{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 2.7266480922698975, "learning_rate": 1.8e-06, "loss": 1.238, "step": 10 }, { "grad_norm": 1.3996137380599976, "learning_rate": 3.8e-06, "loss": 1.2263, "step": 20 }, { "grad_norm": 0.6298696398735046, "learning_rate": 5.8e-06, "loss": 1.1558, "step": 30 }, { "grad_norm": 0.4681908190250397, "learning_rate": 7.8e-06, "loss": 1.1569, "step": 40 }, { "grad_norm": 0.9008304476737976, "learning_rate": 9.800000000000001e-06, "loss": 1.1582, "step": 50 }, { "grad_norm": 0.8695996999740601, "learning_rate": 1.18e-05, "loss": 1.1469, "step": 60 }, { "grad_norm": 0.754650890827179, "learning_rate": 1.3800000000000002e-05, "loss": 1.0852, "step": 70 }, { "grad_norm": 0.7990859746932983, "learning_rate": 1.58e-05, "loss": 1.0661, "step": 80 }, { "grad_norm": 0.6180333495140076, "learning_rate": 1.78e-05, "loss": 1.0627, "step": 90 }, { "grad_norm": 0.46167218685150146, "learning_rate": 1.9800000000000004e-05, "loss": 1.053, "step": 100 }, { "grad_norm": 0.5125234723091125, "learning_rate": 2.18e-05, "loss": 1.0584, "step": 110 }, { "grad_norm": 0.615847647190094, "learning_rate": 2.38e-05, "loss": 1.0547, "step": 120 }, { "grad_norm": 0.7808847427368164, "learning_rate": 2.58e-05, "loss": 1.0613, "step": 130 }, { "grad_norm": 0.6452605128288269, "learning_rate": 2.7800000000000005e-05, "loss": 1.0389, "step": 140 }, { "grad_norm": 0.7127711176872253, "learning_rate": 2.98e-05, "loss": 1.0285, "step": 150 }, { "grad_norm": 1.9395393133163452, "learning_rate": 3.18e-05, "loss": 0.9987, "step": 160 }, { "grad_norm": 2.0809590816497803, "learning_rate": 3.38e-05, "loss": 0.9503, "step": 170 }, { "grad_norm": 1.116188406944275, "learning_rate": 3.58e-05, "loss": 0.8721, "step": 180 }, { "grad_norm": 1.0570509433746338, "learning_rate": 3.7800000000000004e-05, "loss": 0.8447, "step": 190 }, { "grad_norm": 2.021897077560425, "learning_rate": 3.9800000000000005e-05, "loss": 0.7972, "step": 200 }, { "grad_norm": 1.9479414224624634, "learning_rate": 4.18e-05, "loss": 0.7084, "step": 210 }, { "grad_norm": 2.420440912246704, "learning_rate": 4.38e-05, "loss": 0.638, "step": 220 }, { "grad_norm": 1.9939135313034058, "learning_rate": 4.58e-05, "loss": 0.5441, "step": 230 }, { "grad_norm": 1.766343116760254, "learning_rate": 4.78e-05, "loss": 0.4735, "step": 240 }, { "grad_norm": 2.4787240028381348, "learning_rate": 4.9800000000000004e-05, "loss": 0.3976, "step": 250 }, { "grad_norm": 1.8052730560302734, "learning_rate": 4.9999557099392795e-05, "loss": 0.2773, "step": 260 }, { "grad_norm": 1.487770915031433, "learning_rate": 4.9998026105095405e-05, "loss": 0.2167, "step": 270 }, { "grad_norm": 1.126673936843872, "learning_rate": 4.9995401616153804e-05, "loss": 0.1938, "step": 280 }, { "grad_norm": 1.2377407550811768, "learning_rate": 4.9991683747371645e-05, "loss": 0.1878, "step": 290 }, { "grad_norm": 2.0436601638793945, "learning_rate": 4.9986872661380534e-05, "loss": 0.1571, "step": 300 }, { "grad_norm": 1.2251367568969727, "learning_rate": 4.998096856863298e-05, "loss": 0.1595, "step": 310 }, { "grad_norm": 1.2823742628097534, "learning_rate": 4.997397172739312e-05, "loss": 0.1464, "step": 320 }, { "grad_norm": 1.0287004709243774, "learning_rate": 4.996588244372545e-05, "loss": 0.1295, "step": 330 }, { "grad_norm": 1.0611475706100464, "learning_rate": 4.995670107148146e-05, "loss": 0.1309, "step": 340 }, { "grad_norm": 1.0434561967849731, "learning_rate": 4.9946428012284095e-05, "loss": 0.1344, "step": 350 }, { "grad_norm": 1.203682780265808, "learning_rate": 4.993506371551025e-05, "loss": 0.1339, "step": 360 }, { "grad_norm": 0.7937515377998352, "learning_rate": 4.992260867827109e-05, "loss": 0.1247, "step": 370 }, { "grad_norm": 0.9882832169532776, "learning_rate": 4.9909063445390285e-05, "loss": 0.1221, "step": 380 }, { "grad_norm": 1.3059258460998535, "learning_rate": 4.9894428609380206e-05, "loss": 0.1215, "step": 390 }, { "grad_norm": 1.3231853246688843, "learning_rate": 4.987870481041599e-05, "loss": 0.1257, "step": 400 }, { "grad_norm": 0.7158247828483582, "learning_rate": 4.986189273630752e-05, "loss": 0.1243, "step": 410 }, { "grad_norm": 1.0682462453842163, "learning_rate": 4.9843993122469424e-05, "loss": 0.1131, "step": 420 }, { "grad_norm": 0.914338231086731, "learning_rate": 4.9825006751888767e-05, "loss": 0.1135, "step": 430 }, { "grad_norm": 0.8019906878471375, "learning_rate": 4.9804934455090915e-05, "loss": 0.1135, "step": 440 }, { "grad_norm": 0.878158450126648, "learning_rate": 4.978377711010315e-05, "loss": 0.1176, "step": 450 }, { "grad_norm": 1.1396379470825195, "learning_rate": 4.976153564241628e-05, "loss": 0.1077, "step": 460 }, { "grad_norm": 1.298904538154602, "learning_rate": 4.973821102494417e-05, "loss": 0.1208, "step": 470 }, { "grad_norm": 0.8515462279319763, "learning_rate": 4.971380427798113e-05, "loss": 0.1024, "step": 480 }, { "grad_norm": 0.8240450620651245, "learning_rate": 4.9688316469157354e-05, "loss": 0.1018, "step": 490 }, { "grad_norm": 0.6559143662452698, "learning_rate": 4.9661748713392165e-05, "loss": 0.0972, "step": 500 }, { "grad_norm": 0.9851711392402649, "learning_rate": 4.963410217284526e-05, "loss": 0.113, "step": 510 }, { "grad_norm": 0.8339484333992004, "learning_rate": 4.960537805686589e-05, "loss": 0.0973, "step": 520 }, { "grad_norm": 0.8457871675491333, "learning_rate": 4.957557762193994e-05, "loss": 0.101, "step": 530 }, { "grad_norm": 0.9261676669120789, "learning_rate": 4.954470217163498e-05, "loss": 0.0894, "step": 540 }, { "grad_norm": 0.8262922763824463, "learning_rate": 4.9512753056543226e-05, "loss": 0.0825, "step": 550 }, { "grad_norm": 0.6502310633659363, "learning_rate": 4.947973167422247e-05, "loss": 0.0876, "step": 560 }, { "grad_norm": 0.7397481203079224, "learning_rate": 4.9445639469134944e-05, "loss": 0.0768, "step": 570 }, { "grad_norm": 0.7882953882217407, "learning_rate": 4.941047793258416e-05, "loss": 0.0969, "step": 580 }, { "grad_norm": 0.79180508852005, "learning_rate": 4.937424860264961e-05, "loss": 0.108, "step": 590 }, { "grad_norm": 0.7894780039787292, "learning_rate": 4.933695306411957e-05, "loss": 0.097, "step": 600 }, { "grad_norm": 0.8765780925750732, "learning_rate": 4.929859294842172e-05, "loss": 0.1013, "step": 610 }, { "grad_norm": 0.9128709435462952, "learning_rate": 4.9259169933551765e-05, "loss": 0.0854, "step": 620 }, { "grad_norm": 0.6108020544052124, "learning_rate": 4.921868574400011e-05, "loss": 0.0809, "step": 630 }, { "grad_norm": 0.7296820878982544, "learning_rate": 4.9177142150676355e-05, "loss": 0.0858, "step": 640 }, { "grad_norm": 0.8388088345527649, "learning_rate": 4.913454097083185e-05, "loss": 0.0974, "step": 650 }, { "grad_norm": 0.920002818107605, "learning_rate": 4.9090884067980205e-05, "loss": 0.0857, "step": 660 }, { "grad_norm": 0.9557753205299377, "learning_rate": 4.9046173351815796e-05, "loss": 0.1021, "step": 670 }, { "grad_norm": 1.0099128484725952, "learning_rate": 4.900041077813017e-05, "loss": 0.0937, "step": 680 }, { "grad_norm": 0.7957805395126343, "learning_rate": 4.895359834872656e-05, "loss": 0.0919, "step": 690 }, { "grad_norm": 0.5897656083106995, "learning_rate": 4.8905738111332275e-05, "loss": 0.0868, "step": 700 }, { "grad_norm": 0.82470703125, "learning_rate": 4.8856832159509156e-05, "loss": 0.0897, "step": 710 }, { "grad_norm": 0.5583815574645996, "learning_rate": 4.880688263256197e-05, "loss": 0.0831, "step": 720 }, { "grad_norm": 0.5602760910987854, "learning_rate": 4.8755891715444814e-05, "loss": 0.0864, "step": 730 }, { "grad_norm": 0.6645472645759583, "learning_rate": 4.870386163866561e-05, "loss": 0.0785, "step": 740 }, { "grad_norm": 0.4997585117816925, "learning_rate": 4.8650794678188485e-05, "loss": 0.0736, "step": 750 }, { "grad_norm": 0.6175997853279114, "learning_rate": 4.859669315533417e-05, "loss": 0.0802, "step": 760 }, { "grad_norm": 0.6164193749427795, "learning_rate": 4.8541559436678563e-05, "loss": 0.0741, "step": 770 }, { "grad_norm": 0.692690908908844, "learning_rate": 4.8485395933949114e-05, "loss": 0.0803, "step": 780 }, { "grad_norm": 0.7447015047073364, "learning_rate": 4.842820510391938e-05, "loss": 0.0747, "step": 790 }, { "grad_norm": 0.9005197882652283, "learning_rate": 4.836998944830155e-05, "loss": 0.0859, "step": 800 }, { "grad_norm": 0.6014283895492554, "learning_rate": 4.8310751513636975e-05, "loss": 0.0847, "step": 810 }, { "grad_norm": 0.5599387884140015, "learning_rate": 4.825049389118484e-05, "loss": 0.0709, "step": 820 }, { "grad_norm": 0.6468709111213684, "learning_rate": 4.818921921680874e-05, "loss": 0.0887, "step": 830 }, { "grad_norm": 0.6334129571914673, "learning_rate": 4.812693017086145e-05, "loss": 0.0689, "step": 840 }, { "grad_norm": 0.6896155476570129, "learning_rate": 4.806362947806763e-05, "loss": 0.0839, "step": 850 }, { "grad_norm": 0.8341414928436279, "learning_rate": 4.799931990740463e-05, "loss": 0.0785, "step": 860 }, { "grad_norm": 0.5337378978729248, "learning_rate": 4.7934004271981415e-05, "loss": 0.0769, "step": 870 }, { "grad_norm": 0.7976155877113342, "learning_rate": 4.786768542891548e-05, "loss": 0.0738, "step": 880 }, { "grad_norm": 0.3165757358074188, "learning_rate": 4.7800366279207854e-05, "loss": 0.0713, "step": 890 }, { "grad_norm": 0.7630752325057983, "learning_rate": 4.773204976761623e-05, "loss": 0.0719, "step": 900 }, { "grad_norm": 0.6145244836807251, "learning_rate": 4.7662738882526146e-05, "loss": 0.0644, "step": 910 }, { "grad_norm": 0.6122027635574341, "learning_rate": 4.759243665582024e-05, "loss": 0.0728, "step": 920 }, { "grad_norm": 0.4458109736442566, "learning_rate": 4.752114616274567e-05, "loss": 0.0678, "step": 930 }, { "grad_norm": 0.6862826347351074, "learning_rate": 4.744887052177954e-05, "loss": 0.0694, "step": 940 }, { "grad_norm": 0.5472699403762817, "learning_rate": 4.7375612894492536e-05, "loss": 0.0739, "step": 950 }, { "grad_norm": 0.6469587087631226, "learning_rate": 4.7301376485410596e-05, "loss": 0.0612, "step": 960 }, { "grad_norm": 0.5633054971694946, "learning_rate": 4.722616454187474e-05, "loss": 0.0752, "step": 970 }, { "grad_norm": 0.611199140548706, "learning_rate": 4.714998035389904e-05, "loss": 0.0682, "step": 980 }, { "grad_norm": 0.6113159656524658, "learning_rate": 4.7072827254026666e-05, "loss": 0.0734, "step": 990 }, { "grad_norm": 0.5616899728775024, "learning_rate": 4.6994708617184155e-05, "loss": 0.0647, "step": 1000 }, { "grad_norm": 0.5405145883560181, "learning_rate": 4.691562786053376e-05, "loss": 0.0652, "step": 1010 }, { "grad_norm": 0.5287688374519348, "learning_rate": 4.6835588443323956e-05, "loss": 0.071, "step": 1020 }, { "grad_norm": 0.6897308230400085, "learning_rate": 4.675459386673815e-05, "loss": 0.0711, "step": 1030 }, { "grad_norm": 0.43904393911361694, "learning_rate": 4.6672647673741485e-05, "loss": 0.0762, "step": 1040 }, { "grad_norm": 0.502697765827179, "learning_rate": 4.658975344892594e-05, "loss": 0.0892, "step": 1050 }, { "grad_norm": 0.718237042427063, "learning_rate": 4.650591481835344e-05, "loss": 0.0862, "step": 1060 }, { "grad_norm": 0.570436954498291, "learning_rate": 4.642113544939728e-05, "loss": 0.07, "step": 1070 }, { "grad_norm": 0.4335796535015106, "learning_rate": 4.63354190505817e-05, "loss": 0.0811, "step": 1080 }, { "grad_norm": 0.6726382374763489, "learning_rate": 4.6248769371419684e-05, "loss": 0.0675, "step": 1090 }, { "grad_norm": 0.5062233805656433, "learning_rate": 4.6161190202248897e-05, "loss": 0.064, "step": 1100 }, { "grad_norm": 0.3789091408252716, "learning_rate": 4.607268537406591e-05, "loss": 0.0706, "step": 1110 }, { "grad_norm": 0.5326701402664185, "learning_rate": 4.598325875835862e-05, "loss": 0.071, "step": 1120 }, { "grad_norm": 0.5220073461532593, "learning_rate": 4.589291426693692e-05, "loss": 0.0727, "step": 1130 }, { "grad_norm": 0.5300417542457581, "learning_rate": 4.580165585176152e-05, "loss": 0.0678, "step": 1140 }, { "grad_norm": 0.524254560470581, "learning_rate": 4.570948750477114e-05, "loss": 0.0652, "step": 1150 }, { "grad_norm": 0.5178232192993164, "learning_rate": 4.561641325770788e-05, "loss": 0.0636, "step": 1160 }, { "grad_norm": 0.35104331374168396, "learning_rate": 4.5522437181940804e-05, "loss": 0.0624, "step": 1170 }, { "grad_norm": 0.5118323564529419, "learning_rate": 4.542756338828791e-05, "loss": 0.0622, "step": 1180 }, { "grad_norm": 0.4421655535697937, "learning_rate": 4.533179602683629e-05, "loss": 0.0629, "step": 1190 }, { "grad_norm": 0.7161966562271118, "learning_rate": 4.523513928676056e-05, "loss": 0.0599, "step": 1200 }, { "grad_norm": 0.4908861219882965, "learning_rate": 4.5137597396139676e-05, "loss": 0.0653, "step": 1210 }, { "grad_norm": 1.035172939300537, "learning_rate": 4.503917462177192e-05, "loss": 0.0592, "step": 1220 }, { "grad_norm": 0.6242174506187439, "learning_rate": 4.4939875268988273e-05, "loss": 0.0679, "step": 1230 }, { "grad_norm": 0.34626179933547974, "learning_rate": 4.483970368146413e-05, "loss": 0.0608, "step": 1240 }, { "grad_norm": 0.41088080406188965, "learning_rate": 4.473866424102923e-05, "loss": 0.0715, "step": 1250 }, { "grad_norm": 0.4913835823535919, "learning_rate": 4.463676136747602e-05, "loss": 0.0519, "step": 1260 }, { "grad_norm": 0.39666661620140076, "learning_rate": 4.453399951836632e-05, "loss": 0.0559, "step": 1270 }, { "grad_norm": 0.5405796766281128, "learning_rate": 4.44303831888363e-05, "loss": 0.0645, "step": 1280 }, { "grad_norm": 0.6450210809707642, "learning_rate": 4.432591691139989e-05, "loss": 0.0692, "step": 1290 }, { "grad_norm": 0.6847130656242371, "learning_rate": 4.422060525575048e-05, "loss": 0.0665, "step": 1300 }, { "grad_norm": 0.5179506540298462, "learning_rate": 4.411445282856106e-05, "loss": 0.0612, "step": 1310 }, { "grad_norm": 0.76016765832901, "learning_rate": 4.400746427328268e-05, "loss": 0.0618, "step": 1320 }, { "grad_norm": 0.522050678730011, "learning_rate": 4.389964426994134e-05, "loss": 0.0632, "step": 1330 }, { "grad_norm": 0.5314463973045349, "learning_rate": 4.3790997534933276e-05, "loss": 0.0628, "step": 1340 }, { "grad_norm": 0.4170549213886261, "learning_rate": 4.368152882081865e-05, "loss": 0.0595, "step": 1350 }, { "grad_norm": 0.5747459530830383, "learning_rate": 4.357124291611363e-05, "loss": 0.0549, "step": 1360 }, { "grad_norm": 0.5837365388870239, "learning_rate": 4.346014464508098e-05, "loss": 0.0519, "step": 1370 }, { "grad_norm": 0.35218408703804016, "learning_rate": 4.3348238867518985e-05, "loss": 0.06, "step": 1380 }, { "grad_norm": 0.40569043159484863, "learning_rate": 4.323553047854887e-05, "loss": 0.0484, "step": 1390 }, { "grad_norm": 0.5539340972900391, "learning_rate": 4.3122024408400693e-05, "loss": 0.0496, "step": 1400 }, { "grad_norm": 0.33733201026916504, "learning_rate": 4.3007725622197674e-05, "loss": 0.0508, "step": 1410 }, { "grad_norm": 0.5494152903556824, "learning_rate": 4.2892639119739e-05, "loss": 0.0621, "step": 1420 }, { "grad_norm": 0.39040982723236084, "learning_rate": 4.277676993528112e-05, "loss": 0.0528, "step": 1430 }, { "grad_norm": 0.5634658932685852, "learning_rate": 4.2660123137317523e-05, "loss": 0.0586, "step": 1440 }, { "grad_norm": 0.5504993796348572, "learning_rate": 4.254270382835703e-05, "loss": 0.0617, "step": 1450 }, { "grad_norm": 0.3419093191623688, "learning_rate": 4.2424517144700603e-05, "loss": 0.0573, "step": 1460 }, { "grad_norm": 0.3770234286785126, "learning_rate": 4.230556825621667e-05, "loss": 0.0552, "step": 1470 }, { "grad_norm": 0.5088812708854675, "learning_rate": 4.2185862366114936e-05, "loss": 0.061, "step": 1480 }, { "grad_norm": 0.5062453746795654, "learning_rate": 4.2065404710718834e-05, "loss": 0.0528, "step": 1490 }, { "grad_norm": 0.7702981233596802, "learning_rate": 4.194420055923644e-05, "loss": 0.0637, "step": 1500 }, { "grad_norm": 0.6036869287490845, "learning_rate": 4.182225521352999e-05, "loss": 0.0702, "step": 1510 }, { "grad_norm": 0.5008228421211243, "learning_rate": 4.169957400788395e-05, "loss": 0.0624, "step": 1520 }, { "grad_norm": 0.41124141216278076, "learning_rate": 4.157616230877169e-05, "loss": 0.0563, "step": 1530 }, { "grad_norm": 0.7010630369186401, "learning_rate": 4.145202551462072e-05, "loss": 0.062, "step": 1540 }, { "grad_norm": 0.4738198220729828, "learning_rate": 4.132716905557658e-05, "loss": 0.0636, "step": 1550 }, { "grad_norm": 0.5558097958564758, "learning_rate": 4.1201598393265244e-05, "loss": 0.0627, "step": 1560 }, { "grad_norm": 0.4446842074394226, "learning_rate": 4.1075319020554286e-05, "loss": 0.0577, "step": 1570 }, { "grad_norm": 0.476675808429718, "learning_rate": 4.094833646131256e-05, "loss": 0.0544, "step": 1580 }, { "grad_norm": 0.4960286617279053, "learning_rate": 4.082065627016858e-05, "loss": 0.0644, "step": 1590 }, { "grad_norm": 0.46870744228363037, "learning_rate": 4.0692284032267516e-05, "loss": 0.054, "step": 1600 }, { "grad_norm": 0.7339263558387756, "learning_rate": 4.056322536302693e-05, "loss": 0.0552, "step": 1610 }, { "grad_norm": 0.3929803669452667, "learning_rate": 4.043348590789111e-05, "loss": 0.0554, "step": 1620 }, { "grad_norm": 0.4788963198661804, "learning_rate": 4.030307134208411e-05, "loss": 0.052, "step": 1630 }, { "grad_norm": 0.4371449649333954, "learning_rate": 4.017198737036154e-05, "loss": 0.0531, "step": 1640 }, { "grad_norm": 0.4106438159942627, "learning_rate": 4.004023972676097e-05, "loss": 0.0569, "step": 1650 }, { "grad_norm": 0.3421051800251007, "learning_rate": 3.9907834174351125e-05, "loss": 0.0509, "step": 1660 }, { "grad_norm": 0.45882418751716614, "learning_rate": 3.9774776504979805e-05, "loss": 0.058, "step": 1670 }, { "grad_norm": 0.5227372646331787, "learning_rate": 3.964107253902052e-05, "loss": 0.0552, "step": 1680 }, { "grad_norm": 0.567558228969574, "learning_rate": 3.950672812511788e-05, "loss": 0.0617, "step": 1690 }, { "grad_norm": 0.6489816904067993, "learning_rate": 3.937174913993177e-05, "loss": 0.0563, "step": 1700 }, { "grad_norm": 0.4380536377429962, "learning_rate": 3.9236141487880265e-05, "loss": 0.0508, "step": 1710 }, { "grad_norm": 0.5422831177711487, "learning_rate": 3.909991110088138e-05, "loss": 0.0541, "step": 1720 }, { "grad_norm": 0.7260904908180237, "learning_rate": 3.896306393809357e-05, "loss": 0.0563, "step": 1730 }, { "grad_norm": 0.4155614972114563, "learning_rate": 3.882560598565504e-05, "loss": 0.0552, "step": 1740 }, { "grad_norm": 0.35693034529685974, "learning_rate": 3.8687543256421955e-05, "loss": 0.0521, "step": 1750 }, { "grad_norm": 0.591141402721405, "learning_rate": 3.8548881789705345e-05, "loss": 0.0571, "step": 1760 }, { "grad_norm": 0.5429609417915344, "learning_rate": 3.840962765100696e-05, "loss": 0.0575, "step": 1770 }, { "grad_norm": 0.4284355938434601, "learning_rate": 3.826978693175395e-05, "loss": 0.0547, "step": 1780 }, { "grad_norm": 0.720148503780365, "learning_rate": 3.81293657490324e-05, "loss": 0.0575, "step": 1790 }, { "grad_norm": 0.49093666672706604, "learning_rate": 3.798837024531974e-05, "loss": 0.0513, "step": 1800 }, { "grad_norm": 0.5715184211730957, "learning_rate": 3.7846806588216055e-05, "loss": 0.0571, "step": 1810 }, { "grad_norm": 0.45693865418434143, "learning_rate": 3.770468097017432e-05, "loss": 0.0481, "step": 1820 }, { "grad_norm": 0.49179983139038086, "learning_rate": 3.7561999608229505e-05, "loss": 0.0528, "step": 1830 }, { "grad_norm": 0.4709620177745819, "learning_rate": 3.741876874372659e-05, "loss": 0.0501, "step": 1840 }, { "grad_norm": 0.4550098776817322, "learning_rate": 3.727499464204758e-05, "loss": 0.0493, "step": 1850 }, { "grad_norm": 0.4054962396621704, "learning_rate": 3.7130683592337465e-05, "loss": 0.0476, "step": 1860 }, { "grad_norm": 0.541397750377655, "learning_rate": 3.698584190722906e-05, "loss": 0.0534, "step": 1870 }, { "grad_norm": 0.36921900510787964, "learning_rate": 3.6840475922566886e-05, "loss": 0.0478, "step": 1880 }, { "grad_norm": 0.5922922492027283, "learning_rate": 3.6694591997130026e-05, "loss": 0.0501, "step": 1890 }, { "grad_norm": 0.43542882800102234, "learning_rate": 3.654819651235401e-05, "loss": 0.0541, "step": 1900 }, { "grad_norm": 0.48189231753349304, "learning_rate": 3.640129587205156e-05, "loss": 0.0521, "step": 1910 }, { "grad_norm": 0.4395929276943207, "learning_rate": 3.625389650213259e-05, "loss": 0.0429, "step": 1920 }, { "grad_norm": 0.5689699649810791, "learning_rate": 3.610600485032305e-05, "loss": 0.0586, "step": 1930 }, { "grad_norm": 0.5108771920204163, "learning_rate": 3.5957627385882884e-05, "loss": 0.0523, "step": 1940 }, { "grad_norm": 0.5463844537734985, "learning_rate": 3.580877059932308e-05, "loss": 0.0449, "step": 1950 }, { "grad_norm": 0.3828030526638031, "learning_rate": 3.5659441002121694e-05, "loss": 0.0509, "step": 1960 }, { "grad_norm": 0.3314559757709503, "learning_rate": 3.550964512643908e-05, "loss": 0.0435, "step": 1970 }, { "grad_norm": 0.8163153529167175, "learning_rate": 3.535938952483211e-05, "loss": 0.0558, "step": 1980 }, { "grad_norm": 0.7200665473937988, "learning_rate": 3.520868076996755e-05, "loss": 0.0432, "step": 1990 }, { "grad_norm": 0.47182565927505493, "learning_rate": 3.5057525454334564e-05, "loss": 0.0455, "step": 2000 }, { "grad_norm": 0.5063982009887695, "learning_rate": 3.490593018995636e-05, "loss": 0.0446, "step": 2010 }, { "grad_norm": 0.44255322217941284, "learning_rate": 3.475390160810087e-05, "loss": 0.0434, "step": 2020 }, { "grad_norm": 0.4013705849647522, "learning_rate": 3.460144635899079e-05, "loss": 0.0501, "step": 2030 }, { "grad_norm": 0.5221493244171143, "learning_rate": 3.444857111151259e-05, "loss": 0.0547, "step": 2040 }, { "grad_norm": 0.4845060110092163, "learning_rate": 3.429528255292485e-05, "loss": 0.0497, "step": 2050 }, { "grad_norm": 0.4617994725704193, "learning_rate": 3.41415873885657e-05, "loss": 0.0489, "step": 2060 }, { "grad_norm": 0.4810671806335449, "learning_rate": 3.3987492341559535e-05, "loss": 0.0478, "step": 2070 }, { "grad_norm": 0.5026255249977112, "learning_rate": 3.383300415252292e-05, "loss": 0.0446, "step": 2080 }, { "grad_norm": 0.477425754070282, "learning_rate": 3.367812957926971e-05, "loss": 0.0517, "step": 2090 }, { "grad_norm": 0.3495548367500305, "learning_rate": 3.352287539651545e-05, "loss": 0.0456, "step": 2100 }, { "grad_norm": 0.45665979385375977, "learning_rate": 3.3367248395581076e-05, "loss": 0.046, "step": 2110 }, { "grad_norm": 0.41686445474624634, "learning_rate": 3.321125538409574e-05, "loss": 0.0486, "step": 2120 }, { "grad_norm": 0.3617282211780548, "learning_rate": 3.3054903185699135e-05, "loss": 0.0476, "step": 2130 }, { "grad_norm": 0.47553274035453796, "learning_rate": 3.2898198639742915e-05, "loss": 0.0432, "step": 2140 }, { "grad_norm": 0.3067774176597595, "learning_rate": 3.2741148600991576e-05, "loss": 0.0522, "step": 2150 }, { "grad_norm": 0.5735144019126892, "learning_rate": 3.258375993932259e-05, "loss": 0.0476, "step": 2160 }, { "grad_norm": 0.2728676497936249, "learning_rate": 3.2426039539425876e-05, "loss": 0.0414, "step": 2170 }, { "grad_norm": 0.6051283478736877, "learning_rate": 3.226799430050268e-05, "loss": 0.0499, "step": 2180 }, { "grad_norm": 0.5630646347999573, "learning_rate": 3.210963113596374e-05, "loss": 0.0418, "step": 2190 }, { "grad_norm": 0.4311167001724243, "learning_rate": 3.1950956973126906e-05, "loss": 0.0499, "step": 2200 }, { "grad_norm": 0.4380161166191101, "learning_rate": 3.1791978752914085e-05, "loss": 0.0401, "step": 2210 }, { "grad_norm": 0.43494531512260437, "learning_rate": 3.163270342954766e-05, "loss": 0.0469, "step": 2220 }, { "grad_norm": 0.5626083612442017, "learning_rate": 3.1473137970246246e-05, "loss": 0.049, "step": 2230 }, { "grad_norm": 0.4352441132068634, "learning_rate": 3.1313289354919945e-05, "loss": 0.0459, "step": 2240 }, { "grad_norm": 0.48047247529029846, "learning_rate": 3.1153164575865046e-05, "loss": 0.0459, "step": 2250 }, { "grad_norm": 0.6736582517623901, "learning_rate": 3.099277063745811e-05, "loss": 0.0539, "step": 2260 }, { "grad_norm": 0.40036192536354065, "learning_rate": 3.0832114555849614e-05, "loss": 0.0438, "step": 2270 }, { "grad_norm": 0.42229387164115906, "learning_rate": 3.0671203358657e-05, "loss": 0.0413, "step": 2280 }, { "grad_norm": 0.3298429250717163, "learning_rate": 3.051004408465733e-05, "loss": 0.04, "step": 2290 }, { "grad_norm": 0.43710318207740784, "learning_rate": 3.034864378347933e-05, "loss": 0.0454, "step": 2300 }, { "grad_norm": 0.4662761390209198, "learning_rate": 3.018700951529504e-05, "loss": 0.0521, "step": 2310 }, { "grad_norm": 0.5804249048233032, "learning_rate": 3.0025148350510974e-05, "loss": 0.0475, "step": 2320 }, { "grad_norm": 0.4282180070877075, "learning_rate": 2.986306736945883e-05, "loss": 0.0466, "step": 2330 }, { "grad_norm": 0.45659002661705017, "learning_rate": 2.970077366208579e-05, "loss": 0.0387, "step": 2340 }, { "grad_norm": 0.5479471683502197, "learning_rate": 2.953827432764438e-05, "loss": 0.0461, "step": 2350 }, { "grad_norm": 0.31041985750198364, "learning_rate": 2.9375576474381905e-05, "loss": 0.042, "step": 2360 }, { "grad_norm": 0.42949941754341125, "learning_rate": 2.921268721922954e-05, "loss": 0.0435, "step": 2370 }, { "grad_norm": 0.4704861044883728, "learning_rate": 2.904961368749099e-05, "loss": 0.0447, "step": 2380 }, { "grad_norm": 0.4041045606136322, "learning_rate": 2.8886363012530826e-05, "loss": 0.0429, "step": 2390 }, { "grad_norm": 0.3385079503059387, "learning_rate": 2.8722942335462415e-05, "loss": 0.0454, "step": 2400 }, { "grad_norm": 0.5993208289146423, "learning_rate": 2.8559358804835597e-05, "loss": 0.0426, "step": 2410 }, { "grad_norm": 0.36963415145874023, "learning_rate": 2.839561957632393e-05, "loss": 0.0515, "step": 2420 }, { "grad_norm": 0.5569792985916138, "learning_rate": 2.823173181241171e-05, "loss": 0.0484, "step": 2430 }, { "grad_norm": 0.4478709101676941, "learning_rate": 2.806770268208066e-05, "loss": 0.0442, "step": 2440 }, { "grad_norm": 0.39773526787757874, "learning_rate": 2.7903539360496323e-05, "loss": 0.0362, "step": 2450 }, { "grad_norm": 0.4185701906681061, "learning_rate": 2.773924902869418e-05, "loss": 0.046, "step": 2460 }, { "grad_norm": 0.43949025869369507, "learning_rate": 2.757483887326559e-05, "loss": 0.0433, "step": 2470 }, { "grad_norm": 0.4299829602241516, "learning_rate": 2.741031608604337e-05, "loss": 0.0394, "step": 2480 }, { "grad_norm": 0.27871155738830566, "learning_rate": 2.7245687863787195e-05, "loss": 0.0367, "step": 2490 }, { "grad_norm": 0.4810972511768341, "learning_rate": 2.7080961407868848e-05, "loss": 0.0428, "step": 2500 }, { "grad_norm": 0.3548348546028137, "learning_rate": 2.6916143923957138e-05, "loss": 0.0435, "step": 2510 }, { "grad_norm": 0.3431377410888672, "learning_rate": 2.675124262170275e-05, "loss": 0.0336, "step": 2520 }, { "grad_norm": 0.4408491551876068, "learning_rate": 2.6586264714422836e-05, "loss": 0.0404, "step": 2530 }, { "grad_norm": 1.0779521465301514, "learning_rate": 2.6421217418785544e-05, "loss": 0.0424, "step": 2540 }, { "grad_norm": 0.40511152148246765, "learning_rate": 2.6256107954494242e-05, "loss": 0.0476, "step": 2550 }, { "grad_norm": 0.36247146129608154, "learning_rate": 2.6090943543971784e-05, "loss": 0.0436, "step": 2560 }, { "grad_norm": 0.3747636079788208, "learning_rate": 2.592573141204455e-05, "loss": 0.0404, "step": 2570 }, { "grad_norm": 0.5357540249824524, "learning_rate": 2.5760478785626397e-05, "loss": 0.0444, "step": 2580 }, { "grad_norm": 0.3767753839492798, "learning_rate": 2.5595192893402553e-05, "loss": 0.0397, "step": 2590 }, { "grad_norm": 0.3285132944583893, "learning_rate": 2.5429880965513386e-05, "loss": 0.0381, "step": 2600 }, { "grad_norm": 0.4897662401199341, "learning_rate": 2.526455023323817e-05, "loss": 0.039, "step": 2610 }, { "grad_norm": 0.4508710205554962, "learning_rate": 2.5099207928678732e-05, "loss": 0.0395, "step": 2620 }, { "grad_norm": 0.3298162519931793, "learning_rate": 2.4933861284443112e-05, "loss": 0.0413, "step": 2630 }, { "grad_norm": 0.42930713295936584, "learning_rate": 2.4768517533329157e-05, "loss": 0.0461, "step": 2640 }, { "grad_norm": 0.43155762553215027, "learning_rate": 2.460318390800819e-05, "loss": 0.0453, "step": 2650 }, { "grad_norm": 0.3605857789516449, "learning_rate": 2.44378676407086e-05, "loss": 0.0445, "step": 2660 }, { "grad_norm": 0.2702181041240692, "learning_rate": 2.4272575962899462e-05, "loss": 0.0421, "step": 2670 }, { "grad_norm": 0.3844555616378784, "learning_rate": 2.410731610497424e-05, "loss": 0.0418, "step": 2680 }, { "grad_norm": 0.4364534914493561, "learning_rate": 2.3942095295934476e-05, "loss": 0.0379, "step": 2690 }, { "grad_norm": 0.33137261867523193, "learning_rate": 2.3776920763073602e-05, "loss": 0.0422, "step": 2700 }, { "grad_norm": 0.2672342360019684, "learning_rate": 2.361179973166078e-05, "loss": 0.0424, "step": 2710 }, { "grad_norm": 0.34315600991249084, "learning_rate": 2.3446739424624827e-05, "loss": 0.0411, "step": 2720 }, { "grad_norm": 0.4923914670944214, "learning_rate": 2.328174706223832e-05, "loss": 0.0439, "step": 2730 }, { "grad_norm": 0.26800429821014404, "learning_rate": 2.3116829861801686e-05, "loss": 0.0391, "step": 2740 }, { "grad_norm": 0.29833143949508667, "learning_rate": 2.2951995037327515e-05, "loss": 0.0397, "step": 2750 }, { "grad_norm": 0.3327856957912445, "learning_rate": 2.2787249799225024e-05, "loss": 0.0381, "step": 2760 }, { "grad_norm": 0.46502891182899475, "learning_rate": 2.2622601353984635e-05, "loss": 0.0411, "step": 2770 }, { "grad_norm": 0.40785571932792664, "learning_rate": 2.2458056903862726e-05, "loss": 0.0379, "step": 2780 }, { "grad_norm": 0.3968541622161865, "learning_rate": 2.229362364656659e-05, "loss": 0.0429, "step": 2790 }, { "grad_norm": 0.40930506587028503, "learning_rate": 2.21293087749396e-05, "loss": 0.0374, "step": 2800 }, { "grad_norm": 0.4295065701007843, "learning_rate": 2.1965119476646547e-05, "loss": 0.0366, "step": 2810 }, { "grad_norm": 0.4438641667366028, "learning_rate": 2.1801062933859235e-05, "loss": 0.0378, "step": 2820 }, { "grad_norm": 0.3975362181663513, "learning_rate": 2.1637146322942315e-05, "loss": 0.036, "step": 2830 }, { "grad_norm": 0.44547799229621887, "learning_rate": 2.1473376814139362e-05, "loss": 0.0395, "step": 2840 }, { "grad_norm": 0.4006638526916504, "learning_rate": 2.1309761571259237e-05, "loss": 0.0367, "step": 2850 }, { "grad_norm": 0.27414795756340027, "learning_rate": 2.1146307751362694e-05, "loss": 0.0312, "step": 2860 }, { "grad_norm": 0.4664889872074127, "learning_rate": 2.098302250444934e-05, "loss": 0.0413, "step": 2870 }, { "grad_norm": 0.19164593517780304, "learning_rate": 2.0819912973144843e-05, "loss": 0.0345, "step": 2880 }, { "grad_norm": 0.33589157462120056, "learning_rate": 2.065698629238851e-05, "loss": 0.0355, "step": 2890 }, { "grad_norm": 0.33917805552482605, "learning_rate": 2.0494249589121157e-05, "loss": 0.0393, "step": 2900 }, { "grad_norm": 0.32117098569869995, "learning_rate": 2.033170998197339e-05, "loss": 0.0448, "step": 2910 }, { "grad_norm": 0.6140998005867004, "learning_rate": 2.0169374580954165e-05, "loss": 0.0378, "step": 2920 }, { "grad_norm": 0.5331240296363831, "learning_rate": 2.000725048713983e-05, "loss": 0.0365, "step": 2930 }, { "grad_norm": 0.3481861352920532, "learning_rate": 1.9845344792363447e-05, "loss": 0.0356, "step": 2940 }, { "grad_norm": 0.5170403718948364, "learning_rate": 1.9683664578904614e-05, "loss": 0.0403, "step": 2950 }, { "grad_norm": 0.31584399938583374, "learning_rate": 1.9522216919179644e-05, "loss": 0.0404, "step": 2960 }, { "grad_norm": 0.3115321099758148, "learning_rate": 1.9361008875432185e-05, "loss": 0.0358, "step": 2970 }, { "grad_norm": 0.338223397731781, "learning_rate": 1.9200047499424308e-05, "loss": 0.0331, "step": 2980 }, { "grad_norm": 0.48482340574264526, "learning_rate": 1.9039339832128057e-05, "loss": 0.0347, "step": 2990 }, { "grad_norm": 0.4181440770626068, "learning_rate": 1.8878892903417404e-05, "loss": 0.0389, "step": 3000 }, { "grad_norm": 0.42534828186035156, "learning_rate": 1.871871373176078e-05, "loss": 0.0392, "step": 3010 }, { "grad_norm": 0.5645197629928589, "learning_rate": 1.8558809323914083e-05, "loss": 0.0461, "step": 3020 }, { "grad_norm": 0.36499112844467163, "learning_rate": 1.8399186674614126e-05, "loss": 0.0343, "step": 3030 }, { "grad_norm": 0.2742806673049927, "learning_rate": 1.823985276627269e-05, "loss": 0.0283, "step": 3040 }, { "grad_norm": 0.39008432626724243, "learning_rate": 1.80808145686711e-05, "loss": 0.0374, "step": 3050 }, { "grad_norm": 0.36456918716430664, "learning_rate": 1.7922079038655326e-05, "loss": 0.037, "step": 3060 }, { "grad_norm": 0.2564813792705536, "learning_rate": 1.7763653119831684e-05, "loss": 0.0374, "step": 3070 }, { "grad_norm": 0.4312773048877716, "learning_rate": 1.7605543742263087e-05, "loss": 0.0353, "step": 3080 }, { "grad_norm": 0.30969807505607605, "learning_rate": 1.744775782216593e-05, "loss": 0.0339, "step": 3090 }, { "grad_norm": 0.3753385543823242, "learning_rate": 1.7290302261607503e-05, "loss": 0.0323, "step": 3100 }, { "grad_norm": 0.3404577672481537, "learning_rate": 1.7133183948204108e-05, "loss": 0.0346, "step": 3110 }, { "grad_norm": 0.33649688959121704, "learning_rate": 1.6976409754819767e-05, "loss": 0.0332, "step": 3120 }, { "grad_norm": 0.7442240715026855, "learning_rate": 1.6819986539265582e-05, "loss": 0.0392, "step": 3130 }, { "grad_norm": 0.3342534899711609, "learning_rate": 1.6663921143999733e-05, "loss": 0.0368, "step": 3140 }, { "grad_norm": 0.32587477564811707, "learning_rate": 1.650822039582819e-05, "loss": 0.0318, "step": 3150 }, { "grad_norm": 0.5487452149391174, "learning_rate": 1.63528911056061e-05, "loss": 0.0376, "step": 3160 }, { "grad_norm": 0.5345733165740967, "learning_rate": 1.619794006793979e-05, "loss": 0.0386, "step": 3170 }, { "grad_norm": 0.3318815231323242, "learning_rate": 1.604337406088963e-05, "loss": 0.0308, "step": 3180 }, { "grad_norm": 0.41096043586730957, "learning_rate": 1.588919984567349e-05, "loss": 0.036, "step": 3190 }, { "grad_norm": 0.49638691544532776, "learning_rate": 1.5735424166371e-05, "loss": 0.0404, "step": 3200 }, { "grad_norm": 0.38568755984306335, "learning_rate": 1.558205374962854e-05, "loss": 0.033, "step": 3210 }, { "grad_norm": 0.34658241271972656, "learning_rate": 1.5429095304364976e-05, "loss": 0.0327, "step": 3220 }, { "grad_norm": 0.33020010590553284, "learning_rate": 1.527655552147824e-05, "loss": 0.0373, "step": 3230 }, { "grad_norm": 0.5657230615615845, "learning_rate": 1.5124441073552585e-05, "loss": 0.0354, "step": 3240 }, { "grad_norm": 0.3239688277244568, "learning_rate": 1.4972758614566746e-05, "loss": 0.0295, "step": 3250 }, { "grad_norm": 0.387555330991745, "learning_rate": 1.4821514779602864e-05, "loss": 0.0436, "step": 3260 }, { "grad_norm": 0.3232133388519287, "learning_rate": 1.467071618455624e-05, "loss": 0.0322, "step": 3270 }, { "grad_norm": 0.2886604964733124, "learning_rate": 1.452036942584595e-05, "loss": 0.0304, "step": 3280 }, { "grad_norm": 0.2882493734359741, "learning_rate": 1.4370481080126248e-05, "loss": 0.0314, "step": 3290 }, { "grad_norm": 0.4715385138988495, "learning_rate": 1.4221057703998985e-05, "loss": 0.0334, "step": 3300 }, { "grad_norm": 0.3316683769226074, "learning_rate": 1.4072105833726684e-05, "loss": 0.0351, "step": 3310 }, { "grad_norm": 0.3624657988548279, "learning_rate": 1.3923631984946672e-05, "loss": 0.035, "step": 3320 }, { "grad_norm": 0.36212462186813354, "learning_rate": 1.3775642652386103e-05, "loss": 0.0301, "step": 3330 }, { "grad_norm": 0.36848610639572144, "learning_rate": 1.3628144309577784e-05, "loss": 0.0324, "step": 3340 }, { "grad_norm": 0.4272776246070862, "learning_rate": 1.3481143408577079e-05, "loss": 0.0373, "step": 3350 }, { "grad_norm": 0.37170878052711487, "learning_rate": 1.3334646379679583e-05, "loss": 0.033, "step": 3360 }, { "grad_norm": 0.48944494128227234, "learning_rate": 1.3188659631139966e-05, "loss": 0.0414, "step": 3370 }, { "grad_norm": 0.3089650869369507, "learning_rate": 1.3043189548891516e-05, "loss": 0.0338, "step": 3380 }, { "grad_norm": 0.3926386535167694, "learning_rate": 1.2898242496266884e-05, "loss": 0.0351, "step": 3390 }, { "grad_norm": 0.40642455220222473, "learning_rate": 1.2753824813719733e-05, "loss": 0.0412, "step": 3400 }, { "grad_norm": 0.35056406259536743, "learning_rate": 1.2609942818547327e-05, "loss": 0.035, "step": 3410 }, { "grad_norm": 0.2923920750617981, "learning_rate": 1.2466602804614266e-05, "loss": 0.035, "step": 3420 }, { "grad_norm": 0.9390926957130432, "learning_rate": 1.2323811042077096e-05, "loss": 0.0298, "step": 3430 }, { "grad_norm": 0.5153149366378784, "learning_rate": 1.2181573777110106e-05, "loss": 0.0376, "step": 3440 }, { "grad_norm": 0.2597624659538269, "learning_rate": 1.2039897231632055e-05, "loss": 0.0333, "step": 3450 }, { "grad_norm": 0.36722615361213684, "learning_rate": 1.1898787603033995e-05, "loss": 0.038, "step": 3460 }, { "grad_norm": 0.23120303452014923, "learning_rate": 1.1758251063908227e-05, "loss": 0.0305, "step": 3470 }, { "grad_norm": 0.4411832094192505, "learning_rate": 1.1618293761778235e-05, "loss": 0.0305, "step": 3480 }, { "grad_norm": 0.36204567551612854, "learning_rate": 1.1478921818829827e-05, "loss": 0.0325, "step": 3490 }, { "grad_norm": 0.38650524616241455, "learning_rate": 1.1340141331643276e-05, "loss": 0.037, "step": 3500 }, { "grad_norm": 0.29249370098114014, "learning_rate": 1.1201958370926682e-05, "loss": 0.0304, "step": 3510 }, { "grad_norm": 0.3050512969493866, "learning_rate": 1.1064378981250409e-05, "loss": 0.0322, "step": 3520 }, { "grad_norm": 0.36122196912765503, "learning_rate": 1.0927409180782638e-05, "loss": 0.0325, "step": 3530 }, { "grad_norm": 0.3820047974586487, "learning_rate": 1.0791054961026182e-05, "loss": 0.0361, "step": 3540 }, { "grad_norm": 0.3520725965499878, "learning_rate": 1.0655322286556318e-05, "loss": 0.0333, "step": 3550 }, { "grad_norm": 0.3393506705760956, "learning_rate": 1.0520217094759962e-05, "loss": 0.0352, "step": 3560 }, { "grad_norm": 0.3986015021800995, "learning_rate": 1.0385745295575866e-05, "loss": 0.0331, "step": 3570 }, { "grad_norm": 0.3583716154098511, "learning_rate": 1.0251912771236159e-05, "loss": 0.0319, "step": 3580 }, { "grad_norm": 0.27882295846939087, "learning_rate": 1.0118725376009025e-05, "loss": 0.0341, "step": 3590 }, { "grad_norm": 0.32304292917251587, "learning_rate": 9.986188935942579e-06, "loss": 0.0291, "step": 3600 }, { "grad_norm": 0.5138395428657532, "learning_rate": 9.854309248610085e-06, "loss": 0.0313, "step": 3610 }, { "grad_norm": 0.31992360949516296, "learning_rate": 9.723092082856294e-06, "loss": 0.0305, "step": 3620 }, { "grad_norm": 0.2782381772994995, "learning_rate": 9.592543178545107e-06, "loss": 0.029, "step": 3630 }, { "grad_norm": 0.27642911672592163, "learning_rate": 9.462668246308529e-06, "loss": 0.0329, "step": 3640 }, { "grad_norm": 0.33659425377845764, "learning_rate": 9.333472967296833e-06, "loss": 0.0328, "step": 3650 }, { "grad_norm": 0.40272244811058044, "learning_rate": 9.204962992930063e-06, "loss": 0.0284, "step": 3660 }, { "grad_norm": 0.4781797528266907, "learning_rate": 9.077143944650802e-06, "loss": 0.0323, "step": 3670 }, { "grad_norm": 0.34181609749794006, "learning_rate": 8.950021413678306e-06, "loss": 0.0305, "step": 3680 }, { "grad_norm": 0.3515138328075409, "learning_rate": 8.8236009607639e-06, "loss": 0.0311, "step": 3690 }, { "grad_norm": 0.42265385389328003, "learning_rate": 8.697888115947725e-06, "loss": 0.0294, "step": 3700 }, { "grad_norm": 0.24089132249355316, "learning_rate": 8.572888378316884e-06, "loss": 0.0346, "step": 3710 }, { "grad_norm": 0.3806060254573822, "learning_rate": 8.448607215764823e-06, "loss": 0.0274, "step": 3720 }, { "grad_norm": 0.40154969692230225, "learning_rate": 8.325050064752238e-06, "loss": 0.0321, "step": 3730 }, { "grad_norm": 0.6564791798591614, "learning_rate": 8.202222330069167e-06, "loss": 0.0324, "step": 3740 }, { "grad_norm": 0.4108399748802185, "learning_rate": 8.080129384598625e-06, "loss": 0.0301, "step": 3750 }, { "grad_norm": 0.19246874749660492, "learning_rate": 7.958776569081586e-06, "loss": 0.0288, "step": 3760 }, { "grad_norm": 0.4110735058784485, "learning_rate": 7.838169191883316e-06, "loss": 0.0382, "step": 3770 }, { "grad_norm": 0.25348952412605286, "learning_rate": 7.718312528761223e-06, "loss": 0.0311, "step": 3780 }, { "grad_norm": 0.3308151662349701, "learning_rate": 7.59921182263402e-06, "loss": 0.0274, "step": 3790 }, { "grad_norm": 0.28520438075065613, "learning_rate": 7.480872283352428e-06, "loss": 0.0309, "step": 3800 }, { "grad_norm": 0.337143212556839, "learning_rate": 7.363299087471276e-06, "loss": 0.0259, "step": 3810 }, { "grad_norm": 0.22851786017417908, "learning_rate": 7.246497378023018e-06, "loss": 0.0323, "step": 3820 }, { "grad_norm": 0.5074667930603027, "learning_rate": 7.130472264292823e-06, "loss": 0.0292, "step": 3830 }, { "grad_norm": 0.2871147096157074, "learning_rate": 7.015228821595024e-06, "loss": 0.0296, "step": 3840 }, { "grad_norm": 0.30016905069351196, "learning_rate": 6.9007720910511554e-06, "loss": 0.0231, "step": 3850 }, { "grad_norm": 0.28288936614990234, "learning_rate": 6.7871070793693815e-06, "loss": 0.033, "step": 3860 }, { "grad_norm": 0.44114962220191956, "learning_rate": 6.6742387586255505e-06, "loss": 0.0343, "step": 3870 }, { "grad_norm": 0.22615277767181396, "learning_rate": 6.562172066045655e-06, "loss": 0.0267, "step": 3880 }, { "grad_norm": 0.24401821196079254, "learning_rate": 6.450911903789863e-06, "loss": 0.0322, "step": 3890 }, { "grad_norm": 0.4524429738521576, "learning_rate": 6.340463138738123e-06, "loss": 0.0292, "step": 3900 }, { "grad_norm": 0.3167012929916382, "learning_rate": 6.230830602277199e-06, "loss": 0.0325, "step": 3910 }, { "grad_norm": 0.2682662010192871, "learning_rate": 6.1220190900894175e-06, "loss": 0.0273, "step": 3920 }, { "grad_norm": 0.29721036553382874, "learning_rate": 6.014033361942806e-06, "loss": 0.0266, "step": 3930 }, { "grad_norm": 0.42319223284721375, "learning_rate": 5.906878141482944e-06, "loss": 0.0322, "step": 3940 }, { "grad_norm": 0.38211673498153687, "learning_rate": 5.800558116026319e-06, "loss": 0.0312, "step": 3950 }, { "grad_norm": 0.27969926595687866, "learning_rate": 5.695077936355259e-06, "loss": 0.0263, "step": 3960 }, { "grad_norm": 0.22936464846134186, "learning_rate": 5.590442216514543e-06, "loss": 0.0309, "step": 3970 }, { "grad_norm": 0.3358422517776489, "learning_rate": 5.48665553360952e-06, "loss": 0.0243, "step": 3980 }, { "grad_norm": 0.2455916702747345, "learning_rate": 5.383722427605931e-06, "loss": 0.0293, "step": 3990 }, { "grad_norm": 0.24449455738067627, "learning_rate": 5.281647401131279e-06, "loss": 0.0283, "step": 4000 }, { "grad_norm": 0.40746790170669556, "learning_rate": 5.1804349192779044e-06, "loss": 0.0324, "step": 4010 }, { "grad_norm": 0.3240891695022583, "learning_rate": 5.0800894094076565e-06, "loss": 0.024, "step": 4020 }, { "grad_norm": 0.25076866149902344, "learning_rate": 4.980615260958193e-06, "loss": 0.0328, "step": 4030 }, { "grad_norm": 0.1966366469860077, "learning_rate": 4.882016825251037e-06, "loss": 0.0303, "step": 4040 }, { "grad_norm": 0.2640022337436676, "learning_rate": 4.784298415301172e-06, "loss": 0.0254, "step": 4050 }, { "grad_norm": 0.33577871322631836, "learning_rate": 4.687464305628405e-06, "loss": 0.0296, "step": 4060 }, { "grad_norm": 0.40677428245544434, "learning_rate": 4.591518732070402e-06, "loss": 0.0278, "step": 4070 }, { "grad_norm": 0.46990591287612915, "learning_rate": 4.4964658915973676e-06, "loss": 0.0279, "step": 4080 }, { "grad_norm": 0.22323285043239594, "learning_rate": 4.40230994212848e-06, "loss": 0.0293, "step": 4090 }, { "grad_norm": 0.2634666860103607, "learning_rate": 4.309055002349987e-06, "loss": 0.0303, "step": 4100 }, { "grad_norm": 0.45921197533607483, "learning_rate": 4.21670515153505e-06, "loss": 0.0297, "step": 4110 }, { "grad_norm": 0.640613853931427, "learning_rate": 4.12526442936533e-06, "loss": 0.0241, "step": 4120 }, { "grad_norm": 0.3607765734195709, "learning_rate": 4.03473683575423e-06, "loss": 0.0263, "step": 4130 }, { "grad_norm": 0.3656626045703888, "learning_rate": 3.945126330671969e-06, "loss": 0.028, "step": 4140 }, { "grad_norm": 0.8340692520141602, "learning_rate": 3.856436833972341e-06, "loss": 0.0298, "step": 4150 }, { "grad_norm": 0.5998902320861816, "learning_rate": 3.7686722252212344e-06, "loss": 0.0285, "step": 4160 }, { "grad_norm": 0.36758774518966675, "learning_rate": 3.681836343526965e-06, "loss": 0.0329, "step": 4170 }, { "grad_norm": 0.31286585330963135, "learning_rate": 3.5959329873723e-06, "loss": 0.0296, "step": 4180 }, { "grad_norm": 0.38393154740333557, "learning_rate": 3.510965914448333e-06, "loss": 0.0329, "step": 4190 }, { "grad_norm": 0.396363228559494, "learning_rate": 3.4269388414900792e-06, "loss": 0.0317, "step": 4200 }, { "grad_norm": 0.222749263048172, "learning_rate": 3.3438554441139247e-06, "loss": 0.0223, "step": 4210 }, { "grad_norm": 0.24695293605327606, "learning_rate": 3.2617193566568283e-06, "loss": 0.0241, "step": 4220 }, { "grad_norm": 0.4028778374195099, "learning_rate": 3.1805341720173326e-06, "loss": 0.0226, "step": 4230 }, { "grad_norm": 0.2558240592479706, "learning_rate": 3.100303441498423e-06, "loss": 0.0315, "step": 4240 }, { "grad_norm": 0.5111129283905029, "learning_rate": 3.021030674652156e-06, "loss": 0.0291, "step": 4250 }, { "grad_norm": 0.28568732738494873, "learning_rate": 2.942719339126171e-06, "loss": 0.0288, "step": 4260 }, { "grad_norm": 0.9272612929344177, "learning_rate": 2.8653728605119695e-06, "loss": 0.0265, "step": 4270 }, { "grad_norm": 0.2502414584159851, "learning_rate": 2.788994622195096e-06, "loss": 0.0257, "step": 4280 }, { "grad_norm": 0.3649044930934906, "learning_rate": 2.7135879652071318e-06, "loss": 0.0257, "step": 4290 }, { "grad_norm": 0.26225656270980835, "learning_rate": 2.6391561880795253e-06, "loss": 0.0256, "step": 4300 }, { "grad_norm": 0.38593482971191406, "learning_rate": 2.5657025466993472e-06, "loss": 0.0238, "step": 4310 }, { "grad_norm": 0.24045054614543915, "learning_rate": 2.493230254166817e-06, "loss": 0.0275, "step": 4320 }, { "grad_norm": 0.33231034874916077, "learning_rate": 2.4217424806547984e-06, "loss": 0.0251, "step": 4330 }, { "grad_norm": 0.31093278527259827, "learning_rate": 2.3512423532700805e-06, "loss": 0.0244, "step": 4340 }, { "grad_norm": 0.3532539904117584, "learning_rate": 2.2817329559166293e-06, "loss": 0.0252, "step": 4350 }, { "grad_norm": 0.5126403570175171, "learning_rate": 2.213217329160672e-06, "loss": 0.0263, "step": 4360 }, { "grad_norm": 0.3334461450576782, "learning_rate": 2.1456984700976733e-06, "loss": 0.0315, "step": 4370 }, { "grad_norm": 0.3070288598537445, "learning_rate": 2.079179332221265e-06, "loss": 0.0283, "step": 4380 }, { "grad_norm": 0.4461181163787842, "learning_rate": 2.0136628252940217e-06, "loss": 0.0274, "step": 4390 }, { "grad_norm": 0.6602442264556885, "learning_rate": 1.9491518152202096e-06, "loss": 0.0317, "step": 4400 }, { "grad_norm": 0.2886788547039032, "learning_rate": 1.8856491239203938e-06, "loss": 0.0253, "step": 4410 }, { "grad_norm": 0.3347417116165161, "learning_rate": 1.8231575292080027e-06, "loss": 0.0316, "step": 4420 }, { "grad_norm": 0.3918112814426422, "learning_rate": 1.7616797646678479e-06, "loss": 0.0289, "step": 4430 }, { "grad_norm": 0.41051027178764343, "learning_rate": 1.7012185195365016e-06, "loss": 0.0304, "step": 4440 }, { "grad_norm": 0.2741791903972626, "learning_rate": 1.6417764385846996e-06, "loss": 0.0241, "step": 4450 }, { "grad_norm": 0.337773859500885, "learning_rate": 1.5833561220016253e-06, "loss": 0.0276, "step": 4460 }, { "grad_norm": 0.394645631313324, "learning_rate": 1.5259601252811755e-06, "loss": 0.0229, "step": 4470 }, { "grad_norm": 0.26460105180740356, "learning_rate": 1.4695909591101924e-06, "loss": 0.0275, "step": 4480 }, { "grad_norm": 0.30859678983688354, "learning_rate": 1.4142510892586113e-06, "loss": 0.0255, "step": 4490 }, { "grad_norm": 0.3854379951953888, "learning_rate": 1.3599429364716144e-06, "loss": 0.0276, "step": 4500 }, { "grad_norm": 0.2807616591453552, "learning_rate": 1.3066688763637453e-06, "loss": 0.028, "step": 4510 }, { "grad_norm": 0.364891916513443, "learning_rate": 1.2544312393149683e-06, "loss": 0.0258, "step": 4520 }, { "grad_norm": 0.8333351016044617, "learning_rate": 1.2032323103687654e-06, "loss": 0.0302, "step": 4530 }, { "grad_norm": 0.31938087940216064, "learning_rate": 1.1530743291321367e-06, "loss": 0.0255, "step": 4540 }, { "grad_norm": 0.2908027172088623, "learning_rate": 1.1039594896776834e-06, "loss": 0.0227, "step": 4550 }, { "grad_norm": 0.2906731963157654, "learning_rate": 1.0558899404475824e-06, "loss": 0.0254, "step": 4560 }, { "grad_norm": 0.4353867471218109, "learning_rate": 1.008867784159645e-06, "loss": 0.0267, "step": 4570 }, { "grad_norm": 0.30952566862106323, "learning_rate": 9.628950777153256e-07, "loss": 0.0262, "step": 4580 }, { "grad_norm": 0.7123703956604004, "learning_rate": 9.179738321097225e-07, "loss": 0.0262, "step": 4590 }, { "grad_norm": 0.4011175334453583, "learning_rate": 8.741060123436535e-07, "loss": 0.0232, "step": 4600 }, { "grad_norm": 0.305926114320755, "learning_rate": 8.312935373376573e-07, "loss": 0.0259, "step": 4610 }, { "grad_norm": 0.33336731791496277, "learning_rate": 7.895382798480927e-07, "loss": 0.0283, "step": 4620 }, { "grad_norm": 0.3118588328361511, "learning_rate": 7.488420663851859e-07, "loss": 0.0224, "step": 4630 }, { "grad_norm": 0.21298760175704956, "learning_rate": 7.092066771331507e-07, "loss": 0.0282, "step": 4640 }, { "grad_norm": 0.4557483196258545, "learning_rate": 6.706338458723205e-07, "loss": 0.0277, "step": 4650 }, { "grad_norm": 0.33916375041007996, "learning_rate": 6.331252599032833e-07, "loss": 0.0322, "step": 4660 }, { "grad_norm": 0.36580178141593933, "learning_rate": 5.96682559973108e-07, "loss": 0.0234, "step": 4670 }, { "grad_norm": 0.23574711382389069, "learning_rate": 5.613073402035429e-07, "loss": 0.0229, "step": 4680 }, { "grad_norm": 0.34600311517715454, "learning_rate": 5.270011480213056e-07, "loss": 0.0332, "step": 4690 }, { "grad_norm": 0.28487834334373474, "learning_rate": 4.937654840903722e-07, "loss": 0.0355, "step": 4700 }, { "grad_norm": 0.36917445063591003, "learning_rate": 4.6160180224635305e-07, "loss": 0.0245, "step": 4710 }, { "grad_norm": 0.2571835517883301, "learning_rate": 4.3051150943289596e-07, "loss": 0.0265, "step": 4720 }, { "grad_norm": 0.3375624716281891, "learning_rate": 4.004959656401186e-07, "loss": 0.0274, "step": 4730 }, { "grad_norm": 0.35926106572151184, "learning_rate": 3.715564838451452e-07, "loss": 0.0266, "step": 4740 }, { "grad_norm": 0.4017363488674164, "learning_rate": 3.436943299546608e-07, "loss": 0.0278, "step": 4750 }, { "grad_norm": 0.2312772423028946, "learning_rate": 3.169107227495388e-07, "loss": 0.0256, "step": 4760 }, { "grad_norm": 0.2088507115840912, "learning_rate": 2.912068338315255e-07, "loss": 0.0277, "step": 4770 }, { "grad_norm": 0.168931245803833, "learning_rate": 2.6658378757198623e-07, "loss": 0.0234, "step": 4780 }, { "grad_norm": 0.43259093165397644, "learning_rate": 2.430426610627395e-07, "loss": 0.0248, "step": 4790 }, { "grad_norm": 0.45266538858413696, "learning_rate": 2.205844840689142e-07, "loss": 0.0298, "step": 4800 }, { "grad_norm": 0.41012969613075256, "learning_rate": 1.9921023898393233e-07, "loss": 0.0258, "step": 4810 }, { "grad_norm": 0.4440428614616394, "learning_rate": 1.7892086078651616e-07, "loss": 0.021, "step": 4820 }, { "grad_norm": 0.3820282518863678, "learning_rate": 1.5971723699979013e-07, "loss": 0.0274, "step": 4830 }, { "grad_norm": 0.166753888130188, "learning_rate": 1.4160020765247862e-07, "loss": 0.0338, "step": 4840 }, { "grad_norm": 0.2630172669887543, "learning_rate": 1.2457056524212696e-07, "loss": 0.0303, "step": 4850 }, { "grad_norm": 0.28150999546051025, "learning_rate": 1.0862905470047091e-07, "loss": 0.0248, "step": 4860 }, { "grad_norm": 0.19534726440906525, "learning_rate": 9.377637336082101e-08, "loss": 0.0218, "step": 4870 }, { "grad_norm": 0.21588121354579926, "learning_rate": 8.001317092757864e-08, "loss": 0.0292, "step": 4880 }, { "grad_norm": 0.3922027349472046, "learning_rate": 6.73400494477977e-08, "loss": 0.0273, "step": 4890 }, { "grad_norm": 0.3174291253089905, "learning_rate": 5.5757563284875026e-08, "loss": 0.0276, "step": 4900 }, { "grad_norm": 0.24576455354690552, "learning_rate": 4.526621909426987e-08, "loss": 0.0259, "step": 4910 }, { "grad_norm": 0.22395384311676025, "learning_rate": 3.586647580136881e-08, "loss": 0.023, "step": 4920 }, { "grad_norm": 0.8619683980941772, "learning_rate": 2.755874458139629e-08, "loss": 0.0288, "step": 4930 }, { "grad_norm": 0.31981348991394043, "learning_rate": 2.034338884142617e-08, "loss": 0.0246, "step": 4940 }, { "grad_norm": 0.27290672063827515, "learning_rate": 1.422072420449172e-08, "loss": 0.0277, "step": 4950 }, { "grad_norm": 0.2460501492023468, "learning_rate": 9.191018495779968e-09, "loss": 0.0247, "step": 4960 }, { "grad_norm": 0.27399739623069763, "learning_rate": 5.254491730916078e-09, "loss": 0.0333, "step": 4970 }, { "grad_norm": 0.3019241988658905, "learning_rate": 2.411316106326611e-09, "loss": 0.0295, "step": 4980 }, { "grad_norm": 0.2698662281036377, "learning_rate": 6.616159917316456e-10, "loss": 0.0259, "step": 4990 }, { "grad_norm": 0.282757043838501, "learning_rate": 5.4679246769273606e-12, "loss": 0.0249, "step": 5000 } ], "logging_steps": 10, "max_steps": 5000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 5000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 32, "trial_name": null, "trial_params": null }