| { | |
| "best_metric": 0.8987888693809509, | |
| "best_model_checkpoint": "./zhko_mbartLarge_100p_run1/checkpoint-141839", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 283678, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.335965407031397e-06, | |
| "loss": 2.879, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6860312088738486e-06, | |
| "loss": 2.0241, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.0360970107162996e-06, | |
| "loss": 1.8819, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.386162812558751e-06, | |
| "loss": 1.8009, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.1736228614401204e-05, | |
| "loss": 1.7072, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.4086294416243657e-05, | |
| "loss": 1.682, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.643636021808611e-05, | |
| "loss": 1.6188, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.878642601992856e-05, | |
| "loss": 1.5802, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.1136491821771008e-05, | |
| "loss": 1.5547, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.348655762361346e-05, | |
| "loss": 1.5422, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.5831923293852228e-05, | |
| "loss": 1.5194, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.8181989095694684e-05, | |
| "loss": 1.4815, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.053205489753713e-05, | |
| "loss": 1.4509, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.288212069937958e-05, | |
| "loss": 1.458, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.5227486369618354e-05, | |
| "loss": 1.4415, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.757285203985712e-05, | |
| "loss": 1.4284, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.991821771009588e-05, | |
| "loss": 1.4297, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.2268283511938334e-05, | |
| "loss": 1.4307, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.461834931378079e-05, | |
| "loss": 1.3971, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.696371498401955e-05, | |
| "loss": 1.4106, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9313780785862004e-05, | |
| "loss": 1.3999, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.998319322943521e-05, | |
| "loss": 1.3937, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.995945485293127e-05, | |
| "loss": 1.3773, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9935716476427316e-05, | |
| "loss": 1.3621, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.991197809992337e-05, | |
| "loss": 1.3552, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.988823972341943e-05, | |
| "loss": 1.3312, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.986450134691549e-05, | |
| "loss": 1.3331, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.984081044716455e-05, | |
| "loss": 1.3259, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9817072070660604e-05, | |
| "loss": 1.2983, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.979333369415666e-05, | |
| "loss": 1.2828, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.976959531765271e-05, | |
| "loss": 1.2793, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9745856941148764e-05, | |
| "loss": 1.2968, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9722118564644826e-05, | |
| "loss": 1.2774, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.969838018814088e-05, | |
| "loss": 1.2507, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.967464181163694e-05, | |
| "loss": 1.2521, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.965090343513299e-05, | |
| "loss": 1.2337, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.962716505862904e-05, | |
| "loss": 1.2238, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.96034266821251e-05, | |
| "loss": 1.2111, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.957968830562115e-05, | |
| "loss": 1.2336, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.955599740587022e-05, | |
| "loss": 1.1975, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9532259029366274e-05, | |
| "loss": 1.2092, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.950852065286233e-05, | |
| "loss": 1.2079, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.948482975311139e-05, | |
| "loss": 1.1831, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.946109137660745e-05, | |
| "loss": 1.1965, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.94373530001035e-05, | |
| "loss": 1.1928, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9413614623599556e-05, | |
| "loss": 1.1779, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.938987624709561e-05, | |
| "loss": 1.1759, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.936613787059167e-05, | |
| "loss": 1.1871, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.934239949408772e-05, | |
| "loss": 1.1782, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.931866111758378e-05, | |
| "loss": 1.1619, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.9294922741079834e-05, | |
| "loss": 1.1615, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.927118436457589e-05, | |
| "loss": 1.1476, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.924749346482495e-05, | |
| "loss": 1.1507, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.922380256507401e-05, | |
| "loss": 1.1489, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.9200064188570064e-05, | |
| "loss": 1.144, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.917632581206613e-05, | |
| "loss": 1.1223, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.915258743556218e-05, | |
| "loss": 1.1319, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.912884905905824e-05, | |
| "loss": 1.1235, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.9105110682554286e-05, | |
| "loss": 1.1408, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.908141978280335e-05, | |
| "loss": 1.1299, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.905772888305242e-05, | |
| "loss": 1.1172, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.903399050654847e-05, | |
| "loss": 1.1316, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.9010299606797535e-05, | |
| "loss": 1.1154, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.8986561230293584e-05, | |
| "loss": 1.1162, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.896282285378964e-05, | |
| "loss": 1.101, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.89390844772857e-05, | |
| "loss": 1.1153, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.891534610078176e-05, | |
| "loss": 1.1023, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.889160772427781e-05, | |
| "loss": 1.0935, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.886786934777387e-05, | |
| "loss": 1.0898, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.884413097126992e-05, | |
| "loss": 1.0846, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.882039259476597e-05, | |
| "loss": 1.0963, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.879665421826203e-05, | |
| "loss": 1.0764, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.877291584175809e-05, | |
| "loss": 1.0649, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8749177465254145e-05, | |
| "loss": 1.0935, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.8725439088750194e-05, | |
| "loss": 1.0739, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.870170071224625e-05, | |
| "loss": 1.0779, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.8677962335742305e-05, | |
| "loss": 1.083, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.865422395923836e-05, | |
| "loss": 1.0729, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.8630485582734416e-05, | |
| "loss": 1.0587, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.860679468298348e-05, | |
| "loss": 1.078, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.858305630647954e-05, | |
| "loss": 1.0802, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.855931792997559e-05, | |
| "loss": 1.0538, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.853557955347164e-05, | |
| "loss": 1.0562, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.85118411769677e-05, | |
| "loss": 1.0516, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.848810280046375e-05, | |
| "loss": 1.0624, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.8464364423959815e-05, | |
| "loss": 1.0587, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.8440626047455864e-05, | |
| "loss": 1.0544, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.841688767095192e-05, | |
| "loss": 1.0467, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.8393291724707e-05, | |
| "loss": 1.0613, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.836955334820306e-05, | |
| "loss": 1.0365, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.834581497169911e-05, | |
| "loss": 1.0424, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.832207659519516e-05, | |
| "loss": 1.0442, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.829833821869122e-05, | |
| "loss": 1.0354, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.8274789749199304e-05, | |
| "loss": 1.0543, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.825105137269536e-05, | |
| "loss": 1.0476, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.8227312996191415e-05, | |
| "loss": 1.0427, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.820357461968748e-05, | |
| "loss": 1.0174, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.8179836243183526e-05, | |
| "loss": 1.0325, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.815609786667958e-05, | |
| "loss": 1.0126, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.813240696692865e-05, | |
| "loss": 1.0335, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.81086685904247e-05, | |
| "loss": 1.0219, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.808493021392075e-05, | |
| "loss": 1.0291, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.8061191837416815e-05, | |
| "loss": 1.0251, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.803745346091287e-05, | |
| "loss": 1.01, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.8013715084408925e-05, | |
| "loss": 1.0224, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.7989976707904974e-05, | |
| "loss": 1.0163, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.796623833140103e-05, | |
| "loss": 1.0214, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.7942499954897085e-05, | |
| "loss": 1.0187, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.791876157839314e-05, | |
| "loss": 1.0091, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.78950232018892e-05, | |
| "loss": 1.0256, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.787128482538525e-05, | |
| "loss": 1.0025, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.784754644888131e-05, | |
| "loss": 1.0013, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.782380807237736e-05, | |
| "loss": 1.0098, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.780006969587342e-05, | |
| "loss": 1.017, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.777637879612248e-05, | |
| "loss": 1.0114, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.775264041961854e-05, | |
| "loss": 0.9977, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.7728902043114596e-05, | |
| "loss": 1.0107, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.7705163666610644e-05, | |
| "loss": 0.9869, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.76814252901067e-05, | |
| "loss": 1.008, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.7657686913602755e-05, | |
| "loss": 1.0025, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.763394853709881e-05, | |
| "loss": 0.9964, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.7610210160594866e-05, | |
| "loss": 0.9921, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.758647178409092e-05, | |
| "loss": 0.9938, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.756273340758698e-05, | |
| "loss": 0.9934, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.753899503108303e-05, | |
| "loss": 1.0249, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.751530413133209e-05, | |
| "loss": 0.9832, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.749156575482815e-05, | |
| "loss": 1.001, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.74678273783242e-05, | |
| "loss": 0.9755, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.7444089001820266e-05, | |
| "loss": 0.995, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.742035062531632e-05, | |
| "loss": 0.9824, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.739661224881237e-05, | |
| "loss": 0.9759, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.7372873872308425e-05, | |
| "loss": 0.98, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.734913549580448e-05, | |
| "loss": 0.9905, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.7325397119300536e-05, | |
| "loss": 0.9811, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.730165874279659e-05, | |
| "loss": 0.9873, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.727792036629265e-05, | |
| "loss": 0.9804, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.7254229466541714e-05, | |
| "loss": 0.9695, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.723049109003776e-05, | |
| "loss": 0.9751, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.720675271353382e-05, | |
| "loss": 0.9735, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.7183061813782885e-05, | |
| "loss": 0.9832, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.7159370914031945e-05, | |
| "loss": 0.9719, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 40.8492, | |
| "eval_gen_len": 13.8028, | |
| "eval_loss": 0.9435123801231384, | |
| "eval_runtime": 10127.6159, | |
| "eval_samples_per_second": 14.005, | |
| "eval_steps_per_second": 1.751, | |
| "step": 70919 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.7135632537528e-05, | |
| "loss": 0.9687, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.7111894161024056e-05, | |
| "loss": 0.9639, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.708815578452011e-05, | |
| "loss": 0.9669, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.7064417408016166e-05, | |
| "loss": 0.956, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.704067903151222e-05, | |
| "loss": 0.9563, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.701694065500828e-05, | |
| "loss": 0.9564, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.699320227850433e-05, | |
| "loss": 0.9555, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.696946390200039e-05, | |
| "loss": 0.9408, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.6945725525496444e-05, | |
| "loss": 0.9347, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.69219871489925e-05, | |
| "loss": 0.9368, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.6898296249241566e-05, | |
| "loss": 0.9401, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.6874557872737615e-05, | |
| "loss": 0.9383, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.685081949623367e-05, | |
| "loss": 0.9232, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.6827081119729726e-05, | |
| "loss": 0.9046, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.680334274322578e-05, | |
| "loss": 0.9236, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.6779604366721837e-05, | |
| "loss": 0.9091, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.675586599021789e-05, | |
| "loss": 0.9036, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.673212761371395e-05, | |
| "loss": 0.909, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.670838923721e-05, | |
| "loss": 0.9003, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.668469833745906e-05, | |
| "loss": 0.8957, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.666095996095512e-05, | |
| "loss": 0.8928, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.66373640147102e-05, | |
| "loss": 0.8842, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.6613625638206256e-05, | |
| "loss": 0.885, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.658988726170231e-05, | |
| "loss": 0.8821, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.656614888519836e-05, | |
| "loss": 0.8769, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.6542410508694416e-05, | |
| "loss": 0.8681, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.651867213219048e-05, | |
| "loss": 0.8767, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.6494933755686534e-05, | |
| "loss": 0.8702, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.647119537918259e-05, | |
| "loss": 0.8753, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.644745700267864e-05, | |
| "loss": 0.8587, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.642371862617469e-05, | |
| "loss": 0.8459, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.639998024967075e-05, | |
| "loss": 0.853, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.6376241873166804e-05, | |
| "loss": 0.8811, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.635255097341587e-05, | |
| "loss": 0.8618, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.6328812596911926e-05, | |
| "loss": 0.8493, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.630507422040798e-05, | |
| "loss": 0.8603, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.628133584390403e-05, | |
| "loss": 0.8333, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.6257597467400086e-05, | |
| "loss": 0.8386, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.623385909089614e-05, | |
| "loss": 0.8424, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.6210120714392204e-05, | |
| "loss": 0.8519, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.618638233788826e-05, | |
| "loss": 0.8297, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.616264396138431e-05, | |
| "loss": 0.8406, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.613890558488036e-05, | |
| "loss": 0.8392, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.611516720837642e-05, | |
| "loss": 0.8269, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.6091476308625486e-05, | |
| "loss": 0.8392, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.6067785408874545e-05, | |
| "loss": 0.8474, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.60440470323706e-05, | |
| "loss": 0.8329, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.6020308655866656e-05, | |
| "loss": 0.8327, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.599657027936271e-05, | |
| "loss": 0.8435, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.597283190285877e-05, | |
| "loss": 0.8343, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.5949141003107834e-05, | |
| "loss": 0.8316, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.592540262660389e-05, | |
| "loss": 0.8339, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.590166425009994e-05, | |
| "loss": 0.8199, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.5877925873595994e-05, | |
| "loss": 0.8162, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.585418749709205e-05, | |
| "loss": 0.8239, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.5830496597341116e-05, | |
| "loss": 0.8197, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.5806805697590176e-05, | |
| "loss": 0.8164, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.578306732108623e-05, | |
| "loss": 0.818, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.575932894458229e-05, | |
| "loss": 0.8189, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.573559056807834e-05, | |
| "loss": 0.8303, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.57118521915744e-05, | |
| "loss": 0.8174, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.568811381507045e-05, | |
| "loss": 0.814, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.566437543856651e-05, | |
| "loss": 0.8226, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.5640637062062564e-05, | |
| "loss": 0.818, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.561689868555862e-05, | |
| "loss": 0.8054, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.5593160309054675e-05, | |
| "loss": 0.8191, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.556946940930374e-05, | |
| "loss": 0.8134, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.554573103279979e-05, | |
| "loss": 0.8077, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.5521992656295846e-05, | |
| "loss": 0.8081, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.54982542797919e-05, | |
| "loss": 0.7985, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.547451590328796e-05, | |
| "loss": 0.8064, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.545077752678401e-05, | |
| "loss": 0.8008, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.542703915028007e-05, | |
| "loss": 0.7934, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.540330077377612e-05, | |
| "loss": 0.7915, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.537956239727218e-05, | |
| "loss": 0.8087, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.5355824020768234e-05, | |
| "loss": 0.7941, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.533208564426429e-05, | |
| "loss": 0.8032, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.5308347267760345e-05, | |
| "loss": 0.8061, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.5284608891256394e-05, | |
| "loss": 0.7974, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.526091799150546e-05, | |
| "loss": 0.7915, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.523722709175453e-05, | |
| "loss": 0.7998, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.5213488715250576e-05, | |
| "loss": 0.8014, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.518975033874663e-05, | |
| "loss": 0.7848, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.516601196224269e-05, | |
| "loss": 0.7914, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.5142321062491754e-05, | |
| "loss": 0.795, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.5118630162740814e-05, | |
| "loss": 0.7913, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.509489178623687e-05, | |
| "loss": 0.7936, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.5071153409732924e-05, | |
| "loss": 0.7932, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.504741503322898e-05, | |
| "loss": 0.7835, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.502367665672504e-05, | |
| "loss": 0.7942, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.499993828022109e-05, | |
| "loss": 0.7771, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.4976199903717146e-05, | |
| "loss": 0.7805, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.4952509003966206e-05, | |
| "loss": 0.7952, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.492877062746226e-05, | |
| "loss": 0.7799, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.490503225095832e-05, | |
| "loss": 0.7932, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.488129387445438e-05, | |
| "loss": 0.7919, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.4857555497950435e-05, | |
| "loss": 0.7776, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.4833864598199495e-05, | |
| "loss": 0.7731, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.481012622169555e-05, | |
| "loss": 0.7778, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.47863878451916e-05, | |
| "loss": 0.7638, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.476264946868766e-05, | |
| "loss": 0.7819, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.473891109218372e-05, | |
| "loss": 0.7717, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.471517271567977e-05, | |
| "loss": 0.7803, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.469143433917583e-05, | |
| "loss": 0.7807, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.4667695962671876e-05, | |
| "loss": 0.7651, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.464395758616793e-05, | |
| "loss": 0.7811, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.4620266686417e-05, | |
| "loss": 0.7794, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.4596528309913054e-05, | |
| "loss": 0.7763, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.457278993340911e-05, | |
| "loss": 0.7757, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.4549051556905165e-05, | |
| "loss": 0.7683, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.4525360657154225e-05, | |
| "loss": 0.7764, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.450162228065028e-05, | |
| "loss": 0.7661, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.4477883904146336e-05, | |
| "loss": 0.7645, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.445414552764239e-05, | |
| "loss": 0.7703, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.443040715113845e-05, | |
| "loss": 0.7791, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.44066687746345e-05, | |
| "loss": 0.7657, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.438293039813056e-05, | |
| "loss": 0.7651, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.435919202162661e-05, | |
| "loss": 0.7803, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.433545364512267e-05, | |
| "loss": 0.7504, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.4311762745371735e-05, | |
| "loss": 0.7785, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.4288024368867784e-05, | |
| "loss": 0.7592, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.426428599236384e-05, | |
| "loss": 0.7714, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.4240547615859895e-05, | |
| "loss": 0.7652, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.421680923935595e-05, | |
| "loss": 0.7608, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.4193070862852006e-05, | |
| "loss": 0.7688, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.416933248634806e-05, | |
| "loss": 0.79, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.414559410984412e-05, | |
| "loss": 0.7526, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.412190321009318e-05, | |
| "loss": 0.765, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.409816483358923e-05, | |
| "loss": 0.7554, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.407442645708529e-05, | |
| "loss": 0.7636, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.4050735557334354e-05, | |
| "loss": 0.7582, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.402699718083041e-05, | |
| "loss": 0.7504, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.400330628107947e-05, | |
| "loss": 0.7591, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.3979567904575525e-05, | |
| "loss": 0.7629, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.395582952807158e-05, | |
| "loss": 0.7543, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.3932091151567636e-05, | |
| "loss": 0.7634, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.390835277506369e-05, | |
| "loss": 0.7553, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.388461439855975e-05, | |
| "loss": 0.7548, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.38608760220558e-05, | |
| "loss": 0.7554, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.383713764555186e-05, | |
| "loss": 0.7539, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.3813399269047914e-05, | |
| "loss": 0.7562, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.378966089254396e-05, | |
| "loss": 0.7537, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 42.7907, | |
| "eval_gen_len": 13.7941, | |
| "eval_loss": 0.8987888693809509, | |
| "eval_runtime": 10044.1408, | |
| "eval_samples_per_second": 14.121, | |
| "eval_steps_per_second": 1.765, | |
| "step": 141839 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.376596999279303e-05, | |
| "loss": 0.7511, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.3742231616289084e-05, | |
| "loss": 0.75, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.371849323978514e-05, | |
| "loss": 0.7489, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.3694754863281195e-05, | |
| "loss": 0.7359, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.367101648677725e-05, | |
| "loss": 0.7416, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.364732558702632e-05, | |
| "loss": 0.7421, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.362358721052237e-05, | |
| "loss": 0.7448, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.359984883401843e-05, | |
| "loss": 0.7314, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.357611045751448e-05, | |
| "loss": 0.731, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.3552419557763544e-05, | |
| "loss": 0.7307, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.35286811812596e-05, | |
| "loss": 0.7411, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.3504942804755655e-05, | |
| "loss": 0.7298, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.348120442825171e-05, | |
| "loss": 0.7144, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.3457466051747766e-05, | |
| "loss": 0.7146, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.343372767524382e-05, | |
| "loss": 0.7157, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.341003677549288e-05, | |
| "loss": 0.7116, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.338629839898894e-05, | |
| "loss": 0.7105, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.336256002248499e-05, | |
| "loss": 0.7096, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.3338916599487063e-05, | |
| "loss": 0.7026, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.331517822298312e-05, | |
| "loss": 0.706, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.3291439846479174e-05, | |
| "loss": 0.6996, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.326770146997523e-05, | |
| "loss": 0.6965, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.3243963093471285e-05, | |
| "loss": 0.699, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.322022471696734e-05, | |
| "loss": 0.6926, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.3196486340463396e-05, | |
| "loss": 0.6925, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.3172747963959445e-05, | |
| "loss": 0.6825, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.31490095874555e-05, | |
| "loss": 0.6869, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.312527121095156e-05, | |
| "loss": 0.691, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.310153283444762e-05, | |
| "loss": 0.682, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.3077794457943673e-05, | |
| "loss": 0.6822, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.305405608143972e-05, | |
| "loss": 0.6666, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.303031770493578e-05, | |
| "loss": 0.6749, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.300657932843183e-05, | |
| "loss": 0.6903, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.29828884286809e-05, | |
| "loss": 0.6816, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.2959150052176955e-05, | |
| "loss": 0.6776, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.2935459152426015e-05, | |
| "loss": 0.6787, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.2911768252675075e-05, | |
| "loss": 0.6559, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.288807735292414e-05, | |
| "loss": 0.668, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.28643389764202e-05, | |
| "loss": 0.6755, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.284060059991625e-05, | |
| "loss": 0.6751, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.281686222341231e-05, | |
| "loss": 0.6608, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.2793123846908364e-05, | |
| "loss": 0.6562, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.276938547040442e-05, | |
| "loss": 0.6672, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.2745647093900475e-05, | |
| "loss": 0.6577, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.272190871739653e-05, | |
| "loss": 0.6714, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.2698170340892586e-05, | |
| "loss": 0.6706, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.2674479441141646e-05, | |
| "loss": 0.6593, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.26507410646377e-05, | |
| "loss": 0.6633, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.262705016488677e-05, | |
| "loss": 0.6672, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.260331178838282e-05, | |
| "loss": 0.6584, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.257957341187888e-05, | |
| "loss": 0.6593, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.255583503537493e-05, | |
| "loss": 0.6651, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.253209665887098e-05, | |
| "loss": 0.6519, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.250835828236704e-05, | |
| "loss": 0.6518, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.24846199058631e-05, | |
| "loss": 0.6598, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.2460881529359156e-05, | |
| "loss": 0.6445, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.2437143152855205e-05, | |
| "loss": 0.6499, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.241340477635126e-05, | |
| "loss": 0.6516, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.2389666399847316e-05, | |
| "loss": 0.6553, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.236592802334337e-05, | |
| "loss": 0.668, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.2342189646839427e-05, | |
| "loss": 0.6447, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.231845127033548e-05, | |
| "loss": 0.6474, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.229471289383154e-05, | |
| "loss": 0.6594, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.227097451732759e-05, | |
| "loss": 0.6516, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.224723614082365e-05, | |
| "loss": 0.6404, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.222354524107271e-05, | |
| "loss": 0.6526, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.2199806864568764e-05, | |
| "loss": 0.6459, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 4.2176068488064826e-05, | |
| "loss": 0.6519, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.2152377588313886e-05, | |
| "loss": 0.6391, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.212863921180994e-05, | |
| "loss": 0.6346, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.2104900835306e-05, | |
| "loss": 0.6395, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.2081162458802046e-05, | |
| "loss": 0.6409, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.20574240822981e-05, | |
| "loss": 0.6285, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.2033685705794163e-05, | |
| "loss": 0.6307, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.200999480604322e-05, | |
| "loss": 0.6506, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.198625642953928e-05, | |
| "loss": 0.6307, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.1962518053035334e-05, | |
| "loss": 0.6446, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.193877967653139e-05, | |
| "loss": 0.6422, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.1915088776780457e-05, | |
| "loss": 0.6398, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.1891397877029516e-05, | |
| "loss": 0.6333, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.186765950052557e-05, | |
| "loss": 0.6363, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.184392112402162e-05, | |
| "loss": 0.6319, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.1820182747517676e-05, | |
| "loss": 0.6294, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.179649184776674e-05, | |
| "loss": 0.6366, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.17727534712628e-05, | |
| "loss": 0.6384, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.1749015094758854e-05, | |
| "loss": 0.6365, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.172527671825491e-05, | |
| "loss": 0.6331, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.1701538341750965e-05, | |
| "loss": 0.6327, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.167779996524701e-05, | |
| "loss": 0.628, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.1654061588743076e-05, | |
| "loss": 0.6309, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.163032321223913e-05, | |
| "loss": 0.6159, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.1606584835735187e-05, | |
| "loss": 0.6307, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.158284645923124e-05, | |
| "loss": 0.6356, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.155910808272729e-05, | |
| "loss": 0.6269, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.1535369706223346e-05, | |
| "loss": 0.6368, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.15116313297194e-05, | |
| "loss": 0.6282, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.1487892953215464e-05, | |
| "loss": 0.6226, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.146415457671152e-05, | |
| "loss": 0.6181, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.144041620020757e-05, | |
| "loss": 0.6138, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.1416677823703623e-05, | |
| "loss": 0.6184, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.139298692395269e-05, | |
| "loss": 0.624, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.136924854744874e-05, | |
| "loss": 0.6125, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.13455101709448e-05, | |
| "loss": 0.6261, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.1321771794440857e-05, | |
| "loss": 0.6189, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.129803341793691e-05, | |
| "loss": 0.6149, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.127429504143297e-05, | |
| "loss": 0.6296, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.1250556664929016e-05, | |
| "loss": 0.6263, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.122681828842507e-05, | |
| "loss": 0.6298, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.120312738867414e-05, | |
| "loss": 0.6102, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.1179389012170194e-05, | |
| "loss": 0.6248, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.115565063566625e-05, | |
| "loss": 0.6121, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.1131912259162305e-05, | |
| "loss": 0.6156, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.1108221359411365e-05, | |
| "loss": 0.6076, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.1084577936413436e-05, | |
| "loss": 0.6231, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.106083955990949e-05, | |
| "loss": 0.6281, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.103710118340555e-05, | |
| "loss": 0.6157, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.10133628069016e-05, | |
| "loss": 0.6086, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.098962443039766e-05, | |
| "loss": 0.6225, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.096588605389371e-05, | |
| "loss": 0.6078, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.094214767738977e-05, | |
| "loss": 0.6217, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.0918409300885824e-05, | |
| "loss": 0.6096, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.089467092438188e-05, | |
| "loss": 0.6253, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.087102750138395e-05, | |
| "loss": 0.6068, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.0847289124880006e-05, | |
| "loss": 0.613, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.082355074837606e-05, | |
| "loss": 0.6179, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.079981237187212e-05, | |
| "loss": 0.6312, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.077612147212118e-05, | |
| "loss": 0.601, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.075238309561723e-05, | |
| "loss": 0.615, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.072864471911329e-05, | |
| "loss": 0.6071, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.0704906342609344e-05, | |
| "loss": 0.6135, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.06811679661054e-05, | |
| "loss": 0.6073, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.065747706635446e-05, | |
| "loss": 0.6073, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.0633738689850515e-05, | |
| "loss": 0.6053, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.061000031334658e-05, | |
| "loss": 0.6116, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.058626193684263e-05, | |
| "loss": 0.606, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.056252356033868e-05, | |
| "loss": 0.6183, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.0538785183834736e-05, | |
| "loss": 0.601, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.051504680733079e-05, | |
| "loss": 0.6041, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.049130843082685e-05, | |
| "loss": 0.6073, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.04675700543229e-05, | |
| "loss": 0.6106, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.044383167781896e-05, | |
| "loss": 0.6105, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.0420093301315014e-05, | |
| "loss": 0.5973, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 43.4697, | |
| "eval_gen_len": 13.6556, | |
| "eval_loss": 0.9142627120018005, | |
| "eval_runtime": 9819.0968, | |
| "eval_samples_per_second": 14.445, | |
| "eval_steps_per_second": 1.806, | |
| "step": 212758 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.039635492481107e-05, | |
| "loss": 0.6014, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.037266402506013e-05, | |
| "loss": 0.6099, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.0348925648556185e-05, | |
| "loss": 0.5951, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.032518727205224e-05, | |
| "loss": 0.597, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.03014488955483e-05, | |
| "loss": 0.596, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.027771051904435e-05, | |
| "loss": 0.5914, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.0253972142540406e-05, | |
| "loss": 0.5979, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.023023376603646e-05, | |
| "loss": 0.5855, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.020649538953252e-05, | |
| "loss": 0.5887, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 4.018275701302857e-05, | |
| "loss": 0.5853, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 4.015901863652463e-05, | |
| "loss": 0.5921, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.013528026002068e-05, | |
| "loss": 0.5873, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.0111589360269744e-05, | |
| "loss": 0.5708, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.00878509837658e-05, | |
| "loss": 0.5777, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.0064112607261855e-05, | |
| "loss": 0.5772, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.004037423075791e-05, | |
| "loss": 0.573, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.0016635854253966e-05, | |
| "loss": 0.5734, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.999289747775002e-05, | |
| "loss": 0.5682, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.996920657799909e-05, | |
| "loss": 0.5628, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.9945468201495136e-05, | |
| "loss": 0.5687, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.992172982499119e-05, | |
| "loss": 0.5569, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.989799144848725e-05, | |
| "loss": 0.5571, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.98742530719833e-05, | |
| "loss": 0.5619, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.9850514695479365e-05, | |
| "loss": 0.5635, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.982677631897542e-05, | |
| "loss": 0.5538, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.980303794247147e-05, | |
| "loss": 0.5494, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.9779299565967525e-05, | |
| "loss": 0.5532, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.975556118946358e-05, | |
| "loss": 0.5576, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.973187028971264e-05, | |
| "loss": 0.5499, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.970822686671472e-05, | |
| "loss": 0.5434, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.968448849021077e-05, | |
| "loss": 0.5421, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.966075011370682e-05, | |
| "loss": 0.5377, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.963701173720288e-05, | |
| "loss": 0.5614, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.961327336069894e-05, | |
| "loss": 0.5458, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.9589534984194995e-05, | |
| "loss": 0.5483, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.9565796607691044e-05, | |
| "loss": 0.5423, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.95420582311871e-05, | |
| "loss": 0.5346, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.9518319854683155e-05, | |
| "loss": 0.5334, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.9494628954932215e-05, | |
| "loss": 0.5451, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.947089057842828e-05, | |
| "loss": 0.5378, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.944715220192433e-05, | |
| "loss": 0.5323, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.942346130217339e-05, | |
| "loss": 0.5346, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.939972292566945e-05, | |
| "loss": 0.5334, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.9375984549165504e-05, | |
| "loss": 0.5292, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.935224617266155e-05, | |
| "loss": 0.5442, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.9328507796157615e-05, | |
| "loss": 0.5423, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.930476941965367e-05, | |
| "loss": 0.53, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.9281031043149725e-05, | |
| "loss": 0.5359, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.9257340143398785e-05, | |
| "loss": 0.5432, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.923360176689484e-05, | |
| "loss": 0.5325, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.9209863390390896e-05, | |
| "loss": 0.53, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.918612501388695e-05, | |
| "loss": 0.5347, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.916238663738301e-05, | |
| "loss": 0.5268, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.913864826087906e-05, | |
| "loss": 0.5233, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.911490988437512e-05, | |
| "loss": 0.5281, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.9091171507871174e-05, | |
| "loss": 0.5191, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.906743313136723e-05, | |
| "loss": 0.5226, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.904369475486328e-05, | |
| "loss": 0.525, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.901995637835934e-05, | |
| "loss": 0.5252, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.8996218001855396e-05, | |
| "loss": 0.5406, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.897257457885746e-05, | |
| "loss": 0.5168, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.8948836202353515e-05, | |
| "loss": 0.5218, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.892509782584958e-05, | |
| "loss": 0.5304, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.890135944934563e-05, | |
| "loss": 0.5217, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.887762107284169e-05, | |
| "loss": 0.5143, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.885388269633774e-05, | |
| "loss": 0.5326, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.883014431983379e-05, | |
| "loss": 0.5152, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.880640594332985e-05, | |
| "loss": 0.5288, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.8782667566825904e-05, | |
| "loss": 0.517, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.8758929190321966e-05, | |
| "loss": 0.5092, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.8735238290571026e-05, | |
| "loss": 0.5147, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.871149991406708e-05, | |
| "loss": 0.5175, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.868780901431614e-05, | |
| "loss": 0.4983, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.866411811456521e-05, | |
| "loss": 0.5145, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.8640379738061264e-05, | |
| "loss": 0.5226, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.861664136155731e-05, | |
| "loss": 0.5148, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.859290298505337e-05, | |
| "loss": 0.5172, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.856916460854942e-05, | |
| "loss": 0.5245, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.854547370879849e-05, | |
| "loss": 0.5164, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.8521735332294545e-05, | |
| "loss": 0.5087, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.84979969557906e-05, | |
| "loss": 0.517, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.8474258579286656e-05, | |
| "loss": 0.5089, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.8450520202782705e-05, | |
| "loss": 0.5045, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.842682930303177e-05, | |
| "loss": 0.5169, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.840309092652783e-05, | |
| "loss": 0.5204, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.837935255002388e-05, | |
| "loss": 0.5064, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.835561417351994e-05, | |
| "loss": 0.5074, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.8331875797015994e-05, | |
| "loss": 0.5147, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.830813742051205e-05, | |
| "loss": 0.5031, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.8284399044008104e-05, | |
| "loss": 0.5046, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.826066066750415e-05, | |
| "loss": 0.5023, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.8236922291000215e-05, | |
| "loss": 0.5053, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.821318391449627e-05, | |
| "loss": 0.5145, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.8189445537992326e-05, | |
| "loss": 0.5037, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.816570716148838e-05, | |
| "loss": 0.5164, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.814196878498443e-05, | |
| "loss": 0.5089, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.8118230408480486e-05, | |
| "loss": 0.499, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.809453950872955e-05, | |
| "loss": 0.5004, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.807084860897861e-05, | |
| "loss": 0.4955, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.804715770922768e-05, | |
| "loss": 0.5021, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.802341933272373e-05, | |
| "loss": 0.5064, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.799968095621979e-05, | |
| "loss": 0.4947, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.7975942579715846e-05, | |
| "loss": 0.5033, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.79522042032119e-05, | |
| "loss": 0.5022, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.792846582670796e-05, | |
| "loss": 0.4956, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.790477492695702e-05, | |
| "loss": 0.5102, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.788103655045307e-05, | |
| "loss": 0.509, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.785729817394913e-05, | |
| "loss": 0.5048, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.783355979744518e-05, | |
| "loss": 0.4954, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.780982142094124e-05, | |
| "loss": 0.5009, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.7786083044437294e-05, | |
| "loss": 0.4909, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.776234466793335e-05, | |
| "loss": 0.5018, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.77386062914294e-05, | |
| "loss": 0.4903, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.7714867914925454e-05, | |
| "loss": 0.504, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.7691129538421516e-05, | |
| "loss": 0.5088, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.7667438638670576e-05, | |
| "loss": 0.5009, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.7643747738919636e-05, | |
| "loss": 0.4924, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.76200568391687e-05, | |
| "loss": 0.4999, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.759631846266476e-05, | |
| "loss": 0.4997, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.7572580086160813e-05, | |
| "loss": 0.4968, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.754884170965687e-05, | |
| "loss": 0.4905, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.7525103333152924e-05, | |
| "loss": 0.5055, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.750136495664898e-05, | |
| "loss": 0.4918, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.747762658014503e-05, | |
| "loss": 0.4947, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.745388820364109e-05, | |
| "loss": 0.5016, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.7430149827137146e-05, | |
| "loss": 0.5084, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.74064114506332e-05, | |
| "loss": 0.4882, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.738272055088226e-05, | |
| "loss": 0.4948, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.735898217437832e-05, | |
| "loss": 0.4971, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.733524379787437e-05, | |
| "loss": 0.4867, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.731150542137043e-05, | |
| "loss": 0.4917, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.7287767044866484e-05, | |
| "loss": 0.4853, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.726402866836254e-05, | |
| "loss": 0.4948, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.7240290291858594e-05, | |
| "loss": 0.4935, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.721655191535465e-05, | |
| "loss": 0.4906, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.71928135388507e-05, | |
| "loss": 0.4992, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.716907516234676e-05, | |
| "loss": 0.4814, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.7145336785842816e-05, | |
| "loss": 0.4903, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.712159840933887e-05, | |
| "loss": 0.4869, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 3.709786003283493e-05, | |
| "loss": 0.493, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.7074121656330976e-05, | |
| "loss": 0.4909, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.705038327982703e-05, | |
| "loss": 0.4873, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 43.6153, | |
| "eval_gen_len": 13.6544, | |
| "eval_loss": 0.9757916927337646, | |
| "eval_runtime": 9810.7798, | |
| "eval_samples_per_second": 14.457, | |
| "eval_steps_per_second": 1.807, | |
| "step": 283678 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1063785, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "total_flos": 9.836272613473649e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |