diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,7 +1,7 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 5.0761421319796955, + "epoch": 6.451612903225806, "global_step": 1000, "is_hyper_param_search": false, "is_local_process_zero": true, @@ -9,6017 +9,6017 @@ "log_history": [ { "epoch": 0.01, - "learning_rate": 0.0005, - "loss": 0.799, + "learning_rate": 5e-05, + "loss": 0.843, "step": 1 }, { "epoch": 0.01, - "learning_rate": 0.001, - "loss": 0.6675, + "learning_rate": 0.0001, + "loss": 0.7773, "step": 2 }, { "epoch": 0.02, - "learning_rate": 0.000999997522701653, - "loss": 0.999, + "learning_rate": 9.999975227016531e-05, + "loss": 0.8884, "step": 3 }, { - "epoch": 0.02, - "learning_rate": 0.0009999900908311602, - "loss": 0.7497, + "epoch": 0.03, + "learning_rate": 9.999900908311602e-05, + "loss": 1.0872, "step": 4 }, { "epoch": 0.03, - "learning_rate": 0.0009999777044621652, - "loss": 0.5151, + "learning_rate": 9.999777044621652e-05, + "loss": 1.1342, "step": 5 }, { - "epoch": 0.03, - "learning_rate": 0.0009999603637174071, - "loss": 0.5689, + "epoch": 0.04, + "learning_rate": 9.999603637174071e-05, + "loss": 0.8538, "step": 6 }, { - "epoch": 0.04, - "learning_rate": 0.0009999380687687187, - "loss": 1.0888, + "epoch": 0.05, + "learning_rate": 9.999380687687188e-05, + "loss": 0.9938, "step": 7 }, { - "epoch": 0.04, - "learning_rate": 0.0009999108198370248, - "loss": 0.7286, + "epoch": 0.05, + "learning_rate": 9.999108198370249e-05, + "loss": 0.8892, "step": 8 }, { - "epoch": 0.05, - "learning_rate": 0.0009998786171923407, - "loss": 0.7832, + "epoch": 0.06, + "learning_rate": 9.998786171923407e-05, + "loss": 0.9229, "step": 9 }, { - "epoch": 0.05, - "learning_rate": 0.000999841461153768, - "loss": 0.5225, + "epoch": 0.06, + "learning_rate": 9.998414611537681e-05, + "loss": 0.8371, "step": 10 }, { - "epoch": 0.06, - "learning_rate": 0.0009997993520894936, - "loss": 0.6154, + "epoch": 0.07, + "learning_rate": 9.997993520894937e-05, + "loss": 0.889, "step": 11 }, { - "epoch": 0.06, - "learning_rate": 0.0009997522904167844, - "loss": 0.822, + "epoch": 0.08, + "learning_rate": 9.997522904167844e-05, + "loss": 0.8277, "step": 12 }, { - "epoch": 0.07, - "learning_rate": 0.0009997002766019831, - "loss": 0.5985, + "epoch": 0.08, + "learning_rate": 9.997002766019832e-05, + "loss": 0.8332, "step": 13 }, { - "epoch": 0.07, - "learning_rate": 0.0009996433111605053, - "loss": 0.6488, + "epoch": 0.09, + "learning_rate": 9.996433111605052e-05, + "loss": 0.6883, "step": 14 }, { - "epoch": 0.08, - "learning_rate": 0.000999581394656832, - "loss": 0.5519, + "epoch": 0.1, + "learning_rate": 9.99581394656832e-05, + "loss": 0.9964, "step": 15 }, { - "epoch": 0.08, - "learning_rate": 0.0009995145277045061, - "loss": 0.8489, + "epoch": 0.1, + "learning_rate": 9.995145277045061e-05, + "loss": 0.7082, "step": 16 }, { - "epoch": 0.09, - "learning_rate": 0.0009994427109661253, - "loss": 0.622, + "epoch": 0.11, + "learning_rate": 9.994427109661253e-05, + "loss": 0.7267, "step": 17 }, { - "epoch": 0.09, - "learning_rate": 0.0009993659451533353, - "loss": 0.7171, + "epoch": 0.12, + "learning_rate": 9.993659451533353e-05, + "loss": 0.8105, "step": 18 }, { - "epoch": 0.1, - "learning_rate": 0.0009992842310268233, - "loss": 0.6803, + "epoch": 0.12, + "learning_rate": 9.992842310268233e-05, + "loss": 0.828, "step": 19 }, { - "epoch": 0.1, - "learning_rate": 0.0009991975693963108, - "loss": 0.7087, + "epoch": 0.13, + "learning_rate": 9.991975693963107e-05, + "loss": 0.6968, "step": 20 }, { - "epoch": 0.11, - "learning_rate": 0.000999105961120544, - "loss": 0.7689, + "epoch": 0.14, + "learning_rate": 9.99105961120544e-05, + "loss": 0.7103, "step": 21 }, { - "epoch": 0.11, - "learning_rate": 0.0009990094071072877, - "loss": 0.5209, + "epoch": 0.14, + "learning_rate": 9.990094071072877e-05, + "loss": 0.5235, "step": 22 }, { - "epoch": 0.12, - "learning_rate": 0.000998907908313314, - "loss": 0.6299, + "epoch": 0.15, + "learning_rate": 9.989079083133139e-05, + "loss": 0.6003, "step": 23 }, { - "epoch": 0.12, - "learning_rate": 0.000998801465744394, - "loss": 0.9556, + "epoch": 0.15, + "learning_rate": 9.988014657443941e-05, + "loss": 0.6947, "step": 24 }, { - "epoch": 0.13, - "learning_rate": 0.0009986900804552878, - "loss": 0.5427, + "epoch": 0.16, + "learning_rate": 9.986900804552878e-05, + "loss": 0.5919, "step": 25 }, { - "epoch": 0.13, - "learning_rate": 0.0009985737535497337, - "loss": 0.7997, + "epoch": 0.17, + "learning_rate": 9.985737535497337e-05, + "loss": 0.7571, "step": 26 }, { - "epoch": 0.14, - "learning_rate": 0.0009984524861804376, - "loss": 0.7101, + "epoch": 0.17, + "learning_rate": 9.984524861804376e-05, + "loss": 0.7387, "step": 27 }, { - "epoch": 0.14, - "learning_rate": 0.0009983262795490613, - "loss": 0.5568, + "epoch": 0.18, + "learning_rate": 9.983262795490613e-05, + "loss": 0.6541, "step": 28 }, { - "epoch": 0.15, - "learning_rate": 0.0009981951349062107, - "loss": 0.7616, + "epoch": 0.19, + "learning_rate": 9.981951349062106e-05, + "loss": 0.4892, "step": 29 }, { - "epoch": 0.15, - "learning_rate": 0.0009980590535514234, - "loss": 0.5599, + "epoch": 0.19, + "learning_rate": 9.980590535514233e-05, + "loss": 0.6266, "step": 30 }, { - "epoch": 0.16, - "learning_rate": 0.0009979180368331559, - "loss": 0.5306, + "epoch": 0.2, + "learning_rate": 9.979180368331558e-05, + "loss": 0.5837, "step": 31 }, { - "epoch": 0.16, - "learning_rate": 0.00099777208614877, - "loss": 0.7418, + "epoch": 0.21, + "learning_rate": 9.9777208614877e-05, + "loss": 0.6601, "step": 32 }, { - "epoch": 0.17, - "learning_rate": 0.0009976212029445194, - "loss": 0.7747, + "epoch": 0.21, + "learning_rate": 9.976212029445194e-05, + "loss": 0.678, "step": 33 }, { - "epoch": 0.17, - "learning_rate": 0.000997465388715535, - "loss": 0.7886, + "epoch": 0.22, + "learning_rate": 9.97465388715535e-05, + "loss": 0.6987, "step": 34 }, { - "epoch": 0.18, - "learning_rate": 0.00099730464500581, - "loss": 0.7207, + "epoch": 0.23, + "learning_rate": 9.9730464500581e-05, + "loss": 0.7587, "step": 35 }, { - "epoch": 0.18, - "learning_rate": 0.0009971389734081848, - "loss": 0.7, + "epoch": 0.23, + "learning_rate": 9.971389734081848e-05, + "loss": 0.5157, "step": 36 }, { - "epoch": 0.19, - "learning_rate": 0.0009969683755643318, - "loss": 0.5443, + "epoch": 0.24, + "learning_rate": 9.969683755643317e-05, + "loss": 0.7689, "step": 37 }, { - "epoch": 0.19, - "learning_rate": 0.0009967928531647372, - "loss": 0.6659, + "epoch": 0.25, + "learning_rate": 9.967928531647374e-05, + "loss": 0.5331, "step": 38 }, { - "epoch": 0.2, - "learning_rate": 0.0009966124079486872, - "loss": 0.6102, + "epoch": 0.25, + "learning_rate": 9.966124079486872e-05, + "loss": 0.9998, "step": 39 }, { - "epoch": 0.2, - "learning_rate": 0.000996427041704248, - "loss": 0.6723, + "epoch": 0.26, + "learning_rate": 9.96427041704248e-05, + "loss": 0.6785, "step": 40 }, { - "epoch": 0.21, - "learning_rate": 0.0009962367562682496, - "loss": 0.5818, + "epoch": 0.26, + "learning_rate": 9.962367562682496e-05, + "loss": 0.6151, "step": 41 }, { - "epoch": 0.21, - "learning_rate": 0.000996041553526267, - "loss": 0.6673, + "epoch": 0.27, + "learning_rate": 9.960415535262671e-05, + "loss": 0.7399, "step": 42 }, { - "epoch": 0.22, - "learning_rate": 0.0009958414354126022, - "loss": 0.7536, + "epoch": 0.28, + "learning_rate": 9.958414354126022e-05, + "loss": 0.5477, "step": 43 }, { - "epoch": 0.22, - "learning_rate": 0.000995636403910264, - "loss": 0.4539, + "epoch": 0.28, + "learning_rate": 9.956364039102642e-05, + "loss": 0.79, "step": 44 }, { - "epoch": 0.23, - "learning_rate": 0.0009954264610509496, - "loss": 0.7296, + "epoch": 0.29, + "learning_rate": 9.954264610509497e-05, + "loss": 0.5478, "step": 45 }, { - "epoch": 0.23, - "learning_rate": 0.0009952116089150232, - "loss": 0.5989, + "epoch": 0.3, + "learning_rate": 9.952116089150232e-05, + "loss": 0.6187, "step": 46 }, { - "epoch": 0.24, - "learning_rate": 0.000994991849631496, - "loss": 0.5712, + "epoch": 0.3, + "learning_rate": 9.94991849631496e-05, + "loss": 0.5635, "step": 47 }, { - "epoch": 0.24, - "learning_rate": 0.0009947671853780054, - "loss": 0.872, + "epoch": 0.31, + "learning_rate": 9.947671853780054e-05, + "loss": 0.612, "step": 48 }, { - "epoch": 0.25, - "learning_rate": 0.000994537618380793, - "loss": 0.7327, + "epoch": 0.32, + "learning_rate": 9.94537618380793e-05, + "loss": 0.5631, "step": 49 }, { - "epoch": 0.25, - "learning_rate": 0.0009943031509146824, - "loss": 0.7512, + "epoch": 0.32, + "learning_rate": 9.943031509146825e-05, + "loss": 0.7695, "step": 50 }, { - "epoch": 0.26, - "learning_rate": 0.0009940637853030573, - "loss": 0.6572, + "epoch": 0.33, + "learning_rate": 9.940637853030572e-05, + "loss": 0.6831, "step": 51 }, { - "epoch": 0.26, - "learning_rate": 0.0009938195239178375, - "loss": 0.6821, + "epoch": 0.34, + "learning_rate": 9.938195239178374e-05, + "loss": 0.7762, "step": 52 }, { - "epoch": 0.27, - "learning_rate": 0.0009935703691794564, - "loss": 0.6785, + "epoch": 0.34, + "learning_rate": 9.935703691794565e-05, + "loss": 0.5958, "step": 53 }, { - "epoch": 0.27, - "learning_rate": 0.0009933163235568369, - "loss": 0.9043, + "epoch": 0.35, + "learning_rate": 9.933163235568367e-05, + "loss": 0.6312, "step": 54 }, { - "epoch": 0.28, - "learning_rate": 0.0009930573895673657, - "loss": 0.6856, + "epoch": 0.35, + "learning_rate": 9.930573895673657e-05, + "loss": 0.5744, "step": 55 }, { - "epoch": 0.28, - "learning_rate": 0.0009927935697768698, - "loss": 0.4684, + "epoch": 0.36, + "learning_rate": 9.927935697768698e-05, + "loss": 0.7623, "step": 56 }, { - "epoch": 0.29, - "learning_rate": 0.0009925248667995907, - "loss": 0.5579, + "epoch": 0.37, + "learning_rate": 9.925248667995907e-05, + "loss": 0.7806, "step": 57 }, { - "epoch": 0.29, - "learning_rate": 0.0009922512832981584, - "loss": 0.4323, + "epoch": 0.37, + "learning_rate": 9.922512832981584e-05, + "loss": 0.7338, "step": 58 }, { - "epoch": 0.3, - "learning_rate": 0.0009919728219835644, - "loss": 1.0761, + "epoch": 0.38, + "learning_rate": 9.919728219835643e-05, + "loss": 0.5437, "step": 59 }, { - "epoch": 0.3, - "learning_rate": 0.0009916894856151356, - "loss": 0.5668, + "epoch": 0.39, + "learning_rate": 9.916894856151357e-05, + "loss": 0.6343, "step": 60 }, { - "epoch": 0.31, - "learning_rate": 0.0009914012770005072, - "loss": 0.6059, + "epoch": 0.39, + "learning_rate": 9.914012770005072e-05, + "loss": 0.5863, "step": 61 }, { - "epoch": 0.31, - "learning_rate": 0.000991108198995594, - "loss": 0.5635, + "epoch": 0.4, + "learning_rate": 9.91108198995594e-05, + "loss": 0.7165, "step": 62 }, { - "epoch": 0.32, - "learning_rate": 0.0009908102545045623, - "loss": 0.7705, + "epoch": 0.41, + "learning_rate": 9.908102545045625e-05, + "loss": 0.7479, "step": 63 }, { - "epoch": 0.32, - "learning_rate": 0.0009905074464798022, - "loss": 0.8812, + "epoch": 0.41, + "learning_rate": 9.905074464798024e-05, + "loss": 0.6124, "step": 64 }, { - "epoch": 0.33, - "learning_rate": 0.0009901997779218968, - "loss": 0.6145, + "epoch": 0.42, + "learning_rate": 9.901997779218967e-05, + "loss": 0.8549, "step": 65 }, { - "epoch": 0.34, - "learning_rate": 0.0009898872518795932, - "loss": 0.5681, + "epoch": 0.43, + "learning_rate": 9.898872518795932e-05, + "loss": 0.7272, "step": 66 }, { - "epoch": 0.34, - "learning_rate": 0.0009895698714497724, - "loss": 0.658, + "epoch": 0.43, + "learning_rate": 9.895698714497724e-05, + "loss": 0.674, "step": 67 }, { - "epoch": 0.35, - "learning_rate": 0.0009892476397774185, - "loss": 0.5969, + "epoch": 0.44, + "learning_rate": 9.892476397774186e-05, + "loss": 0.7458, "step": 68 }, { - "epoch": 0.35, - "learning_rate": 0.0009889205600555875, - "loss": 0.5991, + "epoch": 0.45, + "learning_rate": 9.889205600555877e-05, + "loss": 0.6658, "step": 69 }, { - "epoch": 0.36, - "learning_rate": 0.0009885886355253757, - "loss": 0.762, + "epoch": 0.45, + "learning_rate": 9.885886355253758e-05, + "loss": 0.5559, "step": 70 }, { - "epoch": 0.36, - "learning_rate": 0.0009882518694758874, - "loss": 0.5969, + "epoch": 0.46, + "learning_rate": 9.882518694758875e-05, + "loss": 0.5863, "step": 71 }, { - "epoch": 0.37, - "learning_rate": 0.0009879102652442023, - "loss": 0.5668, + "epoch": 0.46, + "learning_rate": 9.879102652442024e-05, + "loss": 0.5821, "step": 72 }, { - "epoch": 0.37, - "learning_rate": 0.0009875638262153432, - "loss": 0.6104, + "epoch": 0.47, + "learning_rate": 9.875638262153431e-05, + "loss": 0.6834, "step": 73 }, { - "epoch": 0.38, - "learning_rate": 0.000987212555822241, - "loss": 0.5512, + "epoch": 0.48, + "learning_rate": 9.872125558222409e-05, + "loss": 0.5755, "step": 74 }, { - "epoch": 0.38, - "learning_rate": 0.0009868564575457022, - "loss": 0.5551, + "epoch": 0.48, + "learning_rate": 9.868564575457023e-05, + "loss": 0.6745, "step": 75 }, { - "epoch": 0.39, - "learning_rate": 0.0009864955349143734, - "loss": 0.6269, + "epoch": 0.49, + "learning_rate": 9.864955349143734e-05, + "loss": 0.5966, "step": 76 }, { - "epoch": 0.39, - "learning_rate": 0.000986129791504707, - "loss": 0.5629, + "epoch": 0.5, + "learning_rate": 9.861297915047069e-05, + "loss": 0.6118, "step": 77 }, { - "epoch": 0.4, - "learning_rate": 0.0009857592309409247, - "loss": 0.7589, + "epoch": 0.5, + "learning_rate": 9.857592309409247e-05, + "loss": 0.5196, "step": 78 }, { - "epoch": 0.4, - "learning_rate": 0.0009853838568949832, - "loss": 0.5747, + "epoch": 0.51, + "learning_rate": 9.853838568949831e-05, + "loss": 0.6412, "step": 79 }, { - "epoch": 0.41, - "learning_rate": 0.0009850036730865363, - "loss": 0.7552, + "epoch": 0.52, + "learning_rate": 9.850036730865364e-05, + "loss": 0.6207, "step": 80 }, { - "epoch": 0.41, - "learning_rate": 0.000984618683282899, - "loss": 0.6728, + "epoch": 0.52, + "learning_rate": 9.846186832828989e-05, + "loss": 0.5705, "step": 81 }, { - "epoch": 0.42, - "learning_rate": 0.0009842288912990096, - "loss": 0.6057, + "epoch": 0.53, + "learning_rate": 9.842288912990096e-05, + "loss": 0.5643, "step": 82 }, { - "epoch": 0.42, - "learning_rate": 0.0009838343009973924, - "loss": 0.545, + "epoch": 0.54, + "learning_rate": 9.838343009973925e-05, + "loss": 0.7729, "step": 83 }, { - "epoch": 0.43, - "learning_rate": 0.0009834349162881188, - "loss": 0.7007, + "epoch": 0.54, + "learning_rate": 9.83434916288119e-05, + "loss": 0.5091, "step": 84 }, { - "epoch": 0.43, - "learning_rate": 0.0009830307411287697, - "loss": 0.5392, + "epoch": 0.55, + "learning_rate": 9.830307411287695e-05, + "loss": 0.7592, "step": 85 }, { - "epoch": 0.44, - "learning_rate": 0.000982621779524394, - "loss": 0.8032, + "epoch": 0.55, + "learning_rate": 9.82621779524394e-05, + "loss": 0.6386, "step": 86 }, { - "epoch": 0.44, - "learning_rate": 0.0009822080355274718, - "loss": 0.5957, + "epoch": 0.56, + "learning_rate": 9.822080355274719e-05, + "loss": 0.5282, "step": 87 }, { - "epoch": 0.45, - "learning_rate": 0.0009817895132378724, - "loss": 0.5639, + "epoch": 0.57, + "learning_rate": 9.817895132378725e-05, + "loss": 0.6444, "step": 88 }, { - "epoch": 0.45, - "learning_rate": 0.0009813662168028144, - "loss": 0.6488, + "epoch": 0.57, + "learning_rate": 9.813662168028144e-05, + "loss": 0.5698, "step": 89 }, { - "epoch": 0.46, - "learning_rate": 0.0009809381504168233, - "loss": 0.8626, + "epoch": 0.58, + "learning_rate": 9.809381504168234e-05, + "loss": 0.4588, "step": 90 }, { - "epoch": 0.46, - "learning_rate": 0.0009805053183216924, - "loss": 0.6872, + "epoch": 0.59, + "learning_rate": 9.805053183216923e-05, + "loss": 0.645, "step": 91 }, { - "epoch": 0.47, - "learning_rate": 0.0009800677248064382, - "loss": 0.6908, + "epoch": 0.59, + "learning_rate": 9.800677248064382e-05, + "loss": 0.5169, "step": 92 }, { - "epoch": 0.47, - "learning_rate": 0.0009796253742072596, - "loss": 0.699, + "epoch": 0.6, + "learning_rate": 9.796253742072596e-05, + "loss": 0.5538, "step": 93 }, { - "epoch": 0.48, - "learning_rate": 0.0009791782709074944, - "loss": 0.6305, + "epoch": 0.61, + "learning_rate": 9.791782709074944e-05, + "loss": 0.549, "step": 94 }, { - "epoch": 0.48, - "learning_rate": 0.0009787264193375754, - "loss": 0.8407, + "epoch": 0.61, + "learning_rate": 9.787264193375753e-05, + "loss": 0.6782, "step": 95 }, { - "epoch": 0.49, - "learning_rate": 0.0009782698239749871, - "loss": 0.7651, + "epoch": 0.62, + "learning_rate": 9.782698239749873e-05, + "loss": 0.6065, "step": 96 }, { - "epoch": 0.49, - "learning_rate": 0.0009778084893442218, - "loss": 0.5649, + "epoch": 0.63, + "learning_rate": 9.778084893442218e-05, + "loss": 0.8218, "step": 97 }, { - "epoch": 0.5, - "learning_rate": 0.000977342420016733, - "loss": 0.6298, + "epoch": 0.63, + "learning_rate": 9.77342420016733e-05, + "loss": 0.5523, "step": 98 }, { - "epoch": 0.5, - "learning_rate": 0.0009768716206108921, - "loss": 0.6444, + "epoch": 0.64, + "learning_rate": 9.768716206108921e-05, + "loss": 0.61, "step": 99 }, { - "epoch": 0.51, - "learning_rate": 0.0009763960957919414, - "loss": 0.5879, + "epoch": 0.65, + "learning_rate": 9.763960957919413e-05, + "loss": 0.8251, "step": 100 }, { - "epoch": 0.51, - "learning_rate": 0.0009759158502719481, - "loss": 0.6006, + "epoch": 0.65, + "learning_rate": 9.759158502719481e-05, + "loss": 0.8027, "step": 101 }, { - "epoch": 0.52, - "learning_rate": 0.0009754308888097583, - "loss": 0.6416, + "epoch": 0.66, + "learning_rate": 9.754308888097583e-05, + "loss": 1.0138, "step": 102 }, { - "epoch": 0.52, - "learning_rate": 0.0009749412162109486, - "loss": 0.8301, + "epoch": 0.66, + "learning_rate": 9.749412162109485e-05, + "loss": 0.5758, "step": 103 }, { - "epoch": 0.53, - "learning_rate": 0.0009744468373277796, - "loss": 0.8659, + "epoch": 0.67, + "learning_rate": 9.744468373277797e-05, + "loss": 0.6591, "step": 104 }, { - "epoch": 0.53, - "learning_rate": 0.0009739477570591473, - "loss": 0.7777, + "epoch": 0.68, + "learning_rate": 9.739477570591473e-05, + "loss": 0.5436, "step": 105 }, { - "epoch": 0.54, - "learning_rate": 0.0009734439803505345, - "loss": 0.5722, + "epoch": 0.68, + "learning_rate": 9.734439803505345e-05, + "loss": 0.589, "step": 106 }, { - "epoch": 0.54, - "learning_rate": 0.000972935512193962, - "loss": 0.6249, + "epoch": 0.69, + "learning_rate": 9.729355121939621e-05, + "loss": 0.54, "step": 107 }, { - "epoch": 0.55, - "learning_rate": 0.0009724223576279394, - "loss": 1.1485, + "epoch": 0.7, + "learning_rate": 9.724223576279395e-05, + "loss": 0.5461, "step": 108 }, { - "epoch": 0.55, - "learning_rate": 0.0009719045217374141, - "loss": 0.6837, + "epoch": 0.7, + "learning_rate": 9.719045217374143e-05, + "loss": 0.626, "step": 109 }, { - "epoch": 0.56, - "learning_rate": 0.0009713820096537225, - "loss": 0.5921, + "epoch": 0.71, + "learning_rate": 9.713820096537225e-05, + "loss": 0.5492, "step": 110 }, { - "epoch": 0.56, - "learning_rate": 0.0009708548265545375, - "loss": 0.8619, + "epoch": 0.72, + "learning_rate": 9.708548265545375e-05, + "loss": 0.63, "step": 111 }, { - "epoch": 0.57, - "learning_rate": 0.0009703229776638185, - "loss": 0.6175, + "epoch": 0.72, + "learning_rate": 9.703229776638185e-05, + "loss": 0.6278, "step": 112 }, { - "epoch": 0.57, - "learning_rate": 0.0009697864682517592, - "loss": 0.6915, + "epoch": 0.73, + "learning_rate": 9.697864682517592e-05, + "loss": 0.5364, "step": 113 }, { - "epoch": 0.58, - "learning_rate": 0.0009692453036347351, - "loss": 0.5714, + "epoch": 0.74, + "learning_rate": 9.692453036347351e-05, + "loss": 0.6641, "step": 114 }, { - "epoch": 0.58, - "learning_rate": 0.0009686994891752507, - "loss": 0.5019, + "epoch": 0.74, + "learning_rate": 9.686994891752508e-05, + "loss": 0.5447, "step": 115 }, { - "epoch": 0.59, - "learning_rate": 0.0009681490302818874, - "loss": 1.1504, + "epoch": 0.75, + "learning_rate": 9.681490302818874e-05, + "loss": 0.9975, "step": 116 }, { - "epoch": 0.59, - "learning_rate": 0.0009675939324092487, - "loss": 0.6648, + "epoch": 0.75, + "learning_rate": 9.675939324092486e-05, + "loss": 0.6382, "step": 117 }, { - "epoch": 0.6, - "learning_rate": 0.0009670342010579065, - "loss": 0.6379, + "epoch": 0.76, + "learning_rate": 9.670342010579065e-05, + "loss": 0.7098, "step": 118 }, { - "epoch": 0.6, - "learning_rate": 0.0009664698417743475, - "loss": 0.6071, + "epoch": 0.77, + "learning_rate": 9.664698417743475e-05, + "loss": 0.6094, "step": 119 }, { - "epoch": 0.61, - "learning_rate": 0.0009659008601509168, - "loss": 0.675, + "epoch": 0.77, + "learning_rate": 9.659008601509168e-05, + "loss": 0.5387, "step": 120 }, { - "epoch": 0.61, - "learning_rate": 0.0009653272618257631, - "loss": 0.5167, + "epoch": 0.78, + "learning_rate": 9.653272618257631e-05, + "loss": 0.6442, "step": 121 }, { - "epoch": 0.62, - "learning_rate": 0.0009647490524827833, - "loss": 0.7457, + "epoch": 0.79, + "learning_rate": 9.647490524827834e-05, + "loss": 0.5843, "step": 122 }, { - "epoch": 0.62, - "learning_rate": 0.0009641662378515658, - "loss": 0.836, + "epoch": 0.79, + "learning_rate": 9.641662378515659e-05, + "loss": 0.6625, "step": 123 }, { - "epoch": 0.63, - "learning_rate": 0.0009635788237073333, - "loss": 0.759, + "epoch": 0.8, + "learning_rate": 9.635788237073334e-05, + "loss": 0.447, "step": 124 }, { - "epoch": 0.63, - "learning_rate": 0.0009629868158708861, - "loss": 0.885, + "epoch": 0.81, + "learning_rate": 9.629868158708861e-05, + "loss": 0.4932, "step": 125 }, { - "epoch": 0.64, - "learning_rate": 0.0009623902202085444, - "loss": 0.7311, + "epoch": 0.81, + "learning_rate": 9.623902202085444e-05, + "loss": 0.6485, "step": 126 }, { - "epoch": 0.64, - "learning_rate": 0.0009617890426320899, - "loss": 0.7275, + "epoch": 0.82, + "learning_rate": 9.617890426320899e-05, + "loss": 0.4969, "step": 127 }, { - "epoch": 0.65, - "learning_rate": 0.0009611832890987075, - "loss": 0.6473, + "epoch": 0.83, + "learning_rate": 9.611832890987076e-05, + "loss": 0.5679, "step": 128 }, { - "epoch": 0.65, - "learning_rate": 0.0009605729656109264, - "loss": 0.6269, + "epoch": 0.83, + "learning_rate": 9.605729656109265e-05, + "loss": 0.7879, "step": 129 }, { - "epoch": 0.66, - "learning_rate": 0.0009599580782165598, - "loss": 0.9734, + "epoch": 0.84, + "learning_rate": 9.599580782165598e-05, + "loss": 0.6588, "step": 130 }, { - "epoch": 0.66, - "learning_rate": 0.0009593386330086457, - "loss": 0.6377, + "epoch": 0.85, + "learning_rate": 9.593386330086458e-05, + "loss": 0.7059, "step": 131 }, { - "epoch": 0.67, - "learning_rate": 0.0009587146361253867, - "loss": 0.7208, + "epoch": 0.85, + "learning_rate": 9.587146361253868e-05, + "loss": 0.625, "step": 132 }, { - "epoch": 0.68, - "learning_rate": 0.0009580860937500883, - "loss": 0.5472, + "epoch": 0.86, + "learning_rate": 9.580860937500884e-05, + "loss": 0.6958, "step": 133 }, { - "epoch": 0.68, - "learning_rate": 0.0009574530121110989, - "loss": 0.8343, + "epoch": 0.86, + "learning_rate": 9.57453012111099e-05, + "loss": 0.6816, "step": 134 }, { - "epoch": 0.69, - "learning_rate": 0.0009568153974817464, - "loss": 0.6048, + "epoch": 0.87, + "learning_rate": 9.568153974817464e-05, + "loss": 0.5775, "step": 135 }, { - "epoch": 0.69, - "learning_rate": 0.0009561732561802779, - "loss": 0.5346, + "epoch": 0.88, + "learning_rate": 9.561732561802778e-05, + "loss": 0.5051, "step": 136 }, { - "epoch": 0.7, - "learning_rate": 0.0009555265945697953, - "loss": 0.6295, + "epoch": 0.88, + "learning_rate": 9.555265945697953e-05, + "loss": 0.5025, "step": 137 }, { - "epoch": 0.7, - "learning_rate": 0.0009548754190581938, - "loss": 0.6079, + "epoch": 0.89, + "learning_rate": 9.548754190581939e-05, + "loss": 0.5407, "step": 138 }, { - "epoch": 0.71, - "learning_rate": 0.0009542197360980978, - "loss": 0.5289, + "epoch": 0.9, + "learning_rate": 9.542197360980978e-05, + "loss": 0.5462, "step": 139 }, { - "epoch": 0.71, - "learning_rate": 0.0009535595521867959, - "loss": 0.7579, + "epoch": 0.9, + "learning_rate": 9.53559552186796e-05, + "loss": 0.6505, "step": 140 }, { - "epoch": 0.72, - "learning_rate": 0.0009528948738661784, - "loss": 0.8158, + "epoch": 0.91, + "learning_rate": 9.528948738661784e-05, + "loss": 0.5869, "step": 141 }, { - "epoch": 0.72, - "learning_rate": 0.0009522257077226717, - "loss": 0.6023, + "epoch": 0.92, + "learning_rate": 9.522257077226717e-05, + "loss": 0.5047, "step": 142 }, { - "epoch": 0.73, - "learning_rate": 0.0009515520603871719, - "loss": 0.6647, + "epoch": 0.92, + "learning_rate": 9.51552060387172e-05, + "loss": 0.7094, "step": 143 }, { - "epoch": 0.73, - "learning_rate": 0.0009508739385349812, - "loss": 1.1108, + "epoch": 0.93, + "learning_rate": 9.508739385349812e-05, + "loss": 0.6079, "step": 144 }, { - "epoch": 0.74, - "learning_rate": 0.0009501913488857399, - "loss": 0.9834, + "epoch": 0.94, + "learning_rate": 9.501913488857399e-05, + "loss": 0.5941, "step": 145 }, { - "epoch": 0.74, - "learning_rate": 0.0009495042982033611, - "loss": 0.9049, + "epoch": 0.94, + "learning_rate": 9.49504298203361e-05, + "loss": 0.5613, "step": 146 }, { - "epoch": 0.75, - "learning_rate": 0.0009488127932959625, - "loss": 0.6604, + "epoch": 0.95, + "learning_rate": 9.488127932959625e-05, + "loss": 0.693, "step": 147 }, { - "epoch": 0.75, - "learning_rate": 0.0009481168410158003, - "loss": 0.8768, + "epoch": 0.95, + "learning_rate": 9.481168410158003e-05, + "loss": 0.6659, "step": 148 }, { - "epoch": 0.76, - "learning_rate": 0.0009474164482592001, - "loss": 0.6579, + "epoch": 0.96, + "learning_rate": 9.474164482592002e-05, + "loss": 0.8608, "step": 149 }, { - "epoch": 0.76, - "learning_rate": 0.0009467116219664893, - "loss": 0.7038, + "epoch": 0.97, + "learning_rate": 9.467116219664894e-05, + "loss": 0.743, "step": 150 }, { - "epoch": 0.77, - "learning_rate": 0.0009460023691219276, - "loss": 0.6726, + "epoch": 0.97, + "learning_rate": 9.460023691219277e-05, + "loss": 0.7069, "step": 151 }, { - "epoch": 0.77, - "learning_rate": 0.0009452886967536389, - "loss": 1.1224, + "epoch": 0.98, + "learning_rate": 9.45288696753639e-05, + "loss": 0.6795, "step": 152 }, { - "epoch": 0.78, - "learning_rate": 0.0009445706119335407, - "loss": 0.7254, + "epoch": 0.99, + "learning_rate": 9.445706119335407e-05, + "loss": 0.4837, "step": 153 }, { - "epoch": 0.78, - "learning_rate": 0.0009438481217772743, - "loss": 0.6464, + "epoch": 0.99, + "learning_rate": 9.438481217772744e-05, + "loss": 0.6796, "step": 154 }, { - "epoch": 0.79, - "learning_rate": 0.0009431212334441342, - "loss": 1.0653, + "epoch": 1.0, + "learning_rate": 9.431212334441343e-05, + "loss": 0.5442, "step": 155 }, { - "epoch": 0.79, - "learning_rate": 0.0009423899541369978, - "loss": 0.7072, + "epoch": 1.01, + "learning_rate": 9.423899541369978e-05, + "loss": 0.4109, "step": 156 }, { - "epoch": 0.8, - "learning_rate": 0.000941654291102253, - "loss": 0.9562, + "epoch": 1.01, + "learning_rate": 9.41654291102253e-05, + "loss": 0.5735, "step": 157 }, { - "epoch": 0.8, - "learning_rate": 0.0009409142516297269, - "loss": 0.5401, + "epoch": 1.02, + "learning_rate": 9.409142516297269e-05, + "loss": 0.5028, "step": 158 }, { - "epoch": 0.81, - "learning_rate": 0.0009401698430526141, - "loss": 0.628, + "epoch": 1.03, + "learning_rate": 9.401698430526142e-05, + "loss": 0.4985, "step": 159 }, { - "epoch": 0.81, - "learning_rate": 0.0009394210727474029, - "loss": 0.6226, + "epoch": 1.03, + "learning_rate": 9.394210727474028e-05, + "loss": 0.5741, "step": 160 }, { - "epoch": 0.82, - "learning_rate": 0.0009386679481338031, - "loss": 0.8939, + "epoch": 1.04, + "learning_rate": 9.386679481338033e-05, + "loss": 0.5671, "step": 161 }, { - "epoch": 0.82, - "learning_rate": 0.0009379104766746722, - "loss": 0.614, + "epoch": 1.05, + "learning_rate": 9.379104766746722e-05, + "loss": 0.5839, "step": 162 }, { - "epoch": 0.83, - "learning_rate": 0.0009371486658759415, - "loss": 0.6249, + "epoch": 1.05, + "learning_rate": 9.371486658759416e-05, + "loss": 0.5261, "step": 163 }, { - "epoch": 0.83, - "learning_rate": 0.0009363825232865413, - "loss": 0.5156, + "epoch": 1.06, + "learning_rate": 9.363825232865413e-05, + "loss": 0.474, "step": 164 }, { - "epoch": 0.84, - "learning_rate": 0.0009356120564983266, - "loss": 0.6214, + "epoch": 1.06, + "learning_rate": 9.356120564983266e-05, + "loss": 0.5641, "step": 165 }, { - "epoch": 0.84, - "learning_rate": 0.0009348372731460022, - "loss": 0.6466, + "epoch": 1.07, + "learning_rate": 9.348372731460023e-05, + "loss": 0.5222, "step": 166 }, { - "epoch": 0.85, - "learning_rate": 0.0009340581809070458, - "loss": 1.3654, + "epoch": 1.08, + "learning_rate": 9.340581809070459e-05, + "loss": 0.562, "step": 167 }, { - "epoch": 0.85, - "learning_rate": 0.0009332747875016332, - "loss": 0.7355, + "epoch": 1.08, + "learning_rate": 9.332747875016332e-05, + "loss": 0.5875, "step": 168 }, { - "epoch": 0.86, - "learning_rate": 0.0009324871006925612, - "loss": 0.6488, + "epoch": 1.09, + "learning_rate": 9.324871006925613e-05, + "loss": 0.4605, "step": 169 }, { - "epoch": 0.86, - "learning_rate": 0.0009316951282851706, - "loss": 0.6619, + "epoch": 1.1, + "learning_rate": 9.316951282851707e-05, + "loss": 0.5808, "step": 170 }, { - "epoch": 0.87, - "learning_rate": 0.0009308988781272693, - "loss": 0.7481, + "epoch": 1.1, + "learning_rate": 9.308988781272694e-05, + "loss": 0.6235, "step": 171 }, { - "epoch": 0.87, - "learning_rate": 0.0009300983581090541, - "loss": 0.7286, + "epoch": 1.11, + "learning_rate": 9.300983581090541e-05, + "loss": 0.684, "step": 172 }, { - "epoch": 0.88, - "learning_rate": 0.0009292935761630325, - "loss": 0.6915, + "epoch": 1.12, + "learning_rate": 9.292935761630326e-05, + "loss": 0.4411, "step": 173 }, { - "epoch": 0.88, - "learning_rate": 0.0009284845402639446, - "loss": 0.7665, + "epoch": 1.12, + "learning_rate": 9.284845402639446e-05, + "loss": 0.5228, "step": 174 }, { - "epoch": 0.89, - "learning_rate": 0.0009276712584286833, - "loss": 0.5691, + "epoch": 1.13, + "learning_rate": 9.276712584286833e-05, + "loss": 0.8171, "step": 175 }, { - "epoch": 0.89, - "learning_rate": 0.000926853738716216, - "loss": 0.9606, + "epoch": 1.14, + "learning_rate": 9.26853738716216e-05, + "loss": 0.5837, "step": 176 }, { - "epoch": 0.9, - "learning_rate": 0.0009260319892275033, - "loss": 0.6955, + "epoch": 1.14, + "learning_rate": 9.260319892275034e-05, + "loss": 0.4187, "step": 177 }, { - "epoch": 0.9, - "learning_rate": 0.00092520601810542, - "loss": 0.7575, + "epoch": 1.15, + "learning_rate": 9.2520601810542e-05, + "loss": 0.4252, "step": 178 }, { - "epoch": 0.91, - "learning_rate": 0.0009243758335346734, - "loss": 0.644, + "epoch": 1.15, + "learning_rate": 9.243758335346735e-05, + "loss": 0.5112, "step": 179 }, { - "epoch": 0.91, - "learning_rate": 0.0009235414437417234, - "loss": 0.6696, + "epoch": 1.16, + "learning_rate": 9.235414437417234e-05, + "loss": 0.5455, "step": 180 }, { - "epoch": 0.92, - "learning_rate": 0.0009227028569946996, - "loss": 0.7746, + "epoch": 1.17, + "learning_rate": 9.227028569946996e-05, + "loss": 0.668, "step": 181 }, { - "epoch": 0.92, - "learning_rate": 0.0009218600816033201, - "loss": 0.7244, + "epoch": 1.17, + "learning_rate": 9.2186008160332e-05, + "loss": 0.4831, "step": 182 }, { - "epoch": 0.93, - "learning_rate": 0.0009210131259188095, - "loss": 0.6162, + "epoch": 1.18, + "learning_rate": 9.210131259188095e-05, + "loss": 0.4275, "step": 183 }, { - "epoch": 0.93, - "learning_rate": 0.0009201619983338152, - "loss": 1.1655, + "epoch": 1.19, + "learning_rate": 9.201619983338153e-05, + "loss": 0.852, "step": 184 }, { - "epoch": 0.94, - "learning_rate": 0.0009193067072823251, - "loss": 0.7024, + "epoch": 1.19, + "learning_rate": 9.193067072823251e-05, + "loss": 0.5029, "step": 185 }, { - "epoch": 0.94, - "learning_rate": 0.000918447261239584, - "loss": 0.8246, + "epoch": 1.2, + "learning_rate": 9.18447261239584e-05, + "loss": 0.534, "step": 186 }, { - "epoch": 0.95, - "learning_rate": 0.0009175836687220084, - "loss": 0.8583, + "epoch": 1.21, + "learning_rate": 9.175836687220084e-05, + "loss": 0.5727, "step": 187 }, { - "epoch": 0.95, - "learning_rate": 0.0009167159382871039, - "loss": 0.6455, + "epoch": 1.21, + "learning_rate": 9.167159382871039e-05, + "loss": 0.5095, "step": 188 }, { - "epoch": 0.96, - "learning_rate": 0.000915844078533379, - "loss": 0.7452, + "epoch": 1.22, + "learning_rate": 9.15844078533379e-05, + "loss": 0.674, "step": 189 }, { - "epoch": 0.96, - "learning_rate": 0.0009149680981002608, - "loss": 0.6813, + "epoch": 1.23, + "learning_rate": 9.149680981002609e-05, + "loss": 0.4672, "step": 190 }, { - "epoch": 0.97, - "learning_rate": 0.0009140880056680088, - "loss": 0.652, + "epoch": 1.23, + "learning_rate": 9.140880056680088e-05, + "loss": 0.578, "step": 191 }, { - "epoch": 0.97, - "learning_rate": 0.000913203809957629, - "loss": 0.6686, + "epoch": 1.24, + "learning_rate": 9.13203809957629e-05, + "loss": 0.5875, "step": 192 }, { - "epoch": 0.98, - "learning_rate": 0.0009123155197307875, - "loss": 0.7548, + "epoch": 1.25, + "learning_rate": 9.123155197307876e-05, + "loss": 0.5596, "step": 193 }, { - "epoch": 0.98, - "learning_rate": 0.0009114231437897244, - "loss": 0.8523, + "epoch": 1.25, + "learning_rate": 9.114231437897244e-05, + "loss": 0.5782, "step": 194 }, { - "epoch": 0.99, - "learning_rate": 0.0009105266909771652, - "loss": 0.5544, + "epoch": 1.26, + "learning_rate": 9.105266909771653e-05, + "loss": 0.4072, "step": 195 }, { - "epoch": 0.99, - "learning_rate": 0.0009096261701762343, - "loss": 0.6712, + "epoch": 1.26, + "learning_rate": 9.096261701762342e-05, + "loss": 0.4472, "step": 196 }, { - "epoch": 1.0, - "learning_rate": 0.0009087215903103663, - "loss": 0.4564, + "epoch": 1.27, + "learning_rate": 9.087215903103662e-05, + "loss": 0.4525, "step": 197 }, { - "epoch": 1.01, - "learning_rate": 0.0009078129603432181, - "loss": 0.3499, + "epoch": 1.28, + "learning_rate": 9.078129603432181e-05, + "loss": 0.4294, "step": 198 }, { - "epoch": 1.01, - "learning_rate": 0.0009069002892785796, - "loss": 0.6096, + "epoch": 1.28, + "learning_rate": 9.069002892785797e-05, + "loss": 0.49, "step": 199 }, { - "epoch": 1.02, - "learning_rate": 0.0009059835861602853, - "loss": 0.1973, + "epoch": 1.29, + "learning_rate": 9.059835861602853e-05, + "loss": 0.6561, "step": 200 }, { - "epoch": 1.02, - "learning_rate": 0.0009050628600721233, - "loss": 0.4716, + "epoch": 1.3, + "learning_rate": 9.050628600721234e-05, + "loss": 0.6434, "step": 201 }, { - "epoch": 1.03, - "learning_rate": 0.0009041381201377467, - "loss": 0.443, + "epoch": 1.3, + "learning_rate": 9.041381201377468e-05, + "loss": 0.5954, "step": 202 }, { - "epoch": 1.03, - "learning_rate": 0.0009032093755205823, - "loss": 0.4729, + "epoch": 1.31, + "learning_rate": 9.032093755205822e-05, + "loss": 0.5783, "step": 203 }, { - "epoch": 1.04, - "learning_rate": 0.0009022766354237399, - "loss": 0.373, + "epoch": 1.32, + "learning_rate": 9.0227663542374e-05, + "loss": 0.7083, "step": 204 }, { - "epoch": 1.04, - "learning_rate": 0.0009013399090899217, - "loss": 0.4033, + "epoch": 1.32, + "learning_rate": 9.013399090899217e-05, + "loss": 0.4703, "step": 205 }, { - "epoch": 1.05, - "learning_rate": 0.0009003992058013302, - "loss": 0.3282, + "epoch": 1.33, + "learning_rate": 9.003992058013302e-05, + "loss": 0.5067, "step": 206 }, { - "epoch": 1.05, - "learning_rate": 0.0008994545348795758, - "loss": 0.4702, + "epoch": 1.34, + "learning_rate": 8.994545348795759e-05, + "loss": 0.5962, "step": 207 }, { - "epoch": 1.06, - "learning_rate": 0.0008985059056855857, - "loss": 0.5088, + "epoch": 1.34, + "learning_rate": 8.985059056855858e-05, + "loss": 0.4955, "step": 208 }, { - "epoch": 1.06, - "learning_rate": 0.0008975533276195101, - "loss": 0.4294, + "epoch": 1.35, + "learning_rate": 8.975533276195102e-05, + "loss": 0.5224, "step": 209 }, { - "epoch": 1.07, - "learning_rate": 0.0008965968101206292, - "loss": 0.2792, + "epoch": 1.35, + "learning_rate": 8.965968101206291e-05, + "loss": 0.4446, "step": 210 }, { - "epoch": 1.07, - "learning_rate": 0.0008956363626672594, - "loss": 0.3841, + "epoch": 1.36, + "learning_rate": 8.956363626672595e-05, + "loss": 0.4902, "step": 211 }, { - "epoch": 1.08, - "learning_rate": 0.0008946719947766611, - "loss": 0.4744, + "epoch": 1.37, + "learning_rate": 8.94671994776661e-05, + "loss": 0.466, "step": 212 }, { - "epoch": 1.08, - "learning_rate": 0.0008937037160049415, - "loss": 0.3686, + "epoch": 1.37, + "learning_rate": 8.937037160049416e-05, + "loss": 0.4933, "step": 213 }, { - "epoch": 1.09, - "learning_rate": 0.0008927315359469625, - "loss": 0.6375, + "epoch": 1.38, + "learning_rate": 8.927315359469626e-05, + "loss": 0.542, "step": 214 }, { - "epoch": 1.09, - "learning_rate": 0.0008917554642362443, - "loss": 0.8962, + "epoch": 1.39, + "learning_rate": 8.917554642362443e-05, + "loss": 0.5227, "step": 215 }, { - "epoch": 1.1, - "learning_rate": 0.0008907755105448703, - "loss": 0.4171, + "epoch": 1.39, + "learning_rate": 8.907755105448704e-05, + "loss": 0.5783, "step": 216 }, { - "epoch": 1.1, - "learning_rate": 0.000889791684583391, - "loss": 0.4826, + "epoch": 1.4, + "learning_rate": 8.89791684583391e-05, + "loss": 0.509, "step": 217 }, { - "epoch": 1.11, - "learning_rate": 0.0008888039961007281, - "loss": 0.5129, + "epoch": 1.41, + "learning_rate": 8.888039961007282e-05, + "loss": 0.4139, "step": 218 }, { - "epoch": 1.11, - "learning_rate": 0.000887812454884078, - "loss": 0.2824, + "epoch": 1.41, + "learning_rate": 8.87812454884078e-05, + "loss": 0.4373, "step": 219 }, { - "epoch": 1.12, - "learning_rate": 0.0008868170707588142, - "loss": 0.5414, + "epoch": 1.42, + "learning_rate": 8.868170707588142e-05, + "loss": 0.5913, "step": 220 }, { - "epoch": 1.12, - "learning_rate": 0.0008858178535883905, - "loss": 0.4396, + "epoch": 1.43, + "learning_rate": 8.858178535883905e-05, + "loss": 0.564, "step": 221 }, { - "epoch": 1.13, - "learning_rate": 0.000884814813274243, - "loss": 0.4208, + "epoch": 1.43, + "learning_rate": 8.848148132742431e-05, + "loss": 0.5669, "step": 222 }, { - "epoch": 1.13, - "learning_rate": 0.0008838079597556925, - "loss": 0.3622, + "epoch": 1.44, + "learning_rate": 8.838079597556925e-05, + "loss": 0.4491, "step": 223 }, { - "epoch": 1.14, - "learning_rate": 0.0008827973030098447, - "loss": 0.5095, + "epoch": 1.45, + "learning_rate": 8.827973030098448e-05, + "loss": 0.4236, "step": 224 }, { - "epoch": 1.14, - "learning_rate": 0.0008817828530514931, - "loss": 0.3444, + "epoch": 1.45, + "learning_rate": 8.81782853051493e-05, + "loss": 0.4449, "step": 225 }, { - "epoch": 1.15, - "learning_rate": 0.0008807646199330186, - "loss": 0.3598, + "epoch": 1.46, + "learning_rate": 8.807646199330187e-05, + "loss": 0.5031, "step": 226 }, { - "epoch": 1.15, - "learning_rate": 0.0008797426137442896, - "loss": 0.2523, + "epoch": 1.46, + "learning_rate": 8.797426137442897e-05, + "loss": 0.5822, "step": 227 }, { - "epoch": 1.16, - "learning_rate": 0.0008787168446125638, - "loss": 0.3462, + "epoch": 1.47, + "learning_rate": 8.787168446125638e-05, + "loss": 0.4124, "step": 228 }, { - "epoch": 1.16, - "learning_rate": 0.0008776873227023851, - "loss": 0.4004, + "epoch": 1.48, + "learning_rate": 8.776873227023852e-05, + "loss": 0.5732, "step": 229 }, { - "epoch": 1.17, - "learning_rate": 0.0008766540582154859, - "loss": 0.4201, + "epoch": 1.48, + "learning_rate": 8.766540582154859e-05, + "loss": 0.5132, "step": 230 }, { - "epoch": 1.17, - "learning_rate": 0.0008756170613906833, - "loss": 0.4267, + "epoch": 1.49, + "learning_rate": 8.756170613906833e-05, + "loss": 0.5151, "step": 231 }, { - "epoch": 1.18, - "learning_rate": 0.0008745763425037796, - "loss": 0.4423, + "epoch": 1.5, + "learning_rate": 8.745763425037797e-05, + "loss": 0.5334, "step": 232 }, { - "epoch": 1.18, - "learning_rate": 0.0008735319118674597, - "loss": 0.5584, + "epoch": 1.5, + "learning_rate": 8.735319118674596e-05, + "loss": 0.5812, "step": 233 }, { - "epoch": 1.19, - "learning_rate": 0.0008724837798311882, - "loss": 0.2896, + "epoch": 1.51, + "learning_rate": 8.724837798311882e-05, + "loss": 0.4281, "step": 234 }, { - "epoch": 1.19, - "learning_rate": 0.0008714319567811089, - "loss": 0.4205, + "epoch": 1.52, + "learning_rate": 8.714319567811088e-05, + "loss": 0.5254, "step": 235 }, { - "epoch": 1.2, - "learning_rate": 0.0008703764531399392, - "loss": 0.38, + "epoch": 1.52, + "learning_rate": 8.703764531399392e-05, + "loss": 0.5139, "step": 236 }, { - "epoch": 1.2, - "learning_rate": 0.000869317279366869, - "loss": 0.4093, + "epoch": 1.53, + "learning_rate": 8.69317279366869e-05, + "loss": 0.5656, "step": 237 }, { - "epoch": 1.21, - "learning_rate": 0.0008682544459574561, - "loss": 0.4539, + "epoch": 1.54, + "learning_rate": 8.682544459574562e-05, + "loss": 0.4704, "step": 238 }, { - "epoch": 1.21, - "learning_rate": 0.0008671879634435224, - "loss": 0.4635, + "epoch": 1.54, + "learning_rate": 8.671879634435224e-05, + "loss": 0.4704, "step": 239 }, { - "epoch": 1.22, - "learning_rate": 0.0008661178423930491, - "loss": 0.3336, + "epoch": 1.55, + "learning_rate": 8.661178423930491e-05, + "loss": 0.4682, "step": 240 }, { - "epoch": 1.22, - "learning_rate": 0.0008650440934100728, - "loss": 0.3342, + "epoch": 1.55, + "learning_rate": 8.650440934100728e-05, + "loss": 0.516, "step": 241 }, { - "epoch": 1.23, - "learning_rate": 0.0008639667271345798, - "loss": 0.4935, + "epoch": 1.56, + "learning_rate": 8.639667271345798e-05, + "loss": 0.5812, "step": 242 }, { - "epoch": 1.23, - "learning_rate": 0.0008628857542424009, - "loss": 0.499, + "epoch": 1.57, + "learning_rate": 8.628857542424009e-05, + "loss": 0.6955, "step": 243 }, { - "epoch": 1.24, - "learning_rate": 0.0008618011854451056, - "loss": 0.6852, + "epoch": 1.57, + "learning_rate": 8.618011854451056e-05, + "loss": 0.4637, "step": 244 }, { - "epoch": 1.24, - "learning_rate": 0.0008607130314898956, - "loss": 0.5297, + "epoch": 1.58, + "learning_rate": 8.607130314898956e-05, + "loss": 0.5001, "step": 245 }, { - "epoch": 1.25, - "learning_rate": 0.000859621303159499, - "loss": 0.3527, + "epoch": 1.59, + "learning_rate": 8.596213031594991e-05, + "loss": 0.5324, "step": 246 }, { - "epoch": 1.25, - "learning_rate": 0.0008585260112720631, - "loss": 0.4286, + "epoch": 1.59, + "learning_rate": 8.585260112720631e-05, + "loss": 0.9362, "step": 247 }, { - "epoch": 1.26, - "learning_rate": 0.0008574271666810469, - "loss": 0.4928, + "epoch": 1.6, + "learning_rate": 8.57427166681047e-05, + "loss": 0.4641, "step": 248 }, { - "epoch": 1.26, - "learning_rate": 0.0008563247802751139, - "loss": 0.5073, + "epoch": 1.61, + "learning_rate": 8.56324780275114e-05, + "loss": 0.4814, "step": 249 }, { - "epoch": 1.27, - "learning_rate": 0.0008552188629780245, - "loss": 0.4953, + "epoch": 1.61, + "learning_rate": 8.552188629780244e-05, + "loss": 0.5503, "step": 250 }, { - "epoch": 1.27, - "learning_rate": 0.0008541094257485265, - "loss": 0.4353, + "epoch": 1.62, + "learning_rate": 8.541094257485265e-05, + "loss": 0.4649, "step": 251 }, { - "epoch": 1.28, - "learning_rate": 0.0008529964795802485, - "loss": 0.4549, + "epoch": 1.63, + "learning_rate": 8.529964795802485e-05, + "loss": 0.3803, "step": 252 }, { - "epoch": 1.28, - "learning_rate": 0.0008518800355015891, - "loss": 0.5555, + "epoch": 1.63, + "learning_rate": 8.518800355015892e-05, + "loss": 0.4112, "step": 253 }, { - "epoch": 1.29, - "learning_rate": 0.0008507601045756085, - "loss": 0.3896, + "epoch": 1.64, + "learning_rate": 8.507601045756085e-05, + "loss": 0.4634, "step": 254 }, { - "epoch": 1.29, - "learning_rate": 0.0008496366978999189, - "loss": 0.3865, + "epoch": 1.65, + "learning_rate": 8.49636697899919e-05, + "loss": 0.5406, "step": 255 }, { - "epoch": 1.3, - "learning_rate": 0.0008485098266065744, - "loss": 0.5022, + "epoch": 1.65, + "learning_rate": 8.485098266065744e-05, + "loss": 0.5009, "step": 256 }, { - "epoch": 1.3, - "learning_rate": 0.0008473795018619604, - "loss": 0.4086, + "epoch": 1.66, + "learning_rate": 8.473795018619604e-05, + "loss": 0.5179, "step": 257 }, { - "epoch": 1.31, - "learning_rate": 0.0008462457348666835, - "loss": 0.4797, + "epoch": 1.66, + "learning_rate": 8.462457348666835e-05, + "loss": 0.4948, "step": 258 }, { - "epoch": 1.31, - "learning_rate": 0.0008451085368554601, - "loss": 0.4812, + "epoch": 1.67, + "learning_rate": 8.4510853685546e-05, + "loss": 0.4825, "step": 259 }, { - "epoch": 1.32, - "learning_rate": 0.0008439679190970051, - "loss": 0.6112, + "epoch": 1.68, + "learning_rate": 8.439679190970052e-05, + "loss": 0.6783, "step": 260 }, { - "epoch": 1.32, - "learning_rate": 0.0008428238928939207, - "loss": 0.3997, + "epoch": 1.68, + "learning_rate": 8.428238928939207e-05, + "loss": 0.5107, "step": 261 }, { - "epoch": 1.33, - "learning_rate": 0.0008416764695825835, - "loss": 0.5654, + "epoch": 1.69, + "learning_rate": 8.416764695825835e-05, + "loss": 0.6285, "step": 262 }, { - "epoch": 1.34, - "learning_rate": 0.0008405256605330332, - "loss": 0.3926, + "epoch": 1.7, + "learning_rate": 8.405256605330331e-05, + "loss": 0.5409, "step": 263 }, { - "epoch": 1.34, - "learning_rate": 0.0008393714771488589, - "loss": 0.4882, + "epoch": 1.7, + "learning_rate": 8.39371477148859e-05, + "loss": 0.4913, "step": 264 }, { - "epoch": 1.35, - "learning_rate": 0.0008382139308670875, - "loss": 0.5966, + "epoch": 1.71, + "learning_rate": 8.382139308670875e-05, + "loss": 0.4905, "step": 265 }, { - "epoch": 1.35, - "learning_rate": 0.0008370530331580686, - "loss": 0.7363, + "epoch": 1.72, + "learning_rate": 8.370530331580686e-05, + "loss": 0.5521, "step": 266 }, { - "epoch": 1.36, - "learning_rate": 0.000835888795525362, - "loss": 0.4962, + "epoch": 1.72, + "learning_rate": 8.35888795525362e-05, + "loss": 0.6787, "step": 267 }, { - "epoch": 1.36, - "learning_rate": 0.0008347212295056239, - "loss": 0.4113, + "epoch": 1.73, + "learning_rate": 8.347212295056239e-05, + "loss": 0.6243, "step": 268 }, { - "epoch": 1.37, - "learning_rate": 0.0008335503466684915, - "loss": 0.6778, + "epoch": 1.74, + "learning_rate": 8.335503466684915e-05, + "loss": 0.5954, "step": 269 }, { - "epoch": 1.37, - "learning_rate": 0.0008323761586164695, - "loss": 0.4551, + "epoch": 1.74, + "learning_rate": 8.323761586164695e-05, + "loss": 0.4916, "step": 270 }, { - "epoch": 1.38, - "learning_rate": 0.0008311986769848141, - "loss": 0.3871, + "epoch": 1.75, + "learning_rate": 8.311986769848141e-05, + "loss": 0.4462, "step": 271 }, { - "epoch": 1.38, - "learning_rate": 0.0008300179134414187, - "loss": 0.521, + "epoch": 1.75, + "learning_rate": 8.300179134414188e-05, + "loss": 0.5472, "step": 272 }, { - "epoch": 1.39, - "learning_rate": 0.0008288338796866977, - "loss": 0.6278, + "epoch": 1.76, + "learning_rate": 8.288338796866976e-05, + "loss": 0.5206, "step": 273 }, { - "epoch": 1.39, - "learning_rate": 0.0008276465874534701, - "loss": 0.3758, + "epoch": 1.77, + "learning_rate": 8.276465874534702e-05, + "loss": 0.541, "step": 274 }, { - "epoch": 1.4, - "learning_rate": 0.0008264560485068446, - "loss": 0.5227, + "epoch": 1.77, + "learning_rate": 8.264560485068446e-05, + "loss": 0.5823, "step": 275 }, { - "epoch": 1.4, - "learning_rate": 0.0008252622746441021, - "loss": 0.3918, + "epoch": 1.78, + "learning_rate": 8.252622746441021e-05, + "loss": 0.5128, "step": 276 }, { - "epoch": 1.41, - "learning_rate": 0.0008240652776945781, - "loss": 0.377, + "epoch": 1.79, + "learning_rate": 8.240652776945781e-05, + "loss": 0.4968, "step": 277 }, { - "epoch": 1.41, - "learning_rate": 0.0008228650695195472, - "loss": 0.5587, + "epoch": 1.79, + "learning_rate": 8.228650695195472e-05, + "loss": 0.6532, "step": 278 }, { - "epoch": 1.42, - "learning_rate": 0.0008216616620121042, - "loss": 0.4371, + "epoch": 1.8, + "learning_rate": 8.216616620121043e-05, + "loss": 0.3638, "step": 279 }, { - "epoch": 1.42, - "learning_rate": 0.0008204550670970469, - "loss": 0.3701, + "epoch": 1.81, + "learning_rate": 8.204550670970469e-05, + "loss": 0.6636, "step": 280 }, { - "epoch": 1.43, - "learning_rate": 0.0008192452967307575, - "loss": 0.3613, + "epoch": 1.81, + "learning_rate": 8.192452967307576e-05, + "loss": 0.5279, "step": 281 }, { - "epoch": 1.43, - "learning_rate": 0.0008180323629010848, - "loss": 0.5874, + "epoch": 1.82, + "learning_rate": 8.180323629010848e-05, + "loss": 0.4919, "step": 282 }, { - "epoch": 1.44, - "learning_rate": 0.0008168162776272244, - "loss": 0.4882, + "epoch": 1.83, + "learning_rate": 8.168162776272244e-05, + "loss": 0.4528, "step": 283 }, { - "epoch": 1.44, - "learning_rate": 0.0008155970529596006, - "loss": 0.3851, + "epoch": 1.83, + "learning_rate": 8.155970529596006e-05, + "loss": 0.5357, "step": 284 }, { - "epoch": 1.45, - "learning_rate": 0.0008143747009797464, - "loss": 0.49, + "epoch": 1.84, + "learning_rate": 8.143747009797464e-05, + "loss": 0.5333, "step": 285 }, { - "epoch": 1.45, - "learning_rate": 0.000813149233800184, - "loss": 0.584, + "epoch": 1.85, + "learning_rate": 8.131492338001839e-05, + "loss": 0.4751, "step": 286 }, { - "epoch": 1.46, - "learning_rate": 0.0008119206635643044, - "loss": 0.4602, + "epoch": 1.85, + "learning_rate": 8.119206635643045e-05, + "loss": 0.5092, "step": 287 }, { - "epoch": 1.46, - "learning_rate": 0.0008106890024462481, - "loss": 0.5223, + "epoch": 1.86, + "learning_rate": 8.106890024462481e-05, + "loss": 0.6139, "step": 288 }, { - "epoch": 1.47, - "learning_rate": 0.0008094542626507828, - "loss": 0.4446, + "epoch": 1.86, + "learning_rate": 8.094542626507828e-05, + "loss": 0.4858, "step": 289 }, { - "epoch": 1.47, - "learning_rate": 0.0008082164564131844, - "loss": 0.4818, + "epoch": 1.87, + "learning_rate": 8.082164564131845e-05, + "loss": 0.5627, "step": 290 }, { - "epoch": 1.48, - "learning_rate": 0.000806975595999114, - "loss": 0.4525, + "epoch": 1.88, + "learning_rate": 8.069755959991142e-05, + "loss": 0.9165, "step": 291 }, { - "epoch": 1.48, - "learning_rate": 0.0008057316937044977, - "loss": 0.5049, + "epoch": 1.88, + "learning_rate": 8.057316937044977e-05, + "loss": 0.6534, "step": 292 }, { - "epoch": 1.49, - "learning_rate": 0.0008044847618554035, - "loss": 0.4057, + "epoch": 1.89, + "learning_rate": 8.044847618554034e-05, + "loss": 0.4697, "step": 293 }, { - "epoch": 1.49, - "learning_rate": 0.0008032348128079204, - "loss": 0.4718, + "epoch": 1.9, + "learning_rate": 8.032348128079203e-05, + "loss": 0.5343, "step": 294 }, { - "epoch": 1.5, - "learning_rate": 0.0008019818589480352, - "loss": 0.43, + "epoch": 1.9, + "learning_rate": 8.019818589480352e-05, + "loss": 0.5272, "step": 295 }, { - "epoch": 1.5, - "learning_rate": 0.0008007259126915101, - "loss": 0.4611, + "epoch": 1.91, + "learning_rate": 8.0072591269151e-05, + "loss": 0.4909, "step": 296 }, { - "epoch": 1.51, - "learning_rate": 0.0007994669864837594, - "loss": 0.5587, + "epoch": 1.92, + "learning_rate": 7.994669864837594e-05, + "loss": 0.5913, "step": 297 }, { - "epoch": 1.51, - "learning_rate": 0.0007982050927997263, - "loss": 0.7014, + "epoch": 1.92, + "learning_rate": 7.982050927997264e-05, + "loss": 0.4666, "step": 298 }, { - "epoch": 1.52, - "learning_rate": 0.0007969402441437595, - "loss": 0.4874, + "epoch": 1.93, + "learning_rate": 7.969402441437594e-05, + "loss": 0.6011, "step": 299 }, { - "epoch": 1.52, - "learning_rate": 0.0007956724530494887, - "loss": 0.6031, + "epoch": 1.94, + "learning_rate": 7.956724530494887e-05, + "loss": 0.6574, "step": 300 }, { - "epoch": 1.53, - "learning_rate": 0.0007944017320797012, - "loss": 0.4922, + "epoch": 1.94, + "learning_rate": 7.944017320797013e-05, + "loss": 0.6198, "step": 301 }, { - "epoch": 1.53, - "learning_rate": 0.0007931280938262169, - "loss": 0.5301, + "epoch": 1.95, + "learning_rate": 7.931280938262169e-05, + "loss": 0.5228, "step": 302 }, { - "epoch": 1.54, - "learning_rate": 0.0007918515509097633, - "loss": 0.5914, + "epoch": 1.95, + "learning_rate": 7.918515509097634e-05, + "loss": 0.5375, "step": 303 }, { - "epoch": 1.54, - "learning_rate": 0.0007905721159798514, - "loss": 0.5355, + "epoch": 1.96, + "learning_rate": 7.905721159798513e-05, + "loss": 0.6888, "step": 304 }, { - "epoch": 1.55, - "learning_rate": 0.000789289801714649, - "loss": 0.9378, + "epoch": 1.97, + "learning_rate": 7.89289801714649e-05, + "loss": 0.4663, "step": 305 }, { - "epoch": 1.55, - "learning_rate": 0.0007880046208208562, - "loss": 0.5175, + "epoch": 1.97, + "learning_rate": 7.880046208208563e-05, + "loss": 0.6382, "step": 306 }, { - "epoch": 1.56, - "learning_rate": 0.0007867165860335792, - "loss": 0.4921, + "epoch": 1.98, + "learning_rate": 7.867165860335792e-05, + "loss": 0.5565, "step": 307 }, { - "epoch": 1.56, - "learning_rate": 0.0007854257101162037, - "loss": 0.3571, + "epoch": 1.99, + "learning_rate": 7.854257101162037e-05, + "loss": 0.4628, "step": 308 }, { - "epoch": 1.57, - "learning_rate": 0.0007841320058602688, - "loss": 0.4316, + "epoch": 1.99, + "learning_rate": 7.841320058602688e-05, + "loss": 0.6364, "step": 309 }, { - "epoch": 1.57, - "learning_rate": 0.0007828354860853399, - "loss": 0.7104, + "epoch": 2.0, + "learning_rate": 7.828354860853399e-05, + "loss": 0.4293, "step": 310 }, { - "epoch": 1.58, - "learning_rate": 0.0007815361636388827, - "loss": 0.5402, + "epoch": 2.01, + "learning_rate": 7.815361636388827e-05, + "loss": 0.3665, "step": 311 }, { - "epoch": 1.58, - "learning_rate": 0.0007802340513961342, - "loss": 0.451, + "epoch": 2.01, + "learning_rate": 7.802340513961342e-05, + "loss": 0.3727, "step": 312 }, { - "epoch": 1.59, - "learning_rate": 0.0007789291622599765, - "loss": 0.4272, + "epoch": 2.02, + "learning_rate": 7.789291622599767e-05, + "loss": 0.3468, "step": 313 }, { - "epoch": 1.59, - "learning_rate": 0.0007776215091608085, - "loss": 0.6552, + "epoch": 2.03, + "learning_rate": 7.776215091608085e-05, + "loss": 0.4416, "step": 314 }, { - "epoch": 1.6, - "learning_rate": 0.0007763111050564178, - "loss": 0.5025, + "epoch": 2.03, + "learning_rate": 7.763111050564178e-05, + "loss": 0.3674, "step": 315 }, { - "epoch": 1.6, - "learning_rate": 0.0007749979629318516, - "loss": 0.4279, + "epoch": 2.04, + "learning_rate": 7.749979629318516e-05, + "loss": 0.3358, "step": 316 }, { - "epoch": 1.61, - "learning_rate": 0.0007736820957992895, - "loss": 0.5845, + "epoch": 2.05, + "learning_rate": 7.736820957992895e-05, + "loss": 0.4428, "step": 317 }, { - "epoch": 1.61, - "learning_rate": 0.0007723635166979133, - "loss": 0.4912, + "epoch": 2.05, + "learning_rate": 7.723635166979133e-05, + "loss": 0.3428, "step": 318 }, { - "epoch": 1.62, - "learning_rate": 0.0007710422386937784, - "loss": 0.4654, + "epoch": 2.06, + "learning_rate": 7.710422386937784e-05, + "loss": 0.3407, "step": 319 }, { - "epoch": 1.62, - "learning_rate": 0.0007697182748796841, - "loss": 0.3999, + "epoch": 2.06, + "learning_rate": 7.697182748796841e-05, + "loss": 0.3479, "step": 320 }, { - "epoch": 1.63, - "learning_rate": 0.0007683916383750436, - "loss": 0.584, + "epoch": 2.07, + "learning_rate": 7.683916383750436e-05, + "loss": 0.341, "step": 321 }, { - "epoch": 1.63, - "learning_rate": 0.0007670623423257548, - "loss": 0.4966, + "epoch": 2.08, + "learning_rate": 7.670623423257548e-05, + "loss": 0.3725, "step": 322 }, { - "epoch": 1.64, - "learning_rate": 0.0007657303999040692, - "loss": 0.3419, + "epoch": 2.08, + "learning_rate": 7.657303999040693e-05, + "loss": 0.3185, "step": 323 }, { - "epoch": 1.64, - "learning_rate": 0.0007643958243084619, - "loss": 0.5106, + "epoch": 2.09, + "learning_rate": 7.64395824308462e-05, + "loss": 0.3545, "step": 324 }, { - "epoch": 1.65, - "learning_rate": 0.0007630586287635007, - "loss": 0.5535, + "epoch": 2.1, + "learning_rate": 7.630586287635008e-05, + "loss": 0.4408, "step": 325 }, { - "epoch": 1.65, - "learning_rate": 0.0007617188265197149, - "loss": 0.4968, + "epoch": 2.1, + "learning_rate": 7.617188265197148e-05, + "loss": 0.3995, "step": 326 }, { - "epoch": 1.66, - "learning_rate": 0.0007603764308534636, - "loss": 0.4221, + "epoch": 2.11, + "learning_rate": 7.603764308534636e-05, + "loss": 0.4507, "step": 327 }, { - "epoch": 1.66, - "learning_rate": 0.0007590314550668054, - "loss": 0.5194, + "epoch": 2.12, + "learning_rate": 7.590314550668054e-05, + "loss": 0.3594, "step": 328 }, { - "epoch": 1.67, - "learning_rate": 0.0007576839124873654, - "loss": 0.4809, + "epoch": 2.12, + "learning_rate": 7.576839124873653e-05, + "loss": 0.2748, "step": 329 }, { - "epoch": 1.68, - "learning_rate": 0.0007563338164682036, - "loss": 0.4809, + "epoch": 2.13, + "learning_rate": 7.563338164682036e-05, + "loss": 0.3803, "step": 330 }, { - "epoch": 1.68, - "learning_rate": 0.0007549811803876825, - "loss": 0.4812, + "epoch": 2.14, + "learning_rate": 7.549811803876825e-05, + "loss": 0.3374, "step": 331 }, { - "epoch": 1.69, - "learning_rate": 0.0007536260176493348, - "loss": 0.4273, + "epoch": 2.14, + "learning_rate": 7.536260176493348e-05, + "loss": 0.3064, "step": 332 }, { - "epoch": 1.69, - "learning_rate": 0.00075226834168173, - "loss": 0.4738, + "epoch": 2.15, + "learning_rate": 7.5226834168173e-05, + "loss": 0.3877, "step": 333 }, { - "epoch": 1.7, - "learning_rate": 0.0007509081659383416, - "loss": 0.448, + "epoch": 2.15, + "learning_rate": 7.509081659383417e-05, + "loss": 0.4996, "step": 334 }, { - "epoch": 1.7, - "learning_rate": 0.0007495455038974146, - "loss": 0.4226, + "epoch": 2.16, + "learning_rate": 7.495455038974146e-05, + "loss": 0.3529, "step": 335 }, { - "epoch": 1.71, - "learning_rate": 0.0007481803690618304, - "loss": 0.4618, + "epoch": 2.17, + "learning_rate": 7.481803690618304e-05, + "loss": 0.3439, "step": 336 }, { - "epoch": 1.71, - "learning_rate": 0.000746812774958974, - "loss": 0.4488, + "epoch": 2.17, + "learning_rate": 7.46812774958974e-05, + "loss": 0.2838, "step": 337 }, { - "epoch": 1.72, - "learning_rate": 0.0007454427351405999, - "loss": 0.4695, + "epoch": 2.18, + "learning_rate": 7.454427351405999e-05, + "loss": 0.3637, "step": 338 }, { - "epoch": 1.72, - "learning_rate": 0.0007440702631826976, - "loss": 0.3598, + "epoch": 2.19, + "learning_rate": 7.440702631826977e-05, + "loss": 0.4256, "step": 339 }, { - "epoch": 1.73, - "learning_rate": 0.0007426953726853573, - "loss": 0.4688, + "epoch": 2.19, + "learning_rate": 7.426953726853574e-05, + "loss": 0.3401, "step": 340 }, { - "epoch": 1.73, - "learning_rate": 0.0007413180772726348, - "loss": 0.3983, + "epoch": 2.2, + "learning_rate": 7.413180772726348e-05, + "loss": 0.3433, "step": 341 }, { - "epoch": 1.74, - "learning_rate": 0.0007399383905924165, - "loss": 0.5252, + "epoch": 2.21, + "learning_rate": 7.399383905924165e-05, + "loss": 0.4915, "step": 342 }, { - "epoch": 1.74, - "learning_rate": 0.0007385563263162847, - "loss": 0.5176, + "epoch": 2.21, + "learning_rate": 7.385563263162847e-05, + "loss": 0.3406, "step": 343 }, { - "epoch": 1.75, - "learning_rate": 0.0007371718981393814, - "loss": 0.6202, + "epoch": 2.22, + "learning_rate": 7.371718981393815e-05, + "loss": 0.3037, "step": 344 }, { - "epoch": 1.75, - "learning_rate": 0.0007357851197802735, - "loss": 0.3895, + "epoch": 2.23, + "learning_rate": 7.357851197802735e-05, + "loss": 0.5157, "step": 345 }, { - "epoch": 1.76, - "learning_rate": 0.0007343960049808155, - "loss": 0.4626, + "epoch": 2.23, + "learning_rate": 7.343960049808156e-05, + "loss": 0.4369, "step": 346 }, { - "epoch": 1.76, - "learning_rate": 0.0007330045675060148, - "loss": 0.4189, + "epoch": 2.24, + "learning_rate": 7.330045675060149e-05, + "loss": 0.3087, "step": 347 }, { - "epoch": 1.77, - "learning_rate": 0.0007316108211438946, - "loss": 0.3928, + "epoch": 2.25, + "learning_rate": 7.316108211438945e-05, + "loss": 0.3795, "step": 348 }, { - "epoch": 1.77, - "learning_rate": 0.0007302147797053569, - "loss": 0.6127, + "epoch": 2.25, + "learning_rate": 7.302147797053569e-05, + "loss": 0.3669, "step": 349 }, { - "epoch": 1.78, - "learning_rate": 0.0007288164570240462, - "loss": 0.4124, + "epoch": 2.26, + "learning_rate": 7.288164570240463e-05, + "loss": 0.6638, "step": 350 }, { - "epoch": 1.78, - "learning_rate": 0.0007274158669562126, - "loss": 0.4845, + "epoch": 2.26, + "learning_rate": 7.274158669562126e-05, + "loss": 0.3923, "step": 351 }, { - "epoch": 1.79, - "learning_rate": 0.0007260130233805739, - "loss": 0.5578, + "epoch": 2.27, + "learning_rate": 7.26013023380574e-05, + "loss": 0.3798, "step": 352 }, { - "epoch": 1.79, - "learning_rate": 0.0007246079401981784, - "loss": 0.9514, + "epoch": 2.28, + "learning_rate": 7.246079401981784e-05, + "loss": 0.439, "step": 353 }, { - "epoch": 1.8, - "learning_rate": 0.0007232006313322668, - "loss": 0.4722, + "epoch": 2.28, + "learning_rate": 7.232006313322667e-05, + "loss": 0.431, "step": 354 }, { - "epoch": 1.8, - "learning_rate": 0.0007217911107281352, - "loss": 0.462, + "epoch": 2.29, + "learning_rate": 7.217911107281352e-05, + "loss": 0.3676, "step": 355 }, { - "epoch": 1.81, - "learning_rate": 0.0007203793923529956, - "loss": 0.5369, + "epoch": 2.3, + "learning_rate": 7.203793923529956e-05, + "loss": 0.5888, "step": 356 }, { - "epoch": 1.81, - "learning_rate": 0.0007189654901958386, - "loss": 0.5005, + "epoch": 2.3, + "learning_rate": 7.189654901958385e-05, + "loss": 0.5541, "step": 357 }, { - "epoch": 1.82, - "learning_rate": 0.0007175494182672939, - "loss": 0.5964, + "epoch": 2.31, + "learning_rate": 7.175494182672939e-05, + "loss": 0.3824, "step": 358 }, { - "epoch": 1.82, - "learning_rate": 0.0007161311905994922, - "loss": 0.4215, + "epoch": 2.32, + "learning_rate": 7.161311905994922e-05, + "loss": 0.342, "step": 359 }, { - "epoch": 1.83, - "learning_rate": 0.0007147108212459257, - "loss": 0.5255, + "epoch": 2.32, + "learning_rate": 7.147108212459257e-05, + "loss": 0.4249, "step": 360 }, { - "epoch": 1.83, - "learning_rate": 0.000713288324281309, - "loss": 0.4609, + "epoch": 2.33, + "learning_rate": 7.13288324281309e-05, + "loss": 0.3621, "step": 361 }, { - "epoch": 1.84, - "learning_rate": 0.0007118637138014396, - "loss": 0.5004, + "epoch": 2.34, + "learning_rate": 7.118637138014396e-05, + "loss": 0.3147, "step": 362 }, { - "epoch": 1.84, - "learning_rate": 0.0007104370039230583, - "loss": 0.5112, + "epoch": 2.34, + "learning_rate": 7.104370039230583e-05, + "loss": 0.4188, "step": 363 }, { - "epoch": 1.85, - "learning_rate": 0.0007090082087837092, - "loss": 0.572, + "epoch": 2.35, + "learning_rate": 7.090082087837092e-05, + "loss": 0.7176, "step": 364 }, { - "epoch": 1.85, - "learning_rate": 0.0007075773425415994, - "loss": 0.3962, + "epoch": 2.35, + "learning_rate": 7.075773425415994e-05, + "loss": 0.3622, "step": 365 }, { - "epoch": 1.86, - "learning_rate": 0.0007061444193754596, - "loss": 0.5081, + "epoch": 2.36, + "learning_rate": 7.061444193754596e-05, + "loss": 0.3533, "step": 366 }, { - "epoch": 1.86, - "learning_rate": 0.0007047094534844022, - "loss": 0.4317, + "epoch": 2.37, + "learning_rate": 7.047094534844023e-05, + "loss": 0.2945, "step": 367 }, { - "epoch": 1.87, - "learning_rate": 0.0007032724590877821, - "loss": 0.434, + "epoch": 2.37, + "learning_rate": 7.032724590877821e-05, + "loss": 0.3296, "step": 368 }, { - "epoch": 1.87, - "learning_rate": 0.0007018334504250545, - "loss": 0.424, + "epoch": 2.38, + "learning_rate": 7.018334504250545e-05, + "loss": 0.3552, "step": 369 }, { - "epoch": 1.88, - "learning_rate": 0.0007003924417556343, - "loss": 0.5263, + "epoch": 2.39, + "learning_rate": 7.003924417556343e-05, + "loss": 0.3736, "step": 370 }, { - "epoch": 1.88, - "learning_rate": 0.0006989494473587554, - "loss": 0.3515, + "epoch": 2.39, + "learning_rate": 6.989494473587554e-05, + "loss": 0.3778, "step": 371 }, { - "epoch": 1.89, - "learning_rate": 0.0006975044815333281, - "loss": 0.5854, + "epoch": 2.4, + "learning_rate": 6.975044815333282e-05, + "loss": 0.3759, "step": 372 }, { - "epoch": 1.89, - "learning_rate": 0.0006960575585977984, - "loss": 0.5951, + "epoch": 2.41, + "learning_rate": 6.960575585977984e-05, + "loss": 0.6022, "step": 373 }, { - "epoch": 1.9, - "learning_rate": 0.0006946086928900053, - "loss": 0.5526, + "epoch": 2.41, + "learning_rate": 6.946086928900054e-05, + "loss": 0.4204, "step": 374 }, { - "epoch": 1.9, - "learning_rate": 0.0006931578987670395, - "loss": 0.4395, + "epoch": 2.42, + "learning_rate": 6.931578987670396e-05, + "loss": 0.3803, "step": 375 }, { - "epoch": 1.91, - "learning_rate": 0.0006917051906051005, - "loss": 0.4951, + "epoch": 2.43, + "learning_rate": 6.917051906051006e-05, + "loss": 0.324, "step": 376 }, { - "epoch": 1.91, - "learning_rate": 0.0006902505827993541, - "loss": 0.5379, + "epoch": 2.43, + "learning_rate": 6.902505827993541e-05, + "loss": 0.3164, "step": 377 }, { - "epoch": 1.92, - "learning_rate": 0.0006887940897637908, - "loss": 0.4688, + "epoch": 2.44, + "learning_rate": 6.887940897637908e-05, + "loss": 0.3777, "step": 378 }, { - "epoch": 1.92, - "learning_rate": 0.0006873357259310815, - "loss": 0.422, + "epoch": 2.45, + "learning_rate": 6.873357259310815e-05, + "loss": 0.3591, "step": 379 }, { - "epoch": 1.93, - "learning_rate": 0.0006858755057524354, - "loss": 0.5442, + "epoch": 2.45, + "learning_rate": 6.858755057524354e-05, + "loss": 0.4109, "step": 380 }, { - "epoch": 1.93, - "learning_rate": 0.0006844134436974567, - "loss": 0.5796, + "epoch": 2.46, + "learning_rate": 6.844134436974567e-05, + "loss": 0.5666, "step": 381 }, { - "epoch": 1.94, - "learning_rate": 0.0006829495542540013, - "loss": 0.3116, + "epoch": 2.46, + "learning_rate": 6.829495542540013e-05, + "loss": 0.4039, "step": 382 }, { - "epoch": 1.94, - "learning_rate": 0.0006814838519280324, - "loss": 0.5884, + "epoch": 2.47, + "learning_rate": 6.814838519280324e-05, + "loss": 0.2895, "step": 383 }, { - "epoch": 1.95, - "learning_rate": 0.000680016351243478, - "loss": 0.4409, + "epoch": 2.48, + "learning_rate": 6.80016351243478e-05, + "loss": 0.4687, "step": 384 }, { - "epoch": 1.95, - "learning_rate": 0.0006785470667420861, - "loss": 0.5072, + "epoch": 2.48, + "learning_rate": 6.785470667420862e-05, + "loss": 0.4379, "step": 385 }, { - "epoch": 1.96, - "learning_rate": 0.000677076012983281, - "loss": 0.6952, + "epoch": 2.49, + "learning_rate": 6.77076012983281e-05, + "loss": 0.4803, "step": 386 }, { - "epoch": 1.96, - "learning_rate": 0.000675603204544019, - "loss": 0.5757, + "epoch": 2.5, + "learning_rate": 6.75603204544019e-05, + "loss": 0.3904, "step": 387 }, { - "epoch": 1.97, - "learning_rate": 0.0006741286560186436, - "loss": 0.4874, + "epoch": 2.5, + "learning_rate": 6.741286560186437e-05, + "loss": 0.4808, "step": 388 }, { - "epoch": 1.97, - "learning_rate": 0.0006726523820187413, - "loss": 0.6527, + "epoch": 2.51, + "learning_rate": 6.726523820187413e-05, + "loss": 0.2799, "step": 389 }, { - "epoch": 1.98, - "learning_rate": 0.0006711743971729967, - "loss": 0.3801, + "epoch": 2.52, + "learning_rate": 6.711743971729967e-05, + "loss": 0.4341, "step": 390 }, { - "epoch": 1.98, - "learning_rate": 0.0006696947161270476, - "loss": 0.5924, + "epoch": 2.52, + "learning_rate": 6.696947161270476e-05, + "loss": 0.3893, "step": 391 }, { - "epoch": 1.99, - "learning_rate": 0.0006682133535433393, - "loss": 0.4358, + "epoch": 2.53, + "learning_rate": 6.682133535433393e-05, + "loss": 0.469, "step": 392 }, { - "epoch": 1.99, - "learning_rate": 0.0006667303241009804, - "loss": 0.3506, + "epoch": 2.54, + "learning_rate": 6.667303241009803e-05, + "loss": 0.298, "step": 393 }, { - "epoch": 2.0, - "learning_rate": 0.0006652456424955964, - "loss": 0.4905, + "epoch": 2.54, + "learning_rate": 6.652456424955963e-05, + "loss": 0.4633, "step": 394 }, { - "epoch": 2.01, - "learning_rate": 0.0006637593234391843, - "loss": 0.207, + "epoch": 2.55, + "learning_rate": 6.637593234391843e-05, + "loss": 0.2929, "step": 395 }, { - "epoch": 2.01, - "learning_rate": 0.0006622713816599673, - "loss": 0.2183, + "epoch": 2.55, + "learning_rate": 6.622713816599673e-05, + "loss": 0.4443, "step": 396 }, { - "epoch": 2.02, - "learning_rate": 0.0006607818319022481, - "loss": 0.2478, + "epoch": 2.56, + "learning_rate": 6.60781831902248e-05, + "loss": 0.454, "step": 397 }, { - "epoch": 2.02, - "learning_rate": 0.0006592906889262632, - "loss": 0.2421, + "epoch": 2.57, + "learning_rate": 6.592906889262632e-05, + "loss": 0.2841, "step": 398 }, { - "epoch": 2.03, - "learning_rate": 0.0006577979675080369, - "loss": 0.2596, + "epoch": 2.57, + "learning_rate": 6.577979675080369e-05, + "loss": 0.292, "step": 399 }, { - "epoch": 2.03, - "learning_rate": 0.0006563036824392345, - "loss": 0.1455, + "epoch": 2.58, + "learning_rate": 6.563036824392344e-05, + "loss": 0.2953, "step": 400 }, { - "epoch": 2.04, - "learning_rate": 0.0006548078485270152, - "loss": 0.2416, + "epoch": 2.59, + "learning_rate": 6.548078485270152e-05, + "loss": 0.4211, "step": 401 }, { - "epoch": 2.04, - "learning_rate": 0.0006533104805938873, - "loss": 0.1597, + "epoch": 2.59, + "learning_rate": 6.533104805938873e-05, + "loss": 0.334, "step": 402 }, { - "epoch": 2.05, - "learning_rate": 0.0006518115934775586, - "loss": 0.2274, + "epoch": 2.6, + "learning_rate": 6.518115934775585e-05, + "loss": 0.3343, "step": 403 }, { - "epoch": 2.05, - "learning_rate": 0.0006503112020307916, - "loss": 0.2043, + "epoch": 2.61, + "learning_rate": 6.503112020307916e-05, + "loss": 0.5768, "step": 404 }, { - "epoch": 2.06, - "learning_rate": 0.0006488093211212554, - "loss": 0.1718, + "epoch": 2.61, + "learning_rate": 6.488093211212555e-05, + "loss": 0.3803, "step": 405 }, { - "epoch": 2.06, - "learning_rate": 0.0006473059656313782, - "loss": 0.2003, + "epoch": 2.62, + "learning_rate": 6.473059656313782e-05, + "loss": 0.3835, "step": 406 }, { - "epoch": 2.07, - "learning_rate": 0.0006458011504582005, - "loss": 0.2682, + "epoch": 2.63, + "learning_rate": 6.458011504582005e-05, + "loss": 0.475, "step": 407 }, { - "epoch": 2.07, - "learning_rate": 0.0006442948905132266, - "loss": 0.3159, + "epoch": 2.63, + "learning_rate": 6.442948905132266e-05, + "loss": 0.3802, "step": 408 }, { - "epoch": 2.08, - "learning_rate": 0.0006427872007222777, - "loss": 0.1926, + "epoch": 2.64, + "learning_rate": 6.427872007222777e-05, + "loss": 0.44, "step": 409 }, { - "epoch": 2.08, - "learning_rate": 0.0006412780960253436, - "loss": 0.1705, + "epoch": 2.65, + "learning_rate": 6.412780960253436e-05, + "loss": 0.3829, "step": 410 }, { - "epoch": 2.09, - "learning_rate": 0.0006397675913764345, - "loss": 0.244, + "epoch": 2.65, + "learning_rate": 6.397675913764347e-05, + "loss": 0.2887, "step": 411 }, { - "epoch": 2.09, - "learning_rate": 0.0006382557017434331, - "loss": 0.263, + "epoch": 2.66, + "learning_rate": 6.382557017434332e-05, + "loss": 0.3647, "step": 412 }, { - "epoch": 2.1, - "learning_rate": 0.0006367424421079462, - "loss": 0.1458, + "epoch": 2.66, + "learning_rate": 6.367424421079463e-05, + "loss": 0.3898, "step": 413 }, { - "epoch": 2.1, - "learning_rate": 0.0006352278274651562, - "loss": 0.2241, + "epoch": 2.67, + "learning_rate": 6.352278274651561e-05, + "loss": 0.3215, "step": 414 }, { - "epoch": 2.11, - "learning_rate": 0.0006337118728236722, - "loss": 0.1758, + "epoch": 2.68, + "learning_rate": 6.337118728236721e-05, + "loss": 0.3979, "step": 415 }, { - "epoch": 2.11, - "learning_rate": 0.0006321945932053822, - "loss": 0.2477, + "epoch": 2.68, + "learning_rate": 6.321945932053822e-05, + "loss": 0.3281, "step": 416 }, { - "epoch": 2.12, - "learning_rate": 0.0006306760036453034, - "loss": 0.2506, + "epoch": 2.69, + "learning_rate": 6.306760036453035e-05, + "loss": 0.3366, "step": 417 }, { - "epoch": 2.12, - "learning_rate": 0.0006291561191914333, - "loss": 0.1463, + "epoch": 2.7, + "learning_rate": 6.291561191914333e-05, + "loss": 0.3521, "step": 418 }, { - "epoch": 2.13, - "learning_rate": 0.0006276349549046008, - "loss": 0.2307, + "epoch": 2.7, + "learning_rate": 6.276349549046007e-05, + "loss": 0.3278, "step": 419 }, { - "epoch": 2.13, - "learning_rate": 0.0006261125258583171, - "loss": 0.2191, + "epoch": 2.71, + "learning_rate": 6.261125258583171e-05, + "loss": 0.3875, "step": 420 }, { - "epoch": 2.14, - "learning_rate": 0.0006245888471386262, - "loss": 0.168, + "epoch": 2.72, + "learning_rate": 6.245888471386263e-05, + "loss": 0.3578, "step": 421 }, { - "epoch": 2.14, - "learning_rate": 0.0006230639338439548, - "loss": 0.2035, + "epoch": 2.72, + "learning_rate": 6.230639338439549e-05, + "loss": 0.3993, "step": 422 }, { - "epoch": 2.15, - "learning_rate": 0.000621537801084964, - "loss": 0.1725, + "epoch": 2.73, + "learning_rate": 6.215378010849641e-05, + "loss": 0.3705, "step": 423 }, { - "epoch": 2.15, - "learning_rate": 0.0006200104639843984, - "loss": 0.2434, + "epoch": 2.74, + "learning_rate": 6.200104639843985e-05, + "loss": 0.3881, "step": 424 }, { - "epoch": 2.16, - "learning_rate": 0.0006184819376769364, - "loss": 0.156, + "epoch": 2.74, + "learning_rate": 6.184819376769364e-05, + "loss": 0.2901, "step": 425 }, { - "epoch": 2.16, - "learning_rate": 0.0006169522373090413, - "loss": 0.2614, + "epoch": 2.75, + "learning_rate": 6.169522373090412e-05, + "loss": 0.4248, "step": 426 }, { - "epoch": 2.17, - "learning_rate": 0.0006154213780388092, - "loss": 0.1728, + "epoch": 2.75, + "learning_rate": 6.154213780388092e-05, + "loss": 0.3594, "step": 427 }, { - "epoch": 2.17, - "learning_rate": 0.0006138893750358212, - "loss": 0.4142, + "epoch": 2.76, + "learning_rate": 6.138893750358212e-05, + "loss": 0.3262, "step": 428 }, { - "epoch": 2.18, - "learning_rate": 0.0006123562434809912, - "loss": 0.2358, + "epoch": 2.77, + "learning_rate": 6.123562434809912e-05, + "loss": 0.3315, "step": 429 }, { - "epoch": 2.18, - "learning_rate": 0.0006108219985664161, - "loss": 0.1511, + "epoch": 2.77, + "learning_rate": 6.108219985664161e-05, + "loss": 0.3014, "step": 430 }, { - "epoch": 2.19, - "learning_rate": 0.0006092866554952256, - "loss": 0.2079, + "epoch": 2.78, + "learning_rate": 6.0928665549522554e-05, + "loss": 0.394, "step": 431 }, { - "epoch": 2.19, - "learning_rate": 0.0006077502294814311, - "loss": 0.2778, + "epoch": 2.79, + "learning_rate": 6.0775022948143115e-05, + "loss": 0.3497, "step": 432 }, { - "epoch": 2.2, - "learning_rate": 0.000606212735749775, - "loss": 0.1745, + "epoch": 2.79, + "learning_rate": 6.06212735749775e-05, + "loss": 0.3582, "step": 433 }, { - "epoch": 2.2, - "learning_rate": 0.0006046741895355802, - "loss": 0.2087, + "epoch": 2.8, + "learning_rate": 6.046741895355802e-05, + "loss": 0.3075, "step": 434 }, { - "epoch": 2.21, - "learning_rate": 0.0006031346060845986, - "loss": 0.3661, + "epoch": 2.81, + "learning_rate": 6.031346060845986e-05, + "loss": 0.4744, "step": 435 }, { - "epoch": 2.21, - "learning_rate": 0.0006015940006528601, - "loss": 0.2497, + "epoch": 2.81, + "learning_rate": 6.015940006528602e-05, + "loss": 0.3455, "step": 436 }, { - "epoch": 2.22, - "learning_rate": 0.0006000523885065223, - "loss": 0.1748, + "epoch": 2.82, + "learning_rate": 6.0005238850652234e-05, + "loss": 0.4062, "step": 437 }, { - "epoch": 2.22, - "learning_rate": 0.0005985097849217179, - "loss": 0.2426, + "epoch": 2.83, + "learning_rate": 5.9850978492171794e-05, + "loss": 0.3979, "step": 438 }, { - "epoch": 2.23, - "learning_rate": 0.0005969662051844041, - "loss": 0.2053, + "epoch": 2.83, + "learning_rate": 5.96966205184404e-05, + "loss": 0.425, "step": 439 }, { - "epoch": 2.23, - "learning_rate": 0.0005954216645902108, - "loss": 0.2444, + "epoch": 2.84, + "learning_rate": 5.954216645902109e-05, + "loss": 0.4698, "step": 440 }, { - "epoch": 2.24, - "learning_rate": 0.00059387617844429, - "loss": 0.2316, + "epoch": 2.85, + "learning_rate": 5.9387617844429e-05, + "loss": 0.3623, "step": 441 }, { - "epoch": 2.24, - "learning_rate": 0.0005923297620611623, - "loss": 0.3341, + "epoch": 2.85, + "learning_rate": 5.923297620611623e-05, + "loss": 0.344, "step": 442 }, { - "epoch": 2.25, - "learning_rate": 0.0005907824307645668, - "loss": 0.2375, + "epoch": 2.86, + "learning_rate": 5.907824307645669e-05, + "loss": 0.3457, "step": 443 }, { - "epoch": 2.25, - "learning_rate": 0.0005892341998873089, - "loss": 0.1812, + "epoch": 2.86, + "learning_rate": 5.892341998873089e-05, + "loss": 0.3953, "step": 444 }, { - "epoch": 2.26, - "learning_rate": 0.0005876850847711072, - "loss": 0.1778, + "epoch": 2.87, + "learning_rate": 5.876850847711073e-05, + "loss": 0.2856, "step": 445 }, { - "epoch": 2.26, - "learning_rate": 0.0005861351007664434, - "loss": 0.2015, + "epoch": 2.88, + "learning_rate": 5.861351007664434e-05, + "loss": 0.385, "step": 446 }, { - "epoch": 2.27, - "learning_rate": 0.0005845842632324087, - "loss": 0.2272, + "epoch": 2.88, + "learning_rate": 5.845842632324088e-05, + "loss": 0.5248, "step": 447 }, { - "epoch": 2.27, - "learning_rate": 0.000583032587536552, - "loss": 0.2465, + "epoch": 2.89, + "learning_rate": 5.83032587536552e-05, + "loss": 0.3359, "step": 448 }, { - "epoch": 2.28, - "learning_rate": 0.0005814800890547278, - "loss": 0.2765, + "epoch": 2.9, + "learning_rate": 5.814800890547278e-05, + "loss": 0.327, "step": 449 }, { - "epoch": 2.28, - "learning_rate": 0.0005799267831709441, - "loss": 0.2826, + "epoch": 2.9, + "learning_rate": 5.799267831709442e-05, + "loss": 0.3425, "step": 450 }, { - "epoch": 2.29, - "learning_rate": 0.000578372685277209, - "loss": 0.2576, + "epoch": 2.91, + "learning_rate": 5.78372685277209e-05, + "loss": 0.3914, "step": 451 }, { - "epoch": 2.29, - "learning_rate": 0.0005768178107733791, - "loss": 0.3645, + "epoch": 2.92, + "learning_rate": 5.7681781077337905e-05, + "loss": 0.3865, "step": 452 }, { - "epoch": 2.3, - "learning_rate": 0.0005752621750670068, - "loss": 0.1209, + "epoch": 2.92, + "learning_rate": 5.752621750670068e-05, + "loss": 0.5089, "step": 453 }, { - "epoch": 2.3, - "learning_rate": 0.0005737057935731867, - "loss": 0.2813, + "epoch": 2.93, + "learning_rate": 5.737057935731868e-05, + "loss": 0.4281, "step": 454 }, { - "epoch": 2.31, - "learning_rate": 0.0005721486817144044, - "loss": 0.2002, + "epoch": 2.94, + "learning_rate": 5.721486817144044e-05, + "loss": 0.3415, "step": 455 }, { - "epoch": 2.31, - "learning_rate": 0.0005705908549203824, - "loss": 0.249, + "epoch": 2.94, + "learning_rate": 5.705908549203823e-05, + "loss": 0.334, "step": 456 }, { - "epoch": 2.32, - "learning_rate": 0.0005690323286279274, - "loss": 0.2097, + "epoch": 2.95, + "learning_rate": 5.690323286279274e-05, + "loss": 0.4623, "step": 457 }, { - "epoch": 2.32, - "learning_rate": 0.0005674731182807781, - "loss": 0.2456, + "epoch": 2.95, + "learning_rate": 5.674731182807781e-05, + "loss": 0.3411, "step": 458 }, { - "epoch": 2.33, - "learning_rate": 0.0005659132393294514, - "loss": 0.2035, + "epoch": 2.96, + "learning_rate": 5.659132393294514e-05, + "loss": 0.3804, "step": 459 }, { - "epoch": 2.34, - "learning_rate": 0.0005643527072310891, - "loss": 0.1656, + "epoch": 2.97, + "learning_rate": 5.643527072310891e-05, + "loss": 0.2749, "step": 460 }, { - "epoch": 2.34, - "learning_rate": 0.0005627915374493061, - "loss": 0.334, + "epoch": 2.97, + "learning_rate": 5.627915374493061e-05, + "loss": 0.3664, "step": 461 }, { - "epoch": 2.35, - "learning_rate": 0.0005612297454540352, - "loss": 0.171, + "epoch": 2.98, + "learning_rate": 5.612297454540352e-05, + "loss": 0.2758, "step": 462 }, { - "epoch": 2.35, - "learning_rate": 0.0005596673467213756, - "loss": 0.1982, + "epoch": 2.99, + "learning_rate": 5.596673467213756e-05, + "loss": 0.3713, "step": 463 }, { - "epoch": 2.36, - "learning_rate": 0.0005581043567334383, - "loss": 0.1921, + "epoch": 2.99, + "learning_rate": 5.581043567334383e-05, + "loss": 0.3681, "step": 464 }, { - "epoch": 2.36, - "learning_rate": 0.0005565407909781934, - "loss": 0.2609, + "epoch": 3.0, + "learning_rate": 5.5654079097819345e-05, + "loss": 0.3195, "step": 465 }, { - "epoch": 2.37, - "learning_rate": 0.0005549766649493165, - "loss": 0.2503, + "epoch": 3.01, + "learning_rate": 5.5497666494931654e-05, + "loss": 0.267, "step": 466 }, { - "epoch": 2.37, - "learning_rate": 0.000553411994146035, - "loss": 0.2269, + "epoch": 3.01, + "learning_rate": 5.5341199414603493e-05, + "loss": 0.2766, "step": 467 }, { - "epoch": 2.38, - "learning_rate": 0.0005518467940729739, - "loss": 0.1349, + "epoch": 3.02, + "learning_rate": 5.518467940729739e-05, + "loss": 0.2919, "step": 468 }, { - "epoch": 2.38, - "learning_rate": 0.0005502810802400039, - "loss": 0.2037, + "epoch": 3.03, + "learning_rate": 5.502810802400039e-05, + "loss": 0.2355, "step": 469 }, { - "epoch": 2.39, - "learning_rate": 0.0005487148681620861, - "loss": 0.1903, + "epoch": 3.03, + "learning_rate": 5.487148681620862e-05, + "loss": 0.2233, "step": 470 }, { - "epoch": 2.39, - "learning_rate": 0.000547148173359119, - "loss": 0.187, + "epoch": 3.04, + "learning_rate": 5.4714817335911894e-05, + "loss": 0.2188, "step": 471 }, { - "epoch": 2.4, - "learning_rate": 0.0005455810113557839, - "loss": 0.211, + "epoch": 3.05, + "learning_rate": 5.455810113557839e-05, + "loss": 0.2827, "step": 472 }, { - "epoch": 2.4, - "learning_rate": 0.0005440133976813926, - "loss": 0.2496, + "epoch": 3.05, + "learning_rate": 5.440133976813926e-05, + "loss": 0.1987, "step": 473 }, { - "epoch": 2.41, - "learning_rate": 0.0005424453478697321, - "loss": 0.2057, + "epoch": 3.06, + "learning_rate": 5.4244534786973214e-05, + "loss": 0.2357, "step": 474 }, { - "epoch": 2.41, - "learning_rate": 0.000540876877458911, - "loss": 0.1973, + "epoch": 3.06, + "learning_rate": 5.40876877458911e-05, + "loss": 0.1815, "step": 475 }, { - "epoch": 2.42, - "learning_rate": 0.0005393080019912061, - "loss": 0.232, + "epoch": 3.07, + "learning_rate": 5.3930800199120616e-05, + "loss": 0.1929, "step": 476 }, { - "epoch": 2.42, - "learning_rate": 0.0005377387370129079, - "loss": 0.1924, + "epoch": 3.08, + "learning_rate": 5.377387370129079e-05, + "loss": 0.1879, "step": 477 }, { - "epoch": 2.43, - "learning_rate": 0.0005361690980741663, - "loss": 0.2347, + "epoch": 3.08, + "learning_rate": 5.361690980741663e-05, + "loss": 0.2746, "step": 478 }, { - "epoch": 2.43, - "learning_rate": 0.0005345991007288371, - "loss": 0.2134, + "epoch": 3.09, + "learning_rate": 5.345991007288371e-05, + "loss": 0.2047, "step": 479 }, { - "epoch": 2.44, - "learning_rate": 0.000533028760534328, - "loss": 0.2552, + "epoch": 3.1, + "learning_rate": 5.330287605343279e-05, + "loss": 0.183, "step": 480 }, { - "epoch": 2.44, - "learning_rate": 0.0005314580930514431, - "loss": 0.3036, + "epoch": 3.1, + "learning_rate": 5.314580930514431e-05, + "loss": 0.1678, "step": 481 }, { - "epoch": 2.45, - "learning_rate": 0.0005298871138442307, - "loss": 0.1862, + "epoch": 3.11, + "learning_rate": 5.298871138442307e-05, + "loss": 0.2222, "step": 482 }, { - "epoch": 2.45, - "learning_rate": 0.0005283158384798275, - "loss": 0.3097, + "epoch": 3.12, + "learning_rate": 5.283158384798275e-05, + "loss": 0.2359, "step": 483 }, { - "epoch": 2.46, - "learning_rate": 0.0005267442825283048, - "loss": 0.2764, + "epoch": 3.12, + "learning_rate": 5.267442825283048e-05, + "loss": 0.4228, "step": 484 }, { - "epoch": 2.46, - "learning_rate": 0.0005251724615625145, - "loss": 0.2332, + "epoch": 3.13, + "learning_rate": 5.2517246156251455e-05, + "loss": 0.1614, "step": 485 }, { - "epoch": 2.47, - "learning_rate": 0.0005236003911579344, - "loss": 0.1589, + "epoch": 3.14, + "learning_rate": 5.236003911579345e-05, + "loss": 0.2356, "step": 486 }, { - "epoch": 2.47, - "learning_rate": 0.0005220280868925145, - "loss": 0.1943, + "epoch": 3.14, + "learning_rate": 5.220280868925145e-05, + "loss": 0.1942, "step": 487 }, { - "epoch": 2.48, - "learning_rate": 0.0005204555643465215, - "loss": 0.1713, + "epoch": 3.15, + "learning_rate": 5.204555643465215e-05, + "loss": 0.2173, "step": 488 }, { - "epoch": 2.48, - "learning_rate": 0.0005188828391023856, - "loss": 0.2151, + "epoch": 3.15, + "learning_rate": 5.1888283910238555e-05, + "loss": 0.2235, "step": 489 }, { - "epoch": 2.49, - "learning_rate": 0.0005173099267445451, - "loss": 0.1708, + "epoch": 3.16, + "learning_rate": 5.173099267445451e-05, + "loss": 0.228, "step": 490 }, { - "epoch": 2.49, - "learning_rate": 0.0005157368428592932, - "loss": 0.1367, + "epoch": 3.17, + "learning_rate": 5.157368428592933e-05, + "loss": 0.1412, "step": 491 }, { - "epoch": 2.5, - "learning_rate": 0.000514163603034622, - "loss": 0.2488, + "epoch": 3.17, + "learning_rate": 5.1416360303462206e-05, + "loss": 0.186, "step": 492 }, { - "epoch": 2.5, - "learning_rate": 0.0005125902228600693, - "loss": 0.1392, + "epoch": 3.18, + "learning_rate": 5.125902228600693e-05, + "loss": 0.1846, "step": 493 }, { - "epoch": 2.51, - "learning_rate": 0.0005110167179265636, - "loss": 0.2407, + "epoch": 3.19, + "learning_rate": 5.110167179265636e-05, + "loss": 0.1597, "step": 494 }, { - "epoch": 2.51, - "learning_rate": 0.0005094431038262692, - "loss": 0.2323, + "epoch": 3.19, + "learning_rate": 5.094431038262693e-05, + "loss": 0.2345, "step": 495 }, { - "epoch": 2.52, - "learning_rate": 0.0005078693961524329, - "loss": 0.236, + "epoch": 3.2, + "learning_rate": 5.078693961524329e-05, + "loss": 0.2507, "step": 496 }, { - "epoch": 2.52, - "learning_rate": 0.0005062956104992285, - "loss": 0.1665, + "epoch": 3.21, + "learning_rate": 5.062956104992285e-05, + "loss": 0.2224, "step": 497 }, { - "epoch": 2.53, - "learning_rate": 0.0005047217624616019, - "loss": 0.2289, + "epoch": 3.21, + "learning_rate": 5.0472176246160184e-05, + "loss": 0.2293, "step": 498 }, { - "epoch": 2.53, - "learning_rate": 0.0005031478676351178, - "loss": 0.6418, + "epoch": 3.22, + "learning_rate": 5.031478676351179e-05, + "loss": 0.251, "step": 499 }, { - "epoch": 2.54, - "learning_rate": 0.0005015739416158049, - "loss": 0.1473, + "epoch": 3.23, + "learning_rate": 5.01573941615805e-05, + "loss": 0.2296, "step": 500 }, { - "epoch": 2.54, - "learning_rate": 0.0005, - "loss": 0.2873, + "epoch": 3.23, + "learning_rate": 5e-05, + "loss": 0.2347, "step": 501 }, { - "epoch": 2.55, - "learning_rate": 0.0004984260583841952, - "loss": 0.1745, + "epoch": 3.24, + "learning_rate": 4.984260583841953e-05, + "loss": 0.2897, "step": 502 }, { - "epoch": 2.55, - "learning_rate": 0.0004968521323648822, - "loss": 0.2458, + "epoch": 3.25, + "learning_rate": 4.9685213236488216e-05, + "loss": 0.2051, "step": 503 }, { - "epoch": 2.56, - "learning_rate": 0.0004952782375383984, - "loss": 0.2808, + "epoch": 3.25, + "learning_rate": 4.9527823753839834e-05, + "loss": 0.2166, "step": 504 }, { - "epoch": 2.56, - "learning_rate": 0.0004937043895007717, - "loss": 0.1909, + "epoch": 3.26, + "learning_rate": 4.937043895007717e-05, + "loss": 0.2186, "step": 505 }, { - "epoch": 2.57, - "learning_rate": 0.0004921306038475671, - "loss": 0.2185, + "epoch": 3.26, + "learning_rate": 4.9213060384756716e-05, + "loss": 0.1655, "step": 506 }, { - "epoch": 2.57, - "learning_rate": 0.0004905568961737309, - "loss": 0.236, + "epoch": 3.27, + "learning_rate": 4.9055689617373084e-05, + "loss": 0.231, "step": 507 }, { - "epoch": 2.58, - "learning_rate": 0.0004889832820734367, - "loss": 0.2238, + "epoch": 3.28, + "learning_rate": 4.8898328207343666e-05, + "loss": 0.1842, "step": 508 }, { - "epoch": 2.58, - "learning_rate": 0.0004874097771399308, - "loss": 0.207, + "epoch": 3.28, + "learning_rate": 4.874097771399308e-05, + "loss": 0.2285, "step": 509 }, { - "epoch": 2.59, - "learning_rate": 0.0004858363969653781, - "loss": 0.2601, + "epoch": 3.29, + "learning_rate": 4.858363969653781e-05, + "loss": 0.205, "step": 510 }, { - "epoch": 2.59, - "learning_rate": 0.00048426315714070683, - "loss": 0.2049, + "epoch": 3.3, + "learning_rate": 4.8426315714070684e-05, + "loss": 0.2345, "step": 511 }, { - "epoch": 2.6, - "learning_rate": 0.00048269007325545506, - "loss": 0.2763, + "epoch": 3.3, + "learning_rate": 4.8269007325545506e-05, + "loss": 0.1897, "step": 512 }, { - "epoch": 2.6, - "learning_rate": 0.00048111716089761455, - "loss": 0.2787, + "epoch": 3.31, + "learning_rate": 4.8111716089761456e-05, + "loss": 0.1877, "step": 513 }, { - "epoch": 2.61, - "learning_rate": 0.00047954443565347867, - "loss": 0.2045, + "epoch": 3.32, + "learning_rate": 4.7954443565347865e-05, + "loss": 0.1964, "step": 514 }, { - "epoch": 2.61, - "learning_rate": 0.00047797191310748567, - "loss": 0.2062, + "epoch": 3.32, + "learning_rate": 4.779719131074857e-05, + "loss": 0.2365, "step": 515 }, { - "epoch": 2.62, - "learning_rate": 0.00047639960884206576, - "loss": 0.1595, + "epoch": 3.33, + "learning_rate": 4.7639960884206576e-05, + "loss": 0.1893, "step": 516 }, { - "epoch": 2.62, - "learning_rate": 0.0004748275384374856, - "loss": 0.1786, + "epoch": 3.34, + "learning_rate": 4.7482753843748564e-05, + "loss": 0.1933, "step": 517 }, { - "epoch": 2.63, - "learning_rate": 0.00047325571747169545, - "loss": 0.2422, + "epoch": 3.34, + "learning_rate": 4.7325571747169545e-05, + "loss": 0.2651, "step": 518 }, { - "epoch": 2.63, - "learning_rate": 0.00047168416152017255, - "loss": 0.1506, + "epoch": 3.35, + "learning_rate": 4.716841615201726e-05, + "loss": 0.2011, "step": 519 }, { - "epoch": 2.64, - "learning_rate": 0.0004701128861557694, - "loss": 0.1656, + "epoch": 3.35, + "learning_rate": 4.7011288615576934e-05, + "loss": 0.2375, "step": 520 }, { - "epoch": 2.64, - "learning_rate": 0.0004685419069485569, - "loss": 0.26, + "epoch": 3.36, + "learning_rate": 4.6854190694855694e-05, + "loss": 0.252, "step": 521 }, { - "epoch": 2.65, - "learning_rate": 0.0004669712394656723, - "loss": 0.3045, + "epoch": 3.37, + "learning_rate": 4.6697123946567227e-05, + "loss": 0.192, "step": 522 }, { - "epoch": 2.65, - "learning_rate": 0.00046540089927116305, - "loss": 0.2051, + "epoch": 3.37, + "learning_rate": 4.65400899271163e-05, + "loss": 0.1752, "step": 523 }, { - "epoch": 2.66, - "learning_rate": 0.00046383090192583397, - "loss": 0.198, + "epoch": 3.38, + "learning_rate": 4.63830901925834e-05, + "loss": 0.3051, "step": 524 }, { - "epoch": 2.66, - "learning_rate": 0.0004622612629870923, - "loss": 0.2119, + "epoch": 3.39, + "learning_rate": 4.6226126298709224e-05, + "loss": 0.1545, "step": 525 }, { - "epoch": 2.67, - "learning_rate": 0.00046069199800879404, - "loss": 0.2543, + "epoch": 3.39, + "learning_rate": 4.60691998008794e-05, + "loss": 0.2186, "step": 526 }, { - "epoch": 2.68, - "learning_rate": 0.00045912312254108907, - "loss": 0.2439, + "epoch": 3.4, + "learning_rate": 4.5912312254108905e-05, + "loss": 0.3145, "step": 527 }, { - "epoch": 2.68, - "learning_rate": 0.00045755465213026807, - "loss": 0.1628, + "epoch": 3.41, + "learning_rate": 4.575546521302681e-05, + "loss": 0.2099, "step": 528 }, { - "epoch": 2.69, - "learning_rate": 0.00045598660231860745, - "loss": 0.2464, + "epoch": 3.41, + "learning_rate": 4.5598660231860746e-05, + "loss": 0.2375, "step": 529 }, { - "epoch": 2.69, - "learning_rate": 0.0004544189886442162, - "loss": 0.1534, + "epoch": 3.42, + "learning_rate": 4.544189886442162e-05, + "loss": 0.2055, "step": 530 }, { - "epoch": 2.7, - "learning_rate": 0.0004528518266408811, - "loss": 0.2754, + "epoch": 3.43, + "learning_rate": 4.528518266408811e-05, + "loss": 0.1854, "step": 531 }, { - "epoch": 2.7, - "learning_rate": 0.00045128513183791385, - "loss": 0.2381, + "epoch": 3.43, + "learning_rate": 4.5128513183791386e-05, + "loss": 0.2062, "step": 532 }, { - "epoch": 2.71, - "learning_rate": 0.000449718919759996, - "loss": 0.3008, + "epoch": 3.44, + "learning_rate": 4.49718919759996e-05, + "loss": 0.1739, "step": 533 }, { - "epoch": 2.71, - "learning_rate": 0.0004481532059270262, - "loss": 0.3641, + "epoch": 3.45, + "learning_rate": 4.481532059270262e-05, + "loss": 0.2428, "step": 534 }, { - "epoch": 2.72, - "learning_rate": 0.0004465880058539652, - "loss": 0.3051, + "epoch": 3.45, + "learning_rate": 4.465880058539652e-05, + "loss": 0.2161, "step": 535 }, { - "epoch": 2.72, - "learning_rate": 0.00044502333505068355, - "loss": 0.2982, + "epoch": 3.46, + "learning_rate": 4.450233350506836e-05, + "loss": 0.2911, "step": 536 }, { - "epoch": 2.73, - "learning_rate": 0.0004434592090218065, - "loss": 0.1558, + "epoch": 3.46, + "learning_rate": 4.4345920902180647e-05, + "loss": 0.252, "step": 537 }, { - "epoch": 2.73, - "learning_rate": 0.0004418956432665618, - "loss": 0.2677, + "epoch": 3.47, + "learning_rate": 4.418956432665618e-05, + "loss": 0.1989, "step": 538 }, { - "epoch": 2.74, - "learning_rate": 0.0004403326532786245, - "loss": 0.2461, + "epoch": 3.48, + "learning_rate": 4.403326532786245e-05, + "loss": 0.1503, "step": 539 }, { - "epoch": 2.74, - "learning_rate": 0.00043877025454596493, - "loss": 0.2472, + "epoch": 3.48, + "learning_rate": 4.387702545459649e-05, + "loss": 0.2055, "step": 540 }, { - "epoch": 2.75, - "learning_rate": 0.000437208462550694, - "loss": 0.2212, + "epoch": 3.49, + "learning_rate": 4.3720846255069406e-05, + "loss": 0.1626, "step": 541 }, { - "epoch": 2.75, - "learning_rate": 0.0004356472927689109, - "loss": 0.198, + "epoch": 3.5, + "learning_rate": 4.356472927689109e-05, + "loss": 0.1413, "step": 542 }, { - "epoch": 2.76, - "learning_rate": 0.00043408676067054866, - "loss": 0.2927, + "epoch": 3.5, + "learning_rate": 4.3408676067054866e-05, + "loss": 0.2321, "step": 543 }, { - "epoch": 2.76, - "learning_rate": 0.000432526881719222, - "loss": 0.2636, + "epoch": 3.51, + "learning_rate": 4.32526881719222e-05, + "loss": 0.1704, "step": 544 }, { - "epoch": 2.77, - "learning_rate": 0.00043096767137207256, - "loss": 0.359, + "epoch": 3.52, + "learning_rate": 4.3096767137207256e-05, + "loss": 0.1724, "step": 545 }, { - "epoch": 2.77, - "learning_rate": 0.0004294091450796177, - "loss": 0.2385, + "epoch": 3.52, + "learning_rate": 4.2940914507961775e-05, + "loss": 0.2181, "step": 546 }, { - "epoch": 2.78, - "learning_rate": 0.0004278513182855956, - "loss": 0.136, + "epoch": 3.53, + "learning_rate": 4.278513182855956e-05, + "loss": 0.2005, "step": 547 }, { - "epoch": 2.78, - "learning_rate": 0.0004262942064268134, - "loss": 0.1927, + "epoch": 3.54, + "learning_rate": 4.262942064268134e-05, + "loss": 0.1698, "step": 548 }, { - "epoch": 2.79, - "learning_rate": 0.00042473782493299324, - "loss": 0.2469, + "epoch": 3.54, + "learning_rate": 4.247378249329933e-05, + "loss": 0.1967, "step": 549 }, { - "epoch": 2.79, - "learning_rate": 0.000423182189226621, - "loss": 0.252, + "epoch": 3.55, + "learning_rate": 4.23182189226621e-05, + "loss": 0.193, "step": 550 }, { - "epoch": 2.8, - "learning_rate": 0.00042162731472279095, - "loss": 0.2133, + "epoch": 3.55, + "learning_rate": 4.21627314722791e-05, + "loss": 0.229, "step": 551 }, { - "epoch": 2.8, - "learning_rate": 0.000420073216829056, - "loss": 0.2453, + "epoch": 3.56, + "learning_rate": 4.20073216829056e-05, + "loss": 0.2073, "step": 552 }, { - "epoch": 2.81, - "learning_rate": 0.0004185199109452721, - "loss": 0.274, + "epoch": 3.57, + "learning_rate": 4.185199109452721e-05, + "loss": 0.2214, "step": 553 }, { - "epoch": 2.81, - "learning_rate": 0.00041696741246344813, - "loss": 0.2342, + "epoch": 3.57, + "learning_rate": 4.169674124634481e-05, + "loss": 0.2383, "step": 554 }, { - "epoch": 2.82, - "learning_rate": 0.00041541573676759125, - "loss": 0.17, + "epoch": 3.58, + "learning_rate": 4.1541573676759126e-05, + "loss": 0.234, "step": 555 }, { - "epoch": 2.82, - "learning_rate": 0.00041386489923355666, - "loss": 0.234, + "epoch": 3.59, + "learning_rate": 4.138648992335566e-05, + "loss": 0.2308, "step": 556 }, { - "epoch": 2.83, - "learning_rate": 0.000412314915228893, - "loss": 0.1674, + "epoch": 3.59, + "learning_rate": 4.12314915228893e-05, + "loss": 0.2476, "step": 557 }, { - "epoch": 2.83, - "learning_rate": 0.00041076580011269127, - "loss": 0.2432, + "epoch": 3.6, + "learning_rate": 4.107658001126913e-05, + "loss": 0.2285, "step": 558 }, { - "epoch": 2.84, - "learning_rate": 0.0004092175692354333, - "loss": 0.2297, + "epoch": 3.61, + "learning_rate": 4.092175692354333e-05, + "loss": 0.2358, "step": 559 }, { - "epoch": 2.84, - "learning_rate": 0.00040767023793883785, - "loss": 0.2347, + "epoch": 3.61, + "learning_rate": 4.0767023793883785e-05, + "loss": 0.2443, "step": 560 }, { - "epoch": 2.85, - "learning_rate": 0.0004061238215557103, - "loss": 0.2065, + "epoch": 3.62, + "learning_rate": 4.0612382155571026e-05, + "loss": 0.1997, "step": 561 }, { - "epoch": 2.85, - "learning_rate": 0.0004045783354097893, - "loss": 0.1912, + "epoch": 3.63, + "learning_rate": 4.045783354097893e-05, + "loss": 0.246, "step": 562 }, { - "epoch": 2.86, - "learning_rate": 0.0004030337948155962, - "loss": 0.2439, + "epoch": 3.63, + "learning_rate": 4.0303379481559623e-05, + "loss": 0.2152, "step": 563 }, { - "epoch": 2.86, - "learning_rate": 0.00040149021507828223, - "loss": 0.1611, + "epoch": 3.64, + "learning_rate": 4.0149021507828224e-05, + "loss": 0.1601, "step": 564 }, { - "epoch": 2.87, - "learning_rate": 0.0003999476114934778, - "loss": 0.2625, + "epoch": 3.65, + "learning_rate": 3.9994761149347784e-05, + "loss": 0.1746, "step": 565 }, { - "epoch": 2.87, - "learning_rate": 0.0003984059993471399, - "loss": 0.1553, + "epoch": 3.65, + "learning_rate": 3.984059993471399e-05, + "loss": 0.2204, "step": 566 }, { - "epoch": 2.88, - "learning_rate": 0.0003968653939154016, - "loss": 0.2644, + "epoch": 3.66, + "learning_rate": 3.968653939154016e-05, + "loss": 0.2831, "step": 567 }, { - "epoch": 2.88, - "learning_rate": 0.00039532581046441994, - "loss": 0.1631, + "epoch": 3.66, + "learning_rate": 3.9532581046442e-05, + "loss": 0.1808, "step": 568 }, { - "epoch": 2.89, - "learning_rate": 0.00039378726425022515, - "loss": 0.2319, + "epoch": 3.67, + "learning_rate": 3.937872642502252e-05, + "loss": 0.1939, "step": 569 }, { - "epoch": 2.89, - "learning_rate": 0.00039224977051856904, - "loss": 0.1823, + "epoch": 3.68, + "learning_rate": 3.9224977051856904e-05, + "loss": 0.2102, "step": 570 }, { - "epoch": 2.9, - "learning_rate": 0.0003907133445047747, - "loss": 0.1732, + "epoch": 3.68, + "learning_rate": 3.907133445047747e-05, + "loss": 0.2231, "step": 571 }, { - "epoch": 2.9, - "learning_rate": 0.000389178001433584, - "loss": 0.1973, + "epoch": 3.69, + "learning_rate": 3.8917800143358404e-05, + "loss": 0.2068, "step": 572 }, { - "epoch": 2.91, - "learning_rate": 0.000387643756519009, - "loss": 0.2714, + "epoch": 3.7, + "learning_rate": 3.8764375651900906e-05, + "loss": 0.2311, "step": 573 }, { - "epoch": 2.91, - "learning_rate": 0.00038611062496417893, - "loss": 0.2124, + "epoch": 3.7, + "learning_rate": 3.861106249641789e-05, + "loss": 0.2331, "step": 574 }, { - "epoch": 2.92, - "learning_rate": 0.000384578621961191, - "loss": 0.3089, + "epoch": 3.71, + "learning_rate": 3.84578621961191e-05, + "loss": 0.255, "step": 575 }, { - "epoch": 2.92, - "learning_rate": 0.00038304776269095886, - "loss": 0.2345, + "epoch": 3.72, + "learning_rate": 3.830477626909589e-05, + "loss": 0.1869, "step": 576 }, { - "epoch": 2.93, - "learning_rate": 0.0003815180623230637, - "loss": 0.1838, + "epoch": 3.72, + "learning_rate": 3.8151806232306374e-05, + "loss": 0.1965, "step": 577 }, { - "epoch": 2.93, - "learning_rate": 0.0003799895360156017, - "loss": 0.1751, + "epoch": 3.73, + "learning_rate": 3.7998953601560175e-05, + "loss": 0.478, "step": 578 }, { - "epoch": 2.94, - "learning_rate": 0.00037846219891503606, - "loss": 0.1985, + "epoch": 3.74, + "learning_rate": 3.784621989150361e-05, + "loss": 0.2627, "step": 579 }, { - "epoch": 2.94, - "learning_rate": 0.00037693606615604524, - "loss": 0.2054, + "epoch": 3.74, + "learning_rate": 3.769360661560453e-05, + "loss": 0.2695, "step": 580 }, { - "epoch": 2.95, - "learning_rate": 0.00037541115286137403, - "loss": 0.3287, + "epoch": 3.75, + "learning_rate": 3.75411152861374e-05, + "loss": 0.2186, "step": 581 }, { - "epoch": 2.95, - "learning_rate": 0.00037388747414168294, - "loss": 0.1547, + "epoch": 3.75, + "learning_rate": 3.73887474141683e-05, + "loss": 0.1926, "step": 582 }, { - "epoch": 2.96, - "learning_rate": 0.00037236504509539936, - "loss": 0.2142, + "epoch": 3.76, + "learning_rate": 3.723650450953994e-05, + "loss": 0.2855, "step": 583 }, { - "epoch": 2.96, - "learning_rate": 0.0003708438808085668, - "loss": 0.2749, + "epoch": 3.77, + "learning_rate": 3.708438808085668e-05, + "loss": 0.2363, "step": 584 }, { - "epoch": 2.97, - "learning_rate": 0.00036932399635469673, - "loss": 0.251, + "epoch": 3.77, + "learning_rate": 3.693239963546967e-05, + "loss": 0.21, "step": 585 }, { - "epoch": 2.97, - "learning_rate": 0.0003678054067946178, - "loss": 0.3888, + "epoch": 3.78, + "learning_rate": 3.6780540679461784e-05, + "loss": 0.2214, "step": 586 }, { - "epoch": 2.98, - "learning_rate": 0.00036628812717632794, - "loss": 0.1405, + "epoch": 3.79, + "learning_rate": 3.662881271763279e-05, + "loss": 0.2023, "step": 587 }, { - "epoch": 2.98, - "learning_rate": 0.00036477217253484397, - "loss": 0.2749, + "epoch": 3.79, + "learning_rate": 3.64772172534844e-05, + "loss": 0.2352, "step": 588 }, { - "epoch": 2.99, - "learning_rate": 0.000363257557892054, - "loss": 0.1246, + "epoch": 3.8, + "learning_rate": 3.63257557892054e-05, + "loss": 0.2042, "step": 589 }, { - "epoch": 2.99, - "learning_rate": 0.00036174429825656685, - "loss": 0.2626, + "epoch": 3.81, + "learning_rate": 3.6174429825656685e-05, + "loss": 0.2864, "step": 590 }, { - "epoch": 3.0, - "learning_rate": 0.0003602324086235655, - "loss": 0.1572, + "epoch": 3.81, + "learning_rate": 3.602324086235655e-05, + "loss": 0.1612, "step": 591 }, { - "epoch": 3.01, - "learning_rate": 0.00035872190397465636, - "loss": 0.0793, + "epoch": 3.82, + "learning_rate": 3.587219039746564e-05, + "loss": 0.1933, "step": 592 }, { - "epoch": 3.01, - "learning_rate": 0.00035721279927772233, - "loss": 0.0366, + "epoch": 3.83, + "learning_rate": 3.572127992777223e-05, + "loss": 0.344, "step": 593 }, { - "epoch": 3.02, - "learning_rate": 0.0003557051094867735, - "loss": 0.0661, + "epoch": 3.83, + "learning_rate": 3.557051094867735e-05, + "loss": 0.1954, "step": 594 }, { - "epoch": 3.02, - "learning_rate": 0.0003541988495417997, - "loss": 0.0447, + "epoch": 3.84, + "learning_rate": 3.541988495417997e-05, + "loss": 0.2301, "step": 595 }, { - "epoch": 3.03, - "learning_rate": 0.00035269403436862174, - "loss": 0.0356, + "epoch": 3.85, + "learning_rate": 3.5269403436862175e-05, + "loss": 0.2151, "step": 596 }, { - "epoch": 3.03, - "learning_rate": 0.0003511906788787447, - "loss": 0.0507, + "epoch": 3.85, + "learning_rate": 3.511906788787447e-05, + "loss": 0.2214, "step": 597 }, { - "epoch": 3.04, - "learning_rate": 0.00034968879796920836, - "loss": 0.0461, + "epoch": 3.86, + "learning_rate": 3.496887979692084e-05, + "loss": 0.245, "step": 598 }, { - "epoch": 3.04, - "learning_rate": 0.0003481884065224415, - "loss": 0.0855, + "epoch": 3.86, + "learning_rate": 3.481884065224415e-05, + "loss": 0.2341, "step": 599 }, { - "epoch": 3.05, - "learning_rate": 0.00034668951940611275, - "loss": 0.0666, + "epoch": 3.87, + "learning_rate": 3.466895194061128e-05, + "loss": 0.2308, "step": 600 }, { - "epoch": 3.05, - "learning_rate": 0.0003451921514729848, - "loss": 0.0708, + "epoch": 3.88, + "learning_rate": 3.451921514729848e-05, + "loss": 0.2219, "step": 601 }, { - "epoch": 3.06, - "learning_rate": 0.0003436963175607656, - "loss": 0.0662, + "epoch": 3.88, + "learning_rate": 3.436963175607656e-05, + "loss": 0.2053, "step": 602 }, { - "epoch": 3.06, - "learning_rate": 0.00034220203249196317, - "loss": 0.0593, + "epoch": 3.89, + "learning_rate": 3.422020324919632e-05, + "loss": 0.3216, "step": 603 }, { - "epoch": 3.07, - "learning_rate": 0.0003407093110737368, - "loss": 0.0654, + "epoch": 3.9, + "learning_rate": 3.4070931107373675e-05, + "loss": 0.2443, "step": 604 }, { - "epoch": 3.07, - "learning_rate": 0.00033921816809775204, - "loss": 0.0445, + "epoch": 3.9, + "learning_rate": 3.39218168097752e-05, + "loss": 0.2597, "step": 605 }, { - "epoch": 3.08, - "learning_rate": 0.00033772861834003274, - "loss": 0.0355, + "epoch": 3.91, + "learning_rate": 3.377286183400328e-05, + "loss": 0.254, "step": 606 }, { - "epoch": 3.08, - "learning_rate": 0.00033624067656081575, - "loss": 0.0426, + "epoch": 3.92, + "learning_rate": 3.362406765608158e-05, + "loss": 0.1677, "step": 607 }, { - "epoch": 3.09, - "learning_rate": 0.0003347543575044035, - "loss": 0.0438, + "epoch": 3.92, + "learning_rate": 3.3475435750440356e-05, + "loss": 0.1773, "step": 608 }, { - "epoch": 3.09, - "learning_rate": 0.00033326967589901967, - "loss": 0.0836, + "epoch": 3.93, + "learning_rate": 3.332696758990197e-05, + "loss": 0.2142, "step": 609 }, { - "epoch": 3.1, - "learning_rate": 0.0003317866464566607, - "loss": 0.0479, + "epoch": 3.94, + "learning_rate": 3.3178664645666066e-05, + "loss": 0.2396, "step": 610 }, { - "epoch": 3.1, - "learning_rate": 0.0003303052838729525, - "loss": 0.1101, + "epoch": 3.94, + "learning_rate": 3.303052838729525e-05, + "loss": 0.2472, "step": 611 }, { - "epoch": 3.11, - "learning_rate": 0.00032882560282700336, - "loss": 0.0661, + "epoch": 3.95, + "learning_rate": 3.2882560282700336e-05, + "loss": 0.2072, "step": 612 }, { - "epoch": 3.11, - "learning_rate": 0.0003273476179812588, - "loss": 0.0452, + "epoch": 3.95, + "learning_rate": 3.273476179812588e-05, + "loss": 0.3123, "step": 613 }, { - "epoch": 3.12, - "learning_rate": 0.0003258713439813565, - "loss": 0.0338, + "epoch": 3.96, + "learning_rate": 3.258713439813566e-05, + "loss": 0.2578, "step": 614 }, { - "epoch": 3.12, - "learning_rate": 0.0003243967954559811, - "loss": 0.0277, + "epoch": 3.97, + "learning_rate": 3.243967954559811e-05, + "loss": 0.1629, "step": 615 }, { - "epoch": 3.13, - "learning_rate": 0.0003229239870167191, - "loss": 0.0559, + "epoch": 3.97, + "learning_rate": 3.229239870167191e-05, + "loss": 0.4055, "step": 616 }, { - "epoch": 3.13, - "learning_rate": 0.00032145293325791394, - "loss": 0.0732, + "epoch": 3.98, + "learning_rate": 3.2145293325791395e-05, + "loss": 0.192, "step": 617 }, { - "epoch": 3.14, - "learning_rate": 0.0003199836487565222, - "loss": 0.123, + "epoch": 3.99, + "learning_rate": 3.199836487565222e-05, + "loss": 0.2333, "step": 618 }, { - "epoch": 3.14, - "learning_rate": 0.0003185161480719677, - "loss": 0.0501, + "epoch": 3.99, + "learning_rate": 3.1851614807196774e-05, + "loss": 0.2915, "step": 619 }, { - "epoch": 3.15, - "learning_rate": 0.0003170504457459989, - "loss": 0.0836, + "epoch": 4.0, + "learning_rate": 3.170504457459989e-05, + "loss": 0.1784, "step": 620 }, { - "epoch": 3.15, - "learning_rate": 0.0003155865563025433, - "loss": 0.0464, + "epoch": 4.01, + "learning_rate": 3.155865563025433e-05, + "loss": 0.1412, "step": 621 }, { - "epoch": 3.16, - "learning_rate": 0.00031412449424756474, - "loss": 0.0927, + "epoch": 4.01, + "learning_rate": 3.1412449424756474e-05, + "loss": 0.1135, "step": 622 }, { - "epoch": 3.16, - "learning_rate": 0.0003126642740689186, - "loss": 0.0291, + "epoch": 4.02, + "learning_rate": 3.1266427406891856e-05, + "loss": 0.1025, "step": 623 }, { - "epoch": 3.17, - "learning_rate": 0.00031120591023620927, - "loss": 0.0793, + "epoch": 4.03, + "learning_rate": 3.112059102362093e-05, + "loss": 0.1087, "step": 624 }, { - "epoch": 3.17, - "learning_rate": 0.0003097494172006459, - "loss": 0.0812, + "epoch": 4.03, + "learning_rate": 3.0974941720064585e-05, + "loss": 0.1515, "step": 625 }, { - "epoch": 3.18, - "learning_rate": 0.00030829480939489965, - "loss": 0.0852, + "epoch": 4.04, + "learning_rate": 3.082948093948997e-05, + "loss": 0.1301, "step": 626 }, { - "epoch": 3.18, - "learning_rate": 0.00030684210123296053, - "loss": 0.1087, + "epoch": 4.05, + "learning_rate": 3.0684210123296055e-05, + "loss": 0.0812, "step": 627 }, { - "epoch": 3.19, - "learning_rate": 0.00030539130710999473, - "loss": 0.074, + "epoch": 4.05, + "learning_rate": 3.053913071099947e-05, + "loss": 0.1231, "step": 628 }, { - "epoch": 3.19, - "learning_rate": 0.0003039424414022016, - "loss": 0.0655, + "epoch": 4.06, + "learning_rate": 3.0394244140220163e-05, + "loss": 0.0913, "step": 629 }, { - "epoch": 3.2, - "learning_rate": 0.00030249551846667206, - "loss": 0.0622, + "epoch": 4.06, + "learning_rate": 3.0249551846667207e-05, + "loss": 0.1416, "step": 630 }, { - "epoch": 3.2, - "learning_rate": 0.0003010505526412447, - "loss": 0.0373, + "epoch": 4.07, + "learning_rate": 3.010505526412447e-05, + "loss": 0.126, "step": 631 }, { - "epoch": 3.21, - "learning_rate": 0.0002996075582443658, - "loss": 0.054, + "epoch": 4.08, + "learning_rate": 2.996075582443658e-05, + "loss": 0.1037, "step": 632 }, { - "epoch": 3.21, - "learning_rate": 0.0002981665495749457, - "loss": 0.0741, + "epoch": 4.08, + "learning_rate": 2.981665495749457e-05, + "loss": 0.0937, "step": 633 }, { - "epoch": 3.22, - "learning_rate": 0.00029672754091221806, - "loss": 0.0931, + "epoch": 4.09, + "learning_rate": 2.9672754091221805e-05, + "loss": 0.1061, "step": 634 }, { - "epoch": 3.22, - "learning_rate": 0.0002952905465155977, - "loss": 0.0934, + "epoch": 4.1, + "learning_rate": 2.9529054651559772e-05, + "loss": 0.1426, "step": 635 }, { - "epoch": 3.23, - "learning_rate": 0.0002938555806245406, - "loss": 0.089, + "epoch": 4.1, + "learning_rate": 2.938555806245406e-05, + "loss": 0.0995, "step": 636 }, { - "epoch": 3.23, - "learning_rate": 0.00029242265745840063, - "loss": 0.0873, + "epoch": 4.11, + "learning_rate": 2.9242265745840063e-05, + "loss": 0.0778, "step": 637 }, { - "epoch": 3.24, - "learning_rate": 0.00029099179121629115, - "loss": 0.0719, + "epoch": 4.12, + "learning_rate": 2.9099179121629117e-05, + "loss": 0.0848, "step": 638 }, { - "epoch": 3.24, - "learning_rate": 0.0002895629960769417, - "loss": 0.1076, + "epoch": 4.12, + "learning_rate": 2.895629960769417e-05, + "loss": 0.0843, "step": 639 }, { - "epoch": 3.25, - "learning_rate": 0.0002881362861985606, - "loss": 0.0738, + "epoch": 4.13, + "learning_rate": 2.881362861985606e-05, + "loss": 0.0779, "step": 640 }, { - "epoch": 3.25, - "learning_rate": 0.0002867116757186911, - "loss": 0.0744, + "epoch": 4.14, + "learning_rate": 2.867116757186911e-05, + "loss": 0.0856, "step": 641 }, { - "epoch": 3.26, - "learning_rate": 0.00028528917875407435, - "loss": 0.0468, + "epoch": 4.14, + "learning_rate": 2.8528917875407433e-05, + "loss": 0.1184, "step": 642 }, { - "epoch": 3.26, - "learning_rate": 0.0002838688094005078, - "loss": 0.0523, + "epoch": 4.15, + "learning_rate": 2.838688094005078e-05, + "loss": 0.0872, "step": 643 }, { - "epoch": 3.27, - "learning_rate": 0.00028245058173270623, - "loss": 0.0498, + "epoch": 4.15, + "learning_rate": 2.8245058173270622e-05, + "loss": 0.1039, "step": 644 }, { - "epoch": 3.27, - "learning_rate": 0.00028103450980416134, - "loss": 0.0621, + "epoch": 4.16, + "learning_rate": 2.8103450980416136e-05, + "loss": 0.1731, "step": 645 }, { - "epoch": 3.28, - "learning_rate": 0.0002796206076470044, - "loss": 0.0792, + "epoch": 4.17, + "learning_rate": 2.796206076470044e-05, + "loss": 0.0944, "step": 646 }, { - "epoch": 3.28, - "learning_rate": 0.0002782088892718648, - "loss": 0.0536, + "epoch": 4.17, + "learning_rate": 2.7820888927186483e-05, + "loss": 0.1147, "step": 647 }, { - "epoch": 3.29, - "learning_rate": 0.00027679936866773316, - "loss": 0.0689, + "epoch": 4.18, + "learning_rate": 2.7679936866773315e-05, + "loss": 0.2891, "step": 648 }, { - "epoch": 3.29, - "learning_rate": 0.0002753920598018217, - "loss": 0.0561, + "epoch": 4.19, + "learning_rate": 2.753920598018217e-05, + "loss": 0.0897, "step": 649 }, { - "epoch": 3.3, - "learning_rate": 0.00027398697661942627, - "loss": 0.0681, + "epoch": 4.19, + "learning_rate": 2.739869766194263e-05, + "loss": 0.1152, "step": 650 }, { - "epoch": 3.3, - "learning_rate": 0.00027258413304378734, - "loss": 0.0969, + "epoch": 4.2, + "learning_rate": 2.7258413304378734e-05, + "loss": 0.1273, "step": 651 }, { - "epoch": 3.31, - "learning_rate": 0.00027118354297595395, - "loss": 0.0473, + "epoch": 4.21, + "learning_rate": 2.7118354297595396e-05, + "loss": 0.1433, "step": 652 }, { - "epoch": 3.31, - "learning_rate": 0.00026978522029464324, - "loss": 0.0433, + "epoch": 4.21, + "learning_rate": 2.6978522029464325e-05, + "loss": 0.0997, "step": 653 }, { - "epoch": 3.32, - "learning_rate": 0.0002683891788561055, - "loss": 0.0564, + "epoch": 4.22, + "learning_rate": 2.683891788561055e-05, + "loss": 0.0809, "step": 654 }, { - "epoch": 3.32, - "learning_rate": 0.0002669954324939852, - "loss": 0.1075, + "epoch": 4.23, + "learning_rate": 2.669954324939852e-05, + "loss": 0.1114, "step": 655 }, { - "epoch": 3.33, - "learning_rate": 0.00026560399501918467, - "loss": 0.0576, + "epoch": 4.23, + "learning_rate": 2.6560399501918465e-05, + "loss": 0.1008, "step": 656 }, { - "epoch": 3.34, - "learning_rate": 0.0002642148802197267, - "loss": 0.0872, + "epoch": 4.24, + "learning_rate": 2.6421488021972673e-05, + "loss": 0.1733, "step": 657 }, { - "epoch": 3.34, - "learning_rate": 0.0002628281018606186, - "loss": 0.1002, + "epoch": 4.25, + "learning_rate": 2.6282810186061862e-05, + "loss": 0.1466, "step": 658 }, { - "epoch": 3.35, - "learning_rate": 0.00026144367368371535, - "loss": 0.0549, + "epoch": 4.25, + "learning_rate": 2.6144367368371535e-05, + "loss": 0.0989, "step": 659 }, { - "epoch": 3.35, - "learning_rate": 0.0002600616094075835, - "loss": 0.0571, + "epoch": 4.26, + "learning_rate": 2.600616094075835e-05, + "loss": 0.1372, "step": 660 }, { - "epoch": 3.36, - "learning_rate": 0.00025868192272736514, - "loss": 0.0875, + "epoch": 4.26, + "learning_rate": 2.5868192272736514e-05, + "loss": 0.0731, "step": 661 }, { - "epoch": 3.36, - "learning_rate": 0.0002573046273146427, - "loss": 0.0623, + "epoch": 4.27, + "learning_rate": 2.5730462731464273e-05, + "loss": 0.1069, "step": 662 }, { - "epoch": 3.37, - "learning_rate": 0.0002559297368173024, - "loss": 0.1347, + "epoch": 4.28, + "learning_rate": 2.5592973681730236e-05, + "loss": 0.1162, "step": 663 }, { - "epoch": 3.37, - "learning_rate": 0.00025455726485940013, - "loss": 0.0528, + "epoch": 4.28, + "learning_rate": 2.5455726485940012e-05, + "loss": 0.1549, "step": 664 }, { - "epoch": 3.38, - "learning_rate": 0.00025318722504102606, - "loss": 0.0508, + "epoch": 4.29, + "learning_rate": 2.5318722504102604e-05, + "loss": 0.0871, "step": 665 }, { - "epoch": 3.38, - "learning_rate": 0.0002518196309381696, - "loss": 0.0697, + "epoch": 4.3, + "learning_rate": 2.5181963093816962e-05, + "loss": 0.0946, "step": 666 }, { - "epoch": 3.39, - "learning_rate": 0.0002504544961025853, - "loss": 0.0655, + "epoch": 4.3, + "learning_rate": 2.504544961025853e-05, + "loss": 0.1013, "step": 667 }, { - "epoch": 3.39, - "learning_rate": 0.00024909183406165834, - "loss": 0.0566, + "epoch": 4.31, + "learning_rate": 2.4909183406165836e-05, + "loss": 0.1005, "step": 668 }, { - "epoch": 3.4, - "learning_rate": 0.0002477316583182702, - "loss": 0.0714, + "epoch": 4.32, + "learning_rate": 2.4773165831827018e-05, + "loss": 0.2741, "step": 669 }, { - "epoch": 3.4, - "learning_rate": 0.00024637398235066526, - "loss": 0.0747, + "epoch": 4.32, + "learning_rate": 2.4637398235066527e-05, + "loss": 0.0816, "step": 670 }, { - "epoch": 3.41, - "learning_rate": 0.0002450188196123177, - "loss": 0.0617, + "epoch": 4.33, + "learning_rate": 2.450188196123177e-05, + "loss": 0.1006, "step": 671 }, { - "epoch": 3.41, - "learning_rate": 0.00024366618353179642, - "loss": 0.0488, + "epoch": 4.34, + "learning_rate": 2.4366618353179644e-05, + "loss": 0.102, "step": 672 }, { - "epoch": 3.42, - "learning_rate": 0.0002423160875126348, - "loss": 0.0622, + "epoch": 4.34, + "learning_rate": 2.423160875126348e-05, + "loss": 0.1042, "step": 673 }, { - "epoch": 3.42, - "learning_rate": 0.00024096854493319475, - "loss": 0.0697, + "epoch": 4.35, + "learning_rate": 2.4096854493319477e-05, + "loss": 0.1389, "step": 674 }, { - "epoch": 3.43, - "learning_rate": 0.00023962356914653655, - "loss": 0.0897, + "epoch": 4.35, + "learning_rate": 2.3962356914653657e-05, + "loss": 0.1044, "step": 675 }, { - "epoch": 3.43, - "learning_rate": 0.00023828117348028528, - "loss": 0.078, + "epoch": 4.36, + "learning_rate": 2.3828117348028528e-05, + "loss": 0.085, "step": 676 }, { - "epoch": 3.44, - "learning_rate": 0.00023694137123649945, - "loss": 0.0958, + "epoch": 4.37, + "learning_rate": 2.3694137123649946e-05, + "loss": 0.0848, "step": 677 }, { - "epoch": 3.44, - "learning_rate": 0.00023560417569153797, - "loss": 0.1041, + "epoch": 4.37, + "learning_rate": 2.3560417569153796e-05, + "loss": 0.0948, "step": 678 }, { - "epoch": 3.45, - "learning_rate": 0.0002342696000959309, - "loss": 0.0466, + "epoch": 4.38, + "learning_rate": 2.342696000959309e-05, + "loss": 0.124, "step": 679 }, { - "epoch": 3.45, - "learning_rate": 0.00023293765767424534, - "loss": 0.0474, + "epoch": 4.39, + "learning_rate": 2.3293765767424537e-05, + "loss": 0.0876, "step": 680 }, { - "epoch": 3.46, - "learning_rate": 0.00023160836162495652, - "loss": 0.0809, + "epoch": 4.39, + "learning_rate": 2.3160836162495653e-05, + "loss": 0.0751, "step": 681 }, { - "epoch": 3.46, - "learning_rate": 0.00023028172512031605, - "loss": 0.0506, + "epoch": 4.4, + "learning_rate": 2.3028172512031604e-05, + "loss": 0.0939, "step": 682 }, { - "epoch": 3.47, - "learning_rate": 0.00022895776130622182, - "loss": 0.0766, + "epoch": 4.41, + "learning_rate": 2.289577613062218e-05, + "loss": 0.2399, "step": 683 }, { - "epoch": 3.47, - "learning_rate": 0.0002276364833020868, - "loss": 0.0307, + "epoch": 4.41, + "learning_rate": 2.276364833020868e-05, + "loss": 0.1172, "step": 684 }, { - "epoch": 3.48, - "learning_rate": 0.00022631790420071064, - "loss": 0.0586, + "epoch": 4.42, + "learning_rate": 2.2631790420071064e-05, + "loss": 0.1124, "step": 685 }, { - "epoch": 3.48, - "learning_rate": 0.00022500203706814854, - "loss": 0.0805, + "epoch": 4.43, + "learning_rate": 2.2500203706814856e-05, + "loss": 0.1065, "step": 686 }, { - "epoch": 3.49, - "learning_rate": 0.00022368889494358237, - "loss": 0.0766, + "epoch": 4.43, + "learning_rate": 2.2368889494358235e-05, + "loss": 0.1028, "step": 687 }, { - "epoch": 3.49, - "learning_rate": 0.00022237849083919144, - "loss": 0.0723, + "epoch": 4.44, + "learning_rate": 2.2237849083919142e-05, + "loss": 0.1354, "step": 688 }, { - "epoch": 3.5, - "learning_rate": 0.00022107083774002361, - "loss": 0.0643, + "epoch": 4.45, + "learning_rate": 2.2107083774002364e-05, + "loss": 0.1364, "step": 689 }, { - "epoch": 3.5, - "learning_rate": 0.00021976594860386594, - "loss": 0.0543, + "epoch": 4.45, + "learning_rate": 2.1976594860386597e-05, + "loss": 0.1475, "step": 690 }, { - "epoch": 3.51, - "learning_rate": 0.00021846383636111745, - "loss": 0.0654, + "epoch": 4.46, + "learning_rate": 2.1846383636111743e-05, + "loss": 0.1157, "step": 691 }, { - "epoch": 3.51, - "learning_rate": 0.0002171645139146601, - "loss": 0.066, + "epoch": 4.46, + "learning_rate": 2.1716451391466008e-05, + "loss": 0.108, "step": 692 }, { - "epoch": 3.52, - "learning_rate": 0.00021586799413973135, - "loss": 0.0478, + "epoch": 4.47, + "learning_rate": 2.1586799413973135e-05, + "loss": 0.0873, "step": 693 }, { - "epoch": 3.52, - "learning_rate": 0.00021457428988379634, - "loss": 0.1587, + "epoch": 4.48, + "learning_rate": 2.1457428988379635e-05, + "loss": 0.1704, "step": 694 }, { - "epoch": 3.53, - "learning_rate": 0.00021328341396642091, - "loss": 0.0895, + "epoch": 4.48, + "learning_rate": 2.1328341396642093e-05, + "loss": 0.1362, "step": 695 }, { - "epoch": 3.53, - "learning_rate": 0.00021199537917914385, - "loss": 0.0559, + "epoch": 4.49, + "learning_rate": 2.1199537917914386e-05, + "loss": 0.1068, "step": 696 }, { - "epoch": 3.54, - "learning_rate": 0.0002107101982853511, - "loss": 0.0638, + "epoch": 4.5, + "learning_rate": 2.107101982853511e-05, + "loss": 0.1443, "step": 697 }, { - "epoch": 3.54, - "learning_rate": 0.00020942788402014868, - "loss": 0.0335, + "epoch": 4.5, + "learning_rate": 2.0942788402014867e-05, + "loss": 0.0976, "step": 698 }, { - "epoch": 3.55, - "learning_rate": 0.0002081484490902366, - "loss": 0.2609, + "epoch": 4.51, + "learning_rate": 2.0814844909023663e-05, + "loss": 0.1091, "step": 699 }, { - "epoch": 3.55, - "learning_rate": 0.0002068719061737831, - "loss": 0.0565, + "epoch": 4.52, + "learning_rate": 2.068719061737831e-05, + "loss": 0.138, "step": 700 }, { - "epoch": 3.56, - "learning_rate": 0.00020559826792029885, - "loss": 0.0791, + "epoch": 4.52, + "learning_rate": 2.0559826792029884e-05, + "loss": 0.1105, "step": 701 }, { - "epoch": 3.56, - "learning_rate": 0.00020432754695051135, - "loss": 0.0252, + "epoch": 4.53, + "learning_rate": 2.0432754695051136e-05, + "loss": 0.0952, "step": 702 }, { - "epoch": 3.57, - "learning_rate": 0.0002030597558562406, - "loss": 0.0817, + "epoch": 4.54, + "learning_rate": 2.0305975585624058e-05, + "loss": 0.1463, "step": 703 }, { - "epoch": 3.57, - "learning_rate": 0.0002017949072002737, - "loss": 0.014, + "epoch": 4.54, + "learning_rate": 2.0179490720027372e-05, + "loss": 0.0833, "step": 704 }, { - "epoch": 3.58, - "learning_rate": 0.0002005330135162408, - "loss": 0.0377, + "epoch": 4.55, + "learning_rate": 2.005330135162408e-05, + "loss": 0.1192, "step": 705 }, { - "epoch": 3.58, - "learning_rate": 0.00019927408730848988, - "loss": 0.0633, + "epoch": 4.55, + "learning_rate": 1.992740873084899e-05, + "loss": 0.1378, "step": 706 }, { - "epoch": 3.59, - "learning_rate": 0.00019801814105196498, - "loss": 0.0667, + "epoch": 4.56, + "learning_rate": 1.9801814105196497e-05, + "loss": 0.0791, "step": 707 }, { - "epoch": 3.59, - "learning_rate": 0.00019676518719207977, - "loss": 0.0506, + "epoch": 4.57, + "learning_rate": 1.9676518719207977e-05, + "loss": 0.1302, "step": 708 }, { - "epoch": 3.6, - "learning_rate": 0.00019551523814459665, - "loss": 0.0661, + "epoch": 4.57, + "learning_rate": 1.9551523814459665e-05, + "loss": 0.1174, "step": 709 }, { - "epoch": 3.6, - "learning_rate": 0.0001942683062955024, - "loss": 0.0808, + "epoch": 4.58, + "learning_rate": 1.9426830629550242e-05, + "loss": 0.0775, "step": 710 }, { - "epoch": 3.61, - "learning_rate": 0.00019302440400088605, - "loss": 0.0402, + "epoch": 4.59, + "learning_rate": 1.9302440400088606e-05, + "loss": 0.0887, "step": 711 }, { - "epoch": 3.61, - "learning_rate": 0.00019178354358681548, - "loss": 0.0523, + "epoch": 4.59, + "learning_rate": 1.917835435868155e-05, + "loss": 0.1085, "step": 712 }, { - "epoch": 3.62, - "learning_rate": 0.00019054573734921716, - "loss": 0.0338, + "epoch": 4.6, + "learning_rate": 1.9054573734921714e-05, + "loss": 0.0957, "step": 713 }, { - "epoch": 3.62, - "learning_rate": 0.00018931099755375202, - "loss": 0.0693, + "epoch": 4.61, + "learning_rate": 1.8931099755375203e-05, + "loss": 0.1174, "step": 714 }, { - "epoch": 3.63, - "learning_rate": 0.00018807933643569559, - "loss": 0.0384, + "epoch": 4.61, + "learning_rate": 1.880793364356956e-05, + "loss": 0.1021, "step": 715 }, { - "epoch": 3.63, - "learning_rate": 0.00018685076619981607, - "loss": 0.0587, + "epoch": 4.62, + "learning_rate": 1.8685076619981608e-05, + "loss": 0.1318, "step": 716 }, { - "epoch": 3.64, - "learning_rate": 0.0001856252990202537, - "loss": 0.0533, + "epoch": 4.63, + "learning_rate": 1.8562529902025372e-05, + "loss": 0.1226, "step": 717 }, { - "epoch": 3.64, - "learning_rate": 0.0001844029470403993, - "loss": 0.0493, + "epoch": 4.63, + "learning_rate": 1.844029470403993e-05, + "loss": 0.0972, "step": 718 }, { - "epoch": 3.65, - "learning_rate": 0.00018318372237277565, - "loss": 0.0897, + "epoch": 4.64, + "learning_rate": 1.8318372237277565e-05, + "loss": 0.2055, "step": 719 }, { - "epoch": 3.65, - "learning_rate": 0.00018196763709891524, - "loss": 0.0726, + "epoch": 4.65, + "learning_rate": 1.8196763709891524e-05, + "loss": 0.0886, "step": 720 }, { - "epoch": 3.66, - "learning_rate": 0.0001807547032692424, - "loss": 0.0668, + "epoch": 4.65, + "learning_rate": 1.8075470326924243e-05, + "loss": 0.1226, "step": 721 }, { - "epoch": 3.66, - "learning_rate": 0.0001795449329029531, - "loss": 0.0753, + "epoch": 4.66, + "learning_rate": 1.795449329029531e-05, + "loss": 0.1223, "step": 722 }, { - "epoch": 3.67, - "learning_rate": 0.00017833833798789594, - "loss": 0.071, + "epoch": 4.66, + "learning_rate": 1.7833833798789595e-05, + "loss": 0.0861, "step": 723 }, { - "epoch": 3.68, - "learning_rate": 0.00017713493048045293, - "loss": 0.0453, + "epoch": 4.67, + "learning_rate": 1.7713493048045294e-05, + "loss": 0.1125, "step": 724 }, { - "epoch": 3.68, - "learning_rate": 0.000175934722305422, - "loss": 0.0406, + "epoch": 4.68, + "learning_rate": 1.7593472230542202e-05, + "loss": 0.1058, "step": 725 }, { - "epoch": 3.69, - "learning_rate": 0.0001747377253558982, - "loss": 0.0635, + "epoch": 4.68, + "learning_rate": 1.747377253558982e-05, + "loss": 0.1082, "step": 726 }, { - "epoch": 3.69, - "learning_rate": 0.00017354395149315533, - "loss": 0.0737, + "epoch": 4.69, + "learning_rate": 1.7354395149315534e-05, + "loss": 0.1018, "step": 727 }, { - "epoch": 3.7, - "learning_rate": 0.00017235341254653004, - "loss": 0.0618, + "epoch": 4.7, + "learning_rate": 1.7235341254653005e-05, + "loss": 0.1446, "step": 728 }, { - "epoch": 3.7, - "learning_rate": 0.00017116612031330253, - "loss": 0.0542, + "epoch": 4.7, + "learning_rate": 1.7116612031330252e-05, + "loss": 0.1118, "step": 729 }, { - "epoch": 3.71, - "learning_rate": 0.00016998208655858137, - "loss": 0.081, + "epoch": 4.71, + "learning_rate": 1.6998208655858137e-05, + "loss": 0.1146, "step": 730 }, { - "epoch": 3.71, - "learning_rate": 0.00016880132301518597, - "loss": 0.0573, + "epoch": 4.72, + "learning_rate": 1.6880132301518598e-05, + "loss": 0.1042, "step": 731 }, { - "epoch": 3.72, - "learning_rate": 0.00016762384138353076, - "loss": 0.0335, + "epoch": 4.72, + "learning_rate": 1.6762384138353078e-05, + "loss": 0.145, "step": 732 }, { - "epoch": 3.72, - "learning_rate": 0.00016644965333150846, - "loss": 0.0498, + "epoch": 4.73, + "learning_rate": 1.6644965333150847e-05, + "loss": 0.0899, "step": 733 }, { - "epoch": 3.73, - "learning_rate": 0.00016527877049437624, - "loss": 0.0684, + "epoch": 4.74, + "learning_rate": 1.6527877049437622e-05, + "loss": 0.1717, "step": 734 }, { - "epoch": 3.73, - "learning_rate": 0.00016411120447463806, - "loss": 0.0448, + "epoch": 4.74, + "learning_rate": 1.6411120447463807e-05, + "loss": 0.0906, "step": 735 }, { - "epoch": 3.74, - "learning_rate": 0.00016294696684193154, - "loss": 0.0779, + "epoch": 4.75, + "learning_rate": 1.6294696684193154e-05, + "loss": 0.1858, "step": 736 }, { - "epoch": 3.74, - "learning_rate": 0.0001617860691329126, - "loss": 0.0652, + "epoch": 4.75, + "learning_rate": 1.617860691329126e-05, + "loss": 0.1178, "step": 737 }, { - "epoch": 3.75, - "learning_rate": 0.00016062852285114122, - "loss": 0.0356, + "epoch": 4.76, + "learning_rate": 1.6062852285114123e-05, + "loss": 0.0934, "step": 738 }, { - "epoch": 3.75, - "learning_rate": 0.00015947433946696693, - "loss": 0.1102, + "epoch": 4.77, + "learning_rate": 1.5947433946696693e-05, + "loss": 0.0981, "step": 739 }, { - "epoch": 3.76, - "learning_rate": 0.0001583235304174167, - "loss": 0.0407, + "epoch": 4.77, + "learning_rate": 1.583235304174167e-05, + "loss": 0.1257, "step": 740 }, { - "epoch": 3.76, - "learning_rate": 0.00015717610710607949, - "loss": 0.0249, + "epoch": 4.78, + "learning_rate": 1.5717610710607948e-05, + "loss": 0.1439, "step": 741 }, { - "epoch": 3.77, - "learning_rate": 0.00015603208090299498, - "loss": 0.053, + "epoch": 4.79, + "learning_rate": 1.5603208090299498e-05, + "loss": 0.0967, "step": 742 }, { - "epoch": 3.77, - "learning_rate": 0.00015489146314454, - "loss": 0.0357, + "epoch": 4.79, + "learning_rate": 1.5489146314454002e-05, + "loss": 0.1511, "step": 743 }, { - "epoch": 3.78, - "learning_rate": 0.0001537542651333167, - "loss": 0.0729, + "epoch": 4.8, + "learning_rate": 1.537542651333167e-05, + "loss": 0.0943, "step": 744 }, { - "epoch": 3.78, - "learning_rate": 0.00015262049813803956, - "loss": 0.0398, + "epoch": 4.81, + "learning_rate": 1.5262049813803958e-05, + "loss": 0.0903, "step": 745 }, { - "epoch": 3.79, - "learning_rate": 0.00015149017339342576, - "loss": 0.0674, + "epoch": 4.81, + "learning_rate": 1.5149017339342574e-05, + "loss": 0.1197, "step": 746 }, { - "epoch": 3.79, - "learning_rate": 0.00015036330210008115, - "loss": 0.0381, + "epoch": 4.82, + "learning_rate": 1.5036330210008115e-05, + "loss": 0.146, "step": 747 }, { - "epoch": 3.8, - "learning_rate": 0.0001492398954243916, - "loss": 0.0765, + "epoch": 4.83, + "learning_rate": 1.4923989542439159e-05, + "loss": 0.0967, "step": 748 }, { - "epoch": 3.8, - "learning_rate": 0.00014811996449841098, - "loss": 0.0245, + "epoch": 4.83, + "learning_rate": 1.4811996449841098e-05, + "loss": 0.1224, "step": 749 }, { - "epoch": 3.81, - "learning_rate": 0.0001470035204197517, - "loss": 0.037, + "epoch": 4.84, + "learning_rate": 1.4700352041975168e-05, + "loss": 0.1215, "step": 750 }, { - "epoch": 3.81, - "learning_rate": 0.0001458905742514734, - "loss": 0.0521, + "epoch": 4.85, + "learning_rate": 1.458905742514734e-05, + "loss": 0.0928, "step": 751 }, { - "epoch": 3.82, - "learning_rate": 0.0001447811370219757, - "loss": 0.0429, + "epoch": 4.85, + "learning_rate": 1.447811370219757e-05, + "loss": 0.1678, "step": 752 }, { - "epoch": 3.82, - "learning_rate": 0.0001436752197248861, - "loss": 0.0589, + "epoch": 4.86, + "learning_rate": 1.4367521972488612e-05, + "loss": 0.1032, "step": 753 }, { - "epoch": 3.83, - "learning_rate": 0.00014257283331895317, - "loss": 0.0922, + "epoch": 4.86, + "learning_rate": 1.4257283331895315e-05, + "loss": 0.1344, "step": 754 }, { - "epoch": 3.83, - "learning_rate": 0.00014147398872793693, - "loss": 0.0742, + "epoch": 4.87, + "learning_rate": 1.4147398872793693e-05, + "loss": 0.0932, "step": 755 }, { - "epoch": 3.84, - "learning_rate": 0.00014037869684050113, - "loss": 0.0461, + "epoch": 4.88, + "learning_rate": 1.4037869684050115e-05, + "loss": 0.104, "step": 756 }, { - "epoch": 3.84, - "learning_rate": 0.0001392869685101044, - "loss": 0.0312, + "epoch": 4.88, + "learning_rate": 1.3928696851010443e-05, + "loss": 0.1034, "step": 757 }, { - "epoch": 3.85, - "learning_rate": 0.00013819881455489458, - "loss": 0.0173, + "epoch": 4.89, + "learning_rate": 1.3819881455489458e-05, + "loss": 0.1104, "step": 758 }, { - "epoch": 3.85, - "learning_rate": 0.00013711424575759911, - "loss": 0.0386, + "epoch": 4.9, + "learning_rate": 1.3711424575759912e-05, + "loss": 0.1111, "step": 759 }, { - "epoch": 3.86, - "learning_rate": 0.00013603327286542023, - "loss": 0.0292, + "epoch": 4.9, + "learning_rate": 1.3603327286542023e-05, + "loss": 0.1067, "step": 760 }, { - "epoch": 3.86, - "learning_rate": 0.00013495590658992718, - "loss": 0.0973, + "epoch": 4.91, + "learning_rate": 1.3495590658992718e-05, + "loss": 0.0965, "step": 761 }, { - "epoch": 3.87, - "learning_rate": 0.000133882157606951, - "loss": 0.0646, + "epoch": 4.92, + "learning_rate": 1.33882157606951e-05, + "loss": 0.0913, "step": 762 }, { - "epoch": 3.87, - "learning_rate": 0.00013281203655647755, - "loss": 0.1611, + "epoch": 4.92, + "learning_rate": 1.3281203655647756e-05, + "loss": 0.1056, "step": 763 }, { - "epoch": 3.88, - "learning_rate": 0.0001317455540425439, - "loss": 0.0415, + "epoch": 4.93, + "learning_rate": 1.317455540425439e-05, + "loss": 0.0977, "step": 764 }, { - "epoch": 3.88, - "learning_rate": 0.00013068272063313102, - "loss": 0.0521, + "epoch": 4.94, + "learning_rate": 1.3068272063313102e-05, + "loss": 0.1095, "step": 765 }, { - "epoch": 3.89, - "learning_rate": 0.00012962354686006083, - "loss": 0.035, + "epoch": 4.94, + "learning_rate": 1.2962354686006084e-05, + "loss": 0.1123, "step": 766 }, { - "epoch": 3.89, - "learning_rate": 0.00012856804321889116, - "loss": 0.0574, + "epoch": 4.95, + "learning_rate": 1.2856804321889115e-05, + "loss": 0.1233, "step": 767 }, { - "epoch": 3.9, - "learning_rate": 0.0001275162201688118, - "loss": 0.0436, + "epoch": 4.95, + "learning_rate": 1.2751622016881182e-05, + "loss": 0.1309, "step": 768 }, { - "epoch": 3.9, - "learning_rate": 0.00012646808813254035, - "loss": 0.0405, + "epoch": 4.96, + "learning_rate": 1.2646808813254035e-05, + "loss": 0.1012, "step": 769 }, { - "epoch": 3.91, - "learning_rate": 0.00012542365749622047, - "loss": 0.0526, + "epoch": 4.97, + "learning_rate": 1.2542365749622049e-05, + "loss": 0.1109, "step": 770 }, { - "epoch": 3.91, - "learning_rate": 0.00012438293860931676, - "loss": 0.0275, + "epoch": 4.97, + "learning_rate": 1.2438293860931677e-05, + "loss": 0.1016, "step": 771 }, { - "epoch": 3.92, - "learning_rate": 0.00012334594178451424, - "loss": 0.0757, + "epoch": 4.98, + "learning_rate": 1.2334594178451425e-05, + "loss": 0.1526, "step": 772 }, { - "epoch": 3.92, - "learning_rate": 0.00012231267729761485, - "loss": 0.0317, + "epoch": 4.99, + "learning_rate": 1.2231267729761487e-05, + "loss": 0.1835, "step": 773 }, { - "epoch": 3.93, - "learning_rate": 0.00012128315538743645, - "loss": 0.0306, + "epoch": 4.99, + "learning_rate": 1.2128315538743646e-05, + "loss": 0.0754, "step": 774 }, { - "epoch": 3.93, - "learning_rate": 0.00012025738625571026, - "loss": 0.0453, + "epoch": 5.0, + "learning_rate": 1.2025738625571026e-05, + "loss": 0.1137, "step": 775 }, { - "epoch": 3.94, - "learning_rate": 0.00011923538006698154, - "loss": 0.044, + "epoch": 5.01, + "learning_rate": 1.1923538006698154e-05, + "loss": 0.0515, "step": 776 }, { - "epoch": 3.94, - "learning_rate": 0.00011821714694850688, - "loss": 0.0719, + "epoch": 5.01, + "learning_rate": 1.1821714694850689e-05, + "loss": 0.0571, "step": 777 }, { - "epoch": 3.95, - "learning_rate": 0.00011720269699015528, - "loss": 0.056, + "epoch": 5.02, + "learning_rate": 1.172026969901553e-05, + "loss": 0.0598, "step": 778 }, { - "epoch": 3.95, - "learning_rate": 0.00011619204024430768, - "loss": 0.0287, + "epoch": 5.03, + "learning_rate": 1.161920402443077e-05, + "loss": 0.0467, "step": 779 }, { - "epoch": 3.96, - "learning_rate": 0.00011518518672575701, - "loss": 0.0672, + "epoch": 5.03, + "learning_rate": 1.1518518672575701e-05, + "loss": 0.0517, "step": 780 }, { - "epoch": 3.96, - "learning_rate": 0.00011418214641160956, - "loss": 0.0477, + "epoch": 5.04, + "learning_rate": 1.1418214641160958e-05, + "loss": 0.0907, "step": 781 }, { - "epoch": 3.97, - "learning_rate": 0.00011318292924118584, - "loss": 0.0515, + "epoch": 5.05, + "learning_rate": 1.1318292924118584e-05, + "loss": 0.0663, "step": 782 }, { - "epoch": 3.97, - "learning_rate": 0.0001121875451159221, - "loss": 0.0937, + "epoch": 5.05, + "learning_rate": 1.121875451159221e-05, + "loss": 0.0828, "step": 783 }, { - "epoch": 3.98, - "learning_rate": 0.00011119600389927181, - "loss": 0.0534, + "epoch": 5.06, + "learning_rate": 1.1119600389927182e-05, + "loss": 0.05, "step": 784 }, { - "epoch": 3.98, - "learning_rate": 0.00011020831541660914, - "loss": 0.0238, + "epoch": 5.06, + "learning_rate": 1.1020831541660915e-05, + "loss": 0.0606, "step": 785 }, { - "epoch": 3.99, - "learning_rate": 0.00010922448945512981, - "loss": 0.0686, + "epoch": 5.07, + "learning_rate": 1.092244894551298e-05, + "loss": 0.0505, "step": 786 }, { - "epoch": 3.99, - "learning_rate": 0.00010824453576375576, - "loss": 0.0209, + "epoch": 5.08, + "learning_rate": 1.0824453576375576e-05, + "loss": 0.0596, "step": 787 }, { - "epoch": 4.0, - "learning_rate": 0.00010726846405303752, - "loss": 0.066, + "epoch": 5.08, + "learning_rate": 1.0726846405303754e-05, + "loss": 0.0663, "step": 788 }, { - "epoch": 4.01, - "learning_rate": 0.0001062962839950587, - "loss": 0.0097, + "epoch": 5.09, + "learning_rate": 1.062962839950587e-05, + "loss": 0.053, "step": 789 }, { - "epoch": 4.01, - "learning_rate": 0.00010532800522333902, - "loss": 0.0172, + "epoch": 5.1, + "learning_rate": 1.0532800522333902e-05, + "loss": 0.0641, "step": 790 }, { - "epoch": 4.02, - "learning_rate": 0.00010436363733274057, - "loss": 0.0185, + "epoch": 5.1, + "learning_rate": 1.0436363733274057e-05, + "loss": 0.0951, "step": 791 }, { - "epoch": 4.02, - "learning_rate": 0.00010340318987937097, - "loss": 0.0164, + "epoch": 5.11, + "learning_rate": 1.0340318987937097e-05, + "loss": 0.0671, "step": 792 }, { - "epoch": 4.03, - "learning_rate": 0.00010244667238048988, - "loss": 0.0069, + "epoch": 5.12, + "learning_rate": 1.0244667238048988e-05, + "loss": 0.0516, "step": 793 }, { - "epoch": 4.03, - "learning_rate": 0.0001014940943144142, - "loss": 0.018, + "epoch": 5.12, + "learning_rate": 1.014940943144142e-05, + "loss": 0.0571, "step": 794 }, { - "epoch": 4.04, - "learning_rate": 0.00010054546512042422, - "loss": 0.0065, + "epoch": 5.13, + "learning_rate": 1.0054546512042424e-05, + "loss": 0.046, "step": 795 }, { - "epoch": 4.04, - "learning_rate": 9.960079419866985e-05, - "loss": 0.0177, + "epoch": 5.14, + "learning_rate": 9.960079419866985e-06, + "loss": 0.0595, "step": 796 }, { - "epoch": 4.05, - "learning_rate": 9.866009091007833e-05, - "loss": 0.0105, + "epoch": 5.14, + "learning_rate": 9.866009091007833e-06, + "loss": 0.0451, "step": 797 }, { - "epoch": 4.05, - "learning_rate": 9.772336457626013e-05, - "loss": 0.0127, + "epoch": 5.15, + "learning_rate": 9.772336457626014e-06, + "loss": 0.038, "step": 798 }, { - "epoch": 4.06, - "learning_rate": 9.679062447941778e-05, - "loss": 0.0111, + "epoch": 5.15, + "learning_rate": 9.679062447941778e-06, + "loss": 0.0682, "step": 799 }, { - "epoch": 4.06, - "learning_rate": 9.586187986225326e-05, - "loss": 0.0114, + "epoch": 5.16, + "learning_rate": 9.586187986225325e-06, + "loss": 0.0433, "step": 800 }, { - "epoch": 4.07, - "learning_rate": 9.493713992787673e-05, - "loss": 0.006, + "epoch": 5.17, + "learning_rate": 9.493713992787672e-06, + "loss": 0.0441, "step": 801 }, { - "epoch": 4.07, - "learning_rate": 9.401641383971476e-05, - "loss": 0.0041, + "epoch": 5.17, + "learning_rate": 9.401641383971477e-06, + "loss": 0.0865, "step": 802 }, { - "epoch": 4.08, - "learning_rate": 9.309971072142037e-05, - "loss": 0.0126, + "epoch": 5.18, + "learning_rate": 9.309971072142038e-06, + "loss": 0.0629, "step": 803 }, { - "epoch": 4.08, - "learning_rate": 9.218703965678204e-05, - "loss": 0.0107, + "epoch": 5.19, + "learning_rate": 9.218703965678204e-06, + "loss": 0.0558, "step": 804 }, { - "epoch": 4.09, - "learning_rate": 9.12784096896338e-05, - "loss": 0.0045, + "epoch": 5.19, + "learning_rate": 9.127840968963381e-06, + "loss": 0.0459, "step": 805 }, { - "epoch": 4.09, - "learning_rate": 9.03738298237658e-05, - "loss": 0.0039, + "epoch": 5.2, + "learning_rate": 9.03738298237658e-06, + "loss": 0.0567, "step": 806 }, { - "epoch": 4.1, - "learning_rate": 8.94733090228349e-05, - "loss": 0.0067, + "epoch": 5.21, + "learning_rate": 8.94733090228349e-06, + "loss": 0.0582, "step": 807 }, { - "epoch": 4.1, - "learning_rate": 8.857685621027567e-05, - "loss": 0.0114, + "epoch": 5.21, + "learning_rate": 8.857685621027568e-06, + "loss": 0.0811, "step": 808 }, { - "epoch": 4.11, - "learning_rate": 8.768448026921244e-05, - "loss": 0.0121, + "epoch": 5.22, + "learning_rate": 8.768448026921245e-06, + "loss": 0.0445, "step": 809 }, { - "epoch": 4.11, - "learning_rate": 8.679619004237111e-05, - "loss": 0.0055, + "epoch": 5.23, + "learning_rate": 8.67961900423711e-06, + "loss": 0.0542, "step": 810 }, { - "epoch": 4.12, - "learning_rate": 8.591199433199127e-05, - "loss": 0.0422, + "epoch": 5.23, + "learning_rate": 8.591199433199126e-06, + "loss": 0.049, "step": 811 }, { - "epoch": 4.12, - "learning_rate": 8.503190189973914e-05, - "loss": 0.0126, + "epoch": 5.24, + "learning_rate": 8.503190189973914e-06, + "loss": 0.0587, "step": 812 }, { - "epoch": 4.13, - "learning_rate": 8.415592146662105e-05, - "loss": 0.0071, + "epoch": 5.25, + "learning_rate": 8.415592146662104e-06, + "loss": 0.0647, "step": 813 }, { - "epoch": 4.13, - "learning_rate": 8.328406171289621e-05, - "loss": 0.0035, + "epoch": 5.25, + "learning_rate": 8.328406171289621e-06, + "loss": 0.0462, "step": 814 }, { - "epoch": 4.14, - "learning_rate": 8.24163312779917e-05, - "loss": 0.0216, + "epoch": 5.26, + "learning_rate": 8.24163312779917e-06, + "loss": 0.0817, "step": 815 }, { - "epoch": 4.14, - "learning_rate": 8.155273876041613e-05, - "loss": 0.0069, + "epoch": 5.26, + "learning_rate": 8.155273876041614e-06, + "loss": 0.0621, "step": 816 }, { - "epoch": 4.15, - "learning_rate": 8.069329271767484e-05, - "loss": 0.0219, + "epoch": 5.27, + "learning_rate": 8.069329271767484e-06, + "loss": 0.0792, "step": 817 }, { - "epoch": 4.15, - "learning_rate": 7.983800166618483e-05, - "loss": 0.0105, + "epoch": 5.28, + "learning_rate": 7.983800166618482e-06, + "loss": 0.2328, "step": 818 }, { - "epoch": 4.16, - "learning_rate": 7.898687408119065e-05, - "loss": 0.0114, + "epoch": 5.28, + "learning_rate": 7.898687408119065e-06, + "loss": 0.0794, "step": 819 }, { - "epoch": 4.16, - "learning_rate": 7.813991839667995e-05, - "loss": 0.0269, + "epoch": 5.29, + "learning_rate": 7.813991839667995e-06, + "loss": 0.0715, "step": 820 }, { - "epoch": 4.17, - "learning_rate": 7.72971430053005e-05, - "loss": 0.0235, + "epoch": 5.3, + "learning_rate": 7.72971430053005e-06, + "loss": 0.0616, "step": 821 }, { - "epoch": 4.17, - "learning_rate": 7.645855625827657e-05, - "loss": 0.0289, + "epoch": 5.3, + "learning_rate": 7.645855625827658e-06, + "loss": 0.0546, "step": 822 }, { - "epoch": 4.18, - "learning_rate": 7.562416646532661e-05, - "loss": 0.0042, + "epoch": 5.31, + "learning_rate": 7.56241664653266e-06, + "loss": 0.0482, "step": 823 }, { - "epoch": 4.18, - "learning_rate": 7.479398189458003e-05, - "loss": 0.0193, + "epoch": 5.32, + "learning_rate": 7.4793981894580034e-06, + "loss": 0.0669, "step": 824 }, { - "epoch": 4.19, - "learning_rate": 7.396801077249677e-05, - "loss": 0.024, + "epoch": 5.32, + "learning_rate": 7.396801077249676e-06, + "loss": 0.0647, "step": 825 }, { - "epoch": 4.19, - "learning_rate": 7.31462612837841e-05, - "loss": 0.006, + "epoch": 5.33, + "learning_rate": 7.3146261283784104e-06, + "loss": 0.0585, "step": 826 }, { - "epoch": 4.2, - "learning_rate": 7.232874157131669e-05, - "loss": 0.0372, + "epoch": 5.34, + "learning_rate": 7.2328741571316696e-06, + "loss": 0.0887, "step": 827 }, { - "epoch": 4.2, - "learning_rate": 7.15154597360555e-05, - "loss": 0.0196, + "epoch": 5.34, + "learning_rate": 7.1515459736055505e-06, + "loss": 0.0517, "step": 828 }, { - "epoch": 4.21, - "learning_rate": 7.070642383696762e-05, - "loss": 0.0154, + "epoch": 5.35, + "learning_rate": 7.070642383696763e-06, + "loss": 0.0747, "step": 829 }, { - "epoch": 4.21, - "learning_rate": 6.99016418909459e-05, - "loss": 0.0152, + "epoch": 5.35, + "learning_rate": 6.990164189094589e-06, + "loss": 0.0676, "step": 830 }, { - "epoch": 4.22, - "learning_rate": 6.910112187273066e-05, - "loss": 0.0112, + "epoch": 5.36, + "learning_rate": 6.910112187273066e-06, + "loss": 0.0648, "step": 831 }, { - "epoch": 4.22, - "learning_rate": 6.830487171482935e-05, - "loss": 0.0072, + "epoch": 5.37, + "learning_rate": 6.830487171482935e-06, + "loss": 0.052, "step": 832 }, { - "epoch": 4.23, - "learning_rate": 6.75128993074388e-05, - "loss": 0.0038, + "epoch": 5.37, + "learning_rate": 6.751289930743882e-06, + "loss": 0.0587, "step": 833 }, { - "epoch": 4.23, - "learning_rate": 6.672521249836688e-05, - "loss": 0.0065, + "epoch": 5.38, + "learning_rate": 6.6725212498366885e-06, + "loss": 0.1961, "step": 834 }, { - "epoch": 4.24, - "learning_rate": 6.594181909295427e-05, - "loss": 0.0224, + "epoch": 5.39, + "learning_rate": 6.594181909295427e-06, + "loss": 0.096, "step": 835 }, { - "epoch": 4.24, - "learning_rate": 6.516272685399794e-05, - "loss": 0.0035, + "epoch": 5.39, + "learning_rate": 6.516272685399793e-06, + "loss": 0.1016, "step": 836 }, { - "epoch": 4.25, - "learning_rate": 6.438794350167337e-05, - "loss": 0.0209, + "epoch": 5.4, + "learning_rate": 6.438794350167337e-06, + "loss": 0.0557, "step": 837 }, { - "epoch": 4.25, - "learning_rate": 6.36174767134588e-05, - "loss": 0.0052, + "epoch": 5.41, + "learning_rate": 6.36174767134588e-06, + "loss": 0.0592, "step": 838 }, { - "epoch": 4.26, - "learning_rate": 6.285133412405858e-05, - "loss": 0.013, + "epoch": 5.41, + "learning_rate": 6.285133412405858e-06, + "loss": 0.0739, "step": 839 }, { - "epoch": 4.26, - "learning_rate": 6.208952332532785e-05, - "loss": 0.0097, + "epoch": 5.42, + "learning_rate": 6.208952332532786e-06, + "loss": 0.0507, "step": 840 }, { - "epoch": 4.27, - "learning_rate": 6.133205186619695e-05, - "loss": 0.0178, + "epoch": 5.43, + "learning_rate": 6.133205186619695e-06, + "loss": 0.0649, "step": 841 }, { - "epoch": 4.27, - "learning_rate": 6.057892725259717e-05, - "loss": 0.0089, + "epoch": 5.43, + "learning_rate": 6.057892725259717e-06, + "loss": 0.0798, "step": 842 }, { - "epoch": 4.28, - "learning_rate": 5.983015694738597e-05, - "loss": 0.0112, + "epoch": 5.44, + "learning_rate": 5.983015694738597e-06, + "loss": 0.0634, "step": 843 }, { - "epoch": 4.28, - "learning_rate": 5.908574837027309e-05, - "loss": 0.0057, + "epoch": 5.45, + "learning_rate": 5.908574837027309e-06, + "loss": 0.0654, "step": 844 }, { - "epoch": 4.29, - "learning_rate": 5.8345708897747095e-05, - "loss": 0.003, + "epoch": 5.45, + "learning_rate": 5.83457088977471e-06, + "loss": 0.0722, "step": 845 }, { - "epoch": 4.29, - "learning_rate": 5.761004586300234e-05, - "loss": 0.0048, + "epoch": 5.46, + "learning_rate": 5.761004586300234e-06, + "loss": 0.0656, "step": 846 }, { - "epoch": 4.3, - "learning_rate": 5.687876655586582e-05, - "loss": 0.0136, + "epoch": 5.46, + "learning_rate": 5.687876655586583e-06, + "loss": 0.0539, "step": 847 }, { - "epoch": 4.3, - "learning_rate": 5.615187822272583e-05, - "loss": 0.0277, + "epoch": 5.47, + "learning_rate": 5.615187822272583e-06, + "loss": 0.065, "step": 848 }, { - "epoch": 4.31, - "learning_rate": 5.5429388066459305e-05, - "loss": 0.0309, + "epoch": 5.48, + "learning_rate": 5.542938806645931e-06, + "loss": 0.0573, "step": 849 }, { - "epoch": 4.31, - "learning_rate": 5.4711303246361146e-05, - "loss": 0.0192, + "epoch": 5.48, + "learning_rate": 5.4711303246361144e-06, + "loss": 0.0561, "step": 850 }, { - "epoch": 4.32, - "learning_rate": 5.3997630878072356e-05, - "loss": 0.0061, + "epoch": 5.49, + "learning_rate": 5.399763087807236e-06, + "loss": 0.0652, "step": 851 }, { - "epoch": 4.32, - "learning_rate": 5.328837803351083e-05, - "loss": 0.0031, + "epoch": 5.5, + "learning_rate": 5.328837803351083e-06, + "loss": 0.0604, "step": 852 }, { - "epoch": 4.33, - "learning_rate": 5.258355174079993e-05, - "loss": 0.0031, + "epoch": 5.5, + "learning_rate": 5.258355174079993e-06, + "loss": 0.06, "step": 853 }, { - "epoch": 4.34, - "learning_rate": 5.1883158984199704e-05, - "loss": 0.0063, + "epoch": 5.51, + "learning_rate": 5.188315898419971e-06, + "loss": 0.0785, "step": 854 }, { - "epoch": 4.34, - "learning_rate": 5.118720670403748e-05, - "loss": 0.0113, + "epoch": 5.52, + "learning_rate": 5.118720670403748e-06, + "loss": 0.0748, "step": 855 }, { - "epoch": 4.35, - "learning_rate": 5.0495701796639096e-05, - "loss": 0.0044, + "epoch": 5.52, + "learning_rate": 5.04957017966391e-06, + "loss": 0.048, "step": 856 }, { - "epoch": 4.35, - "learning_rate": 4.980865111426003e-05, - "loss": 0.0022, + "epoch": 5.53, + "learning_rate": 4.980865111426003e-06, + "loss": 0.0843, "step": 857 }, { - "epoch": 4.36, - "learning_rate": 4.9126061465018856e-05, - "loss": 0.0045, + "epoch": 5.54, + "learning_rate": 4.912606146501886e-06, + "loss": 0.042, "step": 858 }, { - "epoch": 4.36, - "learning_rate": 4.844793961282812e-05, - "loss": 0.0099, + "epoch": 5.54, + "learning_rate": 4.844793961282812e-06, + "loss": 0.0764, "step": 859 }, { - "epoch": 4.37, - "learning_rate": 4.777429227732843e-05, - "loss": 0.0075, + "epoch": 5.55, + "learning_rate": 4.777429227732844e-06, + "loss": 0.0874, "step": 860 }, { - "epoch": 4.37, - "learning_rate": 4.710512613382151e-05, - "loss": 0.014, + "epoch": 5.55, + "learning_rate": 4.710512613382151e-06, + "loss": 0.0941, "step": 861 }, { - "epoch": 4.38, - "learning_rate": 4.6440447813204225e-05, - "loss": 0.005, + "epoch": 5.56, + "learning_rate": 4.644044781320422e-06, + "loss": 0.14, "step": 862 }, { - "epoch": 4.38, - "learning_rate": 4.578026390190232e-05, - "loss": 0.0102, + "epoch": 5.57, + "learning_rate": 4.578026390190232e-06, + "loss": 0.0905, "step": 863 }, { - "epoch": 4.39, - "learning_rate": 4.512458094180616e-05, - "loss": 0.0027, + "epoch": 5.57, + "learning_rate": 4.5124580941806165e-06, + "loss": 0.068, "step": 864 }, { - "epoch": 4.39, - "learning_rate": 4.447340543020473e-05, - "loss": 0.0066, + "epoch": 5.58, + "learning_rate": 4.447340543020473e-06, + "loss": 0.057, "step": 865 }, { - "epoch": 4.4, - "learning_rate": 4.3826743819722236e-05, - "loss": 0.0089, + "epoch": 5.59, + "learning_rate": 4.382674381972224e-06, + "loss": 0.0571, "step": 866 }, { - "epoch": 4.4, - "learning_rate": 4.318460251825357e-05, - "loss": 0.008, + "epoch": 5.59, + "learning_rate": 4.318460251825357e-06, + "loss": 0.0642, "step": 867 }, { - "epoch": 4.41, - "learning_rate": 4.2546987888901266e-05, - "loss": 0.0097, + "epoch": 5.6, + "learning_rate": 4.254698788890127e-06, + "loss": 0.0557, "step": 868 }, { - "epoch": 4.41, - "learning_rate": 4.191390624991159e-05, - "loss": 0.0037, + "epoch": 5.61, + "learning_rate": 4.191390624991159e-06, + "loss": 0.0752, "step": 869 }, { - "epoch": 4.42, - "learning_rate": 4.12853638746134e-05, - "loss": 0.0416, + "epoch": 5.61, + "learning_rate": 4.12853638746134e-06, + "loss": 0.1361, "step": 870 }, { - "epoch": 4.42, - "learning_rate": 4.066136699135436e-05, - "loss": 0.0102, + "epoch": 5.62, + "learning_rate": 4.0661366991354365e-06, + "loss": 0.048, "step": 871 }, { - "epoch": 4.43, - "learning_rate": 4.004192178344029e-05, - "loss": 0.0159, + "epoch": 5.63, + "learning_rate": 4.004192178344029e-06, + "loss": 0.0381, "step": 872 }, { - "epoch": 4.43, - "learning_rate": 3.942703438907358e-05, - "loss": 0.0045, + "epoch": 5.63, + "learning_rate": 3.942703438907358e-06, + "loss": 0.0516, "step": 873 }, { - "epoch": 4.44, - "learning_rate": 3.881671090129246e-05, - "loss": 0.0112, + "epoch": 5.64, + "learning_rate": 3.881671090129247e-06, + "loss": 0.07, "step": 874 }, { - "epoch": 4.44, - "learning_rate": 3.821095736791008e-05, - "loss": 0.0061, + "epoch": 5.65, + "learning_rate": 3.821095736791008e-06, + "loss": 0.058, "step": 875 }, { - "epoch": 4.45, - "learning_rate": 3.7609779791455744e-05, - "loss": 0.0139, + "epoch": 5.65, + "learning_rate": 3.7609779791455744e-06, + "loss": 0.065, "step": 876 }, { - "epoch": 4.45, - "learning_rate": 3.7013184129113974e-05, - "loss": 0.0227, + "epoch": 5.66, + "learning_rate": 3.7013184129113976e-06, + "loss": 0.0421, "step": 877 }, { - "epoch": 4.46, - "learning_rate": 3.642117629266678e-05, - "loss": 0.0184, + "epoch": 5.66, + "learning_rate": 3.6421176292666783e-06, + "loss": 0.0721, "step": 878 }, { - "epoch": 4.46, - "learning_rate": 3.58337621484342e-05, - "loss": 0.0063, + "epoch": 5.67, + "learning_rate": 3.58337621484342e-06, + "loss": 0.0599, "step": 879 }, { - "epoch": 4.47, - "learning_rate": 3.5250947517216634e-05, - "loss": 0.0095, + "epoch": 5.68, + "learning_rate": 3.5250947517216637e-06, + "loss": 0.0571, "step": 880 }, { - "epoch": 4.47, - "learning_rate": 3.467273817423688e-05, - "loss": 0.0451, + "epoch": 5.68, + "learning_rate": 3.4672738174236884e-06, + "loss": 0.0589, "step": 881 }, { - "epoch": 4.48, - "learning_rate": 3.40991398490833e-05, - "loss": 0.0086, + "epoch": 5.69, + "learning_rate": 3.4099139849083307e-06, + "loss": 0.0922, "step": 882 }, { - "epoch": 4.48, - "learning_rate": 3.353015822565253e-05, - "loss": 0.0197, + "epoch": 5.7, + "learning_rate": 3.353015822565253e-06, + "loss": 0.0413, "step": 883 }, { - "epoch": 4.49, - "learning_rate": 3.296579894209345e-05, - "loss": 0.0232, + "epoch": 5.7, + "learning_rate": 3.296579894209345e-06, + "loss": 0.0732, "step": 884 }, { - "epoch": 4.49, - "learning_rate": 3.240606759075143e-05, - "loss": 0.004, + "epoch": 5.71, + "learning_rate": 3.2406067590751433e-06, + "loss": 0.051, "step": 885 }, { - "epoch": 4.5, - "learning_rate": 3.185096971811274e-05, - "loss": 0.0071, + "epoch": 5.72, + "learning_rate": 3.1850969718112745e-06, + "loss": 0.058, "step": 886 }, { - "epoch": 4.5, - "learning_rate": 3.1300510824749276e-05, - "loss": 0.0077, + "epoch": 5.72, + "learning_rate": 3.1300510824749273e-06, + "loss": 0.0939, "step": 887 }, { - "epoch": 4.51, - "learning_rate": 3.0754696365265064e-05, - "loss": 0.0156, + "epoch": 5.73, + "learning_rate": 3.0754696365265068e-06, + "loss": 0.0825, "step": 888 }, { - "epoch": 4.51, - "learning_rate": 3.0213531748240763e-05, - "loss": 0.0067, + "epoch": 5.74, + "learning_rate": 3.0213531748240764e-06, + "loss": 0.0542, "step": 889 }, { - "epoch": 4.52, - "learning_rate": 2.9677022336181414e-05, - "loss": 0.0304, + "epoch": 5.74, + "learning_rate": 2.9677022336181413e-06, + "loss": 0.057, "step": 890 }, { - "epoch": 4.52, - "learning_rate": 2.9145173445462582e-05, - "loss": 0.006, + "epoch": 5.75, + "learning_rate": 2.914517344546258e-06, + "loss": 0.0492, "step": 891 }, { - "epoch": 4.53, - "learning_rate": 2.8617990346277655e-05, - "loss": 0.0188, + "epoch": 5.75, + "learning_rate": 2.8617990346277657e-06, + "loss": 0.0601, "step": 892 }, { - "epoch": 4.53, - "learning_rate": 2.8095478262585904e-05, - "loss": 0.0145, + "epoch": 5.76, + "learning_rate": 2.8095478262585907e-06, + "loss": 0.067, "step": 893 }, { - "epoch": 4.54, - "learning_rate": 2.7577642372060676e-05, - "loss": 0.0076, + "epoch": 5.77, + "learning_rate": 2.7577642372060673e-06, + "loss": 0.0573, "step": 894 }, { - "epoch": 4.54, - "learning_rate": 2.7064487806037985e-05, - "loss": 0.0043, + "epoch": 5.77, + "learning_rate": 2.7064487806037985e-06, + "loss": 0.0573, "step": 895 }, { - "epoch": 4.55, - "learning_rate": 2.6556019649465524e-05, - "loss": 0.0237, + "epoch": 5.78, + "learning_rate": 2.6556019649465525e-06, + "loss": 0.1025, "step": 896 }, { - "epoch": 4.55, - "learning_rate": 2.605224294085279e-05, - "loss": 0.0117, + "epoch": 5.79, + "learning_rate": 2.6052242940852787e-06, + "loss": 0.0734, "step": 897 }, { - "epoch": 4.56, - "learning_rate": 2.5553162672220465e-05, - "loss": 0.0166, + "epoch": 5.79, + "learning_rate": 2.5553162672220465e-06, + "loss": 0.1048, "step": 898 }, { - "epoch": 4.56, - "learning_rate": 2.5058783789051463e-05, - "loss": 0.0019, + "epoch": 5.8, + "learning_rate": 2.5058783789051467e-06, + "loss": 0.0557, "step": 899 }, { - "epoch": 4.57, - "learning_rate": 2.45691111902418e-05, - "loss": 0.0276, + "epoch": 5.81, + "learning_rate": 2.45691111902418e-06, + "loss": 0.0607, "step": 900 }, { - "epoch": 4.57, - "learning_rate": 2.4084149728051953e-05, - "loss": 0.0064, + "epoch": 5.81, + "learning_rate": 2.4084149728051952e-06, + "loss": 0.0662, "step": 901 }, { - "epoch": 4.58, - "learning_rate": 2.3603904208058692e-05, - "loss": 0.0092, + "epoch": 5.82, + "learning_rate": 2.360390420805869e-06, + "loss": 0.0689, "step": 902 }, { - "epoch": 4.58, - "learning_rate": 2.3128379389108e-05, - "loss": 0.0086, + "epoch": 5.83, + "learning_rate": 2.3128379389108e-06, + "loss": 0.0692, "step": 903 }, { - "epoch": 4.59, - "learning_rate": 2.2657579983267064e-05, - "loss": 0.0065, + "epoch": 5.83, + "learning_rate": 2.2657579983267064e-06, + "loss": 0.0494, "step": 904 }, { - "epoch": 4.59, - "learning_rate": 2.219151065577829e-05, - "loss": 0.0017, + "epoch": 5.84, + "learning_rate": 2.219151065577829e-06, + "loss": 0.0668, "step": 905 }, { - "epoch": 4.6, - "learning_rate": 2.1730176025012817e-05, - "loss": 0.0046, + "epoch": 5.85, + "learning_rate": 2.1730176025012816e-06, + "loss": 0.067, "step": 906 }, { - "epoch": 4.6, - "learning_rate": 2.1273580662424797e-05, - "loss": 0.0067, + "epoch": 5.85, + "learning_rate": 2.1273580662424796e-06, + "loss": 0.0899, "step": 907 }, { - "epoch": 4.61, - "learning_rate": 2.082172909250568e-05, - "loss": 0.0064, + "epoch": 5.86, + "learning_rate": 2.082172909250568e-06, + "loss": 0.0672, "step": 908 }, { - "epoch": 4.61, - "learning_rate": 2.0374625792740464e-05, - "loss": 0.0069, + "epoch": 5.86, + "learning_rate": 2.0374625792740464e-06, + "loss": 0.0559, "step": 909 }, { - "epoch": 4.62, - "learning_rate": 1.993227519356189e-05, - "loss": 0.0041, + "epoch": 5.87, + "learning_rate": 1.993227519356189e-06, + "loss": 0.0682, "step": 910 }, { - "epoch": 4.62, - "learning_rate": 1.94946816783077e-05, - "loss": 0.0047, + "epoch": 5.88, + "learning_rate": 1.9494681678307703e-06, + "loss": 0.0748, "step": 911 }, { - "epoch": 4.63, - "learning_rate": 1.906184958317664e-05, - "loss": 0.0057, + "epoch": 5.88, + "learning_rate": 1.906184958317664e-06, + "loss": 0.0641, "step": 912 }, { - "epoch": 4.63, - "learning_rate": 1.863378319718578e-05, - "loss": 0.0069, + "epoch": 5.89, + "learning_rate": 1.8633783197185783e-06, + "loss": 0.0729, "step": 913 }, { - "epoch": 4.64, - "learning_rate": 1.82104867621275e-05, - "loss": 0.0217, + "epoch": 5.9, + "learning_rate": 1.8210486762127499e-06, + "loss": 0.0673, "step": 914 }, { - "epoch": 4.64, - "learning_rate": 1.7791964472528233e-05, - "loss": 0.0219, + "epoch": 5.9, + "learning_rate": 1.7791964472528232e-06, + "loss": 0.064, "step": 915 }, { - "epoch": 4.65, - "learning_rate": 1.737822047560611e-05, - "loss": 0.0068, + "epoch": 5.91, + "learning_rate": 1.737822047560611e-06, + "loss": 0.0555, "step": 916 }, { - "epoch": 4.65, - "learning_rate": 1.696925887123052e-05, - "loss": 0.0136, + "epoch": 5.92, + "learning_rate": 1.696925887123052e-06, + "loss": 0.0539, "step": 917 }, { - "epoch": 4.66, - "learning_rate": 1.656508371188109e-05, - "loss": 0.0102, + "epoch": 5.92, + "learning_rate": 1.656508371188109e-06, + "loss": 0.0565, "step": 918 }, { - "epoch": 4.66, - "learning_rate": 1.616569900260767e-05, - "loss": 0.0051, + "epoch": 5.93, + "learning_rate": 1.6165699002607671e-06, + "loss": 0.0623, "step": 919 }, { - "epoch": 4.67, - "learning_rate": 1.577110870099041e-05, - "loss": 0.0154, + "epoch": 5.94, + "learning_rate": 1.5771108700990412e-06, + "loss": 0.0788, "step": 920 }, { - "epoch": 4.68, - "learning_rate": 1.538131671710108e-05, - "loss": 0.0713, + "epoch": 5.94, + "learning_rate": 1.538131671710108e-06, + "loss": 0.0686, "step": 921 }, { - "epoch": 4.68, - "learning_rate": 1.4996326913463754e-05, - "loss": 0.0157, + "epoch": 5.95, + "learning_rate": 1.4996326913463754e-06, + "loss": 0.0602, "step": 922 }, { - "epoch": 4.69, - "learning_rate": 1.461614310501691e-05, - "loss": 0.019, + "epoch": 5.95, + "learning_rate": 1.461614310501691e-06, + "loss": 0.0658, "step": 923 }, { - "epoch": 4.69, - "learning_rate": 1.4240769059075342e-05, - "loss": 0.0123, + "epoch": 5.96, + "learning_rate": 1.4240769059075342e-06, + "loss": 0.0539, "step": 924 }, { - "epoch": 4.7, - "learning_rate": 1.387020849529319e-05, - "loss": 0.0032, + "epoch": 5.97, + "learning_rate": 1.387020849529319e-06, + "loss": 0.0487, "step": 925 }, { - "epoch": 4.7, - "learning_rate": 1.3504465085626639e-05, - "loss": 0.0044, + "epoch": 5.97, + "learning_rate": 1.3504465085626638e-06, + "loss": 0.0557, "step": 926 }, { - "epoch": 4.71, - "learning_rate": 1.3143542454297885e-05, - "loss": 0.0061, + "epoch": 5.98, + "learning_rate": 1.3143542454297885e-06, + "loss": 0.0489, "step": 927 }, { - "epoch": 4.71, - "learning_rate": 1.2787444177759067e-05, - "loss": 0.0145, + "epoch": 5.99, + "learning_rate": 1.2787444177759068e-06, + "loss": 0.0567, "step": 928 }, { - "epoch": 4.72, - "learning_rate": 1.243617378465689e-05, - "loss": 0.0038, + "epoch": 5.99, + "learning_rate": 1.243617378465689e-06, + "loss": 0.0802, "step": 929 }, { - "epoch": 4.72, - "learning_rate": 1.208973475579761e-05, - "loss": 0.0049, + "epoch": 6.0, + "learning_rate": 1.208973475579761e-06, + "loss": 0.038, "step": 930 }, { - "epoch": 4.73, - "learning_rate": 1.1748130524112665e-05, - "loss": 0.0022, + "epoch": 6.01, + "learning_rate": 1.1748130524112666e-06, + "loss": 0.0769, "step": 931 }, { - "epoch": 4.73, - "learning_rate": 1.1411364474624264e-05, - "loss": 0.0133, + "epoch": 6.01, + "learning_rate": 1.1411364474624264e-06, + "loss": 0.0503, "step": 932 }, { - "epoch": 4.74, - "learning_rate": 1.1079439944412406e-05, - "loss": 0.0191, + "epoch": 6.02, + "learning_rate": 1.1079439944412406e-06, + "loss": 0.0544, "step": 933 }, { - "epoch": 4.74, - "learning_rate": 1.0752360222581469e-05, - "loss": 0.0045, + "epoch": 6.03, + "learning_rate": 1.075236022258147e-06, + "loss": 0.0557, "step": 934 }, { - "epoch": 4.75, - "learning_rate": 1.0430128550227624e-05, - "loss": 0.008, + "epoch": 6.03, + "learning_rate": 1.0430128550227625e-06, + "loss": 0.0401, "step": 935 }, { - "epoch": 4.75, - "learning_rate": 1.0112748120406857e-05, - "loss": 0.0066, + "epoch": 6.04, + "learning_rate": 1.0112748120406856e-06, + "loss": 0.0525, "step": 936 }, { - "epoch": 4.76, - "learning_rate": 9.800222078103272e-06, - "loss": 0.0041, + "epoch": 6.05, + "learning_rate": 9.800222078103271e-07, + "loss": 0.1297, "step": 937 }, { - "epoch": 4.76, - "learning_rate": 9.492553520197733e-06, - "loss": 0.0113, + "epoch": 6.05, + "learning_rate": 9.492553520197733e-07, + "loss": 0.0372, "step": 938 }, { - "epoch": 4.77, - "learning_rate": 9.189745495437607e-06, - "loss": 0.0131, + "epoch": 6.06, + "learning_rate": 9.189745495437608e-07, + "loss": 0.0314, "step": 939 }, { - "epoch": 4.77, - "learning_rate": 8.89180100440612e-06, - "loss": 0.0025, + "epoch": 6.06, + "learning_rate": 8.891801004406119e-07, + "loss": 0.0446, "step": 940 }, { - "epoch": 4.78, - "learning_rate": 8.59872299949288e-06, - "loss": 0.0092, + "epoch": 6.07, + "learning_rate": 8.59872299949288e-07, + "loss": 0.0515, "step": 941 }, { - "epoch": 4.78, - "learning_rate": 8.31051438486441e-06, - "loss": 0.0018, + "epoch": 6.08, + "learning_rate": 8.31051438486441e-07, + "loss": 0.0603, "step": 942 }, { - "epoch": 4.79, - "learning_rate": 8.027178016435766e-06, - "loss": 0.0059, + "epoch": 6.08, + "learning_rate": 8.027178016435765e-07, + "loss": 0.0486, "step": 943 }, { - "epoch": 4.79, - "learning_rate": 7.748716701841684e-06, - "loss": 0.018, + "epoch": 6.09, + "learning_rate": 7.748716701841685e-07, + "loss": 0.0474, "step": 944 }, { - "epoch": 4.8, - "learning_rate": 7.475133200409212e-06, - "loss": 0.0161, + "epoch": 6.1, + "learning_rate": 7.475133200409212e-07, + "loss": 0.0592, "step": 945 }, { - "epoch": 4.8, - "learning_rate": 7.206430223130278e-06, - "loss": 0.017, + "epoch": 6.1, + "learning_rate": 7.206430223130278e-07, + "loss": 0.2176, "step": 946 }, { - "epoch": 4.81, - "learning_rate": 6.9426104326345e-06, - "loss": 0.003, + "epoch": 6.11, + "learning_rate": 6.9426104326345e-07, + "loss": 0.0607, "step": 947 }, { - "epoch": 4.81, - "learning_rate": 6.683676443163312e-06, - "loss": 0.0087, + "epoch": 6.12, + "learning_rate": 6.683676443163311e-07, + "loss": 0.0523, "step": 948 }, { - "epoch": 4.82, - "learning_rate": 6.429630820543597e-06, - "loss": 0.0113, + "epoch": 6.12, + "learning_rate": 6.429630820543598e-07, + "loss": 0.0521, "step": 949 }, { - "epoch": 4.82, - "learning_rate": 6.180476082162656e-06, - "loss": 0.0156, + "epoch": 6.13, + "learning_rate": 6.180476082162656e-07, + "loss": 0.0504, "step": 950 }, { - "epoch": 4.83, - "learning_rate": 5.936214696942887e-06, - "loss": 0.005, + "epoch": 6.14, + "learning_rate": 5.936214696942887e-07, + "loss": 0.0565, "step": 951 }, { - "epoch": 4.83, - "learning_rate": 5.696849085317646e-06, - "loss": 0.0041, + "epoch": 6.14, + "learning_rate": 5.696849085317646e-07, + "loss": 0.0505, "step": 952 }, { - "epoch": 4.84, - "learning_rate": 5.462381619207091e-06, - "loss": 0.0068, + "epoch": 6.15, + "learning_rate": 5.462381619207091e-07, + "loss": 0.0641, "step": 953 }, { - "epoch": 4.84, - "learning_rate": 5.232814621994597e-06, - "loss": 0.0027, + "epoch": 6.15, + "learning_rate": 5.232814621994598e-07, + "loss": 0.0585, "step": 954 }, { - "epoch": 4.85, - "learning_rate": 5.008150368503994e-06, - "loss": 0.0037, + "epoch": 6.16, + "learning_rate": 5.008150368503994e-07, + "loss": 0.0443, "step": 955 }, { - "epoch": 4.85, - "learning_rate": 4.788391084976862e-06, - "loss": 0.0125, + "epoch": 6.17, + "learning_rate": 4.788391084976862e-07, + "loss": 0.0452, "step": 956 }, { - "epoch": 4.86, - "learning_rate": 4.5735389490503265e-06, - "loss": 0.0126, + "epoch": 6.17, + "learning_rate": 4.573538949050327e-07, + "loss": 0.0449, "step": 957 }, { - "epoch": 4.86, - "learning_rate": 4.36359608973591e-06, - "loss": 0.0088, + "epoch": 6.18, + "learning_rate": 4.363596089735911e-07, + "loss": 0.0738, "step": 958 }, { - "epoch": 4.87, - "learning_rate": 4.158564587397828e-06, - "loss": 0.0032, + "epoch": 6.19, + "learning_rate": 4.1585645873978284e-07, + "loss": 0.0449, "step": 959 }, { - "epoch": 4.87, - "learning_rate": 3.9584464737330016e-06, - "loss": 0.0066, + "epoch": 6.19, + "learning_rate": 3.958446473733002e-07, + "loss": 0.0445, "step": 960 }, { - "epoch": 4.88, - "learning_rate": 3.763243731750521e-06, - "loss": 0.0061, + "epoch": 6.2, + "learning_rate": 3.7632437317505207e-07, + "loss": 0.0343, "step": 961 }, { - "epoch": 4.88, - "learning_rate": 3.572958295752049e-06, - "loss": 0.0032, + "epoch": 6.21, + "learning_rate": 3.572958295752049e-07, + "loss": 0.0462, "step": 962 }, { - "epoch": 4.89, - "learning_rate": 3.387592051312782e-06, - "loss": 0.0064, + "epoch": 6.21, + "learning_rate": 3.387592051312782e-07, + "loss": 0.0579, "step": 963 }, { - "epoch": 4.89, - "learning_rate": 3.207146835262742e-06, - "loss": 0.0031, + "epoch": 6.22, + "learning_rate": 3.207146835262742e-07, + "loss": 0.0615, "step": 964 }, { - "epoch": 4.9, - "learning_rate": 3.031624435668345e-06, - "loss": 0.0152, + "epoch": 6.23, + "learning_rate": 3.0316244356683454e-07, + "loss": 0.0409, "step": 965 }, { - "epoch": 4.9, - "learning_rate": 2.861026591815141e-06, - "loss": 0.0139, + "epoch": 6.23, + "learning_rate": 2.8610265918151414e-07, + "loss": 0.0358, "step": 966 }, { - "epoch": 4.91, - "learning_rate": 2.6953549941900467e-06, - "loss": 0.0061, + "epoch": 6.24, + "learning_rate": 2.695354994190047e-07, + "loss": 0.0412, "step": 967 }, { - "epoch": 4.91, - "learning_rate": 2.5346112844650826e-06, - "loss": 0.0033, + "epoch": 6.25, + "learning_rate": 2.534611284465083e-07, + "loss": 0.0611, "step": 968 }, { - "epoch": 4.92, - "learning_rate": 2.378797055480608e-06, - "loss": 0.0026, + "epoch": 6.25, + "learning_rate": 2.3787970554806084e-07, + "loss": 0.0416, "step": 969 }, { - "epoch": 4.92, - "learning_rate": 2.227913851230057e-06, - "loss": 0.0067, + "epoch": 6.26, + "learning_rate": 2.2279138512300567e-07, + "loss": 0.0458, "step": 970 }, { - "epoch": 4.93, - "learning_rate": 2.081963166844225e-06, - "loss": 0.0045, + "epoch": 6.26, + "learning_rate": 2.0819631668442253e-07, + "loss": 0.0507, "step": 971 }, { - "epoch": 4.93, - "learning_rate": 1.9409464485766747e-06, - "loss": 0.0126, + "epoch": 6.27, + "learning_rate": 1.940946448576675e-07, + "loss": 0.0527, "step": 972 }, { - "epoch": 4.94, - "learning_rate": 1.804865093789354e-06, - "loss": 0.0036, + "epoch": 6.28, + "learning_rate": 1.8048650937893542e-07, + "loss": 0.0879, "step": 973 }, { - "epoch": 4.94, - "learning_rate": 1.6737204509387205e-06, - "loss": 0.0019, + "epoch": 6.28, + "learning_rate": 1.6737204509387206e-07, + "loss": 0.039, "step": 974 }, { - "epoch": 4.95, - "learning_rate": 1.5475138195623628e-06, - "loss": 0.0085, + "epoch": 6.29, + "learning_rate": 1.5475138195623629e-07, + "loss": 0.0455, "step": 975 }, { - "epoch": 4.95, - "learning_rate": 1.4262464502663442e-06, - "loss": 0.0063, + "epoch": 6.3, + "learning_rate": 1.4262464502663443e-07, + "loss": 0.0433, "step": 976 }, { - "epoch": 4.96, - "learning_rate": 1.309919544712268e-06, - "loss": 0.0052, + "epoch": 6.3, + "learning_rate": 1.309919544712268e-07, + "loss": 0.1652, "step": 977 }, { - "epoch": 4.96, - "learning_rate": 1.1985342556060653e-06, - "loss": 0.0166, + "epoch": 6.31, + "learning_rate": 1.1985342556060652e-07, + "loss": 0.0449, "step": 978 }, { - "epoch": 4.97, - "learning_rate": 1.092091686686114e-06, - "loss": 0.0158, + "epoch": 6.32, + "learning_rate": 1.0920916866861142e-07, + "loss": 0.0421, "step": 979 }, { - "epoch": 4.97, - "learning_rate": 9.905928927123608e-07, - "loss": 0.0166, + "epoch": 6.32, + "learning_rate": 9.905928927123609e-08, + "loss": 0.0377, "step": 980 }, { - "epoch": 4.98, - "learning_rate": 8.94038879455994e-07, - "loss": 0.0074, + "epoch": 6.33, + "learning_rate": 8.940388794559939e-08, + "loss": 0.0516, "step": 981 }, { - "epoch": 4.98, - "learning_rate": 8.024306036893969e-07, - "loss": 0.0084, + "epoch": 6.34, + "learning_rate": 8.02430603689397e-08, + "loss": 0.0473, "step": 982 }, { - "epoch": 4.99, - "learning_rate": 7.157689731767669e-07, - "loss": 0.0071, + "epoch": 6.34, + "learning_rate": 7.157689731767669e-08, + "loss": 0.0416, "step": 983 }, { - "epoch": 4.99, - "learning_rate": 6.340548466648444e-07, - "loss": 0.0031, + "epoch": 6.35, + "learning_rate": 6.340548466648443e-08, + "loss": 0.038, "step": 984 }, { - "epoch": 5.0, - "learning_rate": 5.572890338748082e-07, - "loss": 0.0131, + "epoch": 6.35, + "learning_rate": 5.572890338748082e-08, + "loss": 0.0422, "step": 985 }, { - "epoch": 5.01, - "learning_rate": 4.854722954938385e-07, - "loss": 0.0016, + "epoch": 6.36, + "learning_rate": 4.8547229549383844e-08, + "loss": 0.0644, "step": 986 }, { - "epoch": 5.01, - "learning_rate": 4.1860534316801037e-07, - "loss": 0.0054, + "epoch": 6.37, + "learning_rate": 4.186053431680104e-08, + "loss": 0.0417, "step": 987 }, { - "epoch": 5.02, - "learning_rate": 3.566888394948009e-07, - "loss": 0.0064, + "epoch": 6.37, + "learning_rate": 3.566888394948009e-08, + "loss": 0.0655, "step": 988 }, { - "epoch": 5.02, - "learning_rate": 2.997233980168157e-07, - "loss": 0.0024, + "epoch": 6.38, + "learning_rate": 2.997233980168157e-08, + "loss": 0.0482, "step": 989 }, { - "epoch": 5.03, - "learning_rate": 2.477095832156828e-07, - "loss": 0.004, + "epoch": 6.39, + "learning_rate": 2.4770958321568283e-08, + "loss": 0.0427, "step": 990 }, { - "epoch": 5.03, - "learning_rate": 2.0064791050633523e-07, - "loss": 0.0041, + "epoch": 6.39, + "learning_rate": 2.0064791050633526e-08, + "loss": 0.0622, "step": 991 }, { - "epoch": 5.04, - "learning_rate": 1.5853884623195924e-07, - "loss": 0.0206, + "epoch": 6.4, + "learning_rate": 1.5853884623195925e-08, + "loss": 0.0509, "step": 992 }, { - "epoch": 5.04, - "learning_rate": 1.2138280765944253e-07, - "loss": 0.002, + "epoch": 6.41, + "learning_rate": 1.2138280765944254e-08, + "loss": 0.0523, "step": 993 }, { - "epoch": 5.05, - "learning_rate": 8.918016297515542e-08, - "loss": 0.005, + "epoch": 6.41, + "learning_rate": 8.918016297515541e-09, + "loss": 0.0627, "step": 994 }, { - "epoch": 5.05, - "learning_rate": 6.193123128134249e-08, - "loss": 0.0032, + "epoch": 6.42, + "learning_rate": 6.193123128134248e-09, + "loss": 0.0431, "step": 995 }, { - "epoch": 5.06, - "learning_rate": 3.963628259290308e-08, - "loss": 0.0051, + "epoch": 6.43, + "learning_rate": 3.963628259290308e-09, + "loss": 0.043, "step": 996 }, { - "epoch": 5.06, - "learning_rate": 2.229553783478222e-08, - "loss": 0.0031, + "epoch": 6.43, + "learning_rate": 2.229553783478222e-09, + "loss": 0.0358, "step": 997 }, { - "epoch": 5.07, - "learning_rate": 9.90916883986115e-09, - "loss": 0.0042, + "epoch": 6.44, + "learning_rate": 9.90916883986115e-10, + "loss": 0.0868, "step": 998 }, { - "epoch": 5.07, - "learning_rate": 2.4772983469589784e-09, - "loss": 0.0022, + "epoch": 6.45, + "learning_rate": 2.477298346958978e-10, + "loss": 0.0718, "step": 999 }, { - "epoch": 5.08, + "epoch": 6.45, "learning_rate": 0.0, - "loss": 0.0035, + "loss": 0.0503, "step": 1000 }, { - "epoch": 5.08, + "epoch": 6.45, "step": 1000, - "total_flos": 4.507227485768909e+16, - "train_loss": 0.2918202262979466, - "train_runtime": 2044.5236, - "train_samples_per_second": 1.956, - "train_steps_per_second": 0.489 + "total_flos": 9.524118332178432e+16, + "train_loss": 0.3133371548131108, + "train_runtime": 3521.0844, + "train_samples_per_second": 2.272, + "train_steps_per_second": 0.284 } ], "max_steps": 1000, - "num_train_epochs": 6, - "total_flos": 4.507227485768909e+16, + "num_train_epochs": 7, + "total_flos": 9.524118332178432e+16, "trial_name": null, "trial_params": null }