-05, + "loss": 0.0375, + "step": 4030 + }, + { + "epoch": 1.9183285849952516, + "grad_norm": 0.09680824681505751, + "learning_rate": 7.744081930804707e-05, + "loss": 0.0376, + "step": 4040 + }, + { + "epoch": 1.9230769230769231, + "grad_norm": 0.10633032181488061, + "learning_rate": 7.730211269722427e-05, + "loss": 0.0377, + "step": 4050 + }, + { + "epoch": 1.9278252611585946, + "grad_norm": 0.10830381359201696, + "learning_rate": 7.716310606389052e-05, + "loss": 0.0379, + "step": 4060 + }, + { + "epoch": 1.932573599240266, + "grad_norm": 0.11169085536566575, + "learning_rate": 7.702380093558765e-05, + "loss": 0.0377, + "step": 4070 + }, + { + "epoch": 1.9373219373219372, + "grad_norm": 0.10595802842347714, + "learning_rate": 7.688419884313764e-05, + "loss": 0.0385, + "step": 4080 + }, + { + "epoch": 1.9420702754036088, + "grad_norm": 0.10946877027262536, + "learning_rate": 7.674430132062585e-05, + "loss": 0.0381, + "step": 4090 + }, + { + "epoch": 1.9468186134852803, + "grad_norm": 0.1308447006135116, + "learning_rate": 7.660410990538403e-05, + "loss": 0.038, + "step": 4100 + }, + { + "epoch": 1.9515669515669516, + "grad_norm": 0.10808573759173556, + "learning_rate": 7.646362613797356e-05, + "loss": 0.0389, + "step": 4110 + }, + { + "epoch": 1.956315289648623, + "grad_norm": 0.10640438972784244, + "learning_rate": 7.63228515621685e-05, + "loss": 0.0379, + "step": 4120 + }, + { + "epoch": 1.9610636277302944, + "grad_norm": 0.096121967911724, + "learning_rate": 7.618178772493856e-05, + "loss": 0.0372, + "step": 4130 + }, + { + "epoch": 1.965811965811966, + "grad_norm": 0.09845563937150903, + "learning_rate": 7.604043617643215e-05, + "loss": 0.0377, + "step": 4140 + }, + { + "epoch": 1.9705603038936372, + "grad_norm": 0.11739429835232018, + "learning_rate": 7.589879846995934e-05, + "loss": 0.0383, + "step": 4150 + }, + { + "epoch": 1.9753086419753085, + "grad_norm": 0.10061903447779745, + "learning_rate": 7.575687616197477e-05, + "loss": 0.0382, + "step": 4160 + }, + { + "epoch": 1.98005698005698, + "grad_norm": 0.09857803787521291, + "learning_rate": 7.561467081206061e-05, + "loss": 0.0375, + "step": 4170 + }, + { + "epoch": 1.9848053181386516, + "grad_norm": 0.1045133963823532, + "learning_rate": 7.547218398290932e-05, + "loss": 0.0371, + "step": 4180 + }, + { + "epoch": 1.9895536562203229, + "grad_norm": 0.11006077796479521, + "learning_rate": 7.532941724030654e-05, + "loss": 0.0376, + "step": 4190 + }, + { + "epoch": 1.9943019943019942, + "grad_norm": 0.09479531156698223, + "learning_rate": 7.518637215311388e-05, + "loss": 0.0366, + "step": 4200 + }, + { + "epoch": 1.9990503323836657, + "grad_norm": 0.11104931355392686, + "learning_rate": 7.504305029325172e-05, + "loss": 0.0373, + "step": 4210 + }, + { + "epoch": 2.003798670465337, + "grad_norm": 0.09901902759873075, + "learning_rate": 7.489945323568178e-05, + "loss": 0.0359, + "step": 4220 + }, + { + "epoch": 2.0085470085470085, + "grad_norm": 0.09044738766499129, + "learning_rate": 7.475558255839006e-05, + "loss": 0.0361, + "step": 4230 + }, + { + "epoch": 2.01329534662868, + "grad_norm": 0.09035558123186714, + "learning_rate": 7.461143984236924e-05, + "loss": 0.0347, + "step": 4240 + }, + { + "epoch": 2.0180436847103516, + "grad_norm": 0.10228786902869633, + "learning_rate": 7.446702667160154e-05, + "loss": 0.036, + "step": 4250 + }, + { + "epoch": 2.022792022792023, + "grad_norm": 0.09137854312594297, + "learning_rate": 7.432234463304111e-05, + "loss": 0.0357, + "step": 4260 + }, + { + "epoch": 2.027540360873694, + "grad_norm": 0.10233750898621503, + "learning_rate": 7.417739531659674e-05, + "loss": 0.0346, + "step": 4270 + }, + { + "epoch": 2.0322886989553655, + "grad_norm": 0.10368178984629588, + "learning_rate": 7.403218031511428e-05, + "loss": 0.0347, + "step": 4280 + }, + { + "epoch": 2.037037037037037, + "grad_norm": 0.09336447522859606, + "learning_rate": 7.388670122435928e-05, + "loss": 0.035, + "step": 4290 + }, + { + "epoch": 2.0417853751187085, + "grad_norm": 0.11221938731289963, + "learning_rate": 7.374095964299928e-05, + "loss": 0.0357, + "step": 4300 + }, + { + "epoch": 2.04653371320038, + "grad_norm": 0.09350916834076561, + "learning_rate": 7.359495717258636e-05, + "loss": 0.0342, + "step": 4310 + }, + { + "epoch": 2.051282051282051, + "grad_norm": 0.08959451641987871, + "learning_rate": 7.34486954175395e-05, + "loss": 0.0347, + "step": 4320 + }, + { + "epoch": 2.056030389363723, + "grad_norm": 0.09792606298597017, + "learning_rate": 7.330217598512695e-05, + "loss": 0.0343, + "step": 4330 + }, + { + "epoch": 2.060778727445394, + "grad_norm": 0.09676674745093139, + "learning_rate": 7.315540048544857e-05, + "loss": 0.035, + "step": 4340 + }, + { + "epoch": 2.0655270655270654, + "grad_norm": 0.09882384161009967, + "learning_rate": 7.300837053141815e-05, + "loss": 0.0354, + "step": 4350 + }, + { + "epoch": 2.0702754036087367, + "grad_norm": 0.10163312651909884, + "learning_rate": 7.286108773874566e-05, + "loss": 0.0355, + "step": 4360 + }, + { + "epoch": 2.0750237416904085, + "grad_norm": 0.08755657890363665, + "learning_rate": 7.271355372591956e-05, + "loss": 0.0351, + "step": 4370 + }, + { + "epoch": 2.07977207977208, + "grad_norm": 0.09616063252354044, + "learning_rate": 7.256577011418888e-05, + "loss": 0.0352, + "step": 4380 + }, + { + "epoch": 2.084520417853751, + "grad_norm": 0.1042085822658408, + "learning_rate": 7.241773852754556e-05, + "loss": 0.0344, + "step": 4390 + }, + { + "epoch": 2.0892687559354224, + "grad_norm": 0.09424372909288635, + "learning_rate": 7.226946059270651e-05, + "loss": 0.0346, + "step": 4400 + }, + { + "epoch": 2.094017094017094, + "grad_norm": 0.08753879194752809, + "learning_rate": 7.212093793909573e-05, + "loss": 0.0344, + "step": 4410 + }, + { + "epoch": 2.0987654320987654, + "grad_norm": 0.09286065824287698, + "learning_rate": 7.197217219882647e-05, + "loss": 0.0359, + "step": 4420 + }, + { + "epoch": 2.1035137701804367, + "grad_norm": 0.09687944356866374, + "learning_rate": 7.182316500668326e-05, + "loss": 0.0338, + "step": 4430 + }, + { + "epoch": 2.1082621082621085, + "grad_norm": 0.1037432099543057, + "learning_rate": 7.167391800010389e-05, + "loss": 0.0342, + "step": 4440 + }, + { + "epoch": 2.1130104463437798, + "grad_norm": 0.09348992840947323, + "learning_rate": 7.152443281916151e-05, + "loss": 0.0345, + "step": 4450 + }, + { + "epoch": 2.117758784425451, + "grad_norm": 0.08519410626476705, + "learning_rate": 7.137471110654656e-05, + "loss": 0.0335, + "step": 4460 + }, + { + "epoch": 2.1225071225071224, + "grad_norm": 0.09600514998461258, + "learning_rate": 7.122475450754869e-05, + "loss": 0.0343, + "step": 4470 + }, + { + "epoch": 2.127255460588794, + "grad_norm": 0.08967447752461549, + "learning_rate": 7.107456467003877e-05, + "loss": 0.0347, + "step": 4480 + }, + { + "epoch": 2.1320037986704654, + "grad_norm": 0.08793399748502817, + "learning_rate": 7.092414324445068e-05, + "loss": 0.0349, + "step": 4490 + }, + { + "epoch": 2.1367521367521367, + "grad_norm": 0.09021397815292795, + "learning_rate": 7.077349188376325e-05, + "loss": 0.0343, + "step": 4500 + }, + { + "epoch": 2.141500474833808, + "grad_norm": 0.09240884777974796, + "learning_rate": 7.062261224348203e-05, + "loss": 0.0353, + "step": 4510 + }, + { + "epoch": 2.1462488129154798, + "grad_norm": 0.0876446538153716, + "learning_rate": 7.047150598162114e-05, + "loss": 0.0346, + "step": 4520 + }, + { + "epoch": 2.150997150997151, + "grad_norm": 0.08557666534843845, + "learning_rate": 7.032017475868505e-05, + "loss": 0.0355, + "step": 4530 + }, + { + "epoch": 2.1557454890788224, + "grad_norm": 0.0981449847862032, + "learning_rate": 7.016862023765033e-05, + "loss": 0.0343, + "step": 4540 + }, + { + "epoch": 2.1604938271604937, + "grad_norm": 0.09344850520986868, + "learning_rate": 7.001684408394735e-05, + "loss": 0.0362, + "step": 4550 + }, + { + "epoch": 2.1652421652421654, + "grad_norm": 0.09302078618237986, + "learning_rate": 6.986484796544203e-05, + "loss": 0.0345, + "step": 4560 + }, + { + "epoch": 2.1699905033238367, + "grad_norm": 0.08954079955060139, + "learning_rate": 6.971263355241742e-05, + "loss": 0.0355, + "step": 4570 + }, + { + "epoch": 2.174738841405508, + "grad_norm": 0.11938380276574542, + "learning_rate": 6.956020251755548e-05, + "loss": 0.0354, + "step": 4580 + }, + { + "epoch": 2.1794871794871793, + "grad_norm": 0.09496069017011385, + "learning_rate": 6.940755653591858e-05, + "loss": 0.0332, + "step": 4590 + }, + { + "epoch": 2.184235517568851, + "grad_norm": 0.11463335680864938, + "learning_rate": 6.925469728493114e-05, + "loss": 0.0352, + "step": 4600 + }, + { + "epoch": 2.1889838556505223, + "grad_norm": 0.09594800187155825, + "learning_rate": 6.91016264443612e-05, + "loss": 0.0332, + "step": 4610 + }, + { + "epoch": 2.1937321937321936, + "grad_norm": 0.09279419149618479, + "learning_rate": 6.894834569630195e-05, + "loss": 0.0342, + "step": 4620 + }, + { + "epoch": 2.198480531813865, + "grad_norm": 0.09573345946376917, + "learning_rate": 6.879485672515327e-05, + "loss": 0.0355, + "step": 4630 + }, + { + "epoch": 2.2032288698955367, + "grad_norm": 0.09889624215928858, + "learning_rate": 6.864116121760316e-05, + "loss": 0.0347, + "step": 4640 + }, + { + "epoch": 2.207977207977208, + "grad_norm": 0.09247023961197472, + "learning_rate": 6.848726086260929e-05, + "loss": 0.0345, + "step": 4650 + }, + { + "epoch": 2.2127255460588793, + "grad_norm": 0.09017590727692242, + "learning_rate": 6.833315735138035e-05, + "loss": 0.0338, + "step": 4660 + }, + { + "epoch": 2.217473884140551, + "grad_norm": 0.09624446296259595, + "learning_rate": 6.817885237735758e-05, + "loss": 0.0345, + "step": 4670 + }, + { + "epoch": 2.2222222222222223, + "grad_norm": 0.08813418195946005, + "learning_rate": 6.802434763619599e-05, + "loss": 0.035, + "step": 4680 + }, + { + "epoch": 2.2269705603038936, + "grad_norm": 0.0928067372275443, + "learning_rate": 6.786964482574594e-05, + "loss": 0.0348, + "step": 4690 + }, + { + "epoch": 2.231718898385565, + "grad_norm": 0.09067064908235425, + "learning_rate": 6.771474564603428e-05, + "loss": 0.0348, + "step": 4700 + }, + { + "epoch": 2.2364672364672367, + "grad_norm": 0.09043961450448, + "learning_rate": 6.755965179924582e-05, + "loss": 0.035, + "step": 4710 + }, + { + "epoch": 2.241215574548908, + "grad_norm": 0.09797850107949971, + "learning_rate": 6.740436498970452e-05, + "loss": 0.0336, + "step": 4720 + }, + { + "epoch": 2.2459639126305793, + "grad_norm": 0.09341151212734902, + "learning_rate": 6.724888692385485e-05, + "loss": 0.0342, + "step": 4730 + }, + { + "epoch": 2.2507122507122506, + "grad_norm": 0.08433145633974774, + "learning_rate": 6.709321931024293e-05, + "loss": 0.0343, + "step": 4740 + }, + { + "epoch": 2.2554605887939223, + "grad_norm": 0.09212629097306463, + "learning_rate": 6.693736385949788e-05, + "loss": 0.0348, + "step": 4750 + }, + { + "epoch": 2.2602089268755936, + "grad_norm": 0.10021414221767545, + "learning_rate": 6.678132228431294e-05, + "loss": 0.0349, + "step": 4760 + }, + { + "epoch": 2.264957264957265, + "grad_norm": 0.09708783825249576, + "learning_rate": 6.662509629942661e-05, + "loss": 0.0351, + "step": 4770 + }, + { + "epoch": 2.269705603038936, + "grad_norm": 0.08251106750374618, + "learning_rate": 6.646868762160399e-05, + "loss": 0.0339, + "step": 4780 + }, + { + "epoch": 2.274453941120608, + "grad_norm": 0.08887253040170329, + "learning_rate": 6.631209796961767e-05, + "loss": 0.0348, + "step": 4790 + }, + { + "epoch": 2.2792022792022792, + "grad_norm": 0.09455801753780131, + "learning_rate": 6.615532906422902e-05, + "loss": 0.034, + "step": 4800 + }, + { + "epoch": 2.2839506172839505, + "grad_norm": 0.09113060797736637, + "learning_rate": 6.599838262816922e-05, + "loss": 0.0333, + "step": 4810 + }, + { + "epoch": 2.288698955365622, + "grad_norm": 0.08867092462982426, + "learning_rate": 6.584126038612033e-05, + "loss": 0.0341, + "step": 4820 + }, + { + "epoch": 2.2934472934472936, + "grad_norm": 0.09330602355448789, + "learning_rate": 6.568396406469632e-05, + "loss": 0.0354, + "step": 4830 + }, + { + "epoch": 2.298195631528965, + "grad_norm": 0.09329260278025318, + "learning_rate": 6.552649539242414e-05, + "loss": 0.0339, + "step": 4840 + }, + { + "epoch": 2.302943969610636, + "grad_norm": 0.09335751440976497, + "learning_rate": 6.536885609972467e-05, + "loss": 0.0332, + "step": 4850 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.08964217300974713, + "learning_rate": 6.521104791889381e-05, + "loss": 0.0335, + "step": 4860 + }, + { + "epoch": 2.3124406457739792, + "grad_norm": 0.08771704198156961, + "learning_rate": 6.505307258408325e-05, + "loss": 0.0328, + "step": 4870 + }, + { + "epoch": 2.3171889838556505, + "grad_norm": 0.08563217815035286, + "learning_rate": 6.489493183128163e-05, + "loss": 0.0349, + "step": 4880 + }, + { + "epoch": 2.321937321937322, + "grad_norm": 0.08703017568879401, + "learning_rate": 6.473662739829533e-05, + "loss": 0.0349, + "step": 4890 + }, + { + "epoch": 2.3266856600189936, + "grad_norm": 0.0903431704137305, + "learning_rate": 6.457816102472942e-05, + "loss": 0.0343, + "step": 4900 + }, + { + "epoch": 2.331433998100665, + "grad_norm": 0.09582830483751793, + "learning_rate": 6.44195344519685e-05, + "loss": 0.0346, + "step": 4910 + }, + { + "epoch": 2.336182336182336, + "grad_norm": 0.09314671437325292, + "learning_rate": 6.426074942315766e-05, + "loss": 0.0346, + "step": 4920 + }, + { + "epoch": 2.3409306742640075, + "grad_norm": 0.09283980937584103, + "learning_rate": 6.410180768318321e-05, + "loss": 0.0343, + "step": 4930 + }, + { + "epoch": 2.3456790123456788, + "grad_norm": 0.09194221689766775, + "learning_rate": 6.394271097865355e-05, + "loss": 0.0344, + "step": 4940 + }, + { + "epoch": 2.3504273504273505, + "grad_norm": 0.09338170017959663, + "learning_rate": 6.378346105788002e-05, + "loss": 0.0343, + "step": 4950 + }, + { + "epoch": 2.355175688509022, + "grad_norm": 0.0843221023820531, + "learning_rate": 6.362405967085765e-05, + "loss": 0.0351, + "step": 4960 + }, + { + "epoch": 2.359924026590693, + "grad_norm": 0.09093072053397361, + "learning_rate": 6.346450856924588e-05, + "loss": 0.0345, + "step": 4970 + }, + { + "epoch": 2.364672364672365, + "grad_norm": 0.08467301735339894, + "learning_rate": 6.330480950634942e-05, + "loss": 0.0354, + "step": 4980 + }, + { + "epoch": 2.369420702754036, + "grad_norm": 0.09220329631608627, + "learning_rate": 6.314496423709887e-05, + "loss": 0.0346, + "step": 4990 + }, + { + "epoch": 2.3741690408357075, + "grad_norm": 0.11669195629783755, + "learning_rate": 6.298497451803157e-05, + "loss": 0.0336, + "step": 5000 + }, + { + "epoch": 2.3789173789173788, + "grad_norm": 0.0837059010274995, + "learning_rate": 6.282484210727213e-05, + "loss": 0.0336, + "step": 5010 + }, + { + "epoch": 2.3836657169990505, + "grad_norm": 0.0777465102926546, + "learning_rate": 6.266456876451324e-05, + "loss": 0.0339, + "step": 5020 + }, + { + "epoch": 2.388414055080722, + "grad_norm": 0.07925241006073065, + "learning_rate": 6.25041562509963e-05, + "loss": 0.0338, + "step": 5030 + }, + { + "epoch": 2.393162393162393, + "grad_norm": 0.08902968852474427, + "learning_rate": 6.234360632949201e-05, + "loss": 0.034, + "step": 5040 + }, + { + "epoch": 2.3979107312440644, + "grad_norm": 0.0850286570272343, + "learning_rate": 6.21829207642811e-05, + "loss": 0.0338, + "step": 5050 + }, + { + "epoch": 2.402659069325736, + "grad_norm": 0.09192464441163431, + "learning_rate": 6.202210132113484e-05, + "loss": 0.034, + "step": 5060 + }, + { + "epoch": 2.4074074074074074, + "grad_norm": 0.0859288767704206, + "learning_rate": 6.186114976729573e-05, + "loss": 0.0333, + "step": 5070 + }, + { + "epoch": 2.4121557454890787, + "grad_norm": 0.09612093422784568, + "learning_rate": 6.170006787145796e-05, + "loss": 0.0343, + "step": 5080 + }, + { + "epoch": 2.41690408357075, + "grad_norm": 0.07697302839474403, + "learning_rate": 6.153885740374813e-05, + "loss": 0.0335, + "step": 5090 + }, + { + "epoch": 2.421652421652422, + "grad_norm": 0.07942688550257924, + "learning_rate": 6.137752013570566e-05, + "loss": 0.0334, + "step": 5100 + }, + { + "epoch": 2.426400759734093, + "grad_norm": 0.08437077188433638, + "learning_rate": 6.121605784026339e-05, + "loss": 0.0331, + "step": 5110 + }, + { + "epoch": 2.4311490978157644, + "grad_norm": 0.09371906221231797, + "learning_rate": 6.10544722917281e-05, + "loss": 0.0333, + "step": 5120 + }, + { + "epoch": 2.435897435897436, + "grad_norm": 0.08190966003954078, + "learning_rate": 6.089276526576094e-05, + "loss": 0.0341, + "step": 5130 + }, + { + "epoch": 2.4406457739791074, + "grad_norm": 0.08682116901407358, + "learning_rate": 6.0730938539358084e-05, + "loss": 0.0336, + "step": 5140 + }, + { + "epoch": 2.4453941120607787, + "grad_norm": 0.08626462798982308, + "learning_rate": 6.056899389083097e-05, + "loss": 0.0344, + "step": 5150 + }, + { + "epoch": 2.45014245014245, + "grad_norm": 0.08310971368425704, + "learning_rate": 6.0406933099786976e-05, + "loss": 0.0333, + "step": 5160 + }, + { + "epoch": 2.4548907882241213, + "grad_norm": 0.0848141697366201, + "learning_rate": 6.02447579471097e-05, + "loss": 0.0339, + "step": 5170 + }, + { + "epoch": 2.459639126305793, + "grad_norm": 0.08365783825318655, + "learning_rate": 6.0082470214939514e-05, + "loss": 0.0344, + "step": 5180 + }, + { + "epoch": 2.4643874643874644, + "grad_norm": 0.08924875057937963, + "learning_rate": 5.9920071686653855e-05, + "loss": 0.0337, + "step": 5190 + }, + { + "epoch": 2.4691358024691357, + "grad_norm": 0.08573158164010335, + "learning_rate": 5.9757564146847764e-05, + "loss": 0.0331, + "step": 5200 + }, + { + "epoch": 2.4738841405508074, + "grad_norm": 0.0873924916278996, + "learning_rate": 5.959494938131418e-05, + "loss": 0.0335, + "step": 5210 + }, + { + "epoch": 2.4786324786324787, + "grad_norm": 0.0943764254085368, + "learning_rate": 5.9432229177024346e-05, + "loss": 0.0346, + "step": 5220 + }, + { + "epoch": 2.48338081671415, + "grad_norm": 0.08688958700772889, + "learning_rate": 5.926940532210817e-05, + "loss": 0.0324, + "step": 5230 + }, + { + "epoch": 2.4881291547958213, + "grad_norm": 0.09474305720603773, + "learning_rate": 5.910647960583457e-05, + "loss": 0.0341, + "step": 5240 + }, + { + "epoch": 2.492877492877493, + "grad_norm": 0.0788004210782281, + "learning_rate": 5.8943453818591845e-05, + "loss": 0.0327, + "step": 5250 + }, + { + "epoch": 2.4976258309591644, + "grad_norm": 0.08035818585954843, + "learning_rate": 5.878032975186791e-05, + "loss": 0.0336, + "step": 5260 + }, + { + "epoch": 2.5023741690408356, + "grad_norm": 0.08272796448240387, + "learning_rate": 5.8617109198230744e-05, + "loss": 0.0342, + "step": 5270 + }, + { + "epoch": 2.5071225071225074, + "grad_norm": 0.07851015077004643, + "learning_rate": 5.845379395130859e-05, + "loss": 0.0338, + "step": 5280 + }, + { + "epoch": 2.5118708452041787, + "grad_norm": 0.08768072861251815, + "learning_rate": 5.829038580577025e-05, + "loss": 0.0339, + "step": 5290 + }, + { + "epoch": 2.51661918328585, + "grad_norm": 0.07602139516611504, + "learning_rate": 5.812688655730542e-05, + "loss": 0.0347, + "step": 5300 + }, + { + "epoch": 2.5213675213675213, + "grad_norm": 0.08168975179045737, + "learning_rate": 5.796329800260489e-05, + "loss": 0.0337, + "step": 5310 + }, + { + "epoch": 2.5261158594491926, + "grad_norm": 0.07586189377954826, + "learning_rate": 5.7799621939340896e-05, + "loss": 0.0346, + "step": 5320 + }, + { + "epoch": 2.5308641975308643, + "grad_norm": 0.10544351179783236, + "learning_rate": 5.7635860166147246e-05, + "loss": 0.0335, + "step": 5330 + }, + { + "epoch": 2.5356125356125356, + "grad_norm": 0.08851930600411412, + "learning_rate": 5.747201448259964e-05, + "loss": 0.0338, + "step": 5340 + }, + { + "epoch": 2.540360873694207, + "grad_norm": 0.07986735105542157, + "learning_rate": 5.730808668919584e-05, + "loss": 0.0329, + "step": 5350 + }, + { + "epoch": 2.5451092117758787, + "grad_norm": 0.09002847625577425, + "learning_rate": 5.714407858733596e-05, + "loss": 0.0337, + "step": 5360 + }, + { + "epoch": 2.54985754985755, + "grad_norm": 0.08302785878003045, + "learning_rate": 5.6979991979302585e-05, + "loss": 0.033, + "step": 5370 + }, + { + "epoch": 2.5546058879392213, + "grad_norm": 0.07101583021993221, + "learning_rate": 5.6815828668241e-05, + "loss": 0.0338, + "step": 5380 + }, + { + "epoch": 2.5593542260208926, + "grad_norm": 0.08490239836935078, + "learning_rate": 5.6651590458139424e-05, + "loss": 0.0345, + "step": 5390 + }, + { + "epoch": 2.564102564102564, + "grad_norm": 0.08794874702660714, + "learning_rate": 5.6487279153809094e-05, + "loss": 0.0351, + "step": 5400 + }, + { + "epoch": 2.5688509021842356, + "grad_norm": 0.08897552796361904, + "learning_rate": 5.632289656086448e-05, + "loss": 0.0335, + "step": 5410 + }, + { + "epoch": 2.573599240265907, + "grad_norm": 0.07744983532156134, + "learning_rate": 5.6158444485703465e-05, + "loss": 0.0333, + "step": 5420 + }, + { + "epoch": 2.578347578347578, + "grad_norm": 0.0850222180212384, + "learning_rate": 5.5993924735487455e-05, + "loss": 0.0348, + "step": 5430 + }, + { + "epoch": 2.58309591642925, + "grad_norm": 0.09546245688140068, + "learning_rate": 5.582933911812155e-05, + "loss": 0.0333, + "step": 5440 + }, + { + "epoch": 2.5878442545109213, + "grad_norm": 0.08868077458189247, + "learning_rate": 5.5664689442234654e-05, + "loss": 0.0326, + "step": 5450 + }, + { + "epoch": 2.5925925925925926, + "grad_norm": 0.08389268094086333, + "learning_rate": 5.549997751715961e-05, + "loss": 0.0348, + "step": 5460 + }, + { + "epoch": 2.597340930674264, + "grad_norm": 0.08439379465615206, + "learning_rate": 5.5335205152913305e-05, + "loss": 0.0331, + "step": 5470 + }, + { + "epoch": 2.602089268755935, + "grad_norm": 0.07893691203141218, + "learning_rate": 5.5170374160176815e-05, + "loss": 0.0325, + "step": 5480 + }, + { + "epoch": 2.606837606837607, + "grad_norm": 0.07782232734328874, + "learning_rate": 5.500548635027547e-05, + "loss": 0.0336, + "step": 5490 + }, + { + "epoch": 2.611585944919278, + "grad_norm": 0.09952115864193736, + "learning_rate": 5.484054353515896e-05, + "loss": 0.0332, + "step": 5500 + }, + { + "epoch": 2.61633428300095, + "grad_norm": 0.07760862072481986, + "learning_rate": 5.467554752738143e-05, + "loss": 0.0334, + "step": 5510 + }, + { + "epoch": 2.6210826210826212, + "grad_norm": 0.0839520510103637, + "learning_rate": 5.451050014008157e-05, + "loss": 0.0328, + "step": 5520 + }, + { + "epoch": 2.6258309591642925, + "grad_norm": 0.09546198327872597, + "learning_rate": 5.434540318696265e-05, + "loss": 0.0334, + "step": 5530 + }, + { + "epoch": 2.630579297245964, + "grad_norm": 0.08894242200177575, + "learning_rate": 5.418025848227265e-05, + "loss": 0.0338, + "step": 5540 + }, + { + "epoch": 2.635327635327635, + "grad_norm": 0.07806515919199125, + "learning_rate": 5.401506784078427e-05, + "loss": 0.0333, + "step": 5550 + }, + { + "epoch": 2.640075973409307, + "grad_norm": 0.08569053282178216, + "learning_rate": 5.384983307777502e-05, + "loss": 0.0334, + "step": 5560 + }, + { + "epoch": 2.644824311490978, + "grad_norm": 0.08412407397057373, + "learning_rate": 5.368455600900725e-05, + "loss": 0.0326, + "step": 5570 + }, + { + "epoch": 2.6495726495726495, + "grad_norm": 0.08913781133577336, + "learning_rate": 5.351923845070822e-05, + "loss": 0.0327, + "step": 5580 + }, + { + "epoch": 2.6543209876543212, + "grad_norm": 0.09035361325193161, + "learning_rate": 5.335388221955012e-05, + "loss": 0.0338, + "step": 5590 + }, + { + "epoch": 2.6590693257359925, + "grad_norm": 0.08113510582952685, + "learning_rate": 5.3188489132630124e-05, + "loss": 0.0335, + "step": 5600 + }, + { + "epoch": 2.663817663817664, + "grad_norm": 0.0863540846158926, + "learning_rate": 5.3023061007450404e-05, + "loss": 0.033, + "step": 5610 + }, + { + "epoch": 2.668566001899335, + "grad_norm": 0.08676058975387343, + "learning_rate": 5.285759966189817e-05, + "loss": 0.0339, + "step": 5620 + }, + { + "epoch": 2.6733143399810064, + "grad_norm": 0.07748461161349562, + "learning_rate": 5.269210691422569e-05, + "loss": 0.0327, + "step": 5630 + }, + { + "epoch": 2.678062678062678, + "grad_norm": 0.08865483692725692, + "learning_rate": 5.252658458303032e-05, + "loss": 0.033, + "step": 5640 + }, + { + "epoch": 2.6828110161443495, + "grad_norm": 0.1103858862256417, + "learning_rate": 5.236103448723448e-05, + "loss": 0.0331, + "step": 5650 + }, + { + "epoch": 2.6875593542260208, + "grad_norm": 0.08498267556088492, + "learning_rate": 5.219545844606573e-05, + "loss": 0.0338, + "step": 5660 + }, + { + "epoch": 2.6923076923076925, + "grad_norm": 0.08695964093895549, + "learning_rate": 5.202985827903672e-05, + "loss": 0.0333, + "step": 5670 + }, + { + "epoch": 2.697056030389364, + "grad_norm": 0.08268158087195521, + "learning_rate": 5.186423580592523e-05, + "loss": 0.0333, + "step": 5680 + }, + { + "epoch": 2.701804368471035, + "grad_norm": 0.08814532688958854, + "learning_rate": 5.1698592846754156e-05, + "loss": 0.0323, + "step": 5690 + }, + { + "epoch": 2.7065527065527064, + "grad_norm": 0.07370283756044843, + "learning_rate": 5.1532931221771515e-05, + "loss": 0.0317, + "step": 5700 + }, + { + "epoch": 2.7113010446343777, + "grad_norm": 0.0750319782142584, + "learning_rate": 5.136725275143045e-05, + "loss": 0.0331, + "step": 5710 + }, + { + "epoch": 2.7160493827160495, + "grad_norm": 0.07793618212842458, + "learning_rate": 5.1201559256369215e-05, + "loss": 0.0333, + "step": 5720 + }, + { + "epoch": 2.7207977207977208, + "grad_norm": 0.08048496157622094, + "learning_rate": 5.103585255739113e-05, + "loss": 0.0339, + "step": 5730 + }, + { + "epoch": 2.725546058879392, + "grad_norm": 0.09199765334597354, + "learning_rate": 5.087013447544469e-05, + "loss": 0.0336, + "step": 5740 + }, + { + "epoch": 2.730294396961064, + "grad_norm": 0.0901164374025455, + "learning_rate": 5.0704406831603414e-05, + "loss": 0.0334, + "step": 5750 + }, + { + "epoch": 2.735042735042735, + "grad_norm": 0.08568998211931343, + "learning_rate": 5.0538671447045936e-05, + "loss": 0.0339, + "step": 5760 + }, + { + "epoch": 2.7397910731244064, + "grad_norm": 0.07776745161612766, + "learning_rate": 5.037293014303592e-05, + "loss": 0.0335, + "step": 5770 + }, + { + "epoch": 2.7445394112060777, + "grad_norm": 0.07869367205887913, + "learning_rate": 5.020718474090212e-05, + "loss": 0.0326, + "step": 5780 + }, + { + "epoch": 2.7492877492877494, + "grad_norm": 0.09333052392470734, + "learning_rate": 5.004143706201826e-05, + "loss": 0.032, + "step": 5790 + }, + { + "epoch": 2.7540360873694207, + "grad_norm": 0.07902272699505779, + "learning_rate": 4.987568892778314e-05, + "loss": 0.0333, + "step": 5800 + }, + { + "epoch": 2.758784425451092, + "grad_norm": 0.08608290579470909, + "learning_rate": 4.970994215960056e-05, + "loss": 0.0334, + "step": 5810 + }, + { + "epoch": 2.763532763532764, + "grad_norm": 0.08622248147245169, + "learning_rate": 4.954419857885926e-05, + "loss": 0.0339, + "step": 5820 + }, + { + "epoch": 2.768281101614435, + "grad_norm": 0.07550573786537534, + "learning_rate": 4.9378460006913035e-05, + "loss": 0.0322, + "step": 5830 + }, + { + "epoch": 2.7730294396961064, + "grad_norm": 0.07179958721667509, + "learning_rate": 4.921272826506054e-05, + "loss": 0.033, + "step": 5840 + }, + { + "epoch": 2.7777777777777777, + "grad_norm": 0.08996649806858173, + "learning_rate": 4.9047005174525454e-05, + "loss": 0.0335, + "step": 5850 + }, + { + "epoch": 2.782526115859449, + "grad_norm": 0.08028115394984245, + "learning_rate": 4.8881292556436354e-05, + "loss": 0.033, + "step": 5860 + }, + { + "epoch": 2.7872744539411207, + "grad_norm": 0.08236732278495794, + "learning_rate": 4.871559223180673e-05, + "loss": 0.0325, + "step": 5870 + }, + { + "epoch": 2.792022792022792, + "grad_norm": 0.07953172482680695, + "learning_rate": 4.854990602151502e-05, + "loss": 0.033, + "step": 5880 + }, + { + "epoch": 2.7967711301044633, + "grad_norm": 0.0864278189435432, + "learning_rate": 4.8384235746284476e-05, + "loss": 0.0332, + "step": 5890 + }, + { + "epoch": 2.801519468186135, + "grad_norm": 0.07441734801951162, + "learning_rate": 4.821858322666333e-05, + "loss": 0.0327, + "step": 5900 + }, + { + "epoch": 2.8062678062678064, + "grad_norm": 0.07016442981647984, + "learning_rate": 4.8052950283004637e-05, + "loss": 0.0321, + "step": 5910 + }, + { + "epoch": 2.8110161443494777, + "grad_norm": 0.08988241361094701, + "learning_rate": 4.7887338735446394e-05, + "loss": 0.0324, + "step": 5920 + }, + { + "epoch": 2.815764482431149, + "grad_norm": 0.08510857900394272, + "learning_rate": 4.772175040389138e-05, + "loss": 0.0322, + "step": 5930 + }, + { + "epoch": 2.8205128205128203, + "grad_norm": 0.08134840628339234, + "learning_rate": 4.7556187107987365e-05, + "loss": 0.033, + "step": 5940 + }, + { + "epoch": 2.825261158594492, + "grad_norm": 0.08178093110739178, + "learning_rate": 4.73906506671069e-05, + "loss": 0.0313, + "step": 5950 + }, + { + "epoch": 2.8300094966761633, + "grad_norm": 0.08331545825081525, + "learning_rate": 4.722514290032753e-05, + "loss": 0.0332, + "step": 5960 + }, + { + "epoch": 2.8347578347578346, + "grad_norm": 0.08440284471680276, + "learning_rate": 4.70596656264116e-05, + "loss": 0.0315, + "step": 5970 + }, + { + "epoch": 2.8395061728395063, + "grad_norm": 0.08453818985943418, + "learning_rate": 4.689422066378644e-05, + "loss": 0.0327, + "step": 5980 + }, + { + "epoch": 2.8442545109211776, + "grad_norm": 0.07900543723847953, + "learning_rate": 4.6728809830524285e-05, + "loss": 0.0327, + "step": 5990 + }, + { + "epoch": 2.849002849002849, + "grad_norm": 0.08483359082820749, + "learning_rate": 4.65634349443223e-05, + "loss": 0.0332, + "step": 6000 + }, + { + "epoch": 2.8537511870845202, + "grad_norm": 0.08792512749424275, + "learning_rate": 4.63980978224827e-05, + "loss": 0.0341, + "step": 6010 + }, + { + "epoch": 2.858499525166192, + "grad_norm": 0.0765735917061693, + "learning_rate": 4.62328002818926e-05, + "loss": 0.0314, + "step": 6020 + }, + { + "epoch": 2.8632478632478633, + "grad_norm": 0.08036249702478072, + "learning_rate": 4.606754413900428e-05, + "loss": 0.0324, + "step": 6030 + }, + { + "epoch": 2.8679962013295346, + "grad_norm": 0.2370846996983532, + "learning_rate": 4.5902331209814995e-05, + "loss": 0.0343, + "step": 6040 + }, + { + "epoch": 2.8727445394112063, + "grad_norm": 0.08002029230906148, + "learning_rate": 4.5737163309847205e-05, + "loss": 0.0327, + "step": 6050 + }, + { + "epoch": 2.8774928774928776, + "grad_norm": 0.07430602370809676, + "learning_rate": 4.557204225412848e-05, + "loss": 0.0323, + "step": 6060 + }, + { + "epoch": 2.882241215574549, + "grad_norm": 0.08952791722219931, + "learning_rate": 4.540696985717168e-05, + "loss": 0.0327, + "step": 6070 + }, + { + "epoch": 2.8869895536562202, + "grad_norm": 0.08145019589336985, + "learning_rate": 4.524194793295488e-05, + "loss": 0.0328, + "step": 6080 + }, + { + "epoch": 2.8917378917378915, + "grad_norm": 0.07296975142079583, + "learning_rate": 4.5076978294901604e-05, + "loss": 0.0332, + "step": 6090 + }, + { + "epoch": 2.8964862298195633, + "grad_norm": 0.0795919132167877, + "learning_rate": 4.491206275586072e-05, + "loss": 0.0331, + "step": 6100 + }, + { + "epoch": 2.9012345679012346, + "grad_norm": 0.08332062451714566, + "learning_rate": 4.4747203128086634e-05, + "loss": 0.0331, + "step": 6110 + }, + { + "epoch": 2.905982905982906, + "grad_norm": 0.07558136239210715, + "learning_rate": 4.458240122321939e-05, + "loss": 0.0318, + "step": 6120 + }, + { + "epoch": 2.9107312440645776, + "grad_norm": 0.08671627490918178, + "learning_rate": 4.4417658852264614e-05, + "loss": 0.0323, + "step": 6130 + }, + { + "epoch": 2.915479582146249, + "grad_norm": 0.07657140735008562, + "learning_rate": 4.425297782557383e-05, + "loss": 0.0322, + "step": 6140 + }, + { + "epoch": 2.92022792022792, + "grad_norm": 0.07733988730279348, + "learning_rate": 4.4088359952824344e-05, + "loss": 0.0321, + "step": 6150 + }, + { + "epoch": 2.9249762583095915, + "grad_norm": 0.08097146522816863, + "learning_rate": 4.392380704299955e-05, + "loss": 0.0331, + "step": 6160 + }, + { + "epoch": 2.929724596391263, + "grad_norm": 0.07138051536222574, + "learning_rate": 4.3759320904368913e-05, + "loss": 0.0319, + "step": 6170 + }, + { + "epoch": 2.9344729344729346, + "grad_norm": 0.0749233445405213, + "learning_rate": 4.359490334446819e-05, + "loss": 0.032, + "step": 6180 + }, + { + "epoch": 2.939221272554606, + "grad_norm": 0.07589504384837518, + "learning_rate": 4.343055617007947e-05, + "loss": 0.0329, + "step": 6190 + }, + { + "epoch": 2.943969610636277, + "grad_norm": 0.08816086253011907, + "learning_rate": 4.326628118721144e-05, + "loss": 0.0332, + "step": 6200 + }, + { + "epoch": 2.948717948717949, + "grad_norm": 0.07583652826801959, + "learning_rate": 4.3102080201079435e-05, + "loss": 0.0322, + "step": 6210 + }, + { + "epoch": 2.95346628679962, + "grad_norm": 0.07666512702883353, + "learning_rate": 4.293795501608565e-05, + "loss": 0.0335, + "step": 6220 + }, + { + "epoch": 2.9582146248812915, + "grad_norm": 0.07543819662947532, + "learning_rate": 4.277390743579932e-05, + "loss": 0.0329, + "step": 6230 + }, + { + "epoch": 2.962962962962963, + "grad_norm": 0.07653679521036545, + "learning_rate": 4.2609939262936835e-05, + "loss": 0.0328, + "step": 6240 + }, + { + "epoch": 2.9677113010446345, + "grad_norm": 0.07870936555696605, + "learning_rate": 4.244605229934206e-05, + "loss": 0.0321, + "step": 6250 + }, + { + "epoch": 2.972459639126306, + "grad_norm": 0.08121549620221498, + "learning_rate": 4.228224834596636e-05, + "loss": 0.0329, + "step": 6260 + }, + { + "epoch": 2.977207977207977, + "grad_norm": 0.0803079693648475, + "learning_rate": 4.211852920284899e-05, + "loss": 0.0317, + "step": 6270 + }, + { + "epoch": 2.981956315289649, + "grad_norm": 0.07063944307674028, + "learning_rate": 4.195489666909713e-05, + "loss": 0.0324, + "step": 6280 + }, + { + "epoch": 2.98670465337132, + "grad_norm": 0.07383966192406598, + "learning_rate": 4.179135254286631e-05, + "loss": 0.0316, + "step": 6290 + }, + { + "epoch": 2.9914529914529915, + "grad_norm": 0.07618791078597378, + "learning_rate": 4.162789862134047e-05, + "loss": 0.0325, + "step": 6300 + }, + { + "epoch": 2.996201329534663, + "grad_norm": 0.07808672389369174, + "learning_rate": 4.146453670071236e-05, + "loss": 0.0325, + "step": 6310 + }, + { + "epoch": 3.000949667616334, + "grad_norm": 0.0710582118314449, + "learning_rate": 4.1301268576163665e-05, + "loss": 0.0328, + "step": 6320 + }, + { + "epoch": 3.005698005698006, + "grad_norm": 0.06589896623166609, + "learning_rate": 4.11380960418454e-05, + "loss": 0.0296, + "step": 6330 + }, + { + "epoch": 3.010446343779677, + "grad_norm": 0.06967237480659132, + "learning_rate": 4.0975020890858115e-05, + "loss": 0.0301, + "step": 6340 + }, + { + "epoch": 3.0151946818613484, + "grad_norm": 0.06553355970224677, + "learning_rate": 4.081204491523222e-05, + "loss": 0.0304, + "step": 6350 + }, + { + "epoch": 3.0199430199430197, + "grad_norm": 0.06743478323451074, + "learning_rate": 4.06491699059083e-05, + "loss": 0.0297, + "step": 6360 + }, + { + "epoch": 3.0246913580246915, + "grad_norm": 0.06560199068132755, + "learning_rate": 4.04863976527174e-05, + "loss": 0.0297, + "step": 6370 + }, + { + "epoch": 3.0294396961063628, + "grad_norm": 0.06731435643345209, + "learning_rate": 4.032372994436142e-05, + "loss": 0.0303, + "step": 6380 + }, + { + "epoch": 3.034188034188034, + "grad_norm": 0.07103307210398627, + "learning_rate": 4.016116856839335e-05, + "loss": 0.0293, + "step": 6390 + }, + { + "epoch": 3.038936372269706, + "grad_norm": 0.06930448226386177, + "learning_rate": 3.9998715311197785e-05, + "loss": 0.0303, + "step": 6400 + }, + { + "epoch": 3.043684710351377, + "grad_norm": 0.07661535606482202, + "learning_rate": 3.9836371957971106e-05, + "loss": 0.0303, + "step": 6410 + }, + { + "epoch": 3.0484330484330484, + "grad_norm": 0.06183916217681361, + "learning_rate": 3.967414029270206e-05, + "loss": 0.0289, + "step": 6420 + }, + { + "epoch": 3.0531813865147197, + "grad_norm": 0.07001122792518905, + "learning_rate": 3.9512022098151966e-05, + "loss": 0.0291, + "step": 6430 + }, + { + "epoch": 3.0579297245963915, + "grad_norm": 0.06801738059574665, + "learning_rate": 3.935001915583526e-05, + "loss": 0.029, + "step": 6440 + }, + { + "epoch": 3.0626780626780628, + "grad_norm": 0.06118791099130076, + "learning_rate": 3.918813324599989e-05, + "loss": 0.0296, + "step": 6450 + }, + { + "epoch": 3.067426400759734, + "grad_norm": 0.07202748144589026, + "learning_rate": 3.902636614760771e-05, + "loss": 0.0294, + "step": 6460 + }, + { + "epoch": 3.0721747388414054, + "grad_norm": 0.06392118267209995, + "learning_rate": 3.8864719638314955e-05, + "loss": 0.0295, + "step": 6470 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.06152961222754858, + "learning_rate": 3.87031954944527e-05, + "loss": 0.0295, + "step": 6480 + }, + { + "epoch": 3.0816714150047484, + "grad_norm": 0.06899224674922493, + "learning_rate": 3.854179549100741e-05, + "loss": 0.0303, + "step": 6490 + }, + { + "epoch": 3.0864197530864197, + "grad_norm": 0.06653447898546623, + "learning_rate": 3.838052140160128e-05, + "loss": 0.0292, + "step": 6500 + }, + { + "epoch": 3.091168091168091, + "grad_norm": 0.06918259568317799, + "learning_rate": 3.821937499847291e-05, + "loss": 0.029, + "step": 6510 + }, + { + "epoch": 3.0959164292497627, + "grad_norm": 0.06825716226772567, + "learning_rate": 3.805835805245772e-05, + "loss": 0.0293, + "step": 6520 + }, + { + "epoch": 3.100664767331434, + "grad_norm": 0.07187414949365206, + "learning_rate": 3.789747233296852e-05, + "loss": 0.0306, + "step": 6530 + }, + { + "epoch": 3.1054131054131053, + "grad_norm": 0.06768996626651275, + "learning_rate": 3.7736719607976125e-05, + "loss": 0.0295, + "step": 6540 + }, + { + "epoch": 3.1101614434947766, + "grad_norm": 0.07154653262346644, + "learning_rate": 3.7576101643989816e-05, + "loss": 0.0292, + "step": 6550 + }, + { + "epoch": 3.1149097815764484, + "grad_norm": 0.07168150943744746, + "learning_rate": 3.7415620206038036e-05, + "loss": 0.0297, + "step": 6560 + }, + { + "epoch": 3.1196581196581197, + "grad_norm": 0.07122268195972503, + "learning_rate": 3.7255277057648906e-05, + "loss": 0.029, + "step": 6570 + }, + { + "epoch": 3.124406457739791, + "grad_norm": 0.06977327334738936, + "learning_rate": 3.7095073960830905e-05, + "loss": 0.0295, + "step": 6580 + }, + { + "epoch": 3.1291547958214627, + "grad_norm": 0.0594616623400008, + "learning_rate": 3.69350126760535e-05, + "loss": 0.0287, + "step": 6590 + }, + { + "epoch": 3.133903133903134, + "grad_norm": 0.07121069085002024, + "learning_rate": 3.677509496222776e-05, + "loss": 0.03, + "step": 6600 + }, + { + "epoch": 3.1386514719848053, + "grad_norm": 0.07319022999980948, + "learning_rate": 3.661532257668705e-05, + "loss": 0.0303, + "step": 6610 + }, + { + "epoch": 3.1433998100664766, + "grad_norm": 0.0712837669338485, + "learning_rate": 3.645569727516779e-05, + "loss": 0.0298, + "step": 6620 + }, + { + "epoch": 3.148148148148148, + "grad_norm": 0.06732748260196915, + "learning_rate": 3.6296220811790015e-05, + "loss": 0.0291, + "step": 6630 + }, + { + "epoch": 3.1528964862298197, + "grad_norm": 0.06664102017573165, + "learning_rate": 3.613689493903819e-05, + "loss": 0.0301, + "step": 6640 + }, + { + "epoch": 3.157644824311491, + "grad_norm": 0.06875731448293337, + "learning_rate": 3.597772140774201e-05, + "loss": 0.0293, + "step": 6650 + }, + { + "epoch": 3.1623931623931623, + "grad_norm": 0.06601941211086071, + "learning_rate": 3.5818701967057014e-05, + "loss": 0.0303, + "step": 6660 + }, + { + "epoch": 3.167141500474834, + "grad_norm": 0.06988087016697422, + "learning_rate": 3.5659838364445505e-05, + "loss": 0.0305, + "step": 6670 + }, + { + "epoch": 3.1718898385565053, + "grad_norm": 0.0638777258174965, + "learning_rate": 3.5501132345657204e-05, + "loss": 0.0295, + "step": 6680 + }, + { + "epoch": 3.1766381766381766, + "grad_norm": 0.07533485880526201, + "learning_rate": 3.5342585654710256e-05, + "loss": 0.0301, + "step": 6690 + }, + { + "epoch": 3.181386514719848, + "grad_norm": 0.06478157389723473, + "learning_rate": 3.518420003387183e-05, + "loss": 0.03, + "step": 6700 + }, + { + "epoch": 3.1861348528015196, + "grad_norm": 0.07292742713533744, + "learning_rate": 3.502597722363921e-05, + "loss": 0.0288, + "step": 6710 + }, + { + "epoch": 3.190883190883191, + "grad_norm": 0.07131010747314463, + "learning_rate": 3.486791896272049e-05, + "loss": 0.0306, + "step": 6720 + }, + { + "epoch": 3.1956315289648622, + "grad_norm": 0.06794303314607975, + "learning_rate": 3.471002698801555e-05, + "loss": 0.0288, + "step": 6730 + }, + { + "epoch": 3.2003798670465335, + "grad_norm": 0.06716891400944741, + "learning_rate": 3.4552303034596965e-05, + "loss": 0.0289, + "step": 6740 + }, + { + "epoch": 3.2051282051282053, + "grad_norm": 0.06790990593985272, + "learning_rate": 3.4394748835690894e-05, + "loss": 0.0291, + "step": 6750 + }, + { + "epoch": 3.2098765432098766, + "grad_norm": 0.0888034212536544, + "learning_rate": 3.423736612265812e-05, + "loss": 0.0299, + "step": 6760 + }, + { + "epoch": 3.214624881291548, + "grad_norm": 0.0747476284402158, + "learning_rate": 3.4080156624974904e-05, + "loss": 0.0289, + "step": 6770 + }, + { + "epoch": 3.219373219373219, + "grad_norm": 0.06494675970539739, + "learning_rate": 3.3923122070214126e-05, + "loss": 0.0298, + "step": 6780 + }, + { + "epoch": 3.224121557454891, + "grad_norm": 0.06660095925682659, + "learning_rate": 3.376626418402612e-05, + "loss": 0.0292, + "step": 6790 + }, + { + "epoch": 3.2288698955365622, + "grad_norm": 0.06275102353559732, + "learning_rate": 3.3609584690119925e-05, + "loss": 0.0298, + "step": 6800 + }, + { + "epoch": 3.2336182336182335, + "grad_norm": 0.06584238818431662, + "learning_rate": 3.345308531024412e-05, + "loss": 0.0299, + "step": 6810 + }, + { + "epoch": 3.238366571699905, + "grad_norm": 0.07161380779517292, + "learning_rate": 3.329676776416811e-05, + "loss": 0.0302, + "step": 6820 + }, + { + "epoch": 3.2431149097815766, + "grad_norm": 0.06893278475106504, + "learning_rate": 3.3140633769663057e-05, + "loss": 0.0301, + "step": 6830 + }, + { + "epoch": 3.247863247863248, + "grad_norm": 0.06548621449780895, + "learning_rate": 3.298468504248313e-05, + "loss": 0.0296, + "step": 6840 + }, + { + "epoch": 3.252611585944919, + "grad_norm": 0.0716775104112885, + "learning_rate": 3.282892329634657e-05, + "loss": 0.031, + "step": 6850 + }, + { + "epoch": 3.2573599240265905, + "grad_norm": 0.06372556616620227, + "learning_rate": 3.267335024291687e-05, + "loss": 0.0288, + "step": 6860 + }, + { + "epoch": 3.262108262108262, + "grad_norm": 0.07073948783542645, + "learning_rate": 3.251796759178403e-05, + "loss": 0.0313, + "step": 6870 + }, + { + "epoch": 3.2668566001899335, + "grad_norm": 0.0684437936834999, + "learning_rate": 3.236277705044568e-05, + "loss": 0.0303, + "step": 6880 + }, + { + "epoch": 3.271604938271605, + "grad_norm": 0.07327844324302656, + "learning_rate": 3.220778032428839e-05, + "loss": 0.0299, + "step": 6890 + }, + { + "epoch": 3.2763532763532766, + "grad_norm": 0.0734448176744976, + "learning_rate": 3.205297911656886e-05, + "loss": 0.03, + "step": 6900 + }, + { + "epoch": 3.281101614434948, + "grad_norm": 0.07455056788951349, + "learning_rate": 3.18983751283953e-05, + "loss": 0.0301, + "step": 6910 + }, + { + "epoch": 3.285849952516619, + "grad_norm": 0.07958798598513538, + "learning_rate": 3.174397005870858e-05, + "loss": 0.0303, + "step": 6920 + }, + { + "epoch": 3.2905982905982905, + "grad_norm": 0.06346210438588853, + "learning_rate": 3.158976560426379e-05, + "loss": 0.0294, + "step": 6930 + }, + { + "epoch": 3.295346628679962, + "grad_norm": 0.07691120817369627, + "learning_rate": 3.143576345961132e-05, + "loss": 0.0301, + "step": 6940 + }, + { + "epoch": 3.3000949667616335, + "grad_norm": 0.07164325872215613, + "learning_rate": 3.128196531707849e-05, + "loss": 0.0293, + "step": 6950 + }, + { + "epoch": 3.304843304843305, + "grad_norm": 0.07233096222509539, + "learning_rate": 3.112837286675082e-05, + "loss": 0.03, + "step": 6960 + }, + { + "epoch": 3.309591642924976, + "grad_norm": 0.07539873084590439, + "learning_rate": 3.097498779645341e-05, + "loss": 0.0294, + "step": 6970 + }, + { + "epoch": 3.314339981006648, + "grad_norm": 0.07223289486669585, + "learning_rate": 3.082181179173257e-05, + "loss": 0.029, + "step": 6980 + }, + { + "epoch": 3.319088319088319, + "grad_norm": 0.07106759032173504, + "learning_rate": 3.0668846535837114e-05, + "loss": 0.0298, + "step": 6990 + }, + { + "epoch": 3.3238366571699904, + "grad_norm": 0.06780319103228508, + "learning_rate": 3.051609370969999e-05, + "loss": 0.0305, + "step": 7000 + }, + { + "epoch": 3.3285849952516617, + "grad_norm": 0.07407007565668865, + "learning_rate": 3.03635549919197e-05, + "loss": 0.0298, + "step": 7010 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.07082227455128665, + "learning_rate": 3.021123205874199e-05, + "loss": 0.0294, + "step": 7020 + }, + { + "epoch": 3.338081671415005, + "grad_norm": 0.07345832518154012, + "learning_rate": 3.0059126584041275e-05, + "loss": 0.0291, + "step": 7030 + }, + { + "epoch": 3.342830009496676, + "grad_norm": 0.06676561509095681, + "learning_rate": 2.9907240239302393e-05, + "loss": 0.029, + "step": 7040 + }, + { + "epoch": 3.347578347578348, + "grad_norm": 0.07407332227063478, + "learning_rate": 2.9755574693602083e-05, + "loss": 0.0293, + "step": 7050 + }, + { + "epoch": 3.352326685660019, + "grad_norm": 0.06615456093639178, + "learning_rate": 2.9604131613590824e-05, + "loss": 0.03, + "step": 7060 + }, + { + "epoch": 3.3570750237416904, + "grad_norm": 0.0675800865410347, + "learning_rate": 2.9452912663474348e-05, + "loss": 0.0298, + "step": 7070 + }, + { + "epoch": 3.3618233618233617, + "grad_norm": 0.07377131577889622, + "learning_rate": 2.9301919504995436e-05, + "loss": 0.0299, + "step": 7080 + }, + { + "epoch": 3.366571699905033, + "grad_norm": 0.0669315018822171, + "learning_rate": 2.915115379741571e-05, + "loss": 0.0295, + "step": 7090 + }, + { + "epoch": 3.3713200379867048, + "grad_norm": 0.06492227051529588, + "learning_rate": 2.900061719749724e-05, + "loss": 0.0295, + "step": 7100 + }, + { + "epoch": 3.376068376068376, + "grad_norm": 0.06978530575950649, + "learning_rate": 2.8850311359484527e-05, + "loss": 0.0293, + "step": 7110 + }, + { + "epoch": 3.3808167141500474, + "grad_norm": 0.08419475579359366, + "learning_rate": 2.870023793508616e-05, + "loss": 0.0308, + "step": 7120 + }, + { + "epoch": 3.385565052231719, + "grad_norm": 0.06274727393894568, + "learning_rate": 2.855039857345681e-05, + "loss": 0.0304, + "step": 7130 + }, + { + "epoch": 3.3903133903133904, + "grad_norm": 0.06888963793040788, + "learning_rate": 2.8400794921178953e-05, + "loss": 0.0291, + "step": 7140 + }, + { + "epoch": 3.3950617283950617, + "grad_norm": 0.06693623126264903, + "learning_rate": 2.825142862224494e-05, + "loss": 0.0296, + "step": 7150 + }, + { + "epoch": 3.399810066476733, + "grad_norm": 0.07078918230797407, + "learning_rate": 2.810230131803878e-05, + "loss": 0.0287, + "step": 7160 + }, + { + "epoch": 3.4045584045584047, + "grad_norm": 0.06536488736351585, + "learning_rate": 2.7953414647318232e-05, + "loss": 0.0301, + "step": 7170 + }, + { + "epoch": 3.409306742640076, + "grad_norm": 0.06126251267507368, + "learning_rate": 2.78047702461967e-05, + "loss": 0.0301, + "step": 7180 + }, + { + "epoch": 3.4140550807217473, + "grad_norm": 0.07079050144077682, + "learning_rate": 2.7656369748125276e-05, + "loss": 0.0297, + "step": 7190 + }, + { + "epoch": 3.4188034188034186, + "grad_norm": 0.07075228556595065, + "learning_rate": 2.7508214783874874e-05, + "loss": 0.0288, + "step": 7200 + }, + { + "epoch": 3.4235517568850904, + "grad_norm": 0.06846796248740335, + "learning_rate": 2.7360306981518146e-05, + "loss": 0.0284, + "step": 7210 + }, + { + "epoch": 3.4283000949667617, + "grad_norm": 0.0721292050646478, + "learning_rate": 2.721264796641178e-05, + "loss": 0.0296, + "step": 7220 + }, + { + "epoch": 3.433048433048433, + "grad_norm": 0.06219252172725369, + "learning_rate": 2.7065239361178464e-05, + "loss": 0.029, + "step": 7230 + }, + { + "epoch": 3.4377967711301043, + "grad_norm": 0.07750464332377213, + "learning_rate": 2.6918082785689208e-05, + "loss": 0.0294, + "step": 7240 + }, + { + "epoch": 3.442545109211776, + "grad_norm": 0.06381800335879921, + "learning_rate": 2.6771179857045396e-05, + "loss": 0.0295, + "step": 7250 + }, + { + "epoch": 3.4472934472934473, + "grad_norm": 0.07106851172959223, + "learning_rate": 2.6624532189561145e-05, + "loss": 0.0305, + "step": 7260 + }, + { + "epoch": 3.4520417853751186, + "grad_norm": 0.07438951208194475, + "learning_rate": 2.6478141394745524e-05, + "loss": 0.0301, + "step": 7270 + }, + { + "epoch": 3.45679012345679, + "grad_norm": 0.07616164287915007, + "learning_rate": 2.6332009081284754e-05, + "loss": 0.0306, + "step": 7280 + }, + { + "epoch": 3.4615384615384617, + "grad_norm": 0.07825537217631368, + "learning_rate": 2.61861368550247e-05, + "loss": 0.0293, + "step": 7290 + }, + { + "epoch": 3.466286799620133, + "grad_norm": 0.06646491610528864, + "learning_rate": 2.6040526318953052e-05, + "loss": 0.0291, + "step": 7300 + }, + { + "epoch": 3.4710351377018043, + "grad_norm": 0.06723874601433405, + "learning_rate": 2.5895179073181863e-05, + "loss": 0.029, + "step": 7310 + }, + { + "epoch": 3.4757834757834756, + "grad_norm": 0.06577255286712014, + "learning_rate": 2.575009671492983e-05, + "loss": 0.0293, + "step": 7320 + }, + { + "epoch": 3.4805318138651473, + "grad_norm": 0.06816455799801772, + "learning_rate": 2.5605280838504863e-05, + "loss": 0.0299, + "step": 7330 + }, + { + "epoch": 3.4852801519468186, + "grad_norm": 0.06675898687092334, + "learning_rate": 2.5460733035286432e-05, + "loss": 0.0294, + "step": 7340 + }, + { + "epoch": 3.49002849002849, + "grad_norm": 0.06318313408947812, + "learning_rate": 2.5316454893708252e-05, + "loss": 0.0294, + "step": 7350 + }, + { + "epoch": 3.4947768281101617, + "grad_norm": 0.07189758151324366, + "learning_rate": 2.517244799924064e-05, + "loss": 0.0294, + "step": 7360 + }, + { + "epoch": 3.499525166191833, + "grad_norm": 0.06896592031007132, + "learning_rate": 2.502871393437325e-05, + "loss": 0.0289, + "step": 7370 + }, + { + "epoch": 3.5042735042735043, + "grad_norm": 0.06993924532105567, + "learning_rate": 2.4885254278597564e-05, + "loss": 0.0299, + "step": 7380 + }, + { + "epoch": 3.5090218423551756, + "grad_norm": 0.07027093754624048, + "learning_rate": 2.47420706083896e-05, + "loss": 0.03, + "step": 7390 + }, + { + "epoch": 3.513770180436847, + "grad_norm": 0.07744740598451617, + "learning_rate": 2.4599164497192617e-05, + "loss": 0.0299, + "step": 7400 + }, + { + "epoch": 3.5185185185185186, + "grad_norm": 0.062220716630855696, + "learning_rate": 2.4456537515399707e-05, + "loss": 0.0286, + "step": 7410 + }, + { + "epoch": 3.52326685660019, + "grad_norm": 0.07238813696823611, + "learning_rate": 2.4314191230336702e-05, + "loss": 0.0299, + "step": 7420 + }, + { + "epoch": 3.528015194681861, + "grad_norm": 0.064218010731912, + "learning_rate": 2.4172127206244766e-05, + "loss": 0.0286, + "step": 7430 + }, + { + "epoch": 3.532763532763533, + "grad_norm": 0.06822123851198789, + "learning_rate": 2.4030347004263392e-05, + "loss": 0.0294, + "step": 7440 + }, + { + "epoch": 3.5375118708452042, + "grad_norm": 0.06927541960016761, + "learning_rate": 2.3888852182413085e-05, + "loss": 0.03, + "step": 7450 + }, + { + "epoch": 3.5422602089268755, + "grad_norm": 0.06823533681937732, + "learning_rate": 2.3747644295578364e-05, + "loss": 0.029, + "step": 7460 + }, + { + "epoch": 3.547008547008547, + "grad_norm": 0.06614012081735723, + "learning_rate": 2.3606724895490568e-05, + "loss": 0.0291, + "step": 7470 + }, + { + "epoch": 3.5517568850902186, + "grad_norm": 0.0691063655206067, + "learning_rate": 2.346609553071093e-05, + "loss": 0.0289, + "step": 7480 + }, + { + "epoch": 3.55650522317189, + "grad_norm": 0.0629969554522321, + "learning_rate": 2.332575774661342e-05, + "loss": 0.0287, + "step": 7490 + }, + { + "epoch": 3.561253561253561, + "grad_norm": 0.06283025277767418, + "learning_rate": 2.3185713085367866e-05, + "loss": 0.0288, + "step": 7500 + }, + { + "epoch": 3.566001899335233, + "grad_norm": 0.07526099854461016, + "learning_rate": 2.3045963085923007e-05, + "loss": 0.0293, + "step": 7510 + }, + { + "epoch": 3.5707502374169042, + "grad_norm": 0.0757324713973307, + "learning_rate": 2.2906509283989462e-05, + "loss": 0.0304, + "step": 7520 + }, + { + "epoch": 3.5754985754985755, + "grad_norm": 0.06697936391273197, + "learning_rate": 2.276735321202303e-05, + "loss": 0.0299, + "step": 7530 + }, + { + "epoch": 3.580246913580247, + "grad_norm": 0.07366321440079159, + "learning_rate": 2.2628496399207683e-05, + "loss": 0.0286, + "step": 7540 + }, + { + "epoch": 3.584995251661918, + "grad_norm": 0.07788259211920452, + "learning_rate": 2.2489940371438912e-05, + "loss": 0.0301, + "step": 7550 + }, + { + "epoch": 3.58974358974359, + "grad_norm": 0.06142858570631925, + "learning_rate": 2.2351686651306815e-05, + "loss": 0.0293, + "step": 7560 + }, + { + "epoch": 3.594491927825261, + "grad_norm": 0.06973016115444622, + "learning_rate": 2.2213736758079494e-05, + "loss": 0.0292, + "step": 7570 + }, + { + "epoch": 3.5992402659069325, + "grad_norm": 0.07313284302706362, + "learning_rate": 2.207609220768625e-05, + "loss": 0.0296, + "step": 7580 + }, + { + "epoch": 3.603988603988604, + "grad_norm": 0.06511509056603623, + "learning_rate": 2.1938754512701042e-05, + "loss": 0.0299, + "step": 7590 + }, + { + "epoch": 3.6087369420702755, + "grad_norm": 0.0778195760372216, + "learning_rate": 2.1801725182325733e-05, + "loss": 0.0288, + "step": 7600 + }, + { + "epoch": 3.613485280151947, + "grad_norm": 0.06892980871588102, + "learning_rate": 2.1665005722373588e-05, + "loss": 0.029, + "step": 7610 + }, + { + "epoch": 3.618233618233618, + "grad_norm": 0.06590637775739139, + "learning_rate": 2.1528597635252756e-05, + "loss": 0.0292, + "step": 7620 + }, + { + "epoch": 3.6229819563152894, + "grad_norm": 0.06266081493755991, + "learning_rate": 2.139250241994964e-05, + "loss": 0.0291, + "step": 7630 + }, + { + "epoch": 3.627730294396961, + "grad_norm": 0.06352050961618583, + "learning_rate": 2.1256721572012584e-05, + "loss": 0.0293, + "step": 7640 + }, + { + "epoch": 3.6324786324786325, + "grad_norm": 0.07218735656724067, + "learning_rate": 2.1121256583535272e-05, + "loss": 0.0294, + "step": 7650 + }, + { + "epoch": 3.6372269705603038, + "grad_norm": 0.06220732014953678, + "learning_rate": 2.098610894314048e-05, + "loss": 0.0287, + "step": 7660 + }, + { + "epoch": 3.6419753086419755, + "grad_norm": 0.06697596958873017, + "learning_rate": 2.085128013596359e-05, + "loss": 0.0289, + "step": 7670 + }, + { + "epoch": 3.646723646723647, + "grad_norm": 0.0741731661357627, + "learning_rate": 2.07167716436364e-05, + "loss": 0.0298, + "step": 7680 + }, + { + "epoch": 3.651471984805318, + "grad_norm": 0.06520991046303531, + "learning_rate": 2.0582584944270694e-05, + "loss": 0.0287, + "step": 7690 + }, + { + "epoch": 3.6562203228869894, + "grad_norm": 0.07077207583237781, + "learning_rate": 2.0448721512442153e-05, + "loss": 0.0284, + "step": 7700 + }, + { + "epoch": 3.6609686609686607, + "grad_norm": 0.07259865787936985, + "learning_rate": 2.031518281917403e-05, + "loss": 0.0292, + "step": 7710 + }, + { + "epoch": 3.6657169990503324, + "grad_norm": 0.06783445679735263, + "learning_rate": 2.018197033192102e-05, + "loss": 0.0295, + "step": 7720 + }, + { + "epoch": 3.6704653371320037, + "grad_norm": 0.06627833755171479, + "learning_rate": 2.0049085514553183e-05, + "loss": 0.0302, + "step": 7730 + }, + { + "epoch": 3.6752136752136755, + "grad_norm": 0.06330461188192492, + "learning_rate": 1.9916529827339825e-05, + "loss": 0.0294, + "step": 7740 + }, + { + "epoch": 3.679962013295347, + "grad_norm": 0.06943978938278865, + "learning_rate": 1.9784304726933383e-05, + "loss": 0.0288, + "step": 7750 + }, + { + "epoch": 3.684710351377018, + "grad_norm": 0.06788683924973546, + "learning_rate": 1.9652411666353547e-05, + "loss": 0.0294, + "step": 7760 + }, + { + "epoch": 3.6894586894586894, + "grad_norm": 0.0698883028022984, + "learning_rate": 1.9520852094971215e-05, + "loss": 0.03, + "step": 7770 + }, + { + "epoch": 3.6942070275403607, + "grad_norm": 0.07496936356578845, + "learning_rate": 1.9389627458492526e-05, + "loss": 0.0295, + "step": 7780 + }, + { + "epoch": 3.6989553656220324, + "grad_norm": 0.06266498939566893, + "learning_rate": 1.9258739198943105e-05, + "loss": 0.0298, + "step": 7790 + }, + { + "epoch": 3.7037037037037037, + "grad_norm": 0.07475395266985198, + "learning_rate": 1.9128188754652042e-05, + "loss": 0.0295, + "step": 7800 + }, + { + "epoch": 3.708452041785375, + "grad_norm": 0.060434153203178666, + "learning_rate": 1.8997977560236276e-05, + "loss": 0.0284, + "step": 7810 + }, + { + "epoch": 3.7132003798670468, + "grad_norm": 0.06995861505168599, + "learning_rate": 1.886810704658467e-05, + "loss": 0.0287, + "step": 7820 + }, + { + "epoch": 3.717948717948718, + "grad_norm": 0.06912380343484231, + "learning_rate": 1.8738578640842353e-05, + "loss": 0.0297, + "step": 7830 + }, + { + "epoch": 3.7226970560303894, + "grad_norm": 0.06144330060834129, + "learning_rate": 1.8609393766395085e-05, + "loss": 0.0286, + "step": 7840 + }, + { + "epoch": 3.7274453941120607, + "grad_norm": 0.07386046991659506, + "learning_rate": 1.8480553842853492e-05, + "loss": 0.0285, + "step": 7850 + }, + { + "epoch": 3.732193732193732, + "grad_norm": 0.07598362423461756, + "learning_rate": 1.8352060286037626e-05, + "loss": 0.0293, + "step": 7860 + }, + { + "epoch": 3.7369420702754037, + "grad_norm": 0.06702972696058551, + "learning_rate": 1.8223914507961232e-05, + "loss": 0.0289, + "step": 7870 + }, + { + "epoch": 3.741690408357075, + "grad_norm": 0.06941334150150881, + "learning_rate": 1.80961179168164e-05, + "loss": 0.0288, + "step": 7880 + }, + { + "epoch": 3.7464387464387463, + "grad_norm": 0.06768336658087552, + "learning_rate": 1.796867191695794e-05, + "loss": 0.0293, + "step": 7890 + }, + { + "epoch": 3.751187084520418, + "grad_norm": 0.06672232876261866, + "learning_rate": 1.784157790888808e-05, + "loss": 0.0301, + "step": 7900 + }, + { + "epoch": 3.7559354226020893, + "grad_norm": 0.06680766630782138, + "learning_rate": 1.7714837289240958e-05, + "loss": 0.0292, + "step": 7910 + }, + { + "epoch": 3.7606837606837606, + "grad_norm": 0.07203414607499374, + "learning_rate": 1.7588451450767396e-05, + "loss": 0.0291, + "step": 7920 + }, + { + "epoch": 3.765432098765432, + "grad_norm": 0.07270120155548507, + "learning_rate": 1.7462421782319472e-05, + "loss": 0.0299, + "step": 7930 + }, + { + "epoch": 3.7701804368471032, + "grad_norm": 0.06863174362037226, + "learning_rate": 1.733674966883533e-05, + "loss": 0.0289, + "step": 7940 + }, + { + "epoch": 3.774928774928775, + "grad_norm": 0.06879625191037461, + "learning_rate": 1.721143649132399e-05, + "loss": 0.0293, + "step": 7950 + }, + { + "epoch": 3.7796771130104463, + "grad_norm": 0.06683421134870436, + "learning_rate": 1.7086483626850063e-05, + "loss": 0.0287, + "step": 7960 + }, + { + "epoch": 3.784425451092118, + "grad_norm": 0.06420241987184519, + "learning_rate": 1.696189244851874e-05, + "loss": 0.0293, + "step": 7970 + }, + { + "epoch": 3.7891737891737893, + "grad_norm": 0.06241916563124866, + "learning_rate": 1.683766432546065e-05, + "loss": 0.0287, + "step": 7980 + }, + { + "epoch": 3.7939221272554606, + "grad_norm": 0.06518918556845223, + "learning_rate": 1.671380062281675e-05, + "loss": 0.0297, + "step": 7990 + }, + { + "epoch": 3.798670465337132, + "grad_norm": 0.06610291541366478, + "learning_rate": 1.6590302701723454e-05, + "loss": 0.0283, + "step": 8000 + }, + { + "epoch": 3.8034188034188032, + "grad_norm": 0.06513753716420254, + "learning_rate": 1.6467171919297598e-05, + "loss": 0.0287, + "step": 8010 + }, + { + "epoch": 3.808167141500475, + "grad_norm": 0.06499189240717512, + "learning_rate": 1.6344409628621484e-05, + "loss": 0.0282, + "step": 8020 + }, + { + "epoch": 3.8129154795821463, + "grad_norm": 0.05791132031418536, + "learning_rate": 1.6222017178728134e-05, + "loss": 0.0287, + "step": 8030 + }, + { + "epoch": 3.8176638176638176, + "grad_norm": 0.06242292802002234, + "learning_rate": 1.6099995914586347e-05, + "loss": 0.0289, + "step": 8040 + }, + { + "epoch": 3.8224121557454893, + "grad_norm": 0.06740341868847345, + "learning_rate": 1.5978347177085967e-05, + "loss": 0.0282, + "step": 8050 + }, + { + "epoch": 3.8271604938271606, + "grad_norm": 0.06061445716699388, + "learning_rate": 1.5857072303023195e-05, + "loss": 0.0296, + "step": 8060 + }, + { + "epoch": 3.831908831908832, + "grad_norm": 0.07131985866133515, + "learning_rate": 1.5736172625085792e-05, + "loss": 0.0299, + "step": 8070 + }, + { + "epoch": 3.836657169990503, + "grad_norm": 0.057874525028645624, + "learning_rate": 1.5615649471838555e-05, + "loss": 0.0287, + "step": 8080 + }, + { + "epoch": 3.8414055080721745, + "grad_norm": 0.06822181603049116, + "learning_rate": 1.5495504167708592e-05, + "loss": 0.0292, + "step": 8090 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.0686307963558664, + "learning_rate": 1.5375738032970893e-05, + "loss": 0.0296, + "step": 8100 + }, + { + "epoch": 3.8509021842355176, + "grad_norm": 0.07306811477030246, + "learning_rate": 1.5256352383733707e-05, + "loss": 0.0298, + "step": 8110 + }, + { + "epoch": 3.855650522317189, + "grad_norm": 0.06865171882288519, + "learning_rate": 1.513734853192419e-05, + "loss": 0.0295, + "step": 8120 + }, + { + "epoch": 3.8603988603988606, + "grad_norm": 0.0697439950068489, + "learning_rate": 1.5018727785273872e-05, + "loss": 0.0289, + "step": 8130 + }, + { + "epoch": 3.865147198480532, + "grad_norm": 0.06298200776535154, + "learning_rate": 1.4900491447304409e-05, + "loss": 0.029, + "step": 8140 + }, + { + "epoch": 3.869895536562203, + "grad_norm": 0.06658680923880524, + "learning_rate": 1.4782640817313149e-05, + "loss": 0.0294, + "step": 8150 + }, + { + "epoch": 3.8746438746438745, + "grad_norm": 0.06928571693945171, + "learning_rate": 1.4665177190358886e-05, + "loss": 0.0289, + "step": 8160 + }, + { + "epoch": 3.879392212725546, + "grad_norm": 0.06469398648739424, + "learning_rate": 1.454810185724772e-05, + "loss": 0.0283, + "step": 8170 + }, + { + "epoch": 3.8841405508072175, + "grad_norm": 0.06651650721363567, + "learning_rate": 1.4431416104518708e-05, + "loss": 0.0292, + "step": 8180 + }, + { + "epoch": 3.888888888888889, + "grad_norm": 0.06366584386337464, + "learning_rate": 1.4315121214429882e-05, + "loss": 0.028, + "step": 8190 + }, + { + "epoch": 3.89363722697056, + "grad_norm": 0.062392434365402885, + "learning_rate": 1.4199218464944042e-05, + "loss": 0.0294, + "step": 8200 + }, + { + "epoch": 3.898385565052232, + "grad_norm": 0.07442066425718363, + "learning_rate": 1.40837091297148e-05, + "loss": 0.0287, + "step": 8210 + }, + { + "epoch": 3.903133903133903, + "grad_norm": 0.06157337156655846, + "learning_rate": 1.39685944780725e-05, + "loss": 0.0287, + "step": 8220 + }, + { + "epoch": 3.9078822412155745, + "grad_norm": 0.07073742286274987, + "learning_rate": 1.3853875775010355e-05, + "loss": 0.0298, + "step": 8230 + }, + { + "epoch": 3.912630579297246, + "grad_norm": 0.06380791295996606, + "learning_rate": 1.3739554281170509e-05, + "loss": 0.0286, + "step": 8240 + }, + { + "epoch": 3.9173789173789175, + "grad_norm": 0.06793318458985868, + "learning_rate": 1.3625631252830118e-05, + "loss": 0.029, + "step": 8250 + }, + { + "epoch": 3.922127255460589, + "grad_norm": 0.06673059984021656, + "learning_rate": 1.3512107941887686e-05, + "loss": 0.0285, + "step": 8260 + }, + { + "epoch": 3.92687559354226, + "grad_norm": 0.06647575139142457, + "learning_rate": 1.3398985595849156e-05, + "loss": 0.0284, + "step": 8270 + }, + { + "epoch": 3.931623931623932, + "grad_norm": 0.07348615772743924, + "learning_rate": 1.3286265457814345e-05, + "loss": 0.0294, + "step": 8280 + }, + { + "epoch": 3.936372269705603, + "grad_norm": 0.07094780772889218, + "learning_rate": 1.3173948766463145e-05, + "loss": 0.0289, + "step": 8290 + }, + { + "epoch": 3.9411206077872745, + "grad_norm": 0.0607391278883314, + "learning_rate": 1.3062036756042034e-05, + "loss": 0.0296, + "step": 8300 + }, + { + "epoch": 3.9458689458689458, + "grad_norm": 0.07017807758243115, + "learning_rate": 1.2950530656350413e-05, + "loss": 0.029, + "step": 8310 + }, + { + "epoch": 3.950617283950617, + "grad_norm": 0.06819643839925313, + "learning_rate": 1.2839431692727172e-05, + "loss": 0.0281, + "step": 8320 + }, + { + "epoch": 3.955365622032289, + "grad_norm": 0.07258576073297215, + "learning_rate": 1.2728741086037149e-05, + "loss": 0.0304, + "step": 8330 + }, + { + "epoch": 3.96011396011396, + "grad_norm": 0.0611463022586769, + "learning_rate": 1.2618460052657804e-05, + "loss": 0.0283, + "step": 8340 + }, + { + "epoch": 3.9648622981956314, + "grad_norm": 0.06596899219347056, + "learning_rate": 1.2508589804465765e-05, + "loss": 0.029, + "step": 8350 + }, + { + "epoch": 3.969610636277303, + "grad_norm": 0.06589360916965809, + "learning_rate": 1.2399131548823539e-05, + "loss": 0.0293, + "step": 8360 + }, + { + "epoch": 3.9743589743589745, + "grad_norm": 0.06236261936331933, + "learning_rate": 1.2290086488566305e-05, + "loss": 0.0285, + "step": 8370 + }, + { + "epoch": 3.9791073124406457, + "grad_norm": 0.062187558778588066, + "learning_rate": 1.2181455821988574e-05, + "loss": 0.0282, + "step": 8380 + }, + { + "epoch": 3.983855650522317, + "grad_norm": 0.06302138878401145, + "learning_rate": 1.2073240742831165e-05, + "loss": 0.0286, + "step": 8390 + }, + { + "epoch": 3.9886039886039883, + "grad_norm": 0.06721258381528698, + "learning_rate": 1.1965442440267943e-05, + "loss": 0.0293, + "step": 8400 + }, + { + "epoch": 3.99335232668566, + "grad_norm": 0.06348684077844487, + "learning_rate": 1.185806209889288e-05, + "loss": 0.0295, + "step": 8410 + }, + { + "epoch": 3.9981006647673314, + "grad_norm": 0.06148372428103636, + "learning_rate": 1.1751100898706929e-05, + "loss": 0.0286, + "step": 8420 + }, + { + "epoch": 4.002849002849003, + "grad_norm": 0.0669770850517368, + "learning_rate": 1.1644560015105155e-05, + "loss": 0.0275, + "step": 8430 + }, + { + "epoch": 4.007597340930674, + "grad_norm": 0.06205564035210006, + "learning_rate": 1.1538440618863722e-05, + "loss": 0.0261, + "step": 8440 + }, + { + "epoch": 4.012345679012346, + "grad_norm": 0.06482505236348735, + "learning_rate": 1.1432743876127122e-05, + "loss": 0.0266, + "step": 8450 + }, + { + "epoch": 4.017094017094017, + "grad_norm": 0.06405429708019271, + "learning_rate": 1.1327470948395263e-05, + "loss": 0.0264, + "step": 8460 + }, + { + "epoch": 4.021842355175688, + "grad_norm": 0.05805392904475434, + "learning_rate": 1.1222622992510807e-05, + "loss": 0.0267, + "step": 8470 + }, + { + "epoch": 4.02659069325736, + "grad_norm": 0.06681419167833032, + "learning_rate": 1.1118201160646386e-05, + "loss": 0.0261, + "step": 8480 + }, + { + "epoch": 4.031339031339031, + "grad_norm": 0.06161997026839342, + "learning_rate": 1.1014206600291938e-05, + "loss": 0.0266, + "step": 8490 + }, + { + "epoch": 4.036087369420703, + "grad_norm": 0.060438863544858334, + "learning_rate": 1.0910640454242172e-05, + "loss": 0.0266, + "step": 8500 + }, + { + "epoch": 4.040835707502374, + "grad_norm": 0.060385389480390225, + "learning_rate": 1.0807503860583895e-05, + "loss": 0.0272, + "step": 8510 + }, + { + "epoch": 4.045584045584046, + "grad_norm": 0.06409819399475297, + "learning_rate": 1.0704797952683638e-05, + "loss": 0.0265, + "step": 8520 + }, + { + "epoch": 4.050332383665717, + "grad_norm": 0.06087736110237468, + "learning_rate": 1.0602523859175061e-05, + "loss": 0.0266, + "step": 8530 + }, + { + "epoch": 4.055080721747388, + "grad_norm": 0.06323515633146604, + "learning_rate": 1.0500682703946695e-05, + "loss": 0.0263, + "step": 8540 + }, + { + "epoch": 4.05982905982906, + "grad_norm": 0.0643452572251499, + "learning_rate": 1.0399275606129449e-05, + "loss": 0.0273, + "step": 8550 + }, + { + "epoch": 4.064577397910731, + "grad_norm": 0.058983725279429286, + "learning_rate": 1.0298303680084448e-05, + "loss": 0.0263, + "step": 8560 + }, + { + "epoch": 4.069325735992402, + "grad_norm": 0.05942588829047056, + "learning_rate": 1.0197768035390665e-05, + "loss": 0.0262, + "step": 8570 + }, + { + "epoch": 4.074074074074074, + "grad_norm": 0.05843383183345459, + "learning_rate": 1.0097669776832801e-05, + "loss": 0.0261, + "step": 8580 + }, + { + "epoch": 4.078822412155746, + "grad_norm": 0.06319978308016973, + "learning_rate": 9.998010004389158e-06, + "loss": 0.0268, + "step": 8590 + }, + { + "epoch": 4.083570750237417, + "grad_norm": 0.06959963023516315, + "learning_rate": 9.898789813219467e-06, + "loss": 0.026, + "step": 8600 + }, + { + "epoch": 4.088319088319088, + "grad_norm": 0.06279160444176832, + "learning_rate": 9.800010293652962e-06, + "loss": 0.0262, + "step": 8610 + }, + { + "epoch": 4.09306742640076, + "grad_norm": 0.06210596724260713, + "learning_rate": 9.701672531176286e-06, + "loss": 0.0268, + "step": 8620 + }, + { + "epoch": 4.097815764482431, + "grad_norm": 0.06804911451660237, + "learning_rate": 9.603777606421682e-06, + "loss": 0.0267, + "step": 8630 + }, + { + "epoch": 4.102564102564102, + "grad_norm": 0.06319750889382578, + "learning_rate": 9.506326595155002e-06, + "loss": 0.0261, + "step": 8640 + }, + { + "epoch": 4.107312440645774, + "grad_norm": 0.06345749812453111, + "learning_rate": 9.40932056826398e-06, + "loss": 0.026, + "step": 8650 + }, + { + "epoch": 4.112060778727446, + "grad_norm": 0.07359741704271591, + "learning_rate": 9.312760591746377e-06, + "loss": 0.0267, + "step": 8660 + }, + { + "epoch": 4.116809116809117, + "grad_norm": 0.06646453154170578, + "learning_rate": 9.216647726698374e-06, + "loss": 0.0265, + "step": 8670 + }, + { + "epoch": 4.121557454890788, + "grad_norm": 0.06399503814526554, + "learning_rate": 9.120983029302805e-06, + "loss": 0.0265, + "step": 8680 + }, + { + "epoch": 4.12630579297246, + "grad_norm": 0.0694116774106874, + "learning_rate": 9.025767550817588e-06, + "loss": 0.027, + "step": 8690 + }, + { + "epoch": 4.131054131054131, + "grad_norm": 0.06513012610669609, + "learning_rate": 8.93100233756422e-06, + "loss": 0.0276, + "step": 8700 + }, + { + "epoch": 4.135802469135802, + "grad_norm": 0.06301088287794292, + "learning_rate": 8.83668843091624e-06, + "loss": 0.0265, + "step": 8710 + }, + { + "epoch": 4.1405508072174735, + "grad_norm": 0.0641064011910522, + "learning_rate": 8.74282686728773e-06, + "loss": 0.0263, + "step": 8720 + }, + { + "epoch": 4.145299145299146, + "grad_norm": 0.06483960459769524, + "learning_rate": 8.64941867812204e-06, + "loss": 0.027, + "step": 8730 + }, + { + "epoch": 4.150047483380817, + "grad_norm": 0.06246982318121049, + "learning_rate": 8.556464889880383e-06, + "loss": 0.0266, + "step": 8740 + }, + { + "epoch": 4.154795821462488, + "grad_norm": 0.06599868674063385, + "learning_rate": 8.46396652403053e-06, + "loss": 0.0258, + "step": 8750 + }, + { + "epoch": 4.15954415954416, + "grad_norm": 0.06953218850505466, + "learning_rate": 8.371924597035685e-06, + "loss": 0.0263, + "step": 8760 + }, + { + "epoch": 4.164292497625831, + "grad_norm": 0.06449112817949024, + "learning_rate": 8.280340120343182e-06, + "loss": 0.0262, + "step": 8770 + }, + { + "epoch": 4.169040835707502, + "grad_norm": 0.059146479371991204, + "learning_rate": 8.189214100373499e-06, + "loss": 0.0265, + "step": 8780 + }, + { + "epoch": 4.1737891737891735, + "grad_norm": 0.06304497308511589, + "learning_rate": 8.09854753850911e-06, + "loss": 0.0269, + "step": 8790 + }, + { + "epoch": 4.178537511870845, + "grad_norm": 0.0658470638909, + "learning_rate": 8.0083414310835e-06, + "loss": 0.0264, + "step": 8800 + }, + { + "epoch": 4.183285849952517, + "grad_norm": 0.06269284470926492, + "learning_rate": 7.918596769370273e-06, + "loss": 0.026, + "step": 8810 + }, + { + "epoch": 4.188034188034188, + "grad_norm": 0.06439020948553288, + "learning_rate": 7.829314539572152e-06, + "loss": 0.0272, + "step": 8820 + }, + { + "epoch": 4.1927825261158596, + "grad_norm": 0.06159800041826075, + "learning_rate": 7.740495722810271e-06, + "loss": 0.0262, + "step": 8830 + }, + { + "epoch": 4.197530864197531, + "grad_norm": 0.07107061401976039, + "learning_rate": 7.652141295113246e-06, + "loss": 0.0265, + "step": 8840 + }, + { + "epoch": 4.202279202279202, + "grad_norm": 0.06883391116772176, + "learning_rate": 7.56425222740661e-06, + "loss": 0.0267, + "step": 8850 + }, + { + "epoch": 4.2070275403608735, + "grad_norm": 0.06366701130324344, + "learning_rate": 7.476829485501996e-06, + "loss": 0.0262, + "step": 8860 + }, + { + "epoch": 4.211775878442545, + "grad_norm": 0.06170067399409082, + "learning_rate": 7.389874030086636e-06, + "loss": 0.0259, + "step": 8870 + }, + { + "epoch": 4.216524216524217, + "grad_norm": 0.06789843480685757, + "learning_rate": 7.303386816712715e-06, + "loss": 0.0265, + "step": 8880 + }, + { + "epoch": 4.221272554605888, + "grad_norm": 0.05923120427788399, + "learning_rate": 7.217368795786961e-06, + "loss": 0.026, + "step": 8890 + }, + { + "epoch": 4.2260208926875595, + "grad_norm": 0.06083093241769399, + "learning_rate": 7.131820912560116e-06, + "loss": 0.0261, + "step": 8900 + }, + { + "epoch": 4.230769230769231, + "grad_norm": 0.06305542194247242, + "learning_rate": 7.046744107116587e-06, + "loss": 0.0262, + "step": 8910 + }, + { + "epoch": 4.235517568850902, + "grad_norm": 0.06567886784210275, + "learning_rate": 6.962139314364152e-06, + "loss": 0.0262, + "step": 8920 + }, + { + "epoch": 4.240265906932573, + "grad_norm": 0.07334798155195307, + "learning_rate": 6.878007464023595e-06, + "loss": 0.0265, + "step": 8930 + }, + { + "epoch": 4.245014245014245, + "grad_norm": 0.062478008317110506, + "learning_rate": 6.794349480618583e-06, + "loss": 0.0259, + "step": 8940 + }, + { + "epoch": 4.249762583095916, + "grad_norm": 0.06331353681637673, + "learning_rate": 6.711166283465464e-06, + "loss": 0.0268, + "step": 8950 + }, + { + "epoch": 4.254510921177588, + "grad_norm": 0.07049738237881796, + "learning_rate": 6.628458786663133e-06, + "loss": 0.0267, + "step": 8960 + }, + { + "epoch": 4.2592592592592595, + "grad_norm": 0.07081958472277415, + "learning_rate": 6.546227899083062e-06, + "loss": 0.0261, + "step": 8970 + }, + { + "epoch": 4.264007597340931, + "grad_norm": 0.0699373748714869, + "learning_rate": 6.464474524359243e-06, + "loss": 0.0262, + "step": 8980 + }, + { + "epoch": 4.268755935422602, + "grad_norm": 0.0643433720902624, + "learning_rate": 6.3831995608783006e-06, + "loss": 0.0266, + "step": 8990 + }, + { + "epoch": 4.273504273504273, + "grad_norm": 0.06199841321149695, + "learning_rate": 6.30240390176961e-06, + "loss": 0.0271, + "step": 9000 + }, + { + "epoch": 4.278252611585945, + "grad_norm": 0.0633912810268076, + "learning_rate": 6.222088434895462e-06, + "loss": 0.0257, + "step": 9010 + }, + { + "epoch": 4.283000949667616, + "grad_norm": 0.06963912675209047, + "learning_rate": 6.142254042841317e-06, + "loss": 0.0266, + "step": 9020 + }, + { + "epoch": 4.287749287749287, + "grad_norm": 0.06130285170370608, + "learning_rate": 6.062901602906146e-06, + "loss": 0.0266, + "step": 9030 + }, + { + "epoch": 4.2924976258309595, + "grad_norm": 0.0603878830236836, + "learning_rate": 5.9840319870927085e-06, + "loss": 0.0254, + "step": 9040 + }, + { + "epoch": 4.297245963912631, + "grad_norm": 0.0677199802022205, + "learning_rate": 5.905646062098069e-06, + "loss": 0.0268, + "step": 9050 + }, + { + "epoch": 4.301994301994302, + "grad_norm": 0.06999900931956758, + "learning_rate": 5.827744689303965e-06, + "loss": 0.0259, + "step": 9060 + }, + { + "epoch": 4.306742640075973, + "grad_norm": 0.06715618982867681, + "learning_rate": 5.750328724767451e-06, + "loss": 0.0263, + "step": 9070 + }, + { + "epoch": 4.311490978157645, + "grad_norm": 0.06916301999009067, + "learning_rate": 5.6733990192113895e-06, + "loss": 0.0265, + "step": 9080 + }, + { + "epoch": 4.316239316239316, + "grad_norm": 0.06887916769392297, + "learning_rate": 5.5969564180151865e-06, + "loss": 0.0256, + "step": 9090 + }, + { + "epoch": 4.320987654320987, + "grad_norm": 0.062391492350832585, + "learning_rate": 5.521001761205441e-06, + "loss": 0.0263, + "step": 9100 + }, + { + "epoch": 4.3257359924026595, + "grad_norm": 0.0636610383829691, + "learning_rate": 5.445535883446767e-06, + "loss": 0.0262, + "step": 9110 + }, + { + "epoch": 4.330484330484331, + "grad_norm": 0.06977578416092847, + "learning_rate": 5.370559614032561e-06, + "loss": 0.0262, + "step": 9120 + }, + { + "epoch": 4.335232668566002, + "grad_norm": 0.06199360304205866, + "learning_rate": 5.296073776875921e-06, + "loss": 0.0267, + "step": 9130 + }, + { + "epoch": 4.339981006647673, + "grad_norm": 0.0731027692101727, + "learning_rate": 5.222079190500634e-06, + "loss": 0.0266, + "step": 9140 + }, + { + "epoch": 4.344729344729345, + "grad_norm": 0.06712212492413415, + "learning_rate": 5.1485766680320845e-06, + "loss": 0.0264, + "step": 9150 + }, + { + "epoch": 4.349477682811016, + "grad_norm": 0.07073796291387278, + "learning_rate": 5.075567017188421e-06, + "loss": 0.0263, + "step": 9160 + }, + { + "epoch": 4.354226020892687, + "grad_norm": 0.057210167816470946, + "learning_rate": 5.003051040271589e-06, + "loss": 0.0261, + "step": 9170 + }, + { + "epoch": 4.358974358974359, + "grad_norm": 0.06177772711231395, + "learning_rate": 4.93102953415861e-06, + "loss": 0.0263, + "step": 9180 + }, + { + "epoch": 4.363722697056031, + "grad_norm": 0.06143324967954758, + "learning_rate": 4.8595032902927216e-06, + "loss": 0.0261, + "step": 9190 + }, + { + "epoch": 4.368471035137702, + "grad_norm": 0.05943976139815956, + "learning_rate": 4.788473094674762e-06, + "loss": 0.0263, + "step": 9200 + }, + { + "epoch": 4.373219373219373, + "grad_norm": 0.07329718614072633, + "learning_rate": 4.717939727854515e-06, + "loss": 0.0259, + "step": 9210 + }, + { + "epoch": 4.377967711301045, + "grad_norm": 0.07061606580847958, + "learning_rate": 4.647903964922068e-06, + "loss": 0.0263, + "step": 9220 + }, + { + "epoch": 4.382716049382716, + "grad_norm": 0.06272935812309667, + "learning_rate": 4.57836657549941e-06, + "loss": 0.0264, + "step": 9230 + }, + { + "epoch": 4.387464387464387, + "grad_norm": 0.07029831538768902, + "learning_rate": 4.509328323731854e-06, + "loss": 0.026, + "step": 9240 + }, + { + "epoch": 4.392212725546059, + "grad_norm": 0.06258669446187784, + "learning_rate": 4.440789968279746e-06, + "loss": 0.0262, + "step": 9250 + }, + { + "epoch": 4.39696106362773, + "grad_norm": 0.06088481096301141, + "learning_rate": 4.372752262310037e-06, + "loss": 0.0255, + "step": 9260 + }, + { + "epoch": 4.401709401709402, + "grad_norm": 0.06766231594488686, + "learning_rate": 4.30521595348809e-06, + "loss": 0.0263, + "step": 9270 + }, + { + "epoch": 4.406457739791073, + "grad_norm": 0.06152626075563466, + "learning_rate": 4.238181783969375e-06, + "loss": 0.0256, + "step": 9280 + }, + { + "epoch": 4.411206077872745, + "grad_norm": 0.06940393084860598, + "learning_rate": 4.171650490391405e-06, + "loss": 0.0262, + "step": 9290 + }, + { + "epoch": 4.415954415954416, + "grad_norm": 0.06313129963095232, + "learning_rate": 4.105622803865561e-06, + "loss": 0.0264, + "step": 9300 + }, + { + "epoch": 4.420702754036087, + "grad_norm": 0.06285184126528884, + "learning_rate": 4.040099449969126e-06, + "loss": 0.0266, + "step": 9310 + }, + { + "epoch": 4.4254510921177586, + "grad_norm": 0.07370504911302739, + "learning_rate": 3.975081148737253e-06, + "loss": 0.0274, + "step": 9320 + }, + { + "epoch": 4.43019943019943, + "grad_norm": 0.06394573721669891, + "learning_rate": 3.910568614655086e-06, + "loss": 0.0264, + "step": 9330 + }, + { + "epoch": 4.434947768281102, + "grad_norm": 0.06685897297105997, + "learning_rate": 3.846562556649913e-06, + "loss": 0.026, + "step": 9340 + }, + { + "epoch": 4.439696106362773, + "grad_norm": 0.06301621849785136, + "learning_rate": 3.7830636780833375e-06, + "loss": 0.0256, + "step": 9350 + }, + { + "epoch": 4.444444444444445, + "grad_norm": 0.06336666020117579, + "learning_rate": 3.720072676743619e-06, + "loss": 0.0264, + "step": 9360 + }, + { + "epoch": 4.449192782526116, + "grad_norm": 0.07057764287909521, + "learning_rate": 3.6575902448379105e-06, + "loss": 0.0267, + "step": 9370 + }, + { + "epoch": 4.453941120607787, + "grad_norm": 0.07613078191752795, + "learning_rate": 3.5956170689847525e-06, + "loss": 0.0267, + "step": 9380 + }, + { + "epoch": 4.4586894586894585, + "grad_norm": 0.07514722653828791, + "learning_rate": 3.534153830206427e-06, + "loss": 0.026, + "step": 9390 + }, + { + "epoch": 4.46343779677113, + "grad_norm": 0.058038302569977704, + "learning_rate": 3.4732012039215776e-06, + "loss": 0.0267, + "step": 9400 + }, + { + "epoch": 4.468186134852801, + "grad_norm": 0.06666136619946722, + "learning_rate": 3.4127598599376866e-06, + "loss": 0.0264, + "step": 9410 + }, + { + "epoch": 4.472934472934473, + "grad_norm": 0.06517476786636972, + "learning_rate": 3.352830462443807e-06, + "loss": 0.0258, + "step": 9420 + }, + { + "epoch": 4.477682811016145, + "grad_norm": 0.06315330494297215, + "learning_rate": 3.2934136700031814e-06, + "loss": 0.0265, + "step": 9430 + }, + { + "epoch": 4.482431149097816, + "grad_norm": 0.06412969084338799, + "learning_rate": 3.2345101355460704e-06, + "loss": 0.0255, + "step": 9440 + }, + { + "epoch": 4.487179487179487, + "grad_norm": 0.0727250748514323, + "learning_rate": 3.176120506362529e-06, + "loss": 0.026, + "step": 9450 + }, + { + "epoch": 4.4919278252611585, + "grad_norm": 0.07521947933626794, + "learning_rate": 3.1182454240953286e-06, + "loss": 0.0261, + "step": 9460 + }, + { + "epoch": 4.49667616334283, + "grad_norm": 0.06082526236301682, + "learning_rate": 3.060885524732904e-06, + "loss": 0.0254, + "step": 9470 + }, + { + "epoch": 4.501424501424501, + "grad_norm": 0.07112065108837298, + "learning_rate": 3.0040414386023117e-06, + "loss": 0.0262, + "step": 9480 + }, + { + "epoch": 4.506172839506172, + "grad_norm": 0.06829956390554137, + "learning_rate": 2.9477137903623887e-06, + "loss": 0.0267, + "step": 9490 + }, + { + "epoch": 4.510921177587845, + "grad_norm": 0.0607207587540697, + "learning_rate": 2.891903198996815e-06, + "loss": 0.0264, + "step": 9500 + }, + { + "epoch": 4.515669515669516, + "grad_norm": 0.06527486026244277, + "learning_rate": 2.8366102778073554e-06, + "loss": 0.0259, + "step": 9510 + }, + { + "epoch": 4.520417853751187, + "grad_norm": 0.06547488953865542, + "learning_rate": 2.7818356344070985e-06, + "loss": 0.026, + "step": 9520 + }, + { + "epoch": 4.5251661918328585, + "grad_norm": 0.07347097536663487, + "learning_rate": 2.7275798707137866e-06, + "loss": 0.0269, + "step": 9530 + }, + { + "epoch": 4.52991452991453, + "grad_norm": 0.07241915094759165, + "learning_rate": 2.6738435829432085e-06, + "loss": 0.0256, + "step": 9540 + }, + { + "epoch": 4.534662867996201, + "grad_norm": 0.06797010357033437, + "learning_rate": 2.6206273616026123e-06, + "loss": 0.0258, + "step": 9550 + }, + { + "epoch": 4.539411206077872, + "grad_norm": 0.06456889557055868, + "learning_rate": 2.567931791484285e-06, + "loss": 0.0256, + "step": 9560 + }, + { + "epoch": 4.544159544159545, + "grad_norm": 0.06144708945522269, + "learning_rate": 2.515757451659051e-06, + "loss": 0.0259, + "step": 9570 + }, + { + "epoch": 4.548907882241216, + "grad_norm": 0.06099158910437045, + "learning_rate": 2.464104915469967e-06, + "loss": 0.0256, + "step": 9580 + }, + { + "epoch": 4.553656220322887, + "grad_norm": 0.0648455343733736, + "learning_rate": 2.4129747505259703e-06, + "loss": 0.0265, + "step": 9590 + }, + { + "epoch": 4.5584045584045585, + "grad_norm": 0.06749234390866253, + "learning_rate": 2.3623675186957073e-06, + "loss": 0.0265, + "step": 9600 + }, + { + "epoch": 4.56315289648623, + "grad_norm": 0.05942386937354067, + "learning_rate": 2.3122837761012707e-06, + "loss": 0.0251, + "step": 9610 + }, + { + "epoch": 4.567901234567901, + "grad_norm": 0.06843582700941665, + "learning_rate": 2.2627240731121822e-06, + "loss": 0.0266, + "step": 9620 + }, + { + "epoch": 4.572649572649572, + "grad_norm": 0.0668613816801879, + "learning_rate": 2.2136889543392715e-06, + "loss": 0.0264, + "step": 9630 + }, + { + "epoch": 4.577397910731244, + "grad_norm": 0.06117340424938218, + "learning_rate": 2.1651789586287442e-06, + "loss": 0.0267, + "step": 9640 + }, + { + "epoch": 4.582146248812915, + "grad_norm": 0.06963558041478746, + "learning_rate": 2.117194619056212e-06, + "loss": 0.0261, + "step": 9650 + }, + { + "epoch": 4.586894586894587, + "grad_norm": 0.06891817829549636, + "learning_rate": 2.0697364629208626e-06, + "loss": 0.0263, + "step": 9660 + }, + { + "epoch": 4.5916429249762585, + "grad_norm": 0.06690608904409279, + "learning_rate": 2.0228050117396813e-06, + "loss": 0.0263, + "step": 9670 + }, + { + "epoch": 4.59639126305793, + "grad_norm": 0.06791072216175136, + "learning_rate": 1.976400781241694e-06, + "loss": 0.0268, + "step": 9680 + }, + { + "epoch": 4.601139601139601, + "grad_norm": 0.06794063027598567, + "learning_rate": 1.930524281362278e-06, + "loss": 0.0264, + "step": 9690 + }, + { + "epoch": 4.605887939221272, + "grad_norm": 0.06868300065773, + "learning_rate": 1.8851760162376275e-06, + "loss": 0.0262, + "step": 9700 + }, + { + "epoch": 4.610636277302944, + "grad_norm": 0.06759599755359404, + "learning_rate": 1.8403564841991515e-06, + "loss": 0.0261, + "step": 9710 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.06613037525460631, + "learning_rate": 1.7960661777680188e-06, + "loss": 0.0267, + "step": 9720 + }, + { + "epoch": 4.620132953466287, + "grad_norm": 0.06143127682397426, + "learning_rate": 1.752305583649755e-06, + "loss": 0.0265, + "step": 9730 + }, + { + "epoch": 4.6248812915479585, + "grad_norm": 0.06046758616448863, + "learning_rate": 1.7090751827288643e-06, + "loss": 0.0261, + "step": 9740 + }, + { + "epoch": 4.62962962962963, + "grad_norm": 0.06387553229166308, + "learning_rate": 1.6663754500635954e-06, + "loss": 0.0257, + "step": 9750 + }, + { + "epoch": 4.634377967711301, + "grad_norm": 0.07136864745542577, + "learning_rate": 1.6242068548806777e-06, + "loss": 0.0264, + "step": 9760 + }, + { + "epoch": 4.639126305792972, + "grad_norm": 0.06362578788009923, + "learning_rate": 1.58256986057016e-06, + "loss": 0.0255, + "step": 9770 + }, + { + "epoch": 4.643874643874644, + "grad_norm": 0.06212751381136991, + "learning_rate": 1.5414649246803693e-06, + "loss": 0.0256, + "step": 9780 + }, + { + "epoch": 4.648622981956315, + "grad_norm": 0.06281729670976327, + "learning_rate": 1.5008924989128258e-06, + "loss": 0.0266, + "step": 9790 + }, + { + "epoch": 4.653371320037987, + "grad_norm": 0.06539493413768768, + "learning_rate": 1.4608530291173317e-06, + "loss": 0.0261, + "step": 9800 + }, + { + "epoch": 4.6581196581196584, + "grad_norm": 0.06753536769995627, + "learning_rate": 1.4213469552870006e-06, + "loss": 0.0263, + "step": 9810 + }, + { + "epoch": 4.66286799620133, + "grad_norm": 0.07182533867167326, + "learning_rate": 1.3823747115535135e-06, + "loss": 0.0264, + "step": 9820 + }, + { + "epoch": 4.667616334283001, + "grad_norm": 0.06505771817557507, + "learning_rate": 1.3439367261822599e-06, + "loss": 0.0272, + "step": 9830 + }, + { + "epoch": 4.672364672364672, + "grad_norm": 0.062414192549997936, + "learning_rate": 1.306033421567704e-06, + "loss": 0.0258, + "step": 9840 + }, + { + "epoch": 4.677113010446344, + "grad_norm": 0.06459192621436703, + "learning_rate": 1.2686652142286926e-06, + "loss": 0.0265, + "step": 9850 + }, + { + "epoch": 4.681861348528015, + "grad_norm": 0.06514878653003345, + "learning_rate": 1.2318325148038989e-06, + "loss": 0.0259, + "step": 9860 + }, + { + "epoch": 4.686609686609686, + "grad_norm": 0.07167434527185934, + "learning_rate": 1.1955357280473145e-06, + "loss": 0.0269, + "step": 9870 + }, + { + "epoch": 4.6913580246913575, + "grad_norm": 0.060691920296567094, + "learning_rate": 1.159775252823786e-06, + "loss": 0.0256, + "step": 9880 + }, + { + "epoch": 4.69610636277303, + "grad_norm": 0.05753931264670235, + "learning_rate": 1.1245514821046467e-06, + "loss": 0.0263, + "step": 9890 + }, + { + "epoch": 4.700854700854701, + "grad_norm": 0.07094991592552678, + "learning_rate": 1.0898648029633807e-06, + "loss": 0.0259, + "step": 9900 + }, + { + "epoch": 4.705603038936372, + "grad_norm": 0.0702583571686615, + "learning_rate": 1.055715596571405e-06, + "loss": 0.0269, + "step": 9910 + }, + { + "epoch": 4.710351377018044, + "grad_norm": 0.06545124129557432, + "learning_rate": 1.0221042381938327e-06, + "loss": 0.0262, + "step": 9920 + }, + { + "epoch": 4.715099715099715, + "grad_norm": 0.06820635884664827, + "learning_rate": 9.890310971853833e-07, + "loss": 0.0263, + "step": 9930 + }, + { + "epoch": 4.719848053181386, + "grad_norm": 0.064561933183304, + "learning_rate": 9.56496536986312e-07, + "loss": 0.0258, + "step": 9940 + }, + { + "epoch": 4.7245963912630575, + "grad_norm": 0.0639617968605123, + "learning_rate": 9.245009151184203e-07, + "loss": 0.0261, + "step": 9950 + }, + { + "epoch": 4.72934472934473, + "grad_norm": 0.06591579572082445, + "learning_rate": 8.930445831811129e-07, + "loss": 0.0257, + "step": 9960 + }, + { + "epoch": 4.734093067426401, + "grad_norm": 0.06044250129539923, + "learning_rate": 8.62127886847569e-07, + "loss": 0.0259, + "step": 9970 + }, + { + "epoch": 4.738841405508072, + "grad_norm": 0.060518122286405074, + "learning_rate": 8.317511658609e-07, + "loss": 0.0266, + "step": 9980 + }, + { + "epoch": 4.743589743589744, + "grad_norm": 0.06682896435975068, + "learning_rate": 8.019147540304306e-07, + "loss": 0.0258, + "step": 9990 + }, + { + "epoch": 4.748338081671415, + "grad_norm": 0.0761615568818603, + "learning_rate": 7.726189792280625e-07, + "loss": 0.0264, + "step": 10000 + }, + { + "epoch": 4.753086419753086, + "grad_norm": 0.06284973785359506, + "learning_rate": 7.438641633846166e-07, + "loss": 0.0261, + "step": 10010 + }, + { + "epoch": 4.7578347578347575, + "grad_norm": 0.06227933184778297, + "learning_rate": 7.156506224863468e-07, + "loss": 0.0257, + "step": 10020 + }, + { + "epoch": 4.76258309591643, + "grad_norm": 0.06686519156608015, + "learning_rate": 6.879786665714316e-07, + "loss": 0.0262, + "step": 10030 + }, + { + "epoch": 4.767331433998101, + "grad_norm": 0.06789117543918491, + "learning_rate": 6.608485997265878e-07, + "loss": 0.0259, + "step": 10040 + }, + { + "epoch": 4.772079772079772, + "grad_norm": 0.05962869200265124, + "learning_rate": 6.34260720083707e-07, + "loss": 0.0262, + "step": 10050 + }, + { + "epoch": 4.776828110161444, + "grad_norm": 0.06300301814231202, + "learning_rate": 6.08215319816613e-07, + "loss": 0.0263, + "step": 10060 + }, + { + "epoch": 4.781576448243115, + "grad_norm": 0.0717569189853655, + "learning_rate": 5.827126851378206e-07, + "loss": 0.0262, + "step": 10070 + }, + { + "epoch": 4.786324786324786, + "grad_norm": 0.06337524763981375, + "learning_rate": 5.577530962953992e-07, + "loss": 0.0256, + "step": 10080 + }, + { + "epoch": 4.7910731244064575, + "grad_norm": 0.06218669461371556, + "learning_rate": 5.33336827569908e-07, + "loss": 0.0255, + "step": 10090 + }, + { + "epoch": 4.795821462488129, + "grad_norm": 0.06112078587918284, + "learning_rate": 5.09464147271349e-07, + "loss": 0.026, + "step": 10100 + }, + { + "epoch": 4.8005698005698, + "grad_norm": 0.06212587641494024, + "learning_rate": 4.861353177362638e-07, + "loss": 0.026, + "step": 10110 + }, + { + "epoch": 4.805318138651472, + "grad_norm": 0.06132639770751258, + "learning_rate": 4.633505953248074e-07, + "loss": 0.0255, + "step": 10120 + }, + { + "epoch": 4.810066476733144, + "grad_norm": 0.07187707472949913, + "learning_rate": 4.411102304179571e-07, + "loss": 0.0262, + "step": 10130 + }, + { + "epoch": 4.814814814814815, + "grad_norm": 0.06185055748956023, + "learning_rate": 4.1941446741475267e-07, + "loss": 0.0263, + "step": 10140 + }, + { + "epoch": 4.819563152896486, + "grad_norm": 0.06252001589738033, + "learning_rate": 3.9826354472961016e-07, + "loss": 0.0265, + "step": 10150 + }, + { + "epoch": 4.8243114909781575, + "grad_norm": 0.06603812262385446, + "learning_rate": 3.7765769478970704e-07, + "loss": 0.026, + "step": 10160 + }, + { + "epoch": 4.829059829059829, + "grad_norm": 0.06477537423872996, + "learning_rate": 3.575971440324233e-07, + "loss": 0.0258, + "step": 10170 + }, + { + "epoch": 4.8338081671415, + "grad_norm": 0.06458805726132752, + "learning_rate": 3.380821129028489e-07, + "loss": 0.0258, + "step": 10180 + }, + { + "epoch": 4.838556505223172, + "grad_norm": 0.0697002624596936, + "learning_rate": 3.191128158513745e-07, + "loss": 0.0258, + "step": 10190 + }, + { + "epoch": 4.843304843304844, + "grad_norm": 0.06790311601058047, + "learning_rate": 3.0068946133132713e-07, + "loss": 0.0261, + "step": 10200 + }, + { + "epoch": 4.848053181386515, + "grad_norm": 0.07075387146391178, + "learning_rate": 2.828122517966658e-07, + "loss": 0.0261, + "step": 10210 + }, + { + "epoch": 4.852801519468186, + "grad_norm": 0.0608513989736112, + "learning_rate": 2.6548138369978935e-07, + "loss": 0.0258, + "step": 10220 + }, + { + "epoch": 4.8575498575498575, + "grad_norm": 0.06440478096022074, + "learning_rate": 2.4869704748934907e-07, + "loss": 0.0259, + "step": 10230 + }, + { + "epoch": 4.862298195631529, + "grad_norm": 0.06709277616276473, + "learning_rate": 2.3245942760816152e-07, + "loss": 0.026, + "step": 10240 + }, + { + "epoch": 4.8670465337132, + "grad_norm": 0.07034232290150784, + "learning_rate": 2.167687024911935e-07, + "loss": 0.027, + "step": 10250 + }, + { + "epoch": 4.871794871794872, + "grad_norm": 0.06610420221452963, + "learning_rate": 2.0162504456359143e-07, + "loss": 0.0254, + "step": 10260 + }, + { + "epoch": 4.8765432098765435, + "grad_norm": 0.07251650143366604, + "learning_rate": 1.8702862023878277e-07, + "loss": 0.0266, + "step": 10270 + }, + { + "epoch": 4.881291547958215, + "grad_norm": 0.06534152636248931, + "learning_rate": 1.7297958991666085e-07, + "loss": 0.0256, + "step": 10280 + }, + { + "epoch": 4.886039886039886, + "grad_norm": 0.06617116515041509, + "learning_rate": 1.5947810798181416e-07, + "loss": 0.0262, + "step": 10290 + }, + { + "epoch": 4.8907882241215574, + "grad_norm": 0.06409761983607384, + "learning_rate": 1.4652432280182203e-07, + "loss": 0.0263, + "step": 10300 + }, + { + "epoch": 4.895536562203229, + "grad_norm": 0.065028269071596, + "learning_rate": 1.3411837672564487e-07, + "loss": 0.027, + "step": 10310 + }, + { + "epoch": 4.9002849002849, + "grad_norm": 0.07435811378768264, + "learning_rate": 1.2226040608202537e-07, + "loss": 0.0263, + "step": 10320 + }, + { + "epoch": 4.905033238366571, + "grad_norm": 0.0673580953614697, + "learning_rate": 1.1095054117804537e-07, + "loss": 0.0262, + "step": 10330 + }, + { + "epoch": 4.909781576448243, + "grad_norm": 0.06359865493219967, + "learning_rate": 1.0018890629762689e-07, + "loss": 0.0261, + "step": 10340 + }, + { + "epoch": 4.914529914529915, + "grad_norm": 0.06354481010434058, + "learning_rate": 8.997561970022217e-08, + "loss": 0.0264, + "step": 10350 + }, + { + "epoch": 4.919278252611586, + "grad_norm": 0.06309081205061498, + "learning_rate": 8.03107936194869e-08, + "loss": 0.0265, + "step": 10360 + }, + { + "epoch": 4.924026590693257, + "grad_norm": 0.06293005230951278, + "learning_rate": 7.119453426205347e-08, + "loss": 0.0254, + "step": 10370 + }, + { + "epoch": 4.928774928774929, + "grad_norm": 0.06623336138338565, + "learning_rate": 6.262694180634853e-08, + "loss": 0.0264, + "step": 10380 + }, + { + "epoch": 4.9335232668566, + "grad_norm": 0.05908514628490637, + "learning_rate": 5.4608110401532796e-08, + "loss": 0.0258, + "step": 10390 + }, + { + "epoch": 4.938271604938271, + "grad_norm": 0.0667055567075738, + "learning_rate": 4.713812816641849e-08, + "loss": 0.0267, + "step": 10400 + }, + { + "epoch": 4.943019943019943, + "grad_norm": 0.06796672206275663, + "learning_rate": 4.0217077188536845e-08, + "loss": 0.0261, + "step": 10410 + }, + { + "epoch": 4.947768281101615, + "grad_norm": 0.06214908179099578, + "learning_rate": 3.38450335231999e-08, + "loss": 0.0267, + "step": 10420 + }, + { + "epoch": 4.952516619183286, + "grad_norm": 0.05932075921731385, + "learning_rate": 2.8022067192712277e-08, + "loss": 0.0251, + "step": 10430 + }, + { + "epoch": 4.957264957264957, + "grad_norm": 0.0660746944898574, + "learning_rate": 2.274824218556071e-08, + "loss": 0.0256, + "step": 10440 + }, + { + "epoch": 4.962013295346629, + "grad_norm": 0.06736128344207919, + "learning_rate": 1.802361645573125e-08, + "loss": 0.0265, + "step": 10450 + }, + { + "epoch": 4.9667616334283, + "grad_norm": 0.06464509279996226, + "learning_rate": 1.3848241922059802e-08, + "loss": 0.0257, + "step": 10460 + }, + { + "epoch": 4.971509971509971, + "grad_norm": 0.06164991150055645, + "learning_rate": 1.0222164467688089e-08, + "loss": 0.0263, + "step": 10470 + }, + { + "epoch": 4.976258309591643, + "grad_norm": 0.06476893734374878, + "learning_rate": 7.1454239395141175e-09, + "loss": 0.0266, + "step": 10480 + }, + { + "epoch": 4.981006647673315, + "grad_norm": 0.06701246000400526, + "learning_rate": 4.618054147803585e-09, + "loss": 0.0262, + "step": 10490 + }, + { + "epoch": 4.985754985754986, + "grad_norm": 0.06969654083865987, + "learning_rate": 2.6400828657791033e-09, + "loss": 0.0251, + "step": 10500 + }, + { + "epoch": 4.990503323836657, + "grad_norm": 0.06214880164097315, + "learning_rate": 1.211531829337087e-09, + "loss": 0.0265, + "step": 10510 + }, + { + "epoch": 4.995251661918329, + "grad_norm": 0.06068832149595113, + "learning_rate": 3.324167368090603e-10, + "loss": 0.0258, + "step": 10520 + }, + { + "epoch": 5.0, + "grad_norm": 0.0639583096546201, + "learning_rate": 2.747248778467437e-12, + "loss": 0.0264, + "step": 10530 + }, + { + "epoch": 5.0, + "step": 10530, + "total_flos": 951688701149184.0, + "train_loss": 0.19493026600377625, + "train_runtime": 28802.8725, + "train_samples_per_second": 187.169, + "train_steps_per_second": 0.366 + } + ], + "logging_steps": 10, + "max_steps": 10530, + "num_input_tokens_seen": 0, + "num_train_epochs": 5, + "save_steps": 1800, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 951688701149184.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +}