{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 232668, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.989255075902144e-05, "loss": 1.8237, "step": 500 }, { "epoch": 0.01, "learning_rate": 4.978510151804288e-05, "loss": 1.8141, "step": 1000 }, { "epoch": 0.02, "learning_rate": 4.967765227706432e-05, "loss": 1.7858, "step": 1500 }, { "epoch": 0.03, "learning_rate": 4.9570203036085755e-05, "loss": 1.7785, "step": 2000 }, { "epoch": 0.03, "learning_rate": 4.946275379510719e-05, "loss": 1.7828, "step": 2500 }, { "epoch": 0.04, "learning_rate": 4.935530455412863e-05, "loss": 1.7669, "step": 3000 }, { "epoch": 0.05, "learning_rate": 4.924785531315007e-05, "loss": 1.7623, "step": 3500 }, { "epoch": 0.05, "learning_rate": 4.914040607217151e-05, "loss": 1.7643, "step": 4000 }, { "epoch": 0.06, "learning_rate": 4.9032956831192946e-05, "loss": 1.7587, "step": 4500 }, { "epoch": 0.06, "learning_rate": 4.8925507590214384e-05, "loss": 1.7611, "step": 5000 }, { "epoch": 0.07, "learning_rate": 4.881805834923582e-05, "loss": 1.7565, "step": 5500 }, { "epoch": 0.08, "learning_rate": 4.871060910825726e-05, "loss": 1.745, "step": 6000 }, { "epoch": 0.08, "learning_rate": 4.8603159867278705e-05, "loss": 1.7496, "step": 6500 }, { "epoch": 0.09, "learning_rate": 4.849571062630014e-05, "loss": 1.7557, "step": 7000 }, { "epoch": 0.1, "learning_rate": 4.8388261385321575e-05, "loss": 1.7467, "step": 7500 }, { "epoch": 0.1, "learning_rate": 4.828081214434301e-05, "loss": 1.739, "step": 8000 }, { "epoch": 0.11, "learning_rate": 4.817336290336446e-05, "loss": 1.7374, "step": 8500 }, { "epoch": 0.12, "learning_rate": 4.806591366238589e-05, "loss": 1.7322, "step": 9000 }, { "epoch": 0.12, "learning_rate": 4.795846442140733e-05, "loss": 1.736, "step": 9500 }, { "epoch": 0.13, "learning_rate": 4.7851015180428766e-05, "loss": 1.7292, "step": 10000 }, { "epoch": 0.14, "learning_rate": 4.774356593945021e-05, "loss": 1.7209, "step": 10500 }, { "epoch": 0.14, "learning_rate": 4.763611669847164e-05, "loss": 1.7185, "step": 11000 }, { "epoch": 0.15, "learning_rate": 4.752866745749308e-05, "loss": 1.7245, "step": 11500 }, { "epoch": 0.15, "learning_rate": 4.7421218216514525e-05, "loss": 1.7284, "step": 12000 }, { "epoch": 0.16, "learning_rate": 4.731376897553596e-05, "loss": 1.7205, "step": 12500 }, { "epoch": 0.17, "learning_rate": 4.7206319734557395e-05, "loss": 1.7117, "step": 13000 }, { "epoch": 0.17, "learning_rate": 4.709887049357883e-05, "loss": 1.7213, "step": 13500 }, { "epoch": 0.18, "learning_rate": 4.699142125260028e-05, "loss": 1.711, "step": 14000 }, { "epoch": 0.19, "learning_rate": 4.6883972011621716e-05, "loss": 1.7087, "step": 14500 }, { "epoch": 0.19, "learning_rate": 4.677652277064315e-05, "loss": 1.7137, "step": 15000 }, { "epoch": 0.2, "learning_rate": 4.6669073529664586e-05, "loss": 1.7007, "step": 15500 }, { "epoch": 0.21, "learning_rate": 4.656162428868603e-05, "loss": 1.7068, "step": 16000 }, { "epoch": 0.21, "learning_rate": 4.645417504770747e-05, "loss": 1.6998, "step": 16500 }, { "epoch": 0.22, "learning_rate": 4.63467258067289e-05, "loss": 1.6966, "step": 17000 }, { "epoch": 0.23, "learning_rate": 4.623927656575034e-05, "loss": 1.6987, "step": 17500 }, { "epoch": 0.23, "learning_rate": 4.613182732477178e-05, "loss": 1.7104, "step": 18000 }, { "epoch": 0.24, "learning_rate": 4.602437808379322e-05, "loss": 1.7054, "step": 18500 }, { "epoch": 0.24, "learning_rate": 4.591692884281465e-05, "loss": 1.7, "step": 19000 }, { "epoch": 0.25, "learning_rate": 4.58094796018361e-05, "loss": 1.7117, "step": 19500 }, { "epoch": 0.26, "learning_rate": 4.5702030360857536e-05, "loss": 1.6933, "step": 20000 }, { "epoch": 0.26, "learning_rate": 4.5594581119878974e-05, "loss": 1.6761, "step": 20500 }, { "epoch": 0.27, "learning_rate": 4.5487131878900405e-05, "loss": 1.691, "step": 21000 }, { "epoch": 0.28, "learning_rate": 4.537968263792185e-05, "loss": 1.6856, "step": 21500 }, { "epoch": 0.28, "learning_rate": 4.527223339694329e-05, "loss": 1.7046, "step": 22000 }, { "epoch": 0.29, "learning_rate": 4.516478415596473e-05, "loss": 1.6825, "step": 22500 }, { "epoch": 0.3, "learning_rate": 4.505733491498616e-05, "loss": 1.6776, "step": 23000 }, { "epoch": 0.3, "learning_rate": 4.49498856740076e-05, "loss": 1.6698, "step": 23500 }, { "epoch": 0.31, "learning_rate": 4.484243643302904e-05, "loss": 1.6715, "step": 24000 }, { "epoch": 0.32, "learning_rate": 4.473498719205047e-05, "loss": 1.6865, "step": 24500 }, { "epoch": 0.32, "learning_rate": 4.462753795107192e-05, "loss": 1.6776, "step": 25000 }, { "epoch": 0.33, "learning_rate": 4.4520088710093356e-05, "loss": 1.6831, "step": 25500 }, { "epoch": 0.34, "learning_rate": 4.4412639469114794e-05, "loss": 1.6748, "step": 26000 }, { "epoch": 0.34, "learning_rate": 4.4305190228136225e-05, "loss": 1.7095, "step": 26500 }, { "epoch": 0.35, "learning_rate": 4.419774098715767e-05, "loss": 1.6873, "step": 27000 }, { "epoch": 0.35, "learning_rate": 4.409029174617911e-05, "loss": 1.6689, "step": 27500 }, { "epoch": 0.36, "learning_rate": 4.3982842505200547e-05, "loss": 1.6597, "step": 28000 }, { "epoch": 0.37, "learning_rate": 4.387539326422198e-05, "loss": 1.6801, "step": 28500 }, { "epoch": 0.37, "learning_rate": 4.376794402324342e-05, "loss": 1.6833, "step": 29000 }, { "epoch": 0.38, "learning_rate": 4.366049478226486e-05, "loss": 1.6795, "step": 29500 }, { "epoch": 0.39, "learning_rate": 4.35530455412863e-05, "loss": 1.6787, "step": 30000 }, { "epoch": 0.39, "learning_rate": 4.344559630030774e-05, "loss": 1.6615, "step": 30500 }, { "epoch": 0.4, "learning_rate": 4.3338147059329176e-05, "loss": 1.6579, "step": 31000 }, { "epoch": 0.41, "learning_rate": 4.3230697818350614e-05, "loss": 1.6736, "step": 31500 }, { "epoch": 0.41, "learning_rate": 4.312324857737205e-05, "loss": 1.6589, "step": 32000 }, { "epoch": 0.42, "learning_rate": 4.301579933639349e-05, "loss": 1.6583, "step": 32500 }, { "epoch": 0.43, "learning_rate": 4.290835009541493e-05, "loss": 1.6816, "step": 33000 }, { "epoch": 0.43, "learning_rate": 4.2800900854436366e-05, "loss": 1.66, "step": 33500 }, { "epoch": 0.44, "learning_rate": 4.2693451613457805e-05, "loss": 1.6679, "step": 34000 }, { "epoch": 0.44, "learning_rate": 4.258600237247924e-05, "loss": 1.6608, "step": 34500 }, { "epoch": 0.45, "learning_rate": 4.247855313150068e-05, "loss": 1.6665, "step": 35000 }, { "epoch": 0.46, "learning_rate": 4.237110389052212e-05, "loss": 1.651, "step": 35500 }, { "epoch": 0.46, "learning_rate": 4.226365464954356e-05, "loss": 1.6525, "step": 36000 }, { "epoch": 0.47, "learning_rate": 4.2156205408564995e-05, "loss": 1.6436, "step": 36500 }, { "epoch": 0.48, "learning_rate": 4.2048756167586434e-05, "loss": 1.656, "step": 37000 }, { "epoch": 0.48, "learning_rate": 4.194130692660787e-05, "loss": 1.6529, "step": 37500 }, { "epoch": 0.49, "learning_rate": 4.183385768562931e-05, "loss": 1.6774, "step": 38000 }, { "epoch": 0.5, "learning_rate": 4.172640844465075e-05, "loss": 1.6629, "step": 38500 }, { "epoch": 0.5, "learning_rate": 4.1618959203672186e-05, "loss": 1.6407, "step": 39000 }, { "epoch": 0.51, "learning_rate": 4.1511509962693624e-05, "loss": 1.6512, "step": 39500 }, { "epoch": 0.52, "learning_rate": 4.140406072171506e-05, "loss": 1.625, "step": 40000 }, { "epoch": 0.52, "learning_rate": 4.12966114807365e-05, "loss": 1.6338, "step": 40500 }, { "epoch": 0.53, "learning_rate": 4.118916223975794e-05, "loss": 1.6484, "step": 41000 }, { "epoch": 0.54, "learning_rate": 4.108171299877938e-05, "loss": 1.6364, "step": 41500 }, { "epoch": 0.54, "learning_rate": 4.0974263757800815e-05, "loss": 1.6493, "step": 42000 }, { "epoch": 0.55, "learning_rate": 4.086681451682225e-05, "loss": 1.6353, "step": 42500 }, { "epoch": 0.55, "learning_rate": 4.075936527584369e-05, "loss": 1.6503, "step": 43000 }, { "epoch": 0.56, "learning_rate": 4.0651916034865136e-05, "loss": 1.6469, "step": 43500 }, { "epoch": 0.57, "learning_rate": 4.054446679388657e-05, "loss": 1.642, "step": 44000 }, { "epoch": 0.57, "learning_rate": 4.0437017552908006e-05, "loss": 1.6494, "step": 44500 }, { "epoch": 0.58, "learning_rate": 4.0329568311929444e-05, "loss": 1.6473, "step": 45000 }, { "epoch": 0.59, "learning_rate": 4.022211907095089e-05, "loss": 1.6371, "step": 45500 }, { "epoch": 0.59, "learning_rate": 4.011466982997232e-05, "loss": 1.6472, "step": 46000 }, { "epoch": 0.6, "learning_rate": 4.000722058899376e-05, "loss": 1.6131, "step": 46500 }, { "epoch": 0.61, "learning_rate": 3.98997713480152e-05, "loss": 1.6361, "step": 47000 }, { "epoch": 0.61, "learning_rate": 3.979232210703664e-05, "loss": 1.6342, "step": 47500 }, { "epoch": 0.62, "learning_rate": 3.968487286605807e-05, "loss": 1.6432, "step": 48000 }, { "epoch": 0.63, "learning_rate": 3.957742362507951e-05, "loss": 1.6355, "step": 48500 }, { "epoch": 0.63, "learning_rate": 3.9469974384100956e-05, "loss": 1.6419, "step": 49000 }, { "epoch": 0.64, "learning_rate": 3.9362525143122394e-05, "loss": 1.6376, "step": 49500 }, { "epoch": 0.64, "learning_rate": 3.9255075902143826e-05, "loss": 1.6357, "step": 50000 }, { "epoch": 0.65, "learning_rate": 3.9147626661165264e-05, "loss": 1.6393, "step": 50500 }, { "epoch": 0.66, "learning_rate": 3.904017742018671e-05, "loss": 1.6097, "step": 51000 }, { "epoch": 0.66, "learning_rate": 3.893272817920815e-05, "loss": 1.6309, "step": 51500 }, { "epoch": 0.67, "learning_rate": 3.882527893822958e-05, "loss": 1.6229, "step": 52000 }, { "epoch": 0.68, "learning_rate": 3.871782969725102e-05, "loss": 1.6481, "step": 52500 }, { "epoch": 0.68, "learning_rate": 3.861038045627246e-05, "loss": 1.6373, "step": 53000 }, { "epoch": 0.69, "learning_rate": 3.85029312152939e-05, "loss": 1.6124, "step": 53500 }, { "epoch": 0.7, "learning_rate": 3.839548197431533e-05, "loss": 1.6233, "step": 54000 }, { "epoch": 0.7, "learning_rate": 3.8288032733336776e-05, "loss": 1.6425, "step": 54500 }, { "epoch": 0.71, "learning_rate": 3.8180583492358214e-05, "loss": 1.6305, "step": 55000 }, { "epoch": 0.72, "learning_rate": 3.807313425137965e-05, "loss": 1.6294, "step": 55500 }, { "epoch": 0.72, "learning_rate": 3.7965685010401084e-05, "loss": 1.6034, "step": 56000 }, { "epoch": 0.73, "learning_rate": 3.785823576942253e-05, "loss": 1.6365, "step": 56500 }, { "epoch": 0.73, "learning_rate": 3.775078652844397e-05, "loss": 1.6147, "step": 57000 }, { "epoch": 0.74, "learning_rate": 3.7643337287465405e-05, "loss": 1.6186, "step": 57500 }, { "epoch": 0.75, "learning_rate": 3.7535888046486837e-05, "loss": 1.618, "step": 58000 }, { "epoch": 0.75, "learning_rate": 3.742843880550828e-05, "loss": 1.6273, "step": 58500 }, { "epoch": 0.76, "learning_rate": 3.732098956452972e-05, "loss": 1.6266, "step": 59000 }, { "epoch": 0.77, "learning_rate": 3.721354032355116e-05, "loss": 1.6176, "step": 59500 }, { "epoch": 0.77, "learning_rate": 3.7106091082572596e-05, "loss": 1.597, "step": 60000 }, { "epoch": 0.78, "learning_rate": 3.6998641841594034e-05, "loss": 1.6108, "step": 60500 }, { "epoch": 0.79, "learning_rate": 3.689119260061547e-05, "loss": 1.5991, "step": 61000 }, { "epoch": 0.79, "learning_rate": 3.678374335963691e-05, "loss": 1.6153, "step": 61500 }, { "epoch": 0.8, "learning_rate": 3.667629411865835e-05, "loss": 1.6101, "step": 62000 }, { "epoch": 0.81, "learning_rate": 3.656884487767979e-05, "loss": 1.6122, "step": 62500 }, { "epoch": 0.81, "learning_rate": 3.6461395636701225e-05, "loss": 1.6068, "step": 63000 }, { "epoch": 0.82, "learning_rate": 3.635394639572266e-05, "loss": 1.6066, "step": 63500 }, { "epoch": 0.83, "learning_rate": 3.62464971547441e-05, "loss": 1.6106, "step": 64000 }, { "epoch": 0.83, "learning_rate": 3.613904791376554e-05, "loss": 1.6088, "step": 64500 }, { "epoch": 0.84, "learning_rate": 3.603159867278698e-05, "loss": 1.6154, "step": 65000 }, { "epoch": 0.84, "learning_rate": 3.5924149431808416e-05, "loss": 1.6271, "step": 65500 }, { "epoch": 0.85, "learning_rate": 3.5816700190829854e-05, "loss": 1.5984, "step": 66000 }, { "epoch": 0.86, "learning_rate": 3.570925094985129e-05, "loss": 1.599, "step": 66500 }, { "epoch": 0.86, "learning_rate": 3.560180170887273e-05, "loss": 1.6013, "step": 67000 }, { "epoch": 0.87, "learning_rate": 3.549435246789417e-05, "loss": 1.6089, "step": 67500 }, { "epoch": 0.88, "learning_rate": 3.538690322691561e-05, "loss": 1.6076, "step": 68000 }, { "epoch": 0.88, "learning_rate": 3.5279453985937045e-05, "loss": 1.6111, "step": 68500 }, { "epoch": 0.89, "learning_rate": 3.517200474495848e-05, "loss": 1.5966, "step": 69000 }, { "epoch": 0.9, "learning_rate": 3.506455550397992e-05, "loss": 1.6066, "step": 69500 }, { "epoch": 0.9, "learning_rate": 3.495710626300136e-05, "loss": 1.6117, "step": 70000 }, { "epoch": 0.91, "learning_rate": 3.48496570220228e-05, "loss": 1.6075, "step": 70500 }, { "epoch": 0.92, "learning_rate": 3.4742207781044236e-05, "loss": 1.6064, "step": 71000 }, { "epoch": 0.92, "learning_rate": 3.4634758540065674e-05, "loss": 1.6041, "step": 71500 }, { "epoch": 0.93, "learning_rate": 3.452730929908711e-05, "loss": 1.6069, "step": 72000 }, { "epoch": 0.93, "learning_rate": 3.441986005810855e-05, "loss": 1.5887, "step": 72500 }, { "epoch": 0.94, "learning_rate": 3.4312410817129995e-05, "loss": 1.6004, "step": 73000 }, { "epoch": 0.95, "learning_rate": 3.4204961576151426e-05, "loss": 1.6039, "step": 73500 }, { "epoch": 0.95, "learning_rate": 3.4097512335172865e-05, "loss": 1.5959, "step": 74000 }, { "epoch": 0.96, "learning_rate": 3.39900630941943e-05, "loss": 1.5957, "step": 74500 }, { "epoch": 0.97, "learning_rate": 3.388261385321575e-05, "loss": 1.6003, "step": 75000 }, { "epoch": 0.97, "learning_rate": 3.377516461223718e-05, "loss": 1.5945, "step": 75500 }, { "epoch": 0.98, "learning_rate": 3.366771537125862e-05, "loss": 1.5928, "step": 76000 }, { "epoch": 0.99, "learning_rate": 3.3560266130280055e-05, "loss": 1.5914, "step": 76500 }, { "epoch": 0.99, "learning_rate": 3.34528168893015e-05, "loss": 1.5959, "step": 77000 }, { "epoch": 1.0, "learning_rate": 3.334536764832293e-05, "loss": 1.5916, "step": 77500 }, { "epoch": 1.01, "learning_rate": 3.323791840734437e-05, "loss": 1.5079, "step": 78000 }, { "epoch": 1.01, "learning_rate": 3.3130469166365815e-05, "loss": 1.4935, "step": 78500 }, { "epoch": 1.02, "learning_rate": 3.3023019925387246e-05, "loss": 1.4985, "step": 79000 }, { "epoch": 1.03, "learning_rate": 3.2915570684408684e-05, "loss": 1.4883, "step": 79500 }, { "epoch": 1.03, "learning_rate": 3.280812144343012e-05, "loss": 1.4964, "step": 80000 }, { "epoch": 1.04, "learning_rate": 3.270067220245157e-05, "loss": 1.4902, "step": 80500 }, { "epoch": 1.04, "learning_rate": 3.2593222961473e-05, "loss": 1.4874, "step": 81000 }, { "epoch": 1.05, "learning_rate": 3.248577372049444e-05, "loss": 1.495, "step": 81500 }, { "epoch": 1.06, "learning_rate": 3.2378324479515875e-05, "loss": 1.4983, "step": 82000 }, { "epoch": 1.06, "learning_rate": 3.227087523853732e-05, "loss": 1.4897, "step": 82500 }, { "epoch": 1.07, "learning_rate": 3.216342599755875e-05, "loss": 1.4987, "step": 83000 }, { "epoch": 1.08, "learning_rate": 3.205597675658019e-05, "loss": 1.4935, "step": 83500 }, { "epoch": 1.08, "learning_rate": 3.1948527515601635e-05, "loss": 1.4903, "step": 84000 }, { "epoch": 1.09, "learning_rate": 3.184107827462307e-05, "loss": 1.5145, "step": 84500 }, { "epoch": 1.1, "learning_rate": 3.1733629033644504e-05, "loss": 1.5038, "step": 85000 }, { "epoch": 1.1, "learning_rate": 3.162617979266594e-05, "loss": 1.4882, "step": 85500 }, { "epoch": 1.11, "learning_rate": 3.151873055168739e-05, "loss": 1.5193, "step": 86000 }, { "epoch": 1.12, "learning_rate": 3.1411281310708826e-05, "loss": 1.5029, "step": 86500 }, { "epoch": 1.12, "learning_rate": 3.130383206973026e-05, "loss": 1.4921, "step": 87000 }, { "epoch": 1.13, "learning_rate": 3.1196382828751695e-05, "loss": 1.4987, "step": 87500 }, { "epoch": 1.13, "learning_rate": 3.108893358777314e-05, "loss": 1.4979, "step": 88000 }, { "epoch": 1.14, "learning_rate": 3.098148434679458e-05, "loss": 1.499, "step": 88500 }, { "epoch": 1.15, "learning_rate": 3.087403510581601e-05, "loss": 1.4909, "step": 89000 }, { "epoch": 1.15, "learning_rate": 3.0766585864837455e-05, "loss": 1.4813, "step": 89500 }, { "epoch": 1.16, "learning_rate": 3.065913662385889e-05, "loss": 1.5043, "step": 90000 }, { "epoch": 1.17, "learning_rate": 3.055168738288033e-05, "loss": 1.4965, "step": 90500 }, { "epoch": 1.17, "learning_rate": 3.0444238141901766e-05, "loss": 1.496, "step": 91000 }, { "epoch": 1.18, "learning_rate": 3.0336788900923207e-05, "loss": 1.5061, "step": 91500 }, { "epoch": 1.19, "learning_rate": 3.0229339659944645e-05, "loss": 1.4969, "step": 92000 }, { "epoch": 1.19, "learning_rate": 3.012189041896608e-05, "loss": 1.4818, "step": 92500 }, { "epoch": 1.2, "learning_rate": 3.001444117798752e-05, "loss": 1.5003, "step": 93000 }, { "epoch": 1.21, "learning_rate": 2.990699193700896e-05, "loss": 1.5095, "step": 93500 }, { "epoch": 1.21, "learning_rate": 2.9799542696030398e-05, "loss": 1.4885, "step": 94000 }, { "epoch": 1.22, "learning_rate": 2.9692093455051833e-05, "loss": 1.5077, "step": 94500 }, { "epoch": 1.22, "learning_rate": 2.958464421407327e-05, "loss": 1.4981, "step": 95000 }, { "epoch": 1.23, "learning_rate": 2.9477194973094713e-05, "loss": 1.4877, "step": 95500 }, { "epoch": 1.24, "learning_rate": 2.936974573211615e-05, "loss": 1.5061, "step": 96000 }, { "epoch": 1.24, "learning_rate": 2.9262296491137586e-05, "loss": 1.4959, "step": 96500 }, { "epoch": 1.25, "learning_rate": 2.9154847250159027e-05, "loss": 1.4918, "step": 97000 }, { "epoch": 1.26, "learning_rate": 2.9047398009180465e-05, "loss": 1.5064, "step": 97500 }, { "epoch": 1.26, "learning_rate": 2.8939948768201903e-05, "loss": 1.5053, "step": 98000 }, { "epoch": 1.27, "learning_rate": 2.8832499527223338e-05, "loss": 1.4978, "step": 98500 }, { "epoch": 1.28, "learning_rate": 2.872505028624478e-05, "loss": 1.5012, "step": 99000 }, { "epoch": 1.28, "learning_rate": 2.8617601045266218e-05, "loss": 1.4973, "step": 99500 }, { "epoch": 1.29, "learning_rate": 2.8510151804287656e-05, "loss": 1.4979, "step": 100000 }, { "epoch": 1.3, "learning_rate": 2.840270256330909e-05, "loss": 1.5095, "step": 100500 }, { "epoch": 1.3, "learning_rate": 2.8295253322330532e-05, "loss": 1.4894, "step": 101000 }, { "epoch": 1.31, "learning_rate": 2.818780408135197e-05, "loss": 1.507, "step": 101500 }, { "epoch": 1.32, "learning_rate": 2.808035484037341e-05, "loss": 1.5054, "step": 102000 }, { "epoch": 1.32, "learning_rate": 2.797290559939485e-05, "loss": 1.4919, "step": 102500 }, { "epoch": 1.33, "learning_rate": 2.7865456358416285e-05, "loss": 1.4976, "step": 103000 }, { "epoch": 1.33, "learning_rate": 2.7758007117437723e-05, "loss": 1.4988, "step": 103500 }, { "epoch": 1.34, "learning_rate": 2.765055787645916e-05, "loss": 1.5031, "step": 104000 }, { "epoch": 1.35, "learning_rate": 2.7543108635480603e-05, "loss": 1.5058, "step": 104500 }, { "epoch": 1.35, "learning_rate": 2.7435659394502038e-05, "loss": 1.5091, "step": 105000 }, { "epoch": 1.36, "learning_rate": 2.7328210153523476e-05, "loss": 1.4957, "step": 105500 }, { "epoch": 1.37, "learning_rate": 2.7220760912544914e-05, "loss": 1.5023, "step": 106000 }, { "epoch": 1.37, "learning_rate": 2.7113311671566356e-05, "loss": 1.4898, "step": 106500 }, { "epoch": 1.38, "learning_rate": 2.700586243058779e-05, "loss": 1.5055, "step": 107000 }, { "epoch": 1.39, "learning_rate": 2.689841318960923e-05, "loss": 1.4897, "step": 107500 }, { "epoch": 1.39, "learning_rate": 2.679096394863067e-05, "loss": 1.5013, "step": 108000 }, { "epoch": 1.4, "learning_rate": 2.6683514707652108e-05, "loss": 1.4905, "step": 108500 }, { "epoch": 1.41, "learning_rate": 2.6576065466673543e-05, "loss": 1.4862, "step": 109000 }, { "epoch": 1.41, "learning_rate": 2.646861622569498e-05, "loss": 1.4913, "step": 109500 }, { "epoch": 1.42, "learning_rate": 2.6361166984716423e-05, "loss": 1.4927, "step": 110000 }, { "epoch": 1.42, "learning_rate": 2.625371774373786e-05, "loss": 1.484, "step": 110500 }, { "epoch": 1.43, "learning_rate": 2.6146268502759296e-05, "loss": 1.4896, "step": 111000 }, { "epoch": 1.44, "learning_rate": 2.6038819261780734e-05, "loss": 1.4967, "step": 111500 }, { "epoch": 1.44, "learning_rate": 2.5931370020802175e-05, "loss": 1.5012, "step": 112000 }, { "epoch": 1.45, "learning_rate": 2.5823920779823614e-05, "loss": 1.5021, "step": 112500 }, { "epoch": 1.46, "learning_rate": 2.571647153884505e-05, "loss": 1.5086, "step": 113000 }, { "epoch": 1.46, "learning_rate": 2.5609022297866493e-05, "loss": 1.4857, "step": 113500 }, { "epoch": 1.47, "learning_rate": 2.5501573056887928e-05, "loss": 1.4873, "step": 114000 }, { "epoch": 1.48, "learning_rate": 2.5394123815909366e-05, "loss": 1.485, "step": 114500 }, { "epoch": 1.48, "learning_rate": 2.52866745749308e-05, "loss": 1.4755, "step": 115000 }, { "epoch": 1.49, "learning_rate": 2.5179225333952246e-05, "loss": 1.4852, "step": 115500 }, { "epoch": 1.5, "learning_rate": 2.507177609297368e-05, "loss": 1.4809, "step": 116000 }, { "epoch": 1.5, "learning_rate": 2.496432685199512e-05, "loss": 1.4845, "step": 116500 }, { "epoch": 1.51, "learning_rate": 2.4856877611016557e-05, "loss": 1.487, "step": 117000 }, { "epoch": 1.52, "learning_rate": 2.4749428370037995e-05, "loss": 1.474, "step": 117500 }, { "epoch": 1.52, "learning_rate": 2.4641979129059433e-05, "loss": 1.5064, "step": 118000 }, { "epoch": 1.53, "learning_rate": 2.453452988808087e-05, "loss": 1.491, "step": 118500 }, { "epoch": 1.53, "learning_rate": 2.442708064710231e-05, "loss": 1.4957, "step": 119000 }, { "epoch": 1.54, "learning_rate": 2.4319631406123748e-05, "loss": 1.4837, "step": 119500 }, { "epoch": 1.55, "learning_rate": 2.4212182165145186e-05, "loss": 1.4846, "step": 120000 }, { "epoch": 1.55, "learning_rate": 2.4104732924166624e-05, "loss": 1.4952, "step": 120500 }, { "epoch": 1.56, "learning_rate": 2.3997283683188062e-05, "loss": 1.4772, "step": 121000 }, { "epoch": 1.57, "learning_rate": 2.38898344422095e-05, "loss": 1.4906, "step": 121500 }, { "epoch": 1.57, "learning_rate": 2.378238520123094e-05, "loss": 1.4916, "step": 122000 }, { "epoch": 1.58, "learning_rate": 2.3674935960252377e-05, "loss": 1.4888, "step": 122500 }, { "epoch": 1.59, "learning_rate": 2.3567486719273815e-05, "loss": 1.4971, "step": 123000 }, { "epoch": 1.59, "learning_rate": 2.3460037478295253e-05, "loss": 1.4929, "step": 123500 }, { "epoch": 1.6, "learning_rate": 2.335258823731669e-05, "loss": 1.4798, "step": 124000 }, { "epoch": 1.61, "learning_rate": 2.324513899633813e-05, "loss": 1.4834, "step": 124500 }, { "epoch": 1.61, "learning_rate": 2.313768975535957e-05, "loss": 1.4751, "step": 125000 }, { "epoch": 1.62, "learning_rate": 2.3030240514381006e-05, "loss": 1.4937, "step": 125500 }, { "epoch": 1.62, "learning_rate": 2.2922791273402448e-05, "loss": 1.4854, "step": 126000 }, { "epoch": 1.63, "learning_rate": 2.2815342032423882e-05, "loss": 1.4865, "step": 126500 }, { "epoch": 1.64, "learning_rate": 2.2707892791445324e-05, "loss": 1.4818, "step": 127000 }, { "epoch": 1.64, "learning_rate": 2.260044355046676e-05, "loss": 1.4887, "step": 127500 }, { "epoch": 1.65, "learning_rate": 2.24929943094882e-05, "loss": 1.4839, "step": 128000 }, { "epoch": 1.66, "learning_rate": 2.2385545068509635e-05, "loss": 1.4854, "step": 128500 }, { "epoch": 1.66, "learning_rate": 2.2278095827531077e-05, "loss": 1.4915, "step": 129000 }, { "epoch": 1.67, "learning_rate": 2.217064658655251e-05, "loss": 1.4768, "step": 129500 }, { "epoch": 1.68, "learning_rate": 2.2063197345573953e-05, "loss": 1.5006, "step": 130000 }, { "epoch": 1.68, "learning_rate": 2.1955748104595388e-05, "loss": 1.4758, "step": 130500 }, { "epoch": 1.69, "learning_rate": 2.184829886361683e-05, "loss": 1.4815, "step": 131000 }, { "epoch": 1.7, "learning_rate": 2.1740849622638267e-05, "loss": 1.4841, "step": 131500 }, { "epoch": 1.7, "learning_rate": 2.1633400381659706e-05, "loss": 1.4846, "step": 132000 }, { "epoch": 1.71, "learning_rate": 2.1525951140681144e-05, "loss": 1.4824, "step": 132500 }, { "epoch": 1.71, "learning_rate": 2.1418501899702582e-05, "loss": 1.4792, "step": 133000 }, { "epoch": 1.72, "learning_rate": 2.131105265872402e-05, "loss": 1.4694, "step": 133500 }, { "epoch": 1.73, "learning_rate": 2.1203603417745458e-05, "loss": 1.4811, "step": 134000 }, { "epoch": 1.73, "learning_rate": 2.1096154176766896e-05, "loss": 1.4829, "step": 134500 }, { "epoch": 1.74, "learning_rate": 2.0988704935788335e-05, "loss": 1.4703, "step": 135000 }, { "epoch": 1.75, "learning_rate": 2.0881255694809773e-05, "loss": 1.4804, "step": 135500 }, { "epoch": 1.75, "learning_rate": 2.077380645383121e-05, "loss": 1.4774, "step": 136000 }, { "epoch": 1.76, "learning_rate": 2.066635721285265e-05, "loss": 1.4853, "step": 136500 }, { "epoch": 1.77, "learning_rate": 2.055890797187409e-05, "loss": 1.4725, "step": 137000 }, { "epoch": 1.77, "learning_rate": 2.0451458730895525e-05, "loss": 1.4884, "step": 137500 }, { "epoch": 1.78, "learning_rate": 2.0344009489916967e-05, "loss": 1.4692, "step": 138000 }, { "epoch": 1.79, "learning_rate": 2.02365602489384e-05, "loss": 1.4681, "step": 138500 }, { "epoch": 1.79, "learning_rate": 2.0129111007959843e-05, "loss": 1.4807, "step": 139000 }, { "epoch": 1.8, "learning_rate": 2.0021661766981278e-05, "loss": 1.4707, "step": 139500 }, { "epoch": 1.81, "learning_rate": 1.991421252600272e-05, "loss": 1.4773, "step": 140000 }, { "epoch": 1.81, "learning_rate": 1.9806763285024154e-05, "loss": 1.4763, "step": 140500 }, { "epoch": 1.82, "learning_rate": 1.9699314044045593e-05, "loss": 1.466, "step": 141000 }, { "epoch": 1.82, "learning_rate": 1.959186480306703e-05, "loss": 1.4824, "step": 141500 }, { "epoch": 1.83, "learning_rate": 1.948441556208847e-05, "loss": 1.4726, "step": 142000 }, { "epoch": 1.84, "learning_rate": 1.9376966321109907e-05, "loss": 1.4636, "step": 142500 }, { "epoch": 1.84, "learning_rate": 1.9269517080131345e-05, "loss": 1.4791, "step": 143000 }, { "epoch": 1.85, "learning_rate": 1.9162067839152787e-05, "loss": 1.4701, "step": 143500 }, { "epoch": 1.86, "learning_rate": 1.905461859817422e-05, "loss": 1.4937, "step": 144000 }, { "epoch": 1.86, "learning_rate": 1.8947169357195663e-05, "loss": 1.493, "step": 144500 }, { "epoch": 1.87, "learning_rate": 1.8839720116217098e-05, "loss": 1.479, "step": 145000 }, { "epoch": 1.88, "learning_rate": 1.873227087523854e-05, "loss": 1.4797, "step": 145500 }, { "epoch": 1.88, "learning_rate": 1.8624821634259974e-05, "loss": 1.4672, "step": 146000 }, { "epoch": 1.89, "learning_rate": 1.8517372393281416e-05, "loss": 1.4731, "step": 146500 }, { "epoch": 1.9, "learning_rate": 1.840992315230285e-05, "loss": 1.4639, "step": 147000 }, { "epoch": 1.9, "learning_rate": 1.8302473911324292e-05, "loss": 1.4727, "step": 147500 }, { "epoch": 1.91, "learning_rate": 1.8195024670345727e-05, "loss": 1.4823, "step": 148000 }, { "epoch": 1.91, "learning_rate": 1.808757542936717e-05, "loss": 1.473, "step": 148500 }, { "epoch": 1.92, "learning_rate": 1.7980126188388607e-05, "loss": 1.4702, "step": 149000 }, { "epoch": 1.93, "learning_rate": 1.7872676947410045e-05, "loss": 1.4748, "step": 149500 }, { "epoch": 1.93, "learning_rate": 1.7765227706431483e-05, "loss": 1.4712, "step": 150000 }, { "epoch": 1.94, "learning_rate": 1.765777846545292e-05, "loss": 1.4874, "step": 150500 }, { "epoch": 1.95, "learning_rate": 1.755032922447436e-05, "loss": 1.4633, "step": 151000 }, { "epoch": 1.95, "learning_rate": 1.7442879983495797e-05, "loss": 1.4532, "step": 151500 }, { "epoch": 1.96, "learning_rate": 1.7335430742517236e-05, "loss": 1.4656, "step": 152000 }, { "epoch": 1.97, "learning_rate": 1.7227981501538674e-05, "loss": 1.4622, "step": 152500 }, { "epoch": 1.97, "learning_rate": 1.7120532260560112e-05, "loss": 1.4802, "step": 153000 }, { "epoch": 1.98, "learning_rate": 1.701308301958155e-05, "loss": 1.4652, "step": 153500 }, { "epoch": 1.99, "learning_rate": 1.6905633778602988e-05, "loss": 1.476, "step": 154000 }, { "epoch": 1.99, "learning_rate": 1.6798184537624426e-05, "loss": 1.4731, "step": 154500 }, { "epoch": 2.0, "learning_rate": 1.6690735296645865e-05, "loss": 1.4571, "step": 155000 }, { "epoch": 2.01, "learning_rate": 1.6583286055667306e-05, "loss": 1.3858, "step": 155500 }, { "epoch": 2.01, "learning_rate": 1.647583681468874e-05, "loss": 1.39, "step": 156000 }, { "epoch": 2.02, "learning_rate": 1.6368387573710182e-05, "loss": 1.3756, "step": 156500 }, { "epoch": 2.02, "learning_rate": 1.6260938332731617e-05, "loss": 1.3864, "step": 157000 }, { "epoch": 2.03, "learning_rate": 1.615348909175306e-05, "loss": 1.3862, "step": 157500 }, { "epoch": 2.04, "learning_rate": 1.6046039850774494e-05, "loss": 1.3787, "step": 158000 }, { "epoch": 2.04, "learning_rate": 1.5938590609795935e-05, "loss": 1.3957, "step": 158500 }, { "epoch": 2.05, "learning_rate": 1.583114136881737e-05, "loss": 1.3872, "step": 159000 }, { "epoch": 2.06, "learning_rate": 1.572369212783881e-05, "loss": 1.3876, "step": 159500 }, { "epoch": 2.06, "learning_rate": 1.5616242886860246e-05, "loss": 1.3809, "step": 160000 }, { "epoch": 2.07, "learning_rate": 1.5508793645881688e-05, "loss": 1.3876, "step": 160500 }, { "epoch": 2.08, "learning_rate": 1.5401344404903126e-05, "loss": 1.3877, "step": 161000 }, { "epoch": 2.08, "learning_rate": 1.5293895163924564e-05, "loss": 1.3849, "step": 161500 }, { "epoch": 2.09, "learning_rate": 1.5186445922946002e-05, "loss": 1.3926, "step": 162000 }, { "epoch": 2.1, "learning_rate": 1.5078996681967439e-05, "loss": 1.3987, "step": 162500 }, { "epoch": 2.1, "learning_rate": 1.4971547440988879e-05, "loss": 1.3969, "step": 163000 }, { "epoch": 2.11, "learning_rate": 1.4864098200010315e-05, "loss": 1.3839, "step": 163500 }, { "epoch": 2.11, "learning_rate": 1.4756648959031755e-05, "loss": 1.3856, "step": 164000 }, { "epoch": 2.12, "learning_rate": 1.4649199718053191e-05, "loss": 1.3782, "step": 164500 }, { "epoch": 2.13, "learning_rate": 1.4541750477074631e-05, "loss": 1.3977, "step": 165000 }, { "epoch": 2.13, "learning_rate": 1.4434301236096068e-05, "loss": 1.3821, "step": 165500 }, { "epoch": 2.14, "learning_rate": 1.4326851995117508e-05, "loss": 1.3767, "step": 166000 }, { "epoch": 2.15, "learning_rate": 1.4219402754138944e-05, "loss": 1.3843, "step": 166500 }, { "epoch": 2.15, "learning_rate": 1.4111953513160384e-05, "loss": 1.4, "step": 167000 }, { "epoch": 2.16, "learning_rate": 1.4004504272181824e-05, "loss": 1.3805, "step": 167500 }, { "epoch": 2.17, "learning_rate": 1.389705503120326e-05, "loss": 1.3773, "step": 168000 }, { "epoch": 2.17, "learning_rate": 1.37896057902247e-05, "loss": 1.3999, "step": 168500 }, { "epoch": 2.18, "learning_rate": 1.3682156549246137e-05, "loss": 1.4037, "step": 169000 }, { "epoch": 2.19, "learning_rate": 1.3574707308267577e-05, "loss": 1.3802, "step": 169500 }, { "epoch": 2.19, "learning_rate": 1.3467258067289013e-05, "loss": 1.3954, "step": 170000 }, { "epoch": 2.2, "learning_rate": 1.3359808826310453e-05, "loss": 1.3859, "step": 170500 }, { "epoch": 2.2, "learning_rate": 1.325235958533189e-05, "loss": 1.3907, "step": 171000 }, { "epoch": 2.21, "learning_rate": 1.314491034435333e-05, "loss": 1.3807, "step": 171500 }, { "epoch": 2.22, "learning_rate": 1.3037461103374766e-05, "loss": 1.386, "step": 172000 }, { "epoch": 2.22, "learning_rate": 1.2930011862396206e-05, "loss": 1.3791, "step": 172500 }, { "epoch": 2.23, "learning_rate": 1.2822562621417644e-05, "loss": 1.3836, "step": 173000 }, { "epoch": 2.24, "learning_rate": 1.271511338043908e-05, "loss": 1.3988, "step": 173500 }, { "epoch": 2.24, "learning_rate": 1.260766413946052e-05, "loss": 1.3981, "step": 174000 }, { "epoch": 2.25, "learning_rate": 1.2500214898481956e-05, "loss": 1.384, "step": 174500 }, { "epoch": 2.26, "learning_rate": 1.2392765657503395e-05, "loss": 1.3789, "step": 175000 }, { "epoch": 2.26, "learning_rate": 1.2285316416524835e-05, "loss": 1.3788, "step": 175500 }, { "epoch": 2.27, "learning_rate": 1.2177867175546273e-05, "loss": 1.3833, "step": 176000 }, { "epoch": 2.28, "learning_rate": 1.207041793456771e-05, "loss": 1.395, "step": 176500 }, { "epoch": 2.28, "learning_rate": 1.1962968693589149e-05, "loss": 1.3814, "step": 177000 }, { "epoch": 2.29, "learning_rate": 1.1855519452610587e-05, "loss": 1.3891, "step": 177500 }, { "epoch": 2.3, "learning_rate": 1.1748070211632025e-05, "loss": 1.3862, "step": 178000 }, { "epoch": 2.3, "learning_rate": 1.1640620970653464e-05, "loss": 1.3825, "step": 178500 }, { "epoch": 2.31, "learning_rate": 1.1533171729674902e-05, "loss": 1.3829, "step": 179000 }, { "epoch": 2.31, "learning_rate": 1.142572248869634e-05, "loss": 1.3902, "step": 179500 }, { "epoch": 2.32, "learning_rate": 1.1318273247717778e-05, "loss": 1.3772, "step": 180000 }, { "epoch": 2.33, "learning_rate": 1.1210824006739216e-05, "loss": 1.3812, "step": 180500 }, { "epoch": 2.33, "learning_rate": 1.1103374765760654e-05, "loss": 1.3847, "step": 181000 }, { "epoch": 2.34, "learning_rate": 1.0995925524782094e-05, "loss": 1.3717, "step": 181500 }, { "epoch": 2.35, "learning_rate": 1.0888476283803532e-05, "loss": 1.3807, "step": 182000 }, { "epoch": 2.35, "learning_rate": 1.078102704282497e-05, "loss": 1.386, "step": 182500 }, { "epoch": 2.36, "learning_rate": 1.0673577801846409e-05, "loss": 1.3868, "step": 183000 }, { "epoch": 2.37, "learning_rate": 1.0566128560867847e-05, "loss": 1.381, "step": 183500 }, { "epoch": 2.37, "learning_rate": 1.0458679319889285e-05, "loss": 1.3821, "step": 184000 }, { "epoch": 2.38, "learning_rate": 1.0351230078910723e-05, "loss": 1.3698, "step": 184500 }, { "epoch": 2.39, "learning_rate": 1.0243780837932161e-05, "loss": 1.3901, "step": 185000 }, { "epoch": 2.39, "learning_rate": 1.01363315969536e-05, "loss": 1.3899, "step": 185500 }, { "epoch": 2.4, "learning_rate": 1.0028882355975038e-05, "loss": 1.4006, "step": 186000 }, { "epoch": 2.4, "learning_rate": 9.921433114996476e-06, "loss": 1.3882, "step": 186500 }, { "epoch": 2.41, "learning_rate": 9.813983874017914e-06, "loss": 1.3883, "step": 187000 }, { "epoch": 2.42, "learning_rate": 9.706534633039354e-06, "loss": 1.3821, "step": 187500 }, { "epoch": 2.42, "learning_rate": 9.599085392060792e-06, "loss": 1.3829, "step": 188000 }, { "epoch": 2.43, "learning_rate": 9.49163615108223e-06, "loss": 1.3901, "step": 188500 }, { "epoch": 2.44, "learning_rate": 9.384186910103668e-06, "loss": 1.3812, "step": 189000 }, { "epoch": 2.44, "learning_rate": 9.276737669125107e-06, "loss": 1.3726, "step": 189500 }, { "epoch": 2.45, "learning_rate": 9.169288428146545e-06, "loss": 1.3926, "step": 190000 }, { "epoch": 2.46, "learning_rate": 9.061839187167983e-06, "loss": 1.3834, "step": 190500 }, { "epoch": 2.46, "learning_rate": 8.954389946189421e-06, "loss": 1.3774, "step": 191000 }, { "epoch": 2.47, "learning_rate": 8.84694070521086e-06, "loss": 1.3773, "step": 191500 }, { "epoch": 2.48, "learning_rate": 8.739491464232297e-06, "loss": 1.3942, "step": 192000 }, { "epoch": 2.48, "learning_rate": 8.632042223253736e-06, "loss": 1.385, "step": 192500 }, { "epoch": 2.49, "learning_rate": 8.524592982275174e-06, "loss": 1.3853, "step": 193000 }, { "epoch": 2.49, "learning_rate": 8.417143741296612e-06, "loss": 1.3675, "step": 193500 }, { "epoch": 2.5, "learning_rate": 8.30969450031805e-06, "loss": 1.3836, "step": 194000 }, { "epoch": 2.51, "learning_rate": 8.202245259339488e-06, "loss": 1.3815, "step": 194500 }, { "epoch": 2.51, "learning_rate": 8.094796018360926e-06, "loss": 1.3839, "step": 195000 }, { "epoch": 2.52, "learning_rate": 7.987346777382365e-06, "loss": 1.376, "step": 195500 }, { "epoch": 2.53, "learning_rate": 7.879897536403803e-06, "loss": 1.3788, "step": 196000 }, { "epoch": 2.53, "learning_rate": 7.772448295425241e-06, "loss": 1.3846, "step": 196500 }, { "epoch": 2.54, "learning_rate": 7.664999054446679e-06, "loss": 1.3923, "step": 197000 }, { "epoch": 2.55, "learning_rate": 7.557549813468118e-06, "loss": 1.3731, "step": 197500 }, { "epoch": 2.55, "learning_rate": 7.450100572489556e-06, "loss": 1.3886, "step": 198000 }, { "epoch": 2.56, "learning_rate": 7.342651331510994e-06, "loss": 1.3817, "step": 198500 }, { "epoch": 2.57, "learning_rate": 7.2352020905324326e-06, "loss": 1.3872, "step": 199000 }, { "epoch": 2.57, "learning_rate": 7.127752849553872e-06, "loss": 1.3788, "step": 199500 }, { "epoch": 2.58, "learning_rate": 7.02030360857531e-06, "loss": 1.3854, "step": 200000 }, { "epoch": 2.59, "learning_rate": 6.912854367596748e-06, "loss": 1.3679, "step": 200500 }, { "epoch": 2.59, "learning_rate": 6.805405126618186e-06, "loss": 1.381, "step": 201000 }, { "epoch": 2.6, "learning_rate": 6.697955885639624e-06, "loss": 1.383, "step": 201500 }, { "epoch": 2.6, "learning_rate": 6.5905066446610624e-06, "loss": 1.3757, "step": 202000 }, { "epoch": 2.61, "learning_rate": 6.483057403682501e-06, "loss": 1.3687, "step": 202500 }, { "epoch": 2.62, "learning_rate": 6.375608162703939e-06, "loss": 1.3739, "step": 203000 }, { "epoch": 2.62, "learning_rate": 6.268158921725377e-06, "loss": 1.3737, "step": 203500 }, { "epoch": 2.63, "learning_rate": 6.160709680746816e-06, "loss": 1.3787, "step": 204000 }, { "epoch": 2.64, "learning_rate": 6.053260439768254e-06, "loss": 1.3832, "step": 204500 }, { "epoch": 2.64, "learning_rate": 5.945811198789692e-06, "loss": 1.3734, "step": 205000 }, { "epoch": 2.65, "learning_rate": 5.8383619578111304e-06, "loss": 1.3847, "step": 205500 }, { "epoch": 2.66, "learning_rate": 5.730912716832569e-06, "loss": 1.3635, "step": 206000 }, { "epoch": 2.66, "learning_rate": 5.623463475854007e-06, "loss": 1.3768, "step": 206500 }, { "epoch": 2.67, "learning_rate": 5.516014234875446e-06, "loss": 1.3644, "step": 207000 }, { "epoch": 2.68, "learning_rate": 5.408564993896884e-06, "loss": 1.3766, "step": 207500 }, { "epoch": 2.68, "learning_rate": 5.301115752918321e-06, "loss": 1.3745, "step": 208000 }, { "epoch": 2.69, "learning_rate": 5.1936665119397594e-06, "loss": 1.3779, "step": 208500 }, { "epoch": 2.69, "learning_rate": 5.086217270961198e-06, "loss": 1.3693, "step": 209000 }, { "epoch": 2.7, "learning_rate": 4.978768029982636e-06, "loss": 1.366, "step": 209500 }, { "epoch": 2.71, "learning_rate": 4.871318789004075e-06, "loss": 1.3817, "step": 210000 }, { "epoch": 2.71, "learning_rate": 4.763869548025513e-06, "loss": 1.384, "step": 210500 }, { "epoch": 2.72, "learning_rate": 4.656420307046951e-06, "loss": 1.3692, "step": 211000 }, { "epoch": 2.73, "learning_rate": 4.548971066068389e-06, "loss": 1.3751, "step": 211500 }, { "epoch": 2.73, "learning_rate": 4.4415218250898275e-06, "loss": 1.3787, "step": 212000 }, { "epoch": 2.74, "learning_rate": 4.334072584111266e-06, "loss": 1.3943, "step": 212500 }, { "epoch": 2.75, "learning_rate": 4.226623343132705e-06, "loss": 1.3679, "step": 213000 }, { "epoch": 2.75, "learning_rate": 4.119174102154143e-06, "loss": 1.3642, "step": 213500 }, { "epoch": 2.76, "learning_rate": 4.011724861175581e-06, "loss": 1.3677, "step": 214000 }, { "epoch": 2.77, "learning_rate": 3.904275620197019e-06, "loss": 1.3725, "step": 214500 }, { "epoch": 2.77, "learning_rate": 3.7968263792184573e-06, "loss": 1.3699, "step": 215000 }, { "epoch": 2.78, "learning_rate": 3.6893771382398955e-06, "loss": 1.3676, "step": 215500 }, { "epoch": 2.79, "learning_rate": 3.581927897261334e-06, "loss": 1.3639, "step": 216000 }, { "epoch": 2.79, "learning_rate": 3.4744786562827722e-06, "loss": 1.3669, "step": 216500 }, { "epoch": 2.8, "learning_rate": 3.3670294153042104e-06, "loss": 1.3725, "step": 217000 }, { "epoch": 2.8, "learning_rate": 3.2595801743256486e-06, "loss": 1.3801, "step": 217500 }, { "epoch": 2.81, "learning_rate": 3.1521309333470867e-06, "loss": 1.3683, "step": 218000 }, { "epoch": 2.82, "learning_rate": 3.0446816923685253e-06, "loss": 1.3764, "step": 218500 }, { "epoch": 2.82, "learning_rate": 2.9372324513899635e-06, "loss": 1.3641, "step": 219000 }, { "epoch": 2.83, "learning_rate": 2.829783210411402e-06, "loss": 1.3707, "step": 219500 }, { "epoch": 2.84, "learning_rate": 2.7223339694328402e-06, "loss": 1.3741, "step": 220000 }, { "epoch": 2.84, "learning_rate": 2.6148847284542784e-06, "loss": 1.3637, "step": 220500 }, { "epoch": 2.85, "learning_rate": 2.5074354874757166e-06, "loss": 1.363, "step": 221000 }, { "epoch": 2.86, "learning_rate": 2.3999862464971547e-06, "loss": 1.3676, "step": 221500 }, { "epoch": 2.86, "learning_rate": 2.292537005518593e-06, "loss": 1.3681, "step": 222000 }, { "epoch": 2.87, "learning_rate": 2.1850877645400315e-06, "loss": 1.3713, "step": 222500 }, { "epoch": 2.88, "learning_rate": 2.0776385235614697e-06, "loss": 1.3704, "step": 223000 }, { "epoch": 2.88, "learning_rate": 1.970189282582908e-06, "loss": 1.3762, "step": 223500 }, { "epoch": 2.89, "learning_rate": 1.8627400416043464e-06, "loss": 1.3644, "step": 224000 }, { "epoch": 2.89, "learning_rate": 1.7552908006257844e-06, "loss": 1.3815, "step": 224500 }, { "epoch": 2.9, "learning_rate": 1.6478415596472225e-06, "loss": 1.3706, "step": 225000 }, { "epoch": 2.91, "learning_rate": 1.540392318668661e-06, "loss": 1.3705, "step": 225500 }, { "epoch": 2.91, "learning_rate": 1.4329430776900993e-06, "loss": 1.376, "step": 226000 }, { "epoch": 2.92, "learning_rate": 1.3254938367115377e-06, "loss": 1.3704, "step": 226500 }, { "epoch": 2.93, "learning_rate": 1.2180445957329758e-06, "loss": 1.3691, "step": 227000 }, { "epoch": 2.93, "learning_rate": 1.110595354754414e-06, "loss": 1.3753, "step": 227500 }, { "epoch": 2.94, "learning_rate": 1.0031461137758524e-06, "loss": 1.3553, "step": 228000 }, { "epoch": 2.95, "learning_rate": 8.956968727972906e-07, "loss": 1.3726, "step": 228500 }, { "epoch": 2.95, "learning_rate": 7.882476318187288e-07, "loss": 1.3704, "step": 229000 }, { "epoch": 2.96, "learning_rate": 6.807983908401671e-07, "loss": 1.3674, "step": 229500 }, { "epoch": 2.97, "learning_rate": 5.733491498616055e-07, "loss": 1.3578, "step": 230000 }, { "epoch": 2.97, "learning_rate": 4.658999088830437e-07, "loss": 1.3761, "step": 230500 }, { "epoch": 2.98, "learning_rate": 3.5845066790448197e-07, "loss": 1.3656, "step": 231000 }, { "epoch": 2.98, "learning_rate": 2.510014269259202e-07, "loss": 1.3663, "step": 231500 }, { "epoch": 2.99, "learning_rate": 1.4355218594735846e-07, "loss": 1.3605, "step": 232000 }, { "epoch": 3.0, "learning_rate": 3.610294496879674e-08, "loss": 1.367, "step": 232500 }, { "epoch": 3.0, "step": 232668, "total_flos": 6.971496629325005e+16, "train_loss": 1.5098053402305296, "train_runtime": 78093.1182, "train_samples_per_second": 47.669, "train_steps_per_second": 2.979 } ], "max_steps": 232668, "num_train_epochs": 3, "total_flos": 6.971496629325005e+16, "trial_name": null, "trial_params": null }