{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.8, "eval_steps": 500, "global_step": 8000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 2.6049234867095947, "learning_rate": 1.8e-06, "loss": 1.1662, "step": 10 }, { "grad_norm": 1.6405389308929443, "learning_rate": 3.8e-06, "loss": 1.1296, "step": 20 }, { "grad_norm": 0.8382836580276489, "learning_rate": 5.8e-06, "loss": 1.0685, "step": 30 }, { "grad_norm": 0.8227774500846863, "learning_rate": 7.8e-06, "loss": 1.0499, "step": 40 }, { "grad_norm": 0.6841750144958496, "learning_rate": 9.800000000000001e-06, "loss": 1.0617, "step": 50 }, { "grad_norm": 0.6164016127586365, "learning_rate": 1.18e-05, "loss": 1.067, "step": 60 }, { "grad_norm": 0.7090155482292175, "learning_rate": 1.3800000000000002e-05, "loss": 1.0641, "step": 70 }, { "grad_norm": 0.7555773854255676, "learning_rate": 1.58e-05, "loss": 1.0681, "step": 80 }, { "grad_norm": 0.5217806696891785, "learning_rate": 1.78e-05, "loss": 1.0462, "step": 90 }, { "grad_norm": 0.4389437139034271, "learning_rate": 1.9800000000000004e-05, "loss": 1.053, "step": 100 }, { "grad_norm": 0.4568557143211365, "learning_rate": 2.18e-05, "loss": 1.047, "step": 110 }, { "grad_norm": 0.6062548756599426, "learning_rate": 2.38e-05, "loss": 1.031, "step": 120 }, { "grad_norm": 1.3553529977798462, "learning_rate": 2.58e-05, "loss": 1.0128, "step": 130 }, { "grad_norm": 1.2336472272872925, "learning_rate": 2.7800000000000005e-05, "loss": 0.9739, "step": 140 }, { "grad_norm": 0.7140560150146484, "learning_rate": 2.98e-05, "loss": 0.9196, "step": 150 }, { "grad_norm": 1.0905883312225342, "learning_rate": 3.18e-05, "loss": 0.8988, "step": 160 }, { "grad_norm": 0.7075161933898926, "learning_rate": 3.38e-05, "loss": 0.8739, "step": 170 }, { "grad_norm": 1.2705228328704834, "learning_rate": 3.58e-05, "loss": 0.8568, "step": 180 }, { "grad_norm": 1.9666781425476074, "learning_rate": 3.7800000000000004e-05, "loss": 0.7906, "step": 190 }, { "grad_norm": 2.7811009883880615, "learning_rate": 3.9800000000000005e-05, "loss": 0.6826, "step": 200 }, { "grad_norm": 1.8301430940628052, "learning_rate": 4.18e-05, "loss": 0.5441, "step": 210 }, { "grad_norm": 2.3606672286987305, "learning_rate": 4.38e-05, "loss": 0.421, "step": 220 }, { "grad_norm": 2.5141494274139404, "learning_rate": 4.58e-05, "loss": 0.3251, "step": 230 }, { "grad_norm": 1.962906837463379, "learning_rate": 4.78e-05, "loss": 0.2312, "step": 240 }, { "grad_norm": 1.4764026403427124, "learning_rate": 4.9800000000000004e-05, "loss": 0.1811, "step": 250 }, { "grad_norm": 1.6381537914276123, "learning_rate": 5.1800000000000005e-05, "loss": 0.1645, "step": 260 }, { "grad_norm": 1.7995526790618896, "learning_rate": 5.380000000000001e-05, "loss": 0.1645, "step": 270 }, { "grad_norm": 0.9505502581596375, "learning_rate": 5.580000000000001e-05, "loss": 0.1424, "step": 280 }, { "grad_norm": 1.2174862623214722, "learning_rate": 5.7799999999999995e-05, "loss": 0.1549, "step": 290 }, { "grad_norm": 2.0738980770111084, "learning_rate": 5.9800000000000003e-05, "loss": 0.1343, "step": 300 }, { "grad_norm": 1.2634624242782593, "learning_rate": 6.18e-05, "loss": 0.1367, "step": 310 }, { "grad_norm": 1.127558946609497, "learning_rate": 6.38e-05, "loss": 0.1196, "step": 320 }, { "grad_norm": 1.222124457359314, "learning_rate": 6.58e-05, "loss": 0.1038, "step": 330 }, { "grad_norm": 1.3810409307479858, "learning_rate": 6.780000000000001e-05, "loss": 0.1135, "step": 340 }, { "grad_norm": 1.218234896659851, "learning_rate": 6.98e-05, "loss": 0.1112, "step": 350 }, { "grad_norm": 0.7986820340156555, "learning_rate": 7.18e-05, "loss": 0.1173, "step": 360 }, { "grad_norm": 1.1645500659942627, "learning_rate": 7.38e-05, "loss": 0.115, "step": 370 }, { "grad_norm": 0.8307706713676453, "learning_rate": 7.58e-05, "loss": 0.1034, "step": 380 }, { "grad_norm": 0.8734357953071594, "learning_rate": 7.780000000000001e-05, "loss": 0.1094, "step": 390 }, { "grad_norm": 0.7837036848068237, "learning_rate": 7.98e-05, "loss": 0.1082, "step": 400 }, { "grad_norm": 0.9293273687362671, "learning_rate": 8.18e-05, "loss": 0.1102, "step": 410 }, { "grad_norm": 0.6570233106613159, "learning_rate": 8.38e-05, "loss": 0.0981, "step": 420 }, { "grad_norm": 1.1479554176330566, "learning_rate": 8.58e-05, "loss": 0.112, "step": 430 }, { "grad_norm": 1.0603384971618652, "learning_rate": 8.78e-05, "loss": 0.1082, "step": 440 }, { "grad_norm": 0.7609360814094543, "learning_rate": 8.98e-05, "loss": 0.1067, "step": 450 }, { "grad_norm": 0.723868191242218, "learning_rate": 9.180000000000001e-05, "loss": 0.1169, "step": 460 }, { "grad_norm": 0.7048567533493042, "learning_rate": 9.38e-05, "loss": 0.1074, "step": 470 }, { "grad_norm": 0.6640262603759766, "learning_rate": 9.58e-05, "loss": 0.1073, "step": 480 }, { "grad_norm": 0.7421982884407043, "learning_rate": 9.78e-05, "loss": 0.1017, "step": 490 }, { "grad_norm": 0.6486691236495972, "learning_rate": 9.98e-05, "loss": 0.1053, "step": 500 }, { "grad_norm": 0.7828828692436218, "learning_rate": 9.9999778549206e-05, "loss": 0.1081, "step": 510 }, { "grad_norm": 0.770983874797821, "learning_rate": 9.999901304280685e-05, "loss": 0.0906, "step": 520 }, { "grad_norm": 0.5513870716094971, "learning_rate": 9.999770075521164e-05, "loss": 0.0977, "step": 530 }, { "grad_norm": 0.41549554467201233, "learning_rate": 9.99958417007713e-05, "loss": 0.086, "step": 540 }, { "grad_norm": 0.624375581741333, "learning_rate": 9.999343589981615e-05, "loss": 0.0899, "step": 550 }, { "grad_norm": 0.5545937418937683, "learning_rate": 9.999048337865568e-05, "loss": 0.0789, "step": 560 }, { "grad_norm": 0.7582664489746094, "learning_rate": 9.998698416957815e-05, "loss": 0.0875, "step": 570 }, { "grad_norm": 0.2887002229690552, "learning_rate": 9.998293831085037e-05, "loss": 0.0838, "step": 580 }, { "grad_norm": 0.7157890200614929, "learning_rate": 9.997834584671719e-05, "loss": 0.084, "step": 590 }, { "grad_norm": 0.48241758346557617, "learning_rate": 9.997320682740107e-05, "loss": 0.0883, "step": 600 }, { "grad_norm": 0.49710673093795776, "learning_rate": 9.996752130910149e-05, "loss": 0.0941, "step": 610 }, { "grad_norm": 0.6502618789672852, "learning_rate": 9.99612893539944e-05, "loss": 0.0811, "step": 620 }, { "grad_norm": 0.48173755407333374, "learning_rate": 9.995451103023144e-05, "loss": 0.0772, "step": 630 }, { "grad_norm": 0.452338844537735, "learning_rate": 9.994718641193928e-05, "loss": 0.0721, "step": 640 }, { "grad_norm": 0.6506068110466003, "learning_rate": 9.993931557921874e-05, "loss": 0.0754, "step": 650 }, { "grad_norm": 0.49155664443969727, "learning_rate": 9.993089861814402e-05, "loss": 0.0793, "step": 660 }, { "grad_norm": 0.5275540351867676, "learning_rate": 9.992193562076166e-05, "loss": 0.0867, "step": 670 }, { "grad_norm": 0.5997722744941711, "learning_rate": 9.991242668508954e-05, "loss": 0.0831, "step": 680 }, { "grad_norm": 0.47254955768585205, "learning_rate": 9.990237191511587e-05, "loss": 0.0767, "step": 690 }, { "grad_norm": 0.5700637698173523, "learning_rate": 9.989177142079802e-05, "loss": 0.079, "step": 700 }, { "grad_norm": 0.5925238132476807, "learning_rate": 9.988062531806126e-05, "loss": 0.0812, "step": 710 }, { "grad_norm": 0.5243363380432129, "learning_rate": 9.986893372879762e-05, "loss": 0.0718, "step": 720 }, { "grad_norm": 0.42261287569999695, "learning_rate": 9.985669678086443e-05, "loss": 0.0677, "step": 730 }, { "grad_norm": 0.4768967032432556, "learning_rate": 9.984391460808298e-05, "loss": 0.0639, "step": 740 }, { "grad_norm": 0.4327617883682251, "learning_rate": 9.983058735023709e-05, "loss": 0.0654, "step": 750 }, { "grad_norm": 0.3972388803958893, "learning_rate": 9.98167151530715e-05, "loss": 0.067, "step": 760 }, { "grad_norm": 0.5228298902511597, "learning_rate": 9.980229816829034e-05, "loss": 0.0686, "step": 770 }, { "grad_norm": 0.693810224533081, "learning_rate": 9.978733655355544e-05, "loss": 0.0693, "step": 780 }, { "grad_norm": 0.7045840620994568, "learning_rate": 9.977183047248464e-05, "loss": 0.0721, "step": 790 }, { "grad_norm": 0.555489718914032, "learning_rate": 9.975578009464992e-05, "loss": 0.0707, "step": 800 }, { "grad_norm": 0.4445370137691498, "learning_rate": 9.97391855955757e-05, "loss": 0.071, "step": 810 }, { "grad_norm": 0.5154445767402649, "learning_rate": 9.972204715673669e-05, "loss": 0.0642, "step": 820 }, { "grad_norm": 0.6851938962936401, "learning_rate": 9.970436496555617e-05, "loss": 0.0753, "step": 830 }, { "grad_norm": 0.4723260998725891, "learning_rate": 9.968613921540373e-05, "loss": 0.0663, "step": 840 }, { "grad_norm": 0.4631851315498352, "learning_rate": 9.966737010559326e-05, "loss": 0.0741, "step": 850 }, { "grad_norm": 0.41511446237564087, "learning_rate": 9.964805784138072e-05, "loss": 0.066, "step": 860 }, { "grad_norm": 0.4786284863948822, "learning_rate": 9.962820263396195e-05, "loss": 0.0614, "step": 870 }, { "grad_norm": 0.5274662375450134, "learning_rate": 9.960780470047033e-05, "loss": 0.0574, "step": 880 }, { "grad_norm": 0.36581721901893616, "learning_rate": 9.958686426397437e-05, "loss": 0.0589, "step": 890 }, { "grad_norm": 0.41146519780158997, "learning_rate": 9.956538155347534e-05, "loss": 0.07, "step": 900 }, { "grad_norm": 0.6479470133781433, "learning_rate": 9.95433568039047e-05, "loss": 0.0686, "step": 910 }, { "grad_norm": 0.5151736736297607, "learning_rate": 9.952079025612162e-05, "loss": 0.0767, "step": 920 }, { "grad_norm": 0.5936368703842163, "learning_rate": 9.949768215691022e-05, "loss": 0.0754, "step": 930 }, { "grad_norm": 0.37914860248565674, "learning_rate": 9.9474032758977e-05, "loss": 0.0741, "step": 940 }, { "grad_norm": 0.538762092590332, "learning_rate": 9.944984232094794e-05, "loss": 0.0711, "step": 950 }, { "grad_norm": 0.34496554732322693, "learning_rate": 9.942511110736584e-05, "loss": 0.0619, "step": 960 }, { "grad_norm": 0.42393702268600464, "learning_rate": 9.939983938868726e-05, "loss": 0.066, "step": 970 }, { "grad_norm": 0.40594398975372314, "learning_rate": 9.93740274412797e-05, "loss": 0.0615, "step": 980 }, { "grad_norm": 0.3449326753616333, "learning_rate": 9.934767554741846e-05, "loss": 0.0661, "step": 990 }, { "grad_norm": 0.3900584876537323, "learning_rate": 9.932078399528361e-05, "loss": 0.063, "step": 1000 }, { "grad_norm": 0.5237119793891907, "learning_rate": 9.929335307895689e-05, "loss": 0.061, "step": 1010 }, { "grad_norm": 0.37187719345092773, "learning_rate": 9.926538309841839e-05, "loss": 0.0618, "step": 1020 }, { "grad_norm": 0.47879940271377563, "learning_rate": 9.923687435954334e-05, "loss": 0.0627, "step": 1030 }, { "grad_norm": 0.4103453755378723, "learning_rate": 9.920782717409873e-05, "loss": 0.0575, "step": 1040 }, { "grad_norm": 0.36794260144233704, "learning_rate": 9.917824185973994e-05, "loss": 0.0554, "step": 1050 }, { "grad_norm": 0.4100784659385681, "learning_rate": 9.914811874000723e-05, "loss": 0.0643, "step": 1060 }, { "grad_norm": 0.5202441215515137, "learning_rate": 9.911745814432218e-05, "loss": 0.0633, "step": 1070 }, { "grad_norm": 0.3642865717411041, "learning_rate": 9.90862604079842e-05, "loss": 0.0672, "step": 1080 }, { "grad_norm": 0.42079517245292664, "learning_rate": 9.90545258721667e-05, "loss": 0.0529, "step": 1090 }, { "grad_norm": 0.5196419358253479, "learning_rate": 9.90222548839135e-05, "loss": 0.0606, "step": 1100 }, { "grad_norm": 0.2861085832118988, "learning_rate": 9.898944779613495e-05, "loss": 0.0677, "step": 1110 }, { "grad_norm": 0.4128720462322235, "learning_rate": 9.89561049676041e-05, "loss": 0.0597, "step": 1120 }, { "grad_norm": 0.42165419459342957, "learning_rate": 9.89222267629528e-05, "loss": 0.069, "step": 1130 }, { "grad_norm": 0.5143876671791077, "learning_rate": 9.888781355266763e-05, "loss": 0.0581, "step": 1140 }, { "grad_norm": 0.36486557126045227, "learning_rate": 9.885286571308598e-05, "loss": 0.0588, "step": 1150 }, { "grad_norm": 0.5994070768356323, "learning_rate": 9.881738362639182e-05, "loss": 0.0655, "step": 1160 }, { "grad_norm": 0.5607981085777283, "learning_rate": 9.878136768061154e-05, "loss": 0.0578, "step": 1170 }, { "grad_norm": 0.2817396819591522, "learning_rate": 9.874481826960979e-05, "loss": 0.0628, "step": 1180 }, { "grad_norm": 0.3436738848686218, "learning_rate": 9.870773579308503e-05, "loss": 0.0636, "step": 1190 }, { "grad_norm": 0.41256147623062134, "learning_rate": 9.867012065656533e-05, "loss": 0.0609, "step": 1200 }, { "grad_norm": 0.47535645961761475, "learning_rate": 9.863197327140376e-05, "loss": 0.0615, "step": 1210 }, { "grad_norm": 0.4813234508037567, "learning_rate": 9.859329405477403e-05, "loss": 0.0596, "step": 1220 }, { "grad_norm": 0.34481319785118103, "learning_rate": 9.855408342966585e-05, "loss": 0.0672, "step": 1230 }, { "grad_norm": 0.3594619929790497, "learning_rate": 9.851434182488033e-05, "loss": 0.0616, "step": 1240 }, { "grad_norm": 0.3774634003639221, "learning_rate": 9.84740696750253e-05, "loss": 0.0618, "step": 1250 }, { "grad_norm": 0.39621350169181824, "learning_rate": 9.843326742051055e-05, "loss": 0.0534, "step": 1260 }, { "grad_norm": 0.35853126645088196, "learning_rate": 9.839193550754297e-05, "loss": 0.06, "step": 1270 }, { "grad_norm": 0.48413267731666565, "learning_rate": 9.835007438812177e-05, "loss": 0.0566, "step": 1280 }, { "grad_norm": 0.5082318782806396, "learning_rate": 9.830768452003341e-05, "loss": 0.064, "step": 1290 }, { "grad_norm": 0.4222351908683777, "learning_rate": 9.826476636684671e-05, "loss": 0.0596, "step": 1300 }, { "grad_norm": 0.42161524295806885, "learning_rate": 9.822132039790773e-05, "loss": 0.0604, "step": 1310 }, { "grad_norm": 0.42876842617988586, "learning_rate": 9.817734708833461e-05, "loss": 0.0605, "step": 1320 }, { "grad_norm": 0.33034342527389526, "learning_rate": 9.813284691901243e-05, "loss": 0.0599, "step": 1330 }, { "grad_norm": 0.30100953578948975, "learning_rate": 9.808782037658792e-05, "loss": 0.0583, "step": 1340 }, { "grad_norm": 0.38982516527175903, "learning_rate": 9.804226795346411e-05, "loss": 0.0499, "step": 1350 }, { "grad_norm": 0.4682359993457794, "learning_rate": 9.799619014779503e-05, "loss": 0.0556, "step": 1360 }, { "grad_norm": 0.2822723090648651, "learning_rate": 9.794958746348013e-05, "loss": 0.0535, "step": 1370 }, { "grad_norm": 0.35257041454315186, "learning_rate": 9.790246041015896e-05, "loss": 0.0538, "step": 1380 }, { "grad_norm": 0.3130662441253662, "learning_rate": 9.785480950320538e-05, "loss": 0.0487, "step": 1390 }, { "grad_norm": 0.3834728002548218, "learning_rate": 9.78066352637221e-05, "loss": 0.0522, "step": 1400 }, { "grad_norm": 0.32492199540138245, "learning_rate": 9.775793821853488e-05, "loss": 0.0509, "step": 1410 }, { "grad_norm": 0.3241478204727173, "learning_rate": 9.77087189001868e-05, "loss": 0.0538, "step": 1420 }, { "grad_norm": 0.392362505197525, "learning_rate": 9.765897784693243e-05, "loss": 0.0559, "step": 1430 }, { "grad_norm": 0.3127867877483368, "learning_rate": 9.760871560273197e-05, "loss": 0.0539, "step": 1440 }, { "grad_norm": 0.42344188690185547, "learning_rate": 9.755793271724526e-05, "loss": 0.0643, "step": 1450 }, { "grad_norm": 0.34725844860076904, "learning_rate": 9.750662974582584e-05, "loss": 0.0561, "step": 1460 }, { "grad_norm": 0.35630473494529724, "learning_rate": 9.745480724951473e-05, "loss": 0.0652, "step": 1470 }, { "grad_norm": 0.5625796318054199, "learning_rate": 9.740246579503447e-05, "loss": 0.0612, "step": 1480 }, { "grad_norm": 0.3609796166419983, "learning_rate": 9.734960595478284e-05, "loss": 0.0534, "step": 1490 }, { "grad_norm": 0.40245792269706726, "learning_rate": 9.729622830682657e-05, "loss": 0.0583, "step": 1500 }, { "grad_norm": 0.34302157163619995, "learning_rate": 9.724233343489504e-05, "loss": 0.053, "step": 1510 }, { "grad_norm": 0.26678526401519775, "learning_rate": 9.718792192837396e-05, "loss": 0.053, "step": 1520 }, { "grad_norm": 0.3920958638191223, "learning_rate": 9.713299438229886e-05, "loss": 0.0492, "step": 1530 }, { "grad_norm": 0.46182528138160706, "learning_rate": 9.707755139734855e-05, "loss": 0.0533, "step": 1540 }, { "grad_norm": 0.48236244916915894, "learning_rate": 9.702159357983866e-05, "loss": 0.0606, "step": 1550 }, { "grad_norm": 0.39831215143203735, "learning_rate": 9.696512154171492e-05, "loss": 0.0537, "step": 1560 }, { "grad_norm": 0.2661665678024292, "learning_rate": 9.690813590054645e-05, "loss": 0.0445, "step": 1570 }, { "grad_norm": 0.36949822306632996, "learning_rate": 9.685063727951914e-05, "loss": 0.0528, "step": 1580 }, { "grad_norm": 0.4774511158466339, "learning_rate": 9.679262630742865e-05, "loss": 0.0572, "step": 1590 }, { "grad_norm": 0.5226991176605225, "learning_rate": 9.673410361867373e-05, "loss": 0.0549, "step": 1600 }, { "grad_norm": 0.39829888939857483, "learning_rate": 9.667506985324909e-05, "loss": 0.056, "step": 1610 }, { "grad_norm": 0.32898303866386414, "learning_rate": 9.661552565673855e-05, "loss": 0.0494, "step": 1620 }, { "grad_norm": 0.3641705811023712, "learning_rate": 9.655547168030789e-05, "loss": 0.0552, "step": 1630 }, { "grad_norm": 0.4412024915218353, "learning_rate": 9.649490858069777e-05, "loss": 0.0524, "step": 1640 }, { "grad_norm": 0.3295491635799408, "learning_rate": 9.643383702021658e-05, "loss": 0.0592, "step": 1650 }, { "grad_norm": 0.3767043948173523, "learning_rate": 9.637225766673307e-05, "loss": 0.0496, "step": 1660 }, { "grad_norm": 0.3970220983028412, "learning_rate": 9.631017119366922e-05, "loss": 0.057, "step": 1670 }, { "grad_norm": 0.2654162645339966, "learning_rate": 9.624757827999273e-05, "loss": 0.0596, "step": 1680 }, { "grad_norm": 0.4113260507583618, "learning_rate": 9.618447961020971e-05, "loss": 0.0571, "step": 1690 }, { "grad_norm": 0.3153928816318512, "learning_rate": 9.612087587435707e-05, "loss": 0.0559, "step": 1700 }, { "grad_norm": 0.2880053222179413, "learning_rate": 9.605676776799508e-05, "loss": 0.0531, "step": 1710 }, { "grad_norm": 0.3524435758590698, "learning_rate": 9.599215599219973e-05, "loss": 0.054, "step": 1720 }, { "grad_norm": 0.5203834176063538, "learning_rate": 9.592704125355505e-05, "loss": 0.0548, "step": 1730 }, { "grad_norm": 0.30815619230270386, "learning_rate": 9.586142426414538e-05, "loss": 0.0477, "step": 1740 }, { "grad_norm": 0.5077760815620422, "learning_rate": 9.57953057415476e-05, "loss": 0.0573, "step": 1750 }, { "grad_norm": 0.21000181138515472, "learning_rate": 9.572868640882328e-05, "loss": 0.0565, "step": 1760 }, { "grad_norm": 0.31183919310569763, "learning_rate": 9.56615669945108e-05, "loss": 0.0569, "step": 1770 }, { "grad_norm": 0.2621975243091583, "learning_rate": 9.55939482326173e-05, "loss": 0.05, "step": 1780 }, { "grad_norm": 0.40494292974472046, "learning_rate": 9.552583086261069e-05, "loss": 0.0572, "step": 1790 }, { "grad_norm": 0.41099870204925537, "learning_rate": 9.545721562941168e-05, "loss": 0.0523, "step": 1800 }, { "grad_norm": 0.31243613362312317, "learning_rate": 9.538810328338543e-05, "loss": 0.0478, "step": 1810 }, { "grad_norm": 0.27863073348999023, "learning_rate": 9.531849458033349e-05, "loss": 0.0478, "step": 1820 }, { "grad_norm": 0.37444883584976196, "learning_rate": 9.524839028148547e-05, "loss": 0.0535, "step": 1830 }, { "grad_norm": 0.3028647005558014, "learning_rate": 9.517779115349077e-05, "loss": 0.0536, "step": 1840 }, { "grad_norm": 0.3779330849647522, "learning_rate": 9.510669796841014e-05, "loss": 0.0587, "step": 1850 }, { "grad_norm": 0.30305418372154236, "learning_rate": 9.503511150370727e-05, "loss": 0.0536, "step": 1860 }, { "grad_norm": 0.416415274143219, "learning_rate": 9.496303254224024e-05, "loss": 0.0522, "step": 1870 }, { "grad_norm": 0.4645114541053772, "learning_rate": 9.489046187225306e-05, "loss": 0.0575, "step": 1880 }, { "grad_norm": 0.3728070855140686, "learning_rate": 9.481740028736692e-05, "loss": 0.0488, "step": 1890 }, { "grad_norm": 0.2728413939476013, "learning_rate": 9.474384858657164e-05, "loss": 0.0592, "step": 1900 }, { "grad_norm": 0.32814422249794006, "learning_rate": 9.466980757421679e-05, "loss": 0.0544, "step": 1910 }, { "grad_norm": 0.35529980063438416, "learning_rate": 9.459527806000305e-05, "loss": 0.0597, "step": 1920 }, { "grad_norm": 0.5990370512008667, "learning_rate": 9.452026085897325e-05, "loss": 0.062, "step": 1930 }, { "grad_norm": 0.4094012975692749, "learning_rate": 9.444475679150348e-05, "loss": 0.054, "step": 1940 }, { "grad_norm": 0.5258490443229675, "learning_rate": 9.436876668329411e-05, "loss": 0.0605, "step": 1950 }, { "grad_norm": 0.4500177800655365, "learning_rate": 9.429229136536079e-05, "loss": 0.0544, "step": 1960 }, { "grad_norm": 0.2816692292690277, "learning_rate": 9.421533167402534e-05, "loss": 0.0536, "step": 1970 }, { "grad_norm": 0.3924410045146942, "learning_rate": 9.413788845090666e-05, "loss": 0.0649, "step": 1980 }, { "grad_norm": 0.2608245015144348, "learning_rate": 9.405996254291136e-05, "loss": 0.0504, "step": 1990 }, { "grad_norm": 0.3233589231967926, "learning_rate": 9.398155480222474e-05, "loss": 0.0507, "step": 2000 }, { "grad_norm": 0.35494449734687805, "learning_rate": 9.390266608630128e-05, "loss": 0.0463, "step": 2010 }, { "grad_norm": 0.24342451989650726, "learning_rate": 9.38232972578553e-05, "loss": 0.0536, "step": 2020 }, { "grad_norm": 0.33717989921569824, "learning_rate": 9.374344918485164e-05, "loss": 0.0543, "step": 2030 }, { "grad_norm": 0.3596448004245758, "learning_rate": 9.366312274049602e-05, "loss": 0.0566, "step": 2040 }, { "grad_norm": 0.3476393520832062, "learning_rate": 9.358231880322554e-05, "loss": 0.0525, "step": 2050 }, { "grad_norm": 0.34942588210105896, "learning_rate": 9.350103825669916e-05, "loss": 0.0461, "step": 2060 }, { "grad_norm": 0.33282092213630676, "learning_rate": 9.341928198978787e-05, "loss": 0.0534, "step": 2070 }, { "grad_norm": 0.20088507235050201, "learning_rate": 9.333705089656512e-05, "loss": 0.046, "step": 2080 }, { "grad_norm": 0.2375725507736206, "learning_rate": 9.325434587629698e-05, "loss": 0.0494, "step": 2090 }, { "grad_norm": 0.2507284879684448, "learning_rate": 9.31711678334323e-05, "loss": 0.0443, "step": 2100 }, { "grad_norm": 0.2696078419685364, "learning_rate": 9.308751767759282e-05, "loss": 0.045, "step": 2110 }, { "grad_norm": 0.2889285981655121, "learning_rate": 9.300339632356325e-05, "loss": 0.0493, "step": 2120 }, { "grad_norm": 0.36062878370285034, "learning_rate": 9.291880469128124e-05, "loss": 0.0476, "step": 2130 }, { "grad_norm": 0.47251686453819275, "learning_rate": 9.283374370582732e-05, "loss": 0.0455, "step": 2140 }, { "grad_norm": 0.38125917315483093, "learning_rate": 9.274821429741482e-05, "loss": 0.0498, "step": 2150 }, { "grad_norm": 0.4520898163318634, "learning_rate": 9.266221740137961e-05, "loss": 0.0466, "step": 2160 }, { "grad_norm": 0.28161367774009705, "learning_rate": 9.257575395817001e-05, "loss": 0.0428, "step": 2170 }, { "grad_norm": 0.3592947721481323, "learning_rate": 9.248882491333637e-05, "loss": 0.0507, "step": 2180 }, { "grad_norm": 0.3055783212184906, "learning_rate": 9.240143121752076e-05, "loss": 0.0528, "step": 2190 }, { "grad_norm": 0.2711855173110962, "learning_rate": 9.23135738264467e-05, "loss": 0.0495, "step": 2200 }, { "grad_norm": 0.22879229485988617, "learning_rate": 9.222525370090849e-05, "loss": 0.0443, "step": 2210 }, { "grad_norm": 0.4473995566368103, "learning_rate": 9.213647180676088e-05, "loss": 0.0417, "step": 2220 }, { "grad_norm": 0.3168101906776428, "learning_rate": 9.204722911490846e-05, "loss": 0.0446, "step": 2230 }, { "grad_norm": 0.2529396116733551, "learning_rate": 9.1957526601295e-05, "loss": 0.0423, "step": 2240 }, { "grad_norm": 0.2650357484817505, "learning_rate": 9.186736524689281e-05, "loss": 0.0518, "step": 2250 }, { "grad_norm": 0.2764832377433777, "learning_rate": 9.177674603769204e-05, "loss": 0.0544, "step": 2260 }, { "grad_norm": 0.36209458112716675, "learning_rate": 9.168566996468983e-05, "loss": 0.0462, "step": 2270 }, { "grad_norm": 0.30518805980682373, "learning_rate": 9.159413802387951e-05, "loss": 0.0564, "step": 2280 }, { "grad_norm": 0.2883245348930359, "learning_rate": 9.150215121623974e-05, "loss": 0.0438, "step": 2290 }, { "grad_norm": 0.25169456005096436, "learning_rate": 9.140971054772349e-05, "loss": 0.0376, "step": 2300 }, { "grad_norm": 0.2580600082874298, "learning_rate": 9.131681702924713e-05, "loss": 0.0503, "step": 2310 }, { "grad_norm": 0.2840389609336853, "learning_rate": 9.122347167667926e-05, "loss": 0.0453, "step": 2320 }, { "grad_norm": 0.2722470760345459, "learning_rate": 9.112967551082973e-05, "loss": 0.053, "step": 2330 }, { "grad_norm": 0.27755656838417053, "learning_rate": 9.103542955743835e-05, "loss": 0.049, "step": 2340 }, { "grad_norm": 0.31788480281829834, "learning_rate": 9.094073484716381e-05, "loss": 0.0536, "step": 2350 }, { "grad_norm": 0.2957926392555237, "learning_rate": 9.084559241557226e-05, "loss": 0.0454, "step": 2360 }, { "grad_norm": 0.4266762137413025, "learning_rate": 9.075000330312608e-05, "loss": 0.0486, "step": 2370 }, { "grad_norm": 0.3195292055606842, "learning_rate": 9.065396855517253e-05, "loss": 0.0479, "step": 2380 }, { "grad_norm": 0.41534286737442017, "learning_rate": 9.055748922193219e-05, "loss": 0.0503, "step": 2390 }, { "grad_norm": 0.379057914018631, "learning_rate": 9.046056635848761e-05, "loss": 0.0535, "step": 2400 }, { "grad_norm": 0.39929866790771484, "learning_rate": 9.036320102477169e-05, "loss": 0.0499, "step": 2410 }, { "grad_norm": 0.3459583818912506, "learning_rate": 9.02653942855561e-05, "loss": 0.0446, "step": 2420 }, { "grad_norm": 0.27777546644210815, "learning_rate": 9.016714721043971e-05, "loss": 0.0437, "step": 2430 }, { "grad_norm": 0.2608919143676758, "learning_rate": 9.006846087383675e-05, "loss": 0.0429, "step": 2440 }, { "grad_norm": 0.32613542675971985, "learning_rate": 8.996933635496523e-05, "loss": 0.0384, "step": 2450 }, { "grad_norm": 0.2358037531375885, "learning_rate": 8.986977473783498e-05, "loss": 0.0432, "step": 2460 }, { "grad_norm": 0.27175092697143555, "learning_rate": 8.97697771112359e-05, "loss": 0.0404, "step": 2470 }, { "grad_norm": 0.36901921033859253, "learning_rate": 8.966934456872602e-05, "loss": 0.0454, "step": 2480 }, { "grad_norm": 0.2279893010854721, "learning_rate": 8.95684782086195e-05, "loss": 0.0482, "step": 2490 }, { "grad_norm": 0.3557843267917633, "learning_rate": 8.946717913397476e-05, "loss": 0.055, "step": 2500 }, { "grad_norm": 0.22711744904518127, "learning_rate": 8.93654484525822e-05, "loss": 0.0462, "step": 2510 }, { "grad_norm": 0.3490443527698517, "learning_rate": 8.926328727695226e-05, "loss": 0.041, "step": 2520 }, { "grad_norm": 0.3079090416431427, "learning_rate": 8.916069672430319e-05, "loss": 0.0397, "step": 2530 }, { "grad_norm": 0.3696517050266266, "learning_rate": 8.905767791654884e-05, "loss": 0.0432, "step": 2540 }, { "grad_norm": 0.4252660870552063, "learning_rate": 8.895423198028638e-05, "loss": 0.0441, "step": 2550 }, { "grad_norm": 0.31201404333114624, "learning_rate": 8.885036004678402e-05, "loss": 0.0438, "step": 2560 }, { "grad_norm": 0.38208797574043274, "learning_rate": 8.874606325196857e-05, "loss": 0.049, "step": 2570 }, { "grad_norm": 0.2732146680355072, "learning_rate": 8.864134273641304e-05, "loss": 0.0531, "step": 2580 }, { "grad_norm": 0.2173292487859726, "learning_rate": 8.853619964532427e-05, "loss": 0.0435, "step": 2590 }, { "grad_norm": 0.22845201194286346, "learning_rate": 8.843063512853019e-05, "loss": 0.0444, "step": 2600 }, { "grad_norm": 0.2683410942554474, "learning_rate": 8.832465034046749e-05, "loss": 0.0417, "step": 2610 }, { "grad_norm": 0.3239377439022064, "learning_rate": 8.821824644016882e-05, "loss": 0.0459, "step": 2620 }, { "grad_norm": 0.3698790967464447, "learning_rate": 8.811142459125019e-05, "loss": 0.0423, "step": 2630 }, { "grad_norm": 0.3114430904388428, "learning_rate": 8.800418596189822e-05, "loss": 0.0415, "step": 2640 }, { "grad_norm": 0.344240665435791, "learning_rate": 8.789653172485737e-05, "loss": 0.05, "step": 2650 }, { "grad_norm": 0.37669122219085693, "learning_rate": 8.778846305741715e-05, "loss": 0.0579, "step": 2660 }, { "grad_norm": 0.24897822737693787, "learning_rate": 8.767998114139918e-05, "loss": 0.0469, "step": 2670 }, { "grad_norm": 0.35893765091896057, "learning_rate": 8.757108716314429e-05, "loss": 0.0501, "step": 2680 }, { "grad_norm": 0.44093087315559387, "learning_rate": 8.746178231349962e-05, "loss": 0.0484, "step": 2690 }, { "grad_norm": 0.361758291721344, "learning_rate": 8.735206778780549e-05, "loss": 0.0507, "step": 2700 }, { "grad_norm": 0.371342271566391, "learning_rate": 8.724194478588234e-05, "loss": 0.0536, "step": 2710 }, { "grad_norm": 0.3777939975261688, "learning_rate": 8.713141451201772e-05, "loss": 0.0462, "step": 2720 }, { "grad_norm": 0.24280275404453278, "learning_rate": 8.702047817495295e-05, "loss": 0.0452, "step": 2730 }, { "grad_norm": 0.330927312374115, "learning_rate": 8.69091369878701e-05, "loss": 0.0453, "step": 2740 }, { "grad_norm": 0.22073939442634583, "learning_rate": 8.679739216837849e-05, "loss": 0.0381, "step": 2750 }, { "grad_norm": 0.30922016501426697, "learning_rate": 8.66852449385016e-05, "loss": 0.043, "step": 2760 }, { "grad_norm": 0.26442134380340576, "learning_rate": 8.657269652466356e-05, "loss": 0.044, "step": 2770 }, { "grad_norm": 0.2650727331638336, "learning_rate": 8.645974815767577e-05, "loss": 0.0472, "step": 2780 }, { "grad_norm": 0.1834416538476944, "learning_rate": 8.634640107272351e-05, "loss": 0.0384, "step": 2790 }, { "grad_norm": 0.34226319193840027, "learning_rate": 8.623265650935234e-05, "loss": 0.0422, "step": 2800 }, { "grad_norm": 0.328534334897995, "learning_rate": 8.611851571145456e-05, "loss": 0.0436, "step": 2810 }, { "grad_norm": 0.31813085079193115, "learning_rate": 8.600397992725566e-05, "loss": 0.046, "step": 2820 }, { "grad_norm": 0.3045029938220978, "learning_rate": 8.588905040930061e-05, "loss": 0.0416, "step": 2830 }, { "grad_norm": 0.2649421989917755, "learning_rate": 8.577372841444022e-05, "loss": 0.046, "step": 2840 }, { "grad_norm": 0.287449449300766, "learning_rate": 8.565801520381736e-05, "loss": 0.0462, "step": 2850 }, { "grad_norm": 0.4459896981716156, "learning_rate": 8.554191204285313e-05, "loss": 0.0421, "step": 2860 }, { "grad_norm": 0.39070722460746765, "learning_rate": 8.542542020123315e-05, "loss": 0.0423, "step": 2870 }, { "grad_norm": 0.20476259291172028, "learning_rate": 8.530854095289347e-05, "loss": 0.0412, "step": 2880 }, { "grad_norm": 0.27058538794517517, "learning_rate": 8.519127557600688e-05, "loss": 0.0401, "step": 2890 }, { "grad_norm": 0.3894350826740265, "learning_rate": 8.507362535296871e-05, "loss": 0.0455, "step": 2900 }, { "grad_norm": 0.3115444779396057, "learning_rate": 8.495559157038299e-05, "loss": 0.0522, "step": 2910 }, { "grad_norm": 0.3186185657978058, "learning_rate": 8.483717551904823e-05, "loss": 0.0371, "step": 2920 }, { "grad_norm": 0.37191417813301086, "learning_rate": 8.47183784939434e-05, "loss": 0.0419, "step": 2930 }, { "grad_norm": 0.20773254334926605, "learning_rate": 8.459920179421374e-05, "loss": 0.0446, "step": 2940 }, { "grad_norm": 0.33358556032180786, "learning_rate": 8.447964672315656e-05, "loss": 0.0405, "step": 2950 }, { "grad_norm": 0.28731569647789, "learning_rate": 8.435971458820692e-05, "loss": 0.0509, "step": 2960 }, { "grad_norm": 0.2441011220216751, "learning_rate": 8.423940670092345e-05, "loss": 0.0519, "step": 2970 }, { "grad_norm": 0.29939723014831543, "learning_rate": 8.411872437697394e-05, "loss": 0.0425, "step": 2980 }, { "grad_norm": 0.2867561876773834, "learning_rate": 8.399766893612096e-05, "loss": 0.0382, "step": 2990 }, { "grad_norm": 0.2843914031982422, "learning_rate": 8.38762417022074e-05, "loss": 0.0443, "step": 3000 }, { "grad_norm": 0.3786162734031677, "learning_rate": 8.375444400314204e-05, "loss": 0.0432, "step": 3010 }, { "grad_norm": 0.3932505249977112, "learning_rate": 8.3632277170885e-05, "loss": 0.0503, "step": 3020 }, { "grad_norm": 0.4066917598247528, "learning_rate": 8.350974254143318e-05, "loss": 0.042, "step": 3030 }, { "grad_norm": 0.30071887373924255, "learning_rate": 8.338684145480566e-05, "loss": 0.0417, "step": 3040 }, { "grad_norm": 0.35279059410095215, "learning_rate": 8.326357525502904e-05, "loss": 0.0462, "step": 3050 }, { "grad_norm": 0.3409857451915741, "learning_rate": 8.313994529012273e-05, "loss": 0.0485, "step": 3060 }, { "grad_norm": 0.24817773699760437, "learning_rate": 8.301595291208422e-05, "loss": 0.0466, "step": 3070 }, { "grad_norm": 0.21267981827259064, "learning_rate": 8.289159947687427e-05, "loss": 0.0471, "step": 3080 }, { "grad_norm": 0.28439196944236755, "learning_rate": 8.276688634440216e-05, "loss": 0.0426, "step": 3090 }, { "grad_norm": 0.40261176228523254, "learning_rate": 8.26418148785107e-05, "loss": 0.0465, "step": 3100 }, { "grad_norm": 0.33959680795669556, "learning_rate": 8.251638644696141e-05, "loss": 0.0459, "step": 3110 }, { "grad_norm": 0.2536005973815918, "learning_rate": 8.23906024214195e-05, "loss": 0.0393, "step": 3120 }, { "grad_norm": 0.20771771669387817, "learning_rate": 8.226446417743897e-05, "loss": 0.0433, "step": 3130 }, { "grad_norm": 0.24254511296749115, "learning_rate": 8.213797309444742e-05, "loss": 0.04, "step": 3140 }, { "grad_norm": 0.5023501515388489, "learning_rate": 8.201113055573105e-05, "loss": 0.0463, "step": 3150 }, { "grad_norm": 0.3516068458557129, "learning_rate": 8.188393794841958e-05, "loss": 0.0427, "step": 3160 }, { "grad_norm": 0.3304799795150757, "learning_rate": 8.175639666347094e-05, "loss": 0.0427, "step": 3170 }, { "grad_norm": 0.23683060705661774, "learning_rate": 8.162850809565623e-05, "loss": 0.0368, "step": 3180 }, { "grad_norm": 0.24009308218955994, "learning_rate": 8.150027364354431e-05, "loss": 0.0378, "step": 3190 }, { "grad_norm": 0.22060438990592957, "learning_rate": 8.137169470948662e-05, "loss": 0.0454, "step": 3200 }, { "grad_norm": 0.2193693220615387, "learning_rate": 8.124277269960179e-05, "loss": 0.0462, "step": 3210 }, { "grad_norm": 0.3058091700077057, "learning_rate": 8.111350902376023e-05, "loss": 0.0416, "step": 3220 }, { "grad_norm": 0.4012700319290161, "learning_rate": 8.098390509556883e-05, "loss": 0.0437, "step": 3230 }, { "grad_norm": 0.30088838934898376, "learning_rate": 8.085396233235536e-05, "loss": 0.0373, "step": 3240 }, { "grad_norm": 0.2540518343448639, "learning_rate": 8.072368215515306e-05, "loss": 0.0363, "step": 3250 }, { "grad_norm": 0.3507944941520691, "learning_rate": 8.059306598868506e-05, "loss": 0.0444, "step": 3260 }, { "grad_norm": 0.26104429364204407, "learning_rate": 8.046211526134888e-05, "loss": 0.0382, "step": 3270 }, { "grad_norm": 0.49699699878692627, "learning_rate": 8.033083140520065e-05, "loss": 0.037, "step": 3280 }, { "grad_norm": 0.2147408425807953, "learning_rate": 8.019921585593962e-05, "loss": 0.0392, "step": 3290 }, { "grad_norm": 0.33502137660980225, "learning_rate": 8.006727005289232e-05, "loss": 0.0347, "step": 3300 }, { "grad_norm": 0.21627382934093475, "learning_rate": 7.993499543899692e-05, "loss": 0.0359, "step": 3310 }, { "grad_norm": 0.28527969121932983, "learning_rate": 7.980239346078742e-05, "loss": 0.0374, "step": 3320 }, { "grad_norm": 0.37590011954307556, "learning_rate": 7.966946556837778e-05, "loss": 0.0444, "step": 3330 }, { "grad_norm": 0.2959820330142975, "learning_rate": 7.953621321544616e-05, "loss": 0.0414, "step": 3340 }, { "grad_norm": 0.2505655288696289, "learning_rate": 7.940263785921896e-05, "loss": 0.0482, "step": 3350 }, { "grad_norm": 0.38306328654289246, "learning_rate": 7.926874096045482e-05, "loss": 0.0367, "step": 3360 }, { "grad_norm": 0.3061180114746094, "learning_rate": 7.913452398342881e-05, "loss": 0.0546, "step": 3370 }, { "grad_norm": 0.261535108089447, "learning_rate": 7.89999883959163e-05, "loss": 0.0444, "step": 3380 }, { "grad_norm": 0.30257314443588257, "learning_rate": 7.886513566917687e-05, "loss": 0.0458, "step": 3390 }, { "grad_norm": 0.35927677154541016, "learning_rate": 7.872996727793838e-05, "loss": 0.046, "step": 3400 }, { "grad_norm": 0.2871660888195038, "learning_rate": 7.859448470038069e-05, "loss": 0.0404, "step": 3410 }, { "grad_norm": 0.40603554248809814, "learning_rate": 7.845868941811956e-05, "loss": 0.0388, "step": 3420 }, { "grad_norm": 0.1779915690422058, "learning_rate": 7.832258291619043e-05, "loss": 0.0365, "step": 3430 }, { "grad_norm": 0.4340699315071106, "learning_rate": 7.81861666830322e-05, "loss": 0.0408, "step": 3440 }, { "grad_norm": 0.23649580776691437, "learning_rate": 7.804944221047097e-05, "loss": 0.0376, "step": 3450 }, { "grad_norm": 0.3731379210948944, "learning_rate": 7.791241099370364e-05, "loss": 0.0536, "step": 3460 }, { "grad_norm": 0.21042008697986603, "learning_rate": 7.777507453128163e-05, "loss": 0.037, "step": 3470 }, { "grad_norm": 0.4684864282608032, "learning_rate": 7.763743432509451e-05, "loss": 0.0423, "step": 3480 }, { "grad_norm": 0.20927050709724426, "learning_rate": 7.749949188035353e-05, "loss": 0.0383, "step": 3490 }, { "grad_norm": 0.2757911682128906, "learning_rate": 7.736124870557516e-05, "loss": 0.0475, "step": 3500 }, { "grad_norm": 0.2273566573858261, "learning_rate": 7.722270631256459e-05, "loss": 0.0437, "step": 3510 }, { "grad_norm": 0.1995142251253128, "learning_rate": 7.708386621639925e-05, "loss": 0.0368, "step": 3520 }, { "grad_norm": 0.17957501113414764, "learning_rate": 7.694472993541219e-05, "loss": 0.0408, "step": 3530 }, { "grad_norm": 0.3512098491191864, "learning_rate": 7.680529899117547e-05, "loss": 0.0449, "step": 3540 }, { "grad_norm": 0.2718364894390106, "learning_rate": 7.666557490848358e-05, "loss": 0.0378, "step": 3550 }, { "grad_norm": 0.23810185492038727, "learning_rate": 7.65255592153367e-05, "loss": 0.0421, "step": 3560 }, { "grad_norm": 0.3234817683696747, "learning_rate": 7.638525344292402e-05, "loss": 0.0429, "step": 3570 }, { "grad_norm": 0.15640801191329956, "learning_rate": 7.624465912560697e-05, "loss": 0.0447, "step": 3580 }, { "grad_norm": 0.21737875044345856, "learning_rate": 7.610377780090249e-05, "loss": 0.0354, "step": 3590 }, { "grad_norm": 0.20516827702522278, "learning_rate": 7.596261100946618e-05, "loss": 0.0408, "step": 3600 }, { "grad_norm": 0.265031099319458, "learning_rate": 7.582116029507542e-05, "loss": 0.0364, "step": 3610 }, { "grad_norm": 0.23448477685451508, "learning_rate": 7.56794272046126e-05, "loss": 0.0396, "step": 3620 }, { "grad_norm": 0.28485211730003357, "learning_rate": 7.55374132880481e-05, "loss": 0.0425, "step": 3630 }, { "grad_norm": 0.25391364097595215, "learning_rate": 7.539512009842333e-05, "loss": 0.0425, "step": 3640 }, { "grad_norm": 0.26370328664779663, "learning_rate": 7.525254919183382e-05, "loss": 0.0435, "step": 3650 }, { "grad_norm": 0.21674665808677673, "learning_rate": 7.510970212741215e-05, "loss": 0.0371, "step": 3660 }, { "grad_norm": 0.313233882188797, "learning_rate": 7.496658046731096e-05, "loss": 0.0359, "step": 3670 }, { "grad_norm": 0.183758944272995, "learning_rate": 7.482318577668578e-05, "loss": 0.0375, "step": 3680 }, { "grad_norm": 0.17781658470630646, "learning_rate": 7.467951962367796e-05, "loss": 0.0384, "step": 3690 }, { "grad_norm": 0.2651655077934265, "learning_rate": 7.453558357939755e-05, "loss": 0.0338, "step": 3700 }, { "grad_norm": 0.30325937271118164, "learning_rate": 7.439137921790606e-05, "loss": 0.0386, "step": 3710 }, { "grad_norm": 0.19928258657455444, "learning_rate": 7.42469081161993e-05, "loss": 0.0441, "step": 3720 }, { "grad_norm": 0.2151193618774414, "learning_rate": 7.410217185419006e-05, "loss": 0.0409, "step": 3730 }, { "grad_norm": 0.25591447949409485, "learning_rate": 7.395717201469095e-05, "loss": 0.0398, "step": 3740 }, { "grad_norm": 0.22096776962280273, "learning_rate": 7.381191018339696e-05, "loss": 0.042, "step": 3750 }, { "grad_norm": 0.2646503150463104, "learning_rate": 7.36663879488682e-05, "loss": 0.0408, "step": 3760 }, { "grad_norm": 0.34191131591796875, "learning_rate": 7.352060690251254e-05, "loss": 0.0407, "step": 3770 }, { "grad_norm": 0.268580824136734, "learning_rate": 7.337456863856811e-05, "loss": 0.0393, "step": 3780 }, { "grad_norm": 0.35366860032081604, "learning_rate": 7.3228274754086e-05, "loss": 0.0405, "step": 3790 }, { "grad_norm": 0.24171938002109528, "learning_rate": 7.308172684891267e-05, "loss": 0.0406, "step": 3800 }, { "grad_norm": 0.3674739897251129, "learning_rate": 7.293492652567255e-05, "loss": 0.0393, "step": 3810 }, { "grad_norm": 0.26678988337516785, "learning_rate": 7.278787538975043e-05, "loss": 0.0404, "step": 3820 }, { "grad_norm": 0.25137579441070557, "learning_rate": 7.2640575049274e-05, "loss": 0.0363, "step": 3830 }, { "grad_norm": 0.26335445046424866, "learning_rate": 7.249302711509616e-05, "loss": 0.0399, "step": 3840 }, { "grad_norm": 0.3272726535797119, "learning_rate": 7.23452332007775e-05, "loss": 0.0382, "step": 3850 }, { "grad_norm": 0.2056971937417984, "learning_rate": 7.219719492256858e-05, "loss": 0.0363, "step": 3860 }, { "grad_norm": 0.2941059172153473, "learning_rate": 7.20489138993923e-05, "loss": 0.0398, "step": 3870 }, { "grad_norm": 0.308083176612854, "learning_rate": 7.190039175282614e-05, "loss": 0.0373, "step": 3880 }, { "grad_norm": 0.20674903690814972, "learning_rate": 7.175163010708455e-05, "loss": 0.0352, "step": 3890 }, { "grad_norm": 0.15629924833774567, "learning_rate": 7.1602630589001e-05, "loss": 0.0385, "step": 3900 }, { "grad_norm": 0.18862292170524597, "learning_rate": 7.14533948280104e-05, "loss": 0.038, "step": 3910 }, { "grad_norm": 0.22169341146945953, "learning_rate": 7.130392445613109e-05, "loss": 0.0334, "step": 3920 }, { "grad_norm": 0.2820945084095001, "learning_rate": 7.115422110794711e-05, "loss": 0.0362, "step": 3930 }, { "grad_norm": 0.3198398947715759, "learning_rate": 7.100428642059033e-05, "loss": 0.0383, "step": 3940 }, { "grad_norm": 0.17828813195228577, "learning_rate": 7.08541220337224e-05, "loss": 0.0379, "step": 3950 }, { "grad_norm": 0.21609224379062653, "learning_rate": 7.070372958951706e-05, "loss": 0.033, "step": 3960 }, { "grad_norm": 0.3003736436367035, "learning_rate": 7.055311073264194e-05, "loss": 0.0429, "step": 3970 }, { "grad_norm": 0.42280134558677673, "learning_rate": 7.040226711024077e-05, "loss": 0.0412, "step": 3980 }, { "grad_norm": 0.18756848573684692, "learning_rate": 7.02512003719152e-05, "loss": 0.0402, "step": 3990 }, { "grad_norm": 0.18985839188098907, "learning_rate": 7.00999121697069e-05, "loss": 0.0417, "step": 4000 }, { "grad_norm": 0.19545404613018036, "learning_rate": 6.99484041580794e-05, "loss": 0.0374, "step": 4010 }, { "grad_norm": 0.26584795117378235, "learning_rate": 6.979667799390004e-05, "loss": 0.035, "step": 4020 }, { "grad_norm": 0.2350008338689804, "learning_rate": 6.964473533642185e-05, "loss": 0.0359, "step": 4030 }, { "grad_norm": 0.2585308849811554, "learning_rate": 6.949257784726539e-05, "loss": 0.0413, "step": 4040 }, { "grad_norm": 0.19493839144706726, "learning_rate": 6.934020719040056e-05, "loss": 0.0371, "step": 4050 }, { "grad_norm": 0.19848328828811646, "learning_rate": 6.918762503212848e-05, "loss": 0.0399, "step": 4060 }, { "grad_norm": 0.23942013084888458, "learning_rate": 6.903483304106319e-05, "loss": 0.039, "step": 4070 }, { "grad_norm": 0.23592859506607056, "learning_rate": 6.888183288811341e-05, "loss": 0.0408, "step": 4080 }, { "grad_norm": 0.34572142362594604, "learning_rate": 6.87286262464643e-05, "loss": 0.0353, "step": 4090 }, { "grad_norm": 0.2366577386856079, "learning_rate": 6.857521479155915e-05, "loss": 0.0404, "step": 4100 }, { "grad_norm": 0.24964816868305206, "learning_rate": 6.842160020108104e-05, "loss": 0.0423, "step": 4110 }, { "grad_norm": 0.2763529121875763, "learning_rate": 6.826778415493455e-05, "loss": 0.0401, "step": 4120 }, { "grad_norm": 0.20391684770584106, "learning_rate": 6.811376833522729e-05, "loss": 0.0375, "step": 4130 }, { "grad_norm": 0.23832693696022034, "learning_rate": 6.795955442625159e-05, "loss": 0.0385, "step": 4140 }, { "grad_norm": 0.4532974064350128, "learning_rate": 6.780514411446608e-05, "loss": 0.0341, "step": 4150 }, { "grad_norm": 0.2244638353586197, "learning_rate": 6.765053908847716e-05, "loss": 0.0331, "step": 4160 }, { "grad_norm": 0.340586394071579, "learning_rate": 6.749574103902064e-05, "loss": 0.0404, "step": 4170 }, { "grad_norm": 0.18987394869327545, "learning_rate": 6.734075165894317e-05, "loss": 0.0368, "step": 4180 }, { "grad_norm": 0.222764253616333, "learning_rate": 6.71855726431838e-05, "loss": 0.0386, "step": 4190 }, { "grad_norm": 0.22654202580451965, "learning_rate": 6.703020568875538e-05, "loss": 0.0349, "step": 4200 }, { "grad_norm": 0.2089991569519043, "learning_rate": 6.687465249472603e-05, "loss": 0.0313, "step": 4210 }, { "grad_norm": 0.42392343282699585, "learning_rate": 6.671891476220055e-05, "loss": 0.0359, "step": 4220 }, { "grad_norm": 0.2129932940006256, "learning_rate": 6.656299419430183e-05, "loss": 0.0325, "step": 4230 }, { "grad_norm": 0.6330697536468506, "learning_rate": 6.640689249615223e-05, "loss": 0.0364, "step": 4240 }, { "grad_norm": 0.26332640647888184, "learning_rate": 6.625061137485491e-05, "loss": 0.0454, "step": 4250 }, { "grad_norm": 0.288442999124527, "learning_rate": 6.609415253947517e-05, "loss": 0.0411, "step": 4260 }, { "grad_norm": 0.275870144367218, "learning_rate": 6.593751770102178e-05, "loss": 0.0401, "step": 4270 }, { "grad_norm": 0.31951430439949036, "learning_rate": 6.578070857242823e-05, "loss": 0.0401, "step": 4280 }, { "grad_norm": 0.24794380366802216, "learning_rate": 6.562372686853402e-05, "loss": 0.0392, "step": 4290 }, { "grad_norm": 0.22700633108615875, "learning_rate": 6.546657430606593e-05, "loss": 0.0374, "step": 4300 }, { "grad_norm": 0.23889751732349396, "learning_rate": 6.530925260361918e-05, "loss": 0.0352, "step": 4310 }, { "grad_norm": 0.23572608828544617, "learning_rate": 6.515176348163871e-05, "loss": 0.0366, "step": 4320 }, { "grad_norm": 0.2380843460559845, "learning_rate": 6.499410866240032e-05, "loss": 0.034, "step": 4330 }, { "grad_norm": 0.20993933081626892, "learning_rate": 6.48362898699919e-05, "loss": 0.0414, "step": 4340 }, { "grad_norm": 0.39562374353408813, "learning_rate": 6.467830883029443e-05, "loss": 0.0427, "step": 4350 }, { "grad_norm": 0.44186779856681824, "learning_rate": 6.452016727096326e-05, "loss": 0.038, "step": 4360 }, { "grad_norm": 0.26088884472846985, "learning_rate": 6.436186692140916e-05, "loss": 0.0365, "step": 4370 }, { "grad_norm": 0.2963751554489136, "learning_rate": 6.420340951277938e-05, "loss": 0.0368, "step": 4380 }, { "grad_norm": 0.3488236963748932, "learning_rate": 6.404479677793874e-05, "loss": 0.0342, "step": 4390 }, { "grad_norm": 0.2727224826812744, "learning_rate": 6.388603045145075e-05, "loss": 0.0391, "step": 4400 }, { "grad_norm": 0.3018597960472107, "learning_rate": 6.372711226955843e-05, "loss": 0.0391, "step": 4410 }, { "grad_norm": 0.24780990183353424, "learning_rate": 6.356804397016564e-05, "loss": 0.0358, "step": 4420 }, { "grad_norm": 0.24078252911567688, "learning_rate": 6.340882729281779e-05, "loss": 0.0414, "step": 4430 }, { "grad_norm": 0.24076621234416962, "learning_rate": 6.324946397868294e-05, "loss": 0.0381, "step": 4440 }, { "grad_norm": 0.14270947873592377, "learning_rate": 6.308995577053276e-05, "loss": 0.0295, "step": 4450 }, { "grad_norm": 0.23324450850486755, "learning_rate": 6.293030441272347e-05, "loss": 0.0365, "step": 4460 }, { "grad_norm": 0.21239663660526276, "learning_rate": 6.277051165117677e-05, "loss": 0.0327, "step": 4470 }, { "grad_norm": 0.2108556032180786, "learning_rate": 6.261057923336064e-05, "loss": 0.0344, "step": 4480 }, { "grad_norm": 0.43279483914375305, "learning_rate": 6.245050890827042e-05, "loss": 0.0393, "step": 4490 }, { "grad_norm": 0.22541944682598114, "learning_rate": 6.229030242640952e-05, "loss": 0.0365, "step": 4500 }, { "grad_norm": 0.3119383752346039, "learning_rate": 6.212996153977037e-05, "loss": 0.0426, "step": 4510 }, { "grad_norm": 0.20606237649917603, "learning_rate": 6.196948800181523e-05, "loss": 0.0326, "step": 4520 }, { "grad_norm": 0.44093871116638184, "learning_rate": 6.180888356745695e-05, "loss": 0.0388, "step": 4530 }, { "grad_norm": 0.1838531643152237, "learning_rate": 6.164814999303995e-05, "loss": 0.0344, "step": 4540 }, { "grad_norm": 0.22674474120140076, "learning_rate": 6.148728903632081e-05, "loss": 0.0329, "step": 4550 }, { "grad_norm": 0.2441047728061676, "learning_rate": 6.132630245644921e-05, "loss": 0.0307, "step": 4560 }, { "grad_norm": 0.20752446353435516, "learning_rate": 6.116519201394857e-05, "loss": 0.0334, "step": 4570 }, { "grad_norm": 0.3370424211025238, "learning_rate": 6.10039594706969e-05, "loss": 0.0335, "step": 4580 }, { "grad_norm": 0.18843048810958862, "learning_rate": 6.084260658990744e-05, "loss": 0.0368, "step": 4590 }, { "grad_norm": 0.1898745745420456, "learning_rate": 6.068113513610943e-05, "loss": 0.0303, "step": 4600 }, { "grad_norm": 0.21216046810150146, "learning_rate": 6.0519546875128876e-05, "loss": 0.0333, "step": 4610 }, { "grad_norm": 0.17116063833236694, "learning_rate": 6.035784357406906e-05, "loss": 0.0326, "step": 4620 }, { "grad_norm": 0.18865202367305756, "learning_rate": 6.01960270012914e-05, "loss": 0.0349, "step": 4630 }, { "grad_norm": 0.1817384958267212, "learning_rate": 6.003409892639599e-05, "loss": 0.0342, "step": 4640 }, { "grad_norm": 0.24883867800235748, "learning_rate": 5.9872061120202336e-05, "loss": 0.0391, "step": 4650 }, { "grad_norm": 0.18746870756149292, "learning_rate": 5.9709915354729914e-05, "loss": 0.0392, "step": 4660 }, { "grad_norm": 0.21675528585910797, "learning_rate": 5.9547663403178824e-05, "loss": 0.0319, "step": 4670 }, { "grad_norm": 0.17447808384895325, "learning_rate": 5.9385307039910445e-05, "loss": 0.0291, "step": 4680 }, { "grad_norm": 0.19924499094486237, "learning_rate": 5.922284804042792e-05, "loss": 0.0362, "step": 4690 }, { "grad_norm": 0.29298165440559387, "learning_rate": 5.906028818135687e-05, "loss": 0.0375, "step": 4700 }, { "grad_norm": 0.27040165662765503, "learning_rate": 5.889762924042585e-05, "loss": 0.0334, "step": 4710 }, { "grad_norm": 0.2434110790491104, "learning_rate": 5.873487299644699e-05, "loss": 0.034, "step": 4720 }, { "grad_norm": 0.29628121852874756, "learning_rate": 5.857202122929649e-05, "loss": 0.0386, "step": 4730 }, { "grad_norm": 0.36151131987571716, "learning_rate": 5.840907571989518e-05, "loss": 0.0332, "step": 4740 }, { "grad_norm": 0.24773883819580078, "learning_rate": 5.824603825018904e-05, "loss": 0.0366, "step": 4750 }, { "grad_norm": 0.22567424178123474, "learning_rate": 5.808291060312975e-05, "loss": 0.0321, "step": 4760 }, { "grad_norm": 0.27796632051467896, "learning_rate": 5.7919694562655083e-05, "loss": 0.0323, "step": 4770 }, { "grad_norm": 0.26154422760009766, "learning_rate": 5.775639191366954e-05, "loss": 0.0366, "step": 4780 }, { "grad_norm": 0.28948602080345154, "learning_rate": 5.75930044420247e-05, "loss": 0.0279, "step": 4790 }, { "grad_norm": 0.18959516286849976, "learning_rate": 5.74295339344998e-05, "loss": 0.0292, "step": 4800 }, { "grad_norm": 0.12758132815361023, "learning_rate": 5.726598217878211e-05, "loss": 0.0274, "step": 4810 }, { "grad_norm": 0.17409145832061768, "learning_rate": 5.71023509634474e-05, "loss": 0.0269, "step": 4820 }, { "grad_norm": 0.25498098134994507, "learning_rate": 5.693864207794049e-05, "loss": 0.0344, "step": 4830 }, { "grad_norm": 0.32012808322906494, "learning_rate": 5.677485731255545e-05, "loss": 0.0388, "step": 4840 }, { "grad_norm": 0.26008155941963196, "learning_rate": 5.6610998458416296e-05, "loss": 0.0312, "step": 4850 }, { "grad_norm": 0.25674015283584595, "learning_rate": 5.644706730745716e-05, "loss": 0.0306, "step": 4860 }, { "grad_norm": 0.23774832487106323, "learning_rate": 5.628306565240287e-05, "loss": 0.0383, "step": 4870 }, { "grad_norm": 0.34003540873527527, "learning_rate": 5.611899528674923e-05, "loss": 0.0443, "step": 4880 }, { "grad_norm": 0.3130679130554199, "learning_rate": 5.595485800474349e-05, "loss": 0.0367, "step": 4890 }, { "grad_norm": 0.20688337087631226, "learning_rate": 5.579065560136467e-05, "loss": 0.0363, "step": 4900 }, { "grad_norm": 0.22606690227985382, "learning_rate": 5.562638987230392e-05, "loss": 0.0322, "step": 4910 }, { "grad_norm": 0.2599315345287323, "learning_rate": 5.546206261394498e-05, "loss": 0.0334, "step": 4920 }, { "grad_norm": 0.2506241202354431, "learning_rate": 5.529767562334437e-05, "loss": 0.0363, "step": 4930 }, { "grad_norm": 0.3361583948135376, "learning_rate": 5.5133230698211926e-05, "loss": 0.0287, "step": 4940 }, { "grad_norm": 0.21972353756427765, "learning_rate": 5.496872963689096e-05, "loss": 0.0375, "step": 4950 }, { "grad_norm": 0.27756720781326294, "learning_rate": 5.4804174238338756e-05, "loss": 0.0314, "step": 4960 }, { "grad_norm": 0.17228463292121887, "learning_rate": 5.463956630210678e-05, "loss": 0.0372, "step": 4970 }, { "grad_norm": 0.15606464445590973, "learning_rate": 5.4474907628321046e-05, "loss": 0.0404, "step": 4980 }, { "grad_norm": 0.19428372383117676, "learning_rate": 5.431020001766244e-05, "loss": 0.0283, "step": 4990 }, { "grad_norm": 0.24708117544651031, "learning_rate": 5.4145445271346986e-05, "loss": 0.0326, "step": 5000 }, { "grad_norm": 0.334903746843338, "learning_rate": 5.398064519110622e-05, "loss": 0.035, "step": 5010 }, { "grad_norm": 0.24107778072357178, "learning_rate": 5.3815801579167394e-05, "loss": 0.0296, "step": 5020 }, { "grad_norm": 0.2224382609128952, "learning_rate": 5.365091623823382e-05, "loss": 0.0304, "step": 5030 }, { "grad_norm": 0.22610916197299957, "learning_rate": 5.348599097146521e-05, "loss": 0.032, "step": 5040 }, { "grad_norm": 0.2989542782306671, "learning_rate": 5.3321027582457836e-05, "loss": 0.0298, "step": 5050 }, { "grad_norm": 0.2430366426706314, "learning_rate": 5.315602787522491e-05, "loss": 0.033, "step": 5060 }, { "grad_norm": 0.216997891664505, "learning_rate": 5.299099365417678e-05, "loss": 0.0322, "step": 5070 }, { "grad_norm": 0.12199892848730087, "learning_rate": 5.2825926724101236e-05, "loss": 0.0326, "step": 5080 }, { "grad_norm": 0.2357574701309204, "learning_rate": 5.26608288901438e-05, "loss": 0.032, "step": 5090 }, { "grad_norm": 0.309891939163208, "learning_rate": 5.24957019577879e-05, "loss": 0.038, "step": 5100 }, { "grad_norm": 0.16880495846271515, "learning_rate": 5.2330547732835266e-05, "loss": 0.0354, "step": 5110 }, { "grad_norm": 0.2748132646083832, "learning_rate": 5.2165368021385996e-05, "loss": 0.0447, "step": 5120 }, { "grad_norm": 0.2322409600019455, "learning_rate": 5.200016462981897e-05, "loss": 0.0326, "step": 5130 }, { "grad_norm": 0.2747410237789154, "learning_rate": 5.1834939364772015e-05, "loss": 0.0293, "step": 5140 }, { "grad_norm": 0.22204075753688812, "learning_rate": 5.166969403312214e-05, "loss": 0.0333, "step": 5150 }, { "grad_norm": 0.17951582372188568, "learning_rate": 5.1504430441965844e-05, "loss": 0.0332, "step": 5160 }, { "grad_norm": 0.2836768925189972, "learning_rate": 5.133915039859923e-05, "loss": 0.028, "step": 5170 }, { "grad_norm": 0.21958065032958984, "learning_rate": 5.1173855710498444e-05, "loss": 0.031, "step": 5180 }, { "grad_norm": 0.2701330780982971, "learning_rate": 5.100854818529967e-05, "loss": 0.0332, "step": 5190 }, { "grad_norm": 0.22045926749706268, "learning_rate": 5.084322963077951e-05, "loss": 0.0327, "step": 5200 }, { "grad_norm": 0.20469629764556885, "learning_rate": 5.067790185483522e-05, "loss": 0.0323, "step": 5210 }, { "grad_norm": 0.25377464294433594, "learning_rate": 5.0512566665464844e-05, "loss": 0.0273, "step": 5220 }, { "grad_norm": 0.22175857424736023, "learning_rate": 5.034722587074755e-05, "loss": 0.0341, "step": 5230 }, { "grad_norm": 0.2506747245788574, "learning_rate": 5.018188127882375e-05, "loss": 0.0327, "step": 5240 }, { "grad_norm": 0.21609587967395782, "learning_rate": 5.0016534697875417e-05, "loss": 0.0246, "step": 5250 }, { "grad_norm": 0.23126336932182312, "learning_rate": 4.9851187936106294e-05, "loss": 0.03, "step": 5260 }, { "grad_norm": 0.2796275317668915, "learning_rate": 4.968584280172206e-05, "loss": 0.0292, "step": 5270 }, { "grad_norm": 0.281387597322464, "learning_rate": 4.95205011029106e-05, "loss": 0.0305, "step": 5280 }, { "grad_norm": 0.25199365615844727, "learning_rate": 4.935516464782227e-05, "loss": 0.0323, "step": 5290 }, { "grad_norm": 0.263265460729599, "learning_rate": 4.918983524455003e-05, "loss": 0.0331, "step": 5300 }, { "grad_norm": 0.15472272038459778, "learning_rate": 4.9024514701109766e-05, "loss": 0.0297, "step": 5310 }, { "grad_norm": 0.21798932552337646, "learning_rate": 4.885920482542043e-05, "loss": 0.0296, "step": 5320 }, { "grad_norm": 0.253448486328125, "learning_rate": 4.869390742528438e-05, "loss": 0.026, "step": 5330 }, { "grad_norm": 0.23189914226531982, "learning_rate": 4.852862430836744e-05, "loss": 0.0374, "step": 5340 }, { "grad_norm": 0.2581343352794647, "learning_rate": 4.836335728217933e-05, "loss": 0.0313, "step": 5350 }, { "grad_norm": 0.20150990784168243, "learning_rate": 4.819810815405379e-05, "loss": 0.0281, "step": 5360 }, { "grad_norm": 0.1782112568616867, "learning_rate": 4.803287873112877e-05, "loss": 0.0273, "step": 5370 }, { "grad_norm": 0.16050639748573303, "learning_rate": 4.786767082032681e-05, "loss": 0.0316, "step": 5380 }, { "grad_norm": 0.23555327951908112, "learning_rate": 4.77024862283351e-05, "loss": 0.0292, "step": 5390 }, { "grad_norm": 0.27532029151916504, "learning_rate": 4.753732676158593e-05, "loss": 0.0318, "step": 5400 }, { "grad_norm": 0.3325684368610382, "learning_rate": 4.737219422623672e-05, "loss": 0.03, "step": 5410 }, { "grad_norm": 0.19526585936546326, "learning_rate": 4.720709042815044e-05, "loss": 0.0301, "step": 5420 }, { "grad_norm": 0.2934618890285492, "learning_rate": 4.704201717287578e-05, "loss": 0.0341, "step": 5430 }, { "grad_norm": 0.2733133137226105, "learning_rate": 4.6876976265627404e-05, "loss": 0.0346, "step": 5440 }, { "grad_norm": 0.3040599226951599, "learning_rate": 4.671196951126626e-05, "loss": 0.0369, "step": 5450 }, { "grad_norm": 0.20710590481758118, "learning_rate": 4.654699871427971e-05, "loss": 0.0316, "step": 5460 }, { "grad_norm": 0.24290040135383606, "learning_rate": 4.6382065678762034e-05, "loss": 0.0329, "step": 5470 }, { "grad_norm": 0.17691636085510254, "learning_rate": 4.6217172208394424e-05, "loss": 0.0366, "step": 5480 }, { "grad_norm": 0.20657895505428314, "learning_rate": 4.605232010642549e-05, "loss": 0.0266, "step": 5490 }, { "grad_norm": 0.2362714558839798, "learning_rate": 4.588751117565142e-05, "loss": 0.0283, "step": 5500 }, { "grad_norm": 0.3155677020549774, "learning_rate": 4.5722747218396214e-05, "loss": 0.0322, "step": 5510 }, { "grad_norm": 0.30780959129333496, "learning_rate": 4.5558030036492194e-05, "loss": 0.03, "step": 5520 }, { "grad_norm": 0.24017433822155, "learning_rate": 4.539336143125999e-05, "loss": 0.0343, "step": 5530 }, { "grad_norm": 0.2758651673793793, "learning_rate": 4.522874320348916e-05, "loss": 0.0326, "step": 5540 }, { "grad_norm": 0.19067950546741486, "learning_rate": 4.506417715341821e-05, "loss": 0.0297, "step": 5550 }, { "grad_norm": 0.3021450638771057, "learning_rate": 4.489966508071511e-05, "loss": 0.0303, "step": 5560 }, { "grad_norm": 0.3083040118217468, "learning_rate": 4.4735208784457575e-05, "loss": 0.0248, "step": 5570 }, { "grad_norm": 0.21832144260406494, "learning_rate": 4.457081006311325e-05, "loss": 0.0301, "step": 5580 }, { "grad_norm": 0.261488139629364, "learning_rate": 4.440647071452027e-05, "loss": 0.0351, "step": 5590 }, { "grad_norm": 0.247087761759758, "learning_rate": 4.424219253586737e-05, "loss": 0.032, "step": 5600 }, { "grad_norm": 0.18506105244159698, "learning_rate": 4.407797732367443e-05, "loss": 0.035, "step": 5610 }, { "grad_norm": 0.2241569310426712, "learning_rate": 4.391382687377268e-05, "loss": 0.029, "step": 5620 }, { "grad_norm": 0.25591757893562317, "learning_rate": 4.374974298128512e-05, "loss": 0.0324, "step": 5630 }, { "grad_norm": 0.28581759333610535, "learning_rate": 4.358572744060699e-05, "loss": 0.0317, "step": 5640 }, { "grad_norm": 0.3334788680076599, "learning_rate": 4.342178204538588e-05, "loss": 0.0273, "step": 5650 }, { "grad_norm": 0.2767829895019531, "learning_rate": 4.325790858850241e-05, "loss": 0.031, "step": 5660 }, { "grad_norm": 0.29938843846321106, "learning_rate": 4.309410886205043e-05, "loss": 0.0336, "step": 5670 }, { "grad_norm": 0.18191973865032196, "learning_rate": 4.293038465731752e-05, "loss": 0.0268, "step": 5680 }, { "grad_norm": 0.25212907791137695, "learning_rate": 4.276673776476533e-05, "loss": 0.0328, "step": 5690 }, { "grad_norm": 0.18657682836055756, "learning_rate": 4.260316997401007e-05, "loss": 0.0263, "step": 5700 }, { "grad_norm": 0.21522128582000732, "learning_rate": 4.243968307380293e-05, "loss": 0.0291, "step": 5710 }, { "grad_norm": 0.22109003365039825, "learning_rate": 4.22762788520104e-05, "loss": 0.0322, "step": 5720 }, { "grad_norm": 0.22448302805423737, "learning_rate": 4.211295909559491e-05, "loss": 0.0324, "step": 5730 }, { "grad_norm": 0.18194463849067688, "learning_rate": 4.194972559059511e-05, "loss": 0.0248, "step": 5740 }, { "grad_norm": 0.19496603310108185, "learning_rate": 4.178658012210651e-05, "loss": 0.0297, "step": 5750 }, { "grad_norm": 0.18776461482048035, "learning_rate": 4.162352447426177e-05, "loss": 0.0288, "step": 5760 }, { "grad_norm": 0.13050873577594757, "learning_rate": 4.146056043021135e-05, "loss": 0.0296, "step": 5770 }, { "grad_norm": 0.1322217583656311, "learning_rate": 4.1297689772103944e-05, "loss": 0.0292, "step": 5780 }, { "grad_norm": 0.18648572266101837, "learning_rate": 4.113491428106694e-05, "loss": 0.0276, "step": 5790 }, { "grad_norm": 0.23980101943016052, "learning_rate": 4.0972235737187055e-05, "loss": 0.0293, "step": 5800 }, { "grad_norm": 0.24723590910434723, "learning_rate": 4.080965591949076e-05, "loss": 0.0267, "step": 5810 }, { "grad_norm": 0.20583786070346832, "learning_rate": 4.0647176605924924e-05, "loss": 0.0328, "step": 5820 }, { "grad_norm": 0.17424654960632324, "learning_rate": 4.0484799573337255e-05, "loss": 0.032, "step": 5830 }, { "grad_norm": 0.20658162236213684, "learning_rate": 4.032252659745699e-05, "loss": 0.025, "step": 5840 }, { "grad_norm": 0.2736555337905884, "learning_rate": 4.016035945287539e-05, "loss": 0.0384, "step": 5850 }, { "grad_norm": 0.2825140058994293, "learning_rate": 3.999829991302635e-05, "loss": 0.0304, "step": 5860 }, { "grad_norm": 0.23452791571617126, "learning_rate": 3.983634975016707e-05, "loss": 0.0293, "step": 5870 }, { "grad_norm": 0.18170562386512756, "learning_rate": 3.967451073535854e-05, "loss": 0.0334, "step": 5880 }, { "grad_norm": 0.21897168457508087, "learning_rate": 3.951278463844633e-05, "loss": 0.027, "step": 5890 }, { "grad_norm": 0.2021874189376831, "learning_rate": 3.935117322804111e-05, "loss": 0.0272, "step": 5900 }, { "grad_norm": 0.18219435214996338, "learning_rate": 3.918967827149938e-05, "loss": 0.0294, "step": 5910 }, { "grad_norm": 0.1835804134607315, "learning_rate": 3.9028301534904094e-05, "loss": 0.0265, "step": 5920 }, { "grad_norm": 0.2243785858154297, "learning_rate": 3.88670447830454e-05, "loss": 0.0252, "step": 5930 }, { "grad_norm": 0.1894894689321518, "learning_rate": 3.870590977940132e-05, "loss": 0.0273, "step": 5940 }, { "grad_norm": 0.1670093834400177, "learning_rate": 3.8544898286118404e-05, "loss": 0.0258, "step": 5950 }, { "grad_norm": 0.223557248711586, "learning_rate": 3.838401206399257e-05, "loss": 0.0312, "step": 5960 }, { "grad_norm": 0.22385331988334656, "learning_rate": 3.822325287244975e-05, "loss": 0.0247, "step": 5970 }, { "grad_norm": 0.4498996138572693, "learning_rate": 3.8062622469526725e-05, "loss": 0.03, "step": 5980 }, { "grad_norm": 0.19648559391498566, "learning_rate": 3.790212261185183e-05, "loss": 0.0243, "step": 5990 }, { "grad_norm": 0.1755073219537735, "learning_rate": 3.7741755054625794e-05, "loss": 0.0257, "step": 6000 }, { "grad_norm": 0.29500216245651245, "learning_rate": 3.758152155160255e-05, "loss": 0.0257, "step": 6010 }, { "grad_norm": 0.35184305906295776, "learning_rate": 3.742142385506999e-05, "loss": 0.0313, "step": 6020 }, { "grad_norm": 0.25773724913597107, "learning_rate": 3.72614637158309e-05, "loss": 0.023, "step": 6030 }, { "grad_norm": 0.39594027400016785, "learning_rate": 3.710164288318371e-05, "loss": 0.0324, "step": 6040 }, { "grad_norm": 0.19185921549797058, "learning_rate": 3.694196310490345e-05, "loss": 0.0237, "step": 6050 }, { "grad_norm": 0.3376231789588928, "learning_rate": 3.678242612722259e-05, "loss": 0.0254, "step": 6060 }, { "grad_norm": 0.5873315930366516, "learning_rate": 3.6623033694811953e-05, "loss": 0.0313, "step": 6070 }, { "grad_norm": 0.1866593211889267, "learning_rate": 3.6463787550761665e-05, "loss": 0.0296, "step": 6080 }, { "grad_norm": 0.22215835750102997, "learning_rate": 3.630468943656202e-05, "loss": 0.0266, "step": 6090 }, { "grad_norm": 0.22308844327926636, "learning_rate": 3.6145741092084523e-05, "loss": 0.0266, "step": 6100 }, { "grad_norm": 0.1451566070318222, "learning_rate": 3.598694425556278e-05, "loss": 0.0221, "step": 6110 }, { "grad_norm": 0.13414019346237183, "learning_rate": 3.58283006635736e-05, "loss": 0.0274, "step": 6120 }, { "grad_norm": 0.1368788629770279, "learning_rate": 3.566981205101781e-05, "loss": 0.0266, "step": 6130 }, { "grad_norm": 0.1688688099384308, "learning_rate": 3.5511480151101556e-05, "loss": 0.0263, "step": 6140 }, { "grad_norm": 0.19245925545692444, "learning_rate": 3.5353306695317104e-05, "loss": 0.0281, "step": 6150 }, { "grad_norm": 0.1715364307165146, "learning_rate": 3.519529341342402e-05, "loss": 0.0312, "step": 6160 }, { "grad_norm": 0.14107900857925415, "learning_rate": 3.503744203343026e-05, "loss": 0.0251, "step": 6170 }, { "grad_norm": 0.15094493329524994, "learning_rate": 3.487975428157318e-05, "loss": 0.0245, "step": 6180 }, { "grad_norm": 0.26728346943855286, "learning_rate": 3.472223188230083e-05, "loss": 0.0281, "step": 6190 }, { "grad_norm": 0.19935046136379242, "learning_rate": 3.4564876558252866e-05, "loss": 0.0275, "step": 6200 }, { "grad_norm": 0.46773508191108704, "learning_rate": 3.440769003024195e-05, "loss": 0.0283, "step": 6210 }, { "grad_norm": 0.1736050248146057, "learning_rate": 3.425067401723477e-05, "loss": 0.0272, "step": 6220 }, { "grad_norm": 0.2410256266593933, "learning_rate": 3.409383023633325e-05, "loss": 0.0269, "step": 6230 }, { "grad_norm": 0.22913150489330292, "learning_rate": 3.3937160402755894e-05, "loss": 0.0276, "step": 6240 }, { "grad_norm": 0.17796209454536438, "learning_rate": 3.378066622981885e-05, "loss": 0.0266, "step": 6250 }, { "grad_norm": 0.1465309113264084, "learning_rate": 3.362434942891738e-05, "loss": 0.0324, "step": 6260 }, { "grad_norm": 0.16443581879138947, "learning_rate": 3.346821170950693e-05, "loss": 0.0292, "step": 6270 }, { "grad_norm": 0.12880082428455353, "learning_rate": 3.3312254779084585e-05, "loss": 0.0242, "step": 6280 }, { "grad_norm": 0.1502242088317871, "learning_rate": 3.315648034317039e-05, "loss": 0.0288, "step": 6290 }, { "grad_norm": 0.21227800846099854, "learning_rate": 3.3000890105288564e-05, "loss": 0.0296, "step": 6300 }, { "grad_norm": 0.25768381357192993, "learning_rate": 3.284548576694908e-05, "loss": 0.0218, "step": 6310 }, { "grad_norm": 0.3948172926902771, "learning_rate": 3.2690269027628815e-05, "loss": 0.0233, "step": 6320 }, { "grad_norm": 0.18972599506378174, "learning_rate": 3.253524158475324e-05, "loss": 0.0257, "step": 6330 }, { "grad_norm": 0.209112286567688, "learning_rate": 3.238040513367757e-05, "loss": 0.0233, "step": 6340 }, { "grad_norm": 0.289010614156723, "learning_rate": 3.222576136766843e-05, "loss": 0.0252, "step": 6350 }, { "grad_norm": 0.15033666789531708, "learning_rate": 3.2071311977885324e-05, "loss": 0.0216, "step": 6360 }, { "grad_norm": 0.174081951379776, "learning_rate": 3.191705865336197e-05, "loss": 0.0313, "step": 6370 }, { "grad_norm": 0.19547191262245178, "learning_rate": 3.1763003080988075e-05, "loss": 0.0207, "step": 6380 }, { "grad_norm": 0.21565386652946472, "learning_rate": 3.160914694549063e-05, "loss": 0.0278, "step": 6390 }, { "grad_norm": 0.13888418674468994, "learning_rate": 3.145549192941573e-05, "loss": 0.0229, "step": 6400 }, { "grad_norm": 0.2355973720550537, "learning_rate": 3.130203971310999e-05, "loss": 0.028, "step": 6410 }, { "grad_norm": 0.3216745853424072, "learning_rate": 3.114879197470225e-05, "loss": 0.0269, "step": 6420 }, { "grad_norm": 0.18165922164916992, "learning_rate": 3.0995750390085285e-05, "loss": 0.0212, "step": 6430 }, { "grad_norm": 0.21047364175319672, "learning_rate": 3.084291663289728e-05, "loss": 0.0229, "step": 6440 }, { "grad_norm": 0.32176026701927185, "learning_rate": 3.069029237450375e-05, "loss": 0.022, "step": 6450 }, { "grad_norm": 0.30684152245521545, "learning_rate": 3.053787928397911e-05, "loss": 0.0224, "step": 6460 }, { "grad_norm": 0.23404596745967865, "learning_rate": 3.0385679028088526e-05, "loss": 0.0279, "step": 6470 }, { "grad_norm": 0.2234848290681839, "learning_rate": 3.023369327126959e-05, "loss": 0.025, "step": 6480 }, { "grad_norm": 0.1955266296863556, "learning_rate": 3.0081923675614198e-05, "loss": 0.0238, "step": 6490 }, { "grad_norm": 0.22963213920593262, "learning_rate": 2.993037190085034e-05, "loss": 0.0229, "step": 6500 }, { "grad_norm": 0.16760241985321045, "learning_rate": 2.977903960432392e-05, "loss": 0.0253, "step": 6510 }, { "grad_norm": 0.2523892819881439, "learning_rate": 2.9627928440980722e-05, "loss": 0.0292, "step": 6520 }, { "grad_norm": 0.2927192747592926, "learning_rate": 2.9477040063348183e-05, "loss": 0.0267, "step": 6530 }, { "grad_norm": 0.1315109133720398, "learning_rate": 2.9326376121517456e-05, "loss": 0.0262, "step": 6540 }, { "grad_norm": 0.22135543823242188, "learning_rate": 2.9175938263125236e-05, "loss": 0.0272, "step": 6550 }, { "grad_norm": 0.1749742180109024, "learning_rate": 2.9025728133335873e-05, "loss": 0.0251, "step": 6560 }, { "grad_norm": 0.19801437854766846, "learning_rate": 2.8875747374823288e-05, "loss": 0.0296, "step": 6570 }, { "grad_norm": 0.2606129050254822, "learning_rate": 2.872599762775298e-05, "loss": 0.0335, "step": 6580 }, { "grad_norm": 0.2292376607656479, "learning_rate": 2.857648052976425e-05, "loss": 0.023, "step": 6590 }, { "grad_norm": 0.18571710586547852, "learning_rate": 2.8427197715952047e-05, "loss": 0.0254, "step": 6600 }, { "grad_norm": 0.3039461672306061, "learning_rate": 2.8278150818849393e-05, "loss": 0.0271, "step": 6610 }, { "grad_norm": 0.22344568371772766, "learning_rate": 2.812934146840922e-05, "loss": 0.0271, "step": 6620 }, { "grad_norm": 0.22500230371952057, "learning_rate": 2.7980771291986764e-05, "loss": 0.0307, "step": 6630 }, { "grad_norm": 0.16794003546237946, "learning_rate": 2.783244191432167e-05, "loss": 0.0261, "step": 6640 }, { "grad_norm": 0.18268553912639618, "learning_rate": 2.768435495752022e-05, "loss": 0.0272, "step": 6650 }, { "grad_norm": 0.17268106341362, "learning_rate": 2.753651204103771e-05, "loss": 0.0256, "step": 6660 }, { "grad_norm": 0.1506766527891159, "learning_rate": 2.7388914781660523e-05, "loss": 0.0226, "step": 6670 }, { "grad_norm": 0.23423531651496887, "learning_rate": 2.7241564793488693e-05, "loss": 0.0271, "step": 6680 }, { "grad_norm": 0.2359861135482788, "learning_rate": 2.7094463687918037e-05, "loss": 0.0252, "step": 6690 }, { "grad_norm": 0.21988804638385773, "learning_rate": 2.694761307362268e-05, "loss": 0.0257, "step": 6700 }, { "grad_norm": 0.22502481937408447, "learning_rate": 2.6801014556537467e-05, "loss": 0.0259, "step": 6710 }, { "grad_norm": 0.19089630246162415, "learning_rate": 2.6654669739840243e-05, "loss": 0.0242, "step": 6720 }, { "grad_norm": 0.19184783101081848, "learning_rate": 2.650858022393451e-05, "loss": 0.0256, "step": 6730 }, { "grad_norm": 0.19900085031986237, "learning_rate": 2.6362747606431747e-05, "loss": 0.025, "step": 6740 }, { "grad_norm": 0.15911638736724854, "learning_rate": 2.6217173482134172e-05, "loss": 0.0241, "step": 6750 }, { "grad_norm": 0.21335040032863617, "learning_rate": 2.6071859443017044e-05, "loss": 0.0233, "step": 6760 }, { "grad_norm": 0.17870742082595825, "learning_rate": 2.5926807078211414e-05, "loss": 0.0218, "step": 6770 }, { "grad_norm": 0.20266583561897278, "learning_rate": 2.5782017973986728e-05, "loss": 0.0224, "step": 6780 }, { "grad_norm": 0.9970222115516663, "learning_rate": 2.5637493713733374e-05, "loss": 0.0276, "step": 6790 }, { "grad_norm": 0.25093814730644226, "learning_rate": 2.549323587794559e-05, "loss": 0.0263, "step": 6800 }, { "grad_norm": 0.17274348437786102, "learning_rate": 2.5349246044203895e-05, "loss": 0.0227, "step": 6810 }, { "grad_norm": 0.20156778395175934, "learning_rate": 2.520552578715808e-05, "loss": 0.023, "step": 6820 }, { "grad_norm": 0.14809125661849976, "learning_rate": 2.506207667850981e-05, "loss": 0.0219, "step": 6830 }, { "grad_norm": 0.17152774333953857, "learning_rate": 2.4918900286995555e-05, "loss": 0.0214, "step": 6840 }, { "grad_norm": 0.14345024526119232, "learning_rate": 2.4775998178369458e-05, "loss": 0.0226, "step": 6850 }, { "grad_norm": 0.18765637278556824, "learning_rate": 2.4633371915386017e-05, "loss": 0.0269, "step": 6860 }, { "grad_norm": 0.23587657511234283, "learning_rate": 2.4491023057783235e-05, "loss": 0.0238, "step": 6870 }, { "grad_norm": 0.19875763356685638, "learning_rate": 2.4348953162265375e-05, "loss": 0.0275, "step": 6880 }, { "grad_norm": 0.15419048070907593, "learning_rate": 2.420716378248607e-05, "loss": 0.0258, "step": 6890 }, { "grad_norm": 0.31757232546806335, "learning_rate": 2.4065656469031266e-05, "loss": 0.0242, "step": 6900 }, { "grad_norm": 0.17979766428470612, "learning_rate": 2.3924432769402268e-05, "loss": 0.0283, "step": 6910 }, { "grad_norm": 0.37366753816604614, "learning_rate": 2.3783494227998844e-05, "loss": 0.0356, "step": 6920 }, { "grad_norm": 0.2540377378463745, "learning_rate": 2.3642842386102264e-05, "loss": 0.0236, "step": 6930 }, { "grad_norm": 0.23288565874099731, "learning_rate": 2.3502478781858567e-05, "loss": 0.0217, "step": 6940 }, { "grad_norm": 0.23346400260925293, "learning_rate": 2.3362404950261628e-05, "loss": 0.0309, "step": 6950 }, { "grad_norm": 0.26651233434677124, "learning_rate": 2.3222622423136458e-05, "loss": 0.0238, "step": 6960 }, { "grad_norm": 0.15820765495300293, "learning_rate": 2.3083132729122332e-05, "loss": 0.0236, "step": 6970 }, { "grad_norm": 0.24089506268501282, "learning_rate": 2.294393739365621e-05, "loss": 0.0245, "step": 6980 }, { "grad_norm": 0.16337230801582336, "learning_rate": 2.2805037938956e-05, "loss": 0.0239, "step": 6990 }, { "grad_norm": 0.18966597318649292, "learning_rate": 2.266643588400386e-05, "loss": 0.0234, "step": 7000 }, { "grad_norm": 0.1657150238752365, "learning_rate": 2.252813274452969e-05, "loss": 0.0214, "step": 7010 }, { "grad_norm": 0.22055990993976593, "learning_rate": 2.2390130032994427e-05, "loss": 0.0205, "step": 7020 }, { "grad_norm": 0.15004153549671173, "learning_rate": 2.2252429258573633e-05, "loss": 0.0219, "step": 7030 }, { "grad_norm": 0.1589592546224594, "learning_rate": 2.2115031927140904e-05, "loss": 0.0249, "step": 7040 }, { "grad_norm": 0.23201224207878113, "learning_rate": 2.1977939541251463e-05, "loss": 0.0273, "step": 7050 }, { "grad_norm": 0.15582187473773956, "learning_rate": 2.1841153600125684e-05, "loss": 0.0228, "step": 7060 }, { "grad_norm": 0.1505759209394455, "learning_rate": 2.170467559963267e-05, "loss": 0.0281, "step": 7070 }, { "grad_norm": 0.14130666851997375, "learning_rate": 2.1568507032273982e-05, "loss": 0.0191, "step": 7080 }, { "grad_norm": 0.21408721804618835, "learning_rate": 2.1432649387167264e-05, "loss": 0.0191, "step": 7090 }, { "grad_norm": 0.17760644853115082, "learning_rate": 2.1297104150029973e-05, "loss": 0.0228, "step": 7100 }, { "grad_norm": 0.17090746760368347, "learning_rate": 2.116187280316307e-05, "loss": 0.0233, "step": 7110 }, { "grad_norm": 0.16623882949352264, "learning_rate": 2.1026956825434908e-05, "loss": 0.0227, "step": 7120 }, { "grad_norm": 0.19975309073925018, "learning_rate": 2.0892357692265017e-05, "loss": 0.021, "step": 7130 }, { "grad_norm": 0.1825215071439743, "learning_rate": 2.0758076875607947e-05, "loss": 0.024, "step": 7140 }, { "grad_norm": 0.11679578572511673, "learning_rate": 2.0624115843937207e-05, "loss": 0.0221, "step": 7150 }, { "grad_norm": 0.13711011409759521, "learning_rate": 2.0490476062229157e-05, "loss": 0.0228, "step": 7160 }, { "grad_norm": 0.1472814679145813, "learning_rate": 2.035715899194704e-05, "loss": 0.0226, "step": 7170 }, { "grad_norm": 0.22399266064167023, "learning_rate": 2.022416609102499e-05, "loss": 0.0223, "step": 7180 }, { "grad_norm": 0.1668604463338852, "learning_rate": 2.009149881385205e-05, "loss": 0.0253, "step": 7190 }, { "grad_norm": 0.20802457630634308, "learning_rate": 1.995915861125634e-05, "loss": 0.0204, "step": 7200 }, { "grad_norm": 0.20994879305362701, "learning_rate": 1.9827146930489065e-05, "loss": 0.0209, "step": 7210 }, { "grad_norm": 0.18538501858711243, "learning_rate": 1.9695465215208848e-05, "loss": 0.0211, "step": 7220 }, { "grad_norm": 0.12344224750995636, "learning_rate": 1.9564114905465813e-05, "loss": 0.0181, "step": 7230 }, { "grad_norm": 0.20994114875793457, "learning_rate": 1.9433097437685936e-05, "loss": 0.0213, "step": 7240 }, { "grad_norm": 0.19194720685482025, "learning_rate": 1.930241424465521e-05, "loss": 0.0173, "step": 7250 }, { "grad_norm": 0.1713903248310089, "learning_rate": 1.9172066755504115e-05, "loss": 0.0185, "step": 7260 }, { "grad_norm": 0.18851259350776672, "learning_rate": 1.9042056395691914e-05, "loss": 0.0206, "step": 7270 }, { "grad_norm": 0.1612647920846939, "learning_rate": 1.8912384586991066e-05, "loss": 0.0192, "step": 7280 }, { "grad_norm": 0.34915706515312195, "learning_rate": 1.8783052747471717e-05, "loss": 0.0241, "step": 7290 }, { "grad_norm": 0.2750299870967865, "learning_rate": 1.865406229148611e-05, "loss": 0.0233, "step": 7300 }, { "grad_norm": 0.11135426163673401, "learning_rate": 1.8525414629653233e-05, "loss": 0.0266, "step": 7310 }, { "grad_norm": 0.16879354417324066, "learning_rate": 1.8397111168843255e-05, "loss": 0.021, "step": 7320 }, { "grad_norm": 0.23469537496566772, "learning_rate": 1.8269153312162323e-05, "loss": 0.0267, "step": 7330 }, { "grad_norm": 0.1478690654039383, "learning_rate": 1.8141542458937054e-05, "loss": 0.0225, "step": 7340 }, { "grad_norm": 0.21817472577095032, "learning_rate": 1.8014280004699268e-05, "loss": 0.0212, "step": 7350 }, { "grad_norm": 0.20167005062103271, "learning_rate": 1.788736734117078e-05, "loss": 0.0193, "step": 7360 }, { "grad_norm": 0.14684022963047028, "learning_rate": 1.7760805856248152e-05, "loss": 0.0221, "step": 7370 }, { "grad_norm": 0.23978343605995178, "learning_rate": 1.7634596933987518e-05, "loss": 0.0209, "step": 7380 }, { "grad_norm": 0.24468277394771576, "learning_rate": 1.7508741954589404e-05, "loss": 0.0288, "step": 7390 }, { "grad_norm": 0.2819056510925293, "learning_rate": 1.7383242294383717e-05, "loss": 0.0266, "step": 7400 }, { "grad_norm": 0.25656548142433167, "learning_rate": 1.7258099325814632e-05, "loss": 0.0216, "step": 7410 }, { "grad_norm": 0.23810173571109772, "learning_rate": 1.7133314417425594e-05, "loss": 0.0245, "step": 7420 }, { "grad_norm": 0.15414395928382874, "learning_rate": 1.7008888933844408e-05, "loss": 0.0221, "step": 7430 }, { "grad_norm": 0.17395943403244019, "learning_rate": 1.6884824235768172e-05, "loss": 0.0227, "step": 7440 }, { "grad_norm": 0.2847509980201721, "learning_rate": 1.6761121679948592e-05, "loss": 0.0229, "step": 7450 }, { "grad_norm": 0.1686169058084488, "learning_rate": 1.663778261917695e-05, "loss": 0.023, "step": 7460 }, { "grad_norm": 0.16578781604766846, "learning_rate": 1.651480840226952e-05, "loss": 0.0225, "step": 7470 }, { "grad_norm": 0.15681475400924683, "learning_rate": 1.639220037405258e-05, "loss": 0.0218, "step": 7480 }, { "grad_norm": 0.13499349355697632, "learning_rate": 1.6269959875347906e-05, "loss": 0.0232, "step": 7490 }, { "grad_norm": 0.26179108023643494, "learning_rate": 1.614808824295802e-05, "loss": 0.0221, "step": 7500 }, { "grad_norm": 0.19313094019889832, "learning_rate": 1.602658680965152e-05, "loss": 0.0208, "step": 7510 }, { "grad_norm": 0.15526092052459717, "learning_rate": 1.5905456904148686e-05, "loss": 0.0229, "step": 7520 }, { "grad_norm": 0.19335737824440002, "learning_rate": 1.57846998511067e-05, "loss": 0.0246, "step": 7530 }, { "grad_norm": 0.10295336693525314, "learning_rate": 1.566431697110538e-05, "loss": 0.0174, "step": 7540 }, { "grad_norm": 0.18111182749271393, "learning_rate": 1.554430958063259e-05, "loss": 0.022, "step": 7550 }, { "grad_norm": 0.21188095211982727, "learning_rate": 1.5424678992069912e-05, "loss": 0.0204, "step": 7560 }, { "grad_norm": 0.27052873373031616, "learning_rate": 1.5305426513678362e-05, "loss": 0.0235, "step": 7570 }, { "grad_norm": 0.19758804142475128, "learning_rate": 1.518655344958388e-05, "loss": 0.022, "step": 7580 }, { "grad_norm": 0.27173346281051636, "learning_rate": 1.5068061099763275e-05, "loss": 0.0209, "step": 7590 }, { "grad_norm": 0.16893143951892853, "learning_rate": 1.494995076002988e-05, "loss": 0.0229, "step": 7600 }, { "grad_norm": 0.139555886387825, "learning_rate": 1.4832223722019456e-05, "loss": 0.018, "step": 7610 }, { "grad_norm": 0.29791831970214844, "learning_rate": 1.4714881273176035e-05, "loss": 0.0197, "step": 7620 }, { "grad_norm": 0.1509089320898056, "learning_rate": 1.4597924696737835e-05, "loss": 0.0213, "step": 7630 }, { "grad_norm": 0.1784256100654602, "learning_rate": 1.4481355271723252e-05, "loss": 0.02, "step": 7640 }, { "grad_norm": 0.19197262823581696, "learning_rate": 1.4365174272916809e-05, "loss": 0.0209, "step": 7650 }, { "grad_norm": 0.14417459070682526, "learning_rate": 1.4249382970855319e-05, "loss": 0.0176, "step": 7660 }, { "grad_norm": 0.13155497610569, "learning_rate": 1.4133982631813903e-05, "loss": 0.0224, "step": 7670 }, { "grad_norm": 0.2259848266839981, "learning_rate": 1.4018974517792194e-05, "loss": 0.0212, "step": 7680 }, { "grad_norm": 0.14182083308696747, "learning_rate": 1.390435988650048e-05, "loss": 0.0202, "step": 7690 }, { "grad_norm": 0.13303646445274353, "learning_rate": 1.3790139991346006e-05, "loss": 0.0196, "step": 7700 }, { "grad_norm": 0.21113501489162445, "learning_rate": 1.367631608141926e-05, "loss": 0.02, "step": 7710 }, { "grad_norm": 0.16198165714740753, "learning_rate": 1.3562889401480278e-05, "loss": 0.0194, "step": 7720 }, { "grad_norm": 0.09534993022680283, "learning_rate": 1.3449861191945074e-05, "loss": 0.02, "step": 7730 }, { "grad_norm": 0.1794901192188263, "learning_rate": 1.3337232688872009e-05, "loss": 0.019, "step": 7740 }, { "grad_norm": 0.14858315885066986, "learning_rate": 1.3225005123948364e-05, "loss": 0.0166, "step": 7750 }, { "grad_norm": 0.32794639468193054, "learning_rate": 1.311317972447681e-05, "loss": 0.0189, "step": 7760 }, { "grad_norm": 0.2175934910774231, "learning_rate": 1.3001757713361996e-05, "loss": 0.0162, "step": 7770 }, { "grad_norm": 0.1590438336133957, "learning_rate": 1.2890740309097204e-05, "loss": 0.0209, "step": 7780 }, { "grad_norm": 0.23166044056415558, "learning_rate": 1.2780128725750944e-05, "loss": 0.0187, "step": 7790 }, { "grad_norm": 0.2898387014865875, "learning_rate": 1.266992417295379e-05, "loss": 0.0222, "step": 7800 }, { "grad_norm": 0.11488961428403854, "learning_rate": 1.2560127855885073e-05, "loss": 0.0225, "step": 7810 }, { "grad_norm": 0.19540274143218994, "learning_rate": 1.2450740975259745e-05, "loss": 0.0235, "step": 7820 }, { "grad_norm": 0.1348259150981903, "learning_rate": 1.234176472731517e-05, "loss": 0.0246, "step": 7830 }, { "grad_norm": 0.18963028490543365, "learning_rate": 1.2233200303798158e-05, "loss": 0.0209, "step": 7840 }, { "grad_norm": 0.2206285297870636, "learning_rate": 1.2125048891951846e-05, "loss": 0.0191, "step": 7850 }, { "grad_norm": 0.7746766805648804, "learning_rate": 1.2017311674502745e-05, "loss": 0.0175, "step": 7860 }, { "grad_norm": 0.19611328840255737, "learning_rate": 1.1909989829647822e-05, "loss": 0.0204, "step": 7870 }, { "grad_norm": 0.23876555263996124, "learning_rate": 1.1803084531041553e-05, "loss": 0.0195, "step": 7880 }, { "grad_norm": 0.1559199094772339, "learning_rate": 1.1696596947783162e-05, "loss": 0.0194, "step": 7890 }, { "grad_norm": 0.16328750550746918, "learning_rate": 1.1590528244403803e-05, "loss": 0.016, "step": 7900 }, { "grad_norm": 0.20925039052963257, "learning_rate": 1.148487958085382e-05, "loss": 0.0198, "step": 7910 }, { "grad_norm": 0.1594330072402954, "learning_rate": 1.1379652112490086e-05, "loss": 0.0188, "step": 7920 }, { "grad_norm": 0.26968154311180115, "learning_rate": 1.1274846990063315e-05, "loss": 0.0226, "step": 7930 }, { "grad_norm": 0.2140229046344757, "learning_rate": 1.117046535970554e-05, "loss": 0.0184, "step": 7940 }, { "grad_norm": 0.2111728936433792, "learning_rate": 1.106650836291755e-05, "loss": 0.0241, "step": 7950 }, { "grad_norm": 0.16894827783107758, "learning_rate": 1.0962977136556418e-05, "loss": 0.0224, "step": 7960 }, { "grad_norm": 0.17922641336917877, "learning_rate": 1.0859872812823024e-05, "loss": 0.0179, "step": 7970 }, { "grad_norm": 0.2044275999069214, "learning_rate": 1.0757196519249747e-05, "loss": 0.023, "step": 7980 }, { "grad_norm": 0.18761326372623444, "learning_rate": 1.0654949378688077e-05, "loss": 0.019, "step": 7990 }, { "grad_norm": 0.16072678565979004, "learning_rate": 1.0553132509296376e-05, "loss": 0.0217, "step": 8000 } ], "logging_steps": 10, "max_steps": 10000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 4000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 32, "trial_name": null, "trial_params": null }