| { | |
| "best_metric": 6.406748294830322, | |
| "best_model_checkpoint": "HBERTv1_emb_compress_48_L12_H64_A2/checkpoint-290000", | |
| "epoch": 5.0, | |
| "eval_steps": 10000, | |
| "global_step": 305145, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 10.3179, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 10.2814, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.5e-06, | |
| "loss": 10.2576, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 10.2271, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.5e-06, | |
| "loss": 10.1895, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3e-06, | |
| "loss": 10.1441, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.5e-06, | |
| "loss": 10.0896, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 10.0257, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.5e-06, | |
| "loss": 9.9517, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-06, | |
| "loss": 9.8689, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 9.7764, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6e-06, | |
| "loss": 9.6759, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 9.5681, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 7e-06, | |
| "loss": 9.4531, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 9.3312, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 9.2039, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.5e-06, | |
| "loss": 9.0723, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9e-06, | |
| "loss": 8.9366, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.5e-06, | |
| "loss": 8.7964, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1e-05, | |
| "loss": 8.6554, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_accuracy": 0.048290781267576265, | |
| "eval_loss": 8.584630012512207, | |
| "eval_runtime": 600.2506, | |
| "eval_samples_per_second": 513.712, | |
| "eval_steps_per_second": 5.353, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.983059174304157e-06, | |
| "loss": 8.5166, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.966118348608312e-06, | |
| "loss": 8.3802, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.949177522912468e-06, | |
| "loss": 8.2563, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.932236697216624e-06, | |
| "loss": 8.1368, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.91529587152078e-06, | |
| "loss": 8.0261, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.898355045824935e-06, | |
| "loss": 7.9253, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.881414220129089e-06, | |
| "loss": 7.8284, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.864473394433246e-06, | |
| "loss": 7.7407, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.847532568737402e-06, | |
| "loss": 7.6611, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.830591743041556e-06, | |
| "loss": 7.59, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.813650917345712e-06, | |
| "loss": 7.5243, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.796710091649867e-06, | |
| "loss": 7.4708, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.779769265954023e-06, | |
| "loss": 7.4185, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.762828440258179e-06, | |
| "loss": 7.3726, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.745887614562334e-06, | |
| "loss": 7.3442, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.72894678886649e-06, | |
| "loss": 7.3171, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.712005963170646e-06, | |
| "loss": 7.2839, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.695065137474802e-06, | |
| "loss": 7.2703, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.678124311778957e-06, | |
| "loss": 7.2537, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.661183486083113e-06, | |
| "loss": 7.2331, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_accuracy": 0.05423351427871706, | |
| "eval_loss": 7.227975845336914, | |
| "eval_runtime": 605.972, | |
| "eval_samples_per_second": 508.862, | |
| "eval_steps_per_second": 5.302, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.644242660387269e-06, | |
| "loss": 7.2207, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.627301834691424e-06, | |
| "loss": 7.2044, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.610361008995578e-06, | |
| "loss": 7.1869, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.593420183299736e-06, | |
| "loss": 7.1736, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.576479357603891e-06, | |
| "loss": 7.1594, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.559538531908045e-06, | |
| "loss": 7.1396, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.542597706212201e-06, | |
| "loss": 7.1306, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.525656880516357e-06, | |
| "loss": 7.1185, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.508716054820512e-06, | |
| "loss": 7.1029, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.491775229124668e-06, | |
| "loss": 7.0927, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.474834403428824e-06, | |
| "loss": 7.0818, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.45789357773298e-06, | |
| "loss": 7.0696, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.440952752037135e-06, | |
| "loss": 7.0553, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.424011926341291e-06, | |
| "loss": 7.0496, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.407071100645447e-06, | |
| "loss": 7.038, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.390130274949602e-06, | |
| "loss": 7.0289, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.373189449253758e-06, | |
| "loss": 7.0244, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.356248623557914e-06, | |
| "loss": 7.0145, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.339307797862068e-06, | |
| "loss": 7.0029, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.322366972166225e-06, | |
| "loss": 7.0014, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_accuracy": 0.06774155914288765, | |
| "eval_loss": 6.992653846740723, | |
| "eval_runtime": 600.1356, | |
| "eval_samples_per_second": 513.811, | |
| "eval_steps_per_second": 5.354, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.305426146470379e-06, | |
| "loss": 6.9903, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.288485320774535e-06, | |
| "loss": 6.9807, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.27154449507869e-06, | |
| "loss": 6.9704, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.254603669382846e-06, | |
| "loss": 6.9613, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.237662843687002e-06, | |
| "loss": 6.9601, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.220722017991158e-06, | |
| "loss": 6.9596, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.203781192295313e-06, | |
| "loss": 6.9484, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.186840366599469e-06, | |
| "loss": 6.9372, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.169899540903625e-06, | |
| "loss": 6.932, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.15295871520778e-06, | |
| "loss": 6.925, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.136017889511936e-06, | |
| "loss": 6.9252, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.11907706381609e-06, | |
| "loss": 6.9177, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.102136238120247e-06, | |
| "loss": 6.906, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.085195412424403e-06, | |
| "loss": 6.8972, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.068254586728557e-06, | |
| "loss": 6.8942, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.051313761032714e-06, | |
| "loss": 6.8912, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.034372935336868e-06, | |
| "loss": 6.8861, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.017432109641024e-06, | |
| "loss": 6.8749, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.00049128394518e-06, | |
| "loss": 6.8718, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.983550458249335e-06, | |
| "loss": 6.8699, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_accuracy": 0.08556003293560886, | |
| "eval_loss": 6.8636794090271, | |
| "eval_runtime": 598.5444, | |
| "eval_samples_per_second": 515.176, | |
| "eval_steps_per_second": 5.368, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.966609632553491e-06, | |
| "loss": 6.8626, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.949668806857647e-06, | |
| "loss": 6.8559, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.932727981161803e-06, | |
| "loss": 6.8509, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.915787155465958e-06, | |
| "loss": 6.844, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.898846329770114e-06, | |
| "loss": 6.8383, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.88190550407427e-06, | |
| "loss": 6.8398, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.864964678378425e-06, | |
| "loss": 6.8233, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.84802385268258e-06, | |
| "loss": 6.8242, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.831083026986737e-06, | |
| "loss": 6.82, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.814142201290892e-06, | |
| "loss": 6.8142, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.797201375595046e-06, | |
| "loss": 6.8122, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.780260549899204e-06, | |
| "loss": 6.8017, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.763319724203358e-06, | |
| "loss": 6.8021, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.746378898507513e-06, | |
| "loss": 6.8006, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.729438072811671e-06, | |
| "loss": 6.7897, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.712497247115825e-06, | |
| "loss": 6.7932, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.69555642141998e-06, | |
| "loss": 6.7881, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.678615595724136e-06, | |
| "loss": 6.7793, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.661674770028292e-06, | |
| "loss": 6.7823, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.644733944332448e-06, | |
| "loss": 6.7777, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_accuracy": 0.09222163830820224, | |
| "eval_loss": 6.772622108459473, | |
| "eval_runtime": 600.0695, | |
| "eval_samples_per_second": 513.867, | |
| "eval_steps_per_second": 5.354, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.627793118636603e-06, | |
| "loss": 6.7669, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.610852292940759e-06, | |
| "loss": 6.7646, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.593911467244915e-06, | |
| "loss": 6.7643, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.576970641549069e-06, | |
| "loss": 6.7579, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.560029815853226e-06, | |
| "loss": 6.753, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.543088990157382e-06, | |
| "loss": 6.7522, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.526148164461536e-06, | |
| "loss": 6.7583, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.509207338765693e-06, | |
| "loss": 6.747, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.492266513069847e-06, | |
| "loss": 6.7433, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.475325687374003e-06, | |
| "loss": 6.7394, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.45838486167816e-06, | |
| "loss": 6.7346, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.441444035982314e-06, | |
| "loss": 6.7408, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.42450321028647e-06, | |
| "loss": 6.7341, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.407562384590626e-06, | |
| "loss": 6.7339, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.390621558894781e-06, | |
| "loss": 6.7284, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.373680733198937e-06, | |
| "loss": 6.732, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.356739907503093e-06, | |
| "loss": 6.7212, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.339799081807248e-06, | |
| "loss": 6.7156, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.322858256111404e-06, | |
| "loss": 6.7182, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.305917430415558e-06, | |
| "loss": 6.7091, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_accuracy": 0.09736718543824835, | |
| "eval_loss": 6.7100605964660645, | |
| "eval_runtime": 601.1731, | |
| "eval_samples_per_second": 512.924, | |
| "eval_steps_per_second": 5.345, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.288976604719715e-06, | |
| "loss": 6.7102, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.272035779023871e-06, | |
| "loss": 6.7105, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.255094953328025e-06, | |
| "loss": 6.7048, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.238154127632183e-06, | |
| "loss": 6.7056, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.221213301936337e-06, | |
| "loss": 6.6984, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.204272476240492e-06, | |
| "loss": 6.6931, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.18733165054465e-06, | |
| "loss": 6.6938, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.170390824848804e-06, | |
| "loss": 6.6919, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.15344999915296e-06, | |
| "loss": 6.6935, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.136509173457115e-06, | |
| "loss": 6.6856, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.11956834776127e-06, | |
| "loss": 6.6865, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.102627522065426e-06, | |
| "loss": 6.6818, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.085686696369582e-06, | |
| "loss": 6.6852, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.068745870673738e-06, | |
| "loss": 6.6818, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.051805044977893e-06, | |
| "loss": 6.6742, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.034864219282047e-06, | |
| "loss": 6.6729, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.017923393586205e-06, | |
| "loss": 6.6691, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.00098256789036e-06, | |
| "loss": 6.6618, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.984041742194514e-06, | |
| "loss": 6.6685, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.967100916498672e-06, | |
| "loss": 6.6626, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "eval_accuracy": 0.1015238380889028, | |
| "eval_loss": 6.661967754364014, | |
| "eval_runtime": 605.0085, | |
| "eval_samples_per_second": 509.672, | |
| "eval_steps_per_second": 5.311, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.950160090802826e-06, | |
| "loss": 6.6613, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.933219265106982e-06, | |
| "loss": 6.6616, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.916278439411137e-06, | |
| "loss": 6.6625, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.899337613715293e-06, | |
| "loss": 6.6572, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.882396788019449e-06, | |
| "loss": 6.652, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.865455962323604e-06, | |
| "loss": 6.6498, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.84851513662776e-06, | |
| "loss": 6.6477, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.831574310931916e-06, | |
| "loss": 6.6456, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.814633485236071e-06, | |
| "loss": 6.6418, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.797692659540227e-06, | |
| "loss": 6.6447, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.780751833844383e-06, | |
| "loss": 6.6428, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.763811008148537e-06, | |
| "loss": 6.635, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.746870182452694e-06, | |
| "loss": 6.6335, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.729929356756848e-06, | |
| "loss": 6.637, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.712988531061004e-06, | |
| "loss": 6.6324, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.696047705365161e-06, | |
| "loss": 6.6344, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.679106879669315e-06, | |
| "loss": 6.629, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.662166053973471e-06, | |
| "loss": 6.6266, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.645225228277627e-06, | |
| "loss": 6.6295, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.628284402581782e-06, | |
| "loss": 6.6279, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_accuracy": 0.10400515267929009, | |
| "eval_loss": 6.625459671020508, | |
| "eval_runtime": 603.1684, | |
| "eval_samples_per_second": 511.227, | |
| "eval_steps_per_second": 5.327, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.611343576885938e-06, | |
| "loss": 6.6241, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 7.5944027511900945e-06, | |
| "loss": 6.6207, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.577461925494249e-06, | |
| "loss": 6.6216, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.560521099798405e-06, | |
| "loss": 6.619, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.54358027410256e-06, | |
| "loss": 6.6153, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 7.5266394484067164e-06, | |
| "loss": 6.6181, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.509698622710871e-06, | |
| "loss": 6.6137, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.492757797015027e-06, | |
| "loss": 6.6157, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.475816971319183e-06, | |
| "loss": 6.6114, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.458876145623338e-06, | |
| "loss": 6.6068, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 7.441935319927493e-06, | |
| "loss": 6.6081, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.42499449423165e-06, | |
| "loss": 6.6026, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 7.4080536685358054e-06, | |
| "loss": 6.6031, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.39111284283996e-06, | |
| "loss": 6.6086, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.374172017144117e-06, | |
| "loss": 6.6048, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.357231191448272e-06, | |
| "loss": 6.6031, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.340290365752427e-06, | |
| "loss": 6.5961, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.323349540056584e-06, | |
| "loss": 6.5977, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.306408714360739e-06, | |
| "loss": 6.598, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.2894678886648936e-06, | |
| "loss": 6.5917, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "eval_accuracy": 0.10684875378443538, | |
| "eval_loss": 6.594837188720703, | |
| "eval_runtime": 601.7144, | |
| "eval_samples_per_second": 512.462, | |
| "eval_steps_per_second": 5.34, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.272527062969049e-06, | |
| "loss": 6.5912, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.255586237273206e-06, | |
| "loss": 6.5903, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.238645411577361e-06, | |
| "loss": 6.5949, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.221704585881516e-06, | |
| "loss": 6.5932, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.204763760185672e-06, | |
| "loss": 6.5887, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.187822934489828e-06, | |
| "loss": 6.5879, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.1708821087939826e-06, | |
| "loss": 6.5892, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 7.153941283098139e-06, | |
| "loss": 6.5873, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.137000457402295e-06, | |
| "loss": 6.5908, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 7.12005963170645e-06, | |
| "loss": 6.5839, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 7.103118806010606e-06, | |
| "loss": 6.5803, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 7.086177980314761e-06, | |
| "loss": 6.5832, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 7.069237154618917e-06, | |
| "loss": 6.5801, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 7.052296328923072e-06, | |
| "loss": 6.5743, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.035355503227228e-06, | |
| "loss": 6.576, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 7.018414677531383e-06, | |
| "loss": 6.5739, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 7.001473851835539e-06, | |
| "loss": 6.5728, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.984533026139695e-06, | |
| "loss": 6.5703, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.96759220044385e-06, | |
| "loss": 6.5732, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.950651374748006e-06, | |
| "loss": 6.5691, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_accuracy": 0.10944561377176425, | |
| "eval_loss": 6.569516181945801, | |
| "eval_runtime": 601.8291, | |
| "eval_samples_per_second": 512.365, | |
| "eval_steps_per_second": 5.339, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.933710549052161e-06, | |
| "loss": 6.5714, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.916769723356317e-06, | |
| "loss": 6.5635, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.899828897660472e-06, | |
| "loss": 6.572, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.8828880719646285e-06, | |
| "loss": 6.5643, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 6.865947246268783e-06, | |
| "loss": 6.5663, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 6.849006420572939e-06, | |
| "loss": 6.5591, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 6.8320655948770955e-06, | |
| "loss": 6.5602, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 6.81512476918125e-06, | |
| "loss": 6.5603, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 6.798183943485406e-06, | |
| "loss": 6.5588, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 6.781243117789562e-06, | |
| "loss": 6.5568, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.7643022920937174e-06, | |
| "loss": 6.5571, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 6.747361466397872e-06, | |
| "loss": 6.5582, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.730420640702028e-06, | |
| "loss": 6.5519, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.7134798150061845e-06, | |
| "loss": 6.5571, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 6.696538989310339e-06, | |
| "loss": 6.5534, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.679598163614495e-06, | |
| "loss": 6.5516, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.662657337918651e-06, | |
| "loss": 6.5497, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.6457165122228064e-06, | |
| "loss": 6.5465, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.628775686526961e-06, | |
| "loss": 6.5517, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 6.611834860831118e-06, | |
| "loss": 6.5486, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "eval_accuracy": 0.11219145006851503, | |
| "eval_loss": 6.546039581298828, | |
| "eval_runtime": 601.6578, | |
| "eval_samples_per_second": 512.511, | |
| "eval_steps_per_second": 5.34, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 6.594894035135273e-06, | |
| "loss": 6.5533, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.577953209439428e-06, | |
| "loss": 6.5467, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.561012383743585e-06, | |
| "loss": 6.5454, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.54407155804774e-06, | |
| "loss": 6.5416, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.5271307323518954e-06, | |
| "loss": 6.5416, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.510189906656051e-06, | |
| "loss": 6.543, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.493249080960207e-06, | |
| "loss": 6.5443, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.476308255264362e-06, | |
| "loss": 6.534, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.459367429568517e-06, | |
| "loss": 6.5395, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.442426603872674e-06, | |
| "loss": 6.5386, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.425485778176829e-06, | |
| "loss": 6.5399, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.4085449524809836e-06, | |
| "loss": 6.5382, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.39160412678514e-06, | |
| "loss": 6.5309, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 6.374663301089296e-06, | |
| "loss": 6.533, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.357722475393451e-06, | |
| "loss": 6.5269, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.340781649697607e-06, | |
| "loss": 6.533, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.323840824001762e-06, | |
| "loss": 6.5268, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.306899998305918e-06, | |
| "loss": 6.5326, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.289959172610074e-06, | |
| "loss": 6.5306, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.273018346914229e-06, | |
| "loss": 6.5246, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_accuracy": 0.11442907817610445, | |
| "eval_loss": 6.527458667755127, | |
| "eval_runtime": 601.5697, | |
| "eval_samples_per_second": 512.586, | |
| "eval_steps_per_second": 5.341, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.256077521218385e-06, | |
| "loss": 6.5326, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.2391366955225405e-06, | |
| "loss": 6.535, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.222195869826696e-06, | |
| "loss": 6.5223, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.205255044130851e-06, | |
| "loss": 6.5225, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.1883142184350076e-06, | |
| "loss": 6.518, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.171373392739162e-06, | |
| "loss": 6.5249, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.154432567043318e-06, | |
| "loss": 6.5216, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.137491741347473e-06, | |
| "loss": 6.5238, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.1205509156516295e-06, | |
| "loss": 6.5162, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.103610089955785e-06, | |
| "loss": 6.5211, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.08666926425994e-06, | |
| "loss": 6.5196, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.0697284385640965e-06, | |
| "loss": 6.5158, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.052787612868251e-06, | |
| "loss": 6.5173, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.035846787172407e-06, | |
| "loss": 6.5195, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.018905961476564e-06, | |
| "loss": 6.525, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.0019651357807185e-06, | |
| "loss": 6.5169, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.985024310084873e-06, | |
| "loss": 6.5191, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.96808348438903e-06, | |
| "loss": 6.5111, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.9511426586931855e-06, | |
| "loss": 6.5128, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.93420183299734e-06, | |
| "loss": 6.5069, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_accuracy": 0.11622728049791442, | |
| "eval_loss": 6.511504650115967, | |
| "eval_runtime": 605.5053, | |
| "eval_samples_per_second": 509.254, | |
| "eval_steps_per_second": 5.306, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.917261007301497e-06, | |
| "loss": 6.5104, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.900320181605652e-06, | |
| "loss": 6.5106, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.8833793559098074e-06, | |
| "loss": 6.5054, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.866438530213962e-06, | |
| "loss": 6.5124, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.849497704518119e-06, | |
| "loss": 6.508, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.8325568788222745e-06, | |
| "loss": 6.5066, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.815616053126429e-06, | |
| "loss": 6.5102, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.798675227430586e-06, | |
| "loss": 6.5045, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.781734401734741e-06, | |
| "loss": 6.5027, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.7647935760388964e-06, | |
| "loss": 6.502, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.747852750343053e-06, | |
| "loss": 6.4952, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.730911924647208e-06, | |
| "loss": 6.5016, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.713971098951363e-06, | |
| "loss": 6.5039, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.697030273255519e-06, | |
| "loss": 6.5077, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.680089447559675e-06, | |
| "loss": 6.5012, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5.66314862186383e-06, | |
| "loss": 6.4987, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.646207796167986e-06, | |
| "loss": 6.4977, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.629266970472141e-06, | |
| "loss": 6.4992, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.612326144776297e-06, | |
| "loss": 6.498, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.595385319080452e-06, | |
| "loss": 6.5001, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "eval_accuracy": 0.1180354638977201, | |
| "eval_loss": 6.496232509613037, | |
| "eval_runtime": 603.9932, | |
| "eval_samples_per_second": 510.529, | |
| "eval_steps_per_second": 5.32, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.578444493384608e-06, | |
| "loss": 6.4969, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.561503667688764e-06, | |
| "loss": 6.4959, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5.544562841992919e-06, | |
| "loss": 6.4939, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5.527622016297075e-06, | |
| "loss": 6.4908, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5.51068119060123e-06, | |
| "loss": 6.4954, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5.493740364905386e-06, | |
| "loss": 6.4911, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5.4767995392095415e-06, | |
| "loss": 6.4886, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5.459858713513697e-06, | |
| "loss": 6.492, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.442917887817852e-06, | |
| "loss": 6.4917, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5.4259770621220086e-06, | |
| "loss": 6.491, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5.409036236426164e-06, | |
| "loss": 6.4872, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5.392095410730319e-06, | |
| "loss": 6.4833, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5.375154585034476e-06, | |
| "loss": 6.4912, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5.3582137593386305e-06, | |
| "loss": 6.4895, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.341272933642786e-06, | |
| "loss": 6.4869, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.324332107946941e-06, | |
| "loss": 6.4875, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.3073912822510975e-06, | |
| "loss": 6.4895, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.290450456555252e-06, | |
| "loss": 6.48, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5.273509630859408e-06, | |
| "loss": 6.486, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.256568805163565e-06, | |
| "loss": 6.4785, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "eval_accuracy": 0.11971311116288914, | |
| "eval_loss": 6.4821858406066895, | |
| "eval_runtime": 600.3142, | |
| "eval_samples_per_second": 513.658, | |
| "eval_steps_per_second": 5.352, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.2396279794677195e-06, | |
| "loss": 6.4843, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.222687153771875e-06, | |
| "loss": 6.4887, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 5.205746328076031e-06, | |
| "loss": 6.4861, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.1888055023801865e-06, | |
| "loss": 6.4865, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.171864676684341e-06, | |
| "loss": 6.4767, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 5.154923850988498e-06, | |
| "loss": 6.4775, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.137983025292654e-06, | |
| "loss": 6.4793, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.1210421995968085e-06, | |
| "loss": 6.479, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 5.104101373900965e-06, | |
| "loss": 6.4783, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.08716054820512e-06, | |
| "loss": 6.479, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.0702197225092755e-06, | |
| "loss": 6.4804, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.053278896813432e-06, | |
| "loss": 6.473, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.036338071117587e-06, | |
| "loss": 6.4769, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.019397245421742e-06, | |
| "loss": 6.4778, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.0024564197258974e-06, | |
| "loss": 6.4764, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.985515594030054e-06, | |
| "loss": 6.47, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.968574768334209e-06, | |
| "loss": 6.4716, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.9516339426383645e-06, | |
| "loss": 6.4754, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.93469311694252e-06, | |
| "loss": 6.4741, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.917752291246676e-06, | |
| "loss": 6.4706, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "eval_accuracy": 0.12115731725892939, | |
| "eval_loss": 6.471449851989746, | |
| "eval_runtime": 601.7368, | |
| "eval_samples_per_second": 512.443, | |
| "eval_steps_per_second": 5.34, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.900811465550832e-06, | |
| "loss": 6.4739, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.8838706398549864e-06, | |
| "loss": 6.4717, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.866929814159143e-06, | |
| "loss": 6.468, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.849988988463299e-06, | |
| "loss": 6.4758, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.8330481627674535e-06, | |
| "loss": 6.4698, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.816107337071609e-06, | |
| "loss": 6.4681, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.799166511375765e-06, | |
| "loss": 6.47, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.782225685679921e-06, | |
| "loss": 6.4711, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.765284859984076e-06, | |
| "loss": 6.4707, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.748344034288231e-06, | |
| "loss": 6.4717, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.731403208592387e-06, | |
| "loss": 6.4677, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.714462382896543e-06, | |
| "loss": 6.4677, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.697521557200698e-06, | |
| "loss": 6.4642, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.680580731504854e-06, | |
| "loss": 6.4587, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.6636399058090096e-06, | |
| "loss": 6.4666, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.646699080113165e-06, | |
| "loss": 6.4691, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.629758254417321e-06, | |
| "loss": 6.4628, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.612817428721476e-06, | |
| "loss": 6.4632, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.5958766030256315e-06, | |
| "loss": 6.4586, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.578935777329788e-06, | |
| "loss": 6.4612, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "eval_accuracy": 0.12253875669316174, | |
| "eval_loss": 6.461043357849121, | |
| "eval_runtime": 602.3343, | |
| "eval_samples_per_second": 511.935, | |
| "eval_steps_per_second": 5.334, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.561994951633943e-06, | |
| "loss": 6.4673, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.5450541259380986e-06, | |
| "loss": 6.4579, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.528113300242254e-06, | |
| "loss": 6.4594, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.51117247454641e-06, | |
| "loss": 6.4634, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.494231648850566e-06, | |
| "loss": 6.4586, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.4772908231547205e-06, | |
| "loss": 6.4587, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.460349997458876e-06, | |
| "loss": 6.4599, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.443409171763033e-06, | |
| "loss": 6.4596, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.4264683460671875e-06, | |
| "loss": 6.4547, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.409527520371343e-06, | |
| "loss": 6.4575, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.392586694675499e-06, | |
| "loss": 6.4556, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.375645868979655e-06, | |
| "loss": 6.4581, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.35870504328381e-06, | |
| "loss": 6.4535, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.341764217587965e-06, | |
| "loss": 6.4617, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.324823391892121e-06, | |
| "loss": 6.4535, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.307882566196277e-06, | |
| "loss": 6.4543, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.290941740500432e-06, | |
| "loss": 6.4535, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.274000914804588e-06, | |
| "loss": 6.4523, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.257060089108744e-06, | |
| "loss": 6.4542, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.240119263412899e-06, | |
| "loss": 6.4485, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "eval_accuracy": 0.1232825828604018, | |
| "eval_loss": 6.4530487060546875, | |
| "eval_runtime": 601.0649, | |
| "eval_samples_per_second": 513.016, | |
| "eval_steps_per_second": 5.346, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.223178437717055e-06, | |
| "loss": 6.455, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.20623761202121e-06, | |
| "loss": 6.4519, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.1892967863253655e-06, | |
| "loss": 6.455, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.172355960629522e-06, | |
| "loss": 6.4511, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.155415134933677e-06, | |
| "loss": 6.4468, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.138474309237833e-06, | |
| "loss": 6.4473, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.121533483541988e-06, | |
| "loss": 6.4471, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.104592657846144e-06, | |
| "loss": 6.4467, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.0876518321503e-06, | |
| "loss": 6.4515, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.0707110064544545e-06, | |
| "loss": 6.448, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.05377018075861e-06, | |
| "loss": 6.4491, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.036829355062766e-06, | |
| "loss": 6.4505, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.019888529366922e-06, | |
| "loss": 6.4471, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.002947703671077e-06, | |
| "loss": 6.4435, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.986006877975233e-06, | |
| "loss": 6.4472, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.969066052279389e-06, | |
| "loss": 6.4453, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.952125226583544e-06, | |
| "loss": 6.449, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.935184400887699e-06, | |
| "loss": 6.4475, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.918243575191855e-06, | |
| "loss": 6.4411, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.9013027494960106e-06, | |
| "loss": 6.4477, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "eval_accuracy": 0.1243327551641345, | |
| "eval_loss": 6.444092273712158, | |
| "eval_runtime": 605.0337, | |
| "eval_samples_per_second": 509.651, | |
| "eval_steps_per_second": 5.31, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.884361923800166e-06, | |
| "loss": 6.4434, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.867421098104322e-06, | |
| "loss": 6.4469, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.850480272408478e-06, | |
| "loss": 6.4424, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.833539446712633e-06, | |
| "loss": 6.4427, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.816598621016789e-06, | |
| "loss": 6.4407, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.799657795320944e-06, | |
| "loss": 6.4434, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.7827169696251e-06, | |
| "loss": 6.442, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.7657761439292557e-06, | |
| "loss": 6.438, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.748835318233411e-06, | |
| "loss": 6.4435, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.7318944925375666e-06, | |
| "loss": 6.4418, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.7149536668417223e-06, | |
| "loss": 6.4368, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.6980128411458776e-06, | |
| "loss": 6.4424, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.6810720154500333e-06, | |
| "loss": 6.4395, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.6641311897541886e-06, | |
| "loss": 6.4424, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.6471903640583447e-06, | |
| "loss": 6.4355, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.6302495383625004e-06, | |
| "loss": 6.4401, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.6133087126666556e-06, | |
| "loss": 6.4408, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.5963678869708113e-06, | |
| "loss": 6.4426, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.579427061274967e-06, | |
| "loss": 6.4364, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.5624862355791223e-06, | |
| "loss": 6.4373, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "eval_accuracy": 0.12513166304842002, | |
| "eval_loss": 6.439530849456787, | |
| "eval_runtime": 605.3063, | |
| "eval_samples_per_second": 509.421, | |
| "eval_steps_per_second": 5.308, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.545545409883278e-06, | |
| "loss": 6.4369, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.5286045841874332e-06, | |
| "loss": 6.4347, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.511663758491589e-06, | |
| "loss": 6.435, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.494722932795745e-06, | |
| "loss": 6.4335, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.4777821070999003e-06, | |
| "loss": 6.4391, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.460841281404056e-06, | |
| "loss": 6.4361, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.4439004557082117e-06, | |
| "loss": 6.4367, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.426959630012367e-06, | |
| "loss": 6.4366, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.4100188043165226e-06, | |
| "loss": 6.4339, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.393077978620678e-06, | |
| "loss": 6.4402, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.3761371529248336e-06, | |
| "loss": 6.433, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.3591963272289897e-06, | |
| "loss": 6.4381, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.342255501533145e-06, | |
| "loss": 6.4369, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.3253146758373007e-06, | |
| "loss": 6.4371, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.3083738501414564e-06, | |
| "loss": 6.4339, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.2914330244456116e-06, | |
| "loss": 6.434, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.2744921987497673e-06, | |
| "loss": 6.4282, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.2575513730539226e-06, | |
| "loss": 6.4306, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.2406105473580783e-06, | |
| "loss": 6.4309, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.2236697216622344e-06, | |
| "loss": 6.4351, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "eval_accuracy": 0.12585777151942737, | |
| "eval_loss": 6.432172775268555, | |
| "eval_runtime": 603.5717, | |
| "eval_samples_per_second": 510.885, | |
| "eval_steps_per_second": 5.323, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.2067288959663897e-06, | |
| "loss": 6.4332, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.1897880702705454e-06, | |
| "loss": 6.435, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.172847244574701e-06, | |
| "loss": 6.4315, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.1559064188788563e-06, | |
| "loss": 6.4349, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.138965593183012e-06, | |
| "loss": 6.4264, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.1220247674871673e-06, | |
| "loss": 6.4345, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.105083941791323e-06, | |
| "loss": 6.4287, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.088143116095479e-06, | |
| "loss": 6.433, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.071202290399634e-06, | |
| "loss": 6.4296, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.05426146470379e-06, | |
| "loss": 6.4261, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.0373206390079457e-06, | |
| "loss": 6.4337, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.020379813312101e-06, | |
| "loss": 6.4286, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.0034389876162567e-06, | |
| "loss": 6.4267, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.986498161920412e-06, | |
| "loss": 6.431, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.9695573362245676e-06, | |
| "loss": 6.4325, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 2.9526165105287238e-06, | |
| "loss": 6.4253, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 2.9356756848328786e-06, | |
| "loss": 6.426, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.9187348591370347e-06, | |
| "loss": 6.4327, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 2.9017940334411904e-06, | |
| "loss": 6.4291, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 2.8848532077453457e-06, | |
| "loss": 6.4273, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "eval_accuracy": 0.1262452488408161, | |
| "eval_loss": 6.426393508911133, | |
| "eval_runtime": 606.2126, | |
| "eval_samples_per_second": 508.66, | |
| "eval_steps_per_second": 5.3, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.8679123820495014e-06, | |
| "loss": 6.4263, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 2.8509715563536566e-06, | |
| "loss": 6.4267, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 2.8340307306578123e-06, | |
| "loss": 6.425, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 2.817089904961968e-06, | |
| "loss": 6.4224, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.8001490792661233e-06, | |
| "loss": 6.4217, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.7832082535702794e-06, | |
| "loss": 6.4258, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 2.766267427874435e-06, | |
| "loss": 6.4253, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 2.7493266021785904e-06, | |
| "loss": 6.4266, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 2.732385776482746e-06, | |
| "loss": 6.4271, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 2.7154449507869017e-06, | |
| "loss": 6.4294, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 2.698504125091057e-06, | |
| "loss": 6.4243, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.6815632993952127e-06, | |
| "loss": 6.4242, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.664622473699368e-06, | |
| "loss": 6.4242, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.647681648003524e-06, | |
| "loss": 6.4262, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.6307408223076798e-06, | |
| "loss": 6.4197, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.613799996611835e-06, | |
| "loss": 6.4246, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.5968591709159907e-06, | |
| "loss": 6.4237, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.5799183452201464e-06, | |
| "loss": 6.421, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 2.5629775195243017e-06, | |
| "loss": 6.4207, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 2.5460366938284574e-06, | |
| "loss": 6.4153, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "eval_accuracy": 0.12687429199787553, | |
| "eval_loss": 6.421925067901611, | |
| "eval_runtime": 603.0076, | |
| "eval_samples_per_second": 511.363, | |
| "eval_steps_per_second": 5.328, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 2.5290958681326126e-06, | |
| "loss": 6.4211, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 2.5121550424367688e-06, | |
| "loss": 6.4257, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 2.495214216740924e-06, | |
| "loss": 6.4207, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.4782733910450797e-06, | |
| "loss": 6.4207, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 2.4613325653492354e-06, | |
| "loss": 6.4252, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 2.4443917396533907e-06, | |
| "loss": 6.4226, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.4274509139575464e-06, | |
| "loss": 6.4271, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.410510088261702e-06, | |
| "loss": 6.4204, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 2.3935692625658577e-06, | |
| "loss": 6.423, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 2.376628436870013e-06, | |
| "loss": 6.4211, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 2.3596876111741687e-06, | |
| "loss": 6.4206, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 2.3427467854783244e-06, | |
| "loss": 6.4194, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 2.32580595978248e-06, | |
| "loss": 6.4166, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 2.3088651340866354e-06, | |
| "loss": 6.4259, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 2.291924308390791e-06, | |
| "loss": 6.4142, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.2749834826949467e-06, | |
| "loss": 6.4215, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 2.2580426569991024e-06, | |
| "loss": 6.4135, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.2411018313032577e-06, | |
| "loss": 6.4174, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.2241610056074134e-06, | |
| "loss": 6.4231, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 2.207220179911569e-06, | |
| "loss": 6.4188, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "eval_accuracy": 0.12740316889553732, | |
| "eval_loss": 6.418206691741943, | |
| "eval_runtime": 604.087, | |
| "eval_samples_per_second": 510.45, | |
| "eval_steps_per_second": 5.319, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 2.1902793542157248e-06, | |
| "loss": 6.4167, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 2.17333852851988e-06, | |
| "loss": 6.4203, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 2.1563977028240357e-06, | |
| "loss": 6.417, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 2.1394568771281914e-06, | |
| "loss": 6.4171, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 2.122516051432347e-06, | |
| "loss": 6.4182, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 2.1055752257365024e-06, | |
| "loss": 6.4181, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 2.088634400040658e-06, | |
| "loss": 6.4159, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2.0716935743448138e-06, | |
| "loss": 6.4162, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.0547527486489694e-06, | |
| "loss": 6.4137, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.0378119229531247e-06, | |
| "loss": 6.4163, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 2.0208710972572804e-06, | |
| "loss": 6.4211, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 2.003930271561436e-06, | |
| "loss": 6.4166, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.986989445865592e-06, | |
| "loss": 6.416, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.970048620169747e-06, | |
| "loss": 6.4162, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.9531077944739027e-06, | |
| "loss": 6.4138, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.9361669687780584e-06, | |
| "loss": 6.411, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.919226143082214e-06, | |
| "loss": 6.4124, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.9022853173863696e-06, | |
| "loss": 6.4154, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.885344491690525e-06, | |
| "loss": 6.4149, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.8684036659946808e-06, | |
| "loss": 6.4128, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "eval_accuracy": 0.12784806575663238, | |
| "eval_loss": 6.414950370788574, | |
| "eval_runtime": 607.4915, | |
| "eval_samples_per_second": 507.589, | |
| "eval_steps_per_second": 5.289, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.8514628402988363e-06, | |
| "loss": 6.414, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 1.8345220146029917e-06, | |
| "loss": 6.41, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 1.8175811889071474e-06, | |
| "loss": 6.4131, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 1.8006403632113031e-06, | |
| "loss": 6.4146, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.7836995375154586e-06, | |
| "loss": 6.4126, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.766758711819614e-06, | |
| "loss": 6.4164, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.7498178861237698e-06, | |
| "loss": 6.4147, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 1.7328770604279255e-06, | |
| "loss": 6.4068, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 1.715936234732081e-06, | |
| "loss": 6.4117, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 1.6989954090362364e-06, | |
| "loss": 6.415, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.6820545833403921e-06, | |
| "loss": 6.4131, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.6651137576445478e-06, | |
| "loss": 6.4125, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.6481729319487033e-06, | |
| "loss": 6.4145, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 1.6312321062528588e-06, | |
| "loss": 6.4155, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 1.6142912805570147e-06, | |
| "loss": 6.4066, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.5973504548611701e-06, | |
| "loss": 6.4129, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 1.5804096291653256e-06, | |
| "loss": 6.4079, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 1.563468803469481e-06, | |
| "loss": 6.4124, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 1.546527977773637e-06, | |
| "loss": 6.4124, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 1.5295871520777925e-06, | |
| "loss": 6.4189, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "eval_accuracy": 0.12798065025917943, | |
| "eval_loss": 6.412052154541016, | |
| "eval_runtime": 604.5339, | |
| "eval_samples_per_second": 510.072, | |
| "eval_steps_per_second": 5.315, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 1.512646326381948e-06, | |
| "loss": 6.4115, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.4957055006861034e-06, | |
| "loss": 6.4086, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.4787646749902593e-06, | |
| "loss": 6.4153, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 1.4618238492944148e-06, | |
| "loss": 6.4042, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 1.4448830235985703e-06, | |
| "loss": 6.4188, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 1.4279421979027258e-06, | |
| "loss": 6.4107, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 1.4110013722068817e-06, | |
| "loss": 6.4104, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 1.3940605465110372e-06, | |
| "loss": 6.4108, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 1.3771197208151926e-06, | |
| "loss": 6.4096, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 1.3601788951193481e-06, | |
| "loss": 6.4139, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.343238069423504e-06, | |
| "loss": 6.4142, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 1.3262972437276595e-06, | |
| "loss": 6.4186, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 1.309356418031815e-06, | |
| "loss": 6.4126, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 1.2924155923359705e-06, | |
| "loss": 6.408, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 1.2754747666401264e-06, | |
| "loss": 6.4109, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 1.2585339409442818e-06, | |
| "loss": 6.4117, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 1.2415931152484373e-06, | |
| "loss": 6.4088, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 1.224652289552593e-06, | |
| "loss": 6.4152, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 1.2077114638567485e-06, | |
| "loss": 6.41, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 1.1907706381609042e-06, | |
| "loss": 6.4102, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "eval_accuracy": 0.12817913105652246, | |
| "eval_loss": 6.411165237426758, | |
| "eval_runtime": 602.0627, | |
| "eval_samples_per_second": 512.166, | |
| "eval_steps_per_second": 5.337, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 1.1738298124650597e-06, | |
| "loss": 6.4107, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 1.1568889867692153e-06, | |
| "loss": 6.4099, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 1.1399481610733708e-06, | |
| "loss": 6.4162, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 1.1230073353775265e-06, | |
| "loss": 6.4103, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 1.106066509681682e-06, | |
| "loss": 6.4159, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 1.0891256839858377e-06, | |
| "loss": 6.4052, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 1.0721848582899932e-06, | |
| "loss": 6.4079, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 1.0552440325941489e-06, | |
| "loss": 6.4071, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 1.0383032068983043e-06, | |
| "loss": 6.4131, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 1.02136238120246e-06, | |
| "loss": 6.4077, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 1.0044215555066155e-06, | |
| "loss": 6.4113, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 9.87480729810771e-07, | |
| "loss": 6.4141, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 9.705399041149267e-07, | |
| "loss": 6.4081, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 9.535990784190823e-07, | |
| "loss": 6.4071, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 9.366582527232377e-07, | |
| "loss": 6.4078, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 9.197174270273934e-07, | |
| "loss": 6.4117, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 9.027766013315489e-07, | |
| "loss": 6.4077, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 8.858357756357046e-07, | |
| "loss": 6.4106, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 8.688949499398601e-07, | |
| "loss": 6.4134, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 8.519541242440158e-07, | |
| "loss": 6.4105, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "eval_accuracy": 0.1284733583672517, | |
| "eval_loss": 6.408677577972412, | |
| "eval_runtime": 603.8755, | |
| "eval_samples_per_second": 510.628, | |
| "eval_steps_per_second": 5.321, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 8.350132985481713e-07, | |
| "loss": 6.4046, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 8.180724728523269e-07, | |
| "loss": 6.4109, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 8.011316471564824e-07, | |
| "loss": 6.4117, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 7.841908214606381e-07, | |
| "loss": 6.4107, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 7.672499957647936e-07, | |
| "loss": 6.415, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 7.503091700689493e-07, | |
| "loss": 6.4066, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 7.333683443731048e-07, | |
| "loss": 6.4075, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 7.164275186772605e-07, | |
| "loss": 6.4086, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 6.994866929814159e-07, | |
| "loss": 6.4054, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 6.825458672855715e-07, | |
| "loss": 6.4022, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 6.656050415897271e-07, | |
| "loss": 6.406, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 6.486642158938827e-07, | |
| "loss": 6.4032, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 6.317233901980383e-07, | |
| "loss": 6.4105, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 6.147825645021939e-07, | |
| "loss": 6.4073, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 5.978417388063494e-07, | |
| "loss": 6.4058, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 5.80900913110505e-07, | |
| "loss": 6.4123, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 5.639600874146606e-07, | |
| "loss": 6.4132, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.470192617188162e-07, | |
| "loss": 6.4072, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.300784360229718e-07, | |
| "loss": 6.4085, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 5.131376103271274e-07, | |
| "loss": 6.4065, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_accuracy": 0.12868705268687342, | |
| "eval_loss": 6.406748294830322, | |
| "eval_runtime": 608.2162, | |
| "eval_samples_per_second": 506.984, | |
| "eval_steps_per_second": 5.283, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 4.96196784631283e-07, | |
| "loss": 6.409, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 4.792559589354385e-07, | |
| "loss": 6.4061, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 4.623151332395941e-07, | |
| "loss": 6.4078, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 4.453743075437497e-07, | |
| "loss": 6.4084, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.284334818479053e-07, | |
| "loss": 6.412, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 4.114926561520609e-07, | |
| "loss": 6.4085, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 3.9455183045621646e-07, | |
| "loss": 6.4022, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 3.7761100476037205e-07, | |
| "loss": 6.4065, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.6067017906452763e-07, | |
| "loss": 6.4114, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.437293533686832e-07, | |
| "loss": 6.4085, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 3.267885276728388e-07, | |
| "loss": 6.4117, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.098477019769944e-07, | |
| "loss": 6.4066, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 2.9290687628114997e-07, | |
| "loss": 6.4074, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 2.7596605058530556e-07, | |
| "loss": 6.4059, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 2.5902522488946114e-07, | |
| "loss": 6.4027, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 2.4208439919361673e-07, | |
| "loss": 6.4034, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 2.2514357349777229e-07, | |
| "loss": 6.4053, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 2.0820274780192787e-07, | |
| "loss": 6.4066, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 1.9126192210608346e-07, | |
| "loss": 6.4098, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 1.7432109641023904e-07, | |
| "loss": 6.4082, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "eval_accuracy": 0.12851127491212921, | |
| "eval_loss": 6.407046794891357, | |
| "eval_runtime": 604.7864, | |
| "eval_samples_per_second": 509.859, | |
| "eval_steps_per_second": 5.313, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 1.5738027071439465e-07, | |
| "loss": 6.4019, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 1.404394450185502e-07, | |
| "loss": 6.4073, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 1.234986193227058e-07, | |
| "loss": 6.403, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 1.0655779362686138e-07, | |
| "loss": 6.4098, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 8.961696793101697e-08, | |
| "loss": 6.4055, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 7.267614223517255e-08, | |
| "loss": 6.409, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 5.5735316539328136e-08, | |
| "loss": 6.4094, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 3.8794490843483714e-08, | |
| "loss": 6.4038, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 2.18536651476393e-08, | |
| "loss": 6.4017, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.912839451794881e-09, | |
| "loss": 6.4089, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 305145, | |
| "total_flos": 6.419869954421555e+16, | |
| "train_loss": 6.680038520005282, | |
| "train_runtime": 123390.5046, | |
| "train_samples_per_second": 237.407, | |
| "train_steps_per_second": 2.473 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 305145, | |
| "num_train_epochs": 5, | |
| "save_steps": 10000, | |
| "total_flos": 6.419869954421555e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |