{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.564102564102564, "eval_steps": 500, "global_step": 5400, "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 } ], "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": false }, "attributes": {} } }, "total_flos": 488054625140736.0, "train_batch_size": 128, "trial_name": null, "trial_params": null }