diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,18106 +1,18136 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.9991549435800566, - "global_step": 301500, + "epoch": 0.9991860934470597, + "global_step": 302000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, - "learning_rate": 4.9983430266275624e-05, - "loss": 3.5378, + "learning_rate": 4.9983457183883334e-05, + "loss": 3.2639, "step": 100 }, { "epoch": 0.0, - "learning_rate": 4.9966860532551245e-05, - "loss": 3.514, + "learning_rate": 4.996691436776665e-05, + "loss": 3.2781, "step": 200 }, { "epoch": 0.0, - "learning_rate": 4.995029079882687e-05, - "loss": 3.5069, + "learning_rate": 4.995037155164998e-05, + "loss": 3.2758, "step": 300 }, { "epoch": 0.0, - "learning_rate": 4.993372106510248e-05, - "loss": 3.5127, + "learning_rate": 4.9933828735533314e-05, + "loss": 3.2775, "step": 400 }, { "epoch": 0.0, - "learning_rate": 4.991715133137811e-05, - "loss": 3.4982, + "learning_rate": 4.991728591941663e-05, + "loss": 3.2774, "step": 500 }, { "epoch": 0.0, - "learning_rate": 4.990058159765373e-05, - "loss": 3.4956, + "learning_rate": 4.990074310329996e-05, + "loss": 3.2595, "step": 600 }, { "epoch": 0.0, - "learning_rate": 4.9884011863929346e-05, - "loss": 3.5026, + "learning_rate": 4.9884200287183294e-05, + "loss": 3.2909, "step": 700 }, { "epoch": 0.0, - "learning_rate": 4.9867442130204974e-05, - "loss": 3.4841, + "learning_rate": 4.986765747106661e-05, + "loss": 3.2913, "step": 800 }, { "epoch": 0.0, - "learning_rate": 4.985087239648059e-05, - "loss": 3.4838, + "learning_rate": 4.9851114654949944e-05, + "loss": 3.2938, "step": 900 }, { "epoch": 0.0, - "learning_rate": 4.983430266275621e-05, - "loss": 3.4808, + "learning_rate": 4.9834571838833275e-05, + "loss": 3.2852, "step": 1000 }, { "epoch": 0.0, - "learning_rate": 4.981773292903183e-05, - "loss": 3.4918, + "learning_rate": 4.98180290227166e-05, + "loss": 3.2939, "step": 1100 }, { "epoch": 0.0, - "learning_rate": 4.980116319530745e-05, - "loss": 3.4975, + "learning_rate": 4.9801486206599924e-05, + "loss": 3.2833, "step": 1200 }, { "epoch": 0.0, - "learning_rate": 4.9784593461583075e-05, - "loss": 3.4854, + "learning_rate": 4.9784943390483255e-05, + "loss": 3.2919, "step": 1300 }, { "epoch": 0.0, - "learning_rate": 4.9768023727858696e-05, - "loss": 3.4789, + "learning_rate": 4.976840057436658e-05, + "loss": 3.2771, "step": 1400 }, { "epoch": 0.0, - "learning_rate": 4.975145399413432e-05, - "loss": 3.4906, + "learning_rate": 4.9751857758249904e-05, + "loss": 3.2949, "step": 1500 }, { "epoch": 0.01, - "learning_rate": 4.973488426040994e-05, - "loss": 3.4578, + "learning_rate": 4.973531494213323e-05, + "loss": 3.2974, "step": 1600 }, { "epoch": 0.01, - "learning_rate": 4.971831452668556e-05, - "loss": 3.4764, + "learning_rate": 4.971877212601656e-05, + "loss": 3.307, "step": 1700 }, { "epoch": 0.01, - "learning_rate": 4.970174479296118e-05, - "loss": 3.4731, + "learning_rate": 4.9702229309899885e-05, + "loss": 3.2939, "step": 1800 }, { "epoch": 0.01, - "learning_rate": 4.96851750592368e-05, - "loss": 3.4722, + "learning_rate": 4.968568649378321e-05, + "loss": 3.2917, "step": 1900 }, { "epoch": 0.01, - "learning_rate": 4.9668605325512425e-05, - "loss": 3.4678, + "learning_rate": 4.966914367766654e-05, + "loss": 3.2832, "step": 2000 }, { "epoch": 0.01, - "learning_rate": 4.9652035591788046e-05, - "loss": 3.469, + "learning_rate": 4.9652600861549865e-05, + "loss": 3.2845, "step": 2100 }, { "epoch": 0.01, - "learning_rate": 4.963546585806366e-05, - "loss": 3.4704, + "learning_rate": 4.963605804543319e-05, + "loss": 3.3021, "step": 2200 }, { "epoch": 0.01, - "learning_rate": 4.961889612433929e-05, - "loss": 3.4727, + "learning_rate": 4.961951522931652e-05, + "loss": 3.2918, "step": 2300 }, { "epoch": 0.01, - "learning_rate": 4.9602326390614904e-05, - "loss": 3.4653, + "learning_rate": 4.9602972413199845e-05, + "loss": 3.2982, "step": 2400 }, { "epoch": 0.01, - "learning_rate": 4.9585756656890526e-05, - "loss": 3.4668, + "learning_rate": 4.958642959708317e-05, + "loss": 3.2903, "step": 2500 }, { "epoch": 0.01, - "learning_rate": 4.956918692316615e-05, - "loss": 3.4759, + "learning_rate": 4.95698867809665e-05, + "loss": 3.2898, "step": 2600 }, { "epoch": 0.01, - "learning_rate": 4.955261718944177e-05, - "loss": 3.4657, + "learning_rate": 4.9553343964849826e-05, + "loss": 3.297, "step": 2700 }, { "epoch": 0.01, - "learning_rate": 4.953604745571739e-05, - "loss": 3.4667, + "learning_rate": 4.953680114873315e-05, + "loss": 3.2864, "step": 2800 }, { "epoch": 0.01, - "learning_rate": 4.9519477721993005e-05, - "loss": 3.4546, + "learning_rate": 4.952025833261648e-05, + "loss": 3.2827, "step": 2900 }, { "epoch": 0.01, - "learning_rate": 4.950290798826863e-05, - "loss": 3.4739, + "learning_rate": 4.9503715516499806e-05, + "loss": 3.2932, "step": 3000 }, { "epoch": 0.01, - "learning_rate": 4.9486338254544254e-05, - "loss": 3.4613, + "learning_rate": 4.948717270038313e-05, + "loss": 3.2809, "step": 3100 }, { "epoch": 0.01, - "learning_rate": 4.946976852081987e-05, - "loss": 3.4704, + "learning_rate": 4.947062988426646e-05, + "loss": 3.2943, "step": 3200 }, { "epoch": 0.01, - "learning_rate": 4.94531987870955e-05, - "loss": 3.4573, + "learning_rate": 4.945408706814979e-05, + "loss": 3.2932, "step": 3300 }, { "epoch": 0.01, - "learning_rate": 4.943662905337111e-05, - "loss": 3.4579, + "learning_rate": 4.943754425203311e-05, + "loss": 3.3063, "step": 3400 }, { "epoch": 0.01, - "learning_rate": 4.9420059319646733e-05, - "loss": 3.458, + "learning_rate": 4.942100143591644e-05, + "loss": 3.2854, "step": 3500 }, { "epoch": 0.01, - "learning_rate": 4.9403489585922355e-05, - "loss": 3.4552, + "learning_rate": 4.9404458619799774e-05, + "loss": 3.2995, "step": 3600 }, { "epoch": 0.01, - "learning_rate": 4.9386919852197976e-05, - "loss": 3.4614, + "learning_rate": 4.938791580368309e-05, + "loss": 3.2967, "step": 3700 }, { "epoch": 0.01, - "learning_rate": 4.93703501184736e-05, - "loss": 3.4539, + "learning_rate": 4.937137298756642e-05, + "loss": 3.3016, "step": 3800 }, { "epoch": 0.01, - "learning_rate": 4.935378038474922e-05, - "loss": 3.4479, + "learning_rate": 4.9354830171449754e-05, + "loss": 3.2916, "step": 3900 }, { "epoch": 0.01, - "learning_rate": 4.933721065102484e-05, - "loss": 3.4557, + "learning_rate": 4.933828735533307e-05, + "loss": 3.2958, "step": 4000 }, { "epoch": 0.01, - "learning_rate": 4.932064091730046e-05, - "loss": 3.4528, + "learning_rate": 4.93217445392164e-05, + "loss": 3.3011, "step": 4100 }, { "epoch": 0.01, - "learning_rate": 4.9304071183576084e-05, - "loss": 3.4577, + "learning_rate": 4.930520172309973e-05, + "loss": 3.2951, "step": 4200 }, { "epoch": 0.01, - "learning_rate": 4.9287501449851705e-05, - "loss": 3.4513, + "learning_rate": 4.928865890698305e-05, + "loss": 3.2985, "step": 4300 }, { "epoch": 0.01, - "learning_rate": 4.927093171612732e-05, - "loss": 3.4553, + "learning_rate": 4.9272116090866384e-05, + "loss": 3.2887, "step": 4400 }, { "epoch": 0.01, - "learning_rate": 4.925436198240295e-05, - "loss": 3.4578, + "learning_rate": 4.925557327474971e-05, + "loss": 3.2952, "step": 4500 }, { "epoch": 0.02, - "learning_rate": 4.923779224867856e-05, - "loss": 3.4565, + "learning_rate": 4.923903045863303e-05, + "loss": 3.2845, "step": 4600 }, { "epoch": 0.02, - "learning_rate": 4.9221222514954184e-05, - "loss": 3.4428, + "learning_rate": 4.9222487642516364e-05, + "loss": 3.2851, "step": 4700 }, { "epoch": 0.02, - "learning_rate": 4.920465278122981e-05, - "loss": 3.4565, + "learning_rate": 4.920594482639969e-05, + "loss": 3.2916, "step": 4800 }, { "epoch": 0.02, - "learning_rate": 4.918808304750543e-05, - "loss": 3.4515, + "learning_rate": 4.918940201028301e-05, + "loss": 3.2851, "step": 4900 }, { "epoch": 0.02, - "learning_rate": 4.917151331378105e-05, - "loss": 3.4574, + "learning_rate": 4.9172859194166344e-05, + "loss": 3.2833, "step": 5000 }, { "epoch": 0.02, - "learning_rate": 4.915494358005667e-05, - "loss": 3.448, + "learning_rate": 4.915631637804967e-05, + "loss": 3.2786, "step": 5100 }, { "epoch": 0.02, - "learning_rate": 4.913837384633229e-05, - "loss": 3.4357, + "learning_rate": 4.9139773561932993e-05, + "loss": 3.2954, "step": 5200 }, { "epoch": 0.02, - "learning_rate": 4.912180411260791e-05, - "loss": 3.4416, + "learning_rate": 4.9123230745816325e-05, + "loss": 3.282, "step": 5300 }, { "epoch": 0.02, - "learning_rate": 4.9105234378883535e-05, - "loss": 3.4606, + "learning_rate": 4.910668792969965e-05, + "loss": 3.2972, "step": 5400 }, { "epoch": 0.02, - "learning_rate": 4.9088664645159156e-05, - "loss": 3.4433, + "learning_rate": 4.909014511358298e-05, + "loss": 3.3043, "step": 5500 }, { "epoch": 0.02, - "learning_rate": 4.907209491143477e-05, - "loss": 3.4473, + "learning_rate": 4.9073602297466305e-05, + "loss": 3.2883, "step": 5600 }, { "epoch": 0.02, - "learning_rate": 4.90555251777104e-05, - "loss": 3.4431, + "learning_rate": 4.905705948134963e-05, + "loss": 3.2923, "step": 5700 }, { "epoch": 0.02, - "learning_rate": 4.903895544398602e-05, - "loss": 3.4304, + "learning_rate": 4.904051666523296e-05, + "loss": 3.2822, "step": 5800 }, { "epoch": 0.02, - "learning_rate": 4.9022385710261635e-05, - "loss": 3.4333, + "learning_rate": 4.9023973849116285e-05, + "loss": 3.2934, "step": 5900 }, { "epoch": 0.02, - "learning_rate": 4.900581597653726e-05, - "loss": 3.4405, + "learning_rate": 4.900743103299961e-05, + "loss": 3.2926, "step": 6000 }, { "epoch": 0.02, - "learning_rate": 4.898924624281288e-05, - "loss": 3.45, + "learning_rate": 4.899088821688294e-05, + "loss": 3.2956, "step": 6100 }, { "epoch": 0.02, - "learning_rate": 4.89726765090885e-05, - "loss": 3.4323, + "learning_rate": 4.8974345400766266e-05, + "loss": 3.2989, "step": 6200 }, { "epoch": 0.02, - "learning_rate": 4.895610677536413e-05, - "loss": 3.4531, + "learning_rate": 4.895780258464959e-05, + "loss": 3.2896, "step": 6300 }, { "epoch": 0.02, - "learning_rate": 4.893953704163974e-05, - "loss": 3.4347, + "learning_rate": 4.894125976853292e-05, + "loss": 3.2996, "step": 6400 }, { "epoch": 0.02, - "learning_rate": 4.8922967307915364e-05, - "loss": 3.4276, + "learning_rate": 4.8924716952416246e-05, + "loss": 3.2976, "step": 6500 }, { "epoch": 0.02, - "learning_rate": 4.8906397574190985e-05, - "loss": 3.4396, + "learning_rate": 4.890817413629957e-05, + "loss": 3.2935, "step": 6600 }, { "epoch": 0.02, - "learning_rate": 4.888982784046661e-05, - "loss": 3.4336, + "learning_rate": 4.88916313201829e-05, + "loss": 3.2942, "step": 6700 }, { "epoch": 0.02, - "learning_rate": 4.887325810674223e-05, - "loss": 3.4227, + "learning_rate": 4.8875088504066227e-05, + "loss": 3.2969, "step": 6800 }, { "epoch": 0.02, - "learning_rate": 4.885668837301785e-05, - "loss": 3.4389, + "learning_rate": 4.885854568794955e-05, + "loss": 3.2932, "step": 6900 }, { "epoch": 0.02, - "learning_rate": 4.884011863929347e-05, - "loss": 3.436, + "learning_rate": 4.884200287183288e-05, + "loss": 3.3085, "step": 7000 }, { "epoch": 0.02, - "learning_rate": 4.8823548905569086e-05, - "loss": 3.4434, + "learning_rate": 4.882546005571621e-05, + "loss": 3.2937, "step": 7100 }, { "epoch": 0.02, - "learning_rate": 4.8806979171844714e-05, - "loss": 3.4485, + "learning_rate": 4.880891723959953e-05, + "loss": 3.295, "step": 7200 }, { "epoch": 0.02, - "learning_rate": 4.8790409438120336e-05, - "loss": 3.4207, + "learning_rate": 4.879237442348286e-05, + "loss": 3.2826, "step": 7300 }, { "epoch": 0.02, - "learning_rate": 4.877383970439595e-05, - "loss": 3.4238, + "learning_rate": 4.877583160736619e-05, + "loss": 3.2794, "step": 7400 }, { "epoch": 0.02, - "learning_rate": 4.875726997067158e-05, - "loss": 3.4369, + "learning_rate": 4.875928879124951e-05, + "loss": 3.2947, "step": 7500 }, { "epoch": 0.03, - "learning_rate": 4.874070023694719e-05, - "loss": 3.4345, + "learning_rate": 4.874274597513284e-05, + "loss": 3.2777, "step": 7600 }, { "epoch": 0.03, - "learning_rate": 4.8724130503222815e-05, - "loss": 3.4325, + "learning_rate": 4.872620315901617e-05, + "loss": 3.2978, "step": 7700 }, { "epoch": 0.03, - "learning_rate": 4.8707560769498436e-05, - "loss": 3.4475, + "learning_rate": 4.870966034289949e-05, + "loss": 3.2885, "step": 7800 }, { "epoch": 0.03, - "learning_rate": 4.869099103577406e-05, - "loss": 3.427, + "learning_rate": 4.8693117526782824e-05, + "loss": 3.2839, "step": 7900 }, { "epoch": 0.03, - "learning_rate": 4.867442130204968e-05, - "loss": 3.4285, + "learning_rate": 4.867657471066615e-05, + "loss": 3.2922, "step": 8000 }, { "epoch": 0.03, - "learning_rate": 4.8657851568325294e-05, - "loss": 3.448, + "learning_rate": 4.866003189454947e-05, + "loss": 3.2874, "step": 8100 }, { "epoch": 0.03, - "learning_rate": 4.864128183460092e-05, - "loss": 3.432, + "learning_rate": 4.8643489078432804e-05, + "loss": 3.2901, "step": 8200 }, { "epoch": 0.03, - "learning_rate": 4.8624712100876544e-05, - "loss": 3.4294, + "learning_rate": 4.862694626231613e-05, + "loss": 3.2914, "step": 8300 }, { "epoch": 0.03, - "learning_rate": 4.860814236715216e-05, - "loss": 3.4335, + "learning_rate": 4.861040344619945e-05, + "loss": 3.291, "step": 8400 }, { "epoch": 0.03, - "learning_rate": 4.8591572633427786e-05, - "loss": 3.4329, + "learning_rate": 4.8593860630082784e-05, + "loss": 3.2921, "step": 8500 }, { "epoch": 0.03, - "learning_rate": 4.85750028997034e-05, - "loss": 3.4347, + "learning_rate": 4.857731781396611e-05, + "loss": 3.3034, "step": 8600 }, { "epoch": 0.03, - "learning_rate": 4.855843316597902e-05, - "loss": 3.4302, + "learning_rate": 4.8560774997849433e-05, + "loss": 3.2853, "step": 8700 }, { "epoch": 0.03, - "learning_rate": 4.8541863432254644e-05, - "loss": 3.4159, + "learning_rate": 4.8544232181732765e-05, + "loss": 3.2932, "step": 8800 }, { "epoch": 0.03, - "learning_rate": 4.8525293698530266e-05, - "loss": 3.4263, + "learning_rate": 4.852768936561609e-05, + "loss": 3.2956, "step": 8900 }, { "epoch": 0.03, - "learning_rate": 4.850872396480589e-05, - "loss": 3.4375, + "learning_rate": 4.8511146549499414e-05, + "loss": 3.2974, "step": 9000 }, { "epoch": 0.03, - "learning_rate": 4.849215423108151e-05, - "loss": 3.4262, + "learning_rate": 4.8494603733382745e-05, + "loss": 3.3028, "step": 9100 }, { "epoch": 0.03, - "learning_rate": 4.847558449735713e-05, - "loss": 3.4365, + "learning_rate": 4.847806091726607e-05, + "loss": 3.2898, "step": 9200 }, { "epoch": 0.03, - "learning_rate": 4.845901476363275e-05, - "loss": 3.4249, + "learning_rate": 4.8461518101149394e-05, + "loss": 3.2808, "step": 9300 }, { "epoch": 0.03, - "learning_rate": 4.844244502990837e-05, - "loss": 3.4318, + "learning_rate": 4.8444975285032725e-05, + "loss": 3.2858, "step": 9400 }, { "epoch": 0.03, - "learning_rate": 4.8425875296183994e-05, - "loss": 3.4298, + "learning_rate": 4.842843246891605e-05, + "loss": 3.2769, "step": 9500 }, { "epoch": 0.03, - "learning_rate": 4.840930556245961e-05, - "loss": 3.418, + "learning_rate": 4.841188965279938e-05, + "loss": 3.2756, "step": 9600 }, { "epoch": 0.03, - "learning_rate": 4.839273582873524e-05, - "loss": 3.4355, + "learning_rate": 4.8395346836682706e-05, + "loss": 3.2888, "step": 9700 }, { "epoch": 0.03, - "learning_rate": 4.837616609501085e-05, - "loss": 3.4177, + "learning_rate": 4.837880402056603e-05, + "loss": 3.2892, "step": 9800 }, { "epoch": 0.03, - "learning_rate": 4.8359596361286473e-05, - "loss": 3.4327, + "learning_rate": 4.836226120444936e-05, + "loss": 3.2951, "step": 9900 }, { "epoch": 0.03, - "learning_rate": 4.83430266275621e-05, - "loss": 3.4235, + "learning_rate": 4.8345718388332686e-05, + "loss": 3.2864, "step": 10000 }, { "epoch": 0.03, - "learning_rate": 4.8326456893837716e-05, - "loss": 3.4211, + "learning_rate": 4.832917557221601e-05, + "loss": 3.3114, "step": 10100 }, { "epoch": 0.03, - "learning_rate": 4.830988716011334e-05, - "loss": 3.4231, + "learning_rate": 4.831263275609934e-05, + "loss": 3.2875, "step": 10200 }, { "epoch": 0.03, - "learning_rate": 4.829331742638896e-05, - "loss": 3.4216, + "learning_rate": 4.829608993998267e-05, + "loss": 3.2978, "step": 10300 }, { "epoch": 0.03, - "learning_rate": 4.827674769266458e-05, - "loss": 3.424, + "learning_rate": 4.827954712386599e-05, + "loss": 3.295, "step": 10400 }, { "epoch": 0.03, - "learning_rate": 4.82601779589402e-05, - "loss": 3.4257, + "learning_rate": 4.826300430774932e-05, + "loss": 3.301, "step": 10500 }, { "epoch": 0.04, - "learning_rate": 4.8243608225215824e-05, - "loss": 3.4235, + "learning_rate": 4.824646149163265e-05, + "loss": 3.2969, "step": 10600 }, { "epoch": 0.04, - "learning_rate": 4.8227038491491445e-05, - "loss": 3.4238, + "learning_rate": 4.822991867551597e-05, + "loss": 3.2677, "step": 10700 }, { "epoch": 0.04, - "learning_rate": 4.821046875776707e-05, - "loss": 3.4137, + "learning_rate": 4.82133758593993e-05, + "loss": 3.2916, "step": 10800 }, { "epoch": 0.04, - "learning_rate": 4.819389902404269e-05, - "loss": 3.4133, + "learning_rate": 4.819683304328263e-05, + "loss": 3.2874, "step": 10900 }, { "epoch": 0.04, - "learning_rate": 4.817732929031831e-05, - "loss": 3.4209, + "learning_rate": 4.818029022716595e-05, + "loss": 3.2896, "step": 11000 }, { "epoch": 0.04, - "learning_rate": 4.8160759556593924e-05, - "loss": 3.4163, + "learning_rate": 4.816374741104928e-05, + "loss": 3.2944, "step": 11100 }, { "epoch": 0.04, - "learning_rate": 4.814418982286955e-05, - "loss": 3.415, + "learning_rate": 4.814720459493261e-05, + "loss": 3.2915, "step": 11200 }, { "epoch": 0.04, - "learning_rate": 4.812762008914517e-05, - "loss": 3.4154, + "learning_rate": 4.813066177881593e-05, + "loss": 3.3023, "step": 11300 }, { "epoch": 0.04, - "learning_rate": 4.811105035542079e-05, - "loss": 3.4228, + "learning_rate": 4.8114118962699264e-05, + "loss": 3.2946, "step": 11400 }, { "epoch": 0.04, - "learning_rate": 4.809448062169642e-05, - "loss": 3.4223, + "learning_rate": 4.809757614658258e-05, + "loss": 3.2983, "step": 11500 }, { "epoch": 0.04, - "learning_rate": 4.807791088797203e-05, - "loss": 3.4198, + "learning_rate": 4.808103333046591e-05, + "loss": 3.3081, "step": 11600 }, { "epoch": 0.04, - "learning_rate": 4.806134115424765e-05, - "loss": 3.425, + "learning_rate": 4.8064490514349244e-05, + "loss": 3.2932, "step": 11700 }, { "epoch": 0.04, - "learning_rate": 4.8044771420523275e-05, - "loss": 3.4335, + "learning_rate": 4.804794769823257e-05, + "loss": 3.2931, "step": 11800 }, { "epoch": 0.04, - "learning_rate": 4.8028201686798896e-05, - "loss": 3.4286, + "learning_rate": 4.803140488211589e-05, + "loss": 3.3013, "step": 11900 }, { "epoch": 0.04, - "learning_rate": 4.801163195307452e-05, - "loss": 3.4065, + "learning_rate": 4.8014862065999224e-05, + "loss": 3.2959, "step": 12000 }, { "epoch": 0.04, - "learning_rate": 4.799506221935014e-05, - "loss": 3.4247, + "learning_rate": 4.799831924988255e-05, + "loss": 3.2952, "step": 12100 }, { "epoch": 0.04, - "learning_rate": 4.797849248562576e-05, - "loss": 3.4133, + "learning_rate": 4.7981776433765873e-05, + "loss": 3.2949, "step": 12200 }, { "epoch": 0.04, - "learning_rate": 4.7961922751901375e-05, - "loss": 3.4108, + "learning_rate": 4.7965233617649205e-05, + "loss": 3.2953, "step": 12300 }, { "epoch": 0.04, - "learning_rate": 4.7945353018177e-05, - "loss": 3.4284, + "learning_rate": 4.794869080153253e-05, + "loss": 3.289, "step": 12400 }, { "epoch": 0.04, - "learning_rate": 4.7928783284452625e-05, - "loss": 3.4118, + "learning_rate": 4.7932147985415854e-05, + "loss": 3.2867, "step": 12500 }, { "epoch": 0.04, - "learning_rate": 4.791221355072824e-05, - "loss": 3.4014, + "learning_rate": 4.7915605169299185e-05, + "loss": 3.2901, "step": 12600 }, { "epoch": 0.04, - "learning_rate": 4.789564381700387e-05, - "loss": 3.4246, + "learning_rate": 4.789906235318251e-05, + "loss": 3.2829, "step": 12700 }, { "epoch": 0.04, - "learning_rate": 4.787907408327948e-05, - "loss": 3.415, + "learning_rate": 4.7882519537065834e-05, + "loss": 3.2903, "step": 12800 }, { "epoch": 0.04, - "learning_rate": 4.7862504349555104e-05, - "loss": 3.4196, + "learning_rate": 4.7865976720949166e-05, + "loss": 3.2918, "step": 12900 }, { "epoch": 0.04, - "learning_rate": 4.7845934615830725e-05, - "loss": 3.402, + "learning_rate": 4.784943390483249e-05, + "loss": 3.284, "step": 13000 }, { "epoch": 0.04, - "learning_rate": 4.782936488210635e-05, - "loss": 3.4146, + "learning_rate": 4.7832891088715815e-05, + "loss": 3.2858, "step": 13100 }, { "epoch": 0.04, - "learning_rate": 4.781279514838197e-05, - "loss": 3.4131, + "learning_rate": 4.7816348272599146e-05, + "loss": 3.2846, "step": 13200 }, { "epoch": 0.04, - "learning_rate": 4.779622541465758e-05, - "loss": 3.4164, + "learning_rate": 4.779980545648247e-05, + "loss": 3.2948, "step": 13300 }, { "epoch": 0.04, - "learning_rate": 4.777965568093321e-05, - "loss": 3.4111, + "learning_rate": 4.7783262640365795e-05, + "loss": 3.2962, "step": 13400 }, { "epoch": 0.04, - "learning_rate": 4.776308594720883e-05, - "loss": 3.4213, + "learning_rate": 4.7766719824249126e-05, + "loss": 3.3008, "step": 13500 }, { - "epoch": 0.05, - "learning_rate": 4.774651621348445e-05, - "loss": 3.4171, + "epoch": 0.04, + "learning_rate": 4.775017700813245e-05, + "loss": 3.2893, "step": 13600 }, { "epoch": 0.05, - "learning_rate": 4.7729946479760076e-05, - "loss": 3.4094, + "learning_rate": 4.7733634192015775e-05, + "loss": 3.2871, "step": 13700 }, { "epoch": 0.05, - "learning_rate": 4.771337674603569e-05, - "loss": 3.4074, + "learning_rate": 4.771709137589911e-05, + "loss": 3.2983, "step": 13800 }, { "epoch": 0.05, - "learning_rate": 4.769680701231131e-05, - "loss": 3.4065, + "learning_rate": 4.770054855978243e-05, + "loss": 3.2774, "step": 13900 }, { "epoch": 0.05, - "learning_rate": 4.768023727858694e-05, - "loss": 3.4266, + "learning_rate": 4.768400574366576e-05, + "loss": 3.2899, "step": 14000 }, { "epoch": 0.05, - "learning_rate": 4.7663667544862555e-05, - "loss": 3.4083, + "learning_rate": 4.766746292754908e-05, + "loss": 3.2957, "step": 14100 }, { "epoch": 0.05, - "learning_rate": 4.7647097811138176e-05, - "loss": 3.4123, + "learning_rate": 4.765092011143241e-05, + "loss": 3.2904, "step": 14200 }, { "epoch": 0.05, - "learning_rate": 4.76305280774138e-05, - "loss": 3.4197, + "learning_rate": 4.763437729531574e-05, + "loss": 3.286, "step": 14300 }, { "epoch": 0.05, - "learning_rate": 4.761395834368942e-05, - "loss": 3.4043, + "learning_rate": 4.761783447919906e-05, + "loss": 3.284, "step": 14400 }, { "epoch": 0.05, - "learning_rate": 4.759738860996504e-05, - "loss": 3.4132, + "learning_rate": 4.760129166308239e-05, + "loss": 3.2834, "step": 14500 }, { "epoch": 0.05, - "learning_rate": 4.758081887624066e-05, - "loss": 3.407, + "learning_rate": 4.758474884696572e-05, + "loss": 3.2954, "step": 14600 }, { "epoch": 0.05, - "learning_rate": 4.7564249142516284e-05, - "loss": 3.4026, + "learning_rate": 4.756820603084904e-05, + "loss": 3.286, "step": 14700 }, { "epoch": 0.05, - "learning_rate": 4.75476794087919e-05, - "loss": 3.4, + "learning_rate": 4.755166321473237e-05, + "loss": 3.2783, "step": 14800 }, { "epoch": 0.05, - "learning_rate": 4.7531109675067526e-05, - "loss": 3.4033, + "learning_rate": 4.7535120398615704e-05, + "loss": 3.2939, "step": 14900 }, { "epoch": 0.05, - "learning_rate": 4.751453994134315e-05, - "loss": 3.4167, + "learning_rate": 4.751857758249902e-05, + "loss": 3.301, "step": 15000 }, { "epoch": 0.05, - "learning_rate": 4.749797020761876e-05, - "loss": 3.4197, + "learning_rate": 4.750203476638235e-05, + "loss": 3.3021, "step": 15100 }, { "epoch": 0.05, - "learning_rate": 4.748140047389439e-05, - "loss": 3.4082, + "learning_rate": 4.7485491950265684e-05, + "loss": 3.2916, "step": 15200 }, { "epoch": 0.05, - "learning_rate": 4.7464830740170006e-05, - "loss": 3.4074, + "learning_rate": 4.7468949134149e-05, + "loss": 3.2896, "step": 15300 }, { "epoch": 0.05, - "learning_rate": 4.744826100644563e-05, - "loss": 3.4169, + "learning_rate": 4.745240631803233e-05, + "loss": 3.295, "step": 15400 }, { "epoch": 0.05, - "learning_rate": 4.743169127272125e-05, - "loss": 3.4184, + "learning_rate": 4.7435863501915664e-05, + "loss": 3.2956, "step": 15500 }, { "epoch": 0.05, - "learning_rate": 4.741512153899687e-05, - "loss": 3.4086, + "learning_rate": 4.741932068579898e-05, + "loss": 3.292, "step": 15600 }, { "epoch": 0.05, - "learning_rate": 4.739855180527249e-05, - "loss": 3.4219, + "learning_rate": 4.7402777869682313e-05, + "loss": 3.289, "step": 15700 }, { "epoch": 0.05, - "learning_rate": 4.738198207154811e-05, - "loss": 3.4063, + "learning_rate": 4.7386235053565645e-05, + "loss": 3.284, "step": 15800 }, { "epoch": 0.05, - "learning_rate": 4.7365412337823734e-05, - "loss": 3.4125, + "learning_rate": 4.736969223744897e-05, + "loss": 3.291, "step": 15900 }, { "epoch": 0.05, - "learning_rate": 4.7348842604099356e-05, - "loss": 3.4047, + "learning_rate": 4.7353149421332294e-05, + "loss": 3.2793, "step": 16000 }, { "epoch": 0.05, - "learning_rate": 4.733227287037498e-05, - "loss": 3.3953, + "learning_rate": 4.7336606605215625e-05, + "loss": 3.2924, "step": 16100 }, { "epoch": 0.05, - "learning_rate": 4.73157031366506e-05, - "loss": 3.4067, + "learning_rate": 4.732006378909895e-05, + "loss": 3.2953, "step": 16200 }, { "epoch": 0.05, - "learning_rate": 4.7299133402926213e-05, - "loss": 3.3954, + "learning_rate": 4.7303520972982274e-05, + "loss": 3.2964, "step": 16300 }, { "epoch": 0.05, - "learning_rate": 4.728256366920184e-05, - "loss": 3.4103, + "learning_rate": 4.7286978156865606e-05, + "loss": 3.2916, "step": 16400 }, { "epoch": 0.05, - "learning_rate": 4.7265993935477456e-05, - "loss": 3.4191, + "learning_rate": 4.727043534074893e-05, + "loss": 3.2854, "step": 16500 }, { - "epoch": 0.06, - "learning_rate": 4.724942420175308e-05, - "loss": 3.4117, + "epoch": 0.05, + "learning_rate": 4.7253892524632255e-05, + "loss": 3.2949, "step": 16600 }, { "epoch": 0.06, - "learning_rate": 4.7232854468028706e-05, - "loss": 3.4006, + "learning_rate": 4.723734970851558e-05, + "loss": 3.289, "step": 16700 }, { "epoch": 0.06, - "learning_rate": 4.721628473430432e-05, - "loss": 3.3984, + "learning_rate": 4.722080689239891e-05, + "loss": 3.2721, "step": 16800 }, { "epoch": 0.06, - "learning_rate": 4.719971500057994e-05, - "loss": 3.3922, + "learning_rate": 4.7204264076282235e-05, + "loss": 3.2848, "step": 16900 }, { "epoch": 0.06, - "learning_rate": 4.7183145266855564e-05, - "loss": 3.4097, + "learning_rate": 4.718772126016556e-05, + "loss": 3.2939, "step": 17000 }, { "epoch": 0.06, - "learning_rate": 4.7166575533131185e-05, - "loss": 3.392, + "learning_rate": 4.717117844404889e-05, + "loss": 3.289, "step": 17100 }, { "epoch": 0.06, - "learning_rate": 4.715000579940681e-05, - "loss": 3.4108, + "learning_rate": 4.7154635627932215e-05, + "loss": 3.2877, "step": 17200 }, { "epoch": 0.06, - "learning_rate": 4.713343606568243e-05, - "loss": 3.4029, + "learning_rate": 4.713809281181554e-05, + "loss": 3.2747, "step": 17300 }, { "epoch": 0.06, - "learning_rate": 4.711686633195805e-05, - "loss": 3.4124, + "learning_rate": 4.712154999569887e-05, + "loss": 3.2915, "step": 17400 }, { "epoch": 0.06, - "learning_rate": 4.7100296598233664e-05, - "loss": 3.4116, + "learning_rate": 4.7105007179582196e-05, + "loss": 3.2763, "step": 17500 }, { "epoch": 0.06, - "learning_rate": 4.708372686450929e-05, - "loss": 3.3832, + "learning_rate": 4.708846436346552e-05, + "loss": 3.2881, "step": 17600 }, { "epoch": 0.06, - "learning_rate": 4.7067157130784914e-05, - "loss": 3.4012, + "learning_rate": 4.707192154734885e-05, + "loss": 3.2975, "step": 17700 }, { "epoch": 0.06, - "learning_rate": 4.705058739706053e-05, - "loss": 3.4056, + "learning_rate": 4.7055378731232176e-05, + "loss": 3.294, "step": 17800 }, { "epoch": 0.06, - "learning_rate": 4.703401766333616e-05, - "loss": 3.3942, + "learning_rate": 4.70388359151155e-05, + "loss": 3.2951, "step": 17900 }, { "epoch": 0.06, - "learning_rate": 4.701744792961177e-05, - "loss": 3.3928, + "learning_rate": 4.702229309899883e-05, + "loss": 3.2806, "step": 18000 }, { "epoch": 0.06, - "learning_rate": 4.700087819588739e-05, - "loss": 3.4021, + "learning_rate": 4.700575028288216e-05, + "loss": 3.2991, "step": 18100 }, { "epoch": 0.06, - "learning_rate": 4.6984308462163015e-05, - "loss": 3.4048, + "learning_rate": 4.698920746676548e-05, + "loss": 3.2838, "step": 18200 }, { "epoch": 0.06, - "learning_rate": 4.6967738728438636e-05, - "loss": 3.4163, + "learning_rate": 4.697266465064881e-05, + "loss": 3.2859, "step": 18300 }, { "epoch": 0.06, - "learning_rate": 4.695116899471426e-05, - "loss": 3.4177, + "learning_rate": 4.6956121834532144e-05, + "loss": 3.2875, "step": 18400 }, { "epoch": 0.06, - "learning_rate": 4.693459926098987e-05, - "loss": 3.396, + "learning_rate": 4.693957901841546e-05, + "loss": 3.2801, "step": 18500 }, { "epoch": 0.06, - "learning_rate": 4.69180295272655e-05, - "loss": 3.4118, + "learning_rate": 4.692303620229879e-05, + "loss": 3.2839, "step": 18600 }, { "epoch": 0.06, - "learning_rate": 4.690145979354112e-05, - "loss": 3.4093, + "learning_rate": 4.6906493386182124e-05, + "loss": 3.2893, "step": 18700 }, { "epoch": 0.06, - "learning_rate": 4.6884890059816737e-05, - "loss": 3.3996, + "learning_rate": 4.688995057006544e-05, + "loss": 3.2755, "step": 18800 }, { "epoch": 0.06, - "learning_rate": 4.6868320326092365e-05, - "loss": 3.3824, + "learning_rate": 4.687340775394877e-05, + "loss": 3.2954, "step": 18900 }, { "epoch": 0.06, - "learning_rate": 4.685175059236798e-05, - "loss": 3.407, + "learning_rate": 4.6856864937832104e-05, + "loss": 3.2875, "step": 19000 }, { "epoch": 0.06, - "learning_rate": 4.68351808586436e-05, - "loss": 3.3897, + "learning_rate": 4.684032212171542e-05, + "loss": 3.2911, "step": 19100 }, { "epoch": 0.06, - "learning_rate": 4.681861112491923e-05, - "loss": 3.4016, + "learning_rate": 4.6823779305598753e-05, + "loss": 3.287, "step": 19200 }, { "epoch": 0.06, - "learning_rate": 4.6802041391194844e-05, - "loss": 3.4079, + "learning_rate": 4.680723648948208e-05, + "loss": 3.2809, "step": 19300 }, { "epoch": 0.06, - "learning_rate": 4.6785471657470465e-05, - "loss": 3.381, + "learning_rate": 4.67906936733654e-05, + "loss": 3.2948, "step": 19400 }, { "epoch": 0.06, - "learning_rate": 4.676890192374609e-05, - "loss": 3.3918, + "learning_rate": 4.6774150857248734e-05, + "loss": 3.2868, "step": 19500 }, { "epoch": 0.06, - "learning_rate": 4.675233219002171e-05, - "loss": 3.401, + "learning_rate": 4.675760804113206e-05, + "loss": 3.2931, "step": 19600 }, { "epoch": 0.07, - "learning_rate": 4.673576245629733e-05, - "loss": 3.4038, + "learning_rate": 4.674106522501538e-05, + "loss": 3.2881, "step": 19700 }, { "epoch": 0.07, - "learning_rate": 4.671919272257295e-05, - "loss": 3.3924, + "learning_rate": 4.6724522408898714e-05, + "loss": 3.2942, "step": 19800 }, { "epoch": 0.07, - "learning_rate": 4.670262298884857e-05, - "loss": 3.3818, + "learning_rate": 4.670797959278204e-05, + "loss": 3.2834, "step": 19900 }, { "epoch": 0.07, - "learning_rate": 4.668605325512419e-05, - "loss": 3.3995, + "learning_rate": 4.669143677666536e-05, + "loss": 3.2865, "step": 20000 }, { "epoch": 0.07, - "learning_rate": 4.6669483521399816e-05, - "loss": 3.3896, + "learning_rate": 4.6674893960548695e-05, + "loss": 3.2815, "step": 20100 }, { "epoch": 0.07, - "learning_rate": 4.665291378767544e-05, - "loss": 3.3971, + "learning_rate": 4.665835114443202e-05, + "loss": 3.3033, "step": 20200 }, { "epoch": 0.07, - "learning_rate": 4.663634405395105e-05, - "loss": 3.3955, + "learning_rate": 4.664180832831535e-05, + "loss": 3.2943, "step": 20300 }, { "epoch": 0.07, - "learning_rate": 4.661977432022668e-05, - "loss": 3.3989, + "learning_rate": 4.6625265512198675e-05, + "loss": 3.2785, "step": 20400 }, { "epoch": 0.07, - "learning_rate": 4.6603204586502295e-05, - "loss": 3.4087, + "learning_rate": 4.6608722696082e-05, + "loss": 3.2879, "step": 20500 }, { "epoch": 0.07, - "learning_rate": 4.6586634852777916e-05, - "loss": 3.3976, + "learning_rate": 4.659217987996533e-05, + "loss": 3.2918, "step": 20600 }, { "epoch": 0.07, - "learning_rate": 4.657006511905354e-05, - "loss": 3.3939, + "learning_rate": 4.6575637063848655e-05, + "loss": 3.2842, "step": 20700 }, { "epoch": 0.07, - "learning_rate": 4.655349538532916e-05, - "loss": 3.3986, + "learning_rate": 4.655909424773198e-05, + "loss": 3.2816, "step": 20800 }, { "epoch": 0.07, - "learning_rate": 4.653692565160478e-05, - "loss": 3.3876, + "learning_rate": 4.654255143161531e-05, + "loss": 3.2917, "step": 20900 }, { "epoch": 0.07, - "learning_rate": 4.65203559178804e-05, - "loss": 3.3986, + "learning_rate": 4.6526008615498636e-05, + "loss": 3.2821, "step": 21000 }, { "epoch": 0.07, - "learning_rate": 4.6503786184156024e-05, - "loss": 3.3863, + "learning_rate": 4.650946579938196e-05, + "loss": 3.2816, "step": 21100 }, { "epoch": 0.07, - "learning_rate": 4.6487216450431645e-05, - "loss": 3.3764, + "learning_rate": 4.649292298326529e-05, + "loss": 3.276, "step": 21200 }, { "epoch": 0.07, - "learning_rate": 4.6470646716707266e-05, - "loss": 3.4007, + "learning_rate": 4.6476380167148616e-05, + "loss": 3.2749, "step": 21300 }, { "epoch": 0.07, - "learning_rate": 4.645407698298289e-05, - "loss": 3.3963, + "learning_rate": 4.645983735103194e-05, + "loss": 3.3036, "step": 21400 }, { "epoch": 0.07, - "learning_rate": 4.64375072492585e-05, - "loss": 3.3888, + "learning_rate": 4.644329453491527e-05, + "loss": 3.2723, "step": 21500 }, { "epoch": 0.07, - "learning_rate": 4.642093751553413e-05, - "loss": 3.3853, + "learning_rate": 4.6426751718798597e-05, + "loss": 3.2959, "step": 21600 }, { "epoch": 0.07, - "learning_rate": 4.6404367781809746e-05, - "loss": 3.391, + "learning_rate": 4.641020890268192e-05, + "loss": 3.2839, "step": 21700 }, { "epoch": 0.07, - "learning_rate": 4.638779804808537e-05, - "loss": 3.389, + "learning_rate": 4.639366608656525e-05, + "loss": 3.2925, "step": 21800 }, { "epoch": 0.07, - "learning_rate": 4.6371228314360995e-05, - "loss": 3.3916, + "learning_rate": 4.637712327044858e-05, + "loss": 3.3013, "step": 21900 }, { "epoch": 0.07, - "learning_rate": 4.635465858063661e-05, - "loss": 3.3929, + "learning_rate": 4.63605804543319e-05, + "loss": 3.2799, "step": 22000 }, { "epoch": 0.07, - "learning_rate": 4.633808884691223e-05, - "loss": 3.399, + "learning_rate": 4.634403763821523e-05, + "loss": 3.288, "step": 22100 }, { "epoch": 0.07, - "learning_rate": 4.632151911318785e-05, - "loss": 3.3957, + "learning_rate": 4.632749482209856e-05, + "loss": 3.3072, "step": 22200 }, { "epoch": 0.07, - "learning_rate": 4.6304949379463474e-05, - "loss": 3.3903, + "learning_rate": 4.631095200598188e-05, + "loss": 3.2815, "step": 22300 }, { "epoch": 0.07, - "learning_rate": 4.6288379645739096e-05, - "loss": 3.3814, + "learning_rate": 4.629440918986521e-05, + "loss": 3.2871, "step": 22400 }, { "epoch": 0.07, - "learning_rate": 4.627180991201472e-05, - "loss": 3.3954, + "learning_rate": 4.627786637374854e-05, + "loss": 3.2812, "step": 22500 }, { "epoch": 0.07, - "learning_rate": 4.625524017829034e-05, - "loss": 3.3799, + "learning_rate": 4.626132355763186e-05, + "loss": 3.2976, "step": 22600 }, { "epoch": 0.08, - "learning_rate": 4.623867044456596e-05, - "loss": 3.3879, + "learning_rate": 4.6244780741515194e-05, + "loss": 3.2819, "step": 22700 }, { "epoch": 0.08, - "learning_rate": 4.622210071084158e-05, - "loss": 3.3831, + "learning_rate": 4.622823792539852e-05, + "loss": 3.2883, "step": 22800 }, { "epoch": 0.08, - "learning_rate": 4.62055309771172e-05, - "loss": 3.38, + "learning_rate": 4.621169510928184e-05, + "loss": 3.2844, "step": 22900 }, { "epoch": 0.08, - "learning_rate": 4.618896124339282e-05, - "loss": 3.3887, + "learning_rate": 4.6195152293165174e-05, + "loss": 3.2909, "step": 23000 }, { "epoch": 0.08, - "learning_rate": 4.6172391509668446e-05, - "loss": 3.3955, + "learning_rate": 4.61786094770485e-05, + "loss": 3.2894, "step": 23100 }, { "epoch": 0.08, - "learning_rate": 4.615582177594406e-05, - "loss": 3.3872, + "learning_rate": 4.616206666093182e-05, + "loss": 3.2877, "step": 23200 }, { "epoch": 0.08, - "learning_rate": 4.613925204221968e-05, - "loss": 3.3873, + "learning_rate": 4.6145523844815154e-05, + "loss": 3.2877, "step": 23300 }, { "epoch": 0.08, - "learning_rate": 4.6122682308495304e-05, - "loss": 3.3933, + "learning_rate": 4.612898102869848e-05, + "loss": 3.2899, "step": 23400 }, { "epoch": 0.08, - "learning_rate": 4.6106112574770925e-05, - "loss": 3.3814, + "learning_rate": 4.61124382125818e-05, + "loss": 3.2926, "step": 23500 }, { "epoch": 0.08, - "learning_rate": 4.608954284104655e-05, - "loss": 3.3858, + "learning_rate": 4.6095895396465135e-05, + "loss": 3.2819, "step": 23600 }, { "epoch": 0.08, - "learning_rate": 4.607297310732217e-05, - "loss": 3.38, + "learning_rate": 4.607935258034846e-05, + "loss": 3.2827, "step": 23700 }, { "epoch": 0.08, - "learning_rate": 4.605640337359779e-05, - "loss": 3.3847, + "learning_rate": 4.6062809764231784e-05, + "loss": 3.2924, "step": 23800 }, { "epoch": 0.08, - "learning_rate": 4.603983363987341e-05, - "loss": 3.3922, + "learning_rate": 4.6046266948115115e-05, + "loss": 3.287, "step": 23900 }, { "epoch": 0.08, - "learning_rate": 4.6023263906149026e-05, - "loss": 3.3749, + "learning_rate": 4.602972413199844e-05, + "loss": 3.2746, "step": 24000 }, { "epoch": 0.08, - "learning_rate": 4.6006694172424654e-05, - "loss": 3.3809, + "learning_rate": 4.6013181315881764e-05, + "loss": 3.2971, "step": 24100 }, { "epoch": 0.08, - "learning_rate": 4.599012443870027e-05, - "loss": 3.4017, + "learning_rate": 4.5996638499765095e-05, + "loss": 3.2789, "step": 24200 }, { "epoch": 0.08, - "learning_rate": 4.597355470497589e-05, - "loss": 3.3997, + "learning_rate": 4.598009568364842e-05, + "loss": 3.2857, "step": 24300 }, { "epoch": 0.08, - "learning_rate": 4.595698497125152e-05, - "loss": 3.3871, + "learning_rate": 4.596355286753175e-05, + "loss": 3.2808, "step": 24400 }, { "epoch": 0.08, - "learning_rate": 4.594041523752713e-05, - "loss": 3.3821, + "learning_rate": 4.5947010051415076e-05, + "loss": 3.2932, "step": 24500 }, { "epoch": 0.08, - "learning_rate": 4.5923845503802755e-05, - "loss": 3.3837, + "learning_rate": 4.59304672352984e-05, + "loss": 3.2946, "step": 24600 }, { "epoch": 0.08, - "learning_rate": 4.5907275770078376e-05, - "loss": 3.3859, + "learning_rate": 4.591392441918173e-05, + "loss": 3.2957, "step": 24700 }, { "epoch": 0.08, - "learning_rate": 4.5890706036354e-05, - "loss": 3.3932, + "learning_rate": 4.5897381603065056e-05, + "loss": 3.2578, "step": 24800 }, { "epoch": 0.08, - "learning_rate": 4.587413630262962e-05, - "loss": 3.3755, + "learning_rate": 4.588083878694838e-05, + "loss": 3.3015, "step": 24900 }, { "epoch": 0.08, - "learning_rate": 4.585756656890524e-05, - "loss": 3.3716, + "learning_rate": 4.586429597083171e-05, + "loss": 3.292, "step": 25000 }, { "epoch": 0.08, - "learning_rate": 4.584099683518086e-05, - "loss": 3.3922, + "learning_rate": 4.5847753154715037e-05, + "loss": 3.2771, "step": 25100 }, { "epoch": 0.08, - "learning_rate": 4.5824427101456477e-05, - "loss": 3.3916, + "learning_rate": 4.583121033859836e-05, + "loss": 3.2868, "step": 25200 }, { "epoch": 0.08, - "learning_rate": 4.5807857367732105e-05, - "loss": 3.3878, + "learning_rate": 4.581466752248169e-05, + "loss": 3.2923, "step": 25300 }, { "epoch": 0.08, - "learning_rate": 4.5791287634007726e-05, - "loss": 3.3905, + "learning_rate": 4.579812470636502e-05, + "loss": 3.2885, "step": 25400 }, { "epoch": 0.08, - "learning_rate": 4.577471790028334e-05, - "loss": 3.3902, + "learning_rate": 4.578158189024834e-05, + "loss": 3.2929, "step": 25500 }, { "epoch": 0.08, - "learning_rate": 4.575814816655897e-05, - "loss": 3.3897, + "learning_rate": 4.576503907413167e-05, + "loss": 3.2829, "step": 25600 }, { "epoch": 0.09, - "learning_rate": 4.5741578432834584e-05, - "loss": 3.3895, + "learning_rate": 4.5748496258015e-05, + "loss": 3.2922, "step": 25700 }, { "epoch": 0.09, - "learning_rate": 4.5725008699110205e-05, - "loss": 3.3796, + "learning_rate": 4.573195344189832e-05, + "loss": 3.2905, "step": 25800 }, { "epoch": 0.09, - "learning_rate": 4.5708438965385834e-05, - "loss": 3.3945, + "learning_rate": 4.571541062578165e-05, + "loss": 3.2845, "step": 25900 }, { "epoch": 0.09, - "learning_rate": 4.569186923166145e-05, - "loss": 3.3824, + "learning_rate": 4.569886780966498e-05, + "loss": 3.297, "step": 26000 }, { "epoch": 0.09, - "learning_rate": 4.567529949793707e-05, - "loss": 3.3792, + "learning_rate": 4.56823249935483e-05, + "loss": 3.281, "step": 26100 }, { "epoch": 0.09, - "learning_rate": 4.565872976421269e-05, - "loss": 3.3954, + "learning_rate": 4.5665782177431634e-05, + "loss": 3.2909, "step": 26200 }, { "epoch": 0.09, - "learning_rate": 4.564216003048831e-05, - "loss": 3.3757, + "learning_rate": 4.564923936131496e-05, + "loss": 3.2849, "step": 26300 }, { "epoch": 0.09, - "learning_rate": 4.5625590296763934e-05, - "loss": 3.3941, + "learning_rate": 4.563269654519828e-05, + "loss": 3.2719, "step": 26400 }, { "epoch": 0.09, - "learning_rate": 4.5609020563039556e-05, - "loss": 3.3836, + "learning_rate": 4.5616153729081614e-05, + "loss": 3.2892, "step": 26500 }, { "epoch": 0.09, - "learning_rate": 4.559245082931518e-05, - "loss": 3.3842, + "learning_rate": 4.559961091296494e-05, + "loss": 3.2945, "step": 26600 }, { "epoch": 0.09, - "learning_rate": 4.557588109559079e-05, - "loss": 3.3851, + "learning_rate": 4.558306809684826e-05, + "loss": 3.275, "step": 26700 }, { "epoch": 0.09, - "learning_rate": 4.555931136186642e-05, - "loss": 3.3826, + "learning_rate": 4.5566525280731594e-05, + "loss": 3.2765, "step": 26800 }, { "epoch": 0.09, - "learning_rate": 4.554274162814204e-05, - "loss": 3.3797, + "learning_rate": 4.554998246461492e-05, + "loss": 3.2737, "step": 26900 }, { "epoch": 0.09, - "learning_rate": 4.5526171894417656e-05, - "loss": 3.3837, + "learning_rate": 4.553343964849824e-05, + "loss": 3.2916, "step": 27000 }, { "epoch": 0.09, - "learning_rate": 4.5509602160693284e-05, - "loss": 3.3823, + "learning_rate": 4.5516896832381575e-05, + "loss": 3.2736, "step": 27100 }, { "epoch": 0.09, - "learning_rate": 4.54930324269689e-05, - "loss": 3.3895, + "learning_rate": 4.55003540162649e-05, + "loss": 3.2818, "step": 27200 }, { "epoch": 0.09, - "learning_rate": 4.547646269324452e-05, - "loss": 3.3909, + "learning_rate": 4.5483811200148224e-05, + "loss": 3.2877, "step": 27300 }, { "epoch": 0.09, - "learning_rate": 4.545989295952014e-05, - "loss": 3.3791, + "learning_rate": 4.5467268384031555e-05, + "loss": 3.2904, "step": 27400 }, { "epoch": 0.09, - "learning_rate": 4.5443323225795764e-05, - "loss": 3.3916, + "learning_rate": 4.545072556791488e-05, + "loss": 3.2939, "step": 27500 }, { "epoch": 0.09, - "learning_rate": 4.5426753492071385e-05, - "loss": 3.389, + "learning_rate": 4.5434182751798204e-05, + "loss": 3.2877, "step": 27600 }, { "epoch": 0.09, - "learning_rate": 4.5410183758347006e-05, - "loss": 3.3773, + "learning_rate": 4.5417639935681535e-05, + "loss": 3.2849, "step": 27700 }, { "epoch": 0.09, - "learning_rate": 4.539361402462263e-05, - "loss": 3.3799, + "learning_rate": 4.540109711956486e-05, + "loss": 3.2945, "step": 27800 }, { "epoch": 0.09, - "learning_rate": 4.537704429089825e-05, - "loss": 3.3973, + "learning_rate": 4.5384554303448185e-05, + "loss": 3.2906, "step": 27900 }, { "epoch": 0.09, - "learning_rate": 4.536047455717387e-05, - "loss": 3.378, + "learning_rate": 4.5368011487331516e-05, + "loss": 3.2866, "step": 28000 }, { "epoch": 0.09, - "learning_rate": 4.534390482344949e-05, - "loss": 3.3853, + "learning_rate": 4.535146867121484e-05, + "loss": 3.2704, "step": 28100 }, { "epoch": 0.09, - "learning_rate": 4.532733508972511e-05, - "loss": 3.3843, + "learning_rate": 4.5334925855098165e-05, + "loss": 3.3005, "step": 28200 }, { "epoch": 0.09, - "learning_rate": 4.5310765356000735e-05, - "loss": 3.3835, + "learning_rate": 4.5318383038981496e-05, + "loss": 3.272, "step": 28300 }, { "epoch": 0.09, - "learning_rate": 4.529419562227635e-05, - "loss": 3.3836, + "learning_rate": 4.530184022286482e-05, + "loss": 3.2867, "step": 28400 }, { "epoch": 0.09, - "learning_rate": 4.527762588855197e-05, - "loss": 3.3765, + "learning_rate": 4.5285297406748145e-05, + "loss": 3.3054, "step": 28500 }, { "epoch": 0.09, - "learning_rate": 4.526105615482759e-05, - "loss": 3.3806, + "learning_rate": 4.5268754590631477e-05, + "loss": 3.2861, "step": 28600 }, { - "epoch": 0.1, - "learning_rate": 4.5244486421103214e-05, - "loss": 3.3771, + "epoch": 0.09, + "learning_rate": 4.52522117745148e-05, + "loss": 3.2856, "step": 28700 }, { "epoch": 0.1, - "learning_rate": 4.5227916687378836e-05, - "loss": 3.3696, + "learning_rate": 4.523566895839813e-05, + "loss": 3.2965, "step": 28800 }, { "epoch": 0.1, - "learning_rate": 4.521134695365446e-05, - "loss": 3.3746, + "learning_rate": 4.521912614228146e-05, + "loss": 3.2931, "step": 28900 }, { "epoch": 0.1, - "learning_rate": 4.519477721993008e-05, - "loss": 3.3839, + "learning_rate": 4.520258332616478e-05, + "loss": 3.2858, "step": 29000 }, { "epoch": 0.1, - "learning_rate": 4.51782074862057e-05, - "loss": 3.3664, + "learning_rate": 4.518604051004811e-05, + "loss": 3.2665, "step": 29100 }, { "epoch": 0.1, - "learning_rate": 4.5161637752481315e-05, - "loss": 3.3917, + "learning_rate": 4.516949769393144e-05, + "loss": 3.2821, "step": 29200 }, { "epoch": 0.1, - "learning_rate": 4.514506801875694e-05, - "loss": 3.3744, + "learning_rate": 4.515295487781476e-05, + "loss": 3.2909, "step": 29300 }, { "epoch": 0.1, - "learning_rate": 4.512849828503256e-05, - "loss": 3.3841, + "learning_rate": 4.513641206169809e-05, + "loss": 3.271, "step": 29400 }, { "epoch": 0.1, - "learning_rate": 4.511192855130818e-05, - "loss": 3.3814, + "learning_rate": 4.511986924558141e-05, + "loss": 3.2822, "step": 29500 }, { "epoch": 0.1, - "learning_rate": 4.509535881758381e-05, - "loss": 3.3754, + "learning_rate": 4.510332642946474e-05, + "loss": 3.2736, "step": 29600 }, { "epoch": 0.1, - "learning_rate": 4.507878908385942e-05, - "loss": 3.3822, + "learning_rate": 4.5086783613348074e-05, + "loss": 3.2868, "step": 29700 }, { "epoch": 0.1, - "learning_rate": 4.5062219350135044e-05, - "loss": 3.3776, + "learning_rate": 4.507024079723139e-05, + "loss": 3.2943, "step": 29800 }, { "epoch": 0.1, - "learning_rate": 4.5045649616410665e-05, - "loss": 3.3866, + "learning_rate": 4.505369798111472e-05, + "loss": 3.3031, "step": 29900 }, { "epoch": 0.1, - "learning_rate": 4.502907988268629e-05, - "loss": 3.3715, + "learning_rate": 4.5037155164998054e-05, + "loss": 3.2958, "step": 30000 }, { "epoch": 0.1, - "learning_rate": 4.501251014896191e-05, - "loss": 3.3865, + "learning_rate": 4.502061234888137e-05, + "loss": 3.2868, "step": 30100 }, { "epoch": 0.1, - "learning_rate": 4.499594041523753e-05, - "loss": 3.3769, + "learning_rate": 4.50040695327647e-05, + "loss": 3.2876, "step": 30200 }, { "epoch": 0.1, - "learning_rate": 4.497937068151315e-05, - "loss": 3.3707, + "learning_rate": 4.4987526716648034e-05, + "loss": 3.2898, "step": 30300 }, { "epoch": 0.1, - "learning_rate": 4.4962800947788766e-05, - "loss": 3.3846, + "learning_rate": 4.497098390053135e-05, + "loss": 3.2959, "step": 30400 }, { "epoch": 0.1, - "learning_rate": 4.4946231214064394e-05, - "loss": 3.3746, + "learning_rate": 4.4954441084414683e-05, + "loss": 3.2966, "step": 30500 }, { "epoch": 0.1, - "learning_rate": 4.4929661480340015e-05, - "loss": 3.3792, + "learning_rate": 4.4937898268298015e-05, + "loss": 3.2712, "step": 30600 }, { "epoch": 0.1, - "learning_rate": 4.491309174661563e-05, - "loss": 3.3696, + "learning_rate": 4.492135545218133e-05, + "loss": 3.2904, "step": 30700 }, { "epoch": 0.1, - "learning_rate": 4.489652201289126e-05, - "loss": 3.3778, + "learning_rate": 4.4904812636064664e-05, + "loss": 3.2859, "step": 30800 }, { "epoch": 0.1, - "learning_rate": 4.487995227916687e-05, - "loss": 3.385, + "learning_rate": 4.4888269819947995e-05, + "loss": 3.2782, "step": 30900 }, { "epoch": 0.1, - "learning_rate": 4.4863382545442495e-05, - "loss": 3.3735, + "learning_rate": 4.487172700383132e-05, + "loss": 3.2856, "step": 31000 }, { "epoch": 0.1, - "learning_rate": 4.484681281171812e-05, - "loss": 3.3848, + "learning_rate": 4.4855184187714644e-05, + "loss": 3.2989, "step": 31100 }, { "epoch": 0.1, - "learning_rate": 4.483024307799374e-05, - "loss": 3.3782, + "learning_rate": 4.4838641371597975e-05, + "loss": 3.2831, "step": 31200 }, { "epoch": 0.1, - "learning_rate": 4.481367334426936e-05, - "loss": 3.3651, + "learning_rate": 4.48220985554813e-05, + "loss": 3.2765, "step": 31300 }, { "epoch": 0.1, - "learning_rate": 4.479710361054498e-05, - "loss": 3.3735, + "learning_rate": 4.4805555739364625e-05, + "loss": 3.2854, "step": 31400 }, { "epoch": 0.1, - "learning_rate": 4.47805338768206e-05, - "loss": 3.3713, + "learning_rate": 4.4789012923247956e-05, + "loss": 3.2868, "step": 31500 }, { "epoch": 0.1, - "learning_rate": 4.476396414309622e-05, - "loss": 3.3738, + "learning_rate": 4.477247010713128e-05, + "loss": 3.2819, "step": 31600 }, { - "epoch": 0.11, - "learning_rate": 4.4747394409371845e-05, - "loss": 3.3613, + "epoch": 0.1, + "learning_rate": 4.4755927291014605e-05, + "loss": 3.2786, "step": 31700 }, { "epoch": 0.11, - "learning_rate": 4.4730824675647466e-05, - "loss": 3.3718, + "learning_rate": 4.4739384474897936e-05, + "loss": 3.2863, "step": 31800 }, { "epoch": 0.11, - "learning_rate": 4.471425494192308e-05, - "loss": 3.3817, + "learning_rate": 4.472284165878126e-05, + "loss": 3.2992, "step": 31900 }, { "epoch": 0.11, - "learning_rate": 4.469768520819871e-05, - "loss": 3.3815, + "learning_rate": 4.4706298842664585e-05, + "loss": 3.3031, "step": 32000 }, { "epoch": 0.11, - "learning_rate": 4.468111547447433e-05, - "loss": 3.3879, + "learning_rate": 4.468975602654791e-05, + "loss": 3.2771, "step": 32100 }, { "epoch": 0.11, - "learning_rate": 4.4664545740749945e-05, - "loss": 3.3864, + "learning_rate": 4.467321321043124e-05, + "loss": 3.2902, "step": 32200 }, { "epoch": 0.11, - "learning_rate": 4.4647976007025574e-05, - "loss": 3.3723, + "learning_rate": 4.4656670394314566e-05, + "loss": 3.2804, "step": 32300 }, { "epoch": 0.11, - "learning_rate": 4.463140627330119e-05, - "loss": 3.3721, + "learning_rate": 4.464012757819789e-05, + "loss": 3.2853, "step": 32400 }, { "epoch": 0.11, - "learning_rate": 4.461483653957681e-05, - "loss": 3.3773, + "learning_rate": 4.462358476208122e-05, + "loss": 3.2879, "step": 32500 }, { "epoch": 0.11, - "learning_rate": 4.459826680585243e-05, - "loss": 3.3717, + "learning_rate": 4.4607041945964546e-05, + "loss": 3.2883, "step": 32600 }, { "epoch": 0.11, - "learning_rate": 4.458169707212805e-05, - "loss": 3.3851, + "learning_rate": 4.459049912984787e-05, + "loss": 3.2902, "step": 32700 }, { "epoch": 0.11, - "learning_rate": 4.4565127338403674e-05, - "loss": 3.3856, + "learning_rate": 4.45739563137312e-05, + "loss": 3.2852, "step": 32800 }, { "epoch": 0.11, - "learning_rate": 4.4548557604679296e-05, - "loss": 3.3751, + "learning_rate": 4.455741349761453e-05, + "loss": 3.2832, "step": 32900 }, { "epoch": 0.11, - "learning_rate": 4.453198787095492e-05, - "loss": 3.3739, + "learning_rate": 4.454087068149785e-05, + "loss": 3.2793, "step": 33000 }, { "epoch": 0.11, - "learning_rate": 4.451541813723054e-05, - "loss": 3.3709, + "learning_rate": 4.452432786538118e-05, + "loss": 3.275, "step": 33100 }, { "epoch": 0.11, - "learning_rate": 4.449884840350616e-05, - "loss": 3.3816, + "learning_rate": 4.4507785049264514e-05, + "loss": 3.281, "step": 33200 }, { "epoch": 0.11, - "learning_rate": 4.448227866978178e-05, - "loss": 3.3694, + "learning_rate": 4.449124223314783e-05, + "loss": 3.284, "step": 33300 }, { "epoch": 0.11, - "learning_rate": 4.4465708936057396e-05, - "loss": 3.3802, + "learning_rate": 4.447469941703116e-05, + "loss": 3.2824, "step": 33400 }, { "epoch": 0.11, - "learning_rate": 4.4449139202333024e-05, - "loss": 3.3768, + "learning_rate": 4.4458156600914494e-05, + "loss": 3.2664, "step": 33500 }, { "epoch": 0.11, - "learning_rate": 4.443256946860864e-05, - "loss": 3.3724, + "learning_rate": 4.444161378479781e-05, + "loss": 3.2768, "step": 33600 }, { "epoch": 0.11, - "learning_rate": 4.441599973488426e-05, - "loss": 3.3795, + "learning_rate": 4.442507096868114e-05, + "loss": 3.2952, "step": 33700 }, { "epoch": 0.11, - "learning_rate": 4.439943000115988e-05, - "loss": 3.3822, + "learning_rate": 4.4408528152564474e-05, + "loss": 3.2902, "step": 33800 }, { "epoch": 0.11, - "learning_rate": 4.4382860267435504e-05, - "loss": 3.3826, + "learning_rate": 4.439198533644779e-05, + "loss": 3.2989, "step": 33900 }, { "epoch": 0.11, - "learning_rate": 4.4366290533711125e-05, - "loss": 3.3671, + "learning_rate": 4.4375442520331123e-05, + "loss": 3.2766, "step": 34000 }, { "epoch": 0.11, - "learning_rate": 4.4349720799986746e-05, - "loss": 3.3808, + "learning_rate": 4.4358899704214455e-05, + "loss": 3.276, "step": 34100 }, { "epoch": 0.11, - "learning_rate": 4.433315106626237e-05, - "loss": 3.389, + "learning_rate": 4.434235688809777e-05, + "loss": 3.2783, "step": 34200 }, { "epoch": 0.11, - "learning_rate": 4.431658133253799e-05, - "loss": 3.368, + "learning_rate": 4.4325814071981104e-05, + "loss": 3.2817, "step": 34300 }, { "epoch": 0.11, - "learning_rate": 4.4300011598813604e-05, - "loss": 3.3826, + "learning_rate": 4.4309271255864435e-05, + "loss": 3.2823, "step": 34400 }, { "epoch": 0.11, - "learning_rate": 4.428344186508923e-05, - "loss": 3.3844, + "learning_rate": 4.429272843974775e-05, + "loss": 3.2793, "step": 34500 }, { "epoch": 0.11, - "learning_rate": 4.4266872131364854e-05, - "loss": 3.3868, + "learning_rate": 4.4276185623631084e-05, + "loss": 3.2741, "step": 34600 }, { "epoch": 0.11, - "learning_rate": 4.425030239764047e-05, - "loss": 3.3844, + "learning_rate": 4.425964280751441e-05, + "loss": 3.2744, "step": 34700 }, { "epoch": 0.12, - "learning_rate": 4.42337326639161e-05, - "loss": 3.3686, + "learning_rate": 4.424309999139773e-05, + "loss": 3.2765, "step": 34800 }, { "epoch": 0.12, - "learning_rate": 4.421716293019171e-05, - "loss": 3.3795, + "learning_rate": 4.4226557175281065e-05, + "loss": 3.285, "step": 34900 }, { "epoch": 0.12, - "learning_rate": 4.420059319646733e-05, - "loss": 3.3693, + "learning_rate": 4.421001435916439e-05, + "loss": 3.2782, "step": 35000 }, { "epoch": 0.12, - "learning_rate": 4.4184023462742954e-05, - "loss": 3.3739, + "learning_rate": 4.419347154304772e-05, + "loss": 3.2879, "step": 35100 }, { "epoch": 0.12, - "learning_rate": 4.4167453729018576e-05, - "loss": 3.359, + "learning_rate": 4.4176928726931045e-05, + "loss": 3.2843, "step": 35200 }, { "epoch": 0.12, - "learning_rate": 4.41508839952942e-05, - "loss": 3.3732, + "learning_rate": 4.416038591081437e-05, + "loss": 3.2782, "step": 35300 }, { "epoch": 0.12, - "learning_rate": 4.413431426156982e-05, - "loss": 3.3698, + "learning_rate": 4.41438430946977e-05, + "loss": 3.2875, "step": 35400 }, { "epoch": 0.12, - "learning_rate": 4.411774452784544e-05, - "loss": 3.3687, + "learning_rate": 4.4127300278581025e-05, + "loss": 3.2691, "step": 35500 }, { "epoch": 0.12, - "learning_rate": 4.410117479412106e-05, - "loss": 3.3669, + "learning_rate": 4.411075746246435e-05, + "loss": 3.2843, "step": 35600 }, { "epoch": 0.12, - "learning_rate": 4.408460506039668e-05, - "loss": 3.3867, + "learning_rate": 4.409421464634768e-05, + "loss": 3.2993, "step": 35700 }, { "epoch": 0.12, - "learning_rate": 4.4068035326672305e-05, - "loss": 3.3566, + "learning_rate": 4.4077671830231006e-05, + "loss": 3.2918, "step": 35800 }, { "epoch": 0.12, - "learning_rate": 4.405146559294792e-05, - "loss": 3.371, + "learning_rate": 4.406112901411433e-05, + "loss": 3.2835, "step": 35900 }, { "epoch": 0.12, - "learning_rate": 4.403489585922355e-05, - "loss": 3.3756, + "learning_rate": 4.404458619799766e-05, + "loss": 3.2957, "step": 36000 }, { "epoch": 0.12, - "learning_rate": 4.401832612549916e-05, - "loss": 3.3673, + "learning_rate": 4.4028043381880986e-05, + "loss": 3.2742, "step": 36100 }, { "epoch": 0.12, - "learning_rate": 4.4001756391774784e-05, - "loss": 3.3673, + "learning_rate": 4.401150056576431e-05, + "loss": 3.2875, "step": 36200 }, { "epoch": 0.12, - "learning_rate": 4.398518665805041e-05, - "loss": 3.3583, + "learning_rate": 4.399495774964764e-05, + "loss": 3.2882, "step": 36300 }, { "epoch": 0.12, - "learning_rate": 4.396861692432603e-05, - "loss": 3.3653, + "learning_rate": 4.3978414933530966e-05, + "loss": 3.271, "step": 36400 }, { "epoch": 0.12, - "learning_rate": 4.395204719060165e-05, - "loss": 3.3724, + "learning_rate": 4.396187211741429e-05, + "loss": 3.2728, "step": 36500 }, { "epoch": 0.12, - "learning_rate": 4.393547745687727e-05, - "loss": 3.3672, + "learning_rate": 4.394532930129762e-05, + "loss": 3.2816, "step": 36600 }, { "epoch": 0.12, - "learning_rate": 4.391890772315289e-05, - "loss": 3.3697, + "learning_rate": 4.392878648518095e-05, + "loss": 3.2792, "step": 36700 }, { "epoch": 0.12, - "learning_rate": 4.390233798942851e-05, - "loss": 3.3775, + "learning_rate": 4.391224366906427e-05, + "loss": 3.2718, "step": 36800 }, { "epoch": 0.12, - "learning_rate": 4.3885768255704134e-05, - "loss": 3.3677, + "learning_rate": 4.38957008529476e-05, + "loss": 3.2781, "step": 36900 }, { "epoch": 0.12, - "learning_rate": 4.3869198521979755e-05, - "loss": 3.3621, + "learning_rate": 4.387915803683093e-05, + "loss": 3.2898, "step": 37000 }, { "epoch": 0.12, - "learning_rate": 4.385262878825537e-05, - "loss": 3.3668, + "learning_rate": 4.386261522071425e-05, + "loss": 3.2766, "step": 37100 }, { "epoch": 0.12, - "learning_rate": 4.3836059054531e-05, - "loss": 3.3578, + "learning_rate": 4.384607240459758e-05, + "loss": 3.272, "step": 37200 }, { "epoch": 0.12, - "learning_rate": 4.381948932080662e-05, - "loss": 3.3661, + "learning_rate": 4.382952958848091e-05, + "loss": 3.2715, "step": 37300 }, { "epoch": 0.12, - "learning_rate": 4.3802919587082235e-05, - "loss": 3.356, + "learning_rate": 4.381298677236423e-05, + "loss": 3.2977, "step": 37400 }, { "epoch": 0.12, - "learning_rate": 4.378634985335786e-05, - "loss": 3.3655, + "learning_rate": 4.3796443956247563e-05, + "loss": 3.2827, "step": 37500 }, { "epoch": 0.12, - "learning_rate": 4.376978011963348e-05, - "loss": 3.3662, + "learning_rate": 4.377990114013089e-05, + "loss": 3.2694, "step": 37600 }, { "epoch": 0.12, - "learning_rate": 4.37532103859091e-05, - "loss": 3.358, + "learning_rate": 4.376335832401421e-05, + "loss": 3.2857, "step": 37700 }, { "epoch": 0.13, - "learning_rate": 4.373664065218473e-05, - "loss": 3.3672, + "learning_rate": 4.3746815507897544e-05, + "loss": 3.3005, "step": 37800 }, { "epoch": 0.13, - "learning_rate": 4.372007091846034e-05, - "loss": 3.3619, + "learning_rate": 4.373027269178087e-05, + "loss": 3.2855, "step": 37900 }, { "epoch": 0.13, - "learning_rate": 4.370350118473596e-05, - "loss": 3.3664, + "learning_rate": 4.371372987566419e-05, + "loss": 3.2917, "step": 38000 }, { "epoch": 0.13, - "learning_rate": 4.3686931451011585e-05, - "loss": 3.3591, + "learning_rate": 4.3697187059547524e-05, + "loss": 3.2836, "step": 38100 }, { "epoch": 0.13, - "learning_rate": 4.3670361717287206e-05, - "loss": 3.3668, + "learning_rate": 4.368064424343085e-05, + "loss": 3.2942, "step": 38200 }, { "epoch": 0.13, - "learning_rate": 4.365379198356283e-05, - "loss": 3.3798, + "learning_rate": 4.366410142731417e-05, + "loss": 3.2996, "step": 38300 }, { "epoch": 0.13, - "learning_rate": 4.363722224983845e-05, - "loss": 3.3659, + "learning_rate": 4.3647558611197505e-05, + "loss": 3.284, "step": 38400 }, { "epoch": 0.13, - "learning_rate": 4.362065251611407e-05, - "loss": 3.3577, + "learning_rate": 4.363101579508083e-05, + "loss": 3.2787, "step": 38500 }, { "epoch": 0.13, - "learning_rate": 4.3604082782389685e-05, - "loss": 3.3606, + "learning_rate": 4.3614472978964154e-05, + "loss": 3.2885, "step": 38600 }, { "epoch": 0.13, - "learning_rate": 4.3587513048665314e-05, - "loss": 3.3677, + "learning_rate": 4.3597930162847485e-05, + "loss": 3.2825, "step": 38700 }, { "epoch": 0.13, - "learning_rate": 4.3570943314940935e-05, - "loss": 3.3597, + "learning_rate": 4.358138734673081e-05, + "loss": 3.2837, "step": 38800 }, { "epoch": 0.13, - "learning_rate": 4.355437358121655e-05, - "loss": 3.3561, + "learning_rate": 4.3564844530614134e-05, + "loss": 3.2908, "step": 38900 }, { "epoch": 0.13, - "learning_rate": 4.353780384749217e-05, - "loss": 3.3694, + "learning_rate": 4.3548301714497465e-05, + "loss": 3.2762, "step": 39000 }, { "epoch": 0.13, - "learning_rate": 4.352123411376779e-05, - "loss": 3.359, + "learning_rate": 4.353175889838079e-05, + "loss": 3.2765, "step": 39100 }, { "epoch": 0.13, - "learning_rate": 4.3504664380043414e-05, - "loss": 3.3635, + "learning_rate": 4.351521608226412e-05, + "loss": 3.2976, "step": 39200 }, { "epoch": 0.13, - "learning_rate": 4.3488094646319036e-05, - "loss": 3.362, + "learning_rate": 4.3498673266147446e-05, + "loss": 3.2692, "step": 39300 }, { "epoch": 0.13, - "learning_rate": 4.347152491259466e-05, - "loss": 3.3827, + "learning_rate": 4.348213045003077e-05, + "loss": 3.2811, "step": 39400 }, { "epoch": 0.13, - "learning_rate": 4.345495517887028e-05, - "loss": 3.3742, + "learning_rate": 4.34655876339141e-05, + "loss": 3.2734, "step": 39500 }, { "epoch": 0.13, - "learning_rate": 4.343838544514589e-05, - "loss": 3.3614, + "learning_rate": 4.3449044817797426e-05, + "loss": 3.2943, "step": 39600 }, { "epoch": 0.13, - "learning_rate": 4.342181571142152e-05, - "loss": 3.3722, + "learning_rate": 4.343250200168075e-05, + "loss": 3.2713, "step": 39700 }, { "epoch": 0.13, - "learning_rate": 4.340524597769714e-05, - "loss": 3.3692, + "learning_rate": 4.341595918556408e-05, + "loss": 3.2813, "step": 39800 }, { "epoch": 0.13, - "learning_rate": 4.338867624397276e-05, - "loss": 3.3579, + "learning_rate": 4.3399416369447406e-05, + "loss": 3.2692, "step": 39900 }, { "epoch": 0.13, - "learning_rate": 4.3372106510248386e-05, - "loss": 3.3648, + "learning_rate": 4.338287355333073e-05, + "loss": 3.2853, "step": 40000 }, { "epoch": 0.13, - "learning_rate": 4.3355536776524e-05, - "loss": 3.3643, + "learning_rate": 4.336633073721406e-05, + "loss": 3.2784, "step": 40100 }, { "epoch": 0.13, - "learning_rate": 4.333896704279962e-05, - "loss": 3.3631, + "learning_rate": 4.334978792109739e-05, + "loss": 3.2939, "step": 40200 }, { "epoch": 0.13, - "learning_rate": 4.3322397309075244e-05, - "loss": 3.3611, + "learning_rate": 4.333324510498071e-05, + "loss": 3.2723, "step": 40300 }, { "epoch": 0.13, - "learning_rate": 4.3305827575350865e-05, - "loss": 3.3628, + "learning_rate": 4.331670228886404e-05, + "loss": 3.277, "step": 40400 }, { "epoch": 0.13, - "learning_rate": 4.3289257841626486e-05, - "loss": 3.3738, + "learning_rate": 4.330015947274737e-05, + "loss": 3.2739, "step": 40500 }, { "epoch": 0.13, - "learning_rate": 4.327268810790211e-05, - "loss": 3.3649, + "learning_rate": 4.328361665663069e-05, + "loss": 3.2745, "step": 40600 }, { "epoch": 0.13, - "learning_rate": 4.325611837417773e-05, - "loss": 3.359, + "learning_rate": 4.326707384051402e-05, + "loss": 3.2738, "step": 40700 }, { - "epoch": 0.14, - "learning_rate": 4.323954864045335e-05, - "loss": 3.3587, + "epoch": 0.13, + "learning_rate": 4.325053102439735e-05, + "loss": 3.2705, "step": 40800 }, { "epoch": 0.14, - "learning_rate": 4.322297890672897e-05, - "loss": 3.3606, + "learning_rate": 4.323398820828067e-05, + "loss": 3.2818, "step": 40900 }, { "epoch": 0.14, - "learning_rate": 4.3206409173004594e-05, - "loss": 3.3781, + "learning_rate": 4.3217445392164003e-05, + "loss": 3.2993, "step": 41000 }, { "epoch": 0.14, - "learning_rate": 4.318983943928021e-05, - "loss": 3.3546, + "learning_rate": 4.320090257604733e-05, + "loss": 3.2867, "step": 41100 }, { "epoch": 0.14, - "learning_rate": 4.317326970555584e-05, - "loss": 3.3595, + "learning_rate": 4.318435975993065e-05, + "loss": 3.2908, "step": 41200 }, { "epoch": 0.14, - "learning_rate": 4.315669997183145e-05, - "loss": 3.3702, + "learning_rate": 4.3167816943813984e-05, + "loss": 3.2906, "step": 41300 }, { "epoch": 0.14, - "learning_rate": 4.314013023810707e-05, - "loss": 3.3741, + "learning_rate": 4.315127412769731e-05, + "loss": 3.2861, "step": 41400 }, { "epoch": 0.14, - "learning_rate": 4.31235605043827e-05, - "loss": 3.3655, + "learning_rate": 4.313473131158063e-05, + "loss": 3.2768, "step": 41500 }, { "epoch": 0.14, - "learning_rate": 4.3106990770658316e-05, - "loss": 3.3779, + "learning_rate": 4.3118188495463964e-05, + "loss": 3.2923, "step": 41600 }, { "epoch": 0.14, - "learning_rate": 4.309042103693394e-05, - "loss": 3.3575, + "learning_rate": 4.310164567934729e-05, + "loss": 3.2733, "step": 41700 }, { "epoch": 0.14, - "learning_rate": 4.307385130320956e-05, - "loss": 3.3653, + "learning_rate": 4.308510286323061e-05, + "loss": 3.2874, "step": 41800 }, { "epoch": 0.14, - "learning_rate": 4.305728156948518e-05, - "loss": 3.3474, + "learning_rate": 4.3068560047113945e-05, + "loss": 3.2837, "step": 41900 }, { "epoch": 0.14, - "learning_rate": 4.30407118357608e-05, - "loss": 3.3698, + "learning_rate": 4.305201723099727e-05, + "loss": 3.2827, "step": 42000 }, { "epoch": 0.14, - "learning_rate": 4.302414210203642e-05, - "loss": 3.3659, + "learning_rate": 4.3035474414880594e-05, + "loss": 3.2816, "step": 42100 }, { "epoch": 0.14, - "learning_rate": 4.3007572368312045e-05, - "loss": 3.3652, + "learning_rate": 4.3018931598763925e-05, + "loss": 3.2883, "step": 42200 }, { "epoch": 0.14, - "learning_rate": 4.299100263458766e-05, - "loss": 3.3659, + "learning_rate": 4.300238878264725e-05, + "loss": 3.2763, "step": 42300 }, { "epoch": 0.14, - "learning_rate": 4.297443290086329e-05, - "loss": 3.3692, + "learning_rate": 4.2985845966530574e-05, + "loss": 3.2642, "step": 42400 }, { "epoch": 0.14, - "learning_rate": 4.295786316713891e-05, - "loss": 3.3607, + "learning_rate": 4.2969303150413905e-05, + "loss": 3.2771, "step": 42500 }, { "epoch": 0.14, - "learning_rate": 4.2941293433414524e-05, - "loss": 3.3552, + "learning_rate": 4.295276033429723e-05, + "loss": 3.2745, "step": 42600 }, { "epoch": 0.14, - "learning_rate": 4.292472369969015e-05, - "loss": 3.3682, + "learning_rate": 4.2936217518180554e-05, + "loss": 3.2742, "step": 42700 }, { "epoch": 0.14, - "learning_rate": 4.290815396596577e-05, - "loss": 3.3593, + "learning_rate": 4.2919674702063886e-05, + "loss": 3.2976, "step": 42800 }, { "epoch": 0.14, - "learning_rate": 4.289158423224139e-05, - "loss": 3.3651, + "learning_rate": 4.290313188594721e-05, + "loss": 3.2856, "step": 42900 }, { "epoch": 0.14, - "learning_rate": 4.2875014498517016e-05, - "loss": 3.3516, + "learning_rate": 4.2886589069830535e-05, + "loss": 3.2877, "step": 43000 }, { "epoch": 0.14, - "learning_rate": 4.285844476479263e-05, - "loss": 3.3787, + "learning_rate": 4.2870046253713866e-05, + "loss": 3.268, "step": 43100 }, { "epoch": 0.14, - "learning_rate": 4.284187503106825e-05, - "loss": 3.3641, + "learning_rate": 4.285350343759719e-05, + "loss": 3.2837, "step": 43200 }, { "epoch": 0.14, - "learning_rate": 4.2825305297343874e-05, - "loss": 3.3722, + "learning_rate": 4.2836960621480515e-05, + "loss": 3.2773, "step": 43300 }, { "epoch": 0.14, - "learning_rate": 4.2808735563619495e-05, - "loss": 3.3714, + "learning_rate": 4.2820417805363847e-05, + "loss": 3.2718, "step": 43400 }, { "epoch": 0.14, - "learning_rate": 4.279216582989512e-05, - "loss": 3.35, + "learning_rate": 4.280387498924717e-05, + "loss": 3.2883, "step": 43500 }, { "epoch": 0.14, - "learning_rate": 4.277559609617074e-05, - "loss": 3.3641, + "learning_rate": 4.27873321731305e-05, + "loss": 3.2788, "step": 43600 }, { "epoch": 0.14, - "learning_rate": 4.275902636244636e-05, - "loss": 3.3642, + "learning_rate": 4.277078935701383e-05, + "loss": 3.2846, "step": 43700 }, { - "epoch": 0.15, - "learning_rate": 4.2742456628721975e-05, - "loss": 3.3644, + "epoch": 0.14, + "learning_rate": 4.275424654089715e-05, + "loss": 3.2723, "step": 43800 }, { "epoch": 0.15, - "learning_rate": 4.27258868949976e-05, - "loss": 3.3621, + "learning_rate": 4.273770372478048e-05, + "loss": 3.2704, "step": 43900 }, { "epoch": 0.15, - "learning_rate": 4.2709317161273224e-05, - "loss": 3.3665, + "learning_rate": 4.272116090866381e-05, + "loss": 3.2752, "step": 44000 }, { "epoch": 0.15, - "learning_rate": 4.269274742754884e-05, - "loss": 3.3653, + "learning_rate": 4.270461809254713e-05, + "loss": 3.2712, "step": 44100 }, { "epoch": 0.15, - "learning_rate": 4.267617769382446e-05, - "loss": 3.3556, + "learning_rate": 4.268807527643046e-05, + "loss": 3.2812, "step": 44200 }, { "epoch": 0.15, - "learning_rate": 4.265960796010008e-05, - "loss": 3.3414, + "learning_rate": 4.267153246031379e-05, + "loss": 3.2719, "step": 44300 }, { "epoch": 0.15, - "learning_rate": 4.26430382263757e-05, - "loss": 3.3707, + "learning_rate": 4.265498964419711e-05, + "loss": 3.2726, "step": 44400 }, { "epoch": 0.15, - "learning_rate": 4.2626468492651325e-05, - "loss": 3.3754, + "learning_rate": 4.2638446828080444e-05, + "loss": 3.2883, "step": 44500 }, { "epoch": 0.15, - "learning_rate": 4.2609898758926946e-05, - "loss": 3.3685, + "learning_rate": 4.262190401196377e-05, + "loss": 3.2805, "step": 44600 }, { "epoch": 0.15, - "learning_rate": 4.259332902520257e-05, - "loss": 3.3681, + "learning_rate": 4.260536119584709e-05, + "loss": 3.3015, "step": 44700 }, { "epoch": 0.15, - "learning_rate": 4.257675929147818e-05, - "loss": 3.3646, + "learning_rate": 4.2588818379730424e-05, + "loss": 3.2709, "step": 44800 }, { "epoch": 0.15, - "learning_rate": 4.256018955775381e-05, - "loss": 3.3509, + "learning_rate": 4.257227556361374e-05, + "loss": 3.2864, "step": 44900 }, { "epoch": 0.15, - "learning_rate": 4.254361982402943e-05, - "loss": 3.3567, + "learning_rate": 4.255573274749707e-05, + "loss": 3.2695, "step": 45000 }, { "epoch": 0.15, - "learning_rate": 4.252705009030505e-05, - "loss": 3.3648, + "learning_rate": 4.2539189931380404e-05, + "loss": 3.276, "step": 45100 }, { "epoch": 0.15, - "learning_rate": 4.2510480356580675e-05, - "loss": 3.3566, + "learning_rate": 4.252264711526372e-05, + "loss": 3.2666, "step": 45200 }, { "epoch": 0.15, - "learning_rate": 4.249391062285629e-05, - "loss": 3.3473, + "learning_rate": 4.250610429914705e-05, + "loss": 3.2815, "step": 45300 }, { "epoch": 0.15, - "learning_rate": 4.247734088913191e-05, - "loss": 3.3616, + "learning_rate": 4.2489561483030385e-05, + "loss": 3.2822, "step": 45400 }, { "epoch": 0.15, - "learning_rate": 4.246077115540753e-05, - "loss": 3.3543, + "learning_rate": 4.24730186669137e-05, + "loss": 3.2792, "step": 45500 }, { "epoch": 0.15, - "learning_rate": 4.2444201421683154e-05, - "loss": 3.3596, + "learning_rate": 4.2456475850797034e-05, + "loss": 3.2849, "step": 45600 }, { "epoch": 0.15, - "learning_rate": 4.2427631687958776e-05, - "loss": 3.3504, + "learning_rate": 4.2439933034680365e-05, + "loss": 3.2803, "step": 45700 }, { "epoch": 0.15, - "learning_rate": 4.24110619542344e-05, - "loss": 3.3509, + "learning_rate": 4.242339021856369e-05, + "loss": 3.2746, "step": 45800 }, { "epoch": 0.15, - "learning_rate": 4.239449222051002e-05, - "loss": 3.3641, + "learning_rate": 4.2406847402447014e-05, + "loss": 3.2647, "step": 45900 }, { "epoch": 0.15, - "learning_rate": 4.237792248678564e-05, - "loss": 3.366, + "learning_rate": 4.2390304586330345e-05, + "loss": 3.2782, "step": 46000 }, { "epoch": 0.15, - "learning_rate": 4.236135275306126e-05, - "loss": 3.3578, + "learning_rate": 4.237376177021367e-05, + "loss": 3.275, "step": 46100 }, { "epoch": 0.15, - "learning_rate": 4.234478301933688e-05, - "loss": 3.3682, + "learning_rate": 4.2357218954096994e-05, + "loss": 3.2574, "step": 46200 }, { "epoch": 0.15, - "learning_rate": 4.23282132856125e-05, - "loss": 3.3578, + "learning_rate": 4.2340676137980326e-05, + "loss": 3.2702, "step": 46300 }, { "epoch": 0.15, - "learning_rate": 4.2311643551888126e-05, - "loss": 3.3676, + "learning_rate": 4.232413332186365e-05, + "loss": 3.2754, "step": 46400 }, { "epoch": 0.15, - "learning_rate": 4.229507381816375e-05, - "loss": 3.3432, + "learning_rate": 4.2307590505746975e-05, + "loss": 3.2867, "step": 46500 }, { "epoch": 0.15, - "learning_rate": 4.227850408443936e-05, - "loss": 3.3726, + "learning_rate": 4.2291047689630306e-05, + "loss": 3.2667, "step": 46600 }, { "epoch": 0.15, - "learning_rate": 4.226193435071499e-05, - "loss": 3.3437, + "learning_rate": 4.227450487351363e-05, + "loss": 3.2795, "step": 46700 }, { - "epoch": 0.16, - "learning_rate": 4.2245364616990605e-05, - "loss": 3.3589, + "epoch": 0.15, + "learning_rate": 4.2257962057396955e-05, + "loss": 3.2763, "step": 46800 }, { "epoch": 0.16, - "learning_rate": 4.2228794883266226e-05, - "loss": 3.3523, + "learning_rate": 4.2241419241280287e-05, + "loss": 3.2788, "step": 46900 }, { "epoch": 0.16, - "learning_rate": 4.221222514954185e-05, - "loss": 3.3557, + "learning_rate": 4.222487642516361e-05, + "loss": 3.2677, "step": 47000 }, { "epoch": 0.16, - "learning_rate": 4.219565541581747e-05, - "loss": 3.3583, + "learning_rate": 4.2208333609046936e-05, + "loss": 3.2683, "step": 47100 }, { "epoch": 0.16, - "learning_rate": 4.217908568209309e-05, - "loss": 3.3515, + "learning_rate": 4.219179079293027e-05, + "loss": 3.2804, "step": 47200 }, { "epoch": 0.16, - "learning_rate": 4.216251594836871e-05, - "loss": 3.3537, + "learning_rate": 4.217524797681359e-05, + "loss": 3.2729, "step": 47300 }, { "epoch": 0.16, - "learning_rate": 4.2145946214644334e-05, - "loss": 3.3638, + "learning_rate": 4.2158705160696916e-05, + "loss": 3.2818, "step": 47400 }, { "epoch": 0.16, - "learning_rate": 4.2129376480919955e-05, - "loss": 3.3708, + "learning_rate": 4.214216234458024e-05, + "loss": 3.2829, "step": 47500 }, { "epoch": 0.16, - "learning_rate": 4.211280674719558e-05, - "loss": 3.3518, + "learning_rate": 4.212561952846357e-05, + "loss": 3.272, "step": 47600 }, { "epoch": 0.16, - "learning_rate": 4.20962370134712e-05, - "loss": 3.3519, + "learning_rate": 4.21090767123469e-05, + "loss": 3.2847, "step": 47700 }, { "epoch": 0.16, - "learning_rate": 4.207966727974681e-05, - "loss": 3.3561, + "learning_rate": 4.209253389623022e-05, + "loss": 3.2814, "step": 47800 }, { "epoch": 0.16, - "learning_rate": 4.206309754602244e-05, - "loss": 3.362, + "learning_rate": 4.207599108011355e-05, + "loss": 3.2718, "step": 47900 }, { "epoch": 0.16, - "learning_rate": 4.2046527812298056e-05, - "loss": 3.3666, + "learning_rate": 4.2059448263996884e-05, + "loss": 3.2912, "step": 48000 }, { "epoch": 0.16, - "learning_rate": 4.202995807857368e-05, - "loss": 3.3634, + "learning_rate": 4.20429054478802e-05, + "loss": 3.2841, "step": 48100 }, { "epoch": 0.16, - "learning_rate": 4.2013388344849306e-05, - "loss": 3.36, + "learning_rate": 4.202636263176353e-05, + "loss": 3.2781, "step": 48200 }, { "epoch": 0.16, - "learning_rate": 4.199681861112492e-05, - "loss": 3.3616, + "learning_rate": 4.2009819815646864e-05, + "loss": 3.2631, "step": 48300 }, { "epoch": 0.16, - "learning_rate": 4.198024887740054e-05, - "loss": 3.3535, + "learning_rate": 4.199327699953018e-05, + "loss": 3.2655, "step": 48400 }, { "epoch": 0.16, - "learning_rate": 4.196367914367616e-05, - "loss": 3.3517, + "learning_rate": 4.197673418341351e-05, + "loss": 3.2698, "step": 48500 }, { "epoch": 0.16, - "learning_rate": 4.1947109409951785e-05, - "loss": 3.3619, + "learning_rate": 4.1960191367296844e-05, + "loss": 3.2651, "step": 48600 }, { "epoch": 0.16, - "learning_rate": 4.1930539676227406e-05, - "loss": 3.3597, + "learning_rate": 4.194364855118016e-05, + "loss": 3.2819, "step": 48700 }, { "epoch": 0.16, - "learning_rate": 4.191396994250303e-05, - "loss": 3.3536, + "learning_rate": 4.192710573506349e-05, + "loss": 3.2709, "step": 48800 }, { "epoch": 0.16, - "learning_rate": 4.189740020877865e-05, - "loss": 3.3646, + "learning_rate": 4.1910562918946825e-05, + "loss": 3.2778, "step": 48900 }, { "epoch": 0.16, - "learning_rate": 4.1880830475054264e-05, - "loss": 3.3594, + "learning_rate": 4.189402010283014e-05, + "loss": 3.2856, "step": 49000 }, { "epoch": 0.16, - "learning_rate": 4.186426074132989e-05, - "loss": 3.3587, + "learning_rate": 4.1877477286713474e-05, + "loss": 3.2965, "step": 49100 }, { "epoch": 0.16, - "learning_rate": 4.1847691007605513e-05, - "loss": 3.3618, + "learning_rate": 4.1860934470596805e-05, + "loss": 3.2803, "step": 49200 }, { "epoch": 0.16, - "learning_rate": 4.183112127388113e-05, - "loss": 3.3501, + "learning_rate": 4.184439165448012e-05, + "loss": 3.263, "step": 49300 }, { "epoch": 0.16, - "learning_rate": 4.181455154015675e-05, - "loss": 3.3584, + "learning_rate": 4.1827848838363454e-05, + "loss": 3.2755, "step": 49400 }, { "epoch": 0.16, - "learning_rate": 4.179798180643237e-05, - "loss": 3.3569, + "learning_rate": 4.1811306022246785e-05, + "loss": 3.2782, "step": 49500 }, { "epoch": 0.16, - "learning_rate": 4.178141207270799e-05, - "loss": 3.3504, + "learning_rate": 4.17947632061301e-05, + "loss": 3.2788, "step": 49600 }, { "epoch": 0.16, - "learning_rate": 4.1764842338983614e-05, - "loss": 3.3556, + "learning_rate": 4.1778220390013434e-05, + "loss": 3.2698, "step": 49700 }, { - "epoch": 0.17, - "learning_rate": 4.1748272605259235e-05, - "loss": 3.357, + "epoch": 0.16, + "learning_rate": 4.1761677573896766e-05, + "loss": 3.2774, "step": 49800 }, { "epoch": 0.17, - "learning_rate": 4.173170287153486e-05, - "loss": 3.3402, + "learning_rate": 4.174513475778009e-05, + "loss": 3.2741, "step": 49900 }, { "epoch": 0.17, - "learning_rate": 4.171513313781047e-05, - "loss": 3.3407, + "learning_rate": 4.1728591941663415e-05, + "loss": 3.287, "step": 50000 }, { "epoch": 0.17, - "learning_rate": 4.16985634040861e-05, - "loss": 3.3431, + "learning_rate": 4.171204912554674e-05, + "loss": 3.2744, "step": 50100 }, { "epoch": 0.17, - "learning_rate": 4.168199367036172e-05, - "loss": 3.3394, + "learning_rate": 4.169550630943007e-05, + "loss": 3.2841, "step": 50200 }, { "epoch": 0.17, - "learning_rate": 4.1665423936637336e-05, - "loss": 3.3449, + "learning_rate": 4.1678963493313395e-05, + "loss": 3.2667, "step": 50300 }, { "epoch": 0.17, - "learning_rate": 4.1648854202912964e-05, - "loss": 3.3435, + "learning_rate": 4.166242067719672e-05, + "loss": 3.2733, "step": 50400 }, { "epoch": 0.17, - "learning_rate": 4.163228446918858e-05, - "loss": 3.371, + "learning_rate": 4.164587786108005e-05, + "loss": 3.2774, "step": 50500 }, { "epoch": 0.17, - "learning_rate": 4.16157147354642e-05, - "loss": 3.3492, + "learning_rate": 4.1629335044963376e-05, + "loss": 3.2722, "step": 50600 }, { "epoch": 0.17, - "learning_rate": 4.159914500173983e-05, - "loss": 3.3462, + "learning_rate": 4.16127922288467e-05, + "loss": 3.2764, "step": 50700 }, { "epoch": 0.17, - "learning_rate": 4.158257526801544e-05, - "loss": 3.3562, + "learning_rate": 4.159624941273003e-05, + "loss": 3.2835, "step": 50800 }, { "epoch": 0.17, - "learning_rate": 4.1566005534291065e-05, - "loss": 3.346, + "learning_rate": 4.1579706596613356e-05, + "loss": 3.2851, "step": 50900 }, { "epoch": 0.17, - "learning_rate": 4.1549435800566686e-05, - "loss": 3.3605, + "learning_rate": 4.156316378049668e-05, + "loss": 3.2814, "step": 51000 }, { "epoch": 0.17, - "learning_rate": 4.153286606684231e-05, - "loss": 3.3543, + "learning_rate": 4.154662096438001e-05, + "loss": 3.2749, "step": 51100 }, { "epoch": 0.17, - "learning_rate": 4.151629633311793e-05, - "loss": 3.3502, + "learning_rate": 4.1530078148263336e-05, + "loss": 3.2696, "step": 51200 }, { "epoch": 0.17, - "learning_rate": 4.149972659939355e-05, - "loss": 3.3494, + "learning_rate": 4.151353533214666e-05, + "loss": 3.2935, "step": 51300 }, { "epoch": 0.17, - "learning_rate": 4.148315686566917e-05, - "loss": 3.3549, + "learning_rate": 4.149699251602999e-05, + "loss": 3.2737, "step": 51400 }, { "epoch": 0.17, - "learning_rate": 4.146658713194479e-05, - "loss": 3.3679, + "learning_rate": 4.148044969991332e-05, + "loss": 3.2831, "step": 51500 }, { "epoch": 0.17, - "learning_rate": 4.1450017398220415e-05, - "loss": 3.3514, + "learning_rate": 4.146390688379664e-05, + "loss": 3.2638, "step": 51600 }, { "epoch": 0.17, - "learning_rate": 4.1433447664496037e-05, - "loss": 3.3521, + "learning_rate": 4.144736406767997e-05, + "loss": 3.273, "step": 51700 }, { "epoch": 0.17, - "learning_rate": 4.141687793077165e-05, - "loss": 3.3588, + "learning_rate": 4.14308212515633e-05, + "loss": 3.2786, "step": 51800 }, { "epoch": 0.17, - "learning_rate": 4.140030819704728e-05, - "loss": 3.3486, + "learning_rate": 4.141427843544662e-05, + "loss": 3.28, "step": 51900 }, { "epoch": 0.17, - "learning_rate": 4.1383738463322894e-05, - "loss": 3.3494, + "learning_rate": 4.139773561932995e-05, + "loss": 3.2827, "step": 52000 }, { "epoch": 0.17, - "learning_rate": 4.1367168729598516e-05, - "loss": 3.3549, + "learning_rate": 4.1381192803213284e-05, + "loss": 3.2688, "step": 52100 }, { "epoch": 0.17, - "learning_rate": 4.135059899587414e-05, - "loss": 3.3449, + "learning_rate": 4.13646499870966e-05, + "loss": 3.2596, "step": 52200 }, { "epoch": 0.17, - "learning_rate": 4.133402926214976e-05, - "loss": 3.3437, + "learning_rate": 4.134810717097993e-05, + "loss": 3.2704, "step": 52300 }, { "epoch": 0.17, - "learning_rate": 4.131745952842538e-05, - "loss": 3.3581, + "learning_rate": 4.1331564354863265e-05, + "loss": 3.2792, "step": 52400 }, { "epoch": 0.17, - "learning_rate": 4.1300889794701e-05, - "loss": 3.3456, + "learning_rate": 4.131502153874658e-05, + "loss": 3.2734, "step": 52500 }, { "epoch": 0.17, - "learning_rate": 4.128432006097662e-05, - "loss": 3.3528, + "learning_rate": 4.1298478722629914e-05, + "loss": 3.2838, "step": 52600 }, { "epoch": 0.17, - "learning_rate": 4.1267750327252244e-05, - "loss": 3.3532, + "learning_rate": 4.128193590651324e-05, + "loss": 3.2762, "step": 52700 }, { "epoch": 0.17, - "learning_rate": 4.1251180593527866e-05, - "loss": 3.3497, + "learning_rate": 4.126539309039656e-05, + "loss": 3.2819, "step": 52800 }, { "epoch": 0.18, - "learning_rate": 4.123461085980349e-05, - "loss": 3.3606, + "learning_rate": 4.1248850274279894e-05, + "loss": 3.2741, "step": 52900 }, { "epoch": 0.18, - "learning_rate": 4.12180411260791e-05, - "loss": 3.3414, + "learning_rate": 4.123230745816322e-05, + "loss": 3.2653, "step": 53000 }, { "epoch": 0.18, - "learning_rate": 4.120147139235473e-05, - "loss": 3.3491, + "learning_rate": 4.121576464204654e-05, + "loss": 3.2647, "step": 53100 }, { "epoch": 0.18, - "learning_rate": 4.1184901658630345e-05, - "loss": 3.352, + "learning_rate": 4.1199221825929875e-05, + "loss": 3.2723, "step": 53200 }, { "epoch": 0.18, - "learning_rate": 4.1168331924905966e-05, - "loss": 3.3549, + "learning_rate": 4.11826790098132e-05, + "loss": 3.2606, "step": 53300 }, { "epoch": 0.18, - "learning_rate": 4.1151762191181595e-05, - "loss": 3.3632, + "learning_rate": 4.1166136193696524e-05, + "loss": 3.2717, "step": 53400 }, { "epoch": 0.18, - "learning_rate": 4.113519245745721e-05, - "loss": 3.3477, + "learning_rate": 4.1149593377579855e-05, + "loss": 3.2796, "step": 53500 }, { "epoch": 0.18, - "learning_rate": 4.111862272373283e-05, - "loss": 3.3553, + "learning_rate": 4.113305056146318e-05, + "loss": 3.2854, "step": 53600 }, { "epoch": 0.18, - "learning_rate": 4.110205299000845e-05, - "loss": 3.3426, + "learning_rate": 4.1116507745346504e-05, + "loss": 3.2598, "step": 53700 }, { "epoch": 0.18, - "learning_rate": 4.1085483256284074e-05, - "loss": 3.3422, + "learning_rate": 4.1099964929229835e-05, + "loss": 3.2887, "step": 53800 }, { "epoch": 0.18, - "learning_rate": 4.1068913522559695e-05, - "loss": 3.3405, + "learning_rate": 4.108342211311316e-05, + "loss": 3.2692, "step": 53900 }, { "epoch": 0.18, - "learning_rate": 4.105234378883532e-05, - "loss": 3.3615, + "learning_rate": 4.1066879296996484e-05, + "loss": 3.2895, "step": 54000 }, { "epoch": 0.18, - "learning_rate": 4.103577405511094e-05, - "loss": 3.362, + "learning_rate": 4.1050336480879816e-05, + "loss": 3.2643, "step": 54100 }, { "epoch": 0.18, - "learning_rate": 4.101920432138655e-05, - "loss": 3.3421, + "learning_rate": 4.103379366476314e-05, + "loss": 3.2739, "step": 54200 }, { "epoch": 0.18, - "learning_rate": 4.100263458766218e-05, - "loss": 3.345, + "learning_rate": 4.101725084864647e-05, + "loss": 3.2844, "step": 54300 }, { "epoch": 0.18, - "learning_rate": 4.09860648539378e-05, - "loss": 3.3491, + "learning_rate": 4.1000708032529796e-05, + "loss": 3.2898, "step": 54400 }, { "epoch": 0.18, - "learning_rate": 4.096949512021342e-05, - "loss": 3.3449, + "learning_rate": 4.098416521641312e-05, + "loss": 3.2769, "step": 54500 }, { "epoch": 0.18, - "learning_rate": 4.095292538648904e-05, - "loss": 3.3471, + "learning_rate": 4.096762240029645e-05, + "loss": 3.276, "step": 54600 }, { "epoch": 0.18, - "learning_rate": 4.093635565276466e-05, - "loss": 3.3436, + "learning_rate": 4.0951079584179776e-05, + "loss": 3.2716, "step": 54700 }, { "epoch": 0.18, - "learning_rate": 4.091978591904028e-05, - "loss": 3.3628, + "learning_rate": 4.09345367680631e-05, + "loss": 3.275, "step": 54800 }, { "epoch": 0.18, - "learning_rate": 4.09032161853159e-05, - "loss": 3.3518, + "learning_rate": 4.091799395194643e-05, + "loss": 3.2815, "step": 54900 }, { "epoch": 0.18, - "learning_rate": 4.0886646451591525e-05, - "loss": 3.3411, + "learning_rate": 4.090145113582976e-05, + "loss": 3.2617, "step": 55000 }, { "epoch": 0.18, - "learning_rate": 4.0870076717867146e-05, - "loss": 3.3333, + "learning_rate": 4.088490831971308e-05, + "loss": 3.2822, "step": 55100 }, { "epoch": 0.18, - "learning_rate": 4.085350698414277e-05, - "loss": 3.3354, + "learning_rate": 4.086836550359641e-05, + "loss": 3.275, "step": 55200 }, { "epoch": 0.18, - "learning_rate": 4.083693725041839e-05, - "loss": 3.3366, + "learning_rate": 4.085182268747974e-05, + "loss": 3.2682, "step": 55300 }, { "epoch": 0.18, - "learning_rate": 4.082036751669401e-05, - "loss": 3.3402, + "learning_rate": 4.083527987136306e-05, + "loss": 3.276, "step": 55400 }, { "epoch": 0.18, - "learning_rate": 4.0803797782969625e-05, - "loss": 3.3381, + "learning_rate": 4.081873705524639e-05, + "loss": 3.2635, "step": 55500 }, { "epoch": 0.18, - "learning_rate": 4.0787228049245253e-05, - "loss": 3.3444, + "learning_rate": 4.080219423912972e-05, + "loss": 3.2695, "step": 55600 }, { "epoch": 0.18, - "learning_rate": 4.077065831552087e-05, - "loss": 3.3459, + "learning_rate": 4.078565142301304e-05, + "loss": 3.2798, "step": 55700 }, { "epoch": 0.18, - "learning_rate": 4.075408858179649e-05, - "loss": 3.3572, + "learning_rate": 4.0769108606896373e-05, + "loss": 3.2667, "step": 55800 }, { - "epoch": 0.19, - "learning_rate": 4.073751884807212e-05, - "loss": 3.344, + "epoch": 0.18, + "learning_rate": 4.07525657907797e-05, + "loss": 3.2704, "step": 55900 }, { "epoch": 0.19, - "learning_rate": 4.072094911434773e-05, - "loss": 3.3549, + "learning_rate": 4.073602297466302e-05, + "loss": 3.2718, "step": 56000 }, { "epoch": 0.19, - "learning_rate": 4.0704379380623354e-05, - "loss": 3.3556, + "learning_rate": 4.0719480158546354e-05, + "loss": 3.2788, "step": 56100 }, { "epoch": 0.19, - "learning_rate": 4.0687809646898975e-05, - "loss": 3.3489, + "learning_rate": 4.070293734242968e-05, + "loss": 3.2715, "step": 56200 }, { "epoch": 0.19, - "learning_rate": 4.06712399131746e-05, - "loss": 3.3473, + "learning_rate": 4.0686394526313e-05, + "loss": 3.2813, "step": 56300 }, { "epoch": 0.19, - "learning_rate": 4.065467017945022e-05, - "loss": 3.342, + "learning_rate": 4.0669851710196334e-05, + "loss": 3.2578, "step": 56400 }, { "epoch": 0.19, - "learning_rate": 4.063810044572584e-05, - "loss": 3.3423, + "learning_rate": 4.065330889407966e-05, + "loss": 3.2768, "step": 56500 }, { "epoch": 0.19, - "learning_rate": 4.062153071200146e-05, - "loss": 3.3532, + "learning_rate": 4.063676607796298e-05, + "loss": 3.2587, "step": 56600 }, { "epoch": 0.19, - "learning_rate": 4.0604960978277076e-05, - "loss": 3.336, + "learning_rate": 4.0620223261846315e-05, + "loss": 3.2674, "step": 56700 }, { "epoch": 0.19, - "learning_rate": 4.0588391244552704e-05, - "loss": 3.3547, + "learning_rate": 4.060368044572964e-05, + "loss": 3.281, "step": 56800 }, { "epoch": 0.19, - "learning_rate": 4.0571821510828326e-05, - "loss": 3.3524, + "learning_rate": 4.0587137629612964e-05, + "loss": 3.2645, "step": 56900 }, { "epoch": 0.19, - "learning_rate": 4.055525177710394e-05, - "loss": 3.3477, + "learning_rate": 4.0570594813496295e-05, + "loss": 3.2587, "step": 57000 }, { "epoch": 0.19, - "learning_rate": 4.053868204337957e-05, - "loss": 3.3553, + "learning_rate": 4.055405199737962e-05, + "loss": 3.2672, "step": 57100 }, { "epoch": 0.19, - "learning_rate": 4.052211230965518e-05, - "loss": 3.3433, + "learning_rate": 4.0537509181262944e-05, + "loss": 3.2611, "step": 57200 }, { "epoch": 0.19, - "learning_rate": 4.0505542575930805e-05, - "loss": 3.3486, + "learning_rate": 4.0520966365146275e-05, + "loss": 3.2823, "step": 57300 }, { "epoch": 0.19, - "learning_rate": 4.0488972842206426e-05, - "loss": 3.3543, + "learning_rate": 4.05044235490296e-05, + "loss": 3.2853, "step": 57400 }, { "epoch": 0.19, - "learning_rate": 4.047240310848205e-05, - "loss": 3.3474, + "learning_rate": 4.0487880732912924e-05, + "loss": 3.2697, "step": 57500 }, { "epoch": 0.19, - "learning_rate": 4.045583337475767e-05, - "loss": 3.3409, + "learning_rate": 4.0471337916796256e-05, + "loss": 3.2624, "step": 57600 }, { "epoch": 0.19, - "learning_rate": 4.043926364103329e-05, - "loss": 3.3515, + "learning_rate": 4.045479510067958e-05, + "loss": 3.2631, "step": 57700 }, { "epoch": 0.19, - "learning_rate": 4.042269390730891e-05, - "loss": 3.353, + "learning_rate": 4.0438252284562905e-05, + "loss": 3.2622, "step": 57800 }, { "epoch": 0.19, - "learning_rate": 4.0406124173584534e-05, - "loss": 3.3503, + "learning_rate": 4.0421709468446236e-05, + "loss": 3.2699, "step": 57900 }, { "epoch": 0.19, - "learning_rate": 4.0389554439860155e-05, - "loss": 3.3446, + "learning_rate": 4.040516665232956e-05, + "loss": 3.2851, "step": 58000 }, { "epoch": 0.19, - "learning_rate": 4.0372984706135777e-05, - "loss": 3.3571, + "learning_rate": 4.0388623836212885e-05, + "loss": 3.2694, "step": 58100 }, { "epoch": 0.19, - "learning_rate": 4.035641497241139e-05, - "loss": 3.3467, + "learning_rate": 4.0372081020096216e-05, + "loss": 3.2844, "step": 58200 }, { "epoch": 0.19, - "learning_rate": 4.033984523868702e-05, - "loss": 3.3667, + "learning_rate": 4.035553820397954e-05, + "loss": 3.2686, "step": 58300 }, { "epoch": 0.19, - "learning_rate": 4.032327550496264e-05, - "loss": 3.3492, + "learning_rate": 4.033899538786287e-05, + "loss": 3.2773, "step": 58400 }, { "epoch": 0.19, - "learning_rate": 4.0306705771238256e-05, - "loss": 3.3445, + "learning_rate": 4.03224525717462e-05, + "loss": 3.2559, "step": 58500 }, { "epoch": 0.19, - "learning_rate": 4.0290136037513884e-05, - "loss": 3.3567, + "learning_rate": 4.030590975562952e-05, + "loss": 3.2661, "step": 58600 }, { "epoch": 0.19, - "learning_rate": 4.02735663037895e-05, - "loss": 3.3497, + "learning_rate": 4.028936693951285e-05, + "loss": 3.2859, "step": 58700 }, { "epoch": 0.19, - "learning_rate": 4.025699657006512e-05, - "loss": 3.3386, + "learning_rate": 4.027282412339618e-05, + "loss": 3.2842, "step": 58800 }, { - "epoch": 0.2, - "learning_rate": 4.024042683634074e-05, - "loss": 3.3426, + "epoch": 0.19, + "learning_rate": 4.02562813072795e-05, + "loss": 3.2864, "step": 58900 }, { "epoch": 0.2, - "learning_rate": 4.022385710261636e-05, - "loss": 3.3341, + "learning_rate": 4.023973849116283e-05, + "loss": 3.2716, "step": 59000 }, { "epoch": 0.2, - "learning_rate": 4.0207287368891984e-05, - "loss": 3.3402, + "learning_rate": 4.022319567504616e-05, + "loss": 3.2941, "step": 59100 }, { "epoch": 0.2, - "learning_rate": 4.0190717635167606e-05, - "loss": 3.3455, + "learning_rate": 4.020665285892948e-05, + "loss": 3.2728, "step": 59200 }, { "epoch": 0.2, - "learning_rate": 4.017414790144323e-05, - "loss": 3.3532, + "learning_rate": 4.0190110042812813e-05, + "loss": 3.2651, "step": 59300 }, { "epoch": 0.2, - "learning_rate": 4.015757816771885e-05, - "loss": 3.3383, + "learning_rate": 4.017356722669614e-05, + "loss": 3.272, "step": 59400 }, { "epoch": 0.2, - "learning_rate": 4.014100843399447e-05, - "loss": 3.3416, + "learning_rate": 4.015702441057946e-05, + "loss": 3.2682, "step": 59500 }, { "epoch": 0.2, - "learning_rate": 4.012443870027009e-05, - "loss": 3.3453, + "learning_rate": 4.0140481594462794e-05, + "loss": 3.2834, "step": 59600 }, { "epoch": 0.2, - "learning_rate": 4.0107868966545707e-05, - "loss": 3.3432, + "learning_rate": 4.012393877834612e-05, + "loss": 3.2768, "step": 59700 }, { "epoch": 0.2, - "learning_rate": 4.009129923282133e-05, - "loss": 3.3476, + "learning_rate": 4.010739596222944e-05, + "loss": 3.2825, "step": 59800 }, { "epoch": 0.2, - "learning_rate": 4.007472949909695e-05, - "loss": 3.3514, + "learning_rate": 4.0090853146112774e-05, + "loss": 3.2888, "step": 59900 }, { "epoch": 0.2, - "learning_rate": 4.005815976537257e-05, - "loss": 3.3329, + "learning_rate": 4.00743103299961e-05, + "loss": 3.2852, "step": 60000 }, { "epoch": 0.2, - "learning_rate": 4.004159003164819e-05, - "loss": 3.3428, + "learning_rate": 4.005776751387942e-05, + "loss": 3.2733, "step": 60100 }, { "epoch": 0.2, - "learning_rate": 4.0025020297923814e-05, - "loss": 3.3456, + "learning_rate": 4.0041224697762755e-05, + "loss": 3.2694, "step": 60200 }, { "epoch": 0.2, - "learning_rate": 4.0008450564199435e-05, - "loss": 3.3509, + "learning_rate": 4.002468188164607e-05, + "loss": 3.2803, "step": 60300 }, { "epoch": 0.2, - "learning_rate": 3.999188083047506e-05, - "loss": 3.3497, + "learning_rate": 4.0008139065529404e-05, + "loss": 3.2738, "step": 60400 }, { "epoch": 0.2, - "learning_rate": 3.997531109675068e-05, - "loss": 3.3476, + "learning_rate": 3.9991596249412735e-05, + "loss": 3.2809, "step": 60500 }, { "epoch": 0.2, - "learning_rate": 3.99587413630263e-05, - "loss": 3.3376, + "learning_rate": 3.997505343329606e-05, + "loss": 3.2728, "step": 60600 }, { "epoch": 0.2, - "learning_rate": 3.9942171629301914e-05, - "loss": 3.3388, + "learning_rate": 3.9958510617179384e-05, + "loss": 3.283, "step": 60700 }, { "epoch": 0.2, - "learning_rate": 3.992560189557754e-05, - "loss": 3.3375, + "learning_rate": 3.9941967801062715e-05, + "loss": 3.2633, "step": 60800 }, { "epoch": 0.2, - "learning_rate": 3.990903216185316e-05, - "loss": 3.3422, + "learning_rate": 3.992542498494604e-05, + "loss": 3.2739, "step": 60900 }, { "epoch": 0.2, - "learning_rate": 3.989246242812878e-05, - "loss": 3.3523, + "learning_rate": 3.9908882168829364e-05, + "loss": 3.2639, "step": 61000 }, { "epoch": 0.2, - "learning_rate": 3.987589269440441e-05, - "loss": 3.3427, + "learning_rate": 3.9892339352712696e-05, + "loss": 3.2659, "step": 61100 }, { "epoch": 0.2, - "learning_rate": 3.985932296068002e-05, - "loss": 3.3457, + "learning_rate": 3.987579653659602e-05, + "loss": 3.2785, "step": 61200 }, { "epoch": 0.2, - "learning_rate": 3.984275322695564e-05, - "loss": 3.3377, + "learning_rate": 3.9859253720479345e-05, + "loss": 3.2706, "step": 61300 }, { "epoch": 0.2, - "learning_rate": 3.9826183493231265e-05, - "loss": 3.3534, + "learning_rate": 3.9842710904362676e-05, + "loss": 3.2732, "step": 61400 }, { "epoch": 0.2, - "learning_rate": 3.9809613759506886e-05, - "loss": 3.3395, + "learning_rate": 3.9826168088246e-05, + "loss": 3.2764, "step": 61500 }, { "epoch": 0.2, - "learning_rate": 3.979304402578251e-05, - "loss": 3.3397, + "learning_rate": 3.9809625272129325e-05, + "loss": 3.2644, "step": 61600 }, { "epoch": 0.2, - "learning_rate": 3.977647429205813e-05, - "loss": 3.3347, + "learning_rate": 3.9793082456012656e-05, + "loss": 3.2631, "step": 61700 }, { "epoch": 0.2, - "learning_rate": 3.975990455833375e-05, - "loss": 3.3465, + "learning_rate": 3.977653963989598e-05, + "loss": 3.2846, "step": 61800 }, { - "epoch": 0.21, - "learning_rate": 3.9743334824609365e-05, - "loss": 3.3446, + "epoch": 0.2, + "learning_rate": 3.9759996823779306e-05, + "loss": 3.2645, "step": 61900 }, { "epoch": 0.21, - "learning_rate": 3.9726765090884993e-05, - "loss": 3.328, + "learning_rate": 3.974345400766264e-05, + "loss": 3.2657, "step": 62000 }, { "epoch": 0.21, - "learning_rate": 3.9710195357160615e-05, - "loss": 3.3345, + "learning_rate": 3.972691119154596e-05, + "loss": 3.2847, "step": 62100 }, { "epoch": 0.21, - "learning_rate": 3.969362562343623e-05, - "loss": 3.3504, + "learning_rate": 3.9710368375429286e-05, + "loss": 3.2714, "step": 62200 }, { "epoch": 0.21, - "learning_rate": 3.967705588971186e-05, - "loss": 3.3307, + "learning_rate": 3.969382555931262e-05, + "loss": 3.2759, "step": 62300 }, { "epoch": 0.21, - "learning_rate": 3.966048615598747e-05, - "loss": 3.3406, + "learning_rate": 3.967728274319594e-05, + "loss": 3.2859, "step": 62400 }, { "epoch": 0.21, - "learning_rate": 3.9643916422263094e-05, - "loss": 3.3441, + "learning_rate": 3.966073992707927e-05, + "loss": 3.2691, "step": 62500 }, { "epoch": 0.21, - "learning_rate": 3.962734668853872e-05, - "loss": 3.3492, + "learning_rate": 3.96441971109626e-05, + "loss": 3.2632, "step": 62600 }, { "epoch": 0.21, - "learning_rate": 3.961077695481434e-05, - "loss": 3.3458, + "learning_rate": 3.962765429484592e-05, + "loss": 3.281, "step": 62700 }, { "epoch": 0.21, - "learning_rate": 3.959420722108996e-05, - "loss": 3.343, + "learning_rate": 3.9611111478729253e-05, + "loss": 3.2685, "step": 62800 }, { "epoch": 0.21, - "learning_rate": 3.957763748736558e-05, - "loss": 3.3499, + "learning_rate": 3.959456866261257e-05, + "loss": 3.2606, "step": 62900 }, { "epoch": 0.21, - "learning_rate": 3.95610677536412e-05, - "loss": 3.35, + "learning_rate": 3.95780258464959e-05, + "loss": 3.2632, "step": 63000 }, { "epoch": 0.21, - "learning_rate": 3.954449801991682e-05, - "loss": 3.33, + "learning_rate": 3.9561483030379234e-05, + "loss": 3.2696, "step": 63100 }, { "epoch": 0.21, - "learning_rate": 3.9527928286192444e-05, - "loss": 3.3379, + "learning_rate": 3.954494021426255e-05, + "loss": 3.2681, "step": 63200 }, { "epoch": 0.21, - "learning_rate": 3.9511358552468066e-05, - "loss": 3.3411, + "learning_rate": 3.952839739814588e-05, + "loss": 3.2706, "step": 63300 }, { "epoch": 0.21, - "learning_rate": 3.949478881874368e-05, - "loss": 3.3388, + "learning_rate": 3.9511854582029214e-05, + "loss": 3.2794, "step": 63400 }, { "epoch": 0.21, - "learning_rate": 3.947821908501931e-05, - "loss": 3.3367, + "learning_rate": 3.949531176591253e-05, + "loss": 3.246, "step": 63500 }, { "epoch": 0.21, - "learning_rate": 3.946164935129493e-05, - "loss": 3.3421, + "learning_rate": 3.947876894979586e-05, + "loss": 3.2586, "step": 63600 }, { "epoch": 0.21, - "learning_rate": 3.9445079617570545e-05, - "loss": 3.3445, + "learning_rate": 3.9462226133679195e-05, + "loss": 3.2695, "step": 63700 }, { "epoch": 0.21, - "learning_rate": 3.942850988384617e-05, - "loss": 3.3475, + "learning_rate": 3.944568331756251e-05, + "loss": 3.2706, "step": 63800 }, { "epoch": 0.21, - "learning_rate": 3.941194015012179e-05, - "loss": 3.3425, + "learning_rate": 3.9429140501445844e-05, + "loss": 3.2697, "step": 63900 }, { "epoch": 0.21, - "learning_rate": 3.939537041639741e-05, - "loss": 3.3314, + "learning_rate": 3.9412597685329175e-05, + "loss": 3.2794, "step": 64000 }, { "epoch": 0.21, - "learning_rate": 3.937880068267303e-05, - "loss": 3.349, + "learning_rate": 3.939605486921249e-05, + "loss": 3.2611, "step": 64100 }, { "epoch": 0.21, - "learning_rate": 3.936223094894865e-05, - "loss": 3.3572, + "learning_rate": 3.9379512053095824e-05, + "loss": 3.2624, "step": 64200 }, { "epoch": 0.21, - "learning_rate": 3.9345661215224274e-05, - "loss": 3.3469, + "learning_rate": 3.9362969236979155e-05, + "loss": 3.2737, "step": 64300 }, { "epoch": 0.21, - "learning_rate": 3.9329091481499895e-05, - "loss": 3.3494, + "learning_rate": 3.934642642086247e-05, + "loss": 3.2777, "step": 64400 }, { "epoch": 0.21, - "learning_rate": 3.9312521747775517e-05, - "loss": 3.3446, + "learning_rate": 3.9329883604745804e-05, + "loss": 3.2828, "step": 64500 }, { "epoch": 0.21, - "learning_rate": 3.929595201405114e-05, - "loss": 3.3409, + "learning_rate": 3.9313340788629136e-05, + "loss": 3.2732, "step": 64600 }, { "epoch": 0.21, - "learning_rate": 3.927938228032676e-05, - "loss": 3.3349, + "learning_rate": 3.929679797251246e-05, + "loss": 3.2743, "step": 64700 }, { "epoch": 0.21, - "learning_rate": 3.926281254660238e-05, - "loss": 3.3382, + "learning_rate": 3.9280255156395785e-05, + "loss": 3.2719, "step": 64800 }, { - "epoch": 0.22, - "learning_rate": 3.9246242812877996e-05, - "loss": 3.3367, + "epoch": 0.21, + "learning_rate": 3.9263712340279116e-05, + "loss": 3.271, "step": 64900 }, { "epoch": 0.22, - "learning_rate": 3.922967307915362e-05, - "loss": 3.3396, + "learning_rate": 3.924716952416244e-05, + "loss": 3.2755, "step": 65000 }, { "epoch": 0.22, - "learning_rate": 3.921310334542924e-05, - "loss": 3.3513, + "learning_rate": 3.9230626708045765e-05, + "loss": 3.2623, "step": 65100 }, { "epoch": 0.22, - "learning_rate": 3.919653361170486e-05, - "loss": 3.3427, + "learning_rate": 3.9214083891929097e-05, + "loss": 3.2473, "step": 65200 }, { "epoch": 0.22, - "learning_rate": 3.917996387798048e-05, - "loss": 3.3574, + "learning_rate": 3.919754107581242e-05, + "loss": 3.265, "step": 65300 }, { "epoch": 0.22, - "learning_rate": 3.91633941442561e-05, - "loss": 3.3354, + "learning_rate": 3.9180998259695746e-05, + "loss": 3.2639, "step": 65400 }, { "epoch": 0.22, - "learning_rate": 3.9146824410531724e-05, - "loss": 3.3383, + "learning_rate": 3.916445544357907e-05, + "loss": 3.2691, "step": 65500 }, { "epoch": 0.22, - "learning_rate": 3.9130254676807346e-05, - "loss": 3.3502, + "learning_rate": 3.91479126274624e-05, + "loss": 3.2797, "step": 65600 }, { "epoch": 0.22, - "learning_rate": 3.911368494308297e-05, - "loss": 3.333, + "learning_rate": 3.9131369811345726e-05, + "loss": 3.2732, "step": 65700 }, { "epoch": 0.22, - "learning_rate": 3.909711520935859e-05, - "loss": 3.328, + "learning_rate": 3.911482699522905e-05, + "loss": 3.2771, "step": 65800 }, { "epoch": 0.22, - "learning_rate": 3.9080545475634204e-05, - "loss": 3.3307, + "learning_rate": 3.909828417911238e-05, + "loss": 3.2604, "step": 65900 }, { "epoch": 0.22, - "learning_rate": 3.906397574190983e-05, - "loss": 3.3385, + "learning_rate": 3.9081741362995706e-05, + "loss": 3.2633, "step": 66000 }, { "epoch": 0.22, - "learning_rate": 3.9047406008185447e-05, - "loss": 3.3348, + "learning_rate": 3.906519854687903e-05, + "loss": 3.2683, "step": 66100 }, { "epoch": 0.22, - "learning_rate": 3.903083627446107e-05, - "loss": 3.3486, + "learning_rate": 3.904865573076236e-05, + "loss": 3.2574, "step": 66200 }, { "epoch": 0.22, - "learning_rate": 3.9014266540736696e-05, - "loss": 3.3428, + "learning_rate": 3.903211291464569e-05, + "loss": 3.2739, "step": 66300 }, { "epoch": 0.22, - "learning_rate": 3.899769680701231e-05, - "loss": 3.3343, + "learning_rate": 3.901557009852901e-05, + "loss": 3.2623, "step": 66400 }, { "epoch": 0.22, - "learning_rate": 3.898112707328793e-05, - "loss": 3.3468, + "learning_rate": 3.899902728241234e-05, + "loss": 3.2689, "step": 66500 }, { "epoch": 0.22, - "learning_rate": 3.8964557339563554e-05, - "loss": 3.3333, + "learning_rate": 3.898248446629567e-05, + "loss": 3.2705, "step": 66600 }, { "epoch": 0.22, - "learning_rate": 3.8947987605839175e-05, - "loss": 3.3464, + "learning_rate": 3.896594165017899e-05, + "loss": 3.2755, "step": 66700 }, { "epoch": 0.22, - "learning_rate": 3.89314178721148e-05, - "loss": 3.3235, + "learning_rate": 3.894939883406232e-05, + "loss": 3.2826, "step": 66800 }, { "epoch": 0.22, - "learning_rate": 3.891484813839042e-05, - "loss": 3.3444, + "learning_rate": 3.8932856017945654e-05, + "loss": 3.2878, "step": 66900 }, { "epoch": 0.22, - "learning_rate": 3.889827840466604e-05, - "loss": 3.3375, + "learning_rate": 3.891631320182897e-05, + "loss": 3.2755, "step": 67000 }, { "epoch": 0.22, - "learning_rate": 3.888170867094166e-05, - "loss": 3.3301, + "learning_rate": 3.88997703857123e-05, + "loss": 3.2695, "step": 67100 }, { "epoch": 0.22, - "learning_rate": 3.886513893721728e-05, - "loss": 3.3401, + "learning_rate": 3.8883227569595635e-05, + "loss": 3.2697, "step": 67200 }, { "epoch": 0.22, - "learning_rate": 3.8848569203492904e-05, - "loss": 3.3465, + "learning_rate": 3.886668475347895e-05, + "loss": 3.253, "step": 67300 }, { "epoch": 0.22, - "learning_rate": 3.883199946976852e-05, - "loss": 3.3431, + "learning_rate": 3.8850141937362284e-05, + "loss": 3.2783, "step": 67400 }, { "epoch": 0.22, - "learning_rate": 3.881542973604415e-05, - "loss": 3.3283, + "learning_rate": 3.8833599121245615e-05, + "loss": 3.2743, "step": 67500 }, { "epoch": 0.22, - "learning_rate": 3.879886000231976e-05, - "loss": 3.3269, + "learning_rate": 3.881705630512893e-05, + "loss": 3.2665, "step": 67600 }, { "epoch": 0.22, - "learning_rate": 3.878229026859538e-05, - "loss": 3.3296, + "learning_rate": 3.8800513489012264e-05, + "loss": 3.2634, "step": 67700 }, { "epoch": 0.22, - "learning_rate": 3.876572053487101e-05, - "loss": 3.3422, + "learning_rate": 3.8783970672895595e-05, + "loss": 3.2797, "step": 67800 }, { - "epoch": 0.23, - "learning_rate": 3.8749150801146626e-05, - "loss": 3.3324, + "epoch": 0.22, + "learning_rate": 3.876742785677891e-05, + "loss": 3.2621, "step": 67900 }, { - "epoch": 0.23, - "learning_rate": 3.873258106742225e-05, - "loss": 3.3241, + "epoch": 0.22, + "learning_rate": 3.8750885040662244e-05, + "loss": 3.2649, "step": 68000 }, { "epoch": 0.23, - "learning_rate": 3.871601133369787e-05, - "loss": 3.3203, + "learning_rate": 3.873434222454557e-05, + "loss": 3.2647, "step": 68100 }, { "epoch": 0.23, - "learning_rate": 3.869944159997349e-05, - "loss": 3.3279, + "learning_rate": 3.8717799408428894e-05, + "loss": 3.276, "step": 68200 }, { "epoch": 0.23, - "learning_rate": 3.868287186624911e-05, - "loss": 3.3448, + "learning_rate": 3.8701256592312225e-05, + "loss": 3.2668, "step": 68300 }, { "epoch": 0.23, - "learning_rate": 3.8666302132524733e-05, - "loss": 3.3417, + "learning_rate": 3.868471377619555e-05, + "loss": 3.2666, "step": 68400 }, { "epoch": 0.23, - "learning_rate": 3.8649732398800355e-05, - "loss": 3.3459, + "learning_rate": 3.8668170960078874e-05, + "loss": 3.2688, "step": 68500 }, { "epoch": 0.23, - "learning_rate": 3.863316266507597e-05, - "loss": 3.3312, + "learning_rate": 3.8651628143962205e-05, + "loss": 3.2889, "step": 68600 }, { "epoch": 0.23, - "learning_rate": 3.86165929313516e-05, - "loss": 3.3411, + "learning_rate": 3.863508532784553e-05, + "loss": 3.2787, "step": 68700 }, { "epoch": 0.23, - "learning_rate": 3.860002319762722e-05, - "loss": 3.3326, + "learning_rate": 3.8618542511728854e-05, + "loss": 3.2738, "step": 68800 }, { "epoch": 0.23, - "learning_rate": 3.8583453463902834e-05, - "loss": 3.341, + "learning_rate": 3.8601999695612186e-05, + "loss": 3.256, "step": 68900 }, { "epoch": 0.23, - "learning_rate": 3.856688373017846e-05, - "loss": 3.3319, + "learning_rate": 3.858545687949551e-05, + "loss": 3.2739, "step": 69000 }, { "epoch": 0.23, - "learning_rate": 3.855031399645408e-05, - "loss": 3.3394, + "learning_rate": 3.856891406337884e-05, + "loss": 3.2672, "step": 69100 }, { "epoch": 0.23, - "learning_rate": 3.85337442627297e-05, - "loss": 3.3334, + "learning_rate": 3.8552371247262166e-05, + "loss": 3.2561, "step": 69200 }, { "epoch": 0.23, - "learning_rate": 3.851717452900532e-05, - "loss": 3.3329, + "learning_rate": 3.853582843114549e-05, + "loss": 3.2633, "step": 69300 }, { "epoch": 0.23, - "learning_rate": 3.850060479528094e-05, - "loss": 3.328, + "learning_rate": 3.851928561502882e-05, + "loss": 3.2617, "step": 69400 }, { "epoch": 0.23, - "learning_rate": 3.848403506155656e-05, - "loss": 3.3484, + "learning_rate": 3.8502742798912146e-05, + "loss": 3.2652, "step": 69500 }, { "epoch": 0.23, - "learning_rate": 3.8467465327832184e-05, - "loss": 3.343, + "learning_rate": 3.848619998279547e-05, + "loss": 3.2603, "step": 69600 }, { "epoch": 0.23, - "learning_rate": 3.8450895594107806e-05, - "loss": 3.3432, + "learning_rate": 3.84696571666788e-05, + "loss": 3.2748, "step": 69700 }, { "epoch": 0.23, - "learning_rate": 3.843432586038343e-05, - "loss": 3.3366, + "learning_rate": 3.845311435056213e-05, + "loss": 3.2633, "step": 69800 }, { "epoch": 0.23, - "learning_rate": 3.841775612665905e-05, - "loss": 3.3426, + "learning_rate": 3.843657153444545e-05, + "loss": 3.2608, "step": 69900 }, { "epoch": 0.23, - "learning_rate": 3.840118639293467e-05, - "loss": 3.3544, + "learning_rate": 3.842002871832878e-05, + "loss": 3.2812, "step": 70000 }, { "epoch": 0.23, - "learning_rate": 3.8384616659210285e-05, - "loss": 3.3428, + "learning_rate": 3.840348590221211e-05, + "loss": 3.2775, "step": 70100 }, { "epoch": 0.23, - "learning_rate": 3.8368046925485906e-05, - "loss": 3.3457, + "learning_rate": 3.838694308609543e-05, + "loss": 3.2708, "step": 70200 }, { "epoch": 0.23, - "learning_rate": 3.8351477191761535e-05, - "loss": 3.3327, + "learning_rate": 3.837040026997876e-05, + "loss": 3.2529, "step": 70300 }, { "epoch": 0.23, - "learning_rate": 3.833490745803715e-05, - "loss": 3.338, + "learning_rate": 3.835385745386209e-05, + "loss": 3.2715, "step": 70400 }, { "epoch": 0.23, - "learning_rate": 3.831833772431277e-05, - "loss": 3.3289, + "learning_rate": 3.833731463774541e-05, + "loss": 3.2628, "step": 70500 }, { "epoch": 0.23, - "learning_rate": 3.830176799058839e-05, - "loss": 3.3466, + "learning_rate": 3.832077182162874e-05, + "loss": 3.2716, "step": 70600 }, { "epoch": 0.23, - "learning_rate": 3.8285198256864014e-05, - "loss": 3.3504, + "learning_rate": 3.830422900551207e-05, + "loss": 3.2707, "step": 70700 }, { "epoch": 0.23, - "learning_rate": 3.8268628523139635e-05, - "loss": 3.3219, + "learning_rate": 3.828768618939539e-05, + "loss": 3.2662, "step": 70800 }, { "epoch": 0.23, - "learning_rate": 3.8252058789415257e-05, - "loss": 3.333, + "learning_rate": 3.8271143373278724e-05, + "loss": 3.2659, "step": 70900 }, { - "epoch": 0.24, - "learning_rate": 3.823548905569088e-05, - "loss": 3.3356, + "epoch": 0.23, + "learning_rate": 3.825460055716205e-05, + "loss": 3.253, "step": 71000 }, { "epoch": 0.24, - "learning_rate": 3.821891932196649e-05, - "loss": 3.3308, + "learning_rate": 3.823805774104537e-05, + "loss": 3.2676, "step": 71100 }, { "epoch": 0.24, - "learning_rate": 3.820234958824212e-05, - "loss": 3.3325, + "learning_rate": 3.8221514924928704e-05, + "loss": 3.2615, "step": 71200 }, { "epoch": 0.24, - "learning_rate": 3.818577985451774e-05, - "loss": 3.3262, + "learning_rate": 3.820497210881203e-05, + "loss": 3.2812, "step": 71300 }, { "epoch": 0.24, - "learning_rate": 3.816921012079336e-05, - "loss": 3.3406, + "learning_rate": 3.818842929269535e-05, + "loss": 3.2681, "step": 71400 }, { "epoch": 0.24, - "learning_rate": 3.8152640387068985e-05, - "loss": 3.3236, + "learning_rate": 3.8171886476578684e-05, + "loss": 3.2791, "step": 71500 }, { "epoch": 0.24, - "learning_rate": 3.81360706533446e-05, - "loss": 3.3329, + "learning_rate": 3.815534366046201e-05, + "loss": 3.2443, "step": 71600 }, { "epoch": 0.24, - "learning_rate": 3.811950091962022e-05, - "loss": 3.3484, + "learning_rate": 3.8138800844345334e-05, + "loss": 3.2713, "step": 71700 }, { "epoch": 0.24, - "learning_rate": 3.810293118589584e-05, - "loss": 3.3376, + "learning_rate": 3.8122258028228665e-05, + "loss": 3.2527, "step": 71800 }, { "epoch": 0.24, - "learning_rate": 3.8086361452171464e-05, - "loss": 3.3271, + "learning_rate": 3.810571521211199e-05, + "loss": 3.2694, "step": 71900 }, { "epoch": 0.24, - "learning_rate": 3.8069791718447086e-05, - "loss": 3.335, + "learning_rate": 3.8089172395995314e-05, + "loss": 3.2703, "step": 72000 }, { "epoch": 0.24, - "learning_rate": 3.805322198472271e-05, - "loss": 3.3288, + "learning_rate": 3.8072629579878645e-05, + "loss": 3.2746, "step": 72100 }, { "epoch": 0.24, - "learning_rate": 3.803665225099833e-05, - "loss": 3.3366, + "learning_rate": 3.805608676376197e-05, + "loss": 3.2656, "step": 72200 }, { "epoch": 0.24, - "learning_rate": 3.802008251727395e-05, - "loss": 3.3329, + "learning_rate": 3.8039543947645294e-05, + "loss": 3.2604, "step": 72300 }, { "epoch": 0.24, - "learning_rate": 3.800351278354957e-05, - "loss": 3.3367, + "learning_rate": 3.8023001131528626e-05, + "loss": 3.2676, "step": 72400 }, { "epoch": 0.24, - "learning_rate": 3.798694304982519e-05, - "loss": 3.337, + "learning_rate": 3.800645831541195e-05, + "loss": 3.2658, "step": 72500 }, { "epoch": 0.24, - "learning_rate": 3.797037331610081e-05, - "loss": 3.3255, + "learning_rate": 3.7989915499295275e-05, + "loss": 3.2714, "step": 72600 }, { "epoch": 0.24, - "learning_rate": 3.7953803582376436e-05, - "loss": 3.3435, + "learning_rate": 3.7973372683178606e-05, + "loss": 3.2689, "step": 72700 }, { "epoch": 0.24, - "learning_rate": 3.793723384865205e-05, - "loss": 3.3137, + "learning_rate": 3.795682986706193e-05, + "loss": 3.275, "step": 72800 }, { "epoch": 0.24, - "learning_rate": 3.792066411492767e-05, - "loss": 3.3252, + "learning_rate": 3.7940287050945255e-05, + "loss": 3.2748, "step": 72900 }, { "epoch": 0.24, - "learning_rate": 3.79040943812033e-05, - "loss": 3.3268, + "learning_rate": 3.7923744234828586e-05, + "loss": 3.2614, "step": 73000 }, { "epoch": 0.24, - "learning_rate": 3.7887524647478915e-05, - "loss": 3.3427, + "learning_rate": 3.790720141871191e-05, + "loss": 3.2676, "step": 73100 }, { "epoch": 0.24, - "learning_rate": 3.787095491375454e-05, - "loss": 3.3262, + "learning_rate": 3.789065860259524e-05, + "loss": 3.2591, "step": 73200 }, { "epoch": 0.24, - "learning_rate": 3.785438518003016e-05, - "loss": 3.3233, + "learning_rate": 3.787411578647857e-05, + "loss": 3.2653, "step": 73300 }, { "epoch": 0.24, - "learning_rate": 3.783781544630578e-05, - "loss": 3.3429, + "learning_rate": 3.785757297036189e-05, + "loss": 3.2654, "step": 73400 }, { "epoch": 0.24, - "learning_rate": 3.78212457125814e-05, - "loss": 3.3269, + "learning_rate": 3.784103015424522e-05, + "loss": 3.2677, "step": 73500 }, { "epoch": 0.24, - "learning_rate": 3.780467597885702e-05, - "loss": 3.3382, + "learning_rate": 3.782448733812855e-05, + "loss": 3.2579, "step": 73600 }, { "epoch": 0.24, - "learning_rate": 3.7788106245132644e-05, - "loss": 3.3293, + "learning_rate": 3.780794452201187e-05, + "loss": 3.2727, "step": 73700 }, { "epoch": 0.24, - "learning_rate": 3.777153651140826e-05, - "loss": 3.327, + "learning_rate": 3.77914017058952e-05, + "loss": 3.2725, "step": 73800 }, { "epoch": 0.24, - "learning_rate": 3.775496677768389e-05, - "loss": 3.3373, + "learning_rate": 3.777485888977853e-05, + "loss": 3.2587, "step": 73900 }, { - "epoch": 0.25, - "learning_rate": 3.773839704395951e-05, - "loss": 3.3305, + "epoch": 0.24, + "learning_rate": 3.775831607366185e-05, + "loss": 3.2678, "step": 74000 }, { "epoch": 0.25, - "learning_rate": 3.772182731023512e-05, - "loss": 3.3295, + "learning_rate": 3.774177325754518e-05, + "loss": 3.2609, "step": 74100 }, { "epoch": 0.25, - "learning_rate": 3.770525757651075e-05, - "loss": 3.3324, + "learning_rate": 3.772523044142851e-05, + "loss": 3.2693, "step": 74200 }, { "epoch": 0.25, - "learning_rate": 3.7688687842786366e-05, - "loss": 3.3251, + "learning_rate": 3.770868762531183e-05, + "loss": 3.2631, "step": 74300 }, { "epoch": 0.25, - "learning_rate": 3.767211810906199e-05, - "loss": 3.3369, + "learning_rate": 3.7692144809195164e-05, + "loss": 3.271, "step": 74400 }, { "epoch": 0.25, - "learning_rate": 3.7655548375337616e-05, - "loss": 3.3277, + "learning_rate": 3.767560199307849e-05, + "loss": 3.2663, "step": 74500 }, { "epoch": 0.25, - "learning_rate": 3.763897864161323e-05, - "loss": 3.3361, + "learning_rate": 3.765905917696181e-05, + "loss": 3.2671, "step": 74600 }, { "epoch": 0.25, - "learning_rate": 3.762240890788885e-05, - "loss": 3.3335, + "learning_rate": 3.7642516360845144e-05, + "loss": 3.2716, "step": 74700 }, { "epoch": 0.25, - "learning_rate": 3.7605839174164473e-05, - "loss": 3.3243, + "learning_rate": 3.762597354472847e-05, + "loss": 3.2721, "step": 74800 }, { "epoch": 0.25, - "learning_rate": 3.7589269440440095e-05, - "loss": 3.3218, + "learning_rate": 3.760943072861179e-05, + "loss": 3.2634, "step": 74900 }, { "epoch": 0.25, - "learning_rate": 3.7572699706715716e-05, - "loss": 3.3278, + "learning_rate": 3.7592887912495125e-05, + "loss": 3.2766, "step": 75000 }, { "epoch": 0.25, - "learning_rate": 3.755612997299134e-05, - "loss": 3.3378, + "learning_rate": 3.757634509637845e-05, + "loss": 3.2582, "step": 75100 }, { "epoch": 0.25, - "learning_rate": 3.753956023926696e-05, - "loss": 3.3288, + "learning_rate": 3.7559802280261774e-05, + "loss": 3.265, "step": 75200 }, { "epoch": 0.25, - "learning_rate": 3.7522990505542574e-05, - "loss": 3.3317, + "learning_rate": 3.7543259464145105e-05, + "loss": 3.2603, "step": 75300 }, { "epoch": 0.25, - "learning_rate": 3.7506420771818196e-05, - "loss": 3.3267, + "learning_rate": 3.752671664802843e-05, + "loss": 3.2786, "step": 75400 }, { "epoch": 0.25, - "learning_rate": 3.7489851038093824e-05, - "loss": 3.334, + "learning_rate": 3.7510173831911754e-05, + "loss": 3.2608, "step": 75500 }, { "epoch": 0.25, - "learning_rate": 3.747328130436944e-05, - "loss": 3.3286, + "learning_rate": 3.7493631015795085e-05, + "loss": 3.2588, "step": 75600 }, { "epoch": 0.25, - "learning_rate": 3.745671157064506e-05, - "loss": 3.3298, + "learning_rate": 3.747708819967841e-05, + "loss": 3.2619, "step": 75700 }, { "epoch": 0.25, - "learning_rate": 3.744014183692068e-05, - "loss": 3.3359, + "learning_rate": 3.7460545383561734e-05, + "loss": 3.2648, "step": 75800 }, { "epoch": 0.25, - "learning_rate": 3.74235721031963e-05, - "loss": 3.3261, + "learning_rate": 3.7444002567445066e-05, + "loss": 3.2686, "step": 75900 }, { "epoch": 0.25, - "learning_rate": 3.7407002369471924e-05, - "loss": 3.3361, + "learning_rate": 3.742745975132839e-05, + "loss": 3.2578, "step": 76000 }, { "epoch": 0.25, - "learning_rate": 3.7390432635747546e-05, - "loss": 3.3195, + "learning_rate": 3.7410916935211715e-05, + "loss": 3.2858, "step": 76100 }, { "epoch": 0.25, - "learning_rate": 3.737386290202317e-05, - "loss": 3.322, + "learning_rate": 3.7394374119095046e-05, + "loss": 3.2529, "step": 76200 }, { "epoch": 0.25, - "learning_rate": 3.735729316829878e-05, - "loss": 3.3272, + "learning_rate": 3.737783130297837e-05, + "loss": 3.2609, "step": 76300 }, { "epoch": 0.25, - "learning_rate": 3.734072343457441e-05, - "loss": 3.3348, + "learning_rate": 3.7361288486861695e-05, + "loss": 3.2576, "step": 76400 }, { "epoch": 0.25, - "learning_rate": 3.732415370085003e-05, - "loss": 3.3327, + "learning_rate": 3.7344745670745026e-05, + "loss": 3.2568, "step": 76500 }, { "epoch": 0.25, - "learning_rate": 3.7307583967125646e-05, - "loss": 3.3307, + "learning_rate": 3.732820285462835e-05, + "loss": 3.2537, "step": 76600 }, { "epoch": 0.25, - "learning_rate": 3.7291014233401275e-05, - "loss": 3.3319, + "learning_rate": 3.7311660038511675e-05, + "loss": 3.2653, "step": 76700 }, { "epoch": 0.25, - "learning_rate": 3.727444449967689e-05, - "loss": 3.3302, + "learning_rate": 3.729511722239501e-05, + "loss": 3.2641, "step": 76800 }, { "epoch": 0.25, - "learning_rate": 3.725787476595251e-05, - "loss": 3.3277, + "learning_rate": 3.727857440627833e-05, + "loss": 3.2592, "step": 76900 }, { - "epoch": 0.26, - "learning_rate": 3.724130503222813e-05, - "loss": 3.3357, + "epoch": 0.25, + "learning_rate": 3.7262031590161656e-05, + "loss": 3.2692, "step": 77000 }, { "epoch": 0.26, - "learning_rate": 3.7224735298503754e-05, - "loss": 3.3164, + "learning_rate": 3.724548877404499e-05, + "loss": 3.2643, "step": 77100 }, { "epoch": 0.26, - "learning_rate": 3.7208165564779375e-05, - "loss": 3.3329, + "learning_rate": 3.722894595792831e-05, + "loss": 3.2477, "step": 77200 }, { "epoch": 0.26, - "learning_rate": 3.7191595831054997e-05, - "loss": 3.3211, + "learning_rate": 3.7212403141811636e-05, + "loss": 3.2741, "step": 77300 }, { "epoch": 0.26, - "learning_rate": 3.717502609733062e-05, - "loss": 3.3378, + "learning_rate": 3.719586032569497e-05, + "loss": 3.2808, "step": 77400 }, { "epoch": 0.26, - "learning_rate": 3.715845636360624e-05, - "loss": 3.3201, + "learning_rate": 3.717931750957829e-05, + "loss": 3.2613, "step": 77500 }, { "epoch": 0.26, - "learning_rate": 3.714188662988186e-05, - "loss": 3.3286, + "learning_rate": 3.7162774693461623e-05, + "loss": 3.2563, "step": 77600 }, { "epoch": 0.26, - "learning_rate": 3.712531689615748e-05, - "loss": 3.3313, + "learning_rate": 3.714623187734495e-05, + "loss": 3.2537, "step": 77700 }, { "epoch": 0.26, - "learning_rate": 3.71087471624331e-05, - "loss": 3.3267, + "learning_rate": 3.712968906122827e-05, + "loss": 3.2654, "step": 77800 }, { "epoch": 0.26, - "learning_rate": 3.7092177428708725e-05, - "loss": 3.3384, + "learning_rate": 3.7113146245111604e-05, + "loss": 3.2727, "step": 77900 }, { "epoch": 0.26, - "learning_rate": 3.707560769498434e-05, - "loss": 3.3209, + "learning_rate": 3.709660342899492e-05, + "loss": 3.2664, "step": 78000 }, { "epoch": 0.26, - "learning_rate": 3.705903796125996e-05, - "loss": 3.3313, + "learning_rate": 3.708006061287825e-05, + "loss": 3.2765, "step": 78100 }, { "epoch": 0.26, - "learning_rate": 3.704246822753559e-05, - "loss": 3.3279, + "learning_rate": 3.7063517796761584e-05, + "loss": 3.2702, "step": 78200 }, { "epoch": 0.26, - "learning_rate": 3.7025898493811205e-05, - "loss": 3.3292, + "learning_rate": 3.70469749806449e-05, + "loss": 3.2621, "step": 78300 }, { "epoch": 0.26, - "learning_rate": 3.7009328760086826e-05, - "loss": 3.3328, + "learning_rate": 3.703043216452823e-05, + "loss": 3.2577, "step": 78400 }, { "epoch": 0.26, - "learning_rate": 3.699275902636245e-05, - "loss": 3.3348, + "learning_rate": 3.7013889348411565e-05, + "loss": 3.2605, "step": 78500 }, { "epoch": 0.26, - "learning_rate": 3.697618929263807e-05, - "loss": 3.3162, + "learning_rate": 3.699734653229488e-05, + "loss": 3.27, "step": 78600 }, { "epoch": 0.26, - "learning_rate": 3.695961955891369e-05, - "loss": 3.3426, + "learning_rate": 3.6980803716178214e-05, + "loss": 3.2683, "step": 78700 }, { "epoch": 0.26, - "learning_rate": 3.694304982518931e-05, - "loss": 3.3336, + "learning_rate": 3.6964260900061545e-05, + "loss": 3.2716, "step": 78800 }, { "epoch": 0.26, - "learning_rate": 3.692648009146493e-05, - "loss": 3.3339, + "learning_rate": 3.694771808394486e-05, + "loss": 3.2534, "step": 78900 }, { "epoch": 0.26, - "learning_rate": 3.6909910357740555e-05, - "loss": 3.3258, + "learning_rate": 3.6931175267828194e-05, + "loss": 3.2561, "step": 79000 }, { "epoch": 0.26, - "learning_rate": 3.6893340624016176e-05, - "loss": 3.3128, + "learning_rate": 3.6914632451711525e-05, + "loss": 3.2701, "step": 79100 }, { "epoch": 0.26, - "learning_rate": 3.68767708902918e-05, - "loss": 3.3304, + "learning_rate": 3.689808963559484e-05, + "loss": 3.2741, "step": 79200 }, { "epoch": 0.26, - "learning_rate": 3.686020115656741e-05, - "loss": 3.3341, + "learning_rate": 3.6881546819478174e-05, + "loss": 3.2624, "step": 79300 }, { "epoch": 0.26, - "learning_rate": 3.684363142284304e-05, - "loss": 3.3222, + "learning_rate": 3.6865004003361506e-05, + "loss": 3.2695, "step": 79400 }, { "epoch": 0.26, - "learning_rate": 3.6827061689118655e-05, - "loss": 3.3322, + "learning_rate": 3.684846118724483e-05, + "loss": 3.2583, "step": 79500 }, { "epoch": 0.26, - "learning_rate": 3.681049195539428e-05, - "loss": 3.3301, + "learning_rate": 3.6831918371128155e-05, + "loss": 3.2651, "step": 79600 }, { "epoch": 0.26, - "learning_rate": 3.6793922221669905e-05, - "loss": 3.3371, + "learning_rate": 3.6815375555011486e-05, + "loss": 3.2663, "step": 79700 }, { "epoch": 0.26, - "learning_rate": 3.677735248794552e-05, - "loss": 3.3345, + "learning_rate": 3.679883273889481e-05, + "loss": 3.2697, "step": 79800 }, { "epoch": 0.26, - "learning_rate": 3.676078275422114e-05, - "loss": 3.3223, + "learning_rate": 3.6782289922778135e-05, + "loss": 3.2784, "step": 79900 }, { - "epoch": 0.27, - "learning_rate": 3.674421302049676e-05, - "loss": 3.3365, + "epoch": 0.26, + "learning_rate": 3.6765747106661466e-05, + "loss": 3.2727, "step": 80000 }, { "epoch": 0.27, - "learning_rate": 3.6727643286772384e-05, - "loss": 3.3213, + "learning_rate": 3.674920429054479e-05, + "loss": 3.2633, "step": 80100 }, { "epoch": 0.27, - "learning_rate": 3.6711073553048006e-05, - "loss": 3.321, + "learning_rate": 3.6732661474428116e-05, + "loss": 3.2647, "step": 80200 }, { "epoch": 0.27, - "learning_rate": 3.669450381932363e-05, - "loss": 3.3389, + "learning_rate": 3.671611865831145e-05, + "loss": 3.2597, "step": 80300 }, { "epoch": 0.27, - "learning_rate": 3.667793408559925e-05, - "loss": 3.3182, + "learning_rate": 3.669957584219477e-05, + "loss": 3.2713, "step": 80400 }, { "epoch": 0.27, - "learning_rate": 3.666136435187486e-05, - "loss": 3.3248, + "learning_rate": 3.6683033026078096e-05, + "loss": 3.2642, "step": 80500 }, { "epoch": 0.27, - "learning_rate": 3.6644794618150485e-05, - "loss": 3.3301, + "learning_rate": 3.666649020996143e-05, + "loss": 3.2745, "step": 80600 }, { "epoch": 0.27, - "learning_rate": 3.662822488442611e-05, - "loss": 3.3352, + "learning_rate": 3.664994739384475e-05, + "loss": 3.2671, "step": 80700 }, { "epoch": 0.27, - "learning_rate": 3.661165515070173e-05, - "loss": 3.32, + "learning_rate": 3.6633404577728076e-05, + "loss": 3.2692, "step": 80800 }, { "epoch": 0.27, - "learning_rate": 3.659508541697735e-05, - "loss": 3.3307, + "learning_rate": 3.66168617616114e-05, + "loss": 3.2746, "step": 80900 }, { "epoch": 0.27, - "learning_rate": 3.657851568325297e-05, - "loss": 3.3326, + "learning_rate": 3.660031894549473e-05, + "loss": 3.273, "step": 81000 }, { "epoch": 0.27, - "learning_rate": 3.656194594952859e-05, - "loss": 3.3224, + "learning_rate": 3.658377612937806e-05, + "loss": 3.2685, "step": 81100 }, { "epoch": 0.27, - "learning_rate": 3.6545376215804213e-05, - "loss": 3.332, + "learning_rate": 3.656723331326138e-05, + "loss": 3.2648, "step": 81200 }, { "epoch": 0.27, - "learning_rate": 3.6528806482079835e-05, - "loss": 3.327, + "learning_rate": 3.655069049714471e-05, + "loss": 3.2646, "step": 81300 }, { "epoch": 0.27, - "learning_rate": 3.6512236748355456e-05, - "loss": 3.324, + "learning_rate": 3.653414768102804e-05, + "loss": 3.267, "step": 81400 }, { "epoch": 0.27, - "learning_rate": 3.649566701463107e-05, - "loss": 3.3348, + "learning_rate": 3.651760486491136e-05, + "loss": 3.2648, "step": 81500 }, { "epoch": 0.27, - "learning_rate": 3.64790972809067e-05, - "loss": 3.3276, + "learning_rate": 3.650106204879469e-05, + "loss": 3.2617, "step": 81600 }, { "epoch": 0.27, - "learning_rate": 3.646252754718232e-05, - "loss": 3.3214, + "learning_rate": 3.6484519232678024e-05, + "loss": 3.2792, "step": 81700 }, { "epoch": 0.27, - "learning_rate": 3.6445957813457936e-05, - "loss": 3.3261, + "learning_rate": 3.646797641656134e-05, + "loss": 3.2633, "step": 81800 }, { "epoch": 0.27, - "learning_rate": 3.6429388079733564e-05, - "loss": 3.3127, + "learning_rate": 3.645143360044467e-05, + "loss": 3.2639, "step": 81900 }, { "epoch": 0.27, - "learning_rate": 3.641281834600918e-05, - "loss": 3.336, + "learning_rate": 3.6434890784328005e-05, + "loss": 3.2649, "step": 82000 }, { "epoch": 0.27, - "learning_rate": 3.63962486122848e-05, - "loss": 3.3286, + "learning_rate": 3.641834796821132e-05, + "loss": 3.2688, "step": 82100 }, { "epoch": 0.27, - "learning_rate": 3.637967887856043e-05, - "loss": 3.3243, + "learning_rate": 3.6401805152094654e-05, + "loss": 3.2652, "step": 82200 }, { "epoch": 0.27, - "learning_rate": 3.636310914483604e-05, - "loss": 3.3281, + "learning_rate": 3.6385262335977985e-05, + "loss": 3.2708, "step": 82300 }, { "epoch": 0.27, - "learning_rate": 3.6346539411111664e-05, - "loss": 3.3248, + "learning_rate": 3.63687195198613e-05, + "loss": 3.2851, "step": 82400 }, { "epoch": 0.27, - "learning_rate": 3.6329969677387286e-05, - "loss": 3.3247, + "learning_rate": 3.6352176703744634e-05, + "loss": 3.2653, "step": 82500 }, { "epoch": 0.27, - "learning_rate": 3.631339994366291e-05, - "loss": 3.3305, + "learning_rate": 3.6335633887627965e-05, + "loss": 3.2717, "step": 82600 }, { "epoch": 0.27, - "learning_rate": 3.629683020993853e-05, - "loss": 3.3295, + "learning_rate": 3.631909107151128e-05, + "loss": 3.2662, "step": 82700 }, { "epoch": 0.27, - "learning_rate": 3.628026047621415e-05, - "loss": 3.3252, + "learning_rate": 3.6302548255394614e-05, + "loss": 3.2613, "step": 82800 }, { "epoch": 0.27, - "learning_rate": 3.626369074248977e-05, - "loss": 3.3288, + "learning_rate": 3.6286005439277946e-05, + "loss": 3.2617, "step": 82900 }, { - "epoch": 0.28, - "learning_rate": 3.6247121008765386e-05, - "loss": 3.3223, + "epoch": 0.27, + "learning_rate": 3.6269462623161263e-05, + "loss": 3.2712, "step": 83000 }, { - "epoch": 0.28, - "learning_rate": 3.6230551275041015e-05, - "loss": 3.3214, + "epoch": 0.27, + "learning_rate": 3.6252919807044595e-05, + "loss": 3.2493, "step": 83100 }, { "epoch": 0.28, - "learning_rate": 3.6213981541316636e-05, - "loss": 3.3156, + "learning_rate": 3.6236376990927926e-05, + "loss": 3.2509, "step": 83200 }, { "epoch": 0.28, - "learning_rate": 3.619741180759225e-05, - "loss": 3.3288, + "learning_rate": 3.6219834174811244e-05, + "loss": 3.2757, "step": 83300 }, { "epoch": 0.28, - "learning_rate": 3.618084207386788e-05, - "loss": 3.3246, + "learning_rate": 3.6203291358694575e-05, + "loss": 3.2585, "step": 83400 }, { "epoch": 0.28, - "learning_rate": 3.6164272340143494e-05, - "loss": 3.321, + "learning_rate": 3.61867485425779e-05, + "loss": 3.2671, "step": 83500 }, { "epoch": 0.28, - "learning_rate": 3.6147702606419115e-05, - "loss": 3.3119, + "learning_rate": 3.6170205726461224e-05, + "loss": 3.2674, "step": 83600 }, { "epoch": 0.28, - "learning_rate": 3.613113287269474e-05, - "loss": 3.3268, + "learning_rate": 3.6153662910344556e-05, + "loss": 3.2688, "step": 83700 }, { "epoch": 0.28, - "learning_rate": 3.611456313897036e-05, - "loss": 3.3184, + "learning_rate": 3.613712009422788e-05, + "loss": 3.2623, "step": 83800 }, { "epoch": 0.28, - "learning_rate": 3.609799340524598e-05, - "loss": 3.3316, + "learning_rate": 3.612057727811121e-05, + "loss": 3.2566, "step": 83900 }, { "epoch": 0.28, - "learning_rate": 3.60814236715216e-05, - "loss": 3.3364, + "learning_rate": 3.6104034461994536e-05, + "loss": 3.2667, "step": 84000 }, { "epoch": 0.28, - "learning_rate": 3.606485393779722e-05, - "loss": 3.3271, + "learning_rate": 3.608749164587786e-05, + "loss": 3.2525, "step": 84100 }, { "epoch": 0.28, - "learning_rate": 3.6048284204072844e-05, - "loss": 3.3219, + "learning_rate": 3.607094882976119e-05, + "loss": 3.2731, "step": 84200 }, { "epoch": 0.28, - "learning_rate": 3.6031714470348465e-05, - "loss": 3.3314, + "learning_rate": 3.6054406013644516e-05, + "loss": 3.2526, "step": 84300 }, { "epoch": 0.28, - "learning_rate": 3.601514473662409e-05, - "loss": 3.3187, + "learning_rate": 3.603786319752784e-05, + "loss": 3.2642, "step": 84400 }, { "epoch": 0.28, - "learning_rate": 3.59985750028997e-05, - "loss": 3.3342, + "learning_rate": 3.602132038141117e-05, + "loss": 3.2553, "step": 84500 }, { "epoch": 0.28, - "learning_rate": 3.598200526917533e-05, - "loss": 3.3324, + "learning_rate": 3.60047775652945e-05, + "loss": 3.2642, "step": 84600 }, { "epoch": 0.28, - "learning_rate": 3.5965435535450945e-05, - "loss": 3.3324, + "learning_rate": 3.598823474917782e-05, + "loss": 3.2649, "step": 84700 }, { "epoch": 0.28, - "learning_rate": 3.5948865801726566e-05, - "loss": 3.3141, + "learning_rate": 3.597169193306115e-05, + "loss": 3.2608, "step": 84800 }, { "epoch": 0.28, - "learning_rate": 3.5932296068002194e-05, - "loss": 3.3204, + "learning_rate": 3.595514911694448e-05, + "loss": 3.2602, "step": 84900 }, { "epoch": 0.28, - "learning_rate": 3.591572633427781e-05, - "loss": 3.3396, + "learning_rate": 3.59386063008278e-05, + "loss": 3.2572, "step": 85000 }, { "epoch": 0.28, - "learning_rate": 3.589915660055343e-05, - "loss": 3.3219, + "learning_rate": 3.592206348471113e-05, + "loss": 3.2625, "step": 85100 }, { "epoch": 0.28, - "learning_rate": 3.588258686682905e-05, - "loss": 3.3259, + "learning_rate": 3.590552066859446e-05, + "loss": 3.2632, "step": 85200 }, { "epoch": 0.28, - "learning_rate": 3.586601713310467e-05, - "loss": 3.3202, + "learning_rate": 3.588897785247778e-05, + "loss": 3.2539, "step": 85300 }, { "epoch": 0.28, - "learning_rate": 3.5849447399380295e-05, - "loss": 3.3128, + "learning_rate": 3.587243503636111e-05, + "loss": 3.2657, "step": 85400 }, { "epoch": 0.28, - "learning_rate": 3.5832877665655916e-05, - "loss": 3.3232, + "learning_rate": 3.585589222024444e-05, + "loss": 3.2631, "step": 85500 }, { "epoch": 0.28, - "learning_rate": 3.581630793193154e-05, - "loss": 3.3121, + "learning_rate": 3.583934940412776e-05, + "loss": 3.2696, "step": 85600 }, { "epoch": 0.28, - "learning_rate": 3.579973819820715e-05, - "loss": 3.3264, + "learning_rate": 3.5822806588011094e-05, + "loss": 3.2661, "step": 85700 }, { "epoch": 0.28, - "learning_rate": 3.5783168464482774e-05, - "loss": 3.327, + "learning_rate": 3.5806263771894425e-05, + "loss": 3.2719, "step": 85800 }, { "epoch": 0.28, - "learning_rate": 3.57665987307584e-05, - "loss": 3.3227, + "learning_rate": 3.578972095577774e-05, + "loss": 3.2622, "step": 85900 }, { "epoch": 0.28, - "learning_rate": 3.575002899703402e-05, - "loss": 3.3195, + "learning_rate": 3.5773178139661074e-05, + "loss": 3.2554, "step": 86000 }, { - "epoch": 0.29, - "learning_rate": 3.573345926330964e-05, - "loss": 3.3264, + "epoch": 0.28, + "learning_rate": 3.57566353235444e-05, + "loss": 3.2551, "step": 86100 }, { "epoch": 0.29, - "learning_rate": 3.571688952958526e-05, - "loss": 3.3231, + "learning_rate": 3.574009250742772e-05, + "loss": 3.2581, "step": 86200 }, { "epoch": 0.29, - "learning_rate": 3.570031979586088e-05, - "loss": 3.3213, + "learning_rate": 3.5723549691311054e-05, + "loss": 3.2558, "step": 86300 }, { "epoch": 0.29, - "learning_rate": 3.56837500621365e-05, - "loss": 3.321, + "learning_rate": 3.570700687519438e-05, + "loss": 3.266, "step": 86400 }, { "epoch": 0.29, - "learning_rate": 3.5667180328412124e-05, - "loss": 3.316, + "learning_rate": 3.5690464059077703e-05, + "loss": 3.2681, "step": 86500 }, { "epoch": 0.29, - "learning_rate": 3.5650610594687746e-05, - "loss": 3.3286, + "learning_rate": 3.5673921242961035e-05, + "loss": 3.2486, "step": 86600 }, { "epoch": 0.29, - "learning_rate": 3.563404086096336e-05, - "loss": 3.3299, + "learning_rate": 3.565737842684436e-05, + "loss": 3.2682, "step": 86700 }, { "epoch": 0.29, - "learning_rate": 3.561747112723899e-05, - "loss": 3.3066, + "learning_rate": 3.5640835610727684e-05, + "loss": 3.2613, "step": 86800 }, { "epoch": 0.29, - "learning_rate": 3.560090139351461e-05, - "loss": 3.3381, + "learning_rate": 3.5624292794611015e-05, + "loss": 3.2576, "step": 86900 }, { "epoch": 0.29, - "learning_rate": 3.5584331659790225e-05, - "loss": 3.322, + "learning_rate": 3.560774997849434e-05, + "loss": 3.2673, "step": 87000 }, { "epoch": 0.29, - "learning_rate": 3.556776192606585e-05, - "loss": 3.3231, + "learning_rate": 3.5591207162377664e-05, + "loss": 3.2628, "step": 87100 }, { "epoch": 0.29, - "learning_rate": 3.555119219234147e-05, - "loss": 3.3151, + "learning_rate": 3.5574664346260996e-05, + "loss": 3.2512, "step": 87200 }, { "epoch": 0.29, - "learning_rate": 3.553462245861709e-05, - "loss": 3.3262, + "learning_rate": 3.555812153014432e-05, + "loss": 3.2615, "step": 87300 }, { "epoch": 0.29, - "learning_rate": 3.551805272489272e-05, - "loss": 3.3438, + "learning_rate": 3.5541578714027645e-05, + "loss": 3.26, "step": 87400 }, { "epoch": 0.29, - "learning_rate": 3.550148299116833e-05, - "loss": 3.3167, + "learning_rate": 3.5525035897910976e-05, + "loss": 3.2657, "step": 87500 }, { "epoch": 0.29, - "learning_rate": 3.5484913257443954e-05, - "loss": 3.3166, + "learning_rate": 3.55084930817943e-05, + "loss": 3.2586, "step": 87600 }, { "epoch": 0.29, - "learning_rate": 3.5468343523719575e-05, - "loss": 3.331, + "learning_rate": 3.5491950265677625e-05, + "loss": 3.2655, "step": 87700 }, { "epoch": 0.29, - "learning_rate": 3.5451773789995196e-05, - "loss": 3.3276, + "learning_rate": 3.5475407449560956e-05, + "loss": 3.2617, "step": 87800 }, { "epoch": 0.29, - "learning_rate": 3.543520405627082e-05, - "loss": 3.3279, + "learning_rate": 3.545886463344428e-05, + "loss": 3.2542, "step": 87900 }, { "epoch": 0.29, - "learning_rate": 3.541863432254644e-05, - "loss": 3.3187, + "learning_rate": 3.544232181732761e-05, + "loss": 3.2411, "step": 88000 }, { "epoch": 0.29, - "learning_rate": 3.540206458882206e-05, - "loss": 3.3198, + "learning_rate": 3.542577900121094e-05, + "loss": 3.2702, "step": 88100 }, { "epoch": 0.29, - "learning_rate": 3.5385494855097676e-05, - "loss": 3.3261, + "learning_rate": 3.540923618509426e-05, + "loss": 3.2572, "step": 88200 }, { "epoch": 0.29, - "learning_rate": 3.5368925121373304e-05, - "loss": 3.3199, + "learning_rate": 3.539269336897759e-05, + "loss": 3.2618, "step": 88300 }, { "epoch": 0.29, - "learning_rate": 3.5352355387648925e-05, - "loss": 3.3075, + "learning_rate": 3.537615055286092e-05, + "loss": 3.2691, "step": 88400 }, { "epoch": 0.29, - "learning_rate": 3.533578565392454e-05, - "loss": 3.3158, + "learning_rate": 3.535960773674424e-05, + "loss": 3.2635, "step": 88500 }, { "epoch": 0.29, - "learning_rate": 3.531921592020017e-05, - "loss": 3.3367, + "learning_rate": 3.534306492062757e-05, + "loss": 3.2694, "step": 88600 }, { "epoch": 0.29, - "learning_rate": 3.530264618647578e-05, - "loss": 3.3163, + "learning_rate": 3.53265221045109e-05, + "loss": 3.2622, "step": 88700 }, { "epoch": 0.29, - "learning_rate": 3.5286076452751404e-05, - "loss": 3.324, + "learning_rate": 3.530997928839422e-05, + "loss": 3.2523, "step": 88800 }, { "epoch": 0.29, - "learning_rate": 3.5269506719027026e-05, - "loss": 3.3144, + "learning_rate": 3.529343647227755e-05, + "loss": 3.275, "step": 88900 }, { "epoch": 0.29, - "learning_rate": 3.525293698530265e-05, - "loss": 3.3271, + "learning_rate": 3.527689365616088e-05, + "loss": 3.2564, "step": 89000 }, { - "epoch": 0.3, - "learning_rate": 3.523636725157827e-05, - "loss": 3.3152, + "epoch": 0.29, + "learning_rate": 3.52603508400442e-05, + "loss": 3.263, "step": 89100 }, { "epoch": 0.3, - "learning_rate": 3.521979751785389e-05, - "loss": 3.3138, + "learning_rate": 3.5243808023927534e-05, + "loss": 3.2685, "step": 89200 }, { "epoch": 0.3, - "learning_rate": 3.520322778412951e-05, - "loss": 3.3274, + "learning_rate": 3.522726520781086e-05, + "loss": 3.2586, "step": 89300 }, { "epoch": 0.3, - "learning_rate": 3.518665805040513e-05, - "loss": 3.3101, + "learning_rate": 3.521072239169418e-05, + "loss": 3.25, "step": 89400 }, { "epoch": 0.3, - "learning_rate": 3.5170088316680755e-05, - "loss": 3.3247, + "learning_rate": 3.5194179575577514e-05, + "loss": 3.2568, "step": 89500 }, { "epoch": 0.3, - "learning_rate": 3.5153518582956376e-05, - "loss": 3.313, + "learning_rate": 3.517763675946084e-05, + "loss": 3.2598, "step": 89600 }, { "epoch": 0.3, - "learning_rate": 3.513694884923199e-05, - "loss": 3.3177, + "learning_rate": 3.516109394334416e-05, + "loss": 3.2488, "step": 89700 }, { "epoch": 0.3, - "learning_rate": 3.512037911550762e-05, - "loss": 3.3164, + "learning_rate": 3.5144551127227494e-05, + "loss": 3.2653, "step": 89800 }, { "epoch": 0.3, - "learning_rate": 3.5103809381783234e-05, - "loss": 3.3345, + "learning_rate": 3.512800831111082e-05, + "loss": 3.2596, "step": 89900 }, { "epoch": 0.3, - "learning_rate": 3.5087239648058855e-05, - "loss": 3.33, + "learning_rate": 3.5111465494994144e-05, + "loss": 3.2585, "step": 90000 }, { "epoch": 0.3, - "learning_rate": 3.5070669914334483e-05, - "loss": 3.3201, + "learning_rate": 3.5094922678877475e-05, + "loss": 3.2571, "step": 90100 }, { "epoch": 0.3, - "learning_rate": 3.50541001806101e-05, - "loss": 3.3233, + "learning_rate": 3.50783798627608e-05, + "loss": 3.2586, "step": 90200 }, { "epoch": 0.3, - "learning_rate": 3.503753044688572e-05, - "loss": 3.3317, + "learning_rate": 3.5061837046644124e-05, + "loss": 3.262, "step": 90300 }, { "epoch": 0.3, - "learning_rate": 3.502096071316134e-05, - "loss": 3.3202, + "learning_rate": 3.5045294230527455e-05, + "loss": 3.2643, "step": 90400 }, { "epoch": 0.3, - "learning_rate": 3.500439097943696e-05, - "loss": 3.313, + "learning_rate": 3.502875141441078e-05, + "loss": 3.257, "step": 90500 }, { "epoch": 0.3, - "learning_rate": 3.4987821245712584e-05, - "loss": 3.3079, + "learning_rate": 3.5012208598294104e-05, + "loss": 3.2634, "step": 90600 }, { "epoch": 0.3, - "learning_rate": 3.4971251511988205e-05, - "loss": 3.3235, + "learning_rate": 3.4995665782177436e-05, + "loss": 3.2642, "step": 90700 }, { "epoch": 0.3, - "learning_rate": 3.495468177826383e-05, - "loss": 3.3296, + "learning_rate": 3.497912296606076e-05, + "loss": 3.2506, "step": 90800 }, { "epoch": 0.3, - "learning_rate": 3.493811204453944e-05, - "loss": 3.3169, + "learning_rate": 3.4962580149944085e-05, + "loss": 3.2665, "step": 90900 }, { "epoch": 0.3, - "learning_rate": 3.492154231081506e-05, - "loss": 3.3343, + "learning_rate": 3.4946037333827416e-05, + "loss": 3.2704, "step": 91000 }, { "epoch": 0.3, - "learning_rate": 3.490497257709069e-05, - "loss": 3.3168, + "learning_rate": 3.492949451771074e-05, + "loss": 3.2664, "step": 91100 }, { "epoch": 0.3, - "learning_rate": 3.4888402843366306e-05, - "loss": 3.3215, + "learning_rate": 3.4912951701594065e-05, + "loss": 3.2599, "step": 91200 }, { "epoch": 0.3, - "learning_rate": 3.487183310964193e-05, - "loss": 3.3127, + "learning_rate": 3.4896408885477396e-05, + "loss": 3.2736, "step": 91300 }, { "epoch": 0.3, - "learning_rate": 3.485526337591755e-05, - "loss": 3.3259, + "learning_rate": 3.487986606936072e-05, + "loss": 3.2523, "step": 91400 }, { "epoch": 0.3, - "learning_rate": 3.483869364219317e-05, - "loss": 3.3221, + "learning_rate": 3.4863323253244045e-05, + "loss": 3.2675, "step": 91500 }, { "epoch": 0.3, - "learning_rate": 3.482212390846879e-05, - "loss": 3.3094, + "learning_rate": 3.484678043712738e-05, + "loss": 3.2675, "step": 91600 }, { "epoch": 0.3, - "learning_rate": 3.480555417474441e-05, - "loss": 3.3338, + "learning_rate": 3.48302376210107e-05, + "loss": 3.2569, "step": 91700 }, { "epoch": 0.3, - "learning_rate": 3.4788984441020035e-05, - "loss": 3.3133, + "learning_rate": 3.4813694804894026e-05, + "loss": 3.2536, "step": 91800 }, { "epoch": 0.3, - "learning_rate": 3.4772414707295656e-05, - "loss": 3.3013, + "learning_rate": 3.479715198877736e-05, + "loss": 3.2658, "step": 91900 }, { "epoch": 0.3, - "learning_rate": 3.475584497357128e-05, - "loss": 3.3232, + "learning_rate": 3.478060917266068e-05, + "loss": 3.2642, "step": 92000 }, { - "epoch": 0.31, - "learning_rate": 3.47392752398469e-05, - "loss": 3.3117, + "epoch": 0.3, + "learning_rate": 3.4764066356544006e-05, + "loss": 3.2714, "step": 92100 }, { "epoch": 0.31, - "learning_rate": 3.4722705506122514e-05, - "loss": 3.3029, + "learning_rate": 3.474752354042734e-05, + "loss": 3.2686, "step": 92200 }, { "epoch": 0.31, - "learning_rate": 3.470613577239814e-05, - "loss": 3.3109, + "learning_rate": 3.473098072431066e-05, + "loss": 3.2464, "step": 92300 }, { "epoch": 0.31, - "learning_rate": 3.468956603867376e-05, - "loss": 3.3227, + "learning_rate": 3.471443790819399e-05, + "loss": 3.2589, "step": 92400 }, { "epoch": 0.31, - "learning_rate": 3.467299630494938e-05, - "loss": 3.3202, + "learning_rate": 3.469789509207732e-05, + "loss": 3.2658, "step": 92500 }, { "epoch": 0.31, - "learning_rate": 3.4656426571225007e-05, - "loss": 3.3221, + "learning_rate": 3.468135227596064e-05, + "loss": 3.2554, "step": 92600 }, { "epoch": 0.31, - "learning_rate": 3.463985683750062e-05, - "loss": 3.3167, + "learning_rate": 3.4664809459843974e-05, + "loss": 3.2661, "step": 92700 }, { "epoch": 0.31, - "learning_rate": 3.462328710377624e-05, - "loss": 3.3083, + "learning_rate": 3.46482666437273e-05, + "loss": 3.2749, "step": 92800 }, { "epoch": 0.31, - "learning_rate": 3.4606717370051864e-05, - "loss": 3.3309, + "learning_rate": 3.463172382761062e-05, + "loss": 3.275, "step": 92900 }, { "epoch": 0.31, - "learning_rate": 3.4590147636327486e-05, - "loss": 3.3098, + "learning_rate": 3.4615181011493954e-05, + "loss": 3.2656, "step": 93000 }, { "epoch": 0.31, - "learning_rate": 3.457357790260311e-05, - "loss": 3.3244, + "learning_rate": 3.459863819537728e-05, + "loss": 3.254, "step": 93100 }, { "epoch": 0.31, - "learning_rate": 3.455700816887873e-05, - "loss": 3.3074, + "learning_rate": 3.45820953792606e-05, + "loss": 3.2676, "step": 93200 }, { "epoch": 0.31, - "learning_rate": 3.454043843515435e-05, - "loss": 3.3182, + "learning_rate": 3.4565552563143934e-05, + "loss": 3.2606, "step": 93300 }, { "epoch": 0.31, - "learning_rate": 3.4523868701429965e-05, - "loss": 3.3184, + "learning_rate": 3.454900974702725e-05, + "loss": 3.2555, "step": 93400 }, { "epoch": 0.31, - "learning_rate": 3.450729896770559e-05, - "loss": 3.3168, + "learning_rate": 3.4532466930910584e-05, + "loss": 3.2576, "step": 93500 }, { "epoch": 0.31, - "learning_rate": 3.4490729233981214e-05, - "loss": 3.3325, + "learning_rate": 3.4515924114793915e-05, + "loss": 3.2762, "step": 93600 }, { "epoch": 0.31, - "learning_rate": 3.447415950025683e-05, - "loss": 3.3161, + "learning_rate": 3.449938129867723e-05, + "loss": 3.258, "step": 93700 }, { "epoch": 0.31, - "learning_rate": 3.445758976653246e-05, - "loss": 3.3101, + "learning_rate": 3.4482838482560564e-05, + "loss": 3.2633, "step": 93800 }, { "epoch": 0.31, - "learning_rate": 3.444102003280807e-05, - "loss": 3.3196, + "learning_rate": 3.4466295666443895e-05, + "loss": 3.2653, "step": 93900 }, { "epoch": 0.31, - "learning_rate": 3.4424450299083694e-05, - "loss": 3.3231, + "learning_rate": 3.444975285032721e-05, + "loss": 3.2522, "step": 94000 }, { "epoch": 0.31, - "learning_rate": 3.4407880565359315e-05, - "loss": 3.3216, + "learning_rate": 3.4433210034210544e-05, + "loss": 3.2494, "step": 94100 }, { "epoch": 0.31, - "learning_rate": 3.4391310831634936e-05, - "loss": 3.3124, + "learning_rate": 3.4416667218093876e-05, + "loss": 3.2713, "step": 94200 }, { "epoch": 0.31, - "learning_rate": 3.437474109791056e-05, - "loss": 3.3241, + "learning_rate": 3.44001244019772e-05, + "loss": 3.2463, "step": 94300 }, { "epoch": 0.31, - "learning_rate": 3.435817136418618e-05, - "loss": 3.319, + "learning_rate": 3.4383581585860525e-05, + "loss": 3.2635, "step": 94400 }, { "epoch": 0.31, - "learning_rate": 3.43416016304618e-05, - "loss": 3.2995, + "learning_rate": 3.4367038769743856e-05, + "loss": 3.2574, "step": 94500 }, { "epoch": 0.31, - "learning_rate": 3.432503189673742e-05, - "loss": 3.3217, + "learning_rate": 3.435049595362718e-05, + "loss": 3.2502, "step": 94600 }, { "epoch": 0.31, - "learning_rate": 3.4308462163013044e-05, - "loss": 3.3154, + "learning_rate": 3.4333953137510505e-05, + "loss": 3.2582, "step": 94700 }, { "epoch": 0.31, - "learning_rate": 3.4291892429288665e-05, - "loss": 3.3202, + "learning_rate": 3.4317410321393836e-05, + "loss": 3.2654, "step": 94800 }, { "epoch": 0.31, - "learning_rate": 3.427532269556428e-05, - "loss": 3.3167, + "learning_rate": 3.430086750527716e-05, + "loss": 3.2643, "step": 94900 }, { "epoch": 0.31, - "learning_rate": 3.425875296183991e-05, - "loss": 3.3108, + "learning_rate": 3.4284324689160485e-05, + "loss": 3.2591, "step": 95000 }, { - "epoch": 0.32, - "learning_rate": 3.424218322811553e-05, - "loss": 3.3242, + "epoch": 0.31, + "learning_rate": 3.426778187304382e-05, + "loss": 3.2538, "step": 95100 }, { - "epoch": 0.32, - "learning_rate": 3.4225613494391144e-05, - "loss": 3.3243, + "epoch": 0.31, + "learning_rate": 3.425123905692714e-05, + "loss": 3.2546, "step": 95200 }, { "epoch": 0.32, - "learning_rate": 3.420904376066677e-05, - "loss": 3.3262, + "learning_rate": 3.4234696240810466e-05, + "loss": 3.2509, "step": 95300 }, { "epoch": 0.32, - "learning_rate": 3.419247402694239e-05, - "loss": 3.3114, + "learning_rate": 3.42181534246938e-05, + "loss": 3.2641, "step": 95400 }, { "epoch": 0.32, - "learning_rate": 3.417590429321801e-05, - "loss": 3.3174, + "learning_rate": 3.420161060857712e-05, + "loss": 3.262, "step": 95500 }, { "epoch": 0.32, - "learning_rate": 3.415933455949363e-05, - "loss": 3.3146, + "learning_rate": 3.4185067792460446e-05, + "loss": 3.2672, "step": 95600 }, { "epoch": 0.32, - "learning_rate": 3.414276482576925e-05, - "loss": 3.3133, + "learning_rate": 3.416852497634378e-05, + "loss": 3.2492, "step": 95700 }, { "epoch": 0.32, - "learning_rate": 3.412619509204487e-05, - "loss": 3.323, + "learning_rate": 3.41519821602271e-05, + "loss": 3.2553, "step": 95800 }, { "epoch": 0.32, - "learning_rate": 3.4109625358320495e-05, - "loss": 3.3273, + "learning_rate": 3.4135439344110427e-05, + "loss": 3.2685, "step": 95900 }, { "epoch": 0.32, - "learning_rate": 3.4093055624596116e-05, - "loss": 3.3172, + "learning_rate": 3.411889652799375e-05, + "loss": 3.2603, "step": 96000 }, { "epoch": 0.32, - "learning_rate": 3.407648589087174e-05, - "loss": 3.3245, + "learning_rate": 3.410235371187708e-05, + "loss": 3.2502, "step": 96100 }, { "epoch": 0.32, - "learning_rate": 3.405991615714735e-05, - "loss": 3.3142, + "learning_rate": 3.408581089576041e-05, + "loss": 3.258, "step": 96200 }, { "epoch": 0.32, - "learning_rate": 3.404334642342298e-05, - "loss": 3.3036, + "learning_rate": 3.406926807964373e-05, + "loss": 3.2508, "step": 96300 }, { "epoch": 0.32, - "learning_rate": 3.4026776689698595e-05, - "loss": 3.3287, + "learning_rate": 3.405272526352706e-05, + "loss": 3.2375, "step": 96400 }, { "epoch": 0.32, - "learning_rate": 3.401020695597422e-05, - "loss": 3.3317, + "learning_rate": 3.4036182447410394e-05, + "loss": 3.2548, "step": 96500 }, { "epoch": 0.32, - "learning_rate": 3.399363722224984e-05, - "loss": 3.3327, + "learning_rate": 3.401963963129371e-05, + "loss": 3.2599, "step": 96600 }, { "epoch": 0.32, - "learning_rate": 3.397706748852546e-05, - "loss": 3.3192, + "learning_rate": 3.400309681517704e-05, + "loss": 3.2427, "step": 96700 }, { "epoch": 0.32, - "learning_rate": 3.396049775480108e-05, - "loss": 3.323, + "learning_rate": 3.3986553999060374e-05, + "loss": 3.2546, "step": 96800 }, { "epoch": 0.32, - "learning_rate": 3.39439280210767e-05, - "loss": 3.3148, + "learning_rate": 3.397001118294369e-05, + "loss": 3.2457, "step": 96900 }, { "epoch": 0.32, - "learning_rate": 3.3927358287352324e-05, - "loss": 3.3175, + "learning_rate": 3.3953468366827024e-05, + "loss": 3.259, "step": 97000 }, { "epoch": 0.32, - "learning_rate": 3.3910788553627945e-05, - "loss": 3.3094, + "learning_rate": 3.3936925550710355e-05, + "loss": 3.2589, "step": 97100 }, { "epoch": 0.32, - "learning_rate": 3.389421881990357e-05, - "loss": 3.3154, + "learning_rate": 3.392038273459367e-05, + "loss": 3.2557, "step": 97200 }, { "epoch": 0.32, - "learning_rate": 3.387764908617919e-05, - "loss": 3.3227, + "learning_rate": 3.3903839918477004e-05, + "loss": 3.2607, "step": 97300 }, { "epoch": 0.32, - "learning_rate": 3.38610793524548e-05, - "loss": 3.3228, + "learning_rate": 3.3887297102360335e-05, + "loss": 3.2706, "step": 97400 }, { "epoch": 0.32, - "learning_rate": 3.384450961873043e-05, - "loss": 3.3391, + "learning_rate": 3.387075428624365e-05, + "loss": 3.2462, "step": 97500 }, { "epoch": 0.32, - "learning_rate": 3.3827939885006046e-05, - "loss": 3.3152, + "learning_rate": 3.3854211470126984e-05, + "loss": 3.2611, "step": 97600 }, { "epoch": 0.32, - "learning_rate": 3.381137015128167e-05, - "loss": 3.3209, + "learning_rate": 3.3837668654010316e-05, + "loss": 3.2471, "step": 97700 }, { "epoch": 0.32, - "learning_rate": 3.3794800417557296e-05, - "loss": 3.3245, + "learning_rate": 3.382112583789363e-05, + "loss": 3.2624, "step": 97800 }, { "epoch": 0.32, - "learning_rate": 3.377823068383291e-05, - "loss": 3.319, + "learning_rate": 3.3804583021776965e-05, + "loss": 3.2702, "step": 97900 }, { "epoch": 0.32, - "learning_rate": 3.376166095010853e-05, - "loss": 3.3214, + "learning_rate": 3.3788040205660296e-05, + "loss": 3.256, "step": 98000 }, { - "epoch": 0.33, - "learning_rate": 3.374509121638415e-05, - "loss": 3.3194, + "epoch": 0.32, + "learning_rate": 3.3771497389543614e-05, + "loss": 3.2601, "step": 98100 }, { - "epoch": 0.33, - "learning_rate": 3.3728521482659775e-05, - "loss": 3.325, + "epoch": 0.32, + "learning_rate": 3.3754954573426945e-05, + "loss": 3.2534, "step": 98200 }, { "epoch": 0.33, - "learning_rate": 3.3711951748935396e-05, - "loss": 3.3147, + "learning_rate": 3.3738411757310276e-05, + "loss": 3.2644, "step": 98300 }, { "epoch": 0.33, - "learning_rate": 3.369538201521102e-05, - "loss": 3.316, + "learning_rate": 3.3721868941193594e-05, + "loss": 3.2516, "step": 98400 }, { "epoch": 0.33, - "learning_rate": 3.367881228148664e-05, - "loss": 3.3197, + "learning_rate": 3.3705326125076925e-05, + "loss": 3.243, "step": 98500 }, { "epoch": 0.33, - "learning_rate": 3.3662242547762254e-05, - "loss": 3.3247, + "learning_rate": 3.368878330896025e-05, + "loss": 3.2404, "step": 98600 }, { "epoch": 0.33, - "learning_rate": 3.364567281403788e-05, - "loss": 3.307, + "learning_rate": 3.367224049284358e-05, + "loss": 3.2613, "step": 98700 }, { "epoch": 0.33, - "learning_rate": 3.3629103080313504e-05, - "loss": 3.3046, + "learning_rate": 3.3655697676726906e-05, + "loss": 3.2706, "step": 98800 }, { "epoch": 0.33, - "learning_rate": 3.361253334658912e-05, - "loss": 3.3116, + "learning_rate": 3.363915486061023e-05, + "loss": 3.2475, "step": 98900 }, { "epoch": 0.33, - "learning_rate": 3.3595963612864747e-05, - "loss": 3.3149, + "learning_rate": 3.362261204449356e-05, + "loss": 3.2509, "step": 99000 }, { "epoch": 0.33, - "learning_rate": 3.357939387914036e-05, - "loss": 3.3129, + "learning_rate": 3.3606069228376886e-05, + "loss": 3.2564, "step": 99100 }, { "epoch": 0.33, - "learning_rate": 3.356282414541598e-05, - "loss": 3.3213, + "learning_rate": 3.358952641226021e-05, + "loss": 3.2648, "step": 99200 }, { "epoch": 0.33, - "learning_rate": 3.354625441169161e-05, - "loss": 3.3049, + "learning_rate": 3.357298359614354e-05, + "loss": 3.2498, "step": 99300 }, { "epoch": 0.33, - "learning_rate": 3.3529684677967226e-05, - "loss": 3.3232, + "learning_rate": 3.3556440780026867e-05, + "loss": 3.2664, "step": 99400 }, { "epoch": 0.33, - "learning_rate": 3.351311494424285e-05, - "loss": 3.3017, + "learning_rate": 3.353989796391019e-05, + "loss": 3.2729, "step": 99500 }, { "epoch": 0.33, - "learning_rate": 3.349654521051847e-05, - "loss": 3.3062, + "learning_rate": 3.352335514779352e-05, + "loss": 3.2632, "step": 99600 }, { "epoch": 0.33, - "learning_rate": 3.347997547679409e-05, - "loss": 3.3099, + "learning_rate": 3.350681233167685e-05, + "loss": 3.2534, "step": 99700 }, { "epoch": 0.33, - "learning_rate": 3.346340574306971e-05, - "loss": 3.3173, + "learning_rate": 3.349026951556017e-05, + "loss": 3.2488, "step": 99800 }, { "epoch": 0.33, - "learning_rate": 3.344683600934533e-05, - "loss": 3.311, + "learning_rate": 3.34737266994435e-05, + "loss": 3.2572, "step": 99900 }, { "epoch": 0.33, - "learning_rate": 3.3430266275620954e-05, - "loss": 3.3054, + "learning_rate": 3.345718388332683e-05, + "loss": 3.2657, "step": 100000 }, { "epoch": 0.33, - "learning_rate": 3.341369654189657e-05, - "loss": 3.3117, + "learning_rate": 3.344064106721015e-05, + "loss": 3.2615, "step": 100100 }, { "epoch": 0.33, - "learning_rate": 3.33971268081722e-05, - "loss": 3.3128, + "learning_rate": 3.342409825109348e-05, + "loss": 3.2519, "step": 100200 }, { "epoch": 0.33, - "learning_rate": 3.338055707444782e-05, - "loss": 3.3185, + "learning_rate": 3.340755543497681e-05, + "loss": 3.2487, "step": 100300 }, { "epoch": 0.33, - "learning_rate": 3.3363987340723434e-05, - "loss": 3.3179, + "learning_rate": 3.339101261886013e-05, + "loss": 3.2669, "step": 100400 }, { "epoch": 0.33, - "learning_rate": 3.334741760699906e-05, - "loss": 3.3199, + "learning_rate": 3.3374469802743464e-05, + "loss": 3.2567, "step": 100500 }, { "epoch": 0.33, - "learning_rate": 3.3330847873274676e-05, - "loss": 3.3197, + "learning_rate": 3.335792698662679e-05, + "loss": 3.266, "step": 100600 }, { "epoch": 0.33, - "learning_rate": 3.33142781395503e-05, - "loss": 3.3131, + "learning_rate": 3.334138417051011e-05, + "loss": 3.2476, "step": 100700 }, { "epoch": 0.33, - "learning_rate": 3.329770840582592e-05, - "loss": 3.3167, + "learning_rate": 3.3324841354393444e-05, + "loss": 3.2634, "step": 100800 }, { "epoch": 0.33, - "learning_rate": 3.328113867210154e-05, - "loss": 3.3213, + "learning_rate": 3.3308298538276775e-05, + "loss": 3.2518, "step": 100900 }, { "epoch": 0.33, - "learning_rate": 3.326456893837716e-05, - "loss": 3.3154, + "learning_rate": 3.329175572216009e-05, + "loss": 3.2505, "step": 101000 }, { - "epoch": 0.34, - "learning_rate": 3.3247999204652784e-05, - "loss": 3.3147, + "epoch": 0.33, + "learning_rate": 3.3275212906043424e-05, + "loss": 3.2456, "step": 101100 }, { - "epoch": 0.34, - "learning_rate": 3.3231429470928405e-05, - "loss": 3.3099, + "epoch": 0.33, + "learning_rate": 3.325867008992675e-05, + "loss": 3.2318, "step": 101200 }, { "epoch": 0.34, - "learning_rate": 3.321485973720403e-05, - "loss": 3.3015, + "learning_rate": 3.3242127273810073e-05, + "loss": 3.2553, "step": 101300 }, { "epoch": 0.34, - "learning_rate": 3.319829000347964e-05, - "loss": 3.3198, + "learning_rate": 3.3225584457693405e-05, + "loss": 3.2426, "step": 101400 }, { "epoch": 0.34, - "learning_rate": 3.318172026975527e-05, - "loss": 3.3168, + "learning_rate": 3.320904164157673e-05, + "loss": 3.2574, "step": 101500 }, { "epoch": 0.34, - "learning_rate": 3.3165150536030884e-05, - "loss": 3.3087, + "learning_rate": 3.3192498825460054e-05, + "loss": 3.271, "step": 101600 }, { "epoch": 0.34, - "learning_rate": 3.3148580802306506e-05, - "loss": 3.3162, + "learning_rate": 3.3175956009343385e-05, + "loss": 3.2631, "step": 101700 }, { "epoch": 0.34, - "learning_rate": 3.313201106858213e-05, - "loss": 3.3189, + "learning_rate": 3.315941319322671e-05, + "loss": 3.2554, "step": 101800 }, { "epoch": 0.34, - "learning_rate": 3.311544133485775e-05, - "loss": 3.3242, + "learning_rate": 3.3142870377110034e-05, + "loss": 3.2509, "step": 101900 }, { "epoch": 0.34, - "learning_rate": 3.309887160113337e-05, - "loss": 3.3201, + "learning_rate": 3.3126327560993365e-05, + "loss": 3.2547, "step": 102000 }, { "epoch": 0.34, - "learning_rate": 3.308230186740899e-05, - "loss": 3.3104, + "learning_rate": 3.310978474487669e-05, + "loss": 3.2674, "step": 102100 }, { "epoch": 0.34, - "learning_rate": 3.306573213368461e-05, - "loss": 3.3114, + "learning_rate": 3.3093241928760015e-05, + "loss": 3.2682, "step": 102200 }, { "epoch": 0.34, - "learning_rate": 3.3049162399960235e-05, - "loss": 3.3129, + "learning_rate": 3.3076699112643346e-05, + "loss": 3.247, "step": 102300 }, { "epoch": 0.34, - "learning_rate": 3.3032592666235856e-05, - "loss": 3.3086, + "learning_rate": 3.306015629652667e-05, + "loss": 3.2585, "step": 102400 }, { "epoch": 0.34, - "learning_rate": 3.301602293251148e-05, - "loss": 3.3129, + "learning_rate": 3.3043613480409995e-05, + "loss": 3.2473, "step": 102500 }, { "epoch": 0.34, - "learning_rate": 3.299945319878709e-05, - "loss": 3.3133, + "learning_rate": 3.3027070664293326e-05, + "loss": 3.2527, "step": 102600 }, { "epoch": 0.34, - "learning_rate": 3.298288346506272e-05, - "loss": 3.3229, + "learning_rate": 3.301052784817665e-05, + "loss": 3.2581, "step": 102700 }, { "epoch": 0.34, - "learning_rate": 3.2966313731338335e-05, - "loss": 3.3265, + "learning_rate": 3.299398503205998e-05, + "loss": 3.2472, "step": 102800 }, { "epoch": 0.34, - "learning_rate": 3.294974399761396e-05, - "loss": 3.3209, + "learning_rate": 3.297744221594331e-05, + "loss": 3.2582, "step": 102900 }, { "epoch": 0.34, - "learning_rate": 3.2933174263889585e-05, - "loss": 3.3068, + "learning_rate": 3.296089939982663e-05, + "loss": 3.2706, "step": 103000 }, { "epoch": 0.34, - "learning_rate": 3.29166045301652e-05, - "loss": 3.3195, + "learning_rate": 3.294435658370996e-05, + "loss": 3.2725, "step": 103100 }, { "epoch": 0.34, - "learning_rate": 3.290003479644082e-05, - "loss": 3.3051, + "learning_rate": 3.292781376759329e-05, + "loss": 3.2477, "step": 103200 }, { "epoch": 0.34, - "learning_rate": 3.288346506271644e-05, - "loss": 3.3192, + "learning_rate": 3.291127095147661e-05, + "loss": 3.2604, "step": 103300 }, { "epoch": 0.34, - "learning_rate": 3.2866895328992064e-05, - "loss": 3.3028, + "learning_rate": 3.289472813535994e-05, + "loss": 3.2636, "step": 103400 }, { "epoch": 0.34, - "learning_rate": 3.2850325595267685e-05, - "loss": 3.2945, + "learning_rate": 3.287818531924327e-05, + "loss": 3.2683, "step": 103500 }, { "epoch": 0.34, - "learning_rate": 3.283375586154331e-05, - "loss": 3.3052, + "learning_rate": 3.286164250312659e-05, + "loss": 3.26, "step": 103600 }, { "epoch": 0.34, - "learning_rate": 3.281718612781893e-05, - "loss": 3.315, + "learning_rate": 3.284509968700992e-05, + "loss": 3.2666, "step": 103700 }, { "epoch": 0.34, - "learning_rate": 3.280061639409455e-05, - "loss": 3.3231, + "learning_rate": 3.282855687089325e-05, + "loss": 3.2449, "step": 103800 }, { "epoch": 0.34, - "learning_rate": 3.278404666037017e-05, - "loss": 3.3184, + "learning_rate": 3.281201405477657e-05, + "loss": 3.2597, "step": 103900 }, { "epoch": 0.34, - "learning_rate": 3.276747692664579e-05, - "loss": 3.3128, + "learning_rate": 3.2795471238659904e-05, + "loss": 3.2521, "step": 104000 }, { "epoch": 0.34, - "learning_rate": 3.275090719292141e-05, - "loss": 3.3073, + "learning_rate": 3.277892842254323e-05, + "loss": 3.2453, "step": 104100 }, { - "epoch": 0.35, - "learning_rate": 3.2734337459197036e-05, - "loss": 3.3116, + "epoch": 0.34, + "learning_rate": 3.276238560642655e-05, + "loss": 3.2639, "step": 104200 }, { "epoch": 0.35, - "learning_rate": 3.271776772547265e-05, - "loss": 3.3108, + "learning_rate": 3.2745842790309884e-05, + "loss": 3.2598, "step": 104300 }, { "epoch": 0.35, - "learning_rate": 3.270119799174827e-05, - "loss": 3.3114, + "learning_rate": 3.272929997419321e-05, + "loss": 3.264, "step": 104400 }, { "epoch": 0.35, - "learning_rate": 3.26846282580239e-05, - "loss": 3.3061, + "learning_rate": 3.271275715807653e-05, + "loss": 3.2736, "step": 104500 }, { "epoch": 0.35, - "learning_rate": 3.2668058524299515e-05, - "loss": 3.3151, + "learning_rate": 3.2696214341959864e-05, + "loss": 3.2444, "step": 104600 }, { "epoch": 0.35, - "learning_rate": 3.2651488790575136e-05, - "loss": 3.3037, + "learning_rate": 3.267967152584319e-05, + "loss": 3.2562, "step": 104700 }, { "epoch": 0.35, - "learning_rate": 3.263491905685076e-05, - "loss": 3.308, + "learning_rate": 3.2663128709726513e-05, + "loss": 3.2625, "step": 104800 }, { "epoch": 0.35, - "learning_rate": 3.261834932312638e-05, - "loss": 3.3096, + "learning_rate": 3.2646585893609845e-05, + "loss": 3.2529, "step": 104900 }, { "epoch": 0.35, - "learning_rate": 3.2601779589402e-05, - "loss": 3.3151, + "learning_rate": 3.263004307749317e-05, + "loss": 3.2427, "step": 105000 }, { "epoch": 0.35, - "learning_rate": 3.258520985567762e-05, - "loss": 3.3104, + "learning_rate": 3.2613500261376494e-05, + "loss": 3.2623, "step": 105100 }, { "epoch": 0.35, - "learning_rate": 3.2568640121953244e-05, - "loss": 3.3078, + "learning_rate": 3.2596957445259825e-05, + "loss": 3.247, "step": 105200 }, { "epoch": 0.35, - "learning_rate": 3.255207038822886e-05, - "loss": 3.3137, + "learning_rate": 3.258041462914315e-05, + "loss": 3.25, "step": 105300 }, { "epoch": 0.35, - "learning_rate": 3.2535500654504487e-05, - "loss": 3.3231, + "learning_rate": 3.2563871813026474e-05, + "loss": 3.2543, "step": 105400 }, { "epoch": 0.35, - "learning_rate": 3.251893092078011e-05, - "loss": 3.3048, + "learning_rate": 3.2547328996909806e-05, + "loss": 3.2667, "step": 105500 }, { "epoch": 0.35, - "learning_rate": 3.250236118705572e-05, - "loss": 3.3142, + "learning_rate": 3.253078618079313e-05, + "loss": 3.2512, "step": 105600 }, { "epoch": 0.35, - "learning_rate": 3.248579145333135e-05, - "loss": 3.3189, + "learning_rate": 3.2514243364676455e-05, + "loss": 3.2645, "step": 105700 }, { "epoch": 0.35, - "learning_rate": 3.2469221719606966e-05, - "loss": 3.3046, + "learning_rate": 3.2497700548559786e-05, + "loss": 3.2437, "step": 105800 }, { "epoch": 0.35, - "learning_rate": 3.245265198588259e-05, - "loss": 3.3154, + "learning_rate": 3.248115773244311e-05, + "loss": 3.2566, "step": 105900 }, { "epoch": 0.35, - "learning_rate": 3.243608225215821e-05, - "loss": 3.305, + "learning_rate": 3.2464614916326435e-05, + "loss": 3.2543, "step": 106000 }, { "epoch": 0.35, - "learning_rate": 3.241951251843383e-05, - "loss": 3.3095, + "learning_rate": 3.2448072100209766e-05, + "loss": 3.2631, "step": 106100 }, { "epoch": 0.35, - "learning_rate": 3.240294278470945e-05, - "loss": 3.3065, + "learning_rate": 3.243152928409309e-05, + "loss": 3.2499, "step": 106200 }, { "epoch": 0.35, - "learning_rate": 3.238637305098507e-05, - "loss": 3.3192, + "learning_rate": 3.2414986467976415e-05, + "loss": 3.2435, "step": 106300 }, { "epoch": 0.35, - "learning_rate": 3.2369803317260694e-05, - "loss": 3.3202, + "learning_rate": 3.239844365185975e-05, + "loss": 3.2612, "step": 106400 }, { "epoch": 0.35, - "learning_rate": 3.2353233583536316e-05, - "loss": 3.3266, + "learning_rate": 3.238190083574307e-05, + "loss": 3.2565, "step": 106500 }, { "epoch": 0.35, - "learning_rate": 3.233666384981193e-05, - "loss": 3.3065, + "learning_rate": 3.2365358019626396e-05, + "loss": 3.2454, "step": 106600 }, { "epoch": 0.35, - "learning_rate": 3.232009411608756e-05, - "loss": 3.3059, + "learning_rate": 3.234881520350973e-05, + "loss": 3.2609, "step": 106700 }, { "epoch": 0.35, - "learning_rate": 3.2303524382363174e-05, - "loss": 3.3187, + "learning_rate": 3.233227238739305e-05, + "loss": 3.2491, "step": 106800 }, { "epoch": 0.35, - "learning_rate": 3.2286954648638795e-05, - "loss": 3.3006, + "learning_rate": 3.2315729571276376e-05, + "loss": 3.2552, "step": 106900 }, { "epoch": 0.35, - "learning_rate": 3.227038491491442e-05, - "loss": 3.3092, + "learning_rate": 3.229918675515971e-05, + "loss": 3.2555, "step": 107000 }, { "epoch": 0.35, - "learning_rate": 3.225381518119004e-05, - "loss": 3.2996, + "learning_rate": 3.228264393904303e-05, + "loss": 3.2598, "step": 107100 }, { - "epoch": 0.36, - "learning_rate": 3.223724544746566e-05, - "loss": 3.3104, + "epoch": 0.35, + "learning_rate": 3.226610112292636e-05, + "loss": 3.275, "step": 107200 }, { "epoch": 0.36, - "learning_rate": 3.222067571374128e-05, - "loss": 3.3, + "learning_rate": 3.224955830680969e-05, + "loss": 3.2497, "step": 107300 }, { "epoch": 0.36, - "learning_rate": 3.22041059800169e-05, - "loss": 3.3112, + "learning_rate": 3.223301549069301e-05, + "loss": 3.2512, "step": 107400 }, { "epoch": 0.36, - "learning_rate": 3.2187536246292524e-05, - "loss": 3.3132, + "learning_rate": 3.2216472674576344e-05, + "loss": 3.2622, "step": 107500 }, { "epoch": 0.36, - "learning_rate": 3.2170966512568145e-05, - "loss": 3.3181, + "learning_rate": 3.219992985845967e-05, + "loss": 3.265, "step": 107600 }, { "epoch": 0.36, - "learning_rate": 3.215439677884377e-05, - "loss": 3.3147, + "learning_rate": 3.218338704234299e-05, + "loss": 3.2593, "step": 107700 }, { "epoch": 0.36, - "learning_rate": 3.213782704511938e-05, - "loss": 3.3159, + "learning_rate": 3.2166844226226324e-05, + "loss": 3.245, "step": 107800 }, { "epoch": 0.36, - "learning_rate": 3.212125731139501e-05, - "loss": 3.3144, + "learning_rate": 3.215030141010965e-05, + "loss": 3.2429, "step": 107900 }, { "epoch": 0.36, - "learning_rate": 3.210468757767063e-05, - "loss": 3.3088, + "learning_rate": 3.213375859399297e-05, + "loss": 3.2386, "step": 108000 }, { "epoch": 0.36, - "learning_rate": 3.2088117843946246e-05, - "loss": 3.317, + "learning_rate": 3.2117215777876304e-05, + "loss": 3.251, "step": 108100 }, { "epoch": 0.36, - "learning_rate": 3.2071548110221874e-05, - "loss": 3.3117, + "learning_rate": 3.210067296175963e-05, + "loss": 3.253, "step": 108200 }, { "epoch": 0.36, - "learning_rate": 3.205497837649749e-05, - "loss": 3.3121, + "learning_rate": 3.2084130145642953e-05, + "loss": 3.2625, "step": 108300 }, { "epoch": 0.36, - "learning_rate": 3.203840864277311e-05, - "loss": 3.3206, + "learning_rate": 3.2067587329526285e-05, + "loss": 3.2448, "step": 108400 }, { "epoch": 0.36, - "learning_rate": 3.202183890904873e-05, - "loss": 3.3157, + "learning_rate": 3.205104451340961e-05, + "loss": 3.252, "step": 108500 }, { "epoch": 0.36, - "learning_rate": 3.200526917532435e-05, - "loss": 3.311, + "learning_rate": 3.2034501697292934e-05, + "loss": 3.244, "step": 108600 }, { "epoch": 0.36, - "learning_rate": 3.1988699441599975e-05, - "loss": 3.3001, + "learning_rate": 3.2017958881176265e-05, + "loss": 3.2716, "step": 108700 }, { "epoch": 0.36, - "learning_rate": 3.1972129707875596e-05, - "loss": 3.304, + "learning_rate": 3.200141606505958e-05, + "loss": 3.2486, "step": 108800 }, { "epoch": 0.36, - "learning_rate": 3.195555997415122e-05, - "loss": 3.3021, + "learning_rate": 3.1984873248942914e-05, + "loss": 3.2509, "step": 108900 }, { "epoch": 0.36, - "learning_rate": 3.193899024042684e-05, - "loss": 3.303, + "learning_rate": 3.1968330432826246e-05, + "loss": 3.2505, "step": 109000 }, { "epoch": 0.36, - "learning_rate": 3.192242050670246e-05, - "loss": 3.3237, + "learning_rate": 3.195178761670957e-05, + "loss": 3.2614, "step": 109100 }, { "epoch": 0.36, - "learning_rate": 3.190585077297808e-05, - "loss": 3.31, + "learning_rate": 3.1935244800592895e-05, + "loss": 3.2498, "step": 109200 }, { "epoch": 0.36, - "learning_rate": 3.18892810392537e-05, - "loss": 3.3169, + "learning_rate": 3.1918701984476226e-05, + "loss": 3.2461, "step": 109300 }, { "epoch": 0.36, - "learning_rate": 3.1872711305529325e-05, - "loss": 3.3045, + "learning_rate": 3.190215916835955e-05, + "loss": 3.252, "step": 109400 }, { "epoch": 0.36, - "learning_rate": 3.185614157180494e-05, - "loss": 3.3158, + "learning_rate": 3.1885616352242875e-05, + "loss": 3.2684, "step": 109500 }, { "epoch": 0.36, - "learning_rate": 3.183957183808056e-05, - "loss": 3.3187, + "learning_rate": 3.1869073536126206e-05, + "loss": 3.2564, "step": 109600 }, { "epoch": 0.36, - "learning_rate": 3.182300210435619e-05, - "loss": 3.2998, + "learning_rate": 3.185253072000953e-05, + "loss": 3.2538, "step": 109700 }, { "epoch": 0.36, - "learning_rate": 3.1806432370631804e-05, - "loss": 3.3015, + "learning_rate": 3.1835987903892855e-05, + "loss": 3.2667, "step": 109800 }, { "epoch": 0.36, - "learning_rate": 3.1789862636907425e-05, - "loss": 3.3058, + "learning_rate": 3.181944508777619e-05, + "loss": 3.268, "step": 109900 }, { "epoch": 0.36, - "learning_rate": 3.177329290318305e-05, - "loss": 3.3215, + "learning_rate": 3.180290227165951e-05, + "loss": 3.2611, "step": 110000 }, { "epoch": 0.36, - "learning_rate": 3.175672316945867e-05, - "loss": 3.3004, + "learning_rate": 3.1786359455542836e-05, + "loss": 3.2614, "step": 110100 }, { - "epoch": 0.37, - "learning_rate": 3.174015343573429e-05, - "loss": 3.2953, + "epoch": 0.36, + "learning_rate": 3.176981663942617e-05, + "loss": 3.2639, "step": 110200 }, { - "epoch": 0.37, - "learning_rate": 3.172358370200991e-05, - "loss": 3.2975, + "epoch": 0.36, + "learning_rate": 3.175327382330949e-05, + "loss": 3.2461, "step": 110300 }, { "epoch": 0.37, - "learning_rate": 3.170701396828553e-05, - "loss": 3.3137, + "learning_rate": 3.1736731007192816e-05, + "loss": 3.2381, "step": 110400 }, { "epoch": 0.37, - "learning_rate": 3.169044423456115e-05, - "loss": 3.3093, + "learning_rate": 3.172018819107615e-05, + "loss": 3.2614, "step": 110500 }, { "epoch": 0.37, - "learning_rate": 3.1673874500836776e-05, - "loss": 3.3037, + "learning_rate": 3.170364537495947e-05, + "loss": 3.2706, "step": 110600 }, { "epoch": 0.37, - "learning_rate": 3.16573047671124e-05, - "loss": 3.3011, + "learning_rate": 3.1687102558842797e-05, + "loss": 3.2566, "step": 110700 }, { "epoch": 0.37, - "learning_rate": 3.164073503338801e-05, - "loss": 3.2956, + "learning_rate": 3.167055974272613e-05, + "loss": 3.251, "step": 110800 }, { "epoch": 0.37, - "learning_rate": 3.162416529966364e-05, - "loss": 3.3024, + "learning_rate": 3.165401692660945e-05, + "loss": 3.2492, "step": 110900 }, { "epoch": 0.37, - "learning_rate": 3.1607595565939255e-05, - "loss": 3.3085, + "learning_rate": 3.163747411049278e-05, + "loss": 3.2478, "step": 111000 }, { "epoch": 0.37, - "learning_rate": 3.1591025832214876e-05, - "loss": 3.3161, + "learning_rate": 3.162093129437611e-05, + "loss": 3.2479, "step": 111100 }, { "epoch": 0.37, - "learning_rate": 3.1574456098490505e-05, - "loss": 3.3101, + "learning_rate": 3.160438847825943e-05, + "loss": 3.2489, "step": 111200 }, { "epoch": 0.37, - "learning_rate": 3.155788636476612e-05, - "loss": 3.3044, + "learning_rate": 3.1587845662142764e-05, + "loss": 3.2688, "step": 111300 }, { "epoch": 0.37, - "learning_rate": 3.154131663104174e-05, - "loss": 3.3141, + "learning_rate": 3.157130284602608e-05, + "loss": 3.2459, "step": 111400 }, { "epoch": 0.37, - "learning_rate": 3.152474689731736e-05, - "loss": 3.3088, + "learning_rate": 3.155476002990941e-05, + "loss": 3.2491, "step": 111500 }, { "epoch": 0.37, - "learning_rate": 3.1508177163592984e-05, - "loss": 3.3124, + "learning_rate": 3.1538217213792744e-05, + "loss": 3.2474, "step": 111600 }, { "epoch": 0.37, - "learning_rate": 3.1491607429868605e-05, - "loss": 3.3131, + "learning_rate": 3.152167439767606e-05, + "loss": 3.2456, "step": 111700 }, { "epoch": 0.37, - "learning_rate": 3.147503769614422e-05, - "loss": 3.3061, + "learning_rate": 3.1505131581559393e-05, + "loss": 3.2515, "step": 111800 }, { "epoch": 0.37, - "learning_rate": 3.145846796241985e-05, - "loss": 3.3065, + "learning_rate": 3.1488588765442725e-05, + "loss": 3.2537, "step": 111900 }, { "epoch": 0.37, - "learning_rate": 3.144189822869546e-05, - "loss": 3.301, + "learning_rate": 3.147204594932604e-05, + "loss": 3.2649, "step": 112000 }, { "epoch": 0.37, - "learning_rate": 3.1425328494971084e-05, - "loss": 3.313, + "learning_rate": 3.1455503133209374e-05, + "loss": 3.2436, "step": 112100 }, { "epoch": 0.37, - "learning_rate": 3.140875876124671e-05, - "loss": 3.3081, + "learning_rate": 3.1438960317092705e-05, + "loss": 3.2633, "step": 112200 }, { "epoch": 0.37, - "learning_rate": 3.139218902752233e-05, - "loss": 3.3031, + "learning_rate": 3.142241750097602e-05, + "loss": 3.2567, "step": 112300 }, { "epoch": 0.37, - "learning_rate": 3.137561929379795e-05, - "loss": 3.3074, + "learning_rate": 3.1405874684859354e-05, + "loss": 3.2506, "step": 112400 }, { "epoch": 0.37, - "learning_rate": 3.135904956007357e-05, - "loss": 3.3072, + "learning_rate": 3.1389331868742686e-05, + "loss": 3.2619, "step": 112500 }, { "epoch": 0.37, - "learning_rate": 3.134247982634919e-05, - "loss": 3.3122, + "learning_rate": 3.1372789052626e-05, + "loss": 3.2337, "step": 112600 }, { "epoch": 0.37, - "learning_rate": 3.132591009262481e-05, - "loss": 3.303, + "learning_rate": 3.1356246236509335e-05, + "loss": 3.2546, "step": 112700 }, { "epoch": 0.37, - "learning_rate": 3.1309340358900434e-05, - "loss": 3.2882, + "learning_rate": 3.1339703420392666e-05, + "loss": 3.2644, "step": 112800 }, { "epoch": 0.37, - "learning_rate": 3.1292770625176056e-05, - "loss": 3.2878, + "learning_rate": 3.1323160604275984e-05, + "loss": 3.2641, "step": 112900 }, { "epoch": 0.37, - "learning_rate": 3.127620089145167e-05, - "loss": 3.3077, + "learning_rate": 3.1306617788159315e-05, + "loss": 3.2572, "step": 113000 }, { "epoch": 0.37, - "learning_rate": 3.12596311577273e-05, - "loss": 3.3056, + "learning_rate": 3.1290074972042646e-05, + "loss": 3.2529, "step": 113100 }, { - "epoch": 0.38, - "learning_rate": 3.124306142400292e-05, - "loss": 3.3048, + "epoch": 0.37, + "learning_rate": 3.1273532155925964e-05, + "loss": 3.2674, "step": 113200 }, { - "epoch": 0.38, - "learning_rate": 3.1226491690278535e-05, - "loss": 3.3073, + "epoch": 0.37, + "learning_rate": 3.1256989339809295e-05, + "loss": 3.2632, "step": 113300 }, { "epoch": 0.38, - "learning_rate": 3.120992195655416e-05, - "loss": 3.3039, + "learning_rate": 3.124044652369263e-05, + "loss": 3.2638, "step": 113400 }, { "epoch": 0.38, - "learning_rate": 3.119335222282978e-05, - "loss": 3.3027, + "learning_rate": 3.122390370757595e-05, + "loss": 3.2595, "step": 113500 }, { "epoch": 0.38, - "learning_rate": 3.11767824891054e-05, - "loss": 3.3099, + "learning_rate": 3.1207360891459276e-05, + "loss": 3.2524, "step": 113600 }, { "epoch": 0.38, - "learning_rate": 3.116021275538102e-05, - "loss": 3.3214, + "learning_rate": 3.119081807534261e-05, + "loss": 3.2444, "step": 113700 }, { "epoch": 0.38, - "learning_rate": 3.114364302165664e-05, - "loss": 3.3078, + "learning_rate": 3.117427525922593e-05, + "loss": 3.2588, "step": 113800 }, { "epoch": 0.38, - "learning_rate": 3.1127073287932264e-05, - "loss": 3.3039, + "learning_rate": 3.1157732443109256e-05, + "loss": 3.2632, "step": 113900 }, { "epoch": 0.38, - "learning_rate": 3.1110503554207885e-05, - "loss": 3.314, + "learning_rate": 3.114118962699258e-05, + "loss": 3.2611, "step": 114000 }, { "epoch": 0.38, - "learning_rate": 3.109393382048351e-05, - "loss": 3.3034, + "learning_rate": 3.112464681087591e-05, + "loss": 3.2551, "step": 114100 }, { "epoch": 0.38, - "learning_rate": 3.107736408675913e-05, - "loss": 3.2986, + "learning_rate": 3.1108103994759237e-05, + "loss": 3.2622, "step": 114200 }, { "epoch": 0.38, - "learning_rate": 3.106079435303475e-05, - "loss": 3.3099, + "learning_rate": 3.109156117864256e-05, + "loss": 3.2489, "step": 114300 }, { "epoch": 0.38, - "learning_rate": 3.104422461931037e-05, - "loss": 3.2958, + "learning_rate": 3.107501836252589e-05, + "loss": 3.2541, "step": 114400 }, { "epoch": 0.38, - "learning_rate": 3.1027654885585986e-05, - "loss": 3.3054, + "learning_rate": 3.105847554640922e-05, + "loss": 3.2511, "step": 114500 }, { "epoch": 0.38, - "learning_rate": 3.1011085151861614e-05, - "loss": 3.299, + "learning_rate": 3.104193273029254e-05, + "loss": 3.2567, "step": 114600 }, { "epoch": 0.38, - "learning_rate": 3.099451541813723e-05, - "loss": 3.3178, + "learning_rate": 3.102538991417587e-05, + "loss": 3.2569, "step": 114700 }, { "epoch": 0.38, - "learning_rate": 3.097794568441285e-05, - "loss": 3.2923, + "learning_rate": 3.10088470980592e-05, + "loss": 3.2586, "step": 114800 }, { "epoch": 0.38, - "learning_rate": 3.096137595068848e-05, - "loss": 3.3082, + "learning_rate": 3.099230428194252e-05, + "loss": 3.2541, "step": 114900 }, { "epoch": 0.38, - "learning_rate": 3.094480621696409e-05, - "loss": 3.3059, + "learning_rate": 3.097576146582585e-05, + "loss": 3.256, "step": 115000 }, { "epoch": 0.38, - "learning_rate": 3.0928236483239715e-05, - "loss": 3.3035, + "learning_rate": 3.095921864970918e-05, + "loss": 3.254, "step": 115100 }, { "epoch": 0.38, - "learning_rate": 3.0911666749515336e-05, - "loss": 3.2962, + "learning_rate": 3.09426758335925e-05, + "loss": 3.2426, "step": 115200 }, { "epoch": 0.38, - "learning_rate": 3.089509701579096e-05, - "loss": 3.3004, + "learning_rate": 3.0926133017475834e-05, + "loss": 3.2499, "step": 115300 }, { "epoch": 0.38, - "learning_rate": 3.087852728206658e-05, - "loss": 3.2886, + "learning_rate": 3.090959020135916e-05, + "loss": 3.2554, "step": 115400 }, { "epoch": 0.38, - "learning_rate": 3.08619575483422e-05, - "loss": 3.3008, + "learning_rate": 3.089304738524248e-05, + "loss": 3.2437, "step": 115500 }, { "epoch": 0.38, - "learning_rate": 3.084538781461782e-05, - "loss": 3.3072, + "learning_rate": 3.0876504569125814e-05, + "loss": 3.2515, "step": 115600 }, { "epoch": 0.38, - "learning_rate": 3.0828818080893443e-05, - "loss": 3.3158, + "learning_rate": 3.0859961753009145e-05, + "loss": 3.2665, "step": 115700 }, { "epoch": 0.38, - "learning_rate": 3.0812248347169065e-05, - "loss": 3.3263, + "learning_rate": 3.084341893689246e-05, + "loss": 3.2462, "step": 115800 }, { "epoch": 0.38, - "learning_rate": 3.0795678613444686e-05, - "loss": 3.3183, + "learning_rate": 3.0826876120775794e-05, + "loss": 3.2389, "step": 115900 }, { "epoch": 0.38, - "learning_rate": 3.07791088797203e-05, - "loss": 3.3044, + "learning_rate": 3.0810333304659126e-05, + "loss": 3.2469, "step": 116000 }, { "epoch": 0.38, - "learning_rate": 3.076253914599593e-05, - "loss": 3.3049, + "learning_rate": 3.079379048854244e-05, + "loss": 3.2634, "step": 116100 }, { - "epoch": 0.39, - "learning_rate": 3.0745969412271544e-05, - "loss": 3.3092, + "epoch": 0.38, + "learning_rate": 3.0777247672425775e-05, + "loss": 3.2555, "step": 116200 }, { - "epoch": 0.39, - "learning_rate": 3.0729399678547165e-05, - "loss": 3.3161, + "epoch": 0.38, + "learning_rate": 3.0760704856309106e-05, + "loss": 3.257, "step": 116300 }, { "epoch": 0.39, - "learning_rate": 3.0712829944822794e-05, - "loss": 3.3024, + "learning_rate": 3.0744162040192424e-05, + "loss": 3.2578, "step": 116400 }, { "epoch": 0.39, - "learning_rate": 3.069626021109841e-05, - "loss": 3.3033, + "learning_rate": 3.0727619224075755e-05, + "loss": 3.2435, "step": 116500 }, { "epoch": 0.39, - "learning_rate": 3.067969047737403e-05, - "loss": 3.2878, + "learning_rate": 3.071107640795908e-05, + "loss": 3.2459, "step": 116600 }, { "epoch": 0.39, - "learning_rate": 3.066312074364965e-05, - "loss": 3.3144, + "learning_rate": 3.0694533591842404e-05, + "loss": 3.2538, "step": 116700 }, { "epoch": 0.39, - "learning_rate": 3.064655100992527e-05, - "loss": 3.3086, + "learning_rate": 3.0677990775725735e-05, + "loss": 3.24, "step": 116800 }, { "epoch": 0.39, - "learning_rate": 3.0629981276200894e-05, - "loss": 3.3005, + "learning_rate": 3.066144795960906e-05, + "loss": 3.2555, "step": 116900 }, { "epoch": 0.39, - "learning_rate": 3.061341154247651e-05, - "loss": 3.2935, + "learning_rate": 3.0644905143492384e-05, + "loss": 3.2574, "step": 117000 }, { "epoch": 0.39, - "learning_rate": 3.059684180875214e-05, - "loss": 3.307, + "learning_rate": 3.0628362327375716e-05, + "loss": 3.247, "step": 117100 }, { "epoch": 0.39, - "learning_rate": 3.058027207502775e-05, - "loss": 3.2984, + "learning_rate": 3.061181951125904e-05, + "loss": 3.2465, "step": 117200 }, { "epoch": 0.39, - "learning_rate": 3.056370234130337e-05, - "loss": 3.3104, + "learning_rate": 3.0595276695142365e-05, + "loss": 3.264, "step": 117300 }, { "epoch": 0.39, - "learning_rate": 3.0547132607579e-05, - "loss": 3.3, + "learning_rate": 3.0578733879025696e-05, + "loss": 3.2506, "step": 117400 }, { "epoch": 0.39, - "learning_rate": 3.0530562873854616e-05, - "loss": 3.3153, + "learning_rate": 3.056219106290902e-05, + "loss": 3.2582, "step": 117500 }, { "epoch": 0.39, - "learning_rate": 3.0513993140130238e-05, - "loss": 3.3047, + "learning_rate": 3.054564824679235e-05, + "loss": 3.25, "step": 117600 }, { "epoch": 0.39, - "learning_rate": 3.0497423406405863e-05, - "loss": 3.312, + "learning_rate": 3.0529105430675677e-05, + "loss": 3.2487, "step": 117700 }, { "epoch": 0.39, - "learning_rate": 3.048085367268148e-05, - "loss": 3.3066, + "learning_rate": 3.0512562614559004e-05, + "loss": 3.242, "step": 117800 }, { "epoch": 0.39, - "learning_rate": 3.04642839389571e-05, - "loss": 3.3156, + "learning_rate": 3.0496019798442332e-05, + "loss": 3.2457, "step": 117900 }, { "epoch": 0.39, - "learning_rate": 3.0447714205232724e-05, - "loss": 3.2915, + "learning_rate": 3.0479476982325657e-05, + "loss": 3.2446, "step": 118000 }, { "epoch": 0.39, - "learning_rate": 3.0431144471508345e-05, - "loss": 3.2989, + "learning_rate": 3.0462934166208985e-05, + "loss": 3.248, "step": 118100 }, { "epoch": 0.39, - "learning_rate": 3.0414574737783963e-05, - "loss": 3.2989, + "learning_rate": 3.0446391350092313e-05, + "loss": 3.2506, "step": 118200 }, { "epoch": 0.39, - "learning_rate": 3.0398005004059588e-05, - "loss": 3.2925, + "learning_rate": 3.0429848533975634e-05, + "loss": 3.2581, "step": 118300 }, { "epoch": 0.39, - "learning_rate": 3.0381435270335206e-05, - "loss": 3.316, + "learning_rate": 3.0413305717858965e-05, + "loss": 3.2469, "step": 118400 }, { "epoch": 0.39, - "learning_rate": 3.0364865536610828e-05, - "loss": 3.3056, + "learning_rate": 3.0396762901742293e-05, + "loss": 3.2618, "step": 118500 }, { "epoch": 0.39, - "learning_rate": 3.034829580288645e-05, - "loss": 3.301, + "learning_rate": 3.0380220085625614e-05, + "loss": 3.2452, "step": 118600 }, { "epoch": 0.39, - "learning_rate": 3.033172606916207e-05, - "loss": 3.3109, + "learning_rate": 3.0363677269508946e-05, + "loss": 3.2527, "step": 118700 }, { "epoch": 0.39, - "learning_rate": 3.031515633543769e-05, - "loss": 3.3054, + "learning_rate": 3.0347134453392274e-05, + "loss": 3.2585, "step": 118800 }, { "epoch": 0.39, - "learning_rate": 3.0298586601713313e-05, - "loss": 3.3163, + "learning_rate": 3.0330591637275595e-05, + "loss": 3.2476, "step": 118900 }, { "epoch": 0.39, - "learning_rate": 3.028201686798893e-05, - "loss": 3.3041, + "learning_rate": 3.0314048821158926e-05, + "loss": 3.2521, "step": 119000 }, { "epoch": 0.39, - "learning_rate": 3.0265447134264553e-05, - "loss": 3.303, + "learning_rate": 3.0297506005042254e-05, + "loss": 3.2575, "step": 119100 }, { - "epoch": 0.4, - "learning_rate": 3.0248877400540178e-05, - "loss": 3.2931, + "epoch": 0.39, + "learning_rate": 3.0280963188925575e-05, + "loss": 3.2661, "step": 119200 }, { - "epoch": 0.4, - "learning_rate": 3.0232307666815796e-05, - "loss": 3.3056, + "epoch": 0.39, + "learning_rate": 3.0264420372808906e-05, + "loss": 3.2576, "step": 119300 }, { "epoch": 0.4, - "learning_rate": 3.0215737933091414e-05, - "loss": 3.3065, + "learning_rate": 3.0247877556692234e-05, + "loss": 3.2403, "step": 119400 }, { "epoch": 0.4, - "learning_rate": 3.019916819936704e-05, - "loss": 3.3067, + "learning_rate": 3.0231334740575555e-05, + "loss": 3.253, "step": 119500 }, { "epoch": 0.4, - "learning_rate": 3.0182598465642657e-05, - "loss": 3.3103, + "learning_rate": 3.0214791924458887e-05, + "loss": 3.2435, "step": 119600 }, { "epoch": 0.4, - "learning_rate": 3.016602873191828e-05, - "loss": 3.2999, + "learning_rate": 3.0198249108342215e-05, + "loss": 3.2379, "step": 119700 }, { "epoch": 0.4, - "learning_rate": 3.0149458998193903e-05, - "loss": 3.3203, + "learning_rate": 3.0181706292225543e-05, + "loss": 3.2457, "step": 119800 }, { "epoch": 0.4, - "learning_rate": 3.013288926446952e-05, - "loss": 3.3062, + "learning_rate": 3.0165163476108864e-05, + "loss": 3.2565, "step": 119900 }, { "epoch": 0.4, - "learning_rate": 3.011631953074514e-05, - "loss": 3.3034, + "learning_rate": 3.0148620659992195e-05, + "loss": 3.2483, "step": 120000 }, { "epoch": 0.4, - "learning_rate": 3.0099749797020764e-05, - "loss": 3.3057, + "learning_rate": 3.0132077843875523e-05, + "loss": 3.2476, "step": 120100 }, { "epoch": 0.4, - "learning_rate": 3.0083180063296386e-05, - "loss": 3.298, + "learning_rate": 3.0115535027758844e-05, + "loss": 3.2481, "step": 120200 }, { "epoch": 0.4, - "learning_rate": 3.0066610329572004e-05, - "loss": 3.3028, + "learning_rate": 3.0098992211642175e-05, + "loss": 3.2485, "step": 120300 }, { "epoch": 0.4, - "learning_rate": 3.005004059584763e-05, - "loss": 3.3102, + "learning_rate": 3.0082449395525503e-05, + "loss": 3.2542, "step": 120400 }, { "epoch": 0.4, - "learning_rate": 3.0033470862123247e-05, - "loss": 3.2998, + "learning_rate": 3.0065906579408825e-05, + "loss": 3.2604, "step": 120500 }, { "epoch": 0.4, - "learning_rate": 3.0016901128398868e-05, - "loss": 3.3043, + "learning_rate": 3.0049363763292156e-05, + "loss": 3.2569, "step": 120600 }, { "epoch": 0.4, - "learning_rate": 3.000033139467449e-05, - "loss": 3.3097, + "learning_rate": 3.0032820947175484e-05, + "loss": 3.2416, "step": 120700 }, { "epoch": 0.4, - "learning_rate": 2.998376166095011e-05, - "loss": 3.2932, + "learning_rate": 3.0016278131058805e-05, + "loss": 3.2531, "step": 120800 }, { "epoch": 0.4, - "learning_rate": 2.996719192722573e-05, - "loss": 3.295, + "learning_rate": 2.9999735314942136e-05, + "loss": 3.2484, "step": 120900 }, { "epoch": 0.4, - "learning_rate": 2.9950622193501354e-05, - "loss": 3.2985, + "learning_rate": 2.9983192498825464e-05, + "loss": 3.2483, "step": 121000 }, { "epoch": 0.4, - "learning_rate": 2.9934052459776972e-05, - "loss": 3.3072, + "learning_rate": 2.9966649682708785e-05, + "loss": 3.2284, "step": 121100 }, { "epoch": 0.4, - "learning_rate": 2.9917482726052594e-05, - "loss": 3.3072, + "learning_rate": 2.9950106866592113e-05, + "loss": 3.2521, "step": 121200 }, { "epoch": 0.4, - "learning_rate": 2.990091299232822e-05, - "loss": 3.2897, + "learning_rate": 2.9933564050475445e-05, + "loss": 3.2531, "step": 121300 }, { "epoch": 0.4, - "learning_rate": 2.9884343258603837e-05, - "loss": 3.2973, + "learning_rate": 2.9917021234358766e-05, + "loss": 3.2729, "step": 121400 }, { "epoch": 0.4, - "learning_rate": 2.9867773524879455e-05, - "loss": 3.2913, + "learning_rate": 2.9900478418242094e-05, + "loss": 3.246, "step": 121500 }, { "epoch": 0.4, - "learning_rate": 2.985120379115508e-05, - "loss": 3.2964, + "learning_rate": 2.9883935602125425e-05, + "loss": 3.2473, "step": 121600 }, { "epoch": 0.4, - "learning_rate": 2.98346340574307e-05, - "loss": 3.3007, + "learning_rate": 2.9867392786008746e-05, + "loss": 3.2486, "step": 121700 }, { "epoch": 0.4, - "learning_rate": 2.981806432370632e-05, - "loss": 3.2948, + "learning_rate": 2.9850849969892074e-05, + "loss": 3.2536, "step": 121800 }, { "epoch": 0.4, - "learning_rate": 2.9801494589981944e-05, - "loss": 3.3018, + "learning_rate": 2.9834307153775405e-05, + "loss": 3.2572, "step": 121900 }, { "epoch": 0.4, - "learning_rate": 2.9784924856257562e-05, - "loss": 3.3179, + "learning_rate": 2.9817764337658733e-05, + "loss": 3.2524, "step": 122000 }, { "epoch": 0.4, - "learning_rate": 2.976835512253318e-05, - "loss": 3.3006, + "learning_rate": 2.9801221521542054e-05, + "loss": 3.2476, "step": 122100 }, { "epoch": 0.4, - "learning_rate": 2.97517853888088e-05, - "loss": 3.3017, + "learning_rate": 2.9784678705425386e-05, + "loss": 3.2344, "step": 122200 }, { - "epoch": 0.41, - "learning_rate": 2.9735215655084426e-05, - "loss": 3.3056, + "epoch": 0.4, + "learning_rate": 2.9768135889308714e-05, + "loss": 3.2537, "step": 122300 }, { - "epoch": 0.41, - "learning_rate": 2.9718645921360044e-05, - "loss": 3.3028, + "epoch": 0.4, + "learning_rate": 2.9751593073192035e-05, + "loss": 3.2488, "step": 122400 }, { "epoch": 0.41, - "learning_rate": 2.9702076187635663e-05, - "loss": 3.2979, + "learning_rate": 2.9735050257075363e-05, + "loss": 3.2463, "step": 122500 }, { "epoch": 0.41, - "learning_rate": 2.9685506453911287e-05, - "loss": 3.3089, + "learning_rate": 2.9718507440958694e-05, + "loss": 3.2449, "step": 122600 }, { "epoch": 0.41, - "learning_rate": 2.966893672018691e-05, - "loss": 3.3045, + "learning_rate": 2.9701964624842015e-05, + "loss": 3.2469, "step": 122700 }, { "epoch": 0.41, - "learning_rate": 2.9652366986462527e-05, - "loss": 3.3037, + "learning_rate": 2.9685421808725343e-05, + "loss": 3.2463, "step": 122800 }, { "epoch": 0.41, - "learning_rate": 2.9635797252738152e-05, - "loss": 3.3018, + "learning_rate": 2.9668878992608674e-05, + "loss": 3.2526, "step": 122900 }, { "epoch": 0.41, - "learning_rate": 2.961922751901377e-05, - "loss": 3.2884, + "learning_rate": 2.9652336176491995e-05, + "loss": 3.2531, "step": 123000 }, { "epoch": 0.41, - "learning_rate": 2.9602657785289388e-05, - "loss": 3.3108, + "learning_rate": 2.9635793360375323e-05, + "loss": 3.2586, "step": 123100 }, { "epoch": 0.41, - "learning_rate": 2.9586088051565013e-05, - "loss": 3.3018, + "learning_rate": 2.9619250544258655e-05, + "loss": 3.2502, "step": 123200 }, { "epoch": 0.41, - "learning_rate": 2.9569518317840634e-05, - "loss": 3.299, + "learning_rate": 2.9602707728141976e-05, + "loss": 3.2462, "step": 123300 }, { "epoch": 0.41, - "learning_rate": 2.9552948584116252e-05, - "loss": 3.3073, + "learning_rate": 2.9586164912025304e-05, + "loss": 3.2505, "step": 123400 }, { "epoch": 0.41, - "learning_rate": 2.9536378850391877e-05, - "loss": 3.3112, + "learning_rate": 2.9569622095908635e-05, + "loss": 3.2433, "step": 123500 }, { "epoch": 0.41, - "learning_rate": 2.9519809116667495e-05, - "loss": 3.3004, + "learning_rate": 2.9553079279791956e-05, + "loss": 3.255, "step": 123600 }, { "epoch": 0.41, - "learning_rate": 2.9503239382943117e-05, - "loss": 3.2962, + "learning_rate": 2.9536536463675284e-05, + "loss": 3.2574, "step": 123700 }, { "epoch": 0.41, - "learning_rate": 2.948666964921874e-05, - "loss": 3.3026, + "learning_rate": 2.9519993647558612e-05, + "loss": 3.2417, "step": 123800 }, { "epoch": 0.41, - "learning_rate": 2.947009991549436e-05, - "loss": 3.3034, + "learning_rate": 2.9503450831441937e-05, + "loss": 3.2446, "step": 123900 }, { "epoch": 0.41, - "learning_rate": 2.9453530181769978e-05, - "loss": 3.3162, + "learning_rate": 2.9486908015325265e-05, + "loss": 3.2475, "step": 124000 }, { "epoch": 0.41, - "learning_rate": 2.9436960448045603e-05, - "loss": 3.3043, + "learning_rate": 2.9470365199208592e-05, + "loss": 3.2457, "step": 124100 }, { "epoch": 0.41, - "learning_rate": 2.942039071432122e-05, - "loss": 3.3121, + "learning_rate": 2.9453822383091924e-05, + "loss": 3.2543, "step": 124200 }, { "epoch": 0.41, - "learning_rate": 2.9403820980596842e-05, - "loss": 3.3003, + "learning_rate": 2.9437279566975245e-05, + "loss": 3.2416, "step": 124300 }, { "epoch": 0.41, - "learning_rate": 2.9387251246872467e-05, - "loss": 3.3004, + "learning_rate": 2.9420736750858573e-05, + "loss": 3.2442, "step": 124400 }, { "epoch": 0.41, - "learning_rate": 2.9370681513148085e-05, - "loss": 3.2974, + "learning_rate": 2.9404193934741904e-05, + "loss": 3.2543, "step": 124500 }, { "epoch": 0.41, - "learning_rate": 2.9354111779423703e-05, - "loss": 3.3003, + "learning_rate": 2.9387651118625225e-05, + "loss": 3.2471, "step": 124600 }, { "epoch": 0.41, - "learning_rate": 2.9337542045699328e-05, - "loss": 3.2996, + "learning_rate": 2.9371108302508553e-05, + "loss": 3.2496, "step": 124700 }, { "epoch": 0.41, - "learning_rate": 2.932097231197495e-05, - "loss": 3.3008, + "learning_rate": 2.9354565486391885e-05, + "loss": 3.2618, "step": 124800 }, { "epoch": 0.41, - "learning_rate": 2.9304402578250568e-05, - "loss": 3.3039, + "learning_rate": 2.9338022670275206e-05, + "loss": 3.2468, "step": 124900 }, { "epoch": 0.41, - "learning_rate": 2.9287832844526192e-05, - "loss": 3.3138, + "learning_rate": 2.9321479854158534e-05, + "loss": 3.2503, "step": 125000 }, { "epoch": 0.41, - "learning_rate": 2.927126311080181e-05, - "loss": 3.2872, + "learning_rate": 2.930493703804186e-05, + "loss": 3.2449, "step": 125100 }, { "epoch": 0.41, - "learning_rate": 2.925469337707743e-05, - "loss": 3.3055, + "learning_rate": 2.9288394221925186e-05, + "loss": 3.2516, "step": 125200 }, { - "epoch": 0.42, - "learning_rate": 2.9238123643353053e-05, - "loss": 3.3003, + "epoch": 0.41, + "learning_rate": 2.9271851405808514e-05, + "loss": 3.2648, "step": 125300 }, { - "epoch": 0.42, - "learning_rate": 2.9221553909628675e-05, - "loss": 3.2977, + "epoch": 0.41, + "learning_rate": 2.9255308589691842e-05, + "loss": 3.2436, "step": 125400 }, { "epoch": 0.42, - "learning_rate": 2.9204984175904293e-05, - "loss": 3.297, + "learning_rate": 2.9238765773575166e-05, + "loss": 3.2397, "step": 125500 }, { "epoch": 0.42, - "learning_rate": 2.9188414442179918e-05, - "loss": 3.3013, + "learning_rate": 2.9222222957458494e-05, + "loss": 3.2397, "step": 125600 }, { "epoch": 0.42, - "learning_rate": 2.9171844708455536e-05, - "loss": 3.3002, + "learning_rate": 2.9205680141341822e-05, + "loss": 3.2437, "step": 125700 }, { "epoch": 0.42, - "learning_rate": 2.9155274974731157e-05, - "loss": 3.3054, + "learning_rate": 2.9189137325225147e-05, + "loss": 3.2428, "step": 125800 }, { "epoch": 0.42, - "learning_rate": 2.9138705241006782e-05, - "loss": 3.3099, + "learning_rate": 2.9172594509108475e-05, + "loss": 3.2562, "step": 125900 }, { "epoch": 0.42, - "learning_rate": 2.91221355072824e-05, - "loss": 3.3053, + "learning_rate": 2.9156051692991803e-05, + "loss": 3.2584, "step": 126000 }, { "epoch": 0.42, - "learning_rate": 2.910556577355802e-05, - "loss": 3.3099, + "learning_rate": 2.9139508876875134e-05, + "loss": 3.2693, "step": 126100 }, { "epoch": 0.42, - "learning_rate": 2.9088996039833643e-05, - "loss": 3.2849, + "learning_rate": 2.9122966060758455e-05, + "loss": 3.2517, "step": 126200 }, { "epoch": 0.42, - "learning_rate": 2.907242630610926e-05, - "loss": 3.3061, + "learning_rate": 2.9106423244641783e-05, + "loss": 3.2631, "step": 126300 }, { "epoch": 0.42, - "learning_rate": 2.9055856572384883e-05, - "loss": 3.3076, + "learning_rate": 2.908988042852511e-05, + "loss": 3.2408, "step": 126400 }, { "epoch": 0.42, - "learning_rate": 2.9039286838660508e-05, - "loss": 3.306, + "learning_rate": 2.9073337612408435e-05, + "loss": 3.2568, "step": 126500 }, { "epoch": 0.42, - "learning_rate": 2.9022717104936126e-05, - "loss": 3.3118, + "learning_rate": 2.9056794796291763e-05, + "loss": 3.2513, "step": 126600 }, { "epoch": 0.42, - "learning_rate": 2.9006147371211744e-05, - "loss": 3.2876, + "learning_rate": 2.904025198017509e-05, + "loss": 3.2604, "step": 126700 }, { "epoch": 0.42, - "learning_rate": 2.898957763748737e-05, - "loss": 3.3001, + "learning_rate": 2.9023709164058416e-05, + "loss": 3.2353, "step": 126800 }, { "epoch": 0.42, - "learning_rate": 2.897300790376299e-05, - "loss": 3.2971, + "learning_rate": 2.9007166347941744e-05, + "loss": 3.2449, "step": 126900 }, { "epoch": 0.42, - "learning_rate": 2.8956438170038608e-05, - "loss": 3.3094, + "learning_rate": 2.8990623531825072e-05, + "loss": 3.2629, "step": 127000 }, { "epoch": 0.42, - "learning_rate": 2.8939868436314233e-05, - "loss": 3.3123, + "learning_rate": 2.8974080715708396e-05, + "loss": 3.2463, "step": 127100 }, { "epoch": 0.42, - "learning_rate": 2.892329870258985e-05, - "loss": 3.2966, + "learning_rate": 2.8957537899591724e-05, + "loss": 3.2455, "step": 127200 }, { "epoch": 0.42, - "learning_rate": 2.890672896886547e-05, - "loss": 3.3004, + "learning_rate": 2.8940995083475052e-05, + "loss": 3.2578, "step": 127300 }, { "epoch": 0.42, - "learning_rate": 2.889015923514109e-05, - "loss": 3.295, + "learning_rate": 2.8924452267358377e-05, + "loss": 3.2416, "step": 127400 }, { "epoch": 0.42, - "learning_rate": 2.8873589501416716e-05, - "loss": 3.2953, + "learning_rate": 2.8907909451241705e-05, + "loss": 3.2529, "step": 127500 }, { "epoch": 0.42, - "learning_rate": 2.8857019767692334e-05, - "loss": 3.2945, + "learning_rate": 2.8891366635125032e-05, + "loss": 3.2408, "step": 127600 }, { "epoch": 0.42, - "learning_rate": 2.8840450033967952e-05, - "loss": 3.287, + "learning_rate": 2.8874823819008357e-05, + "loss": 3.2489, "step": 127700 }, { "epoch": 0.42, - "learning_rate": 2.8823880300243577e-05, - "loss": 3.2975, + "learning_rate": 2.8858281002891685e-05, + "loss": 3.2468, "step": 127800 }, { "epoch": 0.42, - "learning_rate": 2.8807310566519198e-05, - "loss": 3.3091, + "learning_rate": 2.8841738186775013e-05, + "loss": 3.252, "step": 127900 }, { "epoch": 0.42, - "learning_rate": 2.8790740832794816e-05, - "loss": 3.3129, + "learning_rate": 2.8825195370658337e-05, + "loss": 3.237, "step": 128000 }, { "epoch": 0.42, - "learning_rate": 2.877417109907044e-05, - "loss": 3.2858, + "learning_rate": 2.8808652554541665e-05, + "loss": 3.2453, "step": 128100 }, { "epoch": 0.42, - "learning_rate": 2.875760136534606e-05, - "loss": 3.3042, + "learning_rate": 2.8792109738424993e-05, + "loss": 3.2345, "step": 128200 }, { - "epoch": 0.43, - "learning_rate": 2.8741031631621677e-05, - "loss": 3.2994, + "epoch": 0.42, + "learning_rate": 2.877556692230832e-05, + "loss": 3.2482, "step": 128300 }, { - "epoch": 0.43, - "learning_rate": 2.8724461897897302e-05, - "loss": 3.3014, + "epoch": 0.42, + "learning_rate": 2.8759024106191646e-05, + "loss": 3.2486, "step": 128400 }, { "epoch": 0.43, - "learning_rate": 2.8707892164172923e-05, - "loss": 3.3026, + "learning_rate": 2.8742481290074974e-05, + "loss": 3.2381, "step": 128500 }, { "epoch": 0.43, - "learning_rate": 2.869132243044854e-05, - "loss": 3.3076, + "learning_rate": 2.87259384739583e-05, + "loss": 3.235, "step": 128600 }, { "epoch": 0.43, - "learning_rate": 2.8674752696724166e-05, - "loss": 3.2989, + "learning_rate": 2.8709395657841626e-05, + "loss": 3.2541, "step": 128700 }, { "epoch": 0.43, - "learning_rate": 2.8658182962999784e-05, - "loss": 3.3039, + "learning_rate": 2.8692852841724954e-05, + "loss": 3.2401, "step": 128800 }, { "epoch": 0.43, - "learning_rate": 2.8641613229275406e-05, - "loss": 3.2934, + "learning_rate": 2.8676310025608282e-05, + "loss": 3.2602, "step": 128900 }, { "epoch": 0.43, - "learning_rate": 2.862504349555103e-05, - "loss": 3.2969, + "learning_rate": 2.8659767209491606e-05, + "loss": 3.2464, "step": 129000 }, { "epoch": 0.43, - "learning_rate": 2.860847376182665e-05, - "loss": 3.3093, + "learning_rate": 2.8643224393374934e-05, + "loss": 3.23, "step": 129100 }, { "epoch": 0.43, - "learning_rate": 2.8591904028102267e-05, - "loss": 3.2908, + "learning_rate": 2.8626681577258262e-05, + "loss": 3.2497, "step": 129200 }, { "epoch": 0.43, - "learning_rate": 2.8575334294377892e-05, - "loss": 3.297, + "learning_rate": 2.8610138761141587e-05, + "loss": 3.2428, "step": 129300 }, { "epoch": 0.43, - "learning_rate": 2.855876456065351e-05, - "loss": 3.3002, + "learning_rate": 2.8593595945024915e-05, + "loss": 3.2494, "step": 129400 }, { "epoch": 0.43, - "learning_rate": 2.854219482692913e-05, - "loss": 3.2863, + "learning_rate": 2.8577053128908243e-05, + "loss": 3.249, "step": 129500 }, { "epoch": 0.43, - "learning_rate": 2.8525625093204756e-05, - "loss": 3.298, + "learning_rate": 2.8560510312791567e-05, + "loss": 3.2527, "step": 129600 }, { "epoch": 0.43, - "learning_rate": 2.8509055359480374e-05, - "loss": 3.3069, + "learning_rate": 2.8543967496674895e-05, + "loss": 3.2407, "step": 129700 }, { "epoch": 0.43, - "learning_rate": 2.8492485625755992e-05, - "loss": 3.3084, + "learning_rate": 2.8527424680558223e-05, + "loss": 3.259, "step": 129800 }, { "epoch": 0.43, - "learning_rate": 2.8475915892031617e-05, - "loss": 3.2898, + "learning_rate": 2.8510881864441548e-05, + "loss": 3.2507, "step": 129900 }, { "epoch": 0.43, - "learning_rate": 2.845934615830724e-05, - "loss": 3.292, + "learning_rate": 2.8494339048324876e-05, + "loss": 3.2468, "step": 130000 }, { "epoch": 0.43, - "learning_rate": 2.8442776424582857e-05, - "loss": 3.3084, + "learning_rate": 2.8477796232208203e-05, + "loss": 3.2577, "step": 130100 }, { "epoch": 0.43, - "learning_rate": 2.842620669085848e-05, - "loss": 3.3206, + "learning_rate": 2.8461253416091528e-05, + "loss": 3.2545, "step": 130200 }, { "epoch": 0.43, - "learning_rate": 2.84096369571341e-05, - "loss": 3.2766, + "learning_rate": 2.8444710599974856e-05, + "loss": 3.2605, "step": 130300 }, { "epoch": 0.43, - "learning_rate": 2.839306722340972e-05, - "loss": 3.2894, + "learning_rate": 2.8428167783858184e-05, + "loss": 3.2484, "step": 130400 }, { "epoch": 0.43, - "learning_rate": 2.8376497489685343e-05, - "loss": 3.2809, + "learning_rate": 2.8411624967741512e-05, + "loss": 3.2433, "step": 130500 }, { "epoch": 0.43, - "learning_rate": 2.8359927755960964e-05, - "loss": 3.2963, + "learning_rate": 2.8395082151624836e-05, + "loss": 3.2461, "step": 130600 }, { "epoch": 0.43, - "learning_rate": 2.8343358022236582e-05, - "loss": 3.2833, + "learning_rate": 2.8378539335508164e-05, + "loss": 3.2434, "step": 130700 }, { "epoch": 0.43, - "learning_rate": 2.8326788288512207e-05, - "loss": 3.3102, + "learning_rate": 2.8361996519391492e-05, + "loss": 3.246, "step": 130800 }, { "epoch": 0.43, - "learning_rate": 2.8310218554787825e-05, - "loss": 3.2951, + "learning_rate": 2.8345453703274817e-05, + "loss": 3.2424, "step": 130900 }, { "epoch": 0.43, - "learning_rate": 2.8293648821063447e-05, - "loss": 3.2916, + "learning_rate": 2.8328910887158145e-05, + "loss": 3.2442, "step": 131000 }, { "epoch": 0.43, - "learning_rate": 2.827707908733907e-05, - "loss": 3.3017, + "learning_rate": 2.8312368071041473e-05, + "loss": 3.2511, "step": 131100 }, { "epoch": 0.43, - "learning_rate": 2.826050935361469e-05, - "loss": 3.295, + "learning_rate": 2.8295825254924797e-05, + "loss": 3.2481, "step": 131200 }, { - "epoch": 0.44, - "learning_rate": 2.8243939619890308e-05, - "loss": 3.2942, + "epoch": 0.43, + "learning_rate": 2.8279282438808125e-05, + "loss": 3.2603, "step": 131300 }, { - "epoch": 0.44, - "learning_rate": 2.8227369886165932e-05, - "loss": 3.2774, + "epoch": 0.43, + "learning_rate": 2.8262739622691453e-05, + "loss": 3.2761, "step": 131400 }, { "epoch": 0.44, - "learning_rate": 2.821080015244155e-05, - "loss": 3.2939, + "learning_rate": 2.8246196806574777e-05, + "loss": 3.2635, "step": 131500 }, { "epoch": 0.44, - "learning_rate": 2.8194230418717172e-05, - "loss": 3.284, + "learning_rate": 2.8229653990458105e-05, + "loss": 3.245, "step": 131600 }, { "epoch": 0.44, - "learning_rate": 2.8177660684992797e-05, - "loss": 3.2989, + "learning_rate": 2.8213111174341433e-05, + "loss": 3.256, "step": 131700 }, { "epoch": 0.44, - "learning_rate": 2.8161090951268415e-05, - "loss": 3.2836, + "learning_rate": 2.8196568358224758e-05, + "loss": 3.2481, "step": 131800 }, { "epoch": 0.44, - "learning_rate": 2.8144521217544033e-05, - "loss": 3.306, + "learning_rate": 2.8180025542108086e-05, + "loss": 3.2468, "step": 131900 }, { "epoch": 0.44, - "learning_rate": 2.8127951483819658e-05, - "loss": 3.2928, + "learning_rate": 2.8163482725991414e-05, + "loss": 3.2462, "step": 132000 }, { "epoch": 0.44, - "learning_rate": 2.811138175009528e-05, - "loss": 3.2873, + "learning_rate": 2.8146939909874738e-05, + "loss": 3.2508, "step": 132100 }, { "epoch": 0.44, - "learning_rate": 2.8094812016370897e-05, - "loss": 3.3073, + "learning_rate": 2.8130397093758066e-05, + "loss": 3.2519, "step": 132200 }, { "epoch": 0.44, - "learning_rate": 2.8078242282646522e-05, - "loss": 3.3014, + "learning_rate": 2.8113854277641394e-05, + "loss": 3.2456, "step": 132300 }, { "epoch": 0.44, - "learning_rate": 2.806167254892214e-05, - "loss": 3.3077, + "learning_rate": 2.8097311461524722e-05, + "loss": 3.2385, "step": 132400 }, { "epoch": 0.44, - "learning_rate": 2.8045102815197762e-05, - "loss": 3.2962, + "learning_rate": 2.8080768645408046e-05, + "loss": 3.2466, "step": 132500 }, { "epoch": 0.44, - "learning_rate": 2.802853308147338e-05, - "loss": 3.2964, + "learning_rate": 2.8064225829291374e-05, + "loss": 3.2453, "step": 132600 }, { "epoch": 0.44, - "learning_rate": 2.8011963347749005e-05, - "loss": 3.2947, + "learning_rate": 2.8047683013174702e-05, + "loss": 3.2365, "step": 132700 }, { "epoch": 0.44, - "learning_rate": 2.7995393614024623e-05, - "loss": 3.3046, + "learning_rate": 2.8031140197058027e-05, + "loss": 3.2478, "step": 132800 }, { "epoch": 0.44, - "learning_rate": 2.797882388030024e-05, - "loss": 3.2973, + "learning_rate": 2.8014597380941355e-05, + "loss": 3.2321, "step": 132900 }, { "epoch": 0.44, - "learning_rate": 2.7962254146575866e-05, - "loss": 3.2993, + "learning_rate": 2.7998054564824683e-05, + "loss": 3.2555, "step": 133000 }, { "epoch": 0.44, - "learning_rate": 2.7945684412851487e-05, - "loss": 3.287, + "learning_rate": 2.7981511748708007e-05, + "loss": 3.2502, "step": 133100 }, { "epoch": 0.44, - "learning_rate": 2.7929114679127105e-05, - "loss": 3.3165, + "learning_rate": 2.7964968932591335e-05, + "loss": 3.2376, "step": 133200 }, { "epoch": 0.44, - "learning_rate": 2.791254494540273e-05, - "loss": 3.3075, + "learning_rate": 2.7948426116474663e-05, + "loss": 3.2628, "step": 133300 }, { "epoch": 0.44, - "learning_rate": 2.7895975211678348e-05, - "loss": 3.2942, + "learning_rate": 2.7931883300357988e-05, + "loss": 3.239, "step": 133400 }, { "epoch": 0.44, - "learning_rate": 2.787940547795397e-05, - "loss": 3.3023, + "learning_rate": 2.7915340484241316e-05, + "loss": 3.2512, "step": 133500 }, { "epoch": 0.44, - "learning_rate": 2.786283574422959e-05, - "loss": 3.2893, + "learning_rate": 2.7898797668124643e-05, + "loss": 3.2568, "step": 133600 }, { "epoch": 0.44, - "learning_rate": 2.7846266010505213e-05, - "loss": 3.2927, + "learning_rate": 2.7882254852007965e-05, + "loss": 3.2496, "step": 133700 }, { "epoch": 0.44, - "learning_rate": 2.782969627678083e-05, - "loss": 3.2997, + "learning_rate": 2.7865712035891296e-05, + "loss": 3.2565, "step": 133800 }, { "epoch": 0.44, - "learning_rate": 2.7813126543056456e-05, - "loss": 3.2947, + "learning_rate": 2.7849169219774624e-05, + "loss": 3.2547, "step": 133900 }, { "epoch": 0.44, - "learning_rate": 2.7796556809332074e-05, - "loss": 3.3037, + "learning_rate": 2.7832626403657945e-05, + "loss": 3.2583, "step": 134000 }, { "epoch": 0.44, - "learning_rate": 2.7779987075607695e-05, - "loss": 3.2895, + "learning_rate": 2.7816083587541276e-05, + "loss": 3.2532, "step": 134100 }, { "epoch": 0.44, - "learning_rate": 2.776341734188332e-05, - "loss": 3.3041, + "learning_rate": 2.7799540771424604e-05, + "loss": 3.2361, "step": 134200 }, { - "epoch": 0.45, - "learning_rate": 2.7746847608158938e-05, - "loss": 3.3082, + "epoch": 0.44, + "learning_rate": 2.7782997955307925e-05, + "loss": 3.2337, "step": 134300 }, { - "epoch": 0.45, - "learning_rate": 2.7730277874434556e-05, - "loss": 3.3116, + "epoch": 0.44, + "learning_rate": 2.7766455139191257e-05, + "loss": 3.2601, "step": 134400 }, { "epoch": 0.45, - "learning_rate": 2.771370814071018e-05, - "loss": 3.287, + "learning_rate": 2.7749912323074585e-05, + "loss": 3.2302, "step": 134500 }, { "epoch": 0.45, - "learning_rate": 2.7697138406985802e-05, - "loss": 3.3, + "learning_rate": 2.7733369506957913e-05, + "loss": 3.2384, "step": 134600 }, { "epoch": 0.45, - "learning_rate": 2.768056867326142e-05, - "loss": 3.3129, + "learning_rate": 2.7716826690841237e-05, + "loss": 3.2524, "step": 134700 }, { "epoch": 0.45, - "learning_rate": 2.7663998939537045e-05, - "loss": 3.2982, + "learning_rate": 2.7700283874724565e-05, + "loss": 3.2321, "step": 134800 }, { "epoch": 0.45, - "learning_rate": 2.7647429205812663e-05, - "loss": 3.307, + "learning_rate": 2.7683741058607893e-05, + "loss": 3.2429, "step": 134900 }, { "epoch": 0.45, - "learning_rate": 2.763085947208828e-05, - "loss": 3.3101, + "learning_rate": 2.7667198242491214e-05, + "loss": 3.2487, "step": 135000 }, { "epoch": 0.45, - "learning_rate": 2.7614289738363906e-05, - "loss": 3.3056, + "learning_rate": 2.7650655426374545e-05, + "loss": 3.2382, "step": 135100 }, { "epoch": 0.45, - "learning_rate": 2.7597720004639528e-05, - "loss": 3.2811, + "learning_rate": 2.7634112610257873e-05, + "loss": 3.2676, "step": 135200 }, { "epoch": 0.45, - "learning_rate": 2.7581150270915146e-05, - "loss": 3.3042, + "learning_rate": 2.7617569794141194e-05, + "loss": 3.2454, "step": 135300 }, { "epoch": 0.45, - "learning_rate": 2.756458053719077e-05, - "loss": 3.2883, + "learning_rate": 2.7601026978024526e-05, + "loss": 3.2544, "step": 135400 }, { "epoch": 0.45, - "learning_rate": 2.754801080346639e-05, - "loss": 3.2926, + "learning_rate": 2.7584484161907854e-05, + "loss": 3.2345, "step": 135500 }, { "epoch": 0.45, - "learning_rate": 2.753144106974201e-05, - "loss": 3.2904, + "learning_rate": 2.7567941345791175e-05, + "loss": 3.2416, "step": 135600 }, { "epoch": 0.45, - "learning_rate": 2.7514871336017635e-05, - "loss": 3.2884, + "learning_rate": 2.7551398529674506e-05, + "loss": 3.2517, "step": 135700 }, { "epoch": 0.45, - "learning_rate": 2.7498301602293253e-05, - "loss": 3.2853, + "learning_rate": 2.7534855713557834e-05, + "loss": 3.2491, "step": 135800 }, { "epoch": 0.45, - "learning_rate": 2.748173186856887e-05, - "loss": 3.2871, + "learning_rate": 2.7518312897441155e-05, + "loss": 3.2563, "step": 135900 }, { "epoch": 0.45, - "learning_rate": 2.7465162134844496e-05, - "loss": 3.2896, + "learning_rate": 2.7501770081324487e-05, + "loss": 3.2486, "step": 136000 }, { "epoch": 0.45, - "learning_rate": 2.7448592401120114e-05, - "loss": 3.2948, + "learning_rate": 2.7485227265207814e-05, + "loss": 3.2351, "step": 136100 }, { "epoch": 0.45, - "learning_rate": 2.7432022667395736e-05, - "loss": 3.2945, + "learning_rate": 2.7468684449091136e-05, + "loss": 3.2413, "step": 136200 }, { "epoch": 0.45, - "learning_rate": 2.741545293367136e-05, - "loss": 3.2969, + "learning_rate": 2.7452141632974464e-05, + "loss": 3.2481, "step": 136300 }, { "epoch": 0.45, - "learning_rate": 2.739888319994698e-05, - "loss": 3.3025, + "learning_rate": 2.7435598816857795e-05, + "loss": 3.2446, "step": 136400 }, { "epoch": 0.45, - "learning_rate": 2.7382313466222597e-05, - "loss": 3.2844, + "learning_rate": 2.7419056000741116e-05, + "loss": 3.2651, "step": 136500 }, { "epoch": 0.45, - "learning_rate": 2.736574373249822e-05, - "loss": 3.2997, + "learning_rate": 2.7402513184624444e-05, + "loss": 3.2315, "step": 136600 }, { "epoch": 0.45, - "learning_rate": 2.7349173998773843e-05, - "loss": 3.2959, + "learning_rate": 2.7385970368507775e-05, + "loss": 3.2531, "step": 136700 }, { "epoch": 0.45, - "learning_rate": 2.733260426504946e-05, - "loss": 3.2987, + "learning_rate": 2.7369427552391103e-05, + "loss": 3.2483, "step": 136800 }, { "epoch": 0.45, - "learning_rate": 2.7316034531325086e-05, - "loss": 3.2923, + "learning_rate": 2.7352884736274424e-05, + "loss": 3.2523, "step": 136900 }, { "epoch": 0.45, - "learning_rate": 2.7299464797600704e-05, - "loss": 3.2997, + "learning_rate": 2.7336341920157756e-05, + "loss": 3.2586, "step": 137000 }, { "epoch": 0.45, - "learning_rate": 2.7282895063876322e-05, - "loss": 3.297, + "learning_rate": 2.7319799104041083e-05, + "loss": 3.2489, "step": 137100 }, { "epoch": 0.45, - "learning_rate": 2.7266325330151947e-05, - "loss": 3.2881, + "learning_rate": 2.7303256287924405e-05, + "loss": 3.2367, "step": 137200 }, { - "epoch": 0.46, - "learning_rate": 2.724975559642757e-05, - "loss": 3.2917, + "epoch": 0.45, + "learning_rate": 2.7286713471807736e-05, + "loss": 3.2464, "step": 137300 }, { - "epoch": 0.46, - "learning_rate": 2.7233185862703187e-05, - "loss": 3.3049, + "epoch": 0.45, + "learning_rate": 2.7270170655691064e-05, + "loss": 3.238, "step": 137400 }, { - "epoch": 0.46, - "learning_rate": 2.721661612897881e-05, - "loss": 3.298, + "epoch": 0.45, + "learning_rate": 2.7253627839574385e-05, + "loss": 3.2439, "step": 137500 }, { "epoch": 0.46, - "learning_rate": 2.720004639525443e-05, - "loss": 3.2935, + "learning_rate": 2.7237085023457713e-05, + "loss": 3.2541, "step": 137600 }, { "epoch": 0.46, - "learning_rate": 2.718347666153005e-05, - "loss": 3.2981, + "learning_rate": 2.7220542207341044e-05, + "loss": 3.2535, "step": 137700 }, { "epoch": 0.46, - "learning_rate": 2.716690692780567e-05, - "loss": 3.2924, + "learning_rate": 2.7203999391224365e-05, + "loss": 3.2317, "step": 137800 }, { "epoch": 0.46, - "learning_rate": 2.7150337194081294e-05, - "loss": 3.2931, + "learning_rate": 2.7187456575107693e-05, + "loss": 3.2556, "step": 137900 }, { "epoch": 0.46, - "learning_rate": 2.7133767460356912e-05, - "loss": 3.2874, + "learning_rate": 2.7170913758991025e-05, + "loss": 3.2629, "step": 138000 }, { "epoch": 0.46, - "learning_rate": 2.711719772663253e-05, - "loss": 3.302, + "learning_rate": 2.7154370942874346e-05, + "loss": 3.2423, "step": 138100 }, { "epoch": 0.46, - "learning_rate": 2.7100627992908155e-05, - "loss": 3.2783, + "learning_rate": 2.7137828126757674e-05, + "loss": 3.2539, "step": 138200 }, { "epoch": 0.46, - "learning_rate": 2.7084058259183776e-05, - "loss": 3.2954, + "learning_rate": 2.7121285310641005e-05, + "loss": 3.2593, "step": 138300 }, { "epoch": 0.46, - "learning_rate": 2.7067488525459394e-05, - "loss": 3.2982, + "learning_rate": 2.7104742494524326e-05, + "loss": 3.2425, "step": 138400 }, { "epoch": 0.46, - "learning_rate": 2.705091879173502e-05, - "loss": 3.2864, + "learning_rate": 2.7088199678407654e-05, + "loss": 3.2566, "step": 138500 }, { "epoch": 0.46, - "learning_rate": 2.7034349058010637e-05, - "loss": 3.2909, + "learning_rate": 2.7071656862290985e-05, + "loss": 3.2485, "step": 138600 }, { "epoch": 0.46, - "learning_rate": 2.701777932428626e-05, - "loss": 3.2948, + "learning_rate": 2.7055114046174307e-05, + "loss": 3.2414, "step": 138700 }, { "epoch": 0.46, - "learning_rate": 2.7001209590561884e-05, - "loss": 3.3039, + "learning_rate": 2.7038571230057634e-05, + "loss": 3.2584, "step": 138800 }, { "epoch": 0.46, - "learning_rate": 2.6984639856837502e-05, - "loss": 3.2887, + "learning_rate": 2.7022028413940962e-05, + "loss": 3.257, "step": 138900 }, { "epoch": 0.46, - "learning_rate": 2.696807012311312e-05, - "loss": 3.2984, + "learning_rate": 2.7005485597824294e-05, + "loss": 3.2496, "step": 139000 }, { "epoch": 0.46, - "learning_rate": 2.6951500389388745e-05, - "loss": 3.2995, + "learning_rate": 2.6988942781707615e-05, + "loss": 3.2441, "step": 139100 }, { "epoch": 0.46, - "learning_rate": 2.6934930655664363e-05, - "loss": 3.3007, + "learning_rate": 2.6972399965590943e-05, + "loss": 3.2497, "step": 139200 }, { "epoch": 0.46, - "learning_rate": 2.6918360921939984e-05, - "loss": 3.2884, + "learning_rate": 2.6955857149474274e-05, + "loss": 3.245, "step": 139300 }, { "epoch": 0.46, - "learning_rate": 2.690179118821561e-05, - "loss": 3.2854, + "learning_rate": 2.6939314333357595e-05, + "loss": 3.2491, "step": 139400 }, { "epoch": 0.46, - "learning_rate": 2.6885221454491227e-05, - "loss": 3.2984, + "learning_rate": 2.6922771517240923e-05, + "loss": 3.2447, "step": 139500 }, { "epoch": 0.46, - "learning_rate": 2.6868651720766845e-05, - "loss": 3.2921, + "learning_rate": 2.6906228701124254e-05, + "loss": 3.241, "step": 139600 }, { "epoch": 0.46, - "learning_rate": 2.685208198704247e-05, - "loss": 3.2915, + "learning_rate": 2.6889685885007576e-05, + "loss": 3.2374, "step": 139700 }, { "epoch": 0.46, - "learning_rate": 2.683551225331809e-05, - "loss": 3.2999, + "learning_rate": 2.6873143068890904e-05, + "loss": 3.2395, "step": 139800 }, { "epoch": 0.46, - "learning_rate": 2.681894251959371e-05, - "loss": 3.3036, + "learning_rate": 2.6856600252774235e-05, + "loss": 3.2537, "step": 139900 }, { "epoch": 0.46, - "learning_rate": 2.6802372785869335e-05, - "loss": 3.2935, + "learning_rate": 2.6840057436657556e-05, + "loss": 3.2505, "step": 140000 }, { "epoch": 0.46, - "learning_rate": 2.6785803052144953e-05, - "loss": 3.2907, + "learning_rate": 2.6823514620540884e-05, + "loss": 3.2474, "step": 140100 }, { "epoch": 0.46, - "learning_rate": 2.676923331842057e-05, - "loss": 3.2788, + "learning_rate": 2.6806971804424212e-05, + "loss": 3.2571, "step": 140200 }, { "epoch": 0.46, - "learning_rate": 2.6752663584696196e-05, - "loss": 3.2891, + "learning_rate": 2.6790428988307536e-05, + "loss": 3.2573, "step": 140300 }, { - "epoch": 0.47, - "learning_rate": 2.6736093850971817e-05, - "loss": 3.3041, + "epoch": 0.46, + "learning_rate": 2.6773886172190864e-05, + "loss": 3.2363, "step": 140400 }, { - "epoch": 0.47, - "learning_rate": 2.6719524117247435e-05, - "loss": 3.2928, + "epoch": 0.46, + "learning_rate": 2.6757343356074192e-05, + "loss": 3.243, "step": 140500 }, { "epoch": 0.47, - "learning_rate": 2.670295438352306e-05, - "loss": 3.3077, + "learning_rate": 2.6740800539957517e-05, + "loss": 3.2355, "step": 140600 }, { "epoch": 0.47, - "learning_rate": 2.6686384649798678e-05, - "loss": 3.2869, + "learning_rate": 2.6724257723840845e-05, + "loss": 3.2466, "step": 140700 }, { "epoch": 0.47, - "learning_rate": 2.66698149160743e-05, - "loss": 3.2886, + "learning_rate": 2.6707714907724173e-05, + "loss": 3.2434, "step": 140800 }, { "epoch": 0.47, - "learning_rate": 2.6653245182349924e-05, - "loss": 3.3035, + "learning_rate": 2.6691172091607504e-05, + "loss": 3.2513, "step": 140900 }, { "epoch": 0.47, - "learning_rate": 2.6636675448625542e-05, - "loss": 3.2928, + "learning_rate": 2.6674629275490825e-05, + "loss": 3.2277, "step": 141000 }, { "epoch": 0.47, - "learning_rate": 2.662010571490116e-05, - "loss": 3.3072, + "learning_rate": 2.6658086459374153e-05, + "loss": 3.2501, "step": 141100 }, { "epoch": 0.47, - "learning_rate": 2.6603535981176785e-05, - "loss": 3.306, + "learning_rate": 2.6641543643257484e-05, + "loss": 3.2502, "step": 141200 }, { "epoch": 0.47, - "learning_rate": 2.6586966247452403e-05, - "loss": 3.2948, + "learning_rate": 2.6625000827140805e-05, + "loss": 3.241, "step": 141300 }, { "epoch": 0.47, - "learning_rate": 2.6570396513728025e-05, - "loss": 3.2868, + "learning_rate": 2.6608458011024133e-05, + "loss": 3.2506, "step": 141400 }, { "epoch": 0.47, - "learning_rate": 2.655382678000365e-05, - "loss": 3.2869, + "learning_rate": 2.659191519490746e-05, + "loss": 3.2479, "step": 141500 }, { "epoch": 0.47, - "learning_rate": 2.6537257046279268e-05, - "loss": 3.3043, + "learning_rate": 2.6575372378790786e-05, + "loss": 3.2357, "step": 141600 }, { "epoch": 0.47, - "learning_rate": 2.6520687312554886e-05, - "loss": 3.2986, + "learning_rate": 2.6558829562674114e-05, + "loss": 3.2455, "step": 141700 }, { "epoch": 0.47, - "learning_rate": 2.650411757883051e-05, - "loss": 3.3005, + "learning_rate": 2.654228674655744e-05, + "loss": 3.2284, "step": 141800 }, { "epoch": 0.47, - "learning_rate": 2.6487547845106132e-05, - "loss": 3.2924, + "learning_rate": 2.6525743930440766e-05, + "loss": 3.2365, "step": 141900 }, { "epoch": 0.47, - "learning_rate": 2.647097811138175e-05, - "loss": 3.3035, + "learning_rate": 2.6509201114324094e-05, + "loss": 3.2395, "step": 142000 }, { "epoch": 0.47, - "learning_rate": 2.6454408377657375e-05, - "loss": 3.301, + "learning_rate": 2.6492658298207422e-05, + "loss": 3.2453, "step": 142100 }, { "epoch": 0.47, - "learning_rate": 2.6437838643932993e-05, - "loss": 3.297, + "learning_rate": 2.6476115482090747e-05, + "loss": 3.2582, "step": 142200 }, { "epoch": 0.47, - "learning_rate": 2.642126891020861e-05, - "loss": 3.2985, + "learning_rate": 2.6459572665974074e-05, + "loss": 3.2534, "step": 142300 }, { "epoch": 0.47, - "learning_rate": 2.6404699176484236e-05, - "loss": 3.2917, + "learning_rate": 2.6443029849857402e-05, + "loss": 3.2494, "step": 142400 }, { "epoch": 0.47, - "learning_rate": 2.6388129442759858e-05, - "loss": 3.2863, + "learning_rate": 2.6426487033740727e-05, + "loss": 3.2501, "step": 142500 }, { "epoch": 0.47, - "learning_rate": 2.6371559709035476e-05, - "loss": 3.3015, + "learning_rate": 2.6409944217624055e-05, + "loss": 3.2483, "step": 142600 }, { "epoch": 0.47, - "learning_rate": 2.63549899753111e-05, - "loss": 3.2924, + "learning_rate": 2.6393401401507383e-05, + "loss": 3.2448, "step": 142700 }, { "epoch": 0.47, - "learning_rate": 2.633842024158672e-05, - "loss": 3.2887, + "learning_rate": 2.6376858585390707e-05, + "loss": 3.2458, "step": 142800 }, { "epoch": 0.47, - "learning_rate": 2.632185050786234e-05, - "loss": 3.2985, + "learning_rate": 2.6360315769274035e-05, + "loss": 3.2339, "step": 142900 }, { "epoch": 0.47, - "learning_rate": 2.6305280774137958e-05, - "loss": 3.2949, + "learning_rate": 2.6343772953157363e-05, + "loss": 3.2389, "step": 143000 }, { "epoch": 0.47, - "learning_rate": 2.6288711040413583e-05, - "loss": 3.2983, + "learning_rate": 2.632723013704069e-05, + "loss": 3.248, "step": 143100 }, { "epoch": 0.47, - "learning_rate": 2.62721413066892e-05, - "loss": 3.292, + "learning_rate": 2.6310687320924016e-05, + "loss": 3.248, "step": 143200 }, { "epoch": 0.47, - "learning_rate": 2.6255571572964823e-05, - "loss": 3.2996, + "learning_rate": 2.6294144504807344e-05, + "loss": 3.2443, "step": 143300 }, { - "epoch": 0.48, - "learning_rate": 2.6239001839240444e-05, - "loss": 3.294, + "epoch": 0.47, + "learning_rate": 2.627760168869067e-05, + "loss": 3.2322, "step": 143400 }, { - "epoch": 0.48, - "learning_rate": 2.6222432105516066e-05, - "loss": 3.2952, + "epoch": 0.47, + "learning_rate": 2.6261058872573996e-05, + "loss": 3.2485, "step": 143500 }, { "epoch": 0.48, - "learning_rate": 2.6205862371791684e-05, - "loss": 3.2972, + "learning_rate": 2.6244516056457324e-05, + "loss": 3.2461, "step": 143600 }, { "epoch": 0.48, - "learning_rate": 2.618929263806731e-05, - "loss": 3.2853, + "learning_rate": 2.6227973240340652e-05, + "loss": 3.2513, "step": 143700 }, { "epoch": 0.48, - "learning_rate": 2.6172722904342927e-05, - "loss": 3.2903, + "learning_rate": 2.6211430424223976e-05, + "loss": 3.2507, "step": 143800 }, { "epoch": 0.48, - "learning_rate": 2.6156153170618548e-05, - "loss": 3.3031, + "learning_rate": 2.6194887608107304e-05, + "loss": 3.2368, "step": 143900 }, { "epoch": 0.48, - "learning_rate": 2.6139583436894173e-05, - "loss": 3.3021, + "learning_rate": 2.6178344791990632e-05, + "loss": 3.2569, "step": 144000 }, { "epoch": 0.48, - "learning_rate": 2.612301370316979e-05, - "loss": 3.2952, + "learning_rate": 2.6161801975873957e-05, + "loss": 3.2391, "step": 144100 }, { "epoch": 0.48, - "learning_rate": 2.610644396944541e-05, - "loss": 3.2861, + "learning_rate": 2.6145259159757285e-05, + "loss": 3.2445, "step": 144200 }, { "epoch": 0.48, - "learning_rate": 2.6089874235721034e-05, - "loss": 3.2972, + "learning_rate": 2.6128716343640613e-05, + "loss": 3.2527, "step": 144300 }, { "epoch": 0.48, - "learning_rate": 2.6073304501996655e-05, - "loss": 3.2912, + "learning_rate": 2.6112173527523937e-05, + "loss": 3.2566, "step": 144400 }, { "epoch": 0.48, - "learning_rate": 2.6056734768272273e-05, - "loss": 3.2879, + "learning_rate": 2.6095630711407265e-05, + "loss": 3.2522, "step": 144500 }, { "epoch": 0.48, - "learning_rate": 2.60401650345479e-05, - "loss": 3.2874, + "learning_rate": 2.6079087895290593e-05, + "loss": 3.2488, "step": 144600 }, { "epoch": 0.48, - "learning_rate": 2.6023595300823516e-05, - "loss": 3.3032, + "learning_rate": 2.6062545079173918e-05, + "loss": 3.2339, "step": 144700 }, { "epoch": 0.48, - "learning_rate": 2.6007025567099134e-05, - "loss": 3.3037, + "learning_rate": 2.6046002263057245e-05, + "loss": 3.2459, "step": 144800 }, { "epoch": 0.48, - "learning_rate": 2.599045583337476e-05, - "loss": 3.2904, + "learning_rate": 2.6029459446940573e-05, + "loss": 3.2285, "step": 144900 }, { "epoch": 0.48, - "learning_rate": 2.597388609965038e-05, - "loss": 3.2851, + "learning_rate": 2.6012916630823898e-05, + "loss": 3.2396, "step": 145000 }, { "epoch": 0.48, - "learning_rate": 2.5957316365926e-05, - "loss": 3.2855, + "learning_rate": 2.5996373814707226e-05, + "loss": 3.2382, "step": 145100 }, { "epoch": 0.48, - "learning_rate": 2.5940746632201624e-05, - "loss": 3.2909, + "learning_rate": 2.5979830998590554e-05, + "loss": 3.2432, "step": 145200 }, { "epoch": 0.48, - "learning_rate": 2.5924176898477242e-05, - "loss": 3.2836, + "learning_rate": 2.596328818247388e-05, + "loss": 3.2278, "step": 145300 }, { "epoch": 0.48, - "learning_rate": 2.5907607164752863e-05, - "loss": 3.2945, + "learning_rate": 2.5946745366357206e-05, + "loss": 3.2369, "step": 145400 }, { "epoch": 0.48, - "learning_rate": 2.5891037431028485e-05, - "loss": 3.2933, + "learning_rate": 2.5930202550240534e-05, + "loss": 3.2431, "step": 145500 }, { "epoch": 0.48, - "learning_rate": 2.5874467697304106e-05, - "loss": 3.2878, + "learning_rate": 2.5913659734123862e-05, + "loss": 3.2232, "step": 145600 }, { "epoch": 0.48, - "learning_rate": 2.5857897963579724e-05, - "loss": 3.2908, + "learning_rate": 2.5897116918007187e-05, + "loss": 3.2485, "step": 145700 }, { "epoch": 0.48, - "learning_rate": 2.584132822985535e-05, - "loss": 3.3071, + "learning_rate": 2.5880574101890515e-05, + "loss": 3.2377, "step": 145800 }, { "epoch": 0.48, - "learning_rate": 2.5824758496130967e-05, - "loss": 3.2979, + "learning_rate": 2.5864031285773842e-05, + "loss": 3.2466, "step": 145900 }, { "epoch": 0.48, - "learning_rate": 2.580818876240659e-05, - "loss": 3.2947, + "learning_rate": 2.5847488469657167e-05, + "loss": 3.2489, "step": 146000 }, { "epoch": 0.48, - "learning_rate": 2.5791619028682214e-05, - "loss": 3.2876, + "learning_rate": 2.5830945653540495e-05, + "loss": 3.2454, "step": 146100 }, { "epoch": 0.48, - "learning_rate": 2.577504929495783e-05, - "loss": 3.2871, + "learning_rate": 2.5814402837423823e-05, + "loss": 3.2343, "step": 146200 }, { "epoch": 0.48, - "learning_rate": 2.575847956123345e-05, - "loss": 3.2901, + "learning_rate": 2.5797860021307147e-05, + "loss": 3.2482, "step": 146300 }, { - "epoch": 0.49, - "learning_rate": 2.5741909827509075e-05, - "loss": 3.2882, + "epoch": 0.48, + "learning_rate": 2.5781317205190475e-05, + "loss": 3.2507, "step": 146400 }, { - "epoch": 0.49, - "learning_rate": 2.5725340093784696e-05, - "loss": 3.2812, + "epoch": 0.48, + "learning_rate": 2.5764774389073803e-05, + "loss": 3.249, "step": 146500 }, { "epoch": 0.49, - "learning_rate": 2.5708770360060314e-05, - "loss": 3.2955, + "learning_rate": 2.5748231572957128e-05, + "loss": 3.242, "step": 146600 }, { "epoch": 0.49, - "learning_rate": 2.569220062633594e-05, - "loss": 3.2957, + "learning_rate": 2.5731688756840456e-05, + "loss": 3.2409, "step": 146700 }, { "epoch": 0.49, - "learning_rate": 2.5675630892611557e-05, - "loss": 3.286, + "learning_rate": 2.5715145940723784e-05, + "loss": 3.2478, "step": 146800 }, { "epoch": 0.49, - "learning_rate": 2.5659061158887175e-05, - "loss": 3.285, + "learning_rate": 2.5698603124607108e-05, + "loss": 3.2529, "step": 146900 }, { "epoch": 0.49, - "learning_rate": 2.56424914251628e-05, - "loss": 3.2879, + "learning_rate": 2.5682060308490436e-05, + "loss": 3.2355, "step": 147000 }, { "epoch": 0.49, - "learning_rate": 2.562592169143842e-05, - "loss": 3.2942, + "learning_rate": 2.5665517492373764e-05, + "loss": 3.2612, "step": 147100 }, { "epoch": 0.49, - "learning_rate": 2.560935195771404e-05, - "loss": 3.2856, + "learning_rate": 2.564897467625709e-05, + "loss": 3.2357, "step": 147200 }, { "epoch": 0.49, - "learning_rate": 2.5592782223989664e-05, - "loss": 3.2979, + "learning_rate": 2.5632431860140416e-05, + "loss": 3.2423, "step": 147300 }, { "epoch": 0.49, - "learning_rate": 2.5576212490265282e-05, - "loss": 3.2765, + "learning_rate": 2.5615889044023744e-05, + "loss": 3.2307, "step": 147400 }, { "epoch": 0.49, - "learning_rate": 2.5559642756540904e-05, - "loss": 3.2969, + "learning_rate": 2.5599346227907072e-05, + "loss": 3.2371, "step": 147500 }, { "epoch": 0.49, - "learning_rate": 2.554307302281653e-05, - "loss": 3.2849, + "learning_rate": 2.5582803411790397e-05, + "loss": 3.2437, "step": 147600 }, { "epoch": 0.49, - "learning_rate": 2.5526503289092147e-05, - "loss": 3.2956, + "learning_rate": 2.5566260595673725e-05, + "loss": 3.2424, "step": 147700 }, { "epoch": 0.49, - "learning_rate": 2.5509933555367765e-05, - "loss": 3.3014, + "learning_rate": 2.5549717779557053e-05, + "loss": 3.2416, "step": 147800 }, { "epoch": 0.49, - "learning_rate": 2.549336382164339e-05, - "loss": 3.2756, + "learning_rate": 2.5533174963440377e-05, + "loss": 3.2388, "step": 147900 }, { "epoch": 0.49, - "learning_rate": 2.5476794087919008e-05, - "loss": 3.2808, + "learning_rate": 2.5516632147323705e-05, + "loss": 3.2432, "step": 148000 }, { "epoch": 0.49, - "learning_rate": 2.546022435419463e-05, - "loss": 3.2833, + "learning_rate": 2.5500089331207033e-05, + "loss": 3.235, "step": 148100 }, { "epoch": 0.49, - "learning_rate": 2.5443654620470247e-05, - "loss": 3.2976, + "learning_rate": 2.5483546515090358e-05, + "loss": 3.2333, "step": 148200 }, { "epoch": 0.49, - "learning_rate": 2.5427084886745872e-05, - "loss": 3.2927, + "learning_rate": 2.5467003698973685e-05, + "loss": 3.2502, "step": 148300 }, { "epoch": 0.49, - "learning_rate": 2.541051515302149e-05, - "loss": 3.2825, + "learning_rate": 2.5450460882857013e-05, + "loss": 3.2537, "step": 148400 }, { "epoch": 0.49, - "learning_rate": 2.5393945419297112e-05, - "loss": 3.2914, + "learning_rate": 2.5433918066740338e-05, + "loss": 3.2347, "step": 148500 }, { "epoch": 0.49, - "learning_rate": 2.5377375685572737e-05, - "loss": 3.2856, + "learning_rate": 2.5417375250623666e-05, + "loss": 3.2508, "step": 148600 }, { "epoch": 0.49, - "learning_rate": 2.5360805951848355e-05, - "loss": 3.2861, + "learning_rate": 2.5400832434506994e-05, + "loss": 3.2562, "step": 148700 }, { "epoch": 0.49, - "learning_rate": 2.5344236218123973e-05, - "loss": 3.2758, + "learning_rate": 2.538428961839032e-05, + "loss": 3.2432, "step": 148800 }, { "epoch": 0.49, - "learning_rate": 2.5327666484399598e-05, - "loss": 3.2882, + "learning_rate": 2.5367746802273646e-05, + "loss": 3.2292, "step": 148900 }, { "epoch": 0.49, - "learning_rate": 2.5311096750675216e-05, - "loss": 3.291, + "learning_rate": 2.5351203986156974e-05, + "loss": 3.2329, "step": 149000 }, { "epoch": 0.49, - "learning_rate": 2.5294527016950837e-05, - "loss": 3.2835, + "learning_rate": 2.5334661170040295e-05, + "loss": 3.2463, "step": 149100 }, { "epoch": 0.49, - "learning_rate": 2.5277957283226462e-05, - "loss": 3.2929, + "learning_rate": 2.5318118353923627e-05, + "loss": 3.2461, "step": 149200 }, { "epoch": 0.49, - "learning_rate": 2.526138754950208e-05, - "loss": 3.2892, + "learning_rate": 2.5301575537806955e-05, + "loss": 3.2421, "step": 149300 }, { - "epoch": 0.5, - "learning_rate": 2.5244817815777698e-05, - "loss": 3.271, + "epoch": 0.49, + "learning_rate": 2.5285032721690282e-05, + "loss": 3.2611, "step": 149400 }, { - "epoch": 0.5, - "learning_rate": 2.5228248082053323e-05, - "loss": 3.2966, + "epoch": 0.49, + "learning_rate": 2.5268489905573607e-05, + "loss": 3.2355, "step": 149500 }, { - "epoch": 0.5, - "learning_rate": 2.5211678348328945e-05, - "loss": 3.2979, + "epoch": 0.49, + "learning_rate": 2.5251947089456935e-05, + "loss": 3.2306, "step": 149600 }, { "epoch": 0.5, - "learning_rate": 2.5195108614604563e-05, - "loss": 3.2853, + "learning_rate": 2.5235404273340263e-05, + "loss": 3.2521, "step": 149700 }, { "epoch": 0.5, - "learning_rate": 2.5178538880880188e-05, - "loss": 3.2897, + "learning_rate": 2.5218861457223587e-05, + "loss": 3.2594, "step": 149800 }, { "epoch": 0.5, - "learning_rate": 2.5161969147155806e-05, - "loss": 3.2922, + "learning_rate": 2.5202318641106915e-05, + "loss": 3.2526, "step": 149900 }, { "epoch": 0.5, - "learning_rate": 2.5145399413431424e-05, - "loss": 3.2803, + "learning_rate": 2.5185775824990243e-05, + "loss": 3.2341, "step": 150000 }, { "epoch": 0.5, - "learning_rate": 2.512882967970705e-05, - "loss": 3.2895, + "learning_rate": 2.5169233008873568e-05, + "loss": 3.2307, "step": 150100 }, { "epoch": 0.5, - "learning_rate": 2.511225994598267e-05, - "loss": 3.2965, + "learning_rate": 2.5152690192756896e-05, + "loss": 3.2473, "step": 150200 }, { "epoch": 0.5, - "learning_rate": 2.5095690212258288e-05, - "loss": 3.2636, + "learning_rate": 2.5136147376640224e-05, + "loss": 3.2472, "step": 150300 }, { "epoch": 0.5, - "learning_rate": 2.5079120478533913e-05, - "loss": 3.2881, + "learning_rate": 2.5119604560523545e-05, + "loss": 3.249, "step": 150400 }, { "epoch": 0.5, - "learning_rate": 2.506255074480953e-05, - "loss": 3.2803, + "learning_rate": 2.5103061744406876e-05, + "loss": 3.2448, "step": 150500 }, { "epoch": 0.5, - "learning_rate": 2.5045981011085152e-05, - "loss": 3.2805, + "learning_rate": 2.5086518928290204e-05, + "loss": 3.2359, "step": 150600 }, { "epoch": 0.5, - "learning_rate": 2.5029411277360777e-05, - "loss": 3.2854, + "learning_rate": 2.5069976112173525e-05, + "loss": 3.2416, "step": 150700 }, { "epoch": 0.5, - "learning_rate": 2.5012841543636395e-05, - "loss": 3.2914, + "learning_rate": 2.5053433296056856e-05, + "loss": 3.2408, "step": 150800 }, { "epoch": 0.5, - "learning_rate": 2.4996271809912017e-05, - "loss": 3.2876, + "learning_rate": 2.5036890479940184e-05, + "loss": 3.2398, "step": 150900 }, { "epoch": 0.5, - "learning_rate": 2.4979702076187635e-05, - "loss": 3.2905, + "learning_rate": 2.5020347663823506e-05, + "loss": 3.2477, "step": 151000 }, { "epoch": 0.5, - "learning_rate": 2.4963132342463256e-05, - "loss": 3.2992, + "learning_rate": 2.5003804847706837e-05, + "loss": 3.2486, "step": 151100 }, { "epoch": 0.5, - "learning_rate": 2.494656260873888e-05, - "loss": 3.2884, + "learning_rate": 2.4987262031590165e-05, + "loss": 3.2229, "step": 151200 }, { "epoch": 0.5, - "learning_rate": 2.49299928750145e-05, - "loss": 3.2947, + "learning_rate": 2.497071921547349e-05, + "loss": 3.2583, "step": 151300 }, { "epoch": 0.5, - "learning_rate": 2.491342314129012e-05, - "loss": 3.3012, + "learning_rate": 2.4954176399356817e-05, + "loss": 3.2412, "step": 151400 }, { "epoch": 0.5, - "learning_rate": 2.4896853407565742e-05, - "loss": 3.2898, + "learning_rate": 2.4937633583240145e-05, + "loss": 3.2396, "step": 151500 }, { "epoch": 0.5, - "learning_rate": 2.488028367384136e-05, - "loss": 3.2881, + "learning_rate": 2.492109076712347e-05, + "loss": 3.24, "step": 151600 }, { "epoch": 0.5, - "learning_rate": 2.4863713940116985e-05, - "loss": 3.2951, + "learning_rate": 2.4904547951006794e-05, + "loss": 3.2408, "step": 151700 }, { "epoch": 0.5, - "learning_rate": 2.4847144206392607e-05, - "loss": 3.2936, + "learning_rate": 2.4888005134890126e-05, + "loss": 3.2473, "step": 151800 }, { "epoch": 0.5, - "learning_rate": 2.4830574472668225e-05, - "loss": 3.2936, + "learning_rate": 2.487146231877345e-05, + "loss": 3.2522, "step": 151900 }, { "epoch": 0.5, - "learning_rate": 2.4814004738943846e-05, - "loss": 3.2883, + "learning_rate": 2.4854919502656778e-05, + "loss": 3.2365, "step": 152000 }, { "epoch": 0.5, - "learning_rate": 2.4797435005219464e-05, - "loss": 3.2834, + "learning_rate": 2.4838376686540106e-05, + "loss": 3.2455, "step": 152100 }, { "epoch": 0.5, - "learning_rate": 2.478086527149509e-05, - "loss": 3.2974, + "learning_rate": 2.482183387042343e-05, + "loss": 3.2273, "step": 152200 }, { "epoch": 0.5, - "learning_rate": 2.476429553777071e-05, - "loss": 3.2838, + "learning_rate": 2.480529105430676e-05, + "loss": 3.249, "step": 152300 }, { - "epoch": 0.51, - "learning_rate": 2.474772580404633e-05, - "loss": 3.2967, + "epoch": 0.5, + "learning_rate": 2.4788748238190086e-05, + "loss": 3.2422, "step": 152400 }, { - "epoch": 0.51, - "learning_rate": 2.473115607032195e-05, - "loss": 3.2877, + "epoch": 0.5, + "learning_rate": 2.477220542207341e-05, + "loss": 3.2458, "step": 152500 }, { - "epoch": 0.51, - "learning_rate": 2.471458633659757e-05, - "loss": 3.2875, + "epoch": 0.5, + "learning_rate": 2.475566260595674e-05, + "loss": 3.2415, "step": 152600 }, { "epoch": 0.51, - "learning_rate": 2.4698016602873193e-05, - "loss": 3.2819, + "learning_rate": 2.4739119789840067e-05, + "loss": 3.236, "step": 152700 }, { "epoch": 0.51, - "learning_rate": 2.4681446869148815e-05, - "loss": 3.2896, + "learning_rate": 2.472257697372339e-05, + "loss": 3.2522, "step": 152800 }, { "epoch": 0.51, - "learning_rate": 2.4664877135424436e-05, - "loss": 3.2978, + "learning_rate": 2.470603415760672e-05, + "loss": 3.2642, "step": 152900 }, { "epoch": 0.51, - "learning_rate": 2.4648307401700054e-05, - "loss": 3.2919, + "learning_rate": 2.4689491341490044e-05, + "loss": 3.2393, "step": 153000 }, { "epoch": 0.51, - "learning_rate": 2.4631737667975676e-05, - "loss": 3.2998, + "learning_rate": 2.4672948525373375e-05, + "loss": 3.2341, "step": 153100 }, { "epoch": 0.51, - "learning_rate": 2.4615167934251297e-05, - "loss": 3.2886, + "learning_rate": 2.46564057092567e-05, + "loss": 3.2309, "step": 153200 }, { "epoch": 0.51, - "learning_rate": 2.459859820052692e-05, - "loss": 3.2825, + "learning_rate": 2.4639862893140024e-05, + "loss": 3.2424, "step": 153300 }, { "epoch": 0.51, - "learning_rate": 2.458202846680254e-05, - "loss": 3.2859, + "learning_rate": 2.4623320077023355e-05, + "loss": 3.2388, "step": 153400 }, { "epoch": 0.51, - "learning_rate": 2.456545873307816e-05, - "loss": 3.2858, + "learning_rate": 2.460677726090668e-05, + "loss": 3.232, "step": 153500 }, { "epoch": 0.51, - "learning_rate": 2.454888899935378e-05, - "loss": 3.2882, + "learning_rate": 2.4590234444790004e-05, + "loss": 3.2512, "step": 153600 }, { "epoch": 0.51, - "learning_rate": 2.45323192656294e-05, - "loss": 3.2905, + "learning_rate": 2.4573691628673336e-05, + "loss": 3.2486, "step": 153700 }, { "epoch": 0.51, - "learning_rate": 2.4515749531905026e-05, - "loss": 3.2871, + "learning_rate": 2.455714881255666e-05, + "loss": 3.2348, "step": 153800 }, { "epoch": 0.51, - "learning_rate": 2.4499179798180644e-05, - "loss": 3.2845, + "learning_rate": 2.4540605996439985e-05, + "loss": 3.2378, "step": 153900 }, { "epoch": 0.51, - "learning_rate": 2.4482610064456265e-05, - "loss": 3.2841, + "learning_rate": 2.4524063180323316e-05, + "loss": 3.2338, "step": 154000 }, { "epoch": 0.51, - "learning_rate": 2.4466040330731887e-05, - "loss": 3.2847, + "learning_rate": 2.450752036420664e-05, + "loss": 3.2389, "step": 154100 }, { "epoch": 0.51, - "learning_rate": 2.4449470597007505e-05, - "loss": 3.2922, + "learning_rate": 2.449097754808997e-05, + "loss": 3.2448, "step": 154200 }, { "epoch": 0.51, - "learning_rate": 2.443290086328313e-05, - "loss": 3.3039, + "learning_rate": 2.4474434731973293e-05, + "loss": 3.2481, "step": 154300 }, { "epoch": 0.51, - "learning_rate": 2.441633112955875e-05, - "loss": 3.2935, + "learning_rate": 2.445789191585662e-05, + "loss": 3.2331, "step": 154400 }, { "epoch": 0.51, - "learning_rate": 2.439976139583437e-05, - "loss": 3.2822, + "learning_rate": 2.444134909973995e-05, + "loss": 3.244, "step": 154500 }, { "epoch": 0.51, - "learning_rate": 2.438319166210999e-05, - "loss": 3.301, + "learning_rate": 2.4424806283623273e-05, + "loss": 3.2427, "step": 154600 }, { "epoch": 0.51, - "learning_rate": 2.4366621928385612e-05, - "loss": 3.2923, + "learning_rate": 2.44082634675066e-05, + "loss": 3.2379, "step": 154700 }, { "epoch": 0.51, - "learning_rate": 2.4350052194661234e-05, - "loss": 3.2712, + "learning_rate": 2.439172065138993e-05, + "loss": 3.2518, "step": 154800 }, { "epoch": 0.51, - "learning_rate": 2.4333482460936855e-05, - "loss": 3.2886, + "learning_rate": 2.4375177835273254e-05, + "loss": 3.2453, "step": 154900 }, { "epoch": 0.51, - "learning_rate": 2.4316912727212473e-05, - "loss": 3.2771, + "learning_rate": 2.4358635019156582e-05, + "loss": 3.257, "step": 155000 }, { "epoch": 0.51, - "learning_rate": 2.4300342993488095e-05, - "loss": 3.2888, + "learning_rate": 2.434209220303991e-05, + "loss": 3.2442, "step": 155100 }, { "epoch": 0.51, - "learning_rate": 2.4283773259763716e-05, - "loss": 3.2664, + "learning_rate": 2.4325549386923234e-05, + "loss": 3.2417, "step": 155200 }, { "epoch": 0.51, - "learning_rate": 2.4267203526039338e-05, - "loss": 3.2847, + "learning_rate": 2.4309006570806566e-05, + "loss": 3.2312, "step": 155300 }, { "epoch": 0.51, - "learning_rate": 2.425063379231496e-05, - "loss": 3.2771, + "learning_rate": 2.429246375468989e-05, + "loss": 3.2477, "step": 155400 }, { - "epoch": 0.52, - "learning_rate": 2.423406405859058e-05, - "loss": 3.2919, + "epoch": 0.51, + "learning_rate": 2.4275920938573215e-05, + "loss": 3.2444, "step": 155500 }, { - "epoch": 0.52, - "learning_rate": 2.42174943248662e-05, - "loss": 3.3008, + "epoch": 0.51, + "learning_rate": 2.4259378122456543e-05, + "loss": 3.2474, "step": 155600 }, { "epoch": 0.52, - "learning_rate": 2.420092459114182e-05, - "loss": 3.285, + "learning_rate": 2.424283530633987e-05, + "loss": 3.2493, "step": 155700 }, { "epoch": 0.52, - "learning_rate": 2.418435485741744e-05, - "loss": 3.2908, + "learning_rate": 2.4226292490223195e-05, + "loss": 3.2478, "step": 155800 }, { "epoch": 0.52, - "learning_rate": 2.4167785123693063e-05, - "loss": 3.2977, + "learning_rate": 2.4209749674106523e-05, + "loss": 3.2394, "step": 155900 }, { "epoch": 0.52, - "learning_rate": 2.4151215389968685e-05, - "loss": 3.2769, + "learning_rate": 2.419320685798985e-05, + "loss": 3.2324, "step": 156000 }, { "epoch": 0.52, - "learning_rate": 2.4134645656244306e-05, - "loss": 3.2821, + "learning_rate": 2.4176664041873175e-05, + "loss": 3.246, "step": 156100 }, { "epoch": 0.52, - "learning_rate": 2.4118075922519924e-05, - "loss": 3.2894, + "learning_rate": 2.4160121225756503e-05, + "loss": 3.2348, "step": 156200 }, { "epoch": 0.52, - "learning_rate": 2.410150618879555e-05, - "loss": 3.2814, + "learning_rate": 2.414357840963983e-05, + "loss": 3.2305, "step": 156300 }, { "epoch": 0.52, - "learning_rate": 2.408493645507117e-05, - "loss": 3.2821, + "learning_rate": 2.412703559352316e-05, + "loss": 3.2474, "step": 156400 }, { "epoch": 0.52, - "learning_rate": 2.406836672134679e-05, - "loss": 3.2914, + "learning_rate": 2.4110492777406484e-05, + "loss": 3.2417, "step": 156500 }, { "epoch": 0.52, - "learning_rate": 2.405179698762241e-05, - "loss": 3.3064, + "learning_rate": 2.409394996128981e-05, + "loss": 3.2415, "step": 156600 }, { "epoch": 0.52, - "learning_rate": 2.403522725389803e-05, - "loss": 3.2822, + "learning_rate": 2.407740714517314e-05, + "loss": 3.2402, "step": 156700 }, { "epoch": 0.52, - "learning_rate": 2.4018657520173653e-05, - "loss": 3.2836, + "learning_rate": 2.4060864329056464e-05, + "loss": 3.2335, "step": 156800 }, { "epoch": 0.52, - "learning_rate": 2.4002087786449274e-05, - "loss": 3.2855, + "learning_rate": 2.4044321512939792e-05, + "loss": 3.237, "step": 156900 }, { "epoch": 0.52, - "learning_rate": 2.3985518052724896e-05, - "loss": 3.2925, + "learning_rate": 2.402777869682312e-05, + "loss": 3.2325, "step": 157000 }, { "epoch": 0.52, - "learning_rate": 2.3968948319000514e-05, - "loss": 3.2881, + "learning_rate": 2.4011235880706444e-05, + "loss": 3.2442, "step": 157100 }, { "epoch": 0.52, - "learning_rate": 2.3952378585276135e-05, - "loss": 3.2888, + "learning_rate": 2.3994693064589772e-05, + "loss": 3.2521, "step": 157200 }, { "epoch": 0.52, - "learning_rate": 2.3935808851551757e-05, - "loss": 3.2933, + "learning_rate": 2.39781502484731e-05, + "loss": 3.2446, "step": 157300 }, { "epoch": 0.52, - "learning_rate": 2.391923911782738e-05, - "loss": 3.2753, + "learning_rate": 2.3961607432356425e-05, + "loss": 3.2415, "step": 157400 }, { "epoch": 0.52, - "learning_rate": 2.3902669384103e-05, - "loss": 3.2907, + "learning_rate": 2.3945064616239753e-05, + "loss": 3.2497, "step": 157500 }, { "epoch": 0.52, - "learning_rate": 2.3886099650378618e-05, - "loss": 3.2759, + "learning_rate": 2.392852180012308e-05, + "loss": 3.2431, "step": 157600 }, { "epoch": 0.52, - "learning_rate": 2.386952991665424e-05, - "loss": 3.2871, + "learning_rate": 2.3911978984006405e-05, + "loss": 3.2538, "step": 157700 }, { "epoch": 0.52, - "learning_rate": 2.385296018292986e-05, - "loss": 3.2896, + "learning_rate": 2.3895436167889733e-05, + "loss": 3.2418, "step": 157800 }, { "epoch": 0.52, - "learning_rate": 2.3836390449205482e-05, - "loss": 3.2737, + "learning_rate": 2.387889335177306e-05, + "loss": 3.2394, "step": 157900 }, { "epoch": 0.52, - "learning_rate": 2.3819820715481104e-05, - "loss": 3.2869, + "learning_rate": 2.3862350535656386e-05, + "loss": 3.2354, "step": 158000 }, { "epoch": 0.52, - "learning_rate": 2.3803250981756725e-05, - "loss": 3.2851, + "learning_rate": 2.3845807719539713e-05, + "loss": 3.2503, "step": 158100 }, { "epoch": 0.52, - "learning_rate": 2.3786681248032343e-05, - "loss": 3.2927, + "learning_rate": 2.382926490342304e-05, + "loss": 3.2458, "step": 158200 }, { "epoch": 0.52, - "learning_rate": 2.3770111514307965e-05, - "loss": 3.2811, + "learning_rate": 2.3812722087306366e-05, + "loss": 3.2393, "step": 158300 }, { "epoch": 0.52, - "learning_rate": 2.375354178058359e-05, - "loss": 3.2817, + "learning_rate": 2.3796179271189694e-05, + "loss": 3.2412, "step": 158400 }, { - "epoch": 0.53, - "learning_rate": 2.3736972046859208e-05, - "loss": 3.2755, + "epoch": 0.52, + "learning_rate": 2.3779636455073022e-05, + "loss": 3.2515, "step": 158500 }, { - "epoch": 0.53, - "learning_rate": 2.372040231313483e-05, - "loss": 3.2903, + "epoch": 0.52, + "learning_rate": 2.376309363895635e-05, + "loss": 3.2411, "step": 158600 }, { "epoch": 0.53, - "learning_rate": 2.370383257941045e-05, - "loss": 3.2895, + "learning_rate": 2.3746550822839674e-05, + "loss": 3.2439, "step": 158700 }, { "epoch": 0.53, - "learning_rate": 2.368726284568607e-05, - "loss": 3.2662, + "learning_rate": 2.3730008006723002e-05, + "loss": 3.2328, "step": 158800 }, { "epoch": 0.53, - "learning_rate": 2.3670693111961694e-05, - "loss": 3.2843, + "learning_rate": 2.371346519060633e-05, + "loss": 3.2493, "step": 158900 }, { "epoch": 0.53, - "learning_rate": 2.3654123378237315e-05, - "loss": 3.285, + "learning_rate": 2.3696922374489655e-05, + "loss": 3.2349, "step": 159000 }, { "epoch": 0.53, - "learning_rate": 2.3637553644512933e-05, - "loss": 3.2847, + "learning_rate": 2.3680379558372983e-05, + "loss": 3.2485, "step": 159100 }, { "epoch": 0.53, - "learning_rate": 2.3620983910788555e-05, - "loss": 3.2952, + "learning_rate": 2.366383674225631e-05, + "loss": 3.2378, "step": 159200 }, { "epoch": 0.53, - "learning_rate": 2.3604414177064176e-05, - "loss": 3.2964, + "learning_rate": 2.3647293926139635e-05, + "loss": 3.2567, "step": 159300 }, { "epoch": 0.53, - "learning_rate": 2.3587844443339798e-05, - "loss": 3.2899, + "learning_rate": 2.3630751110022963e-05, + "loss": 3.2457, "step": 159400 }, { "epoch": 0.53, - "learning_rate": 2.357127470961542e-05, - "loss": 3.2791, + "learning_rate": 2.361420829390629e-05, + "loss": 3.2408, "step": 159500 }, { "epoch": 0.53, - "learning_rate": 2.355470497589104e-05, - "loss": 3.2832, + "learning_rate": 2.3597665477789615e-05, + "loss": 3.2504, "step": 159600 }, { "epoch": 0.53, - "learning_rate": 2.353813524216666e-05, - "loss": 3.2758, + "learning_rate": 2.3581122661672943e-05, + "loss": 3.2181, "step": 159700 }, { "epoch": 0.53, - "learning_rate": 2.352156550844228e-05, - "loss": 3.2839, + "learning_rate": 2.356457984555627e-05, + "loss": 3.2452, "step": 159800 }, { "epoch": 0.53, - "learning_rate": 2.35049957747179e-05, - "loss": 3.2769, + "learning_rate": 2.3548037029439596e-05, + "loss": 3.2495, "step": 159900 }, { "epoch": 0.53, - "learning_rate": 2.3488426040993523e-05, - "loss": 3.2795, + "learning_rate": 2.3531494213322924e-05, + "loss": 3.2336, "step": 160000 }, { "epoch": 0.53, - "learning_rate": 2.3471856307269144e-05, - "loss": 3.2731, + "learning_rate": 2.351495139720625e-05, + "loss": 3.2378, "step": 160100 }, { "epoch": 0.53, - "learning_rate": 2.3455286573544762e-05, - "loss": 3.2861, + "learning_rate": 2.3498408581089576e-05, + "loss": 3.2335, "step": 160200 }, { "epoch": 0.53, - "learning_rate": 2.3438716839820384e-05, - "loss": 3.2805, + "learning_rate": 2.3481865764972904e-05, + "loss": 3.2521, "step": 160300 }, { "epoch": 0.53, - "learning_rate": 2.3422147106096005e-05, - "loss": 3.2756, + "learning_rate": 2.3465322948856232e-05, + "loss": 3.2371, "step": 160400 }, { "epoch": 0.53, - "learning_rate": 2.3405577372371627e-05, - "loss": 3.2833, + "learning_rate": 2.344878013273956e-05, + "loss": 3.243, "step": 160500 }, { "epoch": 0.53, - "learning_rate": 2.338900763864725e-05, - "loss": 3.2805, + "learning_rate": 2.3432237316622884e-05, + "loss": 3.2391, "step": 160600 }, { "epoch": 0.53, - "learning_rate": 2.337243790492287e-05, - "loss": 3.2825, + "learning_rate": 2.341569450050621e-05, + "loss": 3.2362, "step": 160700 }, { "epoch": 0.53, - "learning_rate": 2.3355868171198488e-05, - "loss": 3.2848, + "learning_rate": 2.339915168438954e-05, + "loss": 3.2409, "step": 160800 }, { "epoch": 0.53, - "learning_rate": 2.333929843747411e-05, - "loss": 3.2831, + "learning_rate": 2.3382608868272865e-05, + "loss": 3.2338, "step": 160900 }, { "epoch": 0.53, - "learning_rate": 2.3322728703749734e-05, - "loss": 3.2864, + "learning_rate": 2.336606605215619e-05, + "loss": 3.2346, "step": 161000 }, { "epoch": 0.53, - "learning_rate": 2.3306158970025352e-05, - "loss": 3.2941, + "learning_rate": 2.334952323603952e-05, + "loss": 3.2359, "step": 161100 }, { "epoch": 0.53, - "learning_rate": 2.3289589236300974e-05, - "loss": 3.2851, + "learning_rate": 2.3332980419922845e-05, + "loss": 3.2422, "step": 161200 }, { "epoch": 0.53, - "learning_rate": 2.3273019502576595e-05, - "loss": 3.2845, + "learning_rate": 2.331643760380617e-05, + "loss": 3.2396, "step": 161300 }, { "epoch": 0.53, - "learning_rate": 2.3256449768852213e-05, - "loss": 3.2823, + "learning_rate": 2.32998947876895e-05, + "loss": 3.2392, "step": 161400 }, { - "epoch": 0.54, - "learning_rate": 2.3239880035127838e-05, - "loss": 3.2893, + "epoch": 0.53, + "learning_rate": 2.3283351971572826e-05, + "loss": 3.2409, "step": 161500 }, { - "epoch": 0.54, - "learning_rate": 2.322331030140346e-05, - "loss": 3.277, + "epoch": 0.53, + "learning_rate": 2.3266809155456154e-05, + "loss": 3.2508, "step": 161600 }, { - "epoch": 0.54, - "learning_rate": 2.3206740567679078e-05, - "loss": 3.279, + "epoch": 0.53, + "learning_rate": 2.325026633933948e-05, + "loss": 3.2347, "step": 161700 }, { "epoch": 0.54, - "learning_rate": 2.31901708339547e-05, - "loss": 3.2878, + "learning_rate": 2.3233723523222806e-05, + "loss": 3.2474, "step": 161800 }, { "epoch": 0.54, - "learning_rate": 2.317360110023032e-05, - "loss": 3.288, + "learning_rate": 2.3217180707106134e-05, + "loss": 3.2473, "step": 161900 }, { "epoch": 0.54, - "learning_rate": 2.3157031366505942e-05, - "loss": 3.2729, + "learning_rate": 2.320063789098946e-05, + "loss": 3.24, "step": 162000 }, { "epoch": 0.54, - "learning_rate": 2.3140461632781564e-05, - "loss": 3.2861, + "learning_rate": 2.3184095074872786e-05, + "loss": 3.2291, "step": 162100 }, { "epoch": 0.54, - "learning_rate": 2.3123891899057185e-05, - "loss": 3.2831, + "learning_rate": 2.3167552258756114e-05, + "loss": 3.2368, "step": 162200 }, { "epoch": 0.54, - "learning_rate": 2.3107322165332803e-05, - "loss": 3.2861, + "learning_rate": 2.315100944263944e-05, + "loss": 3.246, "step": 162300 }, { "epoch": 0.54, - "learning_rate": 2.3090752431608425e-05, - "loss": 3.2863, + "learning_rate": 2.3134466626522767e-05, + "loss": 3.2411, "step": 162400 }, { "epoch": 0.54, - "learning_rate": 2.3074182697884046e-05, - "loss": 3.2806, + "learning_rate": 2.3117923810406095e-05, + "loss": 3.2544, "step": 162500 }, { "epoch": 0.54, - "learning_rate": 2.3057612964159668e-05, - "loss": 3.2929, + "learning_rate": 2.310138099428942e-05, + "loss": 3.2261, "step": 162600 }, { "epoch": 0.54, - "learning_rate": 2.304104323043529e-05, - "loss": 3.2824, + "learning_rate": 2.308483817817275e-05, + "loss": 3.2463, "step": 162700 }, { "epoch": 0.54, - "learning_rate": 2.3024473496710907e-05, - "loss": 3.2857, + "learning_rate": 2.3068295362056075e-05, + "loss": 3.2444, "step": 162800 }, { "epoch": 0.54, - "learning_rate": 2.300790376298653e-05, - "loss": 3.2773, + "learning_rate": 2.30517525459394e-05, + "loss": 3.2472, "step": 162900 }, { "epoch": 0.54, - "learning_rate": 2.299133402926215e-05, - "loss": 3.2766, + "learning_rate": 2.303520972982273e-05, + "loss": 3.2331, "step": 163000 }, { "epoch": 0.54, - "learning_rate": 2.297476429553777e-05, - "loss": 3.2782, + "learning_rate": 2.3018666913706055e-05, + "loss": 3.2528, "step": 163100 }, { "epoch": 0.54, - "learning_rate": 2.2958194561813393e-05, - "loss": 3.2789, + "learning_rate": 2.300212409758938e-05, + "loss": 3.2472, "step": 163200 }, { "epoch": 0.54, - "learning_rate": 2.2941624828089014e-05, - "loss": 3.2865, + "learning_rate": 2.2985581281472708e-05, + "loss": 3.2289, "step": 163300 }, { "epoch": 0.54, - "learning_rate": 2.2925055094364632e-05, - "loss": 3.2794, + "learning_rate": 2.2969038465356036e-05, + "loss": 3.2396, "step": 163400 }, { "epoch": 0.54, - "learning_rate": 2.2908485360640254e-05, - "loss": 3.2685, + "learning_rate": 2.295249564923936e-05, + "loss": 3.236, "step": 163500 }, { "epoch": 0.54, - "learning_rate": 2.289191562691588e-05, - "loss": 3.2993, + "learning_rate": 2.2935952833122688e-05, + "loss": 3.2404, "step": 163600 }, { "epoch": 0.54, - "learning_rate": 2.2875345893191497e-05, - "loss": 3.2837, + "learning_rate": 2.2919410017006016e-05, + "loss": 3.2232, "step": 163700 }, { "epoch": 0.54, - "learning_rate": 2.285877615946712e-05, - "loss": 3.2783, + "learning_rate": 2.2902867200889344e-05, + "loss": 3.2322, "step": 163800 }, { "epoch": 0.54, - "learning_rate": 2.284220642574274e-05, - "loss": 3.2909, + "learning_rate": 2.288632438477267e-05, + "loss": 3.2399, "step": 163900 }, { "epoch": 0.54, - "learning_rate": 2.2825636692018358e-05, - "loss": 3.2844, + "learning_rate": 2.2869781568655997e-05, + "loss": 3.2422, "step": 164000 }, { "epoch": 0.54, - "learning_rate": 2.2809066958293983e-05, - "loss": 3.2822, + "learning_rate": 2.2853238752539324e-05, + "loss": 3.2393, "step": 164100 }, { "epoch": 0.54, - "learning_rate": 2.2792497224569604e-05, - "loss": 3.2899, + "learning_rate": 2.283669593642265e-05, + "loss": 3.2434, "step": 164200 }, { "epoch": 0.54, - "learning_rate": 2.2775927490845222e-05, - "loss": 3.282, + "learning_rate": 2.2820153120305977e-05, + "loss": 3.2474, "step": 164300 }, { "epoch": 0.54, - "learning_rate": 2.2759357757120844e-05, - "loss": 3.2589, + "learning_rate": 2.2803610304189305e-05, + "loss": 3.2384, "step": 164400 }, { - "epoch": 0.55, - "learning_rate": 2.2742788023396465e-05, - "loss": 3.2706, + "epoch": 0.54, + "learning_rate": 2.278706748807263e-05, + "loss": 3.2381, "step": 164500 }, { - "epoch": 0.55, - "learning_rate": 2.2726218289672087e-05, - "loss": 3.2807, + "epoch": 0.54, + "learning_rate": 2.2770524671955957e-05, + "loss": 3.2439, "step": 164600 }, { - "epoch": 0.55, - "learning_rate": 2.2709648555947708e-05, - "loss": 3.2747, + "epoch": 0.54, + "learning_rate": 2.2753981855839285e-05, + "loss": 3.2491, "step": 164700 }, { "epoch": 0.55, - "learning_rate": 2.269307882222333e-05, - "loss": 3.2863, + "learning_rate": 2.273743903972261e-05, + "loss": 3.24, "step": 164800 }, { "epoch": 0.55, - "learning_rate": 2.2676509088498948e-05, - "loss": 3.287, + "learning_rate": 2.2720896223605938e-05, + "loss": 3.2394, "step": 164900 }, { "epoch": 0.55, - "learning_rate": 2.265993935477457e-05, - "loss": 3.2848, + "learning_rate": 2.2704353407489266e-05, + "loss": 3.2299, "step": 165000 }, { "epoch": 0.55, - "learning_rate": 2.264336962105019e-05, - "loss": 3.2724, + "learning_rate": 2.268781059137259e-05, + "loss": 3.2377, "step": 165100 }, { "epoch": 0.55, - "learning_rate": 2.2626799887325812e-05, - "loss": 3.2838, + "learning_rate": 2.2671267775255918e-05, + "loss": 3.2351, "step": 165200 }, { "epoch": 0.55, - "learning_rate": 2.2610230153601434e-05, - "loss": 3.2744, + "learning_rate": 2.2654724959139246e-05, + "loss": 3.2439, "step": 165300 }, { "epoch": 0.55, - "learning_rate": 2.259366041987705e-05, - "loss": 3.2855, + "learning_rate": 2.263818214302257e-05, + "loss": 3.2256, "step": 165400 }, { "epoch": 0.55, - "learning_rate": 2.2577090686152673e-05, - "loss": 3.2843, + "learning_rate": 2.26216393269059e-05, + "loss": 3.2266, "step": 165500 }, { "epoch": 0.55, - "learning_rate": 2.2560520952428295e-05, - "loss": 3.2717, + "learning_rate": 2.2605096510789226e-05, + "loss": 3.2459, "step": 165600 }, { "epoch": 0.55, - "learning_rate": 2.2543951218703916e-05, - "loss": 3.2861, + "learning_rate": 2.258855369467255e-05, + "loss": 3.2388, "step": 165700 }, { "epoch": 0.55, - "learning_rate": 2.2527381484979538e-05, - "loss": 3.2767, + "learning_rate": 2.257201087855588e-05, + "loss": 3.2449, "step": 165800 }, { "epoch": 0.55, - "learning_rate": 2.251081175125516e-05, - "loss": 3.2805, + "learning_rate": 2.2555468062439207e-05, + "loss": 3.2476, "step": 165900 }, { "epoch": 0.55, - "learning_rate": 2.2494242017530777e-05, - "loss": 3.2862, + "learning_rate": 2.2538925246322535e-05, + "loss": 3.2432, "step": 166000 }, { "epoch": 0.55, - "learning_rate": 2.24776722838064e-05, - "loss": 3.2846, + "learning_rate": 2.252238243020586e-05, + "loss": 3.235, "step": 166100 }, { "epoch": 0.55, - "learning_rate": 2.2461102550082023e-05, - "loss": 3.2897, + "learning_rate": 2.2505839614089187e-05, + "loss": 3.2409, "step": 166200 }, { "epoch": 0.55, - "learning_rate": 2.244453281635764e-05, - "loss": 3.2737, + "learning_rate": 2.2489296797972515e-05, + "loss": 3.2407, "step": 166300 }, { "epoch": 0.55, - "learning_rate": 2.2427963082633263e-05, - "loss": 3.2872, + "learning_rate": 2.247275398185584e-05, + "loss": 3.2224, "step": 166400 }, { "epoch": 0.55, - "learning_rate": 2.2411393348908884e-05, - "loss": 3.2867, + "learning_rate": 2.2456211165739168e-05, + "loss": 3.2423, "step": 166500 }, { "epoch": 0.55, - "learning_rate": 2.2394823615184506e-05, - "loss": 3.294, + "learning_rate": 2.2439668349622495e-05, + "loss": 3.2434, "step": 166600 }, { "epoch": 0.55, - "learning_rate": 2.2378253881460127e-05, - "loss": 3.2776, + "learning_rate": 2.242312553350582e-05, + "loss": 3.2469, "step": 166700 }, { "epoch": 0.55, - "learning_rate": 2.236168414773575e-05, - "loss": 3.2909, + "learning_rate": 2.2406582717389148e-05, + "loss": 3.2291, "step": 166800 }, { "epoch": 0.55, - "learning_rate": 2.2345114414011367e-05, - "loss": 3.2777, + "learning_rate": 2.2390039901272476e-05, + "loss": 3.2298, "step": 166900 }, { "epoch": 0.55, - "learning_rate": 2.232854468028699e-05, - "loss": 3.2889, + "learning_rate": 2.23734970851558e-05, + "loss": 3.241, "step": 167000 }, { "epoch": 0.55, - "learning_rate": 2.231197494656261e-05, - "loss": 3.2854, + "learning_rate": 2.2356954269039128e-05, + "loss": 3.2396, "step": 167100 }, { "epoch": 0.55, - "learning_rate": 2.229540521283823e-05, - "loss": 3.2827, + "learning_rate": 2.2340411452922456e-05, + "loss": 3.2544, "step": 167200 }, { "epoch": 0.55, - "learning_rate": 2.2278835479113853e-05, - "loss": 3.3006, + "learning_rate": 2.232386863680578e-05, + "loss": 3.2422, "step": 167300 }, { "epoch": 0.55, - "learning_rate": 2.2262265745389474e-05, - "loss": 3.2783, + "learning_rate": 2.230732582068911e-05, + "loss": 3.2417, "step": 167400 }, { - "epoch": 0.56, - "learning_rate": 2.2245696011665092e-05, - "loss": 3.2888, + "epoch": 0.55, + "learning_rate": 2.2290783004572437e-05, + "loss": 3.2348, "step": 167500 }, { - "epoch": 0.56, - "learning_rate": 2.2229126277940714e-05, - "loss": 3.272, + "epoch": 0.55, + "learning_rate": 2.227424018845576e-05, + "loss": 3.2486, "step": 167600 }, { - "epoch": 0.56, - "learning_rate": 2.2212556544216335e-05, - "loss": 3.2821, + "epoch": 0.55, + "learning_rate": 2.225769737233909e-05, + "loss": 3.2375, "step": 167700 }, { "epoch": 0.56, - "learning_rate": 2.2195986810491957e-05, - "loss": 3.2795, + "learning_rate": 2.2241154556222417e-05, + "loss": 3.2442, "step": 167800 }, { "epoch": 0.56, - "learning_rate": 2.2179417076767578e-05, - "loss": 3.2776, + "learning_rate": 2.2224611740105745e-05, + "loss": 3.2392, "step": 167900 }, { "epoch": 0.56, - "learning_rate": 2.2162847343043196e-05, - "loss": 3.2793, + "learning_rate": 2.220806892398907e-05, + "loss": 3.2413, "step": 168000 }, { "epoch": 0.56, - "learning_rate": 2.2146277609318818e-05, - "loss": 3.2818, + "learning_rate": 2.2191526107872397e-05, + "loss": 3.2351, "step": 168100 }, { "epoch": 0.56, - "learning_rate": 2.2129707875594443e-05, - "loss": 3.2793, + "learning_rate": 2.2174983291755725e-05, + "loss": 3.2392, "step": 168200 }, { "epoch": 0.56, - "learning_rate": 2.211313814187006e-05, - "loss": 3.2892, + "learning_rate": 2.215844047563905e-05, + "loss": 3.251, "step": 168300 }, { "epoch": 0.56, - "learning_rate": 2.2096568408145682e-05, - "loss": 3.2896, + "learning_rate": 2.2141897659522374e-05, + "loss": 3.2369, "step": 168400 }, { "epoch": 0.56, - "learning_rate": 2.2079998674421304e-05, - "loss": 3.2771, + "learning_rate": 2.2125354843405706e-05, + "loss": 3.2283, "step": 168500 }, { "epoch": 0.56, - "learning_rate": 2.206342894069692e-05, - "loss": 3.275, + "learning_rate": 2.210881202728903e-05, + "loss": 3.2428, "step": 168600 }, { "epoch": 0.56, - "learning_rate": 2.2046859206972547e-05, - "loss": 3.2791, + "learning_rate": 2.2092269211172355e-05, + "loss": 3.2285, "step": 168700 }, { "epoch": 0.56, - "learning_rate": 2.2030289473248168e-05, - "loss": 3.2656, + "learning_rate": 2.2075726395055686e-05, + "loss": 3.2345, "step": 168800 }, { "epoch": 0.56, - "learning_rate": 2.2013719739523786e-05, - "loss": 3.2706, + "learning_rate": 2.205918357893901e-05, + "loss": 3.2399, "step": 168900 }, { "epoch": 0.56, - "learning_rate": 2.1997150005799408e-05, - "loss": 3.2837, + "learning_rate": 2.204264076282234e-05, + "loss": 3.2292, "step": 169000 }, { "epoch": 0.56, - "learning_rate": 2.198058027207503e-05, - "loss": 3.2636, + "learning_rate": 2.2026097946705666e-05, + "loss": 3.2366, "step": 169100 }, { "epoch": 0.56, - "learning_rate": 2.196401053835065e-05, - "loss": 3.2828, + "learning_rate": 2.200955513058899e-05, + "loss": 3.2502, "step": 169200 }, { "epoch": 0.56, - "learning_rate": 2.1947440804626272e-05, - "loss": 3.2835, + "learning_rate": 2.199301231447232e-05, + "loss": 3.2386, "step": 169300 }, { "epoch": 0.56, - "learning_rate": 2.1930871070901893e-05, - "loss": 3.283, + "learning_rate": 2.1976469498355647e-05, + "loss": 3.2445, "step": 169400 }, { "epoch": 0.56, - "learning_rate": 2.191430133717751e-05, - "loss": 3.2923, + "learning_rate": 2.195992668223897e-05, + "loss": 3.2265, "step": 169500 }, { "epoch": 0.56, - "learning_rate": 2.1897731603453133e-05, - "loss": 3.2938, + "learning_rate": 2.19433838661223e-05, + "loss": 3.2321, "step": 169600 }, { "epoch": 0.56, - "learning_rate": 2.1881161869728754e-05, - "loss": 3.29, + "learning_rate": 2.1926841050005624e-05, + "loss": 3.2246, "step": 169700 }, { "epoch": 0.56, - "learning_rate": 2.1864592136004376e-05, - "loss": 3.2878, + "learning_rate": 2.1910298233888952e-05, + "loss": 3.2327, "step": 169800 }, { "epoch": 0.56, - "learning_rate": 2.1848022402279997e-05, - "loss": 3.2827, + "learning_rate": 2.189375541777228e-05, + "loss": 3.2388, "step": 169900 }, { "epoch": 0.56, - "learning_rate": 2.183145266855562e-05, - "loss": 3.2757, + "learning_rate": 2.1877212601655604e-05, + "loss": 3.2386, "step": 170000 }, { "epoch": 0.56, - "learning_rate": 2.1814882934831237e-05, - "loss": 3.287, + "learning_rate": 2.1860669785538935e-05, + "loss": 3.242, "step": 170100 }, { "epoch": 0.56, - "learning_rate": 2.179831320110686e-05, - "loss": 3.2788, + "learning_rate": 2.184412696942226e-05, + "loss": 3.2284, "step": 170200 }, { "epoch": 0.56, - "learning_rate": 2.178174346738248e-05, - "loss": 3.2798, + "learning_rate": 2.1827584153305585e-05, + "loss": 3.2349, "step": 170300 }, { "epoch": 0.56, - "learning_rate": 2.17651737336581e-05, - "loss": 3.2871, + "learning_rate": 2.1811041337188916e-05, + "loss": 3.2226, "step": 170400 }, { - "epoch": 0.57, - "learning_rate": 2.1748603999933723e-05, - "loss": 3.286, + "epoch": 0.56, + "learning_rate": 2.179449852107224e-05, + "loss": 3.2311, "step": 170500 }, { - "epoch": 0.57, - "learning_rate": 2.173203426620934e-05, - "loss": 3.3001, + "epoch": 0.56, + "learning_rate": 2.1777955704955565e-05, + "loss": 3.2425, "step": 170600 }, { - "epoch": 0.57, - "learning_rate": 2.1715464532484962e-05, - "loss": 3.2867, + "epoch": 0.56, + "learning_rate": 2.1761412888838896e-05, + "loss": 3.2293, "step": 170700 }, { "epoch": 0.57, - "learning_rate": 2.1698894798760587e-05, - "loss": 3.29, + "learning_rate": 2.174487007272222e-05, + "loss": 3.2412, "step": 170800 }, { "epoch": 0.57, - "learning_rate": 2.1682325065036205e-05, - "loss": 3.289, + "learning_rate": 2.1728327256605545e-05, + "loss": 3.2343, "step": 170900 }, { "epoch": 0.57, - "learning_rate": 2.1665755331311827e-05, - "loss": 3.2888, + "learning_rate": 2.1711784440488873e-05, + "loss": 3.2274, "step": 171000 }, { "epoch": 0.57, - "learning_rate": 2.1649185597587448e-05, - "loss": 3.2808, + "learning_rate": 2.16952416243722e-05, + "loss": 3.2412, "step": 171100 }, { "epoch": 0.57, - "learning_rate": 2.1632615863863066e-05, - "loss": 3.2751, + "learning_rate": 2.167869880825553e-05, + "loss": 3.2346, "step": 171200 }, { "epoch": 0.57, - "learning_rate": 2.161604613013869e-05, - "loss": 3.2808, + "learning_rate": 2.1662155992138854e-05, + "loss": 3.232, "step": 171300 }, { "epoch": 0.57, - "learning_rate": 2.1599476396414313e-05, - "loss": 3.2784, + "learning_rate": 2.164561317602218e-05, + "loss": 3.2322, "step": 171400 }, { "epoch": 0.57, - "learning_rate": 2.158290666268993e-05, - "loss": 3.2842, + "learning_rate": 2.162907035990551e-05, + "loss": 3.2558, "step": 171500 }, { "epoch": 0.57, - "learning_rate": 2.1566336928965552e-05, - "loss": 3.2589, + "learning_rate": 2.1612527543788834e-05, + "loss": 3.2349, "step": 171600 }, { "epoch": 0.57, - "learning_rate": 2.1549767195241174e-05, - "loss": 3.2719, + "learning_rate": 2.1595984727672162e-05, + "loss": 3.2243, "step": 171700 }, { "epoch": 0.57, - "learning_rate": 2.1533197461516795e-05, - "loss": 3.2891, + "learning_rate": 2.157944191155549e-05, + "loss": 3.2361, "step": 171800 }, { "epoch": 0.57, - "learning_rate": 2.1516627727792417e-05, - "loss": 3.2631, + "learning_rate": 2.1562899095438814e-05, + "loss": 3.2343, "step": 171900 }, { "epoch": 0.57, - "learning_rate": 2.1500057994068038e-05, - "loss": 3.2826, + "learning_rate": 2.1546356279322142e-05, + "loss": 3.2517, "step": 172000 }, { "epoch": 0.57, - "learning_rate": 2.1483488260343656e-05, - "loss": 3.2756, + "learning_rate": 2.152981346320547e-05, + "loss": 3.2364, "step": 172100 }, { "epoch": 0.57, - "learning_rate": 2.1466918526619278e-05, - "loss": 3.2796, + "learning_rate": 2.1513270647088795e-05, + "loss": 3.2368, "step": 172200 }, { "epoch": 0.57, - "learning_rate": 2.14503487928949e-05, - "loss": 3.3116, + "learning_rate": 2.1496727830972123e-05, + "loss": 3.234, "step": 172300 }, { "epoch": 0.57, - "learning_rate": 2.143377905917052e-05, - "loss": 3.2819, + "learning_rate": 2.148018501485545e-05, + "loss": 3.2417, "step": 172400 }, { "epoch": 0.57, - "learning_rate": 2.1417209325446142e-05, - "loss": 3.2934, + "learning_rate": 2.1463642198738775e-05, + "loss": 3.2377, "step": 172500 }, { "epoch": 0.57, - "learning_rate": 2.1400639591721763e-05, - "loss": 3.2748, + "learning_rate": 2.1447099382622103e-05, + "loss": 3.2333, "step": 172600 }, { "epoch": 0.57, - "learning_rate": 2.138406985799738e-05, - "loss": 3.2857, + "learning_rate": 2.143055656650543e-05, + "loss": 3.2384, "step": 172700 }, { "epoch": 0.57, - "learning_rate": 2.1367500124273003e-05, - "loss": 3.2928, + "learning_rate": 2.1414013750388755e-05, + "loss": 3.2461, "step": 172800 }, { "epoch": 0.57, - "learning_rate": 2.1350930390548624e-05, - "loss": 3.2837, + "learning_rate": 2.1397470934272083e-05, + "loss": 3.238, "step": 172900 }, { "epoch": 0.57, - "learning_rate": 2.1334360656824246e-05, - "loss": 3.2744, + "learning_rate": 2.138092811815541e-05, + "loss": 3.2408, "step": 173000 }, { "epoch": 0.57, - "learning_rate": 2.1317790923099867e-05, - "loss": 3.2811, + "learning_rate": 2.1364385302038736e-05, + "loss": 3.2326, "step": 173100 }, { "epoch": 0.57, - "learning_rate": 2.1301221189375485e-05, - "loss": 3.288, + "learning_rate": 2.1347842485922064e-05, + "loss": 3.2286, "step": 173200 }, { "epoch": 0.57, - "learning_rate": 2.1284651455651107e-05, - "loss": 3.2829, + "learning_rate": 2.1331299669805392e-05, + "loss": 3.2332, "step": 173300 }, { "epoch": 0.57, - "learning_rate": 2.1268081721926732e-05, - "loss": 3.2829, + "learning_rate": 2.131475685368872e-05, + "loss": 3.2429, "step": 173400 }, { "epoch": 0.57, - "learning_rate": 2.125151198820235e-05, - "loss": 3.2831, + "learning_rate": 2.1298214037572044e-05, + "loss": 3.2444, "step": 173500 }, { - "epoch": 0.58, - "learning_rate": 2.123494225447797e-05, - "loss": 3.2837, + "epoch": 0.57, + "learning_rate": 2.1281671221455372e-05, + "loss": 3.2354, "step": 173600 }, { - "epoch": 0.58, - "learning_rate": 2.1218372520753593e-05, - "loss": 3.2809, + "epoch": 0.57, + "learning_rate": 2.12651284053387e-05, + "loss": 3.2424, "step": 173700 }, { "epoch": 0.58, - "learning_rate": 2.120180278702921e-05, - "loss": 3.2842, + "learning_rate": 2.1248585589222025e-05, + "loss": 3.2391, "step": 173800 }, { "epoch": 0.58, - "learning_rate": 2.1185233053304836e-05, - "loss": 3.2762, + "learning_rate": 2.1232042773105352e-05, + "loss": 3.247, "step": 173900 }, { "epoch": 0.58, - "learning_rate": 2.1168663319580457e-05, - "loss": 3.289, + "learning_rate": 2.121549995698868e-05, + "loss": 3.233, "step": 174000 }, { "epoch": 0.58, - "learning_rate": 2.1152093585856075e-05, - "loss": 3.2765, + "learning_rate": 2.1198957140872005e-05, + "loss": 3.2309, "step": 174100 }, { "epoch": 0.58, - "learning_rate": 2.1135523852131697e-05, - "loss": 3.278, + "learning_rate": 2.1182414324755333e-05, + "loss": 3.2377, "step": 174200 }, { "epoch": 0.58, - "learning_rate": 2.1118954118407318e-05, - "loss": 3.2938, + "learning_rate": 2.116587150863866e-05, + "loss": 3.2439, "step": 174300 }, { "epoch": 0.58, - "learning_rate": 2.110238438468294e-05, - "loss": 3.268, + "learning_rate": 2.1149328692521985e-05, + "loss": 3.2305, "step": 174400 }, { "epoch": 0.58, - "learning_rate": 2.108581465095856e-05, - "loss": 3.2875, + "learning_rate": 2.1132785876405313e-05, + "loss": 3.237, "step": 174500 }, { "epoch": 0.58, - "learning_rate": 2.1069244917234183e-05, - "loss": 3.2822, + "learning_rate": 2.111624306028864e-05, + "loss": 3.2385, "step": 174600 }, { "epoch": 0.58, - "learning_rate": 2.10526751835098e-05, - "loss": 3.2809, + "learning_rate": 2.1099700244171966e-05, + "loss": 3.2425, "step": 174700 }, { "epoch": 0.58, - "learning_rate": 2.1036105449785422e-05, - "loss": 3.2876, + "learning_rate": 2.1083157428055294e-05, + "loss": 3.2439, "step": 174800 }, { "epoch": 0.58, - "learning_rate": 2.1019535716061044e-05, - "loss": 3.282, + "learning_rate": 2.106661461193862e-05, + "loss": 3.2435, "step": 174900 }, { "epoch": 0.58, - "learning_rate": 2.1002965982336665e-05, - "loss": 3.2731, + "learning_rate": 2.1050071795821946e-05, + "loss": 3.2269, "step": 175000 }, { "epoch": 0.58, - "learning_rate": 2.0986396248612287e-05, - "loss": 3.2717, + "learning_rate": 2.1033528979705274e-05, + "loss": 3.2308, "step": 175100 }, { "epoch": 0.58, - "learning_rate": 2.0969826514887908e-05, - "loss": 3.2753, + "learning_rate": 2.1016986163588602e-05, + "loss": 3.236, "step": 175200 }, { "epoch": 0.58, - "learning_rate": 2.0953256781163526e-05, - "loss": 3.2735, + "learning_rate": 2.100044334747193e-05, + "loss": 3.2448, "step": 175300 }, { "epoch": 0.58, - "learning_rate": 2.0936687047439148e-05, - "loss": 3.291, + "learning_rate": 2.0983900531355254e-05, + "loss": 3.2304, "step": 175400 }, { "epoch": 0.58, - "learning_rate": 2.092011731371477e-05, - "loss": 3.2831, + "learning_rate": 2.0967357715238582e-05, + "loss": 3.2432, "step": 175500 }, { "epoch": 0.58, - "learning_rate": 2.090354757999039e-05, - "loss": 3.2756, + "learning_rate": 2.095081489912191e-05, + "loss": 3.2307, "step": 175600 }, { "epoch": 0.58, - "learning_rate": 2.0886977846266012e-05, - "loss": 3.2659, + "learning_rate": 2.0934272083005235e-05, + "loss": 3.2353, "step": 175700 }, { "epoch": 0.58, - "learning_rate": 2.087040811254163e-05, - "loss": 3.2736, + "learning_rate": 2.0917729266888563e-05, + "loss": 3.2319, "step": 175800 }, { "epoch": 0.58, - "learning_rate": 2.085383837881725e-05, - "loss": 3.273, + "learning_rate": 2.090118645077189e-05, + "loss": 3.241, "step": 175900 }, { "epoch": 0.58, - "learning_rate": 2.0837268645092876e-05, - "loss": 3.2855, + "learning_rate": 2.0884643634655215e-05, + "loss": 3.2408, "step": 176000 }, { "epoch": 0.58, - "learning_rate": 2.0820698911368494e-05, - "loss": 3.2834, + "learning_rate": 2.086810081853854e-05, + "loss": 3.2493, "step": 176100 }, { "epoch": 0.58, - "learning_rate": 2.0804129177644116e-05, - "loss": 3.2762, + "learning_rate": 2.085155800242187e-05, + "loss": 3.2384, "step": 176200 }, { "epoch": 0.58, - "learning_rate": 2.0787559443919737e-05, - "loss": 3.2869, + "learning_rate": 2.0835015186305196e-05, + "loss": 3.2481, "step": 176300 }, { "epoch": 0.58, - "learning_rate": 2.0770989710195355e-05, - "loss": 3.277, + "learning_rate": 2.0818472370188523e-05, + "loss": 3.2388, "step": 176400 }, { "epoch": 0.58, - "learning_rate": 2.075441997647098e-05, - "loss": 3.2718, + "learning_rate": 2.080192955407185e-05, + "loss": 3.242, "step": 176500 }, { - "epoch": 0.59, - "learning_rate": 2.0737850242746602e-05, - "loss": 3.2905, + "epoch": 0.58, + "learning_rate": 2.0785386737955176e-05, + "loss": 3.2267, "step": 176600 }, { - "epoch": 0.59, - "learning_rate": 2.072128050902222e-05, - "loss": 3.2746, + "epoch": 0.58, + "learning_rate": 2.0768843921838504e-05, + "loss": 3.2264, "step": 176700 }, { - "epoch": 0.59, - "learning_rate": 2.070471077529784e-05, - "loss": 3.2803, + "epoch": 0.58, + "learning_rate": 2.0752301105721832e-05, + "loss": 3.2495, "step": 176800 }, { "epoch": 0.59, - "learning_rate": 2.0688141041573463e-05, - "loss": 3.27, + "learning_rate": 2.0735758289605156e-05, + "loss": 3.2315, "step": 176900 }, { "epoch": 0.59, - "learning_rate": 2.0671571307849084e-05, - "loss": 3.2631, + "learning_rate": 2.0719215473488484e-05, + "loss": 3.2329, "step": 177000 }, { "epoch": 0.59, - "learning_rate": 2.0655001574124706e-05, - "loss": 3.2897, + "learning_rate": 2.0702672657371812e-05, + "loss": 3.2295, "step": 177100 }, { "epoch": 0.59, - "learning_rate": 2.0638431840400327e-05, - "loss": 3.2775, + "learning_rate": 2.0686129841255137e-05, + "loss": 3.2336, "step": 177200 }, { "epoch": 0.59, - "learning_rate": 2.0621862106675945e-05, - "loss": 3.2823, + "learning_rate": 2.0669587025138465e-05, + "loss": 3.2401, "step": 177300 }, { "epoch": 0.59, - "learning_rate": 2.0605292372951567e-05, - "loss": 3.2917, + "learning_rate": 2.065304420902179e-05, + "loss": 3.2386, "step": 177400 }, { "epoch": 0.59, - "learning_rate": 2.0588722639227188e-05, - "loss": 3.274, + "learning_rate": 2.063650139290512e-05, + "loss": 3.2328, "step": 177500 }, { "epoch": 0.59, - "learning_rate": 2.057215290550281e-05, - "loss": 3.2779, + "learning_rate": 2.0619958576788445e-05, + "loss": 3.2413, "step": 177600 }, { "epoch": 0.59, - "learning_rate": 2.055558317177843e-05, - "loss": 3.2712, + "learning_rate": 2.060341576067177e-05, + "loss": 3.2395, "step": 177700 }, { "epoch": 0.59, - "learning_rate": 2.0539013438054053e-05, - "loss": 3.2744, + "learning_rate": 2.05868729445551e-05, + "loss": 3.2279, "step": 177800 }, { "epoch": 0.59, - "learning_rate": 2.052244370432967e-05, - "loss": 3.2781, + "learning_rate": 2.0570330128438425e-05, + "loss": 3.2368, "step": 177900 }, { "epoch": 0.59, - "learning_rate": 2.0505873970605292e-05, - "loss": 3.2889, + "learning_rate": 2.055378731232175e-05, + "loss": 3.2239, "step": 178000 }, { "epoch": 0.59, - "learning_rate": 2.0489304236880914e-05, - "loss": 3.2706, + "learning_rate": 2.053724449620508e-05, + "loss": 3.2394, "step": 178100 }, { "epoch": 0.59, - "learning_rate": 2.0472734503156535e-05, - "loss": 3.2853, + "learning_rate": 2.0520701680088406e-05, + "loss": 3.2384, "step": 178200 }, { "epoch": 0.59, - "learning_rate": 2.0456164769432157e-05, - "loss": 3.2737, + "learning_rate": 2.050415886397173e-05, + "loss": 3.2459, "step": 178300 }, { "epoch": 0.59, - "learning_rate": 2.0439595035707775e-05, - "loss": 3.2676, + "learning_rate": 2.048761604785506e-05, + "loss": 3.2515, "step": 178400 }, { "epoch": 0.59, - "learning_rate": 2.04230253019834e-05, - "loss": 3.2788, + "learning_rate": 2.0471073231738386e-05, + "loss": 3.2241, "step": 178500 }, { "epoch": 0.59, - "learning_rate": 2.040645556825902e-05, - "loss": 3.2943, + "learning_rate": 2.0454530415621714e-05, + "loss": 3.2369, "step": 178600 }, { "epoch": 0.59, - "learning_rate": 2.038988583453464e-05, - "loss": 3.269, + "learning_rate": 2.043798759950504e-05, + "loss": 3.2441, "step": 178700 }, { "epoch": 0.59, - "learning_rate": 2.037331610081026e-05, - "loss": 3.265, + "learning_rate": 2.0421444783388366e-05, + "loss": 3.2333, "step": 178800 }, { "epoch": 0.59, - "learning_rate": 2.0356746367085882e-05, - "loss": 3.2816, + "learning_rate": 2.0404901967271694e-05, + "loss": 3.2325, "step": 178900 }, { "epoch": 0.59, - "learning_rate": 2.0340176633361503e-05, - "loss": 3.2758, + "learning_rate": 2.038835915115502e-05, + "loss": 3.2315, "step": 179000 }, { "epoch": 0.59, - "learning_rate": 2.0323606899637125e-05, - "loss": 3.2659, + "learning_rate": 2.0371816335038347e-05, + "loss": 3.2207, "step": 179100 }, { "epoch": 0.59, - "learning_rate": 2.0307037165912746e-05, - "loss": 3.2791, + "learning_rate": 2.0355273518921675e-05, + "loss": 3.2392, "step": 179200 }, { "epoch": 0.59, - "learning_rate": 2.0290467432188364e-05, - "loss": 3.2586, + "learning_rate": 2.0338730702805e-05, + "loss": 3.243, "step": 179300 }, { "epoch": 0.59, - "learning_rate": 2.0273897698463986e-05, - "loss": 3.2832, + "learning_rate": 2.0322187886688327e-05, + "loss": 3.2299, "step": 179400 }, { "epoch": 0.59, - "learning_rate": 2.0257327964739607e-05, - "loss": 3.2727, + "learning_rate": 2.0305645070571655e-05, + "loss": 3.229, "step": 179500 }, { - "epoch": 0.6, - "learning_rate": 2.024075823101523e-05, - "loss": 3.2878, + "epoch": 0.59, + "learning_rate": 2.028910225445498e-05, + "loss": 3.25, "step": 179600 }, { - "epoch": 0.6, - "learning_rate": 2.022418849729085e-05, - "loss": 3.2717, + "epoch": 0.59, + "learning_rate": 2.027255943833831e-05, + "loss": 3.2382, "step": 179700 }, { - "epoch": 0.6, - "learning_rate": 2.0207618763566472e-05, - "loss": 3.2736, + "epoch": 0.59, + "learning_rate": 2.0256016622221636e-05, + "loss": 3.2351, "step": 179800 }, { "epoch": 0.6, - "learning_rate": 2.019104902984209e-05, - "loss": 3.2894, + "learning_rate": 2.023947380610496e-05, + "loss": 3.2375, "step": 179900 }, { "epoch": 0.6, - "learning_rate": 2.017447929611771e-05, - "loss": 3.2825, + "learning_rate": 2.0222930989988288e-05, + "loss": 3.2347, "step": 180000 }, { "epoch": 0.6, - "learning_rate": 2.0157909562393336e-05, - "loss": 3.2744, + "learning_rate": 2.0206388173871616e-05, + "loss": 3.2456, "step": 180100 }, { "epoch": 0.6, - "learning_rate": 2.0141339828668954e-05, - "loss": 3.2714, + "learning_rate": 2.018984535775494e-05, + "loss": 3.2344, "step": 180200 }, { "epoch": 0.6, - "learning_rate": 2.0124770094944576e-05, - "loss": 3.2727, + "learning_rate": 2.017330254163827e-05, + "loss": 3.2231, "step": 180300 }, { "epoch": 0.6, - "learning_rate": 2.0108200361220197e-05, - "loss": 3.2714, + "learning_rate": 2.0156759725521596e-05, + "loss": 3.2298, "step": 180400 }, { "epoch": 0.6, - "learning_rate": 2.0091630627495815e-05, - "loss": 3.2763, + "learning_rate": 2.014021690940492e-05, + "loss": 3.2341, "step": 180500 }, { "epoch": 0.6, - "learning_rate": 2.007506089377144e-05, - "loss": 3.2797, + "learning_rate": 2.012367409328825e-05, + "loss": 3.2487, "step": 180600 }, { "epoch": 0.6, - "learning_rate": 2.0058491160047058e-05, - "loss": 3.2749, + "learning_rate": 2.0107131277171577e-05, + "loss": 3.2404, "step": 180700 }, { "epoch": 0.6, - "learning_rate": 2.004192142632268e-05, - "loss": 3.279, + "learning_rate": 2.0090588461054905e-05, + "loss": 3.2449, "step": 180800 }, { "epoch": 0.6, - "learning_rate": 2.00253516925983e-05, - "loss": 3.2736, + "learning_rate": 2.007404564493823e-05, + "loss": 3.2524, "step": 180900 }, { "epoch": 0.6, - "learning_rate": 2.000878195887392e-05, - "loss": 3.2795, + "learning_rate": 2.0057502828821557e-05, + "loss": 3.2329, "step": 181000 }, { "epoch": 0.6, - "learning_rate": 1.9992212225149544e-05, - "loss": 3.2749, + "learning_rate": 2.0040960012704885e-05, + "loss": 3.2242, "step": 181100 }, { "epoch": 0.6, - "learning_rate": 1.9975642491425166e-05, - "loss": 3.2735, + "learning_rate": 2.002441719658821e-05, + "loss": 3.2354, "step": 181200 }, { "epoch": 0.6, - "learning_rate": 1.9959072757700784e-05, - "loss": 3.2742, + "learning_rate": 2.0007874380471537e-05, + "loss": 3.2452, "step": 181300 }, { "epoch": 0.6, - "learning_rate": 1.9942503023976405e-05, - "loss": 3.2768, + "learning_rate": 1.9991331564354865e-05, + "loss": 3.2281, "step": 181400 }, { "epoch": 0.6, - "learning_rate": 1.9925933290252027e-05, - "loss": 3.2842, + "learning_rate": 1.997478874823819e-05, + "loss": 3.243, "step": 181500 }, { "epoch": 0.6, - "learning_rate": 1.9909363556527648e-05, - "loss": 3.2691, + "learning_rate": 1.9958245932121518e-05, + "loss": 3.2374, "step": 181600 }, { "epoch": 0.6, - "learning_rate": 1.989279382280327e-05, - "loss": 3.2688, + "learning_rate": 1.9941703116004846e-05, + "loss": 3.2342, "step": 181700 }, { "epoch": 0.6, - "learning_rate": 1.987622408907889e-05, - "loss": 3.2819, + "learning_rate": 1.992516029988817e-05, + "loss": 3.2345, "step": 181800 }, { "epoch": 0.6, - "learning_rate": 1.985965435535451e-05, - "loss": 3.2958, + "learning_rate": 1.9908617483771498e-05, + "loss": 3.2341, "step": 181900 }, { "epoch": 0.6, - "learning_rate": 1.984308462163013e-05, - "loss": 3.2733, + "learning_rate": 1.9892074667654826e-05, + "loss": 3.2379, "step": 182000 }, { "epoch": 0.6, - "learning_rate": 1.9826514887905752e-05, - "loss": 3.284, + "learning_rate": 1.987553185153815e-05, + "loss": 3.2395, "step": 182100 }, { "epoch": 0.6, - "learning_rate": 1.9809945154181373e-05, - "loss": 3.2737, + "learning_rate": 1.985898903542148e-05, + "loss": 3.2301, "step": 182200 }, { "epoch": 0.6, - "learning_rate": 1.9793375420456995e-05, - "loss": 3.2678, + "learning_rate": 1.9842446219304807e-05, + "loss": 3.2414, "step": 182300 }, { "epoch": 0.6, - "learning_rate": 1.9776805686732616e-05, - "loss": 3.2785, + "learning_rate": 1.982590340318813e-05, + "loss": 3.2423, "step": 182400 }, { "epoch": 0.6, - "learning_rate": 1.9760235953008234e-05, - "loss": 3.2794, + "learning_rate": 1.980936058707146e-05, + "loss": 3.2311, "step": 182500 }, { - "epoch": 0.61, - "learning_rate": 1.9743666219283856e-05, - "loss": 3.2725, + "epoch": 0.6, + "learning_rate": 1.9792817770954787e-05, + "loss": 3.2344, "step": 182600 }, { - "epoch": 0.61, - "learning_rate": 1.972709648555948e-05, - "loss": 3.2826, + "epoch": 0.6, + "learning_rate": 1.9776274954838115e-05, + "loss": 3.2277, "step": 182700 }, { - "epoch": 0.61, - "learning_rate": 1.97105267518351e-05, - "loss": 3.2711, + "epoch": 0.6, + "learning_rate": 1.975973213872144e-05, + "loss": 3.2379, "step": 182800 }, { "epoch": 0.61, - "learning_rate": 1.969395701811072e-05, - "loss": 3.28, + "learning_rate": 1.9743189322604767e-05, + "loss": 3.2414, "step": 182900 }, { "epoch": 0.61, - "learning_rate": 1.9677387284386342e-05, - "loss": 3.2706, + "learning_rate": 1.9726646506488095e-05, + "loss": 3.2279, "step": 183000 }, { "epoch": 0.61, - "learning_rate": 1.966081755066196e-05, - "loss": 3.2668, + "learning_rate": 1.971010369037142e-05, + "loss": 3.247, "step": 183100 }, { "epoch": 0.61, - "learning_rate": 1.9644247816937585e-05, - "loss": 3.2762, + "learning_rate": 1.9693560874254748e-05, + "loss": 3.2421, "step": 183200 }, { "epoch": 0.61, - "learning_rate": 1.9627678083213203e-05, - "loss": 3.2763, + "learning_rate": 1.9677018058138076e-05, + "loss": 3.2455, "step": 183300 }, { "epoch": 0.61, - "learning_rate": 1.9611108349488824e-05, - "loss": 3.2811, + "learning_rate": 1.96604752420214e-05, + "loss": 3.2471, "step": 183400 }, { "epoch": 0.61, - "learning_rate": 1.9594538615764446e-05, - "loss": 3.2608, + "learning_rate": 1.9643932425904728e-05, + "loss": 3.2436, "step": 183500 }, { "epoch": 0.61, - "learning_rate": 1.9577968882040064e-05, - "loss": 3.2719, + "learning_rate": 1.9627389609788056e-05, + "loss": 3.2282, "step": 183600 }, { "epoch": 0.61, - "learning_rate": 1.956139914831569e-05, - "loss": 3.2689, + "learning_rate": 1.961084679367138e-05, + "loss": 3.2397, "step": 183700 }, { "epoch": 0.61, - "learning_rate": 1.954482941459131e-05, - "loss": 3.272, + "learning_rate": 1.959430397755471e-05, + "loss": 3.2301, "step": 183800 }, { "epoch": 0.61, - "learning_rate": 1.9528259680866928e-05, - "loss": 3.2739, + "learning_rate": 1.9577761161438036e-05, + "loss": 3.2323, "step": 183900 }, { "epoch": 0.61, - "learning_rate": 1.951168994714255e-05, - "loss": 3.2628, + "learning_rate": 1.956121834532136e-05, + "loss": 3.2347, "step": 184000 }, { "epoch": 0.61, - "learning_rate": 1.949512021341817e-05, - "loss": 3.2716, + "learning_rate": 1.954467552920469e-05, + "loss": 3.2405, "step": 184100 }, { "epoch": 0.61, - "learning_rate": 1.9478550479693793e-05, - "loss": 3.2695, + "learning_rate": 1.9528132713088017e-05, + "loss": 3.2259, "step": 184200 }, { "epoch": 0.61, - "learning_rate": 1.9461980745969414e-05, - "loss": 3.2758, + "learning_rate": 1.951158989697134e-05, + "loss": 3.2456, "step": 184300 }, { "epoch": 0.61, - "learning_rate": 1.9445411012245036e-05, - "loss": 3.2661, + "learning_rate": 1.949504708085467e-05, + "loss": 3.2307, "step": 184400 }, { "epoch": 0.61, - "learning_rate": 1.9428841278520654e-05, - "loss": 3.2644, + "learning_rate": 1.9478504264737997e-05, + "loss": 3.2445, "step": 184500 }, { "epoch": 0.61, - "learning_rate": 1.9412271544796275e-05, - "loss": 3.2676, + "learning_rate": 1.946196144862132e-05, + "loss": 3.2443, "step": 184600 }, { "epoch": 0.61, - "learning_rate": 1.9395701811071897e-05, - "loss": 3.2685, + "learning_rate": 1.944541863250465e-05, + "loss": 3.2369, "step": 184700 }, { "epoch": 0.61, - "learning_rate": 1.9379132077347518e-05, - "loss": 3.2738, + "learning_rate": 1.9428875816387977e-05, + "loss": 3.2373, "step": 184800 }, { "epoch": 0.61, - "learning_rate": 1.936256234362314e-05, - "loss": 3.2685, + "learning_rate": 1.9412333000271305e-05, + "loss": 3.2426, "step": 184900 }, { "epoch": 0.61, - "learning_rate": 1.934599260989876e-05, - "loss": 3.2767, + "learning_rate": 1.939579018415463e-05, + "loss": 3.2375, "step": 185000 }, { "epoch": 0.61, - "learning_rate": 1.932942287617438e-05, - "loss": 3.283, + "learning_rate": 1.9379247368037954e-05, + "loss": 3.236, "step": 185100 }, { "epoch": 0.61, - "learning_rate": 1.931285314245e-05, - "loss": 3.2907, + "learning_rate": 1.9362704551921286e-05, + "loss": 3.2431, "step": 185200 }, { "epoch": 0.61, - "learning_rate": 1.9296283408725625e-05, - "loss": 3.2751, + "learning_rate": 1.934616173580461e-05, + "loss": 3.2381, "step": 185300 }, { "epoch": 0.61, - "learning_rate": 1.9279713675001243e-05, - "loss": 3.2794, + "learning_rate": 1.9329618919687935e-05, + "loss": 3.2361, "step": 185400 }, { "epoch": 0.61, - "learning_rate": 1.9263143941276865e-05, - "loss": 3.2789, + "learning_rate": 1.9313076103571266e-05, + "loss": 3.2282, "step": 185500 }, { - "epoch": 0.62, - "learning_rate": 1.9246574207552486e-05, - "loss": 3.2792, + "epoch": 0.61, + "learning_rate": 1.929653328745459e-05, + "loss": 3.2299, "step": 185600 }, { - "epoch": 0.62, - "learning_rate": 1.9230004473828104e-05, - "loss": 3.2708, + "epoch": 0.61, + "learning_rate": 1.9279990471337915e-05, + "loss": 3.2396, "step": 185700 }, { - "epoch": 0.62, - "learning_rate": 1.921343474010373e-05, - "loss": 3.2807, + "epoch": 0.61, + "learning_rate": 1.9263447655221247e-05, + "loss": 3.2286, "step": 185800 }, { "epoch": 0.62, - "learning_rate": 1.9196865006379347e-05, - "loss": 3.2696, + "learning_rate": 1.924690483910457e-05, + "loss": 3.2218, "step": 185900 }, { "epoch": 0.62, - "learning_rate": 1.918029527265497e-05, - "loss": 3.2728, + "learning_rate": 1.92303620229879e-05, + "loss": 3.226, "step": 186000 }, { "epoch": 0.62, - "learning_rate": 1.916372553893059e-05, - "loss": 3.2733, + "learning_rate": 1.9213819206871227e-05, + "loss": 3.2288, "step": 186100 }, { "epoch": 0.62, - "learning_rate": 1.914715580520621e-05, - "loss": 3.275, + "learning_rate": 1.919727639075455e-05, + "loss": 3.2416, "step": 186200 }, { "epoch": 0.62, - "learning_rate": 1.9130586071481833e-05, - "loss": 3.2787, + "learning_rate": 1.918073357463788e-05, + "loss": 3.2373, "step": 186300 }, { "epoch": 0.62, - "learning_rate": 1.9114016337757455e-05, - "loss": 3.2734, + "learning_rate": 1.9164190758521204e-05, + "loss": 3.2304, "step": 186400 }, { "epoch": 0.62, - "learning_rate": 1.9097446604033073e-05, - "loss": 3.2839, + "learning_rate": 1.9147647942404532e-05, + "loss": 3.2258, "step": 186500 }, { "epoch": 0.62, - "learning_rate": 1.9080876870308694e-05, - "loss": 3.2824, + "learning_rate": 1.913110512628786e-05, + "loss": 3.2384, "step": 186600 }, { "epoch": 0.62, - "learning_rate": 1.9064307136584316e-05, - "loss": 3.2762, + "learning_rate": 1.9114562310171184e-05, + "loss": 3.2358, "step": 186700 }, { "epoch": 0.62, - "learning_rate": 1.9047737402859937e-05, - "loss": 3.2758, + "learning_rate": 1.9098019494054512e-05, + "loss": 3.2259, "step": 186800 }, { "epoch": 0.62, - "learning_rate": 1.903116766913556e-05, - "loss": 3.2933, + "learning_rate": 1.908147667793784e-05, + "loss": 3.2362, "step": 186900 }, { "epoch": 0.62, - "learning_rate": 1.901459793541118e-05, - "loss": 3.2803, + "learning_rate": 1.9064933861821165e-05, + "loss": 3.2283, "step": 187000 }, { "epoch": 0.62, - "learning_rate": 1.8998028201686798e-05, - "loss": 3.2726, + "learning_rate": 1.9048391045704496e-05, + "loss": 3.2307, "step": 187100 }, { "epoch": 0.62, - "learning_rate": 1.898145846796242e-05, - "loss": 3.2732, + "learning_rate": 1.903184822958782e-05, + "loss": 3.2333, "step": 187200 }, { "epoch": 0.62, - "learning_rate": 1.896488873423804e-05, - "loss": 3.2787, + "learning_rate": 1.9015305413471145e-05, + "loss": 3.2289, "step": 187300 }, { "epoch": 0.62, - "learning_rate": 1.8948319000513663e-05, - "loss": 3.2736, + "learning_rate": 1.8998762597354476e-05, + "loss": 3.2501, "step": 187400 }, { "epoch": 0.62, - "learning_rate": 1.8931749266789284e-05, - "loss": 3.2773, + "learning_rate": 1.89822197812378e-05, + "loss": 3.2408, "step": 187500 }, { "epoch": 0.62, - "learning_rate": 1.8915179533064906e-05, - "loss": 3.2698, + "learning_rate": 1.8965676965121125e-05, + "loss": 3.2438, "step": 187600 }, { "epoch": 0.62, - "learning_rate": 1.8898609799340524e-05, - "loss": 3.2671, + "learning_rate": 1.8949134149004453e-05, + "loss": 3.2267, "step": 187700 }, { "epoch": 0.62, - "learning_rate": 1.8882040065616145e-05, - "loss": 3.281, + "learning_rate": 1.893259133288778e-05, + "loss": 3.2329, "step": 187800 }, { "epoch": 0.62, - "learning_rate": 1.886547033189177e-05, - "loss": 3.2684, + "learning_rate": 1.8916048516771106e-05, + "loss": 3.22, "step": 187900 }, { "epoch": 0.62, - "learning_rate": 1.8848900598167388e-05, - "loss": 3.2603, + "learning_rate": 1.8899505700654434e-05, + "loss": 3.2369, "step": 188000 }, { "epoch": 0.62, - "learning_rate": 1.883233086444301e-05, - "loss": 3.2864, + "learning_rate": 1.888296288453776e-05, + "loss": 3.2316, "step": 188100 }, { "epoch": 0.62, - "learning_rate": 1.881576113071863e-05, - "loss": 3.2795, + "learning_rate": 1.886642006842109e-05, + "loss": 3.2531, "step": 188200 }, { "epoch": 0.62, - "learning_rate": 1.879919139699425e-05, - "loss": 3.296, + "learning_rate": 1.8849877252304414e-05, + "loss": 3.2321, "step": 188300 }, { "epoch": 0.62, - "learning_rate": 1.8782621663269874e-05, - "loss": 3.2697, + "learning_rate": 1.8833334436187742e-05, + "loss": 3.2312, "step": 188400 }, { "epoch": 0.62, - "learning_rate": 1.8766051929545492e-05, - "loss": 3.2742, + "learning_rate": 1.881679162007107e-05, + "loss": 3.2272, "step": 188500 }, { - "epoch": 0.63, - "learning_rate": 1.8749482195821113e-05, - "loss": 3.2781, + "epoch": 0.62, + "learning_rate": 1.8800248803954394e-05, + "loss": 3.2453, "step": 188600 }, { - "epoch": 0.63, - "learning_rate": 1.8732912462096735e-05, - "loss": 3.2788, + "epoch": 0.62, + "learning_rate": 1.8783705987837722e-05, + "loss": 3.2434, "step": 188700 }, { - "epoch": 0.63, - "learning_rate": 1.8716342728372356e-05, - "loss": 3.2663, + "epoch": 0.62, + "learning_rate": 1.876716317172105e-05, + "loss": 3.2363, "step": 188800 }, { - "epoch": 0.63, - "learning_rate": 1.8699772994647978e-05, - "loss": 3.2897, + "epoch": 0.62, + "learning_rate": 1.8750620355604375e-05, + "loss": 3.2344, "step": 188900 }, { "epoch": 0.63, - "learning_rate": 1.86832032609236e-05, - "loss": 3.2654, + "learning_rate": 1.8734077539487703e-05, + "loss": 3.2372, "step": 189000 }, { "epoch": 0.63, - "learning_rate": 1.8666633527199217e-05, - "loss": 3.2786, + "learning_rate": 1.871753472337103e-05, + "loss": 3.2379, "step": 189100 }, { "epoch": 0.63, - "learning_rate": 1.865006379347484e-05, - "loss": 3.2806, + "learning_rate": 1.8700991907254355e-05, + "loss": 3.2254, "step": 189200 }, { "epoch": 0.63, - "learning_rate": 1.863349405975046e-05, - "loss": 3.2782, + "learning_rate": 1.8684449091137683e-05, + "loss": 3.238, "step": 189300 }, { "epoch": 0.63, - "learning_rate": 1.8616924326026082e-05, - "loss": 3.2772, + "learning_rate": 1.866790627502101e-05, + "loss": 3.2365, "step": 189400 }, { "epoch": 0.63, - "learning_rate": 1.8600354592301703e-05, - "loss": 3.2688, + "learning_rate": 1.8651363458904336e-05, + "loss": 3.2299, "step": 189500 }, { "epoch": 0.63, - "learning_rate": 1.8583784858577325e-05, - "loss": 3.2742, + "learning_rate": 1.8634820642787664e-05, + "loss": 3.2445, "step": 189600 }, { "epoch": 0.63, - "learning_rate": 1.8567215124852943e-05, - "loss": 3.2817, + "learning_rate": 1.861827782667099e-05, + "loss": 3.2299, "step": 189700 }, { "epoch": 0.63, - "learning_rate": 1.8550645391128564e-05, - "loss": 3.2615, + "learning_rate": 1.8601735010554316e-05, + "loss": 3.2337, "step": 189800 }, { "epoch": 0.63, - "learning_rate": 1.8534075657404186e-05, - "loss": 3.2737, + "learning_rate": 1.8585192194437644e-05, + "loss": 3.241, "step": 189900 }, { "epoch": 0.63, - "learning_rate": 1.8517505923679807e-05, - "loss": 3.2839, + "learning_rate": 1.8568649378320972e-05, + "loss": 3.2336, "step": 190000 }, { "epoch": 0.63, - "learning_rate": 1.850093618995543e-05, - "loss": 3.2694, + "learning_rate": 1.85521065622043e-05, + "loss": 3.2475, "step": 190100 }, { "epoch": 0.63, - "learning_rate": 1.848436645623105e-05, - "loss": 3.2874, + "learning_rate": 1.8535563746087624e-05, + "loss": 3.2282, "step": 190200 }, { "epoch": 0.63, - "learning_rate": 1.8467796722506668e-05, - "loss": 3.2645, + "learning_rate": 1.8519020929970952e-05, + "loss": 3.2336, "step": 190300 }, { "epoch": 0.63, - "learning_rate": 1.8451226988782293e-05, - "loss": 3.2906, + "learning_rate": 1.850247811385428e-05, + "loss": 3.2223, "step": 190400 }, { "epoch": 0.63, - "learning_rate": 1.8434657255057915e-05, - "loss": 3.2579, + "learning_rate": 1.8485935297737605e-05, + "loss": 3.237, "step": 190500 }, { "epoch": 0.63, - "learning_rate": 1.8418087521333533e-05, - "loss": 3.2554, + "learning_rate": 1.8469392481620933e-05, + "loss": 3.2268, "step": 190600 }, { "epoch": 0.63, - "learning_rate": 1.8401517787609154e-05, - "loss": 3.2818, + "learning_rate": 1.845284966550426e-05, + "loss": 3.2294, "step": 190700 }, { "epoch": 0.63, - "learning_rate": 1.8384948053884776e-05, - "loss": 3.2676, + "learning_rate": 1.8436306849387585e-05, + "loss": 3.2349, "step": 190800 }, { "epoch": 0.63, - "learning_rate": 1.8368378320160397e-05, - "loss": 3.2777, + "learning_rate": 1.8419764033270913e-05, + "loss": 3.239, "step": 190900 }, { "epoch": 0.63, - "learning_rate": 1.835180858643602e-05, - "loss": 3.2698, + "learning_rate": 1.840322121715424e-05, + "loss": 3.2249, "step": 191000 }, { "epoch": 0.63, - "learning_rate": 1.8335238852711637e-05, - "loss": 3.2867, + "learning_rate": 1.8386678401037565e-05, + "loss": 3.2335, "step": 191100 }, { "epoch": 0.63, - "learning_rate": 1.8318669118987258e-05, - "loss": 3.2775, + "learning_rate": 1.8370135584920893e-05, + "loss": 3.2271, "step": 191200 }, { "epoch": 0.63, - "learning_rate": 1.830209938526288e-05, - "loss": 3.2695, + "learning_rate": 1.835359276880422e-05, + "loss": 3.232, "step": 191300 }, { "epoch": 0.63, - "learning_rate": 1.82855296515385e-05, - "loss": 3.2669, + "learning_rate": 1.8337049952687546e-05, + "loss": 3.2273, "step": 191400 }, { "epoch": 0.63, - "learning_rate": 1.8268959917814122e-05, - "loss": 3.2675, + "learning_rate": 1.8320507136570874e-05, + "loss": 3.2404, "step": 191500 }, { "epoch": 0.63, - "learning_rate": 1.8252390184089744e-05, - "loss": 3.2769, + "learning_rate": 1.83039643204542e-05, + "loss": 3.2454, "step": 191600 }, { - "epoch": 0.64, - "learning_rate": 1.8235820450365362e-05, - "loss": 3.2674, + "epoch": 0.63, + "learning_rate": 1.8287421504337526e-05, + "loss": 3.2313, "step": 191700 }, { - "epoch": 0.64, - "learning_rate": 1.8219250716640983e-05, - "loss": 3.2807, + "epoch": 0.63, + "learning_rate": 1.8270878688220854e-05, + "loss": 3.2249, "step": 191800 }, { - "epoch": 0.64, - "learning_rate": 1.8202680982916605e-05, - "loss": 3.2706, + "epoch": 0.63, + "learning_rate": 1.8254335872104182e-05, + "loss": 3.2219, "step": 191900 }, { "epoch": 0.64, - "learning_rate": 1.8186111249192226e-05, - "loss": 3.2708, + "learning_rate": 1.8237793055987507e-05, + "loss": 3.2312, "step": 192000 }, { "epoch": 0.64, - "learning_rate": 1.8169541515467848e-05, - "loss": 3.2712, + "learning_rate": 1.8221250239870835e-05, + "loss": 3.2344, "step": 192100 }, { "epoch": 0.64, - "learning_rate": 1.815297178174347e-05, - "loss": 3.2755, + "learning_rate": 1.8204707423754162e-05, + "loss": 3.2349, "step": 192200 }, { "epoch": 0.64, - "learning_rate": 1.8136402048019087e-05, - "loss": 3.289, + "learning_rate": 1.818816460763749e-05, + "loss": 3.2447, "step": 192300 }, { "epoch": 0.64, - "learning_rate": 1.811983231429471e-05, - "loss": 3.2719, + "learning_rate": 1.8171621791520815e-05, + "loss": 3.2212, "step": 192400 }, { "epoch": 0.64, - "learning_rate": 1.8103262580570334e-05, - "loss": 3.2822, + "learning_rate": 1.8155078975404143e-05, + "loss": 3.2284, "step": 192500 }, { "epoch": 0.64, - "learning_rate": 1.8086692846845952e-05, - "loss": 3.2781, + "learning_rate": 1.813853615928747e-05, + "loss": 3.2247, "step": 192600 }, { "epoch": 0.64, - "learning_rate": 1.8070123113121573e-05, - "loss": 3.2812, + "learning_rate": 1.8121993343170795e-05, + "loss": 3.2227, "step": 192700 }, { "epoch": 0.64, - "learning_rate": 1.8053553379397195e-05, - "loss": 3.2678, + "learning_rate": 1.810545052705412e-05, + "loss": 3.2372, "step": 192800 }, { "epoch": 0.64, - "learning_rate": 1.8036983645672813e-05, - "loss": 3.2609, + "learning_rate": 1.808890771093745e-05, + "loss": 3.2332, "step": 192900 }, { "epoch": 0.64, - "learning_rate": 1.8020413911948438e-05, - "loss": 3.2811, + "learning_rate": 1.8072364894820776e-05, + "loss": 3.2189, "step": 193000 }, { "epoch": 0.64, - "learning_rate": 1.800384417822406e-05, - "loss": 3.2712, + "learning_rate": 1.80558220787041e-05, + "loss": 3.2395, "step": 193100 }, { "epoch": 0.64, - "learning_rate": 1.7987274444499677e-05, - "loss": 3.2702, + "learning_rate": 1.803927926258743e-05, + "loss": 3.2343, "step": 193200 }, { "epoch": 0.64, - "learning_rate": 1.79707047107753e-05, - "loss": 3.2763, + "learning_rate": 1.8022736446470756e-05, + "loss": 3.2339, "step": 193300 }, { "epoch": 0.64, - "learning_rate": 1.795413497705092e-05, - "loss": 3.2628, + "learning_rate": 1.8006193630354084e-05, + "loss": 3.2441, "step": 193400 }, { "epoch": 0.64, - "learning_rate": 1.793756524332654e-05, - "loss": 3.2723, + "learning_rate": 1.7989650814237412e-05, + "loss": 3.2307, "step": 193500 }, { "epoch": 0.64, - "learning_rate": 1.7920995509602163e-05, - "loss": 3.2768, + "learning_rate": 1.7973107998120736e-05, + "loss": 3.224, "step": 193600 }, { "epoch": 0.64, - "learning_rate": 1.790442577587778e-05, - "loss": 3.2732, + "learning_rate": 1.7956565182004064e-05, + "loss": 3.2356, "step": 193700 }, { "epoch": 0.64, - "learning_rate": 1.7887856042153403e-05, - "loss": 3.2828, + "learning_rate": 1.7940022365887392e-05, + "loss": 3.2271, "step": 193800 }, { "epoch": 0.64, - "learning_rate": 1.7871286308429024e-05, - "loss": 3.2745, + "learning_rate": 1.7923479549770717e-05, + "loss": 3.2234, "step": 193900 }, { "epoch": 0.64, - "learning_rate": 1.7854716574704646e-05, - "loss": 3.2677, + "learning_rate": 1.7906936733654045e-05, + "loss": 3.223, "step": 194000 }, { "epoch": 0.64, - "learning_rate": 1.7838146840980267e-05, - "loss": 3.2841, + "learning_rate": 1.789039391753737e-05, + "loss": 3.2417, "step": 194100 }, { "epoch": 0.64, - "learning_rate": 1.782157710725589e-05, - "loss": 3.2735, + "learning_rate": 1.7873851101420697e-05, + "loss": 3.225, "step": 194200 }, { "epoch": 0.64, - "learning_rate": 1.7805007373531507e-05, - "loss": 3.2743, + "learning_rate": 1.7857308285304025e-05, + "loss": 3.2354, "step": 194300 }, { "epoch": 0.64, - "learning_rate": 1.7788437639807128e-05, - "loss": 3.2792, + "learning_rate": 1.784076546918735e-05, + "loss": 3.2295, "step": 194400 }, { "epoch": 0.64, - "learning_rate": 1.777186790608275e-05, - "loss": 3.2849, + "learning_rate": 1.782422265307068e-05, + "loss": 3.2377, "step": 194500 }, { "epoch": 0.64, - "learning_rate": 1.775529817235837e-05, - "loss": 3.2608, + "learning_rate": 1.7807679836954005e-05, + "loss": 3.232, "step": 194600 }, { - "epoch": 0.65, - "learning_rate": 1.7738728438633992e-05, - "loss": 3.2722, + "epoch": 0.64, + "learning_rate": 1.779113702083733e-05, + "loss": 3.2402, "step": 194700 }, { - "epoch": 0.65, - "learning_rate": 1.7722158704909614e-05, - "loss": 3.2696, + "epoch": 0.64, + "learning_rate": 1.777459420472066e-05, + "loss": 3.2288, "step": 194800 }, { - "epoch": 0.65, - "learning_rate": 1.7705588971185232e-05, - "loss": 3.2643, + "epoch": 0.64, + "learning_rate": 1.7758051388603986e-05, + "loss": 3.2274, "step": 194900 }, { "epoch": 0.65, - "learning_rate": 1.7689019237460853e-05, - "loss": 3.276, + "learning_rate": 1.774150857248731e-05, + "loss": 3.2249, "step": 195000 }, { "epoch": 0.65, - "learning_rate": 1.7672449503736478e-05, - "loss": 3.2827, + "learning_rate": 1.7724965756370642e-05, + "loss": 3.2447, "step": 195100 }, { "epoch": 0.65, - "learning_rate": 1.7655879770012096e-05, - "loss": 3.2694, + "learning_rate": 1.7708422940253966e-05, + "loss": 3.2294, "step": 195200 }, { "epoch": 0.65, - "learning_rate": 1.7639310036287718e-05, - "loss": 3.2827, + "learning_rate": 1.769188012413729e-05, + "loss": 3.2179, "step": 195300 }, { "epoch": 0.65, - "learning_rate": 1.762274030256334e-05, - "loss": 3.2794, + "learning_rate": 1.767533730802062e-05, + "loss": 3.2449, "step": 195400 }, { "epoch": 0.65, - "learning_rate": 1.7606170568838957e-05, - "loss": 3.2792, + "learning_rate": 1.7658794491903947e-05, + "loss": 3.2291, "step": 195500 }, { "epoch": 0.65, - "learning_rate": 1.7589600835114582e-05, - "loss": 3.2782, + "learning_rate": 1.7642251675787275e-05, + "loss": 3.2413, "step": 195600 }, { "epoch": 0.65, - "learning_rate": 1.7573031101390204e-05, - "loss": 3.2843, + "learning_rate": 1.76257088596706e-05, + "loss": 3.2333, "step": 195700 }, { "epoch": 0.65, - "learning_rate": 1.7556461367665822e-05, - "loss": 3.269, + "learning_rate": 1.7609166043553927e-05, + "loss": 3.2383, "step": 195800 }, { "epoch": 0.65, - "learning_rate": 1.7539891633941443e-05, - "loss": 3.2571, + "learning_rate": 1.7592623227437255e-05, + "loss": 3.2328, "step": 195900 }, { "epoch": 0.65, - "learning_rate": 1.7523321900217065e-05, - "loss": 3.2737, + "learning_rate": 1.757608041132058e-05, + "loss": 3.223, "step": 196000 }, { "epoch": 0.65, - "learning_rate": 1.7506752166492686e-05, - "loss": 3.2767, + "learning_rate": 1.7559537595203907e-05, + "loss": 3.2298, "step": 196100 }, { "epoch": 0.65, - "learning_rate": 1.7490182432768308e-05, - "loss": 3.284, + "learning_rate": 1.7542994779087235e-05, + "loss": 3.2373, "step": 196200 }, { "epoch": 0.65, - "learning_rate": 1.7473612699043926e-05, - "loss": 3.2747, + "learning_rate": 1.752645196297056e-05, + "loss": 3.2341, "step": 196300 }, { "epoch": 0.65, - "learning_rate": 1.7457042965319547e-05, - "loss": 3.2852, + "learning_rate": 1.7509909146853888e-05, + "loss": 3.2228, "step": 196400 }, { "epoch": 0.65, - "learning_rate": 1.744047323159517e-05, - "loss": 3.2653, + "learning_rate": 1.7493366330737216e-05, + "loss": 3.2404, "step": 196500 }, { "epoch": 0.65, - "learning_rate": 1.742390349787079e-05, - "loss": 3.2773, + "learning_rate": 1.747682351462054e-05, + "loss": 3.2297, "step": 196600 }, { "epoch": 0.65, - "learning_rate": 1.740733376414641e-05, - "loss": 3.2758, + "learning_rate": 1.7460280698503868e-05, + "loss": 3.2413, "step": 196700 }, { "epoch": 0.65, - "learning_rate": 1.7390764030422033e-05, - "loss": 3.2789, + "learning_rate": 1.7443737882387196e-05, + "loss": 3.2234, "step": 196800 }, { "epoch": 0.65, - "learning_rate": 1.737419429669765e-05, - "loss": 3.2661, + "learning_rate": 1.742719506627052e-05, + "loss": 3.2257, "step": 196900 }, { "epoch": 0.65, - "learning_rate": 1.7357624562973273e-05, - "loss": 3.2812, + "learning_rate": 1.741065225015385e-05, + "loss": 3.2256, "step": 197000 }, { "epoch": 0.65, - "learning_rate": 1.7341054829248894e-05, - "loss": 3.262, + "learning_rate": 1.7394109434037176e-05, + "loss": 3.2354, "step": 197100 }, { "epoch": 0.65, - "learning_rate": 1.7324485095524516e-05, - "loss": 3.2723, + "learning_rate": 1.73775666179205e-05, + "loss": 3.2416, "step": 197200 }, { "epoch": 0.65, - "learning_rate": 1.7307915361800137e-05, - "loss": 3.2711, + "learning_rate": 1.736102380180383e-05, + "loss": 3.2235, "step": 197300 }, { "epoch": 0.65, - "learning_rate": 1.729134562807576e-05, - "loss": 3.2912, + "learning_rate": 1.7344480985687157e-05, + "loss": 3.2341, "step": 197400 }, { "epoch": 0.65, - "learning_rate": 1.7274775894351377e-05, - "loss": 3.2658, + "learning_rate": 1.732793816957048e-05, + "loss": 3.2408, "step": 197500 }, { "epoch": 0.65, - "learning_rate": 1.7258206160626998e-05, - "loss": 3.2698, + "learning_rate": 1.731139535345381e-05, + "loss": 3.233, "step": 197600 }, { - "epoch": 0.66, - "learning_rate": 1.7241636426902623e-05, - "loss": 3.2782, + "epoch": 0.65, + "learning_rate": 1.7294852537337137e-05, + "loss": 3.2129, "step": 197700 }, { - "epoch": 0.66, - "learning_rate": 1.722506669317824e-05, - "loss": 3.2734, + "epoch": 0.65, + "learning_rate": 1.7278309721220465e-05, + "loss": 3.2282, "step": 197800 }, { - "epoch": 0.66, - "learning_rate": 1.7208496959453862e-05, - "loss": 3.2854, + "epoch": 0.65, + "learning_rate": 1.726176690510379e-05, + "loss": 3.2286, "step": 197900 }, { "epoch": 0.66, - "learning_rate": 1.7191927225729484e-05, - "loss": 3.278, + "learning_rate": 1.7245224088987118e-05, + "loss": 3.2334, "step": 198000 }, { "epoch": 0.66, - "learning_rate": 1.7175357492005102e-05, - "loss": 3.2804, + "learning_rate": 1.7228681272870446e-05, + "loss": 3.2379, "step": 198100 }, { "epoch": 0.66, - "learning_rate": 1.7158787758280727e-05, - "loss": 3.2614, + "learning_rate": 1.721213845675377e-05, + "loss": 3.2225, "step": 198200 }, { "epoch": 0.66, - "learning_rate": 1.7142218024556348e-05, - "loss": 3.2862, + "learning_rate": 1.7195595640637098e-05, + "loss": 3.2176, "step": 198300 }, { "epoch": 0.66, - "learning_rate": 1.7125648290831966e-05, - "loss": 3.2717, + "learning_rate": 1.7179052824520426e-05, + "loss": 3.2342, "step": 198400 }, { "epoch": 0.66, - "learning_rate": 1.7109078557107588e-05, - "loss": 3.2834, + "learning_rate": 1.716251000840375e-05, + "loss": 3.21, "step": 198500 }, { "epoch": 0.66, - "learning_rate": 1.709250882338321e-05, - "loss": 3.2916, + "learning_rate": 1.714596719228708e-05, + "loss": 3.2301, "step": 198600 }, { "epoch": 0.66, - "learning_rate": 1.707593908965883e-05, - "loss": 3.2828, + "learning_rate": 1.7129424376170406e-05, + "loss": 3.247, "step": 198700 }, { "epoch": 0.66, - "learning_rate": 1.7059369355934452e-05, - "loss": 3.282, + "learning_rate": 1.711288156005373e-05, + "loss": 3.2358, "step": 198800 }, { "epoch": 0.66, - "learning_rate": 1.704279962221007e-05, - "loss": 3.262, + "learning_rate": 1.709633874393706e-05, + "loss": 3.2371, "step": 198900 }, { "epoch": 0.66, - "learning_rate": 1.7026229888485692e-05, - "loss": 3.2708, + "learning_rate": 1.7079795927820387e-05, + "loss": 3.2258, "step": 199000 }, { "epoch": 0.66, - "learning_rate": 1.7009660154761313e-05, - "loss": 3.2679, + "learning_rate": 1.706325311170371e-05, + "loss": 3.2237, "step": 199100 }, { "epoch": 0.66, - "learning_rate": 1.6993090421036935e-05, - "loss": 3.2758, + "learning_rate": 1.704671029558704e-05, + "loss": 3.2227, "step": 199200 }, { "epoch": 0.66, - "learning_rate": 1.6976520687312556e-05, - "loss": 3.2707, + "learning_rate": 1.7030167479470367e-05, + "loss": 3.2289, "step": 199300 }, { "epoch": 0.66, - "learning_rate": 1.6959950953588178e-05, - "loss": 3.2676, + "learning_rate": 1.701362466335369e-05, + "loss": 3.2273, "step": 199400 }, { "epoch": 0.66, - "learning_rate": 1.6943381219863796e-05, - "loss": 3.2816, + "learning_rate": 1.699708184723702e-05, + "loss": 3.2309, "step": 199500 }, { "epoch": 0.66, - "learning_rate": 1.6926811486139417e-05, - "loss": 3.2743, + "learning_rate": 1.6980539031120347e-05, + "loss": 3.2358, "step": 199600 }, { "epoch": 0.66, - "learning_rate": 1.691024175241504e-05, - "loss": 3.2635, + "learning_rate": 1.6963996215003675e-05, + "loss": 3.2187, "step": 199700 }, { "epoch": 0.66, - "learning_rate": 1.689367201869066e-05, - "loss": 3.2796, + "learning_rate": 1.6947453398887e-05, + "loss": 3.2218, "step": 199800 }, { "epoch": 0.66, - "learning_rate": 1.687710228496628e-05, - "loss": 3.272, + "learning_rate": 1.6930910582770328e-05, + "loss": 3.2304, "step": 199900 }, { "epoch": 0.66, - "learning_rate": 1.6860532551241903e-05, - "loss": 3.2711, + "learning_rate": 1.6914367766653656e-05, + "loss": 3.2219, "step": 200000 }, { "epoch": 0.66, - "learning_rate": 1.684396281751752e-05, - "loss": 3.2647, + "learning_rate": 1.689782495053698e-05, + "loss": 3.2404, "step": 200100 }, { "epoch": 0.66, - "learning_rate": 1.6827393083793143e-05, - "loss": 3.274, + "learning_rate": 1.6881282134420308e-05, + "loss": 3.2395, "step": 200200 }, { "epoch": 0.66, - "learning_rate": 1.6810823350068767e-05, - "loss": 3.2763, + "learning_rate": 1.6864739318303636e-05, + "loss": 3.2354, "step": 200300 }, { "epoch": 0.66, - "learning_rate": 1.6794253616344386e-05, - "loss": 3.2745, + "learning_rate": 1.684819650218696e-05, + "loss": 3.2193, "step": 200400 }, { "epoch": 0.66, - "learning_rate": 1.6777683882620007e-05, - "loss": 3.2519, + "learning_rate": 1.6831653686070285e-05, + "loss": 3.2225, "step": 200500 }, { "epoch": 0.66, - "learning_rate": 1.676111414889563e-05, - "loss": 3.2775, + "learning_rate": 1.6815110869953616e-05, + "loss": 3.2327, "step": 200600 }, { - "epoch": 0.67, - "learning_rate": 1.674454441517125e-05, - "loss": 3.2831, + "epoch": 0.66, + "learning_rate": 1.679856805383694e-05, + "loss": 3.2374, "step": 200700 }, { - "epoch": 0.67, - "learning_rate": 1.672797468144687e-05, - "loss": 3.2673, + "epoch": 0.66, + "learning_rate": 1.678202523772027e-05, + "loss": 3.2281, "step": 200800 }, { - "epoch": 0.67, - "learning_rate": 1.6711404947722493e-05, - "loss": 3.2622, + "epoch": 0.66, + "learning_rate": 1.6765482421603597e-05, + "loss": 3.2368, "step": 200900 }, { "epoch": 0.67, - "learning_rate": 1.669483521399811e-05, - "loss": 3.2647, + "learning_rate": 1.674893960548692e-05, + "loss": 3.2386, "step": 201000 }, { "epoch": 0.67, - "learning_rate": 1.6678265480273732e-05, - "loss": 3.26, + "learning_rate": 1.673239678937025e-05, + "loss": 3.2207, "step": 201100 }, { "epoch": 0.67, - "learning_rate": 1.6661695746549354e-05, - "loss": 3.2538, + "learning_rate": 1.6715853973253577e-05, + "loss": 3.2254, "step": 201200 }, { "epoch": 0.67, - "learning_rate": 1.6645126012824975e-05, - "loss": 3.2711, + "learning_rate": 1.6699311157136902e-05, + "loss": 3.2274, "step": 201300 }, { "epoch": 0.67, - "learning_rate": 1.6628556279100597e-05, - "loss": 3.2769, + "learning_rate": 1.668276834102023e-05, + "loss": 3.2266, "step": 201400 }, { "epoch": 0.67, - "learning_rate": 1.6611986545376215e-05, - "loss": 3.272, + "learning_rate": 1.6666225524903558e-05, + "loss": 3.2328, "step": 201500 }, { "epoch": 0.67, - "learning_rate": 1.6595416811651836e-05, - "loss": 3.2759, + "learning_rate": 1.6649682708786882e-05, + "loss": 3.2137, "step": 201600 }, { "epoch": 0.67, - "learning_rate": 1.6578847077927458e-05, - "loss": 3.2674, + "learning_rate": 1.663313989267021e-05, + "loss": 3.237, "step": 201700 }, { "epoch": 0.67, - "learning_rate": 1.656227734420308e-05, - "loss": 3.2609, + "learning_rate": 1.6616597076553535e-05, + "loss": 3.2324, "step": 201800 }, { "epoch": 0.67, - "learning_rate": 1.65457076104787e-05, - "loss": 3.277, + "learning_rate": 1.6600054260436866e-05, + "loss": 3.2334, "step": 201900 }, { "epoch": 0.67, - "learning_rate": 1.6529137876754322e-05, - "loss": 3.2703, + "learning_rate": 1.658351144432019e-05, + "loss": 3.2387, "step": 202000 }, { "epoch": 0.67, - "learning_rate": 1.651256814302994e-05, - "loss": 3.2654, + "learning_rate": 1.6566968628203515e-05, + "loss": 3.2386, "step": 202100 }, { "epoch": 0.67, - "learning_rate": 1.6495998409305562e-05, - "loss": 3.2692, + "learning_rate": 1.6550425812086846e-05, + "loss": 3.2219, "step": 202200 }, { "epoch": 0.67, - "learning_rate": 1.6479428675581187e-05, - "loss": 3.2681, + "learning_rate": 1.653388299597017e-05, + "loss": 3.2428, "step": 202300 }, { "epoch": 0.67, - "learning_rate": 1.6462858941856805e-05, - "loss": 3.2711, + "learning_rate": 1.6517340179853495e-05, + "loss": 3.2337, "step": 202400 }, { "epoch": 0.67, - "learning_rate": 1.6446289208132426e-05, - "loss": 3.2859, + "learning_rate": 1.6500797363736827e-05, + "loss": 3.2371, "step": 202500 }, { "epoch": 0.67, - "learning_rate": 1.6429719474408048e-05, - "loss": 3.2762, + "learning_rate": 1.648425454762015e-05, + "loss": 3.2357, "step": 202600 }, { "epoch": 0.67, - "learning_rate": 1.6413149740683666e-05, - "loss": 3.2669, + "learning_rate": 1.6467711731503476e-05, + "loss": 3.2267, "step": 202700 }, { "epoch": 0.67, - "learning_rate": 1.639658000695929e-05, - "loss": 3.2609, + "learning_rate": 1.6451168915386807e-05, + "loss": 3.2392, "step": 202800 }, { "epoch": 0.67, - "learning_rate": 1.6380010273234912e-05, - "loss": 3.2642, + "learning_rate": 1.643462609927013e-05, + "loss": 3.2313, "step": 202900 }, { "epoch": 0.67, - "learning_rate": 1.636344053951053e-05, - "loss": 3.2758, + "learning_rate": 1.641808328315346e-05, + "loss": 3.237, "step": 203000 }, { "epoch": 0.67, - "learning_rate": 1.634687080578615e-05, - "loss": 3.2725, + "learning_rate": 1.6401540467036784e-05, + "loss": 3.2302, "step": 203100 }, { "epoch": 0.67, - "learning_rate": 1.6330301072061773e-05, - "loss": 3.255, + "learning_rate": 1.6384997650920112e-05, + "loss": 3.2379, "step": 203200 }, { "epoch": 0.67, - "learning_rate": 1.6313731338337395e-05, - "loss": 3.285, + "learning_rate": 1.636845483480344e-05, + "loss": 3.2171, "step": 203300 }, { "epoch": 0.67, - "learning_rate": 1.6297161604613016e-05, - "loss": 3.2618, + "learning_rate": 1.6351912018686764e-05, + "loss": 3.2237, "step": 203400 }, { "epoch": 0.67, - "learning_rate": 1.6280591870888637e-05, - "loss": 3.2692, + "learning_rate": 1.6335369202570092e-05, + "loss": 3.2424, "step": 203500 }, { "epoch": 0.67, - "learning_rate": 1.6264022137164256e-05, - "loss": 3.2772, + "learning_rate": 1.631882638645342e-05, + "loss": 3.2292, "step": 203600 }, { - "epoch": 0.68, - "learning_rate": 1.6247452403439877e-05, - "loss": 3.2657, + "epoch": 0.67, + "learning_rate": 1.6302283570336745e-05, + "loss": 3.2262, "step": 203700 }, { - "epoch": 0.68, - "learning_rate": 1.62308826697155e-05, - "loss": 3.2651, + "epoch": 0.67, + "learning_rate": 1.6285740754220073e-05, + "loss": 3.2236, "step": 203800 }, { - "epoch": 0.68, - "learning_rate": 1.621431293599112e-05, - "loss": 3.2729, + "epoch": 0.67, + "learning_rate": 1.62691979381034e-05, + "loss": 3.2241, "step": 203900 }, { - "epoch": 0.68, - "learning_rate": 1.619774320226674e-05, - "loss": 3.2661, + "epoch": 0.67, + "learning_rate": 1.6252655121986725e-05, + "loss": 3.2275, "step": 204000 }, { "epoch": 0.68, - "learning_rate": 1.618117346854236e-05, - "loss": 3.2808, + "learning_rate": 1.6236112305870056e-05, + "loss": 3.2283, "step": 204100 }, { "epoch": 0.68, - "learning_rate": 1.616460373481798e-05, - "loss": 3.2671, + "learning_rate": 1.621956948975338e-05, + "loss": 3.2188, "step": 204200 }, { "epoch": 0.68, - "learning_rate": 1.6148034001093602e-05, - "loss": 3.2564, + "learning_rate": 1.6203026673636706e-05, + "loss": 3.2077, "step": 204300 }, { "epoch": 0.68, - "learning_rate": 1.6131464267369224e-05, - "loss": 3.2659, + "learning_rate": 1.6186483857520033e-05, + "loss": 3.2344, "step": 204400 }, { "epoch": 0.68, - "learning_rate": 1.6114894533644845e-05, - "loss": 3.2725, + "learning_rate": 1.616994104140336e-05, + "loss": 3.2209, "step": 204500 }, { "epoch": 0.68, - "learning_rate": 1.6098324799920467e-05, - "loss": 3.2711, + "learning_rate": 1.6153398225286686e-05, + "loss": 3.2377, "step": 204600 }, { "epoch": 0.68, - "learning_rate": 1.6081755066196085e-05, - "loss": 3.2673, + "learning_rate": 1.6136855409170014e-05, + "loss": 3.2275, "step": 204700 }, { "epoch": 0.68, - "learning_rate": 1.6065185332471706e-05, - "loss": 3.2573, + "learning_rate": 1.6120312593053342e-05, + "loss": 3.2346, "step": 204800 }, { "epoch": 0.68, - "learning_rate": 1.604861559874733e-05, - "loss": 3.272, + "learning_rate": 1.6103769776936666e-05, + "loss": 3.2306, "step": 204900 }, { "epoch": 0.68, - "learning_rate": 1.603204586502295e-05, - "loss": 3.2781, + "learning_rate": 1.6087226960819994e-05, + "loss": 3.2046, "step": 205000 }, { "epoch": 0.68, - "learning_rate": 1.601547613129857e-05, - "loss": 3.267, + "learning_rate": 1.6070684144703322e-05, + "loss": 3.2228, "step": 205100 }, { "epoch": 0.68, - "learning_rate": 1.5998906397574192e-05, - "loss": 3.2583, + "learning_rate": 1.605414132858665e-05, + "loss": 3.2538, "step": 205200 }, { "epoch": 0.68, - "learning_rate": 1.598233666384981e-05, - "loss": 3.2733, + "learning_rate": 1.6037598512469975e-05, + "loss": 3.2279, "step": 205300 }, { "epoch": 0.68, - "learning_rate": 1.5965766930125435e-05, - "loss": 3.2601, + "learning_rate": 1.6021055696353303e-05, + "loss": 3.233, "step": 205400 }, { "epoch": 0.68, - "learning_rate": 1.5949197196401057e-05, - "loss": 3.2657, + "learning_rate": 1.600451288023663e-05, + "loss": 3.2272, "step": 205500 }, { "epoch": 0.68, - "learning_rate": 1.5932627462676675e-05, - "loss": 3.2769, + "learning_rate": 1.5987970064119955e-05, + "loss": 3.224, "step": 205600 }, { "epoch": 0.68, - "learning_rate": 1.5916057728952296e-05, - "loss": 3.265, + "learning_rate": 1.5971427248003283e-05, + "loss": 3.2302, "step": 205700 }, { "epoch": 0.68, - "learning_rate": 1.5899487995227918e-05, - "loss": 3.272, + "learning_rate": 1.595488443188661e-05, + "loss": 3.2254, "step": 205800 }, { "epoch": 0.68, - "learning_rate": 1.588291826150354e-05, - "loss": 3.278, + "learning_rate": 1.5938341615769935e-05, + "loss": 3.2294, "step": 205900 }, { "epoch": 0.68, - "learning_rate": 1.586634852777916e-05, - "loss": 3.2666, + "learning_rate": 1.5921798799653263e-05, + "loss": 3.2239, "step": 206000 }, { "epoch": 0.68, - "learning_rate": 1.5849778794054782e-05, - "loss": 3.2741, + "learning_rate": 1.590525598353659e-05, + "loss": 3.2282, "step": 206100 }, { "epoch": 0.68, - "learning_rate": 1.58332090603304e-05, - "loss": 3.26, + "learning_rate": 1.5888713167419916e-05, + "loss": 3.225, "step": 206200 }, { "epoch": 0.68, - "learning_rate": 1.581663932660602e-05, - "loss": 3.2676, + "learning_rate": 1.5872170351303244e-05, + "loss": 3.2174, "step": 206300 }, { "epoch": 0.68, - "learning_rate": 1.5800069592881643e-05, - "loss": 3.2819, + "learning_rate": 1.585562753518657e-05, + "loss": 3.218, "step": 206400 }, { "epoch": 0.68, - "learning_rate": 1.5783499859157265e-05, - "loss": 3.2651, + "learning_rate": 1.5839084719069896e-05, + "loss": 3.2394, "step": 206500 }, { "epoch": 0.68, - "learning_rate": 1.5766930125432886e-05, - "loss": 3.2679, + "learning_rate": 1.5822541902953224e-05, + "loss": 3.2201, "step": 206600 }, { "epoch": 0.68, - "learning_rate": 1.5750360391708504e-05, - "loss": 3.2687, + "learning_rate": 1.5805999086836552e-05, + "loss": 3.2181, "step": 206700 }, { - "epoch": 0.69, - "learning_rate": 1.5733790657984126e-05, - "loss": 3.2705, + "epoch": 0.68, + "learning_rate": 1.5789456270719877e-05, + "loss": 3.2116, "step": 206800 }, { - "epoch": 0.69, - "learning_rate": 1.5717220924259747e-05, - "loss": 3.2602, + "epoch": 0.68, + "learning_rate": 1.5772913454603204e-05, + "loss": 3.214, "step": 206900 }, { - "epoch": 0.69, - "learning_rate": 1.570065119053537e-05, - "loss": 3.28, + "epoch": 0.68, + "learning_rate": 1.5756370638486532e-05, + "loss": 3.2215, "step": 207000 }, { "epoch": 0.69, - "learning_rate": 1.568408145681099e-05, - "loss": 3.2859, + "learning_rate": 1.573982782236986e-05, + "loss": 3.2275, "step": 207100 }, { "epoch": 0.69, - "learning_rate": 1.566751172308661e-05, - "loss": 3.267, + "learning_rate": 1.5723285006253185e-05, + "loss": 3.2301, "step": 207200 }, { "epoch": 0.69, - "learning_rate": 1.565094198936223e-05, - "loss": 3.2582, + "learning_rate": 1.5706742190136513e-05, + "loss": 3.2208, "step": 207300 }, { "epoch": 0.69, - "learning_rate": 1.563437225563785e-05, - "loss": 3.2728, + "learning_rate": 1.569019937401984e-05, + "loss": 3.2345, "step": 207400 }, { "epoch": 0.69, - "learning_rate": 1.5617802521913476e-05, - "loss": 3.2592, + "learning_rate": 1.5673656557903165e-05, + "loss": 3.2377, "step": 207500 }, { "epoch": 0.69, - "learning_rate": 1.5601232788189094e-05, - "loss": 3.2644, + "learning_rate": 1.5657113741786493e-05, + "loss": 3.226, "step": 207600 }, { "epoch": 0.69, - "learning_rate": 1.5584663054464715e-05, - "loss": 3.2588, + "learning_rate": 1.564057092566982e-05, + "loss": 3.2392, "step": 207700 }, { "epoch": 0.69, - "learning_rate": 1.5568093320740337e-05, - "loss": 3.2614, + "learning_rate": 1.5624028109553146e-05, + "loss": 3.2296, "step": 207800 }, { "epoch": 0.69, - "learning_rate": 1.5551523587015955e-05, - "loss": 3.2743, + "learning_rate": 1.560748529343647e-05, + "loss": 3.2288, "step": 207900 }, { "epoch": 0.69, - "learning_rate": 1.553495385329158e-05, - "loss": 3.2708, + "learning_rate": 1.55909424773198e-05, + "loss": 3.228, "step": 208000 }, { "epoch": 0.69, - "learning_rate": 1.55183841195672e-05, - "loss": 3.2642, + "learning_rate": 1.5574399661203126e-05, + "loss": 3.2338, "step": 208100 }, { "epoch": 0.69, - "learning_rate": 1.550181438584282e-05, - "loss": 3.2638, + "learning_rate": 1.5557856845086454e-05, + "loss": 3.2245, "step": 208200 }, { "epoch": 0.69, - "learning_rate": 1.548524465211844e-05, - "loss": 3.2632, + "learning_rate": 1.5541314028969782e-05, + "loss": 3.2278, "step": 208300 }, { "epoch": 0.69, - "learning_rate": 1.5468674918394062e-05, - "loss": 3.2665, + "learning_rate": 1.5524771212853106e-05, + "loss": 3.228, "step": 208400 }, { "epoch": 0.69, - "learning_rate": 1.5452105184669684e-05, - "loss": 3.2573, + "learning_rate": 1.5508228396736434e-05, + "loss": 3.2347, "step": 208500 }, { "epoch": 0.69, - "learning_rate": 1.5435535450945305e-05, - "loss": 3.2811, + "learning_rate": 1.5491685580619762e-05, + "loss": 3.2341, "step": 208600 }, { "epoch": 0.69, - "learning_rate": 1.5418965717220927e-05, - "loss": 3.2649, + "learning_rate": 1.5475142764503087e-05, + "loss": 3.2229, "step": 208700 }, { "epoch": 0.69, - "learning_rate": 1.5402395983496545e-05, - "loss": 3.2668, + "learning_rate": 1.5458599948386415e-05, + "loss": 3.2169, "step": 208800 }, { "epoch": 0.69, - "learning_rate": 1.5385826249772166e-05, - "loss": 3.2554, + "learning_rate": 1.5442057132269743e-05, + "loss": 3.2493, "step": 208900 }, { "epoch": 0.69, - "learning_rate": 1.5369256516047788e-05, - "loss": 3.2817, + "learning_rate": 1.5425514316153067e-05, + "loss": 3.2421, "step": 209000 }, { "epoch": 0.69, - "learning_rate": 1.535268678232341e-05, - "loss": 3.2759, + "learning_rate": 1.5408971500036395e-05, + "loss": 3.2345, "step": 209100 }, { "epoch": 0.69, - "learning_rate": 1.533611704859903e-05, - "loss": 3.2723, + "learning_rate": 1.539242868391972e-05, + "loss": 3.2206, "step": 209200 }, { "epoch": 0.69, - "learning_rate": 1.531954731487465e-05, - "loss": 3.2701, + "learning_rate": 1.537588586780305e-05, + "loss": 3.2227, "step": 209300 }, { "epoch": 0.69, - "learning_rate": 1.530297758115027e-05, - "loss": 3.2646, + "learning_rate": 1.5359343051686375e-05, + "loss": 3.2274, "step": 209400 }, { "epoch": 0.69, - "learning_rate": 1.528640784742589e-05, - "loss": 3.2729, + "learning_rate": 1.53428002355697e-05, + "loss": 3.2289, "step": 209500 }, { "epoch": 0.69, - "learning_rate": 1.5269838113701513e-05, - "loss": 3.2627, + "learning_rate": 1.532625741945303e-05, + "loss": 3.223, "step": 209600 }, { "epoch": 0.69, - "learning_rate": 1.5253268379977135e-05, - "loss": 3.2799, + "learning_rate": 1.5309714603336356e-05, + "loss": 3.211, "step": 209700 }, { - "epoch": 0.7, - "learning_rate": 1.5236698646252756e-05, - "loss": 3.2756, + "epoch": 0.69, + "learning_rate": 1.529317178721968e-05, + "loss": 3.2255, "step": 209800 }, { - "epoch": 0.7, - "learning_rate": 1.5220128912528376e-05, - "loss": 3.276, + "epoch": 0.69, + "learning_rate": 1.527662897110301e-05, + "loss": 3.2144, "step": 209900 }, { - "epoch": 0.7, - "learning_rate": 1.5203559178803997e-05, - "loss": 3.2698, + "epoch": 0.69, + "learning_rate": 1.5260086154986336e-05, + "loss": 3.2107, "step": 210000 }, { "epoch": 0.7, - "learning_rate": 1.5186989445079619e-05, - "loss": 3.2557, + "learning_rate": 1.5243543338869662e-05, + "loss": 3.2262, "step": 210100 }, { "epoch": 0.7, - "learning_rate": 1.5170419711355239e-05, - "loss": 3.2655, + "learning_rate": 1.522700052275299e-05, + "loss": 3.2254, "step": 210200 }, { "epoch": 0.7, - "learning_rate": 1.515384997763086e-05, - "loss": 3.2645, + "learning_rate": 1.5210457706636317e-05, + "loss": 3.2262, "step": 210300 }, { "epoch": 0.7, - "learning_rate": 1.5137280243906481e-05, - "loss": 3.2626, + "learning_rate": 1.5193914890519644e-05, + "loss": 3.213, "step": 210400 }, { "epoch": 0.7, - "learning_rate": 1.5120710510182101e-05, - "loss": 3.2686, + "learning_rate": 1.517737207440297e-05, + "loss": 3.2287, "step": 210500 }, { "epoch": 0.7, - "learning_rate": 1.5104140776457723e-05, - "loss": 3.2641, + "learning_rate": 1.5160829258286297e-05, + "loss": 3.2259, "step": 210600 }, { "epoch": 0.7, - "learning_rate": 1.5087571042733344e-05, - "loss": 3.2622, + "learning_rate": 1.5144286442169625e-05, + "loss": 3.2214, "step": 210700 }, { "epoch": 0.7, - "learning_rate": 1.5071001309008964e-05, - "loss": 3.267, + "learning_rate": 1.5127743626052951e-05, + "loss": 3.2184, "step": 210800 }, { "epoch": 0.7, - "learning_rate": 1.5054431575284585e-05, - "loss": 3.2524, + "learning_rate": 1.5111200809936277e-05, + "loss": 3.2174, "step": 210900 }, { "epoch": 0.7, - "learning_rate": 1.5037861841560209e-05, - "loss": 3.2742, + "learning_rate": 1.5094657993819605e-05, + "loss": 3.2122, "step": 211000 }, { "epoch": 0.7, - "learning_rate": 1.5021292107835827e-05, - "loss": 3.2675, + "learning_rate": 1.5078115177702931e-05, + "loss": 3.2344, "step": 211100 }, { "epoch": 0.7, - "learning_rate": 1.500472237411145e-05, - "loss": 3.266, + "learning_rate": 1.5061572361586258e-05, + "loss": 3.2194, "step": 211200 }, { "epoch": 0.7, - "learning_rate": 1.4988152640387071e-05, - "loss": 3.2709, + "learning_rate": 1.5045029545469586e-05, + "loss": 3.2268, "step": 211300 }, { "epoch": 0.7, - "learning_rate": 1.497158290666269e-05, - "loss": 3.2726, + "learning_rate": 1.5028486729352912e-05, + "loss": 3.2215, "step": 211400 }, { "epoch": 0.7, - "learning_rate": 1.4955013172938312e-05, - "loss": 3.2645, + "learning_rate": 1.501194391323624e-05, + "loss": 3.2137, "step": 211500 }, { "epoch": 0.7, - "learning_rate": 1.4938443439213934e-05, - "loss": 3.2738, + "learning_rate": 1.4995401097119566e-05, + "loss": 3.2224, "step": 211600 }, { "epoch": 0.7, - "learning_rate": 1.4921873705489554e-05, - "loss": 3.2913, + "learning_rate": 1.4978858281002892e-05, + "loss": 3.2359, "step": 211700 }, { "epoch": 0.7, - "learning_rate": 1.4905303971765175e-05, - "loss": 3.252, + "learning_rate": 1.496231546488622e-05, + "loss": 3.2345, "step": 211800 }, { "epoch": 0.7, - "learning_rate": 1.4888734238040793e-05, - "loss": 3.2723, + "learning_rate": 1.4945772648769546e-05, + "loss": 3.2309, "step": 211900 }, { "epoch": 0.7, - "learning_rate": 1.4872164504316416e-05, - "loss": 3.2825, + "learning_rate": 1.4929229832652873e-05, + "loss": 3.2221, "step": 212000 }, { "epoch": 0.7, - "learning_rate": 1.4855594770592038e-05, - "loss": 3.2527, + "learning_rate": 1.49126870165362e-05, + "loss": 3.2354, "step": 212100 }, { "epoch": 0.7, - "learning_rate": 1.4839025036867658e-05, - "loss": 3.2679, + "learning_rate": 1.4896144200419527e-05, + "loss": 3.239, "step": 212200 }, { "epoch": 0.7, - "learning_rate": 1.4822455303143279e-05, - "loss": 3.2703, + "learning_rate": 1.4879601384302851e-05, + "loss": 3.2341, "step": 212300 }, { "epoch": 0.7, - "learning_rate": 1.48058855694189e-05, - "loss": 3.2642, + "learning_rate": 1.4863058568186181e-05, + "loss": 3.2317, "step": 212400 }, { "epoch": 0.7, - "learning_rate": 1.478931583569452e-05, - "loss": 3.2626, + "learning_rate": 1.4846515752069507e-05, + "loss": 3.2175, "step": 212500 }, { "epoch": 0.7, - "learning_rate": 1.4772746101970142e-05, - "loss": 3.2632, + "learning_rate": 1.4829972935952835e-05, + "loss": 3.2318, "step": 212600 }, { "epoch": 0.7, - "learning_rate": 1.4756176368245763e-05, - "loss": 3.2841, + "learning_rate": 1.4813430119836161e-05, + "loss": 3.2361, "step": 212700 }, { - "epoch": 0.71, - "learning_rate": 1.4739606634521383e-05, - "loss": 3.2687, + "epoch": 0.7, + "learning_rate": 1.4796887303719486e-05, + "loss": 3.2198, "step": 212800 }, { - "epoch": 0.71, - "learning_rate": 1.4723036900797005e-05, - "loss": 3.2673, + "epoch": 0.7, + "learning_rate": 1.4780344487602815e-05, + "loss": 3.2249, "step": 212900 }, { - "epoch": 0.71, - "learning_rate": 1.4706467167072626e-05, - "loss": 3.2596, + "epoch": 0.7, + "learning_rate": 1.4763801671486142e-05, + "loss": 3.2325, "step": 213000 }, { "epoch": 0.71, - "learning_rate": 1.4689897433348246e-05, - "loss": 3.2696, + "learning_rate": 1.4747258855369466e-05, + "loss": 3.2207, "step": 213100 }, { "epoch": 0.71, - "learning_rate": 1.4673327699623867e-05, - "loss": 3.2732, + "learning_rate": 1.4730716039252796e-05, + "loss": 3.216, "step": 213200 }, { "epoch": 0.71, - "learning_rate": 1.465675796589949e-05, - "loss": 3.2571, + "learning_rate": 1.4714173223136122e-05, + "loss": 3.2278, "step": 213300 }, { "epoch": 0.71, - "learning_rate": 1.4640188232175109e-05, - "loss": 3.2726, + "learning_rate": 1.469763040701945e-05, + "loss": 3.2046, "step": 213400 }, { "epoch": 0.71, - "learning_rate": 1.462361849845073e-05, - "loss": 3.2675, + "learning_rate": 1.4681087590902776e-05, + "loss": 3.2214, "step": 213500 }, { "epoch": 0.71, - "learning_rate": 1.4607048764726353e-05, - "loss": 3.2576, + "learning_rate": 1.46645447747861e-05, + "loss": 3.2349, "step": 213600 }, { "epoch": 0.71, - "learning_rate": 1.4590479031001971e-05, - "loss": 3.2547, + "learning_rate": 1.464800195866943e-05, + "loss": 3.2295, "step": 213700 }, { "epoch": 0.71, - "learning_rate": 1.4573909297277594e-05, - "loss": 3.282, + "learning_rate": 1.4631459142552757e-05, + "loss": 3.2238, "step": 213800 }, { "epoch": 0.71, - "learning_rate": 1.4557339563553216e-05, - "loss": 3.2658, + "learning_rate": 1.4614916326436081e-05, + "loss": 3.2328, "step": 213900 }, { "epoch": 0.71, - "learning_rate": 1.4540769829828834e-05, - "loss": 3.2729, + "learning_rate": 1.459837351031941e-05, + "loss": 3.2284, "step": 214000 }, { "epoch": 0.71, - "learning_rate": 1.4524200096104457e-05, - "loss": 3.2687, + "learning_rate": 1.4581830694202735e-05, + "loss": 3.2291, "step": 214100 }, { "epoch": 0.71, - "learning_rate": 1.4507630362380079e-05, - "loss": 3.2659, + "learning_rate": 1.4565287878086061e-05, + "loss": 3.2241, "step": 214200 }, { "epoch": 0.71, - "learning_rate": 1.4491060628655698e-05, - "loss": 3.2719, + "learning_rate": 1.4548745061969391e-05, + "loss": 3.2261, "step": 214300 }, { "epoch": 0.71, - "learning_rate": 1.447449089493132e-05, - "loss": 3.2633, + "learning_rate": 1.4532202245852716e-05, + "loss": 3.2266, "step": 214400 }, { "epoch": 0.71, - "learning_rate": 1.445792116120694e-05, - "loss": 3.2628, + "learning_rate": 1.4515659429736045e-05, + "loss": 3.2286, "step": 214500 }, { "epoch": 0.71, - "learning_rate": 1.4441351427482561e-05, - "loss": 3.2652, + "learning_rate": 1.4499116613619371e-05, + "loss": 3.2204, "step": 214600 }, { "epoch": 0.71, - "learning_rate": 1.4424781693758182e-05, - "loss": 3.2631, + "learning_rate": 1.4482573797502696e-05, + "loss": 3.2378, "step": 214700 }, { "epoch": 0.71, - "learning_rate": 1.4408211960033802e-05, - "loss": 3.2642, + "learning_rate": 1.4466030981386026e-05, + "loss": 3.2296, "step": 214800 }, { "epoch": 0.71, - "learning_rate": 1.4391642226309424e-05, - "loss": 3.267, + "learning_rate": 1.444948816526935e-05, + "loss": 3.2164, "step": 214900 }, { "epoch": 0.71, - "learning_rate": 1.4375072492585045e-05, - "loss": 3.2726, + "learning_rate": 1.4432945349152676e-05, + "loss": 3.2139, "step": 215000 }, { "epoch": 0.71, - "learning_rate": 1.4358502758860665e-05, - "loss": 3.2743, + "learning_rate": 1.4416402533036006e-05, + "loss": 3.2196, "step": 215100 }, { "epoch": 0.71, - "learning_rate": 1.4341933025136286e-05, - "loss": 3.2726, + "learning_rate": 1.439985971691933e-05, + "loss": 3.2271, "step": 215200 }, { "epoch": 0.71, - "learning_rate": 1.4325363291411908e-05, - "loss": 3.2818, + "learning_rate": 1.4383316900802657e-05, + "loss": 3.2253, "step": 215300 }, { "epoch": 0.71, - "learning_rate": 1.4308793557687528e-05, - "loss": 3.2623, + "learning_rate": 1.4366774084685985e-05, + "loss": 3.22, "step": 215400 }, { "epoch": 0.71, - "learning_rate": 1.4292223823963149e-05, - "loss": 3.2673, + "learning_rate": 1.4350231268569311e-05, + "loss": 3.2166, "step": 215500 }, { "epoch": 0.71, - "learning_rate": 1.427565409023877e-05, - "loss": 3.2711, + "learning_rate": 1.433368845245264e-05, + "loss": 3.2289, "step": 215600 }, { "epoch": 0.71, - "learning_rate": 1.425908435651439e-05, - "loss": 3.2587, + "learning_rate": 1.4317145636335965e-05, + "loss": 3.2332, "step": 215700 }, { - "epoch": 0.72, - "learning_rate": 1.4242514622790012e-05, - "loss": 3.2702, + "epoch": 0.71, + "learning_rate": 1.4300602820219291e-05, + "loss": 3.2239, "step": 215800 }, { - "epoch": 0.72, - "learning_rate": 1.4225944889065635e-05, - "loss": 3.274, + "epoch": 0.71, + "learning_rate": 1.4284060004102621e-05, + "loss": 3.2292, "step": 215900 }, { - "epoch": 0.72, - "learning_rate": 1.4209375155341253e-05, - "loss": 3.2676, + "epoch": 0.71, + "learning_rate": 1.4267517187985945e-05, + "loss": 3.2264, "step": 216000 }, { - "epoch": 0.72, - "learning_rate": 1.4192805421616876e-05, - "loss": 3.2663, + "epoch": 0.71, + "learning_rate": 1.4250974371869272e-05, + "loss": 3.2311, "step": 216100 }, { "epoch": 0.72, - "learning_rate": 1.4176235687892498e-05, - "loss": 3.2568, + "learning_rate": 1.42344315557526e-05, + "loss": 3.2259, "step": 216200 }, { "epoch": 0.72, - "learning_rate": 1.4159665954168116e-05, - "loss": 3.2576, + "learning_rate": 1.4217888739635926e-05, + "loss": 3.2308, "step": 216300 }, { "epoch": 0.72, - "learning_rate": 1.4143096220443739e-05, - "loss": 3.262, + "learning_rate": 1.4201345923519252e-05, + "loss": 3.2292, "step": 216400 }, { "epoch": 0.72, - "learning_rate": 1.412652648671936e-05, - "loss": 3.2699, + "learning_rate": 1.418480310740258e-05, + "loss": 3.2345, "step": 216500 }, { "epoch": 0.72, - "learning_rate": 1.410995675299498e-05, - "loss": 3.2637, + "learning_rate": 1.4168260291285906e-05, + "loss": 3.2297, "step": 216600 }, { "epoch": 0.72, - "learning_rate": 1.4093387019270602e-05, - "loss": 3.2598, + "learning_rate": 1.4151717475169234e-05, + "loss": 3.2273, "step": 216700 }, { "epoch": 0.72, - "learning_rate": 1.4076817285546223e-05, - "loss": 3.2566, + "learning_rate": 1.413517465905256e-05, + "loss": 3.2256, "step": 216800 }, { "epoch": 0.72, - "learning_rate": 1.4060247551821843e-05, - "loss": 3.2691, + "learning_rate": 1.4118631842935887e-05, + "loss": 3.2297, "step": 216900 }, { "epoch": 0.72, - "learning_rate": 1.4043677818097464e-05, - "loss": 3.2843, + "learning_rate": 1.4102089026819215e-05, + "loss": 3.214, "step": 217000 }, { "epoch": 0.72, - "learning_rate": 1.4027108084373084e-05, - "loss": 3.2771, + "learning_rate": 1.408554621070254e-05, + "loss": 3.2228, "step": 217100 }, { "epoch": 0.72, - "learning_rate": 1.4010538350648706e-05, - "loss": 3.2689, + "learning_rate": 1.4069003394585867e-05, + "loss": 3.2289, "step": 217200 }, { "epoch": 0.72, - "learning_rate": 1.3993968616924327e-05, - "loss": 3.2526, + "learning_rate": 1.4052460578469195e-05, + "loss": 3.2406, "step": 217300 }, { "epoch": 0.72, - "learning_rate": 1.3977398883199947e-05, - "loss": 3.26, + "learning_rate": 1.4035917762352521e-05, + "loss": 3.2446, "step": 217400 }, { "epoch": 0.72, - "learning_rate": 1.3960829149475568e-05, - "loss": 3.2756, + "learning_rate": 1.4019374946235847e-05, + "loss": 3.2331, "step": 217500 }, { "epoch": 0.72, - "learning_rate": 1.394425941575119e-05, - "loss": 3.2708, + "learning_rate": 1.4002832130119175e-05, + "loss": 3.2407, "step": 217600 }, { "epoch": 0.72, - "learning_rate": 1.392768968202681e-05, - "loss": 3.2598, + "learning_rate": 1.3986289314002502e-05, + "loss": 3.2233, "step": 217700 }, { "epoch": 0.72, - "learning_rate": 1.3911119948302431e-05, - "loss": 3.2672, + "learning_rate": 1.396974649788583e-05, + "loss": 3.2134, "step": 217800 }, { "epoch": 0.72, - "learning_rate": 1.3894550214578053e-05, - "loss": 3.2649, + "learning_rate": 1.3953203681769156e-05, + "loss": 3.2192, "step": 217900 }, { "epoch": 0.72, - "learning_rate": 1.3877980480853672e-05, - "loss": 3.268, + "learning_rate": 1.3936660865652482e-05, + "loss": 3.2357, "step": 218000 }, { "epoch": 0.72, - "learning_rate": 1.3861410747129294e-05, - "loss": 3.2598, + "learning_rate": 1.392011804953581e-05, + "loss": 3.2293, "step": 218100 }, { "epoch": 0.72, - "learning_rate": 1.3844841013404917e-05, - "loss": 3.258, + "learning_rate": 1.3903575233419136e-05, + "loss": 3.2336, "step": 218200 }, { "epoch": 0.72, - "learning_rate": 1.3828271279680535e-05, - "loss": 3.2624, + "learning_rate": 1.3887032417302462e-05, + "loss": 3.2268, "step": 218300 }, { "epoch": 0.72, - "learning_rate": 1.3811701545956156e-05, - "loss": 3.2686, + "learning_rate": 1.387048960118579e-05, + "loss": 3.2261, "step": 218400 }, { "epoch": 0.72, - "learning_rate": 1.379513181223178e-05, - "loss": 3.2719, + "learning_rate": 1.3853946785069116e-05, + "loss": 3.2152, "step": 218500 }, { "epoch": 0.72, - "learning_rate": 1.3778562078507398e-05, - "loss": 3.2599, + "learning_rate": 1.3837403968952443e-05, + "loss": 3.2371, "step": 218600 }, { "epoch": 0.72, - "learning_rate": 1.3761992344783021e-05, - "loss": 3.2622, + "learning_rate": 1.382086115283577e-05, + "loss": 3.2242, "step": 218700 }, { - "epoch": 0.73, - "learning_rate": 1.3745422611058642e-05, - "loss": 3.2629, + "epoch": 0.72, + "learning_rate": 1.3804318336719097e-05, + "loss": 3.2144, "step": 218800 }, { - "epoch": 0.73, - "learning_rate": 1.372885287733426e-05, - "loss": 3.266, + "epoch": 0.72, + "learning_rate": 1.3787775520602425e-05, + "loss": 3.2265, "step": 218900 }, { - "epoch": 0.73, - "learning_rate": 1.3712283143609884e-05, - "loss": 3.2661, + "epoch": 0.72, + "learning_rate": 1.3771232704485751e-05, + "loss": 3.2288, "step": 219000 }, { - "epoch": 0.73, - "learning_rate": 1.3695713409885505e-05, - "loss": 3.2568, + "epoch": 0.72, + "learning_rate": 1.3754689888369077e-05, + "loss": 3.2301, "step": 219100 }, { "epoch": 0.73, - "learning_rate": 1.3679143676161125e-05, - "loss": 3.2693, + "learning_rate": 1.3738147072252405e-05, + "loss": 3.2292, "step": 219200 }, { "epoch": 0.73, - "learning_rate": 1.3662573942436746e-05, - "loss": 3.2605, + "learning_rate": 1.3721604256135731e-05, + "loss": 3.2243, "step": 219300 }, { "epoch": 0.73, - "learning_rate": 1.3646004208712368e-05, - "loss": 3.2639, + "learning_rate": 1.3705061440019058e-05, + "loss": 3.2145, "step": 219400 }, { "epoch": 0.73, - "learning_rate": 1.3629434474987988e-05, - "loss": 3.2749, + "learning_rate": 1.3688518623902385e-05, + "loss": 3.2151, "step": 219500 }, { "epoch": 0.73, - "learning_rate": 1.3612864741263609e-05, - "loss": 3.2569, + "learning_rate": 1.3671975807785712e-05, + "loss": 3.2129, "step": 219600 }, { "epoch": 0.73, - "learning_rate": 1.3596295007539229e-05, - "loss": 3.2594, + "learning_rate": 1.3655432991669038e-05, + "loss": 3.2263, "step": 219700 }, { "epoch": 0.73, - "learning_rate": 1.357972527381485e-05, - "loss": 3.2759, + "learning_rate": 1.3638890175552366e-05, + "loss": 3.2195, "step": 219800 }, { "epoch": 0.73, - "learning_rate": 1.3563155540090472e-05, - "loss": 3.2674, + "learning_rate": 1.3622347359435692e-05, + "loss": 3.2189, "step": 219900 }, { "epoch": 0.73, - "learning_rate": 1.3546585806366091e-05, - "loss": 3.2588, + "learning_rate": 1.360580454331902e-05, + "loss": 3.2307, "step": 220000 }, { "epoch": 0.73, - "learning_rate": 1.3530016072641713e-05, - "loss": 3.274, + "learning_rate": 1.3589261727202346e-05, + "loss": 3.2292, "step": 220100 }, { "epoch": 0.73, - "learning_rate": 1.3513446338917334e-05, - "loss": 3.2662, + "learning_rate": 1.3572718911085672e-05, + "loss": 3.2097, "step": 220200 }, { "epoch": 0.73, - "learning_rate": 1.3496876605192954e-05, - "loss": 3.2666, + "learning_rate": 1.3556176094969e-05, + "loss": 3.2338, "step": 220300 }, { "epoch": 0.73, - "learning_rate": 1.3480306871468576e-05, - "loss": 3.2626, + "learning_rate": 1.3539633278852327e-05, + "loss": 3.2049, "step": 220400 }, { "epoch": 0.73, - "learning_rate": 1.3463737137744197e-05, - "loss": 3.2532, + "learning_rate": 1.3523090462735651e-05, + "loss": 3.2335, "step": 220500 }, { "epoch": 0.73, - "learning_rate": 1.3447167404019817e-05, - "loss": 3.262, + "learning_rate": 1.350654764661898e-05, + "loss": 3.2323, "step": 220600 }, { "epoch": 0.73, - "learning_rate": 1.3430597670295438e-05, - "loss": 3.2657, + "learning_rate": 1.3490004830502307e-05, + "loss": 3.2232, "step": 220700 }, { "epoch": 0.73, - "learning_rate": 1.3414027936571061e-05, - "loss": 3.2636, + "learning_rate": 1.3473462014385632e-05, + "loss": 3.2379, "step": 220800 }, { "epoch": 0.73, - "learning_rate": 1.339745820284668e-05, - "loss": 3.2704, + "learning_rate": 1.3456919198268961e-05, + "loss": 3.224, "step": 220900 }, { "epoch": 0.73, - "learning_rate": 1.3380888469122301e-05, - "loss": 3.254, + "learning_rate": 1.3440376382152287e-05, + "loss": 3.2179, "step": 221000 }, { "epoch": 0.73, - "learning_rate": 1.3364318735397924e-05, - "loss": 3.266, + "learning_rate": 1.3423833566035615e-05, + "loss": 3.2167, "step": 221100 }, { "epoch": 0.73, - "learning_rate": 1.3347749001673542e-05, - "loss": 3.2574, + "learning_rate": 1.3407290749918942e-05, + "loss": 3.2261, "step": 221200 }, { "epoch": 0.73, - "learning_rate": 1.3331179267949165e-05, - "loss": 3.2608, + "learning_rate": 1.3390747933802266e-05, + "loss": 3.2298, "step": 221300 }, { "epoch": 0.73, - "learning_rate": 1.3314609534224787e-05, - "loss": 3.2712, + "learning_rate": 1.3374205117685596e-05, + "loss": 3.2209, "step": 221400 }, { "epoch": 0.73, - "learning_rate": 1.3298039800500407e-05, - "loss": 3.266, + "learning_rate": 1.3357662301568922e-05, + "loss": 3.23, "step": 221500 }, { "epoch": 0.73, - "learning_rate": 1.3281470066776028e-05, - "loss": 3.267, + "learning_rate": 1.3341119485452246e-05, + "loss": 3.2163, "step": 221600 }, { "epoch": 0.73, - "learning_rate": 1.326490033305165e-05, - "loss": 3.262, + "learning_rate": 1.3324576669335576e-05, + "loss": 3.2371, "step": 221700 }, { - "epoch": 0.74, - "learning_rate": 1.324833059932727e-05, - "loss": 3.2646, + "epoch": 0.73, + "learning_rate": 1.33080338532189e-05, + "loss": 3.2248, "step": 221800 }, { - "epoch": 0.74, - "learning_rate": 1.3231760865602891e-05, - "loss": 3.2729, + "epoch": 0.73, + "learning_rate": 1.329149103710223e-05, + "loss": 3.2067, "step": 221900 }, { - "epoch": 0.74, - "learning_rate": 1.3215191131878512e-05, - "loss": 3.2749, + "epoch": 0.73, + "learning_rate": 1.3274948220985556e-05, + "loss": 3.2273, "step": 222000 }, { - "epoch": 0.74, - "learning_rate": 1.3198621398154132e-05, - "loss": 3.2606, + "epoch": 0.73, + "learning_rate": 1.3258405404868881e-05, + "loss": 3.2296, "step": 222100 }, { "epoch": 0.74, - "learning_rate": 1.3182051664429754e-05, - "loss": 3.2684, + "learning_rate": 1.324186258875221e-05, + "loss": 3.2314, "step": 222200 }, { "epoch": 0.74, - "learning_rate": 1.3165481930705373e-05, - "loss": 3.2667, + "learning_rate": 1.3225319772635537e-05, + "loss": 3.2147, "step": 222300 }, { "epoch": 0.74, - "learning_rate": 1.3148912196980995e-05, - "loss": 3.2734, + "learning_rate": 1.3208776956518861e-05, + "loss": 3.2219, "step": 222400 }, { "epoch": 0.74, - "learning_rate": 1.3132342463256616e-05, - "loss": 3.2592, + "learning_rate": 1.3192234140402191e-05, + "loss": 3.2168, "step": 222500 }, { "epoch": 0.74, - "learning_rate": 1.3115772729532236e-05, - "loss": 3.2665, + "learning_rate": 1.3175691324285516e-05, + "loss": 3.2188, "step": 222600 }, { "epoch": 0.74, - "learning_rate": 1.3099202995807858e-05, - "loss": 3.2707, + "learning_rate": 1.3159148508168842e-05, + "loss": 3.213, "step": 222700 }, { "epoch": 0.74, - "learning_rate": 1.3082633262083479e-05, - "loss": 3.2675, + "learning_rate": 1.3142605692052171e-05, + "loss": 3.2197, "step": 222800 }, { "epoch": 0.74, - "learning_rate": 1.3066063528359099e-05, - "loss": 3.2628, + "learning_rate": 1.3126062875935496e-05, + "loss": 3.2143, "step": 222900 }, { "epoch": 0.74, - "learning_rate": 1.304949379463472e-05, - "loss": 3.2799, + "learning_rate": 1.3109520059818826e-05, + "loss": 3.215, "step": 223000 }, { "epoch": 0.74, - "learning_rate": 1.3032924060910343e-05, - "loss": 3.269, + "learning_rate": 1.309297724370215e-05, + "loss": 3.2176, "step": 223100 }, { "epoch": 0.74, - "learning_rate": 1.3016354327185961e-05, - "loss": 3.2689, + "learning_rate": 1.3076434427585476e-05, + "loss": 3.2314, "step": 223200 }, { "epoch": 0.74, - "learning_rate": 1.2999784593461583e-05, - "loss": 3.2529, + "learning_rate": 1.3059891611468806e-05, + "loss": 3.233, "step": 223300 }, { "epoch": 0.74, - "learning_rate": 1.2983214859737206e-05, - "loss": 3.2642, + "learning_rate": 1.304334879535213e-05, + "loss": 3.2213, "step": 223400 }, { "epoch": 0.74, - "learning_rate": 1.2966645126012824e-05, - "loss": 3.2623, + "learning_rate": 1.3026805979235457e-05, + "loss": 3.2222, "step": 223500 }, { "epoch": 0.74, - "learning_rate": 1.2950075392288447e-05, - "loss": 3.279, + "learning_rate": 1.3010263163118786e-05, + "loss": 3.2205, "step": 223600 }, { "epoch": 0.74, - "learning_rate": 1.2933505658564069e-05, - "loss": 3.2653, + "learning_rate": 1.299372034700211e-05, + "loss": 3.22, "step": 223700 }, { "epoch": 0.74, - "learning_rate": 1.2916935924839687e-05, - "loss": 3.2653, + "learning_rate": 1.2977177530885437e-05, + "loss": 3.2314, "step": 223800 }, { "epoch": 0.74, - "learning_rate": 1.290036619111531e-05, - "loss": 3.266, + "learning_rate": 1.2960634714768765e-05, + "loss": 3.2242, "step": 223900 }, { "epoch": 0.74, - "learning_rate": 1.2883796457390931e-05, - "loss": 3.2617, + "learning_rate": 1.2944091898652091e-05, + "loss": 3.2299, "step": 224000 }, { "epoch": 0.74, - "learning_rate": 1.2867226723666551e-05, - "loss": 3.2672, + "learning_rate": 1.292754908253542e-05, + "loss": 3.2339, "step": 224100 }, { "epoch": 0.74, - "learning_rate": 1.2850656989942173e-05, - "loss": 3.267, + "learning_rate": 1.2911006266418745e-05, + "loss": 3.23, "step": 224200 }, { "epoch": 0.74, - "learning_rate": 1.2834087256217794e-05, - "loss": 3.2694, + "learning_rate": 1.2894463450302072e-05, + "loss": 3.2215, "step": 224300 }, { "epoch": 0.74, - "learning_rate": 1.2817517522493414e-05, - "loss": 3.263, + "learning_rate": 1.28779206341854e-05, + "loss": 3.2294, "step": 224400 }, { "epoch": 0.74, - "learning_rate": 1.2800947788769035e-05, - "loss": 3.2549, + "learning_rate": 1.2861377818068726e-05, + "loss": 3.2321, "step": 224500 }, { "epoch": 0.74, - "learning_rate": 1.2784378055044657e-05, - "loss": 3.2633, + "learning_rate": 1.2844835001952052e-05, + "loss": 3.2227, "step": 224600 }, { "epoch": 0.74, - "learning_rate": 1.2767808321320277e-05, - "loss": 3.2643, + "learning_rate": 1.282829218583538e-05, + "loss": 3.2287, "step": 224700 }, { "epoch": 0.74, - "learning_rate": 1.2751238587595898e-05, - "loss": 3.2598, + "learning_rate": 1.2811749369718706e-05, + "loss": 3.231, "step": 224800 }, { - "epoch": 0.75, - "learning_rate": 1.2734668853871518e-05, - "loss": 3.2547, + "epoch": 0.74, + "learning_rate": 1.2795206553602032e-05, + "loss": 3.2334, "step": 224900 }, { - "epoch": 0.75, - "learning_rate": 1.271809912014714e-05, - "loss": 3.2544, + "epoch": 0.74, + "learning_rate": 1.277866373748536e-05, + "loss": 3.2392, "step": 225000 }, { - "epoch": 0.75, - "learning_rate": 1.2701529386422761e-05, - "loss": 3.2636, + "epoch": 0.74, + "learning_rate": 1.2762120921368686e-05, + "loss": 3.2337, "step": 225100 }, { "epoch": 0.75, - "learning_rate": 1.268495965269838e-05, - "loss": 3.265, + "learning_rate": 1.2745578105252014e-05, + "loss": 3.2214, "step": 225200 }, { "epoch": 0.75, - "learning_rate": 1.2668389918974002e-05, - "loss": 3.2606, + "learning_rate": 1.272903528913534e-05, + "loss": 3.2287, "step": 225300 }, { "epoch": 0.75, - "learning_rate": 1.2651820185249624e-05, - "loss": 3.2538, + "learning_rate": 1.2712492473018667e-05, + "loss": 3.2202, "step": 225400 }, { "epoch": 0.75, - "learning_rate": 1.2635250451525243e-05, - "loss": 3.2654, + "learning_rate": 1.2695949656901995e-05, + "loss": 3.2234, "step": 225500 }, { "epoch": 0.75, - "learning_rate": 1.2618680717800865e-05, - "loss": 3.2661, + "learning_rate": 1.2679406840785321e-05, + "loss": 3.236, "step": 225600 }, { "epoch": 0.75, - "learning_rate": 1.2602110984076488e-05, - "loss": 3.2572, + "learning_rate": 1.2662864024668647e-05, + "loss": 3.2189, "step": 225700 }, { "epoch": 0.75, - "learning_rate": 1.2585541250352106e-05, - "loss": 3.2721, + "learning_rate": 1.2646321208551975e-05, + "loss": 3.2014, "step": 225800 }, { "epoch": 0.75, - "learning_rate": 1.2568971516627728e-05, - "loss": 3.257, + "learning_rate": 1.2629778392435301e-05, + "loss": 3.2467, "step": 225900 }, { "epoch": 0.75, - "learning_rate": 1.255240178290335e-05, - "loss": 3.2626, + "learning_rate": 1.2613235576318628e-05, + "loss": 3.231, "step": 226000 }, { "epoch": 0.75, - "learning_rate": 1.2535832049178969e-05, - "loss": 3.2656, + "learning_rate": 1.2596692760201956e-05, + "loss": 3.2339, "step": 226100 }, { "epoch": 0.75, - "learning_rate": 1.2519262315454592e-05, - "loss": 3.2557, + "learning_rate": 1.2580149944085282e-05, + "loss": 3.2132, "step": 226200 }, { "epoch": 0.75, - "learning_rate": 1.2502692581730213e-05, - "loss": 3.2552, + "learning_rate": 1.256360712796861e-05, + "loss": 3.2232, "step": 226300 }, { "epoch": 0.75, - "learning_rate": 1.2486122848005833e-05, - "loss": 3.2574, + "learning_rate": 1.2547064311851936e-05, + "loss": 3.2083, "step": 226400 }, { "epoch": 0.75, - "learning_rate": 1.2469553114281455e-05, - "loss": 3.2704, + "learning_rate": 1.2530521495735262e-05, + "loss": 3.2257, "step": 226500 }, { "epoch": 0.75, - "learning_rate": 1.2452983380557074e-05, - "loss": 3.2673, + "learning_rate": 1.251397867961859e-05, + "loss": 3.2228, "step": 226600 }, { "epoch": 0.75, - "learning_rate": 1.2436413646832696e-05, - "loss": 3.2593, + "learning_rate": 1.2497435863501916e-05, + "loss": 3.216, "step": 226700 }, { "epoch": 0.75, - "learning_rate": 1.2419843913108317e-05, - "loss": 3.2577, + "learning_rate": 1.2480893047385244e-05, + "loss": 3.229, "step": 226800 }, { "epoch": 0.75, - "learning_rate": 1.2403274179383937e-05, - "loss": 3.2597, + "learning_rate": 1.246435023126857e-05, + "loss": 3.2197, "step": 226900 }, { "epoch": 0.75, - "learning_rate": 1.2386704445659559e-05, - "loss": 3.2691, + "learning_rate": 1.2447807415151897e-05, + "loss": 3.2022, "step": 227000 }, { "epoch": 0.75, - "learning_rate": 1.237013471193518e-05, - "loss": 3.2644, + "learning_rate": 1.2431264599035223e-05, + "loss": 3.2261, "step": 227100 }, { "epoch": 0.75, - "learning_rate": 1.2353564978210802e-05, - "loss": 3.2677, + "learning_rate": 1.241472178291855e-05, + "loss": 3.2306, "step": 227200 }, { "epoch": 0.75, - "learning_rate": 1.2336995244486421e-05, - "loss": 3.2728, + "learning_rate": 1.2398178966801877e-05, + "loss": 3.2291, "step": 227300 }, { "epoch": 0.75, - "learning_rate": 1.2320425510762043e-05, - "loss": 3.2588, + "learning_rate": 1.2381636150685203e-05, + "loss": 3.2243, "step": 227400 }, { "epoch": 0.75, - "learning_rate": 1.2303855777037664e-05, - "loss": 3.2624, + "learning_rate": 1.2365093334568531e-05, + "loss": 3.2242, "step": 227500 }, { "epoch": 0.75, - "learning_rate": 1.2287286043313284e-05, - "loss": 3.2602, + "learning_rate": 1.2348550518451857e-05, + "loss": 3.2226, "step": 227600 }, { "epoch": 0.75, - "learning_rate": 1.2270716309588905e-05, - "loss": 3.2576, + "learning_rate": 1.2332007702335184e-05, + "loss": 3.2229, "step": 227700 }, { "epoch": 0.75, - "learning_rate": 1.2254146575864527e-05, - "loss": 3.2722, + "learning_rate": 1.2315464886218512e-05, + "loss": 3.2123, "step": 227800 }, { - "epoch": 0.76, - "learning_rate": 1.2237576842140147e-05, - "loss": 3.2601, + "epoch": 0.75, + "learning_rate": 1.2298922070101838e-05, + "loss": 3.2171, "step": 227900 }, { - "epoch": 0.76, - "learning_rate": 1.222100710841577e-05, - "loss": 3.2778, + "epoch": 0.75, + "learning_rate": 1.2282379253985166e-05, + "loss": 3.2277, "step": 228000 }, { - "epoch": 0.76, - "learning_rate": 1.220443737469139e-05, - "loss": 3.2706, + "epoch": 0.75, + "learning_rate": 1.2265836437868492e-05, + "loss": 3.2214, "step": 228100 }, { "epoch": 0.76, - "learning_rate": 1.218786764096701e-05, - "loss": 3.263, + "learning_rate": 1.2249293621751818e-05, + "loss": 3.2311, "step": 228200 }, { "epoch": 0.76, - "learning_rate": 1.2171297907242631e-05, - "loss": 3.2633, + "learning_rate": 1.2232750805635146e-05, + "loss": 3.2288, "step": 228300 }, { "epoch": 0.76, - "learning_rate": 1.2154728173518252e-05, - "loss": 3.2645, + "learning_rate": 1.2216207989518472e-05, + "loss": 3.2256, "step": 228400 }, { "epoch": 0.76, - "learning_rate": 1.2138158439793874e-05, - "loss": 3.2625, + "learning_rate": 1.2199665173401799e-05, + "loss": 3.2279, "step": 228500 }, { "epoch": 0.76, - "learning_rate": 1.2121588706069494e-05, - "loss": 3.2656, + "learning_rate": 1.2183122357285127e-05, + "loss": 3.2084, "step": 228600 }, { "epoch": 0.76, - "learning_rate": 1.2105018972345115e-05, - "loss": 3.2531, + "learning_rate": 1.2166579541168453e-05, + "loss": 3.2236, "step": 228700 }, { "epoch": 0.76, - "learning_rate": 1.2088449238620737e-05, - "loss": 3.2526, + "learning_rate": 1.2150036725051779e-05, + "loss": 3.2173, "step": 228800 }, { "epoch": 0.76, - "learning_rate": 1.2071879504896356e-05, - "loss": 3.2629, + "learning_rate": 1.2133493908935107e-05, + "loss": 3.2402, "step": 228900 }, { "epoch": 0.76, - "learning_rate": 1.2055309771171978e-05, - "loss": 3.2644, + "learning_rate": 1.2116951092818433e-05, + "loss": 3.2264, "step": 229000 }, { "epoch": 0.76, - "learning_rate": 1.20387400374476e-05, - "loss": 3.2548, + "learning_rate": 1.2100408276701761e-05, + "loss": 3.2104, "step": 229100 }, { "epoch": 0.76, - "learning_rate": 1.2022170303723219e-05, - "loss": 3.2718, + "learning_rate": 1.2083865460585087e-05, + "loss": 3.2274, "step": 229200 }, { "epoch": 0.76, - "learning_rate": 1.2005600569998842e-05, - "loss": 3.2608, + "learning_rate": 1.2067322644468413e-05, + "loss": 3.224, "step": 229300 }, { "epoch": 0.76, - "learning_rate": 1.1989030836274462e-05, - "loss": 3.2688, + "learning_rate": 1.2050779828351741e-05, + "loss": 3.2091, "step": 229400 }, { "epoch": 0.76, - "learning_rate": 1.1972461102550082e-05, - "loss": 3.2618, + "learning_rate": 1.2034237012235066e-05, + "loss": 3.2274, "step": 229500 }, { "epoch": 0.76, - "learning_rate": 1.1955891368825703e-05, - "loss": 3.2599, + "learning_rate": 1.2017694196118394e-05, + "loss": 3.227, "step": 229600 }, { "epoch": 0.76, - "learning_rate": 1.1939321635101325e-05, - "loss": 3.2763, + "learning_rate": 1.2001151380001722e-05, + "loss": 3.2294, "step": 229700 }, { "epoch": 0.76, - "learning_rate": 1.1922751901376946e-05, - "loss": 3.2567, + "learning_rate": 1.1984608563885048e-05, + "loss": 3.228, "step": 229800 }, { "epoch": 0.76, - "learning_rate": 1.1906182167652566e-05, - "loss": 3.2587, + "learning_rate": 1.1968065747768374e-05, + "loss": 3.2269, "step": 229900 }, { "epoch": 0.76, - "learning_rate": 1.1889612433928187e-05, - "loss": 3.2432, + "learning_rate": 1.1951522931651702e-05, + "loss": 3.2256, "step": 230000 }, { "epoch": 0.76, - "learning_rate": 1.1873042700203809e-05, - "loss": 3.2744, + "learning_rate": 1.1934980115535028e-05, + "loss": 3.213, "step": 230100 }, { "epoch": 0.76, - "learning_rate": 1.1856472966479429e-05, - "loss": 3.2633, + "learning_rate": 1.1918437299418356e-05, + "loss": 3.2329, "step": 230200 }, { "epoch": 0.76, - "learning_rate": 1.183990323275505e-05, - "loss": 3.2665, + "learning_rate": 1.1901894483301681e-05, + "loss": 3.231, "step": 230300 }, { "epoch": 0.76, - "learning_rate": 1.1823333499030672e-05, - "loss": 3.2705, + "learning_rate": 1.1885351667185009e-05, + "loss": 3.2333, "step": 230400 }, { "epoch": 0.76, - "learning_rate": 1.1806763765306291e-05, - "loss": 3.2464, + "learning_rate": 1.1868808851068337e-05, + "loss": 3.217, "step": 230500 }, { "epoch": 0.76, - "learning_rate": 1.1790194031581914e-05, - "loss": 3.263, + "learning_rate": 1.1852266034951661e-05, + "loss": 3.22, "step": 230600 }, { "epoch": 0.76, - "learning_rate": 1.1773624297857534e-05, - "loss": 3.2483, + "learning_rate": 1.183572321883499e-05, + "loss": 3.2179, "step": 230700 }, { "epoch": 0.76, - "learning_rate": 1.1757054564133154e-05, - "loss": 3.2757, + "learning_rate": 1.1819180402718315e-05, + "loss": 3.221, "step": 230800 }, { - "epoch": 0.77, - "learning_rate": 1.1740484830408775e-05, - "loss": 3.2705, + "epoch": 0.76, + "learning_rate": 1.1802637586601643e-05, + "loss": 3.2246, "step": 230900 }, { - "epoch": 0.77, - "learning_rate": 1.1723915096684397e-05, - "loss": 3.2578, + "epoch": 0.76, + "learning_rate": 1.178609477048497e-05, + "loss": 3.2326, "step": 231000 }, { - "epoch": 0.77, - "learning_rate": 1.1707345362960018e-05, - "loss": 3.2782, + "epoch": 0.76, + "learning_rate": 1.1769551954368296e-05, + "loss": 3.2094, "step": 231100 }, { - "epoch": 0.77, - "learning_rate": 1.1690775629235638e-05, - "loss": 3.2617, + "epoch": 0.76, + "learning_rate": 1.1753009138251624e-05, + "loss": 3.2129, "step": 231200 }, { "epoch": 0.77, - "learning_rate": 1.167420589551126e-05, - "loss": 3.2722, + "learning_rate": 1.1736466322134952e-05, + "loss": 3.229, "step": 231300 }, { "epoch": 0.77, - "learning_rate": 1.1657636161786881e-05, - "loss": 3.2749, + "learning_rate": 1.1719923506018276e-05, + "loss": 3.2283, "step": 231400 }, { "epoch": 0.77, - "learning_rate": 1.1641066428062501e-05, - "loss": 3.2623, + "learning_rate": 1.1703380689901604e-05, + "loss": 3.2207, "step": 231500 }, { "epoch": 0.77, - "learning_rate": 1.1624496694338122e-05, - "loss": 3.2635, + "learning_rate": 1.168683787378493e-05, + "loss": 3.2489, "step": 231600 }, { "epoch": 0.77, - "learning_rate": 1.1607926960613744e-05, - "loss": 3.2798, + "learning_rate": 1.1670295057668258e-05, + "loss": 3.2253, "step": 231700 }, { "epoch": 0.77, - "learning_rate": 1.1591357226889364e-05, - "loss": 3.26, + "learning_rate": 1.1653752241551584e-05, + "loss": 3.2299, "step": 231800 }, { "epoch": 0.77, - "learning_rate": 1.1574787493164987e-05, - "loss": 3.2534, + "learning_rate": 1.163720942543491e-05, + "loss": 3.2038, "step": 231900 }, { "epoch": 0.77, - "learning_rate": 1.1558217759440607e-05, - "loss": 3.2755, + "learning_rate": 1.1620666609318239e-05, + "loss": 3.2242, "step": 232000 }, { "epoch": 0.77, - "learning_rate": 1.1541648025716226e-05, - "loss": 3.2607, + "learning_rate": 1.1604123793201565e-05, + "loss": 3.2254, "step": 232100 }, { "epoch": 0.77, - "learning_rate": 1.1525078291991848e-05, - "loss": 3.2682, + "learning_rate": 1.1587580977084891e-05, + "loss": 3.2248, "step": 232200 }, { "epoch": 0.77, - "learning_rate": 1.150850855826747e-05, - "loss": 3.2503, + "learning_rate": 1.1571038160968219e-05, + "loss": 3.2163, "step": 232300 }, { "epoch": 0.77, - "learning_rate": 1.149193882454309e-05, - "loss": 3.2657, + "learning_rate": 1.1554495344851545e-05, + "loss": 3.2423, "step": 232400 }, { "epoch": 0.77, - "learning_rate": 1.147536909081871e-05, - "loss": 3.2631, + "learning_rate": 1.1537952528734871e-05, + "loss": 3.2317, "step": 232500 }, { "epoch": 0.77, - "learning_rate": 1.1458799357094332e-05, - "loss": 3.2601, + "learning_rate": 1.15214097126182e-05, + "loss": 3.2249, "step": 232600 }, { "epoch": 0.77, - "learning_rate": 1.1442229623369953e-05, - "loss": 3.2585, + "learning_rate": 1.1504866896501526e-05, + "loss": 3.2198, "step": 232700 }, { "epoch": 0.77, - "learning_rate": 1.1425659889645573e-05, - "loss": 3.2559, + "learning_rate": 1.1488324080384854e-05, + "loss": 3.2232, "step": 232800 }, { "epoch": 0.77, - "learning_rate": 1.1409090155921195e-05, - "loss": 3.2561, + "learning_rate": 1.147178126426818e-05, + "loss": 3.2176, "step": 232900 }, { "epoch": 0.77, - "learning_rate": 1.1392520422196816e-05, - "loss": 3.275, + "learning_rate": 1.1455238448151506e-05, + "loss": 3.2313, "step": 233000 }, { "epoch": 0.77, - "learning_rate": 1.1375950688472436e-05, - "loss": 3.2491, + "learning_rate": 1.1438695632034834e-05, + "loss": 3.2256, "step": 233100 }, { "epoch": 0.77, - "learning_rate": 1.1359380954748059e-05, - "loss": 3.253, + "learning_rate": 1.1422152815918158e-05, + "loss": 3.2252, "step": 233200 }, { "epoch": 0.77, - "learning_rate": 1.1342811221023679e-05, - "loss": 3.2698, + "learning_rate": 1.1405609999801486e-05, + "loss": 3.2195, "step": 233300 }, { "epoch": 0.77, - "learning_rate": 1.13262414872993e-05, - "loss": 3.2598, + "learning_rate": 1.1389067183684814e-05, + "loss": 3.2287, "step": 233400 }, { "epoch": 0.77, - "learning_rate": 1.130967175357492e-05, - "loss": 3.2548, + "learning_rate": 1.137252436756814e-05, + "loss": 3.2274, "step": 233500 }, { "epoch": 0.77, - "learning_rate": 1.1293102019850542e-05, - "loss": 3.263, + "learning_rate": 1.1355981551451467e-05, + "loss": 3.2185, "step": 233600 }, { "epoch": 0.77, - "learning_rate": 1.1276532286126163e-05, - "loss": 3.2648, + "learning_rate": 1.1339438735334795e-05, + "loss": 3.2278, "step": 233700 }, { "epoch": 0.77, - "learning_rate": 1.1259962552401783e-05, - "loss": 3.265, + "learning_rate": 1.1322895919218121e-05, + "loss": 3.2185, "step": 233800 }, { - "epoch": 0.78, - "learning_rate": 1.1243392818677404e-05, - "loss": 3.2794, + "epoch": 0.77, + "learning_rate": 1.1306353103101449e-05, + "loss": 3.2237, "step": 233900 }, { - "epoch": 0.78, - "learning_rate": 1.1226823084953026e-05, - "loss": 3.2626, + "epoch": 0.77, + "learning_rate": 1.1289810286984773e-05, + "loss": 3.221, "step": 234000 }, { - "epoch": 0.78, - "learning_rate": 1.1210253351228645e-05, - "loss": 3.2679, + "epoch": 0.77, + "learning_rate": 1.1273267470868101e-05, + "loss": 3.2152, "step": 234100 }, { - "epoch": 0.78, - "learning_rate": 1.1193683617504269e-05, - "loss": 3.2612, + "epoch": 0.77, + "learning_rate": 1.125672465475143e-05, + "loss": 3.23, "step": 234200 }, { "epoch": 0.78, - "learning_rate": 1.1177113883779888e-05, - "loss": 3.2558, + "learning_rate": 1.1240181838634754e-05, + "loss": 3.2167, "step": 234300 }, { "epoch": 0.78, - "learning_rate": 1.1160544150055508e-05, - "loss": 3.264, + "learning_rate": 1.1223639022518082e-05, + "loss": 3.219, "step": 234400 }, { "epoch": 0.78, - "learning_rate": 1.1143974416331131e-05, - "loss": 3.2554, + "learning_rate": 1.1207096206401408e-05, + "loss": 3.2259, "step": 234500 }, { "epoch": 0.78, - "learning_rate": 1.1127404682606751e-05, - "loss": 3.2616, + "learning_rate": 1.1190553390284736e-05, + "loss": 3.2177, "step": 234600 }, { "epoch": 0.78, - "learning_rate": 1.1110834948882373e-05, - "loss": 3.2619, + "learning_rate": 1.1174010574168062e-05, + "loss": 3.2199, "step": 234700 }, { "epoch": 0.78, - "learning_rate": 1.1094265215157992e-05, - "loss": 3.2507, + "learning_rate": 1.1157467758051388e-05, + "loss": 3.2249, "step": 234800 }, { "epoch": 0.78, - "learning_rate": 1.1077695481433614e-05, - "loss": 3.2593, + "learning_rate": 1.1140924941934716e-05, + "loss": 3.2231, "step": 234900 }, { "epoch": 0.78, - "learning_rate": 1.1061125747709235e-05, - "loss": 3.2572, + "learning_rate": 1.1124382125818044e-05, + "loss": 3.2161, "step": 235000 }, { "epoch": 0.78, - "learning_rate": 1.1044556013984855e-05, - "loss": 3.2724, + "learning_rate": 1.1107839309701369e-05, + "loss": 3.223, "step": 235100 }, { "epoch": 0.78, - "learning_rate": 1.1027986280260477e-05, - "loss": 3.2596, + "learning_rate": 1.1091296493584697e-05, + "loss": 3.2142, "step": 235200 }, { "epoch": 0.78, - "learning_rate": 1.1011416546536098e-05, - "loss": 3.258, + "learning_rate": 1.1074753677468023e-05, + "loss": 3.2126, "step": 235300 }, { "epoch": 0.78, - "learning_rate": 1.0994846812811718e-05, - "loss": 3.2637, + "learning_rate": 1.105821086135135e-05, + "loss": 3.2136, "step": 235400 }, { "epoch": 0.78, - "learning_rate": 1.0978277079087341e-05, - "loss": 3.2707, + "learning_rate": 1.1041668045234677e-05, + "loss": 3.2069, "step": 235500 }, { "epoch": 0.78, - "learning_rate": 1.096170734536296e-05, - "loss": 3.2546, + "learning_rate": 1.1025125229118003e-05, + "loss": 3.2034, "step": 235600 }, { "epoch": 0.78, - "learning_rate": 1.094513761163858e-05, - "loss": 3.2528, + "learning_rate": 1.1008582413001331e-05, + "loss": 3.2187, "step": 235700 }, { "epoch": 0.78, - "learning_rate": 1.0928567877914204e-05, - "loss": 3.2718, + "learning_rate": 1.0992039596884657e-05, + "loss": 3.2117, "step": 235800 }, { "epoch": 0.78, - "learning_rate": 1.0911998144189823e-05, - "loss": 3.246, + "learning_rate": 1.0975496780767984e-05, + "loss": 3.2301, "step": 235900 }, { "epoch": 0.78, - "learning_rate": 1.0895428410465445e-05, - "loss": 3.2584, + "learning_rate": 1.0958953964651311e-05, + "loss": 3.2285, "step": 236000 }, { "epoch": 0.78, - "learning_rate": 1.0878858676741065e-05, - "loss": 3.2703, + "learning_rate": 1.0942411148534638e-05, + "loss": 3.2254, "step": 236100 }, { "epoch": 0.78, - "learning_rate": 1.0862288943016686e-05, - "loss": 3.2502, + "learning_rate": 1.0925868332417964e-05, + "loss": 3.2316, "step": 236200 }, { "epoch": 0.78, - "learning_rate": 1.0845719209292308e-05, - "loss": 3.2637, + "learning_rate": 1.0909325516301292e-05, + "loss": 3.2238, "step": 236300 }, { "epoch": 0.78, - "learning_rate": 1.0829149475567927e-05, - "loss": 3.2497, + "learning_rate": 1.0892782700184618e-05, + "loss": 3.2149, "step": 236400 }, { "epoch": 0.78, - "learning_rate": 1.0812579741843549e-05, - "loss": 3.2672, + "learning_rate": 1.0876239884067946e-05, + "loss": 3.216, "step": 236500 }, { "epoch": 0.78, - "learning_rate": 1.079601000811917e-05, - "loss": 3.2641, + "learning_rate": 1.0859697067951272e-05, + "loss": 3.2164, "step": 236600 }, { "epoch": 0.78, - "learning_rate": 1.077944027439479e-05, - "loss": 3.2619, + "learning_rate": 1.0843154251834598e-05, + "loss": 3.2226, "step": 236700 }, { "epoch": 0.78, - "learning_rate": 1.0762870540670413e-05, - "loss": 3.2614, + "learning_rate": 1.0826611435717926e-05, + "loss": 3.2252, "step": 236800 }, { - "epoch": 0.79, - "learning_rate": 1.0746300806946033e-05, - "loss": 3.2453, + "epoch": 0.78, + "learning_rate": 1.0810068619601253e-05, + "loss": 3.2375, "step": 236900 }, { - "epoch": 0.79, - "learning_rate": 1.0729731073221653e-05, - "loss": 3.2767, + "epoch": 0.78, + "learning_rate": 1.0793525803484579e-05, + "loss": 3.2314, "step": 237000 }, { - "epoch": 0.79, - "learning_rate": 1.0713161339497276e-05, - "loss": 3.2725, + "epoch": 0.78, + "learning_rate": 1.0776982987367907e-05, + "loss": 3.2316, "step": 237100 }, { - "epoch": 0.79, - "learning_rate": 1.0696591605772896e-05, - "loss": 3.2679, + "epoch": 0.78, + "learning_rate": 1.0760440171251233e-05, + "loss": 3.2203, "step": 237200 }, { "epoch": 0.79, - "learning_rate": 1.0680021872048517e-05, - "loss": 3.2508, + "learning_rate": 1.074389735513456e-05, + "loss": 3.2212, "step": 237300 }, { "epoch": 0.79, - "learning_rate": 1.0663452138324137e-05, - "loss": 3.2544, + "learning_rate": 1.0727354539017887e-05, + "loss": 3.2242, "step": 237400 }, { "epoch": 0.79, - "learning_rate": 1.0646882404599758e-05, - "loss": 3.2497, + "learning_rate": 1.0710811722901213e-05, + "loss": 3.2361, "step": 237500 }, { "epoch": 0.79, - "learning_rate": 1.063031267087538e-05, - "loss": 3.2581, + "learning_rate": 1.0694268906784541e-05, + "loss": 3.2217, "step": 237600 }, { "epoch": 0.79, - "learning_rate": 1.0613742937151e-05, - "loss": 3.2723, + "learning_rate": 1.0677726090667866e-05, + "loss": 3.2175, "step": 237700 }, { "epoch": 0.79, - "learning_rate": 1.0597173203426621e-05, - "loss": 3.2564, + "learning_rate": 1.0661183274551194e-05, + "loss": 3.2224, "step": 237800 }, { "epoch": 0.79, - "learning_rate": 1.0580603469702243e-05, - "loss": 3.2642, + "learning_rate": 1.0644640458434522e-05, + "loss": 3.2317, "step": 237900 }, { "epoch": 0.79, - "learning_rate": 1.0564033735977862e-05, - "loss": 3.2776, + "learning_rate": 1.0628097642317846e-05, + "loss": 3.2147, "step": 238000 }, { "epoch": 0.79, - "learning_rate": 1.0547464002253486e-05, - "loss": 3.264, + "learning_rate": 1.0611554826201174e-05, + "loss": 3.2249, "step": 238100 }, { "epoch": 0.79, - "learning_rate": 1.0530894268529105e-05, - "loss": 3.2597, + "learning_rate": 1.0595012010084502e-05, + "loss": 3.2261, "step": 238200 }, { "epoch": 0.79, - "learning_rate": 1.0514324534804727e-05, - "loss": 3.2609, + "learning_rate": 1.0578469193967828e-05, + "loss": 3.2294, "step": 238300 }, { "epoch": 0.79, - "learning_rate": 1.0497754801080348e-05, - "loss": 3.2654, + "learning_rate": 1.0561926377851155e-05, + "loss": 3.2178, "step": 238400 }, { "epoch": 0.79, - "learning_rate": 1.0481185067355968e-05, - "loss": 3.2496, + "learning_rate": 1.054538356173448e-05, + "loss": 3.228, "step": 238500 }, { "epoch": 0.79, - "learning_rate": 1.046461533363159e-05, - "loss": 3.2799, + "learning_rate": 1.0528840745617809e-05, + "loss": 3.2237, "step": 238600 }, { "epoch": 0.79, - "learning_rate": 1.044804559990721e-05, - "loss": 3.2657, + "learning_rate": 1.0512297929501137e-05, + "loss": 3.2282, "step": 238700 }, { "epoch": 0.79, - "learning_rate": 1.043147586618283e-05, - "loss": 3.2495, + "learning_rate": 1.0495755113384461e-05, + "loss": 3.2142, "step": 238800 }, { "epoch": 0.79, - "learning_rate": 1.0414906132458452e-05, - "loss": 3.2575, + "learning_rate": 1.0479212297267789e-05, + "loss": 3.2121, "step": 238900 }, { "epoch": 0.79, - "learning_rate": 1.0398336398734072e-05, - "loss": 3.2421, + "learning_rate": 1.0462669481151115e-05, + "loss": 3.2224, "step": 239000 }, { "epoch": 0.79, - "learning_rate": 1.0381766665009695e-05, - "loss": 3.2636, + "learning_rate": 1.0446126665034443e-05, + "loss": 3.2293, "step": 239100 }, { "epoch": 0.79, - "learning_rate": 1.0365196931285315e-05, - "loss": 3.2623, + "learning_rate": 1.042958384891777e-05, + "loss": 3.223, "step": 239200 }, { "epoch": 0.79, - "learning_rate": 1.0348627197560935e-05, - "loss": 3.2671, + "learning_rate": 1.0413041032801096e-05, + "loss": 3.2256, "step": 239300 }, { "epoch": 0.79, - "learning_rate": 1.0332057463836558e-05, - "loss": 3.247, + "learning_rate": 1.0396498216684424e-05, + "loss": 3.2416, "step": 239400 }, { "epoch": 0.79, - "learning_rate": 1.0315487730112178e-05, - "loss": 3.2798, + "learning_rate": 1.037995540056775e-05, + "loss": 3.2155, "step": 239500 }, { "epoch": 0.79, - "learning_rate": 1.0298917996387799e-05, - "loss": 3.2811, + "learning_rate": 1.0363412584451076e-05, + "loss": 3.2162, "step": 239600 }, { "epoch": 0.79, - "learning_rate": 1.028234826266342e-05, - "loss": 3.2641, + "learning_rate": 1.0346869768334404e-05, + "loss": 3.2125, "step": 239700 }, { "epoch": 0.79, - "learning_rate": 1.026577852893904e-05, - "loss": 3.2678, + "learning_rate": 1.033032695221773e-05, + "loss": 3.2126, "step": 239800 }, { - "epoch": 0.8, - "learning_rate": 1.0249208795214662e-05, - "loss": 3.2535, + "epoch": 0.79, + "learning_rate": 1.0313784136101056e-05, + "loss": 3.2255, "step": 239900 }, { - "epoch": 0.8, - "learning_rate": 1.0232639061490282e-05, - "loss": 3.2556, + "epoch": 0.79, + "learning_rate": 1.0297241319984384e-05, + "loss": 3.2178, "step": 240000 }, { - "epoch": 0.8, - "learning_rate": 1.0216069327765903e-05, - "loss": 3.2472, + "epoch": 0.79, + "learning_rate": 1.028069850386771e-05, + "loss": 3.2192, "step": 240100 }, { - "epoch": 0.8, - "learning_rate": 1.0199499594041524e-05, - "loss": 3.2581, + "epoch": 0.79, + "learning_rate": 1.0264155687751038e-05, + "loss": 3.2138, "step": 240200 }, { "epoch": 0.8, - "learning_rate": 1.0182929860317144e-05, - "loss": 3.2605, + "learning_rate": 1.0247612871634365e-05, + "loss": 3.2154, "step": 240300 }, { "epoch": 0.8, - "learning_rate": 1.0166360126592767e-05, - "loss": 3.2525, + "learning_rate": 1.0231070055517691e-05, + "loss": 3.2144, "step": 240400 }, { "epoch": 0.8, - "learning_rate": 1.0149790392868387e-05, - "loss": 3.2509, + "learning_rate": 1.0214527239401019e-05, + "loss": 3.2252, "step": 240500 }, { "epoch": 0.8, - "learning_rate": 1.0133220659144007e-05, - "loss": 3.2673, + "learning_rate": 1.0197984423284345e-05, + "loss": 3.2331, "step": 240600 }, { "epoch": 0.8, - "learning_rate": 1.011665092541963e-05, - "loss": 3.2623, + "learning_rate": 1.0181441607167671e-05, + "loss": 3.236, "step": 240700 }, { "epoch": 0.8, - "learning_rate": 1.010008119169525e-05, - "loss": 3.2658, + "learning_rate": 1.0164898791051e-05, + "loss": 3.2265, "step": 240800 }, { "epoch": 0.8, - "learning_rate": 1.0083511457970871e-05, - "loss": 3.2704, + "learning_rate": 1.0148355974934325e-05, + "loss": 3.2162, "step": 240900 }, { "epoch": 0.8, - "learning_rate": 1.0066941724246493e-05, - "loss": 3.2569, + "learning_rate": 1.0131813158817652e-05, + "loss": 3.2271, "step": 241000 }, { "epoch": 0.8, - "learning_rate": 1.0050371990522113e-05, - "loss": 3.2645, + "learning_rate": 1.011527034270098e-05, + "loss": 3.221, "step": 241100 }, { "epoch": 0.8, - "learning_rate": 1.0033802256797734e-05, - "loss": 3.2605, + "learning_rate": 1.0098727526584306e-05, + "loss": 3.2314, "step": 241200 }, { "epoch": 0.8, - "learning_rate": 1.0017232523073354e-05, - "loss": 3.2629, + "learning_rate": 1.0082184710467634e-05, + "loss": 3.2172, "step": 241300 }, { "epoch": 0.8, - "learning_rate": 1.0000662789348975e-05, - "loss": 3.2647, + "learning_rate": 1.006564189435096e-05, + "loss": 3.2033, "step": 241400 }, { "epoch": 0.8, - "learning_rate": 9.984093055624597e-06, - "loss": 3.2554, + "learning_rate": 1.0049099078234286e-05, + "loss": 3.2219, "step": 241500 }, { "epoch": 0.8, - "learning_rate": 9.967523321900217e-06, - "loss": 3.2658, + "learning_rate": 1.0032556262117614e-05, + "loss": 3.2107, "step": 241600 }, { "epoch": 0.8, - "learning_rate": 9.95095358817584e-06, - "loss": 3.256, + "learning_rate": 1.0016013446000939e-05, + "loss": 3.2272, "step": 241700 }, { "epoch": 0.8, - "learning_rate": 9.93438385445146e-06, - "loss": 3.2568, + "learning_rate": 9.999470629884267e-06, + "loss": 3.2188, "step": 241800 }, { "epoch": 0.8, - "learning_rate": 9.91781412072708e-06, - "loss": 3.2656, + "learning_rate": 9.982927813767595e-06, + "loss": 3.2327, "step": 241900 }, { "epoch": 0.8, - "learning_rate": 9.901244387002702e-06, - "loss": 3.2557, + "learning_rate": 9.96638499765092e-06, + "loss": 3.2225, "step": 242000 }, { "epoch": 0.8, - "learning_rate": 9.884674653278322e-06, - "loss": 3.2377, + "learning_rate": 9.949842181534247e-06, + "loss": 3.2252, "step": 242100 }, { "epoch": 0.8, - "learning_rate": 9.868104919553944e-06, - "loss": 3.2719, + "learning_rate": 9.933299365417573e-06, + "loss": 3.2331, "step": 242200 }, { "epoch": 0.8, - "learning_rate": 9.851535185829565e-06, - "loss": 3.2636, + "learning_rate": 9.916756549300901e-06, + "loss": 3.2099, "step": 242300 }, { "epoch": 0.8, - "learning_rate": 9.834965452105185e-06, - "loss": 3.2632, + "learning_rate": 9.900213733184229e-06, + "loss": 3.2063, "step": 242400 }, { "epoch": 0.8, - "learning_rate": 9.818395718380806e-06, - "loss": 3.2589, + "learning_rate": 9.883670917067554e-06, + "loss": 3.2239, "step": 242500 }, { "epoch": 0.8, - "learning_rate": 9.801825984656426e-06, - "loss": 3.2642, + "learning_rate": 9.867128100950882e-06, + "loss": 3.2276, "step": 242600 }, { "epoch": 0.8, - "learning_rate": 9.785256250932048e-06, - "loss": 3.2494, + "learning_rate": 9.85058528483421e-06, + "loss": 3.2067, "step": 242700 }, { "epoch": 0.8, - "learning_rate": 9.768686517207669e-06, - "loss": 3.2668, + "learning_rate": 9.834042468717536e-06, + "loss": 3.2311, "step": 242800 }, { "epoch": 0.8, - "learning_rate": 9.752116783483289e-06, - "loss": 3.2728, + "learning_rate": 9.817499652600862e-06, + "loss": 3.2166, "step": 242900 }, { - "epoch": 0.81, - "learning_rate": 9.735547049758912e-06, - "loss": 3.2542, + "epoch": 0.8, + "learning_rate": 9.800956836484188e-06, + "loss": 3.2267, "step": 243000 }, { - "epoch": 0.81, - "learning_rate": 9.718977316034532e-06, - "loss": 3.2544, + "epoch": 0.8, + "learning_rate": 9.784414020367516e-06, + "loss": 3.2151, "step": 243100 }, { - "epoch": 0.81, - "learning_rate": 9.702407582310152e-06, - "loss": 3.2593, + "epoch": 0.8, + "learning_rate": 9.767871204250842e-06, + "loss": 3.2389, "step": 243200 }, { - "epoch": 0.81, - "learning_rate": 9.685837848585775e-06, - "loss": 3.26, + "epoch": 0.8, + "learning_rate": 9.751328388134169e-06, + "loss": 3.2212, "step": 243300 }, { "epoch": 0.81, - "learning_rate": 9.669268114861394e-06, - "loss": 3.2395, + "learning_rate": 9.734785572017496e-06, + "loss": 3.2187, "step": 243400 }, { "epoch": 0.81, - "learning_rate": 9.652698381137016e-06, - "loss": 3.2452, + "learning_rate": 9.718242755900823e-06, + "loss": 3.2265, "step": 243500 }, { "epoch": 0.81, - "learning_rate": 9.636128647412637e-06, - "loss": 3.2545, + "learning_rate": 9.701699939784149e-06, + "loss": 3.2133, "step": 243600 }, { "epoch": 0.81, - "learning_rate": 9.619558913688257e-06, - "loss": 3.2534, + "learning_rate": 9.685157123667477e-06, + "loss": 3.2243, "step": 243700 }, { "epoch": 0.81, - "learning_rate": 9.602989179963879e-06, - "loss": 3.2474, + "learning_rate": 9.668614307550803e-06, + "loss": 3.2045, "step": 243800 }, { "epoch": 0.81, - "learning_rate": 9.586419446239498e-06, - "loss": 3.2655, + "learning_rate": 9.652071491434131e-06, + "loss": 3.2233, "step": 243900 }, { "epoch": 0.81, - "learning_rate": 9.56984971251512e-06, - "loss": 3.2599, + "learning_rate": 9.635528675317457e-06, + "loss": 3.2073, "step": 244000 }, { "epoch": 0.81, - "learning_rate": 9.553279978790741e-06, - "loss": 3.2589, + "learning_rate": 9.618985859200783e-06, + "loss": 3.2173, "step": 244100 }, { "epoch": 0.81, - "learning_rate": 9.536710245066361e-06, - "loss": 3.2568, + "learning_rate": 9.602443043084111e-06, + "loss": 3.2233, "step": 244200 }, { "epoch": 0.81, - "learning_rate": 9.520140511341984e-06, - "loss": 3.2597, + "learning_rate": 9.585900226967438e-06, + "loss": 3.2304, "step": 244300 }, { "epoch": 0.81, - "learning_rate": 9.503570777617604e-06, - "loss": 3.2552, + "learning_rate": 9.569357410850764e-06, + "loss": 3.213, "step": 244400 }, { "epoch": 0.81, - "learning_rate": 9.487001043893226e-06, - "loss": 3.2448, + "learning_rate": 9.552814594734092e-06, + "loss": 3.2104, "step": 244500 }, { "epoch": 0.81, - "learning_rate": 9.470431310168847e-06, - "loss": 3.2578, + "learning_rate": 9.536271778617418e-06, + "loss": 3.215, "step": 244600 }, { "epoch": 0.81, - "learning_rate": 9.453861576444467e-06, - "loss": 3.2493, + "learning_rate": 9.519728962500744e-06, + "loss": 3.2168, "step": 244700 }, { "epoch": 0.81, - "learning_rate": 9.437291842720088e-06, - "loss": 3.2493, + "learning_rate": 9.503186146384072e-06, + "loss": 3.2142, "step": 244800 }, { "epoch": 0.81, - "learning_rate": 9.42072210899571e-06, - "loss": 3.2673, + "learning_rate": 9.486643330267398e-06, + "loss": 3.2159, "step": 244900 }, { "epoch": 0.81, - "learning_rate": 9.40415237527133e-06, - "loss": 3.2691, + "learning_rate": 9.470100514150726e-06, + "loss": 3.2112, "step": 245000 }, { "epoch": 0.81, - "learning_rate": 9.387582641546951e-06, - "loss": 3.2496, + "learning_rate": 9.453557698034052e-06, + "loss": 3.2136, "step": 245100 }, { "epoch": 0.81, - "learning_rate": 9.37101290782257e-06, - "loss": 3.2671, + "learning_rate": 9.437014881917379e-06, + "loss": 3.2208, "step": 245200 }, { "epoch": 0.81, - "learning_rate": 9.354443174098194e-06, - "loss": 3.2581, + "learning_rate": 9.420472065800707e-06, + "loss": 3.2191, "step": 245300 }, { "epoch": 0.81, - "learning_rate": 9.337873440373814e-06, - "loss": 3.255, + "learning_rate": 9.403929249684031e-06, + "loss": 3.2225, "step": 245400 }, { "epoch": 0.81, - "learning_rate": 9.321303706649433e-06, - "loss": 3.2606, + "learning_rate": 9.387386433567359e-06, + "loss": 3.2305, "step": 245500 }, { "epoch": 0.81, - "learning_rate": 9.304733972925057e-06, - "loss": 3.27, + "learning_rate": 9.370843617450687e-06, + "loss": 3.2402, "step": 245600 }, { "epoch": 0.81, - "learning_rate": 9.288164239200676e-06, - "loss": 3.2683, + "learning_rate": 9.354300801334013e-06, + "loss": 3.2247, "step": 245700 }, { "epoch": 0.81, - "learning_rate": 9.271594505476298e-06, - "loss": 3.2596, + "learning_rate": 9.33775798521734e-06, + "loss": 3.2249, "step": 245800 }, { "epoch": 0.81, - "learning_rate": 9.25502477175192e-06, - "loss": 3.2673, + "learning_rate": 9.321215169100667e-06, + "loss": 3.2293, "step": 245900 }, { - "epoch": 0.82, - "learning_rate": 9.238455038027539e-06, - "loss": 3.2601, + "epoch": 0.81, + "learning_rate": 9.304672352983994e-06, + "loss": 3.2142, "step": 246000 }, { - "epoch": 0.82, - "learning_rate": 9.22188530430316e-06, - "loss": 3.2592, + "epoch": 0.81, + "learning_rate": 9.288129536867322e-06, + "loss": 3.23, "step": 246100 }, { - "epoch": 0.82, - "learning_rate": 9.205315570578782e-06, - "loss": 3.2513, + "epoch": 0.81, + "learning_rate": 9.271586720750646e-06, + "loss": 3.2142, "step": 246200 }, { - "epoch": 0.82, - "learning_rate": 9.188745836854402e-06, - "loss": 3.2455, + "epoch": 0.81, + "learning_rate": 9.255043904633974e-06, + "loss": 3.2255, "step": 246300 }, { "epoch": 0.82, - "learning_rate": 9.172176103130023e-06, - "loss": 3.2609, + "learning_rate": 9.238501088517302e-06, + "loss": 3.2204, "step": 246400 }, { "epoch": 0.82, - "learning_rate": 9.155606369405643e-06, - "loss": 3.2354, + "learning_rate": 9.221958272400628e-06, + "loss": 3.22, "step": 246500 }, { "epoch": 0.82, - "learning_rate": 9.139036635681266e-06, - "loss": 3.2476, + "learning_rate": 9.205415456283954e-06, + "loss": 3.2341, "step": 246600 }, { "epoch": 0.82, - "learning_rate": 9.122466901956886e-06, - "loss": 3.2546, + "learning_rate": 9.18887264016728e-06, + "loss": 3.2416, "step": 246700 }, { "epoch": 0.82, - "learning_rate": 9.105897168232506e-06, - "loss": 3.2549, + "learning_rate": 9.172329824050609e-06, + "loss": 3.2219, "step": 246800 }, { "epoch": 0.82, - "learning_rate": 9.089327434508129e-06, - "loss": 3.2424, + "learning_rate": 9.155787007933935e-06, + "loss": 3.2225, "step": 246900 }, { "epoch": 0.82, - "learning_rate": 9.072757700783749e-06, - "loss": 3.2676, + "learning_rate": 9.139244191817261e-06, + "loss": 3.2175, "step": 247000 }, { "epoch": 0.82, - "learning_rate": 9.05618796705937e-06, - "loss": 3.2484, + "learning_rate": 9.122701375700589e-06, + "loss": 3.2151, "step": 247100 }, { "epoch": 0.82, - "learning_rate": 9.039618233334992e-06, - "loss": 3.2649, + "learning_rate": 9.106158559583917e-06, + "loss": 3.2106, "step": 247200 }, { "epoch": 0.82, - "learning_rate": 9.023048499610611e-06, - "loss": 3.2618, + "learning_rate": 9.089615743467241e-06, + "loss": 3.2113, "step": 247300 }, { "epoch": 0.82, - "learning_rate": 9.006478765886233e-06, - "loss": 3.2714, + "learning_rate": 9.07307292735057e-06, + "loss": 3.2067, "step": 247400 }, { "epoch": 0.82, - "learning_rate": 8.989909032161854e-06, - "loss": 3.2691, + "learning_rate": 9.056530111233896e-06, + "loss": 3.2171, "step": 247500 }, { "epoch": 0.82, - "learning_rate": 8.973339298437474e-06, - "loss": 3.2655, + "learning_rate": 9.039987295117223e-06, + "loss": 3.218, "step": 247600 }, { "epoch": 0.82, - "learning_rate": 8.956769564713096e-06, - "loss": 3.2716, + "learning_rate": 9.02344447900055e-06, + "loss": 3.2266, "step": 247700 }, { "epoch": 0.82, - "learning_rate": 8.940199830988715e-06, - "loss": 3.2435, + "learning_rate": 9.006901662883876e-06, + "loss": 3.219, "step": 247800 }, { "epoch": 0.82, - "learning_rate": 8.923630097264338e-06, - "loss": 3.2454, + "learning_rate": 8.990358846767204e-06, + "loss": 3.2154, "step": 247900 }, { "epoch": 0.82, - "learning_rate": 8.907060363539958e-06, - "loss": 3.2509, + "learning_rate": 8.97381603065053e-06, + "loss": 3.2089, "step": 248000 }, { "epoch": 0.82, - "learning_rate": 8.890490629815578e-06, - "loss": 3.2453, + "learning_rate": 8.957273214533856e-06, + "loss": 3.2259, "step": 248100 }, { "epoch": 0.82, - "learning_rate": 8.873920896091201e-06, - "loss": 3.266, + "learning_rate": 8.940730398417184e-06, + "loss": 3.2254, "step": 248200 }, { "epoch": 0.82, - "learning_rate": 8.857351162366821e-06, - "loss": 3.2502, + "learning_rate": 8.92418758230051e-06, + "loss": 3.2102, "step": 248300 }, { "epoch": 0.82, - "learning_rate": 8.840781428642442e-06, - "loss": 3.262, + "learning_rate": 8.907644766183837e-06, + "loss": 3.2187, "step": 248400 }, { "epoch": 0.82, - "learning_rate": 8.824211694918064e-06, - "loss": 3.2582, + "learning_rate": 8.891101950067165e-06, + "loss": 3.2163, "step": 248500 }, { "epoch": 0.82, - "learning_rate": 8.807641961193684e-06, - "loss": 3.2581, + "learning_rate": 8.87455913395049e-06, + "loss": 3.218, "step": 248600 }, { "epoch": 0.82, - "learning_rate": 8.791072227469305e-06, - "loss": 3.2618, + "learning_rate": 8.858016317833819e-06, + "loss": 3.2241, "step": 248700 }, { "epoch": 0.82, - "learning_rate": 8.774502493744927e-06, - "loss": 3.2507, + "learning_rate": 8.841473501717145e-06, + "loss": 3.2193, "step": 248800 }, { "epoch": 0.82, - "learning_rate": 8.757932760020546e-06, - "loss": 3.2429, + "learning_rate": 8.824930685600471e-06, + "loss": 3.231, "step": 248900 }, { - "epoch": 0.83, - "learning_rate": 8.741363026296168e-06, - "loss": 3.2569, + "epoch": 0.82, + "learning_rate": 8.808387869483799e-06, + "loss": 3.2135, "step": 249000 }, { - "epoch": 0.83, - "learning_rate": 8.724793292571788e-06, - "loss": 3.245, + "epoch": 0.82, + "learning_rate": 8.791845053367125e-06, + "loss": 3.2221, "step": 249100 }, { - "epoch": 0.83, - "learning_rate": 8.70822355884741e-06, - "loss": 3.2469, + "epoch": 0.82, + "learning_rate": 8.775302237250452e-06, + "loss": 3.2015, "step": 249200 }, { - "epoch": 0.83, - "learning_rate": 8.69165382512303e-06, - "loss": 3.2576, + "epoch": 0.82, + "learning_rate": 8.75875942113378e-06, + "loss": 3.2196, "step": 249300 }, { "epoch": 0.83, - "learning_rate": 8.67508409139865e-06, - "loss": 3.2481, + "learning_rate": 8.742216605017106e-06, + "loss": 3.2229, "step": 249400 }, { "epoch": 0.83, - "learning_rate": 8.658514357674273e-06, - "loss": 3.2513, + "learning_rate": 8.725673788900432e-06, + "loss": 3.2164, "step": 249500 }, { "epoch": 0.83, - "learning_rate": 8.641944623949893e-06, - "loss": 3.2557, + "learning_rate": 8.70913097278376e-06, + "loss": 3.2257, "step": 249600 }, { "epoch": 0.83, - "learning_rate": 8.625374890225515e-06, - "loss": 3.2621, + "learning_rate": 8.692588156667086e-06, + "loss": 3.2164, "step": 249700 }, { "epoch": 0.83, - "learning_rate": 8.608805156501136e-06, - "loss": 3.2489, + "learning_rate": 8.676045340550414e-06, + "loss": 3.215, "step": 249800 }, { "epoch": 0.83, - "learning_rate": 8.592235422776756e-06, - "loss": 3.247, + "learning_rate": 8.659502524433739e-06, + "loss": 3.205, "step": 249900 }, { "epoch": 0.83, - "learning_rate": 8.575665689052377e-06, - "loss": 3.2597, + "learning_rate": 8.642959708317066e-06, + "loss": 3.2089, "step": 250000 }, { "epoch": 0.83, - "learning_rate": 8.559095955327999e-06, - "loss": 3.244, + "learning_rate": 8.626416892200394e-06, + "loss": 3.2043, "step": 250100 }, { "epoch": 0.83, - "learning_rate": 8.542526221603619e-06, - "loss": 3.2742, + "learning_rate": 8.60987407608372e-06, + "loss": 3.2072, "step": 250200 }, { "epoch": 0.83, - "learning_rate": 8.52595648787924e-06, - "loss": 3.2509, + "learning_rate": 8.593331259967047e-06, + "loss": 3.2192, "step": 250300 }, { "epoch": 0.83, - "learning_rate": 8.50938675415486e-06, - "loss": 3.2494, + "learning_rate": 8.576788443850375e-06, + "loss": 3.2194, "step": 250400 }, { "epoch": 0.83, - "learning_rate": 8.492817020430483e-06, - "loss": 3.2521, + "learning_rate": 8.560245627733701e-06, + "loss": 3.2241, "step": 250500 }, { "epoch": 0.83, - "learning_rate": 8.476247286706103e-06, - "loss": 3.2508, + "learning_rate": 8.543702811617027e-06, + "loss": 3.2131, "step": 250600 }, { "epoch": 0.83, - "learning_rate": 8.459677552981724e-06, - "loss": 3.263, + "learning_rate": 8.527159995500353e-06, + "loss": 3.2192, "step": 250700 }, { "epoch": 0.83, - "learning_rate": 8.443107819257346e-06, - "loss": 3.2574, + "learning_rate": 8.510617179383681e-06, + "loss": 3.215, "step": 250800 }, { "epoch": 0.83, - "learning_rate": 8.426538085532966e-06, - "loss": 3.2547, + "learning_rate": 8.49407436326701e-06, + "loss": 3.228, "step": 250900 }, { "epoch": 0.83, - "learning_rate": 8.409968351808587e-06, - "loss": 3.2543, + "learning_rate": 8.477531547150334e-06, + "loss": 3.2162, "step": 251000 }, { "epoch": 0.83, - "learning_rate": 8.393398618084208e-06, - "loss": 3.2522, + "learning_rate": 8.460988731033662e-06, + "loss": 3.2225, "step": 251100 }, { "epoch": 0.83, - "learning_rate": 8.376828884359828e-06, - "loss": 3.2499, + "learning_rate": 8.444445914916988e-06, + "loss": 3.2252, "step": 251200 }, { "epoch": 0.83, - "learning_rate": 8.36025915063545e-06, - "loss": 3.2579, + "learning_rate": 8.427903098800316e-06, + "loss": 3.2164, "step": 251300 }, { "epoch": 0.83, - "learning_rate": 8.343689416911071e-06, - "loss": 3.2581, + "learning_rate": 8.411360282683642e-06, + "loss": 3.2267, "step": 251400 }, { "epoch": 0.83, - "learning_rate": 8.327119683186693e-06, - "loss": 3.256, + "learning_rate": 8.394817466566968e-06, + "loss": 3.2188, "step": 251500 }, { "epoch": 0.83, - "learning_rate": 8.310549949462312e-06, - "loss": 3.2564, + "learning_rate": 8.378274650450296e-06, + "loss": 3.2228, "step": 251600 }, { "epoch": 0.83, - "learning_rate": 8.293980215737932e-06, - "loss": 3.2578, + "learning_rate": 8.361731834333623e-06, + "loss": 3.2283, "step": 251700 }, { "epoch": 0.83, - "learning_rate": 8.277410482013555e-06, - "loss": 3.2483, + "learning_rate": 8.345189018216949e-06, + "loss": 3.2204, "step": 251800 }, { "epoch": 0.83, - "learning_rate": 8.260840748289175e-06, - "loss": 3.2581, + "learning_rate": 8.328646202100277e-06, + "loss": 3.2194, "step": 251900 }, { - "epoch": 0.84, - "learning_rate": 8.244271014564797e-06, - "loss": 3.2656, + "epoch": 0.83, + "learning_rate": 8.312103385983603e-06, + "loss": 3.2219, "step": 252000 }, { - "epoch": 0.84, - "learning_rate": 8.227701280840418e-06, - "loss": 3.2674, + "epoch": 0.83, + "learning_rate": 8.29556056986693e-06, + "loss": 3.2085, "step": 252100 }, { - "epoch": 0.84, - "learning_rate": 8.211131547116038e-06, - "loss": 3.2455, + "epoch": 0.83, + "learning_rate": 8.279017753750257e-06, + "loss": 3.2133, "step": 252200 }, { - "epoch": 0.84, - "learning_rate": 8.19456181339166e-06, - "loss": 3.2503, + "epoch": 0.83, + "learning_rate": 8.262474937633583e-06, + "loss": 3.2244, "step": 252300 }, { "epoch": 0.84, - "learning_rate": 8.17799207966728e-06, - "loss": 3.2462, + "learning_rate": 8.245932121516911e-06, + "loss": 3.2245, "step": 252400 }, { "epoch": 0.84, - "learning_rate": 8.1614223459429e-06, - "loss": 3.2531, + "learning_rate": 8.229389305400237e-06, + "loss": 3.2169, "step": 252500 }, { "epoch": 0.84, - "learning_rate": 8.144852612218522e-06, - "loss": 3.2386, + "learning_rate": 8.212846489283564e-06, + "loss": 3.2202, "step": 252600 }, { "epoch": 0.84, - "learning_rate": 8.128282878494143e-06, - "loss": 3.2602, + "learning_rate": 8.196303673166892e-06, + "loss": 3.213, "step": 252700 }, { "epoch": 0.84, - "learning_rate": 8.111713144769765e-06, - "loss": 3.2487, + "learning_rate": 8.179760857050218e-06, + "loss": 3.2145, "step": 252800 }, { "epoch": 0.84, - "learning_rate": 8.095143411045385e-06, - "loss": 3.2579, + "learning_rate": 8.163218040933544e-06, + "loss": 3.233, "step": 252900 }, { "epoch": 0.84, - "learning_rate": 8.078573677321004e-06, - "loss": 3.2419, + "learning_rate": 8.146675224816872e-06, + "loss": 3.2078, "step": 253000 }, { "epoch": 0.84, - "learning_rate": 8.062003943596628e-06, - "loss": 3.2558, + "learning_rate": 8.130132408700198e-06, + "loss": 3.205, "step": 253100 }, { "epoch": 0.84, - "learning_rate": 8.045434209872247e-06, - "loss": 3.2537, + "learning_rate": 8.113589592583524e-06, + "loss": 3.2204, "step": 253200 }, { "epoch": 0.84, - "learning_rate": 8.028864476147869e-06, - "loss": 3.2619, + "learning_rate": 8.097046776466852e-06, + "loss": 3.223, "step": 253300 }, { "epoch": 0.84, - "learning_rate": 8.01229474242349e-06, - "loss": 3.2587, + "learning_rate": 8.080503960350179e-06, + "loss": 3.2109, "step": 253400 }, { "epoch": 0.84, - "learning_rate": 7.99572500869911e-06, - "loss": 3.2565, + "learning_rate": 8.063961144233507e-06, + "loss": 3.2133, "step": 253500 }, { "epoch": 0.84, - "learning_rate": 7.979155274974732e-06, - "loss": 3.26, + "learning_rate": 8.047418328116833e-06, + "loss": 3.2281, "step": 253600 }, { "epoch": 0.84, - "learning_rate": 7.962585541250353e-06, - "loss": 3.2609, + "learning_rate": 8.030875512000159e-06, + "loss": 3.2132, "step": 253700 }, { "epoch": 0.84, - "learning_rate": 7.946015807525973e-06, - "loss": 3.2486, + "learning_rate": 8.014332695883487e-06, + "loss": 3.223, "step": 253800 }, { "epoch": 0.84, - "learning_rate": 7.929446073801594e-06, - "loss": 3.2567, + "learning_rate": 7.997789879766811e-06, + "loss": 3.2231, "step": 253900 }, { "epoch": 0.84, - "learning_rate": 7.912876340077216e-06, - "loss": 3.2568, + "learning_rate": 7.98124706365014e-06, + "loss": 3.2159, "step": 254000 }, { "epoch": 0.84, - "learning_rate": 7.896306606352837e-06, - "loss": 3.2586, + "learning_rate": 7.964704247533467e-06, + "loss": 3.2223, "step": 254100 }, { "epoch": 0.84, - "learning_rate": 7.879736872628457e-06, - "loss": 3.2645, + "learning_rate": 7.948161431416794e-06, + "loss": 3.2098, "step": 254200 }, { "epoch": 0.84, - "learning_rate": 7.863167138904077e-06, - "loss": 3.2574, + "learning_rate": 7.93161861530012e-06, + "loss": 3.2113, "step": 254300 }, { "epoch": 0.84, - "learning_rate": 7.8465974051797e-06, - "loss": 3.2594, + "learning_rate": 7.915075799183446e-06, + "loss": 3.2099, "step": 254400 }, { "epoch": 0.84, - "learning_rate": 7.83002767145532e-06, - "loss": 3.2492, + "learning_rate": 7.898532983066774e-06, + "loss": 3.2104, "step": 254500 }, { "epoch": 0.84, - "learning_rate": 7.813457937730941e-06, - "loss": 3.2527, + "learning_rate": 7.881990166950102e-06, + "loss": 3.2137, "step": 254600 }, { "epoch": 0.84, - "learning_rate": 7.796888204006563e-06, - "loss": 3.2576, + "learning_rate": 7.865447350833426e-06, + "loss": 3.217, "step": 254700 }, { "epoch": 0.84, - "learning_rate": 7.780318470282182e-06, - "loss": 3.2318, + "learning_rate": 7.848904534716754e-06, + "loss": 3.2263, "step": 254800 }, { "epoch": 0.84, - "learning_rate": 7.763748736557804e-06, - "loss": 3.261, + "learning_rate": 7.832361718600082e-06, + "loss": 3.2169, "step": 254900 }, { - "epoch": 0.85, - "learning_rate": 7.747179002833425e-06, - "loss": 3.2483, + "epoch": 0.84, + "learning_rate": 7.815818902483408e-06, + "loss": 3.2145, "step": 255000 }, { - "epoch": 0.85, - "learning_rate": 7.730609269109045e-06, - "loss": 3.2741, + "epoch": 0.84, + "learning_rate": 7.799276086366735e-06, + "loss": 3.2145, "step": 255100 }, { - "epoch": 0.85, - "learning_rate": 7.714039535384667e-06, - "loss": 3.252, + "epoch": 0.84, + "learning_rate": 7.782733270250061e-06, + "loss": 3.2234, "step": 255200 }, { - "epoch": 0.85, - "learning_rate": 7.697469801660288e-06, - "loss": 3.2647, + "epoch": 0.84, + "learning_rate": 7.766190454133389e-06, + "loss": 3.2194, "step": 255300 }, { "epoch": 0.85, - "learning_rate": 7.68090006793591e-06, - "loss": 3.2475, + "learning_rate": 7.749647638016715e-06, + "loss": 3.2206, "step": 255400 }, { "epoch": 0.85, - "learning_rate": 7.66433033421153e-06, - "loss": 3.2631, + "learning_rate": 7.733104821900041e-06, + "loss": 3.2108, "step": 255500 }, { "epoch": 0.85, - "learning_rate": 7.64776060048715e-06, - "loss": 3.2574, + "learning_rate": 7.71656200578337e-06, + "loss": 3.2199, "step": 255600 }, { "epoch": 0.85, - "learning_rate": 7.631190866762772e-06, - "loss": 3.2533, + "learning_rate": 7.700019189666695e-06, + "loss": 3.2261, "step": 255700 }, { "epoch": 0.85, - "learning_rate": 7.614621133038392e-06, - "loss": 3.2632, + "learning_rate": 7.683476373550022e-06, + "loss": 3.2189, "step": 255800 }, { "epoch": 0.85, - "learning_rate": 7.598051399314013e-06, - "loss": 3.2413, + "learning_rate": 7.66693355743335e-06, + "loss": 3.2146, "step": 255900 }, { "epoch": 0.85, - "learning_rate": 7.581481665589635e-06, - "loss": 3.2713, + "learning_rate": 7.650390741316676e-06, + "loss": 3.2167, "step": 256000 }, { "epoch": 0.85, - "learning_rate": 7.5649119318652555e-06, - "loss": 3.2554, + "learning_rate": 7.633847925200004e-06, + "loss": 3.2153, "step": 256100 }, { "epoch": 0.85, - "learning_rate": 7.548342198140876e-06, - "loss": 3.2495, + "learning_rate": 7.617305109083329e-06, + "loss": 3.2149, "step": 256200 }, { "epoch": 0.85, - "learning_rate": 7.531772464416498e-06, - "loss": 3.2518, + "learning_rate": 7.600762292966657e-06, + "loss": 3.224, "step": 256300 }, { "epoch": 0.85, - "learning_rate": 7.515202730692118e-06, - "loss": 3.2425, + "learning_rate": 7.584219476849984e-06, + "loss": 3.2284, "step": 256400 }, { "epoch": 0.85, - "learning_rate": 7.498632996967739e-06, - "loss": 3.2487, + "learning_rate": 7.5676766607333095e-06, + "loss": 3.2244, "step": 256500 }, { "epoch": 0.85, - "learning_rate": 7.48206326324336e-06, - "loss": 3.2548, + "learning_rate": 7.5511338446166366e-06, + "loss": 3.2194, "step": 256600 }, { "epoch": 0.85, - "learning_rate": 7.465493529518981e-06, - "loss": 3.263, + "learning_rate": 7.5345910284999645e-06, + "loss": 3.2141, "step": 256700 }, { "epoch": 0.85, - "learning_rate": 7.448923795794602e-06, - "loss": 3.2506, + "learning_rate": 7.5180482123832916e-06, + "loss": 3.2233, "step": 256800 }, { "epoch": 0.85, - "learning_rate": 7.432354062070222e-06, - "loss": 3.2519, + "learning_rate": 7.501505396266617e-06, + "loss": 3.2097, "step": 256900 }, { "epoch": 0.85, - "learning_rate": 7.4157843283458445e-06, - "loss": 3.2595, + "learning_rate": 7.484962580149944e-06, + "loss": 3.2061, "step": 257000 }, { "epoch": 0.85, - "learning_rate": 7.399214594621464e-06, - "loss": 3.2773, + "learning_rate": 7.468419764033271e-06, + "loss": 3.2181, "step": 257100 }, { "epoch": 0.85, - "learning_rate": 7.382644860897085e-06, - "loss": 3.2458, + "learning_rate": 7.451876947916599e-06, + "loss": 3.2228, "step": 257200 }, { "epoch": 0.85, - "learning_rate": 7.366075127172707e-06, - "loss": 3.2609, + "learning_rate": 7.435334131799924e-06, + "loss": 3.2272, "step": 257300 }, { "epoch": 0.85, - "learning_rate": 7.349505393448328e-06, - "loss": 3.2643, + "learning_rate": 7.4187913156832515e-06, + "loss": 3.235, "step": 257400 }, { "epoch": 0.85, - "learning_rate": 7.3329356597239485e-06, - "loss": 3.2548, + "learning_rate": 7.4022484995665785e-06, + "loss": 3.2288, "step": 257500 }, { "epoch": 0.85, - "learning_rate": 7.31636592599957e-06, - "loss": 3.2508, + "learning_rate": 7.385705683449905e-06, + "loss": 3.2106, "step": 257600 }, { "epoch": 0.85, - "learning_rate": 7.2997961922751906e-06, - "loss": 3.2571, + "learning_rate": 7.369162867333232e-06, + "loss": 3.2053, "step": 257700 }, { "epoch": 0.85, - "learning_rate": 7.283226458550811e-06, - "loss": 3.2534, + "learning_rate": 7.352620051216559e-06, + "loss": 3.215, "step": 257800 }, { "epoch": 0.85, - "learning_rate": 7.266656724826433e-06, - "loss": 3.2628, + "learning_rate": 7.336077235099886e-06, + "loss": 3.2134, "step": 257900 }, { "epoch": 0.85, - "learning_rate": 7.250086991102053e-06, - "loss": 3.2537, + "learning_rate": 7.319534418983212e-06, + "loss": 3.2145, "step": 258000 }, { - "epoch": 0.86, - "learning_rate": 7.233517257377674e-06, - "loss": 3.2551, + "epoch": 0.85, + "learning_rate": 7.302991602866539e-06, + "loss": 3.2059, "step": 258100 }, { - "epoch": 0.86, - "learning_rate": 7.2169475236532945e-06, - "loss": 3.2408, + "epoch": 0.85, + "learning_rate": 7.286448786749866e-06, + "loss": 3.2141, "step": 258200 }, { - "epoch": 0.86, - "learning_rate": 7.200377789928917e-06, - "loss": 3.2502, + "epoch": 0.85, + "learning_rate": 7.2699059706331935e-06, + "loss": 3.2285, "step": 258300 }, { - "epoch": 0.86, - "learning_rate": 7.1838080562045374e-06, - "loss": 3.2546, + "epoch": 0.85, + "learning_rate": 7.25336315451652e-06, + "loss": 3.2386, "step": 258400 }, { "epoch": 0.86, - "learning_rate": 7.167238322480157e-06, - "loss": 3.2692, + "learning_rate": 7.236820338399847e-06, + "loss": 3.2227, "step": 258500 }, { "epoch": 0.86, - "learning_rate": 7.1506685887557795e-06, - "loss": 3.2444, + "learning_rate": 7.220277522283174e-06, + "loss": 3.2176, "step": 258600 }, { "epoch": 0.86, - "learning_rate": 7.1340988550314e-06, - "loss": 3.2558, + "learning_rate": 7.203734706166501e-06, + "loss": 3.218, "step": 258700 }, { "epoch": 0.86, - "learning_rate": 7.117529121307021e-06, - "loss": 3.2513, + "learning_rate": 7.187191890049827e-06, + "loss": 3.2293, "step": 258800 }, { "epoch": 0.86, - "learning_rate": 7.100959387582642e-06, - "loss": 3.2618, + "learning_rate": 7.170649073933154e-06, + "loss": 3.2278, "step": 258900 }, { "epoch": 0.86, - "learning_rate": 7.084389653858263e-06, - "loss": 3.2684, + "learning_rate": 7.154106257816481e-06, + "loss": 3.2057, "step": 259000 }, { "epoch": 0.86, - "learning_rate": 7.0678199201338835e-06, - "loss": 3.2547, + "learning_rate": 7.1375634416998075e-06, + "loss": 3.2027, "step": 259100 }, { "epoch": 0.86, - "learning_rate": 7.051250186409506e-06, - "loss": 3.2509, + "learning_rate": 7.121020625583135e-06, + "loss": 3.2224, "step": 259200 }, { "epoch": 0.86, - "learning_rate": 7.0346804526851256e-06, - "loss": 3.2637, + "learning_rate": 7.104477809466462e-06, + "loss": 3.2053, "step": 259300 }, { "epoch": 0.86, - "learning_rate": 7.018110718960746e-06, - "loss": 3.2464, + "learning_rate": 7.087934993349789e-06, + "loss": 3.2237, "step": 259400 }, { "epoch": 0.86, - "learning_rate": 7.001540985236367e-06, - "loss": 3.2607, + "learning_rate": 7.071392177233115e-06, + "loss": 3.2155, "step": 259500 }, { "epoch": 0.86, - "learning_rate": 6.984971251511989e-06, - "loss": 3.2488, + "learning_rate": 7.054849361116442e-06, + "loss": 3.2046, "step": 259600 }, { "epoch": 0.86, - "learning_rate": 6.96840151778761e-06, - "loss": 3.2616, + "learning_rate": 7.038306544999769e-06, + "loss": 3.2, "step": 259700 }, { "epoch": 0.86, - "learning_rate": 6.95183178406323e-06, - "loss": 3.2526, + "learning_rate": 7.021763728883096e-06, + "loss": 3.2218, "step": 259800 }, { "epoch": 0.86, - "learning_rate": 6.935262050338852e-06, - "loss": 3.2454, + "learning_rate": 7.0052209127664224e-06, + "loss": 3.2195, "step": 259900 }, { "epoch": 0.86, - "learning_rate": 6.9186923166144724e-06, - "loss": 3.2558, + "learning_rate": 6.9886780966497495e-06, + "loss": 3.2079, "step": 260000 }, { "epoch": 0.86, - "learning_rate": 6.902122582890093e-06, - "loss": 3.2624, + "learning_rate": 6.972135280533077e-06, + "loss": 3.206, "step": 260100 }, { "epoch": 0.86, - "learning_rate": 6.8855528491657145e-06, - "loss": 3.2744, + "learning_rate": 6.955592464416402e-06, + "loss": 3.2265, "step": 260200 }, { "epoch": 0.86, - "learning_rate": 6.868983115441335e-06, - "loss": 3.2541, + "learning_rate": 6.939049648299729e-06, + "loss": 3.2202, "step": 260300 }, { "epoch": 0.86, - "learning_rate": 6.852413381716956e-06, - "loss": 3.244, + "learning_rate": 6.922506832183057e-06, + "loss": 3.2185, "step": 260400 }, { "epoch": 0.86, - "learning_rate": 6.835843647992578e-06, - "loss": 3.2503, + "learning_rate": 6.905964016066384e-06, + "loss": 3.2105, "step": 260500 }, { "epoch": 0.86, - "learning_rate": 6.819273914268199e-06, - "loss": 3.2638, + "learning_rate": 6.889421199949709e-06, + "loss": 3.2258, "step": 260600 }, { "epoch": 0.86, - "learning_rate": 6.8027041805438185e-06, - "loss": 3.254, + "learning_rate": 6.8728783838330365e-06, + "loss": 3.2235, "step": 260700 }, { "epoch": 0.86, - "learning_rate": 6.786134446819439e-06, - "loss": 3.2487, + "learning_rate": 6.856335567716364e-06, + "loss": 3.2096, "step": 260800 }, { "epoch": 0.86, - "learning_rate": 6.769564713095061e-06, - "loss": 3.2601, + "learning_rate": 6.8397927515996915e-06, + "loss": 3.2049, "step": 260900 }, { "epoch": 0.86, - "learning_rate": 6.752994979370682e-06, - "loss": 3.2533, + "learning_rate": 6.823249935483017e-06, + "loss": 3.2196, "step": 261000 }, { - "epoch": 0.87, - "learning_rate": 6.736425245646303e-06, - "loss": 3.249, + "epoch": 0.86, + "learning_rate": 6.806707119366344e-06, + "loss": 3.229, "step": 261100 }, { - "epoch": 0.87, - "learning_rate": 6.719855511921924e-06, - "loss": 3.2583, + "epoch": 0.86, + "learning_rate": 6.790164303249672e-06, + "loss": 3.2036, "step": 261200 }, { - "epoch": 0.87, - "learning_rate": 6.703285778197545e-06, - "loss": 3.2441, + "epoch": 0.86, + "learning_rate": 6.773621487132997e-06, + "loss": 3.2263, "step": 261300 }, { - "epoch": 0.87, - "learning_rate": 6.686716044473165e-06, - "loss": 3.2649, + "epoch": 0.86, + "learning_rate": 6.757078671016324e-06, + "loss": 3.2097, "step": 261400 }, { "epoch": 0.87, - "learning_rate": 6.670146310748787e-06, - "loss": 3.2538, + "learning_rate": 6.740535854899651e-06, + "loss": 3.2201, "step": 261500 }, { "epoch": 0.87, - "learning_rate": 6.6535765770244074e-06, - "loss": 3.2581, + "learning_rate": 6.7239930387829785e-06, + "loss": 3.2176, "step": 261600 }, { "epoch": 0.87, - "learning_rate": 6.637006843300028e-06, - "loss": 3.2575, + "learning_rate": 6.707450222666305e-06, + "loss": 3.2269, "step": 261700 }, { "epoch": 0.87, - "learning_rate": 6.62043710957565e-06, - "loss": 3.256, + "learning_rate": 6.690907406549632e-06, + "loss": 3.2038, "step": 261800 }, { "epoch": 0.87, - "learning_rate": 6.603867375851271e-06, - "loss": 3.2603, + "learning_rate": 6.674364590432959e-06, + "loss": 3.1985, "step": 261900 }, { "epoch": 0.87, - "learning_rate": 6.587297642126891e-06, - "loss": 3.2464, + "learning_rate": 6.657821774316286e-06, + "loss": 3.2328, "step": 262000 }, { "epoch": 0.87, - "learning_rate": 6.570727908402511e-06, - "loss": 3.2558, + "learning_rate": 6.641278958199612e-06, + "loss": 3.2153, "step": 262100 }, { "epoch": 0.87, - "learning_rate": 6.554158174678134e-06, - "loss": 3.2585, + "learning_rate": 6.624736142082939e-06, + "loss": 3.2044, "step": 262200 }, { "epoch": 0.87, - "learning_rate": 6.537588440953754e-06, - "loss": 3.2559, + "learning_rate": 6.608193325966266e-06, + "loss": 3.2121, "step": 262300 }, { "epoch": 0.87, - "learning_rate": 6.521018707229375e-06, - "loss": 3.2551, + "learning_rate": 6.591650509849593e-06, + "loss": 3.2118, "step": 262400 }, { "epoch": 0.87, - "learning_rate": 6.504448973504996e-06, - "loss": 3.2528, + "learning_rate": 6.57510769373292e-06, + "loss": 3.2136, "step": 262500 }, { "epoch": 0.87, - "learning_rate": 6.487879239780617e-06, - "loss": 3.2655, + "learning_rate": 6.558564877616247e-06, + "loss": 3.2199, "step": 262600 }, { "epoch": 0.87, - "learning_rate": 6.471309506056238e-06, - "loss": 3.2426, + "learning_rate": 6.542022061499574e-06, + "loss": 3.233, "step": 262700 }, { "epoch": 0.87, - "learning_rate": 6.454739772331859e-06, - "loss": 3.2412, + "learning_rate": 6.5254792453829e-06, + "loss": 3.2242, "step": 262800 }, { "epoch": 0.87, - "learning_rate": 6.43817003860748e-06, - "loss": 3.2536, + "learning_rate": 6.508936429266227e-06, + "loss": 3.1949, "step": 262900 }, { "epoch": 0.87, - "learning_rate": 6.4216003048831e-06, - "loss": 3.2496, + "learning_rate": 6.492393613149554e-06, + "loss": 3.2118, "step": 263000 }, { "epoch": 0.87, - "learning_rate": 6.405030571158723e-06, - "loss": 3.2526, + "learning_rate": 6.475850797032881e-06, + "loss": 3.2181, "step": 263100 }, { "epoch": 0.87, - "learning_rate": 6.388460837434343e-06, - "loss": 3.242, + "learning_rate": 6.4593079809162075e-06, + "loss": 3.2172, "step": 263200 }, { "epoch": 0.87, - "learning_rate": 6.371891103709964e-06, - "loss": 3.2421, + "learning_rate": 6.4427651647995345e-06, + "loss": 3.2177, "step": 263300 }, { "epoch": 0.87, - "learning_rate": 6.355321369985584e-06, - "loss": 3.2572, + "learning_rate": 6.426222348682862e-06, + "loss": 3.2102, "step": 263400 }, { "epoch": 0.87, - "learning_rate": 6.338751636261206e-06, - "loss": 3.2512, + "learning_rate": 6.409679532566189e-06, + "loss": 3.2177, "step": 263500 }, { "epoch": 0.87, - "learning_rate": 6.322181902536827e-06, - "loss": 3.249, + "learning_rate": 6.393136716449515e-06, + "loss": 3.2259, "step": 263600 }, { "epoch": 0.87, - "learning_rate": 6.305612168812447e-06, - "loss": 3.2579, + "learning_rate": 6.376593900332842e-06, + "loss": 3.2305, "step": 263700 }, { "epoch": 0.87, - "learning_rate": 6.289042435088069e-06, - "loss": 3.2446, + "learning_rate": 6.360051084216169e-06, + "loss": 3.2247, "step": 263800 }, { "epoch": 0.87, - "learning_rate": 6.272472701363689e-06, - "loss": 3.2631, + "learning_rate": 6.3435082680994944e-06, + "loss": 3.2086, "step": 263900 }, { "epoch": 0.87, - "learning_rate": 6.25590296763931e-06, - "loss": 3.253, + "learning_rate": 6.326965451982822e-06, + "loss": 3.2164, "step": 264000 }, { - "epoch": 0.88, - "learning_rate": 6.239333233914931e-06, - "loss": 3.25, + "epoch": 0.87, + "learning_rate": 6.3104226358661494e-06, + "loss": 3.2158, "step": 264100 }, { - "epoch": 0.88, - "learning_rate": 6.222763500190552e-06, - "loss": 3.2476, + "epoch": 0.87, + "learning_rate": 6.2938798197494765e-06, + "loss": 3.2034, "step": 264200 }, { - "epoch": 0.88, - "learning_rate": 6.206193766466173e-06, - "loss": 3.2657, + "epoch": 0.87, + "learning_rate": 6.277337003632802e-06, + "loss": 3.2282, "step": 264300 }, { - "epoch": 0.88, - "learning_rate": 6.189624032741794e-06, - "loss": 3.2558, + "epoch": 0.87, + "learning_rate": 6.260794187516129e-06, + "loss": 3.2125, "step": 264400 }, { "epoch": 0.88, - "learning_rate": 6.173054299017416e-06, - "loss": 3.2506, + "learning_rate": 6.244251371399457e-06, + "loss": 3.2086, "step": 264500 }, { "epoch": 0.88, - "learning_rate": 6.156484565293036e-06, - "loss": 3.2467, + "learning_rate": 6.227708555282783e-06, + "loss": 3.2104, "step": 264600 }, { "epoch": 0.88, - "learning_rate": 6.139914831568657e-06, - "loss": 3.25, + "learning_rate": 6.21116573916611e-06, + "loss": 3.2256, "step": 264700 }, { "epoch": 0.88, - "learning_rate": 6.1233450978442774e-06, - "loss": 3.2539, + "learning_rate": 6.1946229230494364e-06, + "loss": 3.2317, "step": 264800 }, { "epoch": 0.88, - "learning_rate": 6.106775364119899e-06, - "loss": 3.2447, + "learning_rate": 6.178080106932764e-06, + "loss": 3.216, "step": 264900 }, { "epoch": 0.88, - "learning_rate": 6.09020563039552e-06, - "loss": 3.2487, + "learning_rate": 6.161537290816091e-06, + "loss": 3.223, "step": 265000 }, { "epoch": 0.88, - "learning_rate": 6.07363589667114e-06, - "loss": 3.2412, + "learning_rate": 6.144994474699417e-06, + "loss": 3.2177, "step": 265100 }, { "epoch": 0.88, - "learning_rate": 6.057066162946762e-06, - "loss": 3.2428, + "learning_rate": 6.128451658582744e-06, + "loss": 3.2064, "step": 265200 }, { "epoch": 0.88, - "learning_rate": 6.040496429222383e-06, - "loss": 3.2591, + "learning_rate": 6.111908842466071e-06, + "loss": 3.2168, "step": 265300 }, { "epoch": 0.88, - "learning_rate": 6.023926695498004e-06, - "loss": 3.2555, + "learning_rate": 6.095366026349398e-06, + "loss": 3.2267, "step": 265400 }, { "epoch": 0.88, - "learning_rate": 6.007356961773624e-06, - "loss": 3.244, + "learning_rate": 6.078823210232724e-06, + "loss": 3.2146, "step": 265500 }, { "epoch": 0.88, - "learning_rate": 5.990787228049245e-06, - "loss": 3.2523, + "learning_rate": 6.062280394116051e-06, + "loss": 3.2213, "step": 265600 }, { "epoch": 0.88, - "learning_rate": 5.974217494324866e-06, - "loss": 3.2583, + "learning_rate": 6.045737577999378e-06, + "loss": 3.2195, "step": 265700 }, { "epoch": 0.88, - "learning_rate": 5.957647760600488e-06, - "loss": 3.2514, + "learning_rate": 6.0291947618827055e-06, + "loss": 3.2131, "step": 265800 }, { "epoch": 0.88, - "learning_rate": 5.9410780268761085e-06, - "loss": 3.257, + "learning_rate": 6.012651945766032e-06, + "loss": 3.229, "step": 265900 }, { "epoch": 0.88, - "learning_rate": 5.924508293151729e-06, - "loss": 3.2544, + "learning_rate": 5.996109129649359e-06, + "loss": 3.2099, "step": 266000 }, { "epoch": 0.88, - "learning_rate": 5.90793855942735e-06, - "loss": 3.2597, + "learning_rate": 5.979566313532686e-06, + "loss": 3.2208, "step": 266100 }, { "epoch": 0.88, - "learning_rate": 5.891368825702971e-06, - "loss": 3.2477, + "learning_rate": 5.963023497416012e-06, + "loss": 3.2222, "step": 266200 }, { "epoch": 0.88, - "learning_rate": 5.874799091978593e-06, - "loss": 3.2501, + "learning_rate": 5.946480681299339e-06, + "loss": 3.2234, "step": 266300 }, { "epoch": 0.88, - "learning_rate": 5.858229358254213e-06, - "loss": 3.2673, + "learning_rate": 5.929937865182665e-06, + "loss": 3.21, "step": 266400 }, { "epoch": 0.88, - "learning_rate": 5.841659624529834e-06, - "loss": 3.2465, + "learning_rate": 5.913395049065993e-06, + "loss": 3.2173, "step": 266500 }, { "epoch": 0.88, - "learning_rate": 5.825089890805455e-06, - "loss": 3.2702, + "learning_rate": 5.8968522329493196e-06, + "loss": 3.2111, "step": 266600 }, { "epoch": 0.88, - "learning_rate": 5.808520157081076e-06, - "loss": 3.258, + "learning_rate": 5.880309416832647e-06, + "loss": 3.2074, "step": 266700 }, { "epoch": 0.88, - "learning_rate": 5.7919504233566975e-06, - "loss": 3.2592, + "learning_rate": 5.863766600715973e-06, + "loss": 3.2177, "step": 266800 }, { "epoch": 0.88, - "learning_rate": 5.775380689632317e-06, - "loss": 3.2501, + "learning_rate": 5.847223784599301e-06, + "loss": 3.2001, "step": 266900 }, { "epoch": 0.88, - "learning_rate": 5.758810955907939e-06, - "loss": 3.2459, + "learning_rate": 5.830680968482627e-06, + "loss": 3.2274, "step": 267000 }, { - "epoch": 0.89, - "learning_rate": 5.74224122218356e-06, - "loss": 3.2495, + "epoch": 0.88, + "learning_rate": 5.814138152365954e-06, + "loss": 3.2166, "step": 267100 }, { - "epoch": 0.89, - "learning_rate": 5.725671488459181e-06, - "loss": 3.2561, + "epoch": 0.88, + "learning_rate": 5.79759533624928e-06, + "loss": 3.2162, "step": 267200 }, { - "epoch": 0.89, - "learning_rate": 5.709101754734801e-06, - "loss": 3.2444, + "epoch": 0.88, + "learning_rate": 5.781052520132607e-06, + "loss": 3.2159, "step": 267300 }, { - "epoch": 0.89, - "learning_rate": 5.692532021010422e-06, - "loss": 3.2715, + "epoch": 0.88, + "learning_rate": 5.7645097040159345e-06, + "loss": 3.2237, "step": 267400 }, { "epoch": 0.89, - "learning_rate": 5.6759622872860435e-06, - "loss": 3.2521, + "learning_rate": 5.747966887899261e-06, + "loss": 3.2126, "step": 267500 }, { "epoch": 0.89, - "learning_rate": 5.659392553561665e-06, - "loss": 3.2609, + "learning_rate": 5.731424071782588e-06, + "loss": 3.2088, "step": 267600 }, { "epoch": 0.89, - "learning_rate": 5.642822819837286e-06, - "loss": 3.2641, + "learning_rate": 5.714881255665915e-06, + "loss": 3.2136, "step": 267700 }, { "epoch": 0.89, - "learning_rate": 5.626253086112906e-06, - "loss": 3.2566, + "learning_rate": 5.698338439549242e-06, + "loss": 3.2237, "step": 267800 }, { "epoch": 0.89, - "learning_rate": 5.609683352388528e-06, - "loss": 3.2538, + "learning_rate": 5.681795623432568e-06, + "loss": 3.2079, "step": 267900 }, { "epoch": 0.89, - "learning_rate": 5.593113618664148e-06, - "loss": 3.2493, + "learning_rate": 5.665252807315895e-06, + "loss": 3.232, "step": 268000 }, { "epoch": 0.89, - "learning_rate": 5.57654388493977e-06, - "loss": 3.2716, + "learning_rate": 5.648709991199222e-06, + "loss": 3.2233, "step": 268100 }, { "epoch": 0.89, - "learning_rate": 5.5599741512153895e-06, - "loss": 3.2475, + "learning_rate": 5.632167175082549e-06, + "loss": 3.2131, "step": 268200 }, { "epoch": 0.89, - "learning_rate": 5.543404417491011e-06, - "loss": 3.247, + "learning_rate": 5.615624358965876e-06, + "loss": 3.2187, "step": 268300 }, { "epoch": 0.89, - "learning_rate": 5.5268346837666325e-06, - "loss": 3.2506, + "learning_rate": 5.599081542849203e-06, + "loss": 3.2156, "step": 268400 }, { "epoch": 0.89, - "learning_rate": 5.510264950042253e-06, - "loss": 3.2475, + "learning_rate": 5.58253872673253e-06, + "loss": 3.2185, "step": 268500 }, { "epoch": 0.89, - "learning_rate": 5.493695216317874e-06, - "loss": 3.2534, + "learning_rate": 5.565995910615857e-06, + "loss": 3.236, "step": 268600 }, { "epoch": 0.89, - "learning_rate": 5.477125482593494e-06, - "loss": 3.2625, + "learning_rate": 5.549453094499183e-06, + "loss": 3.2232, "step": 268700 }, { "epoch": 0.89, - "learning_rate": 5.460555748869116e-06, - "loss": 3.2509, + "learning_rate": 5.532910278382509e-06, + "loss": 3.2201, "step": 268800 }, { "epoch": 0.89, - "learning_rate": 5.443986015144737e-06, - "loss": 3.2337, + "learning_rate": 5.516367462265836e-06, + "loss": 3.2054, "step": 268900 }, { "epoch": 0.89, - "learning_rate": 5.427416281420358e-06, - "loss": 3.2574, + "learning_rate": 5.4998246461491634e-06, + "loss": 3.2212, "step": 269000 }, { "epoch": 0.89, - "learning_rate": 5.4108465476959785e-06, - "loss": 3.2448, + "learning_rate": 5.4832818300324905e-06, + "loss": 3.2065, "step": 269100 }, { "epoch": 0.89, - "learning_rate": 5.3942768139716e-06, - "loss": 3.2639, + "learning_rate": 5.466739013915817e-06, + "loss": 3.2176, "step": 269200 }, { "epoch": 0.89, - "learning_rate": 5.377707080247221e-06, - "loss": 3.2423, + "learning_rate": 5.450196197799144e-06, + "loss": 3.2282, "step": 269300 }, { "epoch": 0.89, - "learning_rate": 5.361137346522842e-06, - "loss": 3.2499, + "learning_rate": 5.433653381682471e-06, + "loss": 3.2042, "step": 269400 }, { "epoch": 0.89, - "learning_rate": 5.344567612798463e-06, - "loss": 3.2569, + "learning_rate": 5.417110565565798e-06, + "loss": 3.2102, "step": 269500 }, { "epoch": 0.89, - "learning_rate": 5.327997879074083e-06, - "loss": 3.247, + "learning_rate": 5.400567749449124e-06, + "loss": 3.2231, "step": 269600 }, { "epoch": 0.89, - "learning_rate": 5.311428145349705e-06, - "loss": 3.2541, + "learning_rate": 5.384024933332451e-06, + "loss": 3.2141, "step": 269700 }, { "epoch": 0.89, - "learning_rate": 5.294858411625325e-06, - "loss": 3.2668, + "learning_rate": 5.367482117215778e-06, + "loss": 3.2144, "step": 269800 }, { "epoch": 0.89, - "learning_rate": 5.278288677900947e-06, - "loss": 3.2492, + "learning_rate": 5.350939301099105e-06, + "loss": 3.2256, "step": 269900 }, { "epoch": 0.89, - "learning_rate": 5.261718944176567e-06, - "loss": 3.2529, + "learning_rate": 5.334396484982432e-06, + "loss": 3.223, "step": 270000 }, { - "epoch": 0.9, - "learning_rate": 5.245149210452188e-06, - "loss": 3.2539, + "epoch": 0.89, + "learning_rate": 5.317853668865759e-06, + "loss": 3.2048, "step": 270100 }, { - "epoch": 0.9, - "learning_rate": 5.2285794767278096e-06, - "loss": 3.263, + "epoch": 0.89, + "learning_rate": 5.301310852749086e-06, + "loss": 3.2175, "step": 270200 }, { - "epoch": 0.9, - "learning_rate": 5.21200974300343e-06, - "loss": 3.2425, + "epoch": 0.89, + "learning_rate": 5.284768036632412e-06, + "loss": 3.2169, "step": 270300 }, { - "epoch": 0.9, - "learning_rate": 5.195440009279051e-06, - "loss": 3.2496, + "epoch": 0.89, + "learning_rate": 5.268225220515739e-06, + "loss": 3.2184, "step": 270400 }, { - "epoch": 0.9, - "learning_rate": 5.178870275554672e-06, - "loss": 3.2654, + "epoch": 0.89, + "learning_rate": 5.251682404399065e-06, + "loss": 3.2173, "step": 270500 }, { "epoch": 0.9, - "learning_rate": 5.162300541830293e-06, - "loss": 3.2359, + "learning_rate": 5.235139588282393e-06, + "loss": 3.22, "step": 270600 }, { "epoch": 0.9, - "learning_rate": 5.145730808105914e-06, - "loss": 3.2727, + "learning_rate": 5.2185967721657195e-06, + "loss": 3.2237, "step": 270700 }, { "epoch": 0.9, - "learning_rate": 5.129161074381535e-06, - "loss": 3.256, + "learning_rate": 5.2020539560490466e-06, + "loss": 3.1947, "step": 270800 }, { "epoch": 0.9, - "learning_rate": 5.112591340657156e-06, - "loss": 3.2564, + "learning_rate": 5.185511139932373e-06, + "loss": 3.2158, "step": 270900 }, { "epoch": 0.9, - "learning_rate": 5.096021606932777e-06, - "loss": 3.2623, + "learning_rate": 5.168968323815701e-06, + "loss": 3.2122, "step": 271000 }, { "epoch": 0.9, - "learning_rate": 5.079451873208398e-06, - "loss": 3.2435, + "learning_rate": 5.152425507699027e-06, + "loss": 3.2191, "step": 271100 }, { "epoch": 0.9, - "learning_rate": 5.062882139484019e-06, - "loss": 3.2518, + "learning_rate": 5.135882691582353e-06, + "loss": 3.2119, "step": 271200 }, { "epoch": 0.9, - "learning_rate": 5.04631240575964e-06, - "loss": 3.2454, + "learning_rate": 5.11933987546568e-06, + "loss": 3.2334, "step": 271300 }, { "epoch": 0.9, - "learning_rate": 5.02974267203526e-06, - "loss": 3.2657, + "learning_rate": 5.102797059349007e-06, + "loss": 3.2109, "step": 271400 }, { "epoch": 0.9, - "learning_rate": 5.013172938310882e-06, - "loss": 3.2521, + "learning_rate": 5.086254243232334e-06, + "loss": 3.2182, "step": 271500 }, { "epoch": 0.9, - "learning_rate": 4.9966032045865025e-06, - "loss": 3.2494, + "learning_rate": 5.069711427115661e-06, + "loss": 3.231, "step": 271600 }, { "epoch": 0.9, - "learning_rate": 4.980033470862124e-06, - "loss": 3.2496, + "learning_rate": 5.053168610998988e-06, + "loss": 3.2235, "step": 271700 }, { "epoch": 0.9, - "learning_rate": 4.9634637371377446e-06, - "loss": 3.2596, + "learning_rate": 5.036625794882315e-06, + "loss": 3.215, "step": 271800 }, { "epoch": 0.9, - "learning_rate": 4.946894003413365e-06, - "loss": 3.2414, + "learning_rate": 5.020082978765642e-06, + "loss": 3.2117, "step": 271900 }, { "epoch": 0.9, - "learning_rate": 4.930324269688987e-06, - "loss": 3.2573, + "learning_rate": 5.003540162648968e-06, + "loss": 3.2244, "step": 272000 }, { "epoch": 0.9, - "learning_rate": 4.913754535964607e-06, - "loss": 3.2522, + "learning_rate": 4.986997346532295e-06, + "loss": 3.2114, "step": 272100 }, { "epoch": 0.9, - "learning_rate": 4.897184802240228e-06, - "loss": 3.2581, + "learning_rate": 4.970454530415622e-06, + "loss": 3.2127, "step": 272200 }, { "epoch": 0.9, - "learning_rate": 4.880615068515849e-06, - "loss": 3.2549, + "learning_rate": 4.9539117142989485e-06, + "loss": 3.2133, "step": 272300 }, { "epoch": 0.9, - "learning_rate": 4.86404533479147e-06, - "loss": 3.2502, + "learning_rate": 4.9373688981822755e-06, + "loss": 3.2087, "step": 272400 }, { "epoch": 0.9, - "learning_rate": 4.8474756010670914e-06, - "loss": 3.2504, + "learning_rate": 4.920826082065602e-06, + "loss": 3.2238, "step": 272500 }, { "epoch": 0.9, - "learning_rate": 4.830905867342712e-06, - "loss": 3.2536, + "learning_rate": 4.90428326594893e-06, + "loss": 3.2152, "step": 272600 }, { "epoch": 0.9, - "learning_rate": 4.814336133618333e-06, - "loss": 3.2529, + "learning_rate": 4.887740449832256e-06, + "loss": 3.2088, "step": 272700 }, { "epoch": 0.9, - "learning_rate": 4.797766399893954e-06, - "loss": 3.2578, + "learning_rate": 4.871197633715583e-06, + "loss": 3.2056, "step": 272800 }, { "epoch": 0.9, - "learning_rate": 4.781196666169575e-06, - "loss": 3.2505, + "learning_rate": 4.854654817598909e-06, + "loss": 3.221, "step": 272900 }, { "epoch": 0.9, - "learning_rate": 4.764626932445196e-06, - "loss": 3.2568, + "learning_rate": 4.838112001482237e-06, + "loss": 3.2242, "step": 273000 }, { - "epoch": 0.91, - "learning_rate": 4.748057198720817e-06, - "loss": 3.265, + "epoch": 0.9, + "learning_rate": 4.821569185365563e-06, + "loss": 3.2055, "step": 273100 }, { - "epoch": 0.91, - "learning_rate": 4.7314874649964375e-06, - "loss": 3.2586, + "epoch": 0.9, + "learning_rate": 4.8050263692488905e-06, + "loss": 3.2109, "step": 273200 }, { - "epoch": 0.91, - "learning_rate": 4.714917731272059e-06, - "loss": 3.2575, + "epoch": 0.9, + "learning_rate": 4.788483553132217e-06, + "loss": 3.2133, "step": 273300 }, { - "epoch": 0.91, - "learning_rate": 4.6983479975476796e-06, - "loss": 3.2483, + "epoch": 0.9, + "learning_rate": 4.771940737015544e-06, + "loss": 3.2159, "step": 273400 }, { - "epoch": 0.91, - "learning_rate": 4.6817782638233e-06, - "loss": 3.2363, + "epoch": 0.9, + "learning_rate": 4.755397920898871e-06, + "loss": 3.2194, "step": 273500 }, { "epoch": 0.91, - "learning_rate": 4.665208530098922e-06, - "loss": 3.2531, + "learning_rate": 4.738855104782197e-06, + "loss": 3.2264, "step": 273600 }, { "epoch": 0.91, - "learning_rate": 4.648638796374542e-06, - "loss": 3.2443, + "learning_rate": 4.722312288665524e-06, + "loss": 3.2145, "step": 273700 }, { "epoch": 0.91, - "learning_rate": 4.632069062650164e-06, - "loss": 3.2613, + "learning_rate": 4.705769472548851e-06, + "loss": 3.201, "step": 273800 }, { "epoch": 0.91, - "learning_rate": 4.615499328925784e-06, - "loss": 3.2519, + "learning_rate": 4.689226656432178e-06, + "loss": 3.2046, "step": 273900 }, { "epoch": 0.91, - "learning_rate": 4.598929595201405e-06, - "loss": 3.2395, + "learning_rate": 4.6726838403155045e-06, + "loss": 3.2163, "step": 274000 }, { "epoch": 0.91, - "learning_rate": 4.5823598614770264e-06, - "loss": 3.2567, + "learning_rate": 4.656141024198832e-06, + "loss": 3.2124, "step": 274100 }, { "epoch": 0.91, - "learning_rate": 4.565790127752647e-06, - "loss": 3.244, + "learning_rate": 4.639598208082159e-06, + "loss": 3.2133, "step": 274200 }, { "epoch": 0.91, - "learning_rate": 4.5492203940282685e-06, - "loss": 3.2568, + "learning_rate": 4.623055391965486e-06, + "loss": 3.2204, "step": 274300 }, { "epoch": 0.91, - "learning_rate": 4.532650660303889e-06, - "loss": 3.2599, + "learning_rate": 4.606512575848812e-06, + "loss": 3.2195, "step": 274400 }, { "epoch": 0.91, - "learning_rate": 4.51608092657951e-06, - "loss": 3.2632, + "learning_rate": 4.589969759732139e-06, + "loss": 3.1909, "step": 274500 }, { "epoch": 0.91, - "learning_rate": 4.499511192855131e-06, - "loss": 3.2513, + "learning_rate": 4.573426943615466e-06, + "loss": 3.2171, "step": 274600 }, { "epoch": 0.91, - "learning_rate": 4.482941459130752e-06, - "loss": 3.2539, + "learning_rate": 4.556884127498793e-06, + "loss": 3.2185, "step": 274700 }, { "epoch": 0.91, - "learning_rate": 4.466371725406373e-06, - "loss": 3.2443, + "learning_rate": 4.5403413113821194e-06, + "loss": 3.2152, "step": 274800 }, { "epoch": 0.91, - "learning_rate": 4.449801991681994e-06, - "loss": 3.2453, + "learning_rate": 4.523798495265446e-06, + "loss": 3.2114, "step": 274900 }, { "epoch": 0.91, - "learning_rate": 4.4332322579576146e-06, - "loss": 3.2433, + "learning_rate": 4.507255679148773e-06, + "loss": 3.225, "step": 275000 }, { "epoch": 0.91, - "learning_rate": 4.416662524233236e-06, - "loss": 3.2525, + "learning_rate": 4.4907128630321e-06, + "loss": 3.2152, "step": 275100 }, { "epoch": 0.91, - "learning_rate": 4.400092790508857e-06, - "loss": 3.2611, + "learning_rate": 4.474170046915427e-06, + "loss": 3.1953, "step": 275200 }, { "epoch": 0.91, - "learning_rate": 4.383523056784477e-06, - "loss": 3.2567, + "learning_rate": 4.457627230798753e-06, + "loss": 3.2129, "step": 275300 }, { "epoch": 0.91, - "learning_rate": 4.366953323060099e-06, - "loss": 3.2598, + "learning_rate": 4.44108441468208e-06, + "loss": 3.2131, "step": 275400 }, { "epoch": 0.91, - "learning_rate": 4.350383589335719e-06, - "loss": 3.2616, + "learning_rate": 4.424541598565407e-06, + "loss": 3.2156, "step": 275500 }, { "epoch": 0.91, - "learning_rate": 4.333813855611341e-06, - "loss": 3.2537, + "learning_rate": 4.407998782448734e-06, + "loss": 3.2232, "step": 275600 }, { "epoch": 0.91, - "learning_rate": 4.3172441218869614e-06, - "loss": 3.2519, + "learning_rate": 4.3914559663320606e-06, + "loss": 3.2089, "step": 275700 }, { "epoch": 0.91, - "learning_rate": 4.300674388162582e-06, - "loss": 3.2588, + "learning_rate": 4.374913150215388e-06, + "loss": 3.2168, "step": 275800 }, { "epoch": 0.91, - "learning_rate": 4.2841046544382035e-06, - "loss": 3.2512, + "learning_rate": 4.358370334098715e-06, + "loss": 3.2052, "step": 275900 }, { "epoch": 0.91, - "learning_rate": 4.267534920713824e-06, - "loss": 3.2612, + "learning_rate": 4.341827517982041e-06, + "loss": 3.2259, "step": 276000 }, { "epoch": 0.91, - "learning_rate": 4.250965186989446e-06, - "loss": 3.2596, + "learning_rate": 4.325284701865368e-06, + "loss": 3.2029, "step": 276100 }, { - "epoch": 0.92, - "learning_rate": 4.234395453265066e-06, - "loss": 3.2445, + "epoch": 0.91, + "learning_rate": 4.308741885748695e-06, + "loss": 3.2275, "step": 276200 }, { - "epoch": 0.92, - "learning_rate": 4.217825719540687e-06, - "loss": 3.2519, + "epoch": 0.91, + "learning_rate": 4.292199069632022e-06, + "loss": 3.2216, "step": 276300 }, { - "epoch": 0.92, - "learning_rate": 4.201255985816308e-06, - "loss": 3.2559, + "epoch": 0.91, + "learning_rate": 4.275656253515348e-06, + "loss": 3.2145, "step": 276400 }, { - "epoch": 0.92, - "learning_rate": 4.184686252091929e-06, - "loss": 3.2577, + "epoch": 0.91, + "learning_rate": 4.2591134373986755e-06, + "loss": 3.212, "step": 276500 }, { "epoch": 0.92, - "learning_rate": 4.1681165183675496e-06, - "loss": 3.2445, + "learning_rate": 4.242570621282002e-06, + "loss": 3.2195, "step": 276600 }, { "epoch": 0.92, - "learning_rate": 4.151546784643171e-06, - "loss": 3.2335, + "learning_rate": 4.22602780516533e-06, + "loss": 3.2062, "step": 276700 }, { "epoch": 0.92, - "learning_rate": 4.134977050918792e-06, - "loss": 3.2363, + "learning_rate": 4.209484989048656e-06, + "loss": 3.2096, "step": 276800 }, { "epoch": 0.92, - "learning_rate": 4.118407317194413e-06, - "loss": 3.2511, + "learning_rate": 4.192942172931983e-06, + "loss": 3.2087, "step": 276900 }, { "epoch": 0.92, - "learning_rate": 4.101837583470034e-06, - "loss": 3.2516, + "learning_rate": 4.176399356815309e-06, + "loss": 3.203, "step": 277000 }, { "epoch": 0.92, - "learning_rate": 4.085267849745654e-06, - "loss": 3.2581, + "learning_rate": 4.159856540698637e-06, + "loss": 3.2134, "step": 277100 }, { "epoch": 0.92, - "learning_rate": 4.068698116021276e-06, - "loss": 3.2456, + "learning_rate": 4.143313724581963e-06, + "loss": 3.2182, "step": 277200 }, { "epoch": 0.92, - "learning_rate": 4.0521283822968964e-06, - "loss": 3.2417, + "learning_rate": 4.1267709084652896e-06, + "loss": 3.208, "step": 277300 }, { "epoch": 0.92, - "learning_rate": 4.035558648572518e-06, - "loss": 3.251, + "learning_rate": 4.110228092348617e-06, + "loss": 3.2121, "step": 277400 }, { "epoch": 0.92, - "learning_rate": 4.0189889148481385e-06, - "loss": 3.2562, + "learning_rate": 4.093685276231944e-06, + "loss": 3.2105, "step": 277500 }, { "epoch": 0.92, - "learning_rate": 4.002419181123759e-06, - "loss": 3.2414, + "learning_rate": 4.077142460115271e-06, + "loss": 3.2228, "step": 277600 }, { "epoch": 0.92, - "learning_rate": 3.985849447399381e-06, - "loss": 3.2486, + "learning_rate": 4.060599643998597e-06, + "loss": 3.2149, "step": 277700 }, { "epoch": 0.92, - "learning_rate": 3.969279713675001e-06, - "loss": 3.2617, + "learning_rate": 4.044056827881924e-06, + "loss": 3.2107, "step": 277800 }, { "epoch": 0.92, - "learning_rate": 3.952709979950623e-06, - "loss": 3.2569, + "learning_rate": 4.027514011765251e-06, + "loss": 3.2071, "step": 277900 }, { "epoch": 0.92, - "learning_rate": 3.936140246226243e-06, - "loss": 3.2565, + "learning_rate": 4.010971195648578e-06, + "loss": 3.2262, "step": 278000 }, { "epoch": 0.92, - "learning_rate": 3.919570512501864e-06, - "loss": 3.2482, + "learning_rate": 3.9944283795319045e-06, + "loss": 3.2158, "step": 278100 }, { "epoch": 0.92, - "learning_rate": 3.903000778777485e-06, - "loss": 3.2436, + "learning_rate": 3.9778855634152315e-06, + "loss": 3.199, "step": 278200 }, { "epoch": 0.92, - "learning_rate": 3.886431045053107e-06, - "loss": 3.2534, + "learning_rate": 3.961342747298559e-06, + "loss": 3.2057, "step": 278300 }, { "epoch": 0.92, - "learning_rate": 3.869861311328727e-06, - "loss": 3.2516, + "learning_rate": 3.944799931181886e-06, + "loss": 3.23, "step": 278400 }, { "epoch": 0.92, - "learning_rate": 3.853291577604348e-06, - "loss": 3.263, + "learning_rate": 3.928257115065212e-06, + "loss": 3.2048, "step": 278500 }, { "epoch": 0.92, - "learning_rate": 3.836721843879969e-06, - "loss": 3.2548, + "learning_rate": 3.911714298948538e-06, + "loss": 3.2243, "step": 278600 }, { "epoch": 0.92, - "learning_rate": 3.82015211015559e-06, - "loss": 3.2499, + "learning_rate": 3.895171482831866e-06, + "loss": 3.2103, "step": 278700 }, { "epoch": 0.92, - "learning_rate": 3.8035823764312113e-06, - "loss": 3.2487, + "learning_rate": 3.878628666715192e-06, + "loss": 3.222, "step": 278800 }, { "epoch": 0.92, - "learning_rate": 3.787012642706832e-06, - "loss": 3.2629, + "learning_rate": 3.862085850598519e-06, + "loss": 3.227, "step": 278900 }, { "epoch": 0.92, - "learning_rate": 3.770442908982453e-06, - "loss": 3.2599, + "learning_rate": 3.845543034481846e-06, + "loss": 3.2258, "step": 279000 }, { "epoch": 0.92, - "learning_rate": 3.7538731752580735e-06, - "loss": 3.2592, + "learning_rate": 3.8290002183651735e-06, + "loss": 3.2118, "step": 279100 }, { - "epoch": 0.93, - "learning_rate": 3.7373034415336946e-06, - "loss": 3.2429, + "epoch": 0.92, + "learning_rate": 3.8124574022484998e-06, + "loss": 3.2124, "step": 279200 }, { - "epoch": 0.93, - "learning_rate": 3.720733707809316e-06, - "loss": 3.2559, + "epoch": 0.92, + "learning_rate": 3.795914586131827e-06, + "loss": 3.2096, "step": 279300 }, { - "epoch": 0.93, - "learning_rate": 3.7041639740849362e-06, - "loss": 3.247, + "epoch": 0.92, + "learning_rate": 3.7793717700151535e-06, + "loss": 3.195, "step": 279400 }, { - "epoch": 0.93, - "learning_rate": 3.6875942403605577e-06, - "loss": 3.2498, + "epoch": 0.92, + "learning_rate": 3.7628289538984806e-06, + "loss": 3.2198, "step": 279500 }, { "epoch": 0.93, - "learning_rate": 3.6710245066361788e-06, - "loss": 3.2522, + "learning_rate": 3.746286137781807e-06, + "loss": 3.2074, "step": 279600 }, { "epoch": 0.93, - "learning_rate": 3.6544547729117994e-06, - "loss": 3.2548, + "learning_rate": 3.7297433216651334e-06, + "loss": 3.1993, "step": 279700 }, { "epoch": 0.93, - "learning_rate": 3.6378850391874204e-06, - "loss": 3.2458, + "learning_rate": 3.713200505548461e-06, + "loss": 3.2058, "step": 279800 }, { "epoch": 0.93, - "learning_rate": 3.621315305463041e-06, - "loss": 3.2569, + "learning_rate": 3.696657689431787e-06, + "loss": 3.2126, "step": 279900 }, { "epoch": 0.93, - "learning_rate": 3.6047455717386625e-06, - "loss": 3.2535, + "learning_rate": 3.6801148733151147e-06, + "loss": 3.2169, "step": 280000 }, { "epoch": 0.93, - "learning_rate": 3.5881758380142835e-06, - "loss": 3.2559, + "learning_rate": 3.663572057198441e-06, + "loss": 3.217, "step": 280100 }, { "epoch": 0.93, - "learning_rate": 3.571606104289904e-06, - "loss": 3.2481, + "learning_rate": 3.647029241081768e-06, + "loss": 3.223, "step": 280200 }, { "epoch": 0.93, - "learning_rate": 3.555036370565525e-06, - "loss": 3.2313, + "learning_rate": 3.6304864249650946e-06, + "loss": 3.208, "step": 280300 }, { "epoch": 0.93, - "learning_rate": 3.538466636841146e-06, - "loss": 3.2657, + "learning_rate": 3.6139436088484217e-06, + "loss": 3.221, "step": 280400 }, { "epoch": 0.93, - "learning_rate": 3.521896903116767e-06, - "loss": 3.2597, + "learning_rate": 3.5974007927317483e-06, + "loss": 3.2079, "step": 280500 }, { "epoch": 0.93, - "learning_rate": 3.5053271693923883e-06, - "loss": 3.2486, + "learning_rate": 3.5808579766150754e-06, + "loss": 3.2119, "step": 280600 }, { "epoch": 0.93, - "learning_rate": 3.488757435668009e-06, - "loss": 3.2732, + "learning_rate": 3.564315160498402e-06, + "loss": 3.2172, "step": 280700 }, { "epoch": 0.93, - "learning_rate": 3.47218770194363e-06, - "loss": 3.2479, + "learning_rate": 3.547772344381729e-06, + "loss": 3.2074, "step": 280800 }, { "epoch": 0.93, - "learning_rate": 3.455617968219251e-06, - "loss": 3.2473, + "learning_rate": 3.531229528265056e-06, + "loss": 3.207, "step": 280900 }, { "epoch": 0.93, - "learning_rate": 3.4390482344948717e-06, - "loss": 3.2428, + "learning_rate": 3.5146867121483825e-06, + "loss": 3.206, "step": 281000 }, { "epoch": 0.93, - "learning_rate": 3.422478500770493e-06, - "loss": 3.2411, + "learning_rate": 3.4981438960317095e-06, + "loss": 3.2186, "step": 281100 }, { "epoch": 0.93, - "learning_rate": 3.4059087670461133e-06, - "loss": 3.2496, + "learning_rate": 3.481601079915036e-06, + "loss": 3.2232, "step": 281200 }, { "epoch": 0.93, - "learning_rate": 3.389339033321735e-06, - "loss": 3.2425, + "learning_rate": 3.4650582637983633e-06, + "loss": 3.2219, "step": 281300 }, { "epoch": 0.93, - "learning_rate": 3.372769299597356e-06, - "loss": 3.2326, + "learning_rate": 3.44851544768169e-06, + "loss": 3.2111, "step": 281400 }, { "epoch": 0.93, - "learning_rate": 3.3561995658729765e-06, - "loss": 3.2615, + "learning_rate": 3.431972631565017e-06, + "loss": 3.2241, "step": 281500 }, { "epoch": 0.93, - "learning_rate": 3.3396298321485975e-06, - "loss": 3.2454, + "learning_rate": 3.4154298154483432e-06, + "loss": 3.2079, "step": 281600 }, { "epoch": 0.93, - "learning_rate": 3.323060098424218e-06, - "loss": 3.2517, + "learning_rate": 3.3988869993316707e-06, + "loss": 3.2109, "step": 281700 }, { "epoch": 0.93, - "learning_rate": 3.306490364699839e-06, - "loss": 3.2438, + "learning_rate": 3.382344183214997e-06, + "loss": 3.2175, "step": 281800 }, { "epoch": 0.93, - "learning_rate": 3.2899206309754606e-06, - "loss": 3.2532, + "learning_rate": 3.3658013670983244e-06, + "loss": 3.2135, "step": 281900 }, { "epoch": 0.93, - "learning_rate": 3.2733508972510813e-06, - "loss": 3.2498, + "learning_rate": 3.3492585509816507e-06, + "loss": 3.2205, "step": 282000 }, { "epoch": 0.93, - "learning_rate": 3.2567811635267023e-06, - "loss": 3.2473, + "learning_rate": 3.332715734864978e-06, + "loss": 3.2031, "step": 282100 }, { - "epoch": 0.94, - "learning_rate": 3.2402114298023233e-06, - "loss": 3.2422, + "epoch": 0.93, + "learning_rate": 3.3161729187483044e-06, + "loss": 3.2165, "step": 282200 }, { - "epoch": 0.94, - "learning_rate": 3.223641696077944e-06, - "loss": 3.2507, + "epoch": 0.93, + "learning_rate": 3.299630102631631e-06, + "loss": 3.2207, "step": 282300 }, { - "epoch": 0.94, - "learning_rate": 3.2070719623535654e-06, - "loss": 3.251, + "epoch": 0.93, + "learning_rate": 3.283087286514958e-06, + "loss": 3.2225, "step": 282400 }, { - "epoch": 0.94, - "learning_rate": 3.1905022286291856e-06, - "loss": 3.2553, + "epoch": 0.93, + "learning_rate": 3.2665444703982848e-06, + "loss": 3.2181, "step": 282500 }, { - "epoch": 0.94, - "learning_rate": 3.173932494904807e-06, - "loss": 3.2451, + "epoch": 0.93, + "learning_rate": 3.250001654281612e-06, + "loss": 3.2178, "step": 282600 }, { "epoch": 0.94, - "learning_rate": 3.157362761180428e-06, - "loss": 3.2596, + "learning_rate": 3.2334588381649385e-06, + "loss": 3.2113, "step": 282700 }, { "epoch": 0.94, - "learning_rate": 3.1407930274560488e-06, - "loss": 3.2536, + "learning_rate": 3.2169160220482656e-06, + "loss": 3.2177, "step": 282800 }, { "epoch": 0.94, - "learning_rate": 3.12422329373167e-06, - "loss": 3.2478, + "learning_rate": 3.2003732059315922e-06, + "loss": 3.2269, "step": 282900 }, { "epoch": 0.94, - "learning_rate": 3.107653560007291e-06, - "loss": 3.2625, + "learning_rate": 3.1838303898149193e-06, + "loss": 3.2226, "step": 283000 }, { "epoch": 0.94, - "learning_rate": 3.091083826282912e-06, - "loss": 3.2486, + "learning_rate": 3.167287573698246e-06, + "loss": 3.2158, "step": 283100 }, { "epoch": 0.94, - "learning_rate": 3.0745140925585325e-06, - "loss": 3.2496, + "learning_rate": 3.150744757581573e-06, + "loss": 3.2321, "step": 283200 }, { "epoch": 0.94, - "learning_rate": 3.057944358834154e-06, - "loss": 3.2507, + "learning_rate": 3.1342019414648997e-06, + "loss": 3.2041, "step": 283300 }, { "epoch": 0.94, - "learning_rate": 3.0413746251097746e-06, - "loss": 3.2544, + "learning_rate": 3.1176591253482263e-06, + "loss": 3.2085, "step": 283400 }, { "epoch": 0.94, - "learning_rate": 3.0248048913853956e-06, - "loss": 3.2437, + "learning_rate": 3.1011163092315534e-06, + "loss": 3.2112, "step": 283500 }, { "epoch": 0.94, - "learning_rate": 3.0082351576610163e-06, - "loss": 3.2602, + "learning_rate": 3.08457349311488e-06, + "loss": 3.2034, "step": 283600 }, { "epoch": 0.94, - "learning_rate": 2.9916654239366377e-06, - "loss": 3.2509, + "learning_rate": 3.068030676998207e-06, + "loss": 3.2191, "step": 283700 }, { "epoch": 0.94, - "learning_rate": 2.9750956902122583e-06, - "loss": 3.2376, + "learning_rate": 3.051487860881534e-06, + "loss": 3.2216, "step": 283800 }, { "epoch": 0.94, - "learning_rate": 2.9585259564878794e-06, - "loss": 3.2576, + "learning_rate": 3.034945044764861e-06, + "loss": 3.2283, "step": 283900 }, { "epoch": 0.94, - "learning_rate": 2.9419562227635004e-06, - "loss": 3.252, + "learning_rate": 3.018402228648187e-06, + "loss": 3.2204, "step": 284000 }, { "epoch": 0.94, - "learning_rate": 2.9253864890391215e-06, - "loss": 3.2466, + "learning_rate": 3.001859412531514e-06, + "loss": 3.2243, "step": 284100 }, { "epoch": 0.94, - "learning_rate": 2.9088167553147425e-06, - "loss": 3.2527, + "learning_rate": 2.985316596414841e-06, + "loss": 3.2203, "step": 284200 }, { "epoch": 0.94, - "learning_rate": 2.892247021590363e-06, - "loss": 3.2576, + "learning_rate": 2.968773780298168e-06, + "loss": 3.2006, "step": 284300 }, { "epoch": 0.94, - "learning_rate": 2.875677287865984e-06, - "loss": 3.2492, + "learning_rate": 2.9522309641814946e-06, + "loss": 3.2136, "step": 284400 }, { "epoch": 0.94, - "learning_rate": 2.859107554141605e-06, - "loss": 3.2415, + "learning_rate": 2.9356881480648216e-06, + "loss": 3.218, "step": 284500 }, { "epoch": 0.94, - "learning_rate": 2.8425378204172263e-06, - "loss": 3.2399, + "learning_rate": 2.9191453319481483e-06, + "loss": 3.214, "step": 284600 }, { "epoch": 0.94, - "learning_rate": 2.825968086692847e-06, - "loss": 3.2563, + "learning_rate": 2.9026025158314754e-06, + "loss": 3.2167, "step": 284700 }, { "epoch": 0.94, - "learning_rate": 2.809398352968468e-06, - "loss": 3.2542, + "learning_rate": 2.886059699714802e-06, + "loss": 3.2243, "step": 284800 }, { "epoch": 0.94, - "learning_rate": 2.7928286192440886e-06, - "loss": 3.2552, + "learning_rate": 2.869516883598129e-06, + "loss": 3.2146, "step": 284900 }, { "epoch": 0.94, - "learning_rate": 2.77625888551971e-06, - "loss": 3.2447, + "learning_rate": 2.8529740674814557e-06, + "loss": 3.2202, "step": 285000 }, { "epoch": 0.94, - "learning_rate": 2.7596891517953306e-06, - "loss": 3.2464, + "learning_rate": 2.836431251364783e-06, + "loss": 3.2094, "step": 285100 }, { - "epoch": 0.95, - "learning_rate": 2.7431194180709517e-06, - "loss": 3.2474, + "epoch": 0.94, + "learning_rate": 2.819888435248109e-06, + "loss": 3.2113, "step": 285200 }, { - "epoch": 0.95, - "learning_rate": 2.7265496843465727e-06, - "loss": 3.2489, + "epoch": 0.94, + "learning_rate": 2.803345619131436e-06, + "loss": 3.2202, "step": 285300 }, { - "epoch": 0.95, - "learning_rate": 2.7099799506221938e-06, - "loss": 3.2457, + "epoch": 0.94, + "learning_rate": 2.7868028030147628e-06, + "loss": 3.2215, "step": 285400 }, { - "epoch": 0.95, - "learning_rate": 2.693410216897815e-06, - "loss": 3.248, + "epoch": 0.94, + "learning_rate": 2.77025998689809e-06, + "loss": 3.2075, "step": 285500 }, { - "epoch": 0.95, - "learning_rate": 2.6768404831734354e-06, - "loss": 3.2506, + "epoch": 0.94, + "learning_rate": 2.7537171707814165e-06, + "loss": 3.2211, "step": 285600 }, { "epoch": 0.95, - "learning_rate": 2.6602707494490565e-06, - "loss": 3.2474, + "learning_rate": 2.7371743546647436e-06, + "loss": 3.2196, "step": 285700 }, { "epoch": 0.95, - "learning_rate": 2.643701015724677e-06, - "loss": 3.2457, + "learning_rate": 2.7206315385480702e-06, + "loss": 3.2289, "step": 285800 }, { "epoch": 0.95, - "learning_rate": 2.6271312820002986e-06, - "loss": 3.2339, + "learning_rate": 2.7040887224313973e-06, + "loss": 3.2023, "step": 285900 }, { "epoch": 0.95, - "learning_rate": 2.610561548275919e-06, - "loss": 3.2555, + "learning_rate": 2.687545906314724e-06, + "loss": 3.2242, "step": 286000 }, { "epoch": 0.95, - "learning_rate": 2.5939918145515402e-06, - "loss": 3.2489, + "learning_rate": 2.6710030901980506e-06, + "loss": 3.2228, "step": 286100 }, { "epoch": 0.95, - "learning_rate": 2.5774220808271613e-06, - "loss": 3.2534, + "learning_rate": 2.6544602740813777e-06, + "loss": 3.2271, "step": 286200 }, { "epoch": 0.95, - "learning_rate": 2.5608523471027823e-06, - "loss": 3.2453, + "learning_rate": 2.6379174579647043e-06, + "loss": 3.211, "step": 286300 }, { "epoch": 0.95, - "learning_rate": 2.5442826133784034e-06, - "loss": 3.2474, + "learning_rate": 2.621374641848031e-06, + "loss": 3.2089, "step": 286400 }, { "epoch": 0.95, - "learning_rate": 2.527712879654024e-06, - "loss": 3.2515, + "learning_rate": 2.604831825731358e-06, + "loss": 3.2045, "step": 286500 }, { "epoch": 0.95, - "learning_rate": 2.511143145929645e-06, - "loss": 3.2505, + "learning_rate": 2.5882890096146847e-06, + "loss": 3.2121, "step": 286600 }, { "epoch": 0.95, - "learning_rate": 2.494573412205266e-06, - "loss": 3.2454, + "learning_rate": 2.571746193498012e-06, + "loss": 3.2064, "step": 286700 }, { "epoch": 0.95, - "learning_rate": 2.478003678480887e-06, - "loss": 3.2544, + "learning_rate": 2.5552033773813384e-06, + "loss": 3.2231, "step": 286800 }, { "epoch": 0.95, - "learning_rate": 2.4614339447565077e-06, - "loss": 3.2502, + "learning_rate": 2.538660561264665e-06, + "loss": 3.2206, "step": 286900 }, { "epoch": 0.95, - "learning_rate": 2.4448642110321288e-06, - "loss": 3.2385, + "learning_rate": 2.522117745147992e-06, + "loss": 3.2181, "step": 287000 }, { "epoch": 0.95, - "learning_rate": 2.42829447730775e-06, - "loss": 3.2442, + "learning_rate": 2.505574929031319e-06, + "loss": 3.2059, "step": 287100 }, { "epoch": 0.95, - "learning_rate": 2.411724743583371e-06, - "loss": 3.2561, + "learning_rate": 2.489032112914646e-06, + "loss": 3.2238, "step": 287200 }, { "epoch": 0.95, - "learning_rate": 2.395155009858992e-06, - "loss": 3.2454, + "learning_rate": 2.4724892967979726e-06, + "loss": 3.204, "step": 287300 }, { "epoch": 0.95, - "learning_rate": 2.3785852761346125e-06, - "loss": 3.2519, + "learning_rate": 2.4559464806812996e-06, + "loss": 3.2031, "step": 287400 }, { "epoch": 0.95, - "learning_rate": 2.3620155424102336e-06, - "loss": 3.2516, + "learning_rate": 2.4394036645646263e-06, + "loss": 3.2129, "step": 287500 }, { "epoch": 0.95, - "learning_rate": 2.3454458086858546e-06, - "loss": 3.2579, + "learning_rate": 2.4228608484479534e-06, + "loss": 3.2166, "step": 287600 }, { "epoch": 0.95, - "learning_rate": 2.3288760749614757e-06, - "loss": 3.2401, + "learning_rate": 2.4063180323312796e-06, + "loss": 3.2136, "step": 287700 }, { "epoch": 0.95, - "learning_rate": 2.3123063412370963e-06, - "loss": 3.255, + "learning_rate": 2.3897752162146067e-06, + "loss": 3.2167, "step": 287800 }, { "epoch": 0.95, - "learning_rate": 2.2957366075127173e-06, - "loss": 3.2459, + "learning_rate": 2.3732324000979333e-06, + "loss": 3.2002, "step": 287900 }, { "epoch": 0.95, - "learning_rate": 2.2791668737883384e-06, - "loss": 3.2576, + "learning_rate": 2.3566895839812604e-06, + "loss": 3.2087, "step": 288000 }, { "epoch": 0.95, - "learning_rate": 2.2625971400639594e-06, - "loss": 3.2478, + "learning_rate": 2.340146767864587e-06, + "loss": 3.2058, "step": 288100 }, { - "epoch": 0.96, - "learning_rate": 2.2460274063395804e-06, - "loss": 3.2433, + "epoch": 0.95, + "learning_rate": 2.323603951747914e-06, + "loss": 3.2142, "step": 288200 }, { - "epoch": 0.96, - "learning_rate": 2.229457672615201e-06, - "loss": 3.2443, + "epoch": 0.95, + "learning_rate": 2.3070611356312408e-06, + "loss": 3.2121, "step": 288300 }, { - "epoch": 0.96, - "learning_rate": 2.212887938890822e-06, - "loss": 3.2512, + "epoch": 0.95, + "learning_rate": 2.290518319514568e-06, + "loss": 3.2165, "step": 288400 }, { - "epoch": 0.96, - "learning_rate": 2.196318205166443e-06, - "loss": 3.2514, + "epoch": 0.95, + "learning_rate": 2.2739755033978945e-06, + "loss": 3.2133, "step": 288500 }, { - "epoch": 0.96, - "learning_rate": 2.179748471442064e-06, - "loss": 3.2551, + "epoch": 0.95, + "learning_rate": 2.2574326872812216e-06, + "loss": 3.2224, "step": 288600 }, { "epoch": 0.96, - "learning_rate": 2.163178737717685e-06, - "loss": 3.2427, + "learning_rate": 2.2408898711645482e-06, + "loss": 3.2014, "step": 288700 }, { "epoch": 0.96, - "learning_rate": 2.146609003993306e-06, - "loss": 3.2456, + "learning_rate": 2.2243470550478753e-06, + "loss": 3.208, "step": 288800 }, { "epoch": 0.96, - "learning_rate": 2.130039270268927e-06, - "loss": 3.229, + "learning_rate": 2.2078042389312015e-06, + "loss": 3.2049, "step": 288900 }, { "epoch": 0.96, - "learning_rate": 2.113469536544548e-06, - "loss": 3.2413, + "learning_rate": 2.1912614228145286e-06, + "loss": 3.211, "step": 289000 }, { "epoch": 0.96, - "learning_rate": 2.0968998028201686e-06, - "loss": 3.244, + "learning_rate": 2.1747186066978553e-06, + "loss": 3.2117, "step": 289100 }, { "epoch": 0.96, - "learning_rate": 2.0803300690957896e-06, - "loss": 3.2637, + "learning_rate": 2.1581757905811823e-06, + "loss": 3.2105, "step": 289200 }, { "epoch": 0.96, - "learning_rate": 2.0637603353714107e-06, - "loss": 3.2523, + "learning_rate": 2.141632974464509e-06, + "loss": 3.2084, "step": 289300 }, { "epoch": 0.96, - "learning_rate": 2.0471906016470317e-06, - "loss": 3.2443, + "learning_rate": 2.125090158347836e-06, + "loss": 3.2132, "step": 289400 }, { "epoch": 0.96, - "learning_rate": 2.0306208679226527e-06, - "loss": 3.2409, + "learning_rate": 2.1085473422311627e-06, + "loss": 3.2126, "step": 289500 }, { "epoch": 0.96, - "learning_rate": 2.0140511341982734e-06, - "loss": 3.2374, + "learning_rate": 2.09200452611449e-06, + "loss": 3.2169, "step": 289600 }, { "epoch": 0.96, - "learning_rate": 1.9974814004738944e-06, - "loss": 3.255, + "learning_rate": 2.0754617099978164e-06, + "loss": 3.2065, "step": 289700 }, { "epoch": 0.96, - "learning_rate": 1.9809116667495154e-06, - "loss": 3.2492, + "learning_rate": 2.0589188938811435e-06, + "loss": 3.2273, "step": 289800 }, { "epoch": 0.96, - "learning_rate": 1.9643419330251365e-06, - "loss": 3.245, + "learning_rate": 2.04237607776447e-06, + "loss": 3.2041, "step": 289900 }, { "epoch": 0.96, - "learning_rate": 1.947772199300757e-06, - "loss": 3.2437, + "learning_rate": 2.0258332616477972e-06, + "loss": 3.2115, "step": 290000 }, { "epoch": 0.96, - "learning_rate": 1.931202465576378e-06, - "loss": 3.2524, + "learning_rate": 2.0092904455311235e-06, + "loss": 3.205, "step": 290100 }, { "epoch": 0.96, - "learning_rate": 1.914632731851999e-06, - "loss": 3.237, + "learning_rate": 1.9927476294144505e-06, + "loss": 3.2109, "step": 290200 }, { "epoch": 0.96, - "learning_rate": 1.8980629981276202e-06, - "loss": 3.2353, + "learning_rate": 1.976204813297777e-06, + "loss": 3.2222, "step": 290300 }, { "epoch": 0.96, - "learning_rate": 1.881493264403241e-06, - "loss": 3.24, + "learning_rate": 1.9596619971811043e-06, + "loss": 3.2167, "step": 290400 }, { "epoch": 0.96, - "learning_rate": 1.864923530678862e-06, - "loss": 3.2448, + "learning_rate": 1.943119181064431e-06, + "loss": 3.2184, "step": 290500 }, { "epoch": 0.96, - "learning_rate": 1.8483537969544832e-06, - "loss": 3.2502, + "learning_rate": 1.926576364947758e-06, + "loss": 3.2063, "step": 290600 }, { "epoch": 0.96, - "learning_rate": 1.831784063230104e-06, - "loss": 3.2392, + "learning_rate": 1.9100335488310847e-06, + "loss": 3.2064, "step": 290700 }, { "epoch": 0.96, - "learning_rate": 1.815214329505725e-06, - "loss": 3.2376, + "learning_rate": 1.8934907327144115e-06, + "loss": 3.2036, "step": 290800 }, { "epoch": 0.96, - "learning_rate": 1.7986445957813459e-06, - "loss": 3.2425, + "learning_rate": 1.8769479165977384e-06, + "loss": 3.2089, "step": 290900 }, { "epoch": 0.96, - "learning_rate": 1.7820748620569667e-06, - "loss": 3.2426, + "learning_rate": 1.8604051004810652e-06, + "loss": 3.2037, "step": 291000 }, { "epoch": 0.96, - "learning_rate": 1.765505128332588e-06, - "loss": 3.2412, + "learning_rate": 1.8438622843643921e-06, + "loss": 3.2097, "step": 291100 }, { - "epoch": 0.97, - "learning_rate": 1.7489353946082088e-06, - "loss": 3.2503, + "epoch": 0.96, + "learning_rate": 1.827319468247719e-06, + "loss": 3.2083, "step": 291200 }, { - "epoch": 0.97, - "learning_rate": 1.7323656608838296e-06, - "loss": 3.2537, + "epoch": 0.96, + "learning_rate": 1.8107766521310458e-06, + "loss": 3.2177, "step": 291300 }, { - "epoch": 0.97, - "learning_rate": 1.7157959271594505e-06, - "loss": 3.2248, + "epoch": 0.96, + "learning_rate": 1.7942338360143723e-06, + "loss": 3.2059, "step": 291400 }, { - "epoch": 0.97, - "learning_rate": 1.6992261934350717e-06, - "loss": 3.241, + "epoch": 0.96, + "learning_rate": 1.7776910198976991e-06, + "loss": 3.2141, "step": 291500 }, { - "epoch": 0.97, - "learning_rate": 1.6826564597106925e-06, - "loss": 3.2459, + "epoch": 0.96, + "learning_rate": 1.761148203781026e-06, + "loss": 3.2057, "step": 291600 }, { "epoch": 0.97, - "learning_rate": 1.6660867259863134e-06, - "loss": 3.2414, + "learning_rate": 1.7446053876643529e-06, + "loss": 3.1882, "step": 291700 }, { "epoch": 0.97, - "learning_rate": 1.6495169922619344e-06, - "loss": 3.2381, + "learning_rate": 1.7280625715476797e-06, + "loss": 3.2158, "step": 291800 }, { "epoch": 0.97, - "learning_rate": 1.6329472585375555e-06, - "loss": 3.2293, + "learning_rate": 1.7115197554310066e-06, + "loss": 3.2132, "step": 291900 }, { "epoch": 0.97, - "learning_rate": 1.6163775248131765e-06, - "loss": 3.2505, + "learning_rate": 1.6949769393143335e-06, + "loss": 3.2246, "step": 292000 }, { "epoch": 0.97, - "learning_rate": 1.5998077910887973e-06, - "loss": 3.2451, + "learning_rate": 1.6784341231976603e-06, + "loss": 3.2225, "step": 292100 }, { "epoch": 0.97, - "learning_rate": 1.5832380573644182e-06, - "loss": 3.2446, + "learning_rate": 1.6618913070809872e-06, + "loss": 3.216, "step": 292200 }, { "epoch": 0.97, - "learning_rate": 1.566668323640039e-06, - "loss": 3.2557, + "learning_rate": 1.645348490964314e-06, + "loss": 3.2302, "step": 292300 }, { "epoch": 0.97, - "learning_rate": 1.55009858991566e-06, - "loss": 3.2457, + "learning_rate": 1.628805674847641e-06, + "loss": 3.2145, "step": 292400 }, { "epoch": 0.97, - "learning_rate": 1.533528856191281e-06, - "loss": 3.2484, + "learning_rate": 1.6122628587309678e-06, + "loss": 3.2149, "step": 292500 }, { "epoch": 0.97, - "learning_rate": 1.516959122466902e-06, - "loss": 3.2435, + "learning_rate": 1.5957200426142942e-06, + "loss": 3.2058, "step": 292600 }, { "epoch": 0.97, - "learning_rate": 1.500389388742523e-06, - "loss": 3.2548, + "learning_rate": 1.579177226497621e-06, + "loss": 3.2138, "step": 292700 }, { "epoch": 0.97, - "learning_rate": 1.483819655018144e-06, - "loss": 3.2529, + "learning_rate": 1.562634410380948e-06, + "loss": 3.2141, "step": 292800 }, { "epoch": 0.97, - "learning_rate": 1.467249921293765e-06, - "loss": 3.2526, + "learning_rate": 1.5460915942642748e-06, + "loss": 3.1934, "step": 292900 }, { "epoch": 0.97, - "learning_rate": 1.4506801875693859e-06, - "loss": 3.2568, + "learning_rate": 1.5295487781476017e-06, + "loss": 3.2085, "step": 293000 }, { "epoch": 0.97, - "learning_rate": 1.434110453845007e-06, - "loss": 3.2403, + "learning_rate": 1.5130059620309285e-06, + "loss": 3.2065, "step": 293100 }, { "epoch": 0.97, - "learning_rate": 1.4175407201206278e-06, - "loss": 3.2485, + "learning_rate": 1.4964631459142554e-06, + "loss": 3.2164, "step": 293200 }, { "epoch": 0.97, - "learning_rate": 1.4009709863962486e-06, - "loss": 3.2338, + "learning_rate": 1.4799203297975823e-06, + "loss": 3.2089, "step": 293300 }, { "epoch": 0.97, - "learning_rate": 1.3844012526718696e-06, - "loss": 3.2501, + "learning_rate": 1.4633775136809091e-06, + "loss": 3.2065, "step": 293400 }, { "epoch": 0.97, - "learning_rate": 1.3678315189474905e-06, - "loss": 3.237, + "learning_rate": 1.4468346975642358e-06, + "loss": 3.2077, "step": 293500 }, { "epoch": 0.97, - "learning_rate": 1.3512617852231115e-06, - "loss": 3.2555, + "learning_rate": 1.4302918814475627e-06, + "loss": 3.2056, "step": 293600 }, { "epoch": 0.97, - "learning_rate": 1.3346920514987323e-06, - "loss": 3.2482, + "learning_rate": 1.4137490653308895e-06, + "loss": 3.2158, "step": 293700 }, { "epoch": 0.97, - "learning_rate": 1.3181223177743534e-06, - "loss": 3.2429, + "learning_rate": 1.3972062492142164e-06, + "loss": 3.208, "step": 293800 }, { "epoch": 0.97, - "learning_rate": 1.3015525840499744e-06, - "loss": 3.2479, + "learning_rate": 1.3806634330975432e-06, + "loss": 3.2034, "step": 293900 }, { "epoch": 0.97, - "learning_rate": 1.2849828503255955e-06, - "loss": 3.2473, + "learning_rate": 1.3641206169808701e-06, + "loss": 3.223, "step": 294000 }, { "epoch": 0.97, - "learning_rate": 1.2684131166012163e-06, - "loss": 3.2655, + "learning_rate": 1.3475778008641968e-06, + "loss": 3.1986, "step": 294100 }, { "epoch": 0.97, - "learning_rate": 1.2518433828768373e-06, - "loss": 3.235, + "learning_rate": 1.3310349847475236e-06, + "loss": 3.213, "step": 294200 }, { - "epoch": 0.98, - "learning_rate": 1.2352736491524582e-06, - "loss": 3.2538, + "epoch": 0.97, + "learning_rate": 1.3144921686308505e-06, + "loss": 3.2191, "step": 294300 }, { - "epoch": 0.98, - "learning_rate": 1.2187039154280792e-06, - "loss": 3.2426, + "epoch": 0.97, + "learning_rate": 1.2979493525141774e-06, + "loss": 3.2027, "step": 294400 }, { - "epoch": 0.98, - "learning_rate": 1.2021341817037e-06, - "loss": 3.243, + "epoch": 0.97, + "learning_rate": 1.2814065363975042e-06, + "loss": 3.2177, "step": 294500 }, { - "epoch": 0.98, - "learning_rate": 1.1855644479793209e-06, - "loss": 3.2503, + "epoch": 0.97, + "learning_rate": 1.2648637202808309e-06, + "loss": 3.1982, "step": 294600 }, { "epoch": 0.98, - "learning_rate": 1.168994714254942e-06, - "loss": 3.2571, + "learning_rate": 1.2483209041641577e-06, + "loss": 3.2195, "step": 294700 }, { "epoch": 0.98, - "learning_rate": 1.1524249805305628e-06, - "loss": 3.2534, + "learning_rate": 1.2317780880474846e-06, + "loss": 3.204, "step": 294800 }, { "epoch": 0.98, - "learning_rate": 1.1358552468061838e-06, - "loss": 3.2477, + "learning_rate": 1.2152352719308115e-06, + "loss": 3.2043, "step": 294900 }, { "epoch": 0.98, - "learning_rate": 1.1192855130818048e-06, - "loss": 3.2456, + "learning_rate": 1.1986924558141381e-06, + "loss": 3.217, "step": 295000 }, { "epoch": 0.98, - "learning_rate": 1.1027157793574259e-06, - "loss": 3.2562, + "learning_rate": 1.182149639697465e-06, + "loss": 3.2007, "step": 295100 }, { "epoch": 0.98, - "learning_rate": 1.0861460456330467e-06, - "loss": 3.2389, + "learning_rate": 1.1656068235807918e-06, + "loss": 3.2129, "step": 295200 }, { "epoch": 0.98, - "learning_rate": 1.0695763119086678e-06, - "loss": 3.2564, + "learning_rate": 1.1490640074641187e-06, + "loss": 3.2117, "step": 295300 }, { "epoch": 0.98, - "learning_rate": 1.0530065781842886e-06, - "loss": 3.2599, + "learning_rate": 1.1325211913474454e-06, + "loss": 3.2091, "step": 295400 }, { "epoch": 0.98, - "learning_rate": 1.0364368444599096e-06, - "loss": 3.2318, + "learning_rate": 1.1159783752307722e-06, + "loss": 3.2112, "step": 295500 }, { "epoch": 0.98, - "learning_rate": 1.0198671107355305e-06, - "loss": 3.247, + "learning_rate": 1.099435559114099e-06, + "loss": 3.2084, "step": 295600 }, { "epoch": 0.98, - "learning_rate": 1.0032973770111515e-06, - "loss": 3.252, + "learning_rate": 1.082892742997426e-06, + "loss": 3.1951, "step": 295700 }, { "epoch": 0.98, - "learning_rate": 9.867276432867723e-07, - "loss": 3.2537, + "learning_rate": 1.0663499268807528e-06, + "loss": 3.2097, "step": 295800 }, { "epoch": 0.98, - "learning_rate": 9.701579095623934e-07, - "loss": 3.2519, + "learning_rate": 1.0498071107640797e-06, + "loss": 3.22, "step": 295900 }, { "epoch": 0.98, - "learning_rate": 9.535881758380143e-07, - "loss": 3.2559, + "learning_rate": 1.0332642946474063e-06, + "loss": 3.2024, "step": 296000 }, { "epoch": 0.98, - "learning_rate": 9.370184421136352e-07, - "loss": 3.2399, + "learning_rate": 1.0167214785307332e-06, + "loss": 3.214, "step": 296100 }, { "epoch": 0.98, - "learning_rate": 9.204487083892562e-07, - "loss": 3.2606, + "learning_rate": 1.00017866241406e-06, + "loss": 3.2218, "step": 296200 }, { "epoch": 0.98, - "learning_rate": 9.038789746648771e-07, - "loss": 3.2415, + "learning_rate": 9.83635846297387e-07, + "loss": 3.2131, "step": 296300 }, { "epoch": 0.98, - "learning_rate": 8.873092409404982e-07, - "loss": 3.2532, + "learning_rate": 9.670930301807138e-07, + "loss": 3.1993, "step": 296400 }, { "epoch": 0.98, - "learning_rate": 8.70739507216119e-07, - "loss": 3.2429, + "learning_rate": 9.505502140640406e-07, + "loss": 3.2195, "step": 296500 }, { "epoch": 0.98, - "learning_rate": 8.541697734917401e-07, - "loss": 3.2389, + "learning_rate": 9.340073979473674e-07, + "loss": 3.2226, "step": 296600 }, { "epoch": 0.98, - "learning_rate": 8.37600039767361e-07, - "loss": 3.2571, + "learning_rate": 9.174645818306942e-07, + "loss": 3.2104, "step": 296700 }, { "epoch": 0.98, - "learning_rate": 8.21030306042982e-07, - "loss": 3.2467, + "learning_rate": 9.00921765714021e-07, + "loss": 3.2096, "step": 296800 }, { "epoch": 0.98, - "learning_rate": 8.044605723186029e-07, - "loss": 3.2457, + "learning_rate": 8.843789495973479e-07, + "loss": 3.2324, "step": 296900 }, { "epoch": 0.98, - "learning_rate": 7.878908385942239e-07, - "loss": 3.2513, + "learning_rate": 8.678361334806748e-07, + "loss": 3.1977, "step": 297000 }, { "epoch": 0.98, - "learning_rate": 7.713211048698447e-07, - "loss": 3.2431, + "learning_rate": 8.512933173640016e-07, + "loss": 3.2092, "step": 297100 }, { "epoch": 0.98, - "learning_rate": 7.547513711454657e-07, - "loss": 3.2323, + "learning_rate": 8.347505012473283e-07, + "loss": 3.2122, "step": 297200 }, { - "epoch": 0.99, - "learning_rate": 7.381816374210867e-07, - "loss": 3.2588, + "epoch": 0.98, + "learning_rate": 8.182076851306551e-07, + "loss": 3.222, "step": 297300 }, { - "epoch": 0.99, - "learning_rate": 7.216119036967077e-07, - "loss": 3.2404, + "epoch": 0.98, + "learning_rate": 8.01664869013982e-07, + "loss": 3.2019, "step": 297400 }, { - "epoch": 0.99, - "learning_rate": 7.050421699723286e-07, - "loss": 3.2661, + "epoch": 0.98, + "learning_rate": 7.851220528973089e-07, + "loss": 3.2155, "step": 297500 }, { - "epoch": 0.99, - "learning_rate": 6.884724362479495e-07, - "loss": 3.2386, + "epoch": 0.98, + "learning_rate": 7.685792367806356e-07, + "loss": 3.2188, "step": 297600 }, { - "epoch": 0.99, - "learning_rate": 6.719027025235705e-07, - "loss": 3.2304, + "epoch": 0.98, + "learning_rate": 7.520364206639625e-07, + "loss": 3.2055, "step": 297700 }, { "epoch": 0.99, - "learning_rate": 6.553329687991914e-07, - "loss": 3.2572, + "learning_rate": 7.354936045472893e-07, + "loss": 3.227, "step": 297800 }, { "epoch": 0.99, - "learning_rate": 6.387632350748124e-07, - "loss": 3.2481, + "learning_rate": 7.189507884306161e-07, + "loss": 3.2068, "step": 297900 }, { "epoch": 0.99, - "learning_rate": 6.221935013504333e-07, - "loss": 3.241, + "learning_rate": 7.02407972313943e-07, + "loss": 3.2258, "step": 298000 }, { "epoch": 0.99, - "learning_rate": 6.056237676260542e-07, - "loss": 3.2435, + "learning_rate": 6.858651561972698e-07, + "loss": 3.2147, "step": 298100 }, { "epoch": 0.99, - "learning_rate": 5.890540339016752e-07, - "loss": 3.2394, + "learning_rate": 6.693223400805967e-07, + "loss": 3.2091, "step": 298200 }, { "epoch": 0.99, - "learning_rate": 5.724843001772962e-07, - "loss": 3.2528, + "learning_rate": 6.527795239639235e-07, + "loss": 3.2025, "step": 298300 }, { "epoch": 0.99, - "learning_rate": 5.559145664529171e-07, - "loss": 3.2504, + "learning_rate": 6.362367078472503e-07, + "loss": 3.22, "step": 298400 }, { "epoch": 0.99, - "learning_rate": 5.393448327285381e-07, - "loss": 3.242, + "learning_rate": 6.196938917305772e-07, + "loss": 3.2106, "step": 298500 }, { "epoch": 0.99, - "learning_rate": 5.22775099004159e-07, - "loss": 3.2582, + "learning_rate": 6.031510756139039e-07, + "loss": 3.215, "step": 298600 }, { "epoch": 0.99, - "learning_rate": 5.0620536527978e-07, - "loss": 3.2659, + "learning_rate": 5.866082594972308e-07, + "loss": 3.2134, "step": 298700 }, { "epoch": 0.99, - "learning_rate": 4.89635631555401e-07, - "loss": 3.2437, + "learning_rate": 5.700654433805577e-07, + "loss": 3.2006, "step": 298800 }, { "epoch": 0.99, - "learning_rate": 4.730658978310219e-07, - "loss": 3.2421, + "learning_rate": 5.535226272638844e-07, + "loss": 3.2093, "step": 298900 }, { "epoch": 0.99, - "learning_rate": 4.564961641066429e-07, - "loss": 3.2466, + "learning_rate": 5.369798111472113e-07, + "loss": 3.219, "step": 299000 }, { "epoch": 0.99, - "learning_rate": 4.399264303822637e-07, - "loss": 3.2463, + "learning_rate": 5.20436995030538e-07, + "loss": 3.2185, "step": 299100 }, { "epoch": 0.99, - "learning_rate": 4.233566966578847e-07, - "loss": 3.2505, + "learning_rate": 5.038941789138649e-07, + "loss": 3.2186, "step": 299200 }, { "epoch": 0.99, - "learning_rate": 4.0678696293350564e-07, - "loss": 3.2535, + "learning_rate": 4.873513627971917e-07, + "loss": 3.2104, "step": 299300 }, { "epoch": 0.99, - "learning_rate": 3.9021722920912663e-07, - "loss": 3.2542, + "learning_rate": 4.708085466805186e-07, + "loss": 3.2135, "step": 299400 }, { "epoch": 0.99, - "learning_rate": 3.7364749548474757e-07, - "loss": 3.244, + "learning_rate": 4.5426573056384535e-07, + "loss": 3.2173, "step": 299500 }, { "epoch": 0.99, - "learning_rate": 3.570777617603685e-07, - "loss": 3.2536, + "learning_rate": 4.377229144471722e-07, + "loss": 3.2153, "step": 299600 }, { "epoch": 0.99, - "learning_rate": 3.405080280359895e-07, - "loss": 3.2654, + "learning_rate": 4.211800983304991e-07, + "loss": 3.2024, "step": 299700 }, { "epoch": 0.99, - "learning_rate": 3.2393829431161043e-07, - "loss": 3.2592, + "learning_rate": 4.0463728221382583e-07, + "loss": 3.2402, "step": 299800 }, { "epoch": 0.99, - "learning_rate": 3.0736856058723137e-07, - "loss": 3.2447, + "learning_rate": 3.880944660971527e-07, + "loss": 3.1929, "step": 299900 }, { "epoch": 0.99, - "learning_rate": 2.9079882686285236e-07, - "loss": 3.245, + "learning_rate": 3.715516499804795e-07, + "loss": 3.2123, "step": 300000 }, { "epoch": 0.99, - "learning_rate": 2.7422909313847324e-07, - "loss": 3.2474, + "learning_rate": 3.550088338638063e-07, + "loss": 3.2161, "step": 300100 }, { "epoch": 0.99, - "learning_rate": 2.5765935941409423e-07, - "loss": 3.2475, + "learning_rate": 3.3846601774713313e-07, + "loss": 3.2216, "step": 300200 }, { - "epoch": 1.0, - "learning_rate": 2.4108962568971517e-07, - "loss": 3.2604, + "epoch": 0.99, + "learning_rate": 3.2192320163045994e-07, + "loss": 3.2204, "step": 300300 }, { - "epoch": 1.0, - "learning_rate": 2.2451989196533614e-07, - "loss": 3.2533, + "epoch": 0.99, + "learning_rate": 3.053803855137868e-07, + "loss": 3.2127, "step": 300400 }, { - "epoch": 1.0, - "learning_rate": 2.0795015824095707e-07, - "loss": 3.2473, + "epoch": 0.99, + "learning_rate": 2.888375693971136e-07, + "loss": 3.2248, "step": 300500 }, { - "epoch": 1.0, - "learning_rate": 1.91380424516578e-07, - "loss": 3.2491, + "epoch": 0.99, + "learning_rate": 2.7229475328044043e-07, + "loss": 3.2068, "step": 300600 }, { - "epoch": 1.0, - "learning_rate": 1.7481069079219898e-07, - "loss": 3.255, + "epoch": 0.99, + "learning_rate": 2.557519371637673e-07, + "loss": 3.2173, "step": 300700 }, { "epoch": 1.0, - "learning_rate": 1.5824095706781994e-07, - "loss": 3.2404, + "learning_rate": 2.392091210470941e-07, + "loss": 3.223, "step": 300800 }, { "epoch": 1.0, - "learning_rate": 1.4167122334344088e-07, - "loss": 3.2458, + "learning_rate": 2.2266630493042091e-07, + "loss": 3.2024, "step": 300900 }, { "epoch": 1.0, - "learning_rate": 1.2510148961906181e-07, - "loss": 3.2493, + "learning_rate": 2.0612348881374775e-07, + "loss": 3.2027, "step": 301000 }, { "epoch": 1.0, - "learning_rate": 1.0853175589468278e-07, - "loss": 3.249, + "learning_rate": 1.8958067269707456e-07, + "loss": 3.2021, "step": 301100 }, { "epoch": 1.0, - "learning_rate": 9.196202217030373e-08, - "loss": 3.2519, + "learning_rate": 1.730378565804014e-07, + "loss": 3.208, "step": 301200 }, { "epoch": 1.0, - "learning_rate": 7.539228844592467e-08, - "loss": 3.2445, + "learning_rate": 1.5649504046372824e-07, + "loss": 3.2202, "step": 301300 }, { "epoch": 1.0, - "learning_rate": 5.882255472154563e-08, - "loss": 3.2429, + "learning_rate": 1.3995222434705505e-07, + "loss": 3.216, "step": 301400 }, { "epoch": 1.0, - "learning_rate": 4.2252820997166576e-08, - "loss": 3.2521, + "learning_rate": 1.2340940823038189e-07, + "loss": 3.214, "step": 301500 + }, + { + "epoch": 1.0, + "learning_rate": 1.0686659211370871e-07, + "loss": 3.2043, + "step": 301600 + }, + { + "epoch": 1.0, + "learning_rate": 9.032377599703553e-08, + "loss": 3.2183, + "step": 301700 + }, + { + "epoch": 1.0, + "learning_rate": 7.378095988036236e-08, + "loss": 3.2225, + "step": 301800 + }, + { + "epoch": 1.0, + "learning_rate": 5.723814376368919e-08, + "loss": 3.2166, + "step": 301900 + }, + { + "epoch": 1.0, + "learning_rate": 4.0695327647016006e-08, + "loss": 3.2158, + "step": 302000 } ], - "max_steps": 301755, + "max_steps": 302246, "num_train_epochs": 1, - "total_flos": 2.575848645328896e+18, + "total_flos": 2.580120367792128e+18, "trial_name": null, "trial_params": null }