{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 80000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 1.0454022884368896, "learning_rate": 2.25e-07, "loss": 1.49, "step": 10 }, { "grad_norm": 0.5293704867362976, "learning_rate": 4.75e-07, "loss": 1.4813, "step": 20 }, { "grad_norm": 1.0630000829696655, "learning_rate": 7.25e-07, "loss": 1.4877, "step": 30 }, { "grad_norm": 0.6214918494224548, "learning_rate": 9.75e-07, "loss": 1.4633, "step": 40 }, { "grad_norm": 0.7802557945251465, "learning_rate": 1.2250000000000001e-06, "loss": 1.4515, "step": 50 }, { "grad_norm": 1.2080212831497192, "learning_rate": 1.475e-06, "loss": 1.4101, "step": 60 }, { "grad_norm": 0.7624619603157043, "learning_rate": 1.7250000000000002e-06, "loss": 1.36, "step": 70 }, { "grad_norm": 1.63241446018219, "learning_rate": 1.975e-06, "loss": 1.3631, "step": 80 }, { "grad_norm": 1.4452098608016968, "learning_rate": 2.225e-06, "loss": 1.2984, "step": 90 }, { "grad_norm": 1.2461909055709839, "learning_rate": 2.4750000000000004e-06, "loss": 1.2621, "step": 100 }, { "grad_norm": 1.286273717880249, "learning_rate": 2.725e-06, "loss": 1.2334, "step": 110 }, { "grad_norm": 1.343520998954773, "learning_rate": 2.975e-06, "loss": 1.1913, "step": 120 }, { "grad_norm": 1.1987358331680298, "learning_rate": 3.225e-06, "loss": 1.1388, "step": 130 }, { "grad_norm": 1.377852439880371, "learning_rate": 3.4750000000000006e-06, "loss": 1.15, "step": 140 }, { "grad_norm": 1.3964060544967651, "learning_rate": 3.725e-06, "loss": 1.1302, "step": 150 }, { "grad_norm": 1.2511546611785889, "learning_rate": 3.975e-06, "loss": 1.1371, "step": 160 }, { "grad_norm": 1.2906032800674438, "learning_rate": 4.225e-06, "loss": 1.1014, "step": 170 }, { "grad_norm": 1.3629199266433716, "learning_rate": 4.475e-06, "loss": 1.1177, "step": 180 }, { "grad_norm": 1.3807164430618286, "learning_rate": 4.7250000000000005e-06, "loss": 1.0927, "step": 190 }, { "grad_norm": 1.0074986219406128, "learning_rate": 4.975000000000001e-06, "loss": 1.0764, "step": 200 }, { "grad_norm": 1.3763624429702759, "learning_rate": 5.225e-06, "loss": 1.0799, "step": 210 }, { "grad_norm": 1.5008262395858765, "learning_rate": 5.475e-06, "loss": 1.0715, "step": 220 }, { "grad_norm": 1.6926239728927612, "learning_rate": 5.725e-06, "loss": 1.0635, "step": 230 }, { "grad_norm": 1.4305245876312256, "learning_rate": 5.975e-06, "loss": 1.0679, "step": 240 }, { "grad_norm": 1.1284204721450806, "learning_rate": 6.2250000000000005e-06, "loss": 1.0604, "step": 250 }, { "grad_norm": 1.1936620473861694, "learning_rate": 6.475000000000001e-06, "loss": 1.038, "step": 260 }, { "grad_norm": 1.6276190280914307, "learning_rate": 6.725000000000001e-06, "loss": 1.065, "step": 270 }, { "grad_norm": 1.5137929916381836, "learning_rate": 6.975000000000001e-06, "loss": 1.0695, "step": 280 }, { "grad_norm": 1.250261902809143, "learning_rate": 7.2249999999999994e-06, "loss": 1.0517, "step": 290 }, { "grad_norm": 1.1109188795089722, "learning_rate": 7.4750000000000004e-06, "loss": 1.0472, "step": 300 }, { "grad_norm": 1.4856215715408325, "learning_rate": 7.725e-06, "loss": 1.0529, "step": 310 }, { "grad_norm": 1.1559423208236694, "learning_rate": 7.975e-06, "loss": 1.0391, "step": 320 }, { "grad_norm": 1.271992802619934, "learning_rate": 8.225e-06, "loss": 1.0358, "step": 330 }, { "grad_norm": 1.2036041021347046, "learning_rate": 8.475000000000001e-06, "loss": 1.0063, "step": 340 }, { "grad_norm": 1.6246967315673828, "learning_rate": 8.725e-06, "loss": 0.9755, "step": 350 }, { "grad_norm": 2.5427818298339844, "learning_rate": 8.975e-06, "loss": 0.9214, "step": 360 }, { "grad_norm": 2.01633620262146, "learning_rate": 9.225e-06, "loss": 0.8493, "step": 370 }, { "grad_norm": 2.0069971084594727, "learning_rate": 9.475e-06, "loss": 0.763, "step": 380 }, { "grad_norm": 2.571446418762207, "learning_rate": 9.725000000000001e-06, "loss": 0.6844, "step": 390 }, { "grad_norm": 2.0261216163635254, "learning_rate": 9.975e-06, "loss": 0.6326, "step": 400 }, { "grad_norm": 1.991512656211853, "learning_rate": 1.0225e-05, "loss": 0.5898, "step": 410 }, { "grad_norm": 2.4101078510284424, "learning_rate": 1.0475e-05, "loss": 0.5337, "step": 420 }, { "grad_norm": 2.4332730770111084, "learning_rate": 1.0725e-05, "loss": 0.5066, "step": 430 }, { "grad_norm": 3.3355984687805176, "learning_rate": 1.0975e-05, "loss": 0.452, "step": 440 }, { "grad_norm": 3.054561138153076, "learning_rate": 1.1225e-05, "loss": 0.4061, "step": 450 }, { "grad_norm": 2.601130247116089, "learning_rate": 1.1475000000000001e-05, "loss": 0.3641, "step": 460 }, { "grad_norm": 2.849320650100708, "learning_rate": 1.1725e-05, "loss": 0.3235, "step": 470 }, { "grad_norm": 1.8070234060287476, "learning_rate": 1.1975e-05, "loss": 0.2709, "step": 480 }, { "grad_norm": 2.1696255207061768, "learning_rate": 1.2225e-05, "loss": 0.2596, "step": 490 }, { "grad_norm": 2.159766912460327, "learning_rate": 1.2475e-05, "loss": 0.2416, "step": 500 }, { "grad_norm": 2.1663081645965576, "learning_rate": 1.2725000000000001e-05, "loss": 0.2371, "step": 510 }, { "grad_norm": 1.9580808877944946, "learning_rate": 1.2975e-05, "loss": 0.2088, "step": 520 }, { "grad_norm": 2.563446044921875, "learning_rate": 1.3225000000000001e-05, "loss": 0.2093, "step": 530 }, { "grad_norm": 2.1268203258514404, "learning_rate": 1.3475000000000002e-05, "loss": 0.1833, "step": 540 }, { "grad_norm": 1.7631834745407104, "learning_rate": 1.3725000000000002e-05, "loss": 0.1923, "step": 550 }, { "grad_norm": 2.1129491329193115, "learning_rate": 1.3975000000000003e-05, "loss": 0.1811, "step": 560 }, { "grad_norm": 2.7737998962402344, "learning_rate": 1.4225e-05, "loss": 0.1773, "step": 570 }, { "grad_norm": 2.7324728965759277, "learning_rate": 1.4475e-05, "loss": 0.1699, "step": 580 }, { "grad_norm": 3.004594326019287, "learning_rate": 1.4725e-05, "loss": 0.1683, "step": 590 }, { "grad_norm": 1.8583347797393799, "learning_rate": 1.4975e-05, "loss": 0.1652, "step": 600 }, { "grad_norm": 2.567894697189331, "learning_rate": 1.5225e-05, "loss": 0.1822, "step": 610 }, { "grad_norm": 2.1974761486053467, "learning_rate": 1.5475e-05, "loss": 0.1466, "step": 620 }, { "grad_norm": 1.8204607963562012, "learning_rate": 1.5725e-05, "loss": 0.1373, "step": 630 }, { "grad_norm": 2.392613649368286, "learning_rate": 1.5975000000000002e-05, "loss": 0.1513, "step": 640 }, { "grad_norm": 1.7030209302902222, "learning_rate": 1.6225e-05, "loss": 0.14, "step": 650 }, { "grad_norm": 1.7965730428695679, "learning_rate": 1.6475e-05, "loss": 0.1377, "step": 660 }, { "grad_norm": 2.000826120376587, "learning_rate": 1.6725000000000003e-05, "loss": 0.1654, "step": 670 }, { "grad_norm": 1.9867973327636719, "learning_rate": 1.6975000000000003e-05, "loss": 0.1375, "step": 680 }, { "grad_norm": 1.8465973138809204, "learning_rate": 1.7225e-05, "loss": 0.1347, "step": 690 }, { "grad_norm": 1.6523014307022095, "learning_rate": 1.7475e-05, "loss": 0.1277, "step": 700 }, { "grad_norm": 1.6305477619171143, "learning_rate": 1.7725e-05, "loss": 0.1442, "step": 710 }, { "grad_norm": 1.5859460830688477, "learning_rate": 1.7975e-05, "loss": 0.1255, "step": 720 }, { "grad_norm": 2.1109094619750977, "learning_rate": 1.8225e-05, "loss": 0.1248, "step": 730 }, { "grad_norm": 1.8730812072753906, "learning_rate": 1.8475000000000002e-05, "loss": 0.1008, "step": 740 }, { "grad_norm": 1.4432870149612427, "learning_rate": 1.8725e-05, "loss": 0.1223, "step": 750 }, { "grad_norm": 1.2769674062728882, "learning_rate": 1.8975e-05, "loss": 0.1105, "step": 760 }, { "grad_norm": 1.2808417081832886, "learning_rate": 1.9225e-05, "loss": 0.1101, "step": 770 }, { "grad_norm": 1.9566380977630615, "learning_rate": 1.9475000000000002e-05, "loss": 0.1161, "step": 780 }, { "grad_norm": 2.0479090213775635, "learning_rate": 1.9725000000000002e-05, "loss": 0.1057, "step": 790 }, { "grad_norm": 1.744713306427002, "learning_rate": 1.9975e-05, "loss": 0.1169, "step": 800 }, { "grad_norm": 1.7394174337387085, "learning_rate": 2.0225000000000004e-05, "loss": 0.1173, "step": 810 }, { "grad_norm": 1.6132285594940186, "learning_rate": 2.0475e-05, "loss": 0.1119, "step": 820 }, { "grad_norm": 1.6296381950378418, "learning_rate": 2.0725e-05, "loss": 0.1126, "step": 830 }, { "grad_norm": 1.9848552942276, "learning_rate": 2.0975e-05, "loss": 0.1063, "step": 840 }, { "grad_norm": 1.879822850227356, "learning_rate": 2.1225e-05, "loss": 0.1195, "step": 850 }, { "grad_norm": 1.7197335958480835, "learning_rate": 2.1475e-05, "loss": 0.1006, "step": 860 }, { "grad_norm": 1.541235327720642, "learning_rate": 2.1725e-05, "loss": 0.107, "step": 870 }, { "grad_norm": 1.6838984489440918, "learning_rate": 2.1975000000000002e-05, "loss": 0.1038, "step": 880 }, { "grad_norm": 1.5231255292892456, "learning_rate": 2.2225e-05, "loss": 0.0935, "step": 890 }, { "grad_norm": 1.4617599248886108, "learning_rate": 2.2475e-05, "loss": 0.1211, "step": 900 }, { "grad_norm": 2.8710057735443115, "learning_rate": 2.2725000000000003e-05, "loss": 0.1228, "step": 910 }, { "grad_norm": 1.6846849918365479, "learning_rate": 2.2975000000000003e-05, "loss": 0.1191, "step": 920 }, { "grad_norm": 1.8310812711715698, "learning_rate": 2.3225000000000002e-05, "loss": 0.1087, "step": 930 }, { "grad_norm": 1.5171942710876465, "learning_rate": 2.3475e-05, "loss": 0.0957, "step": 940 }, { "grad_norm": 1.4970916509628296, "learning_rate": 2.3725e-05, "loss": 0.1108, "step": 950 }, { "grad_norm": 2.5020132064819336, "learning_rate": 2.3975e-05, "loss": 0.0938, "step": 960 }, { "grad_norm": 2.019414186477661, "learning_rate": 2.4225e-05, "loss": 0.1005, "step": 970 }, { "grad_norm": 1.4795217514038086, "learning_rate": 2.4475000000000002e-05, "loss": 0.1091, "step": 980 }, { "grad_norm": 1.6160516738891602, "learning_rate": 2.4725e-05, "loss": 0.1165, "step": 990 }, { "grad_norm": 1.4791314601898193, "learning_rate": 2.4975e-05, "loss": 0.0958, "step": 1000 }, { "grad_norm": 1.3359657526016235, "learning_rate": 2.5225e-05, "loss": 0.0932, "step": 1010 }, { "grad_norm": 1.8482234477996826, "learning_rate": 2.5475e-05, "loss": 0.0917, "step": 1020 }, { "grad_norm": 1.4831860065460205, "learning_rate": 2.5725e-05, "loss": 0.0968, "step": 1030 }, { "grad_norm": 1.2481017112731934, "learning_rate": 2.5974999999999998e-05, "loss": 0.1047, "step": 1040 }, { "grad_norm": 1.5080252885818481, "learning_rate": 2.6225e-05, "loss": 0.0977, "step": 1050 }, { "grad_norm": 1.6915346384048462, "learning_rate": 2.6475e-05, "loss": 0.1111, "step": 1060 }, { "grad_norm": 1.1251699924468994, "learning_rate": 2.6725e-05, "loss": 0.0991, "step": 1070 }, { "grad_norm": 1.8935409784317017, "learning_rate": 2.6975000000000002e-05, "loss": 0.0976, "step": 1080 }, { "grad_norm": 1.0620390176773071, "learning_rate": 2.7225e-05, "loss": 0.08, "step": 1090 }, { "grad_norm": 1.7925835847854614, "learning_rate": 2.7475e-05, "loss": 0.0901, "step": 1100 }, { "grad_norm": 1.5456117391586304, "learning_rate": 2.7725e-05, "loss": 0.096, "step": 1110 }, { "grad_norm": 0.8869717121124268, "learning_rate": 2.7975000000000002e-05, "loss": 0.0851, "step": 1120 }, { "grad_norm": 1.3169152736663818, "learning_rate": 2.8225e-05, "loss": 0.0932, "step": 1130 }, { "grad_norm": 1.3773142099380493, "learning_rate": 2.8475e-05, "loss": 0.0925, "step": 1140 }, { "grad_norm": 1.4980037212371826, "learning_rate": 2.8725e-05, "loss": 0.0876, "step": 1150 }, { "grad_norm": 1.1834743022918701, "learning_rate": 2.8975000000000003e-05, "loss": 0.0911, "step": 1160 }, { "grad_norm": 2.2275588512420654, "learning_rate": 2.9225000000000002e-05, "loss": 0.087, "step": 1170 }, { "grad_norm": 1.207052230834961, "learning_rate": 2.9475e-05, "loss": 0.0952, "step": 1180 }, { "grad_norm": 1.4662206172943115, "learning_rate": 2.9725000000000004e-05, "loss": 0.0899, "step": 1190 }, { "grad_norm": 1.3991010189056396, "learning_rate": 2.9975000000000004e-05, "loss": 0.0944, "step": 1200 }, { "grad_norm": 1.2508325576782227, "learning_rate": 3.0225000000000003e-05, "loss": 0.0942, "step": 1210 }, { "grad_norm": 1.3787848949432373, "learning_rate": 3.0475000000000002e-05, "loss": 0.0822, "step": 1220 }, { "grad_norm": 1.4246500730514526, "learning_rate": 3.0725e-05, "loss": 0.0901, "step": 1230 }, { "grad_norm": 1.3118200302124023, "learning_rate": 3.0975e-05, "loss": 0.0834, "step": 1240 }, { "grad_norm": 1.1814906597137451, "learning_rate": 3.122500000000001e-05, "loss": 0.0952, "step": 1250 }, { "grad_norm": 1.3346397876739502, "learning_rate": 3.1475e-05, "loss": 0.0876, "step": 1260 }, { "grad_norm": 1.1218125820159912, "learning_rate": 3.1725e-05, "loss": 0.0902, "step": 1270 }, { "grad_norm": 1.476595401763916, "learning_rate": 3.1975e-05, "loss": 0.0903, "step": 1280 }, { "grad_norm": 1.5931370258331299, "learning_rate": 3.2225e-05, "loss": 0.093, "step": 1290 }, { "grad_norm": 1.311688780784607, "learning_rate": 3.2474999999999997e-05, "loss": 0.0932, "step": 1300 }, { "grad_norm": 0.870827853679657, "learning_rate": 3.2725e-05, "loss": 0.0861, "step": 1310 }, { "grad_norm": 1.7754698991775513, "learning_rate": 3.2975e-05, "loss": 0.0929, "step": 1320 }, { "grad_norm": 1.0749738216400146, "learning_rate": 3.3225e-05, "loss": 0.0778, "step": 1330 }, { "grad_norm": 1.3157168626785278, "learning_rate": 3.3475e-05, "loss": 0.0881, "step": 1340 }, { "grad_norm": 1.3468003273010254, "learning_rate": 3.3725e-05, "loss": 0.0768, "step": 1350 }, { "grad_norm": 1.1502091884613037, "learning_rate": 3.3975e-05, "loss": 0.0846, "step": 1360 }, { "grad_norm": 0.9342949986457825, "learning_rate": 3.4225e-05, "loss": 0.0777, "step": 1370 }, { "grad_norm": 1.1844979524612427, "learning_rate": 3.4475000000000005e-05, "loss": 0.0848, "step": 1380 }, { "grad_norm": 1.0852068662643433, "learning_rate": 3.4725000000000004e-05, "loss": 0.0717, "step": 1390 }, { "grad_norm": 1.184860110282898, "learning_rate": 3.4975e-05, "loss": 0.087, "step": 1400 }, { "grad_norm": 0.9572165012359619, "learning_rate": 3.5225e-05, "loss": 0.0745, "step": 1410 }, { "grad_norm": 1.168735384941101, "learning_rate": 3.5475e-05, "loss": 0.083, "step": 1420 }, { "grad_norm": 1.0272977352142334, "learning_rate": 3.5725e-05, "loss": 0.078, "step": 1430 }, { "grad_norm": 1.1205570697784424, "learning_rate": 3.5975e-05, "loss": 0.085, "step": 1440 }, { "grad_norm": 1.4445632696151733, "learning_rate": 3.6225000000000006e-05, "loss": 0.0847, "step": 1450 }, { "grad_norm": 0.9800775647163391, "learning_rate": 3.6475000000000006e-05, "loss": 0.0781, "step": 1460 }, { "grad_norm": 1.1699436902999878, "learning_rate": 3.6725000000000005e-05, "loss": 0.0746, "step": 1470 }, { "grad_norm": 1.5346860885620117, "learning_rate": 3.6975000000000004e-05, "loss": 0.0851, "step": 1480 }, { "grad_norm": 1.2641719579696655, "learning_rate": 3.7225000000000004e-05, "loss": 0.0789, "step": 1490 }, { "grad_norm": 0.9718949198722839, "learning_rate": 3.7475e-05, "loss": 0.0969, "step": 1500 }, { "grad_norm": 1.7564356327056885, "learning_rate": 3.7725e-05, "loss": 0.0805, "step": 1510 }, { "grad_norm": 0.8439446687698364, "learning_rate": 3.7975e-05, "loss": 0.0727, "step": 1520 }, { "grad_norm": 0.8165621757507324, "learning_rate": 3.8225e-05, "loss": 0.0777, "step": 1530 }, { "grad_norm": 1.73514723777771, "learning_rate": 3.8475e-05, "loss": 0.0752, "step": 1540 }, { "grad_norm": 0.9040842652320862, "learning_rate": 3.8725e-05, "loss": 0.0779, "step": 1550 }, { "grad_norm": 1.3095040321350098, "learning_rate": 3.8975e-05, "loss": 0.0807, "step": 1560 }, { "grad_norm": 0.9831851720809937, "learning_rate": 3.9225e-05, "loss": 0.0746, "step": 1570 }, { "grad_norm": 0.977741003036499, "learning_rate": 3.9475000000000004e-05, "loss": 0.0802, "step": 1580 }, { "grad_norm": 1.1530652046203613, "learning_rate": 3.9725e-05, "loss": 0.0827, "step": 1590 }, { "grad_norm": 0.990468442440033, "learning_rate": 3.9975e-05, "loss": 0.0791, "step": 1600 }, { "grad_norm": 1.0648311376571655, "learning_rate": 4.0225e-05, "loss": 0.0787, "step": 1610 }, { "grad_norm": 0.850773811340332, "learning_rate": 4.0475e-05, "loss": 0.0758, "step": 1620 }, { "grad_norm": 1.0055062770843506, "learning_rate": 4.0725e-05, "loss": 0.0718, "step": 1630 }, { "grad_norm": 1.28902006149292, "learning_rate": 4.0975e-05, "loss": 0.0787, "step": 1640 }, { "grad_norm": 1.013387680053711, "learning_rate": 4.1225e-05, "loss": 0.0908, "step": 1650 }, { "grad_norm": 0.7828097343444824, "learning_rate": 4.1475000000000005e-05, "loss": 0.0676, "step": 1660 }, { "grad_norm": 0.8467814326286316, "learning_rate": 4.1725000000000005e-05, "loss": 0.0827, "step": 1670 }, { "grad_norm": 0.8000909090042114, "learning_rate": 4.1975000000000004e-05, "loss": 0.0804, "step": 1680 }, { "grad_norm": 1.0793185234069824, "learning_rate": 4.2225e-05, "loss": 0.0803, "step": 1690 }, { "grad_norm": 1.1036640405654907, "learning_rate": 4.2475e-05, "loss": 0.0747, "step": 1700 }, { "grad_norm": 1.0465623140335083, "learning_rate": 4.2725e-05, "loss": 0.0759, "step": 1710 }, { "grad_norm": 1.1286619901657104, "learning_rate": 4.2975e-05, "loss": 0.0755, "step": 1720 }, { "grad_norm": 1.0894825458526611, "learning_rate": 4.322500000000001e-05, "loss": 0.0834, "step": 1730 }, { "grad_norm": 1.1862915754318237, "learning_rate": 4.3475000000000006e-05, "loss": 0.0755, "step": 1740 }, { "grad_norm": 0.9766193628311157, "learning_rate": 4.3725000000000006e-05, "loss": 0.0792, "step": 1750 }, { "grad_norm": 0.7859284281730652, "learning_rate": 4.3975e-05, "loss": 0.0832, "step": 1760 }, { "grad_norm": 0.8069933652877808, "learning_rate": 4.4225e-05, "loss": 0.0727, "step": 1770 }, { "grad_norm": 0.9457748532295227, "learning_rate": 4.4475e-05, "loss": 0.0719, "step": 1780 }, { "grad_norm": 0.8864979147911072, "learning_rate": 4.4725e-05, "loss": 0.0821, "step": 1790 }, { "grad_norm": 0.8066006302833557, "learning_rate": 4.4975e-05, "loss": 0.0725, "step": 1800 }, { "grad_norm": 1.238815188407898, "learning_rate": 4.5225e-05, "loss": 0.0745, "step": 1810 }, { "grad_norm": 0.8671537041664124, "learning_rate": 4.5475e-05, "loss": 0.0711, "step": 1820 }, { "grad_norm": 0.8506547212600708, "learning_rate": 4.5725e-05, "loss": 0.0601, "step": 1830 }, { "grad_norm": 1.0611209869384766, "learning_rate": 4.5975e-05, "loss": 0.0719, "step": 1840 }, { "grad_norm": 1.2312718629837036, "learning_rate": 4.6225e-05, "loss": 0.0736, "step": 1850 }, { "grad_norm": 0.7747061848640442, "learning_rate": 4.6475000000000005e-05, "loss": 0.0732, "step": 1860 }, { "grad_norm": 0.7648390531539917, "learning_rate": 4.6725000000000004e-05, "loss": 0.0836, "step": 1870 }, { "grad_norm": 0.9208672642707825, "learning_rate": 4.6975000000000003e-05, "loss": 0.0659, "step": 1880 }, { "grad_norm": 0.8406191468238831, "learning_rate": 4.7225e-05, "loss": 0.0658, "step": 1890 }, { "grad_norm": 1.1677839756011963, "learning_rate": 4.7475e-05, "loss": 0.0823, "step": 1900 }, { "grad_norm": 0.7670629620552063, "learning_rate": 4.7725e-05, "loss": 0.0725, "step": 1910 }, { "grad_norm": 0.8392388820648193, "learning_rate": 4.7975e-05, "loss": 0.0716, "step": 1920 }, { "grad_norm": 0.9049583077430725, "learning_rate": 4.822500000000001e-05, "loss": 0.0782, "step": 1930 }, { "grad_norm": 0.8056450486183167, "learning_rate": 4.8475000000000006e-05, "loss": 0.0706, "step": 1940 }, { "grad_norm": 0.8772521615028381, "learning_rate": 4.8725000000000005e-05, "loss": 0.0793, "step": 1950 }, { "grad_norm": 0.8424368500709534, "learning_rate": 4.8975000000000005e-05, "loss": 0.0647, "step": 1960 }, { "grad_norm": 0.7635156512260437, "learning_rate": 4.9225000000000004e-05, "loss": 0.0769, "step": 1970 }, { "grad_norm": 0.8873920440673828, "learning_rate": 4.9475e-05, "loss": 0.08, "step": 1980 }, { "grad_norm": 1.0546540021896362, "learning_rate": 4.9725e-05, "loss": 0.0662, "step": 1990 }, { "grad_norm": 0.6179728507995605, "learning_rate": 4.9975e-05, "loss": 0.0657, "step": 2000 }, { "grad_norm": 0.7596228122711182, "learning_rate": 5.0225e-05, "loss": 0.0577, "step": 2010 }, { "grad_norm": 0.5574071407318115, "learning_rate": 5.047500000000001e-05, "loss": 0.0672, "step": 2020 }, { "grad_norm": 0.9461479783058167, "learning_rate": 5.0725e-05, "loss": 0.0687, "step": 2030 }, { "grad_norm": 0.7017711997032166, "learning_rate": 5.0975000000000006e-05, "loss": 0.0807, "step": 2040 }, { "grad_norm": 0.8145959377288818, "learning_rate": 5.1225e-05, "loss": 0.0663, "step": 2050 }, { "grad_norm": 0.6763783097267151, "learning_rate": 5.1475000000000004e-05, "loss": 0.0577, "step": 2060 }, { "grad_norm": 0.8021596074104309, "learning_rate": 5.1725000000000004e-05, "loss": 0.0662, "step": 2070 }, { "grad_norm": 0.6172182559967041, "learning_rate": 5.197500000000001e-05, "loss": 0.0678, "step": 2080 }, { "grad_norm": 0.8727259039878845, "learning_rate": 5.2225e-05, "loss": 0.0624, "step": 2090 }, { "grad_norm": 0.9176621437072754, "learning_rate": 5.247500000000001e-05, "loss": 0.0633, "step": 2100 }, { "grad_norm": 0.8655284643173218, "learning_rate": 5.2725e-05, "loss": 0.0649, "step": 2110 }, { "grad_norm": 0.8628140091896057, "learning_rate": 5.297500000000001e-05, "loss": 0.0612, "step": 2120 }, { "grad_norm": 0.8523488640785217, "learning_rate": 5.3225e-05, "loss": 0.0629, "step": 2130 }, { "grad_norm": 0.6831222176551819, "learning_rate": 5.3475e-05, "loss": 0.0733, "step": 2140 }, { "grad_norm": 1.1561083793640137, "learning_rate": 5.3725000000000005e-05, "loss": 0.0741, "step": 2150 }, { "grad_norm": 0.8757587671279907, "learning_rate": 5.3975e-05, "loss": 0.062, "step": 2160 }, { "grad_norm": 0.6811245679855347, "learning_rate": 5.4225000000000003e-05, "loss": 0.0592, "step": 2170 }, { "grad_norm": 1.138044834136963, "learning_rate": 5.4474999999999996e-05, "loss": 0.0752, "step": 2180 }, { "grad_norm": 0.7270435094833374, "learning_rate": 5.4725e-05, "loss": 0.0631, "step": 2190 }, { "grad_norm": 0.7265561819076538, "learning_rate": 5.4975e-05, "loss": 0.069, "step": 2200 }, { "grad_norm": 1.0902704000473022, "learning_rate": 5.522500000000001e-05, "loss": 0.0644, "step": 2210 }, { "grad_norm": 0.981400728225708, "learning_rate": 5.5475e-05, "loss": 0.0611, "step": 2220 }, { "grad_norm": 0.8785156011581421, "learning_rate": 5.5725000000000006e-05, "loss": 0.0666, "step": 2230 }, { "grad_norm": 0.8795560002326965, "learning_rate": 5.5975e-05, "loss": 0.0692, "step": 2240 }, { "grad_norm": 0.9034879207611084, "learning_rate": 5.6225000000000005e-05, "loss": 0.0741, "step": 2250 }, { "grad_norm": 1.022779941558838, "learning_rate": 5.6475e-05, "loss": 0.0692, "step": 2260 }, { "grad_norm": 0.6120957136154175, "learning_rate": 5.6725e-05, "loss": 0.068, "step": 2270 }, { "grad_norm": 0.5666636824607849, "learning_rate": 5.6975e-05, "loss": 0.0696, "step": 2280 }, { "grad_norm": 0.5946398973464966, "learning_rate": 5.722500000000001e-05, "loss": 0.0701, "step": 2290 }, { "grad_norm": 1.5492078065872192, "learning_rate": 5.7475e-05, "loss": 0.0578, "step": 2300 }, { "grad_norm": 0.8053454756736755, "learning_rate": 5.772500000000001e-05, "loss": 0.072, "step": 2310 }, { "grad_norm": 0.9543825387954712, "learning_rate": 5.7975e-05, "loss": 0.0649, "step": 2320 }, { "grad_norm": 0.6245023012161255, "learning_rate": 5.8225000000000006e-05, "loss": 0.0702, "step": 2330 }, { "grad_norm": 0.7600482106208801, "learning_rate": 5.8475000000000005e-05, "loss": 0.0603, "step": 2340 }, { "grad_norm": 0.871504008769989, "learning_rate": 5.8725000000000004e-05, "loss": 0.0673, "step": 2350 }, { "grad_norm": 0.6561927795410156, "learning_rate": 5.8975000000000004e-05, "loss": 0.058, "step": 2360 }, { "grad_norm": 0.7128649353981018, "learning_rate": 5.922500000000001e-05, "loss": 0.0576, "step": 2370 }, { "grad_norm": 0.8923407793045044, "learning_rate": 5.9475e-05, "loss": 0.0532, "step": 2380 }, { "grad_norm": 0.6691806316375732, "learning_rate": 5.9724999999999995e-05, "loss": 0.0594, "step": 2390 }, { "grad_norm": 0.7226727604866028, "learning_rate": 5.9975e-05, "loss": 0.0713, "step": 2400 }, { "grad_norm": 0.662397563457489, "learning_rate": 6.0225e-05, "loss": 0.0555, "step": 2410 }, { "grad_norm": 0.8201700448989868, "learning_rate": 6.0475000000000006e-05, "loss": 0.0573, "step": 2420 }, { "grad_norm": 0.6086867451667786, "learning_rate": 6.0725e-05, "loss": 0.0586, "step": 2430 }, { "grad_norm": 0.6775134205818176, "learning_rate": 6.0975000000000005e-05, "loss": 0.0609, "step": 2440 }, { "grad_norm": 0.510739803314209, "learning_rate": 6.1225e-05, "loss": 0.0547, "step": 2450 }, { "grad_norm": 1.4645036458969116, "learning_rate": 6.1475e-05, "loss": 0.0655, "step": 2460 }, { "grad_norm": 0.5904291868209839, "learning_rate": 6.1725e-05, "loss": 0.0641, "step": 2470 }, { "grad_norm": 0.9463523626327515, "learning_rate": 6.1975e-05, "loss": 0.0614, "step": 2480 }, { "grad_norm": 0.7405305504798889, "learning_rate": 6.2225e-05, "loss": 0.0613, "step": 2490 }, { "grad_norm": 0.8312345743179321, "learning_rate": 6.2475e-05, "loss": 0.0696, "step": 2500 }, { "grad_norm": 0.7351598739624023, "learning_rate": 6.2725e-05, "loss": 0.0649, "step": 2510 }, { "grad_norm": 0.6981844305992126, "learning_rate": 6.297500000000001e-05, "loss": 0.0747, "step": 2520 }, { "grad_norm": 0.8756694793701172, "learning_rate": 6.3225e-05, "loss": 0.0648, "step": 2530 }, { "grad_norm": 0.6754956245422363, "learning_rate": 6.347500000000001e-05, "loss": 0.0694, "step": 2540 }, { "grad_norm": 0.8116153478622437, "learning_rate": 6.3725e-05, "loss": 0.0778, "step": 2550 }, { "grad_norm": 0.6851396560668945, "learning_rate": 6.397500000000001e-05, "loss": 0.0693, "step": 2560 }, { "grad_norm": 0.8753108382225037, "learning_rate": 6.4225e-05, "loss": 0.0629, "step": 2570 }, { "grad_norm": 0.7132512927055359, "learning_rate": 6.447500000000001e-05, "loss": 0.0704, "step": 2580 }, { "grad_norm": 0.6241723299026489, "learning_rate": 6.4725e-05, "loss": 0.0612, "step": 2590 }, { "grad_norm": 0.7012420892715454, "learning_rate": 6.497500000000001e-05, "loss": 0.0662, "step": 2600 }, { "grad_norm": 0.5857554078102112, "learning_rate": 6.5225e-05, "loss": 0.059, "step": 2610 }, { "grad_norm": 0.49916931986808777, "learning_rate": 6.5475e-05, "loss": 0.0612, "step": 2620 }, { "grad_norm": 0.6326761245727539, "learning_rate": 6.5725e-05, "loss": 0.0598, "step": 2630 }, { "grad_norm": 0.7376649379730225, "learning_rate": 6.5975e-05, "loss": 0.0637, "step": 2640 }, { "grad_norm": 0.5665512084960938, "learning_rate": 6.6225e-05, "loss": 0.0632, "step": 2650 }, { "grad_norm": 0.45966506004333496, "learning_rate": 6.6475e-05, "loss": 0.0662, "step": 2660 }, { "grad_norm": 0.7144515514373779, "learning_rate": 6.672500000000001e-05, "loss": 0.0638, "step": 2670 }, { "grad_norm": 0.9156882166862488, "learning_rate": 6.6975e-05, "loss": 0.0667, "step": 2680 }, { "grad_norm": 0.6789060831069946, "learning_rate": 6.722500000000001e-05, "loss": 0.0582, "step": 2690 }, { "grad_norm": 0.7964689135551453, "learning_rate": 6.7475e-05, "loss": 0.0658, "step": 2700 }, { "grad_norm": 0.63087397813797, "learning_rate": 6.7725e-05, "loss": 0.0635, "step": 2710 }, { "grad_norm": 0.7177891135215759, "learning_rate": 6.7975e-05, "loss": 0.0634, "step": 2720 }, { "grad_norm": 0.7916864156723022, "learning_rate": 6.8225e-05, "loss": 0.0633, "step": 2730 }, { "grad_norm": 0.7287747263908386, "learning_rate": 6.8475e-05, "loss": 0.0589, "step": 2740 }, { "grad_norm": 0.6120500564575195, "learning_rate": 6.8725e-05, "loss": 0.0503, "step": 2750 }, { "grad_norm": 0.7189498543739319, "learning_rate": 6.8975e-05, "loss": 0.056, "step": 2760 }, { "grad_norm": 0.7007486820220947, "learning_rate": 6.9225e-05, "loss": 0.0565, "step": 2770 }, { "grad_norm": 0.7845317721366882, "learning_rate": 6.9475e-05, "loss": 0.0608, "step": 2780 }, { "grad_norm": 0.8940625190734863, "learning_rate": 6.9725e-05, "loss": 0.0599, "step": 2790 }, { "grad_norm": 0.7644454836845398, "learning_rate": 6.997500000000001e-05, "loss": 0.0601, "step": 2800 }, { "grad_norm": 0.6220541000366211, "learning_rate": 7.022500000000001e-05, "loss": 0.0668, "step": 2810 }, { "grad_norm": 0.6295808553695679, "learning_rate": 7.0475e-05, "loss": 0.0613, "step": 2820 }, { "grad_norm": 0.5783026814460754, "learning_rate": 7.072500000000001e-05, "loss": 0.0594, "step": 2830 }, { "grad_norm": 0.5059423446655273, "learning_rate": 7.0975e-05, "loss": 0.0562, "step": 2840 }, { "grad_norm": 0.9372662305831909, "learning_rate": 7.122500000000001e-05, "loss": 0.0618, "step": 2850 }, { "grad_norm": 0.6058995723724365, "learning_rate": 7.1475e-05, "loss": 0.0539, "step": 2860 }, { "grad_norm": 0.578026294708252, "learning_rate": 7.172500000000001e-05, "loss": 0.0514, "step": 2870 }, { "grad_norm": 0.4643220901489258, "learning_rate": 7.1975e-05, "loss": 0.0509, "step": 2880 }, { "grad_norm": 0.48842570185661316, "learning_rate": 7.2225e-05, "loss": 0.0483, "step": 2890 }, { "grad_norm": 0.48140162229537964, "learning_rate": 7.2475e-05, "loss": 0.055, "step": 2900 }, { "grad_norm": 0.5442922115325928, "learning_rate": 7.272499999999999e-05, "loss": 0.0685, "step": 2910 }, { "grad_norm": 0.5714932680130005, "learning_rate": 7.2975e-05, "loss": 0.0535, "step": 2920 }, { "grad_norm": 0.4898555874824524, "learning_rate": 7.3225e-05, "loss": 0.0484, "step": 2930 }, { "grad_norm": 0.5388959646224976, "learning_rate": 7.347500000000001e-05, "loss": 0.0489, "step": 2940 }, { "grad_norm": 0.6738954186439514, "learning_rate": 7.3725e-05, "loss": 0.0449, "step": 2950 }, { "grad_norm": 0.6073933243751526, "learning_rate": 7.397500000000001e-05, "loss": 0.06, "step": 2960 }, { "grad_norm": 0.491477370262146, "learning_rate": 7.4225e-05, "loss": 0.0608, "step": 2970 }, { "grad_norm": 0.6820594668388367, "learning_rate": 7.447500000000001e-05, "loss": 0.0541, "step": 2980 }, { "grad_norm": 0.8678102493286133, "learning_rate": 7.4725e-05, "loss": 0.0584, "step": 2990 }, { "grad_norm": 0.6566895246505737, "learning_rate": 7.4975e-05, "loss": 0.0592, "step": 3000 }, { "grad_norm": 0.7871794104576111, "learning_rate": 7.5225e-05, "loss": 0.0567, "step": 3010 }, { "grad_norm": 0.902736246585846, "learning_rate": 7.5475e-05, "loss": 0.0659, "step": 3020 }, { "grad_norm": 0.6606013178825378, "learning_rate": 7.5725e-05, "loss": 0.0601, "step": 3030 }, { "grad_norm": 0.5749215483665466, "learning_rate": 7.5975e-05, "loss": 0.0539, "step": 3040 }, { "grad_norm": 0.7995256185531616, "learning_rate": 7.6225e-05, "loss": 0.0597, "step": 3050 }, { "grad_norm": 0.6864184141159058, "learning_rate": 7.6475e-05, "loss": 0.0589, "step": 3060 }, { "grad_norm": 0.7562260627746582, "learning_rate": 7.672500000000001e-05, "loss": 0.0581, "step": 3070 }, { "grad_norm": 0.6359729170799255, "learning_rate": 7.697500000000001e-05, "loss": 0.054, "step": 3080 }, { "grad_norm": 0.5408867597579956, "learning_rate": 7.722500000000001e-05, "loss": 0.0525, "step": 3090 }, { "grad_norm": 0.8332677483558655, "learning_rate": 7.747500000000001e-05, "loss": 0.0524, "step": 3100 }, { "grad_norm": 0.5660290718078613, "learning_rate": 7.7725e-05, "loss": 0.0607, "step": 3110 }, { "grad_norm": 0.6253294348716736, "learning_rate": 7.797500000000001e-05, "loss": 0.0513, "step": 3120 }, { "grad_norm": 0.5392642617225647, "learning_rate": 7.8225e-05, "loss": 0.0593, "step": 3130 }, { "grad_norm": 0.5520174503326416, "learning_rate": 7.8475e-05, "loss": 0.0526, "step": 3140 }, { "grad_norm": 0.6274023056030273, "learning_rate": 7.8725e-05, "loss": 0.0523, "step": 3150 }, { "grad_norm": 0.6960418224334717, "learning_rate": 7.8975e-05, "loss": 0.0536, "step": 3160 }, { "grad_norm": 0.6309225559234619, "learning_rate": 7.9225e-05, "loss": 0.0573, "step": 3170 }, { "grad_norm": 0.5085383653640747, "learning_rate": 7.9475e-05, "loss": 0.062, "step": 3180 }, { "grad_norm": 0.49904394149780273, "learning_rate": 7.9725e-05, "loss": 0.0621, "step": 3190 }, { "grad_norm": 0.6936240792274475, "learning_rate": 7.9975e-05, "loss": 0.0567, "step": 3200 }, { "grad_norm": 0.6120238900184631, "learning_rate": 8.022500000000001e-05, "loss": 0.0509, "step": 3210 }, { "grad_norm": 0.5574119687080383, "learning_rate": 8.0475e-05, "loss": 0.0487, "step": 3220 }, { "grad_norm": 0.6512274146080017, "learning_rate": 8.072500000000001e-05, "loss": 0.051, "step": 3230 }, { "grad_norm": 0.5463530421257019, "learning_rate": 8.0975e-05, "loss": 0.0504, "step": 3240 }, { "grad_norm": 0.7753533720970154, "learning_rate": 8.122500000000001e-05, "loss": 0.0518, "step": 3250 }, { "grad_norm": 0.6174612045288086, "learning_rate": 8.1475e-05, "loss": 0.0629, "step": 3260 }, { "grad_norm": 0.4632275700569153, "learning_rate": 8.172500000000001e-05, "loss": 0.053, "step": 3270 }, { "grad_norm": 0.6308867335319519, "learning_rate": 8.1975e-05, "loss": 0.0526, "step": 3280 }, { "grad_norm": 0.4319891631603241, "learning_rate": 8.2225e-05, "loss": 0.056, "step": 3290 }, { "grad_norm": 0.6512047648429871, "learning_rate": 8.2475e-05, "loss": 0.0572, "step": 3300 }, { "grad_norm": 0.5488726496696472, "learning_rate": 8.2725e-05, "loss": 0.0595, "step": 3310 }, { "grad_norm": 0.6080116629600525, "learning_rate": 8.2975e-05, "loss": 0.0539, "step": 3320 }, { "grad_norm": 0.7783663868904114, "learning_rate": 8.3225e-05, "loss": 0.0535, "step": 3330 }, { "grad_norm": 0.5180625915527344, "learning_rate": 8.347500000000001e-05, "loss": 0.0521, "step": 3340 }, { "grad_norm": 0.6729445457458496, "learning_rate": 8.3725e-05, "loss": 0.0579, "step": 3350 }, { "grad_norm": 0.6281179189682007, "learning_rate": 8.397500000000001e-05, "loss": 0.0594, "step": 3360 }, { "grad_norm": 0.5863791108131409, "learning_rate": 8.422500000000001e-05, "loss": 0.0672, "step": 3370 }, { "grad_norm": 0.6020815968513489, "learning_rate": 8.447500000000001e-05, "loss": 0.0555, "step": 3380 }, { "grad_norm": 0.6027818918228149, "learning_rate": 8.4725e-05, "loss": 0.0615, "step": 3390 }, { "grad_norm": 0.6394437551498413, "learning_rate": 8.4975e-05, "loss": 0.0654, "step": 3400 }, { "grad_norm": 0.5336827039718628, "learning_rate": 8.5225e-05, "loss": 0.0565, "step": 3410 }, { "grad_norm": 0.5645209550857544, "learning_rate": 8.5475e-05, "loss": 0.0508, "step": 3420 }, { "grad_norm": 0.5860723853111267, "learning_rate": 8.5725e-05, "loss": 0.0478, "step": 3430 }, { "grad_norm": 0.5780758857727051, "learning_rate": 8.5975e-05, "loss": 0.0539, "step": 3440 }, { "grad_norm": 0.4362058937549591, "learning_rate": 8.6225e-05, "loss": 0.0448, "step": 3450 }, { "grad_norm": 0.6643604636192322, "learning_rate": 8.6475e-05, "loss": 0.0547, "step": 3460 }, { "grad_norm": 0.605261504650116, "learning_rate": 8.672500000000001e-05, "loss": 0.0559, "step": 3470 }, { "grad_norm": 0.6710433959960938, "learning_rate": 8.6975e-05, "loss": 0.058, "step": 3480 }, { "grad_norm": 0.4203641712665558, "learning_rate": 8.7225e-05, "loss": 0.049, "step": 3490 }, { "grad_norm": 0.5524789690971375, "learning_rate": 8.747500000000001e-05, "loss": 0.0637, "step": 3500 }, { "grad_norm": 0.49877336621284485, "learning_rate": 8.7725e-05, "loss": 0.0629, "step": 3510 }, { "grad_norm": 0.5321910381317139, "learning_rate": 8.797500000000001e-05, "loss": 0.0532, "step": 3520 }, { "grad_norm": 0.6420599818229675, "learning_rate": 8.8225e-05, "loss": 0.0548, "step": 3530 }, { "grad_norm": 0.5223504304885864, "learning_rate": 8.847500000000001e-05, "loss": 0.0612, "step": 3540 }, { "grad_norm": 0.6315040588378906, "learning_rate": 8.8725e-05, "loss": 0.0559, "step": 3550 }, { "grad_norm": 0.7180589437484741, "learning_rate": 8.897500000000001e-05, "loss": 0.0587, "step": 3560 }, { "grad_norm": 0.6227295994758606, "learning_rate": 8.9225e-05, "loss": 0.0583, "step": 3570 }, { "grad_norm": 0.5337017178535461, "learning_rate": 8.9475e-05, "loss": 0.0557, "step": 3580 }, { "grad_norm": 0.5040066838264465, "learning_rate": 8.9725e-05, "loss": 0.0501, "step": 3590 }, { "grad_norm": 0.47838112711906433, "learning_rate": 8.9975e-05, "loss": 0.0558, "step": 3600 }, { "grad_norm": 0.8950636386871338, "learning_rate": 9.0225e-05, "loss": 0.0483, "step": 3610 }, { "grad_norm": 0.4878252148628235, "learning_rate": 9.0475e-05, "loss": 0.0538, "step": 3620 }, { "grad_norm": 0.6371610164642334, "learning_rate": 9.072500000000001e-05, "loss": 0.055, "step": 3630 }, { "grad_norm": 0.47139838337898254, "learning_rate": 9.0975e-05, "loss": 0.0609, "step": 3640 }, { "grad_norm": 0.543911874294281, "learning_rate": 9.122500000000001e-05, "loss": 0.0599, "step": 3650 }, { "grad_norm": 0.468002587556839, "learning_rate": 9.1475e-05, "loss": 0.0603, "step": 3660 }, { "grad_norm": 0.5300518870353699, "learning_rate": 9.172500000000001e-05, "loss": 0.0576, "step": 3670 }, { "grad_norm": 0.5648208856582642, "learning_rate": 9.1975e-05, "loss": 0.0623, "step": 3680 }, { "grad_norm": 0.5496950745582581, "learning_rate": 9.2225e-05, "loss": 0.0685, "step": 3690 }, { "grad_norm": 0.5805874466896057, "learning_rate": 9.2475e-05, "loss": 0.0661, "step": 3700 }, { "grad_norm": 0.647465705871582, "learning_rate": 9.2725e-05, "loss": 0.0701, "step": 3710 }, { "grad_norm": 0.6350710988044739, "learning_rate": 9.2975e-05, "loss": 0.0745, "step": 3720 }, { "grad_norm": 0.740679144859314, "learning_rate": 9.3225e-05, "loss": 0.06, "step": 3730 }, { "grad_norm": 0.8047658801078796, "learning_rate": 9.3475e-05, "loss": 0.072, "step": 3740 }, { "grad_norm": 0.8473596572875977, "learning_rate": 9.3725e-05, "loss": 0.0661, "step": 3750 }, { "grad_norm": 0.4504777193069458, "learning_rate": 9.397500000000001e-05, "loss": 0.0581, "step": 3760 }, { "grad_norm": 0.6258001327514648, "learning_rate": 9.422500000000001e-05, "loss": 0.0659, "step": 3770 }, { "grad_norm": 0.5916931629180908, "learning_rate": 9.4475e-05, "loss": 0.0662, "step": 3780 }, { "grad_norm": 0.505774199962616, "learning_rate": 9.472500000000001e-05, "loss": 0.0513, "step": 3790 }, { "grad_norm": 0.5445688962936401, "learning_rate": 9.4975e-05, "loss": 0.0549, "step": 3800 }, { "grad_norm": 0.6559740304946899, "learning_rate": 9.522500000000001e-05, "loss": 0.0521, "step": 3810 }, { "grad_norm": 0.6327717304229736, "learning_rate": 9.5475e-05, "loss": 0.0608, "step": 3820 }, { "grad_norm": 0.43546006083488464, "learning_rate": 9.572500000000001e-05, "loss": 0.0593, "step": 3830 }, { "grad_norm": 0.38084667921066284, "learning_rate": 9.5975e-05, "loss": 0.0516, "step": 3840 }, { "grad_norm": 0.7762323021888733, "learning_rate": 9.622500000000001e-05, "loss": 0.0523, "step": 3850 }, { "grad_norm": 0.8990876078605652, "learning_rate": 9.6475e-05, "loss": 0.0589, "step": 3860 }, { "grad_norm": 0.5860038995742798, "learning_rate": 9.6725e-05, "loss": 0.0656, "step": 3870 }, { "grad_norm": 0.5412526726722717, "learning_rate": 9.6975e-05, "loss": 0.0608, "step": 3880 }, { "grad_norm": 0.6437020301818848, "learning_rate": 9.7225e-05, "loss": 0.0543, "step": 3890 }, { "grad_norm": 0.5436424016952515, "learning_rate": 9.747500000000001e-05, "loss": 0.0633, "step": 3900 }, { "grad_norm": 0.6487393379211426, "learning_rate": 9.7725e-05, "loss": 0.0571, "step": 3910 }, { "grad_norm": 0.4998205006122589, "learning_rate": 9.797500000000001e-05, "loss": 0.0526, "step": 3920 }, { "grad_norm": 0.5062453150749207, "learning_rate": 9.8225e-05, "loss": 0.0577, "step": 3930 }, { "grad_norm": 0.4613572657108307, "learning_rate": 9.847500000000001e-05, "loss": 0.0555, "step": 3940 }, { "grad_norm": 0.48340198397636414, "learning_rate": 9.8725e-05, "loss": 0.0548, "step": 3950 }, { "grad_norm": 0.4401691257953644, "learning_rate": 9.897500000000001e-05, "loss": 0.0505, "step": 3960 }, { "grad_norm": 0.4968670904636383, "learning_rate": 9.9225e-05, "loss": 0.0584, "step": 3970 }, { "grad_norm": 0.54523766040802, "learning_rate": 9.9475e-05, "loss": 0.0484, "step": 3980 }, { "grad_norm": 0.4423448145389557, "learning_rate": 9.9725e-05, "loss": 0.0516, "step": 3990 }, { "grad_norm": 0.52967369556427, "learning_rate": 9.9975e-05, "loss": 0.0597, "step": 4000 }, { "grad_norm": 0.45253342390060425, "learning_rate": 9.999999653982884e-05, "loss": 0.0557, "step": 4010 }, { "grad_norm": 0.7881225347518921, "learning_rate": 9.999998457874392e-05, "loss": 0.0565, "step": 4020 }, { "grad_norm": 0.7265608310699463, "learning_rate": 9.999996407402913e-05, "loss": 0.0632, "step": 4030 }, { "grad_norm": 0.5752301812171936, "learning_rate": 9.999993502568801e-05, "loss": 0.0617, "step": 4040 }, { "grad_norm": 0.4884452819824219, "learning_rate": 9.999989743372548e-05, "loss": 0.0494, "step": 4050 }, { "grad_norm": 0.5648279190063477, "learning_rate": 9.999985129814798e-05, "loss": 0.058, "step": 4060 }, { "grad_norm": 0.38567984104156494, "learning_rate": 9.99997966189634e-05, "loss": 0.0533, "step": 4070 }, { "grad_norm": 0.4979497194290161, "learning_rate": 9.999973339618107e-05, "loss": 0.0532, "step": 4080 }, { "grad_norm": 0.6147907376289368, "learning_rate": 9.999966162981179e-05, "loss": 0.0584, "step": 4090 }, { "grad_norm": 0.4407958388328552, "learning_rate": 9.999958131986784e-05, "loss": 0.0591, "step": 4100 }, { "grad_norm": 0.5341190695762634, "learning_rate": 9.999949246636293e-05, "loss": 0.06, "step": 4110 }, { "grad_norm": 0.5020150542259216, "learning_rate": 9.999939506931224e-05, "loss": 0.0546, "step": 4120 }, { "grad_norm": 0.5923051834106445, "learning_rate": 9.999928912873243e-05, "loss": 0.0557, "step": 4130 }, { "grad_norm": 0.708077609539032, "learning_rate": 9.999917464464159e-05, "loss": 0.0534, "step": 4140 }, { "grad_norm": 0.5489857196807861, "learning_rate": 9.999905161705929e-05, "loss": 0.0548, "step": 4150 }, { "grad_norm": 0.447113960981369, "learning_rate": 9.999892004600653e-05, "loss": 0.0502, "step": 4160 }, { "grad_norm": 0.4188302457332611, "learning_rate": 9.999877993150581e-05, "loss": 0.0569, "step": 4170 }, { "grad_norm": 0.4106309711933136, "learning_rate": 9.999863127358108e-05, "loss": 0.0525, "step": 4180 }, { "grad_norm": 0.619684636592865, "learning_rate": 9.999847407225773e-05, "loss": 0.0589, "step": 4190 }, { "grad_norm": 0.43691667914390564, "learning_rate": 9.999830832756262e-05, "loss": 0.0504, "step": 4200 }, { "grad_norm": 0.4357869327068329, "learning_rate": 9.999813403952407e-05, "loss": 0.044, "step": 4210 }, { "grad_norm": 0.7126026749610901, "learning_rate": 9.999795120817187e-05, "loss": 0.0474, "step": 4220 }, { "grad_norm": 0.30909568071365356, "learning_rate": 9.999775983353725e-05, "loss": 0.048, "step": 4230 }, { "grad_norm": 0.46768003702163696, "learning_rate": 9.999755991565292e-05, "loss": 0.0511, "step": 4240 }, { "grad_norm": 0.48844611644744873, "learning_rate": 9.999735145455303e-05, "loss": 0.054, "step": 4250 }, { "grad_norm": 0.4724677503108978, "learning_rate": 9.99971344502732e-05, "loss": 0.0569, "step": 4260 }, { "grad_norm": 0.4460655450820923, "learning_rate": 9.999690890285053e-05, "loss": 0.0529, "step": 4270 }, { "grad_norm": 0.5206024050712585, "learning_rate": 9.999667481232356e-05, "loss": 0.0546, "step": 4280 }, { "grad_norm": 0.4885442852973938, "learning_rate": 9.999643217873225e-05, "loss": 0.0539, "step": 4290 }, { "grad_norm": 0.5157468318939209, "learning_rate": 9.999618100211809e-05, "loss": 0.0525, "step": 4300 }, { "grad_norm": 0.5613343119621277, "learning_rate": 9.999592128252402e-05, "loss": 0.0439, "step": 4310 }, { "grad_norm": 0.2732822000980377, "learning_rate": 9.999565301999437e-05, "loss": 0.048, "step": 4320 }, { "grad_norm": 0.3976214528083801, "learning_rate": 9.999537621457502e-05, "loss": 0.047, "step": 4330 }, { "grad_norm": 0.4027195870876312, "learning_rate": 9.999509086631323e-05, "loss": 0.0418, "step": 4340 }, { "grad_norm": 0.5277324318885803, "learning_rate": 9.99947969752578e-05, "loss": 0.0472, "step": 4350 }, { "grad_norm": 0.5152819752693176, "learning_rate": 9.999449454145891e-05, "loss": 0.0456, "step": 4360 }, { "grad_norm": 0.43630075454711914, "learning_rate": 9.999418356496827e-05, "loss": 0.0484, "step": 4370 }, { "grad_norm": 0.401637464761734, "learning_rate": 9.999386404583899e-05, "loss": 0.0456, "step": 4380 }, { "grad_norm": 0.4351535737514496, "learning_rate": 9.999353598412568e-05, "loss": 0.0512, "step": 4390 }, { "grad_norm": 0.5282711982727051, "learning_rate": 9.999319937988442e-05, "loss": 0.0566, "step": 4400 }, { "grad_norm": 0.4086388647556305, "learning_rate": 9.999285423317268e-05, "loss": 0.054, "step": 4410 }, { "grad_norm": 0.4592735767364502, "learning_rate": 9.999250054404947e-05, "loss": 0.0534, "step": 4420 }, { "grad_norm": 0.3674286901950836, "learning_rate": 9.99921383125752e-05, "loss": 0.0508, "step": 4430 }, { "grad_norm": 0.6955254673957825, "learning_rate": 9.99917675388118e-05, "loss": 0.0585, "step": 4440 }, { "grad_norm": 0.44191768765449524, "learning_rate": 9.99913882228226e-05, "loss": 0.0498, "step": 4450 }, { "grad_norm": 0.473449170589447, "learning_rate": 9.999100036467242e-05, "loss": 0.0545, "step": 4460 }, { "grad_norm": 0.38755524158477783, "learning_rate": 9.999060396442753e-05, "loss": 0.0459, "step": 4470 }, { "grad_norm": 0.35153552889823914, "learning_rate": 9.999019902215566e-05, "loss": 0.0467, "step": 4480 }, { "grad_norm": 0.28706613183021545, "learning_rate": 9.998978553792602e-05, "loss": 0.046, "step": 4490 }, { "grad_norm": 0.3076217472553253, "learning_rate": 9.998936351180926e-05, "loss": 0.0459, "step": 4500 }, { "grad_norm": 0.44720253348350525, "learning_rate": 9.998893294387747e-05, "loss": 0.0529, "step": 4510 }, { "grad_norm": 0.42659151554107666, "learning_rate": 9.998849383420426e-05, "loss": 0.0465, "step": 4520 }, { "grad_norm": 0.31520795822143555, "learning_rate": 9.998804618286465e-05, "loss": 0.0473, "step": 4530 }, { "grad_norm": 0.5367616415023804, "learning_rate": 9.99875899899351e-05, "loss": 0.0447, "step": 4540 }, { "grad_norm": 0.41338980197906494, "learning_rate": 9.99871252554936e-05, "loss": 0.0468, "step": 4550 }, { "grad_norm": 0.4059814214706421, "learning_rate": 9.998665197961955e-05, "loss": 0.0398, "step": 4560 }, { "grad_norm": 0.3609146773815155, "learning_rate": 9.998617016239379e-05, "loss": 0.0442, "step": 4570 }, { "grad_norm": 0.3619050681591034, "learning_rate": 9.998567980389869e-05, "loss": 0.0474, "step": 4580 }, { "grad_norm": 0.38218244910240173, "learning_rate": 9.998518090421802e-05, "loss": 0.0465, "step": 4590 }, { "grad_norm": 0.4719562828540802, "learning_rate": 9.998467346343703e-05, "loss": 0.0461, "step": 4600 }, { "grad_norm": 0.3982713520526886, "learning_rate": 9.998415748164243e-05, "loss": 0.0473, "step": 4610 }, { "grad_norm": 0.4845476448535919, "learning_rate": 9.998363295892238e-05, "loss": 0.0466, "step": 4620 }, { "grad_norm": 0.4581291079521179, "learning_rate": 9.998309989536652e-05, "loss": 0.0441, "step": 4630 }, { "grad_norm": 0.48923057317733765, "learning_rate": 9.998255829106593e-05, "loss": 0.0495, "step": 4640 }, { "grad_norm": 0.4628112316131592, "learning_rate": 9.998200814611316e-05, "loss": 0.0551, "step": 4650 }, { "grad_norm": 0.31906723976135254, "learning_rate": 9.998144946060219e-05, "loss": 0.0532, "step": 4660 }, { "grad_norm": 0.4874865710735321, "learning_rate": 9.998088223462852e-05, "loss": 0.0514, "step": 4670 }, { "grad_norm": 0.32196107506752014, "learning_rate": 9.998030646828905e-05, "loss": 0.0429, "step": 4680 }, { "grad_norm": 0.4173595905303955, "learning_rate": 9.997972216168217e-05, "loss": 0.0513, "step": 4690 }, { "grad_norm": 0.2987581193447113, "learning_rate": 9.997912931490771e-05, "loss": 0.0551, "step": 4700 }, { "grad_norm": 0.4745461046695709, "learning_rate": 9.9978527928067e-05, "loss": 0.0521, "step": 4710 }, { "grad_norm": 0.47733983397483826, "learning_rate": 9.997791800126277e-05, "loss": 0.0493, "step": 4720 }, { "grad_norm": 0.3070448935031891, "learning_rate": 9.997729953459927e-05, "loss": 0.0481, "step": 4730 }, { "grad_norm": 0.4483560025691986, "learning_rate": 9.997667252818214e-05, "loss": 0.0446, "step": 4740 }, { "grad_norm": 0.42794933915138245, "learning_rate": 9.997603698211855e-05, "loss": 0.0504, "step": 4750 }, { "grad_norm": 0.29295727610588074, "learning_rate": 9.99753928965171e-05, "loss": 0.0509, "step": 4760 }, { "grad_norm": 0.4279433786869049, "learning_rate": 9.997474027148781e-05, "loss": 0.0495, "step": 4770 }, { "grad_norm": 0.33933937549591064, "learning_rate": 9.997407910714223e-05, "loss": 0.0489, "step": 4780 }, { "grad_norm": 0.30896031856536865, "learning_rate": 9.997340940359332e-05, "loss": 0.0483, "step": 4790 }, { "grad_norm": 0.5223690867424011, "learning_rate": 9.997273116095552e-05, "loss": 0.0505, "step": 4800 }, { "grad_norm": 0.35150349140167236, "learning_rate": 9.997204437934473e-05, "loss": 0.0453, "step": 4810 }, { "grad_norm": 0.4011611342430115, "learning_rate": 9.997134905887829e-05, "loss": 0.0417, "step": 4820 }, { "grad_norm": 0.4544658362865448, "learning_rate": 9.997064519967501e-05, "loss": 0.0485, "step": 4830 }, { "grad_norm": 0.5231920480728149, "learning_rate": 9.996993280185517e-05, "loss": 0.0504, "step": 4840 }, { "grad_norm": 0.6007421016693115, "learning_rate": 9.99692118655405e-05, "loss": 0.0525, "step": 4850 }, { "grad_norm": 0.36336588859558105, "learning_rate": 9.996848239085417e-05, "loss": 0.0516, "step": 4860 }, { "grad_norm": 0.5642521977424622, "learning_rate": 9.996774437792085e-05, "loss": 0.0515, "step": 4870 }, { "grad_norm": 0.4976515769958496, "learning_rate": 9.996699782686664e-05, "loss": 0.05, "step": 4880 }, { "grad_norm": 0.4070780575275421, "learning_rate": 9.996624273781909e-05, "loss": 0.0565, "step": 4890 }, { "grad_norm": 0.45501193404197693, "learning_rate": 9.996547911090725e-05, "loss": 0.0509, "step": 4900 }, { "grad_norm": 0.2923313081264496, "learning_rate": 9.996470694626157e-05, "loss": 0.0457, "step": 4910 }, { "grad_norm": 0.365083783864975, "learning_rate": 9.996392624401403e-05, "loss": 0.0446, "step": 4920 }, { "grad_norm": 0.4851283133029938, "learning_rate": 9.996313700429801e-05, "loss": 0.0506, "step": 4930 }, { "grad_norm": 0.49396422505378723, "learning_rate": 9.996233922724836e-05, "loss": 0.048, "step": 4940 }, { "grad_norm": 0.46892744302749634, "learning_rate": 9.996153291300141e-05, "loss": 0.0571, "step": 4950 }, { "grad_norm": 0.40098556876182556, "learning_rate": 9.996071806169494e-05, "loss": 0.0453, "step": 4960 }, { "grad_norm": 0.3641745448112488, "learning_rate": 9.995989467346817e-05, "loss": 0.0531, "step": 4970 }, { "grad_norm": 0.363136351108551, "learning_rate": 9.995906274846183e-05, "loss": 0.0479, "step": 4980 }, { "grad_norm": 0.3835716247558594, "learning_rate": 9.995822228681803e-05, "loss": 0.0454, "step": 4990 }, { "grad_norm": 0.33988553285598755, "learning_rate": 9.99573732886804e-05, "loss": 0.0421, "step": 5000 }, { "grad_norm": 0.361171156167984, "learning_rate": 9.995651575419402e-05, "loss": 0.04, "step": 5010 }, { "grad_norm": 0.5401168465614319, "learning_rate": 9.995564968350541e-05, "loss": 0.0381, "step": 5020 }, { "grad_norm": 0.43454062938690186, "learning_rate": 9.995477507676256e-05, "loss": 0.0428, "step": 5030 }, { "grad_norm": 0.5630381107330322, "learning_rate": 9.995389193411493e-05, "loss": 0.0399, "step": 5040 }, { "grad_norm": 0.4581230580806732, "learning_rate": 9.995300025571339e-05, "loss": 0.0439, "step": 5050 }, { "grad_norm": 0.3493824899196625, "learning_rate": 9.995210004171034e-05, "loss": 0.0423, "step": 5060 }, { "grad_norm": 0.28571557998657227, "learning_rate": 9.995119129225956e-05, "loss": 0.0372, "step": 5070 }, { "grad_norm": 0.47328120470046997, "learning_rate": 9.995027400751637e-05, "loss": 0.0425, "step": 5080 }, { "grad_norm": 0.3777240812778473, "learning_rate": 9.994934818763751e-05, "loss": 0.0402, "step": 5090 }, { "grad_norm": 0.36875951290130615, "learning_rate": 9.994841383278115e-05, "loss": 0.0491, "step": 5100 }, { "grad_norm": 0.40168654918670654, "learning_rate": 9.994747094310695e-05, "loss": 0.0495, "step": 5110 }, { "grad_norm": 0.42780715227127075, "learning_rate": 9.994651951877604e-05, "loss": 0.0585, "step": 5120 }, { "grad_norm": 0.4117453396320343, "learning_rate": 9.994555955995099e-05, "loss": 0.0503, "step": 5130 }, { "grad_norm": 0.4469781219959259, "learning_rate": 9.994459106679581e-05, "loss": 0.0498, "step": 5140 }, { "grad_norm": 0.4226412773132324, "learning_rate": 9.994361403947603e-05, "loss": 0.0483, "step": 5150 }, { "grad_norm": 0.45401260256767273, "learning_rate": 9.994262847815854e-05, "loss": 0.048, "step": 5160 }, { "grad_norm": 0.4074515104293823, "learning_rate": 9.99416343830118e-05, "loss": 0.0425, "step": 5170 }, { "grad_norm": 0.31235575675964355, "learning_rate": 9.994063175420565e-05, "loss": 0.0429, "step": 5180 }, { "grad_norm": 0.3987966775894165, "learning_rate": 9.99396205919114e-05, "loss": 0.0457, "step": 5190 }, { "grad_norm": 0.32162803411483765, "learning_rate": 9.993860089630185e-05, "loss": 0.045, "step": 5200 }, { "grad_norm": 0.3575228452682495, "learning_rate": 9.993757266755123e-05, "loss": 0.0408, "step": 5210 }, { "grad_norm": 0.36386141180992126, "learning_rate": 9.993653590583522e-05, "loss": 0.0426, "step": 5220 }, { "grad_norm": 0.3633628189563751, "learning_rate": 9.993549061133102e-05, "loss": 0.0399, "step": 5230 }, { "grad_norm": 0.6027148962020874, "learning_rate": 9.993443678421719e-05, "loss": 0.0465, "step": 5240 }, { "grad_norm": 0.38467225432395935, "learning_rate": 9.993337442467384e-05, "loss": 0.0385, "step": 5250 }, { "grad_norm": 0.26795998215675354, "learning_rate": 9.993230353288248e-05, "loss": 0.0445, "step": 5260 }, { "grad_norm": 0.3457346558570862, "learning_rate": 9.993122410902608e-05, "loss": 0.0414, "step": 5270 }, { "grad_norm": 0.4184693992137909, "learning_rate": 9.993013615328912e-05, "loss": 0.0455, "step": 5280 }, { "grad_norm": 0.3301035761833191, "learning_rate": 9.992903966585747e-05, "loss": 0.0414, "step": 5290 }, { "grad_norm": 0.4609445035457611, "learning_rate": 9.992793464691852e-05, "loss": 0.043, "step": 5300 }, { "grad_norm": 0.4286223351955414, "learning_rate": 9.992682109666105e-05, "loss": 0.0454, "step": 5310 }, { "grad_norm": 0.5829910039901733, "learning_rate": 9.992569901527538e-05, "loss": 0.0509, "step": 5320 }, { "grad_norm": 0.363817036151886, "learning_rate": 9.99245684029532e-05, "loss": 0.045, "step": 5330 }, { "grad_norm": 0.39406439661979675, "learning_rate": 9.992342925988774e-05, "loss": 0.0446, "step": 5340 }, { "grad_norm": 0.4749637842178345, "learning_rate": 9.992228158627361e-05, "loss": 0.0428, "step": 5350 }, { "grad_norm": 0.43053799867630005, "learning_rate": 9.992112538230693e-05, "loss": 0.0477, "step": 5360 }, { "grad_norm": 0.3121701180934906, "learning_rate": 9.991996064818527e-05, "loss": 0.0449, "step": 5370 }, { "grad_norm": 0.362546443939209, "learning_rate": 9.991878738410768e-05, "loss": 0.0457, "step": 5380 }, { "grad_norm": 0.5231050252914429, "learning_rate": 9.991760559027457e-05, "loss": 0.0444, "step": 5390 }, { "grad_norm": 0.38914182782173157, "learning_rate": 9.991641526688793e-05, "loss": 0.0419, "step": 5400 }, { "grad_norm": 0.40705570578575134, "learning_rate": 9.991521641415113e-05, "loss": 0.0446, "step": 5410 }, { "grad_norm": 0.3367755115032196, "learning_rate": 9.991400903226904e-05, "loss": 0.0411, "step": 5420 }, { "grad_norm": 0.45072588324546814, "learning_rate": 9.991279312144794e-05, "loss": 0.042, "step": 5430 }, { "grad_norm": 0.3340522050857544, "learning_rate": 9.991156868189564e-05, "loss": 0.0416, "step": 5440 }, { "grad_norm": 0.44809675216674805, "learning_rate": 9.991033571382131e-05, "loss": 0.0434, "step": 5450 }, { "grad_norm": 0.277905136346817, "learning_rate": 9.990909421743569e-05, "loss": 0.0376, "step": 5460 }, { "grad_norm": 0.3569411337375641, "learning_rate": 9.990784419295085e-05, "loss": 0.0369, "step": 5470 }, { "grad_norm": 0.31797173619270325, "learning_rate": 9.990658564058044e-05, "loss": 0.039, "step": 5480 }, { "grad_norm": 0.2884273827075958, "learning_rate": 9.990531856053948e-05, "loss": 0.0371, "step": 5490 }, { "grad_norm": 0.29545846581459045, "learning_rate": 9.99040429530445e-05, "loss": 0.0369, "step": 5500 }, { "grad_norm": 0.3935755789279938, "learning_rate": 9.990275881831346e-05, "loss": 0.0399, "step": 5510 }, { "grad_norm": 0.36305564641952515, "learning_rate": 9.990146615656577e-05, "loss": 0.0364, "step": 5520 }, { "grad_norm": 0.38898640871047974, "learning_rate": 9.990016496802233e-05, "loss": 0.0437, "step": 5530 }, { "grad_norm": 0.39391523599624634, "learning_rate": 9.989885525290548e-05, "loss": 0.0488, "step": 5540 }, { "grad_norm": 0.2768344283103943, "learning_rate": 9.989753701143897e-05, "loss": 0.0394, "step": 5550 }, { "grad_norm": 0.40026187896728516, "learning_rate": 9.989621024384812e-05, "loss": 0.044, "step": 5560 }, { "grad_norm": 0.49375370144844055, "learning_rate": 9.989487495035959e-05, "loss": 0.0409, "step": 5570 }, { "grad_norm": 0.5607299208641052, "learning_rate": 9.989353113120156e-05, "loss": 0.0437, "step": 5580 }, { "grad_norm": 0.4465269148349762, "learning_rate": 9.989217878660366e-05, "loss": 0.0457, "step": 5590 }, { "grad_norm": 0.3882370591163635, "learning_rate": 9.989081791679695e-05, "loss": 0.0527, "step": 5600 }, { "grad_norm": 0.37474697828292847, "learning_rate": 9.988944852201397e-05, "loss": 0.0447, "step": 5610 }, { "grad_norm": 0.3734793961048126, "learning_rate": 9.988807060248873e-05, "loss": 0.0476, "step": 5620 }, { "grad_norm": 0.4472992420196533, "learning_rate": 9.988668415845665e-05, "loss": 0.0412, "step": 5630 }, { "grad_norm": 0.43122655153274536, "learning_rate": 9.988528919015466e-05, "loss": 0.052, "step": 5640 }, { "grad_norm": 0.4088887870311737, "learning_rate": 9.988388569782112e-05, "loss": 0.0401, "step": 5650 }, { "grad_norm": 0.3591986894607544, "learning_rate": 9.988247368169583e-05, "loss": 0.0429, "step": 5660 }, { "grad_norm": 0.32930299639701843, "learning_rate": 9.988105314202007e-05, "loss": 0.0425, "step": 5670 }, { "grad_norm": 0.35162585973739624, "learning_rate": 9.987962407903659e-05, "loss": 0.0391, "step": 5680 }, { "grad_norm": 0.3327031135559082, "learning_rate": 9.987818649298957e-05, "loss": 0.0448, "step": 5690 }, { "grad_norm": 0.3103363811969757, "learning_rate": 9.987674038412465e-05, "loss": 0.0396, "step": 5700 }, { "grad_norm": 0.3816494643688202, "learning_rate": 9.987528575268891e-05, "loss": 0.0426, "step": 5710 }, { "grad_norm": 0.4012691378593445, "learning_rate": 9.987382259893095e-05, "loss": 0.047, "step": 5720 }, { "grad_norm": 0.6018237471580505, "learning_rate": 9.987235092310074e-05, "loss": 0.0543, "step": 5730 }, { "grad_norm": 0.31186750531196594, "learning_rate": 9.987087072544978e-05, "loss": 0.0451, "step": 5740 }, { "grad_norm": 0.5536149144172668, "learning_rate": 9.9869382006231e-05, "loss": 0.0361, "step": 5750 }, { "grad_norm": 0.34462684392929077, "learning_rate": 9.986788476569875e-05, "loss": 0.046, "step": 5760 }, { "grad_norm": 0.3721707761287689, "learning_rate": 9.986637900410887e-05, "loss": 0.0415, "step": 5770 }, { "grad_norm": 0.37931710481643677, "learning_rate": 9.986486472171869e-05, "loss": 0.0432, "step": 5780 }, { "grad_norm": 0.29469746351242065, "learning_rate": 9.986334191878692e-05, "loss": 0.0393, "step": 5790 }, { "grad_norm": 0.372286319732666, "learning_rate": 9.986181059557378e-05, "loss": 0.04, "step": 5800 }, { "grad_norm": 0.3165684938430786, "learning_rate": 9.986027075234094e-05, "loss": 0.0409, "step": 5810 }, { "grad_norm": 0.34192875027656555, "learning_rate": 9.985872238935152e-05, "loss": 0.0394, "step": 5820 }, { "grad_norm": 0.36455318331718445, "learning_rate": 9.985716550687008e-05, "loss": 0.039, "step": 5830 }, { "grad_norm": 0.33722156286239624, "learning_rate": 9.985560010516264e-05, "loss": 0.0361, "step": 5840 }, { "grad_norm": 0.35575202107429504, "learning_rate": 9.985402618449668e-05, "loss": 0.0448, "step": 5850 }, { "grad_norm": 0.3604450821876526, "learning_rate": 9.985244374514118e-05, "loss": 0.0369, "step": 5860 }, { "grad_norm": 0.4006516933441162, "learning_rate": 9.985085278736651e-05, "loss": 0.0401, "step": 5870 }, { "grad_norm": 0.4137957692146301, "learning_rate": 9.984925331144452e-05, "loss": 0.0534, "step": 5880 }, { "grad_norm": 0.4221997559070587, "learning_rate": 9.984764531764851e-05, "loss": 0.0421, "step": 5890 }, { "grad_norm": 0.3300626873970032, "learning_rate": 9.984602880625326e-05, "loss": 0.0423, "step": 5900 }, { "grad_norm": 0.36521828174591064, "learning_rate": 9.9844403777535e-05, "loss": 0.0383, "step": 5910 }, { "grad_norm": 0.44008249044418335, "learning_rate": 9.984277023177135e-05, "loss": 0.0503, "step": 5920 }, { "grad_norm": 0.4247158169746399, "learning_rate": 9.984112816924148e-05, "loss": 0.0366, "step": 5930 }, { "grad_norm": 0.3234817087650299, "learning_rate": 9.983947759022596e-05, "loss": 0.0416, "step": 5940 }, { "grad_norm": 0.33165714144706726, "learning_rate": 9.983781849500682e-05, "loss": 0.0423, "step": 5950 }, { "grad_norm": 0.43759462237358093, "learning_rate": 9.98361508838676e-05, "loss": 0.0431, "step": 5960 }, { "grad_norm": 0.29049134254455566, "learning_rate": 9.98344747570932e-05, "loss": 0.0395, "step": 5970 }, { "grad_norm": 0.41675928235054016, "learning_rate": 9.983279011497004e-05, "loss": 0.0451, "step": 5980 }, { "grad_norm": 0.3443704843521118, "learning_rate": 9.983109695778596e-05, "loss": 0.043, "step": 5990 }, { "grad_norm": 0.4752723276615143, "learning_rate": 9.982939528583032e-05, "loss": 0.0453, "step": 6000 }, { "grad_norm": 0.2915893495082855, "learning_rate": 9.982768509939385e-05, "loss": 0.0435, "step": 6010 }, { "grad_norm": 0.3078188896179199, "learning_rate": 9.982596639876879e-05, "loss": 0.0445, "step": 6020 }, { "grad_norm": 0.3562212586402893, "learning_rate": 9.982423918424881e-05, "loss": 0.0408, "step": 6030 }, { "grad_norm": 0.3048346936702728, "learning_rate": 9.982250345612908e-05, "loss": 0.0475, "step": 6040 }, { "grad_norm": 0.40832701325416565, "learning_rate": 9.982075921470611e-05, "loss": 0.047, "step": 6050 }, { "grad_norm": 0.3176436424255371, "learning_rate": 9.981900646027802e-05, "loss": 0.0472, "step": 6060 }, { "grad_norm": 0.4630812406539917, "learning_rate": 9.981724519314425e-05, "loss": 0.0496, "step": 6070 }, { "grad_norm": 0.47071516513824463, "learning_rate": 9.981547541360581e-05, "loss": 0.0573, "step": 6080 }, { "grad_norm": 0.32655128836631775, "learning_rate": 9.981369712196508e-05, "loss": 0.0474, "step": 6090 }, { "grad_norm": 0.3364417254924774, "learning_rate": 9.981191031852592e-05, "loss": 0.0417, "step": 6100 }, { "grad_norm": 0.3629354238510132, "learning_rate": 9.981011500359362e-05, "loss": 0.0363, "step": 6110 }, { "grad_norm": 0.3025952875614166, "learning_rate": 9.9808311177475e-05, "loss": 0.0395, "step": 6120 }, { "grad_norm": 0.4049331247806549, "learning_rate": 9.980649884047826e-05, "loss": 0.0379, "step": 6130 }, { "grad_norm": 0.48186346888542175, "learning_rate": 9.980467799291307e-05, "loss": 0.04, "step": 6140 }, { "grad_norm": 0.38822826743125916, "learning_rate": 9.980284863509058e-05, "loss": 0.0432, "step": 6150 }, { "grad_norm": 0.3638334572315216, "learning_rate": 9.980101076732334e-05, "loss": 0.0482, "step": 6160 }, { "grad_norm": 0.421002596616745, "learning_rate": 9.979916438992544e-05, "loss": 0.0461, "step": 6170 }, { "grad_norm": 0.450176477432251, "learning_rate": 9.979730950321237e-05, "loss": 0.0482, "step": 6180 }, { "grad_norm": 0.37306129932403564, "learning_rate": 9.979544610750104e-05, "loss": 0.0432, "step": 6190 }, { "grad_norm": 0.34368234872817993, "learning_rate": 9.97935742031099e-05, "loss": 0.0494, "step": 6200 }, { "grad_norm": 0.306571364402771, "learning_rate": 9.979169379035878e-05, "loss": 0.0484, "step": 6210 }, { "grad_norm": 0.3902214765548706, "learning_rate": 9.978980486956899e-05, "loss": 0.0403, "step": 6220 }, { "grad_norm": 0.4099586606025696, "learning_rate": 9.978790744106332e-05, "loss": 0.0454, "step": 6230 }, { "grad_norm": 0.30369120836257935, "learning_rate": 9.978600150516594e-05, "loss": 0.0448, "step": 6240 }, { "grad_norm": 0.36814719438552856, "learning_rate": 9.978408706220259e-05, "loss": 0.0422, "step": 6250 }, { "grad_norm": 0.364088237285614, "learning_rate": 9.978216411250032e-05, "loss": 0.0443, "step": 6260 }, { "grad_norm": 0.3563232123851776, "learning_rate": 9.978023265638778e-05, "loss": 0.0361, "step": 6270 }, { "grad_norm": 0.3781141936779022, "learning_rate": 9.977829269419495e-05, "loss": 0.0441, "step": 6280 }, { "grad_norm": 0.3377148509025574, "learning_rate": 9.977634422625335e-05, "loss": 0.0413, "step": 6290 }, { "grad_norm": 0.35092300176620483, "learning_rate": 9.97743872528959e-05, "loss": 0.0412, "step": 6300 }, { "grad_norm": 0.3934015929698944, "learning_rate": 9.9772421774457e-05, "loss": 0.0366, "step": 6310 }, { "grad_norm": 0.37159264087677, "learning_rate": 9.977044779127252e-05, "loss": 0.034, "step": 6320 }, { "grad_norm": 0.3719487488269806, "learning_rate": 9.976846530367971e-05, "loss": 0.0402, "step": 6330 }, { "grad_norm": 0.45234403014183044, "learning_rate": 9.976647431201735e-05, "loss": 0.0451, "step": 6340 }, { "grad_norm": 0.3245910108089447, "learning_rate": 9.976447481662568e-05, "loss": 0.0412, "step": 6350 }, { "grad_norm": 0.3159354329109192, "learning_rate": 9.976246681784629e-05, "loss": 0.0421, "step": 6360 }, { "grad_norm": 0.4204883277416229, "learning_rate": 9.976045031602234e-05, "loss": 0.044, "step": 6370 }, { "grad_norm": 0.5012466907501221, "learning_rate": 9.975842531149837e-05, "loss": 0.0549, "step": 6380 }, { "grad_norm": 0.48796209692955017, "learning_rate": 9.975639180462043e-05, "loss": 0.0499, "step": 6390 }, { "grad_norm": 0.2977941930294037, "learning_rate": 9.975434979573596e-05, "loss": 0.0424, "step": 6400 }, { "grad_norm": 0.41893959045410156, "learning_rate": 9.97522992851939e-05, "loss": 0.0444, "step": 6410 }, { "grad_norm": 0.5032734870910645, "learning_rate": 9.975024027334461e-05, "loss": 0.0462, "step": 6420 }, { "grad_norm": 0.36634474992752075, "learning_rate": 9.974817276053993e-05, "loss": 0.0374, "step": 6430 }, { "grad_norm": 0.5573779940605164, "learning_rate": 9.974609674713315e-05, "loss": 0.0428, "step": 6440 }, { "grad_norm": 0.3549627959728241, "learning_rate": 9.9744012233479e-05, "loss": 0.0366, "step": 6450 }, { "grad_norm": 0.4631361961364746, "learning_rate": 9.974191921993366e-05, "loss": 0.0444, "step": 6460 }, { "grad_norm": 0.5011640191078186, "learning_rate": 9.973981770685474e-05, "loss": 0.0525, "step": 6470 }, { "grad_norm": 0.427012175321579, "learning_rate": 9.97377076946014e-05, "loss": 0.048, "step": 6480 }, { "grad_norm": 0.31270265579223633, "learning_rate": 9.973558918353412e-05, "loss": 0.0452, "step": 6490 }, { "grad_norm": 0.4167088270187378, "learning_rate": 9.973346217401494e-05, "loss": 0.039, "step": 6500 }, { "grad_norm": 0.4728192090988159, "learning_rate": 9.973132666640726e-05, "loss": 0.0492, "step": 6510 }, { "grad_norm": 0.44456881284713745, "learning_rate": 9.972918266107602e-05, "loss": 0.0493, "step": 6520 }, { "grad_norm": 0.3474763035774231, "learning_rate": 9.972703015838756e-05, "loss": 0.0503, "step": 6530 }, { "grad_norm": 0.38429638743400574, "learning_rate": 9.97248691587097e-05, "loss": 0.0444, "step": 6540 }, { "grad_norm": 0.33965805172920227, "learning_rate": 9.972269966241166e-05, "loss": 0.0463, "step": 6550 }, { "grad_norm": 0.3735905885696411, "learning_rate": 9.972052166986417e-05, "loss": 0.0437, "step": 6560 }, { "grad_norm": 0.31133875250816345, "learning_rate": 9.971833518143938e-05, "loss": 0.0415, "step": 6570 }, { "grad_norm": 0.42698100209236145, "learning_rate": 9.971614019751093e-05, "loss": 0.0508, "step": 6580 }, { "grad_norm": 0.4601062536239624, "learning_rate": 9.971393671845383e-05, "loss": 0.0404, "step": 6590 }, { "grad_norm": 0.3369807004928589, "learning_rate": 9.971172474464464e-05, "loss": 0.0387, "step": 6600 }, { "grad_norm": 0.2581539750099182, "learning_rate": 9.97095042764613e-05, "loss": 0.0391, "step": 6610 }, { "grad_norm": 0.3017904758453369, "learning_rate": 9.970727531428324e-05, "loss": 0.0379, "step": 6620 }, { "grad_norm": 0.4501408338546753, "learning_rate": 9.970503785849132e-05, "loss": 0.047, "step": 6630 }, { "grad_norm": 0.3074027895927429, "learning_rate": 9.970279190946788e-05, "loss": 0.0391, "step": 6640 }, { "grad_norm": 0.41706687211990356, "learning_rate": 9.970053746759667e-05, "loss": 0.0408, "step": 6650 }, { "grad_norm": 0.34895360469818115, "learning_rate": 9.969827453326292e-05, "loss": 0.0342, "step": 6660 }, { "grad_norm": 0.395799458026886, "learning_rate": 9.969600310685332e-05, "loss": 0.036, "step": 6670 }, { "grad_norm": 0.3794163167476654, "learning_rate": 9.969372318875596e-05, "loss": 0.0391, "step": 6680 }, { "grad_norm": 0.2733258903026581, "learning_rate": 9.969143477936043e-05, "loss": 0.0383, "step": 6690 }, { "grad_norm": 0.40507176518440247, "learning_rate": 9.968913787905775e-05, "loss": 0.0399, "step": 6700 }, { "grad_norm": 0.442362517118454, "learning_rate": 9.968683248824045e-05, "loss": 0.0415, "step": 6710 }, { "grad_norm": 0.38291501998901367, "learning_rate": 9.968451860730238e-05, "loss": 0.0426, "step": 6720 }, { "grad_norm": 0.40462973713874817, "learning_rate": 9.968219623663896e-05, "loss": 0.0393, "step": 6730 }, { "grad_norm": 0.2885414958000183, "learning_rate": 9.967986537664702e-05, "loss": 0.041, "step": 6740 }, { "grad_norm": 0.2354590743780136, "learning_rate": 9.967752602772483e-05, "loss": 0.0328, "step": 6750 }, { "grad_norm": 0.3538680672645569, "learning_rate": 9.967517819027212e-05, "loss": 0.0373, "step": 6760 }, { "grad_norm": 0.36376211047172546, "learning_rate": 9.967282186469009e-05, "loss": 0.0434, "step": 6770 }, { "grad_norm": 0.32360905408859253, "learning_rate": 9.967045705138135e-05, "loss": 0.0404, "step": 6780 }, { "grad_norm": 0.43166881799697876, "learning_rate": 9.966808375074998e-05, "loss": 0.0424, "step": 6790 }, { "grad_norm": 0.40831807255744934, "learning_rate": 9.966570196320154e-05, "loss": 0.0512, "step": 6800 }, { "grad_norm": 0.3280826807022095, "learning_rate": 9.966331168914299e-05, "loss": 0.0381, "step": 6810 }, { "grad_norm": 0.3753024637699127, "learning_rate": 9.966091292898277e-05, "loss": 0.0404, "step": 6820 }, { "grad_norm": 0.3888947665691376, "learning_rate": 9.965850568313076e-05, "loss": 0.0356, "step": 6830 }, { "grad_norm": 0.4097350537776947, "learning_rate": 9.965608995199827e-05, "loss": 0.0416, "step": 6840 }, { "grad_norm": 0.38388657569885254, "learning_rate": 9.965366573599812e-05, "loss": 0.0344, "step": 6850 }, { "grad_norm": 0.3979587256908417, "learning_rate": 9.965123303554453e-05, "loss": 0.0422, "step": 6860 }, { "grad_norm": 0.3239588439464569, "learning_rate": 9.964879185105317e-05, "loss": 0.0422, "step": 6870 }, { "grad_norm": 0.4585088789463043, "learning_rate": 9.964634218294119e-05, "loss": 0.0431, "step": 6880 }, { "grad_norm": 0.24072344601154327, "learning_rate": 9.964388403162714e-05, "loss": 0.0396, "step": 6890 }, { "grad_norm": 0.38076138496398926, "learning_rate": 9.96414173975311e-05, "loss": 0.0363, "step": 6900 }, { "grad_norm": 0.42047369480133057, "learning_rate": 9.963894228107451e-05, "loss": 0.0371, "step": 6910 }, { "grad_norm": 0.39748886227607727, "learning_rate": 9.963645868268032e-05, "loss": 0.0434, "step": 6920 }, { "grad_norm": 0.35684797167778015, "learning_rate": 9.963396660277289e-05, "loss": 0.0373, "step": 6930 }, { "grad_norm": 0.42453813552856445, "learning_rate": 9.963146604177807e-05, "loss": 0.0365, "step": 6940 }, { "grad_norm": 0.28357452154159546, "learning_rate": 9.962895700012311e-05, "loss": 0.0405, "step": 6950 }, { "grad_norm": 0.2812986671924591, "learning_rate": 9.962643947823677e-05, "loss": 0.0353, "step": 6960 }, { "grad_norm": 0.3241969347000122, "learning_rate": 9.962391347654921e-05, "loss": 0.0367, "step": 6970 }, { "grad_norm": 0.3503277599811554, "learning_rate": 9.962137899549204e-05, "loss": 0.0384, "step": 6980 }, { "grad_norm": 0.35699397325515747, "learning_rate": 9.961883603549835e-05, "loss": 0.0431, "step": 6990 }, { "grad_norm": 0.38515597581863403, "learning_rate": 9.961628459700267e-05, "loss": 0.0372, "step": 7000 }, { "grad_norm": 0.4170619249343872, "learning_rate": 9.961372468044095e-05, "loss": 0.0379, "step": 7010 }, { "grad_norm": 0.35797959566116333, "learning_rate": 9.961115628625062e-05, "loss": 0.035, "step": 7020 }, { "grad_norm": 0.23833660781383514, "learning_rate": 9.960857941487056e-05, "loss": 0.0411, "step": 7030 }, { "grad_norm": 0.352483332157135, "learning_rate": 9.960599406674106e-05, "loss": 0.0403, "step": 7040 }, { "grad_norm": 0.3047409653663635, "learning_rate": 9.960340024230393e-05, "loss": 0.0435, "step": 7050 }, { "grad_norm": 0.3204633891582489, "learning_rate": 9.960079794200232e-05, "loss": 0.0356, "step": 7060 }, { "grad_norm": 0.3600463271141052, "learning_rate": 9.959818716628096e-05, "loss": 0.0433, "step": 7070 }, { "grad_norm": 0.37854111194610596, "learning_rate": 9.95955679155859e-05, "loss": 0.036, "step": 7080 }, { "grad_norm": 0.3832847774028778, "learning_rate": 9.959294019036472e-05, "loss": 0.0373, "step": 7090 }, { "grad_norm": 0.3869566023349762, "learning_rate": 9.959030399106646e-05, "loss": 0.0435, "step": 7100 }, { "grad_norm": 0.28280699253082275, "learning_rate": 9.958765931814153e-05, "loss": 0.0353, "step": 7110 }, { "grad_norm": 0.3883381187915802, "learning_rate": 9.958500617204184e-05, "loss": 0.0391, "step": 7120 }, { "grad_norm": 0.36908477544784546, "learning_rate": 9.958234455322075e-05, "loss": 0.0351, "step": 7130 }, { "grad_norm": 0.3705887198448181, "learning_rate": 9.957967446213308e-05, "loss": 0.0397, "step": 7140 }, { "grad_norm": 0.3438790440559387, "learning_rate": 9.957699589923501e-05, "loss": 0.0448, "step": 7150 }, { "grad_norm": 0.37888669967651367, "learning_rate": 9.957430886498431e-05, "loss": 0.0411, "step": 7160 }, { "grad_norm": 0.4722294211387634, "learning_rate": 9.957161335984008e-05, "loss": 0.0474, "step": 7170 }, { "grad_norm": 0.3684791922569275, "learning_rate": 9.956890938426291e-05, "loss": 0.0463, "step": 7180 }, { "grad_norm": 0.3661195933818817, "learning_rate": 9.956619693871482e-05, "loss": 0.0489, "step": 7190 }, { "grad_norm": 0.37388530373573303, "learning_rate": 9.956347602365934e-05, "loss": 0.0456, "step": 7200 }, { "grad_norm": 0.44921472668647766, "learning_rate": 9.956074663956135e-05, "loss": 0.0403, "step": 7210 }, { "grad_norm": 0.4754917621612549, "learning_rate": 9.955800878688726e-05, "loss": 0.0584, "step": 7220 }, { "grad_norm": 0.39969485998153687, "learning_rate": 9.955526246610489e-05, "loss": 0.0473, "step": 7230 }, { "grad_norm": 0.4229693114757538, "learning_rate": 9.955250767768349e-05, "loss": 0.0441, "step": 7240 }, { "grad_norm": 0.4558241367340088, "learning_rate": 9.95497444220938e-05, "loss": 0.0408, "step": 7250 }, { "grad_norm": 0.35203254222869873, "learning_rate": 9.954697269980797e-05, "loss": 0.0441, "step": 7260 }, { "grad_norm": 0.33152034878730774, "learning_rate": 9.954419251129962e-05, "loss": 0.0399, "step": 7270 }, { "grad_norm": 0.25496384501457214, "learning_rate": 9.95414038570438e-05, "loss": 0.0375, "step": 7280 }, { "grad_norm": 0.34752437472343445, "learning_rate": 9.953860673751703e-05, "loss": 0.0331, "step": 7290 }, { "grad_norm": 0.3835650384426117, "learning_rate": 9.953580115319725e-05, "loss": 0.0401, "step": 7300 }, { "grad_norm": 0.22476553916931152, "learning_rate": 9.953298710456387e-05, "loss": 0.0407, "step": 7310 }, { "grad_norm": 0.30182141065597534, "learning_rate": 9.953016459209771e-05, "loss": 0.035, "step": 7320 }, { "grad_norm": 0.2945469617843628, "learning_rate": 9.952733361628108e-05, "loss": 0.037, "step": 7330 }, { "grad_norm": 0.38660064339637756, "learning_rate": 9.952449417759772e-05, "loss": 0.0344, "step": 7340 }, { "grad_norm": 0.30430328845977783, "learning_rate": 9.952164627653279e-05, "loss": 0.0347, "step": 7350 }, { "grad_norm": 0.32091617584228516, "learning_rate": 9.951878991357292e-05, "loss": 0.0327, "step": 7360 }, { "grad_norm": 0.2756192684173584, "learning_rate": 9.951592508920622e-05, "loss": 0.0371, "step": 7370 }, { "grad_norm": 0.2622993290424347, "learning_rate": 9.951305180392219e-05, "loss": 0.0344, "step": 7380 }, { "grad_norm": 0.27111271023750305, "learning_rate": 9.951017005821178e-05, "loss": 0.0417, "step": 7390 }, { "grad_norm": 0.2987191379070282, "learning_rate": 9.95072798525674e-05, "loss": 0.0331, "step": 7400 }, { "grad_norm": 0.26892706751823425, "learning_rate": 9.950438118748293e-05, "loss": 0.0343, "step": 7410 }, { "grad_norm": 0.33188143372535706, "learning_rate": 9.950147406345366e-05, "loss": 0.0407, "step": 7420 }, { "grad_norm": 0.3683417737483978, "learning_rate": 9.949855848097635e-05, "loss": 0.0384, "step": 7430 }, { "grad_norm": 0.3701768219470978, "learning_rate": 9.949563444054916e-05, "loss": 0.0403, "step": 7440 }, { "grad_norm": 0.2561810314655304, "learning_rate": 9.949270194267178e-05, "loss": 0.0346, "step": 7450 }, { "grad_norm": 0.3299402594566345, "learning_rate": 9.948976098784526e-05, "loss": 0.0381, "step": 7460 }, { "grad_norm": 0.3086710274219513, "learning_rate": 9.948681157657213e-05, "loss": 0.0352, "step": 7470 }, { "grad_norm": 0.41160640120506287, "learning_rate": 9.948385370935638e-05, "loss": 0.0381, "step": 7480 }, { "grad_norm": 0.3819289207458496, "learning_rate": 9.94808873867034e-05, "loss": 0.0404, "step": 7490 }, { "grad_norm": 0.3236503303050995, "learning_rate": 9.947791260912009e-05, "loss": 0.0397, "step": 7500 }, { "grad_norm": 0.451627254486084, "learning_rate": 9.947492937711474e-05, "loss": 0.0354, "step": 7510 }, { "grad_norm": 0.43028661608695984, "learning_rate": 9.947193769119707e-05, "loss": 0.0402, "step": 7520 }, { "grad_norm": 0.34661442041397095, "learning_rate": 9.946893755187834e-05, "loss": 0.0442, "step": 7530 }, { "grad_norm": 0.37197718024253845, "learning_rate": 9.946592895967115e-05, "loss": 0.0343, "step": 7540 }, { "grad_norm": 0.2687303125858307, "learning_rate": 9.94629119150896e-05, "loss": 0.033, "step": 7550 }, { "grad_norm": 0.4141524136066437, "learning_rate": 9.94598864186492e-05, "loss": 0.0351, "step": 7560 }, { "grad_norm": 0.41414833068847656, "learning_rate": 9.945685247086696e-05, "loss": 0.0414, "step": 7570 }, { "grad_norm": 0.3453611433506012, "learning_rate": 9.945381007226129e-05, "loss": 0.0453, "step": 7580 }, { "grad_norm": 0.2700868844985962, "learning_rate": 9.945075922335203e-05, "loss": 0.044, "step": 7590 }, { "grad_norm": 0.575219988822937, "learning_rate": 9.944769992466049e-05, "loss": 0.0401, "step": 7600 }, { "grad_norm": 0.38385283946990967, "learning_rate": 9.944463217670945e-05, "loss": 0.0338, "step": 7610 }, { "grad_norm": 0.32480281591415405, "learning_rate": 9.944155598002307e-05, "loss": 0.0338, "step": 7620 }, { "grad_norm": 0.40061330795288086, "learning_rate": 9.943847133512701e-05, "loss": 0.0354, "step": 7630 }, { "grad_norm": 0.3616059124469757, "learning_rate": 9.943537824254834e-05, "loss": 0.0339, "step": 7640 }, { "grad_norm": 0.24759641289710999, "learning_rate": 9.943227670281559e-05, "loss": 0.0385, "step": 7650 }, { "grad_norm": 0.23583410680294037, "learning_rate": 9.942916671645873e-05, "loss": 0.0316, "step": 7660 }, { "grad_norm": 0.39507508277893066, "learning_rate": 9.942604828400916e-05, "loss": 0.04, "step": 7670 }, { "grad_norm": 0.40761932730674744, "learning_rate": 9.942292140599975e-05, "loss": 0.0377, "step": 7680 }, { "grad_norm": 0.22918328642845154, "learning_rate": 9.94197860829648e-05, "loss": 0.0386, "step": 7690 }, { "grad_norm": 0.4628041088581085, "learning_rate": 9.941664231544004e-05, "loss": 0.0369, "step": 7700 }, { "grad_norm": 0.47798171639442444, "learning_rate": 9.941349010396264e-05, "loss": 0.0377, "step": 7710 }, { "grad_norm": 0.27687785029411316, "learning_rate": 9.941032944907125e-05, "loss": 0.031, "step": 7720 }, { "grad_norm": 0.2707415521144867, "learning_rate": 9.940716035130596e-05, "loss": 0.0349, "step": 7730 }, { "grad_norm": 0.3251076638698578, "learning_rate": 9.940398281120821e-05, "loss": 0.033, "step": 7740 }, { "grad_norm": 0.3003835082054138, "learning_rate": 9.940079682932102e-05, "loss": 0.0328, "step": 7750 }, { "grad_norm": 0.2532021105289459, "learning_rate": 9.939760240618877e-05, "loss": 0.0296, "step": 7760 }, { "grad_norm": 0.23798850178718567, "learning_rate": 9.939439954235729e-05, "loss": 0.0271, "step": 7770 }, { "grad_norm": 0.2873820960521698, "learning_rate": 9.939118823837387e-05, "loss": 0.0338, "step": 7780 }, { "grad_norm": 0.3406454920768738, "learning_rate": 9.938796849478725e-05, "loss": 0.0271, "step": 7790 }, { "grad_norm": 0.31799256801605225, "learning_rate": 9.938474031214755e-05, "loss": 0.03, "step": 7800 }, { "grad_norm": 0.3144230544567108, "learning_rate": 9.938150369100643e-05, "loss": 0.0337, "step": 7810 }, { "grad_norm": 0.43177130818367004, "learning_rate": 9.93782586319169e-05, "loss": 0.0383, "step": 7820 }, { "grad_norm": 0.2726548910140991, "learning_rate": 9.937500513543348e-05, "loss": 0.048, "step": 7830 }, { "grad_norm": 0.31150153279304504, "learning_rate": 9.937174320211207e-05, "loss": 0.0366, "step": 7840 }, { "grad_norm": 0.29621267318725586, "learning_rate": 9.936847283251009e-05, "loss": 0.034, "step": 7850 }, { "grad_norm": 0.28057944774627686, "learning_rate": 9.936519402718632e-05, "loss": 0.0306, "step": 7860 }, { "grad_norm": 0.27676814794540405, "learning_rate": 9.936190678670102e-05, "loss": 0.0307, "step": 7870 }, { "grad_norm": 0.23225843906402588, "learning_rate": 9.935861111161593e-05, "loss": 0.0312, "step": 7880 }, { "grad_norm": 0.2906934916973114, "learning_rate": 9.935530700249416e-05, "loss": 0.035, "step": 7890 }, { "grad_norm": 0.2678881585597992, "learning_rate": 9.935199445990028e-05, "loss": 0.0303, "step": 7900 }, { "grad_norm": 0.4902949631214142, "learning_rate": 9.934867348440033e-05, "loss": 0.0427, "step": 7910 }, { "grad_norm": 0.2800944745540619, "learning_rate": 9.934534407656176e-05, "loss": 0.031, "step": 7920 }, { "grad_norm": 0.4261672794818878, "learning_rate": 9.93420062369535e-05, "loss": 0.033, "step": 7930 }, { "grad_norm": 0.2834247946739197, "learning_rate": 9.933865996614589e-05, "loss": 0.0367, "step": 7940 }, { "grad_norm": 0.32554763555526733, "learning_rate": 9.933530526471068e-05, "loss": 0.0404, "step": 7950 }, { "grad_norm": 0.4175547659397125, "learning_rate": 9.933194213322114e-05, "loss": 0.0449, "step": 7960 }, { "grad_norm": 0.39316925406455994, "learning_rate": 9.932857057225192e-05, "loss": 0.0395, "step": 7970 }, { "grad_norm": 0.38587918877601624, "learning_rate": 9.932519058237912e-05, "loss": 0.0414, "step": 7980 }, { "grad_norm": 0.3113575279712677, "learning_rate": 9.932180216418032e-05, "loss": 0.0369, "step": 7990 }, { "grad_norm": 0.3087592124938965, "learning_rate": 9.931840531823446e-05, "loss": 0.0402, "step": 8000 }, { "grad_norm": 0.2149409055709839, "learning_rate": 9.9315000045122e-05, "loss": 0.0411, "step": 8010 }, { "grad_norm": 0.43835538625717163, "learning_rate": 9.931158634542481e-05, "loss": 0.0335, "step": 8020 }, { "grad_norm": 0.4416879117488861, "learning_rate": 9.930816421972617e-05, "loss": 0.0388, "step": 8030 }, { "grad_norm": 0.348726749420166, "learning_rate": 9.930473366861086e-05, "loss": 0.0314, "step": 8040 }, { "grad_norm": 0.3106415867805481, "learning_rate": 9.930129469266505e-05, "loss": 0.0324, "step": 8050 }, { "grad_norm": 0.32965490221977234, "learning_rate": 9.929784729247638e-05, "loss": 0.0365, "step": 8060 }, { "grad_norm": 0.32523274421691895, "learning_rate": 9.929439146863389e-05, "loss": 0.0393, "step": 8070 }, { "grad_norm": 0.4379739463329315, "learning_rate": 9.92909272217281e-05, "loss": 0.0363, "step": 8080 }, { "grad_norm": 0.3430045247077942, "learning_rate": 9.928745455235097e-05, "loss": 0.0333, "step": 8090 }, { "grad_norm": 0.3513079285621643, "learning_rate": 9.928397346109588e-05, "loss": 0.0328, "step": 8100 }, { "grad_norm": 0.39850613474845886, "learning_rate": 9.928048394855762e-05, "loss": 0.0363, "step": 8110 }, { "grad_norm": 0.37340080738067627, "learning_rate": 9.92769860153325e-05, "loss": 0.0334, "step": 8120 }, { "grad_norm": 0.2752595841884613, "learning_rate": 9.927347966201819e-05, "loss": 0.0337, "step": 8130 }, { "grad_norm": 0.3753063678741455, "learning_rate": 9.926996488921383e-05, "loss": 0.0394, "step": 8140 }, { "grad_norm": 0.4984237551689148, "learning_rate": 9.926644169752001e-05, "loss": 0.0395, "step": 8150 }, { "grad_norm": 0.24436978995800018, "learning_rate": 9.926291008753875e-05, "loss": 0.0404, "step": 8160 }, { "grad_norm": 0.33383992314338684, "learning_rate": 9.92593700598735e-05, "loss": 0.0346, "step": 8170 }, { "grad_norm": 0.34191441535949707, "learning_rate": 9.925582161512915e-05, "loss": 0.0391, "step": 8180 }, { "grad_norm": 0.2756718099117279, "learning_rate": 9.925226475391205e-05, "loss": 0.0392, "step": 8190 }, { "grad_norm": 0.2403572052717209, "learning_rate": 9.924869947682993e-05, "loss": 0.0362, "step": 8200 }, { "grad_norm": 0.2513570487499237, "learning_rate": 9.924512578449204e-05, "loss": 0.0352, "step": 8210 }, { "grad_norm": 0.34933850169181824, "learning_rate": 9.924154367750901e-05, "loss": 0.0426, "step": 8220 }, { "grad_norm": 0.43718379735946655, "learning_rate": 9.923795315649293e-05, "loss": 0.0373, "step": 8230 }, { "grad_norm": 0.22932593524456024, "learning_rate": 9.92343542220573e-05, "loss": 0.0328, "step": 8240 }, { "grad_norm": 0.3651770353317261, "learning_rate": 9.92307468748171e-05, "loss": 0.0375, "step": 8250 }, { "grad_norm": 0.32542312145233154, "learning_rate": 9.922713111538873e-05, "loss": 0.0362, "step": 8260 }, { "grad_norm": 0.3609435558319092, "learning_rate": 9.922350694439003e-05, "loss": 0.05, "step": 8270 }, { "grad_norm": 0.3379375636577606, "learning_rate": 9.921987436244024e-05, "loss": 0.0386, "step": 8280 }, { "grad_norm": 0.4778507947921753, "learning_rate": 9.921623337016008e-05, "loss": 0.0344, "step": 8290 }, { "grad_norm": 0.29295870661735535, "learning_rate": 9.921258396817172e-05, "loss": 0.0297, "step": 8300 }, { "grad_norm": 0.2346227467060089, "learning_rate": 9.920892615709874e-05, "loss": 0.0347, "step": 8310 }, { "grad_norm": 0.3978520333766937, "learning_rate": 9.920525993756612e-05, "loss": 0.0344, "step": 8320 }, { "grad_norm": 0.349475622177124, "learning_rate": 9.920158531020036e-05, "loss": 0.0348, "step": 8330 }, { "grad_norm": 0.2939419150352478, "learning_rate": 9.919790227562933e-05, "loss": 0.0316, "step": 8340 }, { "grad_norm": 0.23334713280200958, "learning_rate": 9.919421083448237e-05, "loss": 0.0296, "step": 8350 }, { "grad_norm": 0.28831034898757935, "learning_rate": 9.919051098739022e-05, "loss": 0.0299, "step": 8360 }, { "grad_norm": 0.4128917157649994, "learning_rate": 9.918680273498514e-05, "loss": 0.031, "step": 8370 }, { "grad_norm": 0.3680189251899719, "learning_rate": 9.918308607790072e-05, "loss": 0.0353, "step": 8380 }, { "grad_norm": 0.2279357761144638, "learning_rate": 9.917936101677205e-05, "loss": 0.0362, "step": 8390 }, { "grad_norm": 0.26002979278564453, "learning_rate": 9.917562755223564e-05, "loss": 0.0301, "step": 8400 }, { "grad_norm": 0.26673340797424316, "learning_rate": 9.917188568492944e-05, "loss": 0.0307, "step": 8410 }, { "grad_norm": 0.24005113542079926, "learning_rate": 9.916813541549283e-05, "loss": 0.0283, "step": 8420 }, { "grad_norm": 0.2421724498271942, "learning_rate": 9.916437674456663e-05, "loss": 0.0284, "step": 8430 }, { "grad_norm": 0.40657755732536316, "learning_rate": 9.916060967279308e-05, "loss": 0.0352, "step": 8440 }, { "grad_norm": 0.26043441891670227, "learning_rate": 9.91568342008159e-05, "loss": 0.0318, "step": 8450 }, { "grad_norm": 0.35964271426200867, "learning_rate": 9.915305032928019e-05, "loss": 0.0327, "step": 8460 }, { "grad_norm": 0.40363553166389465, "learning_rate": 9.914925805883253e-05, "loss": 0.0318, "step": 8470 }, { "grad_norm": 0.3536289930343628, "learning_rate": 9.914545739012088e-05, "loss": 0.0487, "step": 8480 }, { "grad_norm": 0.8202237486839294, "learning_rate": 9.91416483237947e-05, "loss": 0.0422, "step": 8490 }, { "grad_norm": 0.32060736417770386, "learning_rate": 9.913783086050485e-05, "loss": 0.0367, "step": 8500 }, { "grad_norm": 0.3770695626735687, "learning_rate": 9.913400500090364e-05, "loss": 0.0399, "step": 8510 }, { "grad_norm": 0.3464851677417755, "learning_rate": 9.913017074564479e-05, "loss": 0.0345, "step": 8520 }, { "grad_norm": 0.32078129053115845, "learning_rate": 9.912632809538348e-05, "loss": 0.0326, "step": 8530 }, { "grad_norm": 0.3177381455898285, "learning_rate": 9.912247705077629e-05, "loss": 0.0318, "step": 8540 }, { "grad_norm": 0.27169474959373474, "learning_rate": 9.911861761248127e-05, "loss": 0.0374, "step": 8550 }, { "grad_norm": 0.27000513672828674, "learning_rate": 9.91147497811579e-05, "loss": 0.0359, "step": 8560 }, { "grad_norm": 0.3137664198875427, "learning_rate": 9.911087355746709e-05, "loss": 0.0332, "step": 8570 }, { "grad_norm": 0.23727527260780334, "learning_rate": 9.910698894207117e-05, "loss": 0.0278, "step": 8580 }, { "grad_norm": 0.2668800950050354, "learning_rate": 9.910309593563392e-05, "loss": 0.0301, "step": 8590 }, { "grad_norm": 0.25694745779037476, "learning_rate": 9.909919453882057e-05, "loss": 0.03, "step": 8600 }, { "grad_norm": 0.2653639018535614, "learning_rate": 9.90952847522977e-05, "loss": 0.0322, "step": 8610 }, { "grad_norm": 0.2605046033859253, "learning_rate": 9.909136657673346e-05, "loss": 0.0353, "step": 8620 }, { "grad_norm": 0.3244588375091553, "learning_rate": 9.908744001279731e-05, "loss": 0.0325, "step": 8630 }, { "grad_norm": 0.2799627184867859, "learning_rate": 9.90835050611602e-05, "loss": 0.0337, "step": 8640 }, { "grad_norm": 0.3005102574825287, "learning_rate": 9.90795617224945e-05, "loss": 0.0375, "step": 8650 }, { "grad_norm": 0.4073898494243622, "learning_rate": 9.907560999747405e-05, "loss": 0.0313, "step": 8660 }, { "grad_norm": 0.36809977889060974, "learning_rate": 9.907164988677408e-05, "loss": 0.0331, "step": 8670 }, { "grad_norm": 0.29263585805892944, "learning_rate": 9.906768139107124e-05, "loss": 0.039, "step": 8680 }, { "grad_norm": 0.3666236400604248, "learning_rate": 9.906370451104367e-05, "loss": 0.0327, "step": 8690 }, { "grad_norm": 0.29641276597976685, "learning_rate": 9.905971924737088e-05, "loss": 0.0317, "step": 8700 }, { "grad_norm": 0.30303412675857544, "learning_rate": 9.905572560073387e-05, "loss": 0.0308, "step": 8710 }, { "grad_norm": 0.2979617714881897, "learning_rate": 9.905172357181501e-05, "loss": 0.0336, "step": 8720 }, { "grad_norm": 0.4379200339317322, "learning_rate": 9.904771316129817e-05, "loss": 0.0346, "step": 8730 }, { "grad_norm": 0.29338833689689636, "learning_rate": 9.904369436986862e-05, "loss": 0.0325, "step": 8740 }, { "grad_norm": 0.30517202615737915, "learning_rate": 9.903966719821303e-05, "loss": 0.0324, "step": 8750 }, { "grad_norm": 0.35514140129089355, "learning_rate": 9.903563164701956e-05, "loss": 0.0377, "step": 8760 }, { "grad_norm": 0.3890049457550049, "learning_rate": 9.903158771697778e-05, "loss": 0.0306, "step": 8770 }, { "grad_norm": 0.3082255721092224, "learning_rate": 9.902753540877867e-05, "loss": 0.0372, "step": 8780 }, { "grad_norm": 0.2941286265850067, "learning_rate": 9.902347472311466e-05, "loss": 0.0351, "step": 8790 }, { "grad_norm": 0.32431310415267944, "learning_rate": 9.901940566067962e-05, "loss": 0.0348, "step": 8800 }, { "grad_norm": 0.29100555181503296, "learning_rate": 9.901532822216883e-05, "loss": 0.0315, "step": 8810 }, { "grad_norm": 0.3088713586330414, "learning_rate": 9.901124240827904e-05, "loss": 0.037, "step": 8820 }, { "grad_norm": 0.39252904057502747, "learning_rate": 9.900714821970835e-05, "loss": 0.034, "step": 8830 }, { "grad_norm": 0.34553468227386475, "learning_rate": 9.900304565715641e-05, "loss": 0.0301, "step": 8840 }, { "grad_norm": 0.2642535865306854, "learning_rate": 9.899893472132419e-05, "loss": 0.0327, "step": 8850 }, { "grad_norm": 0.288168340921402, "learning_rate": 9.899481541291415e-05, "loss": 0.0416, "step": 8860 }, { "grad_norm": 0.21529342234134674, "learning_rate": 9.899068773263016e-05, "loss": 0.031, "step": 8870 }, { "grad_norm": 0.2294350117444992, "learning_rate": 9.898655168117754e-05, "loss": 0.0333, "step": 8880 }, { "grad_norm": 0.2333253026008606, "learning_rate": 9.898240725926302e-05, "loss": 0.0269, "step": 8890 }, { "grad_norm": 0.5516211986541748, "learning_rate": 9.897825446759478e-05, "loss": 0.0351, "step": 8900 }, { "grad_norm": 0.2619836628437042, "learning_rate": 9.897409330688241e-05, "loss": 0.032, "step": 8910 }, { "grad_norm": 0.2825385630130768, "learning_rate": 9.896992377783692e-05, "loss": 0.0328, "step": 8920 }, { "grad_norm": 0.2182321846485138, "learning_rate": 9.89657458811708e-05, "loss": 0.0284, "step": 8930 }, { "grad_norm": 0.28471556305885315, "learning_rate": 9.896155961759792e-05, "loss": 0.0309, "step": 8940 }, { "grad_norm": 0.24628183245658875, "learning_rate": 9.895736498783361e-05, "loss": 0.0316, "step": 8950 }, { "grad_norm": 0.2700500786304474, "learning_rate": 9.895316199259462e-05, "loss": 0.0342, "step": 8960 }, { "grad_norm": 0.2792443037033081, "learning_rate": 9.894895063259909e-05, "loss": 0.0281, "step": 8970 }, { "grad_norm": 0.577218234539032, "learning_rate": 9.894473090856667e-05, "loss": 0.0334, "step": 8980 }, { "grad_norm": 0.29739195108413696, "learning_rate": 9.894050282121839e-05, "loss": 0.0319, "step": 8990 }, { "grad_norm": 0.2696210741996765, "learning_rate": 9.893626637127668e-05, "loss": 0.04, "step": 9000 }, { "grad_norm": 0.2181677371263504, "learning_rate": 9.893202155946546e-05, "loss": 0.0298, "step": 9010 }, { "grad_norm": 0.37576788663864136, "learning_rate": 9.892776838651006e-05, "loss": 0.0313, "step": 9020 }, { "grad_norm": 0.3952059745788574, "learning_rate": 9.892350685313722e-05, "loss": 0.0318, "step": 9030 }, { "grad_norm": 0.2298465073108673, "learning_rate": 9.891923696007513e-05, "loss": 0.0312, "step": 9040 }, { "grad_norm": 0.2515447735786438, "learning_rate": 9.891495870805336e-05, "loss": 0.0282, "step": 9050 }, { "grad_norm": 0.3489874303340912, "learning_rate": 9.891067209780298e-05, "loss": 0.034, "step": 9060 }, { "grad_norm": 0.3867812156677246, "learning_rate": 9.890637713005646e-05, "loss": 0.0339, "step": 9070 }, { "grad_norm": 0.3505154848098755, "learning_rate": 9.890207380554767e-05, "loss": 0.0334, "step": 9080 }, { "grad_norm": 0.36440080404281616, "learning_rate": 9.889776212501196e-05, "loss": 0.0327, "step": 9090 }, { "grad_norm": 0.34296301007270813, "learning_rate": 9.889344208918605e-05, "loss": 0.0313, "step": 9100 }, { "grad_norm": 0.32208749651908875, "learning_rate": 9.888911369880812e-05, "loss": 0.033, "step": 9110 }, { "grad_norm": 0.2785504460334778, "learning_rate": 9.888477695461777e-05, "loss": 0.0326, "step": 9120 }, { "grad_norm": 0.2885165214538574, "learning_rate": 9.888043185735607e-05, "loss": 0.0276, "step": 9130 }, { "grad_norm": 0.2292015254497528, "learning_rate": 9.887607840776542e-05, "loss": 0.0331, "step": 9140 }, { "grad_norm": 0.3258777856826782, "learning_rate": 9.887171660658975e-05, "loss": 0.0356, "step": 9150 }, { "grad_norm": 0.2763221859931946, "learning_rate": 9.886734645457435e-05, "loss": 0.0317, "step": 9160 }, { "grad_norm": 0.33391091227531433, "learning_rate": 9.886296795246597e-05, "loss": 0.032, "step": 9170 }, { "grad_norm": 0.3411232829093933, "learning_rate": 9.885858110101276e-05, "loss": 0.0311, "step": 9180 }, { "grad_norm": 0.21126164495944977, "learning_rate": 9.885418590096434e-05, "loss": 0.0236, "step": 9190 }, { "grad_norm": 0.32119306921958923, "learning_rate": 9.88497823530717e-05, "loss": 0.0359, "step": 9200 }, { "grad_norm": 0.2914391756057739, "learning_rate": 9.884537045808732e-05, "loss": 0.0359, "step": 9210 }, { "grad_norm": 0.47597816586494446, "learning_rate": 9.884095021676502e-05, "loss": 0.0449, "step": 9220 }, { "grad_norm": 0.3806690573692322, "learning_rate": 9.883652162986017e-05, "loss": 0.048, "step": 9230 }, { "grad_norm": 0.32294365763664246, "learning_rate": 9.883208469812943e-05, "loss": 0.041, "step": 9240 }, { "grad_norm": 0.39550766348838806, "learning_rate": 9.882763942233098e-05, "loss": 0.0488, "step": 9250 }, { "grad_norm": 0.6173681020736694, "learning_rate": 9.882318580322441e-05, "loss": 0.0433, "step": 9260 }, { "grad_norm": 0.6025459170341492, "learning_rate": 9.881872384157067e-05, "loss": 0.0408, "step": 9270 }, { "grad_norm": 0.41118451952934265, "learning_rate": 9.881425353813225e-05, "loss": 0.0384, "step": 9280 }, { "grad_norm": 0.6367096900939941, "learning_rate": 9.880977489367296e-05, "loss": 0.0432, "step": 9290 }, { "grad_norm": 0.29893529415130615, "learning_rate": 9.88052879089581e-05, "loss": 0.0465, "step": 9300 }, { "grad_norm": 0.34736815094947815, "learning_rate": 9.880079258475434e-05, "loss": 0.0384, "step": 9310 }, { "grad_norm": 0.42013972997665405, "learning_rate": 9.879628892182985e-05, "loss": 0.0377, "step": 9320 }, { "grad_norm": 0.4000377058982849, "learning_rate": 9.879177692095416e-05, "loss": 0.0423, "step": 9330 }, { "grad_norm": 0.37932008504867554, "learning_rate": 9.878725658289825e-05, "loss": 0.0376, "step": 9340 }, { "grad_norm": 0.25054123997688293, "learning_rate": 9.878272790843454e-05, "loss": 0.0295, "step": 9350 }, { "grad_norm": 0.3356637954711914, "learning_rate": 9.877819089833682e-05, "loss": 0.0372, "step": 9360 }, { "grad_norm": 0.27055492997169495, "learning_rate": 9.877364555338038e-05, "loss": 0.0368, "step": 9370 }, { "grad_norm": 0.3326558470726013, "learning_rate": 9.876909187434186e-05, "loss": 0.0484, "step": 9380 }, { "grad_norm": 0.4081169366836548, "learning_rate": 9.876452986199939e-05, "loss": 0.0465, "step": 9390 }, { "grad_norm": 0.6230624318122864, "learning_rate": 9.875995951713248e-05, "loss": 0.0392, "step": 9400 }, { "grad_norm": 0.2787381410598755, "learning_rate": 9.875538084052207e-05, "loss": 0.0374, "step": 9410 }, { "grad_norm": 0.381122350692749, "learning_rate": 9.875079383295053e-05, "loss": 0.0332, "step": 9420 }, { "grad_norm": 0.31308087706565857, "learning_rate": 9.874619849520167e-05, "loss": 0.0324, "step": 9430 }, { "grad_norm": 0.4457867443561554, "learning_rate": 9.874159482806069e-05, "loss": 0.0422, "step": 9440 }, { "grad_norm": 0.27630963921546936, "learning_rate": 9.873698283231426e-05, "loss": 0.0327, "step": 9450 }, { "grad_norm": 0.4598402678966522, "learning_rate": 9.87323625087504e-05, "loss": 0.0338, "step": 9460 }, { "grad_norm": 0.30069947242736816, "learning_rate": 9.872773385815863e-05, "loss": 0.0429, "step": 9470 }, { "grad_norm": 0.2842298746109009, "learning_rate": 9.872309688132986e-05, "loss": 0.0442, "step": 9480 }, { "grad_norm": 0.3976864516735077, "learning_rate": 9.871845157905639e-05, "loss": 0.0468, "step": 9490 }, { "grad_norm": 0.34357255697250366, "learning_rate": 9.871379795213201e-05, "loss": 0.0434, "step": 9500 }, { "grad_norm": 1.0675992965698242, "learning_rate": 9.87091360013519e-05, "loss": 0.0508, "step": 9510 }, { "grad_norm": 0.6012641191482544, "learning_rate": 9.870446572751262e-05, "loss": 0.0501, "step": 9520 }, { "grad_norm": 0.4618903398513794, "learning_rate": 9.869978713141224e-05, "loss": 0.0488, "step": 9530 }, { "grad_norm": 0.45645591616630554, "learning_rate": 9.869510021385016e-05, "loss": 0.0472, "step": 9540 }, { "grad_norm": 0.3386933505535126, "learning_rate": 9.869040497562727e-05, "loss": 0.0449, "step": 9550 }, { "grad_norm": 0.341007262468338, "learning_rate": 9.868570141754587e-05, "loss": 0.0423, "step": 9560 }, { "grad_norm": 0.3424454629421234, "learning_rate": 9.868098954040965e-05, "loss": 0.0383, "step": 9570 }, { "grad_norm": 0.33508726954460144, "learning_rate": 9.867626934502374e-05, "loss": 0.0417, "step": 9580 }, { "grad_norm": 0.29727041721343994, "learning_rate": 9.86715408321947e-05, "loss": 0.0279, "step": 9590 }, { "grad_norm": 0.4235117435455322, "learning_rate": 9.86668040027305e-05, "loss": 0.0453, "step": 9600 }, { "grad_norm": 0.4473966658115387, "learning_rate": 9.866205885744053e-05, "loss": 0.0429, "step": 9610 }, { "grad_norm": 0.44303038716316223, "learning_rate": 9.865730539713563e-05, "loss": 0.0444, "step": 9620 }, { "grad_norm": 0.32018107175827026, "learning_rate": 9.8652543622628e-05, "loss": 0.0437, "step": 9630 }, { "grad_norm": 0.3434239327907562, "learning_rate": 9.864777353473132e-05, "loss": 0.0359, "step": 9640 }, { "grad_norm": 0.4276004433631897, "learning_rate": 9.864299513426068e-05, "loss": 0.0425, "step": 9650 }, { "grad_norm": 0.3734433650970459, "learning_rate": 9.863820842203254e-05, "loss": 0.0321, "step": 9660 }, { "grad_norm": 0.2906720042228699, "learning_rate": 9.863341339886483e-05, "loss": 0.0333, "step": 9670 }, { "grad_norm": 0.4694768786430359, "learning_rate": 9.86286100655769e-05, "loss": 0.04, "step": 9680 }, { "grad_norm": 0.28952717781066895, "learning_rate": 9.862379842298953e-05, "loss": 0.0388, "step": 9690 }, { "grad_norm": 0.177528515458107, "learning_rate": 9.861897847192485e-05, "loss": 0.0344, "step": 9700 }, { "grad_norm": 0.2721542716026306, "learning_rate": 9.86141502132065e-05, "loss": 0.0351, "step": 9710 }, { "grad_norm": 0.4153067469596863, "learning_rate": 9.860931364765946e-05, "loss": 0.0351, "step": 9720 }, { "grad_norm": 0.24094222486019135, "learning_rate": 9.860446877611021e-05, "loss": 0.0325, "step": 9730 }, { "grad_norm": 0.31710779666900635, "learning_rate": 9.859961559938655e-05, "loss": 0.0382, "step": 9740 }, { "grad_norm": 0.28319206833839417, "learning_rate": 9.85947541183178e-05, "loss": 0.0349, "step": 9750 }, { "grad_norm": 0.27591732144355774, "learning_rate": 9.858988433373463e-05, "loss": 0.0299, "step": 9760 }, { "grad_norm": 0.34469929337501526, "learning_rate": 9.858500624646918e-05, "loss": 0.0301, "step": 9770 }, { "grad_norm": 0.24903236329555511, "learning_rate": 9.858011985735497e-05, "loss": 0.0312, "step": 9780 }, { "grad_norm": 0.3282131552696228, "learning_rate": 9.857522516722693e-05, "loss": 0.0374, "step": 9790 }, { "grad_norm": 0.3527752757072449, "learning_rate": 9.857032217692145e-05, "loss": 0.0293, "step": 9800 }, { "grad_norm": 0.3369750380516052, "learning_rate": 9.856541088727631e-05, "loss": 0.0301, "step": 9810 }, { "grad_norm": 0.39671987295150757, "learning_rate": 9.856049129913072e-05, "loss": 0.0315, "step": 9820 }, { "grad_norm": 0.3592272996902466, "learning_rate": 9.85555634133253e-05, "loss": 0.0431, "step": 9830 }, { "grad_norm": 0.2916005253791809, "learning_rate": 9.855062723070208e-05, "loss": 0.0343, "step": 9840 }, { "grad_norm": 0.30579954385757446, "learning_rate": 9.854568275210454e-05, "loss": 0.0399, "step": 9850 }, { "grad_norm": 0.30766379833221436, "learning_rate": 9.854072997837754e-05, "loss": 0.0366, "step": 9860 }, { "grad_norm": 0.3130258619785309, "learning_rate": 9.853576891036737e-05, "loss": 0.0309, "step": 9870 }, { "grad_norm": 0.20744241774082184, "learning_rate": 9.853079954892177e-05, "loss": 0.0308, "step": 9880 }, { "grad_norm": 0.25589218735694885, "learning_rate": 9.852582189488983e-05, "loss": 0.0391, "step": 9890 }, { "grad_norm": 0.29537326097488403, "learning_rate": 9.852083594912212e-05, "loss": 0.0377, "step": 9900 }, { "grad_norm": 0.3098631203174591, "learning_rate": 9.851584171247058e-05, "loss": 0.0357, "step": 9910 }, { "grad_norm": 0.3342215418815613, "learning_rate": 9.851083918578863e-05, "loss": 0.0383, "step": 9920 }, { "grad_norm": 0.2670917212963104, "learning_rate": 9.850582836993103e-05, "loss": 0.0293, "step": 9930 }, { "grad_norm": 0.39509308338165283, "learning_rate": 9.850080926575397e-05, "loss": 0.0367, "step": 9940 }, { "grad_norm": 0.2593704164028168, "learning_rate": 9.849578187411515e-05, "loss": 0.0378, "step": 9950 }, { "grad_norm": 0.37677618861198425, "learning_rate": 9.849074619587354e-05, "loss": 0.0358, "step": 9960 }, { "grad_norm": 0.26485031843185425, "learning_rate": 9.848570223188964e-05, "loss": 0.0331, "step": 9970 }, { "grad_norm": 0.28589510917663574, "learning_rate": 9.848064998302531e-05, "loss": 0.0356, "step": 9980 }, { "grad_norm": 0.25352728366851807, "learning_rate": 9.847558945014386e-05, "loss": 0.0351, "step": 9990 }, { "grad_norm": 0.2835133671760559, "learning_rate": 9.847052063410996e-05, "loss": 0.0337, "step": 10000 }, { "grad_norm": 0.2192440778017044, "learning_rate": 9.846544353578977e-05, "loss": 0.0314, "step": 10010 }, { "grad_norm": 0.26591914892196655, "learning_rate": 9.846035815605081e-05, "loss": 0.0301, "step": 10020 }, { "grad_norm": 0.28409749269485474, "learning_rate": 9.845526449576204e-05, "loss": 0.0314, "step": 10030 }, { "grad_norm": 0.31617850065231323, "learning_rate": 9.845016255579383e-05, "loss": 0.0338, "step": 10040 }, { "grad_norm": 0.284469336271286, "learning_rate": 9.844505233701794e-05, "loss": 0.0296, "step": 10050 }, { "grad_norm": 0.2623935639858246, "learning_rate": 9.843993384030757e-05, "loss": 0.0304, "step": 10060 }, { "grad_norm": 0.3327840566635132, "learning_rate": 9.843480706653737e-05, "loss": 0.0315, "step": 10070 }, { "grad_norm": 0.2660188376903534, "learning_rate": 9.84296720165833e-05, "loss": 0.0311, "step": 10080 }, { "grad_norm": 0.26006194949150085, "learning_rate": 9.842452869132286e-05, "loss": 0.0292, "step": 10090 }, { "grad_norm": 0.45546358823776245, "learning_rate": 9.841937709163489e-05, "loss": 0.0314, "step": 10100 }, { "grad_norm": 0.2835926413536072, "learning_rate": 9.841421721839962e-05, "loss": 0.0285, "step": 10110 }, { "grad_norm": 0.2647816240787506, "learning_rate": 9.840904907249879e-05, "loss": 0.0358, "step": 10120 }, { "grad_norm": 0.22858856618404388, "learning_rate": 9.840387265481545e-05, "loss": 0.0296, "step": 10130 }, { "grad_norm": 0.2855270206928253, "learning_rate": 9.839868796623411e-05, "loss": 0.0309, "step": 10140 }, { "grad_norm": 0.2349906712770462, "learning_rate": 9.839349500764072e-05, "loss": 0.0277, "step": 10150 }, { "grad_norm": 0.3040671646595001, "learning_rate": 9.83882937799226e-05, "loss": 0.0304, "step": 10160 }, { "grad_norm": 0.2410794198513031, "learning_rate": 9.838308428396849e-05, "loss": 0.0343, "step": 10170 }, { "grad_norm": 0.22997114062309265, "learning_rate": 9.837786652066854e-05, "loss": 0.0306, "step": 10180 }, { "grad_norm": 0.3126216530799866, "learning_rate": 9.837264049091437e-05, "loss": 0.0288, "step": 10190 }, { "grad_norm": 0.20954976975917816, "learning_rate": 9.836740619559893e-05, "loss": 0.0331, "step": 10200 }, { "grad_norm": 0.41519296169281006, "learning_rate": 9.836216363561659e-05, "loss": 0.0382, "step": 10210 }, { "grad_norm": 0.3067454397678375, "learning_rate": 9.835691281186322e-05, "loss": 0.0348, "step": 10220 }, { "grad_norm": 0.2636464536190033, "learning_rate": 9.8351653725236e-05, "loss": 0.0319, "step": 10230 }, { "grad_norm": 0.3039206266403198, "learning_rate": 9.83463863766336e-05, "loss": 0.0318, "step": 10240 }, { "grad_norm": 0.38088303804397583, "learning_rate": 9.834111076695602e-05, "loss": 0.0414, "step": 10250 }, { "grad_norm": 0.3452463746070862, "learning_rate": 9.833582689710477e-05, "loss": 0.035, "step": 10260 }, { "grad_norm": 0.38169923424720764, "learning_rate": 9.833053476798268e-05, "loss": 0.0339, "step": 10270 }, { "grad_norm": 0.3345685601234436, "learning_rate": 9.832523438049404e-05, "loss": 0.0404, "step": 10280 }, { "grad_norm": 0.2578211724758148, "learning_rate": 9.831992573554454e-05, "loss": 0.0352, "step": 10290 }, { "grad_norm": 0.4189004898071289, "learning_rate": 9.831460883404128e-05, "loss": 0.0329, "step": 10300 }, { "grad_norm": 0.19545909762382507, "learning_rate": 9.830928367689278e-05, "loss": 0.0321, "step": 10310 }, { "grad_norm": 0.3239494562149048, "learning_rate": 9.830395026500896e-05, "loss": 0.0325, "step": 10320 }, { "grad_norm": 0.21674814820289612, "learning_rate": 9.829860859930115e-05, "loss": 0.0294, "step": 10330 }, { "grad_norm": 0.24569949507713318, "learning_rate": 9.829325868068212e-05, "loss": 0.0321, "step": 10340 }, { "grad_norm": 0.2988359034061432, "learning_rate": 9.8287900510066e-05, "loss": 0.0351, "step": 10350 }, { "grad_norm": 0.24910788238048553, "learning_rate": 9.828253408836834e-05, "loss": 0.0257, "step": 10360 }, { "grad_norm": 0.3345392644405365, "learning_rate": 9.827715941650615e-05, "loss": 0.0357, "step": 10370 }, { "grad_norm": 0.38457542657852173, "learning_rate": 9.82717764953978e-05, "loss": 0.0325, "step": 10380 }, { "grad_norm": 0.31023654341697693, "learning_rate": 9.826638532596308e-05, "loss": 0.0307, "step": 10390 }, { "grad_norm": 0.2882497012615204, "learning_rate": 9.82609859091232e-05, "loss": 0.0317, "step": 10400 }, { "grad_norm": 0.26852771639823914, "learning_rate": 9.825557824580076e-05, "loss": 0.032, "step": 10410 }, { "grad_norm": 0.23426155745983124, "learning_rate": 9.82501623369198e-05, "loss": 0.0306, "step": 10420 }, { "grad_norm": 0.1986299753189087, "learning_rate": 9.824473818340574e-05, "loss": 0.0293, "step": 10430 }, { "grad_norm": 0.2029925435781479, "learning_rate": 9.823930578618541e-05, "loss": 0.0276, "step": 10440 }, { "grad_norm": 0.22095704078674316, "learning_rate": 9.823386514618709e-05, "loss": 0.0317, "step": 10450 }, { "grad_norm": 0.2631254196166992, "learning_rate": 9.82284162643404e-05, "loss": 0.0285, "step": 10460 }, { "grad_norm": 0.23297441005706787, "learning_rate": 9.822295914157642e-05, "loss": 0.0323, "step": 10470 }, { "grad_norm": 0.26262736320495605, "learning_rate": 9.821749377882763e-05, "loss": 0.0279, "step": 10480 }, { "grad_norm": 0.3062179982662201, "learning_rate": 9.821202017702791e-05, "loss": 0.0335, "step": 10490 }, { "grad_norm": 0.2880789041519165, "learning_rate": 9.820653833711253e-05, "loss": 0.0318, "step": 10500 }, { "grad_norm": 0.2144428938627243, "learning_rate": 9.820104826001822e-05, "loss": 0.0345, "step": 10510 }, { "grad_norm": 0.37298059463500977, "learning_rate": 9.819554994668305e-05, "loss": 0.0355, "step": 10520 }, { "grad_norm": 0.2689935266971588, "learning_rate": 9.819004339804654e-05, "loss": 0.0346, "step": 10530 }, { "grad_norm": 0.30581676959991455, "learning_rate": 9.818452861504961e-05, "loss": 0.0295, "step": 10540 }, { "grad_norm": 0.321038156747818, "learning_rate": 9.81790055986346e-05, "loss": 0.0292, "step": 10550 }, { "grad_norm": 0.2494737058877945, "learning_rate": 9.817347434974523e-05, "loss": 0.0333, "step": 10560 }, { "grad_norm": 0.63808274269104, "learning_rate": 9.816793486932664e-05, "loss": 0.0397, "step": 10570 }, { "grad_norm": 0.565896213054657, "learning_rate": 9.816238715832538e-05, "loss": 0.0447, "step": 10580 }, { "grad_norm": 0.3128828704357147, "learning_rate": 9.815683121768939e-05, "loss": 0.0333, "step": 10590 }, { "grad_norm": 0.361129492521286, "learning_rate": 9.815126704836804e-05, "loss": 0.0395, "step": 10600 }, { "grad_norm": 0.2364974468946457, "learning_rate": 9.81456946513121e-05, "loss": 0.0355, "step": 10610 }, { "grad_norm": 0.3076198101043701, "learning_rate": 9.814011402747373e-05, "loss": 0.0322, "step": 10620 }, { "grad_norm": 0.2510896325111389, "learning_rate": 9.813452517780651e-05, "loss": 0.0335, "step": 10630 }, { "grad_norm": 0.3160039186477661, "learning_rate": 9.81289281032654e-05, "loss": 0.0338, "step": 10640 }, { "grad_norm": 0.3054080903530121, "learning_rate": 9.812332280480683e-05, "loss": 0.0278, "step": 10650 }, { "grad_norm": 0.3210267424583435, "learning_rate": 9.811770928338854e-05, "loss": 0.0299, "step": 10660 }, { "grad_norm": 0.2249358594417572, "learning_rate": 9.811208753996979e-05, "loss": 0.0307, "step": 10670 }, { "grad_norm": 0.28639349341392517, "learning_rate": 9.810645757551113e-05, "loss": 0.0318, "step": 10680 }, { "grad_norm": 0.27169865369796753, "learning_rate": 9.810081939097459e-05, "loss": 0.0259, "step": 10690 }, { "grad_norm": 0.27313706278800964, "learning_rate": 9.809517298732356e-05, "loss": 0.0282, "step": 10700 }, { "grad_norm": 0.36891376972198486, "learning_rate": 9.80895183655229e-05, "loss": 0.0306, "step": 10710 }, { "grad_norm": 0.45647192001342773, "learning_rate": 9.808385552653877e-05, "loss": 0.0321, "step": 10720 }, { "grad_norm": 0.5423809885978699, "learning_rate": 9.807818447133886e-05, "loss": 0.0307, "step": 10730 }, { "grad_norm": 0.31897205114364624, "learning_rate": 9.807250520089215e-05, "loss": 0.0304, "step": 10740 }, { "grad_norm": 0.3268202543258667, "learning_rate": 9.806681771616908e-05, "loss": 0.0282, "step": 10750 }, { "grad_norm": 0.3271597623825073, "learning_rate": 9.80611220181415e-05, "loss": 0.0375, "step": 10760 }, { "grad_norm": 0.2737480103969574, "learning_rate": 9.805541810778264e-05, "loss": 0.0322, "step": 10770 }, { "grad_norm": 0.3337687849998474, "learning_rate": 9.804970598606716e-05, "loss": 0.0291, "step": 10780 }, { "grad_norm": 0.2721922695636749, "learning_rate": 9.804398565397106e-05, "loss": 0.0316, "step": 10790 }, { "grad_norm": 0.3151012659072876, "learning_rate": 9.803825711247183e-05, "loss": 0.0306, "step": 10800 }, { "grad_norm": 0.2216194123029709, "learning_rate": 9.803252036254831e-05, "loss": 0.033, "step": 10810 }, { "grad_norm": 0.27712729573249817, "learning_rate": 9.802677540518076e-05, "loss": 0.0374, "step": 10820 }, { "grad_norm": 0.2649964392185211, "learning_rate": 9.802102224135081e-05, "loss": 0.0322, "step": 10830 }, { "grad_norm": 0.26435554027557373, "learning_rate": 9.801526087204155e-05, "loss": 0.0298, "step": 10840 }, { "grad_norm": 0.35966476798057556, "learning_rate": 9.800949129823743e-05, "loss": 0.0311, "step": 10850 }, { "grad_norm": 0.3450068235397339, "learning_rate": 9.80037135209243e-05, "loss": 0.0374, "step": 10860 }, { "grad_norm": 0.25566133856773376, "learning_rate": 9.799792754108946e-05, "loss": 0.0319, "step": 10870 }, { "grad_norm": 0.24255894124507904, "learning_rate": 9.799213335972152e-05, "loss": 0.0296, "step": 10880 }, { "grad_norm": 0.2733111083507538, "learning_rate": 9.798633097781058e-05, "loss": 0.0285, "step": 10890 }, { "grad_norm": 0.2846207022666931, "learning_rate": 9.79805203963481e-05, "loss": 0.0299, "step": 10900 }, { "grad_norm": 0.257918119430542, "learning_rate": 9.797470161632697e-05, "loss": 0.0265, "step": 10910 }, { "grad_norm": 0.28876540064811707, "learning_rate": 9.796887463874145e-05, "loss": 0.0231, "step": 10920 }, { "grad_norm": 0.293607234954834, "learning_rate": 9.796303946458718e-05, "loss": 0.0295, "step": 10930 }, { "grad_norm": 0.224348783493042, "learning_rate": 9.795719609486127e-05, "loss": 0.029, "step": 10940 }, { "grad_norm": 0.32155805826187134, "learning_rate": 9.795134453056219e-05, "loss": 0.0278, "step": 10950 }, { "grad_norm": 0.2509315013885498, "learning_rate": 9.794548477268979e-05, "loss": 0.0266, "step": 10960 }, { "grad_norm": 0.22646242380142212, "learning_rate": 9.793961682224537e-05, "loss": 0.0264, "step": 10970 }, { "grad_norm": 0.26088809967041016, "learning_rate": 9.793374068023156e-05, "loss": 0.0277, "step": 10980 }, { "grad_norm": 0.20703370869159698, "learning_rate": 9.792785634765247e-05, "loss": 0.0275, "step": 10990 }, { "grad_norm": 0.3148171007633209, "learning_rate": 9.792196382551357e-05, "loss": 0.0355, "step": 11000 }, { "grad_norm": 0.35040226578712463, "learning_rate": 9.791606311482171e-05, "loss": 0.0258, "step": 11010 }, { "grad_norm": 0.2738809585571289, "learning_rate": 9.791015421658518e-05, "loss": 0.0333, "step": 11020 }, { "grad_norm": 0.3234941363334656, "learning_rate": 9.790423713181362e-05, "loss": 0.0384, "step": 11030 }, { "grad_norm": 0.2273460179567337, "learning_rate": 9.789831186151814e-05, "loss": 0.0293, "step": 11040 }, { "grad_norm": 0.3131672739982605, "learning_rate": 9.789237840671118e-05, "loss": 0.0305, "step": 11050 }, { "grad_norm": 0.9604679942131042, "learning_rate": 9.78864367684066e-05, "loss": 0.0252, "step": 11060 }, { "grad_norm": 0.4421369433403015, "learning_rate": 9.788048694761968e-05, "loss": 0.0366, "step": 11070 }, { "grad_norm": 0.36801809072494507, "learning_rate": 9.787452894536709e-05, "loss": 0.0364, "step": 11080 }, { "grad_norm": 0.3059307336807251, "learning_rate": 9.786856276266685e-05, "loss": 0.0326, "step": 11090 }, { "grad_norm": 0.24067875742912292, "learning_rate": 9.786258840053845e-05, "loss": 0.0295, "step": 11100 }, { "grad_norm": 0.25058868527412415, "learning_rate": 9.785660586000273e-05, "loss": 0.0361, "step": 11110 }, { "grad_norm": 0.2839607298374176, "learning_rate": 9.785061514208196e-05, "loss": 0.0316, "step": 11120 }, { "grad_norm": 0.24126982688903809, "learning_rate": 9.784461624779977e-05, "loss": 0.0275, "step": 11130 }, { "grad_norm": 0.2251640260219574, "learning_rate": 9.783860917818123e-05, "loss": 0.0279, "step": 11140 }, { "grad_norm": 0.4134131968021393, "learning_rate": 9.783259393425277e-05, "loss": 0.0367, "step": 11150 }, { "grad_norm": 0.6557153463363647, "learning_rate": 9.782657051704221e-05, "loss": 0.034, "step": 11160 }, { "grad_norm": 0.28896793723106384, "learning_rate": 9.782053892757883e-05, "loss": 0.0325, "step": 11170 }, { "grad_norm": 0.555789053440094, "learning_rate": 9.781449916689324e-05, "loss": 0.0382, "step": 11180 }, { "grad_norm": 0.35265037417411804, "learning_rate": 9.780845123601746e-05, "loss": 0.0476, "step": 11190 }, { "grad_norm": 0.277374267578125, "learning_rate": 9.780239513598492e-05, "loss": 0.037, "step": 11200 }, { "grad_norm": 0.5574532747268677, "learning_rate": 9.779633086783047e-05, "loss": 0.0408, "step": 11210 }, { "grad_norm": 0.3277866840362549, "learning_rate": 9.779025843259031e-05, "loss": 0.0303, "step": 11220 }, { "grad_norm": 0.2866462767124176, "learning_rate": 9.778417783130204e-05, "loss": 0.0339, "step": 11230 }, { "grad_norm": 0.2773391902446747, "learning_rate": 9.777808906500468e-05, "loss": 0.0269, "step": 11240 }, { "grad_norm": 0.3125341236591339, "learning_rate": 9.777199213473862e-05, "loss": 0.0366, "step": 11250 }, { "grad_norm": 0.2684602737426758, "learning_rate": 9.77658870415457e-05, "loss": 0.0266, "step": 11260 }, { "grad_norm": 0.2423272281885147, "learning_rate": 9.775977378646906e-05, "loss": 0.0302, "step": 11270 }, { "grad_norm": 0.4044184982776642, "learning_rate": 9.775365237055331e-05, "loss": 0.0326, "step": 11280 }, { "grad_norm": 0.440172016620636, "learning_rate": 9.774752279484445e-05, "loss": 0.0386, "step": 11290 }, { "grad_norm": 0.35120734572410583, "learning_rate": 9.774138506038984e-05, "loss": 0.0398, "step": 11300 }, { "grad_norm": 0.3252304792404175, "learning_rate": 9.773523916823826e-05, "loss": 0.0346, "step": 11310 }, { "grad_norm": 0.31477391719818115, "learning_rate": 9.772908511943986e-05, "loss": 0.0323, "step": 11320 }, { "grad_norm": 0.30161505937576294, "learning_rate": 9.77229229150462e-05, "loss": 0.0406, "step": 11330 }, { "grad_norm": 0.3888211250305176, "learning_rate": 9.771675255611024e-05, "loss": 0.0356, "step": 11340 }, { "grad_norm": 0.2657642066478729, "learning_rate": 9.771057404368632e-05, "loss": 0.0335, "step": 11350 }, { "grad_norm": 0.3156362473964691, "learning_rate": 9.770438737883018e-05, "loss": 0.0324, "step": 11360 }, { "grad_norm": 0.3447135090827942, "learning_rate": 9.769819256259898e-05, "loss": 0.0375, "step": 11370 }, { "grad_norm": 0.31816741824150085, "learning_rate": 9.769198959605119e-05, "loss": 0.0287, "step": 11380 }, { "grad_norm": 0.3068659007549286, "learning_rate": 9.768577848024678e-05, "loss": 0.03, "step": 11390 }, { "grad_norm": 0.2673933207988739, "learning_rate": 9.767955921624702e-05, "loss": 0.0274, "step": 11400 }, { "grad_norm": 0.34780675172805786, "learning_rate": 9.767333180511465e-05, "loss": 0.031, "step": 11410 }, { "grad_norm": 0.397223562002182, "learning_rate": 9.766709624791373e-05, "loss": 0.0361, "step": 11420 }, { "grad_norm": 0.2315216362476349, "learning_rate": 9.766085254570975e-05, "loss": 0.0307, "step": 11430 }, { "grad_norm": 0.28375205397605896, "learning_rate": 9.76546006995696e-05, "loss": 0.0278, "step": 11440 }, { "grad_norm": 0.25494512915611267, "learning_rate": 9.764834071056155e-05, "loss": 0.0317, "step": 11450 }, { "grad_norm": 0.3509029150009155, "learning_rate": 9.764207257975526e-05, "loss": 0.0428, "step": 11460 }, { "grad_norm": 0.3366115391254425, "learning_rate": 9.763579630822179e-05, "loss": 0.0393, "step": 11470 }, { "grad_norm": 0.29481780529022217, "learning_rate": 9.762951189703356e-05, "loss": 0.0345, "step": 11480 }, { "grad_norm": 0.515710175037384, "learning_rate": 9.762321934726442e-05, "loss": 0.0382, "step": 11490 }, { "grad_norm": 0.44019806385040283, "learning_rate": 9.761691865998959e-05, "loss": 0.0388, "step": 11500 }, { "grad_norm": 0.4698610007762909, "learning_rate": 9.76106098362857e-05, "loss": 0.0313, "step": 11510 }, { "grad_norm": 0.22200779616832733, "learning_rate": 9.760429287723072e-05, "loss": 0.0331, "step": 11520 }, { "grad_norm": 0.3198166787624359, "learning_rate": 9.759796778390406e-05, "loss": 0.0328, "step": 11530 }, { "grad_norm": 0.3505387306213379, "learning_rate": 9.759163455738653e-05, "loss": 0.0287, "step": 11540 }, { "grad_norm": 0.33846017718315125, "learning_rate": 9.75852931987603e-05, "loss": 0.0333, "step": 11550 }, { "grad_norm": 0.31763219833374023, "learning_rate": 9.757894370910891e-05, "loss": 0.0331, "step": 11560 }, { "grad_norm": 0.23093102872371674, "learning_rate": 9.757258608951733e-05, "loss": 0.029, "step": 11570 }, { "grad_norm": 0.23723429441452026, "learning_rate": 9.75662203410719e-05, "loss": 0.0252, "step": 11580 }, { "grad_norm": 0.35671311616897583, "learning_rate": 9.755984646486034e-05, "loss": 0.0304, "step": 11590 }, { "grad_norm": 0.3455841541290283, "learning_rate": 9.75534644619718e-05, "loss": 0.0283, "step": 11600 }, { "grad_norm": 0.254768043756485, "learning_rate": 9.754707433349676e-05, "loss": 0.0305, "step": 11610 }, { "grad_norm": 0.30056852102279663, "learning_rate": 9.754067608052715e-05, "loss": 0.0279, "step": 11620 }, { "grad_norm": 0.27831023931503296, "learning_rate": 9.753426970415622e-05, "loss": 0.0284, "step": 11630 }, { "grad_norm": 0.3915758430957794, "learning_rate": 9.752785520547868e-05, "loss": 0.0353, "step": 11640 }, { "grad_norm": 0.2665162682533264, "learning_rate": 9.752143258559056e-05, "loss": 0.0298, "step": 11650 }, { "grad_norm": 0.4768524467945099, "learning_rate": 9.751500184558933e-05, "loss": 0.0302, "step": 11660 }, { "grad_norm": 0.3566735088825226, "learning_rate": 9.750856298657383e-05, "loss": 0.0324, "step": 11670 }, { "grad_norm": 0.43871161341667175, "learning_rate": 9.750211600964428e-05, "loss": 0.0302, "step": 11680 }, { "grad_norm": 0.2941441237926483, "learning_rate": 9.749566091590226e-05, "loss": 0.0358, "step": 11690 }, { "grad_norm": 0.390023797750473, "learning_rate": 9.748919770645083e-05, "loss": 0.0342, "step": 11700 }, { "grad_norm": 0.24266140162944794, "learning_rate": 9.748272638239432e-05, "loss": 0.0327, "step": 11710 }, { "grad_norm": 0.39491546154022217, "learning_rate": 9.747624694483855e-05, "loss": 0.0287, "step": 11720 }, { "grad_norm": 0.4103579819202423, "learning_rate": 9.746975939489065e-05, "loss": 0.0259, "step": 11730 }, { "grad_norm": 0.2927623987197876, "learning_rate": 9.746326373365918e-05, "loss": 0.0284, "step": 11740 }, { "grad_norm": 0.33976536989212036, "learning_rate": 9.745675996225403e-05, "loss": 0.0285, "step": 11750 }, { "grad_norm": 0.47776368260383606, "learning_rate": 9.745024808178657e-05, "loss": 0.0423, "step": 11760 }, { "grad_norm": 0.4081599712371826, "learning_rate": 9.744372809336947e-05, "loss": 0.0298, "step": 11770 }, { "grad_norm": 0.2758410573005676, "learning_rate": 9.743719999811682e-05, "loss": 0.029, "step": 11780 }, { "grad_norm": 0.24958926439285278, "learning_rate": 9.743066379714412e-05, "loss": 0.0312, "step": 11790 }, { "grad_norm": 0.20267635583877563, "learning_rate": 9.74241194915682e-05, "loss": 0.0281, "step": 11800 }, { "grad_norm": 0.4323677718639374, "learning_rate": 9.741756708250731e-05, "loss": 0.0338, "step": 11810 }, { "grad_norm": 0.26702189445495605, "learning_rate": 9.741100657108109e-05, "loss": 0.0373, "step": 11820 }, { "grad_norm": 0.3740193843841553, "learning_rate": 9.740443795841054e-05, "loss": 0.0328, "step": 11830 }, { "grad_norm": 0.2511082589626312, "learning_rate": 9.739786124561805e-05, "loss": 0.0308, "step": 11840 }, { "grad_norm": 0.2261427938938141, "learning_rate": 9.73912764338274e-05, "loss": 0.0322, "step": 11850 }, { "grad_norm": 0.26456260681152344, "learning_rate": 9.738468352416377e-05, "loss": 0.0343, "step": 11860 }, { "grad_norm": 0.1755346655845642, "learning_rate": 9.737808251775369e-05, "loss": 0.031, "step": 11870 }, { "grad_norm": 0.2479168325662613, "learning_rate": 9.737147341572512e-05, "loss": 0.0265, "step": 11880 }, { "grad_norm": 0.2528288960456848, "learning_rate": 9.736485621920735e-05, "loss": 0.0283, "step": 11890 }, { "grad_norm": 0.2314392477273941, "learning_rate": 9.735823092933108e-05, "loss": 0.0281, "step": 11900 }, { "grad_norm": 0.2746792733669281, "learning_rate": 9.735159754722838e-05, "loss": 0.0267, "step": 11910 }, { "grad_norm": 0.35754454135894775, "learning_rate": 9.734495607403275e-05, "loss": 0.0275, "step": 11920 }, { "grad_norm": 0.3087783753871918, "learning_rate": 9.733830651087901e-05, "loss": 0.0256, "step": 11930 }, { "grad_norm": 0.40688735246658325, "learning_rate": 9.733164885890338e-05, "loss": 0.0343, "step": 11940 }, { "grad_norm": 0.24491791427135468, "learning_rate": 9.732498311924349e-05, "loss": 0.0306, "step": 11950 }, { "grad_norm": 0.2708950340747833, "learning_rate": 9.731830929303833e-05, "loss": 0.031, "step": 11960 }, { "grad_norm": 0.33031055331230164, "learning_rate": 9.731162738142827e-05, "loss": 0.0296, "step": 11970 }, { "grad_norm": 0.3672483563423157, "learning_rate": 9.730493738555506e-05, "loss": 0.024, "step": 11980 }, { "grad_norm": 0.24786360561847687, "learning_rate": 9.729823930656186e-05, "loss": 0.025, "step": 11990 }, { "grad_norm": 0.24462518095970154, "learning_rate": 9.729153314559316e-05, "loss": 0.0247, "step": 12000 }, { "grad_norm": 0.3106611371040344, "learning_rate": 9.728481890379486e-05, "loss": 0.0259, "step": 12010 }, { "grad_norm": 0.2687280476093292, "learning_rate": 9.727809658231428e-05, "loss": 0.0322, "step": 12020 }, { "grad_norm": 0.17889295518398285, "learning_rate": 9.727136618230003e-05, "loss": 0.0306, "step": 12030 }, { "grad_norm": 0.23530317842960358, "learning_rate": 9.726462770490219e-05, "loss": 0.024, "step": 12040 }, { "grad_norm": 0.25986236333847046, "learning_rate": 9.725788115127214e-05, "loss": 0.0256, "step": 12050 }, { "grad_norm": 0.4299505352973938, "learning_rate": 9.725112652256274e-05, "loss": 0.029, "step": 12060 }, { "grad_norm": 0.23207825422286987, "learning_rate": 9.724436381992812e-05, "loss": 0.0257, "step": 12070 }, { "grad_norm": 0.32259538769721985, "learning_rate": 9.723759304452387e-05, "loss": 0.0295, "step": 12080 }, { "grad_norm": 0.2580174505710602, "learning_rate": 9.72308141975069e-05, "loss": 0.0288, "step": 12090 }, { "grad_norm": 0.3039058446884155, "learning_rate": 9.722402728003557e-05, "loss": 0.0296, "step": 12100 }, { "grad_norm": 0.19820885360240936, "learning_rate": 9.721723229326953e-05, "loss": 0.0249, "step": 12110 }, { "grad_norm": 0.27386096119880676, "learning_rate": 9.721042923836992e-05, "loss": 0.0303, "step": 12120 }, { "grad_norm": 0.30376094579696655, "learning_rate": 9.720361811649914e-05, "loss": 0.0319, "step": 12130 }, { "grad_norm": 0.3111371397972107, "learning_rate": 9.719679892882106e-05, "loss": 0.0319, "step": 12140 }, { "grad_norm": 0.47678902745246887, "learning_rate": 9.718997167650085e-05, "loss": 0.035, "step": 12150 }, { "grad_norm": 0.41998952627182007, "learning_rate": 9.718313636070515e-05, "loss": 0.038, "step": 12160 }, { "grad_norm": 0.24788817763328552, "learning_rate": 9.717629298260192e-05, "loss": 0.0285, "step": 12170 }, { "grad_norm": 0.2322855144739151, "learning_rate": 9.716944154336047e-05, "loss": 0.0269, "step": 12180 }, { "grad_norm": 0.3614344596862793, "learning_rate": 9.716258204415157e-05, "loss": 0.0318, "step": 12190 }, { "grad_norm": 0.20508471131324768, "learning_rate": 9.715571448614728e-05, "loss": 0.0225, "step": 12200 }, { "grad_norm": 0.284298300743103, "learning_rate": 9.71488388705211e-05, "loss": 0.0312, "step": 12210 }, { "grad_norm": 0.23639027774333954, "learning_rate": 9.714195519844788e-05, "loss": 0.0295, "step": 12220 }, { "grad_norm": 0.2603820562362671, "learning_rate": 9.713506347110386e-05, "loss": 0.0331, "step": 12230 }, { "grad_norm": 0.2610636353492737, "learning_rate": 9.712816368966662e-05, "loss": 0.0329, "step": 12240 }, { "grad_norm": 0.29143181443214417, "learning_rate": 9.712125585531517e-05, "loss": 0.0333, "step": 12250 }, { "grad_norm": 0.32064691185951233, "learning_rate": 9.711433996922988e-05, "loss": 0.0309, "step": 12260 }, { "grad_norm": 0.3061397969722748, "learning_rate": 9.710741603259245e-05, "loss": 0.0277, "step": 12270 }, { "grad_norm": 0.3117597997188568, "learning_rate": 9.710048404658603e-05, "loss": 0.0271, "step": 12280 }, { "grad_norm": 0.260535329580307, "learning_rate": 9.709354401239508e-05, "loss": 0.0251, "step": 12290 }, { "grad_norm": 0.2914033830165863, "learning_rate": 9.708659593120546e-05, "loss": 0.025, "step": 12300 }, { "grad_norm": 0.2800244092941284, "learning_rate": 9.707963980420443e-05, "loss": 0.0273, "step": 12310 }, { "grad_norm": 0.24898786842823029, "learning_rate": 9.707267563258058e-05, "loss": 0.0326, "step": 12320 }, { "grad_norm": 0.2911490499973297, "learning_rate": 9.70657034175239e-05, "loss": 0.0265, "step": 12330 }, { "grad_norm": 0.3439300060272217, "learning_rate": 9.705872316022577e-05, "loss": 0.0317, "step": 12340 }, { "grad_norm": 0.4041587710380554, "learning_rate": 9.705173486187891e-05, "loss": 0.032, "step": 12350 }, { "grad_norm": 0.3908872604370117, "learning_rate": 9.704473852367741e-05, "loss": 0.0253, "step": 12360 }, { "grad_norm": 0.31258419156074524, "learning_rate": 9.70377341468168e-05, "loss": 0.0293, "step": 12370 }, { "grad_norm": 0.33507439494132996, "learning_rate": 9.703072173249389e-05, "loss": 0.0337, "step": 12380 }, { "grad_norm": 0.24582047760486603, "learning_rate": 9.702370128190693e-05, "loss": 0.026, "step": 12390 }, { "grad_norm": 0.30701538920402527, "learning_rate": 9.701667279625552e-05, "loss": 0.0302, "step": 12400 }, { "grad_norm": 0.2685164511203766, "learning_rate": 9.700963627674065e-05, "loss": 0.0317, "step": 12410 }, { "grad_norm": 0.3492773175239563, "learning_rate": 9.700259172456466e-05, "loss": 0.0282, "step": 12420 }, { "grad_norm": 0.2299525886774063, "learning_rate": 9.699553914093124e-05, "loss": 0.0279, "step": 12430 }, { "grad_norm": 0.3850462734699249, "learning_rate": 9.698847852704553e-05, "loss": 0.0274, "step": 12440 }, { "grad_norm": 0.21249845623970032, "learning_rate": 9.6981409884114e-05, "loss": 0.0216, "step": 12450 }, { "grad_norm": 0.2464178055524826, "learning_rate": 9.697433321334443e-05, "loss": 0.0286, "step": 12460 }, { "grad_norm": 0.21667639911174774, "learning_rate": 9.696724851594607e-05, "loss": 0.0263, "step": 12470 }, { "grad_norm": 0.4936833679676056, "learning_rate": 9.696015579312952e-05, "loss": 0.0265, "step": 12480 }, { "grad_norm": 0.28276464343070984, "learning_rate": 9.695305504610668e-05, "loss": 0.0263, "step": 12490 }, { "grad_norm": 0.3075583875179291, "learning_rate": 9.694594627609092e-05, "loss": 0.0231, "step": 12500 }, { "grad_norm": 0.3004347085952759, "learning_rate": 9.693882948429691e-05, "loss": 0.0215, "step": 12510 }, { "grad_norm": 0.19834859669208527, "learning_rate": 9.693170467194071e-05, "loss": 0.0253, "step": 12520 }, { "grad_norm": 0.2339932769536972, "learning_rate": 9.692457184023977e-05, "loss": 0.0271, "step": 12530 }, { "grad_norm": 0.2990257740020752, "learning_rate": 9.691743099041291e-05, "loss": 0.0271, "step": 12540 }, { "grad_norm": 0.31014588475227356, "learning_rate": 9.691028212368027e-05, "loss": 0.0305, "step": 12550 }, { "grad_norm": 0.3606867790222168, "learning_rate": 9.690312524126342e-05, "loss": 0.0325, "step": 12560 }, { "grad_norm": 0.3045312166213989, "learning_rate": 9.689596034438527e-05, "loss": 0.0343, "step": 12570 }, { "grad_norm": 0.22202999889850616, "learning_rate": 9.688878743427012e-05, "loss": 0.0284, "step": 12580 }, { "grad_norm": 0.17010244727134705, "learning_rate": 9.688160651214359e-05, "loss": 0.0233, "step": 12590 }, { "grad_norm": 0.35912057757377625, "learning_rate": 9.687441757923273e-05, "loss": 0.0276, "step": 12600 }, { "grad_norm": 0.33835670351982117, "learning_rate": 9.68672206367659e-05, "loss": 0.0272, "step": 12610 }, { "grad_norm": 0.24104353785514832, "learning_rate": 9.686001568597291e-05, "loss": 0.0252, "step": 12620 }, { "grad_norm": 0.2671768069267273, "learning_rate": 9.685280272808486e-05, "loss": 0.0343, "step": 12630 }, { "grad_norm": 0.20876947045326233, "learning_rate": 9.684558176433424e-05, "loss": 0.024, "step": 12640 }, { "grad_norm": 0.26368188858032227, "learning_rate": 9.683835279595495e-05, "loss": 0.0239, "step": 12650 }, { "grad_norm": 0.25793877243995667, "learning_rate": 9.683111582418216e-05, "loss": 0.025, "step": 12660 }, { "grad_norm": 0.3307102620601654, "learning_rate": 9.682387085025254e-05, "loss": 0.029, "step": 12670 }, { "grad_norm": 0.35724255442619324, "learning_rate": 9.681661787540401e-05, "loss": 0.0258, "step": 12680 }, { "grad_norm": 0.24443303048610687, "learning_rate": 9.680935690087593e-05, "loss": 0.0322, "step": 12690 }, { "grad_norm": 0.3120298683643341, "learning_rate": 9.680208792790901e-05, "loss": 0.0339, "step": 12700 }, { "grad_norm": 0.26863059401512146, "learning_rate": 9.679481095774529e-05, "loss": 0.0282, "step": 12710 }, { "grad_norm": 0.36861827969551086, "learning_rate": 9.678752599162822e-05, "loss": 0.0316, "step": 12720 }, { "grad_norm": 0.2169095277786255, "learning_rate": 9.678023303080259e-05, "loss": 0.0248, "step": 12730 }, { "grad_norm": 0.25601035356521606, "learning_rate": 9.677293207651459e-05, "loss": 0.0293, "step": 12740 }, { "grad_norm": 0.2042754739522934, "learning_rate": 9.676562313001173e-05, "loss": 0.0229, "step": 12750 }, { "grad_norm": 0.25338178873062134, "learning_rate": 9.675830619254293e-05, "loss": 0.0262, "step": 12760 }, { "grad_norm": 0.1933155655860901, "learning_rate": 9.675098126535843e-05, "loss": 0.0223, "step": 12770 }, { "grad_norm": 0.29384762048721313, "learning_rate": 9.674364834970988e-05, "loss": 0.0273, "step": 12780 }, { "grad_norm": 0.25862833857536316, "learning_rate": 9.673630744685028e-05, "loss": 0.0315, "step": 12790 }, { "grad_norm": 0.2793572247028351, "learning_rate": 9.672895855803397e-05, "loss": 0.0297, "step": 12800 }, { "grad_norm": 0.2138955444097519, "learning_rate": 9.672160168451667e-05, "loss": 0.0293, "step": 12810 }, { "grad_norm": 0.293349951505661, "learning_rate": 9.671423682755549e-05, "loss": 0.0257, "step": 12820 }, { "grad_norm": 0.310886412858963, "learning_rate": 9.670686398840888e-05, "loss": 0.0299, "step": 12830 }, { "grad_norm": 0.3636886477470398, "learning_rate": 9.669948316833664e-05, "loss": 0.0291, "step": 12840 }, { "grad_norm": 0.3387503921985626, "learning_rate": 9.669209436859997e-05, "loss": 0.0293, "step": 12850 }, { "grad_norm": 0.19194085896015167, "learning_rate": 9.66846975904614e-05, "loss": 0.03, "step": 12860 }, { "grad_norm": 0.2875995337963104, "learning_rate": 9.667729283518483e-05, "loss": 0.0258, "step": 12870 }, { "grad_norm": 0.24163974821567535, "learning_rate": 9.666988010403557e-05, "loss": 0.0266, "step": 12880 }, { "grad_norm": 0.26542794704437256, "learning_rate": 9.66624593982802e-05, "loss": 0.0271, "step": 12890 }, { "grad_norm": 0.21995609998703003, "learning_rate": 9.665503071918675e-05, "loss": 0.0259, "step": 12900 }, { "grad_norm": 0.31772851943969727, "learning_rate": 9.664759406802456e-05, "loss": 0.0268, "step": 12910 }, { "grad_norm": 0.2922224998474121, "learning_rate": 9.664014944606437e-05, "loss": 0.029, "step": 12920 }, { "grad_norm": 0.32959064841270447, "learning_rate": 9.663269685457822e-05, "loss": 0.0265, "step": 12930 }, { "grad_norm": 0.3976649045944214, "learning_rate": 9.662523629483962e-05, "loss": 0.0266, "step": 12940 }, { "grad_norm": 0.3041779100894928, "learning_rate": 9.661776776812333e-05, "loss": 0.0292, "step": 12950 }, { "grad_norm": 0.2345142513513565, "learning_rate": 9.661029127570553e-05, "loss": 0.0256, "step": 12960 }, { "grad_norm": 0.3005853593349457, "learning_rate": 9.660280681886373e-05, "loss": 0.0312, "step": 12970 }, { "grad_norm": 0.34665942192077637, "learning_rate": 9.659531439887685e-05, "loss": 0.03, "step": 12980 }, { "grad_norm": 0.3216816782951355, "learning_rate": 9.658781401702511e-05, "loss": 0.028, "step": 12990 }, { "grad_norm": 0.25968828797340393, "learning_rate": 9.658030567459015e-05, "loss": 0.0265, "step": 13000 }, { "grad_norm": 0.2917861342430115, "learning_rate": 9.65727893728549e-05, "loss": 0.0282, "step": 13010 }, { "grad_norm": 0.29361769556999207, "learning_rate": 9.656526511310375e-05, "loss": 0.0258, "step": 13020 }, { "grad_norm": 0.3238969147205353, "learning_rate": 9.655773289662233e-05, "loss": 0.0276, "step": 13030 }, { "grad_norm": 0.18199674785137177, "learning_rate": 9.655019272469772e-05, "loss": 0.0242, "step": 13040 }, { "grad_norm": 0.362564355134964, "learning_rate": 9.654264459861832e-05, "loss": 0.028, "step": 13050 }, { "grad_norm": 0.20156820118427277, "learning_rate": 9.653508851967391e-05, "loss": 0.0237, "step": 13060 }, { "grad_norm": 0.23539674282073975, "learning_rate": 9.65275244891556e-05, "loss": 0.0223, "step": 13070 }, { "grad_norm": 0.3210219740867615, "learning_rate": 9.651995250835591e-05, "loss": 0.0253, "step": 13080 }, { "grad_norm": 0.3303537666797638, "learning_rate": 9.651237257856862e-05, "loss": 0.0348, "step": 13090 }, { "grad_norm": 0.2131270468235016, "learning_rate": 9.6504784701089e-05, "loss": 0.0266, "step": 13100 }, { "grad_norm": 0.30442577600479126, "learning_rate": 9.649718887721357e-05, "loss": 0.0313, "step": 13110 }, { "grad_norm": 0.3679083287715912, "learning_rate": 9.648958510824028e-05, "loss": 0.0374, "step": 13120 }, { "grad_norm": 0.325057715177536, "learning_rate": 9.648197339546837e-05, "loss": 0.0344, "step": 13130 }, { "grad_norm": 0.26263168454170227, "learning_rate": 9.647435374019851e-05, "loss": 0.0272, "step": 13140 }, { "grad_norm": 0.27184218168258667, "learning_rate": 9.646672614373266e-05, "loss": 0.0304, "step": 13150 }, { "grad_norm": 0.233912855386734, "learning_rate": 9.645909060737418e-05, "loss": 0.0266, "step": 13160 }, { "grad_norm": 0.19068916141986847, "learning_rate": 9.645144713242778e-05, "loss": 0.0319, "step": 13170 }, { "grad_norm": 0.26010459661483765, "learning_rate": 9.64437957201995e-05, "loss": 0.0294, "step": 13180 }, { "grad_norm": 0.24298255145549774, "learning_rate": 9.643613637199678e-05, "loss": 0.0262, "step": 13190 }, { "grad_norm": 0.23841194808483124, "learning_rate": 9.642846908912839e-05, "loss": 0.0305, "step": 13200 }, { "grad_norm": 0.2514127194881439, "learning_rate": 9.642079387290444e-05, "loss": 0.0277, "step": 13210 }, { "grad_norm": 0.1679222136735916, "learning_rate": 9.641311072463644e-05, "loss": 0.0277, "step": 13220 }, { "grad_norm": 0.22982122004032135, "learning_rate": 9.640541964563722e-05, "loss": 0.026, "step": 13230 }, { "grad_norm": 0.2617689073085785, "learning_rate": 9.639772063722096e-05, "loss": 0.0313, "step": 13240 }, { "grad_norm": 0.2439662367105484, "learning_rate": 9.639001370070324e-05, "loss": 0.0335, "step": 13250 }, { "grad_norm": 0.26424849033355713, "learning_rate": 9.638229883740095e-05, "loss": 0.0365, "step": 13260 }, { "grad_norm": 0.3244189918041229, "learning_rate": 9.637457604863233e-05, "loss": 0.033, "step": 13270 }, { "grad_norm": 0.3116227388381958, "learning_rate": 9.636684533571703e-05, "loss": 0.0288, "step": 13280 }, { "grad_norm": 0.3180735111236572, "learning_rate": 9.635910669997599e-05, "loss": 0.0315, "step": 13290 }, { "grad_norm": 0.2665703296661377, "learning_rate": 9.635136014273154e-05, "loss": 0.0306, "step": 13300 }, { "grad_norm": 0.22573140263557434, "learning_rate": 9.634360566530735e-05, "loss": 0.024, "step": 13310 }, { "grad_norm": 0.28621014952659607, "learning_rate": 9.633584326902845e-05, "loss": 0.0271, "step": 13320 }, { "grad_norm": 0.3191778361797333, "learning_rate": 9.632807295522124e-05, "loss": 0.0309, "step": 13330 }, { "grad_norm": 0.27286049723625183, "learning_rate": 9.632029472521342e-05, "loss": 0.0306, "step": 13340 }, { "grad_norm": 0.300373911857605, "learning_rate": 9.631250858033409e-05, "loss": 0.0224, "step": 13350 }, { "grad_norm": 0.2900296151638031, "learning_rate": 9.630471452191371e-05, "loss": 0.0215, "step": 13360 }, { "grad_norm": 0.3673747181892395, "learning_rate": 9.629691255128405e-05, "loss": 0.0226, "step": 13370 }, { "grad_norm": 0.226399227976799, "learning_rate": 9.628910266977825e-05, "loss": 0.0231, "step": 13380 }, { "grad_norm": 0.20749834179878235, "learning_rate": 9.628128487873083e-05, "loss": 0.0244, "step": 13390 }, { "grad_norm": 0.22985023260116577, "learning_rate": 9.627345917947761e-05, "loss": 0.0251, "step": 13400 }, { "grad_norm": 0.27480289340019226, "learning_rate": 9.626562557335579e-05, "loss": 0.0242, "step": 13410 }, { "grad_norm": 0.21274136006832123, "learning_rate": 9.625778406170393e-05, "loss": 0.0265, "step": 13420 }, { "grad_norm": 0.2523308992385864, "learning_rate": 9.624993464586193e-05, "loss": 0.03, "step": 13430 }, { "grad_norm": 0.18568569421768188, "learning_rate": 9.624207732717105e-05, "loss": 0.0301, "step": 13440 }, { "grad_norm": 0.2837805151939392, "learning_rate": 9.623421210697386e-05, "loss": 0.0289, "step": 13450 }, { "grad_norm": 0.35700416564941406, "learning_rate": 9.622633898661434e-05, "loss": 0.0302, "step": 13460 }, { "grad_norm": 0.3793489336967468, "learning_rate": 9.621845796743778e-05, "loss": 0.0303, "step": 13470 }, { "grad_norm": 0.2506324052810669, "learning_rate": 9.621056905079082e-05, "loss": 0.0265, "step": 13480 }, { "grad_norm": 0.25908997654914856, "learning_rate": 9.620267223802149e-05, "loss": 0.0372, "step": 13490 }, { "grad_norm": 0.49512508511543274, "learning_rate": 9.619476753047911e-05, "loss": 0.0312, "step": 13500 }, { "grad_norm": 0.24450235068798065, "learning_rate": 9.618685492951438e-05, "loss": 0.0324, "step": 13510 }, { "grad_norm": 0.3106596767902374, "learning_rate": 9.617893443647938e-05, "loss": 0.0237, "step": 13520 }, { "grad_norm": 0.23896312713623047, "learning_rate": 9.617100605272746e-05, "loss": 0.026, "step": 13530 }, { "grad_norm": 0.3401068449020386, "learning_rate": 9.616306977961338e-05, "loss": 0.0342, "step": 13540 }, { "grad_norm": 0.24085278809070587, "learning_rate": 9.615512561849326e-05, "loss": 0.0298, "step": 13550 }, { "grad_norm": 0.2411985546350479, "learning_rate": 9.61471735707245e-05, "loss": 0.0327, "step": 13560 }, { "grad_norm": 0.39052051305770874, "learning_rate": 9.613921363766592e-05, "loss": 0.0323, "step": 13570 }, { "grad_norm": 0.26341870427131653, "learning_rate": 9.613124582067763e-05, "loss": 0.0335, "step": 13580 }, { "grad_norm": 0.2634727954864502, "learning_rate": 9.612327012112112e-05, "loss": 0.03, "step": 13590 }, { "grad_norm": 0.2581135630607605, "learning_rate": 9.611528654035921e-05, "loss": 0.029, "step": 13600 }, { "grad_norm": 0.3027373254299164, "learning_rate": 9.610729507975611e-05, "loss": 0.0315, "step": 13610 }, { "grad_norm": 0.2864629924297333, "learning_rate": 9.609929574067731e-05, "loss": 0.0296, "step": 13620 }, { "grad_norm": 0.28826698660850525, "learning_rate": 9.609128852448967e-05, "loss": 0.0345, "step": 13630 }, { "grad_norm": 0.25473925471305847, "learning_rate": 9.608327343256143e-05, "loss": 0.0334, "step": 13640 }, { "grad_norm": 0.18919208645820618, "learning_rate": 9.607525046626216e-05, "loss": 0.0262, "step": 13650 }, { "grad_norm": 0.21399493515491486, "learning_rate": 9.606721962696272e-05, "loss": 0.0298, "step": 13660 }, { "grad_norm": 0.23309439420700073, "learning_rate": 9.60591809160354e-05, "loss": 0.0268, "step": 13670 }, { "grad_norm": 0.27616626024246216, "learning_rate": 9.605113433485378e-05, "loss": 0.0321, "step": 13680 }, { "grad_norm": 0.39365822076797485, "learning_rate": 9.604307988479279e-05, "loss": 0.0258, "step": 13690 }, { "grad_norm": 0.2478085607290268, "learning_rate": 9.603501756722876e-05, "loss": 0.0284, "step": 13700 }, { "grad_norm": 0.27301472425460815, "learning_rate": 9.602694738353927e-05, "loss": 0.0326, "step": 13710 }, { "grad_norm": 0.33375048637390137, "learning_rate": 9.601886933510331e-05, "loss": 0.0291, "step": 13720 }, { "grad_norm": 0.2764897346496582, "learning_rate": 9.60107834233012e-05, "loss": 0.0289, "step": 13730 }, { "grad_norm": 0.30620163679122925, "learning_rate": 9.60026896495146e-05, "loss": 0.0285, "step": 13740 }, { "grad_norm": 0.30222076177597046, "learning_rate": 9.599458801512652e-05, "loss": 0.0311, "step": 13750 }, { "grad_norm": 0.2832421064376831, "learning_rate": 9.598647852152129e-05, "loss": 0.0302, "step": 13760 }, { "grad_norm": 0.2324761599302292, "learning_rate": 9.597836117008462e-05, "loss": 0.0274, "step": 13770 }, { "grad_norm": 0.27857494354248047, "learning_rate": 9.597023596220356e-05, "loss": 0.0284, "step": 13780 }, { "grad_norm": 0.3212699294090271, "learning_rate": 9.596210289926643e-05, "loss": 0.0295, "step": 13790 }, { "grad_norm": 0.20497871935367584, "learning_rate": 9.5953961982663e-05, "loss": 0.0261, "step": 13800 }, { "grad_norm": 0.2883126735687256, "learning_rate": 9.594581321378431e-05, "loss": 0.0288, "step": 13810 }, { "grad_norm": 0.2645220160484314, "learning_rate": 9.593765659402276e-05, "loss": 0.0266, "step": 13820 }, { "grad_norm": 0.30330920219421387, "learning_rate": 9.59294921247721e-05, "loss": 0.0263, "step": 13830 }, { "grad_norm": 0.25182870030403137, "learning_rate": 9.59213198074274e-05, "loss": 0.0266, "step": 13840 }, { "grad_norm": 0.2915378510951996, "learning_rate": 9.59131396433851e-05, "loss": 0.026, "step": 13850 }, { "grad_norm": 0.2741748094558716, "learning_rate": 9.590495163404297e-05, "loss": 0.029, "step": 13860 }, { "grad_norm": 0.24225859344005585, "learning_rate": 9.589675578080009e-05, "loss": 0.0264, "step": 13870 }, { "grad_norm": 0.3193283677101135, "learning_rate": 9.588855208505694e-05, "loss": 0.0293, "step": 13880 }, { "grad_norm": 0.26073864102363586, "learning_rate": 9.588034054821529e-05, "loss": 0.0314, "step": 13890 }, { "grad_norm": 0.29397132992744446, "learning_rate": 9.587212117167826e-05, "loss": 0.0369, "step": 13900 }, { "grad_norm": 0.3074644207954407, "learning_rate": 9.586389395685033e-05, "loss": 0.0256, "step": 13910 }, { "grad_norm": 0.3925539255142212, "learning_rate": 9.585565890513733e-05, "loss": 0.0322, "step": 13920 }, { "grad_norm": 0.2347671389579773, "learning_rate": 9.584741601794636e-05, "loss": 0.0285, "step": 13930 }, { "grad_norm": 0.26773902773857117, "learning_rate": 9.58391652966859e-05, "loss": 0.0317, "step": 13940 }, { "grad_norm": 0.315123975276947, "learning_rate": 9.583090674276583e-05, "loss": 0.0268, "step": 13950 }, { "grad_norm": 0.22538752853870392, "learning_rate": 9.582264035759726e-05, "loss": 0.0243, "step": 13960 }, { "grad_norm": 0.2557484805583954, "learning_rate": 9.58143661425927e-05, "loss": 0.0286, "step": 13970 }, { "grad_norm": 0.24743540585041046, "learning_rate": 9.580608409916601e-05, "loss": 0.0328, "step": 13980 }, { "grad_norm": 0.28197264671325684, "learning_rate": 9.579779422873233e-05, "loss": 0.0272, "step": 13990 }, { "grad_norm": 0.3645443916320801, "learning_rate": 9.578949653270819e-05, "loss": 0.026, "step": 14000 }, { "grad_norm": 0.2384033501148224, "learning_rate": 9.578119101251144e-05, "loss": 0.0243, "step": 14010 }, { "grad_norm": 0.2446710169315338, "learning_rate": 9.577287766956127e-05, "loss": 0.0295, "step": 14020 }, { "grad_norm": 0.3408454656600952, "learning_rate": 9.57645565052782e-05, "loss": 0.024, "step": 14030 }, { "grad_norm": 0.24009723961353302, "learning_rate": 9.575622752108407e-05, "loss": 0.0261, "step": 14040 }, { "grad_norm": 0.21856169402599335, "learning_rate": 9.57478907184021e-05, "loss": 0.0242, "step": 14050 }, { "grad_norm": 0.25806257128715515, "learning_rate": 9.573954609865681e-05, "loss": 0.0301, "step": 14060 }, { "grad_norm": 0.22692550718784332, "learning_rate": 9.573119366327408e-05, "loss": 0.0259, "step": 14070 }, { "grad_norm": 0.20064756274223328, "learning_rate": 9.57228334136811e-05, "loss": 0.0238, "step": 14080 }, { "grad_norm": 0.3740139305591583, "learning_rate": 9.571446535130641e-05, "loss": 0.0253, "step": 14090 }, { "grad_norm": 0.4953579604625702, "learning_rate": 9.570608947757988e-05, "loss": 0.0286, "step": 14100 }, { "grad_norm": 0.32080966234207153, "learning_rate": 9.569770579393274e-05, "loss": 0.024, "step": 14110 }, { "grad_norm": 0.3911086618900299, "learning_rate": 9.56893143017975e-05, "loss": 0.0284, "step": 14120 }, { "grad_norm": 0.34388765692710876, "learning_rate": 9.568091500260806e-05, "loss": 0.0309, "step": 14130 }, { "grad_norm": 0.263009637594223, "learning_rate": 9.567250789779961e-05, "loss": 0.0321, "step": 14140 }, { "grad_norm": 0.3572722375392914, "learning_rate": 9.566409298880872e-05, "loss": 0.0244, "step": 14150 }, { "grad_norm": 0.21793481707572937, "learning_rate": 9.565567027707326e-05, "loss": 0.025, "step": 14160 }, { "grad_norm": 0.2227770984172821, "learning_rate": 9.56472397640324e-05, "loss": 0.0272, "step": 14170 }, { "grad_norm": 0.33686089515686035, "learning_rate": 9.563880145112675e-05, "loss": 0.0278, "step": 14180 }, { "grad_norm": 0.2193404585123062, "learning_rate": 9.563035533979814e-05, "loss": 0.0239, "step": 14190 }, { "grad_norm": 0.31221169233322144, "learning_rate": 9.562190143148981e-05, "loss": 0.0228, "step": 14200 }, { "grad_norm": 0.2464016228914261, "learning_rate": 9.561343972764627e-05, "loss": 0.0246, "step": 14210 }, { "grad_norm": 0.3180670440196991, "learning_rate": 9.560497022971343e-05, "loss": 0.0291, "step": 14220 }, { "grad_norm": 0.2742231786251068, "learning_rate": 9.559649293913847e-05, "loss": 0.0296, "step": 14230 }, { "grad_norm": 0.26323387026786804, "learning_rate": 9.558800785736993e-05, "loss": 0.0244, "step": 14240 }, { "grad_norm": 0.18385040760040283, "learning_rate": 9.557951498585767e-05, "loss": 0.0268, "step": 14250 }, { "grad_norm": 0.18348637223243713, "learning_rate": 9.557101432605293e-05, "loss": 0.028, "step": 14260 }, { "grad_norm": 0.3456602990627289, "learning_rate": 9.556250587940818e-05, "loss": 0.0257, "step": 14270 }, { "grad_norm": 0.3160270154476166, "learning_rate": 9.555398964737734e-05, "loss": 0.026, "step": 14280 }, { "grad_norm": 0.2373627871274948, "learning_rate": 9.554546563141555e-05, "loss": 0.0201, "step": 14290 }, { "grad_norm": 0.200822651386261, "learning_rate": 9.553693383297937e-05, "loss": 0.0218, "step": 14300 }, { "grad_norm": 0.2053089141845703, "learning_rate": 9.552839425352663e-05, "loss": 0.02, "step": 14310 }, { "grad_norm": 0.1648436337709427, "learning_rate": 9.551984689451652e-05, "loss": 0.0229, "step": 14320 }, { "grad_norm": 0.24106308817863464, "learning_rate": 9.551129175740953e-05, "loss": 0.0272, "step": 14330 }, { "grad_norm": 0.21652856469154358, "learning_rate": 9.550272884366754e-05, "loss": 0.0296, "step": 14340 }, { "grad_norm": 0.22610220313072205, "learning_rate": 9.549415815475369e-05, "loss": 0.0271, "step": 14350 }, { "grad_norm": 0.2527647912502289, "learning_rate": 9.548557969213247e-05, "loss": 0.032, "step": 14360 }, { "grad_norm": 0.2790871262550354, "learning_rate": 9.547699345726972e-05, "loss": 0.0271, "step": 14370 }, { "grad_norm": 0.39406707882881165, "learning_rate": 9.546839945163257e-05, "loss": 0.0339, "step": 14380 }, { "grad_norm": 0.27816227078437805, "learning_rate": 9.545979767668953e-05, "loss": 0.0278, "step": 14390 }, { "grad_norm": 0.3536466062068939, "learning_rate": 9.54511881339104e-05, "loss": 0.0275, "step": 14400 }, { "grad_norm": 0.31266582012176514, "learning_rate": 9.54425708247663e-05, "loss": 0.0279, "step": 14410 }, { "grad_norm": 0.21170534193515778, "learning_rate": 9.543394575072972e-05, "loss": 0.0257, "step": 14420 }, { "grad_norm": 0.28161269426345825, "learning_rate": 9.542531291327441e-05, "loss": 0.0225, "step": 14430 }, { "grad_norm": 0.28435447812080383, "learning_rate": 9.541667231387552e-05, "loss": 0.0229, "step": 14440 }, { "grad_norm": 0.3029438257217407, "learning_rate": 9.540802395400949e-05, "loss": 0.0288, "step": 14450 }, { "grad_norm": 0.3463348150253296, "learning_rate": 9.539936783515406e-05, "loss": 0.031, "step": 14460 }, { "grad_norm": 0.38547810912132263, "learning_rate": 9.539070395878835e-05, "loss": 0.0375, "step": 14470 }, { "grad_norm": 0.3212260603904724, "learning_rate": 9.538203232639277e-05, "loss": 0.0368, "step": 14480 }, { "grad_norm": 0.25584346055984497, "learning_rate": 9.537335293944907e-05, "loss": 0.0312, "step": 14490 }, { "grad_norm": 0.3192524313926697, "learning_rate": 9.536466579944032e-05, "loss": 0.0261, "step": 14500 }, { "grad_norm": 0.30106064677238464, "learning_rate": 9.535597090785091e-05, "loss": 0.0284, "step": 14510 }, { "grad_norm": 0.3265521228313446, "learning_rate": 9.534726826616656e-05, "loss": 0.0241, "step": 14520 }, { "grad_norm": 0.2745601236820221, "learning_rate": 9.53385578758743e-05, "loss": 0.0313, "step": 14530 }, { "grad_norm": 1.1004053354263306, "learning_rate": 9.532983973846252e-05, "loss": 0.0321, "step": 14540 }, { "grad_norm": 0.42510369420051575, "learning_rate": 9.53211138554209e-05, "loss": 0.0355, "step": 14550 }, { "grad_norm": 0.29102760553359985, "learning_rate": 9.531238022824047e-05, "loss": 0.0282, "step": 14560 }, { "grad_norm": 0.32034680247306824, "learning_rate": 9.530363885841355e-05, "loss": 0.0279, "step": 14570 }, { "grad_norm": 0.28964924812316895, "learning_rate": 9.52948897474338e-05, "loss": 0.0243, "step": 14580 }, { "grad_norm": 0.2825934886932373, "learning_rate": 9.528613289679622e-05, "loss": 0.0232, "step": 14590 }, { "grad_norm": 0.321977823972702, "learning_rate": 9.52773683079971e-05, "loss": 0.0304, "step": 14600 }, { "grad_norm": 0.2995043992996216, "learning_rate": 9.526859598253407e-05, "loss": 0.0325, "step": 14610 }, { "grad_norm": 0.29960185289382935, "learning_rate": 9.525981592190609e-05, "loss": 0.0234, "step": 14620 }, { "grad_norm": 0.23956096172332764, "learning_rate": 9.525102812761342e-05, "loss": 0.0336, "step": 14630 }, { "grad_norm": 0.2972114384174347, "learning_rate": 9.524223260115768e-05, "loss": 0.0275, "step": 14640 }, { "grad_norm": 0.22316111624240875, "learning_rate": 9.523342934404175e-05, "loss": 0.0273, "step": 14650 }, { "grad_norm": 0.23218001425266266, "learning_rate": 9.522461835776989e-05, "loss": 0.0249, "step": 14660 }, { "grad_norm": 0.2600511610507965, "learning_rate": 9.521579964384764e-05, "loss": 0.0242, "step": 14670 }, { "grad_norm": 0.2825453579425812, "learning_rate": 9.52069732037819e-05, "loss": 0.0242, "step": 14680 }, { "grad_norm": 0.2572757303714752, "learning_rate": 9.519813903908083e-05, "loss": 0.0229, "step": 14690 }, { "grad_norm": 0.2683745324611664, "learning_rate": 9.5189297151254e-05, "loss": 0.0283, "step": 14700 }, { "grad_norm": 0.3854447901248932, "learning_rate": 9.518044754181218e-05, "loss": 0.0369, "step": 14710 }, { "grad_norm": 0.32187408208847046, "learning_rate": 9.51715902122676e-05, "loss": 0.0357, "step": 14720 }, { "grad_norm": 0.2743450105190277, "learning_rate": 9.516272516413368e-05, "loss": 0.0317, "step": 14730 }, { "grad_norm": 0.32666656374931335, "learning_rate": 9.515385239892525e-05, "loss": 0.0363, "step": 14740 }, { "grad_norm": 0.34297698736190796, "learning_rate": 9.514497191815839e-05, "loss": 0.0315, "step": 14750 }, { "grad_norm": 0.2896521985530853, "learning_rate": 9.513608372335055e-05, "loss": 0.0314, "step": 14760 }, { "grad_norm": 0.2618980407714844, "learning_rate": 9.512718781602045e-05, "loss": 0.036, "step": 14770 }, { "grad_norm": 0.2716728746891022, "learning_rate": 9.511828419768823e-05, "loss": 0.0255, "step": 14780 }, { "grad_norm": 0.20564135909080505, "learning_rate": 9.510937286987521e-05, "loss": 0.0271, "step": 14790 }, { "grad_norm": 0.3071112632751465, "learning_rate": 9.510045383410408e-05, "loss": 0.0217, "step": 14800 }, { "grad_norm": 0.2784254252910614, "learning_rate": 9.509152709189892e-05, "loss": 0.0276, "step": 14810 }, { "grad_norm": 0.235801562666893, "learning_rate": 9.508259264478504e-05, "loss": 0.0246, "step": 14820 }, { "grad_norm": 0.26777610182762146, "learning_rate": 9.507365049428909e-05, "loss": 0.0294, "step": 14830 }, { "grad_norm": 0.2928476333618164, "learning_rate": 9.506470064193902e-05, "loss": 0.0272, "step": 14840 }, { "grad_norm": 0.1483408361673355, "learning_rate": 9.505574308926414e-05, "loss": 0.0248, "step": 14850 }, { "grad_norm": 0.229909747838974, "learning_rate": 9.504677783779505e-05, "loss": 0.0261, "step": 14860 }, { "grad_norm": 0.28275367617607117, "learning_rate": 9.503780488906365e-05, "loss": 0.0264, "step": 14870 }, { "grad_norm": 0.3400011360645294, "learning_rate": 9.502882424460319e-05, "loss": 0.0236, "step": 14880 }, { "grad_norm": 0.25005894899368286, "learning_rate": 9.501983590594821e-05, "loss": 0.0219, "step": 14890 }, { "grad_norm": 0.2909989058971405, "learning_rate": 9.501083987463455e-05, "loss": 0.0229, "step": 14900 }, { "grad_norm": 0.34094345569610596, "learning_rate": 9.500183615219942e-05, "loss": 0.0239, "step": 14910 }, { "grad_norm": 0.2708975672721863, "learning_rate": 9.49928247401813e-05, "loss": 0.0307, "step": 14920 }, { "grad_norm": 0.23673568665981293, "learning_rate": 9.498380564011997e-05, "loss": 0.0243, "step": 14930 }, { "grad_norm": 0.3182630240917206, "learning_rate": 9.497477885355656e-05, "loss": 0.0287, "step": 14940 }, { "grad_norm": 0.2634812593460083, "learning_rate": 9.496574438203353e-05, "loss": 0.024, "step": 14950 }, { "grad_norm": 0.2295188456773758, "learning_rate": 9.495670222709459e-05, "loss": 0.0223, "step": 14960 }, { "grad_norm": 0.15259572863578796, "learning_rate": 9.494765239028483e-05, "loss": 0.0243, "step": 14970 }, { "grad_norm": 0.23815128207206726, "learning_rate": 9.493859487315057e-05, "loss": 0.0212, "step": 14980 }, { "grad_norm": 0.316815584897995, "learning_rate": 9.492952967723953e-05, "loss": 0.0296, "step": 14990 }, { "grad_norm": 0.32911092042922974, "learning_rate": 9.492045680410068e-05, "loss": 0.0254, "step": 15000 }, { "grad_norm": 0.23140747845172882, "learning_rate": 9.491137625528436e-05, "loss": 0.0216, "step": 15010 }, { "grad_norm": 0.22999396920204163, "learning_rate": 9.490228803234215e-05, "loss": 0.0189, "step": 15020 }, { "grad_norm": 0.21836718916893005, "learning_rate": 9.489319213682701e-05, "loss": 0.0233, "step": 15030 }, { "grad_norm": 0.25499141216278076, "learning_rate": 9.488408857029316e-05, "loss": 0.0214, "step": 15040 }, { "grad_norm": 0.1613171547651291, "learning_rate": 9.487497733429616e-05, "loss": 0.0238, "step": 15050 }, { "grad_norm": 0.20813670754432678, "learning_rate": 9.486585843039286e-05, "loss": 0.019, "step": 15060 }, { "grad_norm": 0.2542968988418579, "learning_rate": 9.485673186014143e-05, "loss": 0.0224, "step": 15070 }, { "grad_norm": 0.23240984976291656, "learning_rate": 9.484759762510137e-05, "loss": 0.0192, "step": 15080 }, { "grad_norm": 0.39378201961517334, "learning_rate": 9.483845572683346e-05, "loss": 0.026, "step": 15090 }, { "grad_norm": 0.2466874122619629, "learning_rate": 9.48293061668998e-05, "loss": 0.0263, "step": 15100 }, { "grad_norm": 0.2522830069065094, "learning_rate": 9.48201489468638e-05, "loss": 0.0282, "step": 15110 }, { "grad_norm": 0.21019168198108673, "learning_rate": 9.481098406829016e-05, "loss": 0.0327, "step": 15120 }, { "grad_norm": 0.2797902822494507, "learning_rate": 9.480181153274495e-05, "loss": 0.0263, "step": 15130 }, { "grad_norm": 0.19547653198242188, "learning_rate": 9.479263134179548e-05, "loss": 0.0209, "step": 15140 }, { "grad_norm": 0.32124972343444824, "learning_rate": 9.478344349701039e-05, "loss": 0.0296, "step": 15150 }, { "grad_norm": 0.35368889570236206, "learning_rate": 9.477424799995964e-05, "loss": 0.0261, "step": 15160 }, { "grad_norm": 0.3034403920173645, "learning_rate": 9.476504485221448e-05, "loss": 0.0216, "step": 15170 }, { "grad_norm": 0.19231781363487244, "learning_rate": 9.475583405534748e-05, "loss": 0.0231, "step": 15180 }, { "grad_norm": 0.22240102291107178, "learning_rate": 9.474661561093251e-05, "loss": 0.0276, "step": 15190 }, { "grad_norm": 0.3882385194301605, "learning_rate": 9.473738952054478e-05, "loss": 0.0274, "step": 15200 }, { "grad_norm": 0.22958582639694214, "learning_rate": 9.472815578576073e-05, "loss": 0.0248, "step": 15210 }, { "grad_norm": 0.1870630532503128, "learning_rate": 9.471891440815817e-05, "loss": 0.0228, "step": 15220 }, { "grad_norm": 0.25836074352264404, "learning_rate": 9.470966538931621e-05, "loss": 0.0234, "step": 15230 }, { "grad_norm": 0.19006645679473877, "learning_rate": 9.470040873081525e-05, "loss": 0.0271, "step": 15240 }, { "grad_norm": 0.32344895601272583, "learning_rate": 9.469114443423698e-05, "loss": 0.0297, "step": 15250 }, { "grad_norm": 0.27396130561828613, "learning_rate": 9.468187250116445e-05, "loss": 0.0227, "step": 15260 }, { "grad_norm": 0.23511303961277008, "learning_rate": 9.467259293318197e-05, "loss": 0.0241, "step": 15270 }, { "grad_norm": 0.22904179990291595, "learning_rate": 9.466330573187514e-05, "loss": 0.0217, "step": 15280 }, { "grad_norm": 0.25826209783554077, "learning_rate": 9.46540108988309e-05, "loss": 0.0256, "step": 15290 }, { "grad_norm": 0.26361989974975586, "learning_rate": 9.46447084356375e-05, "loss": 0.0258, "step": 15300 }, { "grad_norm": 0.27907922863960266, "learning_rate": 9.463539834388447e-05, "loss": 0.032, "step": 15310 }, { "grad_norm": 0.26184314489364624, "learning_rate": 9.462608062516263e-05, "loss": 0.0297, "step": 15320 }, { "grad_norm": 0.3391001224517822, "learning_rate": 9.461675528106413e-05, "loss": 0.033, "step": 15330 }, { "grad_norm": 0.2479788362979889, "learning_rate": 9.460742231318244e-05, "loss": 0.0304, "step": 15340 }, { "grad_norm": 0.27849337458610535, "learning_rate": 9.459808172311229e-05, "loss": 0.0252, "step": 15350 }, { "grad_norm": 0.22447428107261658, "learning_rate": 9.458873351244972e-05, "loss": 0.0268, "step": 15360 }, { "grad_norm": 0.40371575951576233, "learning_rate": 9.457937768279211e-05, "loss": 0.0265, "step": 15370 }, { "grad_norm": 0.3596045970916748, "learning_rate": 9.45700142357381e-05, "loss": 0.0301, "step": 15380 }, { "grad_norm": 0.30764973163604736, "learning_rate": 9.456064317288765e-05, "loss": 0.0307, "step": 15390 }, { "grad_norm": 0.37838736176490784, "learning_rate": 9.455126449584201e-05, "loss": 0.0321, "step": 15400 }, { "grad_norm": 0.34422025084495544, "learning_rate": 9.454187820620375e-05, "loss": 0.0373, "step": 15410 }, { "grad_norm": 0.247725710272789, "learning_rate": 9.453248430557673e-05, "loss": 0.0311, "step": 15420 }, { "grad_norm": 0.22943779826164246, "learning_rate": 9.452308279556611e-05, "loss": 0.0247, "step": 15430 }, { "grad_norm": 0.3754647672176361, "learning_rate": 9.451367367777835e-05, "loss": 0.0328, "step": 15440 }, { "grad_norm": 0.33154991269111633, "learning_rate": 9.450425695382122e-05, "loss": 0.0294, "step": 15450 }, { "grad_norm": 0.3258044123649597, "learning_rate": 9.449483262530375e-05, "loss": 0.0292, "step": 15460 }, { "grad_norm": 0.24841289222240448, "learning_rate": 9.448540069383633e-05, "loss": 0.0267, "step": 15470 }, { "grad_norm": 0.280929297208786, "learning_rate": 9.447596116103061e-05, "loss": 0.0222, "step": 15480 }, { "grad_norm": 0.2375214397907257, "learning_rate": 9.446651402849955e-05, "loss": 0.0286, "step": 15490 }, { "grad_norm": 0.2486443668603897, "learning_rate": 9.44570592978574e-05, "loss": 0.0274, "step": 15500 }, { "grad_norm": 0.21351224184036255, "learning_rate": 9.444759697071972e-05, "loss": 0.0226, "step": 15510 }, { "grad_norm": 0.31631699204444885, "learning_rate": 9.443812704870336e-05, "loss": 0.0255, "step": 15520 }, { "grad_norm": 0.30617931485176086, "learning_rate": 9.442864953342649e-05, "loss": 0.0366, "step": 15530 }, { "grad_norm": 0.19523510336875916, "learning_rate": 9.441916442650852e-05, "loss": 0.0275, "step": 15540 }, { "grad_norm": 0.2630978226661682, "learning_rate": 9.440967172957023e-05, "loss": 0.0236, "step": 15550 }, { "grad_norm": 0.27368220686912537, "learning_rate": 9.440017144423364e-05, "loss": 0.023, "step": 15560 }, { "grad_norm": 0.22068637609481812, "learning_rate": 9.439066357212209e-05, "loss": 0.021, "step": 15570 }, { "grad_norm": 0.25164005160331726, "learning_rate": 9.438114811486022e-05, "loss": 0.024, "step": 15580 }, { "grad_norm": 0.26132190227508545, "learning_rate": 9.4371625074074e-05, "loss": 0.0233, "step": 15590 }, { "grad_norm": 0.22359289228916168, "learning_rate": 9.436209445139059e-05, "loss": 0.0269, "step": 15600 }, { "grad_norm": 0.2565712332725525, "learning_rate": 9.435255624843855e-05, "loss": 0.0258, "step": 15610 }, { "grad_norm": 0.2840675711631775, "learning_rate": 9.43430104668477e-05, "loss": 0.0221, "step": 15620 }, { "grad_norm": 0.2063858062028885, "learning_rate": 9.433345710824914e-05, "loss": 0.0201, "step": 15630 }, { "grad_norm": 0.21558105945587158, "learning_rate": 9.432389617427529e-05, "loss": 0.0202, "step": 15640 }, { "grad_norm": 0.25894463062286377, "learning_rate": 9.431432766655984e-05, "loss": 0.023, "step": 15650 }, { "grad_norm": 0.3443044126033783, "learning_rate": 9.430475158673778e-05, "loss": 0.0251, "step": 15660 }, { "grad_norm": 0.22329695522785187, "learning_rate": 9.429516793644542e-05, "loss": 0.023, "step": 15670 }, { "grad_norm": 0.2129734307527542, "learning_rate": 9.428557671732034e-05, "loss": 0.0221, "step": 15680 }, { "grad_norm": 0.226537823677063, "learning_rate": 9.42759779310014e-05, "loss": 0.0225, "step": 15690 }, { "grad_norm": 0.21131792664527893, "learning_rate": 9.426637157912879e-05, "loss": 0.0225, "step": 15700 }, { "grad_norm": 0.23749010264873505, "learning_rate": 9.425675766334397e-05, "loss": 0.0245, "step": 15710 }, { "grad_norm": 0.1795959174633026, "learning_rate": 9.424713618528968e-05, "loss": 0.0233, "step": 15720 }, { "grad_norm": 0.2826479971408844, "learning_rate": 9.423750714661e-05, "loss": 0.0225, "step": 15730 }, { "grad_norm": 0.24406734108924866, "learning_rate": 9.422787054895022e-05, "loss": 0.0223, "step": 15740 }, { "grad_norm": 0.30678218603134155, "learning_rate": 9.4218226393957e-05, "loss": 0.0274, "step": 15750 }, { "grad_norm": 0.22177821397781372, "learning_rate": 9.420857468327828e-05, "loss": 0.0238, "step": 15760 }, { "grad_norm": 0.25752773880958557, "learning_rate": 9.419891541856323e-05, "loss": 0.0275, "step": 15770 }, { "grad_norm": 0.24463215470314026, "learning_rate": 9.41892486014624e-05, "loss": 0.0257, "step": 15780 }, { "grad_norm": 0.23821939527988434, "learning_rate": 9.417957423362756e-05, "loss": 0.0247, "step": 15790 }, { "grad_norm": 0.35571667551994324, "learning_rate": 9.416989231671178e-05, "loss": 0.0252, "step": 15800 }, { "grad_norm": 0.2848917841911316, "learning_rate": 9.416020285236946e-05, "loss": 0.0247, "step": 15810 }, { "grad_norm": 0.3426716923713684, "learning_rate": 9.415050584225626e-05, "loss": 0.0288, "step": 15820 }, { "grad_norm": 0.26042088866233826, "learning_rate": 9.414080128802914e-05, "loss": 0.0271, "step": 15830 }, { "grad_norm": 0.24665158987045288, "learning_rate": 9.413108919134632e-05, "loss": 0.0225, "step": 15840 }, { "grad_norm": 0.15703995525836945, "learning_rate": 9.412136955386734e-05, "loss": 0.0226, "step": 15850 }, { "grad_norm": 0.2909108102321625, "learning_rate": 9.411164237725303e-05, "loss": 0.0216, "step": 15860 }, { "grad_norm": 0.20797601342201233, "learning_rate": 9.41019076631655e-05, "loss": 0.0205, "step": 15870 }, { "grad_norm": 0.260836124420166, "learning_rate": 9.409216541326815e-05, "loss": 0.0262, "step": 15880 }, { "grad_norm": 0.3238610029220581, "learning_rate": 9.408241562922564e-05, "loss": 0.0227, "step": 15890 }, { "grad_norm": 0.26216375827789307, "learning_rate": 9.407265831270395e-05, "loss": 0.0227, "step": 15900 }, { "grad_norm": 0.49739956855773926, "learning_rate": 9.406289346537035e-05, "loss": 0.0278, "step": 15910 }, { "grad_norm": 0.4832365810871124, "learning_rate": 9.405312108889339e-05, "loss": 0.0305, "step": 15920 }, { "grad_norm": 0.29385730624198914, "learning_rate": 9.404334118494288e-05, "loss": 0.0347, "step": 15930 }, { "grad_norm": 0.3248961269855499, "learning_rate": 9.403355375518995e-05, "loss": 0.0309, "step": 15940 }, { "grad_norm": 0.24160245060920715, "learning_rate": 9.4023758801307e-05, "loss": 0.0316, "step": 15950 }, { "grad_norm": 0.5201680064201355, "learning_rate": 9.401395632496774e-05, "loss": 0.0392, "step": 15960 }, { "grad_norm": 0.2660062611103058, "learning_rate": 9.400414632784711e-05, "loss": 0.0288, "step": 15970 }, { "grad_norm": 0.23983627557754517, "learning_rate": 9.39943288116214e-05, "loss": 0.0275, "step": 15980 }, { "grad_norm": 0.29688897728919983, "learning_rate": 9.398450377796815e-05, "loss": 0.0249, "step": 15990 }, { "grad_norm": 0.2278788834810257, "learning_rate": 9.397467122856616e-05, "loss": 0.0247, "step": 16000 }, { "grad_norm": 0.21504709124565125, "learning_rate": 9.396483116509558e-05, "loss": 0.022, "step": 16010 }, { "grad_norm": 0.20170243084430695, "learning_rate": 9.39549835892378e-05, "loss": 0.0332, "step": 16020 }, { "grad_norm": 0.2643338739871979, "learning_rate": 9.39451285026755e-05, "loss": 0.0222, "step": 16030 }, { "grad_norm": 0.20799148082733154, "learning_rate": 9.393526590709262e-05, "loss": 0.0263, "step": 16040 }, { "grad_norm": 0.23049567639827728, "learning_rate": 9.392539580417444e-05, "loss": 0.0275, "step": 16050 }, { "grad_norm": 0.28480064868927, "learning_rate": 9.391551819560747e-05, "loss": 0.0278, "step": 16060 }, { "grad_norm": 0.29662367701530457, "learning_rate": 9.390563308307955e-05, "loss": 0.0292, "step": 16070 }, { "grad_norm": 0.3465023934841156, "learning_rate": 9.389574046827974e-05, "loss": 0.0269, "step": 16080 }, { "grad_norm": 0.25530463457107544, "learning_rate": 9.388584035289845e-05, "loss": 0.0295, "step": 16090 }, { "grad_norm": 0.31320658326148987, "learning_rate": 9.387593273862732e-05, "loss": 0.0252, "step": 16100 }, { "grad_norm": 0.2814061641693115, "learning_rate": 9.386601762715929e-05, "loss": 0.0292, "step": 16110 }, { "grad_norm": 0.3492545783519745, "learning_rate": 9.38560950201886e-05, "loss": 0.0286, "step": 16120 }, { "grad_norm": 0.21864330768585205, "learning_rate": 9.384616491941071e-05, "loss": 0.0256, "step": 16130 }, { "grad_norm": 0.2554105818271637, "learning_rate": 9.383622732652245e-05, "loss": 0.0259, "step": 16140 }, { "grad_norm": 0.23201295733451843, "learning_rate": 9.382628224322187e-05, "loss": 0.0283, "step": 16150 }, { "grad_norm": 0.24489712715148926, "learning_rate": 9.381632967120829e-05, "loss": 0.0279, "step": 16160 }, { "grad_norm": 0.3146250545978546, "learning_rate": 9.380636961218235e-05, "loss": 0.0224, "step": 16170 }, { "grad_norm": 0.2389271855354309, "learning_rate": 9.379640206784597e-05, "loss": 0.0278, "step": 16180 }, { "grad_norm": 0.25793999433517456, "learning_rate": 9.378642703990229e-05, "loss": 0.0252, "step": 16190 }, { "grad_norm": 0.2886514961719513, "learning_rate": 9.37764445300558e-05, "loss": 0.0255, "step": 16200 }, { "grad_norm": 0.1915409415960312, "learning_rate": 9.376645454001222e-05, "loss": 0.0181, "step": 16210 }, { "grad_norm": 0.22627076506614685, "learning_rate": 9.375645707147858e-05, "loss": 0.0277, "step": 16220 }, { "grad_norm": 0.3127160668373108, "learning_rate": 9.374645212616316e-05, "loss": 0.0264, "step": 16230 }, { "grad_norm": 0.31248801946640015, "learning_rate": 9.373643970577555e-05, "loss": 0.0288, "step": 16240 }, { "grad_norm": 0.29894161224365234, "learning_rate": 9.372641981202659e-05, "loss": 0.0281, "step": 16250 }, { "grad_norm": 0.2326967716217041, "learning_rate": 9.37163924466284e-05, "loss": 0.0287, "step": 16260 }, { "grad_norm": 0.2774870693683624, "learning_rate": 9.370635761129438e-05, "loss": 0.0246, "step": 16270 }, { "grad_norm": 0.24844464659690857, "learning_rate": 9.36963153077392e-05, "loss": 0.0289, "step": 16280 }, { "grad_norm": 0.1950620412826538, "learning_rate": 9.368626553767888e-05, "loss": 0.029, "step": 16290 }, { "grad_norm": 0.2850017845630646, "learning_rate": 9.367620830283057e-05, "loss": 0.0266, "step": 16300 }, { "grad_norm": 0.29735615849494934, "learning_rate": 9.366614360491281e-05, "loss": 0.0308, "step": 16310 }, { "grad_norm": 0.2844884395599365, "learning_rate": 9.365607144564539e-05, "loss": 0.0286, "step": 16320 }, { "grad_norm": 0.24116669595241547, "learning_rate": 9.364599182674934e-05, "loss": 0.0304, "step": 16330 }, { "grad_norm": 0.2802724242210388, "learning_rate": 9.3635904749947e-05, "loss": 0.0279, "step": 16340 }, { "grad_norm": 0.2943790555000305, "learning_rate": 9.362581021696202e-05, "loss": 0.0295, "step": 16350 }, { "grad_norm": 0.22848200798034668, "learning_rate": 9.361570822951921e-05, "loss": 0.0266, "step": 16360 }, { "grad_norm": 0.3355381488800049, "learning_rate": 9.360559878934476e-05, "loss": 0.0283, "step": 16370 }, { "grad_norm": 0.3477747440338135, "learning_rate": 9.359548189816611e-05, "loss": 0.0371, "step": 16380 }, { "grad_norm": 0.5402414798736572, "learning_rate": 9.358535755771193e-05, "loss": 0.0306, "step": 16390 }, { "grad_norm": 0.2903274595737457, "learning_rate": 9.357522576971221e-05, "loss": 0.0266, "step": 16400 }, { "grad_norm": 0.26595693826675415, "learning_rate": 9.356508653589819e-05, "loss": 0.0309, "step": 16410 }, { "grad_norm": 0.3083755373954773, "learning_rate": 9.355493985800237e-05, "loss": 0.027, "step": 16420 }, { "grad_norm": 0.18608316779136658, "learning_rate": 9.354478573775857e-05, "loss": 0.0302, "step": 16430 }, { "grad_norm": 0.30678921937942505, "learning_rate": 9.353462417690186e-05, "loss": 0.0293, "step": 16440 }, { "grad_norm": 0.29712745547294617, "learning_rate": 9.352445517716853e-05, "loss": 0.03, "step": 16450 }, { "grad_norm": 0.3283911347389221, "learning_rate": 9.351427874029621e-05, "loss": 0.0285, "step": 16460 }, { "grad_norm": 0.22471868991851807, "learning_rate": 9.350409486802379e-05, "loss": 0.0279, "step": 16470 }, { "grad_norm": 0.48283499479293823, "learning_rate": 9.349390356209138e-05, "loss": 0.0252, "step": 16480 }, { "grad_norm": 0.2793569266796112, "learning_rate": 9.348370482424042e-05, "loss": 0.0296, "step": 16490 }, { "grad_norm": 0.26782211661338806, "learning_rate": 9.347349865621357e-05, "loss": 0.0257, "step": 16500 }, { "grad_norm": 0.3173290491104126, "learning_rate": 9.346328505975481e-05, "loss": 0.026, "step": 16510 }, { "grad_norm": 0.3059878647327423, "learning_rate": 9.345306403660936e-05, "loss": 0.024, "step": 16520 }, { "grad_norm": 0.3385074734687805, "learning_rate": 9.344283558852371e-05, "loss": 0.0228, "step": 16530 }, { "grad_norm": 0.3415232002735138, "learning_rate": 9.343259971724563e-05, "loss": 0.0258, "step": 16540 }, { "grad_norm": 0.3037389814853668, "learning_rate": 9.342235642452413e-05, "loss": 0.0303, "step": 16550 }, { "grad_norm": 0.18789075314998627, "learning_rate": 9.341210571210954e-05, "loss": 0.0265, "step": 16560 }, { "grad_norm": 0.26909947395324707, "learning_rate": 9.340184758175338e-05, "loss": 0.0249, "step": 16570 }, { "grad_norm": 0.23551107943058014, "learning_rate": 9.339158203520854e-05, "loss": 0.0298, "step": 16580 }, { "grad_norm": 0.24116002023220062, "learning_rate": 9.338130907422908e-05, "loss": 0.0229, "step": 16590 }, { "grad_norm": 0.20747777819633484, "learning_rate": 9.337102870057037e-05, "loss": 0.0222, "step": 16600 }, { "grad_norm": 0.29346737265586853, "learning_rate": 9.336074091598907e-05, "loss": 0.0261, "step": 16610 }, { "grad_norm": 0.2001093477010727, "learning_rate": 9.335044572224306e-05, "loss": 0.0243, "step": 16620 }, { "grad_norm": 0.23148173093795776, "learning_rate": 9.334014312109151e-05, "loss": 0.0225, "step": 16630 }, { "grad_norm": 0.1789635419845581, "learning_rate": 9.332983311429486e-05, "loss": 0.0215, "step": 16640 }, { "grad_norm": 0.2619292140007019, "learning_rate": 9.33195157036148e-05, "loss": 0.0267, "step": 16650 }, { "grad_norm": 0.19358953833580017, "learning_rate": 9.330919089081432e-05, "loss": 0.0207, "step": 16660 }, { "grad_norm": 0.2394278645515442, "learning_rate": 9.32988586776576e-05, "loss": 0.0205, "step": 16670 }, { "grad_norm": 0.3519458472728729, "learning_rate": 9.328851906591016e-05, "loss": 0.0232, "step": 16680 }, { "grad_norm": 0.2729833126068115, "learning_rate": 9.327817205733875e-05, "loss": 0.0281, "step": 16690 }, { "grad_norm": 0.4454149603843689, "learning_rate": 9.326781765371142e-05, "loss": 0.0285, "step": 16700 }, { "grad_norm": 0.1903945654630661, "learning_rate": 9.325745585679741e-05, "loss": 0.021, "step": 16710 }, { "grad_norm": 0.3056403398513794, "learning_rate": 9.32470866683673e-05, "loss": 0.0273, "step": 16720 }, { "grad_norm": 0.23792746663093567, "learning_rate": 9.323671009019288e-05, "loss": 0.0196, "step": 16730 }, { "grad_norm": 0.2401159405708313, "learning_rate": 9.322632612404725e-05, "loss": 0.0216, "step": 16740 }, { "grad_norm": 0.30400845408439636, "learning_rate": 9.321593477170471e-05, "loss": 0.0251, "step": 16750 }, { "grad_norm": 0.3234749138355255, "learning_rate": 9.320553603494088e-05, "loss": 0.029, "step": 16760 }, { "grad_norm": 0.2704163193702698, "learning_rate": 9.319512991553261e-05, "loss": 0.0298, "step": 16770 }, { "grad_norm": 0.23721525073051453, "learning_rate": 9.318471641525803e-05, "loss": 0.0238, "step": 16780 }, { "grad_norm": 0.3320481777191162, "learning_rate": 9.317429553589652e-05, "loss": 0.0273, "step": 16790 }, { "grad_norm": 0.30207568407058716, "learning_rate": 9.316386727922873e-05, "loss": 0.0296, "step": 16800 }, { "grad_norm": 0.2653302550315857, "learning_rate": 9.315343164703656e-05, "loss": 0.0295, "step": 16810 }, { "grad_norm": 0.32996705174446106, "learning_rate": 9.314298864110316e-05, "loss": 0.0266, "step": 16820 }, { "grad_norm": 0.2632696032524109, "learning_rate": 9.313253826321295e-05, "loss": 0.022, "step": 16830 }, { "grad_norm": 0.21970853209495544, "learning_rate": 9.312208051515165e-05, "loss": 0.0213, "step": 16840 }, { "grad_norm": 0.25400593876838684, "learning_rate": 9.311161539870618e-05, "loss": 0.0222, "step": 16850 }, { "grad_norm": 0.2737797498703003, "learning_rate": 9.310114291566474e-05, "loss": 0.0221, "step": 16860 }, { "grad_norm": 0.22405867278575897, "learning_rate": 9.309066306781679e-05, "loss": 0.0217, "step": 16870 }, { "grad_norm": 0.2783918082714081, "learning_rate": 9.308017585695306e-05, "loss": 0.0271, "step": 16880 }, { "grad_norm": 0.2961682379245758, "learning_rate": 9.306968128486552e-05, "loss": 0.0268, "step": 16890 }, { "grad_norm": 0.26839032769203186, "learning_rate": 9.30591793533474e-05, "loss": 0.0283, "step": 16900 }, { "grad_norm": 0.191897913813591, "learning_rate": 9.304867006419321e-05, "loss": 0.0242, "step": 16910 }, { "grad_norm": 0.23249883949756622, "learning_rate": 9.303815341919868e-05, "loss": 0.0285, "step": 16920 }, { "grad_norm": 0.2285514920949936, "learning_rate": 9.302762942016084e-05, "loss": 0.0225, "step": 16930 }, { "grad_norm": 0.21760523319244385, "learning_rate": 9.301709806887792e-05, "loss": 0.0223, "step": 16940 }, { "grad_norm": 0.29544419050216675, "learning_rate": 9.300655936714948e-05, "loss": 0.0224, "step": 16950 }, { "grad_norm": 0.27680522203445435, "learning_rate": 9.299601331677627e-05, "loss": 0.0277, "step": 16960 }, { "grad_norm": 0.3118705153465271, "learning_rate": 9.298545991956033e-05, "loss": 0.0281, "step": 16970 }, { "grad_norm": 0.3492506444454193, "learning_rate": 9.297489917730493e-05, "loss": 0.023, "step": 16980 }, { "grad_norm": 0.3854568898677826, "learning_rate": 9.296433109181464e-05, "loss": 0.0253, "step": 16990 }, { "grad_norm": 0.21172600984573364, "learning_rate": 9.295375566489523e-05, "loss": 0.0269, "step": 17000 }, { "grad_norm": 0.1611899733543396, "learning_rate": 9.294317289835379e-05, "loss": 0.0203, "step": 17010 }, { "grad_norm": 0.23469464480876923, "learning_rate": 9.293258279399859e-05, "loss": 0.0239, "step": 17020 }, { "grad_norm": 0.2184944748878479, "learning_rate": 9.292198535363919e-05, "loss": 0.0241, "step": 17030 }, { "grad_norm": 0.2588784694671631, "learning_rate": 9.291138057908641e-05, "loss": 0.0237, "step": 17040 }, { "grad_norm": 0.1937776505947113, "learning_rate": 9.290076847215234e-05, "loss": 0.0217, "step": 17050 }, { "grad_norm": 0.3323674201965332, "learning_rate": 9.289014903465025e-05, "loss": 0.0226, "step": 17060 }, { "grad_norm": 0.20258642733097076, "learning_rate": 9.287952226839475e-05, "loss": 0.0244, "step": 17070 }, { "grad_norm": 0.5509234070777893, "learning_rate": 9.286888817520164e-05, "loss": 0.0231, "step": 17080 }, { "grad_norm": 0.30711829662323, "learning_rate": 9.285824675688803e-05, "loss": 0.0199, "step": 17090 }, { "grad_norm": 0.18504266440868378, "learning_rate": 9.28475980152722e-05, "loss": 0.0251, "step": 17100 }, { "grad_norm": 0.28505638241767883, "learning_rate": 9.283694195217379e-05, "loss": 0.0257, "step": 17110 }, { "grad_norm": 0.2582760453224182, "learning_rate": 9.282627856941356e-05, "loss": 0.022, "step": 17120 }, { "grad_norm": 0.2084549367427826, "learning_rate": 9.281560786881363e-05, "loss": 0.0292, "step": 17130 }, { "grad_norm": 0.23069645464420319, "learning_rate": 9.280492985219733e-05, "loss": 0.0281, "step": 17140 }, { "grad_norm": 0.2774372696876526, "learning_rate": 9.279424452138924e-05, "loss": 0.0229, "step": 17150 }, { "grad_norm": 0.33392149209976196, "learning_rate": 9.278355187821517e-05, "loss": 0.0242, "step": 17160 }, { "grad_norm": 0.20806217193603516, "learning_rate": 9.277285192450224e-05, "loss": 0.0225, "step": 17170 }, { "grad_norm": 0.18746894598007202, "learning_rate": 9.276214466207875e-05, "loss": 0.0216, "step": 17180 }, { "grad_norm": 0.26723992824554443, "learning_rate": 9.275143009277427e-05, "loss": 0.0219, "step": 17190 }, { "grad_norm": 0.33547869324684143, "learning_rate": 9.274070821841964e-05, "loss": 0.0197, "step": 17200 }, { "grad_norm": 0.2792176306247711, "learning_rate": 9.272997904084696e-05, "loss": 0.0252, "step": 17210 }, { "grad_norm": 0.19615086913108826, "learning_rate": 9.271924256188951e-05, "loss": 0.0216, "step": 17220 }, { "grad_norm": 0.17272698879241943, "learning_rate": 9.270849878338189e-05, "loss": 0.0233, "step": 17230 }, { "grad_norm": 0.2485893815755844, "learning_rate": 9.269774770715991e-05, "loss": 0.0249, "step": 17240 }, { "grad_norm": 0.21903683245182037, "learning_rate": 9.268698933506061e-05, "loss": 0.0221, "step": 17250 }, { "grad_norm": 0.3031317889690399, "learning_rate": 9.267622366892235e-05, "loss": 0.0223, "step": 17260 }, { "grad_norm": 0.32002562284469604, "learning_rate": 9.266545071058465e-05, "loss": 0.0243, "step": 17270 }, { "grad_norm": 0.293346107006073, "learning_rate": 9.265467046188833e-05, "loss": 0.0267, "step": 17280 }, { "grad_norm": 0.23179320991039276, "learning_rate": 9.264388292467543e-05, "loss": 0.026, "step": 17290 }, { "grad_norm": 0.23556306958198547, "learning_rate": 9.263308810078926e-05, "loss": 0.0294, "step": 17300 }, { "grad_norm": 0.27298086881637573, "learning_rate": 9.262228599207434e-05, "loss": 0.0258, "step": 17310 }, { "grad_norm": 0.32608383893966675, "learning_rate": 9.261147660037647e-05, "loss": 0.0202, "step": 17320 }, { "grad_norm": 0.29095128178596497, "learning_rate": 9.26006599275427e-05, "loss": 0.0252, "step": 17330 }, { "grad_norm": 0.21895155310630798, "learning_rate": 9.258983597542124e-05, "loss": 0.0221, "step": 17340 }, { "grad_norm": 0.31819042563438416, "learning_rate": 9.257900474586167e-05, "loss": 0.0247, "step": 17350 }, { "grad_norm": 0.2983265221118927, "learning_rate": 9.256816624071471e-05, "loss": 0.0274, "step": 17360 }, { "grad_norm": 0.2241099625825882, "learning_rate": 9.25573204618324e-05, "loss": 0.0238, "step": 17370 }, { "grad_norm": 0.19234424829483032, "learning_rate": 9.254646741106796e-05, "loss": 0.0209, "step": 17380 }, { "grad_norm": 0.23255868256092072, "learning_rate": 9.253560709027589e-05, "loss": 0.0174, "step": 17390 }, { "grad_norm": 0.21318812668323517, "learning_rate": 9.252473950131192e-05, "loss": 0.0182, "step": 17400 }, { "grad_norm": 0.2524592876434326, "learning_rate": 9.251386464603302e-05, "loss": 0.0199, "step": 17410 }, { "grad_norm": 0.2758283317089081, "learning_rate": 9.250298252629741e-05, "loss": 0.0232, "step": 17420 }, { "grad_norm": 0.2702920436859131, "learning_rate": 9.249209314396454e-05, "loss": 0.0237, "step": 17430 }, { "grad_norm": 0.25283119082450867, "learning_rate": 9.248119650089513e-05, "loss": 0.0224, "step": 17440 }, { "grad_norm": 0.21021904051303864, "learning_rate": 9.247029259895108e-05, "loss": 0.0216, "step": 17450 }, { "grad_norm": 0.24374662339687347, "learning_rate": 9.24593814399956e-05, "loss": 0.0217, "step": 17460 }, { "grad_norm": 0.24932485818862915, "learning_rate": 9.244846302589309e-05, "loss": 0.0281, "step": 17470 }, { "grad_norm": 0.241657093167305, "learning_rate": 9.243753735850923e-05, "loss": 0.0248, "step": 17480 }, { "grad_norm": 0.23799660801887512, "learning_rate": 9.24266044397109e-05, "loss": 0.0221, "step": 17490 }, { "grad_norm": 0.30374521017074585, "learning_rate": 9.241566427136624e-05, "loss": 0.0197, "step": 17500 }, { "grad_norm": 0.17099490761756897, "learning_rate": 9.240471685534463e-05, "loss": 0.0227, "step": 17510 }, { "grad_norm": 0.26220473647117615, "learning_rate": 9.239376219351667e-05, "loss": 0.0226, "step": 17520 }, { "grad_norm": 0.3266932964324951, "learning_rate": 9.238280028775425e-05, "loss": 0.0245, "step": 17530 }, { "grad_norm": 0.24394740164279938, "learning_rate": 9.237183113993041e-05, "loss": 0.0266, "step": 17540 }, { "grad_norm": 0.2464722990989685, "learning_rate": 9.236085475191952e-05, "loss": 0.0249, "step": 17550 }, { "grad_norm": 0.3359009921550751, "learning_rate": 9.234987112559709e-05, "loss": 0.0275, "step": 17560 }, { "grad_norm": 0.2729979455471039, "learning_rate": 9.233888026283999e-05, "loss": 0.0233, "step": 17570 }, { "grad_norm": 0.2208847552537918, "learning_rate": 9.232788216552619e-05, "loss": 0.0292, "step": 17580 }, { "grad_norm": 0.20761540532112122, "learning_rate": 9.231687683553502e-05, "loss": 0.0213, "step": 17590 }, { "grad_norm": 0.15836307406425476, "learning_rate": 9.230586427474698e-05, "loss": 0.0178, "step": 17600 }, { "grad_norm": 0.36384671926498413, "learning_rate": 9.229484448504379e-05, "loss": 0.0224, "step": 17610 }, { "grad_norm": 0.2252160757780075, "learning_rate": 9.228381746830843e-05, "loss": 0.0212, "step": 17620 }, { "grad_norm": 0.32432785630226135, "learning_rate": 9.227278322642514e-05, "loss": 0.0225, "step": 17630 }, { "grad_norm": 0.19814340770244598, "learning_rate": 9.226174176127937e-05, "loss": 0.0244, "step": 17640 }, { "grad_norm": 0.20068567991256714, "learning_rate": 9.22506930747578e-05, "loss": 0.0206, "step": 17650 }, { "grad_norm": 0.28988543152809143, "learning_rate": 9.223963716874831e-05, "loss": 0.0242, "step": 17660 }, { "grad_norm": 0.25245246291160583, "learning_rate": 9.222857404514012e-05, "loss": 0.0284, "step": 17670 }, { "grad_norm": 0.2212437093257904, "learning_rate": 9.221750370582355e-05, "loss": 0.0231, "step": 17680 }, { "grad_norm": 0.27447178959846497, "learning_rate": 9.220642615269028e-05, "loss": 0.0202, "step": 17690 }, { "grad_norm": 0.2878243625164032, "learning_rate": 9.219534138763311e-05, "loss": 0.0298, "step": 17700 }, { "grad_norm": 0.17785677313804626, "learning_rate": 9.218424941254613e-05, "loss": 0.0237, "step": 17710 }, { "grad_norm": 0.30483219027519226, "learning_rate": 9.217315022932468e-05, "loss": 0.031, "step": 17720 }, { "grad_norm": 0.28830090165138245, "learning_rate": 9.216204383986528e-05, "loss": 0.0294, "step": 17730 }, { "grad_norm": 0.33323201537132263, "learning_rate": 9.215093024606574e-05, "loss": 0.0259, "step": 17740 }, { "grad_norm": 0.2769988775253296, "learning_rate": 9.213980944982506e-05, "loss": 0.0252, "step": 17750 }, { "grad_norm": 0.23259878158569336, "learning_rate": 9.212868145304346e-05, "loss": 0.0258, "step": 17760 }, { "grad_norm": 0.40520980954170227, "learning_rate": 9.211754625762241e-05, "loss": 0.0271, "step": 17770 }, { "grad_norm": 0.29957830905914307, "learning_rate": 9.210640386546463e-05, "loss": 0.0254, "step": 17780 }, { "grad_norm": 0.24731171131134033, "learning_rate": 9.209525427847405e-05, "loss": 0.0268, "step": 17790 }, { "grad_norm": 0.3840257525444031, "learning_rate": 9.208409749855583e-05, "loss": 0.0288, "step": 17800 }, { "grad_norm": 0.26244908571243286, "learning_rate": 9.207293352761633e-05, "loss": 0.0277, "step": 17810 }, { "grad_norm": 0.2860012352466583, "learning_rate": 9.206176236756319e-05, "loss": 0.028, "step": 17820 }, { "grad_norm": 0.3312965929508209, "learning_rate": 9.205058402030525e-05, "loss": 0.0251, "step": 17830 }, { "grad_norm": 0.27350595593452454, "learning_rate": 9.203939848775259e-05, "loss": 0.0243, "step": 17840 }, { "grad_norm": 0.300057053565979, "learning_rate": 9.202820577181652e-05, "loss": 0.0404, "step": 17850 }, { "grad_norm": 0.33404046297073364, "learning_rate": 9.201700587440953e-05, "loss": 0.025, "step": 17860 }, { "grad_norm": 0.21197180449962616, "learning_rate": 9.200579879744544e-05, "loss": 0.0218, "step": 17870 }, { "grad_norm": 0.25455716252326965, "learning_rate": 9.199458454283918e-05, "loss": 0.0265, "step": 17880 }, { "grad_norm": 0.2838894724845886, "learning_rate": 9.198336311250697e-05, "loss": 0.0196, "step": 17890 }, { "grad_norm": 0.2621755301952362, "learning_rate": 9.197213450836626e-05, "loss": 0.0264, "step": 17900 }, { "grad_norm": 0.21383404731750488, "learning_rate": 9.19608987323357e-05, "loss": 0.0217, "step": 17910 }, { "grad_norm": 0.18582575023174286, "learning_rate": 9.194965578633517e-05, "loss": 0.0271, "step": 17920 }, { "grad_norm": 0.3096994459629059, "learning_rate": 9.193840567228582e-05, "loss": 0.0281, "step": 17930 }, { "grad_norm": 0.26396945118904114, "learning_rate": 9.192714839210994e-05, "loss": 0.0244, "step": 17940 }, { "grad_norm": 0.2616465389728546, "learning_rate": 9.19158839477311e-05, "loss": 0.0239, "step": 17950 }, { "grad_norm": 0.20745326578617096, "learning_rate": 9.190461234107411e-05, "loss": 0.024, "step": 17960 }, { "grad_norm": 0.23413147032260895, "learning_rate": 9.189333357406496e-05, "loss": 0.0282, "step": 17970 }, { "grad_norm": 0.3255471885204315, "learning_rate": 9.188204764863089e-05, "loss": 0.0219, "step": 17980 }, { "grad_norm": 0.21496200561523438, "learning_rate": 9.187075456670033e-05, "loss": 0.0286, "step": 17990 }, { "grad_norm": 0.5349766612052917, "learning_rate": 9.1859454330203e-05, "loss": 0.0241, "step": 18000 }, { "grad_norm": 0.2771739065647125, "learning_rate": 9.18481469410698e-05, "loss": 0.0258, "step": 18010 }, { "grad_norm": 0.32810351252555847, "learning_rate": 9.183683240123281e-05, "loss": 0.0266, "step": 18020 }, { "grad_norm": 0.2048949897289276, "learning_rate": 9.182551071262541e-05, "loss": 0.0308, "step": 18030 }, { "grad_norm": 0.20849108695983887, "learning_rate": 9.181418187718218e-05, "loss": 0.0262, "step": 18040 }, { "grad_norm": 0.20626069605350494, "learning_rate": 9.180284589683888e-05, "loss": 0.024, "step": 18050 }, { "grad_norm": 0.17479054629802704, "learning_rate": 9.17915027735325e-05, "loss": 0.0248, "step": 18060 }, { "grad_norm": 0.2319326549768448, "learning_rate": 9.178015250920133e-05, "loss": 0.0216, "step": 18070 }, { "grad_norm": 0.29115426540374756, "learning_rate": 9.176879510578477e-05, "loss": 0.0197, "step": 18080 }, { "grad_norm": 0.23112262785434723, "learning_rate": 9.17574305652235e-05, "loss": 0.0211, "step": 18090 }, { "grad_norm": 0.2150081843137741, "learning_rate": 9.174605888945942e-05, "loss": 0.0223, "step": 18100 }, { "grad_norm": 0.6334205269813538, "learning_rate": 9.173468008043564e-05, "loss": 0.0229, "step": 18110 }, { "grad_norm": 0.9967339634895325, "learning_rate": 9.172329414009648e-05, "loss": 0.0326, "step": 18120 }, { "grad_norm": 0.2619599997997284, "learning_rate": 9.171190107038747e-05, "loss": 0.0231, "step": 18130 }, { "grad_norm": 0.296147882938385, "learning_rate": 9.170050087325541e-05, "loss": 0.0316, "step": 18140 }, { "grad_norm": 0.2630503177642822, "learning_rate": 9.168909355064824e-05, "loss": 0.0264, "step": 18150 }, { "grad_norm": 0.2258332073688507, "learning_rate": 9.167767910451519e-05, "loss": 0.0267, "step": 18160 }, { "grad_norm": 0.30102071166038513, "learning_rate": 9.166625753680669e-05, "loss": 0.0229, "step": 18170 }, { "grad_norm": 0.2187262773513794, "learning_rate": 9.165482884947431e-05, "loss": 0.0286, "step": 18180 }, { "grad_norm": 0.2531013786792755, "learning_rate": 9.164339304447098e-05, "loss": 0.0296, "step": 18190 }, { "grad_norm": 0.22159449756145477, "learning_rate": 9.163195012375072e-05, "loss": 0.0226, "step": 18200 }, { "grad_norm": 0.31822681427001953, "learning_rate": 9.16205000892688e-05, "loss": 0.0225, "step": 18210 }, { "grad_norm": 0.301421195268631, "learning_rate": 9.160904294298175e-05, "loss": 0.0235, "step": 18220 }, { "grad_norm": 0.3154153823852539, "learning_rate": 9.159757868684727e-05, "loss": 0.0277, "step": 18230 }, { "grad_norm": 0.24188007414340973, "learning_rate": 9.15861073228243e-05, "loss": 0.0262, "step": 18240 }, { "grad_norm": 0.2733789384365082, "learning_rate": 9.157462885287296e-05, "loss": 0.0262, "step": 18250 }, { "grad_norm": 0.23812457919120789, "learning_rate": 9.156314327895461e-05, "loss": 0.0247, "step": 18260 }, { "grad_norm": 0.18854312598705292, "learning_rate": 9.155165060303185e-05, "loss": 0.0228, "step": 18270 }, { "grad_norm": 0.19789950549602509, "learning_rate": 9.154015082706841e-05, "loss": 0.0209, "step": 18280 }, { "grad_norm": 0.3562614619731903, "learning_rate": 9.152864395302936e-05, "loss": 0.0284, "step": 18290 }, { "grad_norm": 0.3520338535308838, "learning_rate": 9.151712998288085e-05, "loss": 0.0274, "step": 18300 }, { "grad_norm": 0.14128467440605164, "learning_rate": 9.150560891859031e-05, "loss": 0.0204, "step": 18310 }, { "grad_norm": 0.23868748545646667, "learning_rate": 9.14940807621264e-05, "loss": 0.0188, "step": 18320 }, { "grad_norm": 0.2072685956954956, "learning_rate": 9.148254551545894e-05, "loss": 0.0251, "step": 18330 }, { "grad_norm": 0.24290594458580017, "learning_rate": 9.147100318055901e-05, "loss": 0.0218, "step": 18340 }, { "grad_norm": 0.2069670855998993, "learning_rate": 9.145945375939888e-05, "loss": 0.0182, "step": 18350 }, { "grad_norm": 0.28751856088638306, "learning_rate": 9.144789725395203e-05, "loss": 0.0209, "step": 18360 }, { "grad_norm": 0.261595755815506, "learning_rate": 9.14363336661931e-05, "loss": 0.024, "step": 18370 }, { "grad_norm": 0.2525525689125061, "learning_rate": 9.142476299809806e-05, "loss": 0.0205, "step": 18380 }, { "grad_norm": 0.2183155119419098, "learning_rate": 9.1413185251644e-05, "loss": 0.0227, "step": 18390 }, { "grad_norm": 0.1964399814605713, "learning_rate": 9.140160042880923e-05, "loss": 0.0242, "step": 18400 }, { "grad_norm": 0.2499321848154068, "learning_rate": 9.139000853157327e-05, "loss": 0.0316, "step": 18410 }, { "grad_norm": 0.2200315296649933, "learning_rate": 9.137840956191688e-05, "loss": 0.0219, "step": 18420 }, { "grad_norm": 0.1891113966703415, "learning_rate": 9.136680352182199e-05, "loss": 0.0222, "step": 18430 }, { "grad_norm": 0.2535722255706787, "learning_rate": 9.135519041327177e-05, "loss": 0.0261, "step": 18440 }, { "grad_norm": 0.2140987664461136, "learning_rate": 9.134357023825058e-05, "loss": 0.0224, "step": 18450 }, { "grad_norm": 0.3280327320098877, "learning_rate": 9.133194299874398e-05, "loss": 0.0277, "step": 18460 }, { "grad_norm": 0.23092466592788696, "learning_rate": 9.132030869673876e-05, "loss": 0.023, "step": 18470 }, { "grad_norm": 0.28355515003204346, "learning_rate": 9.130866733422288e-05, "loss": 0.0239, "step": 18480 }, { "grad_norm": 0.20973797142505646, "learning_rate": 9.129701891318556e-05, "loss": 0.0272, "step": 18490 }, { "grad_norm": 0.33312052488327026, "learning_rate": 9.128536343561718e-05, "loss": 0.0257, "step": 18500 }, { "grad_norm": 0.32931336760520935, "learning_rate": 9.127370090350934e-05, "loss": 0.0285, "step": 18510 }, { "grad_norm": 0.28975993394851685, "learning_rate": 9.126203131885487e-05, "loss": 0.0287, "step": 18520 }, { "grad_norm": 0.23027139902114868, "learning_rate": 9.125035468364775e-05, "loss": 0.0266, "step": 18530 }, { "grad_norm": 0.2967835068702698, "learning_rate": 9.123867099988322e-05, "loss": 0.028, "step": 18540 }, { "grad_norm": 0.2921924889087677, "learning_rate": 9.122698026955769e-05, "loss": 0.0218, "step": 18550 }, { "grad_norm": 0.27313733100891113, "learning_rate": 9.12152824946688e-05, "loss": 0.0288, "step": 18560 }, { "grad_norm": 0.3052126169204712, "learning_rate": 9.120357767721538e-05, "loss": 0.0302, "step": 18570 }, { "grad_norm": 0.25991445779800415, "learning_rate": 9.119186581919745e-05, "loss": 0.0241, "step": 18580 }, { "grad_norm": 0.32471218705177307, "learning_rate": 9.118014692261624e-05, "loss": 0.0265, "step": 18590 }, { "grad_norm": 0.2923130691051483, "learning_rate": 9.116842098947422e-05, "loss": 0.0269, "step": 18600 }, { "grad_norm": 0.28718048334121704, "learning_rate": 9.115668802177499e-05, "loss": 0.0265, "step": 18610 }, { "grad_norm": 0.2503790259361267, "learning_rate": 9.114494802152342e-05, "loss": 0.0191, "step": 18620 }, { "grad_norm": 0.27321717143058777, "learning_rate": 9.113320099072555e-05, "loss": 0.0192, "step": 18630 }, { "grad_norm": 0.26555997133255005, "learning_rate": 9.112144693138864e-05, "loss": 0.0188, "step": 18640 }, { "grad_norm": 0.2414524406194687, "learning_rate": 9.110968584552111e-05, "loss": 0.0211, "step": 18650 }, { "grad_norm": 0.233529195189476, "learning_rate": 9.109791773513264e-05, "loss": 0.0202, "step": 18660 }, { "grad_norm": 0.25848710536956787, "learning_rate": 9.108614260223403e-05, "loss": 0.0295, "step": 18670 }, { "grad_norm": 0.15290167927742004, "learning_rate": 9.107436044883738e-05, "loss": 0.0252, "step": 18680 }, { "grad_norm": 0.22545555233955383, "learning_rate": 9.10625712769559e-05, "loss": 0.0267, "step": 18690 }, { "grad_norm": 0.32614070177078247, "learning_rate": 9.105077508860406e-05, "loss": 0.0245, "step": 18700 }, { "grad_norm": 0.30078059434890747, "learning_rate": 9.103897188579751e-05, "loss": 0.0257, "step": 18710 }, { "grad_norm": 0.2435106486082077, "learning_rate": 9.102716167055308e-05, "loss": 0.0254, "step": 18720 }, { "grad_norm": 0.22691428661346436, "learning_rate": 9.10153444448888e-05, "loss": 0.0221, "step": 18730 }, { "grad_norm": 0.22265012562274933, "learning_rate": 9.100352021082393e-05, "loss": 0.0252, "step": 18740 }, { "grad_norm": 0.24961860477924347, "learning_rate": 9.099168897037891e-05, "loss": 0.0227, "step": 18750 }, { "grad_norm": 0.15787820518016815, "learning_rate": 9.097985072557538e-05, "loss": 0.0228, "step": 18760 }, { "grad_norm": 0.2536594867706299, "learning_rate": 9.096800547843615e-05, "loss": 0.024, "step": 18770 }, { "grad_norm": 0.327959805727005, "learning_rate": 9.095615323098526e-05, "loss": 0.0222, "step": 18780 }, { "grad_norm": 0.24749942123889923, "learning_rate": 9.094429398524795e-05, "loss": 0.0209, "step": 18790 }, { "grad_norm": 0.3288480043411255, "learning_rate": 9.093242774325061e-05, "loss": 0.023, "step": 18800 }, { "grad_norm": 0.24280959367752075, "learning_rate": 9.092055450702088e-05, "loss": 0.0254, "step": 18810 }, { "grad_norm": 0.3543587923049927, "learning_rate": 9.090867427858756e-05, "loss": 0.0262, "step": 18820 }, { "grad_norm": 0.2910216152667999, "learning_rate": 9.089678705998066e-05, "loss": 0.021, "step": 18830 }, { "grad_norm": 0.2006133496761322, "learning_rate": 9.088489285323139e-05, "loss": 0.0199, "step": 18840 }, { "grad_norm": 0.26169726252555847, "learning_rate": 9.087299166037212e-05, "loss": 0.0207, "step": 18850 }, { "grad_norm": 0.18433444201946259, "learning_rate": 9.086108348343647e-05, "loss": 0.0188, "step": 18860 }, { "grad_norm": 0.17593646049499512, "learning_rate": 9.08491683244592e-05, "loss": 0.0212, "step": 18870 }, { "grad_norm": 0.22935689985752106, "learning_rate": 9.08372461854763e-05, "loss": 0.0237, "step": 18880 }, { "grad_norm": 0.18811504542827606, "learning_rate": 9.082531706852492e-05, "loss": 0.0176, "step": 18890 }, { "grad_norm": 0.21344634890556335, "learning_rate": 9.081338097564342e-05, "loss": 0.0224, "step": 18900 }, { "grad_norm": 0.29025107622146606, "learning_rate": 9.080143790887137e-05, "loss": 0.0213, "step": 18910 }, { "grad_norm": 0.2883836328983307, "learning_rate": 9.07894878702495e-05, "loss": 0.0276, "step": 18920 }, { "grad_norm": 0.25536859035491943, "learning_rate": 9.077753086181974e-05, "loss": 0.0215, "step": 18930 }, { "grad_norm": 0.21017472445964813, "learning_rate": 9.076556688562524e-05, "loss": 0.0217, "step": 18940 }, { "grad_norm": 0.24542339146137238, "learning_rate": 9.075359594371029e-05, "loss": 0.0196, "step": 18950 }, { "grad_norm": 0.29242682456970215, "learning_rate": 9.07416180381204e-05, "loss": 0.0253, "step": 18960 }, { "grad_norm": 0.23785562813282013, "learning_rate": 9.072963317090228e-05, "loss": 0.022, "step": 18970 }, { "grad_norm": 0.1502210795879364, "learning_rate": 9.071764134410382e-05, "loss": 0.0202, "step": 18980 }, { "grad_norm": 0.18187958002090454, "learning_rate": 9.070564255977407e-05, "loss": 0.0173, "step": 18990 }, { "grad_norm": 0.21067626774311066, "learning_rate": 9.06936368199633e-05, "loss": 0.0184, "step": 19000 }, { "grad_norm": 0.2925430238246918, "learning_rate": 9.0681624126723e-05, "loss": 0.0208, "step": 19010 }, { "grad_norm": 0.26509058475494385, "learning_rate": 9.066960448210576e-05, "loss": 0.0258, "step": 19020 }, { "grad_norm": 0.357952356338501, "learning_rate": 9.065757788816543e-05, "loss": 0.0247, "step": 19030 }, { "grad_norm": 0.26483362913131714, "learning_rate": 9.064554434695705e-05, "loss": 0.0219, "step": 19040 }, { "grad_norm": 0.19425641000270844, "learning_rate": 9.063350386053677e-05, "loss": 0.0224, "step": 19050 }, { "grad_norm": 0.3246089518070221, "learning_rate": 9.062145643096202e-05, "loss": 0.0215, "step": 19060 }, { "grad_norm": 0.28926947712898254, "learning_rate": 9.060940206029136e-05, "loss": 0.0228, "step": 19070 }, { "grad_norm": 0.20744740962982178, "learning_rate": 9.059734075058457e-05, "loss": 0.0198, "step": 19080 }, { "grad_norm": 0.2960156798362732, "learning_rate": 9.058527250390257e-05, "loss": 0.0195, "step": 19090 }, { "grad_norm": 0.26484736800193787, "learning_rate": 9.057319732230752e-05, "loss": 0.0192, "step": 19100 }, { "grad_norm": 0.22578749060630798, "learning_rate": 9.056111520786273e-05, "loss": 0.0193, "step": 19110 }, { "grad_norm": 0.15966959297657013, "learning_rate": 9.054902616263268e-05, "loss": 0.0187, "step": 19120 }, { "grad_norm": 0.2835562527179718, "learning_rate": 9.05369301886831e-05, "loss": 0.019, "step": 19130 }, { "grad_norm": 0.20320965349674225, "learning_rate": 9.052482728808083e-05, "loss": 0.0205, "step": 19140 }, { "grad_norm": 0.2024783492088318, "learning_rate": 9.051271746289391e-05, "loss": 0.0208, "step": 19150 }, { "grad_norm": 0.2849581241607666, "learning_rate": 9.050060071519162e-05, "loss": 0.0235, "step": 19160 }, { "grad_norm": 0.16197721660137177, "learning_rate": 9.048847704704437e-05, "loss": 0.0204, "step": 19170 }, { "grad_norm": 0.25219962000846863, "learning_rate": 9.047634646052376e-05, "loss": 0.0215, "step": 19180 }, { "grad_norm": 0.213912233710289, "learning_rate": 9.046420895770256e-05, "loss": 0.0214, "step": 19190 }, { "grad_norm": 0.22491829097270966, "learning_rate": 9.045206454065473e-05, "loss": 0.0214, "step": 19200 }, { "grad_norm": 0.24232536554336548, "learning_rate": 9.043991321145546e-05, "loss": 0.0206, "step": 19210 }, { "grad_norm": 0.3069482147693634, "learning_rate": 9.042775497218105e-05, "loss": 0.026, "step": 19220 }, { "grad_norm": 0.4542599618434906, "learning_rate": 9.041558982490901e-05, "loss": 0.0271, "step": 19230 }, { "grad_norm": 0.2817145884037018, "learning_rate": 9.040341777171805e-05, "loss": 0.024, "step": 19240 }, { "grad_norm": 0.253379225730896, "learning_rate": 9.039123881468802e-05, "loss": 0.028, "step": 19250 }, { "grad_norm": 0.29403117299079895, "learning_rate": 9.037905295589998e-05, "loss": 0.0303, "step": 19260 }, { "grad_norm": 0.25689196586608887, "learning_rate": 9.036686019743617e-05, "loss": 0.0224, "step": 19270 }, { "grad_norm": 0.3132834732532501, "learning_rate": 9.035466054137997e-05, "loss": 0.0268, "step": 19280 }, { "grad_norm": 0.44622722268104553, "learning_rate": 9.0342453989816e-05, "loss": 0.024, "step": 19290 }, { "grad_norm": 0.22423385083675385, "learning_rate": 9.033024054483e-05, "loss": 0.0201, "step": 19300 }, { "grad_norm": 0.2311951071023941, "learning_rate": 9.031802020850894e-05, "loss": 0.0214, "step": 19310 }, { "grad_norm": 0.28402256965637207, "learning_rate": 9.030579298294092e-05, "loss": 0.0174, "step": 19320 }, { "grad_norm": 0.2192980796098709, "learning_rate": 9.029355887021524e-05, "loss": 0.0223, "step": 19330 }, { "grad_norm": 0.15537013113498688, "learning_rate": 9.028131787242238e-05, "loss": 0.0222, "step": 19340 }, { "grad_norm": 0.24947184324264526, "learning_rate": 9.026906999165399e-05, "loss": 0.0199, "step": 19350 }, { "grad_norm": 0.22368048131465912, "learning_rate": 9.025681523000291e-05, "loss": 0.0214, "step": 19360 }, { "grad_norm": 0.2482028752565384, "learning_rate": 9.024455358956315e-05, "loss": 0.0204, "step": 19370 }, { "grad_norm": 0.2252022624015808, "learning_rate": 9.023228507242984e-05, "loss": 0.0206, "step": 19380 }, { "grad_norm": 0.2657785713672638, "learning_rate": 9.022000968069937e-05, "loss": 0.0247, "step": 19390 }, { "grad_norm": 0.2900642156600952, "learning_rate": 9.020772741646928e-05, "loss": 0.0257, "step": 19400 }, { "grad_norm": 0.3620211184024811, "learning_rate": 9.019543828183826e-05, "loss": 0.0287, "step": 19410 }, { "grad_norm": 0.21383444964885712, "learning_rate": 9.018314227890616e-05, "loss": 0.0214, "step": 19420 }, { "grad_norm": 0.2339257001876831, "learning_rate": 9.017083940977408e-05, "loss": 0.0225, "step": 19430 }, { "grad_norm": 0.2737490236759186, "learning_rate": 9.015852967654422e-05, "loss": 0.0224, "step": 19440 }, { "grad_norm": 0.2753620147705078, "learning_rate": 9.014621308131996e-05, "loss": 0.0242, "step": 19450 }, { "grad_norm": 0.2906714379787445, "learning_rate": 9.01338896262059e-05, "loss": 0.0174, "step": 19460 }, { "grad_norm": 0.27779194712638855, "learning_rate": 9.012155931330777e-05, "loss": 0.0268, "step": 19470 }, { "grad_norm": 0.21307647228240967, "learning_rate": 9.010922214473246e-05, "loss": 0.03, "step": 19480 }, { "grad_norm": 0.259922057390213, "learning_rate": 9.009687812258808e-05, "loss": 0.0314, "step": 19490 }, { "grad_norm": 0.47821542620658875, "learning_rate": 9.00845272489839e-05, "loss": 0.0242, "step": 19500 }, { "grad_norm": 0.2383827120065689, "learning_rate": 9.007216952603031e-05, "loss": 0.0226, "step": 19510 }, { "grad_norm": 0.2403079867362976, "learning_rate": 9.005980495583894e-05, "loss": 0.0216, "step": 19520 }, { "grad_norm": 0.26981231570243835, "learning_rate": 9.004743354052252e-05, "loss": 0.0236, "step": 19530 }, { "grad_norm": 0.22654901444911957, "learning_rate": 9.003505528219503e-05, "loss": 0.0195, "step": 19540 }, { "grad_norm": 0.2406911700963974, "learning_rate": 9.002267018297154e-05, "loss": 0.0257, "step": 19550 }, { "grad_norm": 0.2417609989643097, "learning_rate": 9.001027824496834e-05, "loss": 0.0223, "step": 19560 }, { "grad_norm": 0.30814820528030396, "learning_rate": 8.999787947030287e-05, "loss": 0.0267, "step": 19570 }, { "grad_norm": 0.34548279643058777, "learning_rate": 8.998547386109376e-05, "loss": 0.024, "step": 19580 }, { "grad_norm": 0.5371661186218262, "learning_rate": 8.997306141946073e-05, "loss": 0.0237, "step": 19590 }, { "grad_norm": 0.3292091190814972, "learning_rate": 8.996064214752481e-05, "loss": 0.0205, "step": 19600 }, { "grad_norm": 0.2836255431175232, "learning_rate": 8.994821604740806e-05, "loss": 0.0193, "step": 19610 }, { "grad_norm": 0.29031333327293396, "learning_rate": 8.993578312123377e-05, "loss": 0.0222, "step": 19620 }, { "grad_norm": 0.2212623655796051, "learning_rate": 8.992334337112639e-05, "loss": 0.0192, "step": 19630 }, { "grad_norm": 0.1872919648885727, "learning_rate": 8.991089679921154e-05, "loss": 0.016, "step": 19640 }, { "grad_norm": 0.18207989633083344, "learning_rate": 8.989844340761599e-05, "loss": 0.0178, "step": 19650 }, { "grad_norm": 0.24840578436851501, "learning_rate": 8.988598319846768e-05, "loss": 0.0178, "step": 19660 }, { "grad_norm": 0.27135762572288513, "learning_rate": 8.987351617389574e-05, "loss": 0.0224, "step": 19670 }, { "grad_norm": 0.27936285734176636, "learning_rate": 8.98610423360304e-05, "loss": 0.0169, "step": 19680 }, { "grad_norm": 0.21589809656143188, "learning_rate": 8.984856168700317e-05, "loss": 0.0189, "step": 19690 }, { "grad_norm": 0.28225985169410706, "learning_rate": 8.983607422894658e-05, "loss": 0.0236, "step": 19700 }, { "grad_norm": 0.18351392447948456, "learning_rate": 8.982357996399442e-05, "loss": 0.0182, "step": 19710 }, { "grad_norm": 0.23356188833713531, "learning_rate": 8.981107889428164e-05, "loss": 0.0199, "step": 19720 }, { "grad_norm": 0.24349546432495117, "learning_rate": 8.979857102194428e-05, "loss": 0.0186, "step": 19730 }, { "grad_norm": 0.27955666184425354, "learning_rate": 8.978605634911968e-05, "loss": 0.021, "step": 19740 }, { "grad_norm": 0.1317054182291031, "learning_rate": 8.977353487794616e-05, "loss": 0.0248, "step": 19750 }, { "grad_norm": 0.2162725031375885, "learning_rate": 8.976100661056334e-05, "loss": 0.0235, "step": 19760 }, { "grad_norm": 0.2400175929069519, "learning_rate": 8.974847154911197e-05, "loss": 0.0218, "step": 19770 }, { "grad_norm": 0.3066961169242859, "learning_rate": 8.973592969573393e-05, "loss": 0.0249, "step": 19780 }, { "grad_norm": 0.2223401963710785, "learning_rate": 8.972338105257228e-05, "loss": 0.0254, "step": 19790 }, { "grad_norm": 0.26803815364837646, "learning_rate": 8.971082562177125e-05, "loss": 0.0212, "step": 19800 }, { "grad_norm": 0.24363774061203003, "learning_rate": 8.96982634054762e-05, "loss": 0.0209, "step": 19810 }, { "grad_norm": 0.30048665404319763, "learning_rate": 8.96856944058337e-05, "loss": 0.0211, "step": 19820 }, { "grad_norm": 0.23962722718715668, "learning_rate": 8.967311862499144e-05, "loss": 0.0243, "step": 19830 }, { "grad_norm": 0.25199535489082336, "learning_rate": 8.966053606509825e-05, "loss": 0.0233, "step": 19840 }, { "grad_norm": 0.2656359374523163, "learning_rate": 8.964794672830417e-05, "loss": 0.0233, "step": 19850 }, { "grad_norm": 0.25133925676345825, "learning_rate": 8.963535061676038e-05, "loss": 0.0226, "step": 19860 }, { "grad_norm": 0.2723986506462097, "learning_rate": 8.962274773261918e-05, "loss": 0.0217, "step": 19870 }, { "grad_norm": 0.28205177187919617, "learning_rate": 8.961013807803409e-05, "loss": 0.0212, "step": 19880 }, { "grad_norm": 0.22158682346343994, "learning_rate": 8.959752165515973e-05, "loss": 0.0229, "step": 19890 }, { "grad_norm": 0.2047394961118698, "learning_rate": 8.958489846615193e-05, "loss": 0.0236, "step": 19900 }, { "grad_norm": 0.17211754620075226, "learning_rate": 8.957226851316762e-05, "loss": 0.0198, "step": 19910 }, { "grad_norm": 0.3390968441963196, "learning_rate": 8.955963179836493e-05, "loss": 0.0211, "step": 19920 }, { "grad_norm": 0.29032260179519653, "learning_rate": 8.954698832390312e-05, "loss": 0.0224, "step": 19930 }, { "grad_norm": 0.2521374821662903, "learning_rate": 8.953433809194263e-05, "loss": 0.0208, "step": 19940 }, { "grad_norm": 0.3048159182071686, "learning_rate": 8.9521681104645e-05, "loss": 0.0263, "step": 19950 }, { "grad_norm": 0.294845849275589, "learning_rate": 8.9509017364173e-05, "loss": 0.0221, "step": 19960 }, { "grad_norm": 0.26413774490356445, "learning_rate": 8.949634687269052e-05, "loss": 0.0217, "step": 19970 }, { "grad_norm": 0.2583146393299103, "learning_rate": 8.948366963236259e-05, "loss": 0.0209, "step": 19980 }, { "grad_norm": 0.2436736822128296, "learning_rate": 8.947098564535538e-05, "loss": 0.0229, "step": 19990 }, { "grad_norm": 0.27061977982521057, "learning_rate": 8.945829491383627e-05, "loss": 0.024, "step": 20000 }, { "grad_norm": 0.237360417842865, "learning_rate": 8.944559743997374e-05, "loss": 0.0219, "step": 20010 }, { "grad_norm": 0.2077520489692688, "learning_rate": 8.943289322593746e-05, "loss": 0.0156, "step": 20020 }, { "grad_norm": 0.19068431854248047, "learning_rate": 8.942018227389821e-05, "loss": 0.0207, "step": 20030 }, { "grad_norm": 0.20571379363536835, "learning_rate": 8.940746458602795e-05, "loss": 0.0201, "step": 20040 }, { "grad_norm": 0.20759248733520508, "learning_rate": 8.939474016449979e-05, "loss": 0.0235, "step": 20050 }, { "grad_norm": 0.2738862633705139, "learning_rate": 8.938200901148799e-05, "loss": 0.0247, "step": 20060 }, { "grad_norm": 0.2626740634441376, "learning_rate": 8.936927112916795e-05, "loss": 0.024, "step": 20070 }, { "grad_norm": 0.209814190864563, "learning_rate": 8.935652651971622e-05, "loss": 0.0203, "step": 20080 }, { "grad_norm": 0.3105640113353729, "learning_rate": 8.934377518531052e-05, "loss": 0.0212, "step": 20090 }, { "grad_norm": 0.3040030896663666, "learning_rate": 8.933101712812967e-05, "loss": 0.0209, "step": 20100 }, { "grad_norm": 0.17084507644176483, "learning_rate": 8.931825235035374e-05, "loss": 0.0196, "step": 20110 }, { "grad_norm": 0.1898270845413208, "learning_rate": 8.930548085416382e-05, "loss": 0.0191, "step": 20120 }, { "grad_norm": 0.32351192831993103, "learning_rate": 8.92927026417422e-05, "loss": 0.0195, "step": 20130 }, { "grad_norm": 0.29283857345581055, "learning_rate": 8.92799177152724e-05, "loss": 0.028, "step": 20140 }, { "grad_norm": 0.21613408625125885, "learning_rate": 8.926712607693895e-05, "loss": 0.0219, "step": 20150 }, { "grad_norm": 0.21639280021190643, "learning_rate": 8.925432772892762e-05, "loss": 0.0212, "step": 20160 }, { "grad_norm": 0.30287161469459534, "learning_rate": 8.924152267342529e-05, "loss": 0.0197, "step": 20170 }, { "grad_norm": 0.2345978319644928, "learning_rate": 8.922871091261998e-05, "loss": 0.0231, "step": 20180 }, { "grad_norm": 0.2737899422645569, "learning_rate": 8.92158924487009e-05, "loss": 0.0211, "step": 20190 }, { "grad_norm": 0.21746636927127838, "learning_rate": 8.920306728385834e-05, "loss": 0.0242, "step": 20200 }, { "grad_norm": 0.287166029214859, "learning_rate": 8.919023542028379e-05, "loss": 0.0213, "step": 20210 }, { "grad_norm": 0.23594796657562256, "learning_rate": 8.917739686016988e-05, "loss": 0.0218, "step": 20220 }, { "grad_norm": 0.1853283941745758, "learning_rate": 8.916455160571033e-05, "loss": 0.0186, "step": 20230 }, { "grad_norm": 0.21769534051418304, "learning_rate": 8.915169965910008e-05, "loss": 0.0233, "step": 20240 }, { "grad_norm": 0.18875305354595184, "learning_rate": 8.913884102253514e-05, "loss": 0.019, "step": 20250 }, { "grad_norm": 0.24849024415016174, "learning_rate": 8.912597569821273e-05, "loss": 0.0228, "step": 20260 }, { "grad_norm": 0.1781679093837738, "learning_rate": 8.911310368833118e-05, "loss": 0.0208, "step": 20270 }, { "grad_norm": 0.2551952302455902, "learning_rate": 8.910022499508994e-05, "loss": 0.0192, "step": 20280 }, { "grad_norm": 0.16808190941810608, "learning_rate": 8.908733962068965e-05, "loss": 0.0174, "step": 20290 }, { "grad_norm": 0.2070717215538025, "learning_rate": 8.907444756733207e-05, "loss": 0.0176, "step": 20300 }, { "grad_norm": 0.28265613317489624, "learning_rate": 8.906154883722006e-05, "loss": 0.0219, "step": 20310 }, { "grad_norm": 0.2702879309654236, "learning_rate": 8.904864343255773e-05, "loss": 0.019, "step": 20320 }, { "grad_norm": 0.17925113439559937, "learning_rate": 8.90357313555502e-05, "loss": 0.0204, "step": 20330 }, { "grad_norm": 0.25360822677612305, "learning_rate": 8.90228126084038e-05, "loss": 0.0246, "step": 20340 }, { "grad_norm": 0.16730234026908875, "learning_rate": 8.900988719332601e-05, "loss": 0.0226, "step": 20350 }, { "grad_norm": 0.23204463720321655, "learning_rate": 8.899695511252542e-05, "loss": 0.0195, "step": 20360 }, { "grad_norm": 0.2155180126428604, "learning_rate": 8.898401636821176e-05, "loss": 0.0198, "step": 20370 }, { "grad_norm": 0.22139325737953186, "learning_rate": 8.897107096259593e-05, "loss": 0.0169, "step": 20380 }, { "grad_norm": 0.2669930160045624, "learning_rate": 8.895811889788994e-05, "loss": 0.0211, "step": 20390 }, { "grad_norm": 0.2144310027360916, "learning_rate": 8.894516017630692e-05, "loss": 0.0151, "step": 20400 }, { "grad_norm": 0.2219594568014145, "learning_rate": 8.893219480006118e-05, "loss": 0.024, "step": 20410 }, { "grad_norm": 0.20933988690376282, "learning_rate": 8.891922277136817e-05, "loss": 0.0182, "step": 20420 }, { "grad_norm": 0.2734142243862152, "learning_rate": 8.890624409244441e-05, "loss": 0.0189, "step": 20430 }, { "grad_norm": 0.2702123522758484, "learning_rate": 8.889325876550763e-05, "loss": 0.0219, "step": 20440 }, { "grad_norm": 0.2079659104347229, "learning_rate": 8.888026679277666e-05, "loss": 0.0204, "step": 20450 }, { "grad_norm": 0.16841579973697662, "learning_rate": 8.886726817647147e-05, "loss": 0.0203, "step": 20460 }, { "grad_norm": 0.2902164161205292, "learning_rate": 8.885426291881319e-05, "loss": 0.0203, "step": 20470 }, { "grad_norm": 0.30141282081604004, "learning_rate": 8.884125102202401e-05, "loss": 0.0189, "step": 20480 }, { "grad_norm": 0.2869926691055298, "learning_rate": 8.882823248832736e-05, "loss": 0.0177, "step": 20490 }, { "grad_norm": 0.21268092095851898, "learning_rate": 8.881520731994772e-05, "loss": 0.0201, "step": 20500 }, { "grad_norm": 0.2598299980163574, "learning_rate": 8.880217551911077e-05, "loss": 0.0222, "step": 20510 }, { "grad_norm": 0.2971009612083435, "learning_rate": 8.878913708804323e-05, "loss": 0.0207, "step": 20520 }, { "grad_norm": 0.163115456700325, "learning_rate": 8.877609202897308e-05, "loss": 0.02, "step": 20530 }, { "grad_norm": 0.26652291417121887, "learning_rate": 8.876304034412933e-05, "loss": 0.0195, "step": 20540 }, { "grad_norm": 0.23539941012859344, "learning_rate": 8.874998203574214e-05, "loss": 0.0215, "step": 20550 }, { "grad_norm": 0.26472094655036926, "learning_rate": 8.873691710604284e-05, "loss": 0.023, "step": 20560 }, { "grad_norm": 0.21555136144161224, "learning_rate": 8.872384555726387e-05, "loss": 0.0257, "step": 20570 }, { "grad_norm": 0.2665577232837677, "learning_rate": 8.871076739163878e-05, "loss": 0.0292, "step": 20580 }, { "grad_norm": 0.26974308490753174, "learning_rate": 8.86976826114023e-05, "loss": 0.0239, "step": 20590 }, { "grad_norm": 0.3664475083351135, "learning_rate": 8.868459121879023e-05, "loss": 0.0308, "step": 20600 }, { "grad_norm": 0.3771047592163086, "learning_rate": 8.867149321603956e-05, "loss": 0.0238, "step": 20610 }, { "grad_norm": 0.3632149398326874, "learning_rate": 8.865838860538835e-05, "loss": 0.024, "step": 20620 }, { "grad_norm": 0.3204464018344879, "learning_rate": 8.864527738907585e-05, "loss": 0.0244, "step": 20630 }, { "grad_norm": 0.24182330071926117, "learning_rate": 8.863215956934239e-05, "loss": 0.0206, "step": 20640 }, { "grad_norm": 0.22469724714756012, "learning_rate": 8.861903514842947e-05, "loss": 0.0221, "step": 20650 }, { "grad_norm": 0.27149128913879395, "learning_rate": 8.860590412857966e-05, "loss": 0.0242, "step": 20660 }, { "grad_norm": 0.26208510994911194, "learning_rate": 8.85927665120367e-05, "loss": 0.0219, "step": 20670 }, { "grad_norm": 0.28036391735076904, "learning_rate": 8.857962230104546e-05, "loss": 0.0192, "step": 20680 }, { "grad_norm": 0.1589844524860382, "learning_rate": 8.856647149785193e-05, "loss": 0.0184, "step": 20690 }, { "grad_norm": 0.24766601622104645, "learning_rate": 8.855331410470322e-05, "loss": 0.0217, "step": 20700 }, { "grad_norm": 0.2660048305988312, "learning_rate": 8.854015012384756e-05, "loss": 0.023, "step": 20710 }, { "grad_norm": 0.2748650014400482, "learning_rate": 8.852697955753433e-05, "loss": 0.019, "step": 20720 }, { "grad_norm": 0.22968021035194397, "learning_rate": 8.851380240801399e-05, "loss": 0.0218, "step": 20730 }, { "grad_norm": 0.18427635729312897, "learning_rate": 8.850061867753818e-05, "loss": 0.0201, "step": 20740 }, { "grad_norm": 0.18725666403770447, "learning_rate": 8.848742836835963e-05, "loss": 0.0202, "step": 20750 }, { "grad_norm": 0.2381531149148941, "learning_rate": 8.847423148273221e-05, "loss": 0.0181, "step": 20760 }, { "grad_norm": 0.2379830926656723, "learning_rate": 8.846102802291092e-05, "loss": 0.022, "step": 20770 }, { "grad_norm": 0.19413411617279053, "learning_rate": 8.844781799115183e-05, "loss": 0.0218, "step": 20780 }, { "grad_norm": 0.24876567721366882, "learning_rate": 8.84346013897122e-05, "loss": 0.0217, "step": 20790 }, { "grad_norm": 0.2138938456773758, "learning_rate": 8.842137822085038e-05, "loss": 0.0222, "step": 20800 }, { "grad_norm": 0.20950990915298462, "learning_rate": 8.840814848682585e-05, "loss": 0.0202, "step": 20810 }, { "grad_norm": 0.19879326224327087, "learning_rate": 8.83949121898992e-05, "loss": 0.0183, "step": 20820 }, { "grad_norm": 0.22776676714420319, "learning_rate": 8.838166933233217e-05, "loss": 0.0168, "step": 20830 }, { "grad_norm": 0.24869610369205475, "learning_rate": 8.83684199163876e-05, "loss": 0.0223, "step": 20840 }, { "grad_norm": 0.22781530022621155, "learning_rate": 8.835516394432943e-05, "loss": 0.0201, "step": 20850 }, { "grad_norm": 0.21269188821315765, "learning_rate": 8.834190141842276e-05, "loss": 0.0165, "step": 20860 }, { "grad_norm": 0.2552459239959717, "learning_rate": 8.83286323409338e-05, "loss": 0.0159, "step": 20870 }, { "grad_norm": 0.251499205827713, "learning_rate": 8.831535671412986e-05, "loss": 0.0194, "step": 20880 }, { "grad_norm": 0.22054460644721985, "learning_rate": 8.830207454027938e-05, "loss": 0.0174, "step": 20890 }, { "grad_norm": 0.2124083787202835, "learning_rate": 8.828878582165192e-05, "loss": 0.0223, "step": 20900 }, { "grad_norm": 0.1410357505083084, "learning_rate": 8.827549056051818e-05, "loss": 0.0176, "step": 20910 }, { "grad_norm": 0.28047657012939453, "learning_rate": 8.826218875914993e-05, "loss": 0.0231, "step": 20920 }, { "grad_norm": 0.28502148389816284, "learning_rate": 8.82488804198201e-05, "loss": 0.0182, "step": 20930 }, { "grad_norm": 0.219553604722023, "learning_rate": 8.82355655448027e-05, "loss": 0.0202, "step": 20940 }, { "grad_norm": 0.22103141248226166, "learning_rate": 8.822224413637293e-05, "loss": 0.0197, "step": 20950 }, { "grad_norm": 0.23798426985740662, "learning_rate": 8.820891619680697e-05, "loss": 0.0214, "step": 20960 }, { "grad_norm": 0.2890033423900604, "learning_rate": 8.819558172838227e-05, "loss": 0.0227, "step": 20970 }, { "grad_norm": 0.26466673612594604, "learning_rate": 8.818224073337731e-05, "loss": 0.0203, "step": 20980 }, { "grad_norm": 0.2600652873516083, "learning_rate": 8.816889321407169e-05, "loss": 0.0211, "step": 20990 }, { "grad_norm": 0.3653309643268585, "learning_rate": 8.815553917274615e-05, "loss": 0.025, "step": 21000 }, { "grad_norm": 0.2215431183576584, "learning_rate": 8.81421786116825e-05, "loss": 0.0192, "step": 21010 }, { "grad_norm": 0.22640807926654816, "learning_rate": 8.812881153316373e-05, "loss": 0.0211, "step": 21020 }, { "grad_norm": 0.2660619914531708, "learning_rate": 8.81154379394739e-05, "loss": 0.0228, "step": 21030 }, { "grad_norm": 0.18221181631088257, "learning_rate": 8.810205783289818e-05, "loss": 0.0195, "step": 21040 }, { "grad_norm": 0.2534482181072235, "learning_rate": 8.808867121572286e-05, "loss": 0.0249, "step": 21050 }, { "grad_norm": 0.25896719098091125, "learning_rate": 8.807527809023537e-05, "loss": 0.0238, "step": 21060 }, { "grad_norm": 0.2434237003326416, "learning_rate": 8.80618784587242e-05, "loss": 0.0204, "step": 21070 }, { "grad_norm": 0.22926780581474304, "learning_rate": 8.804847232347902e-05, "loss": 0.0254, "step": 21080 }, { "grad_norm": 0.15908855199813843, "learning_rate": 8.803505968679054e-05, "loss": 0.02, "step": 21090 }, { "grad_norm": 0.2363239824771881, "learning_rate": 8.802164055095061e-05, "loss": 0.0244, "step": 21100 }, { "grad_norm": 0.24176642298698425, "learning_rate": 8.80082149182522e-05, "loss": 0.0203, "step": 21110 }, { "grad_norm": 0.2699148654937744, "learning_rate": 8.79947827909894e-05, "loss": 0.0212, "step": 21120 }, { "grad_norm": 0.38628557324409485, "learning_rate": 8.798134417145738e-05, "loss": 0.0227, "step": 21130 }, { "grad_norm": 0.24390067160129547, "learning_rate": 8.796789906195243e-05, "loss": 0.0195, "step": 21140 }, { "grad_norm": 0.20364564657211304, "learning_rate": 8.795444746477195e-05, "loss": 0.0214, "step": 21150 }, { "grad_norm": 0.18752656877040863, "learning_rate": 8.794098938221446e-05, "loss": 0.0201, "step": 21160 }, { "grad_norm": 0.28268080949783325, "learning_rate": 8.792752481657957e-05, "loss": 0.0221, "step": 21170 }, { "grad_norm": 0.19254721701145172, "learning_rate": 8.791405377016802e-05, "loss": 0.0167, "step": 21180 }, { "grad_norm": 0.25592735409736633, "learning_rate": 8.790057624528163e-05, "loss": 0.0175, "step": 21190 }, { "grad_norm": 0.2769049108028412, "learning_rate": 8.788709224422333e-05, "loss": 0.0233, "step": 21200 }, { "grad_norm": 0.2760973572731018, "learning_rate": 8.787360176929717e-05, "loss": 0.0187, "step": 21210 }, { "grad_norm": 0.2607189118862152, "learning_rate": 8.786010482280834e-05, "loss": 0.0193, "step": 21220 }, { "grad_norm": 0.6759591102600098, "learning_rate": 8.784660140706306e-05, "loss": 0.0258, "step": 21230 }, { "grad_norm": 0.469552218914032, "learning_rate": 8.783309152436872e-05, "loss": 0.0279, "step": 21240 }, { "grad_norm": 0.3683631122112274, "learning_rate": 8.781957517703375e-05, "loss": 0.0252, "step": 21250 }, { "grad_norm": 0.3043948709964752, "learning_rate": 8.780605236736776e-05, "loss": 0.025, "step": 21260 }, { "grad_norm": 0.24501201510429382, "learning_rate": 8.779252309768142e-05, "loss": 0.0257, "step": 21270 }, { "grad_norm": 0.22188737988471985, "learning_rate": 8.777898737028652e-05, "loss": 0.024, "step": 21280 }, { "grad_norm": 0.2056303471326828, "learning_rate": 8.776544518749591e-05, "loss": 0.0211, "step": 21290 }, { "grad_norm": 0.16011179983615875, "learning_rate": 8.775189655162364e-05, "loss": 0.026, "step": 21300 }, { "grad_norm": 0.26560917496681213, "learning_rate": 8.773834146498474e-05, "loss": 0.0179, "step": 21310 }, { "grad_norm": 0.212201327085495, "learning_rate": 8.772477992989545e-05, "loss": 0.0218, "step": 21320 }, { "grad_norm": 0.6200379729270935, "learning_rate": 8.771121194867304e-05, "loss": 0.0227, "step": 21330 }, { "grad_norm": 0.21607132256031036, "learning_rate": 8.769763752363589e-05, "loss": 0.0236, "step": 21340 }, { "grad_norm": 0.6950702667236328, "learning_rate": 8.768405665710352e-05, "loss": 0.0335, "step": 21350 }, { "grad_norm": 0.4166836738586426, "learning_rate": 8.767046935139655e-05, "loss": 0.0327, "step": 21360 }, { "grad_norm": 0.2346080094575882, "learning_rate": 8.765687560883666e-05, "loss": 0.0236, "step": 21370 }, { "grad_norm": 0.22992004454135895, "learning_rate": 8.764327543174664e-05, "loss": 0.0237, "step": 21380 }, { "grad_norm": 0.20608720183372498, "learning_rate": 8.762966882245038e-05, "loss": 0.019, "step": 21390 }, { "grad_norm": 0.17734673619270325, "learning_rate": 8.761605578327291e-05, "loss": 0.0197, "step": 21400 }, { "grad_norm": 0.36629408597946167, "learning_rate": 8.76024363165403e-05, "loss": 0.023, "step": 21410 }, { "grad_norm": 0.21861626207828522, "learning_rate": 8.758881042457976e-05, "loss": 0.0203, "step": 21420 }, { "grad_norm": 0.2824137210845947, "learning_rate": 8.757517810971957e-05, "loss": 0.0223, "step": 21430 }, { "grad_norm": 0.2856622040271759, "learning_rate": 8.756153937428913e-05, "loss": 0.0264, "step": 21440 }, { "grad_norm": 0.28119662404060364, "learning_rate": 8.754789422061889e-05, "loss": 0.0222, "step": 21450 }, { "grad_norm": 0.4046682119369507, "learning_rate": 8.753424265104052e-05, "loss": 0.0206, "step": 21460 }, { "grad_norm": 0.24282342195510864, "learning_rate": 8.752058466788659e-05, "loss": 0.0223, "step": 21470 }, { "grad_norm": 0.26378557085990906, "learning_rate": 8.750692027349097e-05, "loss": 0.0233, "step": 21480 }, { "grad_norm": 0.37928855419158936, "learning_rate": 8.749324947018847e-05, "loss": 0.0251, "step": 21490 }, { "grad_norm": 0.336551308631897, "learning_rate": 8.747957226031507e-05, "loss": 0.0269, "step": 21500 }, { "grad_norm": 0.25948381423950195, "learning_rate": 8.746588864620787e-05, "loss": 0.0261, "step": 21510 }, { "grad_norm": 0.24111981689929962, "learning_rate": 8.745219863020498e-05, "loss": 0.0242, "step": 21520 }, { "grad_norm": 0.20958225429058075, "learning_rate": 8.743850221464564e-05, "loss": 0.0187, "step": 21530 }, { "grad_norm": 0.1903398633003235, "learning_rate": 8.742479940187026e-05, "loss": 0.0227, "step": 21540 }, { "grad_norm": 0.24891561269760132, "learning_rate": 8.74110901942202e-05, "loss": 0.0178, "step": 21550 }, { "grad_norm": 0.2647970914840698, "learning_rate": 8.739737459403803e-05, "loss": 0.0168, "step": 21560 }, { "grad_norm": 0.2409583330154419, "learning_rate": 8.738365260366737e-05, "loss": 0.0197, "step": 21570 }, { "grad_norm": 0.2589280903339386, "learning_rate": 8.736992422545292e-05, "loss": 0.019, "step": 21580 }, { "grad_norm": 0.2461319863796234, "learning_rate": 8.73561894617405e-05, "loss": 0.0231, "step": 21590 }, { "grad_norm": 0.22629372775554657, "learning_rate": 8.734244831487697e-05, "loss": 0.0167, "step": 21600 }, { "grad_norm": 0.24502290785312653, "learning_rate": 8.732870078721035e-05, "loss": 0.0202, "step": 21610 }, { "grad_norm": 0.190662682056427, "learning_rate": 8.731494688108972e-05, "loss": 0.0213, "step": 21620 }, { "grad_norm": 0.22066070139408112, "learning_rate": 8.730118659886523e-05, "loss": 0.0194, "step": 21630 }, { "grad_norm": 0.26329275965690613, "learning_rate": 8.728741994288814e-05, "loss": 0.0193, "step": 21640 }, { "grad_norm": 0.23869307339191437, "learning_rate": 8.727364691551079e-05, "loss": 0.0192, "step": 21650 }, { "grad_norm": 0.19360998272895813, "learning_rate": 8.725986751908661e-05, "loss": 0.0175, "step": 21660 }, { "grad_norm": 0.3024452328681946, "learning_rate": 8.724608175597016e-05, "loss": 0.0187, "step": 21670 }, { "grad_norm": 0.28541556000709534, "learning_rate": 8.723228962851699e-05, "loss": 0.0195, "step": 21680 }, { "grad_norm": 0.21431507170200348, "learning_rate": 8.721849113908385e-05, "loss": 0.0234, "step": 21690 }, { "grad_norm": 0.20579200983047485, "learning_rate": 8.720468629002848e-05, "loss": 0.0215, "step": 21700 }, { "grad_norm": 0.20988287031650543, "learning_rate": 8.719087508370978e-05, "loss": 0.0185, "step": 21710 }, { "grad_norm": 0.3740880787372589, "learning_rate": 8.717705752248772e-05, "loss": 0.021, "step": 21720 }, { "grad_norm": 0.27141907811164856, "learning_rate": 8.71632336087233e-05, "loss": 0.0222, "step": 21730 }, { "grad_norm": 0.2672840654850006, "learning_rate": 8.71494033447787e-05, "loss": 0.0202, "step": 21740 }, { "grad_norm": 0.28947848081588745, "learning_rate": 8.713556673301708e-05, "loss": 0.0236, "step": 21750 }, { "grad_norm": 0.1475818008184433, "learning_rate": 8.712172377580278e-05, "loss": 0.0205, "step": 21760 }, { "grad_norm": 0.17130650579929352, "learning_rate": 8.710787447550114e-05, "loss": 0.0171, "step": 21770 }, { "grad_norm": 0.6430314779281616, "learning_rate": 8.70940188344787e-05, "loss": 0.0224, "step": 21780 }, { "grad_norm": 0.2810151278972626, "learning_rate": 8.708015685510293e-05, "loss": 0.0271, "step": 21790 }, { "grad_norm": 0.23189255595207214, "learning_rate": 8.706628853974252e-05, "loss": 0.0253, "step": 21800 }, { "grad_norm": 0.26943638920783997, "learning_rate": 8.705241389076715e-05, "loss": 0.0248, "step": 21810 }, { "grad_norm": 0.26521122455596924, "learning_rate": 8.703853291054764e-05, "loss": 0.0187, "step": 21820 }, { "grad_norm": 0.2525568902492523, "learning_rate": 8.702464560145587e-05, "loss": 0.0252, "step": 21830 }, { "grad_norm": 0.24167244136333466, "learning_rate": 8.701075196586476e-05, "loss": 0.0236, "step": 21840 }, { "grad_norm": 0.42913687229156494, "learning_rate": 8.699685200614842e-05, "loss": 0.0264, "step": 21850 }, { "grad_norm": 0.2533644437789917, "learning_rate": 8.698294572468193e-05, "loss": 0.0215, "step": 21860 }, { "grad_norm": 0.28311291337013245, "learning_rate": 8.696903312384148e-05, "loss": 0.0212, "step": 21870 }, { "grad_norm": 0.21342727541923523, "learning_rate": 8.695511420600439e-05, "loss": 0.0232, "step": 21880 }, { "grad_norm": 0.30942076444625854, "learning_rate": 8.694118897354901e-05, "loss": 0.0192, "step": 21890 }, { "grad_norm": 0.19302941858768463, "learning_rate": 8.692725742885478e-05, "loss": 0.0208, "step": 21900 }, { "grad_norm": 0.3821578621864319, "learning_rate": 8.691331957430221e-05, "loss": 0.0244, "step": 21910 }, { "grad_norm": 0.24941493570804596, "learning_rate": 8.68993754122729e-05, "loss": 0.0245, "step": 21920 }, { "grad_norm": 0.19914399087429047, "learning_rate": 8.688542494514955e-05, "loss": 0.0259, "step": 21930 }, { "grad_norm": 0.2850590944290161, "learning_rate": 8.68714681753159e-05, "loss": 0.0275, "step": 21940 }, { "grad_norm": 0.27538588643074036, "learning_rate": 8.685750510515676e-05, "loss": 0.0247, "step": 21950 }, { "grad_norm": 0.2757968604564667, "learning_rate": 8.684353573705805e-05, "loss": 0.0249, "step": 21960 }, { "grad_norm": 0.26191890239715576, "learning_rate": 8.682956007340677e-05, "loss": 0.0232, "step": 21970 }, { "grad_norm": 0.2045707255601883, "learning_rate": 8.681557811659095e-05, "loss": 0.0282, "step": 21980 }, { "grad_norm": 0.20084574818611145, "learning_rate": 8.680158986899974e-05, "loss": 0.0234, "step": 21990 }, { "grad_norm": 0.2719271779060364, "learning_rate": 8.678759533302335e-05, "loss": 0.0205, "step": 22000 }, { "grad_norm": 0.2636520564556122, "learning_rate": 8.677359451105308e-05, "loss": 0.0209, "step": 22010 }, { "grad_norm": 0.23023797571659088, "learning_rate": 8.675958740548123e-05, "loss": 0.0213, "step": 22020 }, { "grad_norm": 0.24141724407672882, "learning_rate": 8.674557401870129e-05, "loss": 0.0239, "step": 22030 }, { "grad_norm": 0.25433045625686646, "learning_rate": 8.673155435310775e-05, "loss": 0.0219, "step": 22040 }, { "grad_norm": 0.26111117005348206, "learning_rate": 8.671752841109617e-05, "loss": 0.0205, "step": 22050 }, { "grad_norm": 0.24722060561180115, "learning_rate": 8.670349619506321e-05, "loss": 0.0181, "step": 22060 }, { "grad_norm": 0.19856299459934235, "learning_rate": 8.66894577074066e-05, "loss": 0.0208, "step": 22070 }, { "grad_norm": 0.2480117231607437, "learning_rate": 8.667541295052513e-05, "loss": 0.0231, "step": 22080 }, { "grad_norm": 0.22242046892642975, "learning_rate": 8.666136192681865e-05, "loss": 0.021, "step": 22090 }, { "grad_norm": 0.20099695026874542, "learning_rate": 8.664730463868811e-05, "loss": 0.0174, "step": 22100 }, { "grad_norm": 0.18567143380641937, "learning_rate": 8.663324108853552e-05, "loss": 0.0219, "step": 22110 }, { "grad_norm": 0.17360031604766846, "learning_rate": 8.661917127876395e-05, "loss": 0.0204, "step": 22120 }, { "grad_norm": 0.21624891459941864, "learning_rate": 8.660509521177754e-05, "loss": 0.016, "step": 22130 }, { "grad_norm": 0.22385288774967194, "learning_rate": 8.65910128899815e-05, "loss": 0.0163, "step": 22140 }, { "grad_norm": 0.29400691390037537, "learning_rate": 8.657692431578214e-05, "loss": 0.0182, "step": 22150 }, { "grad_norm": 0.21942837536334991, "learning_rate": 8.656282949158679e-05, "loss": 0.0192, "step": 22160 }, { "grad_norm": 0.21034224331378937, "learning_rate": 8.654872841980388e-05, "loss": 0.0174, "step": 22170 }, { "grad_norm": 0.22951339185237885, "learning_rate": 8.653462110284289e-05, "loss": 0.0157, "step": 22180 }, { "grad_norm": 0.22349593043327332, "learning_rate": 8.652050754311437e-05, "loss": 0.0171, "step": 22190 }, { "grad_norm": 0.1671716868877411, "learning_rate": 8.650638774302995e-05, "loss": 0.0183, "step": 22200 }, { "grad_norm": 0.22496747970581055, "learning_rate": 8.649226170500233e-05, "loss": 0.0185, "step": 22210 }, { "grad_norm": 0.27359673380851746, "learning_rate": 8.647812943144524e-05, "loss": 0.0225, "step": 22220 }, { "grad_norm": 0.17948560416698456, "learning_rate": 8.646399092477351e-05, "loss": 0.0233, "step": 22230 }, { "grad_norm": 0.2815065085887909, "learning_rate": 8.644984618740301e-05, "loss": 0.0229, "step": 22240 }, { "grad_norm": 0.2334451526403427, "learning_rate": 8.643569522175073e-05, "loss": 0.0219, "step": 22250 }, { "grad_norm": 0.20213207602500916, "learning_rate": 8.642153803023463e-05, "loss": 0.0214, "step": 22260 }, { "grad_norm": 0.25514355301856995, "learning_rate": 8.640737461527383e-05, "loss": 0.0222, "step": 22270 }, { "grad_norm": 0.2783779501914978, "learning_rate": 8.639320497928845e-05, "loss": 0.026, "step": 22280 }, { "grad_norm": 0.2803547978401184, "learning_rate": 8.637902912469969e-05, "loss": 0.022, "step": 22290 }, { "grad_norm": 0.21871094405651093, "learning_rate": 8.636484705392982e-05, "loss": 0.0257, "step": 22300 }, { "grad_norm": 0.27685657143592834, "learning_rate": 8.635065876940216e-05, "loss": 0.0223, "step": 22310 }, { "grad_norm": 0.22217264771461487, "learning_rate": 8.633646427354112e-05, "loss": 0.0227, "step": 22320 }, { "grad_norm": 0.23572276532649994, "learning_rate": 8.632226356877213e-05, "loss": 0.0249, "step": 22330 }, { "grad_norm": 0.23286522924900055, "learning_rate": 8.630805665752173e-05, "loss": 0.0189, "step": 22340 }, { "grad_norm": 0.18592631816864014, "learning_rate": 8.629384354221748e-05, "loss": 0.0196, "step": 22350 }, { "grad_norm": 0.26284661889076233, "learning_rate": 8.627962422528797e-05, "loss": 0.0158, "step": 22360 }, { "grad_norm": 0.2720807194709778, "learning_rate": 8.626539870916296e-05, "loss": 0.023, "step": 22370 }, { "grad_norm": 0.19562466442584991, "learning_rate": 8.625116699627317e-05, "loss": 0.0219, "step": 22380 }, { "grad_norm": 0.2148367017507553, "learning_rate": 8.623692908905041e-05, "loss": 0.0227, "step": 22390 }, { "grad_norm": 0.2844529151916504, "learning_rate": 8.622268498992755e-05, "loss": 0.0279, "step": 22400 }, { "grad_norm": 0.26661163568496704, "learning_rate": 8.620843470133851e-05, "loss": 0.023, "step": 22410 }, { "grad_norm": 0.22624832391738892, "learning_rate": 8.619417822571829e-05, "loss": 0.0202, "step": 22420 }, { "grad_norm": 0.194148451089859, "learning_rate": 8.617991556550292e-05, "loss": 0.017, "step": 22430 }, { "grad_norm": 0.20443280041217804, "learning_rate": 8.616564672312952e-05, "loss": 0.0183, "step": 22440 }, { "grad_norm": 0.23485299944877625, "learning_rate": 8.61513717010362e-05, "loss": 0.0241, "step": 22450 }, { "grad_norm": 0.2461165487766266, "learning_rate": 8.613709050166221e-05, "loss": 0.0205, "step": 22460 }, { "grad_norm": 0.20115619897842407, "learning_rate": 8.61228031274478e-05, "loss": 0.0199, "step": 22470 }, { "grad_norm": 0.218122661113739, "learning_rate": 8.610850958083431e-05, "loss": 0.0173, "step": 22480 }, { "grad_norm": 0.2786535620689392, "learning_rate": 8.609420986426409e-05, "loss": 0.0182, "step": 22490 }, { "grad_norm": 0.21855027973651886, "learning_rate": 8.60799039801806e-05, "loss": 0.0172, "step": 22500 }, { "grad_norm": 0.2091936618089676, "learning_rate": 8.606559193102828e-05, "loss": 0.0186, "step": 22510 }, { "grad_norm": 0.20342639088630676, "learning_rate": 8.605127371925273e-05, "loss": 0.0179, "step": 22520 }, { "grad_norm": 0.282008558511734, "learning_rate": 8.603694934730047e-05, "loss": 0.0217, "step": 22530 }, { "grad_norm": 0.2869277596473694, "learning_rate": 8.602261881761919e-05, "loss": 0.0179, "step": 22540 }, { "grad_norm": 0.2561073303222656, "learning_rate": 8.600828213265759e-05, "loss": 0.016, "step": 22550 }, { "grad_norm": 0.600421667098999, "learning_rate": 8.599393929486539e-05, "loss": 0.0202, "step": 22560 }, { "grad_norm": 0.28179097175598145, "learning_rate": 8.59795903066934e-05, "loss": 0.0177, "step": 22570 }, { "grad_norm": 0.2217061072587967, "learning_rate": 8.596523517059347e-05, "loss": 0.0177, "step": 22580 }, { "grad_norm": 0.25600650906562805, "learning_rate": 8.59508738890185e-05, "loss": 0.0179, "step": 22590 }, { "grad_norm": 0.18468111753463745, "learning_rate": 8.593650646442246e-05, "loss": 0.0148, "step": 22600 }, { "grad_norm": 0.19955310225486755, "learning_rate": 8.59221328992603e-05, "loss": 0.0193, "step": 22610 }, { "grad_norm": 0.19594377279281616, "learning_rate": 8.590775319598813e-05, "loss": 0.0173, "step": 22620 }, { "grad_norm": 0.29661715030670166, "learning_rate": 8.589336735706301e-05, "loss": 0.0174, "step": 22630 }, { "grad_norm": 0.22251491248607635, "learning_rate": 8.587897538494307e-05, "loss": 0.0182, "step": 22640 }, { "grad_norm": 0.22868309915065765, "learning_rate": 8.586457728208756e-05, "loss": 0.0197, "step": 22650 }, { "grad_norm": 0.22954028844833374, "learning_rate": 8.585017305095667e-05, "loss": 0.0183, "step": 22660 }, { "grad_norm": 0.20957550406455994, "learning_rate": 8.583576269401173e-05, "loss": 0.0201, "step": 22670 }, { "grad_norm": 0.2087884545326233, "learning_rate": 8.582134621371504e-05, "loss": 0.0192, "step": 22680 }, { "grad_norm": 0.1878557801246643, "learning_rate": 8.580692361253e-05, "loss": 0.0225, "step": 22690 }, { "grad_norm": 0.3345602750778198, "learning_rate": 8.579249489292104e-05, "loss": 0.0211, "step": 22700 }, { "grad_norm": 0.2534683644771576, "learning_rate": 8.577806005735363e-05, "loss": 0.017, "step": 22710 }, { "grad_norm": 0.18612150847911835, "learning_rate": 8.576361910829429e-05, "loss": 0.0164, "step": 22720 }, { "grad_norm": 0.2327825278043747, "learning_rate": 8.574917204821057e-05, "loss": 0.0205, "step": 22730 }, { "grad_norm": 0.23025736212730408, "learning_rate": 8.57347188795711e-05, "loss": 0.0184, "step": 22740 }, { "grad_norm": 0.18324773013591766, "learning_rate": 8.572025960484551e-05, "loss": 0.021, "step": 22750 }, { "grad_norm": 0.2660672068595886, "learning_rate": 8.57057942265045e-05, "loss": 0.0177, "step": 22760 }, { "grad_norm": 0.2766777276992798, "learning_rate": 8.569132274701984e-05, "loss": 0.0213, "step": 22770 }, { "grad_norm": 0.2501513957977295, "learning_rate": 8.567684516886427e-05, "loss": 0.0191, "step": 22780 }, { "grad_norm": 0.2505580186843872, "learning_rate": 8.56623614945116e-05, "loss": 0.0218, "step": 22790 }, { "grad_norm": 0.2579123079776764, "learning_rate": 8.564787172643675e-05, "loss": 0.0208, "step": 22800 }, { "grad_norm": 0.24564464390277863, "learning_rate": 8.563337586711559e-05, "loss": 0.0197, "step": 22810 }, { "grad_norm": 0.17690086364746094, "learning_rate": 8.561887391902506e-05, "loss": 0.0167, "step": 22820 }, { "grad_norm": 0.2222740650177002, "learning_rate": 8.560436588464316e-05, "loss": 0.0167, "step": 22830 }, { "grad_norm": 0.3614647686481476, "learning_rate": 8.55898517664489e-05, "loss": 0.0231, "step": 22840 }, { "grad_norm": 0.24511082470417023, "learning_rate": 8.557533156692236e-05, "loss": 0.0202, "step": 22850 }, { "grad_norm": 0.22002163529396057, "learning_rate": 8.556080528854467e-05, "loss": 0.0186, "step": 22860 }, { "grad_norm": 0.159353569149971, "learning_rate": 8.554627293379791e-05, "loss": 0.017, "step": 22870 }, { "grad_norm": 0.1549973338842392, "learning_rate": 8.553173450516531e-05, "loss": 0.02, "step": 22880 }, { "grad_norm": 0.18668219447135925, "learning_rate": 8.551719000513108e-05, "loss": 0.0245, "step": 22890 }, { "grad_norm": 0.22290372848510742, "learning_rate": 8.550263943618049e-05, "loss": 0.0171, "step": 22900 }, { "grad_norm": 0.18135742843151093, "learning_rate": 8.54880828007998e-05, "loss": 0.0202, "step": 22910 }, { "grad_norm": 0.21899621188640594, "learning_rate": 8.547352010147637e-05, "loss": 0.0181, "step": 22920 }, { "grad_norm": 0.23137560486793518, "learning_rate": 8.545895134069855e-05, "loss": 0.0225, "step": 22930 }, { "grad_norm": 0.20995964109897614, "learning_rate": 8.544437652095576e-05, "loss": 0.0165, "step": 22940 }, { "grad_norm": 0.2324972003698349, "learning_rate": 8.542979564473843e-05, "loss": 0.0176, "step": 22950 }, { "grad_norm": 0.38623759150505066, "learning_rate": 8.541520871453802e-05, "loss": 0.0206, "step": 22960 }, { "grad_norm": 0.1451282948255539, "learning_rate": 8.540061573284705e-05, "loss": 0.0165, "step": 22970 }, { "grad_norm": 0.21881277859210968, "learning_rate": 8.538601670215906e-05, "loss": 0.0205, "step": 22980 }, { "grad_norm": 0.19454623758792877, "learning_rate": 8.537141162496864e-05, "loss": 0.0198, "step": 22990 }, { "grad_norm": 0.2381395697593689, "learning_rate": 8.535680050377137e-05, "loss": 0.0193, "step": 23000 }, { "grad_norm": 0.34124866127967834, "learning_rate": 8.534218334106391e-05, "loss": 0.0192, "step": 23010 }, { "grad_norm": 0.25585171580314636, "learning_rate": 8.532756013934393e-05, "loss": 0.0187, "step": 23020 }, { "grad_norm": 0.2329947054386139, "learning_rate": 8.531293090111012e-05, "loss": 0.0193, "step": 23030 }, { "grad_norm": 0.19881847500801086, "learning_rate": 8.529829562886225e-05, "loss": 0.0179, "step": 23040 }, { "grad_norm": 0.26680123805999756, "learning_rate": 8.528365432510105e-05, "loss": 0.0228, "step": 23050 }, { "grad_norm": 0.2766922414302826, "learning_rate": 8.526900699232833e-05, "loss": 0.0198, "step": 23060 }, { "grad_norm": 0.1994091421365738, "learning_rate": 8.525435363304695e-05, "loss": 0.0237, "step": 23070 }, { "grad_norm": 0.22645901143550873, "learning_rate": 8.523969424976072e-05, "loss": 0.017, "step": 23080 }, { "grad_norm": 0.25675392150878906, "learning_rate": 8.522502884497457e-05, "loss": 0.0178, "step": 23090 }, { "grad_norm": 0.20748034119606018, "learning_rate": 8.521035742119437e-05, "loss": 0.019, "step": 23100 }, { "grad_norm": 0.19972997903823853, "learning_rate": 8.519567998092712e-05, "loss": 0.0186, "step": 23110 }, { "grad_norm": 0.2927330434322357, "learning_rate": 8.518099652668075e-05, "loss": 0.0194, "step": 23120 }, { "grad_norm": 0.2670433819293976, "learning_rate": 8.516630706096429e-05, "loss": 0.0199, "step": 23130 }, { "grad_norm": 0.19297604262828827, "learning_rate": 8.515161158628773e-05, "loss": 0.0213, "step": 23140 }, { "grad_norm": 0.2385142743587494, "learning_rate": 8.513691010516216e-05, "loss": 0.0203, "step": 23150 }, { "grad_norm": 0.2150987982749939, "learning_rate": 8.512220262009966e-05, "loss": 0.0177, "step": 23160 }, { "grad_norm": 0.1928580105304718, "learning_rate": 8.510748913361332e-05, "loss": 0.0171, "step": 23170 }, { "grad_norm": 0.2791593372821808, "learning_rate": 8.509276964821726e-05, "loss": 0.0185, "step": 23180 }, { "grad_norm": 0.18199573457241058, "learning_rate": 8.507804416642669e-05, "loss": 0.0181, "step": 23190 }, { "grad_norm": 0.19096766412258148, "learning_rate": 8.506331269075774e-05, "loss": 0.017, "step": 23200 }, { "grad_norm": 0.2591412365436554, "learning_rate": 8.504857522372765e-05, "loss": 0.0217, "step": 23210 }, { "grad_norm": 0.1961229294538498, "learning_rate": 8.503383176785461e-05, "loss": 0.0207, "step": 23220 }, { "grad_norm": 0.30536389350891113, "learning_rate": 8.501908232565792e-05, "loss": 0.0257, "step": 23230 }, { "grad_norm": 0.16230408847332, "learning_rate": 8.50043268996578e-05, "loss": 0.0196, "step": 23240 }, { "grad_norm": 0.2713712751865387, "learning_rate": 8.498956549237562e-05, "loss": 0.0236, "step": 23250 }, { "grad_norm": 0.2395186424255371, "learning_rate": 8.497479810633366e-05, "loss": 0.0225, "step": 23260 }, { "grad_norm": 0.2938089370727539, "learning_rate": 8.496002474405525e-05, "loss": 0.018, "step": 23270 }, { "grad_norm": 0.15693600475788116, "learning_rate": 8.494524540806478e-05, "loss": 0.0236, "step": 23280 }, { "grad_norm": 0.2650126814842224, "learning_rate": 8.493046010088761e-05, "loss": 0.0193, "step": 23290 }, { "grad_norm": 0.15904337167739868, "learning_rate": 8.491566882505018e-05, "loss": 0.0181, "step": 23300 }, { "grad_norm": 0.20504456758499146, "learning_rate": 8.490087158307988e-05, "loss": 0.0208, "step": 23310 }, { "grad_norm": 0.2394779920578003, "learning_rate": 8.488606837750518e-05, "loss": 0.0192, "step": 23320 }, { "grad_norm": 0.17277070879936218, "learning_rate": 8.487125921085552e-05, "loss": 0.0184, "step": 23330 }, { "grad_norm": 0.23299534618854523, "learning_rate": 8.485644408566141e-05, "loss": 0.0184, "step": 23340 }, { "grad_norm": 0.30608710646629333, "learning_rate": 8.484162300445431e-05, "loss": 0.0208, "step": 23350 }, { "grad_norm": 0.24244137108325958, "learning_rate": 8.482679596976676e-05, "loss": 0.0146, "step": 23360 }, { "grad_norm": 0.2921494245529175, "learning_rate": 8.48119629841323e-05, "loss": 0.0181, "step": 23370 }, { "grad_norm": 0.1949458122253418, "learning_rate": 8.479712405008547e-05, "loss": 0.0187, "step": 23380 }, { "grad_norm": 0.298787385225296, "learning_rate": 8.478227917016184e-05, "loss": 0.0198, "step": 23390 }, { "grad_norm": 0.2727213501930237, "learning_rate": 8.476742834689801e-05, "loss": 0.0209, "step": 23400 }, { "grad_norm": 0.22138485312461853, "learning_rate": 8.475257158283157e-05, "loss": 0.019, "step": 23410 }, { "grad_norm": 0.2155129611492157, "learning_rate": 8.473770888050112e-05, "loss": 0.0191, "step": 23420 }, { "grad_norm": 0.13589662313461304, "learning_rate": 8.47228402424463e-05, "loss": 0.0167, "step": 23430 }, { "grad_norm": 0.2487872838973999, "learning_rate": 8.470796567120775e-05, "loss": 0.0186, "step": 23440 }, { "grad_norm": 0.21053805947303772, "learning_rate": 8.469308516932714e-05, "loss": 0.0176, "step": 23450 }, { "grad_norm": 0.2260991483926773, "learning_rate": 8.467819873934714e-05, "loss": 0.0232, "step": 23460 }, { "grad_norm": 0.40896129608154297, "learning_rate": 8.466330638381143e-05, "loss": 0.022, "step": 23470 }, { "grad_norm": 0.39802226424217224, "learning_rate": 8.464840810526469e-05, "loss": 0.026, "step": 23480 }, { "grad_norm": 0.2237107753753662, "learning_rate": 8.463350390625264e-05, "loss": 0.0225, "step": 23490 }, { "grad_norm": 0.24667687714099884, "learning_rate": 8.4618593789322e-05, "loss": 0.0236, "step": 23500 }, { "grad_norm": 0.20951741933822632, "learning_rate": 8.46036777570205e-05, "loss": 0.0231, "step": 23510 }, { "grad_norm": 0.3616527318954468, "learning_rate": 8.458875581189688e-05, "loss": 0.0231, "step": 23520 }, { "grad_norm": 0.46639484167099, "learning_rate": 8.457382795650092e-05, "loss": 0.0187, "step": 23530 }, { "grad_norm": 0.28223446011543274, "learning_rate": 8.455889419338335e-05, "loss": 0.0251, "step": 23540 }, { "grad_norm": 0.25225216150283813, "learning_rate": 8.454395452509593e-05, "loss": 0.0261, "step": 23550 }, { "grad_norm": 0.29250311851501465, "learning_rate": 8.452900895419146e-05, "loss": 0.0214, "step": 23560 }, { "grad_norm": 0.2674331068992615, "learning_rate": 8.451405748322376e-05, "loss": 0.0188, "step": 23570 }, { "grad_norm": 0.22527803480625153, "learning_rate": 8.449910011474759e-05, "loss": 0.0244, "step": 23580 }, { "grad_norm": 0.29579806327819824, "learning_rate": 8.448413685131876e-05, "loss": 0.0269, "step": 23590 }, { "grad_norm": 0.2510516941547394, "learning_rate": 8.446916769549407e-05, "loss": 0.0221, "step": 23600 }, { "grad_norm": 0.20906086266040802, "learning_rate": 8.445419264983136e-05, "loss": 0.0211, "step": 23610 }, { "grad_norm": 0.209906667470932, "learning_rate": 8.443921171688947e-05, "loss": 0.0191, "step": 23620 }, { "grad_norm": 0.24914509057998657, "learning_rate": 8.442422489922819e-05, "loss": 0.0166, "step": 23630 }, { "grad_norm": 0.25114554166793823, "learning_rate": 8.440923219940838e-05, "loss": 0.0242, "step": 23640 }, { "grad_norm": 0.23472140729427338, "learning_rate": 8.439423361999189e-05, "loss": 0.0199, "step": 23650 }, { "grad_norm": 0.18849298357963562, "learning_rate": 8.437922916354155e-05, "loss": 0.0209, "step": 23660 }, { "grad_norm": 0.2441105991601944, "learning_rate": 8.436421883262123e-05, "loss": 0.0197, "step": 23670 }, { "grad_norm": 0.20147863030433655, "learning_rate": 8.434920262979577e-05, "loss": 0.0246, "step": 23680 }, { "grad_norm": 0.19169023633003235, "learning_rate": 8.433418055763104e-05, "loss": 0.0204, "step": 23690 }, { "grad_norm": 0.2401863932609558, "learning_rate": 8.431915261869389e-05, "loss": 0.019, "step": 23700 }, { "grad_norm": 0.21249403059482574, "learning_rate": 8.43041188155522e-05, "loss": 0.0211, "step": 23710 }, { "grad_norm": 0.2336207777261734, "learning_rate": 8.428907915077481e-05, "loss": 0.0226, "step": 23720 }, { "grad_norm": 0.2878505289554596, "learning_rate": 8.42740336269316e-05, "loss": 0.0284, "step": 23730 }, { "grad_norm": 0.25387436151504517, "learning_rate": 8.425898224659345e-05, "loss": 0.0182, "step": 23740 }, { "grad_norm": 0.27297863364219666, "learning_rate": 8.42439250123322e-05, "loss": 0.0259, "step": 23750 }, { "grad_norm": 0.4414334297180176, "learning_rate": 8.422886192672076e-05, "loss": 0.0238, "step": 23760 }, { "grad_norm": 0.2660689651966095, "learning_rate": 8.421379299233297e-05, "loss": 0.0226, "step": 23770 }, { "grad_norm": 0.1925434023141861, "learning_rate": 8.419871821174371e-05, "loss": 0.0201, "step": 23780 }, { "grad_norm": 0.1861099749803543, "learning_rate": 8.418363758752884e-05, "loss": 0.0216, "step": 23790 }, { "grad_norm": 0.2788638174533844, "learning_rate": 8.416855112226523e-05, "loss": 0.0222, "step": 23800 }, { "grad_norm": 0.24553731083869934, "learning_rate": 8.415345881853075e-05, "loss": 0.0204, "step": 23810 }, { "grad_norm": 0.2720705270767212, "learning_rate": 8.413836067890426e-05, "loss": 0.0204, "step": 23820 }, { "grad_norm": 0.3041309416294098, "learning_rate": 8.41232567059656e-05, "loss": 0.0301, "step": 23830 }, { "grad_norm": 0.26286581158638, "learning_rate": 8.410814690229565e-05, "loss": 0.0221, "step": 23840 }, { "grad_norm": 0.1967349648475647, "learning_rate": 8.409303127047626e-05, "loss": 0.0269, "step": 23850 }, { "grad_norm": 0.17864570021629333, "learning_rate": 8.407790981309028e-05, "loss": 0.023, "step": 23860 }, { "grad_norm": 0.26664528250694275, "learning_rate": 8.406278253272153e-05, "loss": 0.0256, "step": 23870 }, { "grad_norm": 0.34697726368904114, "learning_rate": 8.404764943195487e-05, "loss": 0.0277, "step": 23880 }, { "grad_norm": 0.23551544547080994, "learning_rate": 8.403251051337613e-05, "loss": 0.0226, "step": 23890 }, { "grad_norm": 0.2671436369419098, "learning_rate": 8.401736577957214e-05, "loss": 0.0176, "step": 23900 }, { "grad_norm": 0.22937290370464325, "learning_rate": 8.40022152331307e-05, "loss": 0.0196, "step": 23910 }, { "grad_norm": 0.2689070999622345, "learning_rate": 8.398705887664064e-05, "loss": 0.0248, "step": 23920 }, { "grad_norm": 0.20942167937755585, "learning_rate": 8.397189671269177e-05, "loss": 0.0209, "step": 23930 }, { "grad_norm": 0.22404085099697113, "learning_rate": 8.395672874387488e-05, "loss": 0.0191, "step": 23940 }, { "grad_norm": 0.28853610157966614, "learning_rate": 8.394155497278177e-05, "loss": 0.0216, "step": 23950 }, { "grad_norm": 0.25083279609680176, "learning_rate": 8.392637540200523e-05, "loss": 0.0215, "step": 23960 }, { "grad_norm": 0.2569904029369354, "learning_rate": 8.391119003413902e-05, "loss": 0.0213, "step": 23970 }, { "grad_norm": 0.16838443279266357, "learning_rate": 8.38959988717779e-05, "loss": 0.0184, "step": 23980 }, { "grad_norm": 0.2031305730342865, "learning_rate": 8.388080191751764e-05, "loss": 0.0193, "step": 23990 }, { "grad_norm": 0.308406263589859, "learning_rate": 8.386559917395496e-05, "loss": 0.0187, "step": 24000 }, { "grad_norm": 0.3045050799846649, "learning_rate": 8.385039064368761e-05, "loss": 0.0221, "step": 24010 }, { "grad_norm": 0.19803978502750397, "learning_rate": 8.383517632931431e-05, "loss": 0.0207, "step": 24020 }, { "grad_norm": 0.28388911485671997, "learning_rate": 8.381995623343477e-05, "loss": 0.0194, "step": 24030 }, { "grad_norm": 0.1889192759990692, "learning_rate": 8.380473035864968e-05, "loss": 0.0188, "step": 24040 }, { "grad_norm": 0.2222190648317337, "learning_rate": 8.378949870756076e-05, "loss": 0.0155, "step": 24050 }, { "grad_norm": 0.16072291135787964, "learning_rate": 8.377426128277063e-05, "loss": 0.0166, "step": 24060 }, { "grad_norm": 0.1941004991531372, "learning_rate": 8.375901808688298e-05, "loss": 0.0168, "step": 24070 }, { "grad_norm": 0.21932272613048553, "learning_rate": 8.374376912250246e-05, "loss": 0.0146, "step": 24080 }, { "grad_norm": 0.2910248339176178, "learning_rate": 8.372851439223468e-05, "loss": 0.0231, "step": 24090 }, { "grad_norm": 0.2057768553495407, "learning_rate": 8.371325389868627e-05, "loss": 0.0183, "step": 24100 }, { "grad_norm": 0.24181237816810608, "learning_rate": 8.369798764446482e-05, "loss": 0.0195, "step": 24110 }, { "grad_norm": 0.20169655978679657, "learning_rate": 8.368271563217893e-05, "loss": 0.0218, "step": 24120 }, { "grad_norm": 0.18227411806583405, "learning_rate": 8.366743786443817e-05, "loss": 0.0179, "step": 24130 }, { "grad_norm": 0.24423272907733917, "learning_rate": 8.365215434385309e-05, "loss": 0.0192, "step": 24140 }, { "grad_norm": 0.2927084267139435, "learning_rate": 8.36368650730352e-05, "loss": 0.0184, "step": 24150 }, { "grad_norm": 0.15725664794445038, "learning_rate": 8.362157005459705e-05, "loss": 0.019, "step": 24160 }, { "grad_norm": 0.2682698369026184, "learning_rate": 8.360626929115213e-05, "loss": 0.0168, "step": 24170 }, { "grad_norm": 0.18283523619174957, "learning_rate": 8.359096278531492e-05, "loss": 0.0177, "step": 24180 }, { "grad_norm": 0.2194872796535492, "learning_rate": 8.357565053970088e-05, "loss": 0.0206, "step": 24190 }, { "grad_norm": 0.1665862500667572, "learning_rate": 8.356033255692647e-05, "loss": 0.0174, "step": 24200 }, { "grad_norm": 0.21816794574260712, "learning_rate": 8.354500883960911e-05, "loss": 0.0177, "step": 24210 }, { "grad_norm": 0.2784527838230133, "learning_rate": 8.352967939036717e-05, "loss": 0.0171, "step": 24220 }, { "grad_norm": 0.2590823471546173, "learning_rate": 8.35143442118201e-05, "loss": 0.0179, "step": 24230 }, { "grad_norm": 0.22396332025527954, "learning_rate": 8.349900330658819e-05, "loss": 0.0188, "step": 24240 }, { "grad_norm": 0.22731368243694305, "learning_rate": 8.348365667729284e-05, "loss": 0.0148, "step": 24250 }, { "grad_norm": 0.2659909725189209, "learning_rate": 8.346830432655633e-05, "loss": 0.0214, "step": 24260 }, { "grad_norm": 0.21121405065059662, "learning_rate": 8.345294625700195e-05, "loss": 0.02, "step": 24270 }, { "grad_norm": 0.2173197716474533, "learning_rate": 8.343758247125402e-05, "loss": 0.0192, "step": 24280 }, { "grad_norm": 0.2283746749162674, "learning_rate": 8.342221297193776e-05, "loss": 0.0188, "step": 24290 }, { "grad_norm": 0.27011364698410034, "learning_rate": 8.34068377616794e-05, "loss": 0.0202, "step": 24300 }, { "grad_norm": 0.21816137433052063, "learning_rate": 8.339145684310615e-05, "loss": 0.017, "step": 24310 }, { "grad_norm": 0.23280476033687592, "learning_rate": 8.337607021884618e-05, "loss": 0.0225, "step": 24320 }, { "grad_norm": 0.20512676239013672, "learning_rate": 8.336067789152867e-05, "loss": 0.0166, "step": 24330 }, { "grad_norm": 0.19617831707000732, "learning_rate": 8.334527986378369e-05, "loss": 0.0153, "step": 24340 }, { "grad_norm": 0.19434699416160583, "learning_rate": 8.332987613824239e-05, "loss": 0.0179, "step": 24350 }, { "grad_norm": 0.23548424243927002, "learning_rate": 8.331446671753685e-05, "loss": 0.0182, "step": 24360 }, { "grad_norm": 0.18043899536132812, "learning_rate": 8.329905160430007e-05, "loss": 0.0152, "step": 24370 }, { "grad_norm": 0.18419180810451508, "learning_rate": 8.328363080116611e-05, "loss": 0.018, "step": 24380 }, { "grad_norm": 0.23323823511600494, "learning_rate": 8.326820431076997e-05, "loss": 0.0218, "step": 24390 }, { "grad_norm": 0.2362087070941925, "learning_rate": 8.325277213574759e-05, "loss": 0.0168, "step": 24400 }, { "grad_norm": 0.16545210778713226, "learning_rate": 8.32373342787359e-05, "loss": 0.0176, "step": 24410 }, { "grad_norm": 0.2735418677330017, "learning_rate": 8.322189074237285e-05, "loss": 0.0199, "step": 24420 }, { "grad_norm": 0.2831237316131592, "learning_rate": 8.32064415292973e-05, "loss": 0.0165, "step": 24430 }, { "grad_norm": 0.2313501089811325, "learning_rate": 8.319098664214907e-05, "loss": 0.0161, "step": 24440 }, { "grad_norm": 0.26331067085266113, "learning_rate": 8.3175526083569e-05, "loss": 0.02, "step": 24450 }, { "grad_norm": 0.21274994313716888, "learning_rate": 8.316005985619889e-05, "loss": 0.0163, "step": 24460 }, { "grad_norm": 0.3305393159389496, "learning_rate": 8.314458796268147e-05, "loss": 0.022, "step": 24470 }, { "grad_norm": 0.3494700491428375, "learning_rate": 8.312911040566047e-05, "loss": 0.0201, "step": 24480 }, { "grad_norm": 0.18642301857471466, "learning_rate": 8.31136271877806e-05, "loss": 0.0173, "step": 24490 }, { "grad_norm": 0.1500970870256424, "learning_rate": 8.309813831168748e-05, "loss": 0.0187, "step": 24500 }, { "grad_norm": 0.35156166553497314, "learning_rate": 8.308264378002777e-05, "loss": 0.0175, "step": 24510 }, { "grad_norm": 0.1986435204744339, "learning_rate": 8.306714359544906e-05, "loss": 0.0199, "step": 24520 }, { "grad_norm": 0.23835834860801697, "learning_rate": 8.30516377605999e-05, "loss": 0.0221, "step": 24530 }, { "grad_norm": 0.31629639863967896, "learning_rate": 8.30361262781298e-05, "loss": 0.0159, "step": 24540 }, { "grad_norm": 0.15285585820674896, "learning_rate": 8.302060915068924e-05, "loss": 0.0162, "step": 24550 }, { "grad_norm": 0.24214980006217957, "learning_rate": 8.300508638092972e-05, "loss": 0.0157, "step": 24560 }, { "grad_norm": 0.16355593502521515, "learning_rate": 8.298955797150361e-05, "loss": 0.0155, "step": 24570 }, { "grad_norm": 0.21050642430782318, "learning_rate": 8.297402392506433e-05, "loss": 0.0146, "step": 24580 }, { "grad_norm": 0.20619624853134155, "learning_rate": 8.295848424426617e-05, "loss": 0.0219, "step": 24590 }, { "grad_norm": 0.19387342035770416, "learning_rate": 8.29429389317645e-05, "loss": 0.0207, "step": 24600 }, { "grad_norm": 0.18777033686637878, "learning_rate": 8.292738799021556e-05, "loss": 0.0208, "step": 24610 }, { "grad_norm": 0.25379958748817444, "learning_rate": 8.291183142227656e-05, "loss": 0.0205, "step": 24620 }, { "grad_norm": 0.2104901224374771, "learning_rate": 8.289626923060572e-05, "loss": 0.0199, "step": 24630 }, { "grad_norm": 0.13502655923366547, "learning_rate": 8.288070141786218e-05, "loss": 0.0161, "step": 24640 }, { "grad_norm": 0.1996404230594635, "learning_rate": 8.286512798670605e-05, "loss": 0.0199, "step": 24650 }, { "grad_norm": 0.2248222529888153, "learning_rate": 8.284954893979842e-05, "loss": 0.0192, "step": 24660 }, { "grad_norm": 0.28556692600250244, "learning_rate": 8.283396427980131e-05, "loss": 0.0193, "step": 24670 }, { "grad_norm": 0.2317713350057602, "learning_rate": 8.281837400937771e-05, "loss": 0.0189, "step": 24680 }, { "grad_norm": 0.30489838123321533, "learning_rate": 8.28027781311916e-05, "loss": 0.0241, "step": 24690 }, { "grad_norm": 0.23562689125537872, "learning_rate": 8.278717664790785e-05, "loss": 0.0197, "step": 24700 }, { "grad_norm": 0.24476349353790283, "learning_rate": 8.277156956219234e-05, "loss": 0.0194, "step": 24710 }, { "grad_norm": 0.19374525547027588, "learning_rate": 8.275595687671189e-05, "loss": 0.0206, "step": 24720 }, { "grad_norm": 0.18054351210594177, "learning_rate": 8.27403385941343e-05, "loss": 0.0191, "step": 24730 }, { "grad_norm": 0.20869436860084534, "learning_rate": 8.272471471712828e-05, "loss": 0.0174, "step": 24740 }, { "grad_norm": 0.21956071257591248, "learning_rate": 8.270908524836355e-05, "loss": 0.0197, "step": 24750 }, { "grad_norm": 0.18763922154903412, "learning_rate": 8.269345019051074e-05, "loss": 0.0148, "step": 24760 }, { "grad_norm": 0.14721687138080597, "learning_rate": 8.267780954624147e-05, "loss": 0.0179, "step": 24770 }, { "grad_norm": 0.18697293102741241, "learning_rate": 8.266216331822827e-05, "loss": 0.0186, "step": 24780 }, { "grad_norm": 0.16234064102172852, "learning_rate": 8.264651150914469e-05, "loss": 0.0166, "step": 24790 }, { "grad_norm": 0.2685803174972534, "learning_rate": 8.263085412166517e-05, "loss": 0.0181, "step": 24800 }, { "grad_norm": 0.188639298081398, "learning_rate": 8.261519115846514e-05, "loss": 0.0172, "step": 24810 }, { "grad_norm": 0.1765759438276291, "learning_rate": 8.259952262222096e-05, "loss": 0.0178, "step": 24820 }, { "grad_norm": 0.6085610389709473, "learning_rate": 8.258384851560997e-05, "loss": 0.0228, "step": 24830 }, { "grad_norm": 0.25242263078689575, "learning_rate": 8.256816884131044e-05, "loss": 0.0173, "step": 24840 }, { "grad_norm": 0.22794415056705475, "learning_rate": 8.255248360200159e-05, "loss": 0.0225, "step": 24850 }, { "grad_norm": 0.361858606338501, "learning_rate": 8.253679280036359e-05, "loss": 0.0185, "step": 24860 }, { "grad_norm": 0.24029025435447693, "learning_rate": 8.252109643907762e-05, "loss": 0.0145, "step": 24870 }, { "grad_norm": 0.2077302783727646, "learning_rate": 8.250539452082569e-05, "loss": 0.0197, "step": 24880 }, { "grad_norm": 0.2297438234090805, "learning_rate": 8.248968704829087e-05, "loss": 0.0188, "step": 24890 }, { "grad_norm": 0.14331863820552826, "learning_rate": 8.247397402415714e-05, "loss": 0.0182, "step": 24900 }, { "grad_norm": 0.23146408796310425, "learning_rate": 8.24582554511094e-05, "loss": 0.0168, "step": 24910 }, { "grad_norm": 0.2109791338443756, "learning_rate": 8.244253133183355e-05, "loss": 0.015, "step": 24920 }, { "grad_norm": 0.3406900465488434, "learning_rate": 8.24268016690164e-05, "loss": 0.0196, "step": 24930 }, { "grad_norm": 0.2085370421409607, "learning_rate": 8.241106646534571e-05, "loss": 0.0202, "step": 24940 }, { "grad_norm": 0.2588016092777252, "learning_rate": 8.23953257235102e-05, "loss": 0.0171, "step": 24950 }, { "grad_norm": 0.16243189573287964, "learning_rate": 8.237957944619956e-05, "loss": 0.0168, "step": 24960 }, { "grad_norm": 0.1870451420545578, "learning_rate": 8.236382763610437e-05, "loss": 0.0207, "step": 24970 }, { "grad_norm": 0.18882843852043152, "learning_rate": 8.234807029591619e-05, "loss": 0.0174, "step": 24980 }, { "grad_norm": 0.21253180503845215, "learning_rate": 8.233230742832752e-05, "loss": 0.0169, "step": 24990 }, { "grad_norm": 0.2789594233036041, "learning_rate": 8.231653903603178e-05, "loss": 0.0175, "step": 25000 }, { "grad_norm": 0.1429053694009781, "learning_rate": 8.23007651217234e-05, "loss": 0.021, "step": 25010 }, { "grad_norm": 0.3114534020423889, "learning_rate": 8.228498568809769e-05, "loss": 0.0188, "step": 25020 }, { "grad_norm": 0.27281108498573303, "learning_rate": 8.22692007378509e-05, "loss": 0.0184, "step": 25030 }, { "grad_norm": 0.2126917541027069, "learning_rate": 8.225341027368028e-05, "loss": 0.015, "step": 25040 }, { "grad_norm": 0.20361310243606567, "learning_rate": 8.223761429828399e-05, "loss": 0.0168, "step": 25050 }, { "grad_norm": 0.31569939851760864, "learning_rate": 8.22218128143611e-05, "loss": 0.0174, "step": 25060 }, { "grad_norm": 0.2056385576725006, "learning_rate": 8.220600582461166e-05, "loss": 0.015, "step": 25070 }, { "grad_norm": 0.2693784832954407, "learning_rate": 8.219019333173668e-05, "loss": 0.0177, "step": 25080 }, { "grad_norm": 0.308123916387558, "learning_rate": 8.217437533843805e-05, "loss": 0.0164, "step": 25090 }, { "grad_norm": 0.25649070739746094, "learning_rate": 8.215855184741867e-05, "loss": 0.0186, "step": 25100 }, { "grad_norm": 0.307101845741272, "learning_rate": 8.21427228613823e-05, "loss": 0.0275, "step": 25110 }, { "grad_norm": 0.3189612925052643, "learning_rate": 8.21268883830337e-05, "loss": 0.0244, "step": 25120 }, { "grad_norm": 0.3245570659637451, "learning_rate": 8.211104841507855e-05, "loss": 0.0189, "step": 25130 }, { "grad_norm": 0.30631372332572937, "learning_rate": 8.209520296022346e-05, "loss": 0.0183, "step": 25140 }, { "grad_norm": 0.2524147033691406, "learning_rate": 8.207935202117599e-05, "loss": 0.0236, "step": 25150 }, { "grad_norm": 0.2958969473838806, "learning_rate": 8.206349560064463e-05, "loss": 0.0199, "step": 25160 }, { "grad_norm": 0.2311261147260666, "learning_rate": 8.204763370133881e-05, "loss": 0.0171, "step": 25170 }, { "grad_norm": 0.17597991228103638, "learning_rate": 8.203176632596892e-05, "loss": 0.0181, "step": 25180 }, { "grad_norm": 0.2115018516778946, "learning_rate": 8.20158934772462e-05, "loss": 0.0179, "step": 25190 }, { "grad_norm": 0.20147086679935455, "learning_rate": 8.200001515788294e-05, "loss": 0.0167, "step": 25200 }, { "grad_norm": 0.19839048385620117, "learning_rate": 8.198413137059228e-05, "loss": 0.0148, "step": 25210 }, { "grad_norm": 0.19525401294231415, "learning_rate": 8.196824211808835e-05, "loss": 0.0142, "step": 25220 }, { "grad_norm": 0.17327308654785156, "learning_rate": 8.195234740308617e-05, "loss": 0.0182, "step": 25230 }, { "grad_norm": 0.2680574953556061, "learning_rate": 8.193644722830171e-05, "loss": 0.0137, "step": 25240 }, { "grad_norm": 0.18652060627937317, "learning_rate": 8.19205415964519e-05, "loss": 0.021, "step": 25250 }, { "grad_norm": 0.22326348721981049, "learning_rate": 8.190463051025456e-05, "loss": 0.0157, "step": 25260 }, { "grad_norm": 0.23746129870414734, "learning_rate": 8.188871397242843e-05, "loss": 0.0187, "step": 25270 }, { "grad_norm": 0.27428337931632996, "learning_rate": 8.187279198569326e-05, "loss": 0.0203, "step": 25280 }, { "grad_norm": 0.165519118309021, "learning_rate": 8.185686455276966e-05, "loss": 0.0155, "step": 25290 }, { "grad_norm": 0.2216275930404663, "learning_rate": 8.184093167637921e-05, "loss": 0.0162, "step": 25300 }, { "grad_norm": 0.3417466878890991, "learning_rate": 8.182499335924437e-05, "loss": 0.0208, "step": 25310 }, { "grad_norm": 0.1834038943052292, "learning_rate": 8.18090496040886e-05, "loss": 0.0224, "step": 25320 }, { "grad_norm": 0.22012348473072052, "learning_rate": 8.179310041363621e-05, "loss": 0.0208, "step": 25330 }, { "grad_norm": 0.25594156980514526, "learning_rate": 8.17771457906125e-05, "loss": 0.0202, "step": 25340 }, { "grad_norm": 0.16551370918750763, "learning_rate": 8.176118573774371e-05, "loss": 0.02, "step": 25350 }, { "grad_norm": 0.2059798240661621, "learning_rate": 8.174522025775692e-05, "loss": 0.0169, "step": 25360 }, { "grad_norm": 0.1886756271123886, "learning_rate": 8.172924935338022e-05, "loss": 0.0273, "step": 25370 }, { "grad_norm": 0.2537521719932556, "learning_rate": 8.171327302734262e-05, "loss": 0.0216, "step": 25380 }, { "grad_norm": 0.15380032360553741, "learning_rate": 8.169729128237401e-05, "loss": 0.0182, "step": 25390 }, { "grad_norm": 0.1896527260541916, "learning_rate": 8.168130412120525e-05, "loss": 0.0166, "step": 25400 }, { "grad_norm": 0.19009706377983093, "learning_rate": 8.16653115465681e-05, "loss": 0.0155, "step": 25410 }, { "grad_norm": 0.1648247092962265, "learning_rate": 8.164931356119526e-05, "loss": 0.0179, "step": 25420 }, { "grad_norm": 0.23372210562229156, "learning_rate": 8.163331016782032e-05, "loss": 0.0187, "step": 25430 }, { "grad_norm": 0.16630758345127106, "learning_rate": 8.161730136917785e-05, "loss": 0.0193, "step": 25440 }, { "grad_norm": 0.2699466347694397, "learning_rate": 8.160128716800333e-05, "loss": 0.0238, "step": 25450 }, { "grad_norm": 0.19532693922519684, "learning_rate": 8.158526756703313e-05, "loss": 0.0237, "step": 25460 }, { "grad_norm": 0.21456776559352875, "learning_rate": 8.156924256900455e-05, "loss": 0.0212, "step": 25470 }, { "grad_norm": 0.2096564918756485, "learning_rate": 8.155321217665584e-05, "loss": 0.0236, "step": 25480 }, { "grad_norm": 0.24692270159721375, "learning_rate": 8.153717639272614e-05, "loss": 0.0204, "step": 25490 }, { "grad_norm": 0.16152384877204895, "learning_rate": 8.152113521995555e-05, "loss": 0.0185, "step": 25500 }, { "grad_norm": 0.2162383645772934, "learning_rate": 8.150508866108505e-05, "loss": 0.0234, "step": 25510 }, { "grad_norm": 0.3018231689929962, "learning_rate": 8.148903671885657e-05, "loss": 0.0201, "step": 25520 }, { "grad_norm": 0.2766837179660797, "learning_rate": 8.147297939601292e-05, "loss": 0.0213, "step": 25530 }, { "grad_norm": 0.25746577978134155, "learning_rate": 8.145691669529792e-05, "loss": 0.021, "step": 25540 }, { "grad_norm": 0.15719744563102722, "learning_rate": 8.144084861945618e-05, "loss": 0.0168, "step": 25550 }, { "grad_norm": 0.16326412558555603, "learning_rate": 8.142477517123333e-05, "loss": 0.0209, "step": 25560 }, { "grad_norm": 0.21071301400661469, "learning_rate": 8.140869635337586e-05, "loss": 0.0213, "step": 25570 }, { "grad_norm": 0.18713998794555664, "learning_rate": 8.139261216863123e-05, "loss": 0.0179, "step": 25580 }, { "grad_norm": 0.26239117980003357, "learning_rate": 8.137652261974776e-05, "loss": 0.0227, "step": 25590 }, { "grad_norm": 0.20023202896118164, "learning_rate": 8.136042770947472e-05, "loss": 0.017, "step": 25600 }, { "grad_norm": 0.2766149342060089, "learning_rate": 8.134432744056228e-05, "loss": 0.0156, "step": 25610 }, { "grad_norm": 0.1656528115272522, "learning_rate": 8.132822181576158e-05, "loss": 0.0176, "step": 25620 }, { "grad_norm": 0.18346892297267914, "learning_rate": 8.131211083782459e-05, "loss": 0.021, "step": 25630 }, { "grad_norm": 0.16977693140506744, "learning_rate": 8.129599450950424e-05, "loss": 0.0184, "step": 25640 }, { "grad_norm": 0.13893020153045654, "learning_rate": 8.127987283355438e-05, "loss": 0.0172, "step": 25650 }, { "grad_norm": 0.2771473526954651, "learning_rate": 8.126374581272976e-05, "loss": 0.0188, "step": 25660 }, { "grad_norm": 0.31161147356033325, "learning_rate": 8.124761344978605e-05, "loss": 0.0189, "step": 25670 }, { "grad_norm": 0.23561473190784454, "learning_rate": 8.12314757474798e-05, "loss": 0.0175, "step": 25680 }, { "grad_norm": 0.20490339398384094, "learning_rate": 8.121533270856856e-05, "loss": 0.0191, "step": 25690 }, { "grad_norm": 0.21198682487010956, "learning_rate": 8.119918433581069e-05, "loss": 0.0228, "step": 25700 }, { "grad_norm": 0.17296667397022247, "learning_rate": 8.118303063196551e-05, "loss": 0.0159, "step": 25710 }, { "grad_norm": 0.18143340945243835, "learning_rate": 8.116687159979326e-05, "loss": 0.0177, "step": 25720 }, { "grad_norm": 0.2468864619731903, "learning_rate": 8.115070724205508e-05, "loss": 0.0206, "step": 25730 }, { "grad_norm": 0.2821013331413269, "learning_rate": 8.113453756151296e-05, "loss": 0.0179, "step": 25740 }, { "grad_norm": 0.2562032639980316, "learning_rate": 8.111836256092995e-05, "loss": 0.019, "step": 25750 }, { "grad_norm": 0.2106027901172638, "learning_rate": 8.110218224306985e-05, "loss": 0.0167, "step": 25760 }, { "grad_norm": 0.1594240963459015, "learning_rate": 8.108599661069745e-05, "loss": 0.016, "step": 25770 }, { "grad_norm": 0.24896754324436188, "learning_rate": 8.106980566657845e-05, "loss": 0.0201, "step": 25780 }, { "grad_norm": 0.3157855272293091, "learning_rate": 8.10536094134794e-05, "loss": 0.02, "step": 25790 }, { "grad_norm": 0.57121741771698, "learning_rate": 8.103740785416783e-05, "loss": 0.0171, "step": 25800 }, { "grad_norm": 0.19349515438079834, "learning_rate": 8.102120099141212e-05, "loss": 0.0181, "step": 25810 }, { "grad_norm": 0.2371644228696823, "learning_rate": 8.100498882798163e-05, "loss": 0.0181, "step": 25820 }, { "grad_norm": 0.2816579043865204, "learning_rate": 8.09887713666465e-05, "loss": 0.0248, "step": 25830 }, { "grad_norm": 0.3029554784297943, "learning_rate": 8.09725486101779e-05, "loss": 0.0238, "step": 25840 }, { "grad_norm": 0.24907954037189484, "learning_rate": 8.095632056134784e-05, "loss": 0.0222, "step": 25850 }, { "grad_norm": 0.23012562096118927, "learning_rate": 8.094008722292925e-05, "loss": 0.0209, "step": 25860 }, { "grad_norm": 0.21729743480682373, "learning_rate": 8.092384859769598e-05, "loss": 0.0203, "step": 25870 }, { "grad_norm": 0.2438456416130066, "learning_rate": 8.090760468842275e-05, "loss": 0.0207, "step": 25880 }, { "grad_norm": 0.2314877063035965, "learning_rate": 8.089135549788521e-05, "loss": 0.0184, "step": 25890 }, { "grad_norm": 0.2355891615152359, "learning_rate": 8.087510102885987e-05, "loss": 0.0208, "step": 25900 }, { "grad_norm": 0.20597456395626068, "learning_rate": 8.085884128412422e-05, "loss": 0.0215, "step": 25910 }, { "grad_norm": 0.20714335143566132, "learning_rate": 8.084257626645659e-05, "loss": 0.0189, "step": 25920 }, { "grad_norm": 0.16847610473632812, "learning_rate": 8.08263059786362e-05, "loss": 0.0187, "step": 25930 }, { "grad_norm": 0.21104513108730316, "learning_rate": 8.081003042344325e-05, "loss": 0.0171, "step": 25940 }, { "grad_norm": 0.2610872983932495, "learning_rate": 8.079374960365872e-05, "loss": 0.0195, "step": 25950 }, { "grad_norm": 0.18586812913417816, "learning_rate": 8.077746352206463e-05, "loss": 0.0181, "step": 25960 }, { "grad_norm": 0.25986847281455994, "learning_rate": 8.076117218144377e-05, "loss": 0.0196, "step": 25970 }, { "grad_norm": 0.17961165308952332, "learning_rate": 8.074487558457991e-05, "loss": 0.0167, "step": 25980 }, { "grad_norm": 0.23414544761180878, "learning_rate": 8.072857373425768e-05, "loss": 0.0181, "step": 25990 }, { "grad_norm": 0.2164238691329956, "learning_rate": 8.071226663326264e-05, "loss": 0.0196, "step": 26000 }, { "grad_norm": 0.1894756406545639, "learning_rate": 8.069595428438121e-05, "loss": 0.0184, "step": 26010 }, { "grad_norm": 0.31135329604148865, "learning_rate": 8.067963669040072e-05, "loss": 0.0201, "step": 26020 }, { "grad_norm": 0.2775935232639313, "learning_rate": 8.066331385410942e-05, "loss": 0.0197, "step": 26030 }, { "grad_norm": 0.29592329263687134, "learning_rate": 8.064698577829641e-05, "loss": 0.0179, "step": 26040 }, { "grad_norm": 0.19603341817855835, "learning_rate": 8.063065246575175e-05, "loss": 0.0158, "step": 26050 }, { "grad_norm": 0.27375635504722595, "learning_rate": 8.061431391926631e-05, "loss": 0.0193, "step": 26060 }, { "grad_norm": 0.16726228594779968, "learning_rate": 8.059797014163195e-05, "loss": 0.0136, "step": 26070 }, { "grad_norm": 0.16520683467388153, "learning_rate": 8.058162113564133e-05, "loss": 0.0198, "step": 26080 }, { "grad_norm": 0.30444589257240295, "learning_rate": 8.056526690408806e-05, "loss": 0.018, "step": 26090 }, { "grad_norm": 0.26608845591545105, "learning_rate": 8.054890744976666e-05, "loss": 0.0215, "step": 26100 }, { "grad_norm": 0.24862916767597198, "learning_rate": 8.053254277547248e-05, "loss": 0.0159, "step": 26110 }, { "grad_norm": 0.22309905290603638, "learning_rate": 8.051617288400182e-05, "loss": 0.0196, "step": 26120 }, { "grad_norm": 0.1762351095676422, "learning_rate": 8.049979777815182e-05, "loss": 0.0159, "step": 26130 }, { "grad_norm": 0.21879160404205322, "learning_rate": 8.048341746072054e-05, "loss": 0.014, "step": 26140 }, { "grad_norm": 0.3480595052242279, "learning_rate": 8.046703193450696e-05, "loss": 0.0177, "step": 26150 }, { "grad_norm": 0.16641171276569366, "learning_rate": 8.04506412023109e-05, "loss": 0.016, "step": 26160 }, { "grad_norm": 0.19904688000679016, "learning_rate": 8.043424526693306e-05, "loss": 0.017, "step": 26170 }, { "grad_norm": 0.21768105030059814, "learning_rate": 8.04178441311751e-05, "loss": 0.0177, "step": 26180 }, { "grad_norm": 0.2980678081512451, "learning_rate": 8.04014377978395e-05, "loss": 0.0164, "step": 26190 }, { "grad_norm": 0.29815128445625305, "learning_rate": 8.038502626972967e-05, "loss": 0.02, "step": 26200 }, { "grad_norm": 0.23433248698711395, "learning_rate": 8.036860954964989e-05, "loss": 0.0161, "step": 26210 }, { "grad_norm": 0.22237755358219147, "learning_rate": 8.035218764040531e-05, "loss": 0.0173, "step": 26220 }, { "grad_norm": 0.19614411890506744, "learning_rate": 8.033576054480199e-05, "loss": 0.0172, "step": 26230 }, { "grad_norm": 0.22104501724243164, "learning_rate": 8.031932826564688e-05, "loss": 0.0145, "step": 26240 }, { "grad_norm": 0.17241786420345306, "learning_rate": 8.030289080574782e-05, "loss": 0.0182, "step": 26250 }, { "grad_norm": 0.13268941640853882, "learning_rate": 8.028644816791349e-05, "loss": 0.0154, "step": 26260 }, { "grad_norm": 0.20860804617404938, "learning_rate": 8.027000035495351e-05, "loss": 0.0194, "step": 26270 }, { "grad_norm": 0.14374350011348724, "learning_rate": 8.025354736967836e-05, "loss": 0.0165, "step": 26280 }, { "grad_norm": 0.42915233969688416, "learning_rate": 8.023708921489941e-05, "loss": 0.0225, "step": 26290 }, { "grad_norm": 0.2367819994688034, "learning_rate": 8.022062589342887e-05, "loss": 0.0191, "step": 26300 }, { "grad_norm": 0.2495443969964981, "learning_rate": 8.020415740807993e-05, "loss": 0.0206, "step": 26310 }, { "grad_norm": 0.19670744240283966, "learning_rate": 8.018768376166656e-05, "loss": 0.0165, "step": 26320 }, { "grad_norm": 0.2566252648830414, "learning_rate": 8.017120495700368e-05, "loss": 0.0227, "step": 26330 }, { "grad_norm": 0.3333962857723236, "learning_rate": 8.015472099690704e-05, "loss": 0.0226, "step": 26340 }, { "grad_norm": 0.21225205063819885, "learning_rate": 8.013823188419332e-05, "loss": 0.0216, "step": 26350 }, { "grad_norm": 0.248921200633049, "learning_rate": 8.012173762168006e-05, "loss": 0.0189, "step": 26360 }, { "grad_norm": 0.2264903038740158, "learning_rate": 8.010523821218567e-05, "loss": 0.0219, "step": 26370 }, { "grad_norm": 0.2102472484111786, "learning_rate": 8.008873365852945e-05, "loss": 0.0229, "step": 26380 }, { "grad_norm": 0.19292129576206207, "learning_rate": 8.007222396353157e-05, "loss": 0.0195, "step": 26390 }, { "grad_norm": 0.23564974963665009, "learning_rate": 8.00557091300131e-05, "loss": 0.0164, "step": 26400 }, { "grad_norm": 0.20661021769046783, "learning_rate": 8.003918916079597e-05, "loss": 0.0176, "step": 26410 }, { "grad_norm": 0.1911749243736267, "learning_rate": 8.002266405870298e-05, "loss": 0.0167, "step": 26420 }, { "grad_norm": 0.16888748109340668, "learning_rate": 8.000613382655782e-05, "loss": 0.0178, "step": 26430 }, { "grad_norm": 0.25090906023979187, "learning_rate": 7.998959846718505e-05, "loss": 0.0183, "step": 26440 }, { "grad_norm": 0.24484489858150482, "learning_rate": 7.997305798341012e-05, "loss": 0.0218, "step": 26450 }, { "grad_norm": 0.1723419725894928, "learning_rate": 7.995651237805937e-05, "loss": 0.0218, "step": 26460 }, { "grad_norm": 0.1316552758216858, "learning_rate": 7.993996165395996e-05, "loss": 0.0159, "step": 26470 }, { "grad_norm": 0.18268074095249176, "learning_rate": 7.992340581393996e-05, "loss": 0.017, "step": 26480 }, { "grad_norm": 0.3038989007472992, "learning_rate": 7.990684486082831e-05, "loss": 0.0162, "step": 26490 }, { "grad_norm": 0.20681537687778473, "learning_rate": 7.989027879745482e-05, "loss": 0.0194, "step": 26500 }, { "grad_norm": 0.22336646914482117, "learning_rate": 7.98737076266502e-05, "loss": 0.018, "step": 26510 }, { "grad_norm": 0.25910040736198425, "learning_rate": 7.985713135124598e-05, "loss": 0.0202, "step": 26520 }, { "grad_norm": 0.21022431552410126, "learning_rate": 7.98405499740746e-05, "loss": 0.0145, "step": 26530 }, { "grad_norm": 0.19865164160728455, "learning_rate": 7.98239634979694e-05, "loss": 0.0216, "step": 26540 }, { "grad_norm": 0.19621139764785767, "learning_rate": 7.98073719257645e-05, "loss": 0.0182, "step": 26550 }, { "grad_norm": 0.24099236726760864, "learning_rate": 7.979077526029499e-05, "loss": 0.0173, "step": 26560 }, { "grad_norm": 0.25916483998298645, "learning_rate": 7.977417350439675e-05, "loss": 0.0227, "step": 26570 }, { "grad_norm": 0.17773818969726562, "learning_rate": 7.97575666609066e-05, "loss": 0.0185, "step": 26580 }, { "grad_norm": 0.15236838161945343, "learning_rate": 7.974095473266216e-05, "loss": 0.0175, "step": 26590 }, { "grad_norm": 0.22125738859176636, "learning_rate": 7.972433772250198e-05, "loss": 0.0184, "step": 26600 }, { "grad_norm": 0.15011683106422424, "learning_rate": 7.970771563326544e-05, "loss": 0.0171, "step": 26610 }, { "grad_norm": 0.21624815464019775, "learning_rate": 7.96910884677928e-05, "loss": 0.0166, "step": 26620 }, { "grad_norm": 0.24172919988632202, "learning_rate": 7.967445622892523e-05, "loss": 0.0175, "step": 26630 }, { "grad_norm": 0.25450485944747925, "learning_rate": 7.965781891950465e-05, "loss": 0.0232, "step": 26640 }, { "grad_norm": 0.2479478269815445, "learning_rate": 7.964117654237397e-05, "loss": 0.0194, "step": 26650 }, { "grad_norm": 0.18213710188865662, "learning_rate": 7.962452910037692e-05, "loss": 0.0215, "step": 26660 }, { "grad_norm": 0.31983011960983276, "learning_rate": 7.96078765963581e-05, "loss": 0.0194, "step": 26670 }, { "grad_norm": 0.26219451427459717, "learning_rate": 7.95912190331629e-05, "loss": 0.0216, "step": 26680 }, { "grad_norm": 0.1854073405265808, "learning_rate": 7.957455641363772e-05, "loss": 0.0159, "step": 26690 }, { "grad_norm": 0.24277281761169434, "learning_rate": 7.955788874062968e-05, "loss": 0.0171, "step": 26700 }, { "grad_norm": 0.1739330291748047, "learning_rate": 7.95412160169869e-05, "loss": 0.0257, "step": 26710 }, { "grad_norm": 0.22668467462062836, "learning_rate": 7.952453824555824e-05, "loss": 0.0218, "step": 26720 }, { "grad_norm": 0.23060792684555054, "learning_rate": 7.95078554291935e-05, "loss": 0.0167, "step": 26730 }, { "grad_norm": 0.2337273359298706, "learning_rate": 7.94911675707433e-05, "loss": 0.0169, "step": 26740 }, { "grad_norm": 0.17089013755321503, "learning_rate": 7.947447467305915e-05, "loss": 0.0173, "step": 26750 }, { "grad_norm": 0.22993820905685425, "learning_rate": 7.94577767389934e-05, "loss": 0.0189, "step": 26760 }, { "grad_norm": 0.14099247753620148, "learning_rate": 7.944107377139928e-05, "loss": 0.0217, "step": 26770 }, { "grad_norm": 0.17087391018867493, "learning_rate": 7.942436577313088e-05, "loss": 0.0224, "step": 26780 }, { "grad_norm": 0.2258823812007904, "learning_rate": 7.940765274704312e-05, "loss": 0.0189, "step": 26790 }, { "grad_norm": 0.21651621162891388, "learning_rate": 7.939093469599181e-05, "loss": 0.0226, "step": 26800 }, { "grad_norm": 0.22304442524909973, "learning_rate": 7.93742116228336e-05, "loss": 0.0183, "step": 26810 }, { "grad_norm": 0.2701457142829895, "learning_rate": 7.935748353042602e-05, "loss": 0.0178, "step": 26820 }, { "grad_norm": 0.2141541689634323, "learning_rate": 7.934075042162744e-05, "loss": 0.0198, "step": 26830 }, { "grad_norm": 0.29137471318244934, "learning_rate": 7.932401229929705e-05, "loss": 0.0186, "step": 26840 }, { "grad_norm": 0.22168105840682983, "learning_rate": 7.9307269166295e-05, "loss": 0.0159, "step": 26850 }, { "grad_norm": 0.2766057252883911, "learning_rate": 7.92905210254822e-05, "loss": 0.0224, "step": 26860 }, { "grad_norm": 0.17617078125476837, "learning_rate": 7.927376787972045e-05, "loss": 0.0191, "step": 26870 }, { "grad_norm": 0.1843852400779724, "learning_rate": 7.92570097318724e-05, "loss": 0.0201, "step": 26880 }, { "grad_norm": 0.20303784310817719, "learning_rate": 7.924024658480158e-05, "loss": 0.0157, "step": 26890 }, { "grad_norm": 0.12588362395763397, "learning_rate": 7.922347844137233e-05, "loss": 0.0168, "step": 26900 }, { "grad_norm": 0.1706882119178772, "learning_rate": 7.92067053044499e-05, "loss": 0.02, "step": 26910 }, { "grad_norm": 0.18885035812854767, "learning_rate": 7.918992717690031e-05, "loss": 0.016, "step": 26920 }, { "grad_norm": 0.1508454829454422, "learning_rate": 7.917314406159053e-05, "loss": 0.0154, "step": 26930 }, { "grad_norm": 0.23955599963665009, "learning_rate": 7.915635596138832e-05, "loss": 0.0218, "step": 26940 }, { "grad_norm": 0.23278740048408508, "learning_rate": 7.913956287916228e-05, "loss": 0.0184, "step": 26950 }, { "grad_norm": 0.22798581421375275, "learning_rate": 7.912276481778193e-05, "loss": 0.0182, "step": 26960 }, { "grad_norm": 0.1793605089187622, "learning_rate": 7.910596178011759e-05, "loss": 0.014, "step": 26970 }, { "grad_norm": 0.17764948308467865, "learning_rate": 7.908915376904043e-05, "loss": 0.0169, "step": 26980 }, { "grad_norm": 0.23837946355342865, "learning_rate": 7.907234078742247e-05, "loss": 0.0195, "step": 26990 }, { "grad_norm": 0.30956047773361206, "learning_rate": 7.90555228381366e-05, "loss": 0.017, "step": 27000 }, { "grad_norm": 0.19932638108730316, "learning_rate": 7.903869992405656e-05, "loss": 0.021, "step": 27010 }, { "grad_norm": 0.1809656172990799, "learning_rate": 7.902187204805691e-05, "loss": 0.0179, "step": 27020 }, { "grad_norm": 0.15412113070487976, "learning_rate": 7.900503921301308e-05, "loss": 0.0155, "step": 27030 }, { "grad_norm": 0.1982208639383316, "learning_rate": 7.898820142180133e-05, "loss": 0.0175, "step": 27040 }, { "grad_norm": 0.23380592465400696, "learning_rate": 7.897135867729879e-05, "loss": 0.0196, "step": 27050 }, { "grad_norm": 0.1525689661502838, "learning_rate": 7.89545109823834e-05, "loss": 0.016, "step": 27060 }, { "grad_norm": 0.2763403654098511, "learning_rate": 7.8937658339934e-05, "loss": 0.0178, "step": 27070 }, { "grad_norm": 0.2590237855911255, "learning_rate": 7.892080075283026e-05, "loss": 0.0178, "step": 27080 }, { "grad_norm": 0.2571032643318176, "learning_rate": 7.890393822395263e-05, "loss": 0.0192, "step": 27090 }, { "grad_norm": 0.20903541147708893, "learning_rate": 7.88870707561825e-05, "loss": 0.0164, "step": 27100 }, { "grad_norm": 0.3969959616661072, "learning_rate": 7.887019835240203e-05, "loss": 0.0187, "step": 27110 }, { "grad_norm": 0.2180691361427307, "learning_rate": 7.885332101549427e-05, "loss": 0.0148, "step": 27120 }, { "grad_norm": 0.11917319893836975, "learning_rate": 7.883643874834308e-05, "loss": 0.0182, "step": 27130 }, { "grad_norm": 0.221603125333786, "learning_rate": 7.881955155383321e-05, "loss": 0.0176, "step": 27140 }, { "grad_norm": 0.263878732919693, "learning_rate": 7.880265943485017e-05, "loss": 0.0182, "step": 27150 }, { "grad_norm": 0.18983975052833557, "learning_rate": 7.878576239428038e-05, "loss": 0.0191, "step": 27160 }, { "grad_norm": 0.277558296918869, "learning_rate": 7.87688604350111e-05, "loss": 0.0179, "step": 27170 }, { "grad_norm": 0.32795295119285583, "learning_rate": 7.875195355993042e-05, "loss": 0.022, "step": 27180 }, { "grad_norm": 0.2696194052696228, "learning_rate": 7.873504177192724e-05, "loss": 0.0212, "step": 27190 }, { "grad_norm": 0.22328290343284607, "learning_rate": 7.87181250738913e-05, "loss": 0.0191, "step": 27200 }, { "grad_norm": 0.1643459051847458, "learning_rate": 7.870120346871324e-05, "loss": 0.0163, "step": 27210 }, { "grad_norm": 0.13250170648097992, "learning_rate": 7.86842769592845e-05, "loss": 0.0159, "step": 27220 }, { "grad_norm": 0.1888604760169983, "learning_rate": 7.866734554849732e-05, "loss": 0.0187, "step": 27230 }, { "grad_norm": 0.25981956720352173, "learning_rate": 7.865040923924486e-05, "loss": 0.0178, "step": 27240 }, { "grad_norm": 0.19562160968780518, "learning_rate": 7.863346803442104e-05, "loss": 0.0184, "step": 27250 }, { "grad_norm": 0.23496872186660767, "learning_rate": 7.861652193692067e-05, "loss": 0.0158, "step": 27260 }, { "grad_norm": 0.1766899973154068, "learning_rate": 7.859957094963937e-05, "loss": 0.0172, "step": 27270 }, { "grad_norm": 0.20161223411560059, "learning_rate": 7.858261507547357e-05, "loss": 0.0207, "step": 27280 }, { "grad_norm": 0.2228918969631195, "learning_rate": 7.856565431732061e-05, "loss": 0.0206, "step": 27290 }, { "grad_norm": 0.26521119475364685, "learning_rate": 7.854868867807859e-05, "loss": 0.0191, "step": 27300 }, { "grad_norm": 0.16465657949447632, "learning_rate": 7.85317181606465e-05, "loss": 0.0165, "step": 27310 }, { "grad_norm": 0.24992656707763672, "learning_rate": 7.85147427679241e-05, "loss": 0.0163, "step": 27320 }, { "grad_norm": 0.28345608711242676, "learning_rate": 7.849776250281205e-05, "loss": 0.0192, "step": 27330 }, { "grad_norm": 0.24517148733139038, "learning_rate": 7.84807773682118e-05, "loss": 0.0171, "step": 27340 }, { "grad_norm": 0.16629482805728912, "learning_rate": 7.846378736702565e-05, "loss": 0.0187, "step": 27350 }, { "grad_norm": 0.2079576551914215, "learning_rate": 7.844679250215671e-05, "loss": 0.0207, "step": 27360 }, { "grad_norm": 0.17714908719062805, "learning_rate": 7.842979277650898e-05, "loss": 0.0204, "step": 27370 }, { "grad_norm": 0.2600739002227783, "learning_rate": 7.84127881929872e-05, "loss": 0.0146, "step": 27380 }, { "grad_norm": 0.1856270730495453, "learning_rate": 7.839577875449704e-05, "loss": 0.0175, "step": 27390 }, { "grad_norm": 0.2227887362241745, "learning_rate": 7.837876446394489e-05, "loss": 0.0182, "step": 27400 }, { "grad_norm": 0.17533469200134277, "learning_rate": 7.836174532423805e-05, "loss": 0.0158, "step": 27410 }, { "grad_norm": 0.28432318568229675, "learning_rate": 7.834472133828466e-05, "loss": 0.0192, "step": 27420 }, { "grad_norm": 0.26406362652778625, "learning_rate": 7.832769250899359e-05, "loss": 0.0179, "step": 27430 }, { "grad_norm": 0.20696599781513214, "learning_rate": 7.831065883927464e-05, "loss": 0.0152, "step": 27440 }, { "grad_norm": 0.220140740275383, "learning_rate": 7.829362033203841e-05, "loss": 0.018, "step": 27450 }, { "grad_norm": 0.2073645293712616, "learning_rate": 7.827657699019628e-05, "loss": 0.0231, "step": 27460 }, { "grad_norm": 0.27027228474617004, "learning_rate": 7.825952881666052e-05, "loss": 0.0183, "step": 27470 }, { "grad_norm": 0.20586511492729187, "learning_rate": 7.824247581434418e-05, "loss": 0.0151, "step": 27480 }, { "grad_norm": 0.2807055711746216, "learning_rate": 7.822541798616116e-05, "loss": 0.022, "step": 27490 }, { "grad_norm": 0.23311340808868408, "learning_rate": 7.820835533502617e-05, "loss": 0.0208, "step": 27500 }, { "grad_norm": 0.2254616767168045, "learning_rate": 7.819128786385475e-05, "loss": 0.0189, "step": 27510 }, { "grad_norm": 0.2439846396446228, "learning_rate": 7.817421557556329e-05, "loss": 0.0197, "step": 27520 }, { "grad_norm": 0.21753011643886566, "learning_rate": 7.815713847306893e-05, "loss": 0.0152, "step": 27530 }, { "grad_norm": 0.19936597347259521, "learning_rate": 7.81400565592897e-05, "loss": 0.0186, "step": 27540 }, { "grad_norm": 0.18053178489208221, "learning_rate": 7.812296983714444e-05, "loss": 0.014, "step": 27550 }, { "grad_norm": 0.16701172292232513, "learning_rate": 7.810587830955281e-05, "loss": 0.0174, "step": 27560 }, { "grad_norm": 0.3363552391529083, "learning_rate": 7.808878197943528e-05, "loss": 0.018, "step": 27570 }, { "grad_norm": 0.22711005806922913, "learning_rate": 7.807168084971312e-05, "loss": 0.0161, "step": 27580 }, { "grad_norm": 0.214948371052742, "learning_rate": 7.805457492330849e-05, "loss": 0.0139, "step": 27590 }, { "grad_norm": 0.19328898191452026, "learning_rate": 7.803746420314428e-05, "loss": 0.0171, "step": 27600 }, { "grad_norm": 0.2892153263092041, "learning_rate": 7.802034869214428e-05, "loss": 0.019, "step": 27610 }, { "grad_norm": 0.2626413106918335, "learning_rate": 7.800322839323303e-05, "loss": 0.017, "step": 27620 }, { "grad_norm": 0.2864058315753937, "learning_rate": 7.798610330933593e-05, "loss": 0.0165, "step": 27630 }, { "grad_norm": 0.27523529529571533, "learning_rate": 7.796897344337922e-05, "loss": 0.0158, "step": 27640 }, { "grad_norm": 0.2516855299472809, "learning_rate": 7.795183879828989e-05, "loss": 0.0209, "step": 27650 }, { "grad_norm": 0.21279339492321014, "learning_rate": 7.793469937699579e-05, "loss": 0.0168, "step": 27660 }, { "grad_norm": 0.28625625371932983, "learning_rate": 7.791755518242558e-05, "loss": 0.017, "step": 27670 }, { "grad_norm": 0.26904070377349854, "learning_rate": 7.790040621750876e-05, "loss": 0.0186, "step": 27680 }, { "grad_norm": 0.29993537068367004, "learning_rate": 7.788325248517558e-05, "loss": 0.0176, "step": 27690 }, { "grad_norm": 0.18057037889957428, "learning_rate": 7.786609398835715e-05, "loss": 0.015, "step": 27700 }, { "grad_norm": 0.16918900609016418, "learning_rate": 7.784893072998541e-05, "loss": 0.0151, "step": 27710 }, { "grad_norm": 0.18634778261184692, "learning_rate": 7.783176271299306e-05, "loss": 0.014, "step": 27720 }, { "grad_norm": 0.20278573036193848, "learning_rate": 7.781458994031368e-05, "loss": 0.0181, "step": 27730 }, { "grad_norm": 0.2508774697780609, "learning_rate": 7.779741241488161e-05, "loss": 0.0151, "step": 27740 }, { "grad_norm": 0.1972876787185669, "learning_rate": 7.7780230139632e-05, "loss": 0.0166, "step": 27750 }, { "grad_norm": 0.27193450927734375, "learning_rate": 7.776304311750087e-05, "loss": 0.0194, "step": 27760 }, { "grad_norm": 0.24764063954353333, "learning_rate": 7.7745851351425e-05, "loss": 0.0191, "step": 27770 }, { "grad_norm": 0.24483706057071686, "learning_rate": 7.772865484434197e-05, "loss": 0.0191, "step": 27780 }, { "grad_norm": 0.20954173803329468, "learning_rate": 7.77114535991902e-05, "loss": 0.0163, "step": 27790 }, { "grad_norm": 0.20929890871047974, "learning_rate": 7.769424761890893e-05, "loss": 0.0158, "step": 27800 }, { "grad_norm": 0.19835373759269714, "learning_rate": 7.767703690643817e-05, "loss": 0.0206, "step": 27810 }, { "grad_norm": 0.16779641807079315, "learning_rate": 7.76598214647188e-05, "loss": 0.0159, "step": 27820 }, { "grad_norm": 0.23552998900413513, "learning_rate": 7.764260129669241e-05, "loss": 0.0168, "step": 27830 }, { "grad_norm": 0.16377878189086914, "learning_rate": 7.76253764053015e-05, "loss": 0.0171, "step": 27840 }, { "grad_norm": 0.19073301553726196, "learning_rate": 7.760814679348932e-05, "loss": 0.0148, "step": 27850 }, { "grad_norm": 0.14500193297863007, "learning_rate": 7.759091246419992e-05, "loss": 0.0138, "step": 27860 }, { "grad_norm": 0.2481333464384079, "learning_rate": 7.75736734203782e-05, "loss": 0.0145, "step": 27870 }, { "grad_norm": 0.2097414880990982, "learning_rate": 7.755642966496985e-05, "loss": 0.0155, "step": 27880 }, { "grad_norm": 0.1816910058259964, "learning_rate": 7.753918120092132e-05, "loss": 0.0174, "step": 27890 }, { "grad_norm": 0.28822267055511475, "learning_rate": 7.752192803117993e-05, "loss": 0.017, "step": 27900 }, { "grad_norm": 0.22015750408172607, "learning_rate": 7.750467015869377e-05, "loss": 0.0184, "step": 27910 }, { "grad_norm": 0.26692986488342285, "learning_rate": 7.748740758641174e-05, "loss": 0.0207, "step": 27920 }, { "grad_norm": 0.27723923325538635, "learning_rate": 7.74701403172835e-05, "loss": 0.0229, "step": 27930 }, { "grad_norm": 0.13533559441566467, "learning_rate": 7.745286835425962e-05, "loss": 0.0177, "step": 27940 }, { "grad_norm": 0.15489225089550018, "learning_rate": 7.743559170029138e-05, "loss": 0.0193, "step": 27950 }, { "grad_norm": 0.18706995248794556, "learning_rate": 7.741831035833087e-05, "loss": 0.0159, "step": 27960 }, { "grad_norm": 0.18695344030857086, "learning_rate": 7.740102433133102e-05, "loss": 0.0174, "step": 27970 }, { "grad_norm": 0.22704751789569855, "learning_rate": 7.738373362224553e-05, "loss": 0.0179, "step": 27980 }, { "grad_norm": 0.23528185486793518, "learning_rate": 7.73664382340289e-05, "loss": 0.0133, "step": 27990 }, { "grad_norm": 0.19135212898254395, "learning_rate": 7.734913816963647e-05, "loss": 0.0177, "step": 28000 }, { "grad_norm": 0.22864465415477753, "learning_rate": 7.73318334320243e-05, "loss": 0.0131, "step": 28010 }, { "grad_norm": 0.27858367562294006, "learning_rate": 7.731452402414934e-05, "loss": 0.0221, "step": 28020 }, { "grad_norm": 0.2354227751493454, "learning_rate": 7.729720994896928e-05, "loss": 0.0203, "step": 28030 }, { "grad_norm": 0.23624329268932343, "learning_rate": 7.727989120944262e-05, "loss": 0.0218, "step": 28040 }, { "grad_norm": 0.17157138884067535, "learning_rate": 7.726256780852865e-05, "loss": 0.0166, "step": 28050 }, { "grad_norm": 0.19175079464912415, "learning_rate": 7.724523974918749e-05, "loss": 0.0155, "step": 28060 }, { "grad_norm": 0.15520350635051727, "learning_rate": 7.722790703438002e-05, "loss": 0.0163, "step": 28070 }, { "grad_norm": 0.18374775350093842, "learning_rate": 7.72105696670679e-05, "loss": 0.0154, "step": 28080 }, { "grad_norm": 0.10928504914045334, "learning_rate": 7.719322765021364e-05, "loss": 0.017, "step": 28090 }, { "grad_norm": 0.23526574671268463, "learning_rate": 7.717588098678051e-05, "loss": 0.0222, "step": 28100 }, { "grad_norm": 0.17588341236114502, "learning_rate": 7.715852967973258e-05, "loss": 0.0166, "step": 28110 }, { "grad_norm": 0.16606505215168, "learning_rate": 7.714117373203474e-05, "loss": 0.0176, "step": 28120 }, { "grad_norm": 0.2378820925951004, "learning_rate": 7.712381314665259e-05, "loss": 0.0192, "step": 28130 }, { "grad_norm": 0.18068408966064453, "learning_rate": 7.710644792655261e-05, "loss": 0.015, "step": 28140 }, { "grad_norm": 0.2356594353914261, "learning_rate": 7.708907807470207e-05, "loss": 0.0147, "step": 28150 }, { "grad_norm": 0.2041177898645401, "learning_rate": 7.707170359406896e-05, "loss": 0.0138, "step": 28160 }, { "grad_norm": 0.25779086351394653, "learning_rate": 7.705432448762213e-05, "loss": 0.0133, "step": 28170 }, { "grad_norm": 0.22474898397922516, "learning_rate": 7.703694075833117e-05, "loss": 0.015, "step": 28180 }, { "grad_norm": 0.23165751993656158, "learning_rate": 7.70195524091665e-05, "loss": 0.0135, "step": 28190 }, { "grad_norm": 0.23351511359214783, "learning_rate": 7.70021594430993e-05, "loss": 0.0164, "step": 28200 }, { "grad_norm": 0.20435777306556702, "learning_rate": 7.698476186310157e-05, "loss": 0.016, "step": 28210 }, { "grad_norm": 0.1782875806093216, "learning_rate": 7.696735967214608e-05, "loss": 0.0164, "step": 28220 }, { "grad_norm": 0.17049841582775116, "learning_rate": 7.694995287320636e-05, "loss": 0.0167, "step": 28230 }, { "grad_norm": 0.17098307609558105, "learning_rate": 7.693254146925679e-05, "loss": 0.0149, "step": 28240 }, { "grad_norm": 0.18039575219154358, "learning_rate": 7.691512546327251e-05, "loss": 0.0173, "step": 28250 }, { "grad_norm": 0.3066008388996124, "learning_rate": 7.689770485822939e-05, "loss": 0.0158, "step": 28260 }, { "grad_norm": 0.21860307455062866, "learning_rate": 7.688027965710416e-05, "loss": 0.0162, "step": 28270 }, { "grad_norm": 0.15824604034423828, "learning_rate": 7.686284986287433e-05, "loss": 0.017, "step": 28280 }, { "grad_norm": 0.2581809461116791, "learning_rate": 7.684541547851817e-05, "loss": 0.0141, "step": 28290 }, { "grad_norm": 0.22271133959293365, "learning_rate": 7.68279765070147e-05, "loss": 0.0148, "step": 28300 }, { "grad_norm": 0.1852310746908188, "learning_rate": 7.68105329513438e-05, "loss": 0.0129, "step": 28310 }, { "grad_norm": 0.1999257355928421, "learning_rate": 7.67930848144861e-05, "loss": 0.0188, "step": 28320 }, { "grad_norm": 0.17374582588672638, "learning_rate": 7.6775632099423e-05, "loss": 0.0149, "step": 28330 }, { "grad_norm": 0.18210244178771973, "learning_rate": 7.675817480913667e-05, "loss": 0.0137, "step": 28340 }, { "grad_norm": 0.1648133546113968, "learning_rate": 7.674071294661011e-05, "loss": 0.0158, "step": 28350 }, { "grad_norm": 0.22614328563213348, "learning_rate": 7.672324651482707e-05, "loss": 0.0187, "step": 28360 }, { "grad_norm": 0.20367814600467682, "learning_rate": 7.670577551677209e-05, "loss": 0.0168, "step": 28370 }, { "grad_norm": 0.23384252190589905, "learning_rate": 7.668829995543047e-05, "loss": 0.0175, "step": 28380 }, { "grad_norm": 0.2942851483821869, "learning_rate": 7.667081983378832e-05, "loss": 0.0174, "step": 28390 }, { "grad_norm": 0.4199114739894867, "learning_rate": 7.66533351548325e-05, "loss": 0.0202, "step": 28400 }, { "grad_norm": 0.1551436185836792, "learning_rate": 7.663584592155069e-05, "loss": 0.0186, "step": 28410 }, { "grad_norm": 0.2078847885131836, "learning_rate": 7.661835213693129e-05, "loss": 0.0167, "step": 28420 }, { "grad_norm": 0.17043060064315796, "learning_rate": 7.660085380396353e-05, "loss": 0.0148, "step": 28430 }, { "grad_norm": 0.21041090786457062, "learning_rate": 7.658335092563738e-05, "loss": 0.0175, "step": 28440 }, { "grad_norm": 0.15283158421516418, "learning_rate": 7.656584350494362e-05, "loss": 0.0156, "step": 28450 }, { "grad_norm": 0.24200581014156342, "learning_rate": 7.654833154487378e-05, "loss": 0.0203, "step": 28460 }, { "grad_norm": 0.2286144345998764, "learning_rate": 7.653081504842017e-05, "loss": 0.0166, "step": 28470 }, { "grad_norm": 0.22240906953811646, "learning_rate": 7.65132940185759e-05, "loss": 0.0183, "step": 28480 }, { "grad_norm": 0.24478933215141296, "learning_rate": 7.649576845833481e-05, "loss": 0.0188, "step": 28490 }, { "grad_norm": 0.1774442046880722, "learning_rate": 7.647823837069156e-05, "loss": 0.0165, "step": 28500 }, { "grad_norm": 0.19650237262248993, "learning_rate": 7.646070375864156e-05, "loss": 0.0138, "step": 28510 }, { "grad_norm": 0.16720226407051086, "learning_rate": 7.644316462518099e-05, "loss": 0.0162, "step": 28520 }, { "grad_norm": 0.16754992306232452, "learning_rate": 7.642562097330679e-05, "loss": 0.014, "step": 28530 }, { "grad_norm": 0.23505759239196777, "learning_rate": 7.640807280601671e-05, "loss": 0.0144, "step": 28540 }, { "grad_norm": 0.28359919786453247, "learning_rate": 7.639052012630927e-05, "loss": 0.0173, "step": 28550 }, { "grad_norm": 0.2182883471250534, "learning_rate": 7.63729629371837e-05, "loss": 0.0126, "step": 28560 }, { "grad_norm": 0.14314959943294525, "learning_rate": 7.635540124164009e-05, "loss": 0.0129, "step": 28570 }, { "grad_norm": 0.20725011825561523, "learning_rate": 7.633783504267922e-05, "loss": 0.0151, "step": 28580 }, { "grad_norm": 0.21820425987243652, "learning_rate": 7.632026434330269e-05, "loss": 0.016, "step": 28590 }, { "grad_norm": 0.1836836338043213, "learning_rate": 7.630268914651282e-05, "loss": 0.0164, "step": 28600 }, { "grad_norm": 0.1522323191165924, "learning_rate": 7.628510945531278e-05, "loss": 0.0184, "step": 28610 }, { "grad_norm": 0.18437445163726807, "learning_rate": 7.626752527270641e-05, "loss": 0.0196, "step": 28620 }, { "grad_norm": 0.24028848111629486, "learning_rate": 7.62499366016984e-05, "loss": 0.0171, "step": 28630 }, { "grad_norm": 0.20290352404117584, "learning_rate": 7.623234344529416e-05, "loss": 0.0176, "step": 28640 }, { "grad_norm": 0.20889298617839813, "learning_rate": 7.62147458064999e-05, "loss": 0.0158, "step": 28650 }, { "grad_norm": 0.2466723471879959, "learning_rate": 7.619714368832254e-05, "loss": 0.0154, "step": 28660 }, { "grad_norm": 0.3315739929676056, "learning_rate": 7.61795370937698e-05, "loss": 0.0173, "step": 28670 }, { "grad_norm": 0.24827443063259125, "learning_rate": 7.61619260258502e-05, "loss": 0.0173, "step": 28680 }, { "grad_norm": 0.22331823408603668, "learning_rate": 7.614431048757298e-05, "loss": 0.0171, "step": 28690 }, { "grad_norm": 0.2104939967393875, "learning_rate": 7.612669048194814e-05, "loss": 0.0169, "step": 28700 }, { "grad_norm": 0.1515272557735443, "learning_rate": 7.610906601198646e-05, "loss": 0.0141, "step": 28710 }, { "grad_norm": 0.138119637966156, "learning_rate": 7.60914370806995e-05, "loss": 0.0163, "step": 28720 }, { "grad_norm": 0.22637459635734558, "learning_rate": 7.607380369109953e-05, "loss": 0.0146, "step": 28730 }, { "grad_norm": 0.21432895958423615, "learning_rate": 7.605616584619961e-05, "loss": 0.0157, "step": 28740 }, { "grad_norm": 0.20013290643692017, "learning_rate": 7.603852354901362e-05, "loss": 0.0157, "step": 28750 }, { "grad_norm": 0.19601787626743317, "learning_rate": 7.602087680255609e-05, "loss": 0.017, "step": 28760 }, { "grad_norm": 0.18898095190525055, "learning_rate": 7.600322560984238e-05, "loss": 0.0189, "step": 28770 }, { "grad_norm": 0.377103328704834, "learning_rate": 7.598556997388863e-05, "loss": 0.0222, "step": 28780 }, { "grad_norm": 0.20937316119670868, "learning_rate": 7.596790989771166e-05, "loss": 0.0168, "step": 28790 }, { "grad_norm": 0.2581440210342407, "learning_rate": 7.595024538432914e-05, "loss": 0.0157, "step": 28800 }, { "grad_norm": 0.3010597825050354, "learning_rate": 7.59325764367594e-05, "loss": 0.0162, "step": 28810 }, { "grad_norm": 0.17481502890586853, "learning_rate": 7.59149030580216e-05, "loss": 0.0133, "step": 28820 }, { "grad_norm": 0.1728338897228241, "learning_rate": 7.589722525113562e-05, "loss": 0.0134, "step": 28830 }, { "grad_norm": 0.18381191790103912, "learning_rate": 7.587954301912216e-05, "loss": 0.0135, "step": 28840 }, { "grad_norm": 0.22321443259716034, "learning_rate": 7.586185636500263e-05, "loss": 0.0163, "step": 28850 }, { "grad_norm": 0.29954975843429565, "learning_rate": 7.584416529179914e-05, "loss": 0.0201, "step": 28860 }, { "grad_norm": 0.26960960030555725, "learning_rate": 7.582646980253465e-05, "loss": 0.02, "step": 28870 }, { "grad_norm": 0.24322021007537842, "learning_rate": 7.580876990023282e-05, "loss": 0.021, "step": 28880 }, { "grad_norm": 0.18112966418266296, "learning_rate": 7.579106558791809e-05, "loss": 0.0158, "step": 28890 }, { "grad_norm": 0.2857503592967987, "learning_rate": 7.577335686861565e-05, "loss": 0.0226, "step": 28900 }, { "grad_norm": 0.23392003774642944, "learning_rate": 7.575564374535141e-05, "loss": 0.0166, "step": 28910 }, { "grad_norm": 0.2678127586841583, "learning_rate": 7.573792622115207e-05, "loss": 0.0167, "step": 28920 }, { "grad_norm": 0.22793817520141602, "learning_rate": 7.572020429904507e-05, "loss": 0.0194, "step": 28930 }, { "grad_norm": 0.23788058757781982, "learning_rate": 7.570247798205861e-05, "loss": 0.0141, "step": 28940 }, { "grad_norm": 0.2243138998746872, "learning_rate": 7.568474727322164e-05, "loss": 0.0149, "step": 28950 }, { "grad_norm": 0.17849977314472198, "learning_rate": 7.566701217556384e-05, "loss": 0.0144, "step": 28960 }, { "grad_norm": 0.18653862178325653, "learning_rate": 7.564927269211564e-05, "loss": 0.0134, "step": 28970 }, { "grad_norm": 0.1660635620355606, "learning_rate": 7.563152882590824e-05, "loss": 0.0136, "step": 28980 }, { "grad_norm": 0.18942807614803314, "learning_rate": 7.56137805799736e-05, "loss": 0.0171, "step": 28990 }, { "grad_norm": 0.20034457743167877, "learning_rate": 7.559602795734439e-05, "loss": 0.0139, "step": 29000 }, { "grad_norm": 0.1881539672613144, "learning_rate": 7.557827096105408e-05, "loss": 0.0179, "step": 29010 }, { "grad_norm": 0.1985853612422943, "learning_rate": 7.55605095941368e-05, "loss": 0.0147, "step": 29020 }, { "grad_norm": 0.1815706193447113, "learning_rate": 7.554274385962752e-05, "loss": 0.0151, "step": 29030 }, { "grad_norm": 0.2738800644874573, "learning_rate": 7.552497376056191e-05, "loss": 0.0174, "step": 29040 }, { "grad_norm": 0.16295196115970612, "learning_rate": 7.550719929997639e-05, "loss": 0.0151, "step": 29050 }, { "grad_norm": 0.12283769249916077, "learning_rate": 7.548942048090813e-05, "loss": 0.0153, "step": 29060 }, { "grad_norm": 0.19756336510181427, "learning_rate": 7.547163730639506e-05, "loss": 0.0143, "step": 29070 }, { "grad_norm": 0.22769524157047272, "learning_rate": 7.545384977947583e-05, "loss": 0.0153, "step": 29080 }, { "grad_norm": 0.18331190943717957, "learning_rate": 7.543605790318981e-05, "loss": 0.018, "step": 29090 }, { "grad_norm": 0.23307950794696808, "learning_rate": 7.54182616805772e-05, "loss": 0.0165, "step": 29100 }, { "grad_norm": 0.21224969625473022, "learning_rate": 7.540046111467885e-05, "loss": 0.0176, "step": 29110 }, { "grad_norm": 0.176890030503273, "learning_rate": 7.53826562085364e-05, "loss": 0.0164, "step": 29120 }, { "grad_norm": 0.21200095117092133, "learning_rate": 7.536484696519221e-05, "loss": 0.0152, "step": 29130 }, { "grad_norm": 0.2353542596101761, "learning_rate": 7.534703338768942e-05, "loss": 0.0177, "step": 29140 }, { "grad_norm": 0.30349698662757874, "learning_rate": 7.532921547907185e-05, "loss": 0.0188, "step": 29150 }, { "grad_norm": 0.19455628097057343, "learning_rate": 7.531139324238412e-05, "loss": 0.0183, "step": 29160 }, { "grad_norm": 0.23248298466205597, "learning_rate": 7.529356668067157e-05, "loss": 0.0195, "step": 29170 }, { "grad_norm": 0.21303494274616241, "learning_rate": 7.527573579698023e-05, "loss": 0.0177, "step": 29180 }, { "grad_norm": 0.21690233051776886, "learning_rate": 7.525790059435693e-05, "loss": 0.0189, "step": 29190 }, { "grad_norm": 0.20451460778713226, "learning_rate": 7.524006107584926e-05, "loss": 0.0191, "step": 29200 }, { "grad_norm": 0.2596215009689331, "learning_rate": 7.522221724450544e-05, "loss": 0.0185, "step": 29210 }, { "grad_norm": 0.21029475331306458, "learning_rate": 7.520436910337451e-05, "loss": 0.0186, "step": 29220 }, { "grad_norm": 0.20148348808288574, "learning_rate": 7.518651665550627e-05, "loss": 0.015, "step": 29230 }, { "grad_norm": 0.26064470410346985, "learning_rate": 7.516865990395117e-05, "loss": 0.0173, "step": 29240 }, { "grad_norm": 0.22721435129642487, "learning_rate": 7.515079885176047e-05, "loss": 0.0133, "step": 29250 }, { "grad_norm": 0.21834099292755127, "learning_rate": 7.513293350198612e-05, "loss": 0.0154, "step": 29260 }, { "grad_norm": 0.2314217984676361, "learning_rate": 7.511506385768081e-05, "loss": 0.0173, "step": 29270 }, { "grad_norm": 0.14462529122829437, "learning_rate": 7.509718992189801e-05, "loss": 0.0157, "step": 29280 }, { "grad_norm": 0.22993847727775574, "learning_rate": 7.507931169769182e-05, "loss": 0.0205, "step": 29290 }, { "grad_norm": 0.17812055349349976, "learning_rate": 7.506142918811722e-05, "loss": 0.0132, "step": 29300 }, { "grad_norm": 0.1938331425189972, "learning_rate": 7.504354239622978e-05, "loss": 0.0143, "step": 29310 }, { "grad_norm": 0.2037251889705658, "learning_rate": 7.50256513250859e-05, "loss": 0.0156, "step": 29320 }, { "grad_norm": 0.1894034445285797, "learning_rate": 7.500775597774265e-05, "loss": 0.0189, "step": 29330 }, { "grad_norm": 0.23213772475719452, "learning_rate": 7.498985635725788e-05, "loss": 0.0154, "step": 29340 }, { "grad_norm": 0.24242962896823883, "learning_rate": 7.497195246669012e-05, "loss": 0.0219, "step": 29350 }, { "grad_norm": 0.1365579515695572, "learning_rate": 7.495404430909868e-05, "loss": 0.0162, "step": 29360 }, { "grad_norm": 0.18208299577236176, "learning_rate": 7.493613188754356e-05, "loss": 0.0133, "step": 29370 }, { "grad_norm": 0.2151040881872177, "learning_rate": 7.49182152050855e-05, "loss": 0.0173, "step": 29380 }, { "grad_norm": 0.2544623613357544, "learning_rate": 7.490029426478598e-05, "loss": 0.0181, "step": 29390 }, { "grad_norm": 0.16920466721057892, "learning_rate": 7.488236906970719e-05, "loss": 0.019, "step": 29400 }, { "grad_norm": 0.24757717549800873, "learning_rate": 7.486443962291207e-05, "loss": 0.018, "step": 29410 }, { "grad_norm": 0.3182554543018341, "learning_rate": 7.484650592746424e-05, "loss": 0.0162, "step": 29420 }, { "grad_norm": 0.26034507155418396, "learning_rate": 7.482856798642811e-05, "loss": 0.0191, "step": 29430 }, { "grad_norm": 0.1899360567331314, "learning_rate": 7.481062580286878e-05, "loss": 0.015, "step": 29440 }, { "grad_norm": 0.24522525072097778, "learning_rate": 7.479267937985208e-05, "loss": 0.0168, "step": 29450 }, { "grad_norm": 0.23996606469154358, "learning_rate": 7.477472872044456e-05, "loss": 0.0161, "step": 29460 }, { "grad_norm": 0.27101126313209534, "learning_rate": 7.475677382771347e-05, "loss": 0.0135, "step": 29470 }, { "grad_norm": 0.14324261248111725, "learning_rate": 7.473881470472683e-05, "loss": 0.0134, "step": 29480 }, { "grad_norm": 0.2253563553094864, "learning_rate": 7.472085135455337e-05, "loss": 0.0161, "step": 29490 }, { "grad_norm": 0.18337319791316986, "learning_rate": 7.470288378026256e-05, "loss": 0.0147, "step": 29500 }, { "grad_norm": 0.2152562290430069, "learning_rate": 7.468491198492451e-05, "loss": 0.018, "step": 29510 }, { "grad_norm": 0.2532086968421936, "learning_rate": 7.466693597161013e-05, "loss": 0.0182, "step": 29520 }, { "grad_norm": 0.2745910882949829, "learning_rate": 7.464895574339104e-05, "loss": 0.0175, "step": 29530 }, { "grad_norm": 0.1719479262828827, "learning_rate": 7.463097130333958e-05, "loss": 0.0189, "step": 29540 }, { "grad_norm": 0.14230994880199432, "learning_rate": 7.461298265452876e-05, "loss": 0.015, "step": 29550 }, { "grad_norm": 0.19993475079536438, "learning_rate": 7.459498980003239e-05, "loss": 0.0176, "step": 29560 }, { "grad_norm": 0.20404842495918274, "learning_rate": 7.457699274292493e-05, "loss": 0.0148, "step": 29570 }, { "grad_norm": 0.2351025938987732, "learning_rate": 7.455899148628159e-05, "loss": 0.0165, "step": 29580 }, { "grad_norm": 0.28364622592926025, "learning_rate": 7.45409860331783e-05, "loss": 0.0183, "step": 29590 }, { "grad_norm": 0.24379302561283112, "learning_rate": 7.452297638669169e-05, "loss": 0.0205, "step": 29600 }, { "grad_norm": 0.22781074047088623, "learning_rate": 7.450496254989911e-05, "loss": 0.018, "step": 29610 }, { "grad_norm": 0.14984136819839478, "learning_rate": 7.448694452587866e-05, "loss": 0.0143, "step": 29620 }, { "grad_norm": 0.23717498779296875, "learning_rate": 7.44689223177091e-05, "loss": 0.0176, "step": 29630 }, { "grad_norm": 0.21530123054981232, "learning_rate": 7.445089592846994e-05, "loss": 0.0165, "step": 29640 }, { "grad_norm": 0.17381811141967773, "learning_rate": 7.443286536124141e-05, "loss": 0.0174, "step": 29650 }, { "grad_norm": 0.1829209178686142, "learning_rate": 7.441483061910443e-05, "loss": 0.0175, "step": 29660 }, { "grad_norm": 0.19616596400737762, "learning_rate": 7.439679170514064e-05, "loss": 0.0147, "step": 29670 }, { "grad_norm": 0.19076453149318695, "learning_rate": 7.43787486224324e-05, "loss": 0.0176, "step": 29680 }, { "grad_norm": 0.2504955530166626, "learning_rate": 7.436070137406276e-05, "loss": 0.0136, "step": 29690 }, { "grad_norm": 0.18568307161331177, "learning_rate": 7.434264996311556e-05, "loss": 0.0161, "step": 29700 }, { "grad_norm": 0.24132277071475983, "learning_rate": 7.432459439267525e-05, "loss": 0.0173, "step": 29710 }, { "grad_norm": 0.20731371641159058, "learning_rate": 7.430653466582701e-05, "loss": 0.0144, "step": 29720 }, { "grad_norm": 0.21597446501255035, "learning_rate": 7.42884707856568e-05, "loss": 0.0139, "step": 29730 }, { "grad_norm": 0.2174377292394638, "learning_rate": 7.427040275525122e-05, "loss": 0.0185, "step": 29740 }, { "grad_norm": 0.11965080350637436, "learning_rate": 7.42523305776976e-05, "loss": 0.0168, "step": 29750 }, { "grad_norm": 0.24644549190998077, "learning_rate": 7.4234254256084e-05, "loss": 0.0162, "step": 29760 }, { "grad_norm": 0.22957997024059296, "learning_rate": 7.421617379349915e-05, "loss": 0.0146, "step": 29770 }, { "grad_norm": 0.19563119113445282, "learning_rate": 7.41980891930325e-05, "loss": 0.0149, "step": 29780 }, { "grad_norm": 0.21293900907039642, "learning_rate": 7.418000045777425e-05, "loss": 0.0146, "step": 29790 }, { "grad_norm": 0.2201579362154007, "learning_rate": 7.416190759081523e-05, "loss": 0.0168, "step": 29800 }, { "grad_norm": 0.2363402247428894, "learning_rate": 7.414381059524704e-05, "loss": 0.0149, "step": 29810 }, { "grad_norm": 0.23852957785129547, "learning_rate": 7.412570947416195e-05, "loss": 0.0136, "step": 29820 }, { "grad_norm": 0.16518226265907288, "learning_rate": 7.410760423065295e-05, "loss": 0.0131, "step": 29830 }, { "grad_norm": 0.23654280602931976, "learning_rate": 7.408949486781372e-05, "loss": 0.0182, "step": 29840 }, { "grad_norm": 0.15997308492660522, "learning_rate": 7.407138138873868e-05, "loss": 0.0139, "step": 29850 }, { "grad_norm": 0.20176146924495697, "learning_rate": 7.405326379652292e-05, "loss": 0.0168, "step": 29860 }, { "grad_norm": 0.274819940328598, "learning_rate": 7.403514209426222e-05, "loss": 0.018, "step": 29870 }, { "grad_norm": 0.24298730492591858, "learning_rate": 7.40170162850531e-05, "loss": 0.0168, "step": 29880 }, { "grad_norm": 0.21843959391117096, "learning_rate": 7.399888637199278e-05, "loss": 0.0156, "step": 29890 }, { "grad_norm": 0.22072571516036987, "learning_rate": 7.398075235817914e-05, "loss": 0.0167, "step": 29900 }, { "grad_norm": 0.14973242580890656, "learning_rate": 7.39626142467108e-05, "loss": 0.0171, "step": 29910 }, { "grad_norm": 0.14933545887470245, "learning_rate": 7.394447204068706e-05, "loss": 0.0143, "step": 29920 }, { "grad_norm": 0.1829163283109665, "learning_rate": 7.392632574320793e-05, "loss": 0.0143, "step": 29930 }, { "grad_norm": 0.23208282887935638, "learning_rate": 7.390817535737411e-05, "loss": 0.0164, "step": 29940 }, { "grad_norm": 0.23050470650196075, "learning_rate": 7.389002088628703e-05, "loss": 0.0146, "step": 29950 }, { "grad_norm": 0.14398479461669922, "learning_rate": 7.387186233304877e-05, "loss": 0.0141, "step": 29960 }, { "grad_norm": 0.2613993287086487, "learning_rate": 7.385369970076212e-05, "loss": 0.013, "step": 29970 }, { "grad_norm": 0.22618389129638672, "learning_rate": 7.38355329925306e-05, "loss": 0.0139, "step": 29980 }, { "grad_norm": 0.17794635891914368, "learning_rate": 7.381736221145838e-05, "loss": 0.0134, "step": 29990 }, { "grad_norm": 0.1745295375585556, "learning_rate": 7.37991873606504e-05, "loss": 0.0141, "step": 30000 }, { "grad_norm": 0.20732100307941437, "learning_rate": 7.378100844321218e-05, "loss": 0.0198, "step": 30010 }, { "grad_norm": 0.2602447271347046, "learning_rate": 7.376282546225004e-05, "loss": 0.0184, "step": 30020 }, { "grad_norm": 0.18260511755943298, "learning_rate": 7.374463842087094e-05, "loss": 0.0153, "step": 30030 }, { "grad_norm": 0.18910811841487885, "learning_rate": 7.372644732218254e-05, "loss": 0.0159, "step": 30040 }, { "grad_norm": 0.20544862747192383, "learning_rate": 7.370825216929322e-05, "loss": 0.016, "step": 30050 }, { "grad_norm": 0.20663560926914215, "learning_rate": 7.369005296531205e-05, "loss": 0.0132, "step": 30060 }, { "grad_norm": 0.17951428890228271, "learning_rate": 7.367184971334873e-05, "loss": 0.0134, "step": 30070 }, { "grad_norm": 0.171540305018425, "learning_rate": 7.365364241651371e-05, "loss": 0.0171, "step": 30080 }, { "grad_norm": 0.19733646512031555, "learning_rate": 7.363543107791815e-05, "loss": 0.0159, "step": 30090 }, { "grad_norm": 0.19676510989665985, "learning_rate": 7.361721570067384e-05, "loss": 0.0193, "step": 30100 }, { "grad_norm": 0.20671071112155914, "learning_rate": 7.359899628789331e-05, "loss": 0.0202, "step": 30110 }, { "grad_norm": 0.4272461235523224, "learning_rate": 7.358077284268974e-05, "loss": 0.0199, "step": 30120 }, { "grad_norm": 0.26575765013694763, "learning_rate": 7.356254536817702e-05, "loss": 0.0182, "step": 30130 }, { "grad_norm": 0.23973128199577332, "learning_rate": 7.354431386746973e-05, "loss": 0.0229, "step": 30140 }, { "grad_norm": 0.3179815709590912, "learning_rate": 7.352607834368316e-05, "loss": 0.0177, "step": 30150 }, { "grad_norm": 0.30998799204826355, "learning_rate": 7.350783879993324e-05, "loss": 0.0207, "step": 30160 }, { "grad_norm": 0.252277672290802, "learning_rate": 7.348959523933658e-05, "loss": 0.0212, "step": 30170 }, { "grad_norm": 0.3155480623245239, "learning_rate": 7.347134766501057e-05, "loss": 0.0213, "step": 30180 }, { "grad_norm": 0.2941843867301941, "learning_rate": 7.345309608007315e-05, "loss": 0.0191, "step": 30190 }, { "grad_norm": 0.21336102485656738, "learning_rate": 7.343484048764308e-05, "loss": 0.0165, "step": 30200 }, { "grad_norm": 0.2984659671783447, "learning_rate": 7.341658089083972e-05, "loss": 0.016, "step": 30210 }, { "grad_norm": 0.19611433148384094, "learning_rate": 7.339831729278313e-05, "loss": 0.02, "step": 30220 }, { "grad_norm": 0.18870456516742706, "learning_rate": 7.338004969659404e-05, "loss": 0.0173, "step": 30230 }, { "grad_norm": 0.19954250752925873, "learning_rate": 7.336177810539391e-05, "loss": 0.0196, "step": 30240 }, { "grad_norm": 0.19971375167369843, "learning_rate": 7.334350252230485e-05, "loss": 0.0174, "step": 30250 }, { "grad_norm": 0.2678895592689514, "learning_rate": 7.332522295044965e-05, "loss": 0.0194, "step": 30260 }, { "grad_norm": 0.22668053209781647, "learning_rate": 7.33069393929518e-05, "loss": 0.0187, "step": 30270 }, { "grad_norm": 0.17713695764541626, "learning_rate": 7.328865185293545e-05, "loss": 0.0151, "step": 30280 }, { "grad_norm": 0.30582043528556824, "learning_rate": 7.327036033352546e-05, "loss": 0.0154, "step": 30290 }, { "grad_norm": 0.2483338713645935, "learning_rate": 7.325206483784733e-05, "loss": 0.0166, "step": 30300 }, { "grad_norm": 0.17165017127990723, "learning_rate": 7.323376536902724e-05, "loss": 0.0154, "step": 30310 }, { "grad_norm": 0.19236646592617035, "learning_rate": 7.321546193019213e-05, "loss": 0.0142, "step": 30320 }, { "grad_norm": 0.21701334416866302, "learning_rate": 7.31971545244695e-05, "loss": 0.0163, "step": 30330 }, { "grad_norm": 0.17504744231700897, "learning_rate": 7.31788431549876e-05, "loss": 0.0135, "step": 30340 }, { "grad_norm": 0.1913650780916214, "learning_rate": 7.316052782487534e-05, "loss": 0.0156, "step": 30350 }, { "grad_norm": 0.19176585972309113, "learning_rate": 7.314220853726234e-05, "loss": 0.0173, "step": 30360 }, { "grad_norm": 0.17262984812259674, "learning_rate": 7.312388529527884e-05, "loss": 0.0175, "step": 30370 }, { "grad_norm": 0.24861690402030945, "learning_rate": 7.310555810205577e-05, "loss": 0.0173, "step": 30380 }, { "grad_norm": 0.15178097784519196, "learning_rate": 7.308722696072476e-05, "loss": 0.0178, "step": 30390 }, { "grad_norm": 0.16590100526809692, "learning_rate": 7.306889187441811e-05, "loss": 0.0137, "step": 30400 }, { "grad_norm": 0.24219569563865662, "learning_rate": 7.305055284626876e-05, "loss": 0.0166, "step": 30410 }, { "grad_norm": 0.2740147113800049, "learning_rate": 7.303220987941037e-05, "loss": 0.0185, "step": 30420 }, { "grad_norm": 0.18574562668800354, "learning_rate": 7.301386297697726e-05, "loss": 0.0145, "step": 30430 }, { "grad_norm": 0.20891322195529938, "learning_rate": 7.299551214210438e-05, "loss": 0.0166, "step": 30440 }, { "grad_norm": 0.20729795098304749, "learning_rate": 7.297715737792738e-05, "loss": 0.0145, "step": 30450 }, { "grad_norm": 0.17728082835674286, "learning_rate": 7.295879868758265e-05, "loss": 0.0161, "step": 30460 }, { "grad_norm": 0.1679074466228485, "learning_rate": 7.294043607420713e-05, "loss": 0.0155, "step": 30470 }, { "grad_norm": 0.27024516463279724, "learning_rate": 7.292206954093852e-05, "loss": 0.018, "step": 30480 }, { "grad_norm": 0.23118357360363007, "learning_rate": 7.290369909091515e-05, "loss": 0.015, "step": 30490 }, { "grad_norm": 0.17509512603282928, "learning_rate": 7.2885324727276e-05, "loss": 0.0125, "step": 30500 }, { "grad_norm": 0.1795777976512909, "learning_rate": 7.286694645316076e-05, "loss": 0.0149, "step": 30510 }, { "grad_norm": 0.17683464288711548, "learning_rate": 7.284856427170982e-05, "loss": 0.0139, "step": 30520 }, { "grad_norm": 0.2031281590461731, "learning_rate": 7.283017818606414e-05, "loss": 0.0185, "step": 30530 }, { "grad_norm": 0.19552303850650787, "learning_rate": 7.28117881993654e-05, "loss": 0.0137, "step": 30540 }, { "grad_norm": 0.3237195909023285, "learning_rate": 7.279339431475598e-05, "loss": 0.0176, "step": 30550 }, { "grad_norm": 0.19571152329444885, "learning_rate": 7.277499653537887e-05, "loss": 0.0133, "step": 30560 }, { "grad_norm": 0.18385231494903564, "learning_rate": 7.275659486437776e-05, "loss": 0.0159, "step": 30570 }, { "grad_norm": 0.216901496052742, "learning_rate": 7.273818930489695e-05, "loss": 0.0166, "step": 30580 }, { "grad_norm": 0.12176478654146194, "learning_rate": 7.271977986008151e-05, "loss": 0.0182, "step": 30590 }, { "grad_norm": 0.12612612545490265, "learning_rate": 7.270136653307705e-05, "loss": 0.0197, "step": 30600 }, { "grad_norm": 0.19748899340629578, "learning_rate": 7.268294932702994e-05, "loss": 0.0151, "step": 30610 }, { "grad_norm": 0.35599038004875183, "learning_rate": 7.266452824508719e-05, "loss": 0.0162, "step": 30620 }, { "grad_norm": 0.2461732029914856, "learning_rate": 7.264610329039643e-05, "loss": 0.0146, "step": 30630 }, { "grad_norm": 0.22221510112285614, "learning_rate": 7.262767446610599e-05, "loss": 0.012, "step": 30640 }, { "grad_norm": 0.16516929864883423, "learning_rate": 7.260924177536485e-05, "loss": 0.0141, "step": 30650 }, { "grad_norm": 0.22474272549152374, "learning_rate": 7.259080522132265e-05, "loss": 0.0168, "step": 30660 }, { "grad_norm": 0.27789708971977234, "learning_rate": 7.257236480712972e-05, "loss": 0.014, "step": 30670 }, { "grad_norm": 0.20590780675411224, "learning_rate": 7.255392053593697e-05, "loss": 0.018, "step": 30680 }, { "grad_norm": 0.210844948887825, "learning_rate": 7.253547241089607e-05, "loss": 0.0159, "step": 30690 }, { "grad_norm": 0.16071687638759613, "learning_rate": 7.251702043515927e-05, "loss": 0.0175, "step": 30700 }, { "grad_norm": 0.219329833984375, "learning_rate": 7.249856461187952e-05, "loss": 0.0198, "step": 30710 }, { "grad_norm": 0.35384872555732727, "learning_rate": 7.248010494421042e-05, "loss": 0.0202, "step": 30720 }, { "grad_norm": 0.19105148315429688, "learning_rate": 7.246164143530622e-05, "loss": 0.0197, "step": 30730 }, { "grad_norm": 0.2600192725658417, "learning_rate": 7.244317408832181e-05, "loss": 0.016, "step": 30740 }, { "grad_norm": 0.21219392120838165, "learning_rate": 7.242470290641279e-05, "loss": 0.0155, "step": 30750 }, { "grad_norm": 0.20028991997241974, "learning_rate": 7.240622789273536e-05, "loss": 0.0126, "step": 30760 }, { "grad_norm": 0.231490820646286, "learning_rate": 7.238774905044638e-05, "loss": 0.0189, "step": 30770 }, { "grad_norm": 0.11372452974319458, "learning_rate": 7.236926638270341e-05, "loss": 0.0137, "step": 30780 }, { "grad_norm": 0.16516467928886414, "learning_rate": 7.23507798926646e-05, "loss": 0.0175, "step": 30790 }, { "grad_norm": 0.1845504641532898, "learning_rate": 7.23322895834888e-05, "loss": 0.0138, "step": 30800 }, { "grad_norm": 0.16229036450386047, "learning_rate": 7.231379545833552e-05, "loss": 0.0124, "step": 30810 }, { "grad_norm": 0.19654397666454315, "learning_rate": 7.229529752036487e-05, "loss": 0.0126, "step": 30820 }, { "grad_norm": 0.18529652059078217, "learning_rate": 7.227679577273765e-05, "loss": 0.0153, "step": 30830 }, { "grad_norm": 0.12784399092197418, "learning_rate": 7.225829021861529e-05, "loss": 0.0173, "step": 30840 }, { "grad_norm": 0.23401415348052979, "learning_rate": 7.223978086115992e-05, "loss": 0.0131, "step": 30850 }, { "grad_norm": 0.1383824348449707, "learning_rate": 7.222126770353425e-05, "loss": 0.0133, "step": 30860 }, { "grad_norm": 0.20478416979312897, "learning_rate": 7.22027507489017e-05, "loss": 0.0194, "step": 30870 }, { "grad_norm": 0.20161625742912292, "learning_rate": 7.218423000042627e-05, "loss": 0.0145, "step": 30880 }, { "grad_norm": 0.17365975677967072, "learning_rate": 7.216570546127268e-05, "loss": 0.0155, "step": 30890 }, { "grad_norm": 0.19224920868873596, "learning_rate": 7.214717713460626e-05, "loss": 0.015, "step": 30900 }, { "grad_norm": 0.1838648021221161, "learning_rate": 7.2128645023593e-05, "loss": 0.0142, "step": 30910 }, { "grad_norm": 0.14288340508937836, "learning_rate": 7.211010913139951e-05, "loss": 0.0114, "step": 30920 }, { "grad_norm": 0.1557616889476776, "learning_rate": 7.209156946119308e-05, "loss": 0.0138, "step": 30930 }, { "grad_norm": 0.1847803294658661, "learning_rate": 7.207302601614166e-05, "loss": 0.0144, "step": 30940 }, { "grad_norm": 0.21859313547611237, "learning_rate": 7.205447879941378e-05, "loss": 0.0203, "step": 30950 }, { "grad_norm": 0.253130167722702, "learning_rate": 7.203592781417866e-05, "loss": 0.0185, "step": 30960 }, { "grad_norm": 0.230371356010437, "learning_rate": 7.201737306360617e-05, "loss": 0.0155, "step": 30970 }, { "grad_norm": 0.2363506555557251, "learning_rate": 7.19988145508668e-05, "loss": 0.0171, "step": 30980 }, { "grad_norm": 0.17850792407989502, "learning_rate": 7.198025227913168e-05, "loss": 0.0177, "step": 30990 }, { "grad_norm": 0.14867395162582397, "learning_rate": 7.196168625157261e-05, "loss": 0.0157, "step": 31000 }, { "grad_norm": 0.23479434847831726, "learning_rate": 7.194311647136201e-05, "loss": 0.015, "step": 31010 }, { "grad_norm": 0.19445711374282837, "learning_rate": 7.192454294167297e-05, "loss": 0.0186, "step": 31020 }, { "grad_norm": 0.2255420982837677, "learning_rate": 7.190596566567917e-05, "loss": 0.0181, "step": 31030 }, { "grad_norm": 0.21820561587810516, "learning_rate": 7.188738464655496e-05, "loss": 0.0166, "step": 31040 }, { "grad_norm": 0.18153417110443115, "learning_rate": 7.186879988747533e-05, "loss": 0.0141, "step": 31050 }, { "grad_norm": 0.21243776381015778, "learning_rate": 7.185021139161592e-05, "loss": 0.0158, "step": 31060 }, { "grad_norm": 0.18130917847156525, "learning_rate": 7.1831619162153e-05, "loss": 0.0183, "step": 31070 }, { "grad_norm": 0.18880543112754822, "learning_rate": 7.181302320226345e-05, "loss": 0.0147, "step": 31080 }, { "grad_norm": 0.17704199254512787, "learning_rate": 7.179442351512482e-05, "loss": 0.0153, "step": 31090 }, { "grad_norm": 0.22649747133255005, "learning_rate": 7.177582010391528e-05, "loss": 0.015, "step": 31100 }, { "grad_norm": 0.2397458404302597, "learning_rate": 7.175721297181366e-05, "loss": 0.0214, "step": 31110 }, { "grad_norm": 0.2484322488307953, "learning_rate": 7.173860212199942e-05, "loss": 0.0142, "step": 31120 }, { "grad_norm": 0.16655510663986206, "learning_rate": 7.171998755765263e-05, "loss": 0.0168, "step": 31130 }, { "grad_norm": 0.23997490108013153, "learning_rate": 7.170136928195398e-05, "loss": 0.0178, "step": 31140 }, { "grad_norm": 0.16947337985038757, "learning_rate": 7.168274729808489e-05, "loss": 0.0195, "step": 31150 }, { "grad_norm": 0.1995472013950348, "learning_rate": 7.166412160922728e-05, "loss": 0.0154, "step": 31160 }, { "grad_norm": 0.2511410713195801, "learning_rate": 7.164549221856382e-05, "loss": 0.012, "step": 31170 }, { "grad_norm": 0.22391772270202637, "learning_rate": 7.162685912927775e-05, "loss": 0.0143, "step": 31180 }, { "grad_norm": 0.1512426733970642, "learning_rate": 7.160822234455294e-05, "loss": 0.0145, "step": 31190 }, { "grad_norm": 0.22132514417171478, "learning_rate": 7.158958186757391e-05, "loss": 0.0147, "step": 31200 }, { "grad_norm": 0.19412200152873993, "learning_rate": 7.157093770152582e-05, "loss": 0.0132, "step": 31210 }, { "grad_norm": 0.18089406192302704, "learning_rate": 7.155228984959446e-05, "loss": 0.0159, "step": 31220 }, { "grad_norm": 0.17697088420391083, "learning_rate": 7.153363831496621e-05, "loss": 0.0117, "step": 31230 }, { "grad_norm": 0.27514973282814026, "learning_rate": 7.151498310082811e-05, "loss": 0.0184, "step": 31240 }, { "grad_norm": 0.18882083892822266, "learning_rate": 7.149632421036784e-05, "loss": 0.0135, "step": 31250 }, { "grad_norm": 0.17364206910133362, "learning_rate": 7.147766164677369e-05, "loss": 0.0145, "step": 31260 }, { "grad_norm": 0.18633945286273956, "learning_rate": 7.145899541323459e-05, "loss": 0.0142, "step": 31270 }, { "grad_norm": 0.22497650980949402, "learning_rate": 7.144032551294007e-05, "loss": 0.0158, "step": 31280 }, { "grad_norm": 0.3316846489906311, "learning_rate": 7.14216519490803e-05, "loss": 0.02, "step": 31290 }, { "grad_norm": 0.20623567700386047, "learning_rate": 7.140297472484609e-05, "loss": 0.0169, "step": 31300 }, { "grad_norm": 0.23369596898555756, "learning_rate": 7.138429384342891e-05, "loss": 0.0137, "step": 31310 }, { "grad_norm": 0.22290104627609253, "learning_rate": 7.136560930802074e-05, "loss": 0.0174, "step": 31320 }, { "grad_norm": 0.17307725548744202, "learning_rate": 7.134692112181431e-05, "loss": 0.0151, "step": 31330 }, { "grad_norm": 0.22154586017131805, "learning_rate": 7.13282292880029e-05, "loss": 0.0145, "step": 31340 }, { "grad_norm": 0.23497851192951202, "learning_rate": 7.130953380978043e-05, "loss": 0.0144, "step": 31350 }, { "grad_norm": 0.18395984172821045, "learning_rate": 7.129083469034144e-05, "loss": 0.0148, "step": 31360 }, { "grad_norm": 0.20909015834331512, "learning_rate": 7.127213193288112e-05, "loss": 0.0167, "step": 31370 }, { "grad_norm": 0.14583559334278107, "learning_rate": 7.125342554059522e-05, "loss": 0.014, "step": 31380 }, { "grad_norm": 0.20696336030960083, "learning_rate": 7.12347155166802e-05, "loss": 0.0185, "step": 31390 }, { "grad_norm": 0.18601706624031067, "learning_rate": 7.121600186433306e-05, "loss": 0.0177, "step": 31400 }, { "grad_norm": 0.22671526670455933, "learning_rate": 7.119728458675148e-05, "loss": 0.0204, "step": 31410 }, { "grad_norm": 0.19202694296836853, "learning_rate": 7.117856368713369e-05, "loss": 0.0146, "step": 31420 }, { "grad_norm": 0.1840973049402237, "learning_rate": 7.115983916867861e-05, "loss": 0.0148, "step": 31430 }, { "grad_norm": 0.1885983943939209, "learning_rate": 7.114111103458574e-05, "loss": 0.0135, "step": 31440 }, { "grad_norm": 0.22112515568733215, "learning_rate": 7.11223792880552e-05, "loss": 0.0176, "step": 31450 }, { "grad_norm": 0.17219115793704987, "learning_rate": 7.110364393228773e-05, "loss": 0.0141, "step": 31460 }, { "grad_norm": 0.1892634928226471, "learning_rate": 7.108490497048471e-05, "loss": 0.0158, "step": 31470 }, { "grad_norm": 0.249062642455101, "learning_rate": 7.10661624058481e-05, "loss": 0.0156, "step": 31480 }, { "grad_norm": 0.27338653802871704, "learning_rate": 7.10474162415805e-05, "loss": 0.0164, "step": 31490 }, { "grad_norm": 0.21250000596046448, "learning_rate": 7.102866648088511e-05, "loss": 0.0193, "step": 31500 }, { "grad_norm": 0.17284634709358215, "learning_rate": 7.100991312696576e-05, "loss": 0.0159, "step": 31510 }, { "grad_norm": 0.18662264943122864, "learning_rate": 7.099115618302686e-05, "loss": 0.0114, "step": 31520 }, { "grad_norm": 0.1783761829137802, "learning_rate": 7.097239565227349e-05, "loss": 0.0153, "step": 31530 }, { "grad_norm": 0.20713087916374207, "learning_rate": 7.09536315379113e-05, "loss": 0.017, "step": 31540 }, { "grad_norm": 0.23775455355644226, "learning_rate": 7.093486384314656e-05, "loss": 0.017, "step": 31550 }, { "grad_norm": 0.2301507592201233, "learning_rate": 7.091609257118616e-05, "loss": 0.0152, "step": 31560 }, { "grad_norm": 0.20189633965492249, "learning_rate": 7.08973177252376e-05, "loss": 0.0207, "step": 31570 }, { "grad_norm": 0.20299343764781952, "learning_rate": 7.087853930850898e-05, "loss": 0.0149, "step": 31580 }, { "grad_norm": 0.1773572564125061, "learning_rate": 7.085975732420903e-05, "loss": 0.014, "step": 31590 }, { "grad_norm": 0.2622652053833008, "learning_rate": 7.084097177554706e-05, "loss": 0.0168, "step": 31600 }, { "grad_norm": 0.25973081588745117, "learning_rate": 7.082218266573301e-05, "loss": 0.0174, "step": 31610 }, { "grad_norm": 0.24823865294456482, "learning_rate": 7.080338999797743e-05, "loss": 0.0146, "step": 31620 }, { "grad_norm": 0.17322294414043427, "learning_rate": 7.07845937754915e-05, "loss": 0.0146, "step": 31630 }, { "grad_norm": 0.21560138463974, "learning_rate": 7.076579400148693e-05, "loss": 0.0194, "step": 31640 }, { "grad_norm": 0.18667848408222198, "learning_rate": 7.074699067917611e-05, "loss": 0.013, "step": 31650 }, { "grad_norm": 0.1867327094078064, "learning_rate": 7.072818381177201e-05, "loss": 0.0175, "step": 31660 }, { "grad_norm": 0.2595457434654236, "learning_rate": 7.070937340248823e-05, "loss": 0.021, "step": 31670 }, { "grad_norm": 0.1867426335811615, "learning_rate": 7.069055945453893e-05, "loss": 0.0142, "step": 31680 }, { "grad_norm": 0.15793773531913757, "learning_rate": 7.067174197113892e-05, "loss": 0.0175, "step": 31690 }, { "grad_norm": 0.15416640043258667, "learning_rate": 7.065292095550355e-05, "loss": 0.0146, "step": 31700 }, { "grad_norm": 0.20821715891361237, "learning_rate": 7.063409641084887e-05, "loss": 0.0181, "step": 31710 }, { "grad_norm": 0.28268688917160034, "learning_rate": 7.061526834039145e-05, "loss": 0.0231, "step": 31720 }, { "grad_norm": 0.24201062321662903, "learning_rate": 7.05964367473485e-05, "loss": 0.0153, "step": 31730 }, { "grad_norm": 0.24673700332641602, "learning_rate": 7.057760163493783e-05, "loss": 0.0165, "step": 31740 }, { "grad_norm": 0.24250547587871552, "learning_rate": 7.055876300637783e-05, "loss": 0.018, "step": 31750 }, { "grad_norm": 0.16682732105255127, "learning_rate": 7.053992086488753e-05, "loss": 0.0176, "step": 31760 }, { "grad_norm": 0.2267983853816986, "learning_rate": 7.052107521368651e-05, "loss": 0.0152, "step": 31770 }, { "grad_norm": 0.25612667202949524, "learning_rate": 7.0502226055995e-05, "loss": 0.0202, "step": 31780 }, { "grad_norm": 0.22113673388957977, "learning_rate": 7.048337339503379e-05, "loss": 0.0135, "step": 31790 }, { "grad_norm": 0.3202968239784241, "learning_rate": 7.046451723402427e-05, "loss": 0.0177, "step": 31800 }, { "grad_norm": 0.15291759371757507, "learning_rate": 7.044565757618848e-05, "loss": 0.0145, "step": 31810 }, { "grad_norm": 0.20816293358802795, "learning_rate": 7.042679442474899e-05, "loss": 0.0159, "step": 31820 }, { "grad_norm": 0.269096702337265, "learning_rate": 7.040792778292902e-05, "loss": 0.0139, "step": 31830 }, { "grad_norm": 0.2096596211194992, "learning_rate": 7.038905765395234e-05, "loss": 0.0154, "step": 31840 }, { "grad_norm": 0.15657606720924377, "learning_rate": 7.037018404104334e-05, "loss": 0.0157, "step": 31850 }, { "grad_norm": 0.17519184947013855, "learning_rate": 7.035130694742702e-05, "loss": 0.0155, "step": 31860 }, { "grad_norm": 0.19085581600666046, "learning_rate": 7.033242637632897e-05, "loss": 0.0169, "step": 31870 }, { "grad_norm": 0.2643909156322479, "learning_rate": 7.031354233097534e-05, "loss": 0.0144, "step": 31880 }, { "grad_norm": 0.15002045035362244, "learning_rate": 7.029465481459289e-05, "loss": 0.0152, "step": 31890 }, { "grad_norm": 0.18768824636936188, "learning_rate": 7.027576383040898e-05, "loss": 0.0194, "step": 31900 }, { "grad_norm": 0.2128220945596695, "learning_rate": 7.025686938165159e-05, "loss": 0.0143, "step": 31910 }, { "grad_norm": 0.2146807610988617, "learning_rate": 7.023797147154924e-05, "loss": 0.0145, "step": 31920 }, { "grad_norm": 0.24301697313785553, "learning_rate": 7.021907010333111e-05, "loss": 0.0133, "step": 31930 }, { "grad_norm": 0.24496467411518097, "learning_rate": 7.020016528022685e-05, "loss": 0.0185, "step": 31940 }, { "grad_norm": 0.2701740264892578, "learning_rate": 7.018125700546683e-05, "loss": 0.0148, "step": 31950 }, { "grad_norm": 0.3105948865413666, "learning_rate": 7.016234528228196e-05, "loss": 0.0175, "step": 31960 }, { "grad_norm": 0.1652555614709854, "learning_rate": 7.014343011390372e-05, "loss": 0.0155, "step": 31970 }, { "grad_norm": 0.2714078426361084, "learning_rate": 7.01245115035642e-05, "loss": 0.0185, "step": 31980 }, { "grad_norm": 0.15535692870616913, "learning_rate": 7.010558945449606e-05, "loss": 0.014, "step": 31990 }, { "grad_norm": 0.21817366778850555, "learning_rate": 7.008666396993258e-05, "loss": 0.0201, "step": 32000 }, { "grad_norm": 0.26270991563796997, "learning_rate": 7.006773505310759e-05, "loss": 0.0134, "step": 32010 }, { "grad_norm": 0.2266518771648407, "learning_rate": 7.004880270725553e-05, "loss": 0.0141, "step": 32020 }, { "grad_norm": 0.17934384942054749, "learning_rate": 7.002986693561144e-05, "loss": 0.0166, "step": 32030 }, { "grad_norm": 0.23692895472049713, "learning_rate": 7.001092774141089e-05, "loss": 0.0161, "step": 32040 }, { "grad_norm": 0.21629638969898224, "learning_rate": 6.999198512789009e-05, "loss": 0.0157, "step": 32050 }, { "grad_norm": 0.25084903836250305, "learning_rate": 6.997303909828584e-05, "loss": 0.0147, "step": 32060 }, { "grad_norm": 0.18708139657974243, "learning_rate": 6.995408965583544e-05, "loss": 0.0139, "step": 32070 }, { "grad_norm": 0.1319880187511444, "learning_rate": 6.993513680377688e-05, "loss": 0.0175, "step": 32080 }, { "grad_norm": 0.24881304800510406, "learning_rate": 6.991618054534868e-05, "loss": 0.0165, "step": 32090 }, { "grad_norm": 0.23619584739208221, "learning_rate": 6.989722088378991e-05, "loss": 0.0148, "step": 32100 }, { "grad_norm": 0.20982608199119568, "learning_rate": 6.987825782234027e-05, "loss": 0.0171, "step": 32110 }, { "grad_norm": 0.2640596926212311, "learning_rate": 6.985929136424006e-05, "loss": 0.0194, "step": 32120 }, { "grad_norm": 0.2372002899646759, "learning_rate": 6.984032151273012e-05, "loss": 0.017, "step": 32130 }, { "grad_norm": 0.23283818364143372, "learning_rate": 6.982134827105186e-05, "loss": 0.0173, "step": 32140 }, { "grad_norm": 0.2083454132080078, "learning_rate": 6.980237164244729e-05, "loss": 0.0161, "step": 32150 }, { "grad_norm": 0.20987024903297424, "learning_rate": 6.9783391630159e-05, "loss": 0.0194, "step": 32160 }, { "grad_norm": 0.18890812993049622, "learning_rate": 6.976440823743015e-05, "loss": 0.0203, "step": 32170 }, { "grad_norm": 0.3619077205657959, "learning_rate": 6.974542146750451e-05, "loss": 0.0205, "step": 32180 }, { "grad_norm": 0.31521090865135193, "learning_rate": 6.972643132362637e-05, "loss": 0.0152, "step": 32190 }, { "grad_norm": 0.2689976692199707, "learning_rate": 6.970743780904064e-05, "loss": 0.0147, "step": 32200 }, { "grad_norm": 0.28120365738868713, "learning_rate": 6.968844092699277e-05, "loss": 0.0157, "step": 32210 }, { "grad_norm": 0.18782000243663788, "learning_rate": 6.966944068072883e-05, "loss": 0.0127, "step": 32220 }, { "grad_norm": 0.19546742737293243, "learning_rate": 6.965043707349545e-05, "loss": 0.0145, "step": 32230 }, { "grad_norm": 0.18235884606838226, "learning_rate": 6.963143010853982e-05, "loss": 0.0142, "step": 32240 }, { "grad_norm": 0.1885063797235489, "learning_rate": 6.961241978910971e-05, "loss": 0.0162, "step": 32250 }, { "grad_norm": 0.21519477665424347, "learning_rate": 6.959340611845344e-05, "loss": 0.0136, "step": 32260 }, { "grad_norm": 0.2720993459224701, "learning_rate": 6.957438909981995e-05, "loss": 0.0191, "step": 32270 }, { "grad_norm": 0.22964197397232056, "learning_rate": 6.955536873645872e-05, "loss": 0.021, "step": 32280 }, { "grad_norm": 0.16769948601722717, "learning_rate": 6.953634503161982e-05, "loss": 0.019, "step": 32290 }, { "grad_norm": 0.1992342621088028, "learning_rate": 6.951731798855387e-05, "loss": 0.0157, "step": 32300 }, { "grad_norm": 0.18061858415603638, "learning_rate": 6.949828761051208e-05, "loss": 0.0146, "step": 32310 }, { "grad_norm": 0.15002918243408203, "learning_rate": 6.947925390074622e-05, "loss": 0.0122, "step": 32320 }, { "grad_norm": 0.15954114496707916, "learning_rate": 6.946021686250863e-05, "loss": 0.0132, "step": 32330 }, { "grad_norm": 0.201379656791687, "learning_rate": 6.94411764990522e-05, "loss": 0.0159, "step": 32340 }, { "grad_norm": 0.4168228507041931, "learning_rate": 6.942213281363044e-05, "loss": 0.013, "step": 32350 }, { "grad_norm": 0.20123109221458435, "learning_rate": 6.940308580949737e-05, "loss": 0.0157, "step": 32360 }, { "grad_norm": 0.23779934644699097, "learning_rate": 6.93840354899076e-05, "loss": 0.0171, "step": 32370 }, { "grad_norm": 0.24636884033679962, "learning_rate": 6.936498185811633e-05, "loss": 0.0152, "step": 32380 }, { "grad_norm": 0.19369351863861084, "learning_rate": 6.93459249173793e-05, "loss": 0.0184, "step": 32390 }, { "grad_norm": 0.21464477479457855, "learning_rate": 6.932686467095279e-05, "loss": 0.0186, "step": 32400 }, { "grad_norm": 0.2206917256116867, "learning_rate": 6.930780112209373e-05, "loss": 0.0153, "step": 32410 }, { "grad_norm": 0.17209112644195557, "learning_rate": 6.92887342740595e-05, "loss": 0.0138, "step": 32420 }, { "grad_norm": 0.1519872546195984, "learning_rate": 6.926966413010816e-05, "loss": 0.0182, "step": 32430 }, { "grad_norm": 0.2263024002313614, "learning_rate": 6.925059069349824e-05, "loss": 0.0174, "step": 32440 }, { "grad_norm": 0.16222304105758667, "learning_rate": 6.923151396748886e-05, "loss": 0.0167, "step": 32450 }, { "grad_norm": 0.20281711220741272, "learning_rate": 6.921243395533974e-05, "loss": 0.0178, "step": 32460 }, { "grad_norm": 0.20760519802570343, "learning_rate": 6.919335066031109e-05, "loss": 0.0139, "step": 32470 }, { "grad_norm": 0.2626643478870392, "learning_rate": 6.917426408566379e-05, "loss": 0.0126, "step": 32480 }, { "grad_norm": 0.304387629032135, "learning_rate": 6.915517423465916e-05, "loss": 0.0177, "step": 32490 }, { "grad_norm": 0.22522248327732086, "learning_rate": 6.913608111055914e-05, "loss": 0.0174, "step": 32500 }, { "grad_norm": 0.2772827446460724, "learning_rate": 6.911698471662623e-05, "loss": 0.0189, "step": 32510 }, { "grad_norm": 0.1790817528963089, "learning_rate": 6.90978850561235e-05, "loss": 0.0153, "step": 32520 }, { "grad_norm": 0.210267573595047, "learning_rate": 6.907878213231454e-05, "loss": 0.0147, "step": 32530 }, { "grad_norm": 0.19324591755867004, "learning_rate": 6.90596759484635e-05, "loss": 0.0162, "step": 32540 }, { "grad_norm": 0.23948903381824493, "learning_rate": 6.904056650783514e-05, "loss": 0.0167, "step": 32550 }, { "grad_norm": 0.1793159395456314, "learning_rate": 6.902145381369471e-05, "loss": 0.0151, "step": 32560 }, { "grad_norm": 0.1739877462387085, "learning_rate": 6.900233786930808e-05, "loss": 0.0176, "step": 32570 }, { "grad_norm": 0.2667791247367859, "learning_rate": 6.898321867794161e-05, "loss": 0.024, "step": 32580 }, { "grad_norm": 0.23187638819217682, "learning_rate": 6.896409624286226e-05, "loss": 0.0173, "step": 32590 }, { "grad_norm": 0.32169970870018005, "learning_rate": 6.894497056733754e-05, "loss": 0.0211, "step": 32600 }, { "grad_norm": 0.24314503371715546, "learning_rate": 6.89258416546355e-05, "loss": 0.0161, "step": 32610 }, { "grad_norm": 0.17203815281391144, "learning_rate": 6.890670950802474e-05, "loss": 0.0169, "step": 32620 }, { "grad_norm": 0.22191686928272247, "learning_rate": 6.88875741307744e-05, "loss": 0.0157, "step": 32630 }, { "grad_norm": 0.2240646779537201, "learning_rate": 6.886843552615425e-05, "loss": 0.0165, "step": 32640 }, { "grad_norm": 0.2150510847568512, "learning_rate": 6.884929369743451e-05, "loss": 0.0169, "step": 32650 }, { "grad_norm": 0.1444389671087265, "learning_rate": 6.8830148647886e-05, "loss": 0.0132, "step": 32660 }, { "grad_norm": 0.24571523070335388, "learning_rate": 6.88110003807801e-05, "loss": 0.0166, "step": 32670 }, { "grad_norm": 0.21593712270259857, "learning_rate": 6.87918488993887e-05, "loss": 0.0161, "step": 32680 }, { "grad_norm": 0.17390486598014832, "learning_rate": 6.877269420698431e-05, "loss": 0.0134, "step": 32690 }, { "grad_norm": 0.2951880991458893, "learning_rate": 6.875353630683989e-05, "loss": 0.015, "step": 32700 }, { "grad_norm": 0.36618363857269287, "learning_rate": 6.873437520222905e-05, "loss": 0.0162, "step": 32710 }, { "grad_norm": 0.2661437690258026, "learning_rate": 6.871521089642585e-05, "loss": 0.0158, "step": 32720 }, { "grad_norm": 0.2348736673593521, "learning_rate": 6.869604339270498e-05, "loss": 0.0125, "step": 32730 }, { "grad_norm": 0.2717404365539551, "learning_rate": 6.867687269434164e-05, "loss": 0.014, "step": 32740 }, { "grad_norm": 0.14561498165130615, "learning_rate": 6.865769880461156e-05, "loss": 0.0126, "step": 32750 }, { "grad_norm": 0.3133457601070404, "learning_rate": 6.863852172679104e-05, "loss": 0.0142, "step": 32760 }, { "grad_norm": 0.2643318474292755, "learning_rate": 6.861934146415693e-05, "loss": 0.0184, "step": 32770 }, { "grad_norm": 0.23616540431976318, "learning_rate": 6.86001580199866e-05, "loss": 0.0125, "step": 32780 }, { "grad_norm": 0.12971867620944977, "learning_rate": 6.858097139755798e-05, "loss": 0.0146, "step": 32790 }, { "grad_norm": 0.1891407072544098, "learning_rate": 6.856178160014955e-05, "loss": 0.0144, "step": 32800 }, { "grad_norm": 0.1948440819978714, "learning_rate": 6.85425886310403e-05, "loss": 0.0152, "step": 32810 }, { "grad_norm": 0.20145153999328613, "learning_rate": 6.852339249350979e-05, "loss": 0.0122, "step": 32820 }, { "grad_norm": 0.1394258737564087, "learning_rate": 6.850419319083812e-05, "loss": 0.0124, "step": 32830 }, { "grad_norm": 0.22020851075649261, "learning_rate": 6.848499072630592e-05, "loss": 0.0124, "step": 32840 }, { "grad_norm": 0.22519157826900482, "learning_rate": 6.846578510319439e-05, "loss": 0.0157, "step": 32850 }, { "grad_norm": 0.17877082526683807, "learning_rate": 6.844657632478519e-05, "loss": 0.0145, "step": 32860 }, { "grad_norm": 0.19132137298583984, "learning_rate": 6.842736439436063e-05, "loss": 0.0147, "step": 32870 }, { "grad_norm": 0.23985810577869415, "learning_rate": 6.84081493152035e-05, "loss": 0.0154, "step": 32880 }, { "grad_norm": 0.9499965906143188, "learning_rate": 6.83889310905971e-05, "loss": 0.0157, "step": 32890 }, { "grad_norm": 0.3855004906654358, "learning_rate": 6.836970972382533e-05, "loss": 0.0194, "step": 32900 }, { "grad_norm": 0.40539249777793884, "learning_rate": 6.835048521817257e-05, "loss": 0.0151, "step": 32910 }, { "grad_norm": 1.2112963199615479, "learning_rate": 6.833125757692379e-05, "loss": 0.0171, "step": 32920 }, { "grad_norm": 0.24030320346355438, "learning_rate": 6.831202680336441e-05, "loss": 0.0151, "step": 32930 }, { "grad_norm": 0.1666414439678192, "learning_rate": 6.829279290078052e-05, "loss": 0.0169, "step": 32940 }, { "grad_norm": 0.1767226755619049, "learning_rate": 6.827355587245863e-05, "loss": 0.0191, "step": 32950 }, { "grad_norm": 0.19351615011692047, "learning_rate": 6.82543157216858e-05, "loss": 0.0161, "step": 32960 }, { "grad_norm": 0.19184187054634094, "learning_rate": 6.823507245174969e-05, "loss": 0.0179, "step": 32970 }, { "grad_norm": 0.23755571246147156, "learning_rate": 6.821582606593841e-05, "loss": 0.0187, "step": 32980 }, { "grad_norm": 0.3077227771282196, "learning_rate": 6.81965765675407e-05, "loss": 0.0166, "step": 32990 }, { "grad_norm": 0.2552628219127655, "learning_rate": 6.81773239598457e-05, "loss": 0.0127, "step": 33000 }, { "grad_norm": 0.2024366408586502, "learning_rate": 6.815806824614319e-05, "loss": 0.0174, "step": 33010 }, { "grad_norm": 0.17838267982006073, "learning_rate": 6.813880942972343e-05, "loss": 0.0143, "step": 33020 }, { "grad_norm": 0.1943020224571228, "learning_rate": 6.811954751387726e-05, "loss": 0.0143, "step": 33030 }, { "grad_norm": 0.21114060282707214, "learning_rate": 6.810028250189598e-05, "loss": 0.0176, "step": 33040 }, { "grad_norm": 0.2362552285194397, "learning_rate": 6.808101439707147e-05, "loss": 0.0146, "step": 33050 }, { "grad_norm": 0.23787108063697815, "learning_rate": 6.806174320269609e-05, "loss": 0.0175, "step": 33060 }, { "grad_norm": 0.1753418892621994, "learning_rate": 6.804246892206281e-05, "loss": 0.0163, "step": 33070 }, { "grad_norm": 0.22706548869609833, "learning_rate": 6.802319155846506e-05, "loss": 0.0161, "step": 33080 }, { "grad_norm": 0.1814766526222229, "learning_rate": 6.800391111519679e-05, "loss": 0.0125, "step": 33090 }, { "grad_norm": 0.18070799112319946, "learning_rate": 6.798462759555253e-05, "loss": 0.0152, "step": 33100 }, { "grad_norm": 0.23632073402404785, "learning_rate": 6.79653410028273e-05, "loss": 0.0151, "step": 33110 }, { "grad_norm": 0.25100091099739075, "learning_rate": 6.794605134031663e-05, "loss": 0.0171, "step": 33120 }, { "grad_norm": 0.16080579161643982, "learning_rate": 6.792675861131661e-05, "loss": 0.0152, "step": 33130 }, { "grad_norm": 0.1645512878894806, "learning_rate": 6.790746281912386e-05, "loss": 0.0168, "step": 33140 }, { "grad_norm": 0.22310176491737366, "learning_rate": 6.788816396703546e-05, "loss": 0.0169, "step": 33150 }, { "grad_norm": 0.17164365947246552, "learning_rate": 6.78688620583491e-05, "loss": 0.0117, "step": 33160 }, { "grad_norm": 0.17006520926952362, "learning_rate": 6.784955709636292e-05, "loss": 0.0176, "step": 33170 }, { "grad_norm": 0.16751591861248016, "learning_rate": 6.783024908437564e-05, "loss": 0.014, "step": 33180 }, { "grad_norm": 0.30053555965423584, "learning_rate": 6.781093802568641e-05, "loss": 0.0188, "step": 33190 }, { "grad_norm": 0.18298207223415375, "learning_rate": 6.779162392359504e-05, "loss": 0.0171, "step": 33200 }, { "grad_norm": 0.18773433566093445, "learning_rate": 6.777230678140172e-05, "loss": 0.0144, "step": 33210 }, { "grad_norm": 0.154659703373909, "learning_rate": 6.775298660240726e-05, "loss": 0.0188, "step": 33220 }, { "grad_norm": 0.16828082501888275, "learning_rate": 6.773366338991292e-05, "loss": 0.0141, "step": 33230 }, { "grad_norm": 0.14031469821929932, "learning_rate": 6.771433714722052e-05, "loss": 0.0128, "step": 33240 }, { "grad_norm": 0.2147185504436493, "learning_rate": 6.769500787763239e-05, "loss": 0.013, "step": 33250 }, { "grad_norm": 0.22807763516902924, "learning_rate": 6.76756755844514e-05, "loss": 0.0177, "step": 33260 }, { "grad_norm": 0.23213137686252594, "learning_rate": 6.765634027098087e-05, "loss": 0.0206, "step": 33270 }, { "grad_norm": 0.3018459975719452, "learning_rate": 6.763700194052468e-05, "loss": 0.0184, "step": 33280 }, { "grad_norm": 0.19254232943058014, "learning_rate": 6.761766059638723e-05, "loss": 0.0122, "step": 33290 }, { "grad_norm": 0.2026410549879074, "learning_rate": 6.759831624187345e-05, "loss": 0.0175, "step": 33300 }, { "grad_norm": 0.2190864235162735, "learning_rate": 6.757896888028871e-05, "loss": 0.0139, "step": 33310 }, { "grad_norm": 0.19972272217273712, "learning_rate": 6.7559618514939e-05, "loss": 0.0159, "step": 33320 }, { "grad_norm": 0.3354247212409973, "learning_rate": 6.754026514913073e-05, "loss": 0.0164, "step": 33330 }, { "grad_norm": 0.2237810343503952, "learning_rate": 6.752090878617087e-05, "loss": 0.0147, "step": 33340 }, { "grad_norm": 0.21851584315299988, "learning_rate": 6.75015494293669e-05, "loss": 0.0148, "step": 33350 }, { "grad_norm": 0.15349233150482178, "learning_rate": 6.74821870820268e-05, "loss": 0.0164, "step": 33360 }, { "grad_norm": 0.29377174377441406, "learning_rate": 6.746282174745907e-05, "loss": 0.0213, "step": 33370 }, { "grad_norm": 0.16280782222747803, "learning_rate": 6.744345342897271e-05, "loss": 0.0149, "step": 33380 }, { "grad_norm": 0.22999581694602966, "learning_rate": 6.742408212987724e-05, "loss": 0.0165, "step": 33390 }, { "grad_norm": 0.18908628821372986, "learning_rate": 6.740470785348269e-05, "loss": 0.0144, "step": 33400 }, { "grad_norm": 0.19398139417171478, "learning_rate": 6.738533060309958e-05, "loss": 0.0142, "step": 33410 }, { "grad_norm": 0.17511001229286194, "learning_rate": 6.736595038203894e-05, "loss": 0.0137, "step": 33420 }, { "grad_norm": 0.25392216444015503, "learning_rate": 6.734656719361236e-05, "loss": 0.0148, "step": 33430 }, { "grad_norm": 0.1929674744606018, "learning_rate": 6.732718104113189e-05, "loss": 0.0164, "step": 33440 }, { "grad_norm": 0.22013399004936218, "learning_rate": 6.730779192791006e-05, "loss": 0.0119, "step": 33450 }, { "grad_norm": 0.32883507013320923, "learning_rate": 6.728839985725997e-05, "loss": 0.0151, "step": 33460 }, { "grad_norm": 0.16023699939250946, "learning_rate": 6.726900483249517e-05, "loss": 0.0128, "step": 33470 }, { "grad_norm": 0.1440904587507248, "learning_rate": 6.724960685692976e-05, "loss": 0.0128, "step": 33480 }, { "grad_norm": 0.15783537924289703, "learning_rate": 6.723020593387833e-05, "loss": 0.0134, "step": 33490 }, { "grad_norm": 0.2546050548553467, "learning_rate": 6.721080206665593e-05, "loss": 0.0177, "step": 33500 }, { "grad_norm": 0.21709905564785004, "learning_rate": 6.719139525857819e-05, "loss": 0.0149, "step": 33510 }, { "grad_norm": 0.35308560729026794, "learning_rate": 6.717198551296117e-05, "loss": 0.0178, "step": 33520 }, { "grad_norm": 0.1883116364479065, "learning_rate": 6.715257283312148e-05, "loss": 0.0158, "step": 33530 }, { "grad_norm": 0.2203894406557083, "learning_rate": 6.713315722237623e-05, "loss": 0.015, "step": 33540 }, { "grad_norm": 0.2610926628112793, "learning_rate": 6.7113738684043e-05, "loss": 0.0164, "step": 33550 }, { "grad_norm": 0.27881357073783875, "learning_rate": 6.709431722143989e-05, "loss": 0.0145, "step": 33560 }, { "grad_norm": 0.13623139262199402, "learning_rate": 6.70748928378855e-05, "loss": 0.0139, "step": 33570 }, { "grad_norm": 0.24442045390605927, "learning_rate": 6.705546553669891e-05, "loss": 0.0185, "step": 33580 }, { "grad_norm": 0.21273252367973328, "learning_rate": 6.703603532119974e-05, "loss": 0.0169, "step": 33590 }, { "grad_norm": 0.20990760624408722, "learning_rate": 6.701660219470808e-05, "loss": 0.0166, "step": 33600 }, { "grad_norm": 0.16790610551834106, "learning_rate": 6.69971661605445e-05, "loss": 0.0154, "step": 33610 }, { "grad_norm": 0.14393827319145203, "learning_rate": 6.697772722203008e-05, "loss": 0.0169, "step": 33620 }, { "grad_norm": 0.1369474083185196, "learning_rate": 6.695828538248643e-05, "loss": 0.0129, "step": 33630 }, { "grad_norm": 0.1714072823524475, "learning_rate": 6.693884064523563e-05, "loss": 0.0177, "step": 33640 }, { "grad_norm": 0.17966711521148682, "learning_rate": 6.691939301360023e-05, "loss": 0.0211, "step": 33650 }, { "grad_norm": 0.24842599034309387, "learning_rate": 6.689994249090333e-05, "loss": 0.0158, "step": 33660 }, { "grad_norm": 0.1763090193271637, "learning_rate": 6.688048908046845e-05, "loss": 0.0171, "step": 33670 }, { "grad_norm": 0.20436687767505646, "learning_rate": 6.686103278561969e-05, "loss": 0.0133, "step": 33680 }, { "grad_norm": 0.21733316779136658, "learning_rate": 6.684157360968156e-05, "loss": 0.0163, "step": 33690 }, { "grad_norm": 0.5298353433609009, "learning_rate": 6.682211155597911e-05, "loss": 0.0141, "step": 33700 }, { "grad_norm": 0.1947278082370758, "learning_rate": 6.680264662783789e-05, "loss": 0.015, "step": 33710 }, { "grad_norm": 0.3479998707771301, "learning_rate": 6.678317882858391e-05, "loss": 0.0144, "step": 33720 }, { "grad_norm": 0.18714849650859833, "learning_rate": 6.67637081615437e-05, "loss": 0.0145, "step": 33730 }, { "grad_norm": 0.19909866154193878, "learning_rate": 6.674423463004427e-05, "loss": 0.0153, "step": 33740 }, { "grad_norm": 0.13359615206718445, "learning_rate": 6.672475823741308e-05, "loss": 0.0124, "step": 33750 }, { "grad_norm": 0.22835904359817505, "learning_rate": 6.670527898697811e-05, "loss": 0.0156, "step": 33760 }, { "grad_norm": 0.23309233784675598, "learning_rate": 6.668579688206788e-05, "loss": 0.0134, "step": 33770 }, { "grad_norm": 0.13502006232738495, "learning_rate": 6.666631192601131e-05, "loss": 0.0129, "step": 33780 }, { "grad_norm": 0.11574584990739822, "learning_rate": 6.664682412213785e-05, "loss": 0.0112, "step": 33790 }, { "grad_norm": 0.19219009578227997, "learning_rate": 6.662733347377745e-05, "loss": 0.017, "step": 33800 }, { "grad_norm": 0.21233269572257996, "learning_rate": 6.660783998426051e-05, "loss": 0.012, "step": 33810 }, { "grad_norm": 0.16521863639354706, "learning_rate": 6.658834365691794e-05, "loss": 0.0137, "step": 33820 }, { "grad_norm": 0.31268537044525146, "learning_rate": 6.656884449508115e-05, "loss": 0.0153, "step": 33830 }, { "grad_norm": 0.13847343623638153, "learning_rate": 6.654934250208198e-05, "loss": 0.0152, "step": 33840 }, { "grad_norm": 0.19848236441612244, "learning_rate": 6.65298376812528e-05, "loss": 0.0122, "step": 33850 }, { "grad_norm": 0.15311187505722046, "learning_rate": 6.651033003592646e-05, "loss": 0.014, "step": 33860 }, { "grad_norm": 0.13510338962078094, "learning_rate": 6.649081956943626e-05, "loss": 0.0098, "step": 33870 }, { "grad_norm": 0.2351563274860382, "learning_rate": 6.647130628511604e-05, "loss": 0.0149, "step": 33880 }, { "grad_norm": 0.14499837160110474, "learning_rate": 6.645179018630005e-05, "loss": 0.0117, "step": 33890 }, { "grad_norm": 0.26689088344573975, "learning_rate": 6.643227127632309e-05, "loss": 0.0159, "step": 33900 }, { "grad_norm": 0.18676164746284485, "learning_rate": 6.641274955852038e-05, "loss": 0.0157, "step": 33910 }, { "grad_norm": 0.246310293674469, "learning_rate": 6.639322503622768e-05, "loss": 0.0126, "step": 33920 }, { "grad_norm": 0.2223706692457199, "learning_rate": 6.637369771278116e-05, "loss": 0.0127, "step": 33930 }, { "grad_norm": 0.15318283438682556, "learning_rate": 6.635416759151751e-05, "loss": 0.0135, "step": 33940 }, { "grad_norm": 0.2455812394618988, "learning_rate": 6.633463467577394e-05, "loss": 0.0134, "step": 33950 }, { "grad_norm": 0.1640731394290924, "learning_rate": 6.631509896888803e-05, "loss": 0.0124, "step": 33960 }, { "grad_norm": 0.19976413249969482, "learning_rate": 6.629556047419794e-05, "loss": 0.0176, "step": 33970 }, { "grad_norm": 0.18958888947963715, "learning_rate": 6.627601919504223e-05, "loss": 0.0158, "step": 33980 }, { "grad_norm": 0.1510247141122818, "learning_rate": 6.625647513476001e-05, "loss": 0.0114, "step": 33990 }, { "grad_norm": 0.1888326108455658, "learning_rate": 6.62369282966908e-05, "loss": 0.0161, "step": 34000 }, { "grad_norm": 0.17969940602779388, "learning_rate": 6.621737868417464e-05, "loss": 0.0145, "step": 34010 }, { "grad_norm": 0.14221276342868805, "learning_rate": 6.619782630055198e-05, "loss": 0.0143, "step": 34020 }, { "grad_norm": 0.1974448263645172, "learning_rate": 6.617827114916382e-05, "loss": 0.0141, "step": 34030 }, { "grad_norm": 0.19068986177444458, "learning_rate": 6.615871323335161e-05, "loss": 0.0137, "step": 34040 }, { "grad_norm": 0.1522132307291031, "learning_rate": 6.613915255645725e-05, "loss": 0.0124, "step": 34050 }, { "grad_norm": 0.18391890823841095, "learning_rate": 6.611958912182312e-05, "loss": 0.0136, "step": 34060 }, { "grad_norm": 0.2022823542356491, "learning_rate": 6.610002293279207e-05, "loss": 0.0124, "step": 34070 }, { "grad_norm": 0.1817331314086914, "learning_rate": 6.608045399270746e-05, "loss": 0.0106, "step": 34080 }, { "grad_norm": 0.16272960603237152, "learning_rate": 6.606088230491304e-05, "loss": 0.0111, "step": 34090 }, { "grad_norm": 0.16670112311840057, "learning_rate": 6.604130787275312e-05, "loss": 0.0126, "step": 34100 }, { "grad_norm": 0.28112339973449707, "learning_rate": 6.602173069957242e-05, "loss": 0.0152, "step": 34110 }, { "grad_norm": 0.12778420746326447, "learning_rate": 6.600215078871612e-05, "loss": 0.0158, "step": 34120 }, { "grad_norm": 0.22684946656227112, "learning_rate": 6.598256814352992e-05, "loss": 0.0125, "step": 34130 }, { "grad_norm": 0.19321613013744354, "learning_rate": 6.596298276735995e-05, "loss": 0.0122, "step": 34140 }, { "grad_norm": 0.22452926635742188, "learning_rate": 6.594339466355282e-05, "loss": 0.0157, "step": 34150 }, { "grad_norm": 0.17687946557998657, "learning_rate": 6.592380383545558e-05, "loss": 0.0129, "step": 34160 }, { "grad_norm": 0.4798657298088074, "learning_rate": 6.590421028641577e-05, "loss": 0.0139, "step": 34170 }, { "grad_norm": 0.20353150367736816, "learning_rate": 6.588461401978143e-05, "loss": 0.015, "step": 34180 }, { "grad_norm": 0.21060454845428467, "learning_rate": 6.586501503890099e-05, "loss": 0.0172, "step": 34190 }, { "grad_norm": 0.18300722539424896, "learning_rate": 6.584541334712338e-05, "loss": 0.0143, "step": 34200 }, { "grad_norm": 0.2020622342824936, "learning_rate": 6.582580894779802e-05, "loss": 0.0165, "step": 34210 }, { "grad_norm": 0.32910799980163574, "learning_rate": 6.580620184427473e-05, "loss": 0.0176, "step": 34220 }, { "grad_norm": 0.2515679895877838, "learning_rate": 6.578659203990385e-05, "loss": 0.012, "step": 34230 }, { "grad_norm": 0.2337794452905655, "learning_rate": 6.576697953803615e-05, "loss": 0.0134, "step": 34240 }, { "grad_norm": 0.2391791045665741, "learning_rate": 6.57473643420229e-05, "loss": 0.0146, "step": 34250 }, { "grad_norm": 0.2269025295972824, "learning_rate": 6.572774645521574e-05, "loss": 0.014, "step": 34260 }, { "grad_norm": 0.2187250256538391, "learning_rate": 6.570812588096688e-05, "loss": 0.0159, "step": 34270 }, { "grad_norm": 0.1589471697807312, "learning_rate": 6.568850262262893e-05, "loss": 0.0146, "step": 34280 }, { "grad_norm": 0.1356225162744522, "learning_rate": 6.566887668355497e-05, "loss": 0.0179, "step": 34290 }, { "grad_norm": 0.20563185214996338, "learning_rate": 6.564924806709851e-05, "loss": 0.0146, "step": 34300 }, { "grad_norm": 0.18706895411014557, "learning_rate": 6.562961677661359e-05, "loss": 0.0169, "step": 34310 }, { "grad_norm": 0.2555154860019684, "learning_rate": 6.56099828154546e-05, "loss": 0.0144, "step": 34320 }, { "grad_norm": 0.2130414992570877, "learning_rate": 6.55903461869765e-05, "loss": 0.0146, "step": 34330 }, { "grad_norm": 0.1927894949913025, "learning_rate": 6.557070689453465e-05, "loss": 0.0186, "step": 34340 }, { "grad_norm": 0.18411187827587128, "learning_rate": 6.555106494148482e-05, "loss": 0.0169, "step": 34350 }, { "grad_norm": 0.23623321950435638, "learning_rate": 6.553142033118333e-05, "loss": 0.0151, "step": 34360 }, { "grad_norm": 0.22376714646816254, "learning_rate": 6.551177306698688e-05, "loss": 0.0142, "step": 34370 }, { "grad_norm": 0.22745607793331146, "learning_rate": 6.549212315225267e-05, "loss": 0.0186, "step": 34380 }, { "grad_norm": 0.2020331472158432, "learning_rate": 6.547247059033833e-05, "loss": 0.0152, "step": 34390 }, { "grad_norm": 0.19458819925785065, "learning_rate": 6.545281538460193e-05, "loss": 0.0168, "step": 34400 }, { "grad_norm": 0.23730237782001495, "learning_rate": 6.543315753840202e-05, "loss": 0.0154, "step": 34410 }, { "grad_norm": 0.15902750194072723, "learning_rate": 6.541349705509758e-05, "loss": 0.0162, "step": 34420 }, { "grad_norm": 0.22860828042030334, "learning_rate": 6.539383393804805e-05, "loss": 0.0202, "step": 34430 }, { "grad_norm": 0.2348070740699768, "learning_rate": 6.537416819061333e-05, "loss": 0.0141, "step": 34440 }, { "grad_norm": 0.1781582087278366, "learning_rate": 6.535449981615375e-05, "loss": 0.0162, "step": 34450 }, { "grad_norm": 0.19162306189537048, "learning_rate": 6.53348288180301e-05, "loss": 0.0148, "step": 34460 }, { "grad_norm": 0.18988007307052612, "learning_rate": 6.531515519960361e-05, "loss": 0.0141, "step": 34470 }, { "grad_norm": 0.22633007168769836, "learning_rate": 6.529547896423597e-05, "loss": 0.014, "step": 34480 }, { "grad_norm": 0.2573505938053131, "learning_rate": 6.52758001152893e-05, "loss": 0.0159, "step": 34490 }, { "grad_norm": 0.19833099842071533, "learning_rate": 6.525611865612618e-05, "loss": 0.0158, "step": 34500 }, { "grad_norm": 0.156576007604599, "learning_rate": 6.523643459010966e-05, "loss": 0.0144, "step": 34510 }, { "grad_norm": 0.25738024711608887, "learning_rate": 6.521674792060317e-05, "loss": 0.0158, "step": 34520 }, { "grad_norm": 0.21598820388317108, "learning_rate": 6.519705865097063e-05, "loss": 0.0133, "step": 34530 }, { "grad_norm": 0.22343404591083527, "learning_rate": 6.517736678457641e-05, "loss": 0.0127, "step": 34540 }, { "grad_norm": 0.22161416709423065, "learning_rate": 6.515767232478534e-05, "loss": 0.0158, "step": 34550 }, { "grad_norm": 0.1773945391178131, "learning_rate": 6.51379752749626e-05, "loss": 0.0154, "step": 34560 }, { "grad_norm": 0.12320670485496521, "learning_rate": 6.511827563847393e-05, "loss": 0.0111, "step": 34570 }, { "grad_norm": 0.16023145616054535, "learning_rate": 6.509857341868542e-05, "loss": 0.0134, "step": 34580 }, { "grad_norm": 0.1861056089401245, "learning_rate": 6.507886861896367e-05, "loss": 0.0126, "step": 34590 }, { "grad_norm": 0.16267569363117218, "learning_rate": 6.505916124267567e-05, "loss": 0.013, "step": 34600 }, { "grad_norm": 0.1667572557926178, "learning_rate": 6.503945129318891e-05, "loss": 0.0138, "step": 34610 }, { "grad_norm": 0.1744374930858612, "learning_rate": 6.501973877387122e-05, "loss": 0.014, "step": 34620 }, { "grad_norm": 0.22729988396167755, "learning_rate": 6.500002368809098e-05, "loss": 0.0139, "step": 34630 }, { "grad_norm": 0.1554151475429535, "learning_rate": 6.498030603921694e-05, "loss": 0.0113, "step": 34640 }, { "grad_norm": 0.18181316554546356, "learning_rate": 6.496058583061832e-05, "loss": 0.0131, "step": 34650 }, { "grad_norm": 0.1868169754743576, "learning_rate": 6.494086306566475e-05, "loss": 0.016, "step": 34660 }, { "grad_norm": 0.16617853939533234, "learning_rate": 6.492113774772632e-05, "loss": 0.0116, "step": 34670 }, { "grad_norm": 0.22123688459396362, "learning_rate": 6.490140988017354e-05, "loss": 0.0132, "step": 34680 }, { "grad_norm": 0.31739678978919983, "learning_rate": 6.488167946637736e-05, "loss": 0.0179, "step": 34690 }, { "grad_norm": 0.23225821554660797, "learning_rate": 6.486194650970915e-05, "loss": 0.014, "step": 34700 }, { "grad_norm": 0.17412827908992767, "learning_rate": 6.48422110135408e-05, "loss": 0.0124, "step": 34710 }, { "grad_norm": 0.1997133493423462, "learning_rate": 6.482247298124451e-05, "loss": 0.0143, "step": 34720 }, { "grad_norm": 0.11810155212879181, "learning_rate": 6.480273241619297e-05, "loss": 0.0128, "step": 34730 }, { "grad_norm": 0.25185179710388184, "learning_rate": 6.478298932175933e-05, "loss": 0.0139, "step": 34740 }, { "grad_norm": 0.19660761952400208, "learning_rate": 6.476324370131712e-05, "loss": 0.0146, "step": 34750 }, { "grad_norm": 0.27280935645103455, "learning_rate": 6.474349555824036e-05, "loss": 0.017, "step": 34760 }, { "grad_norm": 0.28514984250068665, "learning_rate": 6.472374489590342e-05, "loss": 0.0147, "step": 34770 }, { "grad_norm": 0.22577717900276184, "learning_rate": 6.470399171768118e-05, "loss": 0.0194, "step": 34780 }, { "grad_norm": 0.24408230185508728, "learning_rate": 6.468423602694891e-05, "loss": 0.0166, "step": 34790 }, { "grad_norm": 0.24298161268234253, "learning_rate": 6.466447782708232e-05, "loss": 0.0228, "step": 34800 }, { "grad_norm": 0.2662496566772461, "learning_rate": 6.464471712145754e-05, "loss": 0.0137, "step": 34810 }, { "grad_norm": 0.16649086773395538, "learning_rate": 6.462495391345114e-05, "loss": 0.0124, "step": 34820 }, { "grad_norm": 0.20631231367588043, "learning_rate": 6.46051882064401e-05, "loss": 0.0117, "step": 34830 }, { "grad_norm": 0.2143995314836502, "learning_rate": 6.458542000380186e-05, "loss": 0.0143, "step": 34840 }, { "grad_norm": 0.20378242433071136, "learning_rate": 6.456564930891424e-05, "loss": 0.0174, "step": 34850 }, { "grad_norm": 0.2522684633731842, "learning_rate": 6.454587612515555e-05, "loss": 0.0152, "step": 34860 }, { "grad_norm": 0.20202374458312988, "learning_rate": 6.452610045590444e-05, "loss": 0.0124, "step": 34870 }, { "grad_norm": 0.17015324532985687, "learning_rate": 6.450632230454005e-05, "loss": 0.0125, "step": 34880 }, { "grad_norm": 0.13252922892570496, "learning_rate": 6.448654167444195e-05, "loss": 0.0156, "step": 34890 }, { "grad_norm": 0.21638357639312744, "learning_rate": 6.446675856899005e-05, "loss": 0.0116, "step": 34900 }, { "grad_norm": 0.257358193397522, "learning_rate": 6.444697299156481e-05, "loss": 0.0138, "step": 34910 }, { "grad_norm": 0.21935899555683136, "learning_rate": 6.442718494554701e-05, "loss": 0.012, "step": 34920 }, { "grad_norm": 0.2721976041793823, "learning_rate": 6.440739443431787e-05, "loss": 0.0164, "step": 34930 }, { "grad_norm": 0.12824499607086182, "learning_rate": 6.438760146125906e-05, "loss": 0.0166, "step": 34940 }, { "grad_norm": 0.19345039129257202, "learning_rate": 6.436780602975267e-05, "loss": 0.0143, "step": 34950 }, { "grad_norm": 0.17476189136505127, "learning_rate": 6.43480081431812e-05, "loss": 0.0125, "step": 34960 }, { "grad_norm": 0.18665699660778046, "learning_rate": 6.432820780492756e-05, "loss": 0.0139, "step": 34970 }, { "grad_norm": 0.20817537605762482, "learning_rate": 6.430840501837506e-05, "loss": 0.0149, "step": 34980 }, { "grad_norm": 0.2703916132450104, "learning_rate": 6.428859978690748e-05, "loss": 0.0129, "step": 34990 }, { "grad_norm": 0.2175086885690689, "learning_rate": 6.426879211390901e-05, "loss": 0.0155, "step": 35000 }, { "grad_norm": 0.2824026942253113, "learning_rate": 6.424898200276422e-05, "loss": 0.012, "step": 35010 }, { "grad_norm": 0.26486027240753174, "learning_rate": 6.42291694568581e-05, "loss": 0.013, "step": 35020 }, { "grad_norm": 0.15101492404937744, "learning_rate": 6.42093544795761e-05, "loss": 0.0131, "step": 35030 }, { "grad_norm": 0.2520304024219513, "learning_rate": 6.418953707430403e-05, "loss": 0.0148, "step": 35040 }, { "grad_norm": 0.19463399052619934, "learning_rate": 6.416971724442819e-05, "loss": 0.0142, "step": 35050 }, { "grad_norm": 0.15944631397724152, "learning_rate": 6.414989499333519e-05, "loss": 0.0157, "step": 35060 }, { "grad_norm": 0.2026177942752838, "learning_rate": 6.413007032441214e-05, "loss": 0.0153, "step": 35070 }, { "grad_norm": 0.32661646604537964, "learning_rate": 6.411024324104653e-05, "loss": 0.015, "step": 35080 }, { "grad_norm": 0.2385907918214798, "learning_rate": 6.409041374662628e-05, "loss": 0.0137, "step": 35090 }, { "grad_norm": 0.25278735160827637, "learning_rate": 6.407058184453967e-05, "loss": 0.0161, "step": 35100 }, { "grad_norm": 0.25185245275497437, "learning_rate": 6.405074753817548e-05, "loss": 0.0187, "step": 35110 }, { "grad_norm": 0.21488043665885925, "learning_rate": 6.40309108309228e-05, "loss": 0.0137, "step": 35120 }, { "grad_norm": 0.19976180791854858, "learning_rate": 6.401107172617122e-05, "loss": 0.0153, "step": 35130 }, { "grad_norm": 0.10084225237369537, "learning_rate": 6.399123022731068e-05, "loss": 0.016, "step": 35140 }, { "grad_norm": 0.1464826762676239, "learning_rate": 6.397138633773157e-05, "loss": 0.0165, "step": 35150 }, { "grad_norm": 0.1627667248249054, "learning_rate": 6.395154006082463e-05, "loss": 0.0123, "step": 35160 }, { "grad_norm": 0.142025426030159, "learning_rate": 6.393169139998109e-05, "loss": 0.0162, "step": 35170 }, { "grad_norm": 0.22777505218982697, "learning_rate": 6.39118403585925e-05, "loss": 0.0129, "step": 35180 }, { "grad_norm": 0.14831386506557465, "learning_rate": 6.38919869400509e-05, "loss": 0.0174, "step": 35190 }, { "grad_norm": 0.16650833189487457, "learning_rate": 6.387213114774865e-05, "loss": 0.0145, "step": 35200 }, { "grad_norm": 0.1745813637971878, "learning_rate": 6.385227298507863e-05, "loss": 0.0135, "step": 35210 }, { "grad_norm": 0.1548675000667572, "learning_rate": 6.3832412455434e-05, "loss": 0.013, "step": 35220 }, { "grad_norm": 0.16410639882087708, "learning_rate": 6.381254956220841e-05, "loss": 0.0119, "step": 35230 }, { "grad_norm": 0.12264292687177658, "learning_rate": 6.379268430879586e-05, "loss": 0.0147, "step": 35240 }, { "grad_norm": 0.15321440994739532, "learning_rate": 6.37728166985908e-05, "loss": 0.0137, "step": 35250 }, { "grad_norm": 0.1902700513601303, "learning_rate": 6.375294673498804e-05, "loss": 0.0148, "step": 35260 }, { "grad_norm": 0.2664663791656494, "learning_rate": 6.373307442138284e-05, "loss": 0.0148, "step": 35270 }, { "grad_norm": 0.2398737072944641, "learning_rate": 6.371319976117081e-05, "loss": 0.0134, "step": 35280 }, { "grad_norm": 0.23059502243995667, "learning_rate": 6.3693322757748e-05, "loss": 0.0139, "step": 35290 }, { "grad_norm": 0.2016781121492386, "learning_rate": 6.367344341451086e-05, "loss": 0.0174, "step": 35300 }, { "grad_norm": 0.21693503856658936, "learning_rate": 6.36535617348562e-05, "loss": 0.0151, "step": 35310 }, { "grad_norm": 0.22140643000602722, "learning_rate": 6.363367772218128e-05, "loss": 0.013, "step": 35320 }, { "grad_norm": 0.19856922328472137, "learning_rate": 6.36137913798837e-05, "loss": 0.015, "step": 35330 }, { "grad_norm": 0.1308463215827942, "learning_rate": 6.359390271136151e-05, "loss": 0.0122, "step": 35340 }, { "grad_norm": 0.18742012977600098, "learning_rate": 6.357401172001314e-05, "loss": 0.013, "step": 35350 }, { "grad_norm": 0.1772637516260147, "learning_rate": 6.355411840923742e-05, "loss": 0.0144, "step": 35360 }, { "grad_norm": 0.21027067303657532, "learning_rate": 6.353422278243358e-05, "loss": 0.0172, "step": 35370 }, { "grad_norm": 0.1541648805141449, "learning_rate": 6.351432484300121e-05, "loss": 0.0144, "step": 35380 }, { "grad_norm": 0.17552915215492249, "learning_rate": 6.349442459434036e-05, "loss": 0.0145, "step": 35390 }, { "grad_norm": 0.19106194376945496, "learning_rate": 6.34745220398514e-05, "loss": 0.0146, "step": 35400 }, { "grad_norm": 0.3066140413284302, "learning_rate": 6.345461718293518e-05, "loss": 0.0172, "step": 35410 }, { "grad_norm": 0.20276112854480743, "learning_rate": 6.343471002699286e-05, "loss": 0.0144, "step": 35420 }, { "grad_norm": 0.20592859387397766, "learning_rate": 6.341480057542602e-05, "loss": 0.0147, "step": 35430 }, { "grad_norm": 0.11992865055799484, "learning_rate": 6.339488883163667e-05, "loss": 0.0136, "step": 35440 }, { "grad_norm": 0.22004787623882294, "learning_rate": 6.337497479902716e-05, "loss": 0.0155, "step": 35450 }, { "grad_norm": 0.1486038714647293, "learning_rate": 6.335505848100027e-05, "loss": 0.0122, "step": 35460 }, { "grad_norm": 0.15921753644943237, "learning_rate": 6.333513988095915e-05, "loss": 0.0123, "step": 35470 }, { "grad_norm": 0.17985810339450836, "learning_rate": 6.331521900230735e-05, "loss": 0.0135, "step": 35480 }, { "grad_norm": 0.1240018755197525, "learning_rate": 6.329529584844878e-05, "loss": 0.012, "step": 35490 }, { "grad_norm": 0.1860046684741974, "learning_rate": 6.327537042278777e-05, "loss": 0.0148, "step": 35500 }, { "grad_norm": 0.148680180311203, "learning_rate": 6.325544272872905e-05, "loss": 0.0135, "step": 35510 }, { "grad_norm": 0.15143483877182007, "learning_rate": 6.323551276967771e-05, "loss": 0.0121, "step": 35520 }, { "grad_norm": 0.2584308087825775, "learning_rate": 6.321558054903922e-05, "loss": 0.0127, "step": 35530 }, { "grad_norm": 0.22163915634155273, "learning_rate": 6.319564607021947e-05, "loss": 0.0133, "step": 35540 }, { "grad_norm": 0.18711644411087036, "learning_rate": 6.31757093366247e-05, "loss": 0.0121, "step": 35550 }, { "grad_norm": 0.14478042721748352, "learning_rate": 6.315577035166154e-05, "loss": 0.0162, "step": 35560 }, { "grad_norm": 0.16978633403778076, "learning_rate": 6.313582911873708e-05, "loss": 0.0122, "step": 35570 }, { "grad_norm": 0.1875237077474594, "learning_rate": 6.311588564125865e-05, "loss": 0.0153, "step": 35580 }, { "grad_norm": 0.20490440726280212, "learning_rate": 6.30959399226341e-05, "loss": 0.0126, "step": 35590 }, { "grad_norm": 0.15956823527812958, "learning_rate": 6.30759919662716e-05, "loss": 0.0146, "step": 35600 }, { "grad_norm": 0.17818209528923035, "learning_rate": 6.30560417755797e-05, "loss": 0.0159, "step": 35610 }, { "grad_norm": 0.1903870403766632, "learning_rate": 6.303608935396735e-05, "loss": 0.012, "step": 35620 }, { "grad_norm": 0.19163349270820618, "learning_rate": 6.301613470484386e-05, "loss": 0.0139, "step": 35630 }, { "grad_norm": 0.15332329273223877, "learning_rate": 6.299617783161893e-05, "loss": 0.0169, "step": 35640 }, { "grad_norm": 0.28000497817993164, "learning_rate": 6.297621873770266e-05, "loss": 0.0169, "step": 35650 }, { "grad_norm": 0.2144908756017685, "learning_rate": 6.29562574265055e-05, "loss": 0.0117, "step": 35660 }, { "grad_norm": 0.21910062432289124, "learning_rate": 6.293629390143834e-05, "loss": 0.0172, "step": 35670 }, { "grad_norm": 0.14307233691215515, "learning_rate": 6.291632816591232e-05, "loss": 0.0125, "step": 35680 }, { "grad_norm": 0.23136818408966064, "learning_rate": 6.28963602233391e-05, "loss": 0.0135, "step": 35690 }, { "grad_norm": 0.19466978311538696, "learning_rate": 6.287639007713062e-05, "loss": 0.015, "step": 35700 }, { "grad_norm": 0.19647517800331116, "learning_rate": 6.285641773069926e-05, "loss": 0.0131, "step": 35710 }, { "grad_norm": 0.13726933300495148, "learning_rate": 6.283644318745773e-05, "loss": 0.0133, "step": 35720 }, { "grad_norm": 0.2054223269224167, "learning_rate": 6.281646645081912e-05, "loss": 0.0161, "step": 35730 }, { "grad_norm": 0.1675463616847992, "learning_rate": 6.279648752419693e-05, "loss": 0.0142, "step": 35740 }, { "grad_norm": 0.1194111555814743, "learning_rate": 6.2776506411005e-05, "loss": 0.01, "step": 35750 }, { "grad_norm": 0.17320559918880463, "learning_rate": 6.275652311465758e-05, "loss": 0.0133, "step": 35760 }, { "grad_norm": 0.21745477616786957, "learning_rate": 6.273653763856926e-05, "loss": 0.0143, "step": 35770 }, { "grad_norm": 0.1557663232088089, "learning_rate": 6.271654998615501e-05, "loss": 0.0139, "step": 35780 }, { "grad_norm": 0.22851525247097015, "learning_rate": 6.269656016083013e-05, "loss": 0.0137, "step": 35790 }, { "grad_norm": 0.23175151646137238, "learning_rate": 6.267656816601038e-05, "loss": 0.0139, "step": 35800 }, { "grad_norm": 0.2428237497806549, "learning_rate": 6.265657400511185e-05, "loss": 0.0122, "step": 35810 }, { "grad_norm": 0.18374361097812653, "learning_rate": 6.263657768155098e-05, "loss": 0.0142, "step": 35820 }, { "grad_norm": 0.16734227538108826, "learning_rate": 6.261657919874457e-05, "loss": 0.0161, "step": 35830 }, { "grad_norm": 0.24022717773914337, "learning_rate": 6.259657856010986e-05, "loss": 0.0164, "step": 35840 }, { "grad_norm": 0.20844770967960358, "learning_rate": 6.257657576906439e-05, "loss": 0.0135, "step": 35850 }, { "grad_norm": 0.18682222068309784, "learning_rate": 6.255657082902609e-05, "loss": 0.0122, "step": 35860 }, { "grad_norm": 0.678239643573761, "learning_rate": 6.253656374341325e-05, "loss": 0.0159, "step": 35870 }, { "grad_norm": 0.1682630181312561, "learning_rate": 6.251655451564457e-05, "loss": 0.0175, "step": 35880 }, { "grad_norm": 0.2335280030965805, "learning_rate": 6.249654314913902e-05, "loss": 0.0144, "step": 35890 }, { "grad_norm": 0.17729529738426208, "learning_rate": 6.247652964731604e-05, "loss": 0.0165, "step": 35900 }, { "grad_norm": 0.2433188557624817, "learning_rate": 6.245651401359537e-05, "loss": 0.0143, "step": 35910 }, { "grad_norm": 0.1832723170518875, "learning_rate": 6.243649625139715e-05, "loss": 0.0157, "step": 35920 }, { "grad_norm": 0.17797228693962097, "learning_rate": 6.241647636414185e-05, "loss": 0.0161, "step": 35930 }, { "grad_norm": 0.16917476058006287, "learning_rate": 6.239645435525034e-05, "loss": 0.0147, "step": 35940 }, { "grad_norm": 0.20952394604682922, "learning_rate": 6.237643022814381e-05, "loss": 0.0117, "step": 35950 }, { "grad_norm": 0.1450124830007553, "learning_rate": 6.235640398624386e-05, "loss": 0.0108, "step": 35960 }, { "grad_norm": 0.1905018538236618, "learning_rate": 6.233637563297243e-05, "loss": 0.0125, "step": 35970 }, { "grad_norm": 0.16875791549682617, "learning_rate": 6.23163451717518e-05, "loss": 0.0168, "step": 35980 }, { "grad_norm": 0.1911802589893341, "learning_rate": 6.229631260600463e-05, "loss": 0.0152, "step": 35990 }, { "grad_norm": 0.21416901051998138, "learning_rate": 6.227627793915392e-05, "loss": 0.017, "step": 36000 }, { "grad_norm": 0.2078629434108734, "learning_rate": 6.225624117462309e-05, "loss": 0.0162, "step": 36010 }, { "grad_norm": 0.27636969089508057, "learning_rate": 6.223620231583586e-05, "loss": 0.0125, "step": 36020 }, { "grad_norm": 0.2364831417798996, "learning_rate": 6.221616136621629e-05, "loss": 0.0134, "step": 36030 }, { "grad_norm": 0.21090605854988098, "learning_rate": 6.219611832918887e-05, "loss": 0.0152, "step": 36040 }, { "grad_norm": 0.1390971839427948, "learning_rate": 6.217607320817838e-05, "loss": 0.0129, "step": 36050 }, { "grad_norm": 0.26557567715644836, "learning_rate": 6.215602600661001e-05, "loss": 0.0165, "step": 36060 }, { "grad_norm": 0.2007860243320465, "learning_rate": 6.213597672790925e-05, "loss": 0.0173, "step": 36070 }, { "grad_norm": 0.19692017138004303, "learning_rate": 6.2115925375502e-05, "loss": 0.0132, "step": 36080 }, { "grad_norm": 0.15450933575630188, "learning_rate": 6.209587195281447e-05, "loss": 0.0152, "step": 36090 }, { "grad_norm": 0.15273648500442505, "learning_rate": 6.207581646327324e-05, "loss": 0.0108, "step": 36100 }, { "grad_norm": 0.1968441605567932, "learning_rate": 6.205575891030526e-05, "loss": 0.0151, "step": 36110 }, { "grad_norm": 0.20644506812095642, "learning_rate": 6.203569929733781e-05, "loss": 0.012, "step": 36120 }, { "grad_norm": 0.2289627492427826, "learning_rate": 6.201563762779852e-05, "loss": 0.0162, "step": 36130 }, { "grad_norm": 0.273600310087204, "learning_rate": 6.199557390511538e-05, "loss": 0.0166, "step": 36140 }, { "grad_norm": 0.19059765338897705, "learning_rate": 6.197550813271675e-05, "loss": 0.0142, "step": 36150 }, { "grad_norm": 0.17608731985092163, "learning_rate": 6.195544031403131e-05, "loss": 0.013, "step": 36160 }, { "grad_norm": 0.19807007908821106, "learning_rate": 6.19353704524881e-05, "loss": 0.016, "step": 36170 }, { "grad_norm": 0.20573337376117706, "learning_rate": 6.191529855151652e-05, "loss": 0.0164, "step": 36180 }, { "grad_norm": 0.25154435634613037, "learning_rate": 6.189522461454629e-05, "loss": 0.0149, "step": 36190 }, { "grad_norm": 0.15948864817619324, "learning_rate": 6.187514864500752e-05, "loss": 0.0112, "step": 36200 }, { "grad_norm": 0.1962079554796219, "learning_rate": 6.185507064633062e-05, "loss": 0.0127, "step": 36210 }, { "grad_norm": 0.20846320688724518, "learning_rate": 6.18349906219464e-05, "loss": 0.0114, "step": 36220 }, { "grad_norm": 0.17528659105300903, "learning_rate": 6.181490857528596e-05, "loss": 0.0131, "step": 36230 }, { "grad_norm": 0.1580164134502411, "learning_rate": 6.179482450978077e-05, "loss": 0.0136, "step": 36240 }, { "grad_norm": 0.18822377920150757, "learning_rate": 6.177473842886269e-05, "loss": 0.0114, "step": 36250 }, { "grad_norm": 0.17568254470825195, "learning_rate": 6.175465033596382e-05, "loss": 0.0159, "step": 36260 }, { "grad_norm": 0.2279297113418579, "learning_rate": 6.173456023451671e-05, "loss": 0.0158, "step": 36270 }, { "grad_norm": 0.4762137830257416, "learning_rate": 6.171446812795422e-05, "loss": 0.0187, "step": 36280 }, { "grad_norm": 0.19377318024635315, "learning_rate": 6.169437401970949e-05, "loss": 0.0142, "step": 36290 }, { "grad_norm": 0.2256266474723816, "learning_rate": 6.16742779132161e-05, "loss": 0.0141, "step": 36300 }, { "grad_norm": 0.23975056409835815, "learning_rate": 6.165417981190789e-05, "loss": 0.0129, "step": 36310 }, { "grad_norm": 0.1880016326904297, "learning_rate": 6.16340797192191e-05, "loss": 0.0135, "step": 36320 }, { "grad_norm": 0.24614860117435455, "learning_rate": 6.161397763858427e-05, "loss": 0.0213, "step": 36330 }, { "grad_norm": 0.11545088142156601, "learning_rate": 6.159387357343834e-05, "loss": 0.0152, "step": 36340 }, { "grad_norm": 0.17524199187755585, "learning_rate": 6.157376752721648e-05, "loss": 0.0168, "step": 36350 }, { "grad_norm": 0.19674408435821533, "learning_rate": 6.155365950335428e-05, "loss": 0.015, "step": 36360 }, { "grad_norm": 0.1602582186460495, "learning_rate": 6.153354950528768e-05, "loss": 0.016, "step": 36370 }, { "grad_norm": 0.16153141856193542, "learning_rate": 6.151343753645293e-05, "loss": 0.0143, "step": 36380 }, { "grad_norm": 0.20568937063217163, "learning_rate": 6.149332360028657e-05, "loss": 0.0165, "step": 36390 }, { "grad_norm": 0.27900272607803345, "learning_rate": 6.147320770022555e-05, "loss": 0.0148, "step": 36400 }, { "grad_norm": 0.16343699395656586, "learning_rate": 6.145308983970715e-05, "loss": 0.0156, "step": 36410 }, { "grad_norm": 0.14771752059459686, "learning_rate": 6.143297002216892e-05, "loss": 0.01, "step": 36420 }, { "grad_norm": 0.2366308569908142, "learning_rate": 6.141284825104882e-05, "loss": 0.0166, "step": 36430 }, { "grad_norm": 0.27784013748168945, "learning_rate": 6.13927245297851e-05, "loss": 0.0142, "step": 36440 }, { "grad_norm": 0.20561352372169495, "learning_rate": 6.137259886181633e-05, "loss": 0.0133, "step": 36450 }, { "grad_norm": 0.2183699756860733, "learning_rate": 6.135247125058145e-05, "loss": 0.0139, "step": 36460 }, { "grad_norm": 0.20054344832897186, "learning_rate": 6.133234169951974e-05, "loss": 0.0153, "step": 36470 }, { "grad_norm": 0.2704838514328003, "learning_rate": 6.131221021207078e-05, "loss": 0.014, "step": 36480 }, { "grad_norm": 0.1455344706773758, "learning_rate": 6.129207679167448e-05, "loss": 0.0138, "step": 36490 }, { "grad_norm": 0.1425405740737915, "learning_rate": 6.127194144177109e-05, "loss": 0.0114, "step": 36500 }, { "grad_norm": 0.139444500207901, "learning_rate": 6.125180416580118e-05, "loss": 0.0144, "step": 36510 }, { "grad_norm": 0.17436832189559937, "learning_rate": 6.123166496720571e-05, "loss": 0.0119, "step": 36520 }, { "grad_norm": 0.1771218627691269, "learning_rate": 6.121152384942588e-05, "loss": 0.0111, "step": 36530 }, { "grad_norm": 0.1740073412656784, "learning_rate": 6.119138081590324e-05, "loss": 0.014, "step": 36540 }, { "grad_norm": 0.20521457493305206, "learning_rate": 6.117123587007971e-05, "loss": 0.0138, "step": 36550 }, { "grad_norm": 0.163853719830513, "learning_rate": 6.11510890153975e-05, "loss": 0.015, "step": 36560 }, { "grad_norm": 0.2408236563205719, "learning_rate": 6.113094025529916e-05, "loss": 0.012, "step": 36570 }, { "grad_norm": 0.1937365084886551, "learning_rate": 6.111078959322757e-05, "loss": 0.0129, "step": 36580 }, { "grad_norm": 0.14309994876384735, "learning_rate": 6.109063703262592e-05, "loss": 0.0128, "step": 36590 }, { "grad_norm": 0.10705702006816864, "learning_rate": 6.107048257693772e-05, "loss": 0.0112, "step": 36600 }, { "grad_norm": 0.18470166623592377, "learning_rate": 6.105032622960683e-05, "loss": 0.0123, "step": 36610 }, { "grad_norm": 0.15557695925235748, "learning_rate": 6.103016799407743e-05, "loss": 0.0128, "step": 36620 }, { "grad_norm": 0.15698783099651337, "learning_rate": 6.1010007873793984e-05, "loss": 0.0115, "step": 36630 }, { "grad_norm": 0.22078511118888855, "learning_rate": 6.098984587220131e-05, "loss": 0.0152, "step": 36640 }, { "grad_norm": 0.22317031025886536, "learning_rate": 6.096968199274456e-05, "loss": 0.0128, "step": 36650 }, { "grad_norm": 0.23464111983776093, "learning_rate": 6.0949516238869166e-05, "loss": 0.0164, "step": 36660 }, { "grad_norm": 0.17365233600139618, "learning_rate": 6.092934861402092e-05, "loss": 0.0125, "step": 36670 }, { "grad_norm": 0.15836432576179504, "learning_rate": 6.0909179121645924e-05, "loss": 0.0144, "step": 36680 }, { "grad_norm": 0.20892854034900665, "learning_rate": 6.0889007765190576e-05, "loss": 0.0135, "step": 36690 }, { "grad_norm": 0.22631610929965973, "learning_rate": 6.086883454810162e-05, "loss": 0.0158, "step": 36700 }, { "grad_norm": 0.13231691718101501, "learning_rate": 6.0848659473826084e-05, "loss": 0.0132, "step": 36710 }, { "grad_norm": 0.161878302693367, "learning_rate": 6.082848254581138e-05, "loss": 0.0134, "step": 36720 }, { "grad_norm": 0.30337515473365784, "learning_rate": 6.080830376750517e-05, "loss": 0.0174, "step": 36730 }, { "grad_norm": 0.19449912011623383, "learning_rate": 6.0788123142355445e-05, "loss": 0.0151, "step": 36740 }, { "grad_norm": 0.25537174940109253, "learning_rate": 6.076794067381052e-05, "loss": 0.0195, "step": 36750 }, { "grad_norm": 0.1935366839170456, "learning_rate": 6.074775636531905e-05, "loss": 0.0141, "step": 36760 }, { "grad_norm": 0.13291314244270325, "learning_rate": 6.072757022032997e-05, "loss": 0.0126, "step": 36770 }, { "grad_norm": 0.26337310671806335, "learning_rate": 6.070738224229253e-05, "loss": 0.015, "step": 36780 }, { "grad_norm": 0.23510555922985077, "learning_rate": 6.0687192434656314e-05, "loss": 0.014, "step": 36790 }, { "grad_norm": 0.19470496475696564, "learning_rate": 6.066700080087121e-05, "loss": 0.015, "step": 36800 }, { "grad_norm": 0.268566370010376, "learning_rate": 6.0646807344387424e-05, "loss": 0.0177, "step": 36810 }, { "grad_norm": 0.18338075280189514, "learning_rate": 6.062661206865543e-05, "loss": 0.0154, "step": 36820 }, { "grad_norm": 0.16040997207164764, "learning_rate": 6.06064149771261e-05, "loss": 0.012, "step": 36830 }, { "grad_norm": 0.2216443419456482, "learning_rate": 6.058621607325051e-05, "loss": 0.0164, "step": 36840 }, { "grad_norm": 0.18702995777130127, "learning_rate": 6.056601536048014e-05, "loss": 0.0146, "step": 36850 }, { "grad_norm": 0.1694028377532959, "learning_rate": 6.0545812842266725e-05, "loss": 0.0101, "step": 36860 }, { "grad_norm": 0.13817870616912842, "learning_rate": 6.052560852206232e-05, "loss": 0.0116, "step": 36870 }, { "grad_norm": 0.12156351655721664, "learning_rate": 6.05054024033193e-05, "loss": 0.0124, "step": 36880 }, { "grad_norm": 0.2562633156776428, "learning_rate": 6.048519448949032e-05, "loss": 0.0137, "step": 36890 }, { "grad_norm": 0.2607855200767517, "learning_rate": 6.046498478402839e-05, "loss": 0.0121, "step": 36900 }, { "grad_norm": 0.2653660476207733, "learning_rate": 6.044477329038677e-05, "loss": 0.0149, "step": 36910 }, { "grad_norm": 0.22044607996940613, "learning_rate": 6.042456001201906e-05, "loss": 0.0169, "step": 36920 }, { "grad_norm": 0.17401738464832306, "learning_rate": 6.040434495237917e-05, "loss": 0.0141, "step": 36930 }, { "grad_norm": 0.15306778252124786, "learning_rate": 6.0384128114921256e-05, "loss": 0.0104, "step": 36940 }, { "grad_norm": 0.31900203227996826, "learning_rate": 6.036390950309987e-05, "loss": 0.0133, "step": 36950 }, { "grad_norm": 0.18678726255893707, "learning_rate": 6.0343689120369805e-05, "loss": 0.0132, "step": 36960 }, { "grad_norm": 0.2315409779548645, "learning_rate": 6.032346697018616e-05, "loss": 0.0148, "step": 36970 }, { "grad_norm": 0.24358032643795013, "learning_rate": 6.0303243056004375e-05, "loss": 0.0112, "step": 36980 }, { "grad_norm": 0.19560469686985016, "learning_rate": 6.0283017381280136e-05, "loss": 0.0145, "step": 36990 }, { "grad_norm": 0.1833096593618393, "learning_rate": 6.026278994946948e-05, "loss": 0.0137, "step": 37000 }, { "grad_norm": 0.25608089566230774, "learning_rate": 6.02425607640287e-05, "loss": 0.0126, "step": 37010 }, { "grad_norm": 0.12321697920560837, "learning_rate": 6.022232982841441e-05, "loss": 0.0146, "step": 37020 }, { "grad_norm": 0.1601116806268692, "learning_rate": 6.020209714608355e-05, "loss": 0.014, "step": 37030 }, { "grad_norm": 0.260835200548172, "learning_rate": 6.018186272049331e-05, "loss": 0.0166, "step": 37040 }, { "grad_norm": 0.21332302689552307, "learning_rate": 6.01616265551012e-05, "loss": 0.0115, "step": 37050 }, { "grad_norm": 0.2255147397518158, "learning_rate": 6.014138865336503e-05, "loss": 0.0148, "step": 37060 }, { "grad_norm": 0.21608221530914307, "learning_rate": 6.0121149018742905e-05, "loss": 0.0164, "step": 37070 }, { "grad_norm": 0.20335039496421814, "learning_rate": 6.010090765469325e-05, "loss": 0.0175, "step": 37080 }, { "grad_norm": 0.24904271960258484, "learning_rate": 6.008066456467473e-05, "loss": 0.0153, "step": 37090 }, { "grad_norm": 0.2564133107662201, "learning_rate": 6.0060419752146335e-05, "loss": 0.0153, "step": 37100 }, { "grad_norm": 0.21808932721614838, "learning_rate": 6.0040173220567353e-05, "loss": 0.0146, "step": 37110 }, { "grad_norm": 0.13731981813907623, "learning_rate": 6.001992497339737e-05, "loss": 0.0134, "step": 37120 }, { "grad_norm": 0.20283924043178558, "learning_rate": 5.999967501409626e-05, "loss": 0.0164, "step": 37130 }, { "grad_norm": 0.26265376806259155, "learning_rate": 5.997942334612418e-05, "loss": 0.0189, "step": 37140 }, { "grad_norm": 0.28972336649894714, "learning_rate": 5.995916997294158e-05, "loss": 0.0145, "step": 37150 }, { "grad_norm": 0.14762476086616516, "learning_rate": 5.9938914898009214e-05, "loss": 0.0116, "step": 37160 }, { "grad_norm": 0.132644921541214, "learning_rate": 5.991865812478813e-05, "loss": 0.013, "step": 37170 }, { "grad_norm": 0.1601286083459854, "learning_rate": 5.989839965673964e-05, "loss": 0.014, "step": 37180 }, { "grad_norm": 0.17139950394630432, "learning_rate": 5.987813949732539e-05, "loss": 0.0141, "step": 37190 }, { "grad_norm": 0.19648310542106628, "learning_rate": 5.9857877650007255e-05, "loss": 0.0132, "step": 37200 }, { "grad_norm": 0.2883033752441406, "learning_rate": 5.983761411824744e-05, "loss": 0.0155, "step": 37210 }, { "grad_norm": 0.1251402348279953, "learning_rate": 5.981734890550844e-05, "loss": 0.0127, "step": 37220 }, { "grad_norm": 0.19828778505325317, "learning_rate": 5.979708201525301e-05, "loss": 0.0133, "step": 37230 }, { "grad_norm": 0.18447795510292053, "learning_rate": 5.977681345094422e-05, "loss": 0.0111, "step": 37240 }, { "grad_norm": 0.15098652243614197, "learning_rate": 5.97565432160454e-05, "loss": 0.0127, "step": 37250 }, { "grad_norm": 0.1907643973827362, "learning_rate": 5.9736271314020186e-05, "loss": 0.0103, "step": 37260 }, { "grad_norm": 0.19419759511947632, "learning_rate": 5.971599774833251e-05, "loss": 0.012, "step": 37270 }, { "grad_norm": 0.19144155085086823, "learning_rate": 5.9695722522446525e-05, "loss": 0.0149, "step": 37280 }, { "grad_norm": 0.1994095742702484, "learning_rate": 5.9675445639826765e-05, "loss": 0.0105, "step": 37290 }, { "grad_norm": 0.13401725888252258, "learning_rate": 5.965516710393796e-05, "loss": 0.0109, "step": 37300 }, { "grad_norm": 0.1902185082435608, "learning_rate": 5.963488691824516e-05, "loss": 0.0133, "step": 37310 }, { "grad_norm": 0.12352244555950165, "learning_rate": 5.96146050862137e-05, "loss": 0.0108, "step": 37320 }, { "grad_norm": 0.13085223734378815, "learning_rate": 5.959432161130919e-05, "loss": 0.0114, "step": 37330 }, { "grad_norm": 0.13259275257587433, "learning_rate": 5.9574036496997545e-05, "loss": 0.0118, "step": 37340 }, { "grad_norm": 0.21151548624038696, "learning_rate": 5.955374974674488e-05, "loss": 0.0129, "step": 37350 }, { "grad_norm": 0.1739271730184555, "learning_rate": 5.9533461364017696e-05, "loss": 0.0134, "step": 37360 }, { "grad_norm": 0.19200588762760162, "learning_rate": 5.9513171352282716e-05, "loss": 0.0128, "step": 37370 }, { "grad_norm": 0.24876365065574646, "learning_rate": 5.949287971500692e-05, "loss": 0.0158, "step": 37380 }, { "grad_norm": 0.21239706873893738, "learning_rate": 5.947258645565762e-05, "loss": 0.0105, "step": 37390 }, { "grad_norm": 0.20227929949760437, "learning_rate": 5.945229157770237e-05, "loss": 0.0144, "step": 37400 }, { "grad_norm": 0.134829580783844, "learning_rate": 5.9431995084609006e-05, "loss": 0.0105, "step": 37410 }, { "grad_norm": 0.24814923107624054, "learning_rate": 5.941169697984564e-05, "loss": 0.0122, "step": 37420 }, { "grad_norm": 0.16266892850399017, "learning_rate": 5.9391397266880675e-05, "loss": 0.0123, "step": 37430 }, { "grad_norm": 0.14353147149085999, "learning_rate": 5.937109594918279e-05, "loss": 0.0125, "step": 37440 }, { "grad_norm": 0.1486886441707611, "learning_rate": 5.9350793030220884e-05, "loss": 0.0098, "step": 37450 }, { "grad_norm": 0.15222033858299255, "learning_rate": 5.933048851346421e-05, "loss": 0.0099, "step": 37460 }, { "grad_norm": 0.28809836506843567, "learning_rate": 5.931018240238224e-05, "loss": 0.0159, "step": 37470 }, { "grad_norm": 0.24897702038288116, "learning_rate": 5.928987470044471e-05, "loss": 0.0136, "step": 37480 }, { "grad_norm": 0.22009040415287018, "learning_rate": 5.9269565411121695e-05, "loss": 0.0133, "step": 37490 }, { "grad_norm": 0.24498477578163147, "learning_rate": 5.924925453788347e-05, "loss": 0.0118, "step": 37500 }, { "grad_norm": 0.1873750388622284, "learning_rate": 5.92289420842006e-05, "loss": 0.0109, "step": 37510 }, { "grad_norm": 0.19560378789901733, "learning_rate": 5.9208628053543945e-05, "loss": 0.0119, "step": 37520 }, { "grad_norm": 0.23566068708896637, "learning_rate": 5.918831244938462e-05, "loss": 0.0115, "step": 37530 }, { "grad_norm": 0.15979653596878052, "learning_rate": 5.9167995275194e-05, "loss": 0.0121, "step": 37540 }, { "grad_norm": 0.18953262269496918, "learning_rate": 5.914767653444373e-05, "loss": 0.0108, "step": 37550 }, { "grad_norm": 0.1913190633058548, "learning_rate": 5.912735623060572e-05, "loss": 0.0144, "step": 37560 }, { "grad_norm": 0.1836622804403305, "learning_rate": 5.910703436715217e-05, "loss": 0.0164, "step": 37570 }, { "grad_norm": 0.1614615023136139, "learning_rate": 5.908671094755552e-05, "loss": 0.0115, "step": 37580 }, { "grad_norm": 0.24567075073719025, "learning_rate": 5.906638597528851e-05, "loss": 0.0167, "step": 37590 }, { "grad_norm": 0.15841542184352875, "learning_rate": 5.9046059453824076e-05, "loss": 0.0127, "step": 37600 }, { "grad_norm": 0.12813493609428406, "learning_rate": 5.9025731386635505e-05, "loss": 0.0137, "step": 37610 }, { "grad_norm": 0.12665462493896484, "learning_rate": 5.900540177719629e-05, "loss": 0.0133, "step": 37620 }, { "grad_norm": 0.12471553683280945, "learning_rate": 5.898507062898021e-05, "loss": 0.0155, "step": 37630 }, { "grad_norm": 0.18152806162834167, "learning_rate": 5.8964737945461316e-05, "loss": 0.0159, "step": 37640 }, { "grad_norm": 0.17731927335262299, "learning_rate": 5.8944403730113885e-05, "loss": 0.0152, "step": 37650 }, { "grad_norm": 0.2507801651954651, "learning_rate": 5.892406798641248e-05, "loss": 0.0131, "step": 37660 }, { "grad_norm": 0.2174723595380783, "learning_rate": 5.890373071783193e-05, "loss": 0.0157, "step": 37670 }, { "grad_norm": 0.20373190939426422, "learning_rate": 5.888339192784732e-05, "loss": 0.0141, "step": 37680 }, { "grad_norm": 0.1976957470178604, "learning_rate": 5.8863051619934003e-05, "loss": 0.0132, "step": 37690 }, { "grad_norm": 0.20486676692962646, "learning_rate": 5.8842709797567554e-05, "loss": 0.0163, "step": 37700 }, { "grad_norm": 0.1699778437614441, "learning_rate": 5.8822366464223855e-05, "loss": 0.011, "step": 37710 }, { "grad_norm": 0.114106684923172, "learning_rate": 5.880202162337901e-05, "loss": 0.0158, "step": 37720 }, { "grad_norm": 0.30487290024757385, "learning_rate": 5.8781675278509405e-05, "loss": 0.0147, "step": 37730 }, { "grad_norm": 0.14777448773384094, "learning_rate": 5.8761327433091696e-05, "loss": 0.013, "step": 37740 }, { "grad_norm": 0.14980687201023102, "learning_rate": 5.874097809060275e-05, "loss": 0.0127, "step": 37750 }, { "grad_norm": 0.13368962705135345, "learning_rate": 5.87206272545197e-05, "loss": 0.0128, "step": 37760 }, { "grad_norm": 0.12962622940540314, "learning_rate": 5.8700274928319955e-05, "loss": 0.0115, "step": 37770 }, { "grad_norm": 0.18448995053768158, "learning_rate": 5.867992111548118e-05, "loss": 0.0092, "step": 37780 }, { "grad_norm": 0.14440114796161652, "learning_rate": 5.865956581948131e-05, "loss": 0.0141, "step": 37790 }, { "grad_norm": 0.16047799587249756, "learning_rate": 5.863920904379845e-05, "loss": 0.0113, "step": 37800 }, { "grad_norm": 0.18156583607196808, "learning_rate": 5.8618850791911064e-05, "loss": 0.0152, "step": 37810 }, { "grad_norm": 0.28128138184547424, "learning_rate": 5.859849106729779e-05, "loss": 0.0152, "step": 37820 }, { "grad_norm": 0.23961392045021057, "learning_rate": 5.857812987343758e-05, "loss": 0.0151, "step": 37830 }, { "grad_norm": 0.2444954514503479, "learning_rate": 5.855776721380957e-05, "loss": 0.0144, "step": 37840 }, { "grad_norm": 0.18796855211257935, "learning_rate": 5.8537403091893217e-05, "loss": 0.0108, "step": 37850 }, { "grad_norm": 0.26162850856781006, "learning_rate": 5.851703751116816e-05, "loss": 0.0126, "step": 37860 }, { "grad_norm": 0.17659494280815125, "learning_rate": 5.8496670475114336e-05, "loss": 0.0131, "step": 37870 }, { "grad_norm": 0.21328571438789368, "learning_rate": 5.84763019872119e-05, "loss": 0.0121, "step": 37880 }, { "grad_norm": 0.17047487199306488, "learning_rate": 5.845593205094131e-05, "loss": 0.011, "step": 37890 }, { "grad_norm": 0.15588156878948212, "learning_rate": 5.843556066978318e-05, "loss": 0.0143, "step": 37900 }, { "grad_norm": 0.22169384360313416, "learning_rate": 5.8415187847218455e-05, "loss": 0.0147, "step": 37910 }, { "grad_norm": 0.15602833032608032, "learning_rate": 5.839481358672827e-05, "loss": 0.0156, "step": 37920 }, { "grad_norm": 0.14080801606178284, "learning_rate": 5.837443789179407e-05, "loss": 0.017, "step": 37930 }, { "grad_norm": 0.13713322579860687, "learning_rate": 5.8354060765897445e-05, "loss": 0.0134, "step": 37940 }, { "grad_norm": 0.15207432210445404, "learning_rate": 5.8333682212520334e-05, "loss": 0.0118, "step": 37950 }, { "grad_norm": 0.16999536752700806, "learning_rate": 5.831330223514486e-05, "loss": 0.0159, "step": 37960 }, { "grad_norm": 0.20544493198394775, "learning_rate": 5.8292920837253396e-05, "loss": 0.014, "step": 37970 }, { "grad_norm": 0.22590669989585876, "learning_rate": 5.827253802232857e-05, "loss": 0.0125, "step": 37980 }, { "grad_norm": 0.21681106090545654, "learning_rate": 5.825215379385327e-05, "loss": 0.0137, "step": 37990 }, { "grad_norm": 0.2292666733264923, "learning_rate": 5.823176815531057e-05, "loss": 0.0114, "step": 38000 }, { "grad_norm": 0.21785122156143188, "learning_rate": 5.8211381110183826e-05, "loss": 0.0146, "step": 38010 }, { "grad_norm": 0.17374475300312042, "learning_rate": 5.8190992661956645e-05, "loss": 0.011, "step": 38020 }, { "grad_norm": 0.4210844039916992, "learning_rate": 5.817060281411284e-05, "loss": 0.0127, "step": 38030 }, { "grad_norm": 0.16983073949813843, "learning_rate": 5.815021157013647e-05, "loss": 0.0128, "step": 38040 }, { "grad_norm": 0.2239963412284851, "learning_rate": 5.8129818933511856e-05, "loss": 0.0111, "step": 38050 }, { "grad_norm": 0.18564337491989136, "learning_rate": 5.8109424907723544e-05, "loss": 0.0135, "step": 38060 }, { "grad_norm": 0.1799973100423813, "learning_rate": 5.80890294962563e-05, "loss": 0.0118, "step": 38070 }, { "grad_norm": 0.21824365854263306, "learning_rate": 5.806863270259515e-05, "loss": 0.0165, "step": 38080 }, { "grad_norm": 0.1440446972846985, "learning_rate": 5.804823453022536e-05, "loss": 0.0134, "step": 38090 }, { "grad_norm": 0.21998487412929535, "learning_rate": 5.80278349826324e-05, "loss": 0.0114, "step": 38100 }, { "grad_norm": 0.21544349193572998, "learning_rate": 5.8007434063302014e-05, "loss": 0.0152, "step": 38110 }, { "grad_norm": 0.19351552426815033, "learning_rate": 5.7987031775720136e-05, "loss": 0.0117, "step": 38120 }, { "grad_norm": 0.21551382541656494, "learning_rate": 5.7966628123372976e-05, "loss": 0.0122, "step": 38130 }, { "grad_norm": 0.14012926816940308, "learning_rate": 5.7946223109746956e-05, "loss": 0.0104, "step": 38140 }, { "grad_norm": 0.2752141058444977, "learning_rate": 5.7925816738328754e-05, "loss": 0.0135, "step": 38150 }, { "grad_norm": 0.184604212641716, "learning_rate": 5.790540901260521e-05, "loss": 0.0112, "step": 38160 }, { "grad_norm": 0.2210693210363388, "learning_rate": 5.788499993606351e-05, "loss": 0.0116, "step": 38170 }, { "grad_norm": 0.15424785017967224, "learning_rate": 5.786458951219096e-05, "loss": 0.0116, "step": 38180 }, { "grad_norm": 0.22170616686344147, "learning_rate": 5.784417774447517e-05, "loss": 0.0121, "step": 38190 }, { "grad_norm": 0.22511950135231018, "learning_rate": 5.782376463640393e-05, "loss": 0.0163, "step": 38200 }, { "grad_norm": 0.19300894439220428, "learning_rate": 5.780335019146531e-05, "loss": 0.018, "step": 38210 }, { "grad_norm": 0.16650491952896118, "learning_rate": 5.778293441314755e-05, "loss": 0.017, "step": 38220 }, { "grad_norm": 0.18716073036193848, "learning_rate": 5.776251730493917e-05, "loss": 0.0145, "step": 38230 }, { "grad_norm": 0.1441522091627121, "learning_rate": 5.774209887032887e-05, "loss": 0.0126, "step": 38240 }, { "grad_norm": 0.1833232194185257, "learning_rate": 5.772167911280565e-05, "loss": 0.0128, "step": 38250 }, { "grad_norm": 0.2022341638803482, "learning_rate": 5.770125803585864e-05, "loss": 0.016, "step": 38260 }, { "grad_norm": 0.16147200763225555, "learning_rate": 5.768083564297726e-05, "loss": 0.012, "step": 38270 }, { "grad_norm": 0.20423643290996552, "learning_rate": 5.766041193765114e-05, "loss": 0.0123, "step": 38280 }, { "grad_norm": 0.2302705943584442, "learning_rate": 5.763998692337015e-05, "loss": 0.0129, "step": 38290 }, { "grad_norm": 0.17931507527828217, "learning_rate": 5.761956060362433e-05, "loss": 0.0135, "step": 38300 }, { "grad_norm": 0.3000761568546295, "learning_rate": 5.7599132981904e-05, "loss": 0.015, "step": 38310 }, { "grad_norm": 0.16478300094604492, "learning_rate": 5.75787040616997e-05, "loss": 0.0108, "step": 38320 }, { "grad_norm": 0.13630224764347076, "learning_rate": 5.755827384650212e-05, "loss": 0.012, "step": 38330 }, { "grad_norm": 0.11740360409021378, "learning_rate": 5.753784233980228e-05, "loss": 0.0098, "step": 38340 }, { "grad_norm": 0.20976552367210388, "learning_rate": 5.751740954509135e-05, "loss": 0.0144, "step": 38350 }, { "grad_norm": 0.49355077743530273, "learning_rate": 5.7496975465860715e-05, "loss": 0.0127, "step": 38360 }, { "grad_norm": 0.26787570118904114, "learning_rate": 5.747654010560202e-05, "loss": 0.0161, "step": 38370 }, { "grad_norm": 0.22073189914226532, "learning_rate": 5.7456103467807097e-05, "loss": 0.0153, "step": 38380 }, { "grad_norm": 0.17009393870830536, "learning_rate": 5.7435665555968046e-05, "loss": 0.0175, "step": 38390 }, { "grad_norm": 0.14712901413440704, "learning_rate": 5.74152263735771e-05, "loss": 0.0138, "step": 38400 }, { "grad_norm": 0.20477280020713806, "learning_rate": 5.739478592412677e-05, "loss": 0.0127, "step": 38410 }, { "grad_norm": 0.16126640141010284, "learning_rate": 5.7374344211109766e-05, "loss": 0.0116, "step": 38420 }, { "grad_norm": 0.23259218037128448, "learning_rate": 5.735390123801904e-05, "loss": 0.0142, "step": 38430 }, { "grad_norm": 0.17671604454517365, "learning_rate": 5.7333457008347704e-05, "loss": 0.0145, "step": 38440 }, { "grad_norm": 0.1712162345647812, "learning_rate": 5.7313011525589156e-05, "loss": 0.0131, "step": 38450 }, { "grad_norm": 0.17671160399913788, "learning_rate": 5.729256479323694e-05, "loss": 0.011, "step": 38460 }, { "grad_norm": 0.1805780529975891, "learning_rate": 5.727211681478485e-05, "loss": 0.0135, "step": 38470 }, { "grad_norm": 0.22223412990570068, "learning_rate": 5.7251667593726886e-05, "loss": 0.0143, "step": 38480 }, { "grad_norm": 0.17624801397323608, "learning_rate": 5.723121713355728e-05, "loss": 0.0162, "step": 38490 }, { "grad_norm": 0.1767607033252716, "learning_rate": 5.721076543777044e-05, "loss": 0.0152, "step": 38500 }, { "grad_norm": 0.14726512134075165, "learning_rate": 5.7190312509860986e-05, "loss": 0.0121, "step": 38510 }, { "grad_norm": 0.15520188212394714, "learning_rate": 5.716985835332379e-05, "loss": 0.0141, "step": 38520 }, { "grad_norm": 0.1528530865907669, "learning_rate": 5.714940297165389e-05, "loss": 0.0101, "step": 38530 }, { "grad_norm": 0.16549842059612274, "learning_rate": 5.712894636834656e-05, "loss": 0.0118, "step": 38540 }, { "grad_norm": 0.2810786962509155, "learning_rate": 5.7108488546897276e-05, "loss": 0.0166, "step": 38550 }, { "grad_norm": 0.19904054701328278, "learning_rate": 5.708802951080172e-05, "loss": 0.0118, "step": 38560 }, { "grad_norm": 0.17655932903289795, "learning_rate": 5.706756926355576e-05, "loss": 0.014, "step": 38570 }, { "grad_norm": 0.16470953822135925, "learning_rate": 5.704710780865554e-05, "loss": 0.0136, "step": 38580 }, { "grad_norm": 0.16451941430568695, "learning_rate": 5.7026645149597325e-05, "loss": 0.013, "step": 38590 }, { "grad_norm": 0.158746138215065, "learning_rate": 5.700618128987764e-05, "loss": 0.0133, "step": 38600 }, { "grad_norm": 0.13545259833335876, "learning_rate": 5.698571623299317e-05, "loss": 0.0124, "step": 38610 }, { "grad_norm": 0.2058403193950653, "learning_rate": 5.696524998244086e-05, "loss": 0.0121, "step": 38620 }, { "grad_norm": 0.2558321952819824, "learning_rate": 5.6944782541717836e-05, "loss": 0.0155, "step": 38630 }, { "grad_norm": 0.2269628494977951, "learning_rate": 5.69243139143214e-05, "loss": 0.0142, "step": 38640 }, { "grad_norm": 0.18371012806892395, "learning_rate": 5.6903844103749125e-05, "loss": 0.0136, "step": 38650 }, { "grad_norm": 0.1729656606912613, "learning_rate": 5.688337311349869e-05, "loss": 0.0131, "step": 38660 }, { "grad_norm": 0.14401285350322723, "learning_rate": 5.6862900947068074e-05, "loss": 0.0156, "step": 38670 }, { "grad_norm": 0.1252593696117401, "learning_rate": 5.6842427607955374e-05, "loss": 0.0132, "step": 38680 }, { "grad_norm": 0.2027907371520996, "learning_rate": 5.682195309965893e-05, "loss": 0.0157, "step": 38690 }, { "grad_norm": 0.2253224402666092, "learning_rate": 5.6801477425677294e-05, "loss": 0.0127, "step": 38700 }, { "grad_norm": 0.2419627159833908, "learning_rate": 5.678100058950917e-05, "loss": 0.0133, "step": 38710 }, { "grad_norm": 0.13980580866336823, "learning_rate": 5.676052259465352e-05, "loss": 0.0119, "step": 38720 }, { "grad_norm": 0.23879839479923248, "learning_rate": 5.674004344460945e-05, "loss": 0.0174, "step": 38730 }, { "grad_norm": 0.20469625294208527, "learning_rate": 5.6719563142876295e-05, "loss": 0.0166, "step": 38740 }, { "grad_norm": 0.26421746611595154, "learning_rate": 5.669908169295359e-05, "loss": 0.0141, "step": 38750 }, { "grad_norm": 0.20727074146270752, "learning_rate": 5.667859909834105e-05, "loss": 0.0156, "step": 38760 }, { "grad_norm": 0.2134256511926651, "learning_rate": 5.6658115362538565e-05, "loss": 0.0156, "step": 38770 }, { "grad_norm": 0.21474730968475342, "learning_rate": 5.663763048904628e-05, "loss": 0.0151, "step": 38780 }, { "grad_norm": 0.1834181696176529, "learning_rate": 5.661714448136447e-05, "loss": 0.0174, "step": 38790 }, { "grad_norm": 0.2240874022245407, "learning_rate": 5.659665734299366e-05, "loss": 0.0127, "step": 38800 }, { "grad_norm": 0.2056601345539093, "learning_rate": 5.6576169077434516e-05, "loss": 0.0126, "step": 38810 }, { "grad_norm": 0.17341282963752747, "learning_rate": 5.6555679688187944e-05, "loss": 0.012, "step": 38820 }, { "grad_norm": 0.19790402054786682, "learning_rate": 5.6535189178755e-05, "loss": 0.0122, "step": 38830 }, { "grad_norm": 0.13855522871017456, "learning_rate": 5.651469755263698e-05, "loss": 0.0152, "step": 38840 }, { "grad_norm": 0.16341358423233032, "learning_rate": 5.6494204813335316e-05, "loss": 0.0129, "step": 38850 }, { "grad_norm": 0.24224188923835754, "learning_rate": 5.647371096435168e-05, "loss": 0.0126, "step": 38860 }, { "grad_norm": 0.13969288766384125, "learning_rate": 5.645321600918788e-05, "loss": 0.0156, "step": 38870 }, { "grad_norm": 0.18496698141098022, "learning_rate": 5.643271995134597e-05, "loss": 0.0172, "step": 38880 }, { "grad_norm": 0.15296220779418945, "learning_rate": 5.641222279432814e-05, "loss": 0.0147, "step": 38890 }, { "grad_norm": 0.11432040482759476, "learning_rate": 5.6391724541636834e-05, "loss": 0.0134, "step": 38900 }, { "grad_norm": 0.18241696059703827, "learning_rate": 5.6371225196774605e-05, "loss": 0.0132, "step": 38910 }, { "grad_norm": 0.18976706266403198, "learning_rate": 5.635072476324423e-05, "loss": 0.0117, "step": 38920 }, { "grad_norm": 0.17198146879673004, "learning_rate": 5.63302232445487e-05, "loss": 0.0116, "step": 38930 }, { "grad_norm": 0.17227701842784882, "learning_rate": 5.6309720644191144e-05, "loss": 0.0132, "step": 38940 }, { "grad_norm": 0.21399007737636566, "learning_rate": 5.628921696567491e-05, "loss": 0.011, "step": 38950 }, { "grad_norm": 0.16370701789855957, "learning_rate": 5.62687122125035e-05, "loss": 0.0121, "step": 38960 }, { "grad_norm": 0.1438024491071701, "learning_rate": 5.624820638818062e-05, "loss": 0.0107, "step": 38970 }, { "grad_norm": 0.15556326508522034, "learning_rate": 5.6227699496210164e-05, "loss": 0.0123, "step": 38980 }, { "grad_norm": 0.14909252524375916, "learning_rate": 5.6207191540096195e-05, "loss": 0.0118, "step": 38990 }, { "grad_norm": 0.1256008893251419, "learning_rate": 5.618668252334296e-05, "loss": 0.0104, "step": 39000 }, { "grad_norm": 0.18572933971881866, "learning_rate": 5.616617244945488e-05, "loss": 0.014, "step": 39010 }, { "grad_norm": 0.13628622889518738, "learning_rate": 5.614566132193656e-05, "loss": 0.0135, "step": 39020 }, { "grad_norm": 0.1747349351644516, "learning_rate": 5.612514914429282e-05, "loss": 0.0094, "step": 39030 }, { "grad_norm": 0.2169615775346756, "learning_rate": 5.610463592002863e-05, "loss": 0.0143, "step": 39040 }, { "grad_norm": 0.2612306773662567, "learning_rate": 5.608412165264909e-05, "loss": 0.0154, "step": 39050 }, { "grad_norm": 0.29271265864372253, "learning_rate": 5.606360634565959e-05, "loss": 0.013, "step": 39060 }, { "grad_norm": 0.19976861774921417, "learning_rate": 5.604309000256559e-05, "loss": 0.0117, "step": 39070 }, { "grad_norm": 0.13376374542713165, "learning_rate": 5.6022572626872785e-05, "loss": 0.0086, "step": 39080 }, { "grad_norm": 0.20637379586696625, "learning_rate": 5.600205422208704e-05, "loss": 0.0121, "step": 39090 }, { "grad_norm": 0.21071486175060272, "learning_rate": 5.59815347917144e-05, "loss": 0.0118, "step": 39100 }, { "grad_norm": 0.17405922710895538, "learning_rate": 5.596101433926103e-05, "loss": 0.0141, "step": 39110 }, { "grad_norm": 0.15011748671531677, "learning_rate": 5.5940492868233364e-05, "loss": 0.0113, "step": 39120 }, { "grad_norm": 0.2249561846256256, "learning_rate": 5.591997038213793e-05, "loss": 0.0108, "step": 39130 }, { "grad_norm": 0.17754799127578735, "learning_rate": 5.5899446884481475e-05, "loss": 0.015, "step": 39140 }, { "grad_norm": 0.18840090930461884, "learning_rate": 5.5878922378770906e-05, "loss": 0.0132, "step": 39150 }, { "grad_norm": 0.20750124752521515, "learning_rate": 5.5858396868513285e-05, "loss": 0.0138, "step": 39160 }, { "grad_norm": 0.21192118525505066, "learning_rate": 5.583787035721586e-05, "loss": 0.0122, "step": 39170 }, { "grad_norm": 0.15691125392913818, "learning_rate": 5.581734284838606e-05, "loss": 0.0103, "step": 39180 }, { "grad_norm": 0.1796252578496933, "learning_rate": 5.579681434553147e-05, "loss": 0.0127, "step": 39190 }, { "grad_norm": 0.2156338393688202, "learning_rate": 5.5776284852159854e-05, "loss": 0.0145, "step": 39200 }, { "grad_norm": 0.18071135878562927, "learning_rate": 5.575575437177913e-05, "loss": 0.0149, "step": 39210 }, { "grad_norm": 0.18541833758354187, "learning_rate": 5.573522290789742e-05, "loss": 0.0158, "step": 39220 }, { "grad_norm": 0.14362649619579315, "learning_rate": 5.571469046402298e-05, "loss": 0.0116, "step": 39230 }, { "grad_norm": 0.204864040017128, "learning_rate": 5.5694157043664205e-05, "loss": 0.0127, "step": 39240 }, { "grad_norm": 0.2027636170387268, "learning_rate": 5.567362265032975e-05, "loss": 0.0138, "step": 39250 }, { "grad_norm": 0.13955917954444885, "learning_rate": 5.565308728752836e-05, "loss": 0.0133, "step": 39260 }, { "grad_norm": 0.19070492684841156, "learning_rate": 5.5632550958768956e-05, "loss": 0.0124, "step": 39270 }, { "grad_norm": 0.22190320491790771, "learning_rate": 5.5612013667560636e-05, "loss": 0.0122, "step": 39280 }, { "grad_norm": 0.2238008826971054, "learning_rate": 5.5591475417412676e-05, "loss": 0.0142, "step": 39290 }, { "grad_norm": 0.17376762628555298, "learning_rate": 5.557093621183451e-05, "loss": 0.0109, "step": 39300 }, { "grad_norm": 0.1639731526374817, "learning_rate": 5.55503960543357e-05, "loss": 0.0137, "step": 39310 }, { "grad_norm": 0.20526640117168427, "learning_rate": 5.552985494842601e-05, "loss": 0.0153, "step": 39320 }, { "grad_norm": 0.1834075152873993, "learning_rate": 5.550931289761534e-05, "loss": 0.0102, "step": 39330 }, { "grad_norm": 0.15771928429603577, "learning_rate": 5.548876990541378e-05, "loss": 0.0155, "step": 39340 }, { "grad_norm": 0.16987408697605133, "learning_rate": 5.5468225975331564e-05, "loss": 0.013, "step": 39350 }, { "grad_norm": 0.1740271896123886, "learning_rate": 5.544768111087909e-05, "loss": 0.0134, "step": 39360 }, { "grad_norm": 0.22146093845367432, "learning_rate": 5.54271353155669e-05, "loss": 0.0102, "step": 39370 }, { "grad_norm": 0.14577235281467438, "learning_rate": 5.5406588592905715e-05, "loss": 0.0118, "step": 39380 }, { "grad_norm": 0.16551904380321503, "learning_rate": 5.5386040946406416e-05, "loss": 0.0114, "step": 39390 }, { "grad_norm": 0.1852780282497406, "learning_rate": 5.536549237958004e-05, "loss": 0.0145, "step": 39400 }, { "grad_norm": 0.1457424908876419, "learning_rate": 5.5344942895937744e-05, "loss": 0.0101, "step": 39410 }, { "grad_norm": 0.1603831946849823, "learning_rate": 5.5324392498990904e-05, "loss": 0.0109, "step": 39420 }, { "grad_norm": 0.1543601006269455, "learning_rate": 5.5303841192251e-05, "loss": 0.0114, "step": 39430 }, { "grad_norm": 0.1998203545808792, "learning_rate": 5.52832889792297e-05, "loss": 0.0136, "step": 39440 }, { "grad_norm": 0.13369929790496826, "learning_rate": 5.526273586343881e-05, "loss": 0.0129, "step": 39450 }, { "grad_norm": 0.1485539972782135, "learning_rate": 5.5242181848390306e-05, "loss": 0.0125, "step": 39460 }, { "grad_norm": 0.36225542426109314, "learning_rate": 5.5221626937596285e-05, "loss": 0.0145, "step": 39470 }, { "grad_norm": 0.1343269646167755, "learning_rate": 5.520107113456903e-05, "loss": 0.0111, "step": 39480 }, { "grad_norm": 0.43724769353866577, "learning_rate": 5.5180514442820974e-05, "loss": 0.0126, "step": 39490 }, { "grad_norm": 0.19720453023910522, "learning_rate": 5.515995686586469e-05, "loss": 0.0121, "step": 39500 }, { "grad_norm": 0.16932810842990875, "learning_rate": 5.5139398407212916e-05, "loss": 0.0152, "step": 39510 }, { "grad_norm": 0.11405874043703079, "learning_rate": 5.511883907037849e-05, "loss": 0.0167, "step": 39520 }, { "grad_norm": 0.24576795101165771, "learning_rate": 5.509827885887449e-05, "loss": 0.0154, "step": 39530 }, { "grad_norm": 0.22630496323108673, "learning_rate": 5.507771777621406e-05, "loss": 0.0118, "step": 39540 }, { "grad_norm": 0.17615428566932678, "learning_rate": 5.505715582591052e-05, "loss": 0.0131, "step": 39550 }, { "grad_norm": 0.225462406873703, "learning_rate": 5.50365930114774e-05, "loss": 0.011, "step": 39560 }, { "grad_norm": 0.15719644725322723, "learning_rate": 5.5016029336428255e-05, "loss": 0.015, "step": 39570 }, { "grad_norm": 0.19241619110107422, "learning_rate": 5.49954648042769e-05, "loss": 0.0104, "step": 39580 }, { "grad_norm": 0.14363877475261688, "learning_rate": 5.4974899418537226e-05, "loss": 0.0161, "step": 39590 }, { "grad_norm": 0.20817521214485168, "learning_rate": 5.4954333182723316e-05, "loss": 0.0114, "step": 39600 }, { "grad_norm": 0.1947782039642334, "learning_rate": 5.493376610034937e-05, "loss": 0.0146, "step": 39610 }, { "grad_norm": 0.11575733870267868, "learning_rate": 5.4913198174929735e-05, "loss": 0.0106, "step": 39620 }, { "grad_norm": 0.21094834804534912, "learning_rate": 5.489262940997889e-05, "loss": 0.0137, "step": 39630 }, { "grad_norm": 0.1911359578371048, "learning_rate": 5.487205980901151e-05, "loss": 0.0111, "step": 39640 }, { "grad_norm": 0.14189456403255463, "learning_rate": 5.485148937554234e-05, "loss": 0.0134, "step": 39650 }, { "grad_norm": 0.15136246383190155, "learning_rate": 5.483091811308635e-05, "loss": 0.0122, "step": 39660 }, { "grad_norm": 0.19117914140224457, "learning_rate": 5.4810346025158564e-05, "loss": 0.0114, "step": 39670 }, { "grad_norm": 0.10063988715410233, "learning_rate": 5.478977311527421e-05, "loss": 0.0107, "step": 39680 }, { "grad_norm": 0.16225682199001312, "learning_rate": 5.476919938694863e-05, "loss": 0.0141, "step": 39690 }, { "grad_norm": 0.23507949709892273, "learning_rate": 5.474862484369733e-05, "loss": 0.0133, "step": 39700 }, { "grad_norm": 0.21459835767745972, "learning_rate": 5.472804948903589e-05, "loss": 0.0138, "step": 39710 }, { "grad_norm": 0.17942525446414948, "learning_rate": 5.470747332648013e-05, "loss": 0.0127, "step": 39720 }, { "grad_norm": 0.22649016976356506, "learning_rate": 5.468689635954591e-05, "loss": 0.0148, "step": 39730 }, { "grad_norm": 0.17261412739753723, "learning_rate": 5.46663185917493e-05, "loss": 0.0151, "step": 39740 }, { "grad_norm": 0.1494942456483841, "learning_rate": 5.464574002660645e-05, "loss": 0.0132, "step": 39750 }, { "grad_norm": 0.19791395962238312, "learning_rate": 5.4625160667633724e-05, "loss": 0.0123, "step": 39760 }, { "grad_norm": 0.17037279903888702, "learning_rate": 5.4604580518347505e-05, "loss": 0.0115, "step": 39770 }, { "grad_norm": 0.16883249580860138, "learning_rate": 5.458399958226442e-05, "loss": 0.0122, "step": 39780 }, { "grad_norm": 0.2094542682170868, "learning_rate": 5.456341786290119e-05, "loss": 0.0166, "step": 39790 }, { "grad_norm": 0.18315985798835754, "learning_rate": 5.454283536377465e-05, "loss": 0.0101, "step": 39800 }, { "grad_norm": 0.19992689788341522, "learning_rate": 5.452225208840179e-05, "loss": 0.0169, "step": 39810 }, { "grad_norm": 0.20067019760608673, "learning_rate": 5.450166804029975e-05, "loss": 0.0122, "step": 39820 }, { "grad_norm": 0.17906542122364044, "learning_rate": 5.448108322298574e-05, "loss": 0.0126, "step": 39830 }, { "grad_norm": 0.16494129598140717, "learning_rate": 5.446049763997717e-05, "loss": 0.0149, "step": 39840 }, { "grad_norm": 0.26363322138786316, "learning_rate": 5.4439911294791546e-05, "loss": 0.0149, "step": 39850 }, { "grad_norm": 0.29292869567871094, "learning_rate": 5.441932419094652e-05, "loss": 0.0174, "step": 39860 }, { "grad_norm": 0.19282887876033783, "learning_rate": 5.439873633195985e-05, "loss": 0.0121, "step": 39870 }, { "grad_norm": 0.17435045540332794, "learning_rate": 5.437814772134947e-05, "loss": 0.0109, "step": 39880 }, { "grad_norm": 0.15466398000717163, "learning_rate": 5.4357558362633366e-05, "loss": 0.0117, "step": 39890 }, { "grad_norm": 0.13881441950798035, "learning_rate": 5.4336968259329726e-05, "loss": 0.0113, "step": 39900 }, { "grad_norm": 0.1254972368478775, "learning_rate": 5.431637741495681e-05, "loss": 0.0151, "step": 39910 }, { "grad_norm": 0.19700796902179718, "learning_rate": 5.429578583303307e-05, "loss": 0.0139, "step": 39920 }, { "grad_norm": 0.17510615289211273, "learning_rate": 5.427519351707701e-05, "loss": 0.0098, "step": 39930 }, { "grad_norm": 0.2261601984500885, "learning_rate": 5.42546004706073e-05, "loss": 0.0132, "step": 39940 }, { "grad_norm": 0.14550161361694336, "learning_rate": 5.4234006697142735e-05, "loss": 0.0123, "step": 39950 }, { "grad_norm": 0.19392383098602295, "learning_rate": 5.421341220020224e-05, "loss": 0.0104, "step": 39960 }, { "grad_norm": 0.15600305795669556, "learning_rate": 5.419281698330482e-05, "loss": 0.0105, "step": 39970 }, { "grad_norm": 0.1360727697610855, "learning_rate": 5.4172221049969665e-05, "loss": 0.0111, "step": 39980 }, { "grad_norm": 0.17522011697292328, "learning_rate": 5.415162440371604e-05, "loss": 0.0113, "step": 39990 }, { "grad_norm": 0.16880866885185242, "learning_rate": 5.413102704806334e-05, "loss": 0.0118, "step": 40000 }, { "grad_norm": 0.18081897497177124, "learning_rate": 5.41104289865311e-05, "loss": 0.0104, "step": 40010 }, { "grad_norm": 0.12360014021396637, "learning_rate": 5.408983022263898e-05, "loss": 0.011, "step": 40020 }, { "grad_norm": 0.20894934237003326, "learning_rate": 5.406923075990673e-05, "loss": 0.0106, "step": 40030 }, { "grad_norm": 0.1723625212907791, "learning_rate": 5.404863060185423e-05, "loss": 0.0104, "step": 40040 }, { "grad_norm": 0.16870278120040894, "learning_rate": 5.40280297520015e-05, "loss": 0.013, "step": 40050 }, { "grad_norm": 0.17256584763526917, "learning_rate": 5.400742821386865e-05, "loss": 0.0135, "step": 40060 }, { "grad_norm": 0.16118375957012177, "learning_rate": 5.398682599097592e-05, "loss": 0.0124, "step": 40070 }, { "grad_norm": 0.24217572808265686, "learning_rate": 5.396622308684367e-05, "loss": 0.0122, "step": 40080 }, { "grad_norm": 0.16121645271778107, "learning_rate": 5.394561950499236e-05, "loss": 0.0107, "step": 40090 }, { "grad_norm": 0.13770568370819092, "learning_rate": 5.39250152489426e-05, "loss": 0.011, "step": 40100 }, { "grad_norm": 0.1476105898618698, "learning_rate": 5.390441032221507e-05, "loss": 0.011, "step": 40110 }, { "grad_norm": 0.15617704391479492, "learning_rate": 5.388380472833062e-05, "loss": 0.0133, "step": 40120 }, { "grad_norm": 0.230956569314003, "learning_rate": 5.386319847081016e-05, "loss": 0.011, "step": 40130 }, { "grad_norm": 0.200876384973526, "learning_rate": 5.384259155317473e-05, "loss": 0.0143, "step": 40140 }, { "grad_norm": 0.20536573231220245, "learning_rate": 5.38219839789455e-05, "loss": 0.012, "step": 40150 }, { "grad_norm": 0.16422487795352936, "learning_rate": 5.380137575164376e-05, "loss": 0.0116, "step": 40160 }, { "grad_norm": 0.19475649297237396, "learning_rate": 5.378076687479085e-05, "loss": 0.0132, "step": 40170 }, { "grad_norm": 0.18454626202583313, "learning_rate": 5.3760157351908305e-05, "loss": 0.0101, "step": 40180 }, { "grad_norm": 0.15269090235233307, "learning_rate": 5.373954718651768e-05, "loss": 0.0119, "step": 40190 }, { "grad_norm": 0.2673567235469818, "learning_rate": 5.371893638214074e-05, "loss": 0.0128, "step": 40200 }, { "grad_norm": 0.18000224232673645, "learning_rate": 5.369832494229927e-05, "loss": 0.0108, "step": 40210 }, { "grad_norm": 0.18481436371803284, "learning_rate": 5.367771287051524e-05, "loss": 0.015, "step": 40220 }, { "grad_norm": 0.21048568189144135, "learning_rate": 5.365710017031065e-05, "loss": 0.0109, "step": 40230 }, { "grad_norm": 0.17131485044956207, "learning_rate": 5.3636486845207654e-05, "loss": 0.0121, "step": 40240 }, { "grad_norm": 0.17767801880836487, "learning_rate": 5.3615872898728526e-05, "loss": 0.0133, "step": 40250 }, { "grad_norm": 0.16123118996620178, "learning_rate": 5.3595258334395614e-05, "loss": 0.0137, "step": 40260 }, { "grad_norm": 0.14955998957157135, "learning_rate": 5.3574643155731396e-05, "loss": 0.0128, "step": 40270 }, { "grad_norm": 0.5318357944488525, "learning_rate": 5.3554027366258405e-05, "loss": 0.0106, "step": 40280 }, { "grad_norm": 0.17828331887722015, "learning_rate": 5.3533410969499355e-05, "loss": 0.0123, "step": 40290 }, { "grad_norm": 0.1850908249616623, "learning_rate": 5.3512793968977e-05, "loss": 0.0136, "step": 40300 }, { "grad_norm": 0.23940353095531464, "learning_rate": 5.349217636821423e-05, "loss": 0.0124, "step": 40310 }, { "grad_norm": 0.17144176363945007, "learning_rate": 5.347155817073405e-05, "loss": 0.0146, "step": 40320 }, { "grad_norm": 0.18070578575134277, "learning_rate": 5.34509393800595e-05, "loss": 0.0124, "step": 40330 }, { "grad_norm": 0.16275887191295624, "learning_rate": 5.343031999971381e-05, "loss": 0.015, "step": 40340 }, { "grad_norm": 0.15844367444515228, "learning_rate": 5.340970003322026e-05, "loss": 0.0133, "step": 40350 }, { "grad_norm": 0.12623819708824158, "learning_rate": 5.33890794841022e-05, "loss": 0.0113, "step": 40360 }, { "grad_norm": 0.20136259496212006, "learning_rate": 5.336845835588318e-05, "loss": 0.0148, "step": 40370 }, { "grad_norm": 0.16542014479637146, "learning_rate": 5.334783665208674e-05, "loss": 0.0117, "step": 40380 }, { "grad_norm": 0.2106655240058899, "learning_rate": 5.332721437623657e-05, "loss": 0.0136, "step": 40390 }, { "grad_norm": 0.16039922833442688, "learning_rate": 5.3306591531856464e-05, "loss": 0.0161, "step": 40400 }, { "grad_norm": 0.16130352020263672, "learning_rate": 5.3285968122470295e-05, "loss": 0.0173, "step": 40410 }, { "grad_norm": 0.2108146697282791, "learning_rate": 5.326534415160207e-05, "loss": 0.0195, "step": 40420 }, { "grad_norm": 0.17417806386947632, "learning_rate": 5.324471962277582e-05, "loss": 0.0136, "step": 40430 }, { "grad_norm": 0.2083776891231537, "learning_rate": 5.3224094539515746e-05, "loss": 0.0153, "step": 40440 }, { "grad_norm": 0.22863101959228516, "learning_rate": 5.3203468905346076e-05, "loss": 0.0134, "step": 40450 }, { "grad_norm": 0.20420533418655396, "learning_rate": 5.3182842723791195e-05, "loss": 0.0139, "step": 40460 }, { "grad_norm": 0.25242987275123596, "learning_rate": 5.316221599837554e-05, "loss": 0.0137, "step": 40470 }, { "grad_norm": 0.14970631897449493, "learning_rate": 5.314158873262366e-05, "loss": 0.0137, "step": 40480 }, { "grad_norm": 0.2328680157661438, "learning_rate": 5.312096093006018e-05, "loss": 0.0143, "step": 40490 }, { "grad_norm": 0.21375033259391785, "learning_rate": 5.3100332594209846e-05, "loss": 0.0128, "step": 40500 }, { "grad_norm": 0.14969304203987122, "learning_rate": 5.3079703728597454e-05, "loss": 0.0145, "step": 40510 }, { "grad_norm": 0.35040196776390076, "learning_rate": 5.305907433674794e-05, "loss": 0.0146, "step": 40520 }, { "grad_norm": 0.17089875042438507, "learning_rate": 5.3038444422186284e-05, "loss": 0.0127, "step": 40530 }, { "grad_norm": 0.2093709409236908, "learning_rate": 5.301781398843757e-05, "loss": 0.0124, "step": 40540 }, { "grad_norm": 0.18037569522857666, "learning_rate": 5.299718303902699e-05, "loss": 0.0112, "step": 40550 }, { "grad_norm": 0.17965225875377655, "learning_rate": 5.297655157747979e-05, "loss": 0.0118, "step": 40560 }, { "grad_norm": 0.1937979757785797, "learning_rate": 5.295591960732136e-05, "loss": 0.0128, "step": 40570 }, { "grad_norm": 0.18418040871620178, "learning_rate": 5.293528713207708e-05, "loss": 0.0153, "step": 40580 }, { "grad_norm": 0.2072903960943222, "learning_rate": 5.291465415527253e-05, "loss": 0.0143, "step": 40590 }, { "grad_norm": 0.17653338611125946, "learning_rate": 5.2894020680433296e-05, "loss": 0.0128, "step": 40600 }, { "grad_norm": 0.17762544751167297, "learning_rate": 5.287338671108507e-05, "loss": 0.0105, "step": 40610 }, { "grad_norm": 0.18640154600143433, "learning_rate": 5.285275225075367e-05, "loss": 0.0131, "step": 40620 }, { "grad_norm": 0.177922323346138, "learning_rate": 5.283211730296492e-05, "loss": 0.0113, "step": 40630 }, { "grad_norm": 0.18860948085784912, "learning_rate": 5.281148187124477e-05, "loss": 0.0111, "step": 40640 }, { "grad_norm": 0.1541023850440979, "learning_rate": 5.279084595911927e-05, "loss": 0.0119, "step": 40650 }, { "grad_norm": 0.141897514462471, "learning_rate": 5.277020957011453e-05, "loss": 0.0099, "step": 40660 }, { "grad_norm": 0.1670914888381958, "learning_rate": 5.274957270775673e-05, "loss": 0.0119, "step": 40670 }, { "grad_norm": 0.15841297805309296, "learning_rate": 5.2728935375572164e-05, "loss": 0.0108, "step": 40680 }, { "grad_norm": 0.3181303143501282, "learning_rate": 5.2708297577087176e-05, "loss": 0.0145, "step": 40690 }, { "grad_norm": 0.13774502277374268, "learning_rate": 5.2687659315828196e-05, "loss": 0.0127, "step": 40700 }, { "grad_norm": 0.1810792088508606, "learning_rate": 5.266702059532175e-05, "loss": 0.012, "step": 40710 }, { "grad_norm": 0.19020502269268036, "learning_rate": 5.264638141909444e-05, "loss": 0.0139, "step": 40720 }, { "grad_norm": 0.25760194659233093, "learning_rate": 5.2625741790672925e-05, "loss": 0.0135, "step": 40730 }, { "grad_norm": 0.1749948412179947, "learning_rate": 5.2605101713583935e-05, "loss": 0.015, "step": 40740 }, { "grad_norm": 0.19287821650505066, "learning_rate": 5.2584461191354315e-05, "loss": 0.0159, "step": 40750 }, { "grad_norm": 0.2441674768924713, "learning_rate": 5.2563820227510964e-05, "loss": 0.0153, "step": 40760 }, { "grad_norm": 0.13890956342220306, "learning_rate": 5.2543178825580864e-05, "loss": 0.012, "step": 40770 }, { "grad_norm": 0.173804372549057, "learning_rate": 5.2522536989091054e-05, "loss": 0.0149, "step": 40780 }, { "grad_norm": 0.17987067997455597, "learning_rate": 5.2501894721568655e-05, "loss": 0.0139, "step": 40790 }, { "grad_norm": 0.2088792622089386, "learning_rate": 5.248125202654089e-05, "loss": 0.0132, "step": 40800 }, { "grad_norm": 0.14948488771915436, "learning_rate": 5.246060890753501e-05, "loss": 0.0102, "step": 40810 }, { "grad_norm": 0.19316907227039337, "learning_rate": 5.243996536807837e-05, "loss": 0.0098, "step": 40820 }, { "grad_norm": 0.18477754294872284, "learning_rate": 5.2419321411698384e-05, "loss": 0.0149, "step": 40830 }, { "grad_norm": 0.20739945769309998, "learning_rate": 5.239867704192253e-05, "loss": 0.0115, "step": 40840 }, { "grad_norm": 0.17183460295200348, "learning_rate": 5.237803226227838e-05, "loss": 0.0107, "step": 40850 }, { "grad_norm": 0.14377377927303314, "learning_rate": 5.235738707629354e-05, "loss": 0.0102, "step": 40860 }, { "grad_norm": 0.2175450623035431, "learning_rate": 5.233674148749575e-05, "loss": 0.0122, "step": 40870 }, { "grad_norm": 0.2273375689983368, "learning_rate": 5.231609549941272e-05, "loss": 0.0123, "step": 40880 }, { "grad_norm": 0.23370897769927979, "learning_rate": 5.229544911557233e-05, "loss": 0.0124, "step": 40890 }, { "grad_norm": 0.27217793464660645, "learning_rate": 5.227480233950246e-05, "loss": 0.0152, "step": 40900 }, { "grad_norm": 0.15927909314632416, "learning_rate": 5.22541551747311e-05, "loss": 0.0114, "step": 40910 }, { "grad_norm": 0.1275848001241684, "learning_rate": 5.223350762478626e-05, "loss": 0.0127, "step": 40920 }, { "grad_norm": 0.13414660096168518, "learning_rate": 5.221285969319606e-05, "loss": 0.0106, "step": 40930 }, { "grad_norm": 0.10892516374588013, "learning_rate": 5.219221138348865e-05, "loss": 0.0092, "step": 40940 }, { "grad_norm": 0.157364159822464, "learning_rate": 5.217156269919228e-05, "loss": 0.0134, "step": 40950 }, { "grad_norm": 0.18185682594776154, "learning_rate": 5.215091364383523e-05, "loss": 0.0115, "step": 40960 }, { "grad_norm": 0.2555910050868988, "learning_rate": 5.213026422094588e-05, "loss": 0.0127, "step": 40970 }, { "grad_norm": 0.1647329479455948, "learning_rate": 5.210961443405262e-05, "loss": 0.011, "step": 40980 }, { "grad_norm": 0.3435846269130707, "learning_rate": 5.208896428668396e-05, "loss": 0.0127, "step": 40990 }, { "grad_norm": 0.19754528999328613, "learning_rate": 5.206831378236845e-05, "loss": 0.01, "step": 41000 }, { "grad_norm": 0.24182939529418945, "learning_rate": 5.2047662924634666e-05, "loss": 0.0141, "step": 41010 }, { "grad_norm": 0.14328384399414062, "learning_rate": 5.20270117170113e-05, "loss": 0.0112, "step": 41020 }, { "grad_norm": 0.16212378442287445, "learning_rate": 5.200636016302707e-05, "loss": 0.0091, "step": 41030 }, { "grad_norm": 0.11485399305820465, "learning_rate": 5.1985708266210754e-05, "loss": 0.0098, "step": 41040 }, { "grad_norm": 0.16171686351299286, "learning_rate": 5.1965056030091206e-05, "loss": 0.0106, "step": 41050 }, { "grad_norm": 0.15856961905956268, "learning_rate": 5.194440345819732e-05, "loss": 0.0112, "step": 41060 }, { "grad_norm": 0.229489266872406, "learning_rate": 5.1923750554058084e-05, "loss": 0.0131, "step": 41070 }, { "grad_norm": 0.17089709639549255, "learning_rate": 5.1903097321202476e-05, "loss": 0.01, "step": 41080 }, { "grad_norm": 0.2009371519088745, "learning_rate": 5.18824437631596e-05, "loss": 0.0096, "step": 41090 }, { "grad_norm": 0.2013687789440155, "learning_rate": 5.186178988345856e-05, "loss": 0.0106, "step": 41100 }, { "grad_norm": 0.1994268298149109, "learning_rate": 5.184113568562855e-05, "loss": 0.0101, "step": 41110 }, { "grad_norm": 0.14378172159194946, "learning_rate": 5.18204811731988e-05, "loss": 0.0131, "step": 41120 }, { "grad_norm": 0.14852888882160187, "learning_rate": 5.179982634969862e-05, "loss": 0.0112, "step": 41130 }, { "grad_norm": 0.14585420489311218, "learning_rate": 5.177917121865734e-05, "loss": 0.0109, "step": 41140 }, { "grad_norm": 0.1280406415462494, "learning_rate": 5.1758515783604346e-05, "loss": 0.0127, "step": 41150 }, { "grad_norm": 0.18436980247497559, "learning_rate": 5.173786004806911e-05, "loss": 0.0129, "step": 41160 }, { "grad_norm": 0.2039393037557602, "learning_rate": 5.1717204015581135e-05, "loss": 0.0101, "step": 41170 }, { "grad_norm": 0.1613379865884781, "learning_rate": 5.169654768966994e-05, "loss": 0.0121, "step": 41180 }, { "grad_norm": 0.14320193231105804, "learning_rate": 5.1675891073865156e-05, "loss": 0.0095, "step": 41190 }, { "grad_norm": 0.14660564064979553, "learning_rate": 5.1655234171696424e-05, "loss": 0.0123, "step": 41200 }, { "grad_norm": 0.20613227784633636, "learning_rate": 5.163457698669343e-05, "loss": 0.0149, "step": 41210 }, { "grad_norm": 0.17009557783603668, "learning_rate": 5.1613919522385936e-05, "loss": 0.0116, "step": 41220 }, { "grad_norm": 0.1421186625957489, "learning_rate": 5.1593261782303746e-05, "loss": 0.0113, "step": 41230 }, { "grad_norm": 0.18627816438674927, "learning_rate": 5.157260376997669e-05, "loss": 0.0104, "step": 41240 }, { "grad_norm": 0.21978361904621124, "learning_rate": 5.155194548893466e-05, "loss": 0.0138, "step": 41250 }, { "grad_norm": 0.17583049833774567, "learning_rate": 5.153128694270759e-05, "loss": 0.0104, "step": 41260 }, { "grad_norm": 0.1789819598197937, "learning_rate": 5.151062813482548e-05, "loss": 0.0089, "step": 41270 }, { "grad_norm": 0.17514441907405853, "learning_rate": 5.148996906881832e-05, "loss": 0.0104, "step": 41280 }, { "grad_norm": 0.24080395698547363, "learning_rate": 5.1469309748216196e-05, "loss": 0.0109, "step": 41290 }, { "grad_norm": 0.15632016956806183, "learning_rate": 5.144865017654923e-05, "loss": 0.0126, "step": 41300 }, { "grad_norm": 0.276393324136734, "learning_rate": 5.1427990357347556e-05, "loss": 0.013, "step": 41310 }, { "grad_norm": 0.1870569884777069, "learning_rate": 5.140733029414139e-05, "loss": 0.0115, "step": 41320 }, { "grad_norm": 0.16165314614772797, "learning_rate": 5.138666999046098e-05, "loss": 0.0117, "step": 41330 }, { "grad_norm": 0.18343029916286469, "learning_rate": 5.136600944983658e-05, "loss": 0.01, "step": 41340 }, { "grad_norm": 0.17439696192741394, "learning_rate": 5.134534867579853e-05, "loss": 0.0117, "step": 41350 }, { "grad_norm": 0.21804161369800568, "learning_rate": 5.1324687671877184e-05, "loss": 0.0124, "step": 41360 }, { "grad_norm": 0.1677095890045166, "learning_rate": 5.130402644160296e-05, "loss": 0.0128, "step": 41370 }, { "grad_norm": 0.194444477558136, "learning_rate": 5.128336498850628e-05, "loss": 0.0116, "step": 41380 }, { "grad_norm": 0.13317766785621643, "learning_rate": 5.126270331611761e-05, "loss": 0.0116, "step": 41390 }, { "grad_norm": 0.1479637771844864, "learning_rate": 5.124204142796748e-05, "loss": 0.0107, "step": 41400 }, { "grad_norm": 0.1480574905872345, "learning_rate": 5.122137932758644e-05, "loss": 0.0127, "step": 41410 }, { "grad_norm": 0.1593521237373352, "learning_rate": 5.1200717018505074e-05, "loss": 0.0112, "step": 41420 }, { "grad_norm": 0.16832007467746735, "learning_rate": 5.118005450425403e-05, "loss": 0.0134, "step": 41430 }, { "grad_norm": 0.08921031653881073, "learning_rate": 5.115939178836392e-05, "loss": 0.0114, "step": 41440 }, { "grad_norm": 0.16614200174808502, "learning_rate": 5.113872887436547e-05, "loss": 0.0114, "step": 41450 }, { "grad_norm": 0.1343393325805664, "learning_rate": 5.11180657657894e-05, "loss": 0.0095, "step": 41460 }, { "grad_norm": 0.18929187953472137, "learning_rate": 5.109740246616649e-05, "loss": 0.0107, "step": 41470 }, { "grad_norm": 0.15272529423236847, "learning_rate": 5.10767389790275e-05, "loss": 0.0087, "step": 41480 }, { "grad_norm": 0.13808126747608185, "learning_rate": 5.1056075307903265e-05, "loss": 0.0096, "step": 41490 }, { "grad_norm": 0.15302196145057678, "learning_rate": 5.1035411456324634e-05, "loss": 0.0102, "step": 41500 }, { "grad_norm": 0.153005450963974, "learning_rate": 5.1014747427822504e-05, "loss": 0.0084, "step": 41510 }, { "grad_norm": 0.14314502477645874, "learning_rate": 5.0994083225927804e-05, "loss": 0.0105, "step": 41520 }, { "grad_norm": 0.17033523321151733, "learning_rate": 5.0973418854171475e-05, "loss": 0.0106, "step": 41530 }, { "grad_norm": 0.17368771135807037, "learning_rate": 5.095275431608447e-05, "loss": 0.0104, "step": 41540 }, { "grad_norm": 0.25115513801574707, "learning_rate": 5.093208961519782e-05, "loss": 0.0143, "step": 41550 }, { "grad_norm": 0.13716283440589905, "learning_rate": 5.091142475504255e-05, "loss": 0.0103, "step": 41560 }, { "grad_norm": 0.19641181826591492, "learning_rate": 5.089075973914971e-05, "loss": 0.0118, "step": 41570 }, { "grad_norm": 0.148319810628891, "learning_rate": 5.08700945710504e-05, "loss": 0.0097, "step": 41580 }, { "grad_norm": 0.14517278969287872, "learning_rate": 5.0849429254275714e-05, "loss": 0.0099, "step": 41590 }, { "grad_norm": 0.1463831663131714, "learning_rate": 5.0828763792356804e-05, "loss": 0.0119, "step": 41600 }, { "grad_norm": 0.14904586970806122, "learning_rate": 5.080809818882483e-05, "loss": 0.0092, "step": 41610 }, { "grad_norm": 0.17540811002254486, "learning_rate": 5.078743244721097e-05, "loss": 0.0161, "step": 41620 }, { "grad_norm": 0.20095545053482056, "learning_rate": 5.0766766571046455e-05, "loss": 0.0123, "step": 41630 }, { "grad_norm": 0.18904076516628265, "learning_rate": 5.07461005638625e-05, "loss": 0.0129, "step": 41640 }, { "grad_norm": 0.21573983132839203, "learning_rate": 5.072543442919037e-05, "loss": 0.0134, "step": 41650 }, { "grad_norm": 0.15055367350578308, "learning_rate": 5.070476817056132e-05, "loss": 0.0107, "step": 41660 }, { "grad_norm": 0.16264040768146515, "learning_rate": 5.068410179150668e-05, "loss": 0.0115, "step": 41670 }, { "grad_norm": 0.19065168499946594, "learning_rate": 5.066343529555775e-05, "loss": 0.0103, "step": 41680 }, { "grad_norm": 0.1755211055278778, "learning_rate": 5.064276868624587e-05, "loss": 0.0125, "step": 41690 }, { "grad_norm": 0.18223513662815094, "learning_rate": 5.0622101967102396e-05, "loss": 0.0106, "step": 41700 }, { "grad_norm": 0.23265951871871948, "learning_rate": 5.0601435141658714e-05, "loss": 0.0133, "step": 41710 }, { "grad_norm": 0.15046049654483795, "learning_rate": 5.058076821344622e-05, "loss": 0.0108, "step": 41720 }, { "grad_norm": 0.17446672916412354, "learning_rate": 5.056010118599632e-05, "loss": 0.0154, "step": 41730 }, { "grad_norm": 0.18216434121131897, "learning_rate": 5.053943406284044e-05, "loss": 0.0112, "step": 41740 }, { "grad_norm": 0.16720199584960938, "learning_rate": 5.051876684751006e-05, "loss": 0.0096, "step": 41750 }, { "grad_norm": 0.1689527928829193, "learning_rate": 5.0498099543536584e-05, "loss": 0.0112, "step": 41760 }, { "grad_norm": 0.10109398514032364, "learning_rate": 5.047743215445152e-05, "loss": 0.011, "step": 41770 }, { "grad_norm": 0.16032080352306366, "learning_rate": 5.045676468378637e-05, "loss": 0.0127, "step": 41780 }, { "grad_norm": 0.16206978261470795, "learning_rate": 5.0436097135072626e-05, "loss": 0.013, "step": 41790 }, { "grad_norm": 0.16700038313865662, "learning_rate": 5.041542951184181e-05, "loss": 0.0125, "step": 41800 }, { "grad_norm": 0.1946980506181717, "learning_rate": 5.039476181762545e-05, "loss": 0.0121, "step": 41810 }, { "grad_norm": 0.14778399467468262, "learning_rate": 5.037409405595508e-05, "loss": 0.0103, "step": 41820 }, { "grad_norm": 0.19884856045246124, "learning_rate": 5.035342623036229e-05, "loss": 0.0121, "step": 41830 }, { "grad_norm": 0.17757849395275116, "learning_rate": 5.033275834437862e-05, "loss": 0.0142, "step": 41840 }, { "grad_norm": 0.12062198668718338, "learning_rate": 5.031209040153564e-05, "loss": 0.0101, "step": 41850 }, { "grad_norm": 0.15465815365314484, "learning_rate": 5.0291422405364955e-05, "loss": 0.0132, "step": 41860 }, { "grad_norm": 0.2653690278530121, "learning_rate": 5.0270754359398133e-05, "loss": 0.0165, "step": 41870 }, { "grad_norm": 0.15705808997154236, "learning_rate": 5.025008626716682e-05, "loss": 0.0112, "step": 41880 }, { "grad_norm": 0.13057254254817963, "learning_rate": 5.0229418132202585e-05, "loss": 0.0126, "step": 41890 }, { "grad_norm": 0.15493252873420715, "learning_rate": 5.020874995803707e-05, "loss": 0.0124, "step": 41900 }, { "grad_norm": 0.18855848908424377, "learning_rate": 5.01880817482019e-05, "loss": 0.0103, "step": 41910 }, { "grad_norm": 0.2129998803138733, "learning_rate": 5.01674135062287e-05, "loss": 0.012, "step": 41920 }, { "grad_norm": 0.16301414370536804, "learning_rate": 5.0146745235649115e-05, "loss": 0.0131, "step": 41930 }, { "grad_norm": 0.154572993516922, "learning_rate": 5.0126076939994795e-05, "loss": 0.0112, "step": 41940 }, { "grad_norm": 0.2760332226753235, "learning_rate": 5.010540862279736e-05, "loss": 0.0143, "step": 41950 }, { "grad_norm": 0.19357715547084808, "learning_rate": 5.008474028758846e-05, "loss": 0.0113, "step": 41960 }, { "grad_norm": 0.09545531868934631, "learning_rate": 5.0064071937899784e-05, "loss": 0.0101, "step": 41970 }, { "grad_norm": 0.1593519002199173, "learning_rate": 5.004340357726296e-05, "loss": 0.0121, "step": 41980 }, { "grad_norm": 0.18995994329452515, "learning_rate": 5.002273520920965e-05, "loss": 0.011, "step": 41990 }, { "grad_norm": 0.18965652585029602, "learning_rate": 5.000206683727151e-05, "loss": 0.0118, "step": 42000 }, { "grad_norm": 0.22292286157608032, "learning_rate": 4.998139846498021e-05, "loss": 0.0134, "step": 42010 }, { "grad_norm": 0.12784557044506073, "learning_rate": 4.9960730095867405e-05, "loss": 0.0096, "step": 42020 }, { "grad_norm": 0.17762936651706696, "learning_rate": 4.9940061733464755e-05, "loss": 0.0103, "step": 42030 }, { "grad_norm": 0.14509420096874237, "learning_rate": 4.991939338130392e-05, "loss": 0.0118, "step": 42040 }, { "grad_norm": 0.15367498993873596, "learning_rate": 4.989872504291653e-05, "loss": 0.0104, "step": 42050 }, { "grad_norm": 0.13455058634281158, "learning_rate": 4.9878056721834273e-05, "loss": 0.0105, "step": 42060 }, { "grad_norm": 0.4168979525566101, "learning_rate": 4.98573884215888e-05, "loss": 0.0142, "step": 42070 }, { "grad_norm": 0.12618814408779144, "learning_rate": 4.9836720145711715e-05, "loss": 0.0108, "step": 42080 }, { "grad_norm": 0.21848450601100922, "learning_rate": 4.9816051897734725e-05, "loss": 0.0111, "step": 42090 }, { "grad_norm": 0.20395608246326447, "learning_rate": 4.979538368118941e-05, "loss": 0.0111, "step": 42100 }, { "grad_norm": 0.14872987568378448, "learning_rate": 4.9774715499607446e-05, "loss": 0.0132, "step": 42110 }, { "grad_norm": 0.16430674493312836, "learning_rate": 4.9754047356520416e-05, "loss": 0.012, "step": 42120 }, { "grad_norm": 0.15791727602481842, "learning_rate": 4.973337925545999e-05, "loss": 0.0109, "step": 42130 }, { "grad_norm": 0.24369916319847107, "learning_rate": 4.9712711199957746e-05, "loss": 0.0115, "step": 42140 }, { "grad_norm": 0.25114017724990845, "learning_rate": 4.969204319354531e-05, "loss": 0.0163, "step": 42150 }, { "grad_norm": 0.2144865095615387, "learning_rate": 4.9671375239754267e-05, "loss": 0.0098, "step": 42160 }, { "grad_norm": 0.16057895123958588, "learning_rate": 4.9650707342116195e-05, "loss": 0.0106, "step": 42170 }, { "grad_norm": 0.16862598061561584, "learning_rate": 4.96300395041627e-05, "loss": 0.0121, "step": 42180 }, { "grad_norm": 0.19736064970493317, "learning_rate": 4.960937172942532e-05, "loss": 0.0115, "step": 42190 }, { "grad_norm": 0.20618745684623718, "learning_rate": 4.958870402143565e-05, "loss": 0.0121, "step": 42200 }, { "grad_norm": 0.12308244407176971, "learning_rate": 4.9568036383725186e-05, "loss": 0.011, "step": 42210 }, { "grad_norm": 0.1461661010980606, "learning_rate": 4.9547368819825516e-05, "loss": 0.0113, "step": 42220 }, { "grad_norm": 0.24236302077770233, "learning_rate": 4.952670133326812e-05, "loss": 0.0089, "step": 42230 }, { "grad_norm": 0.15693247318267822, "learning_rate": 4.950603392758453e-05, "loss": 0.0127, "step": 42240 }, { "grad_norm": 0.13708914816379547, "learning_rate": 4.948536660630621e-05, "loss": 0.0114, "step": 42250 }, { "grad_norm": 0.23867851495742798, "learning_rate": 4.9464699372964685e-05, "loss": 0.0139, "step": 42260 }, { "grad_norm": 0.22267784178256989, "learning_rate": 4.9444032231091395e-05, "loss": 0.0139, "step": 42270 }, { "grad_norm": 0.16014109551906586, "learning_rate": 4.9423365184217776e-05, "loss": 0.0122, "step": 42280 }, { "grad_norm": 0.15721642971038818, "learning_rate": 4.940269823587529e-05, "loss": 0.0132, "step": 42290 }, { "grad_norm": 0.13681598007678986, "learning_rate": 4.938203138959533e-05, "loss": 0.0118, "step": 42300 }, { "grad_norm": 0.19361527264118195, "learning_rate": 4.9361364648909325e-05, "loss": 0.0117, "step": 42310 }, { "grad_norm": 0.1882167011499405, "learning_rate": 4.9340698017348605e-05, "loss": 0.0113, "step": 42320 }, { "grad_norm": 0.28432464599609375, "learning_rate": 4.932003149844458e-05, "loss": 0.0113, "step": 42330 }, { "grad_norm": 0.17546279728412628, "learning_rate": 4.929936509572857e-05, "loss": 0.0129, "step": 42340 }, { "grad_norm": 0.16645638644695282, "learning_rate": 4.927869881273191e-05, "loss": 0.0093, "step": 42350 }, { "grad_norm": 0.18845629692077637, "learning_rate": 4.9258032652985894e-05, "loss": 0.0126, "step": 42360 }, { "grad_norm": 0.23562397062778473, "learning_rate": 4.9237366620021786e-05, "loss": 0.0117, "step": 42370 }, { "grad_norm": 0.23630109429359436, "learning_rate": 4.921670071737089e-05, "loss": 0.0152, "step": 42380 }, { "grad_norm": 0.1096118688583374, "learning_rate": 4.91960349485644e-05, "loss": 0.0124, "step": 42390 }, { "grad_norm": 0.17990292608737946, "learning_rate": 4.9175369317133566e-05, "loss": 0.0108, "step": 42400 }, { "grad_norm": 0.1622486263513565, "learning_rate": 4.915470382660953e-05, "loss": 0.011, "step": 42410 }, { "grad_norm": 0.13474783301353455, "learning_rate": 4.9134038480523524e-05, "loss": 0.0114, "step": 42420 }, { "grad_norm": 0.13783785700798035, "learning_rate": 4.911337328240664e-05, "loss": 0.0093, "step": 42430 }, { "grad_norm": 0.17279647290706635, "learning_rate": 4.909270823579003e-05, "loss": 0.0123, "step": 42440 }, { "grad_norm": 0.1792650669813156, "learning_rate": 4.907204334420476e-05, "loss": 0.0104, "step": 42450 }, { "grad_norm": 0.19790716469287872, "learning_rate": 4.9051378611181884e-05, "loss": 0.0144, "step": 42460 }, { "grad_norm": 0.3220561444759369, "learning_rate": 4.9030714040252486e-05, "loss": 0.0125, "step": 42470 }, { "grad_norm": 0.19602161645889282, "learning_rate": 4.901004963494752e-05, "loss": 0.0146, "step": 42480 }, { "grad_norm": 0.1450146734714508, "learning_rate": 4.898938539879802e-05, "loss": 0.0153, "step": 42490 }, { "grad_norm": 0.22558997571468353, "learning_rate": 4.8968721335334904e-05, "loss": 0.0135, "step": 42500 }, { "grad_norm": 0.19230137765407562, "learning_rate": 4.894805744808912e-05, "loss": 0.0165, "step": 42510 }, { "grad_norm": 0.16574198007583618, "learning_rate": 4.8927393740591524e-05, "loss": 0.011, "step": 42520 }, { "grad_norm": 0.1679432988166809, "learning_rate": 4.890673021637302e-05, "loss": 0.0128, "step": 42530 }, { "grad_norm": 0.22569654881954193, "learning_rate": 4.888606687896442e-05, "loss": 0.0114, "step": 42540 }, { "grad_norm": 0.12239440530538559, "learning_rate": 4.886540373189652e-05, "loss": 0.0106, "step": 42550 }, { "grad_norm": 0.21438264846801758, "learning_rate": 4.88447407787001e-05, "loss": 0.0104, "step": 42560 }, { "grad_norm": 0.16453976929187775, "learning_rate": 4.8824078022905864e-05, "loss": 0.011, "step": 42570 }, { "grad_norm": 0.14279058575630188, "learning_rate": 4.880341546804455e-05, "loss": 0.0111, "step": 42580 }, { "grad_norm": 0.12385577708482742, "learning_rate": 4.878275311764679e-05, "loss": 0.0131, "step": 42590 }, { "grad_norm": 0.17211568355560303, "learning_rate": 4.8762090975243254e-05, "loss": 0.0106, "step": 42600 }, { "grad_norm": 0.10547741502523422, "learning_rate": 4.874142904436448e-05, "loss": 0.0126, "step": 42610 }, { "grad_norm": 0.19123022258281708, "learning_rate": 4.8720767328541094e-05, "loss": 0.0109, "step": 42620 }, { "grad_norm": 0.17223909497261047, "learning_rate": 4.870010583130358e-05, "loss": 0.0113, "step": 42630 }, { "grad_norm": 0.17946618795394897, "learning_rate": 4.867944455618241e-05, "loss": 0.0108, "step": 42640 }, { "grad_norm": 0.17604166269302368, "learning_rate": 4.865878350670807e-05, "loss": 0.0111, "step": 42650 }, { "grad_norm": 0.18538334965705872, "learning_rate": 4.8638122686410914e-05, "loss": 0.0122, "step": 42660 }, { "grad_norm": 0.20154595375061035, "learning_rate": 4.861746209882137e-05, "loss": 0.01, "step": 42670 }, { "grad_norm": 0.2023003101348877, "learning_rate": 4.859680174746972e-05, "loss": 0.0104, "step": 42680 }, { "grad_norm": 0.17378434538841248, "learning_rate": 4.857614163588629e-05, "loss": 0.0125, "step": 42690 }, { "grad_norm": 0.18590855598449707, "learning_rate": 4.855548176760131e-05, "loss": 0.0121, "step": 42700 }, { "grad_norm": 0.15308599174022675, "learning_rate": 4.8534822146144986e-05, "loss": 0.0107, "step": 42710 }, { "grad_norm": 0.20167022943496704, "learning_rate": 4.851416277504749e-05, "loss": 0.0146, "step": 42720 }, { "grad_norm": 0.2313937246799469, "learning_rate": 4.8493503657838923e-05, "loss": 0.0098, "step": 42730 }, { "grad_norm": 0.15281055867671967, "learning_rate": 4.84728447980494e-05, "loss": 0.0113, "step": 42740 }, { "grad_norm": 0.16902820765972137, "learning_rate": 4.8452186199208914e-05, "loss": 0.0115, "step": 42750 }, { "grad_norm": 0.1820291131734848, "learning_rate": 4.843152786484749e-05, "loss": 0.0118, "step": 42760 }, { "grad_norm": 0.20298436284065247, "learning_rate": 4.8410869798495035e-05, "loss": 0.0084, "step": 42770 }, { "grad_norm": 0.16229578852653503, "learning_rate": 4.8390212003681486e-05, "loss": 0.009, "step": 42780 }, { "grad_norm": 0.2545814514160156, "learning_rate": 4.836955448393667e-05, "loss": 0.0119, "step": 42790 }, { "grad_norm": 0.22819505631923676, "learning_rate": 4.8348897242790406e-05, "loss": 0.0106, "step": 42800 }, { "grad_norm": 0.20528987050056458, "learning_rate": 4.832824028377243e-05, "loss": 0.0116, "step": 42810 }, { "grad_norm": 0.14724576473236084, "learning_rate": 4.830758361041249e-05, "loss": 0.0101, "step": 42820 }, { "grad_norm": 0.17210865020751953, "learning_rate": 4.828692722624022e-05, "loss": 0.0113, "step": 42830 }, { "grad_norm": 0.18038500845432281, "learning_rate": 4.826627113478522e-05, "loss": 0.0131, "step": 42840 }, { "grad_norm": 0.13577914237976074, "learning_rate": 4.824561533957708e-05, "loss": 0.0116, "step": 42850 }, { "grad_norm": 0.19492220878601074, "learning_rate": 4.822495984414527e-05, "loss": 0.011, "step": 42860 }, { "grad_norm": 0.12461059540510178, "learning_rate": 4.8204304652019304e-05, "loss": 0.0094, "step": 42870 }, { "grad_norm": 0.2820148468017578, "learning_rate": 4.8183649766728537e-05, "loss": 0.012, "step": 42880 }, { "grad_norm": 0.15367823839187622, "learning_rate": 4.816299519180237e-05, "loss": 0.0087, "step": 42890 }, { "grad_norm": 0.16884945333003998, "learning_rate": 4.8142340930770066e-05, "loss": 0.0107, "step": 42900 }, { "grad_norm": 0.1353236883878708, "learning_rate": 4.81216869871609e-05, "loss": 0.0093, "step": 42910 }, { "grad_norm": 0.20220914483070374, "learning_rate": 4.8101033364504064e-05, "loss": 0.0103, "step": 42920 }, { "grad_norm": 0.14174784719944, "learning_rate": 4.808038006632866e-05, "loss": 0.0082, "step": 42930 }, { "grad_norm": 0.17358386516571045, "learning_rate": 4.805972709616383e-05, "loss": 0.0136, "step": 42940 }, { "grad_norm": 0.14063352346420288, "learning_rate": 4.8039074457538556e-05, "loss": 0.0111, "step": 42950 }, { "grad_norm": 0.27338382601737976, "learning_rate": 4.801842215398184e-05, "loss": 0.013, "step": 42960 }, { "grad_norm": 0.14246852695941925, "learning_rate": 4.799777018902256e-05, "loss": 0.0137, "step": 42970 }, { "grad_norm": 0.23042073845863342, "learning_rate": 4.797711856618961e-05, "loss": 0.0142, "step": 42980 }, { "grad_norm": 0.14061373472213745, "learning_rate": 4.7956467289011765e-05, "loss": 0.0107, "step": 42990 }, { "grad_norm": 0.13580088317394257, "learning_rate": 4.793581636101778e-05, "loss": 0.0099, "step": 43000 }, { "grad_norm": 0.20885537564754486, "learning_rate": 4.7915165785736326e-05, "loss": 0.0129, "step": 43010 }, { "grad_norm": 0.231458842754364, "learning_rate": 4.789451556669599e-05, "loss": 0.0134, "step": 43020 }, { "grad_norm": 0.16594775021076202, "learning_rate": 4.7873865707425384e-05, "loss": 0.012, "step": 43030 }, { "grad_norm": 0.22671857476234436, "learning_rate": 4.7853216211452975e-05, "loss": 0.0126, "step": 43040 }, { "grad_norm": 0.18549519777297974, "learning_rate": 4.78325670823072e-05, "loss": 0.0112, "step": 43050 }, { "grad_norm": 0.21523843705654144, "learning_rate": 4.781191832351641e-05, "loss": 0.0149, "step": 43060 }, { "grad_norm": 0.16970469057559967, "learning_rate": 4.7791269938608965e-05, "loss": 0.0137, "step": 43070 }, { "grad_norm": 0.12729108333587646, "learning_rate": 4.777062193111305e-05, "loss": 0.0131, "step": 43080 }, { "grad_norm": 0.2308991253376007, "learning_rate": 4.77499743045569e-05, "loss": 0.0106, "step": 43090 }, { "grad_norm": 0.18457511067390442, "learning_rate": 4.7729327062468596e-05, "loss": 0.0104, "step": 43100 }, { "grad_norm": 0.19571255147457123, "learning_rate": 4.770868020837619e-05, "loss": 0.0127, "step": 43110 }, { "grad_norm": 0.16294409334659576, "learning_rate": 4.768803374580768e-05, "loss": 0.0114, "step": 43120 }, { "grad_norm": 0.1457650065422058, "learning_rate": 4.766738767829094e-05, "loss": 0.0111, "step": 43130 }, { "grad_norm": 0.16931946575641632, "learning_rate": 4.764674200935388e-05, "loss": 0.0085, "step": 43140 }, { "grad_norm": 0.1620219349861145, "learning_rate": 4.762609674252424e-05, "loss": 0.0099, "step": 43150 }, { "grad_norm": 0.1434582769870758, "learning_rate": 4.760545188132974e-05, "loss": 0.011, "step": 43160 }, { "grad_norm": 0.17757800221443176, "learning_rate": 4.758480742929801e-05, "loss": 0.0111, "step": 43170 }, { "grad_norm": 0.16144467890262604, "learning_rate": 4.756416338995664e-05, "loss": 0.0101, "step": 43180 }, { "grad_norm": 0.1880951076745987, "learning_rate": 4.7543519766833145e-05, "loss": 0.0109, "step": 43190 }, { "grad_norm": 0.1271429806947708, "learning_rate": 4.752287656345492e-05, "loss": 0.0092, "step": 43200 }, { "grad_norm": 0.19760145246982574, "learning_rate": 4.7502233783349355e-05, "loss": 0.0093, "step": 43210 }, { "grad_norm": 0.1543220430612564, "learning_rate": 4.7481591430043694e-05, "loss": 0.0149, "step": 43220 }, { "grad_norm": 0.4847354590892792, "learning_rate": 4.746094950706521e-05, "loss": 0.0167, "step": 43230 }, { "grad_norm": 0.1246863305568695, "learning_rate": 4.744030801794099e-05, "loss": 0.0087, "step": 43240 }, { "grad_norm": 0.21078145503997803, "learning_rate": 4.741966696619813e-05, "loss": 0.0118, "step": 43250 }, { "grad_norm": 0.13441012799739838, "learning_rate": 4.739902635536361e-05, "loss": 0.0101, "step": 43260 }, { "grad_norm": 0.1779281347990036, "learning_rate": 4.737838618896436e-05, "loss": 0.0125, "step": 43270 }, { "grad_norm": 0.16399621963500977, "learning_rate": 4.7357746470527203e-05, "loss": 0.012, "step": 43280 }, { "grad_norm": 0.1837254762649536, "learning_rate": 4.73371072035789e-05, "loss": 0.0098, "step": 43290 }, { "grad_norm": 0.18130245804786682, "learning_rate": 4.731646839164616e-05, "loss": 0.0127, "step": 43300 }, { "grad_norm": 0.11119712889194489, "learning_rate": 4.729583003825556e-05, "loss": 0.0106, "step": 43310 }, { "grad_norm": 0.1611916422843933, "learning_rate": 4.727519214693367e-05, "loss": 0.0091, "step": 43320 }, { "grad_norm": 0.1651010364294052, "learning_rate": 4.725455472120689e-05, "loss": 0.0091, "step": 43330 }, { "grad_norm": 0.19555969536304474, "learning_rate": 4.723391776460164e-05, "loss": 0.0105, "step": 43340 }, { "grad_norm": 0.2444525957107544, "learning_rate": 4.7213281280644186e-05, "loss": 0.0116, "step": 43350 }, { "grad_norm": 0.1973051130771637, "learning_rate": 4.719264527286075e-05, "loss": 0.0096, "step": 43360 }, { "grad_norm": 0.17807415127754211, "learning_rate": 4.717200974477744e-05, "loss": 0.0107, "step": 43370 }, { "grad_norm": 0.10088803619146347, "learning_rate": 4.715137469992034e-05, "loss": 0.0086, "step": 43380 }, { "grad_norm": 0.11647122353315353, "learning_rate": 4.713074014181539e-05, "loss": 0.01, "step": 43390 }, { "grad_norm": 0.22930005192756653, "learning_rate": 4.711010607398846e-05, "loss": 0.0109, "step": 43400 }, { "grad_norm": 0.18474362790584564, "learning_rate": 4.708947249996538e-05, "loss": 0.0092, "step": 43410 }, { "grad_norm": 0.18935729563236237, "learning_rate": 4.706883942327183e-05, "loss": 0.0113, "step": 43420 }, { "grad_norm": 0.14571937918663025, "learning_rate": 4.704820684743347e-05, "loss": 0.0122, "step": 43430 }, { "grad_norm": 0.14836761355400085, "learning_rate": 4.702757477597581e-05, "loss": 0.0114, "step": 43440 }, { "grad_norm": 0.14262919127941132, "learning_rate": 4.700694321242435e-05, "loss": 0.0116, "step": 43450 }, { "grad_norm": 0.20447148382663727, "learning_rate": 4.698631216030441e-05, "loss": 0.0119, "step": 43460 }, { "grad_norm": 0.21403750777244568, "learning_rate": 4.6965681623141314e-05, "loss": 0.0131, "step": 43470 }, { "grad_norm": 0.14367671310901642, "learning_rate": 4.694505160446024e-05, "loss": 0.0095, "step": 43480 }, { "grad_norm": 0.21443627774715424, "learning_rate": 4.692442210778628e-05, "loss": 0.0166, "step": 43490 }, { "grad_norm": 0.21743233501911163, "learning_rate": 4.690379313664448e-05, "loss": 0.0121, "step": 43500 }, { "grad_norm": 0.11707980930805206, "learning_rate": 4.688316469455973e-05, "loss": 0.0132, "step": 43510 }, { "grad_norm": 0.15956582129001617, "learning_rate": 4.6862536785056914e-05, "loss": 0.0139, "step": 43520 }, { "grad_norm": 0.15687739849090576, "learning_rate": 4.684190941166073e-05, "loss": 0.0122, "step": 43530 }, { "grad_norm": 0.15541525185108185, "learning_rate": 4.682128257789587e-05, "loss": 0.015, "step": 43540 }, { "grad_norm": 0.17194494605064392, "learning_rate": 4.680065628728687e-05, "loss": 0.0148, "step": 43550 }, { "grad_norm": 0.15093982219696045, "learning_rate": 4.678003054335822e-05, "loss": 0.0138, "step": 43560 }, { "grad_norm": 0.14548274874687195, "learning_rate": 4.675940534963428e-05, "loss": 0.0114, "step": 43570 }, { "grad_norm": 0.16228072345256805, "learning_rate": 4.673878070963931e-05, "loss": 0.0111, "step": 43580 }, { "grad_norm": 0.13798296451568604, "learning_rate": 4.671815662689756e-05, "loss": 0.0111, "step": 43590 }, { "grad_norm": 0.14061403274536133, "learning_rate": 4.669753310493306e-05, "loss": 0.0131, "step": 43600 }, { "grad_norm": 0.14243492484092712, "learning_rate": 4.6676910147269845e-05, "loss": 0.0119, "step": 43610 }, { "grad_norm": 0.1441746950149536, "learning_rate": 4.665628775743177e-05, "loss": 0.0115, "step": 43620 }, { "grad_norm": 0.14233796298503876, "learning_rate": 4.66356659389427e-05, "loss": 0.0121, "step": 43630 }, { "grad_norm": 0.1702367663383484, "learning_rate": 4.661504469532629e-05, "loss": 0.0117, "step": 43640 }, { "grad_norm": 0.24085500836372375, "learning_rate": 4.659442403010617e-05, "loss": 0.0118, "step": 43650 }, { "grad_norm": 0.1971554011106491, "learning_rate": 4.6573803946805845e-05, "loss": 0.0107, "step": 43660 }, { "grad_norm": 0.17554520070552826, "learning_rate": 4.6553184448948695e-05, "loss": 0.0102, "step": 43670 }, { "grad_norm": 0.2553006708621979, "learning_rate": 4.653256554005807e-05, "loss": 0.0116, "step": 43680 }, { "grad_norm": 0.12480898946523666, "learning_rate": 4.6511947223657145e-05, "loss": 0.0114, "step": 43690 }, { "grad_norm": 0.127364844083786, "learning_rate": 4.649132950326906e-05, "loss": 0.0142, "step": 43700 }, { "grad_norm": 0.13552553951740265, "learning_rate": 4.647071238241679e-05, "loss": 0.0103, "step": 43710 }, { "grad_norm": 0.18398313224315643, "learning_rate": 4.6450095864623264e-05, "loss": 0.0127, "step": 43720 }, { "grad_norm": 0.1743319034576416, "learning_rate": 4.642947995341123e-05, "loss": 0.013, "step": 43730 }, { "grad_norm": 0.1635948121547699, "learning_rate": 4.640886465230345e-05, "loss": 0.0124, "step": 43740 }, { "grad_norm": 0.21174566447734833, "learning_rate": 4.6388249964822485e-05, "loss": 0.0129, "step": 43750 }, { "grad_norm": 0.16947385668754578, "learning_rate": 4.63676358944908e-05, "loss": 0.01, "step": 43760 }, { "grad_norm": 0.167998269200325, "learning_rate": 4.634702244483081e-05, "loss": 0.0107, "step": 43770 }, { "grad_norm": 0.2110143005847931, "learning_rate": 4.632640961936475e-05, "loss": 0.0105, "step": 43780 }, { "grad_norm": 0.15560075640678406, "learning_rate": 4.6305797421614835e-05, "loss": 0.0089, "step": 43790 }, { "grad_norm": 0.14135737717151642, "learning_rate": 4.6285185855103105e-05, "loss": 0.0103, "step": 43800 }, { "grad_norm": 0.13305015861988068, "learning_rate": 4.626457492335151e-05, "loss": 0.0087, "step": 43810 }, { "grad_norm": 0.1737978458404541, "learning_rate": 4.624396462988188e-05, "loss": 0.0129, "step": 43820 }, { "grad_norm": 0.18501554429531097, "learning_rate": 4.6223354978215985e-05, "loss": 0.0101, "step": 43830 }, { "grad_norm": 0.13942871987819672, "learning_rate": 4.620274597187544e-05, "loss": 0.0113, "step": 43840 }, { "grad_norm": 0.27390754222869873, "learning_rate": 4.6182137614381736e-05, "loss": 0.0123, "step": 43850 }, { "grad_norm": 0.20658983290195465, "learning_rate": 4.616152990925631e-05, "loss": 0.0125, "step": 43860 }, { "grad_norm": 0.17572121322155, "learning_rate": 4.614092286002041e-05, "loss": 0.0134, "step": 43870 }, { "grad_norm": 0.18813928961753845, "learning_rate": 4.612031647019528e-05, "loss": 0.009, "step": 43880 }, { "grad_norm": 0.19122625887393951, "learning_rate": 4.609971074330193e-05, "loss": 0.0122, "step": 43890 }, { "grad_norm": 0.2054794430732727, "learning_rate": 4.607910568286136e-05, "loss": 0.013, "step": 43900 }, { "grad_norm": 0.18129664659500122, "learning_rate": 4.605850129239438e-05, "loss": 0.0092, "step": 43910 }, { "grad_norm": 0.13905592262744904, "learning_rate": 4.603789757542174e-05, "loss": 0.0107, "step": 43920 }, { "grad_norm": 0.2017045021057129, "learning_rate": 4.601729453546404e-05, "loss": 0.0108, "step": 43930 }, { "grad_norm": 0.13804708421230316, "learning_rate": 4.599669217604177e-05, "loss": 0.0108, "step": 43940 }, { "grad_norm": 0.09097320586442947, "learning_rate": 4.597609050067532e-05, "loss": 0.0113, "step": 43950 }, { "grad_norm": 0.1492644101381302, "learning_rate": 4.595548951288495e-05, "loss": 0.0103, "step": 43960 }, { "grad_norm": 0.11528419703245163, "learning_rate": 4.593488921619081e-05, "loss": 0.0115, "step": 43970 }, { "grad_norm": 0.17725761234760284, "learning_rate": 4.591428961411289e-05, "loss": 0.0099, "step": 43980 }, { "grad_norm": 0.12638136744499207, "learning_rate": 4.589369071017117e-05, "loss": 0.0113, "step": 43990 }, { "grad_norm": 0.1672971248626709, "learning_rate": 4.587309250788538e-05, "loss": 0.0112, "step": 44000 }, { "grad_norm": 0.154683455824852, "learning_rate": 4.585249501077522e-05, "loss": 0.0074, "step": 44010 }, { "grad_norm": 0.14349351823329926, "learning_rate": 4.583189822236021e-05, "loss": 0.0111, "step": 44020 }, { "grad_norm": 0.15731501579284668, "learning_rate": 4.5811302146159816e-05, "loss": 0.0093, "step": 44030 }, { "grad_norm": 0.21817699074745178, "learning_rate": 4.579070678569332e-05, "loss": 0.0104, "step": 44040 }, { "grad_norm": 0.25144827365875244, "learning_rate": 4.5770112144479904e-05, "loss": 0.0106, "step": 44050 }, { "grad_norm": 0.1311858743429184, "learning_rate": 4.5749518226038645e-05, "loss": 0.009, "step": 44060 }, { "grad_norm": 0.1337963491678238, "learning_rate": 4.572892503388845e-05, "loss": 0.0124, "step": 44070 }, { "grad_norm": 0.2529774606227875, "learning_rate": 4.570833257154817e-05, "loss": 0.0107, "step": 44080 }, { "grad_norm": 0.1724669188261032, "learning_rate": 4.568774084253646e-05, "loss": 0.01, "step": 44090 }, { "grad_norm": 0.1583145260810852, "learning_rate": 4.566714985037191e-05, "loss": 0.0112, "step": 44100 }, { "grad_norm": 0.1806280016899109, "learning_rate": 4.564655959857295e-05, "loss": 0.0099, "step": 44110 }, { "grad_norm": 0.11873580515384674, "learning_rate": 4.5625970090657893e-05, "loss": 0.0092, "step": 44120 }, { "grad_norm": 0.15681765973567963, "learning_rate": 4.560538133014491e-05, "loss": 0.0095, "step": 44130 }, { "grad_norm": 0.10362102091312408, "learning_rate": 4.5584793320552055e-05, "loss": 0.0107, "step": 44140 }, { "grad_norm": 0.1782158762216568, "learning_rate": 4.556420606539728e-05, "loss": 0.0122, "step": 44150 }, { "grad_norm": 0.1381598860025406, "learning_rate": 4.554361956819836e-05, "loss": 0.0112, "step": 44160 }, { "grad_norm": 0.26088157296180725, "learning_rate": 4.552303383247299e-05, "loss": 0.0098, "step": 44170 }, { "grad_norm": 0.15320749580860138, "learning_rate": 4.5502448861738664e-05, "loss": 0.01, "step": 44180 }, { "grad_norm": 0.15272365510463715, "learning_rate": 4.548186465951284e-05, "loss": 0.0108, "step": 44190 }, { "grad_norm": 0.12729433178901672, "learning_rate": 4.546128122931277e-05, "loss": 0.0079, "step": 44200 }, { "grad_norm": 0.2235291451215744, "learning_rate": 4.544069857465561e-05, "loss": 0.0112, "step": 44210 }, { "grad_norm": 0.13755165040493011, "learning_rate": 4.542011669905837e-05, "loss": 0.0105, "step": 44220 }, { "grad_norm": 0.17613595724105835, "learning_rate": 4.53995356060379e-05, "loss": 0.0097, "step": 44230 }, { "grad_norm": 0.16078312695026398, "learning_rate": 4.5378955299110994e-05, "loss": 0.0109, "step": 44240 }, { "grad_norm": 0.15237602591514587, "learning_rate": 4.5358375781794233e-05, "loss": 0.009, "step": 44250 }, { "grad_norm": 0.1334577202796936, "learning_rate": 4.533779705760411e-05, "loss": 0.0113, "step": 44260 }, { "grad_norm": 0.15409904718399048, "learning_rate": 4.5317219130056934e-05, "loss": 0.0089, "step": 44270 }, { "grad_norm": 0.2207636535167694, "learning_rate": 4.5296642002668964e-05, "loss": 0.0099, "step": 44280 }, { "grad_norm": 0.161219984292984, "learning_rate": 4.5276065678956216e-05, "loss": 0.0099, "step": 44290 }, { "grad_norm": 0.18392252922058105, "learning_rate": 4.525549016243466e-05, "loss": 0.0107, "step": 44300 }, { "grad_norm": 0.22223971784114838, "learning_rate": 4.523491545662008e-05, "loss": 0.0149, "step": 44310 }, { "grad_norm": 0.15094536542892456, "learning_rate": 4.5214341565028104e-05, "loss": 0.0088, "step": 44320 }, { "grad_norm": 0.1906883865594864, "learning_rate": 4.519376849117428e-05, "loss": 0.0088, "step": 44330 }, { "grad_norm": 0.14957350492477417, "learning_rate": 4.517319623857395e-05, "loss": 0.0138, "step": 44340 }, { "grad_norm": 0.16305135190486908, "learning_rate": 4.5152624810742394e-05, "loss": 0.0085, "step": 44350 }, { "grad_norm": 0.1854967623949051, "learning_rate": 4.5132054211194664e-05, "loss": 0.012, "step": 44360 }, { "grad_norm": 0.15709879994392395, "learning_rate": 4.511148444344574e-05, "loss": 0.0112, "step": 44370 }, { "grad_norm": 0.13931135833263397, "learning_rate": 4.509091551101041e-05, "loss": 0.0127, "step": 44380 }, { "grad_norm": 0.2223278433084488, "learning_rate": 4.507034741740338e-05, "loss": 0.0141, "step": 44390 }, { "grad_norm": 0.196381613612175, "learning_rate": 4.5049780166139145e-05, "loss": 0.0098, "step": 44400 }, { "grad_norm": 0.1710164099931717, "learning_rate": 4.5029213760732075e-05, "loss": 0.0112, "step": 44410 }, { "grad_norm": 0.11970005184412003, "learning_rate": 4.5008648204696434e-05, "loss": 0.0142, "step": 44420 }, { "grad_norm": 0.13551436364650726, "learning_rate": 4.4988083501546284e-05, "loss": 0.014, "step": 44430 }, { "grad_norm": 0.16396993398666382, "learning_rate": 4.4967519654795606e-05, "loss": 0.0092, "step": 44440 }, { "grad_norm": 0.13521559536457062, "learning_rate": 4.494695666795816e-05, "loss": 0.0099, "step": 44450 }, { "grad_norm": 0.18112772703170776, "learning_rate": 4.4926394544547644e-05, "loss": 0.0108, "step": 44460 }, { "grad_norm": 0.17707230150699615, "learning_rate": 4.490583328807752e-05, "loss": 0.0108, "step": 44470 }, { "grad_norm": 0.15942959487438202, "learning_rate": 4.488527290206117e-05, "loss": 0.0093, "step": 44480 }, { "grad_norm": 0.1934536248445511, "learning_rate": 4.48647133900118e-05, "loss": 0.0107, "step": 44490 }, { "grad_norm": 0.16013723611831665, "learning_rate": 4.484415475544244e-05, "loss": 0.0094, "step": 44500 }, { "grad_norm": 0.11657578498125076, "learning_rate": 4.4823597001866035e-05, "loss": 0.0071, "step": 44510 }, { "grad_norm": 0.15014347434043884, "learning_rate": 4.480304013279532e-05, "loss": 0.0083, "step": 44520 }, { "grad_norm": 0.21723996102809906, "learning_rate": 4.478248415174292e-05, "loss": 0.0098, "step": 44530 }, { "grad_norm": 0.11553546041250229, "learning_rate": 4.476192906222126e-05, "loss": 0.0093, "step": 44540 }, { "grad_norm": 0.16960860788822174, "learning_rate": 4.474137486774268e-05, "loss": 0.0094, "step": 44550 }, { "grad_norm": 0.12900498509407043, "learning_rate": 4.4720821571819296e-05, "loss": 0.0093, "step": 44560 }, { "grad_norm": 0.19575026631355286, "learning_rate": 4.470026917796314e-05, "loss": 0.01, "step": 44570 }, { "grad_norm": 0.1995389312505722, "learning_rate": 4.4679717689686005e-05, "loss": 0.0108, "step": 44580 }, { "grad_norm": 0.18082307279109955, "learning_rate": 4.465916711049963e-05, "loss": 0.0144, "step": 44590 }, { "grad_norm": 0.11778576672077179, "learning_rate": 4.4638617443915524e-05, "loss": 0.0073, "step": 44600 }, { "grad_norm": 0.18141856789588928, "learning_rate": 4.4618068693445055e-05, "loss": 0.0085, "step": 44610 }, { "grad_norm": 0.17067831754684448, "learning_rate": 4.459752086259946e-05, "loss": 0.0122, "step": 44620 }, { "grad_norm": 0.1381574124097824, "learning_rate": 4.457697395488977e-05, "loss": 0.01, "step": 44630 }, { "grad_norm": 0.12761260569095612, "learning_rate": 4.455642797382693e-05, "loss": 0.0112, "step": 44640 }, { "grad_norm": 0.11813529580831528, "learning_rate": 4.4535882922921643e-05, "loss": 0.0128, "step": 44650 }, { "grad_norm": 0.1763768345117569, "learning_rate": 4.451533880568455e-05, "loss": 0.0115, "step": 44660 }, { "grad_norm": 0.1585688292980194, "learning_rate": 4.449479562562603e-05, "loss": 0.0087, "step": 44670 }, { "grad_norm": 0.16262339055538177, "learning_rate": 4.447425338625639e-05, "loss": 0.0117, "step": 44680 }, { "grad_norm": 0.18815359473228455, "learning_rate": 4.4453712091085705e-05, "loss": 0.0101, "step": 44690 }, { "grad_norm": 0.1805875599384308, "learning_rate": 4.443317174362392e-05, "loss": 0.0128, "step": 44700 }, { "grad_norm": 0.14755836129188538, "learning_rate": 4.441263234738085e-05, "loss": 0.0096, "step": 44710 }, { "grad_norm": 0.13129736483097076, "learning_rate": 4.4392093905866086e-05, "loss": 0.0082, "step": 44720 }, { "grad_norm": 0.1463422328233719, "learning_rate": 4.437155642258911e-05, "loss": 0.0111, "step": 44730 }, { "grad_norm": 0.16541670262813568, "learning_rate": 4.4351019901059177e-05, "loss": 0.0115, "step": 44740 }, { "grad_norm": 0.27479836344718933, "learning_rate": 4.433048434478547e-05, "loss": 0.0119, "step": 44750 }, { "grad_norm": 0.17380580306053162, "learning_rate": 4.4309949757276906e-05, "loss": 0.0098, "step": 44760 }, { "grad_norm": 0.2978500425815582, "learning_rate": 4.428941614204233e-05, "loss": 0.0122, "step": 44770 }, { "grad_norm": 0.12130792438983917, "learning_rate": 4.426888350259034e-05, "loss": 0.0107, "step": 44780 }, { "grad_norm": 0.17564930021762848, "learning_rate": 4.424835184242939e-05, "loss": 0.0104, "step": 44790 }, { "grad_norm": 0.14418840408325195, "learning_rate": 4.422782116506784e-05, "loss": 0.0113, "step": 44800 }, { "grad_norm": 0.17686505615711212, "learning_rate": 4.4207291474013754e-05, "loss": 0.0113, "step": 44810 }, { "grad_norm": 0.1381523609161377, "learning_rate": 4.418676277277514e-05, "loss": 0.0145, "step": 44820 }, { "grad_norm": 0.10697007179260254, "learning_rate": 4.416623506485976e-05, "loss": 0.0123, "step": 44830 }, { "grad_norm": 0.17074476182460785, "learning_rate": 4.414570835377527e-05, "loss": 0.0118, "step": 44840 }, { "grad_norm": 0.1961248815059662, "learning_rate": 4.412518264302909e-05, "loss": 0.0113, "step": 44850 }, { "grad_norm": 0.12298719584941864, "learning_rate": 4.4104657936128535e-05, "loss": 0.012, "step": 44860 }, { "grad_norm": 0.12769746780395508, "learning_rate": 4.408413423658071e-05, "loss": 0.0146, "step": 44870 }, { "grad_norm": 0.10846399515867233, "learning_rate": 4.406361154789253e-05, "loss": 0.0105, "step": 44880 }, { "grad_norm": 0.2014184445142746, "learning_rate": 4.4043089873570776e-05, "loss": 0.0149, "step": 44890 }, { "grad_norm": 0.13736343383789062, "learning_rate": 4.4022569217122025e-05, "loss": 0.0113, "step": 44900 }, { "grad_norm": 0.12874823808670044, "learning_rate": 4.400204958205274e-05, "loss": 0.0113, "step": 44910 }, { "grad_norm": 0.14529794454574585, "learning_rate": 4.3981530971869125e-05, "loss": 0.0108, "step": 44920 }, { "grad_norm": 0.12431306391954422, "learning_rate": 4.396101339007727e-05, "loss": 0.011, "step": 44930 }, { "grad_norm": 0.14396266639232635, "learning_rate": 4.394049684018304e-05, "loss": 0.0108, "step": 44940 }, { "grad_norm": 0.14126954972743988, "learning_rate": 4.3919981325692186e-05, "loss": 0.0103, "step": 44950 }, { "grad_norm": 0.14904190599918365, "learning_rate": 4.389946685011024e-05, "loss": 0.0117, "step": 44960 }, { "grad_norm": 0.15567050874233246, "learning_rate": 4.387895341694255e-05, "loss": 0.0104, "step": 44970 }, { "grad_norm": 0.14273253083229065, "learning_rate": 4.3858441029694324e-05, "loss": 0.0096, "step": 44980 }, { "grad_norm": 0.14751876890659332, "learning_rate": 4.3837929691870527e-05, "loss": 0.0095, "step": 44990 }, { "grad_norm": 0.11628670245409012, "learning_rate": 4.381741940697604e-05, "loss": 0.0092, "step": 45000 }, { "grad_norm": 0.17438361048698425, "learning_rate": 4.379691017851547e-05, "loss": 0.0098, "step": 45010 }, { "grad_norm": 0.15265090763568878, "learning_rate": 4.3776402009993304e-05, "loss": 0.0118, "step": 45020 }, { "grad_norm": 0.18567605316638947, "learning_rate": 4.3755894904913794e-05, "loss": 0.013, "step": 45030 }, { "grad_norm": 0.1714726835489273, "learning_rate": 4.373538886678109e-05, "loss": 0.0109, "step": 45040 }, { "grad_norm": 0.17719508707523346, "learning_rate": 4.371488389909909e-05, "loss": 0.0074, "step": 45050 }, { "grad_norm": 0.16104358434677124, "learning_rate": 4.3694380005371515e-05, "loss": 0.0094, "step": 45060 }, { "grad_norm": 0.19359342753887177, "learning_rate": 4.367387718910196e-05, "loss": 0.0127, "step": 45070 }, { "grad_norm": 0.1550254225730896, "learning_rate": 4.3653375453793764e-05, "loss": 0.0093, "step": 45080 }, { "grad_norm": 0.22054791450500488, "learning_rate": 4.3632874802950136e-05, "loss": 0.0106, "step": 45090 }, { "grad_norm": 0.19206035137176514, "learning_rate": 4.3612375240074034e-05, "loss": 0.0124, "step": 45100 }, { "grad_norm": 0.18458154797554016, "learning_rate": 4.3591876768668325e-05, "loss": 0.0082, "step": 45110 }, { "grad_norm": 0.17245611548423767, "learning_rate": 4.3571379392235605e-05, "loss": 0.0082, "step": 45120 }, { "grad_norm": 0.11487797647714615, "learning_rate": 4.3550883114278335e-05, "loss": 0.01, "step": 45130 }, { "grad_norm": 0.19023549556732178, "learning_rate": 4.353038793829876e-05, "loss": 0.0101, "step": 45140 }, { "grad_norm": 0.19424749910831451, "learning_rate": 4.350989386779891e-05, "loss": 0.0116, "step": 45150 }, { "grad_norm": 0.23081077635288239, "learning_rate": 4.3489400906280724e-05, "loss": 0.0094, "step": 45160 }, { "grad_norm": 0.15166573226451874, "learning_rate": 4.3468909057245845e-05, "loss": 0.008, "step": 45170 }, { "grad_norm": 0.22364141047000885, "learning_rate": 4.3448418324195794e-05, "loss": 0.0108, "step": 45180 }, { "grad_norm": 0.1250305473804474, "learning_rate": 4.342792871063184e-05, "loss": 0.0093, "step": 45190 }, { "grad_norm": 0.16250130534172058, "learning_rate": 4.3407440220055145e-05, "loss": 0.0076, "step": 45200 }, { "grad_norm": 0.18464724719524384, "learning_rate": 4.33869528559666e-05, "loss": 0.0133, "step": 45210 }, { "grad_norm": 0.21383526921272278, "learning_rate": 4.336646662186696e-05, "loss": 0.0118, "step": 45220 }, { "grad_norm": 0.15421606600284576, "learning_rate": 4.334598152125672e-05, "loss": 0.0107, "step": 45230 }, { "grad_norm": 0.13251197338104248, "learning_rate": 4.3325497557636276e-05, "loss": 0.0136, "step": 45240 }, { "grad_norm": 0.20831234753131866, "learning_rate": 4.330501473450574e-05, "loss": 0.012, "step": 45250 }, { "grad_norm": 0.15564951300621033, "learning_rate": 4.328453305536507e-05, "loss": 0.0106, "step": 45260 }, { "grad_norm": 0.16711948812007904, "learning_rate": 4.326405252371404e-05, "loss": 0.0137, "step": 45270 }, { "grad_norm": 0.20916076004505157, "learning_rate": 4.324357314305221e-05, "loss": 0.0088, "step": 45280 }, { "grad_norm": 0.24050508439540863, "learning_rate": 4.3223094916878945e-05, "loss": 0.0155, "step": 45290 }, { "grad_norm": 0.1879194974899292, "learning_rate": 4.320261784869338e-05, "loss": 0.0124, "step": 45300 }, { "grad_norm": 0.1533265858888626, "learning_rate": 4.318214194199455e-05, "loss": 0.0133, "step": 45310 }, { "grad_norm": 0.16517998278141022, "learning_rate": 4.316166720028118e-05, "loss": 0.01, "step": 45320 }, { "grad_norm": 0.1483725607395172, "learning_rate": 4.3141193627051864e-05, "loss": 0.0126, "step": 45330 }, { "grad_norm": 0.24912787973880768, "learning_rate": 4.312072122580496e-05, "loss": 0.0092, "step": 45340 }, { "grad_norm": 0.206825390458107, "learning_rate": 4.3100250000038646e-05, "loss": 0.0118, "step": 45350 }, { "grad_norm": 0.23135177791118622, "learning_rate": 4.307977995325091e-05, "loss": 0.0119, "step": 45360 }, { "grad_norm": 0.2521440386772156, "learning_rate": 4.30593110889395e-05, "loss": 0.014, "step": 45370 }, { "grad_norm": 0.19854965806007385, "learning_rate": 4.3038843410602016e-05, "loss": 0.0145, "step": 45380 }, { "grad_norm": 0.266191303730011, "learning_rate": 4.3018376921735774e-05, "loss": 0.013, "step": 45390 }, { "grad_norm": 0.23052754998207092, "learning_rate": 4.299791162583799e-05, "loss": 0.01, "step": 45400 }, { "grad_norm": 0.23945283889770508, "learning_rate": 4.29774475264056e-05, "loss": 0.013, "step": 45410 }, { "grad_norm": 0.1688099205493927, "learning_rate": 4.2956984626935365e-05, "loss": 0.0113, "step": 45420 }, { "grad_norm": 0.16835519671440125, "learning_rate": 4.293652293092383e-05, "loss": 0.011, "step": 45430 }, { "grad_norm": 0.17172479629516602, "learning_rate": 4.2916062441867324e-05, "loss": 0.0153, "step": 45440 }, { "grad_norm": 0.15922844409942627, "learning_rate": 4.289560316326201e-05, "loss": 0.0093, "step": 45450 }, { "grad_norm": 0.11024890840053558, "learning_rate": 4.28751450986038e-05, "loss": 0.0094, "step": 45460 }, { "grad_norm": 0.16755059361457825, "learning_rate": 4.2854688251388444e-05, "loss": 0.0098, "step": 45470 }, { "grad_norm": 0.18040287494659424, "learning_rate": 4.2834232625111425e-05, "loss": 0.0119, "step": 45480 }, { "grad_norm": 0.20917414128780365, "learning_rate": 4.2813778223268086e-05, "loss": 0.0115, "step": 45490 }, { "grad_norm": 0.236368790268898, "learning_rate": 4.2793325049353477e-05, "loss": 0.0115, "step": 45500 }, { "grad_norm": 0.20491698384284973, "learning_rate": 4.2772873106862535e-05, "loss": 0.0137, "step": 45510 }, { "grad_norm": 0.1343742311000824, "learning_rate": 4.275242239928993e-05, "loss": 0.0088, "step": 45520 }, { "grad_norm": 0.1892155408859253, "learning_rate": 4.273197293013009e-05, "loss": 0.011, "step": 45530 }, { "grad_norm": 0.24498356878757477, "learning_rate": 4.271152470287731e-05, "loss": 0.0114, "step": 45540 }, { "grad_norm": 0.13657715916633606, "learning_rate": 4.2691077721025594e-05, "loss": 0.0137, "step": 45550 }, { "grad_norm": 0.19283854961395264, "learning_rate": 4.267063198806883e-05, "loss": 0.0103, "step": 45560 }, { "grad_norm": 0.12401548773050308, "learning_rate": 4.2650187507500574e-05, "loss": 0.0113, "step": 45570 }, { "grad_norm": 0.2037477046251297, "learning_rate": 4.2629744282814275e-05, "loss": 0.014, "step": 45580 }, { "grad_norm": 0.1765618473291397, "learning_rate": 4.2609302317503074e-05, "loss": 0.0099, "step": 45590 }, { "grad_norm": 0.17258144915103912, "learning_rate": 4.258886161505999e-05, "loss": 0.0109, "step": 45600 }, { "grad_norm": 0.1868695318698883, "learning_rate": 4.2568422178977775e-05, "loss": 0.0112, "step": 45610 }, { "grad_norm": 0.17665967345237732, "learning_rate": 4.254798401274894e-05, "loss": 0.0114, "step": 45620 }, { "grad_norm": 0.24264897406101227, "learning_rate": 4.252754711986583e-05, "loss": 0.013, "step": 45630 }, { "grad_norm": 0.13024848699569702, "learning_rate": 4.250711150382052e-05, "loss": 0.008, "step": 45640 }, { "grad_norm": 0.12680716812610626, "learning_rate": 4.248667716810495e-05, "loss": 0.0075, "step": 45650 }, { "grad_norm": 0.1863763928413391, "learning_rate": 4.246624411621074e-05, "loss": 0.0105, "step": 45660 }, { "grad_norm": 0.20356231927871704, "learning_rate": 4.244581235162938e-05, "loss": 0.0112, "step": 45670 }, { "grad_norm": 0.12759371101856232, "learning_rate": 4.2425381877852075e-05, "loss": 0.009, "step": 45680 }, { "grad_norm": 0.12206628918647766, "learning_rate": 4.2404952698369856e-05, "loss": 0.011, "step": 45690 }, { "grad_norm": 0.22296862304210663, "learning_rate": 4.238452481667349e-05, "loss": 0.0105, "step": 45700 }, { "grad_norm": 0.14664588868618011, "learning_rate": 4.2364098236253526e-05, "loss": 0.011, "step": 45710 }, { "grad_norm": 0.12227997183799744, "learning_rate": 4.2343672960600356e-05, "loss": 0.0113, "step": 45720 }, { "grad_norm": 0.17922788858413696, "learning_rate": 4.232324899320406e-05, "loss": 0.01, "step": 45730 }, { "grad_norm": 0.16432306170463562, "learning_rate": 4.230282633755457e-05, "loss": 0.0096, "step": 45740 }, { "grad_norm": 0.1711791604757309, "learning_rate": 4.2282404997141515e-05, "loss": 0.0109, "step": 45750 }, { "grad_norm": 0.19721141457557678, "learning_rate": 4.2261984975454397e-05, "loss": 0.0111, "step": 45760 }, { "grad_norm": 0.11106406897306442, "learning_rate": 4.224156627598239e-05, "loss": 0.009, "step": 45770 }, { "grad_norm": 0.17166942358016968, "learning_rate": 4.222114890221453e-05, "loss": 0.0097, "step": 45780 }, { "grad_norm": 0.20985959470272064, "learning_rate": 4.2200732857639546e-05, "loss": 0.0111, "step": 45790 }, { "grad_norm": 0.1259932816028595, "learning_rate": 4.2180318145746035e-05, "loss": 0.0106, "step": 45800 }, { "grad_norm": 0.14073455333709717, "learning_rate": 4.215990477002227e-05, "loss": 0.0102, "step": 45810 }, { "grad_norm": 0.1737595498561859, "learning_rate": 4.2139492733956356e-05, "loss": 0.0098, "step": 45820 }, { "grad_norm": 0.12360317260026932, "learning_rate": 4.211908204103615e-05, "loss": 0.0096, "step": 45830 }, { "grad_norm": 0.1151604950428009, "learning_rate": 4.2098672694749265e-05, "loss": 0.0138, "step": 45840 }, { "grad_norm": 0.1773110181093216, "learning_rate": 4.2078264698583133e-05, "loss": 0.014, "step": 45850 }, { "grad_norm": 0.11246626824140549, "learning_rate": 4.205785805602488e-05, "loss": 0.0131, "step": 45860 }, { "grad_norm": 0.1453339159488678, "learning_rate": 4.203745277056149e-05, "loss": 0.0119, "step": 45870 }, { "grad_norm": 0.15506696701049805, "learning_rate": 4.201704884567964e-05, "loss": 0.0109, "step": 45880 }, { "grad_norm": 0.18647125363349915, "learning_rate": 4.1996646284865816e-05, "loss": 0.0097, "step": 45890 }, { "grad_norm": 0.13427816331386566, "learning_rate": 4.197624509160625e-05, "loss": 0.0089, "step": 45900 }, { "grad_norm": 0.1693539172410965, "learning_rate": 4.195584526938692e-05, "loss": 0.0096, "step": 45910 }, { "grad_norm": 0.1367332637310028, "learning_rate": 4.193544682169365e-05, "loss": 0.0108, "step": 45920 }, { "grad_norm": 0.11949004232883453, "learning_rate": 4.1915049752011946e-05, "loss": 0.0101, "step": 45930 }, { "grad_norm": 0.1395503431558609, "learning_rate": 4.189465406382712e-05, "loss": 0.0122, "step": 45940 }, { "grad_norm": 0.1861497014760971, "learning_rate": 4.187425976062422e-05, "loss": 0.0095, "step": 45950 }, { "grad_norm": 0.12974554300308228, "learning_rate": 4.18538668458881e-05, "loss": 0.0131, "step": 45960 }, { "grad_norm": 0.17585590481758118, "learning_rate": 4.183347532310333e-05, "loss": 0.0104, "step": 45970 }, { "grad_norm": 0.2104526162147522, "learning_rate": 4.181308519575429e-05, "loss": 0.012, "step": 45980 }, { "grad_norm": 0.16876360774040222, "learning_rate": 4.179269646732507e-05, "loss": 0.0112, "step": 45990 }, { "grad_norm": 0.1980876475572586, "learning_rate": 4.177230914129954e-05, "loss": 0.0088, "step": 46000 }, { "grad_norm": 0.1840796172618866, "learning_rate": 4.175192322116136e-05, "loss": 0.0093, "step": 46010 }, { "grad_norm": 0.29366233944892883, "learning_rate": 4.173153871039391e-05, "loss": 0.013, "step": 46020 }, { "grad_norm": 0.1779244840145111, "learning_rate": 4.171115561248036e-05, "loss": 0.0097, "step": 46030 }, { "grad_norm": 0.1539543867111206, "learning_rate": 4.16907739309036e-05, "loss": 0.0111, "step": 46040 }, { "grad_norm": 0.16396179795265198, "learning_rate": 4.167039366914633e-05, "loss": 0.0104, "step": 46050 }, { "grad_norm": 0.17355719208717346, "learning_rate": 4.165001483069096e-05, "loss": 0.0128, "step": 46060 }, { "grad_norm": 0.16671858727931976, "learning_rate": 4.16296374190197e-05, "loss": 0.0113, "step": 46070 }, { "grad_norm": 0.16091865301132202, "learning_rate": 4.1609261437614464e-05, "loss": 0.0102, "step": 46080 }, { "grad_norm": 0.18829867243766785, "learning_rate": 4.158888688995696e-05, "loss": 0.0143, "step": 46090 }, { "grad_norm": 0.13565780222415924, "learning_rate": 4.1568513779528645e-05, "loss": 0.0107, "step": 46100 }, { "grad_norm": 0.17250461876392365, "learning_rate": 4.1548142109810704e-05, "loss": 0.0109, "step": 46110 }, { "grad_norm": 0.18094737827777863, "learning_rate": 4.152777188428414e-05, "loss": 0.0114, "step": 46120 }, { "grad_norm": 0.139034703373909, "learning_rate": 4.1507403106429646e-05, "loss": 0.0101, "step": 46130 }, { "grad_norm": 0.17791292071342468, "learning_rate": 4.148703577972768e-05, "loss": 0.0094, "step": 46140 }, { "grad_norm": 0.13813865184783936, "learning_rate": 4.146666990765846e-05, "loss": 0.0091, "step": 46150 }, { "grad_norm": 0.14030858874320984, "learning_rate": 4.1446305493701986e-05, "loss": 0.0102, "step": 46160 }, { "grad_norm": 0.18211010098457336, "learning_rate": 4.142594254133796e-05, "loss": 0.0116, "step": 46170 }, { "grad_norm": 0.16032730042934418, "learning_rate": 4.1405581054045835e-05, "loss": 0.0089, "step": 46180 }, { "grad_norm": 0.2159392386674881, "learning_rate": 4.1385221035304864e-05, "loss": 0.0108, "step": 46190 }, { "grad_norm": 0.17635343968868256, "learning_rate": 4.136486248859398e-05, "loss": 0.0125, "step": 46200 }, { "grad_norm": 0.18050993978977203, "learning_rate": 4.1344505417391955e-05, "loss": 0.011, "step": 46210 }, { "grad_norm": 0.1551889330148697, "learning_rate": 4.132414982517721e-05, "loss": 0.0105, "step": 46220 }, { "grad_norm": 0.1570926010608673, "learning_rate": 4.130379571542798e-05, "loss": 0.0095, "step": 46230 }, { "grad_norm": 0.22486479580402374, "learning_rate": 4.12834430916222e-05, "loss": 0.0086, "step": 46240 }, { "grad_norm": 0.18021611869335175, "learning_rate": 4.126309195723763e-05, "loss": 0.0109, "step": 46250 }, { "grad_norm": 0.22866132855415344, "learning_rate": 4.124274231575168e-05, "loss": 0.0094, "step": 46260 }, { "grad_norm": 0.19340942800045013, "learning_rate": 4.122239417064154e-05, "loss": 0.0084, "step": 46270 }, { "grad_norm": 0.11439710110425949, "learning_rate": 4.1202047525384184e-05, "loss": 0.0088, "step": 46280 }, { "grad_norm": 0.15810959041118622, "learning_rate": 4.118170238345627e-05, "loss": 0.0134, "step": 46290 }, { "grad_norm": 0.2053525149822235, "learning_rate": 4.1161358748334256e-05, "loss": 0.0102, "step": 46300 }, { "grad_norm": 0.11215868592262268, "learning_rate": 4.1141016623494266e-05, "loss": 0.0109, "step": 46310 }, { "grad_norm": 0.16979315876960754, "learning_rate": 4.112067601241227e-05, "loss": 0.01, "step": 46320 }, { "grad_norm": 0.20315773785114288, "learning_rate": 4.110033691856387e-05, "loss": 0.0092, "step": 46330 }, { "grad_norm": 0.20027433335781097, "learning_rate": 4.107999934542451e-05, "loss": 0.0108, "step": 46340 }, { "grad_norm": 0.18789473176002502, "learning_rate": 4.105966329646928e-05, "loss": 0.01, "step": 46350 }, { "grad_norm": 0.16417323052883148, "learning_rate": 4.103932877517308e-05, "loss": 0.0103, "step": 46360 }, { "grad_norm": 0.13244378566741943, "learning_rate": 4.101899578501052e-05, "loss": 0.0096, "step": 46370 }, { "grad_norm": 0.1663385033607483, "learning_rate": 4.099866432945595e-05, "loss": 0.0102, "step": 46380 }, { "grad_norm": 0.17160050570964813, "learning_rate": 4.097833441198346e-05, "loss": 0.0113, "step": 46390 }, { "grad_norm": 0.17163148522377014, "learning_rate": 4.095800603606685e-05, "loss": 0.0087, "step": 46400 }, { "grad_norm": 0.11052019149065018, "learning_rate": 4.093767920517975e-05, "loss": 0.0082, "step": 46410 }, { "grad_norm": 0.11372335255146027, "learning_rate": 4.091735392279539e-05, "loss": 0.0092, "step": 46420 }, { "grad_norm": 0.16266421973705292, "learning_rate": 4.089703019238685e-05, "loss": 0.0135, "step": 46430 }, { "grad_norm": 0.17307636141777039, "learning_rate": 4.0876708017426866e-05, "loss": 0.0123, "step": 46440 }, { "grad_norm": 0.14238035678863525, "learning_rate": 4.085638740138798e-05, "loss": 0.0088, "step": 46450 }, { "grad_norm": 0.1909932941198349, "learning_rate": 4.08360683477424e-05, "loss": 0.012, "step": 46460 }, { "grad_norm": 0.24571527540683746, "learning_rate": 4.0815750859962085e-05, "loss": 0.0123, "step": 46470 }, { "grad_norm": 0.20869502425193787, "learning_rate": 4.079543494151879e-05, "loss": 0.0121, "step": 46480 }, { "grad_norm": 0.16892796754837036, "learning_rate": 4.07751205958839e-05, "loss": 0.0089, "step": 46490 }, { "grad_norm": 0.13868914544582367, "learning_rate": 4.0754807826528615e-05, "loss": 0.0116, "step": 46500 }, { "grad_norm": 0.20603050291538239, "learning_rate": 4.073449663692379e-05, "loss": 0.0111, "step": 46510 }, { "grad_norm": 0.15595506131649017, "learning_rate": 4.07141870305401e-05, "loss": 0.0108, "step": 46520 }, { "grad_norm": 0.1918809860944748, "learning_rate": 4.0693879010847866e-05, "loss": 0.0097, "step": 46530 }, { "grad_norm": 0.14240778982639313, "learning_rate": 4.06735725813172e-05, "loss": 0.0093, "step": 46540 }, { "grad_norm": 0.14832928776741028, "learning_rate": 4.0653267745417903e-05, "loss": 0.0112, "step": 46550 }, { "grad_norm": 0.1507827192544937, "learning_rate": 4.063296450661949e-05, "loss": 0.0148, "step": 46560 }, { "grad_norm": 0.13079161942005157, "learning_rate": 4.061266286839128e-05, "loss": 0.0093, "step": 46570 }, { "grad_norm": 0.12008380144834518, "learning_rate": 4.0592362834202225e-05, "loss": 0.0101, "step": 46580 }, { "grad_norm": 0.12890858948230743, "learning_rate": 4.057206440752107e-05, "loss": 0.0129, "step": 46590 }, { "grad_norm": 0.1480385661125183, "learning_rate": 4.0551767591816245e-05, "loss": 0.0136, "step": 46600 }, { "grad_norm": 0.20652516186237335, "learning_rate": 4.0531472390555935e-05, "loss": 0.0135, "step": 46610 }, { "grad_norm": 0.1419609934091568, "learning_rate": 4.051117880720802e-05, "loss": 0.01, "step": 46620 }, { "grad_norm": 0.14606618881225586, "learning_rate": 4.049088684524015e-05, "loss": 0.01, "step": 46630 }, { "grad_norm": 0.15794813632965088, "learning_rate": 4.0470596508119636e-05, "loss": 0.0109, "step": 46640 }, { "grad_norm": 0.2390931248664856, "learning_rate": 4.0450307799313524e-05, "loss": 0.0116, "step": 46650 }, { "grad_norm": 0.18077634274959564, "learning_rate": 4.0430020722288656e-05, "loss": 0.0086, "step": 46660 }, { "grad_norm": 0.19351357221603394, "learning_rate": 4.040973528051148e-05, "loss": 0.0125, "step": 46670 }, { "grad_norm": 0.2063419669866562, "learning_rate": 4.038945147744827e-05, "loss": 0.0122, "step": 46680 }, { "grad_norm": 0.20142655074596405, "learning_rate": 4.0369169316564945e-05, "loss": 0.0114, "step": 46690 }, { "grad_norm": 0.07511163502931595, "learning_rate": 4.034888880132718e-05, "loss": 0.0082, "step": 46700 }, { "grad_norm": 0.17036019265651703, "learning_rate": 4.032860993520035e-05, "loss": 0.0114, "step": 46710 }, { "grad_norm": 0.1178407296538353, "learning_rate": 4.030833272164959e-05, "loss": 0.0093, "step": 46720 }, { "grad_norm": 0.13129261136054993, "learning_rate": 4.0288057164139705e-05, "loss": 0.0105, "step": 46730 }, { "grad_norm": 0.1410490870475769, "learning_rate": 4.0267783266135205e-05, "loss": 0.0115, "step": 46740 }, { "grad_norm": 0.1703910529613495, "learning_rate": 4.024751103110039e-05, "loss": 0.0103, "step": 46750 }, { "grad_norm": 0.14309975504875183, "learning_rate": 4.0227240462499176e-05, "loss": 0.009, "step": 46760 }, { "grad_norm": 0.21560212969779968, "learning_rate": 4.020697156379531e-05, "loss": 0.0103, "step": 46770 }, { "grad_norm": 0.20232698321342468, "learning_rate": 4.018670433845215e-05, "loss": 0.0114, "step": 46780 }, { "grad_norm": 0.15957942605018616, "learning_rate": 4.016643878993284e-05, "loss": 0.0089, "step": 46790 }, { "grad_norm": 0.1284429281949997, "learning_rate": 4.014617492170017e-05, "loss": 0.0106, "step": 46800 }, { "grad_norm": 0.17304164171218872, "learning_rate": 4.0125912737216726e-05, "loss": 0.0119, "step": 46810 }, { "grad_norm": 0.09431274980306625, "learning_rate": 4.0105652239944735e-05, "loss": 0.0098, "step": 46820 }, { "grad_norm": 0.19360540807247162, "learning_rate": 4.0085393433346144e-05, "loss": 0.0119, "step": 46830 }, { "grad_norm": 0.10246790200471878, "learning_rate": 4.006513632088268e-05, "loss": 0.0115, "step": 46840 }, { "grad_norm": 0.1482086479663849, "learning_rate": 4.004488090601567e-05, "loss": 0.0107, "step": 46850 }, { "grad_norm": 0.1383722871541977, "learning_rate": 4.002462719220626e-05, "loss": 0.0083, "step": 46860 }, { "grad_norm": 0.11712215840816498, "learning_rate": 4.000437518291522e-05, "loss": 0.0097, "step": 46870 }, { "grad_norm": 0.17484962940216064, "learning_rate": 3.9984124881603094e-05, "loss": 0.0104, "step": 46880 }, { "grad_norm": 0.12055528908967972, "learning_rate": 3.9963876291730086e-05, "loss": 0.0081, "step": 46890 }, { "grad_norm": 0.14219054579734802, "learning_rate": 3.994362941675614e-05, "loss": 0.0086, "step": 46900 }, { "grad_norm": 0.11036362498998642, "learning_rate": 3.992338426014088e-05, "loss": 0.0089, "step": 46910 }, { "grad_norm": 0.13148926198482513, "learning_rate": 3.9903140825343636e-05, "loss": 0.0118, "step": 46920 }, { "grad_norm": 0.10244853794574738, "learning_rate": 3.98828991158235e-05, "loss": 0.0116, "step": 46930 }, { "grad_norm": 0.13433296978473663, "learning_rate": 3.9862659135039185e-05, "loss": 0.0124, "step": 46940 }, { "grad_norm": 0.19516116380691528, "learning_rate": 3.984242088644918e-05, "loss": 0.0101, "step": 46950 }, { "grad_norm": 0.1596812903881073, "learning_rate": 3.9822184373511615e-05, "loss": 0.0109, "step": 46960 }, { "grad_norm": 0.16004805266857147, "learning_rate": 3.98019495996844e-05, "loss": 0.0102, "step": 46970 }, { "grad_norm": 0.17497988045215607, "learning_rate": 3.978171656842507e-05, "loss": 0.0118, "step": 46980 }, { "grad_norm": 0.196904256939888, "learning_rate": 3.976148528319091e-05, "loss": 0.0096, "step": 46990 }, { "grad_norm": 0.12186874449253082, "learning_rate": 3.974125574743888e-05, "loss": 0.0099, "step": 47000 }, { "grad_norm": 0.190314382314682, "learning_rate": 3.9721027964625686e-05, "loss": 0.0093, "step": 47010 }, { "grad_norm": 0.1413278430700302, "learning_rate": 3.9700801938207676e-05, "loss": 0.0081, "step": 47020 }, { "grad_norm": 0.15673764050006866, "learning_rate": 3.9680577671640916e-05, "loss": 0.012, "step": 47030 }, { "grad_norm": 0.17924873530864716, "learning_rate": 3.966035516838121e-05, "loss": 0.0113, "step": 47040 }, { "grad_norm": 0.2374975085258484, "learning_rate": 3.9640134431884014e-05, "loss": 0.0113, "step": 47050 }, { "grad_norm": 0.13535475730895996, "learning_rate": 3.961991546560451e-05, "loss": 0.0114, "step": 47060 }, { "grad_norm": 0.2592535614967346, "learning_rate": 3.959969827299753e-05, "loss": 0.0123, "step": 47070 }, { "grad_norm": 0.1920681744813919, "learning_rate": 3.9579482857517684e-05, "loss": 0.0117, "step": 47080 }, { "grad_norm": 0.22208847105503082, "learning_rate": 3.955926922261921e-05, "loss": 0.0125, "step": 47090 }, { "grad_norm": 0.13829176127910614, "learning_rate": 3.9539057371756084e-05, "loss": 0.01, "step": 47100 }, { "grad_norm": 0.22630594670772552, "learning_rate": 3.951884730838195e-05, "loss": 0.0119, "step": 47110 }, { "grad_norm": 0.15255826711654663, "learning_rate": 3.949863903595012e-05, "loss": 0.0104, "step": 47120 }, { "grad_norm": 0.16182434558868408, "learning_rate": 3.947843255791369e-05, "loss": 0.0155, "step": 47130 }, { "grad_norm": 0.1573086977005005, "learning_rate": 3.9458227877725364e-05, "loss": 0.0091, "step": 47140 }, { "grad_norm": 0.14829273521900177, "learning_rate": 3.943802499883758e-05, "loss": 0.01, "step": 47150 }, { "grad_norm": 0.17219586670398712, "learning_rate": 3.9417823924702437e-05, "loss": 0.0098, "step": 47160 }, { "grad_norm": 0.18441693484783173, "learning_rate": 3.939762465877178e-05, "loss": 0.0095, "step": 47170 }, { "grad_norm": 0.1231648325920105, "learning_rate": 3.937742720449708e-05, "loss": 0.008, "step": 47180 }, { "grad_norm": 0.15643595159053802, "learning_rate": 3.9357231565329563e-05, "loss": 0.0112, "step": 47190 }, { "grad_norm": 0.15718500316143036, "learning_rate": 3.933703774472008e-05, "loss": 0.0101, "step": 47200 }, { "grad_norm": 0.1683652400970459, "learning_rate": 3.93168457461192e-05, "loss": 0.01, "step": 47210 }, { "grad_norm": 0.14697429537773132, "learning_rate": 3.9296655572977216e-05, "loss": 0.0104, "step": 47220 }, { "grad_norm": 0.1823432892560959, "learning_rate": 3.927646722874404e-05, "loss": 0.0126, "step": 47230 }, { "grad_norm": 0.17178380489349365, "learning_rate": 3.925628071686934e-05, "loss": 0.011, "step": 47240 }, { "grad_norm": 0.16604259610176086, "learning_rate": 3.9236096040802415e-05, "loss": 0.0114, "step": 47250 }, { "grad_norm": 0.16303569078445435, "learning_rate": 3.9215913203992294e-05, "loss": 0.0078, "step": 47260 }, { "grad_norm": 0.1887277215719223, "learning_rate": 3.9195732209887645e-05, "loss": 0.0136, "step": 47270 }, { "grad_norm": 0.11992961913347244, "learning_rate": 3.9175553061936875e-05, "loss": 0.0101, "step": 47280 }, { "grad_norm": 0.21502958238124847, "learning_rate": 3.9155375763588045e-05, "loss": 0.0119, "step": 47290 }, { "grad_norm": 0.2001992017030716, "learning_rate": 3.913520031828889e-05, "loss": 0.0122, "step": 47300 }, { "grad_norm": 0.12095862627029419, "learning_rate": 3.911502672948685e-05, "loss": 0.0115, "step": 47310 }, { "grad_norm": 0.16518911719322205, "learning_rate": 3.9094855000629014e-05, "loss": 0.0122, "step": 47320 }, { "grad_norm": 0.22762010991573334, "learning_rate": 3.907468513516223e-05, "loss": 0.012, "step": 47330 }, { "grad_norm": 0.2514907419681549, "learning_rate": 3.905451713653294e-05, "loss": 0.0109, "step": 47340 }, { "grad_norm": 0.1636316329240799, "learning_rate": 3.903435100818731e-05, "loss": 0.0071, "step": 47350 }, { "grad_norm": 0.12021040916442871, "learning_rate": 3.901418675357117e-05, "loss": 0.0081, "step": 47360 }, { "grad_norm": 0.16777929663658142, "learning_rate": 3.8994024376130075e-05, "loss": 0.0091, "step": 47370 }, { "grad_norm": 0.2447761595249176, "learning_rate": 3.8973863879309194e-05, "loss": 0.0107, "step": 47380 }, { "grad_norm": 0.16887986660003662, "learning_rate": 3.8953705266553394e-05, "loss": 0.0095, "step": 47390 }, { "grad_norm": 0.15245769917964935, "learning_rate": 3.893354854130727e-05, "loss": 0.0095, "step": 47400 }, { "grad_norm": 0.1069255992770195, "learning_rate": 3.8913393707015006e-05, "loss": 0.0086, "step": 47410 }, { "grad_norm": 0.23366157710552216, "learning_rate": 3.889324076712056e-05, "loss": 0.0145, "step": 47420 }, { "grad_norm": 0.15441809594631195, "learning_rate": 3.8873089725067476e-05, "loss": 0.0119, "step": 47430 }, { "grad_norm": 0.14032235741615295, "learning_rate": 3.885294058429905e-05, "loss": 0.0099, "step": 47440 }, { "grad_norm": 0.17386946082115173, "learning_rate": 3.8832793348258206e-05, "loss": 0.009, "step": 47450 }, { "grad_norm": 0.13192275166511536, "learning_rate": 3.881264802038756e-05, "loss": 0.0126, "step": 47460 }, { "grad_norm": 0.1757449209690094, "learning_rate": 3.879250460412939e-05, "loss": 0.0107, "step": 47470 }, { "grad_norm": 0.14620603621006012, "learning_rate": 3.8772363102925644e-05, "loss": 0.0095, "step": 47480 }, { "grad_norm": 0.2264312207698822, "learning_rate": 3.875222352021798e-05, "loss": 0.0123, "step": 47490 }, { "grad_norm": 0.14079739153385162, "learning_rate": 3.8732085859447686e-05, "loss": 0.0113, "step": 47500 }, { "grad_norm": 0.16071952879428864, "learning_rate": 3.871195012405575e-05, "loss": 0.0114, "step": 47510 }, { "grad_norm": 0.25993555784225464, "learning_rate": 3.869181631748278e-05, "loss": 0.0123, "step": 47520 }, { "grad_norm": 0.38268136978149414, "learning_rate": 3.867168444316915e-05, "loss": 0.0074, "step": 47530 }, { "grad_norm": 0.19072262942790985, "learning_rate": 3.865155450455481e-05, "loss": 0.009, "step": 47540 }, { "grad_norm": 0.31113603711128235, "learning_rate": 3.8631426505079426e-05, "loss": 0.0127, "step": 47550 }, { "grad_norm": 0.14742648601531982, "learning_rate": 3.8611300448182304e-05, "loss": 0.0128, "step": 47560 }, { "grad_norm": 0.25317269563674927, "learning_rate": 3.859117633730248e-05, "loss": 0.0133, "step": 47570 }, { "grad_norm": 0.2720481753349304, "learning_rate": 3.857105417587858e-05, "loss": 0.0109, "step": 47580 }, { "grad_norm": 0.22045475244522095, "learning_rate": 3.855093396734894e-05, "loss": 0.0104, "step": 47590 }, { "grad_norm": 0.11871044337749481, "learning_rate": 3.8530815715151545e-05, "loss": 0.0088, "step": 47600 }, { "grad_norm": 0.15697553753852844, "learning_rate": 3.851069942272405e-05, "loss": 0.0091, "step": 47610 }, { "grad_norm": 0.1857444941997528, "learning_rate": 3.849058509350382e-05, "loss": 0.0151, "step": 47620 }, { "grad_norm": 0.1863778531551361, "learning_rate": 3.8470472730927783e-05, "loss": 0.0103, "step": 47630 }, { "grad_norm": 0.15481802821159363, "learning_rate": 3.845036233843264e-05, "loss": 0.0108, "step": 47640 }, { "grad_norm": 0.16423827409744263, "learning_rate": 3.843025391945469e-05, "loss": 0.013, "step": 47650 }, { "grad_norm": 0.10237938910722733, "learning_rate": 3.841014747742992e-05, "loss": 0.0112, "step": 47660 }, { "grad_norm": 0.15254928171634674, "learning_rate": 3.839004301579397e-05, "loss": 0.0101, "step": 47670 }, { "grad_norm": 0.15596774220466614, "learning_rate": 3.8369940537982097e-05, "loss": 0.0091, "step": 47680 }, { "grad_norm": 0.12520164251327515, "learning_rate": 3.834984004742933e-05, "loss": 0.0108, "step": 47690 }, { "grad_norm": 0.14887844026088715, "learning_rate": 3.832974154757026e-05, "loss": 0.01, "step": 47700 }, { "grad_norm": 0.09087371081113815, "learning_rate": 3.830964504183919e-05, "loss": 0.0089, "step": 47710 }, { "grad_norm": 0.1381336897611618, "learning_rate": 3.828955053367003e-05, "loss": 0.0101, "step": 47720 }, { "grad_norm": 0.1741122156381607, "learning_rate": 3.826945802649642e-05, "loss": 0.0095, "step": 47730 }, { "grad_norm": 0.15303358435630798, "learning_rate": 3.824936752375159e-05, "loss": 0.0103, "step": 47740 }, { "grad_norm": 0.12425360083580017, "learning_rate": 3.822927902886848e-05, "loss": 0.0085, "step": 47750 }, { "grad_norm": 0.11107776314020157, "learning_rate": 3.8209192545279653e-05, "loss": 0.0085, "step": 47760 }, { "grad_norm": 0.1571808159351349, "learning_rate": 3.8189108076417326e-05, "loss": 0.0104, "step": 47770 }, { "grad_norm": 0.10450328886508942, "learning_rate": 3.816902562571342e-05, "loss": 0.0075, "step": 47780 }, { "grad_norm": 0.22902801632881165, "learning_rate": 3.814894519659944e-05, "loss": 0.0141, "step": 47790 }, { "grad_norm": 0.15494146943092346, "learning_rate": 3.812886679250661e-05, "loss": 0.0107, "step": 47800 }, { "grad_norm": 0.10161469876766205, "learning_rate": 3.810879041686575e-05, "loss": 0.0102, "step": 47810 }, { "grad_norm": 0.12771053612232208, "learning_rate": 3.808871607310741e-05, "loss": 0.0094, "step": 47820 }, { "grad_norm": 0.17328616976737976, "learning_rate": 3.806864376466169e-05, "loss": 0.0083, "step": 47830 }, { "grad_norm": 0.24971003830432892, "learning_rate": 3.804857349495845e-05, "loss": 0.009, "step": 47840 }, { "grad_norm": 0.14189742505550385, "learning_rate": 3.802850526742713e-05, "loss": 0.0087, "step": 47850 }, { "grad_norm": 0.20522241294384003, "learning_rate": 3.800843908549683e-05, "loss": 0.0104, "step": 47860 }, { "grad_norm": 0.1373799741268158, "learning_rate": 3.7988374952596325e-05, "loss": 0.0113, "step": 47870 }, { "grad_norm": 0.13116642832756042, "learning_rate": 3.7968312872154e-05, "loss": 0.0098, "step": 47880 }, { "grad_norm": 0.15952721238136292, "learning_rate": 3.7948252847597965e-05, "loss": 0.0102, "step": 47890 }, { "grad_norm": 0.1540338397026062, "learning_rate": 3.7928194882355885e-05, "loss": 0.0109, "step": 47900 }, { "grad_norm": 0.12819503247737885, "learning_rate": 3.790813897985515e-05, "loss": 0.0103, "step": 47910 }, { "grad_norm": 0.15699464082717896, "learning_rate": 3.7888085143522726e-05, "loss": 0.0094, "step": 47920 }, { "grad_norm": 0.21010319888591766, "learning_rate": 3.7868033376785314e-05, "loss": 0.0144, "step": 47930 }, { "grad_norm": 0.21210332214832306, "learning_rate": 3.784798368306919e-05, "loss": 0.0117, "step": 47940 }, { "grad_norm": 0.18684698641300201, "learning_rate": 3.782793606580029e-05, "loss": 0.0122, "step": 47950 }, { "grad_norm": 0.2099505066871643, "learning_rate": 3.7807890528404205e-05, "loss": 0.0117, "step": 47960 }, { "grad_norm": 0.18753103911876678, "learning_rate": 3.778784707430616e-05, "loss": 0.0138, "step": 47970 }, { "grad_norm": 0.14915326237678528, "learning_rate": 3.776780570693107e-05, "loss": 0.0123, "step": 47980 }, { "grad_norm": 0.10331472754478455, "learning_rate": 3.774776642970342e-05, "loss": 0.0093, "step": 47990 }, { "grad_norm": 0.1760881394147873, "learning_rate": 3.77277292460474e-05, "loss": 0.0103, "step": 48000 }, { "grad_norm": 0.21665973961353302, "learning_rate": 3.770769415938678e-05, "loss": 0.0123, "step": 48010 }, { "grad_norm": 0.1226678341627121, "learning_rate": 3.768766117314506e-05, "loss": 0.0101, "step": 48020 }, { "grad_norm": 0.1298130452632904, "learning_rate": 3.76676302907453e-05, "loss": 0.0109, "step": 48030 }, { "grad_norm": 0.22237950563430786, "learning_rate": 3.764760151561021e-05, "loss": 0.0104, "step": 48040 }, { "grad_norm": 0.15988315641880035, "learning_rate": 3.76275748511622e-05, "loss": 0.0103, "step": 48050 }, { "grad_norm": 0.17441776394844055, "learning_rate": 3.7607550300823255e-05, "loss": 0.0111, "step": 48060 }, { "grad_norm": 0.14484384655952454, "learning_rate": 3.7587527868015044e-05, "loss": 0.0104, "step": 48070 }, { "grad_norm": 0.15575668215751648, "learning_rate": 3.756750755615881e-05, "loss": 0.0102, "step": 48080 }, { "grad_norm": 0.14075031876564026, "learning_rate": 3.754748936867553e-05, "loss": 0.0138, "step": 48090 }, { "grad_norm": 0.10664781183004379, "learning_rate": 3.752747330898573e-05, "loss": 0.0085, "step": 48100 }, { "grad_norm": 0.12695243954658508, "learning_rate": 3.750745938050962e-05, "loss": 0.009, "step": 48110 }, { "grad_norm": 0.179496169090271, "learning_rate": 3.7487447586667025e-05, "loss": 0.0141, "step": 48120 }, { "grad_norm": 0.19222573935985565, "learning_rate": 3.7467437930877426e-05, "loss": 0.0122, "step": 48130 }, { "grad_norm": 0.12214070558547974, "learning_rate": 3.744743041655992e-05, "loss": 0.0087, "step": 48140 }, { "grad_norm": 0.22211278975009918, "learning_rate": 3.742742504713324e-05, "loss": 0.0083, "step": 48150 }, { "grad_norm": 0.17807415127754211, "learning_rate": 3.740742182601576e-05, "loss": 0.0106, "step": 48160 }, { "grad_norm": 0.19623900949954987, "learning_rate": 3.7387420756625464e-05, "loss": 0.0103, "step": 48170 }, { "grad_norm": 0.16798563301563263, "learning_rate": 3.736742184238002e-05, "loss": 0.0098, "step": 48180 }, { "grad_norm": 0.16753898561000824, "learning_rate": 3.7347425086696684e-05, "loss": 0.0099, "step": 48190 }, { "grad_norm": 0.183559387922287, "learning_rate": 3.732743049299235e-05, "loss": 0.0147, "step": 48200 }, { "grad_norm": 0.1377570629119873, "learning_rate": 3.730743806468354e-05, "loss": 0.0085, "step": 48210 }, { "grad_norm": 0.1716151088476181, "learning_rate": 3.7287447805186436e-05, "loss": 0.0141, "step": 48220 }, { "grad_norm": 0.16939841210842133, "learning_rate": 3.726745971791682e-05, "loss": 0.0096, "step": 48230 }, { "grad_norm": 0.15394209325313568, "learning_rate": 3.724747380629008e-05, "loss": 0.0113, "step": 48240 }, { "grad_norm": 0.16006939113140106, "learning_rate": 3.72274900737213e-05, "loss": 0.0087, "step": 48250 }, { "grad_norm": 0.14047938585281372, "learning_rate": 3.7207508523625123e-05, "loss": 0.0114, "step": 48260 }, { "grad_norm": 0.302266001701355, "learning_rate": 3.718752915941588e-05, "loss": 0.0104, "step": 48270 }, { "grad_norm": 0.1653343141078949, "learning_rate": 3.7167551984507464e-05, "loss": 0.0093, "step": 48280 }, { "grad_norm": 0.17029689252376556, "learning_rate": 3.714757700231346e-05, "loss": 0.0104, "step": 48290 }, { "grad_norm": 0.19530069828033447, "learning_rate": 3.712760421624703e-05, "loss": 0.0122, "step": 48300 }, { "grad_norm": 0.17122742533683777, "learning_rate": 3.710763362972099e-05, "loss": 0.0145, "step": 48310 }, { "grad_norm": 0.17320825159549713, "learning_rate": 3.708766524614774e-05, "loss": 0.0101, "step": 48320 }, { "grad_norm": 0.1838853657245636, "learning_rate": 3.7067699068939335e-05, "loss": 0.0114, "step": 48330 }, { "grad_norm": 0.13364331424236298, "learning_rate": 3.704773510150748e-05, "loss": 0.0104, "step": 48340 }, { "grad_norm": 0.2644636929035187, "learning_rate": 3.702777334726344e-05, "loss": 0.0102, "step": 48350 }, { "grad_norm": 0.25194546580314636, "learning_rate": 3.7007813809618164e-05, "loss": 0.0117, "step": 48360 }, { "grad_norm": 0.15134136378765106, "learning_rate": 3.6987856491982145e-05, "loss": 0.0064, "step": 48370 }, { "grad_norm": 0.10562381893396378, "learning_rate": 3.696790139776558e-05, "loss": 0.0077, "step": 48380 }, { "grad_norm": 0.15999332070350647, "learning_rate": 3.6947948530378235e-05, "loss": 0.0129, "step": 48390 }, { "grad_norm": 0.15334029495716095, "learning_rate": 3.6927997893229516e-05, "loss": 0.0125, "step": 48400 }, { "grad_norm": 0.1502513885498047, "learning_rate": 3.690804948972845e-05, "loss": 0.0087, "step": 48410 }, { "grad_norm": 0.1577829122543335, "learning_rate": 3.688810332328363e-05, "loss": 0.0078, "step": 48420 }, { "grad_norm": 0.16018059849739075, "learning_rate": 3.686815939730336e-05, "loss": 0.0101, "step": 48430 }, { "grad_norm": 0.14863747358322144, "learning_rate": 3.684821771519548e-05, "loss": 0.0098, "step": 48440 }, { "grad_norm": 0.17387598752975464, "learning_rate": 3.68282782803675e-05, "loss": 0.0133, "step": 48450 }, { "grad_norm": 0.11117701232433319, "learning_rate": 3.6808341096226504e-05, "loss": 0.0113, "step": 48460 }, { "grad_norm": 0.08047572523355484, "learning_rate": 3.678840616617924e-05, "loss": 0.0086, "step": 48470 }, { "grad_norm": 0.1548406034708023, "learning_rate": 3.6768473493632e-05, "loss": 0.0084, "step": 48480 }, { "grad_norm": 0.1644556224346161, "learning_rate": 3.6748543081990783e-05, "loss": 0.0085, "step": 48490 }, { "grad_norm": 0.14680896699428558, "learning_rate": 3.672861493466112e-05, "loss": 0.0089, "step": 48500 }, { "grad_norm": 0.12126560509204865, "learning_rate": 3.670868905504818e-05, "loss": 0.0066, "step": 48510 }, { "grad_norm": 0.10732558369636536, "learning_rate": 3.6688765446556784e-05, "loss": 0.0078, "step": 48520 }, { "grad_norm": 0.13210588693618774, "learning_rate": 3.6668844112591276e-05, "loss": 0.0087, "step": 48530 }, { "grad_norm": 0.14319318532943726, "learning_rate": 3.664892505655573e-05, "loss": 0.0098, "step": 48540 }, { "grad_norm": 0.17078804969787598, "learning_rate": 3.662900828185373e-05, "loss": 0.0102, "step": 48550 }, { "grad_norm": 0.1458834409713745, "learning_rate": 3.6609093791888516e-05, "loss": 0.0082, "step": 48560 }, { "grad_norm": 0.2196558266878128, "learning_rate": 3.658918159006292e-05, "loss": 0.0089, "step": 48570 }, { "grad_norm": 0.14856386184692383, "learning_rate": 3.656927167977942e-05, "loss": 0.0098, "step": 48580 }, { "grad_norm": 0.1299753487110138, "learning_rate": 3.654936406444006e-05, "loss": 0.0118, "step": 48590 }, { "grad_norm": 0.19059468805789948, "learning_rate": 3.65294587474465e-05, "loss": 0.0121, "step": 48600 }, { "grad_norm": 0.15028947591781616, "learning_rate": 3.650955573220002e-05, "loss": 0.0122, "step": 48610 }, { "grad_norm": 0.1512671858072281, "learning_rate": 3.648965502210149e-05, "loss": 0.0096, "step": 48620 }, { "grad_norm": 0.16623680293560028, "learning_rate": 3.646975662055142e-05, "loss": 0.0108, "step": 48630 }, { "grad_norm": 0.2245660275220871, "learning_rate": 3.644986053094987e-05, "loss": 0.0087, "step": 48640 }, { "grad_norm": 0.1315881907939911, "learning_rate": 3.642996675669659e-05, "loss": 0.0092, "step": 48650 }, { "grad_norm": 0.13329415023326874, "learning_rate": 3.641007530119083e-05, "loss": 0.0092, "step": 48660 }, { "grad_norm": 0.11741691082715988, "learning_rate": 3.639018616783153e-05, "loss": 0.0091, "step": 48670 }, { "grad_norm": 0.2038756161928177, "learning_rate": 3.637029936001719e-05, "loss": 0.0131, "step": 48680 }, { "grad_norm": 0.15230558812618256, "learning_rate": 3.6350414881145886e-05, "loss": 0.0101, "step": 48690 }, { "grad_norm": 0.15089410543441772, "learning_rate": 3.6330532734615386e-05, "loss": 0.0111, "step": 48700 }, { "grad_norm": 0.13718879222869873, "learning_rate": 3.6310652923822975e-05, "loss": 0.0101, "step": 48710 }, { "grad_norm": 0.16197240352630615, "learning_rate": 3.629077545216558e-05, "loss": 0.0102, "step": 48720 }, { "grad_norm": 0.15081535279750824, "learning_rate": 3.627090032303969e-05, "loss": 0.0111, "step": 48730 }, { "grad_norm": 0.13379259407520294, "learning_rate": 3.625102753984146e-05, "loss": 0.0091, "step": 48740 }, { "grad_norm": 0.12501485645771027, "learning_rate": 3.623115710596659e-05, "loss": 0.0081, "step": 48750 }, { "grad_norm": 0.18060682713985443, "learning_rate": 3.6211289024810395e-05, "loss": 0.0117, "step": 48760 }, { "grad_norm": 0.14502771198749542, "learning_rate": 3.619142329976777e-05, "loss": 0.0099, "step": 48770 }, { "grad_norm": 0.23579005897045135, "learning_rate": 3.6171559934233247e-05, "loss": 0.0111, "step": 48780 }, { "grad_norm": 0.13869212567806244, "learning_rate": 3.615169893160093e-05, "loss": 0.0094, "step": 48790 }, { "grad_norm": 0.12448867410421371, "learning_rate": 3.61318402952645e-05, "loss": 0.0114, "step": 48800 }, { "grad_norm": 0.13936784863471985, "learning_rate": 3.6111984028617285e-05, "loss": 0.0145, "step": 48810 }, { "grad_norm": 0.16121411323547363, "learning_rate": 3.6092130135052134e-05, "loss": 0.0093, "step": 48820 }, { "grad_norm": 0.14251816272735596, "learning_rate": 3.6072278617961584e-05, "loss": 0.0087, "step": 48830 }, { "grad_norm": 0.15724080801010132, "learning_rate": 3.605242948073767e-05, "loss": 0.0078, "step": 48840 }, { "grad_norm": 0.13312527537345886, "learning_rate": 3.603258272677212e-05, "loss": 0.0083, "step": 48850 }, { "grad_norm": 0.1784452497959137, "learning_rate": 3.601273835945616e-05, "loss": 0.0084, "step": 48860 }, { "grad_norm": 0.23001588881015778, "learning_rate": 3.5992896382180664e-05, "loss": 0.0106, "step": 48870 }, { "grad_norm": 0.2103416919708252, "learning_rate": 3.597305679833609e-05, "loss": 0.0119, "step": 48880 }, { "grad_norm": 0.16484244167804718, "learning_rate": 3.595321961131245e-05, "loss": 0.0118, "step": 48890 }, { "grad_norm": 0.17399044334888458, "learning_rate": 3.593338482449942e-05, "loss": 0.01, "step": 48900 }, { "grad_norm": 0.17628341913223267, "learning_rate": 3.591355244128618e-05, "loss": 0.0081, "step": 48910 }, { "grad_norm": 0.11717627942562103, "learning_rate": 3.589372246506158e-05, "loss": 0.0077, "step": 48920 }, { "grad_norm": 0.12637388706207275, "learning_rate": 3.5873894899213984e-05, "loss": 0.0077, "step": 48930 }, { "grad_norm": 0.07968056946992874, "learning_rate": 3.5854069747131416e-05, "loss": 0.0073, "step": 48940 }, { "grad_norm": 0.15272776782512665, "learning_rate": 3.583424701220143e-05, "loss": 0.0102, "step": 48950 }, { "grad_norm": 0.16774658858776093, "learning_rate": 3.581442669781121e-05, "loss": 0.0079, "step": 48960 }, { "grad_norm": 0.14030425250530243, "learning_rate": 3.579460880734749e-05, "loss": 0.0078, "step": 48970 }, { "grad_norm": 0.11983626335859299, "learning_rate": 3.577479334419657e-05, "loss": 0.0116, "step": 48980 }, { "grad_norm": 0.15019962191581726, "learning_rate": 3.575498031174444e-05, "loss": 0.0099, "step": 48990 }, { "grad_norm": 0.19588600099086761, "learning_rate": 3.573516971337657e-05, "loss": 0.0087, "step": 49000 }, { "grad_norm": 0.16637271642684937, "learning_rate": 3.5715361552478046e-05, "loss": 0.011, "step": 49010 }, { "grad_norm": 0.2153029590845108, "learning_rate": 3.5695555832433536e-05, "loss": 0.0098, "step": 49020 }, { "grad_norm": 0.12679052352905273, "learning_rate": 3.5675752556627325e-05, "loss": 0.0091, "step": 49030 }, { "grad_norm": 0.1504746377468109, "learning_rate": 3.565595172844322e-05, "loss": 0.0077, "step": 49040 }, { "grad_norm": 0.1645805537700653, "learning_rate": 3.5636153351264666e-05, "loss": 0.0095, "step": 49050 }, { "grad_norm": 0.22853390872478485, "learning_rate": 3.5616357428474655e-05, "loss": 0.0086, "step": 49060 }, { "grad_norm": 0.1456431895494461, "learning_rate": 3.559656396345575e-05, "loss": 0.0099, "step": 49070 }, { "grad_norm": 0.14040738344192505, "learning_rate": 3.5576772959590146e-05, "loss": 0.0093, "step": 49080 }, { "grad_norm": 0.16582949459552765, "learning_rate": 3.5556984420259545e-05, "loss": 0.0098, "step": 49090 }, { "grad_norm": 0.22947633266448975, "learning_rate": 3.5537198348845305e-05, "loss": 0.0102, "step": 49100 }, { "grad_norm": 0.0923970490694046, "learning_rate": 3.551741474872831e-05, "loss": 0.008, "step": 49110 }, { "grad_norm": 0.18024630844593048, "learning_rate": 3.549763362328903e-05, "loss": 0.0096, "step": 49120 }, { "grad_norm": 0.08632457256317139, "learning_rate": 3.5477854975907515e-05, "loss": 0.0076, "step": 49130 }, { "grad_norm": 0.10717609524726868, "learning_rate": 3.5458078809963416e-05, "loss": 0.009, "step": 49140 }, { "grad_norm": 0.16743099689483643, "learning_rate": 3.543830512883594e-05, "loss": 0.0099, "step": 49150 }, { "grad_norm": 0.1670006513595581, "learning_rate": 3.5418533935903824e-05, "loss": 0.0098, "step": 49160 }, { "grad_norm": 0.162772536277771, "learning_rate": 3.539876523454547e-05, "loss": 0.0082, "step": 49170 }, { "grad_norm": 0.1489986926317215, "learning_rate": 3.537899902813878e-05, "loss": 0.0078, "step": 49180 }, { "grad_norm": 0.15894249081611633, "learning_rate": 3.5359235320061293e-05, "loss": 0.0073, "step": 49190 }, { "grad_norm": 0.1798473596572876, "learning_rate": 3.533947411369003e-05, "loss": 0.0093, "step": 49200 }, { "grad_norm": 0.12797455489635468, "learning_rate": 3.53197154124017e-05, "loss": 0.0086, "step": 49210 }, { "grad_norm": 0.23332224786281586, "learning_rate": 3.52999592195725e-05, "loss": 0.0103, "step": 49220 }, { "grad_norm": 0.1900467574596405, "learning_rate": 3.5280205538578224e-05, "loss": 0.0086, "step": 49230 }, { "grad_norm": 0.1500455141067505, "learning_rate": 3.5260454372794236e-05, "loss": 0.012, "step": 49240 }, { "grad_norm": 0.0843009203672409, "learning_rate": 3.524070572559545e-05, "loss": 0.0101, "step": 49250 }, { "grad_norm": 0.14860934019088745, "learning_rate": 3.5220959600356395e-05, "loss": 0.0111, "step": 49260 }, { "grad_norm": 0.1898363083600998, "learning_rate": 3.5201216000451145e-05, "loss": 0.0101, "step": 49270 }, { "grad_norm": 0.1573033630847931, "learning_rate": 3.5181474929253335e-05, "loss": 0.0091, "step": 49280 }, { "grad_norm": 0.18582892417907715, "learning_rate": 3.516173639013615e-05, "loss": 0.0072, "step": 49290 }, { "grad_norm": 0.1215420812368393, "learning_rate": 3.5142000386472406e-05, "loss": 0.0088, "step": 49300 }, { "grad_norm": 0.12359955906867981, "learning_rate": 3.5122266921634427e-05, "loss": 0.0122, "step": 49310 }, { "grad_norm": 0.1706608384847641, "learning_rate": 3.510253599899413e-05, "loss": 0.0117, "step": 49320 }, { "grad_norm": 0.25648772716522217, "learning_rate": 3.5082807621922965e-05, "loss": 0.0112, "step": 49330 }, { "grad_norm": 0.13214011490345, "learning_rate": 3.506308179379201e-05, "loss": 0.0084, "step": 49340 }, { "grad_norm": 0.15281017124652863, "learning_rate": 3.5043358517971844e-05, "loss": 0.0104, "step": 49350 }, { "grad_norm": 0.15159165859222412, "learning_rate": 3.502363779783264e-05, "loss": 0.0073, "step": 49360 }, { "grad_norm": 0.17036472260951996, "learning_rate": 3.500391963674415e-05, "loss": 0.0092, "step": 49370 }, { "grad_norm": 0.15806438028812408, "learning_rate": 3.4984204038075615e-05, "loss": 0.0091, "step": 49380 }, { "grad_norm": 0.17583487927913666, "learning_rate": 3.496449100519595e-05, "loss": 0.0096, "step": 49390 }, { "grad_norm": 0.21366094052791595, "learning_rate": 3.494478054147354e-05, "loss": 0.0103, "step": 49400 }, { "grad_norm": 0.18711188435554504, "learning_rate": 3.4925072650276395e-05, "loss": 0.0105, "step": 49410 }, { "grad_norm": 0.2181912213563919, "learning_rate": 3.4905367334972016e-05, "loss": 0.0122, "step": 49420 }, { "grad_norm": 0.16277070343494415, "learning_rate": 3.488566459892752e-05, "loss": 0.0114, "step": 49430 }, { "grad_norm": 0.17128770053386688, "learning_rate": 3.4865964445509585e-05, "loss": 0.0105, "step": 49440 }, { "grad_norm": 0.23461218178272247, "learning_rate": 3.484626687808438e-05, "loss": 0.0078, "step": 49450 }, { "grad_norm": 0.2061574012041092, "learning_rate": 3.4826571900017735e-05, "loss": 0.01, "step": 49460 }, { "grad_norm": 0.15720391273498535, "learning_rate": 3.480687951467495e-05, "loss": 0.0085, "step": 49470 }, { "grad_norm": 0.11696213483810425, "learning_rate": 3.4787189725420925e-05, "loss": 0.0104, "step": 49480 }, { "grad_norm": 0.15950754284858704, "learning_rate": 3.4767502535620086e-05, "loss": 0.0092, "step": 49490 }, { "grad_norm": 0.13617713749408722, "learning_rate": 3.474781794863648e-05, "loss": 0.0085, "step": 49500 }, { "grad_norm": 0.11846500635147095, "learning_rate": 3.472813596783363e-05, "loss": 0.0082, "step": 49510 }, { "grad_norm": 0.08785895258188248, "learning_rate": 3.470845659657466e-05, "loss": 0.011, "step": 49520 }, { "grad_norm": 0.2017519623041153, "learning_rate": 3.468877983822223e-05, "loss": 0.0101, "step": 49530 }, { "grad_norm": 0.16705754399299622, "learning_rate": 3.466910569613855e-05, "loss": 0.0095, "step": 49540 }, { "grad_norm": 0.20800668001174927, "learning_rate": 3.464943417368542e-05, "loss": 0.0087, "step": 49550 }, { "grad_norm": 0.1019013524055481, "learning_rate": 3.462976527422415e-05, "loss": 0.0106, "step": 49560 }, { "grad_norm": 0.1495170295238495, "learning_rate": 3.461009900111562e-05, "loss": 0.008, "step": 49570 }, { "grad_norm": 0.13252002000808716, "learning_rate": 3.459043535772023e-05, "loss": 0.0104, "step": 49580 }, { "grad_norm": 0.0886739045381546, "learning_rate": 3.4570774347398014e-05, "loss": 0.0087, "step": 49590 }, { "grad_norm": 0.11995959281921387, "learning_rate": 3.4551115973508454e-05, "loss": 0.0109, "step": 49600 }, { "grad_norm": 0.1471114158630371, "learning_rate": 3.453146023941066e-05, "loss": 0.0087, "step": 49610 }, { "grad_norm": 0.11415193974971771, "learning_rate": 3.451180714846325e-05, "loss": 0.0112, "step": 49620 }, { "grad_norm": 0.13272418081760406, "learning_rate": 3.449215670402438e-05, "loss": 0.01, "step": 49630 }, { "grad_norm": 0.19778990745544434, "learning_rate": 3.447250890945181e-05, "loss": 0.0092, "step": 49640 }, { "grad_norm": 0.17190298438072205, "learning_rate": 3.4452863768102754e-05, "loss": 0.0092, "step": 49650 }, { "grad_norm": 0.14886309206485748, "learning_rate": 3.443322128333409e-05, "loss": 0.0085, "step": 49660 }, { "grad_norm": 0.18088732659816742, "learning_rate": 3.441358145850215e-05, "loss": 0.0107, "step": 49670 }, { "grad_norm": 0.17451222240924835, "learning_rate": 3.439394429696286e-05, "loss": 0.0113, "step": 49680 }, { "grad_norm": 0.14124490320682526, "learning_rate": 3.4374309802071644e-05, "loss": 0.0093, "step": 49690 }, { "grad_norm": 0.14243212342262268, "learning_rate": 3.435467797718353e-05, "loss": 0.0102, "step": 49700 }, { "grad_norm": 0.1588287353515625, "learning_rate": 3.433504882565306e-05, "loss": 0.014, "step": 49710 }, { "grad_norm": 0.1907816231250763, "learning_rate": 3.43154223508343e-05, "loss": 0.0101, "step": 49720 }, { "grad_norm": 0.20805968344211578, "learning_rate": 3.429579855608089e-05, "loss": 0.0117, "step": 49730 }, { "grad_norm": 0.291475385427475, "learning_rate": 3.427617744474597e-05, "loss": 0.0133, "step": 49740 }, { "grad_norm": 0.11971600353717804, "learning_rate": 3.425655902018231e-05, "loss": 0.0093, "step": 49750 }, { "grad_norm": 0.19777442514896393, "learning_rate": 3.423694328574211e-05, "loss": 0.0145, "step": 49760 }, { "grad_norm": 0.14640210568904877, "learning_rate": 3.42173302447772e-05, "loss": 0.0117, "step": 49770 }, { "grad_norm": 0.1840050220489502, "learning_rate": 3.419771990063886e-05, "loss": 0.0102, "step": 49780 }, { "grad_norm": 0.13861799240112305, "learning_rate": 3.417811225667803e-05, "loss": 0.0106, "step": 49790 }, { "grad_norm": 0.18575765192508698, "learning_rate": 3.415850731624508e-05, "loss": 0.0101, "step": 49800 }, { "grad_norm": 0.15073081851005554, "learning_rate": 3.4138905082689945e-05, "loss": 0.0143, "step": 49810 }, { "grad_norm": 0.16349093616008759, "learning_rate": 3.4119305559362145e-05, "loss": 0.0097, "step": 49820 }, { "grad_norm": 0.16204586625099182, "learning_rate": 3.4099708749610684e-05, "loss": 0.0105, "step": 49830 }, { "grad_norm": 0.12089245021343231, "learning_rate": 3.408011465678413e-05, "loss": 0.011, "step": 49840 }, { "grad_norm": 0.21868997812271118, "learning_rate": 3.406052328423055e-05, "loss": 0.0115, "step": 49850 }, { "grad_norm": 0.14233709871768951, "learning_rate": 3.4040934635297615e-05, "loss": 0.0093, "step": 49860 }, { "grad_norm": 0.22317874431610107, "learning_rate": 3.4021348713332466e-05, "loss": 0.0096, "step": 49870 }, { "grad_norm": 0.16964693367481232, "learning_rate": 3.4001765521681807e-05, "loss": 0.0111, "step": 49880 }, { "grad_norm": 0.14784419536590576, "learning_rate": 3.398218506369188e-05, "loss": 0.0088, "step": 49890 }, { "grad_norm": 0.10305622220039368, "learning_rate": 3.3962607342708404e-05, "loss": 0.0076, "step": 49900 }, { "grad_norm": 0.12033605575561523, "learning_rate": 3.394303236207673e-05, "loss": 0.011, "step": 49910 }, { "grad_norm": 0.10359948873519897, "learning_rate": 3.392346012514166e-05, "loss": 0.0102, "step": 49920 }, { "grad_norm": 0.2010994851589203, "learning_rate": 3.390389063524757e-05, "loss": 0.0093, "step": 49930 }, { "grad_norm": 0.16082622110843658, "learning_rate": 3.3884323895738324e-05, "loss": 0.0099, "step": 49940 }, { "grad_norm": 0.1985778510570526, "learning_rate": 3.386475990995738e-05, "loss": 0.0096, "step": 49950 }, { "grad_norm": 0.19461748003959656, "learning_rate": 3.384519868124765e-05, "loss": 0.0108, "step": 49960 }, { "grad_norm": 0.16365820169448853, "learning_rate": 3.3825640212951645e-05, "loss": 0.0085, "step": 49970 }, { "grad_norm": 0.1296503245830536, "learning_rate": 3.380608450841134e-05, "loss": 0.009, "step": 49980 }, { "grad_norm": 0.15846508741378784, "learning_rate": 3.3786531570968305e-05, "loss": 0.0135, "step": 49990 }, { "grad_norm": 0.14779137074947357, "learning_rate": 3.3766981403963584e-05, "loss": 0.0111, "step": 50000 }, { "grad_norm": 0.14353303611278534, "learning_rate": 3.374743401073775e-05, "loss": 0.0091, "step": 50010 }, { "grad_norm": 0.199258953332901, "learning_rate": 3.372788939463095e-05, "loss": 0.0075, "step": 50020 }, { "grad_norm": 0.13596634566783905, "learning_rate": 3.370834755898281e-05, "loss": 0.0092, "step": 50030 }, { "grad_norm": 0.14463135600090027, "learning_rate": 3.3688808507132493e-05, "loss": 0.0083, "step": 50040 }, { "grad_norm": 0.12495934963226318, "learning_rate": 3.3669272242418685e-05, "loss": 0.0094, "step": 50050 }, { "grad_norm": 0.13643933832645416, "learning_rate": 3.364973876817961e-05, "loss": 0.0103, "step": 50060 }, { "grad_norm": 0.13922768831253052, "learning_rate": 3.363020808775299e-05, "loss": 0.0095, "step": 50070 }, { "grad_norm": 0.1498144268989563, "learning_rate": 3.361068020447611e-05, "loss": 0.0088, "step": 50080 }, { "grad_norm": 0.1435793936252594, "learning_rate": 3.3591155121685724e-05, "loss": 0.0084, "step": 50090 }, { "grad_norm": 0.16702137887477875, "learning_rate": 3.357163284271814e-05, "loss": 0.0086, "step": 50100 }, { "grad_norm": 0.16912706196308136, "learning_rate": 3.355211337090919e-05, "loss": 0.0115, "step": 50110 }, { "grad_norm": 0.12968644499778748, "learning_rate": 3.353259670959421e-05, "loss": 0.0086, "step": 50120 }, { "grad_norm": 0.13210001587867737, "learning_rate": 3.351308286210808e-05, "loss": 0.0067, "step": 50130 }, { "grad_norm": 0.11092536151409149, "learning_rate": 3.3493571831785156e-05, "loss": 0.0087, "step": 50140 }, { "grad_norm": 0.6555966138839722, "learning_rate": 3.347406362195936e-05, "loss": 0.0125, "step": 50150 }, { "grad_norm": 0.20194056630134583, "learning_rate": 3.345455823596411e-05, "loss": 0.0089, "step": 50160 }, { "grad_norm": 0.11893169581890106, "learning_rate": 3.3435055677132346e-05, "loss": 0.008, "step": 50170 }, { "grad_norm": 0.1808474212884903, "learning_rate": 3.3415555948796505e-05, "loss": 0.0104, "step": 50180 }, { "grad_norm": 0.11058581620454788, "learning_rate": 3.3396059054288556e-05, "loss": 0.0092, "step": 50190 }, { "grad_norm": 0.1500617116689682, "learning_rate": 3.3376564996940015e-05, "loss": 0.0116, "step": 50200 }, { "grad_norm": 0.22651949524879456, "learning_rate": 3.3357073780081836e-05, "loss": 0.0092, "step": 50210 }, { "grad_norm": 0.15156632661819458, "learning_rate": 3.333758540704459e-05, "loss": 0.0085, "step": 50220 }, { "grad_norm": 0.12031612545251846, "learning_rate": 3.331809988115827e-05, "loss": 0.0088, "step": 50230 }, { "grad_norm": 0.10898476094007492, "learning_rate": 3.329861720575244e-05, "loss": 0.0075, "step": 50240 }, { "grad_norm": 0.13465234637260437, "learning_rate": 3.3279137384156126e-05, "loss": 0.0077, "step": 50250 }, { "grad_norm": 0.11477522552013397, "learning_rate": 3.3259660419697934e-05, "loss": 0.0091, "step": 50260 }, { "grad_norm": 0.12483382225036621, "learning_rate": 3.3240186315705926e-05, "loss": 0.0109, "step": 50270 }, { "grad_norm": 0.11417987197637558, "learning_rate": 3.322071507550769e-05, "loss": 0.006, "step": 50280 }, { "grad_norm": 0.18232840299606323, "learning_rate": 3.320124670243033e-05, "loss": 0.0113, "step": 50290 }, { "grad_norm": 0.14149273931980133, "learning_rate": 3.318178119980045e-05, "loss": 0.007, "step": 50300 }, { "grad_norm": 0.19704358279705048, "learning_rate": 3.316231857094421e-05, "loss": 0.009, "step": 50310 }, { "grad_norm": 0.19634386897087097, "learning_rate": 3.31428588191872e-05, "loss": 0.0111, "step": 50320 }, { "grad_norm": 0.14960476756095886, "learning_rate": 3.312340194785458e-05, "loss": 0.0086, "step": 50330 }, { "grad_norm": 0.15545445680618286, "learning_rate": 3.3103947960270975e-05, "loss": 0.0083, "step": 50340 }, { "grad_norm": 0.1631743460893631, "learning_rate": 3.308449685976058e-05, "loss": 0.0095, "step": 50350 }, { "grad_norm": 0.12714405357837677, "learning_rate": 3.3065048649647024e-05, "loss": 0.0103, "step": 50360 }, { "grad_norm": 0.13697193562984467, "learning_rate": 3.304560333325348e-05, "loss": 0.0097, "step": 50370 }, { "grad_norm": 0.43003353476524353, "learning_rate": 3.3026160913902624e-05, "loss": 0.0084, "step": 50380 }, { "grad_norm": 0.15611766278743744, "learning_rate": 3.300672139491662e-05, "loss": 0.0087, "step": 50390 }, { "grad_norm": 0.08516093343496323, "learning_rate": 3.298728477961717e-05, "loss": 0.0086, "step": 50400 }, { "grad_norm": 0.20558100938796997, "learning_rate": 3.296785107132544e-05, "loss": 0.0097, "step": 50410 }, { "grad_norm": 0.12933774292469025, "learning_rate": 3.2948420273362156e-05, "loss": 0.0106, "step": 50420 }, { "grad_norm": 0.20110170543193817, "learning_rate": 3.292899238904747e-05, "loss": 0.0094, "step": 50430 }, { "grad_norm": 0.1529223620891571, "learning_rate": 3.29095674217011e-05, "loss": 0.0087, "step": 50440 }, { "grad_norm": 0.13782964646816254, "learning_rate": 3.289014537464224e-05, "loss": 0.009, "step": 50450 }, { "grad_norm": 0.09828437119722366, "learning_rate": 3.287072625118955e-05, "loss": 0.0095, "step": 50460 }, { "grad_norm": 0.12732180953025818, "learning_rate": 3.285131005466129e-05, "loss": 0.0114, "step": 50470 }, { "grad_norm": 0.09689991921186447, "learning_rate": 3.2831896788375105e-05, "loss": 0.0074, "step": 50480 }, { "grad_norm": 0.1458553522825241, "learning_rate": 3.281248645564822e-05, "loss": 0.0103, "step": 50490 }, { "grad_norm": 0.15294910967350006, "learning_rate": 3.2793079059797306e-05, "loss": 0.0072, "step": 50500 }, { "grad_norm": 0.13001301884651184, "learning_rate": 3.277367460413859e-05, "loss": 0.0091, "step": 50510 }, { "grad_norm": 0.09970924258232117, "learning_rate": 3.275427309198773e-05, "loss": 0.0095, "step": 50520 }, { "grad_norm": 0.15765191614627838, "learning_rate": 3.273487452665993e-05, "loss": 0.0117, "step": 50530 }, { "grad_norm": 0.14622052013874054, "learning_rate": 3.271547891146986e-05, "loss": 0.0093, "step": 50540 }, { "grad_norm": 0.1492994725704193, "learning_rate": 3.269608624973173e-05, "loss": 0.0065, "step": 50550 }, { "grad_norm": 0.10496730357408524, "learning_rate": 3.26766965447592e-05, "loss": 0.0099, "step": 50560 }, { "grad_norm": 0.16385433077812195, "learning_rate": 3.2657309799865424e-05, "loss": 0.0091, "step": 50570 }, { "grad_norm": 0.15097306668758392, "learning_rate": 3.2637926018363084e-05, "loss": 0.0096, "step": 50580 }, { "grad_norm": 0.1558261513710022, "learning_rate": 3.261854520356432e-05, "loss": 0.0082, "step": 50590 }, { "grad_norm": 0.11519981175661087, "learning_rate": 3.2599167358780816e-05, "loss": 0.0097, "step": 50600 }, { "grad_norm": 0.15354333817958832, "learning_rate": 3.2579792487323676e-05, "loss": 0.0085, "step": 50610 }, { "grad_norm": 0.13546143472194672, "learning_rate": 3.256042059250358e-05, "loss": 0.0098, "step": 50620 }, { "grad_norm": 0.1567630022764206, "learning_rate": 3.254105167763062e-05, "loss": 0.011, "step": 50630 }, { "grad_norm": 0.14699296653270721, "learning_rate": 3.252168574601443e-05, "loss": 0.0121, "step": 50640 }, { "grad_norm": 0.12070540338754654, "learning_rate": 3.2502322800964115e-05, "loss": 0.0087, "step": 50650 }, { "grad_norm": 0.22278788685798645, "learning_rate": 3.248296284578825e-05, "loss": 0.0111, "step": 50660 }, { "grad_norm": 0.10147441923618317, "learning_rate": 3.246360588379497e-05, "loss": 0.0087, "step": 50670 }, { "grad_norm": 0.16478553414344788, "learning_rate": 3.24442519182918e-05, "loss": 0.0115, "step": 50680 }, { "grad_norm": 0.17651203274726868, "learning_rate": 3.2424900952585845e-05, "loss": 0.0092, "step": 50690 }, { "grad_norm": 0.0938774049282074, "learning_rate": 3.240555298998362e-05, "loss": 0.0107, "step": 50700 }, { "grad_norm": 0.11397454142570496, "learning_rate": 3.2386208033791194e-05, "loss": 0.0078, "step": 50710 }, { "grad_norm": 0.10497745126485825, "learning_rate": 3.2366866087314064e-05, "loss": 0.0073, "step": 50720 }, { "grad_norm": 0.12974333763122559, "learning_rate": 3.234752715385727e-05, "loss": 0.0079, "step": 50730 }, { "grad_norm": 0.21028824150562286, "learning_rate": 3.232819123672529e-05, "loss": 0.0134, "step": 50740 }, { "grad_norm": 0.15297366678714752, "learning_rate": 3.2308858339222076e-05, "loss": 0.0079, "step": 50750 }, { "grad_norm": 0.1311057209968567, "learning_rate": 3.2289528464651144e-05, "loss": 0.0071, "step": 50760 }, { "grad_norm": 0.21245430409908295, "learning_rate": 3.22702016163154e-05, "loss": 0.0092, "step": 50770 }, { "grad_norm": 0.11048123240470886, "learning_rate": 3.2250877797517295e-05, "loss": 0.0089, "step": 50780 }, { "grad_norm": 0.12973561882972717, "learning_rate": 3.223155701155872e-05, "loss": 0.0092, "step": 50790 }, { "grad_norm": 0.1757882982492447, "learning_rate": 3.2212239261741105e-05, "loss": 0.0084, "step": 50800 }, { "grad_norm": 0.14877469837665558, "learning_rate": 3.219292455136528e-05, "loss": 0.0107, "step": 50810 }, { "grad_norm": 0.12967883050441742, "learning_rate": 3.217361288373165e-05, "loss": 0.0099, "step": 50820 }, { "grad_norm": 0.10693301260471344, "learning_rate": 3.215430426214002e-05, "loss": 0.0109, "step": 50830 }, { "grad_norm": 0.10401840507984161, "learning_rate": 3.2134998689889705e-05, "loss": 0.0121, "step": 50840 }, { "grad_norm": 0.15392355620861053, "learning_rate": 3.211569617027952e-05, "loss": 0.0108, "step": 50850 }, { "grad_norm": 0.12281934916973114, "learning_rate": 3.209639670660769e-05, "loss": 0.0078, "step": 50860 }, { "grad_norm": 0.11601944267749786, "learning_rate": 3.207710030217202e-05, "loss": 0.0068, "step": 50870 }, { "grad_norm": 0.12351157516241074, "learning_rate": 3.205780696026972e-05, "loss": 0.0092, "step": 50880 }, { "grad_norm": 0.10835900902748108, "learning_rate": 3.203851668419749e-05, "loss": 0.0102, "step": 50890 }, { "grad_norm": 0.17355333268642426, "learning_rate": 3.201922947725149e-05, "loss": 0.0088, "step": 50900 }, { "grad_norm": 0.12795554101467133, "learning_rate": 3.199994534272742e-05, "loss": 0.0087, "step": 50910 }, { "grad_norm": 0.125840961933136, "learning_rate": 3.1980664283920394e-05, "loss": 0.0084, "step": 50920 }, { "grad_norm": 0.15397144854068756, "learning_rate": 3.196138630412499e-05, "loss": 0.0106, "step": 50930 }, { "grad_norm": 0.1738707721233368, "learning_rate": 3.194211140663532e-05, "loss": 0.0073, "step": 50940 }, { "grad_norm": 0.16144788265228271, "learning_rate": 3.1922839594744914e-05, "loss": 0.0099, "step": 50950 }, { "grad_norm": 0.16600018739700317, "learning_rate": 3.190357087174683e-05, "loss": 0.011, "step": 50960 }, { "grad_norm": 0.19355100393295288, "learning_rate": 3.188430524093353e-05, "loss": 0.0117, "step": 50970 }, { "grad_norm": 0.11661142110824585, "learning_rate": 3.1865042705597014e-05, "loss": 0.0091, "step": 50980 }, { "grad_norm": 0.12255197018384933, "learning_rate": 3.184578326902868e-05, "loss": 0.0104, "step": 50990 }, { "grad_norm": 0.10782759636640549, "learning_rate": 3.182652693451949e-05, "loss": 0.0101, "step": 51000 }, { "grad_norm": 0.12485391646623611, "learning_rate": 3.180727370535981e-05, "loss": 0.0089, "step": 51010 }, { "grad_norm": 0.22404885292053223, "learning_rate": 3.178802358483945e-05, "loss": 0.0121, "step": 51020 }, { "grad_norm": 0.14899297058582306, "learning_rate": 3.17687765762478e-05, "loss": 0.0115, "step": 51030 }, { "grad_norm": 0.1698337197303772, "learning_rate": 3.174953268287357e-05, "loss": 0.0102, "step": 51040 }, { "grad_norm": 0.10896759480237961, "learning_rate": 3.173029190800508e-05, "loss": 0.0114, "step": 51050 }, { "grad_norm": 0.1672913134098053, "learning_rate": 3.171105425493e-05, "loss": 0.0116, "step": 51060 }, { "grad_norm": 0.10502325743436813, "learning_rate": 3.169181972693557e-05, "loss": 0.0091, "step": 51070 }, { "grad_norm": 0.20200511813163757, "learning_rate": 3.1672588327308404e-05, "loss": 0.0095, "step": 51080 }, { "grad_norm": 0.19141773879528046, "learning_rate": 3.1653360059334646e-05, "loss": 0.0088, "step": 51090 }, { "grad_norm": 0.12907063961029053, "learning_rate": 3.163413492629985e-05, "loss": 0.0084, "step": 51100 }, { "grad_norm": 0.14561624825000763, "learning_rate": 3.161491293148912e-05, "loss": 0.0097, "step": 51110 }, { "grad_norm": 0.13553260266780853, "learning_rate": 3.1595694078186925e-05, "loss": 0.0102, "step": 51120 }, { "grad_norm": 0.17303143441677094, "learning_rate": 3.157647836967724e-05, "loss": 0.0075, "step": 51130 }, { "grad_norm": 0.17471939325332642, "learning_rate": 3.155726580924354e-05, "loss": 0.0095, "step": 51140 }, { "grad_norm": 0.09401269257068634, "learning_rate": 3.1538056400168676e-05, "loss": 0.0064, "step": 51150 }, { "grad_norm": 0.11676361411809921, "learning_rate": 3.151885014573506e-05, "loss": 0.0078, "step": 51160 }, { "grad_norm": 0.14999400079250336, "learning_rate": 3.149964704922448e-05, "loss": 0.0109, "step": 51170 }, { "grad_norm": 0.15303488075733185, "learning_rate": 3.148044711391825e-05, "loss": 0.0108, "step": 51180 }, { "grad_norm": 0.12814117968082428, "learning_rate": 3.146125034309707e-05, "loss": 0.0102, "step": 51190 }, { "grad_norm": 0.14392197132110596, "learning_rate": 3.1442056740041195e-05, "loss": 0.0093, "step": 51200 }, { "grad_norm": 0.09365329146385193, "learning_rate": 3.142286630803026e-05, "loss": 0.0089, "step": 51210 }, { "grad_norm": 0.16904471814632416, "learning_rate": 3.140367905034337e-05, "loss": 0.0122, "step": 51220 }, { "grad_norm": 0.21272261440753937, "learning_rate": 3.138449497025914e-05, "loss": 0.0095, "step": 51230 }, { "grad_norm": 0.22421963512897491, "learning_rate": 3.136531407105557e-05, "loss": 0.0079, "step": 51240 }, { "grad_norm": 0.12459389120340347, "learning_rate": 3.1346136356010184e-05, "loss": 0.0079, "step": 51250 }, { "grad_norm": 0.1720452606678009, "learning_rate": 3.132696182839988e-05, "loss": 0.012, "step": 51260 }, { "grad_norm": 0.1589793562889099, "learning_rate": 3.130779049150111e-05, "loss": 0.0079, "step": 51270 }, { "grad_norm": 0.1106981709599495, "learning_rate": 3.128862234858971e-05, "loss": 0.0072, "step": 51280 }, { "grad_norm": 0.09645865112543106, "learning_rate": 3.1269457402941005e-05, "loss": 0.0063, "step": 51290 }, { "grad_norm": 0.19001170992851257, "learning_rate": 3.125029565782974e-05, "loss": 0.0084, "step": 51300 }, { "grad_norm": 0.19925233721733093, "learning_rate": 3.123113711653013e-05, "loss": 0.0095, "step": 51310 }, { "grad_norm": 0.2630423903465271, "learning_rate": 3.121198178231587e-05, "loss": 0.0085, "step": 51320 }, { "grad_norm": 0.1590784788131714, "learning_rate": 3.119282965846006e-05, "loss": 0.008, "step": 51330 }, { "grad_norm": 0.11106228828430176, "learning_rate": 3.1173680748235303e-05, "loss": 0.0077, "step": 51340 }, { "grad_norm": 0.13086655735969543, "learning_rate": 3.115453505491358e-05, "loss": 0.0061, "step": 51350 }, { "grad_norm": 0.14754736423492432, "learning_rate": 3.1135392581766404e-05, "loss": 0.0095, "step": 51360 }, { "grad_norm": 0.14695873856544495, "learning_rate": 3.111625333206467e-05, "loss": 0.0095, "step": 51370 }, { "grad_norm": 0.14737091958522797, "learning_rate": 3.109711730907881e-05, "loss": 0.0081, "step": 51380 }, { "grad_norm": 0.12009315937757492, "learning_rate": 3.107798451607859e-05, "loss": 0.0084, "step": 51390 }, { "grad_norm": 0.17461158335208893, "learning_rate": 3.105885495633328e-05, "loss": 0.008, "step": 51400 }, { "grad_norm": 0.10367889702320099, "learning_rate": 3.103972863311163e-05, "loss": 0.0081, "step": 51410 }, { "grad_norm": 0.16353163123130798, "learning_rate": 3.102060554968178e-05, "loss": 0.011, "step": 51420 }, { "grad_norm": 0.16297651827335358, "learning_rate": 3.100148570931137e-05, "loss": 0.0076, "step": 51430 }, { "grad_norm": 0.12467494606971741, "learning_rate": 3.098236911526744e-05, "loss": 0.0079, "step": 51440 }, { "grad_norm": 0.1439349353313446, "learning_rate": 3.0963255770816495e-05, "loss": 0.0083, "step": 51450 }, { "grad_norm": 0.17059339582920074, "learning_rate": 3.094414567922447e-05, "loss": 0.0067, "step": 51460 }, { "grad_norm": 0.18195343017578125, "learning_rate": 3.092503884375678e-05, "loss": 0.0096, "step": 51470 }, { "grad_norm": 0.14971253275871277, "learning_rate": 3.0905935267678255e-05, "loss": 0.0102, "step": 51480 }, { "grad_norm": 0.12921194732189178, "learning_rate": 3.0886834954253166e-05, "loss": 0.0097, "step": 51490 }, { "grad_norm": 0.11709599196910858, "learning_rate": 3.086773790674524e-05, "loss": 0.0084, "step": 51500 }, { "grad_norm": 0.10501154512166977, "learning_rate": 3.0848644128417617e-05, "loss": 0.0083, "step": 51510 }, { "grad_norm": 0.15968647599220276, "learning_rate": 3.082955362253294e-05, "loss": 0.0093, "step": 51520 }, { "grad_norm": 0.14799527823925018, "learning_rate": 3.081046639235323e-05, "loss": 0.0081, "step": 51530 }, { "grad_norm": 0.19882217049598694, "learning_rate": 3.0791382441139986e-05, "loss": 0.009, "step": 51540 }, { "grad_norm": 0.17738264799118042, "learning_rate": 3.077230177215411e-05, "loss": 0.0078, "step": 51550 }, { "grad_norm": 0.12392445653676987, "learning_rate": 3.0753224388656e-05, "loss": 0.0084, "step": 51560 }, { "grad_norm": 0.1331956684589386, "learning_rate": 3.073415029390544e-05, "loss": 0.0087, "step": 51570 }, { "grad_norm": 0.19245725870132446, "learning_rate": 3.071507949116166e-05, "loss": 0.0114, "step": 51580 }, { "grad_norm": 0.11504878848791122, "learning_rate": 3.069601198368337e-05, "loss": 0.01, "step": 51590 }, { "grad_norm": 0.18118427693843842, "learning_rate": 3.067694777472864e-05, "loss": 0.0084, "step": 51600 }, { "grad_norm": 0.12133785337209702, "learning_rate": 3.065788686755508e-05, "loss": 0.0077, "step": 51610 }, { "grad_norm": 0.14760221540927887, "learning_rate": 3.063882926541961e-05, "loss": 0.0068, "step": 51620 }, { "grad_norm": 0.09936250001192093, "learning_rate": 3.061977497157872e-05, "loss": 0.0092, "step": 51630 }, { "grad_norm": 0.12481900304555893, "learning_rate": 3.0600723989288224e-05, "loss": 0.0102, "step": 51640 }, { "grad_norm": 0.13493110239505768, "learning_rate": 3.058167632180343e-05, "loss": 0.0088, "step": 51650 }, { "grad_norm": 0.08851908892393112, "learning_rate": 3.0562631972379074e-05, "loss": 0.0086, "step": 51660 }, { "grad_norm": 0.13774225115776062, "learning_rate": 3.0543590944269276e-05, "loss": 0.0113, "step": 51670 }, { "grad_norm": 0.15178214013576508, "learning_rate": 3.052455324072766e-05, "loss": 0.0101, "step": 51680 }, { "grad_norm": 0.14912143349647522, "learning_rate": 3.0505518865007244e-05, "loss": 0.0083, "step": 51690 }, { "grad_norm": 0.1914832442998886, "learning_rate": 3.048648782036048e-05, "loss": 0.0091, "step": 51700 }, { "grad_norm": 0.14598235487937927, "learning_rate": 3.0467460110039236e-05, "loss": 0.0066, "step": 51710 }, { "grad_norm": 0.11970360577106476, "learning_rate": 3.0448435737294866e-05, "loss": 0.0082, "step": 51720 }, { "grad_norm": 0.1546279340982437, "learning_rate": 3.042941470537808e-05, "loss": 0.0066, "step": 51730 }, { "grad_norm": 0.21523772180080414, "learning_rate": 3.0410397017539072e-05, "loss": 0.0131, "step": 51740 }, { "grad_norm": 0.1887500286102295, "learning_rate": 3.039138267702742e-05, "loss": 0.0078, "step": 51750 }, { "grad_norm": 0.1563006192445755, "learning_rate": 3.0372371687092193e-05, "loss": 0.0109, "step": 51760 }, { "grad_norm": 0.08525047451257706, "learning_rate": 3.0353364050981823e-05, "loss": 0.0099, "step": 51770 }, { "grad_norm": 0.17400917410850525, "learning_rate": 3.033435977194418e-05, "loss": 0.0105, "step": 51780 }, { "grad_norm": 0.13042905926704407, "learning_rate": 3.0315358853226618e-05, "loss": 0.0091, "step": 51790 }, { "grad_norm": 0.17952580749988556, "learning_rate": 3.0296361298075844e-05, "loss": 0.0096, "step": 51800 }, { "grad_norm": 0.11919897049665451, "learning_rate": 3.027736710973803e-05, "loss": 0.0075, "step": 51810 }, { "grad_norm": 0.1093360185623169, "learning_rate": 3.025837629145875e-05, "loss": 0.0092, "step": 51820 }, { "grad_norm": 0.15496757626533508, "learning_rate": 3.0239388846483048e-05, "loss": 0.008, "step": 51830 }, { "grad_norm": 0.14043690264225006, "learning_rate": 3.022040477805532e-05, "loss": 0.0111, "step": 51840 }, { "grad_norm": 0.12493961304426193, "learning_rate": 3.020142408941946e-05, "loss": 0.01, "step": 51850 }, { "grad_norm": 0.3536372184753418, "learning_rate": 3.018244678381873e-05, "loss": 0.0108, "step": 51860 }, { "grad_norm": 0.10657049715518951, "learning_rate": 3.0163472864495812e-05, "loss": 0.0113, "step": 51870 }, { "grad_norm": 0.12504702806472778, "learning_rate": 3.014450233469287e-05, "loss": 0.012, "step": 51880 }, { "grad_norm": 0.24023929238319397, "learning_rate": 3.0125535197651422e-05, "loss": 0.0115, "step": 51890 }, { "grad_norm": 0.12334780395030975, "learning_rate": 3.0106571456612448e-05, "loss": 0.0056, "step": 51900 }, { "grad_norm": 0.1617041975259781, "learning_rate": 3.0087611114816305e-05, "loss": 0.012, "step": 51910 }, { "grad_norm": 0.1323474496603012, "learning_rate": 3.006865417550284e-05, "loss": 0.0086, "step": 51920 }, { "grad_norm": 0.1854534149169922, "learning_rate": 3.0049700641911242e-05, "loss": 0.0089, "step": 51930 }, { "grad_norm": 0.11624165624380112, "learning_rate": 3.0030750517280175e-05, "loss": 0.0081, "step": 51940 }, { "grad_norm": 0.1574701964855194, "learning_rate": 3.0011803804847682e-05, "loss": 0.0094, "step": 51950 }, { "grad_norm": 0.14421987533569336, "learning_rate": 2.9992860507851224e-05, "loss": 0.0118, "step": 51960 }, { "grad_norm": 0.17329376935958862, "learning_rate": 2.9973920629527723e-05, "loss": 0.0115, "step": 51970 }, { "grad_norm": 0.10105249285697937, "learning_rate": 2.9954984173113453e-05, "loss": 0.009, "step": 51980 }, { "grad_norm": 0.11330248415470123, "learning_rate": 2.993605114184418e-05, "loss": 0.0097, "step": 51990 }, { "grad_norm": 0.17730554938316345, "learning_rate": 2.9917121538955005e-05, "loss": 0.0097, "step": 52000 }, { "grad_norm": 0.21701598167419434, "learning_rate": 2.98981953676805e-05, "loss": 0.0098, "step": 52010 }, { "grad_norm": 0.10539201647043228, "learning_rate": 2.9879272631254594e-05, "loss": 0.0109, "step": 52020 }, { "grad_norm": 0.1034068688750267, "learning_rate": 2.986035333291073e-05, "loss": 0.0087, "step": 52030 }, { "grad_norm": 0.13442805409431458, "learning_rate": 2.9841437475881646e-05, "loss": 0.0081, "step": 52040 }, { "grad_norm": 0.13290198147296906, "learning_rate": 2.9822525063399555e-05, "loss": 0.0099, "step": 52050 }, { "grad_norm": 0.09606993943452835, "learning_rate": 2.9803616098696087e-05, "loss": 0.0072, "step": 52060 }, { "grad_norm": 0.14235326647758484, "learning_rate": 2.9784710585002228e-05, "loss": 0.0098, "step": 52070 }, { "grad_norm": 0.16018609702587128, "learning_rate": 2.9765808525548467e-05, "loss": 0.0104, "step": 52080 }, { "grad_norm": 0.27581924200057983, "learning_rate": 2.974690992356461e-05, "loss": 0.0087, "step": 52090 }, { "grad_norm": 0.1326647847890854, "learning_rate": 2.9728014782279934e-05, "loss": 0.0081, "step": 52100 }, { "grad_norm": 0.12253156304359436, "learning_rate": 2.970912310492307e-05, "loss": 0.0094, "step": 52110 }, { "grad_norm": 0.1503557562828064, "learning_rate": 2.9690234894722134e-05, "loss": 0.0092, "step": 52120 }, { "grad_norm": 0.1526864916086197, "learning_rate": 2.9671350154904577e-05, "loss": 0.0089, "step": 52130 }, { "grad_norm": 0.12077183276414871, "learning_rate": 2.9652468888697282e-05, "loss": 0.0097, "step": 52140 }, { "grad_norm": 0.13970039784908295, "learning_rate": 2.9633591099326562e-05, "loss": 0.009, "step": 52150 }, { "grad_norm": 0.12995891273021698, "learning_rate": 2.9614716790018072e-05, "loss": 0.0094, "step": 52160 }, { "grad_norm": 0.11959940940141678, "learning_rate": 2.959584596399697e-05, "loss": 0.0067, "step": 52170 }, { "grad_norm": 0.16991037130355835, "learning_rate": 2.9576978624487717e-05, "loss": 0.0097, "step": 52180 }, { "grad_norm": 0.13036508858203888, "learning_rate": 2.9558114774714263e-05, "loss": 0.0088, "step": 52190 }, { "grad_norm": 0.11688625812530518, "learning_rate": 2.9539254417899897e-05, "loss": 0.0086, "step": 52200 }, { "grad_norm": 0.18075542151927948, "learning_rate": 2.9520397557267365e-05, "loss": 0.0081, "step": 52210 }, { "grad_norm": 0.16172805428504944, "learning_rate": 2.9501544196038765e-05, "loss": 0.0085, "step": 52220 }, { "grad_norm": 0.16606634855270386, "learning_rate": 2.9482694337435614e-05, "loss": 0.01, "step": 52230 }, { "grad_norm": 0.20160739123821259, "learning_rate": 2.946384798467887e-05, "loss": 0.0087, "step": 52240 }, { "grad_norm": 0.15317963063716888, "learning_rate": 2.9445005140988824e-05, "loss": 0.0078, "step": 52250 }, { "grad_norm": 0.15496551990509033, "learning_rate": 2.942616580958524e-05, "loss": 0.0082, "step": 52260 }, { "grad_norm": 0.14257696270942688, "learning_rate": 2.9407329993687193e-05, "loss": 0.0081, "step": 52270 }, { "grad_norm": 0.1635769009590149, "learning_rate": 2.938849769651326e-05, "loss": 0.0089, "step": 52280 }, { "grad_norm": 0.16325080394744873, "learning_rate": 2.9369668921281345e-05, "loss": 0.0058, "step": 52290 }, { "grad_norm": 0.11809957772493362, "learning_rate": 2.9350843671208773e-05, "loss": 0.0096, "step": 52300 }, { "grad_norm": 0.1301375776529312, "learning_rate": 2.9332021949512245e-05, "loss": 0.0071, "step": 52310 }, { "grad_norm": 0.08537758886814117, "learning_rate": 2.9313203759407908e-05, "loss": 0.0074, "step": 52320 }, { "grad_norm": 0.1853998303413391, "learning_rate": 2.929438910411127e-05, "loss": 0.0108, "step": 52330 }, { "grad_norm": 0.11081638932228088, "learning_rate": 2.927557798683722e-05, "loss": 0.0073, "step": 52340 }, { "grad_norm": 0.13334627449512482, "learning_rate": 2.9256770410800095e-05, "loss": 0.0082, "step": 52350 }, { "grad_norm": 0.1428966075181961, "learning_rate": 2.9237966379213554e-05, "loss": 0.0089, "step": 52360 }, { "grad_norm": 0.11618567258119583, "learning_rate": 2.9219165895290736e-05, "loss": 0.0089, "step": 52370 }, { "grad_norm": 0.16741526126861572, "learning_rate": 2.9200368962244083e-05, "loss": 0.0095, "step": 52380 }, { "grad_norm": 0.18223446607589722, "learning_rate": 2.9181575583285536e-05, "loss": 0.01, "step": 52390 }, { "grad_norm": 0.23156778514385223, "learning_rate": 2.916278576162632e-05, "loss": 0.0117, "step": 52400 }, { "grad_norm": 0.2669926881790161, "learning_rate": 2.9143999500477116e-05, "loss": 0.013, "step": 52410 }, { "grad_norm": 0.1327514499425888, "learning_rate": 2.9125216803048004e-05, "loss": 0.011, "step": 52420 }, { "grad_norm": 0.16012102365493774, "learning_rate": 2.9106437672548403e-05, "loss": 0.0104, "step": 52430 }, { "grad_norm": 0.12730185687541962, "learning_rate": 2.9087662112187154e-05, "loss": 0.0105, "step": 52440 }, { "grad_norm": 0.1737573742866516, "learning_rate": 2.9068890125172498e-05, "loss": 0.0089, "step": 52450 }, { "grad_norm": 0.09521849453449249, "learning_rate": 2.9050121714712058e-05, "loss": 0.0109, "step": 52460 }, { "grad_norm": 0.1026139184832573, "learning_rate": 2.9031356884012835e-05, "loss": 0.0085, "step": 52470 }, { "grad_norm": 0.15525546669960022, "learning_rate": 2.9012595636281247e-05, "loss": 0.0092, "step": 52480 }, { "grad_norm": 0.15150633454322815, "learning_rate": 2.8993837974723016e-05, "loss": 0.0078, "step": 52490 }, { "grad_norm": 0.19954028725624084, "learning_rate": 2.8975083902543394e-05, "loss": 0.009, "step": 52500 }, { "grad_norm": 0.18607980012893677, "learning_rate": 2.895633342294688e-05, "loss": 0.0134, "step": 52510 }, { "grad_norm": 0.14628881216049194, "learning_rate": 2.893758653913744e-05, "loss": 0.0099, "step": 52520 }, { "grad_norm": 0.1347910612821579, "learning_rate": 2.8918843254318423e-05, "loss": 0.0082, "step": 52530 }, { "grad_norm": 0.1477653682231903, "learning_rate": 2.8900103571692483e-05, "loss": 0.0106, "step": 52540 }, { "grad_norm": 0.23011760413646698, "learning_rate": 2.8881367494461796e-05, "loss": 0.0107, "step": 52550 }, { "grad_norm": 0.18639381229877472, "learning_rate": 2.886263502582779e-05, "loss": 0.0091, "step": 52560 }, { "grad_norm": 0.13721857964992523, "learning_rate": 2.8843906168991353e-05, "loss": 0.0107, "step": 52570 }, { "grad_norm": 0.1406027227640152, "learning_rate": 2.8825180927152728e-05, "loss": 0.0098, "step": 52580 }, { "grad_norm": 0.18585477769374847, "learning_rate": 2.8806459303511547e-05, "loss": 0.0101, "step": 52590 }, { "grad_norm": 0.10749708116054535, "learning_rate": 2.8787741301266852e-05, "loss": 0.0078, "step": 52600 }, { "grad_norm": 0.1496100276708603, "learning_rate": 2.876902692361697e-05, "loss": 0.0128, "step": 52610 }, { "grad_norm": 0.16734562814235687, "learning_rate": 2.875031617375976e-05, "loss": 0.0102, "step": 52620 }, { "grad_norm": 0.1343514770269394, "learning_rate": 2.8731609054892318e-05, "loss": 0.0108, "step": 52630 }, { "grad_norm": 0.12684917449951172, "learning_rate": 2.871290557021119e-05, "loss": 0.008, "step": 52640 }, { "grad_norm": 0.12423443794250488, "learning_rate": 2.8694205722912298e-05, "loss": 0.0115, "step": 52650 }, { "grad_norm": 0.19191338121891022, "learning_rate": 2.8675509516190936e-05, "loss": 0.0097, "step": 52660 }, { "grad_norm": 0.17306703329086304, "learning_rate": 2.8656816953241772e-05, "loss": 0.0121, "step": 52670 }, { "grad_norm": 0.1670544296503067, "learning_rate": 2.863812803725887e-05, "loss": 0.0123, "step": 52680 }, { "grad_norm": 0.15957236289978027, "learning_rate": 2.8619442771435623e-05, "loss": 0.0073, "step": 52690 }, { "grad_norm": 0.16815081238746643, "learning_rate": 2.860076115896484e-05, "loss": 0.0082, "step": 52700 }, { "grad_norm": 0.1288667470216751, "learning_rate": 2.8582083203038712e-05, "loss": 0.0091, "step": 52710 }, { "grad_norm": 0.12679365277290344, "learning_rate": 2.8563408906848778e-05, "loss": 0.0076, "step": 52720 }, { "grad_norm": 0.1476312279701233, "learning_rate": 2.8544738273585993e-05, "loss": 0.008, "step": 52730 }, { "grad_norm": 0.15475495159626007, "learning_rate": 2.8526071306440595e-05, "loss": 0.0083, "step": 52740 }, { "grad_norm": 0.10980208963155746, "learning_rate": 2.850740800860233e-05, "loss": 0.0075, "step": 52750 }, { "grad_norm": 0.1861594319343567, "learning_rate": 2.8488748383260178e-05, "loss": 0.0088, "step": 52760 }, { "grad_norm": 0.12241591513156891, "learning_rate": 2.8470092433602635e-05, "loss": 0.0097, "step": 52770 }, { "grad_norm": 0.15636177361011505, "learning_rate": 2.845144016281745e-05, "loss": 0.0105, "step": 52780 }, { "grad_norm": 0.12691070139408112, "learning_rate": 2.8432791574091744e-05, "loss": 0.0068, "step": 52790 }, { "grad_norm": 0.1390954852104187, "learning_rate": 2.8414146670612134e-05, "loss": 0.0139, "step": 52800 }, { "grad_norm": 0.11187423765659332, "learning_rate": 2.8395505455564446e-05, "loss": 0.0081, "step": 52810 }, { "grad_norm": 0.13089004158973694, "learning_rate": 2.837686793213403e-05, "loss": 0.0076, "step": 52820 }, { "grad_norm": 0.14737826585769653, "learning_rate": 2.8358234103505477e-05, "loss": 0.0079, "step": 52830 }, { "grad_norm": 0.18673090636730194, "learning_rate": 2.8339603972862806e-05, "loss": 0.0086, "step": 52840 }, { "grad_norm": 0.11473962664604187, "learning_rate": 2.83209775433894e-05, "loss": 0.0088, "step": 52850 }, { "grad_norm": 0.1223047524690628, "learning_rate": 2.8302354818268013e-05, "loss": 0.0096, "step": 52860 }, { "grad_norm": 0.13552029430866241, "learning_rate": 2.8283735800680754e-05, "loss": 0.0078, "step": 52870 }, { "grad_norm": 0.1772853285074234, "learning_rate": 2.826512049380913e-05, "loss": 0.0086, "step": 52880 }, { "grad_norm": 0.14234890043735504, "learning_rate": 2.8246508900833935e-05, "loss": 0.0076, "step": 52890 }, { "grad_norm": 0.13749948143959045, "learning_rate": 2.822790102493541e-05, "loss": 0.0067, "step": 52900 }, { "grad_norm": 0.1717512160539627, "learning_rate": 2.820929686929314e-05, "loss": 0.01, "step": 52910 }, { "grad_norm": 0.09986614435911179, "learning_rate": 2.819069643708605e-05, "loss": 0.0077, "step": 52920 }, { "grad_norm": 0.11069967597723007, "learning_rate": 2.8172099731492484e-05, "loss": 0.0084, "step": 52930 }, { "grad_norm": 0.2212199568748474, "learning_rate": 2.815350675569003e-05, "loss": 0.0109, "step": 52940 }, { "grad_norm": 0.17061452567577362, "learning_rate": 2.8134917512855807e-05, "loss": 0.0095, "step": 52950 }, { "grad_norm": 0.14961300790309906, "learning_rate": 2.8116332006166134e-05, "loss": 0.0078, "step": 52960 }, { "grad_norm": 0.12291331589221954, "learning_rate": 2.809775023879685e-05, "loss": 0.0094, "step": 52970 }, { "grad_norm": 0.15827317535877228, "learning_rate": 2.807917221392299e-05, "loss": 0.0093, "step": 52980 }, { "grad_norm": 0.12655596435070038, "learning_rate": 2.8060597934719067e-05, "loss": 0.0068, "step": 52990 }, { "grad_norm": 0.14506351947784424, "learning_rate": 2.8042027404358927e-05, "loss": 0.0109, "step": 53000 }, { "grad_norm": 0.16455507278442383, "learning_rate": 2.802346062601572e-05, "loss": 0.0081, "step": 53010 }, { "grad_norm": 0.13288263976573944, "learning_rate": 2.800489760286207e-05, "loss": 0.0107, "step": 53020 }, { "grad_norm": 0.18658994138240814, "learning_rate": 2.7986338338069834e-05, "loss": 0.0073, "step": 53030 }, { "grad_norm": 0.14827609062194824, "learning_rate": 2.7967782834810297e-05, "loss": 0.009, "step": 53040 }, { "grad_norm": 0.16236521303653717, "learning_rate": 2.794923109625409e-05, "loss": 0.0105, "step": 53050 }, { "grad_norm": 0.12533625960350037, "learning_rate": 2.7930683125571205e-05, "loss": 0.0076, "step": 53060 }, { "grad_norm": 0.1696304827928543, "learning_rate": 2.7912138925930997e-05, "loss": 0.0059, "step": 53070 }, { "grad_norm": 0.10304633527994156, "learning_rate": 2.7893598500502117e-05, "loss": 0.0083, "step": 53080 }, { "grad_norm": 0.1483040601015091, "learning_rate": 2.7875061852452644e-05, "loss": 0.0106, "step": 53090 }, { "grad_norm": 0.366931676864624, "learning_rate": 2.7856528984949982e-05, "loss": 0.0107, "step": 53100 }, { "grad_norm": 0.18337345123291016, "learning_rate": 2.7837999901160888e-05, "loss": 0.0093, "step": 53110 }, { "grad_norm": 0.12178833037614822, "learning_rate": 2.7819474604251484e-05, "loss": 0.0097, "step": 53120 }, { "grad_norm": 0.18796633183956146, "learning_rate": 2.780095309738725e-05, "loss": 0.0105, "step": 53130 }, { "grad_norm": 0.167694091796875, "learning_rate": 2.778243538373294e-05, "loss": 0.0111, "step": 53140 }, { "grad_norm": 0.21616296470165253, "learning_rate": 2.7763921466452826e-05, "loss": 0.0152, "step": 53150 }, { "grad_norm": 0.13116195797920227, "learning_rate": 2.7745411348710336e-05, "loss": 0.01, "step": 53160 }, { "grad_norm": 0.0918051153421402, "learning_rate": 2.7726905033668395e-05, "loss": 0.0087, "step": 53170 }, { "grad_norm": 0.21082624793052673, "learning_rate": 2.7708402524489214e-05, "loss": 0.0093, "step": 53180 }, { "grad_norm": 0.15994399785995483, "learning_rate": 2.7689903824334364e-05, "loss": 0.0112, "step": 53190 }, { "grad_norm": 0.15913261473178864, "learning_rate": 2.7671408936364785e-05, "loss": 0.0103, "step": 53200 }, { "grad_norm": 0.1966407746076584, "learning_rate": 2.765291786374069e-05, "loss": 0.01, "step": 53210 }, { "grad_norm": 0.14629140496253967, "learning_rate": 2.7634430609621775e-05, "loss": 0.0084, "step": 53220 }, { "grad_norm": 0.11625228077173233, "learning_rate": 2.7615947177166956e-05, "loss": 0.0075, "step": 53230 }, { "grad_norm": 0.17556753754615784, "learning_rate": 2.7597467569534553e-05, "loss": 0.0094, "step": 53240 }, { "grad_norm": 0.11882472038269043, "learning_rate": 2.757899178988226e-05, "loss": 0.0068, "step": 53250 }, { "grad_norm": 0.13058947026729584, "learning_rate": 2.7560519841367005e-05, "loss": 0.0106, "step": 53260 }, { "grad_norm": 0.1859717071056366, "learning_rate": 2.7542051727145237e-05, "loss": 0.0098, "step": 53270 }, { "grad_norm": 0.15635287761688232, "learning_rate": 2.7523587450372578e-05, "loss": 0.0091, "step": 53280 }, { "grad_norm": 0.15445232391357422, "learning_rate": 2.750512701420409e-05, "loss": 0.0086, "step": 53290 }, { "grad_norm": 0.10385224968194962, "learning_rate": 2.7486670421794158e-05, "loss": 0.0086, "step": 53300 }, { "grad_norm": 0.190942645072937, "learning_rate": 2.7468217676296515e-05, "loss": 0.0118, "step": 53310 }, { "grad_norm": 0.16813667118549347, "learning_rate": 2.7449768780864226e-05, "loss": 0.0084, "step": 53320 }, { "grad_norm": 0.13891638815402985, "learning_rate": 2.7431323738649724e-05, "loss": 0.0077, "step": 53330 }, { "grad_norm": 0.10687632113695145, "learning_rate": 2.7412882552804713e-05, "loss": 0.008, "step": 53340 }, { "grad_norm": 0.14358210563659668, "learning_rate": 2.739444522648032e-05, "loss": 0.0103, "step": 53350 }, { "grad_norm": 0.08487504720687866, "learning_rate": 2.737601176282697e-05, "loss": 0.0075, "step": 53360 }, { "grad_norm": 0.11446764320135117, "learning_rate": 2.735758216499445e-05, "loss": 0.0072, "step": 53370 }, { "grad_norm": 0.09306786954402924, "learning_rate": 2.7339156436131864e-05, "loss": 0.0086, "step": 53380 }, { "grad_norm": 0.14706751704216003, "learning_rate": 2.7320734579387663e-05, "loss": 0.0084, "step": 53390 }, { "grad_norm": 0.07964206486940384, "learning_rate": 2.730231659790966e-05, "loss": 0.0085, "step": 53400 }, { "grad_norm": 0.12218552082777023, "learning_rate": 2.728390249484494e-05, "loss": 0.0075, "step": 53410 }, { "grad_norm": 0.08658856898546219, "learning_rate": 2.7265492273340032e-05, "loss": 0.0081, "step": 53420 }, { "grad_norm": 0.12136873602867126, "learning_rate": 2.7247085936540678e-05, "loss": 0.0071, "step": 53430 }, { "grad_norm": 0.1457306444644928, "learning_rate": 2.722868348759205e-05, "loss": 0.0081, "step": 53440 }, { "grad_norm": 0.16894036531448364, "learning_rate": 2.7210284929638635e-05, "loss": 0.0098, "step": 53450 }, { "grad_norm": 0.19437198340892792, "learning_rate": 2.7191890265824183e-05, "loss": 0.0114, "step": 53460 }, { "grad_norm": 0.10212140530347824, "learning_rate": 2.7173499499291926e-05, "loss": 0.0075, "step": 53470 }, { "grad_norm": 0.10043533146381378, "learning_rate": 2.7155112633184277e-05, "loss": 0.0092, "step": 53480 }, { "grad_norm": 0.14565150439739227, "learning_rate": 2.713672967064307e-05, "loss": 0.0097, "step": 53490 }, { "grad_norm": 0.09112875908613205, "learning_rate": 2.711835061480945e-05, "loss": 0.0074, "step": 53500 }, { "grad_norm": 0.12295489758253098, "learning_rate": 2.7099975468823896e-05, "loss": 0.0107, "step": 53510 }, { "grad_norm": 0.15490099787712097, "learning_rate": 2.708160423582622e-05, "loss": 0.0107, "step": 53520 }, { "grad_norm": 0.12915126979351044, "learning_rate": 2.706323691895557e-05, "loss": 0.011, "step": 53530 }, { "grad_norm": 0.12431284785270691, "learning_rate": 2.70448735213504e-05, "loss": 0.0069, "step": 53540 }, { "grad_norm": 0.12032841145992279, "learning_rate": 2.702651404614852e-05, "loss": 0.0108, "step": 53550 }, { "grad_norm": 0.11945623904466629, "learning_rate": 2.7008158496487056e-05, "loss": 0.0078, "step": 53560 }, { "grad_norm": 0.1346902996301651, "learning_rate": 2.6989806875502487e-05, "loss": 0.0132, "step": 53570 }, { "grad_norm": 0.11533203721046448, "learning_rate": 2.6971459186330584e-05, "loss": 0.0103, "step": 53580 }, { "grad_norm": 0.12668220698833466, "learning_rate": 2.695311543210648e-05, "loss": 0.0096, "step": 53590 }, { "grad_norm": 0.15006014704704285, "learning_rate": 2.693477561596463e-05, "loss": 0.0104, "step": 53600 }, { "grad_norm": 0.16959069669246674, "learning_rate": 2.6916439741038756e-05, "loss": 0.0089, "step": 53610 }, { "grad_norm": 0.13315463066101074, "learning_rate": 2.689810781046203e-05, "loss": 0.0101, "step": 53620 }, { "grad_norm": 0.12390974909067154, "learning_rate": 2.6879779827366823e-05, "loss": 0.0077, "step": 53630 }, { "grad_norm": 0.10389737784862518, "learning_rate": 2.6861455794884904e-05, "loss": 0.0067, "step": 53640 }, { "grad_norm": 0.18833936750888824, "learning_rate": 2.6843135716147373e-05, "loss": 0.009, "step": 53650 }, { "grad_norm": 0.08992522209882736, "learning_rate": 2.6824819594284556e-05, "loss": 0.0071, "step": 53660 }, { "grad_norm": 0.08057937026023865, "learning_rate": 2.6806507432426275e-05, "loss": 0.0082, "step": 53670 }, { "grad_norm": 0.12583863735198975, "learning_rate": 2.6788199233701512e-05, "loss": 0.0093, "step": 53680 }, { "grad_norm": 0.18457189202308655, "learning_rate": 2.6769895001238652e-05, "loss": 0.0079, "step": 53690 }, { "grad_norm": 0.11687811464071274, "learning_rate": 2.67515947381654e-05, "loss": 0.0067, "step": 53700 }, { "grad_norm": 0.2707419693470001, "learning_rate": 2.6733298447608745e-05, "loss": 0.0092, "step": 53710 }, { "grad_norm": 0.10957754403352737, "learning_rate": 2.6715006132695074e-05, "loss": 0.0079, "step": 53720 }, { "grad_norm": 0.13381463289260864, "learning_rate": 2.6696717796549985e-05, "loss": 0.0082, "step": 53730 }, { "grad_norm": 0.10556566715240479, "learning_rate": 2.667843344229848e-05, "loss": 0.0074, "step": 53740 }, { "grad_norm": 0.10574512183666229, "learning_rate": 2.6660153073064846e-05, "loss": 0.0075, "step": 53750 }, { "grad_norm": 0.1807015985250473, "learning_rate": 2.664187669197271e-05, "loss": 0.0088, "step": 53760 }, { "grad_norm": 0.10422425717115402, "learning_rate": 2.6623604302145005e-05, "loss": 0.0089, "step": 53770 }, { "grad_norm": 0.11698248237371445, "learning_rate": 2.6605335906703975e-05, "loss": 0.0066, "step": 53780 }, { "grad_norm": 0.16033321619033813, "learning_rate": 2.65870715087712e-05, "loss": 0.0061, "step": 53790 }, { "grad_norm": 0.14072197675704956, "learning_rate": 2.6568811111467573e-05, "loss": 0.0083, "step": 53800 }, { "grad_norm": 0.07991160452365875, "learning_rate": 2.6550554717913258e-05, "loss": 0.0097, "step": 53810 }, { "grad_norm": 0.1132279708981514, "learning_rate": 2.6532302331227805e-05, "loss": 0.006, "step": 53820 }, { "grad_norm": 0.13436082005500793, "learning_rate": 2.651405395453004e-05, "loss": 0.008, "step": 53830 }, { "grad_norm": 0.11510460823774338, "learning_rate": 2.6495809590938115e-05, "loss": 0.0068, "step": 53840 }, { "grad_norm": 0.13511478900909424, "learning_rate": 2.64775692435695e-05, "loss": 0.007, "step": 53850 }, { "grad_norm": 0.17330659925937653, "learning_rate": 2.6459332915540928e-05, "loss": 0.0085, "step": 53860 }, { "grad_norm": 0.09895966202020645, "learning_rate": 2.644110060996856e-05, "loss": 0.0071, "step": 53870 }, { "grad_norm": 0.17191845178604126, "learning_rate": 2.642287232996774e-05, "loss": 0.0081, "step": 53880 }, { "grad_norm": 0.10843543708324432, "learning_rate": 2.6404648078653205e-05, "loss": 0.0093, "step": 53890 }, { "grad_norm": 0.43328267335891724, "learning_rate": 2.6386427859139002e-05, "loss": 0.0061, "step": 53900 }, { "grad_norm": 0.09077738970518112, "learning_rate": 2.63682116745384e-05, "loss": 0.0072, "step": 53910 }, { "grad_norm": 0.23431135714054108, "learning_rate": 2.6349999527964138e-05, "loss": 0.0095, "step": 53920 }, { "grad_norm": 0.1182260513305664, "learning_rate": 2.63317914225281e-05, "loss": 0.006, "step": 53930 }, { "grad_norm": 0.18219712376594543, "learning_rate": 2.6313587361341585e-05, "loss": 0.0067, "step": 53940 }, { "grad_norm": 0.12712177634239197, "learning_rate": 2.6295387347515165e-05, "loss": 0.0074, "step": 53950 }, { "grad_norm": 0.2132532000541687, "learning_rate": 2.6277191384158727e-05, "loss": 0.0095, "step": 53960 }, { "grad_norm": 0.15919148921966553, "learning_rate": 2.6258999474381457e-05, "loss": 0.0076, "step": 53970 }, { "grad_norm": 0.17439371347427368, "learning_rate": 2.624081162129186e-05, "loss": 0.0105, "step": 53980 }, { "grad_norm": 0.18353192508220673, "learning_rate": 2.6222627827997765e-05, "loss": 0.0099, "step": 53990 }, { "grad_norm": 0.19849014282226562, "learning_rate": 2.6204448097606236e-05, "loss": 0.009, "step": 54000 }, { "grad_norm": 0.20394383370876312, "learning_rate": 2.6186272433223726e-05, "loss": 0.0085, "step": 54010 }, { "grad_norm": 0.11370202898979187, "learning_rate": 2.6168100837955943e-05, "loss": 0.0092, "step": 54020 }, { "grad_norm": 0.145579993724823, "learning_rate": 2.6149933314907926e-05, "loss": 0.0094, "step": 54030 }, { "grad_norm": 0.1455608457326889, "learning_rate": 2.6131769867184e-05, "loss": 0.0105, "step": 54040 }, { "grad_norm": 0.12636761367321014, "learning_rate": 2.611361049788783e-05, "loss": 0.0102, "step": 54050 }, { "grad_norm": 0.1503313034772873, "learning_rate": 2.6095455210122292e-05, "loss": 0.0106, "step": 54060 }, { "grad_norm": 0.15843528509140015, "learning_rate": 2.6077304006989712e-05, "loss": 0.0089, "step": 54070 }, { "grad_norm": 0.1411719173192978, "learning_rate": 2.6059156891591562e-05, "loss": 0.0111, "step": 54080 }, { "grad_norm": 0.09529969096183777, "learning_rate": 2.6041013867028718e-05, "loss": 0.008, "step": 54090 }, { "grad_norm": 0.15902279317378998, "learning_rate": 2.6022874936401347e-05, "loss": 0.0085, "step": 54100 }, { "grad_norm": 0.21252432465553284, "learning_rate": 2.6004740102808832e-05, "loss": 0.0124, "step": 54110 }, { "grad_norm": 0.14946042001247406, "learning_rate": 2.598660936935e-05, "loss": 0.0086, "step": 54120 }, { "grad_norm": 0.11217524856328964, "learning_rate": 2.5968482739122845e-05, "loss": 0.0107, "step": 54130 }, { "grad_norm": 0.17581234872341156, "learning_rate": 2.595036021522472e-05, "loss": 0.0113, "step": 54140 }, { "grad_norm": 0.10417606681585312, "learning_rate": 2.5932241800752278e-05, "loss": 0.009, "step": 54150 }, { "grad_norm": 0.13575704395771027, "learning_rate": 2.5914127498801453e-05, "loss": 0.0091, "step": 54160 }, { "grad_norm": 0.15380287170410156, "learning_rate": 2.5896017312467497e-05, "loss": 0.0078, "step": 54170 }, { "grad_norm": 0.16639508306980133, "learning_rate": 2.587791124484493e-05, "loss": 0.0078, "step": 54180 }, { "grad_norm": 0.09210605174303055, "learning_rate": 2.5859809299027615e-05, "loss": 0.0083, "step": 54190 }, { "grad_norm": 0.1183788999915123, "learning_rate": 2.5841711478108632e-05, "loss": 0.0068, "step": 54200 }, { "grad_norm": 0.1345994621515274, "learning_rate": 2.582361778518043e-05, "loss": 0.0076, "step": 54210 }, { "grad_norm": 0.12218373268842697, "learning_rate": 2.580552822333472e-05, "loss": 0.0096, "step": 54220 }, { "grad_norm": 0.1387796849012375, "learning_rate": 2.578744279566252e-05, "loss": 0.0091, "step": 54230 }, { "grad_norm": 0.06662791967391968, "learning_rate": 2.576936150525413e-05, "loss": 0.0069, "step": 54240 }, { "grad_norm": 0.2521359622478485, "learning_rate": 2.5751284355199168e-05, "loss": 0.0083, "step": 54250 }, { "grad_norm": 0.13724055886268616, "learning_rate": 2.573321134858646e-05, "loss": 0.0076, "step": 54260 }, { "grad_norm": 0.09333471953868866, "learning_rate": 2.5715142488504286e-05, "loss": 0.0082, "step": 54270 }, { "grad_norm": 0.13618597388267517, "learning_rate": 2.5697077778040042e-05, "loss": 0.0081, "step": 54280 }, { "grad_norm": 0.19106023013591766, "learning_rate": 2.5679017220280522e-05, "loss": 0.0095, "step": 54290 }, { "grad_norm": 0.10721602290868759, "learning_rate": 2.5660960818311796e-05, "loss": 0.0078, "step": 54300 }, { "grad_norm": 0.1345309019088745, "learning_rate": 2.564290857521915e-05, "loss": 0.0089, "step": 54310 }, { "grad_norm": 0.17511464655399323, "learning_rate": 2.5624860494087298e-05, "loss": 0.0074, "step": 54320 }, { "grad_norm": 0.11439161002635956, "learning_rate": 2.5606816578000115e-05, "loss": 0.0087, "step": 54330 }, { "grad_norm": 0.11204027384519577, "learning_rate": 2.558877683004082e-05, "loss": 0.0089, "step": 54340 }, { "grad_norm": 0.11765147745609283, "learning_rate": 2.557074125329192e-05, "loss": 0.008, "step": 54350 }, { "grad_norm": 0.20256924629211426, "learning_rate": 2.5552709850835195e-05, "loss": 0.0091, "step": 54360 }, { "grad_norm": 0.11006181687116623, "learning_rate": 2.5534682625751738e-05, "loss": 0.0088, "step": 54370 }, { "grad_norm": 0.19646939635276794, "learning_rate": 2.551665958112186e-05, "loss": 0.0089, "step": 54380 }, { "grad_norm": 0.13174232840538025, "learning_rate": 2.549864072002527e-05, "loss": 0.0091, "step": 54390 }, { "grad_norm": 0.16359329223632812, "learning_rate": 2.5480626045540858e-05, "loss": 0.009, "step": 54400 }, { "grad_norm": 0.150444895029068, "learning_rate": 2.546261556074684e-05, "loss": 0.0079, "step": 54410 }, { "grad_norm": 0.14231976866722107, "learning_rate": 2.5444609268720726e-05, "loss": 0.0099, "step": 54420 }, { "grad_norm": 0.15378721058368683, "learning_rate": 2.5426607172539297e-05, "loss": 0.0077, "step": 54430 }, { "grad_norm": 0.18385566771030426, "learning_rate": 2.5408609275278617e-05, "loss": 0.0097, "step": 54440 }, { "grad_norm": 0.1446652114391327, "learning_rate": 2.5390615580014055e-05, "loss": 0.0083, "step": 54450 }, { "grad_norm": 0.14631906151771545, "learning_rate": 2.5372626089820207e-05, "loss": 0.0092, "step": 54460 }, { "grad_norm": 0.13445700705051422, "learning_rate": 2.5354640807770997e-05, "loss": 0.0095, "step": 54470 }, { "grad_norm": 0.13863889873027802, "learning_rate": 2.5336659736939622e-05, "loss": 0.0083, "step": 54480 }, { "grad_norm": 0.19066694378852844, "learning_rate": 2.5318682880398554e-05, "loss": 0.0082, "step": 54490 }, { "grad_norm": 0.151421919465065, "learning_rate": 2.530071024121956e-05, "loss": 0.0092, "step": 54500 }, { "grad_norm": 0.2369295060634613, "learning_rate": 2.5282741822473627e-05, "loss": 0.0083, "step": 54510 }, { "grad_norm": 0.10691206157207489, "learning_rate": 2.526477762723114e-05, "loss": 0.0086, "step": 54520 }, { "grad_norm": 0.169490784406662, "learning_rate": 2.5246817658561618e-05, "loss": 0.0088, "step": 54530 }, { "grad_norm": 0.10014723241329193, "learning_rate": 2.5228861919533965e-05, "loss": 0.0067, "step": 54540 }, { "grad_norm": 0.16906197369098663, "learning_rate": 2.5210910413216326e-05, "loss": 0.0111, "step": 54550 }, { "grad_norm": 0.10842262953519821, "learning_rate": 2.5192963142676086e-05, "loss": 0.0074, "step": 54560 }, { "grad_norm": 0.13253721594810486, "learning_rate": 2.517502011098001e-05, "loss": 0.0091, "step": 54570 }, { "grad_norm": 0.1358378380537033, "learning_rate": 2.5157081321193987e-05, "loss": 0.0076, "step": 54580 }, { "grad_norm": 0.15188969671726227, "learning_rate": 2.5139146776383356e-05, "loss": 0.0108, "step": 54590 }, { "grad_norm": 0.12802818417549133, "learning_rate": 2.5121216479612575e-05, "loss": 0.0061, "step": 54600 }, { "grad_norm": 0.16750502586364746, "learning_rate": 2.510329043394546e-05, "loss": 0.0116, "step": 54610 }, { "grad_norm": 0.11811714619398117, "learning_rate": 2.508536864244508e-05, "loss": 0.007, "step": 54620 }, { "grad_norm": 0.14989405870437622, "learning_rate": 2.5067451108173778e-05, "loss": 0.0078, "step": 54630 }, { "grad_norm": 0.14664065837860107, "learning_rate": 2.5049537834193204e-05, "loss": 0.0081, "step": 54640 }, { "grad_norm": 0.1661720871925354, "learning_rate": 2.5031628823564194e-05, "loss": 0.0089, "step": 54650 }, { "grad_norm": 0.12404104322195053, "learning_rate": 2.5013724079346933e-05, "loss": 0.0103, "step": 54660 }, { "grad_norm": 0.2107698917388916, "learning_rate": 2.4995823604600854e-05, "loss": 0.0074, "step": 54670 }, { "grad_norm": 0.09619729965925217, "learning_rate": 2.497792740238465e-05, "loss": 0.0082, "step": 54680 }, { "grad_norm": 0.12600168585777283, "learning_rate": 2.49600354757563e-05, "loss": 0.0068, "step": 54690 }, { "grad_norm": 0.12485017627477646, "learning_rate": 2.494214782777306e-05, "loss": 0.009, "step": 54700 }, { "grad_norm": 0.09650926291942596, "learning_rate": 2.4924264461491386e-05, "loss": 0.0068, "step": 54710 }, { "grad_norm": 0.12382872402667999, "learning_rate": 2.4906385379967133e-05, "loss": 0.0089, "step": 54720 }, { "grad_norm": 0.1405172497034073, "learning_rate": 2.4888510586255285e-05, "loss": 0.0073, "step": 54730 }, { "grad_norm": 0.12728047370910645, "learning_rate": 2.487064008341018e-05, "loss": 0.0085, "step": 54740 }, { "grad_norm": 0.15431062877178192, "learning_rate": 2.4852773874485407e-05, "loss": 0.0088, "step": 54750 }, { "grad_norm": 0.15109366178512573, "learning_rate": 2.483491196253377e-05, "loss": 0.0076, "step": 54760 }, { "grad_norm": 0.12492655217647552, "learning_rate": 2.4817054350607443e-05, "loss": 0.0079, "step": 54770 }, { "grad_norm": 0.13730590045452118, "learning_rate": 2.4799201041757742e-05, "loss": 0.0072, "step": 54780 }, { "grad_norm": 0.10449308902025223, "learning_rate": 2.4781352039035373e-05, "loss": 0.008, "step": 54790 }, { "grad_norm": 0.1423531472682953, "learning_rate": 2.4763507345490194e-05, "loss": 0.0104, "step": 54800 }, { "grad_norm": 0.18562622368335724, "learning_rate": 2.4745666964171386e-05, "loss": 0.0086, "step": 54810 }, { "grad_norm": 0.13130667805671692, "learning_rate": 2.4727830898127387e-05, "loss": 0.0075, "step": 54820 }, { "grad_norm": 0.14131830632686615, "learning_rate": 2.4709999150405895e-05, "loss": 0.0075, "step": 54830 }, { "grad_norm": 0.15825097262859344, "learning_rate": 2.469217172405388e-05, "loss": 0.0061, "step": 54840 }, { "grad_norm": 0.11372584849596024, "learning_rate": 2.4674348622117527e-05, "loss": 0.0079, "step": 54850 }, { "grad_norm": 0.15501096844673157, "learning_rate": 2.465652984764234e-05, "loss": 0.0103, "step": 54860 }, { "grad_norm": 0.15516628324985504, "learning_rate": 2.4638715403673056e-05, "loss": 0.0078, "step": 54870 }, { "grad_norm": 0.21680255234241486, "learning_rate": 2.4620905293253676e-05, "loss": 0.0083, "step": 54880 }, { "grad_norm": 0.09602051973342896, "learning_rate": 2.4603099519427463e-05, "loss": 0.0096, "step": 54890 }, { "grad_norm": 0.2526922821998596, "learning_rate": 2.458529808523695e-05, "loss": 0.0085, "step": 54900 }, { "grad_norm": 0.11366382986307144, "learning_rate": 2.456750099372387e-05, "loss": 0.0077, "step": 54910 }, { "grad_norm": 0.1455395668745041, "learning_rate": 2.4549708247929327e-05, "loss": 0.01, "step": 54920 }, { "grad_norm": 0.11091073602437973, "learning_rate": 2.4531919850893554e-05, "loss": 0.0091, "step": 54930 }, { "grad_norm": 0.0740213394165039, "learning_rate": 2.4514135805656125e-05, "loss": 0.0067, "step": 54940 }, { "grad_norm": 0.11334960907697678, "learning_rate": 2.449635611525587e-05, "loss": 0.0075, "step": 54950 }, { "grad_norm": 0.2731685936450958, "learning_rate": 2.447858078273079e-05, "loss": 0.0078, "step": 54960 }, { "grad_norm": 0.0930841714143753, "learning_rate": 2.4460809811118275e-05, "loss": 0.0081, "step": 54970 }, { "grad_norm": 0.09454677253961563, "learning_rate": 2.444304320345483e-05, "loss": 0.0092, "step": 54980 }, { "grad_norm": 0.08323196321725845, "learning_rate": 2.4425280962776354e-05, "loss": 0.0072, "step": 54990 }, { "grad_norm": 0.07433663308620453, "learning_rate": 2.4407523092117875e-05, "loss": 0.0084, "step": 55000 }, { "grad_norm": 0.09254482388496399, "learning_rate": 2.438976959451374e-05, "loss": 0.0092, "step": 55010 }, { "grad_norm": 0.12036140263080597, "learning_rate": 2.4372020472997565e-05, "loss": 0.0085, "step": 55020 }, { "grad_norm": 0.17027047276496887, "learning_rate": 2.4354275730602122e-05, "loss": 0.0085, "step": 55030 }, { "grad_norm": 0.22471539676189423, "learning_rate": 2.433653537035958e-05, "loss": 0.0088, "step": 55040 }, { "grad_norm": 0.13477753102779388, "learning_rate": 2.431879939530123e-05, "loss": 0.0068, "step": 55050 }, { "grad_norm": 0.1605420857667923, "learning_rate": 2.4301067808457684e-05, "loss": 0.0064, "step": 55060 }, { "grad_norm": 0.10107483714818954, "learning_rate": 2.428334061285878e-05, "loss": 0.0078, "step": 55070 }, { "grad_norm": 0.085471972823143, "learning_rate": 2.426561781153361e-05, "loss": 0.0071, "step": 55080 }, { "grad_norm": 0.11141866445541382, "learning_rate": 2.424789940751052e-05, "loss": 0.0058, "step": 55090 }, { "grad_norm": 0.15480799973011017, "learning_rate": 2.423018540381712e-05, "loss": 0.0068, "step": 55100 }, { "grad_norm": 0.1385871320962906, "learning_rate": 2.421247580348021e-05, "loss": 0.0064, "step": 55110 }, { "grad_norm": 0.22577889263629913, "learning_rate": 2.4194770609525897e-05, "loss": 0.0087, "step": 55120 }, { "grad_norm": 0.20264078676700592, "learning_rate": 2.417706982497951e-05, "loss": 0.0072, "step": 55130 }, { "grad_norm": 0.12242049723863602, "learning_rate": 2.415937345286563e-05, "loss": 0.007, "step": 55140 }, { "grad_norm": 0.1297103762626648, "learning_rate": 2.4141681496208087e-05, "loss": 0.0071, "step": 55150 }, { "grad_norm": 0.11346897482872009, "learning_rate": 2.4123993958029946e-05, "loss": 0.0068, "step": 55160 }, { "grad_norm": 0.1243123933672905, "learning_rate": 2.4106310841353548e-05, "loss": 0.0075, "step": 55170 }, { "grad_norm": 0.10803265124559402, "learning_rate": 2.4088632149200398e-05, "loss": 0.0082, "step": 55180 }, { "grad_norm": 0.11193531006574631, "learning_rate": 2.4070957884591367e-05, "loss": 0.0075, "step": 55190 }, { "grad_norm": 0.15317471325397491, "learning_rate": 2.4053288050546464e-05, "loss": 0.0093, "step": 55200 }, { "grad_norm": 0.1695728600025177, "learning_rate": 2.403562265008498e-05, "loss": 0.0077, "step": 55210 }, { "grad_norm": 0.17140668630599976, "learning_rate": 2.4017961686225483e-05, "loss": 0.0085, "step": 55220 }, { "grad_norm": 0.13212642073631287, "learning_rate": 2.400030516198568e-05, "loss": 0.01, "step": 55230 }, { "grad_norm": 0.17789649963378906, "learning_rate": 2.3982653080382673e-05, "loss": 0.0092, "step": 55240 }, { "grad_norm": 0.106368787586689, "learning_rate": 2.396500544443266e-05, "loss": 0.0076, "step": 55250 }, { "grad_norm": 0.22428776323795319, "learning_rate": 2.3947362257151156e-05, "loss": 0.0085, "step": 55260 }, { "grad_norm": 0.13620345294475555, "learning_rate": 2.39297235215529e-05, "loss": 0.0097, "step": 55270 }, { "grad_norm": 0.12306786328554153, "learning_rate": 2.3912089240651873e-05, "loss": 0.0086, "step": 55280 }, { "grad_norm": 0.10750513523817062, "learning_rate": 2.389445941746129e-05, "loss": 0.0074, "step": 55290 }, { "grad_norm": 0.10765483230352402, "learning_rate": 2.3876834054993625e-05, "loss": 0.0081, "step": 55300 }, { "grad_norm": 0.17618313431739807, "learning_rate": 2.3859213156260522e-05, "loss": 0.0102, "step": 55310 }, { "grad_norm": 0.10864990204572678, "learning_rate": 2.3841596724272948e-05, "loss": 0.0096, "step": 55320 }, { "grad_norm": 0.11775844544172287, "learning_rate": 2.382398476204106e-05, "loss": 0.0057, "step": 55330 }, { "grad_norm": 0.16755962371826172, "learning_rate": 2.3806377272574254e-05, "loss": 0.0071, "step": 55340 }, { "grad_norm": 0.3076106309890747, "learning_rate": 2.3788774258881174e-05, "loss": 0.0086, "step": 55350 }, { "grad_norm": 0.13935546576976776, "learning_rate": 2.37711757239697e-05, "loss": 0.011, "step": 55360 }, { "grad_norm": 0.1267903745174408, "learning_rate": 2.3753581670846954e-05, "loss": 0.0105, "step": 55370 }, { "grad_norm": 0.1091521680355072, "learning_rate": 2.3735992102519216e-05, "loss": 0.0084, "step": 55380 }, { "grad_norm": 0.1298101544380188, "learning_rate": 2.371840702199215e-05, "loss": 0.0087, "step": 55390 }, { "grad_norm": 0.08235429227352142, "learning_rate": 2.3700826432270494e-05, "loss": 0.0069, "step": 55400 }, { "grad_norm": 0.1547851264476776, "learning_rate": 2.3683250336358326e-05, "loss": 0.0064, "step": 55410 }, { "grad_norm": 0.08709214627742767, "learning_rate": 2.3665678737258923e-05, "loss": 0.008, "step": 55420 }, { "grad_norm": 0.14346109330654144, "learning_rate": 2.3648111637974745e-05, "loss": 0.009, "step": 55430 }, { "grad_norm": 0.16916565597057343, "learning_rate": 2.36305490415076e-05, "loss": 0.0102, "step": 55440 }, { "grad_norm": 0.12273170799016953, "learning_rate": 2.3612990950858392e-05, "loss": 0.0078, "step": 55450 }, { "grad_norm": 0.0921773836016655, "learning_rate": 2.359543736902735e-05, "loss": 0.0101, "step": 55460 }, { "grad_norm": 0.09852108359336853, "learning_rate": 2.3577888299013896e-05, "loss": 0.0077, "step": 55470 }, { "grad_norm": 0.17087538540363312, "learning_rate": 2.3560343743816683e-05, "loss": 0.0082, "step": 55480 }, { "grad_norm": 0.1811332404613495, "learning_rate": 2.354280370643362e-05, "loss": 0.0079, "step": 55490 }, { "grad_norm": 0.11977436393499374, "learning_rate": 2.3525268189861777e-05, "loss": 0.0095, "step": 55500 }, { "grad_norm": 0.15423054993152618, "learning_rate": 2.3507737197097513e-05, "loss": 0.0101, "step": 55510 }, { "grad_norm": 0.1816304475069046, "learning_rate": 2.34902107311364e-05, "loss": 0.0115, "step": 55520 }, { "grad_norm": 0.1280542016029358, "learning_rate": 2.3472688794973225e-05, "loss": 0.0082, "step": 55530 }, { "grad_norm": 0.12770232558250427, "learning_rate": 2.3455171391602016e-05, "loss": 0.0082, "step": 55540 }, { "grad_norm": 0.15655620396137238, "learning_rate": 2.343765852401601e-05, "loss": 0.0096, "step": 55550 }, { "grad_norm": 0.1221468523144722, "learning_rate": 2.342015019520768e-05, "loss": 0.0078, "step": 55560 }, { "grad_norm": 0.1626727432012558, "learning_rate": 2.3402646408168742e-05, "loss": 0.0111, "step": 55570 }, { "grad_norm": 0.10644719004631042, "learning_rate": 2.3385147165890074e-05, "loss": 0.0082, "step": 55580 }, { "grad_norm": 0.25374239683151245, "learning_rate": 2.336765247136184e-05, "loss": 0.0093, "step": 55590 }, { "grad_norm": 0.14682061970233917, "learning_rate": 2.33501623275734e-05, "loss": 0.008, "step": 55600 }, { "grad_norm": 0.10803636163473129, "learning_rate": 2.333267673751334e-05, "loss": 0.0073, "step": 55610 }, { "grad_norm": 0.12224270403385162, "learning_rate": 2.33151957041695e-05, "loss": 0.0071, "step": 55620 }, { "grad_norm": 0.1818639636039734, "learning_rate": 2.329771923052884e-05, "loss": 0.0078, "step": 55630 }, { "grad_norm": 0.17331594228744507, "learning_rate": 2.3280247319577697e-05, "loss": 0.0125, "step": 55640 }, { "grad_norm": 0.16255009174346924, "learning_rate": 2.3262779974301473e-05, "loss": 0.0101, "step": 55650 }, { "grad_norm": 0.13703615963459015, "learning_rate": 2.3245317197684895e-05, "loss": 0.0094, "step": 55660 }, { "grad_norm": 0.22335080802440643, "learning_rate": 2.322785899271188e-05, "loss": 0.0074, "step": 55670 }, { "grad_norm": 0.11019491404294968, "learning_rate": 2.3210405362365507e-05, "loss": 0.0099, "step": 55680 }, { "grad_norm": 0.14217360317707062, "learning_rate": 2.31929563096282e-05, "loss": 0.0064, "step": 55690 }, { "grad_norm": 0.11400473862886429, "learning_rate": 2.317551183748146e-05, "loss": 0.0078, "step": 55700 }, { "grad_norm": 0.06470572203397751, "learning_rate": 2.3158071948906103e-05, "loss": 0.009, "step": 55710 }, { "grad_norm": 0.1443793922662735, "learning_rate": 2.314063664688212e-05, "loss": 0.0095, "step": 55720 }, { "grad_norm": 0.11103716492652893, "learning_rate": 2.3123205934388725e-05, "loss": 0.0081, "step": 55730 }, { "grad_norm": 0.11610709130764008, "learning_rate": 2.310577981440436e-05, "loss": 0.0085, "step": 55740 }, { "grad_norm": 0.0997595563530922, "learning_rate": 2.3088358289906665e-05, "loss": 0.0086, "step": 55750 }, { "grad_norm": 0.13017266988754272, "learning_rate": 2.307094136387252e-05, "loss": 0.0094, "step": 55760 }, { "grad_norm": 0.08308067172765732, "learning_rate": 2.305352903927796e-05, "loss": 0.0067, "step": 55770 }, { "grad_norm": 0.13992774486541748, "learning_rate": 2.303612131909831e-05, "loss": 0.0116, "step": 55780 }, { "grad_norm": 0.09115521609783173, "learning_rate": 2.3018718206308054e-05, "loss": 0.0086, "step": 55790 }, { "grad_norm": 0.18034672737121582, "learning_rate": 2.3001319703880925e-05, "loss": 0.009, "step": 55800 }, { "grad_norm": 0.15143024921417236, "learning_rate": 2.2983925814789835e-05, "loss": 0.0092, "step": 55810 }, { "grad_norm": 0.09068125486373901, "learning_rate": 2.2966536542006957e-05, "loss": 0.0113, "step": 55820 }, { "grad_norm": 0.15388059616088867, "learning_rate": 2.294915188850358e-05, "loss": 0.0103, "step": 55830 }, { "grad_norm": 0.1671861857175827, "learning_rate": 2.2931771857250333e-05, "loss": 0.0116, "step": 55840 }, { "grad_norm": 0.13042622804641724, "learning_rate": 2.2914396451216946e-05, "loss": 0.0076, "step": 55850 }, { "grad_norm": 0.11888247728347778, "learning_rate": 2.2897025673372412e-05, "loss": 0.007, "step": 55860 }, { "grad_norm": 0.141778826713562, "learning_rate": 2.287965952668494e-05, "loss": 0.0092, "step": 55870 }, { "grad_norm": 0.14136971533298492, "learning_rate": 2.2862298014121873e-05, "loss": 0.0082, "step": 55880 }, { "grad_norm": 0.11497362703084946, "learning_rate": 2.28449411386499e-05, "loss": 0.0069, "step": 55890 }, { "grad_norm": 0.13318322598934174, "learning_rate": 2.282758890323477e-05, "loss": 0.0076, "step": 55900 }, { "grad_norm": 0.15704022347927094, "learning_rate": 2.2810241310841528e-05, "loss": 0.011, "step": 55910 }, { "grad_norm": 0.13916149735450745, "learning_rate": 2.2792898364434412e-05, "loss": 0.0081, "step": 55920 }, { "grad_norm": 0.11578933149576187, "learning_rate": 2.2775560066976843e-05, "loss": 0.0068, "step": 55930 }, { "grad_norm": 0.1506570279598236, "learning_rate": 2.275822642143147e-05, "loss": 0.0067, "step": 55940 }, { "grad_norm": 0.1779950112104416, "learning_rate": 2.274089743076014e-05, "loss": 0.0082, "step": 55950 }, { "grad_norm": 0.0763968974351883, "learning_rate": 2.2723573097923917e-05, "loss": 0.0055, "step": 55960 }, { "grad_norm": 0.11267293989658356, "learning_rate": 2.2706253425883024e-05, "loss": 0.0082, "step": 55970 }, { "grad_norm": 0.10082443803548813, "learning_rate": 2.2688938417596933e-05, "loss": 0.0101, "step": 55980 }, { "grad_norm": 0.08925297856330872, "learning_rate": 2.2671628076024305e-05, "loss": 0.0105, "step": 55990 }, { "grad_norm": 0.1453109234571457, "learning_rate": 2.2654322404123008e-05, "loss": 0.009, "step": 56000 }, { "grad_norm": 0.15721486508846283, "learning_rate": 2.2637021404850105e-05, "loss": 0.0075, "step": 56010 }, { "grad_norm": 0.15637654066085815, "learning_rate": 2.2619725081161885e-05, "loss": 0.0064, "step": 56020 }, { "grad_norm": 0.13225030899047852, "learning_rate": 2.260243343601376e-05, "loss": 0.0073, "step": 56030 }, { "grad_norm": 0.08872693032026291, "learning_rate": 2.2585146472360473e-05, "loss": 0.0071, "step": 56040 }, { "grad_norm": 0.11723263561725616, "learning_rate": 2.2567864193155834e-05, "loss": 0.0061, "step": 56050 }, { "grad_norm": 0.09848960489034653, "learning_rate": 2.255058660135294e-05, "loss": 0.006, "step": 56060 }, { "grad_norm": 0.11021184176206589, "learning_rate": 2.253331369990407e-05, "loss": 0.008, "step": 56070 }, { "grad_norm": 0.09827437996864319, "learning_rate": 2.251604549176063e-05, "loss": 0.0098, "step": 56080 }, { "grad_norm": 0.11706741899251938, "learning_rate": 2.249878197987337e-05, "loss": 0.0069, "step": 56090 }, { "grad_norm": 0.08421961963176727, "learning_rate": 2.2481523167192087e-05, "loss": 0.0086, "step": 56100 }, { "grad_norm": 0.13462066650390625, "learning_rate": 2.2464269056665867e-05, "loss": 0.0116, "step": 56110 }, { "grad_norm": 0.15356209874153137, "learning_rate": 2.2447019651242958e-05, "loss": 0.0069, "step": 56120 }, { "grad_norm": 0.10657504945993423, "learning_rate": 2.2429774953870818e-05, "loss": 0.0064, "step": 56130 }, { "grad_norm": 0.12101120501756668, "learning_rate": 2.241253496749611e-05, "loss": 0.0073, "step": 56140 }, { "grad_norm": 0.15654541552066803, "learning_rate": 2.2395299695064614e-05, "loss": 0.0098, "step": 56150 }, { "grad_norm": 0.1836496889591217, "learning_rate": 2.237806913952145e-05, "loss": 0.0077, "step": 56160 }, { "grad_norm": 0.10866865515708923, "learning_rate": 2.2360843303810798e-05, "loss": 0.0072, "step": 56170 }, { "grad_norm": 0.19955679774284363, "learning_rate": 2.2343622190876084e-05, "loss": 0.0103, "step": 56180 }, { "grad_norm": 0.17257699370384216, "learning_rate": 2.2326405803659935e-05, "loss": 0.009, "step": 56190 }, { "grad_norm": 0.08865620940923691, "learning_rate": 2.230919414510416e-05, "loss": 0.0079, "step": 56200 }, { "grad_norm": 0.13765549659729004, "learning_rate": 2.229198721814976e-05, "loss": 0.011, "step": 56210 }, { "grad_norm": 0.1510147899389267, "learning_rate": 2.2274785025736948e-05, "loss": 0.0117, "step": 56220 }, { "grad_norm": 0.1503959596157074, "learning_rate": 2.225758757080507e-05, "loss": 0.0082, "step": 56230 }, { "grad_norm": 0.16044214367866516, "learning_rate": 2.2240394856292723e-05, "loss": 0.0094, "step": 56240 }, { "grad_norm": 0.11444883048534393, "learning_rate": 2.2223206885137664e-05, "loss": 0.011, "step": 56250 }, { "grad_norm": 0.09789030998945236, "learning_rate": 2.2206023660276853e-05, "loss": 0.0078, "step": 56260 }, { "grad_norm": 0.2057076394557953, "learning_rate": 2.218884518464645e-05, "loss": 0.0086, "step": 56270 }, { "grad_norm": 0.12784551084041595, "learning_rate": 2.2171671461181732e-05, "loss": 0.0079, "step": 56280 }, { "grad_norm": 0.18757347762584686, "learning_rate": 2.2154502492817292e-05, "loss": 0.01, "step": 56290 }, { "grad_norm": 0.11519037932157516, "learning_rate": 2.2137338282486782e-05, "loss": 0.0101, "step": 56300 }, { "grad_norm": 0.12307172268629074, "learning_rate": 2.2120178833123113e-05, "loss": 0.0088, "step": 56310 }, { "grad_norm": 0.06885325163602829, "learning_rate": 2.210302414765838e-05, "loss": 0.0077, "step": 56320 }, { "grad_norm": 0.165543332695961, "learning_rate": 2.2085874229023808e-05, "loss": 0.0078, "step": 56330 }, { "grad_norm": 0.09941581636667252, "learning_rate": 2.2068729080149907e-05, "loss": 0.0082, "step": 56340 }, { "grad_norm": 0.10414629429578781, "learning_rate": 2.205158870396625e-05, "loss": 0.01, "step": 56350 }, { "grad_norm": 0.12159319967031479, "learning_rate": 2.2034453103401732e-05, "loss": 0.0081, "step": 56360 }, { "grad_norm": 0.10720949620008469, "learning_rate": 2.20173222813843e-05, "loss": 0.0066, "step": 56370 }, { "grad_norm": 0.17097850143909454, "learning_rate": 2.200019624084116e-05, "loss": 0.0078, "step": 56380 }, { "grad_norm": 0.12346325069665909, "learning_rate": 2.1983074984698687e-05, "loss": 0.0059, "step": 56390 }, { "grad_norm": 0.13473284244537354, "learning_rate": 2.1965958515882433e-05, "loss": 0.0067, "step": 56400 }, { "grad_norm": 0.10976729542016983, "learning_rate": 2.1948846837317162e-05, "loss": 0.0076, "step": 56410 }, { "grad_norm": 0.13503196835517883, "learning_rate": 2.1931739951926738e-05, "loss": 0.0079, "step": 56420 }, { "grad_norm": 0.1298687756061554, "learning_rate": 2.191463786263429e-05, "loss": 0.0085, "step": 56430 }, { "grad_norm": 0.12647481262683868, "learning_rate": 2.1897540572362095e-05, "loss": 0.0071, "step": 56440 }, { "grad_norm": 0.1502555012702942, "learning_rate": 2.1880448084031614e-05, "loss": 0.007, "step": 56450 }, { "grad_norm": 0.1330055147409439, "learning_rate": 2.1863360400563482e-05, "loss": 0.0065, "step": 56460 }, { "grad_norm": 0.16612808406352997, "learning_rate": 2.184627752487754e-05, "loss": 0.0106, "step": 56470 }, { "grad_norm": 0.10420649498701096, "learning_rate": 2.1829199459892725e-05, "loss": 0.0081, "step": 56480 }, { "grad_norm": 0.10558091849088669, "learning_rate": 2.1812126208527282e-05, "loss": 0.0108, "step": 56490 }, { "grad_norm": 0.12054342776536942, "learning_rate": 2.179505777369852e-05, "loss": 0.0102, "step": 56500 }, { "grad_norm": 0.1234046071767807, "learning_rate": 2.1777994158322974e-05, "loss": 0.0082, "step": 56510 }, { "grad_norm": 0.15914373099803925, "learning_rate": 2.176093536531637e-05, "loss": 0.009, "step": 56520 }, { "grad_norm": 0.181375652551651, "learning_rate": 2.1743881397593537e-05, "loss": 0.0106, "step": 56530 }, { "grad_norm": 0.10125993937253952, "learning_rate": 2.1726832258068595e-05, "loss": 0.0086, "step": 56540 }, { "grad_norm": 0.09442010521888733, "learning_rate": 2.170978794965472e-05, "loss": 0.0076, "step": 56550 }, { "grad_norm": 0.12603655457496643, "learning_rate": 2.169274847526438e-05, "loss": 0.0083, "step": 56560 }, { "grad_norm": 0.17327068746089935, "learning_rate": 2.1675713837809103e-05, "loss": 0.0103, "step": 56570 }, { "grad_norm": 0.0975097194314003, "learning_rate": 2.1658684040199655e-05, "loss": 0.0067, "step": 56580 }, { "grad_norm": 0.12694190442562103, "learning_rate": 2.1641659085345974e-05, "loss": 0.0068, "step": 56590 }, { "grad_norm": 0.0965246856212616, "learning_rate": 2.1624638976157154e-05, "loss": 0.0073, "step": 56600 }, { "grad_norm": 0.14546391367912292, "learning_rate": 2.1607623715541476e-05, "loss": 0.0084, "step": 56610 }, { "grad_norm": 0.09888768196105957, "learning_rate": 2.159061330640636e-05, "loss": 0.009, "step": 56620 }, { "grad_norm": 0.16240140795707703, "learning_rate": 2.1573607751658425e-05, "loss": 0.0086, "step": 56630 }, { "grad_norm": 0.14572802186012268, "learning_rate": 2.155660705420347e-05, "loss": 0.0093, "step": 56640 }, { "grad_norm": 0.170190691947937, "learning_rate": 2.153961121694644e-05, "loss": 0.0074, "step": 56650 }, { "grad_norm": 0.11734595149755478, "learning_rate": 2.152262024279145e-05, "loss": 0.0079, "step": 56660 }, { "grad_norm": 0.12364987283945084, "learning_rate": 2.150563413464183e-05, "loss": 0.0094, "step": 56670 }, { "grad_norm": 0.13789401948451996, "learning_rate": 2.1488652895399974e-05, "loss": 0.0075, "step": 56680 }, { "grad_norm": 0.12756845355033875, "learning_rate": 2.1471676527967587e-05, "loss": 0.0076, "step": 56690 }, { "grad_norm": 0.13964132964611053, "learning_rate": 2.1454705035245414e-05, "loss": 0.0077, "step": 56700 }, { "grad_norm": 0.12913069128990173, "learning_rate": 2.143773842013343e-05, "loss": 0.0074, "step": 56710 }, { "grad_norm": 0.1718546748161316, "learning_rate": 2.1420776685530796e-05, "loss": 0.0081, "step": 56720 }, { "grad_norm": 0.1314714103937149, "learning_rate": 2.1403819834335742e-05, "loss": 0.0066, "step": 56730 }, { "grad_norm": 0.09952346235513687, "learning_rate": 2.1386867869445808e-05, "loss": 0.0061, "step": 56740 }, { "grad_norm": 0.14809945225715637, "learning_rate": 2.1369920793757548e-05, "loss": 0.0067, "step": 56750 }, { "grad_norm": 0.1500210165977478, "learning_rate": 2.1352978610166828e-05, "loss": 0.0093, "step": 56760 }, { "grad_norm": 0.15005382895469666, "learning_rate": 2.1336041321568546e-05, "loss": 0.0077, "step": 56770 }, { "grad_norm": 0.1224994957447052, "learning_rate": 2.131910893085684e-05, "loss": 0.0093, "step": 56780 }, { "grad_norm": 0.06596618890762329, "learning_rate": 2.1302181440925012e-05, "loss": 0.0067, "step": 56790 }, { "grad_norm": 0.1265271008014679, "learning_rate": 2.128525885466546e-05, "loss": 0.0072, "step": 56800 }, { "grad_norm": 0.2271190881729126, "learning_rate": 2.1268341174969847e-05, "loss": 0.0096, "step": 56810 }, { "grad_norm": 0.15633119642734528, "learning_rate": 2.1251428404728894e-05, "loss": 0.0073, "step": 56820 }, { "grad_norm": 0.10171512514352798, "learning_rate": 2.123452054683256e-05, "loss": 0.0058, "step": 56830 }, { "grad_norm": 0.10816559940576553, "learning_rate": 2.1217617604169927e-05, "loss": 0.0088, "step": 56840 }, { "grad_norm": 0.14453864097595215, "learning_rate": 2.120071957962924e-05, "loss": 0.008, "step": 56850 }, { "grad_norm": 0.11133727431297302, "learning_rate": 2.1183826476097917e-05, "loss": 0.0058, "step": 56860 }, { "grad_norm": 0.1230446994304657, "learning_rate": 2.116693829646254e-05, "loss": 0.0072, "step": 56870 }, { "grad_norm": 0.2371629923582077, "learning_rate": 2.1150055043608806e-05, "loss": 0.0084, "step": 56880 }, { "grad_norm": 0.11188804358243942, "learning_rate": 2.113317672042162e-05, "loss": 0.0068, "step": 56890 }, { "grad_norm": 0.2105761021375656, "learning_rate": 2.1116303329785025e-05, "loss": 0.0072, "step": 56900 }, { "grad_norm": 0.11719592660665512, "learning_rate": 2.109943487458222e-05, "loss": 0.0107, "step": 56910 }, { "grad_norm": 0.14053526520729065, "learning_rate": 2.1082571357695574e-05, "loss": 0.0076, "step": 56920 }, { "grad_norm": 0.10180213302373886, "learning_rate": 2.1065712782006557e-05, "loss": 0.0089, "step": 56930 }, { "grad_norm": 0.12773895263671875, "learning_rate": 2.104885915039591e-05, "loss": 0.0079, "step": 56940 }, { "grad_norm": 0.11469591408967972, "learning_rate": 2.103201046574338e-05, "loss": 0.0077, "step": 56950 }, { "grad_norm": 0.10323365032672882, "learning_rate": 2.1015166730928022e-05, "loss": 0.0078, "step": 56960 }, { "grad_norm": 0.15042683482170105, "learning_rate": 2.0998327948827912e-05, "loss": 0.0086, "step": 56970 }, { "grad_norm": 0.17921693623065948, "learning_rate": 2.0981494122320363e-05, "loss": 0.0078, "step": 56980 }, { "grad_norm": 0.09571448713541031, "learning_rate": 2.0964665254281822e-05, "loss": 0.0068, "step": 56990 }, { "grad_norm": 0.1448332667350769, "learning_rate": 2.094784134758784e-05, "loss": 0.0087, "step": 57000 }, { "grad_norm": 0.1212635487318039, "learning_rate": 2.0931022405113226e-05, "loss": 0.0079, "step": 57010 }, { "grad_norm": 0.09153009206056595, "learning_rate": 2.091420842973183e-05, "loss": 0.0075, "step": 57020 }, { "grad_norm": 0.11821256577968597, "learning_rate": 2.0897399424316715e-05, "loss": 0.0077, "step": 57030 }, { "grad_norm": 0.1365649253129959, "learning_rate": 2.0880595391740078e-05, "loss": 0.0077, "step": 57040 }, { "grad_norm": 0.16986851394176483, "learning_rate": 2.0863796334873277e-05, "loss": 0.0075, "step": 57050 }, { "grad_norm": 0.11380542814731598, "learning_rate": 2.08470022565868e-05, "loss": 0.0077, "step": 57060 }, { "grad_norm": 0.13251475989818573, "learning_rate": 2.0830213159750317e-05, "loss": 0.0076, "step": 57070 }, { "grad_norm": 0.159101203083992, "learning_rate": 2.0813429047232596e-05, "loss": 0.0072, "step": 57080 }, { "grad_norm": 0.1449088752269745, "learning_rate": 2.0796649921901594e-05, "loss": 0.007, "step": 57090 }, { "grad_norm": 0.13119740784168243, "learning_rate": 2.077987578662441e-05, "loss": 0.0074, "step": 57100 }, { "grad_norm": 0.08861392736434937, "learning_rate": 2.0763106644267277e-05, "loss": 0.0087, "step": 57110 }, { "grad_norm": 0.1036280170083046, "learning_rate": 2.0746342497695607e-05, "loss": 0.0075, "step": 57120 }, { "grad_norm": 0.13470347225666046, "learning_rate": 2.0729583349773886e-05, "loss": 0.009, "step": 57130 }, { "grad_norm": 0.11533740907907486, "learning_rate": 2.0712829203365853e-05, "loss": 0.0071, "step": 57140 }, { "grad_norm": 0.18148189783096313, "learning_rate": 2.0696080061334267e-05, "loss": 0.011, "step": 57150 }, { "grad_norm": 0.13306193053722382, "learning_rate": 2.067933592654117e-05, "loss": 0.008, "step": 57160 }, { "grad_norm": 0.21345558762550354, "learning_rate": 2.066259680184763e-05, "loss": 0.0057, "step": 57170 }, { "grad_norm": 0.15078651905059814, "learning_rate": 2.0645862690113908e-05, "loss": 0.0083, "step": 57180 }, { "grad_norm": 0.1679978370666504, "learning_rate": 2.0629133594199436e-05, "loss": 0.0074, "step": 57190 }, { "grad_norm": 0.09340739250183105, "learning_rate": 2.0612409516962704e-05, "loss": 0.0096, "step": 57200 }, { "grad_norm": 0.11683040857315063, "learning_rate": 2.0595690461261467e-05, "loss": 0.0067, "step": 57210 }, { "grad_norm": 0.12062804400920868, "learning_rate": 2.0578976429952503e-05, "loss": 0.0079, "step": 57220 }, { "grad_norm": 0.10704754292964935, "learning_rate": 2.0562267425891802e-05, "loss": 0.0077, "step": 57230 }, { "grad_norm": 0.15703386068344116, "learning_rate": 2.0545563451934467e-05, "loss": 0.0074, "step": 57240 }, { "grad_norm": 0.11886090785264969, "learning_rate": 2.0528864510934764e-05, "loss": 0.0082, "step": 57250 }, { "grad_norm": 0.12585769593715668, "learning_rate": 2.0512170605746096e-05, "loss": 0.007, "step": 57260 }, { "grad_norm": 0.19602225720882416, "learning_rate": 2.0495481739220963e-05, "loss": 0.0072, "step": 57270 }, { "grad_norm": 0.10422110557556152, "learning_rate": 2.0478797914211045e-05, "loss": 0.0065, "step": 57280 }, { "grad_norm": 0.18629735708236694, "learning_rate": 2.046211913356716e-05, "loss": 0.0073, "step": 57290 }, { "grad_norm": 0.13167858123779297, "learning_rate": 2.0445445400139247e-05, "loss": 0.0071, "step": 57300 }, { "grad_norm": 0.08330002427101135, "learning_rate": 2.0428776716776405e-05, "loss": 0.0073, "step": 57310 }, { "grad_norm": 0.1813400387763977, "learning_rate": 2.0412113086326856e-05, "loss": 0.0068, "step": 57320 }, { "grad_norm": 0.1502014845609665, "learning_rate": 2.0395454511637918e-05, "loss": 0.0089, "step": 57330 }, { "grad_norm": 0.16749368607997894, "learning_rate": 2.037880099555616e-05, "loss": 0.0106, "step": 57340 }, { "grad_norm": 0.18851631879806519, "learning_rate": 2.0362152540927144e-05, "loss": 0.0073, "step": 57350 }, { "grad_norm": 0.12182223796844482, "learning_rate": 2.0345509150595666e-05, "loss": 0.0072, "step": 57360 }, { "grad_norm": 0.14188729226589203, "learning_rate": 2.0328870827405617e-05, "loss": 0.0085, "step": 57370 }, { "grad_norm": 0.0928051695227623, "learning_rate": 2.0312237574200043e-05, "loss": 0.007, "step": 57380 }, { "grad_norm": 0.13968637585639954, "learning_rate": 2.029560939382112e-05, "loss": 0.0081, "step": 57390 }, { "grad_norm": 0.09706242382526398, "learning_rate": 2.0278986289110097e-05, "loss": 0.006, "step": 57400 }, { "grad_norm": 0.13376273214817047, "learning_rate": 2.0262368262907484e-05, "loss": 0.0088, "step": 57410 }, { "grad_norm": 0.1089470311999321, "learning_rate": 2.024575531805279e-05, "loss": 0.0085, "step": 57420 }, { "grad_norm": 0.14701738953590393, "learning_rate": 2.0229147457384735e-05, "loss": 0.0072, "step": 57430 }, { "grad_norm": 0.10405580699443817, "learning_rate": 2.0212544683741157e-05, "loss": 0.007, "step": 57440 }, { "grad_norm": 0.13156679272651672, "learning_rate": 2.0195946999958976e-05, "loss": 0.0085, "step": 57450 }, { "grad_norm": 0.09609730541706085, "learning_rate": 2.017935440887434e-05, "loss": 0.0087, "step": 57460 }, { "grad_norm": 0.14106687903404236, "learning_rate": 2.0162766913322423e-05, "loss": 0.0082, "step": 57470 }, { "grad_norm": 0.1314324587583542, "learning_rate": 2.0146184516137588e-05, "loss": 0.0079, "step": 57480 }, { "grad_norm": 0.16524289548397064, "learning_rate": 2.012960722015332e-05, "loss": 0.0083, "step": 57490 }, { "grad_norm": 0.12956157326698303, "learning_rate": 2.0113035028202214e-05, "loss": 0.008, "step": 57500 }, { "grad_norm": 0.14425018429756165, "learning_rate": 2.009646794311602e-05, "loss": 0.0107, "step": 57510 }, { "grad_norm": 0.106838159263134, "learning_rate": 2.007990596772559e-05, "loss": 0.0074, "step": 57520 }, { "grad_norm": 0.09918879717588425, "learning_rate": 2.0063349104860923e-05, "loss": 0.0065, "step": 57530 }, { "grad_norm": 0.11405349522829056, "learning_rate": 2.0046797357351116e-05, "loss": 0.0078, "step": 57540 }, { "grad_norm": 0.10456936061382294, "learning_rate": 2.0030250728024412e-05, "loss": 0.0064, "step": 57550 }, { "grad_norm": 0.18537026643753052, "learning_rate": 2.001370921970819e-05, "loss": 0.0073, "step": 57560 }, { "grad_norm": 0.1443081647157669, "learning_rate": 1.9997172835228932e-05, "loss": 0.0087, "step": 57570 }, { "grad_norm": 0.11764086782932281, "learning_rate": 1.9980641577412262e-05, "loss": 0.0058, "step": 57580 }, { "grad_norm": 0.11485451459884644, "learning_rate": 1.9964115449082925e-05, "loss": 0.0071, "step": 57590 }, { "grad_norm": 0.13935686647891998, "learning_rate": 1.9947594453064742e-05, "loss": 0.0069, "step": 57600 }, { "grad_norm": 0.14895637333393097, "learning_rate": 1.9931078592180774e-05, "loss": 0.0087, "step": 57610 }, { "grad_norm": 0.12019339203834534, "learning_rate": 1.9914567869253065e-05, "loss": 0.0077, "step": 57620 }, { "grad_norm": 0.14998941123485565, "learning_rate": 1.989806228710287e-05, "loss": 0.0101, "step": 57630 }, { "grad_norm": 0.10488917678594589, "learning_rate": 1.9881561848550555e-05, "loss": 0.0076, "step": 57640 }, { "grad_norm": 0.18018528819084167, "learning_rate": 1.9865066556415544e-05, "loss": 0.0085, "step": 57650 }, { "grad_norm": 0.152784526348114, "learning_rate": 1.98485764135165e-05, "loss": 0.0062, "step": 57660 }, { "grad_norm": 0.11549048870801926, "learning_rate": 1.983209142267109e-05, "loss": 0.0078, "step": 57670 }, { "grad_norm": 0.19103458523750305, "learning_rate": 1.9815611586696165e-05, "loss": 0.0067, "step": 57680 }, { "grad_norm": 0.13526079058647156, "learning_rate": 1.9799136908407667e-05, "loss": 0.0085, "step": 57690 }, { "grad_norm": 0.08599065244197845, "learning_rate": 1.9782667390620678e-05, "loss": 0.0069, "step": 57700 }, { "grad_norm": 0.11890315264463425, "learning_rate": 1.976620303614939e-05, "loss": 0.0095, "step": 57710 }, { "grad_norm": 0.10835115611553192, "learning_rate": 1.9749743847807108e-05, "loss": 0.0067, "step": 57720 }, { "grad_norm": 0.16582298278808594, "learning_rate": 1.9733289828406272e-05, "loss": 0.0086, "step": 57730 }, { "grad_norm": 0.17139840126037598, "learning_rate": 1.9716840980758382e-05, "loss": 0.0119, "step": 57740 }, { "grad_norm": 0.08858689665794373, "learning_rate": 1.9700397307674134e-05, "loss": 0.0059, "step": 57750 }, { "grad_norm": 0.14637288451194763, "learning_rate": 1.968395881196328e-05, "loss": 0.0101, "step": 57760 }, { "grad_norm": 0.1006770208477974, "learning_rate": 1.966752549643473e-05, "loss": 0.0069, "step": 57770 }, { "grad_norm": 0.13675647974014282, "learning_rate": 1.965109736389647e-05, "loss": 0.0063, "step": 57780 }, { "grad_norm": 0.12504352629184723, "learning_rate": 1.9634674417155645e-05, "loss": 0.0074, "step": 57790 }, { "grad_norm": 0.11353828758001328, "learning_rate": 1.9618256659018434e-05, "loss": 0.0087, "step": 57800 }, { "grad_norm": 0.10563524067401886, "learning_rate": 1.9601844092290257e-05, "loss": 0.0093, "step": 57810 }, { "grad_norm": 0.16538642346858978, "learning_rate": 1.9585436719775512e-05, "loss": 0.0104, "step": 57820 }, { "grad_norm": 0.10301098227500916, "learning_rate": 1.9569034544277793e-05, "loss": 0.0064, "step": 57830 }, { "grad_norm": 0.09318075329065323, "learning_rate": 1.9552637568599798e-05, "loss": 0.0076, "step": 57840 }, { "grad_norm": 0.07307589054107666, "learning_rate": 1.953624579554327e-05, "loss": 0.0073, "step": 57850 }, { "grad_norm": 0.12367477267980576, "learning_rate": 1.951985922790918e-05, "loss": 0.0073, "step": 57860 }, { "grad_norm": 0.15313933789730072, "learning_rate": 1.9503477868497505e-05, "loss": 0.0094, "step": 57870 }, { "grad_norm": 0.15528364479541779, "learning_rate": 1.9487101720107375e-05, "loss": 0.0069, "step": 57880 }, { "grad_norm": 0.12067221105098724, "learning_rate": 1.9470730785537032e-05, "loss": 0.0082, "step": 57890 }, { "grad_norm": 0.11484372615814209, "learning_rate": 1.9454365067583823e-05, "loss": 0.0074, "step": 57900 }, { "grad_norm": 0.11617341637611389, "learning_rate": 1.9438004569044215e-05, "loss": 0.0079, "step": 57910 }, { "grad_norm": 0.20140106976032257, "learning_rate": 1.9421649292713724e-05, "loss": 0.0072, "step": 57920 }, { "grad_norm": 0.13064004480838776, "learning_rate": 1.9405299241387076e-05, "loss": 0.0071, "step": 57930 }, { "grad_norm": 0.1309690773487091, "learning_rate": 1.9388954417858007e-05, "loss": 0.0068, "step": 57940 }, { "grad_norm": 0.1849096417427063, "learning_rate": 1.9372614824919417e-05, "loss": 0.0077, "step": 57950 }, { "grad_norm": 0.1081213504076004, "learning_rate": 1.9356280465363284e-05, "loss": 0.0076, "step": 57960 }, { "grad_norm": 0.14703525602817535, "learning_rate": 1.9339951341980723e-05, "loss": 0.0074, "step": 57970 }, { "grad_norm": 0.12671563029289246, "learning_rate": 1.9323627457561916e-05, "loss": 0.0064, "step": 57980 }, { "grad_norm": 0.13748349249362946, "learning_rate": 1.9307308814896198e-05, "loss": 0.0074, "step": 57990 }, { "grad_norm": 0.17262467741966248, "learning_rate": 1.9290995416771935e-05, "loss": 0.007, "step": 58000 }, { "grad_norm": 0.08005020022392273, "learning_rate": 1.9274687265976665e-05, "loss": 0.0092, "step": 58010 }, { "grad_norm": 0.1146157830953598, "learning_rate": 1.9258384365297e-05, "loss": 0.0062, "step": 58020 }, { "grad_norm": 0.15046750009059906, "learning_rate": 1.924208671751866e-05, "loss": 0.0064, "step": 58030 }, { "grad_norm": 0.1082858294248581, "learning_rate": 1.9225794325426492e-05, "loss": 0.0066, "step": 58040 }, { "grad_norm": 0.09015929698944092, "learning_rate": 1.920950719180436e-05, "loss": 0.009, "step": 58050 }, { "grad_norm": 0.12570419907569885, "learning_rate": 1.919322531943536e-05, "loss": 0.0071, "step": 58060 }, { "grad_norm": 0.1304909586906433, "learning_rate": 1.917694871110157e-05, "loss": 0.0073, "step": 58070 }, { "grad_norm": 0.11424674093723297, "learning_rate": 1.9160677369584234e-05, "loss": 0.0078, "step": 58080 }, { "grad_norm": 0.11905576288700104, "learning_rate": 1.9144411297663694e-05, "loss": 0.006, "step": 58090 }, { "grad_norm": 0.16409289836883545, "learning_rate": 1.9128150498119328e-05, "loss": 0.008, "step": 58100 }, { "grad_norm": 0.11181636154651642, "learning_rate": 1.9111894973729726e-05, "loss": 0.0074, "step": 58110 }, { "grad_norm": 0.1273747682571411, "learning_rate": 1.9095644727272454e-05, "loss": 0.0078, "step": 58120 }, { "grad_norm": 0.13976433873176575, "learning_rate": 1.907939976152429e-05, "loss": 0.0078, "step": 58130 }, { "grad_norm": 0.1411312073469162, "learning_rate": 1.906316007926101e-05, "loss": 0.0061, "step": 58140 }, { "grad_norm": 0.09335871040821075, "learning_rate": 1.904692568325755e-05, "loss": 0.0063, "step": 58150 }, { "grad_norm": 0.10466072708368301, "learning_rate": 1.9030696576287925e-05, "loss": 0.0075, "step": 58160 }, { "grad_norm": 0.1178559735417366, "learning_rate": 1.9014472761125242e-05, "loss": 0.0076, "step": 58170 }, { "grad_norm": 0.10852651298046112, "learning_rate": 1.899825424054172e-05, "loss": 0.0071, "step": 58180 }, { "grad_norm": 0.1728370487689972, "learning_rate": 1.898204101730863e-05, "loss": 0.0071, "step": 58190 }, { "grad_norm": 0.07197733968496323, "learning_rate": 1.8965833094196394e-05, "loss": 0.0058, "step": 58200 }, { "grad_norm": 0.11086256057024002, "learning_rate": 1.8949630473974495e-05, "loss": 0.0056, "step": 58210 }, { "grad_norm": 0.15064065158367157, "learning_rate": 1.8933433159411517e-05, "loss": 0.0068, "step": 58220 }, { "grad_norm": 0.10053396224975586, "learning_rate": 1.891724115327514e-05, "loss": 0.0067, "step": 58230 }, { "grad_norm": 0.13725851476192474, "learning_rate": 1.8901054458332152e-05, "loss": 0.0071, "step": 58240 }, { "grad_norm": 0.15813201665878296, "learning_rate": 1.8884873077348364e-05, "loss": 0.0079, "step": 58250 }, { "grad_norm": 0.12518222630023956, "learning_rate": 1.88686970130888e-05, "loss": 0.006, "step": 58260 }, { "grad_norm": 0.1626112312078476, "learning_rate": 1.8852526268317456e-05, "loss": 0.0079, "step": 58270 }, { "grad_norm": 0.11595416814088821, "learning_rate": 1.883636084579749e-05, "loss": 0.0072, "step": 58280 }, { "grad_norm": 0.09749028086662292, "learning_rate": 1.8820200748291145e-05, "loss": 0.0093, "step": 58290 }, { "grad_norm": 0.16164514422416687, "learning_rate": 1.8804045978559686e-05, "loss": 0.0072, "step": 58300 }, { "grad_norm": 0.10543621331453323, "learning_rate": 1.8787896539363594e-05, "loss": 0.0078, "step": 58310 }, { "grad_norm": 0.17698591947555542, "learning_rate": 1.8771752433462298e-05, "loss": 0.0074, "step": 58320 }, { "grad_norm": 0.07214333862066269, "learning_rate": 1.8755613663614447e-05, "loss": 0.0066, "step": 58330 }, { "grad_norm": 0.14678263664245605, "learning_rate": 1.873948023257767e-05, "loss": 0.0071, "step": 58340 }, { "grad_norm": 0.12883105874061584, "learning_rate": 1.8723352143108747e-05, "loss": 0.0097, "step": 58350 }, { "grad_norm": 0.09496111422777176, "learning_rate": 1.870722939796352e-05, "loss": 0.0075, "step": 58360 }, { "grad_norm": 0.13568192720413208, "learning_rate": 1.869111199989693e-05, "loss": 0.0064, "step": 58370 }, { "grad_norm": 0.10222215205430984, "learning_rate": 1.8674999951663018e-05, "loss": 0.0075, "step": 58380 }, { "grad_norm": 0.10552340745925903, "learning_rate": 1.8658893256014857e-05, "loss": 0.0076, "step": 58390 }, { "grad_norm": 0.14326095581054688, "learning_rate": 1.8642791915704655e-05, "loss": 0.0077, "step": 58400 }, { "grad_norm": 0.09533694386482239, "learning_rate": 1.8626695933483697e-05, "loss": 0.0063, "step": 58410 }, { "grad_norm": 0.12887240946292877, "learning_rate": 1.8610605312102335e-05, "loss": 0.0076, "step": 58420 }, { "grad_norm": 0.13667471706867218, "learning_rate": 1.8594520054310032e-05, "loss": 0.0076, "step": 58430 }, { "grad_norm": 0.15492276847362518, "learning_rate": 1.8578440162855326e-05, "loss": 0.0091, "step": 58440 }, { "grad_norm": 0.11111614108085632, "learning_rate": 1.8562365640485774e-05, "loss": 0.0072, "step": 58450 }, { "grad_norm": 0.29232919216156006, "learning_rate": 1.854629648994815e-05, "loss": 0.0093, "step": 58460 }, { "grad_norm": 0.10953623801469803, "learning_rate": 1.8530232713988183e-05, "loss": 0.0077, "step": 58470 }, { "grad_norm": 0.09449335932731628, "learning_rate": 1.851417431535074e-05, "loss": 0.0096, "step": 58480 }, { "grad_norm": 0.125105082988739, "learning_rate": 1.8498121296779785e-05, "loss": 0.008, "step": 58490 }, { "grad_norm": 0.10710940510034561, "learning_rate": 1.848207366101829e-05, "loss": 0.009, "step": 58500 }, { "grad_norm": 0.11779169738292694, "learning_rate": 1.8466031410808422e-05, "loss": 0.0093, "step": 58510 }, { "grad_norm": 0.12305325269699097, "learning_rate": 1.844999454889129e-05, "loss": 0.0095, "step": 58520 }, { "grad_norm": 0.08873730897903442, "learning_rate": 1.8433963078007237e-05, "loss": 0.0073, "step": 58530 }, { "grad_norm": 0.19407512247562408, "learning_rate": 1.841793700089554e-05, "loss": 0.008, "step": 58540 }, { "grad_norm": 0.27246955037117004, "learning_rate": 1.8401916320294644e-05, "loss": 0.0081, "step": 58550 }, { "grad_norm": 0.17153826355934143, "learning_rate": 1.838590103894205e-05, "loss": 0.0095, "step": 58560 }, { "grad_norm": 0.09262704104185104, "learning_rate": 1.8369891159574292e-05, "loss": 0.009, "step": 58570 }, { "grad_norm": 0.11734256148338318, "learning_rate": 1.835388668492708e-05, "loss": 0.0086, "step": 58580 }, { "grad_norm": 0.15357469022274017, "learning_rate": 1.8337887617735095e-05, "loss": 0.0088, "step": 58590 }, { "grad_norm": 0.2152613401412964, "learning_rate": 1.8321893960732157e-05, "loss": 0.0108, "step": 58600 }, { "grad_norm": 0.14166991412639618, "learning_rate": 1.8305905716651138e-05, "loss": 0.0072, "step": 58610 }, { "grad_norm": 0.16088955104351044, "learning_rate": 1.8289922888224e-05, "loss": 0.0071, "step": 58620 }, { "grad_norm": 0.14008791744709015, "learning_rate": 1.8273945478181765e-05, "loss": 0.0108, "step": 58630 }, { "grad_norm": 0.15445932745933533, "learning_rate": 1.8257973489254558e-05, "loss": 0.0084, "step": 58640 }, { "grad_norm": 0.12186018377542496, "learning_rate": 1.824200692417152e-05, "loss": 0.0064, "step": 58650 }, { "grad_norm": 0.17225094139575958, "learning_rate": 1.8226045785660912e-05, "loss": 0.0083, "step": 58660 }, { "grad_norm": 0.20584756135940552, "learning_rate": 1.821009007645006e-05, "loss": 0.0106, "step": 58670 }, { "grad_norm": 0.10859659314155579, "learning_rate": 1.8194139799265357e-05, "loss": 0.0072, "step": 58680 }, { "grad_norm": 0.10527843981981277, "learning_rate": 1.8178194956832295e-05, "loss": 0.0103, "step": 58690 }, { "grad_norm": 0.10813170671463013, "learning_rate": 1.8162255551875346e-05, "loss": 0.0071, "step": 58700 }, { "grad_norm": 0.14412535727024078, "learning_rate": 1.8146321587118194e-05, "loss": 0.0084, "step": 58710 }, { "grad_norm": 0.1115204244852066, "learning_rate": 1.8130393065283448e-05, "loss": 0.0078, "step": 58720 }, { "grad_norm": 0.11640587449073792, "learning_rate": 1.8114469989092925e-05, "loss": 0.0071, "step": 58730 }, { "grad_norm": 0.11931728571653366, "learning_rate": 1.8098552361267397e-05, "loss": 0.0064, "step": 58740 }, { "grad_norm": 0.10388480126857758, "learning_rate": 1.8082640184526763e-05, "loss": 0.0076, "step": 58750 }, { "grad_norm": 0.14741209149360657, "learning_rate": 1.806673346158999e-05, "loss": 0.0095, "step": 58760 }, { "grad_norm": 0.11331182718276978, "learning_rate": 1.8050832195175067e-05, "loss": 0.0084, "step": 58770 }, { "grad_norm": 0.30106475949287415, "learning_rate": 1.8034936387999136e-05, "loss": 0.0092, "step": 58780 }, { "grad_norm": 0.14254283905029297, "learning_rate": 1.8019046042778315e-05, "loss": 0.0064, "step": 58790 }, { "grad_norm": 0.1448299139738083, "learning_rate": 1.800316116222785e-05, "loss": 0.0079, "step": 58800 }, { "grad_norm": 0.10817305743694305, "learning_rate": 1.7987281749062018e-05, "loss": 0.008, "step": 58810 }, { "grad_norm": 0.10979536920785904, "learning_rate": 1.7971407805994195e-05, "loss": 0.0078, "step": 58820 }, { "grad_norm": 0.4085473120212555, "learning_rate": 1.7955539335736787e-05, "loss": 0.0086, "step": 58830 }, { "grad_norm": 0.15552474558353424, "learning_rate": 1.7939676341001304e-05, "loss": 0.0063, "step": 58840 }, { "grad_norm": 0.09026901423931122, "learning_rate": 1.7923818824498275e-05, "loss": 0.0069, "step": 58850 }, { "grad_norm": 0.2840483784675598, "learning_rate": 1.7907966788937315e-05, "loss": 0.0129, "step": 58860 }, { "grad_norm": 0.1368281990289688, "learning_rate": 1.7892120237027116e-05, "loss": 0.0058, "step": 58870 }, { "grad_norm": 0.09262620657682419, "learning_rate": 1.7876279171475413e-05, "loss": 0.005, "step": 58880 }, { "grad_norm": 0.20695285499095917, "learning_rate": 1.7860443594989028e-05, "loss": 0.008, "step": 58890 }, { "grad_norm": 0.16286173462867737, "learning_rate": 1.784461351027379e-05, "loss": 0.0075, "step": 58900 }, { "grad_norm": 0.1150301918387413, "learning_rate": 1.7828788920034677e-05, "loss": 0.0066, "step": 58910 }, { "grad_norm": 0.22187195718288422, "learning_rate": 1.7812969826975623e-05, "loss": 0.0094, "step": 58920 }, { "grad_norm": 0.12371481209993362, "learning_rate": 1.7797156233799738e-05, "loss": 0.0058, "step": 58930 }, { "grad_norm": 0.07242710888385773, "learning_rate": 1.7781348143209094e-05, "loss": 0.0077, "step": 58940 }, { "grad_norm": 0.15803934633731842, "learning_rate": 1.7765545557904862e-05, "loss": 0.0078, "step": 58950 }, { "grad_norm": 0.08234938234090805, "learning_rate": 1.7749748480587302e-05, "loss": 0.0075, "step": 58960 }, { "grad_norm": 0.18288879096508026, "learning_rate": 1.773395691395564e-05, "loss": 0.0072, "step": 58970 }, { "grad_norm": 0.1331803947687149, "learning_rate": 1.7718170860708305e-05, "loss": 0.0067, "step": 58980 }, { "grad_norm": 0.15007176995277405, "learning_rate": 1.770239032354264e-05, "loss": 0.0083, "step": 58990 }, { "grad_norm": 0.1164565458893776, "learning_rate": 1.7686615305155124e-05, "loss": 0.0076, "step": 59000 }, { "grad_norm": 0.10398858785629272, "learning_rate": 1.767084580824128e-05, "loss": 0.0057, "step": 59010 }, { "grad_norm": 0.10083144158124924, "learning_rate": 1.765508183549569e-05, "loss": 0.009, "step": 59020 }, { "grad_norm": 0.13447657227516174, "learning_rate": 1.763932338961199e-05, "loss": 0.0099, "step": 59030 }, { "grad_norm": 0.07597346603870392, "learning_rate": 1.762357047328284e-05, "loss": 0.0067, "step": 59040 }, { "grad_norm": 0.12329069525003433, "learning_rate": 1.7607823089199997e-05, "loss": 0.01, "step": 59050 }, { "grad_norm": 0.10011323541402817, "learning_rate": 1.7592081240054265e-05, "loss": 0.0079, "step": 59060 }, { "grad_norm": 0.09062182158231735, "learning_rate": 1.7576344928535488e-05, "loss": 0.0062, "step": 59070 }, { "grad_norm": 0.20916655659675598, "learning_rate": 1.7560614157332573e-05, "loss": 0.007, "step": 59080 }, { "grad_norm": 0.1382821649312973, "learning_rate": 1.7544888929133495e-05, "loss": 0.0077, "step": 59090 }, { "grad_norm": 0.1186937615275383, "learning_rate": 1.752916924662522e-05, "loss": 0.0055, "step": 59100 }, { "grad_norm": 0.14670376479625702, "learning_rate": 1.751345511249387e-05, "loss": 0.007, "step": 59110 }, { "grad_norm": 0.1312362104654312, "learning_rate": 1.749774652942452e-05, "loss": 0.0065, "step": 59120 }, { "grad_norm": 0.12990765273571014, "learning_rate": 1.748204350010135e-05, "loss": 0.0076, "step": 59130 }, { "grad_norm": 0.1334749311208725, "learning_rate": 1.7466346027207574e-05, "loss": 0.0067, "step": 59140 }, { "grad_norm": 0.11560852825641632, "learning_rate": 1.745065411342547e-05, "loss": 0.0092, "step": 59150 }, { "grad_norm": 0.1445266604423523, "learning_rate": 1.7434967761436366e-05, "loss": 0.008, "step": 59160 }, { "grad_norm": 0.06124561280012131, "learning_rate": 1.741928697392058e-05, "loss": 0.0071, "step": 59170 }, { "grad_norm": 0.0974557027220726, "learning_rate": 1.7403611753557597e-05, "loss": 0.0062, "step": 59180 }, { "grad_norm": 0.14755892753601074, "learning_rate": 1.738794210302584e-05, "loss": 0.0084, "step": 59190 }, { "grad_norm": 0.07500240206718445, "learning_rate": 1.7372278025002837e-05, "loss": 0.0063, "step": 59200 }, { "grad_norm": 0.13530845940113068, "learning_rate": 1.7356619522165164e-05, "loss": 0.0061, "step": 59210 }, { "grad_norm": 0.088593989610672, "learning_rate": 1.7340966597188378e-05, "loss": 0.0061, "step": 59220 }, { "grad_norm": 0.174618661403656, "learning_rate": 1.732531925274722e-05, "loss": 0.0077, "step": 59230 }, { "grad_norm": 0.11470843106508255, "learning_rate": 1.7309677491515318e-05, "loss": 0.0081, "step": 59240 }, { "grad_norm": 0.12324650585651398, "learning_rate": 1.7294041316165455e-05, "loss": 0.0069, "step": 59250 }, { "grad_norm": 0.17241109907627106, "learning_rate": 1.727841072936942e-05, "loss": 0.0099, "step": 59260 }, { "grad_norm": 0.10808909684419632, "learning_rate": 1.7262785733798058e-05, "loss": 0.0062, "step": 59270 }, { "grad_norm": 0.1673278659582138, "learning_rate": 1.7247166332121246e-05, "loss": 0.007, "step": 59280 }, { "grad_norm": 0.12599699199199677, "learning_rate": 1.7231552527007933e-05, "loss": 0.0056, "step": 59290 }, { "grad_norm": 0.16382715106010437, "learning_rate": 1.721594432112606e-05, "loss": 0.0075, "step": 59300 }, { "grad_norm": 0.15270209312438965, "learning_rate": 1.7200341717142655e-05, "loss": 0.0068, "step": 59310 }, { "grad_norm": 0.11159294098615646, "learning_rate": 1.718474471772377e-05, "loss": 0.0086, "step": 59320 }, { "grad_norm": 0.11436381191015244, "learning_rate": 1.7169153325534528e-05, "loss": 0.0069, "step": 59330 }, { "grad_norm": 0.08750990778207779, "learning_rate": 1.7153567543239045e-05, "loss": 0.0054, "step": 59340 }, { "grad_norm": 0.1568642407655716, "learning_rate": 1.7137987373500525e-05, "loss": 0.0064, "step": 59350 }, { "grad_norm": 0.17800848186016083, "learning_rate": 1.7122412818981198e-05, "loss": 0.0072, "step": 59360 }, { "grad_norm": 0.13061992824077606, "learning_rate": 1.7106843882342283e-05, "loss": 0.0082, "step": 59370 }, { "grad_norm": 0.109474778175354, "learning_rate": 1.709128056624415e-05, "loss": 0.0049, "step": 59380 }, { "grad_norm": 0.14478163421154022, "learning_rate": 1.70757228733461e-05, "loss": 0.008, "step": 59390 }, { "grad_norm": 0.09882907569408417, "learning_rate": 1.706017080630653e-05, "loss": 0.009, "step": 59400 }, { "grad_norm": 0.10502864420413971, "learning_rate": 1.7044624367782873e-05, "loss": 0.0088, "step": 59410 }, { "grad_norm": 0.14032316207885742, "learning_rate": 1.7029083560431553e-05, "loss": 0.0072, "step": 59420 }, { "grad_norm": 0.12915346026420593, "learning_rate": 1.701354838690813e-05, "loss": 0.008, "step": 59430 }, { "grad_norm": 0.2014646679162979, "learning_rate": 1.6998018849867087e-05, "loss": 0.0085, "step": 59440 }, { "grad_norm": 0.15088364481925964, "learning_rate": 1.698249495196202e-05, "loss": 0.0072, "step": 59450 }, { "grad_norm": 0.163038969039917, "learning_rate": 1.6966976695845528e-05, "loss": 0.0076, "step": 59460 }, { "grad_norm": 0.10711783915758133, "learning_rate": 1.6951464084169268e-05, "loss": 0.0064, "step": 59470 }, { "grad_norm": 0.14528502523899078, "learning_rate": 1.6935957119583916e-05, "loss": 0.0074, "step": 59480 }, { "grad_norm": 0.1707136034965515, "learning_rate": 1.6920455804739205e-05, "loss": 0.0096, "step": 59490 }, { "grad_norm": 0.1604880690574646, "learning_rate": 1.6904960142283855e-05, "loss": 0.01, "step": 59500 }, { "grad_norm": 0.14765328168869019, "learning_rate": 1.6889470134865666e-05, "loss": 0.0093, "step": 59510 }, { "grad_norm": 0.1342283934354782, "learning_rate": 1.687398578513145e-05, "loss": 0.0078, "step": 59520 }, { "grad_norm": 0.09549161046743393, "learning_rate": 1.6858507095727066e-05, "loss": 0.0091, "step": 59530 }, { "grad_norm": 0.14222033321857452, "learning_rate": 1.6843034069297403e-05, "loss": 0.0071, "step": 59540 }, { "grad_norm": 0.1792348474264145, "learning_rate": 1.682756670848637e-05, "loss": 0.0098, "step": 59550 }, { "grad_norm": 0.15222322940826416, "learning_rate": 1.6812105015936937e-05, "loss": 0.01, "step": 59560 }, { "grad_norm": 0.1446392834186554, "learning_rate": 1.6796648994291027e-05, "loss": 0.0077, "step": 59570 }, { "grad_norm": 0.1356877237558365, "learning_rate": 1.678119864618973e-05, "loss": 0.0066, "step": 59580 }, { "grad_norm": 0.16994495689868927, "learning_rate": 1.6765753974273023e-05, "loss": 0.0075, "step": 59590 }, { "grad_norm": 0.11387874186038971, "learning_rate": 1.675031498118001e-05, "loss": 0.0075, "step": 59600 }, { "grad_norm": 0.11230447888374329, "learning_rate": 1.6734881669548804e-05, "loss": 0.0108, "step": 59610 }, { "grad_norm": 0.10675123333930969, "learning_rate": 1.671945404201648e-05, "loss": 0.0073, "step": 59620 }, { "grad_norm": 0.09521540254354477, "learning_rate": 1.670403210121927e-05, "loss": 0.0058, "step": 59630 }, { "grad_norm": 0.1969800591468811, "learning_rate": 1.6688615849792312e-05, "loss": 0.012, "step": 59640 }, { "grad_norm": 0.09141869097948074, "learning_rate": 1.6673205290369832e-05, "loss": 0.0057, "step": 59650 }, { "grad_norm": 0.10737227648496628, "learning_rate": 1.6657800425585086e-05, "loss": 0.0062, "step": 59660 }, { "grad_norm": 0.13224560022354126, "learning_rate": 1.6642401258070327e-05, "loss": 0.0076, "step": 59670 }, { "grad_norm": 0.09932629019021988, "learning_rate": 1.6627007790456884e-05, "loss": 0.0093, "step": 59680 }, { "grad_norm": 0.14189709722995758, "learning_rate": 1.6611620025375037e-05, "loss": 0.0085, "step": 59690 }, { "grad_norm": 0.13194561004638672, "learning_rate": 1.6596237965454154e-05, "loss": 0.0086, "step": 59700 }, { "grad_norm": 0.10037019103765488, "learning_rate": 1.6580861613322607e-05, "loss": 0.0088, "step": 59710 }, { "grad_norm": 0.09420958906412125, "learning_rate": 1.6565490971607796e-05, "loss": 0.0074, "step": 59720 }, { "grad_norm": 0.12226289510726929, "learning_rate": 1.655012604293614e-05, "loss": 0.0076, "step": 59730 }, { "grad_norm": 0.1316644847393036, "learning_rate": 1.653476682993309e-05, "loss": 0.0067, "step": 59740 }, { "grad_norm": 0.10928046703338623, "learning_rate": 1.651941333522311e-05, "loss": 0.0079, "step": 59750 }, { "grad_norm": 0.11220955848693848, "learning_rate": 1.6504065561429715e-05, "loss": 0.0075, "step": 59760 }, { "grad_norm": 0.14682981371879578, "learning_rate": 1.6488723511175385e-05, "loss": 0.0087, "step": 59770 }, { "grad_norm": 0.19007866084575653, "learning_rate": 1.6473387187081668e-05, "loss": 0.0108, "step": 59780 }, { "grad_norm": 0.11797046661376953, "learning_rate": 1.6458056591769123e-05, "loss": 0.0086, "step": 59790 }, { "grad_norm": 0.13865695893764496, "learning_rate": 1.6442731727857335e-05, "loss": 0.0068, "step": 59800 }, { "grad_norm": 0.1182093620300293, "learning_rate": 1.642741259796492e-05, "loss": 0.0061, "step": 59810 }, { "grad_norm": 0.22003501653671265, "learning_rate": 1.641209920470944e-05, "loss": 0.0075, "step": 59820 }, { "grad_norm": 0.164928138256073, "learning_rate": 1.639679155070762e-05, "loss": 0.0095, "step": 59830 }, { "grad_norm": 0.0824459046125412, "learning_rate": 1.6381489638575048e-05, "loss": 0.0092, "step": 59840 }, { "grad_norm": 0.08783243596553802, "learning_rate": 1.636619347092643e-05, "loss": 0.0103, "step": 59850 }, { "grad_norm": 0.1730486899614334, "learning_rate": 1.6350903050375476e-05, "loss": 0.0081, "step": 59860 }, { "grad_norm": 0.19452621042728424, "learning_rate": 1.6335618379534856e-05, "loss": 0.0084, "step": 59870 }, { "grad_norm": 0.1273188441991806, "learning_rate": 1.6320339461016364e-05, "loss": 0.0079, "step": 59880 }, { "grad_norm": 0.13859041035175323, "learning_rate": 1.6305066297430687e-05, "loss": 0.0072, "step": 59890 }, { "grad_norm": 0.27357712388038635, "learning_rate": 1.6289798891387654e-05, "loss": 0.0092, "step": 59900 }, { "grad_norm": 0.11897286772727966, "learning_rate": 1.6274537245495995e-05, "loss": 0.0069, "step": 59910 }, { "grad_norm": 0.113404281437397, "learning_rate": 1.6259281362363527e-05, "loss": 0.0064, "step": 59920 }, { "grad_norm": 0.08994641155004501, "learning_rate": 1.6244031244597068e-05, "loss": 0.0066, "step": 59930 }, { "grad_norm": 0.1000523790717125, "learning_rate": 1.622878689480244e-05, "loss": 0.0059, "step": 59940 }, { "grad_norm": 0.08033177256584167, "learning_rate": 1.6213548315584498e-05, "loss": 0.0066, "step": 59950 }, { "grad_norm": 0.2001868337392807, "learning_rate": 1.6198315509547074e-05, "loss": 0.0053, "step": 59960 }, { "grad_norm": 0.10394587367773056, "learning_rate": 1.6183088479293056e-05, "loss": 0.0056, "step": 59970 }, { "grad_norm": 0.12977154552936554, "learning_rate": 1.6167867227424316e-05, "loss": 0.0061, "step": 59980 }, { "grad_norm": 0.11151678115129471, "learning_rate": 1.615265175654176e-05, "loss": 0.0078, "step": 59990 }, { "grad_norm": 0.14410418272018433, "learning_rate": 1.613744206924529e-05, "loss": 0.0058, "step": 60000 }, { "grad_norm": 0.15090219676494598, "learning_rate": 1.6122238168133845e-05, "loss": 0.009, "step": 60010 }, { "grad_norm": 0.11700543761253357, "learning_rate": 1.6107040055805305e-05, "loss": 0.0096, "step": 60020 }, { "grad_norm": 0.17821532487869263, "learning_rate": 1.6091847734856673e-05, "loss": 0.0088, "step": 60030 }, { "grad_norm": 0.09075810760259628, "learning_rate": 1.6076661207883865e-05, "loss": 0.0072, "step": 60040 }, { "grad_norm": 0.10353308171033859, "learning_rate": 1.6061480477481848e-05, "loss": 0.0074, "step": 60050 }, { "grad_norm": 0.10156417638063431, "learning_rate": 1.604630554624461e-05, "loss": 0.0057, "step": 60060 }, { "grad_norm": 0.11271177232265472, "learning_rate": 1.603113641676509e-05, "loss": 0.0096, "step": 60070 }, { "grad_norm": 0.09787759184837341, "learning_rate": 1.6015973091635338e-05, "loss": 0.0058, "step": 60080 }, { "grad_norm": 0.19190779328346252, "learning_rate": 1.6000815573446288e-05, "loss": 0.0079, "step": 60090 }, { "grad_norm": 0.10620525479316711, "learning_rate": 1.5985663864788002e-05, "loss": 0.006, "step": 60100 }, { "grad_norm": 0.11008070409297943, "learning_rate": 1.597051796824946e-05, "loss": 0.0052, "step": 60110 }, { "grad_norm": 0.17424997687339783, "learning_rate": 1.5955377886418682e-05, "loss": 0.0075, "step": 60120 }, { "grad_norm": 0.08664824068546295, "learning_rate": 1.5940243621882704e-05, "loss": 0.0073, "step": 60130 }, { "grad_norm": 0.14659255743026733, "learning_rate": 1.5925115177227555e-05, "loss": 0.0093, "step": 60140 }, { "grad_norm": 0.09978985041379929, "learning_rate": 1.5909992555038288e-05, "loss": 0.008, "step": 60150 }, { "grad_norm": 0.12978604435920715, "learning_rate": 1.5894875757898912e-05, "loss": 0.0073, "step": 60160 }, { "grad_norm": 0.11948299407958984, "learning_rate": 1.58797647883925e-05, "loss": 0.0063, "step": 60170 }, { "grad_norm": 0.12814924120903015, "learning_rate": 1.586465964910109e-05, "loss": 0.0096, "step": 60180 }, { "grad_norm": 0.14528191089630127, "learning_rate": 1.5849560342605734e-05, "loss": 0.0103, "step": 60190 }, { "grad_norm": 0.1856812983751297, "learning_rate": 1.583446687148651e-05, "loss": 0.0089, "step": 60200 }, { "grad_norm": 0.0693582221865654, "learning_rate": 1.581937923832248e-05, "loss": 0.0071, "step": 60210 }, { "grad_norm": 0.1355593055486679, "learning_rate": 1.5804297445691663e-05, "loss": 0.0073, "step": 60220 }, { "grad_norm": 0.10694271326065063, "learning_rate": 1.578922149617119e-05, "loss": 0.0072, "step": 60230 }, { "grad_norm": 0.14313258230686188, "learning_rate": 1.5774151392337084e-05, "loss": 0.0058, "step": 60240 }, { "grad_norm": 0.15147818624973297, "learning_rate": 1.575908713676442e-05, "loss": 0.0071, "step": 60250 }, { "grad_norm": 0.12088742107152939, "learning_rate": 1.5744028732027293e-05, "loss": 0.0074, "step": 60260 }, { "grad_norm": 0.25976622104644775, "learning_rate": 1.5728976180698723e-05, "loss": 0.006, "step": 60270 }, { "grad_norm": 0.13248541951179504, "learning_rate": 1.5713929485350836e-05, "loss": 0.0062, "step": 60280 }, { "grad_norm": 0.10341402888298035, "learning_rate": 1.569888864855465e-05, "loss": 0.0057, "step": 60290 }, { "grad_norm": 0.1186731681227684, "learning_rate": 1.568385367288028e-05, "loss": 0.0086, "step": 60300 }, { "grad_norm": 0.1155214011669159, "learning_rate": 1.5668824560896755e-05, "loss": 0.0081, "step": 60310 }, { "grad_norm": 0.12089289724826813, "learning_rate": 1.5653801315172156e-05, "loss": 0.0061, "step": 60320 }, { "grad_norm": 0.08577431738376617, "learning_rate": 1.5638783938273554e-05, "loss": 0.0064, "step": 60330 }, { "grad_norm": 0.11869765818119049, "learning_rate": 1.5623772432766966e-05, "loss": 0.0068, "step": 60340 }, { "grad_norm": 0.13633984327316284, "learning_rate": 1.5608766801217507e-05, "loss": 0.0079, "step": 60350 }, { "grad_norm": 0.1221836507320404, "learning_rate": 1.5593767046189184e-05, "loss": 0.0069, "step": 60360 }, { "grad_norm": 0.08686850965023041, "learning_rate": 1.5578773170245064e-05, "loss": 0.0058, "step": 60370 }, { "grad_norm": 0.1757706105709076, "learning_rate": 1.5563785175947182e-05, "loss": 0.0065, "step": 60380 }, { "grad_norm": 0.19325734674930573, "learning_rate": 1.5548803065856582e-05, "loss": 0.0089, "step": 60390 }, { "grad_norm": 0.08468885719776154, "learning_rate": 1.55338268425333e-05, "loss": 0.0058, "step": 60400 }, { "grad_norm": 0.14040449261665344, "learning_rate": 1.5518856508536373e-05, "loss": 0.0094, "step": 60410 }, { "grad_norm": 0.20997998118400574, "learning_rate": 1.5503892066423786e-05, "loss": 0.0084, "step": 60420 }, { "grad_norm": 0.0936800092458725, "learning_rate": 1.548893351875258e-05, "loss": 0.0059, "step": 60430 }, { "grad_norm": 0.21669067442417145, "learning_rate": 1.547398086807875e-05, "loss": 0.0073, "step": 60440 }, { "grad_norm": 0.11642152070999146, "learning_rate": 1.5459034116957305e-05, "loss": 0.0059, "step": 60450 }, { "grad_norm": 0.16539479792118073, "learning_rate": 1.544409326794225e-05, "loss": 0.0074, "step": 60460 }, { "grad_norm": 0.1013946384191513, "learning_rate": 1.542915832358651e-05, "loss": 0.0089, "step": 60470 }, { "grad_norm": 0.18599703907966614, "learning_rate": 1.541422928644214e-05, "loss": 0.0104, "step": 60480 }, { "grad_norm": 0.1564996838569641, "learning_rate": 1.5399306159060024e-05, "loss": 0.0097, "step": 60490 }, { "grad_norm": 0.16405089199543, "learning_rate": 1.5384388943990185e-05, "loss": 0.0064, "step": 60500 }, { "grad_norm": 0.1590985208749771, "learning_rate": 1.5369477643781526e-05, "loss": 0.0064, "step": 60510 }, { "grad_norm": 0.23807662725448608, "learning_rate": 1.5354572260981985e-05, "loss": 0.0059, "step": 60520 }, { "grad_norm": 0.12914492189884186, "learning_rate": 1.533967279813851e-05, "loss": 0.0078, "step": 60530 }, { "grad_norm": 0.09783292561769485, "learning_rate": 1.5324779257796956e-05, "loss": 0.0075, "step": 60540 }, { "grad_norm": 0.10686344653367996, "learning_rate": 1.5309891642502293e-05, "loss": 0.0082, "step": 60550 }, { "grad_norm": 0.17596220970153809, "learning_rate": 1.5295009954798357e-05, "loss": 0.0063, "step": 60560 }, { "grad_norm": 0.099618099629879, "learning_rate": 1.5280134197228036e-05, "loss": 0.0077, "step": 60570 }, { "grad_norm": 0.10286998748779297, "learning_rate": 1.526526437233319e-05, "loss": 0.0061, "step": 60580 }, { "grad_norm": 0.1836262196302414, "learning_rate": 1.525040048265466e-05, "loss": 0.0084, "step": 60590 }, { "grad_norm": 0.17139354348182678, "learning_rate": 1.5235542530732288e-05, "loss": 0.0066, "step": 60600 }, { "grad_norm": 0.08501509577035904, "learning_rate": 1.5220690519104901e-05, "loss": 0.0058, "step": 60610 }, { "grad_norm": 0.10927315801382065, "learning_rate": 1.5205844450310275e-05, "loss": 0.0075, "step": 60620 }, { "grad_norm": 0.11222812533378601, "learning_rate": 1.519100432688521e-05, "loss": 0.0072, "step": 60630 }, { "grad_norm": 0.13491500914096832, "learning_rate": 1.5176170151365476e-05, "loss": 0.0057, "step": 60640 }, { "grad_norm": 0.09074302017688751, "learning_rate": 1.5161341926285833e-05, "loss": 0.0065, "step": 60650 }, { "grad_norm": 0.11471735686063766, "learning_rate": 1.5146519654180025e-05, "loss": 0.0091, "step": 60660 }, { "grad_norm": 0.17131061851978302, "learning_rate": 1.5131703337580739e-05, "loss": 0.0068, "step": 60670 }, { "grad_norm": 0.16956382989883423, "learning_rate": 1.5116892979019731e-05, "loss": 0.0083, "step": 60680 }, { "grad_norm": 0.09105345606803894, "learning_rate": 1.5102088581027623e-05, "loss": 0.0096, "step": 60690 }, { "grad_norm": 0.09078479558229446, "learning_rate": 1.5087290146134154e-05, "loss": 0.0066, "step": 60700 }, { "grad_norm": 0.0833481103181839, "learning_rate": 1.5072497676867915e-05, "loss": 0.0057, "step": 60710 }, { "grad_norm": 0.18339717388153076, "learning_rate": 1.505771117575655e-05, "loss": 0.0095, "step": 60720 }, { "grad_norm": 0.08737876266241074, "learning_rate": 1.5042930645326691e-05, "loss": 0.0054, "step": 60730 }, { "grad_norm": 0.13118785619735718, "learning_rate": 1.5028156088103879e-05, "loss": 0.0075, "step": 60740 }, { "grad_norm": 0.07333072274923325, "learning_rate": 1.5013387506612735e-05, "loss": 0.0061, "step": 60750 }, { "grad_norm": 0.09606026858091354, "learning_rate": 1.4998624903376767e-05, "loss": 0.0079, "step": 60760 }, { "grad_norm": 0.1642058938741684, "learning_rate": 1.4983868280918517e-05, "loss": 0.006, "step": 60770 }, { "grad_norm": 0.0583561547100544, "learning_rate": 1.4969117641759478e-05, "loss": 0.0056, "step": 60780 }, { "grad_norm": 0.1259188950061798, "learning_rate": 1.495437298842014e-05, "loss": 0.007, "step": 60790 }, { "grad_norm": 0.20968566834926605, "learning_rate": 1.4939634323419976e-05, "loss": 0.0108, "step": 60800 }, { "grad_norm": 0.08965563029050827, "learning_rate": 1.4924901649277384e-05, "loss": 0.0099, "step": 60810 }, { "grad_norm": 0.08857887983322144, "learning_rate": 1.4910174968509793e-05, "loss": 0.0084, "step": 60820 }, { "grad_norm": 0.11150664836168289, "learning_rate": 1.4895454283633598e-05, "loss": 0.0059, "step": 60830 }, { "grad_norm": 0.1463012993335724, "learning_rate": 1.488073959716415e-05, "loss": 0.0063, "step": 60840 }, { "grad_norm": 0.06512371450662613, "learning_rate": 1.4866030911615791e-05, "loss": 0.0075, "step": 60850 }, { "grad_norm": 0.1134723648428917, "learning_rate": 1.4851328229501849e-05, "loss": 0.0065, "step": 60860 }, { "grad_norm": 0.0936429351568222, "learning_rate": 1.4836631553334562e-05, "loss": 0.0047, "step": 60870 }, { "grad_norm": 0.16671717166900635, "learning_rate": 1.4821940885625251e-05, "loss": 0.0066, "step": 60880 }, { "grad_norm": 0.06961824744939804, "learning_rate": 1.4807256228884109e-05, "loss": 0.0065, "step": 60890 }, { "grad_norm": 0.09031431376934052, "learning_rate": 1.4792577585620353e-05, "loss": 0.0068, "step": 60900 }, { "grad_norm": 0.15555483102798462, "learning_rate": 1.4777904958342164e-05, "loss": 0.007, "step": 60910 }, { "grad_norm": 0.10344758629798889, "learning_rate": 1.4763238349556691e-05, "loss": 0.0052, "step": 60920 }, { "grad_norm": 0.1123596653342247, "learning_rate": 1.4748577761770072e-05, "loss": 0.0058, "step": 60930 }, { "grad_norm": 0.11813365668058395, "learning_rate": 1.4733923197487354e-05, "loss": 0.0054, "step": 60940 }, { "grad_norm": 0.09418714046478271, "learning_rate": 1.4719274659212662e-05, "loss": 0.0065, "step": 60950 }, { "grad_norm": 0.1713634878396988, "learning_rate": 1.4704632149448983e-05, "loss": 0.0069, "step": 60960 }, { "grad_norm": 0.15839113295078278, "learning_rate": 1.4689995670698343e-05, "loss": 0.0055, "step": 60970 }, { "grad_norm": 0.18640382587909698, "learning_rate": 1.4675365225461728e-05, "loss": 0.0071, "step": 60980 }, { "grad_norm": 0.08341354131698608, "learning_rate": 1.4660740816239032e-05, "loss": 0.0062, "step": 60990 }, { "grad_norm": 0.1075768992304802, "learning_rate": 1.464612244552923e-05, "loss": 0.0046, "step": 61000 }, { "grad_norm": 0.1898023635149002, "learning_rate": 1.4631510115830161e-05, "loss": 0.0046, "step": 61010 }, { "grad_norm": 0.11688042432069778, "learning_rate": 1.4616903829638679e-05, "loss": 0.0083, "step": 61020 }, { "grad_norm": 0.10016936808824539, "learning_rate": 1.46023035894506e-05, "loss": 0.0098, "step": 61030 }, { "grad_norm": 0.11996138840913773, "learning_rate": 1.4587709397760713e-05, "loss": 0.0063, "step": 61040 }, { "grad_norm": 0.08330871164798737, "learning_rate": 1.4573121257062755e-05, "loss": 0.0074, "step": 61050 }, { "grad_norm": 0.12546707689762115, "learning_rate": 1.4558539169849472e-05, "loss": 0.0073, "step": 61060 }, { "grad_norm": 0.0720815435051918, "learning_rate": 1.4543963138612499e-05, "loss": 0.0074, "step": 61070 }, { "grad_norm": 0.19379064440727234, "learning_rate": 1.4529393165842498e-05, "loss": 0.0105, "step": 61080 }, { "grad_norm": 0.1309579312801361, "learning_rate": 1.4514829254029084e-05, "loss": 0.0069, "step": 61090 }, { "grad_norm": 0.13168732821941376, "learning_rate": 1.450027140566083e-05, "loss": 0.0085, "step": 61100 }, { "grad_norm": 0.12654227018356323, "learning_rate": 1.4485719623225268e-05, "loss": 0.0072, "step": 61110 }, { "grad_norm": 0.09142458438873291, "learning_rate": 1.4471173909208912e-05, "loss": 0.0053, "step": 61120 }, { "grad_norm": 0.10887029021978378, "learning_rate": 1.4456634266097236e-05, "loss": 0.0063, "step": 61130 }, { "grad_norm": 0.13918037712574005, "learning_rate": 1.444210069637461e-05, "loss": 0.0068, "step": 61140 }, { "grad_norm": 0.15891313552856445, "learning_rate": 1.4427573202524502e-05, "loss": 0.0058, "step": 61150 }, { "grad_norm": 0.13343298435211182, "learning_rate": 1.4413051787029208e-05, "loss": 0.0062, "step": 61160 }, { "grad_norm": 0.16490668058395386, "learning_rate": 1.439853645237006e-05, "loss": 0.0064, "step": 61170 }, { "grad_norm": 0.17877240478992462, "learning_rate": 1.4384027201027344e-05, "loss": 0.0064, "step": 61180 }, { "grad_norm": 0.10476931184530258, "learning_rate": 1.4369524035480253e-05, "loss": 0.0055, "step": 61190 }, { "grad_norm": 0.13780498504638672, "learning_rate": 1.4355026958207035e-05, "loss": 0.0073, "step": 61200 }, { "grad_norm": 0.11924269795417786, "learning_rate": 1.434053597168481e-05, "loss": 0.0069, "step": 61210 }, { "grad_norm": 0.14469440281391144, "learning_rate": 1.432605107838969e-05, "loss": 0.0071, "step": 61220 }, { "grad_norm": 0.12776146829128265, "learning_rate": 1.4311572280796759e-05, "loss": 0.0061, "step": 61230 }, { "grad_norm": 0.1798306256532669, "learning_rate": 1.4297099581380047e-05, "loss": 0.0067, "step": 61240 }, { "grad_norm": 0.08763042837381363, "learning_rate": 1.4282632982612538e-05, "loss": 0.0069, "step": 61250 }, { "grad_norm": 0.15718518197536469, "learning_rate": 1.4268172486966197e-05, "loss": 0.0054, "step": 61260 }, { "grad_norm": 0.11067840456962585, "learning_rate": 1.4253718096911894e-05, "loss": 0.0037, "step": 61270 }, { "grad_norm": 0.08708188682794571, "learning_rate": 1.4239269814919504e-05, "loss": 0.0068, "step": 61280 }, { "grad_norm": 0.1393703818321228, "learning_rate": 1.4224827643457843e-05, "loss": 0.0062, "step": 61290 }, { "grad_norm": 0.16642612218856812, "learning_rate": 1.4210391584994686e-05, "loss": 0.0073, "step": 61300 }, { "grad_norm": 0.0957452803850174, "learning_rate": 1.4195961641996763e-05, "loss": 0.0052, "step": 61310 }, { "grad_norm": 0.13256525993347168, "learning_rate": 1.4181537816929751e-05, "loss": 0.0072, "step": 61320 }, { "grad_norm": 0.09082549810409546, "learning_rate": 1.4167120112258303e-05, "loss": 0.0081, "step": 61330 }, { "grad_norm": 0.08554109185934067, "learning_rate": 1.4152708530445963e-05, "loss": 0.0069, "step": 61340 }, { "grad_norm": 0.19240674376487732, "learning_rate": 1.4138303073955344e-05, "loss": 0.008, "step": 61350 }, { "grad_norm": 0.14193972945213318, "learning_rate": 1.4123903745247897e-05, "loss": 0.0062, "step": 61360 }, { "grad_norm": 0.14572124183177948, "learning_rate": 1.4109510546784082e-05, "loss": 0.0062, "step": 61370 }, { "grad_norm": 0.09709011018276215, "learning_rate": 1.4095123481023325e-05, "loss": 0.0055, "step": 61380 }, { "grad_norm": 0.14838235080242157, "learning_rate": 1.4080742550423936e-05, "loss": 0.0063, "step": 61390 }, { "grad_norm": 0.10972820222377777, "learning_rate": 1.4066367757443278e-05, "loss": 0.0078, "step": 61400 }, { "grad_norm": 0.11269954591989517, "learning_rate": 1.4051999104537572e-05, "loss": 0.0077, "step": 61410 }, { "grad_norm": 0.1476793736219406, "learning_rate": 1.4037636594162034e-05, "loss": 0.0088, "step": 61420 }, { "grad_norm": 0.09435363858938217, "learning_rate": 1.402328022877083e-05, "loss": 0.0057, "step": 61430 }, { "grad_norm": 0.11901181191205978, "learning_rate": 1.4008930010817073e-05, "loss": 0.0054, "step": 61440 }, { "grad_norm": 0.13481682538986206, "learning_rate": 1.3994585942752831e-05, "loss": 0.0074, "step": 61450 }, { "grad_norm": 0.11556590348482132, "learning_rate": 1.398024802702909e-05, "loss": 0.0093, "step": 61460 }, { "grad_norm": 0.08077353239059448, "learning_rate": 1.3965916266095819e-05, "loss": 0.0053, "step": 61470 }, { "grad_norm": 0.1161566972732544, "learning_rate": 1.395159066240192e-05, "loss": 0.0078, "step": 61480 }, { "grad_norm": 0.1181768998503685, "learning_rate": 1.3937271218395248e-05, "loss": 0.0075, "step": 61490 }, { "grad_norm": 0.11545499414205551, "learning_rate": 1.3922957936522613e-05, "loss": 0.0076, "step": 61500 }, { "grad_norm": 0.07593037933111191, "learning_rate": 1.3908650819229758e-05, "loss": 0.0055, "step": 61510 }, { "grad_norm": 0.1399618536233902, "learning_rate": 1.389434986896137e-05, "loss": 0.0076, "step": 61520 }, { "grad_norm": 0.17340320348739624, "learning_rate": 1.388005508816112e-05, "loss": 0.0066, "step": 61530 }, { "grad_norm": 0.09973734617233276, "learning_rate": 1.3865766479271557e-05, "loss": 0.0058, "step": 61540 }, { "grad_norm": 0.0953281819820404, "learning_rate": 1.3851484044734225e-05, "loss": 0.007, "step": 61550 }, { "grad_norm": 0.1706375926733017, "learning_rate": 1.3837207786989608e-05, "loss": 0.0066, "step": 61560 }, { "grad_norm": 0.119709812104702, "learning_rate": 1.382293770847713e-05, "loss": 0.0065, "step": 61570 }, { "grad_norm": 0.11972395330667496, "learning_rate": 1.3808673811635159e-05, "loss": 0.0077, "step": 61580 }, { "grad_norm": 0.09543448686599731, "learning_rate": 1.3794416098900975e-05, "loss": 0.0058, "step": 61590 }, { "grad_norm": 0.14466530084609985, "learning_rate": 1.3780164572710886e-05, "loss": 0.0082, "step": 61600 }, { "grad_norm": 0.12007902562618256, "learning_rate": 1.3765919235500035e-05, "loss": 0.0067, "step": 61610 }, { "grad_norm": 0.08696576207876205, "learning_rate": 1.375168008970259e-05, "loss": 0.0083, "step": 61620 }, { "grad_norm": 0.1309143453836441, "learning_rate": 1.3737447137751635e-05, "loss": 0.0081, "step": 61630 }, { "grad_norm": 0.1519271433353424, "learning_rate": 1.372322038207915e-05, "loss": 0.0059, "step": 61640 }, { "grad_norm": 0.1552470326423645, "learning_rate": 1.3708999825116159e-05, "loss": 0.006, "step": 61650 }, { "grad_norm": 0.18096914887428284, "learning_rate": 1.3694785469292526e-05, "loss": 0.0062, "step": 61660 }, { "grad_norm": 0.0681207999587059, "learning_rate": 1.3680577317037101e-05, "loss": 0.0069, "step": 61670 }, { "grad_norm": 0.12869566679000854, "learning_rate": 1.3666375370777678e-05, "loss": 0.0057, "step": 61680 }, { "grad_norm": 0.11457830667495728, "learning_rate": 1.3652179632940981e-05, "loss": 0.0069, "step": 61690 }, { "grad_norm": 0.15622565150260925, "learning_rate": 1.3637990105952664e-05, "loss": 0.007, "step": 61700 }, { "grad_norm": 0.11430820822715759, "learning_rate": 1.3623806792237337e-05, "loss": 0.0064, "step": 61710 }, { "grad_norm": 0.08943382650613785, "learning_rate": 1.3609629694218551e-05, "loss": 0.0077, "step": 61720 }, { "grad_norm": 0.14961698651313782, "learning_rate": 1.3595458814318762e-05, "loss": 0.0066, "step": 61730 }, { "grad_norm": 0.2464928925037384, "learning_rate": 1.3581294154959389e-05, "loss": 0.0057, "step": 61740 }, { "grad_norm": 0.08497917652130127, "learning_rate": 1.3567135718560792e-05, "loss": 0.0066, "step": 61750 }, { "grad_norm": 0.12453684210777283, "learning_rate": 1.3552983507542261e-05, "loss": 0.007, "step": 61760 }, { "grad_norm": 0.2315596640110016, "learning_rate": 1.3538837524322023e-05, "loss": 0.0084, "step": 61770 }, { "grad_norm": 0.10321587324142456, "learning_rate": 1.3524697771317251e-05, "loss": 0.0084, "step": 61780 }, { "grad_norm": 0.18988196551799774, "learning_rate": 1.3510564250943997e-05, "loss": 0.0089, "step": 61790 }, { "grad_norm": 0.15095986425876617, "learning_rate": 1.3496436965617353e-05, "loss": 0.0095, "step": 61800 }, { "grad_norm": 0.12971708178520203, "learning_rate": 1.3482315917751243e-05, "loss": 0.0048, "step": 61810 }, { "grad_norm": 0.10081643611192703, "learning_rate": 1.3468201109758582e-05, "loss": 0.0076, "step": 61820 }, { "grad_norm": 0.16306783258914948, "learning_rate": 1.3454092544051222e-05, "loss": 0.0073, "step": 61830 }, { "grad_norm": 0.09385571628808975, "learning_rate": 1.3439990223039878e-05, "loss": 0.0061, "step": 61840 }, { "grad_norm": 0.1337089091539383, "learning_rate": 1.3425894149134321e-05, "loss": 0.0065, "step": 61850 }, { "grad_norm": 0.09700801968574524, "learning_rate": 1.3411804324743132e-05, "loss": 0.0061, "step": 61860 }, { "grad_norm": 0.11181089282035828, "learning_rate": 1.3397720752273896e-05, "loss": 0.0079, "step": 61870 }, { "grad_norm": 0.1579919308423996, "learning_rate": 1.3383643434133108e-05, "loss": 0.0065, "step": 61880 }, { "grad_norm": 0.17525658011436462, "learning_rate": 1.3369572372726196e-05, "loss": 0.0097, "step": 61890 }, { "grad_norm": 0.174546480178833, "learning_rate": 1.3355507570457525e-05, "loss": 0.0085, "step": 61900 }, { "grad_norm": 0.16972020268440247, "learning_rate": 1.334144902973038e-05, "loss": 0.0097, "step": 61910 }, { "grad_norm": 0.14986161887645721, "learning_rate": 1.3327396752946997e-05, "loss": 0.0079, "step": 61920 }, { "grad_norm": 0.11885512620210648, "learning_rate": 1.3313350742508496e-05, "loss": 0.0064, "step": 61930 }, { "grad_norm": 0.14300672709941864, "learning_rate": 1.3299311000814973e-05, "loss": 0.006, "step": 61940 }, { "grad_norm": 0.1920953392982483, "learning_rate": 1.3285277530265432e-05, "loss": 0.0076, "step": 61950 }, { "grad_norm": 0.1540173441171646, "learning_rate": 1.3271250333257813e-05, "loss": 0.0074, "step": 61960 }, { "grad_norm": 0.14888642728328705, "learning_rate": 1.3257229412188977e-05, "loss": 0.007, "step": 61970 }, { "grad_norm": 0.15510571002960205, "learning_rate": 1.3243214769454726e-05, "loss": 0.0077, "step": 61980 }, { "grad_norm": 0.11526571214199066, "learning_rate": 1.3229206407449751e-05, "loss": 0.0075, "step": 61990 }, { "grad_norm": 0.07039511948823929, "learning_rate": 1.3215204328567742e-05, "loss": 0.0057, "step": 62000 }, { "grad_norm": 0.10167764872312546, "learning_rate": 1.3201208535201232e-05, "loss": 0.0065, "step": 62010 }, { "grad_norm": 0.08750161528587341, "learning_rate": 1.3187219029741737e-05, "loss": 0.0064, "step": 62020 }, { "grad_norm": 0.10405953228473663, "learning_rate": 1.3173235814579693e-05, "loss": 0.0079, "step": 62030 }, { "grad_norm": 0.12394294887781143, "learning_rate": 1.3159258892104398e-05, "loss": 0.0059, "step": 62040 }, { "grad_norm": 0.1193172037601471, "learning_rate": 1.3145288264704198e-05, "loss": 0.0077, "step": 62050 }, { "grad_norm": 0.17307716608047485, "learning_rate": 1.3131323934766237e-05, "loss": 0.0068, "step": 62060 }, { "grad_norm": 0.09401040524244308, "learning_rate": 1.311736590467666e-05, "loss": 0.0067, "step": 62070 }, { "grad_norm": 0.12229176610708237, "learning_rate": 1.3103414176820505e-05, "loss": 0.0074, "step": 62080 }, { "grad_norm": 0.07231966406106949, "learning_rate": 1.308946875358174e-05, "loss": 0.0074, "step": 62090 }, { "grad_norm": 0.1235133484005928, "learning_rate": 1.3075529637343276e-05, "loss": 0.0063, "step": 62100 }, { "grad_norm": 0.1305513232946396, "learning_rate": 1.3061596830486883e-05, "loss": 0.0086, "step": 62110 }, { "grad_norm": 0.18567006289958954, "learning_rate": 1.3047670335393353e-05, "loss": 0.0086, "step": 62120 }, { "grad_norm": 0.11872832477092743, "learning_rate": 1.3033750154442298e-05, "loss": 0.0054, "step": 62130 }, { "grad_norm": 0.10100843012332916, "learning_rate": 1.3019836290012316e-05, "loss": 0.0062, "step": 62140 }, { "grad_norm": 0.13773997128009796, "learning_rate": 1.30059287444809e-05, "loss": 0.0084, "step": 62150 }, { "grad_norm": 0.13311034440994263, "learning_rate": 1.299202752022447e-05, "loss": 0.0068, "step": 62160 }, { "grad_norm": 0.13662366569042206, "learning_rate": 1.2978132619618371e-05, "loss": 0.0085, "step": 62170 }, { "grad_norm": 0.10871808230876923, "learning_rate": 1.2964244045036866e-05, "loss": 0.0066, "step": 62180 }, { "grad_norm": 0.09896518290042877, "learning_rate": 1.295036179885311e-05, "loss": 0.0068, "step": 62190 }, { "grad_norm": 0.09040534496307373, "learning_rate": 1.293648588343922e-05, "loss": 0.0074, "step": 62200 }, { "grad_norm": 0.11283554881811142, "learning_rate": 1.2922616301166196e-05, "loss": 0.0059, "step": 62210 }, { "grad_norm": 0.1055799126625061, "learning_rate": 1.2908753054403976e-05, "loss": 0.0062, "step": 62220 }, { "grad_norm": 0.11097899824380875, "learning_rate": 1.2894896145521429e-05, "loss": 0.0091, "step": 62230 }, { "grad_norm": 0.14707481861114502, "learning_rate": 1.2881045576886275e-05, "loss": 0.0065, "step": 62240 }, { "grad_norm": 0.15438847243785858, "learning_rate": 1.2867201350865254e-05, "loss": 0.0064, "step": 62250 }, { "grad_norm": 0.10821251571178436, "learning_rate": 1.2853363469823914e-05, "loss": 0.005, "step": 62260 }, { "grad_norm": 0.12689967453479767, "learning_rate": 1.283953193612682e-05, "loss": 0.0056, "step": 62270 }, { "grad_norm": 0.10008242726325989, "learning_rate": 1.2825706752137372e-05, "loss": 0.0076, "step": 62280 }, { "grad_norm": 0.13348349928855896, "learning_rate": 1.2811887920217896e-05, "loss": 0.0082, "step": 62290 }, { "grad_norm": 0.07708575576543808, "learning_rate": 1.2798075442729707e-05, "loss": 0.0077, "step": 62300 }, { "grad_norm": 0.07877866923809052, "learning_rate": 1.2784269322032922e-05, "loss": 0.0066, "step": 62310 }, { "grad_norm": 0.15006208419799805, "learning_rate": 1.2770469560486686e-05, "loss": 0.0072, "step": 62320 }, { "grad_norm": 0.1045626625418663, "learning_rate": 1.2756676160448956e-05, "loss": 0.0085, "step": 62330 }, { "grad_norm": 0.1367652267217636, "learning_rate": 1.2742889124276663e-05, "loss": 0.0063, "step": 62340 }, { "grad_norm": 0.13651931285858154, "learning_rate": 1.2729108454325639e-05, "loss": 0.0095, "step": 62350 }, { "grad_norm": 0.11779078841209412, "learning_rate": 1.2715334152950614e-05, "loss": 0.0068, "step": 62360 }, { "grad_norm": 0.1270628273487091, "learning_rate": 1.2701566222505246e-05, "loss": 0.0094, "step": 62370 }, { "grad_norm": 0.153047114610672, "learning_rate": 1.2687804665342107e-05, "loss": 0.0089, "step": 62380 }, { "grad_norm": 0.14184828102588654, "learning_rate": 1.267404948381265e-05, "loss": 0.0079, "step": 62390 }, { "grad_norm": 0.12412340939044952, "learning_rate": 1.2660300680267267e-05, "loss": 0.006, "step": 62400 }, { "grad_norm": 0.17721149325370789, "learning_rate": 1.2646558257055257e-05, "loss": 0.0066, "step": 62410 }, { "grad_norm": 0.10640458762645721, "learning_rate": 1.263282221652482e-05, "loss": 0.0051, "step": 62420 }, { "grad_norm": 0.20363551378250122, "learning_rate": 1.2619092561023088e-05, "loss": 0.0096, "step": 62430 }, { "grad_norm": 0.15108011662960052, "learning_rate": 1.2605369292896036e-05, "loss": 0.0057, "step": 62440 }, { "grad_norm": 0.15031929314136505, "learning_rate": 1.2591652414488658e-05, "loss": 0.0071, "step": 62450 }, { "grad_norm": 0.213747039437294, "learning_rate": 1.2577941928144732e-05, "loss": 0.0076, "step": 62460 }, { "grad_norm": 0.10105389356613159, "learning_rate": 1.256423783620706e-05, "loss": 0.0065, "step": 62470 }, { "grad_norm": 0.12138763815164566, "learning_rate": 1.2550540141017264e-05, "loss": 0.0065, "step": 62480 }, { "grad_norm": 0.1469411849975586, "learning_rate": 1.253684884491591e-05, "loss": 0.0063, "step": 62490 }, { "grad_norm": 0.11461368203163147, "learning_rate": 1.2523163950242483e-05, "loss": 0.0088, "step": 62500 }, { "grad_norm": 0.11296316236257553, "learning_rate": 1.2509485459335313e-05, "loss": 0.0057, "step": 62510 }, { "grad_norm": 0.08008943498134613, "learning_rate": 1.2495813374531739e-05, "loss": 0.0052, "step": 62520 }, { "grad_norm": 0.1323947310447693, "learning_rate": 1.2482147698167907e-05, "loss": 0.0083, "step": 62530 }, { "grad_norm": 0.12265634536743164, "learning_rate": 1.2468488432578911e-05, "loss": 0.0075, "step": 62540 }, { "grad_norm": 0.12839171290397644, "learning_rate": 1.2454835580098761e-05, "loss": 0.0075, "step": 62550 }, { "grad_norm": 0.14963403344154358, "learning_rate": 1.2441189143060338e-05, "loss": 0.0088, "step": 62560 }, { "grad_norm": 0.09619087725877762, "learning_rate": 1.242754912379548e-05, "loss": 0.0067, "step": 62570 }, { "grad_norm": 0.11188222467899323, "learning_rate": 1.2413915524634844e-05, "loss": 0.0079, "step": 62580 }, { "grad_norm": 0.08326747268438339, "learning_rate": 1.2400288347908073e-05, "loss": 0.0042, "step": 62590 }, { "grad_norm": 0.09523298591375351, "learning_rate": 1.2386667595943663e-05, "loss": 0.0053, "step": 62600 }, { "grad_norm": 0.20784366130828857, "learning_rate": 1.2373053271069035e-05, "loss": 0.0074, "step": 62610 }, { "grad_norm": 0.07381901890039444, "learning_rate": 1.2359445375610501e-05, "loss": 0.0064, "step": 62620 }, { "grad_norm": 0.0896824449300766, "learning_rate": 1.2345843911893301e-05, "loss": 0.0065, "step": 62630 }, { "grad_norm": 0.16097675263881683, "learning_rate": 1.2332248882241498e-05, "loss": 0.0066, "step": 62640 }, { "grad_norm": 0.14072546362876892, "learning_rate": 1.2318660288978178e-05, "loss": 0.0078, "step": 62650 }, { "grad_norm": 0.09229101240634918, "learning_rate": 1.2305078134425213e-05, "loss": 0.0056, "step": 62660 }, { "grad_norm": 0.09698309004306793, "learning_rate": 1.2291502420903434e-05, "loss": 0.0073, "step": 62670 }, { "grad_norm": 0.076294906437397, "learning_rate": 1.2277933150732567e-05, "loss": 0.0075, "step": 62680 }, { "grad_norm": 0.2605094909667969, "learning_rate": 1.2264370326231216e-05, "loss": 0.0055, "step": 62690 }, { "grad_norm": 0.18567171692848206, "learning_rate": 1.2250813949716927e-05, "loss": 0.0067, "step": 62700 }, { "grad_norm": 0.09312427788972855, "learning_rate": 1.2237264023506063e-05, "loss": 0.0074, "step": 62710 }, { "grad_norm": 0.0681115910410881, "learning_rate": 1.2223720549913987e-05, "loss": 0.007, "step": 62720 }, { "grad_norm": 0.16010014712810516, "learning_rate": 1.221018353125487e-05, "loss": 0.0084, "step": 62730 }, { "grad_norm": 0.1549854427576065, "learning_rate": 1.2196652969841837e-05, "loss": 0.0069, "step": 62740 }, { "grad_norm": 0.14324165880680084, "learning_rate": 1.2183128867986904e-05, "loss": 0.0046, "step": 62750 }, { "grad_norm": 0.08210809528827667, "learning_rate": 1.2169611228000927e-05, "loss": 0.006, "step": 62760 }, { "grad_norm": 0.07535839080810547, "learning_rate": 1.2156100052193752e-05, "loss": 0.0073, "step": 62770 }, { "grad_norm": 0.10426604747772217, "learning_rate": 1.214259534287403e-05, "loss": 0.0068, "step": 62780 }, { "grad_norm": 0.18235571682453156, "learning_rate": 1.2129097102349363e-05, "loss": 0.0066, "step": 62790 }, { "grad_norm": 0.09801461547613144, "learning_rate": 1.2115605332926227e-05, "loss": 0.007, "step": 62800 }, { "grad_norm": 0.08344347029924393, "learning_rate": 1.2102120036910003e-05, "loss": 0.008, "step": 62810 }, { "grad_norm": 0.12819647789001465, "learning_rate": 1.2088641216604956e-05, "loss": 0.0097, "step": 62820 }, { "grad_norm": 0.10718166083097458, "learning_rate": 1.2075168874314264e-05, "loss": 0.0051, "step": 62830 }, { "grad_norm": 0.13153213262557983, "learning_rate": 1.2061703012339942e-05, "loss": 0.0065, "step": 62840 }, { "grad_norm": 0.11770479381084442, "learning_rate": 1.204824363298297e-05, "loss": 0.0062, "step": 62850 }, { "grad_norm": 0.10761761665344238, "learning_rate": 1.2034790738543173e-05, "loss": 0.0055, "step": 62860 }, { "grad_norm": 0.12002243846654892, "learning_rate": 1.2021344331319284e-05, "loss": 0.0063, "step": 62870 }, { "grad_norm": 0.12935325503349304, "learning_rate": 1.200790441360894e-05, "loss": 0.0078, "step": 62880 }, { "grad_norm": 0.12042434513568878, "learning_rate": 1.199447098770864e-05, "loss": 0.0067, "step": 62890 }, { "grad_norm": 0.09458949416875839, "learning_rate": 1.198104405591381e-05, "loss": 0.0061, "step": 62900 }, { "grad_norm": 0.13359826803207397, "learning_rate": 1.1967623620518697e-05, "loss": 0.0046, "step": 62910 }, { "grad_norm": 0.15893155336380005, "learning_rate": 1.1954209683816554e-05, "loss": 0.0061, "step": 62920 }, { "grad_norm": 0.15342268347740173, "learning_rate": 1.1940802248099402e-05, "loss": 0.0072, "step": 62930 }, { "grad_norm": 0.15359224379062653, "learning_rate": 1.1927401315658232e-05, "loss": 0.0086, "step": 62940 }, { "grad_norm": 0.15065230429172516, "learning_rate": 1.1914006888782898e-05, "loss": 0.0057, "step": 62950 }, { "grad_norm": 0.10137335956096649, "learning_rate": 1.1900618969762106e-05, "loss": 0.0061, "step": 62960 }, { "grad_norm": 0.13054867088794708, "learning_rate": 1.188723756088354e-05, "loss": 0.0071, "step": 62970 }, { "grad_norm": 0.1541265845298767, "learning_rate": 1.1873862664433672e-05, "loss": 0.0063, "step": 62980 }, { "grad_norm": 0.0762268528342247, "learning_rate": 1.1860494282697927e-05, "loss": 0.0054, "step": 62990 }, { "grad_norm": 0.0966593325138092, "learning_rate": 1.1847132417960588e-05, "loss": 0.0057, "step": 63000 }, { "grad_norm": 0.11610014736652374, "learning_rate": 1.1833777072504832e-05, "loss": 0.0049, "step": 63010 }, { "grad_norm": 0.15053126215934753, "learning_rate": 1.1820428248612731e-05, "loss": 0.0069, "step": 63020 }, { "grad_norm": 0.11357849836349487, "learning_rate": 1.1807085948565245e-05, "loss": 0.0058, "step": 63030 }, { "grad_norm": 0.13450618088245392, "learning_rate": 1.1793750174642172e-05, "loss": 0.008, "step": 63040 }, { "grad_norm": 0.13846121728420258, "learning_rate": 1.1780420929122254e-05, "loss": 0.0066, "step": 63050 }, { "grad_norm": 0.14648690819740295, "learning_rate": 1.176709821428309e-05, "loss": 0.0063, "step": 63060 }, { "grad_norm": 0.16742224991321564, "learning_rate": 1.1753782032401173e-05, "loss": 0.0061, "step": 63070 }, { "grad_norm": 0.10536471754312515, "learning_rate": 1.1740472385751866e-05, "loss": 0.0055, "step": 63080 }, { "grad_norm": 0.09299509972333908, "learning_rate": 1.172716927660943e-05, "loss": 0.0093, "step": 63090 }, { "grad_norm": 0.10850745439529419, "learning_rate": 1.1713872707247015e-05, "loss": 0.0067, "step": 63100 }, { "grad_norm": 0.1281987726688385, "learning_rate": 1.1700582679936595e-05, "loss": 0.0066, "step": 63110 }, { "grad_norm": 0.1055404394865036, "learning_rate": 1.1687299196949136e-05, "loss": 0.005, "step": 63120 }, { "grad_norm": 0.129064679145813, "learning_rate": 1.1674022260554374e-05, "loss": 0.0062, "step": 63130 }, { "grad_norm": 0.1444539874792099, "learning_rate": 1.1660751873020987e-05, "loss": 0.0069, "step": 63140 }, { "grad_norm": 0.15660201013088226, "learning_rate": 1.1647488036616538e-05, "loss": 0.0063, "step": 63150 }, { "grad_norm": 0.11388815939426422, "learning_rate": 1.1634230753607417e-05, "loss": 0.0065, "step": 63160 }, { "grad_norm": 0.17407414317131042, "learning_rate": 1.1620980026258982e-05, "loss": 0.0083, "step": 63170 }, { "grad_norm": 0.1276513934135437, "learning_rate": 1.1607735856835373e-05, "loss": 0.0071, "step": 63180 }, { "grad_norm": 0.16230814158916473, "learning_rate": 1.1594498247599677e-05, "loss": 0.0077, "step": 63190 }, { "grad_norm": 0.14057597517967224, "learning_rate": 1.158126720081384e-05, "loss": 0.0058, "step": 63200 }, { "grad_norm": 0.11481940001249313, "learning_rate": 1.156804271873868e-05, "loss": 0.0085, "step": 63210 }, { "grad_norm": 0.1560056209564209, "learning_rate": 1.1554824803633924e-05, "loss": 0.0062, "step": 63220 }, { "grad_norm": 0.12476728856563568, "learning_rate": 1.154161345775811e-05, "loss": 0.0063, "step": 63230 }, { "grad_norm": 0.1236332356929779, "learning_rate": 1.1528408683368724e-05, "loss": 0.0063, "step": 63240 }, { "grad_norm": 0.10716178268194199, "learning_rate": 1.1515210482722088e-05, "loss": 0.0062, "step": 63250 }, { "grad_norm": 0.1597336381673813, "learning_rate": 1.150201885807342e-05, "loss": 0.0082, "step": 63260 }, { "grad_norm": 0.1098288744688034, "learning_rate": 1.1488833811676807e-05, "loss": 0.006, "step": 63270 }, { "grad_norm": 0.15507900714874268, "learning_rate": 1.1475655345785213e-05, "loss": 0.006, "step": 63280 }, { "grad_norm": 0.12273629754781723, "learning_rate": 1.1462483462650481e-05, "loss": 0.0072, "step": 63290 }, { "grad_norm": 0.11334136128425598, "learning_rate": 1.1449318164523331e-05, "loss": 0.006, "step": 63300 }, { "grad_norm": 0.10965801775455475, "learning_rate": 1.1436159453653334e-05, "loss": 0.007, "step": 63310 }, { "grad_norm": 0.14077146351337433, "learning_rate": 1.1423007332288955e-05, "loss": 0.0062, "step": 63320 }, { "grad_norm": 0.1174660250544548, "learning_rate": 1.1409861802677546e-05, "loss": 0.0062, "step": 63330 }, { "grad_norm": 0.1317596435546875, "learning_rate": 1.1396722867065313e-05, "loss": 0.0074, "step": 63340 }, { "grad_norm": 0.12993642687797546, "learning_rate": 1.1383590527697352e-05, "loss": 0.0068, "step": 63350 }, { "grad_norm": 0.1401626020669937, "learning_rate": 1.137046478681758e-05, "loss": 0.0053, "step": 63360 }, { "grad_norm": 0.09637603163719177, "learning_rate": 1.1357345646668888e-05, "loss": 0.0062, "step": 63370 }, { "grad_norm": 0.11281886696815491, "learning_rate": 1.1344233109492924e-05, "loss": 0.0058, "step": 63380 }, { "grad_norm": 0.07617239654064178, "learning_rate": 1.1331127177530292e-05, "loss": 0.0058, "step": 63390 }, { "grad_norm": 0.15517142415046692, "learning_rate": 1.1318027853020441e-05, "loss": 0.0065, "step": 63400 }, { "grad_norm": 0.13400313258171082, "learning_rate": 1.1304935138201645e-05, "loss": 0.0059, "step": 63410 }, { "grad_norm": 0.07470029592514038, "learning_rate": 1.1291849035311153e-05, "loss": 0.0062, "step": 63420 }, { "grad_norm": 0.13532298803329468, "learning_rate": 1.1278769546584972e-05, "loss": 0.0064, "step": 63430 }, { "grad_norm": 0.12970499694347382, "learning_rate": 1.1265696674258052e-05, "loss": 0.0056, "step": 63440 }, { "grad_norm": 0.13640545308589935, "learning_rate": 1.1252630420564186e-05, "loss": 0.006, "step": 63450 }, { "grad_norm": 0.10588134080171585, "learning_rate": 1.1239570787736036e-05, "loss": 0.0067, "step": 63460 }, { "grad_norm": 0.13575154542922974, "learning_rate": 1.1226517778005135e-05, "loss": 0.006, "step": 63470 }, { "grad_norm": 0.12147995829582214, "learning_rate": 1.1213471393601893e-05, "loss": 0.0107, "step": 63480 }, { "grad_norm": 0.13878387212753296, "learning_rate": 1.1200431636755587e-05, "loss": 0.0064, "step": 63490 }, { "grad_norm": 0.17076294124126434, "learning_rate": 1.1187398509694336e-05, "loss": 0.0078, "step": 63500 }, { "grad_norm": 0.07571649551391602, "learning_rate": 1.1174372014645146e-05, "loss": 0.0052, "step": 63510 }, { "grad_norm": 0.1159290075302124, "learning_rate": 1.1161352153833899e-05, "loss": 0.0056, "step": 63520 }, { "grad_norm": 0.1123928353190422, "learning_rate": 1.1148338929485325e-05, "loss": 0.006, "step": 63530 }, { "grad_norm": 0.08490172773599625, "learning_rate": 1.113533234382304e-05, "loss": 0.0062, "step": 63540 }, { "grad_norm": 0.08374852687120438, "learning_rate": 1.1122332399069513e-05, "loss": 0.0077, "step": 63550 }, { "grad_norm": 0.08773158490657806, "learning_rate": 1.1109339097446047e-05, "loss": 0.0079, "step": 63560 }, { "grad_norm": 0.08713136613368988, "learning_rate": 1.1096352441172897e-05, "loss": 0.0072, "step": 63570 }, { "grad_norm": 0.18486231565475464, "learning_rate": 1.1083372432469086e-05, "loss": 0.0078, "step": 63580 }, { "grad_norm": 0.09968196600675583, "learning_rate": 1.107039907355255e-05, "loss": 0.0063, "step": 63590 }, { "grad_norm": 0.14356358349323273, "learning_rate": 1.1057432366640103e-05, "loss": 0.0069, "step": 63600 }, { "grad_norm": 0.13216851651668549, "learning_rate": 1.1044472313947352e-05, "loss": 0.006, "step": 63610 }, { "grad_norm": 0.1011708676815033, "learning_rate": 1.1031518917688877e-05, "loss": 0.0072, "step": 63620 }, { "grad_norm": 0.15219749510288239, "learning_rate": 1.1018572180078007e-05, "loss": 0.007, "step": 63630 }, { "grad_norm": 0.12702395021915436, "learning_rate": 1.1005632103327018e-05, "loss": 0.0053, "step": 63640 }, { "grad_norm": 0.08541528880596161, "learning_rate": 1.0992698689646996e-05, "loss": 0.0103, "step": 63650 }, { "grad_norm": 0.07577216625213623, "learning_rate": 1.0979771941247919e-05, "loss": 0.0049, "step": 63660 }, { "grad_norm": 0.12595029175281525, "learning_rate": 1.0966851860338611e-05, "loss": 0.0066, "step": 63670 }, { "grad_norm": 0.1316825896501541, "learning_rate": 1.0953938449126766e-05, "loss": 0.0077, "step": 63680 }, { "grad_norm": 0.16563871502876282, "learning_rate": 1.0941031709818933e-05, "loss": 0.0062, "step": 63690 }, { "grad_norm": 0.13447147607803345, "learning_rate": 1.0928131644620509e-05, "loss": 0.008, "step": 63700 }, { "grad_norm": 0.08105767518281937, "learning_rate": 1.0915238255735766e-05, "loss": 0.0076, "step": 63710 }, { "grad_norm": 0.11888866871595383, "learning_rate": 1.0902351545367833e-05, "loss": 0.0083, "step": 63720 }, { "grad_norm": 0.1276337057352066, "learning_rate": 1.0889471515718702e-05, "loss": 0.0054, "step": 63730 }, { "grad_norm": 0.1272878646850586, "learning_rate": 1.087659816898921e-05, "loss": 0.0064, "step": 63740 }, { "grad_norm": 0.09362837672233582, "learning_rate": 1.0863731507379082e-05, "loss": 0.0061, "step": 63750 }, { "grad_norm": 0.11589616537094116, "learning_rate": 1.0850871533086827e-05, "loss": 0.0059, "step": 63760 }, { "grad_norm": 0.09501906484365463, "learning_rate": 1.0838018248309927e-05, "loss": 0.0058, "step": 63770 }, { "grad_norm": 0.07217013090848923, "learning_rate": 1.0825171655244615e-05, "loss": 0.0064, "step": 63780 }, { "grad_norm": 0.06781160086393356, "learning_rate": 1.0812331756086025e-05, "loss": 0.006, "step": 63790 }, { "grad_norm": 0.10362133383750916, "learning_rate": 1.079949855302817e-05, "loss": 0.0073, "step": 63800 }, { "grad_norm": 0.10774409770965576, "learning_rate": 1.0786672048263852e-05, "loss": 0.007, "step": 63810 }, { "grad_norm": 0.1166471540927887, "learning_rate": 1.0773852243984817e-05, "loss": 0.0061, "step": 63820 }, { "grad_norm": 0.17775078117847443, "learning_rate": 1.0761039142381586e-05, "loss": 0.0066, "step": 63830 }, { "grad_norm": 0.0822681114077568, "learning_rate": 1.0748232745643577e-05, "loss": 0.0042, "step": 63840 }, { "grad_norm": 0.10066165030002594, "learning_rate": 1.0735433055959055e-05, "loss": 0.0057, "step": 63850 }, { "grad_norm": 0.082685686647892, "learning_rate": 1.0722640075515133e-05, "loss": 0.005, "step": 63860 }, { "grad_norm": 0.08983573317527771, "learning_rate": 1.0709853806497795e-05, "loss": 0.0089, "step": 63870 }, { "grad_norm": 0.2836025059223175, "learning_rate": 1.0697074251091831e-05, "loss": 0.0066, "step": 63880 }, { "grad_norm": 0.1721571981906891, "learning_rate": 1.0684301411480962e-05, "loss": 0.0085, "step": 63890 }, { "grad_norm": 0.09728005528450012, "learning_rate": 1.067153528984769e-05, "loss": 0.0078, "step": 63900 }, { "grad_norm": 0.10513421148061752, "learning_rate": 1.0658775888373395e-05, "loss": 0.0052, "step": 63910 }, { "grad_norm": 0.088872529566288, "learning_rate": 1.0646023209238314e-05, "loss": 0.0061, "step": 63920 }, { "grad_norm": 0.11260867118835449, "learning_rate": 1.0633277254621537e-05, "loss": 0.0061, "step": 63930 }, { "grad_norm": 0.11873926222324371, "learning_rate": 1.0620538026700994e-05, "loss": 0.0082, "step": 63940 }, { "grad_norm": 0.09842941164970398, "learning_rate": 1.0607805527653486e-05, "loss": 0.0062, "step": 63950 }, { "grad_norm": 0.10851970314979553, "learning_rate": 1.059507975965462e-05, "loss": 0.0066, "step": 63960 }, { "grad_norm": 0.18701498210430145, "learning_rate": 1.0582360724878898e-05, "loss": 0.0071, "step": 63970 }, { "grad_norm": 0.07648220658302307, "learning_rate": 1.0569648425499651e-05, "loss": 0.0059, "step": 63980 }, { "grad_norm": 0.11300641298294067, "learning_rate": 1.0556942863689063e-05, "loss": 0.0062, "step": 63990 }, { "grad_norm": 0.18234218657016754, "learning_rate": 1.054424404161819e-05, "loss": 0.0084, "step": 64000 }, { "grad_norm": 0.0640457272529602, "learning_rate": 1.053155196145686e-05, "loss": 0.0073, "step": 64010 }, { "grad_norm": 0.11167380213737488, "learning_rate": 1.0518866625373863e-05, "loss": 0.0087, "step": 64020 }, { "grad_norm": 0.4566241502761841, "learning_rate": 1.0506188035536735e-05, "loss": 0.0082, "step": 64030 }, { "grad_norm": 0.17995232343673706, "learning_rate": 1.0493516194111919e-05, "loss": 0.0088, "step": 64040 }, { "grad_norm": 0.15897336602210999, "learning_rate": 1.0480851103264688e-05, "loss": 0.0075, "step": 64050 }, { "grad_norm": 0.16588181257247925, "learning_rate": 1.046819276515913e-05, "loss": 0.0081, "step": 64060 }, { "grad_norm": 0.09686122089624405, "learning_rate": 1.0455541181958256e-05, "loss": 0.0064, "step": 64070 }, { "grad_norm": 0.10547114163637161, "learning_rate": 1.0442896355823822e-05, "loss": 0.0089, "step": 64080 }, { "grad_norm": 0.12790940701961517, "learning_rate": 1.0430258288916539e-05, "loss": 0.0072, "step": 64090 }, { "grad_norm": 0.19931913912296295, "learning_rate": 1.0417626983395868e-05, "loss": 0.0099, "step": 64100 }, { "grad_norm": 0.10345298051834106, "learning_rate": 1.0405002441420165e-05, "loss": 0.0056, "step": 64110 }, { "grad_norm": 0.10838305205106735, "learning_rate": 1.039238466514662e-05, "loss": 0.0052, "step": 64120 }, { "grad_norm": 0.12543939054012299, "learning_rate": 1.0379773656731262e-05, "loss": 0.0063, "step": 64130 }, { "grad_norm": 0.09520011395215988, "learning_rate": 1.0367169418328986e-05, "loss": 0.0064, "step": 64140 }, { "grad_norm": 0.07589061558246613, "learning_rate": 1.0354571952093484e-05, "loss": 0.0061, "step": 64150 }, { "grad_norm": 0.1300322264432907, "learning_rate": 1.034198126017733e-05, "loss": 0.0086, "step": 64160 }, { "grad_norm": 0.09762367606163025, "learning_rate": 1.0329397344731928e-05, "loss": 0.0074, "step": 64170 }, { "grad_norm": 0.14421986043453217, "learning_rate": 1.0316820207907524e-05, "loss": 0.0056, "step": 64180 }, { "grad_norm": 0.11299893260002136, "learning_rate": 1.0304249851853215e-05, "loss": 0.0058, "step": 64190 }, { "grad_norm": 0.14166133105754852, "learning_rate": 1.0291686278716933e-05, "loss": 0.0066, "step": 64200 }, { "grad_norm": 0.11547807604074478, "learning_rate": 1.0279129490645418e-05, "loss": 0.0058, "step": 64210 }, { "grad_norm": 0.16918271780014038, "learning_rate": 1.0266579489784328e-05, "loss": 0.0086, "step": 64220 }, { "grad_norm": 0.13078145682811737, "learning_rate": 1.0254036278278084e-05, "loss": 0.0072, "step": 64230 }, { "grad_norm": 0.17711171507835388, "learning_rate": 1.0241499858269982e-05, "loss": 0.0062, "step": 64240 }, { "grad_norm": 0.13467882573604584, "learning_rate": 1.0228970231902169e-05, "loss": 0.0066, "step": 64250 }, { "grad_norm": 0.10214739292860031, "learning_rate": 1.0216447401315582e-05, "loss": 0.0063, "step": 64260 }, { "grad_norm": 0.09940043836832047, "learning_rate": 1.020393136865007e-05, "loss": 0.0054, "step": 64270 }, { "grad_norm": 0.13022321462631226, "learning_rate": 1.0191422136044242e-05, "loss": 0.0058, "step": 64280 }, { "grad_norm": 0.13213273882865906, "learning_rate": 1.017891970563563e-05, "loss": 0.0061, "step": 64290 }, { "grad_norm": 0.10683199018239975, "learning_rate": 1.0166424079560516e-05, "loss": 0.0046, "step": 64300 }, { "grad_norm": 0.14494507014751434, "learning_rate": 1.0153935259954078e-05, "loss": 0.0056, "step": 64310 }, { "grad_norm": 0.09092829376459122, "learning_rate": 1.0141453248950311e-05, "loss": 0.0097, "step": 64320 }, { "grad_norm": 0.06929909437894821, "learning_rate": 1.0128978048682054e-05, "loss": 0.0049, "step": 64330 }, { "grad_norm": 0.13498921692371368, "learning_rate": 1.0116509661280982e-05, "loss": 0.0062, "step": 64340 }, { "grad_norm": 0.14856211841106415, "learning_rate": 1.0104048088877576e-05, "loss": 0.0069, "step": 64350 }, { "grad_norm": 0.22047144174575806, "learning_rate": 1.0091593333601201e-05, "loss": 0.0049, "step": 64360 }, { "grad_norm": 0.10132738947868347, "learning_rate": 1.0079145397580031e-05, "loss": 0.0055, "step": 64370 }, { "grad_norm": 0.10151782631874084, "learning_rate": 1.006670428294107e-05, "loss": 0.0067, "step": 64380 }, { "grad_norm": 0.08216679096221924, "learning_rate": 1.0054269991810166e-05, "loss": 0.0052, "step": 64390 }, { "grad_norm": 0.11838483810424805, "learning_rate": 1.0041842526312024e-05, "loss": 0.0071, "step": 64400 }, { "grad_norm": 0.154100239276886, "learning_rate": 1.0029421888570101e-05, "loss": 0.0078, "step": 64410 }, { "grad_norm": 0.0812615230679512, "learning_rate": 1.0017008080706813e-05, "loss": 0.0062, "step": 64420 }, { "grad_norm": 0.13939893245697021, "learning_rate": 1.0004601104843287e-05, "loss": 0.0078, "step": 64430 }, { "grad_norm": 0.12131763994693756, "learning_rate": 9.992200963099562e-06, "loss": 0.0063, "step": 64440 }, { "grad_norm": 0.14578932523727417, "learning_rate": 9.979807657594486e-06, "loss": 0.0071, "step": 64450 }, { "grad_norm": 0.17971894145011902, "learning_rate": 9.967421190445703e-06, "loss": 0.0061, "step": 64460 }, { "grad_norm": 0.10623050481081009, "learning_rate": 9.955041563769769e-06, "loss": 0.0075, "step": 64470 }, { "grad_norm": 0.11197856813669205, "learning_rate": 9.942668779681974e-06, "loss": 0.0062, "step": 64480 }, { "grad_norm": 0.07479029893875122, "learning_rate": 9.930302840296541e-06, "loss": 0.0063, "step": 64490 }, { "grad_norm": 0.1181001365184784, "learning_rate": 9.917943747726426e-06, "loss": 0.0057, "step": 64500 }, { "grad_norm": 0.09342893958091736, "learning_rate": 9.905591504083484e-06, "loss": 0.0078, "step": 64510 }, { "grad_norm": 0.1450112760066986, "learning_rate": 9.893246111478382e-06, "loss": 0.0078, "step": 64520 }, { "grad_norm": 0.15724506974220276, "learning_rate": 9.88090757202057e-06, "loss": 0.0067, "step": 64530 }, { "grad_norm": 0.10338714718818665, "learning_rate": 9.868575887818421e-06, "loss": 0.0062, "step": 64540 }, { "grad_norm": 0.10929691046476364, "learning_rate": 9.856251060979044e-06, "loss": 0.0074, "step": 64550 }, { "grad_norm": 0.1448298841714859, "learning_rate": 9.843933093608426e-06, "loss": 0.006, "step": 64560 }, { "grad_norm": 0.1765521913766861, "learning_rate": 9.831621987811368e-06, "loss": 0.0074, "step": 64570 }, { "grad_norm": 0.16622523963451385, "learning_rate": 9.819317745691509e-06, "loss": 0.0086, "step": 64580 }, { "grad_norm": 0.13234184682369232, "learning_rate": 9.8070203693513e-06, "loss": 0.0051, "step": 64590 }, { "grad_norm": 0.09714270383119583, "learning_rate": 9.794729860892048e-06, "loss": 0.0078, "step": 64600 }, { "grad_norm": 0.1839306503534317, "learning_rate": 9.782446222413827e-06, "loss": 0.0111, "step": 64610 }, { "grad_norm": 0.1547512412071228, "learning_rate": 9.770169456015598e-06, "loss": 0.0078, "step": 64620 }, { "grad_norm": 0.12179365009069443, "learning_rate": 9.75789956379512e-06, "loss": 0.0095, "step": 64630 }, { "grad_norm": 0.07909803092479706, "learning_rate": 9.74563654784898e-06, "loss": 0.0064, "step": 64640 }, { "grad_norm": 0.10587349534034729, "learning_rate": 9.733380410272596e-06, "loss": 0.0057, "step": 64650 }, { "grad_norm": 0.07735349237918854, "learning_rate": 9.721131153160207e-06, "loss": 0.0076, "step": 64660 }, { "grad_norm": 0.12489654123783112, "learning_rate": 9.708888778604886e-06, "loss": 0.0065, "step": 64670 }, { "grad_norm": 0.1396065652370453, "learning_rate": 9.696653288698488e-06, "loss": 0.0083, "step": 64680 }, { "grad_norm": 0.12920457124710083, "learning_rate": 9.684424685531763e-06, "loss": 0.0086, "step": 64690 }, { "grad_norm": 0.18902525305747986, "learning_rate": 9.672202971194216e-06, "loss": 0.0068, "step": 64700 }, { "grad_norm": 0.09151861816644669, "learning_rate": 9.659988147774213e-06, "loss": 0.0053, "step": 64710 }, { "grad_norm": 0.09408912807703018, "learning_rate": 9.647780217358942e-06, "loss": 0.0099, "step": 64720 }, { "grad_norm": 0.13340534269809723, "learning_rate": 9.635579182034376e-06, "loss": 0.0073, "step": 64730 }, { "grad_norm": 0.12190432101488113, "learning_rate": 9.623385043885386e-06, "loss": 0.006, "step": 64740 }, { "grad_norm": 0.06589911878108978, "learning_rate": 9.61119780499557e-06, "loss": 0.0059, "step": 64750 }, { "grad_norm": 0.13018949329853058, "learning_rate": 9.599017467447418e-06, "loss": 0.0071, "step": 64760 }, { "grad_norm": 0.07267996668815613, "learning_rate": 9.586844033322206e-06, "loss": 0.0074, "step": 64770 }, { "grad_norm": 0.05875396728515625, "learning_rate": 9.574677504700052e-06, "loss": 0.0055, "step": 64780 }, { "grad_norm": 0.19362035393714905, "learning_rate": 9.562517883659877e-06, "loss": 0.0082, "step": 64790 }, { "grad_norm": 0.11716242134571075, "learning_rate": 9.55036517227944e-06, "loss": 0.0062, "step": 64800 }, { "grad_norm": 0.21710357069969177, "learning_rate": 9.538219372635282e-06, "loss": 0.0093, "step": 64810 }, { "grad_norm": 0.14472649991512299, "learning_rate": 9.526080486802802e-06, "loss": 0.0082, "step": 64820 }, { "grad_norm": 0.11860925704240799, "learning_rate": 9.513948516856203e-06, "loss": 0.0083, "step": 64830 }, { "grad_norm": 0.11994551867246628, "learning_rate": 9.501823464868504e-06, "loss": 0.0074, "step": 64840 }, { "grad_norm": 0.11007997393608093, "learning_rate": 9.489705332911547e-06, "loss": 0.0064, "step": 64850 }, { "grad_norm": 0.1089385524392128, "learning_rate": 9.477594123055994e-06, "loss": 0.0078, "step": 64860 }, { "grad_norm": 0.12180691212415695, "learning_rate": 9.465489837371321e-06, "loss": 0.0066, "step": 64870 }, { "grad_norm": 0.09891953319311142, "learning_rate": 9.453392477925794e-06, "loss": 0.0073, "step": 64880 }, { "grad_norm": 0.0849374309182167, "learning_rate": 9.441302046786566e-06, "loss": 0.0057, "step": 64890 }, { "grad_norm": 0.07713228464126587, "learning_rate": 9.429218546019519e-06, "loss": 0.0055, "step": 64900 }, { "grad_norm": 0.07335074990987778, "learning_rate": 9.41714197768941e-06, "loss": 0.0058, "step": 64910 }, { "grad_norm": 0.12651792168617249, "learning_rate": 9.405072343859805e-06, "loss": 0.0064, "step": 64920 }, { "grad_norm": 0.08732690662145615, "learning_rate": 9.393009646593043e-06, "loss": 0.0061, "step": 64930 }, { "grad_norm": 0.08602283895015717, "learning_rate": 9.38095388795035e-06, "loss": 0.006, "step": 64940 }, { "grad_norm": 0.12084396928548813, "learning_rate": 9.36890506999169e-06, "loss": 0.0061, "step": 64950 }, { "grad_norm": 0.16126343607902527, "learning_rate": 9.356863194775894e-06, "loss": 0.0074, "step": 64960 }, { "grad_norm": 0.07840710878372192, "learning_rate": 9.344828264360583e-06, "loss": 0.0071, "step": 64970 }, { "grad_norm": 0.11091464757919312, "learning_rate": 9.332800280802201e-06, "loss": 0.0067, "step": 64980 }, { "grad_norm": 0.11774643510580063, "learning_rate": 9.32077924615602e-06, "loss": 0.0085, "step": 64990 }, { "grad_norm": 0.1053471639752388, "learning_rate": 9.308765162476063e-06, "loss": 0.0081, "step": 65000 }, { "grad_norm": 0.12583261728286743, "learning_rate": 9.296758031815239e-06, "loss": 0.0095, "step": 65010 }, { "grad_norm": 0.13103817403316498, "learning_rate": 9.284757856225229e-06, "loss": 0.0081, "step": 65020 }, { "grad_norm": 0.12376955151557922, "learning_rate": 9.272764637756538e-06, "loss": 0.0067, "step": 65030 }, { "grad_norm": 0.11008753627538681, "learning_rate": 9.260778378458479e-06, "loss": 0.0076, "step": 65040 }, { "grad_norm": 0.15501052141189575, "learning_rate": 9.248799080379172e-06, "loss": 0.0099, "step": 65050 }, { "grad_norm": 0.15876248478889465, "learning_rate": 9.236826745565558e-06, "loss": 0.0053, "step": 65060 }, { "grad_norm": 0.1297333538532257, "learning_rate": 9.224861376063388e-06, "loss": 0.0064, "step": 65070 }, { "grad_norm": 0.11075621098279953, "learning_rate": 9.212902973917192e-06, "loss": 0.0071, "step": 65080 }, { "grad_norm": 0.0687684565782547, "learning_rate": 9.20095154117035e-06, "loss": 0.0054, "step": 65090 }, { "grad_norm": 0.06735516339540482, "learning_rate": 9.189007079865036e-06, "loss": 0.0074, "step": 65100 }, { "grad_norm": 0.05804868042469025, "learning_rate": 9.177069592042226e-06, "loss": 0.0057, "step": 65110 }, { "grad_norm": 0.15363162755966187, "learning_rate": 9.165139079741724e-06, "loss": 0.0099, "step": 65120 }, { "grad_norm": 0.1322382688522339, "learning_rate": 9.153215545002098e-06, "loss": 0.008, "step": 65130 }, { "grad_norm": 0.08563601970672607, "learning_rate": 9.141298989860798e-06, "loss": 0.0085, "step": 65140 }, { "grad_norm": 0.1301046758890152, "learning_rate": 9.129389416353994e-06, "loss": 0.0081, "step": 65150 }, { "grad_norm": 0.14297638833522797, "learning_rate": 9.11748682651673e-06, "loss": 0.0086, "step": 65160 }, { "grad_norm": 0.13713648915290833, "learning_rate": 9.105591222382837e-06, "loss": 0.008, "step": 65170 }, { "grad_norm": 0.1433352530002594, "learning_rate": 9.093702605984915e-06, "loss": 0.0068, "step": 65180 }, { "grad_norm": 0.12646546959877014, "learning_rate": 9.081820979354455e-06, "loss": 0.0053, "step": 65190 }, { "grad_norm": 0.1378961205482483, "learning_rate": 9.069946344521663e-06, "loss": 0.0049, "step": 65200 }, { "grad_norm": 0.10501887649297714, "learning_rate": 9.058078703515598e-06, "loss": 0.0081, "step": 65210 }, { "grad_norm": 0.10054676234722137, "learning_rate": 9.046218058364125e-06, "loss": 0.0071, "step": 65220 }, { "grad_norm": 0.09256161749362946, "learning_rate": 9.034364411093893e-06, "loss": 0.0054, "step": 65230 }, { "grad_norm": 0.10862752050161362, "learning_rate": 9.022517763730371e-06, "loss": 0.0049, "step": 65240 }, { "grad_norm": 0.08801387250423431, "learning_rate": 9.010678118297827e-06, "loss": 0.0072, "step": 65250 }, { "grad_norm": 0.08920418471097946, "learning_rate": 8.998845476819345e-06, "loss": 0.0058, "step": 65260 }, { "grad_norm": 0.11942123621702194, "learning_rate": 8.987019841316773e-06, "loss": 0.007, "step": 65270 }, { "grad_norm": 0.1019507497549057, "learning_rate": 8.975201213810802e-06, "loss": 0.0056, "step": 65280 }, { "grad_norm": 0.09898205101490021, "learning_rate": 8.963389596320915e-06, "loss": 0.0061, "step": 65290 }, { "grad_norm": 0.10919046401977539, "learning_rate": 8.951584990865391e-06, "loss": 0.0093, "step": 65300 }, { "grad_norm": 0.09933799505233765, "learning_rate": 8.939787399461319e-06, "loss": 0.0065, "step": 65310 }, { "grad_norm": 0.08491770923137665, "learning_rate": 8.927996824124591e-06, "loss": 0.0062, "step": 65320 }, { "grad_norm": 0.10423329472541809, "learning_rate": 8.916213266869854e-06, "loss": 0.0055, "step": 65330 }, { "grad_norm": 0.13405382633209229, "learning_rate": 8.904436729710658e-06, "loss": 0.0065, "step": 65340 }, { "grad_norm": 0.08696304261684418, "learning_rate": 8.892667214659245e-06, "loss": 0.0051, "step": 65350 }, { "grad_norm": 0.16378311812877655, "learning_rate": 8.880904723726713e-06, "loss": 0.0079, "step": 65360 }, { "grad_norm": 0.10473835468292236, "learning_rate": 8.869149258922971e-06, "loss": 0.0066, "step": 65370 }, { "grad_norm": 0.11224114894866943, "learning_rate": 8.857400822256662e-06, "loss": 0.007, "step": 65380 }, { "grad_norm": 0.10894991457462311, "learning_rate": 8.845659415735324e-06, "loss": 0.0069, "step": 65390 }, { "grad_norm": 0.1431736797094345, "learning_rate": 8.83392504136521e-06, "loss": 0.0076, "step": 65400 }, { "grad_norm": 0.07650841772556305, "learning_rate": 8.822197701151407e-06, "loss": 0.0057, "step": 65410 }, { "grad_norm": 0.133589968085289, "learning_rate": 8.810477397097804e-06, "loss": 0.0052, "step": 65420 }, { "grad_norm": 0.07114217430353165, "learning_rate": 8.798764131207077e-06, "loss": 0.0063, "step": 65430 }, { "grad_norm": 0.08529608696699142, "learning_rate": 8.787057905480706e-06, "loss": 0.0044, "step": 65440 }, { "grad_norm": 0.07056637853384018, "learning_rate": 8.775358721918958e-06, "loss": 0.0054, "step": 65450 }, { "grad_norm": 0.09572569280862808, "learning_rate": 8.763666582520923e-06, "loss": 0.0063, "step": 65460 }, { "grad_norm": 0.08074464648962021, "learning_rate": 8.751981489284445e-06, "loss": 0.0074, "step": 65470 }, { "grad_norm": 0.08287356048822403, "learning_rate": 8.740303444206188e-06, "loss": 0.0052, "step": 65480 }, { "grad_norm": 0.06886790692806244, "learning_rate": 8.72863244928162e-06, "loss": 0.006, "step": 65490 }, { "grad_norm": 0.09253671765327454, "learning_rate": 8.716968506504991e-06, "loss": 0.0049, "step": 65500 }, { "grad_norm": 0.10336844623088837, "learning_rate": 8.70531161786935e-06, "loss": 0.0067, "step": 65510 }, { "grad_norm": 0.1510448157787323, "learning_rate": 8.693661785366558e-06, "loss": 0.0067, "step": 65520 }, { "grad_norm": 0.12918975949287415, "learning_rate": 8.682019010987208e-06, "loss": 0.0084, "step": 65530 }, { "grad_norm": 0.1673896163702011, "learning_rate": 8.670383296720786e-06, "loss": 0.0066, "step": 65540 }, { "grad_norm": 0.13391155004501343, "learning_rate": 8.658754644555478e-06, "loss": 0.0079, "step": 65550 }, { "grad_norm": 0.15068583190441132, "learning_rate": 8.647133056478313e-06, "loss": 0.0064, "step": 65560 }, { "grad_norm": 0.14229485392570496, "learning_rate": 8.635518534475123e-06, "loss": 0.0088, "step": 65570 }, { "grad_norm": 0.13037671148777008, "learning_rate": 8.623911080530467e-06, "loss": 0.007, "step": 65580 }, { "grad_norm": 0.12676464021205902, "learning_rate": 8.61231069662779e-06, "loss": 0.0054, "step": 65590 }, { "grad_norm": 0.09526067972183228, "learning_rate": 8.600717384749252e-06, "loss": 0.0052, "step": 65600 }, { "grad_norm": 0.30342841148376465, "learning_rate": 8.58913114687584e-06, "loss": 0.0075, "step": 65610 }, { "grad_norm": 0.11231106519699097, "learning_rate": 8.57755198498732e-06, "loss": 0.0061, "step": 65620 }, { "grad_norm": 0.12755392491817474, "learning_rate": 8.565979901062265e-06, "loss": 0.0067, "step": 65630 }, { "grad_norm": 0.18546177446842194, "learning_rate": 8.554414897078033e-06, "loss": 0.007, "step": 65640 }, { "grad_norm": 0.2092367261648178, "learning_rate": 8.542856975010727e-06, "loss": 0.0052, "step": 65650 }, { "grad_norm": 0.20339429378509521, "learning_rate": 8.531306136835337e-06, "loss": 0.0061, "step": 65660 }, { "grad_norm": 0.13111676275730133, "learning_rate": 8.519762384525548e-06, "loss": 0.0065, "step": 65670 }, { "grad_norm": 0.14274905622005463, "learning_rate": 8.508225720053875e-06, "loss": 0.0072, "step": 65680 }, { "grad_norm": 0.09146128594875336, "learning_rate": 8.496696145391625e-06, "loss": 0.0053, "step": 65690 }, { "grad_norm": 0.19813701510429382, "learning_rate": 8.485173662508889e-06, "loss": 0.0096, "step": 65700 }, { "grad_norm": 0.0745663270354271, "learning_rate": 8.473658273374536e-06, "loss": 0.0044, "step": 65710 }, { "grad_norm": 0.19380579888820648, "learning_rate": 8.462149979956253e-06, "loss": 0.0073, "step": 65720 }, { "grad_norm": 0.286114901304245, "learning_rate": 8.450648784220461e-06, "loss": 0.0065, "step": 65730 }, { "grad_norm": 0.11142926663160324, "learning_rate": 8.439154688132417e-06, "loss": 0.0064, "step": 65740 }, { "grad_norm": 0.15541952848434448, "learning_rate": 8.427667693656143e-06, "loss": 0.0069, "step": 65750 }, { "grad_norm": 0.14933302998542786, "learning_rate": 8.416187802754454e-06, "loss": 0.0064, "step": 65760 }, { "grad_norm": 0.12135180830955505, "learning_rate": 8.404715017388965e-06, "loss": 0.0069, "step": 65770 }, { "grad_norm": 0.1643294245004654, "learning_rate": 8.393249339520015e-06, "loss": 0.0067, "step": 65780 }, { "grad_norm": 0.13982060551643372, "learning_rate": 8.381790771106834e-06, "loss": 0.0064, "step": 65790 }, { "grad_norm": 0.09609866142272949, "learning_rate": 8.370339314107339e-06, "loss": 0.0067, "step": 65800 }, { "grad_norm": 0.14076103270053864, "learning_rate": 8.358894970478281e-06, "loss": 0.0067, "step": 65810 }, { "grad_norm": 0.11810784041881561, "learning_rate": 8.347457742175196e-06, "loss": 0.0072, "step": 65820 }, { "grad_norm": 0.10450443625450134, "learning_rate": 8.33602763115236e-06, "loss": 0.0066, "step": 65830 }, { "grad_norm": 0.10195790976285934, "learning_rate": 8.324604639362916e-06, "loss": 0.0043, "step": 65840 }, { "grad_norm": 0.08147723227739334, "learning_rate": 8.31318876875869e-06, "loss": 0.006, "step": 65850 }, { "grad_norm": 0.1410893350839615, "learning_rate": 8.30178002129039e-06, "loss": 0.0079, "step": 65860 }, { "grad_norm": 0.13025957345962524, "learning_rate": 8.290378398907423e-06, "loss": 0.0065, "step": 65870 }, { "grad_norm": 0.10671830177307129, "learning_rate": 8.278983903558029e-06, "loss": 0.0065, "step": 65880 }, { "grad_norm": 0.10315249115228653, "learning_rate": 8.26759653718921e-06, "loss": 0.0056, "step": 65890 }, { "grad_norm": 0.1743730753660202, "learning_rate": 8.25621630174676e-06, "loss": 0.0071, "step": 65900 }, { "grad_norm": 0.13386374711990356, "learning_rate": 8.244843199175261e-06, "loss": 0.0051, "step": 65910 }, { "grad_norm": 0.1342494934797287, "learning_rate": 8.233477231418046e-06, "loss": 0.0072, "step": 65920 }, { "grad_norm": 0.10189267992973328, "learning_rate": 8.22211840041725e-06, "loss": 0.0065, "step": 65930 }, { "grad_norm": 0.09863460808992386, "learning_rate": 8.210766708113792e-06, "loss": 0.0072, "step": 65940 }, { "grad_norm": 0.057107020169496536, "learning_rate": 8.199422156447367e-06, "loss": 0.0035, "step": 65950 }, { "grad_norm": 0.10595393180847168, "learning_rate": 8.188084747356451e-06, "loss": 0.0051, "step": 65960 }, { "grad_norm": 0.10637208074331284, "learning_rate": 8.176754482778299e-06, "loss": 0.0062, "step": 65970 }, { "grad_norm": 0.13353648781776428, "learning_rate": 8.165431364648918e-06, "loss": 0.0066, "step": 65980 }, { "grad_norm": 0.21423223614692688, "learning_rate": 8.154115394903162e-06, "loss": 0.0066, "step": 65990 }, { "grad_norm": 0.14073893427848816, "learning_rate": 8.142806575474582e-06, "loss": 0.0055, "step": 66000 }, { "grad_norm": 0.21554960310459137, "learning_rate": 8.131504908295562e-06, "loss": 0.0056, "step": 66010 }, { "grad_norm": 0.15400800108909607, "learning_rate": 8.120210395297262e-06, "loss": 0.006, "step": 66020 }, { "grad_norm": 0.06406990438699722, "learning_rate": 8.108923038409565e-06, "loss": 0.0041, "step": 66030 }, { "grad_norm": 0.11114349216222763, "learning_rate": 8.097642839561226e-06, "loss": 0.0046, "step": 66040 }, { "grad_norm": 0.11397046595811844, "learning_rate": 8.08636980067966e-06, "loss": 0.0053, "step": 66050 }, { "grad_norm": 0.09619339555501938, "learning_rate": 8.075103923691186e-06, "loss": 0.0055, "step": 66060 }, { "grad_norm": 0.059990011155605316, "learning_rate": 8.063845210520793e-06, "loss": 0.0045, "step": 66070 }, { "grad_norm": 0.08723434060811996, "learning_rate": 8.052593663092295e-06, "loss": 0.0062, "step": 66080 }, { "grad_norm": 0.11181793361902237, "learning_rate": 8.04134928332827e-06, "loss": 0.0044, "step": 66090 }, { "grad_norm": 0.07525095343589783, "learning_rate": 8.030112073150086e-06, "loss": 0.0066, "step": 66100 }, { "grad_norm": 0.08072611689567566, "learning_rate": 8.018882034477881e-06, "loss": 0.0055, "step": 66110 }, { "grad_norm": 0.1441071480512619, "learning_rate": 8.007659169230541e-06, "loss": 0.006, "step": 66120 }, { "grad_norm": 0.08195093274116516, "learning_rate": 7.996443479325755e-06, "loss": 0.0051, "step": 66130 }, { "grad_norm": 0.09804128110408783, "learning_rate": 7.985234966679977e-06, "loss": 0.0056, "step": 66140 }, { "grad_norm": 0.11881784349679947, "learning_rate": 7.974033633208438e-06, "loss": 0.0067, "step": 66150 }, { "grad_norm": 0.16878263652324677, "learning_rate": 7.962839480825135e-06, "loss": 0.0079, "step": 66160 }, { "grad_norm": 0.12629154324531555, "learning_rate": 7.951652511442858e-06, "loss": 0.006, "step": 66170 }, { "grad_norm": 0.08303108811378479, "learning_rate": 7.940472726973125e-06, "loss": 0.0077, "step": 66180 }, { "grad_norm": 0.09789632260799408, "learning_rate": 7.929300129326289e-06, "loss": 0.0063, "step": 66190 }, { "grad_norm": 0.12584975361824036, "learning_rate": 7.91813472041142e-06, "loss": 0.0066, "step": 66200 }, { "grad_norm": 0.15334461629390717, "learning_rate": 7.906976502136376e-06, "loss": 0.0071, "step": 66210 }, { "grad_norm": 0.1026487648487091, "learning_rate": 7.89582547640782e-06, "loss": 0.0081, "step": 66220 }, { "grad_norm": 0.19761225581169128, "learning_rate": 7.884681645131115e-06, "loss": 0.0058, "step": 66230 }, { "grad_norm": 0.09795992821455002, "learning_rate": 7.87354501021048e-06, "loss": 0.0062, "step": 66240 }, { "grad_norm": 0.12271381169557571, "learning_rate": 7.86241557354882e-06, "loss": 0.0061, "step": 66250 }, { "grad_norm": 0.08298242837190628, "learning_rate": 7.8512933370479e-06, "loss": 0.0061, "step": 66260 }, { "grad_norm": 0.10607089847326279, "learning_rate": 7.840178302608158e-06, "loss": 0.005, "step": 66270 }, { "grad_norm": 0.10421518236398697, "learning_rate": 7.82907047212888e-06, "loss": 0.0049, "step": 66280 }, { "grad_norm": 0.08962974697351456, "learning_rate": 7.81796984750809e-06, "loss": 0.0054, "step": 66290 }, { "grad_norm": 0.044978391379117966, "learning_rate": 7.806876430642546e-06, "loss": 0.0062, "step": 66300 }, { "grad_norm": 0.1172885075211525, "learning_rate": 7.795790223427862e-06, "loss": 0.0068, "step": 66310 }, { "grad_norm": 0.07462570816278458, "learning_rate": 7.784711227758324e-06, "loss": 0.0053, "step": 66320 }, { "grad_norm": 0.13065075874328613, "learning_rate": 7.773639445527053e-06, "loss": 0.005, "step": 66330 }, { "grad_norm": 0.12560197710990906, "learning_rate": 7.762574878625905e-06, "loss": 0.0089, "step": 66340 }, { "grad_norm": 0.12042900174856186, "learning_rate": 7.751517528945513e-06, "loss": 0.0062, "step": 66350 }, { "grad_norm": 0.12490951269865036, "learning_rate": 7.740467398375278e-06, "loss": 0.0051, "step": 66360 }, { "grad_norm": 0.10630281269550323, "learning_rate": 7.729424488803378e-06, "loss": 0.0083, "step": 66370 }, { "grad_norm": 0.1392573118209839, "learning_rate": 7.71838880211671e-06, "loss": 0.0085, "step": 66380 }, { "grad_norm": 0.11627200245857239, "learning_rate": 7.707360340200997e-06, "loss": 0.0072, "step": 66390 }, { "grad_norm": 0.0748174786567688, "learning_rate": 7.696339104940697e-06, "loss": 0.0057, "step": 66400 }, { "grad_norm": 0.09884359687566757, "learning_rate": 7.685325098219038e-06, "loss": 0.0097, "step": 66410 }, { "grad_norm": 0.09944774210453033, "learning_rate": 7.674318321918017e-06, "loss": 0.0069, "step": 66420 }, { "grad_norm": 0.09587261080741882, "learning_rate": 7.663318777918366e-06, "loss": 0.0075, "step": 66430 }, { "grad_norm": 0.11601150035858154, "learning_rate": 7.652326468099647e-06, "loss": 0.0069, "step": 66440 }, { "grad_norm": 0.10933615267276764, "learning_rate": 7.641341394340096e-06, "loss": 0.0074, "step": 66450 }, { "grad_norm": 0.10851206630468369, "learning_rate": 7.630363558516818e-06, "loss": 0.0078, "step": 66460 }, { "grad_norm": 0.13572734594345093, "learning_rate": 7.619392962505578e-06, "loss": 0.0059, "step": 66470 }, { "grad_norm": 0.13620802760124207, "learning_rate": 7.6084296081809725e-06, "loss": 0.0091, "step": 66480 }, { "grad_norm": 0.16528883576393127, "learning_rate": 7.597473497416347e-06, "loss": 0.0074, "step": 66490 }, { "grad_norm": 0.1715632677078247, "learning_rate": 7.586524632083764e-06, "loss": 0.0058, "step": 66500 }, { "grad_norm": 0.0838128849864006, "learning_rate": 7.5755830140541326e-06, "loss": 0.007, "step": 66510 }, { "grad_norm": 0.1573840230703354, "learning_rate": 7.5646486451970475e-06, "loss": 0.0054, "step": 66520 }, { "grad_norm": 0.13193732500076294, "learning_rate": 7.553721527380897e-06, "loss": 0.0063, "step": 66530 }, { "grad_norm": 0.08707915246486664, "learning_rate": 7.542801662472826e-06, "loss": 0.0071, "step": 66540 }, { "grad_norm": 0.09209952503442764, "learning_rate": 7.5318890523387475e-06, "loss": 0.0039, "step": 66550 }, { "grad_norm": 0.1408960074186325, "learning_rate": 7.52098369884332e-06, "loss": 0.006, "step": 66560 }, { "grad_norm": 0.11947312951087952, "learning_rate": 7.510085603849992e-06, "loss": 0.0056, "step": 66570 }, { "grad_norm": 0.1784605085849762, "learning_rate": 7.499194769220918e-06, "loss": 0.0093, "step": 66580 }, { "grad_norm": 0.06847098469734192, "learning_rate": 7.48831119681706e-06, "loss": 0.0072, "step": 66590 }, { "grad_norm": 0.11215118318796158, "learning_rate": 7.477434888498119e-06, "loss": 0.0065, "step": 66600 }, { "grad_norm": 0.09088050574064255, "learning_rate": 7.466565846122564e-06, "loss": 0.0076, "step": 66610 }, { "grad_norm": 0.15088577568531036, "learning_rate": 7.455704071547626e-06, "loss": 0.008, "step": 66620 }, { "grad_norm": 0.17739719152450562, "learning_rate": 7.444849566629247e-06, "loss": 0.0065, "step": 66630 }, { "grad_norm": 0.11036292463541031, "learning_rate": 7.434002333222212e-06, "loss": 0.0059, "step": 66640 }, { "grad_norm": 0.1624198704957962, "learning_rate": 7.423162373179976e-06, "loss": 0.0094, "step": 66650 }, { "grad_norm": 0.06769165396690369, "learning_rate": 7.412329688354835e-06, "loss": 0.0039, "step": 66660 }, { "grad_norm": 0.1029626727104187, "learning_rate": 7.40150428059776e-06, "loss": 0.0084, "step": 66670 }, { "grad_norm": 0.15974536538124084, "learning_rate": 7.3906861517585354e-06, "loss": 0.0051, "step": 66680 }, { "grad_norm": 0.13611184060573578, "learning_rate": 7.37987530368569e-06, "loss": 0.0051, "step": 66690 }, { "grad_norm": 0.0731804221868515, "learning_rate": 7.369071738226474e-06, "loss": 0.0073, "step": 66700 }, { "grad_norm": 0.09280718117952347, "learning_rate": 7.358275457226954e-06, "loss": 0.0054, "step": 66710 }, { "grad_norm": 0.07727409899234772, "learning_rate": 7.347486462531899e-06, "loss": 0.0074, "step": 66720 }, { "grad_norm": 0.08311787992715836, "learning_rate": 7.336704755984858e-06, "loss": 0.0061, "step": 66730 }, { "grad_norm": 0.09097499400377274, "learning_rate": 7.325930339428133e-06, "loss": 0.0056, "step": 66740 }, { "grad_norm": 0.10931100696325302, "learning_rate": 7.315163214702769e-06, "loss": 0.0075, "step": 66750 }, { "grad_norm": 0.07605443149805069, "learning_rate": 7.304403383648595e-06, "loss": 0.0052, "step": 66760 }, { "grad_norm": 0.0832507461309433, "learning_rate": 7.293650848104139e-06, "loss": 0.007, "step": 66770 }, { "grad_norm": 0.08194600045681, "learning_rate": 7.2829056099067374e-06, "loss": 0.0047, "step": 66780 }, { "grad_norm": 0.12203248590230942, "learning_rate": 7.2721676708924494e-06, "loss": 0.0061, "step": 66790 }, { "grad_norm": 0.09527835994958878, "learning_rate": 7.261437032896096e-06, "loss": 0.0067, "step": 66800 }, { "grad_norm": 0.13449349999427795, "learning_rate": 7.250713697751255e-06, "loss": 0.0076, "step": 66810 }, { "grad_norm": 0.14350071549415588, "learning_rate": 7.239997667290255e-06, "loss": 0.0062, "step": 66820 }, { "grad_norm": 0.13342350721359253, "learning_rate": 7.2292889433441425e-06, "loss": 0.0069, "step": 66830 }, { "grad_norm": 0.06814400106668472, "learning_rate": 7.218587527742793e-06, "loss": 0.0057, "step": 66840 }, { "grad_norm": 0.09240850061178207, "learning_rate": 7.207893422314749e-06, "loss": 0.0041, "step": 66850 }, { "grad_norm": 0.09828779101371765, "learning_rate": 7.1972066288873545e-06, "loss": 0.0062, "step": 66860 }, { "grad_norm": 0.09137636423110962, "learning_rate": 7.186527149286687e-06, "loss": 0.0096, "step": 66870 }, { "grad_norm": 0.1580628901720047, "learning_rate": 7.175854985337576e-06, "loss": 0.0063, "step": 66880 }, { "grad_norm": 0.09200670570135117, "learning_rate": 7.16519013886362e-06, "loss": 0.0062, "step": 66890 }, { "grad_norm": 0.10098131746053696, "learning_rate": 7.154532611687109e-06, "loss": 0.0073, "step": 66900 }, { "grad_norm": 0.07432423532009125, "learning_rate": 7.143882405629177e-06, "loss": 0.0058, "step": 66910 }, { "grad_norm": 0.12441837042570114, "learning_rate": 7.133239522509605e-06, "loss": 0.0059, "step": 66920 }, { "grad_norm": 0.10568181425333023, "learning_rate": 7.1226039641469956e-06, "loss": 0.0064, "step": 66930 }, { "grad_norm": 0.09099920839071274, "learning_rate": 7.111975732358678e-06, "loss": 0.005, "step": 66940 }, { "grad_norm": 0.07686493545770645, "learning_rate": 7.101354828960693e-06, "loss": 0.0063, "step": 66950 }, { "grad_norm": 0.05579354241490364, "learning_rate": 7.090741255767918e-06, "loss": 0.0056, "step": 66960 }, { "grad_norm": 0.09745355695486069, "learning_rate": 7.080135014593875e-06, "loss": 0.0062, "step": 66970 }, { "grad_norm": 0.05413778871297836, "learning_rate": 7.069536107250896e-06, "loss": 0.0049, "step": 66980 }, { "grad_norm": 0.11062846332788467, "learning_rate": 7.058944535550049e-06, "loss": 0.0049, "step": 66990 }, { "grad_norm": 0.06933420151472092, "learning_rate": 7.048360301301138e-06, "loss": 0.0043, "step": 67000 }, { "grad_norm": 0.08647534251213074, "learning_rate": 7.03778340631272e-06, "loss": 0.006, "step": 67010 }, { "grad_norm": 0.08573900908231735, "learning_rate": 7.0272138523921e-06, "loss": 0.0055, "step": 67020 }, { "grad_norm": 0.11077279597520828, "learning_rate": 7.016651641345334e-06, "loss": 0.0083, "step": 67030 }, { "grad_norm": 0.08280877768993378, "learning_rate": 7.0060967749772e-06, "loss": 0.0058, "step": 67040 }, { "grad_norm": 0.11848456412553787, "learning_rate": 6.995549255091238e-06, "loss": 0.0061, "step": 67050 }, { "grad_norm": 0.17615468800067902, "learning_rate": 6.98500908348973e-06, "loss": 0.0059, "step": 67060 }, { "grad_norm": 0.09687437117099762, "learning_rate": 6.974476261973711e-06, "loss": 0.0054, "step": 67070 }, { "grad_norm": 0.066740483045578, "learning_rate": 6.963950792342949e-06, "loss": 0.0056, "step": 67080 }, { "grad_norm": 0.11340686678886414, "learning_rate": 6.9534326763959715e-06, "loss": 0.0049, "step": 67090 }, { "grad_norm": 0.07200829684734344, "learning_rate": 6.942921915929995e-06, "loss": 0.0067, "step": 67100 }, { "grad_norm": 0.07395389676094055, "learning_rate": 6.9324185127410735e-06, "loss": 0.0041, "step": 67110 }, { "grad_norm": 0.18259793519973755, "learning_rate": 6.921922468623921e-06, "loss": 0.006, "step": 67120 }, { "grad_norm": 0.08521291613578796, "learning_rate": 6.911433785372023e-06, "loss": 0.0072, "step": 67130 }, { "grad_norm": 0.0848226398229599, "learning_rate": 6.900952464777632e-06, "loss": 0.0054, "step": 67140 }, { "grad_norm": 0.12612730264663696, "learning_rate": 6.8904785086316815e-06, "loss": 0.0057, "step": 67150 }, { "grad_norm": 0.10118938237428665, "learning_rate": 6.880011918723927e-06, "loss": 0.0074, "step": 67160 }, { "grad_norm": 0.10184956341981888, "learning_rate": 6.86955269684279e-06, "loss": 0.006, "step": 67170 }, { "grad_norm": 0.12220960110425949, "learning_rate": 6.859100844775473e-06, "loss": 0.0064, "step": 67180 }, { "grad_norm": 0.07468807697296143, "learning_rate": 6.84865636430792e-06, "loss": 0.0052, "step": 67190 }, { "grad_norm": 0.13770878314971924, "learning_rate": 6.838219257224804e-06, "loss": 0.0062, "step": 67200 }, { "grad_norm": 0.07497981190681458, "learning_rate": 6.827789525309536e-06, "loss": 0.0054, "step": 67210 }, { "grad_norm": 0.11704042553901672, "learning_rate": 6.81736717034428e-06, "loss": 0.0058, "step": 67220 }, { "grad_norm": 0.06326339393854141, "learning_rate": 6.806952194109933e-06, "loss": 0.0056, "step": 67230 }, { "grad_norm": 0.0797201544046402, "learning_rate": 6.796544598386112e-06, "loss": 0.0077, "step": 67240 }, { "grad_norm": 0.14098401367664337, "learning_rate": 6.786144384951204e-06, "loss": 0.0085, "step": 67250 }, { "grad_norm": 0.20316767692565918, "learning_rate": 6.775751555582322e-06, "loss": 0.0064, "step": 67260 }, { "grad_norm": 0.09969202429056168, "learning_rate": 6.76536611205531e-06, "loss": 0.0046, "step": 67270 }, { "grad_norm": 0.128652885556221, "learning_rate": 6.754988056144762e-06, "loss": 0.0093, "step": 67280 }, { "grad_norm": 0.1345537155866623, "learning_rate": 6.744617389624014e-06, "loss": 0.0065, "step": 67290 }, { "grad_norm": 0.19551172852516174, "learning_rate": 6.734254114265087e-06, "loss": 0.0087, "step": 67300 }, { "grad_norm": 0.15187862515449524, "learning_rate": 6.723898231838843e-06, "loss": 0.0084, "step": 67310 }, { "grad_norm": 0.11123475432395935, "learning_rate": 6.713549744114766e-06, "loss": 0.0055, "step": 67320 }, { "grad_norm": 0.08245950937271118, "learning_rate": 6.703208652861159e-06, "loss": 0.0057, "step": 67330 }, { "grad_norm": 0.09134593605995178, "learning_rate": 6.69287495984503e-06, "loss": 0.0053, "step": 67340 }, { "grad_norm": 0.11437999457120895, "learning_rate": 6.6825486668321e-06, "loss": 0.0064, "step": 67350 }, { "grad_norm": 0.12371183186769485, "learning_rate": 6.672229775586886e-06, "loss": 0.0081, "step": 67360 }, { "grad_norm": 0.12538525462150574, "learning_rate": 6.661918287872576e-06, "loss": 0.006, "step": 67370 }, { "grad_norm": 0.08089082688093185, "learning_rate": 6.6516142054511346e-06, "loss": 0.0057, "step": 67380 }, { "grad_norm": 0.1408071666955948, "learning_rate": 6.641317530083241e-06, "loss": 0.008, "step": 67390 }, { "grad_norm": 0.09195517003536224, "learning_rate": 6.631028263528322e-06, "loss": 0.0063, "step": 67400 }, { "grad_norm": 0.0993531122803688, "learning_rate": 6.620746407544537e-06, "loss": 0.0072, "step": 67410 }, { "grad_norm": 0.06948203593492508, "learning_rate": 6.610471963888742e-06, "loss": 0.0053, "step": 67420 }, { "grad_norm": 0.12527501583099365, "learning_rate": 6.600204934316606e-06, "loss": 0.006, "step": 67430 }, { "grad_norm": 0.11383728682994843, "learning_rate": 6.589945320582452e-06, "loss": 0.0073, "step": 67440 }, { "grad_norm": 0.0906723216176033, "learning_rate": 6.579693124439374e-06, "loss": 0.0045, "step": 67450 }, { "grad_norm": 0.13000693917274475, "learning_rate": 6.56944834763919e-06, "loss": 0.006, "step": 67460 }, { "grad_norm": 0.07689883559942245, "learning_rate": 6.5592109919324575e-06, "loss": 0.0063, "step": 67470 }, { "grad_norm": 0.11414958536624908, "learning_rate": 6.54898105906846e-06, "loss": 0.0069, "step": 67480 }, { "grad_norm": 0.12023609131574631, "learning_rate": 6.5387585507952155e-06, "loss": 0.0077, "step": 67490 }, { "grad_norm": 0.11329898983240128, "learning_rate": 6.528543468859461e-06, "loss": 0.0062, "step": 67500 }, { "grad_norm": 0.12551210820674896, "learning_rate": 6.518335815006682e-06, "loss": 0.0065, "step": 67510 }, { "grad_norm": 0.11698201298713684, "learning_rate": 6.5081355909810845e-06, "loss": 0.0073, "step": 67520 }, { "grad_norm": 0.11045945435762405, "learning_rate": 6.497942798525608e-06, "loss": 0.0048, "step": 67530 }, { "grad_norm": 0.06777746975421906, "learning_rate": 6.487757439381936e-06, "loss": 0.005, "step": 67540 }, { "grad_norm": 0.1390896737575531, "learning_rate": 6.477579515290433e-06, "loss": 0.0084, "step": 67550 }, { "grad_norm": 0.08686929941177368, "learning_rate": 6.467409027990273e-06, "loss": 0.0054, "step": 67560 }, { "grad_norm": 0.09431102126836777, "learning_rate": 6.457245979219279e-06, "loss": 0.0074, "step": 67570 }, { "grad_norm": 0.08216726779937744, "learning_rate": 6.447090370714054e-06, "loss": 0.0056, "step": 67580 }, { "grad_norm": 0.12552209198474884, "learning_rate": 6.436942204209917e-06, "loss": 0.0056, "step": 67590 }, { "grad_norm": 0.11992699652910233, "learning_rate": 6.4268014814408804e-06, "loss": 0.0071, "step": 67600 }, { "grad_norm": 0.08301329612731934, "learning_rate": 6.41666820413977e-06, "loss": 0.0069, "step": 67610 }, { "grad_norm": 0.08920196443796158, "learning_rate": 6.406542374038033e-06, "loss": 0.0053, "step": 67620 }, { "grad_norm": 0.10756463557481766, "learning_rate": 6.396423992865935e-06, "loss": 0.0058, "step": 67630 }, { "grad_norm": 0.10856829583644867, "learning_rate": 6.386313062352412e-06, "loss": 0.0044, "step": 67640 }, { "grad_norm": 0.11400957405567169, "learning_rate": 6.376209584225152e-06, "loss": 0.0055, "step": 67650 }, { "grad_norm": 0.08003991842269897, "learning_rate": 6.366113560210557e-06, "loss": 0.0054, "step": 67660 }, { "grad_norm": 0.12225547432899475, "learning_rate": 6.356024992033766e-06, "loss": 0.0056, "step": 67670 }, { "grad_norm": 0.09724835306406021, "learning_rate": 6.345943881418648e-06, "loss": 0.0059, "step": 67680 }, { "grad_norm": 0.09970936924219131, "learning_rate": 6.33587023008777e-06, "loss": 0.0056, "step": 67690 }, { "grad_norm": 0.10497895628213882, "learning_rate": 6.32580403976245e-06, "loss": 0.0049, "step": 67700 }, { "grad_norm": 0.12747107446193695, "learning_rate": 6.31574531216273e-06, "loss": 0.0073, "step": 67710 }, { "grad_norm": 0.09661749005317688, "learning_rate": 6.305694049007371e-06, "loss": 0.0047, "step": 67720 }, { "grad_norm": 0.09074199944734573, "learning_rate": 6.2956502520138575e-06, "loss": 0.0056, "step": 67730 }, { "grad_norm": 0.1730847954750061, "learning_rate": 6.285613922898409e-06, "loss": 0.0084, "step": 67740 }, { "grad_norm": 0.10450442880392075, "learning_rate": 6.275585063375927e-06, "loss": 0.0043, "step": 67750 }, { "grad_norm": 0.11512558907270432, "learning_rate": 6.265563675160113e-06, "loss": 0.0055, "step": 67760 }, { "grad_norm": 0.03832417353987694, "learning_rate": 6.25554975996332e-06, "loss": 0.0079, "step": 67770 }, { "grad_norm": 0.09743283689022064, "learning_rate": 6.245543319496661e-06, "loss": 0.006, "step": 67780 }, { "grad_norm": 0.12195522338151932, "learning_rate": 6.2355443554699685e-06, "loss": 0.006, "step": 67790 }, { "grad_norm": 0.13597218692302704, "learning_rate": 6.225552869591766e-06, "loss": 0.0062, "step": 67800 }, { "grad_norm": 0.1503504067659378, "learning_rate": 6.215568863569365e-06, "loss": 0.0063, "step": 67810 }, { "grad_norm": 0.053992860019207, "learning_rate": 6.205592339108712e-06, "loss": 0.0054, "step": 67820 }, { "grad_norm": 0.08793985098600388, "learning_rate": 6.195623297914577e-06, "loss": 0.0046, "step": 67830 }, { "grad_norm": 0.1429556906223297, "learning_rate": 6.185661741690357e-06, "loss": 0.0051, "step": 67840 }, { "grad_norm": 0.09303241223096848, "learning_rate": 6.1757076721382145e-06, "loss": 0.0056, "step": 67850 }, { "grad_norm": 0.08834455907344818, "learning_rate": 6.165761090959038e-06, "loss": 0.0065, "step": 67860 }, { "grad_norm": 0.11822805553674698, "learning_rate": 6.155821999852424e-06, "loss": 0.0056, "step": 67870 }, { "grad_norm": 0.12221843004226685, "learning_rate": 6.145890400516696e-06, "loss": 0.0097, "step": 67880 }, { "grad_norm": 0.08334722369909286, "learning_rate": 6.1359662946488816e-06, "loss": 0.006, "step": 67890 }, { "grad_norm": 0.08549880236387253, "learning_rate": 6.12604968394474e-06, "loss": 0.0055, "step": 67900 }, { "grad_norm": 0.11731625348329544, "learning_rate": 6.11614057009875e-06, "loss": 0.009, "step": 67910 }, { "grad_norm": 0.07543738931417465, "learning_rate": 6.106238954804111e-06, "loss": 0.0074, "step": 67920 }, { "grad_norm": 0.12775328755378723, "learning_rate": 6.096344839752738e-06, "loss": 0.0079, "step": 67930 }, { "grad_norm": 0.09410545974969864, "learning_rate": 6.086458226635278e-06, "loss": 0.005, "step": 67940 }, { "grad_norm": 0.12600472569465637, "learning_rate": 6.076579117141046e-06, "loss": 0.0074, "step": 67950 }, { "grad_norm": 0.1349714696407318, "learning_rate": 6.066707512958153e-06, "loss": 0.0053, "step": 67960 }, { "grad_norm": 0.06858479976654053, "learning_rate": 6.056843415773361e-06, "loss": 0.006, "step": 67970 }, { "grad_norm": 0.08731766790151596, "learning_rate": 6.0469868272721776e-06, "loss": 0.005, "step": 67980 }, { "grad_norm": 0.11096678674221039, "learning_rate": 6.037137749138844e-06, "loss": 0.0054, "step": 67990 }, { "grad_norm": 0.10543126612901688, "learning_rate": 6.027296183056252e-06, "loss": 0.0056, "step": 68000 }, { "grad_norm": 0.12348238378763199, "learning_rate": 6.017462130706114e-06, "loss": 0.0065, "step": 68010 }, { "grad_norm": 0.11583581566810608, "learning_rate": 6.0076355937687516e-06, "loss": 0.0057, "step": 68020 }, { "grad_norm": 0.09548505395650864, "learning_rate": 5.9978165739232925e-06, "loss": 0.0055, "step": 68030 }, { "grad_norm": 0.1131211444735527, "learning_rate": 5.988005072847508e-06, "loss": 0.007, "step": 68040 }, { "grad_norm": 0.17244119942188263, "learning_rate": 5.978201092217928e-06, "loss": 0.0063, "step": 68050 }, { "grad_norm": 0.15796402096748352, "learning_rate": 5.9684046337097895e-06, "loss": 0.0056, "step": 68060 }, { "grad_norm": 0.17315950989723206, "learning_rate": 5.958615698997016e-06, "loss": 0.0053, "step": 68070 }, { "grad_norm": 0.14525869488716125, "learning_rate": 5.948834289752303e-06, "loss": 0.0062, "step": 68080 }, { "grad_norm": 0.12285970896482468, "learning_rate": 5.939060407646996e-06, "loss": 0.0076, "step": 68090 }, { "grad_norm": 0.11547025293111801, "learning_rate": 5.929294054351198e-06, "loss": 0.0051, "step": 68100 }, { "grad_norm": 0.06043177843093872, "learning_rate": 5.919535231533707e-06, "loss": 0.0077, "step": 68110 }, { "grad_norm": 0.1934865266084671, "learning_rate": 5.909783940862046e-06, "loss": 0.0061, "step": 68120 }, { "grad_norm": 0.08116970956325531, "learning_rate": 5.900040184002436e-06, "loss": 0.0049, "step": 68130 }, { "grad_norm": 0.14193123579025269, "learning_rate": 5.890303962619831e-06, "loss": 0.0055, "step": 68140 }, { "grad_norm": 0.16365167498588562, "learning_rate": 5.88057527837787e-06, "loss": 0.0049, "step": 68150 }, { "grad_norm": 0.11437302827835083, "learning_rate": 5.870854132938924e-06, "loss": 0.0046, "step": 68160 }, { "grad_norm": 0.1819181591272354, "learning_rate": 5.861140527964071e-06, "loss": 0.0092, "step": 68170 }, { "grad_norm": 0.08394356071949005, "learning_rate": 5.851434465113098e-06, "loss": 0.0052, "step": 68180 }, { "grad_norm": 0.12516161799430847, "learning_rate": 5.841735946044524e-06, "loss": 0.0042, "step": 68190 }, { "grad_norm": 0.12525568902492523, "learning_rate": 5.832044972415523e-06, "loss": 0.0052, "step": 68200 }, { "grad_norm": 0.07477275282144547, "learning_rate": 5.8223615458820655e-06, "loss": 0.0052, "step": 68210 }, { "grad_norm": 0.13306181132793427, "learning_rate": 5.812685668098733e-06, "loss": 0.0046, "step": 68220 }, { "grad_norm": 0.055557072162628174, "learning_rate": 5.803017340718913e-06, "loss": 0.0044, "step": 68230 }, { "grad_norm": 0.09655983000993729, "learning_rate": 5.793356565394636e-06, "loss": 0.0045, "step": 68240 }, { "grad_norm": 0.12149965018033981, "learning_rate": 5.783703343776658e-06, "loss": 0.0064, "step": 68250 }, { "grad_norm": 0.1710568070411682, "learning_rate": 5.774057677514477e-06, "loss": 0.0066, "step": 68260 }, { "grad_norm": 0.13738037645816803, "learning_rate": 5.764419568256235e-06, "loss": 0.0049, "step": 68270 }, { "grad_norm": 0.17861013114452362, "learning_rate": 5.754789017648865e-06, "loss": 0.0071, "step": 68280 }, { "grad_norm": 0.09412147849798203, "learning_rate": 5.745166027337934e-06, "loss": 0.0046, "step": 68290 }, { "grad_norm": 0.16847530007362366, "learning_rate": 5.735550598967754e-06, "loss": 0.0054, "step": 68300 }, { "grad_norm": 0.08715786039829254, "learning_rate": 5.725942734181339e-06, "loss": 0.0056, "step": 68310 }, { "grad_norm": 0.11249921470880508, "learning_rate": 5.7163424346204146e-06, "loss": 0.0073, "step": 68320 }, { "grad_norm": 0.1947953850030899, "learning_rate": 5.706749701925407e-06, "loss": 0.0062, "step": 68330 }, { "grad_norm": 0.09570833295583725, "learning_rate": 5.697164537735461e-06, "loss": 0.006, "step": 68340 }, { "grad_norm": 0.09514325857162476, "learning_rate": 5.687586943688406e-06, "loss": 0.0061, "step": 68350 }, { "grad_norm": 0.10578252375125885, "learning_rate": 5.678016921420792e-06, "loss": 0.0057, "step": 68360 }, { "grad_norm": 0.124508336186409, "learning_rate": 5.668454472567875e-06, "loss": 0.0082, "step": 68370 }, { "grad_norm": 0.11126866936683655, "learning_rate": 5.658899598763617e-06, "loss": 0.0067, "step": 68380 }, { "grad_norm": 0.09869793057441711, "learning_rate": 5.6493523016407035e-06, "loss": 0.0059, "step": 68390 }, { "grad_norm": 0.14812630414962769, "learning_rate": 5.6398125828304615e-06, "loss": 0.0088, "step": 68400 }, { "grad_norm": 0.13188178837299347, "learning_rate": 5.630280443963015e-06, "loss": 0.0085, "step": 68410 }, { "grad_norm": 0.09762341529130936, "learning_rate": 5.620755886667112e-06, "loss": 0.0052, "step": 68420 }, { "grad_norm": 0.09884628653526306, "learning_rate": 5.611238912570271e-06, "loss": 0.0079, "step": 68430 }, { "grad_norm": 0.08027556538581848, "learning_rate": 5.601729523298649e-06, "loss": 0.0053, "step": 68440 }, { "grad_norm": 0.11467627435922623, "learning_rate": 5.592227720477161e-06, "loss": 0.0069, "step": 68450 }, { "grad_norm": 0.13773809373378754, "learning_rate": 5.582733505729415e-06, "loss": 0.0059, "step": 68460 }, { "grad_norm": 0.0960000678896904, "learning_rate": 5.573246880677668e-06, "loss": 0.0059, "step": 68470 }, { "grad_norm": 0.09526869654655457, "learning_rate": 5.56376784694298e-06, "loss": 0.005, "step": 68480 }, { "grad_norm": 0.1231173649430275, "learning_rate": 5.554296406145027e-06, "loss": 0.0062, "step": 68490 }, { "grad_norm": 0.11626993864774704, "learning_rate": 5.544832559902219e-06, "loss": 0.0062, "step": 68500 }, { "grad_norm": 0.09697849303483963, "learning_rate": 5.53537630983168e-06, "loss": 0.0047, "step": 68510 }, { "grad_norm": 0.0740026906132698, "learning_rate": 5.5259276575492125e-06, "loss": 0.0058, "step": 68520 }, { "grad_norm": 0.08156967908143997, "learning_rate": 5.516486604669357e-06, "loss": 0.0042, "step": 68530 }, { "grad_norm": 0.0577608086168766, "learning_rate": 5.507053152805303e-06, "loss": 0.0055, "step": 68540 }, { "grad_norm": 0.11359752714633942, "learning_rate": 5.497627303568975e-06, "loss": 0.0048, "step": 68550 }, { "grad_norm": 0.09739018976688385, "learning_rate": 5.488209058571003e-06, "loss": 0.0059, "step": 68560 }, { "grad_norm": 0.11631833761930466, "learning_rate": 5.4787984194207054e-06, "loss": 0.0063, "step": 68570 }, { "grad_norm": 0.1333731710910797, "learning_rate": 5.469395387726095e-06, "loss": 0.0056, "step": 68580 }, { "grad_norm": 0.14069244265556335, "learning_rate": 5.45999996509391e-06, "loss": 0.0071, "step": 68590 }, { "grad_norm": 0.13593849539756775, "learning_rate": 5.450612153129536e-06, "loss": 0.0051, "step": 68600 }, { "grad_norm": 0.15911436080932617, "learning_rate": 5.4412319534371426e-06, "loss": 0.0065, "step": 68610 }, { "grad_norm": 0.1458415389060974, "learning_rate": 5.431859367619513e-06, "loss": 0.0071, "step": 68620 }, { "grad_norm": 0.12804977595806122, "learning_rate": 5.422494397278172e-06, "loss": 0.007, "step": 68630 }, { "grad_norm": 0.10905634611845016, "learning_rate": 5.413137044013344e-06, "loss": 0.0063, "step": 68640 }, { "grad_norm": 0.15726983547210693, "learning_rate": 5.403787309423941e-06, "loss": 0.0076, "step": 68650 }, { "grad_norm": 0.20486074686050415, "learning_rate": 5.3944451951075835e-06, "loss": 0.0077, "step": 68660 }, { "grad_norm": 0.1946125030517578, "learning_rate": 5.385110702660562e-06, "loss": 0.009, "step": 68670 }, { "grad_norm": 0.1585768610239029, "learning_rate": 5.375783833677922e-06, "loss": 0.0074, "step": 68680 }, { "grad_norm": 0.09023807942867279, "learning_rate": 5.36646458975334e-06, "loss": 0.0058, "step": 68690 }, { "grad_norm": 0.11345356702804565, "learning_rate": 5.3571529724792294e-06, "loss": 0.0084, "step": 68700 }, { "grad_norm": 0.17494092881679535, "learning_rate": 5.347848983446702e-06, "loss": 0.0064, "step": 68710 }, { "grad_norm": 0.05191732943058014, "learning_rate": 5.3385526242455185e-06, "loss": 0.0074, "step": 68720 }, { "grad_norm": 0.07064533978700638, "learning_rate": 5.329263896464226e-06, "loss": 0.0049, "step": 68730 }, { "grad_norm": 0.1002504825592041, "learning_rate": 5.3199828016899715e-06, "loss": 0.0062, "step": 68740 }, { "grad_norm": 0.07087689638137817, "learning_rate": 5.310709341508657e-06, "loss": 0.0042, "step": 68750 }, { "grad_norm": 0.08299539238214493, "learning_rate": 5.301443517504861e-06, "loss": 0.0067, "step": 68760 }, { "grad_norm": 0.11737237870693207, "learning_rate": 5.292185331261862e-06, "loss": 0.0075, "step": 68770 }, { "grad_norm": 0.059911858290433884, "learning_rate": 5.28293478436162e-06, "loss": 0.0046, "step": 68780 }, { "grad_norm": 0.12685976922512054, "learning_rate": 5.273691878384829e-06, "loss": 0.0063, "step": 68790 }, { "grad_norm": 0.2033957690000534, "learning_rate": 5.264456614910812e-06, "loss": 0.0057, "step": 68800 }, { "grad_norm": 0.1064559817314148, "learning_rate": 5.255228995517647e-06, "loss": 0.0087, "step": 68810 }, { "grad_norm": 0.09809686988592148, "learning_rate": 5.246009021782067e-06, "loss": 0.0069, "step": 68820 }, { "grad_norm": 0.12470041960477829, "learning_rate": 5.2367966952795225e-06, "loss": 0.0055, "step": 68830 }, { "grad_norm": 0.34921762347221375, "learning_rate": 5.2275920175841485e-06, "loss": 0.0069, "step": 68840 }, { "grad_norm": 0.06792579591274261, "learning_rate": 5.218394990268771e-06, "loss": 0.0056, "step": 68850 }, { "grad_norm": 0.09734850376844406, "learning_rate": 5.209205614904916e-06, "loss": 0.0053, "step": 68860 }, { "grad_norm": 0.09631146490573883, "learning_rate": 5.200023893062777e-06, "loss": 0.0056, "step": 68870 }, { "grad_norm": 0.10855080932378769, "learning_rate": 5.190849826311289e-06, "loss": 0.0072, "step": 68880 }, { "grad_norm": 0.07261756807565689, "learning_rate": 5.181683416218025e-06, "loss": 0.0044, "step": 68890 }, { "grad_norm": 0.09473758935928345, "learning_rate": 5.172524664349276e-06, "loss": 0.0048, "step": 68900 }, { "grad_norm": 0.09563300013542175, "learning_rate": 5.1633735722700416e-06, "loss": 0.0047, "step": 68910 }, { "grad_norm": 0.12116235494613647, "learning_rate": 5.154230141543958e-06, "loss": 0.0057, "step": 68920 }, { "grad_norm": 0.04672279953956604, "learning_rate": 5.145094373733433e-06, "loss": 0.0057, "step": 68930 }, { "grad_norm": 0.12341974675655365, "learning_rate": 5.135966270399478e-06, "loss": 0.0063, "step": 68940 }, { "grad_norm": 0.11690814048051834, "learning_rate": 5.126845833101857e-06, "loss": 0.0066, "step": 68950 }, { "grad_norm": 0.056633394211530685, "learning_rate": 5.1177330633989994e-06, "loss": 0.0053, "step": 68960 }, { "grad_norm": 0.11050450801849365, "learning_rate": 5.108627962848033e-06, "loss": 0.0062, "step": 68970 }, { "grad_norm": 0.19458343088626862, "learning_rate": 5.099530533004759e-06, "loss": 0.0061, "step": 68980 }, { "grad_norm": 0.05968352407217026, "learning_rate": 5.090440775423699e-06, "loss": 0.0071, "step": 68990 }, { "grad_norm": 0.14878152310848236, "learning_rate": 5.081358691658022e-06, "loss": 0.0056, "step": 69000 }, { "grad_norm": 0.1544198840856552, "learning_rate": 5.072284283259621e-06, "loss": 0.0061, "step": 69010 }, { "grad_norm": 0.13209103047847748, "learning_rate": 5.063217551779053e-06, "loss": 0.0055, "step": 69020 }, { "grad_norm": 0.07890110462903976, "learning_rate": 5.054158498765588e-06, "loss": 0.006, "step": 69030 }, { "grad_norm": 0.07772171497344971, "learning_rate": 5.0451071257671625e-06, "loss": 0.0062, "step": 69040 }, { "grad_norm": 0.1953461468219757, "learning_rate": 5.036063434330407e-06, "loss": 0.0069, "step": 69050 }, { "grad_norm": 0.08872156590223312, "learning_rate": 5.027027426000652e-06, "loss": 0.0085, "step": 69060 }, { "grad_norm": 0.13960371911525726, "learning_rate": 5.017999102321886e-06, "loss": 0.0078, "step": 69070 }, { "grad_norm": 0.09699589014053345, "learning_rate": 5.0089784648368224e-06, "loss": 0.0042, "step": 69080 }, { "grad_norm": 0.06822865456342697, "learning_rate": 4.999965515086829e-06, "loss": 0.0077, "step": 69090 }, { "grad_norm": 0.11388923972845078, "learning_rate": 4.990960254611976e-06, "loss": 0.007, "step": 69100 }, { "grad_norm": 0.15967361629009247, "learning_rate": 4.9819626849510194e-06, "loss": 0.0049, "step": 69110 }, { "grad_norm": 0.16515479981899261, "learning_rate": 4.97297280764138e-06, "loss": 0.0073, "step": 69120 }, { "grad_norm": 0.10848111659288406, "learning_rate": 4.963990624219211e-06, "loss": 0.005, "step": 69130 }, { "grad_norm": 0.12952575087547302, "learning_rate": 4.955016136219298e-06, "loss": 0.0069, "step": 69140 }, { "grad_norm": 0.08967354148626328, "learning_rate": 4.946049345175146e-06, "loss": 0.0061, "step": 69150 }, { "grad_norm": 0.08901200443506241, "learning_rate": 4.937090252618937e-06, "loss": 0.0057, "step": 69160 }, { "grad_norm": 0.11044269055128098, "learning_rate": 4.928138860081521e-06, "loss": 0.0082, "step": 69170 }, { "grad_norm": 0.0826537013053894, "learning_rate": 4.919195169092472e-06, "loss": 0.0079, "step": 69180 }, { "grad_norm": 0.08078809827566147, "learning_rate": 4.910259181179994e-06, "loss": 0.0048, "step": 69190 }, { "grad_norm": 0.09507793188095093, "learning_rate": 4.901330897871015e-06, "loss": 0.007, "step": 69200 }, { "grad_norm": 0.06085929274559021, "learning_rate": 4.8924103206911375e-06, "loss": 0.0056, "step": 69210 }, { "grad_norm": 0.06022636592388153, "learning_rate": 4.883497451164637e-06, "loss": 0.0051, "step": 69220 }, { "grad_norm": 0.14064215123653412, "learning_rate": 4.874592290814484e-06, "loss": 0.0085, "step": 69230 }, { "grad_norm": 0.09809162467718124, "learning_rate": 4.865694841162327e-06, "loss": 0.008, "step": 69240 }, { "grad_norm": 0.14028605818748474, "learning_rate": 4.856805103728496e-06, "loss": 0.0064, "step": 69250 }, { "grad_norm": 0.11021445691585541, "learning_rate": 4.847923080032007e-06, "loss": 0.0077, "step": 69260 }, { "grad_norm": 0.06331007182598114, "learning_rate": 4.8390487715905486e-06, "loss": 0.0051, "step": 69270 }, { "grad_norm": 0.15010663866996765, "learning_rate": 4.8301821799205e-06, "loss": 0.0052, "step": 69280 }, { "grad_norm": 0.21305948495864868, "learning_rate": 4.821323306536918e-06, "loss": 0.0053, "step": 69290 }, { "grad_norm": 0.10107839107513428, "learning_rate": 4.8124721529535455e-06, "loss": 0.0062, "step": 69300 }, { "grad_norm": 0.09648123383522034, "learning_rate": 4.803628720682807e-06, "loss": 0.0039, "step": 69310 }, { "grad_norm": 0.10448157042264938, "learning_rate": 4.794793011235776e-06, "loss": 0.0048, "step": 69320 }, { "grad_norm": 0.07130225002765656, "learning_rate": 4.78596502612228e-06, "loss": 0.0048, "step": 69330 }, { "grad_norm": 0.133016899228096, "learning_rate": 4.777144766850738e-06, "loss": 0.0058, "step": 69340 }, { "grad_norm": 0.12479941546916962, "learning_rate": 4.768332234928313e-06, "loss": 0.0063, "step": 69350 }, { "grad_norm": 0.10650033503770828, "learning_rate": 4.759527431860828e-06, "loss": 0.0043, "step": 69360 }, { "grad_norm": 0.10525579750537872, "learning_rate": 4.750730359152755e-06, "loss": 0.0054, "step": 69370 }, { "grad_norm": 0.08122555911540985, "learning_rate": 4.741941018307311e-06, "loss": 0.0042, "step": 69380 }, { "grad_norm": 0.10282838344573975, "learning_rate": 4.733159410826321e-06, "loss": 0.0047, "step": 69390 }, { "grad_norm": 0.11699098348617554, "learning_rate": 4.724385538210357e-06, "loss": 0.0063, "step": 69400 }, { "grad_norm": 0.07769222557544708, "learning_rate": 4.715619401958599e-06, "loss": 0.0057, "step": 69410 }, { "grad_norm": 0.09786135703325272, "learning_rate": 4.706861003568958e-06, "loss": 0.0043, "step": 69420 }, { "grad_norm": 0.06489010155200958, "learning_rate": 4.698110344538003e-06, "loss": 0.0058, "step": 69430 }, { "grad_norm": 0.09169984608888626, "learning_rate": 4.689367426360975e-06, "loss": 0.0049, "step": 69440 }, { "grad_norm": 0.055365193635225296, "learning_rate": 4.680632250531819e-06, "loss": 0.0046, "step": 69450 }, { "grad_norm": 0.07821185886859894, "learning_rate": 4.671904818543115e-06, "loss": 0.0064, "step": 69460 }, { "grad_norm": 0.181191548705101, "learning_rate": 4.66318513188615e-06, "loss": 0.0066, "step": 69470 }, { "grad_norm": 0.09721329063177109, "learning_rate": 4.654473192050884e-06, "loss": 0.0057, "step": 69480 }, { "grad_norm": 0.11756564676761627, "learning_rate": 4.64576900052594e-06, "loss": 0.0059, "step": 69490 }, { "grad_norm": 0.12604647874832153, "learning_rate": 4.637072558798638e-06, "loss": 0.0075, "step": 69500 }, { "grad_norm": 0.13880272209644318, "learning_rate": 4.628383868354969e-06, "loss": 0.0066, "step": 69510 }, { "grad_norm": 0.10189623385667801, "learning_rate": 4.6197029306795595e-06, "loss": 0.0094, "step": 69520 }, { "grad_norm": 0.07348693162202835, "learning_rate": 4.611029747255779e-06, "loss": 0.0051, "step": 69530 }, { "grad_norm": 0.14053460955619812, "learning_rate": 4.6023643195656164e-06, "loss": 0.0052, "step": 69540 }, { "grad_norm": 0.09832392632961273, "learning_rate": 4.593706649089768e-06, "loss": 0.0069, "step": 69550 }, { "grad_norm": 0.09958355128765106, "learning_rate": 4.585056737307597e-06, "loss": 0.0049, "step": 69560 }, { "grad_norm": 0.15482661128044128, "learning_rate": 4.576414585697103e-06, "loss": 0.0056, "step": 69570 }, { "grad_norm": 0.0941481664776802, "learning_rate": 4.567780195735044e-06, "loss": 0.0065, "step": 69580 }, { "grad_norm": 0.08339086174964905, "learning_rate": 4.559153568896757e-06, "loss": 0.0058, "step": 69590 }, { "grad_norm": 0.0903439000248909, "learning_rate": 4.550534706656329e-06, "loss": 0.0058, "step": 69600 }, { "grad_norm": 0.05791918933391571, "learning_rate": 4.541923610486465e-06, "loss": 0.0062, "step": 69610 }, { "grad_norm": 0.0775686651468277, "learning_rate": 4.533320281858578e-06, "loss": 0.0061, "step": 69620 }, { "grad_norm": 0.11546394973993301, "learning_rate": 4.52472472224274e-06, "loss": 0.0054, "step": 69630 }, { "grad_norm": 0.054117362946271896, "learning_rate": 4.51613693310769e-06, "loss": 0.0057, "step": 69640 }, { "grad_norm": 0.0815228670835495, "learning_rate": 4.507556915920868e-06, "loss": 0.0055, "step": 69650 }, { "grad_norm": 0.08061571419239044, "learning_rate": 4.498984672148332e-06, "loss": 0.0049, "step": 69660 }, { "grad_norm": 0.09641323238611221, "learning_rate": 4.490420203254864e-06, "loss": 0.0048, "step": 69670 }, { "grad_norm": 0.06346229463815689, "learning_rate": 4.481863510703893e-06, "loss": 0.0076, "step": 69680 }, { "grad_norm": 0.10744519531726837, "learning_rate": 4.473314595957523e-06, "loss": 0.0063, "step": 69690 }, { "grad_norm": 0.12162676453590393, "learning_rate": 4.464773460476535e-06, "loss": 0.0063, "step": 69700 }, { "grad_norm": 0.16561932861804962, "learning_rate": 4.456240105720372e-06, "loss": 0.0084, "step": 69710 }, { "grad_norm": 0.050253093242645264, "learning_rate": 4.4477145331471405e-06, "loss": 0.0056, "step": 69720 }, { "grad_norm": 0.10997051000595093, "learning_rate": 4.439196744213653e-06, "loss": 0.0054, "step": 69730 }, { "grad_norm": 0.09566586464643478, "learning_rate": 4.430686740375339e-06, "loss": 0.0049, "step": 69740 }, { "grad_norm": 0.07022597640752792, "learning_rate": 4.422184523086342e-06, "loss": 0.0062, "step": 69750 }, { "grad_norm": 0.08808184415102005, "learning_rate": 4.41369009379946e-06, "loss": 0.0083, "step": 69760 }, { "grad_norm": 0.08100542426109314, "learning_rate": 4.405203453966139e-06, "loss": 0.0048, "step": 69770 }, { "grad_norm": 0.11562565714120865, "learning_rate": 4.396724605036539e-06, "loss": 0.0059, "step": 69780 }, { "grad_norm": 0.07920826226472855, "learning_rate": 4.388253548459437e-06, "loss": 0.0063, "step": 69790 }, { "grad_norm": 0.07929709553718567, "learning_rate": 4.3797902856823395e-06, "loss": 0.0062, "step": 69800 }, { "grad_norm": 0.07916606217622757, "learning_rate": 4.371334818151357e-06, "loss": 0.0049, "step": 69810 }, { "grad_norm": 0.10204948484897614, "learning_rate": 4.362887147311306e-06, "loss": 0.0067, "step": 69820 }, { "grad_norm": 0.10461822152137756, "learning_rate": 4.354447274605672e-06, "loss": 0.0055, "step": 69830 }, { "grad_norm": 0.10009944438934326, "learning_rate": 4.346015201476572e-06, "loss": 0.0067, "step": 69840 }, { "grad_norm": 0.1286870539188385, "learning_rate": 4.337590929364855e-06, "loss": 0.0047, "step": 69850 }, { "grad_norm": 0.10695795714855194, "learning_rate": 4.329174459709973e-06, "loss": 0.0066, "step": 69860 }, { "grad_norm": 0.11612372100353241, "learning_rate": 4.320765793950071e-06, "loss": 0.0059, "step": 69870 }, { "grad_norm": 0.10538390278816223, "learning_rate": 4.312364933521962e-06, "loss": 0.0044, "step": 69880 }, { "grad_norm": 0.2466292679309845, "learning_rate": 4.303971879861129e-06, "loss": 0.0065, "step": 69890 }, { "grad_norm": 0.13911312818527222, "learning_rate": 4.295586634401716e-06, "loss": 0.0061, "step": 69900 }, { "grad_norm": 0.12494786828756332, "learning_rate": 4.287209198576536e-06, "loss": 0.0045, "step": 69910 }, { "grad_norm": 0.0888662114739418, "learning_rate": 4.278839573817045e-06, "loss": 0.0056, "step": 69920 }, { "grad_norm": 0.09024092555046082, "learning_rate": 4.270477761553399e-06, "loss": 0.0067, "step": 69930 }, { "grad_norm": 0.05500852316617966, "learning_rate": 4.262123763214393e-06, "loss": 0.0056, "step": 69940 }, { "grad_norm": 0.07999961078166962, "learning_rate": 4.253777580227508e-06, "loss": 0.0075, "step": 69950 }, { "grad_norm": 0.09846070408821106, "learning_rate": 4.2454392140188755e-06, "loss": 0.0047, "step": 69960 }, { "grad_norm": 0.09723536670207977, "learning_rate": 4.2371086660132785e-06, "loss": 0.0054, "step": 69970 }, { "grad_norm": 0.24411700665950775, "learning_rate": 4.228785937634205e-06, "loss": 0.0053, "step": 69980 }, { "grad_norm": 0.17986783385276794, "learning_rate": 4.220471030303758e-06, "loss": 0.0078, "step": 69990 }, { "grad_norm": 0.11174017935991287, "learning_rate": 4.212163945442754e-06, "loss": 0.0055, "step": 70000 }, { "grad_norm": 0.14109665155410767, "learning_rate": 4.203864684470621e-06, "loss": 0.0055, "step": 70010 }, { "grad_norm": 0.07778338342905045, "learning_rate": 4.19557324880549e-06, "loss": 0.0055, "step": 70020 }, { "grad_norm": 0.13083146512508392, "learning_rate": 4.187289639864145e-06, "loss": 0.0065, "step": 70030 }, { "grad_norm": 0.20573829114437103, "learning_rate": 4.1790138590619974e-06, "loss": 0.0058, "step": 70040 }, { "grad_norm": 0.06082593649625778, "learning_rate": 4.170745907813195e-06, "loss": 0.0079, "step": 70050 }, { "grad_norm": 0.12601087987422943, "learning_rate": 4.162485787530479e-06, "loss": 0.0061, "step": 70060 }, { "grad_norm": 0.04754842445254326, "learning_rate": 4.154233499625282e-06, "loss": 0.0076, "step": 70070 }, { "grad_norm": 0.11828090250492096, "learning_rate": 4.145989045507692e-06, "loss": 0.0058, "step": 70080 }, { "grad_norm": 0.07787318527698517, "learning_rate": 4.1377524265864666e-06, "loss": 0.0059, "step": 70090 }, { "grad_norm": 0.10968350619077682, "learning_rate": 4.1295236442690175e-06, "loss": 0.0077, "step": 70100 }, { "grad_norm": 0.11536943167448044, "learning_rate": 4.121302699961421e-06, "loss": 0.0059, "step": 70110 }, { "grad_norm": 0.11220812052488327, "learning_rate": 4.113089595068403e-06, "loss": 0.006, "step": 70120 }, { "grad_norm": 0.10496360063552856, "learning_rate": 4.104884330993364e-06, "loss": 0.005, "step": 70130 }, { "grad_norm": 0.0718340054154396, "learning_rate": 4.0966869091383585e-06, "loss": 0.0048, "step": 70140 }, { "grad_norm": 0.06621269136667252, "learning_rate": 4.088497330904101e-06, "loss": 0.0052, "step": 70150 }, { "grad_norm": 0.04991472139954567, "learning_rate": 4.080315597689976e-06, "loss": 0.0055, "step": 70160 }, { "grad_norm": 0.11280010640621185, "learning_rate": 4.072141710893995e-06, "loss": 0.0053, "step": 70170 }, { "grad_norm": 0.050906602293252945, "learning_rate": 4.063975671912879e-06, "loss": 0.0049, "step": 70180 }, { "grad_norm": 0.13265393674373627, "learning_rate": 4.055817482141949e-06, "loss": 0.0061, "step": 70190 }, { "grad_norm": 0.1455838978290558, "learning_rate": 4.047667142975259e-06, "loss": 0.0054, "step": 70200 }, { "grad_norm": 0.12921804189682007, "learning_rate": 4.039524655805443e-06, "loss": 0.0059, "step": 70210 }, { "grad_norm": 0.16653572022914886, "learning_rate": 4.03139002202384e-06, "loss": 0.0059, "step": 70220 }, { "grad_norm": 0.14856167137622833, "learning_rate": 4.023263243020447e-06, "loss": 0.0072, "step": 70230 }, { "grad_norm": 0.06946595013141632, "learning_rate": 4.015144320183884e-06, "loss": 0.0059, "step": 70240 }, { "grad_norm": 0.06791620701551437, "learning_rate": 4.007033254901482e-06, "loss": 0.0062, "step": 70250 }, { "grad_norm": 0.08136873692274094, "learning_rate": 3.9989300485591795e-06, "loss": 0.0056, "step": 70260 }, { "grad_norm": 0.0892406702041626, "learning_rate": 3.990834702541601e-06, "loss": 0.0053, "step": 70270 }, { "grad_norm": 0.13447795808315277, "learning_rate": 3.982747218232019e-06, "loss": 0.0049, "step": 70280 }, { "grad_norm": 0.13668516278266907, "learning_rate": 3.97466759701236e-06, "loss": 0.0057, "step": 70290 }, { "grad_norm": 0.08602000772953033, "learning_rate": 3.966595840263226e-06, "loss": 0.0051, "step": 70300 }, { "grad_norm": 0.11333537101745605, "learning_rate": 3.958531949363836e-06, "loss": 0.0057, "step": 70310 }, { "grad_norm": 0.09779413044452667, "learning_rate": 3.950475925692098e-06, "loss": 0.0062, "step": 70320 }, { "grad_norm": 0.0764365866780281, "learning_rate": 3.9424277706245685e-06, "loss": 0.0045, "step": 70330 }, { "grad_norm": 0.11158131808042526, "learning_rate": 3.934387485536451e-06, "loss": 0.0046, "step": 70340 }, { "grad_norm": 0.09651098400354385, "learning_rate": 3.926355071801619e-06, "loss": 0.0055, "step": 70350 }, { "grad_norm": 0.10994335263967514, "learning_rate": 3.9183305307925965e-06, "loss": 0.0051, "step": 70360 }, { "grad_norm": 0.12153644114732742, "learning_rate": 3.910313863880533e-06, "loss": 0.005, "step": 70370 }, { "grad_norm": 0.09422636777162552, "learning_rate": 3.902305072435292e-06, "loss": 0.0058, "step": 70380 }, { "grad_norm": 0.14287249743938446, "learning_rate": 3.894304157825329e-06, "loss": 0.006, "step": 70390 }, { "grad_norm": 0.04821208491921425, "learning_rate": 3.886311121417791e-06, "loss": 0.0048, "step": 70400 }, { "grad_norm": 0.11357258260250092, "learning_rate": 3.878325964578472e-06, "loss": 0.0069, "step": 70410 }, { "grad_norm": 0.10606390982866287, "learning_rate": 3.870348688671815e-06, "loss": 0.0054, "step": 70420 }, { "grad_norm": 0.14346826076507568, "learning_rate": 3.862379295060931e-06, "loss": 0.0056, "step": 70430 }, { "grad_norm": 0.1825941801071167, "learning_rate": 3.854417785107534e-06, "loss": 0.0066, "step": 70440 }, { "grad_norm": 0.08391599357128143, "learning_rate": 3.8464641601720755e-06, "loss": 0.0046, "step": 70450 }, { "grad_norm": 0.10291021317243576, "learning_rate": 3.838518421613579e-06, "loss": 0.0049, "step": 70460 }, { "grad_norm": 0.12864169478416443, "learning_rate": 3.830580570789766e-06, "loss": 0.0067, "step": 70470 }, { "grad_norm": 0.0812043622136116, "learning_rate": 3.822650609057005e-06, "loss": 0.005, "step": 70480 }, { "grad_norm": 0.08297043293714523, "learning_rate": 3.814728537770285e-06, "loss": 0.0067, "step": 70490 }, { "grad_norm": 0.06444323807954788, "learning_rate": 3.806814358283306e-06, "loss": 0.0063, "step": 70500 }, { "grad_norm": 0.06258975714445114, "learning_rate": 3.7989080719483596e-06, "loss": 0.0071, "step": 70510 }, { "grad_norm": 0.09279119968414307, "learning_rate": 3.7910096801164143e-06, "loss": 0.0064, "step": 70520 }, { "grad_norm": 0.0752415657043457, "learning_rate": 3.7831191841371016e-06, "loss": 0.0048, "step": 70530 }, { "grad_norm": 0.092196024954319, "learning_rate": 3.775236585358688e-06, "loss": 0.0058, "step": 70540 }, { "grad_norm": 0.1196640208363533, "learning_rate": 3.7673618851280843e-06, "loss": 0.0061, "step": 70550 }, { "grad_norm": 0.07659509032964706, "learning_rate": 3.759495084790887e-06, "loss": 0.0056, "step": 70560 }, { "grad_norm": 0.07723543792963028, "learning_rate": 3.751636185691282e-06, "loss": 0.0062, "step": 70570 }, { "grad_norm": 0.05855610966682434, "learning_rate": 3.7437851891721607e-06, "loss": 0.0045, "step": 70580 }, { "grad_norm": 0.10288385301828384, "learning_rate": 3.7359420965750404e-06, "loss": 0.006, "step": 70590 }, { "grad_norm": 0.07723471522331238, "learning_rate": 3.7281069092400922e-06, "loss": 0.0073, "step": 70600 }, { "grad_norm": 0.11747856438159943, "learning_rate": 3.7202796285061348e-06, "loss": 0.0048, "step": 70610 }, { "grad_norm": 0.10137812048196793, "learning_rate": 3.712460255710637e-06, "loss": 0.0047, "step": 70620 }, { "grad_norm": 0.10276234894990921, "learning_rate": 3.704648792189719e-06, "loss": 0.0056, "step": 70630 }, { "grad_norm": 0.0970131978392601, "learning_rate": 3.696845239278124e-06, "loss": 0.0047, "step": 70640 }, { "grad_norm": 0.10201660543680191, "learning_rate": 3.689049598309302e-06, "loss": 0.0059, "step": 70650 }, { "grad_norm": 0.08346179127693176, "learning_rate": 3.681261870615288e-06, "loss": 0.0066, "step": 70660 }, { "grad_norm": 0.13812236487865448, "learning_rate": 3.6734820575268004e-06, "loss": 0.0063, "step": 70670 }, { "grad_norm": 0.19921424984931946, "learning_rate": 3.665710160373209e-06, "loss": 0.0056, "step": 70680 }, { "grad_norm": 0.12958167493343353, "learning_rate": 3.65794618048248e-06, "loss": 0.0069, "step": 70690 }, { "grad_norm": 0.0896218866109848, "learning_rate": 3.6501901191813125e-06, "loss": 0.0074, "step": 70700 }, { "grad_norm": 0.15768171846866608, "learning_rate": 3.642441977794975e-06, "loss": 0.0061, "step": 70710 }, { "grad_norm": 0.12646375596523285, "learning_rate": 3.634701757647424e-06, "loss": 0.0067, "step": 70720 }, { "grad_norm": 0.12683390080928802, "learning_rate": 3.6269694600612468e-06, "loss": 0.0087, "step": 70730 }, { "grad_norm": 0.07595036178827286, "learning_rate": 3.619245086357681e-06, "loss": 0.0056, "step": 70740 }, { "grad_norm": 0.10465666651725769, "learning_rate": 3.611528637856615e-06, "loss": 0.0058, "step": 70750 }, { "grad_norm": 0.1511157900094986, "learning_rate": 3.6038201158765884e-06, "loss": 0.008, "step": 70760 }, { "grad_norm": 0.050116151571273804, "learning_rate": 3.5961195217347534e-06, "loss": 0.0046, "step": 70770 }, { "grad_norm": 0.06548141688108444, "learning_rate": 3.588426856746946e-06, "loss": 0.0051, "step": 70780 }, { "grad_norm": 0.2079913318157196, "learning_rate": 3.5807421222276316e-06, "loss": 0.0063, "step": 70790 }, { "grad_norm": 0.12492455542087555, "learning_rate": 3.5730653194899156e-06, "loss": 0.0064, "step": 70800 }, { "grad_norm": 0.142196387052536, "learning_rate": 3.565396449845554e-06, "loss": 0.0075, "step": 70810 }, { "grad_norm": 0.10415945947170258, "learning_rate": 3.557735514604954e-06, "loss": 0.0055, "step": 70820 }, { "grad_norm": 0.11846822500228882, "learning_rate": 3.5500825150771633e-06, "loss": 0.0053, "step": 70830 }, { "grad_norm": 0.2093353122472763, "learning_rate": 3.5424374525698466e-06, "loss": 0.0056, "step": 70840 }, { "grad_norm": 0.10244611650705338, "learning_rate": 3.5348003283893704e-06, "loss": 0.0053, "step": 70850 }, { "grad_norm": 0.1362837553024292, "learning_rate": 3.527171143840685e-06, "loss": 0.0076, "step": 70860 }, { "grad_norm": 0.08340969681739807, "learning_rate": 3.519549900227409e-06, "loss": 0.0057, "step": 70870 }, { "grad_norm": 0.07210836559534073, "learning_rate": 3.511936598851828e-06, "loss": 0.0061, "step": 70880 }, { "grad_norm": 0.06689418107271194, "learning_rate": 3.504331241014813e-06, "loss": 0.0049, "step": 70890 }, { "grad_norm": 0.07733846455812454, "learning_rate": 3.4967338280159414e-06, "loss": 0.0063, "step": 70900 }, { "grad_norm": 0.11847623437643051, "learning_rate": 3.4891443611533846e-06, "loss": 0.0051, "step": 70910 }, { "grad_norm": 0.10620979219675064, "learning_rate": 3.481562841723984e-06, "loss": 0.0051, "step": 70920 }, { "grad_norm": 0.07039748132228851, "learning_rate": 3.473989271023215e-06, "loss": 0.0064, "step": 70930 }, { "grad_norm": 0.04965681955218315, "learning_rate": 3.4664236503451854e-06, "loss": 0.007, "step": 70940 }, { "grad_norm": 0.10082518309354782, "learning_rate": 3.4588659809826674e-06, "loss": 0.0047, "step": 70950 }, { "grad_norm": 0.08709461241960526, "learning_rate": 3.451316264227039e-06, "loss": 0.0054, "step": 70960 }, { "grad_norm": 0.0633816123008728, "learning_rate": 3.4437745013683574e-06, "loss": 0.0056, "step": 70970 }, { "grad_norm": 0.06427665054798126, "learning_rate": 3.4362406936952916e-06, "loss": 0.0052, "step": 70980 }, { "grad_norm": 0.07652031630277634, "learning_rate": 3.4287148424951786e-06, "loss": 0.0044, "step": 70990 }, { "grad_norm": 0.08310063928365707, "learning_rate": 3.4211969490539617e-06, "loss": 0.0055, "step": 71000 }, { "grad_norm": 0.12826742231845856, "learning_rate": 3.4136870146562584e-06, "loss": 0.005, "step": 71010 }, { "grad_norm": 0.1602802276611328, "learning_rate": 3.406185040585308e-06, "loss": 0.0059, "step": 71020 }, { "grad_norm": 0.09592320024967194, "learning_rate": 3.3986910281229966e-06, "loss": 0.0057, "step": 71030 }, { "grad_norm": 0.07095733284950256, "learning_rate": 3.391204978549828e-06, "loss": 0.0065, "step": 71040 }, { "grad_norm": 0.12899352610111237, "learning_rate": 3.3837268931449785e-06, "loss": 0.006, "step": 71050 }, { "grad_norm": 0.09289135783910751, "learning_rate": 3.376256773186248e-06, "loss": 0.0054, "step": 71060 }, { "grad_norm": 0.06056448817253113, "learning_rate": 3.3687946199500664e-06, "loss": 0.0056, "step": 71070 }, { "grad_norm": 0.14863941073417664, "learning_rate": 3.3613404347115295e-06, "loss": 0.0063, "step": 71080 }, { "grad_norm": 0.10703592747449875, "learning_rate": 3.35389421874433e-06, "loss": 0.0049, "step": 71090 }, { "grad_norm": 0.12473031133413315, "learning_rate": 3.3464559733208446e-06, "loss": 0.0056, "step": 71100 }, { "grad_norm": 0.056665822863578796, "learning_rate": 3.3390256997120505e-06, "loss": 0.0057, "step": 71110 }, { "grad_norm": 0.14238670468330383, "learning_rate": 3.331603399187583e-06, "loss": 0.0052, "step": 71120 }, { "grad_norm": 0.08065049350261688, "learning_rate": 3.324189073015721e-06, "loss": 0.0042, "step": 71130 }, { "grad_norm": 0.1597290337085724, "learning_rate": 3.3167827224633465e-06, "loss": 0.0061, "step": 71140 }, { "grad_norm": 0.10551304370164871, "learning_rate": 3.309384348796024e-06, "loss": 0.007, "step": 71150 }, { "grad_norm": 0.09747893363237381, "learning_rate": 3.301993953277921e-06, "loss": 0.0051, "step": 71160 }, { "grad_norm": 0.1561765968799591, "learning_rate": 3.29461153717186e-06, "loss": 0.0068, "step": 71170 }, { "grad_norm": 0.09073055535554886, "learning_rate": 3.287237101739293e-06, "loss": 0.0053, "step": 71180 }, { "grad_norm": 0.19464844465255737, "learning_rate": 3.2798706482403075e-06, "loss": 0.0066, "step": 71190 }, { "grad_norm": 0.0915236622095108, "learning_rate": 3.2725121779336285e-06, "loss": 0.0051, "step": 71200 }, { "grad_norm": 0.07110033929347992, "learning_rate": 3.265161692076618e-06, "loss": 0.0048, "step": 71210 }, { "grad_norm": 0.1309632956981659, "learning_rate": 3.257819191925282e-06, "loss": 0.0058, "step": 71220 }, { "grad_norm": 0.09061887115240097, "learning_rate": 3.2504846787342392e-06, "loss": 0.0052, "step": 71230 }, { "grad_norm": 0.07759381830692291, "learning_rate": 3.243158153756759e-06, "loss": 0.0071, "step": 71240 }, { "grad_norm": 0.08678951859474182, "learning_rate": 3.2358396182447514e-06, "loss": 0.0044, "step": 71250 }, { "grad_norm": 0.14333254098892212, "learning_rate": 3.2285290734487496e-06, "loss": 0.0053, "step": 71260 }, { "grad_norm": 0.07253151386976242, "learning_rate": 3.221226520617926e-06, "loss": 0.006, "step": 71270 }, { "grad_norm": 0.11455698311328888, "learning_rate": 3.213931961000094e-06, "loss": 0.0045, "step": 71280 }, { "grad_norm": 0.12347786873579025, "learning_rate": 3.2066453958416733e-06, "loss": 0.0058, "step": 71290 }, { "grad_norm": 0.10834459215402603, "learning_rate": 3.199366826387773e-06, "loss": 0.0053, "step": 71300 }, { "grad_norm": 0.17672812938690186, "learning_rate": 3.192096253882071e-06, "loss": 0.0051, "step": 71310 }, { "grad_norm": 0.11429213732481003, "learning_rate": 3.1848336795669177e-06, "loss": 0.0051, "step": 71320 }, { "grad_norm": 0.15953053534030914, "learning_rate": 3.1775791046833035e-06, "loss": 0.0066, "step": 71330 }, { "grad_norm": 0.08616693317890167, "learning_rate": 3.170332530470804e-06, "loss": 0.0059, "step": 71340 }, { "grad_norm": 0.10376930981874466, "learning_rate": 3.1630939581677012e-06, "loss": 0.0062, "step": 71350 }, { "grad_norm": 0.22537901997566223, "learning_rate": 3.155863389010838e-06, "loss": 0.0058, "step": 71360 }, { "grad_norm": 0.13160286843776703, "learning_rate": 3.1486408242357323e-06, "loss": 0.0065, "step": 71370 }, { "grad_norm": 0.08321331441402435, "learning_rate": 3.1414262650765248e-06, "loss": 0.0061, "step": 71380 }, { "grad_norm": 0.07795413583517075, "learning_rate": 3.134219712765979e-06, "loss": 0.0052, "step": 71390 }, { "grad_norm": 0.23283693194389343, "learning_rate": 3.1270211685355044e-06, "loss": 0.0057, "step": 71400 }, { "grad_norm": 0.1121383085846901, "learning_rate": 3.1198306336151338e-06, "loss": 0.0075, "step": 71410 }, { "grad_norm": 0.10386469215154648, "learning_rate": 3.11264810923354e-06, "loss": 0.0053, "step": 71420 }, { "grad_norm": 0.10051410645246506, "learning_rate": 3.105473596618008e-06, "loss": 0.0052, "step": 71430 }, { "grad_norm": 0.10209038108587265, "learning_rate": 3.0983070969944683e-06, "loss": 0.0041, "step": 71440 }, { "grad_norm": 0.08708439022302628, "learning_rate": 3.09114861158748e-06, "loss": 0.0054, "step": 71450 }, { "grad_norm": 0.11694662272930145, "learning_rate": 3.0839981416202314e-06, "loss": 0.0047, "step": 71460 }, { "grad_norm": 0.054031744599342346, "learning_rate": 3.0768556883145507e-06, "loss": 0.0054, "step": 71470 }, { "grad_norm": 0.062113843858242035, "learning_rate": 3.0697212528908834e-06, "loss": 0.0057, "step": 71480 }, { "grad_norm": 0.07607146352529526, "learning_rate": 3.0625948365683e-06, "loss": 0.0061, "step": 71490 }, { "grad_norm": 0.0503055639564991, "learning_rate": 3.055476440564531e-06, "loss": 0.0051, "step": 71500 }, { "grad_norm": 0.12243430316448212, "learning_rate": 3.0483660660958924e-06, "loss": 0.0062, "step": 71510 }, { "grad_norm": 0.11216774582862854, "learning_rate": 3.0412637143773624e-06, "loss": 0.0068, "step": 71520 }, { "grad_norm": 0.08213809877634048, "learning_rate": 3.034169386622554e-06, "loss": 0.0045, "step": 71530 }, { "grad_norm": 0.1417425125837326, "learning_rate": 3.027083084043658e-06, "loss": 0.0055, "step": 71540 }, { "grad_norm": 0.12946264445781708, "learning_rate": 3.0200048078515676e-06, "loss": 0.0068, "step": 71550 }, { "grad_norm": 0.12178407609462738, "learning_rate": 3.012934559255742e-06, "loss": 0.0077, "step": 71560 }, { "grad_norm": 0.10275059193372726, "learning_rate": 3.0058723394642994e-06, "loss": 0.005, "step": 71570 }, { "grad_norm": 0.14347757399082184, "learning_rate": 2.998818149683985e-06, "loss": 0.0052, "step": 71580 }, { "grad_norm": 0.10256267338991165, "learning_rate": 2.9917719911201627e-06, "loss": 0.0066, "step": 71590 }, { "grad_norm": 0.08665009588003159, "learning_rate": 2.9847338649768352e-06, "loss": 0.0062, "step": 71600 }, { "grad_norm": 0.13521650433540344, "learning_rate": 2.977703772456608e-06, "loss": 0.0056, "step": 71610 }, { "grad_norm": 0.06468693166971207, "learning_rate": 2.9706817147607535e-06, "loss": 0.0053, "step": 71620 }, { "grad_norm": 0.0813271552324295, "learning_rate": 2.9636676930891393e-06, "loss": 0.0065, "step": 71630 }, { "grad_norm": 0.14400914311408997, "learning_rate": 2.9566617086402625e-06, "loss": 0.0047, "step": 71640 }, { "grad_norm": 0.09646604210138321, "learning_rate": 2.9496637626112655e-06, "loss": 0.0053, "step": 71650 }, { "grad_norm": 0.07990887761116028, "learning_rate": 2.9426738561979027e-06, "loss": 0.0065, "step": 71660 }, { "grad_norm": 0.10921602696180344, "learning_rate": 2.9356919905945524e-06, "loss": 0.007, "step": 71670 }, { "grad_norm": 0.08001745492219925, "learning_rate": 2.928718166994243e-06, "loss": 0.0048, "step": 71680 }, { "grad_norm": 0.20871615409851074, "learning_rate": 2.9217523865885833e-06, "loss": 0.0067, "step": 71690 }, { "grad_norm": 0.13012181222438812, "learning_rate": 2.9147946505678537e-06, "loss": 0.0063, "step": 71700 }, { "grad_norm": 0.19164611399173737, "learning_rate": 2.9078449601209313e-06, "loss": 0.0049, "step": 71710 }, { "grad_norm": 0.10498059540987015, "learning_rate": 2.900903316435333e-06, "loss": 0.0053, "step": 71720 }, { "grad_norm": 0.10643946379423141, "learning_rate": 2.8939697206971983e-06, "loss": 0.0073, "step": 71730 }, { "grad_norm": 0.13802677392959595, "learning_rate": 2.8870441740912746e-06, "loss": 0.0057, "step": 71740 }, { "grad_norm": 0.06626691669225693, "learning_rate": 2.8801266778009762e-06, "loss": 0.0046, "step": 71750 }, { "grad_norm": 0.07258986681699753, "learning_rate": 2.8732172330082798e-06, "loss": 0.0041, "step": 71760 }, { "grad_norm": 0.07669583708047867, "learning_rate": 2.8663158408938517e-06, "loss": 0.0053, "step": 71770 }, { "grad_norm": 0.0946287214756012, "learning_rate": 2.859422502636938e-06, "loss": 0.0053, "step": 71780 }, { "grad_norm": 0.09042541682720184, "learning_rate": 2.8525372194154076e-06, "loss": 0.006, "step": 71790 }, { "grad_norm": 0.08196108788251877, "learning_rate": 2.8456599924057913e-06, "loss": 0.0062, "step": 71800 }, { "grad_norm": 0.09885133802890778, "learning_rate": 2.8387908227831995e-06, "loss": 0.006, "step": 71810 }, { "grad_norm": 0.13484209775924683, "learning_rate": 2.831929711721404e-06, "loss": 0.0078, "step": 71820 }, { "grad_norm": 0.1149747371673584, "learning_rate": 2.825076660392767e-06, "loss": 0.0058, "step": 71830 }, { "grad_norm": 0.08282430469989777, "learning_rate": 2.8182316699682908e-06, "loss": 0.0051, "step": 71840 }, { "grad_norm": 0.07705481350421906, "learning_rate": 2.811394741617601e-06, "loss": 0.0043, "step": 71850 }, { "grad_norm": 0.19086246192455292, "learning_rate": 2.8045658765089356e-06, "loss": 0.0068, "step": 71860 }, { "grad_norm": 0.13071508705615997, "learning_rate": 2.7977450758091605e-06, "loss": 0.0055, "step": 71870 }, { "grad_norm": 0.06330659240484238, "learning_rate": 2.790932340683783e-06, "loss": 0.0069, "step": 71880 }, { "grad_norm": 0.07510199397802353, "learning_rate": 2.7841276722968823e-06, "loss": 0.0056, "step": 71890 }, { "grad_norm": 0.11031391471624374, "learning_rate": 2.7773310718112067e-06, "loss": 0.0051, "step": 71900 }, { "grad_norm": 0.047706834971904755, "learning_rate": 2.7705425403881102e-06, "loss": 0.0046, "step": 71910 }, { "grad_norm": 0.05523642525076866, "learning_rate": 2.763762079187565e-06, "loss": 0.0047, "step": 71920 }, { "grad_norm": 0.16307711601257324, "learning_rate": 2.756989689368178e-06, "loss": 0.0051, "step": 71930 }, { "grad_norm": 0.11869834363460541, "learning_rate": 2.750225372087134e-06, "loss": 0.0047, "step": 71940 }, { "grad_norm": 0.13473138213157654, "learning_rate": 2.7434691285003033e-06, "loss": 0.0074, "step": 71950 }, { "grad_norm": 0.07861053198575974, "learning_rate": 2.7367209597621178e-06, "loss": 0.0065, "step": 71960 }, { "grad_norm": 0.07965363562107086, "learning_rate": 2.729980867025683e-06, "loss": 0.0049, "step": 71970 }, { "grad_norm": 0.059939198195934296, "learning_rate": 2.7232488514426724e-06, "loss": 0.0043, "step": 71980 }, { "grad_norm": 0.0752192959189415, "learning_rate": 2.7165249141634097e-06, "loss": 0.0088, "step": 71990 }, { "grad_norm": 0.06332498043775558, "learning_rate": 2.709809056336837e-06, "loss": 0.0071, "step": 72000 }, { "grad_norm": 0.13112348318099976, "learning_rate": 2.703101279110498e-06, "loss": 0.0067, "step": 72010 }, { "grad_norm": 0.11903338134288788, "learning_rate": 2.696401583630587e-06, "loss": 0.0053, "step": 72020 }, { "grad_norm": 0.0678485706448555, "learning_rate": 2.689709971041887e-06, "loss": 0.0051, "step": 72030 }, { "grad_norm": 0.09143077582120895, "learning_rate": 2.6830264424878126e-06, "loss": 0.0064, "step": 72040 }, { "grad_norm": 0.11157674342393875, "learning_rate": 2.676350999110394e-06, "loss": 0.0056, "step": 72050 }, { "grad_norm": 0.08160211145877838, "learning_rate": 2.6696836420502856e-06, "loss": 0.0068, "step": 72060 }, { "grad_norm": 0.07596014440059662, "learning_rate": 2.663024372446765e-06, "loss": 0.0062, "step": 72070 }, { "grad_norm": 0.08422856032848358, "learning_rate": 2.6563731914377055e-06, "loss": 0.0066, "step": 72080 }, { "grad_norm": 0.0795002207159996, "learning_rate": 2.6497301001596087e-06, "loss": 0.0061, "step": 72090 }, { "grad_norm": 0.10358788073062897, "learning_rate": 2.643095099747611e-06, "loss": 0.0061, "step": 72100 }, { "grad_norm": 0.09063420444726944, "learning_rate": 2.636468191335445e-06, "loss": 0.0057, "step": 72110 }, { "grad_norm": 0.09861262142658234, "learning_rate": 2.629849376055471e-06, "loss": 0.0044, "step": 72120 }, { "grad_norm": 0.08536950498819351, "learning_rate": 2.6232386550386678e-06, "loss": 0.005, "step": 72130 }, { "grad_norm": 0.08828376233577728, "learning_rate": 2.6166360294146097e-06, "loss": 0.0052, "step": 72140 }, { "grad_norm": 0.10423818230628967, "learning_rate": 2.6100415003115275e-06, "loss": 0.0067, "step": 72150 }, { "grad_norm": 0.06808558851480484, "learning_rate": 2.603455068856225e-06, "loss": 0.0045, "step": 72160 }, { "grad_norm": 0.06638096272945404, "learning_rate": 2.5968767361741584e-06, "loss": 0.0066, "step": 72170 }, { "grad_norm": 0.14865337312221527, "learning_rate": 2.590306503389378e-06, "loss": 0.0053, "step": 72180 }, { "grad_norm": 0.07919252663850784, "learning_rate": 2.583744371624558e-06, "loss": 0.0052, "step": 72190 }, { "grad_norm": 0.12386642396450043, "learning_rate": 2.57719034200099e-06, "loss": 0.0045, "step": 72200 }, { "grad_norm": 0.12454010546207428, "learning_rate": 2.570644415638568e-06, "loss": 0.006, "step": 72210 }, { "grad_norm": 0.08077952265739441, "learning_rate": 2.5641065936558295e-06, "loss": 0.0078, "step": 72220 }, { "grad_norm": 0.12454590946435928, "learning_rate": 2.557576877169898e-06, "loss": 0.0071, "step": 72230 }, { "grad_norm": 0.11492366343736649, "learning_rate": 2.5510552672965205e-06, "loss": 0.005, "step": 72240 }, { "grad_norm": 0.09193979203701019, "learning_rate": 2.544541765150077e-06, "loss": 0.0061, "step": 72250 }, { "grad_norm": 0.11783935874700546, "learning_rate": 2.5380363718435165e-06, "loss": 0.0037, "step": 72260 }, { "grad_norm": 0.11907900869846344, "learning_rate": 2.5315390884884714e-06, "loss": 0.0066, "step": 72270 }, { "grad_norm": 0.0928320586681366, "learning_rate": 2.5250499161951214e-06, "loss": 0.0042, "step": 72280 }, { "grad_norm": 0.09046527743339539, "learning_rate": 2.518568856072301e-06, "loss": 0.0057, "step": 72290 }, { "grad_norm": 0.08078266680240631, "learning_rate": 2.512095909227441e-06, "loss": 0.0041, "step": 72300 }, { "grad_norm": 0.09973569959402084, "learning_rate": 2.50563107676659e-06, "loss": 0.0058, "step": 72310 }, { "grad_norm": 0.08650730550289154, "learning_rate": 2.4991743597944152e-06, "loss": 0.0048, "step": 72320 }, { "grad_norm": 0.12274691462516785, "learning_rate": 2.492725759414205e-06, "loss": 0.0051, "step": 72330 }, { "grad_norm": 0.155926913022995, "learning_rate": 2.4862852767278234e-06, "loss": 0.0075, "step": 72340 }, { "grad_norm": 0.1502123922109604, "learning_rate": 2.479852912835784e-06, "loss": 0.0078, "step": 72350 }, { "grad_norm": 0.1515466868877411, "learning_rate": 2.4734286688372075e-06, "loss": 0.0069, "step": 72360 }, { "grad_norm": 0.05932176485657692, "learning_rate": 2.467012545829811e-06, "loss": 0.0037, "step": 72370 }, { "grad_norm": 0.05043158307671547, "learning_rate": 2.460604544909945e-06, "loss": 0.0056, "step": 72380 }, { "grad_norm": 0.1351238340139389, "learning_rate": 2.454204667172555e-06, "loss": 0.0043, "step": 72390 }, { "grad_norm": 0.08070078492164612, "learning_rate": 2.447812913711217e-06, "loss": 0.0056, "step": 72400 }, { "grad_norm": 0.1689220815896988, "learning_rate": 2.44142928561808e-06, "loss": 0.0063, "step": 72410 }, { "grad_norm": 0.1561882197856903, "learning_rate": 2.435053783983965e-06, "loss": 0.0053, "step": 72420 }, { "grad_norm": 0.06302369385957718, "learning_rate": 2.4286864098982453e-06, "loss": 0.0047, "step": 72430 }, { "grad_norm": 0.08478536456823349, "learning_rate": 2.4223271644489397e-06, "loss": 0.005, "step": 72440 }, { "grad_norm": 0.12523460388183594, "learning_rate": 2.415976048722679e-06, "loss": 0.0055, "step": 72450 }, { "grad_norm": 0.1064072847366333, "learning_rate": 2.4096330638046673e-06, "loss": 0.0068, "step": 72460 }, { "grad_norm": 0.10283811390399933, "learning_rate": 2.4032982107787816e-06, "loss": 0.0056, "step": 72470 }, { "grad_norm": 0.09009476751089096, "learning_rate": 2.396971490727451e-06, "loss": 0.0069, "step": 72480 }, { "grad_norm": 0.12719793617725372, "learning_rate": 2.3906529047317493e-06, "loss": 0.0054, "step": 72490 }, { "grad_norm": 0.16087761521339417, "learning_rate": 2.3843424538713465e-06, "loss": 0.0064, "step": 72500 }, { "grad_norm": 0.09792747348546982, "learning_rate": 2.3780401392245298e-06, "loss": 0.0053, "step": 72510 }, { "grad_norm": 0.06233082339167595, "learning_rate": 2.3717459618681883e-06, "loss": 0.0055, "step": 72520 }, { "grad_norm": 0.1184554174542427, "learning_rate": 2.3654599228778274e-06, "loss": 0.0064, "step": 72530 }, { "grad_norm": 0.10664477944374084, "learning_rate": 2.3591820233275607e-06, "loss": 0.006, "step": 72540 }, { "grad_norm": 0.1723613291978836, "learning_rate": 2.3529122642901024e-06, "loss": 0.0083, "step": 72550 }, { "grad_norm": 0.16618794202804565, "learning_rate": 2.3466506468367845e-06, "loss": 0.005, "step": 72560 }, { "grad_norm": 0.17077408730983734, "learning_rate": 2.3403971720375506e-06, "loss": 0.0065, "step": 72570 }, { "grad_norm": 0.08494328707456589, "learning_rate": 2.334151840960952e-06, "loss": 0.0056, "step": 72580 }, { "grad_norm": 0.06513206660747528, "learning_rate": 2.3279146546741347e-06, "loss": 0.0071, "step": 72590 }, { "grad_norm": 0.09176542609930038, "learning_rate": 2.32168561424288e-06, "loss": 0.0074, "step": 72600 }, { "grad_norm": 0.11885417252779007, "learning_rate": 2.3154647207315307e-06, "loss": 0.0053, "step": 72610 }, { "grad_norm": 0.11439889669418335, "learning_rate": 2.309251975203103e-06, "loss": 0.0058, "step": 72620 }, { "grad_norm": 0.07325771450996399, "learning_rate": 2.303047378719159e-06, "loss": 0.0068, "step": 72630 }, { "grad_norm": 0.058332063257694244, "learning_rate": 2.2968509323399e-06, "loss": 0.005, "step": 72640 }, { "grad_norm": 0.10513363033533096, "learning_rate": 2.290662637124147e-06, "loss": 0.0056, "step": 72650 }, { "grad_norm": 0.10386431962251663, "learning_rate": 2.2844824941292807e-06, "loss": 0.0076, "step": 72660 }, { "grad_norm": 0.1165202409029007, "learning_rate": 2.2783105044113406e-06, "loss": 0.0053, "step": 72670 }, { "grad_norm": 0.14746710658073425, "learning_rate": 2.272146669024944e-06, "loss": 0.0066, "step": 72680 }, { "grad_norm": 0.0847557783126831, "learning_rate": 2.26599098902332e-06, "loss": 0.007, "step": 72690 }, { "grad_norm": 0.12122275680303574, "learning_rate": 2.2598434654583113e-06, "loss": 0.0064, "step": 72700 }, { "grad_norm": 0.08992850035429001, "learning_rate": 2.253704099380355e-06, "loss": 0.005, "step": 72710 }, { "grad_norm": 0.04838186502456665, "learning_rate": 2.247572891838512e-06, "loss": 0.0049, "step": 72720 }, { "grad_norm": 0.08319731801748276, "learning_rate": 2.241449843880422e-06, "loss": 0.006, "step": 72730 }, { "grad_norm": 0.11625447869300842, "learning_rate": 2.235334956552354e-06, "loss": 0.0093, "step": 72740 }, { "grad_norm": 0.16059467196464539, "learning_rate": 2.2292282308991775e-06, "loss": 0.0062, "step": 72750 }, { "grad_norm": 0.07375587522983551, "learning_rate": 2.223129667964363e-06, "loss": 0.0064, "step": 72760 }, { "grad_norm": 0.1505623459815979, "learning_rate": 2.2170392687899834e-06, "loss": 0.0057, "step": 72770 }, { "grad_norm": 0.11494939029216766, "learning_rate": 2.210957034416733e-06, "loss": 0.0087, "step": 72780 }, { "grad_norm": 0.13673198223114014, "learning_rate": 2.2048829658838867e-06, "loss": 0.0068, "step": 72790 }, { "grad_norm": 0.12183809280395508, "learning_rate": 2.1988170642293525e-06, "loss": 0.0067, "step": 72800 }, { "grad_norm": 0.17790955305099487, "learning_rate": 2.1927593304896075e-06, "loss": 0.0062, "step": 72810 }, { "grad_norm": 0.09855537861585617, "learning_rate": 2.1867097656997626e-06, "loss": 0.0048, "step": 72820 }, { "grad_norm": 0.11107858270406723, "learning_rate": 2.180668370893524e-06, "loss": 0.0064, "step": 72830 }, { "grad_norm": 0.1395488977432251, "learning_rate": 2.174635147103199e-06, "loss": 0.0061, "step": 72840 }, { "grad_norm": 0.15901263058185577, "learning_rate": 2.1686100953597075e-06, "loss": 0.0057, "step": 72850 }, { "grad_norm": 0.0461711585521698, "learning_rate": 2.1625932166925433e-06, "loss": 0.0051, "step": 72860 }, { "grad_norm": 0.11894496530294418, "learning_rate": 2.1565845121298556e-06, "loss": 0.0061, "step": 72870 }, { "grad_norm": 0.15863336622714996, "learning_rate": 2.150583982698351e-06, "loss": 0.0058, "step": 72880 }, { "grad_norm": 0.15244744718074799, "learning_rate": 2.144591629423359e-06, "loss": 0.0054, "step": 72890 }, { "grad_norm": 0.08900335431098938, "learning_rate": 2.138607453328817e-06, "loss": 0.0047, "step": 72900 }, { "grad_norm": 0.05331018567085266, "learning_rate": 2.132631455437234e-06, "loss": 0.0056, "step": 72910 }, { "grad_norm": 0.14285831153392792, "learning_rate": 2.1266636367697714e-06, "loss": 0.0051, "step": 72920 }, { "grad_norm": 0.09791738539934158, "learning_rate": 2.120703998346152e-06, "loss": 0.0061, "step": 72930 }, { "grad_norm": 0.0721726194024086, "learning_rate": 2.1147525411847114e-06, "loss": 0.0033, "step": 72940 }, { "grad_norm": 0.11701597273349762, "learning_rate": 2.108809266302403e-06, "loss": 0.0063, "step": 72950 }, { "grad_norm": 0.10989538580179214, "learning_rate": 2.102874174714764e-06, "loss": 0.0053, "step": 72960 }, { "grad_norm": 0.11323146522045135, "learning_rate": 2.0969472674359393e-06, "loss": 0.008, "step": 72970 }, { "grad_norm": 0.09349962323904037, "learning_rate": 2.091028545478668e-06, "loss": 0.0067, "step": 72980 }, { "grad_norm": 0.08411692827939987, "learning_rate": 2.08511800985432e-06, "loss": 0.0044, "step": 72990 }, { "grad_norm": 0.18207380175590515, "learning_rate": 2.07921566157282e-06, "loss": 0.0054, "step": 73000 }, { "grad_norm": 0.07427729666233063, "learning_rate": 2.073321501642728e-06, "loss": 0.0055, "step": 73010 }, { "grad_norm": 0.10284818708896637, "learning_rate": 2.0674355310711937e-06, "loss": 0.005, "step": 73020 }, { "grad_norm": 0.11176858842372894, "learning_rate": 2.0615577508639682e-06, "loss": 0.0073, "step": 73030 }, { "grad_norm": 0.09048686921596527, "learning_rate": 2.055688162025399e-06, "loss": 0.0053, "step": 73040 }, { "grad_norm": 0.06862715631723404, "learning_rate": 2.0498267655584546e-06, "loss": 0.0055, "step": 73050 }, { "grad_norm": 0.08931977301836014, "learning_rate": 2.0439735624646626e-06, "loss": 0.007, "step": 73060 }, { "grad_norm": 0.08214706927537918, "learning_rate": 2.0381285537442008e-06, "loss": 0.0044, "step": 73070 }, { "grad_norm": 0.11554542928934097, "learning_rate": 2.032291740395803e-06, "loss": 0.0056, "step": 73080 }, { "grad_norm": 0.08052445948123932, "learning_rate": 2.0264631234168276e-06, "loss": 0.0063, "step": 73090 }, { "grad_norm": 0.16791535913944244, "learning_rate": 2.0206427038032326e-06, "loss": 0.0045, "step": 73100 }, { "grad_norm": 0.09771635383367538, "learning_rate": 2.0148304825495457e-06, "loss": 0.0048, "step": 73110 }, { "grad_norm": 0.07972514629364014, "learning_rate": 2.0090264606489496e-06, "loss": 0.0041, "step": 73120 }, { "grad_norm": 0.05531252920627594, "learning_rate": 2.003230639093162e-06, "loss": 0.0058, "step": 73130 }, { "grad_norm": 0.10509967058897018, "learning_rate": 1.9974430188725525e-06, "loss": 0.006, "step": 73140 }, { "grad_norm": 0.06182785704731941, "learning_rate": 1.9916636009760513e-06, "loss": 0.0061, "step": 73150 }, { "grad_norm": 0.14580048620700836, "learning_rate": 1.985892386391208e-06, "loss": 0.0064, "step": 73160 }, { "grad_norm": 0.1390351504087448, "learning_rate": 1.980129376104173e-06, "loss": 0.0069, "step": 73170 }, { "grad_norm": 0.10809417068958282, "learning_rate": 1.9743745710996796e-06, "loss": 0.006, "step": 73180 }, { "grad_norm": 0.0670594647526741, "learning_rate": 1.9686279723610757e-06, "loss": 0.0053, "step": 73190 }, { "grad_norm": 0.12120150774717331, "learning_rate": 1.962889580870281e-06, "loss": 0.0052, "step": 73200 }, { "grad_norm": 0.14549687504768372, "learning_rate": 1.9571593976078385e-06, "loss": 0.0055, "step": 73210 }, { "grad_norm": 0.12206590175628662, "learning_rate": 1.9514374235528876e-06, "loss": 0.0043, "step": 73220 }, { "grad_norm": 0.072635218501091, "learning_rate": 1.9457236596831408e-06, "loss": 0.005, "step": 73230 }, { "grad_norm": 0.1025901511311531, "learning_rate": 1.940018106974939e-06, "loss": 0.0059, "step": 73240 }, { "grad_norm": 0.08081553876399994, "learning_rate": 1.9343207664032025e-06, "loss": 0.0066, "step": 73250 }, { "grad_norm": 0.17243506014347076, "learning_rate": 1.928631638941436e-06, "loss": 0.0056, "step": 73260 }, { "grad_norm": 0.17934787273406982, "learning_rate": 1.922950725561784e-06, "loss": 0.0075, "step": 73270 }, { "grad_norm": 0.1403682678937912, "learning_rate": 1.9172780272349312e-06, "loss": 0.0073, "step": 73280 }, { "grad_norm": 0.1314323991537094, "learning_rate": 1.911613544930202e-06, "loss": 0.0064, "step": 73290 }, { "grad_norm": 0.1581193506717682, "learning_rate": 1.9059572796155e-06, "loss": 0.0065, "step": 73300 }, { "grad_norm": 0.0728515237569809, "learning_rate": 1.9003092322573192e-06, "loss": 0.0062, "step": 73310 }, { "grad_norm": 0.07335276156663895, "learning_rate": 1.8946694038207647e-06, "loss": 0.0049, "step": 73320 }, { "grad_norm": 0.06887276470661163, "learning_rate": 1.8890377952695215e-06, "loss": 0.0056, "step": 73330 }, { "grad_norm": 0.11755683273077011, "learning_rate": 1.8834144075658865e-06, "loss": 0.0079, "step": 73340 }, { "grad_norm": 0.10989996790885925, "learning_rate": 1.8777992416707357e-06, "loss": 0.0058, "step": 73350 }, { "grad_norm": 0.04157223552465439, "learning_rate": 1.8721922985435458e-06, "loss": 0.0045, "step": 73360 }, { "grad_norm": 0.052759330719709396, "learning_rate": 1.8665935791424006e-06, "loss": 0.0051, "step": 73370 }, { "grad_norm": 0.11056128889322281, "learning_rate": 1.8610030844239512e-06, "loss": 0.0053, "step": 73380 }, { "grad_norm": 0.1092318668961525, "learning_rate": 1.8554208153434838e-06, "loss": 0.0056, "step": 73390 }, { "grad_norm": 0.11276749521493912, "learning_rate": 1.8498467728548296e-06, "loss": 0.006, "step": 73400 }, { "grad_norm": 0.06181059032678604, "learning_rate": 1.8442809579104547e-06, "loss": 0.0084, "step": 73410 }, { "grad_norm": 0.19320793449878693, "learning_rate": 1.8387233714614039e-06, "loss": 0.0063, "step": 73420 }, { "grad_norm": 0.09761764854192734, "learning_rate": 1.8331740144573174e-06, "loss": 0.0064, "step": 73430 }, { "grad_norm": 0.06362862139940262, "learning_rate": 1.8276328878464255e-06, "loss": 0.0058, "step": 73440 }, { "grad_norm": 0.06291187554597855, "learning_rate": 1.82209999257556e-06, "loss": 0.0068, "step": 73450 }, { "grad_norm": 0.08824345469474792, "learning_rate": 1.8165753295901311e-06, "loss": 0.0065, "step": 73460 }, { "grad_norm": 0.08566910773515701, "learning_rate": 1.8110588998341616e-06, "loss": 0.0061, "step": 73470 }, { "grad_norm": 0.08121778815984726, "learning_rate": 1.805550704250253e-06, "loss": 0.005, "step": 73480 }, { "grad_norm": 0.10100387781858444, "learning_rate": 1.8000507437796077e-06, "loss": 0.0041, "step": 73490 }, { "grad_norm": 0.16279266774654388, "learning_rate": 1.7945590193620242e-06, "loss": 0.006, "step": 73500 }, { "grad_norm": 0.09297078102827072, "learning_rate": 1.7890755319358742e-06, "loss": 0.005, "step": 73510 }, { "grad_norm": 0.07415027916431427, "learning_rate": 1.7836002824381525e-06, "loss": 0.0049, "step": 73520 }, { "grad_norm": 0.17178624868392944, "learning_rate": 1.7781332718044165e-06, "loss": 0.0067, "step": 73530 }, { "grad_norm": 0.15442708134651184, "learning_rate": 1.772674500968835e-06, "loss": 0.0061, "step": 73540 }, { "grad_norm": 0.1146111711859703, "learning_rate": 1.7672239708641624e-06, "loss": 0.0053, "step": 73550 }, { "grad_norm": 0.07608465850353241, "learning_rate": 1.7617816824217371e-06, "loss": 0.0055, "step": 73560 }, { "grad_norm": 0.13709555566310883, "learning_rate": 1.7563476365715148e-06, "loss": 0.0061, "step": 73570 }, { "grad_norm": 0.11975632607936859, "learning_rate": 1.750921834242003e-06, "loss": 0.0046, "step": 73580 }, { "grad_norm": 0.05521666258573532, "learning_rate": 1.7455042763603436e-06, "loss": 0.0039, "step": 73590 }, { "grad_norm": 0.07532007247209549, "learning_rate": 1.7400949638522345e-06, "loss": 0.0057, "step": 73600 }, { "grad_norm": 0.10695847868919373, "learning_rate": 1.734693897641987e-06, "loss": 0.0059, "step": 73610 }, { "grad_norm": 0.08098779618740082, "learning_rate": 1.7293010786524955e-06, "loss": 0.0085, "step": 73620 }, { "grad_norm": 0.10535097867250443, "learning_rate": 1.72391650780524e-06, "loss": 0.0055, "step": 73630 }, { "grad_norm": 0.1322144865989685, "learning_rate": 1.718540186020301e-06, "loss": 0.0057, "step": 73640 }, { "grad_norm": 0.15942136943340302, "learning_rate": 1.7131721142163437e-06, "loss": 0.0058, "step": 73650 }, { "grad_norm": 0.09433111548423767, "learning_rate": 1.7078122933106233e-06, "loss": 0.0052, "step": 73660 }, { "grad_norm": 0.08381475508213043, "learning_rate": 1.7024607242189905e-06, "loss": 0.0064, "step": 73670 }, { "grad_norm": 0.11618505418300629, "learning_rate": 1.6971174078558749e-06, "loss": 0.0051, "step": 73680 }, { "grad_norm": 0.13112422823905945, "learning_rate": 1.6917823451343073e-06, "loss": 0.0046, "step": 73690 }, { "grad_norm": 0.28479912877082825, "learning_rate": 1.6864555369659142e-06, "loss": 0.0058, "step": 73700 }, { "grad_norm": 0.096336230635643, "learning_rate": 1.681136984260878e-06, "loss": 0.0063, "step": 73710 }, { "grad_norm": 0.10535489767789841, "learning_rate": 1.6758266879280172e-06, "loss": 0.0079, "step": 73720 }, { "grad_norm": 0.12373458594083786, "learning_rate": 1.6705246488746996e-06, "loss": 0.0046, "step": 73730 }, { "grad_norm": 0.11892662197351456, "learning_rate": 1.6652308680069062e-06, "loss": 0.005, "step": 73740 }, { "grad_norm": 0.1316543072462082, "learning_rate": 1.6599453462292081e-06, "loss": 0.004, "step": 73750 }, { "grad_norm": 0.10905557125806808, "learning_rate": 1.6546680844447326e-06, "loss": 0.0049, "step": 73760 }, { "grad_norm": 0.062199145555496216, "learning_rate": 1.6493990835552475e-06, "loss": 0.0063, "step": 73770 }, { "grad_norm": 0.17830625176429749, "learning_rate": 1.6441383444610547e-06, "loss": 0.0077, "step": 73780 }, { "grad_norm": 0.0578877329826355, "learning_rate": 1.6388858680610908e-06, "loss": 0.0045, "step": 73790 }, { "grad_norm": 0.06872562319040298, "learning_rate": 1.6336416552528543e-06, "loss": 0.007, "step": 73800 }, { "grad_norm": 0.07734757661819458, "learning_rate": 1.6284057069324288e-06, "loss": 0.0051, "step": 73810 }, { "grad_norm": 0.08768851310014725, "learning_rate": 1.623178023994504e-06, "loss": 0.0058, "step": 73820 }, { "grad_norm": 0.2040807157754898, "learning_rate": 1.6179586073323483e-06, "loss": 0.0063, "step": 73830 }, { "grad_norm": 0.0888972207903862, "learning_rate": 1.612747457837821e-06, "loss": 0.0071, "step": 73840 }, { "grad_norm": 0.05757838115096092, "learning_rate": 1.607544576401354e-06, "loss": 0.0041, "step": 73850 }, { "grad_norm": 0.10269956290721893, "learning_rate": 1.6023499639119754e-06, "loss": 0.0053, "step": 73860 }, { "grad_norm": 0.08826889097690582, "learning_rate": 1.5971636212573138e-06, "loss": 0.005, "step": 73870 }, { "grad_norm": 0.08000966906547546, "learning_rate": 1.5919855493235714e-06, "loss": 0.0042, "step": 73880 }, { "grad_norm": 0.06699005514383316, "learning_rate": 1.586815748995535e-06, "loss": 0.0056, "step": 73890 }, { "grad_norm": 0.2096516191959381, "learning_rate": 1.5816542211565866e-06, "loss": 0.0039, "step": 73900 }, { "grad_norm": 0.06700240820646286, "learning_rate": 1.5765009666886765e-06, "loss": 0.0056, "step": 73910 }, { "grad_norm": 0.11059357970952988, "learning_rate": 1.5713559864723781e-06, "loss": 0.0069, "step": 73920 }, { "grad_norm": 0.1181207150220871, "learning_rate": 1.5662192813868104e-06, "loss": 0.005, "step": 73930 }, { "grad_norm": 0.12439980357885361, "learning_rate": 1.561090852309699e-06, "loss": 0.0043, "step": 73940 }, { "grad_norm": 0.08597816526889801, "learning_rate": 1.5559707001173651e-06, "loss": 0.0061, "step": 73950 }, { "grad_norm": 0.14423806965351105, "learning_rate": 1.5508588256846757e-06, "loss": 0.006, "step": 73960 }, { "grad_norm": 0.1108202412724495, "learning_rate": 1.5457552298851319e-06, "loss": 0.0048, "step": 73970 }, { "grad_norm": 0.05949276685714722, "learning_rate": 1.5406599135907918e-06, "loss": 0.0045, "step": 73980 }, { "grad_norm": 0.12904591858386993, "learning_rate": 1.5355728776723088e-06, "loss": 0.0058, "step": 73990 }, { "grad_norm": 0.08524516224861145, "learning_rate": 1.5304941229989155e-06, "loss": 0.0052, "step": 74000 }, { "grad_norm": 0.07451487332582474, "learning_rate": 1.5254236504384345e-06, "loss": 0.0046, "step": 74010 }, { "grad_norm": 0.49921831488609314, "learning_rate": 1.5203614608572726e-06, "loss": 0.0052, "step": 74020 }, { "grad_norm": 0.09555806964635849, "learning_rate": 1.5153075551204044e-06, "loss": 0.0041, "step": 74030 }, { "grad_norm": 0.09997159987688065, "learning_rate": 1.5102619340914225e-06, "loss": 0.0052, "step": 74040 }, { "grad_norm": 0.13111701607704163, "learning_rate": 1.505224598632482e-06, "loss": 0.006, "step": 74050 }, { "grad_norm": 0.106523297727108, "learning_rate": 1.5001955496043164e-06, "loss": 0.0057, "step": 74060 }, { "grad_norm": 0.145300954580307, "learning_rate": 1.4951747878662604e-06, "loss": 0.0066, "step": 74070 }, { "grad_norm": 0.09604056179523468, "learning_rate": 1.4901623142762221e-06, "loss": 0.0072, "step": 74080 }, { "grad_norm": 0.077010378241539, "learning_rate": 1.4851581296907001e-06, "loss": 0.0047, "step": 74090 }, { "grad_norm": 0.08521629869937897, "learning_rate": 1.4801622349647714e-06, "loss": 0.0054, "step": 74100 }, { "grad_norm": 0.08044733852148056, "learning_rate": 1.4751746309520976e-06, "loss": 0.0075, "step": 74110 }, { "grad_norm": 0.15328630805015564, "learning_rate": 1.4701953185049132e-06, "loss": 0.0067, "step": 74120 }, { "grad_norm": 0.0782121941447258, "learning_rate": 1.4652242984740661e-06, "loss": 0.0063, "step": 74130 }, { "grad_norm": 0.06582991033792496, "learning_rate": 1.4602615717089484e-06, "loss": 0.0055, "step": 74140 }, { "grad_norm": 0.09995695948600769, "learning_rate": 1.4553071390575657e-06, "loss": 0.0055, "step": 74150 }, { "grad_norm": 0.12209933996200562, "learning_rate": 1.450361001366496e-06, "loss": 0.0058, "step": 74160 }, { "grad_norm": 0.1265784502029419, "learning_rate": 1.4454231594809021e-06, "loss": 0.0061, "step": 74170 }, { "grad_norm": 0.08191704750061035, "learning_rate": 1.4404936142445036e-06, "loss": 0.0054, "step": 74180 }, { "grad_norm": 0.05248648300766945, "learning_rate": 1.4355723664996546e-06, "loss": 0.005, "step": 74190 }, { "grad_norm": 0.12297865748405457, "learning_rate": 1.4306594170872433e-06, "loss": 0.0044, "step": 74200 }, { "grad_norm": 0.09946416318416595, "learning_rate": 1.4257547668467598e-06, "loss": 0.0048, "step": 74210 }, { "grad_norm": 0.08537543565034866, "learning_rate": 1.4208584166162886e-06, "loss": 0.0069, "step": 74220 }, { "grad_norm": 0.1599758416414261, "learning_rate": 1.4159703672324553e-06, "loss": 0.0051, "step": 74230 }, { "grad_norm": 0.1335819661617279, "learning_rate": 1.4110906195305252e-06, "loss": 0.0049, "step": 74240 }, { "grad_norm": 0.11016996949911118, "learning_rate": 1.406219174344292e-06, "loss": 0.0057, "step": 74250 }, { "grad_norm": 0.12032245099544525, "learning_rate": 1.401356032506157e-06, "loss": 0.0055, "step": 74260 }, { "grad_norm": 0.09973245859146118, "learning_rate": 1.3965011948471051e-06, "loss": 0.0067, "step": 74270 }, { "grad_norm": 0.07621905207633972, "learning_rate": 1.3916546621966842e-06, "loss": 0.0055, "step": 74280 }, { "grad_norm": 0.09378542751073837, "learning_rate": 1.3868164353830482e-06, "loss": 0.0051, "step": 74290 }, { "grad_norm": 0.057481251657009125, "learning_rate": 1.3819865152329082e-06, "loss": 0.004, "step": 74300 }, { "grad_norm": 0.10624954849481583, "learning_rate": 1.377164902571565e-06, "loss": 0.006, "step": 74310 }, { "grad_norm": 0.09113990515470505, "learning_rate": 1.372351598222904e-06, "loss": 0.0046, "step": 74320 }, { "grad_norm": 0.13878530263900757, "learning_rate": 1.3675466030093843e-06, "loss": 0.0053, "step": 74330 }, { "grad_norm": 0.08019386976957321, "learning_rate": 1.3627499177520486e-06, "loss": 0.0043, "step": 74340 }, { "grad_norm": 0.056232914328575134, "learning_rate": 1.3579615432705196e-06, "loss": 0.0059, "step": 74350 }, { "grad_norm": 0.07416803389787674, "learning_rate": 1.3531814803829978e-06, "loss": 0.0049, "step": 74360 }, { "grad_norm": 0.083086296916008, "learning_rate": 1.3484097299062803e-06, "loss": 0.0054, "step": 74370 }, { "grad_norm": 0.14399486780166626, "learning_rate": 1.343646292655698e-06, "loss": 0.0072, "step": 74380 }, { "grad_norm": 0.07709522545337677, "learning_rate": 1.3388911694452277e-06, "loss": 0.0076, "step": 74390 }, { "grad_norm": 0.08761224150657654, "learning_rate": 1.3341443610873638e-06, "loss": 0.0058, "step": 74400 }, { "grad_norm": 0.042956989258527756, "learning_rate": 1.329405868393213e-06, "loss": 0.0061, "step": 74410 }, { "grad_norm": 0.08857432007789612, "learning_rate": 1.3246756921724613e-06, "loss": 0.0043, "step": 74420 }, { "grad_norm": 0.09967011213302612, "learning_rate": 1.3199538332333506e-06, "loss": 0.0046, "step": 74430 }, { "grad_norm": 0.11052609980106354, "learning_rate": 1.3152402923827411e-06, "loss": 0.0056, "step": 74440 }, { "grad_norm": 0.2637864947319031, "learning_rate": 1.3105350704260277e-06, "loss": 0.0059, "step": 74450 }, { "grad_norm": 0.155509814620018, "learning_rate": 1.305838168167206e-06, "loss": 0.0053, "step": 74460 }, { "grad_norm": 0.07357630878686905, "learning_rate": 1.301149586408862e-06, "loss": 0.0067, "step": 74470 }, { "grad_norm": 0.09336449205875397, "learning_rate": 1.2964693259521322e-06, "loss": 0.0052, "step": 74480 }, { "grad_norm": 0.12579403817653656, "learning_rate": 1.2917973875967548e-06, "loss": 0.0045, "step": 74490 }, { "grad_norm": 0.17402102053165436, "learning_rate": 1.2871337721410249e-06, "loss": 0.0073, "step": 74500 }, { "grad_norm": 0.06766858696937561, "learning_rate": 1.2824784803818379e-06, "loss": 0.0052, "step": 74510 }, { "grad_norm": 0.049412958323955536, "learning_rate": 1.2778315131146467e-06, "loss": 0.005, "step": 74520 }, { "grad_norm": 0.10508320480585098, "learning_rate": 1.2731928711334994e-06, "loss": 0.0061, "step": 74530 }, { "grad_norm": 0.062014807015657425, "learning_rate": 1.2685625552310122e-06, "loss": 0.0082, "step": 74540 }, { "grad_norm": 0.12071686238050461, "learning_rate": 1.263940566198374e-06, "loss": 0.0043, "step": 74550 }, { "grad_norm": 0.1341678649187088, "learning_rate": 1.259326904825353e-06, "loss": 0.0055, "step": 74560 }, { "grad_norm": 0.13995124399662018, "learning_rate": 1.2547215719003137e-06, "loss": 0.0058, "step": 74570 }, { "grad_norm": 0.13875551521778107, "learning_rate": 1.2501245682101703e-06, "loss": 0.0066, "step": 74580 }, { "grad_norm": 0.06942742317914963, "learning_rate": 1.2455358945404171e-06, "loss": 0.0054, "step": 74590 }, { "grad_norm": 0.13129304349422455, "learning_rate": 1.2409555516751493e-06, "loss": 0.006, "step": 74600 }, { "grad_norm": 0.07220458984375, "learning_rate": 1.2363835403970125e-06, "loss": 0.0066, "step": 74610 }, { "grad_norm": 0.10113615542650223, "learning_rate": 1.2318198614872489e-06, "loss": 0.0039, "step": 74620 }, { "grad_norm": 0.10998120158910751, "learning_rate": 1.2272645157256457e-06, "loss": 0.0062, "step": 74630 }, { "grad_norm": 0.06106709688901901, "learning_rate": 1.222717503890608e-06, "loss": 0.0063, "step": 74640 }, { "grad_norm": 0.04931388422846794, "learning_rate": 1.218178826759081e-06, "loss": 0.0042, "step": 74650 }, { "grad_norm": 0.1289490908384323, "learning_rate": 1.213648485106611e-06, "loss": 0.0063, "step": 74660 }, { "grad_norm": 0.10559434443712234, "learning_rate": 1.2091264797073066e-06, "loss": 0.0045, "step": 74670 }, { "grad_norm": 0.10569224506616592, "learning_rate": 1.2046128113338494e-06, "loss": 0.0065, "step": 74680 }, { "grad_norm": 0.08562058955430984, "learning_rate": 1.200107480757512e-06, "loss": 0.0051, "step": 74690 }, { "grad_norm": 0.08470632880926132, "learning_rate": 1.1956104887481168e-06, "loss": 0.0053, "step": 74700 }, { "grad_norm": 0.11403661221265793, "learning_rate": 1.191121836074094e-06, "loss": 0.0044, "step": 74710 }, { "grad_norm": 0.0838080644607544, "learning_rate": 1.1866415235024186e-06, "loss": 0.0056, "step": 74720 }, { "grad_norm": 0.09694193303585052, "learning_rate": 1.182169551798662e-06, "loss": 0.0054, "step": 74730 }, { "grad_norm": 0.10314243286848068, "learning_rate": 1.177705921726957e-06, "loss": 0.0065, "step": 74740 }, { "grad_norm": 0.05074272304773331, "learning_rate": 1.173250634050016e-06, "loss": 0.0051, "step": 74750 }, { "grad_norm": 0.0829746425151825, "learning_rate": 1.1688036895291354e-06, "loss": 0.005, "step": 74760 }, { "grad_norm": 0.13104283809661865, "learning_rate": 1.1643650889241574e-06, "loss": 0.0049, "step": 74770 }, { "grad_norm": 0.11441925168037415, "learning_rate": 1.1599348329935311e-06, "loss": 0.0046, "step": 74780 }, { "grad_norm": 0.08601890504360199, "learning_rate": 1.1555129224942673e-06, "loss": 0.006, "step": 74790 }, { "grad_norm": 0.0929507240653038, "learning_rate": 1.1510993581819396e-06, "loss": 0.0041, "step": 74800 }, { "grad_norm": 0.07625174522399902, "learning_rate": 1.1466941408107112e-06, "loss": 0.004, "step": 74810 }, { "grad_norm": 0.1935838907957077, "learning_rate": 1.1422972711333247e-06, "loss": 0.0058, "step": 74820 }, { "grad_norm": 0.07995229214429855, "learning_rate": 1.1379087499010565e-06, "loss": 0.0065, "step": 74830 }, { "grad_norm": 0.07700946182012558, "learning_rate": 1.1335285778638127e-06, "loss": 0.0051, "step": 74840 }, { "grad_norm": 0.1213860884308815, "learning_rate": 1.1291567557700333e-06, "loss": 0.0051, "step": 74850 }, { "grad_norm": 0.10609034448862076, "learning_rate": 1.124793284366743e-06, "loss": 0.0058, "step": 74860 }, { "grad_norm": 0.08011945337057114, "learning_rate": 1.1204381643995455e-06, "loss": 0.005, "step": 74870 }, { "grad_norm": 0.07525264471769333, "learning_rate": 1.116091396612595e-06, "loss": 0.0056, "step": 74880 }, { "grad_norm": 0.13448628783226013, "learning_rate": 1.1117529817486538e-06, "loss": 0.0064, "step": 74890 }, { "grad_norm": 0.08938349038362503, "learning_rate": 1.107422920549034e-06, "loss": 0.0041, "step": 74900 }, { "grad_norm": 0.04932351037859917, "learning_rate": 1.1031012137536157e-06, "loss": 0.0037, "step": 74910 }, { "grad_norm": 0.08159513771533966, "learning_rate": 1.0987878621008695e-06, "loss": 0.0056, "step": 74920 }, { "grad_norm": 0.06846623867750168, "learning_rate": 1.0944828663278284e-06, "loss": 0.0049, "step": 74930 }, { "grad_norm": 0.11700572818517685, "learning_rate": 1.0901862271700925e-06, "loss": 0.0051, "step": 74940 }, { "grad_norm": 0.079573854804039, "learning_rate": 1.0858979453618467e-06, "loss": 0.0054, "step": 74950 }, { "grad_norm": 0.043462321162223816, "learning_rate": 1.0816180216358441e-06, "loss": 0.0051, "step": 74960 }, { "grad_norm": 0.14852866530418396, "learning_rate": 1.0773464567233937e-06, "loss": 0.0046, "step": 74970 }, { "grad_norm": 0.10351154953241348, "learning_rate": 1.0730832513543953e-06, "loss": 0.006, "step": 74980 }, { "grad_norm": 0.1124805212020874, "learning_rate": 1.0688284062573218e-06, "loss": 0.0076, "step": 74990 }, { "grad_norm": 0.11830601096153259, "learning_rate": 1.0645819221591969e-06, "loss": 0.0059, "step": 75000 }, { "grad_norm": 0.107840895652771, "learning_rate": 1.0603437997856347e-06, "loss": 0.007, "step": 75010 }, { "grad_norm": 0.12668417394161224, "learning_rate": 1.0561140398608228e-06, "loss": 0.0055, "step": 75020 }, { "grad_norm": 0.10380426049232483, "learning_rate": 1.0518926431074937e-06, "loss": 0.0065, "step": 75030 }, { "grad_norm": 0.14650970697402954, "learning_rate": 1.0476796102469877e-06, "loss": 0.0058, "step": 75040 }, { "grad_norm": 0.1410260796546936, "learning_rate": 1.0434749419991786e-06, "loss": 0.0046, "step": 75050 }, { "grad_norm": 0.05757239833474159, "learning_rate": 1.0392786390825415e-06, "loss": 0.0065, "step": 75060 }, { "grad_norm": 0.07266825437545776, "learning_rate": 1.0350907022141087e-06, "loss": 0.0052, "step": 75070 }, { "grad_norm": 0.12755495309829712, "learning_rate": 1.0309111321094744e-06, "loss": 0.0072, "step": 75080 }, { "grad_norm": 0.14875908195972443, "learning_rate": 1.026739929482834e-06, "loss": 0.0055, "step": 75090 }, { "grad_norm": 0.06088259071111679, "learning_rate": 1.022577095046906e-06, "loss": 0.0048, "step": 75100 }, { "grad_norm": 0.15138570964336395, "learning_rate": 1.0184226295130217e-06, "loss": 0.0054, "step": 75110 }, { "grad_norm": 0.09246646612882614, "learning_rate": 1.0142765335910575e-06, "loss": 0.0062, "step": 75120 }, { "grad_norm": 0.10943177342414856, "learning_rate": 1.0101388079894746e-06, "loss": 0.0063, "step": 75130 }, { "grad_norm": 0.17026327550411224, "learning_rate": 1.0060094534152908e-06, "loss": 0.0061, "step": 75140 }, { "grad_norm": 0.05118057131767273, "learning_rate": 1.0018884705741028e-06, "loss": 0.0052, "step": 75150 }, { "grad_norm": 0.10816052556037903, "learning_rate": 9.977758601700805e-07, "loss": 0.0055, "step": 75160 }, { "grad_norm": 0.1757807433605194, "learning_rate": 9.93671622905945e-07, "loss": 0.0062, "step": 75170 }, { "grad_norm": 0.09035356342792511, "learning_rate": 9.895757594830024e-07, "loss": 0.0061, "step": 75180 }, { "grad_norm": 0.1488254964351654, "learning_rate": 9.854882706011204e-07, "loss": 0.0053, "step": 75190 }, { "grad_norm": 0.10875631123781204, "learning_rate": 9.814091569587513e-07, "loss": 0.0078, "step": 75200 }, { "grad_norm": 0.10092882812023163, "learning_rate": 9.77338419252888e-07, "loss": 0.0041, "step": 75210 }, { "grad_norm": 0.13467800617218018, "learning_rate": 9.732760581791289e-07, "loss": 0.0065, "step": 75220 }, { "grad_norm": 0.13362352550029755, "learning_rate": 9.692220744315972e-07, "loss": 0.0058, "step": 75230 }, { "grad_norm": 0.12065107375383377, "learning_rate": 9.651764687030162e-07, "loss": 0.0057, "step": 75240 }, { "grad_norm": 0.09641612321138382, "learning_rate": 9.611392416846776e-07, "loss": 0.0048, "step": 75250 }, { "grad_norm": 0.13019734621047974, "learning_rate": 9.571103940664238e-07, "loss": 0.0047, "step": 75260 }, { "grad_norm": 0.08738961070775986, "learning_rate": 9.530899265366822e-07, "loss": 0.0044, "step": 75270 }, { "grad_norm": 0.21846851706504822, "learning_rate": 9.490778397824307e-07, "loss": 0.0046, "step": 75280 }, { "grad_norm": 0.11188942939043045, "learning_rate": 9.450741344892322e-07, "loss": 0.005, "step": 75290 }, { "grad_norm": 0.09784606844186783, "learning_rate": 9.410788113412116e-07, "loss": 0.007, "step": 75300 }, { "grad_norm": 0.05171563848853111, "learning_rate": 9.370918710210563e-07, "loss": 0.0052, "step": 75310 }, { "grad_norm": 0.08560234308242798, "learning_rate": 9.331133142100268e-07, "loss": 0.0038, "step": 75320 }, { "grad_norm": 0.09197299927473068, "learning_rate": 9.291431415879459e-07, "loss": 0.0053, "step": 75330 }, { "grad_norm": 0.05231054127216339, "learning_rate": 9.251813538332155e-07, "loss": 0.0045, "step": 75340 }, { "grad_norm": 0.13104431331157684, "learning_rate": 9.212279516227884e-07, "loss": 0.0055, "step": 75350 }, { "grad_norm": 0.10733109712600708, "learning_rate": 9.172829356322021e-07, "loss": 0.0048, "step": 75360 }, { "grad_norm": 0.11429977416992188, "learning_rate": 9.133463065355452e-07, "loss": 0.004, "step": 75370 }, { "grad_norm": 0.14969556033611298, "learning_rate": 9.094180650054796e-07, "loss": 0.0069, "step": 75380 }, { "grad_norm": 0.11383223533630371, "learning_rate": 9.054982117132404e-07, "loss": 0.0053, "step": 75390 }, { "grad_norm": 0.11854812502861023, "learning_rate": 9.015867473286143e-07, "loss": 0.0069, "step": 75400 }, { "grad_norm": 0.07474000751972198, "learning_rate": 8.976836725199778e-07, "loss": 0.0048, "step": 75410 }, { "grad_norm": 0.09015891700983047, "learning_rate": 8.937889879542416e-07, "loss": 0.0068, "step": 75420 }, { "grad_norm": 0.13602803647518158, "learning_rate": 8.899026942969068e-07, "loss": 0.0054, "step": 75430 }, { "grad_norm": 0.13555189967155457, "learning_rate": 8.860247922120424e-07, "loss": 0.0041, "step": 75440 }, { "grad_norm": 0.13457904756069183, "learning_rate": 8.821552823622737e-07, "loss": 0.0057, "step": 75450 }, { "grad_norm": 0.09444008767604828, "learning_rate": 8.782941654087884e-07, "loss": 0.0046, "step": 75460 }, { "grad_norm": 0.07938921451568604, "learning_rate": 8.744414420113478e-07, "loss": 0.0064, "step": 75470 }, { "grad_norm": 0.06921878457069397, "learning_rate": 8.705971128282753e-07, "loss": 0.0066, "step": 75480 }, { "grad_norm": 0.12557409703731537, "learning_rate": 8.66761178516473e-07, "loss": 0.0063, "step": 75490 }, { "grad_norm": 0.06438504159450531, "learning_rate": 8.629336397313781e-07, "loss": 0.0041, "step": 75500 }, { "grad_norm": 0.06302520632743835, "learning_rate": 8.591144971270282e-07, "loss": 0.0044, "step": 75510 }, { "grad_norm": 0.14041589200496674, "learning_rate": 8.553037513560069e-07, "loss": 0.0059, "step": 75520 }, { "grad_norm": 0.17951957881450653, "learning_rate": 8.515014030694546e-07, "loss": 0.0053, "step": 75530 }, { "grad_norm": 0.16834643483161926, "learning_rate": 8.47707452917107e-07, "loss": 0.0051, "step": 75540 }, { "grad_norm": 0.05265158414840698, "learning_rate": 8.439219015472343e-07, "loss": 0.0059, "step": 75550 }, { "grad_norm": 0.04918736219406128, "learning_rate": 8.401447496066861e-07, "loss": 0.0053, "step": 75560 }, { "grad_norm": 0.1217593401670456, "learning_rate": 8.363759977408792e-07, "loss": 0.0044, "step": 75570 }, { "grad_norm": 0.09531937539577484, "learning_rate": 8.326156465937817e-07, "loss": 0.0057, "step": 75580 }, { "grad_norm": 0.17653916776180267, "learning_rate": 8.28863696807941e-07, "loss": 0.0051, "step": 75590 }, { "grad_norm": 0.053649939596652985, "learning_rate": 8.251201490244609e-07, "loss": 0.0044, "step": 75600 }, { "grad_norm": 0.1921738237142563, "learning_rate": 8.213850038830129e-07, "loss": 0.0061, "step": 75610 }, { "grad_norm": 0.10532928258180618, "learning_rate": 8.176582620218309e-07, "loss": 0.0037, "step": 75620 }, { "grad_norm": 0.12880466878414154, "learning_rate": 8.139399240777057e-07, "loss": 0.0046, "step": 75630 }, { "grad_norm": 0.07866173982620239, "learning_rate": 8.102299906860122e-07, "loss": 0.0051, "step": 75640 }, { "grad_norm": 0.08070670068264008, "learning_rate": 8.065284624806657e-07, "loss": 0.0067, "step": 75650 }, { "grad_norm": 0.12460153549909592, "learning_rate": 8.028353400941601e-07, "loss": 0.0071, "step": 75660 }, { "grad_norm": 0.09088928252458572, "learning_rate": 7.991506241575575e-07, "loss": 0.0068, "step": 75670 }, { "grad_norm": 0.133449524641037, "learning_rate": 7.954743153004541e-07, "loss": 0.0061, "step": 75680 }, { "grad_norm": 0.0708434209227562, "learning_rate": 7.918064141510528e-07, "loss": 0.0054, "step": 75690 }, { "grad_norm": 0.08569797873497009, "learning_rate": 7.881469213360859e-07, "loss": 0.0082, "step": 75700 }, { "grad_norm": 0.10670623183250427, "learning_rate": 7.844958374808642e-07, "loss": 0.0053, "step": 75710 }, { "grad_norm": 0.10108534246683121, "learning_rate": 7.808531632092608e-07, "loss": 0.005, "step": 75720 }, { "grad_norm": 0.1828058511018753, "learning_rate": 7.772188991436946e-07, "loss": 0.0057, "step": 75730 }, { "grad_norm": 0.12372155487537384, "learning_rate": 7.735930459051799e-07, "loss": 0.0061, "step": 75740 }, { "grad_norm": 0.10090842097997665, "learning_rate": 7.699756041132655e-07, "loss": 0.0053, "step": 75750 }, { "grad_norm": 0.11601923406124115, "learning_rate": 7.663665743860793e-07, "loss": 0.0054, "step": 75760 }, { "grad_norm": 0.08826585114002228, "learning_rate": 7.627659573403001e-07, "loss": 0.005, "step": 75770 }, { "grad_norm": 0.13413454592227936, "learning_rate": 7.591737535911802e-07, "loss": 0.0047, "step": 75780 }, { "grad_norm": 0.14425340294837952, "learning_rate": 7.555899637525288e-07, "loss": 0.0057, "step": 75790 }, { "grad_norm": 0.16183748841285706, "learning_rate": 7.520145884367058e-07, "loss": 0.0055, "step": 75800 }, { "grad_norm": 0.06965753436088562, "learning_rate": 7.484476282546615e-07, "loss": 0.0048, "step": 75810 }, { "grad_norm": 0.13519640266895294, "learning_rate": 7.448890838158806e-07, "loss": 0.0065, "step": 75820 }, { "grad_norm": 0.10632186383008957, "learning_rate": 7.413389557284267e-07, "loss": 0.0057, "step": 75830 }, { "grad_norm": 0.13860666751861572, "learning_rate": 7.377972445989201e-07, "loss": 0.0073, "step": 75840 }, { "grad_norm": 0.11810240149497986, "learning_rate": 7.342639510325377e-07, "loss": 0.0075, "step": 75850 }, { "grad_norm": 0.1400493085384369, "learning_rate": 7.307390756330246e-07, "loss": 0.0067, "step": 75860 }, { "grad_norm": 0.050606388598680496, "learning_rate": 7.272226190026876e-07, "loss": 0.0057, "step": 75870 }, { "grad_norm": 0.0778622180223465, "learning_rate": 7.237145817423907e-07, "loss": 0.0032, "step": 75880 }, { "grad_norm": 0.12290889769792557, "learning_rate": 7.202149644515654e-07, "loss": 0.0062, "step": 75890 }, { "grad_norm": 0.10711806267499924, "learning_rate": 7.167237677281946e-07, "loss": 0.0053, "step": 75900 }, { "grad_norm": 0.0980086624622345, "learning_rate": 7.132409921688288e-07, "loss": 0.0061, "step": 75910 }, { "grad_norm": 0.08431627601385117, "learning_rate": 7.09766638368592e-07, "loss": 0.004, "step": 75920 }, { "grad_norm": 0.09512527287006378, "learning_rate": 7.063007069211313e-07, "loss": 0.0041, "step": 75930 }, { "grad_norm": 0.16310015320777893, "learning_rate": 7.028431984187067e-07, "loss": 0.0067, "step": 75940 }, { "grad_norm": 0.06479614973068237, "learning_rate": 6.993941134520898e-07, "loss": 0.0041, "step": 75950 }, { "grad_norm": 0.08733174949884415, "learning_rate": 6.959534526106537e-07, "loss": 0.0061, "step": 75960 }, { "grad_norm": 0.09224937856197357, "learning_rate": 6.925212164822948e-07, "loss": 0.007, "step": 75970 }, { "grad_norm": 0.05899543687701225, "learning_rate": 6.890974056535049e-07, "loss": 0.0063, "step": 75980 }, { "grad_norm": 0.0675906389951706, "learning_rate": 6.856820207093106e-07, "loss": 0.0058, "step": 75990 }, { "grad_norm": 0.06059154495596886, "learning_rate": 6.822750622333063e-07, "loss": 0.0068, "step": 76000 }, { "grad_norm": 0.09522102773189545, "learning_rate": 6.788765308076539e-07, "loss": 0.0048, "step": 76010 }, { "grad_norm": 0.09891790896654129, "learning_rate": 6.754864270130668e-07, "loss": 0.0047, "step": 76020 }, { "grad_norm": 0.4826730787754059, "learning_rate": 6.721047514288203e-07, "loss": 0.0045, "step": 76030 }, { "grad_norm": 0.11105000972747803, "learning_rate": 6.687315046327469e-07, "loss": 0.006, "step": 76040 }, { "grad_norm": 0.054715655744075775, "learning_rate": 6.65366687201252e-07, "loss": 0.0058, "step": 76050 }, { "grad_norm": 0.1197877749800682, "learning_rate": 6.620102997092814e-07, "loss": 0.0041, "step": 76060 }, { "grad_norm": 0.11280026286840439, "learning_rate": 6.586623427303596e-07, "loss": 0.005, "step": 76070 }, { "grad_norm": 0.13091163337230682, "learning_rate": 6.553228168365455e-07, "loss": 0.0048, "step": 76080 }, { "grad_norm": 0.15439048409461975, "learning_rate": 6.519917225984884e-07, "loss": 0.0073, "step": 76090 }, { "grad_norm": 0.10142394155263901, "learning_rate": 6.486690605853718e-07, "loss": 0.0046, "step": 76100 }, { "grad_norm": 0.048803769052028656, "learning_rate": 6.453548313649527e-07, "loss": 0.0053, "step": 76110 }, { "grad_norm": 0.16956347227096558, "learning_rate": 6.420490355035446e-07, "loss": 0.0072, "step": 76120 }, { "grad_norm": 0.13083067536354065, "learning_rate": 6.387516735660071e-07, "loss": 0.006, "step": 76130 }, { "grad_norm": 0.13802681863307953, "learning_rate": 6.35462746115778e-07, "loss": 0.0058, "step": 76140 }, { "grad_norm": 0.07513614743947983, "learning_rate": 6.321822537148358e-07, "loss": 0.0061, "step": 76150 }, { "grad_norm": 0.09059088677167892, "learning_rate": 6.289101969237432e-07, "loss": 0.0058, "step": 76160 }, { "grad_norm": 0.13744062185287476, "learning_rate": 6.256465763015918e-07, "loss": 0.0052, "step": 76170 }, { "grad_norm": 0.04851220175623894, "learning_rate": 6.22391392406052e-07, "loss": 0.0058, "step": 76180 }, { "grad_norm": 0.06212158501148224, "learning_rate": 6.191446457933403e-07, "loss": 0.0053, "step": 76190 }, { "grad_norm": 0.04824089631438255, "learning_rate": 6.159063370182406e-07, "loss": 0.0062, "step": 76200 }, { "grad_norm": 0.15405286848545074, "learning_rate": 6.126764666340879e-07, "loss": 0.0055, "step": 76210 }, { "grad_norm": 0.04908437281847, "learning_rate": 6.094550351927852e-07, "loss": 0.006, "step": 76220 }, { "grad_norm": 0.058954644948244095, "learning_rate": 6.062420432447757e-07, "loss": 0.0048, "step": 76230 }, { "grad_norm": 0.053234219551086426, "learning_rate": 6.030374913390813e-07, "loss": 0.0043, "step": 76240 }, { "grad_norm": 0.12665440142154694, "learning_rate": 5.998413800232694e-07, "loss": 0.0078, "step": 76250 }, { "grad_norm": 0.11638090014457703, "learning_rate": 5.966537098434755e-07, "loss": 0.005, "step": 76260 }, { "grad_norm": 0.12618272006511688, "learning_rate": 5.934744813443694e-07, "loss": 0.0053, "step": 76270 }, { "grad_norm": 0.13093292713165283, "learning_rate": 5.903036950692054e-07, "loss": 0.0053, "step": 76280 }, { "grad_norm": 0.12995722889900208, "learning_rate": 5.871413515597835e-07, "loss": 0.007, "step": 76290 }, { "grad_norm": 0.12136723846197128, "learning_rate": 5.839874513564547e-07, "loss": 0.0045, "step": 76300 }, { "grad_norm": 0.08986756950616837, "learning_rate": 5.808419949981436e-07, "loss": 0.0046, "step": 76310 }, { "grad_norm": 0.04737069830298424, "learning_rate": 5.777049830223257e-07, "loss": 0.0059, "step": 76320 }, { "grad_norm": 0.15063980221748352, "learning_rate": 5.745764159650114e-07, "loss": 0.0049, "step": 76330 }, { "grad_norm": 0.09503097087144852, "learning_rate": 5.71456294360806e-07, "loss": 0.0044, "step": 76340 }, { "grad_norm": 0.08499383181333542, "learning_rate": 5.683446187428443e-07, "loss": 0.0053, "step": 76350 }, { "grad_norm": 0.17412786185741425, "learning_rate": 5.652413896428288e-07, "loss": 0.0064, "step": 76360 }, { "grad_norm": 0.08402565866708755, "learning_rate": 5.621466075910131e-07, "loss": 0.0062, "step": 76370 }, { "grad_norm": 0.11915072053670883, "learning_rate": 5.590602731162187e-07, "loss": 0.0053, "step": 76380 }, { "grad_norm": 0.07986970990896225, "learning_rate": 5.559823867458125e-07, "loss": 0.0042, "step": 76390 }, { "grad_norm": 0.08055516332387924, "learning_rate": 5.529129490057128e-07, "loss": 0.0061, "step": 76400 }, { "grad_norm": 0.08441344648599625, "learning_rate": 5.498519604204167e-07, "loss": 0.0052, "step": 76410 }, { "grad_norm": 0.04705921933054924, "learning_rate": 5.467994215129501e-07, "loss": 0.0052, "step": 76420 }, { "grad_norm": 0.06074262782931328, "learning_rate": 5.437553328049183e-07, "loss": 0.0043, "step": 76430 }, { "grad_norm": 0.07353762537240982, "learning_rate": 5.407196948164661e-07, "loss": 0.0062, "step": 76440 }, { "grad_norm": 0.0785958543419838, "learning_rate": 5.376925080663009e-07, "loss": 0.0061, "step": 76450 }, { "grad_norm": 0.12856611609458923, "learning_rate": 5.346737730716977e-07, "loss": 0.0065, "step": 76460 }, { "grad_norm": 0.06877520680427551, "learning_rate": 5.316634903484607e-07, "loss": 0.0037, "step": 76470 }, { "grad_norm": 0.056232381612062454, "learning_rate": 5.286616604109728e-07, "loss": 0.0031, "step": 76480 }, { "grad_norm": 0.07742445170879364, "learning_rate": 5.256682837721627e-07, "loss": 0.0055, "step": 76490 }, { "grad_norm": 0.12344095855951309, "learning_rate": 5.226833609435156e-07, "loss": 0.0061, "step": 76500 }, { "grad_norm": 0.17441228032112122, "learning_rate": 5.19706892435079e-07, "loss": 0.0047, "step": 76510 }, { "grad_norm": 0.10355468094348907, "learning_rate": 5.167388787554406e-07, "loss": 0.0072, "step": 76520 }, { "grad_norm": 0.05960612744092941, "learning_rate": 5.137793204117614e-07, "loss": 0.0049, "step": 76530 }, { "grad_norm": 0.14478760957717896, "learning_rate": 5.108282179097479e-07, "loss": 0.0057, "step": 76540 }, { "grad_norm": 0.1241113469004631, "learning_rate": 5.078855717536523e-07, "loss": 0.0053, "step": 76550 }, { "grad_norm": 0.08637150377035141, "learning_rate": 5.049513824463059e-07, "loss": 0.0055, "step": 76560 }, { "grad_norm": 0.0536356195807457, "learning_rate": 5.020256504890742e-07, "loss": 0.0043, "step": 76570 }, { "grad_norm": 0.06718094646930695, "learning_rate": 4.991083763818849e-07, "loss": 0.0056, "step": 76580 }, { "grad_norm": 0.05779512599110603, "learning_rate": 4.961995606232228e-07, "loss": 0.0064, "step": 76590 }, { "grad_norm": 0.1218542829155922, "learning_rate": 4.932992037101236e-07, "loss": 0.0074, "step": 76600 }, { "grad_norm": 0.18881529569625854, "learning_rate": 4.904073061381798e-07, "loss": 0.0057, "step": 76610 }, { "grad_norm": 0.15379323065280914, "learning_rate": 4.875238684015349e-07, "loss": 0.0073, "step": 76620 }, { "grad_norm": 0.04065421596169472, "learning_rate": 4.846488909928948e-07, "loss": 0.0052, "step": 76630 }, { "grad_norm": 0.09358085691928864, "learning_rate": 4.81782374403511e-07, "loss": 0.0066, "step": 76640 }, { "grad_norm": 0.09969694167375565, "learning_rate": 4.789243191231918e-07, "loss": 0.005, "step": 76650 }, { "grad_norm": 0.1773691475391388, "learning_rate": 4.76074725640302e-07, "loss": 0.0055, "step": 76660 }, { "grad_norm": 0.09905452281236649, "learning_rate": 4.732335944417632e-07, "loss": 0.0049, "step": 76670 }, { "grad_norm": 0.08996456116437912, "learning_rate": 4.704009260130371e-07, "loss": 0.0058, "step": 76680 }, { "grad_norm": 0.1299271434545517, "learning_rate": 4.675767208381587e-07, "loss": 0.0045, "step": 76690 }, { "grad_norm": 0.10284890979528427, "learning_rate": 4.647609793997032e-07, "loss": 0.0061, "step": 76700 }, { "grad_norm": 0.07580771297216415, "learning_rate": 4.619537021788023e-07, "loss": 0.0071, "step": 76710 }, { "grad_norm": 0.07691458612680435, "learning_rate": 4.5915488965515005e-07, "loss": 0.0049, "step": 76720 }, { "grad_norm": 0.13809463381767273, "learning_rate": 4.563645423069807e-07, "loss": 0.0046, "step": 76730 }, { "grad_norm": 0.044742003083229065, "learning_rate": 4.535826606110849e-07, "loss": 0.0052, "step": 76740 }, { "grad_norm": 0.13008776307106018, "learning_rate": 4.508092450428214e-07, "loss": 0.0045, "step": 76750 }, { "grad_norm": 0.10005848854780197, "learning_rate": 4.480442960760778e-07, "loss": 0.0063, "step": 76760 }, { "grad_norm": 0.1790791004896164, "learning_rate": 4.4528781418332053e-07, "loss": 0.0055, "step": 76770 }, { "grad_norm": 0.08223308622837067, "learning_rate": 4.4253979983555073e-07, "loss": 0.0051, "step": 76780 }, { "grad_norm": 0.07662277668714523, "learning_rate": 4.3980025350233154e-07, "loss": 0.0079, "step": 76790 }, { "grad_norm": 0.13249434530735016, "learning_rate": 4.370691756517664e-07, "loss": 0.0054, "step": 76800 }, { "grad_norm": 0.06264500319957733, "learning_rate": 4.3434656675053753e-07, "loss": 0.0044, "step": 76810 }, { "grad_norm": 0.13801230490207672, "learning_rate": 4.3163242726385613e-07, "loss": 0.0051, "step": 76820 }, { "grad_norm": 0.10453495383262634, "learning_rate": 4.289267576554956e-07, "loss": 0.0066, "step": 76830 }, { "grad_norm": 0.11434975266456604, "learning_rate": 4.262295583877807e-07, "loss": 0.0057, "step": 76840 }, { "grad_norm": 0.19859850406646729, "learning_rate": 4.235408299215815e-07, "loss": 0.005, "step": 76850 }, { "grad_norm": 0.10525423288345337, "learning_rate": 4.2086057271634173e-07, "loss": 0.0065, "step": 76860 }, { "grad_norm": 0.1085466593503952, "learning_rate": 4.181887872300394e-07, "loss": 0.0034, "step": 76870 }, { "grad_norm": 0.06348701566457748, "learning_rate": 4.155254739192038e-07, "loss": 0.0042, "step": 76880 }, { "grad_norm": 0.2576218247413635, "learning_rate": 4.128706332389265e-07, "loss": 0.0059, "step": 76890 }, { "grad_norm": 0.055986519902944565, "learning_rate": 4.1022426564284453e-07, "loss": 0.0058, "step": 76900 }, { "grad_norm": 0.160281240940094, "learning_rate": 4.075863715831574e-07, "loss": 0.0059, "step": 76910 }, { "grad_norm": 0.15858303010463715, "learning_rate": 4.0495695151059887e-07, "loss": 0.0044, "step": 76920 }, { "grad_norm": 0.0650760680437088, "learning_rate": 4.023360058744763e-07, "loss": 0.0046, "step": 76930 }, { "grad_norm": 0.07845142483711243, "learning_rate": 3.997235351226314e-07, "loss": 0.006, "step": 76940 }, { "grad_norm": 0.12132667750120163, "learning_rate": 3.971195397014571e-07, "loss": 0.0055, "step": 76950 }, { "grad_norm": 0.09678328037261963, "learning_rate": 3.9452402005591417e-07, "loss": 0.005, "step": 76960 }, { "grad_norm": 0.09832692891359329, "learning_rate": 3.9193697662950334e-07, "loss": 0.006, "step": 76970 }, { "grad_norm": 0.0912812128663063, "learning_rate": 3.893584098642822e-07, "loss": 0.0052, "step": 76980 }, { "grad_norm": 0.10806851089000702, "learning_rate": 3.867883202008538e-07, "loss": 0.0065, "step": 76990 }, { "grad_norm": 0.04455463960766792, "learning_rate": 3.842267080783779e-07, "loss": 0.004, "step": 77000 }, { "grad_norm": 0.09109377861022949, "learning_rate": 3.816735739345656e-07, "loss": 0.0059, "step": 77010 }, { "grad_norm": 0.21540606021881104, "learning_rate": 3.791289182056679e-07, "loss": 0.0054, "step": 77020 }, { "grad_norm": 0.11120627820491791, "learning_rate": 3.7659274132650913e-07, "loss": 0.005, "step": 77030 }, { "grad_norm": 0.0907476469874382, "learning_rate": 3.7406504373044826e-07, "loss": 0.0045, "step": 77040 }, { "grad_norm": 0.06748156994581223, "learning_rate": 3.7154582584939533e-07, "loss": 0.0056, "step": 77050 }, { "grad_norm": 0.10493823140859604, "learning_rate": 3.690350881138227e-07, "loss": 0.0051, "step": 77060 }, { "grad_norm": 0.09269355237483978, "learning_rate": 3.665328309527427e-07, "loss": 0.005, "step": 77070 }, { "grad_norm": 0.07880836725234985, "learning_rate": 3.640390547937245e-07, "loss": 0.0052, "step": 77080 }, { "grad_norm": 0.07405096292495728, "learning_rate": 3.615537600628882e-07, "loss": 0.0062, "step": 77090 }, { "grad_norm": 0.07358390092849731, "learning_rate": 3.5907694718489427e-07, "loss": 0.0054, "step": 77100 }, { "grad_norm": 0.09510383754968643, "learning_rate": 3.566086165829707e-07, "loss": 0.0059, "step": 77110 }, { "grad_norm": 0.13244320452213287, "learning_rate": 3.5414876867888024e-07, "loss": 0.0054, "step": 77120 }, { "grad_norm": 0.08162926882505417, "learning_rate": 3.5169740389295326e-07, "loss": 0.0065, "step": 77130 }, { "grad_norm": 0.07701985538005829, "learning_rate": 3.4925452264405466e-07, "loss": 0.0065, "step": 77140 }, { "grad_norm": 0.10428781807422638, "learning_rate": 3.468201253496062e-07, "loss": 0.0065, "step": 77150 }, { "grad_norm": 0.08870097994804382, "learning_rate": 3.443942124255861e-07, "loss": 0.0044, "step": 77160 }, { "grad_norm": 0.10080152750015259, "learning_rate": 3.4197678428650183e-07, "loss": 0.0053, "step": 77170 }, { "grad_norm": 0.10477650910615921, "learning_rate": 3.3956784134544504e-07, "loss": 0.0049, "step": 77180 }, { "grad_norm": 0.17078517377376556, "learning_rate": 3.3716738401402547e-07, "loss": 0.0055, "step": 77190 }, { "grad_norm": 0.09241551905870438, "learning_rate": 3.3477541270241495e-07, "loss": 0.0057, "step": 77200 }, { "grad_norm": 0.13286982476711273, "learning_rate": 3.3239192781934215e-07, "loss": 0.0074, "step": 77210 }, { "grad_norm": 0.1517079919576645, "learning_rate": 3.3001692977207563e-07, "loss": 0.0056, "step": 77220 }, { "grad_norm": 0.1284855157136917, "learning_rate": 3.276504189664409e-07, "loss": 0.0053, "step": 77230 }, { "grad_norm": 0.13080370426177979, "learning_rate": 3.252923958068088e-07, "loss": 0.0046, "step": 77240 }, { "grad_norm": 0.05067615956068039, "learning_rate": 3.2294286069609046e-07, "loss": 0.0044, "step": 77250 }, { "grad_norm": 0.141229048371315, "learning_rate": 3.2060181403577583e-07, "loss": 0.0044, "step": 77260 }, { "grad_norm": 0.1268821358680725, "learning_rate": 3.1826925622587823e-07, "loss": 0.0062, "step": 77270 }, { "grad_norm": 0.1416439712047577, "learning_rate": 3.159451876649622e-07, "loss": 0.0062, "step": 77280 }, { "grad_norm": 0.10873349756002426, "learning_rate": 3.1362960875015444e-07, "loss": 0.0054, "step": 77290 }, { "grad_norm": 0.09244469553232193, "learning_rate": 3.1132251987711637e-07, "loss": 0.0068, "step": 77300 }, { "grad_norm": 0.055492520332336426, "learning_rate": 3.0902392144007695e-07, "loss": 0.0067, "step": 77310 }, { "grad_norm": 0.08574386686086655, "learning_rate": 3.0673381383179436e-07, "loss": 0.0051, "step": 77320 }, { "grad_norm": 0.0824633315205574, "learning_rate": 3.0445219744358875e-07, "loss": 0.0054, "step": 77330 }, { "grad_norm": 0.10360068827867508, "learning_rate": 3.021790726653262e-07, "loss": 0.0036, "step": 77340 }, { "grad_norm": 0.15109877288341522, "learning_rate": 2.9991443988542366e-07, "loss": 0.007, "step": 77350 }, { "grad_norm": 0.11249243468046188, "learning_rate": 2.976582994908439e-07, "loss": 0.005, "step": 77360 }, { "grad_norm": 0.13157857954502106, "learning_rate": 2.954106518671007e-07, "loss": 0.0041, "step": 77370 }, { "grad_norm": 0.11857617646455765, "learning_rate": 2.9317149739825356e-07, "loss": 0.0051, "step": 77380 }, { "grad_norm": 0.0845552384853363, "learning_rate": 2.909408364669075e-07, "loss": 0.0037, "step": 77390 }, { "grad_norm": 0.206607386469841, "learning_rate": 2.887186694542299e-07, "loss": 0.008, "step": 77400 }, { "grad_norm": 0.05941065773367882, "learning_rate": 2.865049967399225e-07, "loss": 0.0037, "step": 77410 }, { "grad_norm": 0.0724002793431282, "learning_rate": 2.842998187022439e-07, "loss": 0.0049, "step": 77420 }, { "grad_norm": 0.12680645287036896, "learning_rate": 2.8210313571800374e-07, "loss": 0.0054, "step": 77430 }, { "grad_norm": 0.08713849633932114, "learning_rate": 2.7991494816255184e-07, "loss": 0.0047, "step": 77440 }, { "grad_norm": 0.10775147378444672, "learning_rate": 2.777352564097779e-07, "loss": 0.0064, "step": 77450 }, { "grad_norm": 0.20793399214744568, "learning_rate": 2.755640608321508e-07, "loss": 0.0058, "step": 77460 }, { "grad_norm": 0.14713075757026672, "learning_rate": 2.73401361800657e-07, "loss": 0.0048, "step": 77470 }, { "grad_norm": 0.13704311847686768, "learning_rate": 2.7124715968484537e-07, "loss": 0.0058, "step": 77480 }, { "grad_norm": 0.1455501765012741, "learning_rate": 2.691014548528104e-07, "loss": 0.005, "step": 77490 }, { "grad_norm": 0.12954574823379517, "learning_rate": 2.669642476711864e-07, "loss": 0.005, "step": 77500 }, { "grad_norm": 0.08867277204990387, "learning_rate": 2.648355385051815e-07, "loss": 0.007, "step": 77510 }, { "grad_norm": 0.05880710482597351, "learning_rate": 2.627153277185157e-07, "loss": 0.0047, "step": 77520 }, { "grad_norm": 0.14382773637771606, "learning_rate": 2.606036156734881e-07, "loss": 0.0056, "step": 77530 }, { "grad_norm": 0.07232549041509628, "learning_rate": 2.5850040273092127e-07, "loss": 0.0047, "step": 77540 }, { "grad_norm": 0.1193719133734703, "learning_rate": 2.5640568925020536e-07, "loss": 0.0055, "step": 77550 }, { "grad_norm": 0.08148100227117538, "learning_rate": 2.5431947558927083e-07, "loss": 0.0044, "step": 77560 }, { "grad_norm": 0.18216252326965332, "learning_rate": 2.522417621045825e-07, "loss": 0.0056, "step": 77570 }, { "grad_norm": 0.10959406942129135, "learning_rate": 2.501725491511786e-07, "loss": 0.004, "step": 77580 }, { "grad_norm": 0.13152547180652618, "learning_rate": 2.481118370826263e-07, "loss": 0.0053, "step": 77590 }, { "grad_norm": 0.061251211911439896, "learning_rate": 2.4605962625104393e-07, "loss": 0.0049, "step": 77600 }, { "grad_norm": 0.11271312832832336, "learning_rate": 2.440159170070955e-07, "loss": 0.0068, "step": 77610 }, { "grad_norm": 0.07263240963220596, "learning_rate": 2.41980709699996e-07, "loss": 0.0048, "step": 77620 }, { "grad_norm": 0.10164746642112732, "learning_rate": 2.3995400467751174e-07, "loss": 0.007, "step": 77630 }, { "grad_norm": 0.09776819497346878, "learning_rate": 2.3793580228594902e-07, "loss": 0.0051, "step": 77640 }, { "grad_norm": 0.10372511297464371, "learning_rate": 2.3592610287015982e-07, "loss": 0.0086, "step": 77650 }, { "grad_norm": 0.19986701011657715, "learning_rate": 2.3392490677354718e-07, "loss": 0.006, "step": 77660 }, { "grad_norm": 0.07613218575716019, "learning_rate": 2.3193221433806533e-07, "loss": 0.0051, "step": 77670 }, { "grad_norm": 0.10703667253255844, "learning_rate": 2.299480259042086e-07, "loss": 0.0071, "step": 77680 }, { "grad_norm": 0.2519834339618683, "learning_rate": 2.2797234181102244e-07, "loss": 0.0057, "step": 77690 }, { "grad_norm": 0.05773976817727089, "learning_rate": 2.2600516239609238e-07, "loss": 0.0081, "step": 77700 }, { "grad_norm": 0.10865725576877594, "learning_rate": 2.240464879955606e-07, "loss": 0.0049, "step": 77710 }, { "grad_norm": 0.1095135509967804, "learning_rate": 2.2209631894410387e-07, "loss": 0.0047, "step": 77720 }, { "grad_norm": 0.22857323288917542, "learning_rate": 2.2015465557496117e-07, "loss": 0.0074, "step": 77730 }, { "grad_norm": 0.12104000896215439, "learning_rate": 2.1822149821990602e-07, "loss": 0.0045, "step": 77740 }, { "grad_norm": 0.05599784478545189, "learning_rate": 2.1629684720926303e-07, "loss": 0.0052, "step": 77750 }, { "grad_norm": 0.1293313056230545, "learning_rate": 2.1438070287189693e-07, "loss": 0.0082, "step": 77760 }, { "grad_norm": 0.16794122755527496, "learning_rate": 2.1247306553523472e-07, "loss": 0.0063, "step": 77770 }, { "grad_norm": 0.07828082889318466, "learning_rate": 2.105739355252323e-07, "loss": 0.0063, "step": 77780 }, { "grad_norm": 0.07927355915307999, "learning_rate": 2.0868331316639678e-07, "loss": 0.0069, "step": 77790 }, { "grad_norm": 0.13175243139266968, "learning_rate": 2.0680119878179193e-07, "loss": 0.0057, "step": 77800 }, { "grad_norm": 0.06003200635313988, "learning_rate": 2.0492759269301054e-07, "loss": 0.0063, "step": 77810 }, { "grad_norm": 0.08637580275535583, "learning_rate": 2.0306249522021315e-07, "loss": 0.0074, "step": 77820 }, { "grad_norm": 0.13034726679325104, "learning_rate": 2.0120590668207816e-07, "loss": 0.007, "step": 77830 }, { "grad_norm": 0.126628577709198, "learning_rate": 1.993578273958574e-07, "loss": 0.0067, "step": 77840 }, { "grad_norm": 0.09913080185651779, "learning_rate": 1.975182576773371e-07, "loss": 0.0055, "step": 77850 }, { "grad_norm": 0.12208022177219391, "learning_rate": 1.9568719784083812e-07, "loss": 0.0077, "step": 77860 }, { "grad_norm": 0.19439220428466797, "learning_rate": 1.9386464819924898e-07, "loss": 0.006, "step": 77870 }, { "grad_norm": 0.07400994002819061, "learning_rate": 1.9205060906399285e-07, "loss": 0.005, "step": 77880 }, { "grad_norm": 0.14772897958755493, "learning_rate": 1.9024508074503845e-07, "loss": 0.0072, "step": 77890 }, { "grad_norm": 0.09030300378799438, "learning_rate": 1.8844806355089452e-07, "loss": 0.0053, "step": 77900 }, { "grad_norm": 0.046580031514167786, "learning_rate": 1.8665955778863208e-07, "loss": 0.0041, "step": 77910 }, { "grad_norm": 0.15138490498065948, "learning_rate": 1.848795637638512e-07, "loss": 0.0064, "step": 77920 }, { "grad_norm": 0.07537177950143814, "learning_rate": 1.831080817807085e-07, "loss": 0.0036, "step": 77930 }, { "grad_norm": 0.04811793193221092, "learning_rate": 1.813451121418952e-07, "loss": 0.0047, "step": 77940 }, { "grad_norm": 0.06349997967481613, "learning_rate": 1.795906551486648e-07, "loss": 0.0037, "step": 77950 }, { "grad_norm": 0.17083971202373505, "learning_rate": 1.7784471110079414e-07, "loss": 0.004, "step": 77960 }, { "grad_norm": 0.06202060729265213, "learning_rate": 1.7610728029662793e-07, "loss": 0.0057, "step": 77970 }, { "grad_norm": 0.0787787213921547, "learning_rate": 1.7437836303303979e-07, "loss": 0.0083, "step": 77980 }, { "grad_norm": 0.11338774859905243, "learning_rate": 1.726579596054545e-07, "loss": 0.0038, "step": 77990 }, { "grad_norm": 0.14158979058265686, "learning_rate": 1.7094607030784803e-07, "loss": 0.0055, "step": 78000 }, { "grad_norm": 0.10567183792591095, "learning_rate": 1.6924269543272532e-07, "loss": 0.0049, "step": 78010 }, { "grad_norm": 0.05434168502688408, "learning_rate": 1.6754783527115348e-07, "loss": 0.0038, "step": 78020 }, { "grad_norm": 0.0863022431731224, "learning_rate": 1.6586149011273422e-07, "loss": 0.0065, "step": 78030 }, { "grad_norm": 0.1750386357307434, "learning_rate": 1.6418366024562038e-07, "loss": 0.0064, "step": 78040 }, { "grad_norm": 0.09551304578781128, "learning_rate": 1.6251434595651037e-07, "loss": 0.0054, "step": 78050 }, { "grad_norm": 0.08303014934062958, "learning_rate": 1.608535475306372e-07, "loss": 0.005, "step": 78060 }, { "grad_norm": 0.09736238420009613, "learning_rate": 1.5920126525179048e-07, "loss": 0.005, "step": 78070 }, { "grad_norm": 0.07844601571559906, "learning_rate": 1.5755749940229435e-07, "loss": 0.0052, "step": 78080 }, { "grad_norm": 0.0655917078256607, "learning_rate": 1.559222502630353e-07, "loss": 0.0061, "step": 78090 }, { "grad_norm": 0.10784902423620224, "learning_rate": 1.5429551811341757e-07, "loss": 0.0044, "step": 78100 }, { "grad_norm": 0.12968234717845917, "learning_rate": 1.5267730323141882e-07, "loss": 0.0051, "step": 78110 }, { "grad_norm": 0.0482487790286541, "learning_rate": 1.5106760589353454e-07, "loss": 0.0043, "step": 78120 }, { "grad_norm": 0.11330214142799377, "learning_rate": 1.4946642637483355e-07, "loss": 0.0064, "step": 78130 }, { "grad_norm": 0.12584024667739868, "learning_rate": 1.4787376494889703e-07, "loss": 0.005, "step": 78140 }, { "grad_norm": 0.13921889662742615, "learning_rate": 1.4628962188787955e-07, "loss": 0.0046, "step": 78150 }, { "grad_norm": 0.046296488493680954, "learning_rate": 1.4471399746247006e-07, "loss": 0.0053, "step": 78160 }, { "grad_norm": 0.20671217143535614, "learning_rate": 1.4314689194188103e-07, "loss": 0.0054, "step": 78170 }, { "grad_norm": 0.10099874436855316, "learning_rate": 1.4158830559390933e-07, "loss": 0.0058, "step": 78180 }, { "grad_norm": 0.11196216940879822, "learning_rate": 1.4003823868486422e-07, "loss": 0.004, "step": 78190 }, { "grad_norm": 0.07645627111196518, "learning_rate": 1.3849669147960598e-07, "loss": 0.004, "step": 78200 }, { "grad_norm": 0.10207990556955338, "learning_rate": 1.3696366424155726e-07, "loss": 0.0048, "step": 78210 }, { "grad_norm": 0.17767933011054993, "learning_rate": 1.35439157232653e-07, "loss": 0.0051, "step": 78220 }, { "grad_norm": 0.13613997399806976, "learning_rate": 1.3392317071340144e-07, "loss": 0.0045, "step": 78230 }, { "grad_norm": 0.06664477288722992, "learning_rate": 1.3241570494283984e-07, "loss": 0.0038, "step": 78240 }, { "grad_norm": 0.14293652772903442, "learning_rate": 1.3091676017855103e-07, "loss": 0.0073, "step": 78250 }, { "grad_norm": 0.039802320301532745, "learning_rate": 1.2942633667666904e-07, "loss": 0.0043, "step": 78260 }, { "grad_norm": 0.07988359779119492, "learning_rate": 1.2794443469185679e-07, "loss": 0.004, "step": 78270 }, { "grad_norm": 0.06082756072282791, "learning_rate": 1.2647105447734508e-07, "loss": 0.0046, "step": 78280 }, { "grad_norm": 0.07443413138389587, "learning_rate": 1.2500619628488254e-07, "loss": 0.0054, "step": 78290 }, { "grad_norm": 0.07191745191812515, "learning_rate": 1.2354986036477446e-07, "loss": 0.0056, "step": 78300 }, { "grad_norm": 0.11202343553304672, "learning_rate": 1.221020469658718e-07, "loss": 0.0051, "step": 78310 }, { "grad_norm": 0.15189379453659058, "learning_rate": 1.2066275633556556e-07, "loss": 0.0056, "step": 78320 }, { "grad_norm": 0.05637203902006149, "learning_rate": 1.192319887197979e-07, "loss": 0.0061, "step": 78330 }, { "grad_norm": 0.08158161491155624, "learning_rate": 1.1780974436303438e-07, "loss": 0.0042, "step": 78340 }, { "grad_norm": 0.044274382293224335, "learning_rate": 1.1639602350830836e-07, "loss": 0.0058, "step": 78350 }, { "grad_norm": 0.11504163593053818, "learning_rate": 1.1499082639718217e-07, "loss": 0.0061, "step": 78360 }, { "grad_norm": 0.07618755102157593, "learning_rate": 1.1359415326976374e-07, "loss": 0.0068, "step": 78370 }, { "grad_norm": 0.06643044948577881, "learning_rate": 1.1220600436470663e-07, "loss": 0.0049, "step": 78380 }, { "grad_norm": 0.11661359667778015, "learning_rate": 1.1082637991920996e-07, "loss": 0.0038, "step": 78390 }, { "grad_norm": 0.11940090358257294, "learning_rate": 1.0945528016901851e-07, "loss": 0.0042, "step": 78400 }, { "grad_norm": 0.12640824913978577, "learning_rate": 1.0809270534840599e-07, "loss": 0.0069, "step": 78410 }, { "grad_norm": 0.09221626818180084, "learning_rate": 1.067386556902028e-07, "loss": 0.0051, "step": 78420 }, { "grad_norm": 0.07103197276592255, "learning_rate": 1.05393131425785e-07, "loss": 0.005, "step": 78430 }, { "grad_norm": 0.06173502281308174, "learning_rate": 1.0405613278505199e-07, "loss": 0.006, "step": 78440 }, { "grad_norm": 0.06409839540719986, "learning_rate": 1.027276599964766e-07, "loss": 0.0052, "step": 78450 }, { "grad_norm": 0.23972435295581818, "learning_rate": 1.0140771328704391e-07, "loss": 0.0057, "step": 78460 }, { "grad_norm": 0.11460128426551819, "learning_rate": 1.0009629288231237e-07, "loss": 0.0059, "step": 78470 }, { "grad_norm": 0.1484505832195282, "learning_rate": 9.879339900635276e-08, "loss": 0.0045, "step": 78480 }, { "grad_norm": 0.10744160413742065, "learning_rate": 9.74990318817981e-08, "loss": 0.0044, "step": 78490 }, { "grad_norm": 0.05205076187849045, "learning_rate": 9.621319172982701e-08, "loss": 0.0052, "step": 78500 }, { "grad_norm": 0.06191228702664375, "learning_rate": 9.493587877015264e-08, "loss": 0.0051, "step": 78510 }, { "grad_norm": 0.09689706563949585, "learning_rate": 9.366709322102263e-08, "loss": 0.0056, "step": 78520 }, { "grad_norm": 0.12192723900079727, "learning_rate": 9.240683529924688e-08, "loss": 0.0051, "step": 78530 }, { "grad_norm": 0.07988476753234863, "learning_rate": 9.115510522016979e-08, "loss": 0.0064, "step": 78540 }, { "grad_norm": 0.15095967054367065, "learning_rate": 8.991190319767583e-08, "loss": 0.0055, "step": 78550 }, { "grad_norm": 0.1285375952720642, "learning_rate": 8.867722944419509e-08, "loss": 0.0045, "step": 78560 }, { "grad_norm": 0.12593774497509003, "learning_rate": 8.745108417069214e-08, "loss": 0.0044, "step": 78570 }, { "grad_norm": 0.11333204805850983, "learning_rate": 8.623346758669381e-08, "loss": 0.0038, "step": 78580 }, { "grad_norm": 0.11047239601612091, "learning_rate": 8.502437990025037e-08, "loss": 0.0064, "step": 78590 }, { "grad_norm": 0.04478451609611511, "learning_rate": 8.382382131795764e-08, "loss": 0.005, "step": 78600 }, { "grad_norm": 0.11576522141695023, "learning_rate": 8.263179204496818e-08, "loss": 0.0073, "step": 78610 }, { "grad_norm": 0.3211509883403778, "learning_rate": 8.144829228496354e-08, "loss": 0.0077, "step": 78620 }, { "grad_norm": 0.07331927120685577, "learning_rate": 8.027332224016526e-08, "loss": 0.0059, "step": 78630 }, { "grad_norm": 0.13101992011070251, "learning_rate": 7.910688211135164e-08, "loss": 0.0051, "step": 78640 }, { "grad_norm": 0.08707757294178009, "learning_rate": 7.794897209783546e-08, "loss": 0.0059, "step": 78650 }, { "grad_norm": 0.1357351392507553, "learning_rate": 7.679959239746403e-08, "loss": 0.0052, "step": 78660 }, { "grad_norm": 0.1355416625738144, "learning_rate": 7.565874320664135e-08, "loss": 0.0058, "step": 78670 }, { "grad_norm": 0.050406407564878464, "learning_rate": 7.45264247203059e-08, "loss": 0.0056, "step": 78680 }, { "grad_norm": 0.17285503447055817, "learning_rate": 7.340263713194184e-08, "loss": 0.0058, "step": 78690 }, { "grad_norm": 0.11475246399641037, "learning_rate": 7.228738063356777e-08, "loss": 0.0063, "step": 78700 }, { "grad_norm": 0.04526393860578537, "learning_rate": 7.118065541575903e-08, "loss": 0.0068, "step": 78710 }, { "grad_norm": 0.157870352268219, "learning_rate": 7.008246166761435e-08, "loss": 0.0063, "step": 78720 }, { "grad_norm": 0.1403883546590805, "learning_rate": 6.899279957679472e-08, "loss": 0.0069, "step": 78730 }, { "grad_norm": 0.07901114225387573, "learning_rate": 6.791166932949011e-08, "loss": 0.0065, "step": 78740 }, { "grad_norm": 0.10735655575990677, "learning_rate": 6.683907111043608e-08, "loss": 0.0051, "step": 78750 }, { "grad_norm": 0.08763178437948227, "learning_rate": 6.577500510290824e-08, "loss": 0.0062, "step": 78760 }, { "grad_norm": 0.12484388798475266, "learning_rate": 6.47194714887278e-08, "loss": 0.006, "step": 78770 }, { "grad_norm": 0.14522193372249603, "learning_rate": 6.367247044825608e-08, "loss": 0.0049, "step": 78780 }, { "grad_norm": 0.12174969166517258, "learning_rate": 6.263400216039994e-08, "loss": 0.0046, "step": 78790 }, { "grad_norm": 0.053393490612506866, "learning_rate": 6.160406680260078e-08, "loss": 0.0071, "step": 78800 }, { "grad_norm": 0.19601204991340637, "learning_rate": 6.058266455084561e-08, "loss": 0.0055, "step": 78810 }, { "grad_norm": 0.13498088717460632, "learning_rate": 5.956979557967257e-08, "loss": 0.0063, "step": 78820 }, { "grad_norm": 0.0910724624991417, "learning_rate": 5.856546006214325e-08, "loss": 0.0048, "step": 78830 }, { "grad_norm": 0.08101898431777954, "learning_rate": 5.7569658169881466e-08, "loss": 0.0041, "step": 78840 }, { "grad_norm": 0.11049889773130417, "learning_rate": 5.658239007303445e-08, "loss": 0.0053, "step": 78850 }, { "grad_norm": 0.1492328643798828, "learning_rate": 5.560365594030059e-08, "loss": 0.0043, "step": 78860 }, { "grad_norm": 0.10868056118488312, "learning_rate": 5.463345593891833e-08, "loss": 0.0049, "step": 78870 }, { "grad_norm": 0.08882341533899307, "learning_rate": 5.367179023467173e-08, "loss": 0.0043, "step": 78880 }, { "grad_norm": 0.08723665773868561, "learning_rate": 5.271865899187378e-08, "loss": 0.0057, "step": 78890 }, { "grad_norm": 0.12289673835039139, "learning_rate": 5.177406237340532e-08, "loss": 0.0063, "step": 78900 }, { "grad_norm": 0.1947496384382248, "learning_rate": 5.083800054065946e-08, "loss": 0.0075, "step": 78910 }, { "grad_norm": 0.06775829195976257, "learning_rate": 4.991047365358603e-08, "loss": 0.0043, "step": 78920 }, { "grad_norm": 0.07719476521015167, "learning_rate": 4.899148187067493e-08, "loss": 0.0047, "step": 78930 }, { "grad_norm": 0.07316138595342636, "learning_rate": 4.808102534895609e-08, "loss": 0.0056, "step": 78940 }, { "grad_norm": 0.10562459379434586, "learning_rate": 4.717910424400507e-08, "loss": 0.0053, "step": 78950 }, { "grad_norm": 0.06958089023828506, "learning_rate": 4.628571870993193e-08, "loss": 0.0043, "step": 78960 }, { "grad_norm": 0.1381629854440689, "learning_rate": 4.5400868899392324e-08, "loss": 0.0053, "step": 78970 }, { "grad_norm": 0.049795426428318024, "learning_rate": 4.4524554963576436e-08, "loss": 0.0046, "step": 78980 }, { "grad_norm": 0.13701914250850677, "learning_rate": 4.36567770522367e-08, "loss": 0.004, "step": 78990 }, { "grad_norm": 0.035836782306432724, "learning_rate": 4.279753531364339e-08, "loss": 0.0044, "step": 79000 }, { "grad_norm": 0.10883336514234543, "learning_rate": 4.1946829894617955e-08, "loss": 0.0063, "step": 79010 }, { "grad_norm": 0.062465593218803406, "learning_rate": 4.11046609405219e-08, "loss": 0.0072, "step": 79020 }, { "grad_norm": 0.040978312492370605, "learning_rate": 4.027102859526233e-08, "loss": 0.0047, "step": 79030 }, { "grad_norm": 0.09773046523332596, "learning_rate": 3.9445933001280856e-08, "loss": 0.005, "step": 79040 }, { "grad_norm": 0.08902810513973236, "learning_rate": 3.8629374299564704e-08, "loss": 0.0068, "step": 79050 }, { "grad_norm": 0.05772840231657028, "learning_rate": 3.78213526296467e-08, "loss": 0.005, "step": 79060 }, { "grad_norm": 0.08125336468219757, "learning_rate": 3.702186812958308e-08, "loss": 0.0079, "step": 79070 }, { "grad_norm": 0.10887984931468964, "learning_rate": 3.623092093599789e-08, "loss": 0.0056, "step": 79080 }, { "grad_norm": 0.12393798679113388, "learning_rate": 3.544851118403303e-08, "loss": 0.0044, "step": 79090 }, { "grad_norm": 0.06856497377157211, "learning_rate": 3.46746390073871e-08, "loss": 0.0053, "step": 79100 }, { "grad_norm": 0.14647303521633148, "learning_rate": 3.390930453829322e-08, "loss": 0.0067, "step": 79110 }, { "grad_norm": 0.06153992936015129, "learning_rate": 3.3152507907519005e-08, "loss": 0.0047, "step": 79120 }, { "grad_norm": 0.14797189831733704, "learning_rate": 3.240424924438323e-08, "loss": 0.0058, "step": 79130 }, { "grad_norm": 0.05029601976275444, "learning_rate": 3.1664528676750295e-08, "loss": 0.0076, "step": 79140 }, { "grad_norm": 0.0640043392777443, "learning_rate": 3.093334633100797e-08, "loss": 0.0056, "step": 79150 }, { "grad_norm": 0.09108184278011322, "learning_rate": 3.021070233210077e-08, "loss": 0.0046, "step": 79160 }, { "grad_norm": 0.09192230552434921, "learning_rate": 2.9496596803507692e-08, "loss": 0.0053, "step": 79170 }, { "grad_norm": 0.14705008268356323, "learning_rate": 2.879102986725335e-08, "loss": 0.0052, "step": 79180 }, { "grad_norm": 0.11421039700508118, "learning_rate": 2.809400164389131e-08, "loss": 0.0057, "step": 79190 }, { "grad_norm": 0.12857244908809662, "learning_rate": 2.740551225253185e-08, "loss": 0.0069, "step": 79200 }, { "grad_norm": 0.10856537520885468, "learning_rate": 2.6725561810819754e-08, "loss": 0.0048, "step": 79210 }, { "grad_norm": 0.11232699453830719, "learning_rate": 2.605415043493431e-08, "loss": 0.0058, "step": 79220 }, { "grad_norm": 0.16139885783195496, "learning_rate": 2.5391278239605965e-08, "loss": 0.0049, "step": 79230 }, { "grad_norm": 0.10120993852615356, "learning_rate": 2.4736945338094118e-08, "loss": 0.0081, "step": 79240 }, { "grad_norm": 0.061301037669181824, "learning_rate": 2.4091151842214887e-08, "loss": 0.004, "step": 79250 }, { "grad_norm": 0.08654067665338516, "learning_rate": 2.3453897862318884e-08, "loss": 0.0063, "step": 79260 }, { "grad_norm": 0.05904562771320343, "learning_rate": 2.2825183507285686e-08, "loss": 0.0042, "step": 79270 }, { "grad_norm": 0.12105465680360794, "learning_rate": 2.220500888455157e-08, "loss": 0.005, "step": 79280 }, { "grad_norm": 0.10650473833084106, "learning_rate": 2.1593374100081777e-08, "loss": 0.0048, "step": 79290 }, { "grad_norm": 0.23578602075576782, "learning_rate": 2.09902792583927e-08, "loss": 0.0072, "step": 79300 }, { "grad_norm": 0.10122404247522354, "learning_rate": 2.0395724462540788e-08, "loss": 0.0065, "step": 79310 }, { "grad_norm": 0.1361081898212433, "learning_rate": 1.9809709814111453e-08, "loss": 0.0052, "step": 79320 }, { "grad_norm": 0.16809368133544922, "learning_rate": 1.923223541324126e-08, "loss": 0.0066, "step": 79330 }, { "grad_norm": 0.12266959249973297, "learning_rate": 1.8663301358606833e-08, "loss": 0.0067, "step": 79340 }, { "grad_norm": 0.07463930547237396, "learning_rate": 1.8102907747419295e-08, "loss": 0.0057, "step": 79350 }, { "grad_norm": 0.09495095908641815, "learning_rate": 1.7551054675435385e-08, "loss": 0.0067, "step": 79360 }, { "grad_norm": 0.1447342038154602, "learning_rate": 1.7007742236957447e-08, "loss": 0.0053, "step": 79370 }, { "grad_norm": 0.10184116661548615, "learning_rate": 1.647297052481678e-08, "loss": 0.0067, "step": 79380 }, { "grad_norm": 0.09167737513780594, "learning_rate": 1.5946739630390285e-08, "loss": 0.0063, "step": 79390 }, { "grad_norm": 0.10460098087787628, "learning_rate": 1.5429049643606032e-08, "loss": 0.0056, "step": 79400 }, { "grad_norm": 0.09164349734783173, "learning_rate": 1.4919900652909935e-08, "loss": 0.0052, "step": 79410 }, { "grad_norm": 0.09084013849496841, "learning_rate": 1.4419292745310175e-08, "loss": 0.0062, "step": 79420 }, { "grad_norm": 0.11355207115411758, "learning_rate": 1.3927226006343885e-08, "loss": 0.0066, "step": 79430 }, { "grad_norm": 0.16964998841285706, "learning_rate": 1.3443700520093805e-08, "loss": 0.0047, "step": 79440 }, { "grad_norm": 0.15721844136714935, "learning_rate": 1.296871636917718e-08, "loss": 0.0052, "step": 79450 }, { "grad_norm": 0.054153986275196075, "learning_rate": 1.2502273634762419e-08, "loss": 0.0058, "step": 79460 }, { "grad_norm": 0.039669983088970184, "learning_rate": 1.2044372396546876e-08, "loss": 0.008, "step": 79470 }, { "grad_norm": 0.17784483730793, "learning_rate": 1.1595012732773524e-08, "loss": 0.0072, "step": 79480 }, { "grad_norm": 0.04886510595679283, "learning_rate": 1.1154194720225386e-08, "loss": 0.0042, "step": 79490 }, { "grad_norm": 0.06971073895692825, "learning_rate": 1.0721918434231093e-08, "loss": 0.0042, "step": 79500 }, { "grad_norm": 0.06226083263754845, "learning_rate": 1.0298183948648232e-08, "loss": 0.0047, "step": 79510 }, { "grad_norm": 0.11411821842193604, "learning_rate": 9.88299133588555e-09, "loss": 0.0061, "step": 79520 }, { "grad_norm": 0.10211104154586792, "learning_rate": 9.476340666891847e-09, "loss": 0.005, "step": 79530 }, { "grad_norm": 0.051422882825136185, "learning_rate": 9.078232011139332e-09, "loss": 0.0055, "step": 79540 }, { "grad_norm": 0.1353507786989212, "learning_rate": 8.68866543666802e-09, "loss": 0.0047, "step": 79550 }, { "grad_norm": 0.07961952686309814, "learning_rate": 8.307641010035782e-09, "loss": 0.0051, "step": 79560 }, { "grad_norm": 0.07475782185792923, "learning_rate": 7.93515879635165e-09, "loss": 0.0056, "step": 79570 }, { "grad_norm": 0.06682596355676651, "learning_rate": 7.571218859264706e-09, "loss": 0.0058, "step": 79580 }, { "grad_norm": 0.08887069672346115, "learning_rate": 7.215821260958544e-09, "loss": 0.0058, "step": 79590 }, { "grad_norm": 0.046809181571006775, "learning_rate": 6.868966062162363e-09, "loss": 0.007, "step": 79600 }, { "grad_norm": 0.11633381992578506, "learning_rate": 6.530653322145419e-09, "loss": 0.0053, "step": 79610 }, { "grad_norm": 0.08757990598678589, "learning_rate": 6.200883098717025e-09, "loss": 0.0054, "step": 79620 }, { "grad_norm": 0.12480597198009491, "learning_rate": 5.879655448226551e-09, "loss": 0.0066, "step": 79630 }, { "grad_norm": 0.1428062468767166, "learning_rate": 5.566970425557872e-09, "loss": 0.0076, "step": 79640 }, { "grad_norm": 0.14014340937137604, "learning_rate": 5.26282808414047e-09, "loss": 0.0052, "step": 79650 }, { "grad_norm": 0.14742130041122437, "learning_rate": 4.967228475949437e-09, "loss": 0.0056, "step": 79660 }, { "grad_norm": 0.11718598753213882, "learning_rate": 4.680171651494369e-09, "loss": 0.006, "step": 79670 }, { "grad_norm": 0.09281902015209198, "learning_rate": 4.40165765981937e-09, "loss": 0.0042, "step": 79680 }, { "grad_norm": 0.07989918440580368, "learning_rate": 4.131686548519698e-09, "loss": 0.0061, "step": 79690 }, { "grad_norm": 0.053028132766485214, "learning_rate": 3.8702583637251214e-09, "loss": 0.0056, "step": 79700 }, { "grad_norm": 0.08180173486471176, "learning_rate": 3.617373150105463e-09, "loss": 0.0057, "step": 79710 }, { "grad_norm": 0.07160034775733948, "learning_rate": 3.3730309508706036e-09, "loss": 0.0047, "step": 79720 }, { "grad_norm": 0.08779224753379822, "learning_rate": 3.137231807781582e-09, "loss": 0.0044, "step": 79730 }, { "grad_norm": 0.07094796746969223, "learning_rate": 2.9099757611172894e-09, "loss": 0.0042, "step": 79740 }, { "grad_norm": 0.06388375908136368, "learning_rate": 2.6912628497133275e-09, "loss": 0.0048, "step": 79750 }, { "grad_norm": 0.0941222533583641, "learning_rate": 2.481093110945354e-09, "loss": 0.0047, "step": 79760 }, { "grad_norm": 0.06935448199510574, "learning_rate": 2.2794665807235327e-09, "loss": 0.004, "step": 79770 }, { "grad_norm": 0.06484304368495941, "learning_rate": 2.0863832934980843e-09, "loss": 0.0042, "step": 79780 }, { "grad_norm": 0.22177167236804962, "learning_rate": 1.901843282264837e-09, "loss": 0.0077, "step": 79790 }, { "grad_norm": 0.0663192868232727, "learning_rate": 1.7258465785541245e-09, "loss": 0.0052, "step": 79800 }, { "grad_norm": 0.09204474836587906, "learning_rate": 1.5583932124418887e-09, "loss": 0.0048, "step": 79810 }, { "grad_norm": 0.1204412505030632, "learning_rate": 1.3994832125441282e-09, "loss": 0.0051, "step": 79820 }, { "grad_norm": 0.09507245570421219, "learning_rate": 1.2491166060057958e-09, "loss": 0.0044, "step": 79830 }, { "grad_norm": 0.16025109589099884, "learning_rate": 1.1072934185230032e-09, "loss": 0.005, "step": 79840 }, { "grad_norm": 0.14352941513061523, "learning_rate": 9.740136743319195e-10, "loss": 0.0034, "step": 79850 }, { "grad_norm": 0.12535881996154785, "learning_rate": 8.492773962087696e-10, "loss": 0.0063, "step": 79860 }, { "grad_norm": 0.11646811664104462, "learning_rate": 7.330846054587337e-10, "loss": 0.0045, "step": 79870 }, { "grad_norm": 0.08881676942110062, "learning_rate": 6.254353219492526e-10, "loss": 0.0058, "step": 79880 }, { "grad_norm": 0.10133565962314606, "learning_rate": 5.263295640600685e-10, "loss": 0.0047, "step": 79890 }, { "grad_norm": 0.05970258638262749, "learning_rate": 4.357673487387359e-10, "loss": 0.0053, "step": 79900 }, { "grad_norm": 0.08171312510967255, "learning_rate": 3.537486914506616e-10, "loss": 0.0055, "step": 79910 }, { "grad_norm": 0.09038796275854111, "learning_rate": 2.8027360621241117e-10, "loss": 0.0058, "step": 79920 }, { "grad_norm": 0.0941799059510231, "learning_rate": 2.153421055806071e-10, "loss": 0.0063, "step": 79930 }, { "grad_norm": 0.0876714438199997, "learning_rate": 1.5895420064637733e-10, "loss": 0.005, "step": 79940 }, { "grad_norm": 0.19313231110572815, "learning_rate": 1.1110990105200891e-10, "loss": 0.0087, "step": 79950 }, { "grad_norm": 0.13903388381004333, "learning_rate": 7.180921496874326e-11, "loss": 0.0087, "step": 79960 }, { "grad_norm": 0.12260116636753082, "learning_rate": 4.10521491134297e-11, "loss": 0.0062, "step": 79970 }, { "grad_norm": 0.06565508246421814, "learning_rate": 1.883870874297422e-11, "loss": 0.0049, "step": 79980 }, { "grad_norm": 0.22094318270683289, "learning_rate": 5.1688976432373584e-12, "loss": 0.0072, "step": 79990 }, { "grad_norm": 0.07973551005125046, "learning_rate": 4.271816234080461e-14, "loss": 0.0056, "step": 80000 } ], "logging_steps": 10, "max_steps": 80000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 40000, "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": 32, "trial_name": null, "trial_params": null }