{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 60000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 0.906653106212616, "learning_rate": 3.0000000000000004e-07, "loss": 1.5679, "step": 10 }, { "grad_norm": 0.8693776726722717, "learning_rate": 6.333333333333333e-07, "loss": 1.5623, "step": 20 }, { "grad_norm": 0.9220455884933472, "learning_rate": 9.666666666666668e-07, "loss": 1.5612, "step": 30 }, { "grad_norm": 0.7960800528526306, "learning_rate": 1.3e-06, "loss": 1.5487, "step": 40 }, { "grad_norm": 0.9122224450111389, "learning_rate": 1.6333333333333333e-06, "loss": 1.5072, "step": 50 }, { "grad_norm": 0.741705596446991, "learning_rate": 1.9666666666666668e-06, "loss": 1.4242, "step": 60 }, { "grad_norm": 1.0284627676010132, "learning_rate": 2.3e-06, "loss": 1.3272, "step": 70 }, { "grad_norm": 0.8586296439170837, "learning_rate": 2.6333333333333337e-06, "loss": 1.1778, "step": 80 }, { "grad_norm": 0.5516090393066406, "learning_rate": 2.966666666666667e-06, "loss": 1.1142, "step": 90 }, { "grad_norm": 0.5271621346473694, "learning_rate": 3.3e-06, "loss": 1.087, "step": 100 }, { "grad_norm": 0.4270395338535309, "learning_rate": 3.633333333333334e-06, "loss": 1.0687, "step": 110 }, { "grad_norm": 0.3792906105518341, "learning_rate": 3.966666666666667e-06, "loss": 1.0592, "step": 120 }, { "grad_norm": 0.48638615012168884, "learning_rate": 4.2999999999999995e-06, "loss": 1.0591, "step": 130 }, { "grad_norm": 0.49145615100860596, "learning_rate": 4.633333333333334e-06, "loss": 1.0574, "step": 140 }, { "grad_norm": 0.47549161314964294, "learning_rate": 4.966666666666667e-06, "loss": 1.05, "step": 150 }, { "grad_norm": 0.519702672958374, "learning_rate": 5.3e-06, "loss": 1.0486, "step": 160 }, { "grad_norm": 0.4958411157131195, "learning_rate": 5.633333333333333e-06, "loss": 1.0373, "step": 170 }, { "grad_norm": 0.46167275309562683, "learning_rate": 5.9666666666666666e-06, "loss": 1.0361, "step": 180 }, { "grad_norm": 0.8041318655014038, "learning_rate": 6.300000000000001e-06, "loss": 1.0248, "step": 190 }, { "grad_norm": 0.6241366267204285, "learning_rate": 6.633333333333333e-06, "loss": 1.0101, "step": 200 }, { "grad_norm": 0.6976537704467773, "learning_rate": 6.966666666666667e-06, "loss": 0.9917, "step": 210 }, { "grad_norm": 1.0786925554275513, "learning_rate": 7.2999999999999996e-06, "loss": 0.9413, "step": 220 }, { "grad_norm": 1.0677145719528198, "learning_rate": 7.633333333333334e-06, "loss": 0.8824, "step": 230 }, { "grad_norm": 1.2146179676055908, "learning_rate": 7.966666666666666e-06, "loss": 0.8175, "step": 240 }, { "grad_norm": 0.9864868521690369, "learning_rate": 8.3e-06, "loss": 0.7308, "step": 250 }, { "grad_norm": 1.229026198387146, "learning_rate": 8.633333333333334e-06, "loss": 0.647, "step": 260 }, { "grad_norm": 1.3828271627426147, "learning_rate": 8.966666666666668e-06, "loss": 0.576, "step": 270 }, { "grad_norm": 1.549126386642456, "learning_rate": 9.3e-06, "loss": 0.5081, "step": 280 }, { "grad_norm": 1.2539232969284058, "learning_rate": 9.633333333333335e-06, "loss": 0.4553, "step": 290 }, { "grad_norm": 1.260278344154358, "learning_rate": 9.966666666666667e-06, "loss": 0.4148, "step": 300 }, { "grad_norm": 1.3500916957855225, "learning_rate": 1.03e-05, "loss": 0.3763, "step": 310 }, { "grad_norm": 1.5110249519348145, "learning_rate": 1.0633333333333334e-05, "loss": 0.3172, "step": 320 }, { "grad_norm": 1.4444080591201782, "learning_rate": 1.0966666666666666e-05, "loss": 0.2915, "step": 330 }, { "grad_norm": 1.6018574237823486, "learning_rate": 1.13e-05, "loss": 0.2776, "step": 340 }, { "grad_norm": 1.6199555397033691, "learning_rate": 1.1633333333333334e-05, "loss": 0.2322, "step": 350 }, { "grad_norm": 1.4146711826324463, "learning_rate": 1.1966666666666668e-05, "loss": 0.2147, "step": 360 }, { "grad_norm": 1.5137650966644287, "learning_rate": 1.23e-05, "loss": 0.1728, "step": 370 }, { "grad_norm": 1.791176676750183, "learning_rate": 1.2633333333333333e-05, "loss": 0.1546, "step": 380 }, { "grad_norm": 1.4381294250488281, "learning_rate": 1.2966666666666669e-05, "loss": 0.1473, "step": 390 }, { "grad_norm": 1.3789844512939453, "learning_rate": 1.3300000000000001e-05, "loss": 0.1338, "step": 400 }, { "grad_norm": 1.1485520601272583, "learning_rate": 1.3633333333333334e-05, "loss": 0.1339, "step": 410 }, { "grad_norm": 1.0684387683868408, "learning_rate": 1.3966666666666666e-05, "loss": 0.1216, "step": 420 }, { "grad_norm": 1.4529162645339966, "learning_rate": 1.43e-05, "loss": 0.1232, "step": 430 }, { "grad_norm": 1.2607598304748535, "learning_rate": 1.4633333333333334e-05, "loss": 0.1109, "step": 440 }, { "grad_norm": 1.601815104484558, "learning_rate": 1.4966666666666668e-05, "loss": 0.111, "step": 450 }, { "grad_norm": 1.3750135898590088, "learning_rate": 1.53e-05, "loss": 0.1023, "step": 460 }, { "grad_norm": 1.0965032577514648, "learning_rate": 1.563333333333333e-05, "loss": 0.0938, "step": 470 }, { "grad_norm": 1.3191807270050049, "learning_rate": 1.5966666666666667e-05, "loss": 0.0983, "step": 480 }, { "grad_norm": 1.4540693759918213, "learning_rate": 1.63e-05, "loss": 0.0942, "step": 490 }, { "grad_norm": 1.4025064706802368, "learning_rate": 1.6633333333333336e-05, "loss": 0.0979, "step": 500 }, { "grad_norm": 1.181281328201294, "learning_rate": 1.6966666666666668e-05, "loss": 0.0871, "step": 510 }, { "grad_norm": 1.5458253622055054, "learning_rate": 1.73e-05, "loss": 0.0884, "step": 520 }, { "grad_norm": 1.375321626663208, "learning_rate": 1.7633333333333336e-05, "loss": 0.0905, "step": 530 }, { "grad_norm": 1.4015212059020996, "learning_rate": 1.796666666666667e-05, "loss": 0.0867, "step": 540 }, { "grad_norm": 1.2696318626403809, "learning_rate": 1.83e-05, "loss": 0.0727, "step": 550 }, { "grad_norm": 1.2652809619903564, "learning_rate": 1.8633333333333333e-05, "loss": 0.084, "step": 560 }, { "grad_norm": 1.2870091199874878, "learning_rate": 1.896666666666667e-05, "loss": 0.0849, "step": 570 }, { "grad_norm": 1.7589884996414185, "learning_rate": 1.93e-05, "loss": 0.0819, "step": 580 }, { "grad_norm": 1.437740683555603, "learning_rate": 1.9633333333333334e-05, "loss": 0.0771, "step": 590 }, { "grad_norm": 1.2285255193710327, "learning_rate": 1.9966666666666666e-05, "loss": 0.0824, "step": 600 }, { "grad_norm": 1.4716416597366333, "learning_rate": 2.0300000000000002e-05, "loss": 0.0727, "step": 610 }, { "grad_norm": 1.3391598463058472, "learning_rate": 2.0633333333333335e-05, "loss": 0.0743, "step": 620 }, { "grad_norm": 0.9616796970367432, "learning_rate": 2.0966666666666667e-05, "loss": 0.0718, "step": 630 }, { "grad_norm": 1.530563473701477, "learning_rate": 2.13e-05, "loss": 0.0762, "step": 640 }, { "grad_norm": 1.4703972339630127, "learning_rate": 2.1633333333333332e-05, "loss": 0.0739, "step": 650 }, { "grad_norm": 1.7144321203231812, "learning_rate": 2.1966666666666668e-05, "loss": 0.0695, "step": 660 }, { "grad_norm": 1.274772047996521, "learning_rate": 2.23e-05, "loss": 0.0664, "step": 670 }, { "grad_norm": 1.7366373538970947, "learning_rate": 2.2633333333333336e-05, "loss": 0.0651, "step": 680 }, { "grad_norm": 1.081359624862671, "learning_rate": 2.2966666666666668e-05, "loss": 0.0609, "step": 690 }, { "grad_norm": 1.2969530820846558, "learning_rate": 2.3300000000000004e-05, "loss": 0.0653, "step": 700 }, { "grad_norm": 1.1062114238739014, "learning_rate": 2.3633333333333336e-05, "loss": 0.0613, "step": 710 }, { "grad_norm": 1.4192845821380615, "learning_rate": 2.396666666666667e-05, "loss": 0.0662, "step": 720 }, { "grad_norm": 1.452603816986084, "learning_rate": 2.43e-05, "loss": 0.0723, "step": 730 }, { "grad_norm": 1.3256062269210815, "learning_rate": 2.4633333333333334e-05, "loss": 0.0701, "step": 740 }, { "grad_norm": 1.3684066534042358, "learning_rate": 2.496666666666667e-05, "loss": 0.0694, "step": 750 }, { "grad_norm": 1.6046570539474487, "learning_rate": 2.5300000000000002e-05, "loss": 0.0618, "step": 760 }, { "grad_norm": 1.3670393228530884, "learning_rate": 2.5633333333333338e-05, "loss": 0.0672, "step": 770 }, { "grad_norm": 1.3244746923446655, "learning_rate": 2.5966666666666667e-05, "loss": 0.0639, "step": 780 }, { "grad_norm": 1.2938417196273804, "learning_rate": 2.6300000000000002e-05, "loss": 0.0548, "step": 790 }, { "grad_norm": 1.5550148487091064, "learning_rate": 2.663333333333333e-05, "loss": 0.058, "step": 800 }, { "grad_norm": 1.404503583908081, "learning_rate": 2.6966666666666667e-05, "loss": 0.0516, "step": 810 }, { "grad_norm": 1.3440333604812622, "learning_rate": 2.7300000000000003e-05, "loss": 0.0548, "step": 820 }, { "grad_norm": 1.0769469738006592, "learning_rate": 2.7633333333333332e-05, "loss": 0.0552, "step": 830 }, { "grad_norm": 1.4326727390289307, "learning_rate": 2.7966666666666668e-05, "loss": 0.0583, "step": 840 }, { "grad_norm": 1.4587595462799072, "learning_rate": 2.83e-05, "loss": 0.0583, "step": 850 }, { "grad_norm": 1.3562586307525635, "learning_rate": 2.8633333333333336e-05, "loss": 0.0558, "step": 860 }, { "grad_norm": 1.3250821828842163, "learning_rate": 2.8966666666666668e-05, "loss": 0.0537, "step": 870 }, { "grad_norm": 1.2373030185699463, "learning_rate": 2.93e-05, "loss": 0.0571, "step": 880 }, { "grad_norm": 1.4397591352462769, "learning_rate": 2.9633333333333336e-05, "loss": 0.0548, "step": 890 }, { "grad_norm": 1.051496982574463, "learning_rate": 2.9966666666666672e-05, "loss": 0.0594, "step": 900 }, { "grad_norm": 1.0722253322601318, "learning_rate": 3.03e-05, "loss": 0.0518, "step": 910 }, { "grad_norm": 1.0756639242172241, "learning_rate": 3.063333333333334e-05, "loss": 0.0497, "step": 920 }, { "grad_norm": 1.261473298072815, "learning_rate": 3.096666666666666e-05, "loss": 0.0547, "step": 930 }, { "grad_norm": 1.4107327461242676, "learning_rate": 3.13e-05, "loss": 0.0508, "step": 940 }, { "grad_norm": 1.2446612119674683, "learning_rate": 3.1633333333333334e-05, "loss": 0.058, "step": 950 }, { "grad_norm": 1.3263245820999146, "learning_rate": 3.196666666666667e-05, "loss": 0.0513, "step": 960 }, { "grad_norm": 1.263375163078308, "learning_rate": 3.2300000000000006e-05, "loss": 0.051, "step": 970 }, { "grad_norm": 0.9138875603675842, "learning_rate": 3.263333333333333e-05, "loss": 0.048, "step": 980 }, { "grad_norm": 0.9764530062675476, "learning_rate": 3.296666666666667e-05, "loss": 0.0551, "step": 990 }, { "grad_norm": 0.838601291179657, "learning_rate": 3.33e-05, "loss": 0.0465, "step": 1000 }, { "grad_norm": 1.1556882858276367, "learning_rate": 3.3633333333333335e-05, "loss": 0.0454, "step": 1010 }, { "grad_norm": 1.3072545528411865, "learning_rate": 3.396666666666667e-05, "loss": 0.0531, "step": 1020 }, { "grad_norm": 1.1065551042556763, "learning_rate": 3.430000000000001e-05, "loss": 0.0458, "step": 1030 }, { "grad_norm": 1.2423796653747559, "learning_rate": 3.463333333333333e-05, "loss": 0.0444, "step": 1040 }, { "grad_norm": 1.0679831504821777, "learning_rate": 3.496666666666667e-05, "loss": 0.0422, "step": 1050 }, { "grad_norm": 0.900173544883728, "learning_rate": 3.53e-05, "loss": 0.0443, "step": 1060 }, { "grad_norm": 1.4439051151275635, "learning_rate": 3.563333333333334e-05, "loss": 0.0459, "step": 1070 }, { "grad_norm": 1.0438175201416016, "learning_rate": 3.596666666666667e-05, "loss": 0.0474, "step": 1080 }, { "grad_norm": 1.236835241317749, "learning_rate": 3.63e-05, "loss": 0.0444, "step": 1090 }, { "grad_norm": 1.100285291671753, "learning_rate": 3.6633333333333334e-05, "loss": 0.0445, "step": 1100 }, { "grad_norm": 1.0615077018737793, "learning_rate": 3.6966666666666666e-05, "loss": 0.0431, "step": 1110 }, { "grad_norm": 0.969318151473999, "learning_rate": 3.73e-05, "loss": 0.0453, "step": 1120 }, { "grad_norm": 1.0389533042907715, "learning_rate": 3.763333333333334e-05, "loss": 0.0468, "step": 1130 }, { "grad_norm": 1.1142019033432007, "learning_rate": 3.796666666666667e-05, "loss": 0.0451, "step": 1140 }, { "grad_norm": 1.0836576223373413, "learning_rate": 3.83e-05, "loss": 0.0421, "step": 1150 }, { "grad_norm": 1.117935061454773, "learning_rate": 3.8633333333333335e-05, "loss": 0.0472, "step": 1160 }, { "grad_norm": 1.2692526578903198, "learning_rate": 3.896666666666667e-05, "loss": 0.0449, "step": 1170 }, { "grad_norm": 1.280261754989624, "learning_rate": 3.9300000000000007e-05, "loss": 0.0457, "step": 1180 }, { "grad_norm": 1.3592414855957031, "learning_rate": 3.963333333333333e-05, "loss": 0.0413, "step": 1190 }, { "grad_norm": 1.1554778814315796, "learning_rate": 3.996666666666667e-05, "loss": 0.0456, "step": 1200 }, { "grad_norm": 1.0000898838043213, "learning_rate": 4.0300000000000004e-05, "loss": 0.0423, "step": 1210 }, { "grad_norm": 0.9291822910308838, "learning_rate": 4.0633333333333336e-05, "loss": 0.0414, "step": 1220 }, { "grad_norm": 1.133824110031128, "learning_rate": 4.096666666666667e-05, "loss": 0.0444, "step": 1230 }, { "grad_norm": 1.1049919128417969, "learning_rate": 4.13e-05, "loss": 0.043, "step": 1240 }, { "grad_norm": 1.2612816095352173, "learning_rate": 4.1633333333333333e-05, "loss": 0.0434, "step": 1250 }, { "grad_norm": 0.9751136302947998, "learning_rate": 4.196666666666667e-05, "loss": 0.0427, "step": 1260 }, { "grad_norm": 0.7762418985366821, "learning_rate": 4.23e-05, "loss": 0.0416, "step": 1270 }, { "grad_norm": 0.8659252524375916, "learning_rate": 4.263333333333334e-05, "loss": 0.0454, "step": 1280 }, { "grad_norm": 0.8201512694358826, "learning_rate": 4.296666666666666e-05, "loss": 0.0424, "step": 1290 }, { "grad_norm": 0.7975085973739624, "learning_rate": 4.33e-05, "loss": 0.0394, "step": 1300 }, { "grad_norm": 1.0157757997512817, "learning_rate": 4.3633333333333335e-05, "loss": 0.0391, "step": 1310 }, { "grad_norm": 1.165273666381836, "learning_rate": 4.396666666666667e-05, "loss": 0.0383, "step": 1320 }, { "grad_norm": 1.088231086730957, "learning_rate": 4.43e-05, "loss": 0.0456, "step": 1330 }, { "grad_norm": 0.8724924325942993, "learning_rate": 4.463333333333334e-05, "loss": 0.0422, "step": 1340 }, { "grad_norm": 0.9821087718009949, "learning_rate": 4.496666666666667e-05, "loss": 0.0433, "step": 1350 }, { "grad_norm": 0.8653689622879028, "learning_rate": 4.53e-05, "loss": 0.0419, "step": 1360 }, { "grad_norm": 0.8438236117362976, "learning_rate": 4.5633333333333336e-05, "loss": 0.0425, "step": 1370 }, { "grad_norm": 0.9100620150566101, "learning_rate": 4.596666666666667e-05, "loss": 0.0362, "step": 1380 }, { "grad_norm": 0.8023723363876343, "learning_rate": 4.630000000000001e-05, "loss": 0.0383, "step": 1390 }, { "grad_norm": 0.8818423748016357, "learning_rate": 4.663333333333333e-05, "loss": 0.036, "step": 1400 }, { "grad_norm": 0.950262725353241, "learning_rate": 4.696666666666667e-05, "loss": 0.0363, "step": 1410 }, { "grad_norm": 1.0039712190628052, "learning_rate": 4.73e-05, "loss": 0.0367, "step": 1420 }, { "grad_norm": 1.0990114212036133, "learning_rate": 4.763333333333334e-05, "loss": 0.0396, "step": 1430 }, { "grad_norm": 0.5855928659439087, "learning_rate": 4.796666666666667e-05, "loss": 0.0369, "step": 1440 }, { "grad_norm": 0.7812920212745667, "learning_rate": 4.83e-05, "loss": 0.0356, "step": 1450 }, { "grad_norm": 0.8480895757675171, "learning_rate": 4.8633333333333334e-05, "loss": 0.0413, "step": 1460 }, { "grad_norm": 0.8607500195503235, "learning_rate": 4.8966666666666667e-05, "loss": 0.0363, "step": 1470 }, { "grad_norm": 0.8618646860122681, "learning_rate": 4.93e-05, "loss": 0.0383, "step": 1480 }, { "grad_norm": 0.9411969780921936, "learning_rate": 4.963333333333334e-05, "loss": 0.0375, "step": 1490 }, { "grad_norm": 1.043501853942871, "learning_rate": 4.996666666666667e-05, "loss": 0.037, "step": 1500 }, { "grad_norm": 0.9865856766700745, "learning_rate": 5.03e-05, "loss": 0.0387, "step": 1510 }, { "grad_norm": 0.818211555480957, "learning_rate": 5.0633333333333335e-05, "loss": 0.0352, "step": 1520 }, { "grad_norm": 0.8160049319267273, "learning_rate": 5.0966666666666674e-05, "loss": 0.0374, "step": 1530 }, { "grad_norm": 0.8237190842628479, "learning_rate": 5.130000000000001e-05, "loss": 0.0367, "step": 1540 }, { "grad_norm": 0.7737480998039246, "learning_rate": 5.163333333333333e-05, "loss": 0.0346, "step": 1550 }, { "grad_norm": 0.8959475755691528, "learning_rate": 5.196666666666667e-05, "loss": 0.0377, "step": 1560 }, { "grad_norm": 0.853604257106781, "learning_rate": 5.2300000000000004e-05, "loss": 0.037, "step": 1570 }, { "grad_norm": 0.7753662467002869, "learning_rate": 5.2633333333333336e-05, "loss": 0.0352, "step": 1580 }, { "grad_norm": 0.9940308928489685, "learning_rate": 5.296666666666666e-05, "loss": 0.0374, "step": 1590 }, { "grad_norm": 0.9186078310012817, "learning_rate": 5.330000000000001e-05, "loss": 0.0379, "step": 1600 }, { "grad_norm": 0.679489016532898, "learning_rate": 5.3633333333333334e-05, "loss": 0.0343, "step": 1610 }, { "grad_norm": 0.7441301345825195, "learning_rate": 5.3966666666666666e-05, "loss": 0.03, "step": 1620 }, { "grad_norm": 0.8201777935028076, "learning_rate": 5.4300000000000005e-05, "loss": 0.0346, "step": 1630 }, { "grad_norm": 0.8654753565788269, "learning_rate": 5.463333333333334e-05, "loss": 0.032, "step": 1640 }, { "grad_norm": 0.8006584048271179, "learning_rate": 5.496666666666666e-05, "loss": 0.0325, "step": 1650 }, { "grad_norm": 0.5706071257591248, "learning_rate": 5.530000000000001e-05, "loss": 0.0297, "step": 1660 }, { "grad_norm": 0.7328195571899414, "learning_rate": 5.5633333333333335e-05, "loss": 0.0326, "step": 1670 }, { "grad_norm": 0.9193377494812012, "learning_rate": 5.596666666666667e-05, "loss": 0.0342, "step": 1680 }, { "grad_norm": 0.8086913228034973, "learning_rate": 5.63e-05, "loss": 0.0377, "step": 1690 }, { "grad_norm": 0.7799825668334961, "learning_rate": 5.663333333333334e-05, "loss": 0.037, "step": 1700 }, { "grad_norm": 0.7533248662948608, "learning_rate": 5.696666666666667e-05, "loss": 0.0354, "step": 1710 }, { "grad_norm": 0.7485305070877075, "learning_rate": 5.73e-05, "loss": 0.0334, "step": 1720 }, { "grad_norm": 0.7816023826599121, "learning_rate": 5.7633333333333336e-05, "loss": 0.032, "step": 1730 }, { "grad_norm": 0.7906578779220581, "learning_rate": 5.796666666666667e-05, "loss": 0.0314, "step": 1740 }, { "grad_norm": 0.7718600034713745, "learning_rate": 5.83e-05, "loss": 0.0378, "step": 1750 }, { "grad_norm": 0.9219616651535034, "learning_rate": 5.863333333333334e-05, "loss": 0.0337, "step": 1760 }, { "grad_norm": 1.0989402532577515, "learning_rate": 5.896666666666667e-05, "loss": 0.0372, "step": 1770 }, { "grad_norm": 0.5807214379310608, "learning_rate": 5.93e-05, "loss": 0.0341, "step": 1780 }, { "grad_norm": 0.6444985270500183, "learning_rate": 5.9633333333333344e-05, "loss": 0.0363, "step": 1790 }, { "grad_norm": 0.9628150463104248, "learning_rate": 5.996666666666667e-05, "loss": 0.0329, "step": 1800 }, { "grad_norm": 0.7154569029808044, "learning_rate": 6.03e-05, "loss": 0.0298, "step": 1810 }, { "grad_norm": 0.8449196815490723, "learning_rate": 6.063333333333333e-05, "loss": 0.0335, "step": 1820 }, { "grad_norm": 0.7128809094429016, "learning_rate": 6.0966666666666674e-05, "loss": 0.0311, "step": 1830 }, { "grad_norm": 0.6906906366348267, "learning_rate": 6.13e-05, "loss": 0.0333, "step": 1840 }, { "grad_norm": 0.7234964966773987, "learning_rate": 6.163333333333333e-05, "loss": 0.034, "step": 1850 }, { "grad_norm": 0.6930971145629883, "learning_rate": 6.196666666666668e-05, "loss": 0.0351, "step": 1860 }, { "grad_norm": 0.8179806470870972, "learning_rate": 6.23e-05, "loss": 0.0329, "step": 1870 }, { "grad_norm": 0.6271581053733826, "learning_rate": 6.263333333333333e-05, "loss": 0.033, "step": 1880 }, { "grad_norm": 0.7783054709434509, "learning_rate": 6.296666666666667e-05, "loss": 0.0364, "step": 1890 }, { "grad_norm": 0.5584641098976135, "learning_rate": 6.330000000000001e-05, "loss": 0.0338, "step": 1900 }, { "grad_norm": 0.5476588010787964, "learning_rate": 6.363333333333334e-05, "loss": 0.0322, "step": 1910 }, { "grad_norm": 0.5505942106246948, "learning_rate": 6.396666666666667e-05, "loss": 0.0311, "step": 1920 }, { "grad_norm": 0.6644856929779053, "learning_rate": 6.43e-05, "loss": 0.0324, "step": 1930 }, { "grad_norm": 0.7084763050079346, "learning_rate": 6.463333333333334e-05, "loss": 0.0305, "step": 1940 }, { "grad_norm": 0.8100665211677551, "learning_rate": 6.496666666666667e-05, "loss": 0.0313, "step": 1950 }, { "grad_norm": 0.4890957176685333, "learning_rate": 6.53e-05, "loss": 0.0304, "step": 1960 }, { "grad_norm": 0.8337940573692322, "learning_rate": 6.563333333333333e-05, "loss": 0.0315, "step": 1970 }, { "grad_norm": 0.6948914527893066, "learning_rate": 6.596666666666667e-05, "loss": 0.0311, "step": 1980 }, { "grad_norm": 0.6527019739151001, "learning_rate": 6.630000000000001e-05, "loss": 0.0288, "step": 1990 }, { "grad_norm": 0.6371620297431946, "learning_rate": 6.663333333333333e-05, "loss": 0.0287, "step": 2000 }, { "grad_norm": 0.7378381490707397, "learning_rate": 6.696666666666666e-05, "loss": 0.0274, "step": 2010 }, { "grad_norm": 0.6902281641960144, "learning_rate": 6.730000000000001e-05, "loss": 0.0322, "step": 2020 }, { "grad_norm": 0.7032964825630188, "learning_rate": 6.763333333333334e-05, "loss": 0.0299, "step": 2030 }, { "grad_norm": 0.7114623188972473, "learning_rate": 6.796666666666666e-05, "loss": 0.0272, "step": 2040 }, { "grad_norm": 0.7903315424919128, "learning_rate": 6.83e-05, "loss": 0.0307, "step": 2050 }, { "grad_norm": 0.8273006081581116, "learning_rate": 6.863333333333334e-05, "loss": 0.0301, "step": 2060 }, { "grad_norm": 0.754179835319519, "learning_rate": 6.896666666666667e-05, "loss": 0.0292, "step": 2070 }, { "grad_norm": 0.6697760820388794, "learning_rate": 6.93e-05, "loss": 0.0271, "step": 2080 }, { "grad_norm": 0.6847582459449768, "learning_rate": 6.963333333333334e-05, "loss": 0.0315, "step": 2090 }, { "grad_norm": 0.7181375026702881, "learning_rate": 6.996666666666667e-05, "loss": 0.0286, "step": 2100 }, { "grad_norm": 0.5362321138381958, "learning_rate": 7.03e-05, "loss": 0.0257, "step": 2110 }, { "grad_norm": 0.7989697456359863, "learning_rate": 7.063333333333333e-05, "loss": 0.0278, "step": 2120 }, { "grad_norm": 0.5767647624015808, "learning_rate": 7.096666666666667e-05, "loss": 0.0317, "step": 2130 }, { "grad_norm": 0.8415308594703674, "learning_rate": 7.13e-05, "loss": 0.0279, "step": 2140 }, { "grad_norm": 0.5926703810691833, "learning_rate": 7.163333333333334e-05, "loss": 0.0273, "step": 2150 }, { "grad_norm": 0.5614122152328491, "learning_rate": 7.196666666666668e-05, "loss": 0.0264, "step": 2160 }, { "grad_norm": 0.46337878704071045, "learning_rate": 7.23e-05, "loss": 0.0263, "step": 2170 }, { "grad_norm": 0.5840409398078918, "learning_rate": 7.263333333333334e-05, "loss": 0.0258, "step": 2180 }, { "grad_norm": 0.6027020215988159, "learning_rate": 7.296666666666667e-05, "loss": 0.0262, "step": 2190 }, { "grad_norm": 0.6225112080574036, "learning_rate": 7.33e-05, "loss": 0.032, "step": 2200 }, { "grad_norm": 0.5141150951385498, "learning_rate": 7.363333333333334e-05, "loss": 0.0303, "step": 2210 }, { "grad_norm": 0.734550952911377, "learning_rate": 7.396666666666667e-05, "loss": 0.033, "step": 2220 }, { "grad_norm": 0.6319429874420166, "learning_rate": 7.43e-05, "loss": 0.0296, "step": 2230 }, { "grad_norm": 0.4286651313304901, "learning_rate": 7.463333333333334e-05, "loss": 0.0269, "step": 2240 }, { "grad_norm": 0.6045241355895996, "learning_rate": 7.496666666666667e-05, "loss": 0.0277, "step": 2250 }, { "grad_norm": 0.5824695825576782, "learning_rate": 7.53e-05, "loss": 0.0287, "step": 2260 }, { "grad_norm": 0.48695918917655945, "learning_rate": 7.563333333333333e-05, "loss": 0.0266, "step": 2270 }, { "grad_norm": 0.5881197452545166, "learning_rate": 7.596666666666668e-05, "loss": 0.0247, "step": 2280 }, { "grad_norm": 0.4567796587944031, "learning_rate": 7.630000000000001e-05, "loss": 0.0276, "step": 2290 }, { "grad_norm": 0.6635883450508118, "learning_rate": 7.663333333333333e-05, "loss": 0.0245, "step": 2300 }, { "grad_norm": 0.6479961276054382, "learning_rate": 7.696666666666668e-05, "loss": 0.0296, "step": 2310 }, { "grad_norm": 0.5224618911743164, "learning_rate": 7.730000000000001e-05, "loss": 0.0274, "step": 2320 }, { "grad_norm": 0.6456857919692993, "learning_rate": 7.763333333333334e-05, "loss": 0.031, "step": 2330 }, { "grad_norm": 0.6729128956794739, "learning_rate": 7.796666666666666e-05, "loss": 0.0279, "step": 2340 }, { "grad_norm": 0.6290175914764404, "learning_rate": 7.83e-05, "loss": 0.0357, "step": 2350 }, { "grad_norm": 0.6106721758842468, "learning_rate": 7.863333333333334e-05, "loss": 0.0288, "step": 2360 }, { "grad_norm": 0.6170771718025208, "learning_rate": 7.896666666666667e-05, "loss": 0.0273, "step": 2370 }, { "grad_norm": 0.6084820628166199, "learning_rate": 7.93e-05, "loss": 0.0267, "step": 2380 }, { "grad_norm": 0.6672490835189819, "learning_rate": 7.963333333333334e-05, "loss": 0.0309, "step": 2390 }, { "grad_norm": 0.4407326579093933, "learning_rate": 7.996666666666667e-05, "loss": 0.0306, "step": 2400 }, { "grad_norm": 0.6859383583068848, "learning_rate": 8.030000000000001e-05, "loss": 0.0257, "step": 2410 }, { "grad_norm": 0.6206133961677551, "learning_rate": 8.063333333333333e-05, "loss": 0.0327, "step": 2420 }, { "grad_norm": 0.509928822517395, "learning_rate": 8.096666666666667e-05, "loss": 0.0287, "step": 2430 }, { "grad_norm": 0.4681433439254761, "learning_rate": 8.13e-05, "loss": 0.0276, "step": 2440 }, { "grad_norm": 0.47213292121887207, "learning_rate": 8.163333333333334e-05, "loss": 0.027, "step": 2450 }, { "grad_norm": 0.6682668328285217, "learning_rate": 8.196666666666668e-05, "loss": 0.0249, "step": 2460 }, { "grad_norm": 0.5247741341590881, "learning_rate": 8.23e-05, "loss": 0.0246, "step": 2470 }, { "grad_norm": 0.45887258648872375, "learning_rate": 8.263333333333334e-05, "loss": 0.0235, "step": 2480 }, { "grad_norm": 0.5065684914588928, "learning_rate": 8.296666666666667e-05, "loss": 0.0303, "step": 2490 }, { "grad_norm": 0.6286656856536865, "learning_rate": 8.33e-05, "loss": 0.0256, "step": 2500 }, { "grad_norm": 0.6097934246063232, "learning_rate": 8.363333333333334e-05, "loss": 0.0274, "step": 2510 }, { "grad_norm": 0.5496993660926819, "learning_rate": 8.396666666666667e-05, "loss": 0.0268, "step": 2520 }, { "grad_norm": 0.5690876245498657, "learning_rate": 8.43e-05, "loss": 0.0281, "step": 2530 }, { "grad_norm": 0.4797924757003784, "learning_rate": 8.463333333333335e-05, "loss": 0.0252, "step": 2540 }, { "grad_norm": 0.46187642216682434, "learning_rate": 8.496666666666667e-05, "loss": 0.0268, "step": 2550 }, { "grad_norm": 0.517298698425293, "learning_rate": 8.53e-05, "loss": 0.028, "step": 2560 }, { "grad_norm": 0.41834643483161926, "learning_rate": 8.563333333333333e-05, "loss": 0.0252, "step": 2570 }, { "grad_norm": 0.5539282560348511, "learning_rate": 8.596666666666668e-05, "loss": 0.0253, "step": 2580 }, { "grad_norm": 0.6864331364631653, "learning_rate": 8.63e-05, "loss": 0.0246, "step": 2590 }, { "grad_norm": 0.5506659746170044, "learning_rate": 8.663333333333333e-05, "loss": 0.0278, "step": 2600 }, { "grad_norm": 0.49274247884750366, "learning_rate": 8.696666666666668e-05, "loss": 0.0273, "step": 2610 }, { "grad_norm": 0.3936481475830078, "learning_rate": 8.730000000000001e-05, "loss": 0.0237, "step": 2620 }, { "grad_norm": 0.4023718535900116, "learning_rate": 8.763333333333334e-05, "loss": 0.0239, "step": 2630 }, { "grad_norm": 0.39274585247039795, "learning_rate": 8.796666666666667e-05, "loss": 0.0247, "step": 2640 }, { "grad_norm": 0.5606030821800232, "learning_rate": 8.83e-05, "loss": 0.0265, "step": 2650 }, { "grad_norm": 0.5088376998901367, "learning_rate": 8.863333333333334e-05, "loss": 0.0244, "step": 2660 }, { "grad_norm": 0.5733055472373962, "learning_rate": 8.896666666666667e-05, "loss": 0.0241, "step": 2670 }, { "grad_norm": 0.5343175530433655, "learning_rate": 8.93e-05, "loss": 0.0275, "step": 2680 }, { "grad_norm": 0.5281231999397278, "learning_rate": 8.963333333333333e-05, "loss": 0.027, "step": 2690 }, { "grad_norm": 0.47041988372802734, "learning_rate": 8.996666666666667e-05, "loss": 0.0256, "step": 2700 }, { "grad_norm": 0.589595377445221, "learning_rate": 9.030000000000001e-05, "loss": 0.0244, "step": 2710 }, { "grad_norm": 0.5070868730545044, "learning_rate": 9.063333333333333e-05, "loss": 0.0247, "step": 2720 }, { "grad_norm": 0.5937284827232361, "learning_rate": 9.096666666666666e-05, "loss": 0.0278, "step": 2730 }, { "grad_norm": 0.6725428700447083, "learning_rate": 9.130000000000001e-05, "loss": 0.0257, "step": 2740 }, { "grad_norm": 0.6287254691123962, "learning_rate": 9.163333333333334e-05, "loss": 0.0244, "step": 2750 }, { "grad_norm": 0.49709782004356384, "learning_rate": 9.196666666666666e-05, "loss": 0.0232, "step": 2760 }, { "grad_norm": 0.44964903593063354, "learning_rate": 9.230000000000001e-05, "loss": 0.0265, "step": 2770 }, { "grad_norm": 0.4872162938117981, "learning_rate": 9.263333333333334e-05, "loss": 0.0229, "step": 2780 }, { "grad_norm": 0.5729479789733887, "learning_rate": 9.296666666666667e-05, "loss": 0.0243, "step": 2790 }, { "grad_norm": 0.42289113998413086, "learning_rate": 9.33e-05, "loss": 0.0255, "step": 2800 }, { "grad_norm": 0.4218228757381439, "learning_rate": 9.363333333333334e-05, "loss": 0.0253, "step": 2810 }, { "grad_norm": 0.48565465211868286, "learning_rate": 9.396666666666667e-05, "loss": 0.0268, "step": 2820 }, { "grad_norm": 0.4074060916900635, "learning_rate": 9.43e-05, "loss": 0.0252, "step": 2830 }, { "grad_norm": 0.5220145583152771, "learning_rate": 9.463333333333333e-05, "loss": 0.026, "step": 2840 }, { "grad_norm": 0.4929032325744629, "learning_rate": 9.496666666666667e-05, "loss": 0.0265, "step": 2850 }, { "grad_norm": 0.7733340263366699, "learning_rate": 9.53e-05, "loss": 0.028, "step": 2860 }, { "grad_norm": 0.6294170022010803, "learning_rate": 9.563333333333334e-05, "loss": 0.0267, "step": 2870 }, { "grad_norm": 0.43131470680236816, "learning_rate": 9.596666666666668e-05, "loss": 0.0237, "step": 2880 }, { "grad_norm": 0.35938671231269836, "learning_rate": 9.63e-05, "loss": 0.0247, "step": 2890 }, { "grad_norm": 0.441954642534256, "learning_rate": 9.663333333333334e-05, "loss": 0.0235, "step": 2900 }, { "grad_norm": 0.5554390549659729, "learning_rate": 9.696666666666667e-05, "loss": 0.0225, "step": 2910 }, { "grad_norm": 0.5399460792541504, "learning_rate": 9.730000000000001e-05, "loss": 0.0255, "step": 2920 }, { "grad_norm": 0.4398443400859833, "learning_rate": 9.763333333333334e-05, "loss": 0.0271, "step": 2930 }, { "grad_norm": 0.6268377900123596, "learning_rate": 9.796666666666667e-05, "loss": 0.0269, "step": 2940 }, { "grad_norm": 0.4373849630355835, "learning_rate": 9.83e-05, "loss": 0.0248, "step": 2950 }, { "grad_norm": 0.44073763489723206, "learning_rate": 9.863333333333334e-05, "loss": 0.0257, "step": 2960 }, { "grad_norm": 0.41025257110595703, "learning_rate": 9.896666666666667e-05, "loss": 0.0229, "step": 2970 }, { "grad_norm": 0.5418692231178284, "learning_rate": 9.93e-05, "loss": 0.0219, "step": 2980 }, { "grad_norm": 0.5484956502914429, "learning_rate": 9.963333333333333e-05, "loss": 0.0227, "step": 2990 }, { "grad_norm": 0.3590196669101715, "learning_rate": 9.996666666666668e-05, "loss": 0.0235, "step": 3000 }, { "grad_norm": 0.48214203119277954, "learning_rate": 9.999999384858465e-05, "loss": 0.0206, "step": 3010 }, { "grad_norm": 0.4876128137111664, "learning_rate": 9.999997258443473e-05, "loss": 0.0222, "step": 3020 }, { "grad_norm": 0.37169092893600464, "learning_rate": 9.999993613161331e-05, "loss": 0.0203, "step": 3030 }, { "grad_norm": 0.3859444558620453, "learning_rate": 9.999988449013146e-05, "loss": 0.0221, "step": 3040 }, { "grad_norm": 0.3617582619190216, "learning_rate": 9.99998176600049e-05, "loss": 0.0201, "step": 3050 }, { "grad_norm": 0.3808351159095764, "learning_rate": 9.999973564125389e-05, "loss": 0.0211, "step": 3060 }, { "grad_norm": 0.35162588953971863, "learning_rate": 9.999963843390335e-05, "loss": 0.0214, "step": 3070 }, { "grad_norm": 0.4366540014743805, "learning_rate": 9.999952603798282e-05, "loss": 0.0243, "step": 3080 }, { "grad_norm": 0.37506943941116333, "learning_rate": 9.999939845352646e-05, "loss": 0.0238, "step": 3090 }, { "grad_norm": 0.3120967149734497, "learning_rate": 9.999925568057298e-05, "loss": 0.0218, "step": 3100 }, { "grad_norm": 0.41679301857948303, "learning_rate": 9.999909771916578e-05, "loss": 0.0211, "step": 3110 }, { "grad_norm": 0.30199912190437317, "learning_rate": 9.999892456935285e-05, "loss": 0.0197, "step": 3120 }, { "grad_norm": 0.40556204319000244, "learning_rate": 9.999873623118679e-05, "loss": 0.0226, "step": 3130 }, { "grad_norm": 0.3389984369277954, "learning_rate": 9.999853270472479e-05, "loss": 0.0214, "step": 3140 }, { "grad_norm": 0.45245930552482605, "learning_rate": 9.999831399002871e-05, "loss": 0.0225, "step": 3150 }, { "grad_norm": 0.39743271470069885, "learning_rate": 9.999808008716494e-05, "loss": 0.0214, "step": 3160 }, { "grad_norm": 0.509181797504425, "learning_rate": 9.999783099620459e-05, "loss": 0.0201, "step": 3170 }, { "grad_norm": 0.33284056186676025, "learning_rate": 9.999756671722328e-05, "loss": 0.0189, "step": 3180 }, { "grad_norm": 0.36040326952934265, "learning_rate": 9.99972872503013e-05, "loss": 0.0207, "step": 3190 }, { "grad_norm": 0.3712449073791504, "learning_rate": 9.999699259552359e-05, "loss": 0.021, "step": 3200 }, { "grad_norm": 0.5042831897735596, "learning_rate": 9.99966827529796e-05, "loss": 0.0201, "step": 3210 }, { "grad_norm": 0.4797256290912628, "learning_rate": 9.999635772276348e-05, "loss": 0.0197, "step": 3220 }, { "grad_norm": 0.36157479882240295, "learning_rate": 9.999601750497396e-05, "loss": 0.0184, "step": 3230 }, { "grad_norm": 0.34613147377967834, "learning_rate": 9.99956620997144e-05, "loss": 0.0197, "step": 3240 }, { "grad_norm": 0.42923906445503235, "learning_rate": 9.999529150709275e-05, "loss": 0.0183, "step": 3250 }, { "grad_norm": 0.3785819709300995, "learning_rate": 9.999490572722158e-05, "loss": 0.0216, "step": 3260 }, { "grad_norm": 0.3847099244594574, "learning_rate": 9.99945047602181e-05, "loss": 0.0196, "step": 3270 }, { "grad_norm": 0.3623664677143097, "learning_rate": 9.99940886062041e-05, "loss": 0.0193, "step": 3280 }, { "grad_norm": 0.3881124258041382, "learning_rate": 9.999365726530599e-05, "loss": 0.0237, "step": 3290 }, { "grad_norm": 0.48550766706466675, "learning_rate": 9.999321073765481e-05, "loss": 0.0193, "step": 3300 }, { "grad_norm": 0.27767398953437805, "learning_rate": 9.99927490233862e-05, "loss": 0.0196, "step": 3310 }, { "grad_norm": 0.5458604097366333, "learning_rate": 9.999227212264043e-05, "loss": 0.0231, "step": 3320 }, { "grad_norm": 0.5505838990211487, "learning_rate": 9.999178003556236e-05, "loss": 0.0197, "step": 3330 }, { "grad_norm": 0.4155043363571167, "learning_rate": 9.999127276230146e-05, "loss": 0.0204, "step": 3340 }, { "grad_norm": 0.36562857031822205, "learning_rate": 9.999075030301184e-05, "loss": 0.0182, "step": 3350 }, { "grad_norm": 0.45053642988204956, "learning_rate": 9.999021265785221e-05, "loss": 0.0194, "step": 3360 }, { "grad_norm": 0.38892242312431335, "learning_rate": 9.998965982698589e-05, "loss": 0.0247, "step": 3370 }, { "grad_norm": 0.5315750241279602, "learning_rate": 9.998909181058082e-05, "loss": 0.0217, "step": 3380 }, { "grad_norm": 0.38353535532951355, "learning_rate": 9.998850860880953e-05, "loss": 0.0194, "step": 3390 }, { "grad_norm": 0.40457838773727417, "learning_rate": 9.998791022184922e-05, "loss": 0.0207, "step": 3400 }, { "grad_norm": 0.33679378032684326, "learning_rate": 9.99872966498816e-05, "loss": 0.0179, "step": 3410 }, { "grad_norm": 0.3411487936973572, "learning_rate": 9.998666789309313e-05, "loss": 0.0177, "step": 3420 }, { "grad_norm": 0.44982147216796875, "learning_rate": 9.998602395167475e-05, "loss": 0.0185, "step": 3430 }, { "grad_norm": 0.42044949531555176, "learning_rate": 9.998536482582213e-05, "loss": 0.0226, "step": 3440 }, { "grad_norm": 0.4197300672531128, "learning_rate": 9.998469051573544e-05, "loss": 0.0196, "step": 3450 }, { "grad_norm": 0.39289572834968567, "learning_rate": 9.998400102161954e-05, "loss": 0.0224, "step": 3460 }, { "grad_norm": 0.29184624552726746, "learning_rate": 9.998329634368388e-05, "loss": 0.017, "step": 3470 }, { "grad_norm": 0.35607680678367615, "learning_rate": 9.998257648214253e-05, "loss": 0.0203, "step": 3480 }, { "grad_norm": 0.34439459443092346, "learning_rate": 9.998184143721417e-05, "loss": 0.0177, "step": 3490 }, { "grad_norm": 0.3407992720603943, "learning_rate": 9.998109120912206e-05, "loss": 0.0182, "step": 3500 }, { "grad_norm": 0.2843284010887146, "learning_rate": 9.998032579809411e-05, "loss": 0.0168, "step": 3510 }, { "grad_norm": 0.40090107917785645, "learning_rate": 9.997954520436286e-05, "loss": 0.0198, "step": 3520 }, { "grad_norm": 0.43687450885772705, "learning_rate": 9.997874942816538e-05, "loss": 0.021, "step": 3530 }, { "grad_norm": 0.4504520893096924, "learning_rate": 9.997793846974345e-05, "loss": 0.0171, "step": 3540 }, { "grad_norm": 0.4449526071548462, "learning_rate": 9.997711232934341e-05, "loss": 0.0173, "step": 3550 }, { "grad_norm": 0.40955591201782227, "learning_rate": 9.99762710072162e-05, "loss": 0.019, "step": 3560 }, { "grad_norm": 0.2853061854839325, "learning_rate": 9.997541450361743e-05, "loss": 0.0185, "step": 3570 }, { "grad_norm": 0.3055476248264313, "learning_rate": 9.997454281880723e-05, "loss": 0.0202, "step": 3580 }, { "grad_norm": 0.3970089852809906, "learning_rate": 9.997365595305044e-05, "loss": 0.0186, "step": 3590 }, { "grad_norm": 0.4560907483100891, "learning_rate": 9.997275390661644e-05, "loss": 0.0201, "step": 3600 }, { "grad_norm": 0.3944167196750641, "learning_rate": 9.997183667977926e-05, "loss": 0.017, "step": 3610 }, { "grad_norm": 0.4034605026245117, "learning_rate": 9.997090427281752e-05, "loss": 0.0162, "step": 3620 }, { "grad_norm": 0.2912987172603607, "learning_rate": 9.996995668601448e-05, "loss": 0.0166, "step": 3630 }, { "grad_norm": 0.3224908411502838, "learning_rate": 9.996899391965798e-05, "loss": 0.016, "step": 3640 }, { "grad_norm": 0.3445907533168793, "learning_rate": 9.996801597404048e-05, "loss": 0.0167, "step": 3650 }, { "grad_norm": 0.33056876063346863, "learning_rate": 9.996702284945905e-05, "loss": 0.0188, "step": 3660 }, { "grad_norm": 0.35129314661026, "learning_rate": 9.996601454621539e-05, "loss": 0.0178, "step": 3670 }, { "grad_norm": 0.32188108563423157, "learning_rate": 9.996499106461577e-05, "loss": 0.0163, "step": 3680 }, { "grad_norm": 0.39511457085609436, "learning_rate": 9.996395240497112e-05, "loss": 0.0167, "step": 3690 }, { "grad_norm": 0.385811984539032, "learning_rate": 9.996289856759696e-05, "loss": 0.0182, "step": 3700 }, { "grad_norm": 0.3520319163799286, "learning_rate": 9.996182955281342e-05, "loss": 0.0183, "step": 3710 }, { "grad_norm": 0.32029423117637634, "learning_rate": 9.996074536094519e-05, "loss": 0.018, "step": 3720 }, { "grad_norm": 0.2860768139362335, "learning_rate": 9.995964599232168e-05, "loss": 0.0175, "step": 3730 }, { "grad_norm": 0.3764471709728241, "learning_rate": 9.995853144727683e-05, "loss": 0.0177, "step": 3740 }, { "grad_norm": 0.34356725215911865, "learning_rate": 9.99574017261492e-05, "loss": 0.0167, "step": 3750 }, { "grad_norm": 0.36886611580848694, "learning_rate": 9.995625682928198e-05, "loss": 0.017, "step": 3760 }, { "grad_norm": 0.2791632115840912, "learning_rate": 9.995509675702295e-05, "loss": 0.0168, "step": 3770 }, { "grad_norm": 0.24454541504383087, "learning_rate": 9.995392150972451e-05, "loss": 0.0152, "step": 3780 }, { "grad_norm": 0.3707715570926666, "learning_rate": 9.995273108774366e-05, "loss": 0.0166, "step": 3790 }, { "grad_norm": 0.3464537262916565, "learning_rate": 9.995152549144205e-05, "loss": 0.0154, "step": 3800 }, { "grad_norm": 0.20957663655281067, "learning_rate": 9.995030472118587e-05, "loss": 0.0127, "step": 3810 }, { "grad_norm": 0.2941529452800751, "learning_rate": 9.9949068777346e-05, "loss": 0.0143, "step": 3820 }, { "grad_norm": 0.3751513361930847, "learning_rate": 9.994781766029786e-05, "loss": 0.0163, "step": 3830 }, { "grad_norm": 0.3027159571647644, "learning_rate": 9.994655137042151e-05, "loss": 0.0153, "step": 3840 }, { "grad_norm": 0.2724834680557251, "learning_rate": 9.99452699081016e-05, "loss": 0.0156, "step": 3850 }, { "grad_norm": 0.24307288229465485, "learning_rate": 9.994397327372743e-05, "loss": 0.0152, "step": 3860 }, { "grad_norm": 0.26712462306022644, "learning_rate": 9.994266146769286e-05, "loss": 0.016, "step": 3870 }, { "grad_norm": 0.2850845456123352, "learning_rate": 9.994133449039642e-05, "loss": 0.0149, "step": 3880 }, { "grad_norm": 0.3900715410709381, "learning_rate": 9.993999234224118e-05, "loss": 0.0144, "step": 3890 }, { "grad_norm": 0.2791450023651123, "learning_rate": 9.993863502363485e-05, "loss": 0.0159, "step": 3900 }, { "grad_norm": 0.2847091853618622, "learning_rate": 9.993726253498976e-05, "loss": 0.017, "step": 3910 }, { "grad_norm": 0.32446303963661194, "learning_rate": 9.993587487672282e-05, "loss": 0.0185, "step": 3920 }, { "grad_norm": 0.23161080479621887, "learning_rate": 9.993447204925558e-05, "loss": 0.0148, "step": 3930 }, { "grad_norm": 0.3096104562282562, "learning_rate": 9.993305405301416e-05, "loss": 0.0154, "step": 3940 }, { "grad_norm": 0.3152487277984619, "learning_rate": 9.993162088842935e-05, "loss": 0.0143, "step": 3950 }, { "grad_norm": 0.3680046796798706, "learning_rate": 9.993017255593646e-05, "loss": 0.016, "step": 3960 }, { "grad_norm": 0.33741575479507446, "learning_rate": 9.992870905597548e-05, "loss": 0.0152, "step": 3970 }, { "grad_norm": 0.4866659641265869, "learning_rate": 9.9927230388991e-05, "loss": 0.0215, "step": 3980 }, { "grad_norm": 0.4477522671222687, "learning_rate": 9.992573655543215e-05, "loss": 0.0226, "step": 3990 }, { "grad_norm": 0.45175525546073914, "learning_rate": 9.992422755575277e-05, "loss": 0.0185, "step": 4000 }, { "grad_norm": 0.27719512581825256, "learning_rate": 9.992270339041123e-05, "loss": 0.0177, "step": 4010 }, { "grad_norm": 0.3620277941226959, "learning_rate": 9.992116405987053e-05, "loss": 0.0158, "step": 4020 }, { "grad_norm": 0.30308952927589417, "learning_rate": 9.991960956459828e-05, "loss": 0.0147, "step": 4030 }, { "grad_norm": 0.4228013753890991, "learning_rate": 9.991803990506669e-05, "loss": 0.0153, "step": 4040 }, { "grad_norm": 0.2697257697582245, "learning_rate": 9.991645508175258e-05, "loss": 0.0134, "step": 4050 }, { "grad_norm": 0.32327428460121155, "learning_rate": 9.99148550951374e-05, "loss": 0.0158, "step": 4060 }, { "grad_norm": 0.285074383020401, "learning_rate": 9.991323994570716e-05, "loss": 0.0143, "step": 4070 }, { "grad_norm": 0.19897910952568054, "learning_rate": 9.99116096339525e-05, "loss": 0.0143, "step": 4080 }, { "grad_norm": 0.3587413728237152, "learning_rate": 9.990996416036869e-05, "loss": 0.0155, "step": 4090 }, { "grad_norm": 0.3771476745605469, "learning_rate": 9.990830352545555e-05, "loss": 0.0148, "step": 4100 }, { "grad_norm": 0.3995601236820221, "learning_rate": 9.990662772971756e-05, "loss": 0.0165, "step": 4110 }, { "grad_norm": 0.2529391050338745, "learning_rate": 9.990493677366376e-05, "loss": 0.017, "step": 4120 }, { "grad_norm": 0.264478862285614, "learning_rate": 9.990323065780786e-05, "loss": 0.0144, "step": 4130 }, { "grad_norm": 0.3624909818172455, "learning_rate": 9.990150938266808e-05, "loss": 0.0188, "step": 4140 }, { "grad_norm": 0.3001759350299835, "learning_rate": 9.989977294876733e-05, "loss": 0.0149, "step": 4150 }, { "grad_norm": 0.2759300768375397, "learning_rate": 9.989802135663308e-05, "loss": 0.0171, "step": 4160 }, { "grad_norm": 0.24909447133541107, "learning_rate": 9.989625460679743e-05, "loss": 0.0148, "step": 4170 }, { "grad_norm": 0.25022226572036743, "learning_rate": 9.989447269979706e-05, "loss": 0.0146, "step": 4180 }, { "grad_norm": 0.26591670513153076, "learning_rate": 9.989267563617328e-05, "loss": 0.0125, "step": 4190 }, { "grad_norm": 0.3029699921607971, "learning_rate": 9.989086341647198e-05, "loss": 0.0137, "step": 4200 }, { "grad_norm": 0.3035736382007599, "learning_rate": 9.988903604124366e-05, "loss": 0.0141, "step": 4210 }, { "grad_norm": 0.28690508008003235, "learning_rate": 9.988719351104343e-05, "loss": 0.0135, "step": 4220 }, { "grad_norm": 0.32911068201065063, "learning_rate": 9.9885335826431e-05, "loss": 0.0125, "step": 4230 }, { "grad_norm": 0.36126384139060974, "learning_rate": 9.988346298797071e-05, "loss": 0.0169, "step": 4240 }, { "grad_norm": 0.3285928964614868, "learning_rate": 9.988157499623146e-05, "loss": 0.0157, "step": 4250 }, { "grad_norm": 0.29585930705070496, "learning_rate": 9.987967185178677e-05, "loss": 0.0153, "step": 4260 }, { "grad_norm": 0.3426804542541504, "learning_rate": 9.987775355521476e-05, "loss": 0.0128, "step": 4270 }, { "grad_norm": 0.2506636381149292, "learning_rate": 9.987582010709817e-05, "loss": 0.0138, "step": 4280 }, { "grad_norm": 0.22547529637813568, "learning_rate": 9.987387150802431e-05, "loss": 0.0137, "step": 4290 }, { "grad_norm": 0.31048324704170227, "learning_rate": 9.987190775858517e-05, "loss": 0.0132, "step": 4300 }, { "grad_norm": 0.2864062488079071, "learning_rate": 9.98699288593772e-05, "loss": 0.0142, "step": 4310 }, { "grad_norm": 0.27879855036735535, "learning_rate": 9.986793481100161e-05, "loss": 0.0127, "step": 4320 }, { "grad_norm": 0.32020094990730286, "learning_rate": 9.986592561406412e-05, "loss": 0.0141, "step": 4330 }, { "grad_norm": 0.2869685888290405, "learning_rate": 9.986390126917503e-05, "loss": 0.0143, "step": 4340 }, { "grad_norm": 0.2390623539686203, "learning_rate": 9.986186177694933e-05, "loss": 0.0133, "step": 4350 }, { "grad_norm": 0.2967565953731537, "learning_rate": 9.985980713800656e-05, "loss": 0.0128, "step": 4360 }, { "grad_norm": 0.28046825528144836, "learning_rate": 9.985773735297084e-05, "loss": 0.0133, "step": 4370 }, { "grad_norm": 0.3282446563243866, "learning_rate": 9.985565242247092e-05, "loss": 0.0145, "step": 4380 }, { "grad_norm": 0.2858050465583801, "learning_rate": 9.985355234714016e-05, "loss": 0.0149, "step": 4390 }, { "grad_norm": 0.28260135650634766, "learning_rate": 9.985143712761652e-05, "loss": 0.0148, "step": 4400 }, { "grad_norm": 0.3402651250362396, "learning_rate": 9.984930676454252e-05, "loss": 0.0174, "step": 4410 }, { "grad_norm": 0.3008532226085663, "learning_rate": 9.984716125856532e-05, "loss": 0.0135, "step": 4420 }, { "grad_norm": 0.27912649512290955, "learning_rate": 9.984500061033667e-05, "loss": 0.0155, "step": 4430 }, { "grad_norm": 0.2840447723865509, "learning_rate": 9.984282482051293e-05, "loss": 0.0146, "step": 4440 }, { "grad_norm": 0.3061172366142273, "learning_rate": 9.9840633889755e-05, "loss": 0.0152, "step": 4450 }, { "grad_norm": 0.2593478262424469, "learning_rate": 9.983842781872848e-05, "loss": 0.0133, "step": 4460 }, { "grad_norm": 0.2853364050388336, "learning_rate": 9.98362066081035e-05, "loss": 0.0135, "step": 4470 }, { "grad_norm": 0.3028251528739929, "learning_rate": 9.983397025855479e-05, "loss": 0.0155, "step": 4480 }, { "grad_norm": 0.2925330102443695, "learning_rate": 9.983171877076171e-05, "loss": 0.0157, "step": 4490 }, { "grad_norm": 0.3344006836414337, "learning_rate": 9.98294521454082e-05, "loss": 0.0165, "step": 4500 }, { "grad_norm": 0.33130329847335815, "learning_rate": 9.98271703831828e-05, "loss": 0.0136, "step": 4510 }, { "grad_norm": 0.2918162941932678, "learning_rate": 9.982487348477865e-05, "loss": 0.0125, "step": 4520 }, { "grad_norm": 0.210161030292511, "learning_rate": 9.982256145089347e-05, "loss": 0.0135, "step": 4530 }, { "grad_norm": 0.31741729378700256, "learning_rate": 9.982023428222962e-05, "loss": 0.0139, "step": 4540 }, { "grad_norm": 0.30784380435943604, "learning_rate": 9.981789197949403e-05, "loss": 0.0133, "step": 4550 }, { "grad_norm": 0.21585719287395477, "learning_rate": 9.98155345433982e-05, "loss": 0.0134, "step": 4560 }, { "grad_norm": 0.30597251653671265, "learning_rate": 9.981316197465831e-05, "loss": 0.0115, "step": 4570 }, { "grad_norm": 0.18654397130012512, "learning_rate": 9.981077427399504e-05, "loss": 0.0147, "step": 4580 }, { "grad_norm": 0.22842897474765778, "learning_rate": 9.980837144213371e-05, "loss": 0.0159, "step": 4590 }, { "grad_norm": 0.2603805661201477, "learning_rate": 9.980595347980426e-05, "loss": 0.0136, "step": 4600 }, { "grad_norm": 0.3307923376560211, "learning_rate": 9.980352038774119e-05, "loss": 0.0156, "step": 4610 }, { "grad_norm": 0.2921556532382965, "learning_rate": 9.98010721666836e-05, "loss": 0.0136, "step": 4620 }, { "grad_norm": 0.2946067154407501, "learning_rate": 9.979860881737523e-05, "loss": 0.0142, "step": 4630 }, { "grad_norm": 0.3355162739753723, "learning_rate": 9.979613034056434e-05, "loss": 0.0132, "step": 4640 }, { "grad_norm": 0.2597655653953552, "learning_rate": 9.979363673700386e-05, "loss": 0.0144, "step": 4650 }, { "grad_norm": 0.2773754596710205, "learning_rate": 9.979112800745124e-05, "loss": 0.0132, "step": 4660 }, { "grad_norm": 0.4004659950733185, "learning_rate": 9.978860415266861e-05, "loss": 0.0133, "step": 4670 }, { "grad_norm": 0.34206724166870117, "learning_rate": 9.978606517342262e-05, "loss": 0.0132, "step": 4680 }, { "grad_norm": 0.2684251070022583, "learning_rate": 9.978351107048456e-05, "loss": 0.0157, "step": 4690 }, { "grad_norm": 0.2419227808713913, "learning_rate": 9.978094184463029e-05, "loss": 0.0131, "step": 4700 }, { "grad_norm": 0.3140287399291992, "learning_rate": 9.977835749664029e-05, "loss": 0.0132, "step": 4710 }, { "grad_norm": 0.2691061496734619, "learning_rate": 9.97757580272996e-05, "loss": 0.0144, "step": 4720 }, { "grad_norm": 0.3429713547229767, "learning_rate": 9.977314343739786e-05, "loss": 0.0135, "step": 4730 }, { "grad_norm": 0.33147162199020386, "learning_rate": 9.977051372772934e-05, "loss": 0.0141, "step": 4740 }, { "grad_norm": 0.29810506105422974, "learning_rate": 9.976786889909286e-05, "loss": 0.0149, "step": 4750 }, { "grad_norm": 0.22380895912647247, "learning_rate": 9.976520895229185e-05, "loss": 0.0147, "step": 4760 }, { "grad_norm": 0.2809971272945404, "learning_rate": 9.976253388813433e-05, "loss": 0.0142, "step": 4770 }, { "grad_norm": 0.2659370005130768, "learning_rate": 9.975984370743293e-05, "loss": 0.0145, "step": 4780 }, { "grad_norm": 0.3158576190471649, "learning_rate": 9.975713841100485e-05, "loss": 0.0149, "step": 4790 }, { "grad_norm": 0.2980402112007141, "learning_rate": 9.975441799967187e-05, "loss": 0.0136, "step": 4800 }, { "grad_norm": 0.28695398569107056, "learning_rate": 9.975168247426039e-05, "loss": 0.0157, "step": 4810 }, { "grad_norm": 0.29537534713745117, "learning_rate": 9.974893183560139e-05, "loss": 0.015, "step": 4820 }, { "grad_norm": 0.32391494512557983, "learning_rate": 9.974616608453045e-05, "loss": 0.0156, "step": 4830 }, { "grad_norm": 0.30940622091293335, "learning_rate": 9.974338522188772e-05, "loss": 0.0142, "step": 4840 }, { "grad_norm": 0.21117357909679413, "learning_rate": 9.974058924851797e-05, "loss": 0.0114, "step": 4850 }, { "grad_norm": 0.23770377039909363, "learning_rate": 9.973777816527051e-05, "loss": 0.0139, "step": 4860 }, { "grad_norm": 0.23051239550113678, "learning_rate": 9.973495197299931e-05, "loss": 0.0134, "step": 4870 }, { "grad_norm": 0.26104339957237244, "learning_rate": 9.973211067256287e-05, "loss": 0.0121, "step": 4880 }, { "grad_norm": 0.26242774724960327, "learning_rate": 9.97292542648243e-05, "loss": 0.0116, "step": 4890 }, { "grad_norm": 0.23617152869701385, "learning_rate": 9.972638275065131e-05, "loss": 0.0144, "step": 4900 }, { "grad_norm": 0.2517353594303131, "learning_rate": 9.972349613091621e-05, "loss": 0.0136, "step": 4910 }, { "grad_norm": 0.27312812209129333, "learning_rate": 9.972059440649584e-05, "loss": 0.0165, "step": 4920 }, { "grad_norm": 0.2591722309589386, "learning_rate": 9.971767757827168e-05, "loss": 0.0147, "step": 4930 }, { "grad_norm": 0.21888680756092072, "learning_rate": 9.971474564712982e-05, "loss": 0.0134, "step": 4940 }, { "grad_norm": 0.3195919394493103, "learning_rate": 9.971179861396084e-05, "loss": 0.014, "step": 4950 }, { "grad_norm": 0.28815561532974243, "learning_rate": 9.970883647966003e-05, "loss": 0.0149, "step": 4960 }, { "grad_norm": 0.209254190325737, "learning_rate": 9.970585924512717e-05, "loss": 0.0128, "step": 4970 }, { "grad_norm": 0.28662344813346863, "learning_rate": 9.970286691126669e-05, "loss": 0.0166, "step": 4980 }, { "grad_norm": 0.3108627498149872, "learning_rate": 9.969985947898756e-05, "loss": 0.0136, "step": 4990 }, { "grad_norm": 0.22628054022789001, "learning_rate": 9.969683694920337e-05, "loss": 0.0149, "step": 5000 }, { "grad_norm": 0.26655545830726624, "learning_rate": 9.969379932283228e-05, "loss": 0.0143, "step": 5010 }, { "grad_norm": 0.24452893435955048, "learning_rate": 9.969074660079704e-05, "loss": 0.0144, "step": 5020 }, { "grad_norm": 0.3246316611766815, "learning_rate": 9.968767878402501e-05, "loss": 0.0131, "step": 5030 }, { "grad_norm": 0.2110854685306549, "learning_rate": 9.968459587344808e-05, "loss": 0.0138, "step": 5040 }, { "grad_norm": 0.24477435648441315, "learning_rate": 9.968149787000278e-05, "loss": 0.0125, "step": 5050 }, { "grad_norm": 0.260122686624527, "learning_rate": 9.967838477463018e-05, "loss": 0.0122, "step": 5060 }, { "grad_norm": 0.33123305439949036, "learning_rate": 9.967525658827597e-05, "loss": 0.0109, "step": 5070 }, { "grad_norm": 0.2738533914089203, "learning_rate": 9.967211331189042e-05, "loss": 0.0108, "step": 5080 }, { "grad_norm": 0.24877673387527466, "learning_rate": 9.966895494642834e-05, "loss": 0.0122, "step": 5090 }, { "grad_norm": 0.31839969754219055, "learning_rate": 9.96657814928492e-05, "loss": 0.013, "step": 5100 }, { "grad_norm": 0.321466326713562, "learning_rate": 9.966259295211697e-05, "loss": 0.0144, "step": 5110 }, { "grad_norm": 0.2485794723033905, "learning_rate": 9.965938932520028e-05, "loss": 0.0136, "step": 5120 }, { "grad_norm": 0.24549199640750885, "learning_rate": 9.965617061307229e-05, "loss": 0.0124, "step": 5130 }, { "grad_norm": 0.30820566415786743, "learning_rate": 9.965293681671077e-05, "loss": 0.0132, "step": 5140 }, { "grad_norm": 0.22428330779075623, "learning_rate": 9.964968793709804e-05, "loss": 0.0128, "step": 5150 }, { "grad_norm": 0.3140605390071869, "learning_rate": 9.964642397522106e-05, "loss": 0.0138, "step": 5160 }, { "grad_norm": 0.2792457640171051, "learning_rate": 9.96431449320713e-05, "loss": 0.0123, "step": 5170 }, { "grad_norm": 0.25067687034606934, "learning_rate": 9.963985080864486e-05, "loss": 0.0103, "step": 5180 }, { "grad_norm": 0.2771768867969513, "learning_rate": 9.96365416059424e-05, "loss": 0.0117, "step": 5190 }, { "grad_norm": 0.27162232995033264, "learning_rate": 9.963321732496919e-05, "loss": 0.0114, "step": 5200 }, { "grad_norm": 0.28037792444229126, "learning_rate": 9.962987796673506e-05, "loss": 0.0126, "step": 5210 }, { "grad_norm": 0.27267295122146606, "learning_rate": 9.962652353225438e-05, "loss": 0.0128, "step": 5220 }, { "grad_norm": 0.23385070264339447, "learning_rate": 9.962315402254619e-05, "loss": 0.0122, "step": 5230 }, { "grad_norm": 0.2320750653743744, "learning_rate": 9.9619769438634e-05, "loss": 0.0128, "step": 5240 }, { "grad_norm": 0.2866482138633728, "learning_rate": 9.9616369781546e-05, "loss": 0.0115, "step": 5250 }, { "grad_norm": 0.28121739625930786, "learning_rate": 9.961295505231491e-05, "loss": 0.0149, "step": 5260 }, { "grad_norm": 0.2835405766963959, "learning_rate": 9.960952525197804e-05, "loss": 0.0128, "step": 5270 }, { "grad_norm": 0.24188794195652008, "learning_rate": 9.960608038157724e-05, "loss": 0.0115, "step": 5280 }, { "grad_norm": 0.3200702965259552, "learning_rate": 9.960262044215901e-05, "loss": 0.0146, "step": 5290 }, { "grad_norm": 0.30075931549072266, "learning_rate": 9.959914543477435e-05, "loss": 0.0147, "step": 5300 }, { "grad_norm": 0.2914126515388489, "learning_rate": 9.959565536047892e-05, "loss": 0.0132, "step": 5310 }, { "grad_norm": 0.1935928761959076, "learning_rate": 9.959215022033288e-05, "loss": 0.0127, "step": 5320 }, { "grad_norm": 0.2510479986667633, "learning_rate": 9.9588630015401e-05, "loss": 0.0125, "step": 5330 }, { "grad_norm": 0.2936236560344696, "learning_rate": 9.958509474675264e-05, "loss": 0.0126, "step": 5340 }, { "grad_norm": 0.2907848656177521, "learning_rate": 9.958154441546171e-05, "loss": 0.0115, "step": 5350 }, { "grad_norm": 0.2776922881603241, "learning_rate": 9.957797902260673e-05, "loss": 0.0129, "step": 5360 }, { "grad_norm": 0.2504800260066986, "learning_rate": 9.957439856927073e-05, "loss": 0.0131, "step": 5370 }, { "grad_norm": 0.2443743497133255, "learning_rate": 9.957080305654139e-05, "loss": 0.0124, "step": 5380 }, { "grad_norm": 0.338456928730011, "learning_rate": 9.956719248551092e-05, "loss": 0.0117, "step": 5390 }, { "grad_norm": 0.30506211519241333, "learning_rate": 9.956356685727612e-05, "loss": 0.0133, "step": 5400 }, { "grad_norm": 0.21705614030361176, "learning_rate": 9.955992617293836e-05, "loss": 0.0129, "step": 5410 }, { "grad_norm": 0.23510167002677917, "learning_rate": 9.955627043360358e-05, "loss": 0.0122, "step": 5420 }, { "grad_norm": 0.19667409360408783, "learning_rate": 9.955259964038231e-05, "loss": 0.0109, "step": 5430 }, { "grad_norm": 0.18108171224594116, "learning_rate": 9.954891379438962e-05, "loss": 0.0104, "step": 5440 }, { "grad_norm": 0.26140815019607544, "learning_rate": 9.954521289674519e-05, "loss": 0.0102, "step": 5450 }, { "grad_norm": 0.2092543989419937, "learning_rate": 9.954149694857325e-05, "loss": 0.0099, "step": 5460 }, { "grad_norm": 0.24797961115837097, "learning_rate": 9.953776595100258e-05, "loss": 0.0111, "step": 5470 }, { "grad_norm": 0.2718297839164734, "learning_rate": 9.95340199051666e-05, "loss": 0.0104, "step": 5480 }, { "grad_norm": 0.26539793610572815, "learning_rate": 9.953025881220325e-05, "loss": 0.0104, "step": 5490 }, { "grad_norm": 0.24245664477348328, "learning_rate": 9.952648267325504e-05, "loss": 0.0119, "step": 5500 }, { "grad_norm": 0.24385614693164825, "learning_rate": 9.952269148946905e-05, "loss": 0.0096, "step": 5510 }, { "grad_norm": 0.2783612906932831, "learning_rate": 9.951888526199697e-05, "loss": 0.0109, "step": 5520 }, { "grad_norm": 0.34346041083335876, "learning_rate": 9.951506399199501e-05, "loss": 0.0099, "step": 5530 }, { "grad_norm": 0.22518707811832428, "learning_rate": 9.951122768062399e-05, "loss": 0.0114, "step": 5540 }, { "grad_norm": 0.19473181664943695, "learning_rate": 9.950737632904927e-05, "loss": 0.0107, "step": 5550 }, { "grad_norm": 0.23703651130199432, "learning_rate": 9.950350993844077e-05, "loss": 0.0119, "step": 5560 }, { "grad_norm": 0.2674230933189392, "learning_rate": 9.949962850997303e-05, "loss": 0.0118, "step": 5570 }, { "grad_norm": 0.25006958842277527, "learning_rate": 9.949573204482512e-05, "loss": 0.0139, "step": 5580 }, { "grad_norm": 0.23594912886619568, "learning_rate": 9.949182054418064e-05, "loss": 0.0109, "step": 5590 }, { "grad_norm": 0.28366562724113464, "learning_rate": 9.948789400922787e-05, "loss": 0.0152, "step": 5600 }, { "grad_norm": 0.36642542481422424, "learning_rate": 9.948395244115953e-05, "loss": 0.0144, "step": 5610 }, { "grad_norm": 0.18197500705718994, "learning_rate": 9.9479995841173e-05, "loss": 0.0134, "step": 5620 }, { "grad_norm": 0.20052175223827362, "learning_rate": 9.947602421047017e-05, "loss": 0.0133, "step": 5630 }, { "grad_norm": 0.24195295572280884, "learning_rate": 9.947203755025753e-05, "loss": 0.0104, "step": 5640 }, { "grad_norm": 0.21863313019275665, "learning_rate": 9.946803586174611e-05, "loss": 0.0112, "step": 5650 }, { "grad_norm": 0.25542113184928894, "learning_rate": 9.946401914615151e-05, "loss": 0.0112, "step": 5660 }, { "grad_norm": 0.2154194712638855, "learning_rate": 9.945998740469394e-05, "loss": 0.0138, "step": 5670 }, { "grad_norm": 0.27246782183647156, "learning_rate": 9.945594063859809e-05, "loss": 0.0112, "step": 5680 }, { "grad_norm": 0.22208519279956818, "learning_rate": 9.94518788490933e-05, "loss": 0.0126, "step": 5690 }, { "grad_norm": 0.22201403975486755, "learning_rate": 9.944780203741341e-05, "loss": 0.0096, "step": 5700 }, { "grad_norm": 0.2565869092941284, "learning_rate": 9.944371020479686e-05, "loss": 0.0126, "step": 5710 }, { "grad_norm": 0.20837312936782837, "learning_rate": 9.943960335248662e-05, "loss": 0.012, "step": 5720 }, { "grad_norm": 0.27542516589164734, "learning_rate": 9.943548148173027e-05, "loss": 0.0134, "step": 5730 }, { "grad_norm": 0.24376772344112396, "learning_rate": 9.943134459377992e-05, "loss": 0.0117, "step": 5740 }, { "grad_norm": 0.2010304182767868, "learning_rate": 9.942719268989222e-05, "loss": 0.0116, "step": 5750 }, { "grad_norm": 0.20107460021972656, "learning_rate": 9.942302577132844e-05, "loss": 0.012, "step": 5760 }, { "grad_norm": 0.24294404685497284, "learning_rate": 9.941884383935438e-05, "loss": 0.0111, "step": 5770 }, { "grad_norm": 0.22166921198368073, "learning_rate": 9.941464689524039e-05, "loss": 0.0115, "step": 5780 }, { "grad_norm": 0.19090259075164795, "learning_rate": 9.941043494026139e-05, "loss": 0.011, "step": 5790 }, { "grad_norm": 0.3306300938129425, "learning_rate": 9.940620797569685e-05, "loss": 0.0123, "step": 5800 }, { "grad_norm": 0.2685251235961914, "learning_rate": 9.940196600283082e-05, "loss": 0.0119, "step": 5810 }, { "grad_norm": 0.2931045889854431, "learning_rate": 9.939770902295192e-05, "loss": 0.0126, "step": 5820 }, { "grad_norm": 0.28906041383743286, "learning_rate": 9.939343703735329e-05, "loss": 0.011, "step": 5830 }, { "grad_norm": 0.23148563504219055, "learning_rate": 9.938915004733264e-05, "loss": 0.0114, "step": 5840 }, { "grad_norm": 0.25450730323791504, "learning_rate": 9.938484805419224e-05, "loss": 0.0115, "step": 5850 }, { "grad_norm": 0.18132035434246063, "learning_rate": 9.938053105923894e-05, "loss": 0.0118, "step": 5860 }, { "grad_norm": 0.2387465238571167, "learning_rate": 9.937619906378413e-05, "loss": 0.0114, "step": 5870 }, { "grad_norm": 0.2718299925327301, "learning_rate": 9.937185206914374e-05, "loss": 0.0108, "step": 5880 }, { "grad_norm": 0.1752624213695526, "learning_rate": 9.936749007663829e-05, "loss": 0.0087, "step": 5890 }, { "grad_norm": 0.2674594521522522, "learning_rate": 9.93631130875928e-05, "loss": 0.0111, "step": 5900 }, { "grad_norm": 0.29471874237060547, "learning_rate": 9.935872110333692e-05, "loss": 0.0115, "step": 5910 }, { "grad_norm": 0.26760125160217285, "learning_rate": 9.935431412520484e-05, "loss": 0.011, "step": 5920 }, { "grad_norm": 0.2951185703277588, "learning_rate": 9.934989215453523e-05, "loss": 0.0136, "step": 5930 }, { "grad_norm": 0.29641392827033997, "learning_rate": 9.934545519267139e-05, "loss": 0.013, "step": 5940 }, { "grad_norm": 0.18544775247573853, "learning_rate": 9.934100324096117e-05, "loss": 0.0136, "step": 5950 }, { "grad_norm": 0.2672192454338074, "learning_rate": 9.933653630075692e-05, "loss": 0.0108, "step": 5960 }, { "grad_norm": 0.22066156566143036, "learning_rate": 9.93320543734156e-05, "loss": 0.0112, "step": 5970 }, { "grad_norm": 0.21750377118587494, "learning_rate": 9.932755746029871e-05, "loss": 0.0105, "step": 5980 }, { "grad_norm": 0.2042769342660904, "learning_rate": 9.932304556277228e-05, "loss": 0.0119, "step": 5990 }, { "grad_norm": 0.2870963513851166, "learning_rate": 9.93185186822069e-05, "loss": 0.011, "step": 6000 }, { "grad_norm": 0.21059495210647583, "learning_rate": 9.931397681997773e-05, "loss": 0.0136, "step": 6010 }, { "grad_norm": 0.23571772873401642, "learning_rate": 9.930941997746446e-05, "loss": 0.0117, "step": 6020 }, { "grad_norm": 0.2426503747701645, "learning_rate": 9.930484815605134e-05, "loss": 0.0123, "step": 6030 }, { "grad_norm": 0.21166139841079712, "learning_rate": 9.930026135712717e-05, "loss": 0.0101, "step": 6040 }, { "grad_norm": 0.314642071723938, "learning_rate": 9.92956595820853e-05, "loss": 0.0116, "step": 6050 }, { "grad_norm": 0.1690446138381958, "learning_rate": 9.929104283232362e-05, "loss": 0.0107, "step": 6060 }, { "grad_norm": 0.2278701812028885, "learning_rate": 9.92864111092446e-05, "loss": 0.0097, "step": 6070 }, { "grad_norm": 0.24926669895648956, "learning_rate": 9.92817644142552e-05, "loss": 0.0135, "step": 6080 }, { "grad_norm": 0.15165188908576965, "learning_rate": 9.927710274876698e-05, "loss": 0.0115, "step": 6090 }, { "grad_norm": 0.19430473446846008, "learning_rate": 9.927242611419603e-05, "loss": 0.0108, "step": 6100 }, { "grad_norm": 0.2211086004972458, "learning_rate": 9.926773451196301e-05, "loss": 0.0108, "step": 6110 }, { "grad_norm": 0.25848832726478577, "learning_rate": 9.926302794349306e-05, "loss": 0.012, "step": 6120 }, { "grad_norm": 0.22855228185653687, "learning_rate": 9.925830641021594e-05, "loss": 0.0112, "step": 6130 }, { "grad_norm": 0.17797555029392242, "learning_rate": 9.925356991356593e-05, "loss": 0.0122, "step": 6140 }, { "grad_norm": 0.27426043152809143, "learning_rate": 9.924881845498184e-05, "loss": 0.0099, "step": 6150 }, { "grad_norm": 0.2742624878883362, "learning_rate": 9.924405203590705e-05, "loss": 0.0106, "step": 6160 }, { "grad_norm": 0.19200335443019867, "learning_rate": 9.923927065778946e-05, "loss": 0.0099, "step": 6170 }, { "grad_norm": 0.22418011724948883, "learning_rate": 9.923447432208154e-05, "loss": 0.0095, "step": 6180 }, { "grad_norm": 0.3045748174190521, "learning_rate": 9.922966303024027e-05, "loss": 0.011, "step": 6190 }, { "grad_norm": 0.34878459572792053, "learning_rate": 9.922483678372721e-05, "loss": 0.0117, "step": 6200 }, { "grad_norm": 0.21053622663021088, "learning_rate": 9.921999558400845e-05, "loss": 0.0117, "step": 6210 }, { "grad_norm": 0.20968590676784515, "learning_rate": 9.92151394325546e-05, "loss": 0.0106, "step": 6220 }, { "grad_norm": 0.23756404221057892, "learning_rate": 9.921026833084084e-05, "loss": 0.01, "step": 6230 }, { "grad_norm": 0.18882854282855988, "learning_rate": 9.920538228034689e-05, "loss": 0.0104, "step": 6240 }, { "grad_norm": 0.2536315619945526, "learning_rate": 9.920048128255699e-05, "loss": 0.0104, "step": 6250 }, { "grad_norm": 0.21862474083900452, "learning_rate": 9.919556533895995e-05, "loss": 0.0091, "step": 6260 }, { "grad_norm": 0.19146029651165009, "learning_rate": 9.919063445104907e-05, "loss": 0.0083, "step": 6270 }, { "grad_norm": 0.1413741111755371, "learning_rate": 9.918568862032227e-05, "loss": 0.0093, "step": 6280 }, { "grad_norm": 0.33060985803604126, "learning_rate": 9.918072784828194e-05, "loss": 0.0104, "step": 6290 }, { "grad_norm": 0.23131898045539856, "learning_rate": 9.917575213643501e-05, "loss": 0.0098, "step": 6300 }, { "grad_norm": 0.21285927295684814, "learning_rate": 9.917076148629302e-05, "loss": 0.0101, "step": 6310 }, { "grad_norm": 0.2728671431541443, "learning_rate": 9.916575589937196e-05, "loss": 0.0106, "step": 6320 }, { "grad_norm": 0.20848846435546875, "learning_rate": 9.916073537719239e-05, "loss": 0.0109, "step": 6330 }, { "grad_norm": 0.20430812239646912, "learning_rate": 9.915569992127944e-05, "loss": 0.0121, "step": 6340 }, { "grad_norm": 0.3142538070678711, "learning_rate": 9.915064953316273e-05, "loss": 0.0146, "step": 6350 }, { "grad_norm": 0.22393396496772766, "learning_rate": 9.914558421437645e-05, "loss": 0.0117, "step": 6360 }, { "grad_norm": 0.25079086422920227, "learning_rate": 9.914050396645929e-05, "loss": 0.0117, "step": 6370 }, { "grad_norm": 0.2913857698440552, "learning_rate": 9.913540879095452e-05, "loss": 0.0107, "step": 6380 }, { "grad_norm": 0.24844977259635925, "learning_rate": 9.913029868940987e-05, "loss": 0.0122, "step": 6390 }, { "grad_norm": 0.26215091347694397, "learning_rate": 9.912517366337772e-05, "loss": 0.0141, "step": 6400 }, { "grad_norm": 0.2342989146709442, "learning_rate": 9.912003371441487e-05, "loss": 0.0127, "step": 6410 }, { "grad_norm": 0.3459029197692871, "learning_rate": 9.911487884408271e-05, "loss": 0.012, "step": 6420 }, { "grad_norm": 0.21722283959388733, "learning_rate": 9.910970905394719e-05, "loss": 0.0099, "step": 6430 }, { "grad_norm": 0.25026699900627136, "learning_rate": 9.91045243455787e-05, "loss": 0.0104, "step": 6440 }, { "grad_norm": 0.17307688295841217, "learning_rate": 9.909932472055225e-05, "loss": 0.0088, "step": 6450 }, { "grad_norm": 0.22708164155483246, "learning_rate": 9.909411018044734e-05, "loss": 0.0101, "step": 6460 }, { "grad_norm": 0.2021331936120987, "learning_rate": 9.908888072684802e-05, "loss": 0.0089, "step": 6470 }, { "grad_norm": 0.369196355342865, "learning_rate": 9.908363636134285e-05, "loss": 0.0104, "step": 6480 }, { "grad_norm": 0.22383593022823334, "learning_rate": 9.907837708552493e-05, "loss": 0.0133, "step": 6490 }, { "grad_norm": 0.2414979785680771, "learning_rate": 9.90731029009919e-05, "loss": 0.0121, "step": 6500 }, { "grad_norm": 0.16609683632850647, "learning_rate": 9.906781380934589e-05, "loss": 0.0131, "step": 6510 }, { "grad_norm": 0.274486243724823, "learning_rate": 9.906250981219362e-05, "loss": 0.0118, "step": 6520 }, { "grad_norm": 0.21607716381549835, "learning_rate": 9.905719091114628e-05, "loss": 0.0116, "step": 6530 }, { "grad_norm": 0.32482439279556274, "learning_rate": 9.905185710781964e-05, "loss": 0.0093, "step": 6540 }, { "grad_norm": 0.2949191629886627, "learning_rate": 9.904650840383392e-05, "loss": 0.0122, "step": 6550 }, { "grad_norm": 0.18507976830005646, "learning_rate": 9.904114480081397e-05, "loss": 0.0113, "step": 6560 }, { "grad_norm": 0.18176154792308807, "learning_rate": 9.903576630038906e-05, "loss": 0.0112, "step": 6570 }, { "grad_norm": 0.1729423552751541, "learning_rate": 9.903037290419309e-05, "loss": 0.0088, "step": 6580 }, { "grad_norm": 0.1651846170425415, "learning_rate": 9.902496461386439e-05, "loss": 0.0103, "step": 6590 }, { "grad_norm": 0.27578118443489075, "learning_rate": 9.901954143104588e-05, "loss": 0.0109, "step": 6600 }, { "grad_norm": 0.23941604793071747, "learning_rate": 9.901410335738496e-05, "loss": 0.0101, "step": 6610 }, { "grad_norm": 0.18468548357486725, "learning_rate": 9.900865039453358e-05, "loss": 0.0128, "step": 6620 }, { "grad_norm": 0.20097686350345612, "learning_rate": 9.900318254414821e-05, "loss": 0.014, "step": 6630 }, { "grad_norm": 0.289951354265213, "learning_rate": 9.899769980788985e-05, "loss": 0.0134, "step": 6640 }, { "grad_norm": 0.2769491374492645, "learning_rate": 9.899220218742398e-05, "loss": 0.0129, "step": 6650 }, { "grad_norm": 0.23806092143058777, "learning_rate": 9.898668968442066e-05, "loss": 0.0101, "step": 6660 }, { "grad_norm": 0.20804210007190704, "learning_rate": 9.898116230055443e-05, "loss": 0.0114, "step": 6670 }, { "grad_norm": 0.1935357302427292, "learning_rate": 9.897562003750437e-05, "loss": 0.0103, "step": 6680 }, { "grad_norm": 0.18162119388580322, "learning_rate": 9.897006289695407e-05, "loss": 0.0096, "step": 6690 }, { "grad_norm": 0.19969996809959412, "learning_rate": 9.896449088059164e-05, "loss": 0.01, "step": 6700 }, { "grad_norm": 0.17136332392692566, "learning_rate": 9.89589039901097e-05, "loss": 0.0101, "step": 6710 }, { "grad_norm": 0.2514260411262512, "learning_rate": 9.895330222720542e-05, "loss": 0.0133, "step": 6720 }, { "grad_norm": 0.16584300994873047, "learning_rate": 9.894768559358047e-05, "loss": 0.0129, "step": 6730 }, { "grad_norm": 0.17270813882350922, "learning_rate": 9.894205409094101e-05, "loss": 0.0119, "step": 6740 }, { "grad_norm": 0.24530959129333496, "learning_rate": 9.893640772099777e-05, "loss": 0.0157, "step": 6750 }, { "grad_norm": 0.22532416880130768, "learning_rate": 9.893074648546595e-05, "loss": 0.0133, "step": 6760 }, { "grad_norm": 0.28840944170951843, "learning_rate": 9.892507038606528e-05, "loss": 0.0125, "step": 6770 }, { "grad_norm": 0.2525266408920288, "learning_rate": 9.891937942452003e-05, "loss": 0.0125, "step": 6780 }, { "grad_norm": 0.2468404620885849, "learning_rate": 9.891367360255895e-05, "loss": 0.0113, "step": 6790 }, { "grad_norm": 0.21942612528800964, "learning_rate": 9.890795292191532e-05, "loss": 0.0124, "step": 6800 }, { "grad_norm": 0.17533065378665924, "learning_rate": 9.890221738432694e-05, "loss": 0.012, "step": 6810 }, { "grad_norm": 0.23471517860889435, "learning_rate": 9.88964669915361e-05, "loss": 0.0115, "step": 6820 }, { "grad_norm": 0.18242962658405304, "learning_rate": 9.889070174528963e-05, "loss": 0.0109, "step": 6830 }, { "grad_norm": 0.22787442803382874, "learning_rate": 9.888492164733883e-05, "loss": 0.0112, "step": 6840 }, { "grad_norm": 0.22420717775821686, "learning_rate": 9.88791266994396e-05, "loss": 0.01, "step": 6850 }, { "grad_norm": 0.17287962138652802, "learning_rate": 9.887331690335223e-05, "loss": 0.0099, "step": 6860 }, { "grad_norm": 0.1824754923582077, "learning_rate": 9.886749226084163e-05, "loss": 0.0098, "step": 6870 }, { "grad_norm": 0.21454459428787231, "learning_rate": 9.886165277367714e-05, "loss": 0.0112, "step": 6880 }, { "grad_norm": 0.20268651843070984, "learning_rate": 9.885579844363265e-05, "loss": 0.0088, "step": 6890 }, { "grad_norm": 0.1923826038837433, "learning_rate": 9.884992927248656e-05, "loss": 0.0106, "step": 6900 }, { "grad_norm": 0.21711242198944092, "learning_rate": 9.884404526202178e-05, "loss": 0.0115, "step": 6910 }, { "grad_norm": 0.18563523888587952, "learning_rate": 9.883814641402568e-05, "loss": 0.0105, "step": 6920 }, { "grad_norm": 0.28178778290748596, "learning_rate": 9.88322327302902e-05, "loss": 0.011, "step": 6930 }, { "grad_norm": 0.23177187144756317, "learning_rate": 9.882630421261176e-05, "loss": 0.01, "step": 6940 }, { "grad_norm": 0.2700766921043396, "learning_rate": 9.88203608627913e-05, "loss": 0.012, "step": 6950 }, { "grad_norm": 0.229451522231102, "learning_rate": 9.881440268263422e-05, "loss": 0.0102, "step": 6960 }, { "grad_norm": 0.14301323890686035, "learning_rate": 9.880842967395048e-05, "loss": 0.0101, "step": 6970 }, { "grad_norm": 0.21275408565998077, "learning_rate": 9.880244183855452e-05, "loss": 0.0108, "step": 6980 }, { "grad_norm": 0.21635325253009796, "learning_rate": 9.879643917826527e-05, "loss": 0.0096, "step": 6990 }, { "grad_norm": 0.27748411893844604, "learning_rate": 9.87904216949062e-05, "loss": 0.0103, "step": 7000 }, { "grad_norm": 0.2489020675420761, "learning_rate": 9.878438939030526e-05, "loss": 0.0107, "step": 7010 }, { "grad_norm": 0.2755835950374603, "learning_rate": 9.877834226629489e-05, "loss": 0.0095, "step": 7020 }, { "grad_norm": 0.14183087646961212, "learning_rate": 9.877228032471206e-05, "loss": 0.0086, "step": 7030 }, { "grad_norm": 0.1850457638502121, "learning_rate": 9.876620356739823e-05, "loss": 0.0087, "step": 7040 }, { "grad_norm": 0.22495532035827637, "learning_rate": 9.876011199619935e-05, "loss": 0.009, "step": 7050 }, { "grad_norm": 0.2325436770915985, "learning_rate": 9.875400561296589e-05, "loss": 0.0081, "step": 7060 }, { "grad_norm": 0.21377263963222504, "learning_rate": 9.874788441955278e-05, "loss": 0.013, "step": 7070 }, { "grad_norm": 0.2864483594894409, "learning_rate": 9.874174841781951e-05, "loss": 0.0119, "step": 7080 }, { "grad_norm": 0.22209839522838593, "learning_rate": 9.873559760963003e-05, "loss": 0.0128, "step": 7090 }, { "grad_norm": 0.1733119785785675, "learning_rate": 9.872943199685278e-05, "loss": 0.0099, "step": 7100 }, { "grad_norm": 0.17068012058734894, "learning_rate": 9.872325158136071e-05, "loss": 0.0111, "step": 7110 }, { "grad_norm": 0.1433611363172531, "learning_rate": 9.871705636503128e-05, "loss": 0.0097, "step": 7120 }, { "grad_norm": 0.24515320360660553, "learning_rate": 9.871084634974641e-05, "loss": 0.0104, "step": 7130 }, { "grad_norm": 0.16447694599628448, "learning_rate": 9.870462153739257e-05, "loss": 0.0097, "step": 7140 }, { "grad_norm": 0.19026611745357513, "learning_rate": 9.869838192986067e-05, "loss": 0.0093, "step": 7150 }, { "grad_norm": 0.239889994263649, "learning_rate": 9.869212752904616e-05, "loss": 0.0108, "step": 7160 }, { "grad_norm": 0.2037869542837143, "learning_rate": 9.868585833684894e-05, "loss": 0.0114, "step": 7170 }, { "grad_norm": 0.24531753361225128, "learning_rate": 9.867957435517342e-05, "loss": 0.0115, "step": 7180 }, { "grad_norm": 0.28841593861579895, "learning_rate": 9.867327558592854e-05, "loss": 0.011, "step": 7190 }, { "grad_norm": 0.22405701875686646, "learning_rate": 9.866696203102766e-05, "loss": 0.0104, "step": 7200 }, { "grad_norm": 0.16235508024692535, "learning_rate": 9.86606336923887e-05, "loss": 0.0101, "step": 7210 }, { "grad_norm": 0.18710583448410034, "learning_rate": 9.865429057193403e-05, "loss": 0.0081, "step": 7220 }, { "grad_norm": 0.25557249784469604, "learning_rate": 9.864793267159053e-05, "loss": 0.0113, "step": 7230 }, { "grad_norm": 0.18187925219535828, "learning_rate": 9.864155999328957e-05, "loss": 0.0125, "step": 7240 }, { "grad_norm": 0.2106826901435852, "learning_rate": 9.8635172538967e-05, "loss": 0.0091, "step": 7250 }, { "grad_norm": 0.21842263638973236, "learning_rate": 9.862877031056312e-05, "loss": 0.0088, "step": 7260 }, { "grad_norm": 0.15585815906524658, "learning_rate": 9.862235331002279e-05, "loss": 0.0104, "step": 7270 }, { "grad_norm": 0.153522789478302, "learning_rate": 9.861592153929533e-05, "loss": 0.0105, "step": 7280 }, { "grad_norm": 0.2110159695148468, "learning_rate": 9.860947500033455e-05, "loss": 0.0098, "step": 7290 }, { "grad_norm": 0.19908937811851501, "learning_rate": 9.86030136950987e-05, "loss": 0.0101, "step": 7300 }, { "grad_norm": 0.22333088517189026, "learning_rate": 9.85965376255506e-05, "loss": 0.0102, "step": 7310 }, { "grad_norm": 0.21908755600452423, "learning_rate": 9.859004679365747e-05, "loss": 0.0116, "step": 7320 }, { "grad_norm": 0.1620989739894867, "learning_rate": 9.858354120139108e-05, "loss": 0.0082, "step": 7330 }, { "grad_norm": 0.20424091815948486, "learning_rate": 9.857702085072764e-05, "loss": 0.0084, "step": 7340 }, { "grad_norm": 0.25713977217674255, "learning_rate": 9.857048574364787e-05, "loss": 0.0093, "step": 7350 }, { "grad_norm": 0.2434779405593872, "learning_rate": 9.856393588213698e-05, "loss": 0.0153, "step": 7360 }, { "grad_norm": 0.22534340620040894, "learning_rate": 9.855737126818458e-05, "loss": 0.0138, "step": 7370 }, { "grad_norm": 0.23254269361495972, "learning_rate": 9.855079190378491e-05, "loss": 0.0122, "step": 7380 }, { "grad_norm": 0.20121580362319946, "learning_rate": 9.854419779093655e-05, "loss": 0.0113, "step": 7390 }, { "grad_norm": 0.19068308174610138, "learning_rate": 9.853758893164264e-05, "loss": 0.0101, "step": 7400 }, { "grad_norm": 0.19366811215877533, "learning_rate": 9.853096532791078e-05, "loss": 0.0097, "step": 7410 }, { "grad_norm": 0.23774310946464539, "learning_rate": 9.852432698175304e-05, "loss": 0.0095, "step": 7420 }, { "grad_norm": 0.24150609970092773, "learning_rate": 9.851767389518597e-05, "loss": 0.0107, "step": 7430 }, { "grad_norm": 0.20313559472560883, "learning_rate": 9.85110060702306e-05, "loss": 0.0103, "step": 7440 }, { "grad_norm": 0.23374426364898682, "learning_rate": 9.850432350891245e-05, "loss": 0.0095, "step": 7450 }, { "grad_norm": 0.18314504623413086, "learning_rate": 9.84976262132615e-05, "loss": 0.0106, "step": 7460 }, { "grad_norm": 0.2967079281806946, "learning_rate": 9.849091418531222e-05, "loss": 0.0102, "step": 7470 }, { "grad_norm": 0.15997156500816345, "learning_rate": 9.848418742710353e-05, "loss": 0.0091, "step": 7480 }, { "grad_norm": 0.18592514097690582, "learning_rate": 9.847744594067885e-05, "loss": 0.009, "step": 7490 }, { "grad_norm": 0.22015932202339172, "learning_rate": 9.847068972808607e-05, "loss": 0.0103, "step": 7500 }, { "grad_norm": 0.2833805978298187, "learning_rate": 9.846391879137756e-05, "loss": 0.0099, "step": 7510 }, { "grad_norm": 0.1647232621908188, "learning_rate": 9.845713313261012e-05, "loss": 0.0115, "step": 7520 }, { "grad_norm": 0.214010551571846, "learning_rate": 9.845033275384505e-05, "loss": 0.0089, "step": 7530 }, { "grad_norm": 0.20872625708580017, "learning_rate": 9.844351765714818e-05, "loss": 0.009, "step": 7540 }, { "grad_norm": 0.15039925277233124, "learning_rate": 9.843668784458971e-05, "loss": 0.0091, "step": 7550 }, { "grad_norm": 0.23181912302970886, "learning_rate": 9.842984331824437e-05, "loss": 0.0114, "step": 7560 }, { "grad_norm": 0.2301759421825409, "learning_rate": 9.842298408019133e-05, "loss": 0.0094, "step": 7570 }, { "grad_norm": 0.24489472806453705, "learning_rate": 9.841611013251429e-05, "loss": 0.0099, "step": 7580 }, { "grad_norm": 0.18132449686527252, "learning_rate": 9.840922147730133e-05, "loss": 0.0089, "step": 7590 }, { "grad_norm": 0.20180903375148773, "learning_rate": 9.840231811664506e-05, "loss": 0.0093, "step": 7600 }, { "grad_norm": 0.1861850470304489, "learning_rate": 9.839540005264252e-05, "loss": 0.0083, "step": 7610 }, { "grad_norm": 0.2381121814250946, "learning_rate": 9.838846728739527e-05, "loss": 0.0094, "step": 7620 }, { "grad_norm": 0.2686672806739807, "learning_rate": 9.838151982300927e-05, "loss": 0.0105, "step": 7630 }, { "grad_norm": 0.27477774024009705, "learning_rate": 9.8374557661595e-05, "loss": 0.0103, "step": 7640 }, { "grad_norm": 0.19893842935562134, "learning_rate": 9.836758080526735e-05, "loss": 0.0107, "step": 7650 }, { "grad_norm": 0.2621132731437683, "learning_rate": 9.836058925614575e-05, "loss": 0.0112, "step": 7660 }, { "grad_norm": 0.16654473543167114, "learning_rate": 9.8353583016354e-05, "loss": 0.0097, "step": 7670 }, { "grad_norm": 0.1834627240896225, "learning_rate": 9.834656208802044e-05, "loss": 0.0089, "step": 7680 }, { "grad_norm": 0.2645115852355957, "learning_rate": 9.833952647327784e-05, "loss": 0.0086, "step": 7690 }, { "grad_norm": 0.22982200980186462, "learning_rate": 9.833247617426342e-05, "loss": 0.0107, "step": 7700 }, { "grad_norm": 0.13729524612426758, "learning_rate": 9.832541119311889e-05, "loss": 0.0106, "step": 7710 }, { "grad_norm": 0.2671828866004944, "learning_rate": 9.83183315319904e-05, "loss": 0.0084, "step": 7720 }, { "grad_norm": 0.22204378247261047, "learning_rate": 9.831123719302855e-05, "loss": 0.0101, "step": 7730 }, { "grad_norm": 0.19302518665790558, "learning_rate": 9.830412817838842e-05, "loss": 0.0099, "step": 7740 }, { "grad_norm": 0.21115466952323914, "learning_rate": 9.829700449022956e-05, "loss": 0.0097, "step": 7750 }, { "grad_norm": 0.1948501616716385, "learning_rate": 9.828986613071593e-05, "loss": 0.0096, "step": 7760 }, { "grad_norm": 0.2595529556274414, "learning_rate": 9.828271310201601e-05, "loss": 0.0139, "step": 7770 }, { "grad_norm": 0.16420646011829376, "learning_rate": 9.827554540630268e-05, "loss": 0.0099, "step": 7780 }, { "grad_norm": 0.15602487325668335, "learning_rate": 9.826836304575329e-05, "loss": 0.0086, "step": 7790 }, { "grad_norm": 0.216202974319458, "learning_rate": 9.826116602254966e-05, "loss": 0.0087, "step": 7800 }, { "grad_norm": 0.2681908905506134, "learning_rate": 9.825395433887805e-05, "loss": 0.0082, "step": 7810 }, { "grad_norm": 0.19040481746196747, "learning_rate": 9.824672799692917e-05, "loss": 0.0075, "step": 7820 }, { "grad_norm": 0.15299873054027557, "learning_rate": 9.823948699889823e-05, "loss": 0.0107, "step": 7830 }, { "grad_norm": 0.22674250602722168, "learning_rate": 9.823223134698483e-05, "loss": 0.0087, "step": 7840 }, { "grad_norm": 0.14766734838485718, "learning_rate": 9.822496104339303e-05, "loss": 0.0093, "step": 7850 }, { "grad_norm": 0.27044588327407837, "learning_rate": 9.821767609033138e-05, "loss": 0.0097, "step": 7860 }, { "grad_norm": 0.27136489748954773, "learning_rate": 9.821037649001284e-05, "loss": 0.0093, "step": 7870 }, { "grad_norm": 0.20458726584911346, "learning_rate": 9.820306224465486e-05, "loss": 0.0095, "step": 7880 }, { "grad_norm": 0.22678154706954956, "learning_rate": 9.819573335647928e-05, "loss": 0.0097, "step": 7890 }, { "grad_norm": 0.3166930377483368, "learning_rate": 9.818838982771246e-05, "loss": 0.0101, "step": 7900 }, { "grad_norm": 0.23094595968723297, "learning_rate": 9.818103166058514e-05, "loss": 0.0101, "step": 7910 }, { "grad_norm": 0.13381284475326538, "learning_rate": 9.817365885733254e-05, "loss": 0.009, "step": 7920 }, { "grad_norm": 0.16007331013679504, "learning_rate": 9.816627142019434e-05, "loss": 0.0098, "step": 7930 }, { "grad_norm": 0.17148540914058685, "learning_rate": 9.815886935141463e-05, "loss": 0.0098, "step": 7940 }, { "grad_norm": 0.22948548197746277, "learning_rate": 9.8151452653242e-05, "loss": 0.0088, "step": 7950 }, { "grad_norm": 0.15800480544567108, "learning_rate": 9.814402132792939e-05, "loss": 0.0092, "step": 7960 }, { "grad_norm": 0.24371029436588287, "learning_rate": 9.813657537773428e-05, "loss": 0.0109, "step": 7970 }, { "grad_norm": 0.2526441514492035, "learning_rate": 9.812911480491854e-05, "loss": 0.0107, "step": 7980 }, { "grad_norm": 0.2680172026157379, "learning_rate": 9.81216396117485e-05, "loss": 0.0095, "step": 7990 }, { "grad_norm": 0.20553508400917053, "learning_rate": 9.811414980049491e-05, "loss": 0.0087, "step": 8000 }, { "grad_norm": 0.17777948081493378, "learning_rate": 9.810664537343301e-05, "loss": 0.0073, "step": 8010 }, { "grad_norm": 0.22883816063404083, "learning_rate": 9.809912633284243e-05, "loss": 0.0091, "step": 8020 }, { "grad_norm": 0.24545183777809143, "learning_rate": 9.809159268100725e-05, "loss": 0.0085, "step": 8030 }, { "grad_norm": 0.2069026678800583, "learning_rate": 9.808404442021599e-05, "loss": 0.0089, "step": 8040 }, { "grad_norm": 0.17950335144996643, "learning_rate": 9.807648155276163e-05, "loss": 0.0081, "step": 8050 }, { "grad_norm": 0.30247387290000916, "learning_rate": 9.806890408094156e-05, "loss": 0.009, "step": 8060 }, { "grad_norm": 0.21063987910747528, "learning_rate": 9.806131200705761e-05, "loss": 0.0088, "step": 8070 }, { "grad_norm": 0.2189946472644806, "learning_rate": 9.805370533341605e-05, "loss": 0.0119, "step": 8080 }, { "grad_norm": 0.25797200202941895, "learning_rate": 9.804608406232762e-05, "loss": 0.0101, "step": 8090 }, { "grad_norm": 0.19559726119041443, "learning_rate": 9.803844819610741e-05, "loss": 0.0098, "step": 8100 }, { "grad_norm": 0.19572825729846954, "learning_rate": 9.803079773707504e-05, "loss": 0.0089, "step": 8110 }, { "grad_norm": 0.20892928540706635, "learning_rate": 9.802313268755447e-05, "loss": 0.0088, "step": 8120 }, { "grad_norm": 0.1921752691268921, "learning_rate": 9.801545304987419e-05, "loss": 0.0096, "step": 8130 }, { "grad_norm": 0.21684056520462036, "learning_rate": 9.800775882636704e-05, "loss": 0.008, "step": 8140 }, { "grad_norm": 0.24276763200759888, "learning_rate": 9.800005001937034e-05, "loss": 0.0104, "step": 8150 }, { "grad_norm": 0.1624593585729599, "learning_rate": 9.79923266312258e-05, "loss": 0.0098, "step": 8160 }, { "grad_norm": 0.16430117189884186, "learning_rate": 9.79845886642796e-05, "loss": 0.0097, "step": 8170 }, { "grad_norm": 0.1657726764678955, "learning_rate": 9.797683612088233e-05, "loss": 0.0089, "step": 8180 }, { "grad_norm": 0.19473975896835327, "learning_rate": 9.796906900338898e-05, "loss": 0.0097, "step": 8190 }, { "grad_norm": 0.22201302647590637, "learning_rate": 9.796128731415903e-05, "loss": 0.0085, "step": 8200 }, { "grad_norm": 0.18315303325653076, "learning_rate": 9.795349105555634e-05, "loss": 0.0079, "step": 8210 }, { "grad_norm": 0.17703647911548615, "learning_rate": 9.794568022994922e-05, "loss": 0.0074, "step": 8220 }, { "grad_norm": 0.21633920073509216, "learning_rate": 9.793785483971034e-05, "loss": 0.0074, "step": 8230 }, { "grad_norm": 0.1363980621099472, "learning_rate": 9.793001488721691e-05, "loss": 0.0086, "step": 8240 }, { "grad_norm": 0.2079075425863266, "learning_rate": 9.792216037485047e-05, "loss": 0.0101, "step": 8250 }, { "grad_norm": 0.1985439956188202, "learning_rate": 9.791429130499704e-05, "loss": 0.009, "step": 8260 }, { "grad_norm": 0.23077088594436646, "learning_rate": 9.790640768004698e-05, "loss": 0.0088, "step": 8270 }, { "grad_norm": 0.1551758348941803, "learning_rate": 9.789850950239518e-05, "loss": 0.0091, "step": 8280 }, { "grad_norm": 0.18689009547233582, "learning_rate": 9.789059677444089e-05, "loss": 0.0096, "step": 8290 }, { "grad_norm": 0.2013704627752304, "learning_rate": 9.788266949858776e-05, "loss": 0.0094, "step": 8300 }, { "grad_norm": 0.24490021169185638, "learning_rate": 9.787472767724392e-05, "loss": 0.0102, "step": 8310 }, { "grad_norm": 0.22172339260578156, "learning_rate": 9.786677131282185e-05, "loss": 0.0091, "step": 8320 }, { "grad_norm": 0.17054429650306702, "learning_rate": 9.785880040773853e-05, "loss": 0.0096, "step": 8330 }, { "grad_norm": 0.17062459886074066, "learning_rate": 9.785081496441527e-05, "loss": 0.0087, "step": 8340 }, { "grad_norm": 0.2062130570411682, "learning_rate": 9.784281498527785e-05, "loss": 0.0094, "step": 8350 }, { "grad_norm": 0.2013336420059204, "learning_rate": 9.783480047275646e-05, "loss": 0.0086, "step": 8360 }, { "grad_norm": 0.18264581263065338, "learning_rate": 9.78267714292857e-05, "loss": 0.0078, "step": 8370 }, { "grad_norm": 0.21823160350322723, "learning_rate": 9.781872785730454e-05, "loss": 0.008, "step": 8380 }, { "grad_norm": 0.1673564612865448, "learning_rate": 9.781066975925646e-05, "loss": 0.009, "step": 8390 }, { "grad_norm": 0.15353608131408691, "learning_rate": 9.780259713758928e-05, "loss": 0.0089, "step": 8400 }, { "grad_norm": 0.19579444825649261, "learning_rate": 9.779450999475524e-05, "loss": 0.0098, "step": 8410 }, { "grad_norm": 0.25259894132614136, "learning_rate": 9.7786408333211e-05, "loss": 0.0082, "step": 8420 }, { "grad_norm": 0.18975409865379333, "learning_rate": 9.777829215541764e-05, "loss": 0.0079, "step": 8430 }, { "grad_norm": 0.20106220245361328, "learning_rate": 9.777016146384064e-05, "loss": 0.0085, "step": 8440 }, { "grad_norm": 0.22724704444408417, "learning_rate": 9.776201626094988e-05, "loss": 0.0115, "step": 8450 }, { "grad_norm": 0.21965228021144867, "learning_rate": 9.775385654921965e-05, "loss": 0.0089, "step": 8460 }, { "grad_norm": 0.22372019290924072, "learning_rate": 9.774568233112868e-05, "loss": 0.0085, "step": 8470 }, { "grad_norm": 0.26249831914901733, "learning_rate": 9.773749360916007e-05, "loss": 0.0098, "step": 8480 }, { "grad_norm": 0.2268304079771042, "learning_rate": 9.772929038580134e-05, "loss": 0.0096, "step": 8490 }, { "grad_norm": 0.26287150382995605, "learning_rate": 9.772107266354439e-05, "loss": 0.0123, "step": 8500 }, { "grad_norm": 0.22434817254543304, "learning_rate": 9.77128404448856e-05, "loss": 0.0099, "step": 8510 }, { "grad_norm": 0.16768772900104523, "learning_rate": 9.770459373232565e-05, "loss": 0.0091, "step": 8520 }, { "grad_norm": 0.17203181982040405, "learning_rate": 9.769633252836969e-05, "loss": 0.0086, "step": 8530 }, { "grad_norm": 0.2011617124080658, "learning_rate": 9.768805683552724e-05, "loss": 0.01, "step": 8540 }, { "grad_norm": 0.15949563682079315, "learning_rate": 9.767976665631228e-05, "loss": 0.0087, "step": 8550 }, { "grad_norm": 0.2560414671897888, "learning_rate": 9.767146199324311e-05, "loss": 0.0098, "step": 8560 }, { "grad_norm": 0.14278870820999146, "learning_rate": 9.766314284884249e-05, "loss": 0.0114, "step": 8570 }, { "grad_norm": 0.2445216029882431, "learning_rate": 9.765480922563752e-05, "loss": 0.0099, "step": 8580 }, { "grad_norm": 0.22695572674274445, "learning_rate": 9.764646112615978e-05, "loss": 0.0096, "step": 8590 }, { "grad_norm": 0.19118037819862366, "learning_rate": 9.763809855294517e-05, "loss": 0.0089, "step": 8600 }, { "grad_norm": 0.1378934234380722, "learning_rate": 9.762972150853404e-05, "loss": 0.0097, "step": 8610 }, { "grad_norm": 0.19689039885997772, "learning_rate": 9.762132999547111e-05, "loss": 0.0083, "step": 8620 }, { "grad_norm": 0.2744772434234619, "learning_rate": 9.761292401630549e-05, "loss": 0.0104, "step": 8630 }, { "grad_norm": 0.1763019859790802, "learning_rate": 9.76045035735907e-05, "loss": 0.0096, "step": 8640 }, { "grad_norm": 0.16152021288871765, "learning_rate": 9.759606866988464e-05, "loss": 0.0099, "step": 8650 }, { "grad_norm": 0.24393711984157562, "learning_rate": 9.758761930774963e-05, "loss": 0.0095, "step": 8660 }, { "grad_norm": 0.23950302600860596, "learning_rate": 9.757915548975235e-05, "loss": 0.0095, "step": 8670 }, { "grad_norm": 0.20611099898815155, "learning_rate": 9.757067721846389e-05, "loss": 0.0093, "step": 8680 }, { "grad_norm": 0.17802837491035461, "learning_rate": 9.756218449645971e-05, "loss": 0.0087, "step": 8690 }, { "grad_norm": 0.1912371814250946, "learning_rate": 9.75536773263197e-05, "loss": 0.0079, "step": 8700 }, { "grad_norm": 0.3140741288661957, "learning_rate": 9.75451557106281e-05, "loss": 0.0084, "step": 8710 }, { "grad_norm": 0.23416782915592194, "learning_rate": 9.753661965197354e-05, "loss": 0.0095, "step": 8720 }, { "grad_norm": 0.2148873656988144, "learning_rate": 9.752806915294908e-05, "loss": 0.008, "step": 8730 }, { "grad_norm": 0.16416548192501068, "learning_rate": 9.75195042161521e-05, "loss": 0.0093, "step": 8740 }, { "grad_norm": 0.10019509494304657, "learning_rate": 9.751092484418442e-05, "loss": 0.0073, "step": 8750 }, { "grad_norm": 0.19408008456230164, "learning_rate": 9.750233103965224e-05, "loss": 0.0073, "step": 8760 }, { "grad_norm": 0.20547275245189667, "learning_rate": 9.749372280516611e-05, "loss": 0.0084, "step": 8770 }, { "grad_norm": 0.24975845217704773, "learning_rate": 9.748510014334097e-05, "loss": 0.0101, "step": 8780 }, { "grad_norm": 0.22200757265090942, "learning_rate": 9.747646305679621e-05, "loss": 0.0096, "step": 8790 }, { "grad_norm": 0.16665823757648468, "learning_rate": 9.74678115481555e-05, "loss": 0.0074, "step": 8800 }, { "grad_norm": 0.1741376519203186, "learning_rate": 9.745914562004696e-05, "loss": 0.0084, "step": 8810 }, { "grad_norm": 0.18316911160945892, "learning_rate": 9.745046527510307e-05, "loss": 0.008, "step": 8820 }, { "grad_norm": 0.18169954419136047, "learning_rate": 9.744177051596068e-05, "loss": 0.0075, "step": 8830 }, { "grad_norm": 0.207261860370636, "learning_rate": 9.743306134526105e-05, "loss": 0.0091, "step": 8840 }, { "grad_norm": 0.15289589762687683, "learning_rate": 9.742433776564977e-05, "loss": 0.009, "step": 8850 }, { "grad_norm": 0.1609528511762619, "learning_rate": 9.741559977977683e-05, "loss": 0.0082, "step": 8860 }, { "grad_norm": 0.24222858250141144, "learning_rate": 9.740684739029661e-05, "loss": 0.0092, "step": 8870 }, { "grad_norm": 0.1670111119747162, "learning_rate": 9.739808059986789e-05, "loss": 0.008, "step": 8880 }, { "grad_norm": 0.1794370859861374, "learning_rate": 9.738929941115373e-05, "loss": 0.0103, "step": 8890 }, { "grad_norm": 0.12583693861961365, "learning_rate": 9.738050382682167e-05, "loss": 0.0095, "step": 8900 }, { "grad_norm": 0.23305585980415344, "learning_rate": 9.737169384954355e-05, "loss": 0.0086, "step": 8910 }, { "grad_norm": 0.20847560465335846, "learning_rate": 9.736286948199562e-05, "loss": 0.0107, "step": 8920 }, { "grad_norm": 0.22775131464004517, "learning_rate": 9.735403072685848e-05, "loss": 0.0126, "step": 8930 }, { "grad_norm": 0.2132289558649063, "learning_rate": 9.734517758681712e-05, "loss": 0.0093, "step": 8940 }, { "grad_norm": 0.18743208050727844, "learning_rate": 9.733631006456088e-05, "loss": 0.0089, "step": 8950 }, { "grad_norm": 0.20613694190979004, "learning_rate": 9.732742816278348e-05, "loss": 0.0098, "step": 8960 }, { "grad_norm": 0.191519096493721, "learning_rate": 9.731853188418302e-05, "loss": 0.0095, "step": 8970 }, { "grad_norm": 0.16560544073581696, "learning_rate": 9.730962123146194e-05, "loss": 0.0096, "step": 8980 }, { "grad_norm": 0.24933350086212158, "learning_rate": 9.730069620732709e-05, "loss": 0.0114, "step": 8990 }, { "grad_norm": 0.2575203776359558, "learning_rate": 9.72917568144896e-05, "loss": 0.0105, "step": 9000 }, { "grad_norm": 0.2678906321525574, "learning_rate": 9.728280305566509e-05, "loss": 0.0105, "step": 9010 }, { "grad_norm": 0.23357273638248444, "learning_rate": 9.727383493357343e-05, "loss": 0.0108, "step": 9020 }, { "grad_norm": 0.17079627513885498, "learning_rate": 9.726485245093891e-05, "loss": 0.0103, "step": 9030 }, { "grad_norm": 0.16885964572429657, "learning_rate": 9.725585561049018e-05, "loss": 0.0078, "step": 9040 }, { "grad_norm": 0.18921607732772827, "learning_rate": 9.724684441496022e-05, "loss": 0.0085, "step": 9050 }, { "grad_norm": 0.2243928462266922, "learning_rate": 9.72378188670864e-05, "loss": 0.0089, "step": 9060 }, { "grad_norm": 0.17640098929405212, "learning_rate": 9.722877896961047e-05, "loss": 0.0079, "step": 9070 }, { "grad_norm": 0.18215858936309814, "learning_rate": 9.721972472527848e-05, "loss": 0.0079, "step": 9080 }, { "grad_norm": 0.12927840650081635, "learning_rate": 9.721065613684089e-05, "loss": 0.0085, "step": 9090 }, { "grad_norm": 0.22945469617843628, "learning_rate": 9.72015732070525e-05, "loss": 0.008, "step": 9100 }, { "grad_norm": 0.1957980841398239, "learning_rate": 9.719247593867244e-05, "loss": 0.007, "step": 9110 }, { "grad_norm": 0.2585962116718292, "learning_rate": 9.718336433446423e-05, "loss": 0.0088, "step": 9120 }, { "grad_norm": 0.1511133462190628, "learning_rate": 9.717423839719574e-05, "loss": 0.0081, "step": 9130 }, { "grad_norm": 0.1701323688030243, "learning_rate": 9.71650981296392e-05, "loss": 0.0072, "step": 9140 }, { "grad_norm": 0.14960132539272308, "learning_rate": 9.715594353457118e-05, "loss": 0.0093, "step": 9150 }, { "grad_norm": 0.13773399591445923, "learning_rate": 9.714677461477257e-05, "loss": 0.0068, "step": 9160 }, { "grad_norm": 0.1306198090314865, "learning_rate": 9.713759137302869e-05, "loss": 0.0079, "step": 9170 }, { "grad_norm": 0.17273500561714172, "learning_rate": 9.712839381212914e-05, "loss": 0.0077, "step": 9180 }, { "grad_norm": 0.2127779871225357, "learning_rate": 9.71191819348679e-05, "loss": 0.0095, "step": 9190 }, { "grad_norm": 0.25119268894195557, "learning_rate": 9.710995574404331e-05, "loss": 0.0093, "step": 9200 }, { "grad_norm": 0.19064481556415558, "learning_rate": 9.710071524245802e-05, "loss": 0.0102, "step": 9210 }, { "grad_norm": 0.19200658798217773, "learning_rate": 9.709146043291906e-05, "loss": 0.0083, "step": 9220 }, { "grad_norm": 0.17374517023563385, "learning_rate": 9.70821913182378e-05, "loss": 0.0089, "step": 9230 }, { "grad_norm": 0.18960197269916534, "learning_rate": 9.707290790122995e-05, "loss": 0.0078, "step": 9240 }, { "grad_norm": 0.20066794753074646, "learning_rate": 9.706361018471557e-05, "loss": 0.0089, "step": 9250 }, { "grad_norm": 0.1902751624584198, "learning_rate": 9.705429817151906e-05, "loss": 0.0106, "step": 9260 }, { "grad_norm": 0.2379666417837143, "learning_rate": 9.704497186446917e-05, "loss": 0.0104, "step": 9270 }, { "grad_norm": 0.18677310645580292, "learning_rate": 9.703563126639896e-05, "loss": 0.0102, "step": 9280 }, { "grad_norm": 0.16183757781982422, "learning_rate": 9.70262763801459e-05, "loss": 0.0082, "step": 9290 }, { "grad_norm": 0.17478598654270172, "learning_rate": 9.701690720855171e-05, "loss": 0.009, "step": 9300 }, { "grad_norm": 0.14422880113124847, "learning_rate": 9.700752375446253e-05, "loss": 0.0085, "step": 9310 }, { "grad_norm": 0.16552671790122986, "learning_rate": 9.69981260207288e-05, "loss": 0.0076, "step": 9320 }, { "grad_norm": 0.1884145736694336, "learning_rate": 9.698871401020529e-05, "loss": 0.0078, "step": 9330 }, { "grad_norm": 0.15756668150424957, "learning_rate": 9.697928772575112e-05, "loss": 0.0121, "step": 9340 }, { "grad_norm": 0.1752808541059494, "learning_rate": 9.696984717022976e-05, "loss": 0.0092, "step": 9350 }, { "grad_norm": 0.1880088448524475, "learning_rate": 9.6960392346509e-05, "loss": 0.0089, "step": 9360 }, { "grad_norm": 0.15989303588867188, "learning_rate": 9.695092325746097e-05, "loss": 0.0086, "step": 9370 }, { "grad_norm": 0.20746932923793793, "learning_rate": 9.694143990596211e-05, "loss": 0.0083, "step": 9380 }, { "grad_norm": 0.1615312099456787, "learning_rate": 9.693194229489325e-05, "loss": 0.0092, "step": 9390 }, { "grad_norm": 0.22494176030158997, "learning_rate": 9.692243042713944e-05, "loss": 0.0093, "step": 9400 }, { "grad_norm": 0.19161087274551392, "learning_rate": 9.691290430559022e-05, "loss": 0.0092, "step": 9410 }, { "grad_norm": 0.14702819287776947, "learning_rate": 9.690336393313932e-05, "loss": 0.0073, "step": 9420 }, { "grad_norm": 0.12548907101154327, "learning_rate": 9.689380931268487e-05, "loss": 0.0077, "step": 9430 }, { "grad_norm": 0.1498330533504486, "learning_rate": 9.688424044712932e-05, "loss": 0.0084, "step": 9440 }, { "grad_norm": 0.17168769240379333, "learning_rate": 9.687465733937942e-05, "loss": 0.0086, "step": 9450 }, { "grad_norm": 0.17755289375782013, "learning_rate": 9.686505999234627e-05, "loss": 0.009, "step": 9460 }, { "grad_norm": 0.16443972289562225, "learning_rate": 9.685544840894529e-05, "loss": 0.0089, "step": 9470 }, { "grad_norm": 0.15925192832946777, "learning_rate": 9.684582259209624e-05, "loss": 0.0076, "step": 9480 }, { "grad_norm": 0.20968161523342133, "learning_rate": 9.683618254472317e-05, "loss": 0.0073, "step": 9490 }, { "grad_norm": 0.1520514041185379, "learning_rate": 9.682652826975449e-05, "loss": 0.0068, "step": 9500 }, { "grad_norm": 0.1837688386440277, "learning_rate": 9.681685977012291e-05, "loss": 0.0095, "step": 9510 }, { "grad_norm": 0.2714928388595581, "learning_rate": 9.680717704876546e-05, "loss": 0.0097, "step": 9520 }, { "grad_norm": 0.21265621483325958, "learning_rate": 9.679748010862349e-05, "loss": 0.0075, "step": 9530 }, { "grad_norm": 0.18434371054172516, "learning_rate": 9.678776895264267e-05, "loss": 0.0081, "step": 9540 }, { "grad_norm": 0.197727769613266, "learning_rate": 9.6778043583773e-05, "loss": 0.0087, "step": 9550 }, { "grad_norm": 0.20220187306404114, "learning_rate": 9.67683040049688e-05, "loss": 0.0079, "step": 9560 }, { "grad_norm": 0.15666186809539795, "learning_rate": 9.675855021918869e-05, "loss": 0.0078, "step": 9570 }, { "grad_norm": 0.1756378412246704, "learning_rate": 9.674878222939561e-05, "loss": 0.0089, "step": 9580 }, { "grad_norm": 0.14880815148353577, "learning_rate": 9.673900003855681e-05, "loss": 0.0088, "step": 9590 }, { "grad_norm": 0.2024814486503601, "learning_rate": 9.672920364964389e-05, "loss": 0.0079, "step": 9600 }, { "grad_norm": 0.11303390562534332, "learning_rate": 9.671939306563269e-05, "loss": 0.0072, "step": 9610 }, { "grad_norm": 0.16505850851535797, "learning_rate": 9.670956828950345e-05, "loss": 0.0088, "step": 9620 }, { "grad_norm": 0.2508787214756012, "learning_rate": 9.669972932424065e-05, "loss": 0.0078, "step": 9630 }, { "grad_norm": 0.17774121463298798, "learning_rate": 9.668987617283312e-05, "loss": 0.0091, "step": 9640 }, { "grad_norm": 0.19980160892009735, "learning_rate": 9.668000883827397e-05, "loss": 0.0078, "step": 9650 }, { "grad_norm": 0.16219568252563477, "learning_rate": 9.667012732356067e-05, "loss": 0.0071, "step": 9660 }, { "grad_norm": 0.17155982553958893, "learning_rate": 9.666023163169493e-05, "loss": 0.0077, "step": 9670 }, { "grad_norm": 0.22763870656490326, "learning_rate": 9.665032176568281e-05, "loss": 0.0074, "step": 9680 }, { "grad_norm": 0.146286740899086, "learning_rate": 9.664039772853469e-05, "loss": 0.0071, "step": 9690 }, { "grad_norm": 0.23459312319755554, "learning_rate": 9.663045952326518e-05, "loss": 0.0075, "step": 9700 }, { "grad_norm": 0.18323279917240143, "learning_rate": 9.662050715289328e-05, "loss": 0.0079, "step": 9710 }, { "grad_norm": 0.13587093353271484, "learning_rate": 9.661054062044226e-05, "loss": 0.0073, "step": 9720 }, { "grad_norm": 0.1844230443239212, "learning_rate": 9.660055992893968e-05, "loss": 0.0072, "step": 9730 }, { "grad_norm": 0.1873893439769745, "learning_rate": 9.659056508141739e-05, "loss": 0.0093, "step": 9740 }, { "grad_norm": 0.20825417339801788, "learning_rate": 9.658055608091161e-05, "loss": 0.0087, "step": 9750 }, { "grad_norm": 0.19465163350105286, "learning_rate": 9.657053293046276e-05, "loss": 0.0086, "step": 9760 }, { "grad_norm": 0.24108390510082245, "learning_rate": 9.656049563311564e-05, "loss": 0.0096, "step": 9770 }, { "grad_norm": 0.20626682043075562, "learning_rate": 9.655044419191929e-05, "loss": 0.0114, "step": 9780 }, { "grad_norm": 0.1941199004650116, "learning_rate": 9.654037860992711e-05, "loss": 0.0106, "step": 9790 }, { "grad_norm": 0.14438331127166748, "learning_rate": 9.653029889019672e-05, "loss": 0.0075, "step": 9800 }, { "grad_norm": 0.158847838640213, "learning_rate": 9.65202050357901e-05, "loss": 0.0076, "step": 9810 }, { "grad_norm": 0.148558109998703, "learning_rate": 9.651009704977347e-05, "loss": 0.0071, "step": 9820 }, { "grad_norm": 0.34624767303466797, "learning_rate": 9.649997493521738e-05, "loss": 0.0105, "step": 9830 }, { "grad_norm": 0.19160129129886627, "learning_rate": 9.64898386951967e-05, "loss": 0.0121, "step": 9840 }, { "grad_norm": 0.2509956657886505, "learning_rate": 9.647968833279049e-05, "loss": 0.0134, "step": 9850 }, { "grad_norm": 0.21037183701992035, "learning_rate": 9.646952385108218e-05, "loss": 0.0102, "step": 9860 }, { "grad_norm": 0.1924392729997635, "learning_rate": 9.645934525315951e-05, "loss": 0.0119, "step": 9870 }, { "grad_norm": 0.16060808300971985, "learning_rate": 9.644915254211442e-05, "loss": 0.0105, "step": 9880 }, { "grad_norm": 0.2482631504535675, "learning_rate": 9.643894572104321e-05, "loss": 0.0122, "step": 9890 }, { "grad_norm": 0.16231736540794373, "learning_rate": 9.642872479304644e-05, "loss": 0.0114, "step": 9900 }, { "grad_norm": 0.19730514287948608, "learning_rate": 9.641848976122895e-05, "loss": 0.0095, "step": 9910 }, { "grad_norm": 0.17056389153003693, "learning_rate": 9.64082406286999e-05, "loss": 0.0112, "step": 9920 }, { "grad_norm": 0.22484782338142395, "learning_rate": 9.639797739857269e-05, "loss": 0.01, "step": 9930 }, { "grad_norm": 0.1937911957502365, "learning_rate": 9.638770007396498e-05, "loss": 0.0079, "step": 9940 }, { "grad_norm": 0.14849326014518738, "learning_rate": 9.63774086579988e-05, "loss": 0.0085, "step": 9950 }, { "grad_norm": 0.13249081373214722, "learning_rate": 9.63671031538004e-05, "loss": 0.0073, "step": 9960 }, { "grad_norm": 0.1353108286857605, "learning_rate": 9.635678356450031e-05, "loss": 0.0072, "step": 9970 }, { "grad_norm": 0.1541292667388916, "learning_rate": 9.634644989323336e-05, "loss": 0.0084, "step": 9980 }, { "grad_norm": 0.2046486884355545, "learning_rate": 9.633610214313861e-05, "loss": 0.0084, "step": 9990 }, { "grad_norm": 0.1277185082435608, "learning_rate": 9.632574031735951e-05, "loss": 0.0077, "step": 10000 }, { "grad_norm": 0.14877693355083466, "learning_rate": 9.631536441904364e-05, "loss": 0.0082, "step": 10010 }, { "grad_norm": 0.1732402741909027, "learning_rate": 9.630497445134293e-05, "loss": 0.0084, "step": 10020 }, { "grad_norm": 0.17450226843357086, "learning_rate": 9.62945704174136e-05, "loss": 0.0083, "step": 10030 }, { "grad_norm": 0.1390739381313324, "learning_rate": 9.628415232041612e-05, "loss": 0.0075, "step": 10040 }, { "grad_norm": 0.23504506051540375, "learning_rate": 9.627372016351524e-05, "loss": 0.0078, "step": 10050 }, { "grad_norm": 0.13311591744422913, "learning_rate": 9.626327394987995e-05, "loss": 0.0076, "step": 10060 }, { "grad_norm": 0.20664839446544647, "learning_rate": 9.625281368268355e-05, "loss": 0.0079, "step": 10070 }, { "grad_norm": 0.24377816915512085, "learning_rate": 9.624233936510357e-05, "loss": 0.01, "step": 10080 }, { "grad_norm": 0.16097991168498993, "learning_rate": 9.623185100032187e-05, "loss": 0.0095, "step": 10090 }, { "grad_norm": 0.22210273146629333, "learning_rate": 9.62213485915245e-05, "loss": 0.0102, "step": 10100 }, { "grad_norm": 0.19515471160411835, "learning_rate": 9.621083214190186e-05, "loss": 0.0092, "step": 10110 }, { "grad_norm": 0.23858733475208282, "learning_rate": 9.62003016546485e-05, "loss": 0.0078, "step": 10120 }, { "grad_norm": 0.16094405949115753, "learning_rate": 9.618975713296339e-05, "loss": 0.007, "step": 10130 }, { "grad_norm": 0.1838267594575882, "learning_rate": 9.61791985800496e-05, "loss": 0.0076, "step": 10140 }, { "grad_norm": 0.14689944684505463, "learning_rate": 9.616862599911458e-05, "loss": 0.0077, "step": 10150 }, { "grad_norm": 0.14380896091461182, "learning_rate": 9.615803939337e-05, "loss": 0.0082, "step": 10160 }, { "grad_norm": 0.21522735059261322, "learning_rate": 9.614743876603178e-05, "loss": 0.0096, "step": 10170 }, { "grad_norm": 0.18471138179302216, "learning_rate": 9.613682412032013e-05, "loss": 0.0079, "step": 10180 }, { "grad_norm": 0.15048225224018097, "learning_rate": 9.612619545945947e-05, "loss": 0.0065, "step": 10190 }, { "grad_norm": 0.17259721457958221, "learning_rate": 9.611555278667852e-05, "loss": 0.0073, "step": 10200 }, { "grad_norm": 0.16143815219402313, "learning_rate": 9.610489610521024e-05, "loss": 0.0079, "step": 10210 }, { "grad_norm": 0.1747552454471588, "learning_rate": 9.609422541829187e-05, "loss": 0.0089, "step": 10220 }, { "grad_norm": 0.1615411788225174, "learning_rate": 9.608354072916486e-05, "loss": 0.0072, "step": 10230 }, { "grad_norm": 0.184183731675148, "learning_rate": 9.607284204107493e-05, "loss": 0.0081, "step": 10240 }, { "grad_norm": 0.1618010699748993, "learning_rate": 9.606212935727208e-05, "loss": 0.0071, "step": 10250 }, { "grad_norm": 0.1594415307044983, "learning_rate": 9.605140268101052e-05, "loss": 0.0093, "step": 10260 }, { "grad_norm": 0.17978312075138092, "learning_rate": 9.604066201554875e-05, "loss": 0.008, "step": 10270 }, { "grad_norm": 0.22130413353443146, "learning_rate": 9.60299073641495e-05, "loss": 0.0078, "step": 10280 }, { "grad_norm": 0.16304796934127808, "learning_rate": 9.601913873007974e-05, "loss": 0.0084, "step": 10290 }, { "grad_norm": 0.234639972448349, "learning_rate": 9.60083561166107e-05, "loss": 0.0076, "step": 10300 }, { "grad_norm": 0.2841924726963043, "learning_rate": 9.599755952701783e-05, "loss": 0.0077, "step": 10310 }, { "grad_norm": 0.16366443037986755, "learning_rate": 9.598674896458089e-05, "loss": 0.0068, "step": 10320 }, { "grad_norm": 0.24552030861377716, "learning_rate": 9.597592443258383e-05, "loss": 0.0086, "step": 10330 }, { "grad_norm": 0.19679199159145355, "learning_rate": 9.596508593431483e-05, "loss": 0.0078, "step": 10340 }, { "grad_norm": 0.16912056505680084, "learning_rate": 9.59542334730664e-05, "loss": 0.008, "step": 10350 }, { "grad_norm": 0.17073877155780792, "learning_rate": 9.594336705213516e-05, "loss": 0.0074, "step": 10360 }, { "grad_norm": 0.21410375833511353, "learning_rate": 9.593248667482208e-05, "loss": 0.0076, "step": 10370 }, { "grad_norm": 0.18286049365997314, "learning_rate": 9.592159234443233e-05, "loss": 0.007, "step": 10380 }, { "grad_norm": 0.16703765094280243, "learning_rate": 9.59106840642753e-05, "loss": 0.0083, "step": 10390 }, { "grad_norm": 0.16797849535942078, "learning_rate": 9.589976183766467e-05, "loss": 0.009, "step": 10400 }, { "grad_norm": 0.2609595060348511, "learning_rate": 9.58888256679183e-05, "loss": 0.0093, "step": 10410 }, { "grad_norm": 0.15310336649417877, "learning_rate": 9.587787555835832e-05, "loss": 0.0088, "step": 10420 }, { "grad_norm": 0.1473853588104248, "learning_rate": 9.586691151231107e-05, "loss": 0.008, "step": 10430 }, { "grad_norm": 0.20532146096229553, "learning_rate": 9.585593353310715e-05, "loss": 0.0074, "step": 10440 }, { "grad_norm": 0.16133126616477966, "learning_rate": 9.58449416240814e-05, "loss": 0.0064, "step": 10450 }, { "grad_norm": 0.15492017567157745, "learning_rate": 9.583393578857283e-05, "loss": 0.008, "step": 10460 }, { "grad_norm": 0.1401054561138153, "learning_rate": 9.582291602992474e-05, "loss": 0.0069, "step": 10470 }, { "grad_norm": 0.17451073229312897, "learning_rate": 9.581188235148466e-05, "loss": 0.0074, "step": 10480 }, { "grad_norm": 0.14397664368152618, "learning_rate": 9.58008347566043e-05, "loss": 0.0075, "step": 10490 }, { "grad_norm": 0.21693894267082214, "learning_rate": 9.578977324863965e-05, "loss": 0.0105, "step": 10500 }, { "grad_norm": 0.16157476603984833, "learning_rate": 9.577869783095089e-05, "loss": 0.0062, "step": 10510 }, { "grad_norm": 0.15006621181964874, "learning_rate": 9.576760850690245e-05, "loss": 0.0071, "step": 10520 }, { "grad_norm": 0.1901611089706421, "learning_rate": 9.575650527986298e-05, "loss": 0.0073, "step": 10530 }, { "grad_norm": 0.21397753059864044, "learning_rate": 9.574538815320531e-05, "loss": 0.0081, "step": 10540 }, { "grad_norm": 0.18866056203842163, "learning_rate": 9.573425713030656e-05, "loss": 0.0081, "step": 10550 }, { "grad_norm": 0.17836225032806396, "learning_rate": 9.572311221454806e-05, "loss": 0.0073, "step": 10560 }, { "grad_norm": 0.21594157814979553, "learning_rate": 9.57119534093153e-05, "loss": 0.0082, "step": 10570 }, { "grad_norm": 0.15842242538928986, "learning_rate": 9.570078071799806e-05, "loss": 0.0072, "step": 10580 }, { "grad_norm": 0.1077301949262619, "learning_rate": 9.568959414399028e-05, "loss": 0.0068, "step": 10590 }, { "grad_norm": 0.16928721964359283, "learning_rate": 9.567839369069018e-05, "loss": 0.0068, "step": 10600 }, { "grad_norm": 0.19963468611240387, "learning_rate": 9.566717936150013e-05, "loss": 0.0061, "step": 10610 }, { "grad_norm": 0.18868370354175568, "learning_rate": 9.565595115982678e-05, "loss": 0.0083, "step": 10620 }, { "grad_norm": 0.11641501635313034, "learning_rate": 9.564470908908094e-05, "loss": 0.0081, "step": 10630 }, { "grad_norm": 0.21289359033107758, "learning_rate": 9.563345315267764e-05, "loss": 0.009, "step": 10640 }, { "grad_norm": 0.19768044352531433, "learning_rate": 9.562218335403616e-05, "loss": 0.0088, "step": 10650 }, { "grad_norm": 0.13821199536323547, "learning_rate": 9.561089969657999e-05, "loss": 0.0068, "step": 10660 }, { "grad_norm": 0.14738482236862183, "learning_rate": 9.559960218373673e-05, "loss": 0.0073, "step": 10670 }, { "grad_norm": 0.2680721879005432, "learning_rate": 9.558829081893836e-05, "loss": 0.0083, "step": 10680 }, { "grad_norm": 0.17825905978679657, "learning_rate": 9.55769656056209e-05, "loss": 0.0077, "step": 10690 }, { "grad_norm": 0.20600773394107819, "learning_rate": 9.556562654722469e-05, "loss": 0.0075, "step": 10700 }, { "grad_norm": 0.1510475128889084, "learning_rate": 9.555427364719422e-05, "loss": 0.0075, "step": 10710 }, { "grad_norm": 0.1682465523481369, "learning_rate": 9.55429069089782e-05, "loss": 0.0084, "step": 10720 }, { "grad_norm": 0.1438562273979187, "learning_rate": 9.553152633602956e-05, "loss": 0.0078, "step": 10730 }, { "grad_norm": 0.16726508736610413, "learning_rate": 9.552013193180543e-05, "loss": 0.0087, "step": 10740 }, { "grad_norm": 0.1994822472333908, "learning_rate": 9.550872369976707e-05, "loss": 0.008, "step": 10750 }, { "grad_norm": 0.17915023863315582, "learning_rate": 9.549730164338007e-05, "loss": 0.0073, "step": 10760 }, { "grad_norm": 0.2013346403837204, "learning_rate": 9.548586576611408e-05, "loss": 0.0073, "step": 10770 }, { "grad_norm": 0.12299773097038269, "learning_rate": 9.54744160714431e-05, "loss": 0.0072, "step": 10780 }, { "grad_norm": 0.16677629947662354, "learning_rate": 9.546295256284516e-05, "loss": 0.007, "step": 10790 }, { "grad_norm": 0.18961666524410248, "learning_rate": 9.545147524380265e-05, "loss": 0.0085, "step": 10800 }, { "grad_norm": 0.12198671698570251, "learning_rate": 9.543998411780201e-05, "loss": 0.0084, "step": 10810 }, { "grad_norm": 0.15893670916557312, "learning_rate": 9.542847918833397e-05, "loss": 0.0082, "step": 10820 }, { "grad_norm": 0.14078204333782196, "learning_rate": 9.541696045889343e-05, "loss": 0.0062, "step": 10830 }, { "grad_norm": 0.17510578036308289, "learning_rate": 9.540542793297947e-05, "loss": 0.0074, "step": 10840 }, { "grad_norm": 0.1377214789390564, "learning_rate": 9.539388161409537e-05, "loss": 0.0089, "step": 10850 }, { "grad_norm": 0.19164493680000305, "learning_rate": 9.538232150574857e-05, "loss": 0.0077, "step": 10860 }, { "grad_norm": 0.17358794808387756, "learning_rate": 9.537074761145076e-05, "loss": 0.0082, "step": 10870 }, { "grad_norm": 0.15221351385116577, "learning_rate": 9.535915993471778e-05, "loss": 0.0081, "step": 10880 }, { "grad_norm": 0.10725689679384232, "learning_rate": 9.534755847906964e-05, "loss": 0.0073, "step": 10890 }, { "grad_norm": 0.2528046667575836, "learning_rate": 9.533594324803057e-05, "loss": 0.0079, "step": 10900 }, { "grad_norm": 0.21275447309017181, "learning_rate": 9.532431424512895e-05, "loss": 0.0087, "step": 10910 }, { "grad_norm": 0.09395437687635422, "learning_rate": 9.531267147389741e-05, "loss": 0.0074, "step": 10920 }, { "grad_norm": 0.21491047739982605, "learning_rate": 9.530101493787266e-05, "loss": 0.0077, "step": 10930 }, { "grad_norm": 0.154053196310997, "learning_rate": 9.528934464059571e-05, "loss": 0.0071, "step": 10940 }, { "grad_norm": 0.19005203247070312, "learning_rate": 9.527766058561163e-05, "loss": 0.0072, "step": 10950 }, { "grad_norm": 0.15363629162311554, "learning_rate": 9.526596277646976e-05, "loss": 0.0067, "step": 10960 }, { "grad_norm": 0.18564030528068542, "learning_rate": 9.525425121672358e-05, "loss": 0.0083, "step": 10970 }, { "grad_norm": 0.1501908153295517, "learning_rate": 9.524252590993074e-05, "loss": 0.007, "step": 10980 }, { "grad_norm": 0.20079569518566132, "learning_rate": 9.523078685965309e-05, "loss": 0.0084, "step": 10990 }, { "grad_norm": 0.19654612243175507, "learning_rate": 9.521903406945664e-05, "loss": 0.0073, "step": 11000 }, { "grad_norm": 0.16472822427749634, "learning_rate": 9.520726754291158e-05, "loss": 0.0078, "step": 11010 }, { "grad_norm": 0.1405155062675476, "learning_rate": 9.519548728359227e-05, "loss": 0.0075, "step": 11020 }, { "grad_norm": 0.20609809458255768, "learning_rate": 9.518369329507726e-05, "loss": 0.0068, "step": 11030 }, { "grad_norm": 0.16839903593063354, "learning_rate": 9.51718855809492e-05, "loss": 0.008, "step": 11040 }, { "grad_norm": 0.2066977620124817, "learning_rate": 9.516006414479502e-05, "loss": 0.008, "step": 11050 }, { "grad_norm": 0.15206675231456757, "learning_rate": 9.514822899020572e-05, "loss": 0.0081, "step": 11060 }, { "grad_norm": 0.2386295348405838, "learning_rate": 9.513638012077654e-05, "loss": 0.0075, "step": 11070 }, { "grad_norm": 0.11901847273111343, "learning_rate": 9.512451754010683e-05, "loss": 0.0075, "step": 11080 }, { "grad_norm": 0.12737704813480377, "learning_rate": 9.511264125180013e-05, "loss": 0.0063, "step": 11090 }, { "grad_norm": 0.1236511841416359, "learning_rate": 9.510075125946414e-05, "loss": 0.0073, "step": 11100 }, { "grad_norm": 0.1847396343946457, "learning_rate": 9.508884756671075e-05, "loss": 0.0066, "step": 11110 }, { "grad_norm": 0.18471145629882812, "learning_rate": 9.507693017715596e-05, "loss": 0.0078, "step": 11120 }, { "grad_norm": 0.1431005895137787, "learning_rate": 9.506499909441997e-05, "loss": 0.0079, "step": 11130 }, { "grad_norm": 0.20400649309158325, "learning_rate": 9.505305432212713e-05, "loss": 0.007, "step": 11140 }, { "grad_norm": 0.17914274334907532, "learning_rate": 9.504109586390595e-05, "loss": 0.0062, "step": 11150 }, { "grad_norm": 0.1463640183210373, "learning_rate": 9.502912372338908e-05, "loss": 0.0064, "step": 11160 }, { "grad_norm": 0.15115781128406525, "learning_rate": 9.501713790421335e-05, "loss": 0.0068, "step": 11170 }, { "grad_norm": 0.1601022481918335, "learning_rate": 9.500513841001974e-05, "loss": 0.0059, "step": 11180 }, { "grad_norm": 0.15264590084552765, "learning_rate": 9.499312524445336e-05, "loss": 0.0079, "step": 11190 }, { "grad_norm": 0.14138618111610413, "learning_rate": 9.498109841116351e-05, "loss": 0.0077, "step": 11200 }, { "grad_norm": 0.2443380057811737, "learning_rate": 9.496905791380363e-05, "loss": 0.0077, "step": 11210 }, { "grad_norm": 0.19472429156303406, "learning_rate": 9.495700375603129e-05, "loss": 0.0067, "step": 11220 }, { "grad_norm": 0.20663347840309143, "learning_rate": 9.494493594150822e-05, "loss": 0.0102, "step": 11230 }, { "grad_norm": 0.17869147658348083, "learning_rate": 9.493285447390032e-05, "loss": 0.0066, "step": 11240 }, { "grad_norm": 0.18590354919433594, "learning_rate": 9.492075935687761e-05, "loss": 0.007, "step": 11250 }, { "grad_norm": 0.14109189808368683, "learning_rate": 9.490865059411427e-05, "loss": 0.0069, "step": 11260 }, { "grad_norm": 0.1524852067232132, "learning_rate": 9.489652818928863e-05, "loss": 0.0073, "step": 11270 }, { "grad_norm": 0.1469593048095703, "learning_rate": 9.488439214608315e-05, "loss": 0.0065, "step": 11280 }, { "grad_norm": 0.14639423787593842, "learning_rate": 9.487224246818444e-05, "loss": 0.0065, "step": 11290 }, { "grad_norm": 0.15060488879680634, "learning_rate": 9.486007915928325e-05, "loss": 0.0064, "step": 11300 }, { "grad_norm": 0.12038417905569077, "learning_rate": 9.484790222307448e-05, "loss": 0.0068, "step": 11310 }, { "grad_norm": 0.1492648869752884, "learning_rate": 9.483571166325716e-05, "loss": 0.0068, "step": 11320 }, { "grad_norm": 0.12764650583267212, "learning_rate": 9.482350748353444e-05, "loss": 0.0068, "step": 11330 }, { "grad_norm": 0.13908953964710236, "learning_rate": 9.481128968761363e-05, "loss": 0.0077, "step": 11340 }, { "grad_norm": 0.14183896780014038, "learning_rate": 9.479905827920621e-05, "loss": 0.0078, "step": 11350 }, { "grad_norm": 0.15708036720752716, "learning_rate": 9.478681326202773e-05, "loss": 0.0079, "step": 11360 }, { "grad_norm": 0.1711425930261612, "learning_rate": 9.477455463979791e-05, "loss": 0.0089, "step": 11370 }, { "grad_norm": 0.1381998211145401, "learning_rate": 9.476228241624059e-05, "loss": 0.0085, "step": 11380 }, { "grad_norm": 0.15988267958164215, "learning_rate": 9.474999659508374e-05, "loss": 0.0076, "step": 11390 }, { "grad_norm": 0.1240556538105011, "learning_rate": 9.47376971800595e-05, "loss": 0.0078, "step": 11400 }, { "grad_norm": 0.1363651603460312, "learning_rate": 9.472538417490409e-05, "loss": 0.0087, "step": 11410 }, { "grad_norm": 0.21648457646369934, "learning_rate": 9.471305758335784e-05, "loss": 0.0067, "step": 11420 }, { "grad_norm": 0.1330215334892273, "learning_rate": 9.47007174091653e-05, "loss": 0.0067, "step": 11430 }, { "grad_norm": 0.10401192307472229, "learning_rate": 9.468836365607507e-05, "loss": 0.0062, "step": 11440 }, { "grad_norm": 0.17066127061843872, "learning_rate": 9.467599632783988e-05, "loss": 0.0071, "step": 11450 }, { "grad_norm": 0.13160260021686554, "learning_rate": 9.466361542821662e-05, "loss": 0.0076, "step": 11460 }, { "grad_norm": 0.14553581178188324, "learning_rate": 9.465122096096625e-05, "loss": 0.0066, "step": 11470 }, { "grad_norm": 0.12321335822343826, "learning_rate": 9.463881292985391e-05, "loss": 0.0059, "step": 11480 }, { "grad_norm": 0.16820929944515228, "learning_rate": 9.462639133864881e-05, "loss": 0.0064, "step": 11490 }, { "grad_norm": 0.1933804154396057, "learning_rate": 9.461395619112432e-05, "loss": 0.0069, "step": 11500 }, { "grad_norm": 0.16274519264698029, "learning_rate": 9.460150749105791e-05, "loss": 0.0072, "step": 11510 }, { "grad_norm": 0.16133961081504822, "learning_rate": 9.458904524223116e-05, "loss": 0.0085, "step": 11520 }, { "grad_norm": 0.2330504059791565, "learning_rate": 9.457656944842976e-05, "loss": 0.0092, "step": 11530 }, { "grad_norm": 0.27903640270233154, "learning_rate": 9.456408011344353e-05, "loss": 0.0095, "step": 11540 }, { "grad_norm": 0.20593151450157166, "learning_rate": 9.455157724106643e-05, "loss": 0.0087, "step": 11550 }, { "grad_norm": 0.13704317808151245, "learning_rate": 9.453906083509647e-05, "loss": 0.0078, "step": 11560 }, { "grad_norm": 0.12454140931367874, "learning_rate": 9.45265308993358e-05, "loss": 0.0072, "step": 11570 }, { "grad_norm": 0.14606046676635742, "learning_rate": 9.451398743759071e-05, "loss": 0.0082, "step": 11580 }, { "grad_norm": 0.2285161316394806, "learning_rate": 9.450143045367156e-05, "loss": 0.0086, "step": 11590 }, { "grad_norm": 0.09291855990886688, "learning_rate": 9.448885995139283e-05, "loss": 0.007, "step": 11600 }, { "grad_norm": 0.16551949083805084, "learning_rate": 9.44762759345731e-05, "loss": 0.0084, "step": 11610 }, { "grad_norm": 0.24187709391117096, "learning_rate": 9.446367840703509e-05, "loss": 0.0101, "step": 11620 }, { "grad_norm": 0.17688533663749695, "learning_rate": 9.445106737260556e-05, "loss": 0.0117, "step": 11630 }, { "grad_norm": 0.16308043897151947, "learning_rate": 9.443844283511543e-05, "loss": 0.0084, "step": 11640 }, { "grad_norm": 0.12256437540054321, "learning_rate": 9.442580479839968e-05, "loss": 0.0074, "step": 11650 }, { "grad_norm": 0.14950814843177795, "learning_rate": 9.441315326629745e-05, "loss": 0.0072, "step": 11660 }, { "grad_norm": 0.21945783495903015, "learning_rate": 9.44004882426519e-05, "loss": 0.0077, "step": 11670 }, { "grad_norm": 0.16672828793525696, "learning_rate": 9.438780973131037e-05, "loss": 0.0066, "step": 11680 }, { "grad_norm": 0.14999136328697205, "learning_rate": 9.437511773612423e-05, "loss": 0.0065, "step": 11690 }, { "grad_norm": 0.12238258123397827, "learning_rate": 9.436241226094896e-05, "loss": 0.0076, "step": 11700 }, { "grad_norm": 0.1199365183711052, "learning_rate": 9.434969330964418e-05, "loss": 0.0071, "step": 11710 }, { "grad_norm": 0.2506870925426483, "learning_rate": 9.433696088607356e-05, "loss": 0.008, "step": 11720 }, { "grad_norm": 0.14209020137786865, "learning_rate": 9.432421499410486e-05, "loss": 0.0077, "step": 11730 }, { "grad_norm": 0.22521211206912994, "learning_rate": 9.431145563760998e-05, "loss": 0.0071, "step": 11740 }, { "grad_norm": 0.1291673481464386, "learning_rate": 9.429868282046484e-05, "loss": 0.0072, "step": 11750 }, { "grad_norm": 0.23795078694820404, "learning_rate": 9.428589654654951e-05, "loss": 0.0075, "step": 11760 }, { "grad_norm": 0.1962433159351349, "learning_rate": 9.42730968197481e-05, "loss": 0.008, "step": 11770 }, { "grad_norm": 0.16383087635040283, "learning_rate": 9.426028364394883e-05, "loss": 0.0067, "step": 11780 }, { "grad_norm": 0.14980100095272064, "learning_rate": 9.424745702304402e-05, "loss": 0.0072, "step": 11790 }, { "grad_norm": 0.14533360302448273, "learning_rate": 9.423461696093006e-05, "loss": 0.0076, "step": 11800 }, { "grad_norm": 0.11927703022956848, "learning_rate": 9.422176346150741e-05, "loss": 0.0068, "step": 11810 }, { "grad_norm": 0.17075565457344055, "learning_rate": 9.420889652868063e-05, "loss": 0.0076, "step": 11820 }, { "grad_norm": 0.16218970715999603, "learning_rate": 9.419601616635836e-05, "loss": 0.0076, "step": 11830 }, { "grad_norm": 0.16034547984600067, "learning_rate": 9.418312237845331e-05, "loss": 0.0076, "step": 11840 }, { "grad_norm": 0.14332519471645355, "learning_rate": 9.417021516888225e-05, "loss": 0.0069, "step": 11850 }, { "grad_norm": 0.1718510091304779, "learning_rate": 9.415729454156608e-05, "loss": 0.0061, "step": 11860 }, { "grad_norm": 0.1537320762872696, "learning_rate": 9.414436050042973e-05, "loss": 0.0087, "step": 11870 }, { "grad_norm": 0.15006513893604279, "learning_rate": 9.413141304940223e-05, "loss": 0.0078, "step": 11880 }, { "grad_norm": 0.1382906138896942, "learning_rate": 9.411845219241666e-05, "loss": 0.0073, "step": 11890 }, { "grad_norm": 0.15428054332733154, "learning_rate": 9.410547793341021e-05, "loss": 0.0076, "step": 11900 }, { "grad_norm": 0.14951461553573608, "learning_rate": 9.409249027632408e-05, "loss": 0.0076, "step": 11910 }, { "grad_norm": 0.1463913917541504, "learning_rate": 9.407948922510362e-05, "loss": 0.0077, "step": 11920 }, { "grad_norm": 0.19131803512573242, "learning_rate": 9.406647478369817e-05, "loss": 0.0088, "step": 11930 }, { "grad_norm": 0.14559558033943176, "learning_rate": 9.405344695606118e-05, "loss": 0.009, "step": 11940 }, { "grad_norm": 0.1257057636976242, "learning_rate": 9.404040574615018e-05, "loss": 0.0078, "step": 11950 }, { "grad_norm": 0.17281369864940643, "learning_rate": 9.402735115792674e-05, "loss": 0.0079, "step": 11960 }, { "grad_norm": 0.17256860435009003, "learning_rate": 9.401428319535649e-05, "loss": 0.0071, "step": 11970 }, { "grad_norm": 0.1465086191892624, "learning_rate": 9.400120186240912e-05, "loss": 0.0066, "step": 11980 }, { "grad_norm": 0.12505146861076355, "learning_rate": 9.398810716305844e-05, "loss": 0.0061, "step": 11990 }, { "grad_norm": 0.1440432369709015, "learning_rate": 9.397499910128222e-05, "loss": 0.0084, "step": 12000 }, { "grad_norm": 0.16327917575836182, "learning_rate": 9.396187768106237e-05, "loss": 0.0078, "step": 12010 }, { "grad_norm": 0.16262125968933105, "learning_rate": 9.394874290638482e-05, "loss": 0.0074, "step": 12020 }, { "grad_norm": 0.17322643101215363, "learning_rate": 9.393559478123959e-05, "loss": 0.0068, "step": 12030 }, { "grad_norm": 0.09363054484128952, "learning_rate": 9.39224333096207e-05, "loss": 0.0061, "step": 12040 }, { "grad_norm": 0.17910003662109375, "learning_rate": 9.390925849552629e-05, "loss": 0.0068, "step": 12050 }, { "grad_norm": 0.16604410111904144, "learning_rate": 9.389607034295849e-05, "loss": 0.0072, "step": 12060 }, { "grad_norm": 0.1463427096605301, "learning_rate": 9.388286885592355e-05, "loss": 0.0062, "step": 12070 }, { "grad_norm": 0.19385825097560883, "learning_rate": 9.386965403843168e-05, "loss": 0.0064, "step": 12080 }, { "grad_norm": 0.184622123837471, "learning_rate": 9.385642589449726e-05, "loss": 0.0057, "step": 12090 }, { "grad_norm": 0.1693432480096817, "learning_rate": 9.38431844281386e-05, "loss": 0.0062, "step": 12100 }, { "grad_norm": 0.1781751960515976, "learning_rate": 9.38299296433781e-05, "loss": 0.0062, "step": 12110 }, { "grad_norm": 0.16767734289169312, "learning_rate": 9.381666154424226e-05, "loss": 0.0068, "step": 12120 }, { "grad_norm": 0.17505338788032532, "learning_rate": 9.380338013476157e-05, "loss": 0.0067, "step": 12130 }, { "grad_norm": 0.2044156938791275, "learning_rate": 9.379008541897054e-05, "loss": 0.0063, "step": 12140 }, { "grad_norm": 0.16815084218978882, "learning_rate": 9.377677740090777e-05, "loss": 0.0061, "step": 12150 }, { "grad_norm": 0.17098890244960785, "learning_rate": 9.376345608461588e-05, "loss": 0.0076, "step": 12160 }, { "grad_norm": 0.17406657338142395, "learning_rate": 9.375012147414155e-05, "loss": 0.0069, "step": 12170 }, { "grad_norm": 0.1885722428560257, "learning_rate": 9.373677357353545e-05, "loss": 0.0082, "step": 12180 }, { "grad_norm": 0.19460640847682953, "learning_rate": 9.372341238685237e-05, "loss": 0.0068, "step": 12190 }, { "grad_norm": 0.11243794113397598, "learning_rate": 9.371003791815102e-05, "loss": 0.0084, "step": 12200 }, { "grad_norm": 0.21392297744750977, "learning_rate": 9.369665017149429e-05, "loss": 0.0089, "step": 12210 }, { "grad_norm": 0.180430606007576, "learning_rate": 9.368324915094895e-05, "loss": 0.006, "step": 12220 }, { "grad_norm": 0.11290343105792999, "learning_rate": 9.366983486058591e-05, "loss": 0.0074, "step": 12230 }, { "grad_norm": 0.15134306252002716, "learning_rate": 9.365640730448009e-05, "loss": 0.0078, "step": 12240 }, { "grad_norm": 0.1465219408273697, "learning_rate": 9.36429664867104e-05, "loss": 0.0064, "step": 12250 }, { "grad_norm": 0.12668655812740326, "learning_rate": 9.362951241135982e-05, "loss": 0.0064, "step": 12260 }, { "grad_norm": 0.22357943654060364, "learning_rate": 9.361604508251534e-05, "loss": 0.0092, "step": 12270 }, { "grad_norm": 0.2453491985797882, "learning_rate": 9.360256450426799e-05, "loss": 0.0074, "step": 12280 }, { "grad_norm": 0.2499997317790985, "learning_rate": 9.358907068071279e-05, "loss": 0.0061, "step": 12290 }, { "grad_norm": 0.19289129972457886, "learning_rate": 9.357556361594882e-05, "loss": 0.0086, "step": 12300 }, { "grad_norm": 0.17823338508605957, "learning_rate": 9.356204331407917e-05, "loss": 0.0076, "step": 12310 }, { "grad_norm": 0.12884674966335297, "learning_rate": 9.354850977921094e-05, "loss": 0.0069, "step": 12320 }, { "grad_norm": 0.17812669277191162, "learning_rate": 9.353496301545529e-05, "loss": 0.0065, "step": 12330 }, { "grad_norm": 0.14366213977336884, "learning_rate": 9.352140302692733e-05, "loss": 0.0069, "step": 12340 }, { "grad_norm": 0.1414911448955536, "learning_rate": 9.350782981774627e-05, "loss": 0.0078, "step": 12350 }, { "grad_norm": 0.12936703860759735, "learning_rate": 9.349424339203526e-05, "loss": 0.0071, "step": 12360 }, { "grad_norm": 0.15067929029464722, "learning_rate": 9.34806437539215e-05, "loss": 0.006, "step": 12370 }, { "grad_norm": 0.17526575922966003, "learning_rate": 9.346703090753622e-05, "loss": 0.0057, "step": 12380 }, { "grad_norm": 0.19499830901622772, "learning_rate": 9.345340485701461e-05, "loss": 0.0065, "step": 12390 }, { "grad_norm": 0.19008158147335052, "learning_rate": 9.343976560649595e-05, "loss": 0.0072, "step": 12400 }, { "grad_norm": 0.190085306763649, "learning_rate": 9.342611316012344e-05, "loss": 0.0063, "step": 12410 }, { "grad_norm": 0.21725183725357056, "learning_rate": 9.341244752204437e-05, "loss": 0.0074, "step": 12420 }, { "grad_norm": 0.1633007973432541, "learning_rate": 9.339876869640995e-05, "loss": 0.0084, "step": 12430 }, { "grad_norm": 0.12932221591472626, "learning_rate": 9.33850766873755e-05, "loss": 0.0064, "step": 12440 }, { "grad_norm": 0.16356875002384186, "learning_rate": 9.337137149910028e-05, "loss": 0.0058, "step": 12450 }, { "grad_norm": 0.15794123709201813, "learning_rate": 9.335765313574753e-05, "loss": 0.0051, "step": 12460 }, { "grad_norm": 0.15112634003162384, "learning_rate": 9.334392160148457e-05, "loss": 0.0073, "step": 12470 }, { "grad_norm": 0.17946302890777588, "learning_rate": 9.333017690048264e-05, "loss": 0.006, "step": 12480 }, { "grad_norm": 0.14079903066158295, "learning_rate": 9.331641903691706e-05, "loss": 0.009, "step": 12490 }, { "grad_norm": 0.19790972769260406, "learning_rate": 9.330264801496707e-05, "loss": 0.0069, "step": 12500 }, { "grad_norm": 0.17030636966228485, "learning_rate": 9.328886383881594e-05, "loss": 0.0093, "step": 12510 }, { "grad_norm": 0.14363013207912445, "learning_rate": 9.327506651265095e-05, "loss": 0.0065, "step": 12520 }, { "grad_norm": 0.1844875067472458, "learning_rate": 9.326125604066338e-05, "loss": 0.0073, "step": 12530 }, { "grad_norm": 0.19230280816555023, "learning_rate": 9.324743242704847e-05, "loss": 0.0058, "step": 12540 }, { "grad_norm": 0.11493001878261566, "learning_rate": 9.323359567600546e-05, "loss": 0.0063, "step": 12550 }, { "grad_norm": 0.27283307909965515, "learning_rate": 9.321974579173761e-05, "loss": 0.009, "step": 12560 }, { "grad_norm": 0.21113546192646027, "learning_rate": 9.320588277845213e-05, "loss": 0.0083, "step": 12570 }, { "grad_norm": 0.14717572927474976, "learning_rate": 9.319200664036026e-05, "loss": 0.0092, "step": 12580 }, { "grad_norm": 0.1940688192844391, "learning_rate": 9.31781173816772e-05, "loss": 0.0066, "step": 12590 }, { "grad_norm": 0.17842546105384827, "learning_rate": 9.316421500662212e-05, "loss": 0.0069, "step": 12600 }, { "grad_norm": 0.14822331070899963, "learning_rate": 9.31502995194182e-05, "loss": 0.0051, "step": 12610 }, { "grad_norm": 0.1503150910139084, "learning_rate": 9.31363709242926e-05, "loss": 0.0059, "step": 12620 }, { "grad_norm": 0.14200299978256226, "learning_rate": 9.312242922547647e-05, "loss": 0.0066, "step": 12630 }, { "grad_norm": 0.15459373593330383, "learning_rate": 9.310847442720492e-05, "loss": 0.0065, "step": 12640 }, { "grad_norm": 0.13565519452095032, "learning_rate": 9.309450653371706e-05, "loss": 0.0063, "step": 12650 }, { "grad_norm": 0.12833379209041595, "learning_rate": 9.308052554925595e-05, "loss": 0.0064, "step": 12660 }, { "grad_norm": 0.14641664922237396, "learning_rate": 9.306653147806867e-05, "loss": 0.0062, "step": 12670 }, { "grad_norm": 0.15507547557353973, "learning_rate": 9.305252432440622e-05, "loss": 0.006, "step": 12680 }, { "grad_norm": 0.19053632020950317, "learning_rate": 9.303850409252361e-05, "loss": 0.0079, "step": 12690 }, { "grad_norm": 0.14489203691482544, "learning_rate": 9.302447078667985e-05, "loss": 0.0061, "step": 12700 }, { "grad_norm": 0.1852709949016571, "learning_rate": 9.301042441113783e-05, "loss": 0.0066, "step": 12710 }, { "grad_norm": 0.17587006092071533, "learning_rate": 9.299636497016451e-05, "loss": 0.0092, "step": 12720 }, { "grad_norm": 0.18914154171943665, "learning_rate": 9.298229246803076e-05, "loss": 0.0083, "step": 12730 }, { "grad_norm": 0.14724212884902954, "learning_rate": 9.296820690901144e-05, "loss": 0.006, "step": 12740 }, { "grad_norm": 0.17917200922966003, "learning_rate": 9.295410829738539e-05, "loss": 0.0065, "step": 12750 }, { "grad_norm": 0.14907144010066986, "learning_rate": 9.293999663743535e-05, "loss": 0.008, "step": 12760 }, { "grad_norm": 0.1564476490020752, "learning_rate": 9.292587193344813e-05, "loss": 0.0074, "step": 12770 }, { "grad_norm": 0.18981020152568817, "learning_rate": 9.291173418971437e-05, "loss": 0.0065, "step": 12780 }, { "grad_norm": 0.1617121696472168, "learning_rate": 9.28975834105288e-05, "loss": 0.0074, "step": 12790 }, { "grad_norm": 0.14393509924411774, "learning_rate": 9.288341960019004e-05, "loss": 0.0074, "step": 12800 }, { "grad_norm": 0.16383635997772217, "learning_rate": 9.286924276300067e-05, "loss": 0.0062, "step": 12810 }, { "grad_norm": 0.17014449834823608, "learning_rate": 9.285505290326726e-05, "loss": 0.008, "step": 12820 }, { "grad_norm": 0.13042590022087097, "learning_rate": 9.284085002530027e-05, "loss": 0.0074, "step": 12830 }, { "grad_norm": 0.12608250975608826, "learning_rate": 9.282663413341422e-05, "loss": 0.0064, "step": 12840 }, { "grad_norm": 0.13917812705039978, "learning_rate": 9.281240523192747e-05, "loss": 0.0064, "step": 12850 }, { "grad_norm": 0.15864966809749603, "learning_rate": 9.279816332516242e-05, "loss": 0.0065, "step": 12860 }, { "grad_norm": 0.16618312895298004, "learning_rate": 9.278390841744536e-05, "loss": 0.0067, "step": 12870 }, { "grad_norm": 0.10801320523023605, "learning_rate": 9.276964051310658e-05, "loss": 0.0055, "step": 12880 }, { "grad_norm": 0.1622803658246994, "learning_rate": 9.275535961648027e-05, "loss": 0.0062, "step": 12890 }, { "grad_norm": 0.13434474170207977, "learning_rate": 9.274106573190459e-05, "loss": 0.0056, "step": 12900 }, { "grad_norm": 0.17306289076805115, "learning_rate": 9.272675886372168e-05, "loss": 0.0055, "step": 12910 }, { "grad_norm": 0.13534604012966156, "learning_rate": 9.271243901627754e-05, "loss": 0.0072, "step": 12920 }, { "grad_norm": 0.19058699905872345, "learning_rate": 9.269810619392219e-05, "loss": 0.0087, "step": 12930 }, { "grad_norm": 0.14378350973129272, "learning_rate": 9.268376040100955e-05, "loss": 0.0071, "step": 12940 }, { "grad_norm": 0.1442684680223465, "learning_rate": 9.266940164189752e-05, "loss": 0.0075, "step": 12950 }, { "grad_norm": 0.14287574589252472, "learning_rate": 9.265502992094787e-05, "loss": 0.0063, "step": 12960 }, { "grad_norm": 0.1520426720380783, "learning_rate": 9.264064524252638e-05, "loss": 0.0061, "step": 12970 }, { "grad_norm": 0.1883792281150818, "learning_rate": 9.262624761100271e-05, "loss": 0.0067, "step": 12980 }, { "grad_norm": 0.12607556581497192, "learning_rate": 9.261183703075051e-05, "loss": 0.0061, "step": 12990 }, { "grad_norm": 0.13036291301250458, "learning_rate": 9.259741350614733e-05, "loss": 0.0077, "step": 13000 }, { "grad_norm": 0.13220156729221344, "learning_rate": 9.258297704157464e-05, "loss": 0.007, "step": 13010 }, { "grad_norm": 0.13834074139595032, "learning_rate": 9.256852764141786e-05, "loss": 0.007, "step": 13020 }, { "grad_norm": 0.1452990025281906, "learning_rate": 9.255406531006634e-05, "loss": 0.007, "step": 13030 }, { "grad_norm": 0.1996668428182602, "learning_rate": 9.253959005191335e-05, "loss": 0.0066, "step": 13040 }, { "grad_norm": 0.1793651580810547, "learning_rate": 9.25251018713561e-05, "loss": 0.0071, "step": 13050 }, { "grad_norm": 0.18872542679309845, "learning_rate": 9.251060077279571e-05, "loss": 0.0063, "step": 13060 }, { "grad_norm": 0.15713800489902496, "learning_rate": 9.249608676063724e-05, "loss": 0.0053, "step": 13070 }, { "grad_norm": 0.18579256534576416, "learning_rate": 9.248155983928964e-05, "loss": 0.0067, "step": 13080 }, { "grad_norm": 0.1676463782787323, "learning_rate": 9.246702001316583e-05, "loss": 0.0067, "step": 13090 }, { "grad_norm": 0.10738570988178253, "learning_rate": 9.245246728668262e-05, "loss": 0.0064, "step": 13100 }, { "grad_norm": 0.118846096098423, "learning_rate": 9.243790166426073e-05, "loss": 0.0065, "step": 13110 }, { "grad_norm": 0.12867479026317596, "learning_rate": 9.242332315032484e-05, "loss": 0.0063, "step": 13120 }, { "grad_norm": 0.13289783895015717, "learning_rate": 9.240873174930349e-05, "loss": 0.007, "step": 13130 }, { "grad_norm": 0.17166031897068024, "learning_rate": 9.239412746562917e-05, "loss": 0.0063, "step": 13140 }, { "grad_norm": 0.15162891149520874, "learning_rate": 9.237951030373828e-05, "loss": 0.007, "step": 13150 }, { "grad_norm": 0.13084948062896729, "learning_rate": 9.236488026807113e-05, "loss": 0.0066, "step": 13160 }, { "grad_norm": 0.13682521879673004, "learning_rate": 9.235023736307193e-05, "loss": 0.0071, "step": 13170 }, { "grad_norm": 0.15575428307056427, "learning_rate": 9.233558159318881e-05, "loss": 0.0062, "step": 13180 }, { "grad_norm": 0.16910164058208466, "learning_rate": 9.232091296287382e-05, "loss": 0.0073, "step": 13190 }, { "grad_norm": 0.1534806787967682, "learning_rate": 9.230623147658288e-05, "loss": 0.0056, "step": 13200 }, { "grad_norm": 0.19949296116828918, "learning_rate": 9.229153713877586e-05, "loss": 0.007, "step": 13210 }, { "grad_norm": 0.13838230073451996, "learning_rate": 9.227682995391649e-05, "loss": 0.0066, "step": 13220 }, { "grad_norm": 0.15017127990722656, "learning_rate": 9.226210992647243e-05, "loss": 0.007, "step": 13230 }, { "grad_norm": 0.1847439706325531, "learning_rate": 9.224737706091525e-05, "loss": 0.0054, "step": 13240 }, { "grad_norm": 0.1644303798675537, "learning_rate": 9.223263136172039e-05, "loss": 0.0059, "step": 13250 }, { "grad_norm": 0.1719895601272583, "learning_rate": 9.22178728333672e-05, "loss": 0.0072, "step": 13260 }, { "grad_norm": 0.13211627304553986, "learning_rate": 9.220310148033897e-05, "loss": 0.0068, "step": 13270 }, { "grad_norm": 0.25040245056152344, "learning_rate": 9.21883173071228e-05, "loss": 0.0071, "step": 13280 }, { "grad_norm": 0.15476050972938538, "learning_rate": 9.217352031820976e-05, "loss": 0.008, "step": 13290 }, { "grad_norm": 0.17415720224380493, "learning_rate": 9.215871051809477e-05, "loss": 0.0077, "step": 13300 }, { "grad_norm": 0.13439618051052094, "learning_rate": 9.214388791127666e-05, "loss": 0.0063, "step": 13310 }, { "grad_norm": 0.16827784478664398, "learning_rate": 9.212905250225814e-05, "loss": 0.0083, "step": 13320 }, { "grad_norm": 0.12802889943122864, "learning_rate": 9.211420429554583e-05, "loss": 0.0071, "step": 13330 }, { "grad_norm": 0.13354554772377014, "learning_rate": 9.209934329565022e-05, "loss": 0.0065, "step": 13340 }, { "grad_norm": 0.16400596499443054, "learning_rate": 9.208446950708568e-05, "loss": 0.0063, "step": 13350 }, { "grad_norm": 0.15052418410778046, "learning_rate": 9.20695829343705e-05, "loss": 0.0074, "step": 13360 }, { "grad_norm": 0.12180141359567642, "learning_rate": 9.205468358202678e-05, "loss": 0.0067, "step": 13370 }, { "grad_norm": 0.16285336017608643, "learning_rate": 9.203977145458059e-05, "loss": 0.0067, "step": 13380 }, { "grad_norm": 0.26275634765625, "learning_rate": 9.202484655656182e-05, "loss": 0.0068, "step": 13390 }, { "grad_norm": 0.20062899589538574, "learning_rate": 9.200990889250427e-05, "loss": 0.0074, "step": 13400 }, { "grad_norm": 0.13696284592151642, "learning_rate": 9.19949584669456e-05, "loss": 0.0073, "step": 13410 }, { "grad_norm": 0.15921203792095184, "learning_rate": 9.197999528442738e-05, "loss": 0.007, "step": 13420 }, { "grad_norm": 0.14847107231616974, "learning_rate": 9.196501934949499e-05, "loss": 0.0081, "step": 13430 }, { "grad_norm": 0.19049206376075745, "learning_rate": 9.195003066669776e-05, "loss": 0.0082, "step": 13440 }, { "grad_norm": 0.1747126281261444, "learning_rate": 9.193502924058884e-05, "loss": 0.006, "step": 13450 }, { "grad_norm": 0.12898847460746765, "learning_rate": 9.192001507572526e-05, "loss": 0.0064, "step": 13460 }, { "grad_norm": 0.15510623157024384, "learning_rate": 9.190498817666793e-05, "loss": 0.0073, "step": 13470 }, { "grad_norm": 0.15048405528068542, "learning_rate": 9.188994854798163e-05, "loss": 0.0071, "step": 13480 }, { "grad_norm": 0.14909416437149048, "learning_rate": 9.187489619423499e-05, "loss": 0.0063, "step": 13490 }, { "grad_norm": 0.14245004951953888, "learning_rate": 9.185983112000056e-05, "loss": 0.0072, "step": 13500 }, { "grad_norm": 0.14598448574543, "learning_rate": 9.184475332985464e-05, "loss": 0.0072, "step": 13510 }, { "grad_norm": 0.1324056088924408, "learning_rate": 9.182966282837754e-05, "loss": 0.005, "step": 13520 }, { "grad_norm": 0.0880085825920105, "learning_rate": 9.18145596201533e-05, "loss": 0.0053, "step": 13530 }, { "grad_norm": 0.1522257775068283, "learning_rate": 9.179944370976991e-05, "loss": 0.0068, "step": 13540 }, { "grad_norm": 0.15083026885986328, "learning_rate": 9.178431510181918e-05, "loss": 0.0088, "step": 13550 }, { "grad_norm": 0.1425086408853531, "learning_rate": 9.176917380089675e-05, "loss": 0.0074, "step": 13560 }, { "grad_norm": 0.15115949511528015, "learning_rate": 9.175401981160219e-05, "loss": 0.0068, "step": 13570 }, { "grad_norm": 0.20139022171497345, "learning_rate": 9.173885313853885e-05, "loss": 0.0065, "step": 13580 }, { "grad_norm": 0.20766586065292358, "learning_rate": 9.172367378631398e-05, "loss": 0.0079, "step": 13590 }, { "grad_norm": 0.14884953200817108, "learning_rate": 9.170848175953866e-05, "loss": 0.0073, "step": 13600 }, { "grad_norm": 0.21074515581130981, "learning_rate": 9.169327706282784e-05, "loss": 0.0062, "step": 13610 }, { "grad_norm": 0.15902255475521088, "learning_rate": 9.167805970080029e-05, "loss": 0.0064, "step": 13620 }, { "grad_norm": 0.18614886701107025, "learning_rate": 9.166282967807864e-05, "loss": 0.0064, "step": 13630 }, { "grad_norm": 0.15012024343013763, "learning_rate": 9.16475869992894e-05, "loss": 0.0058, "step": 13640 }, { "grad_norm": 0.1300547569990158, "learning_rate": 9.163233166906284e-05, "loss": 0.0063, "step": 13650 }, { "grad_norm": 0.14613236486911774, "learning_rate": 9.161706369203317e-05, "loss": 0.0055, "step": 13660 }, { "grad_norm": 0.15966565907001495, "learning_rate": 9.16017830728384e-05, "loss": 0.0058, "step": 13670 }, { "grad_norm": 0.1304013431072235, "learning_rate": 9.158648981612035e-05, "loss": 0.0073, "step": 13680 }, { "grad_norm": 0.11670041084289551, "learning_rate": 9.157118392652472e-05, "loss": 0.0064, "step": 13690 }, { "grad_norm": 0.1004563644528389, "learning_rate": 9.155586540870104e-05, "loss": 0.0067, "step": 13700 }, { "grad_norm": 0.15330870449543, "learning_rate": 9.154053426730267e-05, "loss": 0.0067, "step": 13710 }, { "grad_norm": 0.14933964610099792, "learning_rate": 9.15251905069868e-05, "loss": 0.0057, "step": 13720 }, { "grad_norm": 0.15307804942131042, "learning_rate": 9.150983413241446e-05, "loss": 0.0081, "step": 13730 }, { "grad_norm": 0.14567573368549347, "learning_rate": 9.149446514825051e-05, "loss": 0.0055, "step": 13740 }, { "grad_norm": 0.1360456496477127, "learning_rate": 9.147908355916365e-05, "loss": 0.0072, "step": 13750 }, { "grad_norm": 0.11761092394590378, "learning_rate": 9.146368936982642e-05, "loss": 0.0061, "step": 13760 }, { "grad_norm": 0.16178421676158905, "learning_rate": 9.144828258491511e-05, "loss": 0.0053, "step": 13770 }, { "grad_norm": 0.1387917697429657, "learning_rate": 9.143286320910996e-05, "loss": 0.005, "step": 13780 }, { "grad_norm": 0.14660824835300446, "learning_rate": 9.141743124709491e-05, "loss": 0.0059, "step": 13790 }, { "grad_norm": 0.1863843947649002, "learning_rate": 9.140198670355784e-05, "loss": 0.0055, "step": 13800 }, { "grad_norm": 0.13181474804878235, "learning_rate": 9.138652958319034e-05, "loss": 0.0053, "step": 13810 }, { "grad_norm": 0.2531883418560028, "learning_rate": 9.137105989068791e-05, "loss": 0.0069, "step": 13820 }, { "grad_norm": 0.12418678402900696, "learning_rate": 9.135557763074983e-05, "loss": 0.0067, "step": 13830 }, { "grad_norm": 0.12571033835411072, "learning_rate": 9.13400828080792e-05, "loss": 0.0062, "step": 13840 }, { "grad_norm": 0.15710081160068512, "learning_rate": 9.132457542738292e-05, "loss": 0.0051, "step": 13850 }, { "grad_norm": 0.12865744531154633, "learning_rate": 9.130905549337174e-05, "loss": 0.0049, "step": 13860 }, { "grad_norm": 0.12492245435714722, "learning_rate": 9.129352301076021e-05, "loss": 0.0069, "step": 13870 }, { "grad_norm": 0.19010740518569946, "learning_rate": 9.127797798426668e-05, "loss": 0.0067, "step": 13880 }, { "grad_norm": 0.12891830503940582, "learning_rate": 9.126242041861333e-05, "loss": 0.0062, "step": 13890 }, { "grad_norm": 0.13775856792926788, "learning_rate": 9.124685031852611e-05, "loss": 0.0054, "step": 13900 }, { "grad_norm": 0.10697411000728607, "learning_rate": 9.123126768873482e-05, "loss": 0.0071, "step": 13910 }, { "grad_norm": 0.17851291596889496, "learning_rate": 9.121567253397308e-05, "loss": 0.006, "step": 13920 }, { "grad_norm": 0.11071033030748367, "learning_rate": 9.120006485897824e-05, "loss": 0.0052, "step": 13930 }, { "grad_norm": 0.16563686728477478, "learning_rate": 9.118444466849152e-05, "loss": 0.0055, "step": 13940 }, { "grad_norm": 0.16375784575939178, "learning_rate": 9.116881196725793e-05, "loss": 0.0065, "step": 13950 }, { "grad_norm": 0.14119267463684082, "learning_rate": 9.115316676002627e-05, "loss": 0.0066, "step": 13960 }, { "grad_norm": 0.14563946425914764, "learning_rate": 9.113750905154911e-05, "loss": 0.0079, "step": 13970 }, { "grad_norm": 0.14841607213020325, "learning_rate": 9.112183884658289e-05, "loss": 0.0068, "step": 13980 }, { "grad_norm": 0.19296756386756897, "learning_rate": 9.11061561498878e-05, "loss": 0.0065, "step": 13990 }, { "grad_norm": 0.13706962764263153, "learning_rate": 9.109046096622779e-05, "loss": 0.0053, "step": 14000 }, { "grad_norm": 0.17651259899139404, "learning_rate": 9.107475330037069e-05, "loss": 0.0075, "step": 14010 }, { "grad_norm": 0.14664499461650848, "learning_rate": 9.105903315708806e-05, "loss": 0.0074, "step": 14020 }, { "grad_norm": 0.10691408067941666, "learning_rate": 9.104330054115524e-05, "loss": 0.0063, "step": 14030 }, { "grad_norm": 0.16464382410049438, "learning_rate": 9.102755545735141e-05, "loss": 0.0053, "step": 14040 }, { "grad_norm": 0.14871208369731903, "learning_rate": 9.10117979104595e-05, "loss": 0.0074, "step": 14050 }, { "grad_norm": 0.11644314974546432, "learning_rate": 9.099602790526624e-05, "loss": 0.0057, "step": 14060 }, { "grad_norm": 0.1068727970123291, "learning_rate": 9.098024544656212e-05, "loss": 0.0044, "step": 14070 }, { "grad_norm": 0.13462242484092712, "learning_rate": 9.096445053914148e-05, "loss": 0.0051, "step": 14080 }, { "grad_norm": 0.11795313656330109, "learning_rate": 9.094864318780236e-05, "loss": 0.0051, "step": 14090 }, { "grad_norm": 0.17095260322093964, "learning_rate": 9.093282339734663e-05, "loss": 0.0052, "step": 14100 }, { "grad_norm": 0.1977066844701767, "learning_rate": 9.091699117257992e-05, "loss": 0.0058, "step": 14110 }, { "grad_norm": 0.11390426754951477, "learning_rate": 9.090114651831163e-05, "loss": 0.0064, "step": 14120 }, { "grad_norm": 0.14341358840465546, "learning_rate": 9.088528943935497e-05, "loss": 0.0062, "step": 14130 }, { "grad_norm": 0.1912917196750641, "learning_rate": 9.086941994052689e-05, "loss": 0.0062, "step": 14140 }, { "grad_norm": 0.15952086448669434, "learning_rate": 9.085353802664813e-05, "loss": 0.0056, "step": 14150 }, { "grad_norm": 0.13711123168468475, "learning_rate": 9.08376437025432e-05, "loss": 0.0059, "step": 14160 }, { "grad_norm": 0.15328150987625122, "learning_rate": 9.082173697304035e-05, "loss": 0.0062, "step": 14170 }, { "grad_norm": 0.16459782421588898, "learning_rate": 9.080581784297166e-05, "loss": 0.0079, "step": 14180 }, { "grad_norm": 0.1141209602355957, "learning_rate": 9.078988631717291e-05, "loss": 0.0111, "step": 14190 }, { "grad_norm": 0.12241442501544952, "learning_rate": 9.077394240048369e-05, "loss": 0.0082, "step": 14200 }, { "grad_norm": 0.1487772911787033, "learning_rate": 9.075798609774736e-05, "loss": 0.009, "step": 14210 }, { "grad_norm": 0.13853006064891815, "learning_rate": 9.0742017413811e-05, "loss": 0.0085, "step": 14220 }, { "grad_norm": 0.1867283135652542, "learning_rate": 9.072603635352548e-05, "loss": 0.0087, "step": 14230 }, { "grad_norm": 0.17566253244876862, "learning_rate": 9.071004292174541e-05, "loss": 0.0072, "step": 14240 }, { "grad_norm": 0.20353031158447266, "learning_rate": 9.06940371233292e-05, "loss": 0.0079, "step": 14250 }, { "grad_norm": 0.16296055912971497, "learning_rate": 9.067801896313898e-05, "loss": 0.0069, "step": 14260 }, { "grad_norm": 0.20336081087589264, "learning_rate": 9.066198844604064e-05, "loss": 0.0072, "step": 14270 }, { "grad_norm": 0.12117595225572586, "learning_rate": 9.06459455769038e-05, "loss": 0.0062, "step": 14280 }, { "grad_norm": 0.08882665634155273, "learning_rate": 9.062989036060193e-05, "loss": 0.0054, "step": 14290 }, { "grad_norm": 0.13228408992290497, "learning_rate": 9.061382280201212e-05, "loss": 0.0059, "step": 14300 }, { "grad_norm": 0.16228187084197998, "learning_rate": 9.059774290601528e-05, "loss": 0.0076, "step": 14310 }, { "grad_norm": 0.14325237274169922, "learning_rate": 9.058165067749606e-05, "loss": 0.0066, "step": 14320 }, { "grad_norm": 0.12749585509300232, "learning_rate": 9.056554612134288e-05, "loss": 0.0054, "step": 14330 }, { "grad_norm": 0.14954228699207306, "learning_rate": 9.054942924244785e-05, "loss": 0.0065, "step": 14340 }, { "grad_norm": 0.12507501244544983, "learning_rate": 9.053330004570686e-05, "loss": 0.0058, "step": 14350 }, { "grad_norm": 0.10446546971797943, "learning_rate": 9.051715853601955e-05, "loss": 0.0055, "step": 14360 }, { "grad_norm": 0.14102093875408173, "learning_rate": 9.050100471828926e-05, "loss": 0.0056, "step": 14370 }, { "grad_norm": 0.15639817714691162, "learning_rate": 9.048483859742311e-05, "loss": 0.0056, "step": 14380 }, { "grad_norm": 0.17347560822963715, "learning_rate": 9.046866017833193e-05, "loss": 0.0066, "step": 14390 }, { "grad_norm": 0.15305174887180328, "learning_rate": 9.045246946593029e-05, "loss": 0.0056, "step": 14400 }, { "grad_norm": 0.1426062136888504, "learning_rate": 9.043626646513652e-05, "loss": 0.0048, "step": 14410 }, { "grad_norm": 0.10367455333471298, "learning_rate": 9.042005118087267e-05, "loss": 0.0046, "step": 14420 }, { "grad_norm": 0.17896917462348938, "learning_rate": 9.040382361806448e-05, "loss": 0.0061, "step": 14430 }, { "grad_norm": 0.1355636715888977, "learning_rate": 9.038758378164148e-05, "loss": 0.0076, "step": 14440 }, { "grad_norm": 0.15587522089481354, "learning_rate": 9.037133167653691e-05, "loss": 0.0063, "step": 14450 }, { "grad_norm": 0.14404824376106262, "learning_rate": 9.035506730768771e-05, "loss": 0.0061, "step": 14460 }, { "grad_norm": 0.12648530304431915, "learning_rate": 9.033879068003458e-05, "loss": 0.0055, "step": 14470 }, { "grad_norm": 0.13159193098545074, "learning_rate": 9.032250179852193e-05, "loss": 0.0045, "step": 14480 }, { "grad_norm": 0.1271722912788391, "learning_rate": 9.030620066809787e-05, "loss": 0.0055, "step": 14490 }, { "grad_norm": 0.1795722395181656, "learning_rate": 9.028988729371428e-05, "loss": 0.0073, "step": 14500 }, { "grad_norm": 0.16909003257751465, "learning_rate": 9.027356168032673e-05, "loss": 0.0066, "step": 14510 }, { "grad_norm": 0.20758114755153656, "learning_rate": 9.02572238328945e-05, "loss": 0.0061, "step": 14520 }, { "grad_norm": 0.14295081794261932, "learning_rate": 9.02408737563806e-05, "loss": 0.0061, "step": 14530 }, { "grad_norm": 0.13836334645748138, "learning_rate": 9.022451145575174e-05, "loss": 0.0052, "step": 14540 }, { "grad_norm": 0.1225593313574791, "learning_rate": 9.02081369359784e-05, "loss": 0.0066, "step": 14550 }, { "grad_norm": 0.12210579961538315, "learning_rate": 9.019175020203465e-05, "loss": 0.0069, "step": 14560 }, { "grad_norm": 0.13558083772659302, "learning_rate": 9.017535125889842e-05, "loss": 0.0053, "step": 14570 }, { "grad_norm": 0.1620301604270935, "learning_rate": 9.015894011155124e-05, "loss": 0.0062, "step": 14580 }, { "grad_norm": 0.12315794080495834, "learning_rate": 9.014251676497838e-05, "loss": 0.0058, "step": 14590 }, { "grad_norm": 0.19408975541591644, "learning_rate": 9.012608122416884e-05, "loss": 0.0057, "step": 14600 }, { "grad_norm": 0.16712318360805511, "learning_rate": 9.010963349411529e-05, "loss": 0.0065, "step": 14610 }, { "grad_norm": 0.13333940505981445, "learning_rate": 9.00931735798141e-05, "loss": 0.0054, "step": 14620 }, { "grad_norm": 0.13167326152324677, "learning_rate": 9.00767014862654e-05, "loss": 0.0057, "step": 14630 }, { "grad_norm": 0.14591127634048462, "learning_rate": 9.006021721847295e-05, "loss": 0.0059, "step": 14640 }, { "grad_norm": 0.15092866122722626, "learning_rate": 9.004372078144423e-05, "loss": 0.0054, "step": 14650 }, { "grad_norm": 0.14945967495441437, "learning_rate": 9.002721218019043e-05, "loss": 0.0056, "step": 14660 }, { "grad_norm": 0.15255342423915863, "learning_rate": 9.001069141972642e-05, "loss": 0.006, "step": 14670 }, { "grad_norm": 0.11461211740970612, "learning_rate": 8.99941585050708e-05, "loss": 0.0055, "step": 14680 }, { "grad_norm": 0.13330726325511932, "learning_rate": 8.997761344124578e-05, "loss": 0.0053, "step": 14690 }, { "grad_norm": 0.1805570125579834, "learning_rate": 8.996105623327737e-05, "loss": 0.0068, "step": 14700 }, { "grad_norm": 0.127656489610672, "learning_rate": 8.994448688619517e-05, "loss": 0.006, "step": 14710 }, { "grad_norm": 0.12109510600566864, "learning_rate": 8.992790540503253e-05, "loss": 0.0057, "step": 14720 }, { "grad_norm": 0.1755305528640747, "learning_rate": 8.991131179482648e-05, "loss": 0.0067, "step": 14730 }, { "grad_norm": 0.11689936369657516, "learning_rate": 8.989470606061768e-05, "loss": 0.0056, "step": 14740 }, { "grad_norm": 0.13286441564559937, "learning_rate": 8.987808820745056e-05, "loss": 0.0052, "step": 14750 }, { "grad_norm": 0.15769051015377045, "learning_rate": 8.986145824037315e-05, "loss": 0.0071, "step": 14760 }, { "grad_norm": 0.15926207602024078, "learning_rate": 8.984481616443721e-05, "loss": 0.0056, "step": 14770 }, { "grad_norm": 0.15566225349903107, "learning_rate": 8.982816198469815e-05, "loss": 0.0058, "step": 14780 }, { "grad_norm": 0.15580399334430695, "learning_rate": 8.98114957062151e-05, "loss": 0.0058, "step": 14790 }, { "grad_norm": 0.11278016865253448, "learning_rate": 8.97948173340508e-05, "loss": 0.0071, "step": 14800 }, { "grad_norm": 0.1571248322725296, "learning_rate": 8.977812687327172e-05, "loss": 0.0053, "step": 14810 }, { "grad_norm": 0.1398119479417801, "learning_rate": 8.976142432894798e-05, "loss": 0.0058, "step": 14820 }, { "grad_norm": 0.11984849721193314, "learning_rate": 8.974470970615336e-05, "loss": 0.0063, "step": 14830 }, { "grad_norm": 0.1856822818517685, "learning_rate": 8.972798300996534e-05, "loss": 0.0061, "step": 14840 }, { "grad_norm": 0.13227061927318573, "learning_rate": 8.971124424546504e-05, "loss": 0.0059, "step": 14850 }, { "grad_norm": 0.11837036162614822, "learning_rate": 8.969449341773724e-05, "loss": 0.006, "step": 14860 }, { "grad_norm": 0.12632200121879578, "learning_rate": 8.967773053187042e-05, "loss": 0.006, "step": 14870 }, { "grad_norm": 0.09961303323507309, "learning_rate": 8.966095559295668e-05, "loss": 0.006, "step": 14880 }, { "grad_norm": 0.15789806842803955, "learning_rate": 8.964416860609184e-05, "loss": 0.0056, "step": 14890 }, { "grad_norm": 0.1709860861301422, "learning_rate": 8.962736957637532e-05, "loss": 0.0062, "step": 14900 }, { "grad_norm": 0.21235159039497375, "learning_rate": 8.96105585089102e-05, "loss": 0.0063, "step": 14910 }, { "grad_norm": 0.1555054783821106, "learning_rate": 8.959373540880329e-05, "loss": 0.0068, "step": 14920 }, { "grad_norm": 0.14368557929992676, "learning_rate": 8.957690028116495e-05, "loss": 0.007, "step": 14930 }, { "grad_norm": 0.09485908597707748, "learning_rate": 8.956005313110928e-05, "loss": 0.0056, "step": 14940 }, { "grad_norm": 0.11007863283157349, "learning_rate": 8.9543193963754e-05, "loss": 0.0051, "step": 14950 }, { "grad_norm": 0.1142236590385437, "learning_rate": 8.952632278422048e-05, "loss": 0.0052, "step": 14960 }, { "grad_norm": 0.14819014072418213, "learning_rate": 8.95094395976337e-05, "loss": 0.0058, "step": 14970 }, { "grad_norm": 0.17635884881019592, "learning_rate": 8.949254440912239e-05, "loss": 0.0068, "step": 14980 }, { "grad_norm": 0.17253822088241577, "learning_rate": 8.94756372238188e-05, "loss": 0.0058, "step": 14990 }, { "grad_norm": 0.1334889531135559, "learning_rate": 8.945871804685892e-05, "loss": 0.0057, "step": 15000 }, { "grad_norm": 0.13638600707054138, "learning_rate": 8.944178688338236e-05, "loss": 0.0073, "step": 15010 }, { "grad_norm": 0.1429094672203064, "learning_rate": 8.942484373853233e-05, "loss": 0.007, "step": 15020 }, { "grad_norm": 0.14473600685596466, "learning_rate": 8.940788861745572e-05, "loss": 0.0058, "step": 15030 }, { "grad_norm": 0.0918731540441513, "learning_rate": 8.939092152530308e-05, "loss": 0.0054, "step": 15040 }, { "grad_norm": 0.14249549806118011, "learning_rate": 8.937394246722853e-05, "loss": 0.0065, "step": 15050 }, { "grad_norm": 0.16728469729423523, "learning_rate": 8.935695144838984e-05, "loss": 0.0065, "step": 15060 }, { "grad_norm": 0.13563118875026703, "learning_rate": 8.933994847394849e-05, "loss": 0.0074, "step": 15070 }, { "grad_norm": 0.16576212644577026, "learning_rate": 8.932293354906949e-05, "loss": 0.0079, "step": 15080 }, { "grad_norm": 0.13486281037330627, "learning_rate": 8.930590667892153e-05, "loss": 0.0082, "step": 15090 }, { "grad_norm": 0.13199132680892944, "learning_rate": 8.928886786867696e-05, "loss": 0.0074, "step": 15100 }, { "grad_norm": 0.25562813878059387, "learning_rate": 8.927181712351168e-05, "loss": 0.0077, "step": 15110 }, { "grad_norm": 0.15852156281471252, "learning_rate": 8.925475444860527e-05, "loss": 0.0076, "step": 15120 }, { "grad_norm": 0.10791701823472977, "learning_rate": 8.923767984914092e-05, "loss": 0.0063, "step": 15130 }, { "grad_norm": 0.17528291046619415, "learning_rate": 8.922059333030545e-05, "loss": 0.0074, "step": 15140 }, { "grad_norm": 0.11638785898685455, "learning_rate": 8.920349489728928e-05, "loss": 0.0052, "step": 15150 }, { "grad_norm": 0.12781599164009094, "learning_rate": 8.918638455528646e-05, "loss": 0.0058, "step": 15160 }, { "grad_norm": 0.14162227511405945, "learning_rate": 8.916926230949468e-05, "loss": 0.0061, "step": 15170 }, { "grad_norm": 0.11936146020889282, "learning_rate": 8.915212816511522e-05, "loss": 0.0052, "step": 15180 }, { "grad_norm": 0.1259629726409912, "learning_rate": 8.913498212735296e-05, "loss": 0.0071, "step": 15190 }, { "grad_norm": 0.1405264139175415, "learning_rate": 8.911782420141643e-05, "loss": 0.0066, "step": 15200 }, { "grad_norm": 0.12354318797588348, "learning_rate": 8.910065439251775e-05, "loss": 0.0057, "step": 15210 }, { "grad_norm": 0.13771268725395203, "learning_rate": 8.908347270587268e-05, "loss": 0.0054, "step": 15220 }, { "grad_norm": 0.16689176857471466, "learning_rate": 8.906627914670054e-05, "loss": 0.0061, "step": 15230 }, { "grad_norm": 0.14972850680351257, "learning_rate": 8.904907372022427e-05, "loss": 0.0057, "step": 15240 }, { "grad_norm": 0.1730159968137741, "learning_rate": 8.903185643167042e-05, "loss": 0.0064, "step": 15250 }, { "grad_norm": 0.11792197078466415, "learning_rate": 8.901462728626919e-05, "loss": 0.0058, "step": 15260 }, { "grad_norm": 0.15520115196704865, "learning_rate": 8.899738628925429e-05, "loss": 0.0067, "step": 15270 }, { "grad_norm": 0.15341272950172424, "learning_rate": 8.898013344586312e-05, "loss": 0.0063, "step": 15280 }, { "grad_norm": 0.10494568943977356, "learning_rate": 8.896286876133661e-05, "loss": 0.0067, "step": 15290 }, { "grad_norm": 0.1477355659008026, "learning_rate": 8.894559224091933e-05, "loss": 0.0077, "step": 15300 }, { "grad_norm": 0.1268368512392044, "learning_rate": 8.892830388985942e-05, "loss": 0.0059, "step": 15310 }, { "grad_norm": 0.14344362914562225, "learning_rate": 8.891100371340864e-05, "loss": 0.0043, "step": 15320 }, { "grad_norm": 0.13896867632865906, "learning_rate": 8.889369171682231e-05, "loss": 0.0064, "step": 15330 }, { "grad_norm": 0.14290951192378998, "learning_rate": 8.887636790535936e-05, "loss": 0.0054, "step": 15340 }, { "grad_norm": 0.12396464496850967, "learning_rate": 8.885903228428231e-05, "loss": 0.0053, "step": 15350 }, { "grad_norm": 0.1225813627243042, "learning_rate": 8.884168485885727e-05, "loss": 0.0058, "step": 15360 }, { "grad_norm": 0.1075747087597847, "learning_rate": 8.882432563435393e-05, "loss": 0.0059, "step": 15370 }, { "grad_norm": 0.13950416445732117, "learning_rate": 8.880695461604556e-05, "loss": 0.0049, "step": 15380 }, { "grad_norm": 0.14975054562091827, "learning_rate": 8.878957180920901e-05, "loss": 0.0072, "step": 15390 }, { "grad_norm": 0.11846451461315155, "learning_rate": 8.877217721912473e-05, "loss": 0.0069, "step": 15400 }, { "grad_norm": 0.14029531180858612, "learning_rate": 8.875477085107673e-05, "loss": 0.0051, "step": 15410 }, { "grad_norm": 0.16113963723182678, "learning_rate": 8.87373527103526e-05, "loss": 0.0085, "step": 15420 }, { "grad_norm": 0.15506118535995483, "learning_rate": 8.871992280224353e-05, "loss": 0.0063, "step": 15430 }, { "grad_norm": 0.13782933354377747, "learning_rate": 8.870248113204422e-05, "loss": 0.0085, "step": 15440 }, { "grad_norm": 0.1671835035085678, "learning_rate": 8.868502770505306e-05, "loss": 0.0057, "step": 15450 }, { "grad_norm": 0.1591596007347107, "learning_rate": 8.86675625265719e-05, "loss": 0.0057, "step": 15460 }, { "grad_norm": 0.1707792431116104, "learning_rate": 8.865008560190618e-05, "loss": 0.0046, "step": 15470 }, { "grad_norm": 0.21535485982894897, "learning_rate": 8.863259693636496e-05, "loss": 0.0076, "step": 15480 }, { "grad_norm": 0.16673964262008667, "learning_rate": 8.861509653526083e-05, "loss": 0.0065, "step": 15490 }, { "grad_norm": 0.11849179118871689, "learning_rate": 8.859758440390993e-05, "loss": 0.0082, "step": 15500 }, { "grad_norm": 0.13664448261260986, "learning_rate": 8.858006054763202e-05, "loss": 0.0059, "step": 15510 }, { "grad_norm": 0.12706434726715088, "learning_rate": 8.856252497175035e-05, "loss": 0.0066, "step": 15520 }, { "grad_norm": 0.13235832750797272, "learning_rate": 8.854497768159178e-05, "loss": 0.0052, "step": 15530 }, { "grad_norm": 0.1321275681257248, "learning_rate": 8.852741868248671e-05, "loss": 0.0055, "step": 15540 }, { "grad_norm": 0.10625782608985901, "learning_rate": 8.85098479797691e-05, "loss": 0.0055, "step": 15550 }, { "grad_norm": 0.1244010329246521, "learning_rate": 8.849226557877646e-05, "loss": 0.0066, "step": 15560 }, { "grad_norm": 0.16954456269741058, "learning_rate": 8.84746714848499e-05, "loss": 0.0055, "step": 15570 }, { "grad_norm": 0.10287507623434067, "learning_rate": 8.845706570333397e-05, "loss": 0.0067, "step": 15580 }, { "grad_norm": 0.1316673904657364, "learning_rate": 8.84394482395769e-05, "loss": 0.0048, "step": 15590 }, { "grad_norm": 0.13689720630645752, "learning_rate": 8.842181909893038e-05, "loss": 0.005, "step": 15600 }, { "grad_norm": 0.13802318274974823, "learning_rate": 8.840417828674969e-05, "loss": 0.0046, "step": 15610 }, { "grad_norm": 0.13048425316810608, "learning_rate": 8.838652580839364e-05, "loss": 0.0054, "step": 15620 }, { "grad_norm": 0.10370530188083649, "learning_rate": 8.836886166922458e-05, "loss": 0.0043, "step": 15630 }, { "grad_norm": 0.13624714314937592, "learning_rate": 8.835118587460844e-05, "loss": 0.0047, "step": 15640 }, { "grad_norm": 0.15753823518753052, "learning_rate": 8.83334984299146e-05, "loss": 0.0055, "step": 15650 }, { "grad_norm": 0.16052605211734772, "learning_rate": 8.83157993405161e-05, "loss": 0.0059, "step": 15660 }, { "grad_norm": 0.1414131224155426, "learning_rate": 8.829808861178943e-05, "loss": 0.005, "step": 15670 }, { "grad_norm": 0.15115085244178772, "learning_rate": 8.828036624911464e-05, "loss": 0.0053, "step": 15680 }, { "grad_norm": 0.11776499450206757, "learning_rate": 8.826263225787532e-05, "loss": 0.0047, "step": 15690 }, { "grad_norm": 0.1139160618185997, "learning_rate": 8.824488664345858e-05, "loss": 0.0054, "step": 15700 }, { "grad_norm": 0.11722808331251144, "learning_rate": 8.822712941125508e-05, "loss": 0.0066, "step": 15710 }, { "grad_norm": 0.07599526643753052, "learning_rate": 8.820936056665898e-05, "loss": 0.0047, "step": 15720 }, { "grad_norm": 0.10804594308137894, "learning_rate": 8.819158011506801e-05, "loss": 0.0056, "step": 15730 }, { "grad_norm": 0.10170990973711014, "learning_rate": 8.81737880618834e-05, "loss": 0.005, "step": 15740 }, { "grad_norm": 0.15340791642665863, "learning_rate": 8.815598441250987e-05, "loss": 0.0062, "step": 15750 }, { "grad_norm": 0.16406160593032837, "learning_rate": 8.813816917235576e-05, "loss": 0.0052, "step": 15760 }, { "grad_norm": 0.1500580608844757, "learning_rate": 8.812034234683282e-05, "loss": 0.0059, "step": 15770 }, { "grad_norm": 0.1229170635342598, "learning_rate": 8.810250394135637e-05, "loss": 0.0054, "step": 15780 }, { "grad_norm": 0.1681613028049469, "learning_rate": 8.808465396134529e-05, "loss": 0.0061, "step": 15790 }, { "grad_norm": 0.1456782966852188, "learning_rate": 8.806679241222189e-05, "loss": 0.0068, "step": 15800 }, { "grad_norm": 0.15481068193912506, "learning_rate": 8.804891929941203e-05, "loss": 0.0056, "step": 15810 }, { "grad_norm": 0.127694234251976, "learning_rate": 8.803103462834514e-05, "loss": 0.0056, "step": 15820 }, { "grad_norm": 0.1589706838130951, "learning_rate": 8.801313840445408e-05, "loss": 0.0056, "step": 15830 }, { "grad_norm": 0.1917734593153, "learning_rate": 8.799523063317524e-05, "loss": 0.006, "step": 15840 }, { "grad_norm": 0.16143187880516052, "learning_rate": 8.797731131994854e-05, "loss": 0.0066, "step": 15850 }, { "grad_norm": 0.1423158198595047, "learning_rate": 8.795938047021739e-05, "loss": 0.0068, "step": 15860 }, { "grad_norm": 0.15010786056518555, "learning_rate": 8.794143808942872e-05, "loss": 0.0054, "step": 15870 }, { "grad_norm": 0.11444635689258575, "learning_rate": 8.792348418303296e-05, "loss": 0.006, "step": 15880 }, { "grad_norm": 0.13557502627372742, "learning_rate": 8.790551875648398e-05, "loss": 0.0061, "step": 15890 }, { "grad_norm": 0.1620696783065796, "learning_rate": 8.788754181523926e-05, "loss": 0.0058, "step": 15900 }, { "grad_norm": 0.23927859961986542, "learning_rate": 8.78695533647597e-05, "loss": 0.0064, "step": 15910 }, { "grad_norm": 0.11849919706583023, "learning_rate": 8.785155341050972e-05, "loss": 0.006, "step": 15920 }, { "grad_norm": 0.15039747953414917, "learning_rate": 8.783354195795721e-05, "loss": 0.0057, "step": 15930 }, { "grad_norm": 0.09629960358142853, "learning_rate": 8.78155190125736e-05, "loss": 0.005, "step": 15940 }, { "grad_norm": 0.12490902096033096, "learning_rate": 8.779748457983378e-05, "loss": 0.0063, "step": 15950 }, { "grad_norm": 0.1635897308588028, "learning_rate": 8.777943866521612e-05, "loss": 0.0053, "step": 15960 }, { "grad_norm": 0.15962404012680054, "learning_rate": 8.77613812742025e-05, "loss": 0.0064, "step": 15970 }, { "grad_norm": 0.15527792274951935, "learning_rate": 8.774331241227829e-05, "loss": 0.0057, "step": 15980 }, { "grad_norm": 0.12908482551574707, "learning_rate": 8.772523208493232e-05, "loss": 0.0061, "step": 15990 }, { "grad_norm": 0.13836371898651123, "learning_rate": 8.770714029765692e-05, "loss": 0.0057, "step": 16000 }, { "grad_norm": 0.13541913032531738, "learning_rate": 8.768903705594789e-05, "loss": 0.0053, "step": 16010 }, { "grad_norm": 0.13319873809814453, "learning_rate": 8.767092236530453e-05, "loss": 0.0063, "step": 16020 }, { "grad_norm": 0.12243034690618515, "learning_rate": 8.76527962312296e-05, "loss": 0.0053, "step": 16030 }, { "grad_norm": 0.13700821995735168, "learning_rate": 8.763465865922934e-05, "loss": 0.0049, "step": 16040 }, { "grad_norm": 0.15338273346424103, "learning_rate": 8.761650965481347e-05, "loss": 0.0058, "step": 16050 }, { "grad_norm": 0.14180724322795868, "learning_rate": 8.759834922349516e-05, "loss": 0.006, "step": 16060 }, { "grad_norm": 0.10958247631788254, "learning_rate": 8.758017737079108e-05, "loss": 0.005, "step": 16070 }, { "grad_norm": 0.09231147170066833, "learning_rate": 8.756199410222137e-05, "loss": 0.0053, "step": 16080 }, { "grad_norm": 0.11724795401096344, "learning_rate": 8.754379942330963e-05, "loss": 0.0062, "step": 16090 }, { "grad_norm": 0.16793586313724518, "learning_rate": 8.75255933395829e-05, "loss": 0.004, "step": 16100 }, { "grad_norm": 0.11935970187187195, "learning_rate": 8.750737585657171e-05, "loss": 0.0047, "step": 16110 }, { "grad_norm": 0.08996416628360748, "learning_rate": 8.748914697981008e-05, "loss": 0.0055, "step": 16120 }, { "grad_norm": 0.14336173236370087, "learning_rate": 8.747090671483542e-05, "loss": 0.0053, "step": 16130 }, { "grad_norm": 0.1523982286453247, "learning_rate": 8.745265506718869e-05, "loss": 0.0051, "step": 16140 }, { "grad_norm": 0.12795311212539673, "learning_rate": 8.74343920424142e-05, "loss": 0.0054, "step": 16150 }, { "grad_norm": 0.11439909040927887, "learning_rate": 8.741611764605982e-05, "loss": 0.0054, "step": 16160 }, { "grad_norm": 0.15238462388515472, "learning_rate": 8.739783188367682e-05, "loss": 0.0059, "step": 16170 }, { "grad_norm": 0.12377002090215683, "learning_rate": 8.737953476081991e-05, "loss": 0.0054, "step": 16180 }, { "grad_norm": 0.15165863931179047, "learning_rate": 8.73612262830473e-05, "loss": 0.0058, "step": 16190 }, { "grad_norm": 0.16488352417945862, "learning_rate": 8.734290645592061e-05, "loss": 0.0073, "step": 16200 }, { "grad_norm": 0.17252075672149658, "learning_rate": 8.732457528500493e-05, "loss": 0.0059, "step": 16210 }, { "grad_norm": 0.13260091841220856, "learning_rate": 8.730623277586875e-05, "loss": 0.0056, "step": 16220 }, { "grad_norm": 0.12914824485778809, "learning_rate": 8.72878789340841e-05, "loss": 0.005, "step": 16230 }, { "grad_norm": 0.11223191767930984, "learning_rate": 8.726951376522635e-05, "loss": 0.0045, "step": 16240 }, { "grad_norm": 0.14624814689159393, "learning_rate": 8.725113727487435e-05, "loss": 0.0058, "step": 16250 }, { "grad_norm": 0.2067084163427353, "learning_rate": 8.723274946861042e-05, "loss": 0.0062, "step": 16260 }, { "grad_norm": 0.20357853174209595, "learning_rate": 8.721435035202026e-05, "loss": 0.0064, "step": 16270 }, { "grad_norm": 0.1888667494058609, "learning_rate": 8.719593993069306e-05, "loss": 0.0062, "step": 16280 }, { "grad_norm": 0.1396559327840805, "learning_rate": 8.717751821022139e-05, "loss": 0.0075, "step": 16290 }, { "grad_norm": 0.1511489301919937, "learning_rate": 8.715908519620134e-05, "loss": 0.0069, "step": 16300 }, { "grad_norm": 0.14609986543655396, "learning_rate": 8.71406408942323e-05, "loss": 0.0055, "step": 16310 }, { "grad_norm": 0.11664610356092453, "learning_rate": 8.712218530991723e-05, "loss": 0.005, "step": 16320 }, { "grad_norm": 0.09308174252510071, "learning_rate": 8.710371844886241e-05, "loss": 0.0052, "step": 16330 }, { "grad_norm": 0.12255854159593582, "learning_rate": 8.708524031667758e-05, "loss": 0.0056, "step": 16340 }, { "grad_norm": 0.13418330252170563, "learning_rate": 8.706675091897592e-05, "loss": 0.0048, "step": 16350 }, { "grad_norm": 0.12816165387630463, "learning_rate": 8.704825026137404e-05, "loss": 0.0054, "step": 16360 }, { "grad_norm": 0.09155205637216568, "learning_rate": 8.702973834949192e-05, "loss": 0.0047, "step": 16370 }, { "grad_norm": 0.10890305042266846, "learning_rate": 8.701121518895301e-05, "loss": 0.0045, "step": 16380 }, { "grad_norm": 0.14512844383716583, "learning_rate": 8.699268078538414e-05, "loss": 0.0074, "step": 16390 }, { "grad_norm": 0.16278144717216492, "learning_rate": 8.69741351444156e-05, "loss": 0.005, "step": 16400 }, { "grad_norm": 0.10238658636808395, "learning_rate": 8.695557827168101e-05, "loss": 0.0049, "step": 16410 }, { "grad_norm": 0.13227859139442444, "learning_rate": 8.693701017281753e-05, "loss": 0.0049, "step": 16420 }, { "grad_norm": 0.13298602402210236, "learning_rate": 8.691843085346563e-05, "loss": 0.0067, "step": 16430 }, { "grad_norm": 0.14277076721191406, "learning_rate": 8.689984031926919e-05, "loss": 0.0051, "step": 16440 }, { "grad_norm": 0.14742155373096466, "learning_rate": 8.688123857587555e-05, "loss": 0.0056, "step": 16450 }, { "grad_norm": 0.12107648700475693, "learning_rate": 8.686262562893544e-05, "loss": 0.0046, "step": 16460 }, { "grad_norm": 0.1315726786851883, "learning_rate": 8.684400148410294e-05, "loss": 0.0043, "step": 16470 }, { "grad_norm": 0.1358935534954071, "learning_rate": 8.682536614703562e-05, "loss": 0.0065, "step": 16480 }, { "grad_norm": 0.1632751077413559, "learning_rate": 8.680671962339437e-05, "loss": 0.0069, "step": 16490 }, { "grad_norm": 0.13016195595264435, "learning_rate": 8.678806191884352e-05, "loss": 0.0048, "step": 16500 }, { "grad_norm": 0.11954138427972794, "learning_rate": 8.67693930390508e-05, "loss": 0.006, "step": 16510 }, { "grad_norm": 0.0966130942106247, "learning_rate": 8.67507129896873e-05, "loss": 0.0053, "step": 16520 }, { "grad_norm": 0.13227058947086334, "learning_rate": 8.673202177642757e-05, "loss": 0.0053, "step": 16530 }, { "grad_norm": 0.13968178629875183, "learning_rate": 8.671331940494945e-05, "loss": 0.0047, "step": 16540 }, { "grad_norm": 0.14719158411026, "learning_rate": 8.669460588093427e-05, "loss": 0.0051, "step": 16550 }, { "grad_norm": 0.13144591450691223, "learning_rate": 8.667588121006667e-05, "loss": 0.0047, "step": 16560 }, { "grad_norm": 0.1462807059288025, "learning_rate": 8.665714539803475e-05, "loss": 0.005, "step": 16570 }, { "grad_norm": 0.11224523931741714, "learning_rate": 8.663839845052993e-05, "loss": 0.0059, "step": 16580 }, { "grad_norm": 0.18611736595630646, "learning_rate": 8.661964037324703e-05, "loss": 0.0054, "step": 16590 }, { "grad_norm": 0.14051610231399536, "learning_rate": 8.660087117188427e-05, "loss": 0.0075, "step": 16600 }, { "grad_norm": 0.09508849680423737, "learning_rate": 8.658209085214325e-05, "loss": 0.0057, "step": 16610 }, { "grad_norm": 0.1146106943488121, "learning_rate": 8.656329941972891e-05, "loss": 0.0055, "step": 16620 }, { "grad_norm": 0.15346196293830872, "learning_rate": 8.654449688034963e-05, "loss": 0.0058, "step": 16630 }, { "grad_norm": 0.15563702583312988, "learning_rate": 8.652568323971706e-05, "loss": 0.0075, "step": 16640 }, { "grad_norm": 0.16389474272727966, "learning_rate": 8.650685850354636e-05, "loss": 0.0066, "step": 16650 }, { "grad_norm": 0.1683148741722107, "learning_rate": 8.648802267755593e-05, "loss": 0.0062, "step": 16660 }, { "grad_norm": 0.11736930906772614, "learning_rate": 8.646917576746764e-05, "loss": 0.0066, "step": 16670 }, { "grad_norm": 0.15174764394760132, "learning_rate": 8.645031777900666e-05, "loss": 0.0061, "step": 16680 }, { "grad_norm": 0.17859888076782227, "learning_rate": 8.643144871790154e-05, "loss": 0.0084, "step": 16690 }, { "grad_norm": 0.12657657265663147, "learning_rate": 8.641256858988424e-05, "loss": 0.0063, "step": 16700 }, { "grad_norm": 0.1467714011669159, "learning_rate": 8.639367740069e-05, "loss": 0.0061, "step": 16710 }, { "grad_norm": 0.16311079263687134, "learning_rate": 8.63747751560575e-05, "loss": 0.0066, "step": 16720 }, { "grad_norm": 0.1346302479505539, "learning_rate": 8.635586186172871e-05, "loss": 0.0061, "step": 16730 }, { "grad_norm": 0.15310123562812805, "learning_rate": 8.633693752344902e-05, "loss": 0.0051, "step": 16740 }, { "grad_norm": 0.11306752264499664, "learning_rate": 8.631800214696713e-05, "loss": 0.007, "step": 16750 }, { "grad_norm": 0.15370652079582214, "learning_rate": 8.629905573803511e-05, "loss": 0.006, "step": 16760 }, { "grad_norm": 0.14629162847995758, "learning_rate": 8.628009830240839e-05, "loss": 0.0061, "step": 16770 }, { "grad_norm": 0.1592395007610321, "learning_rate": 8.626112984584571e-05, "loss": 0.0061, "step": 16780 }, { "grad_norm": 0.18656198680400848, "learning_rate": 8.62421503741092e-05, "loss": 0.0061, "step": 16790 }, { "grad_norm": 0.13458506762981415, "learning_rate": 8.622315989296432e-05, "loss": 0.0062, "step": 16800 }, { "grad_norm": 0.10142704844474792, "learning_rate": 8.62041584081799e-05, "loss": 0.0066, "step": 16810 }, { "grad_norm": 0.116159588098526, "learning_rate": 8.618514592552807e-05, "loss": 0.0067, "step": 16820 }, { "grad_norm": 0.10579659789800644, "learning_rate": 8.616612245078431e-05, "loss": 0.006, "step": 16830 }, { "grad_norm": 0.12020207196474075, "learning_rate": 8.614708798972746e-05, "loss": 0.0053, "step": 16840 }, { "grad_norm": 0.1661788821220398, "learning_rate": 8.61280425481397e-05, "loss": 0.0053, "step": 16850 }, { "grad_norm": 0.15245439112186432, "learning_rate": 8.61089861318065e-05, "loss": 0.0053, "step": 16860 }, { "grad_norm": 0.1338823288679123, "learning_rate": 8.608991874651673e-05, "loss": 0.0064, "step": 16870 }, { "grad_norm": 0.19924673438072205, "learning_rate": 8.607084039806255e-05, "loss": 0.0056, "step": 16880 }, { "grad_norm": 0.10350311547517776, "learning_rate": 8.605175109223944e-05, "loss": 0.0056, "step": 16890 }, { "grad_norm": 0.13995248079299927, "learning_rate": 8.603265083484624e-05, "loss": 0.0066, "step": 16900 }, { "grad_norm": 0.11158924549818039, "learning_rate": 8.60135396316851e-05, "loss": 0.0051, "step": 16910 }, { "grad_norm": 0.11393169313669205, "learning_rate": 8.599441748856152e-05, "loss": 0.0057, "step": 16920 }, { "grad_norm": 0.13039371371269226, "learning_rate": 8.597528441128427e-05, "loss": 0.0056, "step": 16930 }, { "grad_norm": 0.11200351268053055, "learning_rate": 8.595614040566549e-05, "loss": 0.0044, "step": 16940 }, { "grad_norm": 0.13897691667079926, "learning_rate": 8.593698547752063e-05, "loss": 0.0046, "step": 16950 }, { "grad_norm": 0.15900687873363495, "learning_rate": 8.591781963266843e-05, "loss": 0.0059, "step": 16960 }, { "grad_norm": 0.14946310222148895, "learning_rate": 8.5898642876931e-05, "loss": 0.0044, "step": 16970 }, { "grad_norm": 0.1419273316860199, "learning_rate": 8.587945521613369e-05, "loss": 0.0056, "step": 16980 }, { "grad_norm": 0.15914732217788696, "learning_rate": 8.586025665610524e-05, "loss": 0.007, "step": 16990 }, { "grad_norm": 0.1261754184961319, "learning_rate": 8.584104720267765e-05, "loss": 0.0066, "step": 17000 }, { "grad_norm": 0.11583907902240753, "learning_rate": 8.582182686168625e-05, "loss": 0.0066, "step": 17010 }, { "grad_norm": 0.11141370236873627, "learning_rate": 8.580259563896967e-05, "loss": 0.0054, "step": 17020 }, { "grad_norm": 0.1226683184504509, "learning_rate": 8.578335354036983e-05, "loss": 0.0053, "step": 17030 }, { "grad_norm": 0.1526629626750946, "learning_rate": 8.576410057173201e-05, "loss": 0.0054, "step": 17040 }, { "grad_norm": 0.1130780577659607, "learning_rate": 8.574483673890474e-05, "loss": 0.0043, "step": 17050 }, { "grad_norm": 0.13227541744709015, "learning_rate": 8.572556204773983e-05, "loss": 0.0058, "step": 17060 }, { "grad_norm": 0.14328451454639435, "learning_rate": 8.570627650409246e-05, "loss": 0.0048, "step": 17070 }, { "grad_norm": 0.11596527695655823, "learning_rate": 8.568698011382107e-05, "loss": 0.0057, "step": 17080 }, { "grad_norm": 0.11340569704771042, "learning_rate": 8.566767288278738e-05, "loss": 0.0054, "step": 17090 }, { "grad_norm": 0.10816922783851624, "learning_rate": 8.56483548168564e-05, "loss": 0.0051, "step": 17100 }, { "grad_norm": 0.14061765372753143, "learning_rate": 8.562902592189648e-05, "loss": 0.0061, "step": 17110 }, { "grad_norm": 0.1132412776350975, "learning_rate": 8.560968620377921e-05, "loss": 0.0049, "step": 17120 }, { "grad_norm": 0.15821801126003265, "learning_rate": 8.559033566837951e-05, "loss": 0.0053, "step": 17130 }, { "grad_norm": 0.11755538731813431, "learning_rate": 8.557097432157551e-05, "loss": 0.0058, "step": 17140 }, { "grad_norm": 0.10606811940670013, "learning_rate": 8.555160216924872e-05, "loss": 0.0052, "step": 17150 }, { "grad_norm": 0.1010679304599762, "learning_rate": 8.55322192172839e-05, "loss": 0.0059, "step": 17160 }, { "grad_norm": 0.11153741180896759, "learning_rate": 8.551282547156902e-05, "loss": 0.005, "step": 17170 }, { "grad_norm": 0.12960731983184814, "learning_rate": 8.549342093799544e-05, "loss": 0.0061, "step": 17180 }, { "grad_norm": 0.17636384069919586, "learning_rate": 8.547400562245773e-05, "loss": 0.0058, "step": 17190 }, { "grad_norm": 0.12351038306951523, "learning_rate": 8.545457953085374e-05, "loss": 0.0054, "step": 17200 }, { "grad_norm": 0.15617306530475616, "learning_rate": 8.543514266908463e-05, "loss": 0.0052, "step": 17210 }, { "grad_norm": 0.11648084223270416, "learning_rate": 8.541569504305478e-05, "loss": 0.006, "step": 17220 }, { "grad_norm": 0.09417236596345901, "learning_rate": 8.539623665867187e-05, "loss": 0.0052, "step": 17230 }, { "grad_norm": 0.1324903815984726, "learning_rate": 8.537676752184685e-05, "loss": 0.005, "step": 17240 }, { "grad_norm": 0.11660132557153702, "learning_rate": 8.53572876384939e-05, "loss": 0.0053, "step": 17250 }, { "grad_norm": 0.11196424812078476, "learning_rate": 8.533779701453056e-05, "loss": 0.0051, "step": 17260 }, { "grad_norm": 0.16767056286334991, "learning_rate": 8.53182956558775e-05, "loss": 0.0045, "step": 17270 }, { "grad_norm": 0.147517591714859, "learning_rate": 8.529878356845877e-05, "loss": 0.005, "step": 17280 }, { "grad_norm": 0.12481088936328888, "learning_rate": 8.527926075820158e-05, "loss": 0.0042, "step": 17290 }, { "grad_norm": 0.12280136346817017, "learning_rate": 8.525972723103648e-05, "loss": 0.0064, "step": 17300 }, { "grad_norm": 0.15549679100513458, "learning_rate": 8.524018299289722e-05, "loss": 0.0061, "step": 17310 }, { "grad_norm": 0.15819813311100006, "learning_rate": 8.522062804972083e-05, "loss": 0.0054, "step": 17320 }, { "grad_norm": 0.1324189007282257, "learning_rate": 8.520106240744759e-05, "loss": 0.0055, "step": 17330 }, { "grad_norm": 0.10023807734251022, "learning_rate": 8.518148607202102e-05, "loss": 0.0046, "step": 17340 }, { "grad_norm": 0.08843084424734116, "learning_rate": 8.51618990493879e-05, "loss": 0.005, "step": 17350 }, { "grad_norm": 0.1587706059217453, "learning_rate": 8.514230134549823e-05, "loss": 0.0046, "step": 17360 }, { "grad_norm": 0.13007818162441254, "learning_rate": 8.51226929663053e-05, "loss": 0.0053, "step": 17370 }, { "grad_norm": 0.13988979160785675, "learning_rate": 8.51030739177656e-05, "loss": 0.0054, "step": 17380 }, { "grad_norm": 0.1114509254693985, "learning_rate": 8.508344420583889e-05, "loss": 0.0063, "step": 17390 }, { "grad_norm": 0.16331839561462402, "learning_rate": 8.506380383648816e-05, "loss": 0.0078, "step": 17400 }, { "grad_norm": 0.1144111156463623, "learning_rate": 8.504415281567963e-05, "loss": 0.0051, "step": 17410 }, { "grad_norm": 0.14225329458713531, "learning_rate": 8.502449114938275e-05, "loss": 0.0049, "step": 17420 }, { "grad_norm": 0.09711595624685287, "learning_rate": 8.500481884357025e-05, "loss": 0.0055, "step": 17430 }, { "grad_norm": 0.14091628789901733, "learning_rate": 8.498513590421801e-05, "loss": 0.0048, "step": 17440 }, { "grad_norm": 0.1254860907793045, "learning_rate": 8.496544233730522e-05, "loss": 0.0067, "step": 17450 }, { "grad_norm": 0.12539657950401306, "learning_rate": 8.494573814881426e-05, "loss": 0.0077, "step": 17460 }, { "grad_norm": 0.15624015033245087, "learning_rate": 8.492602334473074e-05, "loss": 0.0065, "step": 17470 }, { "grad_norm": 0.17239880561828613, "learning_rate": 8.49062979310435e-05, "loss": 0.0061, "step": 17480 }, { "grad_norm": 0.1198827475309372, "learning_rate": 8.488656191374458e-05, "loss": 0.0067, "step": 17490 }, { "grad_norm": 0.16573956608772278, "learning_rate": 8.48668152988293e-05, "loss": 0.0051, "step": 17500 }, { "grad_norm": 0.16630002856254578, "learning_rate": 8.484705809229612e-05, "loss": 0.0061, "step": 17510 }, { "grad_norm": 0.08972948789596558, "learning_rate": 8.482729030014677e-05, "loss": 0.0062, "step": 17520 }, { "grad_norm": 0.13251982629299164, "learning_rate": 8.48075119283862e-05, "loss": 0.0061, "step": 17530 }, { "grad_norm": 0.12141115218400955, "learning_rate": 8.478772298302254e-05, "loss": 0.0058, "step": 17540 }, { "grad_norm": 0.09919685870409012, "learning_rate": 8.476792347006716e-05, "loss": 0.0046, "step": 17550 }, { "grad_norm": 0.14007160067558289, "learning_rate": 8.474811339553462e-05, "loss": 0.0051, "step": 17560 }, { "grad_norm": 0.10816889256238937, "learning_rate": 8.47282927654427e-05, "loss": 0.0063, "step": 17570 }, { "grad_norm": 0.1305587738752365, "learning_rate": 8.470846158581238e-05, "loss": 0.0056, "step": 17580 }, { "grad_norm": 0.12226222455501556, "learning_rate": 8.468861986266787e-05, "loss": 0.0058, "step": 17590 }, { "grad_norm": 0.15601205825805664, "learning_rate": 8.466876760203654e-05, "loss": 0.0062, "step": 17600 }, { "grad_norm": 0.1459972858428955, "learning_rate": 8.464890480994898e-05, "loss": 0.0053, "step": 17610 }, { "grad_norm": 0.12483672797679901, "learning_rate": 8.462903149243899e-05, "loss": 0.0045, "step": 17620 }, { "grad_norm": 0.1250029355287552, "learning_rate": 8.460914765554357e-05, "loss": 0.005, "step": 17630 }, { "grad_norm": 0.12042227387428284, "learning_rate": 8.458925330530288e-05, "loss": 0.0057, "step": 17640 }, { "grad_norm": 0.0920533761382103, "learning_rate": 8.456934844776032e-05, "loss": 0.0049, "step": 17650 }, { "grad_norm": 0.12227769196033478, "learning_rate": 8.454943308896246e-05, "loss": 0.0055, "step": 17660 }, { "grad_norm": 0.16628198325634003, "learning_rate": 8.452950723495905e-05, "loss": 0.0042, "step": 17670 }, { "grad_norm": 0.11223672330379486, "learning_rate": 8.450957089180303e-05, "loss": 0.0041, "step": 17680 }, { "grad_norm": 0.10128932446241379, "learning_rate": 8.448962406555055e-05, "loss": 0.0069, "step": 17690 }, { "grad_norm": 0.13291560113430023, "learning_rate": 8.446966676226093e-05, "loss": 0.0056, "step": 17700 }, { "grad_norm": 0.13856172561645508, "learning_rate": 8.444969898799667e-05, "loss": 0.0059, "step": 17710 }, { "grad_norm": 0.18483726680278778, "learning_rate": 8.442972074882343e-05, "loss": 0.0055, "step": 17720 }, { "grad_norm": 0.14297139644622803, "learning_rate": 8.44097320508101e-05, "loss": 0.0051, "step": 17730 }, { "grad_norm": 0.13405489921569824, "learning_rate": 8.43897329000287e-05, "loss": 0.0045, "step": 17740 }, { "grad_norm": 0.1063595861196518, "learning_rate": 8.436972330255448e-05, "loss": 0.0047, "step": 17750 }, { "grad_norm": 0.12371940165758133, "learning_rate": 8.434970326446579e-05, "loss": 0.0048, "step": 17760 }, { "grad_norm": 0.19933241605758667, "learning_rate": 8.432967279184418e-05, "loss": 0.0055, "step": 17770 }, { "grad_norm": 0.13168714940547943, "learning_rate": 8.430963189077441e-05, "loss": 0.0049, "step": 17780 }, { "grad_norm": 0.14970533549785614, "learning_rate": 8.428958056734437e-05, "loss": 0.005, "step": 17790 }, { "grad_norm": 0.17981652915477753, "learning_rate": 8.426951882764513e-05, "loss": 0.0055, "step": 17800 }, { "grad_norm": 0.12232601642608643, "learning_rate": 8.424944667777089e-05, "loss": 0.0079, "step": 17810 }, { "grad_norm": 0.1581636369228363, "learning_rate": 8.422936412381905e-05, "loss": 0.0059, "step": 17820 }, { "grad_norm": 0.14144784212112427, "learning_rate": 8.420927117189017e-05, "loss": 0.0052, "step": 17830 }, { "grad_norm": 0.11854933202266693, "learning_rate": 8.418916782808795e-05, "loss": 0.0055, "step": 17840 }, { "grad_norm": 0.13232985138893127, "learning_rate": 8.416905409851926e-05, "loss": 0.0058, "step": 17850 }, { "grad_norm": 0.1498449444770813, "learning_rate": 8.41489299892941e-05, "loss": 0.0047, "step": 17860 }, { "grad_norm": 0.13479827344417572, "learning_rate": 8.412879550652566e-05, "loss": 0.0056, "step": 17870 }, { "grad_norm": 0.09958785772323608, "learning_rate": 8.410865065633029e-05, "loss": 0.0051, "step": 17880 }, { "grad_norm": 0.11211567372083664, "learning_rate": 8.408849544482742e-05, "loss": 0.0043, "step": 17890 }, { "grad_norm": 0.15940943360328674, "learning_rate": 8.406832987813968e-05, "loss": 0.0047, "step": 17900 }, { "grad_norm": 0.1046198159456253, "learning_rate": 8.404815396239286e-05, "loss": 0.0045, "step": 17910 }, { "grad_norm": 0.14519041776657104, "learning_rate": 8.402796770371587e-05, "loss": 0.0052, "step": 17920 }, { "grad_norm": 0.09552135318517685, "learning_rate": 8.400777110824071e-05, "loss": 0.0048, "step": 17930 }, { "grad_norm": 0.15784572064876556, "learning_rate": 8.398756418210263e-05, "loss": 0.0049, "step": 17940 }, { "grad_norm": 0.09828948229551315, "learning_rate": 8.396734693143993e-05, "loss": 0.0043, "step": 17950 }, { "grad_norm": 0.12524932622909546, "learning_rate": 8.39471193623941e-05, "loss": 0.004, "step": 17960 }, { "grad_norm": 0.13264253735542297, "learning_rate": 8.392688148110974e-05, "loss": 0.0052, "step": 17970 }, { "grad_norm": 0.13014428317546844, "learning_rate": 8.390663329373456e-05, "loss": 0.0042, "step": 17980 }, { "grad_norm": 0.14081566035747528, "learning_rate": 8.388637480641944e-05, "loss": 0.0043, "step": 17990 }, { "grad_norm": 0.10673496127128601, "learning_rate": 8.386610602531837e-05, "loss": 0.0046, "step": 18000 }, { "grad_norm": 0.10066929459571838, "learning_rate": 8.384582695658847e-05, "loss": 0.0048, "step": 18010 }, { "grad_norm": 0.12579528987407684, "learning_rate": 8.382553760638999e-05, "loss": 0.0047, "step": 18020 }, { "grad_norm": 0.1299908459186554, "learning_rate": 8.380523798088631e-05, "loss": 0.0051, "step": 18030 }, { "grad_norm": 0.11472393572330475, "learning_rate": 8.378492808624389e-05, "loss": 0.0054, "step": 18040 }, { "grad_norm": 0.1071930080652237, "learning_rate": 8.376460792863237e-05, "loss": 0.0061, "step": 18050 }, { "grad_norm": 0.10635554045438766, "learning_rate": 8.374427751422444e-05, "loss": 0.0047, "step": 18060 }, { "grad_norm": 0.11322063207626343, "learning_rate": 8.3723936849196e-05, "loss": 0.0052, "step": 18070 }, { "grad_norm": 0.14777879416942596, "learning_rate": 8.370358593972595e-05, "loss": 0.0048, "step": 18080 }, { "grad_norm": 0.13016892969608307, "learning_rate": 8.36832247919964e-05, "loss": 0.005, "step": 18090 }, { "grad_norm": 0.10385660082101822, "learning_rate": 8.36628534121925e-05, "loss": 0.0041, "step": 18100 }, { "grad_norm": 0.1187315583229065, "learning_rate": 8.364247180650254e-05, "loss": 0.0051, "step": 18110 }, { "grad_norm": 0.12811096012592316, "learning_rate": 8.362207998111794e-05, "loss": 0.0044, "step": 18120 }, { "grad_norm": 0.10719167441129684, "learning_rate": 8.360167794223318e-05, "loss": 0.0043, "step": 18130 }, { "grad_norm": 0.1745486706495285, "learning_rate": 8.358126569604586e-05, "loss": 0.0065, "step": 18140 }, { "grad_norm": 0.1117403581738472, "learning_rate": 8.356084324875668e-05, "loss": 0.0041, "step": 18150 }, { "grad_norm": 0.10326717048883438, "learning_rate": 8.354041060656945e-05, "loss": 0.0044, "step": 18160 }, { "grad_norm": 0.10791341960430145, "learning_rate": 8.351996777569106e-05, "loss": 0.0067, "step": 18170 }, { "grad_norm": 0.09140375256538391, "learning_rate": 8.349951476233148e-05, "loss": 0.0048, "step": 18180 }, { "grad_norm": 0.13937237858772278, "learning_rate": 8.347905157270386e-05, "loss": 0.0046, "step": 18190 }, { "grad_norm": 0.16493289172649384, "learning_rate": 8.345857821302432e-05, "loss": 0.0047, "step": 18200 }, { "grad_norm": 0.16343514621257782, "learning_rate": 8.343809468951213e-05, "loss": 0.0049, "step": 18210 }, { "grad_norm": 0.10505083203315735, "learning_rate": 8.341760100838965e-05, "loss": 0.0054, "step": 18220 }, { "grad_norm": 0.11423405259847641, "learning_rate": 8.339709717588233e-05, "loss": 0.0051, "step": 18230 }, { "grad_norm": 0.12064660340547562, "learning_rate": 8.33765831982187e-05, "loss": 0.0051, "step": 18240 }, { "grad_norm": 0.16841204464435577, "learning_rate": 8.335605908163035e-05, "loss": 0.0046, "step": 18250 }, { "grad_norm": 0.13804352283477783, "learning_rate": 8.333552483235196e-05, "loss": 0.0052, "step": 18260 }, { "grad_norm": 0.15134494006633759, "learning_rate": 8.33149804566213e-05, "loss": 0.0057, "step": 18270 }, { "grad_norm": 0.15208619832992554, "learning_rate": 8.329442596067921e-05, "loss": 0.0057, "step": 18280 }, { "grad_norm": 0.1825433075428009, "learning_rate": 8.32738613507696e-05, "loss": 0.0054, "step": 18290 }, { "grad_norm": 0.13447539508342743, "learning_rate": 8.325328663313946e-05, "loss": 0.005, "step": 18300 }, { "grad_norm": 0.13669344782829285, "learning_rate": 8.323270181403884e-05, "loss": 0.0055, "step": 18310 }, { "grad_norm": 0.13302269577980042, "learning_rate": 8.321210689972086e-05, "loss": 0.005, "step": 18320 }, { "grad_norm": 0.14229074120521545, "learning_rate": 8.319150189644174e-05, "loss": 0.0058, "step": 18330 }, { "grad_norm": 0.10851461440324783, "learning_rate": 8.31708868104607e-05, "loss": 0.0049, "step": 18340 }, { "grad_norm": 0.12380348891019821, "learning_rate": 8.315026164804007e-05, "loss": 0.0048, "step": 18350 }, { "grad_norm": 0.12899619340896606, "learning_rate": 8.312962641544524e-05, "loss": 0.0043, "step": 18360 }, { "grad_norm": 0.09719009697437286, "learning_rate": 8.310898111894465e-05, "loss": 0.0046, "step": 18370 }, { "grad_norm": 0.15128940343856812, "learning_rate": 8.308832576480977e-05, "loss": 0.0056, "step": 18380 }, { "grad_norm": 0.11996312439441681, "learning_rate": 8.306766035931519e-05, "loss": 0.0049, "step": 18390 }, { "grad_norm": 0.12031252682209015, "learning_rate": 8.304698490873847e-05, "loss": 0.0048, "step": 18400 }, { "grad_norm": 0.1433563232421875, "learning_rate": 8.30262994193603e-05, "loss": 0.0065, "step": 18410 }, { "grad_norm": 0.12389174103736877, "learning_rate": 8.300560389746438e-05, "loss": 0.0048, "step": 18420 }, { "grad_norm": 0.1618342399597168, "learning_rate": 8.298489834933745e-05, "loss": 0.0063, "step": 18430 }, { "grad_norm": 0.125018909573555, "learning_rate": 8.296418278126934e-05, "loss": 0.0041, "step": 18440 }, { "grad_norm": 0.1156134158372879, "learning_rate": 8.294345719955284e-05, "loss": 0.0058, "step": 18450 }, { "grad_norm": 0.13189837336540222, "learning_rate": 8.29227216104839e-05, "loss": 0.0061, "step": 18460 }, { "grad_norm": 0.12790213525295258, "learning_rate": 8.290197602036137e-05, "loss": 0.0055, "step": 18470 }, { "grad_norm": 0.12402161955833435, "learning_rate": 8.288122043548725e-05, "loss": 0.0044, "step": 18480 }, { "grad_norm": 0.09819725155830383, "learning_rate": 8.286045486216657e-05, "loss": 0.0049, "step": 18490 }, { "grad_norm": 0.12116826325654984, "learning_rate": 8.283967930670733e-05, "loss": 0.0053, "step": 18500 }, { "grad_norm": 0.1225460097193718, "learning_rate": 8.281889377542058e-05, "loss": 0.0058, "step": 18510 }, { "grad_norm": 0.10606766492128372, "learning_rate": 8.279809827462045e-05, "loss": 0.0063, "step": 18520 }, { "grad_norm": 0.10471025854349136, "learning_rate": 8.277729281062402e-05, "loss": 0.0043, "step": 18530 }, { "grad_norm": 0.105790875852108, "learning_rate": 8.27564773897515e-05, "loss": 0.0051, "step": 18540 }, { "grad_norm": 0.11032555252313614, "learning_rate": 8.273565201832602e-05, "loss": 0.0046, "step": 18550 }, { "grad_norm": 0.11379606276750565, "learning_rate": 8.27148167026738e-05, "loss": 0.0045, "step": 18560 }, { "grad_norm": 0.14655344188213348, "learning_rate": 8.269397144912405e-05, "loss": 0.0063, "step": 18570 }, { "grad_norm": 0.13799504935741425, "learning_rate": 8.267311626400899e-05, "loss": 0.0039, "step": 18580 }, { "grad_norm": 0.11483294516801834, "learning_rate": 8.26522511536639e-05, "loss": 0.0054, "step": 18590 }, { "grad_norm": 0.13940191268920898, "learning_rate": 8.263137612442706e-05, "loss": 0.0049, "step": 18600 }, { "grad_norm": 0.10853126645088196, "learning_rate": 8.261049118263971e-05, "loss": 0.006, "step": 18610 }, { "grad_norm": 0.10080940276384354, "learning_rate": 8.258959633464619e-05, "loss": 0.0044, "step": 18620 }, { "grad_norm": 0.11248516291379929, "learning_rate": 8.256869158679377e-05, "loss": 0.0047, "step": 18630 }, { "grad_norm": 0.11745750904083252, "learning_rate": 8.254777694543278e-05, "loss": 0.0051, "step": 18640 }, { "grad_norm": 0.1940585821866989, "learning_rate": 8.252685241691651e-05, "loss": 0.0044, "step": 18650 }, { "grad_norm": 0.0893968790769577, "learning_rate": 8.250591800760133e-05, "loss": 0.0045, "step": 18660 }, { "grad_norm": 0.13065215945243835, "learning_rate": 8.248497372384649e-05, "loss": 0.0047, "step": 18670 }, { "grad_norm": 0.20832493901252747, "learning_rate": 8.246401957201437e-05, "loss": 0.0061, "step": 18680 }, { "grad_norm": 0.15563459694385529, "learning_rate": 8.244305555847027e-05, "loss": 0.0054, "step": 18690 }, { "grad_norm": 0.12977628409862518, "learning_rate": 8.24220816895825e-05, "loss": 0.005, "step": 18700 }, { "grad_norm": 0.0769093930721283, "learning_rate": 8.240109797172237e-05, "loss": 0.0045, "step": 18710 }, { "grad_norm": 0.10265180468559265, "learning_rate": 8.238010441126416e-05, "loss": 0.005, "step": 18720 }, { "grad_norm": 0.14538060128688812, "learning_rate": 8.23591010145852e-05, "loss": 0.0044, "step": 18730 }, { "grad_norm": 0.1315654218196869, "learning_rate": 8.233808778806571e-05, "loss": 0.0063, "step": 18740 }, { "grad_norm": 0.19353154301643372, "learning_rate": 8.231706473808903e-05, "loss": 0.0056, "step": 18750 }, { "grad_norm": 0.10712548345327377, "learning_rate": 8.229603187104133e-05, "loss": 0.005, "step": 18760 }, { "grad_norm": 0.16423346102237701, "learning_rate": 8.22749891933119e-05, "loss": 0.0072, "step": 18770 }, { "grad_norm": 0.12629786133766174, "learning_rate": 8.225393671129291e-05, "loss": 0.0066, "step": 18780 }, { "grad_norm": 0.169621080160141, "learning_rate": 8.223287443137957e-05, "loss": 0.0066, "step": 18790 }, { "grad_norm": 0.10039521008729935, "learning_rate": 8.221180235997004e-05, "loss": 0.0064, "step": 18800 }, { "grad_norm": 0.09959439188241959, "learning_rate": 8.219072050346544e-05, "loss": 0.0057, "step": 18810 }, { "grad_norm": 0.12553896009922028, "learning_rate": 8.216962886826992e-05, "loss": 0.0055, "step": 18820 }, { "grad_norm": 0.08906561881303787, "learning_rate": 8.214852746079054e-05, "loss": 0.005, "step": 18830 }, { "grad_norm": 0.12072406709194183, "learning_rate": 8.212741628743732e-05, "loss": 0.0049, "step": 18840 }, { "grad_norm": 0.15830902755260468, "learning_rate": 8.210629535462333e-05, "loss": 0.0059, "step": 18850 }, { "grad_norm": 0.13273373246192932, "learning_rate": 8.208516466876453e-05, "loss": 0.0049, "step": 18860 }, { "grad_norm": 0.1346953958272934, "learning_rate": 8.206402423627986e-05, "loss": 0.0045, "step": 18870 }, { "grad_norm": 0.1088934913277626, "learning_rate": 8.204287406359124e-05, "loss": 0.0041, "step": 18880 }, { "grad_norm": 0.13443349301815033, "learning_rate": 8.20217141571235e-05, "loss": 0.0044, "step": 18890 }, { "grad_norm": 0.12654121220111847, "learning_rate": 8.200054452330449e-05, "loss": 0.0042, "step": 18900 }, { "grad_norm": 0.1745271384716034, "learning_rate": 8.197936516856499e-05, "loss": 0.0051, "step": 18910 }, { "grad_norm": 0.11543574929237366, "learning_rate": 8.195817609933871e-05, "loss": 0.0061, "step": 18920 }, { "grad_norm": 0.11783380806446075, "learning_rate": 8.193697732206233e-05, "loss": 0.0047, "step": 18930 }, { "grad_norm": 0.08201044052839279, "learning_rate": 8.19157688431755e-05, "loss": 0.0041, "step": 18940 }, { "grad_norm": 0.10619185119867325, "learning_rate": 8.189455066912077e-05, "loss": 0.0055, "step": 18950 }, { "grad_norm": 0.1368713229894638, "learning_rate": 8.187332280634369e-05, "loss": 0.0044, "step": 18960 }, { "grad_norm": 0.11612719297409058, "learning_rate": 8.18520852612927e-05, "loss": 0.0064, "step": 18970 }, { "grad_norm": 0.15578250586986542, "learning_rate": 8.183083804041921e-05, "loss": 0.0048, "step": 18980 }, { "grad_norm": 0.12651853263378143, "learning_rate": 8.180958115017757e-05, "loss": 0.0048, "step": 18990 }, { "grad_norm": 0.08797446638345718, "learning_rate": 8.178831459702505e-05, "loss": 0.0049, "step": 19000 }, { "grad_norm": 0.1868748515844345, "learning_rate": 8.17670383874219e-05, "loss": 0.0044, "step": 19010 }, { "grad_norm": 0.12090858817100525, "learning_rate": 8.174575252783124e-05, "loss": 0.004, "step": 19020 }, { "grad_norm": 0.09096134454011917, "learning_rate": 8.172445702471914e-05, "loss": 0.0051, "step": 19030 }, { "grad_norm": 0.10263065248727798, "learning_rate": 8.170315188455466e-05, "loss": 0.0053, "step": 19040 }, { "grad_norm": 0.09770806133747101, "learning_rate": 8.168183711380969e-05, "loss": 0.0052, "step": 19050 }, { "grad_norm": 0.11148817837238312, "learning_rate": 8.166051271895913e-05, "loss": 0.0063, "step": 19060 }, { "grad_norm": 0.13792644441127777, "learning_rate": 8.163917870648075e-05, "loss": 0.0061, "step": 19070 }, { "grad_norm": 0.1148388534784317, "learning_rate": 8.161783508285526e-05, "loss": 0.0064, "step": 19080 }, { "grad_norm": 0.1404326856136322, "learning_rate": 8.159648185456628e-05, "loss": 0.0053, "step": 19090 }, { "grad_norm": 0.15902985632419586, "learning_rate": 8.157511902810038e-05, "loss": 0.0054, "step": 19100 }, { "grad_norm": 0.17307549715042114, "learning_rate": 8.155374660994701e-05, "loss": 0.0071, "step": 19110 }, { "grad_norm": 0.11331210285425186, "learning_rate": 8.153236460659857e-05, "loss": 0.0051, "step": 19120 }, { "grad_norm": 0.12461800873279572, "learning_rate": 8.151097302455031e-05, "loss": 0.0054, "step": 19130 }, { "grad_norm": 0.14678697288036346, "learning_rate": 8.148957187030044e-05, "loss": 0.0058, "step": 19140 }, { "grad_norm": 0.1199473887681961, "learning_rate": 8.146816115035006e-05, "loss": 0.0063, "step": 19150 }, { "grad_norm": 0.15286631882190704, "learning_rate": 8.14467408712032e-05, "loss": 0.0061, "step": 19160 }, { "grad_norm": 0.10758288204669952, "learning_rate": 8.142531103936678e-05, "loss": 0.0048, "step": 19170 }, { "grad_norm": 0.13335594534873962, "learning_rate": 8.14038716613506e-05, "loss": 0.0054, "step": 19180 }, { "grad_norm": 0.11543095111846924, "learning_rate": 8.138242274366736e-05, "loss": 0.0059, "step": 19190 }, { "grad_norm": 0.16977964341640472, "learning_rate": 8.136096429283271e-05, "loss": 0.0062, "step": 19200 }, { "grad_norm": 0.1370437890291214, "learning_rate": 8.133949631536515e-05, "loss": 0.0045, "step": 19210 }, { "grad_norm": 0.1220831573009491, "learning_rate": 8.131801881778607e-05, "loss": 0.0052, "step": 19220 }, { "grad_norm": 0.13638974726200104, "learning_rate": 8.129653180661978e-05, "loss": 0.0053, "step": 19230 }, { "grad_norm": 0.12648577988147736, "learning_rate": 8.127503528839346e-05, "loss": 0.0049, "step": 19240 }, { "grad_norm": 0.1786518692970276, "learning_rate": 8.125352926963721e-05, "loss": 0.0059, "step": 19250 }, { "grad_norm": 0.15240177512168884, "learning_rate": 8.123201375688395e-05, "loss": 0.0057, "step": 19260 }, { "grad_norm": 0.12930451333522797, "learning_rate": 8.121048875666954e-05, "loss": 0.0045, "step": 19270 }, { "grad_norm": 0.19413773715496063, "learning_rate": 8.118895427553274e-05, "loss": 0.0062, "step": 19280 }, { "grad_norm": 0.13300421833992004, "learning_rate": 8.116741032001511e-05, "loss": 0.0051, "step": 19290 }, { "grad_norm": 0.12122360616922379, "learning_rate": 8.114585689666114e-05, "loss": 0.0052, "step": 19300 }, { "grad_norm": 0.10667137056589127, "learning_rate": 8.112429401201821e-05, "loss": 0.0049, "step": 19310 }, { "grad_norm": 0.13478370010852814, "learning_rate": 8.110272167263656e-05, "loss": 0.0062, "step": 19320 }, { "grad_norm": 0.12481177598237991, "learning_rate": 8.108113988506929e-05, "loss": 0.0044, "step": 19330 }, { "grad_norm": 0.15710194408893585, "learning_rate": 8.105954865587235e-05, "loss": 0.0051, "step": 19340 }, { "grad_norm": 0.09836911410093307, "learning_rate": 8.103794799160463e-05, "loss": 0.004, "step": 19350 }, { "grad_norm": 0.09088324755430222, "learning_rate": 8.101633789882781e-05, "loss": 0.004, "step": 19360 }, { "grad_norm": 0.14271606504917145, "learning_rate": 8.099471838410648e-05, "loss": 0.0068, "step": 19370 }, { "grad_norm": 0.08520358055830002, "learning_rate": 8.097308945400806e-05, "loss": 0.0051, "step": 19380 }, { "grad_norm": 0.11337148398160934, "learning_rate": 8.095145111510288e-05, "loss": 0.0069, "step": 19390 }, { "grad_norm": 0.14636580646038055, "learning_rate": 8.092980337396406e-05, "loss": 0.0053, "step": 19400 }, { "grad_norm": 0.14585290849208832, "learning_rate": 8.090814623716763e-05, "loss": 0.0059, "step": 19410 }, { "grad_norm": 0.10453341156244278, "learning_rate": 8.088647971129246e-05, "loss": 0.0044, "step": 19420 }, { "grad_norm": 0.10915446281433105, "learning_rate": 8.086480380292026e-05, "loss": 0.0045, "step": 19430 }, { "grad_norm": 0.12952055037021637, "learning_rate": 8.084311851863562e-05, "loss": 0.0061, "step": 19440 }, { "grad_norm": 0.12670087814331055, "learning_rate": 8.082142386502591e-05, "loss": 0.0044, "step": 19450 }, { "grad_norm": 0.16822823882102966, "learning_rate": 8.079971984868145e-05, "loss": 0.0046, "step": 19460 }, { "grad_norm": 0.1043931245803833, "learning_rate": 8.077800647619532e-05, "loss": 0.0045, "step": 19470 }, { "grad_norm": 0.09571532160043716, "learning_rate": 8.075628375416345e-05, "loss": 0.0043, "step": 19480 }, { "grad_norm": 0.0770951360464096, "learning_rate": 8.073455168918464e-05, "loss": 0.0039, "step": 19490 }, { "grad_norm": 0.11346016079187393, "learning_rate": 8.071281028786055e-05, "loss": 0.0039, "step": 19500 }, { "grad_norm": 0.17208901047706604, "learning_rate": 8.069105955679562e-05, "loss": 0.006, "step": 19510 }, { "grad_norm": 0.14662432670593262, "learning_rate": 8.066929950259713e-05, "loss": 0.0044, "step": 19520 }, { "grad_norm": 0.10210877656936646, "learning_rate": 8.064753013187522e-05, "loss": 0.0057, "step": 19530 }, { "grad_norm": 0.1339118778705597, "learning_rate": 8.062575145124289e-05, "loss": 0.0038, "step": 19540 }, { "grad_norm": 0.09193965792655945, "learning_rate": 8.060396346731587e-05, "loss": 0.0043, "step": 19550 }, { "grad_norm": 0.10558807104825974, "learning_rate": 8.058216618671281e-05, "loss": 0.0049, "step": 19560 }, { "grad_norm": 0.1612900197505951, "learning_rate": 8.056035961605514e-05, "loss": 0.0052, "step": 19570 }, { "grad_norm": 0.1588786542415619, "learning_rate": 8.05385437619671e-05, "loss": 0.0049, "step": 19580 }, { "grad_norm": 0.11747302114963531, "learning_rate": 8.05167186310758e-05, "loss": 0.0036, "step": 19590 }, { "grad_norm": 0.12513144314289093, "learning_rate": 8.049488423001113e-05, "loss": 0.0046, "step": 19600 }, { "grad_norm": 0.11611704528331757, "learning_rate": 8.047304056540581e-05, "loss": 0.0048, "step": 19610 }, { "grad_norm": 0.10802311450242996, "learning_rate": 8.045118764389534e-05, "loss": 0.0043, "step": 19620 }, { "grad_norm": 0.10956285893917084, "learning_rate": 8.042932547211809e-05, "loss": 0.0053, "step": 19630 }, { "grad_norm": 0.14515653252601624, "learning_rate": 8.04074540567152e-05, "loss": 0.0037, "step": 19640 }, { "grad_norm": 0.10209812223911285, "learning_rate": 8.038557340433063e-05, "loss": 0.0056, "step": 19650 }, { "grad_norm": 0.1279546618461609, "learning_rate": 8.036368352161115e-05, "loss": 0.0061, "step": 19660 }, { "grad_norm": 0.15137866139411926, "learning_rate": 8.034178441520633e-05, "loss": 0.0068, "step": 19670 }, { "grad_norm": 0.12733231484889984, "learning_rate": 8.031987609176852e-05, "loss": 0.0043, "step": 19680 }, { "grad_norm": 0.11236051470041275, "learning_rate": 8.02979585579529e-05, "loss": 0.0044, "step": 19690 }, { "grad_norm": 0.12176381051540375, "learning_rate": 8.027603182041745e-05, "loss": 0.0041, "step": 19700 }, { "grad_norm": 0.07899688929319382, "learning_rate": 8.025409588582292e-05, "loss": 0.004, "step": 19710 }, { "grad_norm": 0.1417156159877777, "learning_rate": 8.023215076083288e-05, "loss": 0.0047, "step": 19720 }, { "grad_norm": 0.11730936169624329, "learning_rate": 8.021019645211367e-05, "loss": 0.0046, "step": 19730 }, { "grad_norm": 0.10869941860437393, "learning_rate": 8.018823296633441e-05, "loss": 0.005, "step": 19740 }, { "grad_norm": 0.14746445417404175, "learning_rate": 8.016626031016708e-05, "loss": 0.0041, "step": 19750 }, { "grad_norm": 0.1656462550163269, "learning_rate": 8.014427849028636e-05, "loss": 0.0062, "step": 19760 }, { "grad_norm": 0.12122154980897903, "learning_rate": 8.012228751336974e-05, "loss": 0.0042, "step": 19770 }, { "grad_norm": 0.12198203802108765, "learning_rate": 8.01002873860975e-05, "loss": 0.0039, "step": 19780 }, { "grad_norm": 0.1290741115808487, "learning_rate": 8.00782781151527e-05, "loss": 0.0047, "step": 19790 }, { "grad_norm": 0.08780614286661148, "learning_rate": 8.005625970722119e-05, "loss": 0.0047, "step": 19800 }, { "grad_norm": 0.11314620822668076, "learning_rate": 8.003423216899158e-05, "loss": 0.0046, "step": 19810 }, { "grad_norm": 0.11690600216388702, "learning_rate": 8.001219550715522e-05, "loss": 0.0039, "step": 19820 }, { "grad_norm": 0.11488846689462662, "learning_rate": 7.999014972840632e-05, "loss": 0.0055, "step": 19830 }, { "grad_norm": 0.08360697329044342, "learning_rate": 7.996809483944174e-05, "loss": 0.0045, "step": 19840 }, { "grad_norm": 0.1326439082622528, "learning_rate": 7.994603084696124e-05, "loss": 0.0045, "step": 19850 }, { "grad_norm": 0.12676291167736053, "learning_rate": 7.992395775766724e-05, "loss": 0.0053, "step": 19860 }, { "grad_norm": 0.07187412679195404, "learning_rate": 7.990187557826497e-05, "loss": 0.0055, "step": 19870 }, { "grad_norm": 0.09741921722888947, "learning_rate": 7.987978431546242e-05, "loss": 0.0055, "step": 19880 }, { "grad_norm": 0.11666185408830643, "learning_rate": 7.985768397597031e-05, "loss": 0.0044, "step": 19890 }, { "grad_norm": 0.10619828850030899, "learning_rate": 7.983557456650216e-05, "loss": 0.0041, "step": 19900 }, { "grad_norm": 0.1140829548239708, "learning_rate": 7.981345609377422e-05, "loss": 0.0047, "step": 19910 }, { "grad_norm": 0.11159661412239075, "learning_rate": 7.97913285645055e-05, "loss": 0.0051, "step": 19920 }, { "grad_norm": 0.13771270215511322, "learning_rate": 7.976919198541776e-05, "loss": 0.0057, "step": 19930 }, { "grad_norm": 0.1274154782295227, "learning_rate": 7.974704636323548e-05, "loss": 0.0045, "step": 19940 }, { "grad_norm": 0.12200955301523209, "learning_rate": 7.972489170468597e-05, "loss": 0.0047, "step": 19950 }, { "grad_norm": 0.1255652904510498, "learning_rate": 7.970272801649918e-05, "loss": 0.0067, "step": 19960 }, { "grad_norm": 0.12467169016599655, "learning_rate": 7.96805553054079e-05, "loss": 0.0046, "step": 19970 }, { "grad_norm": 0.08556169271469116, "learning_rate": 7.965837357814756e-05, "loss": 0.0051, "step": 19980 }, { "grad_norm": 0.09165595471858978, "learning_rate": 7.963618284145643e-05, "loss": 0.0044, "step": 19990 }, { "grad_norm": 0.09269537031650543, "learning_rate": 7.961398310207544e-05, "loss": 0.0054, "step": 20000 }, { "grad_norm": 0.09166662395000458, "learning_rate": 7.95917743667483e-05, "loss": 0.0039, "step": 20010 }, { "grad_norm": 0.11079510301351547, "learning_rate": 7.956955664222144e-05, "loss": 0.0041, "step": 20020 }, { "grad_norm": 0.10443605482578278, "learning_rate": 7.954732993524399e-05, "loss": 0.0043, "step": 20030 }, { "grad_norm": 0.13122469186782837, "learning_rate": 7.952509425256786e-05, "loss": 0.0044, "step": 20040 }, { "grad_norm": 0.10046558827161789, "learning_rate": 7.950284960094767e-05, "loss": 0.0034, "step": 20050 }, { "grad_norm": 0.0989207997918129, "learning_rate": 7.948059598714076e-05, "loss": 0.0034, "step": 20060 }, { "grad_norm": 0.10260037332773209, "learning_rate": 7.945833341790717e-05, "loss": 0.0041, "step": 20070 }, { "grad_norm": 0.13625651597976685, "learning_rate": 7.94360619000097e-05, "loss": 0.004, "step": 20080 }, { "grad_norm": 0.08820246160030365, "learning_rate": 7.941378144021381e-05, "loss": 0.004, "step": 20090 }, { "grad_norm": 0.07492834329605103, "learning_rate": 7.939149204528777e-05, "loss": 0.0044, "step": 20100 }, { "grad_norm": 0.10251862555742264, "learning_rate": 7.936919372200246e-05, "loss": 0.0044, "step": 20110 }, { "grad_norm": 0.11496394872665405, "learning_rate": 7.934688647713158e-05, "loss": 0.0046, "step": 20120 }, { "grad_norm": 0.11540678143501282, "learning_rate": 7.932457031745143e-05, "loss": 0.0049, "step": 20130 }, { "grad_norm": 0.10543014854192734, "learning_rate": 7.930224524974108e-05, "loss": 0.0042, "step": 20140 }, { "grad_norm": 0.1252148300409317, "learning_rate": 7.927991128078232e-05, "loss": 0.0047, "step": 20150 }, { "grad_norm": 0.1274178922176361, "learning_rate": 7.925756841735958e-05, "loss": 0.0045, "step": 20160 }, { "grad_norm": 0.12980866432189941, "learning_rate": 7.923521666626008e-05, "loss": 0.0054, "step": 20170 }, { "grad_norm": 0.12015727907419205, "learning_rate": 7.921285603427366e-05, "loss": 0.0049, "step": 20180 }, { "grad_norm": 0.10685417056083679, "learning_rate": 7.91904865281929e-05, "loss": 0.0047, "step": 20190 }, { "grad_norm": 0.07967758178710938, "learning_rate": 7.916810815481307e-05, "loss": 0.004, "step": 20200 }, { "grad_norm": 0.1162666380405426, "learning_rate": 7.914572092093211e-05, "loss": 0.0045, "step": 20210 }, { "grad_norm": 0.1278110146522522, "learning_rate": 7.912332483335068e-05, "loss": 0.0047, "step": 20220 }, { "grad_norm": 0.09123413264751434, "learning_rate": 7.910091989887213e-05, "loss": 0.0043, "step": 20230 }, { "grad_norm": 0.11756336688995361, "learning_rate": 7.907850612430248e-05, "loss": 0.0045, "step": 20240 }, { "grad_norm": 0.11875477433204651, "learning_rate": 7.905608351645044e-05, "loss": 0.0045, "step": 20250 }, { "grad_norm": 0.1485244780778885, "learning_rate": 7.90336520821274e-05, "loss": 0.0049, "step": 20260 }, { "grad_norm": 0.0692322850227356, "learning_rate": 7.901121182814746e-05, "loss": 0.0048, "step": 20270 }, { "grad_norm": 0.08476005494594574, "learning_rate": 7.898876276132736e-05, "loss": 0.0047, "step": 20280 }, { "grad_norm": 0.13677085936069489, "learning_rate": 7.896630488848654e-05, "loss": 0.0042, "step": 20290 }, { "grad_norm": 0.1342737376689911, "learning_rate": 7.89438382164471e-05, "loss": 0.0054, "step": 20300 }, { "grad_norm": 0.10511422902345657, "learning_rate": 7.892136275203383e-05, "loss": 0.0048, "step": 20310 }, { "grad_norm": 0.09764520823955536, "learning_rate": 7.889887850207418e-05, "loss": 0.0035, "step": 20320 }, { "grad_norm": 0.10938043147325516, "learning_rate": 7.887638547339827e-05, "loss": 0.0039, "step": 20330 }, { "grad_norm": 0.10028979927301407, "learning_rate": 7.885388367283891e-05, "loss": 0.0042, "step": 20340 }, { "grad_norm": 0.10710888355970383, "learning_rate": 7.88313731072315e-05, "loss": 0.0042, "step": 20350 }, { "grad_norm": 0.10620861500501633, "learning_rate": 7.88088537834142e-05, "loss": 0.0058, "step": 20360 }, { "grad_norm": 0.13577638566493988, "learning_rate": 7.878632570822778e-05, "loss": 0.0054, "step": 20370 }, { "grad_norm": 0.1512564867734909, "learning_rate": 7.876378888851567e-05, "loss": 0.0065, "step": 20380 }, { "grad_norm": 0.09824591130018234, "learning_rate": 7.874124333112396e-05, "loss": 0.0045, "step": 20390 }, { "grad_norm": 0.13106068968772888, "learning_rate": 7.871868904290138e-05, "loss": 0.005, "step": 20400 }, { "grad_norm": 0.11701405048370361, "learning_rate": 7.869612603069935e-05, "loss": 0.0041, "step": 20410 }, { "grad_norm": 0.08838962018489838, "learning_rate": 7.867355430137192e-05, "loss": 0.0041, "step": 20420 }, { "grad_norm": 0.10352303832769394, "learning_rate": 7.865097386177577e-05, "loss": 0.0039, "step": 20430 }, { "grad_norm": 0.09559262543916702, "learning_rate": 7.862838471877023e-05, "loss": 0.004, "step": 20440 }, { "grad_norm": 0.10916736721992493, "learning_rate": 7.860578687921731e-05, "loss": 0.0044, "step": 20450 }, { "grad_norm": 0.0820680558681488, "learning_rate": 7.858318034998164e-05, "loss": 0.0038, "step": 20460 }, { "grad_norm": 0.08464673906564713, "learning_rate": 7.856056513793046e-05, "loss": 0.0042, "step": 20470 }, { "grad_norm": 0.11159292608499527, "learning_rate": 7.85379412499337e-05, "loss": 0.0064, "step": 20480 }, { "grad_norm": 0.1631421446800232, "learning_rate": 7.851530869286389e-05, "loss": 0.0047, "step": 20490 }, { "grad_norm": 0.09300641715526581, "learning_rate": 7.849266747359619e-05, "loss": 0.0037, "step": 20500 }, { "grad_norm": 0.09915776550769806, "learning_rate": 7.847001759900843e-05, "loss": 0.0042, "step": 20510 }, { "grad_norm": 0.08600058406591415, "learning_rate": 7.844735907598102e-05, "loss": 0.0035, "step": 20520 }, { "grad_norm": 0.0780024603009224, "learning_rate": 7.842469191139703e-05, "loss": 0.0043, "step": 20530 }, { "grad_norm": 0.16968964040279388, "learning_rate": 7.840201611214215e-05, "loss": 0.0051, "step": 20540 }, { "grad_norm": 0.16390706598758698, "learning_rate": 7.837933168510469e-05, "loss": 0.0046, "step": 20550 }, { "grad_norm": 0.1435943841934204, "learning_rate": 7.835663863717559e-05, "loss": 0.0048, "step": 20560 }, { "grad_norm": 0.1522931456565857, "learning_rate": 7.833393697524838e-05, "loss": 0.0045, "step": 20570 }, { "grad_norm": 0.11244078725576401, "learning_rate": 7.831122670621922e-05, "loss": 0.0054, "step": 20580 }, { "grad_norm": 0.11028560250997543, "learning_rate": 7.82885078369869e-05, "loss": 0.0047, "step": 20590 }, { "grad_norm": 0.12226146459579468, "learning_rate": 7.826578037445283e-05, "loss": 0.0049, "step": 20600 }, { "grad_norm": 0.09699321538209915, "learning_rate": 7.824304432552097e-05, "loss": 0.0056, "step": 20610 }, { "grad_norm": 0.07450979202985764, "learning_rate": 7.822029969709798e-05, "loss": 0.0037, "step": 20620 }, { "grad_norm": 0.1120205819606781, "learning_rate": 7.819754649609306e-05, "loss": 0.0049, "step": 20630 }, { "grad_norm": 0.13145561516284943, "learning_rate": 7.817478472941802e-05, "loss": 0.0039, "step": 20640 }, { "grad_norm": 0.09815659373998642, "learning_rate": 7.815201440398727e-05, "loss": 0.0041, "step": 20650 }, { "grad_norm": 0.1432807594537735, "learning_rate": 7.812923552671789e-05, "loss": 0.0043, "step": 20660 }, { "grad_norm": 0.10269825905561447, "learning_rate": 7.810644810452945e-05, "loss": 0.0052, "step": 20670 }, { "grad_norm": 0.15828962624073029, "learning_rate": 7.808365214434417e-05, "loss": 0.0056, "step": 20680 }, { "grad_norm": 0.1391712874174118, "learning_rate": 7.80608476530869e-05, "loss": 0.0053, "step": 20690 }, { "grad_norm": 0.09289497137069702, "learning_rate": 7.8038034637685e-05, "loss": 0.0056, "step": 20700 }, { "grad_norm": 0.11996901780366898, "learning_rate": 7.801521310506848e-05, "loss": 0.0054, "step": 20710 }, { "grad_norm": 0.10969211161136627, "learning_rate": 7.799238306216994e-05, "loss": 0.0055, "step": 20720 }, { "grad_norm": 0.09220799058675766, "learning_rate": 7.796954451592448e-05, "loss": 0.0047, "step": 20730 }, { "grad_norm": 0.10057450085878372, "learning_rate": 7.794669747326992e-05, "loss": 0.0036, "step": 20740 }, { "grad_norm": 0.09036066383123398, "learning_rate": 7.792384194114654e-05, "loss": 0.0045, "step": 20750 }, { "grad_norm": 0.13540144264698029, "learning_rate": 7.790097792649729e-05, "loss": 0.0039, "step": 20760 }, { "grad_norm": 0.12061364948749542, "learning_rate": 7.787810543626762e-05, "loss": 0.0047, "step": 20770 }, { "grad_norm": 0.13933967053890228, "learning_rate": 7.785522447740558e-05, "loss": 0.0044, "step": 20780 }, { "grad_norm": 0.0939669981598854, "learning_rate": 7.783233505686182e-05, "loss": 0.0039, "step": 20790 }, { "grad_norm": 0.0915951058268547, "learning_rate": 7.780943718158955e-05, "loss": 0.0042, "step": 20800 }, { "grad_norm": 0.148742213845253, "learning_rate": 7.778653085854453e-05, "loss": 0.0049, "step": 20810 }, { "grad_norm": 0.11277909576892853, "learning_rate": 7.77636160946851e-05, "loss": 0.0045, "step": 20820 }, { "grad_norm": 0.1401667445898056, "learning_rate": 7.774069289697215e-05, "loss": 0.0054, "step": 20830 }, { "grad_norm": 0.09413991868495941, "learning_rate": 7.771776127236913e-05, "loss": 0.0044, "step": 20840 }, { "grad_norm": 0.10045252740383148, "learning_rate": 7.769482122784212e-05, "loss": 0.0051, "step": 20850 }, { "grad_norm": 0.09261435270309448, "learning_rate": 7.767187277035963e-05, "loss": 0.0047, "step": 20860 }, { "grad_norm": 0.09642026573419571, "learning_rate": 7.764891590689285e-05, "loss": 0.0044, "step": 20870 }, { "grad_norm": 0.10654828697443008, "learning_rate": 7.762595064441542e-05, "loss": 0.0041, "step": 20880 }, { "grad_norm": 0.08920629322528839, "learning_rate": 7.760297698990362e-05, "loss": 0.0039, "step": 20890 }, { "grad_norm": 0.1533055603504181, "learning_rate": 7.757999495033623e-05, "loss": 0.0044, "step": 20900 }, { "grad_norm": 0.13554339110851288, "learning_rate": 7.755700453269456e-05, "loss": 0.0056, "step": 20910 }, { "grad_norm": 0.11549925804138184, "learning_rate": 7.753400574396254e-05, "loss": 0.0054, "step": 20920 }, { "grad_norm": 0.08970524370670319, "learning_rate": 7.751099859112655e-05, "loss": 0.0042, "step": 20930 }, { "grad_norm": 0.13848228752613068, "learning_rate": 7.748798308117557e-05, "loss": 0.0059, "step": 20940 }, { "grad_norm": 0.09918537735939026, "learning_rate": 7.746495922110112e-05, "loss": 0.0047, "step": 20950 }, { "grad_norm": 0.11693954467773438, "learning_rate": 7.744192701789723e-05, "loss": 0.0046, "step": 20960 }, { "grad_norm": 0.09934603422880173, "learning_rate": 7.741888647856046e-05, "loss": 0.0046, "step": 20970 }, { "grad_norm": 0.09099900722503662, "learning_rate": 7.739583761008994e-05, "loss": 0.0046, "step": 20980 }, { "grad_norm": 0.0979282483458519, "learning_rate": 7.73727804194873e-05, "loss": 0.0045, "step": 20990 }, { "grad_norm": 0.09612713009119034, "learning_rate": 7.734971491375671e-05, "loss": 0.0038, "step": 21000 }, { "grad_norm": 0.13253089785575867, "learning_rate": 7.732664109990485e-05, "loss": 0.0046, "step": 21010 }, { "grad_norm": 0.11761732399463654, "learning_rate": 7.730355898494095e-05, "loss": 0.0047, "step": 21020 }, { "grad_norm": 0.12074774503707886, "learning_rate": 7.728046857587673e-05, "loss": 0.0042, "step": 21030 }, { "grad_norm": 0.11098058521747589, "learning_rate": 7.725736987972647e-05, "loss": 0.0043, "step": 21040 }, { "grad_norm": 0.11080074310302734, "learning_rate": 7.723426290350691e-05, "loss": 0.0039, "step": 21050 }, { "grad_norm": 0.10189326107501984, "learning_rate": 7.721114765423736e-05, "loss": 0.004, "step": 21060 }, { "grad_norm": 0.09275773912668228, "learning_rate": 7.718802413893963e-05, "loss": 0.0034, "step": 21070 }, { "grad_norm": 0.11444063484668732, "learning_rate": 7.716489236463802e-05, "loss": 0.0047, "step": 21080 }, { "grad_norm": 0.1427702009677887, "learning_rate": 7.714175233835936e-05, "loss": 0.004, "step": 21090 }, { "grad_norm": 0.11103298515081406, "learning_rate": 7.711860406713299e-05, "loss": 0.0041, "step": 21100 }, { "grad_norm": 0.07817143946886063, "learning_rate": 7.70954475579907e-05, "loss": 0.0035, "step": 21110 }, { "grad_norm": 0.09775322675704956, "learning_rate": 7.707228281796688e-05, "loss": 0.004, "step": 21120 }, { "grad_norm": 0.11584832519292831, "learning_rate": 7.704910985409833e-05, "loss": 0.0047, "step": 21130 }, { "grad_norm": 0.11595668643712997, "learning_rate": 7.702592867342439e-05, "loss": 0.0041, "step": 21140 }, { "grad_norm": 0.12551242113113403, "learning_rate": 7.700273928298691e-05, "loss": 0.0042, "step": 21150 }, { "grad_norm": 0.10026446729898453, "learning_rate": 7.697954168983021e-05, "loss": 0.0042, "step": 21160 }, { "grad_norm": 0.12787725031375885, "learning_rate": 7.695633590100109e-05, "loss": 0.0043, "step": 21170 }, { "grad_norm": 0.08370531350374222, "learning_rate": 7.693312192354886e-05, "loss": 0.0038, "step": 21180 }, { "grad_norm": 0.17118331789970398, "learning_rate": 7.690989976452532e-05, "loss": 0.0038, "step": 21190 }, { "grad_norm": 0.10372193157672882, "learning_rate": 7.688666943098475e-05, "loss": 0.004, "step": 21200 }, { "grad_norm": 0.0822589322924614, "learning_rate": 7.686343092998389e-05, "loss": 0.0043, "step": 21210 }, { "grad_norm": 0.12133604288101196, "learning_rate": 7.684018426858202e-05, "loss": 0.0039, "step": 21220 }, { "grad_norm": 0.12112562358379364, "learning_rate": 7.681692945384084e-05, "loss": 0.0048, "step": 21230 }, { "grad_norm": 0.10938675701618195, "learning_rate": 7.679366649282456e-05, "loss": 0.0047, "step": 21240 }, { "grad_norm": 0.10531029105186462, "learning_rate": 7.677039539259983e-05, "loss": 0.0038, "step": 21250 }, { "grad_norm": 0.07967208325862885, "learning_rate": 7.674711616023581e-05, "loss": 0.0037, "step": 21260 }, { "grad_norm": 0.09451541304588318, "learning_rate": 7.672382880280413e-05, "loss": 0.0041, "step": 21270 }, { "grad_norm": 0.0937623679637909, "learning_rate": 7.670053332737885e-05, "loss": 0.0037, "step": 21280 }, { "grad_norm": 0.12740746140480042, "learning_rate": 7.667722974103654e-05, "loss": 0.0038, "step": 21290 }, { "grad_norm": 0.068968765437603, "learning_rate": 7.66539180508562e-05, "loss": 0.004, "step": 21300 }, { "grad_norm": 0.11503585427999496, "learning_rate": 7.663059826391932e-05, "loss": 0.0064, "step": 21310 }, { "grad_norm": 0.10776820033788681, "learning_rate": 7.660727038730981e-05, "loss": 0.0049, "step": 21320 }, { "grad_norm": 0.12830446660518646, "learning_rate": 7.65839344281141e-05, "loss": 0.0051, "step": 21330 }, { "grad_norm": 0.11449817568063736, "learning_rate": 7.656059039342101e-05, "loss": 0.005, "step": 21340 }, { "grad_norm": 0.11006271094083786, "learning_rate": 7.653723829032187e-05, "loss": 0.0053, "step": 21350 }, { "grad_norm": 0.109852135181427, "learning_rate": 7.65138781259104e-05, "loss": 0.0044, "step": 21360 }, { "grad_norm": 0.09107454866170883, "learning_rate": 7.649050990728279e-05, "loss": 0.0042, "step": 21370 }, { "grad_norm": 0.0922350212931633, "learning_rate": 7.646713364153774e-05, "loss": 0.004, "step": 21380 }, { "grad_norm": 0.10476882010698318, "learning_rate": 7.64437493357763e-05, "loss": 0.0042, "step": 21390 }, { "grad_norm": 0.09962932765483856, "learning_rate": 7.642035699710202e-05, "loss": 0.0049, "step": 21400 }, { "grad_norm": 0.13227708637714386, "learning_rate": 7.639695663262089e-05, "loss": 0.0044, "step": 21410 }, { "grad_norm": 0.10051936656236649, "learning_rate": 7.637354824944128e-05, "loss": 0.0048, "step": 21420 }, { "grad_norm": 0.13482819497585297, "learning_rate": 7.635013185467408e-05, "loss": 0.0044, "step": 21430 }, { "grad_norm": 0.10943890362977982, "learning_rate": 7.632670745543256e-05, "loss": 0.0041, "step": 21440 }, { "grad_norm": 0.08129284530878067, "learning_rate": 7.630327505883242e-05, "loss": 0.0036, "step": 21450 }, { "grad_norm": 0.14055223762989044, "learning_rate": 7.627983467199182e-05, "loss": 0.0043, "step": 21460 }, { "grad_norm": 0.11135319620370865, "learning_rate": 7.625638630203132e-05, "loss": 0.005, "step": 21470 }, { "grad_norm": 0.09030419588088989, "learning_rate": 7.623292995607394e-05, "loss": 0.0042, "step": 21480 }, { "grad_norm": 0.11595139652490616, "learning_rate": 7.620946564124507e-05, "loss": 0.0043, "step": 21490 }, { "grad_norm": 0.08424574136734009, "learning_rate": 7.618599336467256e-05, "loss": 0.004, "step": 21500 }, { "grad_norm": 0.1801808625459671, "learning_rate": 7.616251313348666e-05, "loss": 0.0046, "step": 21510 }, { "grad_norm": 0.11832361668348312, "learning_rate": 7.613902495482005e-05, "loss": 0.0048, "step": 21520 }, { "grad_norm": 0.12461630254983902, "learning_rate": 7.611552883580784e-05, "loss": 0.0041, "step": 21530 }, { "grad_norm": 0.06875711679458618, "learning_rate": 7.609202478358748e-05, "loss": 0.0037, "step": 21540 }, { "grad_norm": 0.12757675349712372, "learning_rate": 7.606851280529895e-05, "loss": 0.0039, "step": 21550 }, { "grad_norm": 0.10005826503038406, "learning_rate": 7.604499290808449e-05, "loss": 0.0038, "step": 21560 }, { "grad_norm": 0.0942414402961731, "learning_rate": 7.602146509908888e-05, "loss": 0.0041, "step": 21570 }, { "grad_norm": 0.1165151372551918, "learning_rate": 7.599792938545921e-05, "loss": 0.0048, "step": 21580 }, { "grad_norm": 0.1310691237449646, "learning_rate": 7.597438577434506e-05, "loss": 0.0044, "step": 21590 }, { "grad_norm": 0.10152129083871841, "learning_rate": 7.595083427289831e-05, "loss": 0.0038, "step": 21600 }, { "grad_norm": 0.13690294325351715, "learning_rate": 7.59272748882733e-05, "loss": 0.0045, "step": 21610 }, { "grad_norm": 0.12338243424892426, "learning_rate": 7.590370762762675e-05, "loss": 0.0037, "step": 21620 }, { "grad_norm": 0.12043513357639313, "learning_rate": 7.588013249811777e-05, "loss": 0.0044, "step": 21630 }, { "grad_norm": 0.115561343729496, "learning_rate": 7.585654950690786e-05, "loss": 0.0036, "step": 21640 }, { "grad_norm": 0.0940556451678276, "learning_rate": 7.583295866116091e-05, "loss": 0.0038, "step": 21650 }, { "grad_norm": 0.07679321616888046, "learning_rate": 7.580935996804321e-05, "loss": 0.0038, "step": 21660 }, { "grad_norm": 0.10354890674352646, "learning_rate": 7.57857534347234e-05, "loss": 0.0042, "step": 21670 }, { "grad_norm": 0.08424321562051773, "learning_rate": 7.576213906837254e-05, "loss": 0.0039, "step": 21680 }, { "grad_norm": 0.072038434445858, "learning_rate": 7.573851687616403e-05, "loss": 0.0037, "step": 21690 }, { "grad_norm": 0.136203333735466, "learning_rate": 7.571488686527368e-05, "loss": 0.0044, "step": 21700 }, { "grad_norm": 0.11180093139410019, "learning_rate": 7.569124904287968e-05, "loss": 0.004, "step": 21710 }, { "grad_norm": 0.1176203191280365, "learning_rate": 7.566760341616254e-05, "loss": 0.0049, "step": 21720 }, { "grad_norm": 0.18779459595680237, "learning_rate": 7.564394999230519e-05, "loss": 0.0047, "step": 21730 }, { "grad_norm": 0.10855009406805038, "learning_rate": 7.562028877849294e-05, "loss": 0.0052, "step": 21740 }, { "grad_norm": 0.14495250582695007, "learning_rate": 7.559661978191341e-05, "loss": 0.0055, "step": 21750 }, { "grad_norm": 0.1158025860786438, "learning_rate": 7.557294300975664e-05, "loss": 0.0047, "step": 21760 }, { "grad_norm": 0.10495083034038544, "learning_rate": 7.554925846921499e-05, "loss": 0.0044, "step": 21770 }, { "grad_norm": 0.08738025277853012, "learning_rate": 7.552556616748321e-05, "loss": 0.0051, "step": 21780 }, { "grad_norm": 0.09116768836975098, "learning_rate": 7.550186611175838e-05, "loss": 0.0034, "step": 21790 }, { "grad_norm": 0.09138435870409012, "learning_rate": 7.547815830923998e-05, "loss": 0.0038, "step": 21800 }, { "grad_norm": 0.1078033372759819, "learning_rate": 7.54544427671298e-05, "loss": 0.0041, "step": 21810 }, { "grad_norm": 0.0828215479850769, "learning_rate": 7.543071949263198e-05, "loss": 0.004, "step": 21820 }, { "grad_norm": 0.12016190588474274, "learning_rate": 7.540698849295305e-05, "loss": 0.0046, "step": 21830 }, { "grad_norm": 0.11962045729160309, "learning_rate": 7.538324977530183e-05, "loss": 0.004, "step": 21840 }, { "grad_norm": 0.0936567485332489, "learning_rate": 7.535950334688955e-05, "loss": 0.005, "step": 21850 }, { "grad_norm": 0.11921956390142441, "learning_rate": 7.533574921492972e-05, "loss": 0.0045, "step": 21860 }, { "grad_norm": 0.09210696816444397, "learning_rate": 7.531198738663824e-05, "loss": 0.0045, "step": 21870 }, { "grad_norm": 0.127737358212471, "learning_rate": 7.528821786923333e-05, "loss": 0.0041, "step": 21880 }, { "grad_norm": 0.10303749889135361, "learning_rate": 7.52644406699355e-05, "loss": 0.0045, "step": 21890 }, { "grad_norm": 0.1254042237997055, "learning_rate": 7.524065579596766e-05, "loss": 0.0038, "step": 21900 }, { "grad_norm": 0.12251941114664078, "learning_rate": 7.521686325455506e-05, "loss": 0.0044, "step": 21910 }, { "grad_norm": 0.15726861357688904, "learning_rate": 7.51930630529252e-05, "loss": 0.0037, "step": 21920 }, { "grad_norm": 0.11191677302122116, "learning_rate": 7.516925519830797e-05, "loss": 0.0046, "step": 21930 }, { "grad_norm": 0.12215246260166168, "learning_rate": 7.514543969793557e-05, "loss": 0.004, "step": 21940 }, { "grad_norm": 0.11304830014705658, "learning_rate": 7.512161655904251e-05, "loss": 0.0041, "step": 21950 }, { "grad_norm": 0.08961133658885956, "learning_rate": 7.509778578886563e-05, "loss": 0.0035, "step": 21960 }, { "grad_norm": 0.1820497065782547, "learning_rate": 7.507394739464412e-05, "loss": 0.0078, "step": 21970 }, { "grad_norm": 0.09963054209947586, "learning_rate": 7.50501013836194e-05, "loss": 0.0066, "step": 21980 }, { "grad_norm": 0.09271962940692902, "learning_rate": 7.50262477630353e-05, "loss": 0.0051, "step": 21990 }, { "grad_norm": 0.13157354295253754, "learning_rate": 7.500238654013794e-05, "loss": 0.0052, "step": 22000 }, { "grad_norm": 0.09894653409719467, "learning_rate": 7.497851772217566e-05, "loss": 0.0042, "step": 22010 }, { "grad_norm": 0.1366175413131714, "learning_rate": 7.495464131639924e-05, "loss": 0.0043, "step": 22020 }, { "grad_norm": 0.11767341941595078, "learning_rate": 7.493075733006166e-05, "loss": 0.0045, "step": 22030 }, { "grad_norm": 0.08400294184684753, "learning_rate": 7.490686577041828e-05, "loss": 0.0037, "step": 22040 }, { "grad_norm": 0.12161237746477127, "learning_rate": 7.488296664472668e-05, "loss": 0.004, "step": 22050 }, { "grad_norm": 0.10109543055295944, "learning_rate": 7.485905996024682e-05, "loss": 0.0035, "step": 22060 }, { "grad_norm": 0.09263516962528229, "learning_rate": 7.483514572424093e-05, "loss": 0.0037, "step": 22070 }, { "grad_norm": 0.1019010841846466, "learning_rate": 7.481122394397349e-05, "loss": 0.0041, "step": 22080 }, { "grad_norm": 0.10604467242956161, "learning_rate": 7.478729462671131e-05, "loss": 0.0043, "step": 22090 }, { "grad_norm": 0.12478172034025192, "learning_rate": 7.47633577797235e-05, "loss": 0.004, "step": 22100 }, { "grad_norm": 0.10098453611135483, "learning_rate": 7.473941341028144e-05, "loss": 0.0041, "step": 22110 }, { "grad_norm": 0.11599148064851761, "learning_rate": 7.471546152565879e-05, "loss": 0.0036, "step": 22120 }, { "grad_norm": 0.12458407133817673, "learning_rate": 7.46915021331315e-05, "loss": 0.0047, "step": 22130 }, { "grad_norm": 0.13333269953727722, "learning_rate": 7.466753523997778e-05, "loss": 0.0042, "step": 22140 }, { "grad_norm": 0.08553732931613922, "learning_rate": 7.464356085347819e-05, "loss": 0.0038, "step": 22150 }, { "grad_norm": 0.11024971306324005, "learning_rate": 7.461957898091548e-05, "loss": 0.0046, "step": 22160 }, { "grad_norm": 0.11233661323785782, "learning_rate": 7.459558962957473e-05, "loss": 0.0039, "step": 22170 }, { "grad_norm": 0.11443407833576202, "learning_rate": 7.457159280674326e-05, "loss": 0.0043, "step": 22180 }, { "grad_norm": 0.08860138803720474, "learning_rate": 7.454758851971066e-05, "loss": 0.0038, "step": 22190 }, { "grad_norm": 0.10148967057466507, "learning_rate": 7.45235767757688e-05, "loss": 0.0043, "step": 22200 }, { "grad_norm": 0.07762236148118973, "learning_rate": 7.449955758221183e-05, "loss": 0.0051, "step": 22210 }, { "grad_norm": 0.10945369303226471, "learning_rate": 7.447553094633615e-05, "loss": 0.004, "step": 22220 }, { "grad_norm": 0.09296092391014099, "learning_rate": 7.445149687544039e-05, "loss": 0.0035, "step": 22230 }, { "grad_norm": 0.09306955337524414, "learning_rate": 7.44274553768255e-05, "loss": 0.0035, "step": 22240 }, { "grad_norm": 0.10284070670604706, "learning_rate": 7.440340645779464e-05, "loss": 0.0038, "step": 22250 }, { "grad_norm": 0.1060362458229065, "learning_rate": 7.437935012565322e-05, "loss": 0.0032, "step": 22260 }, { "grad_norm": 0.12612669169902802, "learning_rate": 7.435528638770893e-05, "loss": 0.0043, "step": 22270 }, { "grad_norm": 0.07252784818410873, "learning_rate": 7.433121525127171e-05, "loss": 0.0045, "step": 22280 }, { "grad_norm": 0.07673047482967377, "learning_rate": 7.430713672365371e-05, "loss": 0.0035, "step": 22290 }, { "grad_norm": 0.09918361902236938, "learning_rate": 7.428305081216938e-05, "loss": 0.0051, "step": 22300 }, { "grad_norm": 0.11641882359981537, "learning_rate": 7.425895752413536e-05, "loss": 0.0042, "step": 22310 }, { "grad_norm": 0.09076382964849472, "learning_rate": 7.423485686687057e-05, "loss": 0.0047, "step": 22320 }, { "grad_norm": 0.14617596566677094, "learning_rate": 7.421074884769616e-05, "loss": 0.0047, "step": 22330 }, { "grad_norm": 0.11842937767505646, "learning_rate": 7.418663347393548e-05, "loss": 0.004, "step": 22340 }, { "grad_norm": 0.11926813423633575, "learning_rate": 7.416251075291418e-05, "loss": 0.0056, "step": 22350 }, { "grad_norm": 0.09532985091209412, "learning_rate": 7.413838069196007e-05, "loss": 0.0035, "step": 22360 }, { "grad_norm": 0.10371214896440506, "learning_rate": 7.411424329840324e-05, "loss": 0.0046, "step": 22370 }, { "grad_norm": 0.12787756323814392, "learning_rate": 7.409009857957601e-05, "loss": 0.004, "step": 22380 }, { "grad_norm": 0.09848977625370026, "learning_rate": 7.40659465428129e-05, "loss": 0.0052, "step": 22390 }, { "grad_norm": 0.11314862966537476, "learning_rate": 7.404178719545063e-05, "loss": 0.0052, "step": 22400 }, { "grad_norm": 0.12320855259895325, "learning_rate": 7.401762054482822e-05, "loss": 0.0061, "step": 22410 }, { "grad_norm": 0.10413375496864319, "learning_rate": 7.39934465982868e-05, "loss": 0.0059, "step": 22420 }, { "grad_norm": 0.10744421184062958, "learning_rate": 7.396926536316984e-05, "loss": 0.0045, "step": 22430 }, { "grad_norm": 0.1094822883605957, "learning_rate": 7.394507684682293e-05, "loss": 0.005, "step": 22440 }, { "grad_norm": 0.13682281970977783, "learning_rate": 7.392088105659393e-05, "loss": 0.0056, "step": 22450 }, { "grad_norm": 0.0989609882235527, "learning_rate": 7.389667799983284e-05, "loss": 0.0052, "step": 22460 }, { "grad_norm": 0.10290133953094482, "learning_rate": 7.387246768389193e-05, "loss": 0.0038, "step": 22470 }, { "grad_norm": 0.09308113157749176, "learning_rate": 7.384825011612563e-05, "loss": 0.0035, "step": 22480 }, { "grad_norm": 0.10000839084386826, "learning_rate": 7.382402530389066e-05, "loss": 0.0034, "step": 22490 }, { "grad_norm": 0.17662574350833893, "learning_rate": 7.379979325454582e-05, "loss": 0.005, "step": 22500 }, { "grad_norm": 0.17986339330673218, "learning_rate": 7.37755539754522e-05, "loss": 0.0056, "step": 22510 }, { "grad_norm": 0.12976837158203125, "learning_rate": 7.375130747397302e-05, "loss": 0.006, "step": 22520 }, { "grad_norm": 0.09118513762950897, "learning_rate": 7.372705375747377e-05, "loss": 0.0042, "step": 22530 }, { "grad_norm": 0.12610511481761932, "learning_rate": 7.370279283332205e-05, "loss": 0.0061, "step": 22540 }, { "grad_norm": 0.10881433635950089, "learning_rate": 7.36785247088877e-05, "loss": 0.0067, "step": 22550 }, { "grad_norm": 0.12405761331319809, "learning_rate": 7.365424939154275e-05, "loss": 0.0061, "step": 22560 }, { "grad_norm": 0.10695803910493851, "learning_rate": 7.362996688866138e-05, "loss": 0.006, "step": 22570 }, { "grad_norm": 0.1201576516032219, "learning_rate": 7.360567720761999e-05, "loss": 0.0058, "step": 22580 }, { "grad_norm": 0.12132127583026886, "learning_rate": 7.358138035579711e-05, "loss": 0.0048, "step": 22590 }, { "grad_norm": 0.0802726224064827, "learning_rate": 7.355707634057354e-05, "loss": 0.0039, "step": 22600 }, { "grad_norm": 0.11836083978414536, "learning_rate": 7.353276516933215e-05, "loss": 0.0054, "step": 22610 }, { "grad_norm": 0.07906433939933777, "learning_rate": 7.350844684945806e-05, "loss": 0.0049, "step": 22620 }, { "grad_norm": 0.09996818006038666, "learning_rate": 7.348412138833851e-05, "loss": 0.0055, "step": 22630 }, { "grad_norm": 0.11711765080690384, "learning_rate": 7.345978879336295e-05, "loss": 0.0042, "step": 22640 }, { "grad_norm": 0.08422345668077469, "learning_rate": 7.343544907192296e-05, "loss": 0.0049, "step": 22650 }, { "grad_norm": 0.1001334860920906, "learning_rate": 7.341110223141235e-05, "loss": 0.0038, "step": 22660 }, { "grad_norm": 0.10070529580116272, "learning_rate": 7.3386748279227e-05, "loss": 0.0053, "step": 22670 }, { "grad_norm": 0.09622703492641449, "learning_rate": 7.336238722276501e-05, "loss": 0.0042, "step": 22680 }, { "grad_norm": 0.11123869568109512, "learning_rate": 7.333801906942663e-05, "loss": 0.0058, "step": 22690 }, { "grad_norm": 0.11932650953531265, "learning_rate": 7.331364382661428e-05, "loss": 0.0048, "step": 22700 }, { "grad_norm": 0.11557947099208832, "learning_rate": 7.328926150173248e-05, "loss": 0.0053, "step": 22710 }, { "grad_norm": 0.10157248377799988, "learning_rate": 7.326487210218795e-05, "loss": 0.0044, "step": 22720 }, { "grad_norm": 0.13795332610607147, "learning_rate": 7.324047563538955e-05, "loss": 0.0039, "step": 22730 }, { "grad_norm": 0.09129158407449722, "learning_rate": 7.321607210874828e-05, "loss": 0.0045, "step": 22740 }, { "grad_norm": 0.08471207320690155, "learning_rate": 7.31916615296773e-05, "loss": 0.0036, "step": 22750 }, { "grad_norm": 0.0787125751376152, "learning_rate": 7.316724390559188e-05, "loss": 0.0038, "step": 22760 }, { "grad_norm": 0.09703853726387024, "learning_rate": 7.314281924390946e-05, "loss": 0.0044, "step": 22770 }, { "grad_norm": 0.11738291382789612, "learning_rate": 7.311838755204959e-05, "loss": 0.0047, "step": 22780 }, { "grad_norm": 0.08947824686765671, "learning_rate": 7.3093948837434e-05, "loss": 0.0039, "step": 22790 }, { "grad_norm": 0.0839603841304779, "learning_rate": 7.306950310748651e-05, "loss": 0.0039, "step": 22800 }, { "grad_norm": 0.09286948293447495, "learning_rate": 7.304505036963311e-05, "loss": 0.004, "step": 22810 }, { "grad_norm": 0.12327149510383606, "learning_rate": 7.302059063130186e-05, "loss": 0.0034, "step": 22820 }, { "grad_norm": 0.07860904932022095, "learning_rate": 7.2996123899923e-05, "loss": 0.0033, "step": 22830 }, { "grad_norm": 0.10442068427801132, "learning_rate": 7.297165018292886e-05, "loss": 0.0036, "step": 22840 }, { "grad_norm": 0.13752537965774536, "learning_rate": 7.294716948775396e-05, "loss": 0.0042, "step": 22850 }, { "grad_norm": 0.1075034961104393, "learning_rate": 7.292268182183484e-05, "loss": 0.0045, "step": 22860 }, { "grad_norm": 0.12658560276031494, "learning_rate": 7.28981871926102e-05, "loss": 0.0043, "step": 22870 }, { "grad_norm": 0.08434674888849258, "learning_rate": 7.28736856075209e-05, "loss": 0.0038, "step": 22880 }, { "grad_norm": 0.0972224548459053, "learning_rate": 7.284917707400985e-05, "loss": 0.0035, "step": 22890 }, { "grad_norm": 0.07164610177278519, "learning_rate": 7.282466159952212e-05, "loss": 0.0038, "step": 22900 }, { "grad_norm": 0.09045467525720596, "learning_rate": 7.280013919150483e-05, "loss": 0.0038, "step": 22910 }, { "grad_norm": 0.08653199672698975, "learning_rate": 7.277560985740728e-05, "loss": 0.0044, "step": 22920 }, { "grad_norm": 0.09796405583620071, "learning_rate": 7.275107360468079e-05, "loss": 0.0047, "step": 22930 }, { "grad_norm": 0.1017056256532669, "learning_rate": 7.272653044077885e-05, "loss": 0.0037, "step": 22940 }, { "grad_norm": 0.10477832704782486, "learning_rate": 7.270198037315703e-05, "loss": 0.0034, "step": 22950 }, { "grad_norm": 0.11753161996603012, "learning_rate": 7.267742340927297e-05, "loss": 0.0039, "step": 22960 }, { "grad_norm": 0.08436734974384308, "learning_rate": 7.265285955658645e-05, "loss": 0.0038, "step": 22970 }, { "grad_norm": 0.1275695264339447, "learning_rate": 7.26282888225593e-05, "loss": 0.004, "step": 22980 }, { "grad_norm": 0.09448487311601639, "learning_rate": 7.260371121465548e-05, "loss": 0.0041, "step": 22990 }, { "grad_norm": 0.10938823968172073, "learning_rate": 7.2579126740341e-05, "loss": 0.0038, "step": 23000 }, { "grad_norm": 0.1361725777387619, "learning_rate": 7.2554535407084e-05, "loss": 0.0039, "step": 23010 }, { "grad_norm": 0.1312008649110794, "learning_rate": 7.252993722235464e-05, "loss": 0.004, "step": 23020 }, { "grad_norm": 0.07067906856536865, "learning_rate": 7.250533219362523e-05, "loss": 0.0044, "step": 23030 }, { "grad_norm": 0.08650915324687958, "learning_rate": 7.248072032837012e-05, "loss": 0.0036, "step": 23040 }, { "grad_norm": 0.07595159113407135, "learning_rate": 7.245610163406575e-05, "loss": 0.0039, "step": 23050 }, { "grad_norm": 0.08129045367240906, "learning_rate": 7.243147611819061e-05, "loss": 0.0034, "step": 23060 }, { "grad_norm": 0.08281997591257095, "learning_rate": 7.240684378822531e-05, "loss": 0.0033, "step": 23070 }, { "grad_norm": 0.11922603845596313, "learning_rate": 7.238220465165248e-05, "loss": 0.0038, "step": 23080 }, { "grad_norm": 0.11724866181612015, "learning_rate": 7.235755871595684e-05, "loss": 0.004, "step": 23090 }, { "grad_norm": 0.11465322971343994, "learning_rate": 7.233290598862517e-05, "loss": 0.0042, "step": 23100 }, { "grad_norm": 0.14165787398815155, "learning_rate": 7.230824647714635e-05, "loss": 0.0041, "step": 23110 }, { "grad_norm": 0.10410644859075546, "learning_rate": 7.228358018901124e-05, "loss": 0.0044, "step": 23120 }, { "grad_norm": 0.09022392332553864, "learning_rate": 7.225890713171286e-05, "loss": 0.0042, "step": 23130 }, { "grad_norm": 0.08865658193826675, "learning_rate": 7.223422731274618e-05, "loss": 0.0036, "step": 23140 }, { "grad_norm": 0.09651846438646317, "learning_rate": 7.220954073960832e-05, "loss": 0.0056, "step": 23150 }, { "grad_norm": 0.12107912451028824, "learning_rate": 7.218484741979838e-05, "loss": 0.0054, "step": 23160 }, { "grad_norm": 0.1001145988702774, "learning_rate": 7.216014736081756e-05, "loss": 0.0037, "step": 23170 }, { "grad_norm": 0.09124014526605606, "learning_rate": 7.213544057016906e-05, "loss": 0.0041, "step": 23180 }, { "grad_norm": 0.1336945742368698, "learning_rate": 7.211072705535819e-05, "loss": 0.0037, "step": 23190 }, { "grad_norm": 0.11540259420871735, "learning_rate": 7.208600682389224e-05, "loss": 0.0033, "step": 23200 }, { "grad_norm": 0.09047982096672058, "learning_rate": 7.206127988328055e-05, "loss": 0.0029, "step": 23210 }, { "grad_norm": 0.08134705573320389, "learning_rate": 7.203654624103453e-05, "loss": 0.0036, "step": 23220 }, { "grad_norm": 0.12373678386211395, "learning_rate": 7.201180590466761e-05, "loss": 0.0032, "step": 23230 }, { "grad_norm": 0.12850841879844666, "learning_rate": 7.198705888169523e-05, "loss": 0.0037, "step": 23240 }, { "grad_norm": 0.08345013111829758, "learning_rate": 7.196230517963491e-05, "loss": 0.0034, "step": 23250 }, { "grad_norm": 0.11053719371557236, "learning_rate": 7.193754480600615e-05, "loss": 0.0037, "step": 23260 }, { "grad_norm": 0.09821537882089615, "learning_rate": 7.19127777683305e-05, "loss": 0.0034, "step": 23270 }, { "grad_norm": 0.10475219041109085, "learning_rate": 7.188800407413156e-05, "loss": 0.0045, "step": 23280 }, { "grad_norm": 0.08742792904376984, "learning_rate": 7.186322373093489e-05, "loss": 0.0034, "step": 23290 }, { "grad_norm": 0.0898851826786995, "learning_rate": 7.18384367462681e-05, "loss": 0.0037, "step": 23300 }, { "grad_norm": 0.10975230485200882, "learning_rate": 7.181364312766085e-05, "loss": 0.0038, "step": 23310 }, { "grad_norm": 0.08266641944646835, "learning_rate": 7.178884288264477e-05, "loss": 0.0035, "step": 23320 }, { "grad_norm": 0.11247510462999344, "learning_rate": 7.176403601875353e-05, "loss": 0.0031, "step": 23330 }, { "grad_norm": 0.1567458063364029, "learning_rate": 7.173922254352279e-05, "loss": 0.0045, "step": 23340 }, { "grad_norm": 0.13994504511356354, "learning_rate": 7.171440246449024e-05, "loss": 0.0037, "step": 23350 }, { "grad_norm": 0.133097842335701, "learning_rate": 7.168957578919555e-05, "loss": 0.0038, "step": 23360 }, { "grad_norm": 0.14215685427188873, "learning_rate": 7.16647425251804e-05, "loss": 0.0048, "step": 23370 }, { "grad_norm": 0.12526974081993103, "learning_rate": 7.163990267998852e-05, "loss": 0.0035, "step": 23380 }, { "grad_norm": 0.1405322402715683, "learning_rate": 7.161505626116556e-05, "loss": 0.0037, "step": 23390 }, { "grad_norm": 0.09648234397172928, "learning_rate": 7.159020327625923e-05, "loss": 0.0031, "step": 23400 }, { "grad_norm": 0.13112832605838776, "learning_rate": 7.15653437328192e-05, "loss": 0.0037, "step": 23410 }, { "grad_norm": 0.1019984781742096, "learning_rate": 7.154047763839713e-05, "loss": 0.0047, "step": 23420 }, { "grad_norm": 0.09339386969804764, "learning_rate": 7.15156050005467e-05, "loss": 0.0045, "step": 23430 }, { "grad_norm": 0.10912835597991943, "learning_rate": 7.149072582682357e-05, "loss": 0.0046, "step": 23440 }, { "grad_norm": 0.1105041354894638, "learning_rate": 7.146584012478535e-05, "loss": 0.0044, "step": 23450 }, { "grad_norm": 0.07785440236330032, "learning_rate": 7.144094790199169e-05, "loss": 0.0052, "step": 23460 }, { "grad_norm": 0.10932967066764832, "learning_rate": 7.141604916600415e-05, "loss": 0.0051, "step": 23470 }, { "grad_norm": 0.08828005939722061, "learning_rate": 7.139114392438635e-05, "loss": 0.0045, "step": 23480 }, { "grad_norm": 0.07504556328058243, "learning_rate": 7.136623218470382e-05, "loss": 0.0035, "step": 23490 }, { "grad_norm": 0.11938444525003433, "learning_rate": 7.13413139545241e-05, "loss": 0.0043, "step": 23500 }, { "grad_norm": 0.11904355883598328, "learning_rate": 7.131638924141668e-05, "loss": 0.0042, "step": 23510 }, { "grad_norm": 0.10132081061601639, "learning_rate": 7.129145805295304e-05, "loss": 0.0044, "step": 23520 }, { "grad_norm": 0.11097460240125656, "learning_rate": 7.126652039670661e-05, "loss": 0.0039, "step": 23530 }, { "grad_norm": 0.15167096257209778, "learning_rate": 7.124157628025278e-05, "loss": 0.0051, "step": 23540 }, { "grad_norm": 0.1283387541770935, "learning_rate": 7.121662571116894e-05, "loss": 0.0045, "step": 23550 }, { "grad_norm": 0.06580912321805954, "learning_rate": 7.119166869703441e-05, "loss": 0.0045, "step": 23560 }, { "grad_norm": 0.09116066992282867, "learning_rate": 7.116670524543044e-05, "loss": 0.0052, "step": 23570 }, { "grad_norm": 0.09855709969997406, "learning_rate": 7.114173536394032e-05, "loss": 0.0039, "step": 23580 }, { "grad_norm": 0.09550781548023224, "learning_rate": 7.111675906014917e-05, "loss": 0.0041, "step": 23590 }, { "grad_norm": 0.09938496351242065, "learning_rate": 7.109177634164421e-05, "loss": 0.0042, "step": 23600 }, { "grad_norm": 0.09437577426433563, "learning_rate": 7.106678721601449e-05, "loss": 0.0039, "step": 23610 }, { "grad_norm": 0.11372186988592148, "learning_rate": 7.104179169085103e-05, "loss": 0.005, "step": 23620 }, { "grad_norm": 0.11671777814626694, "learning_rate": 7.101678977374683e-05, "loss": 0.0045, "step": 23630 }, { "grad_norm": 0.11318673193454742, "learning_rate": 7.099178147229685e-05, "loss": 0.0036, "step": 23640 }, { "grad_norm": 0.12088803946971893, "learning_rate": 7.096676679409789e-05, "loss": 0.0042, "step": 23650 }, { "grad_norm": 0.0770561471581459, "learning_rate": 7.094174574674877e-05, "loss": 0.0049, "step": 23660 }, { "grad_norm": 0.10900821536779404, "learning_rate": 7.091671833785025e-05, "loss": 0.0045, "step": 23670 }, { "grad_norm": 0.10583657026290894, "learning_rate": 7.089168457500493e-05, "loss": 0.0035, "step": 23680 }, { "grad_norm": 0.08501455187797546, "learning_rate": 7.086664446581747e-05, "loss": 0.0034, "step": 23690 }, { "grad_norm": 0.09171874076128006, "learning_rate": 7.084159801789438e-05, "loss": 0.0037, "step": 23700 }, { "grad_norm": 0.08357805758714676, "learning_rate": 7.081654523884411e-05, "loss": 0.0032, "step": 23710 }, { "grad_norm": 0.09987840801477432, "learning_rate": 7.0791486136277e-05, "loss": 0.0045, "step": 23720 }, { "grad_norm": 0.08141104876995087, "learning_rate": 7.07664207178054e-05, "loss": 0.0038, "step": 23730 }, { "grad_norm": 0.09369786828756332, "learning_rate": 7.074134899104345e-05, "loss": 0.0039, "step": 23740 }, { "grad_norm": 0.10285733640193939, "learning_rate": 7.071627096360735e-05, "loss": 0.0048, "step": 23750 }, { "grad_norm": 0.10805657505989075, "learning_rate": 7.069118664311511e-05, "loss": 0.0055, "step": 23760 }, { "grad_norm": 0.11309747397899628, "learning_rate": 7.06660960371867e-05, "loss": 0.0043, "step": 23770 }, { "grad_norm": 0.1014593169093132, "learning_rate": 7.064099915344396e-05, "loss": 0.0046, "step": 23780 }, { "grad_norm": 0.08534266799688339, "learning_rate": 7.061589599951066e-05, "loss": 0.0038, "step": 23790 }, { "grad_norm": 0.10747137665748596, "learning_rate": 7.05907865830125e-05, "loss": 0.0037, "step": 23800 }, { "grad_norm": 0.11432048678398132, "learning_rate": 7.056567091157703e-05, "loss": 0.0042, "step": 23810 }, { "grad_norm": 0.10644227266311646, "learning_rate": 7.054054899283375e-05, "loss": 0.0034, "step": 23820 }, { "grad_norm": 0.09497716277837753, "learning_rate": 7.051542083441403e-05, "loss": 0.004, "step": 23830 }, { "grad_norm": 0.08996168524026871, "learning_rate": 7.049028644395113e-05, "loss": 0.0037, "step": 23840 }, { "grad_norm": 0.1590958833694458, "learning_rate": 7.046514582908024e-05, "loss": 0.0041, "step": 23850 }, { "grad_norm": 0.10651428997516632, "learning_rate": 7.043999899743838e-05, "loss": 0.0043, "step": 23860 }, { "grad_norm": 0.10153781622648239, "learning_rate": 7.041484595666451e-05, "loss": 0.0035, "step": 23870 }, { "grad_norm": 0.08889392763376236, "learning_rate": 7.038968671439948e-05, "loss": 0.0036, "step": 23880 }, { "grad_norm": 0.08052600920200348, "learning_rate": 7.036452127828596e-05, "loss": 0.0033, "step": 23890 }, { "grad_norm": 0.11777567118406296, "learning_rate": 7.033934965596859e-05, "loss": 0.0037, "step": 23900 }, { "grad_norm": 0.12035134434700012, "learning_rate": 7.031417185509381e-05, "loss": 0.0034, "step": 23910 }, { "grad_norm": 0.11320974677801132, "learning_rate": 7.028898788331e-05, "loss": 0.0032, "step": 23920 }, { "grad_norm": 0.13572195172309875, "learning_rate": 7.026379774826736e-05, "loss": 0.0037, "step": 23930 }, { "grad_norm": 0.1145256832242012, "learning_rate": 7.0238601457618e-05, "loss": 0.0034, "step": 23940 }, { "grad_norm": 0.12098664045333862, "learning_rate": 7.02133990190159e-05, "loss": 0.0043, "step": 23950 }, { "grad_norm": 0.09799734503030777, "learning_rate": 7.018819044011687e-05, "loss": 0.0039, "step": 23960 }, { "grad_norm": 0.07843978703022003, "learning_rate": 7.016297572857863e-05, "loss": 0.0045, "step": 23970 }, { "grad_norm": 0.09576071053743362, "learning_rate": 7.013775489206072e-05, "loss": 0.0034, "step": 23980 }, { "grad_norm": 0.07303248345851898, "learning_rate": 7.01125279382246e-05, "loss": 0.0037, "step": 23990 }, { "grad_norm": 0.09193909913301468, "learning_rate": 7.008729487473351e-05, "loss": 0.0037, "step": 24000 }, { "grad_norm": 0.0871644988656044, "learning_rate": 7.006205570925263e-05, "loss": 0.0042, "step": 24010 }, { "grad_norm": 0.07623252272605896, "learning_rate": 7.003681044944892e-05, "loss": 0.0033, "step": 24020 }, { "grad_norm": 0.11333737522363663, "learning_rate": 7.001155910299126e-05, "loss": 0.0032, "step": 24030 }, { "grad_norm": 0.13782572746276855, "learning_rate": 6.99863016775503e-05, "loss": 0.0032, "step": 24040 }, { "grad_norm": 0.10731124877929688, "learning_rate": 6.996103818079859e-05, "loss": 0.0034, "step": 24050 }, { "grad_norm": 0.11588946729898453, "learning_rate": 6.993576862041054e-05, "loss": 0.0033, "step": 24060 }, { "grad_norm": 0.0795399472117424, "learning_rate": 6.991049300406235e-05, "loss": 0.0031, "step": 24070 }, { "grad_norm": 0.09492699056863785, "learning_rate": 6.988521133943209e-05, "loss": 0.0035, "step": 24080 }, { "grad_norm": 0.09317857027053833, "learning_rate": 6.985992363419966e-05, "loss": 0.0032, "step": 24090 }, { "grad_norm": 0.10934636741876602, "learning_rate": 6.983462989604682e-05, "loss": 0.004, "step": 24100 }, { "grad_norm": 0.0863257497549057, "learning_rate": 6.980933013265709e-05, "loss": 0.0042, "step": 24110 }, { "grad_norm": 0.0906074196100235, "learning_rate": 6.978402435171592e-05, "loss": 0.0032, "step": 24120 }, { "grad_norm": 0.09510710090398788, "learning_rate": 6.975871256091052e-05, "loss": 0.0035, "step": 24130 }, { "grad_norm": 0.08344444632530212, "learning_rate": 6.973339476792995e-05, "loss": 0.003, "step": 24140 }, { "grad_norm": 0.12388379126787186, "learning_rate": 6.970807098046505e-05, "loss": 0.0033, "step": 24150 }, { "grad_norm": 0.08184759318828583, "learning_rate": 6.968274120620858e-05, "loss": 0.0056, "step": 24160 }, { "grad_norm": 0.14732420444488525, "learning_rate": 6.965740545285499e-05, "loss": 0.0045, "step": 24170 }, { "grad_norm": 0.12024132907390594, "learning_rate": 6.963206372810068e-05, "loss": 0.0051, "step": 24180 }, { "grad_norm": 0.1269758641719818, "learning_rate": 6.960671603964375e-05, "loss": 0.0037, "step": 24190 }, { "grad_norm": 0.10335615277290344, "learning_rate": 6.958136239518418e-05, "loss": 0.0055, "step": 24200 }, { "grad_norm": 0.09815077483654022, "learning_rate": 6.955600280242371e-05, "loss": 0.0043, "step": 24210 }, { "grad_norm": 0.09454613924026489, "learning_rate": 6.953063726906596e-05, "loss": 0.0039, "step": 24220 }, { "grad_norm": 0.1255740076303482, "learning_rate": 6.950526580281626e-05, "loss": 0.0034, "step": 24230 }, { "grad_norm": 0.09403326362371445, "learning_rate": 6.947988841138184e-05, "loss": 0.0031, "step": 24240 }, { "grad_norm": 0.08992529660463333, "learning_rate": 6.945450510247165e-05, "loss": 0.0034, "step": 24250 }, { "grad_norm": 0.09870894253253937, "learning_rate": 6.942911588379647e-05, "loss": 0.0034, "step": 24260 }, { "grad_norm": 0.0952792763710022, "learning_rate": 6.940372076306888e-05, "loss": 0.0035, "step": 24270 }, { "grad_norm": 0.08247236162424088, "learning_rate": 6.937831974800326e-05, "loss": 0.0035, "step": 24280 }, { "grad_norm": 0.10931402444839478, "learning_rate": 6.935291284631574e-05, "loss": 0.0036, "step": 24290 }, { "grad_norm": 0.14396023750305176, "learning_rate": 6.932750006572428e-05, "loss": 0.0045, "step": 24300 }, { "grad_norm": 0.08799847960472107, "learning_rate": 6.930208141394863e-05, "loss": 0.0049, "step": 24310 }, { "grad_norm": 0.15139080584049225, "learning_rate": 6.927665689871026e-05, "loss": 0.0038, "step": 24320 }, { "grad_norm": 0.09756830334663391, "learning_rate": 6.925122652773253e-05, "loss": 0.0037, "step": 24330 }, { "grad_norm": 0.11472410708665848, "learning_rate": 6.922579030874046e-05, "loss": 0.0044, "step": 24340 }, { "grad_norm": 0.09006772190332413, "learning_rate": 6.920034824946093e-05, "loss": 0.0043, "step": 24350 }, { "grad_norm": 0.05735763907432556, "learning_rate": 6.917490035762255e-05, "loss": 0.0044, "step": 24360 }, { "grad_norm": 0.09189494699239731, "learning_rate": 6.914944664095573e-05, "loss": 0.0033, "step": 24370 }, { "grad_norm": 0.09287689626216888, "learning_rate": 6.912398710719264e-05, "loss": 0.0028, "step": 24380 }, { "grad_norm": 0.11311379820108414, "learning_rate": 6.90985217640672e-05, "loss": 0.004, "step": 24390 }, { "grad_norm": 0.1297132670879364, "learning_rate": 6.90730506193151e-05, "loss": 0.0043, "step": 24400 }, { "grad_norm": 0.10573319345712662, "learning_rate": 6.904757368067384e-05, "loss": 0.0045, "step": 24410 }, { "grad_norm": 0.08173314481973648, "learning_rate": 6.90220909558826e-05, "loss": 0.0035, "step": 24420 }, { "grad_norm": 0.11556992679834366, "learning_rate": 6.899660245268237e-05, "loss": 0.0039, "step": 24430 }, { "grad_norm": 0.0890306830406189, "learning_rate": 6.897110817881592e-05, "loss": 0.0042, "step": 24440 }, { "grad_norm": 0.13493123650550842, "learning_rate": 6.894560814202769e-05, "loss": 0.005, "step": 24450 }, { "grad_norm": 0.10525880008935928, "learning_rate": 6.892010235006394e-05, "loss": 0.003, "step": 24460 }, { "grad_norm": 0.08921964466571808, "learning_rate": 6.889459081067264e-05, "loss": 0.0031, "step": 24470 }, { "grad_norm": 0.12770351767539978, "learning_rate": 6.886907353160356e-05, "loss": 0.0036, "step": 24480 }, { "grad_norm": 0.10317334532737732, "learning_rate": 6.884355052060814e-05, "loss": 0.003, "step": 24490 }, { "grad_norm": 0.08861854672431946, "learning_rate": 6.88180217854396e-05, "loss": 0.0038, "step": 24500 }, { "grad_norm": 0.11338762938976288, "learning_rate": 6.87924873338529e-05, "loss": 0.0039, "step": 24510 }, { "grad_norm": 0.07883963733911514, "learning_rate": 6.876694717360475e-05, "loss": 0.0032, "step": 24520 }, { "grad_norm": 0.05144086107611656, "learning_rate": 6.874140131245355e-05, "loss": 0.0032, "step": 24530 }, { "grad_norm": 0.0765475332736969, "learning_rate": 6.871584975815948e-05, "loss": 0.0031, "step": 24540 }, { "grad_norm": 0.09597765654325485, "learning_rate": 6.86902925184844e-05, "loss": 0.0037, "step": 24550 }, { "grad_norm": 0.09993255138397217, "learning_rate": 6.866472960119195e-05, "loss": 0.0043, "step": 24560 }, { "grad_norm": 0.07546520233154297, "learning_rate": 6.863916101404748e-05, "loss": 0.0034, "step": 24570 }, { "grad_norm": 0.10789632052183151, "learning_rate": 6.8613586764818e-05, "loss": 0.0048, "step": 24580 }, { "grad_norm": 0.09163948893547058, "learning_rate": 6.858800686127233e-05, "loss": 0.0032, "step": 24590 }, { "grad_norm": 0.07916485518217087, "learning_rate": 6.856242131118097e-05, "loss": 0.0036, "step": 24600 }, { "grad_norm": 0.14687074720859528, "learning_rate": 6.853683012231614e-05, "loss": 0.0032, "step": 24610 }, { "grad_norm": 0.08913138508796692, "learning_rate": 6.851123330245173e-05, "loss": 0.004, "step": 24620 }, { "grad_norm": 0.08561370521783829, "learning_rate": 6.848563085936343e-05, "loss": 0.0038, "step": 24630 }, { "grad_norm": 0.11358146369457245, "learning_rate": 6.846002280082853e-05, "loss": 0.0037, "step": 24640 }, { "grad_norm": 0.09833899885416031, "learning_rate": 6.843440913462614e-05, "loss": 0.0033, "step": 24650 }, { "grad_norm": 0.0946739912033081, "learning_rate": 6.840878986853698e-05, "loss": 0.004, "step": 24660 }, { "grad_norm": 0.08391616493463516, "learning_rate": 6.838316501034352e-05, "loss": 0.0029, "step": 24670 }, { "grad_norm": 0.08226411044597626, "learning_rate": 6.83575345678299e-05, "loss": 0.0028, "step": 24680 }, { "grad_norm": 0.09755837172269821, "learning_rate": 6.833189854878196e-05, "loss": 0.0032, "step": 24690 }, { "grad_norm": 0.08582141995429993, "learning_rate": 6.83062569609873e-05, "loss": 0.003, "step": 24700 }, { "grad_norm": 0.08056675642728806, "learning_rate": 6.828060981223512e-05, "loss": 0.0042, "step": 24710 }, { "grad_norm": 0.0982280895113945, "learning_rate": 6.825495711031634e-05, "loss": 0.0034, "step": 24720 }, { "grad_norm": 0.08978161215782166, "learning_rate": 6.822929886302359e-05, "loss": 0.0032, "step": 24730 }, { "grad_norm": 0.10053493827581406, "learning_rate": 6.820363507815116e-05, "loss": 0.0036, "step": 24740 }, { "grad_norm": 0.09624748677015305, "learning_rate": 6.817796576349501e-05, "loss": 0.0036, "step": 24750 }, { "grad_norm": 0.10873208194971085, "learning_rate": 6.815229092685285e-05, "loss": 0.0034, "step": 24760 }, { "grad_norm": 0.09527893364429474, "learning_rate": 6.812661057602399e-05, "loss": 0.0044, "step": 24770 }, { "grad_norm": 0.09798077493906021, "learning_rate": 6.810092471880943e-05, "loss": 0.0042, "step": 24780 }, { "grad_norm": 0.09976854920387268, "learning_rate": 6.807523336301187e-05, "loss": 0.0044, "step": 24790 }, { "grad_norm": 0.11300519108772278, "learning_rate": 6.804953651643566e-05, "loss": 0.0039, "step": 24800 }, { "grad_norm": 0.10190540552139282, "learning_rate": 6.802383418688685e-05, "loss": 0.0037, "step": 24810 }, { "grad_norm": 0.13694460690021515, "learning_rate": 6.799812638217309e-05, "loss": 0.0045, "step": 24820 }, { "grad_norm": 0.07107050716876984, "learning_rate": 6.797241311010373e-05, "loss": 0.0037, "step": 24830 }, { "grad_norm": 0.08525345474481583, "learning_rate": 6.794669437848982e-05, "loss": 0.0034, "step": 24840 }, { "grad_norm": 0.10837383568286896, "learning_rate": 6.792097019514402e-05, "loss": 0.0035, "step": 24850 }, { "grad_norm": 0.10877406597137451, "learning_rate": 6.789524056788064e-05, "loss": 0.0051, "step": 24860 }, { "grad_norm": 0.08740849047899246, "learning_rate": 6.786950550451567e-05, "loss": 0.0036, "step": 24870 }, { "grad_norm": 0.0956282690167427, "learning_rate": 6.784376501286676e-05, "loss": 0.0044, "step": 24880 }, { "grad_norm": 0.09242737293243408, "learning_rate": 6.781801910075316e-05, "loss": 0.0041, "step": 24890 }, { "grad_norm": 0.06903569400310516, "learning_rate": 6.779226777599581e-05, "loss": 0.0037, "step": 24900 }, { "grad_norm": 0.1535203754901886, "learning_rate": 6.776651104641729e-05, "loss": 0.0056, "step": 24910 }, { "grad_norm": 0.17223557829856873, "learning_rate": 6.774074891984183e-05, "loss": 0.005, "step": 24920 }, { "grad_norm": 0.15252462029457092, "learning_rate": 6.771498140409526e-05, "loss": 0.0046, "step": 24930 }, { "grad_norm": 0.11552929878234863, "learning_rate": 6.768920850700506e-05, "loss": 0.0051, "step": 24940 }, { "grad_norm": 0.09238611161708832, "learning_rate": 6.766343023640039e-05, "loss": 0.0033, "step": 24950 }, { "grad_norm": 0.09259651601314545, "learning_rate": 6.763764660011198e-05, "loss": 0.0051, "step": 24960 }, { "grad_norm": 0.13512074947357178, "learning_rate": 6.761185760597223e-05, "loss": 0.0044, "step": 24970 }, { "grad_norm": 0.0943504199385643, "learning_rate": 6.758606326181515e-05, "loss": 0.0035, "step": 24980 }, { "grad_norm": 0.08887108415365219, "learning_rate": 6.75602635754764e-05, "loss": 0.0033, "step": 24990 }, { "grad_norm": 0.10151594877243042, "learning_rate": 6.75344585547932e-05, "loss": 0.0055, "step": 25000 }, { "grad_norm": 0.08752438426017761, "learning_rate": 6.750864820760449e-05, "loss": 0.0033, "step": 25010 }, { "grad_norm": 0.09322156757116318, "learning_rate": 6.748283254175072e-05, "loss": 0.0035, "step": 25020 }, { "grad_norm": 0.10731224715709686, "learning_rate": 6.745701156507404e-05, "loss": 0.0041, "step": 25030 }, { "grad_norm": 0.09364540129899979, "learning_rate": 6.743118528541818e-05, "loss": 0.0041, "step": 25040 }, { "grad_norm": 0.08949775993824005, "learning_rate": 6.740535371062846e-05, "loss": 0.0035, "step": 25050 }, { "grad_norm": 0.06964123249053955, "learning_rate": 6.737951684855185e-05, "loss": 0.0039, "step": 25060 }, { "grad_norm": 0.09350074082612991, "learning_rate": 6.735367470703691e-05, "loss": 0.0037, "step": 25070 }, { "grad_norm": 0.10331191122531891, "learning_rate": 6.732782729393379e-05, "loss": 0.0052, "step": 25080 }, { "grad_norm": 0.11096806824207306, "learning_rate": 6.730197461709425e-05, "loss": 0.0035, "step": 25090 }, { "grad_norm": 0.10170231759548187, "learning_rate": 6.727611668437164e-05, "loss": 0.0036, "step": 25100 }, { "grad_norm": 0.1151621863245964, "learning_rate": 6.725025350362094e-05, "loss": 0.0043, "step": 25110 }, { "grad_norm": 0.08629091829061508, "learning_rate": 6.72243850826987e-05, "loss": 0.0035, "step": 25120 }, { "grad_norm": 0.09738361835479736, "learning_rate": 6.719851142946305e-05, "loss": 0.0041, "step": 25130 }, { "grad_norm": 0.09864665567874908, "learning_rate": 6.717263255177372e-05, "loss": 0.0038, "step": 25140 }, { "grad_norm": 0.07949113100767136, "learning_rate": 6.714674845749205e-05, "loss": 0.0042, "step": 25150 }, { "grad_norm": 0.07536553591489792, "learning_rate": 6.712085915448092e-05, "loss": 0.0038, "step": 25160 }, { "grad_norm": 0.07909311354160309, "learning_rate": 6.709496465060486e-05, "loss": 0.0032, "step": 25170 }, { "grad_norm": 0.11236262321472168, "learning_rate": 6.706906495372987e-05, "loss": 0.0033, "step": 25180 }, { "grad_norm": 0.10356815159320831, "learning_rate": 6.704316007172365e-05, "loss": 0.0032, "step": 25190 }, { "grad_norm": 0.08554394543170929, "learning_rate": 6.701725001245539e-05, "loss": 0.0031, "step": 25200 }, { "grad_norm": 0.07480963319540024, "learning_rate": 6.699133478379588e-05, "loss": 0.0044, "step": 25210 }, { "grad_norm": 0.12559111416339874, "learning_rate": 6.69654143936175e-05, "loss": 0.0035, "step": 25220 }, { "grad_norm": 0.09897017478942871, "learning_rate": 6.693948884979419e-05, "loss": 0.0028, "step": 25230 }, { "grad_norm": 0.09743663668632507, "learning_rate": 6.691355816020142e-05, "loss": 0.0035, "step": 25240 }, { "grad_norm": 0.1060958206653595, "learning_rate": 6.688762233271624e-05, "loss": 0.0039, "step": 25250 }, { "grad_norm": 0.07719089835882187, "learning_rate": 6.68616813752173e-05, "loss": 0.0033, "step": 25260 }, { "grad_norm": 0.11226644366979599, "learning_rate": 6.683573529558477e-05, "loss": 0.0043, "step": 25270 }, { "grad_norm": 0.09137935936450958, "learning_rate": 6.680978410170037e-05, "loss": 0.003, "step": 25280 }, { "grad_norm": 0.08853121101856232, "learning_rate": 6.678382780144741e-05, "loss": 0.0036, "step": 25290 }, { "grad_norm": 0.07925929129123688, "learning_rate": 6.675786640271071e-05, "loss": 0.0029, "step": 25300 }, { "grad_norm": 0.08431787043809891, "learning_rate": 6.673189991337665e-05, "loss": 0.0036, "step": 25310 }, { "grad_norm": 0.08636844903230667, "learning_rate": 6.670592834133317e-05, "loss": 0.0035, "step": 25320 }, { "grad_norm": 0.14677344262599945, "learning_rate": 6.667995169446979e-05, "loss": 0.004, "step": 25330 }, { "grad_norm": 0.14075684547424316, "learning_rate": 6.665396998067747e-05, "loss": 0.004, "step": 25340 }, { "grad_norm": 0.10410565137863159, "learning_rate": 6.66279832078488e-05, "loss": 0.0048, "step": 25350 }, { "grad_norm": 0.11211065948009491, "learning_rate": 6.660199138387786e-05, "loss": 0.0039, "step": 25360 }, { "grad_norm": 0.10197128355503082, "learning_rate": 6.65759945166603e-05, "loss": 0.0036, "step": 25370 }, { "grad_norm": 0.0756995752453804, "learning_rate": 6.654999261409326e-05, "loss": 0.0034, "step": 25380 }, { "grad_norm": 0.2292712926864624, "learning_rate": 6.652398568407544e-05, "loss": 0.0038, "step": 25390 }, { "grad_norm": 0.12563657760620117, "learning_rate": 6.649797373450707e-05, "loss": 0.0036, "step": 25400 }, { "grad_norm": 0.13913948833942413, "learning_rate": 6.647195677328988e-05, "loss": 0.0037, "step": 25410 }, { "grad_norm": 0.12654168903827667, "learning_rate": 6.644593480832712e-05, "loss": 0.0045, "step": 25420 }, { "grad_norm": 0.08535333722829819, "learning_rate": 6.641990784752363e-05, "loss": 0.0042, "step": 25430 }, { "grad_norm": 0.0831611156463623, "learning_rate": 6.639387589878566e-05, "loss": 0.0031, "step": 25440 }, { "grad_norm": 0.10239295661449432, "learning_rate": 6.636783897002103e-05, "loss": 0.0046, "step": 25450 }, { "grad_norm": 0.12421279400587082, "learning_rate": 6.63417970691391e-05, "loss": 0.0034, "step": 25460 }, { "grad_norm": 0.08992721885442734, "learning_rate": 6.63157502040507e-05, "loss": 0.0044, "step": 25470 }, { "grad_norm": 0.09407735615968704, "learning_rate": 6.628969838266819e-05, "loss": 0.005, "step": 25480 }, { "grad_norm": 0.09569279104471207, "learning_rate": 6.626364161290541e-05, "loss": 0.0046, "step": 25490 }, { "grad_norm": 0.0791441947221756, "learning_rate": 6.623757990267774e-05, "loss": 0.0048, "step": 25500 }, { "grad_norm": 0.08128484338521957, "learning_rate": 6.621151325990201e-05, "loss": 0.0034, "step": 25510 }, { "grad_norm": 0.09317872673273087, "learning_rate": 6.618544169249657e-05, "loss": 0.0036, "step": 25520 }, { "grad_norm": 0.09328952431678772, "learning_rate": 6.615936520838133e-05, "loss": 0.005, "step": 25530 }, { "grad_norm": 0.09565892070531845, "learning_rate": 6.613328381547759e-05, "loss": 0.0038, "step": 25540 }, { "grad_norm": 0.090150848031044, "learning_rate": 6.610719752170821e-05, "loss": 0.0041, "step": 25550 }, { "grad_norm": 0.1170755997300148, "learning_rate": 6.60811063349975e-05, "loss": 0.0043, "step": 25560 }, { "grad_norm": 0.10462500900030136, "learning_rate": 6.605501026327127e-05, "loss": 0.0046, "step": 25570 }, { "grad_norm": 0.11868873238563538, "learning_rate": 6.602890931445685e-05, "loss": 0.0036, "step": 25580 }, { "grad_norm": 0.1017790213227272, "learning_rate": 6.6002803496483e-05, "loss": 0.0054, "step": 25590 }, { "grad_norm": 0.11260076612234116, "learning_rate": 6.597669281727997e-05, "loss": 0.0037, "step": 25600 }, { "grad_norm": 0.10783389210700989, "learning_rate": 6.595057728477949e-05, "loss": 0.0033, "step": 25610 }, { "grad_norm": 0.08832535892724991, "learning_rate": 6.59244569069148e-05, "loss": 0.0033, "step": 25620 }, { "grad_norm": 0.10355649143457413, "learning_rate": 6.589833169162054e-05, "loss": 0.0039, "step": 25630 }, { "grad_norm": 0.08543998748064041, "learning_rate": 6.587220164683291e-05, "loss": 0.0033, "step": 25640 }, { "grad_norm": 0.08244156092405319, "learning_rate": 6.58460667804895e-05, "loss": 0.0029, "step": 25650 }, { "grad_norm": 0.07632128894329071, "learning_rate": 6.581992710052938e-05, "loss": 0.0026, "step": 25660 }, { "grad_norm": 0.07372214645147324, "learning_rate": 6.579378261489311e-05, "loss": 0.0033, "step": 25670 }, { "grad_norm": 0.12224562466144562, "learning_rate": 6.576763333152268e-05, "loss": 0.0036, "step": 25680 }, { "grad_norm": 0.0999365895986557, "learning_rate": 6.574147925836159e-05, "loss": 0.0038, "step": 25690 }, { "grad_norm": 0.08849378675222397, "learning_rate": 6.571532040335472e-05, "loss": 0.0038, "step": 25700 }, { "grad_norm": 0.14237917959690094, "learning_rate": 6.568915677444845e-05, "loss": 0.0046, "step": 25710 }, { "grad_norm": 0.12570875883102417, "learning_rate": 6.56629883795906e-05, "loss": 0.0038, "step": 25720 }, { "grad_norm": 0.08501864969730377, "learning_rate": 6.563681522673043e-05, "loss": 0.003, "step": 25730 }, { "grad_norm": 0.09742697328329086, "learning_rate": 6.561063732381867e-05, "loss": 0.0038, "step": 25740 }, { "grad_norm": 0.1154140755534172, "learning_rate": 6.558445467880745e-05, "loss": 0.0034, "step": 25750 }, { "grad_norm": 0.09335342049598694, "learning_rate": 6.55582672996504e-05, "loss": 0.0035, "step": 25760 }, { "grad_norm": 0.09416765719652176, "learning_rate": 6.553207519430253e-05, "loss": 0.0036, "step": 25770 }, { "grad_norm": 0.08742601424455643, "learning_rate": 6.550587837072032e-05, "loss": 0.0029, "step": 25780 }, { "grad_norm": 0.11340853571891785, "learning_rate": 6.547967683686166e-05, "loss": 0.0056, "step": 25790 }, { "grad_norm": 0.08540675044059753, "learning_rate": 6.545347060068591e-05, "loss": 0.0041, "step": 25800 }, { "grad_norm": 0.10695073753595352, "learning_rate": 6.542725967015382e-05, "loss": 0.0032, "step": 25810 }, { "grad_norm": 0.07267740368843079, "learning_rate": 6.540104405322757e-05, "loss": 0.0032, "step": 25820 }, { "grad_norm": 0.0865553766489029, "learning_rate": 6.537482375787077e-05, "loss": 0.0032, "step": 25830 }, { "grad_norm": 0.06598778069019318, "learning_rate": 6.534859879204845e-05, "loss": 0.0039, "step": 25840 }, { "grad_norm": 0.08849557489156723, "learning_rate": 6.532236916372709e-05, "loss": 0.0032, "step": 25850 }, { "grad_norm": 0.08146296441555023, "learning_rate": 6.529613488087454e-05, "loss": 0.0034, "step": 25860 }, { "grad_norm": 0.08641819655895233, "learning_rate": 6.526989595146009e-05, "loss": 0.0031, "step": 25870 }, { "grad_norm": 0.08839485049247742, "learning_rate": 6.524365238345441e-05, "loss": 0.0035, "step": 25880 }, { "grad_norm": 0.09233582764863968, "learning_rate": 6.521740418482964e-05, "loss": 0.0037, "step": 25890 }, { "grad_norm": 0.11229106038808823, "learning_rate": 6.519115136355925e-05, "loss": 0.0035, "step": 25900 }, { "grad_norm": 0.09208161383867264, "learning_rate": 6.51648939276182e-05, "loss": 0.0028, "step": 25910 }, { "grad_norm": 0.09764065593481064, "learning_rate": 6.513863188498277e-05, "loss": 0.0034, "step": 25920 }, { "grad_norm": 0.08819198608398438, "learning_rate": 6.511236524363068e-05, "loss": 0.0032, "step": 25930 }, { "grad_norm": 0.08193749934434891, "learning_rate": 6.508609401154104e-05, "loss": 0.0038, "step": 25940 }, { "grad_norm": 0.09382788836956024, "learning_rate": 6.505981819669439e-05, "loss": 0.0028, "step": 25950 }, { "grad_norm": 0.10078175365924835, "learning_rate": 6.503353780707258e-05, "loss": 0.0037, "step": 25960 }, { "grad_norm": 0.07588844001293182, "learning_rate": 6.500725285065895e-05, "loss": 0.0031, "step": 25970 }, { "grad_norm": 0.07668987661600113, "learning_rate": 6.498096333543813e-05, "loss": 0.0043, "step": 25980 }, { "grad_norm": 0.0696195736527443, "learning_rate": 6.49546692693962e-05, "loss": 0.0042, "step": 25990 }, { "grad_norm": 0.1061038225889206, "learning_rate": 6.492837066052059e-05, "loss": 0.0033, "step": 26000 }, { "grad_norm": 0.12182526290416718, "learning_rate": 6.490206751680014e-05, "loss": 0.0038, "step": 26010 }, { "grad_norm": 0.11306622624397278, "learning_rate": 6.487575984622505e-05, "loss": 0.0033, "step": 26020 }, { "grad_norm": 0.07898914813995361, "learning_rate": 6.484944765678689e-05, "loss": 0.0032, "step": 26030 }, { "grad_norm": 0.06345740705728531, "learning_rate": 6.482313095647861e-05, "loss": 0.0036, "step": 26040 }, { "grad_norm": 0.05783192813396454, "learning_rate": 6.479680975329451e-05, "loss": 0.0028, "step": 26050 }, { "grad_norm": 0.10517408698797226, "learning_rate": 6.477048405523031e-05, "loss": 0.0038, "step": 26060 }, { "grad_norm": 0.0921502485871315, "learning_rate": 6.474415387028304e-05, "loss": 0.0027, "step": 26070 }, { "grad_norm": 0.0974380224943161, "learning_rate": 6.471781920645114e-05, "loss": 0.0035, "step": 26080 }, { "grad_norm": 0.06339175254106522, "learning_rate": 6.469148007173434e-05, "loss": 0.0027, "step": 26090 }, { "grad_norm": 0.07327601313591003, "learning_rate": 6.466513647413381e-05, "loss": 0.0029, "step": 26100 }, { "grad_norm": 0.09115177392959595, "learning_rate": 6.463878842165203e-05, "loss": 0.0032, "step": 26110 }, { "grad_norm": 0.08704008162021637, "learning_rate": 6.461243592229286e-05, "loss": 0.0035, "step": 26120 }, { "grad_norm": 0.09355030953884125, "learning_rate": 6.458607898406146e-05, "loss": 0.0029, "step": 26130 }, { "grad_norm": 0.07996867597103119, "learning_rate": 6.455971761496439e-05, "loss": 0.0028, "step": 26140 }, { "grad_norm": 0.08812375366687775, "learning_rate": 6.453335182300953e-05, "loss": 0.0032, "step": 26150 }, { "grad_norm": 0.10610785335302353, "learning_rate": 6.450698161620612e-05, "loss": 0.0032, "step": 26160 }, { "grad_norm": 0.06911955028772354, "learning_rate": 6.448060700256473e-05, "loss": 0.0032, "step": 26170 }, { "grad_norm": 0.061920128762722015, "learning_rate": 6.445422799009726e-05, "loss": 0.0026, "step": 26180 }, { "grad_norm": 0.08061034977436066, "learning_rate": 6.442784458681699e-05, "loss": 0.0032, "step": 26190 }, { "grad_norm": 0.1290789544582367, "learning_rate": 6.440145680073847e-05, "loss": 0.0038, "step": 26200 }, { "grad_norm": 0.08053785562515259, "learning_rate": 6.437506463987762e-05, "loss": 0.0028, "step": 26210 }, { "grad_norm": 0.07217683643102646, "learning_rate": 6.434866811225168e-05, "loss": 0.0036, "step": 26220 }, { "grad_norm": 0.07742109894752502, "learning_rate": 6.432226722587923e-05, "loss": 0.0035, "step": 26230 }, { "grad_norm": 0.12617501616477966, "learning_rate": 6.429586198878015e-05, "loss": 0.0034, "step": 26240 }, { "grad_norm": 0.0864112451672554, "learning_rate": 6.426945240897566e-05, "loss": 0.0035, "step": 26250 }, { "grad_norm": 0.10673404484987259, "learning_rate": 6.424303849448829e-05, "loss": 0.0053, "step": 26260 }, { "grad_norm": 0.09511922299861908, "learning_rate": 6.42166202533419e-05, "loss": 0.0033, "step": 26270 }, { "grad_norm": 0.12108545750379562, "learning_rate": 6.419019769356164e-05, "loss": 0.0038, "step": 26280 }, { "grad_norm": 0.1054314374923706, "learning_rate": 6.416377082317398e-05, "loss": 0.0031, "step": 26290 }, { "grad_norm": 0.11346045136451721, "learning_rate": 6.413733965020674e-05, "loss": 0.0027, "step": 26300 }, { "grad_norm": 0.09616339951753616, "learning_rate": 6.411090418268896e-05, "loss": 0.0053, "step": 26310 }, { "grad_norm": 0.10447492450475693, "learning_rate": 6.408446442865109e-05, "loss": 0.0034, "step": 26320 }, { "grad_norm": 0.10724011808633804, "learning_rate": 6.405802039612479e-05, "loss": 0.0036, "step": 26330 }, { "grad_norm": 0.12309885770082474, "learning_rate": 6.403157209314308e-05, "loss": 0.0059, "step": 26340 }, { "grad_norm": 0.11345037817955017, "learning_rate": 6.400511952774024e-05, "loss": 0.0036, "step": 26350 }, { "grad_norm": 0.08967434614896774, "learning_rate": 6.397866270795187e-05, "loss": 0.0027, "step": 26360 }, { "grad_norm": 0.08533687889575958, "learning_rate": 6.395220164181489e-05, "loss": 0.0032, "step": 26370 }, { "grad_norm": 0.1105530858039856, "learning_rate": 6.39257363373674e-05, "loss": 0.0036, "step": 26380 }, { "grad_norm": 0.11083925515413284, "learning_rate": 6.389926680264892e-05, "loss": 0.0048, "step": 26390 }, { "grad_norm": 0.12687990069389343, "learning_rate": 6.387279304570017e-05, "loss": 0.0044, "step": 26400 }, { "grad_norm": 0.11249039322137833, "learning_rate": 6.384631507456319e-05, "loss": 0.0039, "step": 26410 }, { "grad_norm": 0.10810809582471848, "learning_rate": 6.381983289728126e-05, "loss": 0.0038, "step": 26420 }, { "grad_norm": 0.09211204200983047, "learning_rate": 6.3793346521899e-05, "loss": 0.0039, "step": 26430 }, { "grad_norm": 0.0823834091424942, "learning_rate": 6.376685595646226e-05, "loss": 0.0038, "step": 26440 }, { "grad_norm": 0.08680987358093262, "learning_rate": 6.374036120901816e-05, "loss": 0.0029, "step": 26450 }, { "grad_norm": 0.07548125088214874, "learning_rate": 6.371386228761514e-05, "loss": 0.0029, "step": 26460 }, { "grad_norm": 0.0680433139204979, "learning_rate": 6.368735920030283e-05, "loss": 0.0035, "step": 26470 }, { "grad_norm": 0.06784749031066895, "learning_rate": 6.366085195513218e-05, "loss": 0.0029, "step": 26480 }, { "grad_norm": 0.09679821878671646, "learning_rate": 6.363434056015543e-05, "loss": 0.0036, "step": 26490 }, { "grad_norm": 0.0872567668557167, "learning_rate": 6.360782502342599e-05, "loss": 0.004, "step": 26500 }, { "grad_norm": 0.09918836504220963, "learning_rate": 6.358130535299862e-05, "loss": 0.0042, "step": 26510 }, { "grad_norm": 0.07376641035079956, "learning_rate": 6.355478155692926e-05, "loss": 0.0031, "step": 26520 }, { "grad_norm": 0.1322176456451416, "learning_rate": 6.352825364327517e-05, "loss": 0.0043, "step": 26530 }, { "grad_norm": 0.07203575223684311, "learning_rate": 6.350172162009482e-05, "loss": 0.0033, "step": 26540 }, { "grad_norm": 0.12143644690513611, "learning_rate": 6.347518549544793e-05, "loss": 0.0036, "step": 26550 }, { "grad_norm": 0.11072021722793579, "learning_rate": 6.344864527739547e-05, "loss": 0.0035, "step": 26560 }, { "grad_norm": 0.09134016931056976, "learning_rate": 6.342210097399966e-05, "loss": 0.0048, "step": 26570 }, { "grad_norm": 0.11944025754928589, "learning_rate": 6.339555259332398e-05, "loss": 0.0047, "step": 26580 }, { "grad_norm": 0.074167899787426, "learning_rate": 6.33690001434331e-05, "loss": 0.0034, "step": 26590 }, { "grad_norm": 0.09292297810316086, "learning_rate": 6.334244363239296e-05, "loss": 0.0051, "step": 26600 }, { "grad_norm": 0.08668912947177887, "learning_rate": 6.331588306827073e-05, "loss": 0.0039, "step": 26610 }, { "grad_norm": 0.09352515637874603, "learning_rate": 6.328931845913483e-05, "loss": 0.0035, "step": 26620 }, { "grad_norm": 0.10452084988355637, "learning_rate": 6.326274981305484e-05, "loss": 0.0042, "step": 26630 }, { "grad_norm": 0.11428911983966827, "learning_rate": 6.323617713810166e-05, "loss": 0.0039, "step": 26640 }, { "grad_norm": 0.08752621710300446, "learning_rate": 6.320960044234734e-05, "loss": 0.0035, "step": 26650 }, { "grad_norm": 0.13019266724586487, "learning_rate": 6.318301973386518e-05, "loss": 0.004, "step": 26660 }, { "grad_norm": 0.08284538239240646, "learning_rate": 6.315643502072971e-05, "loss": 0.0033, "step": 26670 }, { "grad_norm": 0.07567287236452103, "learning_rate": 6.312984631101667e-05, "loss": 0.0034, "step": 26680 }, { "grad_norm": 0.08771825581789017, "learning_rate": 6.310325361280297e-05, "loss": 0.0031, "step": 26690 }, { "grad_norm": 0.10956794023513794, "learning_rate": 6.30766569341668e-05, "loss": 0.0032, "step": 26700 }, { "grad_norm": 0.0812644362449646, "learning_rate": 6.305005628318753e-05, "loss": 0.0035, "step": 26710 }, { "grad_norm": 0.0668686106801033, "learning_rate": 6.302345166794572e-05, "loss": 0.0033, "step": 26720 }, { "grad_norm": 0.12930522859096527, "learning_rate": 6.299684309652316e-05, "loss": 0.0051, "step": 26730 }, { "grad_norm": 0.09216716140508652, "learning_rate": 6.297023057700283e-05, "loss": 0.0033, "step": 26740 }, { "grad_norm": 0.07941469550132751, "learning_rate": 6.294361411746891e-05, "loss": 0.0031, "step": 26750 }, { "grad_norm": 0.0887795090675354, "learning_rate": 6.291699372600677e-05, "loss": 0.0034, "step": 26760 }, { "grad_norm": 0.08927451819181442, "learning_rate": 6.2890369410703e-05, "loss": 0.0037, "step": 26770 }, { "grad_norm": 0.08747363090515137, "learning_rate": 6.286374117964534e-05, "loss": 0.0029, "step": 26780 }, { "grad_norm": 0.14267286658287048, "learning_rate": 6.283710904092277e-05, "loss": 0.005, "step": 26790 }, { "grad_norm": 0.12861067056655884, "learning_rate": 6.281047300262542e-05, "loss": 0.0041, "step": 26800 }, { "grad_norm": 0.11127923429012299, "learning_rate": 6.278383307284461e-05, "loss": 0.0034, "step": 26810 }, { "grad_norm": 0.10140615701675415, "learning_rate": 6.275718925967284e-05, "loss": 0.0034, "step": 26820 }, { "grad_norm": 0.08495113998651505, "learning_rate": 6.273054157120382e-05, "loss": 0.0036, "step": 26830 }, { "grad_norm": 0.07336421310901642, "learning_rate": 6.270389001553238e-05, "loss": 0.0027, "step": 26840 }, { "grad_norm": 0.11110903322696686, "learning_rate": 6.26772346007546e-05, "loss": 0.0035, "step": 26850 }, { "grad_norm": 0.088985376060009, "learning_rate": 6.265057533496767e-05, "loss": 0.0036, "step": 26860 }, { "grad_norm": 0.08699354529380798, "learning_rate": 6.262391222626997e-05, "loss": 0.0028, "step": 26870 }, { "grad_norm": 0.10084008425474167, "learning_rate": 6.259724528276106e-05, "loss": 0.0031, "step": 26880 }, { "grad_norm": 0.07686762511730194, "learning_rate": 6.257057451254162e-05, "loss": 0.0027, "step": 26890 }, { "grad_norm": 0.06372152268886566, "learning_rate": 6.254389992371357e-05, "loss": 0.003, "step": 26900 }, { "grad_norm": 0.07562585920095444, "learning_rate": 6.25172215243799e-05, "loss": 0.0037, "step": 26910 }, { "grad_norm": 0.09197675436735153, "learning_rate": 6.249053932264486e-05, "loss": 0.0032, "step": 26920 }, { "grad_norm": 0.09267592430114746, "learning_rate": 6.246385332661376e-05, "loss": 0.0035, "step": 26930 }, { "grad_norm": 0.07965932041406631, "learning_rate": 6.24371635443931e-05, "loss": 0.0029, "step": 26940 }, { "grad_norm": 0.10861317813396454, "learning_rate": 6.241046998409054e-05, "loss": 0.0029, "step": 26950 }, { "grad_norm": 0.10842045396566391, "learning_rate": 6.238377265381489e-05, "loss": 0.0032, "step": 26960 }, { "grad_norm": 0.07028055191040039, "learning_rate": 6.235707156167607e-05, "loss": 0.0027, "step": 26970 }, { "grad_norm": 0.09579256922006607, "learning_rate": 6.233036671578519e-05, "loss": 0.0035, "step": 26980 }, { "grad_norm": 0.11452043801546097, "learning_rate": 6.230365812425445e-05, "loss": 0.0037, "step": 26990 }, { "grad_norm": 0.07512276619672775, "learning_rate": 6.227694579519724e-05, "loss": 0.0037, "step": 27000 }, { "grad_norm": 0.06442194432020187, "learning_rate": 6.225022973672805e-05, "loss": 0.0033, "step": 27010 }, { "grad_norm": 0.07426450401544571, "learning_rate": 6.222350995696253e-05, "loss": 0.0043, "step": 27020 }, { "grad_norm": 0.07488783448934555, "learning_rate": 6.21967864640174e-05, "loss": 0.005, "step": 27030 }, { "grad_norm": 0.09248219430446625, "learning_rate": 6.217005926601059e-05, "loss": 0.0042, "step": 27040 }, { "grad_norm": 0.10651209205389023, "learning_rate": 6.214332837106111e-05, "loss": 0.0038, "step": 27050 }, { "grad_norm": 0.08330981433391571, "learning_rate": 6.21165937872891e-05, "loss": 0.0036, "step": 27060 }, { "grad_norm": 0.10615605115890503, "learning_rate": 6.208985552281582e-05, "loss": 0.0039, "step": 27070 }, { "grad_norm": 0.10323060303926468, "learning_rate": 6.206311358576364e-05, "loss": 0.004, "step": 27080 }, { "grad_norm": 0.07933149486780167, "learning_rate": 6.203636798425608e-05, "loss": 0.0033, "step": 27090 }, { "grad_norm": 0.07642421126365662, "learning_rate": 6.20096187264177e-05, "loss": 0.0032, "step": 27100 }, { "grad_norm": 0.08397579193115234, "learning_rate": 6.198286582037425e-05, "loss": 0.003, "step": 27110 }, { "grad_norm": 0.054705142974853516, "learning_rate": 6.195610927425256e-05, "loss": 0.0031, "step": 27120 }, { "grad_norm": 0.12815669178962708, "learning_rate": 6.192934909618056e-05, "loss": 0.0037, "step": 27130 }, { "grad_norm": 0.0997408926486969, "learning_rate": 6.190258529428728e-05, "loss": 0.0033, "step": 27140 }, { "grad_norm": 0.10731660574674606, "learning_rate": 6.187581787670285e-05, "loss": 0.0029, "step": 27150 }, { "grad_norm": 0.06253734976053238, "learning_rate": 6.184904685155852e-05, "loss": 0.0024, "step": 27160 }, { "grad_norm": 0.08937884867191315, "learning_rate": 6.18222722269866e-05, "loss": 0.0027, "step": 27170 }, { "grad_norm": 0.09209015965461731, "learning_rate": 6.179549401112053e-05, "loss": 0.0034, "step": 27180 }, { "grad_norm": 0.08406312018632889, "learning_rate": 6.176871221209482e-05, "loss": 0.0029, "step": 27190 }, { "grad_norm": 0.09715115278959274, "learning_rate": 6.174192683804508e-05, "loss": 0.0031, "step": 27200 }, { "grad_norm": 0.09505660086870193, "learning_rate": 6.1715137897108e-05, "loss": 0.0039, "step": 27210 }, { "grad_norm": 0.14835520088672638, "learning_rate": 6.168834539742134e-05, "loss": 0.0045, "step": 27220 }, { "grad_norm": 0.07551039010286331, "learning_rate": 6.166154934712397e-05, "loss": 0.004, "step": 27230 }, { "grad_norm": 0.08824001252651215, "learning_rate": 6.163474975435581e-05, "loss": 0.0036, "step": 27240 }, { "grad_norm": 0.12120774388313293, "learning_rate": 6.160794662725787e-05, "loss": 0.0046, "step": 27250 }, { "grad_norm": 0.1074206531047821, "learning_rate": 6.158113997397222e-05, "loss": 0.0032, "step": 27260 }, { "grad_norm": 0.08390028029680252, "learning_rate": 6.155432980264205e-05, "loss": 0.0033, "step": 27270 }, { "grad_norm": 0.10465911030769348, "learning_rate": 6.152751612141156e-05, "loss": 0.0036, "step": 27280 }, { "grad_norm": 0.08654304593801498, "learning_rate": 6.150069893842602e-05, "loss": 0.0032, "step": 27290 }, { "grad_norm": 0.08632276207208633, "learning_rate": 6.147387826183182e-05, "loss": 0.0037, "step": 27300 }, { "grad_norm": 0.07613120973110199, "learning_rate": 6.144705409977635e-05, "loss": 0.0031, "step": 27310 }, { "grad_norm": 0.0818396508693695, "learning_rate": 6.142022646040808e-05, "loss": 0.0029, "step": 27320 }, { "grad_norm": 0.06873863935470581, "learning_rate": 6.139339535187653e-05, "loss": 0.0034, "step": 27330 }, { "grad_norm": 0.07180017232894897, "learning_rate": 6.136656078233232e-05, "loss": 0.0033, "step": 27340 }, { "grad_norm": 0.07715154439210892, "learning_rate": 6.133972275992707e-05, "loss": 0.0027, "step": 27350 }, { "grad_norm": 0.07172761857509613, "learning_rate": 6.131288129281342e-05, "loss": 0.0029, "step": 27360 }, { "grad_norm": 0.09119707345962524, "learning_rate": 6.128603638914516e-05, "loss": 0.0027, "step": 27370 }, { "grad_norm": 0.09247050434350967, "learning_rate": 6.125918805707704e-05, "loss": 0.0028, "step": 27380 }, { "grad_norm": 0.059450019150972366, "learning_rate": 6.123233630476485e-05, "loss": 0.0021, "step": 27390 }, { "grad_norm": 0.06442841142416, "learning_rate": 6.120548114036547e-05, "loss": 0.0026, "step": 27400 }, { "grad_norm": 0.09965792298316956, "learning_rate": 6.117862257203679e-05, "loss": 0.0027, "step": 27410 }, { "grad_norm": 0.053747039288282394, "learning_rate": 6.115176060793771e-05, "loss": 0.0024, "step": 27420 }, { "grad_norm": 0.08584802597761154, "learning_rate": 6.112489525622822e-05, "loss": 0.0031, "step": 27430 }, { "grad_norm": 0.07227820158004761, "learning_rate": 6.109802652506928e-05, "loss": 0.004, "step": 27440 }, { "grad_norm": 0.11351548135280609, "learning_rate": 6.107115442262291e-05, "loss": 0.004, "step": 27450 }, { "grad_norm": 0.0738888531923294, "learning_rate": 6.104427895705214e-05, "loss": 0.0038, "step": 27460 }, { "grad_norm": 0.06578067690134048, "learning_rate": 6.101740013652103e-05, "loss": 0.0036, "step": 27470 }, { "grad_norm": 0.04338997229933739, "learning_rate": 6.099051796919465e-05, "loss": 0.0031, "step": 27480 }, { "grad_norm": 0.07222351431846619, "learning_rate": 6.096363246323911e-05, "loss": 0.003, "step": 27490 }, { "grad_norm": 0.08048726618289948, "learning_rate": 6.0936743626821504e-05, "loss": 0.0025, "step": 27500 }, { "grad_norm": 0.06818561255931854, "learning_rate": 6.090985146810996e-05, "loss": 0.003, "step": 27510 }, { "grad_norm": 0.0845467820763588, "learning_rate": 6.088295599527357e-05, "loss": 0.0037, "step": 27520 }, { "grad_norm": 0.062451571226119995, "learning_rate": 6.085605721648252e-05, "loss": 0.0028, "step": 27530 }, { "grad_norm": 0.09454899281263351, "learning_rate": 6.082915513990792e-05, "loss": 0.0029, "step": 27540 }, { "grad_norm": 0.10547351092100143, "learning_rate": 6.080224977372192e-05, "loss": 0.0023, "step": 27550 }, { "grad_norm": 0.08965689688920975, "learning_rate": 6.0775341126097666e-05, "loss": 0.0035, "step": 27560 }, { "grad_norm": 0.07591979950666428, "learning_rate": 6.074842920520926e-05, "loss": 0.0033, "step": 27570 }, { "grad_norm": 0.0838463306427002, "learning_rate": 6.072151401923186e-05, "loss": 0.0027, "step": 27580 }, { "grad_norm": 0.07876330614089966, "learning_rate": 6.069459557634159e-05, "loss": 0.003, "step": 27590 }, { "grad_norm": 0.09553840011358261, "learning_rate": 6.066767388471557e-05, "loss": 0.0031, "step": 27600 }, { "grad_norm": 0.107811838388443, "learning_rate": 6.064074895253188e-05, "loss": 0.0029, "step": 27610 }, { "grad_norm": 0.09240636974573135, "learning_rate": 6.061382078796961e-05, "loss": 0.0038, "step": 27620 }, { "grad_norm": 0.09095104783773422, "learning_rate": 6.0586889399208814e-05, "loss": 0.0035, "step": 27630 }, { "grad_norm": 0.1271321028470993, "learning_rate": 6.0559954794430565e-05, "loss": 0.0039, "step": 27640 }, { "grad_norm": 0.11581595242023468, "learning_rate": 6.053301698181687e-05, "loss": 0.0033, "step": 27650 }, { "grad_norm": 0.06968104094266891, "learning_rate": 6.0506075969550725e-05, "loss": 0.0034, "step": 27660 }, { "grad_norm": 0.08756158500909805, "learning_rate": 6.047913176581609e-05, "loss": 0.0031, "step": 27670 }, { "grad_norm": 0.07955053448677063, "learning_rate": 6.0452184378797904e-05, "loss": 0.0031, "step": 27680 }, { "grad_norm": 0.11758486926555634, "learning_rate": 6.042523381668209e-05, "loss": 0.0046, "step": 27690 }, { "grad_norm": 0.07785062491893768, "learning_rate": 6.03982800876555e-05, "loss": 0.0029, "step": 27700 }, { "grad_norm": 0.08095085620880127, "learning_rate": 6.0371323199905975e-05, "loss": 0.0049, "step": 27710 }, { "grad_norm": 0.150390625, "learning_rate": 6.03443631616223e-05, "loss": 0.0051, "step": 27720 }, { "grad_norm": 0.09310659766197205, "learning_rate": 6.031739998099421e-05, "loss": 0.0036, "step": 27730 }, { "grad_norm": 0.10321471095085144, "learning_rate": 6.029043366621243e-05, "loss": 0.0033, "step": 27740 }, { "grad_norm": 0.07468841224908829, "learning_rate": 6.0263464225468615e-05, "loss": 0.0038, "step": 27750 }, { "grad_norm": 0.1139456257224083, "learning_rate": 6.023649166695534e-05, "loss": 0.0038, "step": 27760 }, { "grad_norm": 0.11295470595359802, "learning_rate": 6.0209515998866186e-05, "loss": 0.0031, "step": 27770 }, { "grad_norm": 0.0843566283583641, "learning_rate": 6.018253722939563e-05, "loss": 0.0043, "step": 27780 }, { "grad_norm": 0.07181258499622345, "learning_rate": 6.015555536673914e-05, "loss": 0.0032, "step": 27790 }, { "grad_norm": 0.15106022357940674, "learning_rate": 6.0128570419093054e-05, "loss": 0.0045, "step": 27800 }, { "grad_norm": 0.10600783675909042, "learning_rate": 6.010158239465471e-05, "loss": 0.0028, "step": 27810 }, { "grad_norm": 0.09014402329921722, "learning_rate": 6.007459130162235e-05, "loss": 0.0034, "step": 27820 }, { "grad_norm": 0.08217724412679672, "learning_rate": 6.004759714819516e-05, "loss": 0.003, "step": 27830 }, { "grad_norm": 0.08221474289894104, "learning_rate": 6.002059994257323e-05, "loss": 0.0023, "step": 27840 }, { "grad_norm": 0.09432736039161682, "learning_rate": 5.999359969295764e-05, "loss": 0.0029, "step": 27850 }, { "grad_norm": 0.09306759387254715, "learning_rate": 5.9966596407550314e-05, "loss": 0.0029, "step": 27860 }, { "grad_norm": 0.08779420703649521, "learning_rate": 5.993959009455416e-05, "loss": 0.0026, "step": 27870 }, { "grad_norm": 0.07403838634490967, "learning_rate": 5.991258076217298e-05, "loss": 0.0041, "step": 27880 }, { "grad_norm": 0.09922095388174057, "learning_rate": 5.988556841861147e-05, "loss": 0.004, "step": 27890 }, { "grad_norm": 0.10472874343395233, "learning_rate": 5.985855307207531e-05, "loss": 0.0037, "step": 27900 }, { "grad_norm": 0.13561904430389404, "learning_rate": 5.9831534730771e-05, "loss": 0.0042, "step": 27910 }, { "grad_norm": 0.1444617211818695, "learning_rate": 5.980451340290605e-05, "loss": 0.0049, "step": 27920 }, { "grad_norm": 0.1080622598528862, "learning_rate": 5.97774890966888e-05, "loss": 0.003, "step": 27930 }, { "grad_norm": 0.1017281636595726, "learning_rate": 5.975046182032851e-05, "loss": 0.0037, "step": 27940 }, { "grad_norm": 0.09137111902236938, "learning_rate": 5.972343158203537e-05, "loss": 0.0028, "step": 27950 }, { "grad_norm": 0.1136079728603363, "learning_rate": 5.969639839002045e-05, "loss": 0.0037, "step": 27960 }, { "grad_norm": 0.09347692877054214, "learning_rate": 5.966936225249572e-05, "loss": 0.0027, "step": 27970 }, { "grad_norm": 0.06828472018241882, "learning_rate": 5.9642323177674044e-05, "loss": 0.0034, "step": 27980 }, { "grad_norm": 0.07000566273927689, "learning_rate": 5.9615281173769154e-05, "loss": 0.0045, "step": 27990 }, { "grad_norm": 0.11460168659687042, "learning_rate": 5.958823624899574e-05, "loss": 0.0038, "step": 28000 }, { "grad_norm": 0.06664735823869705, "learning_rate": 5.956118841156933e-05, "loss": 0.0037, "step": 28010 }, { "grad_norm": 0.09272653609514236, "learning_rate": 5.953413766970631e-05, "loss": 0.0038, "step": 28020 }, { "grad_norm": 0.07810530066490173, "learning_rate": 5.9507084031624e-05, "loss": 0.0034, "step": 28030 }, { "grad_norm": 0.12326904386281967, "learning_rate": 5.948002750554058e-05, "loss": 0.0047, "step": 28040 }, { "grad_norm": 0.09980030357837677, "learning_rate": 5.9452968099675124e-05, "loss": 0.0036, "step": 28050 }, { "grad_norm": 0.09222137182950974, "learning_rate": 5.9425905822247527e-05, "loss": 0.0043, "step": 28060 }, { "grad_norm": 0.07992402464151382, "learning_rate": 5.939884068147864e-05, "loss": 0.0029, "step": 28070 }, { "grad_norm": 0.07398588210344315, "learning_rate": 5.937177268559011e-05, "loss": 0.0042, "step": 28080 }, { "grad_norm": 0.10543950647115707, "learning_rate": 5.934470184280448e-05, "loss": 0.0035, "step": 28090 }, { "grad_norm": 0.06866272538900375, "learning_rate": 5.931762816134516e-05, "loss": 0.0032, "step": 28100 }, { "grad_norm": 0.10603678226470947, "learning_rate": 5.9290551649436434e-05, "loss": 0.004, "step": 28110 }, { "grad_norm": 0.09136892855167389, "learning_rate": 5.9263472315303416e-05, "loss": 0.0035, "step": 28120 }, { "grad_norm": 0.058811672031879425, "learning_rate": 5.9236390167172096e-05, "loss": 0.0036, "step": 28130 }, { "grad_norm": 0.07639066874980927, "learning_rate": 5.920930521326932e-05, "loss": 0.004, "step": 28140 }, { "grad_norm": 0.10603076964616776, "learning_rate": 5.918221746182276e-05, "loss": 0.0033, "step": 28150 }, { "grad_norm": 0.08737289905548096, "learning_rate": 5.9155126921061e-05, "loss": 0.003, "step": 28160 }, { "grad_norm": 0.08338155597448349, "learning_rate": 5.91280335992134e-05, "loss": 0.0032, "step": 28170 }, { "grad_norm": 0.07445930689573288, "learning_rate": 5.91009375045102e-05, "loss": 0.0038, "step": 28180 }, { "grad_norm": 0.08424690365791321, "learning_rate": 5.9073838645182476e-05, "loss": 0.003, "step": 28190 }, { "grad_norm": 0.07055103033781052, "learning_rate": 5.904673702946217e-05, "loss": 0.0025, "step": 28200 }, { "grad_norm": 0.09559841454029083, "learning_rate": 5.9019632665582004e-05, "loss": 0.0027, "step": 28210 }, { "grad_norm": 0.0647473931312561, "learning_rate": 5.899252556177559e-05, "loss": 0.0024, "step": 28220 }, { "grad_norm": 0.07324642688035965, "learning_rate": 5.896541572627735e-05, "loss": 0.0024, "step": 28230 }, { "grad_norm": 0.06546119600534439, "learning_rate": 5.893830316732253e-05, "loss": 0.0031, "step": 28240 }, { "grad_norm": 0.09195452928543091, "learning_rate": 5.8911187893147214e-05, "loss": 0.0043, "step": 28250 }, { "grad_norm": 0.12869946658611298, "learning_rate": 5.888406991198828e-05, "loss": 0.0043, "step": 28260 }, { "grad_norm": 0.07474249601364136, "learning_rate": 5.885694923208349e-05, "loss": 0.004, "step": 28270 }, { "grad_norm": 0.13440127670764923, "learning_rate": 5.882982586167138e-05, "loss": 0.0033, "step": 28280 }, { "grad_norm": 0.1278272122144699, "learning_rate": 5.880269980899131e-05, "loss": 0.0035, "step": 28290 }, { "grad_norm": 0.08178836107254028, "learning_rate": 5.8775571082283465e-05, "loss": 0.0028, "step": 28300 }, { "grad_norm": 0.10433200001716614, "learning_rate": 5.8748439689788824e-05, "loss": 0.0039, "step": 28310 }, { "grad_norm": 0.06033588573336601, "learning_rate": 5.87213056397492e-05, "loss": 0.0027, "step": 28320 }, { "grad_norm": 0.06601276993751526, "learning_rate": 5.869416894040719e-05, "loss": 0.0028, "step": 28330 }, { "grad_norm": 0.07036713510751724, "learning_rate": 5.866702960000621e-05, "loss": 0.0037, "step": 28340 }, { "grad_norm": 0.1048002690076828, "learning_rate": 5.863988762679048e-05, "loss": 0.0028, "step": 28350 }, { "grad_norm": 0.09363327920436859, "learning_rate": 5.8612743029005e-05, "loss": 0.0031, "step": 28360 }, { "grad_norm": 0.056990016251802444, "learning_rate": 5.858559581489561e-05, "loss": 0.003, "step": 28370 }, { "grad_norm": 0.1000407487154007, "learning_rate": 5.85584459927089e-05, "loss": 0.003, "step": 28380 }, { "grad_norm": 0.09320258349180222, "learning_rate": 5.853129357069227e-05, "loss": 0.0035, "step": 28390 }, { "grad_norm": 0.07430203258991241, "learning_rate": 5.8504138557093913e-05, "loss": 0.0025, "step": 28400 }, { "grad_norm": 0.07402394711971283, "learning_rate": 5.8476980960162784e-05, "loss": 0.0026, "step": 28410 }, { "grad_norm": 0.08215559273958206, "learning_rate": 5.844982078814868e-05, "loss": 0.0028, "step": 28420 }, { "grad_norm": 0.07791217416524887, "learning_rate": 5.842265804930211e-05, "loss": 0.003, "step": 28430 }, { "grad_norm": 0.07207804918289185, "learning_rate": 5.839549275187444e-05, "loss": 0.0025, "step": 28440 }, { "grad_norm": 0.09973877668380737, "learning_rate": 5.836832490411771e-05, "loss": 0.0027, "step": 28450 }, { "grad_norm": 0.10177882760763168, "learning_rate": 5.834115451428485e-05, "loss": 0.0023, "step": 28460 }, { "grad_norm": 0.06301014125347137, "learning_rate": 5.831398159062946e-05, "loss": 0.0032, "step": 28470 }, { "grad_norm": 0.06592923402786255, "learning_rate": 5.828680614140599e-05, "loss": 0.0032, "step": 28480 }, { "grad_norm": 0.06499461829662323, "learning_rate": 5.825962817486962e-05, "loss": 0.0026, "step": 28490 }, { "grad_norm": 0.10867203772068024, "learning_rate": 5.823244769927629e-05, "loss": 0.0034, "step": 28500 }, { "grad_norm": 0.07972925901412964, "learning_rate": 5.8205264722882716e-05, "loss": 0.0031, "step": 28510 }, { "grad_norm": 0.07003040611743927, "learning_rate": 5.817807925394636e-05, "loss": 0.0028, "step": 28520 }, { "grad_norm": 0.06713811308145523, "learning_rate": 5.815089130072546e-05, "loss": 0.0023, "step": 28530 }, { "grad_norm": 0.09446865320205688, "learning_rate": 5.8123700871479e-05, "loss": 0.0031, "step": 28540 }, { "grad_norm": 0.07467010617256165, "learning_rate": 5.809650797446671e-05, "loss": 0.0023, "step": 28550 }, { "grad_norm": 0.0893702581524849, "learning_rate": 5.806931261794907e-05, "loss": 0.0029, "step": 28560 }, { "grad_norm": 0.08077873289585114, "learning_rate": 5.804211481018731e-05, "loss": 0.0034, "step": 28570 }, { "grad_norm": 0.12111235409975052, "learning_rate": 5.801491455944341e-05, "loss": 0.0043, "step": 28580 }, { "grad_norm": 0.08619200438261032, "learning_rate": 5.79877118739801e-05, "loss": 0.003, "step": 28590 }, { "grad_norm": 0.07695353031158447, "learning_rate": 5.7960506762060816e-05, "loss": 0.0038, "step": 28600 }, { "grad_norm": 0.10011881589889526, "learning_rate": 5.793329923194977e-05, "loss": 0.0026, "step": 28610 }, { "grad_norm": 0.07463487237691879, "learning_rate": 5.790608929191187e-05, "loss": 0.0027, "step": 28620 }, { "grad_norm": 0.05811859667301178, "learning_rate": 5.78788769502128e-05, "loss": 0.0029, "step": 28630 }, { "grad_norm": 0.07419785112142563, "learning_rate": 5.785166221511894e-05, "loss": 0.0028, "step": 28640 }, { "grad_norm": 0.08759347349405289, "learning_rate": 5.7824445094897415e-05, "loss": 0.0029, "step": 28650 }, { "grad_norm": 0.08990435302257538, "learning_rate": 5.7797225597816065e-05, "loss": 0.0029, "step": 28660 }, { "grad_norm": 0.1007116362452507, "learning_rate": 5.777000373214345e-05, "loss": 0.0037, "step": 28670 }, { "grad_norm": 0.05588287115097046, "learning_rate": 5.774277950614885e-05, "loss": 0.0023, "step": 28680 }, { "grad_norm": 0.08644215017557144, "learning_rate": 5.771555292810227e-05, "loss": 0.003, "step": 28690 }, { "grad_norm": 0.09438773989677429, "learning_rate": 5.768832400627444e-05, "loss": 0.0026, "step": 28700 }, { "grad_norm": 0.08451443165540695, "learning_rate": 5.7661092748936775e-05, "loss": 0.0026, "step": 28710 }, { "grad_norm": 0.084688700735569, "learning_rate": 5.76338591643614e-05, "loss": 0.0029, "step": 28720 }, { "grad_norm": 0.07217828929424286, "learning_rate": 5.760662326082118e-05, "loss": 0.0028, "step": 28730 }, { "grad_norm": 0.08163123577833176, "learning_rate": 5.757938504658965e-05, "loss": 0.0034, "step": 28740 }, { "grad_norm": 0.09829612821340561, "learning_rate": 5.755214452994107e-05, "loss": 0.004, "step": 28750 }, { "grad_norm": 0.10674062371253967, "learning_rate": 5.752490171915039e-05, "loss": 0.005, "step": 28760 }, { "grad_norm": 0.09864719957113266, "learning_rate": 5.749765662249324e-05, "loss": 0.0032, "step": 28770 }, { "grad_norm": 0.08083617687225342, "learning_rate": 5.747040924824596e-05, "loss": 0.0031, "step": 28780 }, { "grad_norm": 0.08335025608539581, "learning_rate": 5.7443159604685613e-05, "loss": 0.0029, "step": 28790 }, { "grad_norm": 0.07519299536943436, "learning_rate": 5.74159077000899e-05, "loss": 0.0027, "step": 28800 }, { "grad_norm": 0.09245875477790833, "learning_rate": 5.7388653542737235e-05, "loss": 0.0038, "step": 28810 }, { "grad_norm": 0.07338842749595642, "learning_rate": 5.736139714090672e-05, "loss": 0.0036, "step": 28820 }, { "grad_norm": 0.11181720346212387, "learning_rate": 5.73341385028781e-05, "loss": 0.0031, "step": 28830 }, { "grad_norm": 0.10324276983737946, "learning_rate": 5.7306877636931855e-05, "loss": 0.0036, "step": 28840 }, { "grad_norm": 0.10030847787857056, "learning_rate": 5.7279614551349125e-05, "loss": 0.0044, "step": 28850 }, { "grad_norm": 0.13690418004989624, "learning_rate": 5.725234925441169e-05, "loss": 0.0042, "step": 28860 }, { "grad_norm": 0.10150916129350662, "learning_rate": 5.7225081754402044e-05, "loss": 0.0051, "step": 28870 }, { "grad_norm": 0.06732869893312454, "learning_rate": 5.7197812059603326e-05, "loss": 0.004, "step": 28880 }, { "grad_norm": 0.06216065213084221, "learning_rate": 5.717054017829934e-05, "loss": 0.0031, "step": 28890 }, { "grad_norm": 0.09184765815734863, "learning_rate": 5.7143266118774584e-05, "loss": 0.0039, "step": 28900 }, { "grad_norm": 0.09092894941568375, "learning_rate": 5.711598988931418e-05, "loss": 0.0045, "step": 28910 }, { "grad_norm": 0.09006525576114655, "learning_rate": 5.7088711498203954e-05, "loss": 0.0035, "step": 28920 }, { "grad_norm": 0.08309999853372574, "learning_rate": 5.706143095373033e-05, "loss": 0.0038, "step": 28930 }, { "grad_norm": 0.0867915228009224, "learning_rate": 5.703414826418042e-05, "loss": 0.0031, "step": 28940 }, { "grad_norm": 0.08510559052228928, "learning_rate": 5.7006863437842007e-05, "loss": 0.0038, "step": 28950 }, { "grad_norm": 0.1000879555940628, "learning_rate": 5.697957648300348e-05, "loss": 0.0041, "step": 28960 }, { "grad_norm": 0.11264339089393616, "learning_rate": 5.695228740795391e-05, "loss": 0.0034, "step": 28970 }, { "grad_norm": 0.08157478272914886, "learning_rate": 5.6924996220982985e-05, "loss": 0.0029, "step": 28980 }, { "grad_norm": 0.08235388249158859, "learning_rate": 5.6897702930381045e-05, "loss": 0.0032, "step": 28990 }, { "grad_norm": 0.09278234839439392, "learning_rate": 5.687040754443908e-05, "loss": 0.0027, "step": 29000 }, { "grad_norm": 0.08451757580041885, "learning_rate": 5.6843110071448725e-05, "loss": 0.0039, "step": 29010 }, { "grad_norm": 0.10981625318527222, "learning_rate": 5.6815810519702194e-05, "loss": 0.0032, "step": 29020 }, { "grad_norm": 0.05694878101348877, "learning_rate": 5.6788508897492396e-05, "loss": 0.0028, "step": 29030 }, { "grad_norm": 0.06003479287028313, "learning_rate": 5.676120521311282e-05, "loss": 0.003, "step": 29040 }, { "grad_norm": 0.0814424380660057, "learning_rate": 5.6733899474857634e-05, "loss": 0.0035, "step": 29050 }, { "grad_norm": 0.07935577630996704, "learning_rate": 5.670659169102157e-05, "loss": 0.0032, "step": 29060 }, { "grad_norm": 0.09262090921401978, "learning_rate": 5.6679281869900044e-05, "loss": 0.003, "step": 29070 }, { "grad_norm": 0.07549337297677994, "learning_rate": 5.6651970019789045e-05, "loss": 0.0033, "step": 29080 }, { "grad_norm": 0.08846556395292282, "learning_rate": 5.662465614898519e-05, "loss": 0.003, "step": 29090 }, { "grad_norm": 0.06528259813785553, "learning_rate": 5.6597340265785695e-05, "loss": 0.0035, "step": 29100 }, { "grad_norm": 0.07799164950847626, "learning_rate": 5.657002237848843e-05, "loss": 0.0031, "step": 29110 }, { "grad_norm": 0.07073277235031128, "learning_rate": 5.654270249539183e-05, "loss": 0.0034, "step": 29120 }, { "grad_norm": 0.09931930899620056, "learning_rate": 5.651538062479498e-05, "loss": 0.0029, "step": 29130 }, { "grad_norm": 0.09860724955797195, "learning_rate": 5.648805677499751e-05, "loss": 0.0037, "step": 29140 }, { "grad_norm": 0.08486355096101761, "learning_rate": 5.646073095429969e-05, "loss": 0.0037, "step": 29150 }, { "grad_norm": 0.13641156256198883, "learning_rate": 5.643340317100241e-05, "loss": 0.0036, "step": 29160 }, { "grad_norm": 0.06420416384935379, "learning_rate": 5.64060734334071e-05, "loss": 0.0033, "step": 29170 }, { "grad_norm": 0.0873589813709259, "learning_rate": 5.637874174981583e-05, "loss": 0.0033, "step": 29180 }, { "grad_norm": 0.10583397001028061, "learning_rate": 5.635140812853124e-05, "loss": 0.0029, "step": 29190 }, { "grad_norm": 0.09737902134656906, "learning_rate": 5.6324072577856544e-05, "loss": 0.0022, "step": 29200 }, { "grad_norm": 0.10459507256746292, "learning_rate": 5.629673510609559e-05, "loss": 0.004, "step": 29210 }, { "grad_norm": 0.0780731663107872, "learning_rate": 5.626939572155276e-05, "loss": 0.0032, "step": 29220 }, { "grad_norm": 0.08103002607822418, "learning_rate": 5.6242054432533054e-05, "loss": 0.0028, "step": 29230 }, { "grad_norm": 0.08309122920036316, "learning_rate": 5.621471124734201e-05, "loss": 0.0034, "step": 29240 }, { "grad_norm": 0.08476749062538147, "learning_rate": 5.6187366174285794e-05, "loss": 0.0027, "step": 29250 }, { "grad_norm": 0.08358164131641388, "learning_rate": 5.616001922167109e-05, "loss": 0.0035, "step": 29260 }, { "grad_norm": 0.07975678145885468, "learning_rate": 5.61326703978052e-05, "loss": 0.003, "step": 29270 }, { "grad_norm": 0.10801216214895248, "learning_rate": 5.6105319710995964e-05, "loss": 0.0034, "step": 29280 }, { "grad_norm": 0.07247752696275711, "learning_rate": 5.60779671695518e-05, "loss": 0.003, "step": 29290 }, { "grad_norm": 0.10870887339115143, "learning_rate": 5.6050612781781684e-05, "loss": 0.0031, "step": 29300 }, { "grad_norm": 0.08434104174375534, "learning_rate": 5.602325655599516e-05, "loss": 0.0027, "step": 29310 }, { "grad_norm": 0.10294057428836823, "learning_rate": 5.599589850050234e-05, "loss": 0.0035, "step": 29320 }, { "grad_norm": 0.07081089913845062, "learning_rate": 5.5968538623613874e-05, "loss": 0.0025, "step": 29330 }, { "grad_norm": 0.07435920834541321, "learning_rate": 5.594117693364095e-05, "loss": 0.0033, "step": 29340 }, { "grad_norm": 0.08621823787689209, "learning_rate": 5.591381343889535e-05, "loss": 0.0036, "step": 29350 }, { "grad_norm": 0.1062784492969513, "learning_rate": 5.5886448147689355e-05, "loss": 0.003, "step": 29360 }, { "grad_norm": 0.10028819739818573, "learning_rate": 5.585908106833585e-05, "loss": 0.0029, "step": 29370 }, { "grad_norm": 0.06888052076101303, "learning_rate": 5.5831712209148226e-05, "loss": 0.0027, "step": 29380 }, { "grad_norm": 0.09643958508968353, "learning_rate": 5.58043415784404e-05, "loss": 0.0032, "step": 29390 }, { "grad_norm": 0.07409124821424484, "learning_rate": 5.577696918452686e-05, "loss": 0.0023, "step": 29400 }, { "grad_norm": 0.10829959809780121, "learning_rate": 5.5749595035722604e-05, "loss": 0.0031, "step": 29410 }, { "grad_norm": 0.09104298800230026, "learning_rate": 5.5722219140343193e-05, "loss": 0.0032, "step": 29420 }, { "grad_norm": 0.13517454266548157, "learning_rate": 5.56948415067047e-05, "loss": 0.0043, "step": 29430 }, { "grad_norm": 0.06474646925926208, "learning_rate": 5.5667462143123704e-05, "loss": 0.0022, "step": 29440 }, { "grad_norm": 0.07642099261283875, "learning_rate": 5.564008105791737e-05, "loss": 0.0026, "step": 29450 }, { "grad_norm": 0.12363921850919724, "learning_rate": 5.5612698259403316e-05, "loss": 0.0042, "step": 29460 }, { "grad_norm": 0.09491749852895737, "learning_rate": 5.5585313755899724e-05, "loss": 0.0036, "step": 29470 }, { "grad_norm": 0.0787387564778328, "learning_rate": 5.5557927555725285e-05, "loss": 0.0034, "step": 29480 }, { "grad_norm": 0.08059460669755936, "learning_rate": 5.55305396671992e-05, "loss": 0.004, "step": 29490 }, { "grad_norm": 0.08942869305610657, "learning_rate": 5.55031500986412e-05, "loss": 0.0038, "step": 29500 }, { "grad_norm": 0.0753943920135498, "learning_rate": 5.547575885837149e-05, "loss": 0.0028, "step": 29510 }, { "grad_norm": 0.08620470762252808, "learning_rate": 5.5448365954710825e-05, "loss": 0.0033, "step": 29520 }, { "grad_norm": 0.06318535655736923, "learning_rate": 5.5420971395980446e-05, "loss": 0.0025, "step": 29530 }, { "grad_norm": 0.06675083935260773, "learning_rate": 5.539357519050209e-05, "loss": 0.0029, "step": 29540 }, { "grad_norm": 0.05982280150055885, "learning_rate": 5.536617734659799e-05, "loss": 0.0025, "step": 29550 }, { "grad_norm": 0.09121628850698471, "learning_rate": 5.533877787259091e-05, "loss": 0.0025, "step": 29560 }, { "grad_norm": 0.06540267169475555, "learning_rate": 5.5311376776804044e-05, "loss": 0.0026, "step": 29570 }, { "grad_norm": 0.07290727645158768, "learning_rate": 5.528397406756118e-05, "loss": 0.0043, "step": 29580 }, { "grad_norm": 0.07917812466621399, "learning_rate": 5.525656975318652e-05, "loss": 0.0028, "step": 29590 }, { "grad_norm": 0.0878433808684349, "learning_rate": 5.522916384200474e-05, "loss": 0.003, "step": 29600 }, { "grad_norm": 0.07575879245996475, "learning_rate": 5.520175634234106e-05, "loss": 0.0027, "step": 29610 }, { "grad_norm": 0.05189410597085953, "learning_rate": 5.517434726252113e-05, "loss": 0.0028, "step": 29620 }, { "grad_norm": 0.08463402837514877, "learning_rate": 5.514693661087113e-05, "loss": 0.0043, "step": 29630 }, { "grad_norm": 0.06807173043489456, "learning_rate": 5.511952439571769e-05, "loss": 0.003, "step": 29640 }, { "grad_norm": 0.08920951187610626, "learning_rate": 5.509211062538791e-05, "loss": 0.0027, "step": 29650 }, { "grad_norm": 0.0560191310942173, "learning_rate": 5.506469530820939e-05, "loss": 0.0025, "step": 29660 }, { "grad_norm": 0.0656784325838089, "learning_rate": 5.503727845251014e-05, "loss": 0.0023, "step": 29670 }, { "grad_norm": 0.08259536325931549, "learning_rate": 5.50098600666187e-05, "loss": 0.0028, "step": 29680 }, { "grad_norm": 0.07978387922048569, "learning_rate": 5.498244015886406e-05, "loss": 0.003, "step": 29690 }, { "grad_norm": 0.07972588390111923, "learning_rate": 5.495501873757565e-05, "loss": 0.0036, "step": 29700 }, { "grad_norm": 0.1195143535733223, "learning_rate": 5.492759581108336e-05, "loss": 0.0035, "step": 29710 }, { "grad_norm": 0.0809992253780365, "learning_rate": 5.490017138771759e-05, "loss": 0.0024, "step": 29720 }, { "grad_norm": 0.0855342373251915, "learning_rate": 5.487274547580912e-05, "loss": 0.0024, "step": 29730 }, { "grad_norm": 0.09316588938236237, "learning_rate": 5.484531808368923e-05, "loss": 0.0026, "step": 29740 }, { "grad_norm": 0.11520792543888092, "learning_rate": 5.4817889219689656e-05, "loss": 0.0032, "step": 29750 }, { "grad_norm": 0.10419368743896484, "learning_rate": 5.4790458892142536e-05, "loss": 0.0029, "step": 29760 }, { "grad_norm": 0.0985916331410408, "learning_rate": 5.476302710938048e-05, "loss": 0.0029, "step": 29770 }, { "grad_norm": 0.08840324729681015, "learning_rate": 5.473559387973657e-05, "loss": 0.0037, "step": 29780 }, { "grad_norm": 0.11202768981456757, "learning_rate": 5.470815921154425e-05, "loss": 0.0032, "step": 29790 }, { "grad_norm": 0.08590782433748245, "learning_rate": 5.468072311313749e-05, "loss": 0.0031, "step": 29800 }, { "grad_norm": 0.10919313877820969, "learning_rate": 5.465328559285063e-05, "loss": 0.004, "step": 29810 }, { "grad_norm": 0.09292511641979218, "learning_rate": 5.462584665901849e-05, "loss": 0.003, "step": 29820 }, { "grad_norm": 0.04974918067455292, "learning_rate": 5.4598406319976235e-05, "loss": 0.0026, "step": 29830 }, { "grad_norm": 0.07373159378767014, "learning_rate": 5.457096458405958e-05, "loss": 0.0034, "step": 29840 }, { "grad_norm": 0.06707640737295151, "learning_rate": 5.454352145960457e-05, "loss": 0.0034, "step": 29850 }, { "grad_norm": 0.06387105584144592, "learning_rate": 5.4516076954947715e-05, "loss": 0.0032, "step": 29860 }, { "grad_norm": 0.07112947106361389, "learning_rate": 5.448863107842591e-05, "loss": 0.0029, "step": 29870 }, { "grad_norm": 0.10043218731880188, "learning_rate": 5.446118383837651e-05, "loss": 0.0028, "step": 29880 }, { "grad_norm": 0.07344444841146469, "learning_rate": 5.443373524313722e-05, "loss": 0.0037, "step": 29890 }, { "grad_norm": 0.05241670459508896, "learning_rate": 5.440628530104626e-05, "loss": 0.003, "step": 29900 }, { "grad_norm": 0.0771351307630539, "learning_rate": 5.4378834020442146e-05, "loss": 0.0027, "step": 29910 }, { "grad_norm": 0.09057626873254776, "learning_rate": 5.4351381409663884e-05, "loss": 0.004, "step": 29920 }, { "grad_norm": 0.07656253129243851, "learning_rate": 5.432392747705084e-05, "loss": 0.004, "step": 29930 }, { "grad_norm": 0.1010526567697525, "learning_rate": 5.429647223094278e-05, "loss": 0.0037, "step": 29940 }, { "grad_norm": 0.09671127051115036, "learning_rate": 5.4269015679679924e-05, "loss": 0.0046, "step": 29950 }, { "grad_norm": 0.07327692955732346, "learning_rate": 5.424155783160281e-05, "loss": 0.0026, "step": 29960 }, { "grad_norm": 0.06910273432731628, "learning_rate": 5.4214098695052415e-05, "loss": 0.003, "step": 29970 }, { "grad_norm": 0.09515377879142761, "learning_rate": 5.418663827837012e-05, "loss": 0.0033, "step": 29980 }, { "grad_norm": 0.06470029801130295, "learning_rate": 5.415917658989763e-05, "loss": 0.0023, "step": 29990 }, { "grad_norm": 0.06537194550037384, "learning_rate": 5.413171363797713e-05, "loss": 0.0023, "step": 30000 }, { "grad_norm": 0.07304520159959793, "learning_rate": 5.4104249430951116e-05, "loss": 0.0033, "step": 30010 }, { "grad_norm": 0.0672677680850029, "learning_rate": 5.4076783977162494e-05, "loss": 0.003, "step": 30020 }, { "grad_norm": 0.096573606133461, "learning_rate": 5.4049317284954525e-05, "loss": 0.003, "step": 30030 }, { "grad_norm": 0.07763859629631042, "learning_rate": 5.4021849362670884e-05, "loss": 0.0034, "step": 30040 }, { "grad_norm": 0.08747932314872742, "learning_rate": 5.3994380218655604e-05, "loss": 0.0026, "step": 30050 }, { "grad_norm": 0.0744764506816864, "learning_rate": 5.396690986125309e-05, "loss": 0.0027, "step": 30060 }, { "grad_norm": 0.12294887751340866, "learning_rate": 5.3939438298808075e-05, "loss": 0.0046, "step": 30070 }, { "grad_norm": 0.06986549496650696, "learning_rate": 5.3911965539665744e-05, "loss": 0.003, "step": 30080 }, { "grad_norm": 0.06699047982692719, "learning_rate": 5.388449159217156e-05, "loss": 0.0029, "step": 30090 }, { "grad_norm": 0.10736153274774551, "learning_rate": 5.3857016464671385e-05, "loss": 0.0056, "step": 30100 }, { "grad_norm": 0.08004549890756607, "learning_rate": 5.382954016551146e-05, "loss": 0.0032, "step": 30110 }, { "grad_norm": 0.08770285546779633, "learning_rate": 5.380206270303835e-05, "loss": 0.0026, "step": 30120 }, { "grad_norm": 0.07001674175262451, "learning_rate": 5.377458408559897e-05, "loss": 0.0033, "step": 30130 }, { "grad_norm": 0.08562340587377548, "learning_rate": 5.374710432154061e-05, "loss": 0.0032, "step": 30140 }, { "grad_norm": 0.09405549615621567, "learning_rate": 5.3719623419210886e-05, "loss": 0.0029, "step": 30150 }, { "grad_norm": 0.061156850308179855, "learning_rate": 5.3692141386957786e-05, "loss": 0.0024, "step": 30160 }, { "grad_norm": 0.1022692322731018, "learning_rate": 5.3664658233129616e-05, "loss": 0.0028, "step": 30170 }, { "grad_norm": 0.07009396702051163, "learning_rate": 5.363717396607504e-05, "loss": 0.003, "step": 30180 }, { "grad_norm": 0.043954942375421524, "learning_rate": 5.360968859414305e-05, "loss": 0.0022, "step": 30190 }, { "grad_norm": 0.07378990203142166, "learning_rate": 5.358220212568295e-05, "loss": 0.0028, "step": 30200 }, { "grad_norm": 0.08200185745954514, "learning_rate": 5.355471456904444e-05, "loss": 0.0036, "step": 30210 }, { "grad_norm": 0.10161449760198593, "learning_rate": 5.3527225932577495e-05, "loss": 0.0021, "step": 30220 }, { "grad_norm": 0.0802127867937088, "learning_rate": 5.349973622463246e-05, "loss": 0.0026, "step": 30230 }, { "grad_norm": 0.0959535539150238, "learning_rate": 5.3472245453559956e-05, "loss": 0.0031, "step": 30240 }, { "grad_norm": 0.08606383204460144, "learning_rate": 5.3444753627710955e-05, "loss": 0.0037, "step": 30250 }, { "grad_norm": 0.07543378323316574, "learning_rate": 5.341726075543676e-05, "loss": 0.0025, "step": 30260 }, { "grad_norm": 0.09052187949419022, "learning_rate": 5.338976684508898e-05, "loss": 0.0045, "step": 30270 }, { "grad_norm": 0.06647658348083496, "learning_rate": 5.336227190501953e-05, "loss": 0.0028, "step": 30280 }, { "grad_norm": 0.0817442461848259, "learning_rate": 5.3334775943580664e-05, "loss": 0.0026, "step": 30290 }, { "grad_norm": 0.10043880343437195, "learning_rate": 5.330727896912491e-05, "loss": 0.0033, "step": 30300 }, { "grad_norm": 0.09555930644273758, "learning_rate": 5.327978099000511e-05, "loss": 0.0027, "step": 30310 }, { "grad_norm": 0.0702461525797844, "learning_rate": 5.3252282014574465e-05, "loss": 0.0024, "step": 30320 }, { "grad_norm": 0.061503730714321136, "learning_rate": 5.322478205118641e-05, "loss": 0.004, "step": 30330 }, { "grad_norm": 0.0859205350279808, "learning_rate": 5.3197281108194704e-05, "loss": 0.0033, "step": 30340 }, { "grad_norm": 0.06216941401362419, "learning_rate": 5.316977919395342e-05, "loss": 0.0032, "step": 30350 }, { "grad_norm": 0.07714907824993134, "learning_rate": 5.314227631681691e-05, "loss": 0.0027, "step": 30360 }, { "grad_norm": 0.07193394750356674, "learning_rate": 5.311477248513982e-05, "loss": 0.0028, "step": 30370 }, { "grad_norm": 0.06874474883079529, "learning_rate": 5.30872677072771e-05, "loss": 0.0029, "step": 30380 }, { "grad_norm": 0.06703098863363266, "learning_rate": 5.3059761991583954e-05, "loss": 0.0025, "step": 30390 }, { "grad_norm": 0.08670962601900101, "learning_rate": 5.303225534641592e-05, "loss": 0.0044, "step": 30400 }, { "grad_norm": 0.06065334379673004, "learning_rate": 5.300474778012875e-05, "loss": 0.0043, "step": 30410 }, { "grad_norm": 0.06637758016586304, "learning_rate": 5.297723930107855e-05, "loss": 0.0033, "step": 30420 }, { "grad_norm": 0.0632854774594307, "learning_rate": 5.294972991762167e-05, "loss": 0.0025, "step": 30430 }, { "grad_norm": 0.08587957918643951, "learning_rate": 5.292221963811472e-05, "loss": 0.0021, "step": 30440 }, { "grad_norm": 0.08216466754674911, "learning_rate": 5.28947084709146e-05, "loss": 0.0027, "step": 30450 }, { "grad_norm": 0.09570375829935074, "learning_rate": 5.2867196424378465e-05, "loss": 0.0027, "step": 30460 }, { "grad_norm": 0.04739360883831978, "learning_rate": 5.2839683506863765e-05, "loss": 0.0023, "step": 30470 }, { "grad_norm": 0.05624457821249962, "learning_rate": 5.281216972672821e-05, "loss": 0.0026, "step": 30480 }, { "grad_norm": 0.10395187139511108, "learning_rate": 5.278465509232973e-05, "loss": 0.0025, "step": 30490 }, { "grad_norm": 0.06966777890920639, "learning_rate": 5.275713961202655e-05, "loss": 0.0022, "step": 30500 }, { "grad_norm": 0.06950273364782333, "learning_rate": 5.2729623294177165e-05, "loss": 0.0018, "step": 30510 }, { "grad_norm": 0.07290726155042648, "learning_rate": 5.270210614714028e-05, "loss": 0.0039, "step": 30520 }, { "grad_norm": 0.08043686300516129, "learning_rate": 5.267458817927491e-05, "loss": 0.0023, "step": 30530 }, { "grad_norm": 0.08670709282159805, "learning_rate": 5.264706939894026e-05, "loss": 0.0029, "step": 30540 }, { "grad_norm": 0.06420284509658813, "learning_rate": 5.261954981449584e-05, "loss": 0.0023, "step": 30550 }, { "grad_norm": 0.0894528478384018, "learning_rate": 5.2592029434301324e-05, "loss": 0.0029, "step": 30560 }, { "grad_norm": 0.06893451511859894, "learning_rate": 5.256450826671672e-05, "loss": 0.0031, "step": 30570 }, { "grad_norm": 0.06795936077833176, "learning_rate": 5.253698632010221e-05, "loss": 0.0024, "step": 30580 }, { "grad_norm": 0.07752522081136703, "learning_rate": 5.2509463602818246e-05, "loss": 0.0027, "step": 30590 }, { "grad_norm": 0.06375737488269806, "learning_rate": 5.248194012322549e-05, "loss": 0.0032, "step": 30600 }, { "grad_norm": 0.0960036888718605, "learning_rate": 5.245441588968486e-05, "loss": 0.0026, "step": 30610 }, { "grad_norm": 0.07590646296739578, "learning_rate": 5.242689091055748e-05, "loss": 0.0026, "step": 30620 }, { "grad_norm": 0.07888061553239822, "learning_rate": 5.239936519420473e-05, "loss": 0.0021, "step": 30630 }, { "grad_norm": 0.09199371188879013, "learning_rate": 5.2371838748988175e-05, "loss": 0.0023, "step": 30640 }, { "grad_norm": 0.046990640461444855, "learning_rate": 5.234431158326965e-05, "loss": 0.0025, "step": 30650 }, { "grad_norm": 0.07568442076444626, "learning_rate": 5.231678370541115e-05, "loss": 0.0026, "step": 30660 }, { "grad_norm": 0.06093497946858406, "learning_rate": 5.228925512377495e-05, "loss": 0.003, "step": 30670 }, { "grad_norm": 0.08824581652879715, "learning_rate": 5.2261725846723465e-05, "loss": 0.0032, "step": 30680 }, { "grad_norm": 0.09894698113203049, "learning_rate": 5.22341958826194e-05, "loss": 0.0028, "step": 30690 }, { "grad_norm": 0.07434555888175964, "learning_rate": 5.22066652398256e-05, "loss": 0.0025, "step": 30700 }, { "grad_norm": 0.11737482249736786, "learning_rate": 5.2179133926705185e-05, "loss": 0.003, "step": 30710 }, { "grad_norm": 0.09326669573783875, "learning_rate": 5.215160195162141e-05, "loss": 0.0026, "step": 30720 }, { "grad_norm": 0.07991283386945724, "learning_rate": 5.212406932293776e-05, "loss": 0.0037, "step": 30730 }, { "grad_norm": 0.1185639351606369, "learning_rate": 5.209653604901795e-05, "loss": 0.0026, "step": 30740 }, { "grad_norm": 0.09284752607345581, "learning_rate": 5.206900213822584e-05, "loss": 0.0032, "step": 30750 }, { "grad_norm": 0.0750393196940422, "learning_rate": 5.204146759892551e-05, "loss": 0.0027, "step": 30760 }, { "grad_norm": 0.057891786098480225, "learning_rate": 5.2013932439481216e-05, "loss": 0.0022, "step": 30770 }, { "grad_norm": 0.08572710305452347, "learning_rate": 5.198639666825743e-05, "loss": 0.0025, "step": 30780 }, { "grad_norm": 0.06970208883285522, "learning_rate": 5.195886029361877e-05, "loss": 0.0024, "step": 30790 }, { "grad_norm": 0.07511906325817108, "learning_rate": 5.193132332393009e-05, "loss": 0.0024, "step": 30800 }, { "grad_norm": 0.05618830397725105, "learning_rate": 5.1903785767556376e-05, "loss": 0.0023, "step": 30810 }, { "grad_norm": 0.07966750115156174, "learning_rate": 5.187624763286282e-05, "loss": 0.0027, "step": 30820 }, { "grad_norm": 0.08229315280914307, "learning_rate": 5.184870892821475e-05, "loss": 0.0031, "step": 30830 }, { "grad_norm": 0.0936862900853157, "learning_rate": 5.182116966197773e-05, "loss": 0.002, "step": 30840 }, { "grad_norm": 0.07443857938051224, "learning_rate": 5.1793629842517466e-05, "loss": 0.0024, "step": 30850 }, { "grad_norm": 0.0620778389275074, "learning_rate": 5.17660894781998e-05, "loss": 0.0027, "step": 30860 }, { "grad_norm": 0.06017431244254112, "learning_rate": 5.173854857739079e-05, "loss": 0.0024, "step": 30870 }, { "grad_norm": 0.09338808059692383, "learning_rate": 5.171100714845661e-05, "loss": 0.0026, "step": 30880 }, { "grad_norm": 0.08167087286710739, "learning_rate": 5.1683465199763646e-05, "loss": 0.0028, "step": 30890 }, { "grad_norm": 0.09827274084091187, "learning_rate": 5.16559227396784e-05, "loss": 0.0039, "step": 30900 }, { "grad_norm": 0.09869759529829025, "learning_rate": 5.1628379776567556e-05, "loss": 0.0022, "step": 30910 }, { "grad_norm": 0.09551709145307541, "learning_rate": 5.160083631879792e-05, "loss": 0.0023, "step": 30920 }, { "grad_norm": 0.07517164200544357, "learning_rate": 5.1573292374736484e-05, "loss": 0.0024, "step": 30930 }, { "grad_norm": 0.08322843164205551, "learning_rate": 5.1545747952750356e-05, "loss": 0.0034, "step": 30940 }, { "grad_norm": 0.1232864260673523, "learning_rate": 5.151820306120682e-05, "loss": 0.0034, "step": 30950 }, { "grad_norm": 0.10737745463848114, "learning_rate": 5.149065770847328e-05, "loss": 0.0039, "step": 30960 }, { "grad_norm": 0.09256232529878616, "learning_rate": 5.1463111902917297e-05, "loss": 0.0022, "step": 30970 }, { "grad_norm": 0.07752591371536255, "learning_rate": 5.143556565290654e-05, "loss": 0.0024, "step": 30980 }, { "grad_norm": 0.09781847149133682, "learning_rate": 5.140801896680882e-05, "loss": 0.0029, "step": 30990 }, { "grad_norm": 0.07505448162555695, "learning_rate": 5.1380471852992144e-05, "loss": 0.0028, "step": 31000 }, { "grad_norm": 0.08654028177261353, "learning_rate": 5.135292431982457e-05, "loss": 0.003, "step": 31010 }, { "grad_norm": 0.07882335782051086, "learning_rate": 5.1325376375674294e-05, "loss": 0.0027, "step": 31020 }, { "grad_norm": 0.07468730956315994, "learning_rate": 5.129782802890968e-05, "loss": 0.0027, "step": 31030 }, { "grad_norm": 0.08653939515352249, "learning_rate": 5.127027928789916e-05, "loss": 0.0023, "step": 31040 }, { "grad_norm": 0.08226797729730606, "learning_rate": 5.124273016101135e-05, "loss": 0.0031, "step": 31050 }, { "grad_norm": 0.09013613313436508, "learning_rate": 5.121518065661492e-05, "loss": 0.0037, "step": 31060 }, { "grad_norm": 0.08441472798585892, "learning_rate": 5.11876307830787e-05, "loss": 0.0022, "step": 31070 }, { "grad_norm": 0.07503450661897659, "learning_rate": 5.1160080548771596e-05, "loss": 0.0035, "step": 31080 }, { "grad_norm": 0.07577574998140335, "learning_rate": 5.1132529962062656e-05, "loss": 0.0029, "step": 31090 }, { "grad_norm": 0.054710324853658676, "learning_rate": 5.110497903132101e-05, "loss": 0.0031, "step": 31100 }, { "grad_norm": 0.10193612426519394, "learning_rate": 5.107742776491592e-05, "loss": 0.0032, "step": 31110 }, { "grad_norm": 0.0923376977443695, "learning_rate": 5.104987617121673e-05, "loss": 0.0036, "step": 31120 }, { "grad_norm": 0.1209535002708435, "learning_rate": 5.102232425859287e-05, "loss": 0.0028, "step": 31130 }, { "grad_norm": 0.09639696031808853, "learning_rate": 5.09947720354139e-05, "loss": 0.0026, "step": 31140 }, { "grad_norm": 0.06550570577383041, "learning_rate": 5.096721951004942e-05, "loss": 0.0028, "step": 31150 }, { "grad_norm": 0.05699020251631737, "learning_rate": 5.0939666690869227e-05, "loss": 0.0028, "step": 31160 }, { "grad_norm": 0.080254465341568, "learning_rate": 5.0912113586243096e-05, "loss": 0.0029, "step": 31170 }, { "grad_norm": 0.06614501774311066, "learning_rate": 5.0884560204540935e-05, "loss": 0.0031, "step": 31180 }, { "grad_norm": 0.06584352254867554, "learning_rate": 5.0857006554132736e-05, "loss": 0.0035, "step": 31190 }, { "grad_norm": 0.09728734195232391, "learning_rate": 5.0829452643388575e-05, "loss": 0.0028, "step": 31200 }, { "grad_norm": 0.08199625462293625, "learning_rate": 5.08018984806786e-05, "loss": 0.0025, "step": 31210 }, { "grad_norm": 0.09308978170156479, "learning_rate": 5.0774344074373036e-05, "loss": 0.0031, "step": 31220 }, { "grad_norm": 0.07943467050790787, "learning_rate": 5.07467894328422e-05, "loss": 0.0024, "step": 31230 }, { "grad_norm": 0.05651777237653732, "learning_rate": 5.0719234564456454e-05, "loss": 0.0022, "step": 31240 }, { "grad_norm": 0.08130941540002823, "learning_rate": 5.0691679477586216e-05, "loss": 0.0029, "step": 31250 }, { "grad_norm": 0.07875695079565048, "learning_rate": 5.0664124180602035e-05, "loss": 0.0022, "step": 31260 }, { "grad_norm": 0.07295268774032593, "learning_rate": 5.063656868187447e-05, "loss": 0.003, "step": 31270 }, { "grad_norm": 0.09463793784379959, "learning_rate": 5.060901298977413e-05, "loss": 0.0023, "step": 31280 }, { "grad_norm": 0.08465894311666489, "learning_rate": 5.0581457112671725e-05, "loss": 0.0028, "step": 31290 }, { "grad_norm": 0.048614222556352615, "learning_rate": 5.0553901058938016e-05, "loss": 0.0022, "step": 31300 }, { "grad_norm": 0.08412332832813263, "learning_rate": 5.052634483694377e-05, "loss": 0.0028, "step": 31310 }, { "grad_norm": 0.09169578552246094, "learning_rate": 5.049878845505988e-05, "loss": 0.0025, "step": 31320 }, { "grad_norm": 0.0953650027513504, "learning_rate": 5.047123192165721e-05, "loss": 0.0035, "step": 31330 }, { "grad_norm": 0.09991307556629181, "learning_rate": 5.0443675245106735e-05, "loss": 0.003, "step": 31340 }, { "grad_norm": 0.09703557938337326, "learning_rate": 5.0416118433779426e-05, "loss": 0.0025, "step": 31350 }, { "grad_norm": 0.10080339014530182, "learning_rate": 5.038856149604633e-05, "loss": 0.0023, "step": 31360 }, { "grad_norm": 0.08782301843166351, "learning_rate": 5.03610044402785e-05, "loss": 0.0033, "step": 31370 }, { "grad_norm": 0.07432329654693604, "learning_rate": 5.033344727484707e-05, "loss": 0.0022, "step": 31380 }, { "grad_norm": 0.10692891478538513, "learning_rate": 5.030589000812315e-05, "loss": 0.003, "step": 31390 }, { "grad_norm": 0.12543459236621857, "learning_rate": 5.027833264847793e-05, "loss": 0.0027, "step": 31400 }, { "grad_norm": 0.13539305329322815, "learning_rate": 5.025077520428258e-05, "loss": 0.0049, "step": 31410 }, { "grad_norm": 0.13554969429969788, "learning_rate": 5.022321768390837e-05, "loss": 0.0038, "step": 31420 }, { "grad_norm": 0.06828651577234268, "learning_rate": 5.0195660095726516e-05, "loss": 0.0038, "step": 31430 }, { "grad_norm": 0.10358162224292755, "learning_rate": 5.016810244810829e-05, "loss": 0.0028, "step": 31440 }, { "grad_norm": 0.09273110330104828, "learning_rate": 5.0140544749424976e-05, "loss": 0.003, "step": 31450 }, { "grad_norm": 0.06077136844396591, "learning_rate": 5.0112987008047874e-05, "loss": 0.0023, "step": 31460 }, { "grad_norm": 0.06566287577152252, "learning_rate": 5.008542923234831e-05, "loss": 0.0026, "step": 31470 }, { "grad_norm": 0.06862899661064148, "learning_rate": 5.00578714306976e-05, "loss": 0.0027, "step": 31480 }, { "grad_norm": 0.08679269254207611, "learning_rate": 5.0030313611467084e-05, "loss": 0.0023, "step": 31490 }, { "grad_norm": 0.0549398809671402, "learning_rate": 5.0002755783028074e-05, "loss": 0.0027, "step": 31500 }, { "grad_norm": 0.08452866971492767, "learning_rate": 4.997519795375194e-05, "loss": 0.0024, "step": 31510 }, { "grad_norm": 0.10818266868591309, "learning_rate": 4.9947640132010016e-05, "loss": 0.0025, "step": 31520 }, { "grad_norm": 0.07081107050180435, "learning_rate": 4.9920082326173625e-05, "loss": 0.0028, "step": 31530 }, { "grad_norm": 0.08881483227014542, "learning_rate": 4.9892524544614114e-05, "loss": 0.0024, "step": 31540 }, { "grad_norm": 0.07846834510564804, "learning_rate": 4.986496679570283e-05, "loss": 0.0034, "step": 31550 }, { "grad_norm": 0.08090400695800781, "learning_rate": 4.983740908781105e-05, "loss": 0.0023, "step": 31560 }, { "grad_norm": 0.12420470267534256, "learning_rate": 4.9809851429310116e-05, "loss": 0.0027, "step": 31570 }, { "grad_norm": 0.07341904938220978, "learning_rate": 4.9782293828571275e-05, "loss": 0.0025, "step": 31580 }, { "grad_norm": 0.05471440777182579, "learning_rate": 4.9754736293965846e-05, "loss": 0.0029, "step": 31590 }, { "grad_norm": 0.06010778993368149, "learning_rate": 4.972717883386502e-05, "loss": 0.0022, "step": 31600 }, { "grad_norm": 0.08243165910243988, "learning_rate": 4.9699621456640075e-05, "loss": 0.0027, "step": 31610 }, { "grad_norm": 0.07731520384550095, "learning_rate": 4.9672064170662214e-05, "loss": 0.0031, "step": 31620 }, { "grad_norm": 0.08430695533752441, "learning_rate": 4.9644506984302583e-05, "loss": 0.0031, "step": 31630 }, { "grad_norm": 0.08092054724693298, "learning_rate": 4.9616949905932356e-05, "loss": 0.0029, "step": 31640 }, { "grad_norm": 0.07246308773756027, "learning_rate": 4.9589392943922615e-05, "loss": 0.003, "step": 31650 }, { "grad_norm": 0.09581148624420166, "learning_rate": 4.956183610664447e-05, "loss": 0.0025, "step": 31660 }, { "grad_norm": 0.08693739026784897, "learning_rate": 4.9534279402468945e-05, "loss": 0.0028, "step": 31670 }, { "grad_norm": 0.0990511104464531, "learning_rate": 4.9506722839767036e-05, "loss": 0.0027, "step": 31680 }, { "grad_norm": 0.12031392753124237, "learning_rate": 4.947916642690972e-05, "loss": 0.0026, "step": 31690 }, { "grad_norm": 0.08810683339834213, "learning_rate": 4.9451610172267874e-05, "loss": 0.0023, "step": 31700 }, { "grad_norm": 0.06820355355739594, "learning_rate": 4.9424054084212376e-05, "loss": 0.0028, "step": 31710 }, { "grad_norm": 0.084250308573246, "learning_rate": 4.939649817111407e-05, "loss": 0.003, "step": 31720 }, { "grad_norm": 0.07910428941249847, "learning_rate": 4.936894244134365e-05, "loss": 0.0027, "step": 31730 }, { "grad_norm": 0.056100938469171524, "learning_rate": 4.9341386903271886e-05, "loss": 0.0037, "step": 31740 }, { "grad_norm": 0.07636798918247223, "learning_rate": 4.931383156526936e-05, "loss": 0.0029, "step": 31750 }, { "grad_norm": 0.06449862569570541, "learning_rate": 4.92862764357067e-05, "loss": 0.0022, "step": 31760 }, { "grad_norm": 0.06939928233623505, "learning_rate": 4.925872152295443e-05, "loss": 0.0025, "step": 31770 }, { "grad_norm": 0.06441429257392883, "learning_rate": 4.923116683538296e-05, "loss": 0.0031, "step": 31780 }, { "grad_norm": 0.10986407846212387, "learning_rate": 4.920361238136273e-05, "loss": 0.0034, "step": 31790 }, { "grad_norm": 0.062386803328990936, "learning_rate": 4.9176058169264014e-05, "loss": 0.0024, "step": 31800 }, { "grad_norm": 0.0777665302157402, "learning_rate": 4.9148504207457074e-05, "loss": 0.0028, "step": 31810 }, { "grad_norm": 0.09602034837007523, "learning_rate": 4.912095050431208e-05, "loss": 0.0026, "step": 31820 }, { "grad_norm": 0.07796424627304077, "learning_rate": 4.909339706819911e-05, "loss": 0.0035, "step": 31830 }, { "grad_norm": 0.09993423521518707, "learning_rate": 4.906584390748819e-05, "loss": 0.0034, "step": 31840 }, { "grad_norm": 0.09696169942617416, "learning_rate": 4.9038291030549195e-05, "loss": 0.0031, "step": 31850 }, { "grad_norm": 0.09505210816860199, "learning_rate": 4.9010738445751995e-05, "loss": 0.003, "step": 31860 }, { "grad_norm": 0.08534402400255203, "learning_rate": 4.8983186161466364e-05, "loss": 0.0027, "step": 31870 }, { "grad_norm": 0.08583473414182663, "learning_rate": 4.89556341860619e-05, "loss": 0.0029, "step": 31880 }, { "grad_norm": 0.07561373710632324, "learning_rate": 4.892808252790822e-05, "loss": 0.0036, "step": 31890 }, { "grad_norm": 0.09834806621074677, "learning_rate": 4.890053119537475e-05, "loss": 0.0029, "step": 31900 }, { "grad_norm": 0.05809527263045311, "learning_rate": 4.887298019683087e-05, "loss": 0.0031, "step": 31910 }, { "grad_norm": 0.06032400578260422, "learning_rate": 4.884542954064587e-05, "loss": 0.0022, "step": 31920 }, { "grad_norm": 0.05102883279323578, "learning_rate": 4.881787923518887e-05, "loss": 0.0021, "step": 31930 }, { "grad_norm": 0.06798514723777771, "learning_rate": 4.879032928882896e-05, "loss": 0.0025, "step": 31940 }, { "grad_norm": 0.08978193998336792, "learning_rate": 4.876277970993505e-05, "loss": 0.0037, "step": 31950 }, { "grad_norm": 0.06396175920963287, "learning_rate": 4.873523050687602e-05, "loss": 0.0024, "step": 31960 }, { "grad_norm": 0.06599079817533493, "learning_rate": 4.870768168802056e-05, "loss": 0.0028, "step": 31970 }, { "grad_norm": 0.06001109629869461, "learning_rate": 4.868013326173728e-05, "loss": 0.0027, "step": 31980 }, { "grad_norm": 0.07871224731206894, "learning_rate": 4.865258523639468e-05, "loss": 0.0033, "step": 31990 }, { "grad_norm": 0.06325352191925049, "learning_rate": 4.862503762036109e-05, "loss": 0.0021, "step": 32000 }, { "grad_norm": 0.08184175193309784, "learning_rate": 4.859749042200478e-05, "loss": 0.0029, "step": 32010 }, { "grad_norm": 0.0892401859164238, "learning_rate": 4.856994364969384e-05, "loss": 0.0032, "step": 32020 }, { "grad_norm": 0.08688018471002579, "learning_rate": 4.854239731179625e-05, "loss": 0.0034, "step": 32030 }, { "grad_norm": 0.07337816804647446, "learning_rate": 4.85148514166799e-05, "loss": 0.0028, "step": 32040 }, { "grad_norm": 0.07485555112361908, "learning_rate": 4.8487305972712456e-05, "loss": 0.0024, "step": 32050 }, { "grad_norm": 0.05444464087486267, "learning_rate": 4.8459760988261526e-05, "loss": 0.002, "step": 32060 }, { "grad_norm": 0.061447445303201675, "learning_rate": 4.843221647169453e-05, "loss": 0.0021, "step": 32070 }, { "grad_norm": 0.10222586989402771, "learning_rate": 4.840467243137878e-05, "loss": 0.0026, "step": 32080 }, { "grad_norm": 0.08486342430114746, "learning_rate": 4.837712887568143e-05, "loss": 0.0025, "step": 32090 }, { "grad_norm": 0.09780439734458923, "learning_rate": 4.8349585812969464e-05, "loss": 0.0028, "step": 32100 }, { "grad_norm": 0.0788474753499031, "learning_rate": 4.8322043251609775e-05, "loss": 0.0037, "step": 32110 }, { "grad_norm": 0.09090208262205124, "learning_rate": 4.8294501199969015e-05, "loss": 0.0028, "step": 32120 }, { "grad_norm": 0.09932246804237366, "learning_rate": 4.826695966641376e-05, "loss": 0.003, "step": 32130 }, { "grad_norm": 0.06691031157970428, "learning_rate": 4.823941865931043e-05, "loss": 0.0025, "step": 32140 }, { "grad_norm": 0.07339594513177872, "learning_rate": 4.82118781870252e-05, "loss": 0.0024, "step": 32150 }, { "grad_norm": 0.07908441871404648, "learning_rate": 4.8184338257924185e-05, "loss": 0.0034, "step": 32160 }, { "grad_norm": 0.0711250975728035, "learning_rate": 4.815679888037324e-05, "loss": 0.0023, "step": 32170 }, { "grad_norm": 0.09313181787729263, "learning_rate": 4.8129260062738135e-05, "loss": 0.0024, "step": 32180 }, { "grad_norm": 0.058421194553375244, "learning_rate": 4.810172181338445e-05, "loss": 0.0023, "step": 32190 }, { "grad_norm": 0.06808267533779144, "learning_rate": 4.807418414067753e-05, "loss": 0.0022, "step": 32200 }, { "grad_norm": 0.08250769972801208, "learning_rate": 4.804664705298264e-05, "loss": 0.0025, "step": 32210 }, { "grad_norm": 0.07869566977024078, "learning_rate": 4.80191105586648e-05, "loss": 0.0029, "step": 32220 }, { "grad_norm": 0.09876225143671036, "learning_rate": 4.799157466608886e-05, "loss": 0.003, "step": 32230 }, { "grad_norm": 0.0937039703130722, "learning_rate": 4.796403938361951e-05, "loss": 0.0025, "step": 32240 }, { "grad_norm": 0.07838811725378036, "learning_rate": 4.793650471962123e-05, "loss": 0.0023, "step": 32250 }, { "grad_norm": 0.08076564222574234, "learning_rate": 4.790897068245835e-05, "loss": 0.0023, "step": 32260 }, { "grad_norm": 0.05443708971142769, "learning_rate": 4.7881437280494954e-05, "loss": 0.0024, "step": 32270 }, { "grad_norm": 0.0810818001627922, "learning_rate": 4.7853904522094965e-05, "loss": 0.0029, "step": 32280 }, { "grad_norm": 0.07654839754104614, "learning_rate": 4.782637241562215e-05, "loss": 0.0028, "step": 32290 }, { "grad_norm": 0.08439671993255615, "learning_rate": 4.779884096943997e-05, "loss": 0.0027, "step": 32300 }, { "grad_norm": 0.09267181903123856, "learning_rate": 4.777131019191182e-05, "loss": 0.0029, "step": 32310 }, { "grad_norm": 0.08732953667640686, "learning_rate": 4.774378009140076e-05, "loss": 0.0028, "step": 32320 }, { "grad_norm": 0.07609785348176956, "learning_rate": 4.7716250676269735e-05, "loss": 0.0024, "step": 32330 }, { "grad_norm": 0.06783746927976608, "learning_rate": 4.7688721954881485e-05, "loss": 0.0028, "step": 32340 }, { "grad_norm": 0.06090264767408371, "learning_rate": 4.7661193935598446e-05, "loss": 0.003, "step": 32350 }, { "grad_norm": 0.06304286420345306, "learning_rate": 4.763366662678296e-05, "loss": 0.0023, "step": 32360 }, { "grad_norm": 0.06548028439283371, "learning_rate": 4.7606140036797064e-05, "loss": 0.0024, "step": 32370 }, { "grad_norm": 0.05838870629668236, "learning_rate": 4.7578614174002614e-05, "loss": 0.0027, "step": 32380 }, { "grad_norm": 0.07967612147331238, "learning_rate": 4.755108904676125e-05, "loss": 0.0026, "step": 32390 }, { "grad_norm": 0.08086542785167694, "learning_rate": 4.752356466343436e-05, "loss": 0.0027, "step": 32400 }, { "grad_norm": 0.07288575917482376, "learning_rate": 4.7496041032383174e-05, "loss": 0.0025, "step": 32410 }, { "grad_norm": 0.11589808017015457, "learning_rate": 4.746851816196858e-05, "loss": 0.0051, "step": 32420 }, { "grad_norm": 0.11354339867830276, "learning_rate": 4.744099606055135e-05, "loss": 0.0033, "step": 32430 }, { "grad_norm": 0.08519214391708374, "learning_rate": 4.741347473649193e-05, "loss": 0.0029, "step": 32440 }, { "grad_norm": 0.05513518303632736, "learning_rate": 4.738595419815058e-05, "loss": 0.0022, "step": 32450 }, { "grad_norm": 0.0711691603064537, "learning_rate": 4.7358434453887365e-05, "loss": 0.0022, "step": 32460 }, { "grad_norm": 0.06670578569173813, "learning_rate": 4.7330915512061976e-05, "loss": 0.0024, "step": 32470 }, { "grad_norm": 0.059633560478687286, "learning_rate": 4.730339738103402e-05, "loss": 0.002, "step": 32480 }, { "grad_norm": 0.06359156966209412, "learning_rate": 4.727588006916271e-05, "loss": 0.0024, "step": 32490 }, { "grad_norm": 0.06301690638065338, "learning_rate": 4.724836358480711e-05, "loss": 0.0029, "step": 32500 }, { "grad_norm": 0.07343751937150955, "learning_rate": 4.722084793632601e-05, "loss": 0.0027, "step": 32510 }, { "grad_norm": 0.06946815550327301, "learning_rate": 4.719333313207792e-05, "loss": 0.0031, "step": 32520 }, { "grad_norm": 0.09064319729804993, "learning_rate": 4.716581918042114e-05, "loss": 0.0026, "step": 32530 }, { "grad_norm": 0.08299359679222107, "learning_rate": 4.7138306089713636e-05, "loss": 0.0026, "step": 32540 }, { "grad_norm": 0.05961161106824875, "learning_rate": 4.7110793868313183e-05, "loss": 0.0025, "step": 32550 }, { "grad_norm": 0.05811651796102524, "learning_rate": 4.708328252457729e-05, "loss": 0.0039, "step": 32560 }, { "grad_norm": 0.07919701188802719, "learning_rate": 4.7055772066863135e-05, "loss": 0.0037, "step": 32570 }, { "grad_norm": 0.09189342707395554, "learning_rate": 4.702826250352771e-05, "loss": 0.003, "step": 32580 }, { "grad_norm": 0.06992459297180176, "learning_rate": 4.7000753842927653e-05, "loss": 0.0021, "step": 32590 }, { "grad_norm": 0.08999006450176239, "learning_rate": 4.6973246093419384e-05, "loss": 0.0022, "step": 32600 }, { "grad_norm": 0.125523179769516, "learning_rate": 4.694573926335906e-05, "loss": 0.0033, "step": 32610 }, { "grad_norm": 0.09655769169330597, "learning_rate": 4.6918233361102476e-05, "loss": 0.0026, "step": 32620 }, { "grad_norm": 0.08742671459913254, "learning_rate": 4.689072839500525e-05, "loss": 0.0028, "step": 32630 }, { "grad_norm": 0.08610501885414124, "learning_rate": 4.6863224373422635e-05, "loss": 0.0023, "step": 32640 }, { "grad_norm": 0.06344447284936905, "learning_rate": 4.683572130470962e-05, "loss": 0.0027, "step": 32650 }, { "grad_norm": 0.06099351495504379, "learning_rate": 4.680821919722094e-05, "loss": 0.0019, "step": 32660 }, { "grad_norm": 0.05495491251349449, "learning_rate": 4.6780718059310975e-05, "loss": 0.0025, "step": 32670 }, { "grad_norm": 0.07834881544113159, "learning_rate": 4.675321789933389e-05, "loss": 0.0027, "step": 32680 }, { "grad_norm": 0.05845635384321213, "learning_rate": 4.6725718725643464e-05, "loss": 0.0024, "step": 32690 }, { "grad_norm": 0.0779728889465332, "learning_rate": 4.669822054659323e-05, "loss": 0.0023, "step": 32700 }, { "grad_norm": 0.07844603061676025, "learning_rate": 4.667072337053644e-05, "loss": 0.0026, "step": 32710 }, { "grad_norm": 0.07565571367740631, "learning_rate": 4.6643227205825965e-05, "loss": 0.0021, "step": 32720 }, { "grad_norm": 0.07034628093242645, "learning_rate": 4.6615732060814454e-05, "loss": 0.0027, "step": 32730 }, { "grad_norm": 0.048919420689344406, "learning_rate": 4.658823794385417e-05, "loss": 0.0025, "step": 32740 }, { "grad_norm": 0.07137156277894974, "learning_rate": 4.6560744863297115e-05, "loss": 0.0024, "step": 32750 }, { "grad_norm": 0.09662455320358276, "learning_rate": 4.653325282749498e-05, "loss": 0.0031, "step": 32760 }, { "grad_norm": 0.0968838483095169, "learning_rate": 4.6505761844799075e-05, "loss": 0.0026, "step": 32770 }, { "grad_norm": 0.07244012504816055, "learning_rate": 4.647827192356048e-05, "loss": 0.0029, "step": 32780 }, { "grad_norm": 0.08419179916381836, "learning_rate": 4.645078307212989e-05, "loss": 0.0037, "step": 32790 }, { "grad_norm": 0.08268938958644867, "learning_rate": 4.642329529885768e-05, "loss": 0.0022, "step": 32800 }, { "grad_norm": 0.07802700251340866, "learning_rate": 4.639580861209393e-05, "loss": 0.0028, "step": 32810 }, { "grad_norm": 0.066217802464962, "learning_rate": 4.636832302018835e-05, "loss": 0.0023, "step": 32820 }, { "grad_norm": 0.06807414442300797, "learning_rate": 4.6340838531490365e-05, "loss": 0.0021, "step": 32830 }, { "grad_norm": 0.08128061145544052, "learning_rate": 4.6313355154349e-05, "loss": 0.003, "step": 32840 }, { "grad_norm": 0.09717847406864166, "learning_rate": 4.6285872897113025e-05, "loss": 0.0033, "step": 32850 }, { "grad_norm": 0.08732513338327408, "learning_rate": 4.625839176813077e-05, "loss": 0.0025, "step": 32860 }, { "grad_norm": 0.0926898866891861, "learning_rate": 4.623091177575031e-05, "loss": 0.0032, "step": 32870 }, { "grad_norm": 0.07506583631038666, "learning_rate": 4.620343292831936e-05, "loss": 0.0022, "step": 32880 }, { "grad_norm": 0.07877253741025925, "learning_rate": 4.6175955234185206e-05, "loss": 0.0031, "step": 32890 }, { "grad_norm": 0.06761536002159119, "learning_rate": 4.614847870169492e-05, "loss": 0.003, "step": 32900 }, { "grad_norm": 0.09991651028394699, "learning_rate": 4.612100333919509e-05, "loss": 0.0031, "step": 32910 }, { "grad_norm": 0.0732988715171814, "learning_rate": 4.609352915503202e-05, "loss": 0.0023, "step": 32920 }, { "grad_norm": 0.07774285227060318, "learning_rate": 4.606605615755166e-05, "loss": 0.0038, "step": 32930 }, { "grad_norm": 0.0705673024058342, "learning_rate": 4.6038584355099576e-05, "loss": 0.0025, "step": 32940 }, { "grad_norm": 0.07217329740524292, "learning_rate": 4.6011113756020964e-05, "loss": 0.0026, "step": 32950 }, { "grad_norm": 0.06914167106151581, "learning_rate": 4.598364436866066e-05, "loss": 0.003, "step": 32960 }, { "grad_norm": 0.0633111223578453, "learning_rate": 4.595617620136316e-05, "loss": 0.0031, "step": 32970 }, { "grad_norm": 0.11257070302963257, "learning_rate": 4.592870926247257e-05, "loss": 0.002, "step": 32980 }, { "grad_norm": 0.06269139051437378, "learning_rate": 4.5901243560332594e-05, "loss": 0.0028, "step": 32990 }, { "grad_norm": 0.08077561855316162, "learning_rate": 4.587377910328662e-05, "loss": 0.0023, "step": 33000 }, { "grad_norm": 0.06738442927598953, "learning_rate": 4.5846315899677586e-05, "loss": 0.0021, "step": 33010 }, { "grad_norm": 0.06453394144773483, "learning_rate": 4.5818853957848114e-05, "loss": 0.002, "step": 33020 }, { "grad_norm": 0.06205286458134651, "learning_rate": 4.579139328614043e-05, "loss": 0.0028, "step": 33030 }, { "grad_norm": 0.05585794895887375, "learning_rate": 4.576393389289633e-05, "loss": 0.0023, "step": 33040 }, { "grad_norm": 0.07041610032320023, "learning_rate": 4.573647578645728e-05, "loss": 0.0026, "step": 33050 }, { "grad_norm": 0.07192961126565933, "learning_rate": 4.57090189751643e-05, "loss": 0.0025, "step": 33060 }, { "grad_norm": 0.06649421155452728, "learning_rate": 4.568156346735806e-05, "loss": 0.0018, "step": 33070 }, { "grad_norm": 0.09024173766374588, "learning_rate": 4.565410927137882e-05, "loss": 0.0024, "step": 33080 }, { "grad_norm": 0.06656955182552338, "learning_rate": 4.562665639556644e-05, "loss": 0.0024, "step": 33090 }, { "grad_norm": 0.10277613997459412, "learning_rate": 4.559920484826037e-05, "loss": 0.0022, "step": 33100 }, { "grad_norm": 0.05236871913075447, "learning_rate": 4.5571754637799665e-05, "loss": 0.0019, "step": 33110 }, { "grad_norm": 0.04722512513399124, "learning_rate": 4.554430577252298e-05, "loss": 0.0017, "step": 33120 }, { "grad_norm": 0.07318031787872314, "learning_rate": 4.551685826076858e-05, "loss": 0.0027, "step": 33130 }, { "grad_norm": 0.0729396641254425, "learning_rate": 4.5489412110874246e-05, "loss": 0.002, "step": 33140 }, { "grad_norm": 0.08127417415380478, "learning_rate": 4.5461967331177444e-05, "loss": 0.0029, "step": 33150 }, { "grad_norm": 0.08366259187459946, "learning_rate": 4.5434523930015115e-05, "loss": 0.0023, "step": 33160 }, { "grad_norm": 0.08105586469173431, "learning_rate": 4.540708191572388e-05, "loss": 0.0035, "step": 33170 }, { "grad_norm": 0.07771191745996475, "learning_rate": 4.537964129663991e-05, "loss": 0.0027, "step": 33180 }, { "grad_norm": 0.06300336867570877, "learning_rate": 4.535220208109889e-05, "loss": 0.0027, "step": 33190 }, { "grad_norm": 0.06170536205172539, "learning_rate": 4.5324764277436194e-05, "loss": 0.0025, "step": 33200 }, { "grad_norm": 0.0873880535364151, "learning_rate": 4.529732789398664e-05, "loss": 0.0032, "step": 33210 }, { "grad_norm": 0.0665176510810852, "learning_rate": 4.526989293908472e-05, "loss": 0.0029, "step": 33220 }, { "grad_norm": 0.07085788995027542, "learning_rate": 4.524245942106442e-05, "loss": 0.0027, "step": 33230 }, { "grad_norm": 0.08004496991634369, "learning_rate": 4.5215027348259345e-05, "loss": 0.0029, "step": 33240 }, { "grad_norm": 0.08347302675247192, "learning_rate": 4.5187596729002616e-05, "loss": 0.0029, "step": 33250 }, { "grad_norm": 0.06968244910240173, "learning_rate": 4.516016757162693e-05, "loss": 0.0031, "step": 33260 }, { "grad_norm": 0.06591594219207764, "learning_rate": 4.513273988446457e-05, "loss": 0.0027, "step": 33270 }, { "grad_norm": 0.08084295690059662, "learning_rate": 4.5105313675847296e-05, "loss": 0.0019, "step": 33280 }, { "grad_norm": 0.10984225571155548, "learning_rate": 4.5077888954106495e-05, "loss": 0.0033, "step": 33290 }, { "grad_norm": 0.09011803567409515, "learning_rate": 4.505046572757309e-05, "loss": 0.0024, "step": 33300 }, { "grad_norm": 0.07456384599208832, "learning_rate": 4.502304400457749e-05, "loss": 0.0028, "step": 33310 }, { "grad_norm": 0.09689709544181824, "learning_rate": 4.499562379344973e-05, "loss": 0.0026, "step": 33320 }, { "grad_norm": 0.08705080300569534, "learning_rate": 4.4968205102519306e-05, "loss": 0.0029, "step": 33330 }, { "grad_norm": 0.06460078805685043, "learning_rate": 4.494078794011532e-05, "loss": 0.0029, "step": 33340 }, { "grad_norm": 0.056954916566610336, "learning_rate": 4.491337231456639e-05, "loss": 0.003, "step": 33350 }, { "grad_norm": 0.07603782415390015, "learning_rate": 4.4885958234200634e-05, "loss": 0.0026, "step": 33360 }, { "grad_norm": 0.09687424451112747, "learning_rate": 4.485854570734575e-05, "loss": 0.0024, "step": 33370 }, { "grad_norm": 0.07662646472454071, "learning_rate": 4.483113474232891e-05, "loss": 0.0021, "step": 33380 }, { "grad_norm": 0.05543316900730133, "learning_rate": 4.480372534747688e-05, "loss": 0.0022, "step": 33390 }, { "grad_norm": 0.08627893775701523, "learning_rate": 4.477631753111588e-05, "loss": 0.0019, "step": 33400 }, { "grad_norm": 0.08138694614171982, "learning_rate": 4.4748911301571686e-05, "loss": 0.0026, "step": 33410 }, { "grad_norm": 0.09694758802652359, "learning_rate": 4.472150666716961e-05, "loss": 0.0025, "step": 33420 }, { "grad_norm": 0.10956079512834549, "learning_rate": 4.469410363623442e-05, "loss": 0.0028, "step": 33430 }, { "grad_norm": 0.08696750551462173, "learning_rate": 4.466670221709044e-05, "loss": 0.0032, "step": 33440 }, { "grad_norm": 0.09273887425661087, "learning_rate": 4.463930241806154e-05, "loss": 0.0026, "step": 33450 }, { "grad_norm": 0.0663202553987503, "learning_rate": 4.4611904247471006e-05, "loss": 0.0032, "step": 33460 }, { "grad_norm": 0.07384441047906876, "learning_rate": 4.458450771364171e-05, "loss": 0.0023, "step": 33470 }, { "grad_norm": 0.07483766973018646, "learning_rate": 4.4557112824895965e-05, "loss": 0.0027, "step": 33480 }, { "grad_norm": 0.07913064956665039, "learning_rate": 4.452971958955563e-05, "loss": 0.0026, "step": 33490 }, { "grad_norm": 0.0815245732665062, "learning_rate": 4.450232801594208e-05, "loss": 0.0023, "step": 33500 }, { "grad_norm": 0.08081204444169998, "learning_rate": 4.447493811237609e-05, "loss": 0.0026, "step": 33510 }, { "grad_norm": 0.06852748245000839, "learning_rate": 4.444754988717804e-05, "loss": 0.0021, "step": 33520 }, { "grad_norm": 0.08287210017442703, "learning_rate": 4.442016334866771e-05, "loss": 0.0027, "step": 33530 }, { "grad_norm": 0.07231669872999191, "learning_rate": 4.4392778505164445e-05, "loss": 0.0021, "step": 33540 }, { "grad_norm": 0.0809064731001854, "learning_rate": 4.436539536498702e-05, "loss": 0.0024, "step": 33550 }, { "grad_norm": 0.09406714141368866, "learning_rate": 4.433801393645369e-05, "loss": 0.0027, "step": 33560 }, { "grad_norm": 0.07484982907772064, "learning_rate": 4.431063422788226e-05, "loss": 0.0024, "step": 33570 }, { "grad_norm": 0.08398978412151337, "learning_rate": 4.428325624758991e-05, "loss": 0.0031, "step": 33580 }, { "grad_norm": 0.07539892196655273, "learning_rate": 4.4255880003893366e-05, "loss": 0.002, "step": 33590 }, { "grad_norm": 0.04835943132638931, "learning_rate": 4.422850550510884e-05, "loss": 0.0021, "step": 33600 }, { "grad_norm": 0.09534858912229538, "learning_rate": 4.4201132759551934e-05, "loss": 0.0022, "step": 33610 }, { "grad_norm": 0.07952828705310822, "learning_rate": 4.4173761775537804e-05, "loss": 0.0035, "step": 33620 }, { "grad_norm": 0.08314218372106552, "learning_rate": 4.414639256138099e-05, "loss": 0.0038, "step": 33630 }, { "grad_norm": 0.0827043205499649, "learning_rate": 4.411902512539557e-05, "loss": 0.0027, "step": 33640 }, { "grad_norm": 0.07995285838842392, "learning_rate": 4.4091659475895044e-05, "loss": 0.0034, "step": 33650 }, { "grad_norm": 0.05662418529391289, "learning_rate": 4.406429562119235e-05, "loss": 0.0025, "step": 33660 }, { "grad_norm": 0.10635367780923843, "learning_rate": 4.4036933569599945e-05, "loss": 0.0024, "step": 33670 }, { "grad_norm": 0.07576724141836166, "learning_rate": 4.400957332942965e-05, "loss": 0.0024, "step": 33680 }, { "grad_norm": 0.05565813183784485, "learning_rate": 4.3982214908992844e-05, "loss": 0.0022, "step": 33690 }, { "grad_norm": 0.089972123503685, "learning_rate": 4.3954858316600235e-05, "loss": 0.004, "step": 33700 }, { "grad_norm": 0.06460265070199966, "learning_rate": 4.392750356056205e-05, "loss": 0.0022, "step": 33710 }, { "grad_norm": 0.08806309849023819, "learning_rate": 4.390015064918798e-05, "loss": 0.0026, "step": 33720 }, { "grad_norm": 0.07420411705970764, "learning_rate": 4.387279959078705e-05, "loss": 0.0023, "step": 33730 }, { "grad_norm": 0.08544747531414032, "learning_rate": 4.384545039366786e-05, "loss": 0.0024, "step": 33740 }, { "grad_norm": 0.07234702259302139, "learning_rate": 4.381810306613831e-05, "loss": 0.0029, "step": 33750 }, { "grad_norm": 0.07397959381341934, "learning_rate": 4.3790757616505826e-05, "loss": 0.0028, "step": 33760 }, { "grad_norm": 0.05886856094002724, "learning_rate": 4.376341405307725e-05, "loss": 0.0021, "step": 33770 }, { "grad_norm": 0.09355885535478592, "learning_rate": 4.37360723841588e-05, "loss": 0.0036, "step": 33780 }, { "grad_norm": 0.0887896791100502, "learning_rate": 4.370873261805619e-05, "loss": 0.003, "step": 33790 }, { "grad_norm": 0.08405144512653351, "learning_rate": 4.368139476307449e-05, "loss": 0.0028, "step": 33800 }, { "grad_norm": 0.070445716381073, "learning_rate": 4.365405882751822e-05, "loss": 0.0023, "step": 33810 }, { "grad_norm": 0.07149772346019745, "learning_rate": 4.3626724819691326e-05, "loss": 0.0022, "step": 33820 }, { "grad_norm": 0.06694146245718002, "learning_rate": 4.359939274789715e-05, "loss": 0.0024, "step": 33830 }, { "grad_norm": 0.0658978819847107, "learning_rate": 4.357206262043848e-05, "loss": 0.0022, "step": 33840 }, { "grad_norm": 0.06365200132131577, "learning_rate": 4.354473444561745e-05, "loss": 0.0021, "step": 33850 }, { "grad_norm": 0.05931221321225166, "learning_rate": 4.3517408231735644e-05, "loss": 0.0024, "step": 33860 }, { "grad_norm": 0.06930296868085861, "learning_rate": 4.3490083987094086e-05, "loss": 0.0029, "step": 33870 }, { "grad_norm": 0.06852799654006958, "learning_rate": 4.34627617199931e-05, "loss": 0.0024, "step": 33880 }, { "grad_norm": 0.06715954095125198, "learning_rate": 4.3435441438732526e-05, "loss": 0.0033, "step": 33890 }, { "grad_norm": 0.07430458813905716, "learning_rate": 4.340812315161149e-05, "loss": 0.0028, "step": 33900 }, { "grad_norm": 0.1067035123705864, "learning_rate": 4.338080686692859e-05, "loss": 0.0022, "step": 33910 }, { "grad_norm": 0.0688645988702774, "learning_rate": 4.3353492592981816e-05, "loss": 0.0021, "step": 33920 }, { "grad_norm": 0.049295488744974136, "learning_rate": 4.3326180338068485e-05, "loss": 0.0021, "step": 33930 }, { "grad_norm": 0.05828654021024704, "learning_rate": 4.3298870110485356e-05, "loss": 0.0028, "step": 33940 }, { "grad_norm": 0.06051325425505638, "learning_rate": 4.3271561918528567e-05, "loss": 0.0024, "step": 33950 }, { "grad_norm": 0.08173513412475586, "learning_rate": 4.324425577049359e-05, "loss": 0.0027, "step": 33960 }, { "grad_norm": 0.06340742111206055, "learning_rate": 4.321695167467535e-05, "loss": 0.0023, "step": 33970 }, { "grad_norm": 0.06788966804742813, "learning_rate": 4.3189649639368093e-05, "loss": 0.0027, "step": 33980 }, { "grad_norm": 0.0593738928437233, "learning_rate": 4.316234967286547e-05, "loss": 0.0022, "step": 33990 }, { "grad_norm": 0.0647362470626831, "learning_rate": 4.313505178346046e-05, "loss": 0.0024, "step": 34000 }, { "grad_norm": 0.08138606697320938, "learning_rate": 4.3107755979445465e-05, "loss": 0.0023, "step": 34010 }, { "grad_norm": 0.06285237520933151, "learning_rate": 4.308046226911224e-05, "loss": 0.0021, "step": 34020 }, { "grad_norm": 0.07268663495779037, "learning_rate": 4.305317066075185e-05, "loss": 0.0028, "step": 34030 }, { "grad_norm": 0.06030505150556564, "learning_rate": 4.302588116265482e-05, "loss": 0.0028, "step": 34040 }, { "grad_norm": 0.0987108051776886, "learning_rate": 4.299859378311094e-05, "loss": 0.0032, "step": 34050 }, { "grad_norm": 0.09006846696138382, "learning_rate": 4.2971308530409424e-05, "loss": 0.0025, "step": 34060 }, { "grad_norm": 0.06337819248437881, "learning_rate": 4.2944025412838765e-05, "loss": 0.0019, "step": 34070 }, { "grad_norm": 0.06362830847501755, "learning_rate": 4.291674443868689e-05, "loss": 0.0022, "step": 34080 }, { "grad_norm": 0.05494798347353935, "learning_rate": 4.288946561624104e-05, "loss": 0.0029, "step": 34090 }, { "grad_norm": 0.08813592046499252, "learning_rate": 4.2862188953787794e-05, "loss": 0.0025, "step": 34100 }, { "grad_norm": 0.08822809159755707, "learning_rate": 4.283491445961308e-05, "loss": 0.0024, "step": 34110 }, { "grad_norm": 0.06664298474788666, "learning_rate": 4.2807642142002155e-05, "loss": 0.0023, "step": 34120 }, { "grad_norm": 0.07732900232076645, "learning_rate": 4.278037200923966e-05, "loss": 0.0023, "step": 34130 }, { "grad_norm": 0.08103061467409134, "learning_rate": 4.275310406960953e-05, "loss": 0.0028, "step": 34140 }, { "grad_norm": 0.08771002292633057, "learning_rate": 4.272583833139502e-05, "loss": 0.0025, "step": 34150 }, { "grad_norm": 0.06518695503473282, "learning_rate": 4.2698574802878794e-05, "loss": 0.0026, "step": 34160 }, { "grad_norm": 0.06840400397777557, "learning_rate": 4.2671313492342734e-05, "loss": 0.0023, "step": 34170 }, { "grad_norm": 0.062018606811761856, "learning_rate": 4.264405440806813e-05, "loss": 0.0023, "step": 34180 }, { "grad_norm": 0.06391141563653946, "learning_rate": 4.26167975583356e-05, "loss": 0.0024, "step": 34190 }, { "grad_norm": 0.08460880815982819, "learning_rate": 4.2589542951425e-05, "loss": 0.0037, "step": 34200 }, { "grad_norm": 0.06516902148723602, "learning_rate": 4.2562290595615615e-05, "loss": 0.0027, "step": 34210 }, { "grad_norm": 0.0636555552482605, "learning_rate": 4.2535040499185946e-05, "loss": 0.0024, "step": 34220 }, { "grad_norm": 0.07541003823280334, "learning_rate": 4.250779267041387e-05, "loss": 0.0031, "step": 34230 }, { "grad_norm": 0.08605224639177322, "learning_rate": 4.248054711757657e-05, "loss": 0.0025, "step": 34240 }, { "grad_norm": 0.063773013651371, "learning_rate": 4.245330384895052e-05, "loss": 0.0022, "step": 34250 }, { "grad_norm": 0.05512603372335434, "learning_rate": 4.242606287281151e-05, "loss": 0.0023, "step": 34260 }, { "grad_norm": 0.07260847836732864, "learning_rate": 4.2398824197434595e-05, "loss": 0.0023, "step": 34270 }, { "grad_norm": 0.07994195818901062, "learning_rate": 4.23715878310942e-05, "loss": 0.0029, "step": 34280 }, { "grad_norm": 0.08137277513742447, "learning_rate": 4.234435378206402e-05, "loss": 0.0024, "step": 34290 }, { "grad_norm": 0.09992581605911255, "learning_rate": 4.2317122058617006e-05, "loss": 0.0032, "step": 34300 }, { "grad_norm": 0.12630103528499603, "learning_rate": 4.2289892669025485e-05, "loss": 0.0038, "step": 34310 }, { "grad_norm": 0.10511402785778046, "learning_rate": 4.226266562156097e-05, "loss": 0.0031, "step": 34320 }, { "grad_norm": 0.061655133962631226, "learning_rate": 4.223544092449435e-05, "loss": 0.0033, "step": 34330 }, { "grad_norm": 0.07982585579156876, "learning_rate": 4.2208218586095784e-05, "loss": 0.0032, "step": 34340 }, { "grad_norm": 0.10219348967075348, "learning_rate": 4.218099861463466e-05, "loss": 0.004, "step": 34350 }, { "grad_norm": 0.09060598164796829, "learning_rate": 4.215378101837972e-05, "loss": 0.0028, "step": 34360 }, { "grad_norm": 0.06738271564245224, "learning_rate": 4.2126565805598937e-05, "loss": 0.0024, "step": 34370 }, { "grad_norm": 0.08102189004421234, "learning_rate": 4.209935298455957e-05, "loss": 0.0026, "step": 34380 }, { "grad_norm": 0.052749693393707275, "learning_rate": 4.207214256352817e-05, "loss": 0.0025, "step": 34390 }, { "grad_norm": 0.06498201191425323, "learning_rate": 4.2044934550770524e-05, "loss": 0.0027, "step": 34400 }, { "grad_norm": 0.07545073330402374, "learning_rate": 4.201772895455174e-05, "loss": 0.0032, "step": 34410 }, { "grad_norm": 0.07432184368371964, "learning_rate": 4.199052578313613e-05, "loss": 0.0021, "step": 34420 }, { "grad_norm": 0.06402278691530228, "learning_rate": 4.1963325044787294e-05, "loss": 0.0022, "step": 34430 }, { "grad_norm": 0.060953546315431595, "learning_rate": 4.193612674776814e-05, "loss": 0.0024, "step": 34440 }, { "grad_norm": 0.05981498584151268, "learning_rate": 4.1908930900340745e-05, "loss": 0.0021, "step": 34450 }, { "grad_norm": 0.06290742754936218, "learning_rate": 4.1881737510766536e-05, "loss": 0.0017, "step": 34460 }, { "grad_norm": 0.06008613854646683, "learning_rate": 4.185454658730609e-05, "loss": 0.0019, "step": 34470 }, { "grad_norm": 0.0835651382803917, "learning_rate": 4.1827358138219355e-05, "loss": 0.0023, "step": 34480 }, { "grad_norm": 0.06661683320999146, "learning_rate": 4.1800172171765404e-05, "loss": 0.002, "step": 34490 }, { "grad_norm": 0.093763068318367, "learning_rate": 4.177298869620264e-05, "loss": 0.0032, "step": 34500 }, { "grad_norm": 0.07077278941869736, "learning_rate": 4.1745807719788705e-05, "loss": 0.0024, "step": 34510 }, { "grad_norm": 0.06996789574623108, "learning_rate": 4.1718629250780445e-05, "loss": 0.0021, "step": 34520 }, { "grad_norm": 0.06562542915344238, "learning_rate": 4.1691453297433956e-05, "loss": 0.0026, "step": 34530 }, { "grad_norm": 0.08188717067241669, "learning_rate": 4.166427986800457e-05, "loss": 0.003, "step": 34540 }, { "grad_norm": 0.07992450147867203, "learning_rate": 4.163710897074688e-05, "loss": 0.0029, "step": 34550 }, { "grad_norm": 0.10375244915485382, "learning_rate": 4.1609940613914686e-05, "loss": 0.0027, "step": 34560 }, { "grad_norm": 0.05900192633271217, "learning_rate": 4.1582774805760996e-05, "loss": 0.0023, "step": 34570 }, { "grad_norm": 0.061490029096603394, "learning_rate": 4.155561155453809e-05, "loss": 0.0024, "step": 34580 }, { "grad_norm": 0.09691547602415085, "learning_rate": 4.15284508684974e-05, "loss": 0.0035, "step": 34590 }, { "grad_norm": 0.0801868885755539, "learning_rate": 4.1501292755889675e-05, "loss": 0.0031, "step": 34600 }, { "grad_norm": 0.08794953674077988, "learning_rate": 4.1474137224964833e-05, "loss": 0.0023, "step": 34610 }, { "grad_norm": 0.054003238677978516, "learning_rate": 4.144698428397197e-05, "loss": 0.0021, "step": 34620 }, { "grad_norm": 0.07193373888731003, "learning_rate": 4.1419833941159466e-05, "loss": 0.002, "step": 34630 }, { "grad_norm": 0.051196325570344925, "learning_rate": 4.1392686204774846e-05, "loss": 0.0021, "step": 34640 }, { "grad_norm": 0.059483785182237625, "learning_rate": 4.13655410830649e-05, "loss": 0.0023, "step": 34650 }, { "grad_norm": 0.05956968292593956, "learning_rate": 4.1338398584275594e-05, "loss": 0.0019, "step": 34660 }, { "grad_norm": 0.08204597234725952, "learning_rate": 4.1311258716652104e-05, "loss": 0.0019, "step": 34670 }, { "grad_norm": 0.06664404273033142, "learning_rate": 4.128412148843881e-05, "loss": 0.002, "step": 34680 }, { "grad_norm": 0.07155760377645493, "learning_rate": 4.125698690787926e-05, "loss": 0.0017, "step": 34690 }, { "grad_norm": 0.09348452091217041, "learning_rate": 4.1229854983216245e-05, "loss": 0.0022, "step": 34700 }, { "grad_norm": 0.050406698137521744, "learning_rate": 4.120272572269175e-05, "loss": 0.0021, "step": 34710 }, { "grad_norm": 0.08637027442455292, "learning_rate": 4.117559913454687e-05, "loss": 0.0022, "step": 34720 }, { "grad_norm": 0.06546526402235031, "learning_rate": 4.114847522702201e-05, "loss": 0.0022, "step": 34730 }, { "grad_norm": 0.07670972496271133, "learning_rate": 4.112135400835664e-05, "loss": 0.0028, "step": 34740 }, { "grad_norm": 0.11025306582450867, "learning_rate": 4.109423548678949e-05, "loss": 0.0029, "step": 34750 }, { "grad_norm": 0.09726342558860779, "learning_rate": 4.106711967055848e-05, "loss": 0.0035, "step": 34760 }, { "grad_norm": 0.09405899792909622, "learning_rate": 4.1040006567900636e-05, "loss": 0.0022, "step": 34770 }, { "grad_norm": 0.04987333342432976, "learning_rate": 4.101289618705224e-05, "loss": 0.0025, "step": 34780 }, { "grad_norm": 0.05490439012646675, "learning_rate": 4.0985788536248675e-05, "loss": 0.0023, "step": 34790 }, { "grad_norm": 0.07477057725191116, "learning_rate": 4.095868362372454e-05, "loss": 0.002, "step": 34800 }, { "grad_norm": 0.06867654621601105, "learning_rate": 4.0931581457713614e-05, "loss": 0.003, "step": 34810 }, { "grad_norm": 0.0706588551402092, "learning_rate": 4.09044820464488e-05, "loss": 0.0025, "step": 34820 }, { "grad_norm": 0.0726253092288971, "learning_rate": 4.087738539816219e-05, "loss": 0.0025, "step": 34830 }, { "grad_norm": 0.056490764021873474, "learning_rate": 4.085029152108501e-05, "loss": 0.002, "step": 34840 }, { "grad_norm": 0.07868839800357819, "learning_rate": 4.0823200423447714e-05, "loss": 0.0023, "step": 34850 }, { "grad_norm": 0.06220174580812454, "learning_rate": 4.079611211347981e-05, "loss": 0.0024, "step": 34860 }, { "grad_norm": 0.09144733101129532, "learning_rate": 4.076902659941002e-05, "loss": 0.002, "step": 34870 }, { "grad_norm": 0.05415239930152893, "learning_rate": 4.074194388946624e-05, "loss": 0.0024, "step": 34880 }, { "grad_norm": 0.060105666518211365, "learning_rate": 4.071486399187545e-05, "loss": 0.002, "step": 34890 }, { "grad_norm": 0.053925298154354095, "learning_rate": 4.0687786914863836e-05, "loss": 0.0019, "step": 34900 }, { "grad_norm": 0.07073581963777542, "learning_rate": 4.0660712666656666e-05, "loss": 0.002, "step": 34910 }, { "grad_norm": 0.05845822021365166, "learning_rate": 4.0633641255478394e-05, "loss": 0.002, "step": 34920 }, { "grad_norm": 0.06754063814878464, "learning_rate": 4.0606572689552624e-05, "loss": 0.0019, "step": 34930 }, { "grad_norm": 0.07935008406639099, "learning_rate": 4.0579506977102036e-05, "loss": 0.0022, "step": 34940 }, { "grad_norm": 0.09223033487796783, "learning_rate": 4.055244412634849e-05, "loss": 0.0023, "step": 34950 }, { "grad_norm": 0.07628549635410309, "learning_rate": 4.052538414551298e-05, "loss": 0.002, "step": 34960 }, { "grad_norm": 0.07738637179136276, "learning_rate": 4.0498327042815596e-05, "loss": 0.003, "step": 34970 }, { "grad_norm": 0.06673309206962585, "learning_rate": 4.047127282647559e-05, "loss": 0.0018, "step": 34980 }, { "grad_norm": 0.04919028282165527, "learning_rate": 4.04442215047113e-05, "loss": 0.0019, "step": 34990 }, { "grad_norm": 0.05301021784543991, "learning_rate": 4.041717308574023e-05, "loss": 0.0024, "step": 35000 }, { "grad_norm": 0.06540387123823166, "learning_rate": 4.039012757777893e-05, "loss": 0.0021, "step": 35010 }, { "grad_norm": 0.09738972783088684, "learning_rate": 4.036308498904314e-05, "loss": 0.0027, "step": 35020 }, { "grad_norm": 0.07595895230770111, "learning_rate": 4.033604532774771e-05, "loss": 0.0029, "step": 35030 }, { "grad_norm": 0.059609510004520416, "learning_rate": 4.030900860210652e-05, "loss": 0.0021, "step": 35040 }, { "grad_norm": 0.07378243654966354, "learning_rate": 4.028197482033266e-05, "loss": 0.0028, "step": 35050 }, { "grad_norm": 0.05769282579421997, "learning_rate": 4.0254943990638246e-05, "loss": 0.002, "step": 35060 }, { "grad_norm": 0.05165422707796097, "learning_rate": 4.022791612123454e-05, "loss": 0.002, "step": 35070 }, { "grad_norm": 0.04545943811535835, "learning_rate": 4.020089122033192e-05, "loss": 0.0019, "step": 35080 }, { "grad_norm": 0.04226585477590561, "learning_rate": 4.01738692961398e-05, "loss": 0.0019, "step": 35090 }, { "grad_norm": 0.10095933824777603, "learning_rate": 4.014685035686675e-05, "loss": 0.0027, "step": 35100 }, { "grad_norm": 0.0934128686785698, "learning_rate": 4.011983441072039e-05, "loss": 0.0022, "step": 35110 }, { "grad_norm": 0.0834289938211441, "learning_rate": 4.0092821465907485e-05, "loss": 0.0024, "step": 35120 }, { "grad_norm": 0.07949189096689224, "learning_rate": 4.006581153063383e-05, "loss": 0.0016, "step": 35130 }, { "grad_norm": 0.06282517313957214, "learning_rate": 4.003880461310432e-05, "loss": 0.0019, "step": 35140 }, { "grad_norm": 0.06255823373794556, "learning_rate": 4.001180072152298e-05, "loss": 0.0019, "step": 35150 }, { "grad_norm": 0.06979767978191376, "learning_rate": 3.998479986409285e-05, "loss": 0.0026, "step": 35160 }, { "grad_norm": 0.06394638121128082, "learning_rate": 3.995780204901607e-05, "loss": 0.0015, "step": 35170 }, { "grad_norm": 0.0603211410343647, "learning_rate": 3.993080728449391e-05, "loss": 0.0018, "step": 35180 }, { "grad_norm": 0.056488826870918274, "learning_rate": 3.990381557872661e-05, "loss": 0.0016, "step": 35190 }, { "grad_norm": 0.052427008748054504, "learning_rate": 3.987682693991359e-05, "loss": 0.0024, "step": 35200 }, { "grad_norm": 0.09199134260416031, "learning_rate": 3.9849841376253226e-05, "loss": 0.0028, "step": 35210 }, { "grad_norm": 0.07752335071563721, "learning_rate": 3.982285889594306e-05, "loss": 0.0022, "step": 35220 }, { "grad_norm": 0.07382897287607193, "learning_rate": 3.9795879507179665e-05, "loss": 0.0027, "step": 35230 }, { "grad_norm": 0.07151331007480621, "learning_rate": 3.9768903218158634e-05, "loss": 0.0026, "step": 35240 }, { "grad_norm": 0.05211174115538597, "learning_rate": 3.974193003707468e-05, "loss": 0.0022, "step": 35250 }, { "grad_norm": 0.09118609130382538, "learning_rate": 3.971495997212152e-05, "loss": 0.0026, "step": 35260 }, { "grad_norm": 0.05597253516316414, "learning_rate": 3.9687993031491985e-05, "loss": 0.0018, "step": 35270 }, { "grad_norm": 0.09431444108486176, "learning_rate": 3.966102922337787e-05, "loss": 0.0033, "step": 35280 }, { "grad_norm": 0.06584782153367996, "learning_rate": 3.963406855597009e-05, "loss": 0.0021, "step": 35290 }, { "grad_norm": 0.08626531809568405, "learning_rate": 3.960711103745861e-05, "loss": 0.0024, "step": 35300 }, { "grad_norm": 0.07448272407054901, "learning_rate": 3.958015667603237e-05, "loss": 0.002, "step": 35310 }, { "grad_norm": 0.06006735935807228, "learning_rate": 3.955320547987943e-05, "loss": 0.0025, "step": 35320 }, { "grad_norm": 0.056775279343128204, "learning_rate": 3.952625745718681e-05, "loss": 0.0035, "step": 35330 }, { "grad_norm": 0.07748935371637344, "learning_rate": 3.949931261614064e-05, "loss": 0.0024, "step": 35340 }, { "grad_norm": 0.058814775198698044, "learning_rate": 3.947237096492605e-05, "loss": 0.0022, "step": 35350 }, { "grad_norm": 0.06637496501207352, "learning_rate": 3.944543251172719e-05, "loss": 0.0023, "step": 35360 }, { "grad_norm": 0.07813270390033722, "learning_rate": 3.941849726472725e-05, "loss": 0.0037, "step": 35370 }, { "grad_norm": 0.05541187524795532, "learning_rate": 3.939156523210846e-05, "loss": 0.0031, "step": 35380 }, { "grad_norm": 0.09495952725410461, "learning_rate": 3.9364636422052046e-05, "loss": 0.0021, "step": 35390 }, { "grad_norm": 0.05201386287808418, "learning_rate": 3.933771084273828e-05, "loss": 0.0035, "step": 35400 }, { "grad_norm": 0.05987868458032608, "learning_rate": 3.931078850234643e-05, "loss": 0.0024, "step": 35410 }, { "grad_norm": 0.07046771794557571, "learning_rate": 3.928386940905483e-05, "loss": 0.0025, "step": 35420 }, { "grad_norm": 0.06984760612249374, "learning_rate": 3.925695357104073e-05, "loss": 0.0023, "step": 35430 }, { "grad_norm": 0.05476153641939163, "learning_rate": 3.923004099648049e-05, "loss": 0.0023, "step": 35440 }, { "grad_norm": 0.08199213445186615, "learning_rate": 3.920313169354944e-05, "loss": 0.0025, "step": 35450 }, { "grad_norm": 0.0639122799038887, "learning_rate": 3.9176225670421897e-05, "loss": 0.0023, "step": 35460 }, { "grad_norm": 0.04787814989686012, "learning_rate": 3.9149322935271224e-05, "loss": 0.0025, "step": 35470 }, { "grad_norm": 0.05697156488895416, "learning_rate": 3.9122423496269725e-05, "loss": 0.0015, "step": 35480 }, { "grad_norm": 0.08143609017133713, "learning_rate": 3.909552736158877e-05, "loss": 0.0022, "step": 35490 }, { "grad_norm": 0.07709486782550812, "learning_rate": 3.90686345393987e-05, "loss": 0.0019, "step": 35500 }, { "grad_norm": 0.07026131451129913, "learning_rate": 3.9041745037868816e-05, "loss": 0.0022, "step": 35510 }, { "grad_norm": 0.06514963507652283, "learning_rate": 3.9014858865167465e-05, "loss": 0.0029, "step": 35520 }, { "grad_norm": 0.0812133327126503, "learning_rate": 3.8987976029461935e-05, "loss": 0.0023, "step": 35530 }, { "grad_norm": 0.08476238697767258, "learning_rate": 3.896109653891853e-05, "loss": 0.0025, "step": 35540 }, { "grad_norm": 0.10699917376041412, "learning_rate": 3.893422040170254e-05, "loss": 0.002, "step": 35550 }, { "grad_norm": 0.0814480260014534, "learning_rate": 3.8907347625978207e-05, "loss": 0.0025, "step": 35560 }, { "grad_norm": 0.04892750456929207, "learning_rate": 3.88804782199088e-05, "loss": 0.002, "step": 35570 }, { "grad_norm": 0.06400253623723984, "learning_rate": 3.8853612191656495e-05, "loss": 0.0025, "step": 35580 }, { "grad_norm": 0.08315558731555939, "learning_rate": 3.88267495493825e-05, "loss": 0.002, "step": 35590 }, { "grad_norm": 0.05730792135000229, "learning_rate": 3.8799890301247004e-05, "loss": 0.0022, "step": 35600 }, { "grad_norm": 0.0833795890212059, "learning_rate": 3.8773034455409096e-05, "loss": 0.0025, "step": 35610 }, { "grad_norm": 0.05571412295103073, "learning_rate": 3.8746182020026904e-05, "loss": 0.002, "step": 35620 }, { "grad_norm": 0.051887139678001404, "learning_rate": 3.871933300325745e-05, "loss": 0.0017, "step": 35630 }, { "grad_norm": 0.057920876890420914, "learning_rate": 3.869248741325679e-05, "loss": 0.0021, "step": 35640 }, { "grad_norm": 0.08193389326334, "learning_rate": 3.866564525817992e-05, "loss": 0.0021, "step": 35650 }, { "grad_norm": 0.06384164094924927, "learning_rate": 3.8638806546180725e-05, "loss": 0.0022, "step": 35660 }, { "grad_norm": 0.08746197819709778, "learning_rate": 3.861197128541213e-05, "loss": 0.0023, "step": 35670 }, { "grad_norm": 0.06763607263565063, "learning_rate": 3.858513948402599e-05, "loss": 0.0023, "step": 35680 }, { "grad_norm": 0.09107321500778198, "learning_rate": 3.8558311150173077e-05, "loss": 0.0029, "step": 35690 }, { "grad_norm": 0.08430837839841843, "learning_rate": 3.853148629200312e-05, "loss": 0.0022, "step": 35700 }, { "grad_norm": 0.08391877263784409, "learning_rate": 3.850466491766482e-05, "loss": 0.0035, "step": 35710 }, { "grad_norm": 0.07422863692045212, "learning_rate": 3.847784703530583e-05, "loss": 0.003, "step": 35720 }, { "grad_norm": 0.05517173558473587, "learning_rate": 3.845103265307266e-05, "loss": 0.0019, "step": 35730 }, { "grad_norm": 0.05348503217101097, "learning_rate": 3.842422177911086e-05, "loss": 0.0026, "step": 35740 }, { "grad_norm": 0.08868199586868286, "learning_rate": 3.8397414421564826e-05, "loss": 0.0019, "step": 35750 }, { "grad_norm": 0.07154972106218338, "learning_rate": 3.8370610588577935e-05, "loss": 0.002, "step": 35760 }, { "grad_norm": 0.07107392698526382, "learning_rate": 3.834381028829251e-05, "loss": 0.0017, "step": 35770 }, { "grad_norm": 0.08878771960735321, "learning_rate": 3.8317013528849745e-05, "loss": 0.002, "step": 35780 }, { "grad_norm": 0.05786818265914917, "learning_rate": 3.8290220318389815e-05, "loss": 0.0017, "step": 35790 }, { "grad_norm": 0.07146910578012466, "learning_rate": 3.8263430665051746e-05, "loss": 0.0024, "step": 35800 }, { "grad_norm": 0.09711597114801407, "learning_rate": 3.8236644576973554e-05, "loss": 0.0026, "step": 35810 }, { "grad_norm": 0.08291884511709213, "learning_rate": 3.820986206229217e-05, "loss": 0.0025, "step": 35820 }, { "grad_norm": 0.05103524401783943, "learning_rate": 3.8183083129143384e-05, "loss": 0.0021, "step": 35830 }, { "grad_norm": 0.0663147047162056, "learning_rate": 3.815630778566193e-05, "loss": 0.0023, "step": 35840 }, { "grad_norm": 0.06484244763851166, "learning_rate": 3.812953603998145e-05, "loss": 0.0029, "step": 35850 }, { "grad_norm": 0.07744389772415161, "learning_rate": 3.8102767900234504e-05, "loss": 0.0021, "step": 35860 }, { "grad_norm": 0.07470757514238358, "learning_rate": 3.807600337455256e-05, "loss": 0.0028, "step": 35870 }, { "grad_norm": 0.059002961963415146, "learning_rate": 3.804924247106593e-05, "loss": 0.0018, "step": 35880 }, { "grad_norm": 0.09037116169929504, "learning_rate": 3.8022485197903925e-05, "loss": 0.002, "step": 35890 }, { "grad_norm": 0.09376098960638046, "learning_rate": 3.799573156319464e-05, "loss": 0.0025, "step": 35900 }, { "grad_norm": 0.08323952555656433, "learning_rate": 3.796898157506515e-05, "loss": 0.0026, "step": 35910 }, { "grad_norm": 0.07568338513374329, "learning_rate": 3.794223524164143e-05, "loss": 0.003, "step": 35920 }, { "grad_norm": 0.06196058169007301, "learning_rate": 3.7915492571048245e-05, "loss": 0.0022, "step": 35930 }, { "grad_norm": 0.08174154162406921, "learning_rate": 3.788875357140937e-05, "loss": 0.002, "step": 35940 }, { "grad_norm": 0.061103370040655136, "learning_rate": 3.786201825084736e-05, "loss": 0.0026, "step": 35950 }, { "grad_norm": 0.06874973326921463, "learning_rate": 3.783528661748372e-05, "loss": 0.0018, "step": 35960 }, { "grad_norm": 0.06279958784580231, "learning_rate": 3.780855867943882e-05, "loss": 0.003, "step": 35970 }, { "grad_norm": 0.07277543097734451, "learning_rate": 3.778183444483189e-05, "loss": 0.0038, "step": 35980 }, { "grad_norm": 0.06593667715787888, "learning_rate": 3.775511392178108e-05, "loss": 0.0022, "step": 35990 }, { "grad_norm": 0.08089979737997055, "learning_rate": 3.772839711840332e-05, "loss": 0.0039, "step": 36000 }, { "grad_norm": 0.08186695724725723, "learning_rate": 3.7701684042814515e-05, "loss": 0.0024, "step": 36010 }, { "grad_norm": 0.07628335058689117, "learning_rate": 3.76749747031294e-05, "loss": 0.0024, "step": 36020 }, { "grad_norm": 0.08225563913583755, "learning_rate": 3.764826910746152e-05, "loss": 0.0025, "step": 36030 }, { "grad_norm": 0.053256817162036896, "learning_rate": 3.762156726392338e-05, "loss": 0.0021, "step": 36040 }, { "grad_norm": 0.0808863639831543, "learning_rate": 3.759486918062625e-05, "loss": 0.0023, "step": 36050 }, { "grad_norm": 0.046536341309547424, "learning_rate": 3.756817486568033e-05, "loss": 0.0023, "step": 36060 }, { "grad_norm": 0.05493054538965225, "learning_rate": 3.7541484327194654e-05, "loss": 0.0022, "step": 36070 }, { "grad_norm": 0.08904695510864258, "learning_rate": 3.751479757327707e-05, "loss": 0.0032, "step": 36080 }, { "grad_norm": 0.04485413804650307, "learning_rate": 3.7488114612034345e-05, "loss": 0.0023, "step": 36090 }, { "grad_norm": 0.08635213971138, "learning_rate": 3.7461435451572044e-05, "loss": 0.0021, "step": 36100 }, { "grad_norm": 0.07310276478528976, "learning_rate": 3.743476009999459e-05, "loss": 0.0021, "step": 36110 }, { "grad_norm": 0.05920466408133507, "learning_rate": 3.7408088565405245e-05, "loss": 0.0017, "step": 36120 }, { "grad_norm": 0.04903051629662514, "learning_rate": 3.738142085590612e-05, "loss": 0.0017, "step": 36130 }, { "grad_norm": 0.04837939143180847, "learning_rate": 3.7354756979598194e-05, "loss": 0.0017, "step": 36140 }, { "grad_norm": 0.07139816135168076, "learning_rate": 3.7328096944581187e-05, "loss": 0.002, "step": 36150 }, { "grad_norm": 0.0641867071390152, "learning_rate": 3.730144075895377e-05, "loss": 0.002, "step": 36160 }, { "grad_norm": 0.07990435510873795, "learning_rate": 3.727478843081335e-05, "loss": 0.0023, "step": 36170 }, { "grad_norm": 0.06317600607872009, "learning_rate": 3.72481399682562e-05, "loss": 0.0022, "step": 36180 }, { "grad_norm": 0.04583510756492615, "learning_rate": 3.722149537937747e-05, "loss": 0.0023, "step": 36190 }, { "grad_norm": 0.06703637540340424, "learning_rate": 3.7194854672271015e-05, "loss": 0.0024, "step": 36200 }, { "grad_norm": 0.07702125608921051, "learning_rate": 3.7168217855029644e-05, "loss": 0.0023, "step": 36210 }, { "grad_norm": 0.07986728847026825, "learning_rate": 3.7141584935744856e-05, "loss": 0.0023, "step": 36220 }, { "grad_norm": 0.07100901752710342, "learning_rate": 3.7114955922507055e-05, "loss": 0.0025, "step": 36230 }, { "grad_norm": 0.06657664477825165, "learning_rate": 3.708833082340545e-05, "loss": 0.003, "step": 36240 }, { "grad_norm": 0.06103576347231865, "learning_rate": 3.7061709646528034e-05, "loss": 0.0021, "step": 36250 }, { "grad_norm": 0.059382230043411255, "learning_rate": 3.7035092399961604e-05, "loss": 0.0019, "step": 36260 }, { "grad_norm": 0.09995030611753464, "learning_rate": 3.700847909179177e-05, "loss": 0.0024, "step": 36270 }, { "grad_norm": 0.08122195303440094, "learning_rate": 3.698186973010297e-05, "loss": 0.0027, "step": 36280 }, { "grad_norm": 0.0830012708902359, "learning_rate": 3.695526432297844e-05, "loss": 0.0025, "step": 36290 }, { "grad_norm": 0.05844705179333687, "learning_rate": 3.692866287850017e-05, "loss": 0.0021, "step": 36300 }, { "grad_norm": 0.07037673890590668, "learning_rate": 3.6902065404749006e-05, "loss": 0.0021, "step": 36310 }, { "grad_norm": 0.058093681931495667, "learning_rate": 3.6875471909804516e-05, "loss": 0.0017, "step": 36320 }, { "grad_norm": 0.054949115961790085, "learning_rate": 3.6848882401745135e-05, "loss": 0.0023, "step": 36330 }, { "grad_norm": 0.057672400027513504, "learning_rate": 3.682229688864806e-05, "loss": 0.0018, "step": 36340 }, { "grad_norm": 0.05948302522301674, "learning_rate": 3.6795715378589235e-05, "loss": 0.0022, "step": 36350 }, { "grad_norm": 0.048646871000528336, "learning_rate": 3.676913787964345e-05, "loss": 0.0016, "step": 36360 }, { "grad_norm": 0.0654473528265953, "learning_rate": 3.674256439988423e-05, "loss": 0.0027, "step": 36370 }, { "grad_norm": 0.08100137114524841, "learning_rate": 3.6715994947383904e-05, "loss": 0.0023, "step": 36380 }, { "grad_norm": 0.07544617354869843, "learning_rate": 3.668942953021357e-05, "loss": 0.003, "step": 36390 }, { "grad_norm": 0.11980357021093369, "learning_rate": 3.66628681564431e-05, "loss": 0.0029, "step": 36400 }, { "grad_norm": 0.07974409312009811, "learning_rate": 3.663631083414114e-05, "loss": 0.002, "step": 36410 }, { "grad_norm": 0.07860191911458969, "learning_rate": 3.660975757137509e-05, "loss": 0.0022, "step": 36420 }, { "grad_norm": 0.04774710536003113, "learning_rate": 3.658320837621114e-05, "loss": 0.0015, "step": 36430 }, { "grad_norm": 0.06502701342105865, "learning_rate": 3.655666325671426e-05, "loss": 0.0022, "step": 36440 }, { "grad_norm": 0.0566183403134346, "learning_rate": 3.65301222209481e-05, "loss": 0.0021, "step": 36450 }, { "grad_norm": 0.0703512579202652, "learning_rate": 3.650358527697519e-05, "loss": 0.0025, "step": 36460 }, { "grad_norm": 0.04754367843270302, "learning_rate": 3.64770524328567e-05, "loss": 0.0019, "step": 36470 }, { "grad_norm": 0.08809972554445267, "learning_rate": 3.645052369665265e-05, "loss": 0.002, "step": 36480 }, { "grad_norm": 0.04871254414319992, "learning_rate": 3.6423999076421724e-05, "loss": 0.0027, "step": 36490 }, { "grad_norm": 0.0716288834810257, "learning_rate": 3.639747858022142e-05, "loss": 0.0021, "step": 36500 }, { "grad_norm": 0.07881477475166321, "learning_rate": 3.637096221610799e-05, "loss": 0.0017, "step": 36510 }, { "grad_norm": 0.07131056487560272, "learning_rate": 3.634444999213638e-05, "loss": 0.0019, "step": 36520 }, { "grad_norm": 0.07522798329591751, "learning_rate": 3.6317941916360296e-05, "loss": 0.0029, "step": 36530 }, { "grad_norm": 0.05007146671414375, "learning_rate": 3.629143799683221e-05, "loss": 0.0018, "step": 36540 }, { "grad_norm": 0.06490505486726761, "learning_rate": 3.626493824160331e-05, "loss": 0.0026, "step": 36550 }, { "grad_norm": 0.05942781642079353, "learning_rate": 3.623844265872352e-05, "loss": 0.0015, "step": 36560 }, { "grad_norm": 0.07406026124954224, "learning_rate": 3.621195125624149e-05, "loss": 0.0018, "step": 36570 }, { "grad_norm": 0.09339601546525955, "learning_rate": 3.618546404220463e-05, "loss": 0.0023, "step": 36580 }, { "grad_norm": 0.07880236208438873, "learning_rate": 3.615898102465903e-05, "loss": 0.0024, "step": 36590 }, { "grad_norm": 0.0673065334558487, "learning_rate": 3.6132502211649544e-05, "loss": 0.0019, "step": 36600 }, { "grad_norm": 0.040564656257629395, "learning_rate": 3.610602761121975e-05, "loss": 0.0017, "step": 36610 }, { "grad_norm": 0.05902021378278732, "learning_rate": 3.6079557231411897e-05, "loss": 0.002, "step": 36620 }, { "grad_norm": 0.07532542198896408, "learning_rate": 3.6053091080267035e-05, "loss": 0.0027, "step": 36630 }, { "grad_norm": 0.06428222358226776, "learning_rate": 3.602662916582483e-05, "loss": 0.003, "step": 36640 }, { "grad_norm": 0.08718755841255188, "learning_rate": 3.600017149612375e-05, "loss": 0.0021, "step": 36650 }, { "grad_norm": 0.0693429633975029, "learning_rate": 3.5973718079200935e-05, "loss": 0.0026, "step": 36660 }, { "grad_norm": 0.048848707228899, "learning_rate": 3.5947268923092216e-05, "loss": 0.002, "step": 36670 }, { "grad_norm": 0.06829449534416199, "learning_rate": 3.592082403583216e-05, "loss": 0.002, "step": 36680 }, { "grad_norm": 0.05545379966497421, "learning_rate": 3.5894383425454004e-05, "loss": 0.0016, "step": 36690 }, { "grad_norm": 0.06608984619379044, "learning_rate": 3.586794709998975e-05, "loss": 0.002, "step": 36700 }, { "grad_norm": 0.05640289559960365, "learning_rate": 3.584151506747002e-05, "loss": 0.0029, "step": 36710 }, { "grad_norm": 0.07119140774011612, "learning_rate": 3.581508733592418e-05, "loss": 0.0018, "step": 36720 }, { "grad_norm": 0.05211130529642105, "learning_rate": 3.5788663913380297e-05, "loss": 0.0021, "step": 36730 }, { "grad_norm": 0.09019307047128677, "learning_rate": 3.576224480786506e-05, "loss": 0.002, "step": 36740 }, { "grad_norm": 0.05965491011738777, "learning_rate": 3.573583002740393e-05, "loss": 0.0023, "step": 36750 }, { "grad_norm": 0.03995395451784134, "learning_rate": 3.570941958002103e-05, "loss": 0.0019, "step": 36760 }, { "grad_norm": 0.09065361320972443, "learning_rate": 3.568301347373912e-05, "loss": 0.0027, "step": 36770 }, { "grad_norm": 0.06110741198062897, "learning_rate": 3.5656611716579726e-05, "loss": 0.0017, "step": 36780 }, { "grad_norm": 0.05350223183631897, "learning_rate": 3.5630214316562946e-05, "loss": 0.0023, "step": 36790 }, { "grad_norm": 0.0514485202729702, "learning_rate": 3.560382128170766e-05, "loss": 0.0015, "step": 36800 }, { "grad_norm": 0.06641087681055069, "learning_rate": 3.5577432620031374e-05, "loss": 0.0034, "step": 36810 }, { "grad_norm": 0.07127060741186142, "learning_rate": 3.5551048339550216e-05, "loss": 0.0016, "step": 36820 }, { "grad_norm": 0.09485141932964325, "learning_rate": 3.55246684482791e-05, "loss": 0.0018, "step": 36830 }, { "grad_norm": 0.0704297125339508, "learning_rate": 3.5498292954231496e-05, "loss": 0.0026, "step": 36840 }, { "grad_norm": 0.08376938849687576, "learning_rate": 3.54719218654196e-05, "loss": 0.0019, "step": 36850 }, { "grad_norm": 0.07583750784397125, "learning_rate": 3.544555518985425e-05, "loss": 0.0025, "step": 36860 }, { "grad_norm": 0.0970420390367508, "learning_rate": 3.541919293554494e-05, "loss": 0.0026, "step": 36870 }, { "grad_norm": 0.09185571223497391, "learning_rate": 3.539283511049985e-05, "loss": 0.0021, "step": 36880 }, { "grad_norm": 0.09605301171541214, "learning_rate": 3.5366481722725755e-05, "loss": 0.0025, "step": 36890 }, { "grad_norm": 0.06488268077373505, "learning_rate": 3.534013278022816e-05, "loss": 0.0029, "step": 36900 }, { "grad_norm": 0.07824139297008514, "learning_rate": 3.531378829101113e-05, "loss": 0.002, "step": 36910 }, { "grad_norm": 0.05043232813477516, "learning_rate": 3.528744826307746e-05, "loss": 0.002, "step": 36920 }, { "grad_norm": 0.06272426247596741, "learning_rate": 3.5261112704428554e-05, "loss": 0.0019, "step": 36930 }, { "grad_norm": 0.054065048694610596, "learning_rate": 3.523478162306443e-05, "loss": 0.0022, "step": 36940 }, { "grad_norm": 0.06356417387723923, "learning_rate": 3.520845502698381e-05, "loss": 0.002, "step": 36950 }, { "grad_norm": 0.05769532173871994, "learning_rate": 3.5182132924184005e-05, "loss": 0.0022, "step": 36960 }, { "grad_norm": 0.06788720935583115, "learning_rate": 3.5155815322660966e-05, "loss": 0.0018, "step": 36970 }, { "grad_norm": 0.05701836943626404, "learning_rate": 3.512950223040931e-05, "loss": 0.0018, "step": 36980 }, { "grad_norm": 0.061899229884147644, "learning_rate": 3.5103193655422216e-05, "loss": 0.0018, "step": 36990 }, { "grad_norm": 0.05125008523464203, "learning_rate": 3.5076889605691596e-05, "loss": 0.0019, "step": 37000 }, { "grad_norm": 0.04331322759389877, "learning_rate": 3.505059008920787e-05, "loss": 0.0022, "step": 37010 }, { "grad_norm": 0.06807170063257217, "learning_rate": 3.502429511396016e-05, "loss": 0.0017, "step": 37020 }, { "grad_norm": 0.04597223550081253, "learning_rate": 3.4998004687936196e-05, "loss": 0.0023, "step": 37030 }, { "grad_norm": 0.05819643661379814, "learning_rate": 3.497171881912229e-05, "loss": 0.0028, "step": 37040 }, { "grad_norm": 0.04903372749686241, "learning_rate": 3.494543751550342e-05, "loss": 0.0016, "step": 37050 }, { "grad_norm": 0.09523677825927734, "learning_rate": 3.491916078506313e-05, "loss": 0.0029, "step": 37060 }, { "grad_norm": 0.06603211164474487, "learning_rate": 3.489288863578361e-05, "loss": 0.0023, "step": 37070 }, { "grad_norm": 0.07199791073799133, "learning_rate": 3.4866621075645646e-05, "loss": 0.0021, "step": 37080 }, { "grad_norm": 0.05356810614466667, "learning_rate": 3.4840358112628614e-05, "loss": 0.0018, "step": 37090 }, { "grad_norm": 0.07752548903226852, "learning_rate": 3.481409975471053e-05, "loss": 0.0019, "step": 37100 }, { "grad_norm": 0.09799838811159134, "learning_rate": 3.4787846009867986e-05, "loss": 0.002, "step": 37110 }, { "grad_norm": 0.081642284989357, "learning_rate": 3.476159688607615e-05, "loss": 0.0021, "step": 37120 }, { "grad_norm": 0.07300399243831635, "learning_rate": 3.4735352391308854e-05, "loss": 0.0022, "step": 37130 }, { "grad_norm": 0.08301874995231628, "learning_rate": 3.4709112533538446e-05, "loss": 0.0015, "step": 37140 }, { "grad_norm": 0.06355521082878113, "learning_rate": 3.4682877320735934e-05, "loss": 0.0016, "step": 37150 }, { "grad_norm": 0.05453703925013542, "learning_rate": 3.465664676087085e-05, "loss": 0.0021, "step": 37160 }, { "grad_norm": 0.08749836683273315, "learning_rate": 3.463042086191136e-05, "loss": 0.002, "step": 37170 }, { "grad_norm": 0.09687354415655136, "learning_rate": 3.460419963182423e-05, "loss": 0.0024, "step": 37180 }, { "grad_norm": 0.08973969519138336, "learning_rate": 3.457798307857473e-05, "loss": 0.003, "step": 37190 }, { "grad_norm": 0.06206395477056503, "learning_rate": 3.455177121012678e-05, "loss": 0.0028, "step": 37200 }, { "grad_norm": 0.07096783816814423, "learning_rate": 3.452556403444285e-05, "loss": 0.0022, "step": 37210 }, { "grad_norm": 0.07382190972566605, "learning_rate": 3.4499361559483975e-05, "loss": 0.0021, "step": 37220 }, { "grad_norm": 0.07692842930555344, "learning_rate": 3.44731637932098e-05, "loss": 0.002, "step": 37230 }, { "grad_norm": 0.07343554496765137, "learning_rate": 3.44469707435785e-05, "loss": 0.0021, "step": 37240 }, { "grad_norm": 0.07662823796272278, "learning_rate": 3.4420782418546835e-05, "loss": 0.0025, "step": 37250 }, { "grad_norm": 0.07183264195919037, "learning_rate": 3.439459882607012e-05, "loss": 0.0022, "step": 37260 }, { "grad_norm": 0.05862714350223541, "learning_rate": 3.436841997410225e-05, "loss": 0.0031, "step": 37270 }, { "grad_norm": 0.062191370874643326, "learning_rate": 3.434224587059567e-05, "loss": 0.002, "step": 37280 }, { "grad_norm": 0.07031922787427902, "learning_rate": 3.431607652350136e-05, "loss": 0.0024, "step": 37290 }, { "grad_norm": 0.08869438618421555, "learning_rate": 3.428991194076891e-05, "loss": 0.0031, "step": 37300 }, { "grad_norm": 0.054879602044820786, "learning_rate": 3.4263752130346394e-05, "loss": 0.0026, "step": 37310 }, { "grad_norm": 0.04224996268749237, "learning_rate": 3.4237597100180515e-05, "loss": 0.0021, "step": 37320 }, { "grad_norm": 0.05367691442370415, "learning_rate": 3.4211446858216427e-05, "loss": 0.0021, "step": 37330 }, { "grad_norm": 0.04903123155236244, "learning_rate": 3.4185301412397915e-05, "loss": 0.0018, "step": 37340 }, { "grad_norm": 0.06711793690919876, "learning_rate": 3.415916077066729e-05, "loss": 0.0021, "step": 37350 }, { "grad_norm": 0.06074720621109009, "learning_rate": 3.413302494096535e-05, "loss": 0.0017, "step": 37360 }, { "grad_norm": 0.059559475630521774, "learning_rate": 3.410689393123151e-05, "loss": 0.0022, "step": 37370 }, { "grad_norm": 0.05511576309800148, "learning_rate": 3.408076774940364e-05, "loss": 0.0021, "step": 37380 }, { "grad_norm": 0.04740074649453163, "learning_rate": 3.40546464034182e-05, "loss": 0.0022, "step": 37390 }, { "grad_norm": 0.05870010331273079, "learning_rate": 3.4028529901210185e-05, "loss": 0.0019, "step": 37400 }, { "grad_norm": 0.0944473072886467, "learning_rate": 3.4002418250713086e-05, "loss": 0.002, "step": 37410 }, { "grad_norm": 0.07377905398607254, "learning_rate": 3.3976311459858936e-05, "loss": 0.0023, "step": 37420 }, { "grad_norm": 0.051065247505903244, "learning_rate": 3.395020953657826e-05, "loss": 0.0016, "step": 37430 }, { "grad_norm": 0.06987753510475159, "learning_rate": 3.3924112488800165e-05, "loss": 0.002, "step": 37440 }, { "grad_norm": 0.0870235487818718, "learning_rate": 3.389802032445225e-05, "loss": 0.0027, "step": 37450 }, { "grad_norm": 0.0602412223815918, "learning_rate": 3.38719330514606e-05, "loss": 0.0021, "step": 37460 }, { "grad_norm": 0.05864214524626732, "learning_rate": 3.3845850677749866e-05, "loss": 0.0019, "step": 37470 }, { "grad_norm": 0.04279478266835213, "learning_rate": 3.3819773211243157e-05, "loss": 0.0017, "step": 37480 }, { "grad_norm": 0.059891227632761, "learning_rate": 3.379370065986213e-05, "loss": 0.0018, "step": 37490 }, { "grad_norm": 0.05254056304693222, "learning_rate": 3.3767633031526955e-05, "loss": 0.0024, "step": 37500 }, { "grad_norm": 0.07184483110904694, "learning_rate": 3.374157033415626e-05, "loss": 0.0037, "step": 37510 }, { "grad_norm": 0.07130324840545654, "learning_rate": 3.371551257566723e-05, "loss": 0.0025, "step": 37520 }, { "grad_norm": 0.09108628332614899, "learning_rate": 3.36894597639755e-05, "loss": 0.0023, "step": 37530 }, { "grad_norm": 0.07507345080375671, "learning_rate": 3.366341190699523e-05, "loss": 0.0024, "step": 37540 }, { "grad_norm": 0.05339984595775604, "learning_rate": 3.36373690126391e-05, "loss": 0.0019, "step": 37550 }, { "grad_norm": 0.05676151439547539, "learning_rate": 3.3611331088818234e-05, "loss": 0.002, "step": 37560 }, { "grad_norm": 0.07378704845905304, "learning_rate": 3.3585298143442265e-05, "loss": 0.0025, "step": 37570 }, { "grad_norm": 0.04964024946093559, "learning_rate": 3.35592701844193e-05, "loss": 0.0018, "step": 37580 }, { "grad_norm": 0.053892310708761215, "learning_rate": 3.353324721965596e-05, "loss": 0.003, "step": 37590 }, { "grad_norm": 0.08185633271932602, "learning_rate": 3.350722925705736e-05, "loss": 0.0027, "step": 37600 }, { "grad_norm": 0.07341974973678589, "learning_rate": 3.348121630452703e-05, "loss": 0.0019, "step": 37610 }, { "grad_norm": 0.07479524612426758, "learning_rate": 3.3455208369967044e-05, "loss": 0.002, "step": 37620 }, { "grad_norm": 0.08476754277944565, "learning_rate": 3.34292054612779e-05, "loss": 0.0023, "step": 37630 }, { "grad_norm": 0.06657405942678452, "learning_rate": 3.340320758635861e-05, "loss": 0.0018, "step": 37640 }, { "grad_norm": 0.0573965422809124, "learning_rate": 3.337721475310666e-05, "loss": 0.0016, "step": 37650 }, { "grad_norm": 0.061002179980278015, "learning_rate": 3.335122696941795e-05, "loss": 0.0022, "step": 37660 }, { "grad_norm": 0.05561527982354164, "learning_rate": 3.332524424318692e-05, "loss": 0.0018, "step": 37670 }, { "grad_norm": 0.06706704944372177, "learning_rate": 3.32992665823064e-05, "loss": 0.0018, "step": 37680 }, { "grad_norm": 0.0825025886297226, "learning_rate": 3.327329399466774e-05, "loss": 0.0023, "step": 37690 }, { "grad_norm": 0.05895915627479553, "learning_rate": 3.324732648816072e-05, "loss": 0.0019, "step": 37700 }, { "grad_norm": 0.05331651121377945, "learning_rate": 3.322136407067358e-05, "loss": 0.0023, "step": 37710 }, { "grad_norm": 0.0696588084101677, "learning_rate": 3.3195406750093036e-05, "loss": 0.0016, "step": 37720 }, { "grad_norm": 0.050865501165390015, "learning_rate": 3.3169454534304205e-05, "loss": 0.002, "step": 37730 }, { "grad_norm": 0.0776946172118187, "learning_rate": 3.3143507431190725e-05, "loss": 0.0024, "step": 37740 }, { "grad_norm": 0.05896120145916939, "learning_rate": 3.311756544863459e-05, "loss": 0.0019, "step": 37750 }, { "grad_norm": 0.05487974360585213, "learning_rate": 3.309162859451633e-05, "loss": 0.002, "step": 37760 }, { "grad_norm": 0.0875469520688057, "learning_rate": 3.306569687671487e-05, "loss": 0.0021, "step": 37770 }, { "grad_norm": 0.08398973941802979, "learning_rate": 3.303977030310756e-05, "loss": 0.002, "step": 37780 }, { "grad_norm": 0.06637947261333466, "learning_rate": 3.3013848881570245e-05, "loss": 0.002, "step": 37790 }, { "grad_norm": 0.045554254204034805, "learning_rate": 3.298793261997712e-05, "loss": 0.002, "step": 37800 }, { "grad_norm": 0.05406993255019188, "learning_rate": 3.2962021526200893e-05, "loss": 0.0018, "step": 37810 }, { "grad_norm": 0.06110101565718651, "learning_rate": 3.293611560811268e-05, "loss": 0.0017, "step": 37820 }, { "grad_norm": 0.07210232317447662, "learning_rate": 3.291021487358199e-05, "loss": 0.0025, "step": 37830 }, { "grad_norm": 0.0891023799777031, "learning_rate": 3.28843193304768e-05, "loss": 0.0018, "step": 37840 }, { "grad_norm": 0.07092341780662537, "learning_rate": 3.2858428986663456e-05, "loss": 0.0027, "step": 37850 }, { "grad_norm": 0.07430004328489304, "learning_rate": 3.283254385000681e-05, "loss": 0.0018, "step": 37860 }, { "grad_norm": 0.05248403549194336, "learning_rate": 3.2806663928370076e-05, "loss": 0.0027, "step": 37870 }, { "grad_norm": 0.08084608614444733, "learning_rate": 3.278078922961485e-05, "loss": 0.0027, "step": 37880 }, { "grad_norm": 0.049392297863960266, "learning_rate": 3.275491976160123e-05, "loss": 0.0013, "step": 37890 }, { "grad_norm": 0.07096367329359055, "learning_rate": 3.2729055532187645e-05, "loss": 0.0018, "step": 37900 }, { "grad_norm": 0.051599930971860886, "learning_rate": 3.270319654923097e-05, "loss": 0.0017, "step": 37910 }, { "grad_norm": 0.08123737573623657, "learning_rate": 3.2677342820586506e-05, "loss": 0.0024, "step": 37920 }, { "grad_norm": 0.07922752946615219, "learning_rate": 3.2651494354107905e-05, "loss": 0.0024, "step": 37930 }, { "grad_norm": 0.07240046560764313, "learning_rate": 3.2625651157647266e-05, "loss": 0.002, "step": 37940 }, { "grad_norm": 0.06397843360900879, "learning_rate": 3.259981323905505e-05, "loss": 0.0019, "step": 37950 }, { "grad_norm": 0.045746076852083206, "learning_rate": 3.257398060618014e-05, "loss": 0.002, "step": 37960 }, { "grad_norm": 0.1084156185388565, "learning_rate": 3.254815326686983e-05, "loss": 0.0023, "step": 37970 }, { "grad_norm": 0.040898390114307404, "learning_rate": 3.2522331228969774e-05, "loss": 0.0021, "step": 37980 }, { "grad_norm": 0.05988457053899765, "learning_rate": 3.2496514500324006e-05, "loss": 0.0013, "step": 37990 }, { "grad_norm": 0.039763227105140686, "learning_rate": 3.247070308877498e-05, "loss": 0.0016, "step": 38000 }, { "grad_norm": 0.05600820854306221, "learning_rate": 3.2444897002163515e-05, "loss": 0.0013, "step": 38010 }, { "grad_norm": 0.07451197504997253, "learning_rate": 3.241909624832885e-05, "loss": 0.0018, "step": 38020 }, { "grad_norm": 0.04933958873152733, "learning_rate": 3.239330083510852e-05, "loss": 0.0016, "step": 38030 }, { "grad_norm": 0.07334558665752411, "learning_rate": 3.236751077033855e-05, "loss": 0.0019, "step": 38040 }, { "grad_norm": 0.04456650838255882, "learning_rate": 3.234172606185322e-05, "loss": 0.0017, "step": 38050 }, { "grad_norm": 0.07442930340766907, "learning_rate": 3.231594671748528e-05, "loss": 0.0019, "step": 38060 }, { "grad_norm": 0.07504904270172119, "learning_rate": 3.2290172745065815e-05, "loss": 0.002, "step": 38070 }, { "grad_norm": 0.07559297233819962, "learning_rate": 3.226440415242426e-05, "loss": 0.0024, "step": 38080 }, { "grad_norm": 0.06057798117399216, "learning_rate": 3.223864094738846e-05, "loss": 0.0019, "step": 38090 }, { "grad_norm": 0.040309954434633255, "learning_rate": 3.221288313778456e-05, "loss": 0.0017, "step": 38100 }, { "grad_norm": 0.05753094330430031, "learning_rate": 3.2187130731437125e-05, "loss": 0.0017, "step": 38110 }, { "grad_norm": 0.04649025946855545, "learning_rate": 3.216138373616905e-05, "loss": 0.0017, "step": 38120 }, { "grad_norm": 0.07052183151245117, "learning_rate": 3.21356421598016e-05, "loss": 0.0023, "step": 38130 }, { "grad_norm": 0.06700261682271957, "learning_rate": 3.210990601015438e-05, "loss": 0.0014, "step": 38140 }, { "grad_norm": 0.049226488918066025, "learning_rate": 3.208417529504535e-05, "loss": 0.0021, "step": 38150 }, { "grad_norm": 0.07641084492206573, "learning_rate": 3.205845002229084e-05, "loss": 0.0024, "step": 38160 }, { "grad_norm": 0.055381614714860916, "learning_rate": 3.203273019970547e-05, "loss": 0.0017, "step": 38170 }, { "grad_norm": 0.06414777785539627, "learning_rate": 3.200701583510227e-05, "loss": 0.0021, "step": 38180 }, { "grad_norm": 0.06035397946834564, "learning_rate": 3.198130693629261e-05, "loss": 0.0028, "step": 38190 }, { "grad_norm": 0.06752682477235794, "learning_rate": 3.195560351108612e-05, "loss": 0.0026, "step": 38200 }, { "grad_norm": 0.047098252922296524, "learning_rate": 3.1929905567290865e-05, "loss": 0.0021, "step": 38210 }, { "grad_norm": 0.06986498832702637, "learning_rate": 3.1904213112713164e-05, "loss": 0.002, "step": 38220 }, { "grad_norm": 0.06592295318841934, "learning_rate": 3.187852615515774e-05, "loss": 0.0017, "step": 38230 }, { "grad_norm": 0.04608063027262688, "learning_rate": 3.1852844702427606e-05, "loss": 0.0017, "step": 38240 }, { "grad_norm": 0.07827163487672806, "learning_rate": 3.18271687623241e-05, "loss": 0.0018, "step": 38250 }, { "grad_norm": 0.06508449465036392, "learning_rate": 3.1801498342646896e-05, "loss": 0.0017, "step": 38260 }, { "grad_norm": 0.059411462396383286, "learning_rate": 3.177583345119398e-05, "loss": 0.0019, "step": 38270 }, { "grad_norm": 0.05106985196471214, "learning_rate": 3.17501740957617e-05, "loss": 0.0017, "step": 38280 }, { "grad_norm": 0.062363624572753906, "learning_rate": 3.172452028414467e-05, "loss": 0.0018, "step": 38290 }, { "grad_norm": 0.07559698075056076, "learning_rate": 3.169887202413583e-05, "loss": 0.0024, "step": 38300 }, { "grad_norm": 0.09992866963148117, "learning_rate": 3.167322932352646e-05, "loss": 0.0019, "step": 38310 }, { "grad_norm": 0.06761830300092697, "learning_rate": 3.164759219010613e-05, "loss": 0.0021, "step": 38320 }, { "grad_norm": 0.0683814287185669, "learning_rate": 3.1621960631662725e-05, "loss": 0.0018, "step": 38330 }, { "grad_norm": 0.08770007640123367, "learning_rate": 3.159633465598245e-05, "loss": 0.0022, "step": 38340 }, { "grad_norm": 0.05130701884627342, "learning_rate": 3.1570714270849767e-05, "loss": 0.0019, "step": 38350 }, { "grad_norm": 0.05266723036766052, "learning_rate": 3.1545099484047516e-05, "loss": 0.0016, "step": 38360 }, { "grad_norm": 0.048487525433301926, "learning_rate": 3.151949030335674e-05, "loss": 0.0015, "step": 38370 }, { "grad_norm": 0.04326670244336128, "learning_rate": 3.149388673655687e-05, "loss": 0.0012, "step": 38380 }, { "grad_norm": 0.08305417746305466, "learning_rate": 3.146828879142559e-05, "loss": 0.0022, "step": 38390 }, { "grad_norm": 0.08935908228158951, "learning_rate": 3.1442696475738866e-05, "loss": 0.0021, "step": 38400 }, { "grad_norm": 0.0747198686003685, "learning_rate": 3.141710979727098e-05, "loss": 0.0016, "step": 38410 }, { "grad_norm": 0.0888444110751152, "learning_rate": 3.139152876379447e-05, "loss": 0.0018, "step": 38420 }, { "grad_norm": 0.07669004052877426, "learning_rate": 3.1365953383080214e-05, "loss": 0.0023, "step": 38430 }, { "grad_norm": 0.05724601820111275, "learning_rate": 3.134038366289731e-05, "loss": 0.0021, "step": 38440 }, { "grad_norm": 0.10718342661857605, "learning_rate": 3.131481961101317e-05, "loss": 0.0032, "step": 38450 }, { "grad_norm": 0.07561296224594116, "learning_rate": 3.128926123519349e-05, "loss": 0.0021, "step": 38460 }, { "grad_norm": 0.05990166217088699, "learning_rate": 3.1263708543202194e-05, "loss": 0.0017, "step": 38470 }, { "grad_norm": 0.09335974603891373, "learning_rate": 3.123816154280155e-05, "loss": 0.0018, "step": 38480 }, { "grad_norm": 0.06026032567024231, "learning_rate": 3.121262024175207e-05, "loss": 0.0021, "step": 38490 }, { "grad_norm": 0.06706468015909195, "learning_rate": 3.118708464781248e-05, "loss": 0.0018, "step": 38500 }, { "grad_norm": 0.08086840808391571, "learning_rate": 3.116155476873987e-05, "loss": 0.0022, "step": 38510 }, { "grad_norm": 0.0534793920814991, "learning_rate": 3.11360306122895e-05, "loss": 0.0017, "step": 38520 }, { "grad_norm": 0.06145884469151497, "learning_rate": 3.1110512186214975e-05, "loss": 0.0022, "step": 38530 }, { "grad_norm": 0.05392623692750931, "learning_rate": 3.1084999498268095e-05, "loss": 0.0021, "step": 38540 }, { "grad_norm": 0.09448720514774323, "learning_rate": 3.1059492556198934e-05, "loss": 0.0019, "step": 38550 }, { "grad_norm": 0.0724886953830719, "learning_rate": 3.103399136775586e-05, "loss": 0.0022, "step": 38560 }, { "grad_norm": 0.06585702300071716, "learning_rate": 3.100849594068541e-05, "loss": 0.002, "step": 38570 }, { "grad_norm": 0.07140171527862549, "learning_rate": 3.0983006282732484e-05, "loss": 0.0017, "step": 38580 }, { "grad_norm": 0.07850217074155807, "learning_rate": 3.0957522401640116e-05, "loss": 0.0022, "step": 38590 }, { "grad_norm": 0.05206010490655899, "learning_rate": 3.0932044305149645e-05, "loss": 0.0018, "step": 38600 }, { "grad_norm": 0.08537526428699493, "learning_rate": 3.090657200100068e-05, "loss": 0.0026, "step": 38610 }, { "grad_norm": 0.06023966521024704, "learning_rate": 3.088110549693099e-05, "loss": 0.0017, "step": 38620 }, { "grad_norm": 0.0640343725681305, "learning_rate": 3.085564480067667e-05, "loss": 0.0015, "step": 38630 }, { "grad_norm": 0.06861406564712524, "learning_rate": 3.0830189919971955e-05, "loss": 0.0016, "step": 38640 }, { "grad_norm": 0.060223814100027084, "learning_rate": 3.080474086254939e-05, "loss": 0.0022, "step": 38650 }, { "grad_norm": 0.08930911123752594, "learning_rate": 3.077929763613975e-05, "loss": 0.0017, "step": 38660 }, { "grad_norm": 0.06553640961647034, "learning_rate": 3.075386024847198e-05, "loss": 0.0019, "step": 38670 }, { "grad_norm": 0.06940306723117828, "learning_rate": 3.072842870727331e-05, "loss": 0.0022, "step": 38680 }, { "grad_norm": 0.07359109073877335, "learning_rate": 3.070300302026916e-05, "loss": 0.0018, "step": 38690 }, { "grad_norm": 0.06883689016103745, "learning_rate": 3.067758319518318e-05, "loss": 0.0024, "step": 38700 }, { "grad_norm": 0.045122526586055756, "learning_rate": 3.065216923973725e-05, "loss": 0.0015, "step": 38710 }, { "grad_norm": 0.05901119112968445, "learning_rate": 3.062676116165145e-05, "loss": 0.0018, "step": 38720 }, { "grad_norm": 0.08426438271999359, "learning_rate": 3.06013589686441e-05, "loss": 0.0023, "step": 38730 }, { "grad_norm": 0.05524551123380661, "learning_rate": 3.05759626684317e-05, "loss": 0.0019, "step": 38740 }, { "grad_norm": 0.11389289796352386, "learning_rate": 3.055057226872896e-05, "loss": 0.0019, "step": 38750 }, { "grad_norm": 0.05483487993478775, "learning_rate": 3.052518777724887e-05, "loss": 0.0026, "step": 38760 }, { "grad_norm": 0.06121932715177536, "learning_rate": 3.04998092017025e-05, "loss": 0.0019, "step": 38770 }, { "grad_norm": 0.05709533020853996, "learning_rate": 3.0474436549799246e-05, "loss": 0.0015, "step": 38780 }, { "grad_norm": 0.06974329799413681, "learning_rate": 3.044906982924661e-05, "loss": 0.0016, "step": 38790 }, { "grad_norm": 0.055903248488903046, "learning_rate": 3.0423709047750337e-05, "loss": 0.0018, "step": 38800 }, { "grad_norm": 0.05732616409659386, "learning_rate": 3.03983542130144e-05, "loss": 0.0021, "step": 38810 }, { "grad_norm": 0.04870752617716789, "learning_rate": 3.0373005332740877e-05, "loss": 0.0026, "step": 38820 }, { "grad_norm": 0.06256067752838135, "learning_rate": 3.034766241463013e-05, "loss": 0.0031, "step": 38830 }, { "grad_norm": 0.049271926283836365, "learning_rate": 3.032232546638064e-05, "loss": 0.0018, "step": 38840 }, { "grad_norm": 0.05274007469415665, "learning_rate": 3.0296994495689114e-05, "loss": 0.0025, "step": 38850 }, { "grad_norm": 0.03624565526843071, "learning_rate": 3.0271669510250444e-05, "loss": 0.0015, "step": 38860 }, { "grad_norm": 0.055468957871198654, "learning_rate": 3.024635051775766e-05, "loss": 0.0031, "step": 38870 }, { "grad_norm": 0.0721917673945427, "learning_rate": 3.022103752590205e-05, "loss": 0.0022, "step": 38880 }, { "grad_norm": 0.05480673164129257, "learning_rate": 3.0195730542372992e-05, "loss": 0.002, "step": 38890 }, { "grad_norm": 0.0701991394162178, "learning_rate": 3.0170429574858084e-05, "loss": 0.002, "step": 38900 }, { "grad_norm": 0.04788541421294212, "learning_rate": 3.0145134631043127e-05, "loss": 0.0021, "step": 38910 }, { "grad_norm": 0.05938534066081047, "learning_rate": 3.0119845718612018e-05, "loss": 0.0018, "step": 38920 }, { "grad_norm": 0.06708680093288422, "learning_rate": 3.009456284524688e-05, "loss": 0.0019, "step": 38930 }, { "grad_norm": 0.048370759934186935, "learning_rate": 3.0069286018627967e-05, "loss": 0.0018, "step": 38940 }, { "grad_norm": 0.04554510861635208, "learning_rate": 3.0044015246433743e-05, "loss": 0.002, "step": 38950 }, { "grad_norm": 0.09878888726234436, "learning_rate": 3.0018750536340755e-05, "loss": 0.0019, "step": 38960 }, { "grad_norm": 0.06652796268463135, "learning_rate": 2.999349189602378e-05, "loss": 0.0014, "step": 38970 }, { "grad_norm": 0.048394329845905304, "learning_rate": 2.9968239333155733e-05, "loss": 0.0013, "step": 38980 }, { "grad_norm": 0.06855449825525284, "learning_rate": 2.994299285540767e-05, "loss": 0.0025, "step": 38990 }, { "grad_norm": 0.06079397350549698, "learning_rate": 2.9917752470448813e-05, "loss": 0.0023, "step": 39000 }, { "grad_norm": 0.06259370595216751, "learning_rate": 2.9892518185946495e-05, "loss": 0.0018, "step": 39010 }, { "grad_norm": 0.056808989495038986, "learning_rate": 2.986729000956624e-05, "loss": 0.0023, "step": 39020 }, { "grad_norm": 0.05299524590373039, "learning_rate": 2.9842067948971736e-05, "loss": 0.0021, "step": 39030 }, { "grad_norm": 0.08663599193096161, "learning_rate": 2.9816852011824727e-05, "loss": 0.0019, "step": 39040 }, { "grad_norm": 0.058486491441726685, "learning_rate": 2.979164220578519e-05, "loss": 0.0017, "step": 39050 }, { "grad_norm": 0.034115180373191833, "learning_rate": 2.9766438538511165e-05, "loss": 0.0013, "step": 39060 }, { "grad_norm": 0.0801079049706459, "learning_rate": 2.9741241017658873e-05, "loss": 0.0021, "step": 39070 }, { "grad_norm": 0.042667243629693985, "learning_rate": 2.971604965088267e-05, "loss": 0.0015, "step": 39080 }, { "grad_norm": 0.048271551728248596, "learning_rate": 2.9690864445835008e-05, "loss": 0.0014, "step": 39090 }, { "grad_norm": 0.05592638999223709, "learning_rate": 2.966568541016651e-05, "loss": 0.0014, "step": 39100 }, { "grad_norm": 0.06460201740264893, "learning_rate": 2.9640512551525867e-05, "loss": 0.0017, "step": 39110 }, { "grad_norm": 0.0729169175028801, "learning_rate": 2.961534587755995e-05, "loss": 0.0018, "step": 39120 }, { "grad_norm": 0.07034803926944733, "learning_rate": 2.959018539591375e-05, "loss": 0.0014, "step": 39130 }, { "grad_norm": 0.07395219802856445, "learning_rate": 2.9565031114230325e-05, "loss": 0.0019, "step": 39140 }, { "grad_norm": 0.056839924305677414, "learning_rate": 2.9539883040150895e-05, "loss": 0.0021, "step": 39150 }, { "grad_norm": 0.046335481107234955, "learning_rate": 2.9514741181314774e-05, "loss": 0.0021, "step": 39160 }, { "grad_norm": 0.050677355378866196, "learning_rate": 2.94896055453594e-05, "loss": 0.0019, "step": 39170 }, { "grad_norm": 0.06418183445930481, "learning_rate": 2.9464476139920332e-05, "loss": 0.0024, "step": 39180 }, { "grad_norm": 0.06270169466733932, "learning_rate": 2.9439352972631186e-05, "loss": 0.0017, "step": 39190 }, { "grad_norm": 0.06358543783426285, "learning_rate": 2.9414236051123757e-05, "loss": 0.0017, "step": 39200 }, { "grad_norm": 0.0732765942811966, "learning_rate": 2.938912538302785e-05, "loss": 0.0025, "step": 39210 }, { "grad_norm": 0.06655871123075485, "learning_rate": 2.9364020975971464e-05, "loss": 0.0025, "step": 39220 }, { "grad_norm": 0.06046941131353378, "learning_rate": 2.9338922837580657e-05, "loss": 0.0019, "step": 39230 }, { "grad_norm": 0.06725911796092987, "learning_rate": 2.931383097547955e-05, "loss": 0.0014, "step": 39240 }, { "grad_norm": 0.05899517610669136, "learning_rate": 2.928874539729043e-05, "loss": 0.0018, "step": 39250 }, { "grad_norm": 0.07495174556970596, "learning_rate": 2.926366611063358e-05, "loss": 0.0022, "step": 39260 }, { "grad_norm": 0.05464685708284378, "learning_rate": 2.9238593123127463e-05, "loss": 0.0021, "step": 39270 }, { "grad_norm": 0.06206563860177994, "learning_rate": 2.9213526442388583e-05, "loss": 0.0016, "step": 39280 }, { "grad_norm": 0.06196412816643715, "learning_rate": 2.9188466076031545e-05, "loss": 0.0017, "step": 39290 }, { "grad_norm": 0.04596443474292755, "learning_rate": 2.9163412031669012e-05, "loss": 0.0015, "step": 39300 }, { "grad_norm": 0.07658747583627701, "learning_rate": 2.913836431691175e-05, "loss": 0.0034, "step": 39310 }, { "grad_norm": 0.09067443013191223, "learning_rate": 2.9113322939368583e-05, "loss": 0.0026, "step": 39320 }, { "grad_norm": 0.08415514975786209, "learning_rate": 2.9088287906646427e-05, "loss": 0.0023, "step": 39330 }, { "grad_norm": 0.06987850368022919, "learning_rate": 2.906325922635024e-05, "loss": 0.0024, "step": 39340 }, { "grad_norm": 0.07359812408685684, "learning_rate": 2.903823690608313e-05, "loss": 0.0025, "step": 39350 }, { "grad_norm": 0.060969457030296326, "learning_rate": 2.9013220953446174e-05, "loss": 0.0021, "step": 39360 }, { "grad_norm": 0.07623238861560822, "learning_rate": 2.8988211376038564e-05, "loss": 0.0027, "step": 39370 }, { "grad_norm": 0.06929678469896317, "learning_rate": 2.8963208181457564e-05, "loss": 0.0019, "step": 39380 }, { "grad_norm": 0.08145448565483093, "learning_rate": 2.8938211377298453e-05, "loss": 0.002, "step": 39390 }, { "grad_norm": 0.05761203542351723, "learning_rate": 2.8913220971154652e-05, "loss": 0.0019, "step": 39400 }, { "grad_norm": 0.06930030882358551, "learning_rate": 2.888823697061753e-05, "loss": 0.0021, "step": 39410 }, { "grad_norm": 0.061621151864528656, "learning_rate": 2.8863259383276618e-05, "loss": 0.0021, "step": 39420 }, { "grad_norm": 0.05429829657077789, "learning_rate": 2.8838288216719395e-05, "loss": 0.0014, "step": 39430 }, { "grad_norm": 0.04792310297489166, "learning_rate": 2.8813323478531484e-05, "loss": 0.0016, "step": 39440 }, { "grad_norm": 0.06192335486412048, "learning_rate": 2.8788365176296496e-05, "loss": 0.0022, "step": 39450 }, { "grad_norm": 0.0515565499663353, "learning_rate": 2.876341331759611e-05, "loss": 0.0018, "step": 39460 }, { "grad_norm": 0.06563284993171692, "learning_rate": 2.8738467910010036e-05, "loss": 0.0015, "step": 39470 }, { "grad_norm": 0.0441201888024807, "learning_rate": 2.8713528961116032e-05, "loss": 0.0013, "step": 39480 }, { "grad_norm": 0.0410403348505497, "learning_rate": 2.8688596478489875e-05, "loss": 0.0013, "step": 39490 }, { "grad_norm": 0.038203295320272446, "learning_rate": 2.8663670469705434e-05, "loss": 0.0011, "step": 39500 }, { "grad_norm": 0.04952717944979668, "learning_rate": 2.8638750942334546e-05, "loss": 0.0015, "step": 39510 }, { "grad_norm": 0.06319110095500946, "learning_rate": 2.8613837903947115e-05, "loss": 0.0013, "step": 39520 }, { "grad_norm": 0.05653717741370201, "learning_rate": 2.858893136211106e-05, "loss": 0.0019, "step": 39530 }, { "grad_norm": 0.053484924137592316, "learning_rate": 2.8564031324392315e-05, "loss": 0.0018, "step": 39540 }, { "grad_norm": 0.04456328600645065, "learning_rate": 2.85391377983549e-05, "loss": 0.0012, "step": 39550 }, { "grad_norm": 0.05680397152900696, "learning_rate": 2.851425079156075e-05, "loss": 0.0017, "step": 39560 }, { "grad_norm": 0.05591423064470291, "learning_rate": 2.848937031156994e-05, "loss": 0.0028, "step": 39570 }, { "grad_norm": 0.046736545860767365, "learning_rate": 2.846449636594044e-05, "loss": 0.0019, "step": 39580 }, { "grad_norm": 0.05178992822766304, "learning_rate": 2.843962896222836e-05, "loss": 0.0014, "step": 39590 }, { "grad_norm": 0.07585121691226959, "learning_rate": 2.8414768107987722e-05, "loss": 0.0019, "step": 39600 }, { "grad_norm": 0.054572608321905136, "learning_rate": 2.838991381077061e-05, "loss": 0.0019, "step": 39610 }, { "grad_norm": 0.07147668302059174, "learning_rate": 2.83650660781271e-05, "loss": 0.0023, "step": 39620 }, { "grad_norm": 0.05577236786484718, "learning_rate": 2.8340224917605285e-05, "loss": 0.0017, "step": 39630 }, { "grad_norm": 0.07810740172863007, "learning_rate": 2.831539033675122e-05, "loss": 0.0018, "step": 39640 }, { "grad_norm": 0.06982266902923584, "learning_rate": 2.8290562343109038e-05, "loss": 0.0025, "step": 39650 }, { "grad_norm": 0.06727434694766998, "learning_rate": 2.826574094422082e-05, "loss": 0.0019, "step": 39660 }, { "grad_norm": 0.04341048002243042, "learning_rate": 2.8240926147626645e-05, "loss": 0.0017, "step": 39670 }, { "grad_norm": 0.04978841543197632, "learning_rate": 2.8216117960864586e-05, "loss": 0.0022, "step": 39680 }, { "grad_norm": 0.09926304966211319, "learning_rate": 2.8191316391470703e-05, "loss": 0.0032, "step": 39690 }, { "grad_norm": 0.06368870288133621, "learning_rate": 2.816652144697911e-05, "loss": 0.0023, "step": 39700 }, { "grad_norm": 0.06400303542613983, "learning_rate": 2.8141733134921783e-05, "loss": 0.0019, "step": 39710 }, { "grad_norm": 0.04785206913948059, "learning_rate": 2.811695146282884e-05, "loss": 0.0018, "step": 39720 }, { "grad_norm": 0.0671839788556099, "learning_rate": 2.8092176438228212e-05, "loss": 0.0016, "step": 39730 }, { "grad_norm": 0.08348608016967773, "learning_rate": 2.806740806864598e-05, "loss": 0.002, "step": 39740 }, { "grad_norm": 0.0646158754825592, "learning_rate": 2.804264636160604e-05, "loss": 0.0014, "step": 39750 }, { "grad_norm": 0.07709211111068726, "learning_rate": 2.8017891324630402e-05, "loss": 0.0019, "step": 39760 }, { "grad_norm": 0.058477722108364105, "learning_rate": 2.7993142965238976e-05, "loss": 0.0022, "step": 39770 }, { "grad_norm": 0.0648910328745842, "learning_rate": 2.7968401290949665e-05, "loss": 0.0019, "step": 39780 }, { "grad_norm": 0.11398688703775406, "learning_rate": 2.7943666309278328e-05, "loss": 0.0023, "step": 39790 }, { "grad_norm": 0.08896782994270325, "learning_rate": 2.7918938027738783e-05, "loss": 0.0022, "step": 39800 }, { "grad_norm": 0.07008612155914307, "learning_rate": 2.789421645384287e-05, "loss": 0.0017, "step": 39810 }, { "grad_norm": 0.06553495675325394, "learning_rate": 2.786950159510032e-05, "loss": 0.0034, "step": 39820 }, { "grad_norm": 0.0734734982252121, "learning_rate": 2.7844793459018876e-05, "loss": 0.0019, "step": 39830 }, { "grad_norm": 0.055852990597486496, "learning_rate": 2.7820092053104195e-05, "loss": 0.0016, "step": 39840 }, { "grad_norm": 0.07333299517631531, "learning_rate": 2.7795397384859933e-05, "loss": 0.0019, "step": 39850 }, { "grad_norm": 0.043061889708042145, "learning_rate": 2.7770709461787638e-05, "loss": 0.0016, "step": 39860 }, { "grad_norm": 0.05875544250011444, "learning_rate": 2.7746028291386915e-05, "loss": 0.0013, "step": 39870 }, { "grad_norm": 0.04194403439760208, "learning_rate": 2.772135388115519e-05, "loss": 0.0019, "step": 39880 }, { "grad_norm": 0.051910776644945145, "learning_rate": 2.7696686238587945e-05, "loss": 0.0018, "step": 39890 }, { "grad_norm": 0.040034689009189606, "learning_rate": 2.7672025371178505e-05, "loss": 0.0018, "step": 39900 }, { "grad_norm": 0.050310198217630386, "learning_rate": 2.7647371286418238e-05, "loss": 0.0016, "step": 39910 }, { "grad_norm": 0.04776759073138237, "learning_rate": 2.762272399179639e-05, "loss": 0.0014, "step": 39920 }, { "grad_norm": 0.045911967754364014, "learning_rate": 2.7598083494800154e-05, "loss": 0.0014, "step": 39930 }, { "grad_norm": 0.05494828149676323, "learning_rate": 2.7573449802914664e-05, "loss": 0.0012, "step": 39940 }, { "grad_norm": 0.04522963613271713, "learning_rate": 2.7548822923622964e-05, "loss": 0.0013, "step": 39950 }, { "grad_norm": 0.04258984327316284, "learning_rate": 2.752420286440609e-05, "loss": 0.0016, "step": 39960 }, { "grad_norm": 0.05347989499568939, "learning_rate": 2.749958963274295e-05, "loss": 0.0017, "step": 39970 }, { "grad_norm": 0.04701005667448044, "learning_rate": 2.747498323611039e-05, "loss": 0.0021, "step": 39980 }, { "grad_norm": 0.067962646484375, "learning_rate": 2.7450383681983184e-05, "loss": 0.0017, "step": 39990 }, { "grad_norm": 0.061716776341199875, "learning_rate": 2.742579097783403e-05, "loss": 0.0026, "step": 40000 }, { "grad_norm": 0.07535188645124435, "learning_rate": 2.7401205131133512e-05, "loss": 0.0019, "step": 40010 }, { "grad_norm": 0.060579001903533936, "learning_rate": 2.7376626149350238e-05, "loss": 0.0017, "step": 40020 }, { "grad_norm": 0.08482371270656586, "learning_rate": 2.735205403995056e-05, "loss": 0.0022, "step": 40030 }, { "grad_norm": 0.06681688874959946, "learning_rate": 2.7327488810398917e-05, "loss": 0.0021, "step": 40040 }, { "grad_norm": 0.06198566406965256, "learning_rate": 2.7302930468157507e-05, "loss": 0.002, "step": 40050 }, { "grad_norm": 0.0862775444984436, "learning_rate": 2.727837902068655e-05, "loss": 0.0017, "step": 40060 }, { "grad_norm": 0.055194515734910965, "learning_rate": 2.7253834475444123e-05, "loss": 0.0019, "step": 40070 }, { "grad_norm": 0.07019700109958649, "learning_rate": 2.7229296839886204e-05, "loss": 0.003, "step": 40080 }, { "grad_norm": 0.049706459045410156, "learning_rate": 2.720476612146668e-05, "loss": 0.0019, "step": 40090 }, { "grad_norm": 0.13638941943645477, "learning_rate": 2.7180242327637317e-05, "loss": 0.0024, "step": 40100 }, { "grad_norm": 0.04436057433485985, "learning_rate": 2.7155725465847826e-05, "loss": 0.002, "step": 40110 }, { "grad_norm": 0.0764082819223404, "learning_rate": 2.713121554354578e-05, "loss": 0.0015, "step": 40120 }, { "grad_norm": 0.06151760742068291, "learning_rate": 2.7106712568176628e-05, "loss": 0.0019, "step": 40130 }, { "grad_norm": 0.07356014102697372, "learning_rate": 2.708221654718374e-05, "loss": 0.0017, "step": 40140 }, { "grad_norm": 0.08108203113079071, "learning_rate": 2.7057727488008357e-05, "loss": 0.0022, "step": 40150 }, { "grad_norm": 0.06788865476846695, "learning_rate": 2.703324539808961e-05, "loss": 0.002, "step": 40160 }, { "grad_norm": 0.061035964637994766, "learning_rate": 2.7008770284864505e-05, "loss": 0.002, "step": 40170 }, { "grad_norm": 0.054529234766960144, "learning_rate": 2.6984302155767916e-05, "loss": 0.0012, "step": 40180 }, { "grad_norm": 0.04249565675854683, "learning_rate": 2.6959841018232683e-05, "loss": 0.0021, "step": 40190 }, { "grad_norm": 0.05477271229028702, "learning_rate": 2.693538687968937e-05, "loss": 0.0016, "step": 40200 }, { "grad_norm": 0.07462922483682632, "learning_rate": 2.6910939747566556e-05, "loss": 0.0018, "step": 40210 }, { "grad_norm": 0.07388240098953247, "learning_rate": 2.6886499629290607e-05, "loss": 0.0016, "step": 40220 }, { "grad_norm": 0.048475608229637146, "learning_rate": 2.6862066532285802e-05, "loss": 0.0014, "step": 40230 }, { "grad_norm": 0.046621307730674744, "learning_rate": 2.6837640463974262e-05, "loss": 0.0025, "step": 40240 }, { "grad_norm": 0.059311408549547195, "learning_rate": 2.681322143177596e-05, "loss": 0.0026, "step": 40250 }, { "grad_norm": 0.08311508595943451, "learning_rate": 2.678880944310882e-05, "loss": 0.002, "step": 40260 }, { "grad_norm": 0.04273460805416107, "learning_rate": 2.6764404505388474e-05, "loss": 0.0017, "step": 40270 }, { "grad_norm": 0.051506947726011276, "learning_rate": 2.6740006626028558e-05, "loss": 0.0017, "step": 40280 }, { "grad_norm": 0.05074632912874222, "learning_rate": 2.671561581244048e-05, "loss": 0.0021, "step": 40290 }, { "grad_norm": 0.05175533890724182, "learning_rate": 2.6691232072033536e-05, "loss": 0.0028, "step": 40300 }, { "grad_norm": 0.08521302789449692, "learning_rate": 2.6666855412214852e-05, "loss": 0.0022, "step": 40310 }, { "grad_norm": 0.06363514810800552, "learning_rate": 2.664248584038942e-05, "loss": 0.0015, "step": 40320 }, { "grad_norm": 0.05018436536192894, "learning_rate": 2.6618123363960047e-05, "loss": 0.0015, "step": 40330 }, { "grad_norm": 0.046772751957178116, "learning_rate": 2.659376799032748e-05, "loss": 0.0019, "step": 40340 }, { "grad_norm": 0.05517091229557991, "learning_rate": 2.6569419726890145e-05, "loss": 0.0017, "step": 40350 }, { "grad_norm": 0.0720307007431984, "learning_rate": 2.654507858104447e-05, "loss": 0.0022, "step": 40360 }, { "grad_norm": 0.08229152113199234, "learning_rate": 2.652074456018463e-05, "loss": 0.0024, "step": 40370 }, { "grad_norm": 0.06704109907150269, "learning_rate": 2.6496417671702646e-05, "loss": 0.0014, "step": 40380 }, { "grad_norm": 0.04852987825870514, "learning_rate": 2.6472097922988427e-05, "loss": 0.0018, "step": 40390 }, { "grad_norm": 0.06703303754329681, "learning_rate": 2.6447785321429607e-05, "loss": 0.0015, "step": 40400 }, { "grad_norm": 0.054793164134025574, "learning_rate": 2.6423479874411784e-05, "loss": 0.0018, "step": 40410 }, { "grad_norm": 0.05872382968664169, "learning_rate": 2.6399181589318234e-05, "loss": 0.0017, "step": 40420 }, { "grad_norm": 0.07635262608528137, "learning_rate": 2.6374890473530188e-05, "loss": 0.0019, "step": 40430 }, { "grad_norm": 0.07163573056459427, "learning_rate": 2.635060653442664e-05, "loss": 0.0019, "step": 40440 }, { "grad_norm": 0.045780379325151443, "learning_rate": 2.6326329779384395e-05, "loss": 0.0011, "step": 40450 }, { "grad_norm": 0.07467598468065262, "learning_rate": 2.63020602157781e-05, "loss": 0.0017, "step": 40460 }, { "grad_norm": 0.06429940462112427, "learning_rate": 2.62777978509802e-05, "loss": 0.0018, "step": 40470 }, { "grad_norm": 0.050816942006349564, "learning_rate": 2.6253542692360954e-05, "loss": 0.0015, "step": 40480 }, { "grad_norm": 0.04947951063513756, "learning_rate": 2.6229294747288458e-05, "loss": 0.0018, "step": 40490 }, { "grad_norm": 0.05087492987513542, "learning_rate": 2.6205054023128596e-05, "loss": 0.0013, "step": 40500 }, { "grad_norm": 0.045039426535367966, "learning_rate": 2.6180820527245043e-05, "loss": 0.0015, "step": 40510 }, { "grad_norm": 0.02986074984073639, "learning_rate": 2.6156594266999313e-05, "loss": 0.0015, "step": 40520 }, { "grad_norm": 0.05904185399413109, "learning_rate": 2.6132375249750672e-05, "loss": 0.0012, "step": 40530 }, { "grad_norm": 0.043338630348443985, "learning_rate": 2.6108163482856286e-05, "loss": 0.0021, "step": 40540 }, { "grad_norm": 0.0659855455160141, "learning_rate": 2.6083958973670964e-05, "loss": 0.0015, "step": 40550 }, { "grad_norm": 0.08389313519001007, "learning_rate": 2.6059761729547483e-05, "loss": 0.0019, "step": 40560 }, { "grad_norm": 0.07440456002950668, "learning_rate": 2.603557175783624e-05, "loss": 0.0018, "step": 40570 }, { "grad_norm": 0.05846723914146423, "learning_rate": 2.601138906588559e-05, "loss": 0.0018, "step": 40580 }, { "grad_norm": 0.06160348281264305, "learning_rate": 2.598721366104152e-05, "loss": 0.0017, "step": 40590 }, { "grad_norm": 0.056220389902591705, "learning_rate": 2.5963045550647945e-05, "loss": 0.0022, "step": 40600 }, { "grad_norm": 0.07064549624919891, "learning_rate": 2.5938884742046466e-05, "loss": 0.0023, "step": 40610 }, { "grad_norm": 0.06326733529567719, "learning_rate": 2.5914731242576507e-05, "loss": 0.002, "step": 40620 }, { "grad_norm": 0.07001452147960663, "learning_rate": 2.5890585059575268e-05, "loss": 0.0015, "step": 40630 }, { "grad_norm": 0.0702277198433876, "learning_rate": 2.5866446200377688e-05, "loss": 0.0017, "step": 40640 }, { "grad_norm": 0.045393284410238266, "learning_rate": 2.5842314672316566e-05, "loss": 0.0012, "step": 40650 }, { "grad_norm": 0.05033227056264877, "learning_rate": 2.581819048272239e-05, "loss": 0.002, "step": 40660 }, { "grad_norm": 0.05577196553349495, "learning_rate": 2.5794073638923478e-05, "loss": 0.0014, "step": 40670 }, { "grad_norm": 0.07194779813289642, "learning_rate": 2.576996414824586e-05, "loss": 0.0018, "step": 40680 }, { "grad_norm": 0.04083656147122383, "learning_rate": 2.574586201801339e-05, "loss": 0.0017, "step": 40690 }, { "grad_norm": 0.06909669935703278, "learning_rate": 2.572176725554762e-05, "loss": 0.0018, "step": 40700 }, { "grad_norm": 0.06071378290653229, "learning_rate": 2.5697679868167966e-05, "loss": 0.002, "step": 40710 }, { "grad_norm": 0.03965817391872406, "learning_rate": 2.5673599863191468e-05, "loss": 0.0019, "step": 40720 }, { "grad_norm": 0.0407954677939415, "learning_rate": 2.564952724793306e-05, "loss": 0.003, "step": 40730 }, { "grad_norm": 0.055606719106435776, "learning_rate": 2.5625462029705306e-05, "loss": 0.0016, "step": 40740 }, { "grad_norm": 0.056380677968263626, "learning_rate": 2.5601404215818624e-05, "loss": 0.0013, "step": 40750 }, { "grad_norm": 0.08551832288503647, "learning_rate": 2.5577353813581144e-05, "loss": 0.0021, "step": 40760 }, { "grad_norm": 0.059899892657995224, "learning_rate": 2.5553310830298733e-05, "loss": 0.002, "step": 40770 }, { "grad_norm": 0.07630915939807892, "learning_rate": 2.5529275273275012e-05, "loss": 0.0016, "step": 40780 }, { "grad_norm": 0.0627322643995285, "learning_rate": 2.550524714981133e-05, "loss": 0.0015, "step": 40790 }, { "grad_norm": 0.054408617317676544, "learning_rate": 2.5481226467206837e-05, "loss": 0.0015, "step": 40800 }, { "grad_norm": 0.05210895836353302, "learning_rate": 2.5457213232758365e-05, "loss": 0.0015, "step": 40810 }, { "grad_norm": 0.06356265395879745, "learning_rate": 2.5433207453760498e-05, "loss": 0.0018, "step": 40820 }, { "grad_norm": 0.05268098786473274, "learning_rate": 2.5409209137505552e-05, "loss": 0.002, "step": 40830 }, { "grad_norm": 0.05702586472034454, "learning_rate": 2.5385218291283597e-05, "loss": 0.0016, "step": 40840 }, { "grad_norm": 0.0424913726747036, "learning_rate": 2.5361234922382383e-05, "loss": 0.0014, "step": 40850 }, { "grad_norm": 0.049688536673784256, "learning_rate": 2.533725903808749e-05, "loss": 0.0016, "step": 40860 }, { "grad_norm": 0.07291921228170395, "learning_rate": 2.5313290645682085e-05, "loss": 0.0015, "step": 40870 }, { "grad_norm": 0.06674496084451675, "learning_rate": 2.52893297524472e-05, "loss": 0.0023, "step": 40880 }, { "grad_norm": 0.07790017127990723, "learning_rate": 2.526537636566145e-05, "loss": 0.0019, "step": 40890 }, { "grad_norm": 0.07537230849266052, "learning_rate": 2.5241430492601305e-05, "loss": 0.0017, "step": 40900 }, { "grad_norm": 0.07948615401983261, "learning_rate": 2.5217492140540867e-05, "loss": 0.0017, "step": 40910 }, { "grad_norm": 0.05388753116130829, "learning_rate": 2.5193561316751967e-05, "loss": 0.0016, "step": 40920 }, { "grad_norm": 0.09977466613054276, "learning_rate": 2.516963802850416e-05, "loss": 0.0018, "step": 40930 }, { "grad_norm": 0.05563345551490784, "learning_rate": 2.5145722283064698e-05, "loss": 0.0016, "step": 40940 }, { "grad_norm": 0.04928652569651604, "learning_rate": 2.5121814087698602e-05, "loss": 0.0024, "step": 40950 }, { "grad_norm": 0.034179650247097015, "learning_rate": 2.509791344966848e-05, "loss": 0.0016, "step": 40960 }, { "grad_norm": 0.038210175931453705, "learning_rate": 2.5074020376234768e-05, "loss": 0.0019, "step": 40970 }, { "grad_norm": 0.06397689133882523, "learning_rate": 2.5050134874655534e-05, "loss": 0.0016, "step": 40980 }, { "grad_norm": 0.08224061876535416, "learning_rate": 2.5026256952186566e-05, "loss": 0.0014, "step": 40990 }, { "grad_norm": 0.04880569875240326, "learning_rate": 2.5002386616081335e-05, "loss": 0.0021, "step": 41000 }, { "grad_norm": 0.05839892849326134, "learning_rate": 2.497852387359103e-05, "loss": 0.0015, "step": 41010 }, { "grad_norm": 0.049272991716861725, "learning_rate": 2.4954668731964496e-05, "loss": 0.0014, "step": 41020 }, { "grad_norm": 0.04388956353068352, "learning_rate": 2.4930821198448364e-05, "loss": 0.0018, "step": 41030 }, { "grad_norm": 0.05697459354996681, "learning_rate": 2.4906981280286796e-05, "loss": 0.0024, "step": 41040 }, { "grad_norm": 0.05088980495929718, "learning_rate": 2.488314898472179e-05, "loss": 0.0023, "step": 41050 }, { "grad_norm": 0.05695926770567894, "learning_rate": 2.485932431899295e-05, "loss": 0.0016, "step": 41060 }, { "grad_norm": 0.04808083921670914, "learning_rate": 2.4835507290337584e-05, "loss": 0.0015, "step": 41070 }, { "grad_norm": 0.055955007672309875, "learning_rate": 2.4811697905990672e-05, "loss": 0.0014, "step": 41080 }, { "grad_norm": 0.05327450856566429, "learning_rate": 2.4787896173184854e-05, "loss": 0.0018, "step": 41090 }, { "grad_norm": 0.046948183327913284, "learning_rate": 2.4764102099150534e-05, "loss": 0.0018, "step": 41100 }, { "grad_norm": 0.0527791865170002, "learning_rate": 2.4740315691115644e-05, "loss": 0.0028, "step": 41110 }, { "grad_norm": 0.08386292308568954, "learning_rate": 2.4716536956305918e-05, "loss": 0.0025, "step": 41120 }, { "grad_norm": 0.07208546996116638, "learning_rate": 2.4692765901944697e-05, "loss": 0.0015, "step": 41130 }, { "grad_norm": 0.05968081206083298, "learning_rate": 2.4669002535253e-05, "loss": 0.0017, "step": 41140 }, { "grad_norm": 0.04660049080848694, "learning_rate": 2.46452468634495e-05, "loss": 0.0015, "step": 41150 }, { "grad_norm": 0.056341733783483505, "learning_rate": 2.462149889375055e-05, "loss": 0.0017, "step": 41160 }, { "grad_norm": 0.06721089780330658, "learning_rate": 2.459775863337014e-05, "loss": 0.0015, "step": 41170 }, { "grad_norm": 0.0821007713675499, "learning_rate": 2.4574026089519985e-05, "loss": 0.0027, "step": 41180 }, { "grad_norm": 0.06633060425519943, "learning_rate": 2.4550301269409333e-05, "loss": 0.0022, "step": 41190 }, { "grad_norm": 0.05427846685051918, "learning_rate": 2.4526584180245216e-05, "loss": 0.0015, "step": 41200 }, { "grad_norm": 0.03544706478714943, "learning_rate": 2.4502874829232236e-05, "loss": 0.0014, "step": 41210 }, { "grad_norm": 0.05742345750331879, "learning_rate": 2.447917322357267e-05, "loss": 0.0016, "step": 41220 }, { "grad_norm": 0.05405206233263016, "learning_rate": 2.4455479370466443e-05, "loss": 0.0014, "step": 41230 }, { "grad_norm": 0.04622355103492737, "learning_rate": 2.4431793277111097e-05, "loss": 0.0013, "step": 41240 }, { "grad_norm": 0.04662320762872696, "learning_rate": 2.4408114950701905e-05, "loss": 0.0013, "step": 41250 }, { "grad_norm": 0.05033839866518974, "learning_rate": 2.4384444398431634e-05, "loss": 0.0021, "step": 41260 }, { "grad_norm": 0.04299694672226906, "learning_rate": 2.4360781627490837e-05, "loss": 0.0016, "step": 41270 }, { "grad_norm": 0.0683845579624176, "learning_rate": 2.433712664506762e-05, "loss": 0.0016, "step": 41280 }, { "grad_norm": 0.05995451286435127, "learning_rate": 2.431347945834774e-05, "loss": 0.0019, "step": 41290 }, { "grad_norm": 0.0825972631573677, "learning_rate": 2.428984007451458e-05, "loss": 0.0022, "step": 41300 }, { "grad_norm": 0.05076994746923447, "learning_rate": 2.426620850074917e-05, "loss": 0.0015, "step": 41310 }, { "grad_norm": 0.0730394646525383, "learning_rate": 2.424258474423014e-05, "loss": 0.0025, "step": 41320 }, { "grad_norm": 0.056533269584178925, "learning_rate": 2.421896881213382e-05, "loss": 0.0019, "step": 41330 }, { "grad_norm": 0.07595330476760864, "learning_rate": 2.419536071163402e-05, "loss": 0.0016, "step": 41340 }, { "grad_norm": 0.04010976478457451, "learning_rate": 2.417176044990233e-05, "loss": 0.002, "step": 41350 }, { "grad_norm": 0.046416983008384705, "learning_rate": 2.4148168034107855e-05, "loss": 0.0014, "step": 41360 }, { "grad_norm": 0.06403342634439468, "learning_rate": 2.4124583471417355e-05, "loss": 0.0019, "step": 41370 }, { "grad_norm": 0.0418921560049057, "learning_rate": 2.41010067689952e-05, "loss": 0.002, "step": 41380 }, { "grad_norm": 0.07692967355251312, "learning_rate": 2.4077437934003338e-05, "loss": 0.0022, "step": 41390 }, { "grad_norm": 0.07518427073955536, "learning_rate": 2.405387697360143e-05, "loss": 0.0024, "step": 41400 }, { "grad_norm": 0.04541578143835068, "learning_rate": 2.4030323894946595e-05, "loss": 0.0016, "step": 41410 }, { "grad_norm": 0.034985002130270004, "learning_rate": 2.40067787051937e-05, "loss": 0.0013, "step": 41420 }, { "grad_norm": 0.06186091527342796, "learning_rate": 2.3983241411495087e-05, "loss": 0.0017, "step": 41430 }, { "grad_norm": 0.06759852170944214, "learning_rate": 2.3959712021000823e-05, "loss": 0.0018, "step": 41440 }, { "grad_norm": 0.04246590659022331, "learning_rate": 2.3936190540858495e-05, "loss": 0.0015, "step": 41450 }, { "grad_norm": 0.037742357701063156, "learning_rate": 2.39126769782133e-05, "loss": 0.0015, "step": 41460 }, { "grad_norm": 0.047571923583745956, "learning_rate": 2.388917134020805e-05, "loss": 0.0018, "step": 41470 }, { "grad_norm": 0.04854568466544151, "learning_rate": 2.3865673633983128e-05, "loss": 0.0017, "step": 41480 }, { "grad_norm": 0.04793943837285042, "learning_rate": 2.3842183866676492e-05, "loss": 0.0018, "step": 41490 }, { "grad_norm": 0.043044112622737885, "learning_rate": 2.381870204542377e-05, "loss": 0.0014, "step": 41500 }, { "grad_norm": 0.06488317251205444, "learning_rate": 2.379522817735808e-05, "loss": 0.0014, "step": 41510 }, { "grad_norm": 0.04416489601135254, "learning_rate": 2.377176226961018e-05, "loss": 0.0017, "step": 41520 }, { "grad_norm": 0.04505199193954468, "learning_rate": 2.3748304329308384e-05, "loss": 0.0015, "step": 41530 }, { "grad_norm": 0.08719655871391296, "learning_rate": 2.372485436357858e-05, "loss": 0.0014, "step": 41540 }, { "grad_norm": 0.04268303140997887, "learning_rate": 2.3701412379544296e-05, "loss": 0.002, "step": 41550 }, { "grad_norm": 0.055333323776721954, "learning_rate": 2.367797838432653e-05, "loss": 0.0018, "step": 41560 }, { "grad_norm": 0.04065929725766182, "learning_rate": 2.3654552385043967e-05, "loss": 0.0014, "step": 41570 }, { "grad_norm": 0.06550721079111099, "learning_rate": 2.3631134388812742e-05, "loss": 0.0015, "step": 41580 }, { "grad_norm": 0.06184295192360878, "learning_rate": 2.3607724402746684e-05, "loss": 0.0017, "step": 41590 }, { "grad_norm": 0.05469321832060814, "learning_rate": 2.35843224339571e-05, "loss": 0.0012, "step": 41600 }, { "grad_norm": 0.04511866718530655, "learning_rate": 2.3560928489552897e-05, "loss": 0.0012, "step": 41610 }, { "grad_norm": 0.05884099379181862, "learning_rate": 2.353754257664053e-05, "loss": 0.0018, "step": 41620 }, { "grad_norm": 0.05596906319260597, "learning_rate": 2.3514164702324037e-05, "loss": 0.002, "step": 41630 }, { "grad_norm": 0.08722042292356491, "learning_rate": 2.3490794873704963e-05, "loss": 0.0015, "step": 41640 }, { "grad_norm": 0.0460403710603714, "learning_rate": 2.3467433097882496e-05, "loss": 0.0013, "step": 41650 }, { "grad_norm": 0.0697915330529213, "learning_rate": 2.34440793819533e-05, "loss": 0.0017, "step": 41660 }, { "grad_norm": 0.05388262867927551, "learning_rate": 2.3420733733011617e-05, "loss": 0.002, "step": 41670 }, { "grad_norm": 0.03623218834400177, "learning_rate": 2.3397396158149243e-05, "loss": 0.0012, "step": 41680 }, { "grad_norm": 0.058994896709918976, "learning_rate": 2.3374066664455498e-05, "loss": 0.002, "step": 41690 }, { "grad_norm": 0.09736678749322891, "learning_rate": 2.3350745259017315e-05, "loss": 0.0019, "step": 41700 }, { "grad_norm": 0.03113873489201069, "learning_rate": 2.332743194891906e-05, "loss": 0.0015, "step": 41710 }, { "grad_norm": 0.06080101430416107, "learning_rate": 2.330412674124276e-05, "loss": 0.0017, "step": 41720 }, { "grad_norm": 0.06240084767341614, "learning_rate": 2.328082964306786e-05, "loss": 0.0019, "step": 41730 }, { "grad_norm": 0.05037274584174156, "learning_rate": 2.325754066147145e-05, "loss": 0.0018, "step": 41740 }, { "grad_norm": 0.03746006637811661, "learning_rate": 2.32342598035281e-05, "loss": 0.0015, "step": 41750 }, { "grad_norm": 0.04217308759689331, "learning_rate": 2.321098707630991e-05, "loss": 0.0016, "step": 41760 }, { "grad_norm": 0.05598224699497223, "learning_rate": 2.318772248688652e-05, "loss": 0.0026, "step": 41770 }, { "grad_norm": 0.06627514958381653, "learning_rate": 2.3164466042325107e-05, "loss": 0.0022, "step": 41780 }, { "grad_norm": 0.04948800802230835, "learning_rate": 2.3141217749690353e-05, "loss": 0.0016, "step": 41790 }, { "grad_norm": 0.06301697343587875, "learning_rate": 2.3117977616044466e-05, "loss": 0.0014, "step": 41800 }, { "grad_norm": 0.037827953696250916, "learning_rate": 2.309474564844722e-05, "loss": 0.0012, "step": 41810 }, { "grad_norm": 0.061289142817258835, "learning_rate": 2.307152185395585e-05, "loss": 0.0011, "step": 41820 }, { "grad_norm": 0.08019273728132248, "learning_rate": 2.3048306239625144e-05, "loss": 0.002, "step": 41830 }, { "grad_norm": 0.0717327892780304, "learning_rate": 2.3025098812507378e-05, "loss": 0.0018, "step": 41840 }, { "grad_norm": 0.0617026686668396, "learning_rate": 2.3001899579652366e-05, "loss": 0.0025, "step": 41850 }, { "grad_norm": 0.0385734848678112, "learning_rate": 2.2978708548107393e-05, "loss": 0.0016, "step": 41860 }, { "grad_norm": 0.03056393750011921, "learning_rate": 2.2955525724917348e-05, "loss": 0.0013, "step": 41870 }, { "grad_norm": 0.045711543411016464, "learning_rate": 2.2932351117124477e-05, "loss": 0.0018, "step": 41880 }, { "grad_norm": 0.05789196863770485, "learning_rate": 2.29091847317687e-05, "loss": 0.0014, "step": 41890 }, { "grad_norm": 0.07339713722467422, "learning_rate": 2.2886026575887277e-05, "loss": 0.0022, "step": 41900 }, { "grad_norm": 0.07295417785644531, "learning_rate": 2.2862876656515094e-05, "loss": 0.0016, "step": 41910 }, { "grad_norm": 0.05247732251882553, "learning_rate": 2.2839734980684464e-05, "loss": 0.0011, "step": 41920 }, { "grad_norm": 0.04409465193748474, "learning_rate": 2.281660155542522e-05, "loss": 0.0017, "step": 41930 }, { "grad_norm": 0.06250740587711334, "learning_rate": 2.279347638776469e-05, "loss": 0.0016, "step": 41940 }, { "grad_norm": 0.06950917094945908, "learning_rate": 2.2770359484727665e-05, "loss": 0.0018, "step": 41950 }, { "grad_norm": 0.06402555853128433, "learning_rate": 2.27472508533365e-05, "loss": 0.0018, "step": 41960 }, { "grad_norm": 0.05865740776062012, "learning_rate": 2.2724150500610948e-05, "loss": 0.0024, "step": 41970 }, { "grad_norm": 0.05431276187300682, "learning_rate": 2.2701058433568302e-05, "loss": 0.0014, "step": 41980 }, { "grad_norm": 0.04078228026628494, "learning_rate": 2.2677974659223318e-05, "loss": 0.0014, "step": 41990 }, { "grad_norm": 0.037460099905729294, "learning_rate": 2.2654899184588235e-05, "loss": 0.0012, "step": 42000 }, { "grad_norm": 0.04277933016419411, "learning_rate": 2.2631832016672756e-05, "loss": 0.0017, "step": 42010 }, { "grad_norm": 0.05132227763533592, "learning_rate": 2.2608773162484127e-05, "loss": 0.0017, "step": 42020 }, { "grad_norm": 0.052710674703121185, "learning_rate": 2.2585722629026958e-05, "loss": 0.0012, "step": 42030 }, { "grad_norm": 0.05393102392554283, "learning_rate": 2.2562680423303457e-05, "loss": 0.0014, "step": 42040 }, { "grad_norm": 0.047110285609960556, "learning_rate": 2.2539646552313165e-05, "loss": 0.0019, "step": 42050 }, { "grad_norm": 0.061021506786346436, "learning_rate": 2.251662102305322e-05, "loss": 0.0021, "step": 42060 }, { "grad_norm": 0.05427911877632141, "learning_rate": 2.2493603842518152e-05, "loss": 0.0029, "step": 42070 }, { "grad_norm": 0.0666823461651802, "learning_rate": 2.2470595017699974e-05, "loss": 0.0017, "step": 42080 }, { "grad_norm": 0.07451517134904861, "learning_rate": 2.244759455558816e-05, "loss": 0.0016, "step": 42090 }, { "grad_norm": 0.032305728644132614, "learning_rate": 2.2424602463169614e-05, "loss": 0.0012, "step": 42100 }, { "grad_norm": 0.05459263175725937, "learning_rate": 2.2401618747428776e-05, "loss": 0.0019, "step": 42110 }, { "grad_norm": 0.050478000193834305, "learning_rate": 2.237864341534747e-05, "loss": 0.0014, "step": 42120 }, { "grad_norm": 0.038751520216464996, "learning_rate": 2.2355676473904998e-05, "loss": 0.0013, "step": 42130 }, { "grad_norm": 0.05812208354473114, "learning_rate": 2.2332717930078108e-05, "loss": 0.0017, "step": 42140 }, { "grad_norm": 0.03959440067410469, "learning_rate": 2.2309767790840992e-05, "loss": 0.0017, "step": 42150 }, { "grad_norm": 0.0315094068646431, "learning_rate": 2.228682606316529e-05, "loss": 0.0014, "step": 42160 }, { "grad_norm": 0.05163620039820671, "learning_rate": 2.2263892754020138e-05, "loss": 0.0021, "step": 42170 }, { "grad_norm": 0.04860633984208107, "learning_rate": 2.2240967870372004e-05, "loss": 0.0015, "step": 42180 }, { "grad_norm": 0.04499819874763489, "learning_rate": 2.2218051419184933e-05, "loss": 0.0015, "step": 42190 }, { "grad_norm": 0.055013179779052734, "learning_rate": 2.219514340742026e-05, "loss": 0.0017, "step": 42200 }, { "grad_norm": 0.0548957996070385, "learning_rate": 2.2172243842036898e-05, "loss": 0.0017, "step": 42210 }, { "grad_norm": 0.03097691386938095, "learning_rate": 2.2149352729991107e-05, "loss": 0.0018, "step": 42220 }, { "grad_norm": 0.0472845733165741, "learning_rate": 2.2126470078236605e-05, "loss": 0.0012, "step": 42230 }, { "grad_norm": 0.06822464615106583, "learning_rate": 2.2103595893724533e-05, "loss": 0.0014, "step": 42240 }, { "grad_norm": 0.0414358451962471, "learning_rate": 2.208073018340345e-05, "loss": 0.0014, "step": 42250 }, { "grad_norm": 0.056209877133369446, "learning_rate": 2.2057872954219405e-05, "loss": 0.002, "step": 42260 }, { "grad_norm": 0.06046012416481972, "learning_rate": 2.203502421311575e-05, "loss": 0.0016, "step": 42270 }, { "grad_norm": 0.05668378248810768, "learning_rate": 2.2012183967033388e-05, "loss": 0.0016, "step": 42280 }, { "grad_norm": 0.04557597637176514, "learning_rate": 2.198935222291056e-05, "loss": 0.0014, "step": 42290 }, { "grad_norm": 0.0687142089009285, "learning_rate": 2.1966528987682948e-05, "loss": 0.0012, "step": 42300 }, { "grad_norm": 0.05834402143955231, "learning_rate": 2.194371426828365e-05, "loss": 0.0018, "step": 42310 }, { "grad_norm": 0.06441590934991837, "learning_rate": 2.192090807164317e-05, "loss": 0.0016, "step": 42320 }, { "grad_norm": 0.05804548040032387, "learning_rate": 2.1898110404689422e-05, "loss": 0.0018, "step": 42330 }, { "grad_norm": 0.06888297200202942, "learning_rate": 2.1875321274347776e-05, "loss": 0.0012, "step": 42340 }, { "grad_norm": 0.06334075331687927, "learning_rate": 2.18525406875409e-05, "loss": 0.0015, "step": 42350 }, { "grad_norm": 0.038360703736543655, "learning_rate": 2.1829768651188997e-05, "loss": 0.0011, "step": 42360 }, { "grad_norm": 0.05173921957612038, "learning_rate": 2.180700517220958e-05, "loss": 0.0022, "step": 42370 }, { "grad_norm": 0.048826683312654495, "learning_rate": 2.1784250257517603e-05, "loss": 0.0017, "step": 42380 }, { "grad_norm": 0.05709327384829521, "learning_rate": 2.1761503914025406e-05, "loss": 0.0015, "step": 42390 }, { "grad_norm": 0.04427650570869446, "learning_rate": 2.1738766148642705e-05, "loss": 0.0015, "step": 42400 }, { "grad_norm": 0.06401632726192474, "learning_rate": 2.1716036968276683e-05, "loss": 0.0016, "step": 42410 }, { "grad_norm": 0.05590800940990448, "learning_rate": 2.1693316379831808e-05, "loss": 0.0016, "step": 42420 }, { "grad_norm": 0.053608499467372894, "learning_rate": 2.1670604390210037e-05, "loss": 0.0021, "step": 42430 }, { "grad_norm": 0.059733130037784576, "learning_rate": 2.1647901006310656e-05, "loss": 0.0018, "step": 42440 }, { "grad_norm": 0.029863528907299042, "learning_rate": 2.1625206235030353e-05, "loss": 0.0018, "step": 42450 }, { "grad_norm": 0.0433429591357708, "learning_rate": 2.160252008326321e-05, "loss": 0.0013, "step": 42460 }, { "grad_norm": 0.04503430798649788, "learning_rate": 2.157984255790067e-05, "loss": 0.0018, "step": 42470 }, { "grad_norm": 0.04554833471775055, "learning_rate": 2.1557173665831553e-05, "loss": 0.0015, "step": 42480 }, { "grad_norm": 0.09277554601430893, "learning_rate": 2.153451341394212e-05, "loss": 0.0017, "step": 42490 }, { "grad_norm": 0.0558389350771904, "learning_rate": 2.151186180911589e-05, "loss": 0.002, "step": 42500 }, { "grad_norm": 0.06661283224821091, "learning_rate": 2.1489218858233877e-05, "loss": 0.0014, "step": 42510 }, { "grad_norm": 0.03987618535757065, "learning_rate": 2.1466584568174392e-05, "loss": 0.0025, "step": 42520 }, { "grad_norm": 0.06726425886154175, "learning_rate": 2.1443958945813132e-05, "loss": 0.002, "step": 42530 }, { "grad_norm": 0.06597264111042023, "learning_rate": 2.1421341998023163e-05, "loss": 0.002, "step": 42540 }, { "grad_norm": 0.06404577940702438, "learning_rate": 2.139873373167491e-05, "loss": 0.0018, "step": 42550 }, { "grad_norm": 0.08023608475923538, "learning_rate": 2.13761341536362e-05, "loss": 0.002, "step": 42560 }, { "grad_norm": 0.07331794500350952, "learning_rate": 2.1353543270772136e-05, "loss": 0.0017, "step": 42570 }, { "grad_norm": 0.06782885640859604, "learning_rate": 2.1330961089945297e-05, "loss": 0.0017, "step": 42580 }, { "grad_norm": 0.060207072645425797, "learning_rate": 2.130838761801548e-05, "loss": 0.0022, "step": 42590 }, { "grad_norm": 0.05373797565698624, "learning_rate": 2.1285822861839966e-05, "loss": 0.0014, "step": 42600 }, { "grad_norm": 0.05066496878862381, "learning_rate": 2.126326682827331e-05, "loss": 0.0016, "step": 42610 }, { "grad_norm": 0.026597006246447563, "learning_rate": 2.124071952416744e-05, "loss": 0.0014, "step": 42620 }, { "grad_norm": 0.05020894110202789, "learning_rate": 2.1218180956371634e-05, "loss": 0.0019, "step": 42630 }, { "grad_norm": 0.058968376368284225, "learning_rate": 2.119565113173252e-05, "loss": 0.0018, "step": 42640 }, { "grad_norm": 0.04242433235049248, "learning_rate": 2.1173130057094033e-05, "loss": 0.0011, "step": 42650 }, { "grad_norm": 0.04423849284648895, "learning_rate": 2.115061773929753e-05, "loss": 0.002, "step": 42660 }, { "grad_norm": 0.052273888140916824, "learning_rate": 2.1128114185181623e-05, "loss": 0.0017, "step": 42670 }, { "grad_norm": 0.03988956660032272, "learning_rate": 2.1105619401582317e-05, "loss": 0.0017, "step": 42680 }, { "grad_norm": 0.05623282119631767, "learning_rate": 2.1083133395332928e-05, "loss": 0.0022, "step": 42690 }, { "grad_norm": 0.043363094329833984, "learning_rate": 2.1060656173264082e-05, "loss": 0.0014, "step": 42700 }, { "grad_norm": 0.0504370853304863, "learning_rate": 2.103818774220383e-05, "loss": 0.0014, "step": 42710 }, { "grad_norm": 0.05103372782468796, "learning_rate": 2.1015728108977412e-05, "loss": 0.0016, "step": 42720 }, { "grad_norm": 0.054425209760665894, "learning_rate": 2.0993277280407548e-05, "loss": 0.0016, "step": 42730 }, { "grad_norm": 0.0391816720366478, "learning_rate": 2.0970835263314132e-05, "loss": 0.0016, "step": 42740 }, { "grad_norm": 0.048368293792009354, "learning_rate": 2.094840206451451e-05, "loss": 0.0025, "step": 42750 }, { "grad_norm": 0.05216412618756294, "learning_rate": 2.0925977690823273e-05, "loss": 0.0015, "step": 42760 }, { "grad_norm": 0.06558272987604141, "learning_rate": 2.0903562149052364e-05, "loss": 0.002, "step": 42770 }, { "grad_norm": 0.0651988759636879, "learning_rate": 2.0881155446011025e-05, "loss": 0.0025, "step": 42780 }, { "grad_norm": 0.09793414920568466, "learning_rate": 2.0858757588505823e-05, "loss": 0.0019, "step": 42790 }, { "grad_norm": 0.06635548174381256, "learning_rate": 2.0836368583340622e-05, "loss": 0.0014, "step": 42800 }, { "grad_norm": 0.06899050623178482, "learning_rate": 2.081398843731664e-05, "loss": 0.0015, "step": 42810 }, { "grad_norm": 0.05967060849070549, "learning_rate": 2.0791617157232357e-05, "loss": 0.0015, "step": 42820 }, { "grad_norm": 0.05495849624276161, "learning_rate": 2.0769254749883576e-05, "loss": 0.0013, "step": 42830 }, { "grad_norm": 0.04277471452951431, "learning_rate": 2.0746901222063415e-05, "loss": 0.0016, "step": 42840 }, { "grad_norm": 0.06765001267194748, "learning_rate": 2.072455658056226e-05, "loss": 0.002, "step": 42850 }, { "grad_norm": 0.047775544226169586, "learning_rate": 2.0702220832167873e-05, "loss": 0.0014, "step": 42860 }, { "grad_norm": 0.039600130170583725, "learning_rate": 2.0679893983665205e-05, "loss": 0.0014, "step": 42870 }, { "grad_norm": 0.03974800184369087, "learning_rate": 2.0657576041836622e-05, "loss": 0.0013, "step": 42880 }, { "grad_norm": 0.059079334139823914, "learning_rate": 2.0635267013461666e-05, "loss": 0.0018, "step": 42890 }, { "grad_norm": 0.05326703190803528, "learning_rate": 2.061296690531728e-05, "loss": 0.0015, "step": 42900 }, { "grad_norm": 0.06002864986658096, "learning_rate": 2.0590675724177622e-05, "loss": 0.0013, "step": 42910 }, { "grad_norm": 0.04325176775455475, "learning_rate": 2.0568393476814167e-05, "loss": 0.0027, "step": 42920 }, { "grad_norm": 0.034511543810367584, "learning_rate": 2.0546120169995685e-05, "loss": 0.0015, "step": 42930 }, { "grad_norm": 0.029956893995404243, "learning_rate": 2.0523855810488214e-05, "loss": 0.0012, "step": 42940 }, { "grad_norm": 0.041956327855587006, "learning_rate": 2.050160040505505e-05, "loss": 0.0012, "step": 42950 }, { "grad_norm": 0.04809104651212692, "learning_rate": 2.0479353960456843e-05, "loss": 0.0014, "step": 42960 }, { "grad_norm": 0.04646660387516022, "learning_rate": 2.0457116483451456e-05, "loss": 0.0015, "step": 42970 }, { "grad_norm": 0.033656295388936996, "learning_rate": 2.0434887980794043e-05, "loss": 0.0013, "step": 42980 }, { "grad_norm": 0.05332230404019356, "learning_rate": 2.0412668459237043e-05, "loss": 0.0013, "step": 42990 }, { "grad_norm": 0.03570106253027916, "learning_rate": 2.039045792553016e-05, "loss": 0.0014, "step": 43000 }, { "grad_norm": 0.057831231504678726, "learning_rate": 2.036825638642036e-05, "loss": 0.0016, "step": 43010 }, { "grad_norm": 0.05907837301492691, "learning_rate": 2.0346063848651868e-05, "loss": 0.0017, "step": 43020 }, { "grad_norm": 0.07320616394281387, "learning_rate": 2.0323880318966254e-05, "loss": 0.0015, "step": 43030 }, { "grad_norm": 0.06848009675741196, "learning_rate": 2.030170580410221e-05, "loss": 0.002, "step": 43040 }, { "grad_norm": 0.06016403064131737, "learning_rate": 2.0279540310795837e-05, "loss": 0.0016, "step": 43050 }, { "grad_norm": 0.055401600897312164, "learning_rate": 2.0257383845780365e-05, "loss": 0.0015, "step": 43060 }, { "grad_norm": 0.07272762060165405, "learning_rate": 2.0235236415786384e-05, "loss": 0.0018, "step": 43070 }, { "grad_norm": 0.04297976940870285, "learning_rate": 2.021309802754169e-05, "loss": 0.0016, "step": 43080 }, { "grad_norm": 0.05516471341252327, "learning_rate": 2.0190968687771332e-05, "loss": 0.0022, "step": 43090 }, { "grad_norm": 0.04382308945059776, "learning_rate": 2.016884840319763e-05, "loss": 0.0013, "step": 43100 }, { "grad_norm": 0.034641753882169724, "learning_rate": 2.0146737180540122e-05, "loss": 0.0011, "step": 43110 }, { "grad_norm": 0.06878966838121414, "learning_rate": 2.012463502651564e-05, "loss": 0.0014, "step": 43120 }, { "grad_norm": 0.048077113926410675, "learning_rate": 2.0102541947838228e-05, "loss": 0.0015, "step": 43130 }, { "grad_norm": 0.045438215136528015, "learning_rate": 2.0080457951219173e-05, "loss": 0.0014, "step": 43140 }, { "grad_norm": 0.05863852798938751, "learning_rate": 2.0058383043367017e-05, "loss": 0.0017, "step": 43150 }, { "grad_norm": 0.06276649981737137, "learning_rate": 2.0036317230987528e-05, "loss": 0.0017, "step": 43160 }, { "grad_norm": 0.053775396198034286, "learning_rate": 2.0014260520783696e-05, "loss": 0.0025, "step": 43170 }, { "grad_norm": 0.053048137575387955, "learning_rate": 1.9992212919455834e-05, "loss": 0.0014, "step": 43180 }, { "grad_norm": 0.06881546229124069, "learning_rate": 1.9970174433701333e-05, "loss": 0.0013, "step": 43190 }, { "grad_norm": 0.05208338424563408, "learning_rate": 1.9948145070214992e-05, "loss": 0.0016, "step": 43200 }, { "grad_norm": 0.06795935332775116, "learning_rate": 1.9926124835688663e-05, "loss": 0.0014, "step": 43210 }, { "grad_norm": 0.06587838381528854, "learning_rate": 1.9904113736811576e-05, "loss": 0.0019, "step": 43220 }, { "grad_norm": 0.07632599771022797, "learning_rate": 1.9882111780270096e-05, "loss": 0.0022, "step": 43230 }, { "grad_norm": 0.06185325235128403, "learning_rate": 1.986011897274784e-05, "loss": 0.0019, "step": 43240 }, { "grad_norm": 0.04949791356921196, "learning_rate": 1.983813532092565e-05, "loss": 0.0015, "step": 43250 }, { "grad_norm": 0.0555272214114666, "learning_rate": 1.981616083148155e-05, "loss": 0.0013, "step": 43260 }, { "grad_norm": 0.04147817939519882, "learning_rate": 1.9794195511090845e-05, "loss": 0.0019, "step": 43270 }, { "grad_norm": 0.0420994833111763, "learning_rate": 1.977223936642601e-05, "loss": 0.0019, "step": 43280 }, { "grad_norm": 0.058428529649972916, "learning_rate": 1.975029240415674e-05, "loss": 0.0012, "step": 43290 }, { "grad_norm": 0.058280277997255325, "learning_rate": 1.9728354630949936e-05, "loss": 0.0014, "step": 43300 }, { "grad_norm": 0.03056374378502369, "learning_rate": 1.9706426053469716e-05, "loss": 0.0013, "step": 43310 }, { "grad_norm": 0.04887150228023529, "learning_rate": 1.9684506678377396e-05, "loss": 0.0017, "step": 43320 }, { "grad_norm": 0.05616060271859169, "learning_rate": 1.9662596512331544e-05, "loss": 0.0015, "step": 43330 }, { "grad_norm": 0.06050432473421097, "learning_rate": 1.964069556198782e-05, "loss": 0.0015, "step": 43340 }, { "grad_norm": 0.03469546139240265, "learning_rate": 1.9618803833999232e-05, "loss": 0.0016, "step": 43350 }, { "grad_norm": 0.05097326263785362, "learning_rate": 1.9596921335015838e-05, "loss": 0.0015, "step": 43360 }, { "grad_norm": 0.04482686147093773, "learning_rate": 1.957504807168501e-05, "loss": 0.002, "step": 43370 }, { "grad_norm": 0.06075972691178322, "learning_rate": 1.9553184050651253e-05, "loss": 0.0018, "step": 43380 }, { "grad_norm": 0.05438069999217987, "learning_rate": 1.953132927855628e-05, "loss": 0.0014, "step": 43390 }, { "grad_norm": 0.052846550941467285, "learning_rate": 1.9509483762038995e-05, "loss": 0.0011, "step": 43400 }, { "grad_norm": 0.05922212079167366, "learning_rate": 1.9487647507735467e-05, "loss": 0.0014, "step": 43410 }, { "grad_norm": 0.0487179234623909, "learning_rate": 1.9465820522279032e-05, "loss": 0.0017, "step": 43420 }, { "grad_norm": 0.06476166099309921, "learning_rate": 1.9444002812300078e-05, "loss": 0.0013, "step": 43430 }, { "grad_norm": 0.0543285496532917, "learning_rate": 1.94221943844263e-05, "loss": 0.0015, "step": 43440 }, { "grad_norm": 0.043457988649606705, "learning_rate": 1.9400395245282515e-05, "loss": 0.0014, "step": 43450 }, { "grad_norm": 0.051995571702718735, "learning_rate": 1.937860540149071e-05, "loss": 0.0013, "step": 43460 }, { "grad_norm": 0.03030664287507534, "learning_rate": 1.9356824859670082e-05, "loss": 0.0012, "step": 43470 }, { "grad_norm": 0.03477759659290314, "learning_rate": 1.9335053626436967e-05, "loss": 0.0015, "step": 43480 }, { "grad_norm": 0.048272937536239624, "learning_rate": 1.9313291708404885e-05, "loss": 0.0015, "step": 43490 }, { "grad_norm": 0.04313355311751366, "learning_rate": 1.9291539112184587e-05, "loss": 0.0014, "step": 43500 }, { "grad_norm": 0.058301590383052826, "learning_rate": 1.9269795844383854e-05, "loss": 0.0016, "step": 43510 }, { "grad_norm": 0.0631060004234314, "learning_rate": 1.9248061911607777e-05, "loss": 0.0014, "step": 43520 }, { "grad_norm": 0.04892804101109505, "learning_rate": 1.9226337320458538e-05, "loss": 0.0014, "step": 43530 }, { "grad_norm": 0.051630668342113495, "learning_rate": 1.9204622077535488e-05, "loss": 0.0014, "step": 43540 }, { "grad_norm": 0.04802330583333969, "learning_rate": 1.9182916189435147e-05, "loss": 0.0019, "step": 43550 }, { "grad_norm": 0.08272916823625565, "learning_rate": 1.916121966275117e-05, "loss": 0.0029, "step": 43560 }, { "grad_norm": 0.07399989664554596, "learning_rate": 1.9139532504074443e-05, "loss": 0.0017, "step": 43570 }, { "grad_norm": 0.05480703338980675, "learning_rate": 1.9117854719992885e-05, "loss": 0.0014, "step": 43580 }, { "grad_norm": 0.03880338370800018, "learning_rate": 1.9096186317091687e-05, "loss": 0.0018, "step": 43590 }, { "grad_norm": 0.0505962111055851, "learning_rate": 1.9074527301953116e-05, "loss": 0.0024, "step": 43600 }, { "grad_norm": 0.05464589223265648, "learning_rate": 1.9052877681156607e-05, "loss": 0.0019, "step": 43610 }, { "grad_norm": 0.06804168969392776, "learning_rate": 1.903123746127875e-05, "loss": 0.0018, "step": 43620 }, { "grad_norm": 0.092018261551857, "learning_rate": 1.900960664889327e-05, "loss": 0.002, "step": 43630 }, { "grad_norm": 0.042749207466840744, "learning_rate": 1.8987985250571015e-05, "loss": 0.0013, "step": 43640 }, { "grad_norm": 0.05340585857629776, "learning_rate": 1.8966373272880054e-05, "loss": 0.0019, "step": 43650 }, { "grad_norm": 0.04982144385576248, "learning_rate": 1.8944770722385462e-05, "loss": 0.0018, "step": 43660 }, { "grad_norm": 0.044676680117845535, "learning_rate": 1.8923177605649576e-05, "loss": 0.0011, "step": 43670 }, { "grad_norm": 0.04759863391518593, "learning_rate": 1.8901593929231802e-05, "loss": 0.0014, "step": 43680 }, { "grad_norm": 0.08208536356687546, "learning_rate": 1.8880019699688684e-05, "loss": 0.0019, "step": 43690 }, { "grad_norm": 0.05547303706407547, "learning_rate": 1.8858454923573904e-05, "loss": 0.0015, "step": 43700 }, { "grad_norm": 0.04405943304300308, "learning_rate": 1.8836899607438253e-05, "loss": 0.0013, "step": 43710 }, { "grad_norm": 0.043257709592580795, "learning_rate": 1.8815353757829723e-05, "loss": 0.0013, "step": 43720 }, { "grad_norm": 0.0464746356010437, "learning_rate": 1.879381738129331e-05, "loss": 0.0013, "step": 43730 }, { "grad_norm": 0.046552401036024094, "learning_rate": 1.8772290484371236e-05, "loss": 0.0012, "step": 43740 }, { "grad_norm": 0.0556759312748909, "learning_rate": 1.8750773073602795e-05, "loss": 0.0017, "step": 43750 }, { "grad_norm": 0.07073306292295456, "learning_rate": 1.8729265155524405e-05, "loss": 0.0018, "step": 43760 }, { "grad_norm": 0.04355809465050697, "learning_rate": 1.8707766736669607e-05, "loss": 0.0014, "step": 43770 }, { "grad_norm": 0.07426019012928009, "learning_rate": 1.8686277823569055e-05, "loss": 0.0014, "step": 43780 }, { "grad_norm": 0.06713676452636719, "learning_rate": 1.8664798422750484e-05, "loss": 0.0011, "step": 43790 }, { "grad_norm": 0.05537419766187668, "learning_rate": 1.8643328540738832e-05, "loss": 0.0022, "step": 43800 }, { "grad_norm": 0.05494308099150658, "learning_rate": 1.862186818405601e-05, "loss": 0.0018, "step": 43810 }, { "grad_norm": 0.060569100081920624, "learning_rate": 1.8600417359221156e-05, "loss": 0.0012, "step": 43820 }, { "grad_norm": 0.06404769420623779, "learning_rate": 1.8578976072750454e-05, "loss": 0.0012, "step": 43830 }, { "grad_norm": 0.04855360463261604, "learning_rate": 1.8557544331157194e-05, "loss": 0.0016, "step": 43840 }, { "grad_norm": 0.04521453008055687, "learning_rate": 1.8536122140951785e-05, "loss": 0.0016, "step": 43850 }, { "grad_norm": 0.04312565550208092, "learning_rate": 1.8514709508641688e-05, "loss": 0.0014, "step": 43860 }, { "grad_norm": 0.04760471358895302, "learning_rate": 1.8493306440731555e-05, "loss": 0.0013, "step": 43870 }, { "grad_norm": 0.05628479644656181, "learning_rate": 1.8471912943723013e-05, "loss": 0.0016, "step": 43880 }, { "grad_norm": 0.05581070855259895, "learning_rate": 1.8450529024114894e-05, "loss": 0.0018, "step": 43890 }, { "grad_norm": 0.036357346922159195, "learning_rate": 1.842915468840301e-05, "loss": 0.0013, "step": 43900 }, { "grad_norm": 0.06629229336977005, "learning_rate": 1.840778994308037e-05, "loss": 0.0019, "step": 43910 }, { "grad_norm": 0.0395112968981266, "learning_rate": 1.8386434794637004e-05, "loss": 0.0014, "step": 43920 }, { "grad_norm": 0.07718892395496368, "learning_rate": 1.8365089249560034e-05, "loss": 0.0015, "step": 43930 }, { "grad_norm": 0.03943856060504913, "learning_rate": 1.8343753314333683e-05, "loss": 0.0014, "step": 43940 }, { "grad_norm": 0.03800894692540169, "learning_rate": 1.8322426995439236e-05, "loss": 0.0013, "step": 43950 }, { "grad_norm": 0.034817714244127274, "learning_rate": 1.8301110299355058e-05, "loss": 0.0014, "step": 43960 }, { "grad_norm": 0.03932677209377289, "learning_rate": 1.8279803232556625e-05, "loss": 0.0013, "step": 43970 }, { "grad_norm": 0.05783187225461006, "learning_rate": 1.8258505801516444e-05, "loss": 0.0015, "step": 43980 }, { "grad_norm": 0.0830911248922348, "learning_rate": 1.8237218012704117e-05, "loss": 0.0016, "step": 43990 }, { "grad_norm": 0.050481513142585754, "learning_rate": 1.821593987258631e-05, "loss": 0.0012, "step": 44000 }, { "grad_norm": 0.040149837732315063, "learning_rate": 1.8194671387626744e-05, "loss": 0.0012, "step": 44010 }, { "grad_norm": 0.07966675609350204, "learning_rate": 1.8173412564286276e-05, "loss": 0.0013, "step": 44020 }, { "grad_norm": 0.03603832423686981, "learning_rate": 1.8152163409022697e-05, "loss": 0.0015, "step": 44030 }, { "grad_norm": 0.04743272438645363, "learning_rate": 1.8130923928291023e-05, "loss": 0.0014, "step": 44040 }, { "grad_norm": 0.064693383872509, "learning_rate": 1.8109694128543163e-05, "loss": 0.0018, "step": 44050 }, { "grad_norm": 0.050035055726766586, "learning_rate": 1.8088474016228237e-05, "loss": 0.0016, "step": 44060 }, { "grad_norm": 0.046434421092271805, "learning_rate": 1.8067263597792328e-05, "loss": 0.0014, "step": 44070 }, { "grad_norm": 0.0469338521361351, "learning_rate": 1.80460628796786e-05, "loss": 0.0015, "step": 44080 }, { "grad_norm": 0.062471430748701096, "learning_rate": 1.8024871868327276e-05, "loss": 0.0019, "step": 44090 }, { "grad_norm": 0.06406975537538528, "learning_rate": 1.8003690570175608e-05, "loss": 0.0016, "step": 44100 }, { "grad_norm": 0.05420534685254097, "learning_rate": 1.7982518991657943e-05, "loss": 0.0021, "step": 44110 }, { "grad_norm": 0.048343926668167114, "learning_rate": 1.7961357139205643e-05, "loss": 0.0015, "step": 44120 }, { "grad_norm": 0.057103291153907776, "learning_rate": 1.7940205019247108e-05, "loss": 0.0014, "step": 44130 }, { "grad_norm": 0.03739093989133835, "learning_rate": 1.79190626382078e-05, "loss": 0.0012, "step": 44140 }, { "grad_norm": 0.04894346743822098, "learning_rate": 1.7897930002510215e-05, "loss": 0.002, "step": 44150 }, { "grad_norm": 0.07559125870466232, "learning_rate": 1.787680711857387e-05, "loss": 0.0022, "step": 44160 }, { "grad_norm": 0.05241446942090988, "learning_rate": 1.7855693992815398e-05, "loss": 0.0022, "step": 44170 }, { "grad_norm": 0.057074662297964096, "learning_rate": 1.7834590631648328e-05, "loss": 0.0018, "step": 44180 }, { "grad_norm": 0.037474773824214935, "learning_rate": 1.7813497041483384e-05, "loss": 0.0012, "step": 44190 }, { "grad_norm": 0.055868227034807205, "learning_rate": 1.779241322872817e-05, "loss": 0.0013, "step": 44200 }, { "grad_norm": 0.030964579433202744, "learning_rate": 1.777133919978744e-05, "loss": 0.0022, "step": 44210 }, { "grad_norm": 0.03600997477769852, "learning_rate": 1.7750274961062912e-05, "loss": 0.0016, "step": 44220 }, { "grad_norm": 0.03979450836777687, "learning_rate": 1.772922051895335e-05, "loss": 0.0018, "step": 44230 }, { "grad_norm": 0.08725495636463165, "learning_rate": 1.770817587985453e-05, "loss": 0.0024, "step": 44240 }, { "grad_norm": 0.07601011544466019, "learning_rate": 1.7687141050159246e-05, "loss": 0.0019, "step": 44250 }, { "grad_norm": 0.06075899302959442, "learning_rate": 1.7666116036257375e-05, "loss": 0.0018, "step": 44260 }, { "grad_norm": 0.0608060248196125, "learning_rate": 1.764510084453569e-05, "loss": 0.0024, "step": 44270 }, { "grad_norm": 0.07930360734462738, "learning_rate": 1.76240954813781e-05, "loss": 0.0018, "step": 44280 }, { "grad_norm": 0.051511697471141815, "learning_rate": 1.7603099953165476e-05, "loss": 0.0014, "step": 44290 }, { "grad_norm": 0.03521774336695671, "learning_rate": 1.7582114266275683e-05, "loss": 0.0012, "step": 44300 }, { "grad_norm": 0.06560646742582321, "learning_rate": 1.756113842708364e-05, "loss": 0.0011, "step": 44310 }, { "grad_norm": 0.04782005026936531, "learning_rate": 1.7540172441961245e-05, "loss": 0.0014, "step": 44320 }, { "grad_norm": 0.03975570946931839, "learning_rate": 1.7519216317277387e-05, "loss": 0.0015, "step": 44330 }, { "grad_norm": 0.06587781757116318, "learning_rate": 1.7498270059398046e-05, "loss": 0.0014, "step": 44340 }, { "grad_norm": 0.04944848641753197, "learning_rate": 1.7477333674686062e-05, "loss": 0.0013, "step": 44350 }, { "grad_norm": 0.0730154886841774, "learning_rate": 1.745640716950142e-05, "loss": 0.0017, "step": 44360 }, { "grad_norm": 0.0539729967713356, "learning_rate": 1.7435490550201017e-05, "loss": 0.0012, "step": 44370 }, { "grad_norm": 0.033096011728048325, "learning_rate": 1.7414583823138762e-05, "loss": 0.0012, "step": 44380 }, { "grad_norm": 0.048359908163547516, "learning_rate": 1.739368699466558e-05, "loss": 0.0013, "step": 44390 }, { "grad_norm": 0.04085792228579521, "learning_rate": 1.737280007112935e-05, "loss": 0.0013, "step": 44400 }, { "grad_norm": 0.04881393164396286, "learning_rate": 1.735192305887502e-05, "loss": 0.0013, "step": 44410 }, { "grad_norm": 0.06143840029835701, "learning_rate": 1.733105596424441e-05, "loss": 0.0019, "step": 44420 }, { "grad_norm": 0.04995344951748848, "learning_rate": 1.7310198793576437e-05, "loss": 0.0014, "step": 44430 }, { "grad_norm": 0.05436168238520622, "learning_rate": 1.7289351553206952e-05, "loss": 0.0016, "step": 44440 }, { "grad_norm": 0.05552405118942261, "learning_rate": 1.7268514249468788e-05, "loss": 0.0018, "step": 44450 }, { "grad_norm": 0.05413612723350525, "learning_rate": 1.7247686888691765e-05, "loss": 0.0013, "step": 44460 }, { "grad_norm": 0.042587511241436005, "learning_rate": 1.7226869477202694e-05, "loss": 0.0013, "step": 44470 }, { "grad_norm": 0.03774147108197212, "learning_rate": 1.7206062021325336e-05, "loss": 0.0013, "step": 44480 }, { "grad_norm": 0.030787702649831772, "learning_rate": 1.7185264527380502e-05, "loss": 0.001, "step": 44490 }, { "grad_norm": 0.06088605895638466, "learning_rate": 1.716447700168584e-05, "loss": 0.0014, "step": 44500 }, { "grad_norm": 0.03890411555767059, "learning_rate": 1.714369945055611e-05, "loss": 0.0011, "step": 44510 }, { "grad_norm": 0.03421345353126526, "learning_rate": 1.7122931880302968e-05, "loss": 0.0011, "step": 44520 }, { "grad_norm": 0.06340976059436798, "learning_rate": 1.710217429723505e-05, "loss": 0.002, "step": 44530 }, { "grad_norm": 0.04551355168223381, "learning_rate": 1.7081426707657972e-05, "loss": 0.0014, "step": 44540 }, { "grad_norm": 0.07132561504840851, "learning_rate": 1.7060689117874275e-05, "loss": 0.0018, "step": 44550 }, { "grad_norm": 0.04365462809801102, "learning_rate": 1.703996153418354e-05, "loss": 0.0017, "step": 44560 }, { "grad_norm": 0.03989589959383011, "learning_rate": 1.7019243962882205e-05, "loss": 0.002, "step": 44570 }, { "grad_norm": 0.04933281987905502, "learning_rate": 1.6998536410263754e-05, "loss": 0.0015, "step": 44580 }, { "grad_norm": 0.039225827902555466, "learning_rate": 1.6977838882618596e-05, "loss": 0.0012, "step": 44590 }, { "grad_norm": 0.03475998342037201, "learning_rate": 1.6957151386234088e-05, "loss": 0.0012, "step": 44600 }, { "grad_norm": 0.03961825370788574, "learning_rate": 1.6936473927394536e-05, "loss": 0.0016, "step": 44610 }, { "grad_norm": 0.04835716262459755, "learning_rate": 1.6915806512381222e-05, "loss": 0.0014, "step": 44620 }, { "grad_norm": 0.05996303632855415, "learning_rate": 1.6895149147472344e-05, "loss": 0.0014, "step": 44630 }, { "grad_norm": 0.05709138140082359, "learning_rate": 1.6874501838943073e-05, "loss": 0.0009, "step": 44640 }, { "grad_norm": 0.05026773363351822, "learning_rate": 1.6853864593065506e-05, "loss": 0.0016, "step": 44650 }, { "grad_norm": 0.04397597163915634, "learning_rate": 1.683323741610871e-05, "loss": 0.0014, "step": 44660 }, { "grad_norm": 0.06328008323907852, "learning_rate": 1.6812620314338674e-05, "loss": 0.0013, "step": 44670 }, { "grad_norm": 0.07102672755718231, "learning_rate": 1.6792013294018326e-05, "loss": 0.0023, "step": 44680 }, { "grad_norm": 0.06523797661066055, "learning_rate": 1.6771416361407526e-05, "loss": 0.0021, "step": 44690 }, { "grad_norm": 0.05692075565457344, "learning_rate": 1.675082952276308e-05, "loss": 0.0017, "step": 44700 }, { "grad_norm": 0.04508525878190994, "learning_rate": 1.6730252784338757e-05, "loss": 0.0012, "step": 44710 }, { "grad_norm": 0.04908231273293495, "learning_rate": 1.6709686152385166e-05, "loss": 0.002, "step": 44720 }, { "grad_norm": 0.07249220460653305, "learning_rate": 1.668912963314998e-05, "loss": 0.0017, "step": 44730 }, { "grad_norm": 0.050883881747722626, "learning_rate": 1.6668583232877653e-05, "loss": 0.0012, "step": 44740 }, { "grad_norm": 0.0714581161737442, "learning_rate": 1.6648046957809698e-05, "loss": 0.0014, "step": 44750 }, { "grad_norm": 0.05166921392083168, "learning_rate": 1.6627520814184462e-05, "loss": 0.0011, "step": 44760 }, { "grad_norm": 0.0776660218834877, "learning_rate": 1.660700480823726e-05, "loss": 0.0014, "step": 44770 }, { "grad_norm": 0.06504262983798981, "learning_rate": 1.65864989462003e-05, "loss": 0.0012, "step": 44780 }, { "grad_norm": 0.03220699727535248, "learning_rate": 1.656600323430273e-05, "loss": 0.001, "step": 44790 }, { "grad_norm": 0.05638829618692398, "learning_rate": 1.654551767877059e-05, "loss": 0.0011, "step": 44800 }, { "grad_norm": 0.052750322967767715, "learning_rate": 1.6525042285826874e-05, "loss": 0.0024, "step": 44810 }, { "grad_norm": 0.041679754853248596, "learning_rate": 1.6504577061691468e-05, "loss": 0.0013, "step": 44820 }, { "grad_norm": 0.028467386960983276, "learning_rate": 1.6484122012581143e-05, "loss": 0.0011, "step": 44830 }, { "grad_norm": 0.04250194877386093, "learning_rate": 1.6463677144709623e-05, "loss": 0.0013, "step": 44840 }, { "grad_norm": 0.036811571568250656, "learning_rate": 1.6443242464287493e-05, "loss": 0.0014, "step": 44850 }, { "grad_norm": 0.04048433527350426, "learning_rate": 1.642281797752232e-05, "loss": 0.0012, "step": 44860 }, { "grad_norm": 0.04514390975236893, "learning_rate": 1.6402403690618456e-05, "loss": 0.0009, "step": 44870 }, { "grad_norm": 0.04887697100639343, "learning_rate": 1.6381999609777295e-05, "loss": 0.0012, "step": 44880 }, { "grad_norm": 0.03904722258448601, "learning_rate": 1.6361605741196983e-05, "loss": 0.0013, "step": 44890 }, { "grad_norm": 0.05504129081964493, "learning_rate": 1.63412220910727e-05, "loss": 0.0016, "step": 44900 }, { "grad_norm": 0.04917781427502632, "learning_rate": 1.6320848665596433e-05, "loss": 0.0014, "step": 44910 }, { "grad_norm": 0.050548143684864044, "learning_rate": 1.6300485470957095e-05, "loss": 0.0018, "step": 44920 }, { "grad_norm": 0.09167925268411636, "learning_rate": 1.6280132513340483e-05, "loss": 0.0015, "step": 44930 }, { "grad_norm": 0.06670098751783371, "learning_rate": 1.62597897989293e-05, "loss": 0.0018, "step": 44940 }, { "grad_norm": 0.057094063609838486, "learning_rate": 1.623945733390309e-05, "loss": 0.0016, "step": 44950 }, { "grad_norm": 0.027947595342993736, "learning_rate": 1.6219135124438374e-05, "loss": 0.0011, "step": 44960 }, { "grad_norm": 0.0545012392103672, "learning_rate": 1.6198823176708465e-05, "loss": 0.0012, "step": 44970 }, { "grad_norm": 0.04103050380945206, "learning_rate": 1.6178521496883613e-05, "loss": 0.0014, "step": 44980 }, { "grad_norm": 0.03654343634843826, "learning_rate": 1.6158230091130926e-05, "loss": 0.0017, "step": 44990 }, { "grad_norm": 0.05494898557662964, "learning_rate": 1.613794896561438e-05, "loss": 0.0016, "step": 45000 }, { "grad_norm": 0.037650592625141144, "learning_rate": 1.6117678126494894e-05, "loss": 0.0023, "step": 45010 }, { "grad_norm": 0.05370176583528519, "learning_rate": 1.6097417579930153e-05, "loss": 0.0021, "step": 45020 }, { "grad_norm": 0.03816768527030945, "learning_rate": 1.6077167332074834e-05, "loss": 0.0015, "step": 45030 }, { "grad_norm": 0.043679580092430115, "learning_rate": 1.605692738908037e-05, "loss": 0.002, "step": 45040 }, { "grad_norm": 0.05782396346330643, "learning_rate": 1.6036697757095176e-05, "loss": 0.0013, "step": 45050 }, { "grad_norm": 0.039250925183296204, "learning_rate": 1.6016478442264428e-05, "loss": 0.0014, "step": 45060 }, { "grad_norm": 0.06256868690252304, "learning_rate": 1.599626945073026e-05, "loss": 0.0018, "step": 45070 }, { "grad_norm": 0.053174037486314774, "learning_rate": 1.597607078863162e-05, "loss": 0.0016, "step": 45080 }, { "grad_norm": 0.040481556206941605, "learning_rate": 1.595588246210432e-05, "loss": 0.0014, "step": 45090 }, { "grad_norm": 0.054496049880981445, "learning_rate": 1.5935704477281048e-05, "loss": 0.0013, "step": 45100 }, { "grad_norm": 0.04602828994393349, "learning_rate": 1.5915536840291323e-05, "loss": 0.0015, "step": 45110 }, { "grad_norm": 0.06190558150410652, "learning_rate": 1.5895379557261576e-05, "loss": 0.0017, "step": 45120 }, { "grad_norm": 0.04892246052622795, "learning_rate": 1.5875232634315033e-05, "loss": 0.0015, "step": 45130 }, { "grad_norm": 0.05096694454550743, "learning_rate": 1.5855096077571812e-05, "loss": 0.0021, "step": 45140 }, { "grad_norm": 0.036368709057569504, "learning_rate": 1.5834969893148855e-05, "loss": 0.0013, "step": 45150 }, { "grad_norm": 0.037325892597436905, "learning_rate": 1.581485408715997e-05, "loss": 0.0013, "step": 45160 }, { "grad_norm": 0.04305501654744148, "learning_rate": 1.5794748665715785e-05, "loss": 0.0011, "step": 45170 }, { "grad_norm": 0.04053005203604698, "learning_rate": 1.5774653634923857e-05, "loss": 0.0011, "step": 45180 }, { "grad_norm": 0.06124608963727951, "learning_rate": 1.575456900088845e-05, "loss": 0.0012, "step": 45190 }, { "grad_norm": 0.05377684906125069, "learning_rate": 1.5734494769710816e-05, "loss": 0.0015, "step": 45200 }, { "grad_norm": 0.05084618926048279, "learning_rate": 1.5714430947488912e-05, "loss": 0.0015, "step": 45210 }, { "grad_norm": 0.051382821053266525, "learning_rate": 1.5694377540317645e-05, "loss": 0.0015, "step": 45220 }, { "grad_norm": 0.03326239809393883, "learning_rate": 1.5674334554288694e-05, "loss": 0.0009, "step": 45230 }, { "grad_norm": 0.03813609853386879, "learning_rate": 1.5654301995490582e-05, "loss": 0.0019, "step": 45240 }, { "grad_norm": 0.038652848452329636, "learning_rate": 1.5634279870008685e-05, "loss": 0.0017, "step": 45250 }, { "grad_norm": 0.05399107187986374, "learning_rate": 1.5614268183925174e-05, "loss": 0.0014, "step": 45260 }, { "grad_norm": 0.06447041779756546, "learning_rate": 1.5594266943319097e-05, "loss": 0.0016, "step": 45270 }, { "grad_norm": 0.03970228508114815, "learning_rate": 1.5574276154266294e-05, "loss": 0.0013, "step": 45280 }, { "grad_norm": 0.054317180067300797, "learning_rate": 1.5554295822839437e-05, "loss": 0.0016, "step": 45290 }, { "grad_norm": 0.04337753728032112, "learning_rate": 1.5534325955108025e-05, "loss": 0.0018, "step": 45300 }, { "grad_norm": 0.046931102871894836, "learning_rate": 1.5514366557138373e-05, "loss": 0.0018, "step": 45310 }, { "grad_norm": 0.04708966240286827, "learning_rate": 1.5494417634993602e-05, "loss": 0.0012, "step": 45320 }, { "grad_norm": 0.04895802214741707, "learning_rate": 1.547447919473372e-05, "loss": 0.0013, "step": 45330 }, { "grad_norm": 0.045900844037532806, "learning_rate": 1.5454551242415434e-05, "loss": 0.0013, "step": 45340 }, { "grad_norm": 0.044772375375032425, "learning_rate": 1.543463378409239e-05, "loss": 0.0017, "step": 45350 }, { "grad_norm": 0.05248226225376129, "learning_rate": 1.541472682581493e-05, "loss": 0.001, "step": 45360 }, { "grad_norm": 0.044492416083812714, "learning_rate": 1.5394830373630298e-05, "loss": 0.0011, "step": 45370 }, { "grad_norm": 0.031625427305698395, "learning_rate": 1.5374944433582506e-05, "loss": 0.001, "step": 45380 }, { "grad_norm": 0.03879636526107788, "learning_rate": 1.5355069011712375e-05, "loss": 0.001, "step": 45390 }, { "grad_norm": 0.042635757476091385, "learning_rate": 1.5335204114057526e-05, "loss": 0.0013, "step": 45400 }, { "grad_norm": 0.0497710220515728, "learning_rate": 1.5315349746652387e-05, "loss": 0.0011, "step": 45410 }, { "grad_norm": 0.03975752741098404, "learning_rate": 1.5295505915528212e-05, "loss": 0.0014, "step": 45420 }, { "grad_norm": 0.03557031601667404, "learning_rate": 1.5275672626713024e-05, "loss": 0.0014, "step": 45430 }, { "grad_norm": 0.031704336404800415, "learning_rate": 1.5255849886231643e-05, "loss": 0.0011, "step": 45440 }, { "grad_norm": 0.0404135137796402, "learning_rate": 1.523603770010571e-05, "loss": 0.0015, "step": 45450 }, { "grad_norm": 0.05348428711295128, "learning_rate": 1.521623607435363e-05, "loss": 0.002, "step": 45460 }, { "grad_norm": 0.0527169331908226, "learning_rate": 1.5196445014990612e-05, "loss": 0.0015, "step": 45470 }, { "grad_norm": 0.06147736683487892, "learning_rate": 1.5176664528028672e-05, "loss": 0.0014, "step": 45480 }, { "grad_norm": 0.05459366738796234, "learning_rate": 1.5156894619476574e-05, "loss": 0.0012, "step": 45490 }, { "grad_norm": 0.04468003660440445, "learning_rate": 1.5137135295339938e-05, "loss": 0.0014, "step": 45500 }, { "grad_norm": 0.04675285518169403, "learning_rate": 1.5117386561621073e-05, "loss": 0.0014, "step": 45510 }, { "grad_norm": 0.07020141184329987, "learning_rate": 1.5097648424319167e-05, "loss": 0.0025, "step": 45520 }, { "grad_norm": 0.05432651937007904, "learning_rate": 1.5077920889430119e-05, "loss": 0.0013, "step": 45530 }, { "grad_norm": 0.040165066719055176, "learning_rate": 1.5058203962946644e-05, "loss": 0.0015, "step": 45540 }, { "grad_norm": 0.05149964243173599, "learning_rate": 1.503849765085822e-05, "loss": 0.0022, "step": 45550 }, { "grad_norm": 0.04404012858867645, "learning_rate": 1.501880195915109e-05, "loss": 0.0011, "step": 45560 }, { "grad_norm": 0.03668006509542465, "learning_rate": 1.499911689380833e-05, "loss": 0.0013, "step": 45570 }, { "grad_norm": 0.038852009922266006, "learning_rate": 1.4979442460809683e-05, "loss": 0.0012, "step": 45580 }, { "grad_norm": 0.04713957756757736, "learning_rate": 1.4959778666131763e-05, "loss": 0.0011, "step": 45590 }, { "grad_norm": 0.03910002112388611, "learning_rate": 1.4940125515747905e-05, "loss": 0.0012, "step": 45600 }, { "grad_norm": 0.047783661633729935, "learning_rate": 1.4920483015628211e-05, "loss": 0.0013, "step": 45610 }, { "grad_norm": 0.05272037163376808, "learning_rate": 1.490085117173956e-05, "loss": 0.0009, "step": 45620 }, { "grad_norm": 0.06203893944621086, "learning_rate": 1.488122999004558e-05, "loss": 0.0014, "step": 45630 }, { "grad_norm": 0.056438446044921875, "learning_rate": 1.486161947650666e-05, "loss": 0.0018, "step": 45640 }, { "grad_norm": 0.047012969851493835, "learning_rate": 1.4842019637079995e-05, "loss": 0.001, "step": 45650 }, { "grad_norm": 0.07177947461605072, "learning_rate": 1.482243047771944e-05, "loss": 0.0014, "step": 45660 }, { "grad_norm": 0.05052267014980316, "learning_rate": 1.4802852004375712e-05, "loss": 0.001, "step": 45670 }, { "grad_norm": 0.0441480353474617, "learning_rate": 1.4783284222996218e-05, "loss": 0.0017, "step": 45680 }, { "grad_norm": 0.05603256821632385, "learning_rate": 1.4763727139525135e-05, "loss": 0.001, "step": 45690 }, { "grad_norm": 0.034328609704971313, "learning_rate": 1.4744180759903392e-05, "loss": 0.0011, "step": 45700 }, { "grad_norm": 0.04018840938806534, "learning_rate": 1.4724645090068635e-05, "loss": 0.0014, "step": 45710 }, { "grad_norm": 0.047793712466955185, "learning_rate": 1.4705120135955341e-05, "loss": 0.0011, "step": 45720 }, { "grad_norm": 0.03837662935256958, "learning_rate": 1.4685605903494614e-05, "loss": 0.001, "step": 45730 }, { "grad_norm": 0.02730490453541279, "learning_rate": 1.46661023986144e-05, "loss": 0.0012, "step": 45740 }, { "grad_norm": 0.04064948484301567, "learning_rate": 1.4646609627239344e-05, "loss": 0.001, "step": 45750 }, { "grad_norm": 0.04414829984307289, "learning_rate": 1.4627127595290835e-05, "loss": 0.0025, "step": 45760 }, { "grad_norm": 0.04489302262663841, "learning_rate": 1.460765630868699e-05, "loss": 0.0012, "step": 45770 }, { "grad_norm": 0.045841529965400696, "learning_rate": 1.4588195773342678e-05, "loss": 0.0017, "step": 45780 }, { "grad_norm": 0.045548662543296814, "learning_rate": 1.4568745995169485e-05, "loss": 0.0013, "step": 45790 }, { "grad_norm": 0.04902748763561249, "learning_rate": 1.4549306980075778e-05, "loss": 0.0011, "step": 45800 }, { "grad_norm": 0.030713507905602455, "learning_rate": 1.4529878733966557e-05, "loss": 0.0013, "step": 45810 }, { "grad_norm": 0.06209154054522514, "learning_rate": 1.4510461262743658e-05, "loss": 0.0012, "step": 45820 }, { "grad_norm": 0.03396577388048172, "learning_rate": 1.4491054572305585e-05, "loss": 0.0009, "step": 45830 }, { "grad_norm": 0.035836417227983475, "learning_rate": 1.4471658668547566e-05, "loss": 0.0012, "step": 45840 }, { "grad_norm": 0.03814072906970978, "learning_rate": 1.4452273557361579e-05, "loss": 0.0011, "step": 45850 }, { "grad_norm": 0.039252009242773056, "learning_rate": 1.4432899244636282e-05, "loss": 0.0009, "step": 45860 }, { "grad_norm": 0.05463814735412598, "learning_rate": 1.4413535736257134e-05, "loss": 0.0013, "step": 45870 }, { "grad_norm": 0.05787325277924538, "learning_rate": 1.439418303810619e-05, "loss": 0.0012, "step": 45880 }, { "grad_norm": 0.04973071441054344, "learning_rate": 1.4374841156062352e-05, "loss": 0.0012, "step": 45890 }, { "grad_norm": 0.09422586113214493, "learning_rate": 1.4355510096001112e-05, "loss": 0.0016, "step": 45900 }, { "grad_norm": 0.06050478667020798, "learning_rate": 1.4336189863794786e-05, "loss": 0.0015, "step": 45910 }, { "grad_norm": 0.0530063696205616, "learning_rate": 1.4316880465312327e-05, "loss": 0.0014, "step": 45920 }, { "grad_norm": 0.047359343618154526, "learning_rate": 1.4297581906419426e-05, "loss": 0.0012, "step": 45930 }, { "grad_norm": 0.07223302870988846, "learning_rate": 1.4278294192978475e-05, "loss": 0.0015, "step": 45940 }, { "grad_norm": 0.04404931142926216, "learning_rate": 1.4259017330848574e-05, "loss": 0.0012, "step": 45950 }, { "grad_norm": 0.034766316413879395, "learning_rate": 1.4239751325885498e-05, "loss": 0.0011, "step": 45960 }, { "grad_norm": 0.03963158279657364, "learning_rate": 1.4220496183941795e-05, "loss": 0.0015, "step": 45970 }, { "grad_norm": 0.053419239819049835, "learning_rate": 1.4201251910866648e-05, "loss": 0.0019, "step": 45980 }, { "grad_norm": 0.0608658529818058, "learning_rate": 1.4182018512505957e-05, "loss": 0.0012, "step": 45990 }, { "grad_norm": 0.05952328070998192, "learning_rate": 1.4162795994702327e-05, "loss": 0.0015, "step": 46000 }, { "grad_norm": 0.05949981510639191, "learning_rate": 1.4143584363295032e-05, "loss": 0.0015, "step": 46010 }, { "grad_norm": 0.06026626005768776, "learning_rate": 1.4124383624120101e-05, "loss": 0.0015, "step": 46020 }, { "grad_norm": 0.03742896020412445, "learning_rate": 1.4105193783010151e-05, "loss": 0.0013, "step": 46030 }, { "grad_norm": 0.07429492473602295, "learning_rate": 1.4086014845794621e-05, "loss": 0.0016, "step": 46040 }, { "grad_norm": 0.062169287353754044, "learning_rate": 1.4066846818299489e-05, "loss": 0.0019, "step": 46050 }, { "grad_norm": 0.04849258437752724, "learning_rate": 1.4047689706347555e-05, "loss": 0.0012, "step": 46060 }, { "grad_norm": 0.036681097000837326, "learning_rate": 1.402854351575822e-05, "loss": 0.001, "step": 46070 }, { "grad_norm": 0.09613797813653946, "learning_rate": 1.4009408252347588e-05, "loss": 0.0011, "step": 46080 }, { "grad_norm": 0.0343964546918869, "learning_rate": 1.399028392192846e-05, "loss": 0.001, "step": 46090 }, { "grad_norm": 0.049311764538288116, "learning_rate": 1.397117053031029e-05, "loss": 0.001, "step": 46100 }, { "grad_norm": 0.05498640984296799, "learning_rate": 1.3952068083299213e-05, "loss": 0.0015, "step": 46110 }, { "grad_norm": 0.06533177196979523, "learning_rate": 1.3932976586698082e-05, "loss": 0.0012, "step": 46120 }, { "grad_norm": 0.04676033556461334, "learning_rate": 1.3913896046306363e-05, "loss": 0.001, "step": 46130 }, { "grad_norm": 0.03556463122367859, "learning_rate": 1.389482646792023e-05, "loss": 0.001, "step": 46140 }, { "grad_norm": 0.043634578585624695, "learning_rate": 1.387576785733251e-05, "loss": 0.0015, "step": 46150 }, { "grad_norm": 0.035450153052806854, "learning_rate": 1.3856720220332703e-05, "loss": 0.0019, "step": 46160 }, { "grad_norm": 0.05393463000655174, "learning_rate": 1.383768356270701e-05, "loss": 0.0019, "step": 46170 }, { "grad_norm": 0.05431723594665527, "learning_rate": 1.3818657890238207e-05, "loss": 0.0012, "step": 46180 }, { "grad_norm": 0.024233099073171616, "learning_rate": 1.3799643208705859e-05, "loss": 0.0011, "step": 46190 }, { "grad_norm": 0.04283570498228073, "learning_rate": 1.3780639523886058e-05, "loss": 0.0013, "step": 46200 }, { "grad_norm": 0.03958264738321304, "learning_rate": 1.3761646841551668e-05, "loss": 0.0013, "step": 46210 }, { "grad_norm": 0.045967210084199905, "learning_rate": 1.3742665167472146e-05, "loss": 0.0015, "step": 46220 }, { "grad_norm": 0.04008070379495621, "learning_rate": 1.372369450741363e-05, "loss": 0.0014, "step": 46230 }, { "grad_norm": 0.05666425824165344, "learning_rate": 1.3704734867138901e-05, "loss": 0.0015, "step": 46240 }, { "grad_norm": 0.035594504326581955, "learning_rate": 1.36857862524074e-05, "loss": 0.0009, "step": 46250 }, { "grad_norm": 0.06022089719772339, "learning_rate": 1.3666848668975213e-05, "loss": 0.0012, "step": 46260 }, { "grad_norm": 0.0391194187104702, "learning_rate": 1.3647922122595063e-05, "loss": 0.0011, "step": 46270 }, { "grad_norm": 0.03477878496050835, "learning_rate": 1.3629006619016366e-05, "loss": 0.0012, "step": 46280 }, { "grad_norm": 0.03763650730252266, "learning_rate": 1.3610102163985139e-05, "loss": 0.001, "step": 46290 }, { "grad_norm": 0.0472593829035759, "learning_rate": 1.3591208763244057e-05, "loss": 0.002, "step": 46300 }, { "grad_norm": 0.03387000411748886, "learning_rate": 1.3572326422532428e-05, "loss": 0.0009, "step": 46310 }, { "grad_norm": 0.0458303801715374, "learning_rate": 1.355345514758622e-05, "loss": 0.0014, "step": 46320 }, { "grad_norm": 0.03686698526144028, "learning_rate": 1.3534594944138007e-05, "loss": 0.0012, "step": 46330 }, { "grad_norm": 0.050882644951343536, "learning_rate": 1.3515745817917069e-05, "loss": 0.0013, "step": 46340 }, { "grad_norm": 0.03472917154431343, "learning_rate": 1.3496907774649208e-05, "loss": 0.0012, "step": 46350 }, { "grad_norm": 0.05019797012209892, "learning_rate": 1.3478080820056987e-05, "loss": 0.0019, "step": 46360 }, { "grad_norm": 0.0775272324681282, "learning_rate": 1.3459264959859474e-05, "loss": 0.0015, "step": 46370 }, { "grad_norm": 0.03242947533726692, "learning_rate": 1.3440460199772487e-05, "loss": 0.0012, "step": 46380 }, { "grad_norm": 0.05210452526807785, "learning_rate": 1.3421666545508382e-05, "loss": 0.0011, "step": 46390 }, { "grad_norm": 0.06945007294416428, "learning_rate": 1.3402884002776194e-05, "loss": 0.0015, "step": 46400 }, { "grad_norm": 0.04760914295911789, "learning_rate": 1.3384112577281555e-05, "loss": 0.0011, "step": 46410 }, { "grad_norm": 0.06069711968302727, "learning_rate": 1.3365352274726711e-05, "loss": 0.0019, "step": 46420 }, { "grad_norm": 0.06441476941108704, "learning_rate": 1.3346603100810578e-05, "loss": 0.0015, "step": 46430 }, { "grad_norm": 0.029946887865662575, "learning_rate": 1.3327865061228645e-05, "loss": 0.0014, "step": 46440 }, { "grad_norm": 0.029154203832149506, "learning_rate": 1.330913816167304e-05, "loss": 0.0015, "step": 46450 }, { "grad_norm": 0.04345632344484329, "learning_rate": 1.3290422407832492e-05, "loss": 0.001, "step": 46460 }, { "grad_norm": 0.05557679384946823, "learning_rate": 1.3271717805392354e-05, "loss": 0.0024, "step": 46470 }, { "grad_norm": 0.054728906601667404, "learning_rate": 1.3253024360034582e-05, "loss": 0.0019, "step": 46480 }, { "grad_norm": 0.050122860819101334, "learning_rate": 1.323434207743779e-05, "loss": 0.0011, "step": 46490 }, { "grad_norm": 0.05127325654029846, "learning_rate": 1.3215670963277105e-05, "loss": 0.0011, "step": 46500 }, { "grad_norm": 0.031031763181090355, "learning_rate": 1.3197011023224376e-05, "loss": 0.0012, "step": 46510 }, { "grad_norm": 0.03388604149222374, "learning_rate": 1.3178362262947941e-05, "loss": 0.001, "step": 46520 }, { "grad_norm": 0.062034282833337784, "learning_rate": 1.3159724688112845e-05, "loss": 0.0011, "step": 46530 }, { "grad_norm": 0.04773632809519768, "learning_rate": 1.3141098304380683e-05, "loss": 0.0018, "step": 46540 }, { "grad_norm": 0.05342680215835571, "learning_rate": 1.3122483117409651e-05, "loss": 0.0013, "step": 46550 }, { "grad_norm": 0.041026487946510315, "learning_rate": 1.3103879132854552e-05, "loss": 0.0012, "step": 46560 }, { "grad_norm": 0.03226933628320694, "learning_rate": 1.3085286356366771e-05, "loss": 0.001, "step": 46570 }, { "grad_norm": 0.02708420157432556, "learning_rate": 1.3066704793594337e-05, "loss": 0.0013, "step": 46580 }, { "grad_norm": 0.037408433854579926, "learning_rate": 1.3048134450181816e-05, "loss": 0.001, "step": 46590 }, { "grad_norm": 0.033359602093696594, "learning_rate": 1.3029575331770394e-05, "loss": 0.0018, "step": 46600 }, { "grad_norm": 0.046063389629125595, "learning_rate": 1.3011027443997837e-05, "loss": 0.0011, "step": 46610 }, { "grad_norm": 0.05158432945609093, "learning_rate": 1.2992490792498507e-05, "loss": 0.0013, "step": 46620 }, { "grad_norm": 0.04485480114817619, "learning_rate": 1.297396538290333e-05, "loss": 0.0011, "step": 46630 }, { "grad_norm": 0.03547678887844086, "learning_rate": 1.2955451220839888e-05, "loss": 0.0015, "step": 46640 }, { "grad_norm": 0.0551641620695591, "learning_rate": 1.2936948311932223e-05, "loss": 0.0015, "step": 46650 }, { "grad_norm": 0.04466808959841728, "learning_rate": 1.2918456661801104e-05, "loss": 0.0016, "step": 46660 }, { "grad_norm": 0.04069352149963379, "learning_rate": 1.2899976276063736e-05, "loss": 0.0013, "step": 46670 }, { "grad_norm": 0.06009450554847717, "learning_rate": 1.2881507160334022e-05, "loss": 0.0017, "step": 46680 }, { "grad_norm": 0.03856296464800835, "learning_rate": 1.286304932022238e-05, "loss": 0.001, "step": 46690 }, { "grad_norm": 0.08288547396659851, "learning_rate": 1.2844602761335806e-05, "loss": 0.0012, "step": 46700 }, { "grad_norm": 0.038071755319833755, "learning_rate": 1.2826167489277885e-05, "loss": 0.0009, "step": 46710 }, { "grad_norm": 0.03314083814620972, "learning_rate": 1.2807743509648745e-05, "loss": 0.0013, "step": 46720 }, { "grad_norm": 0.06175389885902405, "learning_rate": 1.2789330828045149e-05, "loss": 0.0015, "step": 46730 }, { "grad_norm": 0.049710534512996674, "learning_rate": 1.2770929450060332e-05, "loss": 0.001, "step": 46740 }, { "grad_norm": 0.03910862281918526, "learning_rate": 1.2752539381284184e-05, "loss": 0.0017, "step": 46750 }, { "grad_norm": 0.057268980890512466, "learning_rate": 1.273416062730311e-05, "loss": 0.0015, "step": 46760 }, { "grad_norm": 0.031452544033527374, "learning_rate": 1.2715793193700088e-05, "loss": 0.0013, "step": 46770 }, { "grad_norm": 0.06542156636714935, "learning_rate": 1.2697437086054664e-05, "loss": 0.0026, "step": 46780 }, { "grad_norm": 0.042816538363695145, "learning_rate": 1.2679092309942937e-05, "loss": 0.0011, "step": 46790 }, { "grad_norm": 0.03587433323264122, "learning_rate": 1.266075887093755e-05, "loss": 0.0012, "step": 46800 }, { "grad_norm": 0.04709906131029129, "learning_rate": 1.2642436774607757e-05, "loss": 0.0018, "step": 46810 }, { "grad_norm": 0.05160772055387497, "learning_rate": 1.2624126026519278e-05, "loss": 0.0014, "step": 46820 }, { "grad_norm": 0.033181142061948776, "learning_rate": 1.2605826632234474e-05, "loss": 0.0011, "step": 46830 }, { "grad_norm": 0.04752103611826897, "learning_rate": 1.2587538597312198e-05, "loss": 0.0012, "step": 46840 }, { "grad_norm": 0.045916829258203506, "learning_rate": 1.2569261927307884e-05, "loss": 0.0014, "step": 46850 }, { "grad_norm": 0.048861030489206314, "learning_rate": 1.2550996627773493e-05, "loss": 0.001, "step": 46860 }, { "grad_norm": 0.03658575564622879, "learning_rate": 1.2532742704257527e-05, "loss": 0.0012, "step": 46870 }, { "grad_norm": 0.05359451845288277, "learning_rate": 1.2514500162305087e-05, "loss": 0.0015, "step": 46880 }, { "grad_norm": 0.03327082842588425, "learning_rate": 1.2496269007457728e-05, "loss": 0.0009, "step": 46890 }, { "grad_norm": 0.047316957265138626, "learning_rate": 1.2478049245253625e-05, "loss": 0.0019, "step": 46900 }, { "grad_norm": 0.052227720618247986, "learning_rate": 1.2459840881227459e-05, "loss": 0.0015, "step": 46910 }, { "grad_norm": 0.05170111358165741, "learning_rate": 1.2441643920910435e-05, "loss": 0.0016, "step": 46920 }, { "grad_norm": 0.040631309151649475, "learning_rate": 1.2423458369830322e-05, "loss": 0.0016, "step": 46930 }, { "grad_norm": 0.048736829310655594, "learning_rate": 1.2405284233511406e-05, "loss": 0.0015, "step": 46940 }, { "grad_norm": 0.06186909228563309, "learning_rate": 1.2387121517474487e-05, "loss": 0.0013, "step": 46950 }, { "grad_norm": 0.06860630214214325, "learning_rate": 1.2368970227236975e-05, "loss": 0.0011, "step": 46960 }, { "grad_norm": 0.044824179261922836, "learning_rate": 1.2350830368312688e-05, "loss": 0.0011, "step": 46970 }, { "grad_norm": 0.03619936481118202, "learning_rate": 1.2332701946212083e-05, "loss": 0.001, "step": 46980 }, { "grad_norm": 0.07004960626363754, "learning_rate": 1.2314584966442077e-05, "loss": 0.0013, "step": 46990 }, { "grad_norm": 0.06333940476179123, "learning_rate": 1.2296479434506136e-05, "loss": 0.0013, "step": 47000 }, { "grad_norm": 0.06054126098752022, "learning_rate": 1.2278385355904232e-05, "loss": 0.001, "step": 47010 }, { "grad_norm": 0.04028157517313957, "learning_rate": 1.2260302736132867e-05, "loss": 0.0012, "step": 47020 }, { "grad_norm": 0.03692275658249855, "learning_rate": 1.2242231580685098e-05, "loss": 0.0011, "step": 47030 }, { "grad_norm": 0.0453733466565609, "learning_rate": 1.2224171895050413e-05, "loss": 0.0011, "step": 47040 }, { "grad_norm": 0.02799728885293007, "learning_rate": 1.2206123684714903e-05, "loss": 0.0009, "step": 47050 }, { "grad_norm": 0.05418253690004349, "learning_rate": 1.2188086955161132e-05, "loss": 0.0011, "step": 47060 }, { "grad_norm": 0.03448612615466118, "learning_rate": 1.2170061711868175e-05, "loss": 0.0009, "step": 47070 }, { "grad_norm": 0.06835225224494934, "learning_rate": 1.215204796031163e-05, "loss": 0.0014, "step": 47080 }, { "grad_norm": 0.05034807696938515, "learning_rate": 1.2134045705963599e-05, "loss": 0.001, "step": 47090 }, { "grad_norm": 0.029787059873342514, "learning_rate": 1.2116054954292689e-05, "loss": 0.0009, "step": 47100 }, { "grad_norm": 0.0315895639359951, "learning_rate": 1.2098075710764011e-05, "loss": 0.0011, "step": 47110 }, { "grad_norm": 0.04864165931940079, "learning_rate": 1.2080107980839183e-05, "loss": 0.0012, "step": 47120 }, { "grad_norm": 0.04892661049962044, "learning_rate": 1.2062151769976343e-05, "loss": 0.001, "step": 47130 }, { "grad_norm": 0.041389621794223785, "learning_rate": 1.204420708363011e-05, "loss": 0.0017, "step": 47140 }, { "grad_norm": 0.042853642255067825, "learning_rate": 1.2026273927251597e-05, "loss": 0.0012, "step": 47150 }, { "grad_norm": 0.04368753731250763, "learning_rate": 1.2008352306288424e-05, "loss": 0.0011, "step": 47160 }, { "grad_norm": 0.04486534371972084, "learning_rate": 1.1990442226184695e-05, "loss": 0.0013, "step": 47170 }, { "grad_norm": 0.06087829917669296, "learning_rate": 1.1972543692381066e-05, "loss": 0.0019, "step": 47180 }, { "grad_norm": 0.037710484117269516, "learning_rate": 1.1954656710314576e-05, "loss": 0.0013, "step": 47190 }, { "grad_norm": 0.045315634459257126, "learning_rate": 1.1936781285418875e-05, "loss": 0.0013, "step": 47200 }, { "grad_norm": 0.03780338540673256, "learning_rate": 1.1918917423123993e-05, "loss": 0.0015, "step": 47210 }, { "grad_norm": 0.03046608716249466, "learning_rate": 1.1901065128856537e-05, "loss": 0.001, "step": 47220 }, { "grad_norm": 0.04212281107902527, "learning_rate": 1.1883224408039551e-05, "loss": 0.0012, "step": 47230 }, { "grad_norm": 0.03553682938218117, "learning_rate": 1.1865395266092578e-05, "loss": 0.001, "step": 47240 }, { "grad_norm": 0.053272176533937454, "learning_rate": 1.1847577708431633e-05, "loss": 0.0013, "step": 47250 }, { "grad_norm": 0.06463871151208878, "learning_rate": 1.1829771740469225e-05, "loss": 0.0011, "step": 47260 }, { "grad_norm": 0.0540807768702507, "learning_rate": 1.1811977367614324e-05, "loss": 0.0019, "step": 47270 }, { "grad_norm": 0.05154254287481308, "learning_rate": 1.1794194595272412e-05, "loss": 0.0012, "step": 47280 }, { "grad_norm": 0.037115924060344696, "learning_rate": 1.1776423428845423e-05, "loss": 0.0011, "step": 47290 }, { "grad_norm": 0.03207267448306084, "learning_rate": 1.1758663873731756e-05, "loss": 0.0009, "step": 47300 }, { "grad_norm": 0.027043282985687256, "learning_rate": 1.1740915935326302e-05, "loss": 0.0009, "step": 47310 }, { "grad_norm": 0.062134113162755966, "learning_rate": 1.1723179619020396e-05, "loss": 0.0012, "step": 47320 }, { "grad_norm": 0.044394005089998245, "learning_rate": 1.1705454930201914e-05, "loss": 0.0014, "step": 47330 }, { "grad_norm": 0.051141951233148575, "learning_rate": 1.1687741874255087e-05, "loss": 0.0013, "step": 47340 }, { "grad_norm": 0.03335706144571304, "learning_rate": 1.1670040456560728e-05, "loss": 0.001, "step": 47350 }, { "grad_norm": 0.04186530038714409, "learning_rate": 1.1652350682496005e-05, "loss": 0.001, "step": 47360 }, { "grad_norm": 0.038836073130369186, "learning_rate": 1.163467255743465e-05, "loss": 0.0011, "step": 47370 }, { "grad_norm": 0.0353303886950016, "learning_rate": 1.1617006086746796e-05, "loss": 0.0012, "step": 47380 }, { "grad_norm": 0.038536328822374344, "learning_rate": 1.1599351275799047e-05, "loss": 0.0017, "step": 47390 }, { "grad_norm": 0.03452679142355919, "learning_rate": 1.1581708129954466e-05, "loss": 0.0013, "step": 47400 }, { "grad_norm": 0.0346282459795475, "learning_rate": 1.1564076654572587e-05, "loss": 0.0015, "step": 47410 }, { "grad_norm": 0.031787384301424026, "learning_rate": 1.1546456855009358e-05, "loss": 0.0009, "step": 47420 }, { "grad_norm": 0.05274383723735809, "learning_rate": 1.1528848736617248e-05, "loss": 0.0019, "step": 47430 }, { "grad_norm": 0.04643828794360161, "learning_rate": 1.1511252304745112e-05, "loss": 0.002, "step": 47440 }, { "grad_norm": 0.029134467244148254, "learning_rate": 1.1493667564738297e-05, "loss": 0.001, "step": 47450 }, { "grad_norm": 0.04763577878475189, "learning_rate": 1.1476094521938574e-05, "loss": 0.001, "step": 47460 }, { "grad_norm": 0.0560605525970459, "learning_rate": 1.1458533181684167e-05, "loss": 0.0016, "step": 47470 }, { "grad_norm": 0.047972626984119415, "learning_rate": 1.1440983549309753e-05, "loss": 0.0011, "step": 47480 }, { "grad_norm": 0.04018077254295349, "learning_rate": 1.1423445630146434e-05, "loss": 0.0021, "step": 47490 }, { "grad_norm": 0.0412001833319664, "learning_rate": 1.1405919429521799e-05, "loss": 0.0012, "step": 47500 }, { "grad_norm": 0.043665118515491486, "learning_rate": 1.1388404952759802e-05, "loss": 0.0014, "step": 47510 }, { "grad_norm": 0.04437163099646568, "learning_rate": 1.1370902205180923e-05, "loss": 0.0013, "step": 47520 }, { "grad_norm": 0.06393054127693176, "learning_rate": 1.1353411192101987e-05, "loss": 0.0013, "step": 47530 }, { "grad_norm": 0.041200943291187286, "learning_rate": 1.133593191883634e-05, "loss": 0.0016, "step": 47540 }, { "grad_norm": 0.05441569909453392, "learning_rate": 1.1318464390693711e-05, "loss": 0.0017, "step": 47550 }, { "grad_norm": 0.034991566091775894, "learning_rate": 1.1301008612980257e-05, "loss": 0.0011, "step": 47560 }, { "grad_norm": 0.038787562400102615, "learning_rate": 1.128356459099863e-05, "loss": 0.0011, "step": 47570 }, { "grad_norm": 0.07340555638074875, "learning_rate": 1.1266132330047802e-05, "loss": 0.0019, "step": 47580 }, { "grad_norm": 0.05034352093935013, "learning_rate": 1.1248711835423281e-05, "loss": 0.0014, "step": 47590 }, { "grad_norm": 0.035317059606313705, "learning_rate": 1.123130311241693e-05, "loss": 0.0018, "step": 47600 }, { "grad_norm": 0.04352203756570816, "learning_rate": 1.1213906166317068e-05, "loss": 0.0012, "step": 47610 }, { "grad_norm": 0.044771913439035416, "learning_rate": 1.1196521002408427e-05, "loss": 0.0014, "step": 47620 }, { "grad_norm": 0.05131899565458298, "learning_rate": 1.1179147625972159e-05, "loss": 0.0016, "step": 47630 }, { "grad_norm": 0.05189485475420952, "learning_rate": 1.1161786042285822e-05, "loss": 0.001, "step": 47640 }, { "grad_norm": 0.04266320541501045, "learning_rate": 1.1144436256623447e-05, "loss": 0.0011, "step": 47650 }, { "grad_norm": 0.036122340708971024, "learning_rate": 1.1127098274255392e-05, "loss": 0.0011, "step": 47660 }, { "grad_norm": 0.030339915305376053, "learning_rate": 1.1109772100448512e-05, "loss": 0.0014, "step": 47670 }, { "grad_norm": 0.03893568366765976, "learning_rate": 1.1092457740466033e-05, "loss": 0.0013, "step": 47680 }, { "grad_norm": 0.04607178270816803, "learning_rate": 1.10751551995676e-05, "loss": 0.0012, "step": 47690 }, { "grad_norm": 0.04170548915863037, "learning_rate": 1.1057864483009262e-05, "loss": 0.001, "step": 47700 }, { "grad_norm": 0.03832218796014786, "learning_rate": 1.1040585596043473e-05, "loss": 0.0009, "step": 47710 }, { "grad_norm": 0.04469848796725273, "learning_rate": 1.1023318543919148e-05, "loss": 0.0011, "step": 47720 }, { "grad_norm": 0.05297667905688286, "learning_rate": 1.10060633318815e-05, "loss": 0.0013, "step": 47730 }, { "grad_norm": 0.04727767035365105, "learning_rate": 1.0988819965172248e-05, "loss": 0.0013, "step": 47740 }, { "grad_norm": 0.044491443783044815, "learning_rate": 1.0971588449029462e-05, "loss": 0.001, "step": 47750 }, { "grad_norm": 0.049741845577955246, "learning_rate": 1.095436878868762e-05, "loss": 0.0011, "step": 47760 }, { "grad_norm": 0.06577306985855103, "learning_rate": 1.0937160989377598e-05, "loss": 0.0015, "step": 47770 }, { "grad_norm": 0.027539830654859543, "learning_rate": 1.0919965056326676e-05, "loss": 0.0012, "step": 47780 }, { "grad_norm": 0.03967664763331413, "learning_rate": 1.0902780994758504e-05, "loss": 0.0009, "step": 47790 }, { "grad_norm": 0.03522935137152672, "learning_rate": 1.0885608809893193e-05, "loss": 0.001, "step": 47800 }, { "grad_norm": 0.02452116273343563, "learning_rate": 1.0868448506947142e-05, "loss": 0.0011, "step": 47810 }, { "grad_norm": 0.046873439103364944, "learning_rate": 1.0851300091133243e-05, "loss": 0.0013, "step": 47820 }, { "grad_norm": 0.028417594730854034, "learning_rate": 1.083416356766071e-05, "loss": 0.0011, "step": 47830 }, { "grad_norm": 0.025165094062685966, "learning_rate": 1.0817038941735175e-05, "loss": 0.0011, "step": 47840 }, { "grad_norm": 0.02308029867708683, "learning_rate": 1.0799926218558642e-05, "loss": 0.001, "step": 47850 }, { "grad_norm": 0.025925323367118835, "learning_rate": 1.0782825403329488e-05, "loss": 0.0012, "step": 47860 }, { "grad_norm": 0.026844892650842667, "learning_rate": 1.076573650124254e-05, "loss": 0.0013, "step": 47870 }, { "grad_norm": 0.03607764467597008, "learning_rate": 1.0748659517488891e-05, "loss": 0.0011, "step": 47880 }, { "grad_norm": 0.03940794989466667, "learning_rate": 1.0731594457256138e-05, "loss": 0.0014, "step": 47890 }, { "grad_norm": 0.029100151732563972, "learning_rate": 1.0714541325728139e-05, "loss": 0.002, "step": 47900 }, { "grad_norm": 0.05778423324227333, "learning_rate": 1.0697500128085231e-05, "loss": 0.0013, "step": 47910 }, { "grad_norm": 0.042310066521167755, "learning_rate": 1.0680470869504055e-05, "loss": 0.0015, "step": 47920 }, { "grad_norm": 0.04990653693675995, "learning_rate": 1.066345355515766e-05, "loss": 0.0013, "step": 47930 }, { "grad_norm": 0.03876606374979019, "learning_rate": 1.0646448190215453e-05, "loss": 0.0013, "step": 47940 }, { "grad_norm": 0.06333891302347183, "learning_rate": 1.0629454779843217e-05, "loss": 0.0015, "step": 47950 }, { "grad_norm": 0.048361510038375854, "learning_rate": 1.0612473329203082e-05, "loss": 0.0012, "step": 47960 }, { "grad_norm": 0.043041180819272995, "learning_rate": 1.0595503843453596e-05, "loss": 0.0014, "step": 47970 }, { "grad_norm": 0.04323961213231087, "learning_rate": 1.0578546327749634e-05, "loss": 0.0011, "step": 47980 }, { "grad_norm": 0.037323325872421265, "learning_rate": 1.0561600787242425e-05, "loss": 0.001, "step": 47990 }, { "grad_norm": 0.0872190073132515, "learning_rate": 1.0544667227079591e-05, "loss": 0.0014, "step": 48000 }, { "grad_norm": 0.033517107367515564, "learning_rate": 1.0527745652405085e-05, "loss": 0.0009, "step": 48010 }, { "grad_norm": 0.05096007138490677, "learning_rate": 1.051083606835927e-05, "loss": 0.0012, "step": 48020 }, { "grad_norm": 0.055037256330251694, "learning_rate": 1.049393848007878e-05, "loss": 0.001, "step": 48030 }, { "grad_norm": 0.03057841956615448, "learning_rate": 1.0477052892696709e-05, "loss": 0.0016, "step": 48040 }, { "grad_norm": 0.023678239434957504, "learning_rate": 1.0460179311342394e-05, "loss": 0.0007, "step": 48050 }, { "grad_norm": 0.04654904827475548, "learning_rate": 1.0443317741141634e-05, "loss": 0.0017, "step": 48060 }, { "grad_norm": 0.04428818076848984, "learning_rate": 1.0426468187216514e-05, "loss": 0.0011, "step": 48070 }, { "grad_norm": 0.028832416981458664, "learning_rate": 1.0409630654685477e-05, "loss": 0.0012, "step": 48080 }, { "grad_norm": 0.04621358588337898, "learning_rate": 1.039280514866332e-05, "loss": 0.0012, "step": 48090 }, { "grad_norm": 0.04286601394414902, "learning_rate": 1.0375991674261198e-05, "loss": 0.0016, "step": 48100 }, { "grad_norm": 0.052294764667749405, "learning_rate": 1.0359190236586575e-05, "loss": 0.0014, "step": 48110 }, { "grad_norm": 0.042175792157649994, "learning_rate": 1.0342400840743322e-05, "loss": 0.0011, "step": 48120 }, { "grad_norm": 0.026841526851058006, "learning_rate": 1.0325623491831593e-05, "loss": 0.001, "step": 48130 }, { "grad_norm": 0.0389617383480072, "learning_rate": 1.0308858194947906e-05, "loss": 0.0012, "step": 48140 }, { "grad_norm": 0.07926053553819656, "learning_rate": 1.0292104955185111e-05, "loss": 0.0014, "step": 48150 }, { "grad_norm": 0.03194769099354744, "learning_rate": 1.0275363777632396e-05, "loss": 0.0009, "step": 48160 }, { "grad_norm": 0.02546665072441101, "learning_rate": 1.0258634667375321e-05, "loss": 0.0011, "step": 48170 }, { "grad_norm": 0.06789340078830719, "learning_rate": 1.02419176294957e-05, "loss": 0.0013, "step": 48180 }, { "grad_norm": 0.031453687697649, "learning_rate": 1.0225212669071782e-05, "loss": 0.001, "step": 48190 }, { "grad_norm": 0.027412211522459984, "learning_rate": 1.0208519791178029e-05, "loss": 0.0009, "step": 48200 }, { "grad_norm": 0.04515842720866203, "learning_rate": 1.019183900088535e-05, "loss": 0.0013, "step": 48210 }, { "grad_norm": 0.04139827936887741, "learning_rate": 1.0175170303260906e-05, "loss": 0.0014, "step": 48220 }, { "grad_norm": 0.0363195575773716, "learning_rate": 1.0158513703368206e-05, "loss": 0.0012, "step": 48230 }, { "grad_norm": 0.04182418808341026, "learning_rate": 1.0141869206267095e-05, "loss": 0.0011, "step": 48240 }, { "grad_norm": 0.06606338918209076, "learning_rate": 1.0125236817013723e-05, "loss": 0.0013, "step": 48250 }, { "grad_norm": 0.04679799824953079, "learning_rate": 1.010861654066056e-05, "loss": 0.0009, "step": 48260 }, { "grad_norm": 0.06578245759010315, "learning_rate": 1.0092008382256434e-05, "loss": 0.0013, "step": 48270 }, { "grad_norm": 0.06244608759880066, "learning_rate": 1.0075412346846458e-05, "loss": 0.001, "step": 48280 }, { "grad_norm": 0.04070896655321121, "learning_rate": 1.0058828439472056e-05, "loss": 0.0018, "step": 48290 }, { "grad_norm": 0.031024275347590446, "learning_rate": 1.0042256665170996e-05, "loss": 0.0013, "step": 48300 }, { "grad_norm": 0.03782900422811508, "learning_rate": 1.0025697028977332e-05, "loss": 0.0015, "step": 48310 }, { "grad_norm": 0.07687662541866302, "learning_rate": 1.0009149535921454e-05, "loss": 0.0018, "step": 48320 }, { "grad_norm": 0.035947900265455246, "learning_rate": 9.992614191030031e-06, "loss": 0.001, "step": 48330 }, { "grad_norm": 0.03455832973122597, "learning_rate": 9.976090999326115e-06, "loss": 0.0009, "step": 48340 }, { "grad_norm": 0.03627992793917656, "learning_rate": 9.959579965828952e-06, "loss": 0.0009, "step": 48350 }, { "grad_norm": 0.036076612770557404, "learning_rate": 9.943081095554218e-06, "loss": 0.001, "step": 48360 }, { "grad_norm": 0.06483978778123856, "learning_rate": 9.926594393513783e-06, "loss": 0.0011, "step": 48370 }, { "grad_norm": 0.06204993277788162, "learning_rate": 9.910119864715906e-06, "loss": 0.0011, "step": 48380 }, { "grad_norm": 0.04839164763689041, "learning_rate": 9.8936575141651e-06, "loss": 0.0008, "step": 48390 }, { "grad_norm": 0.04321664199233055, "learning_rate": 9.877207346862194e-06, "loss": 0.0009, "step": 48400 }, { "grad_norm": 0.07397144287824631, "learning_rate": 9.860769367804312e-06, "loss": 0.0013, "step": 48410 }, { "grad_norm": 0.045037828385829926, "learning_rate": 9.844343581984877e-06, "loss": 0.001, "step": 48420 }, { "grad_norm": 0.06107260286808014, "learning_rate": 9.82792999439362e-06, "loss": 0.0013, "step": 48430 }, { "grad_norm": 0.03921382129192352, "learning_rate": 9.811528610016546e-06, "loss": 0.0011, "step": 48440 }, { "grad_norm": 0.05590790882706642, "learning_rate": 9.79513943383597e-06, "loss": 0.0015, "step": 48450 }, { "grad_norm": 0.046472758054733276, "learning_rate": 9.778762470830489e-06, "loss": 0.0011, "step": 48460 }, { "grad_norm": 0.042588308453559875, "learning_rate": 9.762397725974982e-06, "loss": 0.0011, "step": 48470 }, { "grad_norm": 0.03578732907772064, "learning_rate": 9.746045204240622e-06, "loss": 0.0015, "step": 48480 }, { "grad_norm": 0.02937445044517517, "learning_rate": 9.729704910594917e-06, "loss": 0.0013, "step": 48490 }, { "grad_norm": 0.03715091571211815, "learning_rate": 9.713376850001554e-06, "loss": 0.0016, "step": 48500 }, { "grad_norm": 0.06770790368318558, "learning_rate": 9.697061027420622e-06, "loss": 0.0013, "step": 48510 }, { "grad_norm": 0.04903491213917732, "learning_rate": 9.680757447808385e-06, "loss": 0.0015, "step": 48520 }, { "grad_norm": 0.03560156375169754, "learning_rate": 9.664466116117488e-06, "loss": 0.0011, "step": 48530 }, { "grad_norm": 0.030350342392921448, "learning_rate": 9.64818703729678e-06, "loss": 0.0007, "step": 48540 }, { "grad_norm": 0.05217041075229645, "learning_rate": 9.631920216291423e-06, "loss": 0.0019, "step": 48550 }, { "grad_norm": 0.037871867418289185, "learning_rate": 9.615665658042849e-06, "loss": 0.0014, "step": 48560 }, { "grad_norm": 0.055032115429639816, "learning_rate": 9.599423367488747e-06, "loss": 0.0018, "step": 48570 }, { "grad_norm": 0.08048520982265472, "learning_rate": 9.583193349563124e-06, "loss": 0.0014, "step": 48580 }, { "grad_norm": 0.03683581203222275, "learning_rate": 9.566975609196216e-06, "loss": 0.0011, "step": 48590 }, { "grad_norm": 0.02787424623966217, "learning_rate": 9.550770151314548e-06, "loss": 0.0015, "step": 48600 }, { "grad_norm": 0.05850536376237869, "learning_rate": 9.53457698084091e-06, "loss": 0.001, "step": 48610 }, { "grad_norm": 0.03456752002239227, "learning_rate": 9.518396102694355e-06, "loss": 0.0016, "step": 48620 }, { "grad_norm": 0.0515015535056591, "learning_rate": 9.502227521790198e-06, "loss": 0.0012, "step": 48630 }, { "grad_norm": 0.034657660871744156, "learning_rate": 9.486071243040063e-06, "loss": 0.0012, "step": 48640 }, { "grad_norm": 0.052675358951091766, "learning_rate": 9.469927271351747e-06, "loss": 0.0017, "step": 48650 }, { "grad_norm": 0.043131571263074875, "learning_rate": 9.453795611629419e-06, "loss": 0.001, "step": 48660 }, { "grad_norm": 0.0362323634326458, "learning_rate": 9.437676268773399e-06, "loss": 0.0008, "step": 48670 }, { "grad_norm": 0.061982665210962296, "learning_rate": 9.421569247680357e-06, "loss": 0.0012, "step": 48680 }, { "grad_norm": 0.037950299680233, "learning_rate": 9.40547455324316e-06, "loss": 0.0011, "step": 48690 }, { "grad_norm": 0.03406378999352455, "learning_rate": 9.389392190350965e-06, "loss": 0.0016, "step": 48700 }, { "grad_norm": 0.03831136226654053, "learning_rate": 9.373322163889153e-06, "loss": 0.0009, "step": 48710 }, { "grad_norm": 0.032248154282569885, "learning_rate": 9.357264478739375e-06, "loss": 0.0012, "step": 48720 }, { "grad_norm": 0.03864055871963501, "learning_rate": 9.341219139779567e-06, "loss": 0.0012, "step": 48730 }, { "grad_norm": 0.040732491761446, "learning_rate": 9.325186151883824e-06, "loss": 0.0009, "step": 48740 }, { "grad_norm": 0.06667274236679077, "learning_rate": 9.30916551992258e-06, "loss": 0.0014, "step": 48750 }, { "grad_norm": 0.05230396240949631, "learning_rate": 9.293157248762479e-06, "loss": 0.0011, "step": 48760 }, { "grad_norm": 0.03255194053053856, "learning_rate": 9.2771613432664e-06, "loss": 0.0013, "step": 48770 }, { "grad_norm": 0.04377840459346771, "learning_rate": 9.261177808293481e-06, "loss": 0.0011, "step": 48780 }, { "grad_norm": 0.09338803589344025, "learning_rate": 9.245206648699096e-06, "loss": 0.0015, "step": 48790 }, { "grad_norm": 0.05323972925543785, "learning_rate": 9.22924786933485e-06, "loss": 0.0015, "step": 48800 }, { "grad_norm": 0.05009164661169052, "learning_rate": 9.213301475048642e-06, "loss": 0.0011, "step": 48810 }, { "grad_norm": 0.052670206874608994, "learning_rate": 9.197367470684504e-06, "loss": 0.001, "step": 48820 }, { "grad_norm": 0.05207394063472748, "learning_rate": 9.181445861082816e-06, "loss": 0.0015, "step": 48830 }, { "grad_norm": 0.03802059218287468, "learning_rate": 9.16553665108012e-06, "loss": 0.0011, "step": 48840 }, { "grad_norm": 0.04582397639751434, "learning_rate": 9.149639845509223e-06, "loss": 0.001, "step": 48850 }, { "grad_norm": 0.044514838606119156, "learning_rate": 9.133755449199144e-06, "loss": 0.0011, "step": 48860 }, { "grad_norm": 0.030723338946700096, "learning_rate": 9.117883466975135e-06, "loss": 0.0008, "step": 48870 }, { "grad_norm": 0.03462297096848488, "learning_rate": 9.10202390365873e-06, "loss": 0.0016, "step": 48880 }, { "grad_norm": 0.027272971346974373, "learning_rate": 9.086176764067583e-06, "loss": 0.0011, "step": 48890 }, { "grad_norm": 0.03540695458650589, "learning_rate": 9.070342053015684e-06, "loss": 0.001, "step": 48900 }, { "grad_norm": 0.060684915632009506, "learning_rate": 9.054519775313187e-06, "loss": 0.0007, "step": 48910 }, { "grad_norm": 0.04042532667517662, "learning_rate": 9.038709935766476e-06, "loss": 0.001, "step": 48920 }, { "grad_norm": 0.03575567156076431, "learning_rate": 9.02291253917817e-06, "loss": 0.0013, "step": 48930 }, { "grad_norm": 0.03645153343677521, "learning_rate": 9.007127590347091e-06, "loss": 0.001, "step": 48940 }, { "grad_norm": 0.04824043810367584, "learning_rate": 8.991355094068288e-06, "loss": 0.001, "step": 48950 }, { "grad_norm": 0.02752477116882801, "learning_rate": 8.975595055133062e-06, "loss": 0.0012, "step": 48960 }, { "grad_norm": 0.05707898363471031, "learning_rate": 8.959847478328848e-06, "loss": 0.0009, "step": 48970 }, { "grad_norm": 0.0462682843208313, "learning_rate": 8.944112368439378e-06, "loss": 0.0014, "step": 48980 }, { "grad_norm": 0.03699992969632149, "learning_rate": 8.928389730244552e-06, "loss": 0.0009, "step": 48990 }, { "grad_norm": 0.03032098151743412, "learning_rate": 8.912679568520494e-06, "loss": 0.0009, "step": 49000 }, { "grad_norm": 0.04025959596037865, "learning_rate": 8.896981888039534e-06, "loss": 0.0013, "step": 49010 }, { "grad_norm": 0.029242143034934998, "learning_rate": 8.881296693570201e-06, "loss": 0.0024, "step": 49020 }, { "grad_norm": 0.06846607476472855, "learning_rate": 8.865623989877281e-06, "loss": 0.0011, "step": 49030 }, { "grad_norm": 0.03249130770564079, "learning_rate": 8.849963781721681e-06, "loss": 0.001, "step": 49040 }, { "grad_norm": 0.0604640431702137, "learning_rate": 8.834316073860588e-06, "loss": 0.0014, "step": 49050 }, { "grad_norm": 0.04598044231534004, "learning_rate": 8.818680871047357e-06, "loss": 0.0015, "step": 49060 }, { "grad_norm": 0.044467318803071976, "learning_rate": 8.803058178031549e-06, "loss": 0.0011, "step": 49070 }, { "grad_norm": 0.03563545644283295, "learning_rate": 8.787447999558922e-06, "loss": 0.0011, "step": 49080 }, { "grad_norm": 0.03107176162302494, "learning_rate": 8.77185034037144e-06, "loss": 0.001, "step": 49090 }, { "grad_norm": 0.04248611629009247, "learning_rate": 8.756265205207259e-06, "loss": 0.0009, "step": 49100 }, { "grad_norm": 0.06531859934329987, "learning_rate": 8.740692598800732e-06, "loss": 0.0013, "step": 49110 }, { "grad_norm": 0.04581757262349129, "learning_rate": 8.72513252588239e-06, "loss": 0.0012, "step": 49120 }, { "grad_norm": 0.026701314374804497, "learning_rate": 8.709584991178998e-06, "loss": 0.0009, "step": 49130 }, { "grad_norm": 0.025530733168125153, "learning_rate": 8.694049999413479e-06, "loss": 0.0008, "step": 49140 }, { "grad_norm": 0.034179504960775375, "learning_rate": 8.678527555304945e-06, "loss": 0.0012, "step": 49150 }, { "grad_norm": 0.039134178310632706, "learning_rate": 8.663017663568712e-06, "loss": 0.0011, "step": 49160 }, { "grad_norm": 0.057160634547472, "learning_rate": 8.647520328916259e-06, "loss": 0.0015, "step": 49170 }, { "grad_norm": 0.04783875122666359, "learning_rate": 8.632035556055307e-06, "loss": 0.0013, "step": 49180 }, { "grad_norm": 0.038005419075489044, "learning_rate": 8.616563349689672e-06, "loss": 0.0014, "step": 49190 }, { "grad_norm": 0.04646870121359825, "learning_rate": 8.601103714519448e-06, "loss": 0.0016, "step": 49200 }, { "grad_norm": 0.035159219056367874, "learning_rate": 8.58565665524082e-06, "loss": 0.0019, "step": 49210 }, { "grad_norm": 0.030024481937289238, "learning_rate": 8.570222176546222e-06, "loss": 0.0009, "step": 49220 }, { "grad_norm": 0.03347392380237579, "learning_rate": 8.554800283124242e-06, "loss": 0.001, "step": 49230 }, { "grad_norm": 0.04237009957432747, "learning_rate": 8.539390979659639e-06, "loss": 0.0014, "step": 49240 }, { "grad_norm": 0.038728345185518265, "learning_rate": 8.523994270833352e-06, "loss": 0.0012, "step": 49250 }, { "grad_norm": 0.02954372577369213, "learning_rate": 8.5086101613225e-06, "loss": 0.0009, "step": 49260 }, { "grad_norm": 0.033371105790138245, "learning_rate": 8.493238655800346e-06, "loss": 0.002, "step": 49270 }, { "grad_norm": 0.04356628283858299, "learning_rate": 8.47787975893638e-06, "loss": 0.001, "step": 49280 }, { "grad_norm": 0.08303876966238022, "learning_rate": 8.462533475396211e-06, "loss": 0.0017, "step": 49290 }, { "grad_norm": 0.0435803197324276, "learning_rate": 8.447199809841643e-06, "loss": 0.0018, "step": 49300 }, { "grad_norm": 0.048453498631715775, "learning_rate": 8.431878766930635e-06, "loss": 0.0009, "step": 49310 }, { "grad_norm": 0.023118484765291214, "learning_rate": 8.416570351317304e-06, "loss": 0.0009, "step": 49320 }, { "grad_norm": 0.032034728676080704, "learning_rate": 8.401274567651973e-06, "loss": 0.0013, "step": 49330 }, { "grad_norm": 0.03188565745949745, "learning_rate": 8.385991420581058e-06, "loss": 0.0009, "step": 49340 }, { "grad_norm": 0.03192419931292534, "learning_rate": 8.370720914747215e-06, "loss": 0.0009, "step": 49350 }, { "grad_norm": 0.034466709941625595, "learning_rate": 8.355463054789181e-06, "loss": 0.0015, "step": 49360 }, { "grad_norm": 0.06193813309073448, "learning_rate": 8.340217845341919e-06, "loss": 0.0011, "step": 49370 }, { "grad_norm": 0.042217329144477844, "learning_rate": 8.324985291036514e-06, "loss": 0.0011, "step": 49380 }, { "grad_norm": 0.05085426941514015, "learning_rate": 8.309765396500213e-06, "loss": 0.0008, "step": 49390 }, { "grad_norm": 0.04705827683210373, "learning_rate": 8.294558166356419e-06, "loss": 0.0008, "step": 49400 }, { "grad_norm": 0.07032257318496704, "learning_rate": 8.279363605224683e-06, "loss": 0.0014, "step": 49410 }, { "grad_norm": 0.05058501288294792, "learning_rate": 8.264181717720704e-06, "loss": 0.001, "step": 49420 }, { "grad_norm": 0.03242718055844307, "learning_rate": 8.249012508456361e-06, "loss": 0.0011, "step": 49430 }, { "grad_norm": 0.03309142217040062, "learning_rate": 8.233855982039646e-06, "loss": 0.0012, "step": 49440 }, { "grad_norm": 0.036096666008234024, "learning_rate": 8.218712143074708e-06, "loss": 0.001, "step": 49450 }, { "grad_norm": 0.03629326820373535, "learning_rate": 8.203580996161858e-06, "loss": 0.0009, "step": 49460 }, { "grad_norm": 0.08924391120672226, "learning_rate": 8.188462545897512e-06, "loss": 0.001, "step": 49470 }, { "grad_norm": 0.03040645457804203, "learning_rate": 8.173356796874304e-06, "loss": 0.001, "step": 49480 }, { "grad_norm": 0.05457489192485809, "learning_rate": 8.158263753680906e-06, "loss": 0.0013, "step": 49490 }, { "grad_norm": 0.033399466425180435, "learning_rate": 8.143183420902239e-06, "loss": 0.0008, "step": 49500 }, { "grad_norm": 0.035375576466321945, "learning_rate": 8.128115803119258e-06, "loss": 0.0009, "step": 49510 }, { "grad_norm": 0.032674286514520645, "learning_rate": 8.11306090490916e-06, "loss": 0.001, "step": 49520 }, { "grad_norm": 0.03723724186420441, "learning_rate": 8.098018730845169e-06, "loss": 0.001, "step": 49530 }, { "grad_norm": 0.04936983063817024, "learning_rate": 8.082989285496745e-06, "loss": 0.0009, "step": 49540 }, { "grad_norm": 0.05639168992638588, "learning_rate": 8.067972573429416e-06, "loss": 0.0015, "step": 49550 }, { "grad_norm": 0.04267599806189537, "learning_rate": 8.052968599204874e-06, "loss": 0.0011, "step": 49560 }, { "grad_norm": 0.037072762846946716, "learning_rate": 8.037977367380922e-06, "loss": 0.0011, "step": 49570 }, { "grad_norm": 0.04822554066777229, "learning_rate": 8.022998882511495e-06, "loss": 0.0014, "step": 49580 }, { "grad_norm": 0.0497615747153759, "learning_rate": 8.008033149146677e-06, "loss": 0.0018, "step": 49590 }, { "grad_norm": 0.04677431285381317, "learning_rate": 7.993080171832656e-06, "loss": 0.0012, "step": 49600 }, { "grad_norm": 0.02920694462954998, "learning_rate": 7.978139955111752e-06, "loss": 0.0009, "step": 49610 }, { "grad_norm": 0.03790777176618576, "learning_rate": 7.9632125035224e-06, "loss": 0.0011, "step": 49620 }, { "grad_norm": 0.028154432773590088, "learning_rate": 7.948297821599177e-06, "loss": 0.0008, "step": 49630 }, { "grad_norm": 0.043277643620967865, "learning_rate": 7.933395913872755e-06, "loss": 0.0015, "step": 49640 }, { "grad_norm": 0.05370832979679108, "learning_rate": 7.918506784869972e-06, "loss": 0.0013, "step": 49650 }, { "grad_norm": 0.03620884567499161, "learning_rate": 7.903630439113707e-06, "loss": 0.001, "step": 49660 }, { "grad_norm": 0.036187805235385895, "learning_rate": 7.888766881123044e-06, "loss": 0.0012, "step": 49670 }, { "grad_norm": 0.033008474856615067, "learning_rate": 7.873916115413099e-06, "loss": 0.001, "step": 49680 }, { "grad_norm": 0.05175033211708069, "learning_rate": 7.85907814649518e-06, "loss": 0.0013, "step": 49690 }, { "grad_norm": 0.03465165197849274, "learning_rate": 7.844252978876649e-06, "loss": 0.0013, "step": 49700 }, { "grad_norm": 0.033041611313819885, "learning_rate": 7.829440617061001e-06, "loss": 0.0012, "step": 49710 }, { "grad_norm": 0.02770611271262169, "learning_rate": 7.814641065547851e-06, "loss": 0.0013, "step": 49720 }, { "grad_norm": 0.04069388657808304, "learning_rate": 7.79985432883289e-06, "loss": 0.0017, "step": 49730 }, { "grad_norm": 0.029441464692354202, "learning_rate": 7.78508041140797e-06, "loss": 0.0008, "step": 49740 }, { "grad_norm": 0.060206491500139236, "learning_rate": 7.770319317760993e-06, "loss": 0.0016, "step": 49750 }, { "grad_norm": 0.030413251370191574, "learning_rate": 7.755571052376004e-06, "loss": 0.001, "step": 49760 }, { "grad_norm": 0.03142942115664482, "learning_rate": 7.740835619733128e-06, "loss": 0.0009, "step": 49770 }, { "grad_norm": 0.03539381921291351, "learning_rate": 7.726113024308601e-06, "loss": 0.0013, "step": 49780 }, { "grad_norm": 0.05234934389591217, "learning_rate": 7.711403270574746e-06, "loss": 0.0009, "step": 49790 }, { "grad_norm": 0.07625474780797958, "learning_rate": 7.696706363000039e-06, "loss": 0.0013, "step": 49800 }, { "grad_norm": 0.03313169255852699, "learning_rate": 7.682022306048959e-06, "loss": 0.0008, "step": 49810 }, { "grad_norm": 0.04055343568325043, "learning_rate": 7.667351104182186e-06, "loss": 0.0015, "step": 49820 }, { "grad_norm": 0.04845814034342766, "learning_rate": 7.652692761856395e-06, "loss": 0.0017, "step": 49830 }, { "grad_norm": 0.030376877635717392, "learning_rate": 7.63804728352444e-06, "loss": 0.0012, "step": 49840 }, { "grad_norm": 0.03815653547644615, "learning_rate": 7.623414673635215e-06, "loss": 0.0015, "step": 49850 }, { "grad_norm": 0.03223491087555885, "learning_rate": 7.608794936633723e-06, "loss": 0.0015, "step": 49860 }, { "grad_norm": 0.05234360322356224, "learning_rate": 7.594188076961056e-06, "loss": 0.0012, "step": 49870 }, { "grad_norm": 0.03155326098203659, "learning_rate": 7.579594099054382e-06, "loss": 0.0011, "step": 49880 }, { "grad_norm": 0.03108488954603672, "learning_rate": 7.565013007346983e-06, "loss": 0.0008, "step": 49890 }, { "grad_norm": 0.029694383963942528, "learning_rate": 7.5504448062682035e-06, "loss": 0.0009, "step": 49900 }, { "grad_norm": 0.027692699804902077, "learning_rate": 7.53588950024347e-06, "loss": 0.0008, "step": 49910 }, { "grad_norm": 0.06570006161928177, "learning_rate": 7.5213470936943145e-06, "loss": 0.0012, "step": 49920 }, { "grad_norm": 0.04623790085315704, "learning_rate": 7.506817591038323e-06, "loss": 0.0008, "step": 49930 }, { "grad_norm": 0.05528118461370468, "learning_rate": 7.492300996689183e-06, "loss": 0.0015, "step": 49940 }, { "grad_norm": 0.02915171906352043, "learning_rate": 7.477797315056645e-06, "loss": 0.0016, "step": 49950 }, { "grad_norm": 0.023781951516866684, "learning_rate": 7.463306550546539e-06, "loss": 0.0007, "step": 49960 }, { "grad_norm": 0.03472723066806793, "learning_rate": 7.448828707560812e-06, "loss": 0.001, "step": 49970 }, { "grad_norm": 0.035539280623197556, "learning_rate": 7.4343637904974e-06, "loss": 0.001, "step": 49980 }, { "grad_norm": 0.05584820732474327, "learning_rate": 7.419911803750401e-06, "loss": 0.0012, "step": 49990 }, { "grad_norm": 0.04276270046830177, "learning_rate": 7.405472751709935e-06, "loss": 0.0009, "step": 50000 }, { "grad_norm": 0.03144659474492073, "learning_rate": 7.3910466387622e-06, "loss": 0.0009, "step": 50010 }, { "grad_norm": 0.05289490520954132, "learning_rate": 7.3766334692894735e-06, "loss": 0.001, "step": 50020 }, { "grad_norm": 0.03841935098171234, "learning_rate": 7.3622332476700865e-06, "loss": 0.0008, "step": 50030 }, { "grad_norm": 0.040543340146541595, "learning_rate": 7.347845978278472e-06, "loss": 0.0015, "step": 50040 }, { "grad_norm": 0.04442823305726051, "learning_rate": 7.333471665485065e-06, "loss": 0.001, "step": 50050 }, { "grad_norm": 0.04929330199956894, "learning_rate": 7.31911031365643e-06, "loss": 0.0016, "step": 50060 }, { "grad_norm": 0.04348622262477875, "learning_rate": 7.304761927155157e-06, "loss": 0.0013, "step": 50070 }, { "grad_norm": 0.03769285976886749, "learning_rate": 7.29042651033991e-06, "loss": 0.0013, "step": 50080 }, { "grad_norm": 0.03525124117732048, "learning_rate": 7.276104067565409e-06, "loss": 0.001, "step": 50090 }, { "grad_norm": 0.025726255029439926, "learning_rate": 7.261794603182431e-06, "loss": 0.001, "step": 50100 }, { "grad_norm": 0.03237666189670563, "learning_rate": 7.24749812153781e-06, "loss": 0.001, "step": 50110 }, { "grad_norm": 0.02687092497944832, "learning_rate": 7.2332146269744605e-06, "loss": 0.001, "step": 50120 }, { "grad_norm": 0.03052232414484024, "learning_rate": 7.218944123831295e-06, "loss": 0.0011, "step": 50130 }, { "grad_norm": 0.06985855102539062, "learning_rate": 7.204686616443351e-06, "loss": 0.0011, "step": 50140 }, { "grad_norm": 0.048115335404872894, "learning_rate": 7.190442109141665e-06, "loss": 0.0011, "step": 50150 }, { "grad_norm": 0.03901464864611626, "learning_rate": 7.176210606253347e-06, "loss": 0.0009, "step": 50160 }, { "grad_norm": 0.028224704787135124, "learning_rate": 7.161992112101551e-06, "loss": 0.0012, "step": 50170 }, { "grad_norm": 0.04551354795694351, "learning_rate": 7.147786631005465e-06, "loss": 0.0009, "step": 50180 }, { "grad_norm": 0.028778553009033203, "learning_rate": 7.1335941672803775e-06, "loss": 0.0014, "step": 50190 }, { "grad_norm": 0.036041732877492905, "learning_rate": 7.1194147252375384e-06, "loss": 0.0017, "step": 50200 }, { "grad_norm": 0.03774736076593399, "learning_rate": 7.10524830918432e-06, "loss": 0.0008, "step": 50210 }, { "grad_norm": 0.03623029217123985, "learning_rate": 7.091094923424097e-06, "loss": 0.0011, "step": 50220 }, { "grad_norm": 0.030503040179610252, "learning_rate": 7.0769545722562894e-06, "loss": 0.0008, "step": 50230 }, { "grad_norm": 0.041251830756664276, "learning_rate": 7.0628272599763675e-06, "loss": 0.0008, "step": 50240 }, { "grad_norm": 0.02141854166984558, "learning_rate": 7.048712990875828e-06, "loss": 0.0008, "step": 50250 }, { "grad_norm": 0.027424197643995285, "learning_rate": 7.034611769242216e-06, "loss": 0.001, "step": 50260 }, { "grad_norm": 0.03378599137067795, "learning_rate": 7.02052359935913e-06, "loss": 0.0012, "step": 50270 }, { "grad_norm": 0.026700271293520927, "learning_rate": 7.006448485506145e-06, "loss": 0.0012, "step": 50280 }, { "grad_norm": 0.03944316878914833, "learning_rate": 6.992386431958942e-06, "loss": 0.001, "step": 50290 }, { "grad_norm": 0.056664206087589264, "learning_rate": 6.978337442989197e-06, "loss": 0.0011, "step": 50300 }, { "grad_norm": 0.01944173313677311, "learning_rate": 6.964301522864608e-06, "loss": 0.0007, "step": 50310 }, { "grad_norm": 0.016171472147107124, "learning_rate": 6.950278675848926e-06, "loss": 0.0012, "step": 50320 }, { "grad_norm": 0.06155562028288841, "learning_rate": 6.9362689062019145e-06, "loss": 0.0011, "step": 50330 }, { "grad_norm": 0.02293674275279045, "learning_rate": 6.922272218179393e-06, "loss": 0.001, "step": 50340 }, { "grad_norm": 0.02910732291638851, "learning_rate": 6.908288616033148e-06, "loss": 0.0008, "step": 50350 }, { "grad_norm": 0.05306606367230415, "learning_rate": 6.894318104011077e-06, "loss": 0.0013, "step": 50360 }, { "grad_norm": 0.08497066795825958, "learning_rate": 6.880360686357007e-06, "loss": 0.0017, "step": 50370 }, { "grad_norm": 0.03849969804286957, "learning_rate": 6.8664163673108575e-06, "loss": 0.0011, "step": 50380 }, { "grad_norm": 0.026827113702893257, "learning_rate": 6.85248515110854e-06, "loss": 0.0009, "step": 50390 }, { "grad_norm": 0.03176678344607353, "learning_rate": 6.838567041981992e-06, "loss": 0.001, "step": 50400 }, { "grad_norm": 0.05933787301182747, "learning_rate": 6.8246620441591634e-06, "loss": 0.0012, "step": 50410 }, { "grad_norm": 0.04096159338951111, "learning_rate": 6.8107701618640275e-06, "loss": 0.001, "step": 50420 }, { "grad_norm": 0.03270798176527023, "learning_rate": 6.796891399316557e-06, "loss": 0.0009, "step": 50430 }, { "grad_norm": 0.015483078546822071, "learning_rate": 6.7830257607327804e-06, "loss": 0.0007, "step": 50440 }, { "grad_norm": 0.05662526190280914, "learning_rate": 6.7691732503247e-06, "loss": 0.0011, "step": 50450 }, { "grad_norm": 0.026355786249041557, "learning_rate": 6.755333872300346e-06, "loss": 0.0018, "step": 50460 }, { "grad_norm": 0.03910528123378754, "learning_rate": 6.741507630863747e-06, "loss": 0.002, "step": 50470 }, { "grad_norm": 0.04867406561970711, "learning_rate": 6.727694530214945e-06, "loss": 0.0011, "step": 50480 }, { "grad_norm": 0.03979106619954109, "learning_rate": 6.713894574550028e-06, "loss": 0.0009, "step": 50490 }, { "grad_norm": 0.026513051241636276, "learning_rate": 6.700107768061015e-06, "loss": 0.0007, "step": 50500 }, { "grad_norm": 0.048517856746912, "learning_rate": 6.686334114936016e-06, "loss": 0.0016, "step": 50510 }, { "grad_norm": 0.03575950115919113, "learning_rate": 6.672573619359063e-06, "loss": 0.0009, "step": 50520 }, { "grad_norm": 0.0559760257601738, "learning_rate": 6.658826285510256e-06, "loss": 0.001, "step": 50530 }, { "grad_norm": 0.018350739032030106, "learning_rate": 6.645092117565666e-06, "loss": 0.0012, "step": 50540 }, { "grad_norm": 0.024899881333112717, "learning_rate": 6.631371119697371e-06, "loss": 0.001, "step": 50550 }, { "grad_norm": 0.027596695348620415, "learning_rate": 6.6176632960734505e-06, "loss": 0.001, "step": 50560 }, { "grad_norm": 0.05009577423334122, "learning_rate": 6.603968650857978e-06, "loss": 0.0013, "step": 50570 }, { "grad_norm": 0.039638061076402664, "learning_rate": 6.5902871882110085e-06, "loss": 0.0009, "step": 50580 }, { "grad_norm": 0.03571370616555214, "learning_rate": 6.576618912288635e-06, "loss": 0.001, "step": 50590 }, { "grad_norm": 0.03530051186680794, "learning_rate": 6.562963827242913e-06, "loss": 0.0009, "step": 50600 }, { "grad_norm": 0.047844886779785156, "learning_rate": 6.549321937221886e-06, "loss": 0.0014, "step": 50610 }, { "grad_norm": 0.02647819183766842, "learning_rate": 6.5356932463696064e-06, "loss": 0.0012, "step": 50620 }, { "grad_norm": 0.054715804755687714, "learning_rate": 6.522077758826101e-06, "loss": 0.0011, "step": 50630 }, { "grad_norm": 0.032782211899757385, "learning_rate": 6.5084754787274275e-06, "loss": 0.0011, "step": 50640 }, { "grad_norm": 0.026713484898209572, "learning_rate": 6.494886410205553e-06, "loss": 0.0017, "step": 50650 }, { "grad_norm": 0.031033499166369438, "learning_rate": 6.481310557388521e-06, "loss": 0.0007, "step": 50660 }, { "grad_norm": 0.037284865975379944, "learning_rate": 6.46774792440028e-06, "loss": 0.0008, "step": 50670 }, { "grad_norm": 0.057025521993637085, "learning_rate": 6.4541985153608206e-06, "loss": 0.0009, "step": 50680 }, { "grad_norm": 0.018161538988351822, "learning_rate": 6.4406623343861e-06, "loss": 0.0007, "step": 50690 }, { "grad_norm": 0.03402042016386986, "learning_rate": 6.427139385588038e-06, "loss": 0.0009, "step": 50700 }, { "grad_norm": 0.06420837342739105, "learning_rate": 6.413629673074561e-06, "loss": 0.0009, "step": 50710 }, { "grad_norm": 0.0346810445189476, "learning_rate": 6.400133200949554e-06, "loss": 0.0016, "step": 50720 }, { "grad_norm": 0.02984681725502014, "learning_rate": 6.386649973312897e-06, "loss": 0.001, "step": 50730 }, { "grad_norm": 0.03460761532187462, "learning_rate": 6.37317999426042e-06, "loss": 0.0008, "step": 50740 }, { "grad_norm": 0.06418287009000778, "learning_rate": 6.359723267883977e-06, "loss": 0.0015, "step": 50750 }, { "grad_norm": 0.05499419569969177, "learning_rate": 6.346279798271343e-06, "loss": 0.0018, "step": 50760 }, { "grad_norm": 0.036941323429346085, "learning_rate": 6.332849589506301e-06, "loss": 0.0016, "step": 50770 }, { "grad_norm": 0.050626542419195175, "learning_rate": 6.319432645668588e-06, "loss": 0.0017, "step": 50780 }, { "grad_norm": 0.07420223206281662, "learning_rate": 6.306028970833922e-06, "loss": 0.0013, "step": 50790 }, { "grad_norm": 0.05079556629061699, "learning_rate": 6.2926385690739665e-06, "loss": 0.0013, "step": 50800 }, { "grad_norm": 0.035004738718271255, "learning_rate": 6.279261444456413e-06, "loss": 0.0008, "step": 50810 }, { "grad_norm": 0.04946739226579666, "learning_rate": 6.265897601044829e-06, "loss": 0.0015, "step": 50820 }, { "grad_norm": 0.02645600400865078, "learning_rate": 6.2525470428988434e-06, "loss": 0.0008, "step": 50830 }, { "grad_norm": 0.05112334340810776, "learning_rate": 6.239209774073962e-06, "loss": 0.001, "step": 50840 }, { "grad_norm": 0.025018814951181412, "learning_rate": 6.225885798621728e-06, "loss": 0.0008, "step": 50850 }, { "grad_norm": 0.03479260578751564, "learning_rate": 6.2125751205895925e-06, "loss": 0.0016, "step": 50860 }, { "grad_norm": 0.026923982426524162, "learning_rate": 6.199277744020998e-06, "loss": 0.001, "step": 50870 }, { "grad_norm": 0.03799337521195412, "learning_rate": 6.185993672955337e-06, "loss": 0.001, "step": 50880 }, { "grad_norm": 0.03840490058064461, "learning_rate": 6.172722911427947e-06, "loss": 0.0009, "step": 50890 }, { "grad_norm": 0.05276567488908768, "learning_rate": 6.159465463470149e-06, "loss": 0.0008, "step": 50900 }, { "grad_norm": 0.03157708793878555, "learning_rate": 6.146221333109204e-06, "loss": 0.0009, "step": 50910 }, { "grad_norm": 0.037033021450042725, "learning_rate": 6.132990524368326e-06, "loss": 0.0015, "step": 50920 }, { "grad_norm": 0.04049244150519371, "learning_rate": 6.119773041266685e-06, "loss": 0.0017, "step": 50930 }, { "grad_norm": 0.05320831760764122, "learning_rate": 6.106568887819402e-06, "loss": 0.0014, "step": 50940 }, { "grad_norm": 0.03839888051152229, "learning_rate": 6.093378068037548e-06, "loss": 0.0012, "step": 50950 }, { "grad_norm": 0.03476700931787491, "learning_rate": 6.080200585928164e-06, "loss": 0.0011, "step": 50960 }, { "grad_norm": 0.0446372926235199, "learning_rate": 6.06703644549419e-06, "loss": 0.0009, "step": 50970 }, { "grad_norm": 0.03485206142067909, "learning_rate": 6.053885650734576e-06, "loss": 0.0012, "step": 50980 }, { "grad_norm": 0.05564066395163536, "learning_rate": 6.040748205644153e-06, "loss": 0.0011, "step": 50990 }, { "grad_norm": 0.06564692407846451, "learning_rate": 6.0276241142137646e-06, "loss": 0.0012, "step": 51000 }, { "grad_norm": 0.027377380058169365, "learning_rate": 6.014513380430142e-06, "loss": 0.0008, "step": 51010 }, { "grad_norm": 0.03304710611701012, "learning_rate": 6.001416008275984e-06, "loss": 0.0007, "step": 51020 }, { "grad_norm": 0.03786558657884598, "learning_rate": 5.988332001729929e-06, "loss": 0.0009, "step": 51030 }, { "grad_norm": 0.030517028644680977, "learning_rate": 5.975261364766543e-06, "loss": 0.0008, "step": 51040 }, { "grad_norm": 0.05982125923037529, "learning_rate": 5.962204101356356e-06, "loss": 0.001, "step": 51050 }, { "grad_norm": 0.06052187830209732, "learning_rate": 5.94916021546581e-06, "loss": 0.001, "step": 51060 }, { "grad_norm": 0.02700977399945259, "learning_rate": 5.936129711057298e-06, "loss": 0.0011, "step": 51070 }, { "grad_norm": 0.022184329107403755, "learning_rate": 5.923112592089142e-06, "loss": 0.001, "step": 51080 }, { "grad_norm": 0.02318272367119789, "learning_rate": 5.9101088625155954e-06, "loss": 0.0008, "step": 51090 }, { "grad_norm": 0.05629848688840866, "learning_rate": 5.897118526286843e-06, "loss": 0.0011, "step": 51100 }, { "grad_norm": 0.0453752838075161, "learning_rate": 5.884141587349035e-06, "loss": 0.0009, "step": 51110 }, { "grad_norm": 0.04641801118850708, "learning_rate": 5.871178049644177e-06, "loss": 0.0009, "step": 51120 }, { "grad_norm": 0.03886459395289421, "learning_rate": 5.858227917110293e-06, "loss": 0.001, "step": 51130 }, { "grad_norm": 0.021676255390048027, "learning_rate": 5.845291193681252e-06, "loss": 0.0008, "step": 51140 }, { "grad_norm": 0.033031512051820755, "learning_rate": 5.832367883286921e-06, "loss": 0.001, "step": 51150 }, { "grad_norm": 0.028898030519485474, "learning_rate": 5.81945798985305e-06, "loss": 0.0011, "step": 51160 }, { "grad_norm": 0.04366634786128998, "learning_rate": 5.806561517301306e-06, "loss": 0.0016, "step": 51170 }, { "grad_norm": 0.029027581214904785, "learning_rate": 5.793678469549335e-06, "loss": 0.0012, "step": 51180 }, { "grad_norm": 0.035512715578079224, "learning_rate": 5.780808850510627e-06, "loss": 0.0008, "step": 51190 }, { "grad_norm": 0.03581016883254051, "learning_rate": 5.767952664094673e-06, "loss": 0.0007, "step": 51200 }, { "grad_norm": 0.028996065258979797, "learning_rate": 5.755109914206791e-06, "loss": 0.0008, "step": 51210 }, { "grad_norm": 0.0345594696700573, "learning_rate": 5.7422806047483125e-06, "loss": 0.0014, "step": 51220 }, { "grad_norm": 0.053586870431900024, "learning_rate": 5.72946473961643e-06, "loss": 0.0012, "step": 51230 }, { "grad_norm": 0.02885325811803341, "learning_rate": 5.716662322704264e-06, "loss": 0.001, "step": 51240 }, { "grad_norm": 0.0200754813849926, "learning_rate": 5.703873357900852e-06, "loss": 0.001, "step": 51250 }, { "grad_norm": 0.03462424874305725, "learning_rate": 5.691097849091143e-06, "loss": 0.0011, "step": 51260 }, { "grad_norm": 0.039232395589351654, "learning_rate": 5.678335800155982e-06, "loss": 0.0008, "step": 51270 }, { "grad_norm": 0.024856548756361008, "learning_rate": 5.665587214972174e-06, "loss": 0.0009, "step": 51280 }, { "grad_norm": 0.02370898239314556, "learning_rate": 5.652852097412386e-06, "loss": 0.0008, "step": 51290 }, { "grad_norm": 0.03898333013057709, "learning_rate": 5.640130451345216e-06, "loss": 0.0014, "step": 51300 }, { "grad_norm": 0.02949342504143715, "learning_rate": 5.627422280635159e-06, "loss": 0.0011, "step": 51310 }, { "grad_norm": 0.032651327550411224, "learning_rate": 5.61472758914261e-06, "loss": 0.0008, "step": 51320 }, { "grad_norm": 0.03346109762787819, "learning_rate": 5.602046380723918e-06, "loss": 0.0008, "step": 51330 }, { "grad_norm": 0.058472562581300735, "learning_rate": 5.5893786592312535e-06, "loss": 0.0011, "step": 51340 }, { "grad_norm": 0.04178192466497421, "learning_rate": 5.576724428512775e-06, "loss": 0.001, "step": 51350 }, { "grad_norm": 0.046703845262527466, "learning_rate": 5.56408369241247e-06, "loss": 0.0009, "step": 51360 }, { "grad_norm": 0.03267389535903931, "learning_rate": 5.5514564547702875e-06, "loss": 0.0009, "step": 51370 }, { "grad_norm": 0.05766155198216438, "learning_rate": 5.538842719422038e-06, "loss": 0.0018, "step": 51380 }, { "grad_norm": 0.019057534635066986, "learning_rate": 5.52624249019944e-06, "loss": 0.0008, "step": 51390 }, { "grad_norm": 0.023077435791492462, "learning_rate": 5.51365577093011e-06, "loss": 0.0009, "step": 51400 }, { "grad_norm": 0.06618846207857132, "learning_rate": 5.501082565437565e-06, "loss": 0.0009, "step": 51410 }, { "grad_norm": 0.02151825837790966, "learning_rate": 5.488522877541202e-06, "loss": 0.0008, "step": 51420 }, { "grad_norm": 0.04139106720685959, "learning_rate": 5.475976711056341e-06, "loss": 0.001, "step": 51430 }, { "grad_norm": 0.02528640255331993, "learning_rate": 5.463444069794166e-06, "loss": 0.0008, "step": 51440 }, { "grad_norm": 0.023754820227622986, "learning_rate": 5.4509249575617594e-06, "loss": 0.0013, "step": 51450 }, { "grad_norm": 0.07106181979179382, "learning_rate": 5.438419378162107e-06, "loss": 0.0009, "step": 51460 }, { "grad_norm": 0.05136674642562866, "learning_rate": 5.425927335394054e-06, "loss": 0.0011, "step": 51470 }, { "grad_norm": 0.04923127219080925, "learning_rate": 5.413448833052387e-06, "loss": 0.0012, "step": 51480 }, { "grad_norm": 0.03429859131574631, "learning_rate": 5.400983874927701e-06, "loss": 0.0011, "step": 51490 }, { "grad_norm": 0.03772222250699997, "learning_rate": 5.388532464806567e-06, "loss": 0.0009, "step": 51500 }, { "grad_norm": 0.02274983748793602, "learning_rate": 5.3760946064713546e-06, "loss": 0.0009, "step": 51510 }, { "grad_norm": 0.03136491775512695, "learning_rate": 5.363670303700386e-06, "loss": 0.0008, "step": 51520 }, { "grad_norm": 0.028152985498309135, "learning_rate": 5.351259560267824e-06, "loss": 0.0009, "step": 51530 }, { "grad_norm": 0.02351577952504158, "learning_rate": 5.338862379943721e-06, "loss": 0.0022, "step": 51540 }, { "grad_norm": 0.021188246086239815, "learning_rate": 5.326478766494025e-06, "loss": 0.0008, "step": 51550 }, { "grad_norm": 0.041034962981939316, "learning_rate": 5.3141087236805385e-06, "loss": 0.0011, "step": 51560 }, { "grad_norm": 0.0372779406607151, "learning_rate": 5.3017522552609615e-06, "loss": 0.0011, "step": 51570 }, { "grad_norm": 0.018901396542787552, "learning_rate": 5.289409364988851e-06, "loss": 0.001, "step": 51580 }, { "grad_norm": 0.03086313232779503, "learning_rate": 5.277080056613671e-06, "loss": 0.0008, "step": 51590 }, { "grad_norm": 0.02911597490310669, "learning_rate": 5.264764333880729e-06, "loss": 0.001, "step": 51600 }, { "grad_norm": 0.0953865721821785, "learning_rate": 5.252462200531216e-06, "loss": 0.0016, "step": 51610 }, { "grad_norm": 0.05548804998397827, "learning_rate": 5.240173660302194e-06, "loss": 0.001, "step": 51620 }, { "grad_norm": 0.04323357716202736, "learning_rate": 5.2278987169266044e-06, "loss": 0.0013, "step": 51630 }, { "grad_norm": 0.046884406358003616, "learning_rate": 5.215637374133231e-06, "loss": 0.0013, "step": 51640 }, { "grad_norm": 0.03671465814113617, "learning_rate": 5.203389635646782e-06, "loss": 0.0014, "step": 51650 }, { "grad_norm": 0.03485998138785362, "learning_rate": 5.191155505187756e-06, "loss": 0.0008, "step": 51660 }, { "grad_norm": 0.03474303334951401, "learning_rate": 5.1789349864726e-06, "loss": 0.0011, "step": 51670 }, { "grad_norm": 0.048340849578380585, "learning_rate": 5.16672808321354e-06, "loss": 0.0009, "step": 51680 }, { "grad_norm": 0.02048676088452339, "learning_rate": 5.154534799118749e-06, "loss": 0.001, "step": 51690 }, { "grad_norm": 0.06153416261076927, "learning_rate": 5.142355137892207e-06, "loss": 0.001, "step": 51700 }, { "grad_norm": 0.048345934599637985, "learning_rate": 5.130189103233779e-06, "loss": 0.001, "step": 51710 }, { "grad_norm": 0.050894543528556824, "learning_rate": 5.118036698839179e-06, "loss": 0.0013, "step": 51720 }, { "grad_norm": 0.025288050994277, "learning_rate": 5.105897928399983e-06, "loss": 0.001, "step": 51730 }, { "grad_norm": 0.033876724541187286, "learning_rate": 5.0937727956036464e-06, "loss": 0.001, "step": 51740 }, { "grad_norm": 0.028628932312130928, "learning_rate": 5.081661304133456e-06, "loss": 0.001, "step": 51750 }, { "grad_norm": 0.018214719370007515, "learning_rate": 5.069563457668558e-06, "loss": 0.0008, "step": 51760 }, { "grad_norm": 0.020574484020471573, "learning_rate": 5.0574792598839624e-06, "loss": 0.001, "step": 51770 }, { "grad_norm": 0.04904620349407196, "learning_rate": 5.0454087144505276e-06, "loss": 0.0007, "step": 51780 }, { "grad_norm": 0.024713266640901566, "learning_rate": 5.0333518250349655e-06, "loss": 0.0011, "step": 51790 }, { "grad_norm": 0.02963227592408657, "learning_rate": 5.021308595299856e-06, "loss": 0.0008, "step": 51800 }, { "grad_norm": 0.037518542259931564, "learning_rate": 5.009279028903585e-06, "loss": 0.0007, "step": 51810 }, { "grad_norm": 0.0364399254322052, "learning_rate": 4.997263129500452e-06, "loss": 0.0008, "step": 51820 }, { "grad_norm": 0.02677188813686371, "learning_rate": 4.985260900740535e-06, "loss": 0.0008, "step": 51830 }, { "grad_norm": 0.03390787914395332, "learning_rate": 4.973272346269814e-06, "loss": 0.001, "step": 51840 }, { "grad_norm": 0.03238474205136299, "learning_rate": 4.961297469730097e-06, "loss": 0.0012, "step": 51850 }, { "grad_norm": 0.025874929502606392, "learning_rate": 4.949336274759031e-06, "loss": 0.0009, "step": 51860 }, { "grad_norm": 0.06577607989311218, "learning_rate": 4.9373887649901144e-06, "loss": 0.0017, "step": 51870 }, { "grad_norm": 0.04045825079083443, "learning_rate": 4.925454944052666e-06, "loss": 0.0009, "step": 51880 }, { "grad_norm": 0.030475474894046783, "learning_rate": 4.913534815571891e-06, "loss": 0.0007, "step": 51890 }, { "grad_norm": 0.039058949798345566, "learning_rate": 4.901628383168805e-06, "loss": 0.0009, "step": 51900 }, { "grad_norm": 0.028824755921959877, "learning_rate": 4.8897356504602585e-06, "loss": 0.0009, "step": 51910 }, { "grad_norm": 0.01848701387643814, "learning_rate": 4.877856621058957e-06, "loss": 0.0007, "step": 51920 }, { "grad_norm": 0.028456691652536392, "learning_rate": 4.86599129857343e-06, "loss": 0.0011, "step": 51930 }, { "grad_norm": 0.03465249761939049, "learning_rate": 4.85413968660805e-06, "loss": 0.0008, "step": 51940 }, { "grad_norm": 0.035051990300416946, "learning_rate": 4.842301788763031e-06, "loss": 0.0008, "step": 51950 }, { "grad_norm": 0.02100704237818718, "learning_rate": 4.830477608634393e-06, "loss": 0.0014, "step": 51960 }, { "grad_norm": 0.02214948646724224, "learning_rate": 4.818667149814049e-06, "loss": 0.0009, "step": 51970 }, { "grad_norm": 0.018639635294675827, "learning_rate": 4.806870415889664e-06, "loss": 0.0015, "step": 51980 }, { "grad_norm": 0.03579149395227432, "learning_rate": 4.795087410444798e-06, "loss": 0.001, "step": 51990 }, { "grad_norm": 0.03561833128333092, "learning_rate": 4.783318137058807e-06, "loss": 0.0012, "step": 52000 }, { "grad_norm": 0.03719456493854523, "learning_rate": 4.771562599306895e-06, "loss": 0.0007, "step": 52010 }, { "grad_norm": 0.06411419063806534, "learning_rate": 4.759820800760073e-06, "loss": 0.0009, "step": 52020 }, { "grad_norm": 0.04117754101753235, "learning_rate": 4.7480927449851834e-06, "loss": 0.0011, "step": 52030 }, { "grad_norm": 0.0651952251791954, "learning_rate": 4.73637843554493e-06, "loss": 0.0013, "step": 52040 }, { "grad_norm": 0.04223375767469406, "learning_rate": 4.724677875997774e-06, "loss": 0.0008, "step": 52050 }, { "grad_norm": 0.02022809535264969, "learning_rate": 4.712991069898065e-06, "loss": 0.001, "step": 52060 }, { "grad_norm": 0.023807860910892487, "learning_rate": 4.7013180207959305e-06, "loss": 0.0009, "step": 52070 }, { "grad_norm": 0.024473868310451508, "learning_rate": 4.6896587322373395e-06, "loss": 0.0005, "step": 52080 }, { "grad_norm": 0.03677014261484146, "learning_rate": 4.678013207764081e-06, "loss": 0.0012, "step": 52090 }, { "grad_norm": 0.058601219207048416, "learning_rate": 4.666381450913748e-06, "loss": 0.0007, "step": 52100 }, { "grad_norm": 0.030743133276700974, "learning_rate": 4.654763465219752e-06, "loss": 0.0007, "step": 52110 }, { "grad_norm": 0.019609151408076286, "learning_rate": 4.643159254211371e-06, "loss": 0.0009, "step": 52120 }, { "grad_norm": 0.027827218174934387, "learning_rate": 4.631568821413606e-06, "loss": 0.0011, "step": 52130 }, { "grad_norm": 0.038754984736442566, "learning_rate": 4.619992170347359e-06, "loss": 0.0007, "step": 52140 }, { "grad_norm": 0.02985881268978119, "learning_rate": 4.608429304529305e-06, "loss": 0.0007, "step": 52150 }, { "grad_norm": 0.02549574337899685, "learning_rate": 4.596880227471928e-06, "loss": 0.0007, "step": 52160 }, { "grad_norm": 0.023144209757447243, "learning_rate": 4.585344942683539e-06, "loss": 0.001, "step": 52170 }, { "grad_norm": 0.025469643995165825, "learning_rate": 4.573823453668241e-06, "loss": 0.001, "step": 52180 }, { "grad_norm": 0.028795862570405006, "learning_rate": 4.562315763925995e-06, "loss": 0.0007, "step": 52190 }, { "grad_norm": 0.021794026717543602, "learning_rate": 4.5508218769524825e-06, "loss": 0.0015, "step": 52200 }, { "grad_norm": 0.021300388500094414, "learning_rate": 4.539341796239277e-06, "loss": 0.0008, "step": 52210 }, { "grad_norm": 0.02159082144498825, "learning_rate": 4.527875525273717e-06, "loss": 0.0013, "step": 52220 }, { "grad_norm": 0.05457726866006851, "learning_rate": 4.51642306753895e-06, "loss": 0.0017, "step": 52230 }, { "grad_norm": 0.016506901010870934, "learning_rate": 4.5049844265139306e-06, "loss": 0.0007, "step": 52240 }, { "grad_norm": 0.01844743639230728, "learning_rate": 4.4935596056734144e-06, "loss": 0.0008, "step": 52250 }, { "grad_norm": 0.06515755504369736, "learning_rate": 4.482148608487957e-06, "loss": 0.0009, "step": 52260 }, { "grad_norm": 0.037729114294052124, "learning_rate": 4.4707514384239365e-06, "loss": 0.001, "step": 52270 }, { "grad_norm": 0.04904092475771904, "learning_rate": 4.459368098943484e-06, "loss": 0.001, "step": 52280 }, { "grad_norm": 0.03515459597110748, "learning_rate": 4.447998593504582e-06, "loss": 0.001, "step": 52290 }, { "grad_norm": 0.02574360929429531, "learning_rate": 4.4366429255609744e-06, "loss": 0.0012, "step": 52300 }, { "grad_norm": 0.02125648222863674, "learning_rate": 4.425301098562212e-06, "loss": 0.0007, "step": 52310 }, { "grad_norm": 0.022435160353779793, "learning_rate": 4.413973115953651e-06, "loss": 0.0011, "step": 52320 }, { "grad_norm": 0.026370972394943237, "learning_rate": 4.402658981176416e-06, "loss": 0.001, "step": 52330 }, { "grad_norm": 0.02275858074426651, "learning_rate": 4.391358697667475e-06, "loss": 0.0008, "step": 52340 }, { "grad_norm": 0.046149637550115585, "learning_rate": 4.3800722688595195e-06, "loss": 0.0011, "step": 52350 }, { "grad_norm": 0.020193492993712425, "learning_rate": 4.368799698181097e-06, "loss": 0.0009, "step": 52360 }, { "grad_norm": 0.029868030920624733, "learning_rate": 4.357540989056486e-06, "loss": 0.001, "step": 52370 }, { "grad_norm": 0.0408797524869442, "learning_rate": 4.346296144905815e-06, "loss": 0.001, "step": 52380 }, { "grad_norm": 0.06763848662376404, "learning_rate": 4.335065169144958e-06, "loss": 0.0011, "step": 52390 }, { "grad_norm": 0.024690521880984306, "learning_rate": 4.323848065185593e-06, "loss": 0.0011, "step": 52400 }, { "grad_norm": 0.05153069645166397, "learning_rate": 4.31264483643517e-06, "loss": 0.0009, "step": 52410 }, { "grad_norm": 0.04338057339191437, "learning_rate": 4.301455486296946e-06, "loss": 0.0013, "step": 52420 }, { "grad_norm": 0.01745661161839962, "learning_rate": 4.290280018169935e-06, "loss": 0.0008, "step": 52430 }, { "grad_norm": 0.023679692298173904, "learning_rate": 4.27911843544897e-06, "loss": 0.0008, "step": 52440 }, { "grad_norm": 0.02597714215517044, "learning_rate": 4.2679707415246294e-06, "loss": 0.0013, "step": 52450 }, { "grad_norm": 0.022787585854530334, "learning_rate": 4.256836939783299e-06, "loss": 0.0007, "step": 52460 }, { "grad_norm": 0.029859058558940887, "learning_rate": 4.245717033607127e-06, "loss": 0.0009, "step": 52470 }, { "grad_norm": 0.02980623207986355, "learning_rate": 4.234611026374035e-06, "loss": 0.0008, "step": 52480 }, { "grad_norm": 0.02946389652788639, "learning_rate": 4.2235189214577694e-06, "loss": 0.0011, "step": 52490 }, { "grad_norm": 0.0372425876557827, "learning_rate": 4.212440722227779e-06, "loss": 0.0011, "step": 52500 }, { "grad_norm": 0.024833273142576218, "learning_rate": 4.201376432049364e-06, "loss": 0.0007, "step": 52510 }, { "grad_norm": 0.0739869773387909, "learning_rate": 4.1903260542835275e-06, "loss": 0.0014, "step": 52520 }, { "grad_norm": 0.027969108894467354, "learning_rate": 4.1792895922871114e-06, "loss": 0.0009, "step": 52530 }, { "grad_norm": 0.03108866699039936, "learning_rate": 4.168267049412694e-06, "loss": 0.0012, "step": 52540 }, { "grad_norm": 0.026513971388339996, "learning_rate": 4.157258429008626e-06, "loss": 0.0007, "step": 52550 }, { "grad_norm": 0.04113384708762169, "learning_rate": 4.146263734419043e-06, "loss": 0.001, "step": 52560 }, { "grad_norm": 0.043070968240499496, "learning_rate": 4.135282968983839e-06, "loss": 0.001, "step": 52570 }, { "grad_norm": 0.036217525601387024, "learning_rate": 4.124316136038675e-06, "loss": 0.0012, "step": 52580 }, { "grad_norm": 0.0732419416308403, "learning_rate": 4.1133632389149965e-06, "loss": 0.0012, "step": 52590 }, { "grad_norm": 0.030316563323140144, "learning_rate": 4.102424280939998e-06, "loss": 0.001, "step": 52600 }, { "grad_norm": 0.028535569086670876, "learning_rate": 4.091499265436649e-06, "loss": 0.0007, "step": 52610 }, { "grad_norm": 0.03292147070169449, "learning_rate": 4.080588195723684e-06, "loss": 0.0012, "step": 52620 }, { "grad_norm": 0.0357048399746418, "learning_rate": 4.069691075115578e-06, "loss": 0.0009, "step": 52630 }, { "grad_norm": 0.044138822704553604, "learning_rate": 4.0588079069226235e-06, "loss": 0.0013, "step": 52640 }, { "grad_norm": 0.048602379858493805, "learning_rate": 4.0479386944508034e-06, "loss": 0.0007, "step": 52650 }, { "grad_norm": 0.04721556976437569, "learning_rate": 4.037083441001932e-06, "loss": 0.0014, "step": 52660 }, { "grad_norm": 0.04222819209098816, "learning_rate": 4.026242149873516e-06, "loss": 0.001, "step": 52670 }, { "grad_norm": 0.032119493931531906, "learning_rate": 4.015414824358871e-06, "loss": 0.0025, "step": 52680 }, { "grad_norm": 0.024459054693579674, "learning_rate": 4.004601467747054e-06, "loss": 0.0008, "step": 52690 }, { "grad_norm": 0.03724493086338043, "learning_rate": 3.993802083322873e-06, "loss": 0.0009, "step": 52700 }, { "grad_norm": 0.025870224460959435, "learning_rate": 3.9830166743668906e-06, "loss": 0.0008, "step": 52710 }, { "grad_norm": 0.04686817154288292, "learning_rate": 3.9722452441554425e-06, "loss": 0.0011, "step": 52720 }, { "grad_norm": 0.04742765799164772, "learning_rate": 3.961487795960584e-06, "loss": 0.0008, "step": 52730 }, { "grad_norm": 0.0309053473174572, "learning_rate": 3.95074433305016e-06, "loss": 0.0011, "step": 52740 }, { "grad_norm": 0.050778578966856, "learning_rate": 3.940014858687752e-06, "loss": 0.0013, "step": 52750 }, { "grad_norm": 0.040192678570747375, "learning_rate": 3.929299376132689e-06, "loss": 0.0011, "step": 52760 }, { "grad_norm": 0.024980036541819572, "learning_rate": 3.918597888640047e-06, "loss": 0.0008, "step": 52770 }, { "grad_norm": 0.022617200389504433, "learning_rate": 3.907910399460657e-06, "loss": 0.0009, "step": 52780 }, { "grad_norm": 0.024453725665807724, "learning_rate": 3.8972369118410956e-06, "loss": 0.0009, "step": 52790 }, { "grad_norm": 0.040228310972452164, "learning_rate": 3.88657742902368e-06, "loss": 0.0009, "step": 52800 }, { "grad_norm": 0.023424575105309486, "learning_rate": 3.875931954246504e-06, "loss": 0.0008, "step": 52810 }, { "grad_norm": 0.03401309251785278, "learning_rate": 3.865300490743351e-06, "loss": 0.0012, "step": 52820 }, { "grad_norm": 0.036039598286151886, "learning_rate": 3.854683041743806e-06, "loss": 0.0007, "step": 52830 }, { "grad_norm": 0.02245367504656315, "learning_rate": 3.844079610473139e-06, "loss": 0.0011, "step": 52840 }, { "grad_norm": 0.03538507595658302, "learning_rate": 3.833490200152423e-06, "loss": 0.0009, "step": 52850 }, { "grad_norm": 0.02935446985065937, "learning_rate": 3.822914813998424e-06, "loss": 0.0007, "step": 52860 }, { "grad_norm": 0.05508047342300415, "learning_rate": 3.812353455223666e-06, "loss": 0.0008, "step": 52870 }, { "grad_norm": 0.025803634896874428, "learning_rate": 3.8018061270364225e-06, "loss": 0.0008, "step": 52880 }, { "grad_norm": 0.04296756908297539, "learning_rate": 3.7912728326406688e-06, "loss": 0.0009, "step": 52890 }, { "grad_norm": 0.03448756784200668, "learning_rate": 3.7807535752361732e-06, "loss": 0.0013, "step": 52900 }, { "grad_norm": 0.02757043018937111, "learning_rate": 3.7702483580183855e-06, "loss": 0.0007, "step": 52910 }, { "grad_norm": 0.023261243477463722, "learning_rate": 3.759757184178525e-06, "loss": 0.0007, "step": 52920 }, { "grad_norm": 0.025489242747426033, "learning_rate": 3.7492800569035312e-06, "loss": 0.0007, "step": 52930 }, { "grad_norm": 0.05044073611497879, "learning_rate": 3.7388169793760754e-06, "loss": 0.0022, "step": 52940 }, { "grad_norm": 0.054202817380428314, "learning_rate": 3.728367954774553e-06, "loss": 0.001, "step": 52950 }, { "grad_norm": 0.022667862474918365, "learning_rate": 3.7179329862731317e-06, "loss": 0.0008, "step": 52960 }, { "grad_norm": 0.06282364577054977, "learning_rate": 3.707512077041647e-06, "loss": 0.001, "step": 52970 }, { "grad_norm": 0.03067658841609955, "learning_rate": 3.6971052302457288e-06, "loss": 0.0006, "step": 52980 }, { "grad_norm": 0.06360284239053726, "learning_rate": 3.6867124490466698e-06, "loss": 0.0009, "step": 52990 }, { "grad_norm": 0.042741820216178894, "learning_rate": 3.6763337366015393e-06, "loss": 0.0007, "step": 53000 }, { "grad_norm": 0.024360433220863342, "learning_rate": 3.665969096063121e-06, "loss": 0.0009, "step": 53010 }, { "grad_norm": 0.01889701373875141, "learning_rate": 3.6556185305799074e-06, "loss": 0.0009, "step": 53020 }, { "grad_norm": 0.027175424620509148, "learning_rate": 3.645282043296133e-06, "loss": 0.001, "step": 53030 }, { "grad_norm": 0.03393441066145897, "learning_rate": 3.6349596373517427e-06, "loss": 0.0018, "step": 53040 }, { "grad_norm": 0.02261178568005562, "learning_rate": 3.6246513158824215e-06, "loss": 0.0009, "step": 53050 }, { "grad_norm": 0.03441939875483513, "learning_rate": 3.6143570820195593e-06, "loss": 0.0008, "step": 53060 }, { "grad_norm": 0.05135928839445114, "learning_rate": 3.6040769388902773e-06, "loss": 0.001, "step": 53070 }, { "grad_norm": 0.05083455890417099, "learning_rate": 3.593810889617405e-06, "loss": 0.0011, "step": 53080 }, { "grad_norm": 0.029180580750107765, "learning_rate": 3.5835589373194978e-06, "loss": 0.0008, "step": 53090 }, { "grad_norm": 0.04461318999528885, "learning_rate": 3.5733210851108257e-06, "loss": 0.0009, "step": 53100 }, { "grad_norm": 0.06254999339580536, "learning_rate": 3.5630973361014008e-06, "loss": 0.0007, "step": 53110 }, { "grad_norm": 0.06006920710206032, "learning_rate": 3.552887693396889e-06, "loss": 0.0012, "step": 53120 }, { "grad_norm": 0.036146923899650574, "learning_rate": 3.542692160098754e-06, "loss": 0.0011, "step": 53130 }, { "grad_norm": 0.02170035056769848, "learning_rate": 3.5325107393040846e-06, "loss": 0.0008, "step": 53140 }, { "grad_norm": 0.03473633900284767, "learning_rate": 3.522343434105757e-06, "loss": 0.0006, "step": 53150 }, { "grad_norm": 0.038307566195726395, "learning_rate": 3.512190247592323e-06, "loss": 0.0009, "step": 53160 }, { "grad_norm": 0.02308538742363453, "learning_rate": 3.502051182848054e-06, "loss": 0.0008, "step": 53170 }, { "grad_norm": 0.023706888779997826, "learning_rate": 3.4919262429529308e-06, "loss": 0.001, "step": 53180 }, { "grad_norm": 0.030351590365171432, "learning_rate": 3.4818154309826325e-06, "loss": 0.0008, "step": 53190 }, { "grad_norm": 0.06554701179265976, "learning_rate": 3.4717187500085734e-06, "loss": 0.0014, "step": 53200 }, { "grad_norm": 0.03335706889629364, "learning_rate": 3.46163620309784e-06, "loss": 0.0008, "step": 53210 }, { "grad_norm": 0.017806394025683403, "learning_rate": 3.4515677933132595e-06, "loss": 0.0007, "step": 53220 }, { "grad_norm": 0.017712166532874107, "learning_rate": 3.4415135237133466e-06, "loss": 0.0007, "step": 53230 }, { "grad_norm": 0.05896391347050667, "learning_rate": 3.4314733973523196e-06, "loss": 0.0014, "step": 53240 }, { "grad_norm": 0.05514756217598915, "learning_rate": 3.4214474172800993e-06, "loss": 0.001, "step": 53250 }, { "grad_norm": 0.020572904497385025, "learning_rate": 3.411435586542322e-06, "loss": 0.001, "step": 53260 }, { "grad_norm": 0.0551375187933445, "learning_rate": 3.4014379081802995e-06, "loss": 0.0012, "step": 53270 }, { "grad_norm": 0.0643918588757515, "learning_rate": 3.391454385231102e-06, "loss": 0.0008, "step": 53280 }, { "grad_norm": 0.0482659675180912, "learning_rate": 3.3814850207274095e-06, "loss": 0.001, "step": 53290 }, { "grad_norm": 0.06531892716884613, "learning_rate": 3.371529817697694e-06, "loss": 0.0009, "step": 53300 }, { "grad_norm": 0.03462494909763336, "learning_rate": 3.3615887791660585e-06, "loss": 0.0012, "step": 53310 }, { "grad_norm": 0.03665409982204437, "learning_rate": 3.3516619081523426e-06, "loss": 0.002, "step": 53320 }, { "grad_norm": 0.028362106531858444, "learning_rate": 3.3417492076720567e-06, "loss": 0.001, "step": 53330 }, { "grad_norm": 0.02561008557677269, "learning_rate": 3.3318506807364147e-06, "loss": 0.0009, "step": 53340 }, { "grad_norm": 0.030925588682293892, "learning_rate": 3.3219663303523553e-06, "loss": 0.0007, "step": 53350 }, { "grad_norm": 0.028017384931445122, "learning_rate": 3.3120961595224374e-06, "loss": 0.0011, "step": 53360 }, { "grad_norm": 0.027101077139377594, "learning_rate": 3.302240171245002e-06, "loss": 0.0011, "step": 53370 }, { "grad_norm": 0.028744840994477272, "learning_rate": 3.29239836851401e-06, "loss": 0.0008, "step": 53380 }, { "grad_norm": 0.04604867100715637, "learning_rate": 3.2825707543191588e-06, "loss": 0.0009, "step": 53390 }, { "grad_norm": 0.02944442816078663, "learning_rate": 3.272757331645804e-06, "loss": 0.0009, "step": 53400 }, { "grad_norm": 0.025604285299777985, "learning_rate": 3.2629581034750166e-06, "loss": 0.0007, "step": 53410 }, { "grad_norm": 0.032554417848587036, "learning_rate": 3.2531730727835218e-06, "loss": 0.0009, "step": 53420 }, { "grad_norm": 0.0313023217022419, "learning_rate": 3.2434022425437914e-06, "loss": 0.0007, "step": 53430 }, { "grad_norm": 0.03379789739847183, "learning_rate": 3.233645615723907e-06, "loss": 0.0008, "step": 53440 }, { "grad_norm": 0.020862968638539314, "learning_rate": 3.2239031952876918e-06, "loss": 0.0009, "step": 53450 }, { "grad_norm": 0.042856551706790924, "learning_rate": 3.21417498419464e-06, "loss": 0.0007, "step": 53460 }, { "grad_norm": 0.026901202276349068, "learning_rate": 3.204460985399921e-06, "loss": 0.0007, "step": 53470 }, { "grad_norm": 0.02638201043009758, "learning_rate": 3.1947612018543903e-06, "loss": 0.0007, "step": 53480 }, { "grad_norm": 0.016344735398888588, "learning_rate": 3.1850756365045753e-06, "loss": 0.0007, "step": 53490 }, { "grad_norm": 0.035520296543836594, "learning_rate": 3.175404292292722e-06, "loss": 0.0009, "step": 53500 }, { "grad_norm": 0.03408680856227875, "learning_rate": 3.1657471721566965e-06, "loss": 0.001, "step": 53510 }, { "grad_norm": 0.06770768016576767, "learning_rate": 3.1561042790300977e-06, "loss": 0.0013, "step": 53520 }, { "grad_norm": 0.035356488078832626, "learning_rate": 3.1464756158421816e-06, "loss": 0.0007, "step": 53530 }, { "grad_norm": 0.03654219210147858, "learning_rate": 3.136861185517875e-06, "loss": 0.0009, "step": 53540 }, { "grad_norm": 0.0714205875992775, "learning_rate": 3.127260990977798e-06, "loss": 0.0009, "step": 53550 }, { "grad_norm": 0.032311953604221344, "learning_rate": 3.1176750351382235e-06, "loss": 0.0008, "step": 53560 }, { "grad_norm": 0.0386841744184494, "learning_rate": 3.1081033209111153e-06, "loss": 0.0008, "step": 53570 }, { "grad_norm": 0.03897203505039215, "learning_rate": 3.0985458512041155e-06, "loss": 0.0007, "step": 53580 }, { "grad_norm": 0.020415576174855232, "learning_rate": 3.089002628920512e-06, "loss": 0.0009, "step": 53590 }, { "grad_norm": 0.028875937685370445, "learning_rate": 3.079473656959303e-06, "loss": 0.0008, "step": 53600 }, { "grad_norm": 0.025963282212615013, "learning_rate": 3.0699589382151393e-06, "loss": 0.0012, "step": 53610 }, { "grad_norm": 0.018286455422639847, "learning_rate": 3.060458475578326e-06, "loss": 0.0009, "step": 53620 }, { "grad_norm": 0.026592673733830452, "learning_rate": 3.0509722719348656e-06, "loss": 0.0012, "step": 53630 }, { "grad_norm": 0.05419062823057175, "learning_rate": 3.041500330166408e-06, "loss": 0.0008, "step": 53640 }, { "grad_norm": 0.02319803088903427, "learning_rate": 3.032042653150291e-06, "loss": 0.0006, "step": 53650 }, { "grad_norm": 0.023760978132486343, "learning_rate": 3.0225992437594887e-06, "loss": 0.0012, "step": 53660 }, { "grad_norm": 0.03697570413351059, "learning_rate": 3.0131701048626783e-06, "loss": 0.0009, "step": 53670 }, { "grad_norm": 0.017846381291747093, "learning_rate": 3.003755239324163e-06, "loss": 0.0015, "step": 53680 }, { "grad_norm": 0.05195264145731926, "learning_rate": 2.9943546500039553e-06, "loss": 0.0009, "step": 53690 }, { "grad_norm": 0.04924846813082695, "learning_rate": 2.9849683397576877e-06, "loss": 0.0009, "step": 53700 }, { "grad_norm": 0.06480170786380768, "learning_rate": 2.975596311436679e-06, "loss": 0.0012, "step": 53710 }, { "grad_norm": 0.019676614552736282, "learning_rate": 2.966238567887902e-06, "loss": 0.0008, "step": 53720 }, { "grad_norm": 0.04882882535457611, "learning_rate": 2.9568951119539943e-06, "loss": 0.001, "step": 53730 }, { "grad_norm": 0.07500794529914856, "learning_rate": 2.947565946473241e-06, "loss": 0.0009, "step": 53740 }, { "grad_norm": 0.018196454271674156, "learning_rate": 2.9382510742796188e-06, "loss": 0.0009, "step": 53750 }, { "grad_norm": 0.026229634881019592, "learning_rate": 2.9289504982027204e-06, "loss": 0.0009, "step": 53760 }, { "grad_norm": 0.06495290249586105, "learning_rate": 2.9196642210678248e-06, "loss": 0.0012, "step": 53770 }, { "grad_norm": 0.025522902607917786, "learning_rate": 2.910392245695853e-06, "loss": 0.0007, "step": 53780 }, { "grad_norm": 0.02003718540072441, "learning_rate": 2.9011345749033804e-06, "loss": 0.0007, "step": 53790 }, { "grad_norm": 0.02968663163483143, "learning_rate": 2.8918912115026677e-06, "loss": 0.0009, "step": 53800 }, { "grad_norm": 0.039632100611925125, "learning_rate": 2.8826621583015743e-06, "loss": 0.0009, "step": 53810 }, { "grad_norm": 0.01804409734904766, "learning_rate": 2.8734474181036643e-06, "loss": 0.0009, "step": 53820 }, { "grad_norm": 0.025607768446207047, "learning_rate": 2.864246993708114e-06, "loss": 0.0009, "step": 53830 }, { "grad_norm": 0.028095562011003494, "learning_rate": 2.8550608879097884e-06, "loss": 0.0008, "step": 53840 }, { "grad_norm": 0.024181010201573372, "learning_rate": 2.845889103499172e-06, "loss": 0.001, "step": 53850 }, { "grad_norm": 0.02547738514840603, "learning_rate": 2.8367316432624138e-06, "loss": 0.0008, "step": 53860 }, { "grad_norm": 0.039049748331308365, "learning_rate": 2.8275885099813105e-06, "loss": 0.0016, "step": 53870 }, { "grad_norm": 0.027009520679712296, "learning_rate": 2.8184597064332963e-06, "loss": 0.0008, "step": 53880 }, { "grad_norm": 0.02164357714354992, "learning_rate": 2.809345235391464e-06, "loss": 0.0009, "step": 53890 }, { "grad_norm": 0.041818249970674515, "learning_rate": 2.80024509962456e-06, "loss": 0.0015, "step": 53900 }, { "grad_norm": 0.04092884808778763, "learning_rate": 2.7911593018969563e-06, "loss": 0.0009, "step": 53910 }, { "grad_norm": 0.02184605598449707, "learning_rate": 2.7820878449686838e-06, "loss": 0.001, "step": 53920 }, { "grad_norm": 0.03002355620265007, "learning_rate": 2.7730307315953995e-06, "loss": 0.0007, "step": 53930 }, { "grad_norm": 0.02272168919444084, "learning_rate": 2.763987964528425e-06, "loss": 0.001, "step": 53940 }, { "grad_norm": 0.02945324219763279, "learning_rate": 2.754959546514718e-06, "loss": 0.0008, "step": 53950 }, { "grad_norm": 0.020530665293335915, "learning_rate": 2.7459454802968576e-06, "loss": 0.0008, "step": 53960 }, { "grad_norm": 0.05433221906423569, "learning_rate": 2.7369457686131028e-06, "loss": 0.0018, "step": 53970 }, { "grad_norm": 0.030971528962254524, "learning_rate": 2.7279604141973004e-06, "loss": 0.001, "step": 53980 }, { "grad_norm": 0.026718944311141968, "learning_rate": 2.7189894197789946e-06, "loss": 0.0008, "step": 53990 }, { "grad_norm": 0.03669071942567825, "learning_rate": 2.7100327880833055e-06, "loss": 0.0008, "step": 54000 }, { "grad_norm": 0.02624659612774849, "learning_rate": 2.70109052183104e-06, "loss": 0.001, "step": 54010 }, { "grad_norm": 0.037404127418994904, "learning_rate": 2.69216262373862e-06, "loss": 0.0013, "step": 54020 }, { "grad_norm": 0.0289319958537817, "learning_rate": 2.6832490965181036e-06, "loss": 0.001, "step": 54030 }, { "grad_norm": 0.027434643357992172, "learning_rate": 2.6743499428771857e-06, "loss": 0.0009, "step": 54040 }, { "grad_norm": 0.023206625133752823, "learning_rate": 2.6654651655191875e-06, "loss": 0.0008, "step": 54050 }, { "grad_norm": 0.026182498782873154, "learning_rate": 2.6565947671430836e-06, "loss": 0.0008, "step": 54060 }, { "grad_norm": 0.02430746890604496, "learning_rate": 2.647738750443457e-06, "loss": 0.0009, "step": 54070 }, { "grad_norm": 0.021619368344545364, "learning_rate": 2.6388971181105393e-06, "loss": 0.0008, "step": 54080 }, { "grad_norm": 0.028917726129293442, "learning_rate": 2.630069872830171e-06, "loss": 0.0019, "step": 54090 }, { "grad_norm": 0.03416483476758003, "learning_rate": 2.6212570172838514e-06, "loss": 0.0008, "step": 54100 }, { "grad_norm": 0.019264286383986473, "learning_rate": 2.6124585541486778e-06, "loss": 0.0011, "step": 54110 }, { "grad_norm": 0.019043730571866035, "learning_rate": 2.6036744860974127e-06, "loss": 0.0008, "step": 54120 }, { "grad_norm": 0.03246016055345535, "learning_rate": 2.594904815798399e-06, "loss": 0.0011, "step": 54130 }, { "grad_norm": 0.024433616548776627, "learning_rate": 2.586149545915656e-06, "loss": 0.0008, "step": 54140 }, { "grad_norm": 0.03952345997095108, "learning_rate": 2.577408679108778e-06, "loss": 0.001, "step": 54150 }, { "grad_norm": 0.03670293465256691, "learning_rate": 2.5686822180330306e-06, "loss": 0.0008, "step": 54160 }, { "grad_norm": 0.03223775699734688, "learning_rate": 2.5599701653392703e-06, "loss": 0.0009, "step": 54170 }, { "grad_norm": 0.04077651724219322, "learning_rate": 2.551272523673992e-06, "loss": 0.0008, "step": 54180 }, { "grad_norm": 0.03723714500665665, "learning_rate": 2.542589295679315e-06, "loss": 0.0006, "step": 54190 }, { "grad_norm": 0.03136210888624191, "learning_rate": 2.5339204839929575e-06, "loss": 0.0009, "step": 54200 }, { "grad_norm": 0.02663559466600418, "learning_rate": 2.525266091248296e-06, "loss": 0.0007, "step": 54210 }, { "grad_norm": 0.0321524478495121, "learning_rate": 2.5166261200743e-06, "loss": 0.0009, "step": 54220 }, { "grad_norm": 0.04953940212726593, "learning_rate": 2.5080005730955646e-06, "loss": 0.0007, "step": 54230 }, { "grad_norm": 0.02948703244328499, "learning_rate": 2.499389452932299e-06, "loss": 0.0012, "step": 54240 }, { "grad_norm": 0.015895556658506393, "learning_rate": 2.4907927622003336e-06, "loss": 0.0012, "step": 54250 }, { "grad_norm": 0.024678237736225128, "learning_rate": 2.4822105035111177e-06, "loss": 0.0007, "step": 54260 }, { "grad_norm": 0.016681553795933723, "learning_rate": 2.4736426794717273e-06, "loss": 0.0008, "step": 54270 }, { "grad_norm": 0.023380227386951447, "learning_rate": 2.4650892926848135e-06, "loss": 0.001, "step": 54280 }, { "grad_norm": 0.018701622262597084, "learning_rate": 2.456550345748704e-06, "loss": 0.0008, "step": 54290 }, { "grad_norm": 0.023961136117577553, "learning_rate": 2.4480258412572733e-06, "loss": 0.001, "step": 54300 }, { "grad_norm": 0.020209599286317825, "learning_rate": 2.4395157818000612e-06, "loss": 0.0007, "step": 54310 }, { "grad_norm": 0.029200823977589607, "learning_rate": 2.431020169962189e-06, "loss": 0.0015, "step": 54320 }, { "grad_norm": 0.01566787250339985, "learning_rate": 2.422539008324409e-06, "loss": 0.001, "step": 54330 }, { "grad_norm": 0.02616254799067974, "learning_rate": 2.414072299463066e-06, "loss": 0.0009, "step": 54340 }, { "grad_norm": 0.0598980113863945, "learning_rate": 2.4056200459501186e-06, "loss": 0.0012, "step": 54350 }, { "grad_norm": 0.0211861040443182, "learning_rate": 2.397182250353147e-06, "loss": 0.0008, "step": 54360 }, { "grad_norm": 0.02947833202779293, "learning_rate": 2.388758915235334e-06, "loss": 0.0009, "step": 54370 }, { "grad_norm": 0.02663767710328102, "learning_rate": 2.380350043155455e-06, "loss": 0.0007, "step": 54380 }, { "grad_norm": 0.017870524898171425, "learning_rate": 2.371955636667911e-06, "loss": 0.001, "step": 54390 }, { "grad_norm": 0.03243039920926094, "learning_rate": 2.3635756983227008e-06, "loss": 0.0009, "step": 54400 }, { "grad_norm": 0.0323343351483345, "learning_rate": 2.3552102306654278e-06, "loss": 0.0008, "step": 54410 }, { "grad_norm": 0.04481949284672737, "learning_rate": 2.346859236237292e-06, "loss": 0.001, "step": 54420 }, { "grad_norm": 0.021843641996383667, "learning_rate": 2.3385227175751145e-06, "loss": 0.0009, "step": 54430 }, { "grad_norm": 0.028417034074664116, "learning_rate": 2.330200677211314e-06, "loss": 0.0008, "step": 54440 }, { "grad_norm": 0.028681021183729172, "learning_rate": 2.3218931176738847e-06, "loss": 0.0008, "step": 54450 }, { "grad_norm": 0.046621497720479965, "learning_rate": 2.313600041486469e-06, "loss": 0.0011, "step": 54460 }, { "grad_norm": 0.01900894194841385, "learning_rate": 2.3053214511682743e-06, "loss": 0.001, "step": 54470 }, { "grad_norm": 0.027289990335702896, "learning_rate": 2.2970573492341163e-06, "loss": 0.0009, "step": 54480 }, { "grad_norm": 0.027571694925427437, "learning_rate": 2.288807738194415e-06, "loss": 0.0008, "step": 54490 }, { "grad_norm": 0.0226018987596035, "learning_rate": 2.2805726205551768e-06, "loss": 0.0007, "step": 54500 }, { "grad_norm": 0.01957131363451481, "learning_rate": 2.272351998818023e-06, "loss": 0.001, "step": 54510 }, { "grad_norm": 0.028492828831076622, "learning_rate": 2.2641458754801505e-06, "loss": 0.0009, "step": 54520 }, { "grad_norm": 0.0255172960460186, "learning_rate": 2.2559542530343756e-06, "loss": 0.0007, "step": 54530 }, { "grad_norm": 0.04509523883461952, "learning_rate": 2.247777133969087e-06, "loss": 0.0008, "step": 54540 }, { "grad_norm": 0.07337993383407593, "learning_rate": 2.2396145207682795e-06, "loss": 0.0012, "step": 54550 }, { "grad_norm": 0.037826165556907654, "learning_rate": 2.231466415911543e-06, "loss": 0.0008, "step": 54560 }, { "grad_norm": 0.01738448068499565, "learning_rate": 2.2233328218740524e-06, "loss": 0.0007, "step": 54570 }, { "grad_norm": 0.03345083072781563, "learning_rate": 2.215213741126576e-06, "loss": 0.0007, "step": 54580 }, { "grad_norm": 0.03868395835161209, "learning_rate": 2.2071091761354912e-06, "loss": 0.0008, "step": 54590 }, { "grad_norm": 0.03512074425816536, "learning_rate": 2.1990191293627337e-06, "loss": 0.001, "step": 54600 }, { "grad_norm": 0.0378439761698246, "learning_rate": 2.1909436032658548e-06, "loss": 0.0006, "step": 54610 }, { "grad_norm": 0.02511933073401451, "learning_rate": 2.1828826002979806e-06, "loss": 0.0015, "step": 54620 }, { "grad_norm": 0.03505455330014229, "learning_rate": 2.1748361229078362e-06, "loss": 0.001, "step": 54630 }, { "grad_norm": 0.02013227343559265, "learning_rate": 2.1668041735397327e-06, "loss": 0.0011, "step": 54640 }, { "grad_norm": 0.024957874789834023, "learning_rate": 2.1587867546335514e-06, "loss": 0.0007, "step": 54650 }, { "grad_norm": 0.026977956295013428, "learning_rate": 2.1507838686247894e-06, "loss": 0.0008, "step": 54660 }, { "grad_norm": 0.022344939410686493, "learning_rate": 2.1427955179444957e-06, "loss": 0.0008, "step": 54670 }, { "grad_norm": 0.020388638600707054, "learning_rate": 2.13482170501933e-06, "loss": 0.0009, "step": 54680 }, { "grad_norm": 0.024005817249417305, "learning_rate": 2.1268624322715202e-06, "loss": 0.0006, "step": 54690 }, { "grad_norm": 0.024533795192837715, "learning_rate": 2.118917702118889e-06, "loss": 0.0008, "step": 54700 }, { "grad_norm": 0.028580540791153908, "learning_rate": 2.1109875169748327e-06, "loss": 0.0012, "step": 54710 }, { "grad_norm": 0.02241702564060688, "learning_rate": 2.103071879248336e-06, "loss": 0.0007, "step": 54720 }, { "grad_norm": 0.044372476637363434, "learning_rate": 2.0951707913439478e-06, "loss": 0.0017, "step": 54730 }, { "grad_norm": 0.04184812679886818, "learning_rate": 2.0872842556618255e-06, "loss": 0.0008, "step": 54740 }, { "grad_norm": 0.028559662401676178, "learning_rate": 2.079412274597686e-06, "loss": 0.0018, "step": 54750 }, { "grad_norm": 0.02742408588528633, "learning_rate": 2.0715548505428284e-06, "loss": 0.0009, "step": 54760 }, { "grad_norm": 0.044857464730739594, "learning_rate": 2.0637119858841258e-06, "loss": 0.0013, "step": 54770 }, { "grad_norm": 0.026097316294908524, "learning_rate": 2.055883683004034e-06, "loss": 0.0009, "step": 54780 }, { "grad_norm": 0.01654931902885437, "learning_rate": 2.0480699442806006e-06, "loss": 0.0009, "step": 54790 }, { "grad_norm": 0.021511735394597054, "learning_rate": 2.0402707720874105e-06, "loss": 0.0008, "step": 54800 }, { "grad_norm": 0.014422528445720673, "learning_rate": 2.032486168793668e-06, "loss": 0.0009, "step": 54810 }, { "grad_norm": 0.027779007330536842, "learning_rate": 2.024716136764104e-06, "loss": 0.0007, "step": 54820 }, { "grad_norm": 0.03246772289276123, "learning_rate": 2.0169606783590757e-06, "loss": 0.0009, "step": 54830 }, { "grad_norm": 0.03592981770634651, "learning_rate": 2.0092197959344638e-06, "loss": 0.0012, "step": 54840 }, { "grad_norm": 0.03755264729261398, "learning_rate": 2.0014934918417606e-06, "loss": 0.0008, "step": 54850 }, { "grad_norm": 0.030260594561696053, "learning_rate": 1.99378176842801e-06, "loss": 0.0008, "step": 54860 }, { "grad_norm": 0.0498470775783062, "learning_rate": 1.9860846280358224e-06, "loss": 0.0009, "step": 54870 }, { "grad_norm": 0.022521834820508957, "learning_rate": 1.978402073003394e-06, "loss": 0.0011, "step": 54880 }, { "grad_norm": 0.021473685279488564, "learning_rate": 1.9707341056644736e-06, "loss": 0.0011, "step": 54890 }, { "grad_norm": 0.025466302409768105, "learning_rate": 1.963080728348399e-06, "loss": 0.0013, "step": 54900 }, { "grad_norm": 0.02065775915980339, "learning_rate": 1.955441943380054e-06, "loss": 0.0007, "step": 54910 }, { "grad_norm": 0.04071376100182533, "learning_rate": 1.947817753079906e-06, "loss": 0.0017, "step": 54920 }, { "grad_norm": 0.020984752103686333, "learning_rate": 1.9402081597639785e-06, "loss": 0.0009, "step": 54930 }, { "grad_norm": 0.016613604500889778, "learning_rate": 1.9326131657438683e-06, "loss": 0.0007, "step": 54940 }, { "grad_norm": 0.05417529493570328, "learning_rate": 1.925032773326724e-06, "loss": 0.0012, "step": 54950 }, { "grad_norm": 0.029544444754719734, "learning_rate": 1.9174669848152916e-06, "loss": 0.0007, "step": 54960 }, { "grad_norm": 0.05639839917421341, "learning_rate": 1.9099158025078336e-06, "loss": 0.0008, "step": 54970 }, { "grad_norm": 0.010859250091016293, "learning_rate": 1.90237922869822e-06, "loss": 0.001, "step": 54980 }, { "grad_norm": 0.01909448206424713, "learning_rate": 1.8948572656758367e-06, "loss": 0.0007, "step": 54990 }, { "grad_norm": 0.019296785816550255, "learning_rate": 1.8873499157256834e-06, "loss": 0.001, "step": 55000 }, { "grad_norm": 0.030191441997885704, "learning_rate": 1.879857181128286e-06, "loss": 0.0013, "step": 55010 }, { "grad_norm": 0.05075785517692566, "learning_rate": 1.8723790641597349e-06, "loss": 0.0009, "step": 55020 }, { "grad_norm": 0.020119303837418556, "learning_rate": 1.8649155670916906e-06, "loss": 0.001, "step": 55030 }, { "grad_norm": 0.04174528643488884, "learning_rate": 1.8574666921913565e-06, "loss": 0.0012, "step": 55040 }, { "grad_norm": 0.031231369823217392, "learning_rate": 1.8500324417215166e-06, "loss": 0.0006, "step": 55050 }, { "grad_norm": 0.022686311975121498, "learning_rate": 1.8426128179404977e-06, "loss": 0.0009, "step": 55060 }, { "grad_norm": 0.032167695462703705, "learning_rate": 1.8352078231021807e-06, "loss": 0.0008, "step": 55070 }, { "grad_norm": 0.021306294947862625, "learning_rate": 1.8278174594560049e-06, "loss": 0.0011, "step": 55080 }, { "grad_norm": 0.022566279396414757, "learning_rate": 1.82044172924698e-06, "loss": 0.001, "step": 55090 }, { "grad_norm": 0.019877037033438683, "learning_rate": 1.813080634715636e-06, "loss": 0.0006, "step": 55100 }, { "grad_norm": 0.025561723858118057, "learning_rate": 1.8057341780981119e-06, "loss": 0.0008, "step": 55110 }, { "grad_norm": 0.023791780695319176, "learning_rate": 1.7984023616260338e-06, "loss": 0.0013, "step": 55120 }, { "grad_norm": 0.017045751214027405, "learning_rate": 1.7910851875266421e-06, "loss": 0.0007, "step": 55130 }, { "grad_norm": 0.04784821346402168, "learning_rate": 1.7837826580226757e-06, "loss": 0.0009, "step": 55140 }, { "grad_norm": 0.026799343526363373, "learning_rate": 1.7764947753324656e-06, "loss": 0.0012, "step": 55150 }, { "grad_norm": 0.03558816760778427, "learning_rate": 1.7692215416698799e-06, "loss": 0.0009, "step": 55160 }, { "grad_norm": 0.04940835386514664, "learning_rate": 1.7619629592443233e-06, "loss": 0.0008, "step": 55170 }, { "grad_norm": 0.053995028138160706, "learning_rate": 1.7547190302607709e-06, "loss": 0.0008, "step": 55180 }, { "grad_norm": 0.023032736033201218, "learning_rate": 1.747489756919729e-06, "loss": 0.0014, "step": 55190 }, { "grad_norm": 0.1310771256685257, "learning_rate": 1.7402751414172802e-06, "loss": 0.0009, "step": 55200 }, { "grad_norm": 0.033029548823833466, "learning_rate": 1.7330751859450044e-06, "loss": 0.0006, "step": 55210 }, { "grad_norm": 0.03335319459438324, "learning_rate": 1.7258898926900801e-06, "loss": 0.0011, "step": 55220 }, { "grad_norm": 0.033262137323617935, "learning_rate": 1.7187192638352002e-06, "loss": 0.001, "step": 55230 }, { "grad_norm": 0.0648866668343544, "learning_rate": 1.7115633015586163e-06, "loss": 0.0009, "step": 55240 }, { "grad_norm": 0.014714865013957024, "learning_rate": 1.7044220080341178e-06, "loss": 0.001, "step": 55250 }, { "grad_norm": 0.05222463235259056, "learning_rate": 1.6972953854310414e-06, "loss": 0.001, "step": 55260 }, { "grad_norm": 0.014791369438171387, "learning_rate": 1.690183435914261e-06, "loss": 0.0007, "step": 55270 }, { "grad_norm": 0.025990763679146767, "learning_rate": 1.6830861616442206e-06, "loss": 0.0013, "step": 55280 }, { "grad_norm": 0.02957993373274803, "learning_rate": 1.6760035647768568e-06, "loss": 0.0012, "step": 55290 }, { "grad_norm": 0.03688570484519005, "learning_rate": 1.6689356474636985e-06, "loss": 0.0009, "step": 55300 }, { "grad_norm": 0.03539380803704262, "learning_rate": 1.6618824118517784e-06, "loss": 0.0012, "step": 55310 }, { "grad_norm": 0.022967496886849403, "learning_rate": 1.654843860083688e-06, "loss": 0.0007, "step": 55320 }, { "grad_norm": 0.06073208153247833, "learning_rate": 1.647819994297556e-06, "loss": 0.0009, "step": 55330 }, { "grad_norm": 0.033607982099056244, "learning_rate": 1.6408108166270363e-06, "loss": 0.0008, "step": 55340 }, { "grad_norm": 0.04303513839840889, "learning_rate": 1.6338163292013486e-06, "loss": 0.0016, "step": 55350 }, { "grad_norm": 0.027787547558546066, "learning_rate": 1.6268365341452208e-06, "loss": 0.0007, "step": 55360 }, { "grad_norm": 0.07347694784402847, "learning_rate": 1.6198714335789345e-06, "loss": 0.0011, "step": 55370 }, { "grad_norm": 0.02181355096399784, "learning_rate": 1.612921029618303e-06, "loss": 0.0006, "step": 55380 }, { "grad_norm": 0.016867276281118393, "learning_rate": 1.6059853243746815e-06, "loss": 0.0014, "step": 55390 }, { "grad_norm": 0.02231440134346485, "learning_rate": 1.5990643199549404e-06, "loss": 0.0011, "step": 55400 }, { "grad_norm": 0.029117902740836143, "learning_rate": 1.5921580184615147e-06, "loss": 0.001, "step": 55410 }, { "grad_norm": 0.020186610519886017, "learning_rate": 1.5852664219923374e-06, "loss": 0.001, "step": 55420 }, { "grad_norm": 0.01958894170820713, "learning_rate": 1.5783895326409172e-06, "loss": 0.0008, "step": 55430 }, { "grad_norm": 0.01776556298136711, "learning_rate": 1.5715273524962438e-06, "loss": 0.0006, "step": 55440 }, { "grad_norm": 0.04229550436139107, "learning_rate": 1.564679883642889e-06, "loss": 0.0007, "step": 55450 }, { "grad_norm": 0.026386693120002747, "learning_rate": 1.5578471281609276e-06, "loss": 0.0008, "step": 55460 }, { "grad_norm": 0.025987980887293816, "learning_rate": 1.551029088125966e-06, "loss": 0.001, "step": 55470 }, { "grad_norm": 0.018119074404239655, "learning_rate": 1.544225765609142e-06, "loss": 0.0007, "step": 55480 }, { "grad_norm": 0.02307765930891037, "learning_rate": 1.53743716267713e-06, "loss": 0.0009, "step": 55490 }, { "grad_norm": 0.03392595052719116, "learning_rate": 1.5306632813921361e-06, "loss": 0.0015, "step": 55500 }, { "grad_norm": 0.04558272659778595, "learning_rate": 1.52390412381187e-06, "loss": 0.0007, "step": 55510 }, { "grad_norm": 0.015582690015435219, "learning_rate": 1.5171596919895948e-06, "loss": 0.0008, "step": 55520 }, { "grad_norm": 0.029409054666757584, "learning_rate": 1.5104299879740936e-06, "loss": 0.0008, "step": 55530 }, { "grad_norm": 0.020921291783452034, "learning_rate": 1.5037150138096701e-06, "loss": 0.0008, "step": 55540 }, { "grad_norm": 0.05082789808511734, "learning_rate": 1.4970147715361538e-06, "loss": 0.0007, "step": 55550 }, { "grad_norm": 0.027085186913609505, "learning_rate": 1.4903292631889054e-06, "loss": 0.0007, "step": 55560 }, { "grad_norm": 0.02681700326502323, "learning_rate": 1.483658490798795e-06, "loss": 0.0013, "step": 55570 }, { "grad_norm": 0.047652531415224075, "learning_rate": 1.4770024563922457e-06, "loss": 0.0011, "step": 55580 }, { "grad_norm": 0.023168673738837242, "learning_rate": 1.4703611619911628e-06, "loss": 0.0012, "step": 55590 }, { "grad_norm": 0.05826636031270027, "learning_rate": 1.4637346096130155e-06, "loss": 0.001, "step": 55600 }, { "grad_norm": 0.023859718814492226, "learning_rate": 1.4571228012707662e-06, "loss": 0.0017, "step": 55610 }, { "grad_norm": 0.03702421486377716, "learning_rate": 1.4505257389729132e-06, "loss": 0.0008, "step": 55620 }, { "grad_norm": 0.027508214116096497, "learning_rate": 1.4439434247234596e-06, "loss": 0.001, "step": 55630 }, { "grad_norm": 0.03187692537903786, "learning_rate": 1.4373758605219445e-06, "loss": 0.0008, "step": 55640 }, { "grad_norm": 0.021638914942741394, "learning_rate": 1.4308230483634333e-06, "loss": 0.0008, "step": 55650 }, { "grad_norm": 0.020386772230267525, "learning_rate": 1.4242849902384724e-06, "loss": 0.0007, "step": 55660 }, { "grad_norm": 0.04210250452160835, "learning_rate": 1.4177616881331734e-06, "loss": 0.0014, "step": 55670 }, { "grad_norm": 0.023593377321958542, "learning_rate": 1.4112531440291233e-06, "loss": 0.0008, "step": 55680 }, { "grad_norm": 0.01686447486281395, "learning_rate": 1.4047593599034624e-06, "loss": 0.0008, "step": 55690 }, { "grad_norm": 0.018028199672698975, "learning_rate": 1.3982803377288246e-06, "loss": 0.0015, "step": 55700 }, { "grad_norm": 0.05109541118144989, "learning_rate": 1.3918160794733681e-06, "loss": 0.0008, "step": 55710 }, { "grad_norm": 0.031270191073417664, "learning_rate": 1.3853665871007615e-06, "loss": 0.0007, "step": 55720 }, { "grad_norm": 0.028823846951127052, "learning_rate": 1.378931862570193e-06, "loss": 0.0018, "step": 55730 }, { "grad_norm": 0.026730025187134743, "learning_rate": 1.372511907836349e-06, "loss": 0.0011, "step": 55740 }, { "grad_norm": 0.04143184423446655, "learning_rate": 1.3661067248494586e-06, "loss": 0.001, "step": 55750 }, { "grad_norm": 0.018535232171416283, "learning_rate": 1.3597163155552429e-06, "loss": 0.0011, "step": 55760 }, { "grad_norm": 0.016117604449391365, "learning_rate": 1.3533406818949434e-06, "loss": 0.0011, "step": 55770 }, { "grad_norm": 0.05247751623392105, "learning_rate": 1.3469798258053002e-06, "loss": 0.0007, "step": 55780 }, { "grad_norm": 0.05210435763001442, "learning_rate": 1.3406337492185672e-06, "loss": 0.0009, "step": 55790 }, { "grad_norm": 0.029369041323661804, "learning_rate": 1.3343024540625414e-06, "loss": 0.0008, "step": 55800 }, { "grad_norm": 0.0242339875549078, "learning_rate": 1.3279859422604735e-06, "loss": 0.0007, "step": 55810 }, { "grad_norm": 0.03161405771970749, "learning_rate": 1.3216842157311781e-06, "loss": 0.0008, "step": 55820 }, { "grad_norm": 0.02469763159751892, "learning_rate": 1.3153972763889355e-06, "loss": 0.001, "step": 55830 }, { "grad_norm": 0.030109846964478493, "learning_rate": 1.3091251261435566e-06, "loss": 0.0007, "step": 55840 }, { "grad_norm": 0.021244866773486137, "learning_rate": 1.3028677669003564e-06, "loss": 0.0007, "step": 55850 }, { "grad_norm": 0.027879424393177032, "learning_rate": 1.2966252005601587e-06, "loss": 0.0008, "step": 55860 }, { "grad_norm": 0.014501994475722313, "learning_rate": 1.2903974290192855e-06, "loss": 0.0009, "step": 55870 }, { "grad_norm": 0.04488762095570564, "learning_rate": 1.284184454169568e-06, "loss": 0.0008, "step": 55880 }, { "grad_norm": 0.056292176246643066, "learning_rate": 1.2779862778983464e-06, "loss": 0.0008, "step": 55890 }, { "grad_norm": 0.01699529029428959, "learning_rate": 1.2718029020884647e-06, "loss": 0.0006, "step": 55900 }, { "grad_norm": 0.020853612571954727, "learning_rate": 1.2656343286182703e-06, "loss": 0.001, "step": 55910 }, { "grad_norm": 0.017151974141597748, "learning_rate": 1.2594805593616088e-06, "loss": 0.0011, "step": 55920 }, { "grad_norm": 0.023327436298131943, "learning_rate": 1.2533415961878404e-06, "loss": 0.0009, "step": 55930 }, { "grad_norm": 0.015302943997085094, "learning_rate": 1.2472174409618009e-06, "loss": 0.0007, "step": 55940 }, { "grad_norm": 0.056057825684547424, "learning_rate": 1.2411080955438747e-06, "loss": 0.0011, "step": 55950 }, { "grad_norm": 0.016959773376584053, "learning_rate": 1.235013561789894e-06, "loss": 0.001, "step": 55960 }, { "grad_norm": 0.023963332176208496, "learning_rate": 1.2289338415512385e-06, "loss": 0.0006, "step": 55970 }, { "grad_norm": 0.015543878078460693, "learning_rate": 1.222868936674748e-06, "loss": 0.0013, "step": 55980 }, { "grad_norm": 0.028161125257611275, "learning_rate": 1.2168188490027876e-06, "loss": 0.0009, "step": 55990 }, { "grad_norm": 0.024815523996949196, "learning_rate": 1.2107835803732204e-06, "loss": 0.0009, "step": 56000 }, { "grad_norm": 0.02850870043039322, "learning_rate": 1.2047631326193964e-06, "loss": 0.0008, "step": 56010 }, { "grad_norm": 0.03900930657982826, "learning_rate": 1.1987575075701696e-06, "loss": 0.0009, "step": 56020 }, { "grad_norm": 0.0248082485049963, "learning_rate": 1.1927667070498916e-06, "loss": 0.0007, "step": 56030 }, { "grad_norm": 0.02010316774249077, "learning_rate": 1.1867907328784067e-06, "loss": 0.0015, "step": 56040 }, { "grad_norm": 0.029290400445461273, "learning_rate": 1.1808295868710518e-06, "loss": 0.0013, "step": 56050 }, { "grad_norm": 0.024142121896147728, "learning_rate": 1.174883270838678e-06, "loss": 0.001, "step": 56060 }, { "grad_norm": 0.02173689939081669, "learning_rate": 1.1689517865876187e-06, "loss": 0.001, "step": 56070 }, { "grad_norm": 0.012393835000693798, "learning_rate": 1.1630351359196933e-06, "loss": 0.0005, "step": 56080 }, { "grad_norm": 0.06714087724685669, "learning_rate": 1.1571333206322255e-06, "loss": 0.0018, "step": 56090 }, { "grad_norm": 0.020438076928257942, "learning_rate": 1.1512463425180365e-06, "loss": 0.0011, "step": 56100 }, { "grad_norm": 0.020999906584620476, "learning_rate": 1.145374203365429e-06, "loss": 0.001, "step": 56110 }, { "grad_norm": 0.029016243293881416, "learning_rate": 1.1395169049582155e-06, "loss": 0.0012, "step": 56120 }, { "grad_norm": 0.02062392234802246, "learning_rate": 1.1336744490756722e-06, "loss": 0.0011, "step": 56130 }, { "grad_norm": 0.03591269254684448, "learning_rate": 1.1278468374925967e-06, "loss": 0.0015, "step": 56140 }, { "grad_norm": 0.03778735175728798, "learning_rate": 1.12203407197925e-06, "loss": 0.001, "step": 56150 }, { "grad_norm": 0.025590425357222557, "learning_rate": 1.116236154301409e-06, "loss": 0.0006, "step": 56160 }, { "grad_norm": 0.020328257232904434, "learning_rate": 1.11045308622032e-06, "loss": 0.0007, "step": 56170 }, { "grad_norm": 0.06055561080574989, "learning_rate": 1.1046848694927337e-06, "loss": 0.0009, "step": 56180 }, { "grad_norm": 0.016071079298853874, "learning_rate": 1.098931505870876e-06, "loss": 0.0007, "step": 56190 }, { "grad_norm": 0.02022407203912735, "learning_rate": 1.0931929971024657e-06, "loss": 0.0009, "step": 56200 }, { "grad_norm": 0.02980746701359749, "learning_rate": 1.0874693449307193e-06, "loss": 0.001, "step": 56210 }, { "grad_norm": 0.019421499222517014, "learning_rate": 1.081760551094324e-06, "loss": 0.0012, "step": 56220 }, { "grad_norm": 0.021514257416129112, "learning_rate": 1.0760666173274592e-06, "loss": 0.0007, "step": 56230 }, { "grad_norm": 0.014692848548293114, "learning_rate": 1.0703875453597967e-06, "loss": 0.0007, "step": 56240 }, { "grad_norm": 0.02517690323293209, "learning_rate": 1.0647233369164845e-06, "loss": 0.0007, "step": 56250 }, { "grad_norm": 0.03238905221223831, "learning_rate": 1.0590739937181625e-06, "loss": 0.0012, "step": 56260 }, { "grad_norm": 0.02590627782046795, "learning_rate": 1.053439517480953e-06, "loss": 0.0013, "step": 56270 }, { "grad_norm": 0.027010470628738403, "learning_rate": 1.047819909916453e-06, "loss": 0.0006, "step": 56280 }, { "grad_norm": 0.04054348170757294, "learning_rate": 1.0422151727317697e-06, "loss": 0.0009, "step": 56290 }, { "grad_norm": 0.024106524884700775, "learning_rate": 1.0366253076294462e-06, "loss": 0.0018, "step": 56300 }, { "grad_norm": 0.0450914092361927, "learning_rate": 1.031050316307558e-06, "loss": 0.0012, "step": 56310 }, { "grad_norm": 0.02462342008948326, "learning_rate": 1.0254902004596333e-06, "loss": 0.0015, "step": 56320 }, { "grad_norm": 0.020546048879623413, "learning_rate": 1.0199449617746882e-06, "loss": 0.001, "step": 56330 }, { "grad_norm": 0.02079073153436184, "learning_rate": 1.0144146019372247e-06, "loss": 0.0009, "step": 56340 }, { "grad_norm": 0.04180765897035599, "learning_rate": 1.0088991226272048e-06, "loss": 0.001, "step": 56350 }, { "grad_norm": 0.08451967686414719, "learning_rate": 1.003398525520105e-06, "loss": 0.0009, "step": 56360 }, { "grad_norm": 0.027002261951565742, "learning_rate": 9.979128122868552e-07, "loss": 0.001, "step": 56370 }, { "grad_norm": 0.026903053745627403, "learning_rate": 9.924419845938614e-07, "loss": 0.0007, "step": 56380 }, { "grad_norm": 0.03307001292705536, "learning_rate": 9.869860441030276e-07, "loss": 0.001, "step": 56390 }, { "grad_norm": 0.03628193587064743, "learning_rate": 9.815449924717169e-07, "loss": 0.0014, "step": 56400 }, { "grad_norm": 0.07498522102832794, "learning_rate": 9.761188313527791e-07, "loss": 0.0008, "step": 56410 }, { "grad_norm": 0.026495223864912987, "learning_rate": 9.70707562394546e-07, "loss": 0.0011, "step": 56420 }, { "grad_norm": 0.023048488423228264, "learning_rate": 9.653111872408027e-07, "loss": 0.0009, "step": 56430 }, { "grad_norm": 0.023212017491459846, "learning_rate": 9.599297075308434e-07, "loss": 0.0011, "step": 56440 }, { "grad_norm": 0.020699486136436462, "learning_rate": 9.545631248994048e-07, "loss": 0.001, "step": 56450 }, { "grad_norm": 0.019112173467874527, "learning_rate": 9.492114409767217e-07, "loss": 0.001, "step": 56460 }, { "grad_norm": 0.018358642235398293, "learning_rate": 9.438746573884938e-07, "loss": 0.0011, "step": 56470 }, { "grad_norm": 0.017715446650981903, "learning_rate": 9.385527757558909e-07, "loss": 0.0008, "step": 56480 }, { "grad_norm": 0.04682343080639839, "learning_rate": 9.332457976955644e-07, "loss": 0.0008, "step": 56490 }, { "grad_norm": 0.02024477906525135, "learning_rate": 9.279537248196247e-07, "loss": 0.0008, "step": 56500 }, { "grad_norm": 0.03628081828355789, "learning_rate": 9.2267655873568e-07, "loss": 0.0008, "step": 56510 }, { "grad_norm": 0.05198562517762184, "learning_rate": 9.174143010467762e-07, "loss": 0.0009, "step": 56520 }, { "grad_norm": 0.03178170695900917, "learning_rate": 9.12166953351462e-07, "loss": 0.0008, "step": 56530 }, { "grad_norm": 0.018652763217687607, "learning_rate": 9.069345172437404e-07, "loss": 0.0017, "step": 56540 }, { "grad_norm": 0.02150568552315235, "learning_rate": 9.017169943130843e-07, "loss": 0.001, "step": 56550 }, { "grad_norm": 0.022071080282330513, "learning_rate": 8.965143861444425e-07, "loss": 0.0007, "step": 56560 }, { "grad_norm": 0.02937060035765171, "learning_rate": 8.91326694318223e-07, "loss": 0.0009, "step": 56570 }, { "grad_norm": 0.0407356359064579, "learning_rate": 8.861539204103098e-07, "loss": 0.0006, "step": 56580 }, { "grad_norm": 0.010561196133494377, "learning_rate": 8.809960659920735e-07, "loss": 0.0008, "step": 56590 }, { "grad_norm": 0.02324056811630726, "learning_rate": 8.758531326303055e-07, "loss": 0.0006, "step": 56600 }, { "grad_norm": 0.018482590094208717, "learning_rate": 8.707251218873169e-07, "loss": 0.0005, "step": 56610 }, { "grad_norm": 0.03957027569413185, "learning_rate": 8.656120353208507e-07, "loss": 0.0014, "step": 56620 }, { "grad_norm": 0.018486393615603447, "learning_rate": 8.605138744841312e-07, "loss": 0.0008, "step": 56630 }, { "grad_norm": 0.020342862233519554, "learning_rate": 8.554306409258417e-07, "loss": 0.001, "step": 56640 }, { "grad_norm": 0.01421213336288929, "learning_rate": 8.503623361901358e-07, "loss": 0.0016, "step": 56650 }, { "grad_norm": 0.048132214695215225, "learning_rate": 8.453089618166377e-07, "loss": 0.0013, "step": 56660 }, { "grad_norm": 0.01917087286710739, "learning_rate": 8.402705193404137e-07, "loss": 0.0009, "step": 56670 }, { "grad_norm": 0.023053893819451332, "learning_rate": 8.352470102920174e-07, "loss": 0.0008, "step": 56680 }, { "grad_norm": 0.021439246833324432, "learning_rate": 8.302384361974669e-07, "loss": 0.0006, "step": 56690 }, { "grad_norm": 0.02189077064394951, "learning_rate": 8.252447985782231e-07, "loss": 0.0007, "step": 56700 }, { "grad_norm": 0.039146699011325836, "learning_rate": 8.202660989512279e-07, "loss": 0.0008, "step": 56710 }, { "grad_norm": 0.05120660364627838, "learning_rate": 8.153023388288772e-07, "loss": 0.0008, "step": 56720 }, { "grad_norm": 0.022791054099798203, "learning_rate": 8.103535197190204e-07, "loss": 0.001, "step": 56730 }, { "grad_norm": 0.02317900024354458, "learning_rate": 8.054196431249993e-07, "loss": 0.0008, "step": 56740 }, { "grad_norm": 0.052933838218450546, "learning_rate": 8.005007105455709e-07, "loss": 0.0008, "step": 56750 }, { "grad_norm": 0.029369402676820755, "learning_rate": 7.95596723474995e-07, "loss": 0.0009, "step": 56760 }, { "grad_norm": 0.025030996650457382, "learning_rate": 7.907076834029692e-07, "loss": 0.0009, "step": 56770 }, { "grad_norm": 0.012739473953843117, "learning_rate": 7.858335918146498e-07, "loss": 0.0007, "step": 56780 }, { "grad_norm": 0.021789664402604103, "learning_rate": 7.809744501906635e-07, "loss": 0.0011, "step": 56790 }, { "grad_norm": 0.014462902210652828, "learning_rate": 7.761302600070797e-07, "loss": 0.0009, "step": 56800 }, { "grad_norm": 0.020118212327361107, "learning_rate": 7.713010227354545e-07, "loss": 0.0008, "step": 56810 }, { "grad_norm": 0.013999760150909424, "learning_rate": 7.664867398427589e-07, "loss": 0.0008, "step": 56820 }, { "grad_norm": 0.016642186790704727, "learning_rate": 7.616874127914619e-07, "loss": 0.0011, "step": 56830 }, { "grad_norm": 0.025324054062366486, "learning_rate": 7.569030430394641e-07, "loss": 0.001, "step": 56840 }, { "grad_norm": 0.03921608626842499, "learning_rate": 7.521336320401306e-07, "loss": 0.0009, "step": 56850 }, { "grad_norm": 0.020126227289438248, "learning_rate": 7.473791812422915e-07, "loss": 0.0008, "step": 56860 }, { "grad_norm": 0.01938488893210888, "learning_rate": 7.426396920902134e-07, "loss": 0.0009, "step": 56870 }, { "grad_norm": 0.01387827843427658, "learning_rate": 7.379151660236283e-07, "loss": 0.0015, "step": 56880 }, { "grad_norm": 0.07034043222665787, "learning_rate": 7.332056044777324e-07, "loss": 0.0009, "step": 56890 }, { "grad_norm": 0.014026197604835033, "learning_rate": 7.285110088831537e-07, "loss": 0.0006, "step": 56900 }, { "grad_norm": 0.07433885335922241, "learning_rate": 7.238313806659902e-07, "loss": 0.0008, "step": 56910 }, { "grad_norm": 0.018412191420793533, "learning_rate": 7.191667212477993e-07, "loss": 0.0007, "step": 56920 }, { "grad_norm": 0.04953375831246376, "learning_rate": 7.145170320455697e-07, "loss": 0.0007, "step": 56930 }, { "grad_norm": 0.022075623273849487, "learning_rate": 7.098823144717604e-07, "loss": 0.0009, "step": 56940 }, { "grad_norm": 0.029974503442645073, "learning_rate": 7.052625699342674e-07, "loss": 0.001, "step": 56950 }, { "grad_norm": 0.013598878867924213, "learning_rate": 7.006577998364628e-07, "loss": 0.0006, "step": 56960 }, { "grad_norm": 0.03732705116271973, "learning_rate": 6.96068005577133e-07, "loss": 0.0008, "step": 56970 }, { "grad_norm": 0.02487182430922985, "learning_rate": 6.914931885505627e-07, "loss": 0.001, "step": 56980 }, { "grad_norm": 0.014176948927342892, "learning_rate": 6.869333501464347e-07, "loss": 0.0006, "step": 56990 }, { "grad_norm": 0.017099523916840553, "learning_rate": 6.823884917499246e-07, "loss": 0.001, "step": 57000 }, { "grad_norm": 0.013369444757699966, "learning_rate": 6.778586147416278e-07, "loss": 0.0007, "step": 57010 }, { "grad_norm": 0.011981801129877567, "learning_rate": 6.733437204976156e-07, "loss": 0.0008, "step": 57020 }, { "grad_norm": 0.0179788488894701, "learning_rate": 6.688438103893857e-07, "loss": 0.0007, "step": 57030 }, { "grad_norm": 0.02410983480513096, "learning_rate": 6.64358885783889e-07, "loss": 0.0015, "step": 57040 }, { "grad_norm": 0.01057495828717947, "learning_rate": 6.598889480435299e-07, "loss": 0.0007, "step": 57050 }, { "grad_norm": 0.02715732343494892, "learning_rate": 6.554339985261615e-07, "loss": 0.0008, "step": 57060 }, { "grad_norm": 0.01833959110081196, "learning_rate": 6.509940385850733e-07, "loss": 0.0011, "step": 57070 }, { "grad_norm": 0.05339796468615532, "learning_rate": 6.465690695690141e-07, "loss": 0.001, "step": 57080 }, { "grad_norm": 0.048636581748723984, "learning_rate": 6.421590928221699e-07, "loss": 0.0009, "step": 57090 }, { "grad_norm": 0.013709889724850655, "learning_rate": 6.377641096841691e-07, "loss": 0.0008, "step": 57100 }, { "grad_norm": 0.01500894594937563, "learning_rate": 6.333841214901048e-07, "loss": 0.001, "step": 57110 }, { "grad_norm": 0.020183324813842773, "learning_rate": 6.290191295704906e-07, "loss": 0.0009, "step": 57120 }, { "grad_norm": 0.021785955876111984, "learning_rate": 6.246691352513046e-07, "loss": 0.0009, "step": 57130 }, { "grad_norm": 0.01347427349537611, "learning_rate": 6.203341398539452e-07, "loss": 0.0006, "step": 57140 }, { "grad_norm": 0.028693972155451775, "learning_rate": 6.160141446952872e-07, "loss": 0.0007, "step": 57150 }, { "grad_norm": 0.06798223406076431, "learning_rate": 6.11709151087625e-07, "loss": 0.0012, "step": 57160 }, { "grad_norm": 0.03456433489918709, "learning_rate": 6.074191603386958e-07, "loss": 0.0008, "step": 57170 }, { "grad_norm": 0.019345466047525406, "learning_rate": 6.031441737516907e-07, "loss": 0.0009, "step": 57180 }, { "grad_norm": 0.025014769285917282, "learning_rate": 5.988841926252431e-07, "loss": 0.0007, "step": 57190 }, { "grad_norm": 0.04328048229217529, "learning_rate": 5.946392182534066e-07, "loss": 0.0009, "step": 57200 }, { "grad_norm": 0.024884896352887154, "learning_rate": 5.90409251925711e-07, "loss": 0.0009, "step": 57210 }, { "grad_norm": 0.020227773115038872, "learning_rate": 5.861942949270949e-07, "loss": 0.0007, "step": 57220 }, { "grad_norm": 0.015037216246128082, "learning_rate": 5.819943485379564e-07, "loss": 0.002, "step": 57230 }, { "grad_norm": 0.026719139888882637, "learning_rate": 5.778094140341306e-07, "loss": 0.0006, "step": 57240 }, { "grad_norm": 0.015732500702142715, "learning_rate": 5.736394926868893e-07, "loss": 0.0009, "step": 57250 }, { "grad_norm": 0.014531264081597328, "learning_rate": 5.694845857629416e-07, "loss": 0.0008, "step": 57260 }, { "grad_norm": 0.010473146103322506, "learning_rate": 5.653446945244334e-07, "loss": 0.0007, "step": 57270 }, { "grad_norm": 0.01724223420023918, "learning_rate": 5.612198202289698e-07, "loss": 0.0012, "step": 57280 }, { "grad_norm": 0.018022166565060616, "learning_rate": 5.571099641295596e-07, "loss": 0.0008, "step": 57290 }, { "grad_norm": 0.01179509423673153, "learning_rate": 5.530151274746875e-07, "loss": 0.0007, "step": 57300 }, { "grad_norm": 0.03208477050065994, "learning_rate": 5.489353115082418e-07, "loss": 0.001, "step": 57310 }, { "grad_norm": 0.07169181853532791, "learning_rate": 5.4487051746957e-07, "loss": 0.0008, "step": 57320 }, { "grad_norm": 0.012730110436677933, "learning_rate": 5.408207465934511e-07, "loss": 0.0007, "step": 57330 }, { "grad_norm": 0.023364122956991196, "learning_rate": 5.3678600011009e-07, "loss": 0.0013, "step": 57340 }, { "grad_norm": 0.013582487590610981, "learning_rate": 5.327662792451449e-07, "loss": 0.0007, "step": 57350 }, { "grad_norm": 0.028490006923675537, "learning_rate": 5.287615852196947e-07, "loss": 0.0009, "step": 57360 }, { "grad_norm": 0.08043074607849121, "learning_rate": 5.247719192502665e-07, "loss": 0.0012, "step": 57370 }, { "grad_norm": 0.026175573468208313, "learning_rate": 5.207972825488128e-07, "loss": 0.0009, "step": 57380 }, { "grad_norm": 0.013428502716124058, "learning_rate": 5.168376763227178e-07, "loss": 0.0007, "step": 57390 }, { "grad_norm": 0.021243499591946602, "learning_rate": 5.12893101774814e-07, "loss": 0.0011, "step": 57400 }, { "grad_norm": 0.07226214557886124, "learning_rate": 5.089635601033483e-07, "loss": 0.0011, "step": 57410 }, { "grad_norm": 0.025519439950585365, "learning_rate": 5.050490525020213e-07, "loss": 0.0007, "step": 57420 }, { "grad_norm": 0.0231047160923481, "learning_rate": 5.011495801599541e-07, "loss": 0.0009, "step": 57430 }, { "grad_norm": 0.035693444311618805, "learning_rate": 4.972651442616994e-07, "loss": 0.001, "step": 57440 }, { "grad_norm": 0.024099085479974747, "learning_rate": 4.933957459872574e-07, "loss": 0.0011, "step": 57450 }, { "grad_norm": 0.019154084846377373, "learning_rate": 4.895413865120324e-07, "loss": 0.0006, "step": 57460 }, { "grad_norm": 0.030415743589401245, "learning_rate": 4.857020670068935e-07, "loss": 0.0007, "step": 57470 }, { "grad_norm": 0.02686520293354988, "learning_rate": 4.818777886381132e-07, "loss": 0.0008, "step": 57480 }, { "grad_norm": 0.023350752890110016, "learning_rate": 4.780685525674122e-07, "loss": 0.0008, "step": 57490 }, { "grad_norm": 0.02210848033428192, "learning_rate": 4.7427435995193724e-07, "loss": 0.0006, "step": 57500 }, { "grad_norm": 0.022626854479312897, "learning_rate": 4.7049521194425515e-07, "loss": 0.0007, "step": 57510 }, { "grad_norm": 0.02720767632126808, "learning_rate": 4.667311096923754e-07, "loss": 0.0006, "step": 57520 }, { "grad_norm": 0.012848918326199055, "learning_rate": 4.6298205433973895e-07, "loss": 0.0007, "step": 57530 }, { "grad_norm": 0.0243093091994524, "learning_rate": 4.5924804702520696e-07, "loss": 0.001, "step": 57540 }, { "grad_norm": 0.02392641454935074, "learning_rate": 4.5552908888306655e-07, "loss": 0.0009, "step": 57550 }, { "grad_norm": 0.020626628771424294, "learning_rate": 4.5182518104304185e-07, "loss": 0.0009, "step": 57560 }, { "grad_norm": 0.01630723476409912, "learning_rate": 4.4813632463028277e-07, "loss": 0.0007, "step": 57570 }, { "grad_norm": 0.01829487830400467, "learning_rate": 4.444625207653763e-07, "loss": 0.0007, "step": 57580 }, { "grad_norm": 0.013172420673072338, "learning_rate": 4.4080377056430753e-07, "loss": 0.0011, "step": 57590 }, { "grad_norm": 0.022809110581874847, "learning_rate": 4.371600751385263e-07, "loss": 0.0007, "step": 57600 }, { "grad_norm": 0.024329084903001785, "learning_rate": 4.3353143559487495e-07, "loss": 0.0008, "step": 57610 }, { "grad_norm": 0.027271118015050888, "learning_rate": 4.2991785303565513e-07, "loss": 0.0007, "step": 57620 }, { "grad_norm": 0.013731502927839756, "learning_rate": 4.2631932855856647e-07, "loss": 0.0007, "step": 57630 }, { "grad_norm": 0.015879452228546143, "learning_rate": 4.2273586325674576e-07, "loss": 0.001, "step": 57640 }, { "grad_norm": 0.013820669613778591, "learning_rate": 4.1916745821876103e-07, "loss": 0.0006, "step": 57650 }, { "grad_norm": 0.01840285025537014, "learning_rate": 4.156141145285897e-07, "loss": 0.0008, "step": 57660 }, { "grad_norm": 0.0305198784917593, "learning_rate": 4.1207583326566267e-07, "loss": 0.0007, "step": 57670 }, { "grad_norm": 0.014680515974760056, "learning_rate": 4.085526155047925e-07, "loss": 0.001, "step": 57680 }, { "grad_norm": 0.02037234976887703, "learning_rate": 4.050444623162564e-07, "loss": 0.0007, "step": 57690 }, { "grad_norm": 0.01766361854970455, "learning_rate": 4.015513747657351e-07, "loss": 0.001, "step": 57700 }, { "grad_norm": 0.03216957673430443, "learning_rate": 3.9807335391433554e-07, "loss": 0.0007, "step": 57710 }, { "grad_norm": 0.013260153122246265, "learning_rate": 3.946104008185847e-07, "loss": 0.0007, "step": 57720 }, { "grad_norm": 0.02028302289545536, "learning_rate": 3.9116251653044113e-07, "loss": 0.0008, "step": 57730 }, { "grad_norm": 0.028574498370289803, "learning_rate": 3.877297020972781e-07, "loss": 0.0007, "step": 57740 }, { "grad_norm": 0.04185471311211586, "learning_rate": 3.8431195856190036e-07, "loss": 0.0009, "step": 57750 }, { "grad_norm": 0.04393179342150688, "learning_rate": 3.8090928696251635e-07, "loss": 0.0008, "step": 57760 }, { "grad_norm": 0.04001732915639877, "learning_rate": 3.77521688332777e-07, "loss": 0.0007, "step": 57770 }, { "grad_norm": 0.04176328331232071, "learning_rate": 3.741491637017425e-07, "loss": 0.0006, "step": 57780 }, { "grad_norm": 0.013315269723534584, "learning_rate": 3.707917140939043e-07, "loss": 0.0009, "step": 57790 }, { "grad_norm": 0.022320963442325592, "learning_rate": 3.6744934052915235e-07, "loss": 0.0008, "step": 57800 }, { "grad_norm": 0.014721088111400604, "learning_rate": 3.641220440228188e-07, "loss": 0.0007, "step": 57810 }, { "grad_norm": 0.016193363815546036, "learning_rate": 3.608098255856562e-07, "loss": 0.0007, "step": 57820 }, { "grad_norm": 0.022135848179459572, "learning_rate": 3.575126862238154e-07, "loss": 0.0008, "step": 57830 }, { "grad_norm": 0.033562809228897095, "learning_rate": 3.5423062693888955e-07, "loss": 0.0007, "step": 57840 }, { "grad_norm": 0.03632469102740288, "learning_rate": 3.509636487278756e-07, "loss": 0.0016, "step": 57850 }, { "grad_norm": 0.023720284923911095, "learning_rate": 3.4771175258320186e-07, "loss": 0.0007, "step": 57860 }, { "grad_norm": 0.04125068336725235, "learning_rate": 3.4447493949270047e-07, "loss": 0.0006, "step": 57870 }, { "grad_norm": 0.0258037056773901, "learning_rate": 3.4125321043964045e-07, "loss": 0.0006, "step": 57880 }, { "grad_norm": 0.015986893326044083, "learning_rate": 3.380465664026833e-07, "loss": 0.0009, "step": 57890 }, { "grad_norm": 0.020974045619368553, "learning_rate": 3.348550083559388e-07, "loss": 0.0008, "step": 57900 }, { "grad_norm": 0.016071951016783714, "learning_rate": 3.316785372689091e-07, "loss": 0.0005, "step": 57910 }, { "grad_norm": 0.01395060122013092, "learning_rate": 3.285171541065224e-07, "loss": 0.0009, "step": 57920 }, { "grad_norm": 0.02357635833323002, "learning_rate": 3.253708598291272e-07, "loss": 0.0007, "step": 57930 }, { "grad_norm": 0.014270360581576824, "learning_rate": 3.2223965539248116e-07, "loss": 0.0007, "step": 57940 }, { "grad_norm": 0.01821996457874775, "learning_rate": 3.1912354174776227e-07, "loss": 0.0006, "step": 57950 }, { "grad_norm": 0.022186970338225365, "learning_rate": 3.1602251984155786e-07, "loss": 0.0007, "step": 57960 }, { "grad_norm": 0.03539460897445679, "learning_rate": 3.1293659061589207e-07, "loss": 0.0009, "step": 57970 }, { "grad_norm": 0.05231976881623268, "learning_rate": 3.098657550081707e-07, "loss": 0.0009, "step": 57980 }, { "grad_norm": 0.02182178385555744, "learning_rate": 3.068100139512475e-07, "loss": 0.0007, "step": 57990 }, { "grad_norm": 0.033768318593502045, "learning_rate": 3.037693683733689e-07, "loss": 0.0007, "step": 58000 }, { "grad_norm": 0.010510694235563278, "learning_rate": 3.007438191982015e-07, "loss": 0.0008, "step": 58010 }, { "grad_norm": 0.014314056374132633, "learning_rate": 2.9773336734482684e-07, "loss": 0.0007, "step": 58020 }, { "grad_norm": 0.021168867126107216, "learning_rate": 2.9473801372774667e-07, "loss": 0.0006, "step": 58030 }, { "grad_norm": 0.054527975618839264, "learning_rate": 2.91757759256861e-07, "loss": 0.0009, "step": 58040 }, { "grad_norm": 0.035892512649297714, "learning_rate": 2.887926048375067e-07, "loss": 0.0012, "step": 58050 }, { "grad_norm": 0.028710346668958664, "learning_rate": 2.858425513704022e-07, "loss": 0.001, "step": 58060 }, { "grad_norm": 0.021670598536729813, "learning_rate": 2.8290759975170834e-07, "loss": 0.0008, "step": 58070 }, { "grad_norm": 0.028783001005649567, "learning_rate": 2.799877508729787e-07, "loss": 0.0008, "step": 58080 }, { "grad_norm": 0.01193590834736824, "learning_rate": 2.770830056211926e-07, "loss": 0.0006, "step": 58090 }, { "grad_norm": 0.024029796943068504, "learning_rate": 2.741933648787331e-07, "loss": 0.0023, "step": 58100 }, { "grad_norm": 0.03953518345952034, "learning_rate": 2.7131882952339263e-07, "loss": 0.0006, "step": 58110 }, { "grad_norm": 0.027089809998869896, "learning_rate": 2.684594004283836e-07, "loss": 0.0009, "step": 58120 }, { "grad_norm": 0.013301211409270763, "learning_rate": 2.6561507846232234e-07, "loss": 0.0008, "step": 58130 }, { "grad_norm": 0.02184465527534485, "learning_rate": 2.6278586448924005e-07, "loss": 0.0006, "step": 58140 }, { "grad_norm": 0.024057161062955856, "learning_rate": 2.5997175936857685e-07, "loss": 0.0006, "step": 58150 }, { "grad_norm": 0.02915639989078045, "learning_rate": 2.57172763955188e-07, "loss": 0.0012, "step": 58160 }, { "grad_norm": 0.04408879578113556, "learning_rate": 2.543888790993265e-07, "loss": 0.0009, "step": 58170 }, { "grad_norm": 0.01425459049642086, "learning_rate": 2.5162010564666607e-07, "loss": 0.0008, "step": 58180 }, { "grad_norm": 0.04378523305058479, "learning_rate": 2.488664444382893e-07, "loss": 0.0014, "step": 58190 }, { "grad_norm": 0.060189321637153625, "learning_rate": 2.461278963106828e-07, "loss": 0.0017, "step": 58200 }, { "grad_norm": 0.012537211179733276, "learning_rate": 2.434044620957421e-07, "loss": 0.0015, "step": 58210 }, { "grad_norm": 0.013486796990036964, "learning_rate": 2.406961426207832e-07, "loss": 0.0007, "step": 58220 }, { "grad_norm": 0.03278035297989845, "learning_rate": 2.380029387085203e-07, "loss": 0.0007, "step": 58230 }, { "grad_norm": 0.05238354206085205, "learning_rate": 2.353248511770767e-07, "loss": 0.0008, "step": 58240 }, { "grad_norm": 0.018339594826102257, "learning_rate": 2.3266188083997942e-07, "loss": 0.0013, "step": 58250 }, { "grad_norm": 0.008380639366805553, "learning_rate": 2.3001402850617027e-07, "loss": 0.0008, "step": 58260 }, { "grad_norm": 0.01577146351337433, "learning_rate": 2.2738129498000581e-07, "loss": 0.0005, "step": 58270 }, { "grad_norm": 0.016962306573987007, "learning_rate": 2.2476368106122414e-07, "loss": 0.0005, "step": 58280 }, { "grad_norm": 0.020171085372567177, "learning_rate": 2.2216118754500582e-07, "loss": 0.0007, "step": 58290 }, { "grad_norm": 0.026708990335464478, "learning_rate": 2.1957381522190735e-07, "loss": 0.0011, "step": 58300 }, { "grad_norm": 0.02821129932999611, "learning_rate": 2.1700156487790558e-07, "loss": 0.0009, "step": 58310 }, { "grad_norm": 0.013279160484671593, "learning_rate": 2.1444443729439213e-07, "loss": 0.0008, "step": 58320 }, { "grad_norm": 0.016370581462979317, "learning_rate": 2.1190243324814007e-07, "loss": 0.0008, "step": 58330 }, { "grad_norm": 0.04296089708805084, "learning_rate": 2.0937555351135395e-07, "loss": 0.0008, "step": 58340 }, { "grad_norm": 0.012732274830341339, "learning_rate": 2.0686379885162532e-07, "loss": 0.001, "step": 58350 }, { "grad_norm": 0.02990344539284706, "learning_rate": 2.0436717003196604e-07, "loss": 0.0009, "step": 58360 }, { "grad_norm": 0.02170366235077381, "learning_rate": 2.0188566781078057e-07, "loss": 0.0006, "step": 58370 }, { "grad_norm": 0.03874712437391281, "learning_rate": 1.994192929418881e-07, "loss": 0.0007, "step": 58380 }, { "grad_norm": 0.02949521131813526, "learning_rate": 1.9696804617451158e-07, "loss": 0.0007, "step": 58390 }, { "grad_norm": 0.04487626627087593, "learning_rate": 1.9453192825326093e-07, "loss": 0.001, "step": 58400 }, { "grad_norm": 0.07646574079990387, "learning_rate": 1.9211093991817753e-07, "loss": 0.0011, "step": 58410 }, { "grad_norm": 0.01203225925564766, "learning_rate": 1.8970508190468973e-07, "loss": 0.0008, "step": 58420 }, { "grad_norm": 0.025939367711544037, "learning_rate": 1.8731435494362958e-07, "loss": 0.0007, "step": 58430 }, { "grad_norm": 0.030960580334067345, "learning_rate": 1.849387597612495e-07, "loss": 0.0009, "step": 58440 }, { "grad_norm": 0.018326127901673317, "learning_rate": 1.8257829707917228e-07, "loss": 0.0006, "step": 58450 }, { "grad_norm": 0.028956076130270958, "learning_rate": 1.8023296761446317e-07, "loss": 0.0007, "step": 58460 }, { "grad_norm": 0.019083397462964058, "learning_rate": 1.7790277207956341e-07, "loss": 0.0007, "step": 58470 }, { "grad_norm": 0.015154946595430374, "learning_rate": 1.7558771118232343e-07, "loss": 0.0012, "step": 58480 }, { "grad_norm": 0.022100744768977165, "learning_rate": 1.7328778562599734e-07, "loss": 0.0006, "step": 58490 }, { "grad_norm": 0.022382492199540138, "learning_rate": 1.7100299610924298e-07, "loss": 0.0006, "step": 58500 }, { "grad_norm": 0.019731638953089714, "learning_rate": 1.6873334332612733e-07, "loss": 0.0007, "step": 58510 }, { "grad_norm": 0.01562606729567051, "learning_rate": 1.6647882796609894e-07, "loss": 0.0007, "step": 58520 }, { "grad_norm": 0.02091275155544281, "learning_rate": 1.6423945071402102e-07, "loss": 0.0009, "step": 58530 }, { "grad_norm": 0.014786571264266968, "learning_rate": 1.6201521225016614e-07, "loss": 0.0016, "step": 58540 }, { "grad_norm": 0.02675425074994564, "learning_rate": 1.598061132501938e-07, "loss": 0.0009, "step": 58550 }, { "grad_norm": 0.016798026859760284, "learning_rate": 1.576121543851672e-07, "loss": 0.0007, "step": 58560 }, { "grad_norm": 0.02058596909046173, "learning_rate": 1.5543333632155876e-07, "loss": 0.0012, "step": 58570 }, { "grad_norm": 0.01897372119128704, "learning_rate": 1.5326965972123352e-07, "loss": 0.0008, "step": 58580 }, { "grad_norm": 0.0364781878888607, "learning_rate": 1.5112112524146016e-07, "loss": 0.0007, "step": 58590 }, { "grad_norm": 0.022569527849555016, "learning_rate": 1.4898773353489992e-07, "loss": 0.0007, "step": 58600 }, { "grad_norm": 0.026280682533979416, "learning_rate": 1.4686948524962884e-07, "loss": 0.0007, "step": 58610 }, { "grad_norm": 0.01322106085717678, "learning_rate": 1.4476638102911556e-07, "loss": 0.0009, "step": 58620 }, { "grad_norm": 0.06749562174081802, "learning_rate": 1.4267842151222123e-07, "loss": 0.0012, "step": 58630 }, { "grad_norm": 0.017565185204148293, "learning_rate": 1.4060560733321626e-07, "loss": 0.0006, "step": 58640 }, { "grad_norm": 0.01596849597990513, "learning_rate": 1.385479391217692e-07, "loss": 0.0005, "step": 58650 }, { "grad_norm": 0.020029032602906227, "learning_rate": 1.365054175029412e-07, "loss": 0.0007, "step": 58660 }, { "grad_norm": 0.04898257926106453, "learning_rate": 1.3447804309719702e-07, "loss": 0.0012, "step": 58670 }, { "grad_norm": 0.025315523147583008, "learning_rate": 1.3246581652040512e-07, "loss": 0.0013, "step": 58680 }, { "grad_norm": 0.04479203000664711, "learning_rate": 1.3046873838381546e-07, "loss": 0.0009, "step": 58690 }, { "grad_norm": 0.017896732315421104, "learning_rate": 1.2848680929409828e-07, "loss": 0.0006, "step": 58700 }, { "grad_norm": 0.01853989064693451, "learning_rate": 1.2652002985331091e-07, "loss": 0.0006, "step": 58710 }, { "grad_norm": 0.034007053822278976, "learning_rate": 1.2456840065889764e-07, "loss": 0.0008, "step": 58720 }, { "grad_norm": 0.016935469582676888, "learning_rate": 1.226319223037231e-07, "loss": 0.0005, "step": 58730 }, { "grad_norm": 0.01716974563896656, "learning_rate": 1.2071059537603902e-07, "loss": 0.001, "step": 58740 }, { "grad_norm": 0.02342022955417633, "learning_rate": 1.1880442045948403e-07, "loss": 0.0008, "step": 58750 }, { "grad_norm": 0.029138054698705673, "learning_rate": 1.1691339813311164e-07, "loss": 0.0007, "step": 58760 }, { "grad_norm": 0.03460279479622841, "learning_rate": 1.1503752897136233e-07, "loss": 0.0011, "step": 58770 }, { "grad_norm": 0.022142350673675537, "learning_rate": 1.1317681354407472e-07, "loss": 0.0008, "step": 58780 }, { "grad_norm": 0.045577820390462875, "learning_rate": 1.1133125241649111e-07, "loss": 0.0009, "step": 58790 }, { "grad_norm": 0.01320070680230856, "learning_rate": 1.0950084614922973e-07, "loss": 0.0014, "step": 58800 }, { "grad_norm": 0.02399781346321106, "learning_rate": 1.0768559529834021e-07, "loss": 0.001, "step": 58810 }, { "grad_norm": 0.023977188393473625, "learning_rate": 1.0588550041522594e-07, "loss": 0.0009, "step": 58820 }, { "grad_norm": 0.024370668455958366, "learning_rate": 1.0410056204672169e-07, "loss": 0.0006, "step": 58830 }, { "grad_norm": 0.019247496500611305, "learning_rate": 1.0233078073504376e-07, "loss": 0.0009, "step": 58840 }, { "grad_norm": 0.02312154322862625, "learning_rate": 1.0057615701780654e-07, "loss": 0.0009, "step": 58850 }, { "grad_norm": 0.02360367961227894, "learning_rate": 9.883669142801144e-08, "loss": 0.0005, "step": 58860 }, { "grad_norm": 0.019299745559692383, "learning_rate": 9.711238449406356e-08, "loss": 0.0006, "step": 58870 }, { "grad_norm": 0.009473179467022419, "learning_rate": 9.540323673976614e-08, "loss": 0.0012, "step": 58880 }, { "grad_norm": 0.021702522411942482, "learning_rate": 9.370924868430942e-08, "loss": 0.001, "step": 58890 }, { "grad_norm": 0.019664596766233444, "learning_rate": 9.203042084228175e-08, "loss": 0.0007, "step": 58900 }, { "grad_norm": 0.04128976911306381, "learning_rate": 9.036675372366965e-08, "loss": 0.0007, "step": 58910 }, { "grad_norm": 0.018633900210261345, "learning_rate": 8.871824783385218e-08, "loss": 0.0008, "step": 58920 }, { "grad_norm": 0.02080259844660759, "learning_rate": 8.7084903673601e-08, "loss": 0.0016, "step": 58930 }, { "grad_norm": 0.016803056001663208, "learning_rate": 8.546672173908032e-08, "loss": 0.0009, "step": 58940 }, { "grad_norm": 0.017253603786230087, "learning_rate": 8.386370252185249e-08, "loss": 0.001, "step": 58950 }, { "grad_norm": 0.01580902375280857, "learning_rate": 8.227584650887243e-08, "loss": 0.0007, "step": 58960 }, { "grad_norm": 0.02810879424214363, "learning_rate": 8.070315418249319e-08, "loss": 0.0008, "step": 58970 }, { "grad_norm": 0.016160640865564346, "learning_rate": 7.914562602044929e-08, "loss": 0.0006, "step": 58980 }, { "grad_norm": 0.011723443865776062, "learning_rate": 7.76032624958789e-08, "loss": 0.0007, "step": 58990 }, { "grad_norm": 0.031761910766363144, "learning_rate": 7.607606407731282e-08, "loss": 0.0008, "step": 59000 }, { "grad_norm": 0.023405853658914566, "learning_rate": 7.45640312286744e-08, "loss": 0.0015, "step": 59010 }, { "grad_norm": 0.03277507796883583, "learning_rate": 7.306716440927952e-08, "loss": 0.0008, "step": 59020 }, { "grad_norm": 0.021557779982686043, "learning_rate": 7.15854640738367e-08, "loss": 0.0007, "step": 59030 }, { "grad_norm": 0.01860639825463295, "learning_rate": 7.011893067244701e-08, "loss": 0.001, "step": 59040 }, { "grad_norm": 0.022624382749199867, "learning_rate": 6.866756465060408e-08, "loss": 0.0007, "step": 59050 }, { "grad_norm": 0.016629211604595184, "learning_rate": 6.723136644918859e-08, "loss": 0.0008, "step": 59060 }, { "grad_norm": 0.020711729303002357, "learning_rate": 6.581033650449042e-08, "loss": 0.0007, "step": 59070 }, { "grad_norm": 0.02008519321680069, "learning_rate": 6.440447524817539e-08, "loss": 0.001, "step": 59080 }, { "grad_norm": 0.028668919578194618, "learning_rate": 6.301378310730743e-08, "loss": 0.0008, "step": 59090 }, { "grad_norm": 0.015255124308168888, "learning_rate": 6.163826050434307e-08, "loss": 0.0005, "step": 59100 }, { "grad_norm": 0.018527450039982796, "learning_rate": 6.027790785713139e-08, "loss": 0.0008, "step": 59110 }, { "grad_norm": 0.01773693971335888, "learning_rate": 5.89327255789085e-08, "loss": 0.0006, "step": 59120 }, { "grad_norm": 0.022713668644428253, "learning_rate": 5.7602714078303085e-08, "loss": 0.0007, "step": 59130 }, { "grad_norm": 0.018843647092580795, "learning_rate": 5.628787375934197e-08, "loss": 0.0008, "step": 59140 }, { "grad_norm": 0.010801052674651146, "learning_rate": 5.498820502143898e-08, "loss": 0.0006, "step": 59150 }, { "grad_norm": 0.02943783439695835, "learning_rate": 5.370370825939497e-08, "loss": 0.0007, "step": 59160 }, { "grad_norm": 0.042801808565855026, "learning_rate": 5.243438386340893e-08, "loss": 0.0011, "step": 59170 }, { "grad_norm": 0.014363701455295086, "learning_rate": 5.118023221907242e-08, "loss": 0.0008, "step": 59180 }, { "grad_norm": 0.02987738512456417, "learning_rate": 4.994125370735292e-08, "loss": 0.0011, "step": 59190 }, { "grad_norm": 0.02679300121963024, "learning_rate": 4.871744870462713e-08, "loss": 0.0023, "step": 59200 }, { "grad_norm": 0.032146237790584564, "learning_rate": 4.7508817582658794e-08, "loss": 0.0009, "step": 59210 }, { "grad_norm": 0.018410947173833847, "learning_rate": 4.631536070858755e-08, "loss": 0.001, "step": 59220 }, { "grad_norm": 0.01440221443772316, "learning_rate": 4.513707844495674e-08, "loss": 0.0006, "step": 59230 }, { "grad_norm": 0.01728389784693718, "learning_rate": 4.3973971149702255e-08, "loss": 0.0007, "step": 59240 }, { "grad_norm": 0.02073039673268795, "learning_rate": 4.2826039176147025e-08, "loss": 0.001, "step": 59250 }, { "grad_norm": 0.02113829366862774, "learning_rate": 4.169328287299545e-08, "loss": 0.0006, "step": 59260 }, { "grad_norm": 0.018285177648067474, "learning_rate": 4.057570258435006e-08, "loss": 0.0012, "step": 59270 }, { "grad_norm": 0.047166548669338226, "learning_rate": 3.947329864970595e-08, "loss": 0.0009, "step": 59280 }, { "grad_norm": 0.0525626577436924, "learning_rate": 3.8386071403939686e-08, "loss": 0.0011, "step": 59290 }, { "grad_norm": 0.00984828732907772, "learning_rate": 3.731402117733152e-08, "loss": 0.0005, "step": 59300 }, { "grad_norm": 0.016130678355693817, "learning_rate": 3.625714829552651e-08, "loss": 0.0008, "step": 59310 }, { "grad_norm": 0.020337490364909172, "learning_rate": 3.5215453079590065e-08, "loss": 0.0006, "step": 59320 }, { "grad_norm": 0.028064850717782974, "learning_rate": 3.4188935845952396e-08, "loss": 0.0006, "step": 59330 }, { "grad_norm": 0.032436553388834, "learning_rate": 3.3177596906447396e-08, "loss": 0.0015, "step": 59340 }, { "grad_norm": 0.035957857966423035, "learning_rate": 3.218143656829043e-08, "loss": 0.0009, "step": 59350 }, { "grad_norm": 0.07845748215913773, "learning_rate": 3.120045513408387e-08, "loss": 0.0011, "step": 59360 }, { "grad_norm": 0.026551933959126472, "learning_rate": 3.02346529018338e-08, "loss": 0.0007, "step": 59370 }, { "grad_norm": 0.03219938650727272, "learning_rate": 2.9284030164922204e-08, "loss": 0.0006, "step": 59380 }, { "grad_norm": 0.01407028641551733, "learning_rate": 2.8348587212123634e-08, "loss": 0.0006, "step": 59390 }, { "grad_norm": 0.0074916197918355465, "learning_rate": 2.7428324327594125e-08, "loss": 0.001, "step": 59400 }, { "grad_norm": 0.021404771134257317, "learning_rate": 2.6523241790893383e-08, "loss": 0.0011, "step": 59410 }, { "grad_norm": 0.020228080451488495, "learning_rate": 2.563333987695704e-08, "loss": 0.0011, "step": 59420 }, { "grad_norm": 0.01329786702990532, "learning_rate": 2.4758618856118852e-08, "loss": 0.0006, "step": 59430 }, { "grad_norm": 0.02178974077105522, "learning_rate": 2.3899078994088497e-08, "loss": 0.0008, "step": 59440 }, { "grad_norm": 0.04324052110314369, "learning_rate": 2.3054720551973775e-08, "loss": 0.0007, "step": 59450 }, { "grad_norm": 0.011033888906240463, "learning_rate": 2.222554378627506e-08, "loss": 0.0006, "step": 59460 }, { "grad_norm": 0.02090030536055565, "learning_rate": 2.1411548948868653e-08, "loss": 0.001, "step": 59470 }, { "grad_norm": 0.03622133657336235, "learning_rate": 2.0612736287023426e-08, "loss": 0.0013, "step": 59480 }, { "grad_norm": 0.0747155249118805, "learning_rate": 1.9829106043400826e-08, "loss": 0.0009, "step": 59490 }, { "grad_norm": 0.02034788206219673, "learning_rate": 1.9060658456043768e-08, "loss": 0.0008, "step": 59500 }, { "grad_norm": 0.013122273609042168, "learning_rate": 1.830739375838775e-08, "loss": 0.0005, "step": 59510 }, { "grad_norm": 0.015044772066175938, "learning_rate": 1.7569312179260832e-08, "loss": 0.0016, "step": 59520 }, { "grad_norm": 0.02920515649020672, "learning_rate": 1.684641394286146e-08, "loss": 0.0009, "step": 59530 }, { "grad_norm": 0.03250971436500549, "learning_rate": 1.6138699268797296e-08, "loss": 0.001, "step": 59540 }, { "grad_norm": 0.02710442617535591, "learning_rate": 1.5446168372046376e-08, "loss": 0.0008, "step": 59550 }, { "grad_norm": 0.026702459901571274, "learning_rate": 1.4768821462984861e-08, "loss": 0.0009, "step": 59560 }, { "grad_norm": 0.02944856882095337, "learning_rate": 1.4106658747370383e-08, "loss": 0.0009, "step": 59570 }, { "grad_norm": 0.021867232397198677, "learning_rate": 1.3459680426353149e-08, "loss": 0.0009, "step": 59580 }, { "grad_norm": 0.029177628457546234, "learning_rate": 1.2827886696464841e-08, "loss": 0.0007, "step": 59590 }, { "grad_norm": 0.02614719420671463, "learning_rate": 1.2211277749635264e-08, "loss": 0.0009, "step": 59600 }, { "grad_norm": 0.022199885919690132, "learning_rate": 1.1609853773164592e-08, "loss": 0.0009, "step": 59610 }, { "grad_norm": 0.0261878352612257, "learning_rate": 1.1023614949751127e-08, "loss": 0.001, "step": 59620 }, { "grad_norm": 0.03658847138285637, "learning_rate": 1.0452561457485744e-08, "loss": 0.0007, "step": 59630 }, { "grad_norm": 0.018664179369807243, "learning_rate": 9.896693469829688e-09, "loss": 0.0008, "step": 59640 }, { "grad_norm": 0.0535174235701561, "learning_rate": 9.35601115564788e-09, "loss": 0.0012, "step": 59650 }, { "grad_norm": 0.01234456617385149, "learning_rate": 8.830514679186719e-09, "loss": 0.0008, "step": 59660 }, { "grad_norm": 0.1072695404291153, "learning_rate": 8.320204200074066e-09, "loss": 0.001, "step": 59670 }, { "grad_norm": 0.021303335204720497, "learning_rate": 7.825079873324814e-09, "loss": 0.0008, "step": 59680 }, { "grad_norm": 0.02905043214559555, "learning_rate": 7.345141849351977e-09, "loss": 0.0007, "step": 59690 }, { "grad_norm": 0.023555628955364227, "learning_rate": 6.880390273944493e-09, "loss": 0.0007, "step": 59700 }, { "grad_norm": 0.02658013254404068, "learning_rate": 6.4308252882838704e-09, "loss": 0.0008, "step": 59710 }, { "grad_norm": 0.011233111843466759, "learning_rate": 5.9964470289386455e-09, "loss": 0.0006, "step": 59720 }, { "grad_norm": 0.010012122802436352, "learning_rate": 5.577255627853273e-09, "loss": 0.0008, "step": 59730 }, { "grad_norm": 0.041300397366285324, "learning_rate": 5.173251212370334e-09, "loss": 0.0009, "step": 59740 }, { "grad_norm": 0.03106926940381527, "learning_rate": 4.784433905219432e-09, "loss": 0.0006, "step": 59750 }, { "grad_norm": 0.02153533510863781, "learning_rate": 4.4108038245060934e-09, "loss": 0.0007, "step": 59760 }, { "grad_norm": 0.03810606151819229, "learning_rate": 4.05236108373952e-09, "loss": 0.0013, "step": 59770 }, { "grad_norm": 0.034351497888565063, "learning_rate": 3.7091057917937324e-09, "loss": 0.0006, "step": 59780 }, { "grad_norm": 0.023258820176124573, "learning_rate": 3.381038052946428e-09, "loss": 0.0009, "step": 59790 }, { "grad_norm": 0.014770102687180042, "learning_rate": 3.0681579668623283e-09, "loss": 0.0007, "step": 59800 }, { "grad_norm": 0.034172043204307556, "learning_rate": 2.7704656285709727e-09, "loss": 0.0009, "step": 59810 }, { "grad_norm": 0.029209543019533157, "learning_rate": 2.4879611285166803e-09, "loss": 0.0008, "step": 59820 }, { "grad_norm": 0.007940158247947693, "learning_rate": 2.2206445525085883e-09, "loss": 0.0012, "step": 59830 }, { "grad_norm": 0.04839251935482025, "learning_rate": 1.9685159817595112e-09, "loss": 0.0009, "step": 59840 }, { "grad_norm": 0.029985995963215828, "learning_rate": 1.7315754928470817e-09, "loss": 0.0005, "step": 59850 }, { "grad_norm": 0.020097434520721436, "learning_rate": 1.5098231577581611e-09, "loss": 0.0009, "step": 59860 }, { "grad_norm": 0.026428166776895523, "learning_rate": 1.3032590438499804e-09, "loss": 0.0009, "step": 59870 }, { "grad_norm": 0.018860362470149994, "learning_rate": 1.1118832138723444e-09, "loss": 0.0006, "step": 59880 }, { "grad_norm": 0.022543566301465034, "learning_rate": 9.356957259620825e-10, "loss": 0.0008, "step": 59890 }, { "grad_norm": 0.03766554966568947, "learning_rate": 7.74696633637495e-10, "loss": 0.001, "step": 59900 }, { "grad_norm": 0.015285993926227093, "learning_rate": 6.288859858039064e-10, "loss": 0.001, "step": 59910 }, { "grad_norm": 0.00859571248292923, "learning_rate": 4.982638267647665e-10, "loss": 0.0007, "step": 59920 }, { "grad_norm": 0.01552325114607811, "learning_rate": 3.8283019618834406e-10, "loss": 0.0007, "step": 59930 }, { "grad_norm": 0.0407278910279274, "learning_rate": 2.825851291410331e-10, "loss": 0.0007, "step": 59940 }, { "grad_norm": 0.024451980367302895, "learning_rate": 1.975286560873535e-10, "loss": 0.0008, "step": 59950 }, { "grad_norm": 0.017012884840369225, "learning_rate": 1.2766080285109284e-10, "loss": 0.0006, "step": 59960 }, { "grad_norm": 0.04203060641884804, "learning_rate": 7.298159065971533e-11, "loss": 0.0014, "step": 59970 }, { "grad_norm": 0.02116401307284832, "learning_rate": 3.349103612770854e-11, "loss": 0.0007, "step": 59980 }, { "grad_norm": 0.05615259334445, "learning_rate": 9.189151245481142e-12, "loss": 0.0009, "step": 59990 }, { "grad_norm": 0.030241385102272034, "learning_rate": 7.594340156735769e-14, "loss": 0.0011, "step": 60000 } ], "logging_steps": 10, "max_steps": 60000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 20000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }