{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 100000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 1.7663888931274414, "learning_rate": 1.8e-07, "loss": 1.2322, "step": 10 }, { "grad_norm": 1.7530381679534912, "learning_rate": 3.8e-07, "loss": 1.235, "step": 20 }, { "grad_norm": 2.104269504547119, "learning_rate": 5.8e-07, "loss": 1.2686, "step": 30 }, { "grad_norm": 1.9471431970596313, "learning_rate": 7.8e-07, "loss": 1.2693, "step": 40 }, { "grad_norm": 1.4843591451644897, "learning_rate": 9.8e-07, "loss": 1.2441, "step": 50 }, { "grad_norm": 1.0599344968795776, "learning_rate": 1.18e-06, "loss": 1.1953, "step": 60 }, { "grad_norm": 1.042344570159912, "learning_rate": 1.3800000000000001e-06, "loss": 1.166, "step": 70 }, { "grad_norm": 0.9658252000808716, "learning_rate": 1.5800000000000003e-06, "loss": 1.195, "step": 80 }, { "grad_norm": 0.8095623254776001, "learning_rate": 1.7800000000000001e-06, "loss": 1.1611, "step": 90 }, { "grad_norm": 0.7211509943008423, "learning_rate": 1.98e-06, "loss": 1.1381, "step": 100 }, { "grad_norm": 0.6569139957427979, "learning_rate": 2.1800000000000003e-06, "loss": 1.1095, "step": 110 }, { "grad_norm": 1.7389661073684692, "learning_rate": 2.38e-06, "loss": 1.2229, "step": 120 }, { "grad_norm": 1.184212565422058, "learning_rate": 2.5800000000000003e-06, "loss": 1.1703, "step": 130 }, { "grad_norm": 0.6066526770591736, "learning_rate": 2.78e-06, "loss": 1.1304, "step": 140 }, { "grad_norm": 0.3084378242492676, "learning_rate": 2.9800000000000003e-06, "loss": 1.1502, "step": 150 }, { "grad_norm": 0.3719276785850525, "learning_rate": 3.1800000000000005e-06, "loss": 1.139, "step": 160 }, { "grad_norm": 0.33172929286956787, "learning_rate": 3.38e-06, "loss": 1.1311, "step": 170 }, { "grad_norm": 0.7391734719276428, "learning_rate": 3.58e-06, "loss": 1.1194, "step": 180 }, { "grad_norm": 0.5027369260787964, "learning_rate": 3.7800000000000002e-06, "loss": 1.0866, "step": 190 }, { "grad_norm": 0.37241944670677185, "learning_rate": 3.98e-06, "loss": 1.0759, "step": 200 }, { "grad_norm": 0.7735804319381714, "learning_rate": 4.18e-06, "loss": 1.1244, "step": 210 }, { "grad_norm": 0.373465359210968, "learning_rate": 4.38e-06, "loss": 1.0969, "step": 220 }, { "grad_norm": 0.31204044818878174, "learning_rate": 4.58e-06, "loss": 1.0726, "step": 230 }, { "grad_norm": 0.31513532996177673, "learning_rate": 4.780000000000001e-06, "loss": 1.0674, "step": 240 }, { "grad_norm": 0.7387045621871948, "learning_rate": 4.98e-06, "loss": 1.1047, "step": 250 }, { "grad_norm": 1.416513442993164, "learning_rate": 5.18e-06, "loss": 1.1011, "step": 260 }, { "grad_norm": 0.4549804925918579, "learning_rate": 5.38e-06, "loss": 1.1019, "step": 270 }, { "grad_norm": 0.9581396579742432, "learning_rate": 5.580000000000001e-06, "loss": 1.0979, "step": 280 }, { "grad_norm": 0.45345205068588257, "learning_rate": 5.78e-06, "loss": 1.0845, "step": 290 }, { "grad_norm": 1.2547475099563599, "learning_rate": 5.98e-06, "loss": 1.0884, "step": 300 }, { "grad_norm": 0.379281610250473, "learning_rate": 6.18e-06, "loss": 1.0688, "step": 310 }, { "grad_norm": 1.1736204624176025, "learning_rate": 6.38e-06, "loss": 1.1252, "step": 320 }, { "grad_norm": 0.32833167910575867, "learning_rate": 6.58e-06, "loss": 1.056, "step": 330 }, { "grad_norm": 0.3011569082736969, "learning_rate": 6.78e-06, "loss": 1.0391, "step": 340 }, { "grad_norm": 1.0326855182647705, "learning_rate": 6.98e-06, "loss": 1.1636, "step": 350 }, { "grad_norm": 0.3672337830066681, "learning_rate": 7.180000000000001e-06, "loss": 1.0686, "step": 360 }, { "grad_norm": 0.8287212252616882, "learning_rate": 7.3800000000000005e-06, "loss": 1.0863, "step": 370 }, { "grad_norm": 1.3298174142837524, "learning_rate": 7.580000000000001e-06, "loss": 1.072, "step": 380 }, { "grad_norm": 0.41273626685142517, "learning_rate": 7.78e-06, "loss": 1.08, "step": 390 }, { "grad_norm": 0.3591606616973877, "learning_rate": 7.98e-06, "loss": 1.0559, "step": 400 }, { "grad_norm": 0.2763373553752899, "learning_rate": 8.18e-06, "loss": 1.0381, "step": 410 }, { "grad_norm": 0.9213012456893921, "learning_rate": 8.380000000000001e-06, "loss": 1.0544, "step": 420 }, { "grad_norm": 0.4057427942752838, "learning_rate": 8.580000000000001e-06, "loss": 1.0456, "step": 430 }, { "grad_norm": 0.7411806583404541, "learning_rate": 8.78e-06, "loss": 1.0596, "step": 440 }, { "grad_norm": 0.49361637234687805, "learning_rate": 8.98e-06, "loss": 1.0198, "step": 450 }, { "grad_norm": 0.6992295980453491, "learning_rate": 9.180000000000002e-06, "loss": 0.9904, "step": 460 }, { "grad_norm": 1.4632469415664673, "learning_rate": 9.38e-06, "loss": 0.9754, "step": 470 }, { "grad_norm": 1.037854552268982, "learning_rate": 9.58e-06, "loss": 0.9714, "step": 480 }, { "grad_norm": 2.027381420135498, "learning_rate": 9.78e-06, "loss": 0.9304, "step": 490 }, { "grad_norm": 0.9999592900276184, "learning_rate": 9.980000000000001e-06, "loss": 0.9452, "step": 500 }, { "grad_norm": 1.3927254676818848, "learning_rate": 1.018e-05, "loss": 0.9084, "step": 510 }, { "grad_norm": 0.9308933615684509, "learning_rate": 1.038e-05, "loss": 0.9057, "step": 520 }, { "grad_norm": 0.7903367877006531, "learning_rate": 1.058e-05, "loss": 0.8721, "step": 530 }, { "grad_norm": 1.1082496643066406, "learning_rate": 1.0780000000000002e-05, "loss": 0.8485, "step": 540 }, { "grad_norm": 1.7645108699798584, "learning_rate": 1.098e-05, "loss": 0.804, "step": 550 }, { "grad_norm": 1.2326195240020752, "learning_rate": 1.118e-05, "loss": 0.7894, "step": 560 }, { "grad_norm": 1.4467304944992065, "learning_rate": 1.1380000000000001e-05, "loss": 0.7744, "step": 570 }, { "grad_norm": 1.5679982900619507, "learning_rate": 1.1580000000000001e-05, "loss": 0.7414, "step": 580 }, { "grad_norm": 1.2100732326507568, "learning_rate": 1.178e-05, "loss": 0.6984, "step": 590 }, { "grad_norm": 1.5589553117752075, "learning_rate": 1.198e-05, "loss": 0.6645, "step": 600 }, { "grad_norm": 1.4419751167297363, "learning_rate": 1.2180000000000002e-05, "loss": 0.6, "step": 610 }, { "grad_norm": 1.2788368463516235, "learning_rate": 1.238e-05, "loss": 0.5578, "step": 620 }, { "grad_norm": 1.693923830986023, "learning_rate": 1.258e-05, "loss": 0.5108, "step": 630 }, { "grad_norm": 1.9621732234954834, "learning_rate": 1.278e-05, "loss": 0.4981, "step": 640 }, { "grad_norm": 1.7812515497207642, "learning_rate": 1.2980000000000001e-05, "loss": 0.4545, "step": 650 }, { "grad_norm": 1.6232179403305054, "learning_rate": 1.3180000000000001e-05, "loss": 0.4277, "step": 660 }, { "grad_norm": 1.8778806924819946, "learning_rate": 1.338e-05, "loss": 0.3963, "step": 670 }, { "grad_norm": 1.9603911638259888, "learning_rate": 1.358e-05, "loss": 0.3735, "step": 680 }, { "grad_norm": 2.148573398590088, "learning_rate": 1.3780000000000002e-05, "loss": 0.3631, "step": 690 }, { "grad_norm": 1.529474139213562, "learning_rate": 1.3980000000000002e-05, "loss": 0.3358, "step": 700 }, { "grad_norm": 2.4208765029907227, "learning_rate": 1.4180000000000001e-05, "loss": 0.3223, "step": 710 }, { "grad_norm": 1.864573359489441, "learning_rate": 1.4380000000000001e-05, "loss": 0.2881, "step": 720 }, { "grad_norm": 1.9378488063812256, "learning_rate": 1.4580000000000003e-05, "loss": 0.2583, "step": 730 }, { "grad_norm": 1.924202799797058, "learning_rate": 1.4779999999999999e-05, "loss": 0.2633, "step": 740 }, { "grad_norm": 1.923717975616455, "learning_rate": 1.4979999999999999e-05, "loss": 0.2531, "step": 750 }, { "grad_norm": 1.8315948247909546, "learning_rate": 1.518e-05, "loss": 0.2325, "step": 760 }, { "grad_norm": 1.9340721368789673, "learning_rate": 1.538e-05, "loss": 0.2339, "step": 770 }, { "grad_norm": 1.935090184211731, "learning_rate": 1.558e-05, "loss": 0.2098, "step": 780 }, { "grad_norm": 1.3275562524795532, "learning_rate": 1.578e-05, "loss": 0.1996, "step": 790 }, { "grad_norm": 1.8103740215301514, "learning_rate": 1.598e-05, "loss": 0.1972, "step": 800 }, { "grad_norm": 1.473107099533081, "learning_rate": 1.618e-05, "loss": 0.1901, "step": 810 }, { "grad_norm": 2.6445741653442383, "learning_rate": 1.6380000000000002e-05, "loss": 0.1863, "step": 820 }, { "grad_norm": 1.712715744972229, "learning_rate": 1.658e-05, "loss": 0.1767, "step": 830 }, { "grad_norm": 1.4788618087768555, "learning_rate": 1.6780000000000002e-05, "loss": 0.1507, "step": 840 }, { "grad_norm": 2.7250616550445557, "learning_rate": 1.698e-05, "loss": 0.1429, "step": 850 }, { "grad_norm": 2.3308191299438477, "learning_rate": 1.718e-05, "loss": 0.1419, "step": 860 }, { "grad_norm": 1.9587215185165405, "learning_rate": 1.7380000000000003e-05, "loss": 0.1431, "step": 870 }, { "grad_norm": 1.407806396484375, "learning_rate": 1.758e-05, "loss": 0.1348, "step": 880 }, { "grad_norm": 1.976709008216858, "learning_rate": 1.7780000000000003e-05, "loss": 0.1262, "step": 890 }, { "grad_norm": 1.982813835144043, "learning_rate": 1.798e-05, "loss": 0.1111, "step": 900 }, { "grad_norm": 1.7246272563934326, "learning_rate": 1.818e-05, "loss": 0.1029, "step": 910 }, { "grad_norm": 1.7602629661560059, "learning_rate": 1.838e-05, "loss": 0.1124, "step": 920 }, { "grad_norm": 1.8974276781082153, "learning_rate": 1.858e-05, "loss": 0.1106, "step": 930 }, { "grad_norm": 1.7436341047286987, "learning_rate": 1.878e-05, "loss": 0.1224, "step": 940 }, { "grad_norm": 1.3792669773101807, "learning_rate": 1.898e-05, "loss": 0.1131, "step": 950 }, { "grad_norm": 1.656153678894043, "learning_rate": 1.918e-05, "loss": 0.1215, "step": 960 }, { "grad_norm": 1.7466992139816284, "learning_rate": 1.938e-05, "loss": 0.112, "step": 970 }, { "grad_norm": 1.7255088090896606, "learning_rate": 1.9580000000000002e-05, "loss": 0.1092, "step": 980 }, { "grad_norm": 1.693135142326355, "learning_rate": 1.978e-05, "loss": 0.1057, "step": 990 }, { "grad_norm": 2.1449785232543945, "learning_rate": 1.9980000000000002e-05, "loss": 0.1046, "step": 1000 }, { "grad_norm": 1.778427243232727, "learning_rate": 2.0180000000000003e-05, "loss": 0.1064, "step": 1010 }, { "grad_norm": 1.3118313550949097, "learning_rate": 2.038e-05, "loss": 0.1053, "step": 1020 }, { "grad_norm": 1.3776055574417114, "learning_rate": 2.0580000000000003e-05, "loss": 0.1037, "step": 1030 }, { "grad_norm": 1.4827171564102173, "learning_rate": 2.078e-05, "loss": 0.103, "step": 1040 }, { "grad_norm": 1.581317663192749, "learning_rate": 2.098e-05, "loss": 0.0983, "step": 1050 }, { "grad_norm": 1.3623675107955933, "learning_rate": 2.118e-05, "loss": 0.0905, "step": 1060 }, { "grad_norm": 1.3035303354263306, "learning_rate": 2.138e-05, "loss": 0.0873, "step": 1070 }, { "grad_norm": 1.7198104858398438, "learning_rate": 2.158e-05, "loss": 0.0873, "step": 1080 }, { "grad_norm": 1.6970012187957764, "learning_rate": 2.178e-05, "loss": 0.0944, "step": 1090 }, { "grad_norm": 2.0383119583129883, "learning_rate": 2.198e-05, "loss": 0.093, "step": 1100 }, { "grad_norm": 1.184504747390747, "learning_rate": 2.218e-05, "loss": 0.0823, "step": 1110 }, { "grad_norm": 1.75649094581604, "learning_rate": 2.2380000000000003e-05, "loss": 0.0977, "step": 1120 }, { "grad_norm": 1.84686279296875, "learning_rate": 2.258e-05, "loss": 0.092, "step": 1130 }, { "grad_norm": 1.7380592823028564, "learning_rate": 2.2780000000000002e-05, "loss": 0.0972, "step": 1140 }, { "grad_norm": 1.0202646255493164, "learning_rate": 2.298e-05, "loss": 0.0932, "step": 1150 }, { "grad_norm": 1.5624462366104126, "learning_rate": 2.318e-05, "loss": 0.0801, "step": 1160 }, { "grad_norm": 1.7367807626724243, "learning_rate": 2.3380000000000003e-05, "loss": 0.0935, "step": 1170 }, { "grad_norm": 1.7396093606948853, "learning_rate": 2.358e-05, "loss": 0.0898, "step": 1180 }, { "grad_norm": 1.6315388679504395, "learning_rate": 2.3780000000000003e-05, "loss": 0.0886, "step": 1190 }, { "grad_norm": 1.270431637763977, "learning_rate": 2.398e-05, "loss": 0.0878, "step": 1200 }, { "grad_norm": 1.5271105766296387, "learning_rate": 2.418e-05, "loss": 0.073, "step": 1210 }, { "grad_norm": 1.3546088933944702, "learning_rate": 2.438e-05, "loss": 0.0949, "step": 1220 }, { "grad_norm": 1.7466727495193481, "learning_rate": 2.4580000000000002e-05, "loss": 0.085, "step": 1230 }, { "grad_norm": 1.5026220083236694, "learning_rate": 2.478e-05, "loss": 0.1001, "step": 1240 }, { "grad_norm": 1.451074242591858, "learning_rate": 2.498e-05, "loss": 0.0913, "step": 1250 }, { "grad_norm": 1.0659507513046265, "learning_rate": 2.5180000000000003e-05, "loss": 0.081, "step": 1260 }, { "grad_norm": 1.5934942960739136, "learning_rate": 2.5380000000000004e-05, "loss": 0.0929, "step": 1270 }, { "grad_norm": 1.4142900705337524, "learning_rate": 2.5580000000000002e-05, "loss": 0.0881, "step": 1280 }, { "grad_norm": 1.4800379276275635, "learning_rate": 2.5779999999999997e-05, "loss": 0.0817, "step": 1290 }, { "grad_norm": 2.670126438140869, "learning_rate": 2.598e-05, "loss": 0.082, "step": 1300 }, { "grad_norm": 1.5576353073120117, "learning_rate": 2.618e-05, "loss": 0.0861, "step": 1310 }, { "grad_norm": 1.2551933526992798, "learning_rate": 2.6379999999999998e-05, "loss": 0.0921, "step": 1320 }, { "grad_norm": 1.2044438123703003, "learning_rate": 2.658e-05, "loss": 0.0774, "step": 1330 }, { "grad_norm": 1.2778264284133911, "learning_rate": 2.678e-05, "loss": 0.0681, "step": 1340 }, { "grad_norm": 0.9613584280014038, "learning_rate": 2.698e-05, "loss": 0.0798, "step": 1350 }, { "grad_norm": 1.1202353239059448, "learning_rate": 2.718e-05, "loss": 0.0785, "step": 1360 }, { "grad_norm": 1.7635245323181152, "learning_rate": 2.738e-05, "loss": 0.0933, "step": 1370 }, { "grad_norm": 1.2184455394744873, "learning_rate": 2.758e-05, "loss": 0.0709, "step": 1380 }, { "grad_norm": 1.222536325454712, "learning_rate": 2.778e-05, "loss": 0.0643, "step": 1390 }, { "grad_norm": 1.092502236366272, "learning_rate": 2.798e-05, "loss": 0.0911, "step": 1400 }, { "grad_norm": 0.9986786842346191, "learning_rate": 2.818e-05, "loss": 0.0747, "step": 1410 }, { "grad_norm": 0.839569091796875, "learning_rate": 2.8380000000000003e-05, "loss": 0.0686, "step": 1420 }, { "grad_norm": 1.6430128812789917, "learning_rate": 2.858e-05, "loss": 0.0791, "step": 1430 }, { "grad_norm": 1.4924696683883667, "learning_rate": 2.8780000000000002e-05, "loss": 0.0785, "step": 1440 }, { "grad_norm": 0.884482204914093, "learning_rate": 2.898e-05, "loss": 0.0714, "step": 1450 }, { "grad_norm": 1.0814307928085327, "learning_rate": 2.9180000000000002e-05, "loss": 0.0763, "step": 1460 }, { "grad_norm": 1.1131699085235596, "learning_rate": 2.9380000000000003e-05, "loss": 0.0654, "step": 1470 }, { "grad_norm": 0.939599335193634, "learning_rate": 2.958e-05, "loss": 0.0642, "step": 1480 }, { "grad_norm": 1.606095790863037, "learning_rate": 2.9780000000000003e-05, "loss": 0.0837, "step": 1490 }, { "grad_norm": 1.0807520151138306, "learning_rate": 2.998e-05, "loss": 0.0714, "step": 1500 }, { "grad_norm": 1.371087908744812, "learning_rate": 3.0180000000000002e-05, "loss": 0.0817, "step": 1510 }, { "grad_norm": 1.0831212997436523, "learning_rate": 3.0380000000000004e-05, "loss": 0.0734, "step": 1520 }, { "grad_norm": 1.2450683116912842, "learning_rate": 3.058e-05, "loss": 0.0782, "step": 1530 }, { "grad_norm": 1.1932432651519775, "learning_rate": 3.078e-05, "loss": 0.0764, "step": 1540 }, { "grad_norm": 0.979270339012146, "learning_rate": 3.0980000000000005e-05, "loss": 0.0778, "step": 1550 }, { "grad_norm": 1.4223650693893433, "learning_rate": 3.118e-05, "loss": 0.0713, "step": 1560 }, { "grad_norm": 1.1900421380996704, "learning_rate": 3.138e-05, "loss": 0.079, "step": 1570 }, { "grad_norm": 1.3434370756149292, "learning_rate": 3.1580000000000006e-05, "loss": 0.0691, "step": 1580 }, { "grad_norm": 0.8439323902130127, "learning_rate": 3.1780000000000004e-05, "loss": 0.068, "step": 1590 }, { "grad_norm": 1.031024694442749, "learning_rate": 3.198e-05, "loss": 0.0671, "step": 1600 }, { "grad_norm": 2.4049487113952637, "learning_rate": 3.218e-05, "loss": 0.0743, "step": 1610 }, { "grad_norm": 1.3034881353378296, "learning_rate": 3.238e-05, "loss": 0.0785, "step": 1620 }, { "grad_norm": 1.214401364326477, "learning_rate": 3.2579999999999996e-05, "loss": 0.0823, "step": 1630 }, { "grad_norm": 1.1928775310516357, "learning_rate": 3.278e-05, "loss": 0.0744, "step": 1640 }, { "grad_norm": 1.0030676126480103, "learning_rate": 3.298e-05, "loss": 0.0729, "step": 1650 }, { "grad_norm": 0.8322781920433044, "learning_rate": 3.318e-05, "loss": 0.0657, "step": 1660 }, { "grad_norm": 1.3139138221740723, "learning_rate": 3.338e-05, "loss": 0.0745, "step": 1670 }, { "grad_norm": 2.221327781677246, "learning_rate": 3.358e-05, "loss": 0.0667, "step": 1680 }, { "grad_norm": 1.09636652469635, "learning_rate": 3.378e-05, "loss": 0.0718, "step": 1690 }, { "grad_norm": 0.987312912940979, "learning_rate": 3.398e-05, "loss": 0.0709, "step": 1700 }, { "grad_norm": 0.9517627954483032, "learning_rate": 3.418e-05, "loss": 0.0607, "step": 1710 }, { "grad_norm": 0.9335981011390686, "learning_rate": 3.438e-05, "loss": 0.0692, "step": 1720 }, { "grad_norm": 1.0573331117630005, "learning_rate": 3.4580000000000004e-05, "loss": 0.0765, "step": 1730 }, { "grad_norm": 1.046739101409912, "learning_rate": 3.478e-05, "loss": 0.0661, "step": 1740 }, { "grad_norm": 0.78840571641922, "learning_rate": 3.498e-05, "loss": 0.0655, "step": 1750 }, { "grad_norm": 1.1352391242980957, "learning_rate": 3.518e-05, "loss": 0.0603, "step": 1760 }, { "grad_norm": 0.9543855786323547, "learning_rate": 3.5380000000000003e-05, "loss": 0.0771, "step": 1770 }, { "grad_norm": 1.0365525484085083, "learning_rate": 3.558e-05, "loss": 0.0829, "step": 1780 }, { "grad_norm": 0.875767171382904, "learning_rate": 3.578e-05, "loss": 0.0651, "step": 1790 }, { "grad_norm": 0.8544085621833801, "learning_rate": 3.5980000000000004e-05, "loss": 0.0681, "step": 1800 }, { "grad_norm": 1.5333058834075928, "learning_rate": 3.618e-05, "loss": 0.0851, "step": 1810 }, { "grad_norm": 0.8085379004478455, "learning_rate": 3.638e-05, "loss": 0.0694, "step": 1820 }, { "grad_norm": 0.7519668340682983, "learning_rate": 3.6580000000000006e-05, "loss": 0.0623, "step": 1830 }, { "grad_norm": 0.8113094568252563, "learning_rate": 3.6780000000000004e-05, "loss": 0.0543, "step": 1840 }, { "grad_norm": 1.1800061464309692, "learning_rate": 3.698e-05, "loss": 0.066, "step": 1850 }, { "grad_norm": 0.966036856174469, "learning_rate": 3.7180000000000007e-05, "loss": 0.0615, "step": 1860 }, { "grad_norm": 1.1421693563461304, "learning_rate": 3.7380000000000005e-05, "loss": 0.0668, "step": 1870 }, { "grad_norm": 1.1653079986572266, "learning_rate": 3.758e-05, "loss": 0.0573, "step": 1880 }, { "grad_norm": 0.9123221635818481, "learning_rate": 3.778000000000001e-05, "loss": 0.0549, "step": 1890 }, { "grad_norm": 1.2325799465179443, "learning_rate": 3.7980000000000006e-05, "loss": 0.0689, "step": 1900 }, { "grad_norm": 0.7221812009811401, "learning_rate": 3.818e-05, "loss": 0.062, "step": 1910 }, { "grad_norm": 1.0094162225723267, "learning_rate": 3.838e-05, "loss": 0.0667, "step": 1920 }, { "grad_norm": 0.8793899416923523, "learning_rate": 3.858e-05, "loss": 0.0662, "step": 1930 }, { "grad_norm": 0.8294569849967957, "learning_rate": 3.878e-05, "loss": 0.0708, "step": 1940 }, { "grad_norm": 1.1158952713012695, "learning_rate": 3.898e-05, "loss": 0.0664, "step": 1950 }, { "grad_norm": 0.6136499643325806, "learning_rate": 3.918e-05, "loss": 0.0596, "step": 1960 }, { "grad_norm": 1.1517845392227173, "learning_rate": 3.938e-05, "loss": 0.0673, "step": 1970 }, { "grad_norm": 0.8856388330459595, "learning_rate": 3.958e-05, "loss": 0.0645, "step": 1980 }, { "grad_norm": 0.8491419553756714, "learning_rate": 3.978e-05, "loss": 0.0698, "step": 1990 }, { "grad_norm": 0.5980357527732849, "learning_rate": 3.998e-05, "loss": 0.0569, "step": 2000 }, { "grad_norm": 0.7713785171508789, "learning_rate": 4.018e-05, "loss": 0.0589, "step": 2010 }, { "grad_norm": 0.695334255695343, "learning_rate": 4.038e-05, "loss": 0.059, "step": 2020 }, { "grad_norm": 0.7896490097045898, "learning_rate": 4.058e-05, "loss": 0.0511, "step": 2030 }, { "grad_norm": 0.7355788946151733, "learning_rate": 4.078e-05, "loss": 0.057, "step": 2040 }, { "grad_norm": 1.4091358184814453, "learning_rate": 4.0980000000000004e-05, "loss": 0.0712, "step": 2050 }, { "grad_norm": 0.7155201435089111, "learning_rate": 4.118e-05, "loss": 0.0596, "step": 2060 }, { "grad_norm": 1.0377463102340698, "learning_rate": 4.138e-05, "loss": 0.0622, "step": 2070 }, { "grad_norm": 1.4332327842712402, "learning_rate": 4.1580000000000005e-05, "loss": 0.0664, "step": 2080 }, { "grad_norm": 0.9463021159172058, "learning_rate": 4.178e-05, "loss": 0.0753, "step": 2090 }, { "grad_norm": 0.7765832543373108, "learning_rate": 4.198e-05, "loss": 0.0574, "step": 2100 }, { "grad_norm": 0.9698790907859802, "learning_rate": 4.2180000000000006e-05, "loss": 0.0572, "step": 2110 }, { "grad_norm": 0.8725900650024414, "learning_rate": 4.2380000000000004e-05, "loss": 0.0588, "step": 2120 }, { "grad_norm": 0.8711625933647156, "learning_rate": 4.258e-05, "loss": 0.0638, "step": 2130 }, { "grad_norm": 0.9733594655990601, "learning_rate": 4.278e-05, "loss": 0.0609, "step": 2140 }, { "grad_norm": 0.9663091897964478, "learning_rate": 4.2980000000000005e-05, "loss": 0.0652, "step": 2150 }, { "grad_norm": 0.6750958561897278, "learning_rate": 4.318e-05, "loss": 0.0481, "step": 2160 }, { "grad_norm": 0.761877179145813, "learning_rate": 4.338e-05, "loss": 0.0537, "step": 2170 }, { "grad_norm": 1.039548397064209, "learning_rate": 4.3580000000000006e-05, "loss": 0.0559, "step": 2180 }, { "grad_norm": 0.8283993005752563, "learning_rate": 4.3780000000000004e-05, "loss": 0.0601, "step": 2190 }, { "grad_norm": 0.6448161005973816, "learning_rate": 4.398e-05, "loss": 0.0495, "step": 2200 }, { "grad_norm": 0.814069926738739, "learning_rate": 4.418000000000001e-05, "loss": 0.0485, "step": 2210 }, { "grad_norm": 0.8077285289764404, "learning_rate": 4.438e-05, "loss": 0.0504, "step": 2220 }, { "grad_norm": 0.5997656583786011, "learning_rate": 4.458e-05, "loss": 0.0489, "step": 2230 }, { "grad_norm": 1.2048861980438232, "learning_rate": 4.478e-05, "loss": 0.0572, "step": 2240 }, { "grad_norm": 0.7595190405845642, "learning_rate": 4.498e-05, "loss": 0.0594, "step": 2250 }, { "grad_norm": 0.9868074655532837, "learning_rate": 4.518e-05, "loss": 0.072, "step": 2260 }, { "grad_norm": 0.6785060167312622, "learning_rate": 4.538e-05, "loss": 0.0634, "step": 2270 }, { "grad_norm": 0.6237939596176147, "learning_rate": 4.558e-05, "loss": 0.06, "step": 2280 }, { "grad_norm": 0.6781964302062988, "learning_rate": 4.578e-05, "loss": 0.0478, "step": 2290 }, { "grad_norm": 0.7752214670181274, "learning_rate": 4.5980000000000004e-05, "loss": 0.0636, "step": 2300 }, { "grad_norm": 0.8410532474517822, "learning_rate": 4.618e-05, "loss": 0.064, "step": 2310 }, { "grad_norm": 0.5645577311515808, "learning_rate": 4.638e-05, "loss": 0.0639, "step": 2320 }, { "grad_norm": 0.8270856738090515, "learning_rate": 4.6580000000000005e-05, "loss": 0.0588, "step": 2330 }, { "grad_norm": 1.0635420083999634, "learning_rate": 4.678e-05, "loss": 0.0578, "step": 2340 }, { "grad_norm": 0.8979659676551819, "learning_rate": 4.698e-05, "loss": 0.0571, "step": 2350 }, { "grad_norm": 1.0830320119857788, "learning_rate": 4.718e-05, "loss": 0.0618, "step": 2360 }, { "grad_norm": 0.7388594746589661, "learning_rate": 4.7380000000000004e-05, "loss": 0.0592, "step": 2370 }, { "grad_norm": 0.8936744332313538, "learning_rate": 4.758e-05, "loss": 0.0679, "step": 2380 }, { "grad_norm": 0.7902635931968689, "learning_rate": 4.778e-05, "loss": 0.0541, "step": 2390 }, { "grad_norm": 0.640809178352356, "learning_rate": 4.7980000000000005e-05, "loss": 0.0717, "step": 2400 }, { "grad_norm": 0.5267622470855713, "learning_rate": 4.818e-05, "loss": 0.0557, "step": 2410 }, { "grad_norm": 0.7142435908317566, "learning_rate": 4.838e-05, "loss": 0.0543, "step": 2420 }, { "grad_norm": 1.001394510269165, "learning_rate": 4.8580000000000006e-05, "loss": 0.0628, "step": 2430 }, { "grad_norm": 0.5647479295730591, "learning_rate": 4.8780000000000004e-05, "loss": 0.0543, "step": 2440 }, { "grad_norm": 0.9400005340576172, "learning_rate": 4.898e-05, "loss": 0.0626, "step": 2450 }, { "grad_norm": 1.2764812707901, "learning_rate": 4.918000000000001e-05, "loss": 0.0632, "step": 2460 }, { "grad_norm": 0.6312351822853088, "learning_rate": 4.9380000000000005e-05, "loss": 0.0577, "step": 2470 }, { "grad_norm": 0.5873410701751709, "learning_rate": 4.958e-05, "loss": 0.0576, "step": 2480 }, { "grad_norm": 0.6290830969810486, "learning_rate": 4.978e-05, "loss": 0.0469, "step": 2490 }, { "grad_norm": 0.9395654797554016, "learning_rate": 4.9980000000000006e-05, "loss": 0.0573, "step": 2500 }, { "grad_norm": 0.6957690715789795, "learning_rate": 5.0180000000000004e-05, "loss": 0.0573, "step": 2510 }, { "grad_norm": 0.7465547323226929, "learning_rate": 5.038e-05, "loss": 0.0487, "step": 2520 }, { "grad_norm": 1.3726977109909058, "learning_rate": 5.058000000000001e-05, "loss": 0.0522, "step": 2530 }, { "grad_norm": 0.6185299754142761, "learning_rate": 5.0780000000000005e-05, "loss": 0.0554, "step": 2540 }, { "grad_norm": 0.6783818006515503, "learning_rate": 5.098e-05, "loss": 0.0577, "step": 2550 }, { "grad_norm": 1.1770944595336914, "learning_rate": 5.118000000000001e-05, "loss": 0.0513, "step": 2560 }, { "grad_norm": 0.6455990672111511, "learning_rate": 5.1380000000000006e-05, "loss": 0.0619, "step": 2570 }, { "grad_norm": 0.7791551947593689, "learning_rate": 5.1580000000000004e-05, "loss": 0.0592, "step": 2580 }, { "grad_norm": 0.778497576713562, "learning_rate": 5.178000000000001e-05, "loss": 0.0546, "step": 2590 }, { "grad_norm": 0.575142502784729, "learning_rate": 5.198000000000001e-05, "loss": 0.0527, "step": 2600 }, { "grad_norm": 0.9282897114753723, "learning_rate": 5.2180000000000005e-05, "loss": 0.0475, "step": 2610 }, { "grad_norm": 0.6134388446807861, "learning_rate": 5.238000000000001e-05, "loss": 0.0441, "step": 2620 }, { "grad_norm": 1.0023173093795776, "learning_rate": 5.258000000000001e-05, "loss": 0.0579, "step": 2630 }, { "grad_norm": 0.3991743326187134, "learning_rate": 5.2780000000000006e-05, "loss": 0.0536, "step": 2640 }, { "grad_norm": 1.1841704845428467, "learning_rate": 5.2980000000000004e-05, "loss": 0.0591, "step": 2650 }, { "grad_norm": 0.6086379885673523, "learning_rate": 5.318000000000001e-05, "loss": 0.057, "step": 2660 }, { "grad_norm": 0.5952802300453186, "learning_rate": 5.338000000000001e-05, "loss": 0.0495, "step": 2670 }, { "grad_norm": 0.8265329599380493, "learning_rate": 5.3580000000000005e-05, "loss": 0.0564, "step": 2680 }, { "grad_norm": 0.6930387616157532, "learning_rate": 5.378e-05, "loss": 0.0716, "step": 2690 }, { "grad_norm": 0.8350011706352234, "learning_rate": 5.3979999999999995e-05, "loss": 0.0615, "step": 2700 }, { "grad_norm": 0.738007128238678, "learning_rate": 5.418e-05, "loss": 0.0641, "step": 2710 }, { "grad_norm": 0.5261081457138062, "learning_rate": 5.438e-05, "loss": 0.0482, "step": 2720 }, { "grad_norm": 1.7368208169937134, "learning_rate": 5.4579999999999996e-05, "loss": 0.0598, "step": 2730 }, { "grad_norm": 0.5668184161186218, "learning_rate": 5.478e-05, "loss": 0.0704, "step": 2740 }, { "grad_norm": 1.340855598449707, "learning_rate": 5.498e-05, "loss": 0.0612, "step": 2750 }, { "grad_norm": 0.7049098014831543, "learning_rate": 5.518e-05, "loss": 0.0678, "step": 2760 }, { "grad_norm": 0.7261455655097961, "learning_rate": 5.538e-05, "loss": 0.0517, "step": 2770 }, { "grad_norm": 0.5463258028030396, "learning_rate": 5.558e-05, "loss": 0.0562, "step": 2780 }, { "grad_norm": 0.5847055315971375, "learning_rate": 5.578e-05, "loss": 0.0613, "step": 2790 }, { "grad_norm": 0.6239318251609802, "learning_rate": 5.5979999999999996e-05, "loss": 0.0517, "step": 2800 }, { "grad_norm": 0.7713963389396667, "learning_rate": 5.618e-05, "loss": 0.0593, "step": 2810 }, { "grad_norm": 0.5577525496482849, "learning_rate": 5.638e-05, "loss": 0.0481, "step": 2820 }, { "grad_norm": 0.9081506729125977, "learning_rate": 5.658e-05, "loss": 0.0678, "step": 2830 }, { "grad_norm": 0.5861201286315918, "learning_rate": 5.678e-05, "loss": 0.0585, "step": 2840 }, { "grad_norm": 0.5515230298042297, "learning_rate": 5.698e-05, "loss": 0.0453, "step": 2850 }, { "grad_norm": 0.8923326134681702, "learning_rate": 5.718e-05, "loss": 0.0585, "step": 2860 }, { "grad_norm": 0.5944814682006836, "learning_rate": 5.738e-05, "loss": 0.0556, "step": 2870 }, { "grad_norm": 0.46186134219169617, "learning_rate": 5.758e-05, "loss": 0.0477, "step": 2880 }, { "grad_norm": 0.4289417862892151, "learning_rate": 5.778e-05, "loss": 0.0423, "step": 2890 }, { "grad_norm": 0.7590463757514954, "learning_rate": 5.7980000000000004e-05, "loss": 0.053, "step": 2900 }, { "grad_norm": 0.9708814024925232, "learning_rate": 5.818e-05, "loss": 0.0532, "step": 2910 }, { "grad_norm": 0.6277782917022705, "learning_rate": 5.838e-05, "loss": 0.057, "step": 2920 }, { "grad_norm": 0.9921399354934692, "learning_rate": 5.858e-05, "loss": 0.0586, "step": 2930 }, { "grad_norm": 0.6499211192131042, "learning_rate": 5.878e-05, "loss": 0.0607, "step": 2940 }, { "grad_norm": 0.7880867123603821, "learning_rate": 5.898e-05, "loss": 0.058, "step": 2950 }, { "grad_norm": 0.6023193597793579, "learning_rate": 5.918e-05, "loss": 0.061, "step": 2960 }, { "grad_norm": 0.9018617868423462, "learning_rate": 5.9380000000000004e-05, "loss": 0.0526, "step": 2970 }, { "grad_norm": 0.6333033442497253, "learning_rate": 5.958e-05, "loss": 0.0522, "step": 2980 }, { "grad_norm": 0.6302904486656189, "learning_rate": 5.978e-05, "loss": 0.0496, "step": 2990 }, { "grad_norm": 0.5083188414573669, "learning_rate": 5.9980000000000005e-05, "loss": 0.0517, "step": 3000 }, { "grad_norm": 1.062338948249817, "learning_rate": 6.018e-05, "loss": 0.0587, "step": 3010 }, { "grad_norm": 0.6466048955917358, "learning_rate": 6.038e-05, "loss": 0.057, "step": 3020 }, { "grad_norm": 0.6986466646194458, "learning_rate": 6.0580000000000006e-05, "loss": 0.0594, "step": 3030 }, { "grad_norm": 0.47053128480911255, "learning_rate": 6.0780000000000004e-05, "loss": 0.0482, "step": 3040 }, { "grad_norm": 0.5492612719535828, "learning_rate": 6.098e-05, "loss": 0.0461, "step": 3050 }, { "grad_norm": 0.4800388216972351, "learning_rate": 6.118000000000001e-05, "loss": 0.0453, "step": 3060 }, { "grad_norm": 0.7649409174919128, "learning_rate": 6.138e-05, "loss": 0.0589, "step": 3070 }, { "grad_norm": 0.5095891952514648, "learning_rate": 6.158e-05, "loss": 0.0595, "step": 3080 }, { "grad_norm": 0.4373922646045685, "learning_rate": 6.178000000000001e-05, "loss": 0.0507, "step": 3090 }, { "grad_norm": 0.580132007598877, "learning_rate": 6.198e-05, "loss": 0.0515, "step": 3100 }, { "grad_norm": 0.4929387867450714, "learning_rate": 6.218e-05, "loss": 0.052, "step": 3110 }, { "grad_norm": 0.43398740887641907, "learning_rate": 6.238000000000001e-05, "loss": 0.0378, "step": 3120 }, { "grad_norm": 0.8539814949035645, "learning_rate": 6.258e-05, "loss": 0.0531, "step": 3130 }, { "grad_norm": 0.45738476514816284, "learning_rate": 6.278e-05, "loss": 0.0436, "step": 3140 }, { "grad_norm": 0.6220096349716187, "learning_rate": 6.298000000000001e-05, "loss": 0.0467, "step": 3150 }, { "grad_norm": 0.8653008341789246, "learning_rate": 6.318e-05, "loss": 0.0463, "step": 3160 }, { "grad_norm": 0.5008379220962524, "learning_rate": 6.338e-05, "loss": 0.0542, "step": 3170 }, { "grad_norm": 0.4816921651363373, "learning_rate": 6.358000000000001e-05, "loss": 0.0559, "step": 3180 }, { "grad_norm": 0.4234481453895569, "learning_rate": 6.378e-05, "loss": 0.0549, "step": 3190 }, { "grad_norm": 0.6644657254219055, "learning_rate": 6.398000000000001e-05, "loss": 0.0551, "step": 3200 }, { "grad_norm": 0.4503837823867798, "learning_rate": 6.418000000000001e-05, "loss": 0.0435, "step": 3210 }, { "grad_norm": 0.6880264282226562, "learning_rate": 6.438e-05, "loss": 0.058, "step": 3220 }, { "grad_norm": 0.5368906855583191, "learning_rate": 6.458000000000001e-05, "loss": 0.052, "step": 3230 }, { "grad_norm": 0.5340503454208374, "learning_rate": 6.478000000000001e-05, "loss": 0.0448, "step": 3240 }, { "grad_norm": 0.6583150625228882, "learning_rate": 6.498e-05, "loss": 0.0562, "step": 3250 }, { "grad_norm": 0.5658986568450928, "learning_rate": 6.518000000000001e-05, "loss": 0.0512, "step": 3260 }, { "grad_norm": 0.46565285325050354, "learning_rate": 6.538000000000001e-05, "loss": 0.0467, "step": 3270 }, { "grad_norm": 0.9076792597770691, "learning_rate": 6.558e-05, "loss": 0.0521, "step": 3280 }, { "grad_norm": 0.49506181478500366, "learning_rate": 6.578000000000001e-05, "loss": 0.0546, "step": 3290 }, { "grad_norm": 0.41324666142463684, "learning_rate": 6.598e-05, "loss": 0.0475, "step": 3300 }, { "grad_norm": 0.4538547396659851, "learning_rate": 6.618e-05, "loss": 0.0437, "step": 3310 }, { "grad_norm": 0.7337180376052856, "learning_rate": 6.638e-05, "loss": 0.0552, "step": 3320 }, { "grad_norm": 0.48878562450408936, "learning_rate": 6.658e-05, "loss": 0.0514, "step": 3330 }, { "grad_norm": 0.6667009592056274, "learning_rate": 6.678e-05, "loss": 0.0528, "step": 3340 }, { "grad_norm": 0.7415311336517334, "learning_rate": 6.698e-05, "loss": 0.0501, "step": 3350 }, { "grad_norm": 0.517810046672821, "learning_rate": 6.718e-05, "loss": 0.051, "step": 3360 }, { "grad_norm": 0.5024856328964233, "learning_rate": 6.738e-05, "loss": 0.0465, "step": 3370 }, { "grad_norm": 0.4108419120311737, "learning_rate": 6.758e-05, "loss": 0.0399, "step": 3380 }, { "grad_norm": 0.7057079076766968, "learning_rate": 6.778e-05, "loss": 0.0475, "step": 3390 }, { "grad_norm": 0.5512890219688416, "learning_rate": 6.798e-05, "loss": 0.0476, "step": 3400 }, { "grad_norm": 0.6559773087501526, "learning_rate": 6.818e-05, "loss": 0.0467, "step": 3410 }, { "grad_norm": 1.3571206331253052, "learning_rate": 6.838e-05, "loss": 0.0466, "step": 3420 }, { "grad_norm": 0.5611778497695923, "learning_rate": 6.858e-05, "loss": 0.0551, "step": 3430 }, { "grad_norm": 1.1759673357009888, "learning_rate": 6.878e-05, "loss": 0.0542, "step": 3440 }, { "grad_norm": 0.6388060450553894, "learning_rate": 6.898e-05, "loss": 0.0462, "step": 3450 }, { "grad_norm": 0.5403591990470886, "learning_rate": 6.918e-05, "loss": 0.0452, "step": 3460 }, { "grad_norm": 0.48924902081489563, "learning_rate": 6.938e-05, "loss": 0.05, "step": 3470 }, { "grad_norm": 0.6280256509780884, "learning_rate": 6.958e-05, "loss": 0.0566, "step": 3480 }, { "grad_norm": 0.6630896925926208, "learning_rate": 6.978e-05, "loss": 0.0463, "step": 3490 }, { "grad_norm": 0.3568757176399231, "learning_rate": 6.998e-05, "loss": 0.0493, "step": 3500 }, { "grad_norm": 0.5662403702735901, "learning_rate": 7.018e-05, "loss": 0.0496, "step": 3510 }, { "grad_norm": 0.4095475971698761, "learning_rate": 7.038e-05, "loss": 0.0512, "step": 3520 }, { "grad_norm": 0.7884876132011414, "learning_rate": 7.058e-05, "loss": 0.0476, "step": 3530 }, { "grad_norm": 0.3781804144382477, "learning_rate": 7.078e-05, "loss": 0.051, "step": 3540 }, { "grad_norm": 0.3647266626358032, "learning_rate": 7.098e-05, "loss": 0.0465, "step": 3550 }, { "grad_norm": 0.4461723566055298, "learning_rate": 7.118e-05, "loss": 0.0429, "step": 3560 }, { "grad_norm": 0.49902278184890747, "learning_rate": 7.138e-05, "loss": 0.0438, "step": 3570 }, { "grad_norm": 0.48085904121398926, "learning_rate": 7.158e-05, "loss": 0.052, "step": 3580 }, { "grad_norm": 0.5880372524261475, "learning_rate": 7.178000000000001e-05, "loss": 0.0422, "step": 3590 }, { "grad_norm": 0.45162951946258545, "learning_rate": 7.198e-05, "loss": 0.0508, "step": 3600 }, { "grad_norm": 0.43089383840560913, "learning_rate": 7.218e-05, "loss": 0.0449, "step": 3610 }, { "grad_norm": 0.4600507318973541, "learning_rate": 7.238000000000001e-05, "loss": 0.0399, "step": 3620 }, { "grad_norm": 0.5375986099243164, "learning_rate": 7.258e-05, "loss": 0.0433, "step": 3630 }, { "grad_norm": 0.46010565757751465, "learning_rate": 7.278e-05, "loss": 0.0486, "step": 3640 }, { "grad_norm": 0.4487146735191345, "learning_rate": 7.298000000000001e-05, "loss": 0.0401, "step": 3650 }, { "grad_norm": 0.36852604150772095, "learning_rate": 7.318e-05, "loss": 0.039, "step": 3660 }, { "grad_norm": 0.8180766105651855, "learning_rate": 7.338e-05, "loss": 0.0395, "step": 3670 }, { "grad_norm": 0.4331512153148651, "learning_rate": 7.358000000000001e-05, "loss": 0.0452, "step": 3680 }, { "grad_norm": 0.6997743248939514, "learning_rate": 7.378e-05, "loss": 0.0586, "step": 3690 }, { "grad_norm": 0.5136622786521912, "learning_rate": 7.398e-05, "loss": 0.044, "step": 3700 }, { "grad_norm": 0.43291357159614563, "learning_rate": 7.418000000000001e-05, "loss": 0.0422, "step": 3710 }, { "grad_norm": 0.29788899421691895, "learning_rate": 7.438e-05, "loss": 0.0402, "step": 3720 }, { "grad_norm": 0.35856306552886963, "learning_rate": 7.458000000000001e-05, "loss": 0.0417, "step": 3730 }, { "grad_norm": 0.526261031627655, "learning_rate": 7.478e-05, "loss": 0.0474, "step": 3740 }, { "grad_norm": 0.30645951628685, "learning_rate": 7.498e-05, "loss": 0.0415, "step": 3750 }, { "grad_norm": 0.3947923481464386, "learning_rate": 7.518000000000001e-05, "loss": 0.0401, "step": 3760 }, { "grad_norm": 0.6234109997749329, "learning_rate": 7.538e-05, "loss": 0.0523, "step": 3770 }, { "grad_norm": 0.4438498914241791, "learning_rate": 7.558e-05, "loss": 0.0499, "step": 3780 }, { "grad_norm": 0.5023193955421448, "learning_rate": 7.578000000000001e-05, "loss": 0.0485, "step": 3790 }, { "grad_norm": 0.45712313055992126, "learning_rate": 7.598e-05, "loss": 0.0478, "step": 3800 }, { "grad_norm": 0.4903082251548767, "learning_rate": 7.618e-05, "loss": 0.0487, "step": 3810 }, { "grad_norm": 0.6939442753791809, "learning_rate": 7.638000000000001e-05, "loss": 0.0656, "step": 3820 }, { "grad_norm": 0.37813666462898254, "learning_rate": 7.658e-05, "loss": 0.0486, "step": 3830 }, { "grad_norm": 0.662528395652771, "learning_rate": 7.678000000000001e-05, "loss": 0.0517, "step": 3840 }, { "grad_norm": 0.39765241742134094, "learning_rate": 7.698000000000001e-05, "loss": 0.05, "step": 3850 }, { "grad_norm": 0.49288663268089294, "learning_rate": 7.718e-05, "loss": 0.0512, "step": 3860 }, { "grad_norm": 0.5524985790252686, "learning_rate": 7.738000000000001e-05, "loss": 0.0417, "step": 3870 }, { "grad_norm": 0.4438939690589905, "learning_rate": 7.758000000000001e-05, "loss": 0.0379, "step": 3880 }, { "grad_norm": 0.6975436210632324, "learning_rate": 7.778e-05, "loss": 0.0636, "step": 3890 }, { "grad_norm": 1.249714732170105, "learning_rate": 7.798000000000001e-05, "loss": 0.0604, "step": 3900 }, { "grad_norm": 0.33487603068351746, "learning_rate": 7.818000000000001e-05, "loss": 0.0609, "step": 3910 }, { "grad_norm": 0.54608154296875, "learning_rate": 7.838e-05, "loss": 0.0597, "step": 3920 }, { "grad_norm": 0.4703153073787689, "learning_rate": 7.858000000000001e-05, "loss": 0.0425, "step": 3930 }, { "grad_norm": 0.3794666826725006, "learning_rate": 7.878e-05, "loss": 0.0495, "step": 3940 }, { "grad_norm": 0.34775128960609436, "learning_rate": 7.897999999999999e-05, "loss": 0.0442, "step": 3950 }, { "grad_norm": 0.38238832354545593, "learning_rate": 7.918e-05, "loss": 0.0413, "step": 3960 }, { "grad_norm": 0.530239462852478, "learning_rate": 7.938e-05, "loss": 0.0399, "step": 3970 }, { "grad_norm": 0.5896754264831543, "learning_rate": 7.958e-05, "loss": 0.0426, "step": 3980 }, { "grad_norm": 0.32772085070610046, "learning_rate": 7.978e-05, "loss": 0.0395, "step": 3990 }, { "grad_norm": 0.9565135836601257, "learning_rate": 7.998e-05, "loss": 0.0422, "step": 4000 }, { "grad_norm": 0.554099977016449, "learning_rate": 8.018e-05, "loss": 0.0442, "step": 4010 }, { "grad_norm": 0.6889604330062866, "learning_rate": 8.038e-05, "loss": 0.0462, "step": 4020 }, { "grad_norm": 0.47866198420524597, "learning_rate": 8.058e-05, "loss": 0.0541, "step": 4030 }, { "grad_norm": 0.9744945168495178, "learning_rate": 8.078e-05, "loss": 0.0474, "step": 4040 }, { "grad_norm": 0.5014967322349548, "learning_rate": 8.098e-05, "loss": 0.0514, "step": 4050 }, { "grad_norm": 0.6364223957061768, "learning_rate": 8.118e-05, "loss": 0.0501, "step": 4060 }, { "grad_norm": 0.33608072996139526, "learning_rate": 8.138e-05, "loss": 0.0416, "step": 4070 }, { "grad_norm": 0.5390483140945435, "learning_rate": 8.158e-05, "loss": 0.0455, "step": 4080 }, { "grad_norm": 1.035232424736023, "learning_rate": 8.178e-05, "loss": 0.0448, "step": 4090 }, { "grad_norm": 0.3061126470565796, "learning_rate": 8.198e-05, "loss": 0.0502, "step": 4100 }, { "grad_norm": 0.6197689175605774, "learning_rate": 8.218e-05, "loss": 0.0507, "step": 4110 }, { "grad_norm": 0.46309706568717957, "learning_rate": 8.238000000000001e-05, "loss": 0.0546, "step": 4120 }, { "grad_norm": 1.0109127759933472, "learning_rate": 8.258e-05, "loss": 0.0518, "step": 4130 }, { "grad_norm": 0.47116008400917053, "learning_rate": 8.278e-05, "loss": 0.048, "step": 4140 }, { "grad_norm": 0.5121886730194092, "learning_rate": 8.298000000000001e-05, "loss": 0.0424, "step": 4150 }, { "grad_norm": 1.1637609004974365, "learning_rate": 8.318e-05, "loss": 0.0425, "step": 4160 }, { "grad_norm": 0.4351927638053894, "learning_rate": 8.338e-05, "loss": 0.0656, "step": 4170 }, { "grad_norm": 0.3786669969558716, "learning_rate": 8.358e-05, "loss": 0.0461, "step": 4180 }, { "grad_norm": 0.35368284583091736, "learning_rate": 8.378e-05, "loss": 0.0525, "step": 4190 }, { "grad_norm": 0.742223858833313, "learning_rate": 8.398e-05, "loss": 0.042, "step": 4200 }, { "grad_norm": 0.347889244556427, "learning_rate": 8.418e-05, "loss": 0.0464, "step": 4210 }, { "grad_norm": 0.5017602443695068, "learning_rate": 8.438e-05, "loss": 0.0422, "step": 4220 }, { "grad_norm": 0.39328524470329285, "learning_rate": 8.458e-05, "loss": 0.0471, "step": 4230 }, { "grad_norm": 0.5187689065933228, "learning_rate": 8.478e-05, "loss": 0.0501, "step": 4240 }, { "grad_norm": 0.3901059329509735, "learning_rate": 8.498e-05, "loss": 0.0506, "step": 4250 }, { "grad_norm": 0.6230197548866272, "learning_rate": 8.518000000000001e-05, "loss": 0.0503, "step": 4260 }, { "grad_norm": 0.5461595058441162, "learning_rate": 8.538e-05, "loss": 0.0495, "step": 4270 }, { "grad_norm": 0.3772653043270111, "learning_rate": 8.558e-05, "loss": 0.0392, "step": 4280 }, { "grad_norm": 0.41432902216911316, "learning_rate": 8.578000000000001e-05, "loss": 0.041, "step": 4290 }, { "grad_norm": 0.36690279841423035, "learning_rate": 8.598e-05, "loss": 0.0414, "step": 4300 }, { "grad_norm": 0.3271716833114624, "learning_rate": 8.618e-05, "loss": 0.0384, "step": 4310 }, { "grad_norm": 0.34115511178970337, "learning_rate": 8.638000000000001e-05, "loss": 0.0413, "step": 4320 }, { "grad_norm": 0.4081493616104126, "learning_rate": 8.658e-05, "loss": 0.0358, "step": 4330 }, { "grad_norm": 0.4732321500778198, "learning_rate": 8.678e-05, "loss": 0.0403, "step": 4340 }, { "grad_norm": 0.5213508605957031, "learning_rate": 8.698000000000001e-05, "loss": 0.0493, "step": 4350 }, { "grad_norm": 0.2872004508972168, "learning_rate": 8.718e-05, "loss": 0.0409, "step": 4360 }, { "grad_norm": 0.5863088965415955, "learning_rate": 8.738000000000001e-05, "loss": 0.0442, "step": 4370 }, { "grad_norm": 0.2722342610359192, "learning_rate": 8.758000000000001e-05, "loss": 0.0459, "step": 4380 }, { "grad_norm": 0.4581398069858551, "learning_rate": 8.778e-05, "loss": 0.0582, "step": 4390 }, { "grad_norm": 0.3983118236064911, "learning_rate": 8.798000000000001e-05, "loss": 0.0458, "step": 4400 }, { "grad_norm": 0.35771721601486206, "learning_rate": 8.818000000000001e-05, "loss": 0.0433, "step": 4410 }, { "grad_norm": 0.6386746168136597, "learning_rate": 8.838e-05, "loss": 0.05, "step": 4420 }, { "grad_norm": 0.34959691762924194, "learning_rate": 8.858000000000001e-05, "loss": 0.0431, "step": 4430 }, { "grad_norm": 0.7233688831329346, "learning_rate": 8.878000000000001e-05, "loss": 0.0502, "step": 4440 }, { "grad_norm": 0.48628634214401245, "learning_rate": 8.898e-05, "loss": 0.0508, "step": 4450 }, { "grad_norm": 0.3960534334182739, "learning_rate": 8.918000000000001e-05, "loss": 0.0473, "step": 4460 }, { "grad_norm": 0.29604828357696533, "learning_rate": 8.938e-05, "loss": 0.0394, "step": 4470 }, { "grad_norm": 0.41287195682525635, "learning_rate": 8.958e-05, "loss": 0.0485, "step": 4480 }, { "grad_norm": 0.4024379253387451, "learning_rate": 8.978000000000001e-05, "loss": 0.0437, "step": 4490 }, { "grad_norm": 0.35168537497520447, "learning_rate": 8.998e-05, "loss": 0.0346, "step": 4500 }, { "grad_norm": 0.3917841613292694, "learning_rate": 9.018000000000001e-05, "loss": 0.0415, "step": 4510 }, { "grad_norm": 0.29779618978500366, "learning_rate": 9.038000000000001e-05, "loss": 0.0379, "step": 4520 }, { "grad_norm": 0.38726162910461426, "learning_rate": 9.058e-05, "loss": 0.0415, "step": 4530 }, { "grad_norm": 0.39981698989868164, "learning_rate": 9.078000000000001e-05, "loss": 0.0406, "step": 4540 }, { "grad_norm": 0.3438284397125244, "learning_rate": 9.098000000000001e-05, "loss": 0.0359, "step": 4550 }, { "grad_norm": 0.32664167881011963, "learning_rate": 9.118e-05, "loss": 0.042, "step": 4560 }, { "grad_norm": 0.4068146049976349, "learning_rate": 9.138e-05, "loss": 0.0427, "step": 4570 }, { "grad_norm": 0.3811015486717224, "learning_rate": 9.158e-05, "loss": 0.0534, "step": 4580 }, { "grad_norm": 0.49980878829956055, "learning_rate": 9.178e-05, "loss": 0.0516, "step": 4590 }, { "grad_norm": 0.438308984041214, "learning_rate": 9.198e-05, "loss": 0.0479, "step": 4600 }, { "grad_norm": 0.5040497183799744, "learning_rate": 9.218e-05, "loss": 0.0564, "step": 4610 }, { "grad_norm": 0.37374773621559143, "learning_rate": 9.238e-05, "loss": 0.0452, "step": 4620 }, { "grad_norm": 0.5445471405982971, "learning_rate": 9.258e-05, "loss": 0.0548, "step": 4630 }, { "grad_norm": 0.42205530405044556, "learning_rate": 9.278e-05, "loss": 0.0484, "step": 4640 }, { "grad_norm": 0.30648472905158997, "learning_rate": 9.298e-05, "loss": 0.0367, "step": 4650 }, { "grad_norm": 0.338426411151886, "learning_rate": 9.318e-05, "loss": 0.0358, "step": 4660 }, { "grad_norm": 0.4062999486923218, "learning_rate": 9.338e-05, "loss": 0.0323, "step": 4670 }, { "grad_norm": 0.2818452715873718, "learning_rate": 9.358e-05, "loss": 0.0291, "step": 4680 }, { "grad_norm": 0.6983944773674011, "learning_rate": 9.378e-05, "loss": 0.0514, "step": 4690 }, { "grad_norm": 0.4307539165019989, "learning_rate": 9.398e-05, "loss": 0.0559, "step": 4700 }, { "grad_norm": 0.3602946698665619, "learning_rate": 9.418e-05, "loss": 0.0419, "step": 4710 }, { "grad_norm": 0.5544717311859131, "learning_rate": 9.438e-05, "loss": 0.0434, "step": 4720 }, { "grad_norm": 0.37220796942710876, "learning_rate": 9.458e-05, "loss": 0.0443, "step": 4730 }, { "grad_norm": 0.2955740988254547, "learning_rate": 9.478e-05, "loss": 0.0386, "step": 4740 }, { "grad_norm": 0.3590521216392517, "learning_rate": 9.498e-05, "loss": 0.0358, "step": 4750 }, { "grad_norm": 0.4666734039783478, "learning_rate": 9.518000000000001e-05, "loss": 0.0442, "step": 4760 }, { "grad_norm": 0.36916157603263855, "learning_rate": 9.538e-05, "loss": 0.0498, "step": 4770 }, { "grad_norm": 0.389873206615448, "learning_rate": 9.558e-05, "loss": 0.0411, "step": 4780 }, { "grad_norm": 0.37865373492240906, "learning_rate": 9.578000000000001e-05, "loss": 0.04, "step": 4790 }, { "grad_norm": 0.49378713965415955, "learning_rate": 9.598e-05, "loss": 0.0437, "step": 4800 }, { "grad_norm": 0.3894253075122833, "learning_rate": 9.618e-05, "loss": 0.0363, "step": 4810 }, { "grad_norm": 0.43769240379333496, "learning_rate": 9.638000000000001e-05, "loss": 0.05, "step": 4820 }, { "grad_norm": 0.3492748439311981, "learning_rate": 9.658e-05, "loss": 0.0378, "step": 4830 }, { "grad_norm": 0.5206526517868042, "learning_rate": 9.678e-05, "loss": 0.0478, "step": 4840 }, { "grad_norm": 0.41593316197395325, "learning_rate": 9.698000000000001e-05, "loss": 0.0408, "step": 4850 }, { "grad_norm": 0.4961851239204407, "learning_rate": 9.718e-05, "loss": 0.0445, "step": 4860 }, { "grad_norm": 0.3802676498889923, "learning_rate": 9.738e-05, "loss": 0.0443, "step": 4870 }, { "grad_norm": 0.339838445186615, "learning_rate": 9.758000000000001e-05, "loss": 0.0379, "step": 4880 }, { "grad_norm": 0.4841865301132202, "learning_rate": 9.778e-05, "loss": 0.0516, "step": 4890 }, { "grad_norm": 0.29389485716819763, "learning_rate": 9.798000000000001e-05, "loss": 0.0415, "step": 4900 }, { "grad_norm": 0.4057120978832245, "learning_rate": 9.818000000000001e-05, "loss": 0.0469, "step": 4910 }, { "grad_norm": 0.30512773990631104, "learning_rate": 9.838e-05, "loss": 0.0433, "step": 4920 }, { "grad_norm": 0.3284855782985687, "learning_rate": 9.858000000000001e-05, "loss": 0.0404, "step": 4930 }, { "grad_norm": 0.35841089487075806, "learning_rate": 9.878e-05, "loss": 0.0345, "step": 4940 }, { "grad_norm": 0.38782528042793274, "learning_rate": 9.898e-05, "loss": 0.0461, "step": 4950 }, { "grad_norm": 0.29737842082977295, "learning_rate": 9.918000000000001e-05, "loss": 0.0403, "step": 4960 }, { "grad_norm": 0.28764522075653076, "learning_rate": 9.938e-05, "loss": 0.0309, "step": 4970 }, { "grad_norm": 0.42731574177742004, "learning_rate": 9.958e-05, "loss": 0.0402, "step": 4980 }, { "grad_norm": 0.3598746657371521, "learning_rate": 9.978000000000001e-05, "loss": 0.0449, "step": 4990 }, { "grad_norm": 0.26696375012397766, "learning_rate": 9.998e-05, "loss": 0.0379, "step": 5000 }, { "grad_norm": 0.44427648186683655, "learning_rate": 9.999999778549045e-05, "loss": 0.0492, "step": 5010 }, { "grad_norm": 0.28616395592689514, "learning_rate": 9.999999013039593e-05, "loss": 0.0426, "step": 5020 }, { "grad_norm": 0.263823002576828, "learning_rate": 9.999997700737766e-05, "loss": 0.033, "step": 5030 }, { "grad_norm": 0.5624625086784363, "learning_rate": 9.999995841643709e-05, "loss": 0.0498, "step": 5040 }, { "grad_norm": 0.28487688302993774, "learning_rate": 9.999993435757623e-05, "loss": 0.0424, "step": 5050 }, { "grad_norm": 0.2673753798007965, "learning_rate": 9.999990483079773e-05, "loss": 0.0434, "step": 5060 }, { "grad_norm": 0.2716826796531677, "learning_rate": 9.999986983610481e-05, "loss": 0.0364, "step": 5070 }, { "grad_norm": 0.38698992133140564, "learning_rate": 9.99998293735013e-05, "loss": 0.0421, "step": 5080 }, { "grad_norm": 0.326809287071228, "learning_rate": 9.999978344299161e-05, "loss": 0.0415, "step": 5090 }, { "grad_norm": 0.5153592228889465, "learning_rate": 9.99997320445808e-05, "loss": 0.0468, "step": 5100 }, { "grad_norm": 0.41509687900543213, "learning_rate": 9.999967517827444e-05, "loss": 0.0409, "step": 5110 }, { "grad_norm": 0.2947898209095001, "learning_rate": 9.999961284407879e-05, "loss": 0.0323, "step": 5120 }, { "grad_norm": 0.45014578104019165, "learning_rate": 9.999954504200067e-05, "loss": 0.0603, "step": 5130 }, { "grad_norm": 0.35461464524269104, "learning_rate": 9.999947177204744e-05, "loss": 0.0471, "step": 5140 }, { "grad_norm": 0.37814247608184814, "learning_rate": 9.999939303422718e-05, "loss": 0.0453, "step": 5150 }, { "grad_norm": 0.24806588888168335, "learning_rate": 9.999930882854847e-05, "loss": 0.0397, "step": 5160 }, { "grad_norm": 0.4387001097202301, "learning_rate": 9.999921915502051e-05, "loss": 0.0457, "step": 5170 }, { "grad_norm": 0.2755850553512573, "learning_rate": 9.99991240136531e-05, "loss": 0.0381, "step": 5180 }, { "grad_norm": 0.27244386076927185, "learning_rate": 9.999902340445668e-05, "loss": 0.0382, "step": 5190 }, { "grad_norm": 0.2566949129104614, "learning_rate": 9.999891732744224e-05, "loss": 0.0337, "step": 5200 }, { "grad_norm": 0.29657644033432007, "learning_rate": 9.999880578262135e-05, "loss": 0.0396, "step": 5210 }, { "grad_norm": 0.2968031167984009, "learning_rate": 9.999868877000624e-05, "loss": 0.0395, "step": 5220 }, { "grad_norm": 0.6268847584724426, "learning_rate": 9.99985662896097e-05, "loss": 0.0469, "step": 5230 }, { "grad_norm": 1.1648569107055664, "learning_rate": 9.999843834144513e-05, "loss": 0.0425, "step": 5240 }, { "grad_norm": 0.47891929745674133, "learning_rate": 9.99983049255265e-05, "loss": 0.0527, "step": 5250 }, { "grad_norm": 0.3365848958492279, "learning_rate": 9.999816604186843e-05, "loss": 0.0399, "step": 5260 }, { "grad_norm": 0.35889875888824463, "learning_rate": 9.999802169048609e-05, "loss": 0.0358, "step": 5270 }, { "grad_norm": 0.3156275749206543, "learning_rate": 9.999787187139527e-05, "loss": 0.0341, "step": 5280 }, { "grad_norm": 0.3288482427597046, "learning_rate": 9.999771658461234e-05, "loss": 0.0331, "step": 5290 }, { "grad_norm": 0.42018741369247437, "learning_rate": 9.999755583015431e-05, "loss": 0.0373, "step": 5300 }, { "grad_norm": 0.31995517015457153, "learning_rate": 9.999738960803874e-05, "loss": 0.0599, "step": 5310 }, { "grad_norm": 0.5587481260299683, "learning_rate": 9.99972179182838e-05, "loss": 0.052, "step": 5320 }, { "grad_norm": 0.32848992943763733, "learning_rate": 9.99970407609083e-05, "loss": 0.0459, "step": 5330 }, { "grad_norm": 0.39096805453300476, "learning_rate": 9.999685813593159e-05, "loss": 0.0436, "step": 5340 }, { "grad_norm": 0.39574888348579407, "learning_rate": 9.999667004337362e-05, "loss": 0.0611, "step": 5350 }, { "grad_norm": 0.3012562096118927, "learning_rate": 9.9996476483255e-05, "loss": 0.0413, "step": 5360 }, { "grad_norm": 0.29174286127090454, "learning_rate": 9.999627745559688e-05, "loss": 0.0349, "step": 5370 }, { "grad_norm": 0.27413925528526306, "learning_rate": 9.999607296042101e-05, "loss": 0.0362, "step": 5380 }, { "grad_norm": 0.36180591583251953, "learning_rate": 9.99958629977498e-05, "loss": 0.0366, "step": 5390 }, { "grad_norm": 0.22423352301120758, "learning_rate": 9.999564756760615e-05, "loss": 0.0334, "step": 5400 }, { "grad_norm": 0.3119376599788666, "learning_rate": 9.999542667001366e-05, "loss": 0.0341, "step": 5410 }, { "grad_norm": 0.3493826985359192, "learning_rate": 9.999520030499647e-05, "loss": 0.0388, "step": 5420 }, { "grad_norm": 0.38909682631492615, "learning_rate": 9.999496847257936e-05, "loss": 0.0471, "step": 5430 }, { "grad_norm": 0.4181475043296814, "learning_rate": 9.999473117278764e-05, "loss": 0.0425, "step": 5440 }, { "grad_norm": 0.377665251493454, "learning_rate": 9.999448840564731e-05, "loss": 0.0439, "step": 5450 }, { "grad_norm": 0.4236927032470703, "learning_rate": 9.999424017118488e-05, "loss": 0.0437, "step": 5460 }, { "grad_norm": 0.35043761134147644, "learning_rate": 9.999398646942751e-05, "loss": 0.0382, "step": 5470 }, { "grad_norm": 0.31696683168411255, "learning_rate": 9.999372730040296e-05, "loss": 0.0356, "step": 5480 }, { "grad_norm": 0.5127854347229004, "learning_rate": 9.999346266413953e-05, "loss": 0.0489, "step": 5490 }, { "grad_norm": 0.8107929229736328, "learning_rate": 9.99931925606662e-05, "loss": 0.0418, "step": 5500 }, { "grad_norm": 0.35530704259872437, "learning_rate": 9.99929169900125e-05, "loss": 0.049, "step": 5510 }, { "grad_norm": 0.26703861355781555, "learning_rate": 9.999263595220855e-05, "loss": 0.0427, "step": 5520 }, { "grad_norm": 0.2305978685617447, "learning_rate": 9.99923494472851e-05, "loss": 0.0354, "step": 5530 }, { "grad_norm": 0.27947357296943665, "learning_rate": 9.999205747527348e-05, "loss": 0.034, "step": 5540 }, { "grad_norm": 0.32300159335136414, "learning_rate": 9.999176003620561e-05, "loss": 0.0347, "step": 5550 }, { "grad_norm": 0.3752649426460266, "learning_rate": 9.999145713011405e-05, "loss": 0.0391, "step": 5560 }, { "grad_norm": 0.34709611535072327, "learning_rate": 9.999114875703186e-05, "loss": 0.0413, "step": 5570 }, { "grad_norm": 0.2571975886821747, "learning_rate": 9.999083491699281e-05, "loss": 0.0321, "step": 5580 }, { "grad_norm": 0.49590668082237244, "learning_rate": 9.999051561003123e-05, "loss": 0.0397, "step": 5590 }, { "grad_norm": 0.2843210995197296, "learning_rate": 9.999019083618202e-05, "loss": 0.041, "step": 5600 }, { "grad_norm": 0.35253021121025085, "learning_rate": 9.99898605954807e-05, "loss": 0.0404, "step": 5610 }, { "grad_norm": 0.3034692704677582, "learning_rate": 9.998952488796338e-05, "loss": 0.042, "step": 5620 }, { "grad_norm": 0.3068755865097046, "learning_rate": 9.998918371366676e-05, "loss": 0.0355, "step": 5630 }, { "grad_norm": 0.29175257682800293, "learning_rate": 9.99888370726282e-05, "loss": 0.0358, "step": 5640 }, { "grad_norm": 0.3031393885612488, "learning_rate": 9.998848496488556e-05, "loss": 0.037, "step": 5650 }, { "grad_norm": 0.3402264416217804, "learning_rate": 9.998812739047736e-05, "loss": 0.0449, "step": 5660 }, { "grad_norm": 0.29063650965690613, "learning_rate": 9.99877643494427e-05, "loss": 0.031, "step": 5670 }, { "grad_norm": 0.2904714345932007, "learning_rate": 9.998739584182128e-05, "loss": 0.0321, "step": 5680 }, { "grad_norm": 0.3563612699508667, "learning_rate": 9.998702186765342e-05, "loss": 0.0281, "step": 5690 }, { "grad_norm": 0.6657412052154541, "learning_rate": 9.998664242698e-05, "loss": 0.0472, "step": 5700 }, { "grad_norm": 0.24317966401576996, "learning_rate": 9.998625751984251e-05, "loss": 0.0397, "step": 5710 }, { "grad_norm": 0.33302614092826843, "learning_rate": 9.998586714628307e-05, "loss": 0.038, "step": 5720 }, { "grad_norm": 0.26986178755760193, "learning_rate": 9.998547130634432e-05, "loss": 0.0334, "step": 5730 }, { "grad_norm": 0.4625245928764343, "learning_rate": 9.99850700000696e-05, "loss": 0.0418, "step": 5740 }, { "grad_norm": 0.2461082935333252, "learning_rate": 9.998466322750278e-05, "loss": 0.0422, "step": 5750 }, { "grad_norm": 0.5352100133895874, "learning_rate": 9.998425098868834e-05, "loss": 0.0393, "step": 5760 }, { "grad_norm": 0.7631039023399353, "learning_rate": 9.998383328367136e-05, "loss": 0.0402, "step": 5770 }, { "grad_norm": 0.3222205936908722, "learning_rate": 9.99834101124975e-05, "loss": 0.0457, "step": 5780 }, { "grad_norm": 0.6851931810379028, "learning_rate": 9.998298147521309e-05, "loss": 0.0436, "step": 5790 }, { "grad_norm": 0.2736760377883911, "learning_rate": 9.998254737186496e-05, "loss": 0.0381, "step": 5800 }, { "grad_norm": 0.2677905559539795, "learning_rate": 9.99821078025006e-05, "loss": 0.03, "step": 5810 }, { "grad_norm": 0.6576104760169983, "learning_rate": 9.998166276716807e-05, "loss": 0.0358, "step": 5820 }, { "grad_norm": 0.2773016691207886, "learning_rate": 9.998121226591606e-05, "loss": 0.0488, "step": 5830 }, { "grad_norm": 0.24664273858070374, "learning_rate": 9.998075629879382e-05, "loss": 0.0375, "step": 5840 }, { "grad_norm": 0.4441488981246948, "learning_rate": 9.99802948658512e-05, "loss": 0.0551, "step": 5850 }, { "grad_norm": 0.26093950867652893, "learning_rate": 9.99798279671387e-05, "loss": 0.0406, "step": 5860 }, { "grad_norm": 0.30389097332954407, "learning_rate": 9.997935560270734e-05, "loss": 0.037, "step": 5870 }, { "grad_norm": 0.38229918479919434, "learning_rate": 9.997887777260879e-05, "loss": 0.0491, "step": 5880 }, { "grad_norm": 0.3296467065811157, "learning_rate": 9.997839447689532e-05, "loss": 0.0366, "step": 5890 }, { "grad_norm": 0.2995877265930176, "learning_rate": 9.997790571561978e-05, "loss": 0.035, "step": 5900 }, { "grad_norm": 0.16084317862987518, "learning_rate": 9.99774114888356e-05, "loss": 0.0329, "step": 5910 }, { "grad_norm": 0.2555338740348816, "learning_rate": 9.997691179659684e-05, "loss": 0.031, "step": 5920 }, { "grad_norm": 0.7143863439559937, "learning_rate": 9.997640663895815e-05, "loss": 0.0318, "step": 5930 }, { "grad_norm": 0.3992026448249817, "learning_rate": 9.997589601597477e-05, "loss": 0.0435, "step": 5940 }, { "grad_norm": 0.3721110224723816, "learning_rate": 9.997537992770252e-05, "loss": 0.0368, "step": 5950 }, { "grad_norm": 0.269164115190506, "learning_rate": 9.997485837419788e-05, "loss": 0.0351, "step": 5960 }, { "grad_norm": 0.5412243604660034, "learning_rate": 9.997433135551786e-05, "loss": 0.0438, "step": 5970 }, { "grad_norm": 0.28377190232276917, "learning_rate": 9.997379887172009e-05, "loss": 0.0357, "step": 5980 }, { "grad_norm": 0.2824063003063202, "learning_rate": 9.997326092286281e-05, "loss": 0.0332, "step": 5990 }, { "grad_norm": 0.6005157828330994, "learning_rate": 9.997271750900486e-05, "loss": 0.0549, "step": 6000 }, { "grad_norm": 0.35313737392425537, "learning_rate": 9.997216863020565e-05, "loss": 0.0448, "step": 6010 }, { "grad_norm": 0.472919762134552, "learning_rate": 9.99716142865252e-05, "loss": 0.0408, "step": 6020 }, { "grad_norm": 0.2766987681388855, "learning_rate": 9.997105447802415e-05, "loss": 0.0345, "step": 6030 }, { "grad_norm": 0.3644707500934601, "learning_rate": 9.997048920476373e-05, "loss": 0.0473, "step": 6040 }, { "grad_norm": 0.2572399377822876, "learning_rate": 9.996991846680572e-05, "loss": 0.0399, "step": 6050 }, { "grad_norm": 0.20419417321681976, "learning_rate": 9.996934226421257e-05, "loss": 0.0324, "step": 6060 }, { "grad_norm": 0.2333398461341858, "learning_rate": 9.996876059704726e-05, "loss": 0.0323, "step": 6070 }, { "grad_norm": 0.3323324918746948, "learning_rate": 9.996817346537343e-05, "loss": 0.033, "step": 6080 }, { "grad_norm": 0.26502060890197754, "learning_rate": 9.996758086925526e-05, "loss": 0.0252, "step": 6090 }, { "grad_norm": 0.6348937749862671, "learning_rate": 9.996698280875759e-05, "loss": 0.0493, "step": 6100 }, { "grad_norm": 0.2630722224712372, "learning_rate": 9.99663792839458e-05, "loss": 0.042, "step": 6110 }, { "grad_norm": 0.25541824102401733, "learning_rate": 9.99657702948859e-05, "loss": 0.0416, "step": 6120 }, { "grad_norm": 0.2482401579618454, "learning_rate": 9.996515584164448e-05, "loss": 0.0333, "step": 6130 }, { "grad_norm": 0.43698930740356445, "learning_rate": 9.996453592428873e-05, "loss": 0.036, "step": 6140 }, { "grad_norm": 0.3637663722038269, "learning_rate": 9.996391054288646e-05, "loss": 0.0404, "step": 6150 }, { "grad_norm": 0.41607585549354553, "learning_rate": 9.996327969750605e-05, "loss": 0.0379, "step": 6160 }, { "grad_norm": 0.32825595140457153, "learning_rate": 9.996264338821649e-05, "loss": 0.0391, "step": 6170 }, { "grad_norm": 0.23452270030975342, "learning_rate": 9.996200161508735e-05, "loss": 0.0324, "step": 6180 }, { "grad_norm": 0.20162396132946014, "learning_rate": 9.996135437818885e-05, "loss": 0.0278, "step": 6190 }, { "grad_norm": 0.6021217703819275, "learning_rate": 9.996070167759175e-05, "loss": 0.0398, "step": 6200 }, { "grad_norm": 0.3880918025970459, "learning_rate": 9.996004351336743e-05, "loss": 0.0432, "step": 6210 }, { "grad_norm": 0.25166428089141846, "learning_rate": 9.995937988558785e-05, "loss": 0.0367, "step": 6220 }, { "grad_norm": 0.2588699162006378, "learning_rate": 9.995871079432561e-05, "loss": 0.0368, "step": 6230 }, { "grad_norm": 0.3005032539367676, "learning_rate": 9.995803623965389e-05, "loss": 0.0373, "step": 6240 }, { "grad_norm": 0.2605488896369934, "learning_rate": 9.995735622164641e-05, "loss": 0.0385, "step": 6250 }, { "grad_norm": 0.38788551092147827, "learning_rate": 9.995667074037758e-05, "loss": 0.043, "step": 6260 }, { "grad_norm": 0.2793295383453369, "learning_rate": 9.995597979592232e-05, "loss": 0.0388, "step": 6270 }, { "grad_norm": 0.20517079532146454, "learning_rate": 9.995528338835625e-05, "loss": 0.0353, "step": 6280 }, { "grad_norm": 0.22532090544700623, "learning_rate": 9.995458151775547e-05, "loss": 0.0298, "step": 6290 }, { "grad_norm": 0.5845895409584045, "learning_rate": 9.995387418419677e-05, "loss": 0.0402, "step": 6300 }, { "grad_norm": 0.320211797952652, "learning_rate": 9.99531613877575e-05, "loss": 0.0417, "step": 6310 }, { "grad_norm": 0.27572837471961975, "learning_rate": 9.995244312851559e-05, "loss": 0.0338, "step": 6320 }, { "grad_norm": 0.2463844120502472, "learning_rate": 9.995171940654961e-05, "loss": 0.0346, "step": 6330 }, { "grad_norm": 0.32994213700294495, "learning_rate": 9.995099022193871e-05, "loss": 0.0352, "step": 6340 }, { "grad_norm": 0.29224729537963867, "learning_rate": 9.995025557476261e-05, "loss": 0.0368, "step": 6350 }, { "grad_norm": 0.20212271809577942, "learning_rate": 9.994951546510165e-05, "loss": 0.0357, "step": 6360 }, { "grad_norm": 0.6411901116371155, "learning_rate": 9.994876989303679e-05, "loss": 0.0379, "step": 6370 }, { "grad_norm": 0.2635655999183655, "learning_rate": 9.994801885864955e-05, "loss": 0.045, "step": 6380 }, { "grad_norm": 0.5868703126907349, "learning_rate": 9.994726236202205e-05, "loss": 0.0393, "step": 6390 }, { "grad_norm": 0.28050100803375244, "learning_rate": 9.994650040323704e-05, "loss": 0.0357, "step": 6400 }, { "grad_norm": 0.27197080850601196, "learning_rate": 9.994573298237784e-05, "loss": 0.0395, "step": 6410 }, { "grad_norm": 0.25543349981307983, "learning_rate": 9.994496009952837e-05, "loss": 0.0347, "step": 6420 }, { "grad_norm": 0.28868117928504944, "learning_rate": 9.994418175477316e-05, "loss": 0.0301, "step": 6430 }, { "grad_norm": 0.26204052567481995, "learning_rate": 9.994339794819733e-05, "loss": 0.0289, "step": 6440 }, { "grad_norm": 0.2551838457584381, "learning_rate": 9.994260867988658e-05, "loss": 0.031, "step": 6450 }, { "grad_norm": 0.258139431476593, "learning_rate": 9.994181394992723e-05, "loss": 0.0315, "step": 6460 }, { "grad_norm": 0.19277505576610565, "learning_rate": 9.994101375840618e-05, "loss": 0.0284, "step": 6470 }, { "grad_norm": 0.18526682257652283, "learning_rate": 9.994020810541098e-05, "loss": 0.0302, "step": 6480 }, { "grad_norm": 0.29654011130332947, "learning_rate": 9.99393969910297e-05, "loss": 0.0308, "step": 6490 }, { "grad_norm": 0.22101271152496338, "learning_rate": 9.993858041535104e-05, "loss": 0.0282, "step": 6500 }, { "grad_norm": 0.423742413520813, "learning_rate": 9.99377583784643e-05, "loss": 0.0341, "step": 6510 }, { "grad_norm": 0.22981004416942596, "learning_rate": 9.993693088045939e-05, "loss": 0.033, "step": 6520 }, { "grad_norm": 0.2572416067123413, "learning_rate": 9.99360979214268e-05, "loss": 0.0295, "step": 6530 }, { "grad_norm": 0.20377400517463684, "learning_rate": 9.99352595014576e-05, "loss": 0.0299, "step": 6540 }, { "grad_norm": 0.41812604665756226, "learning_rate": 9.993441562064354e-05, "loss": 0.0347, "step": 6550 }, { "grad_norm": 0.307941198348999, "learning_rate": 9.993356627907685e-05, "loss": 0.0373, "step": 6560 }, { "grad_norm": 0.25384944677352905, "learning_rate": 9.99327114768504e-05, "loss": 0.0347, "step": 6570 }, { "grad_norm": 0.22155553102493286, "learning_rate": 9.99318512140577e-05, "loss": 0.0281, "step": 6580 }, { "grad_norm": 0.4039427638053894, "learning_rate": 9.993098549079284e-05, "loss": 0.0362, "step": 6590 }, { "grad_norm": 0.27183854579925537, "learning_rate": 9.993011430715047e-05, "loss": 0.0405, "step": 6600 }, { "grad_norm": 0.5084763169288635, "learning_rate": 9.992923766322586e-05, "loss": 0.0471, "step": 6610 }, { "grad_norm": 0.2870672643184662, "learning_rate": 9.99283555591149e-05, "loss": 0.0378, "step": 6620 }, { "grad_norm": 0.24601082503795624, "learning_rate": 9.992746799491404e-05, "loss": 0.043, "step": 6630 }, { "grad_norm": 0.24255813658237457, "learning_rate": 9.992657497072033e-05, "loss": 0.034, "step": 6640 }, { "grad_norm": 0.23237363994121552, "learning_rate": 9.992567648663147e-05, "loss": 0.0284, "step": 6650 }, { "grad_norm": 0.3794462978839874, "learning_rate": 9.992477254274568e-05, "loss": 0.0467, "step": 6660 }, { "grad_norm": 0.2227509766817093, "learning_rate": 9.992386313916183e-05, "loss": 0.0392, "step": 6670 }, { "grad_norm": 0.3137526214122772, "learning_rate": 9.992294827597934e-05, "loss": 0.0448, "step": 6680 }, { "grad_norm": 0.34251531958580017, "learning_rate": 9.992202795329831e-05, "loss": 0.0409, "step": 6690 }, { "grad_norm": 0.183243989944458, "learning_rate": 9.992110217121936e-05, "loss": 0.0386, "step": 6700 }, { "grad_norm": 0.5287876725196838, "learning_rate": 9.992017092984372e-05, "loss": 0.0404, "step": 6710 }, { "grad_norm": 0.3547166883945465, "learning_rate": 9.991923422927326e-05, "loss": 0.0381, "step": 6720 }, { "grad_norm": 0.3421657979488373, "learning_rate": 9.991829206961037e-05, "loss": 0.0282, "step": 6730 }, { "grad_norm": 0.3305041790008545, "learning_rate": 9.991734445095813e-05, "loss": 0.038, "step": 6740 }, { "grad_norm": 0.2789493203163147, "learning_rate": 9.991639137342015e-05, "loss": 0.0413, "step": 6750 }, { "grad_norm": 0.3134191334247589, "learning_rate": 9.991543283710064e-05, "loss": 0.0336, "step": 6760 }, { "grad_norm": 0.6020836234092712, "learning_rate": 9.991446884210445e-05, "loss": 0.0379, "step": 6770 }, { "grad_norm": 0.4108709692955017, "learning_rate": 9.9913499388537e-05, "loss": 0.0447, "step": 6780 }, { "grad_norm": 0.28356269001960754, "learning_rate": 9.99125244765043e-05, "loss": 0.0358, "step": 6790 }, { "grad_norm": 0.7604698538780212, "learning_rate": 9.991154410611296e-05, "loss": 0.0398, "step": 6800 }, { "grad_norm": 0.27237603068351746, "learning_rate": 9.99105582774702e-05, "loss": 0.0471, "step": 6810 }, { "grad_norm": 0.2719794809818268, "learning_rate": 9.990956699068384e-05, "loss": 0.0406, "step": 6820 }, { "grad_norm": 0.282389760017395, "learning_rate": 9.990857024586224e-05, "loss": 0.0314, "step": 6830 }, { "grad_norm": 0.4825808107852936, "learning_rate": 9.990756804311446e-05, "loss": 0.0339, "step": 6840 }, { "grad_norm": 0.2721031606197357, "learning_rate": 9.990656038255006e-05, "loss": 0.0428, "step": 6850 }, { "grad_norm": 0.39156442880630493, "learning_rate": 9.990554726427926e-05, "loss": 0.0476, "step": 6860 }, { "grad_norm": 0.28590095043182373, "learning_rate": 9.990452868841284e-05, "loss": 0.0364, "step": 6870 }, { "grad_norm": 0.5877484679222107, "learning_rate": 9.99035046550622e-05, "loss": 0.0475, "step": 6880 }, { "grad_norm": 0.3432537019252777, "learning_rate": 9.99024751643393e-05, "loss": 0.0384, "step": 6890 }, { "grad_norm": 0.2290383130311966, "learning_rate": 9.990144021635677e-05, "loss": 0.0309, "step": 6900 }, { "grad_norm": 0.38641026616096497, "learning_rate": 9.990039981122775e-05, "loss": 0.0369, "step": 6910 }, { "grad_norm": 0.3620187044143677, "learning_rate": 9.989935394906602e-05, "loss": 0.0433, "step": 6920 }, { "grad_norm": 0.4211719334125519, "learning_rate": 9.989830262998598e-05, "loss": 0.0401, "step": 6930 }, { "grad_norm": 0.2736668586730957, "learning_rate": 9.989724585410259e-05, "loss": 0.0345, "step": 6940 }, { "grad_norm": 0.2706553041934967, "learning_rate": 9.989618362153139e-05, "loss": 0.0416, "step": 6950 }, { "grad_norm": 0.1842433661222458, "learning_rate": 9.989511593238859e-05, "loss": 0.0359, "step": 6960 }, { "grad_norm": 0.153316929936409, "learning_rate": 9.98940427867909e-05, "loss": 0.0299, "step": 6970 }, { "grad_norm": 0.41219693422317505, "learning_rate": 9.989296418485573e-05, "loss": 0.0423, "step": 6980 }, { "grad_norm": 0.26928403973579407, "learning_rate": 9.989188012670101e-05, "loss": 0.0354, "step": 6990 }, { "grad_norm": 0.41836440563201904, "learning_rate": 9.989079061244528e-05, "loss": 0.0389, "step": 7000 }, { "grad_norm": 0.5844922661781311, "learning_rate": 9.988969564220769e-05, "loss": 0.0431, "step": 7010 }, { "grad_norm": 0.3730790913105011, "learning_rate": 9.988859521610801e-05, "loss": 0.0412, "step": 7020 }, { "grad_norm": 0.4652373492717743, "learning_rate": 9.988748933426656e-05, "loss": 0.0329, "step": 7030 }, { "grad_norm": 0.3337727189064026, "learning_rate": 9.988637799680428e-05, "loss": 0.0344, "step": 7040 }, { "grad_norm": 0.29365643858909607, "learning_rate": 9.98852612038427e-05, "loss": 0.0297, "step": 7050 }, { "grad_norm": 0.6779335737228394, "learning_rate": 9.988413895550397e-05, "loss": 0.0338, "step": 7060 }, { "grad_norm": 0.34515467286109924, "learning_rate": 9.98830112519108e-05, "loss": 0.0382, "step": 7070 }, { "grad_norm": 0.1784866899251938, "learning_rate": 9.98818780931865e-05, "loss": 0.032, "step": 7080 }, { "grad_norm": 0.21257033944129944, "learning_rate": 9.988073947945502e-05, "loss": 0.0303, "step": 7090 }, { "grad_norm": 0.39139559864997864, "learning_rate": 9.987959541084087e-05, "loss": 0.0327, "step": 7100 }, { "grad_norm": 0.2719114422798157, "learning_rate": 9.987844588746915e-05, "loss": 0.0396, "step": 7110 }, { "grad_norm": 0.25185254216194153, "learning_rate": 9.987729090946558e-05, "loss": 0.0339, "step": 7120 }, { "grad_norm": 0.3755461871623993, "learning_rate": 9.987613047695647e-05, "loss": 0.0313, "step": 7130 }, { "grad_norm": 0.3084902763366699, "learning_rate": 9.987496459006871e-05, "loss": 0.0373, "step": 7140 }, { "grad_norm": 0.5531898140907288, "learning_rate": 9.987379324892982e-05, "loss": 0.0319, "step": 7150 }, { "grad_norm": 0.31776919960975647, "learning_rate": 9.987261645366788e-05, "loss": 0.0331, "step": 7160 }, { "grad_norm": 0.21583890914916992, "learning_rate": 9.987143420441158e-05, "loss": 0.0271, "step": 7170 }, { "grad_norm": 0.7494268417358398, "learning_rate": 9.987024650129022e-05, "loss": 0.0324, "step": 7180 }, { "grad_norm": 0.3492940366268158, "learning_rate": 9.986905334443368e-05, "loss": 0.0375, "step": 7190 }, { "grad_norm": 0.4215664863586426, "learning_rate": 9.986785473397245e-05, "loss": 0.036, "step": 7200 }, { "grad_norm": 0.27195483446121216, "learning_rate": 9.98666506700376e-05, "loss": 0.0402, "step": 7210 }, { "grad_norm": 0.29118847846984863, "learning_rate": 9.986544115276081e-05, "loss": 0.0344, "step": 7220 }, { "grad_norm": 0.2549228370189667, "learning_rate": 9.986422618227433e-05, "loss": 0.031, "step": 7230 }, { "grad_norm": 0.3062067925930023, "learning_rate": 9.986300575871106e-05, "loss": 0.0394, "step": 7240 }, { "grad_norm": 0.20692333579063416, "learning_rate": 9.986177988220444e-05, "loss": 0.0322, "step": 7250 }, { "grad_norm": 0.39814212918281555, "learning_rate": 9.986054855288856e-05, "loss": 0.0399, "step": 7260 }, { "grad_norm": 0.3707737624645233, "learning_rate": 9.985931177089802e-05, "loss": 0.0315, "step": 7270 }, { "grad_norm": 0.2681118845939636, "learning_rate": 9.985806953636814e-05, "loss": 0.0254, "step": 7280 }, { "grad_norm": 0.3069770336151123, "learning_rate": 9.985682184943471e-05, "loss": 0.0398, "step": 7290 }, { "grad_norm": 0.3362145721912384, "learning_rate": 9.98555687102342e-05, "loss": 0.0371, "step": 7300 }, { "grad_norm": 0.2568901479244232, "learning_rate": 9.985431011890367e-05, "loss": 0.0354, "step": 7310 }, { "grad_norm": 0.23296138644218445, "learning_rate": 9.985304607558075e-05, "loss": 0.0319, "step": 7320 }, { "grad_norm": 0.22223301231861115, "learning_rate": 9.985177658040364e-05, "loss": 0.0338, "step": 7330 }, { "grad_norm": 0.3545326292514801, "learning_rate": 9.985050163351119e-05, "loss": 0.0341, "step": 7340 }, { "grad_norm": 0.30784186720848083, "learning_rate": 9.984922123504286e-05, "loss": 0.0365, "step": 7350 }, { "grad_norm": 0.3175942003726959, "learning_rate": 9.984793538513862e-05, "loss": 0.0366, "step": 7360 }, { "grad_norm": 0.24929219484329224, "learning_rate": 9.984664408393912e-05, "loss": 0.0326, "step": 7370 }, { "grad_norm": 0.37538158893585205, "learning_rate": 9.984534733158556e-05, "loss": 0.0365, "step": 7380 }, { "grad_norm": 0.27328452467918396, "learning_rate": 9.984404512821977e-05, "loss": 0.0314, "step": 7390 }, { "grad_norm": 0.23738035559654236, "learning_rate": 9.984273747398411e-05, "loss": 0.0278, "step": 7400 }, { "grad_norm": 0.4158852696418762, "learning_rate": 9.984142436902165e-05, "loss": 0.0401, "step": 7410 }, { "grad_norm": 0.2690909206867218, "learning_rate": 9.984010581347596e-05, "loss": 0.0326, "step": 7420 }, { "grad_norm": 0.3282727599143982, "learning_rate": 9.983878180749121e-05, "loss": 0.0441, "step": 7430 }, { "grad_norm": 0.27532491087913513, "learning_rate": 9.983745235121222e-05, "loss": 0.0368, "step": 7440 }, { "grad_norm": 0.37745150923728943, "learning_rate": 9.983611744478438e-05, "loss": 0.0372, "step": 7450 }, { "grad_norm": 0.20027321577072144, "learning_rate": 9.983477708835365e-05, "loss": 0.0317, "step": 7460 }, { "grad_norm": 0.2864284813404083, "learning_rate": 9.983343128206664e-05, "loss": 0.0355, "step": 7470 }, { "grad_norm": 0.519778847694397, "learning_rate": 9.983208002607049e-05, "loss": 0.0384, "step": 7480 }, { "grad_norm": 0.259616494178772, "learning_rate": 9.9830723320513e-05, "loss": 0.0364, "step": 7490 }, { "grad_norm": 0.3611123263835907, "learning_rate": 9.982936116554254e-05, "loss": 0.0412, "step": 7500 }, { "grad_norm": 0.2299681156873703, "learning_rate": 9.982799356130803e-05, "loss": 0.0314, "step": 7510 }, { "grad_norm": 0.4699168801307678, "learning_rate": 9.982662050795908e-05, "loss": 0.0378, "step": 7520 }, { "grad_norm": 0.2888623774051666, "learning_rate": 9.982524200564583e-05, "loss": 0.0347, "step": 7530 }, { "grad_norm": 0.22246316075325012, "learning_rate": 9.982385805451901e-05, "loss": 0.0287, "step": 7540 }, { "grad_norm": 0.20738917589187622, "learning_rate": 9.982246865472998e-05, "loss": 0.0263, "step": 7550 }, { "grad_norm": 0.27011027932167053, "learning_rate": 9.982107380643069e-05, "loss": 0.0285, "step": 7560 }, { "grad_norm": 0.22082927823066711, "learning_rate": 9.981967350977368e-05, "loss": 0.023, "step": 7570 }, { "grad_norm": 0.6308929920196533, "learning_rate": 9.981826776491208e-05, "loss": 0.0442, "step": 7580 }, { "grad_norm": 0.22371616959571838, "learning_rate": 9.98168565719996e-05, "loss": 0.0364, "step": 7590 }, { "grad_norm": 0.385815292596817, "learning_rate": 9.98154399311906e-05, "loss": 0.0363, "step": 7600 }, { "grad_norm": 0.2500925064086914, "learning_rate": 9.981401784263997e-05, "loss": 0.0333, "step": 7610 }, { "grad_norm": 0.4583759307861328, "learning_rate": 9.981259030650326e-05, "loss": 0.0394, "step": 7620 }, { "grad_norm": 0.2194536030292511, "learning_rate": 9.981115732293655e-05, "loss": 0.0325, "step": 7630 }, { "grad_norm": 0.28733474016189575, "learning_rate": 9.980971889209659e-05, "loss": 0.0368, "step": 7640 }, { "grad_norm": 0.15653665363788605, "learning_rate": 9.980827501414064e-05, "loss": 0.0333, "step": 7650 }, { "grad_norm": 0.34324905276298523, "learning_rate": 9.980682568922663e-05, "loss": 0.0307, "step": 7660 }, { "grad_norm": 0.22281065583229065, "learning_rate": 9.980537091751304e-05, "loss": 0.0347, "step": 7670 }, { "grad_norm": 0.4241080582141876, "learning_rate": 9.980391069915897e-05, "loss": 0.0402, "step": 7680 }, { "grad_norm": 0.24152278900146484, "learning_rate": 9.98024450343241e-05, "loss": 0.0365, "step": 7690 }, { "grad_norm": 0.2148476243019104, "learning_rate": 9.980097392316872e-05, "loss": 0.0302, "step": 7700 }, { "grad_norm": 0.36256974935531616, "learning_rate": 9.97994973658537e-05, "loss": 0.0362, "step": 7710 }, { "grad_norm": 0.2579337954521179, "learning_rate": 9.979801536254054e-05, "loss": 0.0321, "step": 7720 }, { "grad_norm": 0.38994717597961426, "learning_rate": 9.979652791339127e-05, "loss": 0.0355, "step": 7730 }, { "grad_norm": 0.19979256391525269, "learning_rate": 9.97950350185686e-05, "loss": 0.031, "step": 7740 }, { "grad_norm": 0.26832345128059387, "learning_rate": 9.979353667823574e-05, "loss": 0.0334, "step": 7750 }, { "grad_norm": 0.1933753341436386, "learning_rate": 9.979203289255658e-05, "loss": 0.033, "step": 7760 }, { "grad_norm": 0.4669051468372345, "learning_rate": 9.979052366169557e-05, "loss": 0.0327, "step": 7770 }, { "grad_norm": 0.26198598742485046, "learning_rate": 9.978900898581775e-05, "loss": 0.031, "step": 7780 }, { "grad_norm": 0.25316694378852844, "learning_rate": 9.978748886508875e-05, "loss": 0.0343, "step": 7790 }, { "grad_norm": 0.19291941821575165, "learning_rate": 9.978596329967484e-05, "loss": 0.0322, "step": 7800 }, { "grad_norm": 0.21753986179828644, "learning_rate": 9.978443228974284e-05, "loss": 0.0301, "step": 7810 }, { "grad_norm": 0.47987812757492065, "learning_rate": 9.978289583546015e-05, "loss": 0.0299, "step": 7820 }, { "grad_norm": 0.21171079576015472, "learning_rate": 9.978135393699484e-05, "loss": 0.0307, "step": 7830 }, { "grad_norm": 0.7340234518051147, "learning_rate": 9.977980659451548e-05, "loss": 0.0328, "step": 7840 }, { "grad_norm": 0.25307750701904297, "learning_rate": 9.977825380819135e-05, "loss": 0.0323, "step": 7850 }, { "grad_norm": 0.2771221101284027, "learning_rate": 9.97766955781922e-05, "loss": 0.0272, "step": 7860 }, { "grad_norm": 0.34203848242759705, "learning_rate": 9.977513190468848e-05, "loss": 0.0244, "step": 7870 }, { "grad_norm": 0.3363509178161621, "learning_rate": 9.977356278785116e-05, "loss": 0.0387, "step": 7880 }, { "grad_norm": 0.2321346551179886, "learning_rate": 9.977198822785184e-05, "loss": 0.0295, "step": 7890 }, { "grad_norm": 0.23269282281398773, "learning_rate": 9.977040822486273e-05, "loss": 0.034, "step": 7900 }, { "grad_norm": 0.3863184154033661, "learning_rate": 9.97688227790566e-05, "loss": 0.0359, "step": 7910 }, { "grad_norm": 0.2525266408920288, "learning_rate": 9.976723189060684e-05, "loss": 0.0312, "step": 7920 }, { "grad_norm": 0.2524719834327698, "learning_rate": 9.976563555968742e-05, "loss": 0.0294, "step": 7930 }, { "grad_norm": 0.2607014775276184, "learning_rate": 9.976403378647292e-05, "loss": 0.0376, "step": 7940 }, { "grad_norm": 0.28670069575309753, "learning_rate": 9.97624265711385e-05, "loss": 0.0262, "step": 7950 }, { "grad_norm": 0.2317831963300705, "learning_rate": 9.976081391385993e-05, "loss": 0.0283, "step": 7960 }, { "grad_norm": 0.24963492155075073, "learning_rate": 9.975919581481356e-05, "loss": 0.0327, "step": 7970 }, { "grad_norm": 0.22054731845855713, "learning_rate": 9.975757227417634e-05, "loss": 0.0304, "step": 7980 }, { "grad_norm": 0.317367285490036, "learning_rate": 9.975594329212586e-05, "loss": 0.0294, "step": 7990 }, { "grad_norm": 0.18375040590763092, "learning_rate": 9.97543088688402e-05, "loss": 0.0255, "step": 8000 }, { "grad_norm": 0.2861368656158447, "learning_rate": 9.975266900449814e-05, "loss": 0.0328, "step": 8010 }, { "grad_norm": 0.1740194708108902, "learning_rate": 9.975102369927898e-05, "loss": 0.0319, "step": 8020 }, { "grad_norm": 0.19911468029022217, "learning_rate": 9.974937295336269e-05, "loss": 0.0254, "step": 8030 }, { "grad_norm": 0.41777169704437256, "learning_rate": 9.974771676692975e-05, "loss": 0.029, "step": 8040 }, { "grad_norm": 0.26208359003067017, "learning_rate": 9.974605514016131e-05, "loss": 0.0298, "step": 8050 }, { "grad_norm": 0.25368624925613403, "learning_rate": 9.974438807323907e-05, "loss": 0.0295, "step": 8060 }, { "grad_norm": 0.25916093587875366, "learning_rate": 9.974271556634535e-05, "loss": 0.0317, "step": 8070 }, { "grad_norm": 0.2069210261106491, "learning_rate": 9.974103761966302e-05, "loss": 0.029, "step": 8080 }, { "grad_norm": 0.212530255317688, "learning_rate": 9.973935423337563e-05, "loss": 0.0233, "step": 8090 }, { "grad_norm": 0.16256047785282135, "learning_rate": 9.973766540766722e-05, "loss": 0.0246, "step": 8100 }, { "grad_norm": 0.21349100768566132, "learning_rate": 9.97359711427225e-05, "loss": 0.0291, "step": 8110 }, { "grad_norm": 0.2931903004646301, "learning_rate": 9.973427143872677e-05, "loss": 0.0293, "step": 8120 }, { "grad_norm": 0.24969616532325745, "learning_rate": 9.973256629586589e-05, "loss": 0.0312, "step": 8130 }, { "grad_norm": 0.19570402801036835, "learning_rate": 9.973085571432632e-05, "loss": 0.0297, "step": 8140 }, { "grad_norm": 0.21634531021118164, "learning_rate": 9.972913969429513e-05, "loss": 0.027, "step": 8150 }, { "grad_norm": 0.3645780384540558, "learning_rate": 9.972741823596e-05, "loss": 0.0288, "step": 8160 }, { "grad_norm": 0.25197383761405945, "learning_rate": 9.972569133950917e-05, "loss": 0.0298, "step": 8170 }, { "grad_norm": 0.3695439100265503, "learning_rate": 9.972395900513151e-05, "loss": 0.033, "step": 8180 }, { "grad_norm": 0.213196262717247, "learning_rate": 9.972222123301645e-05, "loss": 0.0306, "step": 8190 }, { "grad_norm": 0.29261064529418945, "learning_rate": 9.972047802335403e-05, "loss": 0.0299, "step": 8200 }, { "grad_norm": 0.2129892259836197, "learning_rate": 9.971872937633488e-05, "loss": 0.0279, "step": 8210 }, { "grad_norm": 0.31256675720214844, "learning_rate": 9.971697529215024e-05, "loss": 0.0287, "step": 8220 }, { "grad_norm": 0.22999991476535797, "learning_rate": 9.971521577099192e-05, "loss": 0.0295, "step": 8230 }, { "grad_norm": 0.23647402226924896, "learning_rate": 9.971345081305236e-05, "loss": 0.0242, "step": 8240 }, { "grad_norm": 0.16042205691337585, "learning_rate": 9.971168041852456e-05, "loss": 0.0242, "step": 8250 }, { "grad_norm": 0.1574583351612091, "learning_rate": 9.970990458760215e-05, "loss": 0.0274, "step": 8260 }, { "grad_norm": 0.24083459377288818, "learning_rate": 9.970812332047929e-05, "loss": 0.0405, "step": 8270 }, { "grad_norm": 0.17347227036952972, "learning_rate": 9.97063366173508e-05, "loss": 0.033, "step": 8280 }, { "grad_norm": 0.15780337154865265, "learning_rate": 9.970454447841207e-05, "loss": 0.029, "step": 8290 }, { "grad_norm": 0.1875482052564621, "learning_rate": 9.970274690385909e-05, "loss": 0.0238, "step": 8300 }, { "grad_norm": 0.24510684609413147, "learning_rate": 9.970094389388844e-05, "loss": 0.0281, "step": 8310 }, { "grad_norm": 0.2671508491039276, "learning_rate": 9.969913544869728e-05, "loss": 0.0264, "step": 8320 }, { "grad_norm": 0.23552794754505157, "learning_rate": 9.96973215684834e-05, "loss": 0.0242, "step": 8330 }, { "grad_norm": 0.22690939903259277, "learning_rate": 9.969550225344513e-05, "loss": 0.0226, "step": 8340 }, { "grad_norm": 0.20769000053405762, "learning_rate": 9.969367750378147e-05, "loss": 0.0222, "step": 8350 }, { "grad_norm": 0.26047247648239136, "learning_rate": 9.969184731969194e-05, "loss": 0.0212, "step": 8360 }, { "grad_norm": 0.19254350662231445, "learning_rate": 9.96900117013767e-05, "loss": 0.0226, "step": 8370 }, { "grad_norm": 0.18528860807418823, "learning_rate": 9.96881706490365e-05, "loss": 0.0217, "step": 8380 }, { "grad_norm": 0.3932465612888336, "learning_rate": 9.968632416287265e-05, "loss": 0.0323, "step": 8390 }, { "grad_norm": 0.27128857374191284, "learning_rate": 9.96844722430871e-05, "loss": 0.0305, "step": 8400 }, { "grad_norm": 0.34456297755241394, "learning_rate": 9.968261488988235e-05, "loss": 0.0334, "step": 8410 }, { "grad_norm": 0.22941960394382477, "learning_rate": 9.968075210346155e-05, "loss": 0.0295, "step": 8420 }, { "grad_norm": 0.21529731154441833, "learning_rate": 9.967888388402839e-05, "loss": 0.0236, "step": 8430 }, { "grad_norm": 0.15563373267650604, "learning_rate": 9.967701023178717e-05, "loss": 0.0259, "step": 8440 }, { "grad_norm": 0.17398950457572937, "learning_rate": 9.967513114694282e-05, "loss": 0.024, "step": 8450 }, { "grad_norm": 0.3129527270793915, "learning_rate": 9.967324662970079e-05, "loss": 0.0325, "step": 8460 }, { "grad_norm": 0.2054041028022766, "learning_rate": 9.96713566802672e-05, "loss": 0.0303, "step": 8470 }, { "grad_norm": 0.2982199192047119, "learning_rate": 9.966946129884873e-05, "loss": 0.0313, "step": 8480 }, { "grad_norm": 0.1948663741350174, "learning_rate": 9.966756048565265e-05, "loss": 0.0254, "step": 8490 }, { "grad_norm": 0.26727887988090515, "learning_rate": 9.966565424088681e-05, "loss": 0.0348, "step": 8500 }, { "grad_norm": 0.2597026824951172, "learning_rate": 9.96637425647597e-05, "loss": 0.031, "step": 8510 }, { "grad_norm": 0.19194579124450684, "learning_rate": 9.966182545748038e-05, "loss": 0.0279, "step": 8520 }, { "grad_norm": 0.1763199269771576, "learning_rate": 9.96599029192585e-05, "loss": 0.0297, "step": 8530 }, { "grad_norm": 0.32415756583213806, "learning_rate": 9.965797495030428e-05, "loss": 0.0312, "step": 8540 }, { "grad_norm": 0.2559697926044464, "learning_rate": 9.96560415508286e-05, "loss": 0.0328, "step": 8550 }, { "grad_norm": 0.3653196394443512, "learning_rate": 9.965410272104286e-05, "loss": 0.0324, "step": 8560 }, { "grad_norm": 0.3061557710170746, "learning_rate": 9.96521584611591e-05, "loss": 0.0267, "step": 8570 }, { "grad_norm": 0.26278385519981384, "learning_rate": 9.965020877138994e-05, "loss": 0.0267, "step": 8580 }, { "grad_norm": 0.2419290393590927, "learning_rate": 9.964825365194861e-05, "loss": 0.0238, "step": 8590 }, { "grad_norm": 0.21749669313430786, "learning_rate": 9.96462931030489e-05, "loss": 0.0204, "step": 8600 }, { "grad_norm": 0.22272999584674835, "learning_rate": 9.96443271249052e-05, "loss": 0.0207, "step": 8610 }, { "grad_norm": 0.6567922830581665, "learning_rate": 9.964235571773255e-05, "loss": 0.0265, "step": 8620 }, { "grad_norm": 0.27312758564949036, "learning_rate": 9.96403788817465e-05, "loss": 0.029, "step": 8630 }, { "grad_norm": 0.1913338005542755, "learning_rate": 9.963839661716325e-05, "loss": 0.0274, "step": 8640 }, { "grad_norm": 0.15483179688453674, "learning_rate": 9.963640892419958e-05, "loss": 0.0258, "step": 8650 }, { "grad_norm": 0.3090114891529083, "learning_rate": 9.963441580307286e-05, "loss": 0.0303, "step": 8660 }, { "grad_norm": 0.2450549304485321, "learning_rate": 9.963241725400104e-05, "loss": 0.0323, "step": 8670 }, { "grad_norm": 0.20816005766391754, "learning_rate": 9.963041327720271e-05, "loss": 0.0267, "step": 8680 }, { "grad_norm": 0.3739282786846161, "learning_rate": 9.962840387289697e-05, "loss": 0.0274, "step": 8690 }, { "grad_norm": 0.2656397223472595, "learning_rate": 9.962638904130363e-05, "loss": 0.0269, "step": 8700 }, { "grad_norm": 0.2411094456911087, "learning_rate": 9.962436878264298e-05, "loss": 0.0307, "step": 8710 }, { "grad_norm": 0.26333221793174744, "learning_rate": 9.962234309713598e-05, "loss": 0.0285, "step": 8720 }, { "grad_norm": 0.3463946580886841, "learning_rate": 9.962031198500414e-05, "loss": 0.0325, "step": 8730 }, { "grad_norm": 0.26919081807136536, "learning_rate": 9.961827544646958e-05, "loss": 0.0302, "step": 8740 }, { "grad_norm": 0.3016621172428131, "learning_rate": 9.961623348175501e-05, "loss": 0.0361, "step": 8750 }, { "grad_norm": 0.21427421271800995, "learning_rate": 9.961418609108377e-05, "loss": 0.0277, "step": 8760 }, { "grad_norm": 0.14483334124088287, "learning_rate": 9.961213327467971e-05, "loss": 0.021, "step": 8770 }, { "grad_norm": 0.1663583219051361, "learning_rate": 9.961007503276736e-05, "loss": 0.0228, "step": 8780 }, { "grad_norm": 0.3427645266056061, "learning_rate": 9.960801136557179e-05, "loss": 0.0255, "step": 8790 }, { "grad_norm": 0.348026841878891, "learning_rate": 9.960594227331866e-05, "loss": 0.031, "step": 8800 }, { "grad_norm": 0.29679763317108154, "learning_rate": 9.960386775623429e-05, "loss": 0.0331, "step": 8810 }, { "grad_norm": 0.2325839251279831, "learning_rate": 9.96017878145455e-05, "loss": 0.028, "step": 8820 }, { "grad_norm": 0.3142769932746887, "learning_rate": 9.959970244847977e-05, "loss": 0.0398, "step": 8830 }, { "grad_norm": 0.21400749683380127, "learning_rate": 9.959761165826518e-05, "loss": 0.0291, "step": 8840 }, { "grad_norm": 0.18945667147636414, "learning_rate": 9.959551544413033e-05, "loss": 0.0298, "step": 8850 }, { "grad_norm": 0.24004776775836945, "learning_rate": 9.959341380630448e-05, "loss": 0.0385, "step": 8860 }, { "grad_norm": 0.23646759986877441, "learning_rate": 9.959130674501746e-05, "loss": 0.0326, "step": 8870 }, { "grad_norm": 0.2065841108560562, "learning_rate": 9.958919426049968e-05, "loss": 0.0254, "step": 8880 }, { "grad_norm": 0.22764067351818085, "learning_rate": 9.958707635298219e-05, "loss": 0.0281, "step": 8890 }, { "grad_norm": 0.2121032327413559, "learning_rate": 9.958495302269657e-05, "loss": 0.0255, "step": 8900 }, { "grad_norm": 0.20343008637428284, "learning_rate": 9.958282426987503e-05, "loss": 0.0242, "step": 8910 }, { "grad_norm": 0.2563984990119934, "learning_rate": 9.95806900947504e-05, "loss": 0.0307, "step": 8920 }, { "grad_norm": 0.24557620286941528, "learning_rate": 9.957855049755604e-05, "loss": 0.0298, "step": 8930 }, { "grad_norm": 0.20293322205543518, "learning_rate": 9.957640547852593e-05, "loss": 0.0282, "step": 8940 }, { "grad_norm": 0.275467187166214, "learning_rate": 9.957425503789466e-05, "loss": 0.0273, "step": 8950 }, { "grad_norm": 0.25449445843696594, "learning_rate": 9.957209917589738e-05, "loss": 0.0307, "step": 8960 }, { "grad_norm": 0.2008606493473053, "learning_rate": 9.956993789276987e-05, "loss": 0.0285, "step": 8970 }, { "grad_norm": 0.26161256432533264, "learning_rate": 9.956777118874847e-05, "loss": 0.0295, "step": 8980 }, { "grad_norm": 0.23192501068115234, "learning_rate": 9.956559906407016e-05, "loss": 0.0294, "step": 8990 }, { "grad_norm": 0.17698311805725098, "learning_rate": 9.956342151897245e-05, "loss": 0.0286, "step": 9000 }, { "grad_norm": 0.13460135459899902, "learning_rate": 9.956123855369346e-05, "loss": 0.0271, "step": 9010 }, { "grad_norm": 0.14787212014198303, "learning_rate": 9.955905016847196e-05, "loss": 0.0205, "step": 9020 }, { "grad_norm": 0.17928056418895721, "learning_rate": 9.955685636354723e-05, "loss": 0.0251, "step": 9030 }, { "grad_norm": 0.1968584954738617, "learning_rate": 9.95546571391592e-05, "loss": 0.0247, "step": 9040 }, { "grad_norm": 0.20046646893024445, "learning_rate": 9.955245249554837e-05, "loss": 0.0209, "step": 9050 }, { "grad_norm": 0.33062663674354553, "learning_rate": 9.955024243295582e-05, "loss": 0.0275, "step": 9060 }, { "grad_norm": 0.2968083918094635, "learning_rate": 9.954802695162328e-05, "loss": 0.026, "step": 9070 }, { "grad_norm": 0.20833157002925873, "learning_rate": 9.954580605179302e-05, "loss": 0.0323, "step": 9080 }, { "grad_norm": 0.21091879904270172, "learning_rate": 9.954357973370788e-05, "loss": 0.0289, "step": 9090 }, { "grad_norm": 0.2874593734741211, "learning_rate": 9.954134799761135e-05, "loss": 0.0296, "step": 9100 }, { "grad_norm": 0.3893241286277771, "learning_rate": 9.953911084374748e-05, "loss": 0.0318, "step": 9110 }, { "grad_norm": 0.29822519421577454, "learning_rate": 9.953686827236093e-05, "loss": 0.0324, "step": 9120 }, { "grad_norm": 0.25114133954048157, "learning_rate": 9.953462028369695e-05, "loss": 0.0308, "step": 9130 }, { "grad_norm": 0.26086506247520447, "learning_rate": 9.953236687800136e-05, "loss": 0.0301, "step": 9140 }, { "grad_norm": 0.496869832277298, "learning_rate": 9.95301080555206e-05, "loss": 0.0276, "step": 9150 }, { "grad_norm": 0.3346482813358307, "learning_rate": 9.952784381650171e-05, "loss": 0.0287, "step": 9160 }, { "grad_norm": 0.24964818358421326, "learning_rate": 9.952557416119226e-05, "loss": 0.0266, "step": 9170 }, { "grad_norm": 0.3432564437389374, "learning_rate": 9.95232990898405e-05, "loss": 0.0273, "step": 9180 }, { "grad_norm": 0.2648157775402069, "learning_rate": 9.95210186026952e-05, "loss": 0.0313, "step": 9190 }, { "grad_norm": 0.3355536460876465, "learning_rate": 9.951873270000576e-05, "loss": 0.0297, "step": 9200 }, { "grad_norm": 0.34145697951316833, "learning_rate": 9.951644138202216e-05, "loss": 0.031, "step": 9210 }, { "grad_norm": 0.26371413469314575, "learning_rate": 9.951414464899498e-05, "loss": 0.0269, "step": 9220 }, { "grad_norm": 0.36502552032470703, "learning_rate": 9.951184250117538e-05, "loss": 0.0296, "step": 9230 }, { "grad_norm": 0.30856022238731384, "learning_rate": 9.950953493881513e-05, "loss": 0.0347, "step": 9240 }, { "grad_norm": 0.3102133572101593, "learning_rate": 9.950722196216658e-05, "loss": 0.026, "step": 9250 }, { "grad_norm": 0.2624090611934662, "learning_rate": 9.950490357148265e-05, "loss": 0.0283, "step": 9260 }, { "grad_norm": 0.2278311401605606, "learning_rate": 9.950257976701692e-05, "loss": 0.0287, "step": 9270 }, { "grad_norm": 0.2873707115650177, "learning_rate": 9.950025054902348e-05, "loss": 0.0299, "step": 9280 }, { "grad_norm": 0.36865153908729553, "learning_rate": 9.949791591775706e-05, "loss": 0.033, "step": 9290 }, { "grad_norm": 0.2166881561279297, "learning_rate": 9.949557587347298e-05, "loss": 0.0284, "step": 9300 }, { "grad_norm": 0.4247724711894989, "learning_rate": 9.949323041642713e-05, "loss": 0.0333, "step": 9310 }, { "grad_norm": 0.2492968589067459, "learning_rate": 9.949087954687602e-05, "loss": 0.0246, "step": 9320 }, { "grad_norm": 0.1893078237771988, "learning_rate": 9.948852326507672e-05, "loss": 0.0206, "step": 9330 }, { "grad_norm": 0.3645373582839966, "learning_rate": 9.948616157128694e-05, "loss": 0.0339, "step": 9340 }, { "grad_norm": 0.18893197178840637, "learning_rate": 9.948379446576493e-05, "loss": 0.0283, "step": 9350 }, { "grad_norm": 0.31828010082244873, "learning_rate": 9.948142194876952e-05, "loss": 0.0323, "step": 9360 }, { "grad_norm": 0.1962258368730545, "learning_rate": 9.947904402056024e-05, "loss": 0.024, "step": 9370 }, { "grad_norm": 0.15474185347557068, "learning_rate": 9.947666068139708e-05, "loss": 0.0244, "step": 9380 }, { "grad_norm": 0.19272875785827637, "learning_rate": 9.947427193154071e-05, "loss": 0.0196, "step": 9390 }, { "grad_norm": 0.2546410858631134, "learning_rate": 9.947187777125233e-05, "loss": 0.0207, "step": 9400 }, { "grad_norm": 0.20354792475700378, "learning_rate": 9.946947820079377e-05, "loss": 0.0163, "step": 9410 }, { "grad_norm": 0.312989741563797, "learning_rate": 9.946707322042747e-05, "loss": 0.0275, "step": 9420 }, { "grad_norm": 0.176559180021286, "learning_rate": 9.94646628304164e-05, "loss": 0.0283, "step": 9430 }, { "grad_norm": 0.1680167317390442, "learning_rate": 9.946224703102418e-05, "loss": 0.0227, "step": 9440 }, { "grad_norm": 0.18376986682415009, "learning_rate": 9.945982582251498e-05, "loss": 0.0241, "step": 9450 }, { "grad_norm": 0.28203025460243225, "learning_rate": 9.94573992051536e-05, "loss": 0.0232, "step": 9460 }, { "grad_norm": 0.411898672580719, "learning_rate": 9.94549671792054e-05, "loss": 0.0252, "step": 9470 }, { "grad_norm": 0.24079129099845886, "learning_rate": 9.945252974493635e-05, "loss": 0.0282, "step": 9480 }, { "grad_norm": 0.4559076428413391, "learning_rate": 9.9450086902613e-05, "loss": 0.0244, "step": 9490 }, { "grad_norm": 0.34305453300476074, "learning_rate": 9.944763865250248e-05, "loss": 0.0254, "step": 9500 }, { "grad_norm": 0.33388394117355347, "learning_rate": 9.944518499487254e-05, "loss": 0.0312, "step": 9510 }, { "grad_norm": 0.16958823800086975, "learning_rate": 9.944272592999151e-05, "loss": 0.025, "step": 9520 }, { "grad_norm": 0.14477455615997314, "learning_rate": 9.94402614581283e-05, "loss": 0.0182, "step": 9530 }, { "grad_norm": 0.17153532803058624, "learning_rate": 9.943779157955244e-05, "loss": 0.0204, "step": 9540 }, { "grad_norm": 0.17398658394813538, "learning_rate": 9.943531629453403e-05, "loss": 0.0266, "step": 9550 }, { "grad_norm": 0.20823988318443298, "learning_rate": 9.943283560334375e-05, "loss": 0.0225, "step": 9560 }, { "grad_norm": 0.325141042470932, "learning_rate": 9.943034950625288e-05, "loss": 0.0255, "step": 9570 }, { "grad_norm": 0.23822179436683655, "learning_rate": 9.942785800353332e-05, "loss": 0.029, "step": 9580 }, { "grad_norm": 0.1995144635438919, "learning_rate": 9.942536109545751e-05, "loss": 0.0318, "step": 9590 }, { "grad_norm": 0.19169843196868896, "learning_rate": 9.942285878229853e-05, "loss": 0.0227, "step": 9600 }, { "grad_norm": 0.2610032260417938, "learning_rate": 9.942035106433001e-05, "loss": 0.029, "step": 9610 }, { "grad_norm": 0.19679105281829834, "learning_rate": 9.94178379418262e-05, "loss": 0.0198, "step": 9620 }, { "grad_norm": 0.27783122658729553, "learning_rate": 9.941531941506194e-05, "loss": 0.0202, "step": 9630 }, { "grad_norm": 0.2560805678367615, "learning_rate": 9.941279548431263e-05, "loss": 0.0318, "step": 9640 }, { "grad_norm": 0.21296542882919312, "learning_rate": 9.941026614985431e-05, "loss": 0.0268, "step": 9650 }, { "grad_norm": 0.16698366403579712, "learning_rate": 9.940773141196357e-05, "loss": 0.0247, "step": 9660 }, { "grad_norm": 0.2604227364063263, "learning_rate": 9.94051912709176e-05, "loss": 0.0239, "step": 9670 }, { "grad_norm": 0.22078010439872742, "learning_rate": 9.940264572699421e-05, "loss": 0.0207, "step": 9680 }, { "grad_norm": 0.2907180190086365, "learning_rate": 9.940009478047174e-05, "loss": 0.0279, "step": 9690 }, { "grad_norm": 0.2052316516637802, "learning_rate": 9.939753843162918e-05, "loss": 0.0228, "step": 9700 }, { "grad_norm": 0.15055590867996216, "learning_rate": 9.939497668074609e-05, "loss": 0.0209, "step": 9710 }, { "grad_norm": 0.34549927711486816, "learning_rate": 9.93924095281026e-05, "loss": 0.03, "step": 9720 }, { "grad_norm": 0.1799856722354889, "learning_rate": 9.938983697397948e-05, "loss": 0.0237, "step": 9730 }, { "grad_norm": 0.29888129234313965, "learning_rate": 9.938725901865805e-05, "loss": 0.0269, "step": 9740 }, { "grad_norm": 0.3033446967601776, "learning_rate": 9.93846756624202e-05, "loss": 0.0294, "step": 9750 }, { "grad_norm": 0.2512753903865814, "learning_rate": 9.938208690554849e-05, "loss": 0.0277, "step": 9760 }, { "grad_norm": 0.16037796437740326, "learning_rate": 9.9379492748326e-05, "loss": 0.0217, "step": 9770 }, { "grad_norm": 0.17044121026992798, "learning_rate": 9.937689319103641e-05, "loss": 0.0195, "step": 9780 }, { "grad_norm": 0.3979007601737976, "learning_rate": 9.937428823396404e-05, "loss": 0.0239, "step": 9790 }, { "grad_norm": 0.2115117460489273, "learning_rate": 9.937167787739372e-05, "loss": 0.0267, "step": 9800 }, { "grad_norm": 0.20926614105701447, "learning_rate": 9.936906212161095e-05, "loss": 0.0302, "step": 9810 }, { "grad_norm": 0.1846960186958313, "learning_rate": 9.936644096690176e-05, "loss": 0.0237, "step": 9820 }, { "grad_norm": 0.22095893323421478, "learning_rate": 9.936381441355282e-05, "loss": 0.0258, "step": 9830 }, { "grad_norm": 0.1896486133337021, "learning_rate": 9.936118246185136e-05, "loss": 0.0259, "step": 9840 }, { "grad_norm": 0.30963897705078125, "learning_rate": 9.935854511208518e-05, "loss": 0.0319, "step": 9850 }, { "grad_norm": 0.17003612220287323, "learning_rate": 9.935590236454272e-05, "loss": 0.0251, "step": 9860 }, { "grad_norm": 0.2727234363555908, "learning_rate": 9.935325421951298e-05, "loss": 0.0225, "step": 9870 }, { "grad_norm": 0.22744227945804596, "learning_rate": 9.935060067728557e-05, "loss": 0.0214, "step": 9880 }, { "grad_norm": 0.24697564542293549, "learning_rate": 9.934794173815067e-05, "loss": 0.0256, "step": 9890 }, { "grad_norm": 0.24065090715885162, "learning_rate": 9.934527740239906e-05, "loss": 0.0228, "step": 9900 }, { "grad_norm": 0.301686555147171, "learning_rate": 9.934260767032209e-05, "loss": 0.0278, "step": 9910 }, { "grad_norm": 0.14746582508087158, "learning_rate": 9.933993254221172e-05, "loss": 0.0229, "step": 9920 }, { "grad_norm": 0.3129289746284485, "learning_rate": 9.933725201836053e-05, "loss": 0.0276, "step": 9930 }, { "grad_norm": 0.217313751578331, "learning_rate": 9.933456609906162e-05, "loss": 0.0198, "step": 9940 }, { "grad_norm": 0.24159760773181915, "learning_rate": 9.933187478460875e-05, "loss": 0.0314, "step": 9950 }, { "grad_norm": 0.16635747253894806, "learning_rate": 9.93291780752962e-05, "loss": 0.0238, "step": 9960 }, { "grad_norm": 0.13701990246772766, "learning_rate": 9.932647597141893e-05, "loss": 0.0199, "step": 9970 }, { "grad_norm": 0.1311885118484497, "learning_rate": 9.932376847327239e-05, "loss": 0.0161, "step": 9980 }, { "grad_norm": 0.23208792507648468, "learning_rate": 9.932105558115268e-05, "loss": 0.0267, "step": 9990 }, { "grad_norm": 0.23632365465164185, "learning_rate": 9.931833729535651e-05, "loss": 0.0252, "step": 10000 }, { "grad_norm": 0.16954945027828217, "learning_rate": 9.931561361618111e-05, "loss": 0.021, "step": 10010 }, { "grad_norm": 0.30670812726020813, "learning_rate": 9.931288454392435e-05, "loss": 0.0239, "step": 10020 }, { "grad_norm": 0.2186318188905716, "learning_rate": 9.931015007888467e-05, "loss": 0.023, "step": 10030 }, { "grad_norm": 0.25679463148117065, "learning_rate": 9.930741022136112e-05, "loss": 0.0282, "step": 10040 }, { "grad_norm": 0.20376114547252655, "learning_rate": 9.930466497165333e-05, "loss": 0.0237, "step": 10050 }, { "grad_norm": 0.13938675820827484, "learning_rate": 9.93019143300615e-05, "loss": 0.0197, "step": 10060 }, { "grad_norm": 0.21367493271827698, "learning_rate": 9.929915829688644e-05, "loss": 0.0185, "step": 10070 }, { "grad_norm": 0.24957598745822906, "learning_rate": 9.929639687242955e-05, "loss": 0.0246, "step": 10080 }, { "grad_norm": 0.18087086081504822, "learning_rate": 9.929363005699281e-05, "loss": 0.0174, "step": 10090 }, { "grad_norm": 0.282581090927124, "learning_rate": 9.92908578508788e-05, "loss": 0.0242, "step": 10100 }, { "grad_norm": 0.20136874914169312, "learning_rate": 9.928808025439069e-05, "loss": 0.0183, "step": 10110 }, { "grad_norm": 0.28805726766586304, "learning_rate": 9.928529726783223e-05, "loss": 0.0264, "step": 10120 }, { "grad_norm": 0.19685378670692444, "learning_rate": 9.928250889150774e-05, "loss": 0.0245, "step": 10130 }, { "grad_norm": 0.19360260665416718, "learning_rate": 9.92797151257222e-05, "loss": 0.0193, "step": 10140 }, { "grad_norm": 0.18174919486045837, "learning_rate": 9.927691597078108e-05, "loss": 0.0177, "step": 10150 }, { "grad_norm": 0.2815585136413574, "learning_rate": 9.927411142699053e-05, "loss": 0.025, "step": 10160 }, { "grad_norm": 0.17455808818340302, "learning_rate": 9.927130149465725e-05, "loss": 0.0229, "step": 10170 }, { "grad_norm": 0.1767825484275818, "learning_rate": 9.92684861740885e-05, "loss": 0.0224, "step": 10180 }, { "grad_norm": 0.15226861834526062, "learning_rate": 9.926566546559217e-05, "loss": 0.02, "step": 10190 }, { "grad_norm": 0.19256654381752014, "learning_rate": 9.926283936947673e-05, "loss": 0.0243, "step": 10200 }, { "grad_norm": 0.27299046516418457, "learning_rate": 9.926000788605126e-05, "loss": 0.0278, "step": 10210 }, { "grad_norm": 0.26848238706588745, "learning_rate": 9.92571710156254e-05, "loss": 0.0275, "step": 10220 }, { "grad_norm": 0.2988009750843048, "learning_rate": 9.925432875850936e-05, "loss": 0.0228, "step": 10230 }, { "grad_norm": 0.2112835943698883, "learning_rate": 9.925148111501396e-05, "loss": 0.0208, "step": 10240 }, { "grad_norm": 0.31140053272247314, "learning_rate": 9.924862808545066e-05, "loss": 0.0234, "step": 10250 }, { "grad_norm": 0.18164464831352234, "learning_rate": 9.924576967013141e-05, "loss": 0.0264, "step": 10260 }, { "grad_norm": 0.35367512702941895, "learning_rate": 9.924290586936887e-05, "loss": 0.0331, "step": 10270 }, { "grad_norm": 0.22131861746311188, "learning_rate": 9.924003668347614e-05, "loss": 0.027, "step": 10280 }, { "grad_norm": 0.17470434308052063, "learning_rate": 9.923716211276704e-05, "loss": 0.0194, "step": 10290 }, { "grad_norm": 0.3780806362628937, "learning_rate": 9.923428215755594e-05, "loss": 0.0302, "step": 10300 }, { "grad_norm": 0.19751223921775818, "learning_rate": 9.923139681815775e-05, "loss": 0.0233, "step": 10310 }, { "grad_norm": 0.43795591592788696, "learning_rate": 9.922850609488801e-05, "loss": 0.031, "step": 10320 }, { "grad_norm": 0.25547128915786743, "learning_rate": 9.922560998806287e-05, "loss": 0.0276, "step": 10330 }, { "grad_norm": 0.24996745586395264, "learning_rate": 9.922270849799905e-05, "loss": 0.0206, "step": 10340 }, { "grad_norm": 0.4629405736923218, "learning_rate": 9.92198016250138e-05, "loss": 0.0346, "step": 10350 }, { "grad_norm": 0.16475413739681244, "learning_rate": 9.921688936942506e-05, "loss": 0.0277, "step": 10360 }, { "grad_norm": 0.26223355531692505, "learning_rate": 9.921397173155129e-05, "loss": 0.0249, "step": 10370 }, { "grad_norm": 0.18552537262439728, "learning_rate": 9.921104871171157e-05, "loss": 0.0186, "step": 10380 }, { "grad_norm": 0.19080938398838043, "learning_rate": 9.920812031022554e-05, "loss": 0.0188, "step": 10390 }, { "grad_norm": 0.3046596348285675, "learning_rate": 9.920518652741348e-05, "loss": 0.0246, "step": 10400 }, { "grad_norm": 0.19589488208293915, "learning_rate": 9.920224736359618e-05, "loss": 0.0231, "step": 10410 }, { "grad_norm": 0.3420576751232147, "learning_rate": 9.91993028190951e-05, "loss": 0.0265, "step": 10420 }, { "grad_norm": 0.26516902446746826, "learning_rate": 9.919635289423222e-05, "loss": 0.0266, "step": 10430 }, { "grad_norm": 0.21621377766132355, "learning_rate": 9.919339758933015e-05, "loss": 0.0218, "step": 10440 }, { "grad_norm": 0.17206692695617676, "learning_rate": 9.919043690471209e-05, "loss": 0.0187, "step": 10450 }, { "grad_norm": 0.1296205222606659, "learning_rate": 9.91874708407018e-05, "loss": 0.0174, "step": 10460 }, { "grad_norm": 0.2534283995628357, "learning_rate": 9.918449939762367e-05, "loss": 0.0243, "step": 10470 }, { "grad_norm": 0.17163006961345673, "learning_rate": 9.91815225758026e-05, "loss": 0.0201, "step": 10480 }, { "grad_norm": 0.21640734374523163, "learning_rate": 9.917854037556419e-05, "loss": 0.0241, "step": 10490 }, { "grad_norm": 0.190263032913208, "learning_rate": 9.917555279723454e-05, "loss": 0.0209, "step": 10500 }, { "grad_norm": 0.26910215616226196, "learning_rate": 9.917255984114036e-05, "loss": 0.027, "step": 10510 }, { "grad_norm": 0.16192549467086792, "learning_rate": 9.916956150760896e-05, "loss": 0.022, "step": 10520 }, { "grad_norm": 0.259982705116272, "learning_rate": 9.916655779696826e-05, "loss": 0.0253, "step": 10530 }, { "grad_norm": 0.22477218508720398, "learning_rate": 9.916354870954671e-05, "loss": 0.0232, "step": 10540 }, { "grad_norm": 0.22101439535617828, "learning_rate": 9.91605342456734e-05, "loss": 0.0247, "step": 10550 }, { "grad_norm": 0.17407552897930145, "learning_rate": 9.915751440567795e-05, "loss": 0.0179, "step": 10560 }, { "grad_norm": 0.3280814588069916, "learning_rate": 9.915448918989066e-05, "loss": 0.0236, "step": 10570 }, { "grad_norm": 0.17292560636997223, "learning_rate": 9.915145859864232e-05, "loss": 0.023, "step": 10580 }, { "grad_norm": 0.3940398693084717, "learning_rate": 9.914842263226437e-05, "loss": 0.0279, "step": 10590 }, { "grad_norm": 0.26382577419281006, "learning_rate": 9.914538129108882e-05, "loss": 0.0252, "step": 10600 }, { "grad_norm": 0.13082589209079742, "learning_rate": 9.914233457544825e-05, "loss": 0.0227, "step": 10610 }, { "grad_norm": 0.37695056200027466, "learning_rate": 9.913928248567586e-05, "loss": 0.0267, "step": 10620 }, { "grad_norm": 0.2695515751838684, "learning_rate": 9.913622502210542e-05, "loss": 0.0245, "step": 10630 }, { "grad_norm": 0.41106176376342773, "learning_rate": 9.913316218507128e-05, "loss": 0.0259, "step": 10640 }, { "grad_norm": 0.15460923314094543, "learning_rate": 9.91300939749084e-05, "loss": 0.0227, "step": 10650 }, { "grad_norm": 0.16657763719558716, "learning_rate": 9.91270203919523e-05, "loss": 0.0203, "step": 10660 }, { "grad_norm": 0.10476817190647125, "learning_rate": 9.912394143653912e-05, "loss": 0.0172, "step": 10670 }, { "grad_norm": 0.15788200497627258, "learning_rate": 9.912085710900555e-05, "loss": 0.0152, "step": 10680 }, { "grad_norm": 0.16187036037445068, "learning_rate": 9.911776740968892e-05, "loss": 0.0176, "step": 10690 }, { "grad_norm": 0.1455938071012497, "learning_rate": 9.911467233892709e-05, "loss": 0.0183, "step": 10700 }, { "grad_norm": 0.21402689814567566, "learning_rate": 9.911157189705853e-05, "loss": 0.015, "step": 10710 }, { "grad_norm": 0.3128691017627716, "learning_rate": 9.910846608442229e-05, "loss": 0.0242, "step": 10720 }, { "grad_norm": 0.19834725558757782, "learning_rate": 9.910535490135805e-05, "loss": 0.0237, "step": 10730 }, { "grad_norm": 0.23154431581497192, "learning_rate": 9.910223834820603e-05, "loss": 0.0183, "step": 10740 }, { "grad_norm": 0.14449267089366913, "learning_rate": 9.909911642530703e-05, "loss": 0.0173, "step": 10750 }, { "grad_norm": 0.3948277533054352, "learning_rate": 9.909598913300249e-05, "loss": 0.0238, "step": 10760 }, { "grad_norm": 0.16771219670772552, "learning_rate": 9.909285647163438e-05, "loss": 0.0239, "step": 10770 }, { "grad_norm": 0.29896122217178345, "learning_rate": 9.908971844154531e-05, "loss": 0.0215, "step": 10780 }, { "grad_norm": 0.2556955814361572, "learning_rate": 9.908657504307843e-05, "loss": 0.0249, "step": 10790 }, { "grad_norm": 0.2709319591522217, "learning_rate": 9.908342627657751e-05, "loss": 0.0239, "step": 10800 }, { "grad_norm": 0.22469723224639893, "learning_rate": 9.908027214238689e-05, "loss": 0.0186, "step": 10810 }, { "grad_norm": 0.25235864520072937, "learning_rate": 9.90771126408515e-05, "loss": 0.0244, "step": 10820 }, { "grad_norm": 0.22628188133239746, "learning_rate": 9.907394777231685e-05, "loss": 0.0238, "step": 10830 }, { "grad_norm": 0.32239529490470886, "learning_rate": 9.907077753712905e-05, "loss": 0.0238, "step": 10840 }, { "grad_norm": 0.3064868152141571, "learning_rate": 9.906760193563482e-05, "loss": 0.0191, "step": 10850 }, { "grad_norm": 0.18491189181804657, "learning_rate": 9.906442096818139e-05, "loss": 0.0264, "step": 10860 }, { "grad_norm": 0.4119745194911957, "learning_rate": 9.906123463511665e-05, "loss": 0.0238, "step": 10870 }, { "grad_norm": 0.2596206068992615, "learning_rate": 9.905804293678907e-05, "loss": 0.0226, "step": 10880 }, { "grad_norm": 0.2745978534221649, "learning_rate": 9.905484587354766e-05, "loss": 0.0254, "step": 10890 }, { "grad_norm": 0.22697225213050842, "learning_rate": 9.905164344574205e-05, "loss": 0.022, "step": 10900 }, { "grad_norm": 0.16891148686408997, "learning_rate": 9.904843565372248e-05, "loss": 0.0205, "step": 10910 }, { "grad_norm": 0.156121626496315, "learning_rate": 9.904522249783972e-05, "loss": 0.0209, "step": 10920 }, { "grad_norm": 0.17759869992733002, "learning_rate": 9.904200397844517e-05, "loss": 0.0191, "step": 10930 }, { "grad_norm": 0.16377747058868408, "learning_rate": 9.903878009589078e-05, "loss": 0.0177, "step": 10940 }, { "grad_norm": 0.28300684690475464, "learning_rate": 9.903555085052915e-05, "loss": 0.0171, "step": 10950 }, { "grad_norm": 0.2844303846359253, "learning_rate": 9.903231624271338e-05, "loss": 0.0195, "step": 10960 }, { "grad_norm": 0.28916603326797485, "learning_rate": 9.902907627279724e-05, "loss": 0.0199, "step": 10970 }, { "grad_norm": 0.15047083795070648, "learning_rate": 9.902583094113504e-05, "loss": 0.0212, "step": 10980 }, { "grad_norm": 0.35270729660987854, "learning_rate": 9.902258024808168e-05, "loss": 0.0305, "step": 10990 }, { "grad_norm": 0.24096031486988068, "learning_rate": 9.901932419399264e-05, "loss": 0.0217, "step": 11000 }, { "grad_norm": 0.17802368104457855, "learning_rate": 9.9016062779224e-05, "loss": 0.0179, "step": 11010 }, { "grad_norm": 0.27600884437561035, "learning_rate": 9.901279600413242e-05, "loss": 0.0243, "step": 11020 }, { "grad_norm": 0.14583764970302582, "learning_rate": 9.900952386907518e-05, "loss": 0.0224, "step": 11030 }, { "grad_norm": 0.23841454088687897, "learning_rate": 9.90062463744101e-05, "loss": 0.0212, "step": 11040 }, { "grad_norm": 0.15537093579769135, "learning_rate": 9.900296352049558e-05, "loss": 0.02, "step": 11050 }, { "grad_norm": 0.2734662592411041, "learning_rate": 9.899967530769065e-05, "loss": 0.0234, "step": 11060 }, { "grad_norm": 0.48869985342025757, "learning_rate": 9.899638173635489e-05, "loss": 0.026, "step": 11070 }, { "grad_norm": 0.289638489484787, "learning_rate": 9.899308280684849e-05, "loss": 0.0276, "step": 11080 }, { "grad_norm": 0.268319308757782, "learning_rate": 9.898977851953222e-05, "loss": 0.0208, "step": 11090 }, { "grad_norm": 0.22246891260147095, "learning_rate": 9.898646887476741e-05, "loss": 0.019, "step": 11100 }, { "grad_norm": 0.16558250784873962, "learning_rate": 9.898315387291603e-05, "loss": 0.0172, "step": 11110 }, { "grad_norm": 0.19285239279270172, "learning_rate": 9.89798335143406e-05, "loss": 0.0161, "step": 11120 }, { "grad_norm": 0.5455577969551086, "learning_rate": 9.897650779940419e-05, "loss": 0.0179, "step": 11130 }, { "grad_norm": 0.1956338733434677, "learning_rate": 9.897317672847054e-05, "loss": 0.0264, "step": 11140 }, { "grad_norm": 0.21253708004951477, "learning_rate": 9.89698403019039e-05, "loss": 0.0232, "step": 11150 }, { "grad_norm": 0.16344167292118073, "learning_rate": 9.896649852006917e-05, "loss": 0.02, "step": 11160 }, { "grad_norm": 0.23383909463882446, "learning_rate": 9.896315138333177e-05, "loss": 0.0227, "step": 11170 }, { "grad_norm": 0.2003825604915619, "learning_rate": 9.895979889205774e-05, "loss": 0.0245, "step": 11180 }, { "grad_norm": 0.20967184007167816, "learning_rate": 9.895644104661372e-05, "loss": 0.0194, "step": 11190 }, { "grad_norm": 0.09092602878808975, "learning_rate": 9.895307784736691e-05, "loss": 0.0181, "step": 11200 }, { "grad_norm": 0.15264849364757538, "learning_rate": 9.894970929468512e-05, "loss": 0.021, "step": 11210 }, { "grad_norm": 0.18769873678684235, "learning_rate": 9.89463353889367e-05, "loss": 0.0158, "step": 11220 }, { "grad_norm": 0.2368600070476532, "learning_rate": 9.894295613049065e-05, "loss": 0.0191, "step": 11230 }, { "grad_norm": 0.22476188838481903, "learning_rate": 9.893957151971649e-05, "loss": 0.0192, "step": 11240 }, { "grad_norm": 0.2330506443977356, "learning_rate": 9.893618155698436e-05, "loss": 0.0181, "step": 11250 }, { "grad_norm": 0.3366660475730896, "learning_rate": 9.8932786242665e-05, "loss": 0.0239, "step": 11260 }, { "grad_norm": 0.2319626659154892, "learning_rate": 9.89293855771297e-05, "loss": 0.0267, "step": 11270 }, { "grad_norm": 0.31264063715934753, "learning_rate": 9.892597956075036e-05, "loss": 0.0212, "step": 11280 }, { "grad_norm": 0.17142347991466522, "learning_rate": 9.892256819389947e-05, "loss": 0.0186, "step": 11290 }, { "grad_norm": 0.24255353212356567, "learning_rate": 9.891915147695006e-05, "loss": 0.021, "step": 11300 }, { "grad_norm": 0.17657792568206787, "learning_rate": 9.891572941027577e-05, "loss": 0.0228, "step": 11310 }, { "grad_norm": 0.3447244465351105, "learning_rate": 9.89123019942509e-05, "loss": 0.0221, "step": 11320 }, { "grad_norm": 0.17933517694473267, "learning_rate": 9.89088692292502e-05, "loss": 0.0223, "step": 11330 }, { "grad_norm": 0.2265593707561493, "learning_rate": 9.89054311156491e-05, "loss": 0.0234, "step": 11340 }, { "grad_norm": 0.2058628648519516, "learning_rate": 9.890198765382357e-05, "loss": 0.0211, "step": 11350 }, { "grad_norm": 0.29202234745025635, "learning_rate": 9.889853884415021e-05, "loss": 0.019, "step": 11360 }, { "grad_norm": 0.14658212661743164, "learning_rate": 9.889508468700614e-05, "loss": 0.0189, "step": 11370 }, { "grad_norm": 0.4220714867115021, "learning_rate": 9.889162518276915e-05, "loss": 0.0282, "step": 11380 }, { "grad_norm": 0.2176695168018341, "learning_rate": 9.888816033181752e-05, "loss": 0.0215, "step": 11390 }, { "grad_norm": 0.12252438068389893, "learning_rate": 9.888469013453018e-05, "loss": 0.0177, "step": 11400 }, { "grad_norm": 0.41428831219673157, "learning_rate": 9.888121459128663e-05, "loss": 0.0269, "step": 11410 }, { "grad_norm": 0.16051961481571198, "learning_rate": 9.887773370246693e-05, "loss": 0.0273, "step": 11420 }, { "grad_norm": 0.2535487413406372, "learning_rate": 9.887424746845177e-05, "loss": 0.0264, "step": 11430 }, { "grad_norm": 0.36589187383651733, "learning_rate": 9.887075588962239e-05, "loss": 0.0226, "step": 11440 }, { "grad_norm": 0.2256285548210144, "learning_rate": 9.88672589663606e-05, "loss": 0.0189, "step": 11450 }, { "grad_norm": 0.3887942135334015, "learning_rate": 9.886375669904886e-05, "loss": 0.0199, "step": 11460 }, { "grad_norm": 0.15021274983882904, "learning_rate": 9.886024908807014e-05, "loss": 0.0238, "step": 11470 }, { "grad_norm": 0.3472394645214081, "learning_rate": 9.885673613380806e-05, "loss": 0.0201, "step": 11480 }, { "grad_norm": 0.1840546429157257, "learning_rate": 9.885321783664676e-05, "loss": 0.019, "step": 11490 }, { "grad_norm": 0.13540059328079224, "learning_rate": 9.884969419697101e-05, "loss": 0.0221, "step": 11500 }, { "grad_norm": 0.12406298518180847, "learning_rate": 9.884616521516614e-05, "loss": 0.0177, "step": 11510 }, { "grad_norm": 0.30854344367980957, "learning_rate": 9.88426308916181e-05, "loss": 0.0191, "step": 11520 }, { "grad_norm": 0.18454217910766602, "learning_rate": 9.883909122671335e-05, "loss": 0.0233, "step": 11530 }, { "grad_norm": 0.4142153859138489, "learning_rate": 9.883554622083904e-05, "loss": 0.0235, "step": 11540 }, { "grad_norm": 0.21038128435611725, "learning_rate": 9.88319958743828e-05, "loss": 0.0197, "step": 11550 }, { "grad_norm": 0.2752332091331482, "learning_rate": 9.882844018773291e-05, "loss": 0.0226, "step": 11560 }, { "grad_norm": 0.308080792427063, "learning_rate": 9.882487916127823e-05, "loss": 0.0313, "step": 11570 }, { "grad_norm": 0.1650962233543396, "learning_rate": 9.882131279540815e-05, "loss": 0.0203, "step": 11580 }, { "grad_norm": 0.18311840295791626, "learning_rate": 9.881774109051271e-05, "loss": 0.0175, "step": 11590 }, { "grad_norm": 0.20585188269615173, "learning_rate": 9.881416404698252e-05, "loss": 0.0189, "step": 11600 }, { "grad_norm": 0.22784966230392456, "learning_rate": 9.881058166520873e-05, "loss": 0.0195, "step": 11610 }, { "grad_norm": 0.14720560610294342, "learning_rate": 9.880699394558311e-05, "loss": 0.0194, "step": 11620 }, { "grad_norm": 0.187533438205719, "learning_rate": 9.880340088849801e-05, "loss": 0.0148, "step": 11630 }, { "grad_norm": 0.12645860016345978, "learning_rate": 9.879980249434637e-05, "loss": 0.0153, "step": 11640 }, { "grad_norm": 0.1471610814332962, "learning_rate": 9.879619876352168e-05, "loss": 0.0161, "step": 11650 }, { "grad_norm": 0.1973087340593338, "learning_rate": 9.879258969641809e-05, "loss": 0.0211, "step": 11660 }, { "grad_norm": 0.15149103105068207, "learning_rate": 9.878897529343023e-05, "loss": 0.0152, "step": 11670 }, { "grad_norm": 0.13621285557746887, "learning_rate": 9.878535555495338e-05, "loss": 0.0148, "step": 11680 }, { "grad_norm": 0.18959759175777435, "learning_rate": 9.87817304813834e-05, "loss": 0.0182, "step": 11690 }, { "grad_norm": 0.2640373408794403, "learning_rate": 9.877810007311671e-05, "loss": 0.0237, "step": 11700 }, { "grad_norm": 0.2317643165588379, "learning_rate": 9.877446433055035e-05, "loss": 0.0197, "step": 11710 }, { "grad_norm": 0.21986468136310577, "learning_rate": 9.877082325408191e-05, "loss": 0.0191, "step": 11720 }, { "grad_norm": 0.11770494282245636, "learning_rate": 9.876717684410954e-05, "loss": 0.0131, "step": 11730 }, { "grad_norm": 0.15293025970458984, "learning_rate": 9.876352510103204e-05, "loss": 0.0173, "step": 11740 }, { "grad_norm": 0.2194814682006836, "learning_rate": 9.875986802524875e-05, "loss": 0.0208, "step": 11750 }, { "grad_norm": 0.2786935567855835, "learning_rate": 9.87562056171596e-05, "loss": 0.0239, "step": 11760 }, { "grad_norm": 0.2854856252670288, "learning_rate": 9.875253787716511e-05, "loss": 0.0237, "step": 11770 }, { "grad_norm": 0.2268439680337906, "learning_rate": 9.874886480566637e-05, "loss": 0.0225, "step": 11780 }, { "grad_norm": 0.2999127507209778, "learning_rate": 9.874518640306507e-05, "loss": 0.0174, "step": 11790 }, { "grad_norm": 0.3126203417778015, "learning_rate": 9.874150266976347e-05, "loss": 0.0261, "step": 11800 }, { "grad_norm": 0.1200733333826065, "learning_rate": 9.873781360616443e-05, "loss": 0.0176, "step": 11810 }, { "grad_norm": 0.5186024904251099, "learning_rate": 9.873411921267137e-05, "loss": 0.0237, "step": 11820 }, { "grad_norm": 0.2743482291698456, "learning_rate": 9.873041948968829e-05, "loss": 0.0257, "step": 11830 }, { "grad_norm": 0.3909783363342285, "learning_rate": 9.872671443761981e-05, "loss": 0.0221, "step": 11840 }, { "grad_norm": 0.19309571385383606, "learning_rate": 9.872300405687109e-05, "loss": 0.0233, "step": 11850 }, { "grad_norm": 0.1709476113319397, "learning_rate": 9.871928834784792e-05, "loss": 0.0217, "step": 11860 }, { "grad_norm": 0.7815452218055725, "learning_rate": 9.871556731095661e-05, "loss": 0.0263, "step": 11870 }, { "grad_norm": 0.2171861082315445, "learning_rate": 9.871184094660411e-05, "loss": 0.0301, "step": 11880 }, { "grad_norm": 0.3249969780445099, "learning_rate": 9.870810925519791e-05, "loss": 0.0269, "step": 11890 }, { "grad_norm": 0.2060440331697464, "learning_rate": 9.870437223714612e-05, "loss": 0.0293, "step": 11900 }, { "grad_norm": 0.18254584074020386, "learning_rate": 9.87006298928574e-05, "loss": 0.0179, "step": 11910 }, { "grad_norm": 0.18321476876735687, "learning_rate": 9.869688222274103e-05, "loss": 0.017, "step": 11920 }, { "grad_norm": 0.24302147328853607, "learning_rate": 9.869312922720681e-05, "loss": 0.0254, "step": 11930 }, { "grad_norm": 0.4965240955352783, "learning_rate": 9.868937090666521e-05, "loss": 0.0235, "step": 11940 }, { "grad_norm": 0.23964571952819824, "learning_rate": 9.86856072615272e-05, "loss": 0.0275, "step": 11950 }, { "grad_norm": 0.1267717182636261, "learning_rate": 9.868183829220438e-05, "loss": 0.0227, "step": 11960 }, { "grad_norm": 0.3866063356399536, "learning_rate": 9.867806399910893e-05, "loss": 0.0207, "step": 11970 }, { "grad_norm": 0.23904570937156677, "learning_rate": 9.867428438265356e-05, "loss": 0.028, "step": 11980 }, { "grad_norm": 0.13797593116760254, "learning_rate": 9.867049944325165e-05, "loss": 0.0203, "step": 11990 }, { "grad_norm": 0.2568812668323517, "learning_rate": 9.86667091813171e-05, "loss": 0.0193, "step": 12000 }, { "grad_norm": 0.2004375457763672, "learning_rate": 9.866291359726438e-05, "loss": 0.0256, "step": 12010 }, { "grad_norm": 0.12685692310333252, "learning_rate": 9.865911269150861e-05, "loss": 0.0219, "step": 12020 }, { "grad_norm": 0.15245625376701355, "learning_rate": 9.865530646446544e-05, "loss": 0.0189, "step": 12030 }, { "grad_norm": 0.3987889587879181, "learning_rate": 9.86514949165511e-05, "loss": 0.0224, "step": 12040 }, { "grad_norm": 0.23416265845298767, "learning_rate": 9.864767804818243e-05, "loss": 0.0212, "step": 12050 }, { "grad_norm": 0.6511924862861633, "learning_rate": 9.86438558597768e-05, "loss": 0.0209, "step": 12060 }, { "grad_norm": 0.2693503797054291, "learning_rate": 9.864002835175225e-05, "loss": 0.0219, "step": 12070 }, { "grad_norm": 0.21751447021961212, "learning_rate": 9.863619552452734e-05, "loss": 0.0187, "step": 12080 }, { "grad_norm": 0.35556524991989136, "learning_rate": 9.863235737852119e-05, "loss": 0.0153, "step": 12090 }, { "grad_norm": 0.21447981894016266, "learning_rate": 9.862851391415356e-05, "loss": 0.0205, "step": 12100 }, { "grad_norm": 0.14211124181747437, "learning_rate": 9.862466513184477e-05, "loss": 0.0153, "step": 12110 }, { "grad_norm": 0.17909149825572968, "learning_rate": 9.86208110320157e-05, "loss": 0.0178, "step": 12120 }, { "grad_norm": 0.26404789090156555, "learning_rate": 9.861695161508784e-05, "loss": 0.0248, "step": 12130 }, { "grad_norm": 0.1896193027496338, "learning_rate": 9.861308688148324e-05, "loss": 0.0173, "step": 12140 }, { "grad_norm": 0.11653625220060349, "learning_rate": 9.860921683162455e-05, "loss": 0.0167, "step": 12150 }, { "grad_norm": 0.14312070608139038, "learning_rate": 9.860534146593499e-05, "loss": 0.0197, "step": 12160 }, { "grad_norm": 0.26735085248947144, "learning_rate": 9.860146078483836e-05, "loss": 0.0254, "step": 12170 }, { "grad_norm": 0.14577631652355194, "learning_rate": 9.859757478875905e-05, "loss": 0.0167, "step": 12180 }, { "grad_norm": 0.14725704491138458, "learning_rate": 9.859368347812204e-05, "loss": 0.0154, "step": 12190 }, { "grad_norm": 0.17509740591049194, "learning_rate": 9.858978685335285e-05, "loss": 0.02, "step": 12200 }, { "grad_norm": 0.1609705537557602, "learning_rate": 9.858588491487763e-05, "loss": 0.021, "step": 12210 }, { "grad_norm": 0.17573346197605133, "learning_rate": 9.858197766312308e-05, "loss": 0.0188, "step": 12220 }, { "grad_norm": 0.16745832562446594, "learning_rate": 9.857806509851649e-05, "loss": 0.0156, "step": 12230 }, { "grad_norm": 0.1698308289051056, "learning_rate": 9.857414722148574e-05, "loss": 0.0145, "step": 12240 }, { "grad_norm": 0.15430398285388947, "learning_rate": 9.857022403245928e-05, "loss": 0.0183, "step": 12250 }, { "grad_norm": 0.17080561816692352, "learning_rate": 9.856629553186615e-05, "loss": 0.0165, "step": 12260 }, { "grad_norm": 0.1816219985485077, "learning_rate": 9.856236172013595e-05, "loss": 0.0192, "step": 12270 }, { "grad_norm": 0.14434964954853058, "learning_rate": 9.85584225976989e-05, "loss": 0.0148, "step": 12280 }, { "grad_norm": 0.12864209711551666, "learning_rate": 9.855447816498575e-05, "loss": 0.0137, "step": 12290 }, { "grad_norm": 0.22310464084148407, "learning_rate": 9.855052842242787e-05, "loss": 0.0194, "step": 12300 }, { "grad_norm": 0.24315392971038818, "learning_rate": 9.85465733704572e-05, "loss": 0.0217, "step": 12310 }, { "grad_norm": 0.20104894042015076, "learning_rate": 9.854261300950624e-05, "loss": 0.0169, "step": 12320 }, { "grad_norm": 0.21688583493232727, "learning_rate": 9.853864734000813e-05, "loss": 0.0197, "step": 12330 }, { "grad_norm": 0.23101864755153656, "learning_rate": 9.85346763623965e-05, "loss": 0.0238, "step": 12340 }, { "grad_norm": 0.14814333617687225, "learning_rate": 9.853070007710564e-05, "loss": 0.0169, "step": 12350 }, { "grad_norm": 0.37691763043403625, "learning_rate": 9.85267184845704e-05, "loss": 0.0263, "step": 12360 }, { "grad_norm": 0.2501724660396576, "learning_rate": 9.852273158522616e-05, "loss": 0.0225, "step": 12370 }, { "grad_norm": 0.14140614867210388, "learning_rate": 9.851873937950896e-05, "loss": 0.019, "step": 12380 }, { "grad_norm": 0.14020660519599915, "learning_rate": 9.851474186785537e-05, "loss": 0.0166, "step": 12390 }, { "grad_norm": 0.2526994049549103, "learning_rate": 9.851073905070254e-05, "loss": 0.0225, "step": 12400 }, { "grad_norm": 0.265520840883255, "learning_rate": 9.850673092848824e-05, "loss": 0.0229, "step": 12410 }, { "grad_norm": 0.1575305312871933, "learning_rate": 9.850271750165077e-05, "loss": 0.0209, "step": 12420 }, { "grad_norm": 0.17628523707389832, "learning_rate": 9.849869877062902e-05, "loss": 0.0206, "step": 12430 }, { "grad_norm": 0.22219698131084442, "learning_rate": 9.849467473586252e-05, "loss": 0.0239, "step": 12440 }, { "grad_norm": 0.22706902027130127, "learning_rate": 9.849064539779127e-05, "loss": 0.0296, "step": 12450 }, { "grad_norm": 0.19717463850975037, "learning_rate": 9.848661075685594e-05, "loss": 0.0188, "step": 12460 }, { "grad_norm": 0.1709710955619812, "learning_rate": 9.848257081349778e-05, "loss": 0.0204, "step": 12470 }, { "grad_norm": 0.22042076289653778, "learning_rate": 9.847852556815856e-05, "loss": 0.0182, "step": 12480 }, { "grad_norm": 0.2500911355018616, "learning_rate": 9.847447502128067e-05, "loss": 0.0208, "step": 12490 }, { "grad_norm": 0.16689568758010864, "learning_rate": 9.847041917330708e-05, "loss": 0.0159, "step": 12500 }, { "grad_norm": 0.3696238696575165, "learning_rate": 9.846635802468132e-05, "loss": 0.0182, "step": 12510 }, { "grad_norm": 0.240104541182518, "learning_rate": 9.84622915758475e-05, "loss": 0.022, "step": 12520 }, { "grad_norm": 0.31369808316230774, "learning_rate": 9.845821982725034e-05, "loss": 0.0249, "step": 12530 }, { "grad_norm": 0.25200751423835754, "learning_rate": 9.845414277933514e-05, "loss": 0.0227, "step": 12540 }, { "grad_norm": 0.18982799351215363, "learning_rate": 9.845006043254771e-05, "loss": 0.0181, "step": 12550 }, { "grad_norm": 0.48194169998168945, "learning_rate": 9.844597278733451e-05, "loss": 0.0194, "step": 12560 }, { "grad_norm": 0.22392135858535767, "learning_rate": 9.844187984414259e-05, "loss": 0.0231, "step": 12570 }, { "grad_norm": 0.1905815750360489, "learning_rate": 9.84377816034195e-05, "loss": 0.0187, "step": 12580 }, { "grad_norm": 0.17828430235385895, "learning_rate": 9.843367806561345e-05, "loss": 0.0167, "step": 12590 }, { "grad_norm": 0.5101708173751831, "learning_rate": 9.842956923117317e-05, "loss": 0.0182, "step": 12600 }, { "grad_norm": 0.21861013770103455, "learning_rate": 9.842545510054802e-05, "loss": 0.0222, "step": 12610 }, { "grad_norm": 0.3056207597255707, "learning_rate": 9.842133567418792e-05, "loss": 0.0246, "step": 12620 }, { "grad_norm": 0.20850209891796112, "learning_rate": 9.841721095254333e-05, "loss": 0.0203, "step": 12630 }, { "grad_norm": 0.3012924790382385, "learning_rate": 9.841308093606537e-05, "loss": 0.0228, "step": 12640 }, { "grad_norm": 0.1865769326686859, "learning_rate": 9.840894562520565e-05, "loss": 0.0195, "step": 12650 }, { "grad_norm": 0.40056607127189636, "learning_rate": 9.840480502041642e-05, "loss": 0.0195, "step": 12660 }, { "grad_norm": 0.17068056762218475, "learning_rate": 9.840065912215049e-05, "loss": 0.0188, "step": 12670 }, { "grad_norm": 0.21025694906711578, "learning_rate": 9.839650793086124e-05, "loss": 0.0176, "step": 12680 }, { "grad_norm": 0.5101359486579895, "learning_rate": 9.839235144700265e-05, "loss": 0.0246, "step": 12690 }, { "grad_norm": 0.2813558578491211, "learning_rate": 9.838818967102926e-05, "loss": 0.0202, "step": 12700 }, { "grad_norm": 0.21767106652259827, "learning_rate": 9.83840226033962e-05, "loss": 0.0204, "step": 12710 }, { "grad_norm": 0.1990998387336731, "learning_rate": 9.837985024455918e-05, "loss": 0.0193, "step": 12720 }, { "grad_norm": 0.13874460756778717, "learning_rate": 9.837567259497447e-05, "loss": 0.0155, "step": 12730 }, { "grad_norm": 0.17480190098285675, "learning_rate": 9.837148965509894e-05, "loss": 0.0175, "step": 12740 }, { "grad_norm": 0.20270809531211853, "learning_rate": 9.836730142539001e-05, "loss": 0.0183, "step": 12750 }, { "grad_norm": 0.15931516885757446, "learning_rate": 9.836310790630574e-05, "loss": 0.0183, "step": 12760 }, { "grad_norm": 0.1919398158788681, "learning_rate": 9.83589090983047e-05, "loss": 0.0186, "step": 12770 }, { "grad_norm": 0.1355653703212738, "learning_rate": 9.835470500184605e-05, "loss": 0.0177, "step": 12780 }, { "grad_norm": 0.1721031665802002, "learning_rate": 9.835049561738957e-05, "loss": 0.014, "step": 12790 }, { "grad_norm": 0.13553310930728912, "learning_rate": 9.834628094539558e-05, "loss": 0.0138, "step": 12800 }, { "grad_norm": 0.16946004331111908, "learning_rate": 9.834206098632499e-05, "loss": 0.0116, "step": 12810 }, { "grad_norm": 0.22395798563957214, "learning_rate": 9.833783574063931e-05, "loss": 0.0157, "step": 12820 }, { "grad_norm": 0.14744028449058533, "learning_rate": 9.833360520880058e-05, "loss": 0.016, "step": 12830 }, { "grad_norm": 0.19244612753391266, "learning_rate": 9.832936939127144e-05, "loss": 0.0198, "step": 12840 }, { "grad_norm": 0.16537708044052124, "learning_rate": 9.832512828851515e-05, "loss": 0.0167, "step": 12850 }, { "grad_norm": 0.24077382683753967, "learning_rate": 9.832088190099546e-05, "loss": 0.0227, "step": 12860 }, { "grad_norm": 0.1557440608739853, "learning_rate": 9.831663022917679e-05, "loss": 0.0167, "step": 12870 }, { "grad_norm": 0.2340036779642105, "learning_rate": 9.831237327352407e-05, "loss": 0.0183, "step": 12880 }, { "grad_norm": 0.19514399766921997, "learning_rate": 9.830811103450286e-05, "loss": 0.0178, "step": 12890 }, { "grad_norm": 0.13195903599262238, "learning_rate": 9.830384351257924e-05, "loss": 0.0155, "step": 12900 }, { "grad_norm": 0.31946367025375366, "learning_rate": 9.829957070821993e-05, "loss": 0.0185, "step": 12910 }, { "grad_norm": 0.1912778913974762, "learning_rate": 9.829529262189218e-05, "loss": 0.0181, "step": 12920 }, { "grad_norm": 0.27803829312324524, "learning_rate": 9.829100925406385e-05, "loss": 0.0203, "step": 12930 }, { "grad_norm": 0.15369756519794464, "learning_rate": 9.828672060520333e-05, "loss": 0.0197, "step": 12940 }, { "grad_norm": 0.12373926490545273, "learning_rate": 9.828242667577966e-05, "loss": 0.0148, "step": 12950 }, { "grad_norm": 0.11108501255512238, "learning_rate": 9.82781274662624e-05, "loss": 0.0156, "step": 12960 }, { "grad_norm": 0.25429391860961914, "learning_rate": 9.82738229771217e-05, "loss": 0.0191, "step": 12970 }, { "grad_norm": 0.17335496842861176, "learning_rate": 9.826951320882829e-05, "loss": 0.0187, "step": 12980 }, { "grad_norm": 0.148893803358078, "learning_rate": 9.826519816185351e-05, "loss": 0.0236, "step": 12990 }, { "grad_norm": 0.13811378180980682, "learning_rate": 9.826087783666921e-05, "loss": 0.014, "step": 13000 }, { "grad_norm": 0.25877270102500916, "learning_rate": 9.825655223374787e-05, "loss": 0.0194, "step": 13010 }, { "grad_norm": 0.40873855352401733, "learning_rate": 9.825222135356253e-05, "loss": 0.0179, "step": 13020 }, { "grad_norm": 0.2152969092130661, "learning_rate": 9.82478851965868e-05, "loss": 0.0188, "step": 13030 }, { "grad_norm": 0.4048287570476532, "learning_rate": 9.82435437632949e-05, "loss": 0.0256, "step": 13040 }, { "grad_norm": 0.2986556589603424, "learning_rate": 9.823919705416158e-05, "loss": 0.0221, "step": 13050 }, { "grad_norm": 0.19612964987754822, "learning_rate": 9.82348450696622e-05, "loss": 0.0159, "step": 13060 }, { "grad_norm": 0.4373127222061157, "learning_rate": 9.823048781027268e-05, "loss": 0.0148, "step": 13070 }, { "grad_norm": 0.2319761961698532, "learning_rate": 9.822612527646953e-05, "loss": 0.0178, "step": 13080 }, { "grad_norm": 0.3806299865245819, "learning_rate": 9.822175746872984e-05, "loss": 0.0204, "step": 13090 }, { "grad_norm": 0.1755758374929428, "learning_rate": 9.821738438753123e-05, "loss": 0.0157, "step": 13100 }, { "grad_norm": 0.18864615261554718, "learning_rate": 9.821300603335196e-05, "loss": 0.0175, "step": 13110 }, { "grad_norm": 0.15662580728530884, "learning_rate": 9.820862240667085e-05, "loss": 0.0149, "step": 13120 }, { "grad_norm": 0.4327693581581116, "learning_rate": 9.820423350796726e-05, "loss": 0.0217, "step": 13130 }, { "grad_norm": 0.2045464962720871, "learning_rate": 9.819983933772118e-05, "loss": 0.0224, "step": 13140 }, { "grad_norm": 0.2171068787574768, "learning_rate": 9.819543989641314e-05, "loss": 0.0194, "step": 13150 }, { "grad_norm": 0.33618947863578796, "learning_rate": 9.819103518452423e-05, "loss": 0.02, "step": 13160 }, { "grad_norm": 0.2014787495136261, "learning_rate": 9.818662520253618e-05, "loss": 0.0223, "step": 13170 }, { "grad_norm": 0.34798115491867065, "learning_rate": 9.818220995093126e-05, "loss": 0.0197, "step": 13180 }, { "grad_norm": 0.19461970031261444, "learning_rate": 9.817778943019228e-05, "loss": 0.0202, "step": 13190 }, { "grad_norm": 0.2331589013338089, "learning_rate": 9.81733636408027e-05, "loss": 0.0209, "step": 13200 }, { "grad_norm": 0.2324354350566864, "learning_rate": 9.816893258324649e-05, "loss": 0.0187, "step": 13210 }, { "grad_norm": 0.17549940943717957, "learning_rate": 9.816449625800823e-05, "loss": 0.0227, "step": 13220 }, { "grad_norm": 0.13915510475635529, "learning_rate": 9.816005466557308e-05, "loss": 0.0191, "step": 13230 }, { "grad_norm": 0.20350901782512665, "learning_rate": 9.815560780642674e-05, "loss": 0.0228, "step": 13240 }, { "grad_norm": 0.20757360756397247, "learning_rate": 9.815115568105555e-05, "loss": 0.0188, "step": 13250 }, { "grad_norm": 0.1913975179195404, "learning_rate": 9.814669828994638e-05, "loss": 0.0178, "step": 13260 }, { "grad_norm": 0.2874213755130768, "learning_rate": 9.814223563358665e-05, "loss": 0.0172, "step": 13270 }, { "grad_norm": 0.24485161900520325, "learning_rate": 9.813776771246443e-05, "loss": 0.0186, "step": 13280 }, { "grad_norm": 0.2238905280828476, "learning_rate": 9.813329452706829e-05, "loss": 0.0244, "step": 13290 }, { "grad_norm": 0.23133045434951782, "learning_rate": 9.812881607788744e-05, "loss": 0.0225, "step": 13300 }, { "grad_norm": 0.16744892299175262, "learning_rate": 9.812433236541163e-05, "loss": 0.0162, "step": 13310 }, { "grad_norm": 0.2613702416419983, "learning_rate": 9.811984339013116e-05, "loss": 0.0245, "step": 13320 }, { "grad_norm": 0.1029210314154625, "learning_rate": 9.811534915253698e-05, "loss": 0.0195, "step": 13330 }, { "grad_norm": 0.1874517798423767, "learning_rate": 9.811084965312056e-05, "loss": 0.0221, "step": 13340 }, { "grad_norm": 0.1617751568555832, "learning_rate": 9.810634489237396e-05, "loss": 0.0156, "step": 13350 }, { "grad_norm": 0.16004310548305511, "learning_rate": 9.81018348707898e-05, "loss": 0.0169, "step": 13360 }, { "grad_norm": 0.16060398519039154, "learning_rate": 9.809731958886131e-05, "loss": 0.0147, "step": 13370 }, { "grad_norm": 0.21584950387477875, "learning_rate": 9.809279904708224e-05, "loss": 0.0196, "step": 13380 }, { "grad_norm": 0.13350053131580353, "learning_rate": 9.808827324594699e-05, "loss": 0.0142, "step": 13390 }, { "grad_norm": 0.19691893458366394, "learning_rate": 9.808374218595046e-05, "loss": 0.0209, "step": 13400 }, { "grad_norm": 0.1336515098810196, "learning_rate": 9.80792058675882e-05, "loss": 0.0142, "step": 13410 }, { "grad_norm": 0.17323370277881622, "learning_rate": 9.807466429135627e-05, "loss": 0.0167, "step": 13420 }, { "grad_norm": 0.12721185386180878, "learning_rate": 9.807011745775132e-05, "loss": 0.0175, "step": 13430 }, { "grad_norm": 0.18406924605369568, "learning_rate": 9.806556536727061e-05, "loss": 0.0223, "step": 13440 }, { "grad_norm": 0.14395345747470856, "learning_rate": 9.806100802041193e-05, "loss": 0.016, "step": 13450 }, { "grad_norm": 0.22365882992744446, "learning_rate": 9.805644541767368e-05, "loss": 0.02, "step": 13460 }, { "grad_norm": 0.1370268017053604, "learning_rate": 9.805187755955478e-05, "loss": 0.0203, "step": 13470 }, { "grad_norm": 0.15814457833766937, "learning_rate": 9.804730444655483e-05, "loss": 0.0149, "step": 13480 }, { "grad_norm": 0.16730482876300812, "learning_rate": 9.804272607917388e-05, "loss": 0.0139, "step": 13490 }, { "grad_norm": 0.16021688282489777, "learning_rate": 9.803814245791265e-05, "loss": 0.0158, "step": 13500 }, { "grad_norm": 0.2651009261608124, "learning_rate": 9.803355358327239e-05, "loss": 0.02, "step": 13510 }, { "grad_norm": 0.17920628190040588, "learning_rate": 9.802895945575492e-05, "loss": 0.0177, "step": 13520 }, { "grad_norm": 0.37003061175346375, "learning_rate": 9.802436007586266e-05, "loss": 0.0208, "step": 13530 }, { "grad_norm": 0.2662172019481659, "learning_rate": 9.801975544409858e-05, "loss": 0.0198, "step": 13540 }, { "grad_norm": 0.19606225192546844, "learning_rate": 9.801514556096625e-05, "loss": 0.0154, "step": 13550 }, { "grad_norm": 0.22975482046604156, "learning_rate": 9.801053042696977e-05, "loss": 0.025, "step": 13560 }, { "grad_norm": 0.26019322872161865, "learning_rate": 9.800591004261388e-05, "loss": 0.0229, "step": 13570 }, { "grad_norm": 0.26303040981292725, "learning_rate": 9.800128440840385e-05, "loss": 0.0205, "step": 13580 }, { "grad_norm": 0.24436511099338531, "learning_rate": 9.799665352484552e-05, "loss": 0.0194, "step": 13590 }, { "grad_norm": 0.187499538064003, "learning_rate": 9.799201739244532e-05, "loss": 0.0147, "step": 13600 }, { "grad_norm": 0.25914496183395386, "learning_rate": 9.798737601171025e-05, "loss": 0.0191, "step": 13610 }, { "grad_norm": 0.20003783702850342, "learning_rate": 9.79827293831479e-05, "loss": 0.0182, "step": 13620 }, { "grad_norm": 0.1476556360721588, "learning_rate": 9.797807750726638e-05, "loss": 0.0164, "step": 13630 }, { "grad_norm": 0.09263674914836884, "learning_rate": 9.797342038457446e-05, "loss": 0.0165, "step": 13640 }, { "grad_norm": 0.22311261296272278, "learning_rate": 9.796875801558141e-05, "loss": 0.019, "step": 13650 }, { "grad_norm": 0.15730084478855133, "learning_rate": 9.79640904007971e-05, "loss": 0.0169, "step": 13660 }, { "grad_norm": 0.143607497215271, "learning_rate": 9.795941754073199e-05, "loss": 0.0155, "step": 13670 }, { "grad_norm": 0.19232995808124542, "learning_rate": 9.795473943589705e-05, "loss": 0.0149, "step": 13680 }, { "grad_norm": 0.14887353777885437, "learning_rate": 9.795005608680394e-05, "loss": 0.0142, "step": 13690 }, { "grad_norm": 0.11241158843040466, "learning_rate": 9.794536749396477e-05, "loss": 0.0128, "step": 13700 }, { "grad_norm": 0.12405125051736832, "learning_rate": 9.79406736578923e-05, "loss": 0.0153, "step": 13710 }, { "grad_norm": 0.200184166431427, "learning_rate": 9.793597457909984e-05, "loss": 0.0195, "step": 13720 }, { "grad_norm": 0.20288127660751343, "learning_rate": 9.793127025810127e-05, "loss": 0.0176, "step": 13730 }, { "grad_norm": 0.1982911080121994, "learning_rate": 9.792656069541104e-05, "loss": 0.0182, "step": 13740 }, { "grad_norm": 0.18594075739383698, "learning_rate": 9.79218458915442e-05, "loss": 0.0191, "step": 13750 }, { "grad_norm": 0.1444884091615677, "learning_rate": 9.791712584701634e-05, "loss": 0.0156, "step": 13760 }, { "grad_norm": 0.145905539393425, "learning_rate": 9.791240056234364e-05, "loss": 0.0166, "step": 13770 }, { "grad_norm": 0.1362251490354538, "learning_rate": 9.790767003804283e-05, "loss": 0.0159, "step": 13780 }, { "grad_norm": 0.39135533571243286, "learning_rate": 9.790293427463126e-05, "loss": 0.0207, "step": 13790 }, { "grad_norm": 0.4826035797595978, "learning_rate": 9.789819327262684e-05, "loss": 0.0209, "step": 13800 }, { "grad_norm": 0.19848951697349548, "learning_rate": 9.7893447032548e-05, "loss": 0.0193, "step": 13810 }, { "grad_norm": 0.5567179918289185, "learning_rate": 9.78886955549138e-05, "loss": 0.0191, "step": 13820 }, { "grad_norm": 0.2372453361749649, "learning_rate": 9.788393884024387e-05, "loss": 0.0197, "step": 13830 }, { "grad_norm": 0.2922316789627075, "learning_rate": 9.787917688905836e-05, "loss": 0.0204, "step": 13840 }, { "grad_norm": 0.21675175428390503, "learning_rate": 9.787440970187807e-05, "loss": 0.0221, "step": 13850 }, { "grad_norm": 0.3078192472457886, "learning_rate": 9.786963727922429e-05, "loss": 0.0229, "step": 13860 }, { "grad_norm": 0.18929141759872437, "learning_rate": 9.786485962161897e-05, "loss": 0.0205, "step": 13870 }, { "grad_norm": 0.28823503851890564, "learning_rate": 9.786007672958455e-05, "loss": 0.0204, "step": 13880 }, { "grad_norm": 0.1516217738389969, "learning_rate": 9.78552886036441e-05, "loss": 0.0156, "step": 13890 }, { "grad_norm": 0.25366127490997314, "learning_rate": 9.785049524432124e-05, "loss": 0.0208, "step": 13900 }, { "grad_norm": 0.16604386270046234, "learning_rate": 9.784569665214016e-05, "loss": 0.0158, "step": 13910 }, { "grad_norm": 0.28348803520202637, "learning_rate": 9.784089282762563e-05, "loss": 0.0174, "step": 13920 }, { "grad_norm": 0.21554499864578247, "learning_rate": 9.7836083771303e-05, "loss": 0.0154, "step": 13930 }, { "grad_norm": 0.12564507126808167, "learning_rate": 9.783126948369817e-05, "loss": 0.0157, "step": 13940 }, { "grad_norm": 0.1530977040529251, "learning_rate": 9.78264499653376e-05, "loss": 0.0177, "step": 13950 }, { "grad_norm": 0.19650143384933472, "learning_rate": 9.782162521674838e-05, "loss": 0.0191, "step": 13960 }, { "grad_norm": 0.23249340057373047, "learning_rate": 9.781679523845812e-05, "loss": 0.0166, "step": 13970 }, { "grad_norm": 0.13783515989780426, "learning_rate": 9.781196003099502e-05, "loss": 0.0159, "step": 13980 }, { "grad_norm": 0.29335594177246094, "learning_rate": 9.780711959488786e-05, "loss": 0.0285, "step": 13990 }, { "grad_norm": 0.23881946504116058, "learning_rate": 9.780227393066599e-05, "loss": 0.0219, "step": 14000 }, { "grad_norm": 0.13911427557468414, "learning_rate": 9.77974230388593e-05, "loss": 0.0132, "step": 14010 }, { "grad_norm": 0.22763068974018097, "learning_rate": 9.779256691999829e-05, "loss": 0.0207, "step": 14020 }, { "grad_norm": 0.23921096324920654, "learning_rate": 9.778770557461403e-05, "loss": 0.0184, "step": 14030 }, { "grad_norm": 0.23751884698867798, "learning_rate": 9.778283900323812e-05, "loss": 0.0208, "step": 14040 }, { "grad_norm": 0.17646801471710205, "learning_rate": 9.777796720640277e-05, "loss": 0.021, "step": 14050 }, { "grad_norm": 0.16033099591732025, "learning_rate": 9.777309018464078e-05, "loss": 0.0195, "step": 14060 }, { "grad_norm": 0.1394784301519394, "learning_rate": 9.776820793848547e-05, "loss": 0.017, "step": 14070 }, { "grad_norm": 0.23261575400829315, "learning_rate": 9.776332046847075e-05, "loss": 0.0171, "step": 14080 }, { "grad_norm": 0.23014257848262787, "learning_rate": 9.775842777513111e-05, "loss": 0.0183, "step": 14090 }, { "grad_norm": 0.23611190915107727, "learning_rate": 9.775352985900163e-05, "loss": 0.0175, "step": 14100 }, { "grad_norm": 0.1929537057876587, "learning_rate": 9.774862672061791e-05, "loss": 0.0163, "step": 14110 }, { "grad_norm": 0.19682523608207703, "learning_rate": 9.774371836051616e-05, "loss": 0.0174, "step": 14120 }, { "grad_norm": 0.22737707197666168, "learning_rate": 9.773880477923315e-05, "loss": 0.0187, "step": 14130 }, { "grad_norm": 0.36170053482055664, "learning_rate": 9.773388597730623e-05, "loss": 0.0177, "step": 14140 }, { "grad_norm": 0.23305867612361908, "learning_rate": 9.77289619552733e-05, "loss": 0.0177, "step": 14150 }, { "grad_norm": 0.14869500696659088, "learning_rate": 9.772403271367285e-05, "loss": 0.0169, "step": 14160 }, { "grad_norm": 0.3720097243785858, "learning_rate": 9.771909825304396e-05, "loss": 0.0164, "step": 14170 }, { "grad_norm": 0.1693776249885559, "learning_rate": 9.771415857392619e-05, "loss": 0.0232, "step": 14180 }, { "grad_norm": 0.39612945914268494, "learning_rate": 9.770921367685978e-05, "loss": 0.0187, "step": 14190 }, { "grad_norm": 0.16522416472434998, "learning_rate": 9.770426356238551e-05, "loss": 0.0224, "step": 14200 }, { "grad_norm": 0.354660302400589, "learning_rate": 9.769930823104469e-05, "loss": 0.0145, "step": 14210 }, { "grad_norm": 0.2256804257631302, "learning_rate": 9.769434768337926e-05, "loss": 0.0159, "step": 14220 }, { "grad_norm": 0.1738726645708084, "learning_rate": 9.768938191993164e-05, "loss": 0.014, "step": 14230 }, { "grad_norm": 0.494756817817688, "learning_rate": 9.768441094124494e-05, "loss": 0.0173, "step": 14240 }, { "grad_norm": 0.18551811575889587, "learning_rate": 9.767943474786275e-05, "loss": 0.0185, "step": 14250 }, { "grad_norm": 0.27186834812164307, "learning_rate": 9.767445334032923e-05, "loss": 0.0179, "step": 14260 }, { "grad_norm": 0.13865384459495544, "learning_rate": 9.766946671918919e-05, "loss": 0.0194, "step": 14270 }, { "grad_norm": 0.3038199245929718, "learning_rate": 9.766447488498796e-05, "loss": 0.0219, "step": 14280 }, { "grad_norm": 0.33400169014930725, "learning_rate": 9.765947783827139e-05, "loss": 0.0199, "step": 14290 }, { "grad_norm": 0.15861234068870544, "learning_rate": 9.765447557958599e-05, "loss": 0.0219, "step": 14300 }, { "grad_norm": 0.17520594596862793, "learning_rate": 9.764946810947879e-05, "loss": 0.0158, "step": 14310 }, { "grad_norm": 0.14425750076770782, "learning_rate": 9.764445542849738e-05, "loss": 0.0194, "step": 14320 }, { "grad_norm": 0.19350981712341309, "learning_rate": 9.763943753718998e-05, "loss": 0.0181, "step": 14330 }, { "grad_norm": 0.1254790872335434, "learning_rate": 9.76344144361053e-05, "loss": 0.0141, "step": 14340 }, { "grad_norm": 0.30219486355781555, "learning_rate": 9.762938612579269e-05, "loss": 0.0161, "step": 14350 }, { "grad_norm": 0.15529057383537292, "learning_rate": 9.762435260680202e-05, "loss": 0.0172, "step": 14360 }, { "grad_norm": 0.14664793014526367, "learning_rate": 9.761931387968373e-05, "loss": 0.0129, "step": 14370 }, { "grad_norm": 0.1354958415031433, "learning_rate": 9.76142699449889e-05, "loss": 0.0131, "step": 14380 }, { "grad_norm": 0.20106326043605804, "learning_rate": 9.760922080326908e-05, "loss": 0.0149, "step": 14390 }, { "grad_norm": 0.15053337812423706, "learning_rate": 9.760416645507644e-05, "loss": 0.0137, "step": 14400 }, { "grad_norm": 0.1727176010608673, "learning_rate": 9.759910690096375e-05, "loss": 0.0149, "step": 14410 }, { "grad_norm": 0.1743425875902176, "learning_rate": 9.759404214148429e-05, "loss": 0.0182, "step": 14420 }, { "grad_norm": 0.35541832447052, "learning_rate": 9.758897217719191e-05, "loss": 0.0151, "step": 14430 }, { "grad_norm": 0.17271468043327332, "learning_rate": 9.758389700864113e-05, "loss": 0.0147, "step": 14440 }, { "grad_norm": 0.16371317207813263, "learning_rate": 9.757881663638688e-05, "loss": 0.0144, "step": 14450 }, { "grad_norm": 0.13180924952030182, "learning_rate": 9.757373106098478e-05, "loss": 0.0124, "step": 14460 }, { "grad_norm": 0.13073116540908813, "learning_rate": 9.756864028299097e-05, "loss": 0.0168, "step": 14470 }, { "grad_norm": 0.28718459606170654, "learning_rate": 9.75635443029622e-05, "loss": 0.014, "step": 14480 }, { "grad_norm": 0.13905274868011475, "learning_rate": 9.755844312145572e-05, "loss": 0.0198, "step": 14490 }, { "grad_norm": 0.1416616588830948, "learning_rate": 9.755333673902941e-05, "loss": 0.0155, "step": 14500 }, { "grad_norm": 0.17076000571250916, "learning_rate": 9.75482251562417e-05, "loss": 0.0152, "step": 14510 }, { "grad_norm": 0.14691996574401855, "learning_rate": 9.754310837365155e-05, "loss": 0.0184, "step": 14520 }, { "grad_norm": 0.17489339411258698, "learning_rate": 9.753798639181856e-05, "loss": 0.0157, "step": 14530 }, { "grad_norm": 0.14919620752334595, "learning_rate": 9.753285921130286e-05, "loss": 0.017, "step": 14540 }, { "grad_norm": 0.15350979566574097, "learning_rate": 9.752772683266512e-05, "loss": 0.0129, "step": 14550 }, { "grad_norm": 0.17137667536735535, "learning_rate": 9.752258925646665e-05, "loss": 0.0172, "step": 14560 }, { "grad_norm": 0.12985040247440338, "learning_rate": 9.751744648326926e-05, "loss": 0.0165, "step": 14570 }, { "grad_norm": 0.16722142696380615, "learning_rate": 9.751229851363536e-05, "loss": 0.014, "step": 14580 }, { "grad_norm": 0.1728852391242981, "learning_rate": 9.750714534812793e-05, "loss": 0.0128, "step": 14590 }, { "grad_norm": 0.1768958419561386, "learning_rate": 9.750198698731053e-05, "loss": 0.0145, "step": 14600 }, { "grad_norm": 0.20268894731998444, "learning_rate": 9.749682343174722e-05, "loss": 0.0169, "step": 14610 }, { "grad_norm": 0.20142340660095215, "learning_rate": 9.749165468200272e-05, "loss": 0.0209, "step": 14620 }, { "grad_norm": 0.16215218603610992, "learning_rate": 9.748648073864229e-05, "loss": 0.0186, "step": 14630 }, { "grad_norm": 0.10155163705348969, "learning_rate": 9.748130160223168e-05, "loss": 0.0153, "step": 14640 }, { "grad_norm": 0.1317264288663864, "learning_rate": 9.747611727333734e-05, "loss": 0.0126, "step": 14650 }, { "grad_norm": 0.11463559418916702, "learning_rate": 9.74709277525262e-05, "loss": 0.0182, "step": 14660 }, { "grad_norm": 0.15209253132343292, "learning_rate": 9.746573304036576e-05, "loss": 0.0123, "step": 14670 }, { "grad_norm": 0.14808334410190582, "learning_rate": 9.746053313742412e-05, "loss": 0.0188, "step": 14680 }, { "grad_norm": 0.18247264623641968, "learning_rate": 9.745532804426994e-05, "loss": 0.0161, "step": 14690 }, { "grad_norm": 0.19414860010147095, "learning_rate": 9.745011776147242e-05, "loss": 0.0183, "step": 14700 }, { "grad_norm": 0.1264808177947998, "learning_rate": 9.744490228960138e-05, "loss": 0.0129, "step": 14710 }, { "grad_norm": 0.20167997479438782, "learning_rate": 9.743968162922713e-05, "loss": 0.0212, "step": 14720 }, { "grad_norm": 0.16246578097343445, "learning_rate": 9.743445578092064e-05, "loss": 0.0158, "step": 14730 }, { "grad_norm": 0.15436959266662598, "learning_rate": 9.742922474525338e-05, "loss": 0.0181, "step": 14740 }, { "grad_norm": 0.15180395543575287, "learning_rate": 9.742398852279741e-05, "loss": 0.0123, "step": 14750 }, { "grad_norm": 0.1274610459804535, "learning_rate": 9.741874711412535e-05, "loss": 0.0116, "step": 14760 }, { "grad_norm": 0.08841408789157867, "learning_rate": 9.741350051981042e-05, "loss": 0.0138, "step": 14770 }, { "grad_norm": 0.09333448112010956, "learning_rate": 9.740824874042633e-05, "loss": 0.0126, "step": 14780 }, { "grad_norm": 0.11543945223093033, "learning_rate": 9.740299177654746e-05, "loss": 0.0133, "step": 14790 }, { "grad_norm": 0.34154853224754333, "learning_rate": 9.739772962874867e-05, "loss": 0.021, "step": 14800 }, { "grad_norm": 0.24180381000041962, "learning_rate": 9.739246229760541e-05, "loss": 0.0159, "step": 14810 }, { "grad_norm": 0.14007379114627838, "learning_rate": 9.738718978369376e-05, "loss": 0.0145, "step": 14820 }, { "grad_norm": 0.24556589126586914, "learning_rate": 9.738191208759025e-05, "loss": 0.0183, "step": 14830 }, { "grad_norm": 0.40016138553619385, "learning_rate": 9.73766292098721e-05, "loss": 0.0182, "step": 14840 }, { "grad_norm": 0.2281074970960617, "learning_rate": 9.737134115111699e-05, "loss": 0.0174, "step": 14850 }, { "grad_norm": 0.23838286101818085, "learning_rate": 9.736604791190323e-05, "loss": 0.0148, "step": 14860 }, { "grad_norm": 0.17910952866077423, "learning_rate": 9.73607494928097e-05, "loss": 0.0153, "step": 14870 }, { "grad_norm": 0.4817381799221039, "learning_rate": 9.735544589441581e-05, "loss": 0.0196, "step": 14880 }, { "grad_norm": 0.22167803347110748, "learning_rate": 9.735013711730154e-05, "loss": 0.02, "step": 14890 }, { "grad_norm": 0.1325007528066635, "learning_rate": 9.734482316204747e-05, "loss": 0.0152, "step": 14900 }, { "grad_norm": 0.2428637146949768, "learning_rate": 9.733950402923473e-05, "loss": 0.012, "step": 14910 }, { "grad_norm": 0.12303132563829422, "learning_rate": 9.7334179719445e-05, "loss": 0.0179, "step": 14920 }, { "grad_norm": 0.1292172372341156, "learning_rate": 9.732885023326053e-05, "loss": 0.0166, "step": 14930 }, { "grad_norm": 0.1287481039762497, "learning_rate": 9.732351557126418e-05, "loss": 0.0172, "step": 14940 }, { "grad_norm": 0.13479310274124146, "learning_rate": 9.731817573403929e-05, "loss": 0.012, "step": 14950 }, { "grad_norm": 0.21532514691352844, "learning_rate": 9.731283072216985e-05, "loss": 0.0131, "step": 14960 }, { "grad_norm": 0.18514126539230347, "learning_rate": 9.730748053624039e-05, "loss": 0.0187, "step": 14970 }, { "grad_norm": 0.10736801475286484, "learning_rate": 9.730212517683598e-05, "loss": 0.013, "step": 14980 }, { "grad_norm": 0.24296939373016357, "learning_rate": 9.729676464454228e-05, "loss": 0.0155, "step": 14990 }, { "grad_norm": 0.17098745703697205, "learning_rate": 9.72913989399455e-05, "loss": 0.0191, "step": 15000 }, { "grad_norm": 0.14995472133159637, "learning_rate": 9.728602806363242e-05, "loss": 0.0145, "step": 15010 }, { "grad_norm": 0.1315172016620636, "learning_rate": 9.728065201619043e-05, "loss": 0.0156, "step": 15020 }, { "grad_norm": 0.3212347626686096, "learning_rate": 9.727527079820742e-05, "loss": 0.0145, "step": 15030 }, { "grad_norm": 0.14101167023181915, "learning_rate": 9.726988441027186e-05, "loss": 0.0115, "step": 15040 }, { "grad_norm": 0.13109265267848969, "learning_rate": 9.726449285297281e-05, "loss": 0.0123, "step": 15050 }, { "grad_norm": 0.1326238363981247, "learning_rate": 9.72590961268999e-05, "loss": 0.0117, "step": 15060 }, { "grad_norm": 0.1292320191860199, "learning_rate": 9.725369423264328e-05, "loss": 0.0125, "step": 15070 }, { "grad_norm": 0.10898526012897491, "learning_rate": 9.72482871707937e-05, "loss": 0.0131, "step": 15080 }, { "grad_norm": 0.3107147216796875, "learning_rate": 9.724287494194247e-05, "loss": 0.0163, "step": 15090 }, { "grad_norm": 0.20400185883045197, "learning_rate": 9.723745754668147e-05, "loss": 0.0177, "step": 15100 }, { "grad_norm": 0.24464558064937592, "learning_rate": 9.723203498560313e-05, "loss": 0.0178, "step": 15110 }, { "grad_norm": 0.2231675088405609, "learning_rate": 9.722660725930046e-05, "loss": 0.0178, "step": 15120 }, { "grad_norm": 0.14842510223388672, "learning_rate": 9.722117436836702e-05, "loss": 0.0137, "step": 15130 }, { "grad_norm": 0.2971552014350891, "learning_rate": 9.721573631339696e-05, "loss": 0.0219, "step": 15140 }, { "grad_norm": 0.1954970508813858, "learning_rate": 9.721029309498494e-05, "loss": 0.0213, "step": 15150 }, { "grad_norm": 0.2644376754760742, "learning_rate": 9.720484471372627e-05, "loss": 0.0192, "step": 15160 }, { "grad_norm": 0.2806616425514221, "learning_rate": 9.719939117021673e-05, "loss": 0.0178, "step": 15170 }, { "grad_norm": 0.14479734003543854, "learning_rate": 9.719393246505275e-05, "loss": 0.0141, "step": 15180 }, { "grad_norm": 0.2702208459377289, "learning_rate": 9.718846859883128e-05, "loss": 0.0183, "step": 15190 }, { "grad_norm": 0.21215912699699402, "learning_rate": 9.718299957214982e-05, "loss": 0.017, "step": 15200 }, { "grad_norm": 0.21810393035411835, "learning_rate": 9.717752538560646e-05, "loss": 0.0185, "step": 15210 }, { "grad_norm": 0.2019103467464447, "learning_rate": 9.717204603979986e-05, "loss": 0.0173, "step": 15220 }, { "grad_norm": 0.2952309250831604, "learning_rate": 9.716656153532922e-05, "loss": 0.0156, "step": 15230 }, { "grad_norm": 0.13948798179626465, "learning_rate": 9.716107187279434e-05, "loss": 0.0171, "step": 15240 }, { "grad_norm": 0.1276768147945404, "learning_rate": 9.715557705279555e-05, "loss": 0.013, "step": 15250 }, { "grad_norm": 0.12591958045959473, "learning_rate": 9.715007707593372e-05, "loss": 0.0145, "step": 15260 }, { "grad_norm": 0.18265806138515472, "learning_rate": 9.714457194281036e-05, "loss": 0.0162, "step": 15270 }, { "grad_norm": 0.1916394978761673, "learning_rate": 9.713906165402751e-05, "loss": 0.0173, "step": 15280 }, { "grad_norm": 0.1532810628414154, "learning_rate": 9.713354621018774e-05, "loss": 0.0127, "step": 15290 }, { "grad_norm": 0.1438584327697754, "learning_rate": 9.712802561189422e-05, "loss": 0.0116, "step": 15300 }, { "grad_norm": 0.2742258906364441, "learning_rate": 9.712249985975069e-05, "loss": 0.0141, "step": 15310 }, { "grad_norm": 0.1727655827999115, "learning_rate": 9.71169689543614e-05, "loss": 0.017, "step": 15320 }, { "grad_norm": 0.25549331307411194, "learning_rate": 9.711143289633123e-05, "loss": 0.0159, "step": 15330 }, { "grad_norm": 0.16718198359012604, "learning_rate": 9.710589168626561e-05, "loss": 0.0169, "step": 15340 }, { "grad_norm": 0.3050893545150757, "learning_rate": 9.710034532477048e-05, "loss": 0.0199, "step": 15350 }, { "grad_norm": 0.18599162995815277, "learning_rate": 9.709479381245239e-05, "loss": 0.0164, "step": 15360 }, { "grad_norm": 0.2763175368309021, "learning_rate": 9.708923714991847e-05, "loss": 0.0217, "step": 15370 }, { "grad_norm": 0.13491277396678925, "learning_rate": 9.708367533777638e-05, "loss": 0.0189, "step": 15380 }, { "grad_norm": 0.281909316778183, "learning_rate": 9.707810837663431e-05, "loss": 0.0188, "step": 15390 }, { "grad_norm": 0.1564362794160843, "learning_rate": 9.707253626710113e-05, "loss": 0.0187, "step": 15400 }, { "grad_norm": 0.1275024563074112, "learning_rate": 9.706695900978613e-05, "loss": 0.0142, "step": 15410 }, { "grad_norm": 0.13397689163684845, "learning_rate": 9.706137660529926e-05, "loss": 0.0135, "step": 15420 }, { "grad_norm": 0.0957958772778511, "learning_rate": 9.705578905425101e-05, "loss": 0.0161, "step": 15430 }, { "grad_norm": 0.1164698526263237, "learning_rate": 9.705019635725241e-05, "loss": 0.0148, "step": 15440 }, { "grad_norm": 0.2536300718784332, "learning_rate": 9.704459851491508e-05, "loss": 0.0165, "step": 15450 }, { "grad_norm": 0.11799290776252747, "learning_rate": 9.703899552785118e-05, "loss": 0.0155, "step": 15460 }, { "grad_norm": 0.23926255106925964, "learning_rate": 9.703338739667346e-05, "loss": 0.0227, "step": 15470 }, { "grad_norm": 0.22734133899211884, "learning_rate": 9.70277741219952e-05, "loss": 0.0182, "step": 15480 }, { "grad_norm": 0.13717232644557953, "learning_rate": 9.702215570443027e-05, "loss": 0.0205, "step": 15490 }, { "grad_norm": 0.29753735661506653, "learning_rate": 9.701653214459309e-05, "loss": 0.022, "step": 15500 }, { "grad_norm": 0.16886907815933228, "learning_rate": 9.701090344309865e-05, "loss": 0.0198, "step": 15510 }, { "grad_norm": 0.32490965723991394, "learning_rate": 9.700526960056247e-05, "loss": 0.0205, "step": 15520 }, { "grad_norm": 0.5014676451683044, "learning_rate": 9.699963061760068e-05, "loss": 0.0189, "step": 15530 }, { "grad_norm": 0.2113606184720993, "learning_rate": 9.699398649482997e-05, "loss": 0.0195, "step": 15540 }, { "grad_norm": 0.1734161674976349, "learning_rate": 9.698833723286753e-05, "loss": 0.0178, "step": 15550 }, { "grad_norm": 0.1270541399717331, "learning_rate": 9.698268283233118e-05, "loss": 0.0136, "step": 15560 }, { "grad_norm": 0.14300018548965454, "learning_rate": 9.697702329383929e-05, "loss": 0.0127, "step": 15570 }, { "grad_norm": 0.24019452929496765, "learning_rate": 9.697135861801074e-05, "loss": 0.0189, "step": 15580 }, { "grad_norm": 0.1730760633945465, "learning_rate": 9.696568880546505e-05, "loss": 0.0156, "step": 15590 }, { "grad_norm": 0.13042227923870087, "learning_rate": 9.696001385682223e-05, "loss": 0.014, "step": 15600 }, { "grad_norm": 0.2145000398159027, "learning_rate": 9.695433377270291e-05, "loss": 0.0216, "step": 15610 }, { "grad_norm": 0.14364609122276306, "learning_rate": 9.694864855372824e-05, "loss": 0.0146, "step": 15620 }, { "grad_norm": 0.19127215445041656, "learning_rate": 9.694295820051995e-05, "loss": 0.0189, "step": 15630 }, { "grad_norm": 0.13924597203731537, "learning_rate": 9.693726271370032e-05, "loss": 0.0167, "step": 15640 }, { "grad_norm": 0.14955772459506989, "learning_rate": 9.693156209389221e-05, "loss": 0.0195, "step": 15650 }, { "grad_norm": 0.2203483134508133, "learning_rate": 9.692585634171905e-05, "loss": 0.0189, "step": 15660 }, { "grad_norm": 0.1756037175655365, "learning_rate": 9.692014545780476e-05, "loss": 0.016, "step": 15670 }, { "grad_norm": 0.22211863100528717, "learning_rate": 9.691442944277393e-05, "loss": 0.0199, "step": 15680 }, { "grad_norm": 0.16637028753757477, "learning_rate": 9.690870829725162e-05, "loss": 0.0155, "step": 15690 }, { "grad_norm": 0.0977361798286438, "learning_rate": 9.69029820218635e-05, "loss": 0.0124, "step": 15700 }, { "grad_norm": 0.160178542137146, "learning_rate": 9.689725061723579e-05, "loss": 0.016, "step": 15710 }, { "grad_norm": 0.16283027827739716, "learning_rate": 9.689151408399527e-05, "loss": 0.0149, "step": 15720 }, { "grad_norm": 0.2372424453496933, "learning_rate": 9.688577242276924e-05, "loss": 0.0189, "step": 15730 }, { "grad_norm": 0.1856587529182434, "learning_rate": 9.688002563418566e-05, "loss": 0.0168, "step": 15740 }, { "grad_norm": 0.15979395806789398, "learning_rate": 9.687427371887293e-05, "loss": 0.0145, "step": 15750 }, { "grad_norm": 0.14811603724956512, "learning_rate": 9.686851667746012e-05, "loss": 0.0157, "step": 15760 }, { "grad_norm": 0.21349769830703735, "learning_rate": 9.686275451057677e-05, "loss": 0.0164, "step": 15770 }, { "grad_norm": 0.23231437802314758, "learning_rate": 9.685698721885308e-05, "loss": 0.0157, "step": 15780 }, { "grad_norm": 0.21491150557994843, "learning_rate": 9.68512148029197e-05, "loss": 0.0176, "step": 15790 }, { "grad_norm": 0.11480478942394257, "learning_rate": 9.684543726340791e-05, "loss": 0.015, "step": 15800 }, { "grad_norm": 0.15045598149299622, "learning_rate": 9.683965460094952e-05, "loss": 0.015, "step": 15810 }, { "grad_norm": 0.13016340136528015, "learning_rate": 9.683386681617694e-05, "loss": 0.0177, "step": 15820 }, { "grad_norm": 0.1725316345691681, "learning_rate": 9.68280739097231e-05, "loss": 0.0116, "step": 15830 }, { "grad_norm": 0.21076466143131256, "learning_rate": 9.682227588222148e-05, "loss": 0.0189, "step": 15840 }, { "grad_norm": 0.09668133407831192, "learning_rate": 9.681647273430618e-05, "loss": 0.0153, "step": 15850 }, { "grad_norm": 0.18286192417144775, "learning_rate": 9.681066446661182e-05, "loss": 0.018, "step": 15860 }, { "grad_norm": 0.16592618823051453, "learning_rate": 9.680485107977357e-05, "loss": 0.0132, "step": 15870 }, { "grad_norm": 0.15865635871887207, "learning_rate": 9.679903257442716e-05, "loss": 0.0149, "step": 15880 }, { "grad_norm": 0.17731989920139313, "learning_rate": 9.679320895120891e-05, "loss": 0.0196, "step": 15890 }, { "grad_norm": 0.15090727806091309, "learning_rate": 9.67873802107557e-05, "loss": 0.0196, "step": 15900 }, { "grad_norm": 0.1576862782239914, "learning_rate": 9.67815463537049e-05, "loss": 0.0153, "step": 15910 }, { "grad_norm": 0.12153247743844986, "learning_rate": 9.677570738069457e-05, "loss": 0.0181, "step": 15920 }, { "grad_norm": 0.33727771043777466, "learning_rate": 9.676986329236318e-05, "loss": 0.0133, "step": 15930 }, { "grad_norm": 0.20109334588050842, "learning_rate": 9.676401408934987e-05, "loss": 0.0195, "step": 15940 }, { "grad_norm": 0.181721493601799, "learning_rate": 9.675815977229428e-05, "loss": 0.0177, "step": 15950 }, { "grad_norm": 0.18403126299381256, "learning_rate": 9.675230034183664e-05, "loss": 0.0198, "step": 15960 }, { "grad_norm": 0.16820403933525085, "learning_rate": 9.674643579861773e-05, "loss": 0.0151, "step": 15970 }, { "grad_norm": 0.15930373966693878, "learning_rate": 9.674056614327886e-05, "loss": 0.0127, "step": 15980 }, { "grad_norm": 0.17542174458503723, "learning_rate": 9.673469137646198e-05, "loss": 0.0172, "step": 15990 }, { "grad_norm": 0.22061395645141602, "learning_rate": 9.67288114988095e-05, "loss": 0.0185, "step": 16000 }, { "grad_norm": 0.16207247972488403, "learning_rate": 9.672292651096447e-05, "loss": 0.0192, "step": 16010 }, { "grad_norm": 0.17851075530052185, "learning_rate": 9.671703641357042e-05, "loss": 0.017, "step": 16020 }, { "grad_norm": 0.1304084211587906, "learning_rate": 9.67111412072715e-05, "loss": 0.0133, "step": 16030 }, { "grad_norm": 0.16096508502960205, "learning_rate": 9.670524089271242e-05, "loss": 0.0128, "step": 16040 }, { "grad_norm": 0.15587657690048218, "learning_rate": 9.669933547053842e-05, "loss": 0.0124, "step": 16050 }, { "grad_norm": 0.21022118628025055, "learning_rate": 9.669342494139531e-05, "loss": 0.0137, "step": 16060 }, { "grad_norm": 0.26741883158683777, "learning_rate": 9.668750930592943e-05, "loss": 0.0176, "step": 16070 }, { "grad_norm": 0.16449519991874695, "learning_rate": 9.668158856478775e-05, "loss": 0.0163, "step": 16080 }, { "grad_norm": 0.1325584501028061, "learning_rate": 9.66756627186177e-05, "loss": 0.0126, "step": 16090 }, { "grad_norm": 0.23523208498954773, "learning_rate": 9.666973176806737e-05, "loss": 0.0131, "step": 16100 }, { "grad_norm": 0.207364022731781, "learning_rate": 9.666379571378534e-05, "loss": 0.0143, "step": 16110 }, { "grad_norm": 0.26169875264167786, "learning_rate": 9.665785455642076e-05, "loss": 0.0169, "step": 16120 }, { "grad_norm": 0.16371750831604004, "learning_rate": 9.665190829662337e-05, "loss": 0.0185, "step": 16130 }, { "grad_norm": 0.22347956895828247, "learning_rate": 9.664595693504342e-05, "loss": 0.0188, "step": 16140 }, { "grad_norm": 0.2219642996788025, "learning_rate": 9.664000047233175e-05, "loss": 0.018, "step": 16150 }, { "grad_norm": 0.16132846474647522, "learning_rate": 9.663403890913976e-05, "loss": 0.0162, "step": 16160 }, { "grad_norm": 0.12044955044984818, "learning_rate": 9.662807224611938e-05, "loss": 0.0127, "step": 16170 }, { "grad_norm": 0.23719993233680725, "learning_rate": 9.662210048392311e-05, "loss": 0.0166, "step": 16180 }, { "grad_norm": 0.13876885175704956, "learning_rate": 9.661612362320405e-05, "loss": 0.0153, "step": 16190 }, { "grad_norm": 0.1577305644750595, "learning_rate": 9.661014166461579e-05, "loss": 0.0129, "step": 16200 }, { "grad_norm": 0.13727830350399017, "learning_rate": 9.66041546088125e-05, "loss": 0.0125, "step": 16210 }, { "grad_norm": 0.2079867720603943, "learning_rate": 9.659816245644895e-05, "loss": 0.0172, "step": 16220 }, { "grad_norm": 0.17092961072921753, "learning_rate": 9.65921652081804e-05, "loss": 0.0151, "step": 16230 }, { "grad_norm": 0.13908012211322784, "learning_rate": 9.658616286466271e-05, "loss": 0.0143, "step": 16240 }, { "grad_norm": 0.133915975689888, "learning_rate": 9.65801554265523e-05, "loss": 0.0105, "step": 16250 }, { "grad_norm": 0.30860939621925354, "learning_rate": 9.657414289450612e-05, "loss": 0.0191, "step": 16260 }, { "grad_norm": 0.1969204843044281, "learning_rate": 9.656812526918171e-05, "loss": 0.0149, "step": 16270 }, { "grad_norm": 0.3078758120536804, "learning_rate": 9.656210255123712e-05, "loss": 0.0211, "step": 16280 }, { "grad_norm": 0.17022846639156342, "learning_rate": 9.6556074741331e-05, "loss": 0.0137, "step": 16290 }, { "grad_norm": 0.187543123960495, "learning_rate": 9.655004184012256e-05, "loss": 0.0182, "step": 16300 }, { "grad_norm": 0.3290688395500183, "learning_rate": 9.654400384827152e-05, "loss": 0.0172, "step": 16310 }, { "grad_norm": 0.23461905121803284, "learning_rate": 9.653796076643818e-05, "loss": 0.0185, "step": 16320 }, { "grad_norm": 0.29496055841445923, "learning_rate": 9.653191259528344e-05, "loss": 0.018, "step": 16330 }, { "grad_norm": 0.1556835174560547, "learning_rate": 9.65258593354687e-05, "loss": 0.0161, "step": 16340 }, { "grad_norm": 0.4489293098449707, "learning_rate": 9.651980098765591e-05, "loss": 0.0215, "step": 16350 }, { "grad_norm": 0.23070476949214935, "learning_rate": 9.651373755250765e-05, "loss": 0.0215, "step": 16360 }, { "grad_norm": 0.17756566405296326, "learning_rate": 9.650766903068697e-05, "loss": 0.0128, "step": 16370 }, { "grad_norm": 0.2865128219127655, "learning_rate": 9.650159542285753e-05, "loss": 0.0147, "step": 16380 }, { "grad_norm": 0.1879929155111313, "learning_rate": 9.649551672968353e-05, "loss": 0.0181, "step": 16390 }, { "grad_norm": 0.13602672517299652, "learning_rate": 9.648943295182973e-05, "loss": 0.0175, "step": 16400 }, { "grad_norm": 0.14591389894485474, "learning_rate": 9.648334408996144e-05, "loss": 0.0145, "step": 16410 }, { "grad_norm": 0.22360140085220337, "learning_rate": 9.647725014474452e-05, "loss": 0.0178, "step": 16420 }, { "grad_norm": 0.1719558835029602, "learning_rate": 9.64711511168454e-05, "loss": 0.0139, "step": 16430 }, { "grad_norm": 0.1906174123287201, "learning_rate": 9.646504700693108e-05, "loss": 0.0166, "step": 16440 }, { "grad_norm": 0.21913772821426392, "learning_rate": 9.645893781566907e-05, "loss": 0.019, "step": 16450 }, { "grad_norm": 0.12066314369440079, "learning_rate": 9.645282354372744e-05, "loss": 0.0152, "step": 16460 }, { "grad_norm": 0.17409802973270416, "learning_rate": 9.644670419177491e-05, "loss": 0.0151, "step": 16470 }, { "grad_norm": 0.1750776469707489, "learning_rate": 9.644057976048062e-05, "loss": 0.0155, "step": 16480 }, { "grad_norm": 0.11735685914754868, "learning_rate": 9.643445025051435e-05, "loss": 0.0139, "step": 16490 }, { "grad_norm": 0.165924534201622, "learning_rate": 9.642831566254641e-05, "loss": 0.0147, "step": 16500 }, { "grad_norm": 0.17543548345565796, "learning_rate": 9.642217599724769e-05, "loss": 0.0125, "step": 16510 }, { "grad_norm": 0.19824710488319397, "learning_rate": 9.64160312552896e-05, "loss": 0.015, "step": 16520 }, { "grad_norm": 0.13606934249401093, "learning_rate": 9.64098814373441e-05, "loss": 0.0135, "step": 16530 }, { "grad_norm": 0.1728261560201645, "learning_rate": 9.640372654408374e-05, "loss": 0.0133, "step": 16540 }, { "grad_norm": 0.156870499253273, "learning_rate": 9.639756657618162e-05, "loss": 0.0125, "step": 16550 }, { "grad_norm": 0.18017955124378204, "learning_rate": 9.639140153431138e-05, "loss": 0.0128, "step": 16560 }, { "grad_norm": 0.11222199350595474, "learning_rate": 9.638523141914721e-05, "loss": 0.0143, "step": 16570 }, { "grad_norm": 0.14482052624225616, "learning_rate": 9.637905623136388e-05, "loss": 0.0108, "step": 16580 }, { "grad_norm": 0.30148744583129883, "learning_rate": 9.637287597163669e-05, "loss": 0.0169, "step": 16590 }, { "grad_norm": 0.15870065987110138, "learning_rate": 9.63666906406415e-05, "loss": 0.0148, "step": 16600 }, { "grad_norm": 0.1852918565273285, "learning_rate": 9.636050023905473e-05, "loss": 0.0141, "step": 16610 }, { "grad_norm": 0.15589900314807892, "learning_rate": 9.635430476755336e-05, "loss": 0.0153, "step": 16620 }, { "grad_norm": 0.28021788597106934, "learning_rate": 9.63481042268149e-05, "loss": 0.0176, "step": 16630 }, { "grad_norm": 0.14903151988983154, "learning_rate": 9.634189861751745e-05, "loss": 0.0145, "step": 16640 }, { "grad_norm": 0.1334490329027176, "learning_rate": 9.633568794033967e-05, "loss": 0.0176, "step": 16650 }, { "grad_norm": 0.32509031891822815, "learning_rate": 9.63294721959607e-05, "loss": 0.0143, "step": 16660 }, { "grad_norm": 0.1667216420173645, "learning_rate": 9.63232513850603e-05, "loss": 0.0158, "step": 16670 }, { "grad_norm": 0.1279415786266327, "learning_rate": 9.631702550831878e-05, "loss": 0.012, "step": 16680 }, { "grad_norm": 0.31122803688049316, "learning_rate": 9.631079456641698e-05, "loss": 0.0133, "step": 16690 }, { "grad_norm": 0.14324277639389038, "learning_rate": 9.630455856003632e-05, "loss": 0.0143, "step": 16700 }, { "grad_norm": 0.1427488625049591, "learning_rate": 9.629831748985876e-05, "loss": 0.0168, "step": 16710 }, { "grad_norm": 0.1394365429878235, "learning_rate": 9.629207135656679e-05, "loss": 0.0122, "step": 16720 }, { "grad_norm": 0.18030449748039246, "learning_rate": 9.628582016084353e-05, "loss": 0.0157, "step": 16730 }, { "grad_norm": 0.12009354680776596, "learning_rate": 9.627956390337254e-05, "loss": 0.0162, "step": 16740 }, { "grad_norm": 0.118726447224617, "learning_rate": 9.627330258483802e-05, "loss": 0.0127, "step": 16750 }, { "grad_norm": 0.13763000071048737, "learning_rate": 9.62670362059247e-05, "loss": 0.0158, "step": 16760 }, { "grad_norm": 0.1244606301188469, "learning_rate": 9.626076476731786e-05, "loss": 0.0141, "step": 16770 }, { "grad_norm": 0.18764236569404602, "learning_rate": 9.625448826970336e-05, "loss": 0.0182, "step": 16780 }, { "grad_norm": 0.12806066870689392, "learning_rate": 9.624820671376755e-05, "loss": 0.013, "step": 16790 }, { "grad_norm": 0.1594417542219162, "learning_rate": 9.62419201001974e-05, "loss": 0.0131, "step": 16800 }, { "grad_norm": 0.09308208525180817, "learning_rate": 9.623562842968037e-05, "loss": 0.0092, "step": 16810 }, { "grad_norm": 0.2032434642314911, "learning_rate": 9.622933170290454e-05, "loss": 0.0165, "step": 16820 }, { "grad_norm": 0.13586285710334778, "learning_rate": 9.622302992055849e-05, "loss": 0.0142, "step": 16830 }, { "grad_norm": 0.16607820987701416, "learning_rate": 9.62167230833314e-05, "loss": 0.0139, "step": 16840 }, { "grad_norm": 0.15098223090171814, "learning_rate": 9.621041119191295e-05, "loss": 0.0143, "step": 16850 }, { "grad_norm": 0.10767407715320587, "learning_rate": 9.620409424699342e-05, "loss": 0.0108, "step": 16860 }, { "grad_norm": 0.207284614443779, "learning_rate": 9.619777224926359e-05, "loss": 0.017, "step": 16870 }, { "grad_norm": 0.1462889164686203, "learning_rate": 9.619144519941485e-05, "loss": 0.0151, "step": 16880 }, { "grad_norm": 0.17473986744880676, "learning_rate": 9.618511309813912e-05, "loss": 0.0163, "step": 16890 }, { "grad_norm": 0.11963033676147461, "learning_rate": 9.617877594612886e-05, "loss": 0.0144, "step": 16900 }, { "grad_norm": 0.11299020051956177, "learning_rate": 9.617243374407707e-05, "loss": 0.0119, "step": 16910 }, { "grad_norm": 0.1176469475030899, "learning_rate": 9.616608649267736e-05, "loss": 0.0163, "step": 16920 }, { "grad_norm": 0.18114715814590454, "learning_rate": 9.615973419262385e-05, "loss": 0.0149, "step": 16930 }, { "grad_norm": 0.13950616121292114, "learning_rate": 9.615337684461119e-05, "loss": 0.0134, "step": 16940 }, { "grad_norm": 0.16498248279094696, "learning_rate": 9.614701444933465e-05, "loss": 0.0167, "step": 16950 }, { "grad_norm": 0.11578100174665451, "learning_rate": 9.614064700748997e-05, "loss": 0.0163, "step": 16960 }, { "grad_norm": 0.20208409428596497, "learning_rate": 9.613427451977352e-05, "loss": 0.0174, "step": 16970 }, { "grad_norm": 0.4016474783420563, "learning_rate": 9.612789698688216e-05, "loss": 0.0163, "step": 16980 }, { "grad_norm": 0.1948927789926529, "learning_rate": 9.612151440951334e-05, "loss": 0.0128, "step": 16990 }, { "grad_norm": 0.14419478178024292, "learning_rate": 9.611512678836506e-05, "loss": 0.0128, "step": 17000 }, { "grad_norm": 0.2335333377122879, "learning_rate": 9.610873412413584e-05, "loss": 0.0192, "step": 17010 }, { "grad_norm": 0.18539735674858093, "learning_rate": 9.610233641752476e-05, "loss": 0.0163, "step": 17020 }, { "grad_norm": 0.2201862335205078, "learning_rate": 9.609593366923151e-05, "loss": 0.0162, "step": 17030 }, { "grad_norm": 0.1592838019132614, "learning_rate": 9.608952587995625e-05, "loss": 0.0152, "step": 17040 }, { "grad_norm": 0.296048641204834, "learning_rate": 9.608311305039972e-05, "loss": 0.018, "step": 17050 }, { "grad_norm": 0.22873494029045105, "learning_rate": 9.607669518126326e-05, "loss": 0.0164, "step": 17060 }, { "grad_norm": 0.17892470955848694, "learning_rate": 9.607027227324866e-05, "loss": 0.0138, "step": 17070 }, { "grad_norm": 0.1942765712738037, "learning_rate": 9.606384432705837e-05, "loss": 0.0145, "step": 17080 }, { "grad_norm": 0.174344003200531, "learning_rate": 9.60574113433953e-05, "loss": 0.0198, "step": 17090 }, { "grad_norm": 0.22215920686721802, "learning_rate": 9.6050973322963e-05, "loss": 0.014, "step": 17100 }, { "grad_norm": 0.22027868032455444, "learning_rate": 9.604453026646547e-05, "loss": 0.0133, "step": 17110 }, { "grad_norm": 0.18599069118499756, "learning_rate": 9.603808217460735e-05, "loss": 0.0172, "step": 17120 }, { "grad_norm": 0.1736038625240326, "learning_rate": 9.603162904809377e-05, "loss": 0.0143, "step": 17130 }, { "grad_norm": 0.12928669154644012, "learning_rate": 9.602517088763045e-05, "loss": 0.0138, "step": 17140 }, { "grad_norm": 0.12131299823522568, "learning_rate": 9.601870769392365e-05, "loss": 0.0128, "step": 17150 }, { "grad_norm": 0.13692158460617065, "learning_rate": 9.601223946768017e-05, "loss": 0.0125, "step": 17160 }, { "grad_norm": 0.09781276434659958, "learning_rate": 9.600576620960734e-05, "loss": 0.0103, "step": 17170 }, { "grad_norm": 0.15759190917015076, "learning_rate": 9.599928792041308e-05, "loss": 0.0147, "step": 17180 }, { "grad_norm": 0.11431118845939636, "learning_rate": 9.599280460080587e-05, "loss": 0.0157, "step": 17190 }, { "grad_norm": 0.14566324651241302, "learning_rate": 9.59863162514947e-05, "loss": 0.0109, "step": 17200 }, { "grad_norm": 0.19663825631141663, "learning_rate": 9.597982287318911e-05, "loss": 0.0153, "step": 17210 }, { "grad_norm": 0.14845165610313416, "learning_rate": 9.597332446659923e-05, "loss": 0.0146, "step": 17220 }, { "grad_norm": 0.18415895104408264, "learning_rate": 9.59668210324357e-05, "loss": 0.0203, "step": 17230 }, { "grad_norm": 0.1486254185438156, "learning_rate": 9.596031257140974e-05, "loss": 0.0159, "step": 17240 }, { "grad_norm": 0.1299394965171814, "learning_rate": 9.59537990842331e-05, "loss": 0.012, "step": 17250 }, { "grad_norm": 0.210699200630188, "learning_rate": 9.594728057161806e-05, "loss": 0.0184, "step": 17260 }, { "grad_norm": 0.19266806542873383, "learning_rate": 9.594075703427752e-05, "loss": 0.0147, "step": 17270 }, { "grad_norm": 0.08639664202928543, "learning_rate": 9.593422847292486e-05, "loss": 0.0125, "step": 17280 }, { "grad_norm": 0.10602188110351562, "learning_rate": 9.592769488827402e-05, "loss": 0.0107, "step": 17290 }, { "grad_norm": 0.16957975924015045, "learning_rate": 9.592115628103952e-05, "loss": 0.0145, "step": 17300 }, { "grad_norm": 0.15032529830932617, "learning_rate": 9.591461265193643e-05, "loss": 0.0129, "step": 17310 }, { "grad_norm": 0.21813708543777466, "learning_rate": 9.590806400168032e-05, "loss": 0.0144, "step": 17320 }, { "grad_norm": 0.15402686595916748, "learning_rate": 9.590151033098735e-05, "loss": 0.0154, "step": 17330 }, { "grad_norm": 0.20461206138134003, "learning_rate": 9.589495164057423e-05, "loss": 0.0149, "step": 17340 }, { "grad_norm": 0.2770480215549469, "learning_rate": 9.58883879311582e-05, "loss": 0.0171, "step": 17350 }, { "grad_norm": 0.14827385544776917, "learning_rate": 9.588181920345705e-05, "loss": 0.015, "step": 17360 }, { "grad_norm": 0.34089195728302, "learning_rate": 9.587524545818913e-05, "loss": 0.0128, "step": 17370 }, { "grad_norm": 0.2032843977212906, "learning_rate": 9.586866669607335e-05, "loss": 0.0153, "step": 17380 }, { "grad_norm": 0.1334264725446701, "learning_rate": 9.586208291782915e-05, "loss": 0.013, "step": 17390 }, { "grad_norm": 0.1236627921462059, "learning_rate": 9.58554941241765e-05, "loss": 0.0138, "step": 17400 }, { "grad_norm": 0.1054222509264946, "learning_rate": 9.584890031583596e-05, "loss": 0.012, "step": 17410 }, { "grad_norm": 0.15154926478862762, "learning_rate": 9.584230149352861e-05, "loss": 0.0113, "step": 17420 }, { "grad_norm": 0.13053663074970245, "learning_rate": 9.58356976579761e-05, "loss": 0.0105, "step": 17430 }, { "grad_norm": 0.13052208721637726, "learning_rate": 9.58290888099006e-05, "loss": 0.012, "step": 17440 }, { "grad_norm": 0.15068960189819336, "learning_rate": 9.582247495002486e-05, "loss": 0.0151, "step": 17450 }, { "grad_norm": 0.13349461555480957, "learning_rate": 9.581585607907214e-05, "loss": 0.014, "step": 17460 }, { "grad_norm": 0.14916084706783295, "learning_rate": 9.580923219776628e-05, "loss": 0.0144, "step": 17470 }, { "grad_norm": 0.11365335434675217, "learning_rate": 9.580260330683167e-05, "loss": 0.0127, "step": 17480 }, { "grad_norm": 0.2615714371204376, "learning_rate": 9.579596940699322e-05, "loss": 0.0147, "step": 17490 }, { "grad_norm": 0.16871896386146545, "learning_rate": 9.578933049897643e-05, "loss": 0.0155, "step": 17500 }, { "grad_norm": 0.11809787154197693, "learning_rate": 9.578268658350728e-05, "loss": 0.0139, "step": 17510 }, { "grad_norm": 0.1425997018814087, "learning_rate": 9.577603766131235e-05, "loss": 0.0095, "step": 17520 }, { "grad_norm": 0.22077956795692444, "learning_rate": 9.576938373311878e-05, "loss": 0.0165, "step": 17530 }, { "grad_norm": 0.1994914710521698, "learning_rate": 9.576272479965421e-05, "loss": 0.0151, "step": 17540 }, { "grad_norm": 0.12257730215787888, "learning_rate": 9.575606086164687e-05, "loss": 0.0129, "step": 17550 }, { "grad_norm": 0.14868766069412231, "learning_rate": 9.57493919198255e-05, "loss": 0.0149, "step": 17560 }, { "grad_norm": 0.15283481776714325, "learning_rate": 9.57427179749194e-05, "loss": 0.0168, "step": 17570 }, { "grad_norm": 0.1484447568655014, "learning_rate": 9.573603902765846e-05, "loss": 0.0159, "step": 17580 }, { "grad_norm": 0.17876526713371277, "learning_rate": 9.572935507877304e-05, "loss": 0.0165, "step": 17590 }, { "grad_norm": 0.14558731019496918, "learning_rate": 9.57226661289941e-05, "loss": 0.0113, "step": 17600 }, { "grad_norm": 0.18669813871383667, "learning_rate": 9.571597217905315e-05, "loss": 0.014, "step": 17610 }, { "grad_norm": 0.15286824107170105, "learning_rate": 9.57092732296822e-05, "loss": 0.0141, "step": 17620 }, { "grad_norm": 0.12437362968921661, "learning_rate": 9.570256928161385e-05, "loss": 0.0147, "step": 17630 }, { "grad_norm": 0.10585683584213257, "learning_rate": 9.569586033558126e-05, "loss": 0.0119, "step": 17640 }, { "grad_norm": 0.17827168107032776, "learning_rate": 9.568914639231807e-05, "loss": 0.0121, "step": 17650 }, { "grad_norm": 0.1957079917192459, "learning_rate": 9.568242745255852e-05, "loss": 0.0127, "step": 17660 }, { "grad_norm": 0.10965403914451599, "learning_rate": 9.567570351703739e-05, "loss": 0.0131, "step": 17670 }, { "grad_norm": 0.12716948986053467, "learning_rate": 9.566897458649001e-05, "loss": 0.0176, "step": 17680 }, { "grad_norm": 0.23182281851768494, "learning_rate": 9.566224066165221e-05, "loss": 0.0157, "step": 17690 }, { "grad_norm": 0.14054644107818604, "learning_rate": 9.565550174326043e-05, "loss": 0.0158, "step": 17700 }, { "grad_norm": 0.11896494030952454, "learning_rate": 9.564875783205162e-05, "loss": 0.012, "step": 17710 }, { "grad_norm": 0.11929409950971603, "learning_rate": 9.564200892876328e-05, "loss": 0.0145, "step": 17720 }, { "grad_norm": 0.26660314202308655, "learning_rate": 9.563525503413348e-05, "loss": 0.0129, "step": 17730 }, { "grad_norm": 0.15379412472248077, "learning_rate": 9.562849614890079e-05, "loss": 0.016, "step": 17740 }, { "grad_norm": 0.15139541029930115, "learning_rate": 9.562173227380436e-05, "loss": 0.0104, "step": 17750 }, { "grad_norm": 0.21009278297424316, "learning_rate": 9.561496340958389e-05, "loss": 0.0175, "step": 17760 }, { "grad_norm": 0.20666219294071198, "learning_rate": 9.560818955697959e-05, "loss": 0.0148, "step": 17770 }, { "grad_norm": 0.16929684579372406, "learning_rate": 9.560141071673228e-05, "loss": 0.0147, "step": 17780 }, { "grad_norm": 0.1754775494337082, "learning_rate": 9.559462688958323e-05, "loss": 0.0144, "step": 17790 }, { "grad_norm": 0.12937700748443604, "learning_rate": 9.558783807627434e-05, "loss": 0.0123, "step": 17800 }, { "grad_norm": 0.10383200645446777, "learning_rate": 9.558104427754801e-05, "loss": 0.015, "step": 17810 }, { "grad_norm": 0.17649544775485992, "learning_rate": 9.557424549414722e-05, "loss": 0.0174, "step": 17820 }, { "grad_norm": 0.13729803264141083, "learning_rate": 9.556744172681546e-05, "loss": 0.0137, "step": 17830 }, { "grad_norm": 0.1615615338087082, "learning_rate": 9.556063297629677e-05, "loss": 0.0149, "step": 17840 }, { "grad_norm": 0.14340335130691528, "learning_rate": 9.555381924333578e-05, "loss": 0.0141, "step": 17850 }, { "grad_norm": 0.21745778620243073, "learning_rate": 9.554700052867758e-05, "loss": 0.0163, "step": 17860 }, { "grad_norm": 0.20425404608249664, "learning_rate": 9.554017683306789e-05, "loss": 0.017, "step": 17870 }, { "grad_norm": 0.15028776228427887, "learning_rate": 9.553334815725294e-05, "loss": 0.0138, "step": 17880 }, { "grad_norm": 0.23290206491947174, "learning_rate": 9.552651450197949e-05, "loss": 0.0135, "step": 17890 }, { "grad_norm": 0.15031106770038605, "learning_rate": 9.551967586799486e-05, "loss": 0.0135, "step": 17900 }, { "grad_norm": 0.11005686223506927, "learning_rate": 9.551283225604692e-05, "loss": 0.0159, "step": 17910 }, { "grad_norm": 0.10313087701797485, "learning_rate": 9.550598366688406e-05, "loss": 0.0141, "step": 17920 }, { "grad_norm": 0.1263742297887802, "learning_rate": 9.549913010125526e-05, "loss": 0.0112, "step": 17930 }, { "grad_norm": 0.3366166651248932, "learning_rate": 9.549227155990999e-05, "loss": 0.0167, "step": 17940 }, { "grad_norm": 0.19227315485477448, "learning_rate": 9.548540804359828e-05, "loss": 0.0134, "step": 17950 }, { "grad_norm": 0.19231270253658295, "learning_rate": 9.547853955307077e-05, "loss": 0.02, "step": 17960 }, { "grad_norm": 0.14816603064537048, "learning_rate": 9.547166608907853e-05, "loss": 0.0135, "step": 17970 }, { "grad_norm": 0.12203823029994965, "learning_rate": 9.546478765237326e-05, "loss": 0.0125, "step": 17980 }, { "grad_norm": 0.14775314927101135, "learning_rate": 9.545790424370715e-05, "loss": 0.0133, "step": 17990 }, { "grad_norm": 0.19711466133594513, "learning_rate": 9.5451015863833e-05, "loss": 0.0185, "step": 18000 }, { "grad_norm": 0.2064550369977951, "learning_rate": 9.544412251350408e-05, "loss": 0.0148, "step": 18010 }, { "grad_norm": 0.2188308984041214, "learning_rate": 9.543722419347422e-05, "loss": 0.0167, "step": 18020 }, { "grad_norm": 0.13905927538871765, "learning_rate": 9.543032090449788e-05, "loss": 0.0122, "step": 18030 }, { "grad_norm": 0.12375077605247498, "learning_rate": 9.542341264732992e-05, "loss": 0.0101, "step": 18040 }, { "grad_norm": 0.17142599821090698, "learning_rate": 9.541649942272585e-05, "loss": 0.0196, "step": 18050 }, { "grad_norm": 0.14814548194408417, "learning_rate": 9.54095812314417e-05, "loss": 0.0167, "step": 18060 }, { "grad_norm": 0.17339149117469788, "learning_rate": 9.540265807423401e-05, "loss": 0.0191, "step": 18070 }, { "grad_norm": 0.19579945504665375, "learning_rate": 9.53957299518599e-05, "loss": 0.0125, "step": 18080 }, { "grad_norm": 0.1360078752040863, "learning_rate": 9.5388796865077e-05, "loss": 0.0128, "step": 18090 }, { "grad_norm": 0.22721385955810547, "learning_rate": 9.538185881464353e-05, "loss": 0.0189, "step": 18100 }, { "grad_norm": 0.12605084478855133, "learning_rate": 9.537491580131821e-05, "loss": 0.0132, "step": 18110 }, { "grad_norm": 0.1204017773270607, "learning_rate": 9.53679678258603e-05, "loss": 0.0166, "step": 18120 }, { "grad_norm": 0.1217615082859993, "learning_rate": 9.536101488902966e-05, "loss": 0.0127, "step": 18130 }, { "grad_norm": 0.11940459907054901, "learning_rate": 9.535405699158663e-05, "loss": 0.0098, "step": 18140 }, { "grad_norm": 0.1315326690673828, "learning_rate": 9.53470941342921e-05, "loss": 0.0099, "step": 18150 }, { "grad_norm": 0.1276845782995224, "learning_rate": 9.534012631790756e-05, "loss": 0.0116, "step": 18160 }, { "grad_norm": 0.10195299983024597, "learning_rate": 9.533315354319494e-05, "loss": 0.0097, "step": 18170 }, { "grad_norm": 0.12007215619087219, "learning_rate": 9.532617581091682e-05, "loss": 0.0115, "step": 18180 }, { "grad_norm": 0.09108338505029678, "learning_rate": 9.531919312183629e-05, "loss": 0.0137, "step": 18190 }, { "grad_norm": 0.28948265314102173, "learning_rate": 9.531220547671688e-05, "loss": 0.0183, "step": 18200 }, { "grad_norm": 0.14999687671661377, "learning_rate": 9.530521287632285e-05, "loss": 0.0144, "step": 18210 }, { "grad_norm": 0.21751797199249268, "learning_rate": 9.529821532141884e-05, "loss": 0.0154, "step": 18220 }, { "grad_norm": 0.4144652187824249, "learning_rate": 9.52912128127701e-05, "loss": 0.0183, "step": 18230 }, { "grad_norm": 0.2242363691329956, "learning_rate": 9.528420535114244e-05, "loss": 0.0199, "step": 18240 }, { "grad_norm": 0.1696610003709793, "learning_rate": 9.527719293730215e-05, "loss": 0.0145, "step": 18250 }, { "grad_norm": 0.1585906594991684, "learning_rate": 9.527017557201611e-05, "loss": 0.0135, "step": 18260 }, { "grad_norm": 0.12642444670200348, "learning_rate": 9.526315325605176e-05, "loss": 0.0194, "step": 18270 }, { "grad_norm": 0.10891667008399963, "learning_rate": 9.525612599017699e-05, "loss": 0.0147, "step": 18280 }, { "grad_norm": 0.10131518542766571, "learning_rate": 9.524909377516033e-05, "loss": 0.0109, "step": 18290 }, { "grad_norm": 0.07733921706676483, "learning_rate": 9.524205661177081e-05, "loss": 0.0107, "step": 18300 }, { "grad_norm": 0.10713929682970047, "learning_rate": 9.523501450077801e-05, "loss": 0.0153, "step": 18310 }, { "grad_norm": 0.1241069808602333, "learning_rate": 9.522796744295202e-05, "loss": 0.0095, "step": 18320 }, { "grad_norm": 0.14465996623039246, "learning_rate": 9.522091543906352e-05, "loss": 0.0141, "step": 18330 }, { "grad_norm": 0.16706477105617523, "learning_rate": 9.521385848988369e-05, "loss": 0.0117, "step": 18340 }, { "grad_norm": 0.10322082787752151, "learning_rate": 9.520679659618428e-05, "loss": 0.0118, "step": 18350 }, { "grad_norm": 0.28596383333206177, "learning_rate": 9.519972975873754e-05, "loss": 0.0169, "step": 18360 }, { "grad_norm": 0.1569071114063263, "learning_rate": 9.519265797831633e-05, "loss": 0.0129, "step": 18370 }, { "grad_norm": 0.14863547682762146, "learning_rate": 9.518558125569399e-05, "loss": 0.0133, "step": 18380 }, { "grad_norm": 0.09497590363025665, "learning_rate": 9.517849959164442e-05, "loss": 0.0111, "step": 18390 }, { "grad_norm": 0.2429877072572708, "learning_rate": 9.517141298694205e-05, "loss": 0.0146, "step": 18400 }, { "grad_norm": 0.1900026500225067, "learning_rate": 9.516432144236188e-05, "loss": 0.0156, "step": 18410 }, { "grad_norm": 0.23721230030059814, "learning_rate": 9.515722495867941e-05, "loss": 0.016, "step": 18420 }, { "grad_norm": 0.14542022347450256, "learning_rate": 9.515012353667072e-05, "loss": 0.0164, "step": 18430 }, { "grad_norm": 0.13708500564098358, "learning_rate": 9.51430171771124e-05, "loss": 0.0172, "step": 18440 }, { "grad_norm": 0.1472712755203247, "learning_rate": 9.513590588078159e-05, "loss": 0.0154, "step": 18450 }, { "grad_norm": 0.0827057957649231, "learning_rate": 9.512878964845597e-05, "loss": 0.0107, "step": 18460 }, { "grad_norm": 0.09551980346441269, "learning_rate": 9.512166848091377e-05, "loss": 0.01, "step": 18470 }, { "grad_norm": 0.09020144492387772, "learning_rate": 9.511454237893376e-05, "loss": 0.0105, "step": 18480 }, { "grad_norm": 0.1241927444934845, "learning_rate": 9.51074113432952e-05, "loss": 0.0146, "step": 18490 }, { "grad_norm": 0.18743476271629333, "learning_rate": 9.510027537477797e-05, "loss": 0.0156, "step": 18500 }, { "grad_norm": 0.13608942925930023, "learning_rate": 9.509313447416242e-05, "loss": 0.0146, "step": 18510 }, { "grad_norm": 0.16859188675880432, "learning_rate": 9.508598864222949e-05, "loss": 0.0155, "step": 18520 }, { "grad_norm": 0.1758957803249359, "learning_rate": 9.507883787976062e-05, "loss": 0.015, "step": 18530 }, { "grad_norm": 0.197950541973114, "learning_rate": 9.507168218753781e-05, "loss": 0.0151, "step": 18540 }, { "grad_norm": 0.19090990722179413, "learning_rate": 9.506452156634362e-05, "loss": 0.0148, "step": 18550 }, { "grad_norm": 0.11980689316987991, "learning_rate": 9.505735601696109e-05, "loss": 0.011, "step": 18560 }, { "grad_norm": 0.16521471738815308, "learning_rate": 9.505018554017385e-05, "loss": 0.0122, "step": 18570 }, { "grad_norm": 0.19960089027881622, "learning_rate": 9.504301013676604e-05, "loss": 0.0124, "step": 18580 }, { "grad_norm": 0.19279640913009644, "learning_rate": 9.503582980752238e-05, "loss": 0.0153, "step": 18590 }, { "grad_norm": 0.20514339208602905, "learning_rate": 9.502864455322809e-05, "loss": 0.0132, "step": 18600 }, { "grad_norm": 0.23738053441047668, "learning_rate": 9.502145437466891e-05, "loss": 0.0189, "step": 18610 }, { "grad_norm": 0.22715741395950317, "learning_rate": 9.501425927263116e-05, "loss": 0.017, "step": 18620 }, { "grad_norm": 0.15758772194385529, "learning_rate": 9.500705924790172e-05, "loss": 0.0132, "step": 18630 }, { "grad_norm": 0.24950577318668365, "learning_rate": 9.499985430126794e-05, "loss": 0.0144, "step": 18640 }, { "grad_norm": 0.15031254291534424, "learning_rate": 9.499264443351775e-05, "loss": 0.0127, "step": 18650 }, { "grad_norm": 0.17883475124835968, "learning_rate": 9.498542964543961e-05, "loss": 0.0142, "step": 18660 }, { "grad_norm": 0.10436303168535233, "learning_rate": 9.497820993782252e-05, "loss": 0.0125, "step": 18670 }, { "grad_norm": 0.12974968552589417, "learning_rate": 9.497098531145601e-05, "loss": 0.0138, "step": 18680 }, { "grad_norm": 0.12724997103214264, "learning_rate": 9.496375576713017e-05, "loss": 0.0119, "step": 18690 }, { "grad_norm": 0.22166164219379425, "learning_rate": 9.49565213056356e-05, "loss": 0.0163, "step": 18700 }, { "grad_norm": 0.13649876415729523, "learning_rate": 9.494928192776342e-05, "loss": 0.013, "step": 18710 }, { "grad_norm": 0.11422639340162277, "learning_rate": 9.494203763430538e-05, "loss": 0.0118, "step": 18720 }, { "grad_norm": 0.2627640962600708, "learning_rate": 9.493478842605366e-05, "loss": 0.016, "step": 18730 }, { "grad_norm": 0.10279325395822525, "learning_rate": 9.492753430380105e-05, "loss": 0.0127, "step": 18740 }, { "grad_norm": 0.2080105096101761, "learning_rate": 9.492027526834083e-05, "loss": 0.014, "step": 18750 }, { "grad_norm": 0.3354438543319702, "learning_rate": 9.491301132046684e-05, "loss": 0.0122, "step": 18760 }, { "grad_norm": 0.17297111451625824, "learning_rate": 9.490574246097345e-05, "loss": 0.0137, "step": 18770 }, { "grad_norm": 0.265390545129776, "learning_rate": 9.48984686906556e-05, "loss": 0.0135, "step": 18780 }, { "grad_norm": 0.1543634682893753, "learning_rate": 9.489119001030871e-05, "loss": 0.0117, "step": 18790 }, { "grad_norm": 0.22723013162612915, "learning_rate": 9.488390642072878e-05, "loss": 0.0154, "step": 18800 }, { "grad_norm": 0.17431381344795227, "learning_rate": 9.48766179227123e-05, "loss": 0.0125, "step": 18810 }, { "grad_norm": 0.19091902673244476, "learning_rate": 9.486932451705636e-05, "loss": 0.0163, "step": 18820 }, { "grad_norm": 0.13735371828079224, "learning_rate": 9.486202620455857e-05, "loss": 0.0128, "step": 18830 }, { "grad_norm": 0.16869747638702393, "learning_rate": 9.485472298601704e-05, "loss": 0.0159, "step": 18840 }, { "grad_norm": 0.13688477873802185, "learning_rate": 9.484741486223043e-05, "loss": 0.013, "step": 18850 }, { "grad_norm": 0.11719440668821335, "learning_rate": 9.484010183399797e-05, "loss": 0.0113, "step": 18860 }, { "grad_norm": 0.2743169665336609, "learning_rate": 9.483278390211938e-05, "loss": 0.0114, "step": 18870 }, { "grad_norm": 0.17138142883777618, "learning_rate": 9.482546106739496e-05, "loss": 0.0152, "step": 18880 }, { "grad_norm": 0.11472739279270172, "learning_rate": 9.48181333306255e-05, "loss": 0.0154, "step": 18890 }, { "grad_norm": 0.15053527057170868, "learning_rate": 9.481080069261237e-05, "loss": 0.0109, "step": 18900 }, { "grad_norm": 0.12490653991699219, "learning_rate": 9.480346315415745e-05, "loss": 0.0116, "step": 18910 }, { "grad_norm": 0.11839121580123901, "learning_rate": 9.479612071606314e-05, "loss": 0.0112, "step": 18920 }, { "grad_norm": 0.13011443614959717, "learning_rate": 9.478877337913244e-05, "loss": 0.0116, "step": 18930 }, { "grad_norm": 0.1451033502817154, "learning_rate": 9.478142114416881e-05, "loss": 0.0106, "step": 18940 }, { "grad_norm": 0.11015048623085022, "learning_rate": 9.47740640119763e-05, "loss": 0.0116, "step": 18950 }, { "grad_norm": 0.13194270431995392, "learning_rate": 9.476670198335947e-05, "loss": 0.0124, "step": 18960 }, { "grad_norm": 0.20331284403800964, "learning_rate": 9.47593350591234e-05, "loss": 0.0144, "step": 18970 }, { "grad_norm": 0.16843266785144806, "learning_rate": 9.475196324007376e-05, "loss": 0.0132, "step": 18980 }, { "grad_norm": 0.15038709342479706, "learning_rate": 9.474458652701669e-05, "loss": 0.0124, "step": 18990 }, { "grad_norm": 0.14288124442100525, "learning_rate": 9.473720492075892e-05, "loss": 0.0124, "step": 19000 }, { "grad_norm": 0.19257856905460358, "learning_rate": 9.472981842210768e-05, "loss": 0.0133, "step": 19010 }, { "grad_norm": 0.19932854175567627, "learning_rate": 9.472242703187074e-05, "loss": 0.0143, "step": 19020 }, { "grad_norm": 0.16519300639629364, "learning_rate": 9.471503075085643e-05, "loss": 0.0146, "step": 19030 }, { "grad_norm": 0.2022336721420288, "learning_rate": 9.470762957987359e-05, "loss": 0.0131, "step": 19040 }, { "grad_norm": 0.12624944746494293, "learning_rate": 9.470022351973158e-05, "loss": 0.0109, "step": 19050 }, { "grad_norm": 0.2658054828643799, "learning_rate": 9.469281257124034e-05, "loss": 0.0121, "step": 19060 }, { "grad_norm": 0.20805221796035767, "learning_rate": 9.46853967352103e-05, "loss": 0.0145, "step": 19070 }, { "grad_norm": 0.25356319546699524, "learning_rate": 9.467797601245246e-05, "loss": 0.0198, "step": 19080 }, { "grad_norm": 0.1339343786239624, "learning_rate": 9.467055040377834e-05, "loss": 0.0143, "step": 19090 }, { "grad_norm": 0.3054198622703552, "learning_rate": 9.466311990999999e-05, "loss": 0.0151, "step": 19100 }, { "grad_norm": 0.11672776192426682, "learning_rate": 9.465568453193e-05, "loss": 0.013, "step": 19110 }, { "grad_norm": 0.15185971558094025, "learning_rate": 9.464824427038148e-05, "loss": 0.0131, "step": 19120 }, { "grad_norm": 0.4072279930114746, "learning_rate": 9.46407991261681e-05, "loss": 0.0166, "step": 19130 }, { "grad_norm": 0.13902580738067627, "learning_rate": 9.463334910010404e-05, "loss": 0.0152, "step": 19140 }, { "grad_norm": 0.11346828192472458, "learning_rate": 9.462589419300403e-05, "loss": 0.0131, "step": 19150 }, { "grad_norm": 0.08568896353244781, "learning_rate": 9.461843440568333e-05, "loss": 0.0098, "step": 19160 }, { "grad_norm": 0.29810142517089844, "learning_rate": 9.461096973895773e-05, "loss": 0.0149, "step": 19170 }, { "grad_norm": 0.15266206860542297, "learning_rate": 9.460350019364355e-05, "loss": 0.0165, "step": 19180 }, { "grad_norm": 0.14619125425815582, "learning_rate": 9.459602577055764e-05, "loss": 0.0172, "step": 19190 }, { "grad_norm": 0.12254755198955536, "learning_rate": 9.45885464705174e-05, "loss": 0.0137, "step": 19200 }, { "grad_norm": 0.16977863013744354, "learning_rate": 9.458106229434076e-05, "loss": 0.0147, "step": 19210 }, { "grad_norm": 0.138040229678154, "learning_rate": 9.457357324284617e-05, "loss": 0.0119, "step": 19220 }, { "grad_norm": 0.14626435935497284, "learning_rate": 9.456607931685262e-05, "loss": 0.0132, "step": 19230 }, { "grad_norm": 0.12011564522981644, "learning_rate": 9.455858051717965e-05, "loss": 0.0117, "step": 19240 }, { "grad_norm": 0.13912072777748108, "learning_rate": 9.45510768446473e-05, "loss": 0.0186, "step": 19250 }, { "grad_norm": 0.12842689454555511, "learning_rate": 9.454356830007618e-05, "loss": 0.0114, "step": 19260 }, { "grad_norm": 0.16778801381587982, "learning_rate": 9.45360548842874e-05, "loss": 0.0152, "step": 19270 }, { "grad_norm": 0.12584245204925537, "learning_rate": 9.452853659810261e-05, "loss": 0.0138, "step": 19280 }, { "grad_norm": 0.09765877574682236, "learning_rate": 9.452101344234401e-05, "loss": 0.0129, "step": 19290 }, { "grad_norm": 0.23901042342185974, "learning_rate": 9.451348541783431e-05, "loss": 0.0098, "step": 19300 }, { "grad_norm": 0.17831389605998993, "learning_rate": 9.450595252539678e-05, "loss": 0.0152, "step": 19310 }, { "grad_norm": 0.15628840029239655, "learning_rate": 9.449841476585518e-05, "loss": 0.0127, "step": 19320 }, { "grad_norm": 0.11326155811548233, "learning_rate": 9.449087214003384e-05, "loss": 0.0131, "step": 19330 }, { "grad_norm": 0.0962860956788063, "learning_rate": 9.448332464875765e-05, "loss": 0.0117, "step": 19340 }, { "grad_norm": 0.09956884384155273, "learning_rate": 9.447577229285192e-05, "loss": 0.0095, "step": 19350 }, { "grad_norm": 0.17092451453208923, "learning_rate": 9.446821507314261e-05, "loss": 0.0104, "step": 19360 }, { "grad_norm": 0.13134799897670746, "learning_rate": 9.446065299045617e-05, "loss": 0.0106, "step": 19370 }, { "grad_norm": 0.11923711001873016, "learning_rate": 9.445308604561955e-05, "loss": 0.0123, "step": 19380 }, { "grad_norm": 0.09066202491521835, "learning_rate": 9.444551423946028e-05, "loss": 0.013, "step": 19390 }, { "grad_norm": 0.14307841658592224, "learning_rate": 9.443793757280638e-05, "loss": 0.0107, "step": 19400 }, { "grad_norm": 0.12345556169748306, "learning_rate": 9.443035604648646e-05, "loss": 0.0121, "step": 19410 }, { "grad_norm": 0.11624568700790405, "learning_rate": 9.44227696613296e-05, "loss": 0.0089, "step": 19420 }, { "grad_norm": 0.18908338248729706, "learning_rate": 9.441517841816542e-05, "loss": 0.0112, "step": 19430 }, { "grad_norm": 0.13135607540607452, "learning_rate": 9.440758231782413e-05, "loss": 0.0097, "step": 19440 }, { "grad_norm": 0.14297446608543396, "learning_rate": 9.439998136113639e-05, "loss": 0.0098, "step": 19450 }, { "grad_norm": 0.2666758894920349, "learning_rate": 9.439237554893344e-05, "loss": 0.0137, "step": 19460 }, { "grad_norm": 0.12027698755264282, "learning_rate": 9.438476488204705e-05, "loss": 0.015, "step": 19470 }, { "grad_norm": 0.14125916361808777, "learning_rate": 9.43771493613095e-05, "loss": 0.013, "step": 19480 }, { "grad_norm": 0.09411866217851639, "learning_rate": 9.436952898755362e-05, "loss": 0.0106, "step": 19490 }, { "grad_norm": 0.30566155910491943, "learning_rate": 9.436190376161276e-05, "loss": 0.0132, "step": 19500 }, { "grad_norm": 0.11824653297662735, "learning_rate": 9.43542736843208e-05, "loss": 0.0168, "step": 19510 }, { "grad_norm": 0.296597421169281, "learning_rate": 9.434663875651216e-05, "loss": 0.0157, "step": 19520 }, { "grad_norm": 0.17526531219482422, "learning_rate": 9.433899897902177e-05, "loss": 0.0139, "step": 19530 }, { "grad_norm": 0.12396197766065598, "learning_rate": 9.433135435268511e-05, "loss": 0.0104, "step": 19540 }, { "grad_norm": 0.3512357771396637, "learning_rate": 9.432370487833819e-05, "loss": 0.0177, "step": 19550 }, { "grad_norm": 0.13219091296195984, "learning_rate": 9.431605055681756e-05, "loss": 0.0139, "step": 19560 }, { "grad_norm": 0.23621033132076263, "learning_rate": 9.430839138896026e-05, "loss": 0.016, "step": 19570 }, { "grad_norm": 0.2805333435535431, "learning_rate": 9.43007273756039e-05, "loss": 0.0117, "step": 19580 }, { "grad_norm": 0.16843059659004211, "learning_rate": 9.429305851758658e-05, "loss": 0.016, "step": 19590 }, { "grad_norm": 0.25097912549972534, "learning_rate": 9.428538481574699e-05, "loss": 0.0156, "step": 19600 }, { "grad_norm": 0.17304478585720062, "learning_rate": 9.42777062709243e-05, "loss": 0.0125, "step": 19610 }, { "grad_norm": 0.28684794902801514, "learning_rate": 9.427002288395821e-05, "loss": 0.0164, "step": 19620 }, { "grad_norm": 0.108683742582798, "learning_rate": 9.426233465568898e-05, "loss": 0.0121, "step": 19630 }, { "grad_norm": 0.20501162111759186, "learning_rate": 9.42546415869574e-05, "loss": 0.0163, "step": 19640 }, { "grad_norm": 0.2827351987361908, "learning_rate": 9.424694367860473e-05, "loss": 0.0138, "step": 19650 }, { "grad_norm": 0.12752343714237213, "learning_rate": 9.423924093147284e-05, "loss": 0.0129, "step": 19660 }, { "grad_norm": 0.12144196033477783, "learning_rate": 9.423153334640407e-05, "loss": 0.0121, "step": 19670 }, { "grad_norm": 0.12487338483333588, "learning_rate": 9.42238209242413e-05, "loss": 0.0127, "step": 19680 }, { "grad_norm": 0.14484384655952454, "learning_rate": 9.421610366582798e-05, "loss": 0.0117, "step": 19690 }, { "grad_norm": 0.17743292450904846, "learning_rate": 9.420838157200803e-05, "loss": 0.0114, "step": 19700 }, { "grad_norm": 0.22589647769927979, "learning_rate": 9.420065464362594e-05, "loss": 0.0155, "step": 19710 }, { "grad_norm": 0.13110335171222687, "learning_rate": 9.419292288152673e-05, "loss": 0.0115, "step": 19720 }, { "grad_norm": 0.12168275564908981, "learning_rate": 9.418518628655588e-05, "loss": 0.0141, "step": 19730 }, { "grad_norm": 0.13716641068458557, "learning_rate": 9.417744485955951e-05, "loss": 0.012, "step": 19740 }, { "grad_norm": 0.09215670078992844, "learning_rate": 9.41696986013842e-05, "loss": 0.0109, "step": 19750 }, { "grad_norm": 0.09303099662065506, "learning_rate": 9.416194751287705e-05, "loss": 0.0095, "step": 19760 }, { "grad_norm": 0.4005914330482483, "learning_rate": 9.415419159488572e-05, "loss": 0.0208, "step": 19770 }, { "grad_norm": 0.17127180099487305, "learning_rate": 9.414643084825837e-05, "loss": 0.0153, "step": 19780 }, { "grad_norm": 0.1321900635957718, "learning_rate": 9.413866527384372e-05, "loss": 0.0124, "step": 19790 }, { "grad_norm": 0.3258173167705536, "learning_rate": 9.4130894872491e-05, "loss": 0.0156, "step": 19800 }, { "grad_norm": 0.15711984038352966, "learning_rate": 9.412311964504998e-05, "loss": 0.0159, "step": 19810 }, { "grad_norm": 0.3477756381034851, "learning_rate": 9.411533959237091e-05, "loss": 0.0164, "step": 19820 }, { "grad_norm": 0.1662907898426056, "learning_rate": 9.410755471530464e-05, "loss": 0.0149, "step": 19830 }, { "grad_norm": 0.1888793706893921, "learning_rate": 9.40997650147025e-05, "loss": 0.0147, "step": 19840 }, { "grad_norm": 0.10699013620615005, "learning_rate": 9.409197049141637e-05, "loss": 0.0113, "step": 19850 }, { "grad_norm": 0.19528940320014954, "learning_rate": 9.408417114629863e-05, "loss": 0.0166, "step": 19860 }, { "grad_norm": 0.235127791762352, "learning_rate": 9.40763669802022e-05, "loss": 0.0156, "step": 19870 }, { "grad_norm": 0.12862779200077057, "learning_rate": 9.406855799398056e-05, "loss": 0.0159, "step": 19880 }, { "grad_norm": 0.16014203429222107, "learning_rate": 9.406074418848767e-05, "loss": 0.013, "step": 19890 }, { "grad_norm": 0.13649283349514008, "learning_rate": 9.405292556457805e-05, "loss": 0.0131, "step": 19900 }, { "grad_norm": 0.2939833700656891, "learning_rate": 9.404510212310671e-05, "loss": 0.019, "step": 19910 }, { "grad_norm": 0.15149454772472382, "learning_rate": 9.403727386492924e-05, "loss": 0.0122, "step": 19920 }, { "grad_norm": 0.1540224850177765, "learning_rate": 9.40294407909017e-05, "loss": 0.0105, "step": 19930 }, { "grad_norm": 0.10377683490514755, "learning_rate": 9.40216029018807e-05, "loss": 0.0114, "step": 19940 }, { "grad_norm": 0.21789762377738953, "learning_rate": 9.401376019872338e-05, "loss": 0.0118, "step": 19950 }, { "grad_norm": 0.13668125867843628, "learning_rate": 9.400591268228746e-05, "loss": 0.0143, "step": 19960 }, { "grad_norm": 0.2567516565322876, "learning_rate": 9.399806035343106e-05, "loss": 0.0162, "step": 19970 }, { "grad_norm": 0.1411179006099701, "learning_rate": 9.399020321301294e-05, "loss": 0.012, "step": 19980 }, { "grad_norm": 0.19975857436656952, "learning_rate": 9.398234126189234e-05, "loss": 0.0134, "step": 19990 }, { "grad_norm": 0.14636534452438354, "learning_rate": 9.397447450092902e-05, "loss": 0.0138, "step": 20000 }, { "grad_norm": 0.10984044522047043, "learning_rate": 9.39666029309833e-05, "loss": 0.016, "step": 20010 }, { "grad_norm": 0.08920703083276749, "learning_rate": 9.395872655291596e-05, "loss": 0.0125, "step": 20020 }, { "grad_norm": 0.21259324252605438, "learning_rate": 9.395084536758838e-05, "loss": 0.0164, "step": 20030 }, { "grad_norm": 0.141640767455101, "learning_rate": 9.394295937586243e-05, "loss": 0.0134, "step": 20040 }, { "grad_norm": 0.12906494736671448, "learning_rate": 9.393506857860052e-05, "loss": 0.0111, "step": 20050 }, { "grad_norm": 0.25108325481414795, "learning_rate": 9.392717297666555e-05, "loss": 0.0145, "step": 20060 }, { "grad_norm": 0.16038592159748077, "learning_rate": 9.391927257092101e-05, "loss": 0.0137, "step": 20070 }, { "grad_norm": 0.26975467801094055, "learning_rate": 9.391136736223085e-05, "loss": 0.0137, "step": 20080 }, { "grad_norm": 0.17789581418037415, "learning_rate": 9.390345735145956e-05, "loss": 0.0136, "step": 20090 }, { "grad_norm": 0.16851532459259033, "learning_rate": 9.389554253947219e-05, "loss": 0.017, "step": 20100 }, { "grad_norm": 0.12233208864927292, "learning_rate": 9.388762292713428e-05, "loss": 0.0143, "step": 20110 }, { "grad_norm": 0.10210433602333069, "learning_rate": 9.38796985153119e-05, "loss": 0.0132, "step": 20120 }, { "grad_norm": 0.09175745397806168, "learning_rate": 9.387176930487169e-05, "loss": 0.0117, "step": 20130 }, { "grad_norm": 0.11222512274980545, "learning_rate": 9.386383529668072e-05, "loss": 0.0101, "step": 20140 }, { "grad_norm": 0.10430796444416046, "learning_rate": 9.385589649160669e-05, "loss": 0.0118, "step": 20150 }, { "grad_norm": 0.16119220852851868, "learning_rate": 9.384795289051775e-05, "loss": 0.0109, "step": 20160 }, { "grad_norm": 0.1951480507850647, "learning_rate": 9.384000449428261e-05, "loss": 0.0121, "step": 20170 }, { "grad_norm": 0.1476118117570877, "learning_rate": 9.383205130377048e-05, "loss": 0.0121, "step": 20180 }, { "grad_norm": 0.10396480560302734, "learning_rate": 9.382409331985114e-05, "loss": 0.0117, "step": 20190 }, { "grad_norm": 0.09688235074281693, "learning_rate": 9.381613054339482e-05, "loss": 0.0095, "step": 20200 }, { "grad_norm": 0.24099567532539368, "learning_rate": 9.380816297527235e-05, "loss": 0.0133, "step": 20210 }, { "grad_norm": 0.20482249557971954, "learning_rate": 9.380019061635506e-05, "loss": 0.0152, "step": 20220 }, { "grad_norm": 0.14042682945728302, "learning_rate": 9.379221346751474e-05, "loss": 0.0104, "step": 20230 }, { "grad_norm": 0.174514502286911, "learning_rate": 9.378423152962382e-05, "loss": 0.0105, "step": 20240 }, { "grad_norm": 0.1934213787317276, "learning_rate": 9.377624480355517e-05, "loss": 0.0129, "step": 20250 }, { "grad_norm": 0.18797217309474945, "learning_rate": 9.376825329018219e-05, "loss": 0.0144, "step": 20260 }, { "grad_norm": 0.1494952142238617, "learning_rate": 9.376025699037884e-05, "loss": 0.0099, "step": 20270 }, { "grad_norm": 0.15665562450885773, "learning_rate": 9.37522559050196e-05, "loss": 0.011, "step": 20280 }, { "grad_norm": 0.2061758190393448, "learning_rate": 9.37442500349794e-05, "loss": 0.0134, "step": 20290 }, { "grad_norm": 0.16564242541790009, "learning_rate": 9.373623938113381e-05, "loss": 0.0142, "step": 20300 }, { "grad_norm": 0.14077012240886688, "learning_rate": 9.372822394435883e-05, "loss": 0.0146, "step": 20310 }, { "grad_norm": 0.15545062720775604, "learning_rate": 9.372020372553102e-05, "loss": 0.0102, "step": 20320 }, { "grad_norm": 0.2960515022277832, "learning_rate": 9.371217872552746e-05, "loss": 0.0178, "step": 20330 }, { "grad_norm": 0.158604696393013, "learning_rate": 9.370414894522576e-05, "loss": 0.0148, "step": 20340 }, { "grad_norm": 0.21210037171840668, "learning_rate": 9.369611438550406e-05, "loss": 0.0167, "step": 20350 }, { "grad_norm": 0.12627814710140228, "learning_rate": 9.368807504724095e-05, "loss": 0.0127, "step": 20360 }, { "grad_norm": 0.2304926961660385, "learning_rate": 9.368003093131565e-05, "loss": 0.0155, "step": 20370 }, { "grad_norm": 0.46739181876182556, "learning_rate": 9.367198203860785e-05, "loss": 0.0171, "step": 20380 }, { "grad_norm": 0.14798377454280853, "learning_rate": 9.366392836999774e-05, "loss": 0.0193, "step": 20390 }, { "grad_norm": 0.37159463763237, "learning_rate": 9.365586992636607e-05, "loss": 0.016, "step": 20400 }, { "grad_norm": 0.1567990779876709, "learning_rate": 9.364780670859412e-05, "loss": 0.0161, "step": 20410 }, { "grad_norm": 0.24635948240756989, "learning_rate": 9.363973871756364e-05, "loss": 0.0103, "step": 20420 }, { "grad_norm": 0.13959506154060364, "learning_rate": 9.363166595415696e-05, "loss": 0.0103, "step": 20430 }, { "grad_norm": 0.11943871527910233, "learning_rate": 9.362358841925686e-05, "loss": 0.0108, "step": 20440 }, { "grad_norm": 0.24005696177482605, "learning_rate": 9.361550611374674e-05, "loss": 0.0107, "step": 20450 }, { "grad_norm": 0.09405238181352615, "learning_rate": 9.360741903851043e-05, "loss": 0.0156, "step": 20460 }, { "grad_norm": 0.12427427619695663, "learning_rate": 9.359932719443236e-05, "loss": 0.0123, "step": 20470 }, { "grad_norm": 0.11724238097667694, "learning_rate": 9.35912305823974e-05, "loss": 0.0123, "step": 20480 }, { "grad_norm": 0.11543998122215271, "learning_rate": 9.358312920329101e-05, "loss": 0.0135, "step": 20490 }, { "grad_norm": 0.12544643878936768, "learning_rate": 9.357502305799914e-05, "loss": 0.0141, "step": 20500 }, { "grad_norm": 0.22047744691371918, "learning_rate": 9.356691214740824e-05, "loss": 0.0111, "step": 20510 }, { "grad_norm": 0.18712978065013885, "learning_rate": 9.355879647240535e-05, "loss": 0.0176, "step": 20520 }, { "grad_norm": 0.4290134608745575, "learning_rate": 9.355067603387798e-05, "loss": 0.0193, "step": 20530 }, { "grad_norm": 0.15275637805461884, "learning_rate": 9.354255083271412e-05, "loss": 0.0136, "step": 20540 }, { "grad_norm": 0.16856802999973297, "learning_rate": 9.353442086980239e-05, "loss": 0.0118, "step": 20550 }, { "grad_norm": 0.27743053436279297, "learning_rate": 9.352628614603185e-05, "loss": 0.0185, "step": 20560 }, { "grad_norm": 0.18338540196418762, "learning_rate": 9.351814666229209e-05, "loss": 0.0168, "step": 20570 }, { "grad_norm": 0.22908726334571838, "learning_rate": 9.351000241947324e-05, "loss": 0.017, "step": 20580 }, { "grad_norm": 0.09960919618606567, "learning_rate": 9.350185341846594e-05, "loss": 0.0127, "step": 20590 }, { "grad_norm": 0.1741703599691391, "learning_rate": 9.349369966016134e-05, "loss": 0.0154, "step": 20600 }, { "grad_norm": 0.14869850873947144, "learning_rate": 9.348554114545117e-05, "loss": 0.0133, "step": 20610 }, { "grad_norm": 0.1755620688199997, "learning_rate": 9.347737787522758e-05, "loss": 0.0114, "step": 20620 }, { "grad_norm": 0.16471093893051147, "learning_rate": 9.346920985038332e-05, "loss": 0.0152, "step": 20630 }, { "grad_norm": 0.1635434925556183, "learning_rate": 9.346103707181162e-05, "loss": 0.0139, "step": 20640 }, { "grad_norm": 0.2605763375759125, "learning_rate": 9.345285954040626e-05, "loss": 0.0162, "step": 20650 }, { "grad_norm": 0.16613809764385223, "learning_rate": 9.34446772570615e-05, "loss": 0.016, "step": 20660 }, { "grad_norm": 0.11983953416347504, "learning_rate": 9.343649022267214e-05, "loss": 0.0121, "step": 20670 }, { "grad_norm": 0.16744017601013184, "learning_rate": 9.342829843813353e-05, "loss": 0.0148, "step": 20680 }, { "grad_norm": 0.2013935148715973, "learning_rate": 9.342010190434149e-05, "loss": 0.0168, "step": 20690 }, { "grad_norm": 0.1824418157339096, "learning_rate": 9.34119006221924e-05, "loss": 0.0162, "step": 20700 }, { "grad_norm": 0.11549786478281021, "learning_rate": 9.340369459258313e-05, "loss": 0.012, "step": 20710 }, { "grad_norm": 0.2849206030368805, "learning_rate": 9.339548381641106e-05, "loss": 0.0168, "step": 20720 }, { "grad_norm": 0.27170270681381226, "learning_rate": 9.338726829457413e-05, "loss": 0.0158, "step": 20730 }, { "grad_norm": 0.18926995992660522, "learning_rate": 9.337904802797078e-05, "loss": 0.0153, "step": 20740 }, { "grad_norm": 0.15827926993370056, "learning_rate": 9.337082301749993e-05, "loss": 0.0125, "step": 20750 }, { "grad_norm": 0.12923066318035126, "learning_rate": 9.336259326406109e-05, "loss": 0.0126, "step": 20760 }, { "grad_norm": 0.1115986779332161, "learning_rate": 9.335435876855427e-05, "loss": 0.0142, "step": 20770 }, { "grad_norm": 0.09256815165281296, "learning_rate": 9.334611953187994e-05, "loss": 0.0086, "step": 20780 }, { "grad_norm": 0.22089022397994995, "learning_rate": 9.333787555493914e-05, "loss": 0.0105, "step": 20790 }, { "grad_norm": 0.17675486207008362, "learning_rate": 9.332962683863345e-05, "loss": 0.0128, "step": 20800 }, { "grad_norm": 0.1780850738286972, "learning_rate": 9.332137338386489e-05, "loss": 0.0114, "step": 20810 }, { "grad_norm": 0.11488503962755203, "learning_rate": 9.33131151915361e-05, "loss": 0.0115, "step": 20820 }, { "grad_norm": 0.10368473827838898, "learning_rate": 9.330485226255012e-05, "loss": 0.0128, "step": 20830 }, { "grad_norm": 0.07965762168169022, "learning_rate": 9.329658459781061e-05, "loss": 0.0094, "step": 20840 }, { "grad_norm": 0.27115872502326965, "learning_rate": 9.328831219822172e-05, "loss": 0.0139, "step": 20850 }, { "grad_norm": 0.14157292246818542, "learning_rate": 9.328003506468808e-05, "loss": 0.0128, "step": 20860 }, { "grad_norm": 0.15667304396629333, "learning_rate": 9.327175319811488e-05, "loss": 0.0106, "step": 20870 }, { "grad_norm": 0.28508269786834717, "learning_rate": 9.326346659940781e-05, "loss": 0.015, "step": 20880 }, { "grad_norm": 0.17740800976753235, "learning_rate": 9.325517526947308e-05, "loss": 0.0132, "step": 20890 }, { "grad_norm": 0.21746845543384552, "learning_rate": 9.32468792092174e-05, "loss": 0.0113, "step": 20900 }, { "grad_norm": 0.33148685097694397, "learning_rate": 9.323857841954803e-05, "loss": 0.0138, "step": 20910 }, { "grad_norm": 0.1396327167749405, "learning_rate": 9.323027290137276e-05, "loss": 0.013, "step": 20920 }, { "grad_norm": 0.12941190600395203, "learning_rate": 9.322196265559981e-05, "loss": 0.0113, "step": 20930 }, { "grad_norm": 0.11173366755247116, "learning_rate": 9.321364768313803e-05, "loss": 0.0139, "step": 20940 }, { "grad_norm": 0.11514464765787125, "learning_rate": 9.32053279848967e-05, "loss": 0.0092, "step": 20950 }, { "grad_norm": 0.19403301179409027, "learning_rate": 9.319700356178567e-05, "loss": 0.0171, "step": 20960 }, { "grad_norm": 0.11661290377378464, "learning_rate": 9.318867441471527e-05, "loss": 0.0128, "step": 20970 }, { "grad_norm": 0.09532665461301804, "learning_rate": 9.318034054459637e-05, "loss": 0.0131, "step": 20980 }, { "grad_norm": 0.12435909360647202, "learning_rate": 9.317200195234034e-05, "loss": 0.0092, "step": 20990 }, { "grad_norm": 0.12808389961719513, "learning_rate": 9.316365863885909e-05, "loss": 0.0121, "step": 21000 }, { "grad_norm": 0.10908785462379456, "learning_rate": 9.315531060506502e-05, "loss": 0.0088, "step": 21010 }, { "grad_norm": 0.15227733552455902, "learning_rate": 9.314695785187108e-05, "loss": 0.015, "step": 21020 }, { "grad_norm": 0.10424824059009552, "learning_rate": 9.313860038019069e-05, "loss": 0.0099, "step": 21030 }, { "grad_norm": 0.13034608960151672, "learning_rate": 9.313023819093782e-05, "loss": 0.0127, "step": 21040 }, { "grad_norm": 0.11397765576839447, "learning_rate": 9.312187128502695e-05, "loss": 0.0144, "step": 21050 }, { "grad_norm": 0.08251112699508667, "learning_rate": 9.311349966337307e-05, "loss": 0.0083, "step": 21060 }, { "grad_norm": 0.1192241758108139, "learning_rate": 9.310512332689169e-05, "loss": 0.0149, "step": 21070 }, { "grad_norm": 0.12197034060955048, "learning_rate": 9.309674227649883e-05, "loss": 0.0113, "step": 21080 }, { "grad_norm": 0.11481665819883347, "learning_rate": 9.308835651311103e-05, "loss": 0.0101, "step": 21090 }, { "grad_norm": 0.102671779692173, "learning_rate": 9.307996603764533e-05, "loss": 0.0105, "step": 21100 }, { "grad_norm": 0.17948363721370697, "learning_rate": 9.307157085101932e-05, "loss": 0.0162, "step": 21110 }, { "grad_norm": 0.1112513393163681, "learning_rate": 9.306317095415109e-05, "loss": 0.0111, "step": 21120 }, { "grad_norm": 0.14036037027835846, "learning_rate": 9.305476634795922e-05, "loss": 0.0127, "step": 21130 }, { "grad_norm": 0.14802710711956024, "learning_rate": 9.304635703336284e-05, "loss": 0.0124, "step": 21140 }, { "grad_norm": 0.2080894112586975, "learning_rate": 9.303794301128157e-05, "loss": 0.0177, "step": 21150 }, { "grad_norm": 0.1172323003411293, "learning_rate": 9.302952428263555e-05, "loss": 0.0118, "step": 21160 }, { "grad_norm": 0.15404029190540314, "learning_rate": 9.302110084834545e-05, "loss": 0.0134, "step": 21170 }, { "grad_norm": 0.1623791605234146, "learning_rate": 9.301267270933245e-05, "loss": 0.013, "step": 21180 }, { "grad_norm": 0.13651925325393677, "learning_rate": 9.300423986651823e-05, "loss": 0.0141, "step": 21190 }, { "grad_norm": 0.2912414073944092, "learning_rate": 9.299580232082501e-05, "loss": 0.0149, "step": 21200 }, { "grad_norm": 0.15002426505088806, "learning_rate": 9.298736007317547e-05, "loss": 0.0121, "step": 21210 }, { "grad_norm": 0.14157196879386902, "learning_rate": 9.297891312449288e-05, "loss": 0.0141, "step": 21220 }, { "grad_norm": 0.18101491034030914, "learning_rate": 9.297046147570094e-05, "loss": 0.0141, "step": 21230 }, { "grad_norm": 0.220697820186615, "learning_rate": 9.296200512772396e-05, "loss": 0.0119, "step": 21240 }, { "grad_norm": 0.11411435157060623, "learning_rate": 9.295354408148668e-05, "loss": 0.0107, "step": 21250 }, { "grad_norm": 0.13013389706611633, "learning_rate": 9.294507833791441e-05, "loss": 0.0118, "step": 21260 }, { "grad_norm": 0.13105487823486328, "learning_rate": 9.293660789793295e-05, "loss": 0.0138, "step": 21270 }, { "grad_norm": 0.143210768699646, "learning_rate": 9.292813276246858e-05, "loss": 0.0099, "step": 21280 }, { "grad_norm": 0.1397722214460373, "learning_rate": 9.291965293244816e-05, "loss": 0.0145, "step": 21290 }, { "grad_norm": 0.13309544324874878, "learning_rate": 9.291116840879904e-05, "loss": 0.0119, "step": 21300 }, { "grad_norm": 0.18346352875232697, "learning_rate": 9.290267919244904e-05, "loss": 0.0137, "step": 21310 }, { "grad_norm": 0.11729457974433899, "learning_rate": 9.289418528432655e-05, "loss": 0.0123, "step": 21320 }, { "grad_norm": 0.09386423975229263, "learning_rate": 9.288568668536045e-05, "loss": 0.008, "step": 21330 }, { "grad_norm": 0.10905823856592178, "learning_rate": 9.287718339648013e-05, "loss": 0.0131, "step": 21340 }, { "grad_norm": 0.10948000848293304, "learning_rate": 9.28686754186155e-05, "loss": 0.0117, "step": 21350 }, { "grad_norm": 0.13726401329040527, "learning_rate": 9.286016275269698e-05, "loss": 0.0109, "step": 21360 }, { "grad_norm": 0.0824316069483757, "learning_rate": 9.285164539965551e-05, "loss": 0.0099, "step": 21370 }, { "grad_norm": 0.1516805738210678, "learning_rate": 9.284312336042251e-05, "loss": 0.01, "step": 21380 }, { "grad_norm": 0.237228125333786, "learning_rate": 9.283459663592996e-05, "loss": 0.0139, "step": 21390 }, { "grad_norm": 0.1810116320848465, "learning_rate": 9.282606522711033e-05, "loss": 0.0119, "step": 21400 }, { "grad_norm": 0.23283782601356506, "learning_rate": 9.281752913489657e-05, "loss": 0.0148, "step": 21410 }, { "grad_norm": 0.1839967519044876, "learning_rate": 9.280898836022222e-05, "loss": 0.0138, "step": 21420 }, { "grad_norm": 0.1509382426738739, "learning_rate": 9.280044290402126e-05, "loss": 0.0112, "step": 21430 }, { "grad_norm": 0.0952736958861351, "learning_rate": 9.279189276722821e-05, "loss": 0.014, "step": 21440 }, { "grad_norm": 0.16072259843349457, "learning_rate": 9.278333795077812e-05, "loss": 0.0176, "step": 21450 }, { "grad_norm": 0.10063473880290985, "learning_rate": 9.27747784556065e-05, "loss": 0.009, "step": 21460 }, { "grad_norm": 0.08509081602096558, "learning_rate": 9.276621428264942e-05, "loss": 0.0089, "step": 21470 }, { "grad_norm": 0.2737087905406952, "learning_rate": 9.275764543284345e-05, "loss": 0.011, "step": 21480 }, { "grad_norm": 0.1317097246646881, "learning_rate": 9.274907190712566e-05, "loss": 0.0114, "step": 21490 }, { "grad_norm": 0.1299164891242981, "learning_rate": 9.274049370643363e-05, "loss": 0.0119, "step": 21500 }, { "grad_norm": 0.16284793615341187, "learning_rate": 9.273191083170547e-05, "loss": 0.0108, "step": 21510 }, { "grad_norm": 0.20837394893169403, "learning_rate": 9.27233232838798e-05, "loss": 0.0135, "step": 21520 }, { "grad_norm": 0.11104953289031982, "learning_rate": 9.27147310638957e-05, "loss": 0.0122, "step": 21530 }, { "grad_norm": 0.1269427090883255, "learning_rate": 9.270613417269286e-05, "loss": 0.012, "step": 21540 }, { "grad_norm": 0.18217070400714874, "learning_rate": 9.269753261121138e-05, "loss": 0.0118, "step": 21550 }, { "grad_norm": 0.13516968488693237, "learning_rate": 9.268892638039194e-05, "loss": 0.0124, "step": 21560 }, { "grad_norm": 0.08990395069122314, "learning_rate": 9.268031548117569e-05, "loss": 0.0092, "step": 21570 }, { "grad_norm": 0.09872375428676605, "learning_rate": 9.26716999145043e-05, "loss": 0.0124, "step": 21580 }, { "grad_norm": 0.12961508333683014, "learning_rate": 9.266307968131998e-05, "loss": 0.011, "step": 21590 }, { "grad_norm": 0.10091212391853333, "learning_rate": 9.26544547825654e-05, "loss": 0.0127, "step": 21600 }, { "grad_norm": 0.2531519830226898, "learning_rate": 9.264582521918376e-05, "loss": 0.0131, "step": 21610 }, { "grad_norm": 0.17339320480823517, "learning_rate": 9.263719099211881e-05, "loss": 0.0145, "step": 21620 }, { "grad_norm": 0.09247047454118729, "learning_rate": 9.262855210231476e-05, "loss": 0.0138, "step": 21630 }, { "grad_norm": 0.1452503204345703, "learning_rate": 9.261990855071633e-05, "loss": 0.0126, "step": 21640 }, { "grad_norm": 0.14293557405471802, "learning_rate": 9.261126033826878e-05, "loss": 0.0132, "step": 21650 }, { "grad_norm": 0.18465636670589447, "learning_rate": 9.260260746591786e-05, "loss": 0.0151, "step": 21660 }, { "grad_norm": 0.20242665708065033, "learning_rate": 9.259394993460985e-05, "loss": 0.0139, "step": 21670 }, { "grad_norm": 0.17232653498649597, "learning_rate": 9.258528774529151e-05, "loss": 0.0144, "step": 21680 }, { "grad_norm": 0.23456458747386932, "learning_rate": 9.257662089891013e-05, "loss": 0.0164, "step": 21690 }, { "grad_norm": 0.13068361580371857, "learning_rate": 9.25679493964135e-05, "loss": 0.0131, "step": 21700 }, { "grad_norm": 0.1571997106075287, "learning_rate": 9.255927323874994e-05, "loss": 0.0144, "step": 21710 }, { "grad_norm": 0.18444646894931793, "learning_rate": 9.255059242686822e-05, "loss": 0.0167, "step": 21720 }, { "grad_norm": 0.12474265694618225, "learning_rate": 9.254190696171769e-05, "loss": 0.0127, "step": 21730 }, { "grad_norm": 0.16263344883918762, "learning_rate": 9.25332168442482e-05, "loss": 0.0122, "step": 21740 }, { "grad_norm": 0.09830310940742493, "learning_rate": 9.252452207541004e-05, "loss": 0.0109, "step": 21750 }, { "grad_norm": 0.5756896138191223, "learning_rate": 9.251582265615409e-05, "loss": 0.0158, "step": 21760 }, { "grad_norm": 0.21380895376205444, "learning_rate": 9.250711858743169e-05, "loss": 0.0184, "step": 21770 }, { "grad_norm": 0.1548900306224823, "learning_rate": 9.24984098701947e-05, "loss": 0.0141, "step": 21780 }, { "grad_norm": 0.6305316090583801, "learning_rate": 9.248969650539552e-05, "loss": 0.0168, "step": 21790 }, { "grad_norm": 0.17716650664806366, "learning_rate": 9.2480978493987e-05, "loss": 0.0157, "step": 21800 }, { "grad_norm": 0.11241883784532547, "learning_rate": 9.247225583692256e-05, "loss": 0.0149, "step": 21810 }, { "grad_norm": 0.10082170367240906, "learning_rate": 9.246352853515607e-05, "loss": 0.0121, "step": 21820 }, { "grad_norm": 0.09201652556657791, "learning_rate": 9.245479658964194e-05, "loss": 0.0119, "step": 21830 }, { "grad_norm": 0.1034126952290535, "learning_rate": 9.244606000133507e-05, "loss": 0.0123, "step": 21840 }, { "grad_norm": 0.3716290593147278, "learning_rate": 9.24373187711909e-05, "loss": 0.0167, "step": 21850 }, { "grad_norm": 0.16836650669574738, "learning_rate": 9.242857290016537e-05, "loss": 0.0137, "step": 21860 }, { "grad_norm": 0.17040832340717316, "learning_rate": 9.241982238921488e-05, "loss": 0.0181, "step": 21870 }, { "grad_norm": 0.1164306104183197, "learning_rate": 9.24110672392964e-05, "loss": 0.0116, "step": 21880 }, { "grad_norm": 0.19410072267055511, "learning_rate": 9.240230745136737e-05, "loss": 0.0185, "step": 21890 }, { "grad_norm": 0.15948373079299927, "learning_rate": 9.239354302638575e-05, "loss": 0.0147, "step": 21900 }, { "grad_norm": 0.13128569722175598, "learning_rate": 9.238477396531e-05, "loss": 0.0114, "step": 21910 }, { "grad_norm": 0.2282831370830536, "learning_rate": 9.23760002690991e-05, "loss": 0.0178, "step": 21920 }, { "grad_norm": 0.14576472342014313, "learning_rate": 9.236722193871252e-05, "loss": 0.0147, "step": 21930 }, { "grad_norm": 0.1967485398054123, "learning_rate": 9.235843897511023e-05, "loss": 0.0172, "step": 21940 }, { "grad_norm": 0.12987935543060303, "learning_rate": 9.234965137925276e-05, "loss": 0.011, "step": 21950 }, { "grad_norm": 0.10306376963853836, "learning_rate": 9.234085915210108e-05, "loss": 0.0114, "step": 21960 }, { "grad_norm": 0.08763039112091064, "learning_rate": 9.23320622946167e-05, "loss": 0.0119, "step": 21970 }, { "grad_norm": 0.11477893590927124, "learning_rate": 9.232326080776163e-05, "loss": 0.0117, "step": 21980 }, { "grad_norm": 0.07728756219148636, "learning_rate": 9.23144546924984e-05, "loss": 0.0098, "step": 21990 }, { "grad_norm": 0.10960038006305695, "learning_rate": 9.230564394979e-05, "loss": 0.0088, "step": 22000 }, { "grad_norm": 0.08677205443382263, "learning_rate": 9.22968285806e-05, "loss": 0.0097, "step": 22010 }, { "grad_norm": 0.108744315803051, "learning_rate": 9.228800858589242e-05, "loss": 0.0098, "step": 22020 }, { "grad_norm": 0.09600085765123367, "learning_rate": 9.227918396663179e-05, "loss": 0.0093, "step": 22030 }, { "grad_norm": 0.10404186695814133, "learning_rate": 9.227035472378319e-05, "loss": 0.0106, "step": 22040 }, { "grad_norm": 0.0988653227686882, "learning_rate": 9.226152085831213e-05, "loss": 0.0093, "step": 22050 }, { "grad_norm": 0.0783657357096672, "learning_rate": 9.22526823711847e-05, "loss": 0.0078, "step": 22060 }, { "grad_norm": 0.15650516748428345, "learning_rate": 9.224383926336745e-05, "loss": 0.0097, "step": 22070 }, { "grad_norm": 0.2808375656604767, "learning_rate": 9.223499153582744e-05, "loss": 0.0137, "step": 22080 }, { "grad_norm": 0.1958950012922287, "learning_rate": 9.222613918953226e-05, "loss": 0.0134, "step": 22090 }, { "grad_norm": 0.10218381881713867, "learning_rate": 9.221728222544999e-05, "loss": 0.0143, "step": 22100 }, { "grad_norm": 0.21736736595630646, "learning_rate": 9.22084206445492e-05, "loss": 0.0125, "step": 22110 }, { "grad_norm": 0.15819883346557617, "learning_rate": 9.2199554447799e-05, "loss": 0.0129, "step": 22120 }, { "grad_norm": 0.10899554938077927, "learning_rate": 9.219068363616897e-05, "loss": 0.016, "step": 22130 }, { "grad_norm": 0.11989667266607285, "learning_rate": 9.218180821062919e-05, "loss": 0.0137, "step": 22140 }, { "grad_norm": 0.15951645374298096, "learning_rate": 9.21729281721503e-05, "loss": 0.0138, "step": 22150 }, { "grad_norm": 0.1791820228099823, "learning_rate": 9.216404352170339e-05, "loss": 0.0132, "step": 22160 }, { "grad_norm": 0.2507162094116211, "learning_rate": 9.215515426026007e-05, "loss": 0.0127, "step": 22170 }, { "grad_norm": 0.1963556706905365, "learning_rate": 9.214626038879246e-05, "loss": 0.0131, "step": 22180 }, { "grad_norm": 0.1488998979330063, "learning_rate": 9.21373619082732e-05, "loss": 0.01, "step": 22190 }, { "grad_norm": 0.25059494376182556, "learning_rate": 9.212845881967535e-05, "loss": 0.0118, "step": 22200 }, { "grad_norm": 0.1522245705127716, "learning_rate": 9.211955112397262e-05, "loss": 0.0129, "step": 22210 }, { "grad_norm": 0.1677144169807434, "learning_rate": 9.211063882213909e-05, "loss": 0.0148, "step": 22220 }, { "grad_norm": 0.14104542136192322, "learning_rate": 9.210172191514942e-05, "loss": 0.011, "step": 22230 }, { "grad_norm": 0.12306071817874908, "learning_rate": 9.209280040397874e-05, "loss": 0.0103, "step": 22240 }, { "grad_norm": 0.21299877762794495, "learning_rate": 9.208387428960268e-05, "loss": 0.0202, "step": 22250 }, { "grad_norm": 0.12000659108161926, "learning_rate": 9.20749435729974e-05, "loss": 0.0143, "step": 22260 }, { "grad_norm": 0.07929424941539764, "learning_rate": 9.206600825513957e-05, "loss": 0.0104, "step": 22270 }, { "grad_norm": 0.09288730472326279, "learning_rate": 9.20570683370063e-05, "loss": 0.0098, "step": 22280 }, { "grad_norm": 0.16478414833545685, "learning_rate": 9.204812381957528e-05, "loss": 0.0144, "step": 22290 }, { "grad_norm": 0.1257760226726532, "learning_rate": 9.203917470382465e-05, "loss": 0.0107, "step": 22300 }, { "grad_norm": 0.10846328735351562, "learning_rate": 9.203022099073309e-05, "loss": 0.0101, "step": 22310 }, { "grad_norm": 0.13590185344219208, "learning_rate": 9.202126268127976e-05, "loss": 0.0128, "step": 22320 }, { "grad_norm": 0.14501436054706573, "learning_rate": 9.20122997764443e-05, "loss": 0.0143, "step": 22330 }, { "grad_norm": 0.12467419356107712, "learning_rate": 9.200333227720692e-05, "loss": 0.011, "step": 22340 }, { "grad_norm": 0.07277674973011017, "learning_rate": 9.199436018454826e-05, "loss": 0.0103, "step": 22350 }, { "grad_norm": 0.21682901680469513, "learning_rate": 9.198538349944952e-05, "loss": 0.0119, "step": 22360 }, { "grad_norm": 0.1311563402414322, "learning_rate": 9.197640222289234e-05, "loss": 0.0123, "step": 22370 }, { "grad_norm": 0.1905561238527298, "learning_rate": 9.196741635585895e-05, "loss": 0.0126, "step": 22380 }, { "grad_norm": 0.16514788568019867, "learning_rate": 9.195842589933199e-05, "loss": 0.0132, "step": 22390 }, { "grad_norm": 0.2718544006347656, "learning_rate": 9.194943085429466e-05, "loss": 0.0145, "step": 22400 }, { "grad_norm": 0.20182783901691437, "learning_rate": 9.194043122173065e-05, "loss": 0.0122, "step": 22410 }, { "grad_norm": 0.13807451725006104, "learning_rate": 9.193142700262413e-05, "loss": 0.0147, "step": 22420 }, { "grad_norm": 0.11995904892683029, "learning_rate": 9.192241819795979e-05, "loss": 0.0124, "step": 22430 }, { "grad_norm": 0.11477921903133392, "learning_rate": 9.191340480872284e-05, "loss": 0.0122, "step": 22440 }, { "grad_norm": 0.12061353772878647, "learning_rate": 9.190438683589895e-05, "loss": 0.0097, "step": 22450 }, { "grad_norm": 0.20550809800624847, "learning_rate": 9.189536428047432e-05, "loss": 0.0117, "step": 22460 }, { "grad_norm": 0.12437774240970612, "learning_rate": 9.188633714343564e-05, "loss": 0.0111, "step": 22470 }, { "grad_norm": 0.11959756165742874, "learning_rate": 9.18773054257701e-05, "loss": 0.0091, "step": 22480 }, { "grad_norm": 0.0972910225391388, "learning_rate": 9.18682691284654e-05, "loss": 0.0084, "step": 22490 }, { "grad_norm": 0.21578764915466309, "learning_rate": 9.185922825250974e-05, "loss": 0.0111, "step": 22500 }, { "grad_norm": 0.13860443234443665, "learning_rate": 9.185018279889181e-05, "loss": 0.0121, "step": 22510 }, { "grad_norm": 0.15886518359184265, "learning_rate": 9.184113276860082e-05, "loss": 0.0112, "step": 22520 }, { "grad_norm": 0.13265903294086456, "learning_rate": 9.183207816262645e-05, "loss": 0.0127, "step": 22530 }, { "grad_norm": 0.14519613981246948, "learning_rate": 9.182301898195891e-05, "loss": 0.0131, "step": 22540 }, { "grad_norm": 0.1122957095503807, "learning_rate": 9.181395522758889e-05, "loss": 0.0106, "step": 22550 }, { "grad_norm": 0.11260323226451874, "learning_rate": 9.180488690050759e-05, "loss": 0.0124, "step": 22560 }, { "grad_norm": 0.18766668438911438, "learning_rate": 9.179581400170671e-05, "loss": 0.0103, "step": 22570 }, { "grad_norm": 0.1535394787788391, "learning_rate": 9.178673653217845e-05, "loss": 0.0099, "step": 22580 }, { "grad_norm": 0.08824606984853745, "learning_rate": 9.177765449291551e-05, "loss": 0.0105, "step": 22590 }, { "grad_norm": 0.11715652793645859, "learning_rate": 9.176856788491109e-05, "loss": 0.0115, "step": 22600 }, { "grad_norm": 0.12226998060941696, "learning_rate": 9.175947670915887e-05, "loss": 0.0094, "step": 22610 }, { "grad_norm": 0.3964008092880249, "learning_rate": 9.175038096665309e-05, "loss": 0.0102, "step": 22620 }, { "grad_norm": 0.24107520282268524, "learning_rate": 9.17412806583884e-05, "loss": 0.0132, "step": 22630 }, { "grad_norm": 0.14399310946464539, "learning_rate": 9.173217578536002e-05, "loss": 0.011, "step": 22640 }, { "grad_norm": 0.48540425300598145, "learning_rate": 9.172306634856362e-05, "loss": 0.0141, "step": 22650 }, { "grad_norm": 0.20208272337913513, "learning_rate": 9.171395234899545e-05, "loss": 0.0158, "step": 22660 }, { "grad_norm": 0.3419220745563507, "learning_rate": 9.170483378765214e-05, "loss": 0.0188, "step": 22670 }, { "grad_norm": 0.2135874480009079, "learning_rate": 9.169571066553091e-05, "loss": 0.016, "step": 22680 }, { "grad_norm": 0.14518676698207855, "learning_rate": 9.168658298362946e-05, "loss": 0.0125, "step": 22690 }, { "grad_norm": 0.14203782379627228, "learning_rate": 9.167745074294598e-05, "loss": 0.0111, "step": 22700 }, { "grad_norm": 0.15089887380599976, "learning_rate": 9.166831394447913e-05, "loss": 0.0179, "step": 22710 }, { "grad_norm": 0.11263056844472885, "learning_rate": 9.165917258922812e-05, "loss": 0.0112, "step": 22720 }, { "grad_norm": 0.07521755248308182, "learning_rate": 9.165002667819262e-05, "loss": 0.0103, "step": 22730 }, { "grad_norm": 0.37041470408439636, "learning_rate": 9.164087621237282e-05, "loss": 0.014, "step": 22740 }, { "grad_norm": 0.15898965299129486, "learning_rate": 9.163172119276942e-05, "loss": 0.0135, "step": 22750 }, { "grad_norm": 0.3491763174533844, "learning_rate": 9.162256162038358e-05, "loss": 0.0114, "step": 22760 }, { "grad_norm": 0.15402700006961823, "learning_rate": 9.161339749621698e-05, "loss": 0.0158, "step": 22770 }, { "grad_norm": 0.1505160927772522, "learning_rate": 9.160422882127177e-05, "loss": 0.0122, "step": 22780 }, { "grad_norm": 0.10224736481904984, "learning_rate": 9.159505559655069e-05, "loss": 0.0106, "step": 22790 }, { "grad_norm": 0.2871202826499939, "learning_rate": 9.158587782305684e-05, "loss": 0.0111, "step": 22800 }, { "grad_norm": 0.1784210205078125, "learning_rate": 9.157669550179391e-05, "loss": 0.0174, "step": 22810 }, { "grad_norm": 0.3775326609611511, "learning_rate": 9.156750863376609e-05, "loss": 0.0119, "step": 22820 }, { "grad_norm": 0.17801722884178162, "learning_rate": 9.155831721997801e-05, "loss": 0.0142, "step": 22830 }, { "grad_norm": 0.2604866921901703, "learning_rate": 9.154912126143484e-05, "loss": 0.0136, "step": 22840 }, { "grad_norm": 0.1488763391971588, "learning_rate": 9.153992075914224e-05, "loss": 0.0143, "step": 22850 }, { "grad_norm": 0.1928025484085083, "learning_rate": 9.153071571410635e-05, "loss": 0.0133, "step": 22860 }, { "grad_norm": 0.12087387591600418, "learning_rate": 9.152150612733384e-05, "loss": 0.0123, "step": 22870 }, { "grad_norm": 0.11041795462369919, "learning_rate": 9.151229199983184e-05, "loss": 0.0115, "step": 22880 }, { "grad_norm": 0.10969806462526321, "learning_rate": 9.150307333260802e-05, "loss": 0.0093, "step": 22890 }, { "grad_norm": 0.11510511487722397, "learning_rate": 9.149385012667048e-05, "loss": 0.0105, "step": 22900 }, { "grad_norm": 0.09224783629179001, "learning_rate": 9.148462238302788e-05, "loss": 0.011, "step": 22910 }, { "grad_norm": 0.16782142221927643, "learning_rate": 9.147539010268936e-05, "loss": 0.013, "step": 22920 }, { "grad_norm": 0.18023009598255157, "learning_rate": 9.14661532866645e-05, "loss": 0.0107, "step": 22930 }, { "grad_norm": 0.1153121292591095, "learning_rate": 9.145691193596348e-05, "loss": 0.0112, "step": 22940 }, { "grad_norm": 0.10171964019536972, "learning_rate": 9.144766605159691e-05, "loss": 0.0122, "step": 22950 }, { "grad_norm": 0.16939079761505127, "learning_rate": 9.14384156345759e-05, "loss": 0.0113, "step": 22960 }, { "grad_norm": 0.11700471490621567, "learning_rate": 9.142916068591204e-05, "loss": 0.0104, "step": 22970 }, { "grad_norm": 0.23367568850517273, "learning_rate": 9.141990120661746e-05, "loss": 0.0113, "step": 22980 }, { "grad_norm": 0.12484566122293472, "learning_rate": 9.141063719770475e-05, "loss": 0.0098, "step": 22990 }, { "grad_norm": 0.10790100693702698, "learning_rate": 9.140136866018704e-05, "loss": 0.0109, "step": 23000 }, { "grad_norm": 0.2233514040708542, "learning_rate": 9.139209559507788e-05, "loss": 0.0117, "step": 23010 }, { "grad_norm": 0.0876663327217102, "learning_rate": 9.13828180033914e-05, "loss": 0.0107, "step": 23020 }, { "grad_norm": 0.3104231655597687, "learning_rate": 9.137353588614212e-05, "loss": 0.0122, "step": 23030 }, { "grad_norm": 0.1286156326532364, "learning_rate": 9.136424924434519e-05, "loss": 0.013, "step": 23040 }, { "grad_norm": 0.1395702213048935, "learning_rate": 9.135495807901615e-05, "loss": 0.0098, "step": 23050 }, { "grad_norm": 0.20998185873031616, "learning_rate": 9.134566239117108e-05, "loss": 0.0134, "step": 23060 }, { "grad_norm": 0.14722765982151031, "learning_rate": 9.13363621818265e-05, "loss": 0.0131, "step": 23070 }, { "grad_norm": 0.11096816509962082, "learning_rate": 9.132705745199953e-05, "loss": 0.0131, "step": 23080 }, { "grad_norm": 0.2505965828895569, "learning_rate": 9.131774820270768e-05, "loss": 0.0136, "step": 23090 }, { "grad_norm": 0.1466771811246872, "learning_rate": 9.130843443496901e-05, "loss": 0.0132, "step": 23100 }, { "grad_norm": 0.20175933837890625, "learning_rate": 9.129911614980206e-05, "loss": 0.0119, "step": 23110 }, { "grad_norm": 0.14963343739509583, "learning_rate": 9.128979334822584e-05, "loss": 0.0099, "step": 23120 }, { "grad_norm": 0.10449140518903732, "learning_rate": 9.128046603125992e-05, "loss": 0.0102, "step": 23130 }, { "grad_norm": 0.16727110743522644, "learning_rate": 9.12711341999243e-05, "loss": 0.0105, "step": 23140 }, { "grad_norm": 0.15241283178329468, "learning_rate": 9.12617978552395e-05, "loss": 0.0134, "step": 23150 }, { "grad_norm": 0.08021757006645203, "learning_rate": 9.12524569982265e-05, "loss": 0.0101, "step": 23160 }, { "grad_norm": 0.0857931599020958, "learning_rate": 9.124311162990684e-05, "loss": 0.0101, "step": 23170 }, { "grad_norm": 0.209361732006073, "learning_rate": 9.12337617513025e-05, "loss": 0.0117, "step": 23180 }, { "grad_norm": 0.14579497277736664, "learning_rate": 9.122440736343596e-05, "loss": 0.0139, "step": 23190 }, { "grad_norm": 0.13280299305915833, "learning_rate": 9.12150484673302e-05, "loss": 0.015, "step": 23200 }, { "grad_norm": 0.09446609765291214, "learning_rate": 9.120568506400873e-05, "loss": 0.0088, "step": 23210 }, { "grad_norm": 0.10336249321699142, "learning_rate": 9.119631715449548e-05, "loss": 0.0082, "step": 23220 }, { "grad_norm": 0.16952340304851532, "learning_rate": 9.118694473981493e-05, "loss": 0.0143, "step": 23230 }, { "grad_norm": 0.14009854197502136, "learning_rate": 9.117756782099203e-05, "loss": 0.0124, "step": 23240 }, { "grad_norm": 0.18449248373508453, "learning_rate": 9.11681863990522e-05, "loss": 0.0119, "step": 23250 }, { "grad_norm": 0.16695807874202728, "learning_rate": 9.115880047502142e-05, "loss": 0.0111, "step": 23260 }, { "grad_norm": 0.11350813508033752, "learning_rate": 9.114941004992609e-05, "loss": 0.01, "step": 23270 }, { "grad_norm": 0.3082275688648224, "learning_rate": 9.114001512479317e-05, "loss": 0.0159, "step": 23280 }, { "grad_norm": 0.11137282103300095, "learning_rate": 9.113061570065003e-05, "loss": 0.0135, "step": 23290 }, { "grad_norm": 0.21933512389659882, "learning_rate": 9.112121177852459e-05, "loss": 0.0139, "step": 23300 }, { "grad_norm": 0.13478043675422668, "learning_rate": 9.111180335944527e-05, "loss": 0.0108, "step": 23310 }, { "grad_norm": 0.15292775630950928, "learning_rate": 9.110239044444093e-05, "loss": 0.0145, "step": 23320 }, { "grad_norm": 0.10775760561227798, "learning_rate": 9.109297303454099e-05, "loss": 0.0095, "step": 23330 }, { "grad_norm": 0.19137251377105713, "learning_rate": 9.108355113077526e-05, "loss": 0.0103, "step": 23340 }, { "grad_norm": 0.14841331541538239, "learning_rate": 9.107412473417419e-05, "loss": 0.0113, "step": 23350 }, { "grad_norm": 0.14027701318264008, "learning_rate": 9.106469384576858e-05, "loss": 0.012, "step": 23360 }, { "grad_norm": 0.08072340488433838, "learning_rate": 9.105525846658978e-05, "loss": 0.0105, "step": 23370 }, { "grad_norm": 0.14110764861106873, "learning_rate": 9.104581859766965e-05, "loss": 0.0151, "step": 23380 }, { "grad_norm": 0.11765013635158539, "learning_rate": 9.10363742400405e-05, "loss": 0.0106, "step": 23390 }, { "grad_norm": 0.18887615203857422, "learning_rate": 9.102692539473518e-05, "loss": 0.0121, "step": 23400 }, { "grad_norm": 0.09920590370893478, "learning_rate": 9.101747206278697e-05, "loss": 0.0119, "step": 23410 }, { "grad_norm": 0.09906342625617981, "learning_rate": 9.100801424522968e-05, "loss": 0.0136, "step": 23420 }, { "grad_norm": 0.21362943947315216, "learning_rate": 9.099855194309762e-05, "loss": 0.0148, "step": 23430 }, { "grad_norm": 0.1479460895061493, "learning_rate": 9.098908515742554e-05, "loss": 0.0121, "step": 23440 }, { "grad_norm": 0.2677358388900757, "learning_rate": 9.097961388924873e-05, "loss": 0.0142, "step": 23450 }, { "grad_norm": 0.15315161645412445, "learning_rate": 9.097013813960298e-05, "loss": 0.0113, "step": 23460 }, { "grad_norm": 0.1332314908504486, "learning_rate": 9.09606579095245e-05, "loss": 0.0107, "step": 23470 }, { "grad_norm": 0.12607933580875397, "learning_rate": 9.095117320005008e-05, "loss": 0.0093, "step": 23480 }, { "grad_norm": 0.09161700308322906, "learning_rate": 9.094168401221691e-05, "loss": 0.0076, "step": 23490 }, { "grad_norm": 0.13299508392810822, "learning_rate": 9.093219034706273e-05, "loss": 0.0108, "step": 23500 }, { "grad_norm": 0.21827886998653412, "learning_rate": 9.092269220562577e-05, "loss": 0.0138, "step": 23510 }, { "grad_norm": 0.16078846156597137, "learning_rate": 9.09131895889447e-05, "loss": 0.0136, "step": 23520 }, { "grad_norm": 0.26539260149002075, "learning_rate": 9.090368249805873e-05, "loss": 0.0137, "step": 23530 }, { "grad_norm": 0.17330330610275269, "learning_rate": 9.089417093400754e-05, "loss": 0.0142, "step": 23540 }, { "grad_norm": 0.13717268407344818, "learning_rate": 9.088465489783131e-05, "loss": 0.0135, "step": 23550 }, { "grad_norm": 0.12035664170980453, "learning_rate": 9.087513439057068e-05, "loss": 0.009, "step": 23560 }, { "grad_norm": 0.095186248421669, "learning_rate": 9.08656094132668e-05, "loss": 0.0114, "step": 23570 }, { "grad_norm": 0.2215002030134201, "learning_rate": 9.085607996696134e-05, "loss": 0.0141, "step": 23580 }, { "grad_norm": 0.1257324367761612, "learning_rate": 9.084654605269639e-05, "loss": 0.0122, "step": 23590 }, { "grad_norm": 0.16960377991199493, "learning_rate": 9.083700767151457e-05, "loss": 0.0105, "step": 23600 }, { "grad_norm": 0.18389073014259338, "learning_rate": 9.082746482445898e-05, "loss": 0.0124, "step": 23610 }, { "grad_norm": 0.16509920358657837, "learning_rate": 9.081791751257325e-05, "loss": 0.01, "step": 23620 }, { "grad_norm": 0.13167475163936615, "learning_rate": 9.080836573690142e-05, "loss": 0.0099, "step": 23630 }, { "grad_norm": 0.1499357521533966, "learning_rate": 9.079880949848805e-05, "loss": 0.0084, "step": 23640 }, { "grad_norm": 0.12997111678123474, "learning_rate": 9.078924879837822e-05, "loss": 0.0082, "step": 23650 }, { "grad_norm": 0.08794985711574554, "learning_rate": 9.077968363761747e-05, "loss": 0.0077, "step": 23660 }, { "grad_norm": 0.14064382016658783, "learning_rate": 9.077011401725182e-05, "loss": 0.0111, "step": 23670 }, { "grad_norm": 0.08737795799970627, "learning_rate": 9.07605399383278e-05, "loss": 0.0088, "step": 23680 }, { "grad_norm": 0.1699075847864151, "learning_rate": 9.075096140189243e-05, "loss": 0.0109, "step": 23690 }, { "grad_norm": 0.11524952948093414, "learning_rate": 9.074137840899318e-05, "loss": 0.0099, "step": 23700 }, { "grad_norm": 0.12808072566986084, "learning_rate": 9.073179096067804e-05, "loss": 0.0085, "step": 23710 }, { "grad_norm": 0.11777198314666748, "learning_rate": 9.072219905799549e-05, "loss": 0.0099, "step": 23720 }, { "grad_norm": 0.08984674513339996, "learning_rate": 9.071260270199447e-05, "loss": 0.0078, "step": 23730 }, { "grad_norm": 0.08660012483596802, "learning_rate": 9.070300189372441e-05, "loss": 0.0103, "step": 23740 }, { "grad_norm": 0.11556016653776169, "learning_rate": 9.069339663423528e-05, "loss": 0.0087, "step": 23750 }, { "grad_norm": 0.12705574929714203, "learning_rate": 9.068378692457747e-05, "loss": 0.011, "step": 23760 }, { "grad_norm": 0.11904217302799225, "learning_rate": 9.067417276580189e-05, "loss": 0.0096, "step": 23770 }, { "grad_norm": 0.17314782738685608, "learning_rate": 9.066455415895993e-05, "loss": 0.0119, "step": 23780 }, { "grad_norm": 0.12925468385219574, "learning_rate": 9.065493110510346e-05, "loss": 0.0109, "step": 23790 }, { "grad_norm": 0.24517367780208588, "learning_rate": 9.064530360528484e-05, "loss": 0.0135, "step": 23800 }, { "grad_norm": 0.15244314074516296, "learning_rate": 9.063567166055695e-05, "loss": 0.0114, "step": 23810 }, { "grad_norm": 0.24155791103839874, "learning_rate": 9.062603527197308e-05, "loss": 0.0164, "step": 23820 }, { "grad_norm": 0.10917861014604568, "learning_rate": 9.06163944405871e-05, "loss": 0.0094, "step": 23830 }, { "grad_norm": 0.11629422754049301, "learning_rate": 9.060674916745327e-05, "loss": 0.0107, "step": 23840 }, { "grad_norm": 0.07450402528047562, "learning_rate": 9.05970994536264e-05, "loss": 0.0108, "step": 23850 }, { "grad_norm": 0.22960780560970306, "learning_rate": 9.05874453001618e-05, "loss": 0.0118, "step": 23860 }, { "grad_norm": 0.12209855765104294, "learning_rate": 9.057778670811517e-05, "loss": 0.0112, "step": 23870 }, { "grad_norm": 0.1018260270357132, "learning_rate": 9.056812367854281e-05, "loss": 0.0088, "step": 23880 }, { "grad_norm": 0.11835982650518417, "learning_rate": 9.055845621250143e-05, "loss": 0.0111, "step": 23890 }, { "grad_norm": 0.1673738956451416, "learning_rate": 9.054878431104825e-05, "loss": 0.0126, "step": 23900 }, { "grad_norm": 0.1411256045103073, "learning_rate": 9.0539107975241e-05, "loss": 0.0104, "step": 23910 }, { "grad_norm": 0.22881463170051575, "learning_rate": 9.052942720613784e-05, "loss": 0.0129, "step": 23920 }, { "grad_norm": 0.2116701751947403, "learning_rate": 9.051974200479745e-05, "loss": 0.0094, "step": 23930 }, { "grad_norm": 0.1301209032535553, "learning_rate": 9.051005237227901e-05, "loss": 0.0088, "step": 23940 }, { "grad_norm": 0.22268453240394592, "learning_rate": 9.050035830964215e-05, "loss": 0.0127, "step": 23950 }, { "grad_norm": 0.2494148463010788, "learning_rate": 9.049065981794698e-05, "loss": 0.0121, "step": 23960 }, { "grad_norm": 0.10862322151660919, "learning_rate": 9.048095689825414e-05, "loss": 0.0108, "step": 23970 }, { "grad_norm": 0.1796126812696457, "learning_rate": 9.047124955162472e-05, "loss": 0.0094, "step": 23980 }, { "grad_norm": 0.11259521543979645, "learning_rate": 9.046153777912028e-05, "loss": 0.013, "step": 23990 }, { "grad_norm": 0.30684587359428406, "learning_rate": 9.045182158180292e-05, "loss": 0.0127, "step": 24000 }, { "grad_norm": 0.1366564929485321, "learning_rate": 9.044210096073516e-05, "loss": 0.0114, "step": 24010 }, { "grad_norm": 0.1152743250131607, "learning_rate": 9.043237591698004e-05, "loss": 0.0115, "step": 24020 }, { "grad_norm": 0.11227177083492279, "learning_rate": 9.04226464516011e-05, "loss": 0.0093, "step": 24030 }, { "grad_norm": 0.12367116659879684, "learning_rate": 9.041291256566229e-05, "loss": 0.0137, "step": 24040 }, { "grad_norm": 0.24333779513835907, "learning_rate": 9.040317426022814e-05, "loss": 0.012, "step": 24050 }, { "grad_norm": 0.13494263589382172, "learning_rate": 9.03934315363636e-05, "loss": 0.0144, "step": 24060 }, { "grad_norm": 0.13272693753242493, "learning_rate": 9.038368439513409e-05, "loss": 0.0097, "step": 24070 }, { "grad_norm": 0.1213563084602356, "learning_rate": 9.03739328376056e-05, "loss": 0.0125, "step": 24080 }, { "grad_norm": 0.17166653275489807, "learning_rate": 9.036417686484451e-05, "loss": 0.0153, "step": 24090 }, { "grad_norm": 0.12130413949489594, "learning_rate": 9.035441647791773e-05, "loss": 0.0122, "step": 24100 }, { "grad_norm": 0.34143441915512085, "learning_rate": 9.034465167789263e-05, "loss": 0.0147, "step": 24110 }, { "grad_norm": 0.15495911240577698, "learning_rate": 9.033488246583706e-05, "loss": 0.0136, "step": 24120 }, { "grad_norm": 0.11847115308046341, "learning_rate": 9.032510884281941e-05, "loss": 0.0093, "step": 24130 }, { "grad_norm": 0.24213920533657074, "learning_rate": 9.031533080990848e-05, "loss": 0.0124, "step": 24140 }, { "grad_norm": 0.16330085694789886, "learning_rate": 9.030554836817358e-05, "loss": 0.0124, "step": 24150 }, { "grad_norm": 0.1199079379439354, "learning_rate": 9.029576151868451e-05, "loss": 0.0112, "step": 24160 }, { "grad_norm": 0.1062031164765358, "learning_rate": 9.028597026251155e-05, "loss": 0.0095, "step": 24170 }, { "grad_norm": 0.13260813057422638, "learning_rate": 9.027617460072547e-05, "loss": 0.0152, "step": 24180 }, { "grad_norm": 0.1418304592370987, "learning_rate": 9.026637453439745e-05, "loss": 0.0128, "step": 24190 }, { "grad_norm": 0.10080551356077194, "learning_rate": 9.025657006459927e-05, "loss": 0.0091, "step": 24200 }, { "grad_norm": 0.24532882869243622, "learning_rate": 9.024676119240311e-05, "loss": 0.0141, "step": 24210 }, { "grad_norm": 0.17517763376235962, "learning_rate": 9.023694791888166e-05, "loss": 0.0165, "step": 24220 }, { "grad_norm": 0.16072486340999603, "learning_rate": 9.022713024510808e-05, "loss": 0.0142, "step": 24230 }, { "grad_norm": 0.1277373731136322, "learning_rate": 9.021730817215601e-05, "loss": 0.0098, "step": 24240 }, { "grad_norm": 0.26368939876556396, "learning_rate": 9.02074817010996e-05, "loss": 0.0108, "step": 24250 }, { "grad_norm": 0.15819425880908966, "learning_rate": 9.019765083301342e-05, "loss": 0.0112, "step": 24260 }, { "grad_norm": 0.1359773874282837, "learning_rate": 9.01878155689726e-05, "loss": 0.0099, "step": 24270 }, { "grad_norm": 0.08336159586906433, "learning_rate": 9.017797591005268e-05, "loss": 0.0102, "step": 24280 }, { "grad_norm": 0.0957246720790863, "learning_rate": 9.016813185732972e-05, "loss": 0.0133, "step": 24290 }, { "grad_norm": 0.09962613135576248, "learning_rate": 9.015828341188027e-05, "loss": 0.011, "step": 24300 }, { "grad_norm": 0.08751974999904633, "learning_rate": 9.01484305747813e-05, "loss": 0.0102, "step": 24310 }, { "grad_norm": 0.09694404900074005, "learning_rate": 9.013857334711033e-05, "loss": 0.0095, "step": 24320 }, { "grad_norm": 0.2165384143590927, "learning_rate": 9.012871172994534e-05, "loss": 0.0122, "step": 24330 }, { "grad_norm": 0.14045023918151855, "learning_rate": 9.011884572436476e-05, "loss": 0.0127, "step": 24340 }, { "grad_norm": 0.11856695264577866, "learning_rate": 9.010897533144754e-05, "loss": 0.0096, "step": 24350 }, { "grad_norm": 0.08690010756254196, "learning_rate": 9.009910055227306e-05, "loss": 0.0116, "step": 24360 }, { "grad_norm": 0.1000102311372757, "learning_rate": 9.008922138792124e-05, "loss": 0.0123, "step": 24370 }, { "grad_norm": 0.15848764777183533, "learning_rate": 9.007933783947244e-05, "loss": 0.0115, "step": 24380 }, { "grad_norm": 0.19676902890205383, "learning_rate": 9.006944990800752e-05, "loss": 0.0131, "step": 24390 }, { "grad_norm": 0.1365555077791214, "learning_rate": 9.005955759460779e-05, "loss": 0.0111, "step": 24400 }, { "grad_norm": 0.134940966963768, "learning_rate": 9.004966090035508e-05, "loss": 0.0119, "step": 24410 }, { "grad_norm": 0.1420154869556427, "learning_rate": 9.003975982633166e-05, "loss": 0.0112, "step": 24420 }, { "grad_norm": 0.11723940074443817, "learning_rate": 9.00298543736203e-05, "loss": 0.0124, "step": 24430 }, { "grad_norm": 0.1228141337633133, "learning_rate": 9.001994454330427e-05, "loss": 0.0115, "step": 24440 }, { "grad_norm": 0.23800642788410187, "learning_rate": 9.001003033646727e-05, "loss": 0.012, "step": 24450 }, { "grad_norm": 0.1143733561038971, "learning_rate": 9.00001117541935e-05, "loss": 0.0094, "step": 24460 }, { "grad_norm": 0.11553522944450378, "learning_rate": 8.999018879756764e-05, "loss": 0.0091, "step": 24470 }, { "grad_norm": 0.1215766966342926, "learning_rate": 8.998026146767487e-05, "loss": 0.0101, "step": 24480 }, { "grad_norm": 0.19761645793914795, "learning_rate": 8.99703297656008e-05, "loss": 0.0138, "step": 24490 }, { "grad_norm": 0.13745588064193726, "learning_rate": 8.996039369243156e-05, "loss": 0.0137, "step": 24500 }, { "grad_norm": 0.10307708382606506, "learning_rate": 8.995045324925378e-05, "loss": 0.0111, "step": 24510 }, { "grad_norm": 0.09031850099563599, "learning_rate": 8.994050843715448e-05, "loss": 0.0107, "step": 24520 }, { "grad_norm": 0.11480671167373657, "learning_rate": 8.993055925722121e-05, "loss": 0.0105, "step": 24530 }, { "grad_norm": 0.13747093081474304, "learning_rate": 8.992060571054202e-05, "loss": 0.0144, "step": 24540 }, { "grad_norm": 0.15488500893115997, "learning_rate": 8.991064779820542e-05, "loss": 0.0117, "step": 24550 }, { "grad_norm": 0.22294272482395172, "learning_rate": 8.990068552130036e-05, "loss": 0.013, "step": 24560 }, { "grad_norm": 0.11436856538057327, "learning_rate": 8.989071888091634e-05, "loss": 0.014, "step": 24570 }, { "grad_norm": 0.18905378878116608, "learning_rate": 8.988074787814329e-05, "loss": 0.0122, "step": 24580 }, { "grad_norm": 0.10102719813585281, "learning_rate": 8.987077251407158e-05, "loss": 0.0092, "step": 24590 }, { "grad_norm": 0.20500347018241882, "learning_rate": 8.986079278979216e-05, "loss": 0.0172, "step": 24600 }, { "grad_norm": 0.10005241632461548, "learning_rate": 8.985080870639635e-05, "loss": 0.0112, "step": 24610 }, { "grad_norm": 0.11584705114364624, "learning_rate": 8.984082026497603e-05, "loss": 0.0114, "step": 24620 }, { "grad_norm": 0.11159808188676834, "learning_rate": 8.98308274666235e-05, "loss": 0.0084, "step": 24630 }, { "grad_norm": 0.11753292381763458, "learning_rate": 8.982083031243155e-05, "loss": 0.0084, "step": 24640 }, { "grad_norm": 0.13508085906505585, "learning_rate": 8.98108288034935e-05, "loss": 0.008, "step": 24650 }, { "grad_norm": 0.19365471601486206, "learning_rate": 8.980082294090305e-05, "loss": 0.0128, "step": 24660 }, { "grad_norm": 0.11921604722738266, "learning_rate": 8.979081272575443e-05, "loss": 0.0104, "step": 24670 }, { "grad_norm": 0.18427255749702454, "learning_rate": 8.978079815914236e-05, "loss": 0.0132, "step": 24680 }, { "grad_norm": 0.11787468940019608, "learning_rate": 8.977077924216202e-05, "loss": 0.0132, "step": 24690 }, { "grad_norm": 0.08493762463331223, "learning_rate": 8.976075597590905e-05, "loss": 0.0093, "step": 24700 }, { "grad_norm": 0.10805793851613998, "learning_rate": 8.975072836147958e-05, "loss": 0.0078, "step": 24710 }, { "grad_norm": 0.1319517344236374, "learning_rate": 8.974069639997025e-05, "loss": 0.0117, "step": 24720 }, { "grad_norm": 0.14684049785137177, "learning_rate": 8.973066009247808e-05, "loss": 0.0123, "step": 24730 }, { "grad_norm": 0.12218012660741806, "learning_rate": 8.972061944010066e-05, "loss": 0.0104, "step": 24740 }, { "grad_norm": 0.14802275598049164, "learning_rate": 8.971057444393603e-05, "loss": 0.0128, "step": 24750 }, { "grad_norm": 0.10733328759670258, "learning_rate": 8.970052510508268e-05, "loss": 0.013, "step": 24760 }, { "grad_norm": 0.076633021235466, "learning_rate": 8.969047142463959e-05, "loss": 0.0096, "step": 24770 }, { "grad_norm": 0.13234427571296692, "learning_rate": 8.968041340370621e-05, "loss": 0.0092, "step": 24780 }, { "grad_norm": 0.18169862031936646, "learning_rate": 8.96703510433825e-05, "loss": 0.0123, "step": 24790 }, { "grad_norm": 0.1321493685245514, "learning_rate": 8.966028434476883e-05, "loss": 0.0096, "step": 24800 }, { "grad_norm": 0.14415998756885529, "learning_rate": 8.96502133089661e-05, "loss": 0.0109, "step": 24810 }, { "grad_norm": 0.1415484994649887, "learning_rate": 8.964013793707564e-05, "loss": 0.0087, "step": 24820 }, { "grad_norm": 0.1398189216852188, "learning_rate": 8.963005823019932e-05, "loss": 0.0083, "step": 24830 }, { "grad_norm": 0.09091292321681976, "learning_rate": 8.961997418943939e-05, "loss": 0.0074, "step": 24840 }, { "grad_norm": 0.130151629447937, "learning_rate": 8.960988581589865e-05, "loss": 0.0113, "step": 24850 }, { "grad_norm": 0.09221066534519196, "learning_rate": 8.959979311068037e-05, "loss": 0.0082, "step": 24860 }, { "grad_norm": 0.13144254684448242, "learning_rate": 8.958969607488823e-05, "loss": 0.0112, "step": 24870 }, { "grad_norm": 0.13768424093723297, "learning_rate": 8.957959470962647e-05, "loss": 0.0101, "step": 24880 }, { "grad_norm": 0.12184949219226837, "learning_rate": 8.956948901599971e-05, "loss": 0.0102, "step": 24890 }, { "grad_norm": 0.11999748647212982, "learning_rate": 8.955937899511315e-05, "loss": 0.012, "step": 24900 }, { "grad_norm": 0.09718360751867294, "learning_rate": 8.954926464807238e-05, "loss": 0.0101, "step": 24910 }, { "grad_norm": 0.1365094631910324, "learning_rate": 8.953914597598347e-05, "loss": 0.0129, "step": 24920 }, { "grad_norm": 0.1601763814687729, "learning_rate": 8.952902297995303e-05, "loss": 0.0114, "step": 24930 }, { "grad_norm": 0.11814699321985245, "learning_rate": 8.951889566108804e-05, "loss": 0.0102, "step": 24940 }, { "grad_norm": 0.10175842046737671, "learning_rate": 8.950876402049606e-05, "loss": 0.0096, "step": 24950 }, { "grad_norm": 0.16797217726707458, "learning_rate": 8.949862805928504e-05, "loss": 0.0124, "step": 24960 }, { "grad_norm": 0.14442218840122223, "learning_rate": 8.948848777856343e-05, "loss": 0.0102, "step": 24970 }, { "grad_norm": 0.12641246616840363, "learning_rate": 8.947834317944017e-05, "loss": 0.0132, "step": 24980 }, { "grad_norm": 0.07513046264648438, "learning_rate": 8.946819426302466e-05, "loss": 0.0104, "step": 24990 }, { "grad_norm": 0.08781909197568893, "learning_rate": 8.945804103042676e-05, "loss": 0.0094, "step": 25000 }, { "grad_norm": 0.0802912563085556, "learning_rate": 8.944788348275681e-05, "loss": 0.0089, "step": 25010 }, { "grad_norm": 0.0987941324710846, "learning_rate": 8.943772162112565e-05, "loss": 0.0093, "step": 25020 }, { "grad_norm": 0.13854442536830902, "learning_rate": 8.942755544664454e-05, "loss": 0.0111, "step": 25030 }, { "grad_norm": 0.08432289212942123, "learning_rate": 8.941738496042525e-05, "loss": 0.0093, "step": 25040 }, { "grad_norm": 0.08007049560546875, "learning_rate": 8.940721016357999e-05, "loss": 0.0104, "step": 25050 }, { "grad_norm": 0.23849640786647797, "learning_rate": 8.939703105722148e-05, "loss": 0.013, "step": 25060 }, { "grad_norm": 0.09030547738075256, "learning_rate": 8.93868476424629e-05, "loss": 0.0135, "step": 25070 }, { "grad_norm": 0.18542131781578064, "learning_rate": 8.937665992041786e-05, "loss": 0.0109, "step": 25080 }, { "grad_norm": 0.16508662700653076, "learning_rate": 8.93664678922005e-05, "loss": 0.0095, "step": 25090 }, { "grad_norm": 0.1289587914943695, "learning_rate": 8.93562715589254e-05, "loss": 0.0093, "step": 25100 }, { "grad_norm": 0.1983964741230011, "learning_rate": 8.934607092170762e-05, "loss": 0.0127, "step": 25110 }, { "grad_norm": 0.3284700810909271, "learning_rate": 8.933586598166266e-05, "loss": 0.0135, "step": 25120 }, { "grad_norm": 0.15019553899765015, "learning_rate": 8.932565673990655e-05, "loss": 0.0117, "step": 25130 }, { "grad_norm": 0.1323680430650711, "learning_rate": 8.931544319755574e-05, "loss": 0.0114, "step": 25140 }, { "grad_norm": 0.3510649800300598, "learning_rate": 8.930522535572718e-05, "loss": 0.0109, "step": 25150 }, { "grad_norm": 0.1573914736509323, "learning_rate": 8.929500321553826e-05, "loss": 0.012, "step": 25160 }, { "grad_norm": 0.21936766803264618, "learning_rate": 8.928477677810686e-05, "loss": 0.0104, "step": 25170 }, { "grad_norm": 0.1439429372549057, "learning_rate": 8.927454604455137e-05, "loss": 0.0136, "step": 25180 }, { "grad_norm": 0.11577156186103821, "learning_rate": 8.926431101599053e-05, "loss": 0.0104, "step": 25190 }, { "grad_norm": 0.10581266134977341, "learning_rate": 8.925407169354369e-05, "loss": 0.0084, "step": 25200 }, { "grad_norm": 0.1374863237142563, "learning_rate": 8.92438280783306e-05, "loss": 0.0111, "step": 25210 }, { "grad_norm": 0.17735086381435394, "learning_rate": 8.923358017147146e-05, "loss": 0.0134, "step": 25220 }, { "grad_norm": 0.17311999201774597, "learning_rate": 8.922332797408697e-05, "loss": 0.0118, "step": 25230 }, { "grad_norm": 0.1803336888551712, "learning_rate": 8.921307148729831e-05, "loss": 0.0118, "step": 25240 }, { "grad_norm": 0.1590627282857895, "learning_rate": 8.920281071222712e-05, "loss": 0.0104, "step": 25250 }, { "grad_norm": 0.11694687604904175, "learning_rate": 8.919254564999548e-05, "loss": 0.0119, "step": 25260 }, { "grad_norm": 0.17058473825454712, "learning_rate": 8.918227630172598e-05, "loss": 0.0125, "step": 25270 }, { "grad_norm": 0.19615763425827026, "learning_rate": 8.917200266854165e-05, "loss": 0.0109, "step": 25280 }, { "grad_norm": 0.13576793670654297, "learning_rate": 8.9161724751566e-05, "loss": 0.012, "step": 25290 }, { "grad_norm": 0.12634135782718658, "learning_rate": 8.915144255192302e-05, "loss": 0.0122, "step": 25300 }, { "grad_norm": 0.12391111254692078, "learning_rate": 8.914115607073714e-05, "loss": 0.0111, "step": 25310 }, { "grad_norm": 0.09817773848772049, "learning_rate": 8.913086530913327e-05, "loss": 0.0104, "step": 25320 }, { "grad_norm": 0.11635806411504745, "learning_rate": 8.912057026823681e-05, "loss": 0.0101, "step": 25330 }, { "grad_norm": 0.08069764077663422, "learning_rate": 8.91102709491736e-05, "loss": 0.01, "step": 25340 }, { "grad_norm": 0.07907871901988983, "learning_rate": 8.909996735306996e-05, "loss": 0.0078, "step": 25350 }, { "grad_norm": 0.11933977901935577, "learning_rate": 8.908965948105268e-05, "loss": 0.009, "step": 25360 }, { "grad_norm": 0.17108213901519775, "learning_rate": 8.907934733424901e-05, "loss": 0.0115, "step": 25370 }, { "grad_norm": 0.10232669860124588, "learning_rate": 8.906903091378666e-05, "loss": 0.0094, "step": 25380 }, { "grad_norm": 0.09573142230510712, "learning_rate": 8.905871022079384e-05, "loss": 0.0093, "step": 25390 }, { "grad_norm": 0.1390911042690277, "learning_rate": 8.90483852563992e-05, "loss": 0.0144, "step": 25400 }, { "grad_norm": 0.15773296356201172, "learning_rate": 8.903805602173185e-05, "loss": 0.0148, "step": 25410 }, { "grad_norm": 0.1706075370311737, "learning_rate": 8.902772251792137e-05, "loss": 0.0154, "step": 25420 }, { "grad_norm": 0.14805464446544647, "learning_rate": 8.901738474609786e-05, "loss": 0.0097, "step": 25430 }, { "grad_norm": 0.10127298533916473, "learning_rate": 8.900704270739179e-05, "loss": 0.0103, "step": 25440 }, { "grad_norm": 0.0718235969543457, "learning_rate": 8.89966964029342e-05, "loss": 0.0065, "step": 25450 }, { "grad_norm": 0.14339934289455414, "learning_rate": 8.898634583385652e-05, "loss": 0.0101, "step": 25460 }, { "grad_norm": 0.0848870798945427, "learning_rate": 8.897599100129065e-05, "loss": 0.012, "step": 25470 }, { "grad_norm": 0.17269279062747955, "learning_rate": 8.896563190636903e-05, "loss": 0.0147, "step": 25480 }, { "grad_norm": 0.11021239310503006, "learning_rate": 8.895526855022448e-05, "loss": 0.0116, "step": 25490 }, { "grad_norm": 0.1933101862668991, "learning_rate": 8.894490093399033e-05, "loss": 0.0128, "step": 25500 }, { "grad_norm": 0.13948285579681396, "learning_rate": 8.893452905880035e-05, "loss": 0.0106, "step": 25510 }, { "grad_norm": 0.1192912757396698, "learning_rate": 8.892415292578883e-05, "loss": 0.0119, "step": 25520 }, { "grad_norm": 0.19362463057041168, "learning_rate": 8.891377253609046e-05, "loss": 0.0137, "step": 25530 }, { "grad_norm": 0.15470516681671143, "learning_rate": 8.890338789084043e-05, "loss": 0.0133, "step": 25540 }, { "grad_norm": 0.08902993053197861, "learning_rate": 8.88929989911744e-05, "loss": 0.0073, "step": 25550 }, { "grad_norm": 0.10222289711236954, "learning_rate": 8.888260583822847e-05, "loss": 0.0084, "step": 25560 }, { "grad_norm": 0.14883849024772644, "learning_rate": 8.887220843313921e-05, "loss": 0.0131, "step": 25570 }, { "grad_norm": 0.14930854737758636, "learning_rate": 8.88618067770437e-05, "loss": 0.0117, "step": 25580 }, { "grad_norm": 0.14176739752292633, "learning_rate": 8.885140087107942e-05, "loss": 0.0152, "step": 25590 }, { "grad_norm": 0.22143249213695526, "learning_rate": 8.884099071638436e-05, "loss": 0.0152, "step": 25600 }, { "grad_norm": 0.17415723204612732, "learning_rate": 8.883057631409695e-05, "loss": 0.0111, "step": 25610 }, { "grad_norm": 0.20560568571090698, "learning_rate": 8.882015766535608e-05, "loss": 0.0127, "step": 25620 }, { "grad_norm": 0.15995453298091888, "learning_rate": 8.880973477130115e-05, "loss": 0.0126, "step": 25630 }, { "grad_norm": 0.11352657526731491, "learning_rate": 8.879930763307197e-05, "loss": 0.0102, "step": 25640 }, { "grad_norm": 0.13067245483398438, "learning_rate": 8.878887625180884e-05, "loss": 0.0102, "step": 25650 }, { "grad_norm": 0.19243992865085602, "learning_rate": 8.877844062865253e-05, "loss": 0.0119, "step": 25660 }, { "grad_norm": 0.14417234063148499, "learning_rate": 8.876800076474424e-05, "loss": 0.0146, "step": 25670 }, { "grad_norm": 0.20220695436000824, "learning_rate": 8.875755666122568e-05, "loss": 0.0138, "step": 25680 }, { "grad_norm": 0.1624273955821991, "learning_rate": 8.8747108319239e-05, "loss": 0.0087, "step": 25690 }, { "grad_norm": 0.13897652924060822, "learning_rate": 8.87366557399268e-05, "loss": 0.0106, "step": 25700 }, { "grad_norm": 0.18140138685703278, "learning_rate": 8.872619892443217e-05, "loss": 0.0155, "step": 25710 }, { "grad_norm": 0.14691957831382751, "learning_rate": 8.871573787389865e-05, "loss": 0.0093, "step": 25720 }, { "grad_norm": 0.09888259321451187, "learning_rate": 8.870527258947024e-05, "loss": 0.0123, "step": 25730 }, { "grad_norm": 0.1476990431547165, "learning_rate": 8.869480307229143e-05, "loss": 0.0124, "step": 25740 }, { "grad_norm": 0.12220276147127151, "learning_rate": 8.868432932350712e-05, "loss": 0.0117, "step": 25750 }, { "grad_norm": 0.20579393208026886, "learning_rate": 8.867385134426272e-05, "loss": 0.0131, "step": 25760 }, { "grad_norm": 0.11515354365110397, "learning_rate": 8.866336913570407e-05, "loss": 0.0108, "step": 25770 }, { "grad_norm": 0.21271838247776031, "learning_rate": 8.865288269897751e-05, "loss": 0.0117, "step": 25780 }, { "grad_norm": 0.25416383147239685, "learning_rate": 8.864239203522981e-05, "loss": 0.0126, "step": 25790 }, { "grad_norm": 0.14566253125667572, "learning_rate": 8.863189714560822e-05, "loss": 0.0097, "step": 25800 }, { "grad_norm": 0.31238773465156555, "learning_rate": 8.862139803126043e-05, "loss": 0.0116, "step": 25810 }, { "grad_norm": 0.14244459569454193, "learning_rate": 8.861089469333463e-05, "loss": 0.0115, "step": 25820 }, { "grad_norm": 0.10359611362218857, "learning_rate": 8.860038713297944e-05, "loss": 0.0124, "step": 25830 }, { "grad_norm": 0.22033758461475372, "learning_rate": 8.858987535134394e-05, "loss": 0.0106, "step": 25840 }, { "grad_norm": 0.10456909239292145, "learning_rate": 8.857935934957769e-05, "loss": 0.0129, "step": 25850 }, { "grad_norm": 0.14453764259815216, "learning_rate": 8.856883912883071e-05, "loss": 0.0135, "step": 25860 }, { "grad_norm": 0.08488878607749939, "learning_rate": 8.855831469025346e-05, "loss": 0.008, "step": 25870 }, { "grad_norm": 0.08421464264392853, "learning_rate": 8.854778603499689e-05, "loss": 0.0131, "step": 25880 }, { "grad_norm": 0.10002200305461884, "learning_rate": 8.85372531642124e-05, "loss": 0.0101, "step": 25890 }, { "grad_norm": 0.175313800573349, "learning_rate": 8.852671607905185e-05, "loss": 0.0112, "step": 25900 }, { "grad_norm": 0.15892983973026276, "learning_rate": 8.851617478066754e-05, "loss": 0.0121, "step": 25910 }, { "grad_norm": 0.1184801459312439, "learning_rate": 8.850562927021227e-05, "loss": 0.0101, "step": 25920 }, { "grad_norm": 0.0760873332619667, "learning_rate": 8.849507954883928e-05, "loss": 0.0076, "step": 25930 }, { "grad_norm": 0.11936367303133011, "learning_rate": 8.848452561770226e-05, "loss": 0.0072, "step": 25940 }, { "grad_norm": 0.25324374437332153, "learning_rate": 8.847396747795538e-05, "loss": 0.011, "step": 25950 }, { "grad_norm": 0.08385801315307617, "learning_rate": 8.846340513075327e-05, "loss": 0.0103, "step": 25960 }, { "grad_norm": 0.10101345926523209, "learning_rate": 8.845283857725099e-05, "loss": 0.0077, "step": 25970 }, { "grad_norm": 0.10953007638454437, "learning_rate": 8.844226781860409e-05, "loss": 0.0092, "step": 25980 }, { "grad_norm": 0.1506512314081192, "learning_rate": 8.84316928559686e-05, "loss": 0.009, "step": 25990 }, { "grad_norm": 0.10435842722654343, "learning_rate": 8.842111369050094e-05, "loss": 0.0077, "step": 26000 }, { "grad_norm": 0.11261238157749176, "learning_rate": 8.841053032335808e-05, "loss": 0.0118, "step": 26010 }, { "grad_norm": 0.22911123931407928, "learning_rate": 8.839994275569735e-05, "loss": 0.0094, "step": 26020 }, { "grad_norm": 0.15504899621009827, "learning_rate": 8.838935098867662e-05, "loss": 0.0129, "step": 26030 }, { "grad_norm": 0.10818526893854141, "learning_rate": 8.837875502345418e-05, "loss": 0.0092, "step": 26040 }, { "grad_norm": 0.10717188566923141, "learning_rate": 8.83681548611888e-05, "loss": 0.0081, "step": 26050 }, { "grad_norm": 0.12154261767864227, "learning_rate": 8.835755050303969e-05, "loss": 0.0107, "step": 26060 }, { "grad_norm": 0.11593106389045715, "learning_rate": 8.834694195016653e-05, "loss": 0.0114, "step": 26070 }, { "grad_norm": 0.11670465022325516, "learning_rate": 8.833632920372942e-05, "loss": 0.0104, "step": 26080 }, { "grad_norm": 0.1690855324268341, "learning_rate": 8.832571226488903e-05, "loss": 0.011, "step": 26090 }, { "grad_norm": 0.12127944827079773, "learning_rate": 8.831509113480634e-05, "loss": 0.0091, "step": 26100 }, { "grad_norm": 0.08790745586156845, "learning_rate": 8.83044658146429e-05, "loss": 0.0083, "step": 26110 }, { "grad_norm": 0.1290716677904129, "learning_rate": 8.829383630556067e-05, "loss": 0.0085, "step": 26120 }, { "grad_norm": 0.15275122225284576, "learning_rate": 8.828320260872207e-05, "loss": 0.0093, "step": 26130 }, { "grad_norm": 0.10608288645744324, "learning_rate": 8.827256472529e-05, "loss": 0.0089, "step": 26140 }, { "grad_norm": 0.11058110743761063, "learning_rate": 8.826192265642778e-05, "loss": 0.0105, "step": 26150 }, { "grad_norm": 0.11925153434276581, "learning_rate": 8.825127640329923e-05, "loss": 0.009, "step": 26160 }, { "grad_norm": 0.1386803686618805, "learning_rate": 8.824062596706861e-05, "loss": 0.0102, "step": 26170 }, { "grad_norm": 0.149741068482399, "learning_rate": 8.822997134890062e-05, "loss": 0.007, "step": 26180 }, { "grad_norm": 0.10067547857761383, "learning_rate": 8.821931254996044e-05, "loss": 0.0077, "step": 26190 }, { "grad_norm": 0.10381399840116501, "learning_rate": 8.82086495714137e-05, "loss": 0.0092, "step": 26200 }, { "grad_norm": 0.2596409320831299, "learning_rate": 8.81979824144265e-05, "loss": 0.0124, "step": 26210 }, { "grad_norm": 0.12517108023166656, "learning_rate": 8.818731108016536e-05, "loss": 0.011, "step": 26220 }, { "grad_norm": 0.1694115698337555, "learning_rate": 8.81766355697973e-05, "loss": 0.0128, "step": 26230 }, { "grad_norm": 0.14230871200561523, "learning_rate": 8.816595588448977e-05, "loss": 0.0094, "step": 26240 }, { "grad_norm": 0.1202329471707344, "learning_rate": 8.81552720254107e-05, "loss": 0.008, "step": 26250 }, { "grad_norm": 0.08850737661123276, "learning_rate": 8.814458399372842e-05, "loss": 0.0084, "step": 26260 }, { "grad_norm": 0.09242189675569534, "learning_rate": 8.813389179061181e-05, "loss": 0.0078, "step": 26270 }, { "grad_norm": 0.07254650443792343, "learning_rate": 8.812319541723012e-05, "loss": 0.0056, "step": 26280 }, { "grad_norm": 0.36002540588378906, "learning_rate": 8.811249487475309e-05, "loss": 0.0082, "step": 26290 }, { "grad_norm": 0.16674891114234924, "learning_rate": 8.810179016435092e-05, "loss": 0.014, "step": 26300 }, { "grad_norm": 0.32874971628189087, "learning_rate": 8.809108128719428e-05, "loss": 0.011, "step": 26310 }, { "grad_norm": 0.14335529506206512, "learning_rate": 8.808036824445424e-05, "loss": 0.0131, "step": 26320 }, { "grad_norm": 0.19946277141571045, "learning_rate": 8.806965103730238e-05, "loss": 0.013, "step": 26330 }, { "grad_norm": 0.14276030659675598, "learning_rate": 8.805892966691074e-05, "loss": 0.0101, "step": 26340 }, { "grad_norm": 0.09016283601522446, "learning_rate": 8.804820413445175e-05, "loss": 0.0122, "step": 26350 }, { "grad_norm": 0.09475644677877426, "learning_rate": 8.803747444109837e-05, "loss": 0.0083, "step": 26360 }, { "grad_norm": 0.1075848937034607, "learning_rate": 8.802674058802399e-05, "loss": 0.008, "step": 26370 }, { "grad_norm": 0.1246752217411995, "learning_rate": 8.801600257640241e-05, "loss": 0.0093, "step": 26380 }, { "grad_norm": 0.153317391872406, "learning_rate": 8.800526040740795e-05, "loss": 0.0133, "step": 26390 }, { "grad_norm": 0.11977206915616989, "learning_rate": 8.799451408221535e-05, "loss": 0.0115, "step": 26400 }, { "grad_norm": 0.2656938433647156, "learning_rate": 8.798376360199982e-05, "loss": 0.0122, "step": 26410 }, { "grad_norm": 0.11229801177978516, "learning_rate": 8.797300896793701e-05, "loss": 0.0121, "step": 26420 }, { "grad_norm": 0.13070377707481384, "learning_rate": 8.796225018120302e-05, "loss": 0.01, "step": 26430 }, { "grad_norm": 0.10204986482858658, "learning_rate": 8.795148724297444e-05, "loss": 0.0117, "step": 26440 }, { "grad_norm": 0.18879128992557526, "learning_rate": 8.794072015442825e-05, "loss": 0.0112, "step": 26450 }, { "grad_norm": 0.15416382253170013, "learning_rate": 8.792994891674198e-05, "loss": 0.0119, "step": 26460 }, { "grad_norm": 0.15966394543647766, "learning_rate": 8.79191735310935e-05, "loss": 0.0118, "step": 26470 }, { "grad_norm": 0.12685801088809967, "learning_rate": 8.790839399866122e-05, "loss": 0.0119, "step": 26480 }, { "grad_norm": 0.21559995412826538, "learning_rate": 8.789761032062397e-05, "loss": 0.013, "step": 26490 }, { "grad_norm": 0.1504899561405182, "learning_rate": 8.788682249816103e-05, "loss": 0.0115, "step": 26500 }, { "grad_norm": 0.1045578196644783, "learning_rate": 8.787603053245215e-05, "loss": 0.0096, "step": 26510 }, { "grad_norm": 0.1199047863483429, "learning_rate": 8.78652344246775e-05, "loss": 0.0101, "step": 26520 }, { "grad_norm": 0.17000284790992737, "learning_rate": 8.785443417601776e-05, "loss": 0.0106, "step": 26530 }, { "grad_norm": 0.1252088099718094, "learning_rate": 8.784362978765401e-05, "loss": 0.0127, "step": 26540 }, { "grad_norm": 0.13214121758937836, "learning_rate": 8.783282126076779e-05, "loss": 0.0118, "step": 26550 }, { "grad_norm": 0.11214496195316315, "learning_rate": 8.782200859654112e-05, "loss": 0.0103, "step": 26560 }, { "grad_norm": 0.13016189634799957, "learning_rate": 8.781119179615646e-05, "loss": 0.0148, "step": 26570 }, { "grad_norm": 0.14185990393161774, "learning_rate": 8.780037086079674e-05, "loss": 0.0124, "step": 26580 }, { "grad_norm": 0.1124902293086052, "learning_rate": 8.778954579164527e-05, "loss": 0.0107, "step": 26590 }, { "grad_norm": 0.09679670631885529, "learning_rate": 8.777871658988588e-05, "loss": 0.0103, "step": 26600 }, { "grad_norm": 0.11333096772432327, "learning_rate": 8.776788325670285e-05, "loss": 0.0114, "step": 26610 }, { "grad_norm": 0.12994174659252167, "learning_rate": 8.775704579328089e-05, "loss": 0.0128, "step": 26620 }, { "grad_norm": 0.15146802365779877, "learning_rate": 8.774620420080517e-05, "loss": 0.0106, "step": 26630 }, { "grad_norm": 0.10493418574333191, "learning_rate": 8.773535848046131e-05, "loss": 0.0094, "step": 26640 }, { "grad_norm": 0.10230204463005066, "learning_rate": 8.772450863343538e-05, "loss": 0.0082, "step": 26650 }, { "grad_norm": 0.06523863971233368, "learning_rate": 8.77136546609139e-05, "loss": 0.0095, "step": 26660 }, { "grad_norm": 0.09505843371152878, "learning_rate": 8.770279656408385e-05, "loss": 0.0095, "step": 26670 }, { "grad_norm": 0.15750038623809814, "learning_rate": 8.769193434413265e-05, "loss": 0.0092, "step": 26680 }, { "grad_norm": 0.10369821637868881, "learning_rate": 8.76810680022482e-05, "loss": 0.0101, "step": 26690 }, { "grad_norm": 0.17530862987041473, "learning_rate": 8.767019753961878e-05, "loss": 0.0106, "step": 26700 }, { "grad_norm": 0.13980595767498016, "learning_rate": 8.765932295743321e-05, "loss": 0.0118, "step": 26710 }, { "grad_norm": 0.14358162879943848, "learning_rate": 8.764844425688068e-05, "loss": 0.0096, "step": 26720 }, { "grad_norm": 0.20378923416137695, "learning_rate": 8.763756143915092e-05, "loss": 0.0098, "step": 26730 }, { "grad_norm": 0.10793934017419815, "learning_rate": 8.7626674505434e-05, "loss": 0.0075, "step": 26740 }, { "grad_norm": 0.11116540431976318, "learning_rate": 8.761578345692053e-05, "loss": 0.0092, "step": 26750 }, { "grad_norm": 0.13140083849430084, "learning_rate": 8.760488829480156e-05, "loss": 0.0099, "step": 26760 }, { "grad_norm": 0.186012864112854, "learning_rate": 8.759398902026854e-05, "loss": 0.0096, "step": 26770 }, { "grad_norm": 0.12051232904195786, "learning_rate": 8.758308563451339e-05, "loss": 0.0125, "step": 26780 }, { "grad_norm": 0.09007807075977325, "learning_rate": 8.75721781387285e-05, "loss": 0.0062, "step": 26790 }, { "grad_norm": 0.08358198404312134, "learning_rate": 8.75612665341067e-05, "loss": 0.0087, "step": 26800 }, { "grad_norm": 0.08494539558887482, "learning_rate": 8.755035082184126e-05, "loss": 0.0083, "step": 26810 }, { "grad_norm": 0.08751951158046722, "learning_rate": 8.753943100312592e-05, "loss": 0.0097, "step": 26820 }, { "grad_norm": 0.16116555035114288, "learning_rate": 8.752850707915484e-05, "loss": 0.0112, "step": 26830 }, { "grad_norm": 0.11523285508155823, "learning_rate": 8.751757905112264e-05, "loss": 0.0088, "step": 26840 }, { "grad_norm": 0.07915844768285751, "learning_rate": 8.75066469202244e-05, "loss": 0.0085, "step": 26850 }, { "grad_norm": 0.10414998233318329, "learning_rate": 8.749571068765567e-05, "loss": 0.0074, "step": 26860 }, { "grad_norm": 0.2257128655910492, "learning_rate": 8.748477035461238e-05, "loss": 0.0123, "step": 26870 }, { "grad_norm": 0.13502438366413116, "learning_rate": 8.747382592229095e-05, "loss": 0.0112, "step": 26880 }, { "grad_norm": 0.18350252509117126, "learning_rate": 8.746287739188828e-05, "loss": 0.0135, "step": 26890 }, { "grad_norm": 0.17059648036956787, "learning_rate": 8.745192476460165e-05, "loss": 0.0094, "step": 26900 }, { "grad_norm": 0.15337057411670685, "learning_rate": 8.744096804162882e-05, "loss": 0.0094, "step": 26910 }, { "grad_norm": 0.2170953005552292, "learning_rate": 8.743000722416804e-05, "loss": 0.0119, "step": 26920 }, { "grad_norm": 0.14192430675029755, "learning_rate": 8.741904231341793e-05, "loss": 0.0106, "step": 26930 }, { "grad_norm": 0.1640240103006363, "learning_rate": 8.740807331057762e-05, "loss": 0.0129, "step": 26940 }, { "grad_norm": 0.1956864297389984, "learning_rate": 8.739710021684667e-05, "loss": 0.0123, "step": 26950 }, { "grad_norm": 0.14552123844623566, "learning_rate": 8.738612303342503e-05, "loss": 0.0106, "step": 26960 }, { "grad_norm": 0.1513448804616928, "learning_rate": 8.73751417615132e-05, "loss": 0.0116, "step": 26970 }, { "grad_norm": 0.14142374694347382, "learning_rate": 8.736415640231208e-05, "loss": 0.0098, "step": 26980 }, { "grad_norm": 0.09712308645248413, "learning_rate": 8.735316695702297e-05, "loss": 0.0104, "step": 26990 }, { "grad_norm": 0.20241183042526245, "learning_rate": 8.734217342684769e-05, "loss": 0.0108, "step": 27000 }, { "grad_norm": 0.12642145156860352, "learning_rate": 8.733117581298847e-05, "loss": 0.0104, "step": 27010 }, { "grad_norm": 0.20425939559936523, "learning_rate": 8.732017411664796e-05, "loss": 0.0104, "step": 27020 }, { "grad_norm": 0.24487286806106567, "learning_rate": 8.730916833902936e-05, "loss": 0.01, "step": 27030 }, { "grad_norm": 0.12336002290248871, "learning_rate": 8.729815848133618e-05, "loss": 0.011, "step": 27040 }, { "grad_norm": 0.12013079971075058, "learning_rate": 8.728714454477247e-05, "loss": 0.0082, "step": 27050 }, { "grad_norm": 0.0868765115737915, "learning_rate": 8.727612653054269e-05, "loss": 0.0115, "step": 27060 }, { "grad_norm": 0.13186515867710114, "learning_rate": 8.726510443985176e-05, "loss": 0.0094, "step": 27070 }, { "grad_norm": 0.15576066076755524, "learning_rate": 8.725407827390503e-05, "loss": 0.0105, "step": 27080 }, { "grad_norm": 0.19032162427902222, "learning_rate": 8.724304803390833e-05, "loss": 0.0092, "step": 27090 }, { "grad_norm": 0.11043088883161545, "learning_rate": 8.723201372106788e-05, "loss": 0.01, "step": 27100 }, { "grad_norm": 0.20686013996601105, "learning_rate": 8.722097533659038e-05, "loss": 0.0117, "step": 27110 }, { "grad_norm": 0.105968177318573, "learning_rate": 8.720993288168299e-05, "loss": 0.0105, "step": 27120 }, { "grad_norm": 0.11658652126789093, "learning_rate": 8.719888635755327e-05, "loss": 0.0101, "step": 27130 }, { "grad_norm": 0.19241902232170105, "learning_rate": 8.718783576540928e-05, "loss": 0.0102, "step": 27140 }, { "grad_norm": 0.13710694015026093, "learning_rate": 8.717678110645948e-05, "loss": 0.0122, "step": 27150 }, { "grad_norm": 0.16389091312885284, "learning_rate": 8.716572238191279e-05, "loss": 0.0129, "step": 27160 }, { "grad_norm": 0.13508234918117523, "learning_rate": 8.715465959297857e-05, "loss": 0.012, "step": 27170 }, { "grad_norm": 0.17419792711734772, "learning_rate": 8.714359274086665e-05, "loss": 0.0094, "step": 27180 }, { "grad_norm": 0.1414772868156433, "learning_rate": 8.713252182678726e-05, "loss": 0.0111, "step": 27190 }, { "grad_norm": 0.10894901305437088, "learning_rate": 8.712144685195112e-05, "loss": 0.0082, "step": 27200 }, { "grad_norm": 0.12642960250377655, "learning_rate": 8.711036781756936e-05, "loss": 0.0089, "step": 27210 }, { "grad_norm": 0.0929078459739685, "learning_rate": 8.709928472485357e-05, "loss": 0.008, "step": 27220 }, { "grad_norm": 0.1857805997133255, "learning_rate": 8.708819757501579e-05, "loss": 0.0094, "step": 27230 }, { "grad_norm": 0.13922710716724396, "learning_rate": 8.707710636926846e-05, "loss": 0.0094, "step": 27240 }, { "grad_norm": 0.15909640491008759, "learning_rate": 8.706601110882455e-05, "loss": 0.0107, "step": 27250 }, { "grad_norm": 0.11961772292852402, "learning_rate": 8.705491179489738e-05, "loss": 0.0102, "step": 27260 }, { "grad_norm": 0.1932588517665863, "learning_rate": 8.704380842870077e-05, "loss": 0.0102, "step": 27270 }, { "grad_norm": 0.12888965010643005, "learning_rate": 8.703270101144895e-05, "loss": 0.0094, "step": 27280 }, { "grad_norm": 0.13078805804252625, "learning_rate": 8.702158954435664e-05, "loss": 0.0091, "step": 27290 }, { "grad_norm": 0.10956736654043198, "learning_rate": 8.701047402863896e-05, "loss": 0.011, "step": 27300 }, { "grad_norm": 0.086058609187603, "learning_rate": 8.699935446551148e-05, "loss": 0.0086, "step": 27310 }, { "grad_norm": 0.12138265371322632, "learning_rate": 8.698823085619022e-05, "loss": 0.0085, "step": 27320 }, { "grad_norm": 0.09103531390428543, "learning_rate": 8.697710320189166e-05, "loss": 0.0064, "step": 27330 }, { "grad_norm": 0.08232194930315018, "learning_rate": 8.696597150383268e-05, "loss": 0.008, "step": 27340 }, { "grad_norm": 0.21809059381484985, "learning_rate": 8.695483576323063e-05, "loss": 0.013, "step": 27350 }, { "grad_norm": 0.16086623072624207, "learning_rate": 8.69436959813033e-05, "loss": 0.0106, "step": 27360 }, { "grad_norm": 0.10360201448202133, "learning_rate": 8.693255215926892e-05, "loss": 0.0098, "step": 27370 }, { "grad_norm": 0.09717720746994019, "learning_rate": 8.692140429834617e-05, "loss": 0.0092, "step": 27380 }, { "grad_norm": 0.14963115751743317, "learning_rate": 8.691025239975415e-05, "loss": 0.0109, "step": 27390 }, { "grad_norm": 0.08664940297603607, "learning_rate": 8.689909646471243e-05, "loss": 0.0096, "step": 27400 }, { "grad_norm": 0.14120176434516907, "learning_rate": 8.688793649444099e-05, "loss": 0.0107, "step": 27410 }, { "grad_norm": 0.08807102590799332, "learning_rate": 8.687677249016029e-05, "loss": 0.0089, "step": 27420 }, { "grad_norm": 0.13522003591060638, "learning_rate": 8.686560445309118e-05, "loss": 0.0116, "step": 27430 }, { "grad_norm": 0.09935542941093445, "learning_rate": 8.685443238445499e-05, "loss": 0.0094, "step": 27440 }, { "grad_norm": 0.13392525911331177, "learning_rate": 8.68432562854735e-05, "loss": 0.0118, "step": 27450 }, { "grad_norm": 0.10796313732862473, "learning_rate": 8.683207615736887e-05, "loss": 0.0091, "step": 27460 }, { "grad_norm": 0.11808650940656662, "learning_rate": 8.682089200136379e-05, "loss": 0.0116, "step": 27470 }, { "grad_norm": 0.10858885943889618, "learning_rate": 8.680970381868132e-05, "loss": 0.0109, "step": 27480 }, { "grad_norm": 0.16560986638069153, "learning_rate": 8.679851161054498e-05, "loss": 0.01, "step": 27490 }, { "grad_norm": 0.1978902965784073, "learning_rate": 8.678731537817873e-05, "loss": 0.0118, "step": 27500 }, { "grad_norm": 0.09958650171756744, "learning_rate": 8.677611512280697e-05, "loss": 0.0087, "step": 27510 }, { "grad_norm": 0.10332149267196655, "learning_rate": 8.676491084565457e-05, "loss": 0.0119, "step": 27520 }, { "grad_norm": 0.11346398293972015, "learning_rate": 8.675370254794678e-05, "loss": 0.0093, "step": 27530 }, { "grad_norm": 0.12709705531597137, "learning_rate": 8.674249023090935e-05, "loss": 0.0094, "step": 27540 }, { "grad_norm": 0.09327760338783264, "learning_rate": 8.673127389576843e-05, "loss": 0.009, "step": 27550 }, { "grad_norm": 0.2553306818008423, "learning_rate": 8.67200535437506e-05, "loss": 0.0085, "step": 27560 }, { "grad_norm": 0.17826002836227417, "learning_rate": 8.670882917608296e-05, "loss": 0.0134, "step": 27570 }, { "grad_norm": 0.1431598961353302, "learning_rate": 8.669760079399292e-05, "loss": 0.0098, "step": 27580 }, { "grad_norm": 0.1403505504131317, "learning_rate": 8.668636839870845e-05, "loss": 0.0118, "step": 27590 }, { "grad_norm": 0.13373495638370514, "learning_rate": 8.667513199145789e-05, "loss": 0.0116, "step": 27600 }, { "grad_norm": 0.21206025779247284, "learning_rate": 8.666389157347002e-05, "loss": 0.0124, "step": 27610 }, { "grad_norm": 0.17236632108688354, "learning_rate": 8.66526471459741e-05, "loss": 0.0137, "step": 27620 }, { "grad_norm": 0.1594385951757431, "learning_rate": 8.66413987101998e-05, "loss": 0.0131, "step": 27630 }, { "grad_norm": 0.18061323463916779, "learning_rate": 8.663014626737723e-05, "loss": 0.0113, "step": 27640 }, { "grad_norm": 0.10740793496370316, "learning_rate": 8.661888981873691e-05, "loss": 0.0088, "step": 27650 }, { "grad_norm": 0.09495784342288971, "learning_rate": 8.660762936550988e-05, "loss": 0.0073, "step": 27660 }, { "grad_norm": 0.16350650787353516, "learning_rate": 8.659636490892753e-05, "loss": 0.0102, "step": 27670 }, { "grad_norm": 0.11346364766359329, "learning_rate": 8.658509645022174e-05, "loss": 0.0086, "step": 27680 }, { "grad_norm": 0.1381663829088211, "learning_rate": 8.657382399062481e-05, "loss": 0.0105, "step": 27690 }, { "grad_norm": 0.10615214705467224, "learning_rate": 8.656254753136946e-05, "loss": 0.0093, "step": 27700 }, { "grad_norm": 0.121695376932621, "learning_rate": 8.655126707368891e-05, "loss": 0.0107, "step": 27710 }, { "grad_norm": 0.11513940989971161, "learning_rate": 8.653998261881672e-05, "loss": 0.0082, "step": 27720 }, { "grad_norm": 0.10391215234994888, "learning_rate": 8.652869416798699e-05, "loss": 0.0093, "step": 27730 }, { "grad_norm": 0.0568237379193306, "learning_rate": 8.651740172243417e-05, "loss": 0.0095, "step": 27740 }, { "grad_norm": 0.10443329811096191, "learning_rate": 8.65061052833932e-05, "loss": 0.0098, "step": 27750 }, { "grad_norm": 0.11696644127368927, "learning_rate": 8.649480485209945e-05, "loss": 0.0101, "step": 27760 }, { "grad_norm": 0.12230750918388367, "learning_rate": 8.64835004297887e-05, "loss": 0.0115, "step": 27770 }, { "grad_norm": 0.09534630924463272, "learning_rate": 8.64721920176972e-05, "loss": 0.008, "step": 27780 }, { "grad_norm": 0.07265443354845047, "learning_rate": 8.646087961706164e-05, "loss": 0.0087, "step": 27790 }, { "grad_norm": 0.08815845847129822, "learning_rate": 8.644956322911908e-05, "loss": 0.0082, "step": 27800 }, { "grad_norm": 0.08981513977050781, "learning_rate": 8.643824285510709e-05, "loss": 0.0079, "step": 27810 }, { "grad_norm": 0.136661097407341, "learning_rate": 8.642691849626364e-05, "loss": 0.0089, "step": 27820 }, { "grad_norm": 0.1275392770767212, "learning_rate": 8.641559015382717e-05, "loss": 0.0082, "step": 27830 }, { "grad_norm": 0.11626189202070236, "learning_rate": 8.640425782903649e-05, "loss": 0.0075, "step": 27840 }, { "grad_norm": 0.08807487040758133, "learning_rate": 8.639292152313091e-05, "loss": 0.0077, "step": 27850 }, { "grad_norm": 0.2775131165981293, "learning_rate": 8.638158123735015e-05, "loss": 0.0089, "step": 27860 }, { "grad_norm": 0.16139398515224457, "learning_rate": 8.637023697293436e-05, "loss": 0.0131, "step": 27870 }, { "grad_norm": 0.11089971661567688, "learning_rate": 8.635888873112414e-05, "loss": 0.0108, "step": 27880 }, { "grad_norm": 0.11203524470329285, "learning_rate": 8.634753651316052e-05, "loss": 0.0118, "step": 27890 }, { "grad_norm": 0.16763152182102203, "learning_rate": 8.633618032028496e-05, "loss": 0.009, "step": 27900 }, { "grad_norm": 0.13835881650447845, "learning_rate": 8.632482015373934e-05, "loss": 0.0131, "step": 27910 }, { "grad_norm": 0.23372936248779297, "learning_rate": 8.6313456014766e-05, "loss": 0.0092, "step": 27920 }, { "grad_norm": 0.13249222934246063, "learning_rate": 8.630208790460771e-05, "loss": 0.0123, "step": 27930 }, { "grad_norm": 0.1176796555519104, "learning_rate": 8.629071582450768e-05, "loss": 0.0084, "step": 27940 }, { "grad_norm": 0.10422120243310928, "learning_rate": 8.62793397757095e-05, "loss": 0.0109, "step": 27950 }, { "grad_norm": 0.09453340619802475, "learning_rate": 8.626795975945729e-05, "loss": 0.0083, "step": 27960 }, { "grad_norm": 0.07491081207990646, "learning_rate": 8.625657577699551e-05, "loss": 0.0095, "step": 27970 }, { "grad_norm": 0.15287303924560547, "learning_rate": 8.624518782956914e-05, "loss": 0.0102, "step": 27980 }, { "grad_norm": 0.1954530030488968, "learning_rate": 8.62337959184235e-05, "loss": 0.011, "step": 27990 }, { "grad_norm": 0.1363072246313095, "learning_rate": 8.622240004480441e-05, "loss": 0.0074, "step": 28000 }, { "grad_norm": 0.11899043619632721, "learning_rate": 8.621100020995814e-05, "loss": 0.0107, "step": 28010 }, { "grad_norm": 0.1128430888056755, "learning_rate": 8.619959641513132e-05, "loss": 0.0114, "step": 28020 }, { "grad_norm": 0.19364532828330994, "learning_rate": 8.618818866157105e-05, "loss": 0.0128, "step": 28030 }, { "grad_norm": 0.09530209004878998, "learning_rate": 8.617677695052487e-05, "loss": 0.0086, "step": 28040 }, { "grad_norm": 0.11246144771575928, "learning_rate": 8.616536128324078e-05, "loss": 0.0115, "step": 28050 }, { "grad_norm": 0.10470226407051086, "learning_rate": 8.615394166096712e-05, "loss": 0.0081, "step": 28060 }, { "grad_norm": 0.09547816216945648, "learning_rate": 8.614251808495279e-05, "loss": 0.0071, "step": 28070 }, { "grad_norm": 0.09906934946775436, "learning_rate": 8.6131090556447e-05, "loss": 0.008, "step": 28080 }, { "grad_norm": 0.2575007975101471, "learning_rate": 8.611965907669947e-05, "loss": 0.009, "step": 28090 }, { "grad_norm": 0.2749984860420227, "learning_rate": 8.610822364696034e-05, "loss": 0.0135, "step": 28100 }, { "grad_norm": 0.1522912085056305, "learning_rate": 8.609678426848015e-05, "loss": 0.0101, "step": 28110 }, { "grad_norm": 0.1370689570903778, "learning_rate": 8.60853409425099e-05, "loss": 0.0098, "step": 28120 }, { "grad_norm": 0.1281532496213913, "learning_rate": 8.607389367030104e-05, "loss": 0.0111, "step": 28130 }, { "grad_norm": 0.11970086395740509, "learning_rate": 8.606244245310538e-05, "loss": 0.0099, "step": 28140 }, { "grad_norm": 0.08697309345006943, "learning_rate": 8.605098729217525e-05, "loss": 0.0067, "step": 28150 }, { "grad_norm": 0.0636046901345253, "learning_rate": 8.603952818876335e-05, "loss": 0.0089, "step": 28160 }, { "grad_norm": 0.10734687000513077, "learning_rate": 8.602806514412281e-05, "loss": 0.0117, "step": 28170 }, { "grad_norm": 0.16729624569416046, "learning_rate": 8.601659815950726e-05, "loss": 0.0093, "step": 28180 }, { "grad_norm": 0.1638997495174408, "learning_rate": 8.600512723617067e-05, "loss": 0.0123, "step": 28190 }, { "grad_norm": 0.11002699285745621, "learning_rate": 8.59936523753675e-05, "loss": 0.0087, "step": 28200 }, { "grad_norm": 0.0850503146648407, "learning_rate": 8.598217357835264e-05, "loss": 0.0073, "step": 28210 }, { "grad_norm": 0.2226981520652771, "learning_rate": 8.597069084638135e-05, "loss": 0.0095, "step": 28220 }, { "grad_norm": 0.10884550958871841, "learning_rate": 8.595920418070939e-05, "loss": 0.0092, "step": 28230 }, { "grad_norm": 0.18654713034629822, "learning_rate": 8.594771358259295e-05, "loss": 0.0113, "step": 28240 }, { "grad_norm": 0.1325371414422989, "learning_rate": 8.593621905328858e-05, "loss": 0.0094, "step": 28250 }, { "grad_norm": 0.25756537914276123, "learning_rate": 8.592472059405333e-05, "loss": 0.0116, "step": 28260 }, { "grad_norm": 0.13947512209415436, "learning_rate": 8.591321820614464e-05, "loss": 0.0088, "step": 28270 }, { "grad_norm": 0.11765573918819427, "learning_rate": 8.590171189082041e-05, "loss": 0.0132, "step": 28280 }, { "grad_norm": 0.22285309433937073, "learning_rate": 8.589020164933894e-05, "loss": 0.0091, "step": 28290 }, { "grad_norm": 0.11694158613681793, "learning_rate": 8.587868748295898e-05, "loss": 0.0116, "step": 28300 }, { "grad_norm": 0.1097918152809143, "learning_rate": 8.586716939293971e-05, "loss": 0.0098, "step": 28310 }, { "grad_norm": 0.10795631259679794, "learning_rate": 8.58556473805407e-05, "loss": 0.01, "step": 28320 }, { "grad_norm": 0.1110009029507637, "learning_rate": 8.584412144702202e-05, "loss": 0.0092, "step": 28330 }, { "grad_norm": 0.11526809632778168, "learning_rate": 8.58325915936441e-05, "loss": 0.008, "step": 28340 }, { "grad_norm": 0.12234240770339966, "learning_rate": 8.582105782166783e-05, "loss": 0.01, "step": 28350 }, { "grad_norm": 0.12136593461036682, "learning_rate": 8.580952013235455e-05, "loss": 0.0079, "step": 28360 }, { "grad_norm": 0.1600155383348465, "learning_rate": 8.579797852696596e-05, "loss": 0.0105, "step": 28370 }, { "grad_norm": 0.13919425010681152, "learning_rate": 8.578643300676428e-05, "loss": 0.0099, "step": 28380 }, { "grad_norm": 0.13800165057182312, "learning_rate": 8.577488357301209e-05, "loss": 0.0087, "step": 28390 }, { "grad_norm": 0.08736803382635117, "learning_rate": 8.576333022697242e-05, "loss": 0.0072, "step": 28400 }, { "grad_norm": 0.08071865141391754, "learning_rate": 8.575177296990873e-05, "loss": 0.0119, "step": 28410 }, { "grad_norm": 0.07645732164382935, "learning_rate": 8.574021180308489e-05, "loss": 0.0085, "step": 28420 }, { "grad_norm": 0.09866584092378616, "learning_rate": 8.572864672776523e-05, "loss": 0.0113, "step": 28430 }, { "grad_norm": 0.11910224705934525, "learning_rate": 8.571707774521447e-05, "loss": 0.0089, "step": 28440 }, { "grad_norm": 0.15730100870132446, "learning_rate": 8.57055048566978e-05, "loss": 0.0094, "step": 28450 }, { "grad_norm": 0.13993704319000244, "learning_rate": 8.569392806348078e-05, "loss": 0.0086, "step": 28460 }, { "grad_norm": 0.1794116497039795, "learning_rate": 8.568234736682947e-05, "loss": 0.0112, "step": 28470 }, { "grad_norm": 0.1644563525915146, "learning_rate": 8.567076276801029e-05, "loss": 0.0135, "step": 28480 }, { "grad_norm": 0.12480651587247849, "learning_rate": 8.565917426829013e-05, "loss": 0.0096, "step": 28490 }, { "grad_norm": 0.15117144584655762, "learning_rate": 8.564758186893628e-05, "loss": 0.0109, "step": 28500 }, { "grad_norm": 0.12675639986991882, "learning_rate": 8.563598557121649e-05, "loss": 0.0082, "step": 28510 }, { "grad_norm": 0.14214052259922028, "learning_rate": 8.562438537639888e-05, "loss": 0.0095, "step": 28520 }, { "grad_norm": 0.14887885749340057, "learning_rate": 8.561278128575206e-05, "loss": 0.0118, "step": 28530 }, { "grad_norm": 0.11021165549755096, "learning_rate": 8.5601173300545e-05, "loss": 0.0098, "step": 28540 }, { "grad_norm": 0.1897747665643692, "learning_rate": 8.558956142204717e-05, "loss": 0.0115, "step": 28550 }, { "grad_norm": 0.1560927778482437, "learning_rate": 8.55779456515284e-05, "loss": 0.0109, "step": 28560 }, { "grad_norm": 0.16075490415096283, "learning_rate": 8.556632599025898e-05, "loss": 0.0108, "step": 28570 }, { "grad_norm": 0.33181455731391907, "learning_rate": 8.555470243950964e-05, "loss": 0.0115, "step": 28580 }, { "grad_norm": 0.2208736091852188, "learning_rate": 8.554307500055148e-05, "loss": 0.0115, "step": 28590 }, { "grad_norm": 0.08159925788640976, "learning_rate": 8.553144367465609e-05, "loss": 0.0085, "step": 28600 }, { "grad_norm": 0.32509154081344604, "learning_rate": 8.551980846309544e-05, "loss": 0.0088, "step": 28610 }, { "grad_norm": 0.12141910195350647, "learning_rate": 8.550816936714193e-05, "loss": 0.0099, "step": 28620 }, { "grad_norm": 0.1641683578491211, "learning_rate": 8.549652638806841e-05, "loss": 0.0128, "step": 28630 }, { "grad_norm": 0.12760187685489655, "learning_rate": 8.548487952714812e-05, "loss": 0.0085, "step": 28640 }, { "grad_norm": 0.21558722853660583, "learning_rate": 8.547322878565478e-05, "loss": 0.0123, "step": 28650 }, { "grad_norm": 0.1263120323419571, "learning_rate": 8.546157416486245e-05, "loss": 0.0082, "step": 28660 }, { "grad_norm": 0.09944704174995422, "learning_rate": 8.54499156660457e-05, "loss": 0.007, "step": 28670 }, { "grad_norm": 0.32912907004356384, "learning_rate": 8.543825329047947e-05, "loss": 0.0135, "step": 28680 }, { "grad_norm": 0.16472263634204865, "learning_rate": 8.542658703943913e-05, "loss": 0.0121, "step": 28690 }, { "grad_norm": 0.16451549530029297, "learning_rate": 8.541491691420051e-05, "loss": 0.0107, "step": 28700 }, { "grad_norm": 0.14869651198387146, "learning_rate": 8.54032429160398e-05, "loss": 0.0116, "step": 28710 }, { "grad_norm": 0.2224055379629135, "learning_rate": 8.539156504623369e-05, "loss": 0.0098, "step": 28720 }, { "grad_norm": 0.09272845089435577, "learning_rate": 8.537988330605923e-05, "loss": 0.0106, "step": 28730 }, { "grad_norm": 0.2809257209300995, "learning_rate": 8.536819769679393e-05, "loss": 0.0107, "step": 28740 }, { "grad_norm": 0.10668574273586273, "learning_rate": 8.53565082197157e-05, "loss": 0.0104, "step": 28750 }, { "grad_norm": 0.11097738146781921, "learning_rate": 8.534481487610289e-05, "loss": 0.0079, "step": 28760 }, { "grad_norm": 0.20447197556495667, "learning_rate": 8.533311766723428e-05, "loss": 0.0113, "step": 28770 }, { "grad_norm": 0.13620591163635254, "learning_rate": 8.532141659438901e-05, "loss": 0.0119, "step": 28780 }, { "grad_norm": 0.1337948888540268, "learning_rate": 8.530971165884675e-05, "loss": 0.0097, "step": 28790 }, { "grad_norm": 0.08238864690065384, "learning_rate": 8.529800286188752e-05, "loss": 0.0086, "step": 28800 }, { "grad_norm": 0.09124238044023514, "learning_rate": 8.528629020479175e-05, "loss": 0.0084, "step": 28810 }, { "grad_norm": 0.07745745778083801, "learning_rate": 8.527457368884033e-05, "loss": 0.0095, "step": 28820 }, { "grad_norm": 0.05779865384101868, "learning_rate": 8.526285331531458e-05, "loss": 0.0067, "step": 28830 }, { "grad_norm": 0.08839535713195801, "learning_rate": 8.525112908549621e-05, "loss": 0.0085, "step": 28840 }, { "grad_norm": 0.09399644285440445, "learning_rate": 8.523940100066735e-05, "loss": 0.0076, "step": 28850 }, { "grad_norm": 0.07534079253673553, "learning_rate": 8.52276690621106e-05, "loss": 0.0072, "step": 28860 }, { "grad_norm": 0.16883111000061035, "learning_rate": 8.521593327110889e-05, "loss": 0.012, "step": 28870 }, { "grad_norm": 0.09649794548749924, "learning_rate": 8.520419362894569e-05, "loss": 0.0084, "step": 28880 }, { "grad_norm": 0.14129279553890228, "learning_rate": 8.51924501369048e-05, "loss": 0.0109, "step": 28890 }, { "grad_norm": 0.10747747868299484, "learning_rate": 8.518070279627047e-05, "loss": 0.0109, "step": 28900 }, { "grad_norm": 0.1063857153058052, "learning_rate": 8.516895160832737e-05, "loss": 0.008, "step": 28910 }, { "grad_norm": 0.09502819180488586, "learning_rate": 8.515719657436061e-05, "loss": 0.0068, "step": 28920 }, { "grad_norm": 0.08647660911083221, "learning_rate": 8.514543769565568e-05, "loss": 0.011, "step": 28930 }, { "grad_norm": 0.06675609201192856, "learning_rate": 8.513367497349853e-05, "loss": 0.0069, "step": 28940 }, { "grad_norm": 0.15525572001934052, "learning_rate": 8.51219084091755e-05, "loss": 0.0105, "step": 28950 }, { "grad_norm": 0.09959091991186142, "learning_rate": 8.511013800397338e-05, "loss": 0.0118, "step": 28960 }, { "grad_norm": 0.1222245991230011, "learning_rate": 8.509836375917937e-05, "loss": 0.011, "step": 28970 }, { "grad_norm": 0.11537321656942368, "learning_rate": 8.508658567608104e-05, "loss": 0.0097, "step": 28980 }, { "grad_norm": 0.11723384261131287, "learning_rate": 8.507480375596647e-05, "loss": 0.0107, "step": 28990 }, { "grad_norm": 0.13069449365139008, "learning_rate": 8.506301800012408e-05, "loss": 0.009, "step": 29000 }, { "grad_norm": 0.07677442580461502, "learning_rate": 8.505122840984278e-05, "loss": 0.0083, "step": 29010 }, { "grad_norm": 0.10776130110025406, "learning_rate": 8.503943498641182e-05, "loss": 0.0112, "step": 29020 }, { "grad_norm": 0.10531342774629593, "learning_rate": 8.502763773112095e-05, "loss": 0.0086, "step": 29030 }, { "grad_norm": 0.15003614127635956, "learning_rate": 8.501583664526026e-05, "loss": 0.0126, "step": 29040 }, { "grad_norm": 0.121978759765625, "learning_rate": 8.500403173012032e-05, "loss": 0.0079, "step": 29050 }, { "grad_norm": 0.09526622295379639, "learning_rate": 8.499222298699211e-05, "loss": 0.0097, "step": 29060 }, { "grad_norm": 0.1502477377653122, "learning_rate": 8.498041041716701e-05, "loss": 0.0111, "step": 29070 }, { "grad_norm": 0.09849437326192856, "learning_rate": 8.496859402193681e-05, "loss": 0.0093, "step": 29080 }, { "grad_norm": 0.10847119241952896, "learning_rate": 8.495677380259374e-05, "loss": 0.0087, "step": 29090 }, { "grad_norm": 0.12425484508275986, "learning_rate": 8.494494976043045e-05, "loss": 0.0091, "step": 29100 }, { "grad_norm": 0.134415403008461, "learning_rate": 8.493312189673998e-05, "loss": 0.0086, "step": 29110 }, { "grad_norm": 0.11964397877454758, "learning_rate": 8.492129021281584e-05, "loss": 0.0094, "step": 29120 }, { "grad_norm": 0.10169361531734467, "learning_rate": 8.490945470995188e-05, "loss": 0.0099, "step": 29130 }, { "grad_norm": 0.08747226744890213, "learning_rate": 8.489761538944247e-05, "loss": 0.0109, "step": 29140 }, { "grad_norm": 0.20186839997768402, "learning_rate": 8.48857722525823e-05, "loss": 0.01, "step": 29150 }, { "grad_norm": 0.13888968527317047, "learning_rate": 8.487392530066652e-05, "loss": 0.0104, "step": 29160 }, { "grad_norm": 0.10549071431159973, "learning_rate": 8.486207453499069e-05, "loss": 0.0086, "step": 29170 }, { "grad_norm": 0.1195632740855217, "learning_rate": 8.485021995685082e-05, "loss": 0.0096, "step": 29180 }, { "grad_norm": 0.08268370479345322, "learning_rate": 8.483836156754328e-05, "loss": 0.0098, "step": 29190 }, { "grad_norm": 0.17179551720619202, "learning_rate": 8.482649936836491e-05, "loss": 0.0084, "step": 29200 }, { "grad_norm": 0.10192204266786575, "learning_rate": 8.481463336061293e-05, "loss": 0.0089, "step": 29210 }, { "grad_norm": 0.10983652621507645, "learning_rate": 8.480276354558496e-05, "loss": 0.0087, "step": 29220 }, { "grad_norm": 0.14196547865867615, "learning_rate": 8.479088992457913e-05, "loss": 0.0098, "step": 29230 }, { "grad_norm": 0.168582484126091, "learning_rate": 8.477901249889387e-05, "loss": 0.0129, "step": 29240 }, { "grad_norm": 0.09660794585943222, "learning_rate": 8.47671312698281e-05, "loss": 0.0103, "step": 29250 }, { "grad_norm": 0.07829874008893967, "learning_rate": 8.475524623868112e-05, "loss": 0.0074, "step": 29260 }, { "grad_norm": 0.22389893233776093, "learning_rate": 8.474335740675266e-05, "loss": 0.0094, "step": 29270 }, { "grad_norm": 0.1370820254087448, "learning_rate": 8.473146477534289e-05, "loss": 0.0092, "step": 29280 }, { "grad_norm": 0.11208658665418625, "learning_rate": 8.471956834575232e-05, "loss": 0.0121, "step": 29290 }, { "grad_norm": 0.11034154146909714, "learning_rate": 8.470766811928197e-05, "loss": 0.0114, "step": 29300 }, { "grad_norm": 0.13357573747634888, "learning_rate": 8.469576409723323e-05, "loss": 0.0092, "step": 29310 }, { "grad_norm": 0.1739218384027481, "learning_rate": 8.468385628090788e-05, "loss": 0.0095, "step": 29320 }, { "grad_norm": 0.09418176859617233, "learning_rate": 8.467194467160815e-05, "loss": 0.0103, "step": 29330 }, { "grad_norm": 0.1616271436214447, "learning_rate": 8.466002927063667e-05, "loss": 0.009, "step": 29340 }, { "grad_norm": 0.135524719953537, "learning_rate": 8.464811007929651e-05, "loss": 0.0094, "step": 29350 }, { "grad_norm": 0.20054516196250916, "learning_rate": 8.463618709889114e-05, "loss": 0.0109, "step": 29360 }, { "grad_norm": 0.11170828342437744, "learning_rate": 8.462426033072442e-05, "loss": 0.0107, "step": 29370 }, { "grad_norm": 0.09938354045152664, "learning_rate": 8.461232977610061e-05, "loss": 0.0083, "step": 29380 }, { "grad_norm": 0.2597292959690094, "learning_rate": 8.46003954363245e-05, "loss": 0.0109, "step": 29390 }, { "grad_norm": 0.15856119990348816, "learning_rate": 8.458845731270115e-05, "loss": 0.012, "step": 29400 }, { "grad_norm": 0.12471599876880646, "learning_rate": 8.45765154065361e-05, "loss": 0.008, "step": 29410 }, { "grad_norm": 0.0838124081492424, "learning_rate": 8.456456971913532e-05, "loss": 0.0074, "step": 29420 }, { "grad_norm": 0.09261032938957214, "learning_rate": 8.455262025180517e-05, "loss": 0.0096, "step": 29430 }, { "grad_norm": 0.06594356149435043, "learning_rate": 8.454066700585242e-05, "loss": 0.0088, "step": 29440 }, { "grad_norm": 0.15317268669605255, "learning_rate": 8.452870998258423e-05, "loss": 0.0087, "step": 29450 }, { "grad_norm": 0.12155333161354065, "learning_rate": 8.451674918330825e-05, "loss": 0.0109, "step": 29460 }, { "grad_norm": 0.13058240711688995, "learning_rate": 8.450478460933246e-05, "loss": 0.0066, "step": 29470 }, { "grad_norm": 0.1670525074005127, "learning_rate": 8.449281626196532e-05, "loss": 0.0096, "step": 29480 }, { "grad_norm": 0.10719538480043411, "learning_rate": 8.448084414251564e-05, "loss": 0.0075, "step": 29490 }, { "grad_norm": 0.10481089353561401, "learning_rate": 8.446886825229271e-05, "loss": 0.0095, "step": 29500 }, { "grad_norm": 0.10591480880975723, "learning_rate": 8.445688859260615e-05, "loss": 0.0091, "step": 29510 }, { "grad_norm": 0.19191209971904755, "learning_rate": 8.444490516476606e-05, "loss": 0.0107, "step": 29520 }, { "grad_norm": 0.09572459757328033, "learning_rate": 8.443291797008293e-05, "loss": 0.0078, "step": 29530 }, { "grad_norm": 0.08680924773216248, "learning_rate": 8.442092700986765e-05, "loss": 0.0118, "step": 29540 }, { "grad_norm": 0.08114168047904968, "learning_rate": 8.440893228543156e-05, "loss": 0.0086, "step": 29550 }, { "grad_norm": 0.22493042051792145, "learning_rate": 8.439693379808638e-05, "loss": 0.0111, "step": 29560 }, { "grad_norm": 0.10690359771251678, "learning_rate": 8.43849315491442e-05, "loss": 0.0085, "step": 29570 }, { "grad_norm": 0.0948764905333519, "learning_rate": 8.437292553991763e-05, "loss": 0.0074, "step": 29580 }, { "grad_norm": 0.20498567819595337, "learning_rate": 8.436091577171959e-05, "loss": 0.0087, "step": 29590 }, { "grad_norm": 0.1254827082157135, "learning_rate": 8.434890224586347e-05, "loss": 0.0102, "step": 29600 }, { "grad_norm": 0.1397402435541153, "learning_rate": 8.433688496366303e-05, "loss": 0.0096, "step": 29610 }, { "grad_norm": 0.11310727149248123, "learning_rate": 8.432486392643248e-05, "loss": 0.009, "step": 29620 }, { "grad_norm": 0.09466815739870071, "learning_rate": 8.431283913548643e-05, "loss": 0.0065, "step": 29630 }, { "grad_norm": 0.1053977832198143, "learning_rate": 8.430081059213985e-05, "loss": 0.0067, "step": 29640 }, { "grad_norm": 0.15346327424049377, "learning_rate": 8.428877829770823e-05, "loss": 0.0093, "step": 29650 }, { "grad_norm": 0.1250534951686859, "learning_rate": 8.427674225350735e-05, "loss": 0.0098, "step": 29660 }, { "grad_norm": 0.15359438955783844, "learning_rate": 8.426470246085347e-05, "loss": 0.0078, "step": 29670 }, { "grad_norm": 0.10646826773881912, "learning_rate": 8.425265892106324e-05, "loss": 0.0083, "step": 29680 }, { "grad_norm": 0.11362212151288986, "learning_rate": 8.424061163545374e-05, "loss": 0.0105, "step": 29690 }, { "grad_norm": 0.10435856133699417, "learning_rate": 8.422856060534243e-05, "loss": 0.0091, "step": 29700 }, { "grad_norm": 0.11423344165086746, "learning_rate": 8.421650583204718e-05, "loss": 0.009, "step": 29710 }, { "grad_norm": 0.10647853463888168, "learning_rate": 8.420444731688633e-05, "loss": 0.0096, "step": 29720 }, { "grad_norm": 0.10812865197658539, "learning_rate": 8.419238506117852e-05, "loss": 0.0124, "step": 29730 }, { "grad_norm": 0.09274021536111832, "learning_rate": 8.418031906624289e-05, "loss": 0.0079, "step": 29740 }, { "grad_norm": 0.09991773217916489, "learning_rate": 8.416824933339898e-05, "loss": 0.0104, "step": 29750 }, { "grad_norm": 0.11435779184103012, "learning_rate": 8.415617586396667e-05, "loss": 0.0097, "step": 29760 }, { "grad_norm": 0.18830496072769165, "learning_rate": 8.414409865926632e-05, "loss": 0.0116, "step": 29770 }, { "grad_norm": 0.13170672953128815, "learning_rate": 8.413201772061867e-05, "loss": 0.008, "step": 29780 }, { "grad_norm": 0.10920295864343643, "learning_rate": 8.411993304934488e-05, "loss": 0.0082, "step": 29790 }, { "grad_norm": 0.18773871660232544, "learning_rate": 8.410784464676654e-05, "loss": 0.013, "step": 29800 }, { "grad_norm": 0.12639087438583374, "learning_rate": 8.409575251420556e-05, "loss": 0.009, "step": 29810 }, { "grad_norm": 0.08840294927358627, "learning_rate": 8.408365665298435e-05, "loss": 0.0085, "step": 29820 }, { "grad_norm": 0.11757820099592209, "learning_rate": 8.40715570644257e-05, "loss": 0.0099, "step": 29830 }, { "grad_norm": 0.16739705204963684, "learning_rate": 8.40594537498528e-05, "loss": 0.0078, "step": 29840 }, { "grad_norm": 0.1602015346288681, "learning_rate": 8.404734671058924e-05, "loss": 0.0092, "step": 29850 }, { "grad_norm": 0.11705495417118073, "learning_rate": 8.403523594795902e-05, "loss": 0.0089, "step": 29860 }, { "grad_norm": 0.26085227727890015, "learning_rate": 8.402312146328659e-05, "loss": 0.0128, "step": 29870 }, { "grad_norm": 0.1453782618045807, "learning_rate": 8.401100325789675e-05, "loss": 0.0132, "step": 29880 }, { "grad_norm": 0.10549622029066086, "learning_rate": 8.399888133311472e-05, "loss": 0.0109, "step": 29890 }, { "grad_norm": 0.12711074948310852, "learning_rate": 8.398675569026613e-05, "loss": 0.0102, "step": 29900 }, { "grad_norm": 0.15606264770030975, "learning_rate": 8.397462633067705e-05, "loss": 0.0128, "step": 29910 }, { "grad_norm": 0.1514575481414795, "learning_rate": 8.396249325567392e-05, "loss": 0.009, "step": 29920 }, { "grad_norm": 0.1399575024843216, "learning_rate": 8.395035646658357e-05, "loss": 0.0126, "step": 29930 }, { "grad_norm": 0.08955927938222885, "learning_rate": 8.39382159647333e-05, "loss": 0.0095, "step": 29940 }, { "grad_norm": 0.08479531854391098, "learning_rate": 8.392607175145075e-05, "loss": 0.0094, "step": 29950 }, { "grad_norm": 0.0766100212931633, "learning_rate": 8.3913923828064e-05, "loss": 0.0059, "step": 29960 }, { "grad_norm": 0.16417357325553894, "learning_rate": 8.390177219590152e-05, "loss": 0.0099, "step": 29970 }, { "grad_norm": 0.15219546854496002, "learning_rate": 8.388961685629222e-05, "loss": 0.0077, "step": 29980 }, { "grad_norm": 0.10215578973293304, "learning_rate": 8.387745781056536e-05, "loss": 0.0071, "step": 29990 }, { "grad_norm": 0.09380725026130676, "learning_rate": 8.386529506005065e-05, "loss": 0.0116, "step": 30000 }, { "grad_norm": 0.08455125242471695, "learning_rate": 8.38531286060782e-05, "loss": 0.0082, "step": 30010 }, { "grad_norm": 0.10370926558971405, "learning_rate": 8.384095844997849e-05, "loss": 0.0116, "step": 30020 }, { "grad_norm": 0.09826396405696869, "learning_rate": 8.382878459308245e-05, "loss": 0.0088, "step": 30030 }, { "grad_norm": 0.06767570227384567, "learning_rate": 8.381660703672138e-05, "loss": 0.0064, "step": 30040 }, { "grad_norm": 0.22215953469276428, "learning_rate": 8.380442578222702e-05, "loss": 0.0105, "step": 30050 }, { "grad_norm": 0.1437527984380722, "learning_rate": 8.379224083093146e-05, "loss": 0.0115, "step": 30060 }, { "grad_norm": 0.0806981548666954, "learning_rate": 8.378005218416727e-05, "loss": 0.0077, "step": 30070 }, { "grad_norm": 0.11728311330080032, "learning_rate": 8.376785984326735e-05, "loss": 0.0074, "step": 30080 }, { "grad_norm": 0.12995947897434235, "learning_rate": 8.375566380956506e-05, "loss": 0.0109, "step": 30090 }, { "grad_norm": 0.31462764739990234, "learning_rate": 8.374346408439411e-05, "loss": 0.0103, "step": 30100 }, { "grad_norm": 0.13082021474838257, "learning_rate": 8.373126066908868e-05, "loss": 0.0106, "step": 30110 }, { "grad_norm": 0.1889440417289734, "learning_rate": 8.371905356498326e-05, "loss": 0.0092, "step": 30120 }, { "grad_norm": 0.08881939202547073, "learning_rate": 8.370684277341288e-05, "loss": 0.011, "step": 30130 }, { "grad_norm": 0.30259954929351807, "learning_rate": 8.369462829571282e-05, "loss": 0.0135, "step": 30140 }, { "grad_norm": 0.15522025525569916, "learning_rate": 8.36824101332189e-05, "loss": 0.0119, "step": 30150 }, { "grad_norm": 0.10995642840862274, "learning_rate": 8.367018828726721e-05, "loss": 0.0093, "step": 30160 }, { "grad_norm": 0.09495164453983307, "learning_rate": 8.365796275919438e-05, "loss": 0.0109, "step": 30170 }, { "grad_norm": 0.09789624065160751, "learning_rate": 8.364573355033734e-05, "loss": 0.0082, "step": 30180 }, { "grad_norm": 0.10655088722705841, "learning_rate": 8.363350066203346e-05, "loss": 0.0087, "step": 30190 }, { "grad_norm": 0.07265953719615936, "learning_rate": 8.362126409562053e-05, "loss": 0.0094, "step": 30200 }, { "grad_norm": 0.11554845422506332, "learning_rate": 8.360902385243667e-05, "loss": 0.012, "step": 30210 }, { "grad_norm": 0.1404176950454712, "learning_rate": 8.359677993382052e-05, "loss": 0.0106, "step": 30220 }, { "grad_norm": 0.1519888937473297, "learning_rate": 8.358453234111103e-05, "loss": 0.0096, "step": 30230 }, { "grad_norm": 0.09984833002090454, "learning_rate": 8.357228107564756e-05, "loss": 0.0089, "step": 30240 }, { "grad_norm": 0.15333645045757294, "learning_rate": 8.356002613876993e-05, "loss": 0.0117, "step": 30250 }, { "grad_norm": 0.16730740666389465, "learning_rate": 8.35477675318183e-05, "loss": 0.0108, "step": 30260 }, { "grad_norm": 0.14540855586528778, "learning_rate": 8.353550525613323e-05, "loss": 0.0096, "step": 30270 }, { "grad_norm": 0.1306300014257431, "learning_rate": 8.352323931305572e-05, "loss": 0.0107, "step": 30280 }, { "grad_norm": 0.12848487496376038, "learning_rate": 8.351096970392717e-05, "loss": 0.0113, "step": 30290 }, { "grad_norm": 0.13243600726127625, "learning_rate": 8.349869643008937e-05, "loss": 0.0104, "step": 30300 }, { "grad_norm": 0.10758640617132187, "learning_rate": 8.348641949288449e-05, "loss": 0.0097, "step": 30310 }, { "grad_norm": 0.1100904792547226, "learning_rate": 8.34741388936551e-05, "loss": 0.0106, "step": 30320 }, { "grad_norm": 0.14707331359386444, "learning_rate": 8.346185463374423e-05, "loss": 0.0113, "step": 30330 }, { "grad_norm": 0.13179424405097961, "learning_rate": 8.344956671449524e-05, "loss": 0.0113, "step": 30340 }, { "grad_norm": 0.10741341859102249, "learning_rate": 8.343727513725192e-05, "loss": 0.0085, "step": 30350 }, { "grad_norm": 0.16671988368034363, "learning_rate": 8.342497990335847e-05, "loss": 0.0112, "step": 30360 }, { "grad_norm": 0.10724025964736938, "learning_rate": 8.341268101415946e-05, "loss": 0.0101, "step": 30370 }, { "grad_norm": 0.15484003722667694, "learning_rate": 8.34003784709999e-05, "loss": 0.0097, "step": 30380 }, { "grad_norm": 0.16891542077064514, "learning_rate": 8.338807227522516e-05, "loss": 0.0098, "step": 30390 }, { "grad_norm": 0.11013290286064148, "learning_rate": 8.337576242818103e-05, "loss": 0.0083, "step": 30400 }, { "grad_norm": 0.10317762196063995, "learning_rate": 8.336344893121372e-05, "loss": 0.0095, "step": 30410 }, { "grad_norm": 0.1353670209646225, "learning_rate": 8.335113178566977e-05, "loss": 0.009, "step": 30420 }, { "grad_norm": 0.09842198342084885, "learning_rate": 8.33388109928962e-05, "loss": 0.0051, "step": 30430 }, { "grad_norm": 0.12570823729038239, "learning_rate": 8.33264865542404e-05, "loss": 0.0067, "step": 30440 }, { "grad_norm": 0.17116065323352814, "learning_rate": 8.331415847105013e-05, "loss": 0.0083, "step": 30450 }, { "grad_norm": 0.09880076348781586, "learning_rate": 8.330182674467357e-05, "loss": 0.0071, "step": 30460 }, { "grad_norm": 0.3302510976791382, "learning_rate": 8.32894913764593e-05, "loss": 0.0092, "step": 30470 }, { "grad_norm": 0.1423574984073639, "learning_rate": 8.327715236775633e-05, "loss": 0.0082, "step": 30480 }, { "grad_norm": 0.16311825811862946, "learning_rate": 8.326480971991398e-05, "loss": 0.0087, "step": 30490 }, { "grad_norm": 0.29003816843032837, "learning_rate": 8.325246343428206e-05, "loss": 0.0095, "step": 30500 }, { "grad_norm": 0.12334951758384705, "learning_rate": 8.324011351221072e-05, "loss": 0.0101, "step": 30510 }, { "grad_norm": 0.09040788561105728, "learning_rate": 8.322775995505057e-05, "loss": 0.0088, "step": 30520 }, { "grad_norm": 0.0775977373123169, "learning_rate": 8.321540276415254e-05, "loss": 0.0091, "step": 30530 }, { "grad_norm": 0.07377380132675171, "learning_rate": 8.320304194086798e-05, "loss": 0.008, "step": 30540 }, { "grad_norm": 0.13089214265346527, "learning_rate": 8.31906774865487e-05, "loss": 0.0105, "step": 30550 }, { "grad_norm": 0.20885516703128815, "learning_rate": 8.317830940254682e-05, "loss": 0.0106, "step": 30560 }, { "grad_norm": 0.14926566183567047, "learning_rate": 8.316593769021491e-05, "loss": 0.013, "step": 30570 }, { "grad_norm": 0.13851143419742584, "learning_rate": 8.315356235090592e-05, "loss": 0.011, "step": 30580 }, { "grad_norm": 0.11154920607805252, "learning_rate": 8.314118338597319e-05, "loss": 0.008, "step": 30590 }, { "grad_norm": 0.15936146676540375, "learning_rate": 8.312880079677048e-05, "loss": 0.0086, "step": 30600 }, { "grad_norm": 0.10834253579378128, "learning_rate": 8.311641458465191e-05, "loss": 0.0087, "step": 30610 }, { "grad_norm": 0.10863306373357773, "learning_rate": 8.310402475097205e-05, "loss": 0.0064, "step": 30620 }, { "grad_norm": 0.13584895431995392, "learning_rate": 8.309163129708581e-05, "loss": 0.0104, "step": 30630 }, { "grad_norm": 0.10099772363901138, "learning_rate": 8.307923422434852e-05, "loss": 0.0098, "step": 30640 }, { "grad_norm": 0.0922430232167244, "learning_rate": 8.30668335341159e-05, "loss": 0.0089, "step": 30650 }, { "grad_norm": 0.13635046780109406, "learning_rate": 8.305442922774408e-05, "loss": 0.0104, "step": 30660 }, { "grad_norm": 0.09778580814599991, "learning_rate": 8.304202130658959e-05, "loss": 0.0072, "step": 30670 }, { "grad_norm": 0.1279422789812088, "learning_rate": 8.302960977200931e-05, "loss": 0.008, "step": 30680 }, { "grad_norm": 0.12031332403421402, "learning_rate": 8.301719462536058e-05, "loss": 0.0092, "step": 30690 }, { "grad_norm": 0.1396738886833191, "learning_rate": 8.300477586800108e-05, "loss": 0.0101, "step": 30700 }, { "grad_norm": 0.10965152084827423, "learning_rate": 8.299235350128892e-05, "loss": 0.0093, "step": 30710 }, { "grad_norm": 0.10092443227767944, "learning_rate": 8.297992752658258e-05, "loss": 0.0102, "step": 30720 }, { "grad_norm": 0.12909847497940063, "learning_rate": 8.296749794524094e-05, "loss": 0.0109, "step": 30730 }, { "grad_norm": 0.09826788306236267, "learning_rate": 8.295506475862332e-05, "loss": 0.01, "step": 30740 }, { "grad_norm": 0.127895325422287, "learning_rate": 8.294262796808933e-05, "loss": 0.007, "step": 30750 }, { "grad_norm": 0.11459565907716751, "learning_rate": 8.293018757499909e-05, "loss": 0.0075, "step": 30760 }, { "grad_norm": 0.1324159950017929, "learning_rate": 8.291774358071305e-05, "loss": 0.0088, "step": 30770 }, { "grad_norm": 0.11817425489425659, "learning_rate": 8.290529598659205e-05, "loss": 0.0121, "step": 30780 }, { "grad_norm": 0.1667945683002472, "learning_rate": 8.289284479399738e-05, "loss": 0.0081, "step": 30790 }, { "grad_norm": 0.13656195998191833, "learning_rate": 8.288039000429064e-05, "loss": 0.011, "step": 30800 }, { "grad_norm": 0.10043580085039139, "learning_rate": 8.286793161883388e-05, "loss": 0.0096, "step": 30810 }, { "grad_norm": 0.16180676221847534, "learning_rate": 8.285546963898954e-05, "loss": 0.0096, "step": 30820 }, { "grad_norm": 0.1396665722131729, "learning_rate": 8.284300406612044e-05, "loss": 0.0112, "step": 30830 }, { "grad_norm": 0.09699498116970062, "learning_rate": 8.283053490158978e-05, "loss": 0.0078, "step": 30840 }, { "grad_norm": 0.06802566349506378, "learning_rate": 8.28180621467612e-05, "loss": 0.0071, "step": 30850 }, { "grad_norm": 0.09719998389482498, "learning_rate": 8.280558580299867e-05, "loss": 0.0091, "step": 30860 }, { "grad_norm": 0.16416236758232117, "learning_rate": 8.279310587166661e-05, "loss": 0.012, "step": 30870 }, { "grad_norm": 0.11069633811712265, "learning_rate": 8.278062235412978e-05, "loss": 0.0094, "step": 30880 }, { "grad_norm": 0.18885204195976257, "learning_rate": 8.276813525175339e-05, "loss": 0.0089, "step": 30890 }, { "grad_norm": 0.09826420992612839, "learning_rate": 8.2755644565903e-05, "loss": 0.009, "step": 30900 }, { "grad_norm": 0.1666344702243805, "learning_rate": 8.274315029794454e-05, "loss": 0.0102, "step": 30910 }, { "grad_norm": 0.08233197033405304, "learning_rate": 8.273065244924443e-05, "loss": 0.0099, "step": 30920 }, { "grad_norm": 0.08965442329645157, "learning_rate": 8.271815102116936e-05, "loss": 0.0086, "step": 30930 }, { "grad_norm": 0.06564803421497345, "learning_rate": 8.270564601508648e-05, "loss": 0.007, "step": 30940 }, { "grad_norm": 0.07257457077503204, "learning_rate": 8.269313743236333e-05, "loss": 0.0066, "step": 30950 }, { "grad_norm": 0.08793266117572784, "learning_rate": 8.268062527436783e-05, "loss": 0.0082, "step": 30960 }, { "grad_norm": 0.09302791208028793, "learning_rate": 8.266810954246828e-05, "loss": 0.0083, "step": 30970 }, { "grad_norm": 0.09925544261932373, "learning_rate": 8.265559023803338e-05, "loss": 0.0075, "step": 30980 }, { "grad_norm": 0.14143262803554535, "learning_rate": 8.264306736243223e-05, "loss": 0.011, "step": 30990 }, { "grad_norm": 0.13789038360118866, "learning_rate": 8.263054091703432e-05, "loss": 0.0091, "step": 31000 }, { "grad_norm": 0.11109968274831772, "learning_rate": 8.26180109032095e-05, "loss": 0.01, "step": 31010 }, { "grad_norm": 0.09094323962926865, "learning_rate": 8.260547732232807e-05, "loss": 0.0093, "step": 31020 }, { "grad_norm": 0.08522950112819672, "learning_rate": 8.259294017576063e-05, "loss": 0.0069, "step": 31030 }, { "grad_norm": 0.1914813071489334, "learning_rate": 8.258039946487828e-05, "loss": 0.0103, "step": 31040 }, { "grad_norm": 0.10172931104898453, "learning_rate": 8.256785519105241e-05, "loss": 0.0084, "step": 31050 }, { "grad_norm": 0.09341657161712646, "learning_rate": 8.255530735565488e-05, "loss": 0.0075, "step": 31060 }, { "grad_norm": 0.11098755896091461, "learning_rate": 8.254275596005787e-05, "loss": 0.0082, "step": 31070 }, { "grad_norm": 0.10518083721399307, "learning_rate": 8.2530201005634e-05, "loss": 0.0101, "step": 31080 }, { "grad_norm": 0.1232626885175705, "learning_rate": 8.251764249375626e-05, "loss": 0.0078, "step": 31090 }, { "grad_norm": 0.15526163578033447, "learning_rate": 8.250508042579803e-05, "loss": 0.0092, "step": 31100 }, { "grad_norm": 0.14684396982192993, "learning_rate": 8.249251480313307e-05, "loss": 0.0079, "step": 31110 }, { "grad_norm": 0.1261819750070572, "learning_rate": 8.247994562713555e-05, "loss": 0.0094, "step": 31120 }, { "grad_norm": 0.19212915003299713, "learning_rate": 8.246737289917998e-05, "loss": 0.0091, "step": 31130 }, { "grad_norm": 0.11805040389299393, "learning_rate": 8.245479662064135e-05, "loss": 0.0086, "step": 31140 }, { "grad_norm": 0.1214311346411705, "learning_rate": 8.244221679289496e-05, "loss": 0.0083, "step": 31150 }, { "grad_norm": 0.1771279275417328, "learning_rate": 8.242963341731652e-05, "loss": 0.0099, "step": 31160 }, { "grad_norm": 0.09309730678796768, "learning_rate": 8.24170464952821e-05, "loss": 0.0081, "step": 31170 }, { "grad_norm": 0.21193353831768036, "learning_rate": 8.240445602816824e-05, "loss": 0.0104, "step": 31180 }, { "grad_norm": 0.09803927689790726, "learning_rate": 8.239186201735179e-05, "loss": 0.0084, "step": 31190 }, { "grad_norm": 0.2292550951242447, "learning_rate": 8.237926446420998e-05, "loss": 0.0099, "step": 31200 }, { "grad_norm": 0.10384569317102432, "learning_rate": 8.236666337012052e-05, "loss": 0.0105, "step": 31210 }, { "grad_norm": 0.23195374011993408, "learning_rate": 8.23540587364614e-05, "loss": 0.0126, "step": 31220 }, { "grad_norm": 0.08800700306892395, "learning_rate": 8.234145056461107e-05, "loss": 0.0099, "step": 31230 }, { "grad_norm": 0.19794626533985138, "learning_rate": 8.232883885594832e-05, "loss": 0.0119, "step": 31240 }, { "grad_norm": 0.09182585775852203, "learning_rate": 8.231622361185236e-05, "loss": 0.0091, "step": 31250 }, { "grad_norm": 0.17343075573444366, "learning_rate": 8.230360483370278e-05, "loss": 0.0082, "step": 31260 }, { "grad_norm": 0.09549783915281296, "learning_rate": 8.229098252287953e-05, "loss": 0.0092, "step": 31270 }, { "grad_norm": 0.08795969188213348, "learning_rate": 8.2278356680763e-05, "loss": 0.0106, "step": 31280 }, { "grad_norm": 0.11694170534610748, "learning_rate": 8.22657273087339e-05, "loss": 0.0098, "step": 31290 }, { "grad_norm": 0.10855381935834885, "learning_rate": 8.225309440817336e-05, "loss": 0.0119, "step": 31300 }, { "grad_norm": 0.10024954378604889, "learning_rate": 8.224045798046293e-05, "loss": 0.0072, "step": 31310 }, { "grad_norm": 0.08531748503446579, "learning_rate": 8.22278180269845e-05, "loss": 0.0081, "step": 31320 }, { "grad_norm": 0.11576776951551437, "learning_rate": 8.221517454912031e-05, "loss": 0.0088, "step": 31330 }, { "grad_norm": 0.12554246187210083, "learning_rate": 8.220252754825308e-05, "loss": 0.0089, "step": 31340 }, { "grad_norm": 0.15942367911338806, "learning_rate": 8.218987702576586e-05, "loss": 0.0105, "step": 31350 }, { "grad_norm": 0.12327072024345398, "learning_rate": 8.217722298304207e-05, "loss": 0.0102, "step": 31360 }, { "grad_norm": 0.12499910593032837, "learning_rate": 8.216456542146557e-05, "loss": 0.0096, "step": 31370 }, { "grad_norm": 0.08541116118431091, "learning_rate": 8.215190434242055e-05, "loss": 0.0075, "step": 31380 }, { "grad_norm": 0.07785410434007645, "learning_rate": 8.213923974729161e-05, "loss": 0.0096, "step": 31390 }, { "grad_norm": 0.1976744830608368, "learning_rate": 8.212657163746373e-05, "loss": 0.014, "step": 31400 }, { "grad_norm": 0.1315833032131195, "learning_rate": 8.211390001432227e-05, "loss": 0.0117, "step": 31410 }, { "grad_norm": 0.15960410237312317, "learning_rate": 8.210122487925297e-05, "loss": 0.0116, "step": 31420 }, { "grad_norm": 0.08122992515563965, "learning_rate": 8.208854623364202e-05, "loss": 0.0094, "step": 31430 }, { "grad_norm": 0.12678281962871552, "learning_rate": 8.207586407887589e-05, "loss": 0.0096, "step": 31440 }, { "grad_norm": 0.11438398063182831, "learning_rate": 8.206317841634148e-05, "loss": 0.0075, "step": 31450 }, { "grad_norm": 0.25821229815483093, "learning_rate": 8.205048924742609e-05, "loss": 0.01, "step": 31460 }, { "grad_norm": 0.10235166549682617, "learning_rate": 8.203779657351738e-05, "loss": 0.0104, "step": 31470 }, { "grad_norm": 0.13525164127349854, "learning_rate": 8.20251003960034e-05, "loss": 0.008, "step": 31480 }, { "grad_norm": 0.24719397723674774, "learning_rate": 8.201240071627258e-05, "loss": 0.0098, "step": 31490 }, { "grad_norm": 0.1518404632806778, "learning_rate": 8.199969753571377e-05, "loss": 0.0098, "step": 31500 }, { "grad_norm": 0.20673444867134094, "learning_rate": 8.198699085571615e-05, "loss": 0.0093, "step": 31510 }, { "grad_norm": 0.10422701388597488, "learning_rate": 8.197428067766928e-05, "loss": 0.0088, "step": 31520 }, { "grad_norm": 0.24780793488025665, "learning_rate": 8.196156700296316e-05, "loss": 0.0107, "step": 31530 }, { "grad_norm": 0.14531850814819336, "learning_rate": 8.194884983298814e-05, "loss": 0.0115, "step": 31540 }, { "grad_norm": 0.10233011841773987, "learning_rate": 8.193612916913491e-05, "loss": 0.011, "step": 31550 }, { "grad_norm": 0.18791955709457397, "learning_rate": 8.192340501279463e-05, "loss": 0.0084, "step": 31560 }, { "grad_norm": 0.12467340379953384, "learning_rate": 8.191067736535876e-05, "loss": 0.0088, "step": 31570 }, { "grad_norm": 0.0823015570640564, "learning_rate": 8.18979462282192e-05, "loss": 0.0071, "step": 31580 }, { "grad_norm": 0.08839594572782516, "learning_rate": 8.188521160276819e-05, "loss": 0.0081, "step": 31590 }, { "grad_norm": 0.22778543829917908, "learning_rate": 8.187247349039837e-05, "loss": 0.0096, "step": 31600 }, { "grad_norm": 0.266325443983078, "learning_rate": 8.185973189250278e-05, "loss": 0.0087, "step": 31610 }, { "grad_norm": 0.1541169285774231, "learning_rate": 8.184698681047482e-05, "loss": 0.0133, "step": 31620 }, { "grad_norm": 0.11423028260469437, "learning_rate": 8.183423824570827e-05, "loss": 0.0067, "step": 31630 }, { "grad_norm": 0.08219517767429352, "learning_rate": 8.182148619959725e-05, "loss": 0.0104, "step": 31640 }, { "grad_norm": 0.15530119836330414, "learning_rate": 8.180873067353636e-05, "loss": 0.0097, "step": 31650 }, { "grad_norm": 0.12728245556354523, "learning_rate": 8.179597166892052e-05, "loss": 0.0101, "step": 31660 }, { "grad_norm": 0.18943794071674347, "learning_rate": 8.178320918714501e-05, "loss": 0.0115, "step": 31670 }, { "grad_norm": 0.09828218817710876, "learning_rate": 8.177044322960554e-05, "loss": 0.0095, "step": 31680 }, { "grad_norm": 0.09810668975114822, "learning_rate": 8.175767379769816e-05, "loss": 0.0072, "step": 31690 }, { "grad_norm": 0.09888925403356552, "learning_rate": 8.174490089281932e-05, "loss": 0.0096, "step": 31700 }, { "grad_norm": 0.14031605422496796, "learning_rate": 8.173212451636584e-05, "loss": 0.0088, "step": 31710 }, { "grad_norm": 0.13754616677761078, "learning_rate": 8.171934466973493e-05, "loss": 0.0083, "step": 31720 }, { "grad_norm": 0.1655488759279251, "learning_rate": 8.170656135432418e-05, "loss": 0.0095, "step": 31730 }, { "grad_norm": 0.08498386293649673, "learning_rate": 8.169377457153155e-05, "loss": 0.0097, "step": 31740 }, { "grad_norm": 0.09907138347625732, "learning_rate": 8.168098432275539e-05, "loss": 0.0094, "step": 31750 }, { "grad_norm": 0.10257411748170853, "learning_rate": 8.166819060939442e-05, "loss": 0.0081, "step": 31760 }, { "grad_norm": 0.18299457430839539, "learning_rate": 8.165539343284772e-05, "loss": 0.0101, "step": 31770 }, { "grad_norm": 0.2806052565574646, "learning_rate": 8.16425927945148e-05, "loss": 0.0094, "step": 31780 }, { "grad_norm": 0.17733262479305267, "learning_rate": 8.162978869579551e-05, "loss": 0.0099, "step": 31790 }, { "grad_norm": 0.1401364803314209, "learning_rate": 8.161698113809007e-05, "loss": 0.0083, "step": 31800 }, { "grad_norm": 0.15931421518325806, "learning_rate": 8.160417012279911e-05, "loss": 0.008, "step": 31810 }, { "grad_norm": 0.11883849650621414, "learning_rate": 8.159135565132363e-05, "loss": 0.0091, "step": 31820 }, { "grad_norm": 0.22738514840602875, "learning_rate": 8.157853772506498e-05, "loss": 0.0087, "step": 31830 }, { "grad_norm": 0.11930641531944275, "learning_rate": 8.156571634542494e-05, "loss": 0.0091, "step": 31840 }, { "grad_norm": 0.1555333286523819, "learning_rate": 8.15528915138056e-05, "loss": 0.0093, "step": 31850 }, { "grad_norm": 0.12494946271181107, "learning_rate": 8.154006323160949e-05, "loss": 0.009, "step": 31860 }, { "grad_norm": 0.09805641323328018, "learning_rate": 8.152723150023949e-05, "loss": 0.0081, "step": 31870 }, { "grad_norm": 0.06711525470018387, "learning_rate": 8.151439632109886e-05, "loss": 0.0069, "step": 31880 }, { "grad_norm": 0.10051631927490234, "learning_rate": 8.150155769559122e-05, "loss": 0.0079, "step": 31890 }, { "grad_norm": 0.15400433540344238, "learning_rate": 8.148871562512058e-05, "loss": 0.0086, "step": 31900 }, { "grad_norm": 0.15804553031921387, "learning_rate": 8.147587011109136e-05, "loss": 0.0109, "step": 31910 }, { "grad_norm": 0.14935490489006042, "learning_rate": 8.14630211549083e-05, "loss": 0.0079, "step": 31920 }, { "grad_norm": 0.10888995230197906, "learning_rate": 8.145016875797655e-05, "loss": 0.0071, "step": 31930 }, { "grad_norm": 0.10424956679344177, "learning_rate": 8.143731292170164e-05, "loss": 0.0096, "step": 31940 }, { "grad_norm": 0.09921438992023468, "learning_rate": 8.142445364748944e-05, "loss": 0.0073, "step": 31950 }, { "grad_norm": 0.11775533109903336, "learning_rate": 8.141159093674624e-05, "loss": 0.0105, "step": 31960 }, { "grad_norm": 0.12074354290962219, "learning_rate": 8.139872479087869e-05, "loss": 0.0095, "step": 31970 }, { "grad_norm": 0.11406475305557251, "learning_rate": 8.13858552112938e-05, "loss": 0.0103, "step": 31980 }, { "grad_norm": 0.09575912356376648, "learning_rate": 8.137298219939895e-05, "loss": 0.0063, "step": 31990 }, { "grad_norm": 0.12848828732967377, "learning_rate": 8.136010575660196e-05, "loss": 0.0098, "step": 32000 }, { "grad_norm": 0.10321440547704697, "learning_rate": 8.134722588431095e-05, "loss": 0.01, "step": 32010 }, { "grad_norm": 0.11652962118387222, "learning_rate": 8.133434258393444e-05, "loss": 0.0092, "step": 32020 }, { "grad_norm": 0.12550042569637299, "learning_rate": 8.132145585688134e-05, "loss": 0.0089, "step": 32030 }, { "grad_norm": 0.1214456558227539, "learning_rate": 8.130856570456093e-05, "loss": 0.009, "step": 32040 }, { "grad_norm": 0.11944735795259476, "learning_rate": 8.129567212838283e-05, "loss": 0.0077, "step": 32050 }, { "grad_norm": 0.1328495591878891, "learning_rate": 8.128277512975708e-05, "loss": 0.0089, "step": 32060 }, { "grad_norm": 0.09913531690835953, "learning_rate": 8.126987471009408e-05, "loss": 0.0061, "step": 32070 }, { "grad_norm": 0.1224772110581398, "learning_rate": 8.125697087080459e-05, "loss": 0.0091, "step": 32080 }, { "grad_norm": 0.08898892253637314, "learning_rate": 8.124406361329976e-05, "loss": 0.0101, "step": 32090 }, { "grad_norm": 0.1313571035861969, "learning_rate": 8.123115293899111e-05, "loss": 0.0087, "step": 32100 }, { "grad_norm": 0.09082163870334625, "learning_rate": 8.121823884929055e-05, "loss": 0.0089, "step": 32110 }, { "grad_norm": 0.1331106573343277, "learning_rate": 8.12053213456103e-05, "loss": 0.0092, "step": 32120 }, { "grad_norm": 0.09649847447872162, "learning_rate": 8.119240042936303e-05, "loss": 0.0085, "step": 32130 }, { "grad_norm": 0.07343059033155441, "learning_rate": 8.117947610196175e-05, "loss": 0.0087, "step": 32140 }, { "grad_norm": 0.09843356907367706, "learning_rate": 8.116654836481982e-05, "loss": 0.0103, "step": 32150 }, { "grad_norm": 0.10476884245872498, "learning_rate": 8.115361721935106e-05, "loss": 0.0083, "step": 32160 }, { "grad_norm": 0.12331274896860123, "learning_rate": 8.114068266696953e-05, "loss": 0.0087, "step": 32170 }, { "grad_norm": 0.12818260490894318, "learning_rate": 8.112774470908978e-05, "loss": 0.0087, "step": 32180 }, { "grad_norm": 0.07846152037382126, "learning_rate": 8.111480334712665e-05, "loss": 0.0061, "step": 32190 }, { "grad_norm": 0.0733424574136734, "learning_rate": 8.110185858249542e-05, "loss": 0.0068, "step": 32200 }, { "grad_norm": 0.14551439881324768, "learning_rate": 8.108891041661168e-05, "loss": 0.0126, "step": 32210 }, { "grad_norm": 0.1091945469379425, "learning_rate": 8.107595885089146e-05, "loss": 0.0095, "step": 32220 }, { "grad_norm": 0.16683857142925262, "learning_rate": 8.106300388675108e-05, "loss": 0.0093, "step": 32230 }, { "grad_norm": 0.12603136897087097, "learning_rate": 8.105004552560731e-05, "loss": 0.008, "step": 32240 }, { "grad_norm": 0.08613547682762146, "learning_rate": 8.103708376887724e-05, "loss": 0.0056, "step": 32250 }, { "grad_norm": 0.09659937024116516, "learning_rate": 8.102411861797836e-05, "loss": 0.0109, "step": 32260 }, { "grad_norm": 0.10668610036373138, "learning_rate": 8.101115007432851e-05, "loss": 0.0082, "step": 32270 }, { "grad_norm": 0.11161945015192032, "learning_rate": 8.09981781393459e-05, "loss": 0.0087, "step": 32280 }, { "grad_norm": 0.17721609771251678, "learning_rate": 8.098520281444915e-05, "loss": 0.0098, "step": 32290 }, { "grad_norm": 0.07572517544031143, "learning_rate": 8.09722241010572e-05, "loss": 0.0093, "step": 32300 }, { "grad_norm": 0.17619876563549042, "learning_rate": 8.095924200058939e-05, "loss": 0.0085, "step": 32310 }, { "grad_norm": 0.10304118692874908, "learning_rate": 8.094625651446541e-05, "loss": 0.0064, "step": 32320 }, { "grad_norm": 0.09126494824886322, "learning_rate": 8.093326764410536e-05, "loss": 0.009, "step": 32330 }, { "grad_norm": 0.06879735738039017, "learning_rate": 8.092027539092966e-05, "loss": 0.0078, "step": 32340 }, { "grad_norm": 0.07172515243291855, "learning_rate": 8.090727975635913e-05, "loss": 0.0064, "step": 32350 }, { "grad_norm": 0.2636140286922455, "learning_rate": 8.089428074181497e-05, "loss": 0.0083, "step": 32360 }, { "grad_norm": 0.12123817205429077, "learning_rate": 8.088127834871871e-05, "loss": 0.0092, "step": 32370 }, { "grad_norm": 0.19290751218795776, "learning_rate": 8.086827257849226e-05, "loss": 0.0091, "step": 32380 }, { "grad_norm": 0.11675050109624863, "learning_rate": 8.085526343255795e-05, "loss": 0.011, "step": 32390 }, { "grad_norm": 0.11104561388492584, "learning_rate": 8.084225091233842e-05, "loss": 0.0097, "step": 32400 }, { "grad_norm": 0.07129671424627304, "learning_rate": 8.082923501925668e-05, "loss": 0.011, "step": 32410 }, { "grad_norm": 0.08939466625452042, "learning_rate": 8.081621575473617e-05, "loss": 0.0093, "step": 32420 }, { "grad_norm": 0.22875380516052246, "learning_rate": 8.080319312020064e-05, "loss": 0.0085, "step": 32430 }, { "grad_norm": 0.12032339721918106, "learning_rate": 8.079016711707421e-05, "loss": 0.0083, "step": 32440 }, { "grad_norm": 0.0904342383146286, "learning_rate": 8.077713774678139e-05, "loss": 0.0091, "step": 32450 }, { "grad_norm": 0.08598200231790543, "learning_rate": 8.076410501074707e-05, "loss": 0.0084, "step": 32460 }, { "grad_norm": 0.22482064366340637, "learning_rate": 8.075106891039647e-05, "loss": 0.0098, "step": 32470 }, { "grad_norm": 0.12739713490009308, "learning_rate": 8.073802944715523e-05, "loss": 0.0093, "step": 32480 }, { "grad_norm": 0.08344835788011551, "learning_rate": 8.072498662244929e-05, "loss": 0.0074, "step": 32490 }, { "grad_norm": 0.06938858330249786, "learning_rate": 8.0711940437705e-05, "loss": 0.007, "step": 32500 }, { "grad_norm": 0.13381801545619965, "learning_rate": 8.06988908943491e-05, "loss": 0.0076, "step": 32510 }, { "grad_norm": 0.09896818548440933, "learning_rate": 8.068583799380863e-05, "loss": 0.0085, "step": 32520 }, { "grad_norm": 0.11709409952163696, "learning_rate": 8.067278173751104e-05, "loss": 0.0063, "step": 32530 }, { "grad_norm": 0.14085690677165985, "learning_rate": 8.065972212688417e-05, "loss": 0.0143, "step": 32540 }, { "grad_norm": 0.12575623393058777, "learning_rate": 8.064665916335618e-05, "loss": 0.0083, "step": 32550 }, { "grad_norm": 0.12546385824680328, "learning_rate": 8.063359284835564e-05, "loss": 0.0084, "step": 32560 }, { "grad_norm": 0.11140091717243195, "learning_rate": 8.062052318331142e-05, "loss": 0.009, "step": 32570 }, { "grad_norm": 0.17398563027381897, "learning_rate": 8.060745016965283e-05, "loss": 0.0122, "step": 32580 }, { "grad_norm": 0.1375882774591446, "learning_rate": 8.059437380880952e-05, "loss": 0.01, "step": 32590 }, { "grad_norm": 0.12753453850746155, "learning_rate": 8.058129410221146e-05, "loss": 0.011, "step": 32600 }, { "grad_norm": 0.1190129891037941, "learning_rate": 8.056821105128908e-05, "loss": 0.0084, "step": 32610 }, { "grad_norm": 0.10250161588191986, "learning_rate": 8.05551246574731e-05, "loss": 0.0067, "step": 32620 }, { "grad_norm": 0.08160922676324844, "learning_rate": 8.05420349221946e-05, "loss": 0.0087, "step": 32630 }, { "grad_norm": 0.08839499950408936, "learning_rate": 8.05289418468851e-05, "loss": 0.0064, "step": 32640 }, { "grad_norm": 0.16472676396369934, "learning_rate": 8.051584543297642e-05, "loss": 0.0083, "step": 32650 }, { "grad_norm": 0.10490241646766663, "learning_rate": 8.050274568190074e-05, "loss": 0.0086, "step": 32660 }, { "grad_norm": 0.0662681981921196, "learning_rate": 8.048964259509067e-05, "loss": 0.008, "step": 32670 }, { "grad_norm": 0.07975159585475922, "learning_rate": 8.047653617397914e-05, "loss": 0.011, "step": 32680 }, { "grad_norm": 0.12136247009038925, "learning_rate": 8.046342641999941e-05, "loss": 0.0098, "step": 32690 }, { "grad_norm": 0.10243873298168182, "learning_rate": 8.045031333458517e-05, "loss": 0.0085, "step": 32700 }, { "grad_norm": 0.11961743235588074, "learning_rate": 8.043719691917047e-05, "loss": 0.0087, "step": 32710 }, { "grad_norm": 0.1388600468635559, "learning_rate": 8.042407717518966e-05, "loss": 0.0099, "step": 32720 }, { "grad_norm": 0.14048916101455688, "learning_rate": 8.041095410407751e-05, "loss": 0.0112, "step": 32730 }, { "grad_norm": 0.15317057073116302, "learning_rate": 8.039782770726913e-05, "loss": 0.0078, "step": 32740 }, { "grad_norm": 0.11371424794197083, "learning_rate": 8.038469798620004e-05, "loss": 0.0094, "step": 32750 }, { "grad_norm": 0.1074197068810463, "learning_rate": 8.037156494230604e-05, "loss": 0.0094, "step": 32760 }, { "grad_norm": 0.1164090484380722, "learning_rate": 8.035842857702338e-05, "loss": 0.0096, "step": 32770 }, { "grad_norm": 0.10330914705991745, "learning_rate": 8.03452888917886e-05, "loss": 0.0087, "step": 32780 }, { "grad_norm": 0.16441023349761963, "learning_rate": 8.033214588803866e-05, "loss": 0.0076, "step": 32790 }, { "grad_norm": 0.0924493670463562, "learning_rate": 8.031899956721083e-05, "loss": 0.0092, "step": 32800 }, { "grad_norm": 0.10101215541362762, "learning_rate": 8.030584993074282e-05, "loss": 0.0081, "step": 32810 }, { "grad_norm": 0.0849481001496315, "learning_rate": 8.02926969800726e-05, "loss": 0.0065, "step": 32820 }, { "grad_norm": 0.16733548045158386, "learning_rate": 8.027954071663859e-05, "loss": 0.0097, "step": 32830 }, { "grad_norm": 0.12648513913154602, "learning_rate": 8.026638114187954e-05, "loss": 0.0076, "step": 32840 }, { "grad_norm": 0.08089247345924377, "learning_rate": 8.025321825723456e-05, "loss": 0.0066, "step": 32850 }, { "grad_norm": 0.09375122934579849, "learning_rate": 8.02400520641431e-05, "loss": 0.0069, "step": 32860 }, { "grad_norm": 0.08332456648349762, "learning_rate": 8.022688256404501e-05, "loss": 0.0082, "step": 32870 }, { "grad_norm": 0.11071076989173889, "learning_rate": 8.02137097583805e-05, "loss": 0.0108, "step": 32880 }, { "grad_norm": 0.11053387075662613, "learning_rate": 8.02005336485901e-05, "loss": 0.0063, "step": 32890 }, { "grad_norm": 0.12845061719417572, "learning_rate": 8.018735423611476e-05, "loss": 0.0087, "step": 32900 }, { "grad_norm": 0.0825246199965477, "learning_rate": 8.017417152239574e-05, "loss": 0.0062, "step": 32910 }, { "grad_norm": 0.13293175399303436, "learning_rate": 8.01609855088747e-05, "loss": 0.0091, "step": 32920 }, { "grad_norm": 0.14042900502681732, "learning_rate": 8.014779619699362e-05, "loss": 0.0072, "step": 32930 }, { "grad_norm": 0.09265702962875366, "learning_rate": 8.013460358819489e-05, "loss": 0.0094, "step": 32940 }, { "grad_norm": 0.0972234234213829, "learning_rate": 8.01214076839212e-05, "loss": 0.0099, "step": 32950 }, { "grad_norm": 0.21337242424488068, "learning_rate": 8.010820848561565e-05, "loss": 0.0084, "step": 32960 }, { "grad_norm": 0.12973180413246155, "learning_rate": 8.009500599472171e-05, "loss": 0.0086, "step": 32970 }, { "grad_norm": 0.21801993250846863, "learning_rate": 8.008180021268314e-05, "loss": 0.0085, "step": 32980 }, { "grad_norm": 0.10413660109043121, "learning_rate": 8.006859114094414e-05, "loss": 0.0095, "step": 32990 }, { "grad_norm": 0.12332499772310257, "learning_rate": 8.005537878094921e-05, "loss": 0.0088, "step": 33000 }, { "grad_norm": 0.11720956116914749, "learning_rate": 8.004216313414323e-05, "loss": 0.0068, "step": 33010 }, { "grad_norm": 0.14436961710453033, "learning_rate": 8.002894420197149e-05, "loss": 0.0092, "step": 33020 }, { "grad_norm": 0.10088036954402924, "learning_rate": 8.001572198587954e-05, "loss": 0.011, "step": 33030 }, { "grad_norm": 0.07965957373380661, "learning_rate": 8.000249648731338e-05, "loss": 0.0095, "step": 33040 }, { "grad_norm": 0.09012477099895477, "learning_rate": 7.998926770771928e-05, "loss": 0.009, "step": 33050 }, { "grad_norm": 0.07674670219421387, "learning_rate": 7.997603564854397e-05, "loss": 0.0088, "step": 33060 }, { "grad_norm": 0.22221767902374268, "learning_rate": 7.996280031123448e-05, "loss": 0.0102, "step": 33070 }, { "grad_norm": 0.10514072328805923, "learning_rate": 7.994956169723818e-05, "loss": 0.0107, "step": 33080 }, { "grad_norm": 0.09064310789108276, "learning_rate": 7.993631980800285e-05, "loss": 0.0105, "step": 33090 }, { "grad_norm": 0.09196671843528748, "learning_rate": 7.992307464497659e-05, "loss": 0.009, "step": 33100 }, { "grad_norm": 0.2015351951122284, "learning_rate": 7.990982620960787e-05, "loss": 0.009, "step": 33110 }, { "grad_norm": 0.12675660848617554, "learning_rate": 7.989657450334554e-05, "loss": 0.009, "step": 33120 }, { "grad_norm": 0.09253226965665817, "learning_rate": 7.988331952763877e-05, "loss": 0.0072, "step": 33130 }, { "grad_norm": 0.1555977165699005, "learning_rate": 7.987006128393709e-05, "loss": 0.0118, "step": 33140 }, { "grad_norm": 0.14834125339984894, "learning_rate": 7.985679977369043e-05, "loss": 0.0103, "step": 33150 }, { "grad_norm": 0.09207659214735031, "learning_rate": 7.984353499834902e-05, "loss": 0.0096, "step": 33160 }, { "grad_norm": 0.14047248661518097, "learning_rate": 7.983026695936351e-05, "loss": 0.0109, "step": 33170 }, { "grad_norm": 0.1290075182914734, "learning_rate": 7.981699565818486e-05, "loss": 0.0098, "step": 33180 }, { "grad_norm": 0.10986002534627914, "learning_rate": 7.980372109626437e-05, "loss": 0.0071, "step": 33190 }, { "grad_norm": 0.11532679200172424, "learning_rate": 7.979044327505375e-05, "loss": 0.0094, "step": 33200 }, { "grad_norm": 0.09718230366706848, "learning_rate": 7.977716219600506e-05, "loss": 0.0113, "step": 33210 }, { "grad_norm": 0.09921254962682724, "learning_rate": 7.97638778605707e-05, "loss": 0.0095, "step": 33220 }, { "grad_norm": 0.13331255316734314, "learning_rate": 7.975059027020338e-05, "loss": 0.0096, "step": 33230 }, { "grad_norm": 0.1355544775724411, "learning_rate": 7.973729942635623e-05, "loss": 0.0097, "step": 33240 }, { "grad_norm": 0.0919639840722084, "learning_rate": 7.972400533048273e-05, "loss": 0.0089, "step": 33250 }, { "grad_norm": 0.10460067540407181, "learning_rate": 7.97107079840367e-05, "loss": 0.0123, "step": 33260 }, { "grad_norm": 0.199758842587471, "learning_rate": 7.969740738847231e-05, "loss": 0.0099, "step": 33270 }, { "grad_norm": 0.14795906841754913, "learning_rate": 7.968410354524411e-05, "loss": 0.0094, "step": 33280 }, { "grad_norm": 0.16156521439552307, "learning_rate": 7.967079645580697e-05, "loss": 0.0085, "step": 33290 }, { "grad_norm": 0.10348358750343323, "learning_rate": 7.965748612161612e-05, "loss": 0.0069, "step": 33300 }, { "grad_norm": 0.1721479594707489, "learning_rate": 7.96441725441272e-05, "loss": 0.0102, "step": 33310 }, { "grad_norm": 0.13684344291687012, "learning_rate": 7.963085572479614e-05, "loss": 0.0074, "step": 33320 }, { "grad_norm": 0.12517355382442474, "learning_rate": 7.961753566507924e-05, "loss": 0.0124, "step": 33330 }, { "grad_norm": 0.24265842139720917, "learning_rate": 7.960421236643316e-05, "loss": 0.0087, "step": 33340 }, { "grad_norm": 0.1554027646780014, "learning_rate": 7.959088583031496e-05, "loss": 0.0077, "step": 33350 }, { "grad_norm": 0.08561460673809052, "learning_rate": 7.957755605818194e-05, "loss": 0.0068, "step": 33360 }, { "grad_norm": 0.24330946803092957, "learning_rate": 7.956422305149185e-05, "loss": 0.0091, "step": 33370 }, { "grad_norm": 0.14741170406341553, "learning_rate": 7.95508868117028e-05, "loss": 0.0097, "step": 33380 }, { "grad_norm": 0.12454840540885925, "learning_rate": 7.953754734027318e-05, "loss": 0.0087, "step": 33390 }, { "grad_norm": 0.13049568235874176, "learning_rate": 7.952420463866182e-05, "loss": 0.008, "step": 33400 }, { "grad_norm": 0.15658971667289734, "learning_rate": 7.951085870832782e-05, "loss": 0.0109, "step": 33410 }, { "grad_norm": 0.09202525019645691, "learning_rate": 7.949750955073067e-05, "loss": 0.0088, "step": 33420 }, { "grad_norm": 0.14031876623630524, "learning_rate": 7.948415716733022e-05, "loss": 0.0084, "step": 33430 }, { "grad_norm": 0.13755685091018677, "learning_rate": 7.947080155958669e-05, "loss": 0.0129, "step": 33440 }, { "grad_norm": 0.08115251362323761, "learning_rate": 7.94574427289606e-05, "loss": 0.0078, "step": 33450 }, { "grad_norm": 0.15702828764915466, "learning_rate": 7.944408067691284e-05, "loss": 0.0086, "step": 33460 }, { "grad_norm": 0.08981701731681824, "learning_rate": 7.943071540490473e-05, "loss": 0.0093, "step": 33470 }, { "grad_norm": 0.15056772530078888, "learning_rate": 7.94173469143978e-05, "loss": 0.0132, "step": 33480 }, { "grad_norm": 0.10599742829799652, "learning_rate": 7.940397520685406e-05, "loss": 0.0079, "step": 33490 }, { "grad_norm": 0.11717620491981506, "learning_rate": 7.939060028373577e-05, "loss": 0.0074, "step": 33500 }, { "grad_norm": 0.11674444377422333, "learning_rate": 7.937722214650565e-05, "loss": 0.0111, "step": 33510 }, { "grad_norm": 0.09401531517505646, "learning_rate": 7.936384079662666e-05, "loss": 0.0063, "step": 33520 }, { "grad_norm": 0.09744195640087128, "learning_rate": 7.93504562355622e-05, "loss": 0.0088, "step": 33530 }, { "grad_norm": 0.07674890011548996, "learning_rate": 7.933706846477599e-05, "loss": 0.0054, "step": 33540 }, { "grad_norm": 0.07069841772317886, "learning_rate": 7.932367748573206e-05, "loss": 0.0064, "step": 33550 }, { "grad_norm": 0.08085320144891739, "learning_rate": 7.931028329989485e-05, "loss": 0.0073, "step": 33560 }, { "grad_norm": 0.10471966117620468, "learning_rate": 7.929688590872913e-05, "loss": 0.0073, "step": 33570 }, { "grad_norm": 0.10712672770023346, "learning_rate": 7.928348531370003e-05, "loss": 0.0081, "step": 33580 }, { "grad_norm": 0.1658473014831543, "learning_rate": 7.927008151627297e-05, "loss": 0.01, "step": 33590 }, { "grad_norm": 0.07723267376422882, "learning_rate": 7.925667451791383e-05, "loss": 0.0074, "step": 33600 }, { "grad_norm": 0.1510164588689804, "learning_rate": 7.924326432008874e-05, "loss": 0.0104, "step": 33610 }, { "grad_norm": 0.08623582869768143, "learning_rate": 7.922985092426422e-05, "loss": 0.0074, "step": 33620 }, { "grad_norm": 0.17557893693447113, "learning_rate": 7.921643433190717e-05, "loss": 0.0101, "step": 33630 }, { "grad_norm": 0.08164992183446884, "learning_rate": 7.920301454448478e-05, "loss": 0.0084, "step": 33640 }, { "grad_norm": 0.14859656989574432, "learning_rate": 7.918959156346461e-05, "loss": 0.0109, "step": 33650 }, { "grad_norm": 0.10071857273578644, "learning_rate": 7.91761653903146e-05, "loss": 0.0085, "step": 33660 }, { "grad_norm": 0.08080220967531204, "learning_rate": 7.916273602650302e-05, "loss": 0.0077, "step": 33670 }, { "grad_norm": 0.06679883599281311, "learning_rate": 7.914930347349847e-05, "loss": 0.0062, "step": 33680 }, { "grad_norm": 0.10956926643848419, "learning_rate": 7.913586773276992e-05, "loss": 0.0095, "step": 33690 }, { "grad_norm": 0.15156619250774384, "learning_rate": 7.912242880578667e-05, "loss": 0.0076, "step": 33700 }, { "grad_norm": 0.10757657885551453, "learning_rate": 7.910898669401839e-05, "loss": 0.0101, "step": 33710 }, { "grad_norm": 0.1333191990852356, "learning_rate": 7.909554139893511e-05, "loss": 0.0086, "step": 33720 }, { "grad_norm": 0.12009499967098236, "learning_rate": 7.908209292200715e-05, "loss": 0.0078, "step": 33730 }, { "grad_norm": 0.16965219378471375, "learning_rate": 7.906864126470523e-05, "loss": 0.007, "step": 33740 }, { "grad_norm": 0.1024789810180664, "learning_rate": 7.905518642850041e-05, "loss": 0.0087, "step": 33750 }, { "grad_norm": 0.06774807721376419, "learning_rate": 7.904172841486409e-05, "loss": 0.0072, "step": 33760 }, { "grad_norm": 0.08035000413656235, "learning_rate": 7.902826722526801e-05, "loss": 0.0074, "step": 33770 }, { "grad_norm": 0.10062751919031143, "learning_rate": 7.901480286118427e-05, "loss": 0.0083, "step": 33780 }, { "grad_norm": 0.10721253603696823, "learning_rate": 7.900133532408531e-05, "loss": 0.0072, "step": 33790 }, { "grad_norm": 0.06572618335485458, "learning_rate": 7.898786461544395e-05, "loss": 0.009, "step": 33800 }, { "grad_norm": 0.06725125759840012, "learning_rate": 7.897439073673325e-05, "loss": 0.0076, "step": 33810 }, { "grad_norm": 0.1129264160990715, "learning_rate": 7.896091368942677e-05, "loss": 0.0067, "step": 33820 }, { "grad_norm": 0.13226298987865448, "learning_rate": 7.894743347499832e-05, "loss": 0.0096, "step": 33830 }, { "grad_norm": 0.09449688345193863, "learning_rate": 7.893395009492203e-05, "loss": 0.0092, "step": 33840 }, { "grad_norm": 0.10875225812196732, "learning_rate": 7.892046355067248e-05, "loss": 0.0093, "step": 33850 }, { "grad_norm": 0.06713800877332687, "learning_rate": 7.890697384372451e-05, "loss": 0.0067, "step": 33860 }, { "grad_norm": 0.0851481482386589, "learning_rate": 7.889348097555336e-05, "loss": 0.006, "step": 33870 }, { "grad_norm": 0.09992887824773788, "learning_rate": 7.887998494763455e-05, "loss": 0.0085, "step": 33880 }, { "grad_norm": 0.13363893330097198, "learning_rate": 7.886648576144404e-05, "loss": 0.0102, "step": 33890 }, { "grad_norm": 0.13277511298656464, "learning_rate": 7.885298341845802e-05, "loss": 0.01, "step": 33900 }, { "grad_norm": 0.10762478411197662, "learning_rate": 7.883947792015311e-05, "loss": 0.0071, "step": 33910 }, { "grad_norm": 0.1116594448685646, "learning_rate": 7.882596926800628e-05, "loss": 0.0086, "step": 33920 }, { "grad_norm": 0.13647493720054626, "learning_rate": 7.881245746349477e-05, "loss": 0.0125, "step": 33930 }, { "grad_norm": 0.12422939389944077, "learning_rate": 7.879894250809623e-05, "loss": 0.0074, "step": 33940 }, { "grad_norm": 0.09331702440977097, "learning_rate": 7.878542440328865e-05, "loss": 0.0066, "step": 33950 }, { "grad_norm": 0.10099990665912628, "learning_rate": 7.877190315055031e-05, "loss": 0.0079, "step": 33960 }, { "grad_norm": 0.10825908929109573, "learning_rate": 7.875837875135991e-05, "loss": 0.0085, "step": 33970 }, { "grad_norm": 0.11945152282714844, "learning_rate": 7.874485120719646e-05, "loss": 0.0108, "step": 33980 }, { "grad_norm": 0.1664886325597763, "learning_rate": 7.873132051953928e-05, "loss": 0.0094, "step": 33990 }, { "grad_norm": 0.1424439251422882, "learning_rate": 7.87177866898681e-05, "loss": 0.0098, "step": 34000 }, { "grad_norm": 0.12403646111488342, "learning_rate": 7.870424971966294e-05, "loss": 0.0062, "step": 34010 }, { "grad_norm": 0.18907544016838074, "learning_rate": 7.869070961040419e-05, "loss": 0.0113, "step": 34020 }, { "grad_norm": 0.1255308985710144, "learning_rate": 7.867716636357257e-05, "loss": 0.0112, "step": 34030 }, { "grad_norm": 0.18999899923801422, "learning_rate": 7.866361998064915e-05, "loss": 0.0095, "step": 34040 }, { "grad_norm": 0.10849270969629288, "learning_rate": 7.865007046311534e-05, "loss": 0.0087, "step": 34050 }, { "grad_norm": 0.12819638848304749, "learning_rate": 7.86365178124529e-05, "loss": 0.0103, "step": 34060 }, { "grad_norm": 0.21360817551612854, "learning_rate": 7.862296203014394e-05, "loss": 0.0084, "step": 34070 }, { "grad_norm": 0.1486465334892273, "learning_rate": 7.860940311767088e-05, "loss": 0.0092, "step": 34080 }, { "grad_norm": 0.13986094295978546, "learning_rate": 7.85958410765165e-05, "loss": 0.012, "step": 34090 }, { "grad_norm": 0.11753338575363159, "learning_rate": 7.858227590816394e-05, "loss": 0.0077, "step": 34100 }, { "grad_norm": 0.19618168473243713, "learning_rate": 7.856870761409664e-05, "loss": 0.0133, "step": 34110 }, { "grad_norm": 0.09325215965509415, "learning_rate": 7.855513619579846e-05, "loss": 0.0109, "step": 34120 }, { "grad_norm": 0.1043400689959526, "learning_rate": 7.85415616547535e-05, "loss": 0.0097, "step": 34130 }, { "grad_norm": 0.07691407948732376, "learning_rate": 7.852798399244627e-05, "loss": 0.0062, "step": 34140 }, { "grad_norm": 0.07573479413986206, "learning_rate": 7.851440321036161e-05, "loss": 0.009, "step": 34150 }, { "grad_norm": 0.07651841640472412, "learning_rate": 7.850081930998468e-05, "loss": 0.0066, "step": 34160 }, { "grad_norm": 0.12990924715995789, "learning_rate": 7.8487232292801e-05, "loss": 0.0095, "step": 34170 }, { "grad_norm": 0.0930420532822609, "learning_rate": 7.847364216029642e-05, "loss": 0.0092, "step": 34180 }, { "grad_norm": 0.08008788526058197, "learning_rate": 7.846004891395716e-05, "loss": 0.0072, "step": 34190 }, { "grad_norm": 0.10521607100963593, "learning_rate": 7.844645255526972e-05, "loss": 0.0069, "step": 34200 }, { "grad_norm": 0.1440340280532837, "learning_rate": 7.843285308572101e-05, "loss": 0.0081, "step": 34210 }, { "grad_norm": 0.18693509697914124, "learning_rate": 7.841925050679823e-05, "loss": 0.0085, "step": 34220 }, { "grad_norm": 0.1126875951886177, "learning_rate": 7.840564481998895e-05, "loss": 0.0097, "step": 34230 }, { "grad_norm": 0.12021292746067047, "learning_rate": 7.839203602678105e-05, "loss": 0.0088, "step": 34240 }, { "grad_norm": 0.09297098219394684, "learning_rate": 7.837842412866279e-05, "loss": 0.0083, "step": 34250 }, { "grad_norm": 0.10306069999933243, "learning_rate": 7.836480912712272e-05, "loss": 0.0088, "step": 34260 }, { "grad_norm": 0.0918264240026474, "learning_rate": 7.835119102364976e-05, "loss": 0.0077, "step": 34270 }, { "grad_norm": 0.21570342779159546, "learning_rate": 7.83375698197332e-05, "loss": 0.0083, "step": 34280 }, { "grad_norm": 0.15760961174964905, "learning_rate": 7.83239455168626e-05, "loss": 0.0091, "step": 34290 }, { "grad_norm": 0.1995164155960083, "learning_rate": 7.83103181165279e-05, "loss": 0.0078, "step": 34300 }, { "grad_norm": 0.10685025155544281, "learning_rate": 7.829668762021937e-05, "loss": 0.0073, "step": 34310 }, { "grad_norm": 0.24927203357219696, "learning_rate": 7.828305402942764e-05, "loss": 0.0114, "step": 34320 }, { "grad_norm": 0.1336195468902588, "learning_rate": 7.826941734564363e-05, "loss": 0.0101, "step": 34330 }, { "grad_norm": 0.10837852954864502, "learning_rate": 7.825577757035865e-05, "loss": 0.0082, "step": 34340 }, { "grad_norm": 0.08230547606945038, "learning_rate": 7.824213470506431e-05, "loss": 0.0081, "step": 34350 }, { "grad_norm": 0.10235603153705597, "learning_rate": 7.822848875125257e-05, "loss": 0.0069, "step": 34360 }, { "grad_norm": 0.0791499987244606, "learning_rate": 7.821483971041576e-05, "loss": 0.0069, "step": 34370 }, { "grad_norm": 0.10750503093004227, "learning_rate": 7.820118758404649e-05, "loss": 0.0074, "step": 34380 }, { "grad_norm": 0.12704579532146454, "learning_rate": 7.818753237363776e-05, "loss": 0.0127, "step": 34390 }, { "grad_norm": 0.13995735347270966, "learning_rate": 7.817387408068286e-05, "loss": 0.0069, "step": 34400 }, { "grad_norm": 0.13139300048351288, "learning_rate": 7.816021270667544e-05, "loss": 0.0072, "step": 34410 }, { "grad_norm": 0.10818792134523392, "learning_rate": 7.81465482531095e-05, "loss": 0.0063, "step": 34420 }, { "grad_norm": 0.09148187190294266, "learning_rate": 7.813288072147938e-05, "loss": 0.007, "step": 34430 }, { "grad_norm": 0.08155499398708344, "learning_rate": 7.811921011327972e-05, "loss": 0.0078, "step": 34440 }, { "grad_norm": 0.22512535750865936, "learning_rate": 7.810553643000549e-05, "loss": 0.0091, "step": 34450 }, { "grad_norm": 0.12909652292728424, "learning_rate": 7.809185967315206e-05, "loss": 0.0106, "step": 34460 }, { "grad_norm": 0.09186401218175888, "learning_rate": 7.80781798442151e-05, "loss": 0.0072, "step": 34470 }, { "grad_norm": 0.1404954493045807, "learning_rate": 7.806449694469061e-05, "loss": 0.0089, "step": 34480 }, { "grad_norm": 0.07370533049106598, "learning_rate": 7.805081097607492e-05, "loss": 0.0067, "step": 34490 }, { "grad_norm": 0.10332534462213516, "learning_rate": 7.803712193986474e-05, "loss": 0.0079, "step": 34500 }, { "grad_norm": 0.08389067649841309, "learning_rate": 7.802342983755702e-05, "loss": 0.0079, "step": 34510 }, { "grad_norm": 0.09503734111785889, "learning_rate": 7.800973467064918e-05, "loss": 0.0083, "step": 34520 }, { "grad_norm": 0.10721582919359207, "learning_rate": 7.799603644063884e-05, "loss": 0.0076, "step": 34530 }, { "grad_norm": 0.06874359399080276, "learning_rate": 7.798233514902405e-05, "loss": 0.0072, "step": 34540 }, { "grad_norm": 0.0826013907790184, "learning_rate": 7.796863079730318e-05, "loss": 0.0064, "step": 34550 }, { "grad_norm": 0.10694440454244614, "learning_rate": 7.795492338697488e-05, "loss": 0.0076, "step": 34560 }, { "grad_norm": 0.4961487054824829, "learning_rate": 7.794121291953819e-05, "loss": 0.0098, "step": 34570 }, { "grad_norm": 0.104887954890728, "learning_rate": 7.792749939649246e-05, "loss": 0.0151, "step": 34580 }, { "grad_norm": 0.204436793923378, "learning_rate": 7.79137828193374e-05, "loss": 0.0128, "step": 34590 }, { "grad_norm": 0.12416942417621613, "learning_rate": 7.790006318957301e-05, "loss": 0.011, "step": 34600 }, { "grad_norm": 0.09299580007791519, "learning_rate": 7.788634050869965e-05, "loss": 0.0114, "step": 34610 }, { "grad_norm": 0.09495510905981064, "learning_rate": 7.787261477821803e-05, "loss": 0.0105, "step": 34620 }, { "grad_norm": 0.14274832606315613, "learning_rate": 7.785888599962916e-05, "loss": 0.0108, "step": 34630 }, { "grad_norm": 0.11927446722984314, "learning_rate": 7.784515417443439e-05, "loss": 0.011, "step": 34640 }, { "grad_norm": 0.102451853454113, "learning_rate": 7.783141930413545e-05, "loss": 0.0122, "step": 34650 }, { "grad_norm": 0.20801061391830444, "learning_rate": 7.78176813902343e-05, "loss": 0.007, "step": 34660 }, { "grad_norm": 0.15145207941532135, "learning_rate": 7.780394043423336e-05, "loss": 0.0088, "step": 34670 }, { "grad_norm": 0.16157697141170502, "learning_rate": 7.77901964376353e-05, "loss": 0.0094, "step": 34680 }, { "grad_norm": 0.13815028965473175, "learning_rate": 7.777644940194316e-05, "loss": 0.0096, "step": 34690 }, { "grad_norm": 0.08415434509515762, "learning_rate": 7.776269932866023e-05, "loss": 0.0092, "step": 34700 }, { "grad_norm": 0.07713495939970016, "learning_rate": 7.774894621929026e-05, "loss": 0.0073, "step": 34710 }, { "grad_norm": 0.11745363473892212, "learning_rate": 7.773519007533725e-05, "loss": 0.0089, "step": 34720 }, { "grad_norm": 0.09748024493455887, "learning_rate": 7.772143089830556e-05, "loss": 0.0079, "step": 34730 }, { "grad_norm": 0.09571143239736557, "learning_rate": 7.770766868969985e-05, "loss": 0.0088, "step": 34740 }, { "grad_norm": 0.11057436466217041, "learning_rate": 7.769390345102518e-05, "loss": 0.0101, "step": 34750 }, { "grad_norm": 0.0981987789273262, "learning_rate": 7.768013518378683e-05, "loss": 0.008, "step": 34760 }, { "grad_norm": 0.21540535986423492, "learning_rate": 7.766636388949053e-05, "loss": 0.0096, "step": 34770 }, { "grad_norm": 0.08225993812084198, "learning_rate": 7.765258956964229e-05, "loss": 0.0068, "step": 34780 }, { "grad_norm": 0.0729193314909935, "learning_rate": 7.76388122257484e-05, "loss": 0.0068, "step": 34790 }, { "grad_norm": 0.17435120046138763, "learning_rate": 7.762503185931558e-05, "loss": 0.0069, "step": 34800 }, { "grad_norm": 0.12768743932247162, "learning_rate": 7.76112484718508e-05, "loss": 0.0109, "step": 34810 }, { "grad_norm": 0.2261008322238922, "learning_rate": 7.75974620648614e-05, "loss": 0.0089, "step": 34820 }, { "grad_norm": 0.1342516839504242, "learning_rate": 7.758367263985503e-05, "loss": 0.0105, "step": 34830 }, { "grad_norm": 0.23454566299915314, "learning_rate": 7.75698801983397e-05, "loss": 0.0086, "step": 34840 }, { "grad_norm": 0.06016826257109642, "learning_rate": 7.755608474182372e-05, "loss": 0.0082, "step": 34850 }, { "grad_norm": 0.15789207816123962, "learning_rate": 7.754228627181574e-05, "loss": 0.0078, "step": 34860 }, { "grad_norm": 0.15788787603378296, "learning_rate": 7.752848478982476e-05, "loss": 0.0107, "step": 34870 }, { "grad_norm": 0.11265634000301361, "learning_rate": 7.751468029736006e-05, "loss": 0.0083, "step": 34880 }, { "grad_norm": 0.0899643748998642, "learning_rate": 7.750087279593129e-05, "loss": 0.0082, "step": 34890 }, { "grad_norm": 0.08349228650331497, "learning_rate": 7.748706228704843e-05, "loss": 0.006, "step": 34900 }, { "grad_norm": 0.17324547469615936, "learning_rate": 7.747324877222176e-05, "loss": 0.0075, "step": 34910 }, { "grad_norm": 0.07985891401767731, "learning_rate": 7.745943225296188e-05, "loss": 0.0073, "step": 34920 }, { "grad_norm": 0.10304027795791626, "learning_rate": 7.744561273077981e-05, "loss": 0.0077, "step": 34930 }, { "grad_norm": 0.09014730155467987, "learning_rate": 7.743179020718678e-05, "loss": 0.0088, "step": 34940 }, { "grad_norm": 0.12996622920036316, "learning_rate": 7.741796468369443e-05, "loss": 0.0081, "step": 34950 }, { "grad_norm": 0.10929817706346512, "learning_rate": 7.740413616181466e-05, "loss": 0.0075, "step": 34960 }, { "grad_norm": 0.07498990744352341, "learning_rate": 7.739030464305978e-05, "loss": 0.0064, "step": 34970 }, { "grad_norm": 0.18658210337162018, "learning_rate": 7.737647012894235e-05, "loss": 0.012, "step": 34980 }, { "grad_norm": 0.12750369310379028, "learning_rate": 7.736263262097532e-05, "loss": 0.0094, "step": 34990 }, { "grad_norm": 0.19930629432201385, "learning_rate": 7.734879212067192e-05, "loss": 0.008, "step": 35000 }, { "grad_norm": 0.11090801656246185, "learning_rate": 7.733494862954573e-05, "loss": 0.0119, "step": 35010 }, { "grad_norm": 0.1115645095705986, "learning_rate": 7.732110214911066e-05, "loss": 0.0075, "step": 35020 }, { "grad_norm": 0.08390175551176071, "learning_rate": 7.730725268088094e-05, "loss": 0.0078, "step": 35030 }, { "grad_norm": 0.10389872640371323, "learning_rate": 7.729340022637112e-05, "loss": 0.0073, "step": 35040 }, { "grad_norm": 0.07240797579288483, "learning_rate": 7.727954478709607e-05, "loss": 0.0059, "step": 35050 }, { "grad_norm": 0.0652870386838913, "learning_rate": 7.726568636457103e-05, "loss": 0.0056, "step": 35060 }, { "grad_norm": 0.11465267837047577, "learning_rate": 7.725182496031153e-05, "loss": 0.0071, "step": 35070 }, { "grad_norm": 0.08568260818719864, "learning_rate": 7.723796057583341e-05, "loss": 0.0101, "step": 35080 }, { "grad_norm": 0.13873963057994843, "learning_rate": 7.722409321265287e-05, "loss": 0.0087, "step": 35090 }, { "grad_norm": 0.1236729547381401, "learning_rate": 7.721022287228645e-05, "loss": 0.0108, "step": 35100 }, { "grad_norm": 0.10416781902313232, "learning_rate": 7.719634955625094e-05, "loss": 0.0114, "step": 35110 }, { "grad_norm": 0.08792342245578766, "learning_rate": 7.718247326606355e-05, "loss": 0.0098, "step": 35120 }, { "grad_norm": 0.17219699919223785, "learning_rate": 7.716859400324175e-05, "loss": 0.0081, "step": 35130 }, { "grad_norm": 0.1439909189939499, "learning_rate": 7.715471176930335e-05, "loss": 0.0082, "step": 35140 }, { "grad_norm": 0.14946752786636353, "learning_rate": 7.714082656576651e-05, "loss": 0.0093, "step": 35150 }, { "grad_norm": 0.1686590164899826, "learning_rate": 7.712693839414968e-05, "loss": 0.0096, "step": 35160 }, { "grad_norm": 0.1015808954834938, "learning_rate": 7.711304725597164e-05, "loss": 0.0077, "step": 35170 }, { "grad_norm": 0.1085176169872284, "learning_rate": 7.709915315275151e-05, "loss": 0.0113, "step": 35180 }, { "grad_norm": 0.13159577548503876, "learning_rate": 7.708525608600876e-05, "loss": 0.011, "step": 35190 }, { "grad_norm": 0.0820453092455864, "learning_rate": 7.707135605726311e-05, "loss": 0.0111, "step": 35200 }, { "grad_norm": 0.0888393446803093, "learning_rate": 7.705745306803466e-05, "loss": 0.008, "step": 35210 }, { "grad_norm": 0.09574603289365768, "learning_rate": 7.704354711984383e-05, "loss": 0.0073, "step": 35220 }, { "grad_norm": 0.08456031233072281, "learning_rate": 7.702963821421133e-05, "loss": 0.0068, "step": 35230 }, { "grad_norm": 0.15053075551986694, "learning_rate": 7.701572635265826e-05, "loss": 0.0089, "step": 35240 }, { "grad_norm": 0.11521869152784348, "learning_rate": 7.700181153670596e-05, "loss": 0.0086, "step": 35250 }, { "grad_norm": 0.12936626374721527, "learning_rate": 7.698789376787616e-05, "loss": 0.0087, "step": 35260 }, { "grad_norm": 0.09766343235969543, "learning_rate": 7.697397304769085e-05, "loss": 0.0079, "step": 35270 }, { "grad_norm": 0.14467047154903412, "learning_rate": 7.696004937767241e-05, "loss": 0.0077, "step": 35280 }, { "grad_norm": 0.07474422454833984, "learning_rate": 7.694612275934352e-05, "loss": 0.0081, "step": 35290 }, { "grad_norm": 0.21145473420619965, "learning_rate": 7.693219319422714e-05, "loss": 0.0108, "step": 35300 }, { "grad_norm": 0.11421375721693039, "learning_rate": 7.69182606838466e-05, "loss": 0.0096, "step": 35310 }, { "grad_norm": 0.20922091603279114, "learning_rate": 7.690432522972558e-05, "loss": 0.011, "step": 35320 }, { "grad_norm": 0.11096680909395218, "learning_rate": 7.689038683338796e-05, "loss": 0.0102, "step": 35330 }, { "grad_norm": 0.12159036844968796, "learning_rate": 7.687644549635808e-05, "loss": 0.0095, "step": 35340 }, { "grad_norm": 0.15017260611057281, "learning_rate": 7.686250122016053e-05, "loss": 0.009, "step": 35350 }, { "grad_norm": 0.1099122017621994, "learning_rate": 7.684855400632022e-05, "loss": 0.0084, "step": 35360 }, { "grad_norm": 0.12095183879137039, "learning_rate": 7.683460385636243e-05, "loss": 0.0108, "step": 35370 }, { "grad_norm": 0.3418296277523041, "learning_rate": 7.68206507718127e-05, "loss": 0.0073, "step": 35380 }, { "grad_norm": 0.169776052236557, "learning_rate": 7.680669475419693e-05, "loss": 0.0114, "step": 35390 }, { "grad_norm": 0.24688886106014252, "learning_rate": 7.679273580504132e-05, "loss": 0.0102, "step": 35400 }, { "grad_norm": 0.09962339699268341, "learning_rate": 7.67787739258724e-05, "loss": 0.0082, "step": 35410 }, { "grad_norm": 0.21383869647979736, "learning_rate": 7.676480911821705e-05, "loss": 0.011, "step": 35420 }, { "grad_norm": 0.09156090021133423, "learning_rate": 7.675084138360238e-05, "loss": 0.0094, "step": 35430 }, { "grad_norm": 0.17427632212638855, "learning_rate": 7.673687072355592e-05, "loss": 0.0122, "step": 35440 }, { "grad_norm": 0.10004299134016037, "learning_rate": 7.67228971396055e-05, "loss": 0.0087, "step": 35450 }, { "grad_norm": 0.202493816614151, "learning_rate": 7.670892063327922e-05, "loss": 0.0109, "step": 35460 }, { "grad_norm": 0.14350947737693787, "learning_rate": 7.669494120610552e-05, "loss": 0.0098, "step": 35470 }, { "grad_norm": 0.1805928498506546, "learning_rate": 7.66809588596132e-05, "loss": 0.009, "step": 35480 }, { "grad_norm": 0.18424759805202484, "learning_rate": 7.666697359533132e-05, "loss": 0.0058, "step": 35490 }, { "grad_norm": 0.09002691507339478, "learning_rate": 7.665298541478932e-05, "loss": 0.0095, "step": 35500 }, { "grad_norm": 0.15351900458335876, "learning_rate": 7.66389943195169e-05, "loss": 0.0113, "step": 35510 }, { "grad_norm": 0.18040794134140015, "learning_rate": 7.66250003110441e-05, "loss": 0.0107, "step": 35520 }, { "grad_norm": 0.11537184566259384, "learning_rate": 7.66110033909013e-05, "loss": 0.0081, "step": 35530 }, { "grad_norm": 0.10370248556137085, "learning_rate": 7.659700356061918e-05, "loss": 0.0081, "step": 35540 }, { "grad_norm": 0.12557029724121094, "learning_rate": 7.658300082172875e-05, "loss": 0.0078, "step": 35550 }, { "grad_norm": 0.10714498162269592, "learning_rate": 7.65689951757613e-05, "loss": 0.0085, "step": 35560 }, { "grad_norm": 0.08768297731876373, "learning_rate": 7.655498662424849e-05, "loss": 0.0091, "step": 35570 }, { "grad_norm": 0.09983183443546295, "learning_rate": 7.654097516872227e-05, "loss": 0.0069, "step": 35580 }, { "grad_norm": 0.11843448132276535, "learning_rate": 7.65269608107149e-05, "loss": 0.008, "step": 35590 }, { "grad_norm": 0.1325942575931549, "learning_rate": 7.6512943551759e-05, "loss": 0.0075, "step": 35600 }, { "grad_norm": 0.08397350460290909, "learning_rate": 7.649892339338743e-05, "loss": 0.0077, "step": 35610 }, { "grad_norm": 0.0971820056438446, "learning_rate": 7.648490033713344e-05, "loss": 0.0093, "step": 35620 }, { "grad_norm": 0.1489066630601883, "learning_rate": 7.647087438453058e-05, "loss": 0.0109, "step": 35630 }, { "grad_norm": 0.11910929530858994, "learning_rate": 7.64568455371127e-05, "loss": 0.0112, "step": 35640 }, { "grad_norm": 0.12351210415363312, "learning_rate": 7.644281379641396e-05, "loss": 0.0082, "step": 35650 }, { "grad_norm": 0.08502305299043655, "learning_rate": 7.642877916396887e-05, "loss": 0.0082, "step": 35660 }, { "grad_norm": 0.17572952806949615, "learning_rate": 7.641474164131221e-05, "loss": 0.0087, "step": 35670 }, { "grad_norm": 0.1233021542429924, "learning_rate": 7.640070122997913e-05, "loss": 0.0085, "step": 35680 }, { "grad_norm": 0.15425381064414978, "learning_rate": 7.638665793150505e-05, "loss": 0.0102, "step": 35690 }, { "grad_norm": 0.1113448515534401, "learning_rate": 7.637261174742574e-05, "loss": 0.0069, "step": 35700 }, { "grad_norm": 0.09409873932600021, "learning_rate": 7.635856267927725e-05, "loss": 0.006, "step": 35710 }, { "grad_norm": 0.10723233222961426, "learning_rate": 7.634451072859597e-05, "loss": 0.0115, "step": 35720 }, { "grad_norm": 0.1305248737335205, "learning_rate": 7.633045589691863e-05, "loss": 0.0097, "step": 35730 }, { "grad_norm": 0.12965629994869232, "learning_rate": 7.63163981857822e-05, "loss": 0.0113, "step": 35740 }, { "grad_norm": 0.1411678045988083, "learning_rate": 7.630233759672403e-05, "loss": 0.0088, "step": 35750 }, { "grad_norm": 0.08288991451263428, "learning_rate": 7.628827413128177e-05, "loss": 0.0066, "step": 35760 }, { "grad_norm": 0.09795473515987396, "learning_rate": 7.627420779099338e-05, "loss": 0.0098, "step": 35770 }, { "grad_norm": 0.0988519936800003, "learning_rate": 7.626013857739711e-05, "loss": 0.0059, "step": 35780 }, { "grad_norm": 0.08406834304332733, "learning_rate": 7.624606649203158e-05, "loss": 0.0066, "step": 35790 }, { "grad_norm": 0.09546440839767456, "learning_rate": 7.623199153643569e-05, "loss": 0.0082, "step": 35800 }, { "grad_norm": 0.15963052213191986, "learning_rate": 7.621791371214863e-05, "loss": 0.0104, "step": 35810 }, { "grad_norm": 0.11400724202394485, "learning_rate": 7.620383302070995e-05, "loss": 0.0075, "step": 35820 }, { "grad_norm": 0.07061773538589478, "learning_rate": 7.61897494636595e-05, "loss": 0.0065, "step": 35830 }, { "grad_norm": 0.08881872147321701, "learning_rate": 7.617566304253739e-05, "loss": 0.0075, "step": 35840 }, { "grad_norm": 0.09770427644252777, "learning_rate": 7.616157375888416e-05, "loss": 0.0083, "step": 35850 }, { "grad_norm": 0.09623198211193085, "learning_rate": 7.614748161424053e-05, "loss": 0.0052, "step": 35860 }, { "grad_norm": 0.08050979673862457, "learning_rate": 7.613338661014763e-05, "loss": 0.0092, "step": 35870 }, { "grad_norm": 0.1513061672449112, "learning_rate": 7.611928874814686e-05, "loss": 0.0075, "step": 35880 }, { "grad_norm": 0.10185951739549637, "learning_rate": 7.610518802977996e-05, "loss": 0.0077, "step": 35890 }, { "grad_norm": 0.07880763709545135, "learning_rate": 7.609108445658893e-05, "loss": 0.0069, "step": 35900 }, { "grad_norm": 0.1082354262471199, "learning_rate": 7.607697803011612e-05, "loss": 0.0097, "step": 35910 }, { "grad_norm": 0.12097165733575821, "learning_rate": 7.606286875190421e-05, "loss": 0.0066, "step": 35920 }, { "grad_norm": 0.06409462541341782, "learning_rate": 7.604875662349617e-05, "loss": 0.0073, "step": 35930 }, { "grad_norm": 0.10481008887290955, "learning_rate": 7.603464164643525e-05, "loss": 0.0091, "step": 35940 }, { "grad_norm": 0.11183230578899384, "learning_rate": 7.602052382226507e-05, "loss": 0.008, "step": 35950 }, { "grad_norm": 0.09210187196731567, "learning_rate": 7.600640315252954e-05, "loss": 0.0062, "step": 35960 }, { "grad_norm": 0.0858895555138588, "learning_rate": 7.599227963877284e-05, "loss": 0.0078, "step": 35970 }, { "grad_norm": 0.18090258538722992, "learning_rate": 7.597815328253954e-05, "loss": 0.0079, "step": 35980 }, { "grad_norm": 0.11020450294017792, "learning_rate": 7.596402408537443e-05, "loss": 0.0083, "step": 35990 }, { "grad_norm": 0.08510739356279373, "learning_rate": 7.59498920488227e-05, "loss": 0.0086, "step": 36000 }, { "grad_norm": 0.09057163447141647, "learning_rate": 7.593575717442979e-05, "loss": 0.0069, "step": 36010 }, { "grad_norm": 0.10002882033586502, "learning_rate": 7.592161946374147e-05, "loss": 0.0068, "step": 36020 }, { "grad_norm": 0.07390019297599792, "learning_rate": 7.590747891830381e-05, "loss": 0.009, "step": 36030 }, { "grad_norm": 0.0586966946721077, "learning_rate": 7.58933355396632e-05, "loss": 0.0071, "step": 36040 }, { "grad_norm": 0.08122901618480682, "learning_rate": 7.587918932936636e-05, "loss": 0.0061, "step": 36050 }, { "grad_norm": 0.050979696214199066, "learning_rate": 7.586504028896028e-05, "loss": 0.0078, "step": 36060 }, { "grad_norm": 0.129257932305336, "learning_rate": 7.585088841999228e-05, "loss": 0.0097, "step": 36070 }, { "grad_norm": 0.1251709908246994, "learning_rate": 7.583673372400999e-05, "loss": 0.0082, "step": 36080 }, { "grad_norm": 0.08898293972015381, "learning_rate": 7.582257620256134e-05, "loss": 0.0088, "step": 36090 }, { "grad_norm": 0.09334811568260193, "learning_rate": 7.580841585719458e-05, "loss": 0.0089, "step": 36100 }, { "grad_norm": 0.07091926783323288, "learning_rate": 7.579425268945825e-05, "loss": 0.0078, "step": 36110 }, { "grad_norm": 0.11769834905862808, "learning_rate": 7.578008670090127e-05, "loss": 0.008, "step": 36120 }, { "grad_norm": 0.09379711747169495, "learning_rate": 7.576591789307272e-05, "loss": 0.008, "step": 36130 }, { "grad_norm": 0.0956188216805458, "learning_rate": 7.575174626752216e-05, "loss": 0.0063, "step": 36140 }, { "grad_norm": 0.12822259962558746, "learning_rate": 7.573757182579934e-05, "loss": 0.0093, "step": 36150 }, { "grad_norm": 0.11423684656620026, "learning_rate": 7.572339456945435e-05, "loss": 0.0075, "step": 36160 }, { "grad_norm": 0.08658719062805176, "learning_rate": 7.570921450003762e-05, "loss": 0.0098, "step": 36170 }, { "grad_norm": 0.0960431694984436, "learning_rate": 7.569503161909983e-05, "loss": 0.0075, "step": 36180 }, { "grad_norm": 0.11032769829034805, "learning_rate": 7.568084592819202e-05, "loss": 0.0077, "step": 36190 }, { "grad_norm": 0.11892127245664597, "learning_rate": 7.566665742886551e-05, "loss": 0.0065, "step": 36200 }, { "grad_norm": 0.0811859667301178, "learning_rate": 7.565246612267194e-05, "loss": 0.0058, "step": 36210 }, { "grad_norm": 0.07677141577005386, "learning_rate": 7.563827201116325e-05, "loss": 0.0058, "step": 36220 }, { "grad_norm": 0.11136874556541443, "learning_rate": 7.562407509589164e-05, "loss": 0.0098, "step": 36230 }, { "grad_norm": 0.12373854219913483, "learning_rate": 7.560987537840973e-05, "loss": 0.0093, "step": 36240 }, { "grad_norm": 0.11502423882484436, "learning_rate": 7.559567286027036e-05, "loss": 0.0082, "step": 36250 }, { "grad_norm": 0.17371466755867004, "learning_rate": 7.558146754302668e-05, "loss": 0.0086, "step": 36260 }, { "grad_norm": 0.13212275505065918, "learning_rate": 7.556725942823217e-05, "loss": 0.0098, "step": 36270 }, { "grad_norm": 0.08371306955814362, "learning_rate": 7.555304851744061e-05, "loss": 0.0089, "step": 36280 }, { "grad_norm": 0.09468469768762589, "learning_rate": 7.553883481220608e-05, "loss": 0.0072, "step": 36290 }, { "grad_norm": 0.07392705231904984, "learning_rate": 7.552461831408298e-05, "loss": 0.0068, "step": 36300 }, { "grad_norm": 0.10475967079401016, "learning_rate": 7.551039902462599e-05, "loss": 0.0079, "step": 36310 }, { "grad_norm": 0.15458714962005615, "learning_rate": 7.549617694539011e-05, "loss": 0.0067, "step": 36320 }, { "grad_norm": 0.12864093482494354, "learning_rate": 7.548195207793066e-05, "loss": 0.0075, "step": 36330 }, { "grad_norm": 0.11242195963859558, "learning_rate": 7.546772442380323e-05, "loss": 0.0083, "step": 36340 }, { "grad_norm": 0.10722086578607559, "learning_rate": 7.545349398456376e-05, "loss": 0.0073, "step": 36350 }, { "grad_norm": 0.10122663527727127, "learning_rate": 7.543926076176845e-05, "loss": 0.0077, "step": 36360 }, { "grad_norm": 0.13371239602565765, "learning_rate": 7.542502475697385e-05, "loss": 0.0091, "step": 36370 }, { "grad_norm": 0.1306959092617035, "learning_rate": 7.541078597173675e-05, "loss": 0.0085, "step": 36380 }, { "grad_norm": 0.12109065800905228, "learning_rate": 7.539654440761431e-05, "loss": 0.0094, "step": 36390 }, { "grad_norm": 0.08469272404909134, "learning_rate": 7.538230006616395e-05, "loss": 0.0072, "step": 36400 }, { "grad_norm": 0.10314090549945831, "learning_rate": 7.536805294894342e-05, "loss": 0.0076, "step": 36410 }, { "grad_norm": 0.06383778899908066, "learning_rate": 7.535380305751076e-05, "loss": 0.0068, "step": 36420 }, { "grad_norm": 0.08551005274057388, "learning_rate": 7.533955039342431e-05, "loss": 0.0092, "step": 36430 }, { "grad_norm": 0.09729611873626709, "learning_rate": 7.532529495824274e-05, "loss": 0.0074, "step": 36440 }, { "grad_norm": 0.1293598860502243, "learning_rate": 7.531103675352497e-05, "loss": 0.0088, "step": 36450 }, { "grad_norm": 0.10325893014669418, "learning_rate": 7.529677578083029e-05, "loss": 0.0062, "step": 36460 }, { "grad_norm": 0.10434333235025406, "learning_rate": 7.528251204171823e-05, "loss": 0.0064, "step": 36470 }, { "grad_norm": 0.08889938145875931, "learning_rate": 7.526824553774867e-05, "loss": 0.007, "step": 36480 }, { "grad_norm": 0.15051957964897156, "learning_rate": 7.525397627048178e-05, "loss": 0.0076, "step": 36490 }, { "grad_norm": 0.14036010205745697, "learning_rate": 7.523970424147802e-05, "loss": 0.0062, "step": 36500 }, { "grad_norm": 0.11247646808624268, "learning_rate": 7.522542945229813e-05, "loss": 0.0051, "step": 36510 }, { "grad_norm": 0.1733456254005432, "learning_rate": 7.521115190450321e-05, "loss": 0.0064, "step": 36520 }, { "grad_norm": 0.0821315348148346, "learning_rate": 7.519687159965462e-05, "loss": 0.0068, "step": 36530 }, { "grad_norm": 0.13536818325519562, "learning_rate": 7.518258853931403e-05, "loss": 0.0099, "step": 36540 }, { "grad_norm": 0.11340636759996414, "learning_rate": 7.516830272504342e-05, "loss": 0.0069, "step": 36550 }, { "grad_norm": 0.06028134748339653, "learning_rate": 7.515401415840509e-05, "loss": 0.0064, "step": 36560 }, { "grad_norm": 0.1287551373243332, "learning_rate": 7.513972284096155e-05, "loss": 0.006, "step": 36570 }, { "grad_norm": 0.07694990187883377, "learning_rate": 7.512542877427574e-05, "loss": 0.0102, "step": 36580 }, { "grad_norm": 0.1349007934331894, "learning_rate": 7.51111319599108e-05, "loss": 0.0096, "step": 36590 }, { "grad_norm": 0.12590086460113525, "learning_rate": 7.509683239943024e-05, "loss": 0.0079, "step": 36600 }, { "grad_norm": 0.184273362159729, "learning_rate": 7.50825300943978e-05, "loss": 0.0095, "step": 36610 }, { "grad_norm": 0.10382778197526932, "learning_rate": 7.50682250463776e-05, "loss": 0.0084, "step": 36620 }, { "grad_norm": 0.10649165511131287, "learning_rate": 7.5053917256934e-05, "loss": 0.0064, "step": 36630 }, { "grad_norm": 0.10424718260765076, "learning_rate": 7.503960672763166e-05, "loss": 0.0063, "step": 36640 }, { "grad_norm": 0.10168680548667908, "learning_rate": 7.502529346003559e-05, "loss": 0.0071, "step": 36650 }, { "grad_norm": 0.1751803159713745, "learning_rate": 7.501097745571106e-05, "loss": 0.0066, "step": 36660 }, { "grad_norm": 0.11365510523319244, "learning_rate": 7.499665871622361e-05, "loss": 0.0073, "step": 36670 }, { "grad_norm": 0.1410752236843109, "learning_rate": 7.498233724313917e-05, "loss": 0.0083, "step": 36680 }, { "grad_norm": 0.1147836446762085, "learning_rate": 7.496801303802387e-05, "loss": 0.0084, "step": 36690 }, { "grad_norm": 0.06423182040452957, "learning_rate": 7.495368610244423e-05, "loss": 0.0073, "step": 36700 }, { "grad_norm": 0.1584523767232895, "learning_rate": 7.493935643796697e-05, "loss": 0.0081, "step": 36710 }, { "grad_norm": 0.13636943697929382, "learning_rate": 7.49250240461592e-05, "loss": 0.0091, "step": 36720 }, { "grad_norm": 0.1689254641532898, "learning_rate": 7.491068892858826e-05, "loss": 0.0083, "step": 36730 }, { "grad_norm": 0.10578036308288574, "learning_rate": 7.489635108682184e-05, "loss": 0.0078, "step": 36740 }, { "grad_norm": 0.0699324905872345, "learning_rate": 7.48820105224279e-05, "loss": 0.0079, "step": 36750 }, { "grad_norm": 0.09713312983512878, "learning_rate": 7.48676672369747e-05, "loss": 0.0105, "step": 36760 }, { "grad_norm": 0.09241382032632828, "learning_rate": 7.485332123203079e-05, "loss": 0.0062, "step": 36770 }, { "grad_norm": 0.13300195336341858, "learning_rate": 7.483897250916504e-05, "loss": 0.0075, "step": 36780 }, { "grad_norm": 0.10397949069738388, "learning_rate": 7.48246210699466e-05, "loss": 0.0079, "step": 36790 }, { "grad_norm": 0.0945972427725792, "learning_rate": 7.481026691594492e-05, "loss": 0.0082, "step": 36800 }, { "grad_norm": 0.06904192268848419, "learning_rate": 7.479591004872975e-05, "loss": 0.0069, "step": 36810 }, { "grad_norm": 0.09362324327230453, "learning_rate": 7.478155046987114e-05, "loss": 0.0091, "step": 36820 }, { "grad_norm": 0.1469176709651947, "learning_rate": 7.476718818093944e-05, "loss": 0.0101, "step": 36830 }, { "grad_norm": 0.1327422559261322, "learning_rate": 7.475282318350527e-05, "loss": 0.0083, "step": 36840 }, { "grad_norm": 0.21942877769470215, "learning_rate": 7.473845547913957e-05, "loss": 0.0098, "step": 36850 }, { "grad_norm": 0.14848804473876953, "learning_rate": 7.472408506941357e-05, "loss": 0.0091, "step": 36860 }, { "grad_norm": 0.14432917535305023, "learning_rate": 7.470971195589881e-05, "loss": 0.0095, "step": 36870 }, { "grad_norm": 0.08530421555042267, "learning_rate": 7.46953361401671e-05, "loss": 0.0081, "step": 36880 }, { "grad_norm": 0.13850729167461395, "learning_rate": 7.468095762379055e-05, "loss": 0.008, "step": 36890 }, { "grad_norm": 0.12570764124393463, "learning_rate": 7.466657640834158e-05, "loss": 0.0071, "step": 36900 }, { "grad_norm": 0.09898924082517624, "learning_rate": 7.46521924953929e-05, "loss": 0.0106, "step": 36910 }, { "grad_norm": 0.12983758747577667, "learning_rate": 7.463780588651752e-05, "loss": 0.0068, "step": 36920 }, { "grad_norm": 0.0809074118733406, "learning_rate": 7.462341658328873e-05, "loss": 0.0061, "step": 36930 }, { "grad_norm": 0.17545634508132935, "learning_rate": 7.460902458728012e-05, "loss": 0.0088, "step": 36940 }, { "grad_norm": 0.12045889347791672, "learning_rate": 7.459462990006558e-05, "loss": 0.0108, "step": 36950 }, { "grad_norm": 0.07667823135852814, "learning_rate": 7.458023252321929e-05, "loss": 0.0124, "step": 36960 }, { "grad_norm": 0.06883683055639267, "learning_rate": 7.456583245831573e-05, "loss": 0.0078, "step": 36970 }, { "grad_norm": 0.14009065926074982, "learning_rate": 7.455142970692968e-05, "loss": 0.0087, "step": 36980 }, { "grad_norm": 0.10416001826524734, "learning_rate": 7.453702427063615e-05, "loss": 0.0073, "step": 36990 }, { "grad_norm": 0.1025748997926712, "learning_rate": 7.452261615101057e-05, "loss": 0.0092, "step": 37000 }, { "grad_norm": 0.14410331845283508, "learning_rate": 7.450820534962856e-05, "loss": 0.0084, "step": 37010 }, { "grad_norm": 0.11322936415672302, "learning_rate": 7.449379186806604e-05, "loss": 0.0069, "step": 37020 }, { "grad_norm": 0.13835926353931427, "learning_rate": 7.44793757078993e-05, "loss": 0.008, "step": 37030 }, { "grad_norm": 0.0885825827717781, "learning_rate": 7.44649568707048e-05, "loss": 0.0099, "step": 37040 }, { "grad_norm": 0.14018243551254272, "learning_rate": 7.445053535805942e-05, "loss": 0.011, "step": 37050 }, { "grad_norm": 0.19818773865699768, "learning_rate": 7.443611117154024e-05, "loss": 0.0111, "step": 37060 }, { "grad_norm": 0.14142002165317535, "learning_rate": 7.442168431272469e-05, "loss": 0.009, "step": 37070 }, { "grad_norm": 0.13163791596889496, "learning_rate": 7.440725478319044e-05, "loss": 0.0089, "step": 37080 }, { "grad_norm": 0.13128511607646942, "learning_rate": 7.439282258451553e-05, "loss": 0.0098, "step": 37090 }, { "grad_norm": 0.16125939786434174, "learning_rate": 7.43783877182782e-05, "loss": 0.0097, "step": 37100 }, { "grad_norm": 0.12137509137392044, "learning_rate": 7.436395018605707e-05, "loss": 0.0074, "step": 37110 }, { "grad_norm": 0.14044387638568878, "learning_rate": 7.434950998943094e-05, "loss": 0.0076, "step": 37120 }, { "grad_norm": 0.12561200559139252, "learning_rate": 7.433506712997902e-05, "loss": 0.0085, "step": 37130 }, { "grad_norm": 0.08485578745603561, "learning_rate": 7.432062160928077e-05, "loss": 0.0066, "step": 37140 }, { "grad_norm": 0.10416287183761597, "learning_rate": 7.430617342891588e-05, "loss": 0.0084, "step": 37150 }, { "grad_norm": 0.0852966383099556, "learning_rate": 7.429172259046442e-05, "loss": 0.0078, "step": 37160 }, { "grad_norm": 0.09549131989479065, "learning_rate": 7.42772690955067e-05, "loss": 0.0089, "step": 37170 }, { "grad_norm": 0.13546325266361237, "learning_rate": 7.426281294562332e-05, "loss": 0.009, "step": 37180 }, { "grad_norm": 0.1266387701034546, "learning_rate": 7.424835414239519e-05, "loss": 0.0104, "step": 37190 }, { "grad_norm": 0.14396928250789642, "learning_rate": 7.423389268740352e-05, "loss": 0.0091, "step": 37200 }, { "grad_norm": 0.08806516975164413, "learning_rate": 7.42194285822298e-05, "loss": 0.0108, "step": 37210 }, { "grad_norm": 0.08426797389984131, "learning_rate": 7.420496182845576e-05, "loss": 0.008, "step": 37220 }, { "grad_norm": 0.08236280828714371, "learning_rate": 7.419049242766348e-05, "loss": 0.0086, "step": 37230 }, { "grad_norm": 0.08086343854665756, "learning_rate": 7.417602038143533e-05, "loss": 0.0057, "step": 37240 }, { "grad_norm": 0.08017671853303909, "learning_rate": 7.416154569135393e-05, "loss": 0.0073, "step": 37250 }, { "grad_norm": 0.20643045008182526, "learning_rate": 7.414706835900222e-05, "loss": 0.0082, "step": 37260 }, { "grad_norm": 0.1545475572347641, "learning_rate": 7.413258838596345e-05, "loss": 0.0091, "step": 37270 }, { "grad_norm": 0.30659806728363037, "learning_rate": 7.411810577382106e-05, "loss": 0.012, "step": 37280 }, { "grad_norm": 0.1336624026298523, "learning_rate": 7.410362052415889e-05, "loss": 0.0075, "step": 37290 }, { "grad_norm": 0.10781055688858032, "learning_rate": 7.408913263856102e-05, "loss": 0.0079, "step": 37300 }, { "grad_norm": 0.19980384409427643, "learning_rate": 7.407464211861183e-05, "loss": 0.008, "step": 37310 }, { "grad_norm": 0.14607840776443481, "learning_rate": 7.406014896589597e-05, "loss": 0.0118, "step": 37320 }, { "grad_norm": 0.14213542640209198, "learning_rate": 7.404565318199838e-05, "loss": 0.0096, "step": 37330 }, { "grad_norm": 0.1273842453956604, "learning_rate": 7.403115476850432e-05, "loss": 0.0091, "step": 37340 }, { "grad_norm": 0.0956386849284172, "learning_rate": 7.401665372699932e-05, "loss": 0.0083, "step": 37350 }, { "grad_norm": 0.11643429845571518, "learning_rate": 7.400215005906916e-05, "loss": 0.0072, "step": 37360 }, { "grad_norm": 0.06730952113866806, "learning_rate": 7.398764376629996e-05, "loss": 0.0056, "step": 37370 }, { "grad_norm": 0.13146184384822845, "learning_rate": 7.39731348502781e-05, "loss": 0.0103, "step": 37380 }, { "grad_norm": 0.10450412333011627, "learning_rate": 7.395862331259028e-05, "loss": 0.0093, "step": 37390 }, { "grad_norm": 0.10157692432403564, "learning_rate": 7.394410915482341e-05, "loss": 0.0086, "step": 37400 }, { "grad_norm": 0.09249590337276459, "learning_rate": 7.392959237856476e-05, "loss": 0.011, "step": 37410 }, { "grad_norm": 0.10571036487817764, "learning_rate": 7.391507298540187e-05, "loss": 0.0079, "step": 37420 }, { "grad_norm": 0.09071440994739532, "learning_rate": 7.390055097692257e-05, "loss": 0.007, "step": 37430 }, { "grad_norm": 0.09318084269762039, "learning_rate": 7.388602635471494e-05, "loss": 0.0107, "step": 37440 }, { "grad_norm": 0.13311877846717834, "learning_rate": 7.387149912036738e-05, "loss": 0.0096, "step": 37450 }, { "grad_norm": 0.10353513062000275, "learning_rate": 7.385696927546854e-05, "loss": 0.0095, "step": 37460 }, { "grad_norm": 0.10184451192617416, "learning_rate": 7.384243682160743e-05, "loss": 0.0095, "step": 37470 }, { "grad_norm": 0.10625948756933212, "learning_rate": 7.382790176037329e-05, "loss": 0.0079, "step": 37480 }, { "grad_norm": 0.10227864235639572, "learning_rate": 7.381336409335562e-05, "loss": 0.0083, "step": 37490 }, { "grad_norm": 0.1275707632303238, "learning_rate": 7.379882382214426e-05, "loss": 0.0077, "step": 37500 }, { "grad_norm": 0.08647702634334564, "learning_rate": 7.378428094832931e-05, "loss": 0.007, "step": 37510 }, { "grad_norm": 0.11166130006313324, "learning_rate": 7.376973547350114e-05, "loss": 0.0102, "step": 37520 }, { "grad_norm": 0.163993701338768, "learning_rate": 7.375518739925046e-05, "loss": 0.0084, "step": 37530 }, { "grad_norm": 0.08333487808704376, "learning_rate": 7.374063672716818e-05, "loss": 0.0074, "step": 37540 }, { "grad_norm": 0.08456651121377945, "learning_rate": 7.372608345884558e-05, "loss": 0.0067, "step": 37550 }, { "grad_norm": 0.0870194360613823, "learning_rate": 7.371152759587415e-05, "loss": 0.0079, "step": 37560 }, { "grad_norm": 0.08882638812065125, "learning_rate": 7.369696913984572e-05, "loss": 0.0091, "step": 37570 }, { "grad_norm": 0.11978007107973099, "learning_rate": 7.368240809235237e-05, "loss": 0.0075, "step": 37580 }, { "grad_norm": 0.12184770405292511, "learning_rate": 7.366784445498649e-05, "loss": 0.0078, "step": 37590 }, { "grad_norm": 0.1501842439174652, "learning_rate": 7.36532782293407e-05, "loss": 0.0057, "step": 37600 }, { "grad_norm": 0.10166030377149582, "learning_rate": 7.363870941700797e-05, "loss": 0.0071, "step": 37610 }, { "grad_norm": 0.10221615433692932, "learning_rate": 7.362413801958152e-05, "loss": 0.0052, "step": 37620 }, { "grad_norm": 0.1431932896375656, "learning_rate": 7.360956403865486e-05, "loss": 0.0069, "step": 37630 }, { "grad_norm": 0.08888299763202667, "learning_rate": 7.359498747582177e-05, "loss": 0.0067, "step": 37640 }, { "grad_norm": 0.158146470785141, "learning_rate": 7.358040833267634e-05, "loss": 0.009, "step": 37650 }, { "grad_norm": 0.06820777803659439, "learning_rate": 7.356582661081288e-05, "loss": 0.0094, "step": 37660 }, { "grad_norm": 0.2062063068151474, "learning_rate": 7.355124231182607e-05, "loss": 0.01, "step": 37670 }, { "grad_norm": 0.09905511140823364, "learning_rate": 7.353665543731083e-05, "loss": 0.0099, "step": 37680 }, { "grad_norm": 0.09062617272138596, "learning_rate": 7.352206598886229e-05, "loss": 0.0087, "step": 37690 }, { "grad_norm": 0.22026802599430084, "learning_rate": 7.350747396807601e-05, "loss": 0.0085, "step": 37700 }, { "grad_norm": 0.11956370621919632, "learning_rate": 7.349287937654772e-05, "loss": 0.0095, "step": 37710 }, { "grad_norm": 0.14810051023960114, "learning_rate": 7.347828221587345e-05, "loss": 0.0075, "step": 37720 }, { "grad_norm": 0.08418197929859161, "learning_rate": 7.346368248764955e-05, "loss": 0.0071, "step": 37730 }, { "grad_norm": 0.12405888736248016, "learning_rate": 7.34490801934726e-05, "loss": 0.0084, "step": 37740 }, { "grad_norm": 0.1033492237329483, "learning_rate": 7.343447533493947e-05, "loss": 0.0077, "step": 37750 }, { "grad_norm": 0.0771622434258461, "learning_rate": 7.341986791364739e-05, "loss": 0.0067, "step": 37760 }, { "grad_norm": 0.13922755420207977, "learning_rate": 7.340525793119374e-05, "loss": 0.0063, "step": 37770 }, { "grad_norm": 0.12009702622890472, "learning_rate": 7.33906453891763e-05, "loss": 0.0104, "step": 37780 }, { "grad_norm": 0.12752334773540497, "learning_rate": 7.337603028919303e-05, "loss": 0.0063, "step": 37790 }, { "grad_norm": 0.07262061536312103, "learning_rate": 7.336141263284225e-05, "loss": 0.0062, "step": 37800 }, { "grad_norm": 0.07311966270208359, "learning_rate": 7.334679242172249e-05, "loss": 0.0074, "step": 37810 }, { "grad_norm": 0.10721264034509659, "learning_rate": 7.33321696574326e-05, "loss": 0.008, "step": 37820 }, { "grad_norm": 0.13642601668834686, "learning_rate": 7.331754434157174e-05, "loss": 0.0059, "step": 37830 }, { "grad_norm": 0.1295596957206726, "learning_rate": 7.330291647573928e-05, "loss": 0.0069, "step": 37840 }, { "grad_norm": 0.09010661393404007, "learning_rate": 7.32882860615349e-05, "loss": 0.0077, "step": 37850 }, { "grad_norm": 0.0644821971654892, "learning_rate": 7.327365310055858e-05, "loss": 0.0068, "step": 37860 }, { "grad_norm": 0.10656626522541046, "learning_rate": 7.325901759441055e-05, "loss": 0.0057, "step": 37870 }, { "grad_norm": 0.07913416624069214, "learning_rate": 7.324437954469132e-05, "loss": 0.0067, "step": 37880 }, { "grad_norm": 0.09518797695636749, "learning_rate": 7.32297389530017e-05, "loss": 0.0066, "step": 37890 }, { "grad_norm": 0.0927918404340744, "learning_rate": 7.321509582094274e-05, "loss": 0.0058, "step": 37900 }, { "grad_norm": 0.07203630357980728, "learning_rate": 7.320045015011582e-05, "loss": 0.007, "step": 37910 }, { "grad_norm": 0.08545726537704468, "learning_rate": 7.318580194212253e-05, "loss": 0.0063, "step": 37920 }, { "grad_norm": 0.1739175170660019, "learning_rate": 7.317115119856483e-05, "loss": 0.0069, "step": 37930 }, { "grad_norm": 0.11898288130760193, "learning_rate": 7.315649792104485e-05, "loss": 0.0066, "step": 37940 }, { "grad_norm": 0.1460115760564804, "learning_rate": 7.314184211116507e-05, "loss": 0.0067, "step": 37950 }, { "grad_norm": 0.06814774125814438, "learning_rate": 7.312718377052823e-05, "loss": 0.0078, "step": 37960 }, { "grad_norm": 0.06828410923480988, "learning_rate": 7.311252290073737e-05, "loss": 0.0078, "step": 37970 }, { "grad_norm": 0.10962127894163132, "learning_rate": 7.309785950339572e-05, "loss": 0.0077, "step": 37980 }, { "grad_norm": 0.09393835812807083, "learning_rate": 7.308319358010692e-05, "loss": 0.0067, "step": 37990 }, { "grad_norm": 0.10899217426776886, "learning_rate": 7.306852513247474e-05, "loss": 0.0095, "step": 38000 }, { "grad_norm": 0.14507222175598145, "learning_rate": 7.305385416210336e-05, "loss": 0.0088, "step": 38010 }, { "grad_norm": 0.09201472997665405, "learning_rate": 7.303918067059711e-05, "loss": 0.0069, "step": 38020 }, { "grad_norm": 0.09945661574602127, "learning_rate": 7.302450465956075e-05, "loss": 0.0054, "step": 38030 }, { "grad_norm": 0.04500769451260567, "learning_rate": 7.300982613059914e-05, "loss": 0.0055, "step": 38040 }, { "grad_norm": 0.12063803523778915, "learning_rate": 7.299514508531757e-05, "loss": 0.0082, "step": 38050 }, { "grad_norm": 0.07497096061706543, "learning_rate": 7.298046152532148e-05, "loss": 0.0063, "step": 38060 }, { "grad_norm": 0.1129738911986351, "learning_rate": 7.296577545221668e-05, "loss": 0.0069, "step": 38070 }, { "grad_norm": 0.11379526555538177, "learning_rate": 7.29510868676092e-05, "loss": 0.0085, "step": 38080 }, { "grad_norm": 0.08862782269716263, "learning_rate": 7.293639577310538e-05, "loss": 0.01, "step": 38090 }, { "grad_norm": 0.09935818612575531, "learning_rate": 7.292170217031179e-05, "loss": 0.0098, "step": 38100 }, { "grad_norm": 0.09818864613771439, "learning_rate": 7.290700606083532e-05, "loss": 0.0081, "step": 38110 }, { "grad_norm": 0.1207319051027298, "learning_rate": 7.28923074462831e-05, "loss": 0.0081, "step": 38120 }, { "grad_norm": 0.12493492662906647, "learning_rate": 7.287760632826255e-05, "loss": 0.0097, "step": 38130 }, { "grad_norm": 0.13452650606632233, "learning_rate": 7.286290270838139e-05, "loss": 0.0062, "step": 38140 }, { "grad_norm": 0.11502450704574585, "learning_rate": 7.284819658824756e-05, "loss": 0.0063, "step": 38150 }, { "grad_norm": 0.10503542423248291, "learning_rate": 7.283348796946929e-05, "loss": 0.008, "step": 38160 }, { "grad_norm": 0.10829725116491318, "learning_rate": 7.28187768536551e-05, "loss": 0.0095, "step": 38170 }, { "grad_norm": 0.1051606759428978, "learning_rate": 7.280406324241379e-05, "loss": 0.0074, "step": 38180 }, { "grad_norm": 0.1423341929912567, "learning_rate": 7.278934713735441e-05, "loss": 0.0092, "step": 38190 }, { "grad_norm": 0.08379257470369339, "learning_rate": 7.277462854008629e-05, "loss": 0.0069, "step": 38200 }, { "grad_norm": 0.11756433546543121, "learning_rate": 7.275990745221904e-05, "loss": 0.0084, "step": 38210 }, { "grad_norm": 0.1683613806962967, "learning_rate": 7.274518387536252e-05, "loss": 0.0078, "step": 38220 }, { "grad_norm": 0.11858659982681274, "learning_rate": 7.273045781112689e-05, "loss": 0.01, "step": 38230 }, { "grad_norm": 0.1027245968580246, "learning_rate": 7.271572926112257e-05, "loss": 0.0071, "step": 38240 }, { "grad_norm": 0.13365979492664337, "learning_rate": 7.270099822696024e-05, "loss": 0.0083, "step": 38250 }, { "grad_norm": 0.13080815970897675, "learning_rate": 7.26862647102509e-05, "loss": 0.0087, "step": 38260 }, { "grad_norm": 0.13113093376159668, "learning_rate": 7.267152871260573e-05, "loss": 0.0078, "step": 38270 }, { "grad_norm": 0.08018212020397186, "learning_rate": 7.265679023563629e-05, "loss": 0.0074, "step": 38280 }, { "grad_norm": 0.10642615705728531, "learning_rate": 7.264204928095433e-05, "loss": 0.0078, "step": 38290 }, { "grad_norm": 0.0718497484922409, "learning_rate": 7.262730585017188e-05, "loss": 0.0059, "step": 38300 }, { "grad_norm": 0.11734361946582794, "learning_rate": 7.261255994490131e-05, "loss": 0.0082, "step": 38310 }, { "grad_norm": 0.1048247292637825, "learning_rate": 7.259781156675517e-05, "loss": 0.0068, "step": 38320 }, { "grad_norm": 0.1059502512216568, "learning_rate": 7.258306071734633e-05, "loss": 0.0092, "step": 38330 }, { "grad_norm": 0.11647682636976242, "learning_rate": 7.256830739828792e-05, "loss": 0.0109, "step": 38340 }, { "grad_norm": 0.11109372228384018, "learning_rate": 7.255355161119336e-05, "loss": 0.0081, "step": 38350 }, { "grad_norm": 0.09778871387243271, "learning_rate": 7.253879335767628e-05, "loss": 0.0075, "step": 38360 }, { "grad_norm": 0.15346649289131165, "learning_rate": 7.252403263935067e-05, "loss": 0.0102, "step": 38370 }, { "grad_norm": 0.10931724309921265, "learning_rate": 7.25092694578307e-05, "loss": 0.0069, "step": 38380 }, { "grad_norm": 0.09111644327640533, "learning_rate": 7.249450381473087e-05, "loss": 0.0069, "step": 38390 }, { "grad_norm": 0.0603577122092247, "learning_rate": 7.247973571166593e-05, "loss": 0.0059, "step": 38400 }, { "grad_norm": 0.1404297649860382, "learning_rate": 7.246496515025089e-05, "loss": 0.0095, "step": 38410 }, { "grad_norm": 0.13969077169895172, "learning_rate": 7.245019213210101e-05, "loss": 0.0083, "step": 38420 }, { "grad_norm": 0.08921067416667938, "learning_rate": 7.24354166588319e-05, "loss": 0.0075, "step": 38430 }, { "grad_norm": 0.11218368262052536, "learning_rate": 7.242063873205935e-05, "loss": 0.0074, "step": 38440 }, { "grad_norm": 0.11996516585350037, "learning_rate": 7.240585835339946e-05, "loss": 0.0079, "step": 38450 }, { "grad_norm": 0.08056526631116867, "learning_rate": 7.239107552446857e-05, "loss": 0.009, "step": 38460 }, { "grad_norm": 0.06486237794160843, "learning_rate": 7.237629024688334e-05, "loss": 0.0071, "step": 38470 }, { "grad_norm": 0.06318623572587967, "learning_rate": 7.236150252226064e-05, "loss": 0.0067, "step": 38480 }, { "grad_norm": 0.12257537245750427, "learning_rate": 7.234671235221765e-05, "loss": 0.009, "step": 38490 }, { "grad_norm": 0.09895758330821991, "learning_rate": 7.233191973837179e-05, "loss": 0.0103, "step": 38500 }, { "grad_norm": 0.0965050533413887, "learning_rate": 7.231712468234078e-05, "loss": 0.0094, "step": 38510 }, { "grad_norm": 0.11944979429244995, "learning_rate": 7.230232718574254e-05, "loss": 0.0063, "step": 38520 }, { "grad_norm": 0.1064925417304039, "learning_rate": 7.228752725019535e-05, "loss": 0.0081, "step": 38530 }, { "grad_norm": 0.09832325577735901, "learning_rate": 7.227272487731769e-05, "loss": 0.0084, "step": 38540 }, { "grad_norm": 0.13490261137485504, "learning_rate": 7.225792006872831e-05, "loss": 0.0109, "step": 38550 }, { "grad_norm": 0.11496173590421677, "learning_rate": 7.224311282604628e-05, "loss": 0.0067, "step": 38560 }, { "grad_norm": 0.09483883529901505, "learning_rate": 7.222830315089085e-05, "loss": 0.0074, "step": 38570 }, { "grad_norm": 0.08476383239030838, "learning_rate": 7.22134910448816e-05, "loss": 0.0061, "step": 38580 }, { "grad_norm": 0.0933663472533226, "learning_rate": 7.219867650963839e-05, "loss": 0.0073, "step": 38590 }, { "grad_norm": 0.09706565737724304, "learning_rate": 7.21838595467813e-05, "loss": 0.0075, "step": 38600 }, { "grad_norm": 0.098162442445755, "learning_rate": 7.216904015793067e-05, "loss": 0.0071, "step": 38610 }, { "grad_norm": 0.09197066724300385, "learning_rate": 7.215421834470713e-05, "loss": 0.0096, "step": 38620 }, { "grad_norm": 0.09828558564186096, "learning_rate": 7.213939410873158e-05, "loss": 0.0061, "step": 38630 }, { "grad_norm": 0.08661271631717682, "learning_rate": 7.212456745162519e-05, "loss": 0.0074, "step": 38640 }, { "grad_norm": 0.10391271859407425, "learning_rate": 7.210973837500937e-05, "loss": 0.0076, "step": 38650 }, { "grad_norm": 0.12307132035493851, "learning_rate": 7.209490688050578e-05, "loss": 0.009, "step": 38660 }, { "grad_norm": 0.14900091290473938, "learning_rate": 7.208007296973641e-05, "loss": 0.0092, "step": 38670 }, { "grad_norm": 0.09373783320188522, "learning_rate": 7.206523664432345e-05, "loss": 0.0086, "step": 38680 }, { "grad_norm": 0.13248662650585175, "learning_rate": 7.205039790588939e-05, "loss": 0.0078, "step": 38690 }, { "grad_norm": 0.1014467254281044, "learning_rate": 7.203555675605697e-05, "loss": 0.0063, "step": 38700 }, { "grad_norm": 0.13774742186069489, "learning_rate": 7.202071319644917e-05, "loss": 0.0083, "step": 38710 }, { "grad_norm": 0.12065736204385757, "learning_rate": 7.200586722868932e-05, "loss": 0.0089, "step": 38720 }, { "grad_norm": 0.24302132427692413, "learning_rate": 7.199101885440088e-05, "loss": 0.0103, "step": 38730 }, { "grad_norm": 0.12215663492679596, "learning_rate": 7.197616807520771e-05, "loss": 0.0109, "step": 38740 }, { "grad_norm": 0.12112792581319809, "learning_rate": 7.196131489273381e-05, "loss": 0.0093, "step": 38750 }, { "grad_norm": 0.0697566345334053, "learning_rate": 7.194645930860355e-05, "loss": 0.006, "step": 38760 }, { "grad_norm": 0.07738252729177475, "learning_rate": 7.19316013244415e-05, "loss": 0.0083, "step": 38770 }, { "grad_norm": 0.05755312368273735, "learning_rate": 7.191674094187248e-05, "loss": 0.006, "step": 38780 }, { "grad_norm": 0.14882704615592957, "learning_rate": 7.190187816252165e-05, "loss": 0.0068, "step": 38790 }, { "grad_norm": 0.09937490522861481, "learning_rate": 7.188701298801435e-05, "loss": 0.0073, "step": 38800 }, { "grad_norm": 0.15799841284751892, "learning_rate": 7.18721454199762e-05, "loss": 0.0112, "step": 38810 }, { "grad_norm": 0.1010185107588768, "learning_rate": 7.185727546003312e-05, "loss": 0.0091, "step": 38820 }, { "grad_norm": 0.155246302485466, "learning_rate": 7.184240310981126e-05, "loss": 0.0105, "step": 38830 }, { "grad_norm": 0.0677548348903656, "learning_rate": 7.182752837093704e-05, "loss": 0.0074, "step": 38840 }, { "grad_norm": 0.09114177525043488, "learning_rate": 7.181265124503711e-05, "loss": 0.0085, "step": 38850 }, { "grad_norm": 0.09993045777082443, "learning_rate": 7.179777173373847e-05, "loss": 0.009, "step": 38860 }, { "grad_norm": 0.0990419015288353, "learning_rate": 7.178288983866826e-05, "loss": 0.0065, "step": 38870 }, { "grad_norm": 0.10850531607866287, "learning_rate": 7.176800556145397e-05, "loss": 0.0097, "step": 38880 }, { "grad_norm": 0.11342908442020416, "learning_rate": 7.175311890372334e-05, "loss": 0.0085, "step": 38890 }, { "grad_norm": 0.09038445353507996, "learning_rate": 7.17382298671043e-05, "loss": 0.0097, "step": 38900 }, { "grad_norm": 0.09704125672578812, "learning_rate": 7.172333845322515e-05, "loss": 0.0092, "step": 38910 }, { "grad_norm": 0.15010564029216766, "learning_rate": 7.170844466371436e-05, "loss": 0.007, "step": 38920 }, { "grad_norm": 0.08503644168376923, "learning_rate": 7.16935485002007e-05, "loss": 0.0051, "step": 38930 }, { "grad_norm": 0.090105801820755, "learning_rate": 7.167864996431319e-05, "loss": 0.0066, "step": 38940 }, { "grad_norm": 0.1267983466386795, "learning_rate": 7.166374905768111e-05, "loss": 0.01, "step": 38950 }, { "grad_norm": 0.11389966309070587, "learning_rate": 7.164884578193401e-05, "loss": 0.0096, "step": 38960 }, { "grad_norm": 0.14420343935489655, "learning_rate": 7.163394013870168e-05, "loss": 0.0097, "step": 38970 }, { "grad_norm": 0.13259436190128326, "learning_rate": 7.161903212961419e-05, "loss": 0.0102, "step": 38980 }, { "grad_norm": 0.09803396463394165, "learning_rate": 7.160412175630183e-05, "loss": 0.0085, "step": 38990 }, { "grad_norm": 0.08299855142831802, "learning_rate": 7.158920902039521e-05, "loss": 0.0071, "step": 39000 }, { "grad_norm": 0.08844611048698425, "learning_rate": 7.157429392352514e-05, "loss": 0.0088, "step": 39010 }, { "grad_norm": 0.14789657294750214, "learning_rate": 7.155937646732275e-05, "loss": 0.0068, "step": 39020 }, { "grad_norm": 0.10959949344396591, "learning_rate": 7.154445665341933e-05, "loss": 0.0069, "step": 39030 }, { "grad_norm": 0.1313672810792923, "learning_rate": 7.152953448344654e-05, "loss": 0.0078, "step": 39040 }, { "grad_norm": 0.1128552258014679, "learning_rate": 7.151460995903624e-05, "loss": 0.0065, "step": 39050 }, { "grad_norm": 0.11243193596601486, "learning_rate": 7.149968308182052e-05, "loss": 0.0112, "step": 39060 }, { "grad_norm": 0.1375812441110611, "learning_rate": 7.14847538534318e-05, "loss": 0.0094, "step": 39070 }, { "grad_norm": 0.10178280621767044, "learning_rate": 7.14698222755027e-05, "loss": 0.0066, "step": 39080 }, { "grad_norm": 0.13412907719612122, "learning_rate": 7.14548883496661e-05, "loss": 0.0097, "step": 39090 }, { "grad_norm": 0.13694241642951965, "learning_rate": 7.143995207755517e-05, "loss": 0.0102, "step": 39100 }, { "grad_norm": 0.10360193997621536, "learning_rate": 7.142501346080333e-05, "loss": 0.0066, "step": 39110 }, { "grad_norm": 0.15127445757389069, "learning_rate": 7.141007250104421e-05, "loss": 0.0103, "step": 39120 }, { "grad_norm": 0.10262184590101242, "learning_rate": 7.139512919991176e-05, "loss": 0.0093, "step": 39130 }, { "grad_norm": 0.15455196797847748, "learning_rate": 7.138018355904014e-05, "loss": 0.0082, "step": 39140 }, { "grad_norm": 0.11095338314771652, "learning_rate": 7.13652355800638e-05, "loss": 0.0074, "step": 39150 }, { "grad_norm": 0.15220876038074493, "learning_rate": 7.135028526461743e-05, "loss": 0.0083, "step": 39160 }, { "grad_norm": 0.08426080644130707, "learning_rate": 7.133533261433594e-05, "loss": 0.0066, "step": 39170 }, { "grad_norm": 0.17389504611492157, "learning_rate": 7.132037763085457e-05, "loss": 0.0073, "step": 39180 }, { "grad_norm": 0.19732214510440826, "learning_rate": 7.130542031580875e-05, "loss": 0.009, "step": 39190 }, { "grad_norm": 0.15609237551689148, "learning_rate": 7.12904606708342e-05, "loss": 0.0092, "step": 39200 }, { "grad_norm": 0.11567670106887817, "learning_rate": 7.127549869756687e-05, "loss": 0.008, "step": 39210 }, { "grad_norm": 0.09434454143047333, "learning_rate": 7.126053439764299e-05, "loss": 0.0074, "step": 39220 }, { "grad_norm": 0.09214373677968979, "learning_rate": 7.124556777269904e-05, "loss": 0.0072, "step": 39230 }, { "grad_norm": 0.14090706408023834, "learning_rate": 7.123059882437174e-05, "loss": 0.0116, "step": 39240 }, { "grad_norm": 0.07131367921829224, "learning_rate": 7.121562755429807e-05, "loss": 0.0049, "step": 39250 }, { "grad_norm": 0.0776553601026535, "learning_rate": 7.120065396411528e-05, "loss": 0.0051, "step": 39260 }, { "grad_norm": 0.0802636668086052, "learning_rate": 7.118567805546084e-05, "loss": 0.0052, "step": 39270 }, { "grad_norm": 0.12246312201023102, "learning_rate": 7.117069982997248e-05, "loss": 0.0103, "step": 39280 }, { "grad_norm": 0.12407134473323822, "learning_rate": 7.115571928928825e-05, "loss": 0.0105, "step": 39290 }, { "grad_norm": 0.10903506726026535, "learning_rate": 7.114073643504635e-05, "loss": 0.0084, "step": 39300 }, { "grad_norm": 0.10778124630451202, "learning_rate": 7.11257512688853e-05, "loss": 0.0077, "step": 39310 }, { "grad_norm": 0.08069368451833725, "learning_rate": 7.111076379244384e-05, "loss": 0.0073, "step": 39320 }, { "grad_norm": 0.08930652588605881, "learning_rate": 7.109577400736101e-05, "loss": 0.0083, "step": 39330 }, { "grad_norm": 0.06981717795133591, "learning_rate": 7.108078191527605e-05, "loss": 0.0069, "step": 39340 }, { "grad_norm": 0.13789929449558258, "learning_rate": 7.106578751782847e-05, "loss": 0.0099, "step": 39350 }, { "grad_norm": 0.11790528148412704, "learning_rate": 7.105079081665803e-05, "loss": 0.0077, "step": 39360 }, { "grad_norm": 0.0787002444267273, "learning_rate": 7.103579181340476e-05, "loss": 0.007, "step": 39370 }, { "grad_norm": 0.09033001959323883, "learning_rate": 7.102079050970893e-05, "loss": 0.006, "step": 39380 }, { "grad_norm": 0.08138567209243774, "learning_rate": 7.100578690721105e-05, "loss": 0.0066, "step": 39390 }, { "grad_norm": 0.0731222853064537, "learning_rate": 7.09907810075519e-05, "loss": 0.0052, "step": 39400 }, { "grad_norm": 0.09845159202814102, "learning_rate": 7.097577281237249e-05, "loss": 0.0071, "step": 39410 }, { "grad_norm": 0.11377938091754913, "learning_rate": 7.09607623233141e-05, "loss": 0.0075, "step": 39420 }, { "grad_norm": 0.13070522248744965, "learning_rate": 7.094574954201827e-05, "loss": 0.0077, "step": 39430 }, { "grad_norm": 0.10991240292787552, "learning_rate": 7.093073447012675e-05, "loss": 0.0089, "step": 39440 }, { "grad_norm": 0.06015729159116745, "learning_rate": 7.09157171092816e-05, "loss": 0.0068, "step": 39450 }, { "grad_norm": 0.10012631118297577, "learning_rate": 7.090069746112504e-05, "loss": 0.0067, "step": 39460 }, { "grad_norm": 0.08744298666715622, "learning_rate": 7.088567552729965e-05, "loss": 0.0063, "step": 39470 }, { "grad_norm": 0.14049409329891205, "learning_rate": 7.087065130944818e-05, "loss": 0.0074, "step": 39480 }, { "grad_norm": 0.09942702949047089, "learning_rate": 7.085562480921366e-05, "loss": 0.0076, "step": 39490 }, { "grad_norm": 0.16070786118507385, "learning_rate": 7.084059602823937e-05, "loss": 0.0076, "step": 39500 }, { "grad_norm": 0.10523779690265656, "learning_rate": 7.082556496816882e-05, "loss": 0.0126, "step": 39510 }, { "grad_norm": 0.09738443791866302, "learning_rate": 7.081053163064582e-05, "loss": 0.0084, "step": 39520 }, { "grad_norm": 0.12004904448986053, "learning_rate": 7.079549601731434e-05, "loss": 0.0059, "step": 39530 }, { "grad_norm": 0.11095120757818222, "learning_rate": 7.07804581298187e-05, "loss": 0.0071, "step": 39540 }, { "grad_norm": 0.07980181276798248, "learning_rate": 7.07654179698034e-05, "loss": 0.0062, "step": 39550 }, { "grad_norm": 0.07766076922416687, "learning_rate": 7.075037553891321e-05, "loss": 0.0053, "step": 39560 }, { "grad_norm": 0.097453773021698, "learning_rate": 7.073533083879315e-05, "loss": 0.0069, "step": 39570 }, { "grad_norm": 0.0869256779551506, "learning_rate": 7.072028387108849e-05, "loss": 0.0064, "step": 39580 }, { "grad_norm": 0.12829984724521637, "learning_rate": 7.070523463744472e-05, "loss": 0.0076, "step": 39590 }, { "grad_norm": 0.15277500450611115, "learning_rate": 7.069018313950763e-05, "loss": 0.0075, "step": 39600 }, { "grad_norm": 0.11555120348930359, "learning_rate": 7.067512937892322e-05, "loss": 0.0085, "step": 39610 }, { "grad_norm": 0.14719073474407196, "learning_rate": 7.066007335733775e-05, "loss": 0.0065, "step": 39620 }, { "grad_norm": 0.1048964411020279, "learning_rate": 7.064501507639772e-05, "loss": 0.0069, "step": 39630 }, { "grad_norm": 0.0712430477142334, "learning_rate": 7.062995453774987e-05, "loss": 0.0062, "step": 39640 }, { "grad_norm": 0.053175169974565506, "learning_rate": 7.061489174304121e-05, "loss": 0.0041, "step": 39650 }, { "grad_norm": 0.08476821333169937, "learning_rate": 7.0599826693919e-05, "loss": 0.0071, "step": 39660 }, { "grad_norm": 0.06396227329969406, "learning_rate": 7.05847593920307e-05, "loss": 0.007, "step": 39670 }, { "grad_norm": 0.09538708627223969, "learning_rate": 7.056968983902406e-05, "loss": 0.008, "step": 39680 }, { "grad_norm": 0.07118043303489685, "learning_rate": 7.055461803654706e-05, "loss": 0.0069, "step": 39690 }, { "grad_norm": 0.07124345004558563, "learning_rate": 7.053954398624794e-05, "loss": 0.0079, "step": 39700 }, { "grad_norm": 0.07554388046264648, "learning_rate": 7.052446768977518e-05, "loss": 0.0081, "step": 39710 }, { "grad_norm": 0.13638252019882202, "learning_rate": 7.050938914877748e-05, "loss": 0.0062, "step": 39720 }, { "grad_norm": 0.12076367437839508, "learning_rate": 7.049430836490382e-05, "loss": 0.0083, "step": 39730 }, { "grad_norm": 0.1922188103199005, "learning_rate": 7.047922533980342e-05, "loss": 0.0102, "step": 39740 }, { "grad_norm": 0.10945487022399902, "learning_rate": 7.046414007512571e-05, "loss": 0.0072, "step": 39750 }, { "grad_norm": 0.0989910289645195, "learning_rate": 7.044905257252042e-05, "loss": 0.0067, "step": 39760 }, { "grad_norm": 0.0791054517030716, "learning_rate": 7.04339628336375e-05, "loss": 0.0064, "step": 39770 }, { "grad_norm": 0.06562136113643646, "learning_rate": 7.041887086012711e-05, "loss": 0.0063, "step": 39780 }, { "grad_norm": 0.0787772387266159, "learning_rate": 7.040377665363969e-05, "loss": 0.008, "step": 39790 }, { "grad_norm": 0.07562655210494995, "learning_rate": 7.038868021582594e-05, "loss": 0.0073, "step": 39800 }, { "grad_norm": 0.09627562761306763, "learning_rate": 7.037358154833679e-05, "loss": 0.0072, "step": 39810 }, { "grad_norm": 0.09451012313365936, "learning_rate": 7.035848065282339e-05, "loss": 0.0078, "step": 39820 }, { "grad_norm": 0.08338277041912079, "learning_rate": 7.034337753093714e-05, "loss": 0.0056, "step": 39830 }, { "grad_norm": 0.1201380267739296, "learning_rate": 7.032827218432972e-05, "loss": 0.0075, "step": 39840 }, { "grad_norm": 0.08549709618091583, "learning_rate": 7.031316461465302e-05, "loss": 0.0097, "step": 39850 }, { "grad_norm": 0.07503890246152878, "learning_rate": 7.029805482355915e-05, "loss": 0.0087, "step": 39860 }, { "grad_norm": 0.06680304557085037, "learning_rate": 7.028294281270055e-05, "loss": 0.0068, "step": 39870 }, { "grad_norm": 0.09562075138092041, "learning_rate": 7.026782858372982e-05, "loss": 0.007, "step": 39880 }, { "grad_norm": 0.2387770414352417, "learning_rate": 7.025271213829982e-05, "loss": 0.0103, "step": 39890 }, { "grad_norm": 0.09805247187614441, "learning_rate": 7.023759347806366e-05, "loss": 0.008, "step": 39900 }, { "grad_norm": 0.09220296889543533, "learning_rate": 7.022247260467472e-05, "loss": 0.0069, "step": 39910 }, { "grad_norm": 0.1555582731962204, "learning_rate": 7.020734951978658e-05, "loss": 0.0076, "step": 39920 }, { "grad_norm": 0.08866551518440247, "learning_rate": 7.019222422505307e-05, "loss": 0.0098, "step": 39930 }, { "grad_norm": 0.1536664366722107, "learning_rate": 7.017709672212827e-05, "loss": 0.0078, "step": 39940 }, { "grad_norm": 0.09222739934921265, "learning_rate": 7.016196701266652e-05, "loss": 0.0067, "step": 39950 }, { "grad_norm": 0.10775768756866455, "learning_rate": 7.014683509832238e-05, "loss": 0.0073, "step": 39960 }, { "grad_norm": 0.09069456160068512, "learning_rate": 7.013170098075063e-05, "loss": 0.0054, "step": 39970 }, { "grad_norm": 0.1076873391866684, "learning_rate": 7.011656466160632e-05, "loss": 0.0068, "step": 39980 }, { "grad_norm": 0.11394545435905457, "learning_rate": 7.010142614254475e-05, "loss": 0.0068, "step": 39990 }, { "grad_norm": 0.164051815867424, "learning_rate": 7.008628542522147e-05, "loss": 0.0063, "step": 40000 }, { "grad_norm": 0.07943133264780045, "learning_rate": 7.007114251129218e-05, "loss": 0.008, "step": 40010 }, { "grad_norm": 0.1343698352575302, "learning_rate": 7.005599740241293e-05, "loss": 0.0119, "step": 40020 }, { "grad_norm": 0.12478714436292648, "learning_rate": 7.004085010023996e-05, "loss": 0.0077, "step": 40030 }, { "grad_norm": 0.10049918293952942, "learning_rate": 7.002570060642976e-05, "loss": 0.0081, "step": 40040 }, { "grad_norm": 0.124668650329113, "learning_rate": 7.001054892263903e-05, "loss": 0.0096, "step": 40050 }, { "grad_norm": 0.12417890131473541, "learning_rate": 6.999539505052477e-05, "loss": 0.0077, "step": 40060 }, { "grad_norm": 0.09422027319669724, "learning_rate": 6.998023899174418e-05, "loss": 0.0057, "step": 40070 }, { "grad_norm": 0.1874893754720688, "learning_rate": 6.996508074795467e-05, "loss": 0.0081, "step": 40080 }, { "grad_norm": 0.12621811032295227, "learning_rate": 6.994992032081396e-05, "loss": 0.0104, "step": 40090 }, { "grad_norm": 0.11400112509727478, "learning_rate": 6.993475771197995e-05, "loss": 0.0095, "step": 40100 }, { "grad_norm": 0.09289725869894028, "learning_rate": 6.991959292311082e-05, "loss": 0.0073, "step": 40110 }, { "grad_norm": 0.08435899019241333, "learning_rate": 6.990442595586495e-05, "loss": 0.0086, "step": 40120 }, { "grad_norm": 0.05321165546774864, "learning_rate": 6.988925681190098e-05, "loss": 0.0093, "step": 40130 }, { "grad_norm": 0.17585530877113342, "learning_rate": 6.987408549287778e-05, "loss": 0.0091, "step": 40140 }, { "grad_norm": 0.10813569277524948, "learning_rate": 6.985891200045449e-05, "loss": 0.0062, "step": 40150 }, { "grad_norm": 0.19988073408603668, "learning_rate": 6.984373633629045e-05, "loss": 0.0083, "step": 40160 }, { "grad_norm": 0.13908618688583374, "learning_rate": 6.98285585020452e-05, "loss": 0.0088, "step": 40170 }, { "grad_norm": 0.10495130717754364, "learning_rate": 6.981337849937864e-05, "loss": 0.0075, "step": 40180 }, { "grad_norm": 0.10277710855007172, "learning_rate": 6.979819632995078e-05, "loss": 0.0065, "step": 40190 }, { "grad_norm": 0.15326935052871704, "learning_rate": 6.978301199542193e-05, "loss": 0.0094, "step": 40200 }, { "grad_norm": 0.07990909367799759, "learning_rate": 6.976782549745263e-05, "loss": 0.0067, "step": 40210 }, { "grad_norm": 0.04780500754714012, "learning_rate": 6.975263683770367e-05, "loss": 0.0064, "step": 40220 }, { "grad_norm": 0.0690944641828537, "learning_rate": 6.973744601783604e-05, "loss": 0.0072, "step": 40230 }, { "grad_norm": 0.12037447839975357, "learning_rate": 6.972225303951097e-05, "loss": 0.0071, "step": 40240 }, { "grad_norm": 0.21190309524536133, "learning_rate": 6.970705790438998e-05, "loss": 0.0054, "step": 40250 }, { "grad_norm": 0.1687343865633011, "learning_rate": 6.969186061413477e-05, "loss": 0.0089, "step": 40260 }, { "grad_norm": 0.0882556214928627, "learning_rate": 6.967666117040727e-05, "loss": 0.0055, "step": 40270 }, { "grad_norm": 0.10198947787284851, "learning_rate": 6.966145957486972e-05, "loss": 0.0075, "step": 40280 }, { "grad_norm": 0.08276767283678055, "learning_rate": 6.964625582918449e-05, "loss": 0.006, "step": 40290 }, { "grad_norm": 0.2174643576145172, "learning_rate": 6.963104993501425e-05, "loss": 0.0056, "step": 40300 }, { "grad_norm": 0.10981482267379761, "learning_rate": 6.961584189402192e-05, "loss": 0.0083, "step": 40310 }, { "grad_norm": 0.13713866472244263, "learning_rate": 6.96006317078706e-05, "loss": 0.0088, "step": 40320 }, { "grad_norm": 0.10382305085659027, "learning_rate": 6.958541937822367e-05, "loss": 0.0091, "step": 40330 }, { "grad_norm": 0.12050294131040573, "learning_rate": 6.957020490674473e-05, "loss": 0.0071, "step": 40340 }, { "grad_norm": 0.0745251402258873, "learning_rate": 6.95549882950976e-05, "loss": 0.007, "step": 40350 }, { "grad_norm": 0.05972050130367279, "learning_rate": 6.953976954494635e-05, "loss": 0.0108, "step": 40360 }, { "grad_norm": 0.06474476307630539, "learning_rate": 6.952454865795528e-05, "loss": 0.007, "step": 40370 }, { "grad_norm": 0.08881273120641708, "learning_rate": 6.950932563578892e-05, "loss": 0.006, "step": 40380 }, { "grad_norm": 0.070468008518219, "learning_rate": 6.949410048011206e-05, "loss": 0.0062, "step": 40390 }, { "grad_norm": 0.09102269262075424, "learning_rate": 6.947887319258966e-05, "loss": 0.0082, "step": 40400 }, { "grad_norm": 0.09054142981767654, "learning_rate": 6.946364377488696e-05, "loss": 0.0064, "step": 40410 }, { "grad_norm": 0.06469036638736725, "learning_rate": 6.944841222866947e-05, "loss": 0.0061, "step": 40420 }, { "grad_norm": 0.07381612062454224, "learning_rate": 6.943317855560284e-05, "loss": 0.0073, "step": 40430 }, { "grad_norm": 0.0718909353017807, "learning_rate": 6.941794275735302e-05, "loss": 0.0077, "step": 40440 }, { "grad_norm": 0.07024773210287094, "learning_rate": 6.94027048355862e-05, "loss": 0.0064, "step": 40450 }, { "grad_norm": 0.1288130283355713, "learning_rate": 6.938746479196871e-05, "loss": 0.0099, "step": 40460 }, { "grad_norm": 0.1873839795589447, "learning_rate": 6.937222262816724e-05, "loss": 0.0067, "step": 40470 }, { "grad_norm": 0.12044249475002289, "learning_rate": 6.935697834584865e-05, "loss": 0.0085, "step": 40480 }, { "grad_norm": 0.13554620742797852, "learning_rate": 6.934173194667998e-05, "loss": 0.0075, "step": 40490 }, { "grad_norm": 0.10460151731967926, "learning_rate": 6.93264834323286e-05, "loss": 0.0073, "step": 40500 }, { "grad_norm": 0.10104013234376907, "learning_rate": 6.931123280446206e-05, "loss": 0.0072, "step": 40510 }, { "grad_norm": 0.17047752439975739, "learning_rate": 6.929598006474811e-05, "loss": 0.0063, "step": 40520 }, { "grad_norm": 0.09574884921312332, "learning_rate": 6.92807252148548e-05, "loss": 0.0099, "step": 40530 }, { "grad_norm": 0.11691995710134506, "learning_rate": 6.926546825645037e-05, "loss": 0.0123, "step": 40540 }, { "grad_norm": 0.10034924745559692, "learning_rate": 6.92502091912033e-05, "loss": 0.0084, "step": 40550 }, { "grad_norm": 0.07750298082828522, "learning_rate": 6.923494802078229e-05, "loss": 0.009, "step": 40560 }, { "grad_norm": 0.19309936463832855, "learning_rate": 6.921968474685628e-05, "loss": 0.0054, "step": 40570 }, { "grad_norm": 0.09134625643491745, "learning_rate": 6.920441937109445e-05, "loss": 0.0066, "step": 40580 }, { "grad_norm": 0.16931265592575073, "learning_rate": 6.91891518951662e-05, "loss": 0.0061, "step": 40590 }, { "grad_norm": 0.12470642477273941, "learning_rate": 6.917388232074114e-05, "loss": 0.0083, "step": 40600 }, { "grad_norm": 0.19493629038333893, "learning_rate": 6.915861064948914e-05, "loss": 0.0076, "step": 40610 }, { "grad_norm": 0.12484418600797653, "learning_rate": 6.914333688308025e-05, "loss": 0.0059, "step": 40620 }, { "grad_norm": 0.21116185188293457, "learning_rate": 6.912806102318487e-05, "loss": 0.0072, "step": 40630 }, { "grad_norm": 0.119169682264328, "learning_rate": 6.911278307147347e-05, "loss": 0.0068, "step": 40640 }, { "grad_norm": 0.20327864587306976, "learning_rate": 6.909750302961684e-05, "loss": 0.0064, "step": 40650 }, { "grad_norm": 0.11790309846401215, "learning_rate": 6.9082220899286e-05, "loss": 0.0062, "step": 40660 }, { "grad_norm": 0.16683970391750336, "learning_rate": 6.906693668215219e-05, "loss": 0.006, "step": 40670 }, { "grad_norm": 0.11766699701547623, "learning_rate": 6.905165037988683e-05, "loss": 0.0066, "step": 40680 }, { "grad_norm": 0.13079077005386353, "learning_rate": 6.903636199416164e-05, "loss": 0.0098, "step": 40690 }, { "grad_norm": 0.12121004611253738, "learning_rate": 6.902107152664851e-05, "loss": 0.0056, "step": 40700 }, { "grad_norm": 0.1387893557548523, "learning_rate": 6.90057789790196e-05, "loss": 0.0083, "step": 40710 }, { "grad_norm": 0.07790316641330719, "learning_rate": 6.899048435294728e-05, "loss": 0.0081, "step": 40720 }, { "grad_norm": 0.061120036989450455, "learning_rate": 6.897518765010415e-05, "loss": 0.0055, "step": 40730 }, { "grad_norm": 0.08325415849685669, "learning_rate": 6.895988887216302e-05, "loss": 0.0085, "step": 40740 }, { "grad_norm": 0.09652239084243774, "learning_rate": 6.894458802079694e-05, "loss": 0.0062, "step": 40750 }, { "grad_norm": 0.0859035849571228, "learning_rate": 6.892928509767922e-05, "loss": 0.0065, "step": 40760 }, { "grad_norm": 0.0729164108633995, "learning_rate": 6.891398010448333e-05, "loss": 0.0037, "step": 40770 }, { "grad_norm": 0.14456301927566528, "learning_rate": 6.8898673042883e-05, "loss": 0.0084, "step": 40780 }, { "grad_norm": 0.1400545984506607, "learning_rate": 6.888336391455222e-05, "loss": 0.008, "step": 40790 }, { "grad_norm": 0.10406126081943512, "learning_rate": 6.886805272116513e-05, "loss": 0.0096, "step": 40800 }, { "grad_norm": 0.0870702937245369, "learning_rate": 6.885273946439617e-05, "loss": 0.0083, "step": 40810 }, { "grad_norm": 0.09343835711479187, "learning_rate": 6.883742414591998e-05, "loss": 0.0074, "step": 40820 }, { "grad_norm": 0.08139649033546448, "learning_rate": 6.88221067674114e-05, "loss": 0.0067, "step": 40830 }, { "grad_norm": 0.10755102336406708, "learning_rate": 6.88067873305455e-05, "loss": 0.0062, "step": 40840 }, { "grad_norm": 0.1612679660320282, "learning_rate": 6.879146583699765e-05, "loss": 0.0063, "step": 40850 }, { "grad_norm": 0.12968923151493073, "learning_rate": 6.877614228844334e-05, "loss": 0.0065, "step": 40860 }, { "grad_norm": 0.11774719506502151, "learning_rate": 6.876081668655832e-05, "loss": 0.0062, "step": 40870 }, { "grad_norm": 0.13959138095378876, "learning_rate": 6.874548903301861e-05, "loss": 0.0072, "step": 40880 }, { "grad_norm": 0.1271592378616333, "learning_rate": 6.873015932950039e-05, "loss": 0.0073, "step": 40890 }, { "grad_norm": 0.07531809061765671, "learning_rate": 6.871482757768012e-05, "loss": 0.0079, "step": 40900 }, { "grad_norm": 0.07569243758916855, "learning_rate": 6.869949377923442e-05, "loss": 0.0079, "step": 40910 }, { "grad_norm": 0.08027806133031845, "learning_rate": 6.868415793584022e-05, "loss": 0.0053, "step": 40920 }, { "grad_norm": 0.07668402045965195, "learning_rate": 6.86688200491746e-05, "loss": 0.0056, "step": 40930 }, { "grad_norm": 0.10495105385780334, "learning_rate": 6.865348012091488e-05, "loss": 0.0069, "step": 40940 }, { "grad_norm": 0.09707125276327133, "learning_rate": 6.86381381527386e-05, "loss": 0.009, "step": 40950 }, { "grad_norm": 0.0788203552365303, "learning_rate": 6.862279414632359e-05, "loss": 0.0073, "step": 40960 }, { "grad_norm": 0.10066135972738266, "learning_rate": 6.860744810334778e-05, "loss": 0.01, "step": 40970 }, { "grad_norm": 0.07547841966152191, "learning_rate": 6.859210002548943e-05, "loss": 0.0054, "step": 40980 }, { "grad_norm": 0.1140541285276413, "learning_rate": 6.8576749914427e-05, "loss": 0.0086, "step": 40990 }, { "grad_norm": 0.10427994281053543, "learning_rate": 6.85613977718391e-05, "loss": 0.0056, "step": 41000 }, { "grad_norm": 0.08464451134204865, "learning_rate": 6.854604359940464e-05, "loss": 0.0073, "step": 41010 }, { "grad_norm": 0.10699477791786194, "learning_rate": 6.853068739880276e-05, "loss": 0.0065, "step": 41020 }, { "grad_norm": 0.10147246718406677, "learning_rate": 6.851532917171276e-05, "loss": 0.0075, "step": 41030 }, { "grad_norm": 0.05891525372862816, "learning_rate": 6.849996891981418e-05, "loss": 0.006, "step": 41040 }, { "grad_norm": 0.06994379311800003, "learning_rate": 6.848460664478684e-05, "loss": 0.0065, "step": 41050 }, { "grad_norm": 0.0777430534362793, "learning_rate": 6.846924234831069e-05, "loss": 0.0054, "step": 41060 }, { "grad_norm": 0.04960641637444496, "learning_rate": 6.845387603206597e-05, "loss": 0.006, "step": 41070 }, { "grad_norm": 0.17164766788482666, "learning_rate": 6.843850769773312e-05, "loss": 0.007, "step": 41080 }, { "grad_norm": 0.1151646301150322, "learning_rate": 6.842313734699278e-05, "loss": 0.008, "step": 41090 }, { "grad_norm": 0.11529479175806046, "learning_rate": 6.840776498152584e-05, "loss": 0.008, "step": 41100 }, { "grad_norm": 0.09865366667509079, "learning_rate": 6.83923906030134e-05, "loss": 0.0064, "step": 41110 }, { "grad_norm": 0.13062259554862976, "learning_rate": 6.837701421313677e-05, "loss": 0.0058, "step": 41120 }, { "grad_norm": 0.0779501274228096, "learning_rate": 6.83616358135775e-05, "loss": 0.0052, "step": 41130 }, { "grad_norm": 0.06218855082988739, "learning_rate": 6.834625540601734e-05, "loss": 0.0051, "step": 41140 }, { "grad_norm": 0.20931175351142883, "learning_rate": 6.833087299213829e-05, "loss": 0.0067, "step": 41150 }, { "grad_norm": 0.10347355902194977, "learning_rate": 6.83154885736225e-05, "loss": 0.0072, "step": 41160 }, { "grad_norm": 0.15094508230686188, "learning_rate": 6.830010215215244e-05, "loss": 0.0083, "step": 41170 }, { "grad_norm": 0.10041618347167969, "learning_rate": 6.828471372941072e-05, "loss": 0.0069, "step": 41180 }, { "grad_norm": 0.19805988669395447, "learning_rate": 6.826932330708018e-05, "loss": 0.0091, "step": 41190 }, { "grad_norm": 0.11757447570562363, "learning_rate": 6.825393088684393e-05, "loss": 0.0109, "step": 41200 }, { "grad_norm": 0.07318616658449173, "learning_rate": 6.823853647038524e-05, "loss": 0.0059, "step": 41210 }, { "grad_norm": 0.06830034404993057, "learning_rate": 6.822314005938762e-05, "loss": 0.0062, "step": 41220 }, { "grad_norm": 0.055534396320581436, "learning_rate": 6.820774165553479e-05, "loss": 0.0047, "step": 41230 }, { "grad_norm": 0.11370885372161865, "learning_rate": 6.819234126051072e-05, "loss": 0.0076, "step": 41240 }, { "grad_norm": 0.08126167953014374, "learning_rate": 6.817693887599956e-05, "loss": 0.0091, "step": 41250 }, { "grad_norm": 0.11378307640552521, "learning_rate": 6.816153450368568e-05, "loss": 0.0068, "step": 41260 }, { "grad_norm": 0.06185431033372879, "learning_rate": 6.814612814525372e-05, "loss": 0.0083, "step": 41270 }, { "grad_norm": 0.15841543674468994, "learning_rate": 6.813071980238846e-05, "loss": 0.0097, "step": 41280 }, { "grad_norm": 0.14565426111221313, "learning_rate": 6.811530947677492e-05, "loss": 0.0053, "step": 41290 }, { "grad_norm": 0.1472836434841156, "learning_rate": 6.809989717009839e-05, "loss": 0.0074, "step": 41300 }, { "grad_norm": 0.1051977276802063, "learning_rate": 6.808448288404431e-05, "loss": 0.0068, "step": 41310 }, { "grad_norm": 0.07314585894346237, "learning_rate": 6.806906662029838e-05, "loss": 0.0072, "step": 41320 }, { "grad_norm": 0.1826484501361847, "learning_rate": 6.805364838054647e-05, "loss": 0.0065, "step": 41330 }, { "grad_norm": 0.1303073763847351, "learning_rate": 6.803822816647474e-05, "loss": 0.0083, "step": 41340 }, { "grad_norm": 0.08049268275499344, "learning_rate": 6.802280597976949e-05, "loss": 0.008, "step": 41350 }, { "grad_norm": 0.078916035592556, "learning_rate": 6.800738182211727e-05, "loss": 0.0066, "step": 41360 }, { "grad_norm": 0.0798073336482048, "learning_rate": 6.799195569520485e-05, "loss": 0.0059, "step": 41370 }, { "grad_norm": 0.13793373107910156, "learning_rate": 6.79765276007192e-05, "loss": 0.0071, "step": 41380 }, { "grad_norm": 0.15136930346488953, "learning_rate": 6.796109754034753e-05, "loss": 0.0119, "step": 41390 }, { "grad_norm": 0.09597548097372055, "learning_rate": 6.794566551577724e-05, "loss": 0.006, "step": 41400 }, { "grad_norm": 0.0876862108707428, "learning_rate": 6.793023152869594e-05, "loss": 0.0083, "step": 41410 }, { "grad_norm": 0.1167382225394249, "learning_rate": 6.791479558079147e-05, "loss": 0.0077, "step": 41420 }, { "grad_norm": 0.11986453831195831, "learning_rate": 6.789935767375191e-05, "loss": 0.0069, "step": 41430 }, { "grad_norm": 0.07781717926263809, "learning_rate": 6.78839178092655e-05, "loss": 0.0074, "step": 41440 }, { "grad_norm": 0.07652812451124191, "learning_rate": 6.786847598902072e-05, "loss": 0.0061, "step": 41450 }, { "grad_norm": 0.16160188615322113, "learning_rate": 6.785303221470627e-05, "loss": 0.0062, "step": 41460 }, { "grad_norm": 0.11453086882829666, "learning_rate": 6.783758648801108e-05, "loss": 0.0088, "step": 41470 }, { "grad_norm": 0.1352231353521347, "learning_rate": 6.782213881062422e-05, "loss": 0.008, "step": 41480 }, { "grad_norm": 0.07753923535346985, "learning_rate": 6.780668918423508e-05, "loss": 0.0077, "step": 41490 }, { "grad_norm": 0.18016080558300018, "learning_rate": 6.779123761053317e-05, "loss": 0.0084, "step": 41500 }, { "grad_norm": 0.09857835620641708, "learning_rate": 6.777578409120826e-05, "loss": 0.0068, "step": 41510 }, { "grad_norm": 0.10512500256299973, "learning_rate": 6.776032862795033e-05, "loss": 0.0062, "step": 41520 }, { "grad_norm": 0.2108439952135086, "learning_rate": 6.774487122244956e-05, "loss": 0.0081, "step": 41530 }, { "grad_norm": 0.11191405355930328, "learning_rate": 6.772941187639637e-05, "loss": 0.0075, "step": 41540 }, { "grad_norm": 0.14060312509536743, "learning_rate": 6.771395059148134e-05, "loss": 0.0088, "step": 41550 }, { "grad_norm": 0.08553213626146317, "learning_rate": 6.76984873693953e-05, "loss": 0.0064, "step": 41560 }, { "grad_norm": 0.09829257428646088, "learning_rate": 6.768302221182931e-05, "loss": 0.0078, "step": 41570 }, { "grad_norm": 0.12724152207374573, "learning_rate": 6.766755512047457e-05, "loss": 0.0074, "step": 41580 }, { "grad_norm": 0.10441654920578003, "learning_rate": 6.765208609702259e-05, "loss": 0.0102, "step": 41590 }, { "grad_norm": 0.16156119108200073, "learning_rate": 6.763661514316499e-05, "loss": 0.0074, "step": 41600 }, { "grad_norm": 0.07391995191574097, "learning_rate": 6.76211422605937e-05, "loss": 0.0059, "step": 41610 }, { "grad_norm": 0.21166662871837616, "learning_rate": 6.760566745100076e-05, "loss": 0.008, "step": 41620 }, { "grad_norm": 0.13306894898414612, "learning_rate": 6.75901907160785e-05, "loss": 0.0055, "step": 41630 }, { "grad_norm": 0.09754572808742523, "learning_rate": 6.757471205751943e-05, "loss": 0.0046, "step": 41640 }, { "grad_norm": 0.18282724916934967, "learning_rate": 6.75592314770163e-05, "loss": 0.0068, "step": 41650 }, { "grad_norm": 0.14133024215698242, "learning_rate": 6.7543748976262e-05, "loss": 0.0067, "step": 41660 }, { "grad_norm": 0.09045789390802383, "learning_rate": 6.752826455694968e-05, "loss": 0.0104, "step": 41670 }, { "grad_norm": 0.13542135059833527, "learning_rate": 6.751277822077271e-05, "loss": 0.0071, "step": 41680 }, { "grad_norm": 0.1188749447464943, "learning_rate": 6.749728996942465e-05, "loss": 0.0089, "step": 41690 }, { "grad_norm": 0.1283278614282608, "learning_rate": 6.748179980459924e-05, "loss": 0.0078, "step": 41700 }, { "grad_norm": 0.0775395855307579, "learning_rate": 6.746630772799052e-05, "loss": 0.0071, "step": 41710 }, { "grad_norm": 0.0784173533320427, "learning_rate": 6.745081374129262e-05, "loss": 0.0065, "step": 41720 }, { "grad_norm": 0.06105516105890274, "learning_rate": 6.743531784619998e-05, "loss": 0.0053, "step": 41730 }, { "grad_norm": 0.09968402236700058, "learning_rate": 6.741982004440719e-05, "loss": 0.007, "step": 41740 }, { "grad_norm": 0.0854405090212822, "learning_rate": 6.740432033760907e-05, "loss": 0.006, "step": 41750 }, { "grad_norm": 0.09922847896814346, "learning_rate": 6.738881872750066e-05, "loss": 0.0064, "step": 41760 }, { "grad_norm": 0.16293463110923767, "learning_rate": 6.737331521577715e-05, "loss": 0.011, "step": 41770 }, { "grad_norm": 0.11796966195106506, "learning_rate": 6.735780980413403e-05, "loss": 0.0077, "step": 41780 }, { "grad_norm": 0.12253914028406143, "learning_rate": 6.734230249426692e-05, "loss": 0.0074, "step": 41790 }, { "grad_norm": 0.07481222599744797, "learning_rate": 6.732679328787168e-05, "loss": 0.0065, "step": 41800 }, { "grad_norm": 0.12243524938821793, "learning_rate": 6.731128218664438e-05, "loss": 0.007, "step": 41810 }, { "grad_norm": 0.08520809561014175, "learning_rate": 6.729576919228129e-05, "loss": 0.0067, "step": 41820 }, { "grad_norm": 0.06498747318983078, "learning_rate": 6.728025430647888e-05, "loss": 0.0073, "step": 41830 }, { "grad_norm": 0.04542247951030731, "learning_rate": 6.726473753093383e-05, "loss": 0.0052, "step": 41840 }, { "grad_norm": 0.07606728374958038, "learning_rate": 6.724921886734305e-05, "loss": 0.009, "step": 41850 }, { "grad_norm": 0.11223706603050232, "learning_rate": 6.723369831740363e-05, "loss": 0.007, "step": 41860 }, { "grad_norm": 0.07007032632827759, "learning_rate": 6.721817588281288e-05, "loss": 0.0054, "step": 41870 }, { "grad_norm": 0.11934028565883636, "learning_rate": 6.720265156526828e-05, "loss": 0.0086, "step": 41880 }, { "grad_norm": 0.1447666734457016, "learning_rate": 6.718712536646758e-05, "loss": 0.0065, "step": 41890 }, { "grad_norm": 0.0936417281627655, "learning_rate": 6.71715972881087e-05, "loss": 0.0068, "step": 41900 }, { "grad_norm": 0.09760195761919022, "learning_rate": 6.715606733188974e-05, "loss": 0.0087, "step": 41910 }, { "grad_norm": 0.09674108028411865, "learning_rate": 6.714053549950909e-05, "loss": 0.0062, "step": 41920 }, { "grad_norm": 0.13050979375839233, "learning_rate": 6.71250017926652e-05, "loss": 0.0061, "step": 41930 }, { "grad_norm": 0.06825686991214752, "learning_rate": 6.71094662130569e-05, "loss": 0.0053, "step": 41940 }, { "grad_norm": 0.09781289845705032, "learning_rate": 6.709392876238307e-05, "loss": 0.007, "step": 41950 }, { "grad_norm": 0.15734955668449402, "learning_rate": 6.707838944234289e-05, "loss": 0.0078, "step": 41960 }, { "grad_norm": 0.12023162096738815, "learning_rate": 6.706284825463574e-05, "loss": 0.0081, "step": 41970 }, { "grad_norm": 0.09402959793806076, "learning_rate": 6.704730520096114e-05, "loss": 0.0071, "step": 41980 }, { "grad_norm": 0.12802597880363464, "learning_rate": 6.703176028301888e-05, "loss": 0.0072, "step": 41990 }, { "grad_norm": 0.05313875898718834, "learning_rate": 6.701621350250892e-05, "loss": 0.0043, "step": 42000 }, { "grad_norm": 0.11989486962556839, "learning_rate": 6.700066486113144e-05, "loss": 0.0063, "step": 42010 }, { "grad_norm": 0.14167824387550354, "learning_rate": 6.69851143605868e-05, "loss": 0.0094, "step": 42020 }, { "grad_norm": 0.11977881193161011, "learning_rate": 6.696956200257561e-05, "loss": 0.0095, "step": 42030 }, { "grad_norm": 0.09861649572849274, "learning_rate": 6.695400778879863e-05, "loss": 0.0084, "step": 42040 }, { "grad_norm": 0.11974193155765533, "learning_rate": 6.693845172095683e-05, "loss": 0.0056, "step": 42050 }, { "grad_norm": 0.08138927817344666, "learning_rate": 6.692289380075142e-05, "loss": 0.0077, "step": 42060 }, { "grad_norm": 0.10264744609594345, "learning_rate": 6.690733402988379e-05, "loss": 0.007, "step": 42070 }, { "grad_norm": 0.0843694806098938, "learning_rate": 6.689177241005553e-05, "loss": 0.0055, "step": 42080 }, { "grad_norm": 0.0696311816573143, "learning_rate": 6.687620894296844e-05, "loss": 0.0053, "step": 42090 }, { "grad_norm": 0.13895510137081146, "learning_rate": 6.686064363032451e-05, "loss": 0.0086, "step": 42100 }, { "grad_norm": 0.08823805302381516, "learning_rate": 6.684507647382596e-05, "loss": 0.0064, "step": 42110 }, { "grad_norm": 0.07985598593950272, "learning_rate": 6.682950747517513e-05, "loss": 0.0055, "step": 42120 }, { "grad_norm": 0.06037231907248497, "learning_rate": 6.68139366360747e-05, "loss": 0.0066, "step": 42130 }, { "grad_norm": 0.06415089964866638, "learning_rate": 6.679836395822744e-05, "loss": 0.0054, "step": 42140 }, { "grad_norm": 0.09589865058660507, "learning_rate": 6.678278944333633e-05, "loss": 0.0059, "step": 42150 }, { "grad_norm": 0.07209637761116028, "learning_rate": 6.676721309310462e-05, "loss": 0.0058, "step": 42160 }, { "grad_norm": 0.12413614243268967, "learning_rate": 6.67516349092357e-05, "loss": 0.0074, "step": 42170 }, { "grad_norm": 0.10938883572816849, "learning_rate": 6.673605489343315e-05, "loss": 0.0066, "step": 42180 }, { "grad_norm": 0.09532558172941208, "learning_rate": 6.672047304740081e-05, "loss": 0.0058, "step": 42190 }, { "grad_norm": 0.14902839064598083, "learning_rate": 6.670488937284268e-05, "loss": 0.0082, "step": 42200 }, { "grad_norm": 0.10222967714071274, "learning_rate": 6.668930387146294e-05, "loss": 0.0057, "step": 42210 }, { "grad_norm": 0.12287641316652298, "learning_rate": 6.667371654496605e-05, "loss": 0.0064, "step": 42220 }, { "grad_norm": 0.07626648247241974, "learning_rate": 6.665812739505659e-05, "loss": 0.0078, "step": 42230 }, { "grad_norm": 0.09846203774213791, "learning_rate": 6.664253642343935e-05, "loss": 0.0067, "step": 42240 }, { "grad_norm": 0.10948774963617325, "learning_rate": 6.662694363181935e-05, "loss": 0.0056, "step": 42250 }, { "grad_norm": 0.09443772584199905, "learning_rate": 6.66113490219018e-05, "loss": 0.0056, "step": 42260 }, { "grad_norm": 0.0845741257071495, "learning_rate": 6.659575259539207e-05, "loss": 0.007, "step": 42270 }, { "grad_norm": 0.2730492949485779, "learning_rate": 6.65801543539958e-05, "loss": 0.0075, "step": 42280 }, { "grad_norm": 0.11976255476474762, "learning_rate": 6.65645542994188e-05, "loss": 0.0071, "step": 42290 }, { "grad_norm": 0.09272312372922897, "learning_rate": 6.654895243336702e-05, "loss": 0.0077, "step": 42300 }, { "grad_norm": 0.0821327194571495, "learning_rate": 6.65333487575467e-05, "loss": 0.0064, "step": 42310 }, { "grad_norm": 0.15129880607128143, "learning_rate": 6.651774327366422e-05, "loss": 0.0102, "step": 42320 }, { "grad_norm": 0.10811948031187057, "learning_rate": 6.650213598342613e-05, "loss": 0.0077, "step": 42330 }, { "grad_norm": 0.11913901567459106, "learning_rate": 6.648652688853931e-05, "loss": 0.0075, "step": 42340 }, { "grad_norm": 0.08090473711490631, "learning_rate": 6.647091599071066e-05, "loss": 0.0085, "step": 42350 }, { "grad_norm": 0.09832499921321869, "learning_rate": 6.645530329164743e-05, "loss": 0.0069, "step": 42360 }, { "grad_norm": 0.07600824534893036, "learning_rate": 6.643968879305694e-05, "loss": 0.0082, "step": 42370 }, { "grad_norm": 0.08926315605640411, "learning_rate": 6.642407249664683e-05, "loss": 0.01, "step": 42380 }, { "grad_norm": 0.09534474462270737, "learning_rate": 6.640845440412483e-05, "loss": 0.0078, "step": 42390 }, { "grad_norm": 0.06518346816301346, "learning_rate": 6.639283451719893e-05, "loss": 0.0045, "step": 42400 }, { "grad_norm": 0.07555816322565079, "learning_rate": 6.637721283757729e-05, "loss": 0.0057, "step": 42410 }, { "grad_norm": 0.08790435642004013, "learning_rate": 6.636158936696828e-05, "loss": 0.008, "step": 42420 }, { "grad_norm": 0.12007711082696915, "learning_rate": 6.634596410708047e-05, "loss": 0.0084, "step": 42430 }, { "grad_norm": 0.0850740522146225, "learning_rate": 6.633033705962257e-05, "loss": 0.0052, "step": 42440 }, { "grad_norm": 0.09660099446773529, "learning_rate": 6.631470822630359e-05, "loss": 0.006, "step": 42450 }, { "grad_norm": 0.09699466079473495, "learning_rate": 6.629907760883263e-05, "loss": 0.0072, "step": 42460 }, { "grad_norm": 0.11507735401391983, "learning_rate": 6.628344520891905e-05, "loss": 0.0088, "step": 42470 }, { "grad_norm": 0.1535254716873169, "learning_rate": 6.62678110282724e-05, "loss": 0.0091, "step": 42480 }, { "grad_norm": 0.07140956819057465, "learning_rate": 6.625217506860237e-05, "loss": 0.0063, "step": 42490 }, { "grad_norm": 0.1263730674982071, "learning_rate": 6.623653733161892e-05, "loss": 0.0085, "step": 42500 }, { "grad_norm": 0.09298498928546906, "learning_rate": 6.622089781903217e-05, "loss": 0.0067, "step": 42510 }, { "grad_norm": 0.14787890017032623, "learning_rate": 6.620525653255241e-05, "loss": 0.0109, "step": 42520 }, { "grad_norm": 0.12421035021543503, "learning_rate": 6.618961347389015e-05, "loss": 0.0075, "step": 42530 }, { "grad_norm": 0.19080069661140442, "learning_rate": 6.617396864475613e-05, "loss": 0.0074, "step": 42540 }, { "grad_norm": 0.11959217488765717, "learning_rate": 6.61583220468612e-05, "loss": 0.0069, "step": 42550 }, { "grad_norm": 0.0809231549501419, "learning_rate": 6.614267368191645e-05, "loss": 0.0063, "step": 42560 }, { "grad_norm": 0.11325255036354065, "learning_rate": 6.61270235516332e-05, "loss": 0.0057, "step": 42570 }, { "grad_norm": 0.14409054815769196, "learning_rate": 6.611137165772287e-05, "loss": 0.0089, "step": 42580 }, { "grad_norm": 0.09855491667985916, "learning_rate": 6.60957180018972e-05, "loss": 0.0068, "step": 42590 }, { "grad_norm": 0.1030203104019165, "learning_rate": 6.608006258586797e-05, "loss": 0.0105, "step": 42600 }, { "grad_norm": 0.1567572057247162, "learning_rate": 6.606440541134731e-05, "loss": 0.0083, "step": 42610 }, { "grad_norm": 0.13422740995883942, "learning_rate": 6.60487464800474e-05, "loss": 0.0073, "step": 42620 }, { "grad_norm": 0.12987381219863892, "learning_rate": 6.603308579368071e-05, "loss": 0.0117, "step": 42630 }, { "grad_norm": 0.14268694818019867, "learning_rate": 6.601742335395987e-05, "loss": 0.0072, "step": 42640 }, { "grad_norm": 0.06414268165826797, "learning_rate": 6.600175916259769e-05, "loss": 0.0067, "step": 42650 }, { "grad_norm": 0.19073839485645294, "learning_rate": 6.598609322130718e-05, "loss": 0.0074, "step": 42660 }, { "grad_norm": 0.11934112757444382, "learning_rate": 6.597042553180155e-05, "loss": 0.0069, "step": 42670 }, { "grad_norm": 0.10984747111797333, "learning_rate": 6.59547560957942e-05, "loss": 0.0065, "step": 42680 }, { "grad_norm": 0.11734600365161896, "learning_rate": 6.59390849149987e-05, "loss": 0.0082, "step": 42690 }, { "grad_norm": 0.12141596525907516, "learning_rate": 6.592341199112886e-05, "loss": 0.0099, "step": 42700 }, { "grad_norm": 0.17809097468852997, "learning_rate": 6.590773732589861e-05, "loss": 0.0097, "step": 42710 }, { "grad_norm": 0.09001266956329346, "learning_rate": 6.589206092102213e-05, "loss": 0.0072, "step": 42720 }, { "grad_norm": 0.198618546128273, "learning_rate": 6.587638277821375e-05, "loss": 0.0095, "step": 42730 }, { "grad_norm": 0.09888534247875214, "learning_rate": 6.586070289918803e-05, "loss": 0.0073, "step": 42740 }, { "grad_norm": 0.08648759126663208, "learning_rate": 6.584502128565968e-05, "loss": 0.0053, "step": 42750 }, { "grad_norm": 0.09076053649187088, "learning_rate": 6.582933793934363e-05, "loss": 0.006, "step": 42760 }, { "grad_norm": 0.09200500696897507, "learning_rate": 6.581365286195498e-05, "loss": 0.0052, "step": 42770 }, { "grad_norm": 0.06967790424823761, "learning_rate": 6.579796605520903e-05, "loss": 0.0062, "step": 42780 }, { "grad_norm": 0.15907584130764008, "learning_rate": 6.578227752082127e-05, "loss": 0.0083, "step": 42790 }, { "grad_norm": 0.12466548383235931, "learning_rate": 6.576658726050735e-05, "loss": 0.0089, "step": 42800 }, { "grad_norm": 0.12187598645687103, "learning_rate": 6.575089527598316e-05, "loss": 0.0058, "step": 42810 }, { "grad_norm": 0.07267260551452637, "learning_rate": 6.573520156896476e-05, "loss": 0.009, "step": 42820 }, { "grad_norm": 0.10124169290065765, "learning_rate": 6.571950614116835e-05, "loss": 0.0088, "step": 42830 }, { "grad_norm": 0.06982063502073288, "learning_rate": 6.570380899431039e-05, "loss": 0.0079, "step": 42840 }, { "grad_norm": 0.13150067627429962, "learning_rate": 6.568811013010749e-05, "loss": 0.0073, "step": 42850 }, { "grad_norm": 0.09651219099760056, "learning_rate": 6.567240955027645e-05, "loss": 0.0077, "step": 42860 }, { "grad_norm": 0.09148044884204865, "learning_rate": 6.565670725653427e-05, "loss": 0.0058, "step": 42870 }, { "grad_norm": 0.06947211921215057, "learning_rate": 6.564100325059812e-05, "loss": 0.0088, "step": 42880 }, { "grad_norm": 0.08074070513248444, "learning_rate": 6.562529753418538e-05, "loss": 0.0055, "step": 42890 }, { "grad_norm": 0.08274848014116287, "learning_rate": 6.56095901090136e-05, "loss": 0.0082, "step": 42900 }, { "grad_norm": 0.09606537222862244, "learning_rate": 6.55938809768005e-05, "loss": 0.0057, "step": 42910 }, { "grad_norm": 0.07819411903619766, "learning_rate": 6.557817013926402e-05, "loss": 0.0064, "step": 42920 }, { "grad_norm": 0.11395267397165298, "learning_rate": 6.556245759812227e-05, "loss": 0.0067, "step": 42930 }, { "grad_norm": 0.1289549171924591, "learning_rate": 6.554674335509357e-05, "loss": 0.007, "step": 42940 }, { "grad_norm": 0.08286314457654953, "learning_rate": 6.553102741189638e-05, "loss": 0.0064, "step": 42950 }, { "grad_norm": 0.13219250738620758, "learning_rate": 6.55153097702494e-05, "loss": 0.0078, "step": 42960 }, { "grad_norm": 0.0954226478934288, "learning_rate": 6.549959043187146e-05, "loss": 0.0072, "step": 42970 }, { "grad_norm": 0.12132316082715988, "learning_rate": 6.54838693984816e-05, "loss": 0.0092, "step": 42980 }, { "grad_norm": 0.07916747778654099, "learning_rate": 6.546814667179909e-05, "loss": 0.0058, "step": 42990 }, { "grad_norm": 0.11271564662456512, "learning_rate": 6.545242225354328e-05, "loss": 0.0059, "step": 43000 }, { "grad_norm": 0.06782085448503494, "learning_rate": 6.543669614543383e-05, "loss": 0.0069, "step": 43010 }, { "grad_norm": 0.12776829302310944, "learning_rate": 6.542096834919049e-05, "loss": 0.0064, "step": 43020 }, { "grad_norm": 0.08213143050670624, "learning_rate": 6.540523886653324e-05, "loss": 0.0056, "step": 43030 }, { "grad_norm": 0.08763021975755692, "learning_rate": 6.538950769918222e-05, "loss": 0.0074, "step": 43040 }, { "grad_norm": 0.10460745543241501, "learning_rate": 6.537377484885779e-05, "loss": 0.0071, "step": 43050 }, { "grad_norm": 0.09774235635995865, "learning_rate": 6.535804031728045e-05, "loss": 0.0069, "step": 43060 }, { "grad_norm": 0.12378769367933273, "learning_rate": 6.534230410617092e-05, "loss": 0.01, "step": 43070 }, { "grad_norm": 0.08770152926445007, "learning_rate": 6.532656621725007e-05, "loss": 0.007, "step": 43080 }, { "grad_norm": 0.07138214260339737, "learning_rate": 6.531082665223898e-05, "loss": 0.0075, "step": 43090 }, { "grad_norm": 0.11866483837366104, "learning_rate": 6.529508541285889e-05, "loss": 0.0069, "step": 43100 }, { "grad_norm": 0.0884690210223198, "learning_rate": 6.527934250083128e-05, "loss": 0.0068, "step": 43110 }, { "grad_norm": 0.12955451011657715, "learning_rate": 6.526359791787775e-05, "loss": 0.0085, "step": 43120 }, { "grad_norm": 0.12176160514354706, "learning_rate": 6.524785166572009e-05, "loss": 0.008, "step": 43130 }, { "grad_norm": 0.08860543370246887, "learning_rate": 6.523210374608029e-05, "loss": 0.0059, "step": 43140 }, { "grad_norm": 0.0933631956577301, "learning_rate": 6.521635416068054e-05, "loss": 0.0073, "step": 43150 }, { "grad_norm": 0.07626239955425262, "learning_rate": 6.520060291124317e-05, "loss": 0.0058, "step": 43160 }, { "grad_norm": 0.08036171644926071, "learning_rate": 6.518484999949072e-05, "loss": 0.0087, "step": 43170 }, { "grad_norm": 0.09292938560247421, "learning_rate": 6.516909542714591e-05, "loss": 0.0084, "step": 43180 }, { "grad_norm": 0.11697519570589066, "learning_rate": 6.515333919593161e-05, "loss": 0.0074, "step": 43190 }, { "grad_norm": 0.1164688691496849, "learning_rate": 6.513758130757094e-05, "loss": 0.0069, "step": 43200 }, { "grad_norm": 0.0864207074046135, "learning_rate": 6.512182176378713e-05, "loss": 0.0073, "step": 43210 }, { "grad_norm": 0.08676426112651825, "learning_rate": 6.510606056630362e-05, "loss": 0.0095, "step": 43220 }, { "grad_norm": 0.18842482566833496, "learning_rate": 6.509029771684405e-05, "loss": 0.0113, "step": 43230 }, { "grad_norm": 0.09231160581111908, "learning_rate": 6.50745332171322e-05, "loss": 0.0077, "step": 43240 }, { "grad_norm": 0.10178966075181961, "learning_rate": 6.505876706889207e-05, "loss": 0.007, "step": 43250 }, { "grad_norm": 0.07136186957359314, "learning_rate": 6.504299927384781e-05, "loss": 0.0063, "step": 43260 }, { "grad_norm": 0.12346676737070084, "learning_rate": 6.502722983372378e-05, "loss": 0.0052, "step": 43270 }, { "grad_norm": 0.07289237529039383, "learning_rate": 6.501145875024446e-05, "loss": 0.0067, "step": 43280 }, { "grad_norm": 0.18262408673763275, "learning_rate": 6.499568602513459e-05, "loss": 0.0072, "step": 43290 }, { "grad_norm": 0.10741240531206131, "learning_rate": 6.497991166011903e-05, "loss": 0.0063, "step": 43300 }, { "grad_norm": 0.1342296153306961, "learning_rate": 6.49641356569229e-05, "loss": 0.0074, "step": 43310 }, { "grad_norm": 0.09687606990337372, "learning_rate": 6.494835801727135e-05, "loss": 0.0075, "step": 43320 }, { "grad_norm": 0.20260833203792572, "learning_rate": 6.493257874288986e-05, "loss": 0.0109, "step": 43330 }, { "grad_norm": 0.08791913092136383, "learning_rate": 6.491679783550399e-05, "loss": 0.0082, "step": 43340 }, { "grad_norm": 0.0891316831111908, "learning_rate": 6.490101529683957e-05, "loss": 0.0078, "step": 43350 }, { "grad_norm": 0.07671427726745605, "learning_rate": 6.488523112862248e-05, "loss": 0.0067, "step": 43360 }, { "grad_norm": 0.13851866126060486, "learning_rate": 6.486944533257891e-05, "loss": 0.0085, "step": 43370 }, { "grad_norm": 0.09154354780912399, "learning_rate": 6.485365791043517e-05, "loss": 0.0064, "step": 43380 }, { "grad_norm": 0.18338997662067413, "learning_rate": 6.483786886391772e-05, "loss": 0.0089, "step": 43390 }, { "grad_norm": 0.08473596721887589, "learning_rate": 6.482207819475323e-05, "loss": 0.0077, "step": 43400 }, { "grad_norm": 0.21674562990665436, "learning_rate": 6.480628590466857e-05, "loss": 0.0084, "step": 43410 }, { "grad_norm": 0.1285100281238556, "learning_rate": 6.479049199539074e-05, "loss": 0.0098, "step": 43420 }, { "grad_norm": 0.05899868533015251, "learning_rate": 6.477469646864693e-05, "loss": 0.0078, "step": 43430 }, { "grad_norm": 0.13010172545909882, "learning_rate": 6.475889932616454e-05, "loss": 0.006, "step": 43440 }, { "grad_norm": 0.1003371998667717, "learning_rate": 6.474310056967111e-05, "loss": 0.006, "step": 43450 }, { "grad_norm": 0.09624239802360535, "learning_rate": 6.472730020089437e-05, "loss": 0.0069, "step": 43460 }, { "grad_norm": 0.06746122241020203, "learning_rate": 6.471149822156222e-05, "loss": 0.0062, "step": 43470 }, { "grad_norm": 0.07910186797380447, "learning_rate": 6.469569463340275e-05, "loss": 0.008, "step": 43480 }, { "grad_norm": 0.07806196063756943, "learning_rate": 6.467988943814418e-05, "loss": 0.0062, "step": 43490 }, { "grad_norm": 0.08255381882190704, "learning_rate": 6.4664082637515e-05, "loss": 0.0065, "step": 43500 }, { "grad_norm": 0.053265754133462906, "learning_rate": 6.464827423324377e-05, "loss": 0.0051, "step": 43510 }, { "grad_norm": 0.15073388814926147, "learning_rate": 6.463246422705931e-05, "loss": 0.0072, "step": 43520 }, { "grad_norm": 0.1033555120229721, "learning_rate": 6.461665262069055e-05, "loss": 0.0068, "step": 43530 }, { "grad_norm": 0.07749827206134796, "learning_rate": 6.460083941586665e-05, "loss": 0.0062, "step": 43540 }, { "grad_norm": 0.06346999108791351, "learning_rate": 6.45850246143169e-05, "loss": 0.0059, "step": 43550 }, { "grad_norm": 0.07168208807706833, "learning_rate": 6.456920821777077e-05, "loss": 0.0062, "step": 43560 }, { "grad_norm": 0.11412904411554337, "learning_rate": 6.455339022795796e-05, "loss": 0.0063, "step": 43570 }, { "grad_norm": 0.09923383593559265, "learning_rate": 6.453757064660825e-05, "loss": 0.0054, "step": 43580 }, { "grad_norm": 0.09835192561149597, "learning_rate": 6.452174947545169e-05, "loss": 0.005, "step": 43590 }, { "grad_norm": 0.10023532062768936, "learning_rate": 6.450592671621842e-05, "loss": 0.0067, "step": 43600 }, { "grad_norm": 0.08452002704143524, "learning_rate": 6.449010237063882e-05, "loss": 0.0054, "step": 43610 }, { "grad_norm": 0.07238200306892395, "learning_rate": 6.447427644044343e-05, "loss": 0.0062, "step": 43620 }, { "grad_norm": 0.0973048061132431, "learning_rate": 6.44584489273629e-05, "loss": 0.0055, "step": 43630 }, { "grad_norm": 0.0880722850561142, "learning_rate": 6.444261983312817e-05, "loss": 0.0074, "step": 43640 }, { "grad_norm": 0.08524292707443237, "learning_rate": 6.442678915947023e-05, "loss": 0.007, "step": 43650 }, { "grad_norm": 0.07652126997709274, "learning_rate": 6.441095690812032e-05, "loss": 0.0066, "step": 43660 }, { "grad_norm": 0.13542436063289642, "learning_rate": 6.439512308080982e-05, "loss": 0.0074, "step": 43670 }, { "grad_norm": 0.09941430389881134, "learning_rate": 6.43792876792703e-05, "loss": 0.0081, "step": 43680 }, { "grad_norm": 0.07439383864402771, "learning_rate": 6.43634507052335e-05, "loss": 0.0075, "step": 43690 }, { "grad_norm": 0.14352594316005707, "learning_rate": 6.434761216043135e-05, "loss": 0.011, "step": 43700 }, { "grad_norm": 0.10897988826036453, "learning_rate": 6.433177204659589e-05, "loss": 0.0083, "step": 43710 }, { "grad_norm": 0.12982216477394104, "learning_rate": 6.431593036545938e-05, "loss": 0.0106, "step": 43720 }, { "grad_norm": 0.08755652606487274, "learning_rate": 6.430008711875426e-05, "loss": 0.0057, "step": 43730 }, { "grad_norm": 0.1077817901968956, "learning_rate": 6.42842423082131e-05, "loss": 0.006, "step": 43740 }, { "grad_norm": 0.11291032284498215, "learning_rate": 6.426839593556868e-05, "loss": 0.0064, "step": 43750 }, { "grad_norm": 0.08771693706512451, "learning_rate": 6.425254800255392e-05, "loss": 0.0056, "step": 43760 }, { "grad_norm": 0.0708744153380394, "learning_rate": 6.423669851090196e-05, "loss": 0.0049, "step": 43770 }, { "grad_norm": 0.0934615284204483, "learning_rate": 6.422084746234604e-05, "loss": 0.0071, "step": 43780 }, { "grad_norm": 0.1142929270863533, "learning_rate": 6.420499485861965e-05, "loss": 0.0079, "step": 43790 }, { "grad_norm": 0.07307058572769165, "learning_rate": 6.418914070145636e-05, "loss": 0.0079, "step": 43800 }, { "grad_norm": 0.07030721753835678, "learning_rate": 6.417328499258996e-05, "loss": 0.0046, "step": 43810 }, { "grad_norm": 0.12295741587877274, "learning_rate": 6.415742773375446e-05, "loss": 0.0085, "step": 43820 }, { "grad_norm": 0.08742213249206543, "learning_rate": 6.414156892668393e-05, "loss": 0.0082, "step": 43830 }, { "grad_norm": 0.15849894285202026, "learning_rate": 6.412570857311267e-05, "loss": 0.0081, "step": 43840 }, { "grad_norm": 0.11189042031764984, "learning_rate": 6.410984667477518e-05, "loss": 0.0081, "step": 43850 }, { "grad_norm": 0.2421284317970276, "learning_rate": 6.409398323340607e-05, "loss": 0.0107, "step": 43860 }, { "grad_norm": 0.1020430400967598, "learning_rate": 6.407811825074012e-05, "loss": 0.0086, "step": 43870 }, { "grad_norm": 0.1415354609489441, "learning_rate": 6.406225172851234e-05, "loss": 0.008, "step": 43880 }, { "grad_norm": 0.12155470997095108, "learning_rate": 6.404638366845786e-05, "loss": 0.005, "step": 43890 }, { "grad_norm": 0.07632791250944138, "learning_rate": 6.403051407231196e-05, "loss": 0.0068, "step": 43900 }, { "grad_norm": 0.0686677023768425, "learning_rate": 6.401464294181016e-05, "loss": 0.0064, "step": 43910 }, { "grad_norm": 0.06008929759263992, "learning_rate": 6.399877027868808e-05, "loss": 0.0054, "step": 43920 }, { "grad_norm": 0.11671821027994156, "learning_rate": 6.39828960846815e-05, "loss": 0.0054, "step": 43930 }, { "grad_norm": 0.0920315533876419, "learning_rate": 6.396702036152644e-05, "loss": 0.0062, "step": 43940 }, { "grad_norm": 0.13671427965164185, "learning_rate": 6.395114311095905e-05, "loss": 0.0099, "step": 43950 }, { "grad_norm": 0.0871855840086937, "learning_rate": 6.393526433471559e-05, "loss": 0.0064, "step": 43960 }, { "grad_norm": 0.08075901865959167, "learning_rate": 6.391938403453262e-05, "loss": 0.0047, "step": 43970 }, { "grad_norm": 0.0769515112042427, "learning_rate": 6.390350221214671e-05, "loss": 0.0073, "step": 43980 }, { "grad_norm": 0.052568551152944565, "learning_rate": 6.38876188692947e-05, "loss": 0.0055, "step": 43990 }, { "grad_norm": 0.2313539683818817, "learning_rate": 6.387173400771358e-05, "loss": 0.009, "step": 44000 }, { "grad_norm": 0.1279897689819336, "learning_rate": 6.385584762914047e-05, "loss": 0.0075, "step": 44010 }, { "grad_norm": 0.17510610818862915, "learning_rate": 6.38399597353127e-05, "loss": 0.0069, "step": 44020 }, { "grad_norm": 0.12260887026786804, "learning_rate": 6.382407032796775e-05, "loss": 0.0057, "step": 44030 }, { "grad_norm": 0.06880616396665573, "learning_rate": 6.380817940884325e-05, "loss": 0.0068, "step": 44040 }, { "grad_norm": 0.2003394067287445, "learning_rate": 6.379228697967702e-05, "loss": 0.0053, "step": 44050 }, { "grad_norm": 0.1267654001712799, "learning_rate": 6.3776393042207e-05, "loss": 0.0082, "step": 44060 }, { "grad_norm": 0.09887261688709259, "learning_rate": 6.376049759817137e-05, "loss": 0.0072, "step": 44070 }, { "grad_norm": 0.09274732321500778, "learning_rate": 6.37446006493084e-05, "loss": 0.0056, "step": 44080 }, { "grad_norm": 0.16161920130252838, "learning_rate": 6.372870219735656e-05, "loss": 0.0051, "step": 44090 }, { "grad_norm": 0.10974406450986862, "learning_rate": 6.371280224405452e-05, "loss": 0.0079, "step": 44100 }, { "grad_norm": 0.13113930821418762, "learning_rate": 6.369690079114101e-05, "loss": 0.0064, "step": 44110 }, { "grad_norm": 0.08996133506298065, "learning_rate": 6.368099784035504e-05, "loss": 0.0059, "step": 44120 }, { "grad_norm": 0.054381631314754486, "learning_rate": 6.366509339343572e-05, "loss": 0.0052, "step": 44130 }, { "grad_norm": 0.26573336124420166, "learning_rate": 6.364918745212232e-05, "loss": 0.0061, "step": 44140 }, { "grad_norm": 0.09926997125148773, "learning_rate": 6.363328001815431e-05, "loss": 0.0074, "step": 44150 }, { "grad_norm": 0.10638139396905899, "learning_rate": 6.361737109327128e-05, "loss": 0.0067, "step": 44160 }, { "grad_norm": 0.07301458716392517, "learning_rate": 6.360146067921301e-05, "loss": 0.0075, "step": 44170 }, { "grad_norm": 0.07641617953777313, "learning_rate": 6.35855487777195e-05, "loss": 0.005, "step": 44180 }, { "grad_norm": 0.08452394604682922, "learning_rate": 6.356963539053076e-05, "loss": 0.0054, "step": 44190 }, { "grad_norm": 0.06142911687493324, "learning_rate": 6.355372051938711e-05, "loss": 0.0071, "step": 44200 }, { "grad_norm": 0.07174204289913177, "learning_rate": 6.353780416602894e-05, "loss": 0.005, "step": 44210 }, { "grad_norm": 0.09730717539787292, "learning_rate": 6.352188633219689e-05, "loss": 0.0058, "step": 44220 }, { "grad_norm": 0.10425909608602524, "learning_rate": 6.350596701963166e-05, "loss": 0.0092, "step": 44230 }, { "grad_norm": 0.08026149123907089, "learning_rate": 6.349004623007419e-05, "loss": 0.005, "step": 44240 }, { "grad_norm": 0.059312108904123306, "learning_rate": 6.347412396526555e-05, "loss": 0.0063, "step": 44250 }, { "grad_norm": 0.08008641749620438, "learning_rate": 6.345820022694696e-05, "loss": 0.007, "step": 44260 }, { "grad_norm": 0.13250118494033813, "learning_rate": 6.344227501685984e-05, "loss": 0.0088, "step": 44270 }, { "grad_norm": 0.09583503752946854, "learning_rate": 6.342634833674572e-05, "loss": 0.0072, "step": 44280 }, { "grad_norm": 0.09548242390155792, "learning_rate": 6.341042018834635e-05, "loss": 0.0073, "step": 44290 }, { "grad_norm": 0.08732694387435913, "learning_rate": 6.339449057340359e-05, "loss": 0.0066, "step": 44300 }, { "grad_norm": 0.09738223254680634, "learning_rate": 6.337855949365945e-05, "loss": 0.006, "step": 44310 }, { "grad_norm": 0.09697266668081284, "learning_rate": 6.336262695085619e-05, "loss": 0.008, "step": 44320 }, { "grad_norm": 0.10198578238487244, "learning_rate": 6.334669294673612e-05, "loss": 0.0075, "step": 44330 }, { "grad_norm": 0.08429153263568878, "learning_rate": 6.333075748304179e-05, "loss": 0.007, "step": 44340 }, { "grad_norm": 0.12470144778490067, "learning_rate": 6.331482056151585e-05, "loss": 0.0051, "step": 44350 }, { "grad_norm": 0.09368941932916641, "learning_rate": 6.329888218390117e-05, "loss": 0.0062, "step": 44360 }, { "grad_norm": 0.08486538380384445, "learning_rate": 6.328294235194072e-05, "loss": 0.0054, "step": 44370 }, { "grad_norm": 0.06740227341651917, "learning_rate": 6.326700106737765e-05, "loss": 0.0071, "step": 44380 }, { "grad_norm": 0.15455113351345062, "learning_rate": 6.325105833195531e-05, "loss": 0.009, "step": 44390 }, { "grad_norm": 0.14759117364883423, "learning_rate": 6.323511414741715e-05, "loss": 0.0061, "step": 44400 }, { "grad_norm": 0.12532898783683777, "learning_rate": 6.321916851550678e-05, "loss": 0.0078, "step": 44410 }, { "grad_norm": 0.07130972295999527, "learning_rate": 6.320322143796806e-05, "loss": 0.0078, "step": 44420 }, { "grad_norm": 0.07890252023935318, "learning_rate": 6.318727291654488e-05, "loss": 0.0065, "step": 44430 }, { "grad_norm": 0.06290890276432037, "learning_rate": 6.317132295298134e-05, "loss": 0.0057, "step": 44440 }, { "grad_norm": 0.08580348640680313, "learning_rate": 6.315537154902173e-05, "loss": 0.0066, "step": 44450 }, { "grad_norm": 0.07077353447675705, "learning_rate": 6.313941870641048e-05, "loss": 0.0062, "step": 44460 }, { "grad_norm": 0.06067853048443794, "learning_rate": 6.312346442689214e-05, "loss": 0.008, "step": 44470 }, { "grad_norm": 0.06232282519340515, "learning_rate": 6.310750871221147e-05, "loss": 0.0068, "step": 44480 }, { "grad_norm": 0.1024649515748024, "learning_rate": 6.309155156411335e-05, "loss": 0.0049, "step": 44490 }, { "grad_norm": 0.0672529861330986, "learning_rate": 6.307559298434284e-05, "loss": 0.0059, "step": 44500 }, { "grad_norm": 0.0783160999417305, "learning_rate": 6.305963297464512e-05, "loss": 0.0041, "step": 44510 }, { "grad_norm": 0.06673123687505722, "learning_rate": 6.304367153676561e-05, "loss": 0.005, "step": 44520 }, { "grad_norm": 0.0782991349697113, "learning_rate": 6.302770867244978e-05, "loss": 0.0044, "step": 44530 }, { "grad_norm": 0.13417752087116241, "learning_rate": 6.301174438344328e-05, "loss": 0.0075, "step": 44540 }, { "grad_norm": 0.05680188536643982, "learning_rate": 6.299577867149202e-05, "loss": 0.005, "step": 44550 }, { "grad_norm": 0.09912358969449997, "learning_rate": 6.297981153834192e-05, "loss": 0.0058, "step": 44560 }, { "grad_norm": 0.12592725455760956, "learning_rate": 6.296384298573916e-05, "loss": 0.0075, "step": 44570 }, { "grad_norm": 0.10092219710350037, "learning_rate": 6.294787301543001e-05, "loss": 0.0093, "step": 44580 }, { "grad_norm": 0.08825309574604034, "learning_rate": 6.293190162916095e-05, "loss": 0.0075, "step": 44590 }, { "grad_norm": 0.06756780296564102, "learning_rate": 6.291592882867855e-05, "loss": 0.006, "step": 44600 }, { "grad_norm": 0.08580698817968369, "learning_rate": 6.28999546157296e-05, "loss": 0.0069, "step": 44610 }, { "grad_norm": 0.09743478149175644, "learning_rate": 6.288397899206102e-05, "loss": 0.0083, "step": 44620 }, { "grad_norm": 0.08561293035745621, "learning_rate": 6.286800195941984e-05, "loss": 0.0062, "step": 44630 }, { "grad_norm": 0.07444513589143753, "learning_rate": 6.285202351955334e-05, "loss": 0.0056, "step": 44640 }, { "grad_norm": 0.08774098753929138, "learning_rate": 6.283604367420887e-05, "loss": 0.007, "step": 44650 }, { "grad_norm": 0.10669650882482529, "learning_rate": 6.282006242513394e-05, "loss": 0.0072, "step": 44660 }, { "grad_norm": 0.08666540682315826, "learning_rate": 6.280407977407628e-05, "loss": 0.0062, "step": 44670 }, { "grad_norm": 0.08227100968360901, "learning_rate": 6.27880957227837e-05, "loss": 0.0072, "step": 44680 }, { "grad_norm": 0.06758656352758408, "learning_rate": 6.27721102730042e-05, "loss": 0.0057, "step": 44690 }, { "grad_norm": 0.1263384222984314, "learning_rate": 6.275612342648592e-05, "loss": 0.0074, "step": 44700 }, { "grad_norm": 0.09987716376781464, "learning_rate": 6.274013518497716e-05, "loss": 0.0071, "step": 44710 }, { "grad_norm": 0.08378023654222488, "learning_rate": 6.272414555022636e-05, "loss": 0.006, "step": 44720 }, { "grad_norm": 0.14915010333061218, "learning_rate": 6.270815452398215e-05, "loss": 0.0079, "step": 44730 }, { "grad_norm": 0.08618690818548203, "learning_rate": 6.269216210799326e-05, "loss": 0.0076, "step": 44740 }, { "grad_norm": 0.14840823411941528, "learning_rate": 6.26761683040086e-05, "loss": 0.0092, "step": 44750 }, { "grad_norm": 0.09586667269468307, "learning_rate": 6.266017311377723e-05, "loss": 0.0056, "step": 44760 }, { "grad_norm": 0.08722354471683502, "learning_rate": 6.264417653904839e-05, "loss": 0.0054, "step": 44770 }, { "grad_norm": 0.1321067363023758, "learning_rate": 6.262817858157139e-05, "loss": 0.0095, "step": 44780 }, { "grad_norm": 0.1089453175663948, "learning_rate": 6.261217924309576e-05, "loss": 0.0055, "step": 44790 }, { "grad_norm": 0.10840824991464615, "learning_rate": 6.259617852537118e-05, "loss": 0.0075, "step": 44800 }, { "grad_norm": 0.08892355114221573, "learning_rate": 6.258017643014747e-05, "loss": 0.005, "step": 44810 }, { "grad_norm": 0.1059589684009552, "learning_rate": 6.256417295917456e-05, "loss": 0.0084, "step": 44820 }, { "grad_norm": 0.07791253924369812, "learning_rate": 6.254816811420258e-05, "loss": 0.0088, "step": 44830 }, { "grad_norm": 0.14869360625743866, "learning_rate": 6.253216189698183e-05, "loss": 0.0077, "step": 44840 }, { "grad_norm": 0.07696865499019623, "learning_rate": 6.251615430926267e-05, "loss": 0.0064, "step": 44850 }, { "grad_norm": 0.11575386673212051, "learning_rate": 6.25001453527957e-05, "loss": 0.0088, "step": 44860 }, { "grad_norm": 0.0908004567027092, "learning_rate": 6.248413502933164e-05, "loss": 0.0079, "step": 44870 }, { "grad_norm": 0.1146695539355278, "learning_rate": 6.246812334062133e-05, "loss": 0.0076, "step": 44880 }, { "grad_norm": 0.09914080053567886, "learning_rate": 6.245211028841579e-05, "loss": 0.0056, "step": 44890 }, { "grad_norm": 0.08457235246896744, "learning_rate": 6.24360958744662e-05, "loss": 0.0065, "step": 44900 }, { "grad_norm": 0.08150917291641235, "learning_rate": 6.242008010052387e-05, "loss": 0.007, "step": 44910 }, { "grad_norm": 0.07210380584001541, "learning_rate": 6.240406296834024e-05, "loss": 0.0054, "step": 44920 }, { "grad_norm": 0.07906410843133926, "learning_rate": 6.238804447966694e-05, "loss": 0.0108, "step": 44930 }, { "grad_norm": 0.08853424340486526, "learning_rate": 6.237202463625573e-05, "loss": 0.0061, "step": 44940 }, { "grad_norm": 0.09411723166704178, "learning_rate": 6.235600343985848e-05, "loss": 0.0054, "step": 44950 }, { "grad_norm": 0.09434688091278076, "learning_rate": 6.233998089222729e-05, "loss": 0.0055, "step": 44960 }, { "grad_norm": 0.13121682405471802, "learning_rate": 6.232395699511433e-05, "loss": 0.0056, "step": 44970 }, { "grad_norm": 0.1109648197889328, "learning_rate": 6.230793175027199e-05, "loss": 0.006, "step": 44980 }, { "grad_norm": 0.07541906833648682, "learning_rate": 6.22919051594527e-05, "loss": 0.0089, "step": 44990 }, { "grad_norm": 0.06921981275081635, "learning_rate": 6.227587722440917e-05, "loss": 0.0057, "step": 45000 }, { "grad_norm": 0.13254229724407196, "learning_rate": 6.225984794689414e-05, "loss": 0.0077, "step": 45010 }, { "grad_norm": 0.12137173116207123, "learning_rate": 6.22438173286606e-05, "loss": 0.0067, "step": 45020 }, { "grad_norm": 0.08073871582746506, "learning_rate": 6.222778537146159e-05, "loss": 0.0056, "step": 45030 }, { "grad_norm": 0.11561062186956406, "learning_rate": 6.221175207705037e-05, "loss": 0.0085, "step": 45040 }, { "grad_norm": 0.08951088786125183, "learning_rate": 6.219571744718027e-05, "loss": 0.0068, "step": 45050 }, { "grad_norm": 0.13016796112060547, "learning_rate": 6.217968148360487e-05, "loss": 0.007, "step": 45060 }, { "grad_norm": 0.07171022891998291, "learning_rate": 6.216364418807782e-05, "loss": 0.0054, "step": 45070 }, { "grad_norm": 0.059934765100479126, "learning_rate": 6.214760556235292e-05, "loss": 0.0068, "step": 45080 }, { "grad_norm": 0.133753702044487, "learning_rate": 6.213156560818414e-05, "loss": 0.0058, "step": 45090 }, { "grad_norm": 0.07850280404090881, "learning_rate": 6.211552432732559e-05, "loss": 0.0074, "step": 45100 }, { "grad_norm": 0.11343055963516235, "learning_rate": 6.20994817215315e-05, "loss": 0.0068, "step": 45110 }, { "grad_norm": 0.07840801030397415, "learning_rate": 6.208343779255628e-05, "loss": 0.0068, "step": 45120 }, { "grad_norm": 0.1608109176158905, "learning_rate": 6.206739254215449e-05, "loss": 0.0065, "step": 45130 }, { "grad_norm": 0.10819541662931442, "learning_rate": 6.205134597208077e-05, "loss": 0.0076, "step": 45140 }, { "grad_norm": 0.154083713889122, "learning_rate": 6.203529808409e-05, "loss": 0.0074, "step": 45150 }, { "grad_norm": 0.09361513704061508, "learning_rate": 6.201924887993712e-05, "loss": 0.0074, "step": 45160 }, { "grad_norm": 0.10704796761274338, "learning_rate": 6.200319836137725e-05, "loss": 0.0054, "step": 45170 }, { "grad_norm": 0.08445041626691818, "learning_rate": 6.198714653016565e-05, "loss": 0.008, "step": 45180 }, { "grad_norm": 0.07388041913509369, "learning_rate": 6.197109338805774e-05, "loss": 0.0066, "step": 45190 }, { "grad_norm": 0.1028764545917511, "learning_rate": 6.195503893680903e-05, "loss": 0.0077, "step": 45200 }, { "grad_norm": 0.08034757524728775, "learning_rate": 6.193898317817524e-05, "loss": 0.0051, "step": 45210 }, { "grad_norm": 0.06262638419866562, "learning_rate": 6.192292611391221e-05, "loss": 0.0049, "step": 45220 }, { "grad_norm": 0.07028037309646606, "learning_rate": 6.190686774577591e-05, "loss": 0.005, "step": 45230 }, { "grad_norm": 0.05341867357492447, "learning_rate": 6.189080807552245e-05, "loss": 0.0056, "step": 45240 }, { "grad_norm": 0.08028637617826462, "learning_rate": 6.187474710490809e-05, "loss": 0.009, "step": 45250 }, { "grad_norm": 0.1771429032087326, "learning_rate": 6.185868483568926e-05, "loss": 0.0083, "step": 45260 }, { "grad_norm": 0.0829041600227356, "learning_rate": 6.184262126962245e-05, "loss": 0.0087, "step": 45270 }, { "grad_norm": 0.1444445252418518, "learning_rate": 6.182655640846442e-05, "loss": 0.0065, "step": 45280 }, { "grad_norm": 0.055077217519283295, "learning_rate": 6.181049025397196e-05, "loss": 0.0065, "step": 45290 }, { "grad_norm": 0.07659221440553665, "learning_rate": 6.179442280790202e-05, "loss": 0.007, "step": 45300 }, { "grad_norm": 0.0668790340423584, "learning_rate": 6.177835407201174e-05, "loss": 0.0047, "step": 45310 }, { "grad_norm": 0.23226414620876312, "learning_rate": 6.176228404805839e-05, "loss": 0.0054, "step": 45320 }, { "grad_norm": 0.06964747607707977, "learning_rate": 6.174621273779932e-05, "loss": 0.0062, "step": 45330 }, { "grad_norm": 0.1716376692056656, "learning_rate": 6.17301401429921e-05, "loss": 0.0066, "step": 45340 }, { "grad_norm": 0.13686823844909668, "learning_rate": 6.171406626539436e-05, "loss": 0.0059, "step": 45350 }, { "grad_norm": 0.0978114977478981, "learning_rate": 6.169799110676398e-05, "loss": 0.0056, "step": 45360 }, { "grad_norm": 0.11439942568540573, "learning_rate": 6.168191466885885e-05, "loss": 0.0076, "step": 45370 }, { "grad_norm": 0.0831083208322525, "learning_rate": 6.166583695343711e-05, "loss": 0.008, "step": 45380 }, { "grad_norm": 0.10683640837669373, "learning_rate": 6.164975796225698e-05, "loss": 0.007, "step": 45390 }, { "grad_norm": 0.10166427493095398, "learning_rate": 6.163367769707683e-05, "loss": 0.0102, "step": 45400 }, { "grad_norm": 0.09713164716959, "learning_rate": 6.161759615965519e-05, "loss": 0.0077, "step": 45410 }, { "grad_norm": 0.12567450106143951, "learning_rate": 6.16015133517507e-05, "loss": 0.0067, "step": 45420 }, { "grad_norm": 0.08491545915603638, "learning_rate": 6.158542927512214e-05, "loss": 0.0052, "step": 45430 }, { "grad_norm": 0.10116135329008102, "learning_rate": 6.156934393152846e-05, "loss": 0.0065, "step": 45440 }, { "grad_norm": 0.08427830785512924, "learning_rate": 6.15532573227287e-05, "loss": 0.0104, "step": 45450 }, { "grad_norm": 0.09264161437749863, "learning_rate": 6.153716945048212e-05, "loss": 0.0057, "step": 45460 }, { "grad_norm": 0.12013769149780273, "learning_rate": 6.152108031654802e-05, "loss": 0.0074, "step": 45470 }, { "grad_norm": 0.0766659528017044, "learning_rate": 6.15049899226859e-05, "loss": 0.0061, "step": 45480 }, { "grad_norm": 0.261549711227417, "learning_rate": 6.148889827065537e-05, "loss": 0.0089, "step": 45490 }, { "grad_norm": 0.11873612552881241, "learning_rate": 6.147280536221622e-05, "loss": 0.0059, "step": 45500 }, { "grad_norm": 0.09150408208370209, "learning_rate": 6.145671119912832e-05, "loss": 0.0061, "step": 45510 }, { "grad_norm": 0.19327664375305176, "learning_rate": 6.144061578315169e-05, "loss": 0.0095, "step": 45520 }, { "grad_norm": 0.1367950141429901, "learning_rate": 6.142451911604654e-05, "loss": 0.0059, "step": 45530 }, { "grad_norm": 0.08751943707466125, "learning_rate": 6.140842119957315e-05, "loss": 0.0074, "step": 45540 }, { "grad_norm": 0.1065751239657402, "learning_rate": 6.139232203549197e-05, "loss": 0.0059, "step": 45550 }, { "grad_norm": 0.0772872120141983, "learning_rate": 6.137622162556357e-05, "loss": 0.0067, "step": 45560 }, { "grad_norm": 0.07182464748620987, "learning_rate": 6.13601199715487e-05, "loss": 0.0069, "step": 45570 }, { "grad_norm": 0.0608583465218544, "learning_rate": 6.134401707520816e-05, "loss": 0.0054, "step": 45580 }, { "grad_norm": 0.0856604278087616, "learning_rate": 6.132791293830298e-05, "loss": 0.0053, "step": 45590 }, { "grad_norm": 0.07700347155332565, "learning_rate": 6.131180756259428e-05, "loss": 0.0076, "step": 45600 }, { "grad_norm": 0.08327271044254303, "learning_rate": 6.129570094984331e-05, "loss": 0.0054, "step": 45610 }, { "grad_norm": 0.0626278817653656, "learning_rate": 6.127959310181145e-05, "loss": 0.0071, "step": 45620 }, { "grad_norm": 0.08151610195636749, "learning_rate": 6.126348402026026e-05, "loss": 0.0049, "step": 45630 }, { "grad_norm": 0.07992425560951233, "learning_rate": 6.12473737069514e-05, "loss": 0.0064, "step": 45640 }, { "grad_norm": 0.06634486466646194, "learning_rate": 6.123126216364665e-05, "loss": 0.005, "step": 45650 }, { "grad_norm": 0.04002508521080017, "learning_rate": 6.121514939210797e-05, "loss": 0.0044, "step": 45660 }, { "grad_norm": 0.09421226382255554, "learning_rate": 6.119903539409741e-05, "loss": 0.0061, "step": 45670 }, { "grad_norm": 0.06156467646360397, "learning_rate": 6.118292017137716e-05, "loss": 0.0064, "step": 45680 }, { "grad_norm": 0.0957682803273201, "learning_rate": 6.116680372570959e-05, "loss": 0.0068, "step": 45690 }, { "grad_norm": 0.07414442300796509, "learning_rate": 6.115068605885713e-05, "loss": 0.0055, "step": 45700 }, { "grad_norm": 0.09023141860961914, "learning_rate": 6.113456717258243e-05, "loss": 0.0087, "step": 45710 }, { "grad_norm": 0.08570981025695801, "learning_rate": 6.11184470686482e-05, "loss": 0.0053, "step": 45720 }, { "grad_norm": 0.07548385858535767, "learning_rate": 6.110232574881731e-05, "loss": 0.0068, "step": 45730 }, { "grad_norm": 0.1418255865573883, "learning_rate": 6.108620321485277e-05, "loss": 0.0077, "step": 45740 }, { "grad_norm": 0.07043951749801636, "learning_rate": 6.107007946851773e-05, "loss": 0.0098, "step": 45750 }, { "grad_norm": 0.10258113592863083, "learning_rate": 6.105395451157542e-05, "loss": 0.007, "step": 45760 }, { "grad_norm": 0.10294012725353241, "learning_rate": 6.103782834578928e-05, "loss": 0.0084, "step": 45770 }, { "grad_norm": 0.10858486592769623, "learning_rate": 6.102170097292281e-05, "loss": 0.0078, "step": 45780 }, { "grad_norm": 0.09067674726247787, "learning_rate": 6.1005572394739716e-05, "loss": 0.007, "step": 45790 }, { "grad_norm": 0.12142910808324814, "learning_rate": 6.0989442613003765e-05, "loss": 0.0097, "step": 45800 }, { "grad_norm": 0.0851183608174324, "learning_rate": 6.097331162947888e-05, "loss": 0.0066, "step": 45810 }, { "grad_norm": 0.12470422685146332, "learning_rate": 6.095717944592914e-05, "loss": 0.0058, "step": 45820 }, { "grad_norm": 0.08413080126047134, "learning_rate": 6.094104606411873e-05, "loss": 0.0058, "step": 45830 }, { "grad_norm": 0.12078917026519775, "learning_rate": 6.0924911485811966e-05, "loss": 0.0065, "step": 45840 }, { "grad_norm": 0.1797839105129242, "learning_rate": 6.090877571277331e-05, "loss": 0.0077, "step": 45850 }, { "grad_norm": 0.0878278985619545, "learning_rate": 6.089263874676736e-05, "loss": 0.0068, "step": 45860 }, { "grad_norm": 0.0781106948852539, "learning_rate": 6.0876500589558796e-05, "loss": 0.0089, "step": 45870 }, { "grad_norm": 0.11951173096895218, "learning_rate": 6.086036124291248e-05, "loss": 0.0072, "step": 45880 }, { "grad_norm": 0.07960835099220276, "learning_rate": 6.084422070859339e-05, "loss": 0.0051, "step": 45890 }, { "grad_norm": 0.0521552674472332, "learning_rate": 6.082807898836663e-05, "loss": 0.005, "step": 45900 }, { "grad_norm": 0.05584930628538132, "learning_rate": 6.081193608399742e-05, "loss": 0.0046, "step": 45910 }, { "grad_norm": 0.08651933819055557, "learning_rate": 6.0795791997251164e-05, "loss": 0.0072, "step": 45920 }, { "grad_norm": 0.11297531425952911, "learning_rate": 6.0779646729893294e-05, "loss": 0.0064, "step": 45930 }, { "grad_norm": 0.08288490027189255, "learning_rate": 6.0763500283689476e-05, "loss": 0.0052, "step": 45940 }, { "grad_norm": 0.07738248258829117, "learning_rate": 6.0747352660405455e-05, "loss": 0.0055, "step": 45950 }, { "grad_norm": 0.07218014448881149, "learning_rate": 6.073120386180709e-05, "loss": 0.0056, "step": 45960 }, { "grad_norm": 0.04692167416214943, "learning_rate": 6.0715053889660425e-05, "loss": 0.0048, "step": 45970 }, { "grad_norm": 0.07399749755859375, "learning_rate": 6.069890274573157e-05, "loss": 0.0045, "step": 45980 }, { "grad_norm": 0.1587076485157013, "learning_rate": 6.068275043178679e-05, "loss": 0.0079, "step": 45990 }, { "grad_norm": 0.1018507331609726, "learning_rate": 6.066659694959248e-05, "loss": 0.0094, "step": 46000 }, { "grad_norm": 0.07331421226263046, "learning_rate": 6.0650442300915176e-05, "loss": 0.0063, "step": 46010 }, { "grad_norm": 0.06829569488763809, "learning_rate": 6.063428648752152e-05, "loss": 0.0052, "step": 46020 }, { "grad_norm": 0.118865467607975, "learning_rate": 6.0618129511178266e-05, "loss": 0.0051, "step": 46030 }, { "grad_norm": 0.09925846755504608, "learning_rate": 6.060197137365234e-05, "loss": 0.0082, "step": 46040 }, { "grad_norm": 0.09819240123033524, "learning_rate": 6.058581207671077e-05, "loss": 0.009, "step": 46050 }, { "grad_norm": 0.09267593920230865, "learning_rate": 6.056965162212072e-05, "loss": 0.0071, "step": 46060 }, { "grad_norm": 0.08113721013069153, "learning_rate": 6.055349001164943e-05, "loss": 0.006, "step": 46070 }, { "grad_norm": 0.15666338801383972, "learning_rate": 6.0537327247064347e-05, "loss": 0.0084, "step": 46080 }, { "grad_norm": 0.09290100634098053, "learning_rate": 6.0521163330133014e-05, "loss": 0.0052, "step": 46090 }, { "grad_norm": 0.07160532474517822, "learning_rate": 6.050499826262306e-05, "loss": 0.0057, "step": 46100 }, { "grad_norm": 0.10895437002182007, "learning_rate": 6.0488832046302294e-05, "loss": 0.0063, "step": 46110 }, { "grad_norm": 0.10158085078001022, "learning_rate": 6.0472664682938626e-05, "loss": 0.0102, "step": 46120 }, { "grad_norm": 0.07661712169647217, "learning_rate": 6.045649617430009e-05, "loss": 0.008, "step": 46130 }, { "grad_norm": 0.09860408306121826, "learning_rate": 6.0440326522154866e-05, "loss": 0.0073, "step": 46140 }, { "grad_norm": 0.0764465257525444, "learning_rate": 6.0424155728271224e-05, "loss": 0.0052, "step": 46150 }, { "grad_norm": 0.05737879127264023, "learning_rate": 6.040798379441758e-05, "loss": 0.0059, "step": 46160 }, { "grad_norm": 0.1242496520280838, "learning_rate": 6.0391810722362485e-05, "loss": 0.0061, "step": 46170 }, { "grad_norm": 0.06433933228254318, "learning_rate": 6.037563651387458e-05, "loss": 0.0058, "step": 46180 }, { "grad_norm": 0.09235336631536484, "learning_rate": 6.0359461170722666e-05, "loss": 0.0079, "step": 46190 }, { "grad_norm": 0.07755031436681747, "learning_rate": 6.034328469467566e-05, "loss": 0.0065, "step": 46200 }, { "grad_norm": 0.07437947392463684, "learning_rate": 6.0327107087502596e-05, "loss": 0.0062, "step": 46210 }, { "grad_norm": 0.06854142993688583, "learning_rate": 6.031092835097262e-05, "loss": 0.0049, "step": 46220 }, { "grad_norm": 0.08395599573850632, "learning_rate": 6.0294748486855024e-05, "loss": 0.0078, "step": 46230 }, { "grad_norm": 0.09422638267278671, "learning_rate": 6.0278567496919216e-05, "loss": 0.0055, "step": 46240 }, { "grad_norm": 0.07025027275085449, "learning_rate": 6.026238538293472e-05, "loss": 0.0053, "step": 46250 }, { "grad_norm": 0.16962410509586334, "learning_rate": 6.024620214667118e-05, "loss": 0.0048, "step": 46260 }, { "grad_norm": 0.11565910279750824, "learning_rate": 6.0230017789898384e-05, "loss": 0.0056, "step": 46270 }, { "grad_norm": 0.08184047043323517, "learning_rate": 6.021383231438622e-05, "loss": 0.0045, "step": 46280 }, { "grad_norm": 0.1175023540854454, "learning_rate": 6.0197645721904704e-05, "loss": 0.0052, "step": 46290 }, { "grad_norm": 0.13245925307273865, "learning_rate": 6.0181458014224e-05, "loss": 0.006, "step": 46300 }, { "grad_norm": 0.10802864283323288, "learning_rate": 6.016526919311435e-05, "loss": 0.0072, "step": 46310 }, { "grad_norm": 0.07964169234037399, "learning_rate": 6.014907926034613e-05, "loss": 0.0062, "step": 46320 }, { "grad_norm": 0.07365813851356506, "learning_rate": 6.0132888217689875e-05, "loss": 0.0068, "step": 46330 }, { "grad_norm": 0.07778025418519974, "learning_rate": 6.01166960669162e-05, "loss": 0.008, "step": 46340 }, { "grad_norm": 0.061063069850206375, "learning_rate": 6.0100502809795845e-05, "loss": 0.0076, "step": 46350 }, { "grad_norm": 0.08824208378791809, "learning_rate": 6.00843084480997e-05, "loss": 0.0065, "step": 46360 }, { "grad_norm": 0.09073241055011749, "learning_rate": 6.0068112983598736e-05, "loss": 0.0072, "step": 46370 }, { "grad_norm": 0.08748552203178406, "learning_rate": 6.005191641806407e-05, "loss": 0.0064, "step": 46380 }, { "grad_norm": 0.05320742726325989, "learning_rate": 6.003571875326694e-05, "loss": 0.0066, "step": 46390 }, { "grad_norm": 0.08783803135156631, "learning_rate": 6.00195199909787e-05, "loss": 0.0074, "step": 46400 }, { "grad_norm": 0.09211467206478119, "learning_rate": 6.000332013297082e-05, "loss": 0.0082, "step": 46410 }, { "grad_norm": 0.0719153881072998, "learning_rate": 5.998711918101487e-05, "loss": 0.0077, "step": 46420 }, { "grad_norm": 0.10880759358406067, "learning_rate": 5.997091713688261e-05, "loss": 0.0086, "step": 46430 }, { "grad_norm": 0.07504887133836746, "learning_rate": 5.9954714002345836e-05, "loss": 0.007, "step": 46440 }, { "grad_norm": 0.08304451406002045, "learning_rate": 5.993850977917649e-05, "loss": 0.0069, "step": 46450 }, { "grad_norm": 0.07093758881092072, "learning_rate": 5.992230446914667e-05, "loss": 0.0065, "step": 46460 }, { "grad_norm": 0.12062501162290573, "learning_rate": 5.990609807402855e-05, "loss": 0.0057, "step": 46470 }, { "grad_norm": 0.060870349407196045, "learning_rate": 5.988989059559443e-05, "loss": 0.0073, "step": 46480 }, { "grad_norm": 0.05628862604498863, "learning_rate": 5.987368203561675e-05, "loss": 0.0052, "step": 46490 }, { "grad_norm": 0.09601637721061707, "learning_rate": 5.9857472395868055e-05, "loss": 0.0052, "step": 46500 }, { "grad_norm": 0.12463709712028503, "learning_rate": 5.9841261678120983e-05, "loss": 0.0085, "step": 46510 }, { "grad_norm": 0.1430073082447052, "learning_rate": 5.982504988414834e-05, "loss": 0.0073, "step": 46520 }, { "grad_norm": 0.09449533373117447, "learning_rate": 5.9808837015723015e-05, "loss": 0.0065, "step": 46530 }, { "grad_norm": 0.10154783725738525, "learning_rate": 5.9792623074618016e-05, "loss": 0.0063, "step": 46540 }, { "grad_norm": 0.10137458145618439, "learning_rate": 5.977640806260648e-05, "loss": 0.0076, "step": 46550 }, { "grad_norm": 0.14574149250984192, "learning_rate": 5.976019198146166e-05, "loss": 0.007, "step": 46560 }, { "grad_norm": 0.05952244624495506, "learning_rate": 5.974397483295692e-05, "loss": 0.0074, "step": 46570 }, { "grad_norm": 0.09787765145301819, "learning_rate": 5.972775661886575e-05, "loss": 0.008, "step": 46580 }, { "grad_norm": 0.12642790377140045, "learning_rate": 5.971153734096173e-05, "loss": 0.0072, "step": 46590 }, { "grad_norm": 0.10735710710287094, "learning_rate": 5.969531700101859e-05, "loss": 0.0064, "step": 46600 }, { "grad_norm": 0.08777185529470444, "learning_rate": 5.9679095600810155e-05, "loss": 0.006, "step": 46610 }, { "grad_norm": 0.08882483094930649, "learning_rate": 5.9662873142110384e-05, "loss": 0.007, "step": 46620 }, { "grad_norm": 0.0828150063753128, "learning_rate": 5.964664962669333e-05, "loss": 0.0063, "step": 46630 }, { "grad_norm": 0.11660727113485336, "learning_rate": 5.9630425056333186e-05, "loss": 0.0052, "step": 46640 }, { "grad_norm": 0.07033193111419678, "learning_rate": 5.961419943280422e-05, "loss": 0.0069, "step": 46650 }, { "grad_norm": 0.10612161457538605, "learning_rate": 5.959797275788087e-05, "loss": 0.0061, "step": 46660 }, { "grad_norm": 0.11510016769170761, "learning_rate": 5.958174503333765e-05, "loss": 0.0074, "step": 46670 }, { "grad_norm": 0.07336162030696869, "learning_rate": 5.9565516260949195e-05, "loss": 0.0058, "step": 46680 }, { "grad_norm": 0.10888271033763885, "learning_rate": 5.9549286442490273e-05, "loss": 0.0055, "step": 46690 }, { "grad_norm": 0.06773297488689423, "learning_rate": 5.953305557973572e-05, "loss": 0.0079, "step": 46700 }, { "grad_norm": 0.06410191208124161, "learning_rate": 5.9516823674460564e-05, "loss": 0.0051, "step": 46710 }, { "grad_norm": 0.13587938249111176, "learning_rate": 5.9500590728439875e-05, "loss": 0.0055, "step": 46720 }, { "grad_norm": 0.06726794689893723, "learning_rate": 5.948435674344886e-05, "loss": 0.0078, "step": 46730 }, { "grad_norm": 0.1963392049074173, "learning_rate": 5.946812172126285e-05, "loss": 0.0079, "step": 46740 }, { "grad_norm": 0.08573063462972641, "learning_rate": 5.945188566365729e-05, "loss": 0.0067, "step": 46750 }, { "grad_norm": 0.14947551488876343, "learning_rate": 5.943564857240773e-05, "loss": 0.0055, "step": 46760 }, { "grad_norm": 0.10045628994703293, "learning_rate": 5.941941044928981e-05, "loss": 0.0063, "step": 46770 }, { "grad_norm": 0.07904680073261261, "learning_rate": 5.940317129607935e-05, "loss": 0.0047, "step": 46780 }, { "grad_norm": 0.16657143831253052, "learning_rate": 5.9386931114552204e-05, "loss": 0.0059, "step": 46790 }, { "grad_norm": 0.11476226150989532, "learning_rate": 5.937068990648438e-05, "loss": 0.0081, "step": 46800 }, { "grad_norm": 0.12261902540922165, "learning_rate": 5.935444767365199e-05, "loss": 0.0063, "step": 46810 }, { "grad_norm": 0.1026865690946579, "learning_rate": 5.933820441783129e-05, "loss": 0.0062, "step": 46820 }, { "grad_norm": 0.1271812468767166, "learning_rate": 5.932196014079857e-05, "loss": 0.006, "step": 46830 }, { "grad_norm": 0.11674979329109192, "learning_rate": 5.930571484433032e-05, "loss": 0.0053, "step": 46840 }, { "grad_norm": 0.07209952175617218, "learning_rate": 5.928946853020309e-05, "loss": 0.004, "step": 46850 }, { "grad_norm": 0.16725295782089233, "learning_rate": 5.927322120019352e-05, "loss": 0.0068, "step": 46860 }, { "grad_norm": 0.09176727384328842, "learning_rate": 5.925697285607844e-05, "loss": 0.0062, "step": 46870 }, { "grad_norm": 0.12673553824424744, "learning_rate": 5.924072349963472e-05, "loss": 0.0079, "step": 46880 }, { "grad_norm": 0.10150653123855591, "learning_rate": 5.922447313263938e-05, "loss": 0.0047, "step": 46890 }, { "grad_norm": 0.09405361860990524, "learning_rate": 5.920822175686952e-05, "loss": 0.0053, "step": 46900 }, { "grad_norm": 0.10743136703968048, "learning_rate": 5.919196937410239e-05, "loss": 0.0056, "step": 46910 }, { "grad_norm": 0.08095792680978775, "learning_rate": 5.917571598611529e-05, "loss": 0.0051, "step": 46920 }, { "grad_norm": 0.090208038687706, "learning_rate": 5.9159461594685686e-05, "loss": 0.0076, "step": 46930 }, { "grad_norm": 0.1261530965566635, "learning_rate": 5.914320620159114e-05, "loss": 0.0075, "step": 46940 }, { "grad_norm": 0.14753612875938416, "learning_rate": 5.912694980860931e-05, "loss": 0.0066, "step": 46950 }, { "grad_norm": 0.0885821133852005, "learning_rate": 5.9110692417517964e-05, "loss": 0.0059, "step": 46960 }, { "grad_norm": 0.09480953216552734, "learning_rate": 5.9094434030095e-05, "loss": 0.0055, "step": 46970 }, { "grad_norm": 0.11407492309808731, "learning_rate": 5.907817464811841e-05, "loss": 0.008, "step": 46980 }, { "grad_norm": 0.09112264961004257, "learning_rate": 5.906191427336627e-05, "loss": 0.005, "step": 46990 }, { "grad_norm": 0.10786859691143036, "learning_rate": 5.9045652907616835e-05, "loss": 0.0054, "step": 47000 }, { "grad_norm": 0.07199642062187195, "learning_rate": 5.902939055264838e-05, "loss": 0.0056, "step": 47010 }, { "grad_norm": 0.0869862362742424, "learning_rate": 5.901312721023935e-05, "loss": 0.0057, "step": 47020 }, { "grad_norm": 0.07269550114870071, "learning_rate": 5.8996862882168294e-05, "loss": 0.0057, "step": 47030 }, { "grad_norm": 0.1336604207754135, "learning_rate": 5.8980597570213826e-05, "loss": 0.0098, "step": 47040 }, { "grad_norm": 0.11853491514921188, "learning_rate": 5.896433127615471e-05, "loss": 0.0064, "step": 47050 }, { "grad_norm": 0.08067891746759415, "learning_rate": 5.894806400176981e-05, "loss": 0.0052, "step": 47060 }, { "grad_norm": 0.08915679156780243, "learning_rate": 5.893179574883808e-05, "loss": 0.0074, "step": 47070 }, { "grad_norm": 0.07497237622737885, "learning_rate": 5.8915526519138585e-05, "loss": 0.0058, "step": 47080 }, { "grad_norm": 0.053433969616889954, "learning_rate": 5.889925631445053e-05, "loss": 0.0063, "step": 47090 }, { "grad_norm": 0.07941121608018875, "learning_rate": 5.888298513655318e-05, "loss": 0.0069, "step": 47100 }, { "grad_norm": 0.1194448471069336, "learning_rate": 5.886671298722592e-05, "loss": 0.0077, "step": 47110 }, { "grad_norm": 0.07872014492750168, "learning_rate": 5.885043986824828e-05, "loss": 0.0066, "step": 47120 }, { "grad_norm": 0.06641273945569992, "learning_rate": 5.883416578139982e-05, "loss": 0.0064, "step": 47130 }, { "grad_norm": 0.06347884237766266, "learning_rate": 5.881789072846029e-05, "loss": 0.0049, "step": 47140 }, { "grad_norm": 0.07107113301753998, "learning_rate": 5.8801614711209474e-05, "loss": 0.0078, "step": 47150 }, { "grad_norm": 0.08234510570764542, "learning_rate": 5.878533773142732e-05, "loss": 0.0061, "step": 47160 }, { "grad_norm": 0.08060149848461151, "learning_rate": 5.8769059790893834e-05, "loss": 0.005, "step": 47170 }, { "grad_norm": 0.08987937122583389, "learning_rate": 5.8752780891389145e-05, "loss": 0.0074, "step": 47180 }, { "grad_norm": 0.0823754221200943, "learning_rate": 5.873650103469351e-05, "loss": 0.0058, "step": 47190 }, { "grad_norm": 0.06295754015445709, "learning_rate": 5.872022022258725e-05, "loss": 0.0078, "step": 47200 }, { "grad_norm": 0.09736496955156326, "learning_rate": 5.8703938456850814e-05, "loss": 0.0078, "step": 47210 }, { "grad_norm": 0.07272882759571075, "learning_rate": 5.868765573926477e-05, "loss": 0.0079, "step": 47220 }, { "grad_norm": 0.09328076988458633, "learning_rate": 5.867137207160974e-05, "loss": 0.0063, "step": 47230 }, { "grad_norm": 0.0648377314209938, "learning_rate": 5.86550874556665e-05, "loss": 0.0071, "step": 47240 }, { "grad_norm": 0.08564937859773636, "learning_rate": 5.863880189321592e-05, "loss": 0.0058, "step": 47250 }, { "grad_norm": 0.08786025643348694, "learning_rate": 5.862251538603894e-05, "loss": 0.0064, "step": 47260 }, { "grad_norm": 0.14858770370483398, "learning_rate": 5.8606227935916656e-05, "loss": 0.0086, "step": 47270 }, { "grad_norm": 0.09718934446573257, "learning_rate": 5.858993954463021e-05, "loss": 0.0057, "step": 47280 }, { "grad_norm": 0.08034922182559967, "learning_rate": 5.8573650213960896e-05, "loss": 0.0073, "step": 47290 }, { "grad_norm": 0.10983504354953766, "learning_rate": 5.8557359945690084e-05, "loss": 0.0048, "step": 47300 }, { "grad_norm": 0.10002657771110535, "learning_rate": 5.8541068741599246e-05, "loss": 0.0059, "step": 47310 }, { "grad_norm": 0.1000237986445427, "learning_rate": 5.8524776603469985e-05, "loss": 0.0077, "step": 47320 }, { "grad_norm": 0.12603837251663208, "learning_rate": 5.850848353308397e-05, "loss": 0.0054, "step": 47330 }, { "grad_norm": 0.13045096397399902, "learning_rate": 5.849218953222297e-05, "loss": 0.005, "step": 47340 }, { "grad_norm": 0.16454178094863892, "learning_rate": 5.847589460266891e-05, "loss": 0.0099, "step": 47350 }, { "grad_norm": 0.08378157019615173, "learning_rate": 5.845959874620375e-05, "loss": 0.0087, "step": 47360 }, { "grad_norm": 0.08253505825996399, "learning_rate": 5.84433019646096e-05, "loss": 0.0064, "step": 47370 }, { "grad_norm": 0.05434223264455795, "learning_rate": 5.842700425966863e-05, "loss": 0.0059, "step": 47380 }, { "grad_norm": 0.09171811491250992, "learning_rate": 5.841070563316315e-05, "loss": 0.007, "step": 47390 }, { "grad_norm": 0.06922703236341476, "learning_rate": 5.839440608687554e-05, "loss": 0.0052, "step": 47400 }, { "grad_norm": 0.13007968664169312, "learning_rate": 5.837810562258831e-05, "loss": 0.0084, "step": 47410 }, { "grad_norm": 0.10375750809907913, "learning_rate": 5.836180424208405e-05, "loss": 0.0053, "step": 47420 }, { "grad_norm": 0.08274782449007034, "learning_rate": 5.834550194714543e-05, "loss": 0.0052, "step": 47430 }, { "grad_norm": 0.10283026844263077, "learning_rate": 5.832919873955528e-05, "loss": 0.0058, "step": 47440 }, { "grad_norm": 0.07381217181682587, "learning_rate": 5.831289462109648e-05, "loss": 0.0058, "step": 47450 }, { "grad_norm": 0.14860087633132935, "learning_rate": 5.829658959355202e-05, "loss": 0.0083, "step": 47460 }, { "grad_norm": 0.15576069056987762, "learning_rate": 5.8280283658704995e-05, "loss": 0.0048, "step": 47470 }, { "grad_norm": 0.08366043120622635, "learning_rate": 5.826397681833859e-05, "loss": 0.0087, "step": 47480 }, { "grad_norm": 0.06661222875118256, "learning_rate": 5.824766907423612e-05, "loss": 0.0054, "step": 47490 }, { "grad_norm": 0.07422739267349243, "learning_rate": 5.8231360428180945e-05, "loss": 0.005, "step": 47500 }, { "grad_norm": 0.11851319670677185, "learning_rate": 5.821505088195658e-05, "loss": 0.0075, "step": 47510 }, { "grad_norm": 0.09044849872589111, "learning_rate": 5.819874043734661e-05, "loss": 0.0055, "step": 47520 }, { "grad_norm": 0.06199093163013458, "learning_rate": 5.8182429096134695e-05, "loss": 0.0052, "step": 47530 }, { "grad_norm": 0.17390607297420502, "learning_rate": 5.816611686010465e-05, "loss": 0.0077, "step": 47540 }, { "grad_norm": 0.1401018649339676, "learning_rate": 5.814980373104033e-05, "loss": 0.0072, "step": 47550 }, { "grad_norm": 0.09061422944068909, "learning_rate": 5.813348971072572e-05, "loss": 0.0066, "step": 47560 }, { "grad_norm": 0.08063162118196487, "learning_rate": 5.811717480094492e-05, "loss": 0.0062, "step": 47570 }, { "grad_norm": 0.06894516944885254, "learning_rate": 5.810085900348209e-05, "loss": 0.0065, "step": 47580 }, { "grad_norm": 0.08463185280561447, "learning_rate": 5.8084542320121483e-05, "loss": 0.005, "step": 47590 }, { "grad_norm": 0.05880383774638176, "learning_rate": 5.8068224752647497e-05, "loss": 0.0038, "step": 47600 }, { "grad_norm": 0.06792597472667694, "learning_rate": 5.805190630284457e-05, "loss": 0.008, "step": 47610 }, { "grad_norm": 0.10080566257238388, "learning_rate": 5.803558697249729e-05, "loss": 0.0063, "step": 47620 }, { "grad_norm": 0.1068839281797409, "learning_rate": 5.8019266763390276e-05, "loss": 0.0058, "step": 47630 }, { "grad_norm": 0.07272562384605408, "learning_rate": 5.800294567730833e-05, "loss": 0.0062, "step": 47640 }, { "grad_norm": 0.08865657448768616, "learning_rate": 5.798662371603628e-05, "loss": 0.0071, "step": 47650 }, { "grad_norm": 0.16492940485477448, "learning_rate": 5.797030088135904e-05, "loss": 0.0059, "step": 47660 }, { "grad_norm": 0.08707407861948013, "learning_rate": 5.7953977175061705e-05, "loss": 0.0059, "step": 47670 }, { "grad_norm": 0.09227748960256577, "learning_rate": 5.793765259892938e-05, "loss": 0.0051, "step": 47680 }, { "grad_norm": 0.16820184886455536, "learning_rate": 5.792132715474729e-05, "loss": 0.0071, "step": 47690 }, { "grad_norm": 0.0904010534286499, "learning_rate": 5.790500084430078e-05, "loss": 0.0054, "step": 47700 }, { "grad_norm": 0.08452684432268143, "learning_rate": 5.7888673669375274e-05, "loss": 0.0063, "step": 47710 }, { "grad_norm": 0.08484092354774475, "learning_rate": 5.787234563175625e-05, "loss": 0.0057, "step": 47720 }, { "grad_norm": 0.06840820610523224, "learning_rate": 5.7856016733229355e-05, "loss": 0.0062, "step": 47730 }, { "grad_norm": 0.06374131888151169, "learning_rate": 5.7839686975580297e-05, "loss": 0.007, "step": 47740 }, { "grad_norm": 0.10756520926952362, "learning_rate": 5.782335636059484e-05, "loss": 0.0084, "step": 47750 }, { "grad_norm": 0.07350286841392517, "learning_rate": 5.780702489005889e-05, "loss": 0.0058, "step": 47760 }, { "grad_norm": 0.20168332755565643, "learning_rate": 5.779069256575845e-05, "loss": 0.0061, "step": 47770 }, { "grad_norm": 0.08668828755617142, "learning_rate": 5.7774359389479574e-05, "loss": 0.0058, "step": 47780 }, { "grad_norm": 0.11196501553058624, "learning_rate": 5.775802536300845e-05, "loss": 0.0059, "step": 47790 }, { "grad_norm": 0.18295995891094208, "learning_rate": 5.774169048813134e-05, "loss": 0.0077, "step": 47800 }, { "grad_norm": 0.07413417845964432, "learning_rate": 5.77253547666346e-05, "loss": 0.0065, "step": 47810 }, { "grad_norm": 0.0782889574766159, "learning_rate": 5.770901820030465e-05, "loss": 0.0072, "step": 47820 }, { "grad_norm": 0.09022845327854156, "learning_rate": 5.769268079092809e-05, "loss": 0.0063, "step": 47830 }, { "grad_norm": 0.13883943855762482, "learning_rate": 5.767634254029151e-05, "loss": 0.0064, "step": 47840 }, { "grad_norm": 0.10272057354450226, "learning_rate": 5.7660003450181655e-05, "loss": 0.0057, "step": 47850 }, { "grad_norm": 0.07807616889476776, "learning_rate": 5.764366352238534e-05, "loss": 0.0051, "step": 47860 }, { "grad_norm": 0.14482781291007996, "learning_rate": 5.7627322758689474e-05, "loss": 0.0078, "step": 47870 }, { "grad_norm": 0.11199285089969635, "learning_rate": 5.761098116088105e-05, "loss": 0.0079, "step": 47880 }, { "grad_norm": 0.1605379283428192, "learning_rate": 5.759463873074717e-05, "loss": 0.0086, "step": 47890 }, { "grad_norm": 0.08155421167612076, "learning_rate": 5.757829547007504e-05, "loss": 0.0056, "step": 47900 }, { "grad_norm": 0.09761674702167511, "learning_rate": 5.756195138065189e-05, "loss": 0.0052, "step": 47910 }, { "grad_norm": 0.11182606220245361, "learning_rate": 5.754560646426511e-05, "loss": 0.0059, "step": 47920 }, { "grad_norm": 0.25318363308906555, "learning_rate": 5.752926072270216e-05, "loss": 0.0068, "step": 47930 }, { "grad_norm": 0.12051049619913101, "learning_rate": 5.7512914157750563e-05, "loss": 0.0059, "step": 47940 }, { "grad_norm": 0.11525359749794006, "learning_rate": 5.749656677119798e-05, "loss": 0.0065, "step": 47950 }, { "grad_norm": 0.06774083524942398, "learning_rate": 5.7480218564832125e-05, "loss": 0.0073, "step": 47960 }, { "grad_norm": 0.21217204630374908, "learning_rate": 5.746386954044082e-05, "loss": 0.0087, "step": 47970 }, { "grad_norm": 0.1387229710817337, "learning_rate": 5.744751969981195e-05, "loss": 0.0058, "step": 47980 }, { "grad_norm": 0.07385514676570892, "learning_rate": 5.7431169044733526e-05, "loss": 0.0071, "step": 47990 }, { "grad_norm": 0.1435597538948059, "learning_rate": 5.741481757699364e-05, "loss": 0.0077, "step": 48000 }, { "grad_norm": 0.06946974992752075, "learning_rate": 5.7398465298380434e-05, "loss": 0.0049, "step": 48010 }, { "grad_norm": 0.18791191279888153, "learning_rate": 5.7382112210682193e-05, "loss": 0.0062, "step": 48020 }, { "grad_norm": 0.05975443497300148, "learning_rate": 5.7365758315687266e-05, "loss": 0.0057, "step": 48030 }, { "grad_norm": 0.08446530997753143, "learning_rate": 5.734940361518407e-05, "loss": 0.0056, "step": 48040 }, { "grad_norm": 0.08152449876070023, "learning_rate": 5.733304811096116e-05, "loss": 0.0056, "step": 48050 }, { "grad_norm": 0.17443618178367615, "learning_rate": 5.731669180480713e-05, "loss": 0.0064, "step": 48060 }, { "grad_norm": 0.10794923454523087, "learning_rate": 5.730033469851067e-05, "loss": 0.0053, "step": 48070 }, { "grad_norm": 0.06389570236206055, "learning_rate": 5.7283976793860607e-05, "loss": 0.0043, "step": 48080 }, { "grad_norm": 0.21169647574424744, "learning_rate": 5.726761809264577e-05, "loss": 0.007, "step": 48090 }, { "grad_norm": 0.12852519750595093, "learning_rate": 5.7251258596655155e-05, "loss": 0.0089, "step": 48100 }, { "grad_norm": 0.10359698534011841, "learning_rate": 5.72348983076778e-05, "loss": 0.0061, "step": 48110 }, { "grad_norm": 0.084420345723629, "learning_rate": 5.7218537227502854e-05, "loss": 0.0061, "step": 48120 }, { "grad_norm": 0.06710296124219894, "learning_rate": 5.720217535791951e-05, "loss": 0.0059, "step": 48130 }, { "grad_norm": 0.0553891696035862, "learning_rate": 5.718581270071711e-05, "loss": 0.0057, "step": 48140 }, { "grad_norm": 0.05334371328353882, "learning_rate": 5.716944925768505e-05, "loss": 0.0062, "step": 48150 }, { "grad_norm": 0.05703217163681984, "learning_rate": 5.7153085030612786e-05, "loss": 0.0049, "step": 48160 }, { "grad_norm": 0.10184566676616669, "learning_rate": 5.713672002128989e-05, "loss": 0.0053, "step": 48170 }, { "grad_norm": 0.059882234781980515, "learning_rate": 5.712035423150602e-05, "loss": 0.0053, "step": 48180 }, { "grad_norm": 0.08964300900697708, "learning_rate": 5.710398766305094e-05, "loss": 0.0048, "step": 48190 }, { "grad_norm": 0.13993844389915466, "learning_rate": 5.708762031771442e-05, "loss": 0.005, "step": 48200 }, { "grad_norm": 0.0957019031047821, "learning_rate": 5.707125219728642e-05, "loss": 0.0059, "step": 48210 }, { "grad_norm": 0.08417119830846786, "learning_rate": 5.7054883303556905e-05, "loss": 0.0056, "step": 48220 }, { "grad_norm": 0.11466433852910995, "learning_rate": 5.703851363831595e-05, "loss": 0.0062, "step": 48230 }, { "grad_norm": 0.11614283919334412, "learning_rate": 5.702214320335374e-05, "loss": 0.0072, "step": 48240 }, { "grad_norm": 0.05607679858803749, "learning_rate": 5.7005772000460514e-05, "loss": 0.0067, "step": 48250 }, { "grad_norm": 0.07031935453414917, "learning_rate": 5.698940003142656e-05, "loss": 0.0054, "step": 48260 }, { "grad_norm": 0.04926789551973343, "learning_rate": 5.697302729804236e-05, "loss": 0.0067, "step": 48270 }, { "grad_norm": 0.13801562786102295, "learning_rate": 5.695665380209837e-05, "loss": 0.006, "step": 48280 }, { "grad_norm": 0.1345580518245697, "learning_rate": 5.6940279545385165e-05, "loss": 0.0058, "step": 48290 }, { "grad_norm": 0.09240389615297318, "learning_rate": 5.692390452969344e-05, "loss": 0.0066, "step": 48300 }, { "grad_norm": 0.15038466453552246, "learning_rate": 5.690752875681392e-05, "loss": 0.0086, "step": 48310 }, { "grad_norm": 0.0666922777891159, "learning_rate": 5.6891152228537435e-05, "loss": 0.0073, "step": 48320 }, { "grad_norm": 0.14758926630020142, "learning_rate": 5.687477494665492e-05, "loss": 0.0058, "step": 48330 }, { "grad_norm": 0.1001620888710022, "learning_rate": 5.685839691295733e-05, "loss": 0.0064, "step": 48340 }, { "grad_norm": 0.15897054970264435, "learning_rate": 5.6842018129235786e-05, "loss": 0.0067, "step": 48350 }, { "grad_norm": 0.09743533283472061, "learning_rate": 5.6825638597281404e-05, "loss": 0.0086, "step": 48360 }, { "grad_norm": 0.07670412957668304, "learning_rate": 5.680925831888546e-05, "loss": 0.0069, "step": 48370 }, { "grad_norm": 0.07878190279006958, "learning_rate": 5.6792877295839274e-05, "loss": 0.0073, "step": 48380 }, { "grad_norm": 0.053014758974313736, "learning_rate": 5.6776495529934224e-05, "loss": 0.0058, "step": 48390 }, { "grad_norm": 0.08144645392894745, "learning_rate": 5.6760113022961824e-05, "loss": 0.0066, "step": 48400 }, { "grad_norm": 0.09770940244197845, "learning_rate": 5.6743729776713617e-05, "loss": 0.0067, "step": 48410 }, { "grad_norm": 0.04741993546485901, "learning_rate": 5.672734579298126e-05, "loss": 0.0055, "step": 48420 }, { "grad_norm": 0.08158435672521591, "learning_rate": 5.671096107355649e-05, "loss": 0.006, "step": 48430 }, { "grad_norm": 0.06771930307149887, "learning_rate": 5.66945756202311e-05, "loss": 0.0081, "step": 48440 }, { "grad_norm": 0.08551964908838272, "learning_rate": 5.667818943479699e-05, "loss": 0.0051, "step": 48450 }, { "grad_norm": 0.0937681794166565, "learning_rate": 5.666180251904612e-05, "loss": 0.0074, "step": 48460 }, { "grad_norm": 0.08215589821338654, "learning_rate": 5.6645414874770555e-05, "loss": 0.0047, "step": 48470 }, { "grad_norm": 0.09302599728107452, "learning_rate": 5.66290265037624e-05, "loss": 0.0044, "step": 48480 }, { "grad_norm": 0.07331863045692444, "learning_rate": 5.661263740781386e-05, "loss": 0.0084, "step": 48490 }, { "grad_norm": 0.0785355195403099, "learning_rate": 5.6596247588717254e-05, "loss": 0.0063, "step": 48500 }, { "grad_norm": 0.060672514140605927, "learning_rate": 5.6579857048264926e-05, "loss": 0.0072, "step": 48510 }, { "grad_norm": 0.091746024787426, "learning_rate": 5.6563465788249314e-05, "loss": 0.0096, "step": 48520 }, { "grad_norm": 0.11121649295091629, "learning_rate": 5.6547073810462956e-05, "loss": 0.0077, "step": 48530 }, { "grad_norm": 0.11089867353439331, "learning_rate": 5.653068111669846e-05, "loss": 0.0061, "step": 48540 }, { "grad_norm": 0.06363646686077118, "learning_rate": 5.651428770874848e-05, "loss": 0.0047, "step": 48550 }, { "grad_norm": 0.09133771061897278, "learning_rate": 5.64978935884058e-05, "loss": 0.0066, "step": 48560 }, { "grad_norm": 0.05377250537276268, "learning_rate": 5.6481498757463244e-05, "loss": 0.0045, "step": 48570 }, { "grad_norm": 0.14171555638313293, "learning_rate": 5.646510321771373e-05, "loss": 0.0071, "step": 48580 }, { "grad_norm": 0.08735951036214828, "learning_rate": 5.644870697095024e-05, "loss": 0.007, "step": 48590 }, { "grad_norm": 0.08824998140335083, "learning_rate": 5.643231001896586e-05, "loss": 0.0056, "step": 48600 }, { "grad_norm": 0.0933084636926651, "learning_rate": 5.6415912363553726e-05, "loss": 0.008, "step": 48610 }, { "grad_norm": 0.1600790023803711, "learning_rate": 5.639951400650706e-05, "loss": 0.0054, "step": 48620 }, { "grad_norm": 0.07880342751741409, "learning_rate": 5.6383114949619165e-05, "loss": 0.0052, "step": 48630 }, { "grad_norm": 0.07457423955202103, "learning_rate": 5.636671519468342e-05, "loss": 0.0057, "step": 48640 }, { "grad_norm": 0.08648239076137543, "learning_rate": 5.635031474349327e-05, "loss": 0.0085, "step": 48650 }, { "grad_norm": 0.1481177806854248, "learning_rate": 5.6333913597842246e-05, "loss": 0.0064, "step": 48660 }, { "grad_norm": 0.11383888870477676, "learning_rate": 5.6317511759523955e-05, "loss": 0.0061, "step": 48670 }, { "grad_norm": 0.07995028048753738, "learning_rate": 5.630110923033207e-05, "loss": 0.0053, "step": 48680 }, { "grad_norm": 0.08119828999042511, "learning_rate": 5.628470601206036e-05, "loss": 0.0048, "step": 48690 }, { "grad_norm": 0.10059516876935959, "learning_rate": 5.626830210650266e-05, "loss": 0.0058, "step": 48700 }, { "grad_norm": 0.06256729364395142, "learning_rate": 5.625189751545285e-05, "loss": 0.0053, "step": 48710 }, { "grad_norm": 0.09352343529462814, "learning_rate": 5.6235492240704936e-05, "loss": 0.0042, "step": 48720 }, { "grad_norm": 0.08825743198394775, "learning_rate": 5.621908628405296e-05, "loss": 0.0063, "step": 48730 }, { "grad_norm": 0.14709050953388214, "learning_rate": 5.620267964729106e-05, "loss": 0.0048, "step": 48740 }, { "grad_norm": 0.09759602695703506, "learning_rate": 5.618627233221344e-05, "loss": 0.0052, "step": 48750 }, { "grad_norm": 0.08095821738243103, "learning_rate": 5.616986434061438e-05, "loss": 0.0053, "step": 48760 }, { "grad_norm": 0.0648711770772934, "learning_rate": 5.615345567428822e-05, "loss": 0.0059, "step": 48770 }, { "grad_norm": 0.07612290233373642, "learning_rate": 5.613704633502941e-05, "loss": 0.0065, "step": 48780 }, { "grad_norm": 0.06885072588920593, "learning_rate": 5.612063632463245e-05, "loss": 0.0052, "step": 48790 }, { "grad_norm": 0.08558649569749832, "learning_rate": 5.610422564489188e-05, "loss": 0.005, "step": 48800 }, { "grad_norm": 0.12383700162172318, "learning_rate": 5.608781429760239e-05, "loss": 0.0055, "step": 48810 }, { "grad_norm": 0.0784924104809761, "learning_rate": 5.607140228455866e-05, "loss": 0.0053, "step": 48820 }, { "grad_norm": 0.10648423433303833, "learning_rate": 5.605498960755553e-05, "loss": 0.006, "step": 48830 }, { "grad_norm": 0.10638344287872314, "learning_rate": 5.603857626838782e-05, "loss": 0.0082, "step": 48840 }, { "grad_norm": 0.100101999938488, "learning_rate": 5.60221622688505e-05, "loss": 0.0084, "step": 48850 }, { "grad_norm": 0.10093581676483154, "learning_rate": 5.6005747610738565e-05, "loss": 0.0064, "step": 48860 }, { "grad_norm": 0.10128932446241379, "learning_rate": 5.59893322958471e-05, "loss": 0.0059, "step": 48870 }, { "grad_norm": 0.097167007625103, "learning_rate": 5.5972916325971256e-05, "loss": 0.0061, "step": 48880 }, { "grad_norm": 0.12108005583286285, "learning_rate": 5.595649970290628e-05, "loss": 0.005, "step": 48890 }, { "grad_norm": 0.07778268307447433, "learning_rate": 5.5940082428447426e-05, "loss": 0.0073, "step": 48900 }, { "grad_norm": 0.07877438515424728, "learning_rate": 5.592366450439012e-05, "loss": 0.0051, "step": 48910 }, { "grad_norm": 0.09142932295799255, "learning_rate": 5.590724593252975e-05, "loss": 0.0062, "step": 48920 }, { "grad_norm": 0.06470005959272385, "learning_rate": 5.589082671466184e-05, "loss": 0.005, "step": 48930 }, { "grad_norm": 0.10961835831403732, "learning_rate": 5.587440685258199e-05, "loss": 0.009, "step": 48940 }, { "grad_norm": 0.09356467425823212, "learning_rate": 5.585798634808583e-05, "loss": 0.0068, "step": 48950 }, { "grad_norm": 0.08625147491693497, "learning_rate": 5.584156520296909e-05, "loss": 0.0047, "step": 48960 }, { "grad_norm": 0.08105389773845673, "learning_rate": 5.582514341902757e-05, "loss": 0.0063, "step": 48970 }, { "grad_norm": 0.0489436499774456, "learning_rate": 5.580872099805713e-05, "loss": 0.0059, "step": 48980 }, { "grad_norm": 0.05958067998290062, "learning_rate": 5.5792297941853674e-05, "loss": 0.0047, "step": 48990 }, { "grad_norm": 0.09900692850351334, "learning_rate": 5.5775874252213247e-05, "loss": 0.0075, "step": 49000 }, { "grad_norm": 0.10535408556461334, "learning_rate": 5.575944993093189e-05, "loss": 0.0062, "step": 49010 }, { "grad_norm": 0.10632442682981491, "learning_rate": 5.574302497980574e-05, "loss": 0.006, "step": 49020 }, { "grad_norm": 0.08805190026760101, "learning_rate": 5.5726599400631005e-05, "loss": 0.0069, "step": 49030 }, { "grad_norm": 0.13542141020298004, "learning_rate": 5.571017319520399e-05, "loss": 0.0062, "step": 49040 }, { "grad_norm": 0.07184035331010818, "learning_rate": 5.5693746365320984e-05, "loss": 0.0074, "step": 49050 }, { "grad_norm": 0.05262838304042816, "learning_rate": 5.567731891277848e-05, "loss": 0.0039, "step": 49060 }, { "grad_norm": 0.05796550214290619, "learning_rate": 5.566089083937288e-05, "loss": 0.0039, "step": 49070 }, { "grad_norm": 0.14554041624069214, "learning_rate": 5.564446214690079e-05, "loss": 0.0091, "step": 49080 }, { "grad_norm": 0.11938869208097458, "learning_rate": 5.56280328371588e-05, "loss": 0.0045, "step": 49090 }, { "grad_norm": 0.06582871079444885, "learning_rate": 5.5611602911943596e-05, "loss": 0.0053, "step": 49100 }, { "grad_norm": 0.0988667756319046, "learning_rate": 5.559517237305194e-05, "loss": 0.0079, "step": 49110 }, { "grad_norm": 0.07820814102888107, "learning_rate": 5.557874122228064e-05, "loss": 0.004, "step": 49120 }, { "grad_norm": 0.0983959212899208, "learning_rate": 5.55623094614266e-05, "loss": 0.0048, "step": 49130 }, { "grad_norm": 0.0749407410621643, "learning_rate": 5.5545877092286743e-05, "loss": 0.0066, "step": 49140 }, { "grad_norm": 0.10720720887184143, "learning_rate": 5.552944411665811e-05, "loss": 0.0054, "step": 49150 }, { "grad_norm": 0.10360642522573471, "learning_rate": 5.5513010536337795e-05, "loss": 0.0068, "step": 49160 }, { "grad_norm": 0.12355047464370728, "learning_rate": 5.5496576353122934e-05, "loss": 0.006, "step": 49170 }, { "grad_norm": 0.06708340346813202, "learning_rate": 5.548014156881074e-05, "loss": 0.0053, "step": 49180 }, { "grad_norm": 0.10511063039302826, "learning_rate": 5.5463706185198494e-05, "loss": 0.0059, "step": 49190 }, { "grad_norm": 0.07260026782751083, "learning_rate": 5.544727020408358e-05, "loss": 0.0067, "step": 49200 }, { "grad_norm": 0.0600508414208889, "learning_rate": 5.5430833627263366e-05, "loss": 0.0049, "step": 49210 }, { "grad_norm": 0.10302677750587463, "learning_rate": 5.5414396456535354e-05, "loss": 0.0055, "step": 49220 }, { "grad_norm": 0.05855655297636986, "learning_rate": 5.539795869369709e-05, "loss": 0.006, "step": 49230 }, { "grad_norm": 0.1314133256673813, "learning_rate": 5.538152034054618e-05, "loss": 0.0065, "step": 49240 }, { "grad_norm": 0.08891425281763077, "learning_rate": 5.536508139888028e-05, "loss": 0.0071, "step": 49250 }, { "grad_norm": 0.11970534920692444, "learning_rate": 5.534864187049716e-05, "loss": 0.0056, "step": 49260 }, { "grad_norm": 0.10521288216114044, "learning_rate": 5.5332201757194615e-05, "loss": 0.007, "step": 49270 }, { "grad_norm": 0.06767702847719193, "learning_rate": 5.531576106077049e-05, "loss": 0.0052, "step": 49280 }, { "grad_norm": 0.12481176108121872, "learning_rate": 5.529931978302272e-05, "loss": 0.0055, "step": 49290 }, { "grad_norm": 0.11764166504144669, "learning_rate": 5.528287792574932e-05, "loss": 0.007, "step": 49300 }, { "grad_norm": 0.09208124130964279, "learning_rate": 5.5266435490748324e-05, "loss": 0.0051, "step": 49310 }, { "grad_norm": 0.07177701592445374, "learning_rate": 5.524999247981787e-05, "loss": 0.0043, "step": 49320 }, { "grad_norm": 0.0659525990486145, "learning_rate": 5.523354889475613e-05, "loss": 0.0046, "step": 49330 }, { "grad_norm": 0.05691954120993614, "learning_rate": 5.521710473736134e-05, "loss": 0.007, "step": 49340 }, { "grad_norm": 0.06605962663888931, "learning_rate": 5.5200660009431826e-05, "loss": 0.0052, "step": 49350 }, { "grad_norm": 0.08732111006975174, "learning_rate": 5.518421471276596e-05, "loss": 0.0055, "step": 49360 }, { "grad_norm": 0.1563643366098404, "learning_rate": 5.5167768849162174e-05, "loss": 0.0057, "step": 49370 }, { "grad_norm": 0.101982980966568, "learning_rate": 5.515132242041893e-05, "loss": 0.0076, "step": 49380 }, { "grad_norm": 0.09322881698608398, "learning_rate": 5.513487542833483e-05, "loss": 0.0079, "step": 49390 }, { "grad_norm": 0.09105533361434937, "learning_rate": 5.5118427874708466e-05, "loss": 0.0049, "step": 49400 }, { "grad_norm": 0.06533670425415039, "learning_rate": 5.510197976133853e-05, "loss": 0.0077, "step": 49410 }, { "grad_norm": 0.06958554685115814, "learning_rate": 5.508553109002376e-05, "loss": 0.0084, "step": 49420 }, { "grad_norm": 0.05022652819752693, "learning_rate": 5.5069081862562957e-05, "loss": 0.0058, "step": 49430 }, { "grad_norm": 0.09362038224935532, "learning_rate": 5.5052632080754965e-05, "loss": 0.0061, "step": 49440 }, { "grad_norm": 0.06223532184958458, "learning_rate": 5.503618174639874e-05, "loss": 0.0045, "step": 49450 }, { "grad_norm": 0.07952805608510971, "learning_rate": 5.501973086129325e-05, "loss": 0.0053, "step": 49460 }, { "grad_norm": 0.06889232993125916, "learning_rate": 5.500327942723753e-05, "loss": 0.0044, "step": 49470 }, { "grad_norm": 0.08190281689167023, "learning_rate": 5.498682744603071e-05, "loss": 0.0069, "step": 49480 }, { "grad_norm": 0.07204695791006088, "learning_rate": 5.497037491947194e-05, "loss": 0.0065, "step": 49490 }, { "grad_norm": 0.07533285021781921, "learning_rate": 5.4953921849360424e-05, "loss": 0.0078, "step": 49500 }, { "grad_norm": 0.07332232594490051, "learning_rate": 5.493746823749547e-05, "loss": 0.0067, "step": 49510 }, { "grad_norm": 0.08380582183599472, "learning_rate": 5.49210140856764e-05, "loss": 0.0047, "step": 49520 }, { "grad_norm": 0.08037548512220383, "learning_rate": 5.4904559395702647e-05, "loss": 0.0047, "step": 49530 }, { "grad_norm": 0.07420177757740021, "learning_rate": 5.488810416937364e-05, "loss": 0.0066, "step": 49540 }, { "grad_norm": 0.17317591607570648, "learning_rate": 5.4871648408488926e-05, "loss": 0.0066, "step": 49550 }, { "grad_norm": 0.10586941242218018, "learning_rate": 5.485519211484807e-05, "loss": 0.005, "step": 49560 }, { "grad_norm": 0.09013570845127106, "learning_rate": 5.4838735290250676e-05, "loss": 0.0054, "step": 49570 }, { "grad_norm": 0.10580822080373764, "learning_rate": 5.482227793649649e-05, "loss": 0.0053, "step": 49580 }, { "grad_norm": 0.08101443946361542, "learning_rate": 5.480582005538524e-05, "loss": 0.0104, "step": 49590 }, { "grad_norm": 0.0681440681219101, "learning_rate": 5.478936164871671e-05, "loss": 0.0058, "step": 49600 }, { "grad_norm": 0.1081363782286644, "learning_rate": 5.477290271829081e-05, "loss": 0.0066, "step": 49610 }, { "grad_norm": 0.08872342109680176, "learning_rate": 5.475644326590744e-05, "loss": 0.0045, "step": 49620 }, { "grad_norm": 0.11846423149108887, "learning_rate": 5.473998329336658e-05, "loss": 0.0062, "step": 49630 }, { "grad_norm": 0.09212376922369003, "learning_rate": 5.4723522802468286e-05, "loss": 0.0053, "step": 49640 }, { "grad_norm": 0.09577935189008713, "learning_rate": 5.470706179501264e-05, "loss": 0.0054, "step": 49650 }, { "grad_norm": 0.11161526292562485, "learning_rate": 5.4690600272799776e-05, "loss": 0.005, "step": 49660 }, { "grad_norm": 0.06764725595712662, "learning_rate": 5.467413823762993e-05, "loss": 0.0077, "step": 49670 }, { "grad_norm": 0.06726779043674469, "learning_rate": 5.465767569130335e-05, "loss": 0.0061, "step": 49680 }, { "grad_norm": 0.06156165897846222, "learning_rate": 5.464121263562036e-05, "loss": 0.0075, "step": 49690 }, { "grad_norm": 0.06508403271436691, "learning_rate": 5.4624749072381343e-05, "loss": 0.0052, "step": 49700 }, { "grad_norm": 0.06412573158740997, "learning_rate": 5.460828500338672e-05, "loss": 0.0054, "step": 49710 }, { "grad_norm": 0.10367027670145035, "learning_rate": 5.459182043043698e-05, "loss": 0.0083, "step": 49720 }, { "grad_norm": 0.09394048154354095, "learning_rate": 5.457535535533265e-05, "loss": 0.0063, "step": 49730 }, { "grad_norm": 0.10252919048070908, "learning_rate": 5.4558889779874334e-05, "loss": 0.008, "step": 49740 }, { "grad_norm": 0.05246991664171219, "learning_rate": 5.454242370586269e-05, "loss": 0.0052, "step": 49750 }, { "grad_norm": 0.07226159423589706, "learning_rate": 5.452595713509843e-05, "loss": 0.0065, "step": 49760 }, { "grad_norm": 0.09299445897340775, "learning_rate": 5.4509490069382274e-05, "loss": 0.0062, "step": 49770 }, { "grad_norm": 0.09566245973110199, "learning_rate": 5.449302251051509e-05, "loss": 0.0061, "step": 49780 }, { "grad_norm": 0.08611899614334106, "learning_rate": 5.44765544602977e-05, "loss": 0.0073, "step": 49790 }, { "grad_norm": 0.12069649249315262, "learning_rate": 5.446008592053107e-05, "loss": 0.007, "step": 49800 }, { "grad_norm": 0.08224045485258102, "learning_rate": 5.4443616893016135e-05, "loss": 0.0056, "step": 49810 }, { "grad_norm": 0.09733958542346954, "learning_rate": 5.4427147379553934e-05, "loss": 0.0056, "step": 49820 }, { "grad_norm": 0.07987083494663239, "learning_rate": 5.441067738194556e-05, "loss": 0.0054, "step": 49830 }, { "grad_norm": 0.07668361067771912, "learning_rate": 5.439420690199214e-05, "loss": 0.0061, "step": 49840 }, { "grad_norm": 0.0753711387515068, "learning_rate": 5.437773594149484e-05, "loss": 0.0064, "step": 49850 }, { "grad_norm": 0.0759597197175026, "learning_rate": 5.436126450225495e-05, "loss": 0.0044, "step": 49860 }, { "grad_norm": 0.09319114685058594, "learning_rate": 5.434479258607371e-05, "loss": 0.0053, "step": 49870 }, { "grad_norm": 0.07509224116802216, "learning_rate": 5.43283201947525e-05, "loss": 0.0054, "step": 49880 }, { "grad_norm": 0.10479556024074554, "learning_rate": 5.431184733009268e-05, "loss": 0.0052, "step": 49890 }, { "grad_norm": 0.12308724224567413, "learning_rate": 5.4295373993895736e-05, "loss": 0.0047, "step": 49900 }, { "grad_norm": 0.08143184334039688, "learning_rate": 5.4278900187963157e-05, "loss": 0.0043, "step": 49910 }, { "grad_norm": 0.13741745054721832, "learning_rate": 5.426242591409646e-05, "loss": 0.0054, "step": 49920 }, { "grad_norm": 0.12766019999980927, "learning_rate": 5.42459511740973e-05, "loss": 0.0089, "step": 49930 }, { "grad_norm": 0.08964943885803223, "learning_rate": 5.422947596976729e-05, "loss": 0.0075, "step": 49940 }, { "grad_norm": 0.24303343892097473, "learning_rate": 5.4213000302908134e-05, "loss": 0.006, "step": 49950 }, { "grad_norm": 0.07988370954990387, "learning_rate": 5.419652417532162e-05, "loss": 0.0068, "step": 49960 }, { "grad_norm": 0.148246631026268, "learning_rate": 5.4180047588809534e-05, "loss": 0.0056, "step": 49970 }, { "grad_norm": 0.1042245477437973, "learning_rate": 5.4163570545173704e-05, "loss": 0.0063, "step": 49980 }, { "grad_norm": 0.14081934094429016, "learning_rate": 5.414709304621608e-05, "loss": 0.0069, "step": 49990 }, { "grad_norm": 0.08336088806390762, "learning_rate": 5.413061509373858e-05, "loss": 0.0059, "step": 50000 }, { "grad_norm": 0.08029274642467499, "learning_rate": 5.411413668954324e-05, "loss": 0.0087, "step": 50010 }, { "grad_norm": 0.12572982907295227, "learning_rate": 5.4097657835432083e-05, "loss": 0.0058, "step": 50020 }, { "grad_norm": 0.08890955150127411, "learning_rate": 5.408117853320723e-05, "loss": 0.0088, "step": 50030 }, { "grad_norm": 0.0693216547369957, "learning_rate": 5.406469878467084e-05, "loss": 0.0067, "step": 50040 }, { "grad_norm": 0.07848190516233444, "learning_rate": 5.404821859162509e-05, "loss": 0.008, "step": 50050 }, { "grad_norm": 0.21355344355106354, "learning_rate": 5.403173795587225e-05, "loss": 0.007, "step": 50060 }, { "grad_norm": 0.07836052775382996, "learning_rate": 5.4015256879214606e-05, "loss": 0.0056, "step": 50070 }, { "grad_norm": 0.1287107914686203, "learning_rate": 5.39987753634545e-05, "loss": 0.0062, "step": 50080 }, { "grad_norm": 0.13610300421714783, "learning_rate": 5.3982293410394335e-05, "loss": 0.007, "step": 50090 }, { "grad_norm": 0.09923036396503448, "learning_rate": 5.396581102183655e-05, "loss": 0.0061, "step": 50100 }, { "grad_norm": 0.0935204029083252, "learning_rate": 5.3949328199583634e-05, "loss": 0.0057, "step": 50110 }, { "grad_norm": 0.06382323801517487, "learning_rate": 5.393284494543813e-05, "loss": 0.0078, "step": 50120 }, { "grad_norm": 0.09739023447036743, "learning_rate": 5.391636126120262e-05, "loss": 0.0088, "step": 50130 }, { "grad_norm": 0.06833621114492416, "learning_rate": 5.389987714867971e-05, "loss": 0.0051, "step": 50140 }, { "grad_norm": 0.07193586230278015, "learning_rate": 5.3883392609672114e-05, "loss": 0.0049, "step": 50150 }, { "grad_norm": 0.07061214745044708, "learning_rate": 5.3866907645982545e-05, "loss": 0.0043, "step": 50160 }, { "grad_norm": 0.06710269302129745, "learning_rate": 5.385042225941375e-05, "loss": 0.005, "step": 50170 }, { "grad_norm": 0.0820130929350853, "learning_rate": 5.3833936451768585e-05, "loss": 0.0053, "step": 50180 }, { "grad_norm": 0.0977894514799118, "learning_rate": 5.3817450224849884e-05, "loss": 0.0066, "step": 50190 }, { "grad_norm": 0.1301395297050476, "learning_rate": 5.380096358046056e-05, "loss": 0.0058, "step": 50200 }, { "grad_norm": 0.056417737156152725, "learning_rate": 5.378447652040359e-05, "loss": 0.0055, "step": 50210 }, { "grad_norm": 0.07571479678153992, "learning_rate": 5.376798904648195e-05, "loss": 0.0049, "step": 50220 }, { "grad_norm": 0.08241195231676102, "learning_rate": 5.375150116049869e-05, "loss": 0.005, "step": 50230 }, { "grad_norm": 0.05363574996590614, "learning_rate": 5.373501286425691e-05, "loss": 0.0047, "step": 50240 }, { "grad_norm": 0.07924467325210571, "learning_rate": 5.3718524159559726e-05, "loss": 0.0039, "step": 50250 }, { "grad_norm": 0.1400996297597885, "learning_rate": 5.370203504821034e-05, "loss": 0.007, "step": 50260 }, { "grad_norm": 0.09281391650438309, "learning_rate": 5.368554553201196e-05, "loss": 0.0069, "step": 50270 }, { "grad_norm": 0.1595853865146637, "learning_rate": 5.3669055612767874e-05, "loss": 0.0071, "step": 50280 }, { "grad_norm": 0.07984550297260284, "learning_rate": 5.3652565292281376e-05, "loss": 0.0057, "step": 50290 }, { "grad_norm": 0.08494489639997482, "learning_rate": 5.3636074572355824e-05, "loss": 0.0059, "step": 50300 }, { "grad_norm": 0.06083560362458229, "learning_rate": 5.361958345479463e-05, "loss": 0.0061, "step": 50310 }, { "grad_norm": 0.07926460355520248, "learning_rate": 5.360309194140123e-05, "loss": 0.0069, "step": 50320 }, { "grad_norm": 0.11088575422763824, "learning_rate": 5.3586600033979106e-05, "loss": 0.0052, "step": 50330 }, { "grad_norm": 0.10124745219945908, "learning_rate": 5.357010773433181e-05, "loss": 0.0059, "step": 50340 }, { "grad_norm": 0.10589981079101562, "learning_rate": 5.3553615044262895e-05, "loss": 0.0049, "step": 50350 }, { "grad_norm": 0.08663994073867798, "learning_rate": 5.3537121965575964e-05, "loss": 0.0067, "step": 50360 }, { "grad_norm": 0.08389584720134735, "learning_rate": 5.352062850007471e-05, "loss": 0.0059, "step": 50370 }, { "grad_norm": 0.067860446870327, "learning_rate": 5.350413464956282e-05, "loss": 0.0053, "step": 50380 }, { "grad_norm": 0.13995245099067688, "learning_rate": 5.348764041584403e-05, "loss": 0.0055, "step": 50390 }, { "grad_norm": 0.10770779848098755, "learning_rate": 5.347114580072212e-05, "loss": 0.006, "step": 50400 }, { "grad_norm": 0.08400558680295944, "learning_rate": 5.345465080600093e-05, "loss": 0.0059, "step": 50410 }, { "grad_norm": 0.15845118463039398, "learning_rate": 5.3438155433484326e-05, "loss": 0.0056, "step": 50420 }, { "grad_norm": 0.09704972058534622, "learning_rate": 5.3421659684976197e-05, "loss": 0.0056, "step": 50430 }, { "grad_norm": 0.09094904363155365, "learning_rate": 5.340516356228052e-05, "loss": 0.0092, "step": 50440 }, { "grad_norm": 0.08561050891876221, "learning_rate": 5.338866706720128e-05, "loss": 0.0065, "step": 50450 }, { "grad_norm": 0.08301212638616562, "learning_rate": 5.337217020154249e-05, "loss": 0.0059, "step": 50460 }, { "grad_norm": 0.07773558795452118, "learning_rate": 5.335567296710825e-05, "loss": 0.0073, "step": 50470 }, { "grad_norm": 0.06396739929914474, "learning_rate": 5.333917536570265e-05, "loss": 0.0047, "step": 50480 }, { "grad_norm": 0.06924635916948318, "learning_rate": 5.332267739912986e-05, "loss": 0.0058, "step": 50490 }, { "grad_norm": 0.10774480551481247, "learning_rate": 5.330617906919405e-05, "loss": 0.0071, "step": 50500 }, { "grad_norm": 0.06818131357431412, "learning_rate": 5.328968037769949e-05, "loss": 0.0068, "step": 50510 }, { "grad_norm": 0.08846920728683472, "learning_rate": 5.327318132645042e-05, "loss": 0.0044, "step": 50520 }, { "grad_norm": 0.06434766203165054, "learning_rate": 5.325668191725116e-05, "loss": 0.005, "step": 50530 }, { "grad_norm": 0.06387355923652649, "learning_rate": 5.324018215190606e-05, "loss": 0.0045, "step": 50540 }, { "grad_norm": 0.09470800310373306, "learning_rate": 5.3223682032219515e-05, "loss": 0.0055, "step": 50550 }, { "grad_norm": 0.0875593051314354, "learning_rate": 5.320718155999595e-05, "loss": 0.006, "step": 50560 }, { "grad_norm": 0.18110710382461548, "learning_rate": 5.3190680737039835e-05, "loss": 0.007, "step": 50570 }, { "grad_norm": 0.07207777351140976, "learning_rate": 5.317417956515567e-05, "loss": 0.0064, "step": 50580 }, { "grad_norm": 0.07520391047000885, "learning_rate": 5.315767804614798e-05, "loss": 0.0046, "step": 50590 }, { "grad_norm": 0.052443768829107285, "learning_rate": 5.3141176181821395e-05, "loss": 0.0059, "step": 50600 }, { "grad_norm": 0.075651615858078, "learning_rate": 5.31246739739805e-05, "loss": 0.0065, "step": 50610 }, { "grad_norm": 0.07442744076251984, "learning_rate": 5.310817142442994e-05, "loss": 0.0049, "step": 50620 }, { "grad_norm": 0.07448754459619522, "learning_rate": 5.309166853497445e-05, "loss": 0.0053, "step": 50630 }, { "grad_norm": 0.08773719519376755, "learning_rate": 5.307516530741873e-05, "loss": 0.006, "step": 50640 }, { "grad_norm": 0.07622720301151276, "learning_rate": 5.305866174356754e-05, "loss": 0.0043, "step": 50650 }, { "grad_norm": 0.12224938720464706, "learning_rate": 5.304215784522571e-05, "loss": 0.0053, "step": 50660 }, { "grad_norm": 0.07054558396339417, "learning_rate": 5.302565361419808e-05, "loss": 0.0046, "step": 50670 }, { "grad_norm": 0.12230922281742096, "learning_rate": 5.3009149052289507e-05, "loss": 0.0057, "step": 50680 }, { "grad_norm": 0.0680338591337204, "learning_rate": 5.299264416130493e-05, "loss": 0.0055, "step": 50690 }, { "grad_norm": 0.11931020021438599, "learning_rate": 5.297613894304928e-05, "loss": 0.0062, "step": 50700 }, { "grad_norm": 0.05041477456688881, "learning_rate": 5.2959633399327534e-05, "loss": 0.0049, "step": 50710 }, { "grad_norm": 0.08540406078100204, "learning_rate": 5.294312753194476e-05, "loss": 0.0049, "step": 50720 }, { "grad_norm": 0.05538768321275711, "learning_rate": 5.292662134270596e-05, "loss": 0.0053, "step": 50730 }, { "grad_norm": 0.06553760170936584, "learning_rate": 5.291011483341626e-05, "loss": 0.0057, "step": 50740 }, { "grad_norm": 0.10859310626983643, "learning_rate": 5.2893608005880767e-05, "loss": 0.0052, "step": 50750 }, { "grad_norm": 0.06853152066469193, "learning_rate": 5.287710086190467e-05, "loss": 0.006, "step": 50760 }, { "grad_norm": 0.10811741650104523, "learning_rate": 5.286059340329314e-05, "loss": 0.0083, "step": 50770 }, { "grad_norm": 0.09498916566371918, "learning_rate": 5.284408563185141e-05, "loss": 0.0064, "step": 50780 }, { "grad_norm": 0.118771493434906, "learning_rate": 5.282757754938476e-05, "loss": 0.0057, "step": 50790 }, { "grad_norm": 0.04540213197469711, "learning_rate": 5.281106915769849e-05, "loss": 0.0046, "step": 50800 }, { "grad_norm": 0.15476833283901215, "learning_rate": 5.2794560458597897e-05, "loss": 0.0064, "step": 50810 }, { "grad_norm": 0.0715632438659668, "learning_rate": 5.27780514538884e-05, "loss": 0.0057, "step": 50820 }, { "grad_norm": 0.10747615247964859, "learning_rate": 5.2761542145375365e-05, "loss": 0.0082, "step": 50830 }, { "grad_norm": 0.05889144539833069, "learning_rate": 5.274503253486421e-05, "loss": 0.0053, "step": 50840 }, { "grad_norm": 0.09321251511573792, "learning_rate": 5.272852262416046e-05, "loss": 0.004, "step": 50850 }, { "grad_norm": 0.09610024839639664, "learning_rate": 5.2712012415069555e-05, "loss": 0.0085, "step": 50860 }, { "grad_norm": 0.06445679813623428, "learning_rate": 5.269550190939705e-05, "loss": 0.0057, "step": 50870 }, { "grad_norm": 0.09690942615270615, "learning_rate": 5.267899110894852e-05, "loss": 0.008, "step": 50880 }, { "grad_norm": 0.06358157098293304, "learning_rate": 5.266248001552955e-05, "loss": 0.0054, "step": 50890 }, { "grad_norm": 0.11621928960084915, "learning_rate": 5.264596863094575e-05, "loss": 0.0075, "step": 50900 }, { "grad_norm": 0.16906046867370605, "learning_rate": 5.262945695700282e-05, "loss": 0.0064, "step": 50910 }, { "grad_norm": 0.10587430745363235, "learning_rate": 5.261294499550643e-05, "loss": 0.0063, "step": 50920 }, { "grad_norm": 0.15401335060596466, "learning_rate": 5.25964327482623e-05, "loss": 0.0062, "step": 50930 }, { "grad_norm": 0.11072362959384918, "learning_rate": 5.257992021707617e-05, "loss": 0.0071, "step": 50940 }, { "grad_norm": 0.0653378888964653, "learning_rate": 5.256340740375387e-05, "loss": 0.0067, "step": 50950 }, { "grad_norm": 0.07433189451694489, "learning_rate": 5.254689431010117e-05, "loss": 0.0044, "step": 50960 }, { "grad_norm": 0.08748501539230347, "learning_rate": 5.253038093792395e-05, "loss": 0.0048, "step": 50970 }, { "grad_norm": 0.06306059658527374, "learning_rate": 5.251386728902806e-05, "loss": 0.0039, "step": 50980 }, { "grad_norm": 0.0626496896147728, "learning_rate": 5.2497353365219446e-05, "loss": 0.0042, "step": 50990 }, { "grad_norm": 0.11103999614715576, "learning_rate": 5.2480839168304e-05, "loss": 0.005, "step": 51000 }, { "grad_norm": 0.21416178345680237, "learning_rate": 5.246432470008773e-05, "loss": 0.0046, "step": 51010 }, { "grad_norm": 0.09966856986284256, "learning_rate": 5.2447809962376616e-05, "loss": 0.0067, "step": 51020 }, { "grad_norm": 0.0941130518913269, "learning_rate": 5.243129495697666e-05, "loss": 0.0054, "step": 51030 }, { "grad_norm": 0.09412357211112976, "learning_rate": 5.241477968569397e-05, "loss": 0.0044, "step": 51040 }, { "grad_norm": 0.07151377201080322, "learning_rate": 5.23982641503346e-05, "loss": 0.0061, "step": 51050 }, { "grad_norm": 0.18438820540905, "learning_rate": 5.238174835270465e-05, "loss": 0.0077, "step": 51060 }, { "grad_norm": 0.15079525113105774, "learning_rate": 5.23652322946103e-05, "loss": 0.0075, "step": 51070 }, { "grad_norm": 0.17836928367614746, "learning_rate": 5.234871597785771e-05, "loss": 0.0066, "step": 51080 }, { "grad_norm": 0.10165293514728546, "learning_rate": 5.233219940425306e-05, "loss": 0.0052, "step": 51090 }, { "grad_norm": 0.15084519982337952, "learning_rate": 5.2315682575602576e-05, "loss": 0.0084, "step": 51100 }, { "grad_norm": 0.09185989946126938, "learning_rate": 5.229916549371254e-05, "loss": 0.006, "step": 51110 }, { "grad_norm": 0.16809901595115662, "learning_rate": 5.228264816038924e-05, "loss": 0.0067, "step": 51120 }, { "grad_norm": 0.08750639110803604, "learning_rate": 5.2266130577438935e-05, "loss": 0.0062, "step": 51130 }, { "grad_norm": 0.1358499377965927, "learning_rate": 5.224961274666802e-05, "loss": 0.0057, "step": 51140 }, { "grad_norm": 0.07496633380651474, "learning_rate": 5.223309466988283e-05, "loss": 0.0054, "step": 51150 }, { "grad_norm": 0.15922442078590393, "learning_rate": 5.221657634888976e-05, "loss": 0.0067, "step": 51160 }, { "grad_norm": 0.049273353070020676, "learning_rate": 5.2200057785495235e-05, "loss": 0.0046, "step": 51170 }, { "grad_norm": 0.15561027824878693, "learning_rate": 5.2183538981505706e-05, "loss": 0.0061, "step": 51180 }, { "grad_norm": 0.07006394863128662, "learning_rate": 5.216701993872762e-05, "loss": 0.0072, "step": 51190 }, { "grad_norm": 0.17638668417930603, "learning_rate": 5.215050065896752e-05, "loss": 0.0067, "step": 51200 }, { "grad_norm": 0.1918557733297348, "learning_rate": 5.2133981144031875e-05, "loss": 0.0054, "step": 51210 }, { "grad_norm": 0.1039872020483017, "learning_rate": 5.211746139572727e-05, "loss": 0.0052, "step": 51220 }, { "grad_norm": 0.0828794464468956, "learning_rate": 5.210094141586026e-05, "loss": 0.005, "step": 51230 }, { "grad_norm": 0.12209328263998032, "learning_rate": 5.2084421206237476e-05, "loss": 0.0073, "step": 51240 }, { "grad_norm": 0.11244485527276993, "learning_rate": 5.20679007686655e-05, "loss": 0.0067, "step": 51250 }, { "grad_norm": 0.2162790447473526, "learning_rate": 5.205138010495102e-05, "loss": 0.0077, "step": 51260 }, { "grad_norm": 0.09329605102539062, "learning_rate": 5.20348592169007e-05, "loss": 0.0066, "step": 51270 }, { "grad_norm": 0.12092269957065582, "learning_rate": 5.201833810632123e-05, "loss": 0.0063, "step": 51280 }, { "grad_norm": 0.07317271083593369, "learning_rate": 5.200181677501933e-05, "loss": 0.0049, "step": 51290 }, { "grad_norm": 0.09144478291273117, "learning_rate": 5.1985295224801777e-05, "loss": 0.0064, "step": 51300 }, { "grad_norm": 0.07564769685268402, "learning_rate": 5.196877345747531e-05, "loss": 0.0064, "step": 51310 }, { "grad_norm": 0.07984443008899689, "learning_rate": 5.195225147484674e-05, "loss": 0.0062, "step": 51320 }, { "grad_norm": 0.05397506430745125, "learning_rate": 5.193572927872288e-05, "loss": 0.0056, "step": 51330 }, { "grad_norm": 0.06410893052816391, "learning_rate": 5.191920687091059e-05, "loss": 0.0058, "step": 51340 }, { "grad_norm": 0.1012912318110466, "learning_rate": 5.1902684253216695e-05, "loss": 0.0058, "step": 51350 }, { "grad_norm": 0.08567503839731216, "learning_rate": 5.1886161427448134e-05, "loss": 0.0062, "step": 51360 }, { "grad_norm": 0.12996582686901093, "learning_rate": 5.186963839541179e-05, "loss": 0.0067, "step": 51370 }, { "grad_norm": 0.11842352896928787, "learning_rate": 5.185311515891459e-05, "loss": 0.0056, "step": 51380 }, { "grad_norm": 0.09573953598737717, "learning_rate": 5.1836591719763516e-05, "loss": 0.0062, "step": 51390 }, { "grad_norm": 0.06813561171293259, "learning_rate": 5.182006807976554e-05, "loss": 0.006, "step": 51400 }, { "grad_norm": 0.08210863173007965, "learning_rate": 5.180354424072763e-05, "loss": 0.0073, "step": 51410 }, { "grad_norm": 0.06986615806818008, "learning_rate": 5.178702020445686e-05, "loss": 0.0059, "step": 51420 }, { "grad_norm": 0.08954325318336487, "learning_rate": 5.177049597276024e-05, "loss": 0.0065, "step": 51430 }, { "grad_norm": 0.09360402077436447, "learning_rate": 5.175397154744482e-05, "loss": 0.0052, "step": 51440 }, { "grad_norm": 0.12521876394748688, "learning_rate": 5.1737446930317736e-05, "loss": 0.0074, "step": 51450 }, { "grad_norm": 0.11602035164833069, "learning_rate": 5.172092212318604e-05, "loss": 0.006, "step": 51460 }, { "grad_norm": 0.12233693897724152, "learning_rate": 5.170439712785692e-05, "loss": 0.0065, "step": 51470 }, { "grad_norm": 0.14897535741329193, "learning_rate": 5.1687871946137467e-05, "loss": 0.0083, "step": 51480 }, { "grad_norm": 0.11581439524888992, "learning_rate": 5.16713465798349e-05, "loss": 0.0048, "step": 51490 }, { "grad_norm": 0.07381191849708557, "learning_rate": 5.165482103075637e-05, "loss": 0.0055, "step": 51500 }, { "grad_norm": 0.05664912238717079, "learning_rate": 5.163829530070909e-05, "loss": 0.0033, "step": 51510 }, { "grad_norm": 0.09427446126937866, "learning_rate": 5.162176939150032e-05, "loss": 0.0056, "step": 51520 }, { "grad_norm": 0.08432378619909286, "learning_rate": 5.160524330493728e-05, "loss": 0.0063, "step": 51530 }, { "grad_norm": 0.05964028462767601, "learning_rate": 5.158871704282724e-05, "loss": 0.0056, "step": 51540 }, { "grad_norm": 0.07154712826013565, "learning_rate": 5.15721906069775e-05, "loss": 0.0047, "step": 51550 }, { "grad_norm": 0.05227944254875183, "learning_rate": 5.155566399919536e-05, "loss": 0.0053, "step": 51560 }, { "grad_norm": 0.13457907736301422, "learning_rate": 5.153913722128813e-05, "loss": 0.0073, "step": 51570 }, { "grad_norm": 0.07670368254184723, "learning_rate": 5.1522610275063176e-05, "loss": 0.0046, "step": 51580 }, { "grad_norm": 0.16701731085777283, "learning_rate": 5.150608316232786e-05, "loss": 0.006, "step": 51590 }, { "grad_norm": 0.1129765659570694, "learning_rate": 5.148955588488955e-05, "loss": 0.0079, "step": 51600 }, { "grad_norm": 0.07068251073360443, "learning_rate": 5.147302844455564e-05, "loss": 0.0047, "step": 51610 }, { "grad_norm": 0.13214105367660522, "learning_rate": 5.145650084313356e-05, "loss": 0.0053, "step": 51620 }, { "grad_norm": 0.10721897333860397, "learning_rate": 5.1439973082430734e-05, "loss": 0.006, "step": 51630 }, { "grad_norm": 0.1668797731399536, "learning_rate": 5.142344516425461e-05, "loss": 0.0073, "step": 51640 }, { "grad_norm": 0.1131814569234848, "learning_rate": 5.1406917090412673e-05, "loss": 0.0064, "step": 51650 }, { "grad_norm": 0.13781948387622833, "learning_rate": 5.1390388862712404e-05, "loss": 0.0061, "step": 51660 }, { "grad_norm": 0.09041334688663483, "learning_rate": 5.137386048296129e-05, "loss": 0.0052, "step": 51670 }, { "grad_norm": 0.05528992414474487, "learning_rate": 5.135733195296686e-05, "loss": 0.0075, "step": 51680 }, { "grad_norm": 0.15017034113407135, "learning_rate": 5.134080327453664e-05, "loss": 0.0062, "step": 51690 }, { "grad_norm": 0.08398114889860153, "learning_rate": 5.1324274449478204e-05, "loss": 0.0042, "step": 51700 }, { "grad_norm": 0.06454594433307648, "learning_rate": 5.130774547959909e-05, "loss": 0.004, "step": 51710 }, { "grad_norm": 0.09350671619176865, "learning_rate": 5.129121636670692e-05, "loss": 0.005, "step": 51720 }, { "grad_norm": 0.07092354446649551, "learning_rate": 5.1274687112609246e-05, "loss": 0.0037, "step": 51730 }, { "grad_norm": 0.059617385268211365, "learning_rate": 5.125815771911372e-05, "loss": 0.0067, "step": 51740 }, { "grad_norm": 0.14808395504951477, "learning_rate": 5.1241628188027956e-05, "loss": 0.0062, "step": 51750 }, { "grad_norm": 0.06410329788923264, "learning_rate": 5.1225098521159585e-05, "loss": 0.0065, "step": 51760 }, { "grad_norm": 0.10285558551549911, "learning_rate": 5.12085687203163e-05, "loss": 0.0072, "step": 51770 }, { "grad_norm": 0.06423057615756989, "learning_rate": 5.119203878730575e-05, "loss": 0.0073, "step": 51780 }, { "grad_norm": 0.0895053893327713, "learning_rate": 5.117550872393564e-05, "loss": 0.0058, "step": 51790 }, { "grad_norm": 0.05951458215713501, "learning_rate": 5.1158978532013645e-05, "loss": 0.0077, "step": 51800 }, { "grad_norm": 0.06517022103071213, "learning_rate": 5.114244821334751e-05, "loss": 0.0039, "step": 51810 }, { "grad_norm": 0.10542982816696167, "learning_rate": 5.112591776974497e-05, "loss": 0.0071, "step": 51820 }, { "grad_norm": 0.08073296397924423, "learning_rate": 5.110938720301374e-05, "loss": 0.0064, "step": 51830 }, { "grad_norm": 0.09063152968883514, "learning_rate": 5.1092856514961595e-05, "loss": 0.0075, "step": 51840 }, { "grad_norm": 0.07842101901769638, "learning_rate": 5.107632570739631e-05, "loss": 0.0065, "step": 51850 }, { "grad_norm": 0.0851413905620575, "learning_rate": 5.1059794782125645e-05, "loss": 0.0061, "step": 51860 }, { "grad_norm": 0.06560453772544861, "learning_rate": 5.104326374095744e-05, "loss": 0.0051, "step": 51870 }, { "grad_norm": 0.07898537069559097, "learning_rate": 5.102673258569947e-05, "loss": 0.0054, "step": 51880 }, { "grad_norm": 0.10349540412425995, "learning_rate": 5.101020131815956e-05, "loss": 0.0052, "step": 51890 }, { "grad_norm": 0.10804913938045502, "learning_rate": 5.099366994014556e-05, "loss": 0.0053, "step": 51900 }, { "grad_norm": 0.05097607150673866, "learning_rate": 5.097713845346532e-05, "loss": 0.0052, "step": 51910 }, { "grad_norm": 0.0790918841958046, "learning_rate": 5.096060685992668e-05, "loss": 0.0061, "step": 51920 }, { "grad_norm": 0.08334661275148392, "learning_rate": 5.0944075161337514e-05, "loss": 0.0048, "step": 51930 }, { "grad_norm": 0.0931013971567154, "learning_rate": 5.09275433595057e-05, "loss": 0.0052, "step": 51940 }, { "grad_norm": 0.06423337012529373, "learning_rate": 5.0911011456239157e-05, "loss": 0.0058, "step": 51950 }, { "grad_norm": 0.08711778372526169, "learning_rate": 5.089447945334574e-05, "loss": 0.0037, "step": 51960 }, { "grad_norm": 0.08622202277183533, "learning_rate": 5.087794735263343e-05, "loss": 0.0042, "step": 51970 }, { "grad_norm": 0.06502759456634521, "learning_rate": 5.086141515591011e-05, "loss": 0.0051, "step": 51980 }, { "grad_norm": 0.07498809695243835, "learning_rate": 5.084488286498371e-05, "loss": 0.0084, "step": 51990 }, { "grad_norm": 0.06954757869243622, "learning_rate": 5.08283504816622e-05, "loss": 0.0058, "step": 52000 }, { "grad_norm": 0.054595496505498886, "learning_rate": 5.081181800775353e-05, "loss": 0.0047, "step": 52010 }, { "grad_norm": 0.06381961703300476, "learning_rate": 5.079528544506565e-05, "loss": 0.0063, "step": 52020 }, { "grad_norm": 0.0712890475988388, "learning_rate": 5.077875279540657e-05, "loss": 0.0054, "step": 52030 }, { "grad_norm": 0.06826768070459366, "learning_rate": 5.076222006058424e-05, "loss": 0.0061, "step": 52040 }, { "grad_norm": 0.0647805705666542, "learning_rate": 5.074568724240667e-05, "loss": 0.0041, "step": 52050 }, { "grad_norm": 0.06537940353155136, "learning_rate": 5.072915434268187e-05, "loss": 0.0062, "step": 52060 }, { "grad_norm": 0.0441078320145607, "learning_rate": 5.071262136321785e-05, "loss": 0.0047, "step": 52070 }, { "grad_norm": 0.07776955515146255, "learning_rate": 5.069608830582262e-05, "loss": 0.0052, "step": 52080 }, { "grad_norm": 0.05763189122080803, "learning_rate": 5.067955517230423e-05, "loss": 0.0048, "step": 52090 }, { "grad_norm": 0.09594716131687164, "learning_rate": 5.06630219644707e-05, "loss": 0.0057, "step": 52100 }, { "grad_norm": 0.09826280921697617, "learning_rate": 5.064648868413008e-05, "loss": 0.0047, "step": 52110 }, { "grad_norm": 0.0822327733039856, "learning_rate": 5.062995533309044e-05, "loss": 0.005, "step": 52120 }, { "grad_norm": 0.057531554251909256, "learning_rate": 5.061342191315983e-05, "loss": 0.0047, "step": 52130 }, { "grad_norm": 0.12050817161798477, "learning_rate": 5.0596888426146325e-05, "loss": 0.0048, "step": 52140 }, { "grad_norm": 0.059439219534397125, "learning_rate": 5.058035487385798e-05, "loss": 0.0054, "step": 52150 }, { "grad_norm": 0.0693645253777504, "learning_rate": 5.0563821258102915e-05, "loss": 0.0054, "step": 52160 }, { "grad_norm": 0.061384525150060654, "learning_rate": 5.05472875806892e-05, "loss": 0.0038, "step": 52170 }, { "grad_norm": 0.05686241388320923, "learning_rate": 5.053075384342494e-05, "loss": 0.0041, "step": 52180 }, { "grad_norm": 0.08048392832279205, "learning_rate": 5.0514220048118234e-05, "loss": 0.0062, "step": 52190 }, { "grad_norm": 0.09233824163675308, "learning_rate": 5.049768619657721e-05, "loss": 0.0059, "step": 52200 }, { "grad_norm": 0.07255811244249344, "learning_rate": 5.048115229060996e-05, "loss": 0.0042, "step": 52210 }, { "grad_norm": 0.06171483173966408, "learning_rate": 5.046461833202462e-05, "loss": 0.0062, "step": 52220 }, { "grad_norm": 0.0633125975728035, "learning_rate": 5.044808432262932e-05, "loss": 0.0056, "step": 52230 }, { "grad_norm": 0.14275093376636505, "learning_rate": 5.043155026423219e-05, "loss": 0.0065, "step": 52240 }, { "grad_norm": 0.06361223012208939, "learning_rate": 5.041501615864137e-05, "loss": 0.0049, "step": 52250 }, { "grad_norm": 0.13556988537311554, "learning_rate": 5.039848200766502e-05, "loss": 0.0054, "step": 52260 }, { "grad_norm": 0.06750934571027756, "learning_rate": 5.038194781311125e-05, "loss": 0.0065, "step": 52270 }, { "grad_norm": 0.05960704758763313, "learning_rate": 5.0365413576788264e-05, "loss": 0.0059, "step": 52280 }, { "grad_norm": 0.05997982993721962, "learning_rate": 5.0348879300504194e-05, "loss": 0.0046, "step": 52290 }, { "grad_norm": 0.07168794423341751, "learning_rate": 5.033234498606721e-05, "loss": 0.0069, "step": 52300 }, { "grad_norm": 0.06721783429384232, "learning_rate": 5.0315810635285476e-05, "loss": 0.0045, "step": 52310 }, { "grad_norm": 0.07879156619310379, "learning_rate": 5.0299276249967164e-05, "loss": 0.0044, "step": 52320 }, { "grad_norm": 0.07372487336397171, "learning_rate": 5.0282741831920454e-05, "loss": 0.0057, "step": 52330 }, { "grad_norm": 0.06904930621385574, "learning_rate": 5.026620738295351e-05, "loss": 0.0061, "step": 52340 }, { "grad_norm": 0.05611816793680191, "learning_rate": 5.024967290487455e-05, "loss": 0.0057, "step": 52350 }, { "grad_norm": 0.051015447825193405, "learning_rate": 5.023313839949172e-05, "loss": 0.0068, "step": 52360 }, { "grad_norm": 0.08957571536302567, "learning_rate": 5.0216603868613224e-05, "loss": 0.0061, "step": 52370 }, { "grad_norm": 0.07165288180112839, "learning_rate": 5.020006931404726e-05, "loss": 0.0048, "step": 52380 }, { "grad_norm": 0.0771292969584465, "learning_rate": 5.0183534737602014e-05, "loss": 0.0043, "step": 52390 }, { "grad_norm": 0.05180674046278, "learning_rate": 5.0167000141085686e-05, "loss": 0.0043, "step": 52400 }, { "grad_norm": 0.04882701486349106, "learning_rate": 5.015046552630647e-05, "loss": 0.0104, "step": 52410 }, { "grad_norm": 0.06791996210813522, "learning_rate": 5.013393089507258e-05, "loss": 0.0066, "step": 52420 }, { "grad_norm": 0.14376239478588104, "learning_rate": 5.011739624919219e-05, "loss": 0.0066, "step": 52430 }, { "grad_norm": 0.09591423720121384, "learning_rate": 5.0100861590473545e-05, "loss": 0.0052, "step": 52440 }, { "grad_norm": 0.0647212490439415, "learning_rate": 5.008432692072482e-05, "loss": 0.0082, "step": 52450 }, { "grad_norm": 0.12610222399234772, "learning_rate": 5.006779224175423e-05, "loss": 0.0063, "step": 52460 }, { "grad_norm": 0.11355756968259811, "learning_rate": 5.005125755536998e-05, "loss": 0.0077, "step": 52470 }, { "grad_norm": 0.07202136516571045, "learning_rate": 5.003472286338029e-05, "loss": 0.0067, "step": 52480 }, { "grad_norm": 0.07010280340909958, "learning_rate": 5.001818816759335e-05, "loss": 0.0067, "step": 52490 }, { "grad_norm": 0.14964659512043, "learning_rate": 5.000165346981738e-05, "loss": 0.0066, "step": 52500 }, { "grad_norm": 0.09817318618297577, "learning_rate": 4.9985118771860595e-05, "loss": 0.0057, "step": 52510 }, { "grad_norm": 0.09030035138130188, "learning_rate": 4.996858407553119e-05, "loss": 0.0052, "step": 52520 }, { "grad_norm": 0.07590778917074203, "learning_rate": 4.99520493826374e-05, "loss": 0.0058, "step": 52530 }, { "grad_norm": 0.06842516362667084, "learning_rate": 4.9935514694987384e-05, "loss": 0.0049, "step": 52540 }, { "grad_norm": 0.07955500483512878, "learning_rate": 4.991898001438941e-05, "loss": 0.0068, "step": 52550 }, { "grad_norm": 0.08786193281412125, "learning_rate": 4.990244534265166e-05, "loss": 0.0066, "step": 52560 }, { "grad_norm": 0.09177850931882858, "learning_rate": 4.988591068158234e-05, "loss": 0.006, "step": 52570 }, { "grad_norm": 0.09352252632379532, "learning_rate": 4.986937603298965e-05, "loss": 0.0051, "step": 52580 }, { "grad_norm": 0.10492177307605743, "learning_rate": 4.985284139868181e-05, "loss": 0.0045, "step": 52590 }, { "grad_norm": 0.09357631951570511, "learning_rate": 4.9836306780467e-05, "loss": 0.0062, "step": 52600 }, { "grad_norm": 0.059729889035224915, "learning_rate": 4.981977218015343e-05, "loss": 0.0063, "step": 52610 }, { "grad_norm": 0.07651776820421219, "learning_rate": 4.980323759954932e-05, "loss": 0.0048, "step": 52620 }, { "grad_norm": 0.06703239679336548, "learning_rate": 4.978670304046284e-05, "loss": 0.0051, "step": 52630 }, { "grad_norm": 0.07430112361907959, "learning_rate": 4.9770168504702195e-05, "loss": 0.0045, "step": 52640 }, { "grad_norm": 0.09302330017089844, "learning_rate": 4.975363399407559e-05, "loss": 0.0042, "step": 52650 }, { "grad_norm": 0.04652919992804527, "learning_rate": 4.9737099510391194e-05, "loss": 0.0034, "step": 52660 }, { "grad_norm": 0.07553795725107193, "learning_rate": 4.972056505545719e-05, "loss": 0.0038, "step": 52670 }, { "grad_norm": 0.05845707282423973, "learning_rate": 4.9704030631081775e-05, "loss": 0.0056, "step": 52680 }, { "grad_norm": 0.07563527673482895, "learning_rate": 4.968749623907313e-05, "loss": 0.0037, "step": 52690 }, { "grad_norm": 0.06882845610380173, "learning_rate": 4.967096188123943e-05, "loss": 0.0047, "step": 52700 }, { "grad_norm": 0.06145431101322174, "learning_rate": 4.965442755938884e-05, "loss": 0.0044, "step": 52710 }, { "grad_norm": 0.08521214127540588, "learning_rate": 4.963789327532954e-05, "loss": 0.0072, "step": 52720 }, { "grad_norm": 0.10909277945756912, "learning_rate": 4.962135903086968e-05, "loss": 0.0043, "step": 52730 }, { "grad_norm": 0.06500539183616638, "learning_rate": 4.960482482781741e-05, "loss": 0.0072, "step": 52740 }, { "grad_norm": 0.1093306913971901, "learning_rate": 4.958829066798092e-05, "loss": 0.0048, "step": 52750 }, { "grad_norm": 0.09195739030838013, "learning_rate": 4.957175655316835e-05, "loss": 0.0088, "step": 52760 }, { "grad_norm": 0.07335206121206284, "learning_rate": 4.9555222485187834e-05, "loss": 0.0061, "step": 52770 }, { "grad_norm": 0.09241339564323425, "learning_rate": 4.953868846584752e-05, "loss": 0.0044, "step": 52780 }, { "grad_norm": 0.10391972213983536, "learning_rate": 4.9522154496955544e-05, "loss": 0.0042, "step": 52790 }, { "grad_norm": 0.13260993361473083, "learning_rate": 4.950562058032001e-05, "loss": 0.0075, "step": 52800 }, { "grad_norm": 0.12876586616039276, "learning_rate": 4.948908671774909e-05, "loss": 0.0052, "step": 52810 }, { "grad_norm": 0.06037571653723717, "learning_rate": 4.947255291105087e-05, "loss": 0.0058, "step": 52820 }, { "grad_norm": 0.06405401974916458, "learning_rate": 4.945601916203348e-05, "loss": 0.0068, "step": 52830 }, { "grad_norm": 0.11089044064283371, "learning_rate": 4.9439485472505016e-05, "loss": 0.0051, "step": 52840 }, { "grad_norm": 0.057115040719509125, "learning_rate": 4.942295184427356e-05, "loss": 0.0056, "step": 52850 }, { "grad_norm": 0.07924337685108185, "learning_rate": 4.9406418279147234e-05, "loss": 0.0052, "step": 52860 }, { "grad_norm": 0.07486560940742493, "learning_rate": 4.9389884778934116e-05, "loss": 0.0059, "step": 52870 }, { "grad_norm": 0.10051867365837097, "learning_rate": 4.937335134544228e-05, "loss": 0.0072, "step": 52880 }, { "grad_norm": 0.09248325973749161, "learning_rate": 4.935681798047979e-05, "loss": 0.0051, "step": 52890 }, { "grad_norm": 0.07784145325422287, "learning_rate": 4.934028468585473e-05, "loss": 0.0037, "step": 52900 }, { "grad_norm": 0.231063574552536, "learning_rate": 4.932375146337514e-05, "loss": 0.0104, "step": 52910 }, { "grad_norm": 0.06238534674048424, "learning_rate": 4.930721831484908e-05, "loss": 0.0091, "step": 52920 }, { "grad_norm": 0.10704046487808228, "learning_rate": 4.929068524208456e-05, "loss": 0.0072, "step": 52930 }, { "grad_norm": 0.06588814407587051, "learning_rate": 4.927415224688965e-05, "loss": 0.0054, "step": 52940 }, { "grad_norm": 0.06813472509384155, "learning_rate": 4.9257619331072375e-05, "loss": 0.0056, "step": 52950 }, { "grad_norm": 0.13877427577972412, "learning_rate": 4.924108649644072e-05, "loss": 0.0046, "step": 52960 }, { "grad_norm": 0.10572914034128189, "learning_rate": 4.922455374480271e-05, "loss": 0.0064, "step": 52970 }, { "grad_norm": 0.10518216341733932, "learning_rate": 4.920802107796634e-05, "loss": 0.0065, "step": 52980 }, { "grad_norm": 0.11528097838163376, "learning_rate": 4.9191488497739564e-05, "loss": 0.0067, "step": 52990 }, { "grad_norm": 0.11252384632825851, "learning_rate": 4.917495600593042e-05, "loss": 0.0047, "step": 53000 }, { "grad_norm": 0.06805615872144699, "learning_rate": 4.915842360434684e-05, "loss": 0.005, "step": 53010 }, { "grad_norm": 0.08099635690450668, "learning_rate": 4.914189129479679e-05, "loss": 0.0055, "step": 53020 }, { "grad_norm": 0.11239262670278549, "learning_rate": 4.912535907908822e-05, "loss": 0.0048, "step": 53030 }, { "grad_norm": 0.06854328513145447, "learning_rate": 4.910882695902907e-05, "loss": 0.0063, "step": 53040 }, { "grad_norm": 0.06833557784557343, "learning_rate": 4.9092294936427234e-05, "loss": 0.0046, "step": 53050 }, { "grad_norm": 0.06720728427171707, "learning_rate": 4.9075763013090685e-05, "loss": 0.007, "step": 53060 }, { "grad_norm": 0.09130488336086273, "learning_rate": 4.9059231190827305e-05, "loss": 0.0066, "step": 53070 }, { "grad_norm": 0.0811457559466362, "learning_rate": 4.904269947144499e-05, "loss": 0.0046, "step": 53080 }, { "grad_norm": 0.06077462434768677, "learning_rate": 4.902616785675161e-05, "loss": 0.0102, "step": 53090 }, { "grad_norm": 0.10844738036394119, "learning_rate": 4.900963634855505e-05, "loss": 0.0048, "step": 53100 }, { "grad_norm": 0.06866439431905746, "learning_rate": 4.8993104948663173e-05, "loss": 0.0052, "step": 53110 }, { "grad_norm": 0.07646557688713074, "learning_rate": 4.8976573658883823e-05, "loss": 0.006, "step": 53120 }, { "grad_norm": 0.08881965279579163, "learning_rate": 4.896004248102484e-05, "loss": 0.0062, "step": 53130 }, { "grad_norm": 0.11473730206489563, "learning_rate": 4.8943511416894045e-05, "loss": 0.006, "step": 53140 }, { "grad_norm": 0.08464627712965012, "learning_rate": 4.892698046829926e-05, "loss": 0.0047, "step": 53150 }, { "grad_norm": 0.08945154398679733, "learning_rate": 4.8910449637048275e-05, "loss": 0.0053, "step": 53160 }, { "grad_norm": 0.1103629395365715, "learning_rate": 4.8893918924948884e-05, "loss": 0.0078, "step": 53170 }, { "grad_norm": 0.08693067729473114, "learning_rate": 4.8877388333808835e-05, "loss": 0.0053, "step": 53180 }, { "grad_norm": 0.055924952030181885, "learning_rate": 4.8860857865435925e-05, "loss": 0.0061, "step": 53190 }, { "grad_norm": 0.06382478028535843, "learning_rate": 4.884432752163789e-05, "loss": 0.0052, "step": 53200 }, { "grad_norm": 0.09293970465660095, "learning_rate": 4.8827797304222466e-05, "loss": 0.0053, "step": 53210 }, { "grad_norm": 0.09011328965425491, "learning_rate": 4.8811267214997366e-05, "loss": 0.0055, "step": 53220 }, { "grad_norm": 0.06724110245704651, "learning_rate": 4.879473725577029e-05, "loss": 0.0046, "step": 53230 }, { "grad_norm": 0.06912654638290405, "learning_rate": 4.877820742834891e-05, "loss": 0.0053, "step": 53240 }, { "grad_norm": 0.06687141209840775, "learning_rate": 4.8761677734540956e-05, "loss": 0.0034, "step": 53250 }, { "grad_norm": 0.0642881989479065, "learning_rate": 4.874514817615407e-05, "loss": 0.0039, "step": 53260 }, { "grad_norm": 0.1099601611495018, "learning_rate": 4.872861875499588e-05, "loss": 0.0054, "step": 53270 }, { "grad_norm": 0.07709970325231552, "learning_rate": 4.871208947287404e-05, "loss": 0.0055, "step": 53280 }, { "grad_norm": 0.1254010945558548, "learning_rate": 4.869556033159615e-05, "loss": 0.0074, "step": 53290 }, { "grad_norm": 0.11080963164567947, "learning_rate": 4.8679031332969835e-05, "loss": 0.0061, "step": 53300 }, { "grad_norm": 0.08444761484861374, "learning_rate": 4.8662502478802635e-05, "loss": 0.0058, "step": 53310 }, { "grad_norm": 0.1080499216914177, "learning_rate": 4.8645973770902176e-05, "loss": 0.0067, "step": 53320 }, { "grad_norm": 0.09133065491914749, "learning_rate": 4.862944521107599e-05, "loss": 0.0072, "step": 53330 }, { "grad_norm": 0.12015558779239655, "learning_rate": 4.861291680113162e-05, "loss": 0.0083, "step": 53340 }, { "grad_norm": 0.10430858284235, "learning_rate": 4.8596388542876563e-05, "loss": 0.0059, "step": 53350 }, { "grad_norm": 0.0692174881696701, "learning_rate": 4.857986043811836e-05, "loss": 0.0054, "step": 53360 }, { "grad_norm": 0.15710313618183136, "learning_rate": 4.856333248866448e-05, "loss": 0.0078, "step": 53370 }, { "grad_norm": 0.08448478579521179, "learning_rate": 4.85468046963224e-05, "loss": 0.006, "step": 53380 }, { "grad_norm": 0.08593381196260452, "learning_rate": 4.853027706289957e-05, "loss": 0.0067, "step": 53390 }, { "grad_norm": 0.06811469793319702, "learning_rate": 4.851374959020344e-05, "loss": 0.0053, "step": 53400 }, { "grad_norm": 0.14152993261814117, "learning_rate": 4.849722228004141e-05, "loss": 0.0061, "step": 53410 }, { "grad_norm": 0.048552025109529495, "learning_rate": 4.848069513422091e-05, "loss": 0.0058, "step": 53420 }, { "grad_norm": 0.08736321330070496, "learning_rate": 4.8464168154549294e-05, "loss": 0.0053, "step": 53430 }, { "grad_norm": 0.1067853718996048, "learning_rate": 4.844764134283392e-05, "loss": 0.0089, "step": 53440 }, { "grad_norm": 0.07060785591602325, "learning_rate": 4.843111470088218e-05, "loss": 0.0073, "step": 53450 }, { "grad_norm": 0.09542785584926605, "learning_rate": 4.841458823050137e-05, "loss": 0.0058, "step": 53460 }, { "grad_norm": 0.07085731625556946, "learning_rate": 4.8398061933498816e-05, "loss": 0.0062, "step": 53470 }, { "grad_norm": 0.04844629764556885, "learning_rate": 4.83815358116818e-05, "loss": 0.0048, "step": 53480 }, { "grad_norm": 0.057432882487773895, "learning_rate": 4.8365009866857584e-05, "loss": 0.0039, "step": 53490 }, { "grad_norm": 0.11843806505203247, "learning_rate": 4.834848410083342e-05, "loss": 0.0049, "step": 53500 }, { "grad_norm": 0.12266328185796738, "learning_rate": 4.833195851541657e-05, "loss": 0.0053, "step": 53510 }, { "grad_norm": 0.12488958239555359, "learning_rate": 4.831543311241423e-05, "loss": 0.0043, "step": 53520 }, { "grad_norm": 0.06464771926403046, "learning_rate": 4.829890789363359e-05, "loss": 0.0049, "step": 53530 }, { "grad_norm": 0.08776939660310745, "learning_rate": 4.828238286088182e-05, "loss": 0.0042, "step": 53540 }, { "grad_norm": 0.06235726922750473, "learning_rate": 4.826585801596609e-05, "loss": 0.009, "step": 53550 }, { "grad_norm": 0.07412629574537277, "learning_rate": 4.8249333360693486e-05, "loss": 0.0065, "step": 53560 }, { "grad_norm": 0.07845770567655563, "learning_rate": 4.8232808896871185e-05, "loss": 0.003, "step": 53570 }, { "grad_norm": 0.07176254689693451, "learning_rate": 4.821628462630624e-05, "loss": 0.0057, "step": 53580 }, { "grad_norm": 0.06641321629285812, "learning_rate": 4.8199760550805735e-05, "loss": 0.0058, "step": 53590 }, { "grad_norm": 0.06733790785074234, "learning_rate": 4.818323667217669e-05, "loss": 0.0046, "step": 53600 }, { "grad_norm": 0.07498440891504288, "learning_rate": 4.8166712992226165e-05, "loss": 0.0036, "step": 53610 }, { "grad_norm": 0.08115088939666748, "learning_rate": 4.8150189512761154e-05, "loss": 0.0068, "step": 53620 }, { "grad_norm": 0.1823841780424118, "learning_rate": 4.813366623558862e-05, "loss": 0.0077, "step": 53630 }, { "grad_norm": 0.06454864889383316, "learning_rate": 4.811714316251554e-05, "loss": 0.0051, "step": 53640 }, { "grad_norm": 0.048185136169195175, "learning_rate": 4.810062029534886e-05, "loss": 0.006, "step": 53650 }, { "grad_norm": 0.07858516275882721, "learning_rate": 4.8084097635895505e-05, "loss": 0.0084, "step": 53660 }, { "grad_norm": 0.0777483806014061, "learning_rate": 4.8067575185962335e-05, "loss": 0.0075, "step": 53670 }, { "grad_norm": 0.11890769749879837, "learning_rate": 4.805105294735625e-05, "loss": 0.0061, "step": 53680 }, { "grad_norm": 0.07794001698493958, "learning_rate": 4.803453092188406e-05, "loss": 0.0045, "step": 53690 }, { "grad_norm": 0.0669962465763092, "learning_rate": 4.801800911135263e-05, "loss": 0.0053, "step": 53700 }, { "grad_norm": 0.06791883707046509, "learning_rate": 4.8001487517568744e-05, "loss": 0.0047, "step": 53710 }, { "grad_norm": 0.07741056382656097, "learning_rate": 4.798496614233918e-05, "loss": 0.006, "step": 53720 }, { "grad_norm": 0.0663587749004364, "learning_rate": 4.796844498747069e-05, "loss": 0.0073, "step": 53730 }, { "grad_norm": 0.06867263466119766, "learning_rate": 4.795192405477e-05, "loss": 0.0079, "step": 53740 }, { "grad_norm": 0.09325243532657623, "learning_rate": 4.79354033460438e-05, "loss": 0.0047, "step": 53750 }, { "grad_norm": 0.06064585596323013, "learning_rate": 4.79188828630988e-05, "loss": 0.0047, "step": 53760 }, { "grad_norm": 0.06383217871189117, "learning_rate": 4.790236260774165e-05, "loss": 0.0047, "step": 53770 }, { "grad_norm": 0.06310650706291199, "learning_rate": 4.788584258177896e-05, "loss": 0.0059, "step": 53780 }, { "grad_norm": 0.0857129767537117, "learning_rate": 4.7869322787017356e-05, "loss": 0.0046, "step": 53790 }, { "grad_norm": 0.0687716007232666, "learning_rate": 4.78528032252634e-05, "loss": 0.0054, "step": 53800 }, { "grad_norm": 0.05442124232649803, "learning_rate": 4.783628389832366e-05, "loss": 0.0057, "step": 53810 }, { "grad_norm": 0.077877938747406, "learning_rate": 4.7819764808004644e-05, "loss": 0.0059, "step": 53820 }, { "grad_norm": 0.08398570865392685, "learning_rate": 4.780324595611289e-05, "loss": 0.0051, "step": 53830 }, { "grad_norm": 0.08080442994832993, "learning_rate": 4.7786727344454854e-05, "loss": 0.0047, "step": 53840 }, { "grad_norm": 0.09588369727134705, "learning_rate": 4.777020897483697e-05, "loss": 0.0067, "step": 53850 }, { "grad_norm": 0.07611370086669922, "learning_rate": 4.775369084906569e-05, "loss": 0.0045, "step": 53860 }, { "grad_norm": 0.08441435545682907, "learning_rate": 4.773717296894742e-05, "loss": 0.0058, "step": 53870 }, { "grad_norm": 0.056048162281513214, "learning_rate": 4.772065533628848e-05, "loss": 0.0057, "step": 53880 }, { "grad_norm": 0.054244931787252426, "learning_rate": 4.7704137952895265e-05, "loss": 0.0041, "step": 53890 }, { "grad_norm": 0.10755371302366257, "learning_rate": 4.768762082057407e-05, "loss": 0.0079, "step": 53900 }, { "grad_norm": 0.08672771602869034, "learning_rate": 4.7671103941131186e-05, "loss": 0.0047, "step": 53910 }, { "grad_norm": 0.07876196503639221, "learning_rate": 4.765458731637287e-05, "loss": 0.0069, "step": 53920 }, { "grad_norm": 0.0780259370803833, "learning_rate": 4.763807094810537e-05, "loss": 0.0039, "step": 53930 }, { "grad_norm": 0.05533144623041153, "learning_rate": 4.762155483813487e-05, "loss": 0.0032, "step": 53940 }, { "grad_norm": 0.06885410845279694, "learning_rate": 4.760503898826754e-05, "loss": 0.0044, "step": 53950 }, { "grad_norm": 0.08462505042552948, "learning_rate": 4.758852340030957e-05, "loss": 0.0049, "step": 53960 }, { "grad_norm": 0.054680995643138885, "learning_rate": 4.757200807606705e-05, "loss": 0.0038, "step": 53970 }, { "grad_norm": 0.0908997431397438, "learning_rate": 4.755549301734607e-05, "loss": 0.0051, "step": 53980 }, { "grad_norm": 0.13264405727386475, "learning_rate": 4.753897822595271e-05, "loss": 0.0046, "step": 53990 }, { "grad_norm": 0.09651672095060349, "learning_rate": 4.7522463703692994e-05, "loss": 0.0045, "step": 54000 }, { "grad_norm": 0.10168299823999405, "learning_rate": 4.75059494523729e-05, "loss": 0.0052, "step": 54010 }, { "grad_norm": 0.08219775557518005, "learning_rate": 4.748943547379843e-05, "loss": 0.005, "step": 54020 }, { "grad_norm": 0.11617492139339447, "learning_rate": 4.7472921769775535e-05, "loss": 0.0055, "step": 54030 }, { "grad_norm": 0.0949246734380722, "learning_rate": 4.745640834211011e-05, "loss": 0.0045, "step": 54040 }, { "grad_norm": 0.0822535827755928, "learning_rate": 4.7439895192608054e-05, "loss": 0.0075, "step": 54050 }, { "grad_norm": 0.08140025287866592, "learning_rate": 4.7423382323075186e-05, "loss": 0.0045, "step": 54060 }, { "grad_norm": 0.08109921216964722, "learning_rate": 4.740686973531737e-05, "loss": 0.0044, "step": 54070 }, { "grad_norm": 0.09135286509990692, "learning_rate": 4.7390357431140375e-05, "loss": 0.0048, "step": 54080 }, { "grad_norm": 0.09404424577951431, "learning_rate": 4.737384541234998e-05, "loss": 0.0054, "step": 54090 }, { "grad_norm": 0.13180899620056152, "learning_rate": 4.735733368075189e-05, "loss": 0.0051, "step": 54100 }, { "grad_norm": 0.07465437799692154, "learning_rate": 4.734082223815182e-05, "loss": 0.0055, "step": 54110 }, { "grad_norm": 0.09540001302957535, "learning_rate": 4.7324311086355436e-05, "loss": 0.005, "step": 54120 }, { "grad_norm": 0.11103196442127228, "learning_rate": 4.730780022716836e-05, "loss": 0.005, "step": 54130 }, { "grad_norm": 0.08326145261526108, "learning_rate": 4.729128966239619e-05, "loss": 0.005, "step": 54140 }, { "grad_norm": 0.08104535192251205, "learning_rate": 4.7274779393844536e-05, "loss": 0.0071, "step": 54150 }, { "grad_norm": 0.07237560302019119, "learning_rate": 4.725826942331891e-05, "loss": 0.0061, "step": 54160 }, { "grad_norm": 0.10465901345014572, "learning_rate": 4.724175975262481e-05, "loss": 0.0053, "step": 54170 }, { "grad_norm": 0.06237513944506645, "learning_rate": 4.722525038356772e-05, "loss": 0.0055, "step": 54180 }, { "grad_norm": 0.09480655193328857, "learning_rate": 4.7208741317953075e-05, "loss": 0.0046, "step": 54190 }, { "grad_norm": 0.07902935892343521, "learning_rate": 4.7192232557586264e-05, "loss": 0.0038, "step": 54200 }, { "grad_norm": 0.05212654545903206, "learning_rate": 4.717572410427269e-05, "loss": 0.0056, "step": 54210 }, { "grad_norm": 0.07259418815374374, "learning_rate": 4.71592159598177e-05, "loss": 0.0046, "step": 54220 }, { "grad_norm": 0.07350248843431473, "learning_rate": 4.714270812602657e-05, "loss": 0.0056, "step": 54230 }, { "grad_norm": 0.11317128688097, "learning_rate": 4.712620060470458e-05, "loss": 0.0056, "step": 54240 }, { "grad_norm": 0.08383072912693024, "learning_rate": 4.7109693397656986e-05, "loss": 0.005, "step": 54250 }, { "grad_norm": 0.0646175965666771, "learning_rate": 4.709318650668894e-05, "loss": 0.0048, "step": 54260 }, { "grad_norm": 0.07817556709051132, "learning_rate": 4.7076679933605676e-05, "loss": 0.0059, "step": 54270 }, { "grad_norm": 0.09192083775997162, "learning_rate": 4.70601736802123e-05, "loss": 0.005, "step": 54280 }, { "grad_norm": 0.07145705819129944, "learning_rate": 4.704366774831391e-05, "loss": 0.006, "step": 54290 }, { "grad_norm": 0.07556382566690445, "learning_rate": 4.702716213971557e-05, "loss": 0.0059, "step": 54300 }, { "grad_norm": 0.05812934786081314, "learning_rate": 4.70106568562223e-05, "loss": 0.0064, "step": 54310 }, { "grad_norm": 0.09056005626916885, "learning_rate": 4.6994151899639105e-05, "loss": 0.0056, "step": 54320 }, { "grad_norm": 0.06355994194746017, "learning_rate": 4.697764727177093e-05, "loss": 0.0035, "step": 54330 }, { "grad_norm": 0.09370465576648712, "learning_rate": 4.696114297442272e-05, "loss": 0.006, "step": 54340 }, { "grad_norm": 0.1052653044462204, "learning_rate": 4.694463900939933e-05, "loss": 0.0072, "step": 54350 }, { "grad_norm": 0.09476678818464279, "learning_rate": 4.692813537850564e-05, "loss": 0.0057, "step": 54360 }, { "grad_norm": 0.07143112272024155, "learning_rate": 4.6911632083546445e-05, "loss": 0.0055, "step": 54370 }, { "grad_norm": 0.04430149868130684, "learning_rate": 4.6895129126326525e-05, "loss": 0.0051, "step": 54380 }, { "grad_norm": 0.07174410670995712, "learning_rate": 4.687862650865059e-05, "loss": 0.0044, "step": 54390 }, { "grad_norm": 0.06636503338813782, "learning_rate": 4.686212423232339e-05, "loss": 0.0057, "step": 54400 }, { "grad_norm": 0.0664130225777626, "learning_rate": 4.6845622299149565e-05, "loss": 0.0054, "step": 54410 }, { "grad_norm": 0.1527503877878189, "learning_rate": 4.682912071093374e-05, "loss": 0.0069, "step": 54420 }, { "grad_norm": 0.08972666412591934, "learning_rate": 4.681261946948052e-05, "loss": 0.0069, "step": 54430 }, { "grad_norm": 0.10075045377016068, "learning_rate": 4.679611857659443e-05, "loss": 0.0059, "step": 54440 }, { "grad_norm": 0.13082818686962128, "learning_rate": 4.6779618034079975e-05, "loss": 0.005, "step": 54450 }, { "grad_norm": 0.10053354501724243, "learning_rate": 4.6763117843741675e-05, "loss": 0.0051, "step": 54460 }, { "grad_norm": 0.07465943694114685, "learning_rate": 4.674661800738394e-05, "loss": 0.0041, "step": 54470 }, { "grad_norm": 0.060472264885902405, "learning_rate": 4.673011852681117e-05, "loss": 0.0062, "step": 54480 }, { "grad_norm": 0.07576721161603928, "learning_rate": 4.671361940382772e-05, "loss": 0.0057, "step": 54490 }, { "grad_norm": 0.07878870517015457, "learning_rate": 4.669712064023791e-05, "loss": 0.0078, "step": 54500 }, { "grad_norm": 0.04670009762048721, "learning_rate": 4.668062223784602e-05, "loss": 0.0053, "step": 54510 }, { "grad_norm": 0.0708826407790184, "learning_rate": 4.6664124198456274e-05, "loss": 0.0051, "step": 54520 }, { "grad_norm": 0.04601117968559265, "learning_rate": 4.6647626523872905e-05, "loss": 0.0051, "step": 54530 }, { "grad_norm": 0.05274408310651779, "learning_rate": 4.6631129215900066e-05, "loss": 0.0047, "step": 54540 }, { "grad_norm": 0.054622720927000046, "learning_rate": 4.661463227634187e-05, "loss": 0.0054, "step": 54550 }, { "grad_norm": 0.08048045635223389, "learning_rate": 4.659813570700238e-05, "loss": 0.0039, "step": 54560 }, { "grad_norm": 0.049778446555137634, "learning_rate": 4.658163950968568e-05, "loss": 0.0043, "step": 54570 }, { "grad_norm": 0.07041634619235992, "learning_rate": 4.6565143686195726e-05, "loss": 0.0056, "step": 54580 }, { "grad_norm": 0.07512505352497101, "learning_rate": 4.654864823833649e-05, "loss": 0.007, "step": 54590 }, { "grad_norm": 0.12262902408838272, "learning_rate": 4.65321531679119e-05, "loss": 0.005, "step": 54600 }, { "grad_norm": 0.08136535435914993, "learning_rate": 4.6515658476725834e-05, "loss": 0.0051, "step": 54610 }, { "grad_norm": 0.10392829030752182, "learning_rate": 4.649916416658212e-05, "loss": 0.0075, "step": 54620 }, { "grad_norm": 0.08593914657831192, "learning_rate": 4.648267023928455e-05, "loss": 0.0046, "step": 54630 }, { "grad_norm": 0.07963419705629349, "learning_rate": 4.646617669663687e-05, "loss": 0.0053, "step": 54640 }, { "grad_norm": 0.06908710300922394, "learning_rate": 4.644968354044277e-05, "loss": 0.0037, "step": 54650 }, { "grad_norm": 0.05107751861214638, "learning_rate": 4.643319077250596e-05, "loss": 0.006, "step": 54660 }, { "grad_norm": 0.06417737156152725, "learning_rate": 4.6416698394630055e-05, "loss": 0.0046, "step": 54670 }, { "grad_norm": 0.067062146961689, "learning_rate": 4.640020640861862e-05, "loss": 0.0045, "step": 54680 }, { "grad_norm": 0.06237225607037544, "learning_rate": 4.6383714816275195e-05, "loss": 0.0033, "step": 54690 }, { "grad_norm": 0.07878319919109344, "learning_rate": 4.636722361940329e-05, "loss": 0.0057, "step": 54700 }, { "grad_norm": 0.09595171362161636, "learning_rate": 4.635073281980632e-05, "loss": 0.0064, "step": 54710 }, { "grad_norm": 0.0469767302274704, "learning_rate": 4.633424241928775e-05, "loss": 0.0053, "step": 54720 }, { "grad_norm": 0.10702560096979141, "learning_rate": 4.6317752419650914e-05, "loss": 0.0061, "step": 54730 }, { "grad_norm": 0.0673726499080658, "learning_rate": 4.630126282269914e-05, "loss": 0.0041, "step": 54740 }, { "grad_norm": 0.05188357084989548, "learning_rate": 4.628477363023569e-05, "loss": 0.0047, "step": 54750 }, { "grad_norm": 0.07581186294555664, "learning_rate": 4.626828484406383e-05, "loss": 0.0046, "step": 54760 }, { "grad_norm": 0.08295218646526337, "learning_rate": 4.625179646598669e-05, "loss": 0.004, "step": 54770 }, { "grad_norm": 0.10271600633859634, "learning_rate": 4.6235308497807487e-05, "loss": 0.0053, "step": 54780 }, { "grad_norm": 0.045817166566848755, "learning_rate": 4.6218820941329276e-05, "loss": 0.0052, "step": 54790 }, { "grad_norm": 0.05754905194044113, "learning_rate": 4.620233379835513e-05, "loss": 0.0042, "step": 54800 }, { "grad_norm": 0.07280323654413223, "learning_rate": 4.618584707068804e-05, "loss": 0.0089, "step": 54810 }, { "grad_norm": 0.08000414073467255, "learning_rate": 4.616936076013099e-05, "loss": 0.0055, "step": 54820 }, { "grad_norm": 0.07353229820728302, "learning_rate": 4.615287486848688e-05, "loss": 0.0054, "step": 54830 }, { "grad_norm": 0.06600016355514526, "learning_rate": 4.6136389397558584e-05, "loss": 0.0044, "step": 54840 }, { "grad_norm": 0.08637452125549316, "learning_rate": 4.6119904349148935e-05, "loss": 0.0071, "step": 54850 }, { "grad_norm": 0.07363321632146835, "learning_rate": 4.6103419725060726e-05, "loss": 0.0052, "step": 54860 }, { "grad_norm": 0.05944423750042915, "learning_rate": 4.608693552709667e-05, "loss": 0.0043, "step": 54870 }, { "grad_norm": 0.12072350084781647, "learning_rate": 4.607045175705947e-05, "loss": 0.0047, "step": 54880 }, { "grad_norm": 0.07026880979537964, "learning_rate": 4.605396841675176e-05, "loss": 0.0046, "step": 54890 }, { "grad_norm": 0.09137940406799316, "learning_rate": 4.6037485507976094e-05, "loss": 0.0041, "step": 54900 }, { "grad_norm": 0.0767289474606514, "learning_rate": 4.602100303253509e-05, "loss": 0.0078, "step": 54910 }, { "grad_norm": 0.05011855065822601, "learning_rate": 4.600452099223122e-05, "loss": 0.0066, "step": 54920 }, { "grad_norm": 0.09863360971212387, "learning_rate": 4.598803938886691e-05, "loss": 0.0049, "step": 54930 }, { "grad_norm": 0.07256504148244858, "learning_rate": 4.597155822424459e-05, "loss": 0.0056, "step": 54940 }, { "grad_norm": 0.08006169646978378, "learning_rate": 4.59550775001666e-05, "loss": 0.0049, "step": 54950 }, { "grad_norm": 0.0450897179543972, "learning_rate": 4.5938597218435234e-05, "loss": 0.0054, "step": 54960 }, { "grad_norm": 0.06040681526064873, "learning_rate": 4.592211738085279e-05, "loss": 0.0056, "step": 54970 }, { "grad_norm": 0.11737664043903351, "learning_rate": 4.590563798922146e-05, "loss": 0.0053, "step": 54980 }, { "grad_norm": 0.05924132466316223, "learning_rate": 4.5889159045343404e-05, "loss": 0.0045, "step": 54990 }, { "grad_norm": 0.08722873777151108, "learning_rate": 4.5872680551020734e-05, "loss": 0.0054, "step": 55000 }, { "grad_norm": 0.059489283710718155, "learning_rate": 4.585620250805552e-05, "loss": 0.0059, "step": 55010 }, { "grad_norm": 0.06421510875225067, "learning_rate": 4.583972491824975e-05, "loss": 0.0059, "step": 55020 }, { "grad_norm": 0.09834785014390945, "learning_rate": 4.582324778340541e-05, "loss": 0.0047, "step": 55030 }, { "grad_norm": 0.07676604390144348, "learning_rate": 4.5806771105324424e-05, "loss": 0.0054, "step": 55040 }, { "grad_norm": 0.09040259569883347, "learning_rate": 4.579029488580865e-05, "loss": 0.0053, "step": 55050 }, { "grad_norm": 0.05375327542424202, "learning_rate": 4.5773819126659876e-05, "loss": 0.0059, "step": 55060 }, { "grad_norm": 0.08383533358573914, "learning_rate": 4.575734382967991e-05, "loss": 0.0052, "step": 55070 }, { "grad_norm": 0.05632239207625389, "learning_rate": 4.574086899667044e-05, "loss": 0.0059, "step": 55080 }, { "grad_norm": 0.11438053846359253, "learning_rate": 4.5724394629433106e-05, "loss": 0.0053, "step": 55090 }, { "grad_norm": 0.07343477755784988, "learning_rate": 4.570792072976958e-05, "loss": 0.0068, "step": 55100 }, { "grad_norm": 0.0568314753472805, "learning_rate": 4.569144729948138e-05, "loss": 0.0052, "step": 55110 }, { "grad_norm": 0.07510174810886383, "learning_rate": 4.5674974340370033e-05, "loss": 0.0042, "step": 55120 }, { "grad_norm": 0.05390086770057678, "learning_rate": 4.565850185423699e-05, "loss": 0.0061, "step": 55130 }, { "grad_norm": 0.10071416944265366, "learning_rate": 4.564202984288365e-05, "loss": 0.0047, "step": 55140 }, { "grad_norm": 0.09959768503904343, "learning_rate": 4.5625558308111356e-05, "loss": 0.0056, "step": 55150 }, { "grad_norm": 0.05981919914484024, "learning_rate": 4.560908725172145e-05, "loss": 0.0062, "step": 55160 }, { "grad_norm": 0.0862855538725853, "learning_rate": 4.559261667551516e-05, "loss": 0.005, "step": 55170 }, { "grad_norm": 0.10897234827280045, "learning_rate": 4.5576146581293685e-05, "loss": 0.006, "step": 55180 }, { "grad_norm": 0.06351712346076965, "learning_rate": 4.5559676970858164e-05, "loss": 0.0054, "step": 55190 }, { "grad_norm": 0.09303160011768341, "learning_rate": 4.5543207846009705e-05, "loss": 0.007, "step": 55200 }, { "grad_norm": 0.08937429636716843, "learning_rate": 4.552673920854933e-05, "loss": 0.0044, "step": 55210 }, { "grad_norm": 0.10350862145423889, "learning_rate": 4.5510271060278006e-05, "loss": 0.0044, "step": 55220 }, { "grad_norm": 0.1101856678724289, "learning_rate": 4.5493803402996704e-05, "loss": 0.0043, "step": 55230 }, { "grad_norm": 0.06710558384656906, "learning_rate": 4.5477336238506295e-05, "loss": 0.004, "step": 55240 }, { "grad_norm": 0.09559115022420883, "learning_rate": 4.546086956860759e-05, "loss": 0.0058, "step": 55250 }, { "grad_norm": 0.08703037351369858, "learning_rate": 4.5444403395101374e-05, "loss": 0.0047, "step": 55260 }, { "grad_norm": 0.0421416275203228, "learning_rate": 4.542793771978834e-05, "loss": 0.0034, "step": 55270 }, { "grad_norm": 0.08965595811605453, "learning_rate": 4.541147254446916e-05, "loss": 0.0047, "step": 55280 }, { "grad_norm": 0.0683586522936821, "learning_rate": 4.539500787094445e-05, "loss": 0.0076, "step": 55290 }, { "grad_norm": 0.0885317325592041, "learning_rate": 4.537854370101475e-05, "loss": 0.0052, "step": 55300 }, { "grad_norm": 0.08542056381702423, "learning_rate": 4.5362080036480566e-05, "loss": 0.0055, "step": 55310 }, { "grad_norm": 0.11007688194513321, "learning_rate": 4.534561687914235e-05, "loss": 0.0058, "step": 55320 }, { "grad_norm": 0.09092286229133606, "learning_rate": 4.5329154230800466e-05, "loss": 0.004, "step": 55330 }, { "grad_norm": 0.05944531038403511, "learning_rate": 4.531269209325525e-05, "loss": 0.0035, "step": 55340 }, { "grad_norm": 0.0935116782784462, "learning_rate": 4.529623046830696e-05, "loss": 0.0078, "step": 55350 }, { "grad_norm": 0.06917735934257507, "learning_rate": 4.5279769357755866e-05, "loss": 0.0044, "step": 55360 }, { "grad_norm": 0.11117275059223175, "learning_rate": 4.5263308763402084e-05, "loss": 0.0055, "step": 55370 }, { "grad_norm": 0.05412599444389343, "learning_rate": 4.524684868704574e-05, "loss": 0.0039, "step": 55380 }, { "grad_norm": 0.1208319365978241, "learning_rate": 4.523038913048688e-05, "loss": 0.0052, "step": 55390 }, { "grad_norm": 0.10369692742824554, "learning_rate": 4.521393009552549e-05, "loss": 0.0062, "step": 55400 }, { "grad_norm": 0.08711005747318268, "learning_rate": 4.5197471583961484e-05, "loss": 0.0057, "step": 55410 }, { "grad_norm": 0.049722522497177124, "learning_rate": 4.51810135975948e-05, "loss": 0.0041, "step": 55420 }, { "grad_norm": 0.06496550887823105, "learning_rate": 4.516455613822521e-05, "loss": 0.0046, "step": 55430 }, { "grad_norm": 0.10288765281438828, "learning_rate": 4.514809920765249e-05, "loss": 0.005, "step": 55440 }, { "grad_norm": 0.08945627510547638, "learning_rate": 4.5131642807676356e-05, "loss": 0.0052, "step": 55450 }, { "grad_norm": 0.06581491976976395, "learning_rate": 4.511518694009644e-05, "loss": 0.0042, "step": 55460 }, { "grad_norm": 0.05970697104930878, "learning_rate": 4.509873160671231e-05, "loss": 0.0042, "step": 55470 }, { "grad_norm": 0.10595677047967911, "learning_rate": 4.508227680932356e-05, "loss": 0.0041, "step": 55480 }, { "grad_norm": 0.10231820493936539, "learning_rate": 4.5065822549729605e-05, "loss": 0.005, "step": 55490 }, { "grad_norm": 0.058172326534986496, "learning_rate": 4.504936882972989e-05, "loss": 0.0064, "step": 55500 }, { "grad_norm": 0.19354256987571716, "learning_rate": 4.503291565112374e-05, "loss": 0.0046, "step": 55510 }, { "grad_norm": 0.1442229002714157, "learning_rate": 4.501646301571048e-05, "loss": 0.0055, "step": 55520 }, { "grad_norm": 0.06276878714561462, "learning_rate": 4.500001092528933e-05, "loss": 0.0034, "step": 55530 }, { "grad_norm": 0.0764424130320549, "learning_rate": 4.4983559381659455e-05, "loss": 0.0055, "step": 55540 }, { "grad_norm": 0.08130823820829391, "learning_rate": 4.496710838661999e-05, "loss": 0.0038, "step": 55550 }, { "grad_norm": 0.08872606605291367, "learning_rate": 4.495065794196999e-05, "loss": 0.0051, "step": 55560 }, { "grad_norm": 0.07745568454265594, "learning_rate": 4.4934208049508455e-05, "loss": 0.0039, "step": 55570 }, { "grad_norm": 0.0981985405087471, "learning_rate": 4.4917758711034303e-05, "loss": 0.0056, "step": 55580 }, { "grad_norm": 0.11393745988607407, "learning_rate": 4.490130992834642e-05, "loss": 0.0052, "step": 55590 }, { "grad_norm": 0.12489616125822067, "learning_rate": 4.488486170324359e-05, "loss": 0.0058, "step": 55600 }, { "grad_norm": 0.09317943453788757, "learning_rate": 4.486841403752463e-05, "loss": 0.0049, "step": 55610 }, { "grad_norm": 0.0542672835290432, "learning_rate": 4.4851966932988174e-05, "loss": 0.0043, "step": 55620 }, { "grad_norm": 0.09275759756565094, "learning_rate": 4.4835520391432894e-05, "loss": 0.0049, "step": 55630 }, { "grad_norm": 0.06373601406812668, "learning_rate": 4.481907441465733e-05, "loss": 0.0054, "step": 55640 }, { "grad_norm": 0.08454827964305878, "learning_rate": 4.480262900446e-05, "loss": 0.0063, "step": 55650 }, { "grad_norm": 0.06969519704580307, "learning_rate": 4.478618416263933e-05, "loss": 0.0037, "step": 55660 }, { "grad_norm": 0.09781406819820404, "learning_rate": 4.476973989099373e-05, "loss": 0.0055, "step": 55670 }, { "grad_norm": 0.05392000824213028, "learning_rate": 4.475329619132153e-05, "loss": 0.0054, "step": 55680 }, { "grad_norm": 0.10209514945745468, "learning_rate": 4.4736853065420956e-05, "loss": 0.0067, "step": 55690 }, { "grad_norm": 0.06577184051275253, "learning_rate": 4.472041051509023e-05, "loss": 0.0066, "step": 55700 }, { "grad_norm": 0.05990896746516228, "learning_rate": 4.470396854212746e-05, "loss": 0.0052, "step": 55710 }, { "grad_norm": 0.07259605079889297, "learning_rate": 4.468752714833072e-05, "loss": 0.0057, "step": 55720 }, { "grad_norm": 0.059212036430835724, "learning_rate": 4.467108633549803e-05, "loss": 0.004, "step": 55730 }, { "grad_norm": 0.06175730749964714, "learning_rate": 4.4654646105427335e-05, "loss": 0.0043, "step": 55740 }, { "grad_norm": 0.053599826991558075, "learning_rate": 4.463820645991651e-05, "loss": 0.0034, "step": 55750 }, { "grad_norm": 0.08209415525197983, "learning_rate": 4.462176740076334e-05, "loss": 0.0049, "step": 55760 }, { "grad_norm": 0.10763558745384216, "learning_rate": 4.4605328929765624e-05, "loss": 0.0053, "step": 55770 }, { "grad_norm": 0.07992292195558548, "learning_rate": 4.458889104872102e-05, "loss": 0.0077, "step": 55780 }, { "grad_norm": 0.06723136454820633, "learning_rate": 4.457245375942715e-05, "loss": 0.0063, "step": 55790 }, { "grad_norm": 0.0615483894944191, "learning_rate": 4.455601706368157e-05, "loss": 0.0047, "step": 55800 }, { "grad_norm": 0.07103653252124786, "learning_rate": 4.453958096328179e-05, "loss": 0.0049, "step": 55810 }, { "grad_norm": 0.09020815789699554, "learning_rate": 4.452314546002523e-05, "loss": 0.0041, "step": 55820 }, { "grad_norm": 0.08152438700199127, "learning_rate": 4.4506710555709246e-05, "loss": 0.006, "step": 55830 }, { "grad_norm": 0.06582469493150711, "learning_rate": 4.449027625213114e-05, "loss": 0.0049, "step": 55840 }, { "grad_norm": 0.06078450754284859, "learning_rate": 4.4473842551088135e-05, "loss": 0.0041, "step": 55850 }, { "grad_norm": 0.10878092795610428, "learning_rate": 4.445740945437738e-05, "loss": 0.0054, "step": 55860 }, { "grad_norm": 0.0784948468208313, "learning_rate": 4.444097696379602e-05, "loss": 0.0062, "step": 55870 }, { "grad_norm": 0.05141441151499748, "learning_rate": 4.442454508114106e-05, "loss": 0.0052, "step": 55880 }, { "grad_norm": 0.06844569742679596, "learning_rate": 4.4408113808209465e-05, "loss": 0.0037, "step": 55890 }, { "grad_norm": 0.07616317272186279, "learning_rate": 4.439168314679813e-05, "loss": 0.0046, "step": 55900 }, { "grad_norm": 0.0640796348452568, "learning_rate": 4.43752530987039e-05, "loss": 0.0067, "step": 55910 }, { "grad_norm": 0.05190158262848854, "learning_rate": 4.4358823665723515e-05, "loss": 0.0043, "step": 55920 }, { "grad_norm": 0.06007637083530426, "learning_rate": 4.434239484965371e-05, "loss": 0.0045, "step": 55930 }, { "grad_norm": 0.09919023513793945, "learning_rate": 4.4325966652291103e-05, "loss": 0.0051, "step": 55940 }, { "grad_norm": 0.10761591792106628, "learning_rate": 4.430953907543225e-05, "loss": 0.0056, "step": 55950 }, { "grad_norm": 0.09438618272542953, "learning_rate": 4.4293112120873645e-05, "loss": 0.0077, "step": 55960 }, { "grad_norm": 0.09246102720499039, "learning_rate": 4.427668579041172e-05, "loss": 0.0056, "step": 55970 }, { "grad_norm": 0.14672404527664185, "learning_rate": 4.4260260085842816e-05, "loss": 0.0037, "step": 55980 }, { "grad_norm": 0.06012731045484543, "learning_rate": 4.4243835008963256e-05, "loss": 0.0047, "step": 55990 }, { "grad_norm": 0.04905606061220169, "learning_rate": 4.422741056156925e-05, "loss": 0.003, "step": 56000 }, { "grad_norm": 0.12276692688465118, "learning_rate": 4.421098674545693e-05, "loss": 0.0057, "step": 56010 }, { "grad_norm": 0.09286129474639893, "learning_rate": 4.419456356242241e-05, "loss": 0.0051, "step": 56020 }, { "grad_norm": 0.060782451182603836, "learning_rate": 4.4178141014261685e-05, "loss": 0.0048, "step": 56030 }, { "grad_norm": 0.06555835902690887, "learning_rate": 4.416171910277071e-05, "loss": 0.0039, "step": 56040 }, { "grad_norm": 0.05850270390510559, "learning_rate": 4.414529782974533e-05, "loss": 0.0046, "step": 56050 }, { "grad_norm": 0.12406694889068604, "learning_rate": 4.41288771969814e-05, "loss": 0.0037, "step": 56060 }, { "grad_norm": 0.07675055414438248, "learning_rate": 4.411245720627464e-05, "loss": 0.0075, "step": 56070 }, { "grad_norm": 0.09844896197319031, "learning_rate": 4.40960378594207e-05, "loss": 0.0065, "step": 56080 }, { "grad_norm": 0.08752098679542542, "learning_rate": 4.407961915821519e-05, "loss": 0.0098, "step": 56090 }, { "grad_norm": 0.07180071622133255, "learning_rate": 4.406320110445362e-05, "loss": 0.006, "step": 56100 }, { "grad_norm": 0.05470027029514313, "learning_rate": 4.404678369993144e-05, "loss": 0.0055, "step": 56110 }, { "grad_norm": 0.07480905205011368, "learning_rate": 4.403036694644406e-05, "loss": 0.0039, "step": 56120 }, { "grad_norm": 0.07948349416255951, "learning_rate": 4.4013950845786764e-05, "loss": 0.0045, "step": 56130 }, { "grad_norm": 0.09058217704296112, "learning_rate": 4.399753539975482e-05, "loss": 0.0045, "step": 56140 }, { "grad_norm": 0.09437571465969086, "learning_rate": 4.398112061014337e-05, "loss": 0.0056, "step": 56150 }, { "grad_norm": 0.06957512348890305, "learning_rate": 4.396470647874753e-05, "loss": 0.004, "step": 56160 }, { "grad_norm": 0.03702164441347122, "learning_rate": 4.394829300736229e-05, "loss": 0.0043, "step": 56170 }, { "grad_norm": 0.12216532975435257, "learning_rate": 4.393188019778265e-05, "loss": 0.0064, "step": 56180 }, { "grad_norm": 0.09555995464324951, "learning_rate": 4.391546805180347e-05, "loss": 0.0057, "step": 56190 }, { "grad_norm": 0.06547928601503372, "learning_rate": 4.389905657121955e-05, "loss": 0.0058, "step": 56200 }, { "grad_norm": 0.08084813505411148, "learning_rate": 4.3882645757825635e-05, "loss": 0.0053, "step": 56210 }, { "grad_norm": 0.06658879667520523, "learning_rate": 4.386623561341637e-05, "loss": 0.0047, "step": 56220 }, { "grad_norm": 0.07557735592126846, "learning_rate": 4.384982613978637e-05, "loss": 0.0066, "step": 56230 }, { "grad_norm": 0.05720683932304382, "learning_rate": 4.383341733873012e-05, "loss": 0.0043, "step": 56240 }, { "grad_norm": 0.06592851877212524, "learning_rate": 4.381700921204209e-05, "loss": 0.0046, "step": 56250 }, { "grad_norm": 0.1690463423728943, "learning_rate": 4.380060176151663e-05, "loss": 0.0052, "step": 56260 }, { "grad_norm": 0.10073742270469666, "learning_rate": 4.378419498894805e-05, "loss": 0.0069, "step": 56270 }, { "grad_norm": 0.10824387520551682, "learning_rate": 4.376778889613056e-05, "loss": 0.0058, "step": 56280 }, { "grad_norm": 0.09406760334968567, "learning_rate": 4.375138348485831e-05, "loss": 0.0061, "step": 56290 }, { "grad_norm": 0.09397848695516586, "learning_rate": 4.3734978756925346e-05, "loss": 0.0048, "step": 56300 }, { "grad_norm": 0.05657079443335533, "learning_rate": 4.3718574714125706e-05, "loss": 0.0032, "step": 56310 }, { "grad_norm": 0.0692141205072403, "learning_rate": 4.370217135825329e-05, "loss": 0.0053, "step": 56320 }, { "grad_norm": 0.08579811453819275, "learning_rate": 4.368576869110194e-05, "loss": 0.004, "step": 56330 }, { "grad_norm": 0.10292776674032211, "learning_rate": 4.366936671446544e-05, "loss": 0.0055, "step": 56340 }, { "grad_norm": 0.06292232871055603, "learning_rate": 4.3652965430137474e-05, "loss": 0.0049, "step": 56350 }, { "grad_norm": 0.06223497912287712, "learning_rate": 4.3636564839911646e-05, "loss": 0.0058, "step": 56360 }, { "grad_norm": 0.08770184963941574, "learning_rate": 4.3620164945581545e-05, "loss": 0.005, "step": 56370 }, { "grad_norm": 0.1156054213643074, "learning_rate": 4.360376574894061e-05, "loss": 0.0054, "step": 56380 }, { "grad_norm": 0.10260125249624252, "learning_rate": 4.358736725178224e-05, "loss": 0.0051, "step": 56390 }, { "grad_norm": 0.12845468521118164, "learning_rate": 4.357096945589974e-05, "loss": 0.0049, "step": 56400 }, { "grad_norm": 0.06958083808422089, "learning_rate": 4.355457236308636e-05, "loss": 0.0052, "step": 56410 }, { "grad_norm": 0.0728265568614006, "learning_rate": 4.353817597513526e-05, "loss": 0.0066, "step": 56420 }, { "grad_norm": 0.08262462168931961, "learning_rate": 4.352178029383948e-05, "loss": 0.0056, "step": 56430 }, { "grad_norm": 0.08846467733383179, "learning_rate": 4.3505385320992105e-05, "loss": 0.0064, "step": 56440 }, { "grad_norm": 0.0495184063911438, "learning_rate": 4.348899105838602e-05, "loss": 0.0044, "step": 56450 }, { "grad_norm": 0.0988035649061203, "learning_rate": 4.3472597507814087e-05, "loss": 0.0049, "step": 56460 }, { "grad_norm": 0.11721570789813995, "learning_rate": 4.3456204671069066e-05, "loss": 0.0056, "step": 56470 }, { "grad_norm": 0.07401982694864273, "learning_rate": 4.343981254994367e-05, "loss": 0.0046, "step": 56480 }, { "grad_norm": 0.06529427319765091, "learning_rate": 4.34234211462305e-05, "loss": 0.0077, "step": 56490 }, { "grad_norm": 0.0823589637875557, "learning_rate": 4.340703046172213e-05, "loss": 0.0066, "step": 56500 }, { "grad_norm": 0.053433630615472794, "learning_rate": 4.339064049821097e-05, "loss": 0.0049, "step": 56510 }, { "grad_norm": 0.07082079350948334, "learning_rate": 4.3374251257489446e-05, "loss": 0.0048, "step": 56520 }, { "grad_norm": 0.07025990635156631, "learning_rate": 4.335786274134984e-05, "loss": 0.0037, "step": 56530 }, { "grad_norm": 0.12268546223640442, "learning_rate": 4.334147495158438e-05, "loss": 0.0059, "step": 56540 }, { "grad_norm": 0.10933796316385269, "learning_rate": 4.33250878899852e-05, "loss": 0.0054, "step": 56550 }, { "grad_norm": 0.06037065386772156, "learning_rate": 4.330870155834437e-05, "loss": 0.0048, "step": 56560 }, { "grad_norm": 0.18125174939632416, "learning_rate": 4.3292315958453884e-05, "loss": 0.0074, "step": 56570 }, { "grad_norm": 0.14126555621623993, "learning_rate": 4.3275931092105646e-05, "loss": 0.0068, "step": 56580 }, { "grad_norm": 0.14528925716876984, "learning_rate": 4.325954696109147e-05, "loss": 0.0063, "step": 56590 }, { "grad_norm": 0.0826430693268776, "learning_rate": 4.32431635672031e-05, "loss": 0.007, "step": 56600 }, { "grad_norm": 0.06009507179260254, "learning_rate": 4.3226780912232214e-05, "loss": 0.0039, "step": 56610 }, { "grad_norm": 0.08132528513669968, "learning_rate": 4.321039899797036e-05, "loss": 0.0059, "step": 56620 }, { "grad_norm": 0.1280679553747177, "learning_rate": 4.319401782620908e-05, "loss": 0.0063, "step": 56630 }, { "grad_norm": 0.09037584066390991, "learning_rate": 4.317763739873978e-05, "loss": 0.0061, "step": 56640 }, { "grad_norm": 0.0622127465903759, "learning_rate": 4.31612577173538e-05, "loss": 0.005, "step": 56650 }, { "grad_norm": 0.08100573718547821, "learning_rate": 4.314487878384239e-05, "loss": 0.0069, "step": 56660 }, { "grad_norm": 0.08861604332923889, "learning_rate": 4.3128500599996736e-05, "loss": 0.0039, "step": 56670 }, { "grad_norm": 0.0826646015048027, "learning_rate": 4.31121231676079e-05, "loss": 0.0048, "step": 56680 }, { "grad_norm": 0.05331113189458847, "learning_rate": 4.309574648846694e-05, "loss": 0.006, "step": 56690 }, { "grad_norm": 0.08287375420331955, "learning_rate": 4.3079370564364755e-05, "loss": 0.0051, "step": 56700 }, { "grad_norm": 0.08545945584774017, "learning_rate": 4.306299539709221e-05, "loss": 0.0058, "step": 56710 }, { "grad_norm": 0.08445336669683456, "learning_rate": 4.3046620988440046e-05, "loss": 0.0063, "step": 56720 }, { "grad_norm": 0.06496837735176086, "learning_rate": 4.303024734019897e-05, "loss": 0.0066, "step": 56730 }, { "grad_norm": 0.05473277345299721, "learning_rate": 4.3013874454159565e-05, "loss": 0.0054, "step": 56740 }, { "grad_norm": 0.09439970552921295, "learning_rate": 4.299750233211233e-05, "loss": 0.0065, "step": 56750 }, { "grad_norm": 0.05554115027189255, "learning_rate": 4.2981130975847715e-05, "loss": 0.0049, "step": 56760 }, { "grad_norm": 0.08244190365076065, "learning_rate": 4.2964760387156076e-05, "loss": 0.0043, "step": 56770 }, { "grad_norm": 0.08829844743013382, "learning_rate": 4.294839056782767e-05, "loss": 0.007, "step": 56780 }, { "grad_norm": 0.09474816918373108, "learning_rate": 4.293202151965266e-05, "loss": 0.0052, "step": 56790 }, { "grad_norm": 0.06557123363018036, "learning_rate": 4.291565324442115e-05, "loss": 0.0056, "step": 56800 }, { "grad_norm": 0.08523651212453842, "learning_rate": 4.2899285743923135e-05, "loss": 0.0049, "step": 56810 }, { "grad_norm": 0.06163790449500084, "learning_rate": 4.288291901994857e-05, "loss": 0.005, "step": 56820 }, { "grad_norm": 0.06941509991884232, "learning_rate": 4.286655307428729e-05, "loss": 0.0047, "step": 56830 }, { "grad_norm": 0.07194022089242935, "learning_rate": 4.285018790872903e-05, "loss": 0.0059, "step": 56840 }, { "grad_norm": 0.05065106600522995, "learning_rate": 4.283382352506347e-05, "loss": 0.0051, "step": 56850 }, { "grad_norm": 0.0774947851896286, "learning_rate": 4.2817459925080205e-05, "loss": 0.0042, "step": 56860 }, { "grad_norm": 0.04945304989814758, "learning_rate": 4.28010971105687e-05, "loss": 0.0042, "step": 56870 }, { "grad_norm": 0.08547990769147873, "learning_rate": 4.278473508331841e-05, "loss": 0.0068, "step": 56880 }, { "grad_norm": 0.045754652470350266, "learning_rate": 4.276837384511864e-05, "loss": 0.0048, "step": 56890 }, { "grad_norm": 0.07457661628723145, "learning_rate": 4.275201339775864e-05, "loss": 0.0052, "step": 56900 }, { "grad_norm": 0.0641399398446083, "learning_rate": 4.273565374302756e-05, "loss": 0.0057, "step": 56910 }, { "grad_norm": 0.07218432426452637, "learning_rate": 4.2719294882714454e-05, "loss": 0.0047, "step": 56920 }, { "grad_norm": 0.07247602939605713, "learning_rate": 4.270293681860831e-05, "loss": 0.0046, "step": 56930 }, { "grad_norm": 0.07244003564119339, "learning_rate": 4.268657955249802e-05, "loss": 0.0053, "step": 56940 }, { "grad_norm": 0.07601705193519592, "learning_rate": 4.267022308617241e-05, "loss": 0.0046, "step": 56950 }, { "grad_norm": 0.08106356859207153, "learning_rate": 4.265386742142018e-05, "loss": 0.0047, "step": 56960 }, { "grad_norm": 0.08234349638223648, "learning_rate": 4.263751256002996e-05, "loss": 0.0058, "step": 56970 }, { "grad_norm": 0.09473985433578491, "learning_rate": 4.26211585037903e-05, "loss": 0.0044, "step": 56980 }, { "grad_norm": 0.04595676809549332, "learning_rate": 4.2604805254489656e-05, "loss": 0.0038, "step": 56990 }, { "grad_norm": 0.08373010158538818, "learning_rate": 4.258845281391638e-05, "loss": 0.0062, "step": 57000 }, { "grad_norm": 0.06474598497152328, "learning_rate": 4.257210118385876e-05, "loss": 0.004, "step": 57010 }, { "grad_norm": 0.059064123779535294, "learning_rate": 4.2555750366104994e-05, "loss": 0.0034, "step": 57020 }, { "grad_norm": 0.0883837416768074, "learning_rate": 4.2539400362443184e-05, "loss": 0.0076, "step": 57030 }, { "grad_norm": 0.058605365455150604, "learning_rate": 4.252305117466133e-05, "loss": 0.0048, "step": 57040 }, { "grad_norm": 0.08791669458150864, "learning_rate": 4.2506702804547356e-05, "loss": 0.0052, "step": 57050 }, { "grad_norm": 0.11778344959020615, "learning_rate": 4.249035525388907e-05, "loss": 0.0057, "step": 57060 }, { "grad_norm": 0.09189805388450623, "learning_rate": 4.247400852447427e-05, "loss": 0.005, "step": 57070 }, { "grad_norm": 0.09594600647687912, "learning_rate": 4.245766261809059e-05, "loss": 0.0048, "step": 57080 }, { "grad_norm": 0.08156637847423553, "learning_rate": 4.244131753652558e-05, "loss": 0.0047, "step": 57090 }, { "grad_norm": 0.07873635739088058, "learning_rate": 4.242497328156672e-05, "loss": 0.0041, "step": 57100 }, { "grad_norm": 0.06196039170026779, "learning_rate": 4.240862985500139e-05, "loss": 0.0042, "step": 57110 }, { "grad_norm": 0.12830372154712677, "learning_rate": 4.2392287258616894e-05, "loss": 0.0053, "step": 57120 }, { "grad_norm": 0.10425934940576553, "learning_rate": 4.2375945494200396e-05, "loss": 0.0072, "step": 57130 }, { "grad_norm": 0.07101543992757797, "learning_rate": 4.2359604563539054e-05, "loss": 0.0043, "step": 57140 }, { "grad_norm": 0.06646593660116196, "learning_rate": 4.2343264468419885e-05, "loss": 0.004, "step": 57150 }, { "grad_norm": 0.13746784627437592, "learning_rate": 4.232692521062979e-05, "loss": 0.0049, "step": 57160 }, { "grad_norm": 0.06770347058773041, "learning_rate": 4.231058679195562e-05, "loss": 0.0044, "step": 57170 }, { "grad_norm": 0.07661186158657074, "learning_rate": 4.229424921418411e-05, "loss": 0.004, "step": 57180 }, { "grad_norm": 0.055630799382925034, "learning_rate": 4.2277912479101916e-05, "loss": 0.0061, "step": 57190 }, { "grad_norm": 0.053864337503910065, "learning_rate": 4.226157658849562e-05, "loss": 0.004, "step": 57200 }, { "grad_norm": 0.15798422694206238, "learning_rate": 4.2245241544151664e-05, "loss": 0.0051, "step": 57210 }, { "grad_norm": 0.08076803386211395, "learning_rate": 4.2228907347856425e-05, "loss": 0.0104, "step": 57220 }, { "grad_norm": 0.05539011210203171, "learning_rate": 4.2212574001396214e-05, "loss": 0.0044, "step": 57230 }, { "grad_norm": 0.06839236617088318, "learning_rate": 4.21962415065572e-05, "loss": 0.0038, "step": 57240 }, { "grad_norm": 0.07178077846765518, "learning_rate": 4.217990986512548e-05, "loss": 0.0053, "step": 57250 }, { "grad_norm": 0.06376443058252335, "learning_rate": 4.216357907888704e-05, "loss": 0.0037, "step": 57260 }, { "grad_norm": 0.0455075278878212, "learning_rate": 4.2147249149627824e-05, "loss": 0.0045, "step": 57270 }, { "grad_norm": 0.0987062156200409, "learning_rate": 4.213092007913364e-05, "loss": 0.0047, "step": 57280 }, { "grad_norm": 0.06165221333503723, "learning_rate": 4.2114591869190205e-05, "loss": 0.0066, "step": 57290 }, { "grad_norm": 0.07530847191810608, "learning_rate": 4.209826452158315e-05, "loss": 0.0044, "step": 57300 }, { "grad_norm": 0.11249879002571106, "learning_rate": 4.2081938038098e-05, "loss": 0.005, "step": 57310 }, { "grad_norm": 0.11139649897813797, "learning_rate": 4.2065612420520185e-05, "loss": 0.0054, "step": 57320 }, { "grad_norm": 0.10008010268211365, "learning_rate": 4.204928767063509e-05, "loss": 0.0054, "step": 57330 }, { "grad_norm": 0.07389522343873978, "learning_rate": 4.2032963790227944e-05, "loss": 0.0039, "step": 57340 }, { "grad_norm": 0.05766060948371887, "learning_rate": 4.2016640781083887e-05, "loss": 0.0074, "step": 57350 }, { "grad_norm": 0.06299947947263718, "learning_rate": 4.2000318644988004e-05, "loss": 0.0057, "step": 57360 }, { "grad_norm": 0.10656619071960449, "learning_rate": 4.198399738372524e-05, "loss": 0.0044, "step": 57370 }, { "grad_norm": 0.07101458311080933, "learning_rate": 4.196767699908045e-05, "loss": 0.008, "step": 57380 }, { "grad_norm": 0.07519327849149704, "learning_rate": 4.195135749283845e-05, "loss": 0.0066, "step": 57390 }, { "grad_norm": 0.07125206291675568, "learning_rate": 4.1935038866783894e-05, "loss": 0.0047, "step": 57400 }, { "grad_norm": 0.07127585262060165, "learning_rate": 4.1918721122701355e-05, "loss": 0.0046, "step": 57410 }, { "grad_norm": 0.05009157583117485, "learning_rate": 4.190240426237533e-05, "loss": 0.0066, "step": 57420 }, { "grad_norm": 0.05238403379917145, "learning_rate": 4.188608828759019e-05, "loss": 0.0052, "step": 57430 }, { "grad_norm": 0.05381438881158829, "learning_rate": 4.1869773200130246e-05, "loss": 0.0056, "step": 57440 }, { "grad_norm": 0.08497905731201172, "learning_rate": 4.185345900177967e-05, "loss": 0.0055, "step": 57450 }, { "grad_norm": 0.09021767973899841, "learning_rate": 4.183714569432258e-05, "loss": 0.0036, "step": 57460 }, { "grad_norm": 0.06914389878511429, "learning_rate": 4.182083327954295e-05, "loss": 0.0043, "step": 57470 }, { "grad_norm": 0.05459616705775261, "learning_rate": 4.1804521759224714e-05, "loss": 0.0038, "step": 57480 }, { "grad_norm": 0.08229110389947891, "learning_rate": 4.178821113515165e-05, "loss": 0.0053, "step": 57490 }, { "grad_norm": 0.06621275097131729, "learning_rate": 4.1771901409107476e-05, "loss": 0.0065, "step": 57500 }, { "grad_norm": 0.0759795680642128, "learning_rate": 4.175559258287577e-05, "loss": 0.0057, "step": 57510 }, { "grad_norm": 0.13061051070690155, "learning_rate": 4.173928465824008e-05, "loss": 0.0068, "step": 57520 }, { "grad_norm": 0.1145525798201561, "learning_rate": 4.17229776369838e-05, "loss": 0.0047, "step": 57530 }, { "grad_norm": 0.06735195964574814, "learning_rate": 4.170667152089025e-05, "loss": 0.0059, "step": 57540 }, { "grad_norm": 0.07809415459632874, "learning_rate": 4.1690366311742626e-05, "loss": 0.0054, "step": 57550 }, { "grad_norm": 0.04804612696170807, "learning_rate": 4.1674062011324055e-05, "loss": 0.0073, "step": 57560 }, { "grad_norm": 0.05902441591024399, "learning_rate": 4.165775862141752e-05, "loss": 0.0038, "step": 57570 }, { "grad_norm": 0.07742872834205627, "learning_rate": 4.164145614380598e-05, "loss": 0.0043, "step": 57580 }, { "grad_norm": 0.08906858414411545, "learning_rate": 4.162515458027223e-05, "loss": 0.0045, "step": 57590 }, { "grad_norm": 0.06946156173944473, "learning_rate": 4.160885393259899e-05, "loss": 0.0056, "step": 57600 }, { "grad_norm": 0.09350252151489258, "learning_rate": 4.159255420256886e-05, "loss": 0.0039, "step": 57610 }, { "grad_norm": 0.06181836873292923, "learning_rate": 4.157625539196436e-05, "loss": 0.008, "step": 57620 }, { "grad_norm": 0.07542537152767181, "learning_rate": 4.155995750256791e-05, "loss": 0.0052, "step": 57630 }, { "grad_norm": 0.09502217173576355, "learning_rate": 4.15436605361618e-05, "loss": 0.0067, "step": 57640 }, { "grad_norm": 0.0569646880030632, "learning_rate": 4.152736449452827e-05, "loss": 0.0055, "step": 57650 }, { "grad_norm": 0.11923142522573471, "learning_rate": 4.151106937944943e-05, "loss": 0.0051, "step": 57660 }, { "grad_norm": 0.08519406616687775, "learning_rate": 4.149477519270727e-05, "loss": 0.0045, "step": 57670 }, { "grad_norm": 0.09460964053869247, "learning_rate": 4.147848193608371e-05, "loss": 0.005, "step": 57680 }, { "grad_norm": 0.09926797449588776, "learning_rate": 4.146218961136056e-05, "loss": 0.0078, "step": 57690 }, { "grad_norm": 0.07565817981958389, "learning_rate": 4.14458982203195e-05, "loss": 0.0059, "step": 57700 }, { "grad_norm": 0.05912530794739723, "learning_rate": 4.1429607764742165e-05, "loss": 0.0067, "step": 57710 }, { "grad_norm": 0.0753636360168457, "learning_rate": 4.141331824641003e-05, "loss": 0.0045, "step": 57720 }, { "grad_norm": 0.06313586980104446, "learning_rate": 4.139702966710452e-05, "loss": 0.005, "step": 57730 }, { "grad_norm": 0.07332559674978256, "learning_rate": 4.13807420286069e-05, "loss": 0.0061, "step": 57740 }, { "grad_norm": 0.14792922139167786, "learning_rate": 4.136445533269837e-05, "loss": 0.0052, "step": 57750 }, { "grad_norm": 0.09186819195747375, "learning_rate": 4.134816958116003e-05, "loss": 0.006, "step": 57760 }, { "grad_norm": 0.07102841883897781, "learning_rate": 4.133188477577283e-05, "loss": 0.0058, "step": 57770 }, { "grad_norm": 0.09916232526302338, "learning_rate": 4.131560091831771e-05, "loss": 0.0045, "step": 57780 }, { "grad_norm": 0.05275842919945717, "learning_rate": 4.1299318010575405e-05, "loss": 0.0044, "step": 57790 }, { "grad_norm": 0.06706730276346207, "learning_rate": 4.128303605432661e-05, "loss": 0.0059, "step": 57800 }, { "grad_norm": 0.06966796517372131, "learning_rate": 4.126675505135188e-05, "loss": 0.0045, "step": 57810 }, { "grad_norm": 0.09181559830904007, "learning_rate": 4.125047500343169e-05, "loss": 0.0065, "step": 57820 }, { "grad_norm": 0.18970245122909546, "learning_rate": 4.1234195912346376e-05, "loss": 0.0048, "step": 57830 }, { "grad_norm": 0.0833803042769432, "learning_rate": 4.1217917779876235e-05, "loss": 0.0076, "step": 57840 }, { "grad_norm": 0.11405646055936813, "learning_rate": 4.120164060780141e-05, "loss": 0.0041, "step": 57850 }, { "grad_norm": 0.06292839348316193, "learning_rate": 4.1185364397901935e-05, "loss": 0.004, "step": 57860 }, { "grad_norm": 0.08977420628070831, "learning_rate": 4.116908915195776e-05, "loss": 0.0046, "step": 57870 }, { "grad_norm": 0.04801870137453079, "learning_rate": 4.115281487174872e-05, "loss": 0.0038, "step": 57880 }, { "grad_norm": 0.07761209458112717, "learning_rate": 4.113654155905452e-05, "loss": 0.0058, "step": 57890 }, { "grad_norm": 0.12776334583759308, "learning_rate": 4.112026921565483e-05, "loss": 0.0044, "step": 57900 }, { "grad_norm": 0.0624484159052372, "learning_rate": 4.1103997843329147e-05, "loss": 0.0056, "step": 57910 }, { "grad_norm": 0.06570397317409515, "learning_rate": 4.108772744385689e-05, "loss": 0.0043, "step": 57920 }, { "grad_norm": 0.05984688177704811, "learning_rate": 4.1071458019017346e-05, "loss": 0.0051, "step": 57930 }, { "grad_norm": 0.0791059136390686, "learning_rate": 4.1055189570589734e-05, "loss": 0.0037, "step": 57940 }, { "grad_norm": 0.049371492117643356, "learning_rate": 4.103892210035315e-05, "loss": 0.0041, "step": 57950 }, { "grad_norm": 0.06726177036762238, "learning_rate": 4.102265561008656e-05, "loss": 0.0042, "step": 57960 }, { "grad_norm": 0.06219016760587692, "learning_rate": 4.100639010156885e-05, "loss": 0.0058, "step": 57970 }, { "grad_norm": 0.05505939945578575, "learning_rate": 4.099012557657882e-05, "loss": 0.0035, "step": 57980 }, { "grad_norm": 0.09404225647449493, "learning_rate": 4.09738620368951e-05, "loss": 0.0053, "step": 57990 }, { "grad_norm": 0.06234084069728851, "learning_rate": 4.095759948429626e-05, "loss": 0.0044, "step": 58000 }, { "grad_norm": 0.04434341564774513, "learning_rate": 4.094133792056075e-05, "loss": 0.0048, "step": 58010 }, { "grad_norm": 0.05820886790752411, "learning_rate": 4.092507734746688e-05, "loss": 0.0039, "step": 58020 }, { "grad_norm": 0.06768742948770523, "learning_rate": 4.090881776679293e-05, "loss": 0.0045, "step": 58030 }, { "grad_norm": 0.08310475945472717, "learning_rate": 4.0892559180316994e-05, "loss": 0.0048, "step": 58040 }, { "grad_norm": 0.053160954266786575, "learning_rate": 4.08763015898171e-05, "loss": 0.0043, "step": 58050 }, { "grad_norm": 0.06413073092699051, "learning_rate": 4.0860044997071145e-05, "loss": 0.0047, "step": 58060 }, { "grad_norm": 0.05232260003685951, "learning_rate": 4.084378940385693e-05, "loss": 0.0046, "step": 58070 }, { "grad_norm": 0.09861932694911957, "learning_rate": 4.082753481195211e-05, "loss": 0.0075, "step": 58080 }, { "grad_norm": 0.06635049730539322, "learning_rate": 4.0811281223134324e-05, "loss": 0.0036, "step": 58090 }, { "grad_norm": 0.0737835019826889, "learning_rate": 4.0795028639181e-05, "loss": 0.006, "step": 58100 }, { "grad_norm": 0.0416790172457695, "learning_rate": 4.077877706186951e-05, "loss": 0.0055, "step": 58110 }, { "grad_norm": 0.0619577057659626, "learning_rate": 4.076252649297709e-05, "loss": 0.0062, "step": 58120 }, { "grad_norm": 0.06864659488201141, "learning_rate": 4.07462769342809e-05, "loss": 0.0069, "step": 58130 }, { "grad_norm": 0.061637941747903824, "learning_rate": 4.073002838755793e-05, "loss": 0.0062, "step": 58140 }, { "grad_norm": 0.07085564732551575, "learning_rate": 4.0713780854585115e-05, "loss": 0.0042, "step": 58150 }, { "grad_norm": 0.06499063968658447, "learning_rate": 4.069753433713927e-05, "loss": 0.0053, "step": 58160 }, { "grad_norm": 0.0816882774233818, "learning_rate": 4.068128883699709e-05, "loss": 0.0052, "step": 58170 }, { "grad_norm": 0.0758492574095726, "learning_rate": 4.0665044355935155e-05, "loss": 0.0049, "step": 58180 }, { "grad_norm": 0.1060686782002449, "learning_rate": 4.064880089572993e-05, "loss": 0.0044, "step": 58190 }, { "grad_norm": 0.10122174769639969, "learning_rate": 4.063255845815779e-05, "loss": 0.0052, "step": 58200 }, { "grad_norm": 0.05294065922498703, "learning_rate": 4.0616317044994945e-05, "loss": 0.0052, "step": 58210 }, { "grad_norm": 0.05047405883669853, "learning_rate": 4.0600076658017585e-05, "loss": 0.0035, "step": 58220 }, { "grad_norm": 0.03760332241654396, "learning_rate": 4.0583837299001706e-05, "loss": 0.0033, "step": 58230 }, { "grad_norm": 0.0634971484541893, "learning_rate": 4.056759896972322e-05, "loss": 0.0048, "step": 58240 }, { "grad_norm": 0.0472700372338295, "learning_rate": 4.055136167195793e-05, "loss": 0.0047, "step": 58250 }, { "grad_norm": 0.09205716103315353, "learning_rate": 4.053512540748152e-05, "loss": 0.0061, "step": 58260 }, { "grad_norm": 0.09476716071367264, "learning_rate": 4.051889017806955e-05, "loss": 0.005, "step": 58270 }, { "grad_norm": 0.06562570482492447, "learning_rate": 4.0502655985497516e-05, "loss": 0.0049, "step": 58280 }, { "grad_norm": 0.06545675545930862, "learning_rate": 4.048642283154074e-05, "loss": 0.0047, "step": 58290 }, { "grad_norm": 0.06058153137564659, "learning_rate": 4.0470190717974464e-05, "loss": 0.0043, "step": 58300 }, { "grad_norm": 0.06988738477230072, "learning_rate": 4.045395964657381e-05, "loss": 0.0048, "step": 58310 }, { "grad_norm": 0.05351496487855911, "learning_rate": 4.043772961911376e-05, "loss": 0.004, "step": 58320 }, { "grad_norm": 0.06588038057088852, "learning_rate": 4.042150063736924e-05, "loss": 0.0047, "step": 58330 }, { "grad_norm": 0.09364523738622665, "learning_rate": 4.0405272703114976e-05, "loss": 0.0055, "step": 58340 }, { "grad_norm": 0.09302471578121185, "learning_rate": 4.038904581812569e-05, "loss": 0.0057, "step": 58350 }, { "grad_norm": 0.06448140740394592, "learning_rate": 4.03728199841759e-05, "loss": 0.0047, "step": 58360 }, { "grad_norm": 0.07017311453819275, "learning_rate": 4.0356595203040044e-05, "loss": 0.0052, "step": 58370 }, { "grad_norm": 0.041924599558115005, "learning_rate": 4.0340371476492426e-05, "loss": 0.0035, "step": 58380 }, { "grad_norm": 0.07529977709054947, "learning_rate": 4.032414880630727e-05, "loss": 0.0041, "step": 58390 }, { "grad_norm": 0.08307303488254547, "learning_rate": 4.0307927194258646e-05, "loss": 0.0082, "step": 58400 }, { "grad_norm": 0.08901418000459671, "learning_rate": 4.029170664212054e-05, "loss": 0.0051, "step": 58410 }, { "grad_norm": 0.09757844358682632, "learning_rate": 4.0275487151666794e-05, "loss": 0.0043, "step": 58420 }, { "grad_norm": 0.08018789440393448, "learning_rate": 4.025926872467116e-05, "loss": 0.0031, "step": 58430 }, { "grad_norm": 0.06554853171110153, "learning_rate": 4.024305136290725e-05, "loss": 0.004, "step": 58440 }, { "grad_norm": 0.12251549959182739, "learning_rate": 4.022683506814858e-05, "loss": 0.0054, "step": 58450 }, { "grad_norm": 0.08816584944725037, "learning_rate": 4.021061984216853e-05, "loss": 0.0069, "step": 58460 }, { "grad_norm": 0.06169553101062775, "learning_rate": 4.019440568674035e-05, "loss": 0.0046, "step": 58470 }, { "grad_norm": 0.07366448640823364, "learning_rate": 4.017819260363724e-05, "loss": 0.0044, "step": 58480 }, { "grad_norm": 0.1412130892276764, "learning_rate": 4.016198059463223e-05, "loss": 0.0047, "step": 58490 }, { "grad_norm": 0.08324939012527466, "learning_rate": 4.0145769661498233e-05, "loss": 0.0053, "step": 58500 }, { "grad_norm": 0.08772353827953339, "learning_rate": 4.012955980600804e-05, "loss": 0.0038, "step": 58510 }, { "grad_norm": 0.11284191906452179, "learning_rate": 4.0113351029934355e-05, "loss": 0.0068, "step": 58520 }, { "grad_norm": 0.08703768253326416, "learning_rate": 4.0097143335049714e-05, "loss": 0.0043, "step": 58530 }, { "grad_norm": 0.07049828767776489, "learning_rate": 4.0080936723126606e-05, "loss": 0.0076, "step": 58540 }, { "grad_norm": 0.06921163201332092, "learning_rate": 4.006473119593735e-05, "loss": 0.0051, "step": 58550 }, { "grad_norm": 0.06494712084531784, "learning_rate": 4.004852675525415e-05, "loss": 0.0049, "step": 58560 }, { "grad_norm": 0.0712813064455986, "learning_rate": 4.00323234028491e-05, "loss": 0.0043, "step": 58570 }, { "grad_norm": 0.08363720029592514, "learning_rate": 4.0016121140494186e-05, "loss": 0.005, "step": 58580 }, { "grad_norm": 0.05700806528329849, "learning_rate": 3.999991996996122e-05, "loss": 0.0042, "step": 58590 }, { "grad_norm": 0.06997139006853104, "learning_rate": 3.9983719893022e-05, "loss": 0.0041, "step": 58600 }, { "grad_norm": 0.07104822248220444, "learning_rate": 3.996752091144812e-05, "loss": 0.0054, "step": 58610 }, { "grad_norm": 0.06404286623001099, "learning_rate": 3.9951323027011055e-05, "loss": 0.0045, "step": 58620 }, { "grad_norm": 0.07331154495477676, "learning_rate": 3.993512624148219e-05, "loss": 0.0049, "step": 58630 }, { "grad_norm": 0.07345818728208542, "learning_rate": 3.99189305566328e-05, "loss": 0.0056, "step": 58640 }, { "grad_norm": 0.08201862871646881, "learning_rate": 3.990273597423401e-05, "loss": 0.0042, "step": 58650 }, { "grad_norm": 0.06275293231010437, "learning_rate": 3.988654249605681e-05, "loss": 0.0051, "step": 58660 }, { "grad_norm": 0.06608876585960388, "learning_rate": 3.987035012387213e-05, "loss": 0.0033, "step": 58670 }, { "grad_norm": 0.048209141939878464, "learning_rate": 3.9854158859450735e-05, "loss": 0.0034, "step": 58680 }, { "grad_norm": 0.05687651038169861, "learning_rate": 3.983796870456328e-05, "loss": 0.0042, "step": 58690 }, { "grad_norm": 0.042978305369615555, "learning_rate": 3.982177966098028e-05, "loss": 0.0051, "step": 58700 }, { "grad_norm": 0.04558302089571953, "learning_rate": 3.980559173047216e-05, "loss": 0.0035, "step": 58710 }, { "grad_norm": 0.03934630751609802, "learning_rate": 3.9789404914809176e-05, "loss": 0.0039, "step": 58720 }, { "grad_norm": 0.05973227694630623, "learning_rate": 3.9773219215761544e-05, "loss": 0.0049, "step": 58730 }, { "grad_norm": 0.05041375383734703, "learning_rate": 3.975703463509928e-05, "loss": 0.0048, "step": 58740 }, { "grad_norm": 0.10546818375587463, "learning_rate": 3.974085117459231e-05, "loss": 0.0048, "step": 58750 }, { "grad_norm": 0.07494434714317322, "learning_rate": 3.972466883601043e-05, "loss": 0.0043, "step": 58760 }, { "grad_norm": 0.1498911827802658, "learning_rate": 3.9708487621123316e-05, "loss": 0.007, "step": 58770 }, { "grad_norm": 0.1004062294960022, "learning_rate": 3.969230753170051e-05, "loss": 0.0051, "step": 58780 }, { "grad_norm": 0.047136034816503525, "learning_rate": 3.967612856951146e-05, "loss": 0.0035, "step": 58790 }, { "grad_norm": 0.08325867354869843, "learning_rate": 3.9659950736325476e-05, "loss": 0.006, "step": 58800 }, { "grad_norm": 0.08022241294384003, "learning_rate": 3.964377403391174e-05, "loss": 0.007, "step": 58810 }, { "grad_norm": 0.13736411929130554, "learning_rate": 3.9627598464039303e-05, "loss": 0.0045, "step": 58820 }, { "grad_norm": 0.2562277615070343, "learning_rate": 3.9611424028477096e-05, "loss": 0.0073, "step": 58830 }, { "grad_norm": 0.1344723254442215, "learning_rate": 3.959525072899394e-05, "loss": 0.005, "step": 58840 }, { "grad_norm": 0.09431572258472443, "learning_rate": 3.9579078567358506e-05, "loss": 0.005, "step": 58850 }, { "grad_norm": 0.15273095667362213, "learning_rate": 3.956290754533939e-05, "loss": 0.0053, "step": 58860 }, { "grad_norm": 0.10288156569004059, "learning_rate": 3.954673766470501e-05, "loss": 0.0067, "step": 58870 }, { "grad_norm": 0.10236953943967819, "learning_rate": 3.9530568927223674e-05, "loss": 0.0071, "step": 58880 }, { "grad_norm": 0.05451863631606102, "learning_rate": 3.95144013346636e-05, "loss": 0.0067, "step": 58890 }, { "grad_norm": 0.05507732927799225, "learning_rate": 3.949823488879282e-05, "loss": 0.005, "step": 58900 }, { "grad_norm": 0.05377039313316345, "learning_rate": 3.9482069591379276e-05, "loss": 0.0042, "step": 58910 }, { "grad_norm": 0.04939576983451843, "learning_rate": 3.9465905444190784e-05, "loss": 0.0034, "step": 58920 }, { "grad_norm": 0.06375693529844284, "learning_rate": 3.944974244899505e-05, "loss": 0.0049, "step": 58930 }, { "grad_norm": 0.05157448723912239, "learning_rate": 3.9433580607559616e-05, "loss": 0.0035, "step": 58940 }, { "grad_norm": 0.06596875190734863, "learning_rate": 3.941741992165193e-05, "loss": 0.007, "step": 58950 }, { "grad_norm": 0.07860590517520905, "learning_rate": 3.940126039303928e-05, "loss": 0.0064, "step": 58960 }, { "grad_norm": 0.07505100965499878, "learning_rate": 3.938510202348886e-05, "loss": 0.0039, "step": 58970 }, { "grad_norm": 0.06830480694770813, "learning_rate": 3.93689448147677e-05, "loss": 0.0067, "step": 58980 }, { "grad_norm": 0.10263535380363464, "learning_rate": 3.935278876864278e-05, "loss": 0.0084, "step": 58990 }, { "grad_norm": 0.11864865571260452, "learning_rate": 3.933663388688087e-05, "loss": 0.0069, "step": 59000 }, { "grad_norm": 0.06177353486418724, "learning_rate": 3.9320480171248645e-05, "loss": 0.0059, "step": 59010 }, { "grad_norm": 0.059212055057287216, "learning_rate": 3.930432762351265e-05, "loss": 0.0055, "step": 59020 }, { "grad_norm": 0.06249019876122475, "learning_rate": 3.928817624543931e-05, "loss": 0.0055, "step": 59030 }, { "grad_norm": 0.07593600451946259, "learning_rate": 3.9272026038794885e-05, "loss": 0.0059, "step": 59040 }, { "grad_norm": 0.06641804426908493, "learning_rate": 3.9255877005345586e-05, "loss": 0.0048, "step": 59050 }, { "grad_norm": 0.04729507863521576, "learning_rate": 3.923972914685743e-05, "loss": 0.0044, "step": 59060 }, { "grad_norm": 0.06356806308031082, "learning_rate": 3.922358246509631e-05, "loss": 0.0038, "step": 59070 }, { "grad_norm": 0.08214789628982544, "learning_rate": 3.920743696182801e-05, "loss": 0.0036, "step": 59080 }, { "grad_norm": 0.05365855619311333, "learning_rate": 3.9191292638818164e-05, "loss": 0.0046, "step": 59090 }, { "grad_norm": 0.07791151851415634, "learning_rate": 3.9175149497832316e-05, "loss": 0.0052, "step": 59100 }, { "grad_norm": 0.06890711188316345, "learning_rate": 3.915900754063584e-05, "loss": 0.0054, "step": 59110 }, { "grad_norm": 0.05289345234632492, "learning_rate": 3.9142866768994014e-05, "loss": 0.0037, "step": 59120 }, { "grad_norm": 0.0572626069188118, "learning_rate": 3.912672718467193e-05, "loss": 0.0034, "step": 59130 }, { "grad_norm": 0.08774880319833755, "learning_rate": 3.911058878943463e-05, "loss": 0.006, "step": 59140 }, { "grad_norm": 0.053353242576122284, "learning_rate": 3.9094451585046974e-05, "loss": 0.0053, "step": 59150 }, { "grad_norm": 0.05768096074461937, "learning_rate": 3.90783155732737e-05, "loss": 0.004, "step": 59160 }, { "grad_norm": 0.06551545113325119, "learning_rate": 3.9062180755879373e-05, "loss": 0.0046, "step": 59170 }, { "grad_norm": 0.054172515869140625, "learning_rate": 3.904604713462855e-05, "loss": 0.0036, "step": 59180 }, { "grad_norm": 0.08037199825048447, "learning_rate": 3.902991471128554e-05, "loss": 0.0038, "step": 59190 }, { "grad_norm": 0.0495479553937912, "learning_rate": 3.901378348761456e-05, "loss": 0.0043, "step": 59200 }, { "grad_norm": 0.04833374544978142, "learning_rate": 3.899765346537971e-05, "loss": 0.0037, "step": 59210 }, { "grad_norm": 0.049525972455739975, "learning_rate": 3.898152464634493e-05, "loss": 0.0048, "step": 59220 }, { "grad_norm": 0.056078698486089706, "learning_rate": 3.896539703227402e-05, "loss": 0.0046, "step": 59230 }, { "grad_norm": 0.0788312554359436, "learning_rate": 3.894927062493073e-05, "loss": 0.0045, "step": 59240 }, { "grad_norm": 0.06356114149093628, "learning_rate": 3.893314542607859e-05, "loss": 0.0043, "step": 59250 }, { "grad_norm": 0.08089616149663925, "learning_rate": 3.891702143748103e-05, "loss": 0.0053, "step": 59260 }, { "grad_norm": 0.06959900259971619, "learning_rate": 3.890089866090133e-05, "loss": 0.0053, "step": 59270 }, { "grad_norm": 0.04653984680771828, "learning_rate": 3.888477709810267e-05, "loss": 0.0044, "step": 59280 }, { "grad_norm": 0.06231003999710083, "learning_rate": 3.886865675084806e-05, "loss": 0.0043, "step": 59290 }, { "grad_norm": 0.05228787660598755, "learning_rate": 3.885253762090043e-05, "loss": 0.0041, "step": 59300 }, { "grad_norm": 0.09803751111030579, "learning_rate": 3.8836419710022534e-05, "loss": 0.0042, "step": 59310 }, { "grad_norm": 0.06679879873991013, "learning_rate": 3.882030301997698e-05, "loss": 0.0036, "step": 59320 }, { "grad_norm": 0.06511357426643372, "learning_rate": 3.880418755252629e-05, "loss": 0.0054, "step": 59330 }, { "grad_norm": 0.05363161861896515, "learning_rate": 3.87880733094328e-05, "loss": 0.0045, "step": 59340 }, { "grad_norm": 0.07082097977399826, "learning_rate": 3.877196029245877e-05, "loss": 0.0041, "step": 59350 }, { "grad_norm": 0.05203055962920189, "learning_rate": 3.875584850336627e-05, "loss": 0.003, "step": 59360 }, { "grad_norm": 0.04389949515461922, "learning_rate": 3.873973794391728e-05, "loss": 0.0064, "step": 59370 }, { "grad_norm": 0.12107236683368683, "learning_rate": 3.872362861587361e-05, "loss": 0.0048, "step": 59380 }, { "grad_norm": 0.05250843986868858, "learning_rate": 3.870752052099698e-05, "loss": 0.0066, "step": 59390 }, { "grad_norm": 0.10565295815467834, "learning_rate": 3.8691413661048916e-05, "loss": 0.0057, "step": 59400 }, { "grad_norm": 0.054206278175115585, "learning_rate": 3.867530803779085e-05, "loss": 0.0044, "step": 59410 }, { "grad_norm": 0.16228529810905457, "learning_rate": 3.8659203652984054e-05, "loss": 0.0057, "step": 59420 }, { "grad_norm": 0.058599527925252914, "learning_rate": 3.864310050838972e-05, "loss": 0.0069, "step": 59430 }, { "grad_norm": 0.08558258414268494, "learning_rate": 3.8626998605768835e-05, "loss": 0.005, "step": 59440 }, { "grad_norm": 0.05390910059213638, "learning_rate": 3.8610897946882294e-05, "loss": 0.0039, "step": 59450 }, { "grad_norm": 0.09882944822311401, "learning_rate": 3.859479853349082e-05, "loss": 0.0034, "step": 59460 }, { "grad_norm": 0.07048241794109344, "learning_rate": 3.857870036735503e-05, "loss": 0.0037, "step": 59470 }, { "grad_norm": 0.060571085661649704, "learning_rate": 3.856260345023538e-05, "loss": 0.0055, "step": 59480 }, { "grad_norm": 0.05043015256524086, "learning_rate": 3.854650778389224e-05, "loss": 0.0042, "step": 59490 }, { "grad_norm": 0.06729618459939957, "learning_rate": 3.8530413370085786e-05, "loss": 0.0044, "step": 59500 }, { "grad_norm": 0.057177767157554626, "learning_rate": 3.851432021057608e-05, "loss": 0.0041, "step": 59510 }, { "grad_norm": 0.06333262473344803, "learning_rate": 3.849822830712306e-05, "loss": 0.0036, "step": 59520 }, { "grad_norm": 0.10264680534601212, "learning_rate": 3.848213766148649e-05, "loss": 0.0052, "step": 59530 }, { "grad_norm": 0.04589351639151573, "learning_rate": 3.846604827542602e-05, "loss": 0.0031, "step": 59540 }, { "grad_norm": 0.07330054044723511, "learning_rate": 3.844996015070117e-05, "loss": 0.0046, "step": 59550 }, { "grad_norm": 0.08091453462839127, "learning_rate": 3.843387328907132e-05, "loss": 0.004, "step": 59560 }, { "grad_norm": 0.05735812708735466, "learning_rate": 3.8417787692295696e-05, "loss": 0.0069, "step": 59570 }, { "grad_norm": 0.07782494276762009, "learning_rate": 3.84017033621334e-05, "loss": 0.0075, "step": 59580 }, { "grad_norm": 0.06621349602937698, "learning_rate": 3.838562030034338e-05, "loss": 0.0039, "step": 59590 }, { "grad_norm": 0.07131002843379974, "learning_rate": 3.836953850868447e-05, "loss": 0.0056, "step": 59600 }, { "grad_norm": 0.08991077542304993, "learning_rate": 3.835345798891533e-05, "loss": 0.0046, "step": 59610 }, { "grad_norm": 0.09956333786249161, "learning_rate": 3.833737874279452e-05, "loss": 0.0063, "step": 59620 }, { "grad_norm": 0.07281657308340073, "learning_rate": 3.8321300772080433e-05, "loss": 0.007, "step": 59630 }, { "grad_norm": 0.08778645843267441, "learning_rate": 3.8305224078531345e-05, "loss": 0.0044, "step": 59640 }, { "grad_norm": 0.07057427614927292, "learning_rate": 3.828914866390537e-05, "loss": 0.0045, "step": 59650 }, { "grad_norm": 0.08419531583786011, "learning_rate": 3.827307452996048e-05, "loss": 0.0045, "step": 59660 }, { "grad_norm": 0.1241096705198288, "learning_rate": 3.825700167845454e-05, "loss": 0.0067, "step": 59670 }, { "grad_norm": 0.08846034854650497, "learning_rate": 3.8240930111145214e-05, "loss": 0.0036, "step": 59680 }, { "grad_norm": 0.10894962400197983, "learning_rate": 3.8224859829790116e-05, "loss": 0.005, "step": 59690 }, { "grad_norm": 0.07447374612092972, "learning_rate": 3.820879083614664e-05, "loss": 0.0045, "step": 59700 }, { "grad_norm": 0.08335833251476288, "learning_rate": 3.819272313197208e-05, "loss": 0.0044, "step": 59710 }, { "grad_norm": 0.06039856746792793, "learning_rate": 3.817665671902355e-05, "loss": 0.0033, "step": 59720 }, { "grad_norm": 0.07036704570055008, "learning_rate": 3.816059159905808e-05, "loss": 0.0032, "step": 59730 }, { "grad_norm": 0.12479573488235474, "learning_rate": 3.8144527773832476e-05, "loss": 0.0044, "step": 59740 }, { "grad_norm": 0.04370313137769699, "learning_rate": 3.812846524510352e-05, "loss": 0.0057, "step": 59750 }, { "grad_norm": 0.047042954713106155, "learning_rate": 3.811240401462775e-05, "loss": 0.0045, "step": 59760 }, { "grad_norm": 0.12381307780742645, "learning_rate": 3.809634408416162e-05, "loss": 0.005, "step": 59770 }, { "grad_norm": 0.056883521378040314, "learning_rate": 3.8080285455461394e-05, "loss": 0.006, "step": 59780 }, { "grad_norm": 0.08516599237918854, "learning_rate": 3.806422813028323e-05, "loss": 0.0041, "step": 59790 }, { "grad_norm": 0.06273284554481506, "learning_rate": 3.8048172110383114e-05, "loss": 0.0041, "step": 59800 }, { "grad_norm": 0.05802585184574127, "learning_rate": 3.8032117397516944e-05, "loss": 0.0046, "step": 59810 }, { "grad_norm": 0.061724502593278885, "learning_rate": 3.801606399344042e-05, "loss": 0.0035, "step": 59820 }, { "grad_norm": 0.04685923084616661, "learning_rate": 3.800001189990913e-05, "loss": 0.0039, "step": 59830 }, { "grad_norm": 0.06371255218982697, "learning_rate": 3.798396111867847e-05, "loss": 0.0043, "step": 59840 }, { "grad_norm": 0.0477682501077652, "learning_rate": 3.796791165150378e-05, "loss": 0.0055, "step": 59850 }, { "grad_norm": 0.0726507306098938, "learning_rate": 3.795186350014018e-05, "loss": 0.0051, "step": 59860 }, { "grad_norm": 0.05059744417667389, "learning_rate": 3.793581666634266e-05, "loss": 0.004, "step": 59870 }, { "grad_norm": 0.08542421460151672, "learning_rate": 3.791977115186609e-05, "loss": 0.0055, "step": 59880 }, { "grad_norm": 0.058011963963508606, "learning_rate": 3.79037269584652e-05, "loss": 0.0052, "step": 59890 }, { "grad_norm": 0.061367060989141464, "learning_rate": 3.7887684087894545e-05, "loss": 0.0045, "step": 59900 }, { "grad_norm": 0.12081936746835709, "learning_rate": 3.787164254190854e-05, "loss": 0.0054, "step": 59910 }, { "grad_norm": 0.08388069272041321, "learning_rate": 3.7855602322261473e-05, "loss": 0.0051, "step": 59920 }, { "grad_norm": 0.10014228522777557, "learning_rate": 3.783956343070746e-05, "loss": 0.0051, "step": 59930 }, { "grad_norm": 0.05436869338154793, "learning_rate": 3.7823525869000526e-05, "loss": 0.0036, "step": 59940 }, { "grad_norm": 0.0463256910443306, "learning_rate": 3.78074896388945e-05, "loss": 0.0036, "step": 59950 }, { "grad_norm": 0.08673442900180817, "learning_rate": 3.7791454742143075e-05, "loss": 0.0038, "step": 59960 }, { "grad_norm": 0.08301756531000137, "learning_rate": 3.7775421180499795e-05, "loss": 0.0041, "step": 59970 }, { "grad_norm": 0.1266101449728012, "learning_rate": 3.7759388955718086e-05, "loss": 0.0067, "step": 59980 }, { "grad_norm": 0.08743958920240402, "learning_rate": 3.774335806955117e-05, "loss": 0.0042, "step": 59990 }, { "grad_norm": 0.09581368416547775, "learning_rate": 3.7727328523752215e-05, "loss": 0.0065, "step": 60000 }, { "grad_norm": 0.05766809359192848, "learning_rate": 3.771130032007415e-05, "loss": 0.0049, "step": 60010 }, { "grad_norm": 0.10159821808338165, "learning_rate": 3.769527346026982e-05, "loss": 0.0078, "step": 60020 }, { "grad_norm": 0.03938784822821617, "learning_rate": 3.767924794609188e-05, "loss": 0.0045, "step": 60030 }, { "grad_norm": 0.08461851626634598, "learning_rate": 3.7663223779292854e-05, "loss": 0.0055, "step": 60040 }, { "grad_norm": 0.0650879442691803, "learning_rate": 3.764720096162512e-05, "loss": 0.0044, "step": 60050 }, { "grad_norm": 0.07878279685974121, "learning_rate": 3.763117949484092e-05, "loss": 0.0072, "step": 60060 }, { "grad_norm": 0.07357071340084076, "learning_rate": 3.7615159380692344e-05, "loss": 0.0042, "step": 60070 }, { "grad_norm": 0.06319206207990646, "learning_rate": 3.759914062093132e-05, "loss": 0.0045, "step": 60080 }, { "grad_norm": 0.10052292048931122, "learning_rate": 3.7583123217309615e-05, "loss": 0.0066, "step": 60090 }, { "grad_norm": 0.07228246331214905, "learning_rate": 3.7567107171578904e-05, "loss": 0.0037, "step": 60100 }, { "grad_norm": 0.12207397818565369, "learning_rate": 3.755109248549066e-05, "loss": 0.0046, "step": 60110 }, { "grad_norm": 0.06576507538557053, "learning_rate": 3.75350791607962e-05, "loss": 0.0045, "step": 60120 }, { "grad_norm": 0.05615287274122238, "learning_rate": 3.751906719924676e-05, "loss": 0.0044, "step": 60130 }, { "grad_norm": 0.07772962749004364, "learning_rate": 3.750305660259337e-05, "loss": 0.0058, "step": 60140 }, { "grad_norm": 0.05900830402970314, "learning_rate": 3.748704737258693e-05, "loss": 0.0045, "step": 60150 }, { "grad_norm": 0.08204735070466995, "learning_rate": 3.747103951097816e-05, "loss": 0.0055, "step": 60160 }, { "grad_norm": 0.09475690126419067, "learning_rate": 3.745503301951768e-05, "loss": 0.0051, "step": 60170 }, { "grad_norm": 0.07046619057655334, "learning_rate": 3.743902789995592e-05, "loss": 0.0046, "step": 60180 }, { "grad_norm": 0.07542845606803894, "learning_rate": 3.7423024154043183e-05, "loss": 0.0042, "step": 60190 }, { "grad_norm": 0.041845664381980896, "learning_rate": 3.740702178352963e-05, "loss": 0.0032, "step": 60200 }, { "grad_norm": 0.07691647857427597, "learning_rate": 3.739102079016523e-05, "loss": 0.0045, "step": 60210 }, { "grad_norm": 0.053174447268247604, "learning_rate": 3.7375021175699846e-05, "loss": 0.0034, "step": 60220 }, { "grad_norm": 0.06341925263404846, "learning_rate": 3.735902294188317e-05, "loss": 0.0041, "step": 60230 }, { "grad_norm": 0.1393747180700302, "learning_rate": 3.734302609046473e-05, "loss": 0.0036, "step": 60240 }, { "grad_norm": 0.06278039515018463, "learning_rate": 3.732703062319391e-05, "loss": 0.0043, "step": 60250 }, { "grad_norm": 0.061574459075927734, "learning_rate": 3.731103654181999e-05, "loss": 0.0042, "step": 60260 }, { "grad_norm": 0.05420595034956932, "learning_rate": 3.729504384809203e-05, "loss": 0.0052, "step": 60270 }, { "grad_norm": 0.07378046214580536, "learning_rate": 3.727905254375898e-05, "loss": 0.005, "step": 60280 }, { "grad_norm": 0.15333716571331024, "learning_rate": 3.7263062630569614e-05, "loss": 0.0057, "step": 60290 }, { "grad_norm": 0.10500520467758179, "learning_rate": 3.724707411027256e-05, "loss": 0.0041, "step": 60300 }, { "grad_norm": 0.05868298560380936, "learning_rate": 3.723108698461631e-05, "loss": 0.0037, "step": 60310 }, { "grad_norm": 0.07014190405607224, "learning_rate": 3.72151012553492e-05, "loss": 0.0053, "step": 60320 }, { "grad_norm": 0.054273780435323715, "learning_rate": 3.719911692421939e-05, "loss": 0.0071, "step": 60330 }, { "grad_norm": 0.05775352939963341, "learning_rate": 3.71831339929749e-05, "loss": 0.0049, "step": 60340 }, { "grad_norm": 0.11987019330263138, "learning_rate": 3.716715246336362e-05, "loss": 0.0059, "step": 60350 }, { "grad_norm": 0.071132130920887, "learning_rate": 3.715117233713324e-05, "loss": 0.0065, "step": 60360 }, { "grad_norm": 0.0729292780160904, "learning_rate": 3.713519361603135e-05, "loss": 0.0052, "step": 60370 }, { "grad_norm": 0.07365214824676514, "learning_rate": 3.711921630180532e-05, "loss": 0.0054, "step": 60380 }, { "grad_norm": 0.10557527840137482, "learning_rate": 3.710324039620245e-05, "loss": 0.0079, "step": 60390 }, { "grad_norm": 0.05006348341703415, "learning_rate": 3.708726590096982e-05, "loss": 0.004, "step": 60400 }, { "grad_norm": 0.07052702456712723, "learning_rate": 3.707129281785437e-05, "loss": 0.0073, "step": 60410 }, { "grad_norm": 0.08288943767547607, "learning_rate": 3.705532114860291e-05, "loss": 0.0054, "step": 60420 }, { "grad_norm": 0.04509555548429489, "learning_rate": 3.703935089496207e-05, "loss": 0.0024, "step": 60430 }, { "grad_norm": 0.05651458352804184, "learning_rate": 3.702338205867829e-05, "loss": 0.0044, "step": 60440 }, { "grad_norm": 0.10025525838136673, "learning_rate": 3.7007414641497976e-05, "loss": 0.0046, "step": 60450 }, { "grad_norm": 0.04258064180612564, "learning_rate": 3.699144864516726e-05, "loss": 0.0055, "step": 60460 }, { "grad_norm": 0.052267368882894516, "learning_rate": 3.6975484071432153e-05, "loss": 0.0031, "step": 60470 }, { "grad_norm": 0.09918919205665588, "learning_rate": 3.695952092203854e-05, "loss": 0.0041, "step": 60480 }, { "grad_norm": 0.08034180849790573, "learning_rate": 3.69435591987321e-05, "loss": 0.0048, "step": 60490 }, { "grad_norm": 0.10845726728439331, "learning_rate": 3.6927598903258374e-05, "loss": 0.005, "step": 60500 }, { "grad_norm": 0.06929008662700653, "learning_rate": 3.691164003736281e-05, "loss": 0.0036, "step": 60510 }, { "grad_norm": 0.05356457456946373, "learning_rate": 3.689568260279061e-05, "loss": 0.0036, "step": 60520 }, { "grad_norm": 0.043831273913383484, "learning_rate": 3.687972660128686e-05, "loss": 0.0037, "step": 60530 }, { "grad_norm": 0.0858667641878128, "learning_rate": 3.686377203459648e-05, "loss": 0.0048, "step": 60540 }, { "grad_norm": 0.06054750829935074, "learning_rate": 3.6847818904464226e-05, "loss": 0.0046, "step": 60550 }, { "grad_norm": 0.05784489959478378, "learning_rate": 3.683186721263474e-05, "loss": 0.0048, "step": 60560 }, { "grad_norm": 0.06399521976709366, "learning_rate": 3.6815916960852435e-05, "loss": 0.0044, "step": 60570 }, { "grad_norm": 0.11648456752300262, "learning_rate": 3.679996815086165e-05, "loss": 0.0039, "step": 60580 }, { "grad_norm": 0.13176408410072327, "learning_rate": 3.678402078440649e-05, "loss": 0.0043, "step": 60590 }, { "grad_norm": 0.07362741976976395, "learning_rate": 3.676807486323096e-05, "loss": 0.0053, "step": 60600 }, { "grad_norm": 0.07935497164726257, "learning_rate": 3.6752130389078864e-05, "loss": 0.0056, "step": 60610 }, { "grad_norm": 0.05313396453857422, "learning_rate": 3.673618736369388e-05, "loss": 0.0065, "step": 60620 }, { "grad_norm": 0.04947073385119438, "learning_rate": 3.6720245788819474e-05, "loss": 0.004, "step": 60630 }, { "grad_norm": 0.08016828447580338, "learning_rate": 3.6704305666199044e-05, "loss": 0.005, "step": 60640 }, { "grad_norm": 0.05527553707361221, "learning_rate": 3.668836699757576e-05, "loss": 0.0046, "step": 60650 }, { "grad_norm": 0.07847294956445694, "learning_rate": 3.667242978469265e-05, "loss": 0.0045, "step": 60660 }, { "grad_norm": 0.06393566727638245, "learning_rate": 3.6656494029292575e-05, "loss": 0.0044, "step": 60670 }, { "grad_norm": 0.05497419461607933, "learning_rate": 3.664055973311825e-05, "loss": 0.0034, "step": 60680 }, { "grad_norm": 0.09305624663829803, "learning_rate": 3.662462689791221e-05, "loss": 0.0051, "step": 60690 }, { "grad_norm": 0.048982612788677216, "learning_rate": 3.660869552541689e-05, "loss": 0.0062, "step": 60700 }, { "grad_norm": 0.0969201996922493, "learning_rate": 3.659276561737448e-05, "loss": 0.0043, "step": 60710 }, { "grad_norm": 0.0705890879034996, "learning_rate": 3.657683717552708e-05, "loss": 0.0034, "step": 60720 }, { "grad_norm": 0.08708607405424118, "learning_rate": 3.6560910201616574e-05, "loss": 0.0072, "step": 60730 }, { "grad_norm": 0.05235714465379715, "learning_rate": 3.654498469738472e-05, "loss": 0.004, "step": 60740 }, { "grad_norm": 0.08428121358156204, "learning_rate": 3.6529060664573126e-05, "loss": 0.0034, "step": 60750 }, { "grad_norm": 0.09064572304487228, "learning_rate": 3.6513138104923176e-05, "loss": 0.0052, "step": 60760 }, { "grad_norm": 0.08976522088050842, "learning_rate": 3.6497217020176176e-05, "loss": 0.0042, "step": 60770 }, { "grad_norm": 0.05286175385117531, "learning_rate": 3.648129741207323e-05, "loss": 0.0035, "step": 60780 }, { "grad_norm": 0.11458950489759445, "learning_rate": 3.646537928235527e-05, "loss": 0.004, "step": 60790 }, { "grad_norm": 0.09148295223712921, "learning_rate": 3.6449462632763075e-05, "loss": 0.0044, "step": 60800 }, { "grad_norm": 0.07218501716852188, "learning_rate": 3.643354746503729e-05, "loss": 0.0049, "step": 60810 }, { "grad_norm": 0.07711002975702286, "learning_rate": 3.6417633780918345e-05, "loss": 0.0037, "step": 60820 }, { "grad_norm": 0.053285617381334305, "learning_rate": 3.640172158214655e-05, "loss": 0.0043, "step": 60830 }, { "grad_norm": 0.05431374907493591, "learning_rate": 3.6385810870462035e-05, "loss": 0.0052, "step": 60840 }, { "grad_norm": 0.04532454535365105, "learning_rate": 3.636990164760479e-05, "loss": 0.0045, "step": 60850 }, { "grad_norm": 0.17679494619369507, "learning_rate": 3.6353993915314596e-05, "loss": 0.0067, "step": 60860 }, { "grad_norm": 0.08316539227962494, "learning_rate": 3.6338087675331124e-05, "loss": 0.0049, "step": 60870 }, { "grad_norm": 0.11334936320781708, "learning_rate": 3.632218292939383e-05, "loss": 0.0053, "step": 60880 }, { "grad_norm": 0.06955081969499588, "learning_rate": 3.6306279679242033e-05, "loss": 0.0038, "step": 60890 }, { "grad_norm": 0.057199034839868546, "learning_rate": 3.629037792661492e-05, "loss": 0.0039, "step": 60900 }, { "grad_norm": 0.042777471244335175, "learning_rate": 3.627447767325147e-05, "loss": 0.0043, "step": 60910 }, { "grad_norm": 0.08009916543960571, "learning_rate": 3.62585789208905e-05, "loss": 0.0046, "step": 60920 }, { "grad_norm": 0.08835753798484802, "learning_rate": 3.624268167127068e-05, "loss": 0.0033, "step": 60930 }, { "grad_norm": 0.06560810655355453, "learning_rate": 3.622678592613051e-05, "loss": 0.0054, "step": 60940 }, { "grad_norm": 0.05251514911651611, "learning_rate": 3.6210891687208304e-05, "loss": 0.005, "step": 60950 }, { "grad_norm": 0.05939716845750809, "learning_rate": 3.619499895624228e-05, "loss": 0.0029, "step": 60960 }, { "grad_norm": 0.052431050688028336, "learning_rate": 3.61791077349704e-05, "loss": 0.004, "step": 60970 }, { "grad_norm": 0.10507570952177048, "learning_rate": 3.616321802513053e-05, "loss": 0.0042, "step": 60980 }, { "grad_norm": 0.06833593547344208, "learning_rate": 3.6147329828460325e-05, "loss": 0.0037, "step": 60990 }, { "grad_norm": 0.07445714622735977, "learning_rate": 3.6131443146697306e-05, "loss": 0.004, "step": 61000 }, { "grad_norm": 0.06541519612073898, "learning_rate": 3.611555798157879e-05, "loss": 0.0074, "step": 61010 }, { "grad_norm": 0.0986061692237854, "learning_rate": 3.6099674334842e-05, "loss": 0.004, "step": 61020 }, { "grad_norm": 0.05412180721759796, "learning_rate": 3.6083792208223934e-05, "loss": 0.0063, "step": 61030 }, { "grad_norm": 0.09201696515083313, "learning_rate": 3.606791160346142e-05, "loss": 0.0037, "step": 61040 }, { "grad_norm": 0.05907505750656128, "learning_rate": 3.6052032522291134e-05, "loss": 0.0038, "step": 61050 }, { "grad_norm": 0.04909920692443848, "learning_rate": 3.603615496644962e-05, "loss": 0.0028, "step": 61060 }, { "grad_norm": 0.10942140221595764, "learning_rate": 3.60202789376732e-05, "loss": 0.0039, "step": 61070 }, { "grad_norm": 0.05460493266582489, "learning_rate": 3.6004404437698036e-05, "loss": 0.006, "step": 61080 }, { "grad_norm": 0.1384759247303009, "learning_rate": 3.5988531468260184e-05, "loss": 0.0044, "step": 61090 }, { "grad_norm": 0.08073113113641739, "learning_rate": 3.597266003109546e-05, "loss": 0.0043, "step": 61100 }, { "grad_norm": 0.06599868834018707, "learning_rate": 3.5956790127939557e-05, "loss": 0.0044, "step": 61110 }, { "grad_norm": 0.04889967292547226, "learning_rate": 3.594092176052796e-05, "loss": 0.0038, "step": 61120 }, { "grad_norm": 0.06694289296865463, "learning_rate": 3.592505493059603e-05, "loss": 0.0036, "step": 61130 }, { "grad_norm": 0.04831418767571449, "learning_rate": 3.5909189639878905e-05, "loss": 0.0038, "step": 61140 }, { "grad_norm": 0.09605247527360916, "learning_rate": 3.5893325890111626e-05, "loss": 0.0036, "step": 61150 }, { "grad_norm": 0.058158423751592636, "learning_rate": 3.5877463683029034e-05, "loss": 0.0044, "step": 61160 }, { "grad_norm": 0.05233682319521904, "learning_rate": 3.586160302036578e-05, "loss": 0.0042, "step": 61170 }, { "grad_norm": 0.06439842283725739, "learning_rate": 3.584574390385636e-05, "loss": 0.0049, "step": 61180 }, { "grad_norm": 0.08004430681467056, "learning_rate": 3.582988633523511e-05, "loss": 0.0046, "step": 61190 }, { "grad_norm": 0.06778216361999512, "learning_rate": 3.581403031623616e-05, "loss": 0.0068, "step": 61200 }, { "grad_norm": 0.06301074475049973, "learning_rate": 3.579817584859354e-05, "loss": 0.0055, "step": 61210 }, { "grad_norm": 0.04540422558784485, "learning_rate": 3.578232293404107e-05, "loss": 0.0041, "step": 61220 }, { "grad_norm": 0.0508372001349926, "learning_rate": 3.5766471574312385e-05, "loss": 0.0062, "step": 61230 }, { "grad_norm": 0.03243400901556015, "learning_rate": 3.5750621771140964e-05, "loss": 0.0043, "step": 61240 }, { "grad_norm": 0.07651741057634354, "learning_rate": 3.5734773526260115e-05, "loss": 0.0055, "step": 61250 }, { "grad_norm": 0.08630629628896713, "learning_rate": 3.5718926841402996e-05, "loss": 0.0041, "step": 61260 }, { "grad_norm": 0.04447408393025398, "learning_rate": 3.5703081718302555e-05, "loss": 0.0041, "step": 61270 }, { "grad_norm": 0.05673797056078911, "learning_rate": 3.568723815869162e-05, "loss": 0.0038, "step": 61280 }, { "grad_norm": 0.1246330738067627, "learning_rate": 3.5671396164302777e-05, "loss": 0.0056, "step": 61290 }, { "grad_norm": 0.06711893528699875, "learning_rate": 3.565555573686853e-05, "loss": 0.0076, "step": 61300 }, { "grad_norm": 0.07382843643426895, "learning_rate": 3.563971687812113e-05, "loss": 0.006, "step": 61310 }, { "grad_norm": 0.06069692596793175, "learning_rate": 3.5623879589792706e-05, "loss": 0.0043, "step": 61320 }, { "grad_norm": 0.054936252534389496, "learning_rate": 3.5608043873615185e-05, "loss": 0.0038, "step": 61330 }, { "grad_norm": 0.03852272778749466, "learning_rate": 3.559220973132035e-05, "loss": 0.0043, "step": 61340 }, { "grad_norm": 0.10859843343496323, "learning_rate": 3.557637716463981e-05, "loss": 0.0059, "step": 61350 }, { "grad_norm": 0.07130713015794754, "learning_rate": 3.556054617530497e-05, "loss": 0.0044, "step": 61360 }, { "grad_norm": 0.08284968882799149, "learning_rate": 3.5544716765047106e-05, "loss": 0.0044, "step": 61370 }, { "grad_norm": 0.0653713271021843, "learning_rate": 3.5528888935597274e-05, "loss": 0.0064, "step": 61380 }, { "grad_norm": 0.055745840072631836, "learning_rate": 3.551306268868636e-05, "loss": 0.0047, "step": 61390 }, { "grad_norm": 0.06107866019010544, "learning_rate": 3.5497238026045166e-05, "loss": 0.0062, "step": 61400 }, { "grad_norm": 0.07730837166309357, "learning_rate": 3.548141494940421e-05, "loss": 0.0041, "step": 61410 }, { "grad_norm": 0.04856954514980316, "learning_rate": 3.546559346049388e-05, "loss": 0.004, "step": 61420 }, { "grad_norm": 0.04585200920701027, "learning_rate": 3.544977356104441e-05, "loss": 0.0033, "step": 61430 }, { "grad_norm": 0.059895701706409454, "learning_rate": 3.543395525278582e-05, "loss": 0.0042, "step": 61440 }, { "grad_norm": 0.05953684076666832, "learning_rate": 3.5418138537447985e-05, "loss": 0.0031, "step": 61450 }, { "grad_norm": 0.062318287789821625, "learning_rate": 3.5402323416760574e-05, "loss": 0.0032, "step": 61460 }, { "grad_norm": 0.07480844110250473, "learning_rate": 3.538650989245316e-05, "loss": 0.0047, "step": 61470 }, { "grad_norm": 0.06235284358263016, "learning_rate": 3.537069796625504e-05, "loss": 0.0052, "step": 61480 }, { "grad_norm": 0.0638476237654686, "learning_rate": 3.53548876398954e-05, "loss": 0.0039, "step": 61490 }, { "grad_norm": 0.05081426352262497, "learning_rate": 3.5339078915103214e-05, "loss": 0.0033, "step": 61500 }, { "grad_norm": 0.08557896316051483, "learning_rate": 3.532327179360733e-05, "loss": 0.0036, "step": 61510 }, { "grad_norm": 0.09801159054040909, "learning_rate": 3.530746627713636e-05, "loss": 0.0033, "step": 61520 }, { "grad_norm": 0.07561657577753067, "learning_rate": 3.5291662367418805e-05, "loss": 0.0046, "step": 61530 }, { "grad_norm": 0.06286149471998215, "learning_rate": 3.527586006618293e-05, "loss": 0.0048, "step": 61540 }, { "grad_norm": 0.07566388696432114, "learning_rate": 3.526005937515687e-05, "loss": 0.0045, "step": 61550 }, { "grad_norm": 0.06669723242521286, "learning_rate": 3.524426029606856e-05, "loss": 0.0072, "step": 61560 }, { "grad_norm": 0.06423290818929672, "learning_rate": 3.5228462830645746e-05, "loss": 0.0047, "step": 61570 }, { "grad_norm": 0.06621081382036209, "learning_rate": 3.5212666980616036e-05, "loss": 0.0033, "step": 61580 }, { "grad_norm": 0.056126918643713, "learning_rate": 3.519687274770681e-05, "loss": 0.0039, "step": 61590 }, { "grad_norm": 0.10233189910650253, "learning_rate": 3.518108013364535e-05, "loss": 0.0044, "step": 61600 }, { "grad_norm": 0.12561851739883423, "learning_rate": 3.516528914015868e-05, "loss": 0.0035, "step": 61610 }, { "grad_norm": 0.07438115775585175, "learning_rate": 3.5149499768973694e-05, "loss": 0.0049, "step": 61620 }, { "grad_norm": 0.06723185628652573, "learning_rate": 3.513371202181708e-05, "loss": 0.0038, "step": 61630 }, { "grad_norm": 0.09838520735502243, "learning_rate": 3.511792590041537e-05, "loss": 0.0049, "step": 61640 }, { "grad_norm": 0.07675892114639282, "learning_rate": 3.510214140649488e-05, "loss": 0.0042, "step": 61650 }, { "grad_norm": 0.06586460024118423, "learning_rate": 3.508635854178184e-05, "loss": 0.0043, "step": 61660 }, { "grad_norm": 0.07381682097911835, "learning_rate": 3.50705773080022e-05, "loss": 0.0036, "step": 61670 }, { "grad_norm": 0.07535415887832642, "learning_rate": 3.505479770688178e-05, "loss": 0.0047, "step": 61680 }, { "grad_norm": 0.05006147176027298, "learning_rate": 3.503901974014622e-05, "loss": 0.0035, "step": 61690 }, { "grad_norm": 0.09005638211965561, "learning_rate": 3.502324340952095e-05, "loss": 0.0055, "step": 61700 }, { "grad_norm": 0.09616304934024811, "learning_rate": 3.500746871673126e-05, "loss": 0.0045, "step": 61710 }, { "grad_norm": 0.0661316066980362, "learning_rate": 3.499169566350227e-05, "loss": 0.0041, "step": 61720 }, { "grad_norm": 0.07358501106500626, "learning_rate": 3.4975924251558876e-05, "loss": 0.0076, "step": 61730 }, { "grad_norm": 0.05160776525735855, "learning_rate": 3.496015448262582e-05, "loss": 0.0038, "step": 61740 }, { "grad_norm": 0.09502244740724564, "learning_rate": 3.494438635842764e-05, "loss": 0.0038, "step": 61750 }, { "grad_norm": 0.10349663347005844, "learning_rate": 3.492861988068874e-05, "loss": 0.0051, "step": 61760 }, { "grad_norm": 0.07752781361341476, "learning_rate": 3.491285505113332e-05, "loss": 0.0042, "step": 61770 }, { "grad_norm": 0.07158132642507553, "learning_rate": 3.4897091871485376e-05, "loss": 0.0043, "step": 61780 }, { "grad_norm": 0.06806057691574097, "learning_rate": 3.488133034346875e-05, "loss": 0.0049, "step": 61790 }, { "grad_norm": 0.05267547816038132, "learning_rate": 3.486557046880713e-05, "loss": 0.0032, "step": 61800 }, { "grad_norm": 0.06556537002325058, "learning_rate": 3.484981224922395e-05, "loss": 0.0048, "step": 61810 }, { "grad_norm": 0.08030356466770172, "learning_rate": 3.4834055686442544e-05, "loss": 0.0059, "step": 61820 }, { "grad_norm": 0.06892119348049164, "learning_rate": 3.4818300782186e-05, "loss": 0.0046, "step": 61830 }, { "grad_norm": 0.06174037978053093, "learning_rate": 3.4802547538177236e-05, "loss": 0.0034, "step": 61840 }, { "grad_norm": 0.06089869141578674, "learning_rate": 3.478679595613905e-05, "loss": 0.0045, "step": 61850 }, { "grad_norm": 0.051158465445041656, "learning_rate": 3.4771046037793975e-05, "loss": 0.0044, "step": 61860 }, { "grad_norm": 0.06975802779197693, "learning_rate": 3.4755297784864416e-05, "loss": 0.0059, "step": 61870 }, { "grad_norm": 0.09464547038078308, "learning_rate": 3.473955119907257e-05, "loss": 0.0042, "step": 61880 }, { "grad_norm": 0.049382880330085754, "learning_rate": 3.472380628214046e-05, "loss": 0.0031, "step": 61890 }, { "grad_norm": 0.11877407133579254, "learning_rate": 3.4708063035789906e-05, "loss": 0.0067, "step": 61900 }, { "grad_norm": 0.10022259503602982, "learning_rate": 3.4692321461742614e-05, "loss": 0.0051, "step": 61910 }, { "grad_norm": 0.04760964587330818, "learning_rate": 3.4676581561720023e-05, "loss": 0.0042, "step": 61920 }, { "grad_norm": 0.06376896053552628, "learning_rate": 3.466084333744344e-05, "loss": 0.0042, "step": 61930 }, { "grad_norm": 0.05375639721751213, "learning_rate": 3.4645106790633964e-05, "loss": 0.0057, "step": 61940 }, { "grad_norm": 0.062196072190999985, "learning_rate": 3.4629371923012526e-05, "loss": 0.0051, "step": 61950 }, { "grad_norm": 0.06557786464691162, "learning_rate": 3.4613638736299856e-05, "loss": 0.0075, "step": 61960 }, { "grad_norm": 0.06532973051071167, "learning_rate": 3.459790723221652e-05, "loss": 0.0057, "step": 61970 }, { "grad_norm": 0.06859750300645828, "learning_rate": 3.45821774124829e-05, "loss": 0.0052, "step": 61980 }, { "grad_norm": 0.061569180339574814, "learning_rate": 3.456644927881918e-05, "loss": 0.0027, "step": 61990 }, { "grad_norm": 0.06260191649198532, "learning_rate": 3.4550722832945354e-05, "loss": 0.0033, "step": 62000 }, { "grad_norm": 0.09867730736732483, "learning_rate": 3.453499807658126e-05, "loss": 0.0033, "step": 62010 }, { "grad_norm": 0.06979941576719284, "learning_rate": 3.451927501144653e-05, "loss": 0.004, "step": 62020 }, { "grad_norm": 0.06038185581564903, "learning_rate": 3.450355363926059e-05, "loss": 0.0031, "step": 62030 }, { "grad_norm": 0.06890132278203964, "learning_rate": 3.448783396174273e-05, "loss": 0.0042, "step": 62040 }, { "grad_norm": 0.07471287995576859, "learning_rate": 3.4472115980612034e-05, "loss": 0.0073, "step": 62050 }, { "grad_norm": 0.06577063351869583, "learning_rate": 3.4456399697587396e-05, "loss": 0.004, "step": 62060 }, { "grad_norm": 0.06086597591638565, "learning_rate": 3.4440685114387506e-05, "loss": 0.0056, "step": 62070 }, { "grad_norm": 0.06132137402892113, "learning_rate": 3.442497223273091e-05, "loss": 0.0053, "step": 62080 }, { "grad_norm": 0.07759929448366165, "learning_rate": 3.4409261054335926e-05, "loss": 0.0033, "step": 62090 }, { "grad_norm": 0.0510445162653923, "learning_rate": 3.439355158092069e-05, "loss": 0.0049, "step": 62100 }, { "grad_norm": 0.06673116236925125, "learning_rate": 3.437784381420322e-05, "loss": 0.0031, "step": 62110 }, { "grad_norm": 0.05958566442131996, "learning_rate": 3.436213775590126e-05, "loss": 0.0048, "step": 62120 }, { "grad_norm": 0.09148725867271423, "learning_rate": 3.4346433407732405e-05, "loss": 0.0049, "step": 62130 }, { "grad_norm": 0.062170691788196564, "learning_rate": 3.433073077141405e-05, "loss": 0.0044, "step": 62140 }, { "grad_norm": 0.07168716937303543, "learning_rate": 3.431502984866343e-05, "loss": 0.005, "step": 62150 }, { "grad_norm": 0.050545480102300644, "learning_rate": 3.429933064119753e-05, "loss": 0.0046, "step": 62160 }, { "grad_norm": 0.032543495297431946, "learning_rate": 3.4283633150733255e-05, "loss": 0.0037, "step": 62170 }, { "grad_norm": 0.08277469873428345, "learning_rate": 3.4267937378987225e-05, "loss": 0.0057, "step": 62180 }, { "grad_norm": 0.06654950231313705, "learning_rate": 3.425224332767592e-05, "loss": 0.003, "step": 62190 }, { "grad_norm": 0.08715860545635223, "learning_rate": 3.4236550998515595e-05, "loss": 0.0043, "step": 62200 }, { "grad_norm": 0.07671041786670685, "learning_rate": 3.422086039322235e-05, "loss": 0.0064, "step": 62210 }, { "grad_norm": 0.06807304173707962, "learning_rate": 3.420517151351209e-05, "loss": 0.0039, "step": 62220 }, { "grad_norm": 0.07255663722753525, "learning_rate": 3.418948436110053e-05, "loss": 0.0053, "step": 62230 }, { "grad_norm": 0.06417336314916611, "learning_rate": 3.417379893770319e-05, "loss": 0.005, "step": 62240 }, { "grad_norm": 0.057929977774620056, "learning_rate": 3.41581152450354e-05, "loss": 0.0035, "step": 62250 }, { "grad_norm": 0.07148100435733795, "learning_rate": 3.414243328481231e-05, "loss": 0.0034, "step": 62260 }, { "grad_norm": 0.03675023466348648, "learning_rate": 3.412675305874888e-05, "loss": 0.0035, "step": 62270 }, { "grad_norm": 0.09053254127502441, "learning_rate": 3.4111074568559876e-05, "loss": 0.0035, "step": 62280 }, { "grad_norm": 0.07187193632125854, "learning_rate": 3.4095397815959846e-05, "loss": 0.0054, "step": 62290 }, { "grad_norm": 0.1083841323852539, "learning_rate": 3.407972280266322e-05, "loss": 0.0057, "step": 62300 }, { "grad_norm": 0.08228138089179993, "learning_rate": 3.4064049530384166e-05, "loss": 0.0044, "step": 62310 }, { "grad_norm": 0.05612320825457573, "learning_rate": 3.40483780008367e-05, "loss": 0.0035, "step": 62320 }, { "grad_norm": 0.0639638900756836, "learning_rate": 3.4032708215734635e-05, "loss": 0.0051, "step": 62330 }, { "grad_norm": 0.053015079349279404, "learning_rate": 3.4017040176791584e-05, "loss": 0.0039, "step": 62340 }, { "grad_norm": 0.04283592849969864, "learning_rate": 3.4001373885720976e-05, "loss": 0.0034, "step": 62350 }, { "grad_norm": 0.06957940012216568, "learning_rate": 3.398570934423608e-05, "loss": 0.0051, "step": 62360 }, { "grad_norm": 0.08221745491027832, "learning_rate": 3.397004655404994e-05, "loss": 0.0059, "step": 62370 }, { "grad_norm": 0.08093387633562088, "learning_rate": 3.395438551687541e-05, "loss": 0.0055, "step": 62380 }, { "grad_norm": 0.07987130433320999, "learning_rate": 3.393872623442515e-05, "loss": 0.0045, "step": 62390 }, { "grad_norm": 0.07365671545267105, "learning_rate": 3.392306870841164e-05, "loss": 0.0038, "step": 62400 }, { "grad_norm": 0.058202099055051804, "learning_rate": 3.390741294054715e-05, "loss": 0.0066, "step": 62410 }, { "grad_norm": 0.055947445333004, "learning_rate": 3.38917589325438e-05, "loss": 0.0045, "step": 62420 }, { "grad_norm": 0.08739685267210007, "learning_rate": 3.387610668611349e-05, "loss": 0.0055, "step": 62430 }, { "grad_norm": 0.05815112590789795, "learning_rate": 3.38604562029679e-05, "loss": 0.003, "step": 62440 }, { "grad_norm": 0.10566051304340363, "learning_rate": 3.384480748481855e-05, "loss": 0.004, "step": 62450 }, { "grad_norm": 0.07386427372694016, "learning_rate": 3.382916053337676e-05, "loss": 0.0046, "step": 62460 }, { "grad_norm": 0.07213285565376282, "learning_rate": 3.381351535035366e-05, "loss": 0.0027, "step": 62470 }, { "grad_norm": 0.08107113093137741, "learning_rate": 3.379787193746019e-05, "loss": 0.0038, "step": 62480 }, { "grad_norm": 0.06951453536748886, "learning_rate": 3.378223029640709e-05, "loss": 0.0034, "step": 62490 }, { "grad_norm": 0.10197566449642181, "learning_rate": 3.376659042890489e-05, "loss": 0.0042, "step": 62500 }, { "grad_norm": 0.07587356120347977, "learning_rate": 3.375095233666397e-05, "loss": 0.0048, "step": 62510 }, { "grad_norm": 0.056069422513246536, "learning_rate": 3.373531602139447e-05, "loss": 0.0029, "step": 62520 }, { "grad_norm": 0.08019605278968811, "learning_rate": 3.371968148480635e-05, "loss": 0.0072, "step": 62530 }, { "grad_norm": 0.07763193547725677, "learning_rate": 3.370404872860937e-05, "loss": 0.0054, "step": 62540 }, { "grad_norm": 0.04946831986308098, "learning_rate": 3.3688417754513145e-05, "loss": 0.0042, "step": 62550 }, { "grad_norm": 0.06789553165435791, "learning_rate": 3.367278856422703e-05, "loss": 0.0068, "step": 62560 }, { "grad_norm": 0.060839395970106125, "learning_rate": 3.365716115946021e-05, "loss": 0.0049, "step": 62570 }, { "grad_norm": 0.0563921295106411, "learning_rate": 3.3641535541921666e-05, "loss": 0.0064, "step": 62580 }, { "grad_norm": 0.06822901964187622, "learning_rate": 3.36259117133202e-05, "loss": 0.0036, "step": 62590 }, { "grad_norm": 0.07746470719575882, "learning_rate": 3.36102896753644e-05, "loss": 0.0053, "step": 62600 }, { "grad_norm": 0.06094985082745552, "learning_rate": 3.35946694297627e-05, "loss": 0.0036, "step": 62610 }, { "grad_norm": 0.04681083932518959, "learning_rate": 3.357905097822328e-05, "loss": 0.0045, "step": 62620 }, { "grad_norm": 0.05401160195469856, "learning_rate": 3.356343432245416e-05, "loss": 0.0044, "step": 62630 }, { "grad_norm": 0.049278054386377335, "learning_rate": 3.354781946416315e-05, "loss": 0.004, "step": 62640 }, { "grad_norm": 0.10844890773296356, "learning_rate": 3.3532206405057864e-05, "loss": 0.0049, "step": 62650 }, { "grad_norm": 0.07275736331939697, "learning_rate": 3.351659514684573e-05, "loss": 0.0055, "step": 62660 }, { "grad_norm": 0.08832277357578278, "learning_rate": 3.350098569123395e-05, "loss": 0.0038, "step": 62670 }, { "grad_norm": 0.07818489521741867, "learning_rate": 3.348537803992958e-05, "loss": 0.0039, "step": 62680 }, { "grad_norm": 0.04710198566317558, "learning_rate": 3.346977219463945e-05, "loss": 0.0051, "step": 62690 }, { "grad_norm": 0.06976444274187088, "learning_rate": 3.345416815707017e-05, "loss": 0.0042, "step": 62700 }, { "grad_norm": 0.06952810287475586, "learning_rate": 3.3438565928928185e-05, "loss": 0.0043, "step": 62710 }, { "grad_norm": 0.0736791267991066, "learning_rate": 3.3422965511919736e-05, "loss": 0.0044, "step": 62720 }, { "grad_norm": 0.06864258646965027, "learning_rate": 3.3407366907750846e-05, "loss": 0.003, "step": 62730 }, { "grad_norm": 0.04954804107546806, "learning_rate": 3.339177011812737e-05, "loss": 0.0038, "step": 62740 }, { "grad_norm": 0.07812879234552383, "learning_rate": 3.3376175144754935e-05, "loss": 0.0042, "step": 62750 }, { "grad_norm": 0.08709880709648132, "learning_rate": 3.3360581989339004e-05, "loss": 0.0045, "step": 62760 }, { "grad_norm": 0.0791008397936821, "learning_rate": 3.3344990653584804e-05, "loss": 0.0055, "step": 62770 }, { "grad_norm": 0.07812098413705826, "learning_rate": 3.332940113919739e-05, "loss": 0.0043, "step": 62780 }, { "grad_norm": 0.06591662019491196, "learning_rate": 3.331381344788161e-05, "loss": 0.0035, "step": 62790 }, { "grad_norm": 0.047227032482624054, "learning_rate": 3.3298227581342076e-05, "loss": 0.0036, "step": 62800 }, { "grad_norm": 0.04016667976975441, "learning_rate": 3.328264354128329e-05, "loss": 0.0042, "step": 62810 }, { "grad_norm": 0.06387531757354736, "learning_rate": 3.3267061329409465e-05, "loss": 0.0045, "step": 62820 }, { "grad_norm": 0.07845418155193329, "learning_rate": 3.325148094742466e-05, "loss": 0.0047, "step": 62830 }, { "grad_norm": 0.06210748106241226, "learning_rate": 3.323590239703271e-05, "loss": 0.0045, "step": 62840 }, { "grad_norm": 0.08028872311115265, "learning_rate": 3.322032567993728e-05, "loss": 0.0052, "step": 62850 }, { "grad_norm": 0.057498667389154434, "learning_rate": 3.3204750797841774e-05, "loss": 0.0027, "step": 62860 }, { "grad_norm": 0.0719296932220459, "learning_rate": 3.31891777524495e-05, "loss": 0.0037, "step": 62870 }, { "grad_norm": 0.08544447273015976, "learning_rate": 3.317360654546347e-05, "loss": 0.0055, "step": 62880 }, { "grad_norm": 0.06093737110495567, "learning_rate": 3.3158037178586534e-05, "loss": 0.0061, "step": 62890 }, { "grad_norm": 0.05383811891078949, "learning_rate": 3.3142469653521316e-05, "loss": 0.004, "step": 62900 }, { "grad_norm": 0.056079331785440445, "learning_rate": 3.3126903971970284e-05, "loss": 0.0045, "step": 62910 }, { "grad_norm": 0.04364873096346855, "learning_rate": 3.311134013563564e-05, "loss": 0.0044, "step": 62920 }, { "grad_norm": 0.06572679430246353, "learning_rate": 3.309577814621947e-05, "loss": 0.0045, "step": 62930 }, { "grad_norm": 0.05771135911345482, "learning_rate": 3.308021800542359e-05, "loss": 0.0048, "step": 62940 }, { "grad_norm": 0.03995067998766899, "learning_rate": 3.306465971494963e-05, "loss": 0.0039, "step": 62950 }, { "grad_norm": 0.049894727766513824, "learning_rate": 3.304910327649901e-05, "loss": 0.004, "step": 62960 }, { "grad_norm": 0.0856931060552597, "learning_rate": 3.303354869177297e-05, "loss": 0.0041, "step": 62970 }, { "grad_norm": 0.058146167546510696, "learning_rate": 3.301799596247255e-05, "loss": 0.0054, "step": 62980 }, { "grad_norm": 0.0688178688287735, "learning_rate": 3.300244509029854e-05, "loss": 0.003, "step": 62990 }, { "grad_norm": 0.07389543950557709, "learning_rate": 3.2986896076951585e-05, "loss": 0.0052, "step": 63000 }, { "grad_norm": 0.06576713919639587, "learning_rate": 3.297134892413211e-05, "loss": 0.0037, "step": 63010 }, { "grad_norm": 0.08130069077014923, "learning_rate": 3.295580363354031e-05, "loss": 0.0055, "step": 63020 }, { "grad_norm": 0.08611300587654114, "learning_rate": 3.2940260206876194e-05, "loss": 0.0053, "step": 63030 }, { "grad_norm": 0.04779786989092827, "learning_rate": 3.292471864583958e-05, "loss": 0.0033, "step": 63040 }, { "grad_norm": 0.0596645288169384, "learning_rate": 3.290917895213004e-05, "loss": 0.0042, "step": 63050 }, { "grad_norm": 0.047644224017858505, "learning_rate": 3.289364112744701e-05, "loss": 0.0039, "step": 63060 }, { "grad_norm": 0.03604913130402565, "learning_rate": 3.287810517348967e-05, "loss": 0.0027, "step": 63070 }, { "grad_norm": 0.048024244606494904, "learning_rate": 3.2862571091957014e-05, "loss": 0.003, "step": 63080 }, { "grad_norm": 0.04608551785349846, "learning_rate": 3.284703888454781e-05, "loss": 0.0036, "step": 63090 }, { "grad_norm": 0.06891579180955887, "learning_rate": 3.2831508552960646e-05, "loss": 0.0055, "step": 63100 }, { "grad_norm": 0.05315803363919258, "learning_rate": 3.281598009889387e-05, "loss": 0.0038, "step": 63110 }, { "grad_norm": 0.05314064398407936, "learning_rate": 3.28004535240457e-05, "loss": 0.0039, "step": 63120 }, { "grad_norm": 0.06234043091535568, "learning_rate": 3.278492883011407e-05, "loss": 0.0033, "step": 63130 }, { "grad_norm": 0.06326299160718918, "learning_rate": 3.276940601879675e-05, "loss": 0.0046, "step": 63140 }, { "grad_norm": 0.06672974675893784, "learning_rate": 3.275388509179127e-05, "loss": 0.0042, "step": 63150 }, { "grad_norm": 0.0526646189391613, "learning_rate": 3.2738366050794985e-05, "loss": 0.0039, "step": 63160 }, { "grad_norm": 0.11726316064596176, "learning_rate": 3.2722848897505034e-05, "loss": 0.0049, "step": 63170 }, { "grad_norm": 0.051923226565122604, "learning_rate": 3.2707333633618356e-05, "loss": 0.0046, "step": 63180 }, { "grad_norm": 0.09922371059656143, "learning_rate": 3.269182026083167e-05, "loss": 0.0057, "step": 63190 }, { "grad_norm": 0.04120950400829315, "learning_rate": 3.2676308780841516e-05, "loss": 0.004, "step": 63200 }, { "grad_norm": 0.0958879142999649, "learning_rate": 3.266079919534416e-05, "loss": 0.004, "step": 63210 }, { "grad_norm": 0.06969248503446579, "learning_rate": 3.264529150603575e-05, "loss": 0.0038, "step": 63220 }, { "grad_norm": 0.05375044047832489, "learning_rate": 3.262978571461217e-05, "loss": 0.003, "step": 63230 }, { "grad_norm": 0.09973420947790146, "learning_rate": 3.26142818227691e-05, "loss": 0.0064, "step": 63240 }, { "grad_norm": 0.060494646430015564, "learning_rate": 3.259877983220203e-05, "loss": 0.0063, "step": 63250 }, { "grad_norm": 0.056206170469522476, "learning_rate": 3.2583279744606246e-05, "loss": 0.0033, "step": 63260 }, { "grad_norm": 0.0658767968416214, "learning_rate": 3.25677815616768e-05, "loss": 0.0028, "step": 63270 }, { "grad_norm": 0.08702419698238373, "learning_rate": 3.255228528510856e-05, "loss": 0.0029, "step": 63280 }, { "grad_norm": 0.08751106262207031, "learning_rate": 3.253679091659617e-05, "loss": 0.0046, "step": 63290 }, { "grad_norm": 0.07218766212463379, "learning_rate": 3.252129845783404e-05, "loss": 0.0035, "step": 63300 }, { "grad_norm": 0.05554194003343582, "learning_rate": 3.2505807910516456e-05, "loss": 0.0046, "step": 63310 }, { "grad_norm": 0.05646703764796257, "learning_rate": 3.249031927633742e-05, "loss": 0.0076, "step": 63320 }, { "grad_norm": 0.07281424850225449, "learning_rate": 3.2474832556990737e-05, "loss": 0.0045, "step": 63330 }, { "grad_norm": 0.07761390507221222, "learning_rate": 3.245934775417002e-05, "loss": 0.0047, "step": 63340 }, { "grad_norm": 0.08159665763378143, "learning_rate": 3.244386486956866e-05, "loss": 0.0057, "step": 63350 }, { "grad_norm": 0.0533316396176815, "learning_rate": 3.2428383904879846e-05, "loss": 0.0033, "step": 63360 }, { "grad_norm": 0.05897517874836922, "learning_rate": 3.241290486179652e-05, "loss": 0.0041, "step": 63370 }, { "grad_norm": 0.03881248086690903, "learning_rate": 3.239742774201151e-05, "loss": 0.0047, "step": 63380 }, { "grad_norm": 0.03874270245432854, "learning_rate": 3.238195254721733e-05, "loss": 0.004, "step": 63390 }, { "grad_norm": 0.10242968797683716, "learning_rate": 3.236647927910633e-05, "loss": 0.0051, "step": 63400 }, { "grad_norm": 0.0612027607858181, "learning_rate": 3.2351007939370646e-05, "loss": 0.0045, "step": 63410 }, { "grad_norm": 0.08792088180780411, "learning_rate": 3.2335538529702184e-05, "loss": 0.0033, "step": 63420 }, { "grad_norm": 0.06835920363664627, "learning_rate": 3.232007105179268e-05, "loss": 0.0028, "step": 63430 }, { "grad_norm": 0.08342137187719345, "learning_rate": 3.230460550733363e-05, "loss": 0.006, "step": 63440 }, { "grad_norm": 0.08684593439102173, "learning_rate": 3.2289141898016315e-05, "loss": 0.005, "step": 63450 }, { "grad_norm": 0.07078571617603302, "learning_rate": 3.227368022553181e-05, "loss": 0.0048, "step": 63460 }, { "grad_norm": 0.06454284489154816, "learning_rate": 3.225822049157099e-05, "loss": 0.0039, "step": 63470 }, { "grad_norm": 0.06061553582549095, "learning_rate": 3.2242762697824494e-05, "loss": 0.0051, "step": 63480 }, { "grad_norm": 0.0436820350587368, "learning_rate": 3.222730684598279e-05, "loss": 0.0042, "step": 63490 }, { "grad_norm": 0.06924960017204285, "learning_rate": 3.221185293773605e-05, "loss": 0.0057, "step": 63500 }, { "grad_norm": 0.09556784480810165, "learning_rate": 3.2196400974774354e-05, "loss": 0.0055, "step": 63510 }, { "grad_norm": 0.07538409531116486, "learning_rate": 3.2180950958787485e-05, "loss": 0.0034, "step": 63520 }, { "grad_norm": 0.06610094010829926, "learning_rate": 3.2165502891465024e-05, "loss": 0.0048, "step": 63530 }, { "grad_norm": 0.05658133700489998, "learning_rate": 3.215005677449636e-05, "loss": 0.0038, "step": 63540 }, { "grad_norm": 0.06243637949228287, "learning_rate": 3.2134612609570644e-05, "loss": 0.0041, "step": 63550 }, { "grad_norm": 0.05735483393073082, "learning_rate": 3.211917039837682e-05, "loss": 0.0051, "step": 63560 }, { "grad_norm": 0.052229590713977814, "learning_rate": 3.210373014260365e-05, "loss": 0.0032, "step": 63570 }, { "grad_norm": 0.05228237807750702, "learning_rate": 3.208829184393965e-05, "loss": 0.006, "step": 63580 }, { "grad_norm": 0.08146093785762787, "learning_rate": 3.207285550407312e-05, "loss": 0.0039, "step": 63590 }, { "grad_norm": 0.06536393612623215, "learning_rate": 3.205742112469217e-05, "loss": 0.0035, "step": 63600 }, { "grad_norm": 0.06991693377494812, "learning_rate": 3.204198870748465e-05, "loss": 0.0029, "step": 63610 }, { "grad_norm": 0.10716225206851959, "learning_rate": 3.202655825413824e-05, "loss": 0.0051, "step": 63620 }, { "grad_norm": 0.05791753903031349, "learning_rate": 3.2011129766340416e-05, "loss": 0.0034, "step": 63630 }, { "grad_norm": 0.06907276064157486, "learning_rate": 3.1995703245778384e-05, "loss": 0.0046, "step": 63640 }, { "grad_norm": 0.07799652963876724, "learning_rate": 3.1980278694139186e-05, "loss": 0.005, "step": 63650 }, { "grad_norm": 0.07659471780061722, "learning_rate": 3.196485611310962e-05, "loss": 0.004, "step": 63660 }, { "grad_norm": 0.06086815893650055, "learning_rate": 3.1949435504376257e-05, "loss": 0.0036, "step": 63670 }, { "grad_norm": 0.04888474941253662, "learning_rate": 3.1934016869625505e-05, "loss": 0.0037, "step": 63680 }, { "grad_norm": 0.053990621119737625, "learning_rate": 3.191860021054349e-05, "loss": 0.0044, "step": 63690 }, { "grad_norm": 0.05932609364390373, "learning_rate": 3.190318552881618e-05, "loss": 0.0039, "step": 63700 }, { "grad_norm": 0.06697670370340347, "learning_rate": 3.1887772826129285e-05, "loss": 0.0046, "step": 63710 }, { "grad_norm": 0.08710700273513794, "learning_rate": 3.1872362104168335e-05, "loss": 0.0037, "step": 63720 }, { "grad_norm": 0.04983529448509216, "learning_rate": 3.185695336461861e-05, "loss": 0.0038, "step": 63730 }, { "grad_norm": 0.07174713164567947, "learning_rate": 3.184154660916519e-05, "loss": 0.0044, "step": 63740 }, { "grad_norm": 0.05893294885754585, "learning_rate": 3.1826141839492906e-05, "loss": 0.0047, "step": 63750 }, { "grad_norm": 0.06422193348407745, "learning_rate": 3.181073905728644e-05, "loss": 0.0035, "step": 63760 }, { "grad_norm": 0.08423750102519989, "learning_rate": 3.1795338264230214e-05, "loss": 0.0038, "step": 63770 }, { "grad_norm": 0.05877207592129707, "learning_rate": 3.1779939462008426e-05, "loss": 0.0045, "step": 63780 }, { "grad_norm": 0.07091251015663147, "learning_rate": 3.1764542652305066e-05, "loss": 0.0029, "step": 63790 }, { "grad_norm": 0.05304628238081932, "learning_rate": 3.17491478368039e-05, "loss": 0.0034, "step": 63800 }, { "grad_norm": 0.08347912132740021, "learning_rate": 3.173375501718847e-05, "loss": 0.0039, "step": 63810 }, { "grad_norm": 0.09174931049346924, "learning_rate": 3.171836419514215e-05, "loss": 0.0042, "step": 63820 }, { "grad_norm": 0.056544266641139984, "learning_rate": 3.170297537234803e-05, "loss": 0.0047, "step": 63830 }, { "grad_norm": 0.0599992610514164, "learning_rate": 3.168758855048901e-05, "loss": 0.0039, "step": 63840 }, { "grad_norm": 0.0407579205930233, "learning_rate": 3.1672203731247785e-05, "loss": 0.0055, "step": 63850 }, { "grad_norm": 0.06992412358522415, "learning_rate": 3.16568209163068e-05, "loss": 0.0062, "step": 63860 }, { "grad_norm": 0.06405138224363327, "learning_rate": 3.1641440107348295e-05, "loss": 0.0042, "step": 63870 }, { "grad_norm": 0.08183427900075912, "learning_rate": 3.162606130605429e-05, "loss": 0.0036, "step": 63880 }, { "grad_norm": 0.048951003700494766, "learning_rate": 3.16106845141066e-05, "loss": 0.0047, "step": 63890 }, { "grad_norm": 0.0574837401509285, "learning_rate": 3.159530973318682e-05, "loss": 0.0048, "step": 63900 }, { "grad_norm": 0.09160003066062927, "learning_rate": 3.1579936964976286e-05, "loss": 0.0053, "step": 63910 }, { "grad_norm": 0.08238576352596283, "learning_rate": 3.156456621115615e-05, "loss": 0.0043, "step": 63920 }, { "grad_norm": 0.08513504266738892, "learning_rate": 3.154919747340734e-05, "loss": 0.0044, "step": 63930 }, { "grad_norm": 0.09003734588623047, "learning_rate": 3.153383075341054e-05, "loss": 0.0055, "step": 63940 }, { "grad_norm": 0.06571044027805328, "learning_rate": 3.1518466052846264e-05, "loss": 0.0047, "step": 63950 }, { "grad_norm": 0.07913246005773544, "learning_rate": 3.150310337339474e-05, "loss": 0.0048, "step": 63960 }, { "grad_norm": 0.060304317623376846, "learning_rate": 3.148774271673603e-05, "loss": 0.0051, "step": 63970 }, { "grad_norm": 0.08699891716241837, "learning_rate": 3.147238408454995e-05, "loss": 0.0045, "step": 63980 }, { "grad_norm": 0.03938807547092438, "learning_rate": 3.1457027478516086e-05, "loss": 0.0032, "step": 63990 }, { "grad_norm": 0.0898825079202652, "learning_rate": 3.144167290031382e-05, "loss": 0.0049, "step": 64000 }, { "grad_norm": 0.06375337392091751, "learning_rate": 3.142632035162227e-05, "loss": 0.0033, "step": 64010 }, { "grad_norm": 0.08234788477420807, "learning_rate": 3.141096983412044e-05, "loss": 0.0033, "step": 64020 }, { "grad_norm": 0.07341889292001724, "learning_rate": 3.139562134948699e-05, "loss": 0.0035, "step": 64030 }, { "grad_norm": 0.06827342510223389, "learning_rate": 3.1380274899400424e-05, "loss": 0.0035, "step": 64040 }, { "grad_norm": 0.0552026629447937, "learning_rate": 3.1364930485539e-05, "loss": 0.0049, "step": 64050 }, { "grad_norm": 0.050622642040252686, "learning_rate": 3.134958810958076e-05, "loss": 0.003, "step": 64060 }, { "grad_norm": 0.0730791762471199, "learning_rate": 3.13342477732035e-05, "loss": 0.0034, "step": 64070 }, { "grad_norm": 0.11045766621828079, "learning_rate": 3.131890947808487e-05, "loss": 0.0057, "step": 64080 }, { "grad_norm": 0.06742708384990692, "learning_rate": 3.13035732259022e-05, "loss": 0.0037, "step": 64090 }, { "grad_norm": 0.047902628779411316, "learning_rate": 3.1288239018332656e-05, "loss": 0.0032, "step": 64100 }, { "grad_norm": 0.04615403339266777, "learning_rate": 3.1272906857053164e-05, "loss": 0.0044, "step": 64110 }, { "grad_norm": 0.04830367863178253, "learning_rate": 3.125757674374042e-05, "loss": 0.0035, "step": 64120 }, { "grad_norm": 0.059731561690568924, "learning_rate": 3.124224868007088e-05, "loss": 0.0055, "step": 64130 }, { "grad_norm": 0.06346411257982254, "learning_rate": 3.1226922667720854e-05, "loss": 0.0044, "step": 64140 }, { "grad_norm": 0.0470520481467247, "learning_rate": 3.121159870836633e-05, "loss": 0.0052, "step": 64150 }, { "grad_norm": 0.05924117937684059, "learning_rate": 3.119627680368312e-05, "loss": 0.0052, "step": 64160 }, { "grad_norm": 0.05720756947994232, "learning_rate": 3.1180956955346816e-05, "loss": 0.0024, "step": 64170 }, { "grad_norm": 0.07130639255046844, "learning_rate": 3.116563916503276e-05, "loss": 0.0054, "step": 64180 }, { "grad_norm": 0.061523132026195526, "learning_rate": 3.1150323434416095e-05, "loss": 0.0053, "step": 64190 }, { "grad_norm": 0.06320202350616455, "learning_rate": 3.1135009765171696e-05, "loss": 0.0061, "step": 64200 }, { "grad_norm": 0.05964222922921181, "learning_rate": 3.111969815897429e-05, "loss": 0.0031, "step": 64210 }, { "grad_norm": 0.08085685223340988, "learning_rate": 3.1104388617498295e-05, "loss": 0.0037, "step": 64220 }, { "grad_norm": 0.15477211773395538, "learning_rate": 3.1089081142417966e-05, "loss": 0.0043, "step": 64230 }, { "grad_norm": 0.09020280838012695, "learning_rate": 3.1073775735407285e-05, "loss": 0.0044, "step": 64240 }, { "grad_norm": 0.05891675502061844, "learning_rate": 3.1058472398140035e-05, "loss": 0.0037, "step": 64250 }, { "grad_norm": 0.08297877013683319, "learning_rate": 3.1043171132289736e-05, "loss": 0.0036, "step": 64260 }, { "grad_norm": 0.06707911938428879, "learning_rate": 3.1027871939529774e-05, "loss": 0.0039, "step": 64270 }, { "grad_norm": 0.07555732131004333, "learning_rate": 3.1012574821533214e-05, "loss": 0.0032, "step": 64280 }, { "grad_norm": 0.08701054751873016, "learning_rate": 3.099727977997291e-05, "loss": 0.0045, "step": 64290 }, { "grad_norm": 0.0770309790968895, "learning_rate": 3.098198681652154e-05, "loss": 0.0054, "step": 64300 }, { "grad_norm": 0.07728508859872818, "learning_rate": 3.0966695932851475e-05, "loss": 0.0038, "step": 64310 }, { "grad_norm": 0.07455035299062729, "learning_rate": 3.0951407130634926e-05, "loss": 0.004, "step": 64320 }, { "grad_norm": 0.055160731077194214, "learning_rate": 3.0936120411543864e-05, "loss": 0.0037, "step": 64330 }, { "grad_norm": 0.08704997599124908, "learning_rate": 3.0920835777250015e-05, "loss": 0.0051, "step": 64340 }, { "grad_norm": 0.051703982055187225, "learning_rate": 3.090555322942488e-05, "loss": 0.0041, "step": 64350 }, { "grad_norm": 0.06807178258895874, "learning_rate": 3.089027276973973e-05, "loss": 0.005, "step": 64360 }, { "grad_norm": 0.08416768163442612, "learning_rate": 3.0874994399865615e-05, "loss": 0.0034, "step": 64370 }, { "grad_norm": 0.0880005955696106, "learning_rate": 3.085971812147337e-05, "loss": 0.0045, "step": 64380 }, { "grad_norm": 0.059577032923698425, "learning_rate": 3.0844443936233557e-05, "loss": 0.008, "step": 64390 }, { "grad_norm": 0.07331658154726028, "learning_rate": 3.082917184581657e-05, "loss": 0.0041, "step": 64400 }, { "grad_norm": 0.0541452132165432, "learning_rate": 3.0813901851892513e-05, "loss": 0.0053, "step": 64410 }, { "grad_norm": 0.07508812099695206, "learning_rate": 3.0798633956131316e-05, "loss": 0.0041, "step": 64420 }, { "grad_norm": 0.06211461126804352, "learning_rate": 3.078336816020264e-05, "loss": 0.0046, "step": 64430 }, { "grad_norm": 0.060559142380952835, "learning_rate": 3.0768104465775935e-05, "loss": 0.0035, "step": 64440 }, { "grad_norm": 0.0553661473095417, "learning_rate": 3.0752842874520396e-05, "loss": 0.0035, "step": 64450 }, { "grad_norm": 0.03876620903611183, "learning_rate": 3.073758338810503e-05, "loss": 0.0038, "step": 64460 }, { "grad_norm": 0.0720440223813057, "learning_rate": 3.072232600819859e-05, "loss": 0.0033, "step": 64470 }, { "grad_norm": 0.0653332993388176, "learning_rate": 3.070707073646959e-05, "loss": 0.0041, "step": 64480 }, { "grad_norm": 0.07036153972148895, "learning_rate": 3.069181757458633e-05, "loss": 0.0045, "step": 64490 }, { "grad_norm": 0.05502946674823761, "learning_rate": 3.067656652421687e-05, "loss": 0.0043, "step": 64500 }, { "grad_norm": 0.07100151479244232, "learning_rate": 3.0661317587029025e-05, "loss": 0.0045, "step": 64510 }, { "grad_norm": 0.05860858038067818, "learning_rate": 3.064607076469043e-05, "loss": 0.0033, "step": 64520 }, { "grad_norm": 0.045938096940517426, "learning_rate": 3.063082605886845e-05, "loss": 0.003, "step": 64530 }, { "grad_norm": 0.08087185025215149, "learning_rate": 3.061558347123019e-05, "loss": 0.0039, "step": 64540 }, { "grad_norm": 0.03901368007063866, "learning_rate": 3.060034300344259e-05, "loss": 0.0035, "step": 64550 }, { "grad_norm": 0.06654166430234909, "learning_rate": 3.058510465717231e-05, "loss": 0.003, "step": 64560 }, { "grad_norm": 0.07057415693998337, "learning_rate": 3.0569868434085804e-05, "loss": 0.0051, "step": 64570 }, { "grad_norm": 0.0603056438267231, "learning_rate": 3.055463433584924e-05, "loss": 0.004, "step": 64580 }, { "grad_norm": 0.08500455319881439, "learning_rate": 3.0539402364128656e-05, "loss": 0.0038, "step": 64590 }, { "grad_norm": 0.08095671236515045, "learning_rate": 3.052417252058977e-05, "loss": 0.0066, "step": 64600 }, { "grad_norm": 0.06090884655714035, "learning_rate": 3.0508944806898092e-05, "loss": 0.0024, "step": 64610 }, { "grad_norm": 0.06443262845277786, "learning_rate": 3.04937192247189e-05, "loss": 0.0047, "step": 64620 }, { "grad_norm": 0.05615818127989769, "learning_rate": 3.047849577571725e-05, "loss": 0.0054, "step": 64630 }, { "grad_norm": 0.05700046196579933, "learning_rate": 3.046327446155795e-05, "loss": 0.0056, "step": 64640 }, { "grad_norm": 0.0507776141166687, "learning_rate": 3.0448055283905587e-05, "loss": 0.0036, "step": 64650 }, { "grad_norm": 0.07296279817819595, "learning_rate": 3.0432838244424494e-05, "loss": 0.0043, "step": 64660 }, { "grad_norm": 0.09733271598815918, "learning_rate": 3.04176233447788e-05, "loss": 0.0049, "step": 64670 }, { "grad_norm": 0.08041001856327057, "learning_rate": 3.040241058663238e-05, "loss": 0.0052, "step": 64680 }, { "grad_norm": 0.03989740461111069, "learning_rate": 3.038719997164887e-05, "loss": 0.0046, "step": 64690 }, { "grad_norm": 0.0767580047249794, "learning_rate": 3.0371991501491683e-05, "loss": 0.0036, "step": 64700 }, { "grad_norm": 0.07067535817623138, "learning_rate": 3.0356785177823975e-05, "loss": 0.0033, "step": 64710 }, { "grad_norm": 0.05630268156528473, "learning_rate": 3.0341581002308726e-05, "loss": 0.0024, "step": 64720 }, { "grad_norm": 0.07812045514583588, "learning_rate": 3.0326378976608627e-05, "loss": 0.0059, "step": 64730 }, { "grad_norm": 0.07829872518777847, "learning_rate": 3.0311179102386146e-05, "loss": 0.0051, "step": 64740 }, { "grad_norm": 0.06196556240320206, "learning_rate": 3.0295981381303513e-05, "loss": 0.0044, "step": 64750 }, { "grad_norm": 0.11766164749860764, "learning_rate": 3.0280785815022727e-05, "loss": 0.0069, "step": 64760 }, { "grad_norm": 0.05851701647043228, "learning_rate": 3.0265592405205544e-05, "loss": 0.0049, "step": 64770 }, { "grad_norm": 0.060470353811979294, "learning_rate": 3.0250401153513517e-05, "loss": 0.0044, "step": 64780 }, { "grad_norm": 0.06688784807920456, "learning_rate": 3.0235212061607936e-05, "loss": 0.0037, "step": 64790 }, { "grad_norm": 0.06876933574676514, "learning_rate": 3.0220025131149842e-05, "loss": 0.0048, "step": 64800 }, { "grad_norm": 0.05533294752240181, "learning_rate": 3.020484036380007e-05, "loss": 0.0042, "step": 64810 }, { "grad_norm": 0.0497862733900547, "learning_rate": 3.0189657761219188e-05, "loss": 0.0051, "step": 64820 }, { "grad_norm": 0.05672124773263931, "learning_rate": 3.017447732506753e-05, "loss": 0.006, "step": 64830 }, { "grad_norm": 0.10531354695558548, "learning_rate": 3.015929905700524e-05, "loss": 0.0038, "step": 64840 }, { "grad_norm": 0.05772288888692856, "learning_rate": 3.014412295869218e-05, "loss": 0.0057, "step": 64850 }, { "grad_norm": 0.09952161461114883, "learning_rate": 3.0128949031787977e-05, "loss": 0.0064, "step": 64860 }, { "grad_norm": 0.05044536292552948, "learning_rate": 3.0113777277952022e-05, "loss": 0.0036, "step": 64870 }, { "grad_norm": 0.07079549133777618, "learning_rate": 3.0098607698843496e-05, "loss": 0.0035, "step": 64880 }, { "grad_norm": 0.09405211359262466, "learning_rate": 3.0083440296121312e-05, "loss": 0.0033, "step": 64890 }, { "grad_norm": 0.03889619931578636, "learning_rate": 3.0068275071444142e-05, "loss": 0.004, "step": 64900 }, { "grad_norm": 0.07214369624853134, "learning_rate": 3.0053112026470442e-05, "loss": 0.0042, "step": 64910 }, { "grad_norm": 0.05134033039212227, "learning_rate": 3.0037951162858434e-05, "loss": 0.005, "step": 64920 }, { "grad_norm": 0.0870383232831955, "learning_rate": 3.002279248226607e-05, "loss": 0.0061, "step": 64930 }, { "grad_norm": 0.06867720931768417, "learning_rate": 3.000763598635109e-05, "loss": 0.0033, "step": 64940 }, { "grad_norm": 0.05820942297577858, "learning_rate": 2.999248167677098e-05, "loss": 0.0033, "step": 64950 }, { "grad_norm": 0.05912372097373009, "learning_rate": 2.9977329555182976e-05, "loss": 0.0056, "step": 64960 }, { "grad_norm": 0.06447017937898636, "learning_rate": 2.9962179623244124e-05, "loss": 0.006, "step": 64970 }, { "grad_norm": 0.06084360554814339, "learning_rate": 2.9947031882611175e-05, "loss": 0.0032, "step": 64980 }, { "grad_norm": 0.05054495856165886, "learning_rate": 2.9931886334940685e-05, "loss": 0.0035, "step": 64990 }, { "grad_norm": 0.07118254899978638, "learning_rate": 2.9916742981888923e-05, "loss": 0.0039, "step": 65000 }, { "grad_norm": 0.07978309690952301, "learning_rate": 2.990160182511196e-05, "loss": 0.0041, "step": 65010 }, { "grad_norm": 0.07259302586317062, "learning_rate": 2.9886462866265574e-05, "loss": 0.0045, "step": 65020 }, { "grad_norm": 0.06677606701850891, "learning_rate": 2.9871326107005392e-05, "loss": 0.0049, "step": 65030 }, { "grad_norm": 0.0823160856962204, "learning_rate": 2.9856191548986735e-05, "loss": 0.0059, "step": 65040 }, { "grad_norm": 0.07236375659704208, "learning_rate": 2.9841059193864672e-05, "loss": 0.0048, "step": 65050 }, { "grad_norm": 0.0728965625166893, "learning_rate": 2.982592904329407e-05, "loss": 0.006, "step": 65060 }, { "grad_norm": 0.07243292033672333, "learning_rate": 2.981080109892954e-05, "loss": 0.0057, "step": 65070 }, { "grad_norm": 0.06338965147733688, "learning_rate": 2.9795675362425434e-05, "loss": 0.0045, "step": 65080 }, { "grad_norm": 0.056695494800806046, "learning_rate": 2.97805518354359e-05, "loss": 0.0039, "step": 65090 }, { "grad_norm": 0.09508385509252548, "learning_rate": 2.9765430519614825e-05, "loss": 0.0052, "step": 65100 }, { "grad_norm": 0.07358376681804657, "learning_rate": 2.975031141661584e-05, "loss": 0.0042, "step": 65110 }, { "grad_norm": 0.06263943761587143, "learning_rate": 2.9735194528092347e-05, "loss": 0.0028, "step": 65120 }, { "grad_norm": 0.04372701793909073, "learning_rate": 2.9720079855697525e-05, "loss": 0.0038, "step": 65130 }, { "grad_norm": 0.07055865228176117, "learning_rate": 2.970496740108427e-05, "loss": 0.0045, "step": 65140 }, { "grad_norm": 0.03986204415559769, "learning_rate": 2.968985716590527e-05, "loss": 0.0032, "step": 65150 }, { "grad_norm": 0.0530298687517643, "learning_rate": 2.9674749151812953e-05, "loss": 0.0033, "step": 65160 }, { "grad_norm": 0.05702392756938934, "learning_rate": 2.965964336045951e-05, "loss": 0.0039, "step": 65170 }, { "grad_norm": 0.0572182722389698, "learning_rate": 2.9644539793496894e-05, "loss": 0.0067, "step": 65180 }, { "grad_norm": 0.06926797330379486, "learning_rate": 2.9629438452576807e-05, "loss": 0.0032, "step": 65190 }, { "grad_norm": 0.059991445392370224, "learning_rate": 2.9614339339350698e-05, "loss": 0.0042, "step": 65200 }, { "grad_norm": 0.05306165665388107, "learning_rate": 2.959924245546979e-05, "loss": 0.0022, "step": 65210 }, { "grad_norm": 0.038623448461294174, "learning_rate": 2.9584147802585038e-05, "loss": 0.003, "step": 65220 }, { "grad_norm": 0.08140438050031662, "learning_rate": 2.956905538234721e-05, "loss": 0.0051, "step": 65230 }, { "grad_norm": 0.05253210291266441, "learning_rate": 2.9553965196406764e-05, "loss": 0.004, "step": 65240 }, { "grad_norm": 0.07201717048883438, "learning_rate": 2.9538877246413943e-05, "loss": 0.0037, "step": 65250 }, { "grad_norm": 0.07725420594215393, "learning_rate": 2.952379153401874e-05, "loss": 0.0034, "step": 65260 }, { "grad_norm": 0.059399377554655075, "learning_rate": 2.9508708060870914e-05, "loss": 0.004, "step": 65270 }, { "grad_norm": 0.054943036288022995, "learning_rate": 2.9493626828619948e-05, "loss": 0.0051, "step": 65280 }, { "grad_norm": 0.04612911865115166, "learning_rate": 2.9478547838915137e-05, "loss": 0.003, "step": 65290 }, { "grad_norm": 0.08661092817783356, "learning_rate": 2.946347109340548e-05, "loss": 0.0055, "step": 65300 }, { "grad_norm": 0.09238843619823456, "learning_rate": 2.944839659373976e-05, "loss": 0.0038, "step": 65310 }, { "grad_norm": 0.09508388489484787, "learning_rate": 2.943332434156647e-05, "loss": 0.0038, "step": 65320 }, { "grad_norm": 0.05843593180179596, "learning_rate": 2.9418254338533912e-05, "loss": 0.0036, "step": 65330 }, { "grad_norm": 0.049815837293863297, "learning_rate": 2.940318658629011e-05, "loss": 0.004, "step": 65340 }, { "grad_norm": 0.07019691169261932, "learning_rate": 2.938812108648287e-05, "loss": 0.005, "step": 65350 }, { "grad_norm": 0.04694925621151924, "learning_rate": 2.937305784075972e-05, "loss": 0.0085, "step": 65360 }, { "grad_norm": 0.05113803595304489, "learning_rate": 2.9357996850767937e-05, "loss": 0.0032, "step": 65370 }, { "grad_norm": 0.06777261942625046, "learning_rate": 2.9342938118154595e-05, "loss": 0.005, "step": 65380 }, { "grad_norm": 0.052244335412979126, "learning_rate": 2.9327881644566485e-05, "loss": 0.0022, "step": 65390 }, { "grad_norm": 0.04803803190588951, "learning_rate": 2.9312827431650157e-05, "loss": 0.0055, "step": 65400 }, { "grad_norm": 0.0619097575545311, "learning_rate": 2.92977754810519e-05, "loss": 0.0048, "step": 65410 }, { "grad_norm": 0.05208057165145874, "learning_rate": 2.928272579441781e-05, "loss": 0.0035, "step": 65420 }, { "grad_norm": 0.0616886243224144, "learning_rate": 2.9267678373393676e-05, "loss": 0.0045, "step": 65430 }, { "grad_norm": 0.04819776490330696, "learning_rate": 2.925263321962507e-05, "loss": 0.0047, "step": 65440 }, { "grad_norm": 0.0673832967877388, "learning_rate": 2.9237590334757292e-05, "loss": 0.0037, "step": 65450 }, { "grad_norm": 0.05757470801472664, "learning_rate": 2.922254972043543e-05, "loss": 0.0044, "step": 65460 }, { "grad_norm": 0.09059024602174759, "learning_rate": 2.9207511378304265e-05, "loss": 0.0053, "step": 65470 }, { "grad_norm": 0.09541887044906616, "learning_rate": 2.9192475310008428e-05, "loss": 0.0043, "step": 65480 }, { "grad_norm": 0.04897056147456169, "learning_rate": 2.917744151719219e-05, "loss": 0.0037, "step": 65490 }, { "grad_norm": 0.0494963675737381, "learning_rate": 2.916241000149964e-05, "loss": 0.0033, "step": 65500 }, { "grad_norm": 0.04396757483482361, "learning_rate": 2.9147380764574622e-05, "loss": 0.0034, "step": 65510 }, { "grad_norm": 0.06449053436517715, "learning_rate": 2.9132353808060674e-05, "loss": 0.0031, "step": 65520 }, { "grad_norm": 0.03519679605960846, "learning_rate": 2.9117329133601134e-05, "loss": 0.0042, "step": 65530 }, { "grad_norm": 0.07169365882873535, "learning_rate": 2.9102306742839093e-05, "loss": 0.0047, "step": 65540 }, { "grad_norm": 0.06407368928194046, "learning_rate": 2.9087286637417375e-05, "loss": 0.0043, "step": 65550 }, { "grad_norm": 0.06763901561498642, "learning_rate": 2.9072268818978537e-05, "loss": 0.005, "step": 65560 }, { "grad_norm": 0.06010938808321953, "learning_rate": 2.905725328916493e-05, "loss": 0.0043, "step": 65570 }, { "grad_norm": 0.045363038778305054, "learning_rate": 2.9042240049618608e-05, "loss": 0.0046, "step": 65580 }, { "grad_norm": 0.07387260347604752, "learning_rate": 2.9027229101981413e-05, "loss": 0.0042, "step": 65590 }, { "grad_norm": 0.08029363304376602, "learning_rate": 2.901222044789488e-05, "loss": 0.005, "step": 65600 }, { "grad_norm": 0.03699725121259689, "learning_rate": 2.89972140890004e-05, "loss": 0.006, "step": 65610 }, { "grad_norm": 0.06532886624336243, "learning_rate": 2.8982210026938988e-05, "loss": 0.003, "step": 65620 }, { "grad_norm": 0.07041723281145096, "learning_rate": 2.896720826335151e-05, "loss": 0.0029, "step": 65630 }, { "grad_norm": 0.04261178523302078, "learning_rate": 2.8952208799878493e-05, "loss": 0.0044, "step": 65640 }, { "grad_norm": 0.0952691063284874, "learning_rate": 2.89372116381603e-05, "loss": 0.0042, "step": 65650 }, { "grad_norm": 0.042656052857637405, "learning_rate": 2.8922216779836932e-05, "loss": 0.0046, "step": 65660 }, { "grad_norm": 0.05906626954674721, "learning_rate": 2.8907224226548285e-05, "loss": 0.0046, "step": 65670 }, { "grad_norm": 0.04205508530139923, "learning_rate": 2.8892233979933868e-05, "loss": 0.0032, "step": 65680 }, { "grad_norm": 0.0882800742983818, "learning_rate": 2.8877246041633026e-05, "loss": 0.0063, "step": 65690 }, { "grad_norm": 0.12873397767543793, "learning_rate": 2.886226041328478e-05, "loss": 0.005, "step": 65700 }, { "grad_norm": 0.04627613723278046, "learning_rate": 2.8847277096527957e-05, "loss": 0.0045, "step": 65710 }, { "grad_norm": 0.05296579375863075, "learning_rate": 2.8832296093001105e-05, "loss": 0.0062, "step": 65720 }, { "grad_norm": 0.04992394894361496, "learning_rate": 2.8817317404342547e-05, "loss": 0.0037, "step": 65730 }, { "grad_norm": 0.0321272611618042, "learning_rate": 2.880234103219029e-05, "loss": 0.0035, "step": 65740 }, { "grad_norm": 0.0694509893655777, "learning_rate": 2.878736697818215e-05, "loss": 0.0033, "step": 65750 }, { "grad_norm": 0.03764773905277252, "learning_rate": 2.877239524395568e-05, "loss": 0.0031, "step": 65760 }, { "grad_norm": 0.05051490291953087, "learning_rate": 2.8757425831148134e-05, "loss": 0.0031, "step": 65770 }, { "grad_norm": 0.0658084899187088, "learning_rate": 2.874245874139658e-05, "loss": 0.0055, "step": 65780 }, { "grad_norm": 0.07788246870040894, "learning_rate": 2.8727493976337737e-05, "loss": 0.0035, "step": 65790 }, { "grad_norm": 0.050147831439971924, "learning_rate": 2.8712531537608218e-05, "loss": 0.0032, "step": 65800 }, { "grad_norm": 0.07735539972782135, "learning_rate": 2.8697571426844217e-05, "loss": 0.0051, "step": 65810 }, { "grad_norm": 0.07757967710494995, "learning_rate": 2.86826136456818e-05, "loss": 0.0041, "step": 65820 }, { "grad_norm": 0.07262107729911804, "learning_rate": 2.8667658195756685e-05, "loss": 0.0038, "step": 65830 }, { "grad_norm": 0.053903691470623016, "learning_rate": 2.8652705078704412e-05, "loss": 0.0035, "step": 65840 }, { "grad_norm": 0.04275275766849518, "learning_rate": 2.863775429616018e-05, "loss": 0.0026, "step": 65850 }, { "grad_norm": 0.045111771672964096, "learning_rate": 2.862280584975906e-05, "loss": 0.0027, "step": 65860 }, { "grad_norm": 0.040582288056612015, "learning_rate": 2.8607859741135735e-05, "loss": 0.0042, "step": 65870 }, { "grad_norm": 0.05945143103599548, "learning_rate": 2.8592915971924723e-05, "loss": 0.005, "step": 65880 }, { "grad_norm": 0.057706449180841446, "learning_rate": 2.8577974543760212e-05, "loss": 0.0036, "step": 65890 }, { "grad_norm": 0.05004267394542694, "learning_rate": 2.8563035458276222e-05, "loss": 0.0044, "step": 65900 }, { "grad_norm": 0.05332145467400551, "learning_rate": 2.8548098717106424e-05, "loss": 0.0049, "step": 65910 }, { "grad_norm": 0.05409618094563484, "learning_rate": 2.85331643218843e-05, "loss": 0.0044, "step": 65920 }, { "grad_norm": 0.09202543646097183, "learning_rate": 2.8518232274243046e-05, "loss": 0.0046, "step": 65930 }, { "grad_norm": 0.04928325116634369, "learning_rate": 2.8503302575815638e-05, "loss": 0.0032, "step": 65940 }, { "grad_norm": 0.045314349234104156, "learning_rate": 2.8488375228234716e-05, "loss": 0.0028, "step": 65950 }, { "grad_norm": 0.05514518916606903, "learning_rate": 2.8473450233132738e-05, "loss": 0.0044, "step": 65960 }, { "grad_norm": 0.049166545271873474, "learning_rate": 2.8458527592141898e-05, "loss": 0.0046, "step": 65970 }, { "grad_norm": 0.07914064079523087, "learning_rate": 2.8443607306894073e-05, "loss": 0.0045, "step": 65980 }, { "grad_norm": 0.06910763680934906, "learning_rate": 2.8428689379020946e-05, "loss": 0.0041, "step": 65990 }, { "grad_norm": 0.061248790472745895, "learning_rate": 2.841377381015391e-05, "loss": 0.003, "step": 66000 }, { "grad_norm": 0.06622251123189926, "learning_rate": 2.8398860601924144e-05, "loss": 0.0025, "step": 66010 }, { "grad_norm": 0.05929822847247124, "learning_rate": 2.838394975596248e-05, "loss": 0.0034, "step": 66020 }, { "grad_norm": 0.056306254118680954, "learning_rate": 2.83690412738996e-05, "loss": 0.0026, "step": 66030 }, { "grad_norm": 0.05926838144659996, "learning_rate": 2.835413515736579e-05, "loss": 0.0033, "step": 66040 }, { "grad_norm": 0.052124496549367905, "learning_rate": 2.8339231407991267e-05, "loss": 0.0039, "step": 66050 }, { "grad_norm": 0.05680587887763977, "learning_rate": 2.832433002740581e-05, "loss": 0.0039, "step": 66060 }, { "grad_norm": 0.08725880086421967, "learning_rate": 2.8309431017239047e-05, "loss": 0.0067, "step": 66070 }, { "grad_norm": 0.09728022664785385, "learning_rate": 2.829453437912028e-05, "loss": 0.004, "step": 66080 }, { "grad_norm": 0.08736602216959, "learning_rate": 2.8279640114678617e-05, "loss": 0.0047, "step": 66090 }, { "grad_norm": 0.054005976766347885, "learning_rate": 2.8264748225542836e-05, "loss": 0.0041, "step": 66100 }, { "grad_norm": 0.05334438011050224, "learning_rate": 2.8249858713341516e-05, "loss": 0.0054, "step": 66110 }, { "grad_norm": 0.04720903933048248, "learning_rate": 2.823497157970294e-05, "loss": 0.0063, "step": 66120 }, { "grad_norm": 0.04569164663553238, "learning_rate": 2.8220086826255164e-05, "loss": 0.0036, "step": 66130 }, { "grad_norm": 0.07365353405475616, "learning_rate": 2.8205204454625934e-05, "loss": 0.0058, "step": 66140 }, { "grad_norm": 0.04733188450336456, "learning_rate": 2.8190324466442798e-05, "loss": 0.0034, "step": 66150 }, { "grad_norm": 0.06537149846553802, "learning_rate": 2.8175446863332965e-05, "loss": 0.004, "step": 66160 }, { "grad_norm": 0.06956147402524948, "learning_rate": 2.816057164692345e-05, "loss": 0.0032, "step": 66170 }, { "grad_norm": 0.0940246656537056, "learning_rate": 2.8145698818840992e-05, "loss": 0.0042, "step": 66180 }, { "grad_norm": 0.06479311734437943, "learning_rate": 2.8130828380712065e-05, "loss": 0.0062, "step": 66190 }, { "grad_norm": 0.057643283158540726, "learning_rate": 2.811596033416285e-05, "loss": 0.0037, "step": 66200 }, { "grad_norm": 0.07238063216209412, "learning_rate": 2.8101094680819307e-05, "loss": 0.0035, "step": 66210 }, { "grad_norm": 0.04301946982741356, "learning_rate": 2.808623142230715e-05, "loss": 0.0037, "step": 66220 }, { "grad_norm": 0.0657348483800888, "learning_rate": 2.8071370560251757e-05, "loss": 0.0032, "step": 66230 }, { "grad_norm": 0.05353859066963196, "learning_rate": 2.805651209627831e-05, "loss": 0.0029, "step": 66240 }, { "grad_norm": 0.0966721698641777, "learning_rate": 2.804165603201171e-05, "loss": 0.0029, "step": 66250 }, { "grad_norm": 0.05782681331038475, "learning_rate": 2.802680236907661e-05, "loss": 0.0058, "step": 66260 }, { "grad_norm": 0.06414904445409775, "learning_rate": 2.8011951109097344e-05, "loss": 0.0027, "step": 66270 }, { "grad_norm": 0.05531008914113045, "learning_rate": 2.7997102253698072e-05, "loss": 0.0029, "step": 66280 }, { "grad_norm": 0.047273147851228714, "learning_rate": 2.7982255804502588e-05, "loss": 0.0036, "step": 66290 }, { "grad_norm": 0.09908521175384521, "learning_rate": 2.7967411763134516e-05, "loss": 0.0031, "step": 66300 }, { "grad_norm": 0.05324234440922737, "learning_rate": 2.7952570131217166e-05, "loss": 0.0062, "step": 66310 }, { "grad_norm": 0.06192532181739807, "learning_rate": 2.7937730910373618e-05, "loss": 0.003, "step": 66320 }, { "grad_norm": 0.06060376390814781, "learning_rate": 2.7922894102226628e-05, "loss": 0.0051, "step": 66330 }, { "grad_norm": 0.045268502086400986, "learning_rate": 2.790805970839876e-05, "loss": 0.0042, "step": 66340 }, { "grad_norm": 0.047789786010980606, "learning_rate": 2.789322773051225e-05, "loss": 0.0029, "step": 66350 }, { "grad_norm": 0.055202171206474304, "learning_rate": 2.7878398170189124e-05, "loss": 0.0027, "step": 66360 }, { "grad_norm": 0.07758799940347672, "learning_rate": 2.7863571029051112e-05, "loss": 0.0056, "step": 66370 }, { "grad_norm": 0.05047674477100372, "learning_rate": 2.784874630871971e-05, "loss": 0.0036, "step": 66380 }, { "grad_norm": 0.06554268300533295, "learning_rate": 2.7833924010816086e-05, "loss": 0.0046, "step": 66390 }, { "grad_norm": 0.06327713280916214, "learning_rate": 2.7819104136961226e-05, "loss": 0.0042, "step": 66400 }, { "grad_norm": 0.044854652136564255, "learning_rate": 2.780428668877577e-05, "loss": 0.0031, "step": 66410 }, { "grad_norm": 0.06042871251702309, "learning_rate": 2.7789471667880152e-05, "loss": 0.0061, "step": 66420 }, { "grad_norm": 0.051133930683135986, "learning_rate": 2.7774659075894523e-05, "loss": 0.0053, "step": 66430 }, { "grad_norm": 0.04528889060020447, "learning_rate": 2.7759848914438778e-05, "loss": 0.0038, "step": 66440 }, { "grad_norm": 0.060198768973350525, "learning_rate": 2.7745041185132496e-05, "loss": 0.0028, "step": 66450 }, { "grad_norm": 0.07076423615217209, "learning_rate": 2.773023588959506e-05, "loss": 0.0069, "step": 66460 }, { "grad_norm": 0.06859242171049118, "learning_rate": 2.7715433029445555e-05, "loss": 0.0041, "step": 66470 }, { "grad_norm": 0.05959610268473625, "learning_rate": 2.7700632606302766e-05, "loss": 0.0054, "step": 66480 }, { "grad_norm": 0.07220522314310074, "learning_rate": 2.768583462178528e-05, "loss": 0.0035, "step": 66490 }, { "grad_norm": 0.0479474700987339, "learning_rate": 2.767103907751136e-05, "loss": 0.0023, "step": 66500 }, { "grad_norm": 0.033255476504564285, "learning_rate": 2.7656245975099065e-05, "loss": 0.0031, "step": 66510 }, { "grad_norm": 0.051462020725011826, "learning_rate": 2.7641455316166093e-05, "loss": 0.0041, "step": 66520 }, { "grad_norm": 0.06314203888177872, "learning_rate": 2.7626667102329973e-05, "loss": 0.0032, "step": 66530 }, { "grad_norm": 0.04383222386240959, "learning_rate": 2.761188133520788e-05, "loss": 0.0039, "step": 66540 }, { "grad_norm": 0.0793316438794136, "learning_rate": 2.7597098016416784e-05, "loss": 0.0041, "step": 66550 }, { "grad_norm": 0.07823357731103897, "learning_rate": 2.7582317147573368e-05, "loss": 0.0047, "step": 66560 }, { "grad_norm": 0.06461622565984726, "learning_rate": 2.7567538730294064e-05, "loss": 0.0027, "step": 66570 }, { "grad_norm": 0.04248431324958801, "learning_rate": 2.7552762766194973e-05, "loss": 0.0036, "step": 66580 }, { "grad_norm": 0.054873209446668625, "learning_rate": 2.753798925689202e-05, "loss": 0.0064, "step": 66590 }, { "grad_norm": 0.06407824903726578, "learning_rate": 2.7523218204000774e-05, "loss": 0.0054, "step": 66600 }, { "grad_norm": 0.07438971847295761, "learning_rate": 2.7508449609136612e-05, "loss": 0.0029, "step": 66610 }, { "grad_norm": 0.044884540140628815, "learning_rate": 2.749368347391454e-05, "loss": 0.0026, "step": 66620 }, { "grad_norm": 0.07647120952606201, "learning_rate": 2.7478919799949443e-05, "loss": 0.0028, "step": 66630 }, { "grad_norm": 0.06724678725004196, "learning_rate": 2.7464158588855803e-05, "loss": 0.0033, "step": 66640 }, { "grad_norm": 0.06097293645143509, "learning_rate": 2.7449399842247918e-05, "loss": 0.0041, "step": 66650 }, { "grad_norm": 0.06795478612184525, "learning_rate": 2.7434643561739736e-05, "loss": 0.0033, "step": 66660 }, { "grad_norm": 0.051918040961027145, "learning_rate": 2.7419889748945016e-05, "loss": 0.0036, "step": 66670 }, { "grad_norm": 0.08252724260091782, "learning_rate": 2.7405138405477203e-05, "loss": 0.0042, "step": 66680 }, { "grad_norm": 0.06865986436605453, "learning_rate": 2.73903895329495e-05, "loss": 0.0041, "step": 66690 }, { "grad_norm": 0.054522912949323654, "learning_rate": 2.737564313297478e-05, "loss": 0.0031, "step": 66700 }, { "grad_norm": 0.04621798172593117, "learning_rate": 2.736089920716572e-05, "loss": 0.0024, "step": 66710 }, { "grad_norm": 0.09526080638170242, "learning_rate": 2.7346157757134706e-05, "loss": 0.0053, "step": 66720 }, { "grad_norm": 0.05585109069943428, "learning_rate": 2.733141878449379e-05, "loss": 0.0044, "step": 66730 }, { "grad_norm": 0.04527711495757103, "learning_rate": 2.7316682290854844e-05, "loss": 0.0034, "step": 66740 }, { "grad_norm": 0.07329697906970978, "learning_rate": 2.7301948277829414e-05, "loss": 0.0068, "step": 66750 }, { "grad_norm": 0.06093493849039078, "learning_rate": 2.7287216747028814e-05, "loss": 0.0051, "step": 66760 }, { "grad_norm": 0.07571981102228165, "learning_rate": 2.7272487700064025e-05, "loss": 0.0051, "step": 66770 }, { "grad_norm": 0.08027268946170807, "learning_rate": 2.725776113854583e-05, "loss": 0.0038, "step": 66780 }, { "grad_norm": 0.060301102697849274, "learning_rate": 2.7243037064084664e-05, "loss": 0.005, "step": 66790 }, { "grad_norm": 0.06734581291675568, "learning_rate": 2.7228315478290766e-05, "loss": 0.0039, "step": 66800 }, { "grad_norm": 0.04233279079198837, "learning_rate": 2.721359638277402e-05, "loss": 0.0047, "step": 66810 }, { "grad_norm": 0.10575257241725922, "learning_rate": 2.7198879779144148e-05, "loss": 0.0036, "step": 66820 }, { "grad_norm": 0.08259551972150803, "learning_rate": 2.7184165669010485e-05, "loss": 0.0033, "step": 66830 }, { "grad_norm": 0.061241816729307175, "learning_rate": 2.716945405398218e-05, "loss": 0.0043, "step": 66840 }, { "grad_norm": 0.06453771889209747, "learning_rate": 2.715474493566803e-05, "loss": 0.0039, "step": 66850 }, { "grad_norm": 0.06573674082756042, "learning_rate": 2.7140038315676652e-05, "loss": 0.0055, "step": 66860 }, { "grad_norm": 0.05406190827488899, "learning_rate": 2.7125334195616275e-05, "loss": 0.0046, "step": 66870 }, { "grad_norm": 0.05361739173531532, "learning_rate": 2.7110632577094997e-05, "loss": 0.0033, "step": 66880 }, { "grad_norm": 0.0428115651011467, "learning_rate": 2.709593346172051e-05, "loss": 0.0055, "step": 66890 }, { "grad_norm": 0.06116181239485741, "learning_rate": 2.708123685110032e-05, "loss": 0.0048, "step": 66900 }, { "grad_norm": 0.05139556899666786, "learning_rate": 2.7066542746841588e-05, "loss": 0.0031, "step": 66910 }, { "grad_norm": 0.04135127365589142, "learning_rate": 2.7051851150551267e-05, "loss": 0.0052, "step": 66920 }, { "grad_norm": 0.08162513375282288, "learning_rate": 2.7037162063835996e-05, "loss": 0.0072, "step": 66930 }, { "grad_norm": 0.04418955370783806, "learning_rate": 2.702247548830218e-05, "loss": 0.0036, "step": 66940 }, { "grad_norm": 0.06477481871843338, "learning_rate": 2.7007791425555883e-05, "loss": 0.0024, "step": 66950 }, { "grad_norm": 0.05569041520357132, "learning_rate": 2.6993109877202945e-05, "loss": 0.0045, "step": 66960 }, { "grad_norm": 0.047119177877902985, "learning_rate": 2.6978430844848935e-05, "loss": 0.0032, "step": 66970 }, { "grad_norm": 0.05063151940703392, "learning_rate": 2.69637543300991e-05, "loss": 0.0025, "step": 66980 }, { "grad_norm": 0.05972612649202347, "learning_rate": 2.694908033455848e-05, "loss": 0.0032, "step": 66990 }, { "grad_norm": 0.06194920465350151, "learning_rate": 2.6934408859831733e-05, "loss": 0.0028, "step": 67000 }, { "grad_norm": 0.06021516025066376, "learning_rate": 2.69197399075234e-05, "loss": 0.0045, "step": 67010 }, { "grad_norm": 0.038970667868852615, "learning_rate": 2.6905073479237584e-05, "loss": 0.0043, "step": 67020 }, { "grad_norm": 0.08149518817663193, "learning_rate": 2.6890409576578235e-05, "loss": 0.0049, "step": 67030 }, { "grad_norm": 0.040818121284246445, "learning_rate": 2.687574820114892e-05, "loss": 0.0038, "step": 67040 }, { "grad_norm": 0.06895624846220016, "learning_rate": 2.686108935455305e-05, "loss": 0.0035, "step": 67050 }, { "grad_norm": 0.0692279264330864, "learning_rate": 2.6846433038393603e-05, "loss": 0.0046, "step": 67060 }, { "grad_norm": 0.05620422586798668, "learning_rate": 2.6831779254273482e-05, "loss": 0.0038, "step": 67070 }, { "grad_norm": 0.046399082988500595, "learning_rate": 2.6817128003795122e-05, "loss": 0.0029, "step": 67080 }, { "grad_norm": 0.05095384642481804, "learning_rate": 2.680247928856081e-05, "loss": 0.0051, "step": 67090 }, { "grad_norm": 0.07991720736026764, "learning_rate": 2.6787833110172466e-05, "loss": 0.0037, "step": 67100 }, { "grad_norm": 0.05309341475367546, "learning_rate": 2.6773189470231807e-05, "loss": 0.0043, "step": 67110 }, { "grad_norm": 0.05541111156344414, "learning_rate": 2.675854837034022e-05, "loss": 0.0036, "step": 67120 }, { "grad_norm": 0.061003800481557846, "learning_rate": 2.6743909812098833e-05, "loss": 0.0028, "step": 67130 }, { "grad_norm": 0.06932616233825684, "learning_rate": 2.67292737971085e-05, "loss": 0.0032, "step": 67140 }, { "grad_norm": 0.04305044189095497, "learning_rate": 2.671464032696982e-05, "loss": 0.0039, "step": 67150 }, { "grad_norm": 0.057446230202913284, "learning_rate": 2.6700009403283055e-05, "loss": 0.0057, "step": 67160 }, { "grad_norm": 0.06243153288960457, "learning_rate": 2.6685381027648214e-05, "loss": 0.0034, "step": 67170 }, { "grad_norm": 0.0349823497235775, "learning_rate": 2.6670755201665086e-05, "loss": 0.0033, "step": 67180 }, { "grad_norm": 0.07753307372331619, "learning_rate": 2.6656131926933066e-05, "loss": 0.0039, "step": 67190 }, { "grad_norm": 0.0701775923371315, "learning_rate": 2.664151120505136e-05, "loss": 0.0037, "step": 67200 }, { "grad_norm": 0.06438162177801132, "learning_rate": 2.6626893037618873e-05, "loss": 0.0041, "step": 67210 }, { "grad_norm": 0.06425192952156067, "learning_rate": 2.6612277426234243e-05, "loss": 0.0039, "step": 67220 }, { "grad_norm": 0.06583108007907867, "learning_rate": 2.659766437249577e-05, "loss": 0.0037, "step": 67230 }, { "grad_norm": 0.06179172173142433, "learning_rate": 2.6583053878001557e-05, "loss": 0.0039, "step": 67240 }, { "grad_norm": 0.0508117750287056, "learning_rate": 2.6568445944349323e-05, "loss": 0.0032, "step": 67250 }, { "grad_norm": 0.08199185132980347, "learning_rate": 2.6553840573136657e-05, "loss": 0.0036, "step": 67260 }, { "grad_norm": 0.05321452394127846, "learning_rate": 2.6539237765960706e-05, "loss": 0.0038, "step": 67270 }, { "grad_norm": 0.05103109031915665, "learning_rate": 2.6524637524418465e-05, "loss": 0.0063, "step": 67280 }, { "grad_norm": 0.06146305799484253, "learning_rate": 2.651003985010655e-05, "loss": 0.007, "step": 67290 }, { "grad_norm": 0.06244271621108055, "learning_rate": 2.649544474462138e-05, "loss": 0.0051, "step": 67300 }, { "grad_norm": 0.046733103692531586, "learning_rate": 2.6480852209559015e-05, "loss": 0.0038, "step": 67310 }, { "grad_norm": 0.05837835371494293, "learning_rate": 2.6466262246515296e-05, "loss": 0.0029, "step": 67320 }, { "grad_norm": 0.06436920166015625, "learning_rate": 2.6451674857085746e-05, "loss": 0.0031, "step": 67330 }, { "grad_norm": 0.06239737942814827, "learning_rate": 2.6437090042865655e-05, "loss": 0.004, "step": 67340 }, { "grad_norm": 0.07588529586791992, "learning_rate": 2.6422507805449947e-05, "loss": 0.0043, "step": 67350 }, { "grad_norm": 0.058702897280454636, "learning_rate": 2.6407928146433357e-05, "loss": 0.0036, "step": 67360 }, { "grad_norm": 0.0628959983587265, "learning_rate": 2.639335106741026e-05, "loss": 0.0034, "step": 67370 }, { "grad_norm": 0.040305349975824356, "learning_rate": 2.6378776569974796e-05, "loss": 0.0043, "step": 67380 }, { "grad_norm": 0.04865231364965439, "learning_rate": 2.6364204655720814e-05, "loss": 0.0043, "step": 67390 }, { "grad_norm": 0.0393994078040123, "learning_rate": 2.6349635326241894e-05, "loss": 0.0029, "step": 67400 }, { "grad_norm": 0.06433365494012833, "learning_rate": 2.633506858313127e-05, "loss": 0.0046, "step": 67410 }, { "grad_norm": 0.0458991639316082, "learning_rate": 2.6320504427981975e-05, "loss": 0.0028, "step": 67420 }, { "grad_norm": 0.09255966544151306, "learning_rate": 2.6305942862386734e-05, "loss": 0.0052, "step": 67430 }, { "grad_norm": 0.08168124407529831, "learning_rate": 2.629138388793793e-05, "loss": 0.0032, "step": 67440 }, { "grad_norm": 0.07076341658830643, "learning_rate": 2.6276827506227743e-05, "loss": 0.0037, "step": 67450 }, { "grad_norm": 0.06722094118595123, "learning_rate": 2.6262273718848037e-05, "loss": 0.0036, "step": 67460 }, { "grad_norm": 0.06632629036903381, "learning_rate": 2.6247722527390407e-05, "loss": 0.0034, "step": 67470 }, { "grad_norm": 0.058481503278017044, "learning_rate": 2.6233173933446104e-05, "loss": 0.0059, "step": 67480 }, { "grad_norm": 0.057955291122198105, "learning_rate": 2.6218627938606188e-05, "loss": 0.0063, "step": 67490 }, { "grad_norm": 0.04731302335858345, "learning_rate": 2.620408454446135e-05, "loss": 0.0035, "step": 67500 }, { "grad_norm": 0.07294480502605438, "learning_rate": 2.6189543752602046e-05, "loss": 0.0033, "step": 67510 }, { "grad_norm": 0.05072832107543945, "learning_rate": 2.6175005564618444e-05, "loss": 0.0054, "step": 67520 }, { "grad_norm": 0.07435009628534317, "learning_rate": 2.6160469982100428e-05, "loss": 0.0053, "step": 67530 }, { "grad_norm": 0.045408133417367935, "learning_rate": 2.6145937006637555e-05, "loss": 0.0025, "step": 67540 }, { "grad_norm": 0.06682887673377991, "learning_rate": 2.6131406639819166e-05, "loss": 0.0027, "step": 67550 }, { "grad_norm": 0.05296851694583893, "learning_rate": 2.611687888323424e-05, "loss": 0.0031, "step": 67560 }, { "grad_norm": 0.03868510201573372, "learning_rate": 2.610235373847153e-05, "loss": 0.0029, "step": 67570 }, { "grad_norm": 0.06508981436491013, "learning_rate": 2.6087831207119494e-05, "loss": 0.0035, "step": 67580 }, { "grad_norm": 0.05713155120611191, "learning_rate": 2.60733112907663e-05, "loss": 0.0037, "step": 67590 }, { "grad_norm": 0.0710706114768982, "learning_rate": 2.6058793990999786e-05, "loss": 0.0034, "step": 67600 }, { "grad_norm": 0.07484403997659683, "learning_rate": 2.6044279309407593e-05, "loss": 0.0035, "step": 67610 }, { "grad_norm": 0.07348080724477768, "learning_rate": 2.6029767247576976e-05, "loss": 0.0039, "step": 67620 }, { "grad_norm": 0.06586482375860214, "learning_rate": 2.601525780709497e-05, "loss": 0.0046, "step": 67630 }, { "grad_norm": 0.06863074004650116, "learning_rate": 2.6000750989548318e-05, "loss": 0.0046, "step": 67640 }, { "grad_norm": 0.0647508054971695, "learning_rate": 2.5986246796523462e-05, "loss": 0.0029, "step": 67650 }, { "grad_norm": 0.038102880120277405, "learning_rate": 2.5971745229606546e-05, "loss": 0.0034, "step": 67660 }, { "grad_norm": 0.035704355686903, "learning_rate": 2.5957246290383442e-05, "loss": 0.0048, "step": 67670 }, { "grad_norm": 0.07295475155115128, "learning_rate": 2.594274998043975e-05, "loss": 0.0055, "step": 67680 }, { "grad_norm": 0.081696055829525, "learning_rate": 2.5928256301360733e-05, "loss": 0.0039, "step": 67690 }, { "grad_norm": 0.0638905018568039, "learning_rate": 2.591376525473142e-05, "loss": 0.0056, "step": 67700 }, { "grad_norm": 0.07885975390672684, "learning_rate": 2.5899276842136523e-05, "loss": 0.004, "step": 67710 }, { "grad_norm": 0.10991445928812027, "learning_rate": 2.5884791065160495e-05, "loss": 0.0049, "step": 67720 }, { "grad_norm": 0.07466631382703781, "learning_rate": 2.5870307925387436e-05, "loss": 0.0051, "step": 67730 }, { "grad_norm": 0.07699732482433319, "learning_rate": 2.585582742440125e-05, "loss": 0.0037, "step": 67740 }, { "grad_norm": 0.07211155444383621, "learning_rate": 2.5841349563785455e-05, "loss": 0.0043, "step": 67750 }, { "grad_norm": 0.0740102231502533, "learning_rate": 2.5826874345123355e-05, "loss": 0.0057, "step": 67760 }, { "grad_norm": 0.07800935953855515, "learning_rate": 2.581240176999792e-05, "loss": 0.0037, "step": 67770 }, { "grad_norm": 0.07050707936286926, "learning_rate": 2.579793183999189e-05, "loss": 0.0033, "step": 67780 }, { "grad_norm": 0.0685257613658905, "learning_rate": 2.5783464556687627e-05, "loss": 0.0035, "step": 67790 }, { "grad_norm": 0.06035678833723068, "learning_rate": 2.576899992166729e-05, "loss": 0.0037, "step": 67800 }, { "grad_norm": 0.048257309943437576, "learning_rate": 2.5754537936512668e-05, "loss": 0.0067, "step": 67810 }, { "grad_norm": 0.035390693694353104, "learning_rate": 2.574007860280535e-05, "loss": 0.0029, "step": 67820 }, { "grad_norm": 0.062107980251312256, "learning_rate": 2.5725621922126518e-05, "loss": 0.0047, "step": 67830 }, { "grad_norm": 0.08706670999526978, "learning_rate": 2.571116789605722e-05, "loss": 0.0035, "step": 67840 }, { "grad_norm": 0.08369853347539902, "learning_rate": 2.5696716526178054e-05, "loss": 0.0037, "step": 67850 }, { "grad_norm": 0.0579906702041626, "learning_rate": 2.568226781406946e-05, "loss": 0.0022, "step": 67860 }, { "grad_norm": 0.09011044353246689, "learning_rate": 2.5667821761311474e-05, "loss": 0.0042, "step": 67870 }, { "grad_norm": 0.040567561984062195, "learning_rate": 2.5653378369483917e-05, "loss": 0.0032, "step": 67880 }, { "grad_norm": 0.05458299443125725, "learning_rate": 2.5638937640166292e-05, "loss": 0.0046, "step": 67890 }, { "grad_norm": 0.08338510245084763, "learning_rate": 2.5624499574937842e-05, "loss": 0.0034, "step": 67900 }, { "grad_norm": 0.04783927649259567, "learning_rate": 2.5610064175377456e-05, "loss": 0.0066, "step": 67910 }, { "grad_norm": 0.06315626949071884, "learning_rate": 2.5595631443063782e-05, "loss": 0.0035, "step": 67920 }, { "grad_norm": 0.06744649261236191, "learning_rate": 2.5581201379575177e-05, "loss": 0.0031, "step": 67930 }, { "grad_norm": 0.050106484442949295, "learning_rate": 2.5566773986489655e-05, "loss": 0.0039, "step": 67940 }, { "grad_norm": 0.05562691390514374, "learning_rate": 2.5552349265385e-05, "loss": 0.0038, "step": 67950 }, { "grad_norm": 0.053996481001377106, "learning_rate": 2.5537927217838675e-05, "loss": 0.003, "step": 67960 }, { "grad_norm": 0.03447466343641281, "learning_rate": 2.5523507845427864e-05, "loss": 0.0049, "step": 67970 }, { "grad_norm": 0.05424566939473152, "learning_rate": 2.5509091149729413e-05, "loss": 0.0033, "step": 67980 }, { "grad_norm": 0.04725269228219986, "learning_rate": 2.5494677132319962e-05, "loss": 0.0042, "step": 67990 }, { "grad_norm": 0.051272422075271606, "learning_rate": 2.548026579477575e-05, "loss": 0.0055, "step": 68000 }, { "grad_norm": 0.05429557338356972, "learning_rate": 2.5465857138672834e-05, "loss": 0.0037, "step": 68010 }, { "grad_norm": 0.04515499621629715, "learning_rate": 2.5451451165586848e-05, "loss": 0.002, "step": 68020 }, { "grad_norm": 0.028360258787870407, "learning_rate": 2.543704787709329e-05, "loss": 0.0025, "step": 68030 }, { "grad_norm": 0.05117388442158699, "learning_rate": 2.542264727476723e-05, "loss": 0.0044, "step": 68040 }, { "grad_norm": 0.08374299854040146, "learning_rate": 2.5408249360183535e-05, "loss": 0.0043, "step": 68050 }, { "grad_norm": 0.07245971262454987, "learning_rate": 2.5393854134916694e-05, "loss": 0.0035, "step": 68060 }, { "grad_norm": 0.06219179183244705, "learning_rate": 2.5379461600540993e-05, "loss": 0.0034, "step": 68070 }, { "grad_norm": 0.0661109983921051, "learning_rate": 2.5365071758630317e-05, "loss": 0.0063, "step": 68080 }, { "grad_norm": 0.03661397844552994, "learning_rate": 2.535068461075839e-05, "loss": 0.004, "step": 68090 }, { "grad_norm": 0.07548267394304276, "learning_rate": 2.5336300158498516e-05, "loss": 0.0033, "step": 68100 }, { "grad_norm": 0.05744712054729462, "learning_rate": 2.5321918403423793e-05, "loss": 0.0035, "step": 68110 }, { "grad_norm": 0.04060915485024452, "learning_rate": 2.5307539347106957e-05, "loss": 0.0031, "step": 68120 }, { "grad_norm": 0.07490085810422897, "learning_rate": 2.529316299112049e-05, "loss": 0.0041, "step": 68130 }, { "grad_norm": 0.08653891086578369, "learning_rate": 2.5278789337036567e-05, "loss": 0.0042, "step": 68140 }, { "grad_norm": 0.06514443457126617, "learning_rate": 2.52644183864271e-05, "loss": 0.0033, "step": 68150 }, { "grad_norm": 0.058287084102630615, "learning_rate": 2.525005014086363e-05, "loss": 0.0036, "step": 68160 }, { "grad_norm": 0.04715602472424507, "learning_rate": 2.5235684601917465e-05, "loss": 0.0037, "step": 68170 }, { "grad_norm": 0.06172798201441765, "learning_rate": 2.5221321771159618e-05, "loss": 0.0035, "step": 68180 }, { "grad_norm": 0.056487251073122025, "learning_rate": 2.5206961650160743e-05, "loss": 0.0026, "step": 68190 }, { "grad_norm": 0.05069964751601219, "learning_rate": 2.519260424049129e-05, "loss": 0.0045, "step": 68200 }, { "grad_norm": 0.08182637393474579, "learning_rate": 2.5178249543721304e-05, "loss": 0.0048, "step": 68210 }, { "grad_norm": 0.04743756353855133, "learning_rate": 2.5163897561420668e-05, "loss": 0.0053, "step": 68220 }, { "grad_norm": 0.04890985041856766, "learning_rate": 2.514954829515883e-05, "loss": 0.0043, "step": 68230 }, { "grad_norm": 0.06903751939535141, "learning_rate": 2.513520174650505e-05, "loss": 0.0051, "step": 68240 }, { "grad_norm": 0.07861047238111496, "learning_rate": 2.5120857917028207e-05, "loss": 0.0048, "step": 68250 }, { "grad_norm": 0.05347684398293495, "learning_rate": 2.5106516808296953e-05, "loss": 0.0031, "step": 68260 }, { "grad_norm": 0.059041932225227356, "learning_rate": 2.5092178421879552e-05, "loss": 0.0031, "step": 68270 }, { "grad_norm": 0.05511687695980072, "learning_rate": 2.5077842759344116e-05, "loss": 0.0037, "step": 68280 }, { "grad_norm": 0.05603078380227089, "learning_rate": 2.5063509822258304e-05, "loss": 0.0056, "step": 68290 }, { "grad_norm": 0.05722152441740036, "learning_rate": 2.5049179612189576e-05, "loss": 0.0032, "step": 68300 }, { "grad_norm": 0.05535028874874115, "learning_rate": 2.5034852130705043e-05, "loss": 0.0041, "step": 68310 }, { "grad_norm": 0.06612256914377213, "learning_rate": 2.5020527379371572e-05, "loss": 0.004, "step": 68320 }, { "grad_norm": 0.05747311934828758, "learning_rate": 2.500620535975564e-05, "loss": 0.0038, "step": 68330 }, { "grad_norm": 0.07395424693822861, "learning_rate": 2.4991886073423515e-05, "loss": 0.0039, "step": 68340 }, { "grad_norm": 0.053540706634521484, "learning_rate": 2.4977569521941134e-05, "loss": 0.0027, "step": 68350 }, { "grad_norm": 0.03256598487496376, "learning_rate": 2.4963255706874143e-05, "loss": 0.0036, "step": 68360 }, { "grad_norm": 0.08193137496709824, "learning_rate": 2.4948944629787853e-05, "loss": 0.0042, "step": 68370 }, { "grad_norm": 0.0528230145573616, "learning_rate": 2.493463629224731e-05, "loss": 0.0031, "step": 68380 }, { "grad_norm": 0.05773645266890526, "learning_rate": 2.492033069581728e-05, "loss": 0.0024, "step": 68390 }, { "grad_norm": 0.19038468599319458, "learning_rate": 2.4906027842062137e-05, "loss": 0.0039, "step": 68400 }, { "grad_norm": 0.060079414397478104, "learning_rate": 2.48917277325461e-05, "loss": 0.0039, "step": 68410 }, { "grad_norm": 0.05993882194161415, "learning_rate": 2.487743036883295e-05, "loss": 0.0047, "step": 68420 }, { "grad_norm": 0.07461880147457123, "learning_rate": 2.486313575248626e-05, "loss": 0.0027, "step": 68430 }, { "grad_norm": 0.0611131452023983, "learning_rate": 2.4848843885069235e-05, "loss": 0.0026, "step": 68440 }, { "grad_norm": 0.08400265872478485, "learning_rate": 2.483455476814484e-05, "loss": 0.0029, "step": 68450 }, { "grad_norm": 0.05212588235735893, "learning_rate": 2.4820268403275672e-05, "loss": 0.0032, "step": 68460 }, { "grad_norm": 0.03316480293869972, "learning_rate": 2.480598479202413e-05, "loss": 0.0027, "step": 68470 }, { "grad_norm": 0.04364202916622162, "learning_rate": 2.4791703935952193e-05, "loss": 0.0035, "step": 68480 }, { "grad_norm": 0.052495524287223816, "learning_rate": 2.477742583662163e-05, "loss": 0.0034, "step": 68490 }, { "grad_norm": 0.052921514958143234, "learning_rate": 2.476315049559384e-05, "loss": 0.0044, "step": 68500 }, { "grad_norm": 0.05948375165462494, "learning_rate": 2.4748877914429987e-05, "loss": 0.0073, "step": 68510 }, { "grad_norm": 0.06439831107854843, "learning_rate": 2.4734608094690857e-05, "loss": 0.0037, "step": 68520 }, { "grad_norm": 0.08477562665939331, "learning_rate": 2.4720341037937006e-05, "loss": 0.0027, "step": 68530 }, { "grad_norm": 0.07968771457672119, "learning_rate": 2.4706076745728645e-05, "loss": 0.0026, "step": 68540 }, { "grad_norm": 0.05333411321043968, "learning_rate": 2.469181521962573e-05, "loss": 0.0032, "step": 68550 }, { "grad_norm": 0.062497448176145554, "learning_rate": 2.467755646118783e-05, "loss": 0.0041, "step": 68560 }, { "grad_norm": 0.049727726727724075, "learning_rate": 2.46633004719743e-05, "loss": 0.0042, "step": 68570 }, { "grad_norm": 0.05134141445159912, "learning_rate": 2.4649047253544123e-05, "loss": 0.0032, "step": 68580 }, { "grad_norm": 0.060374557971954346, "learning_rate": 2.4634796807456016e-05, "loss": 0.0028, "step": 68590 }, { "grad_norm": 0.04918896406888962, "learning_rate": 2.46205491352684e-05, "loss": 0.005, "step": 68600 }, { "grad_norm": 0.05797470360994339, "learning_rate": 2.4606304238539375e-05, "loss": 0.0028, "step": 68610 }, { "grad_norm": 0.04654047638177872, "learning_rate": 2.4592062118826753e-05, "loss": 0.0038, "step": 68620 }, { "grad_norm": 0.04620082676410675, "learning_rate": 2.4577822777688004e-05, "loss": 0.0027, "step": 68630 }, { "grad_norm": 0.05627596378326416, "learning_rate": 2.4563586216680347e-05, "loss": 0.0052, "step": 68640 }, { "grad_norm": 0.0724298283457756, "learning_rate": 2.4549352437360622e-05, "loss": 0.004, "step": 68650 }, { "grad_norm": 0.05470665916800499, "learning_rate": 2.4535121441285493e-05, "loss": 0.0031, "step": 68660 }, { "grad_norm": 0.08771094679832458, "learning_rate": 2.4520893230011172e-05, "loss": 0.0042, "step": 68670 }, { "grad_norm": 0.06253550201654434, "learning_rate": 2.450666780509368e-05, "loss": 0.003, "step": 68680 }, { "grad_norm": 0.0693158432841301, "learning_rate": 2.4492445168088656e-05, "loss": 0.0029, "step": 68690 }, { "grad_norm": 0.04898999631404877, "learning_rate": 2.447822532055149e-05, "loss": 0.0029, "step": 68700 }, { "grad_norm": 0.07568448781967163, "learning_rate": 2.4464008264037212e-05, "loss": 0.0058, "step": 68710 }, { "grad_norm": 0.057164888828992844, "learning_rate": 2.4449794000100605e-05, "loss": 0.0053, "step": 68720 }, { "grad_norm": 0.06866094470024109, "learning_rate": 2.4435582530296108e-05, "loss": 0.0031, "step": 68730 }, { "grad_norm": 0.062408480793237686, "learning_rate": 2.4421373856177887e-05, "loss": 0.0051, "step": 68740 }, { "grad_norm": 0.05955486372113228, "learning_rate": 2.440716797929975e-05, "loss": 0.0036, "step": 68750 }, { "grad_norm": 0.07111980766057968, "learning_rate": 2.439296490121526e-05, "loss": 0.003, "step": 68760 }, { "grad_norm": 0.06514578312635422, "learning_rate": 2.4378764623477608e-05, "loss": 0.0049, "step": 68770 }, { "grad_norm": 0.04254527762532234, "learning_rate": 2.436456714763974e-05, "loss": 0.0025, "step": 68780 }, { "grad_norm": 0.07544425874948502, "learning_rate": 2.4350372475254262e-05, "loss": 0.0031, "step": 68790 }, { "grad_norm": 0.054649144411087036, "learning_rate": 2.433618060787351e-05, "loss": 0.0036, "step": 68800 }, { "grad_norm": 0.042592573910951614, "learning_rate": 2.432199154704944e-05, "loss": 0.0037, "step": 68810 }, { "grad_norm": 0.041662149131298065, "learning_rate": 2.430780529433377e-05, "loss": 0.0051, "step": 68820 }, { "grad_norm": 0.061887968331575394, "learning_rate": 2.42936218512779e-05, "loss": 0.0064, "step": 68830 }, { "grad_norm": 0.05305812880396843, "learning_rate": 2.427944121943288e-05, "loss": 0.004, "step": 68840 }, { "grad_norm": 0.043042294681072235, "learning_rate": 2.4265263400349497e-05, "loss": 0.0037, "step": 68850 }, { "grad_norm": 0.08749192208051682, "learning_rate": 2.4251088395578213e-05, "loss": 0.0052, "step": 68860 }, { "grad_norm": 0.04031553119421005, "learning_rate": 2.4236916206669214e-05, "loss": 0.0036, "step": 68870 }, { "grad_norm": 0.03742007911205292, "learning_rate": 2.4222746835172304e-05, "loss": 0.0028, "step": 68880 }, { "grad_norm": 0.06019243225455284, "learning_rate": 2.4208580282637065e-05, "loss": 0.0045, "step": 68890 }, { "grad_norm": 0.05743729695677757, "learning_rate": 2.4194416550612687e-05, "loss": 0.0036, "step": 68900 }, { "grad_norm": 0.07137097418308258, "learning_rate": 2.418025564064812e-05, "loss": 0.0032, "step": 68910 }, { "grad_norm": 0.04843303561210632, "learning_rate": 2.4166097554291978e-05, "loss": 0.004, "step": 68920 }, { "grad_norm": 0.07745169848203659, "learning_rate": 2.4151942293092583e-05, "loss": 0.0032, "step": 68930 }, { "grad_norm": 0.07193559408187866, "learning_rate": 2.41377898585979e-05, "loss": 0.0051, "step": 68940 }, { "grad_norm": 0.10362565517425537, "learning_rate": 2.4123640252355652e-05, "loss": 0.0046, "step": 68950 }, { "grad_norm": 0.04717939719557762, "learning_rate": 2.410949347591318e-05, "loss": 0.003, "step": 68960 }, { "grad_norm": 0.06291782855987549, "learning_rate": 2.4095349530817578e-05, "loss": 0.0042, "step": 68970 }, { "grad_norm": 0.07155065983533859, "learning_rate": 2.4081208418615603e-05, "loss": 0.0038, "step": 68980 }, { "grad_norm": 0.04211590439081192, "learning_rate": 2.4067070140853725e-05, "loss": 0.0037, "step": 68990 }, { "grad_norm": 0.06858491897583008, "learning_rate": 2.405293469907805e-05, "loss": 0.0045, "step": 69000 }, { "grad_norm": 0.05700577422976494, "learning_rate": 2.4038802094834446e-05, "loss": 0.0039, "step": 69010 }, { "grad_norm": 0.09750714898109436, "learning_rate": 2.4024672329668395e-05, "loss": 0.0041, "step": 69020 }, { "grad_norm": 0.06894423812627792, "learning_rate": 2.4010545405125125e-05, "loss": 0.0032, "step": 69030 }, { "grad_norm": 0.06929809600114822, "learning_rate": 2.3996421322749528e-05, "loss": 0.0025, "step": 69040 }, { "grad_norm": 0.06021217629313469, "learning_rate": 2.3982300084086224e-05, "loss": 0.0054, "step": 69050 }, { "grad_norm": 0.07485062628984451, "learning_rate": 2.396818169067944e-05, "loss": 0.0028, "step": 69060 }, { "grad_norm": 0.05252905562520027, "learning_rate": 2.3954066144073178e-05, "loss": 0.0024, "step": 69070 }, { "grad_norm": 0.058340612798929214, "learning_rate": 2.39399534458111e-05, "loss": 0.0031, "step": 69080 }, { "grad_norm": 0.04214544594287872, "learning_rate": 2.392584359743651e-05, "loss": 0.0029, "step": 69090 }, { "grad_norm": 0.034884262830019, "learning_rate": 2.3911736600492463e-05, "loss": 0.004, "step": 69100 }, { "grad_norm": 0.0810484066605568, "learning_rate": 2.3897632456521672e-05, "loss": 0.0037, "step": 69110 }, { "grad_norm": 0.06910083442926407, "learning_rate": 2.388353116706658e-05, "loss": 0.0032, "step": 69120 }, { "grad_norm": 0.0529010184109211, "learning_rate": 2.3869432733669232e-05, "loss": 0.0035, "step": 69130 }, { "grad_norm": 0.06217228248715401, "learning_rate": 2.385533715787145e-05, "loss": 0.0053, "step": 69140 }, { "grad_norm": 0.05758824199438095, "learning_rate": 2.3841244441214673e-05, "loss": 0.0025, "step": 69150 }, { "grad_norm": 0.08089102804660797, "learning_rate": 2.3827154585240073e-05, "loss": 0.0054, "step": 69160 }, { "grad_norm": 0.0487816259264946, "learning_rate": 2.3813067591488498e-05, "loss": 0.0033, "step": 69170 }, { "grad_norm": 0.02648666873574257, "learning_rate": 2.3798983461500494e-05, "loss": 0.0022, "step": 69180 }, { "grad_norm": 0.09599839150905609, "learning_rate": 2.3784902196816256e-05, "loss": 0.0032, "step": 69190 }, { "grad_norm": 0.04705299437046051, "learning_rate": 2.3770823798975716e-05, "loss": 0.003, "step": 69200 }, { "grad_norm": 0.049979280680418015, "learning_rate": 2.3756748269518424e-05, "loss": 0.0035, "step": 69210 }, { "grad_norm": 0.037595294415950775, "learning_rate": 2.374267560998371e-05, "loss": 0.0026, "step": 69220 }, { "grad_norm": 0.07679342478513718, "learning_rate": 2.3728605821910477e-05, "loss": 0.0041, "step": 69230 }, { "grad_norm": 0.04003475606441498, "learning_rate": 2.3714538906837453e-05, "loss": 0.0034, "step": 69240 }, { "grad_norm": 0.05474553629755974, "learning_rate": 2.3700474866302913e-05, "loss": 0.0037, "step": 69250 }, { "grad_norm": 0.035927943885326385, "learning_rate": 2.3686413701844917e-05, "loss": 0.0034, "step": 69260 }, { "grad_norm": 0.06870471686124802, "learning_rate": 2.367235541500114e-05, "loss": 0.0039, "step": 69270 }, { "grad_norm": 0.040030766278505325, "learning_rate": 2.3658300007308993e-05, "loss": 0.0044, "step": 69280 }, { "grad_norm": 0.0635734498500824, "learning_rate": 2.364424748030555e-05, "loss": 0.0035, "step": 69290 }, { "grad_norm": 0.06348370760679245, "learning_rate": 2.3630197835527595e-05, "loss": 0.0041, "step": 69300 }, { "grad_norm": 0.07733673602342606, "learning_rate": 2.3616151074511546e-05, "loss": 0.0038, "step": 69310 }, { "grad_norm": 0.046659454703330994, "learning_rate": 2.3602107198793545e-05, "loss": 0.0032, "step": 69320 }, { "grad_norm": 0.053672485053539276, "learning_rate": 2.3588066209909427e-05, "loss": 0.0041, "step": 69330 }, { "grad_norm": 0.06619108468294144, "learning_rate": 2.3574028109394662e-05, "loss": 0.0033, "step": 69340 }, { "grad_norm": 0.036890216171741486, "learning_rate": 2.3559992898784454e-05, "loss": 0.0025, "step": 69350 }, { "grad_norm": 0.06099271401762962, "learning_rate": 2.354596057961367e-05, "loss": 0.0038, "step": 69360 }, { "grad_norm": 0.052514031529426575, "learning_rate": 2.353193115341688e-05, "loss": 0.0026, "step": 69370 }, { "grad_norm": 0.0495724156498909, "learning_rate": 2.3517904621728294e-05, "loss": 0.004, "step": 69380 }, { "grad_norm": 0.05174310877919197, "learning_rate": 2.3503880986081855e-05, "loss": 0.0042, "step": 69390 }, { "grad_norm": 0.08435472100973129, "learning_rate": 2.3489860248011136e-05, "loss": 0.0028, "step": 69400 }, { "grad_norm": 0.07674519717693329, "learning_rate": 2.3475842409049464e-05, "loss": 0.004, "step": 69410 }, { "grad_norm": 0.04652480408549309, "learning_rate": 2.346182747072975e-05, "loss": 0.0039, "step": 69420 }, { "grad_norm": 0.039908990263938904, "learning_rate": 2.3447815434584718e-05, "loss": 0.0024, "step": 69430 }, { "grad_norm": 0.06882314383983612, "learning_rate": 2.3433806302146656e-05, "loss": 0.0022, "step": 69440 }, { "grad_norm": 0.0545860230922699, "learning_rate": 2.3419800074947612e-05, "loss": 0.0054, "step": 69450 }, { "grad_norm": 0.0520210824906826, "learning_rate": 2.3405796754519248e-05, "loss": 0.0025, "step": 69460 }, { "grad_norm": 0.05361893028020859, "learning_rate": 2.3391796342392992e-05, "loss": 0.0039, "step": 69470 }, { "grad_norm": 0.07605461031198502, "learning_rate": 2.337779884009984e-05, "loss": 0.0032, "step": 69480 }, { "grad_norm": 0.04241131618618965, "learning_rate": 2.3363804249170624e-05, "loss": 0.0035, "step": 69490 }, { "grad_norm": 0.06264795362949371, "learning_rate": 2.3349812571135703e-05, "loss": 0.0043, "step": 69500 }, { "grad_norm": 0.03526058793067932, "learning_rate": 2.333582380752523e-05, "loss": 0.0033, "step": 69510 }, { "grad_norm": 0.039180509746074677, "learning_rate": 2.3321837959868963e-05, "loss": 0.0032, "step": 69520 }, { "grad_norm": 0.0621863454580307, "learning_rate": 2.3307855029696383e-05, "loss": 0.0029, "step": 69530 }, { "grad_norm": 0.07188748568296432, "learning_rate": 2.3293875018536648e-05, "loss": 0.0037, "step": 69540 }, { "grad_norm": 0.04457325488328934, "learning_rate": 2.327989792791861e-05, "loss": 0.0025, "step": 69550 }, { "grad_norm": 0.06442669779062271, "learning_rate": 2.326592375937074e-05, "loss": 0.005, "step": 69560 }, { "grad_norm": 0.06694293022155762, "learning_rate": 2.3251952514421248e-05, "loss": 0.0036, "step": 69570 }, { "grad_norm": 0.057328857481479645, "learning_rate": 2.323798419459804e-05, "loss": 0.0024, "step": 69580 }, { "grad_norm": 0.059997279196977615, "learning_rate": 2.3224018801428625e-05, "loss": 0.0026, "step": 69590 }, { "grad_norm": 0.034655846655368805, "learning_rate": 2.3210056336440273e-05, "loss": 0.0048, "step": 69600 }, { "grad_norm": 0.06691533327102661, "learning_rate": 2.3196096801159844e-05, "loss": 0.0041, "step": 69610 }, { "grad_norm": 0.1179208755493164, "learning_rate": 2.3182140197114012e-05, "loss": 0.0047, "step": 69620 }, { "grad_norm": 0.04586833715438843, "learning_rate": 2.3168186525828985e-05, "loss": 0.0025, "step": 69630 }, { "grad_norm": 0.06297102570533752, "learning_rate": 2.3154235788830757e-05, "loss": 0.0073, "step": 69640 }, { "grad_norm": 0.14573290944099426, "learning_rate": 2.3140287987644927e-05, "loss": 0.0032, "step": 69650 }, { "grad_norm": 0.09958687424659729, "learning_rate": 2.3126343123796833e-05, "loss": 0.0031, "step": 69660 }, { "grad_norm": 0.07689161598682404, "learning_rate": 2.3112401198811418e-05, "loss": 0.0031, "step": 69670 }, { "grad_norm": 0.0769355446100235, "learning_rate": 2.3098462214213418e-05, "loss": 0.0028, "step": 69680 }, { "grad_norm": 0.03547224774956703, "learning_rate": 2.308452617152712e-05, "loss": 0.0032, "step": 69690 }, { "grad_norm": 0.07905087620019913, "learning_rate": 2.3070593072276603e-05, "loss": 0.0033, "step": 69700 }, { "grad_norm": 0.03793778270483017, "learning_rate": 2.3056662917985518e-05, "loss": 0.0033, "step": 69710 }, { "grad_norm": 0.06641113013029099, "learning_rate": 2.3042735710177283e-05, "loss": 0.0046, "step": 69720 }, { "grad_norm": 0.07033105194568634, "learning_rate": 2.302881145037492e-05, "loss": 0.0041, "step": 69730 }, { "grad_norm": 0.05128297954797745, "learning_rate": 2.3014890140101192e-05, "loss": 0.0037, "step": 69740 }, { "grad_norm": 0.051192544400691986, "learning_rate": 2.30009717808785e-05, "loss": 0.0039, "step": 69750 }, { "grad_norm": 0.04507644474506378, "learning_rate": 2.2987056374228967e-05, "loss": 0.0029, "step": 69760 }, { "grad_norm": 0.042095862329006195, "learning_rate": 2.297314392167431e-05, "loss": 0.0047, "step": 69770 }, { "grad_norm": 0.027266765013337135, "learning_rate": 2.295923442473601e-05, "loss": 0.0035, "step": 69780 }, { "grad_norm": 0.06492974609136581, "learning_rate": 2.294532788493519e-05, "loss": 0.0027, "step": 69790 }, { "grad_norm": 0.05789269879460335, "learning_rate": 2.293142430379262e-05, "loss": 0.0036, "step": 69800 }, { "grad_norm": 0.06594566255807877, "learning_rate": 2.291752368282879e-05, "loss": 0.0032, "step": 69810 }, { "grad_norm": 0.06464514136314392, "learning_rate": 2.290362602356385e-05, "loss": 0.0029, "step": 69820 }, { "grad_norm": 0.06530607491731644, "learning_rate": 2.288973132751765e-05, "loss": 0.0045, "step": 69830 }, { "grad_norm": 0.0520385280251503, "learning_rate": 2.287583959620965e-05, "loss": 0.0042, "step": 69840 }, { "grad_norm": 0.08006312698125839, "learning_rate": 2.2861950831159074e-05, "loss": 0.0037, "step": 69850 }, { "grad_norm": 0.04558514803647995, "learning_rate": 2.284806503388472e-05, "loss": 0.0036, "step": 69860 }, { "grad_norm": 0.05365310236811638, "learning_rate": 2.283418220590518e-05, "loss": 0.0041, "step": 69870 }, { "grad_norm": 0.05475035309791565, "learning_rate": 2.282030234873862e-05, "loss": 0.0033, "step": 69880 }, { "grad_norm": 0.05337129533290863, "learning_rate": 2.2806425463902948e-05, "loss": 0.0056, "step": 69890 }, { "grad_norm": 0.04776670038700104, "learning_rate": 2.279255155291568e-05, "loss": 0.0032, "step": 69900 }, { "grad_norm": 0.05640561878681183, "learning_rate": 2.277868061729409e-05, "loss": 0.0038, "step": 69910 }, { "grad_norm": 0.05702785775065422, "learning_rate": 2.276481265855504e-05, "loss": 0.0028, "step": 69920 }, { "grad_norm": 0.07012659311294556, "learning_rate": 2.2750947678215128e-05, "loss": 0.0044, "step": 69930 }, { "grad_norm": 0.06141406297683716, "learning_rate": 2.2737085677790616e-05, "loss": 0.0031, "step": 69940 }, { "grad_norm": 0.06544587016105652, "learning_rate": 2.2723226658797443e-05, "loss": 0.0035, "step": 69950 }, { "grad_norm": 0.06632682681083679, "learning_rate": 2.270937062275117e-05, "loss": 0.0035, "step": 69960 }, { "grad_norm": 0.0373779833316803, "learning_rate": 2.2695517571167113e-05, "loss": 0.0028, "step": 69970 }, { "grad_norm": 0.06787820160388947, "learning_rate": 2.268166750556019e-05, "loss": 0.0034, "step": 69980 }, { "grad_norm": 0.0500805489718914, "learning_rate": 2.266782042744503e-05, "loss": 0.0044, "step": 69990 }, { "grad_norm": 0.07949486374855042, "learning_rate": 2.2653976338335935e-05, "loss": 0.0044, "step": 70000 }, { "grad_norm": 0.0482972078025341, "learning_rate": 2.2640135239746895e-05, "loss": 0.0022, "step": 70010 }, { "grad_norm": 0.048898447304964066, "learning_rate": 2.2626297133191515e-05, "loss": 0.0033, "step": 70020 }, { "grad_norm": 0.04932928457856178, "learning_rate": 2.261246202018312e-05, "loss": 0.0046, "step": 70030 }, { "grad_norm": 0.03860192745923996, "learning_rate": 2.2598629902234718e-05, "loss": 0.0036, "step": 70040 }, { "grad_norm": 0.05439833924174309, "learning_rate": 2.258480078085894e-05, "loss": 0.0041, "step": 70050 }, { "grad_norm": 0.07890335470438004, "learning_rate": 2.2570974657568126e-05, "loss": 0.0026, "step": 70060 }, { "grad_norm": 0.053378742188215256, "learning_rate": 2.2557151533874286e-05, "loss": 0.0035, "step": 70070 }, { "grad_norm": 0.046233948320150375, "learning_rate": 2.2543331411289116e-05, "loss": 0.0029, "step": 70080 }, { "grad_norm": 0.06089852750301361, "learning_rate": 2.252951429132392e-05, "loss": 0.0043, "step": 70090 }, { "grad_norm": 0.06720112264156342, "learning_rate": 2.2515700175489762e-05, "loss": 0.0038, "step": 70100 }, { "grad_norm": 0.06003754213452339, "learning_rate": 2.2501889065297288e-05, "loss": 0.0044, "step": 70110 }, { "grad_norm": 0.0781460627913475, "learning_rate": 2.2488080962256887e-05, "loss": 0.0043, "step": 70120 }, { "grad_norm": 0.05540552735328674, "learning_rate": 2.2474275867878585e-05, "loss": 0.0039, "step": 70130 }, { "grad_norm": 0.042445939034223557, "learning_rate": 2.246047378367211e-05, "loss": 0.0032, "step": 70140 }, { "grad_norm": 0.054926738142967224, "learning_rate": 2.2446674711146798e-05, "loss": 0.0028, "step": 70150 }, { "grad_norm": 0.07863136380910873, "learning_rate": 2.2432878651811734e-05, "loss": 0.0046, "step": 70160 }, { "grad_norm": 0.0649069994688034, "learning_rate": 2.2419085607175594e-05, "loss": 0.0023, "step": 70170 }, { "grad_norm": 0.07251179218292236, "learning_rate": 2.240529557874678e-05, "loss": 0.0042, "step": 70180 }, { "grad_norm": 0.05218115821480751, "learning_rate": 2.2391508568033358e-05, "loss": 0.005, "step": 70190 }, { "grad_norm": 0.08296170085668564, "learning_rate": 2.237772457654307e-05, "loss": 0.0036, "step": 70200 }, { "grad_norm": 0.04858231544494629, "learning_rate": 2.236394360578327e-05, "loss": 0.0056, "step": 70210 }, { "grad_norm": 0.05827389657497406, "learning_rate": 2.2350165657261073e-05, "loss": 0.004, "step": 70220 }, { "grad_norm": 0.09643720090389252, "learning_rate": 2.2336390732483164e-05, "loss": 0.0055, "step": 70230 }, { "grad_norm": 0.051168251782655716, "learning_rate": 2.2322618832955976e-05, "loss": 0.0042, "step": 70240 }, { "grad_norm": 0.07545819133520126, "learning_rate": 2.2308849960185578e-05, "loss": 0.0033, "step": 70250 }, { "grad_norm": 0.042352717369794846, "learning_rate": 2.229508411567774e-05, "loss": 0.0036, "step": 70260 }, { "grad_norm": 0.0427115336060524, "learning_rate": 2.2281321300937823e-05, "loss": 0.0024, "step": 70270 }, { "grad_norm": 0.05637168884277344, "learning_rate": 2.2267561517470937e-05, "loss": 0.0034, "step": 70280 }, { "grad_norm": 0.0734546035528183, "learning_rate": 2.2253804766781845e-05, "loss": 0.0026, "step": 70290 }, { "grad_norm": 0.050568364560604095, "learning_rate": 2.2240051050374933e-05, "loss": 0.0042, "step": 70300 }, { "grad_norm": 0.03856190666556358, "learning_rate": 2.22263003697543e-05, "loss": 0.0032, "step": 70310 }, { "grad_norm": 0.046437643468379974, "learning_rate": 2.2212552726423702e-05, "loss": 0.0034, "step": 70320 }, { "grad_norm": 0.07748883962631226, "learning_rate": 2.2198808121886578e-05, "loss": 0.004, "step": 70330 }, { "grad_norm": 0.06636824458837509, "learning_rate": 2.218506655764598e-05, "loss": 0.0035, "step": 70340 }, { "grad_norm": 0.09674481302499771, "learning_rate": 2.2171328035204704e-05, "loss": 0.0028, "step": 70350 }, { "grad_norm": 0.04063928872346878, "learning_rate": 2.2157592556065133e-05, "loss": 0.0029, "step": 70360 }, { "grad_norm": 0.04083158075809479, "learning_rate": 2.2143860121729382e-05, "loss": 0.0035, "step": 70370 }, { "grad_norm": 0.054759349673986435, "learning_rate": 2.2130130733699206e-05, "loss": 0.004, "step": 70380 }, { "grad_norm": 0.05445054918527603, "learning_rate": 2.2116404393476052e-05, "loss": 0.0035, "step": 70390 }, { "grad_norm": 0.070438452064991, "learning_rate": 2.2102681102560967e-05, "loss": 0.0037, "step": 70400 }, { "grad_norm": 0.04454707354307175, "learning_rate": 2.208896086245476e-05, "loss": 0.0035, "step": 70410 }, { "grad_norm": 0.05369239300489426, "learning_rate": 2.2075243674657808e-05, "loss": 0.0035, "step": 70420 }, { "grad_norm": 0.08158610016107559, "learning_rate": 2.206152954067024e-05, "loss": 0.0044, "step": 70430 }, { "grad_norm": 0.06805894523859024, "learning_rate": 2.2047818461991758e-05, "loss": 0.0027, "step": 70440 }, { "grad_norm": 0.047837994992733, "learning_rate": 2.2034110440121864e-05, "loss": 0.0051, "step": 70450 }, { "grad_norm": 0.06039274483919144, "learning_rate": 2.2020405476559585e-05, "loss": 0.0032, "step": 70460 }, { "grad_norm": 0.05292675271630287, "learning_rate": 2.2006703572803722e-05, "loss": 0.0041, "step": 70470 }, { "grad_norm": 0.06331058591604233, "learning_rate": 2.1993004730352646e-05, "loss": 0.0035, "step": 70480 }, { "grad_norm": 0.07956646382808685, "learning_rate": 2.1979308950704464e-05, "loss": 0.0035, "step": 70490 }, { "grad_norm": 0.0555691234767437, "learning_rate": 2.1965616235356923e-05, "loss": 0.0032, "step": 70500 }, { "grad_norm": 0.03394375741481781, "learning_rate": 2.1951926585807465e-05, "loss": 0.0031, "step": 70510 }, { "grad_norm": 0.052511557936668396, "learning_rate": 2.193824000355313e-05, "loss": 0.0033, "step": 70520 }, { "grad_norm": 0.0465628057718277, "learning_rate": 2.192455649009067e-05, "loss": 0.0056, "step": 70530 }, { "grad_norm": 0.07271728664636612, "learning_rate": 2.1910876046916523e-05, "loss": 0.0031, "step": 70540 }, { "grad_norm": 0.06752869486808777, "learning_rate": 2.1897198675526725e-05, "loss": 0.0029, "step": 70550 }, { "grad_norm": 0.051010631024837494, "learning_rate": 2.1883524377417024e-05, "loss": 0.0028, "step": 70560 }, { "grad_norm": 0.04373009130358696, "learning_rate": 2.1869853154082825e-05, "loss": 0.0028, "step": 70570 }, { "grad_norm": 0.04704230651259422, "learning_rate": 2.1856185007019215e-05, "loss": 0.0051, "step": 70580 }, { "grad_norm": 0.056808196008205414, "learning_rate": 2.1842519937720874e-05, "loss": 0.0053, "step": 70590 }, { "grad_norm": 0.042858295142650604, "learning_rate": 2.182885794768224e-05, "loss": 0.003, "step": 70600 }, { "grad_norm": 0.04245218262076378, "learning_rate": 2.1815199038397326e-05, "loss": 0.0037, "step": 70610 }, { "grad_norm": 0.07214953005313873, "learning_rate": 2.1801543211359893e-05, "loss": 0.0037, "step": 70620 }, { "grad_norm": 0.06300563365221024, "learning_rate": 2.178789046806326e-05, "loss": 0.004, "step": 70630 }, { "grad_norm": 0.06203790381550789, "learning_rate": 2.1774240810000547e-05, "loss": 0.0029, "step": 70640 }, { "grad_norm": 0.040606021881103516, "learning_rate": 2.17605942386644e-05, "loss": 0.0027, "step": 70650 }, { "grad_norm": 0.05365661531686783, "learning_rate": 2.174695075554723e-05, "loss": 0.0029, "step": 70660 }, { "grad_norm": 0.03849610313773155, "learning_rate": 2.173331036214103e-05, "loss": 0.0028, "step": 70670 }, { "grad_norm": 0.042288485914468765, "learning_rate": 2.171967305993753e-05, "loss": 0.0037, "step": 70680 }, { "grad_norm": 0.028278984129428864, "learning_rate": 2.1706038850428024e-05, "loss": 0.0032, "step": 70690 }, { "grad_norm": 0.0786486268043518, "learning_rate": 2.169240773510361e-05, "loss": 0.0041, "step": 70700 }, { "grad_norm": 0.06207556650042534, "learning_rate": 2.1678779715454906e-05, "loss": 0.0063, "step": 70710 }, { "grad_norm": 0.0641036108136177, "learning_rate": 2.166515479297229e-05, "loss": 0.006, "step": 70720 }, { "grad_norm": 0.06908608973026276, "learning_rate": 2.1651532969145732e-05, "loss": 0.0032, "step": 70730 }, { "grad_norm": 0.06072888895869255, "learning_rate": 2.1637914245464903e-05, "loss": 0.0033, "step": 70740 }, { "grad_norm": 0.06055086478590965, "learning_rate": 2.1624298623419133e-05, "loss": 0.0041, "step": 70750 }, { "grad_norm": 0.0429280661046505, "learning_rate": 2.1610686104497413e-05, "loss": 0.0035, "step": 70760 }, { "grad_norm": 0.0590876080095768, "learning_rate": 2.1597076690188366e-05, "loss": 0.0046, "step": 70770 }, { "grad_norm": 0.05185779184103012, "learning_rate": 2.1583470381980315e-05, "loss": 0.0046, "step": 70780 }, { "grad_norm": 0.06258239597082138, "learning_rate": 2.156986718136123e-05, "loss": 0.0029, "step": 70790 }, { "grad_norm": 0.06396149098873138, "learning_rate": 2.155626708981871e-05, "loss": 0.0038, "step": 70800 }, { "grad_norm": 0.051305681467056274, "learning_rate": 2.1542670108840075e-05, "loss": 0.0028, "step": 70810 }, { "grad_norm": 0.06695818156003952, "learning_rate": 2.152907623991221e-05, "loss": 0.0045, "step": 70820 }, { "grad_norm": 0.06968887895345688, "learning_rate": 2.1515485484521802e-05, "loss": 0.0046, "step": 70830 }, { "grad_norm": 0.0433061346411705, "learning_rate": 2.150189784415506e-05, "loss": 0.0042, "step": 70840 }, { "grad_norm": 0.07143576443195343, "learning_rate": 2.148831332029794e-05, "loss": 0.0035, "step": 70850 }, { "grad_norm": 0.04080790653824806, "learning_rate": 2.147473191443599e-05, "loss": 0.0026, "step": 70860 }, { "grad_norm": 0.051046937704086304, "learning_rate": 2.1461153628054492e-05, "loss": 0.0023, "step": 70870 }, { "grad_norm": 0.051045726984739304, "learning_rate": 2.1447578462638284e-05, "loss": 0.0037, "step": 70880 }, { "grad_norm": 0.04335199296474457, "learning_rate": 2.143400641967201e-05, "loss": 0.0031, "step": 70890 }, { "grad_norm": 0.06591477245092392, "learning_rate": 2.1420437500639823e-05, "loss": 0.0043, "step": 70900 }, { "grad_norm": 0.06758064031600952, "learning_rate": 2.1406871707025646e-05, "loss": 0.003, "step": 70910 }, { "grad_norm": 0.0755959004163742, "learning_rate": 2.1393309040312963e-05, "loss": 0.0024, "step": 70920 }, { "grad_norm": 0.07287575304508209, "learning_rate": 2.137974950198502e-05, "loss": 0.0046, "step": 70930 }, { "grad_norm": 0.08338741213083267, "learning_rate": 2.1366193093524617e-05, "loss": 0.0047, "step": 70940 }, { "grad_norm": 0.06545405089855194, "learning_rate": 2.1352639816414288e-05, "loss": 0.0029, "step": 70950 }, { "grad_norm": 0.045727796852588654, "learning_rate": 2.1339089672136202e-05, "loss": 0.0026, "step": 70960 }, { "grad_norm": 0.07722962647676468, "learning_rate": 2.1325542662172193e-05, "loss": 0.0033, "step": 70970 }, { "grad_norm": 0.07403630763292313, "learning_rate": 2.1311998788003712e-05, "loss": 0.0044, "step": 70980 }, { "grad_norm": 0.08895818889141083, "learning_rate": 2.1298458051111907e-05, "loss": 0.0042, "step": 70990 }, { "grad_norm": 0.049020782113075256, "learning_rate": 2.1284920452977602e-05, "loss": 0.0038, "step": 71000 }, { "grad_norm": 0.06515365839004517, "learning_rate": 2.12713859950812e-05, "loss": 0.0034, "step": 71010 }, { "grad_norm": 0.07558786123991013, "learning_rate": 2.1257854678902832e-05, "loss": 0.0042, "step": 71020 }, { "grad_norm": 0.06340939551591873, "learning_rate": 2.124432650592227e-05, "loss": 0.0037, "step": 71030 }, { "grad_norm": 0.05211186036467552, "learning_rate": 2.1230801477618937e-05, "loss": 0.0034, "step": 71040 }, { "grad_norm": 0.05234374850988388, "learning_rate": 2.121727959547189e-05, "loss": 0.0035, "step": 71050 }, { "grad_norm": 0.04459633305668831, "learning_rate": 2.1203760860959886e-05, "loss": 0.0031, "step": 71060 }, { "grad_norm": 0.03902577981352806, "learning_rate": 2.119024527556127e-05, "loss": 0.0041, "step": 71070 }, { "grad_norm": 0.04976290464401245, "learning_rate": 2.117673284075415e-05, "loss": 0.0052, "step": 71080 }, { "grad_norm": 0.08120544999837875, "learning_rate": 2.1163223558016165e-05, "loss": 0.0084, "step": 71090 }, { "grad_norm": 0.06297092139720917, "learning_rate": 2.1149717428824722e-05, "loss": 0.0034, "step": 71100 }, { "grad_norm": 0.07420752197504044, "learning_rate": 2.1136214454656784e-05, "loss": 0.0042, "step": 71110 }, { "grad_norm": 0.0461604967713356, "learning_rate": 2.1122714636989056e-05, "loss": 0.004, "step": 71120 }, { "grad_norm": 0.051105573773384094, "learning_rate": 2.1109217977297823e-05, "loss": 0.0034, "step": 71130 }, { "grad_norm": 0.04579402506351471, "learning_rate": 2.1095724477059077e-05, "loss": 0.0039, "step": 71140 }, { "grad_norm": 0.05479393154382706, "learning_rate": 2.1082234137748434e-05, "loss": 0.0038, "step": 71150 }, { "grad_norm": 0.04731207340955734, "learning_rate": 2.106874696084122e-05, "loss": 0.0037, "step": 71160 }, { "grad_norm": 0.04306790605187416, "learning_rate": 2.1055262947812315e-05, "loss": 0.0025, "step": 71170 }, { "grad_norm": 0.04471487179398537, "learning_rate": 2.104178210013636e-05, "loss": 0.0025, "step": 71180 }, { "grad_norm": 0.03243514522910118, "learning_rate": 2.1028304419287554e-05, "loss": 0.0034, "step": 71190 }, { "grad_norm": 0.062770314514637, "learning_rate": 2.101482990673982e-05, "loss": 0.0038, "step": 71200 }, { "grad_norm": 0.04708278924226761, "learning_rate": 2.1001358563966707e-05, "loss": 0.0032, "step": 71210 }, { "grad_norm": 0.07481221854686737, "learning_rate": 2.098789039244144e-05, "loss": 0.003, "step": 71220 }, { "grad_norm": 0.05307013541460037, "learning_rate": 2.097442539363685e-05, "loss": 0.0034, "step": 71230 }, { "grad_norm": 0.04031238704919815, "learning_rate": 2.0960963569025448e-05, "loss": 0.003, "step": 71240 }, { "grad_norm": 0.046616360545158386, "learning_rate": 2.0947504920079435e-05, "loss": 0.0039, "step": 71250 }, { "grad_norm": 0.05080205574631691, "learning_rate": 2.093404944827059e-05, "loss": 0.0053, "step": 71260 }, { "grad_norm": 0.05617836117744446, "learning_rate": 2.0920597155070394e-05, "loss": 0.0027, "step": 71270 }, { "grad_norm": 0.03524743765592575, "learning_rate": 2.090714804194997e-05, "loss": 0.0031, "step": 71280 }, { "grad_norm": 0.04795578867197037, "learning_rate": 2.0893702110380114e-05, "loss": 0.0039, "step": 71290 }, { "grad_norm": 0.0533800944685936, "learning_rate": 2.0880259361831216e-05, "loss": 0.0038, "step": 71300 }, { "grad_norm": 0.030457250773906708, "learning_rate": 2.0866819797773397e-05, "loss": 0.0027, "step": 71310 }, { "grad_norm": 0.04051443561911583, "learning_rate": 2.085338341967634e-05, "loss": 0.0029, "step": 71320 }, { "grad_norm": 0.052841026335954666, "learning_rate": 2.083995022900946e-05, "loss": 0.0045, "step": 71330 }, { "grad_norm": 0.05679060146212578, "learning_rate": 2.0826520227241768e-05, "loss": 0.0027, "step": 71340 }, { "grad_norm": 0.0852782279253006, "learning_rate": 2.081309341584199e-05, "loss": 0.0044, "step": 71350 }, { "grad_norm": 0.06041337549686432, "learning_rate": 2.0799669796278413e-05, "loss": 0.0048, "step": 71360 }, { "grad_norm": 0.04599592834711075, "learning_rate": 2.078624937001906e-05, "loss": 0.0026, "step": 71370 }, { "grad_norm": 0.06010683625936508, "learning_rate": 2.0772832138531544e-05, "loss": 0.0042, "step": 71380 }, { "grad_norm": 0.04955654591321945, "learning_rate": 2.0759418103283158e-05, "loss": 0.0028, "step": 71390 }, { "grad_norm": 0.06567558646202087, "learning_rate": 2.0746007265740846e-05, "loss": 0.0035, "step": 71400 }, { "grad_norm": 0.06095850095152855, "learning_rate": 2.0732599627371215e-05, "loss": 0.0038, "step": 71410 }, { "grad_norm": 0.03988879546523094, "learning_rate": 2.0719195189640468e-05, "loss": 0.003, "step": 71420 }, { "grad_norm": 0.06064867973327637, "learning_rate": 2.0705793954014528e-05, "loss": 0.0039, "step": 71430 }, { "grad_norm": 0.0624065101146698, "learning_rate": 2.06923959219589e-05, "loss": 0.0032, "step": 71440 }, { "grad_norm": 0.05129474028944969, "learning_rate": 2.0679001094938793e-05, "loss": 0.0038, "step": 71450 }, { "grad_norm": 0.053785163909196854, "learning_rate": 2.066560947441904e-05, "loss": 0.0032, "step": 71460 }, { "grad_norm": 0.09647063910961151, "learning_rate": 2.0652221061864152e-05, "loss": 0.0036, "step": 71470 }, { "grad_norm": 0.07071565836668015, "learning_rate": 2.0638835858738226e-05, "loss": 0.0044, "step": 71480 }, { "grad_norm": 0.06441047042608261, "learning_rate": 2.0625453866505068e-05, "loss": 0.004, "step": 71490 }, { "grad_norm": 0.04524748772382736, "learning_rate": 2.0612075086628123e-05, "loss": 0.0032, "step": 71500 }, { "grad_norm": 0.06496914476156235, "learning_rate": 2.059869952057045e-05, "loss": 0.0057, "step": 71510 }, { "grad_norm": 0.07803594321012497, "learning_rate": 2.0585327169794793e-05, "loss": 0.0036, "step": 71520 }, { "grad_norm": 0.07234246283769608, "learning_rate": 2.0571958035763534e-05, "loss": 0.0029, "step": 71530 }, { "grad_norm": 0.059536490589380264, "learning_rate": 2.0558592119938715e-05, "loss": 0.005, "step": 71540 }, { "grad_norm": 0.08520426601171494, "learning_rate": 2.054522942378198e-05, "loss": 0.0031, "step": 71550 }, { "grad_norm": 0.08295934647321701, "learning_rate": 2.053186994875469e-05, "loss": 0.0026, "step": 71560 }, { "grad_norm": 0.058143652975559235, "learning_rate": 2.0518513696317775e-05, "loss": 0.0069, "step": 71570 }, { "grad_norm": 0.04611005261540413, "learning_rate": 2.050516066793188e-05, "loss": 0.0034, "step": 71580 }, { "grad_norm": 0.07462847232818604, "learning_rate": 2.0491810865057276e-05, "loss": 0.004, "step": 71590 }, { "grad_norm": 0.06044371426105499, "learning_rate": 2.0478464289153882e-05, "loss": 0.0031, "step": 71600 }, { "grad_norm": 0.0679219514131546, "learning_rate": 2.0465120941681232e-05, "loss": 0.0037, "step": 71610 }, { "grad_norm": 0.08978874236345291, "learning_rate": 2.0451780824098566e-05, "loss": 0.0045, "step": 71620 }, { "grad_norm": 0.05303678661584854, "learning_rate": 2.043844393786471e-05, "loss": 0.0032, "step": 71630 }, { "grad_norm": 0.07186707109212875, "learning_rate": 2.0425110284438193e-05, "loss": 0.0034, "step": 71640 }, { "grad_norm": 0.05240573361515999, "learning_rate": 2.041177986527712e-05, "loss": 0.0035, "step": 71650 }, { "grad_norm": 0.04461078718304634, "learning_rate": 2.0398452681839347e-05, "loss": 0.0045, "step": 71660 }, { "grad_norm": 0.046752601861953735, "learning_rate": 2.0385128735582272e-05, "loss": 0.0023, "step": 71670 }, { "grad_norm": 0.053897470235824585, "learning_rate": 2.0371808027963003e-05, "loss": 0.0036, "step": 71680 }, { "grad_norm": 0.06891529262065887, "learning_rate": 2.035849056043825e-05, "loss": 0.0043, "step": 71690 }, { "grad_norm": 0.06343714147806168, "learning_rate": 2.03451763344644e-05, "loss": 0.004, "step": 71700 }, { "grad_norm": 0.03756963834166527, "learning_rate": 2.0331865351497482e-05, "loss": 0.0022, "step": 71710 }, { "grad_norm": 0.04960313066840172, "learning_rate": 2.0318557612993185e-05, "loss": 0.0032, "step": 71720 }, { "grad_norm": 0.08736897259950638, "learning_rate": 2.0305253120406782e-05, "loss": 0.003, "step": 71730 }, { "grad_norm": 0.08242157101631165, "learning_rate": 2.029195187519325e-05, "loss": 0.0041, "step": 71740 }, { "grad_norm": 0.06618504226207733, "learning_rate": 2.0278653878807217e-05, "loss": 0.0034, "step": 71750 }, { "grad_norm": 0.055031705647706985, "learning_rate": 2.0265359132702892e-05, "loss": 0.0028, "step": 71760 }, { "grad_norm": 0.08158232271671295, "learning_rate": 2.0252067638334182e-05, "loss": 0.0038, "step": 71770 }, { "grad_norm": 0.03663557022809982, "learning_rate": 2.023877939715464e-05, "loss": 0.0025, "step": 71780 }, { "grad_norm": 0.06423982232809067, "learning_rate": 2.0225494410617448e-05, "loss": 0.0038, "step": 71790 }, { "grad_norm": 0.04739992693066597, "learning_rate": 2.02122126801754e-05, "loss": 0.0039, "step": 71800 }, { "grad_norm": 0.048370834439992905, "learning_rate": 2.0198934207281018e-05, "loss": 0.0029, "step": 71810 }, { "grad_norm": 0.061363253742456436, "learning_rate": 2.0185658993386357e-05, "loss": 0.0034, "step": 71820 }, { "grad_norm": 0.04304715618491173, "learning_rate": 2.0172387039943226e-05, "loss": 0.004, "step": 71830 }, { "grad_norm": 0.06391818821430206, "learning_rate": 2.0159118348402968e-05, "loss": 0.0037, "step": 71840 }, { "grad_norm": 0.08585850149393082, "learning_rate": 2.0145852920216697e-05, "loss": 0.0038, "step": 71850 }, { "grad_norm": 0.0368262343108654, "learning_rate": 2.0132590756835045e-05, "loss": 0.0037, "step": 71860 }, { "grad_norm": 0.059598375111818314, "learning_rate": 2.0119331859708386e-05, "loss": 0.0047, "step": 71870 }, { "grad_norm": 0.08056779950857162, "learning_rate": 2.0106076230286654e-05, "loss": 0.0045, "step": 71880 }, { "grad_norm": 0.07678236067295074, "learning_rate": 2.0092823870019488e-05, "loss": 0.0041, "step": 71890 }, { "grad_norm": 0.07493352144956589, "learning_rate": 2.0079574780356115e-05, "loss": 0.0033, "step": 71900 }, { "grad_norm": 0.04403087869286537, "learning_rate": 2.0066328962745486e-05, "loss": 0.0032, "step": 71910 }, { "grad_norm": 0.07228290289640427, "learning_rate": 2.00530864186361e-05, "loss": 0.0043, "step": 71920 }, { "grad_norm": 0.07088839262723923, "learning_rate": 2.003984714947618e-05, "loss": 0.003, "step": 71930 }, { "grad_norm": 0.05972662940621376, "learning_rate": 2.0026611156713515e-05, "loss": 0.0024, "step": 71940 }, { "grad_norm": 0.08991512656211853, "learning_rate": 2.0013378441795584e-05, "loss": 0.0036, "step": 71950 }, { "grad_norm": 0.07797543704509735, "learning_rate": 2.0000149006169506e-05, "loss": 0.0072, "step": 71960 }, { "grad_norm": 0.07572651654481888, "learning_rate": 1.9986922851282048e-05, "loss": 0.0039, "step": 71970 }, { "grad_norm": 0.054717183113098145, "learning_rate": 1.997369997857956e-05, "loss": 0.0041, "step": 71980 }, { "grad_norm": 0.04819191247224808, "learning_rate": 1.9960480389508103e-05, "loss": 0.0032, "step": 71990 }, { "grad_norm": 0.061677251011133194, "learning_rate": 1.9947264085513363e-05, "loss": 0.0029, "step": 72000 }, { "grad_norm": 0.04775407165288925, "learning_rate": 1.993405106804062e-05, "loss": 0.0033, "step": 72010 }, { "grad_norm": 0.06903567910194397, "learning_rate": 1.9920841338534867e-05, "loss": 0.0053, "step": 72020 }, { "grad_norm": 0.0459347702562809, "learning_rate": 1.9907634898440646e-05, "loss": 0.0026, "step": 72030 }, { "grad_norm": 0.047032617032527924, "learning_rate": 1.989443174920227e-05, "loss": 0.0051, "step": 72040 }, { "grad_norm": 0.06888120621442795, "learning_rate": 1.988123189226355e-05, "loss": 0.0038, "step": 72050 }, { "grad_norm": 0.05577676743268967, "learning_rate": 1.986803532906805e-05, "loss": 0.0028, "step": 72060 }, { "grad_norm": 0.03848963603377342, "learning_rate": 1.985484206105889e-05, "loss": 0.0031, "step": 72070 }, { "grad_norm": 0.07563251256942749, "learning_rate": 1.9841652089678887e-05, "loss": 0.0034, "step": 72080 }, { "grad_norm": 0.05077947676181793, "learning_rate": 1.9828465416370435e-05, "loss": 0.0033, "step": 72090 }, { "grad_norm": 0.033591125160455704, "learning_rate": 1.9815282042575685e-05, "loss": 0.0029, "step": 72100 }, { "grad_norm": 0.059713877737522125, "learning_rate": 1.9802101969736285e-05, "loss": 0.006, "step": 72110 }, { "grad_norm": 0.0490376353263855, "learning_rate": 1.9788925199293633e-05, "loss": 0.0028, "step": 72120 }, { "grad_norm": 0.06066516414284706, "learning_rate": 1.977575173268868e-05, "loss": 0.0033, "step": 72130 }, { "grad_norm": 0.10695818811655045, "learning_rate": 1.9762581571362094e-05, "loss": 0.0048, "step": 72140 }, { "grad_norm": 0.06258919090032578, "learning_rate": 1.9749414716754112e-05, "loss": 0.0035, "step": 72150 }, { "grad_norm": 0.03781343623995781, "learning_rate": 1.973625117030465e-05, "loss": 0.0034, "step": 72160 }, { "grad_norm": 0.05025364086031914, "learning_rate": 1.972309093345326e-05, "loss": 0.0045, "step": 72170 }, { "grad_norm": 0.03216182067990303, "learning_rate": 1.970993400763915e-05, "loss": 0.0031, "step": 72180 }, { "grad_norm": 0.0371704138815403, "learning_rate": 1.969678039430109e-05, "loss": 0.0047, "step": 72190 }, { "grad_norm": 0.03784919157624245, "learning_rate": 1.9683630094877564e-05, "loss": 0.0029, "step": 72200 }, { "grad_norm": 0.06265800446271896, "learning_rate": 1.9670483110806696e-05, "loss": 0.003, "step": 72210 }, { "grad_norm": 0.05152462050318718, "learning_rate": 1.9657339443526172e-05, "loss": 0.0032, "step": 72220 }, { "grad_norm": 0.041101180016994476, "learning_rate": 1.964419909447339e-05, "loss": 0.0036, "step": 72230 }, { "grad_norm": 0.05273044854402542, "learning_rate": 1.963106206508536e-05, "loss": 0.0031, "step": 72240 }, { "grad_norm": 0.056829556822776794, "learning_rate": 1.9617928356798738e-05, "loss": 0.0028, "step": 72250 }, { "grad_norm": 0.045051753520965576, "learning_rate": 1.960479797104977e-05, "loss": 0.0047, "step": 72260 }, { "grad_norm": 0.057316653430461884, "learning_rate": 1.9591670909274418e-05, "loss": 0.0029, "step": 72270 }, { "grad_norm": 0.058077432215213776, "learning_rate": 1.9578547172908184e-05, "loss": 0.0028, "step": 72280 }, { "grad_norm": 0.049718741327524185, "learning_rate": 1.9565426763386325e-05, "loss": 0.0034, "step": 72290 }, { "grad_norm": 0.048437222838401794, "learning_rate": 1.9552309682143622e-05, "loss": 0.0022, "step": 72300 }, { "grad_norm": 0.033523429185152054, "learning_rate": 1.9539195930614574e-05, "loss": 0.0029, "step": 72310 }, { "grad_norm": 0.05362777039408684, "learning_rate": 1.952608551023324e-05, "loss": 0.0052, "step": 72320 }, { "grad_norm": 0.04312025010585785, "learning_rate": 1.9512978422433397e-05, "loss": 0.0047, "step": 72330 }, { "grad_norm": 0.046899132430553436, "learning_rate": 1.9499874668648376e-05, "loss": 0.0034, "step": 72340 }, { "grad_norm": 0.051853377372026443, "learning_rate": 1.9486774250311213e-05, "loss": 0.0025, "step": 72350 }, { "grad_norm": 0.04835980385541916, "learning_rate": 1.9473677168854537e-05, "loss": 0.0029, "step": 72360 }, { "grad_norm": 0.06233367696404457, "learning_rate": 1.946058342571065e-05, "loss": 0.0039, "step": 72370 }, { "grad_norm": 0.03874778375029564, "learning_rate": 1.9447493022311424e-05, "loss": 0.0038, "step": 72380 }, { "grad_norm": 0.04914108291268349, "learning_rate": 1.9434405960088442e-05, "loss": 0.0023, "step": 72390 }, { "grad_norm": 0.055216945707798004, "learning_rate": 1.9421322240472857e-05, "loss": 0.0051, "step": 72400 }, { "grad_norm": 0.05706264078617096, "learning_rate": 1.9408241864895494e-05, "loss": 0.0022, "step": 72410 }, { "grad_norm": 0.052528589963912964, "learning_rate": 1.939516483478681e-05, "loss": 0.0032, "step": 72420 }, { "grad_norm": 0.03788239508867264, "learning_rate": 1.9382091151576902e-05, "loss": 0.005, "step": 72430 }, { "grad_norm": 0.04077909514307976, "learning_rate": 1.9369020816695454e-05, "loss": 0.0032, "step": 72440 }, { "grad_norm": 0.062269169837236404, "learning_rate": 1.935595383157184e-05, "loss": 0.0032, "step": 72450 }, { "grad_norm": 0.06956877559423447, "learning_rate": 1.9342890197635057e-05, "loss": 0.0043, "step": 72460 }, { "grad_norm": 0.06405127048492432, "learning_rate": 1.9329829916313685e-05, "loss": 0.0065, "step": 72470 }, { "grad_norm": 0.06584437191486359, "learning_rate": 1.9316772989036013e-05, "loss": 0.0028, "step": 72480 }, { "grad_norm": 0.04559742659330368, "learning_rate": 1.930371941722991e-05, "loss": 0.003, "step": 72490 }, { "grad_norm": 0.05626058578491211, "learning_rate": 1.929066920232292e-05, "loss": 0.0039, "step": 72500 }, { "grad_norm": 0.04257083311676979, "learning_rate": 1.9277622345742152e-05, "loss": 0.002, "step": 72510 }, { "grad_norm": 0.042950306087732315, "learning_rate": 1.926457884891444e-05, "loss": 0.0034, "step": 72520 }, { "grad_norm": 0.04132293909788132, "learning_rate": 1.9251538713266147e-05, "loss": 0.0038, "step": 72530 }, { "grad_norm": 0.06029539927840233, "learning_rate": 1.9238501940223358e-05, "loss": 0.0036, "step": 72540 }, { "grad_norm": 0.05886450409889221, "learning_rate": 1.922546853121175e-05, "loss": 0.0036, "step": 72550 }, { "grad_norm": 0.07429676502943039, "learning_rate": 1.9212438487656647e-05, "loss": 0.0039, "step": 72560 }, { "grad_norm": 0.029844172298908234, "learning_rate": 1.9199411810982965e-05, "loss": 0.0026, "step": 72570 }, { "grad_norm": 0.09023772925138474, "learning_rate": 1.918638850261532e-05, "loss": 0.004, "step": 72580 }, { "grad_norm": 0.06419488787651062, "learning_rate": 1.9173368563977885e-05, "loss": 0.0027, "step": 72590 }, { "grad_norm": 0.05962991714477539, "learning_rate": 1.9160351996494513e-05, "loss": 0.0041, "step": 72600 }, { "grad_norm": 0.06278033554553986, "learning_rate": 1.9147338801588677e-05, "loss": 0.0041, "step": 72610 }, { "grad_norm": 0.07783512026071548, "learning_rate": 1.913432898068351e-05, "loss": 0.0032, "step": 72620 }, { "grad_norm": 0.04217972606420517, "learning_rate": 1.912132253520169e-05, "loss": 0.0055, "step": 72630 }, { "grad_norm": 0.0341201052069664, "learning_rate": 1.910831946656564e-05, "loss": 0.0035, "step": 72640 }, { "grad_norm": 0.06087488308548927, "learning_rate": 1.909531977619731e-05, "loss": 0.0032, "step": 72650 }, { "grad_norm": 0.06545517593622208, "learning_rate": 1.908232346551834e-05, "loss": 0.0028, "step": 72660 }, { "grad_norm": 0.052589815109968185, "learning_rate": 1.9069330535949993e-05, "loss": 0.0022, "step": 72670 }, { "grad_norm": 0.04455430805683136, "learning_rate": 1.9056340988913167e-05, "loss": 0.0042, "step": 72680 }, { "grad_norm": 0.0506042055785656, "learning_rate": 1.9043354825828357e-05, "loss": 0.0034, "step": 72690 }, { "grad_norm": 0.030301030725240707, "learning_rate": 1.9030372048115724e-05, "loss": 0.0024, "step": 72700 }, { "grad_norm": 0.03931983932852745, "learning_rate": 1.9017392657195056e-05, "loss": 0.0024, "step": 72710 }, { "grad_norm": 0.04831136390566826, "learning_rate": 1.9004416654485724e-05, "loss": 0.003, "step": 72720 }, { "grad_norm": 0.07592733949422836, "learning_rate": 1.8991444041406785e-05, "loss": 0.0028, "step": 72730 }, { "grad_norm": 0.03719174489378929, "learning_rate": 1.8978474819376912e-05, "loss": 0.0027, "step": 72740 }, { "grad_norm": 0.08233114331960678, "learning_rate": 1.8965508989814406e-05, "loss": 0.0041, "step": 72750 }, { "grad_norm": 0.051133252680301666, "learning_rate": 1.8952546554137158e-05, "loss": 0.0026, "step": 72760 }, { "grad_norm": 0.061994701623916626, "learning_rate": 1.8939587513762757e-05, "loss": 0.0027, "step": 72770 }, { "grad_norm": 0.049497563391923904, "learning_rate": 1.892663187010835e-05, "loss": 0.0031, "step": 72780 }, { "grad_norm": 0.0952250063419342, "learning_rate": 1.891367962459077e-05, "loss": 0.004, "step": 72790 }, { "grad_norm": 0.05744398385286331, "learning_rate": 1.890073077862644e-05, "loss": 0.0038, "step": 72800 }, { "grad_norm": 0.04631669074296951, "learning_rate": 1.888778533363146e-05, "loss": 0.0027, "step": 72810 }, { "grad_norm": 0.06917741149663925, "learning_rate": 1.8874843291021478e-05, "loss": 0.0031, "step": 72820 }, { "grad_norm": 0.030810866504907608, "learning_rate": 1.8861904652211853e-05, "loss": 0.0046, "step": 72830 }, { "grad_norm": 0.04954130947589874, "learning_rate": 1.8848969418617507e-05, "loss": 0.0026, "step": 72840 }, { "grad_norm": 0.03998202830553055, "learning_rate": 1.8836037591653044e-05, "loss": 0.0049, "step": 72850 }, { "grad_norm": 0.038101062178611755, "learning_rate": 1.882310917273262e-05, "loss": 0.0036, "step": 72860 }, { "grad_norm": 0.04498506709933281, "learning_rate": 1.881018416327014e-05, "loss": 0.0042, "step": 72870 }, { "grad_norm": 0.043414823710918427, "learning_rate": 1.8797262564679014e-05, "loss": 0.0036, "step": 72880 }, { "grad_norm": 0.03497651591897011, "learning_rate": 1.8784344378372355e-05, "loss": 0.0024, "step": 72890 }, { "grad_norm": 0.056747645139694214, "learning_rate": 1.8771429605762843e-05, "loss": 0.0034, "step": 72900 }, { "grad_norm": 0.043333668261766434, "learning_rate": 1.8758518248262835e-05, "loss": 0.0048, "step": 72910 }, { "grad_norm": 0.05694954842329025, "learning_rate": 1.8745610307284305e-05, "loss": 0.0028, "step": 72920 }, { "grad_norm": 0.040892552584409714, "learning_rate": 1.8732705784238862e-05, "loss": 0.0026, "step": 72930 }, { "grad_norm": 0.06272284686565399, "learning_rate": 1.8719804680537682e-05, "loss": 0.0032, "step": 72940 }, { "grad_norm": 0.07469848543405533, "learning_rate": 1.870690699759164e-05, "loss": 0.004, "step": 72950 }, { "grad_norm": 0.03686196357011795, "learning_rate": 1.8694012736811213e-05, "loss": 0.002, "step": 72960 }, { "grad_norm": 0.045215409249067307, "learning_rate": 1.8681121899606463e-05, "loss": 0.0052, "step": 72970 }, { "grad_norm": 0.06503912806510925, "learning_rate": 1.8668234487387144e-05, "loss": 0.0041, "step": 72980 }, { "grad_norm": 0.03422660008072853, "learning_rate": 1.8655350501562592e-05, "loss": 0.0065, "step": 72990 }, { "grad_norm": 0.050945889204740524, "learning_rate": 1.8642469943541796e-05, "loss": 0.0029, "step": 73000 }, { "grad_norm": 0.06691086292266846, "learning_rate": 1.8629592814733328e-05, "loss": 0.0038, "step": 73010 }, { "grad_norm": 0.05922841280698776, "learning_rate": 1.8616719116545438e-05, "loss": 0.0051, "step": 73020 }, { "grad_norm": 0.03050910122692585, "learning_rate": 1.8603848850385952e-05, "loss": 0.0043, "step": 73030 }, { "grad_norm": 0.03924354910850525, "learning_rate": 1.859098201766236e-05, "loss": 0.0036, "step": 73040 }, { "grad_norm": 0.050468046218156815, "learning_rate": 1.8578118619781726e-05, "loss": 0.0029, "step": 73050 }, { "grad_norm": 0.04517103731632233, "learning_rate": 1.856525865815083e-05, "loss": 0.002, "step": 73060 }, { "grad_norm": 0.050936345010995865, "learning_rate": 1.8552402134175966e-05, "loss": 0.0032, "step": 73070 }, { "grad_norm": 0.04442968592047691, "learning_rate": 1.8539549049263143e-05, "loss": 0.0036, "step": 73080 }, { "grad_norm": 0.04374353215098381, "learning_rate": 1.8526699404817917e-05, "loss": 0.0026, "step": 73090 }, { "grad_norm": 0.0737387165427208, "learning_rate": 1.8513853202245546e-05, "loss": 0.0026, "step": 73100 }, { "grad_norm": 0.05033831670880318, "learning_rate": 1.850101044295081e-05, "loss": 0.0033, "step": 73110 }, { "grad_norm": 0.08387717604637146, "learning_rate": 1.848817112833825e-05, "loss": 0.0034, "step": 73120 }, { "grad_norm": 0.05743671581149101, "learning_rate": 1.8475335259811905e-05, "loss": 0.0026, "step": 73130 }, { "grad_norm": 0.06332802772521973, "learning_rate": 1.8462502838775508e-05, "loss": 0.0026, "step": 73140 }, { "grad_norm": 0.058703403919935226, "learning_rate": 1.8449673866632378e-05, "loss": 0.0036, "step": 73150 }, { "grad_norm": 0.07237105816602707, "learning_rate": 1.8436848344785473e-05, "loss": 0.0036, "step": 73160 }, { "grad_norm": 0.05295068398118019, "learning_rate": 1.8424026274637378e-05, "loss": 0.0036, "step": 73170 }, { "grad_norm": 0.055747825652360916, "learning_rate": 1.8411207657590312e-05, "loss": 0.0058, "step": 73180 }, { "grad_norm": 0.06725611537694931, "learning_rate": 1.839839249504607e-05, "loss": 0.006, "step": 73190 }, { "grad_norm": 0.048084408044815063, "learning_rate": 1.8385580788406108e-05, "loss": 0.0024, "step": 73200 }, { "grad_norm": 0.05205981805920601, "learning_rate": 1.837277253907152e-05, "loss": 0.0034, "step": 73210 }, { "grad_norm": 0.09178459644317627, "learning_rate": 1.8359967748442957e-05, "loss": 0.0024, "step": 73220 }, { "grad_norm": 0.036026615649461746, "learning_rate": 1.8347166417920774e-05, "loss": 0.0054, "step": 73230 }, { "grad_norm": 0.08321591466665268, "learning_rate": 1.833436854890484e-05, "loss": 0.0035, "step": 73240 }, { "grad_norm": 0.06889913231134415, "learning_rate": 1.8321574142794796e-05, "loss": 0.0034, "step": 73250 }, { "grad_norm": 0.0405447892844677, "learning_rate": 1.8308783200989748e-05, "loss": 0.004, "step": 73260 }, { "grad_norm": 0.06230342388153076, "learning_rate": 1.829599572488855e-05, "loss": 0.0051, "step": 73270 }, { "grad_norm": 0.029903775081038475, "learning_rate": 1.828321171588957e-05, "loss": 0.0027, "step": 73280 }, { "grad_norm": 0.07772606611251831, "learning_rate": 1.827043117539089e-05, "loss": 0.0025, "step": 73290 }, { "grad_norm": 0.09160926938056946, "learning_rate": 1.8257654104790116e-05, "loss": 0.0036, "step": 73300 }, { "grad_norm": 0.04072890058159828, "learning_rate": 1.82448805054846e-05, "loss": 0.0025, "step": 73310 }, { "grad_norm": 0.0576549656689167, "learning_rate": 1.823211037887119e-05, "loss": 0.0032, "step": 73320 }, { "grad_norm": 0.06528189778327942, "learning_rate": 1.821934372634645e-05, "loss": 0.0047, "step": 73330 }, { "grad_norm": 0.04335188865661621, "learning_rate": 1.8206580549306472e-05, "loss": 0.0042, "step": 73340 }, { "grad_norm": 0.05047134310007095, "learning_rate": 1.819382084914707e-05, "loss": 0.003, "step": 73350 }, { "grad_norm": 0.06600434333086014, "learning_rate": 1.8181064627263577e-05, "loss": 0.004, "step": 73360 }, { "grad_norm": 0.04628555476665497, "learning_rate": 1.8168311885051022e-05, "loss": 0.0038, "step": 73370 }, { "grad_norm": 0.08810107409954071, "learning_rate": 1.815556262390402e-05, "loss": 0.0034, "step": 73380 }, { "grad_norm": 0.06363809108734131, "learning_rate": 1.8142816845216826e-05, "loss": 0.0047, "step": 73390 }, { "grad_norm": 0.039715804159641266, "learning_rate": 1.8130074550383276e-05, "loss": 0.0027, "step": 73400 }, { "grad_norm": 0.08304252475500107, "learning_rate": 1.8117335740796848e-05, "loss": 0.0029, "step": 73410 }, { "grad_norm": 0.059417273849248886, "learning_rate": 1.810460041785067e-05, "loss": 0.0028, "step": 73420 }, { "grad_norm": 0.03775830939412117, "learning_rate": 1.809186858293742e-05, "loss": 0.0032, "step": 73430 }, { "grad_norm": 0.05504471808671951, "learning_rate": 1.8079140237449448e-05, "loss": 0.0026, "step": 73440 }, { "grad_norm": 0.04479227960109711, "learning_rate": 1.8066415382778707e-05, "loss": 0.0051, "step": 73450 }, { "grad_norm": 0.06939394026994705, "learning_rate": 1.8053694020316786e-05, "loss": 0.003, "step": 73460 }, { "grad_norm": 0.0531628243625164, "learning_rate": 1.8040976151454837e-05, "loss": 0.0034, "step": 73470 }, { "grad_norm": 0.06405665725469589, "learning_rate": 1.8028261777583704e-05, "loss": 0.0033, "step": 73480 }, { "grad_norm": 0.07277371734380722, "learning_rate": 1.8015550900093758e-05, "loss": 0.0051, "step": 73490 }, { "grad_norm": 0.0387323796749115, "learning_rate": 1.800284352037512e-05, "loss": 0.0035, "step": 73500 }, { "grad_norm": 0.04470705986022949, "learning_rate": 1.799013963981738e-05, "loss": 0.0022, "step": 73510 }, { "grad_norm": 0.051471415907144547, "learning_rate": 1.797743925980987e-05, "loss": 0.0035, "step": 73520 }, { "grad_norm": 0.04308846965432167, "learning_rate": 1.7964742381741435e-05, "loss": 0.0035, "step": 73530 }, { "grad_norm": 0.030191339552402496, "learning_rate": 1.7952049007000627e-05, "loss": 0.0046, "step": 73540 }, { "grad_norm": 0.06439214199781418, "learning_rate": 1.7939359136975535e-05, "loss": 0.0031, "step": 73550 }, { "grad_norm": 0.07356243580579758, "learning_rate": 1.7926672773053932e-05, "loss": 0.0052, "step": 73560 }, { "grad_norm": 0.0926215797662735, "learning_rate": 1.791398991662317e-05, "loss": 0.003, "step": 73570 }, { "grad_norm": 0.08869657665491104, "learning_rate": 1.7901310569070244e-05, "loss": 0.004, "step": 73580 }, { "grad_norm": 0.06848745793104172, "learning_rate": 1.7888634731781724e-05, "loss": 0.005, "step": 73590 }, { "grad_norm": 0.07130637764930725, "learning_rate": 1.7875962406143844e-05, "loss": 0.0038, "step": 73600 }, { "grad_norm": 0.06799644231796265, "learning_rate": 1.7863293593542403e-05, "loss": 0.002, "step": 73610 }, { "grad_norm": 0.04630255699157715, "learning_rate": 1.7850628295362853e-05, "loss": 0.0024, "step": 73620 }, { "grad_norm": 0.09412531554698944, "learning_rate": 1.7837966512990263e-05, "loss": 0.0042, "step": 73630 }, { "grad_norm": 0.06264808028936386, "learning_rate": 1.782530824780931e-05, "loss": 0.0032, "step": 73640 }, { "grad_norm": 0.05553179606795311, "learning_rate": 1.7812653501204263e-05, "loss": 0.0027, "step": 73650 }, { "grad_norm": 0.04305879771709442, "learning_rate": 1.780000227455903e-05, "loss": 0.0047, "step": 73660 }, { "grad_norm": 0.039184682071208954, "learning_rate": 1.778735456925716e-05, "loss": 0.0032, "step": 73670 }, { "grad_norm": 0.08451637625694275, "learning_rate": 1.7774710386681737e-05, "loss": 0.0043, "step": 73680 }, { "grad_norm": 0.04192013293504715, "learning_rate": 1.7762069728215545e-05, "loss": 0.0027, "step": 73690 }, { "grad_norm": 0.05247919633984566, "learning_rate": 1.7749432595240932e-05, "loss": 0.0029, "step": 73700 }, { "grad_norm": 0.03529355302453041, "learning_rate": 1.773679898913991e-05, "loss": 0.0052, "step": 73710 }, { "grad_norm": 0.05698050558567047, "learning_rate": 1.7724168911294015e-05, "loss": 0.0033, "step": 73720 }, { "grad_norm": 0.07133688032627106, "learning_rate": 1.7711542363084504e-05, "loss": 0.0039, "step": 73730 }, { "grad_norm": 0.07132021337747574, "learning_rate": 1.7698919345892158e-05, "loss": 0.0048, "step": 73740 }, { "grad_norm": 0.047371093183755875, "learning_rate": 1.768629986109743e-05, "loss": 0.0029, "step": 73750 }, { "grad_norm": 0.042984094470739365, "learning_rate": 1.7673683910080357e-05, "loss": 0.0032, "step": 73760 }, { "grad_norm": 0.07485732436180115, "learning_rate": 1.7661071494220638e-05, "loss": 0.0039, "step": 73770 }, { "grad_norm": 0.0719359740614891, "learning_rate": 1.7648462614897498e-05, "loss": 0.0029, "step": 73780 }, { "grad_norm": 0.05668492615222931, "learning_rate": 1.7635857273489863e-05, "loss": 0.0029, "step": 73790 }, { "grad_norm": 0.050466202199459076, "learning_rate": 1.7623255471376198e-05, "loss": 0.0056, "step": 73800 }, { "grad_norm": 0.05472920462489128, "learning_rate": 1.7610657209934646e-05, "loss": 0.0028, "step": 73810 }, { "grad_norm": 0.054456502199172974, "learning_rate": 1.7598062490542915e-05, "loss": 0.0041, "step": 73820 }, { "grad_norm": 0.057817280292510986, "learning_rate": 1.7585471314578377e-05, "loss": 0.0034, "step": 73830 }, { "grad_norm": 0.0504763089120388, "learning_rate": 1.7572883683417945e-05, "loss": 0.003, "step": 73840 }, { "grad_norm": 0.05066767334938049, "learning_rate": 1.7560299598438212e-05, "loss": 0.0051, "step": 73850 }, { "grad_norm": 0.07085468620061874, "learning_rate": 1.7547719061015334e-05, "loss": 0.0032, "step": 73860 }, { "grad_norm": 0.034823227673769, "learning_rate": 1.75351420725251e-05, "loss": 0.0044, "step": 73870 }, { "grad_norm": 0.03655168414115906, "learning_rate": 1.7522568634342923e-05, "loss": 0.0051, "step": 73880 }, { "grad_norm": 0.045157525688409805, "learning_rate": 1.7509998747843825e-05, "loss": 0.0039, "step": 73890 }, { "grad_norm": 0.04394611343741417, "learning_rate": 1.7497432414402403e-05, "loss": 0.003, "step": 73900 }, { "grad_norm": 0.04373862221837044, "learning_rate": 1.7484869635392903e-05, "loss": 0.0023, "step": 73910 }, { "grad_norm": 0.06423746049404144, "learning_rate": 1.747231041218919e-05, "loss": 0.002, "step": 73920 }, { "grad_norm": 0.03368525952100754, "learning_rate": 1.7459754746164692e-05, "loss": 0.0028, "step": 73930 }, { "grad_norm": 0.051784079521894455, "learning_rate": 1.744720263869249e-05, "loss": 0.0028, "step": 73940 }, { "grad_norm": 0.03842446208000183, "learning_rate": 1.7434654091145263e-05, "loss": 0.0021, "step": 73950 }, { "grad_norm": 0.045214585959911346, "learning_rate": 1.742210910489532e-05, "loss": 0.0046, "step": 73960 }, { "grad_norm": 0.04807240888476372, "learning_rate": 1.7409567681314532e-05, "loss": 0.0024, "step": 73970 }, { "grad_norm": 0.05337647348642349, "learning_rate": 1.7397029821774434e-05, "loss": 0.0031, "step": 73980 }, { "grad_norm": 0.052211545407772064, "learning_rate": 1.7384495527646126e-05, "loss": 0.006, "step": 73990 }, { "grad_norm": 0.05803774669766426, "learning_rate": 1.737196480030035e-05, "loss": 0.0024, "step": 74000 }, { "grad_norm": 0.02882535383105278, "learning_rate": 1.7359437641107447e-05, "loss": 0.0028, "step": 74010 }, { "grad_norm": 0.07623717933893204, "learning_rate": 1.7346914051437387e-05, "loss": 0.0039, "step": 74020 }, { "grad_norm": 0.05260467901825905, "learning_rate": 1.733439403265969e-05, "loss": 0.003, "step": 74030 }, { "grad_norm": 0.07944028824567795, "learning_rate": 1.732187758614357e-05, "loss": 0.0026, "step": 74040 }, { "grad_norm": 0.044839948415756226, "learning_rate": 1.730936471325777e-05, "loss": 0.0033, "step": 74050 }, { "grad_norm": 0.06584887951612473, "learning_rate": 1.7296855415370718e-05, "loss": 0.0027, "step": 74060 }, { "grad_norm": 0.061708156019449234, "learning_rate": 1.728434969385035e-05, "loss": 0.0028, "step": 74070 }, { "grad_norm": 0.04353911057114601, "learning_rate": 1.7271847550064346e-05, "loss": 0.0027, "step": 74080 }, { "grad_norm": 0.036580298095941544, "learning_rate": 1.725934898537987e-05, "loss": 0.0034, "step": 74090 }, { "grad_norm": 0.06963017582893372, "learning_rate": 1.7246854001163783e-05, "loss": 0.0047, "step": 74100 }, { "grad_norm": 0.08656927943229675, "learning_rate": 1.7234362598782476e-05, "loss": 0.0042, "step": 74110 }, { "grad_norm": 0.06895411014556885, "learning_rate": 1.722187477960202e-05, "loss": 0.0034, "step": 74120 }, { "grad_norm": 0.08268719911575317, "learning_rate": 1.7209390544988057e-05, "loss": 0.0034, "step": 74130 }, { "grad_norm": 0.06839025765657425, "learning_rate": 1.7196909896305856e-05, "loss": 0.0042, "step": 74140 }, { "grad_norm": 0.05436558648943901, "learning_rate": 1.718443283492026e-05, "loss": 0.0054, "step": 74150 }, { "grad_norm": 0.09745283424854279, "learning_rate": 1.7171959362195746e-05, "loss": 0.0032, "step": 74160 }, { "grad_norm": 0.06311625987291336, "learning_rate": 1.715948947949642e-05, "loss": 0.0052, "step": 74170 }, { "grad_norm": 0.05727612227201462, "learning_rate": 1.714702318818593e-05, "loss": 0.0033, "step": 74180 }, { "grad_norm": 0.06194325163960457, "learning_rate": 1.7134560489627592e-05, "loss": 0.0029, "step": 74190 }, { "grad_norm": 0.0696493610739708, "learning_rate": 1.712210138518431e-05, "loss": 0.004, "step": 74200 }, { "grad_norm": 0.04392446577548981, "learning_rate": 1.71096458762186e-05, "loss": 0.0038, "step": 74210 }, { "grad_norm": 0.056775737553834915, "learning_rate": 1.7097193964092552e-05, "loss": 0.0056, "step": 74220 }, { "grad_norm": 0.06588111817836761, "learning_rate": 1.708474565016792e-05, "loss": 0.0031, "step": 74230 }, { "grad_norm": 0.07525540143251419, "learning_rate": 1.7072300935806e-05, "loss": 0.0039, "step": 74240 }, { "grad_norm": 0.06596248596906662, "learning_rate": 1.705985982236776e-05, "loss": 0.0037, "step": 74250 }, { "grad_norm": 0.04293346032500267, "learning_rate": 1.7047422311213697e-05, "loss": 0.0047, "step": 74260 }, { "grad_norm": 0.05117369443178177, "learning_rate": 1.703498840370401e-05, "loss": 0.0023, "step": 74270 }, { "grad_norm": 0.0552983358502388, "learning_rate": 1.702255810119842e-05, "loss": 0.0026, "step": 74280 }, { "grad_norm": 0.06853367388248444, "learning_rate": 1.7010131405056305e-05, "loss": 0.0026, "step": 74290 }, { "grad_norm": 0.04445771127939224, "learning_rate": 1.6997708316636613e-05, "loss": 0.0053, "step": 74300 }, { "grad_norm": 0.05738389492034912, "learning_rate": 1.6985288837297932e-05, "loss": 0.0044, "step": 74310 }, { "grad_norm": 0.0787765309214592, "learning_rate": 1.6972872968398394e-05, "loss": 0.0054, "step": 74320 }, { "grad_norm": 0.03483441472053528, "learning_rate": 1.6960460711295845e-05, "loss": 0.0039, "step": 74330 }, { "grad_norm": 0.07982172816991806, "learning_rate": 1.6948052067347623e-05, "loss": 0.0024, "step": 74340 }, { "grad_norm": 0.05853322893381119, "learning_rate": 1.693564703791075e-05, "loss": 0.0035, "step": 74350 }, { "grad_norm": 0.03451082855463028, "learning_rate": 1.6923245624341793e-05, "loss": 0.0037, "step": 74360 }, { "grad_norm": 0.04044124484062195, "learning_rate": 1.6910847827996962e-05, "loss": 0.0025, "step": 74370 }, { "grad_norm": 0.029700644314289093, "learning_rate": 1.689845365023206e-05, "loss": 0.0034, "step": 74380 }, { "grad_norm": 0.07293170690536499, "learning_rate": 1.6886063092402505e-05, "loss": 0.0036, "step": 74390 }, { "grad_norm": 0.02967042848467827, "learning_rate": 1.6873676155863317e-05, "loss": 0.003, "step": 74400 }, { "grad_norm": 0.03319968655705452, "learning_rate": 1.6861292841969085e-05, "loss": 0.0023, "step": 74410 }, { "grad_norm": 0.07060456275939941, "learning_rate": 1.684891315207406e-05, "loss": 0.0029, "step": 74420 }, { "grad_norm": 0.03292229771614075, "learning_rate": 1.6836537087532034e-05, "loss": 0.0034, "step": 74430 }, { "grad_norm": 0.05962138995528221, "learning_rate": 1.6824164649696465e-05, "loss": 0.0038, "step": 74440 }, { "grad_norm": 0.03728325292468071, "learning_rate": 1.681179583992034e-05, "loss": 0.0025, "step": 74450 }, { "grad_norm": 0.030360836535692215, "learning_rate": 1.679943065955636e-05, "loss": 0.003, "step": 74460 }, { "grad_norm": 0.0626855194568634, "learning_rate": 1.678706910995671e-05, "loss": 0.0034, "step": 74470 }, { "grad_norm": 0.0739888995885849, "learning_rate": 1.6774711192473262e-05, "loss": 0.0032, "step": 74480 }, { "grad_norm": 0.05380020663142204, "learning_rate": 1.676235690845743e-05, "loss": 0.0041, "step": 74490 }, { "grad_norm": 0.061192646622657776, "learning_rate": 1.6750006259260285e-05, "loss": 0.004, "step": 74500 }, { "grad_norm": 0.09382887184619904, "learning_rate": 1.673765924623244e-05, "loss": 0.0036, "step": 74510 }, { "grad_norm": 0.07769384980201721, "learning_rate": 1.67253158707242e-05, "loss": 0.0026, "step": 74520 }, { "grad_norm": 0.04671730473637581, "learning_rate": 1.671297613408537e-05, "loss": 0.0024, "step": 74530 }, { "grad_norm": 0.04846768453717232, "learning_rate": 1.6700640037665444e-05, "loss": 0.0021, "step": 74540 }, { "grad_norm": 0.056602008640766144, "learning_rate": 1.668830758281344e-05, "loss": 0.0037, "step": 74550 }, { "grad_norm": 0.03720482438802719, "learning_rate": 1.6675978770878052e-05, "loss": 0.0036, "step": 74560 }, { "grad_norm": 0.031170262023806572, "learning_rate": 1.6663653603207506e-05, "loss": 0.0037, "step": 74570 }, { "grad_norm": 0.05496222525835037, "learning_rate": 1.6651332081149685e-05, "loss": 0.003, "step": 74580 }, { "grad_norm": 0.044858504086732864, "learning_rate": 1.6639014206052046e-05, "loss": 0.0031, "step": 74590 }, { "grad_norm": 0.07116517424583435, "learning_rate": 1.6626699979261657e-05, "loss": 0.0043, "step": 74600 }, { "grad_norm": 0.06366587430238724, "learning_rate": 1.6614389402125197e-05, "loss": 0.0049, "step": 74610 }, { "grad_norm": 0.036974091082811356, "learning_rate": 1.66020824759889e-05, "loss": 0.0029, "step": 74620 }, { "grad_norm": 0.05273599922657013, "learning_rate": 1.6589779202198675e-05, "loss": 0.0024, "step": 74630 }, { "grad_norm": 0.036755118519067764, "learning_rate": 1.6577479582099924e-05, "loss": 0.0043, "step": 74640 }, { "grad_norm": 0.041017938405275345, "learning_rate": 1.6565183617037788e-05, "loss": 0.0027, "step": 74650 }, { "grad_norm": 0.07166627794504166, "learning_rate": 1.6552891308356894e-05, "loss": 0.0032, "step": 74660 }, { "grad_norm": 0.06238851323723793, "learning_rate": 1.6540602657401526e-05, "loss": 0.0037, "step": 74670 }, { "grad_norm": 0.061522405594587326, "learning_rate": 1.6528317665515537e-05, "loss": 0.0031, "step": 74680 }, { "grad_norm": 0.060339003801345825, "learning_rate": 1.6516036334042416e-05, "loss": 0.003, "step": 74690 }, { "grad_norm": 0.05350790545344353, "learning_rate": 1.650375866432519e-05, "loss": 0.0028, "step": 74700 }, { "grad_norm": 0.054945629090070724, "learning_rate": 1.649148465770659e-05, "loss": 0.0022, "step": 74710 }, { "grad_norm": 0.059057947248220444, "learning_rate": 1.647921431552883e-05, "loss": 0.0062, "step": 74720 }, { "grad_norm": 0.0648876428604126, "learning_rate": 1.646694763913381e-05, "loss": 0.0043, "step": 74730 }, { "grad_norm": 0.08125144243240356, "learning_rate": 1.6454684629862966e-05, "loss": 0.0043, "step": 74740 }, { "grad_norm": 0.07983669638633728, "learning_rate": 1.6442425289057388e-05, "loss": 0.0035, "step": 74750 }, { "grad_norm": 0.045749396085739136, "learning_rate": 1.6430169618057718e-05, "loss": 0.0053, "step": 74760 }, { "grad_norm": 0.04298620671033859, "learning_rate": 1.641791761820422e-05, "loss": 0.0021, "step": 74770 }, { "grad_norm": 0.05419696494936943, "learning_rate": 1.640566929083676e-05, "loss": 0.0029, "step": 74780 }, { "grad_norm": 0.053934793919324875, "learning_rate": 1.6393424637294823e-05, "loss": 0.0066, "step": 74790 }, { "grad_norm": 0.06584141403436661, "learning_rate": 1.638118365891742e-05, "loss": 0.0033, "step": 74800 }, { "grad_norm": 0.04902113601565361, "learning_rate": 1.636894635704323e-05, "loss": 0.0032, "step": 74810 }, { "grad_norm": 0.06035120412707329, "learning_rate": 1.6356712733010517e-05, "loss": 0.0021, "step": 74820 }, { "grad_norm": 0.07376157492399216, "learning_rate": 1.6344482788157096e-05, "loss": 0.0042, "step": 74830 }, { "grad_norm": 0.04699721187353134, "learning_rate": 1.6332256523820444e-05, "loss": 0.0056, "step": 74840 }, { "grad_norm": 0.06603637337684631, "learning_rate": 1.6320033941337597e-05, "loss": 0.0044, "step": 74850 }, { "grad_norm": 0.05034460872411728, "learning_rate": 1.6307815042045216e-05, "loss": 0.0028, "step": 74860 }, { "grad_norm": 0.04198654368519783, "learning_rate": 1.6295599827279513e-05, "loss": 0.0031, "step": 74870 }, { "grad_norm": 0.05555370822548866, "learning_rate": 1.6283388298376356e-05, "loss": 0.0025, "step": 74880 }, { "grad_norm": 0.039688751101493835, "learning_rate": 1.6271180456671125e-05, "loss": 0.0028, "step": 74890 }, { "grad_norm": 0.03252673149108887, "learning_rate": 1.625897630349893e-05, "loss": 0.0031, "step": 74900 }, { "grad_norm": 0.06406185030937195, "learning_rate": 1.6246775840194335e-05, "loss": 0.0051, "step": 74910 }, { "grad_norm": 0.03585806116461754, "learning_rate": 1.6234579068091604e-05, "loss": 0.0028, "step": 74920 }, { "grad_norm": 0.04870543256402016, "learning_rate": 1.6222385988524524e-05, "loss": 0.0029, "step": 74930 }, { "grad_norm": 0.06284572929143906, "learning_rate": 1.6210196602826544e-05, "loss": 0.0023, "step": 74940 }, { "grad_norm": 0.07617267221212387, "learning_rate": 1.6198010912330646e-05, "loss": 0.0049, "step": 74950 }, { "grad_norm": 0.07998497039079666, "learning_rate": 1.6185828918369456e-05, "loss": 0.0042, "step": 74960 }, { "grad_norm": 0.06228525564074516, "learning_rate": 1.617365062227517e-05, "loss": 0.004, "step": 74970 }, { "grad_norm": 0.04084093123674393, "learning_rate": 1.616147602537962e-05, "loss": 0.0034, "step": 74980 }, { "grad_norm": 0.04365445300936699, "learning_rate": 1.614930512901415e-05, "loss": 0.0025, "step": 74990 }, { "grad_norm": 0.06328431516885757, "learning_rate": 1.6137137934509804e-05, "loss": 0.0025, "step": 75000 }, { "grad_norm": 0.06575899571180344, "learning_rate": 1.6124974443197123e-05, "loss": 0.0051, "step": 75010 }, { "grad_norm": 0.04234626144170761, "learning_rate": 1.6112814656406306e-05, "loss": 0.0035, "step": 75020 }, { "grad_norm": 0.04450676217675209, "learning_rate": 1.6100658575467125e-05, "loss": 0.0046, "step": 75030 }, { "grad_norm": 0.0533718466758728, "learning_rate": 1.6088506201708974e-05, "loss": 0.0023, "step": 75040 }, { "grad_norm": 0.06018826365470886, "learning_rate": 1.6076357536460785e-05, "loss": 0.0033, "step": 75050 }, { "grad_norm": 0.040873851627111435, "learning_rate": 1.606421258105113e-05, "loss": 0.0048, "step": 75060 }, { "grad_norm": 0.04126700386404991, "learning_rate": 1.6052071336808183e-05, "loss": 0.0032, "step": 75070 }, { "grad_norm": 0.06388528645038605, "learning_rate": 1.6039933805059658e-05, "loss": 0.0056, "step": 75080 }, { "grad_norm": 0.07664613425731659, "learning_rate": 1.6027799987132915e-05, "loss": 0.006, "step": 75090 }, { "grad_norm": 0.03094286099076271, "learning_rate": 1.6015669884354888e-05, "loss": 0.0037, "step": 75100 }, { "grad_norm": 0.058738842606544495, "learning_rate": 1.6003543498052125e-05, "loss": 0.0046, "step": 75110 }, { "grad_norm": 0.05650153383612633, "learning_rate": 1.5991420829550715e-05, "loss": 0.0025, "step": 75120 }, { "grad_norm": 0.07103748619556427, "learning_rate": 1.5979301880176405e-05, "loss": 0.0038, "step": 75130 }, { "grad_norm": 0.0612751841545105, "learning_rate": 1.5967186651254485e-05, "loss": 0.0036, "step": 75140 }, { "grad_norm": 0.06588425487279892, "learning_rate": 1.5955075144109853e-05, "loss": 0.0046, "step": 75150 }, { "grad_norm": 0.05364542081952095, "learning_rate": 1.594296736006703e-05, "loss": 0.0032, "step": 75160 }, { "grad_norm": 0.07303159683942795, "learning_rate": 1.59308633004501e-05, "loss": 0.0032, "step": 75170 }, { "grad_norm": 0.05201466754078865, "learning_rate": 1.5918762966582724e-05, "loss": 0.0031, "step": 75180 }, { "grad_norm": 0.042479291558265686, "learning_rate": 1.5906666359788203e-05, "loss": 0.0032, "step": 75190 }, { "grad_norm": 0.05590615049004555, "learning_rate": 1.589457348138938e-05, "loss": 0.0055, "step": 75200 }, { "grad_norm": 0.06212214007973671, "learning_rate": 1.588248433270872e-05, "loss": 0.0021, "step": 75210 }, { "grad_norm": 0.04031018912792206, "learning_rate": 1.587039891506828e-05, "loss": 0.0034, "step": 75220 }, { "grad_norm": 0.084856316447258, "learning_rate": 1.5858317229789715e-05, "loss": 0.005, "step": 75230 }, { "grad_norm": 0.057767532765865326, "learning_rate": 1.584623927819423e-05, "loss": 0.0035, "step": 75240 }, { "grad_norm": 0.05270104482769966, "learning_rate": 1.583416506160269e-05, "loss": 0.0054, "step": 75250 }, { "grad_norm": 0.03256944566965103, "learning_rate": 1.5822094581335472e-05, "loss": 0.0038, "step": 75260 }, { "grad_norm": 0.05588745325803757, "learning_rate": 1.5810027838712604e-05, "loss": 0.0025, "step": 75270 }, { "grad_norm": 0.08407142013311386, "learning_rate": 1.5797964835053687e-05, "loss": 0.0038, "step": 75280 }, { "grad_norm": 0.036591868847608566, "learning_rate": 1.5785905571677933e-05, "loss": 0.0036, "step": 75290 }, { "grad_norm": 0.03504021093249321, "learning_rate": 1.5773850049904086e-05, "loss": 0.0028, "step": 75300 }, { "grad_norm": 0.05824173241853714, "learning_rate": 1.5761798271050543e-05, "loss": 0.0036, "step": 75310 }, { "grad_norm": 0.06177709624171257, "learning_rate": 1.5749750236435277e-05, "loss": 0.0022, "step": 75320 }, { "grad_norm": 0.051511526107788086, "learning_rate": 1.5737705947375815e-05, "loss": 0.0034, "step": 75330 }, { "grad_norm": 0.055805500596761703, "learning_rate": 1.5725665405189322e-05, "loss": 0.0028, "step": 75340 }, { "grad_norm": 0.04481714591383934, "learning_rate": 1.571362861119253e-05, "loss": 0.0031, "step": 75350 }, { "grad_norm": 0.07815235108137131, "learning_rate": 1.570159556670177e-05, "loss": 0.0046, "step": 75360 }, { "grad_norm": 0.06615374982357025, "learning_rate": 1.5689566273032946e-05, "loss": 0.003, "step": 75370 }, { "grad_norm": 0.033883024007081985, "learning_rate": 1.5677540731501583e-05, "loss": 0.0054, "step": 75380 }, { "grad_norm": 0.050185926258563995, "learning_rate": 1.5665518943422746e-05, "loss": 0.0031, "step": 75390 }, { "grad_norm": 0.036914099007844925, "learning_rate": 1.5653500910111134e-05, "loss": 0.0028, "step": 75400 }, { "grad_norm": 0.049055904150009155, "learning_rate": 1.5641486632881023e-05, "loss": 0.0024, "step": 75410 }, { "grad_norm": 0.07528384029865265, "learning_rate": 1.56294761130463e-05, "loss": 0.0031, "step": 75420 }, { "grad_norm": 0.039494838565588, "learning_rate": 1.5617469351920373e-05, "loss": 0.0035, "step": 75430 }, { "grad_norm": 0.08506558835506439, "learning_rate": 1.5605466350816315e-05, "loss": 0.0032, "step": 75440 }, { "grad_norm": 0.06651367992162704, "learning_rate": 1.559346711104674e-05, "loss": 0.0039, "step": 75450 }, { "grad_norm": 0.043531812727451324, "learning_rate": 1.5581471633923883e-05, "loss": 0.0026, "step": 75460 }, { "grad_norm": 0.04033684358000755, "learning_rate": 1.5569479920759505e-05, "loss": 0.0017, "step": 75470 }, { "grad_norm": 0.03287440538406372, "learning_rate": 1.5557491972865072e-05, "loss": 0.0023, "step": 75480 }, { "grad_norm": 0.05649729073047638, "learning_rate": 1.554550779155152e-05, "loss": 0.0057, "step": 75490 }, { "grad_norm": 0.06898097693920135, "learning_rate": 1.5533527378129453e-05, "loss": 0.0031, "step": 75500 }, { "grad_norm": 0.044377487152814865, "learning_rate": 1.552155073390899e-05, "loss": 0.0039, "step": 75510 }, { "grad_norm": 0.05593252182006836, "learning_rate": 1.5509577860199903e-05, "loss": 0.0031, "step": 75520 }, { "grad_norm": 0.04922134056687355, "learning_rate": 1.5497608758311525e-05, "loss": 0.005, "step": 75530 }, { "grad_norm": 0.07498645782470703, "learning_rate": 1.5485643429552797e-05, "loss": 0.003, "step": 75540 }, { "grad_norm": 0.06778869777917862, "learning_rate": 1.5473681875232197e-05, "loss": 0.0049, "step": 75550 }, { "grad_norm": 0.06058648228645325, "learning_rate": 1.5461724096657837e-05, "loss": 0.0034, "step": 75560 }, { "grad_norm": 0.05375564470887184, "learning_rate": 1.5449770095137416e-05, "loss": 0.0034, "step": 75570 }, { "grad_norm": 0.07439995557069778, "learning_rate": 1.5437819871978178e-05, "loss": 0.0024, "step": 75580 }, { "grad_norm": 0.08452409505844116, "learning_rate": 1.5425873428486992e-05, "loss": 0.0024, "step": 75590 }, { "grad_norm": 0.04968058690428734, "learning_rate": 1.5413930765970312e-05, "loss": 0.0021, "step": 75600 }, { "grad_norm": 0.037174660712480545, "learning_rate": 1.5401991885734175e-05, "loss": 0.0028, "step": 75610 }, { "grad_norm": 0.06753513216972351, "learning_rate": 1.5390056789084174e-05, "loss": 0.0037, "step": 75620 }, { "grad_norm": 0.07405730336904526, "learning_rate": 1.5378125477325545e-05, "loss": 0.0035, "step": 75630 }, { "grad_norm": 0.04734795168042183, "learning_rate": 1.5366197951763046e-05, "loss": 0.004, "step": 75640 }, { "grad_norm": 0.0624627061188221, "learning_rate": 1.5354274213701075e-05, "loss": 0.003, "step": 75650 }, { "grad_norm": 0.04859844222664833, "learning_rate": 1.5342354264443566e-05, "loss": 0.0031, "step": 75660 }, { "grad_norm": 0.037361808121204376, "learning_rate": 1.533043810529411e-05, "loss": 0.0034, "step": 75670 }, { "grad_norm": 0.04381125792860985, "learning_rate": 1.5318525737555807e-05, "loss": 0.0033, "step": 75680 }, { "grad_norm": 0.03831274062395096, "learning_rate": 1.53066171625314e-05, "loss": 0.0039, "step": 75690 }, { "grad_norm": 0.06572374701499939, "learning_rate": 1.529471238152317e-05, "loss": 0.002, "step": 75700 }, { "grad_norm": 0.05538395419716835, "learning_rate": 1.5282811395833023e-05, "loss": 0.0022, "step": 75710 }, { "grad_norm": 0.03793419525027275, "learning_rate": 1.5270914206762397e-05, "loss": 0.0042, "step": 75720 }, { "grad_norm": 0.03427645564079285, "learning_rate": 1.525902081561241e-05, "loss": 0.0037, "step": 75730 }, { "grad_norm": 0.0672672912478447, "learning_rate": 1.5247131223683664e-05, "loss": 0.0032, "step": 75740 }, { "grad_norm": 0.05022541806101799, "learning_rate": 1.5235245432276408e-05, "loss": 0.0025, "step": 75750 }, { "grad_norm": 0.037793390452861786, "learning_rate": 1.5223363442690424e-05, "loss": 0.0036, "step": 75760 }, { "grad_norm": 0.038148753345012665, "learning_rate": 1.5211485256225133e-05, "loss": 0.0028, "step": 75770 }, { "grad_norm": 0.040468282997608185, "learning_rate": 1.519961087417951e-05, "loss": 0.0061, "step": 75780 }, { "grad_norm": 0.07483833283185959, "learning_rate": 1.5187740297852137e-05, "loss": 0.0036, "step": 75790 }, { "grad_norm": 0.062351733446121216, "learning_rate": 1.517587352854113e-05, "loss": 0.0036, "step": 75800 }, { "grad_norm": 0.06258001178503036, "learning_rate": 1.516401056754423e-05, "loss": 0.0031, "step": 75810 }, { "grad_norm": 0.04497429355978966, "learning_rate": 1.5152151416158783e-05, "loss": 0.0031, "step": 75820 }, { "grad_norm": 0.06161196529865265, "learning_rate": 1.5140296075681642e-05, "loss": 0.0057, "step": 75830 }, { "grad_norm": 0.04768846929073334, "learning_rate": 1.5128444547409331e-05, "loss": 0.0024, "step": 75840 }, { "grad_norm": 0.023411830887198448, "learning_rate": 1.5116596832637864e-05, "loss": 0.0032, "step": 75850 }, { "grad_norm": 0.036863502115011215, "learning_rate": 1.5104752932662952e-05, "loss": 0.0033, "step": 75860 }, { "grad_norm": 0.07457339018583298, "learning_rate": 1.5092912848779778e-05, "loss": 0.0026, "step": 75870 }, { "grad_norm": 0.06279824674129486, "learning_rate": 1.5081076582283194e-05, "loss": 0.0025, "step": 75880 }, { "grad_norm": 0.047731660306453705, "learning_rate": 1.5069244134467553e-05, "loss": 0.0033, "step": 75890 }, { "grad_norm": 0.0438041165471077, "learning_rate": 1.5057415506626876e-05, "loss": 0.0023, "step": 75900 }, { "grad_norm": 0.043565478175878525, "learning_rate": 1.5045590700054668e-05, "loss": 0.003, "step": 75910 }, { "grad_norm": 0.0565909780561924, "learning_rate": 1.5033769716044144e-05, "loss": 0.0038, "step": 75920 }, { "grad_norm": 0.08000032603740692, "learning_rate": 1.5021952555887975e-05, "loss": 0.0037, "step": 75930 }, { "grad_norm": 0.05182865634560585, "learning_rate": 1.50101392208785e-05, "loss": 0.0032, "step": 75940 }, { "grad_norm": 0.06840092688798904, "learning_rate": 1.4998329712307579e-05, "loss": 0.0037, "step": 75950 }, { "grad_norm": 0.061098989099264145, "learning_rate": 1.4986524031466708e-05, "loss": 0.0027, "step": 75960 }, { "grad_norm": 0.06069161742925644, "learning_rate": 1.4974722179646916e-05, "loss": 0.004, "step": 75970 }, { "grad_norm": 0.05600333586335182, "learning_rate": 1.4962924158138847e-05, "loss": 0.0029, "step": 75980 }, { "grad_norm": 0.05836816877126694, "learning_rate": 1.4951129968232714e-05, "loss": 0.0024, "step": 75990 }, { "grad_norm": 0.032282400876283646, "learning_rate": 1.4939339611218329e-05, "loss": 0.0048, "step": 76000 }, { "grad_norm": 0.04577535390853882, "learning_rate": 1.4927553088385032e-05, "loss": 0.0027, "step": 76010 }, { "grad_norm": 0.0660557672381401, "learning_rate": 1.4915770401021801e-05, "loss": 0.0028, "step": 76020 }, { "grad_norm": 0.03002103976905346, "learning_rate": 1.4903991550417185e-05, "loss": 0.0035, "step": 76030 }, { "grad_norm": 0.05438901111483574, "learning_rate": 1.4892216537859277e-05, "loss": 0.0023, "step": 76040 }, { "grad_norm": 0.038046881556510925, "learning_rate": 1.4880445364635775e-05, "loss": 0.004, "step": 76050 }, { "grad_norm": 0.07349962741136551, "learning_rate": 1.4868678032033973e-05, "loss": 0.0028, "step": 76060 }, { "grad_norm": 0.08159039169549942, "learning_rate": 1.4856914541340739e-05, "loss": 0.0031, "step": 76070 }, { "grad_norm": 0.05858206748962402, "learning_rate": 1.4845154893842472e-05, "loss": 0.0023, "step": 76080 }, { "grad_norm": 0.06236813962459564, "learning_rate": 1.4833399090825234e-05, "loss": 0.0039, "step": 76090 }, { "grad_norm": 0.04138234630227089, "learning_rate": 1.4821647133574562e-05, "loss": 0.0019, "step": 76100 }, { "grad_norm": 0.05692322179675102, "learning_rate": 1.4809899023375706e-05, "loss": 0.0035, "step": 76110 }, { "grad_norm": 0.08482838422060013, "learning_rate": 1.4798154761513366e-05, "loss": 0.0033, "step": 76120 }, { "grad_norm": 0.06719670444726944, "learning_rate": 1.4786414349271916e-05, "loss": 0.0033, "step": 76130 }, { "grad_norm": 0.046376943588256836, "learning_rate": 1.4774677787935226e-05, "loss": 0.0034, "step": 76140 }, { "grad_norm": 0.04126846790313721, "learning_rate": 1.4762945078786838e-05, "loss": 0.0034, "step": 76150 }, { "grad_norm": 0.05990204960107803, "learning_rate": 1.4751216223109777e-05, "loss": 0.0035, "step": 76160 }, { "grad_norm": 0.036952391266822815, "learning_rate": 1.4739491222186714e-05, "loss": 0.004, "step": 76170 }, { "grad_norm": 0.07901475578546524, "learning_rate": 1.4727770077299879e-05, "loss": 0.0021, "step": 76180 }, { "grad_norm": 0.08938652276992798, "learning_rate": 1.4716052789731094e-05, "loss": 0.0026, "step": 76190 }, { "grad_norm": 0.07891502976417542, "learning_rate": 1.470433936076171e-05, "loss": 0.0025, "step": 76200 }, { "grad_norm": 0.10803228616714478, "learning_rate": 1.4692629791672719e-05, "loss": 0.0031, "step": 76210 }, { "grad_norm": 0.06991485506296158, "learning_rate": 1.468092408374464e-05, "loss": 0.0042, "step": 76220 }, { "grad_norm": 0.07101801782846451, "learning_rate": 1.4669222238257602e-05, "loss": 0.0043, "step": 76230 }, { "grad_norm": 0.06964319199323654, "learning_rate": 1.4657524256491301e-05, "loss": 0.0028, "step": 76240 }, { "grad_norm": 0.03347795829176903, "learning_rate": 1.464583013972502e-05, "loss": 0.0028, "step": 76250 }, { "grad_norm": 0.06816459447145462, "learning_rate": 1.4634139889237591e-05, "loss": 0.003, "step": 76260 }, { "grad_norm": 0.08107919991016388, "learning_rate": 1.4622453506307448e-05, "loss": 0.0035, "step": 76270 }, { "grad_norm": 0.06320025771856308, "learning_rate": 1.461077099221262e-05, "loss": 0.0029, "step": 76280 }, { "grad_norm": 0.06471503525972366, "learning_rate": 1.4599092348230652e-05, "loss": 0.0034, "step": 76290 }, { "grad_norm": 0.05926648527383804, "learning_rate": 1.4587417575638724e-05, "loss": 0.0058, "step": 76300 }, { "grad_norm": 0.06324592232704163, "learning_rate": 1.4575746675713563e-05, "loss": 0.0047, "step": 76310 }, { "grad_norm": 0.09383026510477066, "learning_rate": 1.4564079649731504e-05, "loss": 0.0031, "step": 76320 }, { "grad_norm": 0.05253484845161438, "learning_rate": 1.4552416498968402e-05, "loss": 0.0028, "step": 76330 }, { "grad_norm": 0.03373672440648079, "learning_rate": 1.4540757224699747e-05, "loss": 0.0023, "step": 76340 }, { "grad_norm": 0.04540606588125229, "learning_rate": 1.452910182820056e-05, "loss": 0.0039, "step": 76350 }, { "grad_norm": 0.05364774167537689, "learning_rate": 1.4517450310745468e-05, "loss": 0.0028, "step": 76360 }, { "grad_norm": 0.05245761200785637, "learning_rate": 1.450580267360866e-05, "loss": 0.0035, "step": 76370 }, { "grad_norm": 0.03894432261586189, "learning_rate": 1.4494158918063922e-05, "loss": 0.0033, "step": 76380 }, { "grad_norm": 0.05435527116060257, "learning_rate": 1.4482519045384568e-05, "loss": 0.0022, "step": 76390 }, { "grad_norm": 0.044559817761182785, "learning_rate": 1.4470883056843548e-05, "loss": 0.0025, "step": 76400 }, { "grad_norm": 0.03766839578747749, "learning_rate": 1.4459250953713322e-05, "loss": 0.0044, "step": 76410 }, { "grad_norm": 0.06293999403715134, "learning_rate": 1.444762273726598e-05, "loss": 0.0026, "step": 76420 }, { "grad_norm": 0.03155336529016495, "learning_rate": 1.443599840877316e-05, "loss": 0.0028, "step": 76430 }, { "grad_norm": 0.03204319626092911, "learning_rate": 1.4424377969506103e-05, "loss": 0.0026, "step": 76440 }, { "grad_norm": 0.03293219581246376, "learning_rate": 1.4412761420735565e-05, "loss": 0.0041, "step": 76450 }, { "grad_norm": 0.05907129496335983, "learning_rate": 1.4401148763731953e-05, "loss": 0.0034, "step": 76460 }, { "grad_norm": 0.058772291988134384, "learning_rate": 1.4389539999765173e-05, "loss": 0.0025, "step": 76470 }, { "grad_norm": 0.07499174773693085, "learning_rate": 1.4377935130104758e-05, "loss": 0.0055, "step": 76480 }, { "grad_norm": 0.06787016987800598, "learning_rate": 1.4366334156019801e-05, "loss": 0.0043, "step": 76490 }, { "grad_norm": 0.0443902313709259, "learning_rate": 1.435473707877898e-05, "loss": 0.004, "step": 76500 }, { "grad_norm": 0.06364859640598297, "learning_rate": 1.4343143899650508e-05, "loss": 0.0029, "step": 76510 }, { "grad_norm": 0.0361689068377018, "learning_rate": 1.4331554619902204e-05, "loss": 0.0026, "step": 76520 }, { "grad_norm": 0.04703517258167267, "learning_rate": 1.4319969240801484e-05, "loss": 0.0042, "step": 76530 }, { "grad_norm": 0.05269886180758476, "learning_rate": 1.430838776361526e-05, "loss": 0.0028, "step": 76540 }, { "grad_norm": 0.04762827232480049, "learning_rate": 1.4296810189610093e-05, "loss": 0.0044, "step": 76550 }, { "grad_norm": 0.050160232931375504, "learning_rate": 1.4285236520052087e-05, "loss": 0.0029, "step": 76560 }, { "grad_norm": 0.038612041622400284, "learning_rate": 1.4273666756206933e-05, "loss": 0.0025, "step": 76570 }, { "grad_norm": 0.030976582318544388, "learning_rate": 1.4262100899339848e-05, "loss": 0.0032, "step": 76580 }, { "grad_norm": 0.06824909895658493, "learning_rate": 1.4250538950715696e-05, "loss": 0.0037, "step": 76590 }, { "grad_norm": 0.03214874118566513, "learning_rate": 1.4238980911598842e-05, "loss": 0.0027, "step": 76600 }, { "grad_norm": 0.06795868277549744, "learning_rate": 1.4227426783253267e-05, "loss": 0.0027, "step": 76610 }, { "grad_norm": 0.07848047465085983, "learning_rate": 1.4215876566942521e-05, "loss": 0.0028, "step": 76620 }, { "grad_norm": 0.057929523289203644, "learning_rate": 1.4204330263929721e-05, "loss": 0.0032, "step": 76630 }, { "grad_norm": 0.04205456003546715, "learning_rate": 1.4192787875477537e-05, "loss": 0.0033, "step": 76640 }, { "grad_norm": 0.05005178228020668, "learning_rate": 1.4181249402848246e-05, "loss": 0.0035, "step": 76650 }, { "grad_norm": 0.04355262219905853, "learning_rate": 1.4169714847303656e-05, "loss": 0.0028, "step": 76660 }, { "grad_norm": 0.02936551347374916, "learning_rate": 1.4158184210105197e-05, "loss": 0.0025, "step": 76670 }, { "grad_norm": 0.052133820950984955, "learning_rate": 1.4146657492513787e-05, "loss": 0.0024, "step": 76680 }, { "grad_norm": 0.048697780817747116, "learning_rate": 1.4135134695790047e-05, "loss": 0.0025, "step": 76690 }, { "grad_norm": 0.07585464417934418, "learning_rate": 1.4123615821194037e-05, "loss": 0.0032, "step": 76700 }, { "grad_norm": 0.03321017697453499, "learning_rate": 1.4112100869985473e-05, "loss": 0.0027, "step": 76710 }, { "grad_norm": 0.04573100432753563, "learning_rate": 1.4100589843423589e-05, "loss": 0.0038, "step": 76720 }, { "grad_norm": 0.05396893620491028, "learning_rate": 1.408908274276723e-05, "loss": 0.0034, "step": 76730 }, { "grad_norm": 0.06505228579044342, "learning_rate": 1.4077579569274786e-05, "loss": 0.0035, "step": 76740 }, { "grad_norm": 0.051894158124923706, "learning_rate": 1.4066080324204245e-05, "loss": 0.0035, "step": 76750 }, { "grad_norm": 0.03882855921983719, "learning_rate": 1.405458500881312e-05, "loss": 0.0027, "step": 76760 }, { "grad_norm": 0.05460336431860924, "learning_rate": 1.4043093624358538e-05, "loss": 0.0022, "step": 76770 }, { "grad_norm": 0.0701584443449974, "learning_rate": 1.4031606172097189e-05, "loss": 0.0043, "step": 76780 }, { "grad_norm": 0.059711754322052, "learning_rate": 1.40201226532853e-05, "loss": 0.0024, "step": 76790 }, { "grad_norm": 0.04710674658417702, "learning_rate": 1.4008643069178701e-05, "loss": 0.0027, "step": 76800 }, { "grad_norm": 0.05642251297831535, "learning_rate": 1.399716742103278e-05, "loss": 0.0029, "step": 76810 }, { "grad_norm": 0.05903032794594765, "learning_rate": 1.398569571010252e-05, "loss": 0.0028, "step": 76820 }, { "grad_norm": 0.049049679189920425, "learning_rate": 1.3974227937642414e-05, "loss": 0.0027, "step": 76830 }, { "grad_norm": 0.06206236779689789, "learning_rate": 1.3962764104906595e-05, "loss": 0.0028, "step": 76840 }, { "grad_norm": 0.042716801166534424, "learning_rate": 1.3951304213148696e-05, "loss": 0.0033, "step": 76850 }, { "grad_norm": 0.06778321415185928, "learning_rate": 1.3939848263621979e-05, "loss": 0.0027, "step": 76860 }, { "grad_norm": 0.05785068869590759, "learning_rate": 1.3928396257579219e-05, "loss": 0.0026, "step": 76870 }, { "grad_norm": 0.041688717901706696, "learning_rate": 1.3916948196272833e-05, "loss": 0.0026, "step": 76880 }, { "grad_norm": 0.04947974532842636, "learning_rate": 1.3905504080954734e-05, "loss": 0.0031, "step": 76890 }, { "grad_norm": 0.06137511134147644, "learning_rate": 1.3894063912876454e-05, "loss": 0.003, "step": 76900 }, { "grad_norm": 0.047526512295007706, "learning_rate": 1.3882627693289046e-05, "loss": 0.0037, "step": 76910 }, { "grad_norm": 0.06110819801688194, "learning_rate": 1.3871195423443183e-05, "loss": 0.0027, "step": 76920 }, { "grad_norm": 0.045386966317892075, "learning_rate": 1.3859767104589044e-05, "loss": 0.0037, "step": 76930 }, { "grad_norm": 0.05299128219485283, "learning_rate": 1.3848342737976467e-05, "loss": 0.003, "step": 76940 }, { "grad_norm": 0.04285881295800209, "learning_rate": 1.3836922324854757e-05, "loss": 0.0029, "step": 76950 }, { "grad_norm": 0.0475105382502079, "learning_rate": 1.382550586647287e-05, "loss": 0.0022, "step": 76960 }, { "grad_norm": 0.04863741993904114, "learning_rate": 1.3814093364079257e-05, "loss": 0.0035, "step": 76970 }, { "grad_norm": 0.03554077818989754, "learning_rate": 1.3802684818921991e-05, "loss": 0.0022, "step": 76980 }, { "grad_norm": 0.035296730697155, "learning_rate": 1.3791280232248687e-05, "loss": 0.0029, "step": 76990 }, { "grad_norm": 0.04949313774704933, "learning_rate": 1.3779879605306555e-05, "loss": 0.0042, "step": 77000 }, { "grad_norm": 0.07094408571720123, "learning_rate": 1.376848293934232e-05, "loss": 0.0025, "step": 77010 }, { "grad_norm": 0.043889276683330536, "learning_rate": 1.3757090235602316e-05, "loss": 0.0035, "step": 77020 }, { "grad_norm": 0.053755905479192734, "learning_rate": 1.3745701495332447e-05, "loss": 0.0027, "step": 77030 }, { "grad_norm": 0.04845187067985535, "learning_rate": 1.3734316719778145e-05, "loss": 0.0024, "step": 77040 }, { "grad_norm": 0.04393734782934189, "learning_rate": 1.3722935910184454e-05, "loss": 0.003, "step": 77050 }, { "grad_norm": 0.0767597034573555, "learning_rate": 1.3711559067795921e-05, "loss": 0.0039, "step": 77060 }, { "grad_norm": 0.026138411834836006, "learning_rate": 1.3700186193856756e-05, "loss": 0.0035, "step": 77070 }, { "grad_norm": 0.045394416898489, "learning_rate": 1.368881728961064e-05, "loss": 0.003, "step": 77080 }, { "grad_norm": 0.02847137488424778, "learning_rate": 1.3677452356300885e-05, "loss": 0.0026, "step": 77090 }, { "grad_norm": 0.07333262264728546, "learning_rate": 1.3666091395170321e-05, "loss": 0.0026, "step": 77100 }, { "grad_norm": 0.03895891085267067, "learning_rate": 1.3654734407461384e-05, "loss": 0.0022, "step": 77110 }, { "grad_norm": 0.03943482041358948, "learning_rate": 1.3643381394416022e-05, "loss": 0.0031, "step": 77120 }, { "grad_norm": 0.050417210906744, "learning_rate": 1.3632032357275838e-05, "loss": 0.0044, "step": 77130 }, { "grad_norm": 0.06000903248786926, "learning_rate": 1.3620687297281898e-05, "loss": 0.0027, "step": 77140 }, { "grad_norm": 0.04618477821350098, "learning_rate": 1.3609346215674917e-05, "loss": 0.0023, "step": 77150 }, { "grad_norm": 0.061301372945308685, "learning_rate": 1.3598009113695109e-05, "loss": 0.0024, "step": 77160 }, { "grad_norm": 0.027349181473255157, "learning_rate": 1.358667599258231e-05, "loss": 0.0024, "step": 77170 }, { "grad_norm": 0.06010274589061737, "learning_rate": 1.357534685357586e-05, "loss": 0.0022, "step": 77180 }, { "grad_norm": 0.051472265273332596, "learning_rate": 1.3564021697914715e-05, "loss": 0.0034, "step": 77190 }, { "grad_norm": 0.07375731319189072, "learning_rate": 1.3552700526837381e-05, "loss": 0.0034, "step": 77200 }, { "grad_norm": 0.03225487470626831, "learning_rate": 1.3541383341581925e-05, "loss": 0.0024, "step": 77210 }, { "grad_norm": 0.04491449519991875, "learning_rate": 1.3530070143385965e-05, "loss": 0.0037, "step": 77220 }, { "grad_norm": 0.07238727062940598, "learning_rate": 1.3518760933486702e-05, "loss": 0.0034, "step": 77230 }, { "grad_norm": 0.04492754861712456, "learning_rate": 1.3507455713120908e-05, "loss": 0.0027, "step": 77240 }, { "grad_norm": 0.022908005863428116, "learning_rate": 1.3496154483524876e-05, "loss": 0.0031, "step": 77250 }, { "grad_norm": 0.03734326735138893, "learning_rate": 1.3484857245934506e-05, "loss": 0.0027, "step": 77260 }, { "grad_norm": 0.06773559749126434, "learning_rate": 1.3473564001585248e-05, "loss": 0.0026, "step": 77270 }, { "grad_norm": 0.02908749133348465, "learning_rate": 1.3462274751712129e-05, "loss": 0.0021, "step": 77280 }, { "grad_norm": 0.030708346515893936, "learning_rate": 1.3450989497549698e-05, "loss": 0.0032, "step": 77290 }, { "grad_norm": 0.046814125031232834, "learning_rate": 1.3439708240332115e-05, "loss": 0.004, "step": 77300 }, { "grad_norm": 0.033451419323682785, "learning_rate": 1.3428430981293043e-05, "loss": 0.0024, "step": 77310 }, { "grad_norm": 0.07138196378946304, "learning_rate": 1.3417157721665808e-05, "loss": 0.0022, "step": 77320 }, { "grad_norm": 0.048701655119657516, "learning_rate": 1.3405888462683186e-05, "loss": 0.0026, "step": 77330 }, { "grad_norm": 0.06046688184142113, "learning_rate": 1.3394623205577595e-05, "loss": 0.0024, "step": 77340 }, { "grad_norm": 0.06388478726148605, "learning_rate": 1.3383361951580963e-05, "loss": 0.0027, "step": 77350 }, { "grad_norm": 0.029884250834584236, "learning_rate": 1.337210470192483e-05, "loss": 0.0028, "step": 77360 }, { "grad_norm": 0.04886353388428688, "learning_rate": 1.3360851457840245e-05, "loss": 0.003, "step": 77370 }, { "grad_norm": 0.03798801451921463, "learning_rate": 1.3349602220557856e-05, "loss": 0.0025, "step": 77380 }, { "grad_norm": 0.04192734882235527, "learning_rate": 1.3338356991307865e-05, "loss": 0.0039, "step": 77390 }, { "grad_norm": 0.05094399303197861, "learning_rate": 1.3327115771320048e-05, "loss": 0.0036, "step": 77400 }, { "grad_norm": 0.04258964583277702, "learning_rate": 1.3315878561823697e-05, "loss": 0.0037, "step": 77410 }, { "grad_norm": 0.04665564000606537, "learning_rate": 1.330464536404773e-05, "loss": 0.003, "step": 77420 }, { "grad_norm": 0.05908884480595589, "learning_rate": 1.3293416179220563e-05, "loss": 0.0033, "step": 77430 }, { "grad_norm": 0.08714038133621216, "learning_rate": 1.328219100857021e-05, "loss": 0.003, "step": 77440 }, { "grad_norm": 0.03227110952138901, "learning_rate": 1.327096985332425e-05, "loss": 0.0021, "step": 77450 }, { "grad_norm": 0.02783994935452938, "learning_rate": 1.3259752714709816e-05, "loss": 0.0023, "step": 77460 }, { "grad_norm": 0.037783801555633545, "learning_rate": 1.3248539593953569e-05, "loss": 0.0067, "step": 77470 }, { "grad_norm": 0.04138918220996857, "learning_rate": 1.323733049228178e-05, "loss": 0.0024, "step": 77480 }, { "grad_norm": 0.0811571404337883, "learning_rate": 1.3226125410920276e-05, "loss": 0.0033, "step": 77490 }, { "grad_norm": 0.030616395175457, "learning_rate": 1.3214924351094388e-05, "loss": 0.0041, "step": 77500 }, { "grad_norm": 0.07614966481924057, "learning_rate": 1.3203727314029069e-05, "loss": 0.0025, "step": 77510 }, { "grad_norm": 0.03404664248228073, "learning_rate": 1.3192534300948811e-05, "loss": 0.0023, "step": 77520 }, { "grad_norm": 0.06320245563983917, "learning_rate": 1.3181345313077675e-05, "loss": 0.0025, "step": 77530 }, { "grad_norm": 0.040832675993442535, "learning_rate": 1.3170160351639243e-05, "loss": 0.0049, "step": 77540 }, { "grad_norm": 0.029253646731376648, "learning_rate": 1.3158979417856721e-05, "loss": 0.0031, "step": 77550 }, { "grad_norm": 0.055104196071624756, "learning_rate": 1.31478025129528e-05, "loss": 0.0029, "step": 77560 }, { "grad_norm": 0.06444871425628662, "learning_rate": 1.3136629638149794e-05, "loss": 0.0025, "step": 77570 }, { "grad_norm": 0.03815362602472305, "learning_rate": 1.3125460794669547e-05, "loss": 0.0055, "step": 77580 }, { "grad_norm": 0.05605974793434143, "learning_rate": 1.3114295983733483e-05, "loss": 0.003, "step": 77590 }, { "grad_norm": 0.03278697654604912, "learning_rate": 1.3103135206562534e-05, "loss": 0.0056, "step": 77600 }, { "grad_norm": 0.029123304411768913, "learning_rate": 1.3091978464377264e-05, "loss": 0.0058, "step": 77610 }, { "grad_norm": 0.06759287416934967, "learning_rate": 1.3080825758397724e-05, "loss": 0.005, "step": 77620 }, { "grad_norm": 0.047539446502923965, "learning_rate": 1.3069677089843568e-05, "loss": 0.0023, "step": 77630 }, { "grad_norm": 0.042169276624917984, "learning_rate": 1.305853245993401e-05, "loss": 0.003, "step": 77640 }, { "grad_norm": 0.06568904966115952, "learning_rate": 1.3047391869887809e-05, "loss": 0.0021, "step": 77650 }, { "grad_norm": 0.07526873797178268, "learning_rate": 1.3036255320923263e-05, "loss": 0.006, "step": 77660 }, { "grad_norm": 0.039404116570949554, "learning_rate": 1.3025122814258273e-05, "loss": 0.0018, "step": 77670 }, { "grad_norm": 0.03616413101553917, "learning_rate": 1.3013994351110249e-05, "loss": 0.0023, "step": 77680 }, { "grad_norm": 0.047480177134275436, "learning_rate": 1.3002869932696194e-05, "loss": 0.0027, "step": 77690 }, { "grad_norm": 0.07366402447223663, "learning_rate": 1.2991749560232664e-05, "loss": 0.0016, "step": 77700 }, { "grad_norm": 0.04190920665860176, "learning_rate": 1.2980633234935768e-05, "loss": 0.0018, "step": 77710 }, { "grad_norm": 0.041584424674510956, "learning_rate": 1.2969520958021153e-05, "loss": 0.0027, "step": 77720 }, { "grad_norm": 0.06589970737695694, "learning_rate": 1.295841273070405e-05, "loss": 0.0039, "step": 77730 }, { "grad_norm": 0.05087461322546005, "learning_rate": 1.2947308554199255e-05, "loss": 0.0032, "step": 77740 }, { "grad_norm": 0.05560385808348656, "learning_rate": 1.2936208429721075e-05, "loss": 0.0033, "step": 77750 }, { "grad_norm": 0.04221703112125397, "learning_rate": 1.2925112358483416e-05, "loss": 0.0034, "step": 77760 }, { "grad_norm": 0.06896819919347763, "learning_rate": 1.2914020341699728e-05, "loss": 0.0029, "step": 77770 }, { "grad_norm": 0.03617573902010918, "learning_rate": 1.2902932380583032e-05, "loss": 0.0024, "step": 77780 }, { "grad_norm": 0.04285399243235588, "learning_rate": 1.2891848476345863e-05, "loss": 0.0042, "step": 77790 }, { "grad_norm": 0.03022989071905613, "learning_rate": 1.2880768630200368e-05, "loss": 0.0026, "step": 77800 }, { "grad_norm": 0.02524876594543457, "learning_rate": 1.286969284335819e-05, "loss": 0.0033, "step": 77810 }, { "grad_norm": 0.0461435429751873, "learning_rate": 1.2858621117030578e-05, "loss": 0.0029, "step": 77820 }, { "grad_norm": 0.029937518760561943, "learning_rate": 1.2847553452428324e-05, "loss": 0.0022, "step": 77830 }, { "grad_norm": 0.0675329938530922, "learning_rate": 1.283648985076178e-05, "loss": 0.0032, "step": 77840 }, { "grad_norm": 0.04159285128116608, "learning_rate": 1.2825430313240817e-05, "loss": 0.0023, "step": 77850 }, { "grad_norm": 0.05185529217123985, "learning_rate": 1.281437484107491e-05, "loss": 0.0027, "step": 77860 }, { "grad_norm": 0.05853372439742088, "learning_rate": 1.2803323435473058e-05, "loss": 0.005, "step": 77870 }, { "grad_norm": 0.07882469147443771, "learning_rate": 1.2792276097643841e-05, "loss": 0.0039, "step": 77880 }, { "grad_norm": 0.03661394491791725, "learning_rate": 1.2781232828795342e-05, "loss": 0.0018, "step": 77890 }, { "grad_norm": 0.04720331355929375, "learning_rate": 1.2770193630135292e-05, "loss": 0.0028, "step": 77900 }, { "grad_norm": 0.0607680082321167, "learning_rate": 1.2759158502870877e-05, "loss": 0.0038, "step": 77910 }, { "grad_norm": 0.033260758966207504, "learning_rate": 1.2748127448208913e-05, "loss": 0.0024, "step": 77920 }, { "grad_norm": 0.08910852670669556, "learning_rate": 1.2737100467355706e-05, "loss": 0.0024, "step": 77930 }, { "grad_norm": 0.08666811138391495, "learning_rate": 1.2726077561517169e-05, "loss": 0.003, "step": 77940 }, { "grad_norm": 0.059932999312877655, "learning_rate": 1.2715058731898755e-05, "loss": 0.0031, "step": 77950 }, { "grad_norm": 0.05267166718840599, "learning_rate": 1.2704043979705471e-05, "loss": 0.0023, "step": 77960 }, { "grad_norm": 0.04013938829302788, "learning_rate": 1.269303330614185e-05, "loss": 0.004, "step": 77970 }, { "grad_norm": 0.06949114054441452, "learning_rate": 1.2682026712412015e-05, "loss": 0.0022, "step": 77980 }, { "grad_norm": 0.05214434862136841, "learning_rate": 1.2671024199719644e-05, "loss": 0.0022, "step": 77990 }, { "grad_norm": 0.0499805323779583, "learning_rate": 1.2660025769267936e-05, "loss": 0.0024, "step": 78000 }, { "grad_norm": 0.05109814926981926, "learning_rate": 1.2649031422259667e-05, "loss": 0.0029, "step": 78010 }, { "grad_norm": 0.05107586085796356, "learning_rate": 1.2638041159897163e-05, "loss": 0.0038, "step": 78020 }, { "grad_norm": 0.06384561210870743, "learning_rate": 1.262705498338232e-05, "loss": 0.0034, "step": 78030 }, { "grad_norm": 0.05682240426540375, "learning_rate": 1.2616072893916542e-05, "loss": 0.0031, "step": 78040 }, { "grad_norm": 0.049633778631687164, "learning_rate": 1.2605094892700836e-05, "loss": 0.0032, "step": 78050 }, { "grad_norm": 0.06436581909656525, "learning_rate": 1.259412098093572e-05, "loss": 0.0074, "step": 78060 }, { "grad_norm": 0.055642303079366684, "learning_rate": 1.258315115982131e-05, "loss": 0.003, "step": 78070 }, { "grad_norm": 0.08589702099561691, "learning_rate": 1.2572185430557203e-05, "loss": 0.0024, "step": 78080 }, { "grad_norm": 0.0481988824903965, "learning_rate": 1.2561223794342659e-05, "loss": 0.0051, "step": 78090 }, { "grad_norm": 0.05494924634695053, "learning_rate": 1.255026625237637e-05, "loss": 0.0036, "step": 78100 }, { "grad_norm": 0.0492105707526207, "learning_rate": 1.2539312805856678e-05, "loss": 0.0036, "step": 78110 }, { "grad_norm": 0.05543147400021553, "learning_rate": 1.25283634559814e-05, "loss": 0.0023, "step": 78120 }, { "grad_norm": 0.054914653301239014, "learning_rate": 1.2517418203947972e-05, "loss": 0.0023, "step": 78130 }, { "grad_norm": 0.04955926537513733, "learning_rate": 1.2506477050953308e-05, "loss": 0.0034, "step": 78140 }, { "grad_norm": 0.04026113077998161, "learning_rate": 1.2495539998193972e-05, "loss": 0.0026, "step": 78150 }, { "grad_norm": 0.04014744609594345, "learning_rate": 1.2484607046865975e-05, "loss": 0.004, "step": 78160 }, { "grad_norm": 0.05537353456020355, "learning_rate": 1.2473678198164968e-05, "loss": 0.003, "step": 78170 }, { "grad_norm": 0.02503344602882862, "learning_rate": 1.2462753453286075e-05, "loss": 0.0025, "step": 78180 }, { "grad_norm": 0.0289724413305521, "learning_rate": 1.2451832813424031e-05, "loss": 0.0023, "step": 78190 }, { "grad_norm": 0.059467241168022156, "learning_rate": 1.2440916279773095e-05, "loss": 0.0025, "step": 78200 }, { "grad_norm": 0.04218417406082153, "learning_rate": 1.2430003853527101e-05, "loss": 0.0031, "step": 78210 }, { "grad_norm": 0.0956086739897728, "learning_rate": 1.2419095535879383e-05, "loss": 0.0044, "step": 78220 }, { "grad_norm": 0.04634840413928032, "learning_rate": 1.2408191328022878e-05, "loss": 0.0039, "step": 78230 }, { "grad_norm": 0.05164247751235962, "learning_rate": 1.2397291231150066e-05, "loss": 0.0035, "step": 78240 }, { "grad_norm": 0.028340721502900124, "learning_rate": 1.2386395246452937e-05, "loss": 0.0032, "step": 78250 }, { "grad_norm": 0.06029204651713371, "learning_rate": 1.2375503375123083e-05, "loss": 0.0039, "step": 78260 }, { "grad_norm": 0.033410876989364624, "learning_rate": 1.2364615618351583e-05, "loss": 0.0019, "step": 78270 }, { "grad_norm": 0.033010244369506836, "learning_rate": 1.2353731977329169e-05, "loss": 0.0026, "step": 78280 }, { "grad_norm": 0.03437322378158569, "learning_rate": 1.2342852453246007e-05, "loss": 0.0034, "step": 78290 }, { "grad_norm": 0.04044085368514061, "learning_rate": 1.2331977047291898e-05, "loss": 0.0033, "step": 78300 }, { "grad_norm": 0.04938848689198494, "learning_rate": 1.2321105760656137e-05, "loss": 0.0019, "step": 78310 }, { "grad_norm": 0.032198164612054825, "learning_rate": 1.2310238594527613e-05, "loss": 0.0028, "step": 78320 }, { "grad_norm": 0.030692875385284424, "learning_rate": 1.2299375550094693e-05, "loss": 0.0021, "step": 78330 }, { "grad_norm": 0.060958657413721085, "learning_rate": 1.2288516628545421e-05, "loss": 0.0034, "step": 78340 }, { "grad_norm": 0.05256922170519829, "learning_rate": 1.2277661831067255e-05, "loss": 0.0024, "step": 78350 }, { "grad_norm": 0.07214286178350449, "learning_rate": 1.2266811158847286e-05, "loss": 0.0047, "step": 78360 }, { "grad_norm": 0.04508307948708534, "learning_rate": 1.2255964613072107e-05, "loss": 0.0036, "step": 78370 }, { "grad_norm": 0.05584880709648132, "learning_rate": 1.2245122194927905e-05, "loss": 0.0039, "step": 78380 }, { "grad_norm": 0.040436048060655594, "learning_rate": 1.2234283905600364e-05, "loss": 0.0032, "step": 78390 }, { "grad_norm": 0.04254689812660217, "learning_rate": 1.222344974627475e-05, "loss": 0.0061, "step": 78400 }, { "grad_norm": 0.07686513662338257, "learning_rate": 1.2212619718135875e-05, "loss": 0.0043, "step": 78410 }, { "grad_norm": 0.050352953374385834, "learning_rate": 1.2201793822368118e-05, "loss": 0.0028, "step": 78420 }, { "grad_norm": 0.05981706082820892, "learning_rate": 1.2190972060155337e-05, "loss": 0.0031, "step": 78430 }, { "grad_norm": 0.03480060026049614, "learning_rate": 1.2180154432681013e-05, "loss": 0.0028, "step": 78440 }, { "grad_norm": 0.034641072154045105, "learning_rate": 1.2169340941128148e-05, "loss": 0.0041, "step": 78450 }, { "grad_norm": 0.02822648175060749, "learning_rate": 1.2158531586679267e-05, "loss": 0.0038, "step": 78460 }, { "grad_norm": 0.06739373505115509, "learning_rate": 1.2147726370516477e-05, "loss": 0.0022, "step": 78470 }, { "grad_norm": 0.04596472159028053, "learning_rate": 1.2136925293821422e-05, "loss": 0.0019, "step": 78480 }, { "grad_norm": 0.031886886805295944, "learning_rate": 1.2126128357775306e-05, "loss": 0.0019, "step": 78490 }, { "grad_norm": 0.04183612018823624, "learning_rate": 1.2115335563558838e-05, "loss": 0.0036, "step": 78500 }, { "grad_norm": 0.0380992516875267, "learning_rate": 1.2104546912352327e-05, "loss": 0.0033, "step": 78510 }, { "grad_norm": 0.07263296842575073, "learning_rate": 1.2093762405335557e-05, "loss": 0.004, "step": 78520 }, { "grad_norm": 0.05389631539583206, "learning_rate": 1.2082982043687974e-05, "loss": 0.0023, "step": 78530 }, { "grad_norm": 0.046260394155979156, "learning_rate": 1.207220582858845e-05, "loss": 0.0023, "step": 78540 }, { "grad_norm": 0.046460237354040146, "learning_rate": 1.206143376121549e-05, "loss": 0.0042, "step": 78550 }, { "grad_norm": 0.04994688555598259, "learning_rate": 1.2050665842747078e-05, "loss": 0.0031, "step": 78560 }, { "grad_norm": 0.08411084860563278, "learning_rate": 1.2039902074360804e-05, "loss": 0.0052, "step": 78570 }, { "grad_norm": 0.0584019236266613, "learning_rate": 1.2029142457233755e-05, "loss": 0.0036, "step": 78580 }, { "grad_norm": 0.07196064293384552, "learning_rate": 1.2018386992542601e-05, "loss": 0.0028, "step": 78590 }, { "grad_norm": 0.041152019053697586, "learning_rate": 1.2007635681463542e-05, "loss": 0.0035, "step": 78600 }, { "grad_norm": 0.0630137249827385, "learning_rate": 1.1996888525172334e-05, "loss": 0.0022, "step": 78610 }, { "grad_norm": 0.03725709021091461, "learning_rate": 1.198614552484425e-05, "loss": 0.0022, "step": 78620 }, { "grad_norm": 0.049332428723573685, "learning_rate": 1.197540668165415e-05, "loss": 0.0025, "step": 78630 }, { "grad_norm": 0.02566814050078392, "learning_rate": 1.1964671996776395e-05, "loss": 0.0019, "step": 78640 }, { "grad_norm": 0.041570041328668594, "learning_rate": 1.1953941471384922e-05, "loss": 0.0025, "step": 78650 }, { "grad_norm": 0.05876960977911949, "learning_rate": 1.1943215106653205e-05, "loss": 0.0027, "step": 78660 }, { "grad_norm": 0.016739312559366226, "learning_rate": 1.193249290375429e-05, "loss": 0.003, "step": 78670 }, { "grad_norm": 0.04546872153878212, "learning_rate": 1.1921774863860696e-05, "loss": 0.0032, "step": 78680 }, { "grad_norm": 0.05880536139011383, "learning_rate": 1.1911060988144556e-05, "loss": 0.0024, "step": 78690 }, { "grad_norm": 0.034964416176080704, "learning_rate": 1.1900351277777533e-05, "loss": 0.003, "step": 78700 }, { "grad_norm": 0.02613493613898754, "learning_rate": 1.1889645733930798e-05, "loss": 0.0023, "step": 78710 }, { "grad_norm": 0.05533187836408615, "learning_rate": 1.1878944357775107e-05, "loss": 0.0038, "step": 78720 }, { "grad_norm": 0.0628223791718483, "learning_rate": 1.1868247150480749e-05, "loss": 0.0026, "step": 78730 }, { "grad_norm": 0.06416812539100647, "learning_rate": 1.1857554113217567e-05, "loss": 0.0037, "step": 78740 }, { "grad_norm": 0.04889374226331711, "learning_rate": 1.1846865247154903e-05, "loss": 0.0037, "step": 78750 }, { "grad_norm": 0.043304406106472015, "learning_rate": 1.1836180553461711e-05, "loss": 0.0017, "step": 78760 }, { "grad_norm": 0.03417598456144333, "learning_rate": 1.1825500033306419e-05, "loss": 0.0039, "step": 78770 }, { "grad_norm": 0.09922873973846436, "learning_rate": 1.1814823687857052e-05, "loss": 0.0036, "step": 78780 }, { "grad_norm": 0.08421804755926132, "learning_rate": 1.1804151518281153e-05, "loss": 0.0018, "step": 78790 }, { "grad_norm": 0.08703706413507462, "learning_rate": 1.179348352574583e-05, "loss": 0.0023, "step": 78800 }, { "grad_norm": 0.02437741309404373, "learning_rate": 1.1782819711417697e-05, "loss": 0.0042, "step": 78810 }, { "grad_norm": 0.034927722066640854, "learning_rate": 1.1772160076462951e-05, "loss": 0.0016, "step": 78820 }, { "grad_norm": 0.05441906675696373, "learning_rate": 1.1761504622047293e-05, "loss": 0.0017, "step": 78830 }, { "grad_norm": 0.03554204851388931, "learning_rate": 1.1750853349335999e-05, "loss": 0.003, "step": 78840 }, { "grad_norm": 0.03727661073207855, "learning_rate": 1.1740206259493874e-05, "loss": 0.0016, "step": 78850 }, { "grad_norm": 0.04123162105679512, "learning_rate": 1.1729563353685286e-05, "loss": 0.0048, "step": 78860 }, { "grad_norm": 0.041883088648319244, "learning_rate": 1.1718924633074091e-05, "loss": 0.0021, "step": 78870 }, { "grad_norm": 0.029667159542441368, "learning_rate": 1.1708290098823765e-05, "loss": 0.0026, "step": 78880 }, { "grad_norm": 0.051551103591918945, "learning_rate": 1.1697659752097245e-05, "loss": 0.0032, "step": 78890 }, { "grad_norm": 0.045534465461969376, "learning_rate": 1.1687033594057068e-05, "loss": 0.0038, "step": 78900 }, { "grad_norm": 0.037483762949705124, "learning_rate": 1.1676411625865297e-05, "loss": 0.0024, "step": 78910 }, { "grad_norm": 0.04823721945285797, "learning_rate": 1.1665793848683542e-05, "loss": 0.0026, "step": 78920 }, { "grad_norm": 0.07495616376399994, "learning_rate": 1.1655180263672927e-05, "loss": 0.0066, "step": 78930 }, { "grad_norm": 0.06933499127626419, "learning_rate": 1.1644570871994142e-05, "loss": 0.0048, "step": 78940 }, { "grad_norm": 0.058203794062137604, "learning_rate": 1.163396567480744e-05, "loss": 0.003, "step": 78950 }, { "grad_norm": 0.03089030086994171, "learning_rate": 1.1623364673272552e-05, "loss": 0.0021, "step": 78960 }, { "grad_norm": 0.025063414126634598, "learning_rate": 1.1612767868548807e-05, "loss": 0.003, "step": 78970 }, { "grad_norm": 0.03971705213189125, "learning_rate": 1.1602175261795057e-05, "loss": 0.0019, "step": 78980 }, { "grad_norm": 0.038871150463819504, "learning_rate": 1.1591586854169696e-05, "loss": 0.0021, "step": 78990 }, { "grad_norm": 0.03445197269320488, "learning_rate": 1.158100264683064e-05, "loss": 0.0048, "step": 79000 }, { "grad_norm": 0.029837781563401222, "learning_rate": 1.1570422640935386e-05, "loss": 0.0019, "step": 79010 }, { "grad_norm": 0.052702371031045914, "learning_rate": 1.155984683764092e-05, "loss": 0.0045, "step": 79020 }, { "grad_norm": 0.025130053982138634, "learning_rate": 1.1549275238103812e-05, "loss": 0.0028, "step": 79030 }, { "grad_norm": 0.061725594103336334, "learning_rate": 1.1538707843480152e-05, "loss": 0.0035, "step": 79040 }, { "grad_norm": 0.06249505653977394, "learning_rate": 1.1528144654925593e-05, "loss": 0.0055, "step": 79050 }, { "grad_norm": 0.0362296923995018, "learning_rate": 1.1517585673595277e-05, "loss": 0.003, "step": 79060 }, { "grad_norm": 0.04864705726504326, "learning_rate": 1.150703090064395e-05, "loss": 0.002, "step": 79070 }, { "grad_norm": 0.08689161390066147, "learning_rate": 1.1496480337225835e-05, "loss": 0.0025, "step": 79080 }, { "grad_norm": 0.040589362382888794, "learning_rate": 1.1485933984494751e-05, "loss": 0.0027, "step": 79090 }, { "grad_norm": 0.029372522607445717, "learning_rate": 1.1475391843604e-05, "loss": 0.0027, "step": 79100 }, { "grad_norm": 0.04510302469134331, "learning_rate": 1.1464853915706503e-05, "loss": 0.0056, "step": 79110 }, { "grad_norm": 0.04144425317645073, "learning_rate": 1.1454320201954626e-05, "loss": 0.0014, "step": 79120 }, { "grad_norm": 0.04029291495680809, "learning_rate": 1.1443790703500356e-05, "loss": 0.0036, "step": 79130 }, { "grad_norm": 0.07075987011194229, "learning_rate": 1.1433265421495154e-05, "loss": 0.0045, "step": 79140 }, { "grad_norm": 0.02986927516758442, "learning_rate": 1.1422744357090065e-05, "loss": 0.0031, "step": 79150 }, { "grad_norm": 0.07047366350889206, "learning_rate": 1.1412227511435647e-05, "loss": 0.0026, "step": 79160 }, { "grad_norm": 0.047882091253995895, "learning_rate": 1.1401714885682025e-05, "loss": 0.0022, "step": 79170 }, { "grad_norm": 0.05024518817663193, "learning_rate": 1.1391206480978823e-05, "loss": 0.0024, "step": 79180 }, { "grad_norm": 0.06383843719959259, "learning_rate": 1.1380702298475227e-05, "loss": 0.0034, "step": 79190 }, { "grad_norm": 0.0388602688908577, "learning_rate": 1.1370202339319985e-05, "loss": 0.0026, "step": 79200 }, { "grad_norm": 0.051436759531497955, "learning_rate": 1.1359706604661319e-05, "loss": 0.0046, "step": 79210 }, { "grad_norm": 0.04857863485813141, "learning_rate": 1.1349215095647042e-05, "loss": 0.0034, "step": 79220 }, { "grad_norm": 0.06907793879508972, "learning_rate": 1.1338727813424493e-05, "loss": 0.0028, "step": 79230 }, { "grad_norm": 0.03980656713247299, "learning_rate": 1.132824475914056e-05, "loss": 0.0018, "step": 79240 }, { "grad_norm": 0.036354295909404755, "learning_rate": 1.131776593394162e-05, "loss": 0.0026, "step": 79250 }, { "grad_norm": 0.04588976129889488, "learning_rate": 1.1307291338973657e-05, "loss": 0.0027, "step": 79260 }, { "grad_norm": 0.05666710063815117, "learning_rate": 1.1296820975382121e-05, "loss": 0.0025, "step": 79270 }, { "grad_norm": 0.029664548113942146, "learning_rate": 1.1286354844312074e-05, "loss": 0.0024, "step": 79280 }, { "grad_norm": 0.03412255272269249, "learning_rate": 1.1275892946908023e-05, "loss": 0.0025, "step": 79290 }, { "grad_norm": 0.036495085805654526, "learning_rate": 1.126543528431413e-05, "loss": 0.0044, "step": 79300 }, { "grad_norm": 0.03330555185675621, "learning_rate": 1.125498185767398e-05, "loss": 0.0052, "step": 79310 }, { "grad_norm": 0.046555694192647934, "learning_rate": 1.1244532668130781e-05, "loss": 0.0042, "step": 79320 }, { "grad_norm": 0.039294276386499405, "learning_rate": 1.1234087716827203e-05, "loss": 0.0028, "step": 79330 }, { "grad_norm": 0.0577361173927784, "learning_rate": 1.1223647004905524e-05, "loss": 0.0022, "step": 79340 }, { "grad_norm": 0.06560607999563217, "learning_rate": 1.1213210533507485e-05, "loss": 0.0042, "step": 79350 }, { "grad_norm": 0.056325677782297134, "learning_rate": 1.1202778303774447e-05, "loss": 0.0028, "step": 79360 }, { "grad_norm": 0.07736638188362122, "learning_rate": 1.1192350316847234e-05, "loss": 0.0034, "step": 79370 }, { "grad_norm": 0.0476844348013401, "learning_rate": 1.1181926573866258e-05, "loss": 0.003, "step": 79380 }, { "grad_norm": 0.04662000387907028, "learning_rate": 1.1171507075971416e-05, "loss": 0.0017, "step": 79390 }, { "grad_norm": 0.06569600850343704, "learning_rate": 1.1161091824302177e-05, "loss": 0.0032, "step": 79400 }, { "grad_norm": 0.07777198404073715, "learning_rate": 1.115068081999755e-05, "loss": 0.0049, "step": 79410 }, { "grad_norm": 0.05308368057012558, "learning_rate": 1.1140274064196071e-05, "loss": 0.003, "step": 79420 }, { "grad_norm": 0.056027840822935104, "learning_rate": 1.112987155803578e-05, "loss": 0.0046, "step": 79430 }, { "grad_norm": 0.07096923142671585, "learning_rate": 1.1119473302654298e-05, "loss": 0.0022, "step": 79440 }, { "grad_norm": 0.07312964648008347, "learning_rate": 1.1109079299188769e-05, "loss": 0.0027, "step": 79450 }, { "grad_norm": 0.05927550420165062, "learning_rate": 1.1098689548775847e-05, "loss": 0.0031, "step": 79460 }, { "grad_norm": 0.06427105516195297, "learning_rate": 1.1088304052551762e-05, "loss": 0.0036, "step": 79470 }, { "grad_norm": 0.041057996451854706, "learning_rate": 1.1077922811652215e-05, "loss": 0.0029, "step": 79480 }, { "grad_norm": 0.05230766907334328, "learning_rate": 1.1067545827212534e-05, "loss": 0.0035, "step": 79490 }, { "grad_norm": 0.030981512740254402, "learning_rate": 1.1057173100367496e-05, "loss": 0.0038, "step": 79500 }, { "grad_norm": 0.06500870734453201, "learning_rate": 1.1046804632251474e-05, "loss": 0.0035, "step": 79510 }, { "grad_norm": 0.02885635569691658, "learning_rate": 1.1036440423998318e-05, "loss": 0.0023, "step": 79520 }, { "grad_norm": 0.04287269338965416, "learning_rate": 1.1026080476741469e-05, "loss": 0.0028, "step": 79530 }, { "grad_norm": 0.050095800310373306, "learning_rate": 1.1015724791613836e-05, "loss": 0.0037, "step": 79540 }, { "grad_norm": 0.030469726771116257, "learning_rate": 1.1005373369747951e-05, "loss": 0.002, "step": 79550 }, { "grad_norm": 0.03123840130865574, "learning_rate": 1.09950262122758e-05, "loss": 0.0034, "step": 79560 }, { "grad_norm": 0.04596463590860367, "learning_rate": 1.0984683320328948e-05, "loss": 0.0029, "step": 79570 }, { "grad_norm": 0.042715735733509064, "learning_rate": 1.0974344695038458e-05, "loss": 0.0015, "step": 79580 }, { "grad_norm": 0.053044792264699936, "learning_rate": 1.0964010337534968e-05, "loss": 0.0023, "step": 79590 }, { "grad_norm": 0.03315860033035278, "learning_rate": 1.0953680248948611e-05, "loss": 0.0025, "step": 79600 }, { "grad_norm": 0.047431495040655136, "learning_rate": 1.0943354430409069e-05, "loss": 0.002, "step": 79610 }, { "grad_norm": 0.028903208673000336, "learning_rate": 1.0933032883045574e-05, "loss": 0.0029, "step": 79620 }, { "grad_norm": 0.05773043632507324, "learning_rate": 1.0922715607986872e-05, "loss": 0.0029, "step": 79630 }, { "grad_norm": 0.042006731033325195, "learning_rate": 1.0912402606361234e-05, "loss": 0.0016, "step": 79640 }, { "grad_norm": 0.05832890793681145, "learning_rate": 1.090209387929647e-05, "loss": 0.0033, "step": 79650 }, { "grad_norm": 0.07731235772371292, "learning_rate": 1.0891789427919957e-05, "loss": 0.0034, "step": 79660 }, { "grad_norm": 0.03220187500119209, "learning_rate": 1.088148925335854e-05, "loss": 0.0019, "step": 79670 }, { "grad_norm": 0.042572371661663055, "learning_rate": 1.087119335673864e-05, "loss": 0.003, "step": 79680 }, { "grad_norm": 0.06009584292769432, "learning_rate": 1.0860901739186208e-05, "loss": 0.0031, "step": 79690 }, { "grad_norm": 0.06510593742132187, "learning_rate": 1.085061440182673e-05, "loss": 0.0031, "step": 79700 }, { "grad_norm": 0.04815106838941574, "learning_rate": 1.0840331345785193e-05, "loss": 0.0024, "step": 79710 }, { "grad_norm": 0.03118581511080265, "learning_rate": 1.083005257218615e-05, "loss": 0.0028, "step": 79720 }, { "grad_norm": 0.038116004317998886, "learning_rate": 1.0819778082153642e-05, "loss": 0.0024, "step": 79730 }, { "grad_norm": 0.03728831186890602, "learning_rate": 1.0809507876811325e-05, "loss": 0.0039, "step": 79740 }, { "grad_norm": 0.04231858253479004, "learning_rate": 1.0799241957282292e-05, "loss": 0.0031, "step": 79750 }, { "grad_norm": 0.07885727286338806, "learning_rate": 1.078898032468923e-05, "loss": 0.0035, "step": 79760 }, { "grad_norm": 0.058005500584840775, "learning_rate": 1.0778722980154315e-05, "loss": 0.0049, "step": 79770 }, { "grad_norm": 0.04729239642620087, "learning_rate": 1.0768469924799302e-05, "loss": 0.0034, "step": 79780 }, { "grad_norm": 0.05856640636920929, "learning_rate": 1.0758221159745419e-05, "loss": 0.0021, "step": 79790 }, { "grad_norm": 0.03644979000091553, "learning_rate": 1.074797668611347e-05, "loss": 0.002, "step": 79800 }, { "grad_norm": 0.0579046756029129, "learning_rate": 1.0737736505023777e-05, "loss": 0.0043, "step": 79810 }, { "grad_norm": 0.038439180701971054, "learning_rate": 1.0727500617596203e-05, "loss": 0.0027, "step": 79820 }, { "grad_norm": 0.05771642550826073, "learning_rate": 1.0717269024950104e-05, "loss": 0.0034, "step": 79830 }, { "grad_norm": 0.045214228332042694, "learning_rate": 1.070704172820442e-05, "loss": 0.0025, "step": 79840 }, { "grad_norm": 0.033741917461156845, "learning_rate": 1.0696818728477559e-05, "loss": 0.0022, "step": 79850 }, { "grad_norm": 0.05432065203785896, "learning_rate": 1.0686600026887511e-05, "loss": 0.0037, "step": 79860 }, { "grad_norm": 0.03185621649026871, "learning_rate": 1.0676385624551777e-05, "loss": 0.0035, "step": 79870 }, { "grad_norm": 0.041399069130420685, "learning_rate": 1.0666175522587402e-05, "loss": 0.0035, "step": 79880 }, { "grad_norm": 0.028137989342212677, "learning_rate": 1.0655969722110925e-05, "loss": 0.0026, "step": 79890 }, { "grad_norm": 0.07137901335954666, "learning_rate": 1.0645768224238439e-05, "loss": 0.0028, "step": 79900 }, { "grad_norm": 0.05061814561486244, "learning_rate": 1.063557103008559e-05, "loss": 0.0023, "step": 79910 }, { "grad_norm": 0.05839911103248596, "learning_rate": 1.0625378140767494e-05, "loss": 0.003, "step": 79920 }, { "grad_norm": 0.06258539110422134, "learning_rate": 1.0615189557398846e-05, "loss": 0.0031, "step": 79930 }, { "grad_norm": 0.05766410380601883, "learning_rate": 1.060500528109386e-05, "loss": 0.002, "step": 79940 }, { "grad_norm": 0.06260787695646286, "learning_rate": 1.0594825312966272e-05, "loss": 0.0029, "step": 79950 }, { "grad_norm": 0.07571157068014145, "learning_rate": 1.0584649654129336e-05, "loss": 0.0042, "step": 79960 }, { "grad_norm": 0.05551358684897423, "learning_rate": 1.0574478305695863e-05, "loss": 0.0029, "step": 79970 }, { "grad_norm": 0.060607485473155975, "learning_rate": 1.0564311268778154e-05, "loss": 0.003, "step": 79980 }, { "grad_norm": 0.07279904186725616, "learning_rate": 1.0554148544488073e-05, "loss": 0.0039, "step": 79990 }, { "grad_norm": 0.054550182074308395, "learning_rate": 1.0543990133937004e-05, "loss": 0.0037, "step": 80000 }, { "grad_norm": 0.05820264294743538, "learning_rate": 1.0533836038235868e-05, "loss": 0.0046, "step": 80010 }, { "grad_norm": 0.044899556785821915, "learning_rate": 1.052368625849507e-05, "loss": 0.0027, "step": 80020 }, { "grad_norm": 0.06353059411048889, "learning_rate": 1.05135407958246e-05, "loss": 0.0035, "step": 80030 }, { "grad_norm": 0.04099064692854881, "learning_rate": 1.0503399651333934e-05, "loss": 0.0037, "step": 80040 }, { "grad_norm": 0.05093922093510628, "learning_rate": 1.04932628261321e-05, "loss": 0.0025, "step": 80050 }, { "grad_norm": 0.04999414458870888, "learning_rate": 1.0483130321327644e-05, "loss": 0.0027, "step": 80060 }, { "grad_norm": 0.058289535343647, "learning_rate": 1.0473002138028654e-05, "loss": 0.0019, "step": 80070 }, { "grad_norm": 0.0588323213160038, "learning_rate": 1.0462878277342713e-05, "loss": 0.0022, "step": 80080 }, { "grad_norm": 0.05701516196131706, "learning_rate": 1.0452758740376972e-05, "loss": 0.005, "step": 80090 }, { "grad_norm": 0.042896728962659836, "learning_rate": 1.0442643528238067e-05, "loss": 0.0024, "step": 80100 }, { "grad_norm": 0.05591531842947006, "learning_rate": 1.0432532642032188e-05, "loss": 0.0025, "step": 80110 }, { "grad_norm": 0.05110284313559532, "learning_rate": 1.042242608286505e-05, "loss": 0.0023, "step": 80120 }, { "grad_norm": 0.05640706047415733, "learning_rate": 1.041232385184191e-05, "loss": 0.0031, "step": 80130 }, { "grad_norm": 0.04609149321913719, "learning_rate": 1.0402225950067495e-05, "loss": 0.0035, "step": 80140 }, { "grad_norm": 0.07094492763280869, "learning_rate": 1.0392132378646119e-05, "loss": 0.0034, "step": 80150 }, { "grad_norm": 0.05278436094522476, "learning_rate": 1.0382043138681607e-05, "loss": 0.0017, "step": 80160 }, { "grad_norm": 0.06786717474460602, "learning_rate": 1.0371958231277284e-05, "loss": 0.0029, "step": 80170 }, { "grad_norm": 0.027274172753095627, "learning_rate": 1.0361877657536024e-05, "loss": 0.0021, "step": 80180 }, { "grad_norm": 0.05830351635813713, "learning_rate": 1.035180141856023e-05, "loss": 0.004, "step": 80190 }, { "grad_norm": 0.04002058506011963, "learning_rate": 1.0341729515451836e-05, "loss": 0.0027, "step": 80200 }, { "grad_norm": 0.04538952559232712, "learning_rate": 1.0331661949312265e-05, "loss": 0.0019, "step": 80210 }, { "grad_norm": 0.053005650639534, "learning_rate": 1.0321598721242514e-05, "loss": 0.0031, "step": 80220 }, { "grad_norm": 0.02958034723997116, "learning_rate": 1.0311539832343064e-05, "loss": 0.0027, "step": 80230 }, { "grad_norm": 0.03849311172962189, "learning_rate": 1.0301485283713942e-05, "loss": 0.0033, "step": 80240 }, { "grad_norm": 0.04782059043645859, "learning_rate": 1.0291435076454703e-05, "loss": 0.0018, "step": 80250 }, { "grad_norm": 0.05492934584617615, "learning_rate": 1.028138921166444e-05, "loss": 0.0023, "step": 80260 }, { "grad_norm": 0.039299607276916504, "learning_rate": 1.0271347690441719e-05, "loss": 0.0049, "step": 80270 }, { "grad_norm": 0.045684851706027985, "learning_rate": 1.0261310513884698e-05, "loss": 0.0025, "step": 80280 }, { "grad_norm": 0.03966078907251358, "learning_rate": 1.0251277683090998e-05, "loss": 0.0032, "step": 80290 }, { "grad_norm": 0.041342783719301224, "learning_rate": 1.0241249199157826e-05, "loss": 0.0021, "step": 80300 }, { "grad_norm": 0.04941603168845177, "learning_rate": 1.0231225063181837e-05, "loss": 0.002, "step": 80310 }, { "grad_norm": 0.0511934831738472, "learning_rate": 1.0221205276259316e-05, "loss": 0.0039, "step": 80320 }, { "grad_norm": 0.058015741407871246, "learning_rate": 1.0211189839485958e-05, "loss": 0.0026, "step": 80330 }, { "grad_norm": 0.06416086852550507, "learning_rate": 1.0201178753957075e-05, "loss": 0.0047, "step": 80340 }, { "grad_norm": 0.05467355251312256, "learning_rate": 1.0191172020767432e-05, "loss": 0.0035, "step": 80350 }, { "grad_norm": 0.05864671990275383, "learning_rate": 1.0181169641011362e-05, "loss": 0.0027, "step": 80360 }, { "grad_norm": 0.04197177663445473, "learning_rate": 1.0171171615782721e-05, "loss": 0.0021, "step": 80370 }, { "grad_norm": 0.028431542217731476, "learning_rate": 1.0161177946174866e-05, "loss": 0.0023, "step": 80380 }, { "grad_norm": 0.03398182988166809, "learning_rate": 1.0151188633280706e-05, "loss": 0.0025, "step": 80390 }, { "grad_norm": 0.05875040590763092, "learning_rate": 1.0141203678192635e-05, "loss": 0.0025, "step": 80400 }, { "grad_norm": 0.031337689608335495, "learning_rate": 1.0131223082002612e-05, "loss": 0.0029, "step": 80410 }, { "grad_norm": 0.038580555468797684, "learning_rate": 1.012124684580208e-05, "loss": 0.0026, "step": 80420 }, { "grad_norm": 0.06199346110224724, "learning_rate": 1.0111274970682039e-05, "loss": 0.0032, "step": 80430 }, { "grad_norm": 0.030328553169965744, "learning_rate": 1.010130745773299e-05, "loss": 0.0029, "step": 80440 }, { "grad_norm": 0.058753181248903275, "learning_rate": 1.0091344308044986e-05, "loss": 0.0022, "step": 80450 }, { "grad_norm": 0.06002825126051903, "learning_rate": 1.0081385522707554e-05, "loss": 0.0025, "step": 80460 }, { "grad_norm": 0.06497983634471893, "learning_rate": 1.0071431102809803e-05, "loss": 0.0027, "step": 80470 }, { "grad_norm": 0.04400823637843132, "learning_rate": 1.0061481049440297e-05, "loss": 0.0043, "step": 80480 }, { "grad_norm": 0.060854677110910416, "learning_rate": 1.0051535363687187e-05, "loss": 0.0055, "step": 80490 }, { "grad_norm": 0.03362738713622093, "learning_rate": 1.0041594046638087e-05, "loss": 0.0037, "step": 80500 }, { "grad_norm": 0.03772392496466637, "learning_rate": 1.0031657099380204e-05, "loss": 0.0027, "step": 80510 }, { "grad_norm": 0.04963395372033119, "learning_rate": 1.0021724523000204e-05, "loss": 0.005, "step": 80520 }, { "grad_norm": 0.031829435378313065, "learning_rate": 1.001179631858431e-05, "loss": 0.0023, "step": 80530 }, { "grad_norm": 0.07149147987365723, "learning_rate": 1.0001872487218245e-05, "loss": 0.0027, "step": 80540 }, { "grad_norm": 0.04862121120095253, "learning_rate": 9.991953029987272e-06, "loss": 0.0026, "step": 80550 }, { "grad_norm": 0.044230181723833084, "learning_rate": 9.982037947976147e-06, "loss": 0.0022, "step": 80560 }, { "grad_norm": 0.057945750653743744, "learning_rate": 9.972127242269208e-06, "loss": 0.005, "step": 80570 }, { "grad_norm": 0.06631313264369965, "learning_rate": 9.962220913950238e-06, "loss": 0.0062, "step": 80580 }, { "grad_norm": 0.0656157061457634, "learning_rate": 9.952318964102591e-06, "loss": 0.003, "step": 80590 }, { "grad_norm": 0.041250478476285934, "learning_rate": 9.942421393809148e-06, "loss": 0.0019, "step": 80600 }, { "grad_norm": 0.06994584947824478, "learning_rate": 9.932528204152264e-06, "loss": 0.0025, "step": 80610 }, { "grad_norm": 0.06235887110233307, "learning_rate": 9.922639396213856e-06, "loss": 0.004, "step": 80620 }, { "grad_norm": 0.033675774931907654, "learning_rate": 9.912754971075344e-06, "loss": 0.0026, "step": 80630 }, { "grad_norm": 0.06288263946771622, "learning_rate": 9.902874929817696e-06, "loss": 0.0039, "step": 80640 }, { "grad_norm": 0.043066468089818954, "learning_rate": 9.892999273521342e-06, "loss": 0.0025, "step": 80650 }, { "grad_norm": 0.06562016904354095, "learning_rate": 9.883128003266302e-06, "loss": 0.0031, "step": 80660 }, { "grad_norm": 0.049220211803913116, "learning_rate": 9.87326112013206e-06, "loss": 0.0031, "step": 80670 }, { "grad_norm": 0.0723041519522667, "learning_rate": 9.863398625197662e-06, "loss": 0.0042, "step": 80680 }, { "grad_norm": 0.04732153192162514, "learning_rate": 9.853540519541615e-06, "loss": 0.0026, "step": 80690 }, { "grad_norm": 0.03741295635700226, "learning_rate": 9.843686804242053e-06, "loss": 0.0028, "step": 80700 }, { "grad_norm": 0.040706466883420944, "learning_rate": 9.833837480376506e-06, "loss": 0.0032, "step": 80710 }, { "grad_norm": 0.05521947145462036, "learning_rate": 9.82399254902212e-06, "loss": 0.0021, "step": 80720 }, { "grad_norm": 0.024919385090470314, "learning_rate": 9.814152011255494e-06, "loss": 0.0017, "step": 80730 }, { "grad_norm": 0.06284165382385254, "learning_rate": 9.804315868152797e-06, "loss": 0.0028, "step": 80740 }, { "grad_norm": 0.05924344062805176, "learning_rate": 9.79448412078966e-06, "loss": 0.0019, "step": 80750 }, { "grad_norm": 0.06054430082440376, "learning_rate": 9.784656770241312e-06, "loss": 0.004, "step": 80760 }, { "grad_norm": 0.04012664780020714, "learning_rate": 9.77483381758243e-06, "loss": 0.0032, "step": 80770 }, { "grad_norm": 0.040374305099248886, "learning_rate": 9.76501526388725e-06, "loss": 0.003, "step": 80780 }, { "grad_norm": 0.030927591025829315, "learning_rate": 9.7552011102295e-06, "loss": 0.0023, "step": 80790 }, { "grad_norm": 0.06432882696390152, "learning_rate": 9.745391357682449e-06, "loss": 0.0021, "step": 80800 }, { "grad_norm": 0.03582296520471573, "learning_rate": 9.735586007318875e-06, "loss": 0.0026, "step": 80810 }, { "grad_norm": 0.07874767482280731, "learning_rate": 9.725785060211096e-06, "loss": 0.0028, "step": 80820 }, { "grad_norm": 0.04330272600054741, "learning_rate": 9.715988517430896e-06, "loss": 0.0034, "step": 80830 }, { "grad_norm": 0.036376260221004486, "learning_rate": 9.706196380049625e-06, "loss": 0.005, "step": 80840 }, { "grad_norm": 0.06186980381608009, "learning_rate": 9.69640864913815e-06, "loss": 0.0025, "step": 80850 }, { "grad_norm": 0.04246589541435242, "learning_rate": 9.686625325766818e-06, "loss": 0.0028, "step": 80860 }, { "grad_norm": 0.03416721895337105, "learning_rate": 9.67684641100554e-06, "loss": 0.004, "step": 80870 }, { "grad_norm": 0.030007170513272285, "learning_rate": 9.667071905923692e-06, "loss": 0.0017, "step": 80880 }, { "grad_norm": 0.09208396077156067, "learning_rate": 9.657301811590242e-06, "loss": 0.0024, "step": 80890 }, { "grad_norm": 0.0690663605928421, "learning_rate": 9.647536129073603e-06, "loss": 0.0029, "step": 80900 }, { "grad_norm": 0.05710677430033684, "learning_rate": 9.637774859441751e-06, "loss": 0.0023, "step": 80910 }, { "grad_norm": 0.046578213572502136, "learning_rate": 9.628018003762152e-06, "loss": 0.0031, "step": 80920 }, { "grad_norm": 0.035355813801288605, "learning_rate": 9.618265563101813e-06, "loss": 0.0022, "step": 80930 }, { "grad_norm": 0.07425820082426071, "learning_rate": 9.608517538527211e-06, "loss": 0.0035, "step": 80940 }, { "grad_norm": 0.026612108573317528, "learning_rate": 9.598773931104431e-06, "loss": 0.0029, "step": 80950 }, { "grad_norm": 0.036358170211315155, "learning_rate": 9.589034741898983e-06, "loss": 0.0036, "step": 80960 }, { "grad_norm": 0.058137446641922, "learning_rate": 9.579299971975947e-06, "loss": 0.0038, "step": 80970 }, { "grad_norm": 0.0723731517791748, "learning_rate": 9.569569622399882e-06, "loss": 0.0024, "step": 80980 }, { "grad_norm": 0.03561469539999962, "learning_rate": 9.559843694234916e-06, "loss": 0.0032, "step": 80990 }, { "grad_norm": 0.04580390453338623, "learning_rate": 9.550122188544625e-06, "loss": 0.004, "step": 81000 }, { "grad_norm": 0.05841705948114395, "learning_rate": 9.540405106392152e-06, "loss": 0.0039, "step": 81010 }, { "grad_norm": 0.02245839312672615, "learning_rate": 9.53069244884015e-06, "loss": 0.0025, "step": 81020 }, { "grad_norm": 0.05184316635131836, "learning_rate": 9.52098421695079e-06, "loss": 0.0021, "step": 81030 }, { "grad_norm": 0.07173819839954376, "learning_rate": 9.511280411785728e-06, "loss": 0.0042, "step": 81040 }, { "grad_norm": 0.05668855458498001, "learning_rate": 9.50158103440616e-06, "loss": 0.0036, "step": 81050 }, { "grad_norm": 0.0439523346722126, "learning_rate": 9.491886085872815e-06, "loss": 0.0033, "step": 81060 }, { "grad_norm": 0.0648767501115799, "learning_rate": 9.482195567245895e-06, "loss": 0.0035, "step": 81070 }, { "grad_norm": 0.03100421465933323, "learning_rate": 9.472509479585151e-06, "loss": 0.0025, "step": 81080 }, { "grad_norm": 0.07382816076278687, "learning_rate": 9.462827823949833e-06, "loss": 0.0033, "step": 81090 }, { "grad_norm": 0.05654039978981018, "learning_rate": 9.453150601398735e-06, "loss": 0.0034, "step": 81100 }, { "grad_norm": 0.08665159344673157, "learning_rate": 9.44347781299011e-06, "loss": 0.0041, "step": 81110 }, { "grad_norm": 0.033429164439439774, "learning_rate": 9.43380945978179e-06, "loss": 0.0023, "step": 81120 }, { "grad_norm": 0.03663567453622818, "learning_rate": 9.424145542831047e-06, "loss": 0.0034, "step": 81130 }, { "grad_norm": 0.043314699083566666, "learning_rate": 9.414486063194778e-06, "loss": 0.0034, "step": 81140 }, { "grad_norm": 0.038234204053878784, "learning_rate": 9.404831021929273e-06, "loss": 0.0035, "step": 81150 }, { "grad_norm": 0.026701997965574265, "learning_rate": 9.395180420090427e-06, "loss": 0.002, "step": 81160 }, { "grad_norm": 0.06363799422979355, "learning_rate": 9.385534258733591e-06, "loss": 0.0027, "step": 81170 }, { "grad_norm": 0.050759363919496536, "learning_rate": 9.375892538913677e-06, "loss": 0.0033, "step": 81180 }, { "grad_norm": 0.07000937312841415, "learning_rate": 9.36625526168507e-06, "loss": 0.0034, "step": 81190 }, { "grad_norm": 0.0695442259311676, "learning_rate": 9.356622428101686e-06, "loss": 0.0035, "step": 81200 }, { "grad_norm": 0.04258909076452255, "learning_rate": 9.346994039216972e-06, "loss": 0.0029, "step": 81210 }, { "grad_norm": 0.05230768397450447, "learning_rate": 9.337370096083886e-06, "loss": 0.0023, "step": 81220 }, { "grad_norm": 0.09746435284614563, "learning_rate": 9.327750599754854e-06, "loss": 0.0024, "step": 81230 }, { "grad_norm": 0.08140022307634354, "learning_rate": 9.318135551281875e-06, "loss": 0.0023, "step": 81240 }, { "grad_norm": 0.06097135692834854, "learning_rate": 9.308524951716424e-06, "loss": 0.0026, "step": 81250 }, { "grad_norm": 0.06944389641284943, "learning_rate": 9.298918802109497e-06, "loss": 0.004, "step": 81260 }, { "grad_norm": 0.07701093703508377, "learning_rate": 9.289317103511625e-06, "loss": 0.0038, "step": 81270 }, { "grad_norm": 0.032968707382678986, "learning_rate": 9.279719856972835e-06, "loss": 0.003, "step": 81280 }, { "grad_norm": 0.0464327447116375, "learning_rate": 9.270127063542645e-06, "loss": 0.0031, "step": 81290 }, { "grad_norm": 0.0623815692961216, "learning_rate": 9.260538724270124e-06, "loss": 0.0027, "step": 81300 }, { "grad_norm": 0.03656524792313576, "learning_rate": 9.250954840203852e-06, "loss": 0.0017, "step": 81310 }, { "grad_norm": 0.05013401806354523, "learning_rate": 9.241375412391878e-06, "loss": 0.0022, "step": 81320 }, { "grad_norm": 0.03911423683166504, "learning_rate": 9.23180044188181e-06, "loss": 0.0035, "step": 81330 }, { "grad_norm": 0.037157874554395676, "learning_rate": 9.222229929720755e-06, "loss": 0.0036, "step": 81340 }, { "grad_norm": 0.07373396307229996, "learning_rate": 9.212663876955336e-06, "loss": 0.0032, "step": 81350 }, { "grad_norm": 0.04233142361044884, "learning_rate": 9.20310228463166e-06, "loss": 0.0021, "step": 81360 }, { "grad_norm": 0.08767074346542358, "learning_rate": 9.19354515379539e-06, "loss": 0.0043, "step": 81370 }, { "grad_norm": 0.04234324395656586, "learning_rate": 9.183992485491666e-06, "loss": 0.0029, "step": 81380 }, { "grad_norm": 0.028673037886619568, "learning_rate": 9.174444280765154e-06, "loss": 0.0036, "step": 81390 }, { "grad_norm": 0.03203614056110382, "learning_rate": 9.164900540660032e-06, "loss": 0.0043, "step": 81400 }, { "grad_norm": 0.03768075257539749, "learning_rate": 9.155361266220008e-06, "loss": 0.002, "step": 81410 }, { "grad_norm": 0.05139245092868805, "learning_rate": 9.145826458488254e-06, "loss": 0.0034, "step": 81420 }, { "grad_norm": 0.052086617797613144, "learning_rate": 9.136296118507509e-06, "loss": 0.0034, "step": 81430 }, { "grad_norm": 0.06099175661802292, "learning_rate": 9.126770247319966e-06, "loss": 0.003, "step": 81440 }, { "grad_norm": 0.08675617724657059, "learning_rate": 9.11724884596738e-06, "loss": 0.0025, "step": 81450 }, { "grad_norm": 0.04735932499170303, "learning_rate": 9.107731915490996e-06, "loss": 0.0021, "step": 81460 }, { "grad_norm": 0.06223732978105545, "learning_rate": 9.098219456931579e-06, "loss": 0.0032, "step": 81470 }, { "grad_norm": 0.04123954474925995, "learning_rate": 9.088711471329376e-06, "loss": 0.0021, "step": 81480 }, { "grad_norm": 0.036954306066036224, "learning_rate": 9.079207959724184e-06, "loss": 0.0029, "step": 81490 }, { "grad_norm": 0.05023018643260002, "learning_rate": 9.069708923155279e-06, "loss": 0.0021, "step": 81500 }, { "grad_norm": 0.06250400096178055, "learning_rate": 9.060214362661468e-06, "loss": 0.0021, "step": 81510 }, { "grad_norm": 0.06333565711975098, "learning_rate": 9.050724279281064e-06, "loss": 0.0025, "step": 81520 }, { "grad_norm": 0.05007501691579819, "learning_rate": 9.041238674051894e-06, "loss": 0.0028, "step": 81530 }, { "grad_norm": 0.0718863382935524, "learning_rate": 9.031757548011277e-06, "loss": 0.0043, "step": 81540 }, { "grad_norm": 0.021090596914291382, "learning_rate": 9.022280902196051e-06, "loss": 0.0032, "step": 81550 }, { "grad_norm": 0.05421983823180199, "learning_rate": 9.012808737642592e-06, "loss": 0.0022, "step": 81560 }, { "grad_norm": 0.03164182975888252, "learning_rate": 9.003341055386739e-06, "loss": 0.005, "step": 81570 }, { "grad_norm": 0.038128964602947235, "learning_rate": 8.993877856463862e-06, "loss": 0.0061, "step": 81580 }, { "grad_norm": 0.03217810019850731, "learning_rate": 8.984419141908856e-06, "loss": 0.0021, "step": 81590 }, { "grad_norm": 0.04120485484600067, "learning_rate": 8.974964912756117e-06, "loss": 0.0027, "step": 81600 }, { "grad_norm": 0.09764048457145691, "learning_rate": 8.965515170039523e-06, "loss": 0.0041, "step": 81610 }, { "grad_norm": 0.048065200448036194, "learning_rate": 8.956069914792514e-06, "loss": 0.0039, "step": 81620 }, { "grad_norm": 0.05145261436700821, "learning_rate": 8.946629148047969e-06, "loss": 0.0035, "step": 81630 }, { "grad_norm": 0.05275561287999153, "learning_rate": 8.93719287083834e-06, "loss": 0.0025, "step": 81640 }, { "grad_norm": 0.03813457116484642, "learning_rate": 8.927761084195569e-06, "loss": 0.0027, "step": 81650 }, { "grad_norm": 0.053975846618413925, "learning_rate": 8.918333789151096e-06, "loss": 0.0029, "step": 81660 }, { "grad_norm": 0.046651940792798996, "learning_rate": 8.90891098673587e-06, "loss": 0.0026, "step": 81670 }, { "grad_norm": 0.054454293102025986, "learning_rate": 8.89949267798037e-06, "loss": 0.0052, "step": 81680 }, { "grad_norm": 0.05909391865134239, "learning_rate": 8.890078863914542e-06, "loss": 0.0057, "step": 81690 }, { "grad_norm": 0.048193808645009995, "learning_rate": 8.880669545567894e-06, "loss": 0.002, "step": 81700 }, { "grad_norm": 0.021586133167147636, "learning_rate": 8.871264723969381e-06, "loss": 0.0022, "step": 81710 }, { "grad_norm": 0.03890522941946983, "learning_rate": 8.86186440014754e-06, "loss": 0.0032, "step": 81720 }, { "grad_norm": 0.03677838668227196, "learning_rate": 8.852468575130351e-06, "loss": 0.0057, "step": 81730 }, { "grad_norm": 0.06015811488032341, "learning_rate": 8.843077249945347e-06, "loss": 0.0023, "step": 81740 }, { "grad_norm": 0.048179224133491516, "learning_rate": 8.833690425619523e-06, "loss": 0.0021, "step": 81750 }, { "grad_norm": 0.0379289872944355, "learning_rate": 8.82430810317942e-06, "loss": 0.0028, "step": 81760 }, { "grad_norm": 0.051809798926115036, "learning_rate": 8.814930283651074e-06, "loss": 0.0025, "step": 81770 }, { "grad_norm": 0.030296016484498978, "learning_rate": 8.805556968060048e-06, "loss": 0.0027, "step": 81780 }, { "grad_norm": 0.040528442710638046, "learning_rate": 8.796188157431362e-06, "loss": 0.0037, "step": 81790 }, { "grad_norm": 0.03109184466302395, "learning_rate": 8.786823852789595e-06, "loss": 0.0036, "step": 81800 }, { "grad_norm": 0.026711389422416687, "learning_rate": 8.777464055158813e-06, "loss": 0.0038, "step": 81810 }, { "grad_norm": 0.044608574360609055, "learning_rate": 8.768108765562578e-06, "loss": 0.0024, "step": 81820 }, { "grad_norm": 0.02120167948305607, "learning_rate": 8.758757985023975e-06, "loss": 0.0022, "step": 81830 }, { "grad_norm": 0.033898379653692245, "learning_rate": 8.749411714565598e-06, "loss": 0.0019, "step": 81840 }, { "grad_norm": 0.04934876412153244, "learning_rate": 8.74006995520954e-06, "loss": 0.0025, "step": 81850 }, { "grad_norm": 0.08409978449344635, "learning_rate": 8.730732707977397e-06, "loss": 0.003, "step": 81860 }, { "grad_norm": 0.03838121145963669, "learning_rate": 8.721399973890281e-06, "loss": 0.0038, "step": 81870 }, { "grad_norm": 0.03812520578503609, "learning_rate": 8.712071753968792e-06, "loss": 0.0025, "step": 81880 }, { "grad_norm": 0.040652237832546234, "learning_rate": 8.702748049233072e-06, "loss": 0.0021, "step": 81890 }, { "grad_norm": 0.07859864085912704, "learning_rate": 8.693428860702718e-06, "loss": 0.0036, "step": 81900 }, { "grad_norm": 0.08062434941530228, "learning_rate": 8.684114189396898e-06, "loss": 0.0028, "step": 81910 }, { "grad_norm": 0.055654071271419525, "learning_rate": 8.674804036334222e-06, "loss": 0.0026, "step": 81920 }, { "grad_norm": 0.04095393046736717, "learning_rate": 8.665498402532857e-06, "loss": 0.003, "step": 81930 }, { "grad_norm": 0.030164213851094246, "learning_rate": 8.656197289010432e-06, "loss": 0.0041, "step": 81940 }, { "grad_norm": 0.04713903367519379, "learning_rate": 8.646900696784122e-06, "loss": 0.0047, "step": 81950 }, { "grad_norm": 0.04345841333270073, "learning_rate": 8.637608626870553e-06, "loss": 0.0023, "step": 81960 }, { "grad_norm": 0.023100297898054123, "learning_rate": 8.628321080285944e-06, "loss": 0.0025, "step": 81970 }, { "grad_norm": 0.05976081266999245, "learning_rate": 8.619038058045925e-06, "loss": 0.0037, "step": 81980 }, { "grad_norm": 0.02946259081363678, "learning_rate": 8.609759561165704e-06, "loss": 0.0048, "step": 81990 }, { "grad_norm": 0.03009764291346073, "learning_rate": 8.600485590659934e-06, "loss": 0.0026, "step": 82000 }, { "grad_norm": 0.039806146174669266, "learning_rate": 8.591216147542813e-06, "loss": 0.0028, "step": 82010 }, { "grad_norm": 0.03481607884168625, "learning_rate": 8.581951232828045e-06, "loss": 0.0019, "step": 82020 }, { "grad_norm": 0.0550541989505291, "learning_rate": 8.572690847528825e-06, "loss": 0.0032, "step": 82030 }, { "grad_norm": 0.052797216922044754, "learning_rate": 8.563434992657838e-06, "loss": 0.0035, "step": 82040 }, { "grad_norm": 0.024083003401756287, "learning_rate": 8.554183669227307e-06, "loss": 0.0033, "step": 82050 }, { "grad_norm": 0.09896209090948105, "learning_rate": 8.544936878248937e-06, "loss": 0.0028, "step": 82060 }, { "grad_norm": 0.038504403084516525, "learning_rate": 8.535694620733942e-06, "loss": 0.0032, "step": 82070 }, { "grad_norm": 0.04663178324699402, "learning_rate": 8.526456897693053e-06, "loss": 0.002, "step": 82080 }, { "grad_norm": 0.03329822048544884, "learning_rate": 8.517223710136456e-06, "loss": 0.0021, "step": 82090 }, { "grad_norm": 0.04287361726164818, "learning_rate": 8.507995059073931e-06, "loss": 0.0033, "step": 82100 }, { "grad_norm": 0.04859457537531853, "learning_rate": 8.498770945514673e-06, "loss": 0.0035, "step": 82110 }, { "grad_norm": 0.03840424865484238, "learning_rate": 8.489551370467436e-06, "loss": 0.0022, "step": 82120 }, { "grad_norm": 0.04164150357246399, "learning_rate": 8.480336334940442e-06, "loss": 0.0022, "step": 82130 }, { "grad_norm": 0.0314461849629879, "learning_rate": 8.471125839941458e-06, "loss": 0.0021, "step": 82140 }, { "grad_norm": 0.05128394439816475, "learning_rate": 8.461919886477682e-06, "loss": 0.0021, "step": 82150 }, { "grad_norm": 0.048505865037441254, "learning_rate": 8.452718475555927e-06, "loss": 0.0029, "step": 82160 }, { "grad_norm": 0.028033453971147537, "learning_rate": 8.443521608182398e-06, "loss": 0.0027, "step": 82170 }, { "grad_norm": 0.0684376209974289, "learning_rate": 8.43432928536288e-06, "loss": 0.0024, "step": 82180 }, { "grad_norm": 0.060867663472890854, "learning_rate": 8.425141508102607e-06, "loss": 0.0022, "step": 82190 }, { "grad_norm": 0.024632280692458153, "learning_rate": 8.415958277406365e-06, "loss": 0.0025, "step": 82200 }, { "grad_norm": 0.08364914357662201, "learning_rate": 8.406779594278392e-06, "loss": 0.0029, "step": 82210 }, { "grad_norm": 0.06925616413354874, "learning_rate": 8.397605459722468e-06, "loss": 0.0024, "step": 82220 }, { "grad_norm": 0.09655017405748367, "learning_rate": 8.388435874741863e-06, "loss": 0.0028, "step": 82230 }, { "grad_norm": 0.03102811798453331, "learning_rate": 8.37927084033936e-06, "loss": 0.0066, "step": 82240 }, { "grad_norm": 0.030737362802028656, "learning_rate": 8.370110357517208e-06, "loss": 0.0018, "step": 82250 }, { "grad_norm": 0.054280735552310944, "learning_rate": 8.360954427277201e-06, "loss": 0.0024, "step": 82260 }, { "grad_norm": 0.037219855934381485, "learning_rate": 8.351803050620627e-06, "loss": 0.0022, "step": 82270 }, { "grad_norm": 0.032388459891080856, "learning_rate": 8.34265622854824e-06, "loss": 0.0032, "step": 82280 }, { "grad_norm": 0.06687033176422119, "learning_rate": 8.333513962060335e-06, "loss": 0.0026, "step": 82290 }, { "grad_norm": 0.038914624601602554, "learning_rate": 8.324376252156701e-06, "loss": 0.0043, "step": 82300 }, { "grad_norm": 0.02818896435201168, "learning_rate": 8.315243099836633e-06, "loss": 0.0026, "step": 82310 }, { "grad_norm": 0.05466456338763237, "learning_rate": 8.3061145060989e-06, "loss": 0.0026, "step": 82320 }, { "grad_norm": 0.045324213802814484, "learning_rate": 8.296990471941806e-06, "loss": 0.0025, "step": 82330 }, { "grad_norm": 0.09206051379442215, "learning_rate": 8.287870998363112e-06, "loss": 0.0054, "step": 82340 }, { "grad_norm": 0.03913034871220589, "learning_rate": 8.278756086360156e-06, "loss": 0.0024, "step": 82350 }, { "grad_norm": 0.06313757598400116, "learning_rate": 8.269645736929693e-06, "loss": 0.0024, "step": 82360 }, { "grad_norm": 0.03452639654278755, "learning_rate": 8.260539951068048e-06, "loss": 0.002, "step": 82370 }, { "grad_norm": 0.0655386671423912, "learning_rate": 8.251438729770983e-06, "loss": 0.0025, "step": 82380 }, { "grad_norm": 0.05936753377318382, "learning_rate": 8.242342074033826e-06, "loss": 0.0036, "step": 82390 }, { "grad_norm": 0.05102138593792915, "learning_rate": 8.233249984851344e-06, "loss": 0.0047, "step": 82400 }, { "grad_norm": 0.06959748268127441, "learning_rate": 8.224162463217849e-06, "loss": 0.0043, "step": 82410 }, { "grad_norm": 0.04747885465621948, "learning_rate": 8.215079510127139e-06, "loss": 0.0033, "step": 82420 }, { "grad_norm": 0.026192808523774147, "learning_rate": 8.206001126572521e-06, "loss": 0.0054, "step": 82430 }, { "grad_norm": 0.03748996928334236, "learning_rate": 8.196927313546776e-06, "loss": 0.0059, "step": 82440 }, { "grad_norm": 0.05419636890292168, "learning_rate": 8.18785807204222e-06, "loss": 0.0024, "step": 82450 }, { "grad_norm": 0.030253862962126732, "learning_rate": 8.178793403050627e-06, "loss": 0.0023, "step": 82460 }, { "grad_norm": 0.06884995102882385, "learning_rate": 8.16973330756331e-06, "loss": 0.0029, "step": 82470 }, { "grad_norm": 0.053737033158540726, "learning_rate": 8.16067778657107e-06, "loss": 0.0026, "step": 82480 }, { "grad_norm": 0.05318194255232811, "learning_rate": 8.151626841064214e-06, "loss": 0.0031, "step": 82490 }, { "grad_norm": 0.05125297233462334, "learning_rate": 8.142580472032518e-06, "loss": 0.004, "step": 82500 }, { "grad_norm": 0.060049619525671005, "learning_rate": 8.133538680465285e-06, "loss": 0.0021, "step": 82510 }, { "grad_norm": 0.04291774705052376, "learning_rate": 8.124501467351336e-06, "loss": 0.0022, "step": 82520 }, { "grad_norm": 0.03516384959220886, "learning_rate": 8.11546883367893e-06, "loss": 0.003, "step": 82530 }, { "grad_norm": 0.06565454602241516, "learning_rate": 8.106440780435882e-06, "loss": 0.0023, "step": 82540 }, { "grad_norm": 0.04238080978393555, "learning_rate": 8.097417308609485e-06, "loss": 0.002, "step": 82550 }, { "grad_norm": 0.03371065482497215, "learning_rate": 8.088398419186544e-06, "loss": 0.0019, "step": 82560 }, { "grad_norm": 0.02294495515525341, "learning_rate": 8.079384113153321e-06, "loss": 0.0027, "step": 82570 }, { "grad_norm": 0.029071081429719925, "learning_rate": 8.070374391495644e-06, "loss": 0.0022, "step": 82580 }, { "grad_norm": 0.061372507363557816, "learning_rate": 8.061369255198764e-06, "loss": 0.0034, "step": 82590 }, { "grad_norm": 0.08183473348617554, "learning_rate": 8.052368705247493e-06, "loss": 0.0021, "step": 82600 }, { "grad_norm": 0.047178253531455994, "learning_rate": 8.043372742626114e-06, "loss": 0.0025, "step": 82610 }, { "grad_norm": 0.052848558872938156, "learning_rate": 8.034381368318422e-06, "loss": 0.0021, "step": 82620 }, { "grad_norm": 0.040206898003816605, "learning_rate": 8.025394583307677e-06, "loss": 0.0022, "step": 82630 }, { "grad_norm": 0.10273769497871399, "learning_rate": 8.016412388576683e-06, "loss": 0.0024, "step": 82640 }, { "grad_norm": 0.06944314390420914, "learning_rate": 8.007434785107699e-06, "loss": 0.0033, "step": 82650 }, { "grad_norm": 0.04920537769794464, "learning_rate": 7.998461773882509e-06, "loss": 0.002, "step": 82660 }, { "grad_norm": 0.0699719786643982, "learning_rate": 7.989493355882393e-06, "loss": 0.0025, "step": 82670 }, { "grad_norm": 0.04313506558537483, "learning_rate": 7.980529532088137e-06, "loss": 0.0023, "step": 82680 }, { "grad_norm": 0.044831834733486176, "learning_rate": 7.97157030347998e-06, "loss": 0.0023, "step": 82690 }, { "grad_norm": 0.041024018079042435, "learning_rate": 7.96261567103772e-06, "loss": 0.003, "step": 82700 }, { "grad_norm": 0.027393104508519173, "learning_rate": 7.953665635740593e-06, "loss": 0.003, "step": 82710 }, { "grad_norm": 0.040570542216300964, "learning_rate": 7.94472019856738e-06, "loss": 0.0025, "step": 82720 }, { "grad_norm": 0.044742651283741, "learning_rate": 7.935779360496337e-06, "loss": 0.0047, "step": 82730 }, { "grad_norm": 0.06179158017039299, "learning_rate": 7.92684312250523e-06, "loss": 0.0024, "step": 82740 }, { "grad_norm": 0.02825240232050419, "learning_rate": 7.917911485571289e-06, "loss": 0.0019, "step": 82750 }, { "grad_norm": 0.062372688204050064, "learning_rate": 7.90898445067128e-06, "loss": 0.0044, "step": 82760 }, { "grad_norm": 0.04401911422610283, "learning_rate": 7.900062018781462e-06, "loss": 0.0027, "step": 82770 }, { "grad_norm": 0.04274768754839897, "learning_rate": 7.891144190877553e-06, "loss": 0.0022, "step": 82780 }, { "grad_norm": 0.05801568925380707, "learning_rate": 7.882230967934806e-06, "loss": 0.0037, "step": 82790 }, { "grad_norm": 0.034949611872434616, "learning_rate": 7.873322350927959e-06, "loss": 0.0022, "step": 82800 }, { "grad_norm": 0.039741188287734985, "learning_rate": 7.864418340831253e-06, "loss": 0.0035, "step": 82810 }, { "grad_norm": 0.05423169955611229, "learning_rate": 7.855518938618395e-06, "loss": 0.0019, "step": 82820 }, { "grad_norm": 0.07780410349369049, "learning_rate": 7.846624145262633e-06, "loss": 0.0032, "step": 82830 }, { "grad_norm": 0.034578174352645874, "learning_rate": 7.837733961736672e-06, "loss": 0.0022, "step": 82840 }, { "grad_norm": 0.04700297862291336, "learning_rate": 7.828848389012727e-06, "loss": 0.0025, "step": 82850 }, { "grad_norm": 0.037541087716817856, "learning_rate": 7.819967428062524e-06, "loss": 0.0021, "step": 82860 }, { "grad_norm": 0.05371968448162079, "learning_rate": 7.811091079857274e-06, "loss": 0.0026, "step": 82870 }, { "grad_norm": 0.026058442890644073, "learning_rate": 7.802219345367667e-06, "loss": 0.0032, "step": 82880 }, { "grad_norm": 0.04958212748169899, "learning_rate": 7.793352225563916e-06, "loss": 0.0016, "step": 82890 }, { "grad_norm": 0.03646925464272499, "learning_rate": 7.7844897214157e-06, "loss": 0.0027, "step": 82900 }, { "grad_norm": 0.038702670484781265, "learning_rate": 7.775631833892228e-06, "loss": 0.0026, "step": 82910 }, { "grad_norm": 0.05811313912272453, "learning_rate": 7.76677856396215e-06, "loss": 0.0037, "step": 82920 }, { "grad_norm": 0.03948419913649559, "learning_rate": 7.757929912593697e-06, "loss": 0.0028, "step": 82930 }, { "grad_norm": 0.03796429932117462, "learning_rate": 7.749085880754509e-06, "loss": 0.0025, "step": 82940 }, { "grad_norm": 0.03247002512216568, "learning_rate": 7.740246469411777e-06, "loss": 0.0014, "step": 82950 }, { "grad_norm": 0.02031794935464859, "learning_rate": 7.731411679532142e-06, "loss": 0.0027, "step": 82960 }, { "grad_norm": 0.052708785980939865, "learning_rate": 7.722581512081779e-06, "loss": 0.0037, "step": 82970 }, { "grad_norm": 0.038181111216545105, "learning_rate": 7.713755968026331e-06, "loss": 0.003, "step": 82980 }, { "grad_norm": 0.05053868144750595, "learning_rate": 7.704935048330969e-06, "loss": 0.0026, "step": 82990 }, { "grad_norm": 0.1054021492600441, "learning_rate": 7.696118753960313e-06, "loss": 0.0025, "step": 83000 }, { "grad_norm": 0.039732035249471664, "learning_rate": 7.687307085878503e-06, "loss": 0.0018, "step": 83010 }, { "grad_norm": 0.0577780082821846, "learning_rate": 7.678500045049186e-06, "loss": 0.0024, "step": 83020 }, { "grad_norm": 0.035197626799345016, "learning_rate": 7.669697632435463e-06, "loss": 0.0026, "step": 83030 }, { "grad_norm": 0.045052386820316315, "learning_rate": 7.660899848999963e-06, "loss": 0.0045, "step": 83040 }, { "grad_norm": 0.037301450967788696, "learning_rate": 7.652106695704803e-06, "loss": 0.0033, "step": 83050 }, { "grad_norm": 0.041238293051719666, "learning_rate": 7.643318173511598e-06, "loss": 0.0033, "step": 83060 }, { "grad_norm": 0.08092731237411499, "learning_rate": 7.634534283381423e-06, "loss": 0.002, "step": 83070 }, { "grad_norm": 0.06071290001273155, "learning_rate": 7.625755026274889e-06, "loss": 0.0023, "step": 83080 }, { "grad_norm": 0.04491659998893738, "learning_rate": 7.616980403152069e-06, "loss": 0.0024, "step": 83090 }, { "grad_norm": 0.02906949259340763, "learning_rate": 7.608210414972561e-06, "loss": 0.0047, "step": 83100 }, { "grad_norm": 0.10402695834636688, "learning_rate": 7.599445062695404e-06, "loss": 0.0035, "step": 83110 }, { "grad_norm": 0.03765382245182991, "learning_rate": 7.590684347279214e-06, "loss": 0.0042, "step": 83120 }, { "grad_norm": 0.02236860990524292, "learning_rate": 7.581928269682004e-06, "loss": 0.0035, "step": 83130 }, { "grad_norm": 0.05094509199261665, "learning_rate": 7.573176830861361e-06, "loss": 0.0025, "step": 83140 }, { "grad_norm": 0.04002302139997482, "learning_rate": 7.5644300317743e-06, "loss": 0.0022, "step": 83150 }, { "grad_norm": 0.06819801032543182, "learning_rate": 7.555687873377376e-06, "loss": 0.0031, "step": 83160 }, { "grad_norm": 0.06446180492639542, "learning_rate": 7.546950356626592e-06, "loss": 0.0041, "step": 83170 }, { "grad_norm": 0.05019285902380943, "learning_rate": 7.538217482477516e-06, "loss": 0.0022, "step": 83180 }, { "grad_norm": 0.0803980752825737, "learning_rate": 7.529489251885119e-06, "loss": 0.0048, "step": 83190 }, { "grad_norm": 0.06622587889432907, "learning_rate": 7.520765665803941e-06, "loss": 0.0019, "step": 83200 }, { "grad_norm": 0.07533016055822372, "learning_rate": 7.512046725187949e-06, "loss": 0.002, "step": 83210 }, { "grad_norm": 0.0757993832230568, "learning_rate": 7.503332430990645e-06, "loss": 0.0022, "step": 83220 }, { "grad_norm": 0.05852792412042618, "learning_rate": 7.494622784165017e-06, "loss": 0.0026, "step": 83230 }, { "grad_norm": 0.04961281269788742, "learning_rate": 7.485917785663543e-06, "loss": 0.0053, "step": 83240 }, { "grad_norm": 0.043285563588142395, "learning_rate": 7.477217436438172e-06, "loss": 0.0033, "step": 83250 }, { "grad_norm": 0.06562795490026474, "learning_rate": 7.46852173744037e-06, "loss": 0.0041, "step": 83260 }, { "grad_norm": 0.05907558277249336, "learning_rate": 7.4598306896211004e-06, "loss": 0.0045, "step": 83270 }, { "grad_norm": 0.037048015743494034, "learning_rate": 7.451144293930773e-06, "loss": 0.002, "step": 83280 }, { "grad_norm": 0.027760475873947144, "learning_rate": 7.4424625513193455e-06, "loss": 0.0019, "step": 83290 }, { "grad_norm": 0.038433220237493515, "learning_rate": 7.433785462736209e-06, "loss": 0.0015, "step": 83300 }, { "grad_norm": 0.044843513518571854, "learning_rate": 7.425113029130315e-06, "loss": 0.0031, "step": 83310 }, { "grad_norm": 0.046470317989587784, "learning_rate": 7.416445251450044e-06, "loss": 0.0038, "step": 83320 }, { "grad_norm": 0.05134887993335724, "learning_rate": 7.4077821306433035e-06, "loss": 0.0032, "step": 83330 }, { "grad_norm": 0.045851971954107285, "learning_rate": 7.399123667657465e-06, "loss": 0.0036, "step": 83340 }, { "grad_norm": 0.03072824887931347, "learning_rate": 7.390469863439426e-06, "loss": 0.0048, "step": 83350 }, { "grad_norm": 0.03038354404270649, "learning_rate": 7.381820718935512e-06, "loss": 0.0015, "step": 83360 }, { "grad_norm": 0.04104151949286461, "learning_rate": 7.373176235091633e-06, "loss": 0.0019, "step": 83370 }, { "grad_norm": 0.02586931549012661, "learning_rate": 7.364536412853101e-06, "loss": 0.0037, "step": 83380 }, { "grad_norm": 0.02922407165169716, "learning_rate": 7.355901253164776e-06, "loss": 0.0021, "step": 83390 }, { "grad_norm": 0.03921383246779442, "learning_rate": 7.34727075697097e-06, "loss": 0.0031, "step": 83400 }, { "grad_norm": 0.045107774436473846, "learning_rate": 7.3386449252155095e-06, "loss": 0.0032, "step": 83410 }, { "grad_norm": 0.02395659312605858, "learning_rate": 7.330023758841692e-06, "loss": 0.0025, "step": 83420 }, { "grad_norm": 0.0386793315410614, "learning_rate": 7.321407258792323e-06, "loss": 0.0028, "step": 83430 }, { "grad_norm": 0.023449860513210297, "learning_rate": 7.312795426009694e-06, "loss": 0.0031, "step": 83440 }, { "grad_norm": 0.07338808476924896, "learning_rate": 7.304188261435585e-06, "loss": 0.003, "step": 83450 }, { "grad_norm": 0.02065438963472843, "learning_rate": 7.295585766011242e-06, "loss": 0.0028, "step": 83460 }, { "grad_norm": 0.05412602797150612, "learning_rate": 7.286987940677436e-06, "loss": 0.003, "step": 83470 }, { "grad_norm": 0.06819306313991547, "learning_rate": 7.278394786374426e-06, "loss": 0.0026, "step": 83480 }, { "grad_norm": 0.02307308092713356, "learning_rate": 7.269806304041915e-06, "loss": 0.0026, "step": 83490 }, { "grad_norm": 0.047444701194763184, "learning_rate": 7.261222494619147e-06, "loss": 0.0033, "step": 83500 }, { "grad_norm": 0.02843196503818035, "learning_rate": 7.252643359044825e-06, "loss": 0.0029, "step": 83510 }, { "grad_norm": 0.05438637360930443, "learning_rate": 7.244068898257172e-06, "loss": 0.0032, "step": 83520 }, { "grad_norm": 0.04748024046421051, "learning_rate": 7.2354991131938495e-06, "loss": 0.0027, "step": 83530 }, { "grad_norm": 0.028061147779226303, "learning_rate": 7.2269340047920574e-06, "loss": 0.0022, "step": 83540 }, { "grad_norm": 0.016242729499936104, "learning_rate": 7.218373573988435e-06, "loss": 0.0032, "step": 83550 }, { "grad_norm": 0.03708527609705925, "learning_rate": 7.209817821719178e-06, "loss": 0.0051, "step": 83560 }, { "grad_norm": 0.019799523055553436, "learning_rate": 7.201266748919894e-06, "loss": 0.0025, "step": 83570 }, { "grad_norm": 0.034627411514520645, "learning_rate": 7.19272035652575e-06, "loss": 0.0019, "step": 83580 }, { "grad_norm": 0.03389899060130119, "learning_rate": 7.184178645471334e-06, "loss": 0.0037, "step": 83590 }, { "grad_norm": 0.04732375964522362, "learning_rate": 7.175641616690775e-06, "loss": 0.0018, "step": 83600 }, { "grad_norm": 0.06523969769477844, "learning_rate": 7.167109271117661e-06, "loss": 0.0056, "step": 83610 }, { "grad_norm": 0.0585736483335495, "learning_rate": 7.158581609685072e-06, "loss": 0.0055, "step": 83620 }, { "grad_norm": 0.05640757083892822, "learning_rate": 7.150058633325591e-06, "loss": 0.002, "step": 83630 }, { "grad_norm": 0.02944653108716011, "learning_rate": 7.141540342971287e-06, "loss": 0.003, "step": 83640 }, { "grad_norm": 0.03158555179834366, "learning_rate": 7.133026739553677e-06, "loss": 0.0022, "step": 83650 }, { "grad_norm": 0.08872894197702408, "learning_rate": 7.124517824003829e-06, "loss": 0.0068, "step": 83660 }, { "grad_norm": 0.02817728742957115, "learning_rate": 7.116013597252236e-06, "loss": 0.0024, "step": 83670 }, { "grad_norm": 0.02310970425605774, "learning_rate": 7.107514060228921e-06, "loss": 0.004, "step": 83680 }, { "grad_norm": 0.027409419417381287, "learning_rate": 7.099019213863378e-06, "loss": 0.0036, "step": 83690 }, { "grad_norm": 0.04284220561385155, "learning_rate": 7.090529059084611e-06, "loss": 0.0029, "step": 83700 }, { "grad_norm": 0.06297098845243454, "learning_rate": 7.082043596821058e-06, "loss": 0.0024, "step": 83710 }, { "grad_norm": 0.038723137229681015, "learning_rate": 7.07356282800069e-06, "loss": 0.0036, "step": 83720 }, { "grad_norm": 0.06051250174641609, "learning_rate": 7.0650867535509645e-06, "loss": 0.0035, "step": 83730 }, { "grad_norm": 0.021052049472928047, "learning_rate": 7.0566153743987895e-06, "loss": 0.0023, "step": 83740 }, { "grad_norm": 0.05369960889220238, "learning_rate": 7.04814869147059e-06, "loss": 0.0033, "step": 83750 }, { "grad_norm": 0.0306012574583292, "learning_rate": 7.0396867056922725e-06, "loss": 0.0022, "step": 83760 }, { "grad_norm": 0.03834167867898941, "learning_rate": 7.031229417989244e-06, "loss": 0.0019, "step": 83770 }, { "grad_norm": 0.040018800646066666, "learning_rate": 7.022776829286348e-06, "loss": 0.0025, "step": 83780 }, { "grad_norm": 0.06000189110636711, "learning_rate": 7.014328940507975e-06, "loss": 0.0029, "step": 83790 }, { "grad_norm": 0.02579854428768158, "learning_rate": 7.005885752577945e-06, "loss": 0.002, "step": 83800 }, { "grad_norm": 0.03485289216041565, "learning_rate": 6.997447266419615e-06, "loss": 0.0023, "step": 83810 }, { "grad_norm": 0.03698756545782089, "learning_rate": 6.989013482955797e-06, "loss": 0.0018, "step": 83820 }, { "grad_norm": 0.027299687266349792, "learning_rate": 6.980584403108808e-06, "loss": 0.0054, "step": 83830 }, { "grad_norm": 0.0526290237903595, "learning_rate": 6.972160027800417e-06, "loss": 0.0024, "step": 83840 }, { "grad_norm": 0.046612296253442764, "learning_rate": 6.963740357951931e-06, "loss": 0.0017, "step": 83850 }, { "grad_norm": 0.06768908351659775, "learning_rate": 6.95532539448408e-06, "loss": 0.0023, "step": 83860 }, { "grad_norm": 0.046100202947854996, "learning_rate": 6.946915138317128e-06, "loss": 0.0027, "step": 83870 }, { "grad_norm": 0.021106144413352013, "learning_rate": 6.9385095903708065e-06, "loss": 0.0028, "step": 83880 }, { "grad_norm": 0.03442111238837242, "learning_rate": 6.930108751564346e-06, "loss": 0.004, "step": 83890 }, { "grad_norm": 0.05467665567994118, "learning_rate": 6.921712622816428e-06, "loss": 0.0032, "step": 83900 }, { "grad_norm": 0.09797613322734833, "learning_rate": 6.913321205045259e-06, "loss": 0.0024, "step": 83910 }, { "grad_norm": 0.032774657011032104, "learning_rate": 6.904934499168491e-06, "loss": 0.0017, "step": 83920 }, { "grad_norm": 0.030045446008443832, "learning_rate": 6.896552506103299e-06, "loss": 0.0026, "step": 83930 }, { "grad_norm": 0.021816417574882507, "learning_rate": 6.888175226766313e-06, "loss": 0.0028, "step": 83940 }, { "grad_norm": 0.035206131637096405, "learning_rate": 6.879802662073686e-06, "loss": 0.0031, "step": 83950 }, { "grad_norm": 0.05823385715484619, "learning_rate": 6.871434812940991e-06, "loss": 0.0029, "step": 83960 }, { "grad_norm": 0.09921819716691971, "learning_rate": 6.863071680283345e-06, "loss": 0.0031, "step": 83970 }, { "grad_norm": 0.05212799087166786, "learning_rate": 6.854713265015339e-06, "loss": 0.002, "step": 83980 }, { "grad_norm": 0.07663991302251816, "learning_rate": 6.8463595680510025e-06, "loss": 0.0037, "step": 83990 }, { "grad_norm": 0.032376762479543686, "learning_rate": 6.838010590303911e-06, "loss": 0.0067, "step": 84000 }, { "grad_norm": 0.0513005256652832, "learning_rate": 6.829666332687085e-06, "loss": 0.002, "step": 84010 }, { "grad_norm": 0.06820821762084961, "learning_rate": 6.821326796113059e-06, "loss": 0.0027, "step": 84020 }, { "grad_norm": 0.041936103254556656, "learning_rate": 6.812991981493799e-06, "loss": 0.0065, "step": 84030 }, { "grad_norm": 0.03325774148106575, "learning_rate": 6.804661889740821e-06, "loss": 0.0041, "step": 84040 }, { "grad_norm": 0.035618044435977936, "learning_rate": 6.7963365217650625e-06, "loss": 0.0036, "step": 84050 }, { "grad_norm": 0.02427474595606327, "learning_rate": 6.788015878476983e-06, "loss": 0.0033, "step": 84060 }, { "grad_norm": 0.033404264599084854, "learning_rate": 6.779699960786523e-06, "loss": 0.0023, "step": 84070 }, { "grad_norm": 0.014419394545257092, "learning_rate": 6.771388769603104e-06, "loss": 0.0021, "step": 84080 }, { "grad_norm": 0.020671887323260307, "learning_rate": 6.763082305835605e-06, "loss": 0.0026, "step": 84090 }, { "grad_norm": 0.03242640942335129, "learning_rate": 6.754780570392433e-06, "loss": 0.0023, "step": 84100 }, { "grad_norm": 0.026614470407366753, "learning_rate": 6.7464835641814215e-06, "loss": 0.0026, "step": 84110 }, { "grad_norm": 0.04264674708247185, "learning_rate": 6.7381912881099515e-06, "loss": 0.0028, "step": 84120 }, { "grad_norm": 0.058043722063302994, "learning_rate": 6.72990374308482e-06, "loss": 0.0022, "step": 84130 }, { "grad_norm": 0.0369160920381546, "learning_rate": 6.721620930012379e-06, "loss": 0.0032, "step": 84140 }, { "grad_norm": 0.04708629474043846, "learning_rate": 6.713342849798393e-06, "loss": 0.0022, "step": 84150 }, { "grad_norm": 0.048715461045503616, "learning_rate": 6.705069503348166e-06, "loss": 0.002, "step": 84160 }, { "grad_norm": 0.03268716111779213, "learning_rate": 6.696800891566435e-06, "loss": 0.0028, "step": 84170 }, { "grad_norm": 0.035676129162311554, "learning_rate": 6.688537015357449e-06, "loss": 0.0045, "step": 84180 }, { "grad_norm": 0.06294553726911545, "learning_rate": 6.680277875624941e-06, "loss": 0.0026, "step": 84190 }, { "grad_norm": 0.07509592175483704, "learning_rate": 6.672023473272121e-06, "loss": 0.0039, "step": 84200 }, { "grad_norm": 0.043482620269060135, "learning_rate": 6.663773809201667e-06, "loss": 0.0032, "step": 84210 }, { "grad_norm": 0.047289036214351654, "learning_rate": 6.655528884315748e-06, "loss": 0.0024, "step": 84220 }, { "grad_norm": 0.07031846046447754, "learning_rate": 6.647288699516036e-06, "loss": 0.0031, "step": 84230 }, { "grad_norm": 0.06984228640794754, "learning_rate": 6.639053255703642e-06, "loss": 0.0046, "step": 84240 }, { "grad_norm": 0.04105360060930252, "learning_rate": 6.6308225537791925e-06, "loss": 0.0031, "step": 84250 }, { "grad_norm": 0.031582094728946686, "learning_rate": 6.622596594642777e-06, "loss": 0.0021, "step": 84260 }, { "grad_norm": 0.06602924317121506, "learning_rate": 6.614375379193999e-06, "loss": 0.0028, "step": 84270 }, { "grad_norm": 0.03776537626981735, "learning_rate": 6.606158908331889e-06, "loss": 0.0028, "step": 84280 }, { "grad_norm": 0.05128682032227516, "learning_rate": 6.597947182955005e-06, "loss": 0.0033, "step": 84290 }, { "grad_norm": 0.05166797712445259, "learning_rate": 6.589740203961359e-06, "loss": 0.0023, "step": 84300 }, { "grad_norm": 0.04676367715001106, "learning_rate": 6.581537972248464e-06, "loss": 0.0035, "step": 84310 }, { "grad_norm": 0.05106736719608307, "learning_rate": 6.573340488713275e-06, "loss": 0.0029, "step": 84320 }, { "grad_norm": 0.03467559069395065, "learning_rate": 6.565147754252304e-06, "loss": 0.0027, "step": 84330 }, { "grad_norm": 0.03202580288052559, "learning_rate": 6.556959769761462e-06, "loss": 0.0026, "step": 84340 }, { "grad_norm": 0.01838076487183571, "learning_rate": 6.548776536136192e-06, "loss": 0.0045, "step": 84350 }, { "grad_norm": 0.021120503544807434, "learning_rate": 6.54059805427138e-06, "loss": 0.0033, "step": 84360 }, { "grad_norm": 0.02725311741232872, "learning_rate": 6.532424325061437e-06, "loss": 0.0027, "step": 84370 }, { "grad_norm": 0.06662454456090927, "learning_rate": 6.5242553494001935e-06, "loss": 0.0045, "step": 84380 }, { "grad_norm": 0.09190945327281952, "learning_rate": 6.516091128181046e-06, "loss": 0.0018, "step": 84390 }, { "grad_norm": 0.061199579387903214, "learning_rate": 6.507931662296785e-06, "loss": 0.0021, "step": 84400 }, { "grad_norm": 0.07972943037748337, "learning_rate": 6.4997769526397354e-06, "loss": 0.0029, "step": 84410 }, { "grad_norm": 0.04284339025616646, "learning_rate": 6.491627000101663e-06, "loss": 0.0023, "step": 84420 }, { "grad_norm": 0.039717551320791245, "learning_rate": 6.4834818055738545e-06, "loss": 0.0018, "step": 84430 }, { "grad_norm": 0.03469003364443779, "learning_rate": 6.4753413699470465e-06, "loss": 0.0027, "step": 84440 }, { "grad_norm": 0.038711968809366226, "learning_rate": 6.467205694111478e-06, "loss": 0.0024, "step": 84450 }, { "grad_norm": 0.05402608960866928, "learning_rate": 6.459074778956836e-06, "loss": 0.0026, "step": 84460 }, { "grad_norm": 0.0747544914484024, "learning_rate": 6.450948625372311e-06, "loss": 0.0056, "step": 84470 }, { "grad_norm": 0.02995450235903263, "learning_rate": 6.442827234246585e-06, "loss": 0.0019, "step": 84480 }, { "grad_norm": 0.05904434621334076, "learning_rate": 6.43471060646777e-06, "loss": 0.0035, "step": 84490 }, { "grad_norm": 0.0317017138004303, "learning_rate": 6.426598742923512e-06, "loss": 0.0024, "step": 84500 }, { "grad_norm": 0.035764798521995544, "learning_rate": 6.418491644500885e-06, "loss": 0.0022, "step": 84510 }, { "grad_norm": 0.022977514192461967, "learning_rate": 6.410389312086512e-06, "loss": 0.0056, "step": 84520 }, { "grad_norm": 0.026474496349692345, "learning_rate": 6.4022917465664135e-06, "loss": 0.003, "step": 84530 }, { "grad_norm": 0.05781778693199158, "learning_rate": 6.3941989488261575e-06, "loss": 0.0034, "step": 84540 }, { "grad_norm": 0.05723346769809723, "learning_rate": 6.386110919750726e-06, "loss": 0.0015, "step": 84550 }, { "grad_norm": 0.04257884994149208, "learning_rate": 6.378027660224651e-06, "loss": 0.0032, "step": 84560 }, { "grad_norm": 0.04970595985651016, "learning_rate": 6.369949171131856e-06, "loss": 0.0026, "step": 84570 }, { "grad_norm": 0.048510074615478516, "learning_rate": 6.361875453355848e-06, "loss": 0.0018, "step": 84580 }, { "grad_norm": 0.03772909566760063, "learning_rate": 6.353806507779525e-06, "loss": 0.0037, "step": 84590 }, { "grad_norm": 0.04744681715965271, "learning_rate": 6.3457423352853044e-06, "loss": 0.0029, "step": 84600 }, { "grad_norm": 0.03371400386095047, "learning_rate": 6.337682936755062e-06, "loss": 0.0023, "step": 84610 }, { "grad_norm": 0.028477605432271957, "learning_rate": 6.329628313070179e-06, "loss": 0.0019, "step": 84620 }, { "grad_norm": 0.09023994952440262, "learning_rate": 6.321578465111477e-06, "loss": 0.0024, "step": 84630 }, { "grad_norm": 0.07103472948074341, "learning_rate": 6.31353339375928e-06, "loss": 0.0021, "step": 84640 }, { "grad_norm": 0.06384149938821793, "learning_rate": 6.305493099893395e-06, "loss": 0.0021, "step": 84650 }, { "grad_norm": 0.06064889580011368, "learning_rate": 6.297457584393096e-06, "loss": 0.0023, "step": 84660 }, { "grad_norm": 0.03495204076170921, "learning_rate": 6.289426848137126e-06, "loss": 0.0021, "step": 84670 }, { "grad_norm": 0.040848031640052795, "learning_rate": 6.281400892003713e-06, "loss": 0.0021, "step": 84680 }, { "grad_norm": 0.06630939990282059, "learning_rate": 6.273379716870581e-06, "loss": 0.0018, "step": 84690 }, { "grad_norm": 0.05031915009021759, "learning_rate": 6.265363323614887e-06, "loss": 0.0029, "step": 84700 }, { "grad_norm": 0.034994564950466156, "learning_rate": 6.257351713113308e-06, "loss": 0.002, "step": 84710 }, { "grad_norm": 0.06744436919689178, "learning_rate": 6.249344886241981e-06, "loss": 0.0026, "step": 84720 }, { "grad_norm": 0.028619196265935898, "learning_rate": 6.2413428438765295e-06, "loss": 0.003, "step": 84730 }, { "grad_norm": 0.052396729588508606, "learning_rate": 6.233345586892026e-06, "loss": 0.0025, "step": 84740 }, { "grad_norm": 0.04569748416543007, "learning_rate": 6.225353116163052e-06, "loss": 0.0022, "step": 84750 }, { "grad_norm": 0.0465962179005146, "learning_rate": 6.217365432563627e-06, "loss": 0.0027, "step": 84760 }, { "grad_norm": 0.056494615972042084, "learning_rate": 6.209382536967318e-06, "loss": 0.0023, "step": 84770 }, { "grad_norm": 0.028655974194407463, "learning_rate": 6.201404430247082e-06, "loss": 0.0046, "step": 84780 }, { "grad_norm": 0.03209497407078743, "learning_rate": 6.1934311132754165e-06, "loss": 0.0021, "step": 84790 }, { "grad_norm": 0.05778489634394646, "learning_rate": 6.185462586924257e-06, "loss": 0.0035, "step": 84800 }, { "grad_norm": 0.06084517017006874, "learning_rate": 6.177498852065039e-06, "loss": 0.0026, "step": 84810 }, { "grad_norm": 0.0314374677836895, "learning_rate": 6.169539909568655e-06, "loss": 0.0021, "step": 84820 }, { "grad_norm": 0.028061533346772194, "learning_rate": 6.161585760305488e-06, "loss": 0.0018, "step": 84830 }, { "grad_norm": 0.04881535843014717, "learning_rate": 6.153636405145391e-06, "loss": 0.0037, "step": 84840 }, { "grad_norm": 0.06663957983255386, "learning_rate": 6.145691844957702e-06, "loss": 0.0022, "step": 84850 }, { "grad_norm": 0.06511278450489044, "learning_rate": 6.1377520806112145e-06, "loss": 0.0028, "step": 84860 }, { "grad_norm": 0.03141718730330467, "learning_rate": 6.129817112974218e-06, "loss": 0.0039, "step": 84870 }, { "grad_norm": 0.037325698882341385, "learning_rate": 6.121886942914457e-06, "loss": 0.0041, "step": 84880 }, { "grad_norm": 0.025671010836958885, "learning_rate": 6.11396157129917e-06, "loss": 0.0019, "step": 84890 }, { "grad_norm": 0.09199358522891998, "learning_rate": 6.106040998995066e-06, "loss": 0.0028, "step": 84900 }, { "grad_norm": 0.036234572529792786, "learning_rate": 6.098125226868329e-06, "loss": 0.0035, "step": 84910 }, { "grad_norm": 0.07072620838880539, "learning_rate": 6.090214255784604e-06, "loss": 0.0049, "step": 84920 }, { "grad_norm": 0.03942515328526497, "learning_rate": 6.0823080866090346e-06, "loss": 0.0027, "step": 84930 }, { "grad_norm": 0.03757782652974129, "learning_rate": 6.074406720206233e-06, "loss": 0.0039, "step": 84940 }, { "grad_norm": 0.0552932545542717, "learning_rate": 6.066510157440259e-06, "loss": 0.0021, "step": 84950 }, { "grad_norm": 0.07351397722959518, "learning_rate": 6.058618399174676e-06, "loss": 0.002, "step": 84960 }, { "grad_norm": 0.08926333487033844, "learning_rate": 6.050731446272528e-06, "loss": 0.0028, "step": 84970 }, { "grad_norm": 0.033209506422281265, "learning_rate": 6.042849299596321e-06, "loss": 0.0027, "step": 84980 }, { "grad_norm": 0.06594928354024887, "learning_rate": 6.034971960008013e-06, "loss": 0.0041, "step": 84990 }, { "grad_norm": 0.07565520703792572, "learning_rate": 6.027099428369082e-06, "loss": 0.003, "step": 85000 }, { "grad_norm": 0.06750355660915375, "learning_rate": 6.019231705540435e-06, "loss": 0.0034, "step": 85010 }, { "grad_norm": 0.02941921167075634, "learning_rate": 6.011368792382488e-06, "loss": 0.0033, "step": 85020 }, { "grad_norm": 0.03809913247823715, "learning_rate": 6.00351068975511e-06, "loss": 0.0036, "step": 85030 }, { "grad_norm": 0.07682668417692184, "learning_rate": 5.9956573985176674e-06, "loss": 0.002, "step": 85040 }, { "grad_norm": 0.03453338146209717, "learning_rate": 5.987808919528959e-06, "loss": 0.0033, "step": 85050 }, { "grad_norm": 0.039342135190963745, "learning_rate": 5.979965253647307e-06, "loss": 0.0022, "step": 85060 }, { "grad_norm": 0.027520783245563507, "learning_rate": 5.9721264017304655e-06, "loss": 0.0025, "step": 85070 }, { "grad_norm": 0.05152871087193489, "learning_rate": 5.964292364635682e-06, "loss": 0.0036, "step": 85080 }, { "grad_norm": 0.022393668070435524, "learning_rate": 5.956463143219682e-06, "loss": 0.0036, "step": 85090 }, { "grad_norm": 0.0655655488371849, "learning_rate": 5.948638738338663e-06, "loss": 0.0037, "step": 85100 }, { "grad_norm": 0.031114473938941956, "learning_rate": 5.940819150848271e-06, "loss": 0.0024, "step": 85110 }, { "grad_norm": 0.055749449878931046, "learning_rate": 5.933004381603663e-06, "loss": 0.0024, "step": 85120 }, { "grad_norm": 0.022891778498888016, "learning_rate": 5.92519443145943e-06, "loss": 0.0043, "step": 85130 }, { "grad_norm": 0.0508512519299984, "learning_rate": 5.917389301269671e-06, "loss": 0.0024, "step": 85140 }, { "grad_norm": 0.05887577682733536, "learning_rate": 5.9095889918879435e-06, "loss": 0.0021, "step": 85150 }, { "grad_norm": 0.03303149342536926, "learning_rate": 5.901793504167286e-06, "loss": 0.0025, "step": 85160 }, { "grad_norm": 0.05351898819208145, "learning_rate": 5.894002838960172e-06, "loss": 0.0034, "step": 85170 }, { "grad_norm": 0.04388446360826492, "learning_rate": 5.886216997118604e-06, "loss": 0.0025, "step": 85180 }, { "grad_norm": 0.050526801496744156, "learning_rate": 5.878435979494023e-06, "loss": 0.0032, "step": 85190 }, { "grad_norm": 0.050328806042671204, "learning_rate": 5.870659786937344e-06, "loss": 0.0034, "step": 85200 }, { "grad_norm": 0.0803474709391594, "learning_rate": 5.862888420298962e-06, "loss": 0.0029, "step": 85210 }, { "grad_norm": 0.02520136348903179, "learning_rate": 5.855121880428743e-06, "loss": 0.0044, "step": 85220 }, { "grad_norm": 0.05214850232005119, "learning_rate": 5.84736016817603e-06, "loss": 0.0037, "step": 85230 }, { "grad_norm": 0.0314793735742569, "learning_rate": 5.839603284389622e-06, "loss": 0.0026, "step": 85240 }, { "grad_norm": 0.03171715885400772, "learning_rate": 5.831851229917812e-06, "loss": 0.0018, "step": 85250 }, { "grad_norm": 0.06482253968715668, "learning_rate": 5.824104005608338e-06, "loss": 0.0014, "step": 85260 }, { "grad_norm": 0.03488479554653168, "learning_rate": 5.81636161230843e-06, "loss": 0.0019, "step": 85270 }, { "grad_norm": 0.052963342517614365, "learning_rate": 5.808624050864791e-06, "loss": 0.0021, "step": 85280 }, { "grad_norm": 0.040521640330553055, "learning_rate": 5.800891322123592e-06, "loss": 0.0039, "step": 85290 }, { "grad_norm": 0.029893608763813972, "learning_rate": 5.793163426930454e-06, "loss": 0.0032, "step": 85300 }, { "grad_norm": 0.07120410352945328, "learning_rate": 5.785440366130512e-06, "loss": 0.0026, "step": 85310 }, { "grad_norm": 0.06885907799005508, "learning_rate": 5.777722140568326e-06, "loss": 0.0024, "step": 85320 }, { "grad_norm": 0.024808114394545555, "learning_rate": 5.7700087510879685e-06, "loss": 0.0028, "step": 85330 }, { "grad_norm": 0.055866021662950516, "learning_rate": 5.762300198532933e-06, "loss": 0.0039, "step": 85340 }, { "grad_norm": 0.04878748208284378, "learning_rate": 5.7545964837462606e-06, "loss": 0.0021, "step": 85350 }, { "grad_norm": 0.032082583755254745, "learning_rate": 5.746897607570384e-06, "loss": 0.0025, "step": 85360 }, { "grad_norm": 0.022744007408618927, "learning_rate": 5.7392035708472626e-06, "loss": 0.002, "step": 85370 }, { "grad_norm": 0.030538130551576614, "learning_rate": 5.731514374418278e-06, "loss": 0.0039, "step": 85380 }, { "grad_norm": 0.03508840873837471, "learning_rate": 5.7238300191243325e-06, "loss": 0.0022, "step": 85390 }, { "grad_norm": 0.06205281615257263, "learning_rate": 5.716150505805762e-06, "loss": 0.0032, "step": 85400 }, { "grad_norm": 0.03503032401204109, "learning_rate": 5.708475835302407e-06, "loss": 0.002, "step": 85410 }, { "grad_norm": 0.021039754152297974, "learning_rate": 5.70080600845353e-06, "loss": 0.0033, "step": 85420 }, { "grad_norm": 0.027685964480042458, "learning_rate": 5.693141026097909e-06, "loss": 0.0027, "step": 85430 }, { "grad_norm": 0.06305708736181259, "learning_rate": 5.6854808890737745e-06, "loss": 0.004, "step": 85440 }, { "grad_norm": 0.0343208871781826, "learning_rate": 5.6778255982188225e-06, "loss": 0.004, "step": 85450 }, { "grad_norm": 0.027518896386027336, "learning_rate": 5.670175154370217e-06, "loss": 0.002, "step": 85460 }, { "grad_norm": 0.04882771149277687, "learning_rate": 5.662529558364616e-06, "loss": 0.0017, "step": 85470 }, { "grad_norm": 0.03560720756649971, "learning_rate": 5.654888811038128e-06, "loss": 0.0033, "step": 85480 }, { "grad_norm": 0.028095290064811707, "learning_rate": 5.647252913226314e-06, "loss": 0.0025, "step": 85490 }, { "grad_norm": 0.06212808936834335, "learning_rate": 5.639621865764255e-06, "loss": 0.0023, "step": 85500 }, { "grad_norm": 0.028453541919589043, "learning_rate": 5.63199566948644e-06, "loss": 0.0033, "step": 85510 }, { "grad_norm": 0.03114200383424759, "learning_rate": 5.6243743252268835e-06, "loss": 0.0019, "step": 85520 }, { "grad_norm": 0.029809780418872833, "learning_rate": 5.616757833819009e-06, "loss": 0.0025, "step": 85530 }, { "grad_norm": 0.05339277163147926, "learning_rate": 5.609146196095794e-06, "loss": 0.0052, "step": 85540 }, { "grad_norm": 0.03112437203526497, "learning_rate": 5.6015394128896005e-06, "loss": 0.0052, "step": 85550 }, { "grad_norm": 0.025739751756191254, "learning_rate": 5.593937485032308e-06, "loss": 0.0023, "step": 85560 }, { "grad_norm": 0.033487822860479355, "learning_rate": 5.586340413355246e-06, "loss": 0.0045, "step": 85570 }, { "grad_norm": 0.050994228571653366, "learning_rate": 5.578748198689227e-06, "loss": 0.0017, "step": 85580 }, { "grad_norm": 0.08675755560398102, "learning_rate": 5.571160841864504e-06, "loss": 0.0025, "step": 85590 }, { "grad_norm": 0.03202984109520912, "learning_rate": 5.563578343710846e-06, "loss": 0.0026, "step": 85600 }, { "grad_norm": 0.03584592416882515, "learning_rate": 5.556000705057446e-06, "loss": 0.0027, "step": 85610 }, { "grad_norm": 0.022561367601156235, "learning_rate": 5.548427926732996e-06, "loss": 0.0018, "step": 85620 }, { "grad_norm": 0.02698501944541931, "learning_rate": 5.540860009565629e-06, "loss": 0.0016, "step": 85630 }, { "grad_norm": 0.044679105281829834, "learning_rate": 5.533296954382966e-06, "loss": 0.0026, "step": 85640 }, { "grad_norm": 0.031428851187229156, "learning_rate": 5.525738762012095e-06, "loss": 0.0031, "step": 85650 }, { "grad_norm": 0.06181950494647026, "learning_rate": 5.518185433279577e-06, "loss": 0.0019, "step": 85660 }, { "grad_norm": 0.05652617663145065, "learning_rate": 5.510636969011407e-06, "loss": 0.003, "step": 85670 }, { "grad_norm": 0.07064187526702881, "learning_rate": 5.503093370033096e-06, "loss": 0.0036, "step": 85680 }, { "grad_norm": 0.04919559881091118, "learning_rate": 5.495554637169603e-06, "loss": 0.0024, "step": 85690 }, { "grad_norm": 0.01900775544345379, "learning_rate": 5.488020771245328e-06, "loss": 0.0024, "step": 85700 }, { "grad_norm": 0.04122666269540787, "learning_rate": 5.480491773084195e-06, "loss": 0.0033, "step": 85710 }, { "grad_norm": 0.036007314920425415, "learning_rate": 5.472967643509525e-06, "loss": 0.0031, "step": 85720 }, { "grad_norm": 0.04752470925450325, "learning_rate": 5.46544838334419e-06, "loss": 0.0023, "step": 85730 }, { "grad_norm": 0.03400038182735443, "learning_rate": 5.4579339934104525e-06, "loss": 0.0028, "step": 85740 }, { "grad_norm": 0.0473962165415287, "learning_rate": 5.450424474530091e-06, "loss": 0.0023, "step": 85750 }, { "grad_norm": 0.060464538633823395, "learning_rate": 5.442919827524323e-06, "loss": 0.0026, "step": 85760 }, { "grad_norm": 0.04476991295814514, "learning_rate": 5.435420053213863e-06, "loss": 0.0027, "step": 85770 }, { "grad_norm": 0.06825564056634903, "learning_rate": 5.427925152418845e-06, "loss": 0.0018, "step": 85780 }, { "grad_norm": 0.039216943085193634, "learning_rate": 5.420435125958934e-06, "loss": 0.0019, "step": 85790 }, { "grad_norm": 0.04048135504126549, "learning_rate": 5.412949974653209e-06, "loss": 0.0034, "step": 85800 }, { "grad_norm": 0.049780331552028656, "learning_rate": 5.405469699320248e-06, "loss": 0.003, "step": 85810 }, { "grad_norm": 0.05840221419930458, "learning_rate": 5.397994300778064e-06, "loss": 0.0053, "step": 85820 }, { "grad_norm": 0.07457157224416733, "learning_rate": 5.39052377984417e-06, "loss": 0.0041, "step": 85830 }, { "grad_norm": 0.07320038229227066, "learning_rate": 5.383058137335523e-06, "loss": 0.0029, "step": 85840 }, { "grad_norm": 0.06120599806308746, "learning_rate": 5.375597374068553e-06, "loss": 0.0021, "step": 85850 }, { "grad_norm": 0.04226559400558472, "learning_rate": 5.368141490859163e-06, "loss": 0.0029, "step": 85860 }, { "grad_norm": 0.03941309452056885, "learning_rate": 5.360690488522729e-06, "loss": 0.0027, "step": 85870 }, { "grad_norm": 0.058273572474718094, "learning_rate": 5.353244367874055e-06, "loss": 0.0025, "step": 85880 }, { "grad_norm": 0.07417801022529602, "learning_rate": 5.345803129727456e-06, "loss": 0.0028, "step": 85890 }, { "grad_norm": 0.07141539454460144, "learning_rate": 5.338366774896697e-06, "loss": 0.0037, "step": 85900 }, { "grad_norm": 0.02400224097073078, "learning_rate": 5.330935304194995e-06, "loss": 0.0025, "step": 85910 }, { "grad_norm": 0.02761424519121647, "learning_rate": 5.323508718435044e-06, "loss": 0.002, "step": 85920 }, { "grad_norm": 0.053986359387636185, "learning_rate": 5.316087018429011e-06, "loss": 0.0026, "step": 85930 }, { "grad_norm": 0.03609619289636612, "learning_rate": 5.30867020498853e-06, "loss": 0.0026, "step": 85940 }, { "grad_norm": 0.05334632471203804, "learning_rate": 5.30125827892467e-06, "loss": 0.0029, "step": 85950 }, { "grad_norm": 0.021510258316993713, "learning_rate": 5.293851241048015e-06, "loss": 0.0041, "step": 85960 }, { "grad_norm": 0.06635324656963348, "learning_rate": 5.286449092168555e-06, "loss": 0.0024, "step": 85970 }, { "grad_norm": 0.07174506783485413, "learning_rate": 5.2790518330958105e-06, "loss": 0.0022, "step": 85980 }, { "grad_norm": 0.07028009742498398, "learning_rate": 5.271659464638717e-06, "loss": 0.0023, "step": 85990 }, { "grad_norm": 0.0674041211605072, "learning_rate": 5.264271987605701e-06, "loss": 0.0024, "step": 86000 }, { "grad_norm": 0.014028818346560001, "learning_rate": 5.256889402804632e-06, "loss": 0.0018, "step": 86010 }, { "grad_norm": 0.02711443044245243, "learning_rate": 5.249511711042881e-06, "loss": 0.0023, "step": 86020 }, { "grad_norm": 0.04954691231250763, "learning_rate": 5.242138913127237e-06, "loss": 0.0035, "step": 86030 }, { "grad_norm": 0.0441257618367672, "learning_rate": 5.234771009863987e-06, "loss": 0.0021, "step": 86040 }, { "grad_norm": 0.05569330230355263, "learning_rate": 5.227408002058881e-06, "loss": 0.0025, "step": 86050 }, { "grad_norm": 0.06468489021062851, "learning_rate": 5.220049890517126e-06, "loss": 0.0034, "step": 86060 }, { "grad_norm": 0.03295812010765076, "learning_rate": 5.212696676043377e-06, "loss": 0.0033, "step": 86070 }, { "grad_norm": 0.027115585282444954, "learning_rate": 5.205348359441798e-06, "loss": 0.0018, "step": 86080 }, { "grad_norm": 0.028235450387001038, "learning_rate": 5.198004941515966e-06, "loss": 0.0022, "step": 86090 }, { "grad_norm": 0.033826038241386414, "learning_rate": 5.190666423068952e-06, "loss": 0.002, "step": 86100 }, { "grad_norm": 0.028406543657183647, "learning_rate": 5.183332804903284e-06, "loss": 0.0024, "step": 86110 }, { "grad_norm": 0.020327521488070488, "learning_rate": 5.176004087820979e-06, "loss": 0.0036, "step": 86120 }, { "grad_norm": 0.06141739338636398, "learning_rate": 5.168680272623455e-06, "loss": 0.0031, "step": 86130 }, { "grad_norm": 0.09930942952632904, "learning_rate": 5.161361360111661e-06, "loss": 0.0024, "step": 86140 }, { "grad_norm": 0.04419457167387009, "learning_rate": 5.154047351085984e-06, "loss": 0.0021, "step": 86150 }, { "grad_norm": 0.03935200721025467, "learning_rate": 5.1467382463462385e-06, "loss": 0.0023, "step": 86160 }, { "grad_norm": 0.03345948085188866, "learning_rate": 5.139434046691788e-06, "loss": 0.0019, "step": 86170 }, { "grad_norm": 0.04966834932565689, "learning_rate": 5.13213475292137e-06, "loss": 0.0021, "step": 86180 }, { "grad_norm": 0.05493877828121185, "learning_rate": 5.12484036583325e-06, "loss": 0.0034, "step": 86190 }, { "grad_norm": 0.03127504512667656, "learning_rate": 5.117550886225109e-06, "loss": 0.0027, "step": 86200 }, { "grad_norm": 0.03801039233803749, "learning_rate": 5.11026631489413e-06, "loss": 0.0016, "step": 86210 }, { "grad_norm": 0.04355722293257713, "learning_rate": 5.102986652636926e-06, "loss": 0.0039, "step": 86220 }, { "grad_norm": 0.02245071902871132, "learning_rate": 5.095711900249606e-06, "loss": 0.0036, "step": 86230 }, { "grad_norm": 0.02720518782734871, "learning_rate": 5.088442058527715e-06, "loss": 0.0021, "step": 86240 }, { "grad_norm": 0.033183034509420395, "learning_rate": 5.08117712826629e-06, "loss": 0.0025, "step": 86250 }, { "grad_norm": 0.03476594761013985, "learning_rate": 5.073917110259785e-06, "loss": 0.0031, "step": 86260 }, { "grad_norm": 0.02682413160800934, "learning_rate": 5.066662005302175e-06, "loss": 0.0021, "step": 86270 }, { "grad_norm": 0.02211243286728859, "learning_rate": 5.059411814186838e-06, "loss": 0.0041, "step": 86280 }, { "grad_norm": 0.02498651295900345, "learning_rate": 5.05216653770666e-06, "loss": 0.0013, "step": 86290 }, { "grad_norm": 0.021931830793619156, "learning_rate": 5.044926176653969e-06, "loss": 0.0023, "step": 86300 }, { "grad_norm": 0.04428894817829132, "learning_rate": 5.037690731820571e-06, "loss": 0.0023, "step": 86310 }, { "grad_norm": 0.043777018785476685, "learning_rate": 5.030460203997711e-06, "loss": 0.0017, "step": 86320 }, { "grad_norm": 0.061562422662973404, "learning_rate": 5.0232345939761174e-06, "loss": 0.003, "step": 86330 }, { "grad_norm": 0.029540373012423515, "learning_rate": 5.016013902545957e-06, "loss": 0.0026, "step": 86340 }, { "grad_norm": 0.04965386539697647, "learning_rate": 5.008798130496889e-06, "loss": 0.0022, "step": 86350 }, { "grad_norm": 0.06919822096824646, "learning_rate": 5.001587278618008e-06, "loss": 0.0022, "step": 86360 }, { "grad_norm": 0.03599907457828522, "learning_rate": 4.994381347697891e-06, "loss": 0.0019, "step": 86370 }, { "grad_norm": 0.06245826929807663, "learning_rate": 4.987180338524572e-06, "loss": 0.002, "step": 86380 }, { "grad_norm": 0.027027839794754982, "learning_rate": 4.9799842518855275e-06, "loss": 0.0028, "step": 86390 }, { "grad_norm": 0.03372389078140259, "learning_rate": 4.9727930885677285e-06, "loss": 0.0045, "step": 86400 }, { "grad_norm": 0.030409805476665497, "learning_rate": 4.965606849357573e-06, "loss": 0.002, "step": 86410 }, { "grad_norm": 0.06487718224525452, "learning_rate": 4.9584255350409445e-06, "loss": 0.0045, "step": 86420 }, { "grad_norm": 0.037994589656591415, "learning_rate": 4.951249146403175e-06, "loss": 0.0045, "step": 86430 }, { "grad_norm": 0.06126632168889046, "learning_rate": 4.9440776842290815e-06, "loss": 0.0022, "step": 86440 }, { "grad_norm": 0.0644264742732048, "learning_rate": 4.936911149302903e-06, "loss": 0.0032, "step": 86450 }, { "grad_norm": 0.06321373581886292, "learning_rate": 4.929749542408379e-06, "loss": 0.0027, "step": 86460 }, { "grad_norm": 0.030680911615490913, "learning_rate": 4.922592864328673e-06, "loss": 0.0027, "step": 86470 }, { "grad_norm": 0.07465813308954239, "learning_rate": 4.915441115846436e-06, "loss": 0.0032, "step": 86480 }, { "grad_norm": 0.03868947550654411, "learning_rate": 4.908294297743782e-06, "loss": 0.0015, "step": 86490 }, { "grad_norm": 0.0256258025765419, "learning_rate": 4.901152410802273e-06, "loss": 0.0043, "step": 86500 }, { "grad_norm": 0.03674192354083061, "learning_rate": 4.894015455802919e-06, "loss": 0.0018, "step": 86510 }, { "grad_norm": 0.01931631565093994, "learning_rate": 4.8868834335262346e-06, "loss": 0.0021, "step": 86520 }, { "grad_norm": 0.04145147278904915, "learning_rate": 4.87975634475214e-06, "loss": 0.0031, "step": 86530 }, { "grad_norm": 0.07829608023166656, "learning_rate": 4.8726341902600614e-06, "loss": 0.0018, "step": 86540 }, { "grad_norm": 0.02580058015882969, "learning_rate": 4.865516970828843e-06, "loss": 0.0022, "step": 86550 }, { "grad_norm": 0.0765685886144638, "learning_rate": 4.85840468723685e-06, "loss": 0.0025, "step": 86560 }, { "grad_norm": 0.032494913786649704, "learning_rate": 4.85129734026184e-06, "loss": 0.0021, "step": 86570 }, { "grad_norm": 0.01584547385573387, "learning_rate": 4.844194930681073e-06, "loss": 0.0017, "step": 86580 }, { "grad_norm": 0.039687227457761765, "learning_rate": 4.837097459271267e-06, "loss": 0.0022, "step": 86590 }, { "grad_norm": 0.025213513523340225, "learning_rate": 4.830004926808573e-06, "loss": 0.0018, "step": 86600 }, { "grad_norm": 0.07478372007608414, "learning_rate": 4.822917334068627e-06, "loss": 0.0026, "step": 86610 }, { "grad_norm": 0.025345847010612488, "learning_rate": 4.81583468182652e-06, "loss": 0.0013, "step": 86620 }, { "grad_norm": 0.04425879940390587, "learning_rate": 4.808756970856809e-06, "loss": 0.0014, "step": 86630 }, { "grad_norm": 0.06673482060432434, "learning_rate": 4.801684201933476e-06, "loss": 0.0059, "step": 86640 }, { "grad_norm": 0.06756241619586945, "learning_rate": 4.794616375830019e-06, "loss": 0.0059, "step": 86650 }, { "grad_norm": 0.026168884709477425, "learning_rate": 4.787553493319341e-06, "loss": 0.0023, "step": 86660 }, { "grad_norm": 0.04765188321471214, "learning_rate": 4.780495555173836e-06, "loss": 0.0028, "step": 86670 }, { "grad_norm": 0.05783502385020256, "learning_rate": 4.7734425621653475e-06, "loss": 0.0024, "step": 86680 }, { "grad_norm": 0.04191957414150238, "learning_rate": 4.766394515065198e-06, "loss": 0.0047, "step": 86690 }, { "grad_norm": 0.03052789717912674, "learning_rate": 4.75935141464412e-06, "loss": 0.0024, "step": 86700 }, { "grad_norm": 0.023972392082214355, "learning_rate": 4.75231326167237e-06, "loss": 0.0033, "step": 86710 }, { "grad_norm": 0.037999607622623444, "learning_rate": 4.745280056919599e-06, "loss": 0.0022, "step": 86720 }, { "grad_norm": 0.07098758965730667, "learning_rate": 4.738251801154975e-06, "loss": 0.0021, "step": 86730 }, { "grad_norm": 0.04990347474813461, "learning_rate": 4.731228495147061e-06, "loss": 0.0013, "step": 86740 }, { "grad_norm": 0.030731666833162308, "learning_rate": 4.7242101396639595e-06, "loss": 0.0031, "step": 86750 }, { "grad_norm": 0.03579730913043022, "learning_rate": 4.717196735473151e-06, "loss": 0.0019, "step": 86760 }, { "grad_norm": 0.038125135004520416, "learning_rate": 4.710188283341643e-06, "loss": 0.0021, "step": 86770 }, { "grad_norm": 0.05648866668343544, "learning_rate": 4.703184784035835e-06, "loss": 0.0026, "step": 86780 }, { "grad_norm": 0.03059482015669346, "learning_rate": 4.696186238321642e-06, "loss": 0.0029, "step": 86790 }, { "grad_norm": 0.02449541538953781, "learning_rate": 4.689192646964408e-06, "loss": 0.003, "step": 86800 }, { "grad_norm": 0.04034603759646416, "learning_rate": 4.682204010728952e-06, "loss": 0.0021, "step": 86810 }, { "grad_norm": 0.05052794888615608, "learning_rate": 4.67522033037952e-06, "loss": 0.0022, "step": 86820 }, { "grad_norm": 0.029590405523777008, "learning_rate": 4.668241606679846e-06, "loss": 0.0018, "step": 86830 }, { "grad_norm": 0.02759273163974285, "learning_rate": 4.66126784039313e-06, "loss": 0.0027, "step": 86840 }, { "grad_norm": 0.030745888128876686, "learning_rate": 4.6542990322819836e-06, "loss": 0.002, "step": 86850 }, { "grad_norm": 0.026865359395742416, "learning_rate": 4.64733518310852e-06, "loss": 0.0041, "step": 86860 }, { "grad_norm": 0.027616454288363457, "learning_rate": 4.640376293634291e-06, "loss": 0.0018, "step": 86870 }, { "grad_norm": 0.04757134988903999, "learning_rate": 4.633422364620327e-06, "loss": 0.0034, "step": 86880 }, { "grad_norm": 0.06179681420326233, "learning_rate": 4.6264733968270745e-06, "loss": 0.0023, "step": 86890 }, { "grad_norm": 0.045347943902015686, "learning_rate": 4.619529391014482e-06, "loss": 0.0024, "step": 86900 }, { "grad_norm": 0.042668115347623825, "learning_rate": 4.61259034794192e-06, "loss": 0.0029, "step": 86910 }, { "grad_norm": 0.027926620095968246, "learning_rate": 4.605656268368247e-06, "loss": 0.0018, "step": 86920 }, { "grad_norm": 0.05266914889216423, "learning_rate": 4.598727153051735e-06, "loss": 0.0031, "step": 86930 }, { "grad_norm": 0.06312093883752823, "learning_rate": 4.591803002750178e-06, "loss": 0.0029, "step": 86940 }, { "grad_norm": 0.03256355971097946, "learning_rate": 4.58488381822077e-06, "loss": 0.0035, "step": 86950 }, { "grad_norm": 0.03058619797229767, "learning_rate": 4.5779696002201946e-06, "loss": 0.0025, "step": 86960 }, { "grad_norm": 0.034203991293907166, "learning_rate": 4.571060349504563e-06, "loss": 0.0037, "step": 86970 }, { "grad_norm": 0.03832349553704262, "learning_rate": 4.564156066829478e-06, "loss": 0.0016, "step": 86980 }, { "grad_norm": 0.05455496907234192, "learning_rate": 4.557256752949957e-06, "loss": 0.0025, "step": 86990 }, { "grad_norm": 0.0344991609454155, "learning_rate": 4.550362408620534e-06, "loss": 0.003, "step": 87000 }, { "grad_norm": 0.07125308364629745, "learning_rate": 4.543473034595136e-06, "loss": 0.0038, "step": 87010 }, { "grad_norm": 0.06625911593437195, "learning_rate": 4.536588631627192e-06, "loss": 0.0021, "step": 87020 }, { "grad_norm": 0.059682950377464294, "learning_rate": 4.5297092004695565e-06, "loss": 0.0028, "step": 87030 }, { "grad_norm": 0.04581134021282196, "learning_rate": 4.522834741874555e-06, "loss": 0.002, "step": 87040 }, { "grad_norm": 0.03202055022120476, "learning_rate": 4.515965256593974e-06, "loss": 0.0028, "step": 87050 }, { "grad_norm": 0.0686088353395462, "learning_rate": 4.509100745379058e-06, "loss": 0.0053, "step": 87060 }, { "grad_norm": 0.04239681735634804, "learning_rate": 4.502241208980479e-06, "loss": 0.0029, "step": 87070 }, { "grad_norm": 0.03472960367798805, "learning_rate": 4.495386648148398e-06, "loss": 0.0024, "step": 87080 }, { "grad_norm": 0.030016325414180756, "learning_rate": 4.488537063632431e-06, "loss": 0.0051, "step": 87090 }, { "grad_norm": 0.04509887844324112, "learning_rate": 4.481692456181608e-06, "loss": 0.0035, "step": 87100 }, { "grad_norm": 0.054703906178474426, "learning_rate": 4.474852826544479e-06, "loss": 0.0026, "step": 87110 }, { "grad_norm": 0.03452537953853607, "learning_rate": 4.468018175468974e-06, "loss": 0.0016, "step": 87120 }, { "grad_norm": 0.06765807420015335, "learning_rate": 4.461188503702568e-06, "loss": 0.0035, "step": 87130 }, { "grad_norm": 0.057142309844493866, "learning_rate": 4.454363811992107e-06, "loss": 0.0022, "step": 87140 }, { "grad_norm": 0.020340895280241966, "learning_rate": 4.447544101083956e-06, "loss": 0.0014, "step": 87150 }, { "grad_norm": 0.0419897586107254, "learning_rate": 4.440729371723884e-06, "loss": 0.0034, "step": 87160 }, { "grad_norm": 0.03920938819646835, "learning_rate": 4.433919624657162e-06, "loss": 0.0024, "step": 87170 }, { "grad_norm": 0.04435570910573006, "learning_rate": 4.427114860628462e-06, "loss": 0.0028, "step": 87180 }, { "grad_norm": 0.06132723018527031, "learning_rate": 4.420315080381982e-06, "loss": 0.0032, "step": 87190 }, { "grad_norm": 0.03442506864666939, "learning_rate": 4.4135202846613045e-06, "loss": 0.0017, "step": 87200 }, { "grad_norm": 0.054821740835905075, "learning_rate": 4.406730474209525e-06, "loss": 0.0027, "step": 87210 }, { "grad_norm": 0.06418764591217041, "learning_rate": 4.399945649769138e-06, "loss": 0.0039, "step": 87220 }, { "grad_norm": 0.048878014087677, "learning_rate": 4.393165812082151e-06, "loss": 0.003, "step": 87230 }, { "grad_norm": 0.029647089540958405, "learning_rate": 4.386390961889969e-06, "loss": 0.0022, "step": 87240 }, { "grad_norm": 0.07735378295183182, "learning_rate": 4.3796210999334895e-06, "loss": 0.0019, "step": 87250 }, { "grad_norm": 0.08064884692430496, "learning_rate": 4.372856226953065e-06, "loss": 0.0022, "step": 87260 }, { "grad_norm": 0.048078503459692, "learning_rate": 4.366096343688486e-06, "loss": 0.0023, "step": 87270 }, { "grad_norm": 0.08888900279998779, "learning_rate": 4.359341450878995e-06, "loss": 0.0034, "step": 87280 }, { "grad_norm": 0.0241124015301466, "learning_rate": 4.352591549263302e-06, "loss": 0.004, "step": 87290 }, { "grad_norm": 0.029434548690915108, "learning_rate": 4.345846639579582e-06, "loss": 0.0036, "step": 87300 }, { "grad_norm": 0.04655930772423744, "learning_rate": 4.339106722565417e-06, "loss": 0.002, "step": 87310 }, { "grad_norm": 0.03515389561653137, "learning_rate": 4.332371798957896e-06, "loss": 0.0024, "step": 87320 }, { "grad_norm": 0.040739819407463074, "learning_rate": 4.3256418694935295e-06, "loss": 0.0026, "step": 87330 }, { "grad_norm": 0.06793061643838882, "learning_rate": 4.3189169349083125e-06, "loss": 0.0018, "step": 87340 }, { "grad_norm": 0.05330134928226471, "learning_rate": 4.312196995937645e-06, "loss": 0.0016, "step": 87350 }, { "grad_norm": 0.02193836309015751, "learning_rate": 4.305482053316434e-06, "loss": 0.002, "step": 87360 }, { "grad_norm": 0.06895066052675247, "learning_rate": 4.2987721077789925e-06, "loss": 0.0027, "step": 87370 }, { "grad_norm": 0.03386683762073517, "learning_rate": 4.2920671600591325e-06, "loss": 0.0023, "step": 87380 }, { "grad_norm": 0.022733386605978012, "learning_rate": 4.285367210890084e-06, "loss": 0.0021, "step": 87390 }, { "grad_norm": 0.037800054997205734, "learning_rate": 4.278672261004551e-06, "loss": 0.0015, "step": 87400 }, { "grad_norm": 0.0365016795694828, "learning_rate": 4.2719823111346704e-06, "loss": 0.0022, "step": 87410 }, { "grad_norm": 0.029032999649643898, "learning_rate": 4.265297362012061e-06, "loss": 0.0021, "step": 87420 }, { "grad_norm": 0.032367341220378876, "learning_rate": 4.258617414367766e-06, "loss": 0.0019, "step": 87430 }, { "grad_norm": 0.035235412418842316, "learning_rate": 4.251942468932296e-06, "loss": 0.0019, "step": 87440 }, { "grad_norm": 0.045181795954704285, "learning_rate": 4.2452725264356175e-06, "loss": 0.0017, "step": 87450 }, { "grad_norm": 0.0856250450015068, "learning_rate": 4.2386075876071575e-06, "loss": 0.0021, "step": 87460 }, { "grad_norm": 0.04072950780391693, "learning_rate": 4.231947653175755e-06, "loss": 0.002, "step": 87470 }, { "grad_norm": 0.04308602586388588, "learning_rate": 4.225292723869761e-06, "loss": 0.0023, "step": 87480 }, { "grad_norm": 0.08447297662496567, "learning_rate": 4.218642800416916e-06, "loss": 0.0022, "step": 87490 }, { "grad_norm": 0.02990705892443657, "learning_rate": 4.211997883544472e-06, "loss": 0.0052, "step": 87500 }, { "grad_norm": 0.03164408355951309, "learning_rate": 4.205357973979096e-06, "loss": 0.0031, "step": 87510 }, { "grad_norm": 0.02730683982372284, "learning_rate": 4.198723072446925e-06, "loss": 0.0027, "step": 87520 }, { "grad_norm": 0.02331615425646305, "learning_rate": 4.192093179673534e-06, "loss": 0.0016, "step": 87530 }, { "grad_norm": 0.031414322555065155, "learning_rate": 4.185468296383959e-06, "loss": 0.0021, "step": 87540 }, { "grad_norm": 0.056287605315446854, "learning_rate": 4.1788484233027005e-06, "loss": 0.0034, "step": 87550 }, { "grad_norm": 0.03367321565747261, "learning_rate": 4.172233561153677e-06, "loss": 0.0024, "step": 87560 }, { "grad_norm": 0.042871780693531036, "learning_rate": 4.165623710660288e-06, "loss": 0.0019, "step": 87570 }, { "grad_norm": 0.05954708904027939, "learning_rate": 4.1590188725453835e-06, "loss": 0.003, "step": 87580 }, { "grad_norm": 0.021173791959881783, "learning_rate": 4.152419047531259e-06, "loss": 0.0019, "step": 87590 }, { "grad_norm": 0.05720863491296768, "learning_rate": 4.145824236339651e-06, "loss": 0.0047, "step": 87600 }, { "grad_norm": 0.06147274002432823, "learning_rate": 4.1392344396917695e-06, "loss": 0.0029, "step": 87610 }, { "grad_norm": 0.053193990141153336, "learning_rate": 4.132649658308252e-06, "loss": 0.0042, "step": 87620 }, { "grad_norm": 0.04512985050678253, "learning_rate": 4.126069892909213e-06, "loss": 0.0024, "step": 87630 }, { "grad_norm": 0.06210790202021599, "learning_rate": 4.119495144214191e-06, "loss": 0.0031, "step": 87640 }, { "grad_norm": 0.06706760823726654, "learning_rate": 4.112925412942214e-06, "loss": 0.0016, "step": 87650 }, { "grad_norm": 0.023764004930853844, "learning_rate": 4.106360699811712e-06, "loss": 0.0023, "step": 87660 }, { "grad_norm": 0.04427873343229294, "learning_rate": 4.099801005540616e-06, "loss": 0.0026, "step": 87670 }, { "grad_norm": 0.04210970923304558, "learning_rate": 4.093246330846256e-06, "loss": 0.0037, "step": 87680 }, { "grad_norm": 0.03314879909157753, "learning_rate": 4.086696676445467e-06, "loss": 0.0042, "step": 87690 }, { "grad_norm": 0.07856100797653198, "learning_rate": 4.0801520430544914e-06, "loss": 0.005, "step": 87700 }, { "grad_norm": 0.06859782338142395, "learning_rate": 4.073612431389062e-06, "loss": 0.0037, "step": 87710 }, { "grad_norm": 0.08553217351436615, "learning_rate": 4.067077842164319e-06, "loss": 0.0027, "step": 87720 }, { "grad_norm": 0.03729292377829552, "learning_rate": 4.0605482760948945e-06, "loss": 0.002, "step": 87730 }, { "grad_norm": 0.051021356135606766, "learning_rate": 4.054023733894829e-06, "loss": 0.0032, "step": 87740 }, { "grad_norm": 0.0691632628440857, "learning_rate": 4.047504216277653e-06, "loss": 0.0016, "step": 87750 }, { "grad_norm": 0.027025822550058365, "learning_rate": 4.040989723956323e-06, "loss": 0.0029, "step": 87760 }, { "grad_norm": 0.01921318657696247, "learning_rate": 4.03448025764327e-06, "loss": 0.0022, "step": 87770 }, { "grad_norm": 0.05091432109475136, "learning_rate": 4.027975818050339e-06, "loss": 0.0019, "step": 87780 }, { "grad_norm": 0.02730243280529976, "learning_rate": 4.021476405888858e-06, "loss": 0.0026, "step": 87790 }, { "grad_norm": 0.03951761871576309, "learning_rate": 4.01498202186959e-06, "loss": 0.0017, "step": 87800 }, { "grad_norm": 0.030681796371936798, "learning_rate": 4.008492666702751e-06, "loss": 0.0018, "step": 87810 }, { "grad_norm": 0.02699960395693779, "learning_rate": 4.002008341098001e-06, "loss": 0.0024, "step": 87820 }, { "grad_norm": 0.05340362340211868, "learning_rate": 3.995529045764462e-06, "loss": 0.0026, "step": 87830 }, { "grad_norm": 0.03803316131234169, "learning_rate": 3.989054781410712e-06, "loss": 0.005, "step": 87840 }, { "grad_norm": 0.0279734767973423, "learning_rate": 3.982585548744744e-06, "loss": 0.0035, "step": 87850 }, { "grad_norm": 0.03541635349392891, "learning_rate": 3.9761213484740435e-06, "loss": 0.0033, "step": 87860 }, { "grad_norm": 0.031665828078985214, "learning_rate": 3.9696621813055055e-06, "loss": 0.0024, "step": 87870 }, { "grad_norm": 0.04349420219659805, "learning_rate": 3.9632080479455044e-06, "loss": 0.0024, "step": 87880 }, { "grad_norm": 0.045649442821741104, "learning_rate": 3.956758949099859e-06, "loss": 0.003, "step": 87890 }, { "grad_norm": 0.06948979943990707, "learning_rate": 3.9503148854738385e-06, "loss": 0.0027, "step": 87900 }, { "grad_norm": 0.06746570020914078, "learning_rate": 3.943875857772139e-06, "loss": 0.0019, "step": 87910 }, { "grad_norm": 0.0259823240339756, "learning_rate": 3.9374418666989375e-06, "loss": 0.0019, "step": 87920 }, { "grad_norm": 0.051154326647520065, "learning_rate": 3.931012912957838e-06, "loss": 0.003, "step": 87930 }, { "grad_norm": 0.03589973971247673, "learning_rate": 3.9245889972519045e-06, "loss": 0.0014, "step": 87940 }, { "grad_norm": 0.05294341966509819, "learning_rate": 3.9181701202836265e-06, "loss": 0.0021, "step": 87950 }, { "grad_norm": 0.07218591123819351, "learning_rate": 3.911756282755003e-06, "loss": 0.003, "step": 87960 }, { "grad_norm": 0.04575704038143158, "learning_rate": 3.905347485367411e-06, "loss": 0.0028, "step": 87970 }, { "grad_norm": 0.03328021615743637, "learning_rate": 3.898943728821725e-06, "loss": 0.0017, "step": 87980 }, { "grad_norm": 0.035099487751722336, "learning_rate": 3.892545013818227e-06, "loss": 0.0037, "step": 87990 }, { "grad_norm": 0.04605838656425476, "learning_rate": 3.886151341056687e-06, "loss": 0.0046, "step": 88000 }, { "grad_norm": 0.027022240683436394, "learning_rate": 3.8797627112363066e-06, "loss": 0.0015, "step": 88010 }, { "grad_norm": 0.03830292448401451, "learning_rate": 3.873379125055743e-06, "loss": 0.0023, "step": 88020 }, { "grad_norm": 0.04738445580005646, "learning_rate": 3.867000583213082e-06, "loss": 0.0014, "step": 88030 }, { "grad_norm": 0.021844130009412766, "learning_rate": 3.860627086405871e-06, "loss": 0.0019, "step": 88040 }, { "grad_norm": 0.03014487773180008, "learning_rate": 3.854258635331126e-06, "loss": 0.0045, "step": 88050 }, { "grad_norm": 0.0870489776134491, "learning_rate": 3.8478952306852665e-06, "loss": 0.0032, "step": 88060 }, { "grad_norm": 0.038319069892168045, "learning_rate": 3.841536873164198e-06, "loss": 0.0022, "step": 88070 }, { "grad_norm": 0.05682750791311264, "learning_rate": 3.8351835634632626e-06, "loss": 0.003, "step": 88080 }, { "grad_norm": 0.05448656901717186, "learning_rate": 3.828835302277245e-06, "loss": 0.0019, "step": 88090 }, { "grad_norm": 0.07314594835042953, "learning_rate": 3.822492090300378e-06, "loss": 0.0028, "step": 88100 }, { "grad_norm": 0.07823728024959564, "learning_rate": 3.8161539282263566e-06, "loss": 0.0028, "step": 88110 }, { "grad_norm": 0.07690925151109695, "learning_rate": 3.8098208167482995e-06, "loss": 0.0018, "step": 88120 }, { "grad_norm": 0.033094294369220734, "learning_rate": 3.803492756558796e-06, "loss": 0.0018, "step": 88130 }, { "grad_norm": 0.059423256665468216, "learning_rate": 3.797169748349855e-06, "loss": 0.0038, "step": 88140 }, { "grad_norm": 0.02663746289908886, "learning_rate": 3.790851792812977e-06, "loss": 0.0028, "step": 88150 }, { "grad_norm": 0.05467129498720169, "learning_rate": 3.784538890639072e-06, "loss": 0.0027, "step": 88160 }, { "grad_norm": 0.023010965436697006, "learning_rate": 3.7782310425185153e-06, "loss": 0.0023, "step": 88170 }, { "grad_norm": 0.0195023100823164, "learning_rate": 3.77192824914111e-06, "loss": 0.0014, "step": 88180 }, { "grad_norm": 0.061031974852085114, "learning_rate": 3.7656305111961333e-06, "loss": 0.0034, "step": 88190 }, { "grad_norm": 0.05295870825648308, "learning_rate": 3.7593378293722782e-06, "loss": 0.0034, "step": 88200 }, { "grad_norm": 0.051758747547864914, "learning_rate": 3.7530502043577333e-06, "loss": 0.0024, "step": 88210 }, { "grad_norm": 0.028094492852687836, "learning_rate": 3.746767636840076e-06, "loss": 0.0017, "step": 88220 }, { "grad_norm": 0.06140318512916565, "learning_rate": 3.74049012750638e-06, "loss": 0.0025, "step": 88230 }, { "grad_norm": 0.08409448713064194, "learning_rate": 3.734217677043128e-06, "loss": 0.0022, "step": 88240 }, { "grad_norm": 0.026656148955225945, "learning_rate": 3.727950286136267e-06, "loss": 0.0016, "step": 88250 }, { "grad_norm": 0.05416570603847504, "learning_rate": 3.7216879554711925e-06, "loss": 0.0046, "step": 88260 }, { "grad_norm": 0.041869234293699265, "learning_rate": 3.715430685732757e-06, "loss": 0.0014, "step": 88270 }, { "grad_norm": 0.024730168282985687, "learning_rate": 3.7091784776052243e-06, "loss": 0.0024, "step": 88280 }, { "grad_norm": 0.04912255331873894, "learning_rate": 3.7029313317723356e-06, "loss": 0.0021, "step": 88290 }, { "grad_norm": 0.036263346672058105, "learning_rate": 3.6966892489172787e-06, "loss": 0.0023, "step": 88300 }, { "grad_norm": 0.04422992467880249, "learning_rate": 3.6904522297226575e-06, "loss": 0.0022, "step": 88310 }, { "grad_norm": 0.03062746301293373, "learning_rate": 3.6842202748705647e-06, "loss": 0.0022, "step": 88320 }, { "grad_norm": 0.02436339296400547, "learning_rate": 3.677993385042494e-06, "loss": 0.0034, "step": 88330 }, { "grad_norm": 0.02335287444293499, "learning_rate": 3.67177156091944e-06, "loss": 0.0019, "step": 88340 }, { "grad_norm": 0.030476221814751625, "learning_rate": 3.6655548031817866e-06, "loss": 0.0024, "step": 88350 }, { "grad_norm": 0.05611847713589668, "learning_rate": 3.6593431125094e-06, "loss": 0.0028, "step": 88360 }, { "grad_norm": 0.04454699903726578, "learning_rate": 3.6531364895815766e-06, "loss": 0.0018, "step": 88370 }, { "grad_norm": 0.04805945232510567, "learning_rate": 3.6469349350770677e-06, "loss": 0.0028, "step": 88380 }, { "grad_norm": 0.031308405101299286, "learning_rate": 3.6407384496740525e-06, "loss": 0.0024, "step": 88390 }, { "grad_norm": 0.07490259408950806, "learning_rate": 3.6345470340501954e-06, "loss": 0.0041, "step": 88400 }, { "grad_norm": 0.03619086742401123, "learning_rate": 3.6283606888825537e-06, "loss": 0.0031, "step": 88410 }, { "grad_norm": 0.052877310663461685, "learning_rate": 3.622179414847676e-06, "loss": 0.0035, "step": 88420 }, { "grad_norm": 0.03324494883418083, "learning_rate": 3.616003212621527e-06, "loss": 0.0024, "step": 88430 }, { "grad_norm": 0.06117795407772064, "learning_rate": 3.609832082879533e-06, "loss": 0.0024, "step": 88440 }, { "grad_norm": 0.05017508566379547, "learning_rate": 3.6036660262965485e-06, "loss": 0.003, "step": 88450 }, { "grad_norm": 0.05670822039246559, "learning_rate": 3.597505043546895e-06, "loss": 0.0023, "step": 88460 }, { "grad_norm": 0.035336509346961975, "learning_rate": 3.5913491353043227e-06, "loss": 0.0028, "step": 88470 }, { "grad_norm": 0.026732271537184715, "learning_rate": 3.5851983022420375e-06, "loss": 0.0024, "step": 88480 }, { "grad_norm": 0.026537874713540077, "learning_rate": 3.5790525450326797e-06, "loss": 0.0022, "step": 88490 }, { "grad_norm": 0.033533982932567596, "learning_rate": 3.572911864348344e-06, "loss": 0.0017, "step": 88500 }, { "grad_norm": 0.031280796974897385, "learning_rate": 3.5667762608605717e-06, "loss": 0.0033, "step": 88510 }, { "grad_norm": 0.05705590546131134, "learning_rate": 3.5606457352403255e-06, "loss": 0.002, "step": 88520 }, { "grad_norm": 0.049176111817359924, "learning_rate": 3.554520288158042e-06, "loss": 0.0051, "step": 88530 }, { "grad_norm": 0.024676810950040817, "learning_rate": 3.5483999202835903e-06, "loss": 0.0017, "step": 88540 }, { "grad_norm": 0.04319687560200691, "learning_rate": 3.5422846322862913e-06, "loss": 0.0026, "step": 88550 }, { "grad_norm": 0.03192703425884247, "learning_rate": 3.5361744248348925e-06, "loss": 0.0022, "step": 88560 }, { "grad_norm": 0.016539903357625008, "learning_rate": 3.5300692985976047e-06, "loss": 0.0021, "step": 88570 }, { "grad_norm": 0.02482330985367298, "learning_rate": 3.52396925424206e-06, "loss": 0.003, "step": 88580 }, { "grad_norm": 0.02111542783677578, "learning_rate": 3.5178742924353747e-06, "loss": 0.002, "step": 88590 }, { "grad_norm": 0.04410766437649727, "learning_rate": 3.511784413844066e-06, "loss": 0.002, "step": 88600 }, { "grad_norm": 0.04914728179574013, "learning_rate": 3.5056996191341287e-06, "loss": 0.0025, "step": 88610 }, { "grad_norm": 0.04676733538508415, "learning_rate": 3.4996199089709692e-06, "loss": 0.0019, "step": 88620 }, { "grad_norm": 0.060962848365306854, "learning_rate": 3.4935452840194725e-06, "loss": 0.003, "step": 88630 }, { "grad_norm": 0.07279569655656815, "learning_rate": 3.48747574494393e-06, "loss": 0.0024, "step": 88640 }, { "grad_norm": 0.0402531735599041, "learning_rate": 3.4814112924081154e-06, "loss": 0.0022, "step": 88650 }, { "grad_norm": 0.023036526516079903, "learning_rate": 3.475351927075221e-06, "loss": 0.0021, "step": 88660 }, { "grad_norm": 0.06266935169696808, "learning_rate": 3.469297649607894e-06, "loss": 0.0024, "step": 88670 }, { "grad_norm": 0.03711804747581482, "learning_rate": 3.463248460668217e-06, "loss": 0.0036, "step": 88680 }, { "grad_norm": 0.03987259045243263, "learning_rate": 3.457204360917726e-06, "loss": 0.0018, "step": 88690 }, { "grad_norm": 0.03818083181977272, "learning_rate": 3.4511653510173827e-06, "loss": 0.0027, "step": 88700 }, { "grad_norm": 0.025671083480119705, "learning_rate": 3.4451314316276074e-06, "loss": 0.0021, "step": 88710 }, { "grad_norm": 0.05163063481450081, "learning_rate": 3.4391026034082674e-06, "loss": 0.0022, "step": 88720 }, { "grad_norm": 0.04179442301392555, "learning_rate": 3.433078867018674e-06, "loss": 0.003, "step": 88730 }, { "grad_norm": 0.04906238242983818, "learning_rate": 3.427060223117551e-06, "loss": 0.0056, "step": 88740 }, { "grad_norm": 0.03337644040584564, "learning_rate": 3.4210466723631042e-06, "loss": 0.0014, "step": 88750 }, { "grad_norm": 0.030260900035500526, "learning_rate": 3.415038215412969e-06, "loss": 0.003, "step": 88760 }, { "grad_norm": 0.046907030045986176, "learning_rate": 3.4090348529242088e-06, "loss": 0.0027, "step": 88770 }, { "grad_norm": 0.05537812039256096, "learning_rate": 3.4030365855533487e-06, "loss": 0.0022, "step": 88780 }, { "grad_norm": 0.023538656532764435, "learning_rate": 3.397043413956347e-06, "loss": 0.0039, "step": 88790 }, { "grad_norm": 0.025227487087249756, "learning_rate": 3.3910553387886237e-06, "loss": 0.0013, "step": 88800 }, { "grad_norm": 0.03893237188458443, "learning_rate": 3.3850723607049996e-06, "loss": 0.0028, "step": 88810 }, { "grad_norm": 0.027275457978248596, "learning_rate": 3.3790944803597902e-06, "loss": 0.002, "step": 88820 }, { "grad_norm": 0.03717436641454697, "learning_rate": 3.373121698406706e-06, "loss": 0.0028, "step": 88830 }, { "grad_norm": 0.054580818861722946, "learning_rate": 3.3671540154989302e-06, "loss": 0.0029, "step": 88840 }, { "grad_norm": 0.05446145310997963, "learning_rate": 3.361191432289079e-06, "loss": 0.0033, "step": 88850 }, { "grad_norm": 0.04117342457175255, "learning_rate": 3.3552339494292207e-06, "loss": 0.0039, "step": 88860 }, { "grad_norm": 0.03934553638100624, "learning_rate": 3.349281567570839e-06, "loss": 0.0025, "step": 88870 }, { "grad_norm": 0.06232587993144989, "learning_rate": 3.3433342873648965e-06, "loss": 0.0031, "step": 88880 }, { "grad_norm": 0.03567634895443916, "learning_rate": 3.337392109461762e-06, "loss": 0.0029, "step": 88890 }, { "grad_norm": 0.030817126855254173, "learning_rate": 3.331455034511266e-06, "loss": 0.0015, "step": 88900 }, { "grad_norm": 0.054628536105155945, "learning_rate": 3.325523063162689e-06, "loss": 0.0025, "step": 88910 }, { "grad_norm": 0.030627984553575516, "learning_rate": 3.3195961960647393e-06, "loss": 0.002, "step": 88920 }, { "grad_norm": 0.025428403168916702, "learning_rate": 3.31367443386556e-06, "loss": 0.003, "step": 88930 }, { "grad_norm": 0.05493614450097084, "learning_rate": 3.3077577772127556e-06, "loss": 0.0022, "step": 88940 }, { "grad_norm": 0.032199677079916, "learning_rate": 3.301846226753358e-06, "loss": 0.0019, "step": 88950 }, { "grad_norm": 0.05018356442451477, "learning_rate": 3.2959397831338444e-06, "loss": 0.0026, "step": 88960 }, { "grad_norm": 0.019181085750460625, "learning_rate": 3.2900384470001432e-06, "loss": 0.0027, "step": 88970 }, { "grad_norm": 0.05910005047917366, "learning_rate": 3.28414221899761e-06, "loss": 0.0019, "step": 88980 }, { "grad_norm": 0.04001850262284279, "learning_rate": 3.278251099771046e-06, "loss": 0.0027, "step": 88990 }, { "grad_norm": 0.017690416425466537, "learning_rate": 3.2723650899646906e-06, "loss": 0.0019, "step": 89000 }, { "grad_norm": 0.052594974637031555, "learning_rate": 3.2664841902222466e-06, "loss": 0.0024, "step": 89010 }, { "grad_norm": 0.038277894258499146, "learning_rate": 3.2606084011868208e-06, "loss": 0.0021, "step": 89020 }, { "grad_norm": 0.038604047149419785, "learning_rate": 3.254737723500989e-06, "loss": 0.0017, "step": 89030 }, { "grad_norm": 0.040571678429841995, "learning_rate": 3.2488721578067595e-06, "loss": 0.0024, "step": 89040 }, { "grad_norm": 0.027020281180739403, "learning_rate": 3.243011704745591e-06, "loss": 0.0023, "step": 89050 }, { "grad_norm": 0.02788620814681053, "learning_rate": 3.237156364958355e-06, "loss": 0.0016, "step": 89060 }, { "grad_norm": 0.04655071347951889, "learning_rate": 3.231306139085394e-06, "loss": 0.003, "step": 89070 }, { "grad_norm": 0.0635191947221756, "learning_rate": 3.2254610277664797e-06, "loss": 0.0029, "step": 89080 }, { "grad_norm": 0.034257180988788605, "learning_rate": 3.2196210316408183e-06, "loss": 0.0023, "step": 89090 }, { "grad_norm": 0.036410797387361526, "learning_rate": 3.2137861513470644e-06, "loss": 0.0025, "step": 89100 }, { "grad_norm": 0.071274533867836, "learning_rate": 3.2079563875233252e-06, "loss": 0.003, "step": 89110 }, { "grad_norm": 0.034082453697919846, "learning_rate": 3.2021317408071184e-06, "loss": 0.0036, "step": 89120 }, { "grad_norm": 0.03792151063680649, "learning_rate": 3.196312211835434e-06, "loss": 0.0016, "step": 89130 }, { "grad_norm": 0.050908781588077545, "learning_rate": 3.1904978012446686e-06, "loss": 0.0028, "step": 89140 }, { "grad_norm": 0.052663180977106094, "learning_rate": 3.184688509670691e-06, "loss": 0.0022, "step": 89150 }, { "grad_norm": 0.030819056555628777, "learning_rate": 3.178884337748783e-06, "loss": 0.0019, "step": 89160 }, { "grad_norm": 0.08758488297462463, "learning_rate": 3.173085286113697e-06, "loss": 0.0031, "step": 89170 }, { "grad_norm": 0.05197008699178696, "learning_rate": 3.1672913553995986e-06, "loss": 0.0023, "step": 89180 }, { "grad_norm": 0.024760844185948372, "learning_rate": 3.1615025462401138e-06, "loss": 0.0022, "step": 89190 }, { "grad_norm": 0.05149642750620842, "learning_rate": 3.1557188592682808e-06, "loss": 0.0023, "step": 89200 }, { "grad_norm": 0.02250826358795166, "learning_rate": 3.1499402951166045e-06, "loss": 0.0024, "step": 89210 }, { "grad_norm": 0.06627755612134933, "learning_rate": 3.144166854417019e-06, "loss": 0.0026, "step": 89220 }, { "grad_norm": 0.06368950009346008, "learning_rate": 3.1383985378009073e-06, "loss": 0.0029, "step": 89230 }, { "grad_norm": 0.03771260008215904, "learning_rate": 3.1326353458990654e-06, "loss": 0.0036, "step": 89240 }, { "grad_norm": 0.044561244547367096, "learning_rate": 3.126877279341761e-06, "loss": 0.0018, "step": 89250 }, { "grad_norm": 0.04950358718633652, "learning_rate": 3.1211243387586964e-06, "loss": 0.0027, "step": 89260 }, { "grad_norm": 0.016100699082016945, "learning_rate": 3.115376524778979e-06, "loss": 0.0019, "step": 89270 }, { "grad_norm": 0.022955289110541344, "learning_rate": 3.109633838031195e-06, "loss": 0.0023, "step": 89280 }, { "grad_norm": 0.03607458993792534, "learning_rate": 3.103896279143359e-06, "loss": 0.0023, "step": 89290 }, { "grad_norm": 0.04906187206506729, "learning_rate": 3.098163848742924e-06, "loss": 0.0022, "step": 89300 }, { "grad_norm": 0.05303225293755531, "learning_rate": 3.0924365474567675e-06, "loss": 0.0032, "step": 89310 }, { "grad_norm": 0.061160773038864136, "learning_rate": 3.086714375911237e-06, "loss": 0.0017, "step": 89320 }, { "grad_norm": 0.05142034962773323, "learning_rate": 3.0809973347320775e-06, "loss": 0.0025, "step": 89330 }, { "grad_norm": 0.033792659640312195, "learning_rate": 3.075285424544516e-06, "loss": 0.0029, "step": 89340 }, { "grad_norm": 0.021056514233350754, "learning_rate": 3.069578645973176e-06, "loss": 0.0014, "step": 89350 }, { "grad_norm": 0.03311356157064438, "learning_rate": 3.0638769996421735e-06, "loss": 0.0021, "step": 89360 }, { "grad_norm": 0.02050468511879444, "learning_rate": 3.058180486175005e-06, "loss": 0.0013, "step": 89370 }, { "grad_norm": 0.04613974317908287, "learning_rate": 3.052489106194645e-06, "loss": 0.0021, "step": 89380 }, { "grad_norm": 0.017863279208540916, "learning_rate": 3.0468028603234946e-06, "loss": 0.0021, "step": 89390 }, { "grad_norm": 0.01964760385453701, "learning_rate": 3.04112174918339e-06, "loss": 0.0021, "step": 89400 }, { "grad_norm": 0.023040330037474632, "learning_rate": 3.0354457733955954e-06, "loss": 0.0017, "step": 89410 }, { "grad_norm": 0.027544699609279633, "learning_rate": 3.029774933580859e-06, "loss": 0.0048, "step": 89420 }, { "grad_norm": 0.06671386957168579, "learning_rate": 3.0241092303593065e-06, "loss": 0.0027, "step": 89430 }, { "grad_norm": 0.019222965463995934, "learning_rate": 3.018448664350554e-06, "loss": 0.002, "step": 89440 }, { "grad_norm": 0.04920774698257446, "learning_rate": 3.0127932361736055e-06, "loss": 0.0024, "step": 89450 }, { "grad_norm": 0.03914395347237587, "learning_rate": 3.00714294644695e-06, "loss": 0.0021, "step": 89460 }, { "grad_norm": 0.016723236069083214, "learning_rate": 3.0014977957884828e-06, "loss": 0.0016, "step": 89470 }, { "grad_norm": 0.06386102735996246, "learning_rate": 2.99585778481557e-06, "loss": 0.0042, "step": 89480 }, { "grad_norm": 0.05928133800625801, "learning_rate": 2.990222914144969e-06, "loss": 0.0019, "step": 89490 }, { "grad_norm": 0.025042438879609108, "learning_rate": 2.984593184392914e-06, "loss": 0.0025, "step": 89500 }, { "grad_norm": 0.04577279090881348, "learning_rate": 2.978968596175069e-06, "loss": 0.0021, "step": 89510 }, { "grad_norm": 0.04248645529150963, "learning_rate": 2.9733491501065137e-06, "loss": 0.0028, "step": 89520 }, { "grad_norm": 0.05270112678408623, "learning_rate": 2.9677348468018005e-06, "loss": 0.0021, "step": 89530 }, { "grad_norm": 0.04745320603251457, "learning_rate": 2.9621256868748726e-06, "loss": 0.0031, "step": 89540 }, { "grad_norm": 0.03911411017179489, "learning_rate": 2.956521670939183e-06, "loss": 0.003, "step": 89550 }, { "grad_norm": 0.028059206902980804, "learning_rate": 2.950922799607542e-06, "loss": 0.0058, "step": 89560 }, { "grad_norm": 0.015234965831041336, "learning_rate": 2.9453290734922535e-06, "loss": 0.0018, "step": 89570 }, { "grad_norm": 0.020480304956436157, "learning_rate": 2.939740493205029e-06, "loss": 0.0025, "step": 89580 }, { "grad_norm": 0.02634398080408573, "learning_rate": 2.9341570593570345e-06, "loss": 0.0029, "step": 89590 }, { "grad_norm": 0.02502588927745819, "learning_rate": 2.928578772558843e-06, "loss": 0.0021, "step": 89600 }, { "grad_norm": 0.04690703749656677, "learning_rate": 2.923005633420528e-06, "loss": 0.0023, "step": 89610 }, { "grad_norm": 0.05355937033891678, "learning_rate": 2.9174376425515237e-06, "loss": 0.0023, "step": 89620 }, { "grad_norm": 0.06995655596256256, "learning_rate": 2.91187480056076e-06, "loss": 0.0024, "step": 89630 }, { "grad_norm": 0.02691389061510563, "learning_rate": 2.9063171080565664e-06, "loss": 0.0016, "step": 89640 }, { "grad_norm": 0.09929507225751877, "learning_rate": 2.900764565646741e-06, "loss": 0.0034, "step": 89650 }, { "grad_norm": 0.017967205494642258, "learning_rate": 2.895217173938475e-06, "loss": 0.0019, "step": 89660 }, { "grad_norm": 0.03297552466392517, "learning_rate": 2.8896749335384445e-06, "loss": 0.002, "step": 89670 }, { "grad_norm": 0.016378697007894516, "learning_rate": 2.884137845052731e-06, "loss": 0.0017, "step": 89680 }, { "grad_norm": 0.05822674185037613, "learning_rate": 2.8786059090868734e-06, "loss": 0.0025, "step": 89690 }, { "grad_norm": 0.03547622635960579, "learning_rate": 2.87307912624582e-06, "loss": 0.0036, "step": 89700 }, { "grad_norm": 0.02391686849296093, "learning_rate": 2.867557497133977e-06, "loss": 0.0021, "step": 89710 }, { "grad_norm": 0.0641510933637619, "learning_rate": 2.862041022355194e-06, "loss": 0.0024, "step": 89720 }, { "grad_norm": 0.07379771023988724, "learning_rate": 2.8565297025127226e-06, "loss": 0.0038, "step": 89730 }, { "grad_norm": 0.04962461441755295, "learning_rate": 2.8510235382092853e-06, "loss": 0.0033, "step": 89740 }, { "grad_norm": 0.03293604403734207, "learning_rate": 2.845522530047029e-06, "loss": 0.0014, "step": 89750 }, { "grad_norm": 0.0869457870721817, "learning_rate": 2.8400266786275387e-06, "loss": 0.0039, "step": 89760 }, { "grad_norm": 0.06270669400691986, "learning_rate": 2.834535984551817e-06, "loss": 0.003, "step": 89770 }, { "grad_norm": 0.04685435816645622, "learning_rate": 2.8290504484203397e-06, "loss": 0.0028, "step": 89780 }, { "grad_norm": 0.015021884813904762, "learning_rate": 2.8235700708329705e-06, "loss": 0.0014, "step": 89790 }, { "grad_norm": 0.032636020332574844, "learning_rate": 2.818094852389058e-06, "loss": 0.0026, "step": 89800 }, { "grad_norm": 0.053303588181734085, "learning_rate": 2.8126247936873516e-06, "loss": 0.0016, "step": 89810 }, { "grad_norm": 0.017972078174352646, "learning_rate": 2.8071598953260614e-06, "loss": 0.0012, "step": 89820 }, { "grad_norm": 0.053048763424158096, "learning_rate": 2.8017001579028035e-06, "loss": 0.002, "step": 89830 }, { "grad_norm": 0.03524443879723549, "learning_rate": 2.7962455820146617e-06, "loss": 0.0032, "step": 89840 }, { "grad_norm": 0.0245505478233099, "learning_rate": 2.7907961682581253e-06, "loss": 0.0028, "step": 89850 }, { "grad_norm": 0.0387805700302124, "learning_rate": 2.7853519172291453e-06, "loss": 0.0024, "step": 89860 }, { "grad_norm": 0.05433417111635208, "learning_rate": 2.7799128295230904e-06, "loss": 0.002, "step": 89870 }, { "grad_norm": 0.05696794018149376, "learning_rate": 2.774478905734784e-06, "loss": 0.0022, "step": 89880 }, { "grad_norm": 0.0456685833632946, "learning_rate": 2.7690501464584516e-06, "loss": 0.0043, "step": 89890 }, { "grad_norm": 0.0321134477853775, "learning_rate": 2.763626552287796e-06, "loss": 0.0027, "step": 89900 }, { "grad_norm": 0.08019303530454636, "learning_rate": 2.7582081238159083e-06, "loss": 0.0037, "step": 89910 }, { "grad_norm": 0.03477271646261215, "learning_rate": 2.7527948616353548e-06, "loss": 0.0026, "step": 89920 }, { "grad_norm": 0.027743013575673103, "learning_rate": 2.7473867663381226e-06, "loss": 0.0025, "step": 89930 }, { "grad_norm": 0.029549503698945045, "learning_rate": 2.7419838385156383e-06, "loss": 0.0031, "step": 89940 }, { "grad_norm": 0.07708154618740082, "learning_rate": 2.7365860787587407e-06, "loss": 0.0028, "step": 89950 }, { "grad_norm": 0.04556615650653839, "learning_rate": 2.7311934876577295e-06, "loss": 0.0028, "step": 89960 }, { "grad_norm": 0.018952537328004837, "learning_rate": 2.725806065802339e-06, "loss": 0.0023, "step": 89970 }, { "grad_norm": 0.03512080758810043, "learning_rate": 2.720423813781714e-06, "loss": 0.0018, "step": 89980 }, { "grad_norm": 0.024314699694514275, "learning_rate": 2.715046732184462e-06, "loss": 0.0023, "step": 89990 }, { "grad_norm": 0.021687079221010208, "learning_rate": 2.7096748215986013e-06, "loss": 0.0037, "step": 90000 }, { "grad_norm": 0.022068260237574577, "learning_rate": 2.704308082611612e-06, "loss": 0.0036, "step": 90010 }, { "grad_norm": 0.02729140967130661, "learning_rate": 2.698946515810374e-06, "loss": 0.0037, "step": 90020 }, { "grad_norm": 0.020975550636649132, "learning_rate": 2.6935901217812367e-06, "loss": 0.0018, "step": 90030 }, { "grad_norm": 0.07077853381633759, "learning_rate": 2.688238901109952e-06, "loss": 0.003, "step": 90040 }, { "grad_norm": 0.0678316131234169, "learning_rate": 2.68289285438173e-06, "loss": 0.0023, "step": 90050 }, { "grad_norm": 0.048526931554079056, "learning_rate": 2.6775519821812033e-06, "loss": 0.0029, "step": 90060 }, { "grad_norm": 0.03773389756679535, "learning_rate": 2.672216285092449e-06, "loss": 0.0021, "step": 90070 }, { "grad_norm": 0.057566188275814056, "learning_rate": 2.666885763698962e-06, "loss": 0.002, "step": 90080 }, { "grad_norm": 0.0560627207159996, "learning_rate": 2.661560418583692e-06, "loss": 0.0018, "step": 90090 }, { "grad_norm": 0.07141503691673279, "learning_rate": 2.656240250328995e-06, "loss": 0.0032, "step": 90100 }, { "grad_norm": 0.04715292155742645, "learning_rate": 2.650925259516679e-06, "loss": 0.0027, "step": 90110 }, { "grad_norm": 0.018386326730251312, "learning_rate": 2.645615446727995e-06, "loss": 0.0015, "step": 90120 }, { "grad_norm": 0.03284139558672905, "learning_rate": 2.6403108125436115e-06, "loss": 0.0016, "step": 90130 }, { "grad_norm": 0.06102212518453598, "learning_rate": 2.6350113575436264e-06, "loss": 0.0027, "step": 90140 }, { "grad_norm": 0.052400991320610046, "learning_rate": 2.629717082307598e-06, "loss": 0.0022, "step": 90150 }, { "grad_norm": 0.048436086624860764, "learning_rate": 2.6244279874144793e-06, "loss": 0.0039, "step": 90160 }, { "grad_norm": 0.04392494633793831, "learning_rate": 2.6191440734426862e-06, "loss": 0.0014, "step": 90170 }, { "grad_norm": 0.022534621879458427, "learning_rate": 2.613865340970062e-06, "loss": 0.0018, "step": 90180 }, { "grad_norm": 0.03662675619125366, "learning_rate": 2.608591790573889e-06, "loss": 0.0016, "step": 90190 }, { "grad_norm": 0.02815156802535057, "learning_rate": 2.603323422830861e-06, "loss": 0.002, "step": 90200 }, { "grad_norm": 0.02637694589793682, "learning_rate": 2.5980602383171183e-06, "loss": 0.0026, "step": 90210 }, { "grad_norm": 0.05856594443321228, "learning_rate": 2.5928022376082496e-06, "loss": 0.0017, "step": 90220 }, { "grad_norm": 0.02380967140197754, "learning_rate": 2.5875494212792505e-06, "loss": 0.0028, "step": 90230 }, { "grad_norm": 0.08325907588005066, "learning_rate": 2.5823017899045565e-06, "loss": 0.0038, "step": 90240 }, { "grad_norm": 0.03662474453449249, "learning_rate": 2.577059344058047e-06, "loss": 0.002, "step": 90250 }, { "grad_norm": 0.030417297035455704, "learning_rate": 2.5718220843130415e-06, "loss": 0.0025, "step": 90260 }, { "grad_norm": 0.06806480139493942, "learning_rate": 2.5665900112422535e-06, "loss": 0.0033, "step": 90270 }, { "grad_norm": 0.017974456772208214, "learning_rate": 2.561363125417876e-06, "loss": 0.0022, "step": 90280 }, { "grad_norm": 0.03708492964506149, "learning_rate": 2.556141427411496e-06, "loss": 0.0024, "step": 90290 }, { "grad_norm": 0.03920460119843483, "learning_rate": 2.550924917794156e-06, "loss": 0.0019, "step": 90300 }, { "grad_norm": 0.06250012665987015, "learning_rate": 2.545713597136329e-06, "loss": 0.0029, "step": 90310 }, { "grad_norm": 0.08178427815437317, "learning_rate": 2.5405074660079298e-06, "loss": 0.0035, "step": 90320 }, { "grad_norm": 0.08579503744840622, "learning_rate": 2.5353065249782647e-06, "loss": 0.0021, "step": 90330 }, { "grad_norm": 0.0534047931432724, "learning_rate": 2.530110774616129e-06, "loss": 0.0025, "step": 90340 }, { "grad_norm": 0.04507967084646225, "learning_rate": 2.524920215489701e-06, "loss": 0.0025, "step": 90350 }, { "grad_norm": 0.03966628015041351, "learning_rate": 2.519734848166627e-06, "loss": 0.0027, "step": 90360 }, { "grad_norm": 0.029278289526700974, "learning_rate": 2.514554673213948e-06, "loss": 0.0022, "step": 90370 }, { "grad_norm": 0.0595962293446064, "learning_rate": 2.5093796911981936e-06, "loss": 0.0033, "step": 90380 }, { "grad_norm": 0.020648781210184097, "learning_rate": 2.5042099026852672e-06, "loss": 0.0013, "step": 90390 }, { "grad_norm": 0.06467917561531067, "learning_rate": 2.499045308240544e-06, "loss": 0.0034, "step": 90400 }, { "grad_norm": 0.0506049320101738, "learning_rate": 2.4938859084288003e-06, "loss": 0.0016, "step": 90410 }, { "grad_norm": 0.06927073746919632, "learning_rate": 2.4887317038142734e-06, "loss": 0.003, "step": 90420 }, { "grad_norm": 0.017959125339984894, "learning_rate": 2.4835826949606124e-06, "loss": 0.0025, "step": 90430 }, { "grad_norm": 0.0674341693520546, "learning_rate": 2.4784388824309167e-06, "loss": 0.0025, "step": 90440 }, { "grad_norm": 0.048317767679691315, "learning_rate": 2.473300266787687e-06, "loss": 0.0029, "step": 90450 }, { "grad_norm": 0.06361183524131775, "learning_rate": 2.46816684859289e-06, "loss": 0.002, "step": 90460 }, { "grad_norm": 0.028017597272992134, "learning_rate": 2.4630386284079043e-06, "loss": 0.0016, "step": 90470 }, { "grad_norm": 0.08085998147726059, "learning_rate": 2.4579156067935427e-06, "loss": 0.0038, "step": 90480 }, { "grad_norm": 0.024723056703805923, "learning_rate": 2.452797784310046e-06, "loss": 0.0018, "step": 90490 }, { "grad_norm": 0.035015035420656204, "learning_rate": 2.447685161517105e-06, "loss": 0.0016, "step": 90500 }, { "grad_norm": 0.024170059710741043, "learning_rate": 2.4425777389738224e-06, "loss": 0.0019, "step": 90510 }, { "grad_norm": 0.06066220998764038, "learning_rate": 2.43747551723873e-06, "loss": 0.0031, "step": 90520 }, { "grad_norm": 0.0351778119802475, "learning_rate": 2.4323784968698136e-06, "loss": 0.004, "step": 90530 }, { "grad_norm": 0.029451929032802582, "learning_rate": 2.4272866784244565e-06, "loss": 0.0018, "step": 90540 }, { "grad_norm": 0.024998942390084267, "learning_rate": 2.422200062459518e-06, "loss": 0.0024, "step": 90550 }, { "grad_norm": 0.07614462077617645, "learning_rate": 2.41711864953123e-06, "loss": 0.0041, "step": 90560 }, { "grad_norm": 0.055534638464450836, "learning_rate": 2.412042440195322e-06, "loss": 0.0052, "step": 90570 }, { "grad_norm": 0.045939184725284576, "learning_rate": 2.4069714350068993e-06, "loss": 0.0054, "step": 90580 }, { "grad_norm": 0.03618434816598892, "learning_rate": 2.4019056345205294e-06, "loss": 0.0025, "step": 90590 }, { "grad_norm": 0.05499684065580368, "learning_rate": 2.396845039290191e-06, "loss": 0.0021, "step": 90600 }, { "grad_norm": 0.015885455533862114, "learning_rate": 2.3917896498693194e-06, "loss": 0.0028, "step": 90610 }, { "grad_norm": 0.05747031047940254, "learning_rate": 2.3867394668107335e-06, "loss": 0.0031, "step": 90620 }, { "grad_norm": 0.039515845477581024, "learning_rate": 2.3816944906667527e-06, "loss": 0.0017, "step": 90630 }, { "grad_norm": 0.031737808138132095, "learning_rate": 2.3766547219890634e-06, "loss": 0.0026, "step": 90640 }, { "grad_norm": 0.04213859513401985, "learning_rate": 2.371620161328819e-06, "loss": 0.0027, "step": 90650 }, { "grad_norm": 0.035508524626493454, "learning_rate": 2.366590809236574e-06, "loss": 0.0033, "step": 90660 }, { "grad_norm": 0.07310141623020172, "learning_rate": 2.3615666662623494e-06, "loss": 0.002, "step": 90670 }, { "grad_norm": 0.03656483441591263, "learning_rate": 2.3565477329555664e-06, "loss": 0.0023, "step": 90680 }, { "grad_norm": 0.023008324205875397, "learning_rate": 2.3515340098650975e-06, "loss": 0.0021, "step": 90690 }, { "grad_norm": 0.03405215963721275, "learning_rate": 2.3465254975392258e-06, "loss": 0.002, "step": 90700 }, { "grad_norm": 0.045755334198474884, "learning_rate": 2.3415221965256805e-06, "loss": 0.0022, "step": 90710 }, { "grad_norm": 0.024586988613009453, "learning_rate": 2.3365241073716237e-06, "loss": 0.0022, "step": 90720 }, { "grad_norm": 0.022731274366378784, "learning_rate": 2.3315312306236183e-06, "loss": 0.0025, "step": 90730 }, { "grad_norm": 0.06221138313412666, "learning_rate": 2.326543566827699e-06, "loss": 0.0023, "step": 90740 }, { "grad_norm": 0.04514839127659798, "learning_rate": 2.3215611165292862e-06, "loss": 0.0035, "step": 90750 }, { "grad_norm": 0.019620543345808983, "learning_rate": 2.3165838802732765e-06, "loss": 0.0035, "step": 90760 }, { "grad_norm": 0.0175140593200922, "learning_rate": 2.311611858603957e-06, "loss": 0.0027, "step": 90770 }, { "grad_norm": 0.03213964030146599, "learning_rate": 2.3066450520650763e-06, "loss": 0.0019, "step": 90780 }, { "grad_norm": 0.016325082629919052, "learning_rate": 2.3016834611997773e-06, "loss": 0.0041, "step": 90790 }, { "grad_norm": 0.060285784304142, "learning_rate": 2.296727086550665e-06, "loss": 0.0042, "step": 90800 }, { "grad_norm": 0.06308981776237488, "learning_rate": 2.2917759286597507e-06, "loss": 0.0038, "step": 90810 }, { "grad_norm": 0.04864254966378212, "learning_rate": 2.2868299880685006e-06, "loss": 0.002, "step": 90820 }, { "grad_norm": 0.06630564481019974, "learning_rate": 2.2818892653177827e-06, "loss": 0.0029, "step": 90830 }, { "grad_norm": 0.028850218281149864, "learning_rate": 2.2769537609479196e-06, "loss": 0.0024, "step": 90840 }, { "grad_norm": 0.02100553549826145, "learning_rate": 2.2720234754986357e-06, "loss": 0.0017, "step": 90850 }, { "grad_norm": 0.031035957857966423, "learning_rate": 2.2670984095091108e-06, "loss": 0.0018, "step": 90860 }, { "grad_norm": 0.05698789283633232, "learning_rate": 2.262178563517936e-06, "loss": 0.0025, "step": 90870 }, { "grad_norm": 0.06973445415496826, "learning_rate": 2.257263938063137e-06, "loss": 0.0041, "step": 90880 }, { "grad_norm": 0.06817272305488586, "learning_rate": 2.2523545336821715e-06, "loss": 0.0038, "step": 90890 }, { "grad_norm": 0.03796526417136192, "learning_rate": 2.247450350911939e-06, "loss": 0.0031, "step": 90900 }, { "grad_norm": 0.06835540384054184, "learning_rate": 2.2425513902887328e-06, "loss": 0.0062, "step": 90910 }, { "grad_norm": 0.03738715499639511, "learning_rate": 2.2376576523483015e-06, "loss": 0.0025, "step": 90920 }, { "grad_norm": 0.02607736364006996, "learning_rate": 2.232769137625829e-06, "loss": 0.0024, "step": 90930 }, { "grad_norm": 0.03835476189851761, "learning_rate": 2.227885846655903e-06, "loss": 0.0028, "step": 90940 }, { "grad_norm": 0.053935643285512924, "learning_rate": 2.2230077799725523e-06, "loss": 0.003, "step": 90950 }, { "grad_norm": 0.034104738384485245, "learning_rate": 2.2181349381092396e-06, "loss": 0.002, "step": 90960 }, { "grad_norm": 0.025582024827599525, "learning_rate": 2.2132673215988552e-06, "loss": 0.0016, "step": 90970 }, { "grad_norm": 0.03729652613401413, "learning_rate": 2.2084049309737065e-06, "loss": 0.0023, "step": 90980 }, { "grad_norm": 0.0368199422955513, "learning_rate": 2.2035477667655513e-06, "loss": 0.0019, "step": 90990 }, { "grad_norm": 0.020631015300750732, "learning_rate": 2.1986958295055315e-06, "loss": 0.0017, "step": 91000 }, { "grad_norm": 0.05756578966975212, "learning_rate": 2.193849119724284e-06, "loss": 0.0019, "step": 91010 }, { "grad_norm": 0.06127852201461792, "learning_rate": 2.1890076379518175e-06, "loss": 0.0025, "step": 91020 }, { "grad_norm": 0.03690226748585701, "learning_rate": 2.1841713847175925e-06, "loss": 0.0018, "step": 91030 }, { "grad_norm": 0.02807183563709259, "learning_rate": 2.1793403605504968e-06, "loss": 0.0026, "step": 91040 }, { "grad_norm": 0.022765984758734703, "learning_rate": 2.1745145659788413e-06, "loss": 0.0025, "step": 91050 }, { "grad_norm": 0.05884197726845741, "learning_rate": 2.1696940015303645e-06, "loss": 0.0026, "step": 91060 }, { "grad_norm": 0.044733256101608276, "learning_rate": 2.1648786677322397e-06, "loss": 0.0041, "step": 91070 }, { "grad_norm": 0.07928960025310516, "learning_rate": 2.1600685651110673e-06, "loss": 0.0029, "step": 91080 }, { "grad_norm": 0.04078712314367294, "learning_rate": 2.1552636941928717e-06, "loss": 0.0024, "step": 91090 }, { "grad_norm": 0.02637302130460739, "learning_rate": 2.1504640555031043e-06, "loss": 0.0031, "step": 91100 }, { "grad_norm": 0.05887063592672348, "learning_rate": 2.1456696495666506e-06, "loss": 0.0033, "step": 91110 }, { "grad_norm": 0.036552608013153076, "learning_rate": 2.1408804769078084e-06, "loss": 0.0014, "step": 91120 }, { "grad_norm": 0.031934063881635666, "learning_rate": 2.136096538050325e-06, "loss": 0.0038, "step": 91130 }, { "grad_norm": 0.04209659621119499, "learning_rate": 2.13131783351736e-06, "loss": 0.0019, "step": 91140 }, { "grad_norm": 0.06749007105827332, "learning_rate": 2.126544363831512e-06, "loss": 0.0024, "step": 91150 }, { "grad_norm": 0.026442186906933784, "learning_rate": 2.121776129514791e-06, "loss": 0.0018, "step": 91160 }, { "grad_norm": 0.03517116606235504, "learning_rate": 2.1170131310886523e-06, "loss": 0.0029, "step": 91170 }, { "grad_norm": 0.044888440519571304, "learning_rate": 2.112255369073968e-06, "loss": 0.0025, "step": 91180 }, { "grad_norm": 0.061233386397361755, "learning_rate": 2.1075028439910328e-06, "loss": 0.0022, "step": 91190 }, { "grad_norm": 0.06149179860949516, "learning_rate": 2.1027555563595813e-06, "loss": 0.0022, "step": 91200 }, { "grad_norm": 0.02034894749522209, "learning_rate": 2.0980135066987706e-06, "loss": 0.0025, "step": 91210 }, { "grad_norm": 0.060466207563877106, "learning_rate": 2.093276695527191e-06, "loss": 0.0027, "step": 91220 }, { "grad_norm": 0.024695049971342087, "learning_rate": 2.0885451233628396e-06, "loss": 0.0044, "step": 91230 }, { "grad_norm": 0.03861813247203827, "learning_rate": 2.0838187907231633e-06, "loss": 0.0035, "step": 91240 }, { "grad_norm": 0.03695095330476761, "learning_rate": 2.0790976981250153e-06, "loss": 0.0031, "step": 91250 }, { "grad_norm": 0.05086292698979378, "learning_rate": 2.0743818460847e-06, "loss": 0.0034, "step": 91260 }, { "grad_norm": 0.031934622675180435, "learning_rate": 2.069671235117926e-06, "loss": 0.0025, "step": 91270 }, { "grad_norm": 0.04086238145828247, "learning_rate": 2.0649658657398484e-06, "loss": 0.0022, "step": 91280 }, { "grad_norm": 0.05212155729532242, "learning_rate": 2.060265738465034e-06, "loss": 0.0017, "step": 91290 }, { "grad_norm": 0.01924266666173935, "learning_rate": 2.055570853807487e-06, "loss": 0.0017, "step": 91300 }, { "grad_norm": 0.022289343178272247, "learning_rate": 2.0508812122806196e-06, "loss": 0.0033, "step": 91310 }, { "grad_norm": 0.022194795310497284, "learning_rate": 2.046196814397294e-06, "loss": 0.0017, "step": 91320 }, { "grad_norm": 0.034535638988018036, "learning_rate": 2.0415176606697896e-06, "loss": 0.002, "step": 91330 }, { "grad_norm": 0.03912574052810669, "learning_rate": 2.0368437516098126e-06, "loss": 0.0031, "step": 91340 }, { "grad_norm": 0.06044400483369827, "learning_rate": 2.032175087728483e-06, "loss": 0.0026, "step": 91350 }, { "grad_norm": 0.03138083219528198, "learning_rate": 2.027511669536375e-06, "loss": 0.0019, "step": 91360 }, { "grad_norm": 0.023450462147593498, "learning_rate": 2.0228534975434643e-06, "loss": 0.0033, "step": 91370 }, { "grad_norm": 0.041796278208494186, "learning_rate": 2.0182005722591548e-06, "loss": 0.0019, "step": 91380 }, { "grad_norm": 0.0547272190451622, "learning_rate": 2.0135528941922944e-06, "loss": 0.0026, "step": 91390 }, { "grad_norm": 0.03428599610924721, "learning_rate": 2.008910463851149e-06, "loss": 0.0018, "step": 91400 }, { "grad_norm": 0.03449595347046852, "learning_rate": 2.0042732817433963e-06, "loss": 0.0017, "step": 91410 }, { "grad_norm": 0.04764179512858391, "learning_rate": 1.999641348376158e-06, "loss": 0.0027, "step": 91420 }, { "grad_norm": 0.0474570132791996, "learning_rate": 1.9950146642559787e-06, "loss": 0.0025, "step": 91430 }, { "grad_norm": 0.02594583109021187, "learning_rate": 1.990393229888815e-06, "loss": 0.0017, "step": 91440 }, { "grad_norm": 0.03366832807660103, "learning_rate": 1.9857770457800684e-06, "loss": 0.0044, "step": 91450 }, { "grad_norm": 0.03345818445086479, "learning_rate": 1.9811661124345517e-06, "loss": 0.0024, "step": 91460 }, { "grad_norm": 0.061606939882040024, "learning_rate": 1.976560430356522e-06, "loss": 0.0048, "step": 91470 }, { "grad_norm": 0.03460020571947098, "learning_rate": 1.9719600000496383e-06, "loss": 0.0024, "step": 91480 }, { "grad_norm": 0.06726010888814926, "learning_rate": 1.967364822017004e-06, "loss": 0.0029, "step": 91490 }, { "grad_norm": 0.029251690953969955, "learning_rate": 1.962774896761127e-06, "loss": 0.0018, "step": 91500 }, { "grad_norm": 0.03287065029144287, "learning_rate": 1.958190224783968e-06, "loss": 0.0015, "step": 91510 }, { "grad_norm": 0.03456360474228859, "learning_rate": 1.9536108065868975e-06, "loss": 0.002, "step": 91520 }, { "grad_norm": 0.03779974952340126, "learning_rate": 1.949036642670715e-06, "loss": 0.0045, "step": 91530 }, { "grad_norm": 0.026868557557463646, "learning_rate": 1.9444677335356377e-06, "loss": 0.0015, "step": 91540 }, { "grad_norm": 0.07127739489078522, "learning_rate": 1.939904079681326e-06, "loss": 0.0022, "step": 91550 }, { "grad_norm": 0.01622352935373783, "learning_rate": 1.935345681606837e-06, "loss": 0.0029, "step": 91560 }, { "grad_norm": 0.05348636209964752, "learning_rate": 1.930792539810694e-06, "loss": 0.0012, "step": 91570 }, { "grad_norm": 0.012411575764417648, "learning_rate": 1.9262446547907877e-06, "loss": 0.0022, "step": 91580 }, { "grad_norm": 0.027175836265087128, "learning_rate": 1.921702027044503e-06, "loss": 0.002, "step": 91590 }, { "grad_norm": 0.05754633992910385, "learning_rate": 1.917164657068593e-06, "loss": 0.0018, "step": 91600 }, { "grad_norm": 0.043592166155576706, "learning_rate": 1.912632545359272e-06, "loss": 0.0021, "step": 91610 }, { "grad_norm": 0.06619671732187271, "learning_rate": 1.9081056924121488e-06, "loss": 0.0031, "step": 91620 }, { "grad_norm": 0.07908129692077637, "learning_rate": 1.9035840987222887e-06, "loss": 0.0027, "step": 91630 }, { "grad_norm": 0.02860272489488125, "learning_rate": 1.8990677647841515e-06, "loss": 0.0034, "step": 91640 }, { "grad_norm": 0.01781284250319004, "learning_rate": 1.8945566910916534e-06, "loss": 0.0016, "step": 91650 }, { "grad_norm": 0.06898529082536697, "learning_rate": 1.8900508781381054e-06, "loss": 0.0031, "step": 91660 }, { "grad_norm": 0.01756206899881363, "learning_rate": 1.8855503264162633e-06, "loss": 0.0023, "step": 91670 }, { "grad_norm": 0.02795029990375042, "learning_rate": 1.8810550364182998e-06, "loss": 0.0026, "step": 91680 }, { "grad_norm": 0.035245683044195175, "learning_rate": 1.8765650086358112e-06, "loss": 0.0016, "step": 91690 }, { "grad_norm": 0.02450065314769745, "learning_rate": 1.8720802435598151e-06, "loss": 0.0029, "step": 91700 }, { "grad_norm": 0.03553939610719681, "learning_rate": 1.8676007416807694e-06, "loss": 0.0017, "step": 91710 }, { "grad_norm": 0.015133429318666458, "learning_rate": 1.8631265034885437e-06, "loss": 0.0035, "step": 91720 }, { "grad_norm": 0.03160567954182625, "learning_rate": 1.858657529472424e-06, "loss": 0.0035, "step": 91730 }, { "grad_norm": 0.0335959754884243, "learning_rate": 1.8541938201211473e-06, "loss": 0.0024, "step": 91740 }, { "grad_norm": 0.0549442321062088, "learning_rate": 1.8497353759228398e-06, "loss": 0.0029, "step": 91750 }, { "grad_norm": 0.04786621406674385, "learning_rate": 1.8452821973650835e-06, "loss": 0.0024, "step": 91760 }, { "grad_norm": 0.04155101627111435, "learning_rate": 1.840834284934856e-06, "loss": 0.0037, "step": 91770 }, { "grad_norm": 0.01744442619383335, "learning_rate": 1.8363916391185954e-06, "loss": 0.0016, "step": 91780 }, { "grad_norm": 0.07367733120918274, "learning_rate": 1.8319542604021245e-06, "loss": 0.0044, "step": 91790 }, { "grad_norm": 0.025067545473575592, "learning_rate": 1.8275221492707273e-06, "loss": 0.0039, "step": 91800 }, { "grad_norm": 0.024914855137467384, "learning_rate": 1.8230953062090717e-06, "loss": 0.0027, "step": 91810 }, { "grad_norm": 0.03560195490717888, "learning_rate": 1.8186737317012926e-06, "loss": 0.0031, "step": 91820 }, { "grad_norm": 0.03554365038871765, "learning_rate": 1.8142574262308977e-06, "loss": 0.0027, "step": 91830 }, { "grad_norm": 0.048723116517066956, "learning_rate": 1.8098463902808727e-06, "loss": 0.0039, "step": 91840 }, { "grad_norm": 0.02157318778336048, "learning_rate": 1.805440624333593e-06, "loss": 0.0014, "step": 91850 }, { "grad_norm": 0.026415430009365082, "learning_rate": 1.8010401288708734e-06, "loss": 0.0017, "step": 91860 }, { "grad_norm": 0.06695389747619629, "learning_rate": 1.7966449043739343e-06, "loss": 0.0026, "step": 91870 }, { "grad_norm": 0.09203029423952103, "learning_rate": 1.79225495132343e-06, "loss": 0.0021, "step": 91880 }, { "grad_norm": 0.05438876152038574, "learning_rate": 1.7878702701994487e-06, "loss": 0.0018, "step": 91890 }, { "grad_norm": 0.07754357904195786, "learning_rate": 1.783490861481496e-06, "loss": 0.0014, "step": 91900 }, { "grad_norm": 0.05091788247227669, "learning_rate": 1.7791167256484832e-06, "loss": 0.0022, "step": 91910 }, { "grad_norm": 0.060816776007413864, "learning_rate": 1.774747863178766e-06, "loss": 0.0037, "step": 91920 }, { "grad_norm": 0.05480858311057091, "learning_rate": 1.7703842745501186e-06, "loss": 0.0021, "step": 91930 }, { "grad_norm": 0.043746158480644226, "learning_rate": 1.7660259602397366e-06, "loss": 0.0019, "step": 91940 }, { "grad_norm": 0.06454399228096008, "learning_rate": 1.7616729207242388e-06, "loss": 0.0038, "step": 91950 }, { "grad_norm": 0.026783550158143044, "learning_rate": 1.7573251564796556e-06, "loss": 0.0021, "step": 91960 }, { "grad_norm": 0.037485282868146896, "learning_rate": 1.7529826679814676e-06, "loss": 0.0021, "step": 91970 }, { "grad_norm": 0.027764683589339256, "learning_rate": 1.7486454557045562e-06, "loss": 0.0044, "step": 91980 }, { "grad_norm": 0.032786544412374496, "learning_rate": 1.7443135201232364e-06, "loss": 0.0022, "step": 91990 }, { "grad_norm": 0.03761330619454384, "learning_rate": 1.7399868617112347e-06, "loss": 0.0028, "step": 92000 }, { "grad_norm": 0.03500337898731232, "learning_rate": 1.7356654809417171e-06, "loss": 0.0024, "step": 92010 }, { "grad_norm": 0.047052279114723206, "learning_rate": 1.7313493782872448e-06, "loss": 0.0048, "step": 92020 }, { "grad_norm": 0.024988235905766487, "learning_rate": 1.7270385542198453e-06, "loss": 0.0023, "step": 92030 }, { "grad_norm": 0.02545008435845375, "learning_rate": 1.7227330092109307e-06, "loss": 0.0029, "step": 92040 }, { "grad_norm": 0.07325705885887146, "learning_rate": 1.718432743731352e-06, "loss": 0.0044, "step": 92050 }, { "grad_norm": 0.05673721060156822, "learning_rate": 1.7141377582513718e-06, "loss": 0.0024, "step": 92060 }, { "grad_norm": 0.06007383018732071, "learning_rate": 1.709848053240698e-06, "loss": 0.0024, "step": 92070 }, { "grad_norm": 0.049948692321777344, "learning_rate": 1.7055636291684328e-06, "loss": 0.0061, "step": 92080 }, { "grad_norm": 0.05645494908094406, "learning_rate": 1.7012844865031185e-06, "loss": 0.002, "step": 92090 }, { "grad_norm": 0.01951483264565468, "learning_rate": 1.6970106257127138e-06, "loss": 0.0015, "step": 92100 }, { "grad_norm": 0.012077251449227333, "learning_rate": 1.6927420472646117e-06, "loss": 0.0024, "step": 92110 }, { "grad_norm": 0.049330953508615494, "learning_rate": 1.6884787516256106e-06, "loss": 0.0018, "step": 92120 }, { "grad_norm": 0.03947839140892029, "learning_rate": 1.6842207392619325e-06, "loss": 0.0028, "step": 92130 }, { "grad_norm": 0.03836270794272423, "learning_rate": 1.679968010639238e-06, "loss": 0.0035, "step": 92140 }, { "grad_norm": 0.020454229786992073, "learning_rate": 1.6757205662225829e-06, "loss": 0.0026, "step": 92150 }, { "grad_norm": 0.02988779917359352, "learning_rate": 1.6714784064764844e-06, "loss": 0.0017, "step": 92160 }, { "grad_norm": 0.07183799892663956, "learning_rate": 1.6672415318648383e-06, "loss": 0.0044, "step": 92170 }, { "grad_norm": 0.02429720014333725, "learning_rate": 1.6630099428510016e-06, "loss": 0.0022, "step": 92180 }, { "grad_norm": 0.052082087844610214, "learning_rate": 1.6587836398977153e-06, "loss": 0.0024, "step": 92190 }, { "grad_norm": 0.06503409892320633, "learning_rate": 1.6545626234671762e-06, "loss": 0.003, "step": 92200 }, { "grad_norm": 0.038251843303442, "learning_rate": 1.650346894020971e-06, "loss": 0.0031, "step": 92210 }, { "grad_norm": 0.029409626498818398, "learning_rate": 1.6461364520201472e-06, "loss": 0.0026, "step": 92220 }, { "grad_norm": 0.0241499412804842, "learning_rate": 1.6419312979251366e-06, "loss": 0.0042, "step": 92230 }, { "grad_norm": 0.04115753248333931, "learning_rate": 1.6377314321958214e-06, "loss": 0.0028, "step": 92240 }, { "grad_norm": 0.03730292618274689, "learning_rate": 1.6335368552914787e-06, "loss": 0.0027, "step": 92250 }, { "grad_norm": 0.05056682974100113, "learning_rate": 1.6293475676708358e-06, "loss": 0.0024, "step": 92260 }, { "grad_norm": 0.025086654350161552, "learning_rate": 1.6251635697920097e-06, "loss": 0.002, "step": 92270 }, { "grad_norm": 0.052878353744745255, "learning_rate": 1.6209848621125624e-06, "loss": 0.0026, "step": 92280 }, { "grad_norm": 0.015883613377809525, "learning_rate": 1.6168114450894778e-06, "loss": 0.0044, "step": 92290 }, { "grad_norm": 0.04980669915676117, "learning_rate": 1.612643319179158e-06, "loss": 0.0019, "step": 92300 }, { "grad_norm": 0.03752264380455017, "learning_rate": 1.6084804848374046e-06, "loss": 0.0034, "step": 92310 }, { "grad_norm": 0.024626726284623146, "learning_rate": 1.6043229425194817e-06, "loss": 0.0024, "step": 92320 }, { "grad_norm": 0.015693357214331627, "learning_rate": 1.6001706926800308e-06, "loss": 0.0034, "step": 92330 }, { "grad_norm": 0.033937159925699234, "learning_rate": 1.5960237357731444e-06, "loss": 0.002, "step": 92340 }, { "grad_norm": 0.06029747799038887, "learning_rate": 1.5918820722523264e-06, "loss": 0.0031, "step": 92350 }, { "grad_norm": 0.04507216066122055, "learning_rate": 1.587745702570509e-06, "loss": 0.0026, "step": 92360 }, { "grad_norm": 0.030342575162649155, "learning_rate": 1.5836146271800356e-06, "loss": 0.0017, "step": 92370 }, { "grad_norm": 0.049873240292072296, "learning_rate": 1.5794888465326728e-06, "loss": 0.0034, "step": 92380 }, { "grad_norm": 0.01567966677248478, "learning_rate": 1.5753683610796155e-06, "loss": 0.0019, "step": 92390 }, { "grad_norm": 0.04033595696091652, "learning_rate": 1.5712531712714528e-06, "loss": 0.002, "step": 92400 }, { "grad_norm": 0.04720183461904526, "learning_rate": 1.5671432775582474e-06, "loss": 0.0015, "step": 92410 }, { "grad_norm": 0.0258990116417408, "learning_rate": 1.563038680389428e-06, "loss": 0.0022, "step": 92420 }, { "grad_norm": 0.02666998840868473, "learning_rate": 1.5589393802138808e-06, "loss": 0.0027, "step": 92430 }, { "grad_norm": 0.04326043650507927, "learning_rate": 1.554845377479891e-06, "loss": 0.002, "step": 92440 }, { "grad_norm": 0.0332099013030529, "learning_rate": 1.550756672635173e-06, "loss": 0.0062, "step": 92450 }, { "grad_norm": 0.04893026128411293, "learning_rate": 1.5466732661268635e-06, "loss": 0.0028, "step": 92460 }, { "grad_norm": 0.021925529465079308, "learning_rate": 1.5425951584015163e-06, "loss": 0.0018, "step": 92470 }, { "grad_norm": 0.030797796323895454, "learning_rate": 1.5385223499051026e-06, "loss": 0.0035, "step": 92480 }, { "grad_norm": 0.12441935390233994, "learning_rate": 1.5344548410830328e-06, "loss": 0.0035, "step": 92490 }, { "grad_norm": 0.027558941394090652, "learning_rate": 1.5303926323801121e-06, "loss": 0.0025, "step": 92500 }, { "grad_norm": 0.04400058835744858, "learning_rate": 1.5263357242405795e-06, "loss": 0.002, "step": 92510 }, { "grad_norm": 0.04476713016629219, "learning_rate": 1.5222841171080915e-06, "loss": 0.0024, "step": 92520 }, { "grad_norm": 0.028358956798911095, "learning_rate": 1.5182378114257268e-06, "loss": 0.0033, "step": 92530 }, { "grad_norm": 0.05827516317367554, "learning_rate": 1.5141968076359869e-06, "loss": 0.002, "step": 92540 }, { "grad_norm": 0.025259729474782944, "learning_rate": 1.5101611061807908e-06, "loss": 0.0024, "step": 92550 }, { "grad_norm": 0.037007782608270645, "learning_rate": 1.506130707501463e-06, "loss": 0.0036, "step": 92560 }, { "grad_norm": 0.05581996589899063, "learning_rate": 1.5021056120387734e-06, "loss": 0.0029, "step": 92570 }, { "grad_norm": 0.04394061118364334, "learning_rate": 1.4980858202329084e-06, "loss": 0.0024, "step": 92580 }, { "grad_norm": 0.04529755562543869, "learning_rate": 1.4940713325234445e-06, "loss": 0.0024, "step": 92590 }, { "grad_norm": 0.013012135401368141, "learning_rate": 1.4900621493494137e-06, "loss": 0.0028, "step": 92600 }, { "grad_norm": 0.03667175769805908, "learning_rate": 1.4860582711492544e-06, "loss": 0.0023, "step": 92610 }, { "grad_norm": 0.04529518634080887, "learning_rate": 1.4820596983608325e-06, "loss": 0.0022, "step": 92620 }, { "grad_norm": 0.03510982543230057, "learning_rate": 1.4780664314214044e-06, "loss": 0.0023, "step": 92630 }, { "grad_norm": 0.03565097600221634, "learning_rate": 1.474078470767687e-06, "loss": 0.0029, "step": 92640 }, { "grad_norm": 0.059108126908540726, "learning_rate": 1.470095816835787e-06, "loss": 0.0034, "step": 92650 }, { "grad_norm": 0.015369156375527382, "learning_rate": 1.4661184700612395e-06, "loss": 0.0022, "step": 92660 }, { "grad_norm": 0.05173742026090622, "learning_rate": 1.4621464308790133e-06, "loss": 0.0019, "step": 92670 }, { "grad_norm": 0.017501475289463997, "learning_rate": 1.4581796997234775e-06, "loss": 0.0044, "step": 92680 }, { "grad_norm": 0.02326822280883789, "learning_rate": 1.4542182770284297e-06, "loss": 0.0029, "step": 92690 }, { "grad_norm": 0.056108258664608, "learning_rate": 1.450262163227084e-06, "loss": 0.0017, "step": 92700 }, { "grad_norm": 0.0582084096968174, "learning_rate": 1.4463113587520671e-06, "loss": 0.003, "step": 92710 }, { "grad_norm": 0.04988273233175278, "learning_rate": 1.4423658640354443e-06, "loss": 0.002, "step": 92720 }, { "grad_norm": 0.048684630542993546, "learning_rate": 1.438425679508687e-06, "loss": 0.0031, "step": 92730 }, { "grad_norm": 0.044997598975896835, "learning_rate": 1.4344908056026895e-06, "loss": 0.0023, "step": 92740 }, { "grad_norm": 0.031079772859811783, "learning_rate": 1.4305612427477521e-06, "loss": 0.0028, "step": 92750 }, { "grad_norm": 0.06083167716860771, "learning_rate": 1.42663699137362e-06, "loss": 0.0022, "step": 92760 }, { "grad_norm": 0.05421121045947075, "learning_rate": 1.4227180519094386e-06, "loss": 0.0024, "step": 92770 }, { "grad_norm": 0.03596322610974312, "learning_rate": 1.4188044247837706e-06, "loss": 0.0024, "step": 92780 }, { "grad_norm": 0.016854528337717056, "learning_rate": 1.4148961104246128e-06, "loss": 0.003, "step": 92790 }, { "grad_norm": 0.03656591475009918, "learning_rate": 1.4109931092593731e-06, "loss": 0.0024, "step": 92800 }, { "grad_norm": 0.030051080510020256, "learning_rate": 1.4070954217148658e-06, "loss": 0.0027, "step": 92810 }, { "grad_norm": 0.058751702308654785, "learning_rate": 1.4032030482173497e-06, "loss": 0.0051, "step": 92820 }, { "grad_norm": 0.04177498817443848, "learning_rate": 1.3993159891924901e-06, "loss": 0.0021, "step": 92830 }, { "grad_norm": 0.021335680037736893, "learning_rate": 1.3954342450653524e-06, "loss": 0.0017, "step": 92840 }, { "grad_norm": 0.03469911590218544, "learning_rate": 1.3915578162604582e-06, "loss": 0.0029, "step": 92850 }, { "grad_norm": 0.04733581468462944, "learning_rate": 1.3876867032017126e-06, "loss": 0.0022, "step": 92860 }, { "grad_norm": 0.07163247466087341, "learning_rate": 1.383820906312472e-06, "loss": 0.0037, "step": 92870 }, { "grad_norm": 0.047892991453409195, "learning_rate": 1.3799604260154752e-06, "loss": 0.0018, "step": 92880 }, { "grad_norm": 0.03188912943005562, "learning_rate": 1.376105262732913e-06, "loss": 0.0012, "step": 92890 }, { "grad_norm": 0.0336490236222744, "learning_rate": 1.37225541688637e-06, "loss": 0.0026, "step": 92900 }, { "grad_norm": 0.04030569642782211, "learning_rate": 1.3684108888968594e-06, "loss": 0.0016, "step": 92910 }, { "grad_norm": 0.01569226011633873, "learning_rate": 1.364571679184823e-06, "loss": 0.002, "step": 92920 }, { "grad_norm": 0.04970218241214752, "learning_rate": 1.3607377881701077e-06, "loss": 0.0022, "step": 92930 }, { "grad_norm": 0.05375358834862709, "learning_rate": 1.3569092162719787e-06, "loss": 0.0037, "step": 92940 }, { "grad_norm": 0.04173501208424568, "learning_rate": 1.353085963909123e-06, "loss": 0.0024, "step": 92950 }, { "grad_norm": 0.03329070657491684, "learning_rate": 1.3492680314996454e-06, "loss": 0.0036, "step": 92960 }, { "grad_norm": 0.023286402225494385, "learning_rate": 1.345455419461078e-06, "loss": 0.0022, "step": 92970 }, { "grad_norm": 0.04138324409723282, "learning_rate": 1.3416481282103431e-06, "loss": 0.0021, "step": 92980 }, { "grad_norm": 0.09411317110061646, "learning_rate": 1.337846158163819e-06, "loss": 0.0026, "step": 92990 }, { "grad_norm": 0.03275774419307709, "learning_rate": 1.334049509737273e-06, "loss": 0.0033, "step": 93000 }, { "grad_norm": 0.02544604241847992, "learning_rate": 1.3302581833459116e-06, "loss": 0.0021, "step": 93010 }, { "grad_norm": 0.01562860794365406, "learning_rate": 1.3264721794043366e-06, "loss": 0.0028, "step": 93020 }, { "grad_norm": 0.03515351191163063, "learning_rate": 1.3226914983265836e-06, "loss": 0.0015, "step": 93030 }, { "grad_norm": 0.026114752516150475, "learning_rate": 1.3189161405260997e-06, "loss": 0.0017, "step": 93040 }, { "grad_norm": 0.053670138120651245, "learning_rate": 1.3151461064157656e-06, "loss": 0.0029, "step": 93050 }, { "grad_norm": 0.05104440450668335, "learning_rate": 1.3113813964078459e-06, "loss": 0.003, "step": 93060 }, { "grad_norm": 0.036843929439783096, "learning_rate": 1.3076220109140614e-06, "loss": 0.0018, "step": 93070 }, { "grad_norm": 0.07107522338628769, "learning_rate": 1.3038679503455275e-06, "loss": 0.0017, "step": 93080 }, { "grad_norm": 0.037763625383377075, "learning_rate": 1.3001192151127773e-06, "loss": 0.0031, "step": 93090 }, { "grad_norm": 0.03190051391720772, "learning_rate": 1.2963758056257657e-06, "loss": 0.0023, "step": 93100 }, { "grad_norm": 0.04754325747489929, "learning_rate": 1.2926377222938768e-06, "loss": 0.0034, "step": 93110 }, { "grad_norm": 0.030189912766218185, "learning_rate": 1.2889049655259e-06, "loss": 0.0024, "step": 93120 }, { "grad_norm": 0.023549260571599007, "learning_rate": 1.2851775357300312e-06, "loss": 0.0023, "step": 93130 }, { "grad_norm": 0.055080924183130264, "learning_rate": 1.281455433313916e-06, "loss": 0.0026, "step": 93140 }, { "grad_norm": 0.018231885507702827, "learning_rate": 1.2777386586845796e-06, "loss": 0.0034, "step": 93150 }, { "grad_norm": 0.07043773680925369, "learning_rate": 1.2740272122484964e-06, "loss": 0.0042, "step": 93160 }, { "grad_norm": 0.06367177516222, "learning_rate": 1.2703210944115308e-06, "loss": 0.0035, "step": 93170 }, { "grad_norm": 0.0536356195807457, "learning_rate": 1.2666203055789916e-06, "loss": 0.0026, "step": 93180 }, { "grad_norm": 0.01995687186717987, "learning_rate": 1.2629248461555832e-06, "loss": 0.002, "step": 93190 }, { "grad_norm": 0.031583271920681, "learning_rate": 1.2592347165454431e-06, "loss": 0.0022, "step": 93200 }, { "grad_norm": 0.029585687443614006, "learning_rate": 1.2555499171521102e-06, "loss": 0.0023, "step": 93210 }, { "grad_norm": 0.05236365273594856, "learning_rate": 1.251870448378556e-06, "loss": 0.0032, "step": 93220 }, { "grad_norm": 0.04395947605371475, "learning_rate": 1.2481963106271478e-06, "loss": 0.0032, "step": 93230 }, { "grad_norm": 0.03210834041237831, "learning_rate": 1.2445275042997084e-06, "loss": 0.0031, "step": 93240 }, { "grad_norm": 0.023023808375000954, "learning_rate": 1.2408640297974283e-06, "loss": 0.0024, "step": 93250 }, { "grad_norm": 0.048439960926771164, "learning_rate": 1.2372058875209535e-06, "loss": 0.0025, "step": 93260 }, { "grad_norm": 0.07110247015953064, "learning_rate": 1.2335530778703252e-06, "loss": 0.0025, "step": 93270 }, { "grad_norm": 0.052048180252313614, "learning_rate": 1.2299056012450182e-06, "loss": 0.0023, "step": 93280 }, { "grad_norm": 0.0327272042632103, "learning_rate": 1.2262634580439026e-06, "loss": 0.0019, "step": 93290 }, { "grad_norm": 0.017379119992256165, "learning_rate": 1.2226266486652926e-06, "loss": 0.0016, "step": 93300 }, { "grad_norm": 0.08653682470321655, "learning_rate": 1.2189951735068927e-06, "loss": 0.0044, "step": 93310 }, { "grad_norm": 0.05088384822010994, "learning_rate": 1.2153690329658407e-06, "loss": 0.0018, "step": 93320 }, { "grad_norm": 0.03495948761701584, "learning_rate": 1.2117482274386859e-06, "loss": 0.0025, "step": 93330 }, { "grad_norm": 0.02501119300723076, "learning_rate": 1.2081327573213953e-06, "loss": 0.0019, "step": 93340 }, { "grad_norm": 0.017634987831115723, "learning_rate": 1.2045226230093465e-06, "loss": 0.0018, "step": 93350 }, { "grad_norm": 0.038696788251399994, "learning_rate": 1.2009178248973296e-06, "loss": 0.0018, "step": 93360 }, { "grad_norm": 0.05131662264466286, "learning_rate": 1.1973183633795849e-06, "loss": 0.0033, "step": 93370 }, { "grad_norm": 0.052869800478219986, "learning_rate": 1.193724238849725e-06, "loss": 0.003, "step": 93380 }, { "grad_norm": 0.07561641186475754, "learning_rate": 1.190135451700808e-06, "loss": 0.002, "step": 93390 }, { "grad_norm": 0.019936637952923775, "learning_rate": 1.186552002325292e-06, "loss": 0.0014, "step": 93400 }, { "grad_norm": 0.04481561854481697, "learning_rate": 1.1829738911150578e-06, "loss": 0.0019, "step": 93410 }, { "grad_norm": 0.03759532421827316, "learning_rate": 1.1794011184613985e-06, "loss": 0.0036, "step": 93420 }, { "grad_norm": 0.13387662172317505, "learning_rate": 1.17583368475504e-06, "loss": 0.002, "step": 93430 }, { "grad_norm": 0.047731395810842514, "learning_rate": 1.1722715903860982e-06, "loss": 0.0016, "step": 93440 }, { "grad_norm": 0.02887100912630558, "learning_rate": 1.168714835744128e-06, "loss": 0.0019, "step": 93450 }, { "grad_norm": 0.024662967771291733, "learning_rate": 1.165163421218085e-06, "loss": 0.0031, "step": 93460 }, { "grad_norm": 0.03267456591129303, "learning_rate": 1.1616173471963477e-06, "loss": 0.0024, "step": 93470 }, { "grad_norm": 0.028701089322566986, "learning_rate": 1.1580766140667109e-06, "loss": 0.0016, "step": 93480 }, { "grad_norm": 0.046266768127679825, "learning_rate": 1.1545412222163764e-06, "loss": 0.0027, "step": 93490 }, { "grad_norm": 0.03423180431127548, "learning_rate": 1.1510111720319795e-06, "loss": 0.0017, "step": 93500 }, { "grad_norm": 0.022929098457098007, "learning_rate": 1.1474864638995663e-06, "loss": 0.0017, "step": 93510 }, { "grad_norm": 0.04631279036402702, "learning_rate": 1.1439670982045735e-06, "loss": 0.0025, "step": 93520 }, { "grad_norm": 0.064286969602108, "learning_rate": 1.1404530753318877e-06, "loss": 0.0042, "step": 93530 }, { "grad_norm": 0.05035454407334328, "learning_rate": 1.136944395665801e-06, "loss": 0.0025, "step": 93540 }, { "grad_norm": 0.027424829080700874, "learning_rate": 1.1334410595900014e-06, "loss": 0.0039, "step": 93550 }, { "grad_norm": 0.019136467948555946, "learning_rate": 1.129943067487621e-06, "loss": 0.0041, "step": 93560 }, { "grad_norm": 0.044836416840553284, "learning_rate": 1.126450419741193e-06, "loss": 0.0019, "step": 93570 }, { "grad_norm": 0.037678416818380356, "learning_rate": 1.1229631167326671e-06, "loss": 0.0018, "step": 93580 }, { "grad_norm": 0.021191978827118874, "learning_rate": 1.1194811588434052e-06, "loss": 0.0016, "step": 93590 }, { "grad_norm": 0.057460423558950424, "learning_rate": 1.1160045464542023e-06, "loss": 0.0017, "step": 93600 }, { "grad_norm": 0.04633593559265137, "learning_rate": 1.1125332799452325e-06, "loss": 0.0034, "step": 93610 }, { "grad_norm": 0.0466233529150486, "learning_rate": 1.109067359696131e-06, "loss": 0.0027, "step": 93620 }, { "grad_norm": 0.03682717680931091, "learning_rate": 1.1056067860859166e-06, "loss": 0.0025, "step": 93630 }, { "grad_norm": 0.044498492032289505, "learning_rate": 1.102151559493031e-06, "loss": 0.0058, "step": 93640 }, { "grad_norm": 0.022688573226332664, "learning_rate": 1.0987016802953332e-06, "loss": 0.0039, "step": 93650 }, { "grad_norm": 0.04557139798998833, "learning_rate": 1.0952571488701048e-06, "loss": 0.0027, "step": 93660 }, { "grad_norm": 0.042708221822977066, "learning_rate": 1.0918179655940163e-06, "loss": 0.0022, "step": 93670 }, { "grad_norm": 0.04846669360995293, "learning_rate": 1.088384130843184e-06, "loss": 0.002, "step": 93680 }, { "grad_norm": 0.06124762445688248, "learning_rate": 1.0849556449931296e-06, "loss": 0.0025, "step": 93690 }, { "grad_norm": 0.0450027696788311, "learning_rate": 1.0815325084187866e-06, "loss": 0.002, "step": 93700 }, { "grad_norm": 0.024758264422416687, "learning_rate": 1.0781147214944943e-06, "loss": 0.0028, "step": 93710 }, { "grad_norm": 0.017157675698399544, "learning_rate": 1.0747022845940313e-06, "loss": 0.0026, "step": 93720 }, { "grad_norm": 0.045251455157995224, "learning_rate": 1.0712951980905605e-06, "loss": 0.0018, "step": 93730 }, { "grad_norm": 0.0512908473610878, "learning_rate": 1.0678934623566839e-06, "loss": 0.0039, "step": 93740 }, { "grad_norm": 0.053296975791454315, "learning_rate": 1.0644970777644091e-06, "loss": 0.0023, "step": 93750 }, { "grad_norm": 0.027480196207761765, "learning_rate": 1.0611060446851673e-06, "loss": 0.0018, "step": 93760 }, { "grad_norm": 0.05488453805446625, "learning_rate": 1.0577203634897836e-06, "loss": 0.0042, "step": 93770 }, { "grad_norm": 0.01902029477059841, "learning_rate": 1.0543400345485178e-06, "loss": 0.0011, "step": 93780 }, { "grad_norm": 0.05720706284046173, "learning_rate": 1.0509650582310404e-06, "loss": 0.0025, "step": 93790 }, { "grad_norm": 0.02316754311323166, "learning_rate": 1.0475954349064287e-06, "loss": 0.0023, "step": 93800 }, { "grad_norm": 0.03619835153222084, "learning_rate": 1.0442311649431825e-06, "loss": 0.0016, "step": 93810 }, { "grad_norm": 0.08162708580493927, "learning_rate": 1.0408722487092071e-06, "loss": 0.0022, "step": 93820 }, { "grad_norm": 0.046734705567359924, "learning_rate": 1.0375186865718424e-06, "loss": 0.0026, "step": 93830 }, { "grad_norm": 0.027310866862535477, "learning_rate": 1.0341704788978113e-06, "loss": 0.0022, "step": 93840 }, { "grad_norm": 0.019304461777210236, "learning_rate": 1.0308276260532879e-06, "loss": 0.002, "step": 93850 }, { "grad_norm": 0.01971939578652382, "learning_rate": 1.0274901284038241e-06, "loss": 0.004, "step": 93860 }, { "grad_norm": 0.08136700093746185, "learning_rate": 1.0241579863144114e-06, "loss": 0.0027, "step": 93870 }, { "grad_norm": 0.0173347070813179, "learning_rate": 1.0208312001494413e-06, "loss": 0.0017, "step": 93880 }, { "grad_norm": 0.056607507169246674, "learning_rate": 1.0175097702727453e-06, "loss": 0.0026, "step": 93890 }, { "grad_norm": 0.06480780988931656, "learning_rate": 1.0141936970475275e-06, "loss": 0.0029, "step": 93900 }, { "grad_norm": 0.04882150515913963, "learning_rate": 1.0108829808364417e-06, "loss": 0.0026, "step": 93910 }, { "grad_norm": 0.03355151042342186, "learning_rate": 1.0075776220015376e-06, "loss": 0.0049, "step": 93920 }, { "grad_norm": 0.02234506979584694, "learning_rate": 1.0042776209042816e-06, "loss": 0.0015, "step": 93930 }, { "grad_norm": 0.0696575939655304, "learning_rate": 1.000982977905568e-06, "loss": 0.0028, "step": 93940 }, { "grad_norm": 0.057423096150159836, "learning_rate": 9.976936933656867e-07, "loss": 0.0029, "step": 93950 }, { "grad_norm": 0.06974713504314423, "learning_rate": 9.944097676443442e-07, "loss": 0.0065, "step": 93960 }, { "grad_norm": 0.05879400670528412, "learning_rate": 9.911312011006757e-07, "loss": 0.0024, "step": 93970 }, { "grad_norm": 0.02491292729973793, "learning_rate": 9.878579940932109e-07, "loss": 0.0024, "step": 93980 }, { "grad_norm": 0.035996343940496445, "learning_rate": 9.84590146979908e-07, "loss": 0.0017, "step": 93990 }, { "grad_norm": 0.04159514978528023, "learning_rate": 9.813276601181365e-07, "loss": 0.0016, "step": 94000 }, { "grad_norm": 0.02705877274274826, "learning_rate": 9.780705338646724e-07, "loss": 0.0016, "step": 94010 }, { "grad_norm": 0.05581314116716385, "learning_rate": 9.748187685757083e-07, "loss": 0.0016, "step": 94020 }, { "grad_norm": 0.0661696344614029, "learning_rate": 9.715723646068543e-07, "loss": 0.002, "step": 94030 }, { "grad_norm": 0.03209679201245308, "learning_rate": 9.68331322313143e-07, "loss": 0.0044, "step": 94040 }, { "grad_norm": 0.01905694045126438, "learning_rate": 9.650956420489909e-07, "loss": 0.0033, "step": 94050 }, { "grad_norm": 0.058087125420570374, "learning_rate": 9.618653241682595e-07, "loss": 0.0028, "step": 94060 }, { "grad_norm": 0.06834718585014343, "learning_rate": 9.586403690242053e-07, "loss": 0.0027, "step": 94070 }, { "grad_norm": 0.060906171798706055, "learning_rate": 9.554207769695122e-07, "loss": 0.003, "step": 94080 }, { "grad_norm": 0.018970174714922905, "learning_rate": 9.522065483562659e-07, "loss": 0.002, "step": 94090 }, { "grad_norm": 0.04434806481003761, "learning_rate": 9.489976835359681e-07, "loss": 0.0021, "step": 94100 }, { "grad_norm": 0.028919586911797523, "learning_rate": 9.457941828595329e-07, "loss": 0.0016, "step": 94110 }, { "grad_norm": 0.05570441111922264, "learning_rate": 9.425960466772965e-07, "loss": 0.0049, "step": 94120 }, { "grad_norm": 0.025591615587472916, "learning_rate": 9.394032753390014e-07, "loss": 0.0026, "step": 94130 }, { "grad_norm": 0.016385655850172043, "learning_rate": 9.362158691938017e-07, "loss": 0.0019, "step": 94140 }, { "grad_norm": 0.05609631538391113, "learning_rate": 9.330338285902685e-07, "loss": 0.0034, "step": 94150 }, { "grad_norm": 0.03456393629312515, "learning_rate": 9.298571538763845e-07, "loss": 0.002, "step": 94160 }, { "grad_norm": 0.020235665142536163, "learning_rate": 9.266858453995441e-07, "loss": 0.0022, "step": 94170 }, { "grad_norm": 0.043272148817777634, "learning_rate": 9.235199035065645e-07, "loss": 0.0017, "step": 94180 }, { "grad_norm": 0.04114726930856705, "learning_rate": 9.203593285436574e-07, "loss": 0.0024, "step": 94190 }, { "grad_norm": 0.03810876980423927, "learning_rate": 9.172041208564686e-07, "loss": 0.0015, "step": 94200 }, { "grad_norm": 0.08039947599172592, "learning_rate": 9.140542807900332e-07, "loss": 0.0032, "step": 94210 }, { "grad_norm": 0.03703958913683891, "learning_rate": 9.109098086888313e-07, "loss": 0.0017, "step": 94220 }, { "grad_norm": 0.02449285425245762, "learning_rate": 9.077707048967265e-07, "loss": 0.0033, "step": 94230 }, { "grad_norm": 0.07745257019996643, "learning_rate": 9.046369697570057e-07, "loss": 0.0027, "step": 94240 }, { "grad_norm": 0.014275058172643185, "learning_rate": 9.015086036123721e-07, "loss": 0.0026, "step": 94250 }, { "grad_norm": 0.038709789514541626, "learning_rate": 8.983856068049413e-07, "loss": 0.0021, "step": 94260 }, { "grad_norm": 0.05077232047915459, "learning_rate": 8.952679796762398e-07, "loss": 0.0019, "step": 94270 }, { "grad_norm": 0.03695070743560791, "learning_rate": 8.921557225671951e-07, "loss": 0.0029, "step": 94280 }, { "grad_norm": 0.027015164494514465, "learning_rate": 8.890488358181792e-07, "loss": 0.0052, "step": 94290 }, { "grad_norm": 0.02042742446064949, "learning_rate": 8.859473197689372e-07, "loss": 0.0021, "step": 94300 }, { "grad_norm": 0.02085852064192295, "learning_rate": 8.828511747586588e-07, "loss": 0.0016, "step": 94310 }, { "grad_norm": 0.054701440036296844, "learning_rate": 8.797604011259287e-07, "loss": 0.0027, "step": 94320 }, { "grad_norm": 0.04409419000148773, "learning_rate": 8.766749992087487e-07, "loss": 0.0017, "step": 94330 }, { "grad_norm": 0.027256522327661514, "learning_rate": 8.735949693445378e-07, "loss": 0.0026, "step": 94340 }, { "grad_norm": 0.057111915200948715, "learning_rate": 8.705203118701211e-07, "loss": 0.0024, "step": 94350 }, { "grad_norm": 0.042511794716119766, "learning_rate": 8.674510271217295e-07, "loss": 0.0014, "step": 94360 }, { "grad_norm": 0.05567888543009758, "learning_rate": 8.643871154350336e-07, "loss": 0.0037, "step": 94370 }, { "grad_norm": 0.036207813769578934, "learning_rate": 8.613285771450818e-07, "loss": 0.0013, "step": 94380 }, { "grad_norm": 0.026675764471292496, "learning_rate": 8.582754125863623e-07, "loss": 0.0035, "step": 94390 }, { "grad_norm": 0.07488755881786346, "learning_rate": 8.55227622092758e-07, "loss": 0.004, "step": 94400 }, { "grad_norm": 0.03664228692650795, "learning_rate": 8.521852059975743e-07, "loss": 0.003, "step": 94410 }, { "grad_norm": 0.04169555380940437, "learning_rate": 8.491481646335231e-07, "loss": 0.0018, "step": 94420 }, { "grad_norm": 0.030653608962893486, "learning_rate": 8.46116498332733e-07, "loss": 0.0019, "step": 94430 }, { "grad_norm": 0.027522960677742958, "learning_rate": 8.430902074267333e-07, "loss": 0.0018, "step": 94440 }, { "grad_norm": 0.07837618142366409, "learning_rate": 8.400692922464925e-07, "loss": 0.0027, "step": 94450 }, { "grad_norm": 0.029083536937832832, "learning_rate": 8.370537531223577e-07, "loss": 0.0023, "step": 94460 }, { "grad_norm": 0.04913517087697983, "learning_rate": 8.340435903841092e-07, "loss": 0.0019, "step": 94470 }, { "grad_norm": 0.016004465520381927, "learning_rate": 8.310388043609341e-07, "loss": 0.0029, "step": 94480 }, { "grad_norm": 0.019406190142035484, "learning_rate": 8.280393953814303e-07, "loss": 0.0041, "step": 94490 }, { "grad_norm": 0.10558506101369858, "learning_rate": 8.250453637736077e-07, "loss": 0.0024, "step": 94500 }, { "grad_norm": 0.06285184621810913, "learning_rate": 8.22056709864899e-07, "loss": 0.0024, "step": 94510 }, { "grad_norm": 0.040307771414518356, "learning_rate": 8.190734339821204e-07, "loss": 0.0014, "step": 94520 }, { "grad_norm": 0.06285180151462555, "learning_rate": 8.160955364515333e-07, "loss": 0.0017, "step": 94530 }, { "grad_norm": 0.02893708273768425, "learning_rate": 8.131230175987936e-07, "loss": 0.0018, "step": 94540 }, { "grad_norm": 0.03561922535300255, "learning_rate": 8.101558777489693e-07, "loss": 0.0027, "step": 94550 }, { "grad_norm": 0.09446302801370621, "learning_rate": 8.071941172265396e-07, "loss": 0.0024, "step": 94560 }, { "grad_norm": 0.0532538965344429, "learning_rate": 8.042377363553955e-07, "loss": 0.0024, "step": 94570 }, { "grad_norm": 0.026169992983341217, "learning_rate": 8.01286735458856e-07, "loss": 0.0025, "step": 94580 }, { "grad_norm": 0.06007642298936844, "learning_rate": 7.983411148596298e-07, "loss": 0.0027, "step": 94590 }, { "grad_norm": 0.05560761317610741, "learning_rate": 7.954008748798425e-07, "loss": 0.0018, "step": 94600 }, { "grad_norm": 0.0275247935205698, "learning_rate": 7.924660158410369e-07, "loss": 0.004, "step": 94610 }, { "grad_norm": 0.037666212767362595, "learning_rate": 7.895365380641728e-07, "loss": 0.0029, "step": 94620 }, { "grad_norm": 0.09149029105901718, "learning_rate": 7.866124418695941e-07, "loss": 0.0019, "step": 94630 }, { "grad_norm": 0.0550488606095314, "learning_rate": 7.83693727577095e-07, "loss": 0.0026, "step": 94640 }, { "grad_norm": 0.037226442247629166, "learning_rate": 7.807803955058535e-07, "loss": 0.0023, "step": 94650 }, { "grad_norm": 0.031984228640794754, "learning_rate": 7.778724459744702e-07, "loss": 0.0016, "step": 94660 }, { "grad_norm": 0.011824831366539001, "learning_rate": 7.749698793009464e-07, "loss": 0.0032, "step": 94670 }, { "grad_norm": 0.05010351911187172, "learning_rate": 7.720726958027169e-07, "loss": 0.0023, "step": 94680 }, { "grad_norm": 0.04884021729230881, "learning_rate": 7.691808957965951e-07, "loss": 0.0017, "step": 94690 }, { "grad_norm": 0.03898537904024124, "learning_rate": 7.662944795988336e-07, "loss": 0.0021, "step": 94700 }, { "grad_norm": 0.07578708231449127, "learning_rate": 7.634134475250854e-07, "loss": 0.002, "step": 94710 }, { "grad_norm": 0.016184642910957336, "learning_rate": 7.605377998904262e-07, "loss": 0.0017, "step": 94720 }, { "grad_norm": 0.021425483748316765, "learning_rate": 7.5766753700931e-07, "loss": 0.0014, "step": 94730 }, { "grad_norm": 0.02641419880092144, "learning_rate": 7.54802659195647e-07, "loss": 0.0043, "step": 94740 }, { "grad_norm": 0.04287385940551758, "learning_rate": 7.519431667627253e-07, "loss": 0.0025, "step": 94750 }, { "grad_norm": 0.05030911788344383, "learning_rate": 7.490890600232503e-07, "loss": 0.0036, "step": 94760 }, { "grad_norm": 0.02519001066684723, "learning_rate": 7.462403392893502e-07, "loss": 0.002, "step": 94770 }, { "grad_norm": 0.020711621269583702, "learning_rate": 7.433970048725536e-07, "loss": 0.0041, "step": 94780 }, { "grad_norm": 0.02013748325407505, "learning_rate": 7.405590570838062e-07, "loss": 0.0039, "step": 94790 }, { "grad_norm": 0.018963726237416267, "learning_rate": 7.377264962334596e-07, "loss": 0.0024, "step": 94800 }, { "grad_norm": 0.08246072381734848, "learning_rate": 7.348993226312773e-07, "loss": 0.0029, "step": 94810 }, { "grad_norm": 0.022057678550481796, "learning_rate": 7.32077536586434e-07, "loss": 0.0016, "step": 94820 }, { "grad_norm": 0.0694965124130249, "learning_rate": 7.292611384075221e-07, "loss": 0.0018, "step": 94830 }, { "grad_norm": 0.03743930533528328, "learning_rate": 7.264501284025338e-07, "loss": 0.003, "step": 94840 }, { "grad_norm": 0.04492906481027603, "learning_rate": 7.236445068788789e-07, "loss": 0.003, "step": 94850 }, { "grad_norm": 0.026473987847566605, "learning_rate": 7.20844274143373e-07, "loss": 0.0018, "step": 94860 }, { "grad_norm": 0.05019211396574974, "learning_rate": 7.180494305022489e-07, "loss": 0.0026, "step": 94870 }, { "grad_norm": 0.036939121782779694, "learning_rate": 7.152599762611456e-07, "loss": 0.0015, "step": 94880 }, { "grad_norm": 0.07354892790317535, "learning_rate": 7.124759117251079e-07, "loss": 0.0036, "step": 94890 }, { "grad_norm": 0.08357948064804077, "learning_rate": 7.096972371986033e-07, "loss": 0.0029, "step": 94900 }, { "grad_norm": 0.051168933510780334, "learning_rate": 7.069239529855054e-07, "loss": 0.0043, "step": 94910 }, { "grad_norm": 0.044410113245248795, "learning_rate": 7.041560593890939e-07, "loss": 0.0029, "step": 94920 }, { "grad_norm": 0.021181700751185417, "learning_rate": 7.013935567120599e-07, "loss": 0.0022, "step": 94930 }, { "grad_norm": 0.02480686455965042, "learning_rate": 6.986364452565009e-07, "loss": 0.0024, "step": 94940 }, { "grad_norm": 0.05250087007880211, "learning_rate": 6.95884725323942e-07, "loss": 0.002, "step": 94950 }, { "grad_norm": 0.05960728973150253, "learning_rate": 6.931383972152982e-07, "loss": 0.0023, "step": 94960 }, { "grad_norm": 0.028528088703751564, "learning_rate": 6.903974612309127e-07, "loss": 0.0024, "step": 94970 }, { "grad_norm": 0.030342958867549896, "learning_rate": 6.876619176705179e-07, "loss": 0.0015, "step": 94980 }, { "grad_norm": 0.04558439180254936, "learning_rate": 6.8493176683328e-07, "loss": 0.0018, "step": 94990 }, { "grad_norm": 0.06076262891292572, "learning_rate": 6.822070090177657e-07, "loss": 0.0024, "step": 95000 }, { "grad_norm": 0.06371033191680908, "learning_rate": 6.794876445219367e-07, "loss": 0.0025, "step": 95010 }, { "grad_norm": 0.023050697520375252, "learning_rate": 6.767736736431885e-07, "loss": 0.0021, "step": 95020 }, { "grad_norm": 0.01892201229929924, "learning_rate": 6.740650966783113e-07, "loss": 0.0019, "step": 95030 }, { "grad_norm": 0.051319703459739685, "learning_rate": 6.713619139235239e-07, "loss": 0.0018, "step": 95040 }, { "grad_norm": 0.032422661781311035, "learning_rate": 6.686641256744286e-07, "loss": 0.0019, "step": 95050 }, { "grad_norm": 0.02481389045715332, "learning_rate": 6.659717322260562e-07, "loss": 0.0037, "step": 95060 }, { "grad_norm": 0.021156756207346916, "learning_rate": 6.632847338728377e-07, "loss": 0.0024, "step": 95070 }, { "grad_norm": 0.024148201569914818, "learning_rate": 6.60603130908627e-07, "loss": 0.0022, "step": 95080 }, { "grad_norm": 0.04597136750817299, "learning_rate": 6.579269236266783e-07, "loss": 0.0022, "step": 95090 }, { "grad_norm": 0.020484203472733498, "learning_rate": 6.552561123196632e-07, "loss": 0.0023, "step": 95100 }, { "grad_norm": 0.0213839802891016, "learning_rate": 6.525906972796425e-07, "loss": 0.0018, "step": 95110 }, { "grad_norm": 0.04193875193595886, "learning_rate": 6.499306787981108e-07, "loss": 0.0022, "step": 95120 }, { "grad_norm": 0.025427015498280525, "learning_rate": 6.472760571659631e-07, "loss": 0.0021, "step": 95130 }, { "grad_norm": 0.02432117983698845, "learning_rate": 6.446268326735061e-07, "loss": 0.0019, "step": 95140 }, { "grad_norm": 0.0379570834338665, "learning_rate": 6.41983005610458e-07, "loss": 0.003, "step": 95150 }, { "grad_norm": 0.03573670983314514, "learning_rate": 6.393445762659378e-07, "loss": 0.0034, "step": 95160 }, { "grad_norm": 0.04969397187232971, "learning_rate": 6.36711544928481e-07, "loss": 0.0046, "step": 95170 }, { "grad_norm": 0.05041061341762543, "learning_rate": 6.340839118860298e-07, "loss": 0.0017, "step": 95180 }, { "grad_norm": 0.05245620012283325, "learning_rate": 6.314616774259485e-07, "loss": 0.0024, "step": 95190 }, { "grad_norm": 0.02608642727136612, "learning_rate": 6.288448418349858e-07, "loss": 0.0023, "step": 95200 }, { "grad_norm": 0.023458331823349, "learning_rate": 6.262334053993235e-07, "loss": 0.0024, "step": 95210 }, { "grad_norm": 0.03659585118293762, "learning_rate": 6.236273684045501e-07, "loss": 0.0035, "step": 95220 }, { "grad_norm": 0.026641221717000008, "learning_rate": 6.210267311356488e-07, "loss": 0.0016, "step": 95230 }, { "grad_norm": 0.06777594983577728, "learning_rate": 6.184314938770197e-07, "loss": 0.0034, "step": 95240 }, { "grad_norm": 0.01178498100489378, "learning_rate": 6.158416569124858e-07, "loss": 0.0019, "step": 95250 }, { "grad_norm": 0.06961125135421753, "learning_rate": 6.132572205252541e-07, "loss": 0.0025, "step": 95260 }, { "grad_norm": 0.0866493284702301, "learning_rate": 6.106781849979648e-07, "loss": 0.0025, "step": 95270 }, { "grad_norm": 0.06824842840433121, "learning_rate": 6.081045506126482e-07, "loss": 0.0033, "step": 95280 }, { "grad_norm": 0.04090472683310509, "learning_rate": 6.055363176507678e-07, "loss": 0.0019, "step": 95290 }, { "grad_norm": 0.02990131825208664, "learning_rate": 6.029734863931602e-07, "loss": 0.0028, "step": 95300 }, { "grad_norm": 0.02139567770063877, "learning_rate": 6.004160571201179e-07, "loss": 0.0022, "step": 95310 }, { "grad_norm": 0.037973761558532715, "learning_rate": 5.978640301112947e-07, "loss": 0.0043, "step": 95320 }, { "grad_norm": 0.058316443115472794, "learning_rate": 5.9531740564579e-07, "loss": 0.0015, "step": 95330 }, { "grad_norm": 0.027439989149570465, "learning_rate": 5.927761840020918e-07, "loss": 0.0019, "step": 95340 }, { "grad_norm": 0.01647316850721836, "learning_rate": 5.902403654581113e-07, "loss": 0.0047, "step": 95350 }, { "grad_norm": 0.013032328337430954, "learning_rate": 5.877099502911542e-07, "loss": 0.0017, "step": 95360 }, { "grad_norm": 0.03314751014113426, "learning_rate": 5.851849387779496e-07, "loss": 0.0021, "step": 95370 }, { "grad_norm": 0.036011967808008194, "learning_rate": 5.826653311946262e-07, "loss": 0.003, "step": 95380 }, { "grad_norm": 0.033189184963703156, "learning_rate": 5.801511278167249e-07, "loss": 0.0016, "step": 95390 }, { "grad_norm": 0.02043783850967884, "learning_rate": 5.776423289191868e-07, "loss": 0.0021, "step": 95400 }, { "grad_norm": 0.021251065656542778, "learning_rate": 5.751389347763869e-07, "loss": 0.0018, "step": 95410 }, { "grad_norm": 0.052819475531578064, "learning_rate": 5.726409456620841e-07, "loss": 0.0026, "step": 95420 }, { "grad_norm": 0.044343143701553345, "learning_rate": 5.701483618494541e-07, "loss": 0.0025, "step": 95430 }, { "grad_norm": 0.03151831775903702, "learning_rate": 5.67661183611079e-07, "loss": 0.004, "step": 95440 }, { "grad_norm": 0.03280593827366829, "learning_rate": 5.651794112189579e-07, "loss": 0.0024, "step": 95450 }, { "grad_norm": 0.03776874020695686, "learning_rate": 5.627030449444903e-07, "loss": 0.0017, "step": 95460 }, { "grad_norm": 0.03161877393722534, "learning_rate": 5.60232085058493e-07, "loss": 0.0015, "step": 95470 }, { "grad_norm": 0.027770912274718285, "learning_rate": 5.577665318311776e-07, "loss": 0.0016, "step": 95480 }, { "grad_norm": 0.016628200188279152, "learning_rate": 5.553063855321838e-07, "loss": 0.0022, "step": 95490 }, { "grad_norm": 0.020950671285390854, "learning_rate": 5.528516464305467e-07, "loss": 0.002, "step": 95500 }, { "grad_norm": 0.06918846815824509, "learning_rate": 5.504023147947068e-07, "loss": 0.0033, "step": 95510 }, { "grad_norm": 0.05135662853717804, "learning_rate": 5.479583908925279e-07, "loss": 0.002, "step": 95520 }, { "grad_norm": 0.01927942782640457, "learning_rate": 5.455198749912626e-07, "loss": 0.0018, "step": 95530 }, { "grad_norm": 0.028649086132645607, "learning_rate": 5.430867673575979e-07, "loss": 0.0017, "step": 95540 }, { "grad_norm": 0.040145646780729294, "learning_rate": 5.406590682575985e-07, "loss": 0.004, "step": 95550 }, { "grad_norm": 0.03039918839931488, "learning_rate": 5.382367779567687e-07, "loss": 0.002, "step": 95560 }, { "grad_norm": 0.02727297507226467, "learning_rate": 5.358198967200023e-07, "loss": 0.0024, "step": 95570 }, { "grad_norm": 0.033438973128795624, "learning_rate": 5.334084248115989e-07, "loss": 0.0017, "step": 95580 }, { "grad_norm": 0.03234800696372986, "learning_rate": 5.310023624952754e-07, "loss": 0.002, "step": 95590 }, { "grad_norm": 0.02328271046280861, "learning_rate": 5.286017100341656e-07, "loss": 0.0018, "step": 95600 }, { "grad_norm": 0.01887989602982998, "learning_rate": 5.262064676907874e-07, "loss": 0.0034, "step": 95610 }, { "grad_norm": 0.01904233545064926, "learning_rate": 5.238166357270924e-07, "loss": 0.0023, "step": 95620 }, { "grad_norm": 0.0454237163066864, "learning_rate": 5.214322144044215e-07, "loss": 0.0034, "step": 95630 }, { "grad_norm": 0.03705010563135147, "learning_rate": 5.190532039835327e-07, "loss": 0.0019, "step": 95640 }, { "grad_norm": 0.0435451865196228, "learning_rate": 5.166796047245903e-07, "loss": 0.0022, "step": 95650 }, { "grad_norm": 0.011418160982429981, "learning_rate": 5.143114168871699e-07, "loss": 0.0015, "step": 95660 }, { "grad_norm": 0.031247088685631752, "learning_rate": 5.119486407302532e-07, "loss": 0.0014, "step": 95670 }, { "grad_norm": 0.07055950164794922, "learning_rate": 5.095912765122335e-07, "loss": 0.0018, "step": 95680 }, { "grad_norm": 0.019844043999910355, "learning_rate": 5.07239324490899e-07, "loss": 0.0017, "step": 95690 }, { "grad_norm": 0.055808596312999725, "learning_rate": 5.048927849234608e-07, "loss": 0.0026, "step": 95700 }, { "grad_norm": 0.05285599082708359, "learning_rate": 5.025516580665358e-07, "loss": 0.0038, "step": 95710 }, { "grad_norm": 0.015031269751489162, "learning_rate": 5.002159441761412e-07, "loss": 0.0019, "step": 95720 }, { "grad_norm": 0.03215087950229645, "learning_rate": 4.978856435077062e-07, "loss": 0.0026, "step": 95730 }, { "grad_norm": 0.09138131141662598, "learning_rate": 4.955607563160769e-07, "loss": 0.002, "step": 95740 }, { "grad_norm": 0.023576144129037857, "learning_rate": 4.932412828554944e-07, "loss": 0.0015, "step": 95750 }, { "grad_norm": 0.03134693205356598, "learning_rate": 4.909272233796114e-07, "loss": 0.0023, "step": 95760 }, { "grad_norm": 0.05801130831241608, "learning_rate": 4.88618578141492e-07, "loss": 0.0023, "step": 95770 }, { "grad_norm": 0.047134485095739365, "learning_rate": 4.863153473936011e-07, "loss": 0.0032, "step": 95780 }, { "grad_norm": 0.02168581821024418, "learning_rate": 4.840175313878259e-07, "loss": 0.0031, "step": 95790 }, { "grad_norm": 0.014146354980766773, "learning_rate": 4.817251303754433e-07, "loss": 0.0033, "step": 95800 }, { "grad_norm": 0.018572846427559853, "learning_rate": 4.794381446071583e-07, "loss": 0.0026, "step": 95810 }, { "grad_norm": 0.03271034359931946, "learning_rate": 4.771565743330597e-07, "loss": 0.0019, "step": 95820 }, { "grad_norm": 0.06453793495893478, "learning_rate": 4.7488041980266463e-07, "loss": 0.0024, "step": 95830 }, { "grad_norm": 0.027773376554250717, "learning_rate": 4.726096812648795e-07, "loss": 0.0019, "step": 95840 }, { "grad_norm": 0.034297458827495575, "learning_rate": 4.703443589680445e-07, "loss": 0.002, "step": 95850 }, { "grad_norm": 0.01800035871565342, "learning_rate": 4.6808445315987826e-07, "loss": 0.0017, "step": 95860 }, { "grad_norm": 0.045136719942092896, "learning_rate": 4.658299640875274e-07, "loss": 0.0023, "step": 95870 }, { "grad_norm": 0.0862639918923378, "learning_rate": 4.635808919975393e-07, "loss": 0.0019, "step": 95880 }, { "grad_norm": 0.03450366482138634, "learning_rate": 4.6133723713586706e-07, "loss": 0.0014, "step": 95890 }, { "grad_norm": 0.04152646288275719, "learning_rate": 4.590989997478756e-07, "loss": 0.0025, "step": 95900 }, { "grad_norm": 0.04895003139972687, "learning_rate": 4.5686618007833027e-07, "loss": 0.0019, "step": 95910 }, { "grad_norm": 0.041797760874032974, "learning_rate": 4.54638778371419e-07, "loss": 0.0014, "step": 95920 }, { "grad_norm": 0.03271431103348732, "learning_rate": 4.524167948707192e-07, "loss": 0.0027, "step": 95930 }, { "grad_norm": 0.04261993616819382, "learning_rate": 4.5020022981921986e-07, "loss": 0.0018, "step": 95940 }, { "grad_norm": 0.08866984397172928, "learning_rate": 4.47989083459327e-07, "loss": 0.0043, "step": 95950 }, { "grad_norm": 0.03793081268668175, "learning_rate": 4.457833560328528e-07, "loss": 0.0048, "step": 95960 }, { "grad_norm": 0.019634978845715523, "learning_rate": 4.4358304778100434e-07, "loss": 0.0031, "step": 95970 }, { "grad_norm": 0.020207645371556282, "learning_rate": 4.413881589444058e-07, "loss": 0.0027, "step": 95980 }, { "grad_norm": 0.025360463187098503, "learning_rate": 4.391986897630873e-07, "loss": 0.0035, "step": 95990 }, { "grad_norm": 0.06165565550327301, "learning_rate": 4.370146404764908e-07, "loss": 0.003, "step": 96000 }, { "grad_norm": 0.030179284512996674, "learning_rate": 4.3483601132345285e-07, "loss": 0.003, "step": 96010 }, { "grad_norm": 0.08491349220275879, "learning_rate": 4.32662802542233e-07, "loss": 0.0045, "step": 96020 }, { "grad_norm": 0.04710880666971207, "learning_rate": 4.3049501437047444e-07, "loss": 0.0018, "step": 96030 }, { "grad_norm": 0.044550396502017975, "learning_rate": 4.283326470452653e-07, "loss": 0.0025, "step": 96040 }, { "grad_norm": 0.04573266580700874, "learning_rate": 4.2617570080306647e-07, "loss": 0.0016, "step": 96050 }, { "grad_norm": 0.025772443041205406, "learning_rate": 4.2402417587975585e-07, "loss": 0.0017, "step": 96060 }, { "grad_norm": 0.07036142796278, "learning_rate": 4.218780725106286e-07, "loss": 0.0023, "step": 96070 }, { "grad_norm": 0.041034482419490814, "learning_rate": 4.197373909303748e-07, "loss": 0.0026, "step": 96080 }, { "grad_norm": 0.05197301134467125, "learning_rate": 4.17602131373096e-07, "loss": 0.002, "step": 96090 }, { "grad_norm": 0.03308779373764992, "learning_rate": 4.154722940723055e-07, "loss": 0.0025, "step": 96100 }, { "grad_norm": 0.012864352203905582, "learning_rate": 4.1334787926091136e-07, "loss": 0.0021, "step": 96110 }, { "grad_norm": 0.01537255197763443, "learning_rate": 4.1122888717124444e-07, "loss": 0.0022, "step": 96120 }, { "grad_norm": 0.05602341145277023, "learning_rate": 4.091153180350249e-07, "loss": 0.0025, "step": 96130 }, { "grad_norm": 0.061740729957818985, "learning_rate": 4.0700717208340125e-07, "loss": 0.0029, "step": 96140 }, { "grad_norm": 0.047477222979068756, "learning_rate": 4.049044495469112e-07, "loss": 0.0018, "step": 96150 }, { "grad_norm": 0.034983791410923004, "learning_rate": 4.0280715065549866e-07, "loss": 0.0036, "step": 96160 }, { "grad_norm": 0.031068259850144386, "learning_rate": 4.0071527563853575e-07, "loss": 0.0023, "step": 96170 }, { "grad_norm": 0.03039800003170967, "learning_rate": 3.9862882472477823e-07, "loss": 0.0016, "step": 96180 }, { "grad_norm": 0.03356829285621643, "learning_rate": 3.9654779814239375e-07, "loss": 0.0023, "step": 96190 }, { "grad_norm": 0.047496289014816284, "learning_rate": 3.944721961189668e-07, "loss": 0.0027, "step": 96200 }, { "grad_norm": 0.01921624131500721, "learning_rate": 3.9240201888147697e-07, "loss": 0.002, "step": 96210 }, { "grad_norm": 0.08661466091871262, "learning_rate": 3.9033726665632096e-07, "loss": 0.0022, "step": 96220 }, { "grad_norm": 0.022727366536855698, "learning_rate": 3.882779396692959e-07, "loss": 0.0023, "step": 96230 }, { "grad_norm": 0.051505714654922485, "learning_rate": 3.8622403814559947e-07, "loss": 0.0017, "step": 96240 }, { "grad_norm": 0.04144386574625969, "learning_rate": 3.841755623098575e-07, "loss": 0.0016, "step": 96250 }, { "grad_norm": 0.024755502119660378, "learning_rate": 3.821325123860742e-07, "loss": 0.0025, "step": 96260 }, { "grad_norm": 0.02684479020535946, "learning_rate": 3.800948885976818e-07, "loss": 0.0019, "step": 96270 }, { "grad_norm": 0.05272011086344719, "learning_rate": 3.780626911675078e-07, "loss": 0.0039, "step": 96280 }, { "grad_norm": 0.05484010651707649, "learning_rate": 3.76035920317791e-07, "loss": 0.0029, "step": 96290 }, { "grad_norm": 0.019867779687047005, "learning_rate": 3.740145762701819e-07, "loss": 0.0018, "step": 96300 }, { "grad_norm": 0.038068901747465134, "learning_rate": 3.7199865924572585e-07, "loss": 0.0031, "step": 96310 }, { "grad_norm": 0.054331663995981216, "learning_rate": 3.699881694648799e-07, "loss": 0.0037, "step": 96320 }, { "grad_norm": 0.05030248314142227, "learning_rate": 3.679831071475126e-07, "loss": 0.005, "step": 96330 }, { "grad_norm": 0.031838804483413696, "learning_rate": 3.65983472512893e-07, "loss": 0.0031, "step": 96340 }, { "grad_norm": 0.045569516718387604, "learning_rate": 3.6398926577969616e-07, "loss": 0.0016, "step": 96350 }, { "grad_norm": 0.015550428070127964, "learning_rate": 3.620004871660032e-07, "loss": 0.0027, "step": 96360 }, { "grad_norm": 0.04467138648033142, "learning_rate": 3.600171368893124e-07, "loss": 0.0031, "step": 96370 }, { "grad_norm": 0.03927168622612953, "learning_rate": 3.580392151665113e-07, "loss": 0.0019, "step": 96380 }, { "grad_norm": 0.023950766772031784, "learning_rate": 3.5606672221391024e-07, "loss": 0.0017, "step": 96390 }, { "grad_norm": 0.016080724075436592, "learning_rate": 3.540996582472089e-07, "loss": 0.0016, "step": 96400 }, { "grad_norm": 0.056090712547302246, "learning_rate": 3.521380234815297e-07, "loss": 0.0028, "step": 96410 }, { "grad_norm": 0.058485500514507294, "learning_rate": 3.5018181813138985e-07, "loss": 0.0022, "step": 96420 }, { "grad_norm": 0.026606621220707893, "learning_rate": 3.482310424107238e-07, "loss": 0.0032, "step": 96430 }, { "grad_norm": 0.016015982255339622, "learning_rate": 3.4628569653285537e-07, "loss": 0.0044, "step": 96440 }, { "grad_norm": 0.022559484466910362, "learning_rate": 3.44345780710531e-07, "loss": 0.0016, "step": 96450 }, { "grad_norm": 0.02451721765100956, "learning_rate": 3.424112951558978e-07, "loss": 0.0025, "step": 96460 }, { "grad_norm": 0.0219111368060112, "learning_rate": 3.40482240080503e-07, "loss": 0.0014, "step": 96470 }, { "grad_norm": 0.025393908843398094, "learning_rate": 3.3855861569531133e-07, "loss": 0.0015, "step": 96480 }, { "grad_norm": 0.04814131557941437, "learning_rate": 3.3664042221068226e-07, "loss": 0.0027, "step": 96490 }, { "grad_norm": 0.02333453856408596, "learning_rate": 3.347276598363924e-07, "loss": 0.0014, "step": 96500 }, { "grad_norm": 0.020593849942088127, "learning_rate": 3.328203287816134e-07, "loss": 0.0024, "step": 96510 }, { "grad_norm": 0.0319550596177578, "learning_rate": 3.3091842925493386e-07, "loss": 0.004, "step": 96520 }, { "grad_norm": 0.03372547775506973, "learning_rate": 3.2902196146433196e-07, "loss": 0.0021, "step": 96530 }, { "grad_norm": 0.02854597568511963, "learning_rate": 3.2713092561720836e-07, "loss": 0.0018, "step": 96540 }, { "grad_norm": 0.02351042628288269, "learning_rate": 3.2524532192036996e-07, "loss": 0.0022, "step": 96550 }, { "grad_norm": 0.06024057790637016, "learning_rate": 3.2336515058001836e-07, "loss": 0.0034, "step": 96560 }, { "grad_norm": 0.016011208295822144, "learning_rate": 3.2149041180176143e-07, "loss": 0.0025, "step": 96570 }, { "grad_norm": 0.09682632982730865, "learning_rate": 3.1962110579062954e-07, "loss": 0.0036, "step": 96580 }, { "grad_norm": 0.04511258378624916, "learning_rate": 3.177572327510314e-07, "loss": 0.0023, "step": 96590 }, { "grad_norm": 0.015485890209674835, "learning_rate": 3.158987928868151e-07, "loss": 0.0024, "step": 96600 }, { "grad_norm": 0.02079617790877819, "learning_rate": 3.1404578640120144e-07, "loss": 0.0029, "step": 96610 }, { "grad_norm": 0.06542806327342987, "learning_rate": 3.1219821349683933e-07, "loss": 0.0025, "step": 96620 }, { "grad_norm": 0.05374599248170853, "learning_rate": 3.103560743757783e-07, "loss": 0.0021, "step": 96630 }, { "grad_norm": 0.03613894805312157, "learning_rate": 3.0851936923946834e-07, "loss": 0.002, "step": 96640 }, { "grad_norm": 0.015496104024350643, "learning_rate": 3.06688098288771e-07, "loss": 0.002, "step": 96650 }, { "grad_norm": 0.06234448030591011, "learning_rate": 3.0486226172394825e-07, "loss": 0.0023, "step": 96660 }, { "grad_norm": 0.023908326402306557, "learning_rate": 3.030418597446738e-07, "loss": 0.0021, "step": 96670 }, { "grad_norm": 0.033006828278303146, "learning_rate": 3.012268925500272e-07, "loss": 0.0014, "step": 96680 }, { "grad_norm": 0.029557788744568825, "learning_rate": 2.994173603384831e-07, "loss": 0.0033, "step": 96690 }, { "grad_norm": 0.08507516235113144, "learning_rate": 2.9761326330793317e-07, "loss": 0.0043, "step": 96700 }, { "grad_norm": 0.03661167994141579, "learning_rate": 2.958146016556751e-07, "loss": 0.0017, "step": 96710 }, { "grad_norm": 0.023570507764816284, "learning_rate": 2.94021375578396e-07, "loss": 0.0033, "step": 96720 }, { "grad_norm": 0.02357937954366207, "learning_rate": 2.9223358527221135e-07, "loss": 0.0022, "step": 96730 }, { "grad_norm": 0.039633773267269135, "learning_rate": 2.904512309326257e-07, "loss": 0.0024, "step": 96740 }, { "grad_norm": 0.03358771651983261, "learning_rate": 2.8867431275455545e-07, "loss": 0.0028, "step": 96750 }, { "grad_norm": 0.03494497016072273, "learning_rate": 2.86902830932323e-07, "loss": 0.0023, "step": 96760 }, { "grad_norm": 0.012270435690879822, "learning_rate": 2.851367856596565e-07, "loss": 0.0034, "step": 96770 }, { "grad_norm": 0.03625021502375603, "learning_rate": 2.83376177129685e-07, "loss": 0.002, "step": 96780 }, { "grad_norm": 0.03413993865251541, "learning_rate": 2.8162100553494884e-07, "loss": 0.0023, "step": 96790 }, { "grad_norm": 0.05962377414107323, "learning_rate": 2.798712710673834e-07, "loss": 0.0042, "step": 96800 }, { "grad_norm": 0.056050896644592285, "learning_rate": 2.7812697391834676e-07, "loss": 0.0016, "step": 96810 }, { "grad_norm": 0.03540138527750969, "learning_rate": 2.7638811427858627e-07, "loss": 0.0016, "step": 96820 }, { "grad_norm": 0.043835435062646866, "learning_rate": 2.746546923382665e-07, "loss": 0.0015, "step": 96830 }, { "grad_norm": 0.05373961105942726, "learning_rate": 2.729267082869469e-07, "loss": 0.0032, "step": 96840 }, { "grad_norm": 0.049764424562454224, "learning_rate": 2.712041623135986e-07, "loss": 0.0021, "step": 96850 }, { "grad_norm": 0.047533344477415085, "learning_rate": 2.694870546065931e-07, "loss": 0.0028, "step": 96860 }, { "grad_norm": 0.028317438438534737, "learning_rate": 2.677753853537246e-07, "loss": 0.0024, "step": 96870 }, { "grad_norm": 0.043840713798999786, "learning_rate": 2.660691547421601e-07, "loss": 0.0033, "step": 96880 }, { "grad_norm": 0.01997937075793743, "learning_rate": 2.6436836295850033e-07, "loss": 0.004, "step": 96890 }, { "grad_norm": 0.055341340601444244, "learning_rate": 2.6267301018874093e-07, "loss": 0.0018, "step": 96900 }, { "grad_norm": 0.017648929730057716, "learning_rate": 2.6098309661828355e-07, "loss": 0.0019, "step": 96910 }, { "grad_norm": 0.047767315059900284, "learning_rate": 2.592986224319305e-07, "loss": 0.0025, "step": 96920 }, { "grad_norm": 0.043039098381996155, "learning_rate": 2.576195878139009e-07, "loss": 0.0028, "step": 96930 }, { "grad_norm": 0.04785962775349617, "learning_rate": 2.5594599294780364e-07, "loss": 0.0013, "step": 96940 }, { "grad_norm": 0.025870613753795624, "learning_rate": 2.542778380166644e-07, "loss": 0.0015, "step": 96950 }, { "grad_norm": 0.025512799620628357, "learning_rate": 2.526151232029095e-07, "loss": 0.0049, "step": 96960 }, { "grad_norm": 0.021858656778931618, "learning_rate": 2.5095784868836567e-07, "loss": 0.0019, "step": 96970 }, { "grad_norm": 0.021188385784626007, "learning_rate": 2.493060146542825e-07, "loss": 0.0041, "step": 96980 }, { "grad_norm": 0.061679642647504807, "learning_rate": 2.476596212812876e-07, "loss": 0.0021, "step": 96990 }, { "grad_norm": 0.06103057041764259, "learning_rate": 2.4601866874943703e-07, "loss": 0.0029, "step": 97000 }, { "grad_norm": 0.026246482506394386, "learning_rate": 2.4438315723818163e-07, "loss": 0.0019, "step": 97010 }, { "grad_norm": 0.042964015156030655, "learning_rate": 2.4275308692637833e-07, "loss": 0.0019, "step": 97020 }, { "grad_norm": 0.020611269399523735, "learning_rate": 2.411284579922901e-07, "loss": 0.0026, "step": 97030 }, { "grad_norm": 0.03355874866247177, "learning_rate": 2.395092706135804e-07, "loss": 0.0031, "step": 97040 }, { "grad_norm": 0.047879207879304886, "learning_rate": 2.3789552496731872e-07, "loss": 0.0043, "step": 97050 }, { "grad_norm": 0.033767569810152054, "learning_rate": 2.3628722122999158e-07, "loss": 0.0017, "step": 97060 }, { "grad_norm": 0.028324328362941742, "learning_rate": 2.3468435957747503e-07, "loss": 0.0033, "step": 97070 }, { "grad_norm": 0.04573633149266243, "learning_rate": 2.330869401850566e-07, "loss": 0.0023, "step": 97080 }, { "grad_norm": 0.08697149902582169, "learning_rate": 2.3149496322742437e-07, "loss": 0.0014, "step": 97090 }, { "grad_norm": 0.03950726240873337, "learning_rate": 2.2990842887868346e-07, "loss": 0.0016, "step": 97100 }, { "grad_norm": 0.012848183512687683, "learning_rate": 2.2832733731232292e-07, "loss": 0.0019, "step": 97110 }, { "grad_norm": 0.035312190651893616, "learning_rate": 2.2675168870124886e-07, "loss": 0.0016, "step": 97120 }, { "grad_norm": 0.02644846774637699, "learning_rate": 2.2518148321778455e-07, "loss": 0.0016, "step": 97130 }, { "grad_norm": 0.03522539138793945, "learning_rate": 2.2361672103363706e-07, "loss": 0.0021, "step": 97140 }, { "grad_norm": 0.05427911505103111, "learning_rate": 2.220574023199251e-07, "loss": 0.003, "step": 97150 }, { "grad_norm": 0.029113294556736946, "learning_rate": 2.2050352724717894e-07, "loss": 0.0029, "step": 97160 }, { "grad_norm": 0.0314309224486351, "learning_rate": 2.1895509598532372e-07, "loss": 0.0019, "step": 97170 }, { "grad_norm": 0.08347928524017334, "learning_rate": 2.1741210870369066e-07, "loss": 0.0023, "step": 97180 }, { "grad_norm": 0.10524626821279526, "learning_rate": 2.1587456557102814e-07, "loss": 0.002, "step": 97190 }, { "grad_norm": 0.014479217119514942, "learning_rate": 2.143424667554683e-07, "loss": 0.0019, "step": 97200 }, { "grad_norm": 0.03654063865542412, "learning_rate": 2.128158124245716e-07, "loss": 0.0024, "step": 97210 }, { "grad_norm": 0.04583508521318436, "learning_rate": 2.1129460274527668e-07, "loss": 0.0028, "step": 97220 }, { "grad_norm": 0.02266627736389637, "learning_rate": 2.0977883788395046e-07, "loss": 0.0029, "step": 97230 }, { "grad_norm": 0.04255325347185135, "learning_rate": 2.0826851800634928e-07, "loss": 0.0024, "step": 97240 }, { "grad_norm": 0.023709535598754883, "learning_rate": 2.0676364327764653e-07, "loss": 0.0022, "step": 97250 }, { "grad_norm": 0.025204118341207504, "learning_rate": 2.0526421386240502e-07, "loss": 0.0013, "step": 97260 }, { "grad_norm": 0.02982226386666298, "learning_rate": 2.0377022992461025e-07, "loss": 0.0024, "step": 97270 }, { "grad_norm": 0.07074984908103943, "learning_rate": 2.0228169162763156e-07, "loss": 0.005, "step": 97280 }, { "grad_norm": 0.02182305045425892, "learning_rate": 2.0079859913425536e-07, "loss": 0.0024, "step": 97290 }, { "grad_norm": 0.05830863118171692, "learning_rate": 1.9932095260667417e-07, "loss": 0.0025, "step": 97300 }, { "grad_norm": 0.06877569109201431, "learning_rate": 1.9784875220648092e-07, "loss": 0.0046, "step": 97310 }, { "grad_norm": 0.08978430926799774, "learning_rate": 1.9638199809466907e-07, "loss": 0.002, "step": 97320 }, { "grad_norm": 0.028305910527706146, "learning_rate": 1.9492069043164364e-07, "loss": 0.0026, "step": 97330 }, { "grad_norm": 0.04877552390098572, "learning_rate": 1.9346482937721567e-07, "loss": 0.0024, "step": 97340 }, { "grad_norm": 0.07246219366788864, "learning_rate": 1.920144150905856e-07, "loss": 0.0022, "step": 97350 }, { "grad_norm": 0.03988852724432945, "learning_rate": 1.9056944773037655e-07, "loss": 0.0019, "step": 97360 }, { "grad_norm": 0.011815112084150314, "learning_rate": 1.8912992745460102e-07, "loss": 0.0013, "step": 97370 }, { "grad_norm": 0.01963530108332634, "learning_rate": 1.876958544206886e-07, "loss": 0.0032, "step": 97380 }, { "grad_norm": 0.04711122438311577, "learning_rate": 1.8626722878546942e-07, "loss": 0.0047, "step": 97390 }, { "grad_norm": 0.04169495031237602, "learning_rate": 1.8484405070516851e-07, "loss": 0.0019, "step": 97400 }, { "grad_norm": 0.04704965278506279, "learning_rate": 1.8342632033542805e-07, "loss": 0.0023, "step": 97410 }, { "grad_norm": 0.03953350707888603, "learning_rate": 1.8201403783129066e-07, "loss": 0.003, "step": 97420 }, { "grad_norm": 0.025133369490504265, "learning_rate": 1.806072033471884e-07, "loss": 0.0045, "step": 97430 }, { "grad_norm": 0.048445820808410645, "learning_rate": 1.79205817036987e-07, "loss": 0.0021, "step": 97440 }, { "grad_norm": 0.01936199888586998, "learning_rate": 1.7780987905393065e-07, "loss": 0.003, "step": 97450 }, { "grad_norm": 0.020023556426167488, "learning_rate": 1.764193895506805e-07, "loss": 0.0023, "step": 97460 }, { "grad_norm": 0.02832924947142601, "learning_rate": 1.7503434867929824e-07, "loss": 0.0021, "step": 97470 }, { "grad_norm": 0.045744217932224274, "learning_rate": 1.7365475659125165e-07, "loss": 0.0038, "step": 97480 }, { "grad_norm": 0.07059945166110992, "learning_rate": 1.7228061343740332e-07, "loss": 0.0083, "step": 97490 }, { "grad_norm": 0.02794918417930603, "learning_rate": 1.7091191936803308e-07, "loss": 0.0022, "step": 97500 }, { "grad_norm": 0.05047759786248207, "learning_rate": 1.6954867453281563e-07, "loss": 0.0022, "step": 97510 }, { "grad_norm": 0.0746237188577652, "learning_rate": 1.6819087908084285e-07, "loss": 0.0042, "step": 97520 }, { "grad_norm": 0.03700945898890495, "learning_rate": 1.6683853316058485e-07, "loss": 0.0015, "step": 97530 }, { "grad_norm": 0.02709764428436756, "learning_rate": 1.6549163691995106e-07, "loss": 0.0021, "step": 97540 }, { "grad_norm": 0.052119724452495575, "learning_rate": 1.641501905062237e-07, "loss": 0.0025, "step": 97550 }, { "grad_norm": 0.055299438536167145, "learning_rate": 1.628141940661021e-07, "loss": 0.0045, "step": 97560 }, { "grad_norm": 0.02325581945478916, "learning_rate": 1.6148364774569158e-07, "loss": 0.0021, "step": 97570 }, { "grad_norm": 0.03244763985276222, "learning_rate": 1.6015855169050355e-07, "loss": 0.0031, "step": 97580 }, { "grad_norm": 0.059332627803087234, "learning_rate": 1.5883890604543873e-07, "loss": 0.0015, "step": 97590 }, { "grad_norm": 0.019381478428840637, "learning_rate": 1.5752471095482057e-07, "loss": 0.0041, "step": 97600 }, { "grad_norm": 0.06302104890346527, "learning_rate": 1.5621596656235638e-07, "loss": 0.0027, "step": 97610 }, { "grad_norm": 0.04756031557917595, "learning_rate": 1.5491267301117608e-07, "loss": 0.0048, "step": 97620 }, { "grad_norm": 0.010802830569446087, "learning_rate": 1.536148304438101e-07, "loss": 0.0034, "step": 97630 }, { "grad_norm": 0.07837165147066116, "learning_rate": 1.5232243900217823e-07, "loss": 0.0041, "step": 97640 }, { "grad_norm": 0.05448540672659874, "learning_rate": 1.51035498827623e-07, "loss": 0.0016, "step": 97650 }, { "grad_norm": 0.030135279521346092, "learning_rate": 1.497540100608763e-07, "loss": 0.0013, "step": 97660 }, { "grad_norm": 0.014423205517232418, "learning_rate": 1.4847797284208153e-07, "loss": 0.0014, "step": 97670 }, { "grad_norm": 0.05942137539386749, "learning_rate": 1.472073873107882e-07, "loss": 0.0018, "step": 97680 }, { "grad_norm": 0.07074867933988571, "learning_rate": 1.459422536059407e-07, "loss": 0.003, "step": 97690 }, { "grad_norm": 0.03261151909828186, "learning_rate": 1.4468257186589508e-07, "loss": 0.0032, "step": 97700 }, { "grad_norm": 0.03696437180042267, "learning_rate": 1.4342834222840217e-07, "loss": 0.0025, "step": 97710 }, { "grad_norm": 0.020303983241319656, "learning_rate": 1.4217956483063566e-07, "loss": 0.0013, "step": 97720 }, { "grad_norm": 0.013537137769162655, "learning_rate": 1.409362398091474e-07, "loss": 0.0021, "step": 97730 }, { "grad_norm": 0.021863970905542374, "learning_rate": 1.3969836729990638e-07, "loss": 0.0023, "step": 97740 }, { "grad_norm": 0.044320572167634964, "learning_rate": 1.384659474382932e-07, "loss": 0.0021, "step": 97750 }, { "grad_norm": 0.014403547160327435, "learning_rate": 1.372389803590779e-07, "loss": 0.0025, "step": 97760 }, { "grad_norm": 0.025726018473505974, "learning_rate": 1.360174661964364e-07, "loss": 0.0021, "step": 97770 }, { "grad_norm": 0.04313318058848381, "learning_rate": 1.3480140508396188e-07, "loss": 0.0025, "step": 97780 }, { "grad_norm": 0.027320178225636482, "learning_rate": 1.335907971546313e-07, "loss": 0.0029, "step": 97790 }, { "grad_norm": 0.010397843085229397, "learning_rate": 1.3238564254083875e-07, "loss": 0.0023, "step": 97800 }, { "grad_norm": 0.01577157713472843, "learning_rate": 1.3118594137437323e-07, "loss": 0.0025, "step": 97810 }, { "grad_norm": 0.020130354911088943, "learning_rate": 1.2999169378644094e-07, "loss": 0.0025, "step": 97820 }, { "grad_norm": 0.020236346870660782, "learning_rate": 1.2880289990763738e-07, "loss": 0.0026, "step": 97830 }, { "grad_norm": 0.01486428827047348, "learning_rate": 1.2761955986796968e-07, "loss": 0.0021, "step": 97840 }, { "grad_norm": 0.023895282298326492, "learning_rate": 1.264416737968399e-07, "loss": 0.0026, "step": 97850 }, { "grad_norm": 0.029302816838026047, "learning_rate": 1.2526924182307275e-07, "loss": 0.0037, "step": 97860 }, { "grad_norm": 0.07127291709184647, "learning_rate": 1.2410226407487124e-07, "loss": 0.002, "step": 97870 }, { "grad_norm": 0.0176461860537529, "learning_rate": 1.229407406798555e-07, "loss": 0.0022, "step": 97880 }, { "grad_norm": 0.029140762984752655, "learning_rate": 1.2178467176505726e-07, "loss": 0.0017, "step": 97890 }, { "grad_norm": 0.02552892081439495, "learning_rate": 1.2063405745689205e-07, "loss": 0.0025, "step": 97900 }, { "grad_norm": 0.039957448840141296, "learning_rate": 1.1948889788119812e-07, "loss": 0.0026, "step": 97910 }, { "grad_norm": 0.04556552320718765, "learning_rate": 1.1834919316320303e-07, "loss": 0.0026, "step": 97920 }, { "grad_norm": 0.02325582318007946, "learning_rate": 1.1721494342754048e-07, "loss": 0.0028, "step": 97930 }, { "grad_norm": 0.018547451123595238, "learning_rate": 1.1608614879825563e-07, "loss": 0.002, "step": 97940 }, { "grad_norm": 0.043716199696063995, "learning_rate": 1.1496280939879422e-07, "loss": 0.0035, "step": 97950 }, { "grad_norm": 0.042825132608413696, "learning_rate": 1.1384492535199687e-07, "loss": 0.0019, "step": 97960 }, { "grad_norm": 0.017138484865427017, "learning_rate": 1.1273249678011578e-07, "loss": 0.0018, "step": 97970 }, { "grad_norm": 0.044086843729019165, "learning_rate": 1.1162552380480362e-07, "loss": 0.004, "step": 97980 }, { "grad_norm": 0.052127987146377563, "learning_rate": 1.1052400654711359e-07, "loss": 0.0027, "step": 97990 }, { "grad_norm": 0.018680719658732414, "learning_rate": 1.0942794512751597e-07, "loss": 0.0032, "step": 98000 }, { "grad_norm": 0.02064567245543003, "learning_rate": 1.0833733966587045e-07, "loss": 0.0038, "step": 98010 }, { "grad_norm": 0.036622852087020874, "learning_rate": 1.0725219028143718e-07, "loss": 0.0021, "step": 98020 }, { "grad_norm": 0.01538633182644844, "learning_rate": 1.0617249709289345e-07, "loss": 0.0038, "step": 98030 }, { "grad_norm": 0.021805431693792343, "learning_rate": 1.0509826021831703e-07, "loss": 0.0017, "step": 98040 }, { "grad_norm": 0.06701724231243134, "learning_rate": 1.0402947977517508e-07, "loss": 0.0026, "step": 98050 }, { "grad_norm": 0.011682609096169472, "learning_rate": 1.0296615588035185e-07, "loss": 0.0017, "step": 98060 }, { "grad_norm": 0.06806302815675735, "learning_rate": 1.0190828865012659e-07, "loss": 0.002, "step": 98070 }, { "grad_norm": 0.07687700539827347, "learning_rate": 1.008558782001956e-07, "loss": 0.0021, "step": 98080 }, { "grad_norm": 0.055407583713531494, "learning_rate": 9.980892464563906e-08, "loss": 0.0019, "step": 98090 }, { "grad_norm": 0.04666544497013092, "learning_rate": 9.876742810095985e-08, "loss": 0.0038, "step": 98100 }, { "grad_norm": 0.05966266989707947, "learning_rate": 9.77313886800446e-08, "loss": 0.0023, "step": 98110 }, { "grad_norm": 0.040763795375823975, "learning_rate": 9.670080649619717e-08, "loss": 0.0017, "step": 98120 }, { "grad_norm": 0.03490854799747467, "learning_rate": 9.567568166212737e-08, "loss": 0.0051, "step": 98130 }, { "grad_norm": 0.013126625679433346, "learning_rate": 9.465601428992888e-08, "loss": 0.0047, "step": 98140 }, { "grad_norm": 0.02088790200650692, "learning_rate": 9.364180449111803e-08, "loss": 0.0027, "step": 98150 }, { "grad_norm": 0.06132994592189789, "learning_rate": 9.263305237661169e-08, "loss": 0.0024, "step": 98160 }, { "grad_norm": 0.022462693974375725, "learning_rate": 9.162975805671603e-08, "loss": 0.0037, "step": 98170 }, { "grad_norm": 0.08655738830566406, "learning_rate": 9.063192164115442e-08, "loss": 0.0031, "step": 98180 }, { "grad_norm": 0.03149227797985077, "learning_rate": 8.963954323904511e-08, "loss": 0.002, "step": 98190 }, { "grad_norm": 0.06934130191802979, "learning_rate": 8.865262295891796e-08, "loss": 0.0035, "step": 98200 }, { "grad_norm": 0.035581767559051514, "learning_rate": 8.767116090870331e-08, "loss": 0.0026, "step": 98210 }, { "grad_norm": 0.01727980561554432, "learning_rate": 8.669515719572086e-08, "loss": 0.0019, "step": 98220 }, { "grad_norm": 0.017431074753403664, "learning_rate": 8.572461192671855e-08, "loss": 0.0031, "step": 98230 }, { "grad_norm": 0.029178068041801453, "learning_rate": 8.475952520782815e-08, "loss": 0.0026, "step": 98240 }, { "grad_norm": 0.030646946281194687, "learning_rate": 8.379989714458747e-08, "loss": 0.0013, "step": 98250 }, { "grad_norm": 0.06988786160945892, "learning_rate": 8.284572784194034e-08, "loss": 0.0021, "step": 98260 }, { "grad_norm": 0.028600439429283142, "learning_rate": 8.189701740424216e-08, "loss": 0.0025, "step": 98270 }, { "grad_norm": 0.052326999604701996, "learning_rate": 8.095376593522663e-08, "loss": 0.0018, "step": 98280 }, { "grad_norm": 0.01815553940832615, "learning_rate": 8.001597353806123e-08, "loss": 0.0021, "step": 98290 }, { "grad_norm": 0.03528902307152748, "learning_rate": 7.90836403152917e-08, "loss": 0.0021, "step": 98300 }, { "grad_norm": 0.07718713581562042, "learning_rate": 7.815676636888092e-08, "loss": 0.0039, "step": 98310 }, { "grad_norm": 0.0631919726729393, "learning_rate": 7.723535180019226e-08, "loss": 0.002, "step": 98320 }, { "grad_norm": 0.034972019493579865, "learning_rate": 7.6319396709984e-08, "loss": 0.0024, "step": 98330 }, { "grad_norm": 0.023211808875203133, "learning_rate": 7.540890119843158e-08, "loss": 0.0013, "step": 98340 }, { "grad_norm": 0.03227873519062996, "learning_rate": 7.450386536509979e-08, "loss": 0.001, "step": 98350 }, { "grad_norm": 0.03056296706199646, "learning_rate": 7.360428930895947e-08, "loss": 0.002, "step": 98360 }, { "grad_norm": 0.021854475140571594, "learning_rate": 7.271017312839302e-08, "loss": 0.0033, "step": 98370 }, { "grad_norm": 0.056301552802324295, "learning_rate": 7.182151692118333e-08, "loss": 0.0027, "step": 98380 }, { "grad_norm": 0.03342689201235771, "learning_rate": 7.093832078449713e-08, "loss": 0.0016, "step": 98390 }, { "grad_norm": 0.040643978863954544, "learning_rate": 7.006058481493493e-08, "loss": 0.0022, "step": 98400 }, { "grad_norm": 0.10312146693468094, "learning_rate": 6.918830910847552e-08, "loss": 0.0018, "step": 98410 }, { "grad_norm": 0.05003601312637329, "learning_rate": 6.832149376052033e-08, "loss": 0.0016, "step": 98420 }, { "grad_norm": 0.059204187244176865, "learning_rate": 6.746013886584912e-08, "loss": 0.0026, "step": 98430 }, { "grad_norm": 0.032966576516628265, "learning_rate": 6.660424451866431e-08, "loss": 0.0024, "step": 98440 }, { "grad_norm": 0.02626805007457733, "learning_rate": 6.575381081256327e-08, "loss": 0.0019, "step": 98450 }, { "grad_norm": 0.029216766357421875, "learning_rate": 6.49088378405549e-08, "loss": 0.0016, "step": 98460 }, { "grad_norm": 0.03173595666885376, "learning_rate": 6.406932569503755e-08, "loss": 0.0012, "step": 98470 }, { "grad_norm": 0.034543491899967194, "learning_rate": 6.323527446782662e-08, "loss": 0.0025, "step": 98480 }, { "grad_norm": 0.05133751779794693, "learning_rate": 6.240668425012142e-08, "loss": 0.0014, "step": 98490 }, { "grad_norm": 0.04327136278152466, "learning_rate": 6.158355513254388e-08, "loss": 0.0021, "step": 98500 }, { "grad_norm": 0.01496540941298008, "learning_rate": 6.076588720510534e-08, "loss": 0.0021, "step": 98510 }, { "grad_norm": 0.020250728353857994, "learning_rate": 5.995368055722316e-08, "loss": 0.0035, "step": 98520 }, { "grad_norm": 0.058389756828546524, "learning_rate": 5.914693527773185e-08, "loss": 0.0022, "step": 98530 }, { "grad_norm": 0.03427356481552124, "learning_rate": 5.834565145483861e-08, "loss": 0.0033, "step": 98540 }, { "grad_norm": 0.04130121320486069, "learning_rate": 5.754982917618446e-08, "loss": 0.0017, "step": 98550 }, { "grad_norm": 0.05923760309815407, "learning_rate": 5.675946852878866e-08, "loss": 0.0023, "step": 98560 }, { "grad_norm": 0.0371319055557251, "learning_rate": 5.59745695990932e-08, "loss": 0.0021, "step": 98570 }, { "grad_norm": 0.01848852075636387, "learning_rate": 5.51951324729294e-08, "loss": 0.0013, "step": 98580 }, { "grad_norm": 0.03751562163233757, "learning_rate": 5.4421157235529093e-08, "loss": 0.0016, "step": 98590 }, { "grad_norm": 0.0609101764857769, "learning_rate": 5.365264397154679e-08, "loss": 0.0018, "step": 98600 }, { "grad_norm": 0.07122135162353516, "learning_rate": 5.288959276501526e-08, "loss": 0.0033, "step": 98610 }, { "grad_norm": 0.011153742671012878, "learning_rate": 5.2132003699378876e-08, "loss": 0.0033, "step": 98620 }, { "grad_norm": 0.026325508952140808, "learning_rate": 5.137987685749912e-08, "loss": 0.0017, "step": 98630 }, { "grad_norm": 0.029167253524065018, "learning_rate": 5.0633212321610227e-08, "loss": 0.0045, "step": 98640 }, { "grad_norm": 0.01758931577205658, "learning_rate": 4.989201017338019e-08, "loss": 0.0028, "step": 98650 }, { "grad_norm": 0.040363866835832596, "learning_rate": 4.9156270493860846e-08, "loss": 0.0032, "step": 98660 }, { "grad_norm": 0.017394205555319786, "learning_rate": 4.842599336351561e-08, "loss": 0.002, "step": 98670 }, { "grad_norm": 0.03888239338994026, "learning_rate": 4.7701178862197274e-08, "loss": 0.0026, "step": 98680 }, { "grad_norm": 0.03867315873503685, "learning_rate": 4.698182706918131e-08, "loss": 0.0022, "step": 98690 }, { "grad_norm": 0.023311873897910118, "learning_rate": 4.6267938063121466e-08, "loss": 0.0039, "step": 98700 }, { "grad_norm": 0.02456883154809475, "learning_rate": 4.555951192209973e-08, "loss": 0.0022, "step": 98710 }, { "grad_norm": 0.03884761035442352, "learning_rate": 4.485654872358747e-08, "loss": 0.0017, "step": 98720 }, { "grad_norm": 0.05615631490945816, "learning_rate": 4.415904854445097e-08, "loss": 0.0042, "step": 98730 }, { "grad_norm": 0.020987940952181816, "learning_rate": 4.34670114609792e-08, "loss": 0.0026, "step": 98740 }, { "grad_norm": 0.054189521819353104, "learning_rate": 4.278043754884498e-08, "loss": 0.0032, "step": 98750 }, { "grad_norm": 0.07261312007904053, "learning_rate": 4.209932688312712e-08, "loss": 0.0027, "step": 98760 }, { "grad_norm": 0.03993108496069908, "learning_rate": 4.14236795383216e-08, "loss": 0.0022, "step": 98770 }, { "grad_norm": 0.028609326109290123, "learning_rate": 4.075349558830821e-08, "loss": 0.0017, "step": 98780 }, { "grad_norm": 0.06383258104324341, "learning_rate": 4.008877510638387e-08, "loss": 0.0017, "step": 98790 }, { "grad_norm": 0.04766970872879028, "learning_rate": 3.942951816523488e-08, "loss": 0.0031, "step": 98800 }, { "grad_norm": 0.05752476677298546, "learning_rate": 3.8775724836959125e-08, "loss": 0.003, "step": 98810 }, { "grad_norm": 0.01511762011796236, "learning_rate": 3.812739519305497e-08, "loss": 0.0029, "step": 98820 }, { "grad_norm": 0.0407753624022007, "learning_rate": 3.748452930442126e-08, "loss": 0.0022, "step": 98830 }, { "grad_norm": 0.028986841440200806, "learning_rate": 3.6847127241362855e-08, "loss": 0.0015, "step": 98840 }, { "grad_norm": 0.026933738961815834, "learning_rate": 3.621518907358512e-08, "loss": 0.0022, "step": 98850 }, { "grad_norm": 0.017373081296682358, "learning_rate": 3.558871487019388e-08, "loss": 0.0043, "step": 98860 }, { "grad_norm": 0.09290582686662674, "learning_rate": 3.4967704699701007e-08, "loss": 0.0028, "step": 98870 }, { "grad_norm": 0.015555351972579956, "learning_rate": 3.435215863001884e-08, "loss": 0.0026, "step": 98880 }, { "grad_norm": 0.050728898495435715, "learning_rate": 3.374207672846019e-08, "loss": 0.0024, "step": 98890 }, { "grad_norm": 0.016560660675168037, "learning_rate": 3.313745906174392e-08, "loss": 0.0023, "step": 98900 }, { "grad_norm": 0.09990046918392181, "learning_rate": 3.253830569599492e-08, "loss": 0.0019, "step": 98910 }, { "grad_norm": 0.04246450215578079, "learning_rate": 3.1944616696727436e-08, "loss": 0.0026, "step": 98920 }, { "grad_norm": 0.056079600006341934, "learning_rate": 3.135639212887287e-08, "loss": 0.0021, "step": 98930 }, { "grad_norm": 0.019625907763838768, "learning_rate": 3.077363205675754e-08, "loss": 0.0018, "step": 98940 }, { "grad_norm": 0.014745845459401608, "learning_rate": 3.0196336544113804e-08, "loss": 0.0025, "step": 98950 }, { "grad_norm": 0.031473156064748764, "learning_rate": 2.9624505654063383e-08, "loss": 0.0021, "step": 98960 }, { "grad_norm": 0.05934307351708412, "learning_rate": 2.90581394491507e-08, "loss": 0.0057, "step": 98970 }, { "grad_norm": 0.07828740030527115, "learning_rate": 2.8497237991309545e-08, "loss": 0.0019, "step": 98980 }, { "grad_norm": 0.04474852979183197, "learning_rate": 2.7941801341879735e-08, "loss": 0.0019, "step": 98990 }, { "grad_norm": 0.02180040255188942, "learning_rate": 2.7391829561601578e-08, "loss": 0.0018, "step": 99000 }, { "grad_norm": 0.049990106374025345, "learning_rate": 2.6847322710621403e-08, "loss": 0.0012, "step": 99010 }, { "grad_norm": 0.052738722413778305, "learning_rate": 2.6308280848486022e-08, "loss": 0.0025, "step": 99020 }, { "grad_norm": 0.05744105949997902, "learning_rate": 2.5774704034137177e-08, "loss": 0.0024, "step": 99030 }, { "grad_norm": 0.04856180399656296, "learning_rate": 2.524659232593929e-08, "loss": 0.0017, "step": 99040 }, { "grad_norm": 0.06347759068012238, "learning_rate": 2.472394578163506e-08, "loss": 0.002, "step": 99050 }, { "grad_norm": 0.029849840328097343, "learning_rate": 2.4206764458378772e-08, "loss": 0.0029, "step": 99060 }, { "grad_norm": 0.06185067072510719, "learning_rate": 2.3695048412736285e-08, "loss": 0.0014, "step": 99070 }, { "grad_norm": 0.07344269007444382, "learning_rate": 2.3188797700668395e-08, "loss": 0.0019, "step": 99080 }, { "grad_norm": 0.027619194239377975, "learning_rate": 2.268801237753082e-08, "loss": 0.0016, "step": 99090 }, { "grad_norm": 0.047669317573308945, "learning_rate": 2.2192692498090862e-08, "loss": 0.0032, "step": 99100 }, { "grad_norm": 0.022580115124583244, "learning_rate": 2.170283811652185e-08, "loss": 0.0027, "step": 99110 }, { "grad_norm": 0.03812472149729729, "learning_rate": 2.12184492863865e-08, "loss": 0.0022, "step": 99120 }, { "grad_norm": 0.012147696688771248, "learning_rate": 2.073952606066465e-08, "loss": 0.0021, "step": 99130 }, { "grad_norm": 0.06546423584222794, "learning_rate": 2.026606849171997e-08, "loss": 0.0028, "step": 99140 }, { "grad_norm": 0.06447058171033859, "learning_rate": 1.9798076631333262e-08, "loss": 0.002, "step": 99150 }, { "grad_norm": 0.03543512523174286, "learning_rate": 1.933555053069136e-08, "loss": 0.0016, "step": 99160 }, { "grad_norm": 0.0174406748265028, "learning_rate": 1.887849024036492e-08, "loss": 0.0019, "step": 99170 }, { "grad_norm": 0.07495015859603882, "learning_rate": 1.8426895810341736e-08, "loss": 0.0038, "step": 99180 }, { "grad_norm": 0.036370113492012024, "learning_rate": 1.798076729000453e-08, "loss": 0.0018, "step": 99190 }, { "grad_norm": 0.04228798672556877, "learning_rate": 1.75401047281476e-08, "loss": 0.0013, "step": 99200 }, { "grad_norm": 0.044953200966119766, "learning_rate": 1.7104908172954626e-08, "loss": 0.0033, "step": 99210 }, { "grad_norm": 0.03993658721446991, "learning_rate": 1.6675177672020868e-08, "loss": 0.0034, "step": 99220 }, { "grad_norm": 0.04435596242547035, "learning_rate": 1.6250913272342073e-08, "loss": 0.0017, "step": 99230 }, { "grad_norm": 0.03019709140062332, "learning_rate": 1.5832115020314453e-08, "loss": 0.0018, "step": 99240 }, { "grad_norm": 0.048142001032829285, "learning_rate": 1.5418782961734712e-08, "loss": 0.0022, "step": 99250 }, { "grad_norm": 0.032437317073345184, "learning_rate": 1.501091714181113e-08, "loss": 0.0025, "step": 99260 }, { "grad_norm": 0.030798370018601418, "learning_rate": 1.460851760513582e-08, "loss": 0.0027, "step": 99270 }, { "grad_norm": 0.04915241152048111, "learning_rate": 1.4211584395723566e-08, "loss": 0.0017, "step": 99280 }, { "grad_norm": 0.03566494584083557, "learning_rate": 1.382011755697854e-08, "loss": 0.0031, "step": 99290 }, { "grad_norm": 0.025821123272180557, "learning_rate": 1.3434117131716494e-08, "loss": 0.0016, "step": 99300 }, { "grad_norm": 0.054893508553504944, "learning_rate": 1.3053583162142557e-08, "loss": 0.003, "step": 99310 }, { "grad_norm": 0.02177971601486206, "learning_rate": 1.2678515689873438e-08, "loss": 0.0025, "step": 99320 }, { "grad_norm": 0.06232763081789017, "learning_rate": 1.2308914755931877e-08, "loss": 0.0024, "step": 99330 }, { "grad_norm": 0.040329307317733765, "learning_rate": 1.1944780400729993e-08, "loss": 0.0019, "step": 99340 }, { "grad_norm": 0.046609606593847275, "learning_rate": 1.1586112664085935e-08, "loss": 0.0048, "step": 99350 }, { "grad_norm": 0.05555790290236473, "learning_rate": 1.1232911585234984e-08, "loss": 0.0021, "step": 99360 }, { "grad_norm": 0.01906220242381096, "learning_rate": 1.0885177202790697e-08, "loss": 0.0043, "step": 99370 }, { "grad_norm": 0.04417794197797775, "learning_rate": 1.0542909554789315e-08, "loss": 0.0024, "step": 99380 }, { "grad_norm": 0.015551368705928326, "learning_rate": 1.0206108678656456e-08, "loss": 0.0036, "step": 99390 }, { "grad_norm": 0.03970441594719887, "learning_rate": 9.874774611223769e-09, "loss": 0.0019, "step": 99400 }, { "grad_norm": 0.057124774903059006, "learning_rate": 9.548907388728933e-09, "loss": 0.0024, "step": 99410 }, { "grad_norm": 0.013277589343488216, "learning_rate": 9.228507046804557e-09, "loss": 0.0029, "step": 99420 }, { "grad_norm": 0.06994612514972687, "learning_rate": 8.91357362048928e-09, "loss": 0.002, "step": 99430 }, { "grad_norm": 0.022245103493332863, "learning_rate": 8.604107144227769e-09, "loss": 0.0021, "step": 99440 }, { "grad_norm": 0.014242121949791908, "learning_rate": 8.300107651859623e-09, "loss": 0.0033, "step": 99450 }, { "grad_norm": 0.04540034756064415, "learning_rate": 8.001575176630472e-09, "loss": 0.0028, "step": 99460 }, { "grad_norm": 0.026765061542391777, "learning_rate": 7.708509751186422e-09, "loss": 0.0021, "step": 99470 }, { "grad_norm": 0.07279933989048004, "learning_rate": 7.420911407579611e-09, "loss": 0.0056, "step": 99480 }, { "grad_norm": 0.022021982818841934, "learning_rate": 7.13878017725711e-09, "loss": 0.0017, "step": 99490 }, { "grad_norm": 0.040894605219364166, "learning_rate": 6.8621160910720125e-09, "loss": 0.0013, "step": 99500 }, { "grad_norm": 0.032191917300224304, "learning_rate": 6.59091917928345e-09, "loss": 0.0021, "step": 99510 }, { "grad_norm": 0.019048303365707397, "learning_rate": 6.32518947155103e-09, "loss": 0.0014, "step": 99520 }, { "grad_norm": 0.02171645127236843, "learning_rate": 6.064926996929288e-09, "loss": 0.004, "step": 99530 }, { "grad_norm": 0.03804267197847366, "learning_rate": 5.810131783884343e-09, "loss": 0.003, "step": 99540 }, { "grad_norm": 0.04846678674221039, "learning_rate": 5.5608038602772415e-09, "loss": 0.0028, "step": 99550 }, { "grad_norm": 0.021658947691321373, "learning_rate": 5.316943253375062e-09, "loss": 0.0022, "step": 99560 }, { "grad_norm": 0.06633356213569641, "learning_rate": 5.07854998984536e-09, "loss": 0.0031, "step": 99570 }, { "grad_norm": 0.03821488842368126, "learning_rate": 4.845624095756174e-09, "loss": 0.0036, "step": 99580 }, { "grad_norm": 0.0217865239828825, "learning_rate": 4.618165596587121e-09, "loss": 0.002, "step": 99590 }, { "grad_norm": 0.048064131289720535, "learning_rate": 4.396174517207197e-09, "loss": 0.0039, "step": 99600 }, { "grad_norm": 0.02443508617579937, "learning_rate": 4.179650881896979e-09, "loss": 0.0014, "step": 99610 }, { "grad_norm": 0.051471635699272156, "learning_rate": 3.968594714331975e-09, "loss": 0.0017, "step": 99620 }, { "grad_norm": 0.05235738307237625, "learning_rate": 3.763006037593719e-09, "loss": 0.0047, "step": 99630 }, { "grad_norm": 0.04636736586689949, "learning_rate": 3.5628848741586786e-09, "loss": 0.0038, "step": 99640 }, { "grad_norm": 0.042956143617630005, "learning_rate": 3.368231245926001e-09, "loss": 0.0028, "step": 99650 }, { "grad_norm": 0.013591980561614037, "learning_rate": 3.179045174167561e-09, "loss": 0.0019, "step": 99660 }, { "grad_norm": 0.013937506824731827, "learning_rate": 2.9953266795834657e-09, "loss": 0.0021, "step": 99670 }, { "grad_norm": 0.014927874319255352, "learning_rate": 2.817075782263201e-09, "loss": 0.0023, "step": 99680 }, { "grad_norm": 0.0663418397307396, "learning_rate": 2.6442925016967325e-09, "loss": 0.002, "step": 99690 }, { "grad_norm": 0.047341469675302505, "learning_rate": 2.4769768567800555e-09, "loss": 0.0035, "step": 99700 }, { "grad_norm": 0.03740203008055687, "learning_rate": 2.315128865809646e-09, "loss": 0.0014, "step": 99710 }, { "grad_norm": 0.03987589478492737, "learning_rate": 2.1587485464880097e-09, "loss": 0.0019, "step": 99720 }, { "grad_norm": 0.01980525068938732, "learning_rate": 2.0078359159125813e-09, "loss": 0.0025, "step": 99730 }, { "grad_norm": 0.017495693638920784, "learning_rate": 1.8623909905923776e-09, "loss": 0.0013, "step": 99740 }, { "grad_norm": 0.02054174430668354, "learning_rate": 1.7224137864257916e-09, "loss": 0.0014, "step": 99750 }, { "grad_norm": 0.025922073051333427, "learning_rate": 1.5879043187283505e-09, "loss": 0.0022, "step": 99760 }, { "grad_norm": 0.029116667807102203, "learning_rate": 1.458862602204958e-09, "loss": 0.0036, "step": 99770 }, { "grad_norm": 0.06226621940732002, "learning_rate": 1.3352886509720996e-09, "loss": 0.0018, "step": 99780 }, { "grad_norm": 0.061199549585580826, "learning_rate": 1.2171824785356389e-09, "loss": 0.002, "step": 99790 }, { "grad_norm": 0.06867440044879913, "learning_rate": 1.1045440978185718e-09, "loss": 0.0019, "step": 99800 }, { "grad_norm": 0.03894897550344467, "learning_rate": 9.973735211332714e-10, "loss": 0.0018, "step": 99810 }, { "grad_norm": 0.013887157663702965, "learning_rate": 8.956707602036929e-10, "loss": 0.0024, "step": 99820 }, { "grad_norm": 0.024418296292424202, "learning_rate": 7.994358261542712e-10, "loss": 0.0016, "step": 99830 }, { "grad_norm": 0.03318384662270546, "learning_rate": 7.086687295043692e-10, "loss": 0.0023, "step": 99840 }, { "grad_norm": 0.05534980818629265, "learning_rate": 6.23369480179381e-10, "loss": 0.0023, "step": 99850 }, { "grad_norm": 0.062029846012592316, "learning_rate": 5.43538087510731e-10, "loss": 0.004, "step": 99860 }, { "grad_norm": 0.019488655030727386, "learning_rate": 4.691745602303233e-10, "loss": 0.0033, "step": 99870 }, { "grad_norm": 0.05203225091099739, "learning_rate": 4.0027890646499034e-10, "loss": 0.0018, "step": 99880 }, { "grad_norm": 0.060758840292692184, "learning_rate": 3.368511337531466e-10, "loss": 0.0035, "step": 99890 }, { "grad_norm": 0.06557714939117432, "learning_rate": 2.788912490281348e-10, "loss": 0.0021, "step": 99900 }, { "grad_norm": 0.0411020927131176, "learning_rate": 2.2639925863487954e-10, "loss": 0.0032, "step": 99910 }, { "grad_norm": 0.03372574597597122, "learning_rate": 1.7937516830213163e-10, "loss": 0.0022, "step": 99920 }, { "grad_norm": 0.04762095957994461, "learning_rate": 1.3781898318687702e-10, "loss": 0.0011, "step": 99930 }, { "grad_norm": 0.02457445114850998, "learning_rate": 1.0173070781882566e-10, "loss": 0.0023, "step": 99940 }, { "grad_norm": 0.049271535128355026, "learning_rate": 7.111034615592261e-11, "loss": 0.0032, "step": 99950 }, { "grad_norm": 0.018767286092042923, "learning_rate": 4.595790153993918e-11, "loss": 0.0016, "step": 99960 }, { "grad_norm": 0.049088794738054276, "learning_rate": 2.6273376729779585e-11, "loss": 0.0024, "step": 99970 }, { "grad_norm": 0.05040724202990532, "learning_rate": 1.2056773868174276e-11, "loss": 0.003, "step": 99980 }, { "grad_norm": 0.03473784402012825, "learning_rate": 3.3080945149865926e-12, "loss": 0.0012, "step": 99990 }, { "grad_norm": 0.04041016474366188, "learning_rate": 2.7339625008337977e-14, "loss": 0.0023, "step": 100000 } ], "logging_steps": 10, "max_steps": 100000, "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": 48, "trial_name": null, "trial_params": null }